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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
65662f65-de60-4599-af2d-05a5c574f6dd | cross-document-temporal-relation-extraction | null | null | https://openreview.net/forum?id=h9FRrlcs8yo | https://openreview.net/pdf?id=h9FRrlcs8yo | Cross-Document Temporal Relation Extraction with Temporal Anchoring Events | Automatically extracting a timeline on a certain topic from multiple documents has been a challenge in natural language processing, partly due to the difficulty of collecting large amounts of training data. In this work, we collect a dataset for cross-document timeline extraction from online news that gives access to metadata such as hyperlinks and publication dates. The metadata allows us to define a set of important events while linking them to time anchors, which opens the opportunity to scale up data collection. Furthermore, with this set of linked news articles, we propose a method to enhance the inference process of temporal relation prediction, by utilizing a model to link events to a set of anchoring events that are added to the inference program. We report performance of common neural models and show that our method can boost the performance of all baseline models. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['temporal-relation-extraction'] | ['natural-language-processing'] | [-5.99973761e-02 1.26621544e-01 -8.18355620e-01 -7.30839312e-01
-8.78302097e-01 -7.44679928e-01 1.15646863e+00 8.31454039e-01
-5.39458215e-01 6.58587754e-01 8.49465132e-01 -1.19781308e-01
-2.05422014e-01 -9.29214478e-01 -8.42947721e-01 -8.25941190e-02
-3.54599208e-01 5.42481482e-01 4.02191073e-01 -1.08939610e-01
8.86439309e-02 3.14156622e-01 -1.24532449e+00 3.72188419e-01
4.13950831e-01 9.10451293e-01 -8.47618002e-03 3.52894306e-01
-3.81077856e-01 9.49581981e-01 -6.17761314e-01 -4.27471638e-01
-1.80094287e-01 -7.75840804e-02 -9.58188474e-01 -5.37789285e-01
3.89878184e-01 -3.72750640e-01 -5.47630548e-01 5.16687930e-01
2.82545034e-02 4.09885138e-01 5.91213584e-01 -1.37924969e+00
-2.59774655e-01 1.36440647e+00 -6.13437116e-01 6.84213519e-01
3.46427321e-01 -6.63364053e-01 1.43309307e+00 -3.05831492e-01
1.13064969e+00 9.34267581e-01 7.39926279e-01 1.23164682e-02
-1.14263296e+00 -4.45465595e-01 4.52727199e-01 2.85025239e-01
-1.02682626e+00 -3.88682902e-01 8.13719690e-01 -3.97125632e-01
1.25976014e+00 2.36227170e-01 6.34561718e-01 1.23273599e+00
1.28484040e-01 7.16633499e-01 4.40399826e-01 -5.15236259e-01
-4.32491489e-02 -2.06668824e-01 6.78552747e-01 2.76915848e-01
3.16787623e-02 -1.27352938e-01 -8.03806782e-01 -2.01510146e-01
2.98832655e-01 6.88670054e-02 -4.48659286e-02 1.49279267e-01
-1.23450935e+00 6.84152603e-01 2.83468395e-01 4.49813098e-01
-3.79031390e-01 2.52596829e-02 5.85243583e-01 7.77607039e-03
8.78443062e-01 6.22872829e-01 -7.26356447e-01 -1.57168657e-01
-1.07058740e+00 4.83478427e-01 1.21101701e+00 9.94898617e-01
4.70545292e-01 -7.01429486e-01 -4.65092182e-01 7.19054401e-01
8.73487517e-02 4.82420400e-02 2.68420160e-01 -7.14006186e-01
8.59600782e-01 5.60213447e-01 2.56161720e-01 -1.18718219e+00
-7.90795088e-01 -2.31228560e-01 -3.28455567e-01 -5.81691027e-01
4.87162858e-01 -1.41125232e-01 -5.49955964e-01 1.95065343e+00
5.13835013e-01 2.17490405e-01 -2.22518176e-01 5.11795402e-01
6.37527347e-01 1.10486650e+00 1.96732596e-01 -4.75864530e-01
1.53049755e+00 -7.79670954e-01 -9.81782496e-01 -8.84557888e-02
6.89085066e-01 -6.09782696e-01 7.30470359e-01 2.43866384e-01
-7.92500317e-01 -1.77722201e-01 -8.53713632e-01 -4.38428164e-01
-7.73991942e-01 -1.59830898e-01 1.03791463e+00 -1.98291942e-01
-4.15972412e-01 7.14664578e-01 -1.13094807e+00 -6.01560533e-01
1.08315945e-01 1.95642505e-02 -1.99734151e-01 2.00521529e-01
-1.55997908e+00 9.59364414e-01 8.45802426e-01 -2.75836140e-01
-3.60107362e-01 -7.69522429e-01 -7.17085779e-01 1.19610012e-01
5.94654977e-01 -4.68751639e-01 1.58387470e+00 -3.32525462e-01
-1.06741726e+00 6.41555786e-01 -2.63953358e-01 -7.76689112e-01
2.19517827e-01 -5.28383970e-01 -8.11678648e-01 1.30164236e-01
3.56681108e-01 4.76803184e-01 4.70009446e-01 -5.85430145e-01
-1.07829893e+00 -2.77408987e-01 9.63050053e-02 -1.19838156e-01
-6.93613708e-01 5.51921070e-01 -9.37585354e-01 -7.80092597e-01
-7.68654123e-02 -8.03754389e-01 -2.09295005e-02 -4.16687906e-01
-6.29089177e-01 -7.46797740e-01 7.31234431e-01 -8.73619020e-01
1.85878003e+00 -1.92589319e+00 1.36259973e-01 8.60674530e-02
7.82287866e-02 -5.32316923e-01 2.05027029e-01 6.31438851e-01
1.28744200e-01 7.03974888e-02 2.06346989e-01 -2.63244778e-01
1.06409825e-01 2.32307523e-01 -8.29630852e-01 1.25733808e-01
7.12588653e-02 8.14484656e-01 -1.01533222e+00 -7.17961669e-01
-4.38561708e-01 2.66343296e-01 -3.73592913e-01 1.50911108e-01
-7.26143956e-01 8.73222277e-02 -5.57839632e-01 2.66607493e-01
3.64901014e-02 -3.23253781e-01 1.98143661e-01 -3.05016041e-01
-2.62521476e-01 1.13329947e+00 -9.59414244e-01 1.67864692e+00
-4.69994485e-01 8.01808655e-01 -6.03141129e-01 -6.53394401e-01
6.52166188e-01 4.61516678e-01 8.60569119e-01 -5.59838891e-01
-6.50741719e-03 -2.11921498e-01 -4.94382560e-01 -5.84879220e-01
8.88002276e-01 3.37093860e-01 -4.35966313e-01 8.69295657e-01
9.26484913e-02 1.48089349e-01 7.10536957e-01 4.43031251e-01
1.15794027e+00 1.82776392e-01 2.65827179e-01 1.57551482e-01
-7.22980872e-02 1.63451359e-01 5.47764659e-01 6.36000216e-01
2.48796031e-01 1.23284884e-01 7.12672114e-01 -6.19517148e-01
-1.19640040e+00 -8.28748345e-01 -2.76600093e-01 1.41107202e+00
-4.63853419e-01 -8.76851857e-01 -3.63020420e-01 -9.06231463e-01
1.74897015e-02 9.84272897e-01 -7.47962594e-01 4.84910794e-02
-8.52239251e-01 -8.16248238e-01 6.14814162e-01 6.94741845e-01
6.49600253e-02 -6.05916381e-01 -3.42853338e-01 5.01319408e-01
-7.35417306e-01 -1.28562260e+00 -6.79494023e-01 4.42497581e-01
-7.24901199e-01 -8.05519164e-01 -5.70679195e-02 -5.79787195e-01
4.00043041e-01 -4.09409180e-02 1.24680459e+00 -3.30764949e-01
2.24405631e-01 -1.06912695e-01 -2.83107698e-01 -7.23536789e-01
-2.27173373e-01 6.71283722e-01 4.40342724e-02 -2.69251406e-01
6.90603733e-01 -5.05413771e-01 -1.17969774e-01 2.01138571e-01
-8.54717016e-01 2.44695365e-01 4.97727003e-03 4.83250350e-01
4.17540371e-01 2.88904727e-01 5.77662706e-01 -1.02209365e+00
4.57025498e-01 -8.49388242e-01 -1.02821040e+00 1.03486866e-01
-6.70172095e-01 3.15161407e-01 5.21940351e-01 -6.01524711e-01
-1.21476805e+00 -1.79799840e-01 2.32849658e-01 2.14948088e-01
1.30878940e-01 1.17048740e+00 6.60994127e-02 8.57558906e-01
5.55638790e-01 -2.99465626e-01 -5.43469787e-01 -7.11925864e-01
7.49335647e-01 5.64561367e-01 7.95386374e-01 -6.50793433e-01
5.75967789e-01 4.24930930e-01 -1.85106412e-01 -3.75413150e-01
-1.62909532e+00 -6.56247735e-01 -7.57702410e-01 2.33250316e-02
5.61484337e-01 -9.66596901e-01 -2.64161289e-01 5.03638089e-02
-1.32998347e+00 -1.12548187e-01 -1.30160198e-01 7.19293773e-01
-2.25754321e-01 -3.12549658e-02 -8.55820239e-01 -2.78738141e-01
-2.66021490e-01 -3.20622414e-01 7.78246880e-01 1.10753834e-01
-7.56395042e-01 -1.08175814e+00 3.23817998e-01 1.45248130e-01
3.83050852e-02 3.76978397e-01 1.02848625e+00 -1.17226088e+00
-4.71219033e-01 -4.04449850e-01 -7.42827654e-02 -3.17925602e-01
1.27227545e-01 3.68077904e-01 -7.93459117e-01 9.81171429e-03
-2.05480397e-01 -4.33272356e-03 8.70570004e-01 4.03633535e-01
1.09827995e+00 -7.13995636e-01 -1.00983822e+00 6.25018477e-01
9.48953807e-01 1.80357382e-01 2.86034942e-01 7.66002595e-01
6.66718960e-01 6.53758764e-01 6.02034330e-01 5.64847827e-01
7.07768917e-01 9.04178917e-01 -2.03253582e-01 1.46119520e-01
1.34250522e-01 -5.25765061e-01 7.62887746e-02 9.18197751e-01
2.72766445e-02 -6.15887582e-01 -8.66587639e-01 7.13949502e-01
-1.96769297e+00 -1.15980721e+00 -1.37176037e-01 2.01730227e+00
1.48409843e+00 4.81187940e-01 2.08415091e-01 1.56106409e-02
6.45753741e-01 2.73656487e-01 -1.62805885e-01 -4.17867079e-02
9.95922461e-02 -2.22856671e-01 5.14069796e-01 3.77389789e-01
-1.40185940e+00 8.54564071e-01 6.28128815e+00 4.72466588e-01
-1.08623409e+00 7.16693923e-02 3.23022813e-01 -3.10373932e-01
-2.01245874e-01 2.00825065e-01 -1.48443246e+00 6.02066696e-01
1.53652525e+00 -4.36438918e-01 2.39595056e-01 6.15785956e-01
3.80896002e-01 -2.57215127e-02 -1.62492514e+00 3.54979724e-01
-3.71712744e-02 -1.52795827e+00 -7.59858415e-02 -6.49117082e-02
6.09313965e-01 1.74866021e-01 -4.91109461e-01 2.55527020e-01
5.90889096e-01 -4.89411235e-01 7.37864196e-01 4.93005663e-01
4.40860301e-01 -6.29216015e-01 4.10930008e-01 3.06406349e-01
-1.00742042e+00 1.08016945e-01 3.83808799e-02 -1.11323588e-01
3.61729264e-01 9.14471865e-01 -1.27440643e+00 4.68463629e-01
6.42580867e-01 9.85778987e-01 -4.31360543e-01 1.01917005e+00
-4.40529913e-01 9.50421751e-01 -8.38178754e-01 -1.82824120e-01
8.63156468e-02 3.07940990e-01 4.61169422e-01 1.39398646e+00
3.07428151e-01 3.13698165e-02 1.34641826e-01 7.17814267e-01
-4.24017608e-01 4.56406921e-02 -5.85819244e-01 -2.41178915e-01
7.56156504e-01 1.28712761e+00 -8.03609908e-01 -4.28738445e-01
-6.59462750e-01 5.00636399e-01 6.43150926e-01 2.65830696e-01
-9.52984571e-01 -5.98684251e-01 1.67469800e-01 1.29019409e-01
1.58758998e-01 -4.38632995e-01 -4.86274920e-02 -1.21944952e+00
4.49142084e-02 -4.35743690e-01 9.90570426e-01 -7.77051389e-01
-1.13569641e+00 3.87216687e-01 4.91731465e-01 -1.04421949e+00
-5.83754122e-01 -1.52138725e-01 -4.86594588e-01 5.94155431e-01
-1.32661700e+00 -1.22326815e+00 -2.33420003e-02 2.05907434e-01
3.36125344e-01 6.75783381e-02 5.39526939e-01 4.86774653e-01
-5.59697807e-01 3.02761346e-01 9.84573215e-02 5.03817320e-01
1.07877207e+00 -1.20409131e+00 7.49262989e-01 9.97247994e-01
6.55121505e-01 8.78499627e-01 8.42325807e-01 -8.59795988e-01
-1.16484821e+00 -1.19300687e+00 1.64532542e+00 -8.97182047e-01
1.12814760e+00 -4.82893646e-01 -9.75610673e-01 1.24600351e+00
3.01152647e-01 -3.03547204e-01 8.19379628e-01 7.94336438e-01
-5.94785154e-01 -2.35179707e-01 -4.29614305e-01 5.02313733e-01
7.20727205e-01 -6.03711903e-01 -1.03979301e+00 6.02687716e-01
1.09356821e+00 -5.41100919e-01 -1.12871587e+00 -1.01669900e-01
5.14509678e-01 6.77393004e-02 9.29681718e-01 -7.87867606e-01
7.27685213e-01 -1.98599994e-01 1.82134077e-01 -1.27324402e+00
-2.65025288e-01 -6.00232184e-01 -7.07518101e-01 1.84831905e+00
7.27344751e-01 -2.99721152e-01 3.69765520e-01 7.11629748e-01
4.78148907e-02 -2.55974859e-01 -6.76557362e-01 -7.18245506e-01
-1.76603168e-01 -6.41562939e-01 6.98675036e-01 1.22215021e+00
4.58536148e-01 7.12239623e-01 -3.25073630e-01 2.89529771e-01
2.99628735e-01 4.08242255e-01 4.12698478e-01 -1.63786042e+00
-1.83666065e-01 -1.79478481e-01 2.73009211e-01 -8.98788333e-01
3.26499999e-01 -1.05698085e+00 -2.80330214e-03 -1.65354538e+00
1.54356360e-01 -5.90448201e-01 -4.18856561e-01 7.78575659e-01
-1.52274266e-01 -2.29410201e-01 -2.12641075e-01 3.25917423e-01
-7.66165257e-01 1.56741068e-01 7.02102184e-01 -2.66589403e-01
-4.11219209e-01 -3.15533835e-04 -5.38586855e-01 6.04609787e-01
5.13518512e-01 -9.44657147e-01 -3.51759046e-01 -5.50152421e-01
8.36215734e-01 2.79928744e-01 6.21307418e-02 -7.44712532e-01
5.58592260e-01 -3.07377756e-01 3.50523710e-01 -9.96226013e-01
6.08484223e-02 -8.01772594e-01 4.96039428e-02 -2.04540845e-02
-9.18926239e-01 1.19606078e-01 2.13922769e-01 7.76196480e-01
-3.13956320e-01 -1.45177603e-01 1.40903428e-01 8.02429095e-02
-6.64841712e-01 2.23065332e-01 -2.27330521e-01 2.99999595e-01
7.13178694e-01 4.45116490e-01 -5.56715071e-01 -2.97297865e-01
-5.19028246e-01 3.82157445e-01 4.28843200e-02 7.84338057e-01
-4.45803255e-02 -1.38256419e+00 -4.90705907e-01 -2.89024621e-01
3.27170908e-01 1.98407359e-02 -4.16165024e-01 7.53724635e-01
-8.05540290e-03 5.89592218e-01 9.44735780e-02 -2.68357068e-01
-1.07586217e+00 8.78104568e-01 -3.44405144e-01 -5.53071499e-01
-8.49292338e-01 5.46863258e-01 -3.85255307e-01 -4.05021198e-02
3.81891161e-01 -8.04911315e-01 -6.20530367e-01 6.88332975e-01
7.73895741e-01 1.44618824e-01 1.48004532e-01 -1.82305098e-01
-1.45235777e-01 6.86295033e-02 -5.14206231e-01 -2.95964956e-01
1.79414487e+00 -2.36720130e-01 -2.43350774e-01 7.54467249e-01
1.17186499e+00 1.45174831e-01 -1.12481499e+00 -5.48399866e-01
7.39784122e-01 -4.39269878e-02 3.00764531e-01 -7.04580069e-01
-6.08469367e-01 2.73546576e-01 -1.75393254e-01 5.49942195e-01
8.36512864e-01 2.38150105e-01 1.07050014e+00 6.72324419e-01
1.23519219e-01 -1.29047120e+00 -2.46558025e-01 8.08743894e-01
5.67152917e-01 -1.14884150e+00 1.97887495e-01 -2.81001002e-01
-1.85438290e-01 1.25224364e+00 4.01695251e-01 4.59606379e-01
7.51040220e-01 5.43158770e-01 -1.17219798e-01 -3.15903395e-01
-1.20834613e+00 7.48634618e-03 4.92710888e-01 9.63492170e-02
6.02732182e-01 -2.07125664e-01 -4.96936768e-01 7.39580035e-01
-2.60150790e-01 1.74172014e-01 4.80716676e-01 8.94818664e-01
-2.85315961e-01 -1.04688835e+00 -1.17686898e-01 5.67019522e-01
-7.50667453e-01 -1.69821337e-01 -7.24730343e-02 6.27074838e-01
-1.35574624e-01 7.21522570e-01 4.00168002e-01 -7.76379704e-02
1.85057744e-01 2.63929516e-01 2.47771312e-02 -4.87223148e-01
-5.23817778e-01 7.99498409e-02 6.97930038e-01 -3.26468587e-01
-4.07163233e-01 -8.42658103e-01 -1.29883361e+00 -7.96809718e-02
-2.80535251e-01 3.84330720e-01 6.58395946e-01 1.08151495e+00
3.95813882e-01 6.33999407e-01 4.33361322e-01 -3.78837794e-01
-1.21270396e-01 -9.29922462e-01 -2.42542848e-01 5.58497429e-01
2.97027141e-01 -5.55288017e-01 -3.60483117e-02 5.63464582e-01] | [9.124344825744629, 9.242755889892578] |
a13c57c6-e7c2-4d64-9e40-d187f938c4b5 | quotes-query-oriented-technical-summarization | 2306.11832 | null | https://arxiv.org/abs/2306.11832v1 | https://arxiv.org/pdf/2306.11832v1.pdf | QuOTeS: Query-Oriented Technical Summarization | Abstract. When writing an academic paper, researchers often spend considerable time reviewing and summarizing papers to extract relevant citations and data to compose the Introduction and Related Work sections. To address this problem, we propose QuOTeS, an interactive system designed to retrieve sentences related to a summary of the research from a collection of potential references and hence assist in the composition of new papers. QuOTeS integrates techniques from Query-Focused Extractive Summarization and High-Recall Information Retrieval to provide Interactive Query-Focused Summarization of scientific documents. To measure the performance of our system, we carried out a comprehensive user study where participants uploaded papers related to their research and evaluated the system in terms of its usability and the quality of the summaries it produces. The results show that QuOTeS provides a positive user experience and consistently provides query-focused summaries that are relevant, concise, and complete. We share the code of our system and the novel Query-Focused Summarization dataset collected during our experiments at https://github.com/jarobyte91/quotes. | ['Evangelos Milios', 'Flavia P. Zanoto', 'Axel J. Soto', 'Ana Maguitman', 'Eduardo Xamena', 'Juan Ramirez-Orta'] | 2023-06-20 | null | null | null | null | ['retrieval', 'information-retrieval', 'extractive-summarization'] | ['methodology', 'natural-language-processing', 'natural-language-processing'] | [ 9.45996195e-02 -9.89191309e-02 -2.39859745e-01 -1.18308989e-02
-1.50738013e+00 -1.00465035e+00 6.10570252e-01 8.32347453e-01
-2.39101291e-01 8.31098378e-01 6.69224322e-01 -4.20093745e-01
-2.91231334e-01 -4.05096382e-01 -3.73545021e-01 -8.85246024e-02
5.09173572e-01 1.40610993e-01 1.26921013e-01 1.42260656e-01
1.28415835e+00 4.45084751e-01 -1.31399548e+00 3.07555020e-01
1.36465585e+00 2.25784361e-01 6.72985792e-01 1.10486913e+00
-4.33179975e-01 2.78724611e-01 -1.39359963e+00 -3.22257310e-01
-3.12122285e-01 -5.12286067e-01 -9.92859244e-01 -2.10707933e-01
6.73047543e-01 -1.65134389e-02 -4.24491242e-02 8.90123546e-01
7.40775764e-01 7.44857341e-02 5.30850291e-01 -1.01910174e+00
-7.58587718e-01 7.60239601e-01 -6.06568217e-01 7.08608508e-01
9.30733562e-01 -1.34192899e-01 9.80420768e-01 -9.58837807e-01
9.75488245e-01 1.13955820e+00 1.11852705e-01 2.72411376e-01
-1.10937846e+00 -5.12476623e-01 -5.16359843e-02 3.50925960e-02
-1.12865698e+00 -8.60662758e-01 6.09460771e-01 -3.86490643e-01
8.75951648e-01 8.75240922e-01 4.81588781e-01 1.05529487e+00
3.42340708e-01 7.66998768e-01 4.67062443e-01 -5.02513289e-01
3.69361937e-01 2.35880882e-01 7.23541200e-01 2.01466486e-01
7.25644886e-01 -7.79758632e-01 -6.55484080e-01 -4.14543331e-01
1.42410472e-01 -1.86728358e-01 -4.64570671e-01 4.64057356e-01
-1.33038676e+00 3.79440844e-01 -4.73965406e-02 3.23782027e-01
-5.77583134e-01 -2.64677018e-01 5.73193312e-01 1.54309630e-01
2.79432118e-01 1.10730910e+00 -7.01418743e-02 -3.75543296e-01
-1.30278647e+00 6.39313161e-01 1.15019643e+00 1.25080085e+00
2.10743725e-01 -5.21040022e-01 -1.00595820e+00 8.43363583e-01
-1.71969533e-01 5.28166711e-01 4.10721868e-01 -1.25704253e+00
5.49287438e-01 8.21537077e-01 2.01434121e-01 -9.91632640e-01
-1.53059587e-01 -4.77662563e-01 -5.04568696e-01 -4.29408044e-01
-2.20799744e-01 -2.51091063e-01 -3.29250813e-01 1.17997158e+00
-1.04011685e-01 -5.89679182e-01 1.41878992e-01 5.59864879e-01
1.73292422e+00 1.02078080e+00 5.07642478e-02 -6.95653200e-01
1.49761105e+00 -8.66301477e-01 -1.14073777e+00 5.20106219e-02
5.00905931e-01 -1.20242500e+00 1.08663285e+00 3.15123856e-01
-1.39906073e+00 -4.10044074e-01 -1.06915438e+00 -3.08915645e-01
-3.05773884e-01 4.18121338e-01 -7.93168768e-02 1.38445243e-01
-9.81222093e-01 5.98344624e-01 -5.14882028e-01 -5.22877038e-01
3.92013550e-01 -2.58263737e-01 -1.29644811e-01 5.87468706e-02
-6.81653261e-01 7.41870701e-01 1.16079137e-01 -4.74320918e-01
-1.78705707e-01 -1.10342133e+00 -3.56633633e-01 2.17825338e-01
4.84551758e-01 -1.08667910e+00 1.56285393e+00 1.26191452e-01
-1.09284616e+00 5.02718985e-01 -6.04112089e-01 2.13532038e-02
1.36478931e-01 -4.95910078e-01 -1.50247961e-01 4.68571991e-01
5.03079772e-01 3.05073231e-01 4.55476381e-02 -9.87880528e-01
-5.03210247e-01 -4.04743671e-01 -2.44226038e-01 2.73064166e-01
-3.57911587e-01 3.14672232e-01 -7.54843533e-01 -7.70422399e-01
-2.95989960e-01 -5.64276993e-01 -8.12888145e-02 -4.98655438e-01
-9.18271542e-01 -6.32953644e-01 7.48647749e-01 -8.68804932e-01
1.90455759e+00 -1.72553563e+00 2.38111198e-01 -1.26502112e-01
3.22136879e-01 3.12385201e-01 -2.08321765e-01 1.12128270e+00
2.77895153e-01 7.96588421e-01 -4.55189683e-02 -2.21993819e-01
-5.64315952e-02 -3.84857148e-01 -3.85213643e-01 -1.79918781e-01
-1.24827035e-01 9.77792382e-01 -1.04116631e+00 -7.06212997e-01
-1.39718130e-01 2.22960204e-01 1.19559863e-03 3.82285893e-01
-9.89079103e-02 1.59704253e-01 -8.69646788e-01 6.12531185e-01
3.59972954e-01 -2.81990975e-01 -2.18566298e-01 -5.12584634e-02
-4.68652010e-01 5.40525973e-01 -8.63763511e-01 1.68377566e+00
-3.36540073e-01 9.26196218e-01 -1.74553916e-01 -5.86593688e-01
7.88043976e-01 2.63168931e-01 1.25307083e-01 -8.00652027e-01
-7.62000009e-02 2.81680644e-01 -6.38750315e-01 -7.85582185e-01
9.10805166e-01 6.35670722e-01 -1.31961688e-01 7.27064848e-01
-2.95190573e-01 -5.62756002e-01 1.02384210e+00 1.09176898e+00
1.28259850e+00 -1.66853100e-01 4.39787447e-01 -3.03161144e-01
4.53600705e-01 1.77869219e-02 -2.03769356e-02 1.12597597e+00
3.37719172e-01 6.63811922e-01 6.33068383e-01 1.17955022e-01
-9.23356593e-01 -7.22554684e-01 9.38211083e-02 7.65303493e-01
-5.57038523e-02 -1.13650012e+00 -7.14789033e-01 -2.97497988e-01
-1.78830400e-01 1.13931251e+00 -2.75175333e-01 -9.73188803e-02
-1.92335874e-01 -2.23424807e-01 2.21119180e-01 2.08258167e-01
2.42662668e-01 -1.26355398e+00 -8.94796431e-01 -1.09094521e-02
-5.14441431e-01 -7.48270333e-01 -7.91073203e-01 -3.11319917e-01
-8.18839788e-01 -9.68106687e-01 -1.09060478e+00 -4.91024762e-01
5.20058393e-01 4.60605741e-01 1.27540982e+00 1.17407016e-01
-3.66073728e-01 3.94285291e-01 -4.25952822e-01 -7.07542121e-01
-5.00813782e-01 6.28217399e-01 -4.11014259e-01 -9.74091589e-01
1.30548999e-01 -4.31004047e-01 -5.84997594e-01 -2.38886550e-01
-1.02909648e+00 5.38081378e-02 6.65250659e-01 4.32497829e-01
3.42404157e-01 -3.38523090e-01 9.14957941e-01 -8.55522811e-01
1.71706843e+00 -3.54455471e-01 -4.80350018e-01 6.44706309e-01
-7.26267695e-01 9.37915295e-02 3.91468287e-01 -8.12066048e-02
-8.71419966e-01 -4.21012133e-01 -5.19894361e-02 1.97083548e-01
-1.58123374e-02 9.24464107e-01 8.16357359e-02 3.56379986e-01
7.44133651e-01 2.29358613e-01 -1.53219700e-01 -6.84294224e-01
2.84401178e-01 1.13705528e+00 5.46448886e-01 -4.95492220e-01
4.32889342e-01 -3.54090989e-01 -2.45409846e-01 -9.84756172e-01
-8.63103747e-01 -7.31716454e-01 -3.21720690e-01 -2.70749807e-01
2.78113812e-01 -5.44775963e-01 -7.70561039e-01 -2.69003540e-01
-1.48497498e+00 2.12161824e-01 -2.93190688e-01 3.22838932e-01
-1.03722990e-01 5.55411339e-01 -3.65870446e-01 -6.75367773e-01
-1.11523819e+00 -1.04725468e+00 1.21318257e+00 7.95834124e-01
-8.91654134e-01 -5.35477459e-01 2.43121684e-01 3.17337543e-01
2.79094756e-01 1.28473908e-01 7.39270568e-01 -6.86806560e-01
-2.05399990e-01 -4.24145341e-01 -3.11434567e-01 -1.37026355e-01
2.42247239e-01 5.00568926e-01 -5.65225542e-01 -2.12640911e-01
-4.58893329e-01 -2.80754957e-02 9.39701200e-01 4.52632159e-01
1.25548458e+00 -7.38671184e-01 -5.69812953e-01 -1.81254204e-02
1.11248708e+00 2.84903198e-01 4.48132575e-01 4.55551982e-01
3.18993866e-01 5.02779543e-01 5.56251705e-01 5.09097338e-01
2.34389797e-01 5.46675146e-01 -2.76769191e-01 3.43378067e-01
-1.24164864e-01 -2.24976782e-02 1.81640878e-01 1.16968656e+00
2.06420898e-01 -6.91507339e-01 -7.56159544e-01 7.90444314e-01
-1.69964480e+00 -9.00851130e-01 -1.75790995e-01 2.00686669e+00
1.07027471e+00 4.03827615e-02 1.69493660e-01 -7.22646862e-02
5.04528880e-01 5.68083189e-02 -3.32146019e-01 -7.52440393e-01
2.01259777e-02 1.04133144e-01 7.42466822e-02 5.26132882e-01
-5.19999743e-01 6.10379696e-01 6.25591898e+00 7.56473184e-01
-9.13450003e-01 -4.06982929e-01 3.29767793e-01 -2.98300654e-01
-5.86071134e-01 -1.68964028e-01 -7.33481884e-01 4.80017304e-01
1.25257695e+00 -1.35835385e+00 -1.59543857e-01 5.78532398e-01
5.82511842e-01 -5.69643378e-01 -1.05369937e+00 7.18047023e-01
-2.90443655e-02 -1.73861361e+00 4.99274731e-01 -1.39482886e-01
5.85316002e-01 -4.19189602e-01 -6.30254224e-02 8.36649612e-02
2.04511285e-02 -7.27800548e-01 5.05105674e-01 7.20781803e-01
6.00627840e-01 -7.06792831e-01 7.46736646e-01 3.72487873e-01
-5.81529200e-01 2.70793229e-01 -2.85690129e-01 -5.47001436e-02
8.53063539e-02 7.81549156e-01 -9.40152168e-01 9.46396947e-01
5.81179023e-01 7.56767750e-01 -9.31245446e-01 1.33250320e+00
-1.79863125e-01 5.43925703e-01 -4.38902043e-02 -5.57045460e-01
-1.34660080e-01 -1.06568478e-01 9.73737359e-01 1.68705618e+00
5.04907727e-01 3.73390526e-01 -6.08987883e-02 1.02301419e+00
-4.50101167e-01 4.58240807e-01 -4.57752466e-01 -5.96234798e-01
9.61242557e-01 1.38708663e+00 -8.23127329e-01 -6.67060673e-01
1.24585927e-01 6.48067951e-01 -4.63864487e-03 3.61773252e-01
-2.76220381e-01 -1.10587049e+00 -1.85332168e-02 -1.15496777e-01
1.16949037e-01 -5.56191243e-02 -6.29524052e-01 -9.77150619e-01
5.00831127e-01 -7.89488733e-01 2.79863000e-01 -1.12369394e+00
-8.71675432e-01 6.15324020e-01 8.61314759e-02 -8.76731753e-01
-8.33742172e-02 1.72664985e-01 -8.73834312e-01 1.05808163e+00
-1.02782369e+00 -3.80236924e-01 -4.22758073e-01 -3.28644097e-01
9.43377256e-01 -1.57257356e-02 7.66690254e-01 6.89206123e-02
-8.75401616e-01 2.29841292e-01 1.37939185e-01 -2.76513606e-01
9.45466399e-01 -1.23478353e+00 4.79771525e-01 9.42388713e-01
-1.38186784e-02 1.18068302e+00 1.00356197e+00 -8.66781533e-01
-1.49739325e+00 -7.39394069e-01 1.29355478e+00 -5.34726739e-01
3.73013139e-01 1.53517678e-01 -9.97776330e-01 1.21674910e-01
7.85042286e-01 -7.63659418e-01 7.34776914e-01 5.07574948e-03
9.31179449e-02 1.79631654e-02 -6.81936920e-01 7.90797830e-01
7.49901295e-01 -2.14000463e-01 -1.07851088e+00 3.13663870e-01
8.23255658e-01 -3.19621265e-01 -8.62293124e-01 9.67364088e-02
5.65009117e-01 -5.14826894e-01 7.81315625e-01 -3.65534514e-01
9.22500968e-01 -1.47905871e-01 3.67216080e-01 -1.41440022e+00
-3.08647037e-01 -6.96376085e-01 -5.04169166e-02 1.49120939e+00
6.68022990e-01 -1.24010347e-01 3.58603865e-01 6.59645498e-01
-2.77212828e-01 -7.43476391e-01 -3.76236171e-01 -4.44812536e-01
-2.18827296e-02 7.48398080e-02 3.67401272e-01 4.89288986e-01
5.39447844e-01 7.01780736e-01 2.30273500e-01 -1.99747294e-01
4.93022859e-01 4.59017575e-01 8.02940786e-01 -1.31262910e+00
4.32808489e-01 -9.01461184e-01 2.61306524e-01 -8.69658589e-01
-1.11214273e-01 -7.89262235e-01 -1.46404788e-01 -2.49102449e+00
7.48772621e-01 2.22093359e-01 8.41935650e-02 3.26793998e-01
-5.24415612e-01 -1.05873846e-01 1.74770460e-01 4.41260040e-01
-1.05692828e+00 3.29222977e-01 1.27085936e+00 -1.70394987e-01
-5.33751369e-01 5.89666851e-02 -1.51314378e+00 6.46499768e-02
6.15454435e-01 -3.53262812e-01 -2.52581924e-01 -9.97432843e-02
3.41973245e-01 3.69367540e-01 7.71961287e-02 -7.10056067e-01
6.14410520e-01 -1.74502954e-01 3.09497923e-01 -1.03289342e+00
-2.18421608e-01 -6.20660931e-02 4.13643681e-02 2.91506261e-01
-7.75230587e-01 4.09408420e-01 5.58121979e-01 1.22081004e-01
-7.89040253e-02 -3.76663148e-01 2.12638929e-01 -1.05979845e-01
-1.60550237e-01 -1.65726170e-01 -5.43235719e-01 2.71779865e-01
5.24220347e-01 1.26180172e-01 -6.89620197e-01 -5.58533072e-01
-2.00577781e-01 5.11424303e-01 4.40448701e-01 6.52121305e-01
7.79040694e-01 -8.77348661e-01 -1.06476378e+00 -3.80399436e-01
1.85051292e-01 6.33713789e-03 -8.10871366e-03 6.07345879e-01
-6.22995853e-01 8.95421326e-01 -1.50612757e-01 -2.50096947e-01
-1.47468829e+00 3.63889486e-01 -4.54759985e-01 -1.49795920e-01
-7.38720655e-01 5.28479159e-01 -2.07441702e-01 4.13498236e-03
4.59805191e-01 -4.19073343e-01 -5.90696156e-01 4.15001661e-01
1.05021119e+00 7.63185322e-01 4.14182633e-01 -8.73336792e-02
-3.54972452e-01 2.94868678e-01 -2.57183641e-01 -2.60789275e-01
1.38825953e+00 -3.61473978e-01 -3.10885519e-01 5.94335258e-01
1.11542022e+00 5.08025885e-01 -3.40040743e-01 -1.05941720e-01
3.70705336e-01 -1.47600681e-01 9.98014361e-02 -1.10899627e+00
-5.82378268e-01 4.51081812e-01 -1.34758174e-01 4.55434948e-01
9.92830753e-01 1.31600320e-01 6.68279171e-01 6.86890781e-01
-3.34678829e-01 -9.54451859e-01 -3.71348746e-02 2.91699737e-01
1.34422314e+00 -1.04728377e+00 6.99686527e-01 -2.39433303e-01
-4.69897985e-01 1.30439460e+00 3.15510213e-01 3.24433088e-01
1.95423871e-01 9.71185565e-02 3.05572748e-02 -4.55111444e-01
-1.01913905e+00 3.59751195e-01 6.68338895e-01 9.43656713e-02
7.57533550e-01 -1.41371444e-01 -8.95952165e-01 6.75186813e-01
-4.69430596e-01 1.70673460e-01 9.56090748e-01 1.05353701e+00
-4.56261277e-01 -9.19133484e-01 -3.16758782e-01 6.18565381e-01
-6.35315001e-01 -3.04717477e-02 -1.01645482e+00 3.62354398e-01
-9.14145887e-01 1.23089910e+00 -2.31528327e-01 7.03075752e-02
6.81110680e-01 -4.70481291e-02 2.70130754e-01 -7.84143448e-01
-6.53917253e-01 7.97001272e-02 7.88970888e-02 -2.68930882e-01
-2.68042505e-01 -7.08893597e-01 -1.12864029e+00 -4.09067422e-01
-1.33627072e-01 9.32252645e-01 8.61702979e-01 5.79565287e-01
1.00434923e+00 9.72049713e-01 4.31703836e-01 -7.04696715e-01
-2.82662004e-01 -1.24237144e+00 -2.20187139e-02 1.02548428e-01
4.08050954e-01 -1.16189465e-01 -1.88577756e-01 2.45272383e-01] | [12.375143051147461, 9.547451972961426] |
d4520950-4120-44c1-b3d5-216a77ace39c | automatic-truss-design-with-reinforcement | 2306.15182 | null | https://arxiv.org/abs/2306.15182v1 | https://arxiv.org/pdf/2306.15182v1.pdf | Automatic Truss Design with Reinforcement Learning | Truss layout design, namely finding a lightweight truss layout satisfying all the physical constraints, is a fundamental problem in the building industry. Generating the optimal layout is a challenging combinatorial optimization problem, which can be extremely expensive to solve by exhaustive search. Directly applying end-to-end reinforcement learning (RL) methods to truss layout design is infeasible either, since only a tiny portion of the entire layout space is valid under the physical constraints, leading to particularly sparse rewards for RL training. In this paper, we develop AutoTruss, a two-stage framework to efficiently generate both lightweight and valid truss layouts. AutoTruss first adopts Monte Carlo tree search to discover a diverse collection of valid layouts. Then RL is applied to iteratively refine the valid solutions. We conduct experiments and ablation studies in popular truss layout design test cases in both 2D and 3D settings. AutoTruss outperforms the best-reported layouts by 25.1% in the most challenging 3D test cases, resulting in the first effective deep-RL-based approach in the truss layout design literature. | ['Yi Wu', 'Xianzhong Zhao', 'Zhiyuan Liu', 'Ruifeng Luo', 'Siyang Wu', 'Zimeng Song', 'Xingcheng Yao', 'Chao Yu', 'Jinglun Zhao', 'Weihua Du'] | 2023-06-27 | null | null | null | null | ['layout-design', 'combinatorial-optimization'] | ['computer-vision', 'methodology'] | [-1.51258007e-01 -6.64519519e-02 1.07056379e-01 1.40953407e-01
-1.05200815e+00 -8.56103063e-01 -1.57732248e-01 -1.65208466e-02
1.77405939e-01 8.82059455e-01 -9.09594595e-02 -5.78995347e-01
-8.04456711e-01 -8.15394104e-01 -8.96258771e-01 -6.10345364e-01
-3.62793684e-01 6.17468476e-01 -3.99900287e-01 -2.75983661e-01
3.25461626e-01 6.97052181e-01 -1.50007343e+00 -2.39936501e-01
9.91098285e-01 7.51610994e-01 4.81870085e-01 6.14859521e-01
2.43776441e-01 4.61693019e-01 -8.36195886e-01 8.67157578e-02
3.55492055e-01 -1.14088185e-01 -7.74319470e-01 2.19789505e-01
5.65268934e-01 -4.69609231e-01 -1.45261616e-01 3.41460019e-01
8.64046574e-01 2.94878215e-01 6.24706864e-01 -8.44826937e-01
-5.70555568e-01 8.68962586e-01 -9.50613141e-01 -4.04734284e-01
1.30537376e-01 2.97442704e-01 1.43536127e+00 -6.54346526e-01
3.44737083e-01 1.07576585e+00 5.64814210e-01 4.01101410e-01
-1.40294266e+00 -4.76240814e-01 6.01471178e-02 -2.26824671e-01
-1.44153988e+00 -4.61998489e-03 9.77583170e-01 -1.44873366e-01
8.75439048e-01 2.48449817e-01 8.41103613e-01 8.22349370e-01
2.48875305e-01 9.24241483e-01 8.61557841e-01 -9.43077728e-02
5.36523640e-01 -7.83354938e-01 -4.28115487e-01 8.91529679e-01
2.92932808e-01 4.27754968e-01 -1.65474460e-01 4.18285541e-02
8.02272201e-01 -3.29349816e-01 1.46461800e-01 -7.72248268e-01
-7.73850024e-01 8.01850080e-01 8.91828835e-01 -1.67864785e-01
-4.19374071e-02 7.34300613e-01 2.37714812e-01 1.26210079e-01
-1.14377551e-01 1.33168113e+00 -3.02684784e-01 -8.39705914e-02
-1.11287522e+00 9.13944960e-01 5.86643994e-01 9.98145878e-01
5.29833794e-01 4.58889455e-01 -9.97273922e-02 1.17462218e+00
1.45395532e-01 6.26514554e-01 -3.72748852e-01 -1.00557244e+00
5.49832821e-01 3.32638711e-01 2.28089586e-01 -7.43299127e-01
-4.70128447e-01 -8.61541033e-01 -6.12201452e-01 2.94953257e-01
1.63062871e-01 -6.57258689e-01 -7.95674920e-01 1.30526626e+00
3.81278634e-01 -1.28705233e-01 -1.30725384e-01 1.06319129e+00
7.89375246e-01 8.05793583e-01 -4.64700788e-01 2.88017899e-01
7.19768226e-01 -1.00850880e+00 -2.07980573e-02 -4.49564725e-01
7.85542786e-01 -7.38529921e-01 1.01477575e+00 5.87815106e-01
-1.22239852e+00 -3.75780880e-01 -1.45204675e+00 2.79934824e-01
1.96771715e-02 5.24458945e-01 7.49975801e-01 6.51949346e-01
-8.12355816e-01 7.72091269e-01 -2.86475450e-01 2.03403994e-01
5.26118398e-01 5.18718541e-01 9.05004069e-02 -3.66390228e-01
-6.59878790e-01 7.12715864e-01 2.38918334e-01 4.57066208e-01
-1.48095989e+00 -8.82693291e-01 -8.54312956e-01 1.22413173e-01
1.06769419e+00 -5.63525021e-01 1.36327350e+00 2.54301459e-01
-1.57050276e+00 3.03647578e-01 4.55897421e-01 -2.75197804e-01
3.69766861e-01 -3.98078173e-01 2.84140438e-01 -2.80530989e-01
-1.33128762e-01 7.03243077e-01 9.41901684e-01 -1.61671162e+00
-3.57102484e-01 3.29058468e-02 2.90966988e-01 1.81241691e-01
2.16817111e-01 -8.90836477e-01 -5.63042946e-02 -3.23298424e-01
-9.56873223e-02 -1.06384397e+00 -7.92754948e-01 -4.67661887e-01
-7.63322592e-01 -2.59566545e-01 7.50241518e-01 -3.92627120e-01
1.18104458e+00 -1.71017754e+00 4.15147245e-01 5.37458897e-01
2.45259088e-02 -1.02788463e-01 -3.69417906e-01 8.26362610e-01
1.29987970e-01 -1.64279826e-02 -1.95185736e-01 -1.68692976e-01
3.78750980e-01 4.83595014e-01 -4.81327772e-01 1.49801135e-01
2.15213820e-01 1.11098444e+00 -1.05779958e+00 -1.97438672e-01
5.54206073e-01 -2.36776054e-01 -8.64581287e-01 2.45774999e-01
-4.32806760e-01 1.62656188e-01 -5.34790456e-01 7.45935678e-01
6.58909202e-01 9.49588344e-02 1.92139387e-01 -1.33932725e-01
-2.23252058e-01 -2.34552603e-02 -1.17858493e+00 2.17964935e+00
-1.07007682e+00 2.24395290e-01 6.44223951e-03 -1.01057446e+00
1.44065392e+00 -3.55414927e-01 5.30316710e-01 -5.07127941e-01
1.74656838e-01 3.48613739e-01 -1.58094149e-02 -2.43299454e-01
8.05869043e-01 1.39766112e-01 -8.19181859e-01 3.15862268e-01
-1.31384671e-01 -1.16183579e+00 5.64420283e-01 6.66825399e-02
1.42101121e+00 4.23635066e-01 -3.13687444e-01 -3.21087122e-01
1.00605443e-01 -1.14428423e-01 3.86943609e-01 9.94318306e-01
5.36606789e-01 6.12335980e-01 6.57734573e-01 -4.77631211e-01
-1.28437054e+00 -1.52108908e+00 2.79287636e-01 4.14725989e-01
1.28843859e-01 -3.41742426e-01 -6.13540769e-01 -7.64925718e-01
2.90272743e-01 8.78122270e-01 -3.95972013e-01 -3.23712677e-01
-6.37666464e-01 -2.92395860e-01 4.04951900e-01 3.94668639e-01
2.46296689e-01 -9.58582759e-01 -1.20854652e+00 3.78165066e-01
2.90143549e-01 -7.96401680e-01 -4.40847129e-01 4.25200582e-01
-7.82895267e-01 -1.02645373e+00 -5.61617017e-01 -6.06157899e-01
7.64099658e-01 2.00822040e-01 1.16403699e+00 3.94795120e-01
-8.25549483e-01 1.73942029e-01 -3.42322469e-01 -1.49428800e-01
-2.46450201e-01 5.76401114e-01 -3.83605242e-01 -7.67672539e-01
-6.68254077e-01 -5.50321162e-01 -6.39830351e-01 4.52543646e-01
-5.20504415e-01 3.44399810e-02 8.86026323e-01 1.01442254e+00
5.61812043e-01 6.48809493e-01 7.12597966e-01 -4.98745054e-01
7.67876387e-01 -1.56891078e-01 -1.08250713e+00 1.84892178e-01
1.07883764e-02 4.68281478e-01 8.31925929e-01 -9.98059437e-02
-7.09481061e-01 1.92615405e-01 -3.72048110e-01 -5.22659898e-01
-1.01492412e-01 8.01143825e-01 -3.69498879e-02 4.91534993e-02
4.65460330e-01 -1.48506179e-01 -3.16025198e-01 -4.57496166e-01
4.28193241e-01 9.12817195e-02 3.62864166e-01 -1.40808129e+00
1.17878568e+00 -3.24587822e-02 5.10449469e-01 -8.45553339e-01
-1.09827530e+00 -3.52711789e-02 -4.72323984e-01 -5.91562986e-01
1.94708288e-01 -2.11874947e-01 -9.61813450e-01 1.34063410e-02
-7.34086037e-01 -9.04357076e-01 -5.54423988e-01 -3.47852930e-02
-7.70582855e-01 2.57309794e-01 -2.13166714e-01 -9.20410991e-01
-4.32769507e-01 -1.43386626e+00 1.37291110e+00 1.15690357e-03
-2.56199151e-01 -4.08921450e-01 7.12805539e-02 2.73769110e-01
7.26039112e-02 6.11161232e-01 1.22655022e+00 2.86964804e-01
-7.81341553e-01 -1.90596327e-01 -2.31859572e-02 1.72623098e-01
1.41164348e-01 2.20493868e-01 -3.94547939e-01 -5.57811081e-01
-7.22046018e-01 -7.10106075e-01 5.77368379e-01 6.39262319e-01
1.45564067e+00 -4.54175025e-02 -1.46985605e-01 2.58112043e-01
1.47412848e+00 1.05375074e-01 5.76389015e-01 -1.00110769e-01
5.40030539e-01 4.70609039e-01 9.90201056e-01 7.33959436e-01
2.64110100e-02 5.42766631e-01 8.42717886e-01 -1.66905031e-01
6.84621409e-02 -6.87048674e-01 6.05021827e-02 4.58441734e-01
3.47350121e-01 -3.12182099e-01 -1.01771569e+00 6.53882504e-01
-1.90171218e+00 -7.21196294e-01 1.98971421e-01 2.19975495e+00
6.17284477e-01 3.81267726e-01 -1.34839388e-02 3.43057513e-01
3.03753287e-01 1.13692261e-01 -6.31044149e-01 -7.34761536e-01
3.16422343e-01 6.20811045e-01 6.11734092e-01 1.65177733e-01
-8.80342364e-01 8.64140272e-01 6.17459583e+00 1.07780826e+00
-8.64341259e-01 -6.81775331e-01 4.66327012e-01 -5.98360479e-01
-2.66712457e-01 1.46203101e-01 -7.01579690e-01 1.35334164e-01
5.31444788e-01 2.23605052e-01 7.98573136e-01 1.01774657e+00
4.54105437e-01 -3.27603638e-01 -1.03599095e+00 8.53382885e-01
-3.14646155e-01 -1.60982430e+00 -2.04871505e-01 2.40656570e-01
1.10265541e+00 -3.98285866e-01 2.33926699e-01 5.65530479e-01
6.81530952e-01 -1.23697722e+00 9.30248976e-01 1.59123778e-01
6.02579892e-01 -1.62581944e+00 4.36031878e-01 2.58668244e-01
-1.26875925e+00 -5.06549656e-01 -4.37642634e-01 1.11368142e-01
4.47319925e-01 8.63363802e-01 -1.07035851e+00 9.46210444e-01
7.70421088e-01 6.15758419e-01 -3.51097375e-01 1.32873011e+00
-3.70335609e-01 4.78678286e-01 -4.75371689e-01 -3.85924399e-01
8.51489484e-01 -2.89773971e-01 4.52524960e-01 5.04559636e-01
5.59586585e-01 -3.92030358e-01 4.45102483e-01 1.22528493e+00
-1.52671576e-01 -8.19907337e-02 -7.72321045e-01 1.36593282e-01
6.53802574e-01 1.35339594e+00 -6.45526350e-01 4.12494510e-01
1.41712204e-01 3.13418388e-01 4.52275395e-01 2.08509192e-01
-1.12158549e+00 -5.69757938e-01 3.01502615e-01 4.54343669e-02
7.64271319e-01 -7.60526955e-01 -3.73820573e-01 -2.80427277e-01
-1.47307768e-01 -8.64126742e-01 1.64841518e-01 -8.45029473e-01
-1.05913019e+00 7.48988837e-02 2.70038903e-01 -1.10777617e+00
-3.31043042e-02 -5.67106783e-01 -6.78151608e-01 4.05694097e-01
-1.28834033e+00 -1.17976224e+00 -1.43280461e-01 -1.97930068e-01
8.72881591e-01 -9.61318836e-02 4.15132046e-01 1.35012269e-01
-8.35478067e-01 5.48333764e-01 1.15157805e-01 -2.99010128e-01
1.51484132e-01 -1.36001253e+00 5.01644552e-01 5.31887650e-01
1.03264406e-01 2.52115816e-01 8.07135761e-01 -6.35217428e-01
-1.92833126e+00 -9.99552131e-01 -1.45428702e-01 -3.62852514e-02
5.35276532e-01 -5.06598890e-01 -4.52302516e-01 -4.16640751e-03
8.97026956e-02 -5.26955128e-01 3.00108552e-01 2.59626120e-01
1.36797413e-01 -1.93495095e-01 -9.77736235e-01 1.10983801e+00
1.26939070e+00 2.46308953e-01 -1.62568271e-01 1.20715722e-01
5.44816494e-01 -7.66686261e-01 -1.07538581e+00 5.75623453e-01
4.75029051e-01 -6.48589373e-01 1.11027443e+00 -2.38833837e-02
8.66908848e-01 -3.93486649e-01 -1.24320481e-02 -1.60696197e+00
-1.77847400e-01 -8.78303528e-01 -1.77375227e-01 1.18745208e+00
5.01627862e-01 1.55635430e-02 1.00820947e+00 2.33232200e-01
-6.56933904e-01 -1.31413412e+00 -8.27595174e-01 -8.89262438e-01
1.63956240e-01 -3.06228608e-01 1.02721953e+00 1.75197020e-01
-1.34679779e-01 3.87973398e-01 -5.15483975e-01 3.40503873e-03
7.87699103e-01 6.02189898e-01 1.10248280e+00 -7.52902865e-01
-5.24120569e-01 -3.66733700e-01 2.71490514e-01 -1.40010893e+00
2.27651715e-01 -8.07521105e-01 5.92676044e-01 -1.90520859e+00
-5.02185047e-01 -9.09607470e-01 2.25074977e-01 4.06315088e-01
3.29148382e-01 -1.87051758e-01 -1.83028472e-03 -4.66185898e-01
-3.97140652e-01 9.85616982e-01 1.63993227e+00 -5.05624354e-01
-3.27763706e-01 5.42582944e-02 -5.21082640e-01 2.21299872e-01
9.55828309e-01 -2.58602202e-01 -7.60273695e-01 -5.90651095e-01
4.79653746e-01 3.87852013e-01 3.64746988e-01 -1.09892333e+00
-2.76258826e-01 -3.44363242e-01 3.18826497e-01 -1.43878531e+00
2.84515113e-01 -8.42187405e-01 -6.44971281e-02 6.12481594e-01
-3.22146773e-01 1.34817302e-01 6.38559699e-01 4.11656678e-01
4.49443609e-01 -6.31056011e-01 7.51330256e-01 -3.17354016e-02
-5.59762537e-01 2.91946024e-01 -4.63422567e-01 1.16692008e-02
9.48052764e-01 -1.91200331e-01 -4.65166382e-02 -1.67934775e-01
-3.43300790e-01 7.42112696e-01 2.58467466e-01 4.85335380e-01
1.02438068e+00 -1.17426717e+00 -5.84438443e-01 -2.31213309e-03
-2.03204930e-01 5.74716508e-01 5.87431371e-01 3.06725442e-01
-7.63658822e-01 1.86292022e-01 -1.35527298e-01 -6.00153625e-01
-9.26812291e-01 4.88168210e-01 3.49306136e-01 -6.10828280e-01
-6.79665089e-01 6.43156469e-01 -1.83193475e-01 -6.10698640e-01
2.52300277e-02 -4.32860076e-01 3.18799555e-01 -3.34752679e-01
-1.71750590e-01 5.64666867e-01 1.19220570e-01 1.15764171e-01
-4.85084727e-02 6.86509490e-01 2.70585027e-02 8.30789935e-03
1.44628882e+00 1.78370073e-01 2.14780450e-01 -1.52728856e-01
9.44256783e-01 -3.06571007e-01 -1.28714454e+00 3.42136681e-01
-1.71100404e-02 -6.93669558e-01 2.49624655e-01 -7.73740590e-01
-1.19637072e+00 7.41016746e-01 8.26139525e-02 -3.98069061e-02
8.34054351e-01 -1.26450792e-01 1.03467822e+00 7.66052604e-01
6.46325886e-01 -1.36959505e+00 5.21612287e-01 4.62166011e-01
1.33510470e+00 -5.20679235e-01 3.99541974e-01 -1.14130035e-01
-4.99005616e-01 1.08309388e+00 9.41500545e-01 -6.78172052e-01
2.20768631e-01 4.30627555e-01 -4.98638242e-01 -3.64894837e-01
-7.75144577e-01 -1.04596309e-01 1.58254206e-01 2.95259804e-01
5.89811355e-02 -7.44350776e-02 4.05934174e-03 1.42092854e-01
-6.69263422e-01 -4.72423106e-01 3.86966944e-01 1.10440755e+00
-3.98398638e-01 -1.29300451e+00 -3.33880037e-01 7.34638095e-01
2.25412771e-01 3.52815449e-01 -1.78421944e-01 7.97476947e-01
-2.00522453e-01 9.62559879e-01 -2.46839598e-01 -3.28222603e-01
7.08792210e-01 -6.46811247e-01 9.09438968e-01 -6.46134377e-01
-6.04070246e-01 4.51036356e-02 3.71292591e-01 -2.65113980e-01
3.70659292e-01 -4.79321390e-01 -1.42660272e+00 -2.72891909e-01
-5.02660692e-01 2.33257189e-01 5.25433242e-01 7.64363587e-01
2.10314602e-01 8.77935052e-01 1.05579233e+00 -1.09912372e+00
-6.45770192e-01 -2.96286017e-01 -5.62377214e-01 -1.70986548e-01
-3.11986730e-02 -1.07111311e+00 1.47227466e-01 -6.89361572e-01] | [5.591645240783691, 3.1136295795440674] |
eebdb397-2a92-4e4d-a867-919baf5544e2 | shape-change-and-control-of-pressure-based | 2205.00467 | null | https://arxiv.org/abs/2205.00467v1 | https://arxiv.org/pdf/2205.00467v1.pdf | Shape Change and Control of Pressure-based Soft Agents | Biological agents possess bodies that are mostly of soft tissues. Researchers have resorted to soft bodies to investigate Artificial Life (ALife)-related questions; similarly, a new era of soft-bodied robots has just begun. Nevertheless, because of their infinite degrees of freedom, soft bodies pose unique challenges in terms of simulation, control, and optimization. Here we propose a novel soft-bodied agents formalism, namely Pressure-based Soft Agents (PSAs): they are bodies of gas enveloped by a chain of springs and masses, with pressure pushing on the masses from inside the body. Pressure endows the agents with structure, while springs and masses simulate softness and allow the agents to assume a large gamut of shapes. Actuation takes place by changing the length of springs or modulating global pressure. We optimize the controller of PSAs for a locomotion task on hilly terrain and an escape task from a cage; the latter is particularly suitable for soft-bodied agents, as it requires the agent to contort itself to squeeze through a small aperture. Our results suggest that PSAs are indeed effective at those tasks and that controlling pressure is fundamental for shape-changing. Looking forward, we envision PSAs to play a role in the modeling of soft-bodied agents, including soft robots and biological cells. Videos of evolved agents are available at https://pressuresoftagents.github.io. | ['Federico Pigozzi'] | 2022-05-01 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [ 1.51122939e-02 6.90316617e-01 2.46899843e-01 4.94012505e-01
5.84047556e-01 -6.15809202e-01 6.49360120e-01 -2.46594369e-01
-3.25317323e-01 1.06729221e+00 -3.70163649e-01 4.15626436e-01
3.32313329e-02 -9.50183570e-01 -9.15646434e-01 -1.26796770e+00
-3.89642358e-01 7.65519977e-01 6.57315493e-01 -9.24874604e-01
-2.94825763e-01 3.81518990e-01 -1.48278320e+00 -3.94614726e-01
9.99000788e-01 7.09603548e-01 4.50133443e-01 5.13259113e-01
6.46736383e-01 2.32799977e-01 -3.64940912e-01 -3.11363369e-01
2.37690628e-01 -1.38788879e-01 -3.76300871e-01 -2.22653016e-01
-4.35415149e-01 1.12088308e-01 -1.07725717e-01 5.30594170e-01
4.52676445e-01 5.81058003e-02 1.01597881e+00 -1.11933458e+00
-5.77216089e-01 4.44171965e-01 -1.04380667e-01 -5.79021931e-01
2.68505692e-01 4.99649167e-01 5.94875515e-01 -5.73157132e-01
8.22418272e-01 1.08862245e+00 7.26137042e-01 1.09140062e+00
-8.69398117e-01 5.10139056e-02 -2.24821627e-01 -4.59822118e-01
-1.00655389e+00 -3.98510516e-01 5.41116476e-01 -6.14237428e-01
7.62285709e-01 5.82265556e-01 1.44512582e+00 1.14401674e+00
9.93253052e-01 4.14139122e-01 9.42087471e-01 -7.52594620e-02
8.15714955e-01 -7.12751895e-02 -5.08371949e-01 7.79316366e-01
7.55923688e-01 2.28310317e-01 -4.56942677e-01 -4.17822927e-01
1.02558386e+00 -1.85694218e-01 -4.14525867e-01 -4.92448181e-01
-1.09917498e+00 4.73532677e-01 4.25256044e-01 1.78459093e-01
-3.86073589e-01 3.86540055e-01 -9.60991532e-02 -5.89609556e-02
4.21967730e-02 8.88613999e-01 -3.50471854e-01 -1.51643641e-02
-2.20365927e-01 7.11635888e-01 1.26163888e+00 5.22913098e-01
6.37821555e-02 1.95099741e-01 4.60813731e-01 6.73356056e-01
4.14412349e-01 9.28723097e-01 2.98677981e-01 -9.88222778e-01
-2.21532181e-01 7.41485178e-01 2.24468976e-01 -8.35463643e-01
-6.11944437e-01 -2.36146316e-01 -8.35717499e-01 8.04303586e-01
4.33067054e-01 -3.96621913e-01 -6.84617460e-01 1.75028658e+00
6.05210543e-01 -5.38971424e-01 1.32055104e-01 1.17466640e+00
5.65436780e-01 6.93144917e-01 -2.95372814e-01 -1.59426346e-01
1.28266692e+00 -6.03660166e-01 -4.10694599e-01 -2.64368415e-01
2.54579663e-01 -6.96103573e-02 1.10124540e+00 2.83333391e-01
-1.49602258e+00 -1.69698875e-02 -9.62266922e-01 5.05594730e-01
-3.92903447e-01 -4.70552653e-01 4.83312756e-01 1.81047976e-01
-9.87805784e-01 7.94878304e-01 -1.17865944e+00 -5.98158300e-01
1.40103444e-01 4.16940629e-01 -9.70211551e-02 8.85598540e-01
-1.17198801e+00 1.07280648e+00 -8.31360742e-03 6.23448677e-02
-7.50958800e-01 -4.24877465e-01 -5.98027527e-01 -2.38341838e-01
3.08274537e-01 -1.33736503e+00 8.80934000e-01 -5.62009096e-01
-2.37336230e+00 9.95646238e-01 6.07104301e-01 -4.07035202e-01
9.28803980e-01 -9.86580476e-02 3.41632485e-01 4.04137224e-02
-2.64110267e-01 5.33414602e-01 8.53479743e-01 -1.45184159e+00
4.51267123e-01 -1.20899148e-01 -2.00611516e-03 2.29269803e-01
-3.43291402e-01 -3.73758376e-01 3.59677434e-01 -7.11538970e-01
-6.90724030e-02 -1.46155739e+00 -4.07076001e-01 4.32493627e-01
-3.81046891e-01 -6.03068136e-02 8.73373866e-01 4.95791212e-02
5.70697844e-01 -1.82156265e+00 1.10580230e+00 1.31879926e-01
2.99561054e-01 1.97233275e-01 2.20462978e-01 8.25521767e-01
7.79696047e-01 -3.92105523e-03 -7.15655744e-01 -7.76577368e-02
1.01450570e-01 4.93879527e-01 -3.84481996e-02 3.47468793e-01
3.47089991e-02 1.02002680e+00 -5.46900094e-01 -5.43291092e-01
-3.42672497e-01 4.94336605e-01 -6.97655141e-01 1.95982799e-01
-4.82770294e-01 5.92219889e-01 -6.02479935e-01 1.01055467e+00
3.79622996e-01 -1.33151710e-01 -1.47912614e-02 2.60822415e-01
-3.66464287e-01 -3.28596711e-01 -7.57269025e-01 1.03900230e+00
-1.68319821e-01 -4.66984361e-02 9.26780820e-01 -5.75045347e-01
1.02196348e+00 1.41528055e-01 5.03332257e-01 -1.92937851e-01
4.96462286e-01 5.96269786e-01 2.47484311e-01 -5.72913826e-01
2.79606551e-01 -4.41126049e-01 -1.12186618e-01 1.26108721e-01
-4.69626218e-01 -1.24167013e+00 4.05406728e-02 -1.35507330e-01
1.32736790e+00 4.79958653e-01 5.48978671e-02 -6.86990976e-01
2.79243797e-01 -1.48887709e-01 5.29241443e-01 2.86818147e-01
-1.56401843e-01 3.42273414e-01 2.15537623e-02 -4.49665904e-01
-1.08183432e+00 -1.28847098e+00 -2.47747898e-01 9.83996511e-01
7.38444567e-01 -1.21350259e-01 -9.22825933e-01 3.63414139e-01
6.22461081e-01 -4.36495468e-02 -7.14120984e-01 -2.88068652e-01
-9.04732466e-01 -9.34907496e-01 2.51778036e-01 1.44584119e-01
2.73853511e-01 -1.46321213e+00 -1.62044394e+00 3.99315655e-01
2.19639093e-01 -7.84208179e-01 2.74057221e-02 3.35690469e-01
-6.25553668e-01 -6.47782087e-01 -1.05028307e+00 -5.55178523e-01
5.46293199e-01 -2.98882395e-01 7.49529600e-01 4.23893869e-01
-4.13191646e-01 1.99102134e-01 -4.38651890e-01 -3.69285315e-01
-6.12528920e-01 2.09901333e-02 5.44501603e-01 -3.86108667e-01
-9.80494499e-01 -1.01826537e+00 -6.91753626e-01 7.51912653e-01
-7.83173323e-01 2.05425739e-01 3.72284442e-01 5.78093410e-01
4.35069889e-01 -2.54518658e-01 5.51249385e-01 -1.64894983e-01
4.13390011e-01 -4.27817315e-01 -4.64833587e-01 -3.85219194e-02
3.48558486e-01 -1.35469139e-01 8.65144432e-01 -8.48596573e-01
-6.84462070e-01 -5.95808588e-03 -1.35320947e-01 1.56293169e-01
2.52794087e-01 2.74430305e-01 2.65740678e-02 -4.62484986e-01
4.26745296e-01 1.35348931e-01 4.19529527e-01 -1.13039888e-01
-1.11958332e-01 3.48914653e-01 2.58776069e-01 -1.05045080e+00
8.52639377e-01 8.07237506e-01 4.00683492e-01 -1.16585588e+00
1.39012784e-01 6.72828674e-01 -2.99782336e-01 -5.76315582e-01
7.90351808e-01 -1.60969913e-01 -1.44680464e+00 8.79096746e-01
-6.28659189e-01 -1.15484524e+00 -6.31232917e-01 9.31302607e-02
-1.02303362e+00 1.21282667e-01 -8.96488070e-01 -8.91805232e-01
-5.56307495e-01 -9.31564748e-01 9.82154429e-01 4.86518323e-01
-6.68182254e-01 -8.80357325e-01 3.72435391e-01 1.69786066e-01
9.61650968e-01 8.08467329e-01 6.92343175e-01 9.08375084e-02
-4.64776844e-01 -9.74859893e-02 8.44778299e-01 -2.82830656e-01
4.15197993e-03 5.06435812e-01 -4.65339541e-01 -5.26898503e-01
1.73172774e-03 -4.99109596e-01 6.34795845e-01 6.51754081e-01
5.37744224e-01 -6.84219122e-01 -7.77723849e-01 5.47618449e-01
1.01954401e+00 2.89193481e-01 3.60621899e-01 4.00973737e-01
2.57908285e-01 7.51156628e-01 2.44323894e-01 8.01100612e-01
1.67335927e-01 8.23283434e-01 9.55677032e-01 -1.24190673e-01
7.27376640e-02 2.74155498e-01 6.20178521e-01 1.05289924e+00
-8.66890490e-01 -4.93306696e-01 -8.83181572e-01 1.81097195e-01
-1.63262343e+00 -6.87262118e-01 2.54078340e-02 1.94014978e+00
1.04115045e+00 2.31387079e-01 4.18199867e-01 -5.23384660e-02
3.47274691e-01 -1.85487032e-01 -8.17635655e-01 -3.25767457e-01
-4.86353606e-01 -4.18973900e-02 4.01547849e-01 5.62099278e-01
-7.61812329e-01 7.35179663e-01 6.03069448e+00 3.22212964e-01
-1.37026846e+00 -2.64418572e-01 6.53009117e-02 -2.28368357e-01
-3.61026168e-01 -1.07667774e-01 -5.00695229e-01 8.67916107e-01
3.56001049e-01 -3.11747864e-02 3.53021264e-01 6.30340815e-01
2.07400948e-01 -4.04291987e-01 -7.40106046e-01 2.07517862e-01
-3.07720721e-01 -1.15183830e+00 -1.22249238e-02 1.87184066e-01
4.85813737e-01 -1.07899621e-01 1.24416910e-01 -3.04199219e-01
3.98820996e-01 -8.28285038e-01 1.19935930e+00 8.96332026e-01
3.90725136e-01 -6.09498620e-02 5.41427910e-01 6.34789705e-01
-1.08699048e+00 -3.60001922e-02 -3.76533359e-01 -4.46298331e-01
6.57307029e-01 5.67097425e-01 -4.09496248e-01 1.00534409e-02
7.95555711e-01 3.97488475e-01 -1.10947482e-01 7.52668321e-01
2.83395767e-01 2.23673239e-01 -7.87045240e-01 -6.76551938e-01
-1.72199845e-01 -6.18950427e-01 1.09394479e+00 6.98293746e-01
3.50724876e-01 3.21641117e-01 6.11603707e-02 1.26523066e+00
3.86378914e-01 -2.82678276e-01 -6.53976560e-01 8.34318693e-04
3.38947624e-01 1.20019734e+00 -8.85096908e-01 -9.50190127e-02
4.36352968e-01 5.46506643e-01 -1.19186789e-01 1.15115181e-01
-1.09639382e+00 -4.35639948e-01 8.43563497e-01 5.15652537e-01
-1.28410473e-01 -6.15983725e-01 -2.82597721e-01 -1.17324221e+00
-1.61211982e-01 -4.88669872e-01 -3.93325120e-01 -7.98181117e-01
-1.16333461e+00 4.24648136e-01 -2.67031789e-01 -9.97815132e-01
2.22695038e-01 -4.35336113e-01 -5.04792571e-01 1.10238470e-01
-9.13037002e-01 -1.02073443e+00 -4.66609269e-01 1.28497541e-01
2.11501151e-01 1.46498024e-01 7.15182304e-01 -3.50834221e-01
-5.03291547e-01 1.82006061e-01 6.77208975e-02 -3.36324155e-01
3.28907549e-01 -9.65397298e-01 1.38762563e-01 2.96831310e-01
-9.18984830e-01 5.19222975e-01 1.27301610e+00 -9.51605082e-01
-1.97300279e+00 -4.81316298e-01 9.89927910e-03 -2.22769290e-01
4.94372219e-01 -5.43665588e-01 -9.44076538e-01 1.58779219e-01
2.61640072e-01 -1.26611784e-01 3.52997720e-01 -6.83031976e-01
3.56624544e-01 2.38536462e-01 -1.25913298e+00 8.07301223e-01
1.34945512e+00 3.06432337e-01 -3.98876101e-01 3.31516027e-01
5.43623388e-01 -5.54644048e-01 -1.07686806e+00 7.50673592e-01
9.37052310e-01 -6.15848541e-01 9.51393127e-01 -7.94864148e-02
3.58401954e-01 -4.27015096e-01 1.23034321e-01 -1.39495242e+00
-3.38707507e-01 -9.90740359e-01 -2.06791103e-01 6.53054953e-01
2.37864390e-01 -1.14742553e+00 6.73714638e-01 3.58757079e-01
-5.49292445e-01 -1.26566267e+00 -1.01803339e+00 -1.13323212e+00
3.84873927e-01 8.74283671e-01 3.57098490e-01 7.10975170e-01
5.73827326e-01 -1.96381658e-01 -7.40523189e-02 -4.18967456e-02
5.59857130e-01 5.82062155e-02 5.71435273e-01 -1.41061866e+00
-5.72737157e-01 -5.48104525e-01 -8.08300897e-02 -6.02928758e-01
-1.53722048e-01 -5.58243930e-01 3.40338409e-01 -1.53212452e+00
-5.53084537e-02 -8.34380448e-01 5.95728040e-01 6.35109663e-01
1.90787718e-01 1.09241098e-01 2.65413821e-01 4.38592672e-01
-2.47416943e-01 9.72255945e-01 1.79468286e+00 5.74698374e-02
-5.13509512e-01 -1.93116106e-02 3.65515612e-02 9.26164865e-01
8.68950665e-01 -1.27517924e-01 1.19131789e-01 -1.03214994e-01
7.06734180e-01 2.96838939e-01 3.71947646e-01 -1.36503685e+00
1.95766926e-01 -6.77046895e-01 -6.36204258e-02 1.96329713e-01
9.41389740e-01 -6.89055502e-01 7.02098548e-01 1.28426421e+00
-5.73333353e-03 -1.66921958e-01 -1.65663078e-01 2.40036428e-01
1.95153296e-01 3.46306711e-02 1.17488062e+00 -3.87097776e-01
3.29109311e-01 -1.89986490e-02 -1.01874781e+00 3.16659524e-03
1.38019419e+00 -5.52964449e-01 -6.58313930e-01 1.95103302e-03
-8.61809492e-01 1.94853678e-01 1.42594862e+00 -2.09292680e-01
4.04632896e-01 -9.33052063e-01 -4.94192243e-01 1.67229608e-01
-4.06049371e-01 3.26074064e-01 -6.47921264e-02 8.54362488e-01
-9.99515831e-01 -1.96164280e-01 -7.95013189e-01 -5.04028082e-01
-1.02556944e+00 9.27984267e-02 7.01065242e-01 1.97544008e-01
-4.67694610e-01 1.04026437e+00 2.06389323e-01 -4.27845120e-01
-1.28964141e-01 -6.52521193e-01 8.06862339e-02 -2.03792989e-01
-3.29170018e-01 4.68309671e-01 -3.22967112e-01 -5.03566980e-01
-5.73638797e-01 1.29010713e+00 6.86482549e-01 1.58474281e-01
1.80896020e+00 2.49815881e-01 -3.92297149e-01 3.33242238e-01
2.08859280e-01 1.70955166e-01 -1.43393874e+00 4.54957902e-01
-5.88386297e-01 1.03367800e-02 -8.03438723e-01 -5.51523626e-01
-8.18259120e-01 4.57360595e-01 -2.23453552e-01 6.95108831e-01
7.47926176e-01 4.79992598e-01 8.75067174e-01 3.21627408e-01
7.35185862e-01 -1.08358645e+00 6.36280060e-01 6.49963558e-01
1.45455682e+00 -5.07170677e-01 -1.19678125e-01 -6.84868455e-01
-3.17162991e-01 9.71676528e-01 6.39058888e-01 -5.06477773e-01
5.49620688e-01 1.08859217e+00 -2.91368544e-01 -1.18754543e-01
-8.89514089e-01 1.46941632e-01 -2.96470791e-01 5.65828979e-01
1.06937541e-02 3.83017421e-01 -5.31191766e-01 4.92619157e-01
-5.87256670e-01 -1.91543430e-01 7.72467315e-01 1.35888755e+00
-1.12745750e+00 -8.33628953e-01 -4.90323484e-01 4.72977050e-02
-1.85881555e-02 5.26870072e-01 -7.40077257e-01 5.51934540e-01
3.50885719e-01 4.58879054e-01 -7.25072101e-02 -1.89007998e-01
3.37246329e-01 -3.10818195e-01 5.53939462e-01 -2.77962297e-01
-5.64560115e-01 -2.32192203e-01 -1.96857929e-01 -2.57692069e-01
-3.41965616e-01 -7.35789478e-01 -1.68831110e+00 -3.35120022e-01
-3.49414229e-01 1.48746893e-01 6.63090646e-01 5.88455439e-01
2.90158391e-01 3.43567789e-01 3.01480561e-01 -1.50328767e+00
-5.66161692e-01 -7.81848609e-01 -6.06579900e-01 2.35878780e-01
3.07063103e-01 -1.20533669e+00 -5.84583044e-01 1.87206835e-01] | [5.640406131744385, 3.998598098754883] |
a130f7af-ac8e-4678-88e8-3122bcb7c5c8 | pre-training-on-high-resource-speech | 1809.01431 | null | http://arxiv.org/abs/1809.01431v2 | http://arxiv.org/pdf/1809.01431v2.pdf | Pre-training on high-resource speech recognition improves low-resource speech-to-text translation | We present a simple approach to improve direct speech-to-text translation
(ST) when the source language is low-resource: we pre-train the model on a
high-resource automatic speech recognition (ASR) task, and then fine-tune its
parameters for ST. We demonstrate that our approach is effective by
pre-training on 300 hours of English ASR data to improve Spanish-English ST
from 10.8 to 20.2 BLEU when only 20 hours of Spanish-English ST training data
are available. Through an ablation study, we find that the pre-trained encoder
(acoustic model) accounts for most of the improvement, despite the fact that
the shared language in these tasks is the target language text, not the source
language audio. Applying this insight, we show that pre-training on ASR helps
ST even when the ASR language differs from both source and target ST languages:
pre-training on French ASR also improves Spanish-English ST. Finally, we show
that the approach improves performance on a true low-resource task:
pre-training on a combination of English ASR and French ASR improves
Mboshi-French ST, where only 4 hours of data are available, from 3.5 to 7.1
BLEU. | ['Sharon Goldwater', 'Sameer Bansal', 'Karen Livescu', 'Adam Lopez', 'Herman Kamper'] | 2018-09-05 | pre-training-on-high-resource-speech-1 | https://aclanthology.org/N19-1006 | https://aclanthology.org/N19-1006.pdf | naacl-2019-6 | ['speech-to-text-translation'] | ['natural-language-processing'] | [ 3.70060772e-01 1.35566011e-01 -6.47498369e-02 -1.98790222e-01
-1.80624354e+00 -9.12781239e-01 6.18148327e-01 -3.15484285e-01
-6.91760659e-01 7.83833921e-01 5.36882222e-01 -7.44964957e-01
5.85275590e-01 -2.07122743e-01 -9.74327445e-01 -4.09029961e-01
3.88527960e-01 5.82158089e-01 -6.97625661e-03 -5.12913108e-01
-2.64493436e-01 3.52364451e-01 -1.05677414e+00 5.45444250e-01
7.14964449e-01 4.92119312e-01 4.25341338e-01 1.07831788e+00
5.81855178e-02 6.92690134e-01 -9.82864916e-01 -2.61881024e-01
3.97220016e-01 -6.25882506e-01 -8.53328109e-01 -2.13820696e-01
2.63712376e-01 -1.72321215e-01 -4.57697362e-01 6.00558460e-01
7.72913456e-01 8.19631442e-02 4.53987181e-01 -5.79101443e-01
-5.88272631e-01 1.04724288e+00 -8.09102952e-02 3.79399478e-01
4.00942028e-01 2.13985011e-01 1.00373411e+00 -1.11914730e+00
5.27457416e-01 1.36487949e+00 5.40538788e-01 7.12457001e-01
-1.19002843e+00 -5.58801115e-01 -6.92107305e-02 -1.27148822e-01
-1.41603482e+00 -1.24240673e+00 2.30914280e-01 -1.28226846e-01
1.56073487e+00 3.00644457e-01 1.93520501e-01 1.43011022e+00
-8.64530802e-02 8.23748171e-01 1.00596499e+00 -7.64601886e-01
-2.20481940e-02 2.49964073e-01 -2.35519797e-01 3.45482796e-01
-3.69449437e-01 3.30341369e-01 -7.87225008e-01 8.42487291e-02
3.51807654e-01 -6.07164085e-01 -3.66631538e-01 4.18550372e-01
-1.53842807e+00 7.35637844e-01 6.81186561e-03 6.03989899e-01
-1.57388359e-01 6.30718842e-02 5.38800061e-01 8.50016177e-01
4.84781414e-01 5.78043163e-01 -8.84733081e-01 -6.23789608e-01
-1.11060500e+00 -3.97387981e-01 8.35169673e-01 1.08068252e+00
6.34207666e-01 4.60427821e-01 -1.20674759e-01 1.25094163e+00
4.22026999e-02 1.22454715e+00 8.76504183e-01 -8.33969414e-01
9.21630144e-01 -1.97816446e-01 -8.26045275e-02 -9.44293737e-02
7.08853081e-02 -5.46346188e-01 -3.98614138e-01 -2.98546851e-01
4.51353669e-01 -5.17848611e-01 -1.08846986e+00 1.78788126e+00
-2.35502928e-01 -2.60101438e-01 5.26997805e-01 5.14327526e-01
6.28117085e-01 1.08110142e+00 -2.88345814e-01 -3.10181439e-01
1.09084284e+00 -1.32484007e+00 -7.44151473e-01 -6.23280704e-01
9.95663047e-01 -1.19240248e+00 1.48567986e+00 1.38938874e-01
-1.38022196e+00 -5.77018976e-01 -9.24446166e-01 -5.76437339e-02
-1.33864135e-01 5.40938079e-01 -1.16106771e-01 6.17025733e-01
-1.38521564e+00 3.19901526e-01 -7.64718473e-01 -4.85286802e-01
-2.19716862e-01 2.90605307e-01 -3.81676286e-01 -2.15449184e-01
-1.37655771e+00 9.95809674e-01 1.55452654e-01 -1.81429833e-01
-1.21881580e+00 -5.46722770e-01 -8.57861221e-01 -1.51812555e-02
3.67404282e-01 -4.51336682e-01 1.81944001e+00 -1.36262238e+00
-2.02953339e+00 6.87911391e-01 -6.45214260e-01 -4.11720008e-01
3.87927890e-01 -2.80731142e-01 -6.67875290e-01 1.01997271e-01
4.34052944e-02 3.68642777e-01 9.03366864e-01 -8.45917702e-01
-6.44793391e-01 -3.00929956e-02 -2.47982070e-01 4.40885723e-01
-4.07703929e-02 5.80915630e-01 -3.37862462e-01 -8.08236420e-01
-2.30502889e-01 -1.04019916e+00 9.87075493e-02 -7.97055900e-01
-2.36533508e-01 4.12029549e-02 4.17244285e-01 -1.12531471e+00
1.22642016e+00 -2.33862495e+00 1.52655169e-01 4.81389165e-02
-3.26865524e-01 3.44543874e-01 -4.99090374e-01 5.31272113e-01
-1.64272889e-01 2.73283809e-01 -3.58205348e-01 -5.35851657e-01
-1.26096383e-01 2.59363621e-01 -4.54448134e-01 1.19210258e-01
3.21942687e-01 9.74697232e-01 -6.71573400e-01 -3.40216793e-02
2.57472508e-02 4.95187730e-01 -2.77225822e-01 2.63998598e-01
4.45910431e-02 6.44743264e-01 8.81172791e-02 3.32776934e-01
1.40581742e-01 1.46947429e-01 -1.73887832e-03 1.68744236e-01
-1.52298048e-01 1.24040115e+00 -6.70404196e-01 1.73567355e+00
-1.05484235e+00 9.74270940e-01 1.94101632e-01 -7.41375148e-01
8.42412174e-01 6.70085490e-01 1.85563177e-01 -9.36041534e-01
-1.54257834e-01 8.47027183e-01 3.92564029e-01 -1.28155515e-01
3.03885370e-01 -2.54039139e-01 -5.44948392e-02 6.36656880e-01
1.07454680e-01 -3.86677474e-01 5.68192750e-02 1.87478840e-01
1.29902053e+00 -2.21516013e-01 4.90065254e-02 -1.01818547e-01
5.84744334e-01 1.74150744e-04 2.50332862e-01 7.06120968e-01
4.52262387e-02 7.74583042e-01 2.52741110e-02 1.57351062e-01
-1.31352079e+00 -1.07384765e+00 4.41976339e-02 1.27832925e+00
-5.90149760e-01 -5.92055261e-01 -8.15468729e-01 -8.26489031e-01
-3.39836687e-01 1.09621990e+00 -1.44675061e-01 -1.42201230e-01
-1.00593984e+00 -3.31800252e-01 1.03457081e+00 3.51003975e-01
1.42624617e-01 -1.10609233e+00 1.87236920e-01 3.03427696e-01
-5.07529318e-01 -1.30346966e+00 -8.75611782e-01 3.42177302e-01
-5.31096041e-01 -3.43415678e-01 -9.79994893e-01 -6.67151690e-01
1.77486733e-01 2.65070677e-01 1.31588423e+00 -1.50533944e-01
2.70470411e-01 2.88929760e-01 -4.93976235e-01 -8.74523968e-02
-1.29298627e+00 4.82085764e-01 1.47355318e-01 -1.52525648e-01
2.38671824e-01 -1.64856881e-01 8.02284852e-03 4.02385682e-01
-4.37554926e-01 -7.34462589e-02 6.99367702e-01 8.28015208e-01
3.43295753e-01 -3.63143742e-01 7.26160288e-01 -6.30046189e-01
2.90349841e-01 -2.46555045e-01 -2.87331790e-01 2.52699614e-01
-4.23183054e-01 2.24354133e-01 7.25157022e-01 -5.22924781e-01
-8.44945431e-01 -3.66908056e-03 -6.82854295e-01 -2.90778190e-01
-1.36030421e-01 4.84328330e-01 -2.59388596e-01 3.46848875e-01
7.53781140e-01 4.67776626e-01 -1.25672251e-01 -5.98893881e-01
3.74105364e-01 1.23413014e+00 3.58169377e-01 -4.74312842e-01
8.16969573e-01 -1.09245017e-01 -6.67051196e-01 -9.86509800e-01
-6.09678566e-01 -4.03793871e-01 -3.65530699e-01 3.38636905e-01
7.04280794e-01 -1.30799758e+00 -2.76927650e-02 1.76163703e-01
-1.32646942e+00 -9.22224581e-01 -5.09576917e-01 7.49320567e-01
-7.62867451e-01 1.30377218e-01 -6.95237815e-01 -6.46787524e-01
-4.41424906e-01 -1.43789971e+00 1.24498177e+00 -6.93765461e-01
-1.89443156e-01 -8.41137648e-01 1.12290077e-01 4.34529483e-01
5.07401347e-01 -8.26544702e-01 7.25498915e-01 -9.36763823e-01
-2.48228058e-01 9.03303549e-02 -1.62085909e-02 8.02736521e-01
3.84523153e-01 -1.37037620e-01 -1.06268001e+00 -5.20986319e-01
-3.28700878e-02 -2.00173289e-01 6.99116647e-01 1.90379456e-01
4.95000392e-01 -4.36488658e-01 1.21948197e-01 4.32755500e-01
9.33490217e-01 2.63514429e-01 4.88301635e-01 5.47816493e-02
7.21284151e-01 4.29882228e-01 5.54146945e-01 -6.44801036e-02
3.82736653e-01 9.25575554e-01 -3.32686126e-01 -2.19087228e-01
-6.03734493e-01 -3.61337900e-01 1.37887347e+00 1.59333491e+00
2.21177965e-01 -4.09328133e-01 -9.96163070e-01 6.86288655e-01
-1.50515497e+00 -6.74647927e-01 6.14587143e-02 2.42483902e+00
1.31845224e+00 8.73216540e-02 3.49542648e-01 -2.61662416e-02
6.21736169e-01 -6.14670180e-02 -1.65356055e-01 -7.83552825e-01
-3.78063619e-01 3.59407753e-01 6.95444405e-01 1.08321297e+00
-6.98145390e-01 1.45745099e+00 6.91484451e+00 1.07904410e+00
-1.39987326e+00 4.96346176e-01 4.23542053e-01 -2.83896297e-01
-4.91608262e-01 -1.21520072e-01 -8.29606652e-01 4.42726433e-01
1.91347885e+00 -1.68004110e-01 1.04300380e+00 5.57254314e-01
3.84052962e-01 2.77172804e-01 -1.37838399e+00 1.10987020e+00
1.33334726e-01 -9.11719441e-01 -6.18018694e-02 -6.80290982e-02
6.76711380e-01 6.51502967e-01 -2.53593456e-02 5.14355838e-01
4.97811317e-01 -1.15990806e+00 1.01472628e+00 -5.47928624e-02
1.31551039e+00 -6.69376254e-01 6.66710138e-01 5.13388216e-01
-9.97219324e-01 1.09801628e-01 -2.19907403e-01 8.54821131e-02
2.47040585e-01 4.47347552e-01 -1.28271031e+00 4.88248944e-01
4.01771575e-01 6.35731041e-01 -4.75842983e-01 3.82123619e-01
-4.79276717e-01 1.20889592e+00 -4.77496266e-01 2.30460346e-01
1.26732334e-01 2.24459276e-01 7.44466960e-01 1.60397696e+00
6.04552865e-01 -3.06921154e-01 7.07289875e-02 3.31113487e-01
-1.64321020e-01 3.63656968e-01 -8.06581259e-01 -3.70074600e-01
5.51984251e-01 6.74993634e-01 -1.14668302e-01 -5.48458576e-01
-4.40567255e-01 1.25772989e+00 2.79943675e-01 5.29993355e-01
-5.77436924e-01 -3.51007879e-01 6.08095169e-01 8.67792405e-03
3.21598858e-01 -5.26712298e-01 -1.04317889e-01 -1.24456120e+00
-8.69455636e-02 -1.58407211e+00 -1.54398871e-03 -8.27800751e-01
-9.24348593e-01 9.10328567e-01 -5.18829167e-01 -9.16454017e-01
-8.36291492e-01 -3.10301483e-01 -1.37225434e-01 1.40057826e+00
-1.50757813e+00 -1.05624521e+00 4.19039637e-01 5.56684792e-01
1.02947164e+00 -6.21816993e-01 8.58863354e-01 5.11835337e-01
-4.56991762e-01 1.01742077e+00 3.58162731e-01 7.55913258e-02
1.03232300e+00 -1.14635944e+00 9.21750486e-01 1.01939654e+00
6.57997429e-01 5.40531099e-01 4.42478448e-01 -5.35669267e-01
-1.36002040e+00 -9.76265848e-01 1.37210667e+00 -7.78254151e-01
7.64726400e-01 -6.66618645e-01 -7.21010804e-01 8.77745330e-01
4.26639497e-01 -3.08531165e-01 5.21551132e-01 1.29699141e-01
-3.85930926e-01 -5.45057021e-02 -7.78047681e-01 6.92815065e-01
9.68074203e-01 -1.18223035e+00 -5.62276423e-01 2.74274886e-01
1.25101435e+00 -3.56852353e-01 -7.36105025e-01 2.04689071e-01
1.94977030e-01 -3.03138554e-01 6.84556186e-01 -5.30337691e-01
2.10567251e-01 -2.40475565e-01 -6.45794690e-01 -1.81038964e+00
-9.64482427e-02 -1.06911683e+00 2.00602040e-01 1.09382057e+00
9.91425455e-01 -6.53126299e-01 1.96802676e-01 3.77934463e-02
-5.13364375e-01 -2.69946992e-01 -1.01947260e+00 -9.32479322e-01
2.96063364e-01 -5.60039937e-01 4.39386725e-01 6.45357788e-01
-2.97010928e-01 7.61049807e-01 -1.80575773e-01 1.39943346e-01
-8.55481178e-02 -4.25099820e-01 8.27832401e-01 -5.82494795e-01
-6.64213002e-01 -2.10134432e-01 1.63342044e-01 -1.33537936e+00
3.71266931e-01 -1.11714578e+00 4.83158201e-01 -1.21871173e+00
-1.49823993e-01 -4.84382451e-01 -2.01952383e-01 6.78504765e-01
-3.46474722e-02 2.17755154e-01 3.47836256e-01 1.87790006e-01
-2.14035079e-01 5.20339549e-01 1.06710505e+00 -1.75476193e-01
-2.50398308e-01 5.27648181e-02 -5.78916550e-01 1.96720511e-01
6.77858591e-01 -6.52099788e-01 -3.14231604e-01 -8.20497632e-01
-1.60755590e-02 1.49515033e-01 -1.35538459e-01 -7.92053521e-01
-9.78891402e-02 1.20354190e-01 -1.05741419e-01 -2.99295753e-01
3.74446452e-01 -6.33504629e-01 2.84047495e-03 2.88754463e-01
-4.30586666e-01 1.93008393e-01 4.61718768e-01 1.66961402e-02
-2.28412196e-01 -2.87588984e-01 7.55305648e-01 5.58729880e-02
-2.85479546e-01 -1.05368145e-01 -7.59561777e-01 3.77848417e-01
3.49223793e-01 4.93637994e-02 -1.60739928e-01 -7.57190704e-01
-6.64582253e-01 -2.57946193e-01 4.06048954e-01 4.92528468e-01
2.75825351e-01 -1.16323292e+00 -1.16963398e+00 5.78092635e-01
-6.40012696e-02 -3.61098886e-01 -2.51734197e-01 7.61546075e-01
-1.98220655e-01 7.15026557e-01 3.22413772e-01 -5.96793056e-01
-1.23316371e+00 3.24356139e-01 3.68810564e-01 -1.58882648e-01
-3.74209821e-01 9.18269813e-01 2.38700703e-01 -6.28799975e-01
2.48335212e-01 -3.31192225e-01 3.48979175e-01 -2.65032321e-01
3.96645784e-01 1.78776190e-01 5.27748585e-01 -9.13201392e-01
-4.41714942e-01 4.91204113e-01 -1.10397898e-01 -9.55765665e-01
9.91123915e-01 -4.92280871e-01 1.94220200e-01 8.17517340e-01
1.38996863e+00 7.38436759e-01 -9.50037122e-01 -2.96998739e-01
-1.05061777e-01 -8.00874829e-02 2.09469609e-02 -1.15307939e+00
-7.10898340e-01 1.12253714e+00 2.85059363e-01 -1.50642216e-01
1.00327444e+00 9.40538272e-02 1.00461066e+00 6.58533812e-01
3.87038618e-01 -1.11640608e+00 4.14363900e-03 1.07821846e+00
1.13055968e+00 -1.11867154e+00 -5.72742462e-01 -1.46593392e-01
-7.95093834e-01 8.30986619e-01 7.44781783e-03 1.38368234e-01
4.56690758e-01 3.83294523e-01 4.29271042e-01 4.44881469e-01
-8.39640319e-01 -2.83617973e-01 3.32630098e-01 4.84812349e-01
6.36986136e-01 2.35127702e-01 1.05315596e-01 4.96084750e-01
-7.67349184e-01 -3.58548433e-01 4.43945259e-01 5.81094563e-01
-2.66861796e-01 -1.33195627e+00 -2.70302802e-01 7.41285607e-02
-8.09169292e-01 -6.98400438e-01 -6.10513270e-01 4.59722400e-01
-3.58645111e-01 1.28766179e+00 -6.26277179e-03 -3.99287730e-01
3.13613772e-01 3.23963732e-01 3.54801208e-01 -1.05722821e+00
-7.35025644e-01 5.26673555e-01 6.13680959e-01 -5.23469627e-01
1.04258552e-01 -6.13422513e-01 -1.20782065e+00 -3.17393929e-01
-3.52707744e-01 2.82368153e-01 9.05298293e-01 1.00250041e+00
4.32444364e-01 4.65811580e-01 9.91199553e-01 -5.00230074e-01
-7.50029922e-01 -1.13686597e+00 -2.09967405e-01 4.23011603e-03
7.25555301e-01 -9.45674181e-02 -6.43675268e-01 1.05351120e-01] | [14.461509704589844, 7.08357048034668] |
62616794-c594-409f-ae7e-d6f71da2d109 | real-time-neural-style-transfer-for-videos | null | null | http://openaccess.thecvf.com/content_cvpr_2017/html/Huang_Real-Time_Neural_Style_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Huang_Real-Time_Neural_Style_CVPR_2017_paper.pdf | Real-Time Neural Style Transfer for Videos | Recent research endeavors have shown the potential of using feed-forward convolutional neural networks to accomplish fast style transfer for images. In this work, we take one step further to explore the possibility of exploiting a feed-forward network to perform style transfer for videos and simultaneously maintain temporal consistency among stylized video frames. Our feed-forward network is trained by enforcing the outputs of consecutive frames to be both well stylized and temporally consistent. More specifically, a hybrid loss is proposed to capitalize on the content information of input frames, the style information of a given style image, and the temporal information of consecutive frames. To calculate the temporal loss during the training stage, a novel two-frame synergic training mechanism is proposed. Compared with directly applying an existing image style transfer method to videos, our proposed method employs the trained network to yield temporally consistent stylized videos which are much more visually pleasant. In contrast to the prior video style transfer method which relies on time-consuming optimization on the fly, our method runs in real time while generating competitive visual results.
| ['Hao-Zhi Huang', 'Xiaolong Zhu', 'Wenhan Luo', 'Lin Ma', 'Wei Liu', 'Hao Wang', 'Zhifeng Li', 'Wenhao Jiang'] | 2017-07-01 | null | null | null | cvpr-2017-7 | ['video-style-transfer'] | ['computer-vision'] | [ 3.96952540e-01 -5.49377836e-02 9.63583365e-02 -4.41108584e-01
-2.84959048e-01 -4.83009279e-01 6.80014193e-01 -2.85013199e-01
-3.31155449e-01 6.78444088e-01 1.24780357e-01 -1.07515007e-01
2.87813902e-01 -7.50592232e-01 -1.05060279e+00 -5.04005611e-01
4.05246496e-01 -2.09908217e-01 1.46875620e-01 -8.84538889e-03
2.48635560e-01 4.95130450e-01 -1.28965592e+00 1.94596767e-01
7.45551586e-01 1.04471552e+00 3.26481164e-01 6.16662264e-01
-1.73211530e-01 9.25233662e-01 -3.09964031e-01 -3.91234070e-01
2.39997178e-01 -8.93803537e-01 -8.59813154e-01 5.36275208e-01
5.83922684e-01 -5.02888799e-01 -4.63511914e-01 9.78202045e-01
3.24601233e-01 4.46019202e-01 4.49712902e-01 -1.11815572e+00
-9.20629501e-01 1.29375875e-01 -5.90691566e-01 -1.92093216e-02
3.65900308e-01 2.86287397e-01 7.18105674e-01 -7.70192325e-01
8.90081286e-01 1.20902181e+00 3.25078696e-01 6.33568287e-01
-1.33953941e+00 -6.67070985e-01 5.10179520e-01 1.84598431e-01
-8.54771316e-01 -3.79593432e-01 1.13552129e+00 -3.66712868e-01
5.79506338e-01 -5.53358458e-02 8.94861877e-01 1.18568015e+00
1.72084257e-01 8.82716835e-01 1.02399766e+00 -5.06416559e-01
2.09937498e-01 3.59311351e-04 -4.89731908e-01 6.72585070e-01
-1.33801907e-01 2.84360796e-01 -7.89047599e-01 4.10175890e-01
1.09703338e+00 9.38932225e-02 -4.07832474e-01 -5.27758062e-01
-1.13139307e+00 6.81924582e-01 4.70622748e-01 3.12200129e-01
-4.86290663e-01 3.44817817e-01 4.69431698e-01 4.13098961e-01
5.57229757e-01 3.58472079e-01 -1.02832936e-01 -1.64798573e-01
-1.27536237e+00 6.78076521e-02 4.76705045e-01 8.33566725e-01
8.49587023e-01 3.46852899e-01 -4.29256946e-01 4.54853565e-01
1.58889204e-01 2.73084134e-01 2.48410538e-01 -1.16804898e+00
2.44995371e-01 2.93141156e-01 2.26150006e-01 -1.05594361e+00
1.85165688e-01 -1.94352418e-01 -6.28478706e-01 5.47688425e-01
3.60533386e-01 -1.56920925e-01 -8.56736600e-01 1.86961055e+00
2.95592189e-01 4.52949196e-01 -1.22229688e-01 9.26054776e-01
4.68392074e-01 6.59700930e-01 2.17563838e-01 -1.43565848e-01
8.40314746e-01 -1.07491946e+00 -8.74016702e-01 2.48354953e-02
1.56470925e-01 -9.31785047e-01 1.24637127e+00 1.75450206e-01
-1.44643927e+00 -6.84135020e-01 -1.01456034e+00 -2.35455334e-01
-6.29917458e-02 1.31141514e-01 5.34204066e-01 2.43121564e-01
-1.03526187e+00 8.71235192e-01 -8.57202530e-01 -3.98321807e-01
4.33459282e-01 2.10427314e-01 -3.95954937e-01 2.04206631e-01
-9.67660129e-01 7.30695128e-01 2.69836098e-01 2.14037821e-01
-9.31394279e-01 -6.88856125e-01 -1.05503678e+00 3.88219133e-02
2.78741151e-01 -8.51603031e-01 1.05711818e+00 -1.85534120e+00
-2.11568737e+00 7.51681447e-01 -2.68356085e-01 -2.89882272e-01
7.39023149e-01 -4.61874783e-01 -1.22284353e-01 3.08621258e-01
-5.51928394e-02 1.10814977e+00 1.04306936e+00 -1.27932894e+00
-5.93699694e-01 1.52467325e-01 2.96711057e-01 2.37363830e-01
-5.57668149e-01 -4.28285450e-02 -6.35636032e-01 -9.55639362e-01
-3.54126841e-01 -8.80801082e-01 -5.56595623e-02 4.83119547e-01
-1.38534799e-01 1.67133138e-01 1.09129012e+00 -7.18583226e-01
9.44718540e-01 -2.10136294e+00 5.99059045e-01 -4.20755371e-02
2.97702034e-03 4.20544326e-01 -3.52633685e-01 3.24492484e-01
-1.30027086e-01 -3.01085770e-01 -2.52266467e-01 -4.76916045e-01
-3.75045210e-01 2.24346504e-01 -3.13731402e-01 3.46894175e-01
6.26960099e-01 9.99946177e-01 -1.04597962e+00 -4.29854095e-01
5.64516306e-01 6.37959540e-01 -8.16577911e-01 5.67716241e-01
-4.28704262e-01 9.80606198e-01 -2.38777548e-01 2.63141423e-01
5.08820534e-01 -1.09165177e-01 9.11776200e-02 -3.82470816e-01
-1.07934460e-01 8.69708806e-02 -7.42114544e-01 2.04260993e+00
-6.90069795e-01 8.02566648e-01 -1.42955156e-02 -1.00219238e+00
7.64678717e-01 3.39489013e-01 4.68645960e-01 -7.17300653e-01
3.25751185e-01 6.74943964e-04 -4.22552079e-01 -3.18147302e-01
4.82468575e-01 -3.15291643e-01 2.01508164e-01 4.09278989e-01
2.52240688e-01 -1.83413073e-01 1.21839941e-01 1.28369257e-01
6.50323570e-01 8.72482240e-01 -9.54339132e-02 -2.34883558e-02
6.85971260e-01 -3.56415480e-01 3.87222797e-01 2.74889320e-01
-1.80088162e-01 7.05057502e-01 3.43643159e-01 -4.27064478e-01
-1.20838463e+00 -1.03717113e+00 4.19446141e-01 8.67176533e-01
1.96151197e-01 -1.42742842e-01 -8.10717225e-01 -6.93401337e-01
-2.30458647e-01 6.24692261e-01 -7.38008142e-01 -3.31327856e-01
-7.50123262e-01 1.70244873e-01 2.05736712e-01 6.16142333e-01
7.58695841e-01 -1.30033851e+00 -5.71101129e-01 2.82019138e-01
-2.30868593e-01 -1.11278224e+00 -1.06633615e+00 -4.79043946e-02
-8.60954762e-01 -7.51535475e-01 -1.12247932e+00 -9.48687792e-01
7.14398146e-01 2.11814240e-01 9.59287822e-01 -8.09761360e-02
-9.08958912e-02 3.46702039e-01 -2.60721564e-01 -8.60959142e-02
-3.07986528e-01 -1.41074464e-01 -1.98504165e-01 3.93542439e-01
-1.36832297e-01 -8.29489827e-01 -7.69726157e-01 1.33538201e-01
-1.13033199e+00 4.16101307e-01 4.91133273e-01 1.07217038e+00
5.21606028e-01 -2.39144593e-01 4.49147969e-01 -7.44480491e-01
2.93826818e-01 -1.67314604e-01 -5.89416564e-01 2.66743213e-01
-3.60980779e-01 2.56061673e-01 7.93466330e-01 -5.60701013e-01
-1.40027225e+00 3.86539042e-01 -4.53790165e-02 -9.64456975e-01
4.57003862e-02 1.96523637e-01 -8.02280009e-02 -2.62556255e-01
2.41084397e-01 2.99106658e-01 3.02014023e-01 -2.33286634e-01
6.41825855e-01 1.58749223e-01 6.55136228e-01 -5.13047814e-01
8.65767777e-01 5.97951233e-01 -4.77102026e-02 -5.14028907e-01
-7.55368531e-01 -1.29116639e-01 -6.86174214e-01 -5.05682766e-01
1.10815704e+00 -7.98987687e-01 -6.10959709e-01 5.75245261e-01
-1.19102180e+00 -5.68271399e-01 -3.23730409e-01 4.47692126e-01
-7.88196087e-01 5.09174407e-01 -6.37251794e-01 -4.68146324e-01
-3.18113506e-01 -9.70933199e-01 1.02199769e+00 1.76109865e-01
-2.70911008e-01 -1.14762723e+00 1.05825614e-03 2.99686333e-03
4.54277962e-01 3.39458257e-01 6.21309757e-01 1.54499859e-01
-5.68902731e-01 6.39374927e-02 -3.71945441e-01 4.84322131e-01
4.72108483e-01 1.84823230e-01 -7.45305240e-01 -3.85483831e-01
-3.90004255e-02 -3.30047309e-01 8.77423882e-01 4.40705925e-01
1.18462574e+00 -2.22485274e-01 7.21493131e-03 7.63859212e-01
1.26710761e+00 3.23829174e-01 6.98664486e-01 3.13319236e-01
8.64120841e-01 6.84731424e-01 5.32308638e-01 1.75351426e-01
1.36989117e-01 7.28116632e-01 2.93485612e-01 -5.67640126e-01
-3.82796139e-01 -4.19595122e-01 4.65390027e-01 5.28028905e-01
-1.44109249e-01 -2.23902091e-01 -2.19204336e-01 4.88411248e-01
-1.76502359e+00 -1.08289647e+00 3.98015082e-01 2.09383464e+00
8.31118941e-01 1.17421500e-01 1.68849781e-01 -4.27780151e-02
7.26981580e-01 2.30604544e-01 -5.74331403e-01 -4.29720879e-01
7.51856789e-02 2.37230971e-01 2.17406973e-01 4.54166353e-01
-1.02693880e+00 1.04109132e+00 6.04552412e+00 5.58753014e-01
-1.68400407e+00 -1.14919975e-01 7.41959572e-01 -3.41733575e-01
-5.02635837e-01 -3.87286544e-02 -1.89952195e-01 5.17301083e-01
6.18868113e-01 -1.73486561e-01 5.55946946e-01 5.82804561e-01
5.62754273e-01 -4.35434841e-02 -1.13727510e+00 8.63816321e-01
-5.57052251e-03 -1.45753491e+00 3.66877168e-01 -1.60382152e-01
9.44969475e-01 -6.15169168e-01 3.86166602e-01 2.04210375e-02
1.85555652e-01 -8.14446867e-01 1.06165957e+00 5.96560717e-01
9.77182865e-01 -8.15892696e-01 3.88769001e-01 -1.93016559e-01
-1.14963210e+00 1.26833737e-01 7.96443447e-02 -2.21334517e-01
4.72116292e-01 3.55749577e-01 -3.88561934e-01 5.87967038e-01
6.40911579e-01 1.09295285e+00 -2.96812564e-01 6.95990860e-01
-2.08597496e-01 4.29200351e-01 2.11250499e-01 2.48006806e-01
5.10849595e-01 -4.19889003e-01 4.30887133e-01 1.16307592e+00
2.95198262e-01 -1.33427009e-01 4.22095768e-02 1.05659795e+00
-7.38070980e-02 -3.79463546e-02 -7.63666451e-01 -2.47897685e-01
1.45771056e-02 1.16428626e+00 -6.34784222e-01 -3.74245256e-01
-4.96479362e-01 1.58156478e+00 4.36235517e-01 5.10421693e-01
-9.08582926e-01 -3.24519277e-01 4.67572480e-01 5.31458296e-02
6.04544044e-01 -4.14324522e-01 -4.22456771e-01 -1.24572110e+00
5.53433858e-02 -5.98264337e-01 5.53911924e-02 -9.41991985e-01
-1.10293818e+00 7.18480706e-01 -2.74504840e-01 -1.36035895e+00
-3.60856950e-01 -4.81600255e-01 -9.70997870e-01 8.62381279e-01
-1.66917002e+00 -1.34518218e+00 -4.37287211e-01 7.15465128e-01
7.35338211e-01 1.08611099e-02 5.39125979e-01 3.33352178e-01
-3.42635810e-01 5.84772766e-01 -1.65174261e-01 -3.10226083e-02
9.13526058e-01 -1.17757165e+00 2.91021258e-01 9.98023927e-01
1.84799805e-02 4.76360977e-01 7.41184592e-01 -6.23916030e-01
-1.12204766e+00 -1.24705637e+00 6.43576324e-01 -6.82121562e-03
4.67036635e-01 -1.54107809e-01 -8.94199371e-01 5.78248501e-01
7.04149783e-01 -6.96542785e-02 2.63898462e-01 -3.12836975e-01
-2.58222014e-01 -2.07494959e-01 -9.81768548e-01 7.46558130e-01
1.05099678e+00 -5.75486064e-01 -3.76011461e-01 -6.76603839e-02
6.65143788e-01 -3.15467805e-01 -5.74237406e-01 1.60891324e-01
6.28739715e-01 -8.46091628e-01 8.49510074e-01 -5.39836645e-01
8.71078968e-01 -4.08480912e-01 2.30950400e-01 -1.31255269e+00
-3.02996278e-01 -7.63542473e-01 1.40180662e-01 1.27370667e+00
5.25554232e-02 -2.70377427e-01 7.06307709e-01 4.99750286e-01
-2.07664832e-01 -5.08076549e-01 -5.93286455e-01 -7.07377136e-01
-4.92877001e-03 2.13377289e-02 1.37074068e-01 7.61116266e-01
-2.12857425e-01 2.91101843e-01 -8.13763380e-01 -1.59807786e-01
4.95108724e-01 2.49233887e-01 6.56065047e-01 -7.30231464e-01
-3.88004571e-01 -3.43084455e-01 -1.90054953e-01 -1.03441727e+00
3.56571287e-01 -6.56246305e-01 1.41530976e-01 -1.31279969e+00
1.12380229e-01 -1.15016110e-01 -2.97827482e-01 4.20683324e-01
-2.43023828e-01 4.12079573e-01 3.17108154e-01 3.42246816e-02
-5.17877221e-01 8.17460179e-01 1.80550420e+00 -3.49420006e-03
-2.03746602e-01 -1.54015198e-01 -4.51723695e-01 5.45955181e-01
5.77861607e-01 -1.64640471e-02 -6.27618253e-01 -5.13346136e-01
-1.56619191e-01 1.63577616e-01 4.50845987e-01 -7.77571321e-01
5.68075329e-02 -1.93469450e-01 6.28841400e-01 -3.81828785e-01
4.42907780e-01 -8.49568963e-01 2.81907976e-01 5.07874250e-01
-4.60410833e-01 5.64275570e-02 3.03786874e-01 7.66086221e-01
-4.71081257e-01 7.40591586e-02 1.02562821e+00 -1.47081539e-01
-6.89056277e-01 3.79904836e-01 -3.21695238e-01 -2.19818398e-01
1.09725785e+00 -2.40118086e-01 1.68881357e-01 -5.71876764e-01
-6.81213677e-01 -3.73262982e-03 7.35237658e-01 4.91068035e-01
6.40667975e-01 -1.44262969e+00 -3.86250943e-01 3.11920315e-01
-1.84704885e-01 -3.39115918e-01 2.98032105e-01 6.56710744e-01
-5.47446549e-01 2.45641753e-01 -7.63203859e-01 -4.79901105e-01
-1.18291271e+00 7.47328520e-01 3.69008452e-01 -7.83933625e-02
-6.77003860e-01 7.42681682e-01 4.85810608e-01 5.14535829e-02
2.59932250e-01 -1.52590096e-01 1.29801277e-02 -1.09181575e-01
3.99207681e-01 8.14219043e-02 -2.00999767e-01 -4.66013521e-01
-8.80754292e-02 6.35784864e-01 -1.29841074e-01 -4.40580040e-01
1.41961932e+00 -2.70680904e-01 4.53830846e-02 3.92909706e-01
1.32544315e+00 -7.64942318e-02 -2.03154850e+00 -2.26738349e-01
-2.25067198e-01 -6.67873621e-01 8.76487046e-02 -6.40308678e-01
-1.39784777e+00 8.08597863e-01 4.68478411e-01 -3.09871733e-01
1.38678050e+00 -3.65915149e-01 1.04479432e+00 -1.33249819e-01
1.72779784e-01 -1.13010466e+00 6.66187584e-01 3.11508149e-01
9.84768867e-01 -1.20973110e+00 -3.02299887e-01 -3.71934772e-01
-7.13006616e-01 1.24542975e+00 6.37079597e-01 -3.99992347e-01
3.36330384e-01 -5.75115383e-02 -8.43729917e-03 8.99071842e-02
-5.84719241e-01 -1.83713064e-02 4.75796342e-01 4.43155646e-01
5.38284719e-01 -2.94157028e-01 -2.89592624e-01 -1.97604802e-02
1.75008193e-01 4.07355666e-01 3.76937717e-01 8.24785054e-01
-8.53642896e-02 -1.22151196e+00 -8.57968181e-02 7.77632743e-02
-3.44320238e-01 9.96684842e-03 -2.92149395e-01 5.21813393e-01
-6.87954426e-02 7.99142241e-01 1.58151433e-01 -2.77464151e-01
2.14307725e-01 -1.49004906e-01 7.75664568e-01 -4.33565408e-01
-4.54438835e-01 1.72377422e-01 -1.33033603e-01 -7.66212046e-01
-7.14562118e-01 -6.00254238e-01 -9.67245996e-01 -3.68405759e-01
-5.59557378e-02 -2.01603681e-01 4.13210303e-01 9.49445724e-01
2.54367948e-01 7.26359248e-01 8.09339404e-01 -1.25294554e+00
-9.03091058e-02 -6.08066022e-01 -3.68985206e-01 6.90334857e-01
3.46681058e-01 -7.69135952e-01 -1.12921238e-01 5.49897134e-01] | [11.195060729980469, -0.7888762354850769] |
4f31924e-6313-4811-89b0-a4e18623fe34 | small-object-detection-using-context-and | 1912.06319 | null | https://arxiv.org/abs/1912.06319v2 | https://arxiv.org/pdf/1912.06319v2.pdf | Small Object Detection using Context and Attention | There are many limitations applying object detection algorithm on various environments. Especially detecting small objects is still challenging because they have low resolution and limited information. We propose an object detection method using context for improving accuracy of detecting small objects. The proposed method uses additional features from different layers as context by concatenating multi-scale features. We also propose object detection with attention mechanism which can focus on the object in image, and it can include contextual information from target layer. Experimental results shows that proposed method also has higher accuracy than conventional SSD on detecting small objects. Also, for 300$\times$300 input, we achieved 78.1% Mean Average Precision (mAP) on the PASCAL VOC2007 test set. | ['Seung-Ik Lee', 'Hyun-Jin Yoon', 'Jeong-Seon Lim', 'Marcella Astrid'] | 2019-12-13 | null | null | null | null | ['small-object-detection'] | ['computer-vision'] | [ 2.86659449e-01 -4.64811474e-01 1.21660724e-01 -4.30573136e-01
-4.40167725e-01 -2.81563371e-01 1.00107625e-01 1.57766864e-01
-7.96219528e-01 4.38149899e-01 6.03439112e-04 1.52583763e-01
1.32020757e-01 -7.83202291e-01 -5.20698369e-01 -6.44029319e-01
1.10320836e-01 -7.52481595e-02 1.38180232e+00 1.55761942e-01
5.75333714e-01 5.96294522e-01 -1.84679151e+00 6.30389631e-01
5.44456303e-01 1.08716846e+00 1.07071519e+00 8.99168015e-01
-2.09146619e-01 5.39383590e-01 -1.00761664e+00 -1.79329719e-02
2.85254717e-01 -1.71554908e-02 -4.24960136e-01 5.44748791e-02
8.66047144e-01 -4.44293171e-01 -1.57619178e-01 1.13953996e+00
6.98551536e-01 1.47372246e-01 5.21124125e-01 -9.71031904e-01
-6.52245581e-01 2.97841161e-01 -1.08271015e+00 9.93346393e-01
1.71430394e-01 1.38160944e-01 6.09344900e-01 -1.05798638e+00
1.77776739e-01 1.35019064e+00 3.73574346e-01 4.55393106e-01
-6.57128096e-01 -9.12113309e-01 2.71899998e-01 6.18993282e-01
-1.46914470e+00 -1.41225541e-02 3.74649912e-01 -2.90217936e-01
1.08028698e+00 3.66020560e-01 2.28660583e-01 3.89019907e-01
1.17284872e-01 9.46977317e-01 1.25303435e+00 -5.05059898e-01
-1.79294229e-01 3.99357319e-01 5.89211106e-01 4.31634605e-01
6.57474637e-01 -1.22968899e-02 -2.11343870e-01 1.89611614e-01
4.76205558e-01 9.65364799e-02 -7.08937347e-02 1.72132179e-01
-1.14514077e+00 5.18354774e-01 7.36956537e-01 2.26143658e-01
-1.47376940e-01 3.03773247e-02 2.76591629e-01 -1.69296488e-01
-1.28366739e-01 -5.22256605e-02 -3.51586074e-01 1.87149659e-01
-4.11775589e-01 6.74035251e-02 2.90829211e-01 1.12381995e+00
5.18094718e-01 -1.11486405e-01 -4.92765188e-01 8.37896705e-01
2.46053845e-01 8.27938139e-01 5.30148149e-01 -6.49360180e-01
5.94837427e-01 6.91662788e-01 2.34093666e-01 -1.07033873e+00
-3.83094579e-01 -4.49566126e-01 -3.37209016e-01 3.04196656e-01
1.67379290e-01 9.06890258e-02 -1.11903214e+00 1.33362901e+00
5.62044799e-01 1.43204883e-01 5.44169769e-02 1.15187085e+00
1.29877400e+00 8.68564904e-01 1.96247369e-01 -1.35791957e-01
1.69408488e+00 -9.61673558e-01 -6.75127804e-01 -5.06971359e-01
1.62265077e-01 -1.11007607e+00 1.11464739e+00 3.61789584e-01
-6.85095072e-01 -1.18799591e+00 -1.25021625e+00 1.57137644e-02
-5.28929889e-01 7.06611872e-01 3.68298084e-01 6.96741819e-01
-7.07714021e-01 1.07897125e-01 -4.49353933e-01 -6.14081323e-01
3.52550685e-01 6.40836656e-01 -7.82342106e-02 -2.67351985e-01
-6.78039551e-01 8.45620871e-01 7.52005279e-01 9.29409638e-02
-6.55558884e-01 -2.98050307e-02 -5.80816746e-01 1.07389197e-01
6.23434842e-01 -4.67757434e-02 1.04775298e+00 -6.81259036e-01
-8.83347750e-01 6.10578299e-01 -2.16949269e-01 -2.74674147e-01
1.63447693e-01 -4.35336709e-01 -5.44513881e-01 1.91084966e-01
9.65683535e-02 7.78983295e-01 6.15492523e-01 -1.00755942e+00
-1.32414579e+00 -4.72431213e-01 4.73620780e-02 4.09760952e-01
-4.54071790e-01 5.90765774e-01 -6.24421895e-01 -3.01580638e-01
1.71900958e-01 -6.55890763e-01 -7.28564896e-03 8.45150352e-02
-3.34480740e-02 -4.95783389e-01 1.60349643e+00 -4.93830055e-01
1.20560181e+00 -2.14632440e+00 -6.11376345e-01 -1.42137110e-01
-5.73912859e-02 8.69499683e-01 -2.40701392e-01 -6.35915622e-02
3.51703018e-01 -1.05170883e-01 1.74198002e-01 2.00109676e-01
-5.27217507e-01 3.81754078e-02 -8.56671855e-02 1.22305408e-01
2.28625268e-01 6.58266723e-01 -6.10163331e-01 -8.73551786e-01
3.50890636e-01 4.33026999e-01 -1.21909305e-01 1.25927076e-01
1.34816706e-01 -1.60844117e-01 -4.91111279e-01 7.76799202e-01
9.92302537e-01 1.96266230e-02 -3.93999755e-01 -3.85802358e-01
-2.45920002e-01 -1.72651857e-01 -1.71678340e+00 1.08065343e+00
-1.36367574e-01 7.35200226e-01 -5.45916073e-02 -6.24061286e-01
1.12071753e+00 -1.29343003e-01 -2.08044007e-01 -5.94425082e-01
3.44115347e-01 7.53795579e-02 4.72429186e-01 -8.08452964e-01
5.75957656e-01 3.32473159e-01 3.78680944e-01 1.21056587e-01
-1.29479870e-01 9.21230614e-02 3.60161215e-01 8.08516145e-02
9.52178299e-01 -7.47057796e-02 3.54098976e-01 -2.07146734e-01
8.45026493e-01 -1.86179001e-02 6.74047828e-01 1.00032163e+00
-6.16005123e-01 6.03634715e-01 -3.70473079e-02 -1.93613961e-01
-6.66720629e-01 -9.10464227e-01 -3.82194519e-01 1.25943875e+00
6.26459301e-01 -2.11715460e-01 -5.90321243e-01 -8.66009355e-01
1.20053925e-01 4.71634537e-01 -7.20536649e-01 -3.01976502e-02
-7.25494027e-01 -8.87307882e-01 1.76133275e-01 1.04832900e+00
8.91969681e-01 -1.31151199e+00 -1.02766371e+00 9.55325551e-03
7.39753842e-02 -1.33920193e+00 -5.51655710e-01 -7.40189329e-02
-8.90332878e-01 -1.23199105e+00 -5.52322268e-01 -1.16187608e+00
6.37087643e-01 8.75211060e-01 5.98058283e-01 1.94503471e-01
-7.68724620e-01 1.35931030e-01 -2.46057317e-01 -9.89401758e-01
7.55885839e-02 -2.84129441e-01 -2.79523805e-02 -3.91959250e-02
8.88033271e-01 1.89619899e-01 -8.07550967e-01 7.17097878e-01
-6.92621589e-01 -1.45425275e-01 9.32640254e-01 5.81031859e-01
5.53794503e-01 4.77840528e-02 6.32572830e-01 -5.60515881e-01
3.43504667e-01 -7.02758785e-03 -7.19700575e-01 2.46119037e-01
-4.23938453e-01 -3.50739330e-01 3.10829729e-01 -6.21176422e-01
-1.27742767e+00 1.24001190e-01 1.93843603e-01 -1.18723303e-01
-3.14662308e-01 -2.31770501e-01 -1.68642268e-01 -2.15423182e-01
7.66320169e-01 2.34408617e-01 -3.39656740e-01 -5.80292523e-01
-1.11706913e-01 1.20181537e+00 5.13249397e-01 -3.07505578e-01
4.85245496e-01 3.47870260e-01 -4.74259220e-02 -8.03889334e-01
-6.35666013e-01 -7.51312494e-01 -6.32844806e-01 -2.51655430e-01
8.43708038e-01 -9.00925338e-01 -8.12598288e-01 4.22662228e-01
-1.03037441e+00 1.54868409e-01 1.46715686e-01 7.34242916e-01
1.09265998e-01 2.81488925e-01 -5.29946268e-01 -9.76225078e-01
-4.85633463e-01 -9.90293264e-01 1.00148618e+00 7.29601383e-01
4.16028202e-01 -1.80762216e-01 -5.23680806e-01 3.01476330e-01
5.21057665e-01 -3.43195125e-02 3.02703172e-01 -5.05537331e-01
-7.23263800e-01 -5.22511423e-01 -9.31533277e-01 4.42423880e-01
2.63298243e-01 6.50221184e-02 -1.12320447e+00 -2.53435194e-01
7.73153093e-04 -9.48598608e-02 1.18429101e+00 4.02688205e-01
1.35006392e+00 -5.22197364e-03 -7.87267804e-01 1.33855492e-01
1.70044625e+00 5.90133011e-01 5.56922376e-01 2.10116744e-01
6.81891143e-01 5.76305926e-01 1.26636302e+00 4.46428433e-02
8.80351942e-03 5.47813475e-01 3.53373349e-01 1.09535605e-01
-4.59314078e-01 2.83521563e-01 2.10695729e-01 3.05969656e-01
-9.70646143e-02 -1.07308164e-01 -8.55680943e-01 5.96066058e-01
-1.66521275e+00 -1.05366719e+00 -4.69627857e-01 2.19009209e+00
4.84217614e-01 5.11975348e-01 6.30279556e-02 7.18018115e-02
1.12607217e+00 -1.46666095e-01 -5.98339081e-01 -2.30862707e-01
-1.36577740e-01 6.08866010e-03 5.06787598e-01 2.46491864e-01
-1.24658692e+00 8.26248527e-01 6.45888805e+00 8.34201813e-01
-1.04944015e+00 1.96004003e-01 3.57817322e-01 -2.94247359e-01
6.11838400e-01 -3.98589611e-01 -1.50367391e+00 6.66049480e-01
4.52657640e-01 -9.62348655e-02 -3.73043358e-01 1.09528589e+00
1.05660930e-01 -6.86643243e-01 -7.64171600e-01 9.50568974e-01
3.33681881e-01 -8.36254179e-01 -1.22147992e-01 -2.64922261e-01
5.84045768e-01 -4.14956966e-03 -2.09305510e-02 2.95296043e-01
-1.42580599e-01 -7.76155472e-01 5.51373720e-01 8.61643255e-02
6.14973664e-01 -6.55856550e-01 9.66684520e-01 4.09429550e-01
-1.54950011e+00 -5.22079229e-01 -8.65481198e-01 -1.36363983e-01
-3.51734579e-01 1.85455546e-01 -1.01243258e+00 1.48321930e-02
1.02966380e+00 3.50906819e-01 -1.09530139e+00 1.54383171e+00
-8.89143944e-02 4.80749905e-01 -4.42646563e-01 -6.38048053e-01
4.65567783e-02 2.68666059e-01 4.52943295e-01 1.68455195e+00
2.93987095e-01 4.40631092e-01 2.32097313e-01 5.03284454e-01
2.86068529e-01 3.13963443e-01 -4.53538954e-01 5.66123486e-01
4.41088110e-01 1.37321830e+00 -1.04107893e+00 -6.78942084e-01
-5.36469460e-01 8.51758122e-01 -1.30710334e-01 1.07854776e-01
-9.13311601e-01 -8.98386717e-01 8.89290720e-02 8.48529413e-02
6.20915413e-01 -5.22323027e-02 -2.05851775e-02 -7.33312428e-01
2.04049438e-01 -5.07050753e-01 6.58689022e-01 -1.12472498e+00
-7.60835230e-01 5.91532350e-01 6.75627142e-02 -1.04619014e+00
4.64550197e-01 -1.05874097e+00 -7.73356318e-01 7.34687865e-01
-1.29931271e+00 -1.29267550e+00 -8.35637689e-01 4.33908850e-01
8.76005054e-01 -5.19415326e-02 2.27519915e-01 2.94657290e-01
-5.86680532e-01 5.19942045e-01 -1.51722627e-02 1.46281764e-01
7.93366194e-01 -1.21621943e+00 2.25225557e-02 1.10787749e+00
-3.18241678e-02 5.69256127e-01 5.29363215e-01 -7.25831568e-01
-1.08731723e+00 -1.12305140e+00 3.72748435e-01 -4.27611291e-01
5.16735502e-02 -2.57697195e-01 -9.43371356e-01 5.35546601e-01
1.79147914e-01 2.59934843e-01 3.02865356e-01 -1.54485986e-01
-1.53024212e-01 -4.54072058e-01 -1.40646684e+00 2.95625329e-01
1.07460284e+00 1.34591600e-02 -7.01974809e-01 2.02975050e-01
7.91392267e-01 -3.90016347e-01 -2.38883063e-01 7.90501237e-01
5.36435425e-01 -9.28147972e-01 9.75719988e-01 -2.22661331e-01
-1.15053058e-01 -8.34986031e-01 -4.45574909e-01 -5.25276661e-01
-4.25238281e-01 8.81303623e-02 -5.30564375e-02 1.40033698e+00
2.23259822e-01 -5.17420650e-01 8.04286718e-01 3.83683860e-01
-3.51153426e-02 -5.81644773e-01 -4.31770444e-01 -8.55052352e-01
-4.11088496e-01 -2.10077226e-01 4.67336416e-01 3.54595304e-01
-5.31908274e-01 4.25294727e-01 4.33743149e-02 4.17329222e-01
5.74258208e-01 1.89189464e-01 7.17176855e-01 -1.04539454e+00
-3.18407118e-01 -1.61169514e-01 -8.18861127e-01 -8.46338451e-01
-5.78824759e-01 -2.76617050e-01 1.65232554e-01 -1.52360582e+00
7.23517001e-01 -3.66615564e-01 -7.56887555e-01 3.73300076e-01
-6.39890254e-01 6.25477850e-01 3.21227133e-01 5.65865114e-02
-8.52355957e-01 4.81353775e-02 1.18665004e+00 -2.30988771e-01
-9.65813026e-02 1.65675506e-01 -4.36529756e-01 7.36885607e-01
7.74055302e-01 -4.22017843e-01 -1.93102032e-01 -4.36294556e-01
-4.39682633e-01 -4.34006184e-01 2.25151882e-01 -1.64084244e+00
2.95993060e-01 -2.67323494e-01 9.05754685e-01 -1.20834351e+00
4.10854101e-01 -7.14147687e-01 -3.60355467e-01 7.63680220e-01
-9.74269807e-02 -1.22181058e-01 6.37901366e-01 6.41673148e-01
-1.11371586e-02 -5.79320490e-01 9.19407547e-01 -3.24193165e-02
-1.32278824e+00 -8.75069872e-02 -5.40120043e-02 -3.21399450e-01
1.41458619e+00 -4.03608859e-01 -4.95398730e-01 2.16839120e-01
-5.77969134e-01 3.83610547e-01 9.48936790e-02 6.12201273e-01
7.96283007e-01 -1.22531962e+00 -7.39217401e-01 1.61542103e-01
3.45345944e-01 -2.10182741e-02 2.17649996e-01 4.49145347e-01
-5.82349896e-01 4.38429952e-01 -3.30430001e-01 -8.28589022e-01
-2.17363954e+00 7.97493219e-01 2.07967713e-01 3.50381374e-01
-2.96206415e-01 1.15113986e+00 5.15742064e-01 8.06743652e-02
3.94109547e-01 -5.15323460e-01 -5.31412542e-01 -1.49937451e-01
1.08872473e+00 5.86976051e-01 -7.82252103e-02 -6.49082005e-01
-6.18935227e-01 9.98466253e-01 -3.04624677e-01 1.39911816e-01
8.11331987e-01 -8.54734778e-02 1.84718221e-01 2.51492947e-01
1.00437522e+00 1.30010117e-02 -1.21584105e+00 -4.61177588e-01
-1.65807024e-01 -9.81065035e-01 -6.38558641e-02 -9.28562701e-01
-8.79002631e-01 1.02012599e+00 1.23756051e+00 3.95238772e-02
1.23580122e+00 5.93440048e-02 5.05243182e-01 6.11599386e-01
3.79502028e-01 -1.17002666e+00 3.44533324e-01 2.83139944e-01
9.69062209e-01 -1.68900120e+00 3.31367940e-01 -6.19759858e-01
-5.21270454e-01 1.05982125e+00 1.43724561e+00 -2.45331287e-01
4.79458690e-01 2.97228873e-01 8.14116374e-02 -3.94909689e-03
-6.46108925e-01 -6.05758250e-01 4.00502443e-01 6.99599683e-01
1.59658998e-01 -6.57916516e-02 -6.06940866e-01 4.79086131e-01
2.95998305e-01 -1.25187933e-01 5.15610278e-01 1.15500438e+00
-1.34813297e+00 -6.72724307e-01 -8.48710835e-01 6.35254323e-01
-7.52688646e-01 6.94867671e-02 -2.52793908e-01 7.63034761e-01
6.01412177e-01 1.13972569e+00 2.09414795e-01 -4.39515769e-01
3.76940727e-01 -1.95272207e-01 5.58510423e-01 -7.60151386e-01
-3.94457608e-01 3.67963426e-02 -2.28753179e-01 -4.26727116e-01
-2.51430899e-01 -6.14070714e-01 -1.32233608e+00 1.15014963e-01
-8.80826592e-01 -2.58836173e-03 8.82034540e-01 5.42214334e-01
2.50154883e-01 5.50376415e-01 4.08969909e-01 -5.97447157e-01
-3.66080225e-01 -1.33495772e+00 -4.15188730e-01 2.55484879e-01
1.27471760e-01 -7.02372730e-01 -1.67787284e-01 -1.30636375e-02] | [8.681130409240723, -0.5508219599723816] |
aa4d68ec-7a90-44f4-8ede-e556212872ef | cr-fiqa-face-image-quality-assessment-by | 2112.06592 | null | https://arxiv.org/abs/2112.06592v2 | https://arxiv.org/pdf/2112.06592v2.pdf | CR-FIQA: Face Image Quality Assessment by Learning Sample Relative Classifiability | The quality of face images significantly influences the performance of underlying face recognition algorithms. Face image quality assessment (FIQA) estimates the utility of the captured image in achieving reliable and accurate recognition performance. In this work, we propose a novel learning paradigm that learns internal network observations during the training process. Based on that, our proposed CR-FIQA uses this paradigm to estimate the face image quality of a sample by predicting its relative classifiability. This classifiability is measured based on the allocation of the training sample feature representation in angular space with respect to its class center and the nearest negative class center. We experimentally illustrate the correlation between the face image quality and the sample relative classifiability. As such property is only observable for the training dataset, we propose to learn this property from the training dataset and utilize it to predict the quality measure on unseen samples. This training is performed simultaneously while optimizing the class centers by an angular margin penalty-based softmax loss used for face recognition model training. Through extensive evaluation experiments on eight benchmarks and four face recognition models, we demonstrate the superiority of our proposed CR-FIQA over state-of-the-art (SOTA) FIQA algorithms. | ['Naser Damer', 'Biying Fu', 'Marcel Klemt', 'Meiling Fang', 'Fadi Boutros'] | 2021-12-13 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Boutros_CR-FIQA_Face_Image_Quality_Assessment_by_Learning_Sample_Relative_Classifiability_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Boutros_CR-FIQA_Face_Image_Quality_Assessment_by_Learning_Sample_Relative_Classifiability_CVPR_2023_paper.pdf | cvpr-2023-1 | ['face-quality-assessement', 'face-image-quality', 'face-image-quality-assessment'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.33560795e-01 -1.13676181e-02 -2.92546004e-01 -8.23817968e-01
-6.51516736e-01 -2.37605274e-01 3.95112127e-01 -3.86291176e-01
-2.79791057e-01 5.90575814e-01 -2.05116943e-01 1.22468218e-01
-4.14991468e-01 -7.47461736e-01 -8.30893755e-01 -9.21766341e-01
-1.15195535e-01 4.31246877e-01 -5.87430894e-01 3.68430525e-01
2.37587124e-01 9.62933600e-01 -1.80110717e+00 3.28369498e-01
6.88583553e-01 1.76164758e+00 -1.95752695e-01 5.65843880e-01
1.31419122e-01 6.63735390e-01 -7.02797294e-01 -4.48148221e-01
3.99197966e-01 -3.74271065e-01 -7.03950882e-01 2.85609305e-01
9.55707014e-01 -2.33945161e-01 -3.30333650e-01 1.01823843e+00
3.48969758e-01 -1.90747213e-02 8.94865215e-01 -1.44086456e+00
-7.77583957e-01 5.30301072e-02 -2.51450449e-01 2.87259191e-01
2.25621045e-01 1.87128291e-01 9.85374212e-01 -1.25968981e+00
4.11099881e-01 1.28643048e+00 3.38333637e-01 7.49969482e-01
-1.13134015e+00 -9.63458359e-01 -1.27885833e-01 3.84431183e-01
-1.48891592e+00 -9.62155282e-01 8.91706645e-01 -4.03557688e-01
4.88026947e-01 2.40609005e-01 3.54810745e-01 6.43685997e-01
3.77550386e-02 5.70670605e-01 1.07400274e+00 -3.69706213e-01
3.44145894e-01 9.76828188e-02 1.71608152e-03 1.09576690e+00
-1.54143302e-02 3.32354248e-01 -7.66331851e-01 5.26046529e-02
6.70987189e-01 -1.29207656e-01 -3.70191276e-01 -4.00348216e-01
-6.13390028e-01 5.92745960e-01 7.62743533e-01 1.57960221e-01
-3.85191470e-01 1.51593924e-01 -5.61624914e-02 5.09284794e-01
3.19886953e-01 2.63997853e-01 -2.33943358e-01 2.08783746e-01
-1.07392693e+00 -2.95502335e-01 5.91141939e-01 3.88021886e-01
8.35050285e-01 1.67251363e-01 -5.15916228e-01 9.20753837e-01
4.96700674e-01 8.64143848e-01 3.97850692e-01 -9.20030892e-01
2.69791663e-01 7.15715468e-01 -2.27294341e-02 -1.12232423e+00
-1.73912898e-01 -6.20745599e-01 -7.81839311e-01 5.32617331e-01
5.94488978e-01 1.31931812e-01 -1.03231585e+00 1.77808881e+00
3.53689522e-01 3.32242221e-01 2.56671626e-02 9.82906401e-01
7.02339411e-01 6.09544575e-01 -3.24016549e-02 -4.53038633e-01
1.11845326e+00 -5.40974081e-01 -6.21000707e-01 4.02255729e-02
4.33418900e-02 -5.66817224e-01 1.07718790e+00 5.52533567e-01
-1.02408361e+00 -8.52225900e-01 -1.26664400e+00 4.33441043e-01
-3.69676985e-02 7.75112987e-01 3.06616753e-01 9.62992787e-01
-9.47484136e-01 7.46333241e-01 -7.60768950e-01 1.52249098e-01
8.88058186e-01 6.47345543e-01 -6.21123374e-01 -1.41431466e-01
-7.25501120e-01 6.03452682e-01 -9.56262182e-03 3.31418127e-01
-1.25027502e+00 -7.04180062e-01 -6.18559301e-01 2.88912117e-01
2.64335424e-01 -3.67444485e-01 8.14222038e-01 -1.51777160e+00
-1.65643990e+00 9.01511252e-01 -2.64282316e-01 -2.65206426e-01
2.19247743e-01 7.34135509e-02 -5.83832979e-01 3.73423338e-01
-1.40648052e-01 5.91137111e-01 1.58171797e+00 -1.24612355e+00
-3.38289559e-01 -8.56470048e-01 -2.23505303e-01 -1.32586315e-01
-5.76421559e-01 -1.94462687e-01 -2.68684626e-01 -3.89467299e-01
1.11777380e-01 -5.66011727e-01 5.48925996e-01 4.95984435e-01
-5.66468984e-02 -4.34390396e-01 8.75932515e-01 -5.25840282e-01
9.70446467e-01 -2.19391060e+00 1.63511947e-01 4.19642270e-01
7.28615224e-02 3.10178608e-01 -4.96375114e-01 -1.41972467e-01
-4.63436358e-02 -1.29476622e-01 -1.00864962e-01 -3.19252998e-01
-6.41435534e-02 5.05995043e-02 -1.80096075e-01 7.81675220e-01
8.30483973e-01 7.01817870e-01 -5.22334933e-01 -3.27276766e-01
9.15669352e-02 7.76386321e-01 -5.01852453e-01 7.07825363e-01
3.55290771e-02 2.70253539e-01 -3.77896011e-01 8.29836071e-01
8.88640046e-01 -2.50716954e-01 9.76356715e-02 -5.98021984e-01
4.01977658e-01 -4.25658785e-02 -1.19222856e+00 1.25388873e+00
-6.40008569e-01 4.12870169e-01 5.57087269e-03 -1.09712934e+00
1.06539655e+00 2.33020544e-01 3.54964375e-01 -9.01018023e-01
1.65274754e-01 1.49474755e-01 5.51612973e-02 -4.23175871e-01
-7.16826692e-02 -1.13637999e-01 5.56618273e-01 3.85976285e-01
5.47365785e-01 2.82933533e-01 -1.49999866e-02 -2.65800178e-01
6.71734035e-01 -1.21696055e-01 2.91945226e-02 -3.62574637e-01
9.51799452e-01 -7.16938674e-01 5.26913226e-01 5.08691192e-01
-5.17955303e-01 3.77241284e-01 3.79611701e-01 -4.94058371e-01
-8.04705203e-01 -9.88847554e-01 -5.23307264e-01 1.02728903e+00
-6.73023090e-02 2.01096442e-02 -1.01268101e+00 -1.07550049e+00
7.02634454e-02 1.98617011e-01 -1.05467987e+00 -5.06514430e-01
-4.24201876e-01 -4.45580572e-01 4.59964663e-01 3.96172941e-01
4.81147617e-01 -1.21638012e+00 -4.22802240e-01 -2.91691005e-01
1.79564998e-01 -8.28275084e-01 -3.70096028e-01 -1.02085933e-01
-6.81033313e-01 -1.14543426e+00 -4.23246592e-01 -7.37207174e-01
9.31406498e-01 -7.83034042e-02 1.10829103e+00 4.00323123e-01
-4.70593810e-01 5.00243962e-01 -4.53877077e-02 -8.43762420e-03
-2.74227023e-01 -2.94920117e-01 2.41270930e-01 7.12159336e-01
2.56372958e-01 -2.59963274e-01 -7.95876503e-01 4.49624658e-01
-8.40840936e-01 -4.87142414e-01 5.09221137e-01 1.05988550e+00
6.38808370e-01 1.87504679e-01 7.37695873e-01 -4.73085344e-01
3.20997745e-01 -2.04572529e-01 -7.03688085e-01 5.71048081e-01
-6.64620817e-01 2.69486874e-01 6.18528485e-01 -2.48671561e-01
-9.29803848e-01 -1.51841687e-02 -4.87527102e-02 -5.60300827e-01
-1.39084876e-01 2.78076798e-01 -4.67969596e-01 -4.23582017e-01
5.46267092e-01 1.38884678e-01 1.74279392e-01 -1.50736645e-01
1.98544860e-01 6.58763170e-01 6.61648214e-01 -5.60600698e-01
7.01759934e-01 5.59634626e-01 3.00394714e-01 -8.17585528e-01
-8.01267862e-01 -1.92963779e-01 -5.00531435e-01 -5.09250462e-01
5.97643733e-01 -6.18525863e-01 -1.22641742e+00 4.78328794e-01
-7.98452914e-01 -5.63467741e-02 -8.94412547e-02 4.08631891e-01
-4.16192144e-01 7.70755634e-02 -2.96852410e-01 -1.03038764e+00
-5.78818619e-01 -1.31308270e+00 1.18325961e+00 2.57516503e-01
4.30231422e-01 -7.05817819e-01 -2.49387860e-01 5.12571931e-01
3.44055116e-01 -2.92007159e-03 7.38546968e-01 -3.17909896e-01
-5.71302354e-01 -2.97186941e-01 -3.86816055e-01 6.63204730e-01
3.15000027e-01 1.11193031e-01 -1.27291405e+00 -6.13553643e-01
9.02040824e-02 -6.27173245e-01 8.97643566e-01 4.12234694e-01
1.45790148e+00 -4.55584377e-01 2.80592591e-02 7.03101456e-01
1.41803098e+00 1.08430281e-01 6.78585231e-01 -1.40585184e-01
4.29336846e-01 5.22079647e-01 6.43746018e-01 3.25231791e-01
-1.04523681e-01 8.25502992e-01 6.14235520e-01 3.09689462e-01
-2.57849663e-01 -1.17431216e-01 4.63584930e-01 4.07370090e-01
7.89785907e-02 -1.22297250e-01 -5.97001076e-01 3.04404926e-02
-1.21407640e+00 -9.35605884e-01 6.36480153e-01 2.40947914e+00
7.64845669e-01 -9.31683555e-02 5.23202978e-02 3.87551188e-01
5.32410741e-01 -5.12424707e-02 -7.38545895e-01 -1.43692881e-01
-1.09990845e-02 4.13360834e-01 4.40550782e-02 5.04065871e-01
-1.06958687e+00 6.24417245e-01 5.79146767e+00 9.86834764e-01
-1.54024851e+00 -1.49649575e-01 1.12671733e+00 -9.54416841e-02
4.24022973e-02 -5.74417174e-01 -7.76625693e-01 3.57176721e-01
8.94271910e-01 1.10058255e-01 7.42701590e-01 7.08990335e-01
-1.13366675e-02 2.00181127e-01 -1.43633616e+00 1.17204726e+00
3.53619963e-01 -1.15377450e+00 2.56805271e-01 -5.62249720e-02
5.15147030e-01 -4.22870159e-01 6.50124311e-01 1.52004823e-01
-3.93022537e-01 -1.54068935e+00 6.96855247e-01 7.53727674e-01
1.08840001e+00 -7.94517756e-01 4.87308204e-01 2.18350142e-02
-1.00368869e+00 -4.29251611e-01 -6.00673497e-01 2.02748403e-01
-5.25879622e-01 6.53110206e-01 -8.75395834e-01 2.93629557e-01
6.37636662e-01 5.11622250e-01 -6.93819344e-01 7.75211334e-01
-8.02689791e-02 6.50355816e-01 -2.95881629e-02 2.50585765e-01
-1.33648947e-01 -1.88621357e-01 3.74335021e-01 7.50466585e-01
2.88489997e-01 1.74552985e-02 -2.23316196e-02 9.62031066e-01
-6.72621071e-01 6.89257607e-02 -1.26427114e-01 -1.48605093e-01
4.77885932e-01 1.30087364e+00 -4.22161996e-01 -1.70147464e-01
-1.91515144e-02 8.47793579e-01 5.80696762e-01 2.22000435e-01
-4.91838157e-01 -1.04344189e-01 7.16589272e-01 -3.33748013e-02
3.65323842e-01 1.92142814e-01 -1.06135085e-01 -9.45830941e-01
3.55634481e-01 -9.59402859e-01 4.65284944e-01 -4.98802215e-01
-1.23259223e+00 1.01140857e+00 -5.14263093e-01 -1.14972186e+00
-1.07048415e-01 -8.81174266e-01 -3.69755387e-01 8.18601489e-01
-1.64095628e+00 -9.95746076e-01 -4.23542202e-01 6.04296923e-01
3.15500796e-01 -4.79149967e-01 6.86647236e-01 3.52920175e-01
-7.07909703e-01 1.20360291e+00 -2.97278296e-02 2.70429373e-01
5.03625751e-01 -1.06293833e+00 -1.28217086e-01 6.88182950e-01
2.94584960e-01 4.44109976e-01 3.92823458e-01 -2.09954038e-01
-1.56067038e+00 -1.09314656e+00 4.16254640e-01 -4.12328571e-01
2.04265863e-01 -4.83467281e-02 -9.44568515e-01 1.82872757e-01
-1.94783792e-01 6.19358659e-01 5.98256350e-01 -1.74561650e-01
-7.32303858e-01 -8.33722830e-01 -1.52324200e+00 -1.42108006e-02
8.83765221e-01 -8.13324213e-01 -1.83067381e-01 1.01994418e-01
1.77624479e-01 9.85670388e-02 -9.57566321e-01 7.03562081e-01
8.51331413e-01 -8.64525139e-01 1.02338266e+00 -6.73671484e-01
4.37733442e-01 -3.52586567e-01 -2.91535974e-01 -1.21379781e+00
-3.00301611e-01 5.08096144e-02 -2.93872118e-01 1.03700531e+00
2.23029718e-01 -2.85974443e-01 9.71483946e-01 4.31583345e-01
2.43289173e-01 -8.94465327e-01 -1.32475030e+00 -6.98621690e-01
3.42957191e-02 -1.88128427e-01 7.45327055e-01 7.00111747e-01
-3.13206673e-01 8.65853205e-02 -3.66502792e-01 4.71954733e-01
9.10392523e-01 1.85934559e-01 3.75312895e-01 -1.25050735e+00
-2.51006126e-01 -4.36963499e-01 -7.37700522e-01 -6.59534752e-01
4.20674801e-01 -9.73189294e-01 -5.51051721e-02 -8.57342660e-01
2.87422091e-01 -5.70491433e-01 -5.65078795e-01 3.99935484e-01
-1.35100722e-01 6.61001623e-01 1.30483359e-01 2.29377225e-01
-6.42498195e-01 6.80717349e-01 1.32025874e+00 -3.82944852e-01
4.59538773e-02 6.63024411e-02 -3.08218598e-01 2.70574987e-01
5.03651142e-01 -3.16602409e-01 -2.45386183e-01 -3.96732122e-01
-1.45281656e-02 1.46541134e-01 4.52837408e-01 -1.27958512e+00
5.76505587e-02 -1.78658664e-01 9.18843925e-01 -1.03065874e-02
4.69966054e-01 -9.98360395e-01 4.53526713e-03 5.72259307e-01
-5.99879801e-01 -5.35035670e-01 4.84280810e-02 3.94188523e-01
-1.94365382e-01 -1.16673864e-01 1.12847412e+00 3.53834003e-01
-3.90907943e-01 5.98019421e-01 2.10150063e-01 -2.64304578e-01
8.06913555e-01 -1.83705658e-01 -2.13642105e-01 -1.93328857e-01
-7.70876944e-01 -5.71757443e-02 1.50999159e-01 3.28014433e-01
9.36903238e-01 -1.60409451e+00 -8.20834756e-01 8.17583323e-01
2.73523301e-01 -6.35414362e-01 1.63071662e-01 4.55468029e-01
-1.73598424e-01 4.40657616e-01 -5.55576563e-01 -6.73697710e-01
-1.42527330e+00 5.42254686e-01 7.83455193e-01 1.08621359e-01
9.61214155e-02 9.19929087e-01 1.17266282e-01 -2.40051761e-01
4.56400007e-01 -7.01555163e-02 -3.11872303e-01 -1.86912000e-01
8.30115497e-01 3.62791896e-01 3.63818854e-01 -8.50558460e-01
-3.83008093e-01 5.74841499e-01 3.38566652e-03 1.86200351e-01
1.16970170e+00 5.58100827e-02 -1.27010152e-01 6.28757328e-02
1.54035008e+00 -2.96259075e-01 -1.49232996e+00 -3.35309535e-01
-1.45637080e-01 -8.35692406e-01 3.73983949e-01 -1.06356645e+00
-1.63101459e+00 9.40435112e-01 1.28574765e+00 -1.54730037e-01
1.50006282e+00 -3.97132523e-03 2.98527718e-01 4.28133637e-01
2.14071214e-01 -9.23267484e-01 5.05638063e-01 -3.28968465e-02
1.18091261e+00 -1.43363285e+00 -2.03538075e-01 -3.14458221e-01
-2.85116673e-01 1.23785746e+00 6.58527792e-01 -1.17497124e-01
8.00177932e-01 -1.76698551e-01 4.22325730e-02 -3.32816869e-01
-6.69824123e-01 1.22956589e-01 8.57717454e-01 4.15585399e-01
2.73387015e-01 2.10704029e-01 2.43469160e-02 4.53948468e-01
-4.05146033e-02 1.15699254e-01 4.64839768e-03 4.61144835e-01
-4.96348023e-01 -8.56173992e-01 -3.74250799e-01 6.42845690e-01
-5.37973642e-01 2.66132623e-01 -2.20701605e-01 1.43779993e-01
1.19372897e-01 9.63481784e-01 2.14544758e-01 -3.54891956e-01
1.91496849e-01 1.56250736e-03 8.31859529e-01 -3.00928086e-01
-2.47559980e-01 -3.87569606e-01 -4.09782469e-01 -5.92391253e-01
-3.15929502e-01 -4.22261745e-01 -9.34268177e-01 -1.97947890e-01
-4.36412901e-01 1.75738230e-01 7.46035695e-01 9.34710503e-01
2.59113818e-01 2.76705235e-01 1.21951044e+00 -6.83933973e-01
-8.28291595e-01 -8.59026790e-01 -6.20914698e-01 4.78412002e-01
5.89146137e-01 -8.77572238e-01 -4.03906226e-01 -2.16010539e-03] | [13.103196144104004, 0.7006977796554565] |
9f75f2af-db50-4cf7-874f-602a83485e6a | document-level-event-extraction-via-human | 2202.03092 | null | https://arxiv.org/abs/2202.03092v1 | https://arxiv.org/pdf/2202.03092v1.pdf | Document-Level Event Extraction via Human-Like Reading Process | Document-level Event Extraction (DEE) is particularly tricky due to the two challenges it poses: scattering-arguments and multi-events. The first challenge means that arguments of one event record could reside in different sentences in the document, while the second one reflects one document may simultaneously contain multiple such event records. Motivated by humans' reading cognitive to extract information of interests, in this paper, we propose a method called HRE (Human Reading inspired Extractor for Document Events), where DEE is decomposed into these two iterative stages, rough reading and elaborate reading. Specifically, the first stage browses the document to detect the occurrence of events, and the second stage serves to extract specific event arguments. For each concrete event role, elaborate reading hierarchically works from sentences to characters to locate arguments across sentences, thus the scattering-arguments problem is tackled. Meanwhile, rough reading is explored in a multi-round manner to discover undetected events, thus the multi-events problem is handled. Experiment results show the superiority of HRE over prior competitive methods. | ['Jinqiao Shi', 'Yucheng Wang', 'Tingwen Liu', 'Bowen Yu', 'Xin Cong', 'Shiyao Cui'] | 2022-02-07 | null | null | null | null | ['document-level-event-extraction'] | ['natural-language-processing'] | [ 7.38499463e-01 2.41833434e-01 -2.57527009e-02 -1.69063464e-01
-1.09654105e+00 -4.28727061e-01 7.06981063e-01 5.93202889e-01
-5.88848591e-01 6.92498982e-01 5.81794143e-01 -1.81758076e-01
-1.56906620e-01 -7.98931062e-01 -4.20029551e-01 -5.79485655e-01
1.05322435e-01 3.89861763e-01 6.61569595e-01 -1.09868329e-02
4.85010833e-01 2.89308399e-01 -1.28579688e+00 7.00810969e-01
5.57620823e-01 8.12788188e-01 3.19839954e-01 4.58651185e-01
-5.28150797e-01 1.21576893e+00 -1.06067133e+00 -3.96766663e-01
-2.81896174e-01 -6.46660924e-01 -1.06335020e+00 2.03326255e-01
-2.18830436e-01 -1.86130106e-01 8.66566747e-02 1.13269234e+00
4.98727858e-01 1.52118295e-01 3.96578282e-01 -1.03879440e+00
-3.30041409e-01 1.13727176e+00 -8.91524255e-01 7.15530276e-01
5.89012802e-01 -4.68146950e-01 1.28456616e+00 -1.32931209e+00
5.35722256e-01 1.40827191e+00 2.34048948e-01 2.35242814e-01
-5.80129981e-01 -4.85103071e-01 5.19867003e-01 1.65659949e-01
-1.14889836e+00 -2.54579663e-01 7.71721125e-01 -1.16716295e-01
8.76455545e-01 4.98149633e-01 5.98893225e-01 1.11984444e+00
6.51698485e-02 1.31211579e+00 7.78306901e-01 -4.63319033e-01
3.91963899e-01 -1.56231165e-01 5.74394286e-01 4.37885672e-01
1.41095787e-01 -3.95620227e-01 -6.49583876e-01 -8.18444043e-02
2.39289656e-01 5.67403547e-02 -3.15432370e-01 5.53506017e-01
-1.45429254e+00 7.29529202e-01 -9.08712596e-02 3.49971771e-01
-6.38899624e-01 -4.39815938e-01 6.04948282e-01 4.17392105e-02
3.27809632e-01 1.48471504e-01 -5.66592932e-01 -1.74879014e-01
-8.07386935e-01 3.58095676e-01 7.90322542e-01 9.42829430e-01
3.60392302e-01 -6.96108222e-01 -5.25149882e-01 7.01290309e-01
3.33018333e-01 1.50137469e-01 4.66458857e-01 -1.26313701e-01
1.01023078e+00 7.71786690e-01 7.45768473e-02 -1.24368346e+00
-3.63687903e-01 -3.89441937e-01 -8.21041167e-01 -4.46455956e-01
2.15797752e-01 -2.17556149e-01 -6.58049107e-01 1.39006317e+00
6.27499044e-01 -7.47656748e-02 1.29471660e-01 7.05535173e-01
1.14734328e+00 1.07180774e+00 7.53608495e-02 -7.12263465e-01
1.98843300e+00 -9.28932250e-01 -1.10615814e+00 -4.75663722e-01
4.98335548e-02 -8.20763707e-01 8.43419254e-01 4.78417844e-01
-1.30641925e+00 -2.60756880e-01 -9.16424215e-01 -1.40671879e-01
-3.08872193e-01 2.37782776e-01 2.76102573e-01 6.59922957e-02
-9.12425071e-02 -1.03028864e-02 -5.25521994e-01 -1.23715840e-01
4.35511410e-01 5.05368412e-02 2.32937500e-01 2.35410929e-01
-1.60887706e+00 6.71277642e-01 8.38109255e-01 2.62104571e-01
-6.04252100e-01 -2.47696996e-01 -7.46314883e-01 1.88351065e-01
1.03295004e+00 -1.89933568e-01 1.34597290e+00 -5.10857344e-01
-1.12421083e+00 8.01371872e-01 -5.13734698e-01 -2.36449048e-01
3.83095026e-01 -3.99022907e-01 -7.11182415e-01 2.49439180e-01
4.22085345e-01 7.66422926e-03 9.33421314e-01 -8.86427879e-01
-1.03087294e+00 -4.03954864e-01 1.41820218e-02 2.47335017e-01
-1.86338305e-01 6.32890701e-01 -7.02823222e-01 -7.53059685e-01
3.39459181e-01 -3.53561342e-01 6.92854747e-02 -7.63223469e-01
-8.48123014e-01 -6.84847653e-01 8.14225614e-01 -7.19882548e-01
1.76371121e+00 -2.19962192e+00 1.27505794e-01 2.08342075e-01
4.48398203e-01 -1.06161810e-01 2.33979464e-01 5.62311172e-01
-2.52382785e-01 1.35100558e-01 -7.59396255e-02 -6.67093322e-02
-7.56119639e-02 1.17766120e-01 -7.16210365e-01 8.86565894e-02
4.19849217e-01 7.60528564e-01 -1.05944419e+00 -9.21812534e-01
-2.10149676e-01 5.55715635e-02 -3.80425006e-02 1.53649747e-01
-2.52627760e-01 2.21906751e-01 -1.06784344e+00 4.94548619e-01
7.36664951e-01 -6.41148627e-01 1.04259877e-02 -2.20472217e-01
-4.27220911e-01 7.34249055e-01 -1.55421138e+00 1.20336258e+00
-1.07099712e-01 2.43358061e-01 -1.26502916e-01 -8.21392119e-01
6.83626592e-01 3.15304577e-01 1.19527377e-01 -5.92779219e-01
2.36645237e-01 -1.48715684e-02 -2.90884584e-01 -5.90154588e-01
6.09976768e-01 2.18568183e-02 -3.45983624e-01 4.79837149e-01
-2.23174691e-01 4.33047205e-01 5.47080040e-01 5.36792338e-01
1.20320809e+00 -1.39352188e-01 7.09621966e-01 -1.73702475e-03
9.08611536e-01 -3.63900810e-02 8.31713617e-01 8.37336838e-01
-2.27453727e-02 4.82304901e-01 7.81326175e-01 -3.79396528e-01
-4.06263739e-01 -1.01396918e+00 8.15617070e-02 1.12583637e+00
6.76745534e-01 -8.24431956e-01 -6.67045534e-01 -9.67282176e-01
-6.03046119e-01 7.43011177e-01 -6.09271884e-01 1.45118505e-01
-9.80251193e-01 -9.48705673e-01 3.42691928e-01 5.02705455e-01
7.22966194e-01 -1.45117342e+00 -9.81006145e-01 5.02961993e-01
-6.34576082e-01 -1.01263869e+00 -6.02328122e-01 2.69793123e-01
-3.68357122e-01 -1.08032596e+00 -5.74221373e-01 -8.64273489e-01
6.34622753e-01 1.86262891e-01 1.07746589e+00 1.37785360e-01
-2.64593989e-01 -1.54197857e-01 -7.59826899e-01 -7.17982352e-01
-2.92039037e-01 -8.97651017e-02 -5.11112452e-01 1.73148528e-01
5.49043953e-01 -2.49499813e-01 -4.92999494e-01 2.59353936e-01
-1.04669237e+00 1.52085781e-01 7.76236832e-01 6.79961562e-01
7.52233982e-01 6.06236398e-01 5.87951362e-01 -1.03713763e+00
8.18970680e-01 -5.98292410e-01 -3.84934157e-01 5.73041320e-01
-1.61439627e-01 -1.26115426e-01 6.85857832e-01 -5.63840568e-01
-1.43083692e+00 -2.48815045e-01 -1.04415819e-01 2.80039161e-01
-2.26547673e-01 7.61854231e-01 -5.77233195e-01 1.02138031e+00
2.55944133e-01 5.23050487e-01 -6.94899678e-01 -3.54293078e-01
4.93724607e-02 6.16839051e-01 6.16597831e-01 -6.57586157e-01
6.67137504e-01 3.81704897e-01 -4.31788504e-01 -7.83339143e-01
-1.44288909e+00 -5.63572824e-01 -3.34077686e-01 -3.21358651e-01
1.15141022e+00 -6.82281077e-01 -5.46889305e-01 4.04129386e-01
-1.42574704e+00 3.41705978e-01 -3.15127760e-01 3.80813628e-01
-5.39588034e-02 3.55012596e-01 -6.71020627e-01 -9.04412687e-01
-3.45407218e-01 -9.19765711e-01 1.24359012e+00 5.65163612e-01
-3.46376121e-01 -5.16340315e-01 -1.77034318e-01 2.82483429e-01
-1.83845267e-01 1.71651110e-01 8.99983585e-01 -1.06879711e+00
-4.93033260e-01 -2.24665105e-01 -2.09561318e-01 -3.22258741e-01
1.56121999e-01 -6.31618276e-02 -5.82216442e-01 -2.02499591e-02
4.71404672e-01 -2.14140132e-01 6.81954503e-01 7.13010579e-02
1.05534923e+00 -3.62571001e-01 -4.23675179e-01 -1.17137693e-01
9.34639335e-01 5.10047674e-01 6.73982084e-01 5.65392852e-01
4.35855120e-01 5.95184803e-01 8.58593643e-01 7.20834732e-01
3.88874948e-01 2.86570996e-01 1.29454181e-01 -1.57596335e-01
1.74297407e-01 -3.73111695e-01 2.28389531e-01 8.77980471e-01
1.42875597e-01 -5.95607102e-01 -7.10117102e-01 4.92821813e-01
-1.98472977e+00 -1.17588854e+00 -4.25657213e-01 1.91790402e+00
8.60025764e-01 6.03888869e-01 2.65599396e-02 4.34545428e-01
9.98215377e-01 4.02983755e-01 -4.33575153e-01 9.92235262e-03
-2.41925493e-01 3.20579894e-02 -5.61326742e-02 -1.02143064e-01
-1.08670855e+00 7.21535563e-01 4.77475309e+00 1.06097698e+00
-5.62223196e-01 -1.95077136e-02 4.61193264e-01 1.09964654e-01
-1.48177952e-01 1.50748506e-01 -1.30203605e+00 5.92507958e-01
4.89417821e-01 -6.77165017e-02 -7.66574368e-02 3.62978965e-01
9.08463225e-02 -4.13342834e-01 -9.62770104e-01 7.40708768e-01
1.44154072e-01 -1.12739933e+00 1.80238843e-01 -1.13552876e-01
2.60280550e-01 -4.62595731e-01 -3.40925783e-01 3.22430253e-01
1.32365644e-01 -4.14851606e-01 1.08558416e+00 3.13369602e-01
1.35525987e-01 -9.38015103e-01 5.58476806e-01 8.07394445e-01
-1.64713776e+00 -1.46861392e-04 -1.77480757e-01 1.83176562e-01
6.10095501e-01 9.44538414e-01 -3.88481349e-01 5.59278190e-01
7.75075138e-01 5.95925570e-01 -3.97407025e-01 6.93746805e-01
-5.79902768e-01 6.95920289e-01 -2.67328560e-01 -4.23907012e-01
2.75436014e-01 -7.19408691e-02 8.49444807e-01 1.39563012e+00
2.85400134e-02 6.02842689e-01 3.43888044e-01 8.66905332e-01
-1.01402611e-01 1.91822588e-01 -3.25536653e-02 3.27242389e-02
5.11421323e-01 1.31714439e+00 -1.23211527e+00 -5.67394912e-01
-5.33612847e-01 1.07239723e+00 3.45471352e-01 1.31055415e-01
-1.10927594e+00 -7.40683436e-01 -1.82183251e-01 -2.23409101e-01
5.61552048e-01 1.53892756e-01 -2.48688050e-02 -1.29074287e+00
4.52314943e-01 -9.66409147e-01 8.45928729e-01 -6.69577479e-01
-1.16497922e+00 6.15817904e-01 4.81416769e-02 -1.13615549e+00
8.50205794e-02 -1.82502106e-01 -8.15965176e-01 6.88779235e-01
-1.32055330e+00 -8.21632147e-01 -1.28700405e-01 6.50385082e-01
1.16712511e+00 2.70242959e-01 4.24664021e-01 1.78850621e-01
-9.82076645e-01 3.04018319e-01 -3.69800776e-01 4.05798286e-01
3.23833615e-01 -1.25740123e+00 4.38713759e-01 1.18713713e+00
3.00084203e-01 7.40936041e-01 4.44298685e-01 -9.75054085e-01
-1.15687990e+00 -8.38697612e-01 1.43296361e+00 -2.51007050e-01
6.80497169e-01 -3.13192040e-01 -9.35313165e-01 5.27344406e-01
1.29330993e-01 -4.70023453e-01 7.02420235e-01 5.93877770e-02
-2.34956861e-01 2.28512272e-01 -7.09423244e-01 6.88352227e-01
8.08630884e-01 -3.94645691e-01 -1.34536207e+00 4.79056627e-01
7.09955931e-01 -5.07950664e-01 -3.61389399e-01 2.33420342e-01
1.49088144e-01 -7.55241752e-01 9.27419066e-01 -5.47206521e-01
7.95490563e-01 -4.93591636e-01 1.39166668e-01 -7.59601176e-01
-1.86483145e-01 -6.90294802e-01 -3.96006286e-01 1.71600032e+00
6.73977137e-01 -3.28667819e-01 1.83138251e-01 3.49513650e-01
4.76492979e-02 -7.68917918e-01 -5.49994707e-01 -4.74818081e-01
-4.99034286e-01 -4.25440490e-01 7.92363584e-01 7.91328788e-01
2.97956198e-01 9.16921437e-01 -2.47627631e-01 5.50082147e-01
4.19623047e-01 5.17957270e-01 3.13871711e-01 -1.12613499e+00
-4.09261197e-01 -3.30268711e-01 3.28017592e-01 -1.50335646e+00
-5.80447204e-02 -6.23279631e-01 4.88142490e-01 -1.56598997e+00
5.42399883e-01 8.85399282e-02 -2.59974331e-01 2.06257075e-01
-7.58187652e-01 -4.02109534e-01 -6.47163158e-03 3.61587822e-01
-1.06766272e+00 3.80741358e-01 1.18990541e+00 -2.39912346e-02
-2.97314376e-01 1.24831349e-01 -7.27146208e-01 1.12270308e+00
4.63876218e-01 -5.96197724e-01 -3.92378509e-01 -2.29695633e-01
6.35996580e-01 4.51460510e-01 1.70231193e-01 -7.11090505e-01
5.98842025e-01 -6.85545653e-02 4.95357275e-01 -1.28901887e+00
-1.17731258e-01 -5.13135672e-01 -3.09392273e-01 1.83254972e-01
-6.05743289e-01 2.61180133e-01 -5.89669542e-03 8.90106201e-01
-4.28714365e-01 -4.53504294e-01 4.14598525e-01 -1.65449843e-01
-6.54727697e-01 7.55547881e-02 -6.21042669e-01 6.10576749e-01
9.62321639e-01 -1.89953774e-01 -3.13894153e-01 -3.75525318e-02
-7.02633381e-01 3.64693254e-01 -3.64108860e-01 3.85971308e-01
8.17771554e-01 -1.03165698e+00 -8.19659352e-01 6.84347302e-02
1.03295162e-01 4.48853970e-01 1.78851649e-01 7.08212852e-01
-6.85170814e-02 -1.85012922e-01 5.34078777e-01 -3.24409097e-01
-1.41581023e+00 6.05984867e-01 -3.74362826e-01 -7.56666601e-01
-1.04981327e+00 1.11994135e+00 2.35345826e-01 1.13695771e-01
3.89364928e-01 -2.74460137e-01 -6.25475824e-01 5.46900094e-01
1.09745634e+00 3.33061576e-01 7.79860318e-02 -5.19473016e-01
-4.66880083e-01 3.65281999e-01 -6.78364038e-01 -1.86648443e-01
1.18489718e+00 -3.10000896e-01 -4.14095730e-01 3.66123080e-01
8.51295948e-01 3.19240063e-01 -9.75341499e-01 -3.86319280e-01
5.28830767e-01 -1.29869923e-01 -1.39731914e-01 -8.63256395e-01
-5.61624765e-01 5.42707622e-01 -1.59873679e-01 6.79621041e-01
1.32208967e+00 3.66847157e-01 1.18717682e+00 3.15065324e-01
1.30912796e-01 -1.14644098e+00 1.00515217e-01 8.70014250e-01
8.21520627e-01 -9.26214457e-01 2.90601049e-02 -6.93781793e-01
-7.03819752e-01 1.09062541e+00 4.72282112e-01 2.47443303e-01
5.03725469e-01 3.82559687e-01 -3.18551183e-01 -6.73567235e-01
-6.49843454e-01 -6.40178770e-02 3.14096183e-01 -2.21014753e-01
1.72210023e-01 -2.53435820e-01 -7.28111982e-01 1.01204550e+00
-1.18625507e-01 -2.50263780e-01 2.94505954e-01 1.27960742e+00
-5.17149150e-01 -8.96278143e-01 -4.31986272e-01 3.60444993e-01
-7.37309873e-01 -3.15460384e-01 -6.78987741e-01 6.73926115e-01
2.34251931e-01 1.21325529e+00 -2.46910118e-02 -1.30054550e-02
4.75014329e-01 -4.12305584e-03 3.39144692e-02 -6.43653393e-01
-7.48610854e-01 4.55535382e-01 2.49248654e-01 -4.31038588e-01
-4.36249077e-01 -9.00519967e-01 -1.80574405e+00 3.04705739e-01
-5.23476779e-01 4.67987418e-01 3.19836497e-01 1.30143797e+00
9.04967412e-02 8.23248327e-01 6.86005473e-01 -3.19972306e-01
-4.81977552e-01 -6.95714533e-01 -3.44946563e-01 3.87661725e-01
3.14043701e-01 -3.97045106e-01 -4.68041390e-01 1.65328175e-01] | [9.102408409118652, 9.204730987548828] |
cef131ea-148a-4a20-ba4c-a62c95764875 | device-depth-and-visual-concepts-aware | 2302.01540 | null | https://arxiv.org/abs/2302.01540v3 | https://arxiv.org/pdf/2302.01540v3.pdf | DEVICE: DEpth and VIsual ConcEpts Aware Transformer for TextCaps | Text-based image captioning is an important but under-explored task, aiming to generate descriptions containing visual objects and scene text. Recent studies have made encouraging progress, but they are still suffering from a lack of overall understanding of scenes and generating inaccurate captions. One possible reason is that current studies mainly focus on constructing the plane-level geometric relationship of scene text without depth information. This leads to insufficient scene text relational reasoning so that models may describe scene text inaccurately. The other possible reason is that existing methods fail to generate fine-grained descriptions of some visual objects. In addition, they may ignore essential visual objects, leading to the scene text belonging to these ignored objects not being utilized. To address the above issues, we propose a DEpth and VIsual ConcEpts Aware Transformer (DEVICE) for TextCaps. Concretely, to construct three-dimensional geometric relations, we introduce depth information and propose a depth-enhanced feature updating module to ameliorate OCR token features. To generate more precise and comprehensive captions, we introduce semantic features of detected visual object concepts as auxiliary information. Our DEVICE is capable of generalizing scenes more comprehensively and boosting the accuracy of described visual entities. Sufficient experiments demonstrate the effectiveness of our proposed DEVICE, which outperforms state-of-the-art models on the TextCaps test set. Our code will be publicly available. | ['Feng Shuang', 'Yi Cai', 'Qingbao Huang', 'Dongsheng Xu'] | 2023-02-03 | null | null | null | null | ['optical-character-recognition', 'relational-reasoning'] | ['computer-vision', 'natural-language-processing'] | [ 2.57663697e-01 1.09169982e-01 7.85803497e-02 -5.03042400e-01
-5.62449574e-01 -5.58956683e-01 8.70274007e-01 6.72077835e-02
-8.99952129e-02 5.89078903e-01 3.77149910e-01 -4.10499163e-02
2.19194621e-01 -8.04243565e-01 -9.25718307e-01 -3.42598855e-01
7.97413528e-01 4.87634420e-01 4.73600000e-01 -3.20040226e-01
3.46239954e-01 4.46148574e-01 -1.79238057e+00 6.51504338e-01
9.26897883e-01 7.62208641e-01 5.50312519e-01 3.07687938e-01
-8.37051511e-01 7.41044641e-01 -7.80402482e-01 -6.71983480e-01
-1.86856151e-01 -3.76816064e-01 -6.81429386e-01 4.25670326e-01
7.72760749e-01 -5.48388660e-01 -3.09367716e-01 1.14881992e+00
2.23210186e-01 2.45323125e-02 7.48288453e-01 -1.54793561e+00
-1.03243041e+00 4.52853799e-01 -4.51881289e-01 -1.83498740e-01
5.34371138e-01 2.81116609e-02 8.06342602e-01 -1.07001746e+00
6.20463610e-01 1.45591426e+00 2.75376439e-01 7.28645504e-01
-9.86504793e-01 -6.87167346e-01 4.45262671e-01 3.41475725e-01
-1.58982754e+00 -2.82853186e-01 1.05334556e+00 -3.32048327e-01
8.25092137e-01 3.88918340e-01 6.81071699e-01 1.24503124e+00
-1.30643517e-01 8.22849274e-01 8.65405798e-01 -3.29728752e-01
3.92937995e-02 5.19989550e-01 -1.36864454e-01 6.22964203e-01
4.71392334e-01 -3.06894004e-01 -3.57566386e-01 3.18907231e-01
7.95421958e-01 1.12782732e-01 -2.75046438e-01 -4.87452835e-01
-1.25683773e+00 5.87523460e-01 5.61649859e-01 3.59301448e-01
-2.24119499e-01 1.67205840e-01 2.54731387e-01 -3.95618081e-01
3.71519655e-01 4.03264165e-01 -7.58615360e-02 -4.61145304e-02
-7.18157649e-01 3.71477515e-01 3.17321062e-01 1.49866152e+00
8.80589545e-01 -3.70351709e-02 -4.44621801e-01 7.16840088e-01
2.88480103e-01 6.23768151e-01 1.95480213e-01 -6.65139496e-01
8.59015405e-01 1.01417685e+00 1.08510323e-01 -1.22919726e+00
-2.89270431e-01 -2.54175842e-01 -7.21655071e-01 -1.01589873e-01
1.43566042e-01 3.64230752e-01 -1.02441216e+00 1.50550389e+00
1.23573445e-01 -5.53967766e-02 1.02004349e-01 1.11092269e+00
1.20601547e+00 8.35180283e-01 4.04104412e-01 1.31126732e-01
1.59734201e+00 -1.08432269e+00 -1.05270207e+00 -4.78550375e-01
5.71126699e-01 -7.35716224e-01 1.36175847e+00 -5.68305654e-03
-8.78819466e-01 -7.31680453e-01 -8.60918224e-01 -4.53268677e-01
-8.07567537e-01 2.44103730e-01 6.64826810e-01 5.10580540e-01
-9.73495245e-01 1.61188915e-02 -4.53635573e-01 -5.19556105e-01
4.30704594e-01 -3.33668925e-02 -3.96090478e-01 -3.10030133e-01
-1.14376664e+00 9.30491745e-01 6.68320656e-01 8.58322233e-02
-6.63968503e-01 -6.16118908e-01 -1.13623285e+00 8.01343843e-02
2.97980756e-01 -9.19652820e-01 1.06688070e+00 -7.97126412e-01
-8.89246345e-01 7.90563643e-01 -2.49362633e-01 -5.65265305e-02
5.27036011e-01 -9.47394371e-02 -4.04443830e-01 2.63710648e-01
2.63432264e-01 1.16403937e+00 6.83938920e-01 -1.87464452e+00
-5.60703754e-01 -3.16648573e-01 2.98045695e-01 6.71155810e-01
-5.93682408e-01 -2.64348060e-01 -8.23251069e-01 -6.87868536e-01
2.70151556e-01 -6.44908071e-01 8.42966326e-03 2.21040621e-01
-6.00657403e-01 -1.59241050e-01 1.04277158e+00 -5.20088792e-01
1.11495376e+00 -2.08976078e+00 -1.26196831e-01 -3.81433994e-01
9.32375938e-02 1.98159799e-01 -1.66191727e-01 3.91775995e-01
6.72745407e-02 2.98017800e-01 -1.26910552e-01 -5.60674310e-01
1.18539464e-02 2.62252241e-01 -5.04594922e-01 -3.96018438e-02
3.81720722e-01 1.00437605e+00 -8.55569720e-01 -7.71498501e-01
7.92226434e-01 7.37546682e-01 -4.13017690e-01 2.15608522e-01
-6.22682571e-01 3.54025781e-01 -6.97408617e-01 6.25160038e-01
9.72785652e-01 -2.02637047e-01 -4.29551870e-01 -6.20947778e-01
-1.20334029e-01 8.54490101e-02 -9.57947433e-01 1.76754081e+00
-5.26157260e-01 7.78902888e-01 -5.65986037e-01 -6.36461258e-01
1.02883577e+00 8.99259895e-02 1.29057989e-01 -8.82689297e-01
1.26096323e-01 -5.59896044e-02 -6.12384021e-01 -7.72218645e-01
7.33245134e-01 -3.00745517e-02 3.02672517e-02 3.85265611e-02
-2.94052988e-01 -6.12067997e-01 1.20755106e-01 3.71618837e-01
6.26780391e-01 5.03443062e-01 5.64128347e-02 2.35594317e-01
5.63850462e-01 3.14174801e-01 7.71603137e-02 5.49776852e-01
-1.06644973e-01 9.92624223e-01 3.13039452e-01 -3.33138108e-01
-1.19138801e+00 -1.04633367e+00 -1.01094574e-01 8.22656572e-01
6.91656530e-01 -6.15152180e-01 -7.46874511e-01 -5.80235958e-01
-3.35410476e-01 1.20804727e+00 -6.59390330e-01 -3.68927307e-02
-2.53591657e-01 -4.18063134e-01 4.40037757e-01 5.58261156e-01
9.06546354e-01 -1.04346561e+00 -5.02053618e-01 5.24908975e-02
-6.53954625e-01 -1.59472394e+00 -3.22201133e-01 -3.74189556e-01
-7.25637317e-01 -8.42692554e-01 -8.78552675e-01 -9.37668741e-01
1.07582128e+00 6.47654653e-01 9.97021317e-01 -4.35694680e-02
-1.73478618e-01 3.15920144e-01 -5.72451413e-01 -5.03380537e-01
-5.02267540e-01 -1.50121406e-01 -3.53828698e-01 -5.37926070e-02
4.20482904e-01 -1.66463450e-01 -6.10417128e-01 3.08977991e-01
-1.12356877e+00 8.83826613e-01 5.45435309e-01 3.61506045e-01
5.28698325e-01 2.57786736e-02 2.36495301e-01 -7.26484358e-01
3.41014445e-01 -1.89154550e-01 -4.34663475e-01 3.74063700e-01
-2.40230531e-01 1.93236306e-01 7.66507983e-01 -3.29908818e-01
-1.37476397e+00 1.80980876e-01 -5.21764830e-02 -5.27420580e-01
-4.21614885e-01 7.96109810e-02 -3.97851169e-01 1.64185792e-01
4.98551965e-01 5.40039480e-01 -4.45556194e-01 -2.81051725e-01
4.90380168e-01 6.97214663e-01 4.51794744e-01 -6.12866700e-01
9.42390859e-01 5.96350074e-01 -3.00118532e-02 -8.43456089e-01
-9.91963387e-01 -4.56576884e-01 -5.07507443e-01 -3.11081767e-01
1.05823731e+00 -1.05766892e+00 -4.23056692e-01 2.27769539e-01
-1.65185928e+00 1.47571087e-01 1.00648694e-01 2.65630990e-01
-6.18933320e-01 3.96191746e-01 -1.53148994e-01 -7.48459280e-01
-1.29573539e-01 -1.08859515e+00 1.65798366e+00 2.77608424e-01
1.29292279e-01 -7.76631176e-01 -3.09218526e-01 5.74797094e-01
3.50554168e-01 3.48220587e-01 9.51645672e-01 -2.50584275e-01
-8.27721119e-01 -3.07714958e-02 -8.17988694e-01 2.28598282e-01
1.36909008e-01 1.07192621e-01 -1.04773164e+00 1.32786870e-01
-3.58490437e-01 -1.79721549e-01 6.36069834e-01 9.38349813e-02
1.40224457e+00 -3.05367023e-01 -5.36930501e-01 5.17301619e-01
1.56260049e+00 2.66090125e-01 7.95610964e-01 4.10441816e-01
1.08320737e+00 8.74207079e-01 7.78068960e-01 3.44036788e-01
7.74012983e-01 8.16635787e-01 6.62883997e-01 -3.92360240e-01
-4.55635875e-01 -6.67620718e-01 2.17980370e-02 4.68956500e-01
1.09813534e-01 -5.28873742e-01 -8.52091968e-01 6.95619345e-01
-1.75306571e+00 -9.89372969e-01 -4.51051742e-01 1.78054154e+00
5.84415853e-01 1.29571278e-03 -2.63195336e-01 5.57880895e-03
9.51313436e-01 -1.84385199e-02 -4.53562587e-01 -3.20575476e-01
-3.10161799e-01 -3.08615863e-01 2.92586714e-01 1.75104126e-01
-8.90097380e-01 1.27383065e+00 5.21700525e+00 7.07272410e-01
-8.78673375e-01 -1.22109629e-01 5.09779871e-01 2.76141912e-01
-6.67904377e-01 6.50472715e-02 -9.54687774e-01 2.38379702e-01
3.41381162e-01 -1.77028269e-01 1.32266834e-01 9.73591506e-01
2.62201130e-01 5.59432544e-02 -1.08086228e+00 1.32704771e+00
5.05208135e-01 -1.31565797e+00 8.04702401e-01 -2.32176647e-01
7.25480795e-01 -4.51596439e-01 4.51609008e-02 3.19889635e-01
-1.28708780e-01 -8.74589264e-01 1.05399036e+00 4.20678884e-01
1.00466394e+00 -6.29518807e-01 6.14345908e-01 1.84374437e-01
-1.25655699e+00 1.50702074e-01 -5.38227499e-01 4.79893871e-02
1.94180369e-01 3.31340253e-01 -9.97454762e-01 6.22658193e-01
6.78569615e-01 6.12225592e-01 -1.02686083e+00 1.02484047e+00
-2.88496822e-01 7.50801042e-02 -7.70894885e-02 -3.29829246e-01
3.70792955e-01 8.47329870e-02 2.99075156e-01 1.15197885e+00
4.53568518e-01 8.50436538e-02 -1.49616852e-01 1.31688499e+00
-2.62823589e-02 2.96036482e-01 -8.38582873e-01 -1.87377371e-02
4.36503291e-01 1.19211221e+00 -7.84591556e-01 -4.36407089e-01
-5.37930310e-01 1.27866018e+00 1.88344806e-01 5.08134067e-01
-1.20614803e+00 -4.34980899e-01 3.62975866e-01 2.98337638e-01
1.62719458e-01 -1.54519901e-01 -4.24117982e-01 -1.25412881e+00
1.95627704e-01 -4.81896937e-01 7.50947222e-02 -1.57816303e+00
-1.09937406e+00 7.07704663e-01 2.95354038e-01 -1.36893928e+00
6.05652519e-02 -6.35753989e-01 -2.28538454e-01 6.30189061e-01
-1.59212387e+00 -1.55943370e+00 -8.83371949e-01 6.24481738e-01
9.64280188e-01 2.19773903e-01 6.70437992e-01 1.94547504e-01
-3.01253051e-01 3.86991948e-01 -3.19748998e-01 7.09951797e-04
7.13497937e-01 -9.85244393e-01 4.53009069e-01 6.64095163e-01
8.74976292e-02 4.83919472e-01 9.09178257e-01 -6.39455557e-01
-1.22917366e+00 -1.37999129e+00 9.29676950e-01 -6.66104615e-01
2.77406633e-01 -6.67790294e-01 -9.85180616e-01 5.07989287e-01
5.11391945e-02 -1.58133581e-01 2.04858884e-01 -3.90447527e-01
-4.75806355e-01 7.07167527e-03 -1.00702965e+00 1.10928822e+00
1.23705637e+00 -4.43367839e-01 -7.26932883e-01 4.63319659e-01
9.79764283e-01 -4.06819195e-01 -3.24412376e-01 4.25549895e-01
2.86225438e-01 -1.03626835e+00 9.42503572e-01 -1.52643725e-01
5.95553875e-01 -6.11851931e-01 -1.71249494e-01 -1.02329838e+00
-5.66657484e-02 -4.89612482e-02 7.94635788e-02 1.50878644e+00
2.89348453e-01 -2.08383903e-01 7.43466377e-01 5.95821857e-01
-3.59416306e-01 -3.48859310e-01 -5.89003742e-01 -6.11771524e-01
-1.93647206e-01 -5.45153201e-01 6.85622334e-01 9.89610791e-01
-1.80052951e-01 4.93832707e-01 -1.80305019e-01 2.38294289e-01
4.49345142e-01 3.72105017e-02 8.38071585e-01 -9.54667807e-01
2.23478809e-01 -4.99273598e-01 -5.84902465e-01 -1.10471344e+00
1.61285937e-01 -6.74288034e-01 1.04440242e-01 -2.12855792e+00
4.57503319e-01 -3.47893715e-01 1.80226907e-01 4.65385646e-01
-2.13866740e-01 4.88283634e-01 2.05801025e-01 -6.64275931e-03
-7.36441970e-01 8.18256557e-01 1.71920347e+00 -3.30712020e-01
-6.29701093e-03 -4.84479576e-01 -8.37008774e-01 6.67122483e-01
6.91741109e-01 -2.64586538e-01 -6.03904605e-01 -7.37613142e-01
1.95266411e-01 -1.94890633e-01 7.09150970e-01 -1.20040214e+00
1.47752613e-01 -3.01852882e-01 6.34227157e-01 -1.02631903e+00
5.86277962e-01 -9.71196532e-01 1.14682905e-01 1.52082607e-01
-2.54727393e-01 4.13323268e-02 4.96973455e-01 4.54423815e-01
-3.28268021e-01 -1.82331696e-01 5.56896806e-01 -1.29177347e-01
-1.01480842e+00 2.26885855e-01 -6.56569228e-02 -1.38711408e-01
1.14525962e+00 -5.42940259e-01 -6.10129535e-01 -4.95315820e-01
-2.49914274e-01 1.36787757e-01 7.28071809e-01 8.63048673e-01
9.23978806e-01 -1.39331734e+00 -6.78159952e-01 1.20872386e-01
7.98747182e-01 1.84742227e-01 5.73280275e-01 1.81763738e-01
-7.58499563e-01 7.30131686e-01 -2.51448840e-01 -6.71405673e-01
-1.07276320e+00 8.11538994e-01 2.24561170e-01 2.55138755e-01
-7.02584445e-01 6.41100585e-01 8.11283529e-01 -1.42713234e-01
2.47495607e-01 -4.64911193e-01 -3.45291197e-01 -2.06355855e-01
5.88224471e-01 -1.31777604e-04 -4.51669730e-02 -8.72744143e-01
-3.70531350e-01 9.76783574e-01 -7.55390301e-02 -1.16405161e-02
8.33484948e-01 -4.40607071e-01 1.46458924e-01 2.80336261e-01
1.09800529e+00 -1.95046842e-01 -1.22900891e+00 2.20808461e-02
-4.05377060e-01 -5.40419996e-01 -7.60747641e-02 -7.55630612e-01
-6.53188884e-01 1.10294342e+00 4.15757120e-01 8.21956433e-03
1.16327083e+00 1.62112683e-01 6.42183185e-01 3.35632533e-01
3.12418133e-01 -9.08547163e-01 3.61147314e-01 4.14023846e-01
1.12327456e+00 -1.24640262e+00 -7.41538033e-02 -8.69585276e-01
-9.23520684e-01 1.17328811e+00 1.11360097e+00 3.67134154e-01
-7.61842504e-02 -7.60885254e-02 2.75025051e-02 -2.08059385e-01
-5.60314178e-01 -4.14395809e-01 4.11805540e-01 8.29555452e-01
3.25988859e-01 -1.70565665e-01 -5.04751988e-02 4.66996729e-01
-1.56532571e-01 -3.32742512e-01 6.75510466e-01 5.82556784e-01
-4.13364351e-01 -7.79256940e-01 -4.60668176e-01 6.38006479e-02
-1.14856280e-01 -2.39371523e-01 -4.43015844e-01 9.42763925e-01
2.83712447e-01 9.13032115e-01 3.08146834e-01 -2.70449102e-01
6.10859275e-01 -8.41416270e-02 4.92542565e-01 -6.43602908e-01
3.67102399e-03 -1.59512043e-01 -6.08180538e-02 -3.82602006e-01
-4.00765121e-01 -2.26704776e-01 -1.38507414e+00 -1.38882130e-01
-2.94029415e-01 -2.31949659e-03 1.02968669e+00 7.66571999e-01
2.39970803e-01 6.51189983e-01 3.90656024e-01 -7.21551180e-01
2.29652617e-02 -8.84485841e-01 -2.03542173e-01 7.66036630e-01
1.54164135e-01 -8.20335448e-01 -2.84942895e-01 2.05558464e-01] | [10.544636726379395, 1.3133841753005981] |
7d11dabe-0587-46f2-ab4b-f813cd6cb942 | bistatic-ofdm-based-joint-radar-communication | 2305.15058 | null | https://arxiv.org/abs/2305.15058v1 | https://arxiv.org/pdf/2305.15058v1.pdf | Bistatic OFDM-based Joint Radar-Communication: Synchronization, Data Communication and Sensing | This article introduces a bistatic joint radar-communication (RadCom) system based on orthogonal frequency-division multiplexing (OFDM). In this context, the adopted OFDM frame structure is described and system model encompassing time, frequency, and sampling synchronization mismatches between the transmitter and receiver of the bistatic system is outlined. Next, the signal processing approaches for synchronization and communication are discussed, and radar sensing processing approaches using either only pilots or a reconstructed OFDM frame based on the estimated receive communication data are presented. Finally, proof-of-concept measurement results are presented to validate the investigated system and a trade-off between frame size and the performance of the aforementioned processing steps is observed. | ['Benjamin Nuss', 'Thomas Zwick', 'Laurent Schmalen', 'Charlotte Muth', 'Axel Diewald', 'David Brunner', 'Lucas Giroto de Oliveira'] | 2023-05-24 | null | null | null | null | ['joint-radar-communication'] | ['robots'] | [ 7.08057404e-01 -1.66747540e-01 -3.12186014e-02 -1.40572667e-01
-3.55446935e-01 -4.69592690e-01 8.25139523e-01 -1.55941680e-01
-3.27443779e-01 9.64036047e-01 -2.08538294e-01 -2.74341494e-01
-5.90498686e-01 -5.21587729e-01 8.72728005e-02 -8.13934624e-01
-5.31045496e-01 -2.47889921e-01 -4.69745278e-01 1.32188186e-01
1.91112176e-01 5.99567175e-01 -1.20658422e+00 -4.67001498e-01
8.96197438e-01 1.45499015e+00 1.65007293e-01 7.92982757e-01
6.49313390e-01 8.83579031e-02 -1.22244513e+00 -3.99600230e-02
6.06758714e-01 -5.83362162e-01 4.34622824e-01 -6.97610453e-02
4.42237467e-01 -4.68291223e-01 -7.90758669e-01 8.64060104e-01
5.06114960e-01 -1.06062368e-01 7.38871574e-01 -7.63764679e-01
-3.83517109e-02 5.40907681e-01 -1.65086538e-01 1.24120757e-01
2.01968268e-01 -2.07879782e-01 3.50691915e-01 -6.86603129e-01
3.00692379e-01 6.93146646e-01 2.78167456e-01 -2.37157896e-01
-9.24267828e-01 -1.14428782e+00 -5.96080482e-01 1.46060392e-01
-1.34636462e+00 -4.59678084e-01 5.09007871e-01 -3.81159216e-01
5.87732911e-01 2.70164281e-01 9.98736322e-01 3.88244152e-01
9.90710616e-01 -3.36761653e-01 1.22150040e+00 -7.73290515e-01
9.46535543e-02 -2.27595299e-01 3.17353189e-01 2.36980468e-01
9.94917154e-01 1.09604836e+00 -4.58143830e-01 -3.16758752e-01
7.81711102e-01 -2.59634167e-01 -6.69027805e-01 -2.37386942e-01
-1.23972023e+00 6.41886890e-01 -1.02664597e-01 4.43719119e-01
-4.51855570e-01 1.37443036e-01 4.56719585e-02 7.56719530e-01
7.24437460e-02 2.29985103e-01 3.16423744e-01 1.17986746e-01
-1.37180793e+00 -1.89595930e-02 9.65542316e-01 1.02622044e+00
1.92187816e-01 1.11173773e+00 -1.00888230e-01 1.21774357e-02
4.43127960e-01 1.45771289e+00 1.33269399e-01 -6.73257172e-01
1.99780852e-01 -6.35258019e-01 2.27640718e-01 -8.86234462e-01
-2.30058566e-01 -1.16484439e+00 -7.96340346e-01 2.18615517e-01
-5.68217114e-02 -8.32394898e-01 -6.28316522e-01 1.05048716e+00
-3.52110080e-02 2.23003298e-01 7.89240837e-01 7.08316684e-01
2.12156534e-01 6.78742588e-01 -6.03045344e-01 -8.13292205e-01
1.51191902e+00 -2.74612159e-01 -1.38734472e+00 -3.10877234e-01
-1.74974158e-01 -1.09236395e+00 -6.82441235e-01 3.52337807e-01
-9.25476193e-01 -8.43772411e-01 -1.75440121e+00 9.57767904e-01
3.00167441e-01 5.66095412e-01 7.43006989e-02 1.12830532e+00
-4.79537904e-01 2.02093631e-01 -4.31936532e-01 9.22428444e-03
-4.90176231e-01 2.01666087e-01 9.18164104e-02 -6.44038245e-02
-1.39858115e+00 9.67431605e-01 2.09489003e-01 4.80235815e-01
-2.56917924e-01 -6.46764934e-01 -1.02196598e+00 -6.26923814e-02
-2.47113809e-01 -6.69995964e-01 1.11881900e+00 -5.92868805e-01
-1.65317822e+00 8.28672498e-02 5.69431782e-02 -1.27806664e+00
1.88076437e-01 -4.15851802e-01 -1.17897785e+00 5.14811635e-01
-1.04314819e-01 1.34049281e-02 1.44803345e+00 -6.22955382e-01
-6.92300200e-01 -9.60743204e-02 -4.61021930e-01 -1.48076750e-02
5.32460272e-01 -4.36349601e-01 4.84467655e-01 -7.94271767e-01
5.90518773e-01 -7.59009600e-01 -8.56102183e-02 -2.66811013e-01
7.24356323e-02 1.05907857e+00 9.57222700e-01 -3.47572654e-01
1.32824421e+00 -2.30555534e+00 -2.96071529e-01 5.17257988e-01
-3.23657691e-01 3.01271021e-01 3.35974157e-01 8.91308904e-01
-4.23523486e-02 -9.53562498e-01 -2.35146239e-01 4.38110679e-01
-2.39810199e-01 -1.99763015e-01 -8.19532514e-01 1.03527343e+00
-5.41569851e-02 8.20367187e-02 -2.97112614e-01 2.42593333e-01
5.18211186e-01 4.76703733e-01 -8.40092823e-02 -8.84469226e-02
2.99407482e-01 5.89493990e-01 -2.45932922e-01 7.66073644e-01
1.04842651e+00 4.76985514e-01 5.53649843e-01 -5.02768278e-01
-8.09891939e-01 1.54100642e-01 -1.16531897e+00 7.70840764e-01
-5.17761648e-01 1.07923663e+00 5.51620781e-01 -1.03476512e+00
1.34073639e+00 5.36859751e-01 4.51888889e-01 -8.03173542e-01
2.69784182e-01 5.17031372e-01 3.51471543e-01 -6.53779209e-02
5.58726192e-01 -1.10675864e-01 -2.35235065e-01 3.66614640e-01
4.27122861e-02 -1.65037751e-01 2.91813970e-01 -2.02295408e-01
8.79789114e-01 -1.16473772e-01 8.27026784e-01 -4.22808230e-01
8.93786073e-01 9.56822336e-02 4.13828433e-01 5.66537201e-01
1.17656654e-02 -2.05543831e-01 -2.77391911e-01 9.17226914e-03
-5.66838086e-01 -9.10791039e-01 -5.85099578e-01 -2.25118116e-01
5.14882743e-01 -1.61394715e-01 -1.23278901e-01 3.63092989e-01
3.05232406e-01 7.07133710e-01 1.50260538e-01 -2.28931345e-02
-5.94732106e-01 -3.54491025e-01 5.00073850e-01 -2.33590841e-01
7.03292072e-01 -1.67308584e-01 -1.23310196e+00 5.67274988e-01
3.35502297e-01 -1.46282089e+00 -3.94432573e-03 2.76256293e-01
-8.46855462e-01 -9.24623430e-01 -6.77250206e-01 -5.72285950e-01
5.65859914e-01 9.72537994e-01 3.06483954e-01 -6.68375373e-01
-3.13572437e-01 8.63658071e-01 -2.62624055e-01 -2.89700687e-01
-1.44511178e-01 -6.42320633e-01 1.54097229e-01 5.65097518e-02
7.37563223e-02 -6.10889077e-01 -4.29526091e-01 2.65537471e-01
-4.07266766e-01 -2.03561291e-01 1.04489267e+00 7.15606868e-01
-1.92102611e-01 1.55771405e-01 5.10006309e-01 -1.90299645e-01
3.83342445e-01 1.41799655e-02 -1.51252067e+00 -8.64816904e-02
-4.98834312e-01 -4.04640436e-02 3.64013702e-01 -6.34550303e-02
-9.60780919e-01 -4.12886329e-02 3.98451477e-01 1.67583767e-02
7.53649622e-02 7.46210039e-01 1.62256867e-01 -6.11729741e-01
3.68154109e-01 6.05134070e-01 3.61659646e-01 1.98289439e-01
1.64200425e-01 8.65599811e-01 8.18319440e-01 -1.52351856e-01
1.40126729e+00 5.62922478e-01 2.19295129e-01 -1.31570840e+00
-4.59728777e-01 -2.24149466e-01 -4.57127243e-01 -5.52644610e-01
4.78686154e-01 -1.33916509e+00 -3.52946252e-01 1.77897036e-01
-1.14973485e+00 4.78976667e-01 -7.43130893e-02 1.64182162e+00
-4.72733319e-01 3.91316026e-01 -8.33596587e-02 -1.19961309e+00
-4.78091210e-01 -6.31894171e-01 5.85327387e-01 1.31263986e-01
-1.86126292e-01 -5.85079789e-01 3.64610180e-02 1.96764916e-01
6.15243971e-01 4.02337134e-01 3.30230534e-01 -1.34755243e-02
-1.20603704e+00 -5.93121171e-01 -9.51715410e-02 -1.50837004e-01
6.81291223e-02 -4.05850768e-01 -4.41225886e-01 -6.59580529e-01
3.92581314e-01 4.47137982e-01 5.12825251e-01 7.55914271e-01
-2.17341021e-01 -6.60340488e-02 -5.39870799e-01 6.49319410e-01
1.64877295e+00 7.44421721e-01 4.79176104e-01 -5.30365556e-02
-4.03105944e-01 1.41666815e-01 1.21998847e+00 5.46332538e-01
-2.78547525e-01 7.11662531e-01 6.34040534e-02 6.23622894e-01
5.38719408e-02 4.32722211e-01 8.11704159e-01 7.05837786e-01
1.56429991e-01 -1.42572716e-01 -5.05431779e-02 -2.07241941e-02
-1.15682065e+00 -1.19575536e+00 -2.90025800e-01 2.42377639e+00
7.36121610e-02 2.16504082e-01 -5.61053157e-01 3.26580077e-01
7.17317879e-01 4.02588725e-01 -3.30422521e-02 -2.56198019e-01
-4.89940085e-02 6.62984371e-01 1.21788013e+00 7.85853505e-01
-1.07194018e+00 5.97837977e-02 6.11752462e+00 5.85055590e-01
-1.40567350e+00 -1.67083293e-01 -7.01422393e-01 3.63759726e-01
4.04922813e-02 2.74704903e-01 -9.35854733e-01 2.98575401e-01
1.25557315e+00 -8.44232619e-01 -1.03141174e-01 2.21176252e-01
3.16253453e-01 -6.22304857e-01 -5.49588025e-01 8.98227692e-01
2.03390867e-01 -1.28929973e+00 -1.58341259e-01 3.35652590e-01
3.72924089e-01 -4.06622142e-01 -1.57753215e-03 2.17997864e-01
-3.07044059e-01 -3.96988481e-01 7.74840534e-01 6.38723135e-01
9.02762353e-01 -5.12707114e-01 7.72857904e-01 2.40498677e-01
-1.47855544e+00 -2.28437051e-01 -9.44830626e-02 -4.61205184e-01
6.91711307e-01 1.16895175e+00 -6.39374197e-01 1.40146983e+00
-2.36763448e-01 3.77588868e-01 2.52317101e-01 1.25698936e+00
-7.96920434e-02 6.24888837e-01 -3.44691098e-01 3.44560295e-02
5.21660410e-02 -5.58202147e-01 1.11001301e+00 1.13584423e+00
1.27039981e+00 4.19410735e-01 2.13745963e-02 3.95875186e-01
7.85593212e-01 -5.56963742e-01 -8.87594759e-01 -8.21937900e-03
8.49441051e-01 9.89162445e-01 -1.24262787e-01 -1.33658499e-01
-4.80833411e-01 1.67536378e-01 -9.75122809e-01 3.12996686e-01
-9.28046346e-01 -8.16898286e-01 5.82046688e-01 2.16499269e-02
6.53721690e-01 -1.14543784e+00 -2.79134475e-02 -6.72688842e-01
-1.86392367e-01 -2.60091722e-01 -6.71081478e-03 -3.81193131e-01
-6.01601243e-01 3.77197832e-01 8.65201429e-02 -2.27580547e+00
-5.21755874e-01 -2.13142201e-01 -2.15287924e-01 8.90297115e-01
-1.62869191e+00 -6.30033672e-01 -2.16089174e-01 4.88307923e-02
1.96138948e-01 -7.31964827e-01 7.94538260e-01 2.30071947e-01
1.42367691e-01 1.32703409e-01 4.99322444e-01 -2.19412789e-01
4.70125526e-01 -4.56391960e-01 -1.05009146e-01 1.18330729e+00
-9.34626460e-02 5.27956903e-01 9.77583826e-01 -8.20362926e-01
-1.66922402e+00 -8.32630754e-01 8.16321731e-01 7.10614502e-01
5.81815481e-01 -7.24278390e-02 6.26440793e-02 2.72411555e-01
8.00652623e-01 -2.71569729e-01 8.09741795e-01 -7.28921115e-01
1.02683313e-01 -2.42266163e-01 -9.46958780e-01 3.73773128e-01
1.63001359e-01 -1.52162686e-01 -6.79114640e-01 -8.60027820e-02
7.39601552e-02 -6.36305034e-01 -8.38412464e-01 5.62797248e-01
9.83139575e-01 -8.73193681e-01 8.47451091e-01 6.16689980e-01
-3.78814310e-01 -7.87869751e-01 -6.27431810e-01 -1.13357699e+00
-2.22341090e-01 -1.02685702e+00 -1.56224638e-01 6.92413270e-01
2.95117823e-03 -9.03481245e-01 4.63027388e-01 -7.49591529e-01
-7.24508837e-02 1.88004859e-02 -1.43219507e+00 -1.13979113e+00
-7.37460732e-01 -1.46127194e-01 -1.10871904e-01 5.06907344e-01
2.19261423e-01 5.42697906e-01 -5.53098440e-01 9.41911697e-01
1.14423037e+00 6.72463179e-01 8.43900621e-01 -1.33169198e+00
-3.52299809e-01 1.35204643e-01 -6.27669632e-01 -1.37689078e+00
-1.12485223e-01 -4.81545001e-01 -2.66073614e-01 -1.00046694e+00
-8.02512825e-01 -5.55640236e-02 2.89027929e-01 -7.44818926e-01
5.54721475e-01 1.24578429e-02 1.81707650e-01 1.28271222e-01
2.87387699e-01 6.97926819e-01 9.26383972e-01 -1.15642203e-02
8.64575803e-02 7.73488641e-01 8.96916613e-02 1.36046886e-01
6.38732135e-01 -1.83695942e-01 -2.88980663e-01 1.03645489e-01
-4.57575411e-01 1.05247974e+00 4.86434728e-01 -1.83087385e+00
6.41065836e-01 2.67265916e-01 2.65983224e-01 -1.03889859e+00
8.12939584e-01 -1.47280145e+00 5.46002686e-01 1.39726913e+00
3.34315956e-01 -2.57704914e-01 9.48165283e-02 8.20525229e-01
-5.88775277e-01 -3.44480127e-02 9.19503927e-01 6.31739080e-01
-5.11351645e-01 -2.76475310e-01 -1.02120686e+00 -5.09819448e-01
1.31184840e+00 -6.49582207e-01 -4.61754978e-01 -6.40200555e-01
-6.95951343e-01 9.50085893e-02 -4.28327829e-01 -1.78375810e-01
5.95999897e-01 -1.27153659e+00 -6.67479038e-01 7.64816463e-01
-1.80871978e-01 -8.92653286e-01 3.48840445e-01 1.04021478e+00
-6.48916960e-01 1.28510702e+00 -4.34765011e-01 -6.76246285e-01
-1.43053973e+00 -2.27734238e-01 3.27920824e-01 1.13444244e-02
-6.29221201e-01 -9.45746154e-02 -3.06163371e-01 2.37896770e-01
1.03625935e-03 -8.88115838e-02 -1.16128005e-01 1.38343364e-01
5.40946841e-01 1.53348610e-01 -1.06101133e-01 -5.00918150e-01
-2.60849595e-01 7.49114275e-01 3.02135706e-01 -5.71227491e-01
7.23947108e-01 -3.44546467e-01 9.06111747e-02 1.52174234e-01
3.87922436e-01 3.48224550e-01 -9.76477742e-01 -4.01969075e-01
-2.08642274e-01 -4.85122621e-01 1.37907550e-01 -2.74377674e-01
-6.57036304e-01 4.04330850e-01 8.07410121e-01 2.59749442e-01
1.01460171e+00 -8.18906426e-01 6.56104684e-01 4.89935338e-01
9.13815618e-01 -7.43775427e-01 -4.52442884e-01 6.07611954e-01
5.68729281e-01 -2.38962829e-01 6.92328215e-01 -7.48168945e-01
5.57306735e-03 1.58292222e+00 -2.11310145e-02 -3.02405536e-01
9.01751637e-01 6.46063268e-01 1.65495127e-01 -1.19171469e-02
-6.35703325e-01 -3.33596796e-01 -3.59140299e-02 9.28970754e-01
3.69808525e-01 1.19433664e-02 -9.76222336e-01 3.11545432e-01
-5.36293745e-01 -2.17639819e-01 8.63370657e-01 1.19847274e+00
-1.04310179e+00 -9.51934040e-01 -1.18110800e+00 1.86744198e-01
-1.63583159e-01 -1.06087746e-03 1.11681513e-01 1.09312677e+00
-1.11474276e-01 9.29857731e-01 4.04015094e-01 -2.74079382e-01
4.41197962e-01 -2.45298177e-01 6.37887657e-01 -2.89844543e-01
-6.47485182e-02 4.45759356e-01 3.69844764e-01 -5.76286800e-02
-6.08642519e-01 -7.87574530e-01 -9.39534426e-01 -1.54991761e-01
-5.57730138e-01 7.06376612e-01 5.47585905e-01 7.52517819e-01
1.69556499e-01 6.70560241e-01 9.82147992e-01 -6.35483444e-01
-6.88074350e-01 -6.64029717e-01 -9.32346880e-01 -1.13800907e+00
6.43405676e-01 -5.61252773e-01 -3.83356959e-01 -2.36330479e-02] | [6.3840765953063965, 1.2439959049224854] |
ce726d0e-4a00-403a-8c02-dd4d3118e8b4 | fusionformer-exploiting-the-joint-motion | 2210.04006 | null | https://arxiv.org/abs/2210.04006v2 | https://arxiv.org/pdf/2210.04006v2.pdf | (Fusionformer):Exploiting the Joint Motion Synergy with Fusion Network Based On Transformer for 3D Human Pose Estimation | For the current 3D human pose estimation task, a group of methods mainly learn the rules of 2D-3D projection from spatial and temporal correlation. However, earlier methods model the global features of the entire body joint in the time domain, but ignore the motion trajectory of individual joint. The recent work [29] considers that there are differences in motion between different joints and deals with the temporal relationship of each joint separately. However, we found that different joints show the same movement trends under some specific actions. Therefore, our proposed Fusionformer method introduces a self-trajectory module and a mutual-trajectory module based on the spatio-temporal module .After that, the global spatio-temporal features and local joint trajectory features are fused through a linear network in a parallel manner. To eliminate the influence of bad 2D poses on 3D projections, finally we also introduce a pose refinement network to balance the consistency of 3D projections. In addition, we evaluate the proposed method on two benchmark datasets (Human3.6M, MPI-INF-3DHP). Comparing our method with the baseline method poseformer, the results show an improvement of 2.4% MPJPE and 4.3% P-MPJPE on the Human3.6M dataset, respectively. | ['Xiaohua Zhang', 'Xinwei Yu'] | 2022-10-08 | null | null | null | null | ['3d-human-pose-estimation'] | ['computer-vision'] | [-5.18623054e-01 -1.07358865e-01 -1.26566440e-01 -3.29496264e-01
-3.75409991e-01 -2.29869336e-01 5.29732049e-01 -2.98253804e-01
-5.56726515e-01 3.42613965e-01 5.29088497e-01 4.72565830e-01
-4.78774076e-03 -4.45412546e-01 -6.29810214e-01 -5.39906383e-01
-2.98579037e-01 6.19925797e-01 6.52796566e-01 -3.08553427e-01
-6.82336986e-02 3.67753774e-01 -9.97197807e-01 8.56330618e-02
4.21580315e-01 6.65542066e-01 2.41061067e-03 3.65604013e-01
3.02088708e-01 4.14271355e-01 -4.05314982e-01 -1.44595727e-01
4.39679831e-01 -2.85019249e-01 -5.18383086e-01 1.83968708e-01
3.79579276e-01 -2.98504740e-01 -5.78410208e-01 7.96829224e-01
8.71861994e-01 3.03124458e-01 4.95034873e-01 -1.20316076e+00
-1.60716474e-01 1.38524249e-01 -9.53808665e-01 -8.56365561e-02
5.22246361e-01 2.49323100e-01 6.05716109e-01 -1.03675044e+00
8.32145929e-01 1.54742205e+00 7.17426300e-01 4.28717166e-01
-9.96066630e-01 -5.86028218e-01 4.22590822e-01 3.10041189e-01
-1.48485470e+00 -1.37456775e-01 9.45156634e-01 -4.80593383e-01
7.93104470e-01 1.08476602e-01 9.78811264e-01 1.03517818e+00
8.17165196e-01 8.75093222e-01 8.52511108e-01 -7.19468072e-02
-2.42855728e-01 -4.44769025e-01 1.03839770e-01 5.57159007e-01
-1.26808122e-01 1.06009722e-01 -7.28099942e-01 -1.96868572e-02
7.34677672e-01 1.83826670e-01 -1.35572642e-01 -6.45326495e-01
-1.49137950e+00 3.79661888e-01 5.19596994e-01 1.07258134e-01
-2.86275864e-01 1.81498840e-01 4.41501051e-01 -3.78799513e-02
3.77124816e-01 -1.63316056e-01 -5.65637767e-01 -4.24709693e-02
-8.09459209e-01 6.70661926e-01 5.66774964e-01 9.89863694e-01
4.39095974e-01 -3.35528165e-01 -3.19090039e-01 6.20419204e-01
6.26716733e-01 5.49920976e-01 3.44373375e-01 -9.76479650e-01
7.34454274e-01 5.31817138e-01 1.82747483e-01 -1.31503284e+00
-9.91310120e-01 -5.93320847e-01 -1.00072277e+00 3.89076658e-02
4.36667353e-01 -9.99806821e-02 -9.18500066e-01 1.92561781e+00
6.74493074e-01 -8.51155743e-02 -4.08356845e-01 1.32017136e+00
5.38535833e-01 5.54177880e-01 -1.12372957e-01 -1.22260608e-01
1.32334900e+00 -1.27392626e+00 -8.03490818e-01 -1.97158813e-01
4.41134661e-01 -7.86571622e-01 6.63798571e-01 2.91559577e-01
-1.03221631e+00 -8.72637153e-01 -9.19326782e-01 -1.51074886e-01
6.69830367e-02 3.03715646e-01 2.55074978e-01 1.42615378e-01
-6.69297516e-01 5.85991263e-01 -1.20709634e+00 -3.40219796e-01
-1.84380636e-01 4.02549952e-01 -6.08680367e-01 6.77979812e-02
-1.27122319e+00 1.06689119e+00 2.41315231e-01 5.88286877e-01
-5.31642735e-01 -4.22955185e-01 -6.32744491e-01 -4.60959077e-01
4.56739634e-01 -9.47910130e-01 8.88502896e-01 -2.44619980e-01
-1.60307169e+00 5.61409712e-01 -1.79735899e-01 -9.99593288e-02
8.66679192e-01 -8.76753926e-01 -1.94990978e-01 -6.40200675e-02
7.48122111e-02 7.44769096e-01 4.49456125e-01 -1.06264472e+00
-5.02447903e-01 -7.34701335e-01 -1.44795135e-01 7.37930357e-01
3.06476895e-02 -2.29020014e-01 -1.20598590e+00 -6.90362990e-01
5.63099623e-01 -1.42684996e+00 -4.26262826e-01 2.35083222e-01
-5.66775620e-01 -1.70476586e-01 7.04804063e-01 -1.00773680e+00
1.26021159e+00 -2.05243301e+00 8.19245994e-01 2.04462081e-01
-3.11989570e-04 -5.91885783e-02 7.60124251e-02 3.53477001e-01
1.13710083e-01 -3.08229685e-01 -2.53250953e-02 -6.54210150e-01
-5.63782267e-02 2.60312408e-01 1.68231562e-01 8.07502866e-01
-1.27469286e-01 8.11826229e-01 -6.03326678e-01 -6.93747461e-01
2.51458764e-01 5.97709894e-01 -5.49079478e-01 1.24538951e-01
1.49746403e-01 8.47778320e-01 -4.55438852e-01 4.30171907e-01
6.15918815e-01 -6.70865411e-03 2.28114873e-01 -5.41021287e-01
-7.35577419e-02 1.96837753e-01 -1.48710489e+00 2.23959851e+00
-2.22902000e-02 -7.52742514e-02 -3.26292738e-02 -5.92575967e-01
7.65423059e-01 4.44456190e-01 9.82176065e-01 -5.35380125e-01
1.40934423e-01 9.18160528e-02 9.14992839e-02 -4.00807321e-01
3.32635164e-01 1.14494398e-01 -1.79650798e-01 1.92454398e-01
-3.91521398e-03 1.98113486e-01 -4.10989895e-02 -3.68661396e-02
8.84827375e-01 7.35175908e-01 2.38526538e-01 -1.68211371e-01
4.74100620e-01 -1.87043056e-01 8.76622200e-01 2.23601460e-01
-4.10712570e-01 1.02623439e+00 2.19506189e-01 -4.67897207e-01
-9.14327443e-01 -1.22495210e+00 1.70015290e-01 6.81520343e-01
4.88051027e-01 -6.04677796e-01 -5.65646291e-01 -6.17235780e-01
3.46640088e-02 1.39606044e-01 -5.20817339e-01 -2.52056062e-01
-1.19713318e+00 -8.34838808e-01 3.36040765e-01 6.42521143e-01
5.63476086e-01 -7.44546175e-01 -6.93279743e-01 2.24687740e-01
-5.14205992e-01 -1.13170815e+00 -9.01198089e-01 -1.86316803e-01
-1.01804483e+00 -9.12067175e-01 -9.88086343e-01 -5.92347026e-01
4.73840415e-01 6.48776665e-02 7.68364012e-01 -2.29412243e-01
-2.25012209e-02 1.50234565e-01 -2.77435243e-01 1.00306235e-01
1.79782093e-01 8.13824311e-02 3.61186743e-01 -1.61828682e-01
2.34432090e-02 -7.57252097e-01 -7.90345192e-01 7.62009919e-01
-2.47432575e-01 2.73604065e-01 5.16160846e-01 7.16245055e-01
6.65664554e-01 -1.19129285e-01 -3.96787338e-02 -2.36597091e-01
2.13555172e-01 -2.46035323e-01 -6.33386970e-02 1.61685482e-01
-2.32992709e-01 -2.01554410e-02 1.80261314e-01 -6.06781483e-01
-1.15400767e+00 4.24767137e-01 -1.43045202e-01 -5.53624809e-01
4.05643694e-02 3.08972031e-01 -5.64184368e-01 2.77439088e-01
3.65221679e-01 1.38091952e-01 1.15784153e-01 -7.79009640e-01
3.09621215e-01 5.06980643e-02 5.67666888e-01 -5.48293471e-01
8.84127915e-01 6.21163964e-01 1.99343085e-01 -4.08428222e-01
-5.19618452e-01 -5.43125808e-01 -1.07316971e+00 -3.48091632e-01
9.98713791e-01 -1.07568514e+00 -6.27744913e-01 6.59860969e-01
-1.37325931e+00 -1.50884882e-01 -3.10187358e-02 8.29961717e-01
-6.09674752e-01 7.06878304e-01 -6.71513736e-01 -5.56317329e-01
-3.46180648e-01 -1.20128453e+00 1.26785409e+00 -1.78971291e-01
-5.37426591e-01 -4.50486094e-01 3.74510258e-01 1.29166007e-01
-1.51017159e-01 4.43814844e-01 4.90871012e-01 -1.68144345e-01
-3.48425597e-01 -1.65269524e-01 1.71829760e-01 5.39816357e-02
1.26330048e-01 -2.64676839e-01 -3.53851557e-01 -2.89361209e-01
4.05715182e-02 9.54864398e-02 7.51389802e-01 5.05416930e-01
5.97300887e-01 -9.65110362e-02 -5.02163589e-01 6.08984113e-01
9.89676833e-01 6.52437145e-03 6.23305857e-01 2.67386824e-01
1.01702833e+00 9.98043358e-01 8.62513602e-01 2.98392862e-01
5.85913062e-01 1.28474820e+00 3.28203589e-01 1.97413564e-01
-2.37745866e-01 -3.62733185e-01 6.02417946e-01 1.14635086e+00
-6.68552101e-01 -8.98679718e-02 -7.49651611e-01 4.10819799e-01
-2.27501988e+00 -6.47599757e-01 -2.31381923e-01 2.10698056e+00
6.25118494e-01 2.31460720e-01 3.44881535e-01 -6.58828691e-02
6.32728934e-01 2.87325978e-01 -4.66350526e-01 2.28341669e-01
4.16327529e-02 -2.71389812e-01 3.73698473e-01 3.59813809e-01
-1.11866248e+00 6.48950756e-01 5.83127308e+00 6.07731879e-01
-8.64092410e-01 1.85809761e-01 8.93514231e-02 -4.53376234e-01
8.53611231e-02 -4.03545760e-02 -8.80741656e-01 5.01599729e-01
3.81259084e-01 2.94191331e-01 -1.19386904e-01 5.15151143e-01
3.29236031e-01 -8.82654339e-02 -1.04743898e+00 8.48632097e-01
-3.21957469e-02 -6.16467655e-01 -1.28422722e-01 1.12907395e-01
4.79207098e-01 -1.44374976e-02 -9.19959918e-02 3.13786231e-02
-1.80493057e-01 -6.87535107e-01 1.03445256e+00 9.21604514e-01
1.59702674e-01 -7.55821049e-01 6.73718512e-01 6.83764160e-01
-1.53940225e+00 1.56050801e-01 -4.86426167e-02 -6.33321144e-03
7.35074997e-01 6.97779715e-01 -1.90441594e-01 8.69050860e-01
9.28640842e-01 9.05787528e-01 -5.01051188e-01 9.17789400e-01
-3.78077537e-01 7.67574906e-02 -5.86736798e-01 3.11198294e-01
-1.24029137e-01 -1.35010764e-01 8.32084954e-01 9.00786281e-01
2.98143506e-01 3.64984758e-02 3.84692580e-01 5.59969842e-01
4.22207266e-01 -3.09831393e-03 -1.88901350e-01 5.26198089e-01
7.25191757e-02 1.05325615e+00 -4.97205824e-01 -8.06249827e-02
-2.90118217e-01 1.17953885e+00 1.35347873e-01 1.99989408e-01
-1.01466465e+00 1.58762485e-01 5.25458992e-01 2.49986604e-01
1.46387607e-01 -7.71077991e-01 -1.41322494e-01 -1.27354991e+00
5.98717868e-01 -7.31659710e-01 3.27325284e-01 -5.70860803e-01
-1.15609598e+00 2.03538224e-01 4.79459137e-01 -1.44839799e+00
-3.60756904e-01 -3.83931577e-01 -4.06250566e-01 7.24637270e-01
-8.09153259e-01 -1.19998991e+00 -8.72861370e-02 5.20300269e-01
4.67875034e-01 2.57222354e-01 4.58264291e-01 3.78796846e-01
-5.00529349e-01 4.32856202e-01 -3.18046123e-01 1.74913436e-01
1.03735089e+00 -9.53834355e-01 6.10341907e-01 6.82102919e-01
4.22621779e-02 7.84523010e-01 6.87525451e-01 -9.79012966e-01
-1.35326457e+00 -9.05910552e-01 1.13198531e+00 -6.01640642e-01
1.65136635e-01 -3.92064512e-01 -8.00371885e-01 6.55710220e-01
2.13654339e-02 1.46975279e-01 7.85784796e-02 1.68265179e-01
-1.13263234e-01 -7.74513483e-02 -8.09418440e-01 6.71892107e-01
1.49855185e+00 -1.09922580e-01 -7.56413221e-01 1.96095541e-01
7.38153875e-01 -7.67527223e-01 -1.04121614e+00 7.31966138e-01
1.08627415e+00 -9.12404478e-01 1.23583472e+00 -3.12871397e-01
2.13268101e-01 -6.85867250e-01 -1.29649475e-01 -1.11371529e+00
-4.42276895e-01 -4.37522382e-01 -3.11159462e-01 8.57968330e-01
-3.75531800e-02 -2.02564403e-01 8.24881196e-01 3.11894894e-01
-4.09899838e-02 -8.86402726e-01 -1.15934134e+00 -8.81332338e-01
-7.89346620e-02 -3.80748630e-01 3.67500246e-01 5.22405207e-01
-7.70508051e-02 3.12159717e-01 -9.53695476e-01 1.35436222e-01
6.54487848e-01 6.72733560e-02 1.07727301e+00 -9.72333848e-01
-4.82158154e-01 -1.30031526e-01 -5.93230486e-01 -1.44358003e+00
-1.33173823e-01 -5.57768464e-01 1.70599803e-01 -1.31504774e+00
3.32559764e-01 5.79928420e-03 -2.65656322e-01 2.58329421e-01
-2.30374023e-01 2.57886350e-02 4.72170204e-01 4.10687327e-01
-5.15954852e-01 7.40613341e-01 1.60671532e+00 3.05393618e-02
-3.24603707e-01 3.11936550e-02 6.76695630e-02 8.77898693e-01
5.70219278e-01 -5.25261462e-01 -3.46075058e-01 -4.41407382e-01
2.63800602e-02 1.07815079e-01 6.54419780e-01 -1.24497402e+00
4.55018908e-01 -1.53004214e-01 6.49086833e-01 -1.20202732e+00
7.00014710e-01 -7.48367667e-01 5.69441855e-01 6.81857288e-01
-1.80637147e-02 2.52720594e-01 -1.10138483e-01 7.06943452e-01
-2.11895570e-01 4.14558947e-01 5.25347948e-01 -2.28037834e-01
-6.33314312e-01 5.60815930e-01 -1.27822697e-01 -9.64439586e-02
8.39576304e-01 -2.74545372e-01 1.61788851e-01 -2.99076855e-01
-1.00797272e+00 4.17681992e-01 3.52241755e-01 6.82078481e-01
6.31554484e-01 -1.68898857e+00 -5.21468461e-01 1.65049490e-02
-7.90412873e-02 3.30776215e-01 5.50791621e-01 1.35352719e+00
-1.69686273e-01 3.46980721e-01 -4.33104932e-01 -9.51341629e-01
-1.22269166e+00 3.29978794e-01 3.52877975e-01 -5.14682293e-01
-8.05301487e-01 6.73681557e-01 3.76167744e-01 -6.95076704e-01
1.95131883e-01 -2.93117225e-01 -1.56847388e-01 -9.55851376e-02
1.80655897e-01 6.59337223e-01 -2.26771668e-01 -1.09441221e+00
-7.31640160e-01 1.14288020e+00 1.69183195e-01 -3.40529501e-01
1.26133156e+00 -2.49433979e-01 -1.02986269e-01 5.46399772e-01
1.39983821e+00 -8.40837955e-02 -1.39617217e+00 -2.65592068e-01
-2.19919235e-01 -3.52736354e-01 -4.51493710e-01 -4.68197703e-01
-1.21501732e+00 8.63377213e-01 7.20877171e-01 -5.56658208e-01
1.04228222e+00 -3.95096354e-02 9.48546648e-01 1.49337023e-01
5.93647003e-01 -1.20492232e+00 2.19314009e-01 6.02992892e-01
1.23921156e+00 -9.29477990e-01 5.42327821e-01 -6.42481387e-01
-5.58864415e-01 9.11819339e-01 8.03584874e-01 -2.56382257e-01
8.79169345e-01 8.19076225e-02 -1.05626740e-01 -1.26292467e-01
-4.86052573e-01 1.01370223e-01 7.93983877e-01 4.06761110e-01
4.32830453e-01 9.44853798e-02 -7.97710836e-01 6.19813323e-01
-1.03283182e-01 -4.31418195e-02 -1.25980064e-01 1.12656355e+00
-2.04379991e-01 -1.28010988e+00 -5.13606966e-01 -1.36872262e-01
-1.63207874e-01 4.17089671e-01 -2.62271613e-01 1.15113449e+00
4.60964710e-01 5.52243710e-01 -2.65063588e-02 -8.70903194e-01
7.38723814e-01 7.00220391e-02 6.86393797e-01 -4.13040042e-01
-5.42863369e-01 6.68945611e-01 5.59076183e-02 -1.07358527e+00
-7.33461678e-01 -9.03583705e-01 -1.45993483e+00 -6.38518333e-02
-1.33754820e-01 -2.73391157e-01 3.66614759e-01 8.45879138e-01
3.54869366e-01 4.49148238e-01 2.14276090e-01 -1.04759741e+00
-5.80681205e-01 -1.08193040e+00 -5.71235597e-01 5.37168384e-01
8.59073177e-02 -1.03137994e+00 -1.06726453e-01 -5.32457568e-02] | [7.125755310058594, -0.6581653952598572] |
c6eadc0b-3de1-4d26-a363-d3c4664f47f5 | aspect-extraction-using-coreference | null | null | https://aclanthology.org/2020.aacl-srw.18 | https://aclanthology.org/2020.aacl-srw.18.pdf | Aspect Extraction Using Coreference Resolution and Unsupervised Filtering | Aspect extraction is a widely researched field of natural language processing in which aspects are identified from the text as a means for information. For example, in aspect-based sentiment analysis (ABSA), aspects need to be first identified. Previous studies have introduced various approaches to increasing accuracy, although leaving room for further improvement. In a practical situation where the examined dataset is lacking labels, to fine-tune the process a novel unsupervised approach is proposed, combining a lexical rule-based approach with coreference resolution. The model increases accuracy through the recognition and removal of coreferring aspects. Experimental evaluations are performed on two benchmark datasets, demonstrating the greater performance of our approach to extracting coherent aspects through outperforming the baseline approaches. | ['Wei Emma Zhang', 'Deon Mai'] | 2020-12-01 | null | null | null | asian-chapter-of-the-association-for | ['aspect-extraction'] | ['natural-language-processing'] | [ 5.01178026e-01 4.56999242e-01 -5.32773495e-01 -5.93702316e-01
-8.74027550e-01 -7.03572512e-01 9.46900904e-01 5.73012054e-01
-4.61795181e-01 7.22135723e-01 6.13833308e-01 -1.43308759e-01
-1.36458009e-01 -7.53034353e-01 -1.40897438e-01 -5.08307159e-01
2.82348037e-01 6.02154613e-01 1.56915888e-01 -3.15833926e-01
6.81099474e-01 2.95443863e-01 -1.66011095e+00 4.90502775e-01
5.43290436e-01 5.72965801e-01 -2.78585732e-01 3.33114237e-01
-8.42165530e-01 6.44887567e-01 -7.27051198e-01 -5.54768443e-01
1.22962251e-01 -4.20523077e-01 -1.15238357e+00 4.14562970e-01
3.03621683e-02 3.10433716e-01 7.15984344e-01 9.93240893e-01
1.59422204e-01 7.16129690e-02 5.29351473e-01 -9.87041235e-01
-2.97082830e-02 8.31353426e-01 -6.78145468e-01 -4.92151380e-02
9.39590454e-01 -4.01630998e-01 1.50013196e+00 -9.27358508e-01
7.06691504e-01 1.17605972e+00 4.44770813e-01 2.87695676e-01
-1.04864585e+00 -4.90410805e-01 3.06481689e-01 1.45700686e-02
-1.07375896e+00 -3.84979576e-01 9.70032692e-01 -1.19531594e-01
1.43246615e+00 3.77152652e-01 7.84434438e-01 5.84998369e-01
3.59669551e-02 8.42441142e-01 1.28276300e+00 -9.86477315e-01
3.32966596e-01 6.44894302e-01 6.04772329e-01 9.72348228e-02
6.65566206e-01 -2.80464739e-01 -8.93213153e-01 -4.81740147e-01
3.84563655e-02 -3.96816492e-01 -3.48671898e-02 -4.14997548e-01
-1.01536131e+00 8.14752042e-01 -3.25071692e-01 6.73725486e-01
-5.70987403e-01 -7.20237494e-01 5.14976919e-01 3.60204875e-01
5.22972107e-01 8.63631666e-01 -6.52424514e-01 -2.88251072e-01
-1.11999619e+00 2.87243038e-01 1.31609809e+00 1.07670927e+00
6.71455085e-01 -4.09875542e-01 -1.28576443e-01 7.10482061e-01
4.16389644e-01 1.96604446e-01 4.73469317e-01 -6.33911848e-01
2.84439415e-01 1.48170149e+00 5.17307632e-02 -8.55211556e-01
-4.00260508e-01 -2.72150397e-01 -3.63143504e-01 4.74199504e-02
1.62280828e-01 6.78238198e-02 -8.67110193e-01 1.22153115e+00
5.01037180e-01 -3.81828159e-01 4.05941904e-01 5.41024923e-01
8.75089884e-01 3.24572235e-01 2.76889205e-01 -4.49137062e-01
1.92862618e+00 -6.97140276e-01 -1.04078829e+00 -2.13222742e-01
4.77326453e-01 -1.34083176e+00 7.81471908e-01 6.76149368e-01
-6.72435880e-01 -1.10465817e-01 -1.09542990e+00 4.29076105e-02
-7.27310896e-01 5.07927909e-02 9.06715155e-01 6.56679392e-01
-7.39210248e-01 1.04534335e-01 -5.35380602e-01 -6.73584044e-01
3.03886265e-01 5.56982279e-01 -5.93560934e-01 1.93418011e-01
-1.01571703e+00 6.00064456e-01 6.05237126e-01 -9.83555615e-02
-1.08584195e-01 -4.31272835e-01 -8.98630798e-01 3.95085709e-03
9.05121088e-01 -6.48895979e-01 1.13565373e+00 -1.08410895e+00
-1.28066015e+00 1.05385709e+00 -3.99026245e-01 -4.19481695e-01
-1.52568072e-01 -4.43439186e-01 -5.17628908e-01 2.62441009e-01
4.06405121e-01 5.17954588e-01 8.17915380e-01 -1.31106019e+00
-9.72467661e-01 -3.88255388e-01 6.05364084e-01 2.95884013e-01
-3.47214550e-01 4.64742780e-01 -4.47622240e-01 -5.22795916e-01
9.03746933e-02 -9.40645516e-01 -2.18602121e-01 -6.26754463e-01
-4.00247037e-01 -4.76318240e-01 9.12304342e-01 -3.79087150e-01
1.30280924e+00 -1.74406242e+00 -4.89408523e-02 3.79869550e-01
1.52796000e-01 4.42415476e-01 1.27015397e-01 6.53593302e-01
-1.26668856e-01 2.11495191e-01 -3.79292011e-01 -1.17604025e-02
-9.91762280e-02 1.17849082e-01 -3.58536214e-01 -1.30462751e-01
3.02983671e-01 5.74260950e-01 -7.09229469e-01 -7.32866526e-01
-6.02915958e-02 3.90998751e-01 -4.05357152e-01 1.85467809e-01
-2.15155125e-01 1.78466082e-01 -6.36734247e-01 7.55073905e-01
4.81747955e-01 -9.88106951e-02 2.74962813e-01 -2.98179507e-01
-6.30438104e-02 7.38254666e-01 -1.41030824e+00 1.41962481e+00
-4.09386605e-01 5.37032366e-01 -1.44206882e-01 -9.14713562e-01
9.60901260e-01 4.15088534e-01 5.35972893e-01 -4.04799730e-01
2.00941086e-01 -1.15932357e-02 1.15588438e-02 -5.18211842e-01
9.78368104e-01 -3.80236298e-01 -2.53534675e-01 6.93426549e-01
1.27893075e-01 -2.58400977e-01 6.31682277e-01 4.69153404e-01
8.96241844e-01 3.42731416e-01 1.05093527e+00 -4.67845708e-01
1.08449233e+00 8.13342154e-01 6.27215743e-01 3.39190423e-01
-7.12771118e-02 2.88017482e-01 7.12571621e-01 -3.80340338e-01
-6.75210714e-01 -2.07784891e-01 -9.13284495e-02 8.68256450e-01
-1.56607255e-02 -1.10384119e+00 -7.53270805e-01 -1.06114960e+00
-4.73376900e-01 9.14621115e-01 -6.38523817e-01 1.24718875e-01
-6.01685286e-01 -8.43843579e-01 3.19719732e-01 2.51484454e-01
3.55421990e-01 -1.16870284e+00 -8.03436816e-01 2.69791454e-01
-2.52256721e-01 -1.27132273e+00 -1.39369190e-01 3.16206664e-01
-8.28422606e-01 -1.33431137e+00 -1.01629853e-01 -5.71248889e-01
7.84943998e-01 3.24139982e-01 1.35824966e+00 1.02188140e-01
-3.97833101e-02 6.35695219e-01 -6.92071617e-01 -7.22595811e-01
-4.32831913e-01 3.21386963e-01 -2.95714855e-01 -5.73587269e-02
1.27264869e+00 -2.88058341e-01 -3.37357938e-01 8.07037810e-04
-1.12192512e+00 -2.06082851e-01 8.99753809e-01 5.93273044e-01
7.34262347e-01 2.43582487e-01 3.32826406e-01 -1.52929449e+00
9.60559607e-01 -2.10590705e-01 -3.89182597e-01 2.39891529e-01
-1.03443789e+00 1.25805646e-01 4.02642161e-01 -1.24865375e-01
-1.34564722e+00 9.04798880e-02 1.86402053e-02 4.73687291e-01
-5.60391903e-01 6.75284624e-01 -2.77645200e-01 1.60101980e-01
2.44535342e-01 7.77452663e-02 -2.26177633e-01 -2.56628543e-01
1.32324368e-01 7.28666067e-01 1.22355325e-02 -4.76776004e-01
6.06716335e-01 4.73932803e-01 1.04961880e-02 -1.04305041e+00
-1.14343202e+00 -1.03465080e+00 -9.27784264e-01 -1.90295324e-01
4.85448480e-01 -8.58835876e-01 -2.48827383e-01 1.00713912e-02
-1.08303416e+00 6.42622709e-01 -5.02146840e-01 5.96612930e-01
-3.45427781e-01 3.51049960e-01 -1.07675582e-01 -8.40106606e-01
-6.59250677e-01 -9.27807093e-01 1.19692874e+00 5.43377042e-01
-9.44304168e-01 -9.19371068e-01 3.34840775e-01 7.06766665e-01
2.18656719e-01 2.77629793e-02 8.96649599e-01 -1.33707118e+00
-3.44943404e-01 -3.26223552e-01 1.31147683e-01 1.00416556e-01
4.93375450e-01 -1.61861107e-01 -1.00321794e+00 9.70367491e-02
2.54745781e-01 5.90930916e-02 4.13851112e-01 -8.20977017e-02
1.88366234e-01 -1.89691469e-01 -4.17064309e-01 -3.22176330e-02
1.33969855e+00 3.45179439e-01 4.65270996e-01 8.86973619e-01
4.17807043e-01 8.92479777e-01 1.19714832e+00 2.65205264e-01
4.76596951e-01 5.19566119e-01 2.81864661e-03 5.55480570e-02
-1.86858252e-01 1.79221734e-01 2.34741136e-01 1.00186539e+00
-1.32138655e-01 1.51159093e-01 -8.56048048e-01 7.83040285e-01
-1.59035873e+00 -8.70862722e-01 -2.26798162e-01 1.97043693e+00
9.17822540e-01 4.87706929e-01 1.17128290e-01 4.77042049e-01
4.30269778e-01 1.72659457e-01 -1.12183996e-01 -6.79638743e-01
-1.98153421e-01 3.40768874e-01 -1.81032345e-02 4.15702552e-01
-1.18197298e+00 1.03227699e+00 6.28561926e+00 5.39442658e-01
-8.09901178e-01 -1.12984523e-01 2.27959707e-01 1.44177571e-01
-4.61456090e-01 4.28388476e-01 -1.03326666e+00 -2.12104991e-01
6.89148486e-01 -3.31406444e-01 -7.54550397e-02 7.81627953e-01
1.59329146e-01 -6.27969921e-01 -8.88713598e-01 5.67773938e-01
5.24794281e-01 -9.23957407e-01 3.13356519e-01 -2.98607815e-02
5.71880221e-01 -4.24270034e-01 -5.72144687e-01 2.89334983e-01
1.92701235e-01 -5.14192283e-01 2.77921408e-01 3.77647609e-01
1.68322504e-01 -9.03059721e-01 1.15670145e+00 1.63180113e-01
-1.32505333e+00 3.20279449e-01 -1.53491199e-01 -5.25040478e-02
-7.59171962e-04 7.33938515e-01 -8.66108239e-01 8.71638834e-01
7.38156438e-01 6.23356342e-01 -6.79098129e-01 8.86962116e-01
-5.79272389e-01 7.06555963e-01 -1.42818451e-01 -2.67487258e-01
1.63442001e-01 -3.42809260e-01 1.02239335e+00 1.30890584e+00
-6.63234368e-02 2.47008055e-01 3.10113225e-02 5.73519289e-01
9.65256840e-02 6.50018573e-01 -7.06050754e-01 -2.15916216e-01
1.05929472e-01 1.67277575e+00 -1.24086320e+00 -5.82349062e-01
-5.95172584e-01 4.33158964e-01 2.15101428e-02 1.24929160e-01
-1.61439463e-01 -6.16769433e-01 3.85005146e-01 5.62757580e-03
6.04406118e-01 -1.06048770e-02 -3.98820788e-01 -1.30118966e+00
2.82380134e-01 -1.33567894e+00 7.08209991e-01 -4.65011865e-01
-8.37460279e-01 1.01923382e+00 2.20468163e-01 -1.33123147e+00
-5.64540327e-01 -3.56497973e-01 -5.91775060e-01 5.18970311e-01
-1.54374492e+00 -1.27787530e+00 1.00027256e-01 2.75564760e-01
7.67380238e-01 -2.47806758e-01 9.69845653e-01 4.52367887e-02
-1.41683504e-01 2.05215484e-01 -6.06559873e-01 -6.76043555e-02
8.55339766e-01 -1.32971883e+00 6.70417771e-02 1.16870821e+00
4.70358163e-01 1.10296106e+00 1.01867557e+00 -7.96991229e-01
-1.38952935e+00 -6.47027373e-01 1.61678910e+00 -4.28077012e-01
7.27669895e-01 -3.01582724e-01 -1.01734293e+00 5.16525626e-01
8.78192186e-01 -6.95442021e-01 1.34912682e+00 4.82710481e-01
-4.04543400e-01 -1.54321743e-02 -1.13310456e+00 4.29406166e-01
5.77071965e-01 -5.33441782e-01 -1.49308395e+00 -1.20591789e-01
4.27262902e-01 -9.95848998e-02 -7.73988307e-01 4.85056937e-01
5.21596253e-01 -8.39284480e-01 5.28078854e-01 -5.26463270e-01
1.08917303e-01 -5.16092300e-01 -1.18240133e-01 -1.20588470e+00
1.87022895e-01 -6.25426292e-01 4.63338085e-02 1.82887316e+00
7.51857877e-01 -4.77088392e-01 5.84519267e-01 7.59805560e-01
4.62655604e-01 -5.60537040e-01 -6.41438961e-01 -2.81117439e-01
-4.53217149e-01 -5.83077192e-01 8.13409507e-01 9.23054814e-01
3.76225740e-01 1.00684834e+00 -4.95842434e-02 6.63080625e-03
5.03172636e-01 5.49793601e-01 7.35586107e-01 -1.34815454e+00
-4.65400070e-02 -4.50719416e-01 -2.68509388e-01 -3.94023746e-01
9.50170234e-02 -6.77874863e-01 -1.31020054e-01 -1.55518758e+00
5.21568000e-01 -1.08431831e-01 -1.59196392e-01 4.20415759e-01
-3.73919696e-01 1.52056903e-01 5.83518595e-02 1.44906282e-01
-8.88218045e-01 2.58562714e-01 8.99529755e-01 -2.46105403e-01
-3.87297779e-01 2.14666858e-01 -1.31774580e+00 1.01582205e+00
7.73599386e-01 -8.72999430e-01 -3.85512471e-01 1.89393654e-01
6.11725330e-01 -5.01276016e-01 -3.25617194e-01 -7.27813661e-01
3.40220183e-01 6.71454589e-04 -7.99524691e-03 -8.21253061e-01
4.98568378e-02 -1.09303904e+00 -9.57971811e-02 2.00570732e-01
-2.86021829e-01 5.56702949e-02 2.93618083e-01 3.12665701e-01
-6.96938276e-01 -4.94530052e-01 2.39473656e-01 -9.07258689e-02
-7.93888927e-01 -3.12932432e-01 -5.63001275e-01 -5.81524055e-03
9.00210083e-01 -1.22035585e-01 -8.00916329e-02 -3.14989507e-01
-5.58149636e-01 -5.59863113e-02 2.91326910e-01 5.19172370e-01
4.83886719e-01 -1.10804391e+00 -3.84245723e-01 1.12525851e-01
4.53490674e-01 -5.36504947e-02 -2.61146814e-01 8.40768337e-01
-8.31254721e-02 6.08058393e-01 -5.66875152e-02 -3.67816567e-01
-1.64241505e+00 4.84619766e-01 -2.32548892e-01 -6.73466682e-01
-5.15898705e-01 1.81555927e-01 -1.33443817e-01 -5.04397273e-01
5.91128543e-02 -2.53319055e-01 -1.14957666e+00 5.79081774e-01
6.98822141e-01 9.99799445e-02 3.31959456e-01 -1.06380296e+00
-5.99516749e-01 8.20489764e-01 -4.21787262e-01 -1.88666359e-01
1.47464013e+00 -4.18692380e-01 -5.07696807e-01 5.08141696e-01
6.29932761e-01 4.26599145e-01 -2.59250015e-01 -3.60991627e-01
7.27718651e-01 -2.37352848e-01 -1.24584781e-02 -8.72893751e-01
-8.08227003e-01 3.91146392e-01 1.20517850e-01 5.89213014e-01
1.31868339e+00 1.64024010e-01 3.85514587e-01 6.16078377e-01
2.27207929e-01 -1.21939230e+00 -3.26619625e-01 5.15104115e-01
6.71153724e-01 -1.47042990e+00 4.51911181e-01 -6.94643676e-01
-7.44660914e-01 1.21623993e+00 4.84807640e-01 3.27504352e-02
8.43638301e-01 5.92286527e-01 4.26019579e-01 -6.59321785e-01
-8.92096937e-01 -5.44548571e-01 5.75096130e-01 4.39496517e-01
6.54781580e-01 -5.97156473e-02 -1.01658297e+00 8.75009418e-01
-2.38492355e-01 -1.63097411e-01 6.11529768e-01 1.25988150e+00
-3.29770625e-01 -1.66559315e+00 -4.48153824e-01 5.33716917e-01
-1.02294302e+00 -2.50308305e-01 -9.79907334e-01 9.35391009e-01
-9.44178253e-02 1.32176590e+00 -3.33143115e-01 -1.96130071e-02
5.43134987e-01 3.15397888e-01 -7.20929261e-03 -7.79072523e-01
-1.05176735e+00 4.94304448e-01 5.59322238e-01 -5.65860331e-01
-1.36407709e+00 -8.83579016e-01 -1.15872133e+00 4.00381088e-01
-4.01018947e-01 7.13211060e-01 6.56206965e-01 1.46138573e+00
4.27039146e-01 4.91507202e-01 3.49244237e-01 -1.55083254e-01
1.35280430e-01 -8.40016067e-01 -3.37840915e-01 4.88655299e-01
1.33596256e-01 -5.56470037e-01 -3.30619067e-01 2.58337200e-01] | [11.274374961853027, 6.790992736816406] |
2c8d3e75-9a7b-4146-b0aa-2974e111c6f6 | segmentation-of-aortic-vessel-tree-in-ct | 2305.09833 | null | https://arxiv.org/abs/2305.09833v1 | https://arxiv.org/pdf/2305.09833v1.pdf | Segmentation of Aortic Vessel Tree in CT Scans with Deep Fully Convolutional Networks | Automatic and accurate segmentation of aortic vessel tree (AVT) in computed tomography (CT) scans is crucial for early detection, diagnosis and prognosis of aortic diseases, such as aneurysms, dissections and stenosis. However, this task remains challenges, due to the complexity of aortic vessel tree and amount of CT angiography data. In this technical report, we use two-stage fully convolutional networks (FCNs) to automatically segment AVT in CTA scans from multiple centers. Specifically, we firstly adopt a 3D FCN with U-shape network architecture to segment AVT in order to produce topology attention and accelerate medical image analysis pipeline. And then another one 3D FCN is trained to segment branches of AVT along the pseudo-centerline of AVT. In the 2023 MICCAI Segmentation of the Aorta (SEG.A.) Challenge , the reported method was evaluated on the public dataset of 56 cases. The resulting Dice Similarity Coefficient (DSC) is 0.920, Jaccard Similarity Coefficient (JSC) is 0.861, Recall is 0.922, and Precision is 0.926 on a 5-fold random split of training and validation set. | ['Feng Yang', 'Shaofeng Yuan'] | 2023-05-16 | null | null | null | null | ['computed-tomography-ct'] | ['methodology'] | [-2.02019259e-01 6.29660189e-02 4.38503996e-02 -5.10797262e-01
-4.37810242e-01 -7.11709797e-01 4.24976982e-02 1.79558352e-01
-3.46890956e-01 5.38122177e-01 4.40621823e-02 -8.82090151e-01
-4.64308411e-02 -5.75852931e-01 -3.75246108e-01 -4.44827408e-01
-6.28897250e-01 9.43680227e-01 3.30302596e-01 1.23066515e-01
1.36176661e-01 1.08422244e+00 -5.83310187e-01 1.39666572e-01
9.54690516e-01 1.28271723e+00 -1.04771040e-01 9.03148413e-01
-2.92649925e-01 8.13861072e-01 -5.36245763e-01 -4.70395416e-01
4.63936269e-01 -2.95067668e-01 -1.12147641e+00 -1.29192725e-01
6.30677760e-01 -3.68375063e-01 -1.89470455e-01 8.65432262e-01
5.90403378e-01 -2.00530306e-01 8.36452425e-01 -7.00171888e-01
-3.00949097e-01 5.34589350e-01 -5.87974250e-01 7.92059600e-01
-7.91743517e-01 1.79374933e-01 7.04569459e-01 -5.97545564e-01
7.75915384e-01 1.11770999e+00 8.09840262e-01 4.10889268e-01
-8.46426368e-01 -5.29903173e-01 -3.07839364e-01 -2.09660735e-02
-9.35364902e-01 1.82121620e-01 4.28181827e-01 -5.94520390e-01
3.48699242e-01 9.81297940e-02 8.62022460e-01 5.75474977e-01
2.88888961e-01 5.30455828e-01 7.84167647e-01 -4.13179770e-02
1.65301025e-01 -2.62032628e-01 2.37570733e-01 7.80535936e-01
3.17882329e-01 5.04597090e-02 5.40035129e-01 -1.60856023e-01
1.39644659e+00 -8.26715156e-02 9.13150832e-02 -4.05841678e-01
-1.13734043e+00 1.00272965e+00 9.96544778e-01 3.28330487e-01
-5.82453787e-01 -1.65193915e-01 7.84885406e-01 1.65349439e-01
2.29112610e-01 6.22788608e-01 -3.08400065e-01 1.89543754e-01
-8.52650225e-01 1.52582556e-01 6.20139182e-01 5.41619539e-01
-3.30843121e-01 2.71639042e-02 -1.58879265e-01 7.52201676e-01
1.53102204e-01 2.97939390e-01 6.10345125e-01 -1.21720386e+00
2.13387042e-01 5.54765821e-01 -2.81335086e-01 -6.74777687e-01
-8.12934041e-01 -8.47738087e-01 -1.24808061e+00 4.26115036e-01
7.33573616e-01 -4.42853570e-01 -1.24356771e+00 1.20772064e+00
5.40136039e-01 2.37567306e-01 -2.84939200e-01 1.41826582e+00
1.26207638e+00 2.07267851e-01 1.56459063e-01 -4.50505391e-02
1.72694123e+00 -1.05732715e+00 -2.92347312e-01 2.23954231e-01
8.86669755e-01 -8.06453049e-01 5.73584378e-01 3.01645249e-02
-1.24789667e+00 -4.79795337e-01 -8.07728767e-01 1.01084106e-01
1.13357082e-01 2.53960371e-01 5.17027020e-01 6.48341715e-01
-9.39642966e-01 5.37618876e-01 -9.95710969e-01 4.38603526e-03
1.16811442e+00 1.45424962e-01 -2.28143230e-01 1.04013167e-01
-1.02618623e+00 9.43167269e-01 2.51227945e-01 5.21731116e-02
-8.28161597e-01 -1.20955443e+00 -3.58889550e-01 8.00736062e-03
2.22077772e-01 -8.73008847e-01 1.19820595e+00 -6.38750970e-01
-1.26610112e+00 1.11352730e+00 3.06248575e-01 -9.71367657e-01
8.33424449e-01 -1.92519709e-01 -2.56100267e-01 6.23180091e-01
1.97576791e-01 6.13599479e-01 4.58159059e-01 -9.33324456e-01
-6.28484428e-01 -5.24265468e-01 -4.77859110e-01 -1.21025719e-01
4.75357175e-01 5.60716353e-02 -1.99805558e-01 -7.33639598e-01
3.12828898e-01 -9.20526147e-01 -8.42521846e-01 3.88077110e-01
-4.16350573e-01 -9.25392658e-02 6.08970702e-01 -9.79938328e-01
9.79016066e-01 -1.91010475e+00 -2.18835801e-01 5.68390548e-01
9.26190913e-01 7.66834140e-01 2.52103299e-01 -5.48728704e-01
-3.35788578e-01 1.84741288e-01 -5.76860845e-01 1.37645841e-01
-8.61200690e-01 1.79137364e-01 9.41965580e-02 3.64339918e-01
1.90800905e-01 1.01272905e+00 -5.69402397e-01 -8.53274286e-01
2.91661292e-01 3.30252677e-01 -7.06230938e-01 1.94124639e-01
7.39381686e-02 1.02273011e+00 -6.95165455e-01 4.63239044e-01
7.85616398e-01 -5.60374379e-01 1.01108156e-01 -2.36441255e-01
-2.03065202e-01 4.82783988e-02 -8.51152241e-01 1.68773937e+00
-4.46813613e-01 5.19748867e-01 9.39060375e-02 -1.15863490e+00
1.02113211e+00 5.96500516e-01 1.04034948e+00 -5.35385728e-01
6.13524616e-01 5.95473528e-01 7.77481139e-01 -5.30596793e-01
-3.64973098e-01 -2.17288416e-02 3.82747978e-01 3.73192251e-01
-9.37488750e-02 -1.05407974e-02 3.01781803e-01 1.94016546e-01
9.36418951e-01 -3.32826674e-01 2.12345734e-01 -4.84031200e-01
5.88757098e-01 3.57775658e-01 4.38096672e-01 6.09923065e-01
-6.42359316e-01 9.56167519e-01 7.32047439e-01 -1.05680585e+00
-1.38501275e+00 -1.15531015e+00 -6.91018283e-01 2.31051326e-01
-4.26599458e-02 -2.38441527e-02 -6.63048685e-01 -6.75773382e-01
-9.42799747e-02 3.80691916e-01 -5.06935358e-01 3.72309089e-02
-1.31954288e+00 -6.39356554e-01 5.71238279e-01 6.87750518e-01
6.33188725e-01 -8.71099591e-01 -1.20930648e+00 3.74750346e-01
-3.94070856e-02 -1.14755595e+00 -1.40270412e-01 -2.29324594e-01
-1.37478042e+00 -1.29874229e+00 -9.91636276e-01 -9.11910832e-01
4.31696355e-01 -1.80764735e-01 1.28826809e+00 3.22237670e-01
-8.03760052e-01 -4.47573781e-01 -1.16534874e-01 -8.51043612e-02
-5.15772700e-01 1.10349633e-01 -5.56641936e-01 -2.22831443e-01
-3.48087326e-02 -4.79488820e-01 -1.05397749e+00 3.38066101e-01
-3.01328421e-01 -1.53062344e-01 6.48147762e-01 8.36682081e-01
6.78863764e-01 -3.39653879e-01 3.92003030e-01 -1.05471981e+00
3.05003434e-01 -5.92103779e-01 -7.22181380e-01 5.33446372e-02
-5.19684196e-01 -3.54693621e-01 6.75527930e-01 -1.42273515e-01
-8.26369047e-01 2.20062360e-02 -2.99642146e-01 -6.78740978e-01
-3.13975304e-01 3.48398328e-01 5.66703320e-01 9.91407856e-02
8.44321012e-01 -2.08243281e-01 5.08106530e-01 -5.17039657e-01
3.14849019e-01 4.21242774e-01 8.06337714e-01 -3.80729586e-01
3.01871926e-01 4.84049112e-01 6.45799816e-01 -5.17325878e-01
-7.26587653e-01 -2.86169142e-01 -9.00257826e-01 -8.86823535e-02
1.16976166e+00 -5.63696206e-01 -6.42390966e-01 -6.80351909e-03
-9.22091901e-01 2.48351414e-02 -2.82089114e-01 7.47565806e-01
-1.94339067e-01 3.44201624e-01 -8.59530091e-01 -2.27477461e-01
-9.10964668e-01 -1.55498636e+00 6.95385814e-01 2.15341195e-01
-2.59489834e-01 -9.42714334e-01 -1.17745206e-01 2.68688530e-01
7.60166168e-01 7.03425109e-01 1.30293727e+00 -1.01849711e+00
-3.44134241e-01 -3.57138902e-01 -5.62056899e-01 3.39185476e-01
8.03191885e-02 1.08509950e-01 -3.75302047e-01 9.92534161e-02
-1.36317760e-01 1.06896266e-01 8.26586604e-01 1.27891791e+00
1.42792785e+00 1.38956591e-01 -2.42366046e-01 9.54543710e-01
9.86589491e-01 3.87453407e-01 2.10263267e-01 2.63582081e-01
8.80444169e-01 4.47775304e-01 3.32685977e-01 3.76481146e-01
-1.21863708e-02 3.69430035e-01 6.24429405e-01 -3.35535794e-01
-4.05890793e-01 4.48552817e-01 -5.94127595e-01 4.40808207e-01
-3.35172296e-01 2.97334760e-01 -1.24559724e+00 3.62330824e-01
-1.24758232e+00 -5.75672567e-01 -9.19664800e-01 1.94010532e+00
6.19167626e-01 1.76666617e-01 3.25532913e-01 -2.75353968e-01
8.15409184e-01 -2.59289920e-01 -6.28612995e-01 -4.05892938e-01
-5.70610166e-02 5.95624864e-01 5.62155962e-01 3.03533942e-01
-1.49023366e+00 6.84499502e-01 5.61129570e+00 4.04931843e-01
-1.36080945e+00 1.78919453e-02 1.12764406e+00 -1.75054874e-02
3.04616779e-01 -1.73030078e-01 -3.30923051e-01 4.67341453e-01
9.89953041e-01 1.92023918e-01 -2.04567581e-01 6.95072174e-01
1.08367257e-01 2.34275252e-01 -5.42594969e-01 6.62855744e-01
-6.67310894e-01 -1.69738126e+00 1.55091714e-02 -7.22849220e-02
6.73318505e-01 2.10507274e-01 4.60716076e-02 4.45732512e-02
4.36065227e-01 -1.24425018e+00 2.17916340e-01 6.24832287e-02
8.61725867e-01 -6.08527422e-01 1.04447806e+00 -7.64899477e-02
-1.04557133e+00 4.99652848e-02 -1.55153304e-01 5.11495590e-01
5.41936100e-01 6.98364615e-01 -1.09156287e+00 4.03609991e-01
8.96369100e-01 6.52262211e-01 -4.11376894e-01 1.54396832e+00
1.13299467e-01 8.13104153e-01 -4.55033600e-01 3.87243807e-01
5.51184356e-01 -3.28267843e-01 6.55257821e-01 8.02581012e-01
2.93577671e-01 2.68399030e-01 7.99812824e-02 8.05359006e-01
-1.43602192e-01 3.70332807e-01 2.87909899e-03 4.73918051e-01
5.68850696e-01 1.35825074e+00 -1.09341443e+00 -6.95940614e-01
-4.79414240e-02 2.97501117e-01 -1.97426416e-02 -9.40642436e-04
-9.26830709e-01 1.18667353e-03 3.51446480e-01 1.91847488e-01
2.80749083e-01 1.11461461e-01 -8.63188386e-01 -7.32883573e-01
-2.06250921e-01 -5.54160595e-01 8.25022697e-01 -3.68998915e-01
-1.21766579e+00 8.46783221e-01 -2.66441554e-01 -1.18885553e+00
7.08649531e-02 -6.74228907e-01 -8.64398897e-01 9.73076761e-01
-1.18297124e+00 -8.94152999e-01 -4.48082864e-01 3.16955388e-01
4.83721882e-01 -3.09136540e-01 6.01473093e-01 5.15547991e-01
-3.82018179e-01 3.20499867e-01 -1.96880661e-02 6.07466578e-01
5.84588408e-01 -1.36878777e+00 5.60898542e-01 5.72185814e-01
-5.73196173e-01 5.16033173e-01 2.50360996e-01 -7.47563660e-01
-6.33407831e-01 -1.40293872e+00 4.19743896e-01 -2.69685149e-01
5.13633072e-01 5.03461778e-01 -9.57137823e-01 5.79208791e-01
-1.18945427e-01 7.22338021e-01 6.72977448e-01 -4.57823902e-01
-2.12165847e-01 2.57451087e-01 -1.28051531e+00 4.69227701e-01
8.38226616e-01 4.03667390e-01 -3.55545044e-01 3.46644729e-01
6.29859447e-01 -8.29492152e-01 -1.56596518e+00 6.64056182e-01
6.70720160e-01 -7.37491965e-01 1.40766549e+00 -1.09896350e+00
8.72384071e-01 -1.57425016e-01 2.09275246e-01 -9.31424916e-01
-3.75824302e-01 -1.90042347e-01 1.25001207e-01 4.87138003e-01
3.61851037e-01 -4.43983644e-01 8.30375969e-01 3.17127079e-01
-8.05780828e-01 -9.28404748e-01 -1.11130619e+00 -3.47957134e-01
7.99880326e-01 -7.82496110e-03 5.10024786e-01 9.95874166e-01
-8.63612115e-01 8.24501440e-02 2.92295939e-03 -2.02946868e-02
8.58325124e-01 3.46873909e-01 3.64275038e-01 -1.71140623e+00
7.52277821e-02 -1.05446208e+00 -3.02090079e-01 -7.86341250e-01
-3.54561567e-01 -1.19546092e+00 -5.68302631e-01 -1.40620387e+00
-1.90097541e-01 -8.49669456e-01 -2.07628295e-01 -6.84844032e-02
-1.73312485e-01 3.43151577e-02 -1.66728228e-01 5.14327228e-01
3.43165547e-02 5.92564084e-02 1.98215938e+00 1.22026503e-01
-1.76611617e-01 4.90277976e-01 -2.92479575e-01 6.35253429e-01
1.06301332e+00 -4.60867345e-01 -1.34934813e-01 -2.22904444e-01
-5.57216346e-01 4.98091400e-01 4.97861356e-01 -7.65023410e-01
-5.52542582e-02 1.57868177e-01 6.14332914e-01 -8.52529824e-01
-7.10518062e-02 -7.25014448e-01 -1.54118896e-01 1.15790856e+00
-4.85403031e-01 9.48028713e-02 -7.06031919e-02 2.40848996e-02
-1.40906185e-01 -2.78258532e-01 1.13081622e+00 -6.14053249e-01
-4.01300699e-01 6.87016487e-01 -3.77058327e-01 3.82169247e-01
1.06821430e+00 -2.76397001e-02 -2.25816876e-01 2.43439347e-01
-1.05328679e+00 3.02013040e-01 -2.09528267e-01 4.74779829e-02
5.23300946e-01 -1.09759474e+00 -1.23874426e+00 1.70429185e-01
-1.22469477e-01 6.15974665e-01 4.73807365e-01 1.48704827e+00
-1.48037374e+00 3.82228613e-01 -4.90557522e-01 -1.08267999e+00
-1.17220521e+00 2.80350357e-01 8.32201302e-01 -3.73209625e-01
-1.34486198e+00 1.04360902e+00 6.05951361e-02 -4.46754932e-01
1.12996548e-01 -4.72094238e-01 -4.91266400e-01 -1.89569175e-01
3.35534304e-01 3.06962818e-01 7.01781064e-02 -6.74744666e-01
-2.44372830e-01 6.56637669e-01 -2.59248734e-01 3.98277372e-01
1.16588259e+00 1.39389381e-01 -1.15037523e-01 -2.29204267e-01
1.19712174e+00 -5.10318100e-01 -8.35005641e-01 -4.40901965e-01
-8.04874673e-02 -3.07346314e-01 1.67110890e-01 -7.53969193e-01
-1.86544359e+00 1.13694751e+00 1.05044675e+00 1.05452195e-01
7.43553042e-01 3.53221297e-02 1.16350353e+00 -1.35151133e-01
-3.00872117e-01 -4.58062708e-01 -2.31064767e-01 3.58346462e-01
5.55801213e-01 -1.38608682e+00 -1.04346111e-01 -5.98962426e-01
-7.88958132e-01 1.53174102e+00 5.88692904e-01 -4.62748915e-01
1.02811635e+00 -4.77266498e-02 3.78450483e-01 -5.47595501e-01
-4.22241122e-01 2.40270168e-01 3.15127164e-01 3.27415884e-01
6.48368359e-01 3.30457002e-01 -2.07350224e-01 3.06104183e-01
-4.84942853e-01 -1.19796008e-01 3.52849394e-01 8.65533710e-01
-5.27157128e-01 -5.67539454e-01 -2.78698117e-01 8.74277949e-01
-8.54676604e-01 -3.15245129e-02 1.74979027e-02 7.20889628e-01
-8.35236162e-02 4.93272275e-01 2.73294806e-01 1.50409132e-01
2.29086503e-01 -2.59600580e-01 7.72245154e-02 -1.43315941e-01
-9.05406475e-01 2.69494265e-01 7.82281160e-02 -4.09402758e-01
-1.18766069e-01 -5.28383315e-01 -1.64852607e+00 -3.06514263e-01
-2.62282393e-03 1.77488148e-01 7.52857566e-01 8.36388052e-01
2.80837178e-01 1.01338780e+00 6.00890160e-01 -3.33120078e-01
-5.81008494e-01 -9.22815502e-01 -3.20178419e-01 6.13499761e-01
3.07915658e-01 -6.90851212e-01 6.15355074e-02 5.71634732e-02] | [14.29101848602295, -2.467310667037964] |
69b5fc7c-3645-4bf3-8d31-d1d7fc83308b | just-ask-for-calibration-strategies-for | 2305.14975 | null | https://arxiv.org/abs/2305.14975v1 | https://arxiv.org/pdf/2305.14975v1.pdf | Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback | A trustworthy real-world prediction system should be well-calibrated; that is, its confidence in an answer is indicative of the likelihood that the answer is correct, enabling deferral to a more expensive expert in cases of low-confidence predictions. While recent studies have shown that unsupervised pre-training produces large language models (LMs) that are remarkably well-calibrated, the most widely-used LMs in practice are fine-tuned with reinforcement learning with human feedback (RLHF-LMs) after the initial unsupervised pre-training stage, and results are mixed as to whether these models preserve the well-calibratedness of their ancestors. In this paper, we conduct a broad evaluation of computationally feasible methods for extracting confidence scores from LLMs fine-tuned with RLHF. We find that with the right prompting strategy, RLHF-LMs verbalize probabilities that are much better calibrated than the model's conditional probabilities, enabling fairly well-calibrated predictions. Through a combination of prompting strategy and temperature scaling, we find that we can reduce the expected calibration error of RLHF-LMs by over 50%. | ['Christopher D. Manning', 'Chelsea Finn', 'Huaxiu Yao', 'Rafael Rafailov', 'Archit Sharma', 'Allan Zhou', 'Eric Mitchell', 'Katherine Tian'] | 2023-05-24 | null | null | null | null | ['unsupervised-pre-training'] | ['methodology'] | [-4.60879095e-02 4.35375124e-01 -3.22275549e-01 -8.68796706e-01
-1.02593362e+00 -6.47254407e-01 3.94928008e-01 3.23948890e-01
-4.61359739e-01 1.04540157e+00 -9.20779479e-04 -6.50461912e-01
1.47575453e-01 -8.02151024e-01 -8.80134761e-01 -2.89351761e-01
2.73197681e-01 7.30787456e-01 4.66283828e-01 -1.36416733e-01
2.91879892e-01 2.14251310e-01 -1.36154521e+00 3.16569097e-02
1.01430357e+00 6.30004227e-01 -1.22518435e-01 9.81924772e-01
-5.39559219e-03 1.15499640e+00 -7.30350673e-01 -7.35199869e-01
-5.01599498e-02 -3.23477000e-01 -8.54563117e-01 -6.79647565e-01
2.38062650e-01 -4.84994560e-01 7.48533756e-02 9.29214537e-01
3.58520299e-02 -8.16115644e-03 6.91612899e-01 -1.21504879e+00
-2.85211176e-01 1.03067148e+00 -2.58004636e-01 5.73607907e-02
4.70485121e-01 9.08090025e-02 1.12387741e+00 -3.97247702e-01
3.66193831e-01 1.19301665e+00 9.61201072e-01 4.55526888e-01
-1.23255348e+00 -9.63797271e-01 1.06015280e-01 -1.10428318e-01
-1.44676316e+00 -5.33389091e-01 1.76358759e-01 -3.76171708e-01
1.14535129e+00 2.33770117e-01 3.53907824e-01 8.32757294e-01
5.08673370e-01 4.12111208e-02 1.11347842e+00 -6.32178843e-01
5.16439140e-01 6.46351993e-01 6.97771311e-02 6.25509441e-01
5.35444915e-01 4.27587777e-01 -8.15646172e-01 -4.31810737e-01
4.90883112e-01 -4.36019272e-01 1.56025440e-01 -5.54069243e-02
-1.01216412e+00 7.04848289e-01 4.72446904e-02 1.70384288e-01
-2.06585079e-01 2.52375484e-01 6.61808997e-02 2.74747044e-01
3.51286262e-01 8.10343623e-01 -7.52819359e-01 -3.01015019e-01
-1.33245409e+00 1.88454479e-01 1.10362160e+00 9.78166223e-01
8.75108242e-01 -5.51911220e-02 9.06676054e-03 4.13734913e-01
5.41816771e-01 6.37382448e-01 3.35446835e-01 -1.23127067e+00
9.26864967e-02 2.69902378e-01 5.26963115e-01 -6.73498511e-01
-1.99452326e-01 -3.81970406e-01 -3.07501793e-01 2.72125453e-01
4.25885081e-01 -4.15984809e-01 -6.17931843e-01 1.99057996e+00
2.17003688e-01 8.80609602e-02 3.50806445e-01 5.59936464e-01
2.75508791e-01 7.51817226e-01 4.22104329e-01 -3.27539563e-01
8.04880202e-01 -5.25597811e-01 -2.43447065e-01 -5.00833571e-01
5.88539958e-01 -7.38520324e-01 1.06825089e+00 6.02633893e-01
-7.27504551e-01 -6.13476574e-01 -1.31854498e+00 2.52346307e-01
1.59318671e-01 -3.21548313e-01 7.06856370e-01 8.99917662e-01
-1.08041024e+00 9.03995275e-01 -7.79437304e-01 -2.47388810e-01
-1.53354645e-01 3.66839767e-01 -3.05353075e-01 2.86466122e-01
-1.44177687e+00 1.30752003e+00 5.00497580e-01 -2.60640800e-01
-7.62186229e-01 -5.13397694e-01 -5.81381500e-01 3.65283191e-02
3.21140409e-01 -3.43022883e-01 1.81466281e+00 -1.06008327e+00
-1.54573119e+00 5.04000306e-01 -4.07832205e-01 -5.75878203e-01
5.92379212e-01 -2.11221322e-01 -6.12197220e-01 1.40310954e-02
7.00209697e-04 7.34873831e-01 9.05759573e-01 -1.16244471e+00
-7.32324839e-01 1.25110701e-01 1.77948289e-02 -1.02377631e-01
-2.10651204e-01 3.01339701e-02 -8.69098306e-02 -1.10204212e-01
-6.08178973e-02 -7.71287739e-01 -1.95208505e-01 -1.81047812e-01
-3.29621941e-01 -1.05055124e-01 3.31610441e-01 -5.66966116e-01
1.50053859e+00 -1.79793262e+00 -4.43599403e-01 5.23803771e-01
1.32990301e-01 1.97583988e-01 6.08558021e-02 4.01862115e-01
2.63764672e-02 3.80829483e-01 -1.40116379e-01 -6.45913780e-02
1.57754257e-01 4.15065318e-01 -5.47052562e-01 2.00374827e-01
3.23724031e-01 6.27609074e-01 -9.29483891e-01 -7.99288630e-01
8.25137049e-02 2.17058226e-01 -6.07823551e-01 5.42103469e-01
-5.42429686e-01 1.84051201e-01 -2.54317880e-01 2.28009522e-01
2.35647365e-01 -5.52624941e-01 4.85631555e-01 1.90519974e-01
-7.89697692e-02 4.49010670e-01 -1.06557405e+00 9.96299386e-01
-4.14854020e-01 3.43436331e-01 -3.24461430e-01 -4.18645144e-01
1.20763969e+00 1.53876111e-01 -7.14668632e-02 -4.77657497e-01
-1.95525542e-01 3.63702714e-01 2.85815150e-02 -9.01442915e-02
7.29084432e-01 -5.95623732e-01 -2.12716728e-01 7.42454171e-01
-1.66184772e-02 -6.16186559e-01 -1.99819133e-02 5.07698357e-01
8.75431001e-01 3.75829220e-01 4.01176274e-01 -1.47952408e-01
2.74307579e-01 -1.93401296e-02 7.23294079e-01 1.04683077e+00
-2.45167077e-01 3.21581513e-01 4.90031153e-01 -1.44986138e-01
-1.15401912e+00 -8.75627339e-01 -4.59619053e-03 1.15355313e+00
-1.21183746e-01 -7.58919120e-01 -9.52510834e-01 -5.36365569e-01
7.88047723e-03 1.37456822e+00 -3.58909637e-01 -3.60298336e-01
-1.71653137e-01 -3.54418278e-01 7.47906744e-01 6.61493957e-01
2.53825694e-01 -7.45563209e-01 -7.19331205e-01 3.92914325e-01
3.01120896e-02 -8.01343024e-01 -1.79279089e-01 6.44093275e-01
-7.60191023e-01 -8.97209942e-01 -3.09306055e-01 -3.42703164e-01
5.25750220e-01 -4.04650778e-01 1.37460423e+00 3.96090418e-01
5.06435096e-01 2.03811094e-01 -1.67884752e-01 -3.24235320e-01
-1.13461411e+00 1.87720433e-02 3.48012388e-01 -6.17287993e-01
4.70610261e-01 -4.03223753e-01 -1.95810303e-01 3.54020834e-01
-4.76891220e-01 -4.04294841e-02 4.50850219e-01 7.34278500e-01
4.59486932e-01 -4.25145179e-02 5.99187553e-01 -1.24244583e+00
6.45129144e-01 -2.56147712e-01 -8.18236351e-01 7.70663440e-01
-1.24705303e+00 5.93483090e-01 8.17609847e-01 -4.58675057e-01
-1.19415307e+00 -7.36826062e-02 -2.19484165e-01 -8.21160451e-02
-2.52395093e-01 4.83059227e-01 1.73115015e-01 1.05024397e-01
9.74233866e-01 -4.94165979e-02 -4.25714761e-01 -1.32185921e-01
2.73157269e-01 7.35790670e-01 7.06750989e-01 -9.56235647e-01
8.15497696e-01 -2.29763910e-01 -4.16112781e-01 -3.28033596e-01
-9.76650894e-01 1.35817185e-01 -4.81922299e-01 -2.61201113e-01
4.07012969e-01 -7.65148878e-01 -7.10377097e-01 1.64925337e-01
-8.76777470e-01 -6.19803131e-01 -1.76421568e-01 3.70267212e-01
-3.73575121e-01 5.02125084e-01 -5.76645076e-01 -1.16527474e+00
-3.70078117e-01 -9.37415957e-01 7.76900232e-01 4.59846884e-01
-1.08896244e+00 -9.47975576e-01 3.38701338e-01 1.37316883e-01
4.23144221e-01 -2.81287134e-01 1.21570659e+00 -9.05970156e-01
-8.93848240e-02 -1.85636804e-01 1.79232866e-01 4.87516463e-01
-1.36555925e-01 6.16650224e-01 -1.17903996e+00 -1.29361689e-01
-2.87066191e-01 -6.06930256e-01 4.88082141e-01 4.76495363e-02
7.39319801e-01 -3.27470034e-01 -2.53746063e-01 1.05042510e-01
1.03190160e+00 8.16722587e-02 3.70187312e-01 8.21419284e-02
7.09931552e-02 3.74284208e-01 6.60462081e-01 4.74439532e-01
3.34883720e-01 1.37634188e-01 -1.23694845e-01 2.52876073e-01
3.77945721e-01 -8.53104711e-01 5.04760861e-01 9.33804214e-01
1.40624568e-01 -2.69938290e-01 -1.04485679e+00 8.85498617e-03
-1.70668364e+00 -8.32151532e-01 2.69214839e-01 2.46259594e+00
1.25001729e+00 8.19011927e-01 -6.05335310e-02 2.55280375e-01
6.38578951e-01 -7.78183192e-02 -8.00264418e-01 -6.77028775e-01
-4.74639330e-03 3.60484689e-01 4.16512489e-01 9.48620975e-01
-7.20973790e-01 9.99995708e-01 8.28637123e+00 6.95742786e-01
-1.08811224e+00 -5.65758944e-02 9.34897184e-01 1.77888215e-01
-7.60257721e-01 4.57936168e-01 -8.43190849e-01 3.29086244e-01
1.66869998e+00 -2.90191442e-01 3.75112891e-01 9.72708941e-01
-8.44666585e-02 -5.43433845e-01 -1.37930882e+00 5.06771863e-01
-4.06389028e-01 -1.15144193e+00 -3.76782954e-01 -5.04611611e-01
6.61941886e-01 -1.77404150e-01 -1.42060637e-01 6.31171823e-01
9.71643269e-01 -1.11737227e+00 1.06734383e+00 7.46868312e-01
9.77314711e-01 -8.88926685e-01 7.56908119e-01 6.87435269e-01
-6.90526664e-01 1.49887890e-01 -5.25736153e-01 -3.10153604e-01
-3.86089869e-02 7.23627448e-01 -1.24824691e+00 9.80368108e-02
5.61386883e-01 3.48198801e-01 -5.96397281e-01 5.41855276e-01
-6.10001624e-01 1.32357287e+00 -5.28853059e-01 -4.30772066e-01
-1.36993080e-01 9.50896591e-02 -8.15519989e-02 1.24079776e+00
1.38297468e-01 1.43754229e-01 -1.32820249e-01 9.85327244e-01
9.46574360e-02 -2.37090915e-01 -7.21808448e-02 -3.01745564e-01
7.98450947e-01 9.06962812e-01 -4.73781884e-01 -6.44524872e-01
-1.42970502e-01 5.47292233e-01 4.46882397e-01 2.20177695e-01
-7.04409480e-01 -1.69751644e-01 2.32681200e-01 -7.96487257e-02
3.97648774e-02 -7.87019283e-02 -3.36542577e-01 -9.36336219e-01
-2.72432178e-01 -1.07168829e+00 2.77166337e-01 -1.10205483e+00
-1.33526540e+00 5.73420107e-01 -8.85633081e-02 -6.42519116e-01
-7.43486881e-01 -4.69838470e-01 -4.38812166e-01 8.63681257e-01
-1.21714532e+00 -6.12224638e-01 1.01650015e-01 2.36149251e-01
-2.34299302e-02 8.40341374e-02 1.20928717e+00 -2.00738758e-01
-4.10985470e-01 8.00676823e-01 -1.67948864e-02 -7.17638433e-02
1.04821122e+00 -1.16956437e+00 3.73128712e-01 8.45519125e-01
1.88265800e-01 9.32233274e-01 1.19821262e+00 -1.00321507e+00
-1.14043927e+00 -7.05502927e-01 1.12228239e+00 -5.91980338e-01
6.68103576e-01 -1.37030616e-01 -1.12252867e+00 7.17858016e-01
-8.32600892e-03 -3.71144116e-01 7.04123020e-01 3.94214571e-01
-8.52013409e-01 -1.72000751e-01 -1.20146072e+00 3.65852028e-01
4.29005712e-01 -8.68111491e-01 -7.42059290e-01 2.82847524e-01
7.49457121e-01 -3.57404351e-01 -1.03623509e+00 2.47683465e-01
6.81580544e-01 -1.08061802e+00 4.61354196e-01 -6.41607702e-01
3.06603134e-01 -3.34324032e-01 -3.25234234e-01 -1.09083533e+00
-3.02649051e-01 -6.20323420e-01 7.45743373e-03 1.17905521e+00
8.95635605e-01 -7.25924909e-01 7.32185125e-01 1.29296565e+00
3.18928331e-01 -4.81952965e-01 -7.14348614e-01 -4.87707376e-01
3.42800647e-01 -7.27261722e-01 6.46393478e-01 6.97781682e-01
3.04721951e-01 2.30425879e-01 -3.77032787e-01 3.84066522e-01
3.98617178e-01 8.23575780e-02 6.40778422e-01 -1.20535374e+00
-6.57933176e-01 3.90872955e-02 3.22021991e-02 -7.87540734e-01
1.60695851e-01 -5.25764704e-01 4.47390765e-01 -1.05666983e+00
3.48511875e-01 -6.35837853e-01 -2.63667703e-01 7.43117332e-01
-2.34860599e-01 -9.54026878e-02 -6.01349361e-02 1.30833447e-01
-7.80033469e-01 4.03399840e-02 6.35219395e-01 1.32294849e-01
8.79848599e-02 -1.30088747e-01 -8.45954657e-01 7.02179372e-01
7.98364341e-01 -6.67500198e-01 -3.14756781e-01 -3.07776909e-02
6.61633253e-01 2.65793085e-01 1.60752147e-01 -1.02048063e+00
3.42464715e-01 -4.19545799e-01 7.27941513e-01 -3.67774874e-01
6.07894175e-02 -4.70624238e-01 4.06293869e-01 4.57987070e-01
-7.47815847e-01 -1.85781962e-03 2.72751391e-01 3.92872512e-01
-4.72229198e-02 -4.88006979e-01 7.67082870e-01 -1.00452350e-02
-6.92765296e-01 -1.49138063e-01 -6.10498548e-01 1.32549062e-01
7.14920819e-01 4.97378036e-02 -2.61774331e-01 -7.01081812e-01
-4.24058259e-01 1.27514005e-01 8.06643367e-01 1.00427702e-01
3.85462612e-01 -6.75899208e-01 -4.83563215e-01 1.76027611e-01
2.34054476e-02 -2.28122115e-01 -2.66825765e-01 2.83601612e-01
-4.83673841e-01 3.59125376e-01 1.30636156e-01 -5.42189360e-01
-9.08443570e-01 4.26500022e-01 3.93561572e-01 -2.22354174e-01
-1.61436260e-01 1.00609374e+00 -3.82381588e-01 -6.35373235e-01
2.12807551e-01 -4.35291737e-01 3.28280538e-01 -4.45455343e-01
4.78142649e-01 7.58594796e-02 1.47835277e-02 -3.10912251e-01
-4.46523428e-01 3.43969315e-01 -1.31502151e-01 -3.44156832e-01
1.01291275e+00 9.63959098e-02 1.55136302e-01 6.11655951e-01
5.45750141e-01 2.74955660e-01 -1.47750509e+00 -1.54830113e-01
2.08281890e-01 -1.85024038e-01 -1.36087209e-01 -1.18045986e+00
-4.75655586e-01 8.80247951e-01 2.11254001e-01 -9.23397094e-02
7.24997461e-01 -7.28384107e-02 4.01265562e-01 8.39668989e-01
7.84322143e-01 -1.21013141e+00 -9.37616527e-02 4.74452764e-01
4.34157312e-01 -1.18193305e+00 2.00712711e-01 -6.49180338e-02
-8.71762872e-01 1.14089620e+00 7.62369156e-01 2.02034488e-01
3.18387896e-01 5.07620335e-01 2.79343665e-01 3.75689805e-01
-1.05917430e+00 5.26393652e-01 1.55729735e-02 4.17377651e-01
5.85588872e-01 3.34610671e-01 2.44999766e-01 8.62476289e-01
-6.80217326e-01 1.19371817e-01 4.04322714e-01 7.59792924e-01
-8.23286235e-01 -1.04792953e+00 -4.72362339e-01 4.98109847e-01
-1.43105090e-01 -1.83828562e-01 -5.25077403e-01 4.97464299e-01
-8.27778801e-02 1.42823875e+00 -1.40308648e-01 -7.45854676e-01
-3.87894623e-02 4.11089301e-01 4.19902891e-01 -6.14867747e-01
-6.82506502e-01 -1.52446881e-01 4.45924431e-01 -4.94324446e-01
-2.30614141e-01 -5.72374344e-01 -1.59846699e+00 -6.37419760e-01
-6.21901810e-01 6.08733296e-01 2.53594786e-01 1.14386201e+00
2.13246852e-01 6.90863952e-02 3.84523958e-01 -3.25476199e-01
-1.01867270e+00 -8.06260169e-01 -4.01885033e-01 -1.89587120e-02
1.08701199e-01 -3.21631610e-01 -4.05016303e-01 -1.33062065e-01] | [9.939848899841309, 7.46804141998291] |
feaf396a-b900-4e2f-af8a-6b5c6b9235ae | fixed-neural-network-steganography-train-the | null | null | https://openreview.net/forum?id=hcMvApxGSzZ | https://openreview.net/pdf?id=hcMvApxGSzZ | Fixed Neural Network Steganography: Train the images, not the network | Recent attempts at image steganography make use of advances in deep learning to train an encoder-decoder network pair to hide and retrieve secret messages in images. These methods are able to hide large amounts of data, but also incur high decoding error rates (around 20\%). In this paper, we propose a novel algorithm for steganography that takes advantage of the fact that neural networks are sensitive to tiny perturbations. Our method, Fixed Neural Network Steganography (FNNS), achieves 0\% error reliably for hiding up to 3 bits per pixel (bpp) of secret information in images and yields significantly lower error rates when compared to prior state of the art methods for hiding more than 3 bpp. FNNS also successfully evades existing statistical steganalysis systems and can be modified to evade neural steganalysis systems as well. Recovering every bit correctly, up to 3 bpp, enables novel applications, e.g. those requiring encryption.
We introduce one specific use case for facilitating anonymized and safe image sharing.
| ['Kilian Q Weinberger', 'Boyi Li', 'Yan Wang', 'Xiangyu Chen', 'Varsha Kishore'] | 2021-09-29 | null | null | null | iclr-2022-4 | ['steganalysis', 'image-steganography'] | ['computer-vision', 'computer-vision'] | [ 1.00715888e+00 5.61542511e-01 5.33385538e-02 -4.98870052e-02
-7.04645038e-01 -4.80237395e-01 4.34811980e-01 -3.15230429e-01
-5.11953831e-01 7.21975982e-01 -1.98974878e-01 -7.63964236e-01
5.52585959e-01 -7.98809469e-01 -1.25695789e+00 -1.06899571e+00
-5.90477943e-01 1.79311130e-02 7.33825713e-02 -3.77331197e-01
1.65017188e-01 3.14490616e-01 -1.28435659e+00 3.97802502e-01
4.29478228e-01 7.93442488e-01 -2.18647629e-01 9.44589913e-01
4.59773451e-01 7.08845556e-01 -7.51640856e-01 -5.20450473e-01
5.74905694e-01 -6.44772470e-01 -6.22143090e-01 -3.51343095e-01
2.99701333e-01 -7.65326977e-01 -7.85926998e-01 1.26161838e+00
4.40946966e-01 -7.05169320e-01 3.42970490e-01 -1.26394331e+00
-7.42117643e-01 8.92295003e-01 -2.66058058e-01 -1.69576615e-01
-8.18586498e-02 4.21448559e-01 5.69356203e-01 -2.26114810e-01
4.75188971e-01 1.01896393e+00 8.07424545e-01 1.06994331e+00
-1.18651450e+00 -1.31580579e+00 -6.88444018e-01 1.68592900e-01
-1.39922667e+00 -9.08205748e-01 5.79294443e-01 -2.33614668e-02
9.44603920e-01 3.49481851e-01 6.33048534e-01 1.22207832e+00
6.87294304e-01 6.55596972e-01 1.16380477e+00 -3.57319355e-01
-3.88043039e-02 3.03242266e-01 -6.91443682e-01 7.97716916e-01
5.76566696e-01 5.10385752e-01 -3.89519781e-01 8.80519077e-02
6.77791953e-01 -1.55282989e-01 -5.07323742e-01 -3.78157079e-01
-1.29825079e+00 1.07950270e+00 3.93719196e-01 2.50347763e-01
2.35720217e-01 8.66426587e-01 4.49350417e-01 9.89057183e-01
1.46087840e-01 4.43970799e-01 -2.21776843e-01 2.77900517e-01
-9.00099814e-01 -8.04544389e-02 1.18228710e+00 8.42754185e-01
6.47777319e-01 4.74750668e-01 4.30322498e-01 -3.35686351e-03
3.19354355e-01 7.52120197e-01 3.11122864e-01 -8.25116038e-01
4.60648656e-01 -1.18975654e-01 -2.60140181e-01 -1.37053156e+00
1.14246927e-01 3.39821465e-02 -1.38670611e+00 5.36967516e-01
7.07792640e-02 -1.87801376e-01 -1.08523309e+00 1.58789492e+00
-1.35568723e-01 5.79532236e-02 5.72829187e-01 2.80933976e-01
5.98077416e-01 6.54625475e-01 -3.52609575e-01 8.65854397e-02
1.09581339e+00 -4.29363459e-01 -6.02312982e-01 -3.77950996e-01
9.41066384e-01 -4.99714613e-01 2.60171682e-01 1.93176508e-01
-1.01819444e+00 -9.70460251e-02 -1.50443733e+00 1.77101791e-01
-4.29939896e-01 -7.58149087e-01 2.85934478e-01 1.40735495e+00
-1.38436759e+00 7.36745954e-01 -7.47351766e-01 2.36342281e-01
9.53362644e-01 1.09557557e+00 -7.14358509e-01 -9.96024534e-02
-1.60391688e+00 7.35168695e-01 8.00992489e-01 -9.21801254e-02
-1.20535135e+00 -4.71724123e-01 -1.29899359e+00 7.95500726e-02
5.70576601e-02 -3.20285141e-01 7.30731666e-01 -1.20322216e+00
-1.43360102e+00 1.12631500e+00 1.67514309e-01 -1.14705181e+00
5.08747101e-01 2.67429560e-01 -5.91662467e-01 2.99141794e-01
-5.38641393e-01 9.38070953e-01 1.40948582e+00 -1.19179165e+00
-2.34033778e-01 -1.40870351e-03 -2.56943405e-01 -3.71938109e-01
-4.17724103e-01 3.99312750e-02 1.65435001e-01 -5.34165919e-01
-1.35539755e-01 -1.15447652e+00 -8.39433149e-02 1.99974284e-01
-5.31447947e-01 5.60858727e-01 1.28038526e+00 -8.75488281e-01
6.81680977e-01 -2.14226365e+00 7.11279958e-02 4.06084418e-01
6.66048825e-01 8.61072183e-01 -1.92480728e-01 4.11836684e-01
-3.08700264e-01 5.83141685e-01 -6.73257530e-01 -3.79386872e-01
-2.24889163e-02 2.45206103e-01 -2.77199417e-01 8.45906854e-01
3.20796156e-04 1.24451578e+00 -6.44569576e-01 -2.82283634e-01
2.37223119e-01 7.99288511e-01 -4.43195999e-01 -1.97508678e-01
-3.83546203e-02 4.64271873e-01 7.85418227e-02 4.28557336e-01
9.63313460e-01 -2.60571867e-01 3.91204327e-01 3.37902158e-01
3.26485634e-01 1.45214289e-01 -4.70859051e-01 1.07921457e+00
-9.20346230e-02 1.21855056e+00 1.81389332e-01 -1.14951158e+00
7.68185675e-01 6.27292454e-01 3.04933518e-01 -4.55807656e-01
5.21172345e-01 4.43284214e-01 1.30561441e-01 -3.24491441e-01
2.18486875e-01 -1.49665356e-01 -1.67013869e-01 7.09631622e-01
-5.06575070e-02 -9.33775902e-02 -5.75010300e-01 -5.17703369e-02
1.22751272e+00 -4.52275097e-01 3.22737962e-01 -4.98817600e-02
3.96267742e-01 -4.71977413e-01 1.57707617e-01 9.25798237e-01
-2.49147803e-01 4.53095555e-01 6.11532331e-01 -4.51149821e-01
-1.53577018e+00 -4.53726381e-01 2.20186308e-01 4.89825726e-01
1.34116128e-01 -1.51635811e-01 -1.00192571e+00 -6.27486944e-01
3.17838714e-02 4.07586485e-01 -4.85802561e-01 -6.04557991e-01
-7.65616834e-01 -4.61318344e-01 1.33077013e+00 -1.52039491e-02
1.02485824e+00 -1.21019566e+00 -5.03920436e-01 -4.49895076e-02
-7.62240263e-03 -1.14422607e+00 -2.33883530e-01 1.05123788e-01
-6.24967515e-01 -8.18936288e-01 -7.49010861e-01 -9.98940587e-01
6.36762798e-01 1.82940617e-01 7.14479268e-01 4.87376302e-01
-9.01605263e-02 -1.51622206e-01 -1.82531059e-01 -4.31710541e-01
-1.36485720e+00 1.53602675e-01 -5.74722216e-02 -4.61224653e-02
3.80219012e-01 -6.80802345e-01 -4.46418136e-01 2.16674641e-01
-1.22666073e+00 4.51244153e-02 8.35343778e-01 8.75975430e-01
2.03199163e-01 4.19978619e-01 1.49027318e-01 -1.20920527e+00
1.04752019e-01 -4.06042963e-01 -7.68834531e-01 2.47681476e-02
-7.77619839e-01 2.02336967e-01 6.49002135e-01 -3.64001095e-01
-9.68412086e-02 -1.79902568e-01 -3.71393472e-01 -3.41320783e-01
1.71689820e-02 8.50076824e-02 -1.83804661e-01 -1.10879898e+00
4.34050053e-01 6.86682403e-01 5.67380130e-01 9.97257158e-02
8.88987258e-03 8.05054188e-01 2.66947091e-01 4.06630814e-01
1.26204538e+00 6.11946404e-01 2.66304523e-01 -7.01510012e-01
-1.51110783e-01 2.02774957e-01 -4.38455582e-01 3.22836339e-01
6.76797569e-01 -9.87226307e-01 -9.68355060e-01 1.14482975e+00
-1.03494811e+00 -3.75241905e-01 5.38911845e-04 6.54881448e-02
-6.83348536e-01 6.38034046e-01 -4.51968938e-01 -4.92177457e-01
-5.08177280e-01 -1.12792742e+00 7.36626089e-01 -3.60228986e-01
2.48992071e-02 -1.18292212e+00 -4.50713813e-01 1.51537389e-01
6.01829469e-01 7.18488455e-01 7.42102742e-01 -5.79465628e-01
-1.01180267e+00 -5.62030375e-01 -2.00390682e-01 8.07068288e-01
-9.63998085e-04 -6.74495518e-01 -8.34817886e-01 -9.08127785e-01
2.77094573e-01 -3.39707732e-01 1.20185030e+00 1.19297408e-01
1.34911048e+00 -1.07553315e+00 -3.38063151e-01 1.28949368e+00
1.49165869e+00 1.56476945e-01 1.52102816e+00 3.90996069e-01
9.09163475e-01 4.22919869e-01 -3.82692575e-01 9.88321081e-02
4.23536077e-02 2.41861835e-01 6.42349005e-01 -3.42042521e-02
-3.21286805e-02 -2.27983370e-01 6.32817566e-01 7.37682700e-01
1.41374305e-01 -7.35486507e-01 -7.59581625e-01 4.34548050e-01
-1.40028572e+00 -1.08886337e+00 8.46343637e-02 2.12076402e+00
9.60943997e-01 2.67562211e-01 -4.25391525e-01 4.13471878e-01
6.66027546e-01 6.28221095e-01 -5.28979123e-01 -4.98165578e-01
-2.45076895e-01 4.31123137e-01 1.49007571e+00 4.21094388e-01
-1.21542311e+00 1.01533961e+00 7.16206551e+00 7.19933391e-01
-1.16658032e+00 1.19541131e-01 5.90120316e-01 9.92850736e-02
-3.76491159e-01 -1.38145357e-01 -6.15130723e-01 6.72865570e-01
1.43094695e+00 8.73000100e-02 6.64656520e-01 3.72915834e-01
-4.91621792e-01 2.92957366e-01 -9.72120702e-01 1.06708241e+00
4.76067126e-01 -1.61602116e+00 1.59037545e-01 6.38071299e-01
9.10108685e-01 7.04839546e-03 6.01861477e-01 -3.49398702e-01
3.79761696e-01 -1.53646553e+00 3.03987682e-01 1.19665094e-01
1.43794918e+00 -9.70806181e-01 9.14966583e-01 2.11427018e-01
-5.56177020e-01 7.67593319e-03 -5.95263898e-01 2.36869380e-01
-1.79665908e-02 1.78618014e-01 -1.04971731e+00 8.02516490e-02
5.95882595e-01 8.42921555e-01 -1.86065048e-01 5.08152843e-01
-2.74996251e-01 8.28871012e-01 -2.98245698e-01 -1.33815706e-01
3.55761319e-01 3.95376205e-01 6.65981591e-01 9.29435551e-01
6.13922417e-01 -1.33719817e-01 -4.06952441e-01 6.66926622e-01
-5.78910589e-01 -5.95674574e-01 -1.04874039e+00 -3.42670470e-01
4.43287551e-01 4.58135545e-01 -5.00277340e-01 -1.59801126e-01
6.87050968e-02 1.36287212e+00 -2.05446675e-01 1.13211900e-01
-5.53657889e-01 -8.64082873e-01 7.68829644e-01 1.84257571e-02
1.01633346e+00 -3.09717298e-01 -1.95102930e-01 -1.09587419e+00
-1.81526661e-01 -1.30391252e+00 -2.39439622e-01 -2.84003943e-01
-6.23819292e-01 3.05636793e-01 -4.17314887e-01 -1.18972290e+00
-3.30636978e-01 -6.28551900e-01 -2.10597575e-01 7.52958834e-01
-1.76943767e+00 -1.19887948e+00 1.81328982e-01 6.56391621e-01
-9.42483544e-02 -6.43536389e-01 1.18166769e+00 5.28482907e-03
-2.47527465e-01 1.05908358e+00 6.40871882e-01 5.64322293e-01
3.56787443e-01 -8.04475904e-01 1.10886514e+00 8.30801308e-01
-1.38687953e-01 4.40214396e-01 8.36823523e-01 -5.26534140e-01
-1.55743468e+00 -1.14271963e+00 9.14049029e-01 -3.33776981e-01
4.74407285e-01 -7.98947453e-01 -7.75471866e-01 9.65455770e-01
3.00542861e-01 -4.02083714e-03 7.61892438e-01 -6.84586525e-01
-7.61100888e-01 8.98498446e-02 -1.58756149e+00 5.77396154e-01
7.95226693e-01 -7.55357504e-01 -1.09372176e-01 3.12121689e-01
1.00894535e+00 -4.60879385e-01 -5.62411904e-01 6.84271455e-02
7.65949965e-01 -1.11596692e+00 1.12013233e+00 -2.22244039e-01
6.16217077e-01 4.64419760e-02 -9.43666622e-02 -1.07059348e+00
1.31637342e-02 -1.37333465e+00 -5.04203558e-01 6.16855741e-01
2.46602371e-01 -9.69755650e-01 1.05725324e+00 7.96707198e-02
3.68247092e-01 -2.14976057e-01 -1.15457225e+00 -7.18884706e-01
2.07578599e-01 -2.24022701e-01 8.02546084e-01 9.20120656e-01
-3.23471248e-01 -4.87485707e-01 -1.25016820e+00 3.55866224e-01
1.15602148e+00 -5.30196667e-01 7.24090576e-01 -9.44891751e-01
-1.00327969e-01 -1.13036595e-01 -1.06315529e+00 -9.07117486e-01
4.35558349e-01 -9.16531801e-01 5.38591594e-02 -8.46289635e-01
-1.66569024e-01 -3.61372083e-01 -2.39288077e-01 5.89591742e-01
4.38115209e-01 1.01930511e+00 1.06626349e-02 2.77177840e-01
-1.50433257e-01 1.71213374e-01 1.09540474e+00 -3.51648569e-01
2.25680888e-01 -1.18691809e-01 -9.34273005e-01 5.12494266e-01
1.08359277e+00 -1.01861227e+00 -1.43140644e-01 -5.11939049e-01
2.70455718e-01 -2.81094968e-01 6.79470241e-01 -1.21851861e+00
1.60436615e-01 1.23928502e-01 3.92940134e-01 -1.84221998e-01
3.56735855e-01 -1.07096553e+00 3.44643772e-01 1.17613244e+00
-4.06966805e-01 -4.74232942e-01 1.39927596e-01 6.45128012e-01
2.66536102e-02 -1.54622123e-01 9.94482815e-01 -4.60333884e-01
-6.46945834e-01 3.06800455e-01 -5.58290958e-01 -2.34040320e-01
1.11440921e+00 -6.02819622e-01 -3.09433430e-01 -7.58693576e-01
-3.28930467e-01 -4.76375706e-02 6.36520445e-01 1.32612124e-01
1.18164551e+00 -1.09194887e+00 -7.75275052e-01 8.51624668e-01
-8.27638134e-02 -4.47228819e-01 1.56792507e-01 5.25797546e-01
-1.09545302e+00 5.73468745e-01 -3.33794296e-01 -3.37610334e-01
-1.56292272e+00 5.56943715e-01 4.25503403e-01 -8.00649971e-02
-8.75266135e-01 1.08694100e+00 -1.65145248e-01 -1.10888153e-01
2.12413877e-01 3.21092419e-02 2.03701258e-01 -5.06751597e-01
8.58404815e-01 -4.42970097e-02 -2.79835492e-01 -7.83687711e-01
-9.88905504e-02 4.14084524e-01 -3.80603075e-01 3.03049088e-02
1.32091618e+00 -2.23695323e-01 -3.79562825e-01 -5.08366525e-02
1.80470860e+00 -2.79494375e-01 -1.26266778e+00 -1.68557733e-01
-3.72281581e-01 -6.38257205e-01 7.69055188e-02 -3.55639696e-01
-1.19939339e+00 9.22949612e-01 6.87606156e-01 3.89587611e-01
9.07681704e-01 -2.07388237e-01 1.61451674e+00 7.56322980e-01
4.73294020e-01 -5.78869224e-01 -2.02891812e-01 4.79695141e-01
2.82942146e-01 -1.32698154e+00 -8.93660262e-02 -2.06325963e-01
-2.58746475e-01 9.83707786e-01 -2.12719575e-01 -2.74499834e-01
7.82156944e-01 4.42203879e-01 -1.79946069e-02 -1.02296375e-01
-4.11465734e-01 4.02875781e-01 -1.38873784e-02 1.04805696e+00
-2.26334408e-01 1.29838409e-02 3.35262567e-01 -3.60945523e-01
-4.70248282e-01 -1.17304586e-01 7.34350920e-01 1.09076071e+00
-5.05317867e-01 -1.13339293e+00 -3.29526454e-01 1.93467662e-01
-8.16690326e-01 -3.60191405e-01 -2.94088125e-01 6.54735863e-01
8.19893032e-02 6.95537388e-01 -2.51900367e-02 -8.09706688e-01
-5.16194463e-01 -1.49000451e-01 3.83674622e-01 -2.64250338e-01
-4.16211516e-01 -5.09860992e-01 -1.83556765e-01 -5.81129909e-01
-6.15394294e-01 -3.58358294e-01 -9.07180429e-01 -1.21814239e+00
-2.39172563e-01 -1.63759470e-01 7.81033933e-01 9.15600955e-01
3.18794191e-01 2.76156902e-01 8.22849095e-01 -7.37895250e-01
-4.31960613e-01 -3.86246175e-01 -7.18912899e-01 7.20666274e-02
1.18659604e+00 1.08785786e-01 -7.76748061e-01 2.28328362e-01] | [4.358722686767578, 8.035676002502441] |
3440fa0e-716f-43f9-9e2b-d83f762c7b4e | robust-direction-of-arrival-estimation-using | 2303.09053 | null | https://arxiv.org/abs/2303.09053v2 | https://arxiv.org/pdf/2303.09053v2.pdf | Robust Direction-of-Arrival Estimation using Array Feedback Beamforming in Low SNR Scenarios | A new spatial IIR beamformer based direction-of-arrival (DoA) estimation method is proposed in this paper. We propose a retransmission based spatial feedback method for an array of transmit and receive antennas that improves the performance parameters of a beamformer, viz. half-power beamwidth (HPBW), side-lobe suppression, and directivity. Through quantitative comparison, we show that our approach outperforms the previous feedback beamforming approach with a single transmit antenna, and the conventional beamformer. We then incorporate a retransmission based minimum variance distortionless response (MVDR) beamformer with the feedback beamforming setup. We propose two approaches, show that one approach is superior in terms of lower estimation error, and use that as the DoA estimation method. We then compare this approach with Multiple Signal Classification (MUSIC), Estimation of Parameters using Rotation Invariant Technique (ESPRIT), robust MVDR, nested-array MVDR, and reduced-dimension MVDR methods. The results show that at SNR levels of -60 dB to -10 dB, the angle estiation error of the proposed method is 20 degree less compared to that of prior methods. | ['Rajbabu Velmurugan', 'Kumar Appaiah', 'Parth Mehta'] | 2023-03-16 | null | null | null | null | ['direction-of-arrival-estimation'] | ['audio'] | [ 2.03959331e-01 -3.30720633e-01 4.63402838e-01 1.50238527e-02
-8.51403236e-01 -9.44162846e-01 4.80403453e-01 -3.76752704e-01
-2.32101172e-01 7.97331989e-01 1.11076677e+00 -5.22509336e-01
-8.64858449e-01 -4.00040835e-01 -1.09760828e-01 -1.06256878e+00
-2.72469312e-01 -4.58272964e-01 -2.40427982e-02 -1.37557045e-01
2.23183885e-01 8.86244416e-01 -9.87834573e-01 -1.34045646e-01
2.38124549e-01 1.08765352e+00 1.16263106e-01 1.10765874e+00
5.45941889e-01 4.39861506e-01 -6.71144664e-01 2.59463668e-01
5.94151020e-01 -4.04578835e-01 1.31702915e-01 -2.83671200e-01
3.82024139e-01 -3.48008960e-01 -4.62354600e-01 5.84071755e-01
1.24738109e+00 2.08401874e-01 7.64360726e-01 -7.09965289e-01
-1.70139328e-01 4.19487536e-01 -7.78304279e-01 4.55956131e-01
6.80807889e-01 -5.79486012e-01 3.49963844e-01 -1.31616020e+00
3.62062633e-01 1.07809746e+00 8.45982492e-01 -1.46886602e-01
-1.03763187e+00 -6.42240882e-01 -4.19717699e-01 -3.17718722e-02
-1.79434931e+00 -8.17680180e-01 6.71754003e-01 -3.47626716e-01
5.01304448e-01 7.85245299e-01 3.51410151e-01 6.04583502e-01
2.41728455e-01 2.17165858e-01 1.19727957e+00 -6.80920303e-01
3.53393495e-01 -3.35909426e-01 -1.20776631e-01 3.17096025e-01
4.78015959e-01 3.70065808e-01 -6.32201314e-01 -3.19635510e-01
9.34373796e-01 -2.30153173e-01 -8.68809700e-01 -2.78812110e-01
-1.50900531e+00 5.27031124e-01 1.26593187e-01 7.79847860e-01
-5.24942994e-01 3.06170970e-01 -2.70449996e-01 5.55268884e-01
2.92361408e-01 2.97013581e-01 -2.51450002e-01 2.13852033e-01
-1.26626754e+00 1.24317348e-01 8.31192911e-01 8.45240653e-01
1.40449226e-01 6.08629167e-01 -4.75228995e-01 7.86586463e-01
9.43753600e-01 1.25939107e+00 2.46368274e-01 -8.52685273e-01
4.68107998e-01 -4.58134979e-01 4.62872356e-01 -1.31749022e+00
-5.96906483e-01 -1.41121197e+00 -1.16696846e+00 1.07808031e-01
2.04683527e-01 -7.25707173e-01 -3.91909838e-01 1.36472273e+00
2.62658060e-01 2.66790539e-01 2.15382695e-01 1.05242181e+00
5.66107333e-01 8.77791226e-01 -8.96913826e-01 -1.03103602e+00
9.40876186e-01 -5.13082147e-01 -1.08437383e+00 -5.25226220e-02
3.31606448e-01 -1.37155724e+00 -1.49594635e-01 1.06640494e+00
-9.40806448e-01 -4.98846322e-01 -1.20739019e+00 7.74233699e-01
3.69529992e-01 2.60913521e-01 2.95602173e-01 1.20314693e+00
-1.22355139e+00 6.12892676e-03 -2.84227669e-01 -1.12395816e-01
-3.46387923e-01 1.39607579e-01 -3.85097146e-01 -1.30861565e-01
-5.41928172e-01 5.85099816e-01 -3.15223575e-01 1.64855242e-01
-3.47449601e-01 -9.40095067e-01 -5.65111339e-01 -5.61192520e-02
-2.03260347e-01 -7.89410949e-01 9.72633064e-01 -6.17276132e-01
-1.56108749e+00 -1.08265474e-01 -3.58258098e-01 -3.10925812e-01
5.69977313e-02 -5.05274296e-01 -7.30741024e-01 1.15637794e-01
-2.46849805e-01 -6.26477823e-02 8.99388611e-01 -1.37117147e+00
-4.12259907e-01 -2.37783223e-01 -6.97057486e-01 2.24056140e-01
1.66146502e-01 5.63257486e-02 2.78490633e-01 -1.10574293e+00
8.51974368e-01 -4.79601651e-01 -3.05222541e-01 -2.16990381e-01
2.40192451e-02 4.10944611e-01 5.29534578e-01 -6.71724737e-01
1.62262976e+00 -2.25753593e+00 1.85098834e-02 5.88179946e-01
-1.34801567e-01 3.17186192e-02 -1.40294641e-01 8.58427763e-01
-3.86783749e-01 -1.89824939e-01 2.81282932e-01 1.02381006e-01
-3.99234623e-01 -1.33340865e-01 -4.21142697e-01 1.05387461e+00
-8.16069126e-01 1.31912783e-01 -5.71132123e-01 1.47486880e-01
1.97281003e-01 5.24901688e-01 -5.42055786e-01 2.91421741e-01
9.73147929e-01 4.95246321e-01 -2.48223364e-01 4.93008405e-01
1.02120018e+00 3.80027354e-01 -9.59111452e-02 -6.43837690e-01
-7.46113896e-01 -1.81028768e-01 -2.16330028e+00 1.53622687e+00
-6.56633735e-01 9.15823579e-01 5.12531936e-01 -7.20070422e-01
1.21291852e+00 6.99473500e-01 4.93454099e-01 -5.59013307e-01
1.10568315e-01 5.12633443e-01 9.07197744e-02 -3.69948715e-01
5.70428558e-02 -1.38935640e-01 2.84012228e-01 5.06115854e-01
4.83586192e-02 5.81013858e-02 -2.80709803e-01 3.63087282e-02
1.34706223e+00 -1.85642824e-01 7.82704949e-01 -5.49617708e-01
7.16430545e-01 -8.08368444e-01 4.53160137e-01 8.85026932e-01
2.02438414e-01 8.86173129e-01 -4.43423301e-01 -3.13221128e-03
-7.93698311e-01 -1.03615332e+00 -9.87306088e-02 7.30807424e-01
1.57256499e-02 -1.65594146e-01 -1.41321450e-01 2.70036787e-01
-3.33994403e-02 8.82389426e-01 -1.76450938e-01 4.44423050e-01
-6.59003079e-01 -6.99566126e-01 2.88863361e-01 1.41106114e-01
3.28118443e-01 3.10463369e-01 -4.51226979e-01 3.71635377e-01
-1.04824841e-01 -5.94207644e-01 -3.36163759e-01 -7.35258870e-03
-8.51814032e-01 -4.35164601e-01 -1.02890420e+00 -2.88226515e-01
4.11654294e-01 9.18120742e-01 6.21779501e-01 -3.83337885e-01
6.64917007e-02 9.61844087e-01 -7.71170795e-01 -1.87272966e-01
2.15367526e-01 -7.64194369e-01 3.28965008e-01 3.84761035e-01
-5.43708503e-01 -1.05996072e+00 -1.11810887e+00 5.90834260e-01
-5.12332976e-01 -2.21354723e-01 1.00164855e+00 5.01709938e-01
2.18297150e-02 -3.19188058e-01 6.70145094e-01 -1.41559411e-02
6.03227913e-01 -4.84846383e-01 -4.76024896e-01 -1.55974030e-01
-5.28914750e-01 -2.47075651e-02 2.87005246e-01 -1.30596206e-01
-1.22564948e+00 2.14434769e-02 -2.41994575e-01 -1.06752785e-02
-6.07742416e-03 4.75885898e-01 4.08246554e-02 -4.73413318e-01
9.18573260e-01 2.94575155e-01 -4.94428843e-01 -6.69338405e-01
4.56524789e-01 9.53433156e-01 4.91865814e-01 1.14116095e-01
1.15823185e+00 3.74443114e-01 5.04800081e-01 -1.08370900e+00
-4.09350336e-01 -1.12128377e+00 -4.30584311e-01 -4.24445063e-01
3.07171136e-01 -1.02911270e+00 -3.13726127e-01 2.36667739e-03
-1.16811121e+00 2.11699113e-01 2.55772293e-01 1.40659487e+00
-3.57172310e-01 3.96352470e-01 1.07203111e-01 -1.31667912e+00
-4.78503555e-01 -5.77971518e-01 8.44001055e-01 -9.86455083e-02
-1.20849706e-01 -8.12466085e-01 3.49274397e-01 -6.60612583e-02
8.92815411e-01 2.54092455e-01 1.49842486e-01 -3.24585021e-01
-4.81014013e-01 -2.03990996e-01 1.04839459e-01 7.69410506e-02
1.35545239e-01 -7.40233243e-01 -6.45115256e-01 -6.04832888e-01
2.21755788e-01 4.73447740e-01 3.92759919e-01 8.50263476e-01
3.03087145e-01 -5.02334714e-01 -3.89905959e-01 6.93043768e-01
1.98306823e+00 3.22767347e-01 7.73428202e-01 5.22845685e-02
-8.82930458e-02 -1.58394188e-01 7.01132894e-01 8.96024466e-01
-1.68064013e-01 7.64213145e-01 3.22549015e-01 9.07385722e-02
-5.16905844e-01 3.05833548e-01 1.91614226e-01 1.07545614e+00
-1.20864339e-01 -9.05127347e-01 -4.54214245e-01 3.12146366e-01
-1.63851321e+00 -9.65124011e-01 -6.36942148e-01 2.50152493e+00
2.84240156e-01 -4.00784463e-01 -1.81784272e-01 5.40435016e-01
2.36416042e-01 3.90017517e-02 2.41234183e-01 -3.49423349e-01
-1.89462021e-01 3.15719724e-01 1.15057898e+00 8.74788880e-01
-8.48273396e-01 -8.78032669e-02 6.36606359e+00 5.34757733e-01
-8.53484809e-01 3.04948151e-01 -2.08083913e-01 1.31369784e-01
-4.81971353e-01 -1.11156747e-01 -5.74134111e-01 1.44075096e-01
7.81792819e-01 -1.01375412e-02 3.22592795e-01 2.89835125e-01
7.93971539e-01 -4.83088076e-01 -7.65995502e-01 1.54291868e+00
5.16269505e-01 -1.03685737e+00 -3.03178519e-01 7.05043525e-02
6.83107436e-01 -2.89550543e-01 -2.03579336e-01 -2.85165727e-01
-6.54514208e-02 -7.99886525e-01 6.82042837e-01 1.13200319e+00
4.02325362e-01 -5.91278732e-01 1.00887036e+00 5.28080940e-01
-1.26419115e+00 -3.22580010e-01 -1.41602680e-01 -2.81414628e-01
4.31242168e-01 1.12238884e+00 -7.37865210e-01 1.01459992e+00
2.79503077e-01 6.98292702e-02 -1.68187022e-01 1.89218223e+00
-8.24679211e-02 1.02723086e+00 -6.17381811e-01 -8.52189213e-02
-5.35267480e-02 -1.63271710e-01 1.05349696e+00 1.35021877e+00
1.13587475e+00 7.94217825e-01 -1.23930611e-01 -7.89000001e-03
5.53323209e-01 2.96270907e-01 -6.06126189e-01 7.76587486e-01
7.63865948e-01 1.00361443e+00 -2.55248487e-01 6.70976043e-02
-3.85634065e-01 6.93786383e-01 -9.44926143e-01 5.26420414e-01
-2.51419157e-01 -5.98093808e-01 1.52113363e-01 1.67764872e-01
5.09153008e-01 -8.49315047e-01 -2.73466319e-01 -7.49177158e-01
-2.69280851e-01 -6.74568117e-01 7.70398676e-02 -1.13449204e+00
-6.52517140e-01 2.78511643e-01 2.83104572e-02 -1.76153696e+00
-5.16145229e-01 -4.04428065e-01 -3.62059116e-01 1.12119865e+00
-1.09687269e+00 -9.06760216e-01 -3.09012104e-02 4.52313095e-01
4.03835416e-01 -5.10646880e-01 6.22452974e-01 6.26685917e-01
2.30012611e-01 5.20841956e-01 5.94374895e-01 -1.63334310e-01
7.43957400e-01 -9.17469263e-01 -6.10284388e-01 1.19288909e+00
6.05821013e-02 6.49012387e-01 1.34808040e+00 -1.95514098e-01
-1.95346487e+00 -7.71747828e-01 7.56681681e-01 5.52263185e-02
4.89272356e-01 -1.56736746e-01 -8.39305744e-02 3.43911618e-01
5.78826368e-01 -7.44264573e-02 1.19106483e+00 -1.38775885e-01
9.13326279e-04 -5.03519654e-01 -9.84545827e-01 3.46333086e-01
9.23713446e-01 3.25613976e-01 -7.79598892e-01 2.29668573e-01
9.28725377e-02 -1.06710546e-01 -8.69957447e-01 6.10800803e-01
8.40578258e-01 -1.00346828e+00 1.15185702e+00 2.91109532e-01
-5.33896029e-01 -8.95539641e-01 -9.31286395e-01 -1.51556575e+00
-9.63826716e-01 -1.04307008e+00 6.23280928e-02 1.15611696e+00
3.37877512e-01 -6.36283338e-01 1.17265664e-01 -6.39272571e-01
-2.89097488e-01 -2.15492487e-01 -1.15655959e+00 -7.87288606e-01
-4.97475982e-01 -5.29691100e-01 -5.71701191e-02 5.00232220e-01
-2.55454659e-01 5.56035519e-01 -6.58811331e-01 8.62188876e-01
8.59139025e-01 3.29651423e-02 8.50852787e-01 -1.02920878e+00
-7.28066087e-01 -5.74483909e-02 -5.60553253e-01 -1.54555643e+00
-7.95042396e-01 -4.63398039e-01 -7.52782747e-02 -1.83957124e+00
-5.13388038e-01 -2.98287958e-01 -1.92600608e-01 -7.32009411e-02
5.13238132e-01 3.82137746e-01 1.10226413e-02 -1.66103020e-01
-1.28686517e-01 1.54422343e-01 8.39595556e-01 1.76232234e-01
-1.50863022e-01 5.83901942e-01 -4.89345461e-01 5.70095479e-01
4.39935923e-01 -4.64219391e-01 -3.91622871e-01 -4.13364887e-01
6.19648218e-01 7.51306653e-01 2.60140151e-01 -1.29788804e+00
4.81327862e-01 1.14552736e-01 6.85415149e-01 -9.27582860e-01
3.15172762e-01 -8.50191534e-01 6.35134578e-01 5.45607388e-01
3.19455191e-02 3.61342262e-03 -8.83329362e-02 6.71429634e-01
-2.23739240e-02 -2.27549359e-01 6.00665748e-01 2.18202129e-01
-2.03209281e-01 -3.66819173e-01 -8.68860722e-01 -6.50806904e-01
6.58919215e-01 -3.53645325e-01 1.28362468e-02 -9.90266204e-01
-6.83668375e-01 -3.58792722e-01 -4.72253323e-01 5.20180799e-02
6.36442006e-01 -1.50357163e+00 -1.14586568e+00 6.23947866e-02
-3.72129470e-01 -6.91685379e-01 8.16116557e-02 1.16605425e+00
-4.26952511e-01 6.97224021e-01 7.87601098e-02 -6.24968588e-01
-1.50949228e+00 9.08466727e-02 3.21413547e-01 2.37931401e-01
-1.32195801e-01 8.58572781e-01 -1.05320551e-01 1.22465916e-01
1.26076818e-01 -1.02108009e-01 -4.41592485e-01 4.71825488e-02
7.61812806e-01 6.51773095e-01 3.64437073e-01 -6.21495485e-01
-4.43011671e-01 9.45337772e-01 8.76316547e-01 -8.59579384e-01
1.24943841e+00 -6.08161271e-01 -2.59594973e-02 1.50117636e-01
1.08953536e+00 1.13315332e+00 -5.62932014e-01 -5.32081462e-02
-3.70003194e-01 -9.50228512e-01 5.68359256e-01 -1.03643751e+00
-6.61091268e-01 5.48934102e-01 1.37732089e+00 1.07154176e-01
1.45987463e+00 -2.36408815e-01 1.16663915e-03 5.35997868e-01
5.90833962e-01 -5.27229726e-01 -2.71912478e-03 2.87908375e-01
1.38796449e+00 -3.37526888e-01 3.43237668e-01 -4.68305618e-01
2.85848230e-01 1.44497728e+00 -2.61059254e-01 -1.48412377e-01
1.04553485e+00 5.60700834e-01 3.49120088e-02 2.60678738e-01
-5.37099957e-01 -1.79142579e-01 5.07685483e-01 7.77125955e-01
6.35899842e-01 1.42038986e-02 -8.77584934e-01 4.49876189e-01
-3.28877062e-01 -1.62505254e-01 6.58751547e-01 8.98981571e-01
-1.12137640e+00 -9.60655868e-01 -1.17792630e+00 2.60816008e-01
-3.46262813e-01 -2.15675637e-01 1.85218811e-01 4.57022309e-01
5.40196970e-02 1.45827568e+00 -4.87762652e-02 -3.34835529e-01
6.29748464e-01 -4.01878476e-01 6.62308931e-01 -2.92551577e-01
-2.25848824e-01 7.50633836e-01 2.59719014e-01 -1.91323027e-01
-4.04498905e-01 -8.09500396e-01 -5.40066242e-01 1.66039959e-01
-4.92384583e-01 2.84755856e-01 8.18330646e-01 8.35471451e-01
3.97091955e-01 4.70897108e-01 1.21222341e+00 -6.33915126e-01
-4.16751385e-01 -1.00472975e+00 -6.60091817e-01 -4.50433165e-01
5.88301003e-01 -5.35922885e-01 -4.37498808e-01 -4.76811640e-02] | [6.477575778961182, 1.3512932062149048] |
772facde-cbd1-4b10-bb54-e9a2a74a7351 | chinese-lexical-simplification | 2010.07048 | null | https://arxiv.org/abs/2010.07048v1 | https://arxiv.org/pdf/2010.07048v1.pdf | Chinese Lexical Simplification | Lexical simplification has attracted much attention in many languages, which is the process of replacing complex words in a given sentence with simpler alternatives of equivalent meaning. Although the richness of vocabulary in Chinese makes the text very difficult to read for children and non-native speakers, there is no research work for Chinese lexical simplification (CLS) task. To circumvent difficulties in acquiring annotations, we manually create the first benchmark dataset for CLS, which can be used for evaluating the lexical simplification systems automatically. In order to acquire more thorough comparison, we present five different types of methods as baselines to generate substitute candidates for the complex word that include synonym-based approach, word embedding-based approach, pretrained language model-based approach, sememe-based approach, and a hybrid approach. Finally, we design the experimental evaluation of these baselines and discuss their advantages and disadvantages. To our best knowledge, this is the first study for CLS task. | ['Xindong Wu', 'Yang Shi', 'Yunhao Yuan', 'Yun Li', 'Xinyu Lu', 'Jipeng Qiang'] | 2020-10-14 | null | null | null | null | ['lexical-simplification'] | ['natural-language-processing'] | [ 1.16623089e-01 -9.65764523e-02 -7.28626475e-02 -3.82536799e-01
-4.93250042e-01 -2.77290732e-01 4.25687730e-01 3.95164818e-01
-9.11564052e-01 1.06734157e+00 5.62994778e-01 -2.59206116e-01
3.55173856e-01 -6.69933200e-01 -2.62434214e-01 -2.83899516e-01
7.12296486e-01 3.41468573e-01 3.25841933e-01 -5.52302897e-01
3.53521019e-01 2.84356087e-01 -1.17340362e+00 2.03655660e-01
1.41740406e+00 6.95787221e-02 5.73863566e-01 1.07334994e-01
-3.60852569e-01 3.03066701e-01 -9.14414108e-01 -8.75419617e-01
8.62697810e-02 -4.64597970e-01 -6.78098023e-01 -5.24277270e-01
2.72816598e-01 -6.82602748e-02 -2.10952256e-02 1.25229537e+00
8.26039553e-01 2.46095195e-01 8.95750523e-01 -6.48650587e-01
-1.13792932e+00 1.06755710e+00 -3.40339392e-01 2.14478180e-01
6.52405322e-01 -2.01592073e-01 8.50953758e-01 -9.12798345e-01
8.69376898e-01 1.50046790e+00 7.06663370e-01 7.40836382e-01
-9.11913574e-01 -8.28090131e-01 4.83333737e-01 1.57252803e-01
-1.58266020e+00 -2.58307457e-01 5.11932611e-01 1.27445464e-03
1.37994814e+00 3.51546973e-01 7.00877964e-01 1.03151131e+00
2.26259828e-01 8.00476074e-01 1.08911431e+00 -9.31752980e-01
-2.41686925e-01 3.51342291e-01 3.94779593e-01 2.51596749e-01
8.33056390e-01 -3.81189674e-01 -4.33534563e-01 3.20513368e-01
2.77317762e-01 -6.93719164e-02 -1.34961843e-01 3.53676647e-01
-1.15230167e+00 8.21181118e-01 6.55023754e-02 5.37243009e-01
-1.81171015e-01 -1.97921470e-01 5.52596807e-01 5.06623745e-01
4.23543423e-01 7.51888335e-01 -5.97772479e-01 5.80793731e-02
-9.38319802e-01 4.66289371e-01 7.86622763e-01 1.42909765e+00
3.99814934e-01 2.32263222e-01 -3.29195291e-01 9.51047063e-01
1.43732175e-01 6.15523756e-01 8.30354095e-01 -3.33397359e-01
6.58517718e-01 4.91634458e-01 -1.44755870e-01 -5.53505182e-01
-2.41236150e-01 -1.11837357e-01 -7.48355508e-01 -2.45508581e-01
-5.41700311e-02 -3.53871644e-01 -6.99539185e-01 1.77619648e+00
7.54958391e-02 -1.69707954e-01 1.27600849e-01 4.82271016e-01
1.18214571e+00 6.64859533e-01 4.16147172e-01 -4.36220318e-01
1.44142985e+00 -1.27370310e+00 -1.22923338e+00 1.23169787e-01
8.71807992e-01 -1.35113680e+00 1.55881000e+00 3.99710357e-01
-1.08732152e+00 -5.17564654e-01 -1.20781326e+00 -4.82580125e-01
-9.05548573e-01 1.40278712e-01 6.42260134e-01 8.68600309e-01
-7.51487970e-01 5.22687018e-01 -6.29723132e-01 -6.59933031e-01
3.45303118e-02 2.39490673e-01 -4.69716609e-01 -5.82355000e-02
-1.70446515e+00 1.52772701e+00 7.54219115e-01 -2.96418458e-01
-2.08762065e-01 -7.52093017e-01 -9.93031740e-01 -4.30166982e-02
1.94219410e-01 -7.14270890e-01 1.21525061e+00 -1.03518045e+00
-1.47896600e+00 8.92975807e-01 -4.02255088e-01 -3.17765206e-01
1.21500701e-01 -5.39449036e-01 -6.00248396e-01 -6.20407701e-01
2.89297134e-01 5.57101011e-01 5.31714261e-01 -7.30450034e-01
-6.37828946e-01 3.36149782e-02 1.03053682e-01 5.65856159e-01
-2.87943095e-01 5.24544060e-01 -5.24175406e-01 -1.34247434e+00
-2.47317120e-01 -8.49750459e-01 -9.83919650e-02 -5.33596575e-01
-5.25305271e-01 -5.50399244e-01 4.57095563e-01 -9.54122603e-01
2.07576942e+00 -2.04277658e+00 2.11040631e-01 -1.73638701e-01
-2.23263517e-01 8.81721675e-01 -1.59437865e-01 8.16945791e-01
-8.68230015e-02 7.72624433e-01 -2.47772500e-01 -5.22268414e-01
5.96875437e-02 2.70932078e-01 -1.28542513e-01 -8.04372355e-02
3.08854461e-01 9.39833164e-01 -8.85002673e-01 -6.48098052e-01
1.41880363e-01 2.95496583e-01 -5.65258980e-01 1.31120443e-01
-1.60418041e-02 1.73800662e-01 -2.79858857e-01 6.75661266e-01
7.56007969e-01 8.04376721e-01 7.02625327e-03 -1.84048265e-01
-3.83271962e-01 6.83654547e-01 -1.33776128e+00 1.68342006e+00
-6.57054961e-01 4.50336814e-01 -4.82992142e-01 -5.64113498e-01
5.45624077e-01 1.96390480e-01 -4.24347073e-01 -6.17543936e-01
1.62591487e-01 3.67765814e-01 3.15900207e-01 -4.40285534e-01
8.45261335e-01 -2.19366342e-01 -2.69679457e-01 3.52544069e-01
8.35937709e-02 -4.98844832e-01 7.57310629e-01 6.93192407e-02
7.50709593e-01 1.98676616e-01 9.30776417e-01 -4.65945035e-01
8.59957874e-01 -1.01995140e-01 6.80164397e-01 5.65006375e-01
6.74428940e-02 6.42872155e-01 2.83613354e-01 -3.94968212e-01
-8.62973034e-01 -1.10828388e+00 1.11268787e-02 1.09034181e+00
2.25789279e-01 -8.23141336e-01 -1.09959102e+00 -5.78335643e-01
-2.55165011e-01 1.23135042e+00 -2.67559290e-01 -1.15021884e-01
-9.94772315e-01 -7.38945723e-01 7.62774348e-01 1.90007418e-01
3.32034409e-01 -1.52413619e+00 -1.95510760e-01 1.57993257e-01
-2.52415657e-01 -9.04963911e-01 -7.98423588e-01 4.82558906e-02
-6.29030108e-01 -8.46507668e-01 -5.00860572e-01 -1.22496974e+00
5.92196226e-01 2.93568671e-01 1.15704381e+00 3.95100564e-01
-1.14430539e-01 -3.30735207e-01 -6.86090887e-01 -1.14461637e+00
-3.24901491e-01 5.77674627e-01 1.63631812e-01 -6.63154423e-01
7.10146308e-01 -3.28424603e-01 -1.92269534e-01 -3.20358247e-01
-1.02596343e+00 3.16829197e-02 7.40220964e-01 7.16833591e-01
7.06851363e-01 -4.97266874e-02 3.83312255e-01 -1.50265181e+00
1.14403915e+00 -3.27463597e-01 -3.18364471e-01 4.82759416e-01
-5.06509423e-01 2.75556415e-01 9.17155921e-01 -5.05814195e-01
-1.26373136e+00 -1.06042914e-01 -5.08157253e-01 3.39701593e-01
-1.40862182e-01 6.70047998e-01 -4.22477752e-01 1.70537814e-01
4.70384687e-01 3.14259827e-02 -5.98964810e-01 -8.37901533e-01
5.47064185e-01 7.21997499e-01 2.94520825e-01 -5.58067501e-01
8.10193658e-01 -1.29091486e-01 -4.19035345e-01 -1.11084294e+00
-9.63211715e-01 -3.15860242e-01 -9.41569626e-01 4.01387185e-01
9.11028266e-01 -8.78371000e-01 4.74235453e-02 3.63303423e-01
-1.71662438e+00 4.80204225e-02 -2.08747283e-01 5.25594532e-01
2.00084209e-01 5.26962459e-01 -4.86783892e-01 -3.50218117e-01
-6.72260523e-01 -1.07636929e+00 7.98300624e-01 3.03467125e-01
-7.11869717e-01 -9.63020265e-01 1.46618113e-01 -6.46373332e-02
5.80749571e-01 2.18336120e-01 1.20129919e+00 -8.90790284e-01
-1.67305082e-01 -1.03478990e-01 -2.39586886e-02 5.43285429e-01
2.19635054e-01 1.86497852e-01 -5.17001510e-01 2.25467663e-02
-1.24636829e-01 -2.00449124e-01 7.40258813e-01 1.78137422e-01
7.79660821e-01 -1.35159329e-01 -1.85707077e-01 6.46042287e-01
1.46964431e+00 3.06561083e-01 7.55316556e-01 4.09861505e-01
8.58918011e-01 2.77199924e-01 6.52970791e-01 -4.85411240e-03
5.25552511e-01 5.74939072e-01 -2.67250896e-01 -1.01610050e-01
-5.32521605e-01 -2.51098603e-01 2.95659870e-01 1.71506441e+00
-1.04575858e-01 -4.81087536e-01 -6.35083139e-01 4.35215086e-01
-1.46116948e+00 -6.73757613e-01 -3.20100665e-01 2.19932938e+00
1.33523631e+00 2.18204826e-01 -2.95383900e-01 1.40821218e-01
8.04299593e-01 9.69071612e-02 7.73001555e-03 -9.93452430e-01
-3.67973328e-01 8.41888726e-01 3.05064708e-01 7.84015477e-01
-1.07328463e+00 1.76964903e+00 6.02513027e+00 1.20556593e+00
-1.07660377e+00 3.64433408e-01 1.13149405e-01 1.95441514e-01
-5.19398272e-01 2.00571388e-01 -1.10969114e+00 3.79723668e-01
6.52534664e-01 -3.82598221e-01 5.00078201e-01 4.92194414e-01
1.90609574e-01 -1.68050364e-01 -1.16365230e+00 1.01221848e+00
2.53127038e-01 -9.18323398e-01 6.54683590e-01 -6.68154895e-01
9.84976649e-01 -1.25071958e-01 -2.98187435e-01 5.44526219e-01
5.08050621e-02 -1.08795953e+00 7.77855217e-01 5.79784811e-01
7.46691585e-01 -8.49107921e-01 9.44531739e-01 3.01473022e-01
-1.11403775e+00 5.86923599e-01 -5.54554880e-01 -5.02468407e-01
2.46259525e-01 3.23924482e-01 -9.21036378e-02 5.82462907e-01
3.69228959e-01 5.58231413e-01 -7.97101974e-01 8.98189366e-01
-8.23334157e-01 5.34088314e-01 -2.03555584e-01 -4.23191905e-01
1.09340467e-01 -3.10725093e-01 5.48916161e-01 1.78847563e+00
2.05733284e-01 1.19086213e-01 -5.97123243e-02 5.41340888e-01
-1.25441223e-01 8.70190740e-01 -7.00257659e-01 2.17826124e-02
8.67640674e-01 1.00582385e+00 -6.87327445e-01 -3.55309725e-01
-6.42982185e-01 1.24849415e+00 5.02529502e-01 2.21160352e-01
-7.17120707e-01 -1.12937701e+00 4.36333507e-01 -8.97977799e-02
3.02025788e-02 -7.87512138e-02 -5.75011075e-01 -1.39584827e+00
1.22741558e-01 -1.09613967e+00 1.20480023e-01 -4.14113671e-01
-1.20192862e+00 7.61894941e-01 3.45128030e-01 -9.39497948e-01
4.39296328e-02 -4.01771009e-01 -8.59640479e-01 1.18852854e+00
-1.57597589e+00 -1.17051089e+00 -3.44018117e-02 2.46695161e-01
9.94118631e-01 -3.05389583e-01 1.02167332e+00 4.40883756e-01
-7.25467384e-01 9.89961505e-01 -1.04795182e-02 -4.51462753e-02
9.72910821e-01 -1.31450248e+00 7.02346146e-01 1.14306140e+00
3.69340964e-02 1.06437111e+00 7.38679886e-01 -8.24225187e-01
-8.92503023e-01 -1.16294539e+00 1.79506981e+00 -4.24933344e-01
3.90702665e-01 -4.35788274e-01 -9.73416805e-01 4.97115076e-01
6.97021008e-01 -6.30916774e-01 6.66957438e-01 9.22384486e-03
-1.05857477e-01 -8.94534141e-02 -9.24746871e-01 1.17138290e+00
1.10589564e+00 -1.30643502e-01 -1.22013450e+00 2.72317439e-01
1.08471572e+00 -4.24729347e-01 -6.05380416e-01 2.88021415e-01
3.39293540e-01 -6.57190323e-01 6.64499640e-01 -4.89734322e-01
2.93694109e-01 -3.79617482e-01 -3.51920887e-03 -1.53009856e+00
-3.47464502e-01 -7.85387039e-01 5.18918157e-01 1.57859588e+00
6.10627949e-01 -5.39111555e-01 1.19260706e-01 2.93119818e-01
-3.95220637e-01 -7.37599492e-01 -6.91064298e-01 -6.98412597e-01
5.42216003e-01 -2.44091719e-01 8.48233581e-01 9.15471792e-01
-1.91129208e-01 8.45692694e-01 -2.25198850e-01 -2.10309252e-01
5.46137467e-02 -3.56468350e-01 6.31279469e-01 -1.03771079e+00
1.92193136e-01 -6.17519796e-01 -1.57588512e-01 -7.36268401e-01
4.96666998e-01 -9.98040020e-01 -2.22929686e-01 -1.56063211e+00
1.35987297e-01 -2.89428592e-01 -7.55704343e-02 1.93903476e-01
-6.40772343e-01 2.28057787e-01 3.94145072e-01 -1.28461853e-01
-5.74269533e-01 6.57786965e-01 1.08984470e+00 1.35931075e-01
-2.78036207e-01 -2.86105931e-01 -9.78813708e-01 8.90127480e-01
7.99960494e-01 -7.62522101e-01 -4.78072077e-01 -8.74039829e-01
3.04293782e-01 -5.44805348e-01 -5.68669856e-01 -7.13456154e-01
1.13017233e-02 -1.96220681e-01 4.37634103e-02 -6.54085994e-01
-1.71348214e-01 -4.58884627e-01 3.84047106e-02 5.64097345e-01
-1.91221774e-01 6.75247490e-01 2.24269465e-01 -6.75711557e-02
-3.67587328e-01 -7.45315611e-01 8.58310163e-01 -3.89325351e-01
-6.39831483e-01 1.42183406e-02 -5.78050315e-01 5.62166870e-01
9.02763426e-01 -2.11878523e-01 2.84354333e-02 -1.24154292e-01
-2.02514485e-01 1.34664774e-01 3.45632046e-01 5.77582836e-01
4.31231856e-01 -1.36433744e+00 -7.83330798e-01 1.37104601e-01
2.33535785e-02 3.95427458e-03 -1.58341780e-01 4.32734132e-01
-1.16900504e+00 4.57247764e-01 -1.88689590e-01 2.78644323e-01
-1.25361824e+00 6.22017562e-01 -1.13235816e-01 -3.88130575e-01
-2.98312783e-01 7.83330500e-01 1.54554889e-01 -6.89224064e-01
3.54818374e-01 -7.20522583e-01 -5.76446772e-01 -9.40967351e-02
6.14052236e-01 4.46443051e-01 1.66318730e-01 -7.83912718e-01
-3.38498652e-01 5.83757818e-01 -3.24152887e-01 1.71261542e-02
9.36926425e-01 -2.73050040e-01 -6.27615452e-01 4.59836364e-01
9.41346169e-01 5.29844582e-01 -1.28927469e-01 -1.29253909e-01
3.00614417e-01 -1.83480620e-01 -3.80734146e-01 -6.90556586e-01
-6.36203825e-01 7.81315506e-01 1.10938743e-01 -1.78347483e-01
1.12552595e+00 -5.25276899e-01 1.00776577e+00 6.73749447e-01
1.13642342e-01 -1.40561426e+00 -4.43009228e-01 1.05768847e+00
9.88243163e-01 -1.17825782e+00 1.69819847e-01 -7.50977993e-01
-8.15747559e-01 1.07576239e+00 7.44092107e-01 -4.49760765e-01
6.90728605e-01 1.35336578e-01 1.31598011e-01 2.61585474e-01
-6.46331847e-01 -2.43314415e-01 3.45787317e-01 8.40383291e-01
1.08723056e+00 1.74585760e-01 -1.39611113e+00 1.02126014e+00
-6.72928572e-01 -3.05261344e-01 4.76536006e-01 6.64839566e-01
-2.73096383e-01 -1.69560707e+00 -1.16447568e-01 4.21041399e-01
-6.97709918e-01 -8.95891368e-01 -5.24287403e-01 1.17482543e+00
5.76109946e-01 7.30613768e-01 -3.86694044e-01 -9.03059095e-02
7.74065375e-01 1.49914950e-01 4.91908491e-01 -1.14738369e+00
-7.78974354e-01 -1.21638663e-01 2.25950703e-01 -9.33628306e-02
-2.54111350e-01 -6.65504515e-01 -1.12184250e+00 -4.43212956e-01
-4.27582443e-01 9.93455574e-02 3.61217976e-01 1.08987045e+00
-1.63111627e-01 3.59714270e-01 2.32977971e-01 -7.80462027e-01
-3.51268649e-01 -1.21248329e+00 -3.93081278e-01 5.76619506e-01
-2.15669617e-01 -5.06274581e-01 1.41799729e-02 2.99437381e-02] | [10.86661434173584, 10.410893440246582] |
f152d8b6-21ef-48f3-b4e5-d50da37b3401 | metam-goal-oriented-data-discovery | 2304.09068 | null | https://arxiv.org/abs/2304.09068v1 | https://arxiv.org/pdf/2304.09068v1.pdf | METAM: Goal-Oriented Data Discovery | Data is a central component of machine learning and causal inference tasks. The availability of large amounts of data from sources such as open data repositories, data lakes and data marketplaces creates an opportunity to augment data and boost those tasks' performance. However, augmentation techniques rely on a user manually discovering and shortlisting useful candidate augmentations. Existing solutions do not leverage the synergy between discovery and augmentation, thus under exploiting data. In this paper, we introduce METAM, a novel goal-oriented framework that queries the downstream task with a candidate dataset, forming a feedback loop that automatically steers the discovery and augmentation process. To select candidates efficiently, METAM leverages properties of the: i) data, ii) utility function, and iii) solution set size. We show METAM's theoretical guarantees and demonstrate those empirically on a broad set of tasks. All in all, we demonstrate the promise of goal-oriented data discovery to modern data science applications. | ['Raul Castro Fernandez', 'Yue Gong', 'Sainyam Galhotra'] | 2023-04-18 | null | null | null | null | ['causal-inference', 'causal-inference'] | ['knowledge-base', 'miscellaneous'] | [ 2.78558403e-01 1.84575066e-01 -8.34764242e-01 -4.17698145e-01
-6.99274898e-01 -7.39996433e-01 7.53130376e-01 6.90346181e-01
-3.76322448e-01 8.41524005e-01 6.90017939e-01 -5.45169652e-01
-4.41149205e-01 -9.69618320e-01 -7.11092651e-01 -2.52813429e-01
-4.10014182e-01 7.52420783e-01 1.15508530e-02 9.88336653e-03
4.08737183e-01 3.40047121e-01 -1.64186561e+00 1.38421610e-01
8.94012630e-01 8.61032426e-01 1.02029741e-02 2.39690915e-01
-2.85645992e-01 7.97647417e-01 4.43302691e-02 -1.48151264e-01
4.27399635e-01 -2.83484787e-01 -9.61517751e-01 -9.55693424e-02
-4.85395826e-02 -8.34391713e-02 -9.59011242e-02 8.32950652e-01
2.25967407e-01 2.08603635e-01 2.38495231e-01 -1.59638608e+00
-5.01923800e-01 1.00230789e+00 -6.70566499e-01 4.23169643e-01
1.46408305e-01 3.67631853e-01 1.40848196e+00 -9.79859173e-01
8.22354138e-01 1.13981807e+00 5.53159952e-01 3.38609248e-01
-1.45372057e+00 -7.45686233e-01 4.37632561e-01 -2.47000143e-01
-1.03754807e+00 -6.01886094e-01 5.40574431e-01 -7.50364423e-01
6.23768747e-01 4.56124425e-01 5.10812938e-01 6.63917065e-01
-5.10255992e-01 7.65354991e-01 7.81231999e-01 -3.23531836e-01
5.43411553e-01 6.45754486e-02 4.75511491e-01 6.05234861e-01
5.18626750e-01 3.07215661e-01 -9.83871758e-01 -5.71681499e-01
4.72856969e-01 1.28378393e-02 -4.56929505e-02 -3.71879011e-01
-1.15552390e+00 8.26259911e-01 3.66961300e-01 -4.01296355e-02
-6.72754645e-01 1.62380472e-01 2.37275854e-01 2.64930844e-01
4.98021573e-01 1.04761994e+00 -6.32812738e-01 -2.29727492e-01
-6.99728847e-01 4.31258470e-01 6.45466387e-01 9.76114869e-01
8.28217685e-01 -4.56286699e-01 -2.68995374e-01 6.23832822e-01
3.18367451e-01 -2.20709667e-02 2.26602212e-01 -8.37799370e-01
4.23007876e-01 1.10515630e+00 2.28041038e-01 -7.19458818e-01
-6.41255617e-01 -4.62786227e-01 -5.70446551e-01 -1.18620381e-01
4.49412614e-01 -6.64936453e-02 -8.85711372e-01 2.05347037e+00
8.00707042e-01 -6.66325837e-02 -5.89878969e-02 9.96372104e-01
4.18528318e-01 2.36212984e-01 4.17116672e-01 -3.65939289e-01
1.20088804e+00 -4.10150468e-01 -7.32611537e-01 -3.53112400e-01
9.85639095e-01 -3.57980907e-01 1.29288268e+00 4.21679050e-01
-9.67563570e-01 6.72647962e-03 -8.66687655e-01 -9.28532630e-02
-3.34595591e-01 -3.79565120e-01 1.19370532e+00 3.89051348e-01
-7.07052469e-01 5.57216525e-01 -7.34060824e-01 -2.66353756e-01
7.57244229e-01 2.79032946e-01 -1.92514062e-01 -1.66151643e-01
-9.78489459e-01 5.18902898e-01 4.21797514e-01 -1.02855884e-01
-8.39024186e-01 -1.23059690e+00 -4.79711473e-01 1.39491349e-01
8.01606178e-01 -8.65078092e-01 1.13419628e+00 -4.94001776e-01
-6.50528371e-01 5.40652156e-01 -8.59867260e-02 -3.81022096e-01
3.77623975e-01 -3.59238356e-01 -1.05258472e-01 -4.91596013e-01
2.65072256e-01 4.12299156e-01 2.00556383e-01 -1.14781988e+00
-9.05801892e-01 -7.18592584e-01 -5.76795191e-02 2.87915338e-02
-5.24402797e-01 4.40339372e-02 -1.99743897e-01 -4.17232245e-01
1.00854866e-01 -5.26858866e-01 -9.06485438e-01 -1.43178642e-01
-5.55658758e-01 -3.25768799e-01 6.52329326e-01 -1.75363809e-01
1.47430062e+00 -1.90463448e+00 -3.84848043e-02 5.27264178e-01
5.82749367e-01 -1.92779586e-01 -1.94747731e-01 4.87212420e-01
-3.26958075e-02 6.68721318e-01 -2.36817166e-01 -1.13542117e-01
-7.10233077e-02 1.50541782e-01 -2.78754234e-01 3.14298958e-01
4.35160130e-01 8.66526008e-01 -1.12646735e+00 -3.30949605e-01
-6.43983036e-02 -1.93017453e-01 -9.43354547e-01 3.46320897e-01
-7.93172956e-01 4.80095029e-01 -7.14417934e-01 6.04296505e-01
2.46921271e-01 -6.91410899e-01 3.21907818e-01 1.87317222e-01
-4.55919325e-01 6.12774789e-01 -1.19415760e+00 1.64078546e+00
-2.40741834e-01 3.30703348e-01 -6.88441098e-03 -8.78814995e-01
8.71504605e-01 2.38419678e-02 7.82651544e-01 -5.09925425e-01
-1.29196361e-01 1.65266678e-01 1.01087026e-01 -5.00594795e-01
3.62199336e-01 -1.56659242e-02 -3.77298892e-02 7.78754532e-01
-2.16882944e-01 3.05902869e-01 5.05721867e-01 4.47581559e-01
1.51657021e+00 -4.04691286e-02 4.51536238e-01 -3.84266376e-01
-9.84455273e-02 5.25651932e-01 6.31377339e-01 8.90241146e-01
3.45452130e-01 1.66533098e-01 7.67324567e-01 -5.68020046e-01
-1.09277928e+00 -7.69672036e-01 -1.46004617e-01 1.31544733e+00
-1.68082193e-01 -6.44472718e-01 -1.08677134e-01 -5.88517547e-01
3.93269986e-01 7.76002049e-01 -7.43596911e-01 -1.25164896e-01
-3.32557797e-01 -1.16957593e+00 2.91408896e-01 5.77288687e-01
-2.94263642e-02 -7.38061368e-01 -5.93758821e-01 2.79545397e-01
-1.34698972e-01 -7.93636024e-01 -2.91831434e-01 3.86238992e-01
-9.81205642e-01 -1.30163920e+00 3.41850191e-01 -1.32913753e-01
6.64230883e-01 1.02111034e-01 1.28969395e+00 1.17137939e-01
-4.50798631e-01 1.01168744e-01 -3.52784902e-01 -6.72205627e-01
-4.41258729e-01 1.67361036e-01 7.55668208e-02 -1.51788205e-01
6.04236722e-01 -7.50077069e-01 -4.90153015e-01 1.24335319e-01
-7.59811103e-01 8.21022317e-02 6.27883255e-01 7.94367552e-01
6.52339816e-01 -1.98439226e-01 1.02567685e+00 -9.94304299e-01
7.34135032e-01 -1.15521169e+00 -1.11019385e+00 7.56555200e-02
-1.25187814e+00 4.31044340e-01 2.22626850e-01 -3.07755470e-01
-7.88479865e-01 3.33589405e-01 2.62812436e-01 -2.85851300e-01
-1.36701325e-02 1.18091822e+00 -2.08396107e-01 5.39786518e-01
1.07283866e+00 -3.33043218e-01 9.47779194e-02 -7.45929420e-01
6.88850939e-01 4.20824319e-01 5.02568066e-01 -6.80724323e-01
6.46223426e-01 3.49383980e-01 6.31885901e-02 -5.06298006e-01
-1.02288878e+00 -6.19721651e-01 -4.59074438e-01 -6.71700314e-02
4.14567173e-01 -7.81021833e-01 -9.90795076e-01 -3.49385887e-01
-8.69009793e-01 -3.49738985e-01 -6.60422146e-01 3.74229044e-01
-1.98905766e-01 -3.36457253e-01 -5.33665493e-02 -1.04804909e+00
-3.31252158e-01 -7.49358058e-01 7.09636807e-01 -1.14149479e-02
-4.15395081e-01 -6.97546005e-01 1.90305591e-01 1.97427988e-01
3.85308474e-01 3.30308467e-01 1.11501026e+00 -1.09661579e+00
-9.32434618e-01 -2.14578733e-01 -3.21411669e-01 -3.19109917e-01
2.11720020e-01 -1.39100522e-01 -8.45689476e-01 1.33602992e-01
-4.52191323e-01 -4.26752359e-01 7.95147359e-01 3.22885990e-01
1.52469003e+00 -5.79038084e-01 -5.08332729e-01 5.46174228e-01
1.15997684e+00 2.24318057e-02 1.83374569e-01 3.64983797e-01
5.79092324e-01 8.27860475e-01 7.33098865e-01 8.93405735e-01
4.72170204e-01 3.84058297e-01 5.86420894e-01 -2.37653747e-01
2.74461746e-01 -4.43339169e-01 -3.33878756e-01 5.43861352e-02
-4.56634052e-02 -2.18567535e-01 -1.32707894e+00 6.14002466e-01
-2.16614509e+00 -9.39013660e-01 -2.09115639e-01 2.45190644e+00
1.30939913e+00 5.65426387e-02 4.25917208e-01 4.22946885e-02
3.93988192e-01 -1.78844735e-01 -8.24338078e-01 4.27139327e-02
3.42428461e-02 5.07673360e-02 6.55173540e-01 3.85175765e-01
-7.96276212e-01 8.36955845e-01 6.65020704e+00 2.89123178e-01
-6.46095216e-01 -9.27954353e-03 7.42031157e-01 -4.28182662e-01
-7.78675914e-01 2.17208073e-01 -7.26423323e-01 1.59404889e-01
9.55873132e-01 -6.40181541e-01 5.23365557e-01 8.53885591e-01
4.72432554e-01 -4.82922308e-02 -1.58678281e+00 4.86891061e-01
-5.45441568e-01 -1.85649860e+00 -8.76540318e-02 4.11657304e-01
5.36699414e-01 2.38999903e-01 -1.56406060e-01 3.01760174e-02
1.19802594e+00 -1.25365686e+00 4.02595997e-01 3.52924883e-01
6.57931209e-01 -6.36518419e-01 2.98919022e-01 3.11133772e-01
-7.31030583e-01 -4.77918923e-01 5.07734381e-02 -2.72905618e-01
1.10844728e-02 1.08707666e+00 -1.19640744e+00 3.62633646e-01
5.61278999e-01 5.96106946e-01 -4.21694219e-01 1.04420924e+00
-1.79522336e-02 9.06192362e-01 -4.62425709e-01 1.30525798e-01
3.95110287e-02 -1.03008047e-01 5.37146509e-01 1.02802885e+00
3.20085213e-02 2.74647862e-01 5.16627073e-01 1.24962854e+00
-3.66399109e-01 1.61955848e-01 -8.82846713e-01 -4.89795774e-01
8.82032692e-01 1.04075730e+00 -3.87919366e-01 -1.59180656e-01
-2.82630056e-01 -2.62744594e-02 4.04436976e-01 2.57357478e-01
-4.11691934e-01 1.68468207e-01 9.88789260e-01 1.73298582e-01
-3.41220081e-01 -1.26320675e-01 -8.12492132e-01 -7.97246337e-01
-5.49322069e-02 -9.29828525e-01 8.95252109e-01 -3.12260330e-01
-1.33321631e+00 2.97363885e-02 1.05485752e-01 -7.27047861e-01
-2.22529158e-01 -2.21555665e-01 -3.26262653e-01 8.21695089e-01
-1.27280450e+00 -8.89156282e-01 -3.15119058e-01 2.87557572e-01
2.87772775e-01 5.41800149e-02 5.76432467e-01 2.06923857e-01
-7.53123701e-01 3.47715557e-01 -1.14702970e-01 -1.42553195e-01
4.83536661e-01 -1.23817825e+00 5.60609937e-01 9.89444375e-01
8.06979463e-02 8.58379185e-01 7.42548645e-01 -8.92782450e-01
-1.63684213e+00 -1.15708470e+00 7.23242164e-01 -5.99152803e-01
8.71206045e-01 -4.14046288e-01 -9.66993928e-01 7.47520626e-01
-2.57878393e-01 3.11192516e-02 8.53085637e-01 1.01254308e+00
-5.70607007e-01 -1.28553346e-01 -9.85356867e-01 5.84816277e-01
1.43501008e+00 -2.14656387e-02 -2.35223427e-01 4.58084434e-01
9.25396264e-01 -5.98935187e-02 -9.95813131e-01 5.06310403e-01
4.58125859e-01 -4.82861578e-01 8.96772206e-01 -1.46913660e+00
6.34107590e-01 -1.05838455e-01 -1.75677150e-01 -1.19838858e+00
-2.28049695e-01 -8.97790551e-01 -2.10324451e-01 1.16297209e+00
8.54055464e-01 -5.40749252e-01 7.52756834e-01 1.31382704e+00
-1.56561404e-01 -6.87420964e-01 -6.22732162e-01 -5.37305653e-01
-1.84179366e-01 -6.39105856e-01 9.14513707e-01 1.28758061e+00
3.14316481e-01 5.27944803e-01 -2.67113507e-01 2.78731048e-01
7.41604447e-01 2.98486263e-01 1.04709804e+00 -1.77668130e+00
-1.57982200e-01 -5.45150816e-01 -2.36567762e-03 -5.79696476e-01
-3.81615162e-01 -1.07867110e+00 -7.90930763e-02 -1.45163608e+00
3.19285959e-01 -1.27338266e+00 -3.83380920e-01 9.68278110e-01
-4.93724942e-01 -3.30196172e-01 -2.06896290e-01 3.06365520e-01
-4.87255633e-01 3.74056607e-01 7.96199441e-01 2.43567452e-01
-7.02737451e-01 -2.34156504e-01 -1.21259809e+00 5.75127184e-01
8.38152111e-01 -6.11630023e-01 -5.68649948e-01 -3.61159563e-01
6.62503600e-01 -1.25599012e-01 4.11344022e-01 -3.78018081e-01
2.24362016e-01 -7.52620518e-01 1.48742691e-01 -5.50498426e-01
-3.06779854e-02 -5.73155105e-01 2.37336978e-01 2.92408347e-01
-9.78457153e-01 -4.78962101e-02 1.93100438e-01 7.52507269e-01
1.58495948e-01 1.44249126e-01 5.60833514e-01 7.97053427e-03
-5.19008517e-01 5.44960856e-01 1.39332367e-02 3.96038085e-01
8.94890964e-01 2.68411249e-01 -6.11022532e-01 -6.20081946e-02
-5.19837797e-01 8.17359626e-01 2.71252155e-01 5.34778535e-01
3.62264484e-01 -1.02020514e+00 -8.95121217e-01 1.97033212e-01
3.20585161e-01 3.95742029e-01 -3.58243108e-01 9.43439305e-01
2.67881483e-01 1.51321828e-01 8.23970214e-02 -4.56227630e-01
-9.67387617e-01 8.44422698e-01 -9.37291682e-02 -2.70006031e-01
-5.87245286e-01 7.66493380e-01 1.18913971e-01 -3.55547160e-01
1.95662931e-01 -2.22906191e-02 8.54587182e-02 2.73960441e-01
6.77062988e-01 2.71600038e-01 -7.16899708e-02 3.39196652e-01
-4.03936654e-01 -4.98396099e-01 -4.20581624e-02 -2.34957248e-01
1.70090365e+00 3.90042216e-02 -3.11815351e-01 3.60482544e-01
6.77712381e-01 -7.63079971e-02 -9.40074563e-01 -5.45775712e-01
8.10077727e-01 -7.71167099e-01 1.34458378e-01 -8.90187204e-01
-7.84195721e-01 3.83258015e-01 5.23733497e-02 3.66923690e-01
1.10585713e+00 2.93283850e-01 1.58875629e-01 2.95736581e-01
1.22034654e-01 -9.69476461e-01 7.98404664e-02 4.39887308e-02
9.10942316e-01 -1.24170101e+00 -1.06569594e-02 -3.04014534e-01
-4.53131497e-01 6.89496756e-01 6.57956541e-01 3.58157575e-01
5.06109059e-01 6.03520095e-01 -3.18746030e-01 -5.52772105e-01
-1.45354843e+00 -5.11485875e-01 1.12150133e-01 3.75403345e-01
4.83164907e-01 -6.48703799e-02 -3.74429554e-01 7.10841238e-01
-1.14201516e-01 1.75059170e-01 4.72776085e-01 9.17939186e-01
-5.94860673e-01 -1.10392499e+00 -3.32288891e-01 1.05856693e+00
-2.40788445e-01 -3.78820509e-01 -5.06590068e-01 6.95739329e-01
-9.05505940e-02 1.09416246e+00 7.43343383e-02 -2.67452925e-01
2.93256640e-01 1.45470813e-01 -1.24549136e-01 -7.20848143e-01
-4.56116766e-01 -1.13709331e-01 4.66381073e-01 -7.28328109e-01
-1.44776091e-01 -7.80060589e-01 -1.16778886e+00 -3.06313246e-01
-3.33153695e-01 3.68343800e-01 7.61957169e-01 9.19247329e-01
8.39880943e-01 4.35380101e-01 6.25362933e-01 -1.76801383e-01
-6.49988770e-01 -7.72322178e-01 -1.44299075e-01 3.21507156e-01
2.16031492e-01 -6.36276901e-01 -1.02087520e-01 8.73820484e-02] | [7.876273155212402, 5.417774200439453] |
2876cd73-988f-423f-ad70-1b8ba034a3bf | atril-an-xml-visualization-system-for-corpus | null | null | https://aclanthology.org/2022.lrec-1.611 | https://aclanthology.org/2022.lrec-1.611.pdf | Atril: an XML Visualization System for Corpus Texts | This paper presents Atril, an XML visualization system for corpus texts, developed for, but not restricted to, the project Corpus de Audiências (CorAuDis), a corpus composed of transcripts of sessions of criminal proceedings recorded at the Coimbra Court. The main aim of the tool is to provide researchers with a web-based environment that allows for an easily customizable visualization of corpus texts with heavy structural annotation. Existing corpus analysis tools such as SketchEngine, TEITOK and CQPweb offer some kind of visualization mechanisms, but, to our knowledge, none meets our project’s main needs. Our requirements are a system that is open-source; that can be easily connected to CQPweb and TEITOK, that provides a full text-view with switchable visualization templates, that allows for the visualization of overlapping utterances. To meet those requirements, we created Atril, a module with a corpus XML file viewer, a visualization management system, and a word alignment tool. | ['Cornelia Plag', 'Conceição Carapinha', 'Andressa Rodrigues Gomide'] | null | null | null | null | lrec-2022-6 | ['word-alignment'] | ['natural-language-processing'] | [-1.26281828e-01 2.44402304e-01 4.93309170e-01 -1.10961221e-01
-2.71219075e-01 -7.08264768e-01 6.88077390e-01 5.68776309e-01
-1.50677517e-01 3.58177185e-01 3.95139396e-01 -8.71255875e-01
-3.84443343e-01 -3.98739010e-01 2.88543582e-01 -1.15449794e-01
-3.12214275e-03 6.33973360e-01 4.46005255e-01 -4.02974695e-01
6.26233101e-01 5.19257843e-01 -1.62769222e+00 4.55818027e-01
5.39855599e-01 3.99304450e-01 7.55719483e-01 7.58499444e-01
-7.06985831e-01 5.40626943e-01 -9.84289646e-01 -1.65440395e-01
-1.66455775e-01 -3.53720665e-01 -1.05544007e+00 -4.17278022e-01
-1.89075351e-01 2.86434907e-02 2.35527754e-01 6.25644743e-01
4.75353658e-01 -2.19405398e-01 4.09747988e-01 -1.18056679e+00
-2.24327132e-01 8.12245607e-01 -2.03728214e-01 4.39462125e-01
9.89834726e-01 -1.18562944e-01 4.95862931e-01 -6.09156251e-01
1.20922506e+00 1.32975960e+00 4.40467000e-01 2.68738955e-01
-9.31187570e-01 -5.88196158e-01 -3.69075209e-01 1.15537643e-01
-8.81374180e-01 -2.89546281e-01 8.01358521e-01 -6.39859319e-01
1.27546346e+00 8.63238633e-01 5.90533435e-01 1.12363482e+00
5.20160887e-04 3.26348931e-01 1.02901077e+00 -1.05966842e+00
1.91313028e-01 6.43264592e-01 4.98120606e-01 3.01767260e-01
3.95629220e-02 -5.87643266e-01 -6.12241447e-01 -4.17142361e-01
5.38094163e-01 -4.26799178e-01 -3.00116181e-01 -7.40025043e-02
-1.05349016e+00 5.42057216e-01 -5.81708729e-01 1.25776517e+00
-1.09494179e-01 -4.35497552e-01 9.90369916e-01 5.38082063e-01
4.01805907e-01 3.37197095e-01 -2.19130337e-01 -8.26752782e-01
-7.29067147e-01 2.73988008e-01 1.04915798e+00 1.03826499e+00
1.99169785e-01 -2.26886272e-01 1.55236512e-01 7.82180130e-01
6.49506152e-01 2.22776994e-01 6.77621245e-01 -5.07346570e-01
6.61642790e-01 8.89727235e-01 -2.84169838e-02 -9.14678574e-01
-6.32102966e-01 1.64655820e-01 -3.06817573e-02 3.94265890e-01
6.35057867e-01 -7.56316036e-02 -1.98551729e-01 1.02737188e+00
3.96751583e-01 -7.02709675e-01 7.56567866e-02 5.79604506e-01
1.10569942e+00 4.89123076e-01 8.44993144e-02 -4.06987101e-01
1.36752355e+00 -1.53523415e-01 -1.34075046e+00 8.64100009e-02
9.64241207e-01 -1.33661211e+00 1.50020075e+00 5.70731282e-01
-1.03905892e+00 -9.73144770e-02 -1.10048020e+00 -1.29956365e-01
-9.84277487e-01 1.97201688e-02 3.79919231e-01 9.38385069e-01
-8.75387669e-01 5.71604192e-01 -7.97212064e-01 -8.48999918e-01
1.00194814e-03 -1.69817492e-01 -3.21472645e-01 5.14723837e-01
-7.97553658e-01 1.23169398e+00 5.27378500e-01 -3.05472642e-01
1.40558466e-01 -4.30381119e-01 -6.47936821e-01 8.65217000e-02
4.07253832e-01 1.63259491e-01 1.23885357e+00 -6.84548080e-01
-1.31384313e+00 8.39108109e-01 1.72548488e-01 1.12468190e-01
7.37393677e-01 -7.73197487e-02 -5.68043411e-01 4.73776609e-02
1.53410390e-01 -2.14236990e-01 1.10620931e-01 -1.05799973e+00
-4.82711345e-01 -6.38915122e-01 -2.17539728e-01 -9.68664289e-02
-5.43330133e-01 9.36694860e-01 -4.19027984e-01 -5.95029294e-01
-5.43415427e-01 -2.44855106e-01 9.77448225e-02 -3.64182264e-01
-2.88923711e-01 -3.65363330e-01 1.46274376e+00 -1.14296091e+00
1.85780287e+00 -2.08966279e+00 -2.72135083e-02 4.90972757e-01
-8.65485519e-02 6.06072664e-01 1.40871838e-01 1.35618317e+00
-1.79675221e-01 3.48073334e-01 -9.65414345e-02 -4.60317254e-01
2.37135589e-01 2.90435642e-01 8.51473510e-02 8.96429177e-03
-2.42929578e-01 5.31219393e-02 -9.42275107e-01 -8.24286282e-01
6.23248994e-01 3.81822109e-01 9.39674973e-02 -4.71725985e-02
-1.31282508e-01 1.80920303e-01 -4.37600851e-01 1.22148417e-01
5.96533477e-01 1.55129388e-01 8.78818631e-01 4.74396348e-01
-9.79139864e-01 4.61365402e-01 -1.70776212e+00 1.94206679e+00
-4.31971490e-01 9.90798056e-01 3.92404348e-01 -6.82587564e-01
1.32282639e+00 7.91540146e-01 3.32814097e-01 -5.92638433e-01
1.97232053e-01 4.47262198e-01 -3.14520359e-01 -9.42206502e-01
7.95379937e-01 6.22888148e-01 2.41536543e-01 9.97007310e-01
-4.35366221e-02 7.09256828e-02 7.42329001e-01 4.36638832e-01
8.70198905e-01 4.29369986e-01 3.76493961e-01 -2.84997046e-01
4.67293471e-01 9.63602308e-03 1.58455908e-01 1.77816376e-01
5.10029614e-01 1.50120258e-01 7.70457029e-01 -2.92089790e-01
-1.41384172e+00 -5.68875849e-01 -4.06263143e-01 8.63696635e-01
-4.58766550e-01 -1.03136539e+00 -1.10411036e+00 -2.19165459e-01
-6.76803231e-01 7.77581573e-01 -3.36669892e-01 8.01020086e-01
-4.47217733e-01 -9.51922592e-03 5.15583396e-01 7.59037212e-02
-8.38692824e-04 -1.67260957e+00 -1.41627514e+00 2.57822216e-01
-7.77080432e-02 -5.67285061e-01 -2.00786963e-01 -1.66691557e-01
-6.65438533e-01 -1.23579538e+00 -2.78539598e-01 -6.35843873e-01
2.09334671e-01 -1.42296955e-01 9.80503201e-01 4.50742483e-01
-7.40845501e-01 4.61635500e-01 -8.99257004e-01 -7.70951807e-01
-8.69224310e-01 -3.53209525e-02 -5.26638687e-01 -6.08664870e-01
3.47207248e-01 -8.36802423e-01 8.37289989e-02 4.17004451e-02
-1.28850508e+00 1.89952970e-01 -1.43926710e-01 3.22637349e-01
-2.95795985e-02 -4.01307702e-01 7.68097699e-01 -9.84317720e-01
1.34851754e+00 -3.61310869e-01 -6.78541601e-01 4.93527919e-01
-5.73487163e-01 -3.86523396e-01 4.44782555e-01 -7.14889988e-02
-1.22191954e+00 -2.40986347e-01 -2.27922946e-01 -1.03432061e-02
-4.92343247e-01 6.37231410e-01 -2.92077869e-01 5.15408337e-01
5.30806124e-01 -1.09166503e-01 1.64687008e-01 -9.38294768e-01
4.23018992e-01 1.37179625e+00 3.69747579e-01 -3.39090317e-01
1.21829778e-01 -3.97300377e-04 -5.59487939e-01 -1.04521501e+00
4.62178826e-01 -7.54925787e-01 -8.69103551e-01 -6.42066181e-01
5.74215412e-01 -7.91110620e-02 -8.24860513e-01 -1.04987308e-01
-1.38578665e+00 -4.05017197e-01 -3.00459832e-01 2.97742397e-01
-3.35367024e-01 7.77411461e-01 -1.32837087e-01 -1.30282187e+00
-6.19966328e-01 -7.70432413e-01 5.77287555e-01 2.34707490e-01
-8.96693468e-01 -1.02888334e+00 4.75146919e-01 5.21891676e-02
3.34256113e-01 5.34551978e-01 7.33407259e-01 -7.69001901e-01
2.48236097e-02 3.08116595e-03 -5.64534590e-02 -8.55391845e-02
-6.09335341e-02 6.98029935e-01 -8.93087864e-01 -9.02712718e-03
-3.62778157e-01 1.64535772e-02 -3.74893844e-02 -1.27257183e-01
6.56632960e-01 -3.98525894e-01 -6.33749843e-01 -1.20757245e-01
1.33159053e+00 9.31617379e-01 7.72793472e-01 7.60592759e-01
-4.51464206e-03 1.01365185e+00 6.59290135e-01 8.52467000e-01
1.64233148e-01 1.04204774e+00 2.57878333e-01 -8.57485980e-02
8.26734006e-02 1.28887981e-01 -7.46919364e-02 9.52727437e-01
-3.03099006e-01 -2.69830555e-01 -1.02536356e+00 5.77018797e-01
-1.97864902e+00 -1.16012287e+00 -8.10742438e-01 2.06145120e+00
4.13264066e-01 -1.18998006e-01 4.52437967e-01 7.50137985e-01
4.39940602e-01 -1.55112604e-02 4.67316836e-01 -1.30862844e+00
4.04400349e-01 5.09901047e-01 -2.23787397e-01 7.12686062e-01
-5.19143105e-01 3.39480639e-01 5.53898287e+00 6.69963419e-01
-1.00512373e+00 4.26974207e-01 -2.15897098e-01 -1.25247344e-01
-4.45167631e-01 -3.71725624e-03 -1.33151025e-01 5.84949970e-01
1.07429051e+00 -1.92424923e-01 3.12334150e-01 8.33789051e-01
5.93244135e-01 -3.24151814e-02 -7.90458977e-01 4.74157721e-01
-1.24196194e-01 -1.73051381e+00 -3.51527452e-01 2.32121617e-01
-1.81880876e-01 -1.08096242e-01 -5.14868915e-01 -4.86410633e-02
5.54930903e-02 -7.09704041e-01 9.35959756e-01 4.92135972e-01
8.78800213e-01 -9.17065084e-01 5.89357138e-01 2.55085975e-01
-1.02401674e+00 1.25876576e-01 -1.30880922e-01 -4.14990261e-02
4.91687387e-01 1.97414875e-01 -9.81247544e-01 9.96387005e-01
9.65768039e-01 2.04427019e-01 -4.37016368e-01 9.49138522e-01
1.85266569e-01 4.61902767e-01 -2.64237463e-01 -4.99997765e-01
-1.01886146e-01 -4.20155436e-01 7.55526721e-01 2.11190248e+00
4.25858438e-01 1.60275903e-02 -1.33966520e-01 7.48284459e-01
7.15036392e-01 8.49216938e-01 -7.02572346e-01 1.57984532e-02
1.09837353e+00 1.37211990e+00 -1.07891607e+00 -5.57847209e-02
-3.42885524e-01 5.44736862e-01 -3.43102291e-02 5.64449467e-04
-3.67708117e-01 -1.01152599e+00 6.43847808e-02 1.02012679e-01
-8.33924115e-03 -2.07267135e-01 -2.99366653e-01 -6.34856820e-01
1.33172482e-01 -1.11734855e+00 4.03538615e-01 -1.00706875e+00
-5.42125583e-01 8.29300880e-01 4.86266524e-01 -7.26684391e-01
-5.34482956e-01 -7.22353399e-01 -1.01281464e+00 1.04086208e+00
-4.19723779e-01 -1.36680329e+00 -1.58162072e-01 4.49630737e-01
4.22723889e-01 -2.77515948e-01 1.41773975e+00 4.76956427e-01
-4.72996801e-01 4.01111357e-02 1.71211019e-01 4.98375520e-02
2.77150840e-01 -1.52919590e+00 3.51153374e-01 5.81255674e-01
2.98080355e-01 6.10368431e-01 9.53069091e-01 -6.23548150e-01
-9.28522289e-01 -1.69184878e-01 1.34280694e+00 -3.36730123e-01
6.70854688e-01 -6.55187428e-01 -9.34144139e-01 6.82025731e-01
8.70854139e-01 -8.57468247e-01 8.82192194e-01 1.67492270e-01
-9.32494029e-02 2.66328275e-01 -1.02013671e+00 5.59694171e-01
6.31880283e-01 -3.66798550e-01 -9.25759256e-01 4.18834358e-01
2.73124963e-01 -3.98867756e-01 -1.03623235e+00 -5.19655585e-01
6.28113031e-01 -1.15225744e+00 5.49981177e-01 -6.80891797e-02
8.78153183e-03 -1.77272588e-01 2.62258798e-01 -1.07995665e+00
4.50143516e-01 -1.05022871e+00 4.87091452e-01 1.93491900e+00
4.72987324e-01 -8.53384137e-01 3.24735284e-01 4.58001018e-01
-4.95107830e-01 -2.66917632e-03 -1.27883017e+00 -4.62074816e-01
-2.82655627e-01 -1.00455678e+00 8.50299954e-01 1.24362016e+00
1.22977805e+00 2.29876786e-01 1.23690009e-01 -2.32420981e-01
-4.22202349e-02 -3.76289070e-01 9.48367119e-01 -1.54639399e+00
-1.88219994e-01 -7.52468348e-01 -1.73476562e-01 -1.22661283e-02
-4.42640364e-01 -6.65033638e-01 -5.91429710e-01 -1.74286330e+00
-2.11138695e-01 -3.48260254e-01 6.70772970e-01 4.72864807e-01
6.97709620e-01 -1.15931198e-01 5.12167454e-01 2.48470038e-01
-9.66666937e-02 -2.90644050e-01 7.94979990e-01 3.13663155e-01
-4.33582067e-01 -4.24464643e-01 -2.45967522e-01 4.44761068e-01
6.79198921e-01 -5.68119705e-01 -2.88716227e-01 -8.44474062e-02
3.64353269e-01 -6.40742248e-03 3.13684251e-03 -7.84857333e-01
1.44229736e-02 4.71447781e-02 -8.51711631e-02 -1.02966952e+00
4.20449972e-02 -8.56350899e-01 6.21129036e-01 4.10542846e-01
-2.16548428e-01 2.96557575e-01 5.49214125e-01 -8.19074214e-02
4.11879197e-02 -8.86007190e-01 -1.89977000e-03 -9.95840058e-02
-1.38658568e-01 -5.95973849e-01 -1.10789776e+00 -3.81972671e-01
1.14703178e+00 -5.95069885e-01 -6.49007142e-01 -9.51236859e-02
-5.82544208e-01 1.41694203e-01 5.00635803e-01 3.86647850e-01
1.74594805e-01 -7.84884155e-01 -2.88612396e-01 -6.02086559e-02
-6.82369899e-03 -1.22158840e-01 7.56109804e-02 2.93902963e-01
-9.21017349e-01 5.62165797e-01 -5.89421272e-01 -1.36803731e-01
-1.86315560e+00 4.77902174e-01 -7.54834935e-02 -4.40702379e-01
-9.23918307e-01 2.42442191e-01 -6.55584216e-01 -5.57317436e-01
1.76535651e-01 -3.74289930e-01 -8.05461466e-01 3.12344491e-01
9.51386690e-01 7.99472988e-01 3.13554317e-01 -5.02013385e-01
-3.36947590e-01 1.99082255e-01 4.03469324e-01 -5.78669846e-01
1.52384198e+00 -2.22681001e-01 -1.69639856e-01 5.84907889e-01
5.33972561e-01 4.37551141e-01 -5.49294412e-01 3.49779159e-01
5.14268100e-01 -3.89863938e-01 -2.72535354e-01 -1.10188937e+00
-3.93155992e-01 8.59201729e-01 6.52221441e-01 9.26067531e-01
8.84662569e-01 -1.18342966e-01 4.88929683e-03 1.74960241e-01
1.70980379e-01 -1.26104343e+00 -3.17813158e-01 1.94044799e-01
1.26321423e+00 -3.59955668e-01 -1.56021267e-01 -3.07875544e-01
-4.89976704e-01 1.83480597e+00 1.02593206e-01 4.96064276e-01
5.36423266e-01 7.29394734e-01 4.22505856e-01 -7.21888602e-01
-7.27566838e-01 -8.96963924e-02 -6.67986721e-02 1.11374807e+00
9.41594720e-01 1.64880250e-02 -1.24324012e+00 5.20249307e-01
-4.92865831e-01 3.15697759e-01 7.38439023e-01 1.11486554e+00
-4.35919762e-01 -1.49306667e+00 -7.53115714e-01 3.31113935e-02
-6.75506115e-01 9.46492478e-02 -7.94815660e-01 1.50475454e+00
2.02532366e-01 1.13995659e+00 5.15184522e-01 -9.81206633e-03
4.22502697e-01 1.39132023e-01 2.16145471e-01 -6.44060194e-01
-1.19993949e+00 6.20013587e-02 7.24130630e-01 -2.83869803e-01
-2.42500752e-01 -9.01439488e-01 -1.31689692e+00 -3.57970327e-01
-3.58620644e-01 7.09504902e-01 1.20423472e+00 7.57634401e-01
1.76913172e-01 4.78176564e-01 3.10286917e-02 -7.75078595e-01
1.72354594e-01 -1.59821892e+00 -6.55597508e-01 2.72843152e-01
-3.10441434e-01 -2.98839122e-01 1.70250952e-01 1.53996885e-01] | [9.503018379211426, 9.15194034576416] |
7dc9adf1-39d2-44d5-bf10-9d5faaf6eadb | dfac-framework-factorizing-the-value-function | 2102.07936 | null | https://arxiv.org/abs/2102.07936v2 | https://arxiv.org/pdf/2102.07936v2.pdf | DFAC Framework: Factorizing the Value Function via Quantile Mixture for Multi-Agent Distributional Q-Learning | In fully cooperative multi-agent reinforcement learning (MARL) settings, the environments are highly stochastic due to the partial observability of each agent and the continuously changing policies of the other agents. To address the above issues, we integrate distributional RL and value function factorization methods by proposing a Distributional Value Function Factorization (DFAC) framework to generalize expected value function factorization methods to their DFAC variants. DFAC extends the individual utility functions from deterministic variables to random variables, and models the quantile function of the total return as a quantile mixture. To validate DFAC, we demonstrate DFAC's ability to factorize a simple two-step matrix game with stochastic rewards and perform experiments on all Super Hard tasks of StarCraft Multi-Agent Challenge, showing that DFAC is able to outperform expected value function factorization baselines. | ['Chun-Yi Lee', 'Cheng-Kuang Lee', 'Wei-Fang Sun'] | 2021-02-16 | null | null | null | null | ['smac-1', 'smac'] | ['playing-games', 'playing-games'] | [-6.88505173e-01 2.55205631e-02 -2.18634114e-01 7.59868249e-02
-7.65216649e-01 -8.09033751e-01 6.41450942e-01 3.39852348e-02
-6.70360088e-01 1.32127357e+00 4.64433253e-01 -3.61968189e-01
-4.47167754e-01 -7.73882091e-01 -7.50552416e-01 -8.82939041e-01
-5.29410481e-01 8.58358979e-01 -1.91670924e-01 -5.66851199e-01
-2.76933938e-01 1.55018717e-01 -9.09303725e-01 1.40428841e-01
6.92461491e-01 8.01520944e-01 1.62018701e-01 9.17534351e-01
2.14114532e-01 1.35078824e+00 -8.59686971e-01 -4.04380292e-01
5.50664783e-01 -1.03546143e-01 -4.62945998e-01 -4.68284264e-02
-5.26220620e-01 -7.86967397e-01 -4.19750154e-01 8.39745879e-01
4.91672754e-01 7.83798695e-01 8.42399597e-01 -1.72555017e+00
-8.09786916e-01 1.09104431e+00 -7.03519106e-01 -6.44541206e-03
9.88654941e-02 5.29262543e-01 1.30969179e+00 -1.79587588e-01
2.32946560e-01 1.61042750e+00 3.55439961e-01 6.55561864e-01
-1.52914393e+00 -2.57145137e-01 5.49954116e-01 -2.81086922e-01
-4.64212656e-01 2.20871955e-01 1.10197254e-01 -4.91193920e-01
1.03414357e+00 -2.58352637e-01 6.25809669e-01 1.29533291e+00
4.28346455e-01 1.26341081e+00 1.22031164e+00 1.63358822e-01
6.37456775e-01 -3.69725943e-01 -2.12956265e-01 4.43616062e-01
1.62033569e-02 5.72549403e-01 -3.57171506e-01 -7.56896436e-01
8.24569702e-01 2.37973467e-01 6.19121119e-02 -5.60618401e-01
-1.13169134e+00 1.14799583e+00 4.94267717e-02 -3.46778393e-01
-8.48814428e-01 9.02481258e-01 5.09683192e-01 7.72973120e-01
2.70817488e-01 6.96346402e-01 -7.50022590e-01 -5.05092740e-01
-2.71075368e-01 9.74574566e-01 8.08138311e-01 8.83004189e-01
5.08026838e-01 4.30786312e-01 -7.50123680e-01 4.23321277e-01
1.85203150e-01 7.62052953e-01 4.43915308e-01 -1.48076832e+00
4.90850776e-01 1.35916829e-01 8.41472924e-01 -4.03380185e-01
-3.92343283e-01 -5.34157038e-01 -2.47167170e-01 7.09549546e-01
6.06226504e-01 -1.01623940e+00 -6.52505338e-01 2.20334601e+00
1.76724225e-01 -3.67921293e-02 3.99552971e-01 8.74359667e-01
1.25810325e-01 5.39423645e-01 1.17898118e-02 -1.95549965e-01
8.91102135e-01 -8.79003823e-01 -4.69114810e-01 -1.63936630e-01
5.45201063e-01 -1.62428141e-01 9.77638066e-01 6.13042116e-01
-1.04637182e+00 -3.72905135e-02 -7.58375466e-01 4.42727804e-01
1.85633942e-01 -1.84037015e-01 9.89245713e-01 3.47547084e-01
-1.04800785e+00 7.50506222e-01 -1.29739940e+00 4.73838478e-01
3.95726174e-01 3.11868131e-01 -1.37095824e-01 1.13738447e-01
-1.02887404e+00 5.78071296e-01 1.48384035e-01 -3.76442313e-01
-1.81457329e+00 -8.23995352e-01 -6.47315502e-01 3.50845009e-01
9.67906296e-01 -7.77904809e-01 1.76126063e+00 -8.46871138e-01
-1.72452676e+00 -3.18598449e-01 5.26969433e-01 -8.88236880e-01
7.75503516e-01 -3.72535795e-01 1.88992366e-01 -6.58151060e-02
2.11784437e-01 2.53350258e-01 9.01618004e-01 -1.03891575e+00
-8.10926378e-01 -5.94206788e-02 5.88348806e-01 4.48970795e-01
-6.94235116e-02 -4.58747119e-01 5.43326914e-01 -6.24599099e-01
-1.00706410e+00 -8.65781367e-01 -7.36078620e-01 -5.53522646e-01
1.23963773e-01 -2.95719504e-01 1.62255570e-01 -3.17154825e-01
8.49619985e-01 -2.01096654e+00 5.65747142e-01 1.11871939e-02
3.12782466e-01 -3.29783589e-01 -6.03130579e-01 6.40041351e-01
1.62033617e-01 -2.14072391e-01 1.46814510e-01 -4.26557809e-01
6.30286932e-01 6.39426410e-01 -4.63865370e-01 3.74188066e-01
-1.87839478e-01 1.09371078e+00 -1.21643806e+00 2.45417550e-01
-9.01630819e-02 5.59435599e-02 -7.95155764e-01 3.95319402e-01
-8.44169796e-01 1.24508545e-01 -6.14944160e-01 2.26696849e-01
3.79193664e-01 8.02671760e-02 3.74171555e-01 6.61069095e-01
1.97841018e-01 -3.12039815e-02 -1.17599654e+00 1.50319481e+00
-4.06131834e-01 4.38668914e-02 3.29170138e-01 -6.98651791e-01
3.67387682e-01 2.51450241e-01 9.63277102e-01 -5.00710726e-01
2.64170766e-01 -2.21127972e-01 1.59785867e-01 -1.04628526e-01
8.90474558e-01 -3.69250625e-01 -4.38170910e-01 1.02751994e+00
4.46829081e-01 -8.53051767e-02 2.65923619e-01 7.59563923e-01
1.30878174e+00 3.69245589e-01 1.38293400e-01 -3.82413745e-01
-7.47819543e-02 -3.03976890e-02 8.66585970e-01 1.13612723e+00
-3.54092807e-01 -2.88441987e-03 1.24196947e+00 -2.48827264e-01
-7.87544072e-01 -1.47355950e+00 7.37942755e-01 1.56680667e+00
-1.69944465e-01 -5.74073076e-01 -4.66881871e-01 -9.33638990e-01
7.31762469e-01 7.87455678e-01 -1.05675578e+00 -1.95475332e-02
-9.44866389e-02 -9.85333502e-01 2.04520434e-01 6.66976273e-01
-1.03422076e-01 -8.19922686e-01 -6.38224900e-01 5.36784828e-01
-4.67869034e-03 -6.72254622e-01 -6.78937674e-01 3.64920616e-01
-3.19901079e-01 -1.00850785e+00 -5.22900939e-01 1.13450609e-01
1.67067364e-01 2.00760603e-01 1.18099380e+00 -3.74137700e-01
9.90476608e-02 1.00753343e+00 -4.63793218e-01 -4.48936701e-01
-3.00261140e-01 -1.49704218e-01 3.97671342e-01 -1.92172602e-01
-1.69991180e-01 -4.97755319e-01 -6.76380873e-01 4.79095168e-02
-8.04810524e-01 -5.21773934e-01 3.14659864e-01 1.29408753e+00
3.12734962e-01 2.35419929e-01 8.60128105e-01 -6.25804722e-01
1.38671243e+00 -8.39767814e-01 -9.24786091e-01 1.38687134e-01
-4.67356384e-01 4.13978577e-01 7.72531390e-01 -7.76314497e-01
-1.12827957e+00 -1.53937280e-01 2.97173768e-01 -3.87826651e-01
4.26824838e-01 4.90796804e-01 9.72181261e-02 4.69054341e-01
6.29797101e-01 9.39625725e-02 6.21949852e-01 -9.15388837e-02
6.62248671e-01 6.62578344e-02 3.60913664e-01 -1.32156742e+00
7.14473248e-01 2.69360840e-01 -1.18285995e-02 -1.65127620e-01
-4.86968011e-01 1.15369841e-01 2.20816448e-01 -3.49844337e-01
6.48325622e-01 -1.31611645e+00 -1.53346288e+00 3.75535011e-01
-7.13558495e-01 -1.07532275e+00 -7.98036993e-01 6.92555428e-01
-1.10736632e+00 1.30739078e-01 -7.90815294e-01 -1.17577684e+00
2.95198169e-02 -1.21848571e+00 6.64740980e-01 2.99419254e-01
3.93962115e-01 -1.10447788e+00 3.14461470e-01 3.83421518e-02
4.73021895e-01 3.94572541e-02 8.08172822e-01 -4.84005958e-01
-3.55659157e-01 3.83506179e-01 3.05377007e-01 3.98939729e-01
-2.20007338e-02 -3.06745350e-01 -4.14486021e-01 -7.37295032e-01
-1.91909820e-01 -8.42722833e-01 8.59398246e-01 5.68342090e-01
5.21274805e-01 -3.93291652e-01 1.06286027e-01 2.51890361e-01
1.17370570e+00 1.63183466e-01 1.44535422e-01 3.61653686e-01
1.63690895e-01 2.47787923e-01 9.12837327e-01 1.28802609e+00
7.86035299e-01 2.49175519e-01 9.48573589e-01 3.64810199e-01
5.13578117e-01 -5.28676689e-01 1.08148551e+00 -1.78566799e-01
-1.71194881e-01 -2.35187858e-01 -2.91913152e-01 4.71185535e-01
-2.44377422e+00 -1.25357485e+00 3.05932611e-01 2.05470228e+00
6.76096976e-01 -3.83349322e-02 1.01981580e+00 -5.90911567e-01
1.57824174e-01 1.26526535e-01 -9.28153038e-01 -4.83093441e-01
-9.14660916e-02 -3.09141930e-02 7.80515611e-01 6.15884244e-01
-8.76801014e-01 9.33263659e-01 6.56739378e+00 1.01609051e+00
-5.08544981e-01 2.27429360e-01 4.94977355e-01 -7.19640672e-01
-5.09753883e-01 -9.40988213e-02 -3.74505490e-01 4.08645540e-01
6.15756571e-01 -4.84452933e-01 1.18261838e+00 1.00002778e+00
1.50302812e-01 -1.24301560e-01 -9.14556205e-01 6.15275681e-01
-5.33353925e-01 -9.91298556e-01 -4.62028652e-01 4.26316977e-01
9.09395278e-01 4.63593096e-01 3.42745990e-01 1.01397872e+00
2.02134395e+00 -1.24164677e+00 7.48977661e-01 4.39926058e-01
3.27791214e-01 -1.20123160e+00 6.44238412e-01 5.14216423e-01
-9.04767692e-01 -7.29910493e-01 -5.67308486e-01 -5.63008845e-01
4.20214273e-02 1.14058845e-01 -7.95679986e-01 4.99194413e-01
3.92344773e-01 3.46709967e-01 -6.38857782e-02 4.56239343e-01
-2.54675925e-01 4.89708871e-01 -2.46779844e-01 -1.27174472e-02
6.73337162e-01 -4.10180628e-01 5.76959491e-01 5.43205380e-01
1.34657428e-01 -4.42727357e-02 7.26220846e-01 8.81195247e-01
-3.74512151e-02 -1.09196536e-01 -2.70622611e-01 -3.40762585e-01
1.74100026e-01 1.36834097e+00 -2.94410288e-01 -2.39511654e-02
-3.90630394e-01 7.61194050e-01 6.42096639e-01 5.20436823e-01
-8.66804659e-01 1.88598838e-02 1.40810633e+00 -3.35711032e-01
4.42138582e-01 -5.67775309e-01 2.87674308e-01 -1.41244292e+00
-2.20171392e-01 -1.05625594e+00 6.37729168e-01 -3.90548378e-01
-1.66058540e+00 3.02391142e-01 -1.29432261e-01 -9.38179374e-01
-8.53673398e-01 -5.45032382e-01 -4.81338769e-01 6.99942827e-01
-1.31599891e+00 -9.66562569e-01 4.87424165e-01 9.11853135e-01
2.57995874e-01 -6.96996272e-01 6.42271280e-01 -3.23847890e-01
-2.59320557e-01 4.72446114e-01 6.15098000e-01 2.96002012e-02
5.59210300e-01 -1.82486796e+00 2.71372378e-01 6.15812063e-01
-8.29875544e-02 2.68640578e-01 7.45410204e-01 -6.47450507e-01
-1.76588249e+00 -8.27306032e-01 -4.67905730e-01 -4.22339797e-01
1.21896100e+00 -4.34927791e-01 -3.86929452e-01 8.39551151e-01
3.92120183e-01 3.76962163e-02 6.40675545e-01 2.17083633e-01
-2.67971754e-01 2.15364665e-01 -1.19221389e+00 4.97209638e-01
7.57439613e-01 -1.88713923e-01 -3.64674479e-01 1.43350437e-01
9.23237145e-01 -2.91710198e-01 -1.03835785e+00 -1.33728832e-01
4.47653741e-01 -7.76549280e-01 7.77957857e-01 -1.11000705e+00
4.20881182e-01 -9.86937582e-02 -4.20976788e-01 -2.22222042e+00
-4.96426642e-01 -1.01763046e+00 -2.84393609e-01 1.07823849e+00
3.52046162e-01 -8.27537656e-01 7.51541495e-01 5.69058359e-01
1.59090832e-01 -5.56690335e-01 -9.70886409e-01 -9.32031929e-01
6.87613964e-01 -2.88102448e-01 8.52594435e-01 5.72071433e-01
1.97707012e-01 2.19676152e-01 -8.74067724e-01 -3.01701147e-02
7.96009779e-01 2.24045336e-01 9.42290127e-01 -7.50568688e-01
-1.38098824e+00 -3.95030230e-01 6.07306845e-02 -1.00998354e+00
4.71327811e-01 -7.47958958e-01 -3.92314838e-03 -1.27897918e+00
2.56870031e-01 -4.48140681e-01 -4.14334178e-01 4.69293177e-01
-2.97123969e-01 -4.05059725e-01 7.17847109e-01 -3.57677788e-01
-9.00325239e-01 1.09763014e+00 1.29062915e+00 -2.13360071e-01
-3.50306362e-01 2.90627509e-01 -1.14738417e+00 4.01565284e-01
6.94160819e-01 -3.06814104e-01 -9.13515329e-01 -4.91892189e-01
4.86960709e-01 6.40438735e-01 5.65717407e-02 -4.31608021e-01
-1.17323831e-01 -9.63639677e-01 2.21480995e-01 -2.54439890e-01
2.45646030e-01 -5.27363956e-01 -2.03115597e-01 4.14465368e-01
-4.14633036e-01 4.30469155e-01 -6.62026852e-02 9.88025129e-01
1.25725612e-01 1.52036995e-01 4.26185101e-01 -1.94731623e-01
-3.64730835e-01 5.47227323e-01 -9.34049726e-01 4.37561303e-01
1.02035761e+00 4.86273408e-01 -4.51377809e-01 -1.02250147e+00
-8.46445024e-01 1.15335357e+00 3.36481899e-01 1.66270629e-01
5.90557396e-01 -1.13178527e+00 -8.40340793e-01 -1.26292109e-01
-2.55974799e-01 -5.70851713e-02 4.73980606e-01 5.09393811e-01
8.58822763e-02 -3.02473009e-01 -4.20952678e-01 3.45001146e-02
-7.68954873e-01 7.23791122e-01 4.98588562e-01 -7.82328248e-01
-3.51698518e-01 4.99127775e-01 3.00701052e-01 -5.74807644e-01
1.01910658e-01 -1.94704413e-01 -8.90732091e-03 5.42751513e-02
6.31074190e-01 4.75517005e-01 -4.92813289e-01 -1.08080328e-01
4.48934026e-02 -4.40661371e-01 -1.94135889e-01 -7.17299402e-01
1.77407396e+00 -2.07486562e-02 2.52040416e-01 1.62316293e-01
3.55396569e-01 -1.27836848e-02 -2.02763891e+00 -3.52567405e-01
-1.61323369e-01 -3.80312204e-01 -9.83418152e-02 -1.12539184e+00
-1.14127350e+00 4.31916714e-01 8.85303915e-02 1.70778304e-01
7.82971799e-01 -2.35154077e-01 4.20606285e-01 3.97741079e-01
8.09608459e-01 -1.26378727e+00 3.34746927e-01 8.62432420e-01
7.83009887e-01 -9.77033675e-01 -7.21733272e-02 5.30871809e-01
-1.48301172e+00 8.38032246e-01 5.65468609e-01 -5.21160424e-01
6.52333260e-01 7.81975269e-01 -7.01905116e-02 2.35613614e-01
-1.54803264e+00 -4.27514434e-01 -4.33243632e-01 9.92012024e-01
-1.39293268e-01 6.56419456e-01 -1.44234702e-01 1.26647878e+00
-3.24651450e-01 -4.23468724e-02 1.04870307e+00 9.11528826e-01
-1.62326112e-01 -1.43559325e+00 -2.96941370e-01 5.82598150e-01
-6.22669101e-01 1.33605957e-01 -3.77842151e-02 6.16137505e-01
-3.69470149e-01 1.12257016e+00 2.57338714e-02 -3.39508116e-01
1.34067714e-01 -2.89513737e-01 5.72606146e-01 -4.61998463e-01
-9.17351305e-01 3.28575402e-01 -1.62430838e-01 -6.10672593e-01
3.79597284e-02 -8.75633061e-01 -1.46929371e+00 -5.59510291e-01
1.28249258e-01 4.41811264e-01 3.27441365e-01 7.52855003e-01
3.58520508e-01 7.61059344e-01 7.32892275e-01 -6.57504320e-01
-1.63144290e+00 -9.24071729e-01 -1.07202089e+00 3.85260761e-01
5.21524370e-01 -1.09583390e+00 -2.99957514e-01 -7.86067128e-01] | [3.709299325942993, 2.018826484680176] |
baae4fa7-9387-4000-a1ef-0ccf6336d277 | speech-denoising-without-clean-training-data | 2104.03838 | null | https://arxiv.org/abs/2104.03838v2 | https://arxiv.org/pdf/2104.03838v2.pdf | Speech Denoising Without Clean Training Data: A Noise2Noise Approach | This paper tackles the problem of the heavy dependence of clean speech data required by deep learning based audio-denoising methods by showing that it is possible to train deep speech denoising networks using only noisy speech samples. Conventional wisdom dictates that in order to achieve good speech denoising performance, there is a requirement for a large quantity of both noisy speech samples and perfectly clean speech samples, resulting in a need for expensive audio recording equipment and extremely controlled soundproof recording studios. These requirements pose significant challenges in data collection, especially in economically disadvantaged regions and for low resource languages. This work shows that speech denoising deep neural networks can be successfully trained utilizing only noisy training audio. Furthermore it is revealed that such training regimes achieve superior denoising performance over conventional training regimes utilizing clean training audio targets, in cases involving complex noise distributions and low Signal-to-Noise ratios (high noise environments). This is demonstrated through experiments studying the efficacy of our proposed approach over both real-world noises and synthetic noises using the 20 layered Deep Complex U-Net architecture. | ['S Natarajan', 'Krishnamoorthy Manohara', 'Anuj Tambwekar', 'Madhav Mahesh Kashyap'] | 2021-04-08 | null | null | null | null | ['audio-denoising', 'speech-denoising'] | ['audio', 'speech'] | [ 3.13599259e-01 -1.09202422e-01 7.21040010e-01 -2.95401573e-01
-1.09811080e+00 -2.13059828e-01 3.66364598e-01 -4.39817868e-02
-7.69819140e-01 6.93543971e-01 2.14618310e-01 -3.46719742e-01
-5.67769744e-02 -7.30063975e-01 -6.24888957e-01 -9.56171095e-01
-1.11770049e-01 -1.58856064e-01 -1.30788118e-01 -6.11817777e-01
-2.06277668e-01 4.38019395e-01 -1.81507695e+00 -4.09135449e-04
6.08305991e-01 9.80913877e-01 5.39443672e-01 1.00501323e+00
2.03571450e-02 5.95831811e-01 -1.18082726e+00 -2.70527244e-01
4.98281211e-01 -3.54900211e-01 -1.05242580e-01 -1.53835699e-01
2.80849099e-01 -6.90423965e-01 -4.79849517e-01 1.35955977e+00
1.04755032e+00 5.74146330e-01 2.28976429e-01 -5.92653632e-01
-3.44718486e-01 5.24270415e-01 -2.89708644e-01 4.38406169e-01
4.38738652e-02 1.78709805e-01 6.24473155e-01 -6.52813673e-01
7.98924342e-02 1.19015288e+00 8.03419709e-01 5.71615160e-01
-1.02554333e+00 -8.01599205e-01 -5.10443076e-02 1.34372234e-01
-9.60035861e-01 -1.13226008e+00 8.28425944e-01 2.69404482e-02
1.06639171e+00 3.29108030e-01 4.02180612e-01 1.42227089e+00
-6.17131367e-02 3.09261292e-01 1.00974441e+00 -5.27075052e-01
5.12185454e-01 -3.02179419e-02 -6.76177219e-02 5.98309971e-02
4.39568087e-02 2.01670483e-01 -5.80382109e-01 1.27153873e-01
5.03562987e-01 -2.91578561e-01 -3.68884683e-01 5.33457339e-01
-5.82451642e-01 5.16177952e-01 2.12817729e-01 6.07094228e-01
-5.56519628e-01 2.46854201e-01 7.97211349e-01 7.49863565e-01
7.81085432e-01 1.45739410e-02 -3.83592606e-01 -4.86520082e-01
-1.14501870e+00 5.08441553e-02 6.87995970e-01 7.26530313e-01
3.35612774e-01 1.02671480e+00 4.40537751e-01 1.26446545e+00
1.86947048e-01 5.16760111e-01 4.41640347e-01 -8.73883188e-01
5.76037765e-01 -4.71144527e-01 7.01207388e-03 -1.00016296e+00
-4.52480875e-02 -9.32778895e-01 -1.36928582e+00 4.94302303e-01
2.72015363e-01 -4.31682736e-01 -1.01864505e+00 1.80971885e+00
2.06004441e-01 2.57740170e-01 3.41474026e-01 8.46806765e-01
6.82970166e-01 9.44159567e-01 -2.87589803e-02 -4.13732946e-01
9.48072076e-01 -4.85017329e-01 -1.13229156e+00 -7.55362958e-02
1.22761481e-01 -9.11612868e-01 1.23664868e+00 8.92971337e-01
-1.19871140e+00 -7.24877596e-01 -1.19269228e+00 5.81165552e-02
-2.93280333e-01 -7.29703233e-02 2.54079849e-01 1.22356772e+00
-1.25919044e+00 6.98718905e-01 -8.94535959e-01 -9.89020169e-02
3.41415882e-01 4.07380879e-01 -4.60326016e-01 -1.30562380e-01
-1.18102062e+00 5.08972466e-01 -4.49412651e-02 8.79343152e-01
-1.27607989e+00 -5.89414239e-01 -9.09817636e-01 4.12557393e-01
8.97249058e-02 -2.43893445e-01 1.30138278e+00 -9.14414346e-01
-1.59730065e+00 2.53787965e-01 -9.54376906e-02 -6.89399660e-01
6.97226524e-01 -5.77824235e-01 -5.28267980e-01 2.18351275e-01
-2.15063170e-01 -5.02981208e-02 1.28002846e+00 -1.34741652e+00
-4.11335438e-01 -3.75510037e-01 -9.05451030e-02 1.97879113e-02
-6.49128377e-01 8.78921002e-02 8.36165398e-02 -8.62273574e-01
2.51923591e-01 -2.79055536e-01 -3.00884604e-01 -4.22164947e-01
-2.12281063e-01 2.71270245e-01 1.02140057e+00 -1.13916016e+00
7.73169756e-01 -2.26388836e+00 -2.03607857e-01 2.34220371e-01
7.64105842e-02 5.96118867e-01 -1.93664387e-01 4.02598828e-01
-1.97293058e-01 1.24197707e-01 -5.19669019e-02 -7.92970598e-01
-5.35347015e-02 3.37091297e-01 -1.77492172e-01 5.14404893e-01
1.01439200e-01 -2.50152461e-02 -7.24012136e-01 -4.48257476e-02
3.60191762e-01 9.37667251e-01 -4.50967312e-01 3.44228804e-01
2.15977520e-01 5.09100080e-01 -5.15950657e-02 5.02494395e-01
8.19690406e-01 8.33942294e-01 -7.04935491e-02 -1.34962082e-01
-4.75956276e-02 2.90440857e-01 -1.45136058e+00 1.40832400e+00
-9.68366623e-01 8.63879263e-01 1.12776065e+00 -1.41057658e+00
1.08477128e+00 8.30639839e-01 1.17593169e-01 -8.86682093e-01
3.26350957e-01 4.25167024e-01 -1.18128702e-01 -6.90781593e-01
3.06084037e-01 -6.21701121e-01 3.89546692e-01 7.54109696e-02
2.22098649e-01 -2.79378861e-01 7.36667216e-02 -8.08634758e-02
1.25604427e+00 -2.44783178e-01 -2.60581255e-01 -1.57662064e-01
5.15211284e-01 -6.21683002e-01 6.65519118e-01 9.34502840e-01
-3.63800079e-01 5.37672102e-01 2.40043908e-01 -6.14955574e-02
-1.15130746e+00 -1.01057625e+00 -4.01118435e-02 1.07951522e+00
-4.51733440e-01 -7.86067452e-03 -9.61883426e-01 7.51850978e-02
-6.02598429e-01 5.17176211e-01 -3.62967938e-01 5.22063999e-03
-7.14884162e-01 -5.15520513e-01 7.91215479e-01 2.21654907e-01
5.74474990e-01 -1.05133224e+00 -2.70270944e-01 3.70891362e-01
-4.95311469e-02 -1.07282090e+00 1.62575040e-02 7.40151823e-01
-8.19665790e-01 -5.70623755e-01 -9.39733803e-01 -1.06165659e+00
4.07788277e-01 3.08035165e-01 9.17679846e-01 8.70170966e-02
-8.47124904e-02 2.15457216e-01 -4.70966995e-01 -5.72280109e-01
-7.79984713e-01 -1.89223200e-01 5.40468633e-01 -8.79761502e-02
1.40742689e-01 -1.27411842e+00 -5.80697775e-01 -1.07033215e-01
-1.08013308e+00 -4.16294724e-01 4.57538903e-01 1.14589500e+00
2.53911823e-01 7.15258837e-01 1.15269458e+00 -3.08144718e-01
1.01002395e+00 -2.78169125e-01 -6.46071792e-01 -1.57268450e-01
1.16387598e-01 -3.68547499e-01 1.03337646e+00 -3.05492550e-01
-1.38802099e+00 -2.59280413e-01 -9.53273177e-01 -2.55226880e-01
-5.30756056e-01 4.82301056e-01 -3.44318092e-01 1.69304878e-01
7.08037972e-01 3.03549320e-01 7.58578405e-02 -8.20594370e-01
7.24584609e-02 1.14662182e+00 7.13402808e-01 -4.95232403e-01
8.52650404e-01 3.53156030e-01 -1.29350677e-01 -1.71210790e+00
-3.72940570e-01 -4.31159496e-01 -4.06561017e-01 -1.93771809e-01
6.07916892e-01 -9.49181020e-01 -6.00584090e-01 6.87003970e-01
-1.08228087e+00 -9.61726531e-02 -1.71185464e-01 5.70358872e-01
-2.47066423e-01 5.27786672e-01 -8.18978667e-01 -1.42366302e+00
-5.70399821e-01 -1.32527518e+00 7.43789971e-01 -1.07500050e-02
3.53884250e-02 -8.15950513e-01 -2.93056220e-01 2.89645642e-01
6.90765321e-01 9.30299684e-02 7.22648561e-01 -4.46848422e-01
-2.59709835e-01 -2.79102176e-01 1.41525567e-01 1.20739281e+00
4.06528294e-01 -2.02482998e-01 -1.43384218e+00 -3.74599010e-01
7.13215411e-01 -4.96304065e-01 7.12701797e-01 6.07974350e-01
9.05298293e-01 -1.77630901e-01 6.07750952e-01 5.79846680e-01
1.33805656e+00 3.69812459e-01 7.19568551e-01 7.42685869e-02
3.25644821e-01 6.46349072e-01 1.75880432e-01 3.59793603e-01
-4.30086434e-01 1.14207119e-01 4.47668761e-01 -2.02281326e-01
-1.97629467e-01 6.70564175e-02 2.72690266e-01 1.39657319e+00
3.17950500e-03 -2.17230499e-01 -6.19827569e-01 9.02121305e-01
-1.22308731e+00 -1.02095735e+00 -6.61831424e-02 2.16434121e+00
9.40543592e-01 2.77404994e-01 -1.22538060e-01 9.42591071e-01
4.00189668e-01 5.69666252e-02 -1.68222159e-01 -4.93573636e-01
-1.53180376e-01 7.89899409e-01 1.77865058e-01 6.53389394e-01
-1.05589890e+00 5.28688490e-01 5.87753201e+00 9.42084968e-01
-1.13652241e+00 2.59895265e-01 6.03353322e-01 -2.14612409e-01
-1.86192334e-01 -6.31766379e-01 -2.38723755e-01 3.37272495e-01
1.29960775e+00 2.53863633e-01 4.61108834e-01 7.01303482e-01
9.71627593e-01 -1.83444396e-01 -7.24794030e-01 1.15270972e+00
-4.02955323e-01 -1.05165470e+00 -2.67129093e-01 -3.40876311e-01
4.12322193e-01 1.08127601e-01 1.86012983e-01 3.06427687e-01
1.22508965e-01 -1.08083129e+00 6.26290977e-01 6.38671070e-02
6.09858155e-01 -8.43437433e-01 8.97457719e-01 3.97349507e-01
-9.00739372e-01 -1.05128542e-01 -5.20973146e-01 -1.90972984e-01
2.33147725e-01 9.64532852e-01 -6.91608191e-01 4.43734288e-01
1.14357448e+00 2.82614976e-01 1.71478420e-01 9.79057312e-01
-6.30621612e-02 1.08725417e+00 -5.37119627e-01 1.00147776e-01
3.19145858e-01 -2.89151818e-01 6.03296518e-01 1.43839502e+00
5.33408523e-01 2.01979708e-02 -3.86803001e-01 3.71719688e-01
-1.64281622e-01 -6.66525587e-02 -7.51461148e-01 4.31580096e-02
2.67357469e-01 9.18535292e-01 -5.68206370e-01 -1.11822657e-01
-2.00949892e-01 7.68719614e-01 -2.07447082e-01 7.00497627e-01
-4.46125388e-01 -6.27788126e-01 8.85141611e-01 -4.35104370e-02
3.02978814e-01 -6.05841160e-01 -5.15335083e-01 -8.44608188e-01
1.59197733e-01 -1.20790172e+00 -2.23395199e-01 -5.97935915e-01
-1.14297152e+00 8.44916105e-01 -4.26884115e-01 -1.01388919e+00
-1.93234071e-01 -4.90622878e-01 -7.08824098e-01 1.21938753e+00
-1.53100014e+00 -7.99029291e-01 -2.80366808e-01 6.50142372e-01
9.19078946e-01 -3.08472574e-01 8.56663048e-01 9.29987848e-01
-4.82202947e-01 4.87224936e-01 5.71030617e-01 1.98049676e-02
3.27079833e-01 -1.17387593e+00 5.00571012e-01 1.08654451e+00
6.23256117e-02 7.28973985e-01 1.29173899e+00 -3.54580402e-01
-1.39533973e+00 -7.94856966e-01 4.88265306e-01 2.44922310e-01
5.41074395e-01 -7.26791501e-01 -1.14895487e+00 9.18739289e-02
4.42454517e-01 -2.56942332e-01 5.35416067e-01 -9.09276381e-02
-2.27739848e-02 -4.99898940e-01 -1.24032736e+00 4.84732777e-01
7.40884364e-01 -6.43603623e-01 -4.32324111e-01 3.63824844e-01
6.57265007e-01 -1.62540033e-01 -5.98927379e-01 2.52459422e-02
3.47840041e-01 -1.03875303e+00 9.27671850e-01 -5.01471758e-01
1.57313317e-01 -1.88091457e-01 -4.70306605e-01 -1.53040671e+00
2.90753752e-01 -9.80886996e-01 2.78225392e-01 1.34212732e+00
2.83301502e-01 -3.83757561e-01 7.77690649e-01 2.39840537e-01
-4.57830906e-01 -3.90381843e-01 -1.34971702e+00 -7.73396134e-01
1.26184791e-01 -9.22209799e-01 1.54661104e-01 6.10696912e-01
-5.41793764e-01 9.63839740e-02 -5.69705844e-01 4.39796656e-01
7.40150571e-01 -9.15754914e-01 6.80881441e-01 -8.76522362e-01
-2.91112393e-01 -1.98408086e-02 -4.03460801e-01 -8.58314514e-01
1.22645445e-01 -9.19643566e-02 3.91060621e-01 -1.58252335e+00
-6.68617904e-01 -3.87550682e-01 -1.92973584e-01 2.65715439e-02
7.81894550e-02 2.92807668e-01 -5.38138784e-02 -2.49999076e-01
6.89305365e-02 8.18813205e-01 8.63074660e-01 -2.52786815e-01
-1.52066216e-01 2.74920344e-01 -4.17137742e-01 7.34140992e-01
7.46792734e-01 -4.68477756e-01 -6.86546862e-01 -7.50322759e-01
4.42847833e-02 -2.02971920e-02 2.26008788e-01 -1.21109283e+00
1.92471832e-01 3.19095820e-01 1.46043748e-01 -4.53004867e-01
7.44939268e-01 -1.16975474e+00 -1.93880014e-02 3.66525143e-01
-2.00682431e-01 -1.82420567e-01 3.50541621e-01 6.47406697e-01
-6.00930512e-01 -3.23534966e-01 8.88762891e-01 -2.46019110e-01
-4.31293994e-01 -3.12389582e-01 -6.20658457e-01 -3.15577507e-01
5.23246169e-01 -3.22500706e-01 8.34792703e-02 -9.60195422e-01
-1.05503702e+00 -2.62276232e-01 -2.79204100e-01 1.82765990e-01
6.87879562e-01 -9.16184783e-01 -7.23852277e-01 4.49715555e-01
-6.84459805e-01 1.13759056e-01 6.41995370e-01 6.59278333e-01
-6.99164927e-01 2.28857920e-02 -3.30749787e-02 -4.36482280e-01
-1.40010524e+00 5.46529219e-02 5.13547361e-01 2.23452896e-01
-7.74132192e-01 1.04534590e+00 -1.22635722e-01 -2.66204953e-01
8.25491488e-01 -4.93231773e-01 1.87919587e-01 -6.98247999e-02
7.64901817e-01 3.98208261e-01 7.60186493e-01 -3.27366620e-01
1.10524490e-01 7.11787939e-02 8.43940526e-02 -3.17676097e-01
1.73601127e+00 -8.00216496e-02 -2.11246386e-02 2.10480943e-01
1.33244014e+00 1.35326192e-01 -1.16078591e+00 -9.46703628e-02
-2.54914016e-01 -4.40426677e-01 6.32628739e-01 -5.38264394e-01
-1.07377672e+00 1.17478025e+00 8.80631030e-01 5.55450797e-01
1.56245100e+00 -6.42646670e-01 9.86523867e-01 7.32640803e-01
2.63622552e-01 -1.27394867e+00 -6.29290789e-02 4.93125588e-01
9.12711084e-01 -1.17997229e+00 -3.60466808e-01 -4.92996052e-02
-1.55124813e-01 9.80921507e-01 1.40675589e-01 -9.36842114e-02
9.29129303e-01 5.94311655e-01 6.16577268e-01 1.02483742e-01
-3.44341427e-01 -6.33975193e-02 -2.80962437e-01 1.00042450e+00
4.00478035e-01 -2.29919031e-01 6.57247454e-02 5.61677396e-01
-5.35577595e-01 -3.58750999e-01 5.81373751e-01 1.03313947e+00
-5.60887516e-01 -9.64371681e-01 -8.06446075e-01 3.91770512e-01
-1.02261567e+00 -2.51168668e-01 1.87512487e-02 5.71211278e-01
-7.90193528e-02 1.49451637e+00 -1.64276049e-01 -2.70275235e-01
5.03061533e-01 -6.42207488e-02 1.93455089e-02 -4.18414950e-01
-8.46131206e-01 4.62619096e-01 2.83230156e-01 -1.08316287e-01
-3.93993884e-01 -2.96077669e-01 -7.66486108e-01 -4.26974446e-01
-3.42691571e-01 2.42570147e-01 1.22817886e+00 8.68755341e-01
1.63230672e-02 8.87986422e-01 6.56122088e-01 -1.10903358e+00
-9.96975839e-01 -1.28276646e+00 -8.34097743e-01 2.42565185e-01
9.63280559e-01 -2.73527324e-01 -6.64445877e-01 1.38565227e-01] | [15.085322380065918, 5.934104919433594] |
59c58e7d-b977-4852-802a-fa2f8021f5aa | kanerva-extending-the-kanerva-machine-with-1 | 2103.03905 | null | https://arxiv.org/abs/2103.03905v3 | https://arxiv.org/pdf/2103.03905v3.pdf | Kanerva++: extending The Kanerva Machine with differentiable, locally block allocated latent memory | Episodic and semantic memory are critical components of the human memory model. The theory of complementary learning systems (McClelland et al., 1995) suggests that the compressed representation produced by a serial event (episodic memory) is later restructured to build a more generalized form of reusable knowledge (semantic memory). In this work we develop a new principled Bayesian memory allocation scheme that bridges the gap between episodic and semantic memory via a hierarchical latent variable model. We take inspiration from traditional heap allocation and extend the idea of locally contiguous memory to the Kanerva Machine, enabling a novel differentiable block allocated latent memory. In contrast to the Kanerva Machine, we simplify the process of memory writing by treating it as a fully feed forward deterministic process, relying on the stochasticity of the read key distribution to disperse information within the memory. We demonstrate that this allocation scheme improves performance in memory conditional image generation, resulting in new state-of-the-art conditional likelihood values on binarized MNIST (<=41.58 nats/image) , binarized Omniglot (<=66.24 nats/image), as well as presenting competitive performance on CIFAR10, DMLab Mazes, Celeb-A and ImageNet32x32. | ['Alexandros Kalousis', 'Yan Wu', 'Jason Ramapuram'] | 2021-02-20 | kanerva-extending-the-kanerva-machine-with | https://openreview.net/forum?id=QoWatN-b8T | https://openreview.net/pdf?id=QoWatN-b8T | iclr-2021-1 | ['conditional-image-generation'] | ['computer-vision'] | [-2.04036776e-02 8.23706090e-02 -1.07279435e-01 -4.42061156e-01
-6.02584958e-01 -1.72655523e-01 9.90558684e-01 2.69775148e-02
-8.44832301e-01 1.09156406e+00 2.15463385e-01 -3.44922990e-01
-4.12861407e-01 -1.13615906e+00 -1.00523317e+00 -8.20434272e-01
-3.57016414e-01 6.78515077e-01 4.44992810e-01 3.55800897e-01
3.46471608e-01 3.40442628e-01 -1.57991683e+00 3.35988283e-01
4.32891399e-01 8.34578097e-01 6.51117206e-01 7.05420494e-01
-5.15712947e-02 1.12863278e+00 -5.20210385e-01 8.43080282e-02
-1.00791916e-01 -2.92754263e-01 -1.15836918e+00 -2.93959230e-01
-7.87293166e-03 -1.87417224e-01 -7.90159225e-01 8.26618433e-01
4.41309184e-01 5.14639854e-01 8.06978643e-01 -9.32355881e-01
-7.54363775e-01 7.14620411e-01 -4.10344303e-01 7.10094631e-01
-2.08790585e-01 -1.70625765e-02 5.72797596e-01 -8.91132295e-01
5.05051851e-01 1.33219302e+00 4.19021547e-01 2.93204814e-01
-1.34360254e+00 -7.33859658e-01 5.06272987e-02 4.49519157e-01
-1.54047120e+00 -5.09408951e-01 8.07397184e-04 -4.26615089e-01
1.61330974e+00 -1.07426532e-02 5.37689507e-01 1.10989666e+00
6.31504297e-01 7.95193374e-01 1.47518504e+00 -5.52256823e-01
4.78577942e-01 -9.11156759e-02 5.94591022e-01 7.69126654e-01
8.97287428e-02 4.32681054e-01 -1.23172331e+00 -2.32214198e-01
7.37954021e-01 -7.95715898e-02 -1.98887158e-02 -3.50871503e-01
-1.15196037e+00 9.40950394e-01 3.28806370e-01 -1.10554181e-01
-3.67996573e-01 6.36269689e-01 1.66929275e-01 1.27201840e-01
2.48961926e-01 -5.94362170e-02 -3.08082521e-01 -1.65963829e-01
-1.21569848e+00 1.27659231e-01 8.14714968e-01 9.80816543e-01
7.36684501e-01 2.46970832e-01 -2.18521103e-01 5.28049052e-01
5.61291695e-01 8.02067459e-01 7.99986780e-01 -8.21928859e-01
1.27252907e-01 -1.99337482e-01 5.44316620e-02 -6.67206764e-01
-3.74256015e-01 -4.89251912e-01 -8.21778476e-01 1.74645796e-01
1.89714748e-02 3.07429254e-01 -1.29215419e+00 1.82618415e+00
-2.93566108e-01 2.98351109e-01 1.81806356e-01 5.78850746e-01
4.75467145e-01 8.87937009e-01 6.19188190e-01 -1.24385186e-01
1.33600664e+00 -1.11215413e+00 -5.45107901e-01 -4.81741220e-01
3.52700263e-01 -2.98808217e-01 7.38371074e-01 3.91522497e-01
-1.24044371e+00 -6.09042525e-01 -1.16961932e+00 -2.27596592e-02
-6.43396914e-01 -2.18091175e-01 9.23915863e-01 7.02263176e-01
-1.28860056e+00 3.94539654e-01 -1.19251144e+00 -1.07853651e-01
5.39447963e-01 2.21378133e-01 -2.44556502e-01 -7.17285275e-02
-1.40190244e+00 1.03375304e+00 9.27722275e-01 -2.23724842e-01
-1.35484600e+00 -5.02004445e-01 -6.73959553e-01 2.18469426e-01
1.43454701e-01 -8.64588439e-01 9.87205386e-01 -5.44153273e-01
-1.35817003e+00 7.64991105e-01 -2.26636767e-01 -1.04024732e+00
3.81665379e-02 -2.97710657e-01 -2.36957058e-01 1.68956786e-01
-2.24049501e-02 1.10084820e+00 9.32065189e-01 -8.79576325e-01
-5.83186686e-01 -2.47289419e-01 -3.85453105e-01 1.97282419e-01
-1.24974838e-02 -1.16596065e-01 -2.03653440e-01 -8.69541526e-01
2.01375842e-01 -9.95846212e-01 5.08255232e-03 -3.94934207e-01
-8.19282085e-02 1.32417651e-02 4.91150379e-01 -5.92033863e-01
1.07068133e+00 -2.28336024e+00 3.14708233e-01 3.06614399e-01
1.28904626e-01 -1.36653408e-01 -1.13607133e-02 1.37543604e-01
-6.20673969e-02 -1.26888111e-01 -2.96524346e-01 -4.27536845e-01
2.28609830e-01 4.43986952e-01 -8.33501518e-01 4.45554763e-01
-3.53774250e-01 1.14042199e+00 -8.03804874e-01 -4.47227240e-01
-1.48508787e-01 4.25375044e-01 -3.33603054e-01 5.13284989e-02
-1.62736639e-01 -2.69077588e-02 -7.60693327e-02 3.21405083e-01
6.31371260e-01 -5.49848914e-01 1.93993732e-01 2.18861952e-01
-5.32620624e-02 4.03462112e-01 -1.14340913e+00 1.78282201e+00
-3.14517796e-01 6.63629413e-01 -2.31876075e-01 -8.16948712e-01
6.47376597e-01 1.68749735e-01 -1.42134592e-01 -9.92245853e-01
-1.76373437e-01 -1.01537175e-01 -3.43738019e-01 -3.70481871e-02
7.98518836e-01 -1.40120938e-01 -1.83610156e-01 9.50913608e-01
6.30815566e-01 3.34000915e-01 -7.72598237e-02 4.33338046e-01
1.10662436e+00 9.84473825e-02 1.67190731e-01 -5.70860624e-01
-1.45058826e-01 -1.65808946e-01 1.56529084e-01 1.52684939e+00
-7.28495838e-03 4.38063145e-01 1.81870759e-01 -2.35055089e-01
-9.01775599e-01 -1.79961991e+00 -3.62676442e-01 1.40621269e+00
-6.67679310e-02 -4.43132788e-01 -6.34150565e-01 -9.66110677e-02
-1.50960073e-01 9.43089187e-01 -6.07799649e-01 -4.89890218e-01
-3.46813738e-01 -1.41028798e+00 7.76970387e-01 8.98914635e-01
9.60687041e-01 -1.40352261e+00 -1.26019096e+00 2.02644959e-01
-1.58417657e-01 -4.78485674e-01 3.95475812e-02 7.99926758e-01
-8.43944669e-01 -5.54410517e-01 -6.06993794e-01 -4.23517555e-01
2.68551052e-01 2.03854382e-01 1.29744458e+00 -1.52682558e-01
-4.45197821e-01 6.83261931e-01 -4.49818745e-02 -1.77121937e-01
-2.90736463e-02 2.21484810e-01 1.13095671e-01 -3.89994383e-01
2.54856706e-01 -6.18057847e-01 -5.54135263e-01 -4.85578887e-02
-9.95620489e-01 3.03846210e-01 6.91796660e-01 1.16802752e+00
7.64106393e-01 3.33897769e-01 5.54367661e-01 -8.30511928e-01
3.48444700e-01 -8.92151475e-01 -6.36772156e-01 2.22505271e-01
-8.25530469e-01 3.73187393e-01 -1.54030785e-01 -5.37025988e-01
-1.47710848e+00 -7.83009827e-02 2.73129076e-01 -1.06568955e-01
-8.54387805e-02 6.37852013e-01 6.26371950e-02 1.59123793e-01
6.30402923e-01 7.46390343e-01 -2.36890480e-01 -4.22618628e-01
5.46807051e-01 4.23042953e-01 6.56832635e-01 -7.55692363e-01
2.59307504e-01 6.21852577e-01 -9.72174555e-02 -7.14429200e-01
-4.73337859e-01 -2.28578985e-01 -7.03679025e-01 1.47068590e-01
9.75109041e-01 -1.07781589e+00 -4.44294244e-01 7.31595814e-01
-1.05200338e+00 -8.64994228e-01 -3.08876097e-01 5.88883221e-01
-9.57212567e-01 -7.62575585e-03 -9.77416635e-01 -5.84510446e-01
-1.89509019e-01 -8.84457827e-01 7.58504868e-01 1.98815137e-01
-2.82639146e-01 -8.86397839e-01 1.31264001e-01 -4.08808105e-02
7.60474324e-01 -5.63781559e-01 1.19697154e+00 -4.41187650e-01
-1.13933432e+00 2.18427181e-01 -3.59448522e-01 1.53894424e-02
-6.60460591e-01 -8.21513832e-01 -1.07496381e+00 -2.19798535e-01
2.75087446e-01 -5.47179937e-01 1.53150308e+00 3.76544088e-01
9.94963408e-01 -1.65204585e-01 -5.06536067e-01 6.35378361e-01
1.33623767e+00 2.31523931e-01 1.11684668e+00 5.66321731e-01
2.41851822e-01 3.49621862e-01 2.72047073e-01 4.17103618e-01
4.02356178e-01 4.36965525e-01 -9.89080817e-02 4.26162481e-01
-1.85333595e-01 -1.85818434e-01 2.26313904e-01 7.63806164e-01
2.84604877e-01 -2.10882425e-01 -1.06293571e+00 6.40421391e-01
-1.96593833e+00 -1.07933390e+00 1.35903046e-01 2.11186004e+00
1.08241606e+00 2.71305889e-01 -4.25616115e-01 -3.20477307e-01
4.15061176e-01 2.48208091e-01 -3.26441497e-01 -3.91047671e-02
-4.27029252e-01 5.55698931e-01 7.65626490e-01 6.02815270e-01
-1.07517838e+00 1.07727730e+00 7.24140644e+00 1.24732065e+00
-7.83110023e-01 6.59415245e-01 6.44879639e-01 -1.80570111e-01
3.35254483e-02 4.41821031e-02 -1.05336761e+00 6.04272783e-01
1.60919571e+00 -6.95513412e-02 6.16014123e-01 7.78514802e-01
-4.70669866e-01 -8.68839324e-01 -7.97751129e-01 1.03336287e+00
1.80603024e-02 -1.50490165e+00 4.44059074e-03 -1.19673712e-02
5.26437104e-01 3.97303253e-01 4.51456219e-01 5.29471993e-01
5.25706887e-01 -1.40608704e+00 9.89247978e-01 1.14509428e+00
4.98121530e-01 -7.19271362e-01 3.95896077e-01 4.21287507e-01
-7.78486192e-01 6.70605078e-02 -6.75497532e-01 -4.49307114e-02
2.16062561e-01 5.20398915e-01 -3.16476583e-01 3.65709625e-02
1.02817547e+00 1.03577025e-01 -7.71349430e-01 8.83835375e-01
-1.34199560e-01 7.21349716e-01 -3.16713035e-01 3.39870989e-01
-1.60179019e-03 3.63641858e-01 2.87328094e-01 1.42031026e+00
2.26994246e-01 2.71294266e-01 -1.55139327e-01 1.01970983e+00
2.48786844e-02 -4.29352522e-01 -4.25974429e-01 1.29695609e-01
6.51216567e-01 8.07859957e-01 -1.14466226e+00 -7.19554067e-01
-1.34106383e-01 1.10558653e+00 4.13765907e-01 4.96232033e-01
-9.67699587e-01 -1.92414716e-01 -1.83364619e-02 -1.86296940e-01
7.16643512e-01 -4.84746456e-01 -2.72894770e-01 -9.09988165e-01
-4.85377103e-01 -4.47322071e-01 3.54702979e-01 -1.12190199e+00
-1.00097585e+00 6.37503684e-01 4.94418561e-01 -4.14164454e-01
-2.15230972e-01 -3.64725888e-01 -2.06346482e-01 1.10961866e+00
-1.37478733e+00 -1.08785820e+00 -1.28394127e-01 5.20841062e-01
3.25762421e-01 -4.14052010e-01 1.31076467e+00 2.20019728e-01
-1.49147604e-02 3.56818318e-01 3.78046662e-01 -1.44639224e-01
5.69737673e-01 -1.10005319e+00 3.91251683e-01 6.99798822e-01
3.55454266e-01 9.96405602e-01 5.61853409e-01 -8.81182790e-01
-1.24139011e+00 -1.08454823e+00 7.90839672e-01 -5.97660780e-01
4.71840054e-01 -4.16965187e-01 -1.07154191e+00 1.05019701e+00
3.56885999e-01 -1.74018711e-01 6.15183413e-01 6.96709678e-02
-6.32798433e-01 3.38929921e-01 -7.50307679e-01 3.52970421e-01
9.67853725e-01 -7.68865526e-01 -9.00704443e-01 1.79947421e-01
5.69669724e-01 -2.95536339e-01 -5.17044425e-01 1.65791035e-01
4.38470572e-01 -9.14614975e-01 1.09743154e+00 -4.18117076e-01
8.66374373e-03 -1.71429902e-01 -4.59064066e-01 -1.02730966e+00
-5.29674113e-01 -2.50330061e-01 -5.38185716e-01 9.46759522e-01
1.40763178e-01 -5.73697507e-01 6.32270336e-01 4.34859186e-01
4.78734374e-02 -3.89364392e-01 -1.07187903e+00 -9.51363742e-01
6.70953616e-02 -5.18733084e-01 2.99528509e-01 5.98456442e-01
-1.31039888e-01 3.41788828e-01 -3.18147659e-01 6.45235553e-02
8.25115621e-01 -2.38767155e-02 3.21963191e-01 -9.02432919e-01
-6.11494422e-01 -2.09323570e-01 -1.94064483e-01 -1.26536560e+00
4.12044406e-01 -1.04141521e+00 1.85606122e-01 -1.13761222e+00
7.29915500e-01 -6.53913975e-01 -4.83147502e-01 5.34290850e-01
1.94384024e-01 2.26684958e-01 1.49899393e-01 4.22330707e-01
-1.04759502e+00 5.75899601e-01 2.25728065e-01 -5.52086234e-02
1.06395870e-01 -4.69069898e-01 -4.70765829e-01 5.76549053e-01
7.57048607e-01 -9.02982891e-01 -5.56380093e-01 -5.31187475e-01
3.58300716e-01 1.22541100e-01 8.27381313e-01 -1.17506230e+00
7.02360988e-01 1.12093918e-01 6.06398880e-01 -7.21661031e-01
4.25929546e-01 -3.43883455e-01 4.73595202e-01 4.43113476e-01
-4.32920963e-01 1.52284533e-01 3.58940572e-01 9.22623515e-01
-1.21703327e-01 -3.66969198e-01 7.36199141e-01 -3.61784279e-01
-1.07291615e+00 -6.15529418e-02 -7.73695707e-01 3.44071239e-02
8.58562350e-01 1.62769686e-02 -5.65717816e-01 -2.51455039e-01
-1.07799256e+00 -1.03716590e-01 8.99729952e-02 2.03246236e-01
7.24306524e-01 -1.21992314e+00 -3.07172269e-01 4.58385587e-01
-3.24464768e-01 -2.59875715e-01 6.00476801e-01 7.85608292e-01
-3.18103522e-01 8.22254598e-01 -4.37721312e-01 -5.16642690e-01
-8.27426314e-01 6.42965913e-01 1.86626226e-01 -1.65113598e-01
-5.85723996e-01 9.26698565e-01 3.42942446e-01 -7.36158192e-02
2.97628313e-01 1.54045522e-01 1.43960401e-01 7.51004219e-02
7.71202922e-01 5.28645217e-01 2.36932971e-02 -3.83518845e-01
-2.26654142e-01 -1.21859714e-01 -2.01521069e-01 -6.43063724e-01
1.27644789e+00 -4.14698571e-01 -4.28853780e-01 6.24328136e-01
9.37981606e-01 -3.19944352e-01 -1.18428707e+00 -2.58036613e-01
1.98275059e-01 -1.94685906e-01 2.13599190e-01 -8.02320063e-01
-5.93176961e-01 1.08058000e+00 9.54582751e-01 -2.07773730e-01
8.18804562e-01 1.68290272e-01 6.58552110e-01 4.16405350e-01
7.54715323e-01 -1.26268744e+00 2.22743869e-01 9.60295856e-01
5.28364658e-01 -8.81809294e-01 9.10796449e-02 2.27650285e-01
-3.92693251e-01 7.57019341e-01 3.92595798e-01 -1.30680293e-01
1.07941353e+00 4.10434842e-01 -5.25611043e-01 -2.79793680e-01
-1.01281881e+00 8.20042044e-02 -6.04194868e-03 5.20456374e-01
1.32801324e-01 2.23421186e-01 -6.42253086e-02 9.88363624e-01
-1.40054077e-01 7.30183423e-02 3.19726497e-01 1.29292369e+00
-8.09407651e-01 -7.51408994e-01 -4.07285035e-01 5.02526343e-01
-3.51501942e-01 -5.97829044e-01 4.61754143e-01 6.40658915e-01
-9.73009691e-02 3.93675268e-01 3.79296750e-01 1.98654905e-02
-3.38629156e-01 5.96506178e-01 7.07150996e-01 -6.52256012e-01
-3.30253094e-02 -1.02404226e-02 -1.52197495e-01 -4.92931068e-01
-2.27766737e-01 -4.98511672e-01 -1.41345143e+00 -4.01466668e-01
-1.22882567e-01 1.49619892e-01 5.67375600e-01 9.10120368e-01
5.48034370e-01 5.78743100e-01 -3.33168715e-01 -8.96077216e-01
-5.86948931e-01 -9.16256130e-01 -7.90155292e-01 -1.25997979e-03
-2.85574477e-02 -9.89564896e-01 -1.14197880e-01 3.55638176e-01] | [7.460334300994873, 3.774923086166382] |
14817d71-743d-4e3f-9d34-0c2a6838b025 | concentric-spherical-gnn-for-3d-1 | 2103.10484 | null | https://arxiv.org/abs/2103.10484v1 | https://arxiv.org/pdf/2103.10484v1.pdf | Concentric Spherical GNN for 3D Representation Learning | Learning 3D representations that generalize well to arbitrarily oriented inputs is a challenge of practical importance in applications varying from computer vision to physics and chemistry. We propose a novel multi-resolution convolutional architecture for learning over concentric spherical feature maps, of which the single sphere representation is a special case. Our hierarchical architecture is based on alternatively learning to incorporate both intra-sphere and inter-sphere information. We show the applicability of our method for two different types of 3D inputs, mesh objects, which can be regularly sampled, and point clouds, which are irregularly distributed. We also propose an efficient mapping of point clouds to concentric spherical images, thereby bridging spherical convolutions on grids with general point clouds. We demonstrate the effectiveness of our approach in improving state-of-the-art performance on 3D classification tasks with rotated data. | ['Le Song', 'Rampi Ramprasad', 'Sivasankaran Rajamanickam', 'Bo Zhao', 'James Fox'] | 2021-03-18 | concentric-spherical-gnn-for-3d | https://openreview.net/forum?id=OItp-Avs6Iy | https://openreview.net/pdf?id=OItp-Avs6Iy | null | ['3d-classification'] | ['computer-vision'] | [-9.33822766e-02 4.63143773e-02 4.19239253e-01 -2.96690524e-01
-6.60269439e-01 -7.84395695e-01 8.73020589e-01 1.81995943e-01
-2.84664094e-01 3.44186813e-01 -1.14313938e-01 -3.81458402e-01
-6.16621822e-02 -1.01340020e+00 -1.26044095e+00 -5.50592303e-01
-2.91100949e-01 7.71566153e-01 3.17709833e-01 -1.58480540e-01
1.12250656e-01 1.47652316e+00 -1.63363457e+00 2.89205074e-01
6.78449512e-01 9.48062778e-01 1.69598967e-01 6.70712411e-01
-3.33572060e-01 8.41395929e-03 -3.86958599e-01 -8.93686041e-02
3.34406435e-01 4.72349405e-01 -8.45660210e-01 1.02764405e-01
6.65978730e-01 1.36185065e-01 -1.55366853e-01 7.41973996e-01
3.97839487e-01 1.28083616e-01 1.15936327e+00 -7.98500001e-01
-9.64511871e-01 -8.08392465e-02 -4.24603462e-01 1.33371666e-01
1.66023642e-01 -3.38726103e-01 7.75314450e-01 -1.41963303e+00
4.91590291e-01 1.50490117e+00 7.90684342e-01 4.16495770e-01
-1.19009340e+00 -4.54064757e-01 3.42707455e-01 -1.68233469e-01
-1.33770180e+00 -7.68820941e-02 6.11021042e-01 -7.57616103e-01
1.27839673e+00 2.09938794e-01 4.65538293e-01 7.83405542e-01
1.87757820e-01 4.16452587e-01 1.07682538e+00 -2.08136633e-01
3.61765414e-01 1.59103330e-02 -1.00690044e-01 4.15096551e-01
3.84150565e-01 -2.27968469e-02 -1.71026617e-01 -1.92605525e-01
1.33357966e+00 2.15047091e-01 -2.58329082e-02 -8.64129901e-01
-1.09862101e+00 8.28086138e-01 8.37949097e-01 1.63658679e-01
-2.34710723e-01 -7.30684251e-02 4.97634672e-02 6.53384477e-02
8.48783076e-01 5.91210663e-01 -6.36183500e-01 3.09906900e-01
-5.17494261e-01 4.57057655e-01 7.41274118e-01 1.25193000e+00
6.97719574e-01 -1.19033732e-01 2.25296065e-01 7.62567699e-01
1.39255032e-01 6.05552852e-01 5.44986613e-02 -8.17264080e-01
1.68771982e-01 7.27051735e-01 4.34453040e-01 -6.35546029e-01
-7.41372764e-01 -5.84980786e-01 -9.51761723e-01 5.85393012e-01
3.21259260e-01 2.19794467e-01 -1.03598487e+00 1.17290080e+00
5.09466946e-01 2.91045129e-01 3.79306749e-02 8.67999256e-01
1.29161656e+00 4.58456337e-01 -2.70639628e-01 3.50556314e-01
1.28651297e+00 -5.23391187e-01 1.30819485e-01 1.77107751e-01
4.22561646e-01 -6.13908589e-01 7.75401115e-01 1.21606767e-01
-1.15125644e+00 -5.83317459e-01 -1.03184354e+00 -3.42765063e-01
-7.14240968e-01 1.34115174e-01 6.74594462e-01 3.35245788e-01
-1.22176111e+00 9.63667452e-01 -9.21170115e-01 -3.88564706e-01
7.15310454e-01 5.29903591e-01 -4.76059318e-01 7.89079815e-02
-7.80367196e-01 8.71295929e-01 -1.15703441e-01 -3.11415613e-01
-7.06669450e-01 -9.87108648e-01 -8.35499525e-01 1.33025900e-01
-2.09094837e-01 -8.95856619e-01 1.23241127e+00 -3.28566074e-01
-1.27785277e+00 1.13513052e+00 -3.20528537e-01 -2.41805106e-01
4.08948481e-01 -5.36969863e-02 1.59722064e-02 1.08240440e-01
3.86674404e-02 4.47301775e-01 7.09695399e-01 -1.32027626e+00
-2.69747764e-01 -6.56327844e-01 3.16198051e-01 5.16065300e-01
-3.62718180e-02 -1.16234265e-01 -1.30808204e-01 -3.41334313e-01
3.80333096e-01 -9.77383673e-01 -4.11348641e-01 2.29788095e-01
-1.05950452e-01 -5.22637486e-01 7.34363735e-01 6.74672425e-02
2.49647468e-01 -1.86158323e+00 3.46976995e-01 2.46579304e-01
4.00071383e-01 -6.74918219e-02 -3.43507715e-03 1.47626042e-01
-3.60185683e-01 2.00873375e-01 -1.14799604e-01 -3.05871755e-01
1.79257747e-02 1.19595796e-01 -3.41904730e-01 6.06857657e-01
6.38987184e-01 8.92585516e-01 -7.49215662e-01 3.58128510e-02
2.60711044e-01 8.36790621e-01 -3.46044451e-01 9.07615796e-02
-2.18468875e-01 4.92262810e-01 -6.36771619e-01 4.65457708e-01
9.06303108e-01 -8.31708491e-01 -4.66518193e-01 -3.24648581e-02
-3.59040111e-01 4.64083403e-01 -1.10743690e+00 1.75168383e+00
-4.79333639e-01 1.92438975e-01 2.46875972e-01 -9.14351881e-01
1.13928664e+00 2.55368143e-01 6.30094647e-01 -1.34785444e-01
-1.56056210e-01 1.84307307e-01 -1.70105368e-01 7.00380728e-02
3.14191043e-01 -2.68394470e-01 1.28781825e-01 3.51157993e-01
3.92146371e-02 -7.24188387e-01 -2.77998626e-01 2.19597891e-02
9.87544835e-01 2.21743360e-01 4.40619886e-02 -6.06188536e-01
5.04295051e-01 -1.41840726e-01 7.14501068e-02 5.65306425e-01
3.01561445e-01 9.79878545e-01 3.92569304e-01 -8.08132827e-01
-1.38162410e+00 -1.26605701e+00 -7.15717554e-01 8.66758585e-01
1.19343720e-01 -8.03777426e-02 -4.92653996e-01 -4.78468865e-01
6.28398657e-01 1.50881976e-01 -7.53299713e-01 1.77310109e-01
-3.28959316e-01 -6.95258141e-01 4.75551262e-02 7.95936346e-01
9.75515507e-03 -8.15792739e-01 -3.75778794e-01 4.77266647e-02
6.30346417e-01 -1.14964557e+00 -2.89457012e-02 3.89767349e-01
-9.86711621e-01 -1.17714334e+00 -8.67957115e-01 -8.40260923e-01
7.33387411e-01 6.78795636e-01 1.45609856e+00 -2.39934742e-01
-2.70648062e-01 5.79795301e-01 -1.44828483e-01 -8.00878286e-01
-1.65490478e-01 2.80064195e-01 3.15984428e-01 -2.41501138e-01
4.05061930e-01 -7.60071099e-01 -5.53214848e-01 2.79290617e-01
-6.71862602e-01 1.80762205e-02 3.64427030e-01 4.29450750e-01
8.37130487e-01 -4.49303001e-01 4.19220239e-01 -8.29417646e-01
3.55016947e-01 -6.42788351e-01 -7.15838075e-01 -1.43369988e-01
1.10874595e-02 7.14592263e-02 6.66332006e-01 -2.09176525e-01
-4.76835757e-01 2.81910688e-01 -3.29745412e-01 -8.61875772e-01
-5.48991740e-01 5.67287691e-02 1.04622222e-01 -6.09214842e-01
8.49859178e-01 1.50978506e-01 2.86720097e-02 -6.16865873e-01
3.59246612e-01 6.27580225e-01 2.75641590e-01 -1.07035887e+00
7.86356986e-01 8.86632025e-01 3.91242623e-01 -1.16925776e+00
-6.88567162e-01 -4.73224699e-01 -1.02377129e+00 1.60273731e-01
8.57552886e-01 -1.16215205e+00 -7.34798610e-01 3.23716730e-01
-1.43897057e+00 -1.82361245e-01 -4.46197271e-01 3.56147021e-01
-8.55531812e-01 1.04427651e-01 -6.32389784e-01 -4.97517616e-01
-3.02894324e-01 -1.26735771e+00 1.60245097e+00 5.17790690e-02
1.77827209e-01 -1.04444492e+00 1.78515119e-03 -2.71552205e-01
4.40717280e-01 3.40490997e-01 8.96663845e-01 -7.20534563e-01
-7.58876145e-01 -6.24785684e-02 -4.46357161e-01 4.07822505e-02
1.45457178e-01 -9.48895514e-02 -1.13261068e+00 -4.77436513e-01
-2.31867105e-01 -4.76076335e-01 8.53892922e-01 5.38293004e-01
1.62461793e+00 2.45714962e-01 -5.56728184e-01 9.69847918e-01
1.15255201e+00 -1.89017937e-01 3.58345985e-01 1.69758752e-01
6.88172519e-01 3.64005923e-01 2.08002195e-01 3.97081643e-01
4.28003848e-01 4.67060983e-01 6.77108943e-01 -3.57793629e-01
8.53531957e-02 1.21448286e-01 -1.59538731e-01 6.43032730e-01
-6.06318474e-01 2.59305984e-01 -8.78952384e-01 4.36982214e-01
-1.47643578e+00 -6.21544778e-01 -2.43583526e-02 2.22289228e+00
5.33326983e-01 1.20762289e-01 9.13587734e-02 -1.77713796e-01
5.97834229e-01 4.43225801e-02 -9.28585172e-01 -3.62328172e-01
-8.45121220e-02 7.85425425e-01 4.77577060e-01 3.81451428e-01
-1.58381975e+00 8.03598464e-01 6.94327641e+00 4.66395527e-01
-1.15001678e+00 -1.39055043e-01 4.51322556e-01 -5.88635877e-02
-4.39239502e-01 -5.08597255e-01 -1.02558720e+00 2.39594102e-01
6.80751562e-01 -1.06608503e-01 2.62822449e-01 9.60881650e-01
-1.49892882e-01 6.11990571e-01 -1.33791983e+00 9.22173679e-01
-1.72357738e-01 -1.71318400e+00 2.63954580e-01 6.74506873e-02
9.00825620e-01 6.39378190e-01 2.43895069e-01 1.04184330e-01
7.59439647e-01 -1.42393839e+00 4.49971557e-01 4.52375472e-01
1.23285520e+00 -7.07219601e-01 2.19989076e-01 2.63156056e-01
-1.42688143e+00 2.93539941e-01 -8.89498532e-01 -1.63318031e-02
-3.77476096e-01 4.48302954e-01 -8.05183351e-01 5.44120073e-01
9.70329523e-01 1.03993690e+00 -2.89533645e-01 1.07077026e+00
5.51216640e-02 -7.65109211e-02 -6.23649478e-01 -1.25098273e-01
3.12313437e-01 -3.76026422e-01 5.88727176e-01 1.22195041e+00
5.42176962e-01 3.30854863e-01 6.34079576e-02 1.15919781e+00
-1.96875021e-01 -1.15120122e-02 -1.20080864e+00 3.90746802e-01
5.66128790e-01 1.29691911e+00 -6.76170290e-01 -1.85695410e-01
-8.18116069e-01 6.01489723e-01 7.05954552e-01 5.32459676e-01
-3.78222048e-01 -3.69969934e-01 1.20113170e+00 3.11931312e-01
8.15364301e-01 -6.03093028e-01 -2.98300445e-01 -1.09502745e+00
-4.99859191e-02 -2.85668701e-01 4.43708487e-02 -9.38547790e-01
-1.50609744e+00 6.09452963e-01 -1.54454067e-01 -1.23239195e+00
1.30358011e-01 -1.21461427e+00 -4.98958647e-01 1.25052416e+00
-1.83927059e+00 -1.26574337e+00 -3.61033797e-01 5.95049381e-01
2.73826957e-01 -1.88918427e-01 1.07911301e+00 -1.47600800e-01
2.31794834e-01 2.29464635e-01 3.90789330e-01 2.93207280e-02
3.46648306e-01 -1.63632834e+00 9.99548554e-01 1.30192086e-01
1.16811506e-01 5.80103278e-01 2.75622606e-01 -2.92199701e-01
-1.55882776e+00 -1.42363155e+00 4.08432424e-01 -7.84129739e-01
5.14788985e-01 -6.78359151e-01 -1.17057824e+00 6.81286752e-01
-3.60212773e-01 4.19379026e-01 6.41149580e-01 2.78225929e-01
-7.52110779e-01 1.50158837e-01 -1.14791775e+00 3.49132746e-01
1.19953966e+00 -6.33026302e-01 -5.34710407e-01 7.50705719e-01
7.69347429e-01 -9.86753345e-01 -1.32189178e+00 5.90765595e-01
2.98824519e-01 -7.62268186e-01 1.41302943e+00 -1.03599870e+00
4.71215636e-01 -3.79191101e-01 -1.13908783e-01 -1.55109107e+00
-6.81824923e-01 -1.65203214e-01 -5.75866066e-02 1.97182581e-01
3.25209081e-01 -6.58922434e-01 9.35677290e-01 2.47033924e-01
-5.54826319e-01 -9.63643789e-01 -1.14706326e+00 -6.47557557e-01
9.05370355e-01 -2.34762967e-01 8.10147822e-01 9.21080828e-01
-2.15888977e-01 3.17506254e-01 4.77864295e-01 5.72783530e-01
5.56675375e-01 5.82657516e-01 6.88102126e-01 -1.85630906e+00
6.95448592e-02 -6.11421168e-01 -6.66157007e-01 -1.44659328e+00
2.53678560e-01 -1.29107893e+00 -2.84230649e-01 -1.53367555e+00
3.27373855e-02 -7.76751697e-01 -1.17685355e-01 1.95202425e-01
1.31947279e-01 2.06087545e-01 -7.04898983e-02 1.29496738e-01
-5.87808788e-01 6.16577268e-01 1.63106620e+00 -7.42145777e-02
-2.98342537e-02 1.17934003e-01 -6.02216780e-01 7.23913848e-01
6.10904634e-01 -8.01837351e-03 -9.77092832e-02 -6.56097591e-01
1.44420624e-01 -1.96572274e-01 4.92429137e-01 -9.68469679e-01
1.60445109e-01 5.47976643e-02 8.40103447e-01 -7.47909665e-01
6.46540403e-01 -8.23130667e-01 -2.00636730e-01 -1.31209627e-01
-8.33002254e-02 1.82697941e-02 4.35671896e-01 6.00496292e-01
1.41735867e-01 1.70154855e-01 1.02192461e+00 -5.21308661e-01
-3.40453945e-02 7.92886913e-01 1.30420759e-01 -1.07225299e-01
9.53738689e-01 -6.44685254e-02 -3.15712631e-01 -8.87090862e-02
-1.05067587e+00 2.02207714e-01 8.41934502e-01 3.62638563e-01
7.97182858e-01 -1.44997633e+00 -8.63867939e-01 5.29641986e-01
9.98977795e-02 8.97104740e-01 -1.02106638e-01 3.68715942e-01
-6.01348996e-01 7.12915182e-01 -5.64501435e-02 -1.06262910e+00
-9.02001143e-01 4.58380133e-01 6.39258027e-01 1.19703643e-01
-7.07228661e-01 9.99423325e-01 5.65472960e-01 -8.08227241e-01
1.82135805e-01 -6.87810183e-01 -3.00891042e-01 -4.22183782e-01
3.77037257e-01 6.71184063e-02 5.58093250e-01 -5.65834761e-01
-2.96965778e-01 9.81249332e-01 -1.86793074e-01 5.48121333e-01
1.63882017e+00 3.32426459e-01 -1.21043719e-01 6.06456578e-01
1.04456031e+00 -3.34890157e-01 -1.52210891e+00 -4.24514949e-01
-3.10730338e-01 -2.72038549e-01 -3.03773373e-01 -3.68737131e-01
-7.48054802e-01 1.18626189e+00 3.11262190e-01 3.92921299e-01
4.24077630e-01 6.11160278e-01 2.96509355e-01 5.70178807e-01
4.11341161e-01 -5.23689568e-01 -8.88180956e-02 9.86599445e-01
1.25405836e+00 -1.17074692e+00 9.85759422e-02 -4.74167734e-01
-2.22046658e-01 1.36564255e+00 4.79086161e-01 -7.78079152e-01
9.82953966e-01 3.78012538e-01 -2.50282556e-01 -4.48496044e-01
-7.35875070e-01 -3.30632657e-01 4.81008768e-01 7.17314124e-01
5.13452232e-01 1.84608027e-01 4.16713357e-01 5.46529770e-01
-3.23478222e-01 -3.44253421e-01 3.25900495e-01 7.28533030e-01
-6.11336887e-01 -7.48660386e-01 -4.58973378e-01 6.89659357e-01
-1.88933313e-01 5.97620895e-03 -4.70443726e-01 7.51490593e-01
-4.16472331e-02 2.19244733e-01 6.67388260e-01 -1.49031222e-01
5.76428056e-01 5.25279418e-02 7.21710742e-01 -8.85858357e-01
-1.50419384e-01 -5.90769164e-02 -3.36072832e-01 -4.22354817e-01
-4.43943560e-01 -8.85388374e-01 -1.36524403e+00 -1.04078718e-01
2.64379326e-02 1.34910330e-01 9.07660663e-01 7.10820556e-01
7.49579787e-01 4.59008247e-01 6.96522176e-01 -1.65014756e+00
-6.84819818e-01 -9.72700179e-01 -7.40786850e-01 3.75144392e-01
6.14452362e-01 -1.04844725e+00 -2.95922726e-01 -2.47179806e-01] | [7.972278118133545, -3.6798954010009766] |
e2bd7222-ef8e-44da-86d2-519d512bf5b8 | peekaboo-text-to-image-diffusion-models-are | 2211.13224 | null | https://arxiv.org/abs/2211.13224v2 | https://arxiv.org/pdf/2211.13224v2.pdf | Peekaboo: Text to Image Diffusion Models are Zero-Shot Segmentors | Recently, text-to-image diffusion models have shown remarkable capabilities in creating realistic images from natural language prompts. However, few works have explored using these models for semantic localization or grounding. In this work, we explore how an off-the-shelf text-to-image diffusion model, trained without exposure to localization information, can ground various semantic phrases without segmentation-specific re-training. We introduce an inference time optimization process capable of generating segmentation masks conditioned on natural language prompts. Our proposal, Peekaboo, is a first-of-its-kind zero-shot, open-vocabulary, unsupervised semantic grounding technique leveraging diffusion models without any training. We evaluate Peekaboo on the Pascal VOC dataset for unsupervised semantic segmentation and the RefCOCO dataset for referring segmentation, showing results competitive with promising results. We also demonstrate how Peekaboo can be used to generate images with transparency, even though the underlying diffusion model was only trained on RGB images - which to our knowledge we are the first to attempt. Please see our project page, including our code: https://ryanndagreat.github.io/peekaboo | ['Michael S. Ryoo', 'Xiang Li', 'Kanchana Ranasinghe', 'Ryan Burgert'] | 2022-11-23 | null | null | null | null | ['unsupervised-semantic-segmentation'] | ['computer-vision'] | [ 4.13515389e-01 6.53108954e-01 1.10201567e-01 -4.01366472e-01
-8.94288421e-01 -6.75260305e-01 7.68107355e-01 -1.33900911e-01
-3.52581680e-01 4.25072581e-01 -5.44914231e-02 -2.84362465e-01
2.02471852e-01 -8.62934232e-01 -9.22484398e-01 -4.71867323e-01
4.73894238e-01 8.34375143e-01 5.91248035e-01 -2.42546394e-01
3.64918321e-01 5.49536534e-02 -1.29573536e+00 4.86686289e-01
8.09183300e-01 8.70394588e-01 6.41741097e-01 8.71673763e-01
-4.64900762e-01 1.02321994e+00 -5.67531765e-01 -3.88219535e-01
2.13261321e-01 -5.76617479e-01 -1.12221539e+00 1.94773048e-01
6.66987896e-01 -2.64963686e-01 -1.45043612e-01 1.05878091e+00
5.23932695e-01 8.61328170e-02 6.14512146e-01 -1.27386856e+00
-1.04202902e+00 8.52411866e-01 -1.75464272e-01 -9.55866724e-02
4.95787203e-01 4.09487665e-01 8.77494276e-01 -8.46932471e-01
1.13294733e+00 1.06633735e+00 4.61382449e-01 9.65049624e-01
-1.27780998e+00 -4.89937693e-01 -8.65354761e-03 8.47039223e-02
-1.32546711e+00 -3.43370557e-01 7.46783793e-01 -5.47122180e-01
1.10753942e+00 -8.50516558e-02 5.61614215e-01 1.62742794e+00
-3.14290881e-01 1.00975597e+00 1.60465550e+00 -5.30184031e-01
3.77229959e-01 2.59506553e-01 -2.80382279e-02 1.08040118e+00
-1.42056376e-01 -1.10529950e-02 -8.28214824e-01 2.49861926e-01
7.90971279e-01 -5.01787245e-01 -1.60149187e-01 -2.66170174e-01
-1.36252701e+00 7.64386356e-01 5.45437217e-01 3.19102734e-01
-2.81890124e-01 6.28133833e-01 1.33423992e-02 2.90135015e-02
5.54829061e-01 4.25278842e-01 -2.47597054e-01 -1.99505374e-01
-1.21204245e+00 2.50041395e-01 8.75669658e-01 1.38057709e+00
9.35055435e-01 -8.79408717e-02 -3.64011586e-01 6.01089895e-01
1.44769341e-01 5.86873233e-01 5.22821724e-01 -1.16163349e+00
4.09621745e-01 2.31570408e-01 6.16931506e-02 -6.80024683e-01
-3.03875744e-01 1.15593933e-01 -3.94185096e-01 7.18477070e-02
4.96421486e-01 -2.23542541e-01 -1.23804963e+00 1.23690879e+00
1.20619394e-01 2.54575431e-01 2.90615886e-01 9.23295975e-01
9.84517694e-01 6.56458855e-01 9.59282890e-02 3.38397622e-01
1.15496027e+00 -1.43908608e+00 -5.61556339e-01 -2.82992333e-01
7.37561524e-01 -7.47817576e-01 1.43813396e+00 5.71514130e-01
-8.68367553e-01 -4.56919402e-01 -7.90975928e-01 -3.66704017e-01
-6.33096933e-01 5.82877249e-02 8.30511510e-01 7.00532377e-01
-1.44783628e+00 6.15430832e-01 -7.82273889e-01 -8.33664894e-01
5.51988721e-01 -3.07122394e-02 -1.16150670e-01 -7.74455294e-02
-8.75796616e-01 6.37803733e-01 5.40569663e-01 -2.49049380e-01
-1.22550249e+00 -4.85848695e-01 -6.12593889e-01 -4.06852186e-01
8.24749708e-01 -7.62297094e-01 1.43329692e+00 -1.06920338e+00
-1.86684763e+00 1.06699991e+00 -1.14483081e-01 -7.08823979e-01
7.29377627e-01 -2.94316411e-01 -8.66679549e-02 5.13538897e-01
4.21546459e-01 1.62443709e+00 9.83645439e-01 -1.39532983e+00
-2.24444255e-01 -6.88908016e-03 2.09562108e-01 7.84458220e-02
-6.15710430e-02 -1.42572626e-01 -6.63811982e-01 -6.45528615e-01
-7.24175200e-02 -9.89269137e-01 -3.77720267e-01 8.47925246e-02
-7.94246614e-01 -2.49911308e-01 7.48709321e-01 -5.45269251e-01
6.04758084e-01 -1.86253464e+00 1.61920920e-01 3.21974047e-03
3.55167165e-02 2.25446336e-02 -2.46042073e-01 3.96124542e-01
4.17348832e-01 3.46768230e-01 -5.43054760e-01 -5.72542012e-01
1.76948011e-01 2.71412730e-01 -5.24897099e-01 1.70369864e-01
2.26549968e-01 1.39888358e+00 -1.15732706e+00 -7.31268466e-01
4.83254820e-01 5.04624605e-01 -4.50430334e-01 1.15545906e-01
-9.42724526e-01 7.36590266e-01 -4.16106939e-01 6.71926737e-01
3.35500628e-01 -3.66467893e-01 -2.76467085e-01 6.20830199e-03
-1.80871990e-02 -4.00593989e-02 -8.60834241e-01 2.56732345e+00
-4.42839712e-01 7.41301000e-01 -1.85152218e-01 -7.08759427e-01
8.18091035e-01 8.41808841e-02 3.22358251e-01 -8.20587933e-01
8.58769342e-02 4.75131035e-01 -6.42750204e-01 -5.01589298e-01
7.15204060e-01 -5.42571619e-02 -1.22252680e-01 4.24761176e-01
4.17657048e-01 -1.01566064e+00 4.00817424e-01 6.04570687e-01
8.94416094e-01 7.19499171e-01 -2.14626551e-01 -1.90020025e-01
1.51655599e-01 4.61447686e-01 -1.53188556e-01 1.13910341e+00
-1.05046265e-01 1.10161221e+00 3.15502465e-01 5.36881424e-02
-9.34352398e-01 -1.03790188e+00 1.51342824e-01 8.86920810e-01
3.33133638e-01 -4.70500201e-01 -1.51374292e+00 -7.74748743e-01
-3.56524944e-01 1.15843320e+00 -4.74299908e-01 4.44057316e-01
1.16338553e-02 -3.37700456e-01 8.42942834e-01 3.44027430e-01
7.24772155e-01 -1.39174020e+00 -6.95855677e-01 1.72706634e-01
-2.45760247e-01 -1.62388027e+00 -3.94908965e-01 7.25353062e-02
-6.63830161e-01 -1.01651776e+00 -1.12453067e+00 -6.16842628e-01
6.62240148e-01 1.39547408e-01 1.10928404e+00 -4.53566946e-02
-3.35239202e-01 1.02846730e+00 -6.04990363e-01 -4.37392324e-01
-4.66819614e-01 1.99266598e-01 -4.74886745e-01 -4.95567732e-02
3.67319763e-01 -2.26962775e-01 -6.41846895e-01 1.49021760e-01
-1.05827820e+00 5.02482712e-01 4.14447993e-01 3.47908586e-01
7.54056633e-01 -2.71996468e-01 2.11104497e-01 -1.08127499e+00
6.02787375e-01 -2.13495985e-01 -6.01748049e-01 7.36493096e-02
-5.24110794e-01 1.15795888e-01 2.19792470e-01 -2.06600085e-01
-1.02625108e+00 2.06469089e-01 -2.46280089e-01 -4.96815801e-01
-6.37009382e-01 1.10896952e-01 1.32819161e-01 -1.67183936e-01
8.15695167e-01 1.99425340e-01 -4.38871413e-01 -1.84404030e-01
1.22711504e+00 6.33331954e-01 5.97693086e-01 -7.95057118e-01
5.98926246e-01 8.87343884e-01 -2.31699511e-01 -9.71749425e-01
-1.01801264e+00 -6.19396329e-01 -7.71470189e-01 -2.91134179e-01
1.52687454e+00 -8.70596409e-01 -2.32983753e-01 7.24789441e-01
-1.46624970e+00 -1.19534290e+00 -6.44568026e-01 7.46354014e-02
-1.04465199e+00 3.54738206e-01 -4.62755442e-01 -5.30838549e-01
-2.50105560e-01 -1.15925300e+00 1.37055635e+00 2.18571886e-01
-2.35831901e-01 -1.04356873e+00 -2.29488775e-01 7.10032105e-01
4.04194534e-01 2.35485256e-01 1.84654146e-01 -4.16126430e-01
-1.02314544e+00 1.80221215e-01 -4.17176992e-01 4.29230988e-01
-2.72201031e-01 -9.91050079e-02 -1.21435440e+00 6.75968677e-02
-2.18611866e-01 -6.64895475e-01 9.27392483e-01 2.19425485e-01
1.26984870e+00 -7.46987313e-02 -2.60000199e-01 6.83950067e-01
1.41374707e+00 -2.04037637e-01 6.98881030e-01 5.09557366e-01
9.76488292e-01 5.87300599e-01 5.70623338e-01 1.40840977e-01
5.86538374e-01 5.50431609e-01 3.93681228e-01 -1.96687296e-01
-7.53276706e-01 -6.07598066e-01 3.16768765e-01 6.76715016e-01
5.70569374e-02 -5.34743309e-01 -1.00061965e+00 7.25385547e-01
-1.62622821e+00 -6.60880268e-01 -4.08111215e-01 1.71155536e+00
7.01829851e-01 9.41504017e-02 -2.25181237e-01 -3.08259010e-01
4.09892768e-01 2.76462823e-01 -4.89971459e-01 -3.98162782e-01
-2.25153446e-01 4.83031660e-01 6.96536005e-01 7.29467809e-01
-9.51077878e-01 1.85876215e+00 5.37164879e+00 9.94378507e-01
-1.13928688e+00 6.89101100e-01 6.36739194e-01 1.08447812e-01
-4.00751233e-01 4.00085449e-01 -7.44637430e-01 1.43410683e-01
7.88250983e-01 2.05836639e-01 6.46143734e-01 7.93677032e-01
1.94000810e-01 -3.62994134e-01 -7.96264887e-01 9.92229939e-01
3.66292685e-01 -1.35295284e+00 3.28707024e-02 -8.19158331e-02
1.03895283e+00 3.95981699e-01 6.00742027e-02 6.19882159e-02
6.46695554e-01 -9.51287866e-01 1.19920516e+00 6.58080518e-01
7.51160324e-01 -1.63932770e-01 2.35682577e-01 2.81001210e-01
-7.31082439e-01 2.99551010e-01 -2.19035327e-01 3.10006917e-01
5.06378114e-01 5.30018628e-01 -9.66872156e-01 4.59160477e-01
6.90587819e-01 7.13387966e-01 -7.67875969e-01 6.49541497e-01
-8.08399439e-01 6.83376610e-01 -3.33912700e-01 -1.44660130e-01
6.21853173e-01 -2.03153536e-01 3.32414597e-01 1.31545770e+00
5.07128239e-01 9.29651707e-02 2.46285617e-01 1.30009019e+00
-1.22491732e-01 2.97756165e-01 -5.80433011e-01 -3.27739447e-01
3.59562524e-02 1.18287766e+00 -1.28258431e+00 -6.44623935e-01
-2.50582427e-01 1.78643298e+00 2.17894867e-01 5.73786497e-01
-9.96426165e-01 -3.80429000e-01 -1.48685694e-01 1.98417604e-01
4.05095160e-01 -4.32305157e-01 -2.75661200e-01 -1.15223467e+00
-2.21649066e-01 -6.34265900e-01 -7.38048106e-02 -1.46636808e+00
-1.17885077e+00 6.56688631e-01 1.31162658e-01 -5.81611991e-01
2.52272729e-02 -6.65975451e-01 -3.32446635e-01 7.49519646e-01
-1.61024606e+00 -1.43778789e+00 -6.12362325e-01 8.03281665e-01
8.27929020e-01 4.10594232e-02 6.26958489e-01 -1.52131483e-01
-1.87113866e-01 1.29319020e-02 -2.12562650e-01 -7.06570894e-02
5.98280907e-01 -1.42631125e+00 6.34089947e-01 6.78645611e-01
6.47529662e-01 2.46392801e-01 8.05359900e-01 -8.23185146e-01
-9.69930589e-01 -1.12703812e+00 6.59214735e-01 -6.94869816e-01
8.46370041e-01 -6.13634288e-01 -6.97748780e-01 6.80339158e-01
6.45209134e-01 -1.14833392e-01 3.07569176e-01 -3.93629164e-01
-2.45808542e-01 3.35051805e-01 -9.42673624e-01 7.76925981e-01
1.29632151e+00 -7.56503046e-01 -4.26651537e-01 8.86796534e-01
1.08558285e+00 -6.81562185e-01 -6.71748757e-01 5.46130501e-02
7.86973760e-02 -1.12529159e+00 7.99565196e-01 1.15381703e-01
4.94432360e-01 -2.42581218e-01 -1.49053022e-01 -1.11942089e+00
3.62330765e-01 -8.89936209e-01 2.32308224e-01 1.32553315e+00
5.05100369e-01 -5.43812394e-01 1.00816989e+00 5.10306001e-01
-1.76800042e-01 -3.04117501e-01 -7.39890099e-01 -7.66843081e-01
8.11693743e-02 -8.57683241e-01 3.12140018e-01 7.69400418e-01
-4.27487344e-01 2.78028816e-01 -1.71954930e-01 8.40964615e-02
6.68559492e-01 -2.20104873e-01 1.04898596e+00 -7.15923667e-01
-4.84030843e-01 -4.10383254e-01 -1.20325886e-01 -1.32219303e+00
2.35239431e-01 -1.06536841e+00 2.56274015e-01 -1.99290264e+00
-1.59332916e-01 -5.71757436e-01 1.50203913e-01 5.82423389e-01
1.55420721e-01 3.44200343e-01 4.76821035e-01 2.28031456e-01
-7.92103231e-01 4.70141351e-01 1.43657041e+00 -3.94943088e-01
-1.75251231e-01 -5.49128771e-01 -5.46899796e-01 6.63118958e-01
1.05008447e+00 -7.00268984e-01 -6.82384312e-01 -6.06440544e-01
6.17979653e-02 -2.68239707e-01 7.09937751e-01 -1.28337729e+00
3.56390119e-01 -9.02773663e-02 -2.35305473e-01 -4.26497310e-01
4.31015283e-01 -4.88263845e-01 8.07199180e-02 1.69698998e-01
-3.55127096e-01 -2.89483070e-01 2.28578642e-01 6.17941976e-01
-5.29811829e-02 -6.53745711e-01 3.58145148e-01 -6.22338057e-01
-1.26796341e+00 1.54538110e-01 -4.18795019e-01 2.05250099e-01
1.15158129e+00 -9.01702568e-02 -4.68069941e-01 -4.15501177e-01
-8.24372470e-01 -1.00998044e-01 7.62924194e-01 5.64504981e-01
6.53220832e-01 -9.09537375e-01 -4.77854878e-01 -7.83486664e-02
2.31452346e-01 2.99213350e-01 2.28658259e-01 6.56271219e-01
-8.55027914e-01 5.86364388e-01 1.40030757e-01 -7.81138897e-01
-8.78713071e-01 7.02107072e-01 2.58215576e-01 -1.08414777e-01
-7.84665823e-01 1.15516675e+00 1.53279260e-01 -5.62863171e-01
4.86512519e-02 -5.53015411e-01 2.68434733e-01 1.45386169e-02
8.85006040e-02 4.94708493e-02 -7.07358643e-02 -6.47119999e-01
-7.17028901e-02 7.31956720e-01 1.22645713e-01 -8.21449578e-01
8.94852877e-01 -2.11881623e-01 -4.92818542e-02 5.13599157e-01
8.83972466e-01 -7.89741054e-02 -1.40308535e+00 2.73321625e-02
4.11202684e-02 -3.43385309e-01 1.76165879e-01 -1.00472188e+00
-9.09559488e-01 1.24274731e+00 5.32478869e-01 1.51535302e-01
8.14641476e-01 2.62635499e-01 9.21763361e-01 3.85732174e-01
8.25525105e-01 -1.20547581e+00 6.49020851e-01 3.75641495e-01
9.06146407e-01 -1.28925812e+00 -3.53992045e-01 -5.05910814e-01
-1.01628935e+00 8.69737148e-01 5.04115641e-01 -1.54812619e-01
3.66967082e-01 -7.87156634e-03 3.90494138e-01 -4.16696161e-01
-3.78073573e-01 -6.24407172e-01 6.83364347e-02 1.01981974e+00
1.44072920e-01 1.15141585e-01 9.90586579e-02 1.47035912e-01
-3.00798327e-01 3.18052679e-01 6.57828629e-01 8.16548765e-01
-3.45474243e-01 -1.13787901e+00 -2.59467959e-01 1.47458771e-02
-1.63464054e-01 -5.61966002e-01 -7.13595271e-01 6.33447587e-01
2.25600541e-01 1.08870733e+00 -1.03128612e-01 -1.14474595e-01
1.82813138e-01 7.42989853e-02 6.48405731e-01 -8.48552287e-01
-4.95319992e-01 1.52538400e-02 5.21576256e-02 -7.93365240e-01
-6.69258177e-01 -4.58024263e-01 -1.52509379e+00 6.06388412e-02
-2.06914514e-01 -3.17532457e-02 1.03107882e+00 7.85005271e-01
3.63300294e-01 5.91995418e-01 -1.00354657e-01 -1.17927277e+00
4.48730439e-02 -8.86270761e-01 -3.71118486e-01 4.56337601e-01
-3.00996955e-02 -4.40940380e-01 -8.05029944e-02 3.62001777e-01] | [9.887088775634766, 0.8136979937553406] |
3de832a9-b0d3-406f-b9cd-01f4cba39018 | anomaly-detection-with-variance-stabilized | 2306.00582 | null | https://arxiv.org/abs/2306.00582v1 | https://arxiv.org/pdf/2306.00582v1.pdf | Anomaly Detection with Variance Stabilized Density Estimation | Density estimation based anomaly detection schemes typically model anomalies as examples that reside in low-density regions. We propose a modified density estimation problem and demonstrate its effectiveness for anomaly detection. Specifically, we assume the density function of normal samples is uniform in some compact domain. This assumption implies the density function is more stable (with lower variance) around normal samples than anomalies. We first corroborate this assumption empirically using a wide range of real-world data. Then, we design a variance stabilized density estimation problem for maximizing the likelihood of the observed samples while minimizing the variance of the density around normal samples. We introduce an ensemble of autoregressive models to learn the variance stabilized distribution. Finally, we perform an extensive benchmark with 52 datasets demonstrating that our method leads to state-of-the-art results while alleviating the need for data-specific hyperparameter tuning. | ['Ofir Lindenbaum', 'Lior Wolf', 'Henry Li', 'Barak Battash', 'Amit Rozner'] | 2023-06-01 | null | null | null | null | ['density-estimation'] | ['methodology'] | [-2.85336584e-01 -1.26543358e-01 -2.30061039e-01 -4.06289935e-01
-8.97670746e-01 -1.47887245e-01 7.01384068e-01 9.76454541e-02
-1.69532374e-01 6.60910070e-01 -6.06027730e-02 -3.42507929e-01
-8.57014954e-02 -7.95338333e-01 -6.63374782e-01 -6.48153007e-01
-3.10232073e-01 5.74898005e-01 3.64288121e-01 1.63213819e-01
3.22142929e-01 4.55883265e-01 -1.16491354e+00 -3.97354931e-01
9.79399383e-01 1.06981492e+00 -4.81143355e-01 5.55231631e-01
-8.51170421e-02 5.39799452e-01 -9.04703617e-01 -2.08085671e-01
2.00255841e-01 -2.90764511e-01 -2.93638915e-01 2.95292228e-01
2.81196028e-01 -5.34746289e-01 -1.63988948e-01 1.19854486e+00
2.41942741e-02 3.33357722e-01 1.20619941e+00 -1.62452042e+00
-5.62680602e-01 5.21714836e-02 -9.78005230e-01 5.85225880e-01
3.62285078e-02 -3.49139608e-03 8.82254064e-01 -8.86287510e-01
-5.71077242e-02 1.02751637e+00 5.91138184e-01 4.29793686e-01
-1.43031955e+00 -5.21456242e-01 2.57627189e-01 5.42547554e-02
-1.59766781e+00 -3.90076578e-01 4.76367563e-01 -3.91447008e-01
7.71456003e-01 6.16237223e-02 4.26735401e-01 1.31744325e+00
1.54082671e-01 6.78951025e-01 4.01953280e-01 -1.94610283e-01
7.13041961e-01 5.89863062e-02 1.44443691e-01 3.18597436e-01
7.26967394e-01 -9.23872367e-02 -1.74497277e-01 -7.95068026e-01
6.51508331e-01 1.93097025e-01 -2.24120602e-01 -6.62824571e-01
-3.95055413e-01 1.08544087e+00 -1.93532839e-01 1.11869156e-01
-3.61627430e-01 1.28657654e-01 2.85953492e-01 1.11616209e-01
7.53764987e-01 1.00099497e-01 -1.81972444e-01 -4.20742631e-01
-7.98837245e-01 2.32208788e-01 8.53435576e-01 8.03341806e-01
6.25686824e-01 5.26215076e-01 -8.44137743e-02 9.54827547e-01
5.53015709e-01 7.40541399e-01 4.16853786e-01 -6.31957829e-01
3.32933694e-01 4.56769347e-01 4.00481641e-01 -7.08124995e-01
-1.38691351e-01 -2.28111684e-01 -8.48706603e-01 -2.74925753e-02
7.62003958e-01 -3.07656497e-01 -8.54027450e-01 1.64396119e+00
3.34186822e-01 8.39574337e-01 -8.82503539e-02 5.80293179e-01
-1.96931735e-01 6.99353397e-01 7.38105848e-02 -1.92213669e-01
9.56360936e-01 -3.53339136e-01 -5.83986402e-01 -2.36130789e-01
6.60845697e-01 -2.07040891e-01 1.22286725e+00 3.96687508e-01
-6.84147775e-01 1.90067351e-01 -9.00334597e-01 6.57997012e-01
-1.03359967e-01 -3.31806391e-01 3.21688652e-01 8.17278743e-01
-6.78916693e-01 3.04646283e-01 -1.13215125e+00 -3.46258610e-01
4.22504365e-01 -9.87437963e-02 -2.38841549e-01 2.03280941e-01
-8.62041831e-01 5.45275927e-01 3.28074366e-01 -3.57502997e-01
-8.38210702e-01 -1.02858531e+00 -9.68172848e-01 3.61059010e-01
2.40125015e-01 -3.38345617e-01 1.33678508e+00 -5.45225739e-01
-1.25190330e+00 4.70690548e-01 -3.06379348e-01 -9.65329111e-01
4.70588654e-01 -5.41882575e-01 -7.51991332e-01 8.21789056e-02
-1.36764228e-01 -1.57114476e-01 1.22805500e+00 -9.98176098e-01
-6.72067642e-01 -1.62307888e-01 -7.25640893e-01 -2.60966450e-01
-4.69945192e-01 -1.12208419e-01 -3.01281363e-01 -6.29777491e-01
-3.82965729e-02 -5.94820619e-01 -1.16660289e-01 -2.06193104e-01
-5.01937807e-01 -1.17179267e-01 1.10508418e+00 -2.95526683e-01
1.64495456e+00 -2.37724876e+00 -5.13882399e-01 8.18401694e-01
2.28640422e-01 6.63710013e-02 2.10553706e-01 1.26539618e-01
-6.29981905e-02 -2.60255039e-02 -5.94943583e-01 -4.66219634e-01
1.29568309e-01 2.24599540e-01 -8.35864127e-01 9.60144401e-01
5.00458002e-01 3.56812179e-01 -6.50931835e-01 -1.32846236e-01
1.92262307e-01 2.67210275e-01 -7.51508951e-01 4.56321150e-01
-1.34722978e-01 1.94253311e-01 -3.67041558e-01 6.57999754e-01
8.34429443e-01 -4.64575440e-01 -1.16357982e-01 4.70758796e-01
3.27734113e-01 -5.67114577e-02 -1.27275443e+00 7.82685280e-01
-1.74924240e-01 4.58877355e-01 -1.35977566e-01 -1.30882561e+00
1.17220080e+00 7.46036991e-02 6.05012119e-01 -2.08737940e-01
1.89876799e-02 1.73261985e-01 -8.51805732e-02 -8.50360841e-02
3.74264270e-01 -1.50370613e-01 -5.99067211e-02 5.18652141e-01
-3.14495340e-02 1.22067453e-02 7.83300772e-02 1.03475682e-01
1.20239258e+00 -4.95812654e-01 5.48837543e-01 -1.95024028e-01
3.64287764e-01 -4.83258039e-01 4.01680827e-01 1.02213347e+00
-2.81131357e-01 5.47301471e-01 1.14306831e+00 -1.68676138e-01
-1.15199018e+00 -1.58591926e+00 -5.36527574e-01 8.64914715e-01
-1.08713403e-01 -3.33129078e-01 -6.76668823e-01 -7.46791542e-01
2.36507997e-01 1.14405990e+00 -7.26424098e-01 -4.92930770e-01
-2.89308548e-01 -1.21078670e+00 5.34553051e-01 4.72579420e-01
3.11877459e-01 -5.78308165e-01 -4.11202013e-01 -5.95762804e-02
2.00010866e-01 -9.90681767e-01 -3.56993526e-01 3.45543772e-02
-8.54079962e-01 -1.03695869e+00 -5.88235557e-01 -4.12864834e-02
6.67468488e-01 -1.96556181e-01 1.19561303e+00 -1.84473380e-01
-7.26976693e-02 5.12696981e-01 -1.89638466e-01 -4.83236760e-01
-2.63195306e-01 -1.31891951e-01 3.35078329e-01 2.22109258e-01
7.92791545e-01 -6.81143343e-01 -3.46683383e-01 3.88138443e-01
-8.74003112e-01 -9.17250991e-01 1.73666552e-01 8.69182229e-01
5.05567551e-01 -1.17426187e-01 8.66855204e-01 -7.34328449e-01
8.56038034e-01 -1.18366790e+00 -9.57049251e-01 -1.67195275e-02
-7.27723062e-01 8.55587199e-02 5.32912433e-01 -5.65900385e-01
-7.67279744e-01 -3.86235654e-01 1.60681903e-02 -6.78826571e-01
-4.45764899e-01 1.84817165e-01 4.01231572e-02 3.54032278e-01
7.12915540e-01 2.97591507e-01 9.30150598e-02 -4.11087215e-01
-1.99541845e-03 5.68989336e-01 5.77034295e-01 -7.07328200e-01
7.79931843e-01 4.53569680e-01 4.17687036e-02 -1.06044126e+00
-7.54291892e-01 -6.47612929e-01 -3.42632495e-02 1.31902933e-01
3.31331015e-01 -7.34904587e-01 -4.51534241e-01 5.06999731e-01
-5.67131877e-01 -3.34608614e-01 -5.38902164e-01 4.37268138e-01
-6.03145599e-01 5.35931587e-01 -4.02464271e-01 -1.45784640e+00
-2.05064476e-01 -7.32993066e-01 1.01951087e+00 1.38784632e-01
-3.44178647e-01 -1.20827532e+00 4.74441350e-01 -5.08653045e-01
4.98339474e-01 1.32233530e-01 8.95228982e-01 -1.34951127e+00
-1.15653053e-01 -3.97982866e-01 -3.13280195e-01 4.58360106e-01
1.62413701e-01 3.69927883e-01 -7.98016012e-01 -1.66458622e-01
-7.88124800e-02 9.26407129e-02 7.62910962e-01 6.49454296e-01
1.48108006e+00 -1.81668684e-01 -2.28747368e-01 6.37533665e-01
1.20453334e+00 2.58730836e-02 7.61784315e-01 3.30717504e-01
3.25703979e-01 1.61703140e-01 4.78375614e-01 1.06874537e+00
1.18273376e-02 4.19554502e-01 3.71241301e-01 2.55823642e-01
8.36222351e-01 -3.07685107e-01 4.78219897e-01 1.51805833e-01
4.97079313e-01 -3.68012846e-01 -1.15372086e+00 8.03699851e-01
-1.87483120e+00 -1.08839321e+00 -4.72764038e-02 2.51672006e+00
5.30526042e-01 3.48882228e-01 7.46627748e-01 -2.42083203e-02
8.00415456e-01 -2.44725030e-03 -8.16275179e-01 -2.93231994e-01
7.37486035e-02 2.31819481e-01 3.75221044e-01 4.59244221e-01
-1.21744788e+00 6.28825486e-01 7.45895815e+00 7.91345239e-01
-8.21589053e-01 -3.85037601e-01 7.62725830e-01 -1.64855763e-01
-1.95745453e-01 -3.04138511e-01 -6.76388204e-01 8.69799972e-01
1.36079812e+00 -5.05542159e-01 3.57653312e-02 1.22966170e+00
1.12623073e-01 -3.29175681e-01 -9.55766559e-01 9.57085252e-01
-3.00490141e-01 -9.71016407e-01 -9.46208462e-02 2.40241364e-01
5.07275045e-01 7.72801489e-02 3.59432489e-01 5.69132626e-01
4.58596468e-01 -1.06653202e+00 2.37327352e-01 6.15611553e-01
5.24115860e-01 -1.07037318e+00 8.13229382e-01 3.12245727e-01
-8.59987855e-01 -7.20161051e-02 -4.71521825e-01 1.62508383e-01
2.73698092e-01 1.02934539e+00 -9.71417248e-01 4.05867323e-02
6.29448652e-01 5.47731876e-01 -5.08130550e-01 1.28267467e+00
1.22505635e-01 1.12613606e+00 -8.32183659e-01 1.56547025e-01
1.53393373e-01 -5.26329637e-01 7.75236547e-01 1.23672533e+00
6.21638179e-01 -4.17868644e-01 -7.62491748e-02 8.72392893e-01
3.84559482e-02 1.50906041e-01 -7.42415249e-01 6.80047870e-02
7.67907143e-01 8.89641225e-01 -4.60945159e-01 -3.12715173e-01
-6.23760223e-01 5.97005665e-01 2.71304846e-01 6.21426225e-01
-9.00205553e-01 -3.20666790e-01 1.09614277e+00 3.08648109e-01
3.63467783e-01 1.36020314e-03 -2.56302267e-01 -1.28057671e+00
-6.55986294e-02 -6.57776535e-01 7.12926865e-01 -3.27416927e-01
-1.86772823e+00 2.62388438e-01 3.46691370e-01 -1.18321812e+00
-7.87619591e-01 -6.29330575e-01 -1.13955748e+00 8.26925874e-01
-1.11635172e+00 -1.05905913e-01 -1.07179634e-01 5.99687159e-01
2.95892447e-01 -3.76162440e-01 7.21235096e-01 5.60929477e-02
-8.72688711e-01 7.95612693e-01 3.06522161e-01 2.17338368e-01
6.80019617e-01 -1.55266976e+00 7.15795755e-01 1.01153910e+00
-3.62699851e-03 5.65203190e-01 7.75361180e-01 -7.47813523e-01
-6.79538488e-01 -1.06371641e+00 1.38124153e-02 -4.98574317e-01
1.09733605e+00 -1.88456669e-01 -1.51814508e+00 7.01764226e-01
-2.83904195e-01 4.17214632e-01 7.24997282e-01 2.80443996e-01
-3.77349317e-01 2.73253918e-01 -1.56248987e+00 5.16652226e-01
4.75008339e-01 -2.05204472e-01 -3.93883944e-01 5.07898293e-02
3.09238255e-01 -1.74718395e-01 -7.03810573e-01 4.67443734e-01
2.17394069e-01 -1.00888896e+00 7.93720961e-01 -8.89352739e-01
1.10073024e-02 -1.89568862e-01 -3.60844404e-01 -1.31044590e+00
1.93361994e-02 -5.91317952e-01 -1.09684217e+00 1.06663144e+00
3.95284325e-01 -1.06567037e+00 8.39512765e-01 7.12461531e-01
2.16630995e-01 -8.07845175e-01 -1.11814666e+00 -8.91959369e-01
1.06887236e-01 -6.99526489e-01 4.85134274e-01 6.84442103e-01
2.37275790e-02 -4.78477329e-02 -3.37770462e-01 3.67614090e-01
6.64991796e-01 -4.21565264e-01 8.60709906e-01 -1.36707830e+00
-1.83530256e-01 -3.04960608e-01 -6.60182536e-01 -1.01152802e+00
3.24452817e-01 -1.52698979e-01 -4.33105715e-02 -7.74262369e-01
1.09990381e-01 -1.38562948e-01 -2.95147896e-01 1.63266435e-01
-3.62084627e-01 -1.52001694e-01 -4.71385479e-01 4.82537784e-02
-6.30651593e-01 8.29219282e-01 4.24609572e-01 4.37313378e-01
-1.12720512e-01 3.71243536e-01 -5.73269069e-01 9.33078051e-01
9.99388933e-01 -4.68621552e-01 -5.21235347e-01 2.67002434e-01
-2.09703147e-02 -2.91621745e-01 1.96710199e-01 -8.84936750e-01
-1.75984159e-01 -3.05219501e-01 3.95362884e-01 -5.99409699e-01
1.96393326e-01 -7.36126661e-01 -3.34581524e-01 4.00264785e-02
3.06522325e-02 2.12146744e-01 2.51927108e-01 9.72821236e-01
-1.57517925e-01 -2.52809465e-01 1.20357764e+00 4.87144381e-01
-4.37164217e-01 4.06842500e-01 -6.41198814e-01 5.82733095e-01
1.12209034e+00 5.38547598e-02 -2.98031092e-01 -8.22365761e-01
-4.67302471e-01 2.81484157e-01 5.47394395e-01 9.90702957e-02
6.09419286e-01 -1.27488232e+00 -6.11710548e-01 5.60560286e-01
3.17367911e-01 -3.20080156e-03 4.84404108e-03 8.88009965e-01
-3.60643893e-01 7.86711578e-04 2.46106282e-01 -9.56046283e-01
-6.02621794e-01 4.57797408e-01 5.60987175e-01 -2.06624433e-01
-7.37415671e-01 4.55832273e-01 2.11033762e-01 -1.38386607e-01
2.82294214e-01 -2.69794255e-01 1.33277223e-01 -2.43852541e-01
8.19045961e-01 5.28587937e-01 -1.05036773e-01 -2.12760255e-01
-3.86036962e-01 5.82273379e-02 -2.68783271e-01 -2.31072456e-01
1.08816373e+00 2.19872016e-02 2.81269908e-01 6.54089689e-01
1.03782344e+00 -6.63549751e-02 -1.51214790e+00 -1.41856864e-01
4.21917289e-01 -6.56734109e-01 -1.54133782e-01 -1.09105699e-01
-8.02578568e-01 5.77858627e-01 5.00750303e-01 6.63267374e-01
1.00926590e+00 2.80530527e-02 4.83883411e-01 4.46294159e-01
-1.47224786e-02 -1.00847340e+00 1.76844493e-01 6.93570793e-01
4.50268120e-01 -1.18893933e+00 -1.29831553e-01 -4.78988513e-02
-8.44261289e-01 8.22493374e-01 7.64696062e-01 -6.03249669e-01
1.00768030e+00 4.59100097e-01 -1.90795302e-01 -8.96824449e-02
-8.18763494e-01 -1.36049222e-02 3.49792719e-01 7.09790051e-01
2.05284432e-01 -1.02882884e-01 2.69259781e-01 5.90498090e-01
-6.03802949e-02 -5.79112828e-01 4.87186193e-01 6.07063115e-01
-6.55926704e-01 -6.64213240e-01 -4.13400710e-01 8.25650930e-01
-6.61097646e-01 3.94196473e-02 -8.80795494e-02 8.51093650e-01
-7.57701516e-01 7.01424956e-01 7.48907626e-01 1.23236895e-01
2.92953879e-01 4.59766775e-01 3.31636369e-02 -5.21432459e-01
2.58364409e-01 1.29570812e-01 -2.48414665e-01 -5.04011750e-01
2.90853947e-01 -7.51821220e-01 -9.41300571e-01 -5.12409389e-01
-1.98265404e-01 2.50706106e-01 3.06799501e-01 9.83673692e-01
3.42279494e-01 2.61193484e-01 5.71749330e-01 -3.63285005e-01
-9.73440945e-01 -1.17703068e+00 -9.81394708e-01 3.77387553e-01
6.18936837e-01 -8.19459856e-01 -9.27293181e-01 -3.64100039e-01] | [7.628641605377197, 2.486440896987915] |
fbc869eb-2592-4ed6-92d4-38fcec6e4ec9 | learning-to-follow-instructions-in-text-based | 2211.04591 | null | https://arxiv.org/abs/2211.04591v1 | https://arxiv.org/pdf/2211.04591v1.pdf | Learning to Follow Instructions in Text-Based Games | Text-based games present a unique class of sequential decision making problem in which agents interact with a partially observable, simulated environment via actions and observations conveyed through natural language. Such observations typically include instructions that, in a reinforcement learning (RL) setting, can directly or indirectly guide a player towards completing reward-worthy tasks. In this work, we study the ability of RL agents to follow such instructions. We conduct experiments that show that the performance of state-of-the-art text-based game agents is largely unaffected by the presence or absence of such instructions, and that these agents are typically unable to execute tasks to completion. To further study and address the task of instruction following, we equip RL agents with an internal structured representation of natural language instructions in the form of Linear Temporal Logic (LTL), a formal language that is increasingly used for temporally extended reward specification in RL. Our framework both supports and highlights the benefit of understanding the temporal semantics of instructions and in measuring progress towards achievement of such a temporally extended behaviour. Experiments with 500+ games in TextWorld demonstrate the superior performance of our approach. | ['Sheila A. McIlraith', 'Scott Sanner', 'Toryn Q. Klassen', 'Pashootan Vaezipoor', 'Andrew C. Li', 'Mathieu Tuli'] | 2022-11-08 | null | null | null | null | ['text-based-games'] | ['playing-games'] | [ 2.25298241e-01 3.25111300e-01 -2.28835985e-01 -2.09134549e-01
-1.83987439e-01 -6.84748828e-01 1.18219244e+00 2.13760570e-01
-7.35026836e-01 9.46243227e-01 3.72204006e-01 -4.78313833e-01
-1.67749137e-01 -8.93830836e-01 -4.14147079e-01 -4.77967232e-01
-5.67085803e-01 5.42521238e-01 4.96956557e-01 -7.24414110e-01
2.82055557e-01 2.57459462e-01 -1.67696202e+00 -1.13750147e-02
3.29435736e-01 6.01376891e-01 3.61805350e-01 9.84131455e-01
9.45500657e-02 1.71625483e+00 -3.05999607e-01 2.01869428e-01
1.63219005e-01 -7.25834668e-01 -1.10450709e+00 1.46523654e-01
-8.07106853e-01 -6.78339005e-01 -3.93903941e-01 6.21026576e-01
2.15582289e-02 5.00550866e-01 3.87667179e-01 -1.46192229e+00
2.90750060e-03 8.71780217e-01 7.89665282e-02 6.73447698e-02
9.29135323e-01 6.36440217e-01 1.05437768e+00 4.21002023e-02
7.39815354e-01 1.14972138e+00 -1.58693910e-01 7.60397553e-01
-1.06444681e+00 -1.31775498e-01 5.34912348e-01 1.21741109e-01
-8.35758388e-01 -4.32217360e-01 4.75926548e-01 -2.76639462e-01
1.38161182e+00 8.00678581e-02 8.14520478e-01 9.37433600e-01
5.53091526e-01 9.44218397e-01 1.38323236e+00 -5.32575309e-01
6.78302765e-01 -4.13532764e-01 -2.01421216e-01 6.83140278e-01
-3.47276330e-01 1.07753479e+00 -8.29625785e-01 -8.15483406e-02
8.74598324e-01 -1.92123890e-01 8.94616246e-02 -5.21995604e-01
-1.24860609e+00 8.77048254e-01 -2.81503919e-04 3.37708354e-01
-7.37688005e-01 6.04603887e-01 5.39759874e-01 7.24703729e-01
-5.94394021e-02 5.22368371e-01 -2.99231410e-01 -7.89486349e-01
-2.96078563e-01 5.08290112e-01 8.89520526e-01 6.96968019e-01
4.67019677e-01 1.23609178e-01 -3.47509354e-01 9.58048254e-02
3.67662102e-01 2.31867835e-01 5.15203834e-01 -1.40002716e+00
1.03112824e-01 4.10307050e-01 7.76619554e-01 -4.19320047e-01
-6.47397459e-01 -7.51504824e-02 -1.44378040e-02 7.27992058e-01
4.02805150e-01 -2.50868469e-01 -3.64324689e-01 1.98388219e+00
2.68131435e-01 1.93469614e-01 5.30200839e-01 7.65065551e-01
1.39225259e-01 5.25529742e-01 1.69087768e-01 -4.84297186e-01
1.13965452e+00 -6.39300644e-01 -6.48710847e-01 -4.47810829e-01
9.63249385e-01 -8.98883939e-02 1.13274562e+00 4.17816907e-01
-1.34958553e+00 -1.18148163e-01 -1.04043555e+00 3.19836438e-01
9.24701020e-02 -5.09897411e-01 7.38386631e-01 3.16539824e-01
-1.27814496e+00 7.02212095e-01 -1.11094260e+00 -3.23341787e-01
-3.78000736e-02 5.43763041e-01 -2.36872464e-01 1.17698550e-01
-1.09800243e+00 8.89368892e-01 4.37420756e-01 -1.86273590e-01
-1.35097134e+00 3.29468250e-02 -1.18225431e+00 -3.58281843e-02
1.05122924e+00 -4.70613807e-01 2.12514639e+00 -7.39334702e-01
-2.17071652e+00 7.44469106e-01 2.40530312e-01 -8.40859711e-01
5.49886048e-01 -2.23605577e-02 -1.26297534e-01 1.29870951e-01
2.39746332e-01 6.87836230e-01 6.07894540e-01 -7.71078289e-01
-9.90089595e-01 -2.83587903e-01 8.26436698e-01 6.37165725e-01
3.35850179e-01 2.14777514e-01 6.09078780e-02 -8.95130113e-02
-3.46986294e-01 -1.15320265e+00 -6.62913203e-01 -4.10636932e-01
8.77670646e-02 -5.70046306e-01 3.76721442e-01 2.51866937e-01
9.02222335e-01 -2.02090406e+00 1.97431128e-02 -6.07659891e-02
2.22715929e-01 -3.47449869e-01 -2.93568850e-01 8.23038399e-01
2.73742020e-01 -2.02087849e-01 2.91128680e-02 -9.32665691e-02
5.61954856e-01 6.06247842e-01 -3.44293505e-01 4.75910038e-01
3.13841738e-02 1.08494818e+00 -1.31005120e+00 -2.92258501e-01
3.56011987e-01 -2.54050463e-01 -5.51211536e-01 5.13164580e-01
-8.19965720e-01 7.43077874e-01 -9.80462015e-01 1.12901695e-01
-2.94874966e-01 -2.92427763e-02 4.69476163e-01 1.07710123e+00
-4.39070523e-01 9.15590048e-01 -1.11344123e+00 1.70934486e+00
-4.03921276e-01 3.70444179e-01 -3.92830856e-02 -6.11132443e-01
5.25422037e-01 5.43807924e-01 4.55585212e-01 -1.15780985e+00
2.90418059e-01 -8.97505954e-02 3.55848581e-01 -3.69675517e-01
6.23478472e-01 -2.94308513e-01 -4.59823459e-01 1.25868893e+00
-4.29076642e-01 -4.05805916e-01 5.36940873e-01 3.07473093e-01
1.52839422e+00 5.95724583e-01 7.29785025e-01 -1.06987819e-01
3.17005306e-01 2.84081399e-01 3.94093096e-01 1.23449051e+00
-4.60222036e-01 -2.96888739e-01 8.45760584e-01 -4.77507710e-01
-7.68759847e-01 -6.94710374e-01 5.54726303e-01 1.61834526e+00
1.29858211e-01 -5.35260320e-01 -3.07258219e-01 -5.03971875e-01
-4.36360508e-01 1.02782774e+00 -6.23807490e-01 -3.26684654e-01
-6.50759876e-01 -8.18076357e-02 4.86680061e-01 4.54017550e-01
2.65347570e-01 -1.82719028e+00 -1.85902882e+00 5.72464585e-01
-1.12514190e-01 -1.05402148e+00 -5.50559103e-01 4.46797788e-01
-7.55453110e-01 -1.14240110e+00 2.11332086e-03 -4.11599398e-01
3.46358716e-01 -5.65010495e-02 1.32055306e+00 2.20308751e-01
1.96096644e-01 1.01047158e+00 -5.17505050e-01 -3.20523143e-01
-5.87991357e-01 -4.22125936e-01 2.05759093e-01 -4.34966803e-01
1.78646624e-01 -4.77843970e-01 -4.51165229e-01 2.79024005e-01
-9.72873449e-01 3.16284090e-01 1.97485179e-01 6.48233414e-01
2.09553763e-01 2.49926522e-01 2.01118425e-01 -5.92441797e-01
1.15109074e+00 -2.32008740e-01 -9.39578593e-01 -1.33912846e-01
-4.27833021e-01 6.37015164e-01 6.73146486e-01 -3.72570366e-01
-1.05050027e+00 3.95245738e-02 3.75134230e-01 2.60103345e-01
-2.83591658e-01 6.78632677e-01 2.31303230e-01 2.33255669e-01
6.08033240e-01 3.96025389e-01 6.15475588e-02 4.12524909e-01
3.33141416e-01 1.35520473e-01 4.21128720e-01 -1.10428488e+00
3.84585470e-01 2.63527602e-01 3.30064297e-01 -2.59157836e-01
-2.12662071e-01 -2.09455982e-01 -1.32313803e-01 -5.04888058e-01
5.52232265e-01 -6.86441004e-01 -1.57569134e+00 2.70977229e-01
-7.48499393e-01 -1.22106898e+00 -6.74436510e-01 4.21705246e-01
-1.41019905e+00 8.53107423e-02 -6.50244594e-01 -1.21759939e+00
3.01075578e-01 -1.40788293e+00 8.83313000e-01 1.69444025e-01
-4.02593225e-01 -8.36839318e-01 2.60254502e-01 -6.83858097e-02
2.19561592e-01 -2.47905287e-03 6.71124876e-01 -4.64730978e-01
-5.88921130e-01 3.22710574e-01 4.96742427e-01 -3.13709289e-01
2.95183845e-02 -4.37791288e-01 -5.19919634e-01 -1.38232097e-01
1.86457574e-01 -7.42737174e-01 -1.43245305e-03 5.30236721e-01
4.21457320e-01 -2.70223975e-01 2.45599728e-03 -2.27898657e-01
1.17321682e+00 6.46013737e-01 5.69581091e-01 6.20655835e-01
-2.86018103e-01 8.34695697e-01 1.25010443e+00 9.65464592e-01
5.80115914e-01 7.37774074e-01 7.76001751e-01 5.57900131e-01
4.08188015e-01 -3.40762705e-01 7.99195707e-01 1.08483560e-01
-2.33538315e-01 -2.15764239e-01 -7.73486555e-01 4.07726347e-01
-2.23739982e+00 -1.04820371e+00 3.43170390e-02 2.19109678e+00
1.04602242e+00 3.90831649e-01 3.59344780e-01 3.31296697e-02
2.04220235e-01 2.41498336e-01 -6.82067394e-01 -6.88412488e-01
3.29623580e-01 2.93303967e-01 3.60065371e-01 7.18412519e-01
-6.82793379e-01 1.20796561e+00 6.47619009e+00 3.92271608e-01
-7.08548009e-01 -1.30311593e-01 1.84742093e-01 -1.66267380e-01
-2.14689318e-02 2.07668155e-01 -2.14077622e-01 1.84174869e-02
1.17123961e+00 -6.10485017e-01 9.80864704e-01 5.52444518e-01
7.54869103e-01 -6.98453426e-01 -1.60876572e+00 5.81019521e-01
-6.49325132e-01 -9.62714851e-01 -5.73101938e-01 1.92562997e-01
4.26596999e-01 -5.96959330e-02 9.47612226e-02 6.84563994e-01
1.34051347e+00 -1.07962132e+00 1.12450743e+00 2.83098429e-01
5.44386506e-01 -7.82965004e-01 4.99706089e-01 8.49846125e-01
-1.15363753e+00 -2.98342168e-01 6.39667064e-02 -9.36140001e-01
1.27741620e-01 -3.79742205e-01 -1.00807178e+00 3.72682780e-01
2.94727415e-01 4.51411247e-01 6.34014327e-03 4.54831541e-01
-7.20489204e-01 3.02951068e-01 -2.13886350e-01 -4.11470056e-01
8.64091039e-01 -3.19899440e-01 4.56231087e-01 5.42139173e-01
1.38475806e-01 6.75272703e-01 5.55931389e-01 6.56638086e-01
3.88992131e-01 -2.29158357e-01 -9.81702924e-01 -2.21289515e-01
1.72526240e-01 6.46826625e-01 -7.92084813e-01 -2.82298535e-01
-1.90399885e-01 8.69844496e-01 2.30580255e-01 3.11689347e-01
-7.06528664e-01 2.31299773e-01 8.53042245e-01 -1.16361134e-01
3.23849618e-02 -5.56686521e-01 9.74079520e-02 -8.10141981e-01
-1.23881146e-01 -1.01080310e+00 2.89564490e-01 -1.10322380e+00
-5.79989612e-01 4.33245659e-01 3.65145355e-02 -1.06933260e+00
-1.11687374e+00 -3.21980625e-01 -5.49770772e-01 4.33759481e-01
-1.36938751e+00 -3.97420645e-01 1.17432892e-01 7.91205525e-01
6.40549183e-01 8.34394097e-02 8.12386692e-01 -5.79096913e-01
-5.71009032e-02 -2.45653406e-01 -4.41785604e-01 -2.11880788e-01
4.04567838e-01 -1.41495430e+00 1.70784697e-01 7.28973329e-01
-2.17668600e-02 3.49912971e-01 1.06471896e+00 -6.16480887e-01
-1.59441018e+00 -7.10007370e-01 5.28363228e-01 -3.36680949e-01
7.85198212e-01 -9.52111208e-04 -2.61992663e-01 9.93289411e-01
3.46325070e-01 -1.56550050e-01 4.28557724e-01 -1.80060104e-01
1.78555116e-01 3.96921754e-01 -1.02697742e+00 1.19196296e+00
1.17435277e+00 -6.05326533e-01 -8.75055969e-01 9.05117095e-02
8.29760075e-01 -5.42494297e-01 -3.37405801e-01 -9.89082456e-02
3.96882325e-01 -1.13892114e+00 6.00783825e-01 -1.00489855e+00
4.96234179e-01 -4.14724827e-01 -1.37973085e-01 -1.47640920e+00
-1.29060209e-01 -1.03937984e+00 5.91624640e-02 4.46101546e-01
-8.24839473e-02 -6.29666328e-01 8.43114734e-01 7.68666446e-01
-6.11111782e-02 -2.41467953e-01 -8.91762018e-01 -7.96891451e-01
-9.94031057e-02 -8.91175151e-01 6.69005692e-01 3.37635189e-01
7.09414124e-01 2.10115403e-01 -3.25507253e-01 -9.57246944e-02
3.45087379e-01 -9.62972641e-03 6.04069710e-01 -1.00022936e+00
-4.68309551e-01 -4.86854106e-01 -1.39920622e-01 -1.23074436e+00
4.41946805e-01 -6.67856455e-01 6.36606097e-01 -1.38227272e+00
-1.23186961e-01 -4.53127801e-01 -1.83624133e-01 6.81389689e-01
2.94816524e-01 -3.88679534e-01 3.48973900e-01 -2.46063247e-02
-1.22060764e+00 6.75506711e-01 1.42789400e+00 1.91224948e-01
-4.60262835e-01 3.98023874e-02 -5.10193288e-01 6.41368449e-01
8.34944248e-01 -4.79149789e-01 -7.79389918e-01 -1.15343995e-01
5.67505002e-01 9.94379103e-01 1.97603837e-01 -7.52513707e-01
5.58375835e-01 -9.27020967e-01 -4.91628736e-01 -2.34971136e-01
2.92813778e-01 -8.61450195e-01 -1.57814790e-02 8.12396765e-01
-7.59596348e-01 3.08634996e-01 1.05170496e-01 5.76658785e-01
-5.62993735e-02 -2.20693275e-01 2.57609069e-01 -3.01171273e-01
-9.29003239e-01 1.43732041e-01 -1.33461666e+00 1.17119402e-01
1.29334581e+00 -5.74542908e-03 -1.65550902e-01 -8.70917916e-01
-6.84296310e-01 6.26241803e-01 4.66923118e-01 1.42906785e-01
6.66553915e-01 -1.07133591e+00 -5.29098749e-01 -9.53588709e-02
1.65391818e-01 -2.58967459e-01 -1.45010069e-01 7.06537783e-01
-2.67346859e-01 6.58800244e-01 -5.54508746e-01 -3.11246604e-01
-1.09469783e+00 4.94570553e-01 3.69355261e-01 -7.89181173e-01
-7.82346189e-01 3.13416600e-01 3.49650979e-01 -4.52197492e-01
1.25079677e-01 -6.07416332e-01 -3.72174323e-01 -3.93933207e-01
5.12738228e-01 -2.30481867e-02 -2.30242074e-01 -2.22499356e-01
-2.68134534e-01 -2.32650384e-01 1.02570266e-01 -9.05519903e-01
1.35299492e+00 -2.21926630e-01 6.66896626e-03 5.62208176e-01
1.37899950e-01 -3.29231232e-01 -1.69206345e+00 -2.94033140e-01
1.63219377e-01 -2.36686260e-01 -2.39106845e-02 -7.43227482e-01
-4.76665378e-01 3.47032428e-01 9.10889506e-02 5.85376263e-01
9.33352768e-01 -3.44292857e-02 3.99386495e-01 5.50136149e-01
1.06406581e+00 -1.12744427e+00 3.58030111e-01 9.40701723e-01
6.02613926e-01 -9.85285759e-01 -4.13908392e-01 1.18376739e-01
-8.27020705e-01 1.06128848e+00 4.76671249e-01 1.06731253e-02
8.68966221e-04 5.34213126e-01 -8.20000544e-02 -3.37023616e-01
-1.56004369e+00 -6.35861158e-01 -4.96912897e-01 7.70213723e-01
1.83206260e-01 3.50057334e-01 -2.72496253e-01 3.88298333e-01
-3.74941707e-01 3.17604721e-01 1.13061094e+00 1.68698514e+00
-5.43896139e-01 -1.35430992e+00 -3.88909370e-01 2.65169293e-01
-3.27006996e-01 9.46537331e-02 -2.72536904e-01 6.61377549e-01
-3.75732750e-01 1.53730094e+00 7.72774443e-02 -1.31722856e-02
2.18813002e-01 2.34133489e-02 7.86972106e-01 -8.85810435e-01
-6.85280442e-01 -1.95754528e-01 1.94213212e-01 -1.08269036e+00
-5.36810637e-01 -8.92028987e-01 -1.78475153e+00 -3.31789017e-01
2.75890946e-01 2.90496677e-01 3.95382315e-01 1.35897517e+00
-1.90402851e-01 8.24510872e-01 5.72618008e-01 -6.82187319e-01
-7.60333121e-01 -4.38106686e-01 -7.31245577e-01 8.50944370e-02
5.45183241e-01 -6.90156281e-01 -1.60649166e-01 -2.46522725e-01] | [3.989110231399536, 1.5103503465652466] |
a1499035-b5a6-48b1-bf92-85f667335eae | learning-to-plan-chemical-syntheses | 1708.04202 | null | http://arxiv.org/abs/1708.04202v1 | http://arxiv.org/pdf/1708.04202v1.pdf | Learning to Plan Chemical Syntheses | From medicines to materials, small organic molecules are indispensable for
human well-being. To plan their syntheses, chemists employ a problem solving
technique called retrosynthesis. In retrosynthesis, target molecules are
recursively transformed into increasingly simpler precursor compounds until a
set of readily available starting materials is obtained. Computer-aided
retrosynthesis would be a highly valuable tool, however, past approaches were
slow and provided results of unsatisfactory quality. Here, we employ Monte
Carlo Tree Search (MCTS) to efficiently discover retrosynthetic routes. MCTS
was combined with an expansion policy network that guides the search, and an
"in-scope" filter network to pre-select the most promising retrosynthetic
steps. These deep neural networks were trained on 12 million reactions, which
represents essentially all reactions ever published in organic chemistry. Our
system solves almost twice as many molecules and is 30 times faster in
comparison to the traditional search method based on extracted rules and
hand-coded heuristics. Finally after a 60 year history of computer-aided
synthesis planning, chemists can no longer distinguish between routes generated
by a computer system and real routes taken from the scientific literature. We
anticipate that our method will accelerate drug and materials discovery by
assisting chemists to plan better syntheses faster, and by enabling fully
automated robot synthesis. | ['Mark P. Waller', 'Mike Preuss', 'Marwin H. S. Segler'] | 2017-08-14 | null | null | null | null | ['retrosynthesis'] | ['medical'] | [ 3.25768232e-01 1.52456909e-01 -5.41210711e-01 2.70767789e-02
-5.73548853e-01 -1.13708663e+00 5.26186705e-01 4.53427076e-01
-5.04905283e-01 1.34204841e+00 1.26538292e-01 -7.53098071e-01
2.60716885e-01 -9.18014526e-01 -6.36801720e-01 -5.23001373e-01
-5.63357249e-02 9.10403728e-01 -7.36200139e-02 -3.56931150e-01
6.03046238e-01 9.56495643e-01 -1.00337517e+00 1.43978506e-01
8.57346356e-01 6.60133243e-01 4.74726051e-01 5.62661767e-01
-1.87546894e-01 5.83815038e-01 -3.83319080e-01 -2.28319749e-01
2.46154964e-01 -4.35627580e-01 -1.03486562e+00 -3.30664217e-01
-1.86507270e-01 -1.82727203e-02 1.13137633e-01 7.50190377e-01
4.82865512e-01 4.43700343e-01 7.20685959e-01 -3.06521416e-01
-4.13823128e-01 7.46966362e-01 2.24032868e-02 -1.06315978e-01
7.13031769e-01 3.00976753e-01 1.07172334e+00 -1.09292817e+00
9.33281183e-01 1.05284083e+00 4.79993641e-01 7.04584360e-01
-1.05092835e+00 -7.74845183e-01 -2.68587563e-02 -3.99200730e-02
-1.29196560e+00 -5.32866359e-01 3.07344198e-01 -4.07168031e-01
1.61138940e+00 5.92160262e-02 1.00756466e+00 9.47136939e-01
7.21823215e-01 2.35671684e-01 5.19138992e-01 -2.63071150e-01
6.48717344e-01 4.14006924e-03 -8.84122312e-01 8.84655058e-01
-2.10454259e-02 3.67159843e-01 -2.77728319e-01 2.69916207e-02
7.67775059e-01 -9.24002826e-02 -1.11647055e-01 -1.34276778e-01
-1.44574285e+00 1.13582778e+00 6.26939237e-01 4.41489369e-01
-5.02574384e-01 1.77872822e-01 4.22302872e-01 -7.52384961e-02
-1.84904560e-02 1.61604583e+00 -6.78625464e-01 1.17534973e-01
-1.12057650e+00 2.75882244e-01 1.04838300e+00 8.47448587e-01
4.37405407e-01 9.88240391e-02 4.28423166e-01 3.21896791e-01
2.21152440e-01 -1.02821112e-01 3.17680359e-01 -9.70357597e-01
2.39855275e-01 2.51696259e-01 3.86114269e-01 -6.54060364e-01
-7.16813862e-01 -2.50315279e-01 -5.52647233e-01 6.95205629e-02
4.89174396e-01 -1.91971645e-01 -1.08107960e+00 1.28374028e+00
4.79883581e-01 -6.87440157e-01 1.55740529e-01 7.45680511e-01
6.69240654e-01 1.31106937e+00 3.19234848e-01 -5.85760176e-01
1.17860830e+00 -1.08132815e+00 -4.82532531e-01 1.86433256e-01
7.55626321e-01 -9.57452774e-01 5.21294236e-01 1.01350403e+00
-1.41452062e+00 -2.44175673e-01 -1.28943729e+00 -1.08955421e-01
-9.91779268e-01 -1.28039911e-01 1.01312733e+00 6.90991759e-01
-8.95013690e-01 1.28698850e+00 -6.85973525e-01 -2.52662927e-01
5.37019312e-01 8.46713841e-01 -2.83652395e-01 9.93582234e-02
-1.17876494e+00 1.23281133e+00 1.01929867e+00 1.71418026e-01
-1.46957409e+00 -7.75876999e-01 -7.15124965e-01 9.50741116e-03
7.89438188e-01 -8.18701506e-01 1.53810871e+00 -7.29315519e-01
-1.93483675e+00 4.43328440e-01 -5.71427196e-02 -3.83529007e-01
1.56159252e-01 3.36883277e-01 -2.09649235e-01 1.79128543e-01
4.52973098e-02 1.17605531e+00 5.11005044e-01 -8.40496838e-01
-5.92567146e-01 9.07431170e-02 -3.61173064e-03 3.08282286e-01
4.85086411e-01 1.49234161e-01 6.59182295e-03 -2.99777806e-01
-8.70481541e-04 -7.40863740e-01 -1.01981604e+00 -1.81064203e-01
-5.25504768e-01 -2.72693604e-01 -5.22352718e-02 -4.50918525e-01
9.03805673e-01 -1.15835321e+00 3.08942258e-01 2.39572749e-01
1.36984855e-01 2.16777503e-01 -1.02845512e-01 9.26659465e-01
-3.93605322e-01 3.54658723e-01 -3.39605696e-02 4.94796097e-01
-2.64753938e-01 -1.18707143e-01 -2.19668806e-01 2.24640101e-01
2.16544330e-01 7.63515115e-01 -1.39735711e+00 -2.68629342e-01
1.89631090e-01 1.74623027e-01 -7.53071666e-01 -1.35181218e-01
-1.04244363e+00 7.05290616e-01 -5.51562428e-01 1.11681378e+00
5.71116693e-02 -3.93641621e-01 5.80819190e-01 -9.26602557e-02
-5.91196835e-01 8.05921197e-01 -6.91738605e-01 2.06743741e+00
-4.87133652e-01 8.15359596e-03 -2.87931383e-01 -7.86352038e-01
8.69630337e-01 5.98689139e-01 5.59760690e-01 -6.33499682e-01
2.62703121e-01 4.94874001e-01 2.01683611e-01 -1.50405973e-01
6.71979547e-01 -6.05146587e-01 1.31374910e-01 2.17813715e-01
-1.49292618e-01 -6.93227708e-01 8.15975904e-01 2.32002325e-02
8.57036710e-01 5.46451092e-01 8.73208106e-01 -2.24828929e-01
6.23254299e-01 7.02333331e-01 4.45921570e-01 5.10555267e-01
3.71063836e-02 1.85476393e-01 2.07541198e-01 -1.07448196e+00
-1.44604909e+00 -9.21206236e-01 7.37740099e-02 9.88124669e-01
-2.61199892e-01 -5.63900709e-01 -4.78248268e-01 -5.06045818e-01
-3.95835251e-01 6.85249984e-01 -2.09845543e-01 7.50960037e-02
-6.23962343e-01 -5.79737067e-01 2.40751013e-01 3.21945965e-01
-3.82201560e-02 -1.33419168e+00 -4.70535278e-01 1.02256036e+00
1.65750608e-01 -5.96980989e-01 -4.71405208e-01 6.49487197e-01
-8.38806748e-01 -1.01689839e+00 -7.12768018e-01 -7.18150318e-01
5.92134118e-01 -3.61160550e-04 9.60891604e-01 -2.48813376e-01
-7.00156033e-01 -4.42011386e-01 -1.79781746e-02 -4.92706299e-01
-9.21306908e-01 2.73855895e-01 7.41390884e-02 -1.04221523e+00
-1.73447460e-01 -5.88201880e-01 -8.40770304e-01 3.61988507e-02
-5.44963539e-01 6.53439984e-02 7.12266743e-01 7.69303739e-01
6.98012769e-01 2.30913490e-01 7.24698722e-01 -6.45986259e-01
5.41026533e-01 -4.06853378e-01 -9.50267375e-01 1.89763531e-01
-8.13563406e-01 3.66948903e-01 1.12051308e+00 -3.00928742e-01
-9.67588186e-01 5.99312961e-01 -3.82008672e-01 9.07959118e-02
-6.35378212e-02 6.32377446e-01 -2.77457722e-02 -2.77006552e-02
8.04117024e-01 4.77669425e-02 -3.22521925e-01 -1.16901919e-01
6.20282948e-01 -3.12058236e-02 2.30102956e-01 -7.90750861e-01
3.33950937e-01 1.27295613e-01 3.43545824e-01 -6.04014337e-01
-5.68452299e-01 -2.45667085e-01 -5.09667754e-01 1.70680925e-01
9.44814026e-01 -6.87645972e-01 -1.23223019e+00 -4.54089642e-01
-1.36997855e+00 -3.35270673e-01 -8.26588124e-02 4.79316384e-01
-7.39465237e-01 2.35700801e-01 -5.11270881e-01 -4.66574699e-01
-5.79674959e-01 -1.52477634e+00 8.05119872e-01 2.55632937e-01
-6.03232801e-01 -8.91466916e-01 2.10472867e-01 2.05292881e-01
3.48993428e-02 2.69101620e-01 1.14899898e+00 -7.29998410e-01
-8.87854517e-01 4.57522459e-02 1.44022983e-02 -3.01674873e-01
4.62046593e-01 1.93036914e-01 -4.97501969e-01 -1.19658243e-02
-4.93060529e-01 -4.75659013e-01 7.09958196e-01 4.21606511e-01
1.04763949e+00 -5.71466565e-01 -7.09720612e-01 3.40759903e-01
1.36051667e+00 1.18874466e+00 4.69116867e-01 4.65952903e-01
6.47485614e-01 7.18557656e-01 6.51133776e-01 4.14707273e-01
-7.20956028e-02 1.82028219e-01 3.90150070e-01 7.76893198e-02
3.73637289e-01 -4.53711629e-01 2.17776120e-01 3.13511252e-01
-4.32041615e-01 -4.46526110e-01 -9.61095750e-01 1.49018735e-01
-1.22774351e+00 -9.41099226e-01 3.80538613e-01 2.11317492e+00
1.37557137e+00 1.76203832e-01 3.13660294e-01 -4.11374718e-02
3.23037535e-01 -2.15020135e-01 -7.53070652e-01 -8.53192449e-01
4.12515908e-01 7.81439424e-01 8.35420251e-01 6.14334643e-01
-1.01274812e+00 1.51471889e+00 7.45688009e+00 7.98641086e-01
-1.23656368e+00 -5.10120809e-01 6.05790317e-01 -1.73545405e-01
-1.31568342e-01 2.33021364e-01 -8.59330475e-01 2.08967745e-01
1.10817480e+00 -2.40769863e-01 8.00495088e-01 8.33396256e-01
5.89572549e-01 -1.68949589e-01 -1.51518738e+00 6.12062573e-01
-5.29610455e-01 -2.23490977e+00 2.50428617e-01 -3.17428894e-02
8.18741620e-01 -2.81454653e-01 -1.71184525e-01 -4.97880718e-03
7.27405131e-01 -1.45177376e+00 6.84489191e-01 1.36860505e-01
7.22440422e-01 -1.06833196e+00 7.27620423e-02 1.61368668e-01
-1.19996274e+00 -8.69918540e-02 -3.39381069e-01 5.29004000e-02
2.25689784e-01 2.61382639e-01 -1.71718132e+00 6.76791430e-01
4.03862633e-02 5.84782124e-01 1.11066185e-01 8.74263167e-01
-2.82392621e-01 2.51177582e-03 -2.21158996e-01 -5.86795568e-01
7.68937290e-01 -3.84906173e-01 1.38075739e-01 1.07211936e+00
3.35138410e-01 2.00091630e-01 2.58931458e-01 1.04959071e+00
-1.43831526e-03 2.10373312e-01 -6.13222480e-01 -8.10308278e-01
5.11945903e-01 1.06650174e+00 -1.33049154e+00 -4.94768977e-01
-6.27843067e-02 6.74017966e-01 -1.61421850e-01 2.22658902e-01
-7.40666866e-01 -3.84708345e-01 1.86855853e-01 -2.02607766e-01
3.35975438e-01 -3.55246723e-01 2.53167767e-02 -6.23602867e-01
-7.51305282e-01 -1.20257759e+00 2.26964995e-01 -7.24652886e-01
-6.97406590e-01 4.03964579e-01 -3.48676771e-01 -6.85349107e-01
-1.90676108e-01 -8.35576177e-01 -4.11358297e-01 6.67792797e-01
-1.08993757e+00 -5.40526867e-01 5.92995405e-01 3.50467898e-02
9.92341042e-01 -6.88537136e-02 9.02951837e-01 9.65793654e-02
-5.44018209e-01 4.46617268e-02 1.26947880e-01 -6.12921357e-01
6.53949082e-01 -1.15181768e+00 4.55635935e-01 2.11715430e-01
-7.54900696e-03 9.79669690e-01 9.08723593e-01 -8.51215303e-01
-1.52068830e+00 -8.46990466e-01 8.66813660e-01 -1.59894690e-01
7.44201660e-01 -1.71414763e-01 -3.09468299e-01 2.43370846e-01
1.65056020e-01 -7.19407380e-01 6.10776544e-01 -1.74527928e-01
-2.41678953e-02 1.82886690e-01 -9.77883339e-01 9.42177653e-01
8.47859681e-01 -1.36858717e-01 -2.61952370e-01 9.99224246e-01
8.98580670e-01 -5.95041573e-01 -1.03110862e+00 1.11950673e-01
8.04479837e-01 -5.36610305e-01 1.15376091e+00 -6.88303530e-01
7.49843597e-01 -3.55324924e-01 1.46845207e-01 -1.10727417e+00
-2.52577960e-01 -1.15365803e+00 3.63199681e-01 3.54688257e-01
1.06043434e+00 -3.81550103e-01 1.02406383e+00 3.98465931e-01
-4.10353750e-01 -9.86078501e-01 -5.83144367e-01 -8.51286292e-01
3.51915300e-01 1.49931787e-02 7.52202511e-01 8.71089160e-01
3.83127004e-01 6.99598193e-01 -1.29142195e-01 -1.28201276e-01
7.32702985e-02 2.00221501e-03 2.13331550e-01 -9.66586828e-01
-1.64678693e-01 -7.60942817e-01 3.29282224e-01 -9.60431695e-01
-1.41871452e-01 -1.05319381e+00 2.72702366e-01 -1.63945568e+00
-6.86876550e-02 -4.47586268e-01 1.23520773e-02 4.53881562e-01
4.43434924e-01 -2.61313379e-01 -1.95040494e-01 -9.32348669e-02
-4.17176783e-01 4.12837535e-01 1.61260545e+00 -3.56615663e-01
-6.71540022e-01 -2.89267927e-01 -7.88366020e-01 6.13266110e-01
9.18510973e-01 -5.77817678e-01 -4.33165491e-01 2.71577418e-01
8.52513671e-01 4.90579844e-01 -3.24986488e-01 -7.72305071e-01
3.13856125e-01 -5.62101483e-01 6.04714453e-01 -9.30067360e-01
2.65922457e-01 -6.18765473e-01 4.01842386e-01 9.06927347e-01
-3.86766791e-01 -2.06295297e-01 2.22065434e-01 3.28159392e-01
2.38443911e-01 -4.18030739e-01 5.11522949e-01 -1.01111805e+00
-4.19925451e-01 4.84030396e-01 -1.04253364e+00 -5.33686757e-01
9.64805841e-01 -3.69877487e-01 -3.18326103e-03 6.79599196e-02
-9.52167392e-01 1.83793530e-01 3.71074110e-01 -5.75557277e-02
5.59549510e-01 -9.17537868e-01 -1.20036807e-02 -3.70511711e-01
-2.22568303e-01 2.80050904e-01 -2.09218621e-01 4.45269465e-01
-1.13550615e+00 1.10225165e+00 -1.33744299e-01 -9.03333500e-02
-7.95064270e-01 1.11765969e+00 3.84349823e-01 -3.93064737e-01
-1.57510102e-01 9.44382548e-01 2.11389765e-01 -3.85828197e-01
2.28299454e-01 -6.84723198e-01 6.37802407e-02 1.11413591e-01
3.91407132e-01 2.63753951e-01 5.73409274e-02 3.76016460e-02
-4.63610530e-01 3.15196544e-01 -4.99648690e-01 -1.20383769e-01
1.25315058e+00 4.38357949e-01 7.47640654e-02 -1.58449844e-01
7.57688999e-01 -1.12523757e-01 -9.96688187e-01 3.33304137e-01
8.85049328e-02 2.14596197e-01 -1.58074602e-01 -1.14960897e+00
-4.08388168e-01 6.15054548e-01 -4.50397246e-02 -1.02801576e-01
7.16413140e-01 -1.63464412e-01 8.78119409e-01 1.10391676e+00
3.53919595e-01 -1.15998518e+00 2.80810177e-01 4.21336204e-01
8.61789167e-01 -1.06500912e+00 5.06791770e-01 -2.88681746e-01
-2.97137082e-01 1.54530168e+00 1.39866307e-01 1.61006585e-01
3.40813063e-02 6.40950277e-02 -5.20256758e-01 -3.64076674e-01
-7.97431946e-01 -2.03956850e-02 -7.67660737e-02 3.90526682e-01
6.89902544e-01 -1.29692510e-01 -3.53205085e-01 1.57665595e-01
-3.59874934e-01 1.31783172e-01 4.54422772e-01 1.13991106e+00
-7.67064869e-01 -1.55422485e+00 -3.78886610e-01 -1.03810653e-01
-5.02865851e-01 -4.98115212e-01 -7.03757465e-01 8.50074649e-01
2.08882496e-01 9.56365228e-01 -4.06717092e-01 -1.97793581e-02
6.94196895e-02 9.37110037e-02 1.06296897e+00 -7.91353285e-01
-6.33141518e-01 2.50394344e-01 5.69002748e-01 -6.74315929e-01
-2.73656160e-01 -3.49063337e-01 -1.61418748e+00 -3.90786737e-01
-3.38338047e-01 5.01436830e-01 8.81562829e-01 8.90639305e-01
7.12985247e-02 4.21127081e-01 6.92093432e-01 -1.12028837e+00
-1.76781818e-01 -2.76054233e-01 -9.62073058e-02 -8.65575492e-01
3.09912604e-03 -4.02954161e-01 1.43877551e-01 2.21394315e-01] | [4.513131141662598, 6.08503532409668] |
d4424e6f-ea72-46e4-93d8-d39149289771 | speech-data-augmentation-for-improving | null | null | https://aclanthology.org/2022.rapid-1.8 | https://aclanthology.org/2022.rapid-1.8.pdf | Speech Data Augmentation for Improving Phoneme Transcriptions of Aphasic Speech Using Wav2Vec 2.0 for the PSST Challenge | As part of the PSST challenge, we explore how data augmentations, data sources, and model size affect phoneme transcription accuracy on speech produced by individuals with aphasia. We evaluate model performance in terms of feature error rate (FER) and phoneme error rate (PER). We find that data augmentations techniques, such as pitch shift, improve model performance. Additionally, increasing the size of the model decreases FER and PER. Our experiments also show that adding manually-transcribed speech from non-aphasic speakers (TIMIT) improves performance when Room Impulse Response is used to augment the data. The best performing model combines aphasic and non-aphasic data and has a 21.0% PER and a 9.2% FER, a relative improvement of 9.8% compared to the baseline model on the primary outcome measurement. We show that data augmentation, larger model size, and additional non-aphasic data sources can be helpful in improving automatic phoneme recognition models for people with aphasia. | ['Jonas Beskow', 'Joakim Gustafson', 'Harm Lameris', 'Ambika Kirkland', 'Shivam Mehta', 'Jim O’Regan', 'Birger Moell'] | null | null | null | null | rapid-lrec-2022-6 | ['room-impulse-response'] | ['audio'] | [ 2.25529745e-01 1.63974568e-01 1.04549557e-01 -1.65740177e-01
-1.09600699e+00 -2.96097845e-01 4.56815302e-01 9.06277522e-02
-7.68656790e-01 4.56062406e-01 1.38911259e+00 -3.22573751e-01
3.22197825e-01 -1.76601544e-01 -9.65254381e-02 -1.92497253e-01
3.97231989e-02 3.42567742e-01 -3.09547901e-01 -3.46777380e-01
7.03893155e-02 2.84562021e-01 -1.18387651e+00 6.27801716e-01
1.06358492e+00 5.76352239e-01 1.25713319e-01 7.42791653e-01
1.74881473e-01 6.77411079e-01 -1.04102969e+00 6.34424835e-02
1.94082946e-01 -6.78846896e-01 -8.38700235e-01 4.10034060e-02
3.81696105e-01 -5.69294453e-01 -6.43628597e-01 4.16517973e-01
1.06921101e+00 3.10598701e-01 5.76521933e-01 -8.24452877e-01
-4.67437714e-01 6.40591025e-01 3.77297252e-02 5.99897861e-01
4.92508024e-01 4.48845834e-01 5.14799476e-01 -9.67915535e-01
4.60254624e-02 1.32260680e+00 1.05639505e+00 6.68678761e-01
-1.32281649e+00 -6.01511419e-01 2.41136238e-01 3.22676033e-01
-1.12737465e+00 -1.23281586e+00 3.34179521e-01 -4.03214782e-01
1.68757439e+00 4.35096830e-01 1.11151826e+00 1.13696074e+00
-1.91937163e-01 5.41500747e-01 1.15385807e+00 -7.21336961e-01
1.08735427e-01 1.42032415e-01 3.30022335e-01 9.08750668e-02
-2.36481950e-01 1.46305636e-01 -9.26312387e-01 -2.01853067e-01
7.19386816e-01 -5.57810545e-01 -6.68816686e-01 6.06555343e-01
-1.30430830e+00 6.57327235e-01 1.21742547e-01 2.96650797e-01
-6.52873039e-01 -1.40154064e-01 4.65553284e-01 5.15954852e-01
2.61229813e-01 9.33799446e-01 -5.65996468e-01 -8.17301273e-01
-7.88315117e-01 -6.91677704e-02 6.24054432e-01 6.71569228e-01
-3.02161306e-01 5.17575502e-01 -5.74835777e-01 1.53202808e+00
-4.38185669e-02 5.38397849e-01 1.13729560e+00 -1.07278502e+00
7.25839257e-01 3.27378511e-01 8.60074461e-02 -4.29379046e-01
-7.58580208e-01 -5.27968585e-01 -5.29442966e-01 -7.03207776e-02
5.69941401e-01 -4.00706649e-01 -1.13604081e+00 2.00845599e+00
-3.49581391e-01 -3.88152480e-01 8.39171633e-02 6.38960898e-01
4.19426680e-01 6.62992418e-01 3.23558271e-01 -5.34191072e-01
1.30465424e+00 -1.00598502e+00 -1.11107564e+00 -7.99836874e-01
1.26296139e+00 -7.47368634e-01 1.67892456e+00 6.61279917e-01
-1.32036841e+00 -4.89135832e-01 -6.52051568e-01 -2.62519382e-02
5.58409207e-02 2.89626181e-01 1.77895412e-01 9.54005718e-01
-1.28463256e+00 2.66872793e-01 -7.41058826e-01 -5.00366747e-01
3.89285862e-01 4.23868269e-01 -5.18537939e-01 1.74425766e-01
-1.21715093e+00 1.14695132e+00 1.70163289e-01 -4.28998202e-01
-3.53469968e-01 -8.93418014e-01 -7.65694737e-01 1.42666236e-01
-3.97392124e-01 -6.61915958e-01 1.52077758e+00 -7.71178365e-01
-1.49523509e+00 4.89626825e-01 -3.50272119e-01 -7.50920177e-01
4.85353887e-01 -5.23450196e-01 -7.35363960e-01 1.80033922e-01
-3.24356467e-01 8.85640919e-01 7.28341103e-01 -5.68710148e-01
-6.50827050e-01 -3.39098096e-01 -5.69908082e-01 6.94209456e-01
-8.21328044e-01 3.46548706e-01 1.52698278e-01 -1.10760832e+00
5.52256331e-02 -9.31620300e-01 1.79282017e-02 -2.09425762e-01
-2.20334515e-01 -1.37767047e-01 3.23717386e-01 -1.67784846e+00
1.49198496e+00 -2.26833200e+00 -6.66944832e-02 -2.99455505e-02
1.68638807e-02 3.88195574e-01 -1.47555202e-01 2.13799983e-01
-2.96637625e-01 4.15642828e-01 -3.98453057e-01 -5.75179577e-01
-1.78486824e-01 -1.70942247e-01 -4.69790623e-02 9.11929533e-02
9.80670005e-02 6.14284277e-01 -3.14551771e-01 -1.84598044e-02
3.08120757e-01 6.13313496e-01 -9.23197925e-01 5.98052964e-02
4.05581087e-01 5.98513484e-01 6.14454091e-01 4.71253581e-02
2.28136659e-01 9.22252610e-02 -3.34590316e-01 1.39373839e-01
7.74320215e-02 9.93356645e-01 -8.02050173e-01 1.43708837e+00
-6.46970689e-01 8.75564396e-01 -5.79084493e-02 -4.16766286e-01
7.00209200e-01 6.14974618e-01 3.02290112e-01 -7.59636819e-01
2.17912450e-01 2.09046602e-01 7.39116967e-01 -3.75669390e-01
3.14419210e-01 -2.40359545e-01 2.53588438e-01 4.56449479e-01
-2.71133602e-01 -5.62686063e-02 -4.24166024e-02 7.21136630e-02
1.26429904e+00 -8.65156353e-01 2.40916565e-01 -1.88183919e-01
1.89394087e-01 -1.29920349e-01 1.66593775e-01 9.22056437e-01
-3.65827143e-01 8.21619689e-01 6.72464073e-02 1.74747765e-01
-1.22562921e+00 -8.67276430e-01 -7.58391172e-02 1.06252170e+00
-9.49690878e-01 -7.54868329e-01 -9.63159323e-01 -1.88910261e-01
-2.56458193e-01 1.19323957e+00 -3.14202398e-01 -6.16649985e-01
-7.70341754e-01 -6.92986429e-01 7.02574909e-01 1.16087127e+00
5.53374410e-01 -1.29635274e+00 -2.91426092e-01 3.74558687e-01
-8.11607122e-01 -1.07602096e+00 -7.90274143e-01 -5.37150428e-02
-9.20929253e-01 -4.47965175e-01 -9.31889117e-01 -7.42891490e-01
2.79265076e-01 5.76712601e-02 6.43155694e-01 -7.12522194e-02
8.42063352e-02 5.21597624e-01 -5.65585315e-01 -3.93287659e-01
-7.18804479e-01 1.81190565e-01 6.08706594e-01 -4.07368094e-01
3.20047826e-01 -5.39369285e-01 -4.53462839e-01 1.42506361e-01
-3.34414989e-01 1.52246282e-01 5.58320224e-01 8.13226223e-01
9.58540291e-02 -4.04597640e-01 9.01010871e-01 -2.21383452e-01
1.10906029e+00 -1.16152331e-01 2.30605111e-01 -2.53577739e-01
-5.83533585e-01 -2.06802785e-01 4.65392470e-01 -8.28142047e-01
-8.86450171e-01 -1.05075315e-01 -5.14394939e-01 -3.76842506e-02
-3.98800492e-01 3.98165822e-01 -8.31047073e-02 2.94047743e-01
1.07715690e+00 2.10222155e-01 4.19094324e-01 -7.44985104e-01
1.57151803e-01 1.29604149e+00 7.34328270e-01 1.23745359e-01
2.47784331e-01 -5.99019937e-02 -7.16911495e-01 -1.16912389e+00
-1.85767829e-01 -3.95974696e-01 -5.21791160e-01 7.61832446e-02
6.28113568e-01 -1.05148029e+00 -1.68329686e-01 8.56168687e-01
-1.07434535e+00 -7.88916647e-01 -5.40607989e-01 9.72782493e-01
-7.35738933e-01 1.31750122e-01 -5.79270840e-01 -8.77136469e-01
-7.43989170e-01 -1.02749944e+00 5.12405455e-01 -1.97436392e-01
-1.06721449e+00 -5.31301022e-01 -1.77136928e-01 6.24227226e-01
7.87887573e-01 -4.70380753e-01 1.02684331e+00 -1.00433123e+00
4.58513409e-01 -3.27995986e-01 1.17382873e-03 6.53442442e-01
6.48907065e-01 -7.98609495e-01 -1.09984899e+00 -3.44830304e-01
2.71963298e-01 -1.64362609e-01 5.42120218e-01 5.23536503e-01
9.39061701e-01 -6.43281817e-01 4.41923365e-02 1.31304264e-01
2.54562259e-01 4.42047417e-01 6.71748281e-01 1.18777044e-01
5.24983585e-01 5.66422164e-01 4.37287167e-02 4.64681089e-01
2.57095665e-01 8.22077572e-01 -4.63855565e-01 1.31604895e-01
-6.89192653e-01 -1.81963786e-01 8.32351327e-01 1.18491518e+00
2.48102382e-01 -3.87113750e-01 -1.24084818e+00 6.97072327e-01
-1.25712216e+00 -8.16608965e-01 -1.77891046e-01 2.31073427e+00
9.38387036e-01 -1.52525364e-03 6.57959819e-01 8.25794160e-01
6.67193890e-01 -2.92667210e-01 -5.16906381e-01 -4.98237908e-01
-2.34774977e-01 1.72667995e-01 1.68132827e-01 8.57354403e-01
-6.27661705e-01 8.88744891e-01 7.19674253e+00 3.97471368e-01
-8.89162242e-01 2.34943092e-01 5.26018560e-01 -5.19515812e-01
1.98287919e-01 -5.63091934e-01 -4.53132331e-01 5.85089803e-01
1.70439327e+00 -1.51990071e-01 7.05460727e-01 3.94588888e-01
6.79684520e-01 -3.81646976e-02 -1.17060268e+00 1.22496724e+00
3.42918932e-01 -7.27502465e-01 1.69483602e-01 2.24404797e-01
2.46954322e-01 1.42523468e-01 2.09674388e-02 4.05737936e-01
-1.17878914e-01 -1.22827685e+00 6.40305519e-01 4.52468961e-01
8.61950755e-01 -5.67547381e-01 6.00770235e-01 2.78577805e-01
-7.06908405e-01 -4.09686208e-01 3.10563415e-01 -3.82005244e-01
4.88380134e-01 -7.56967515e-02 -1.05738008e+00 -2.59797603e-01
6.46773219e-01 3.18987966e-01 -6.52446508e-01 1.29285431e+00
-5.08076362e-02 1.09098387e+00 -7.19104826e-01 3.63956839e-01
-2.74021327e-01 4.65260565e-01 6.53692186e-01 1.36520565e+00
4.42200989e-01 4.61498559e-01 -1.60811812e-01 3.39004606e-01
2.68076472e-02 3.92659903e-01 -2.97277659e-01 -1.80324346e-01
9.11190569e-01 4.54094231e-01 -1.35503188e-01 -4.13281798e-01
-2.86305934e-01 1.15020812e+00 8.54736418e-02 1.67543277e-01
-2.43190125e-01 -4.24287915e-01 6.00448132e-01 2.88090438e-01
-3.72031748e-01 -1.91465795e-01 -7.19552577e-01 -8.16922963e-01
1.12668186e-01 -1.45117366e+00 3.26610118e-01 -9.13697481e-01
-1.07022524e+00 6.29101396e-01 -1.17311403e-01 -1.10109663e+00
-6.18284762e-01 -5.13478518e-01 -4.02526170e-01 1.22403395e+00
-5.91385484e-01 -5.77392817e-01 -4.12181973e-01 5.28161764e-01
8.48997295e-01 -3.98302406e-01 1.12937856e+00 2.85563052e-01
-5.25170803e-01 1.11163723e+00 -1.82181045e-01 1.80311278e-01
7.99353004e-01 -1.03056049e+00 8.28095555e-01 8.15219343e-01
-6.70688376e-02 6.19540274e-01 6.56024635e-01 -7.23202646e-01
-7.08443761e-01 -8.30504477e-01 1.08712316e+00 -7.48878777e-01
3.47245246e-01 -2.92350888e-01 -1.13775301e+00 6.06209576e-01
8.67101327e-02 -6.49272680e-01 8.15734923e-01 1.15539774e-01
-1.26179338e-01 1.88861385e-01 -1.31563759e+00 8.36901367e-01
1.31204867e+00 -6.92469597e-01 -7.95412540e-01 3.78754944e-01
9.20046210e-01 -7.85536095e-02 -1.06722713e+00 4.09739017e-01
5.36245704e-01 -4.09258962e-01 7.67123342e-01 -7.93435216e-01
-3.57821733e-02 3.10764164e-01 -1.94806218e-01 -1.99349523e+00
-5.03795266e-01 -7.34181881e-01 -6.97266608e-02 9.78092432e-01
6.69236898e-01 -8.49467695e-01 4.04538780e-01 1.09377360e+00
-5.43164134e-01 -2.11816847e-01 -9.18247461e-01 -9.30032790e-01
5.68718612e-01 -7.18750656e-01 3.25569391e-01 7.24828660e-01
7.06866026e-01 4.09629405e-01 -3.38527948e-01 -2.72526275e-02
-2.20031306e-01 -1.13317978e+00 3.65220845e-01 -1.23505199e+00
-1.61788598e-01 -7.22407043e-01 -3.32326978e-01 -8.13222468e-01
-1.55606821e-01 -7.80112147e-01 1.00800321e-02 -1.73384738e+00
-2.58307964e-01 -3.00255805e-01 -3.07714306e-02 8.56309772e-01
-1.48472175e-01 -2.87881158e-02 5.72754383e-01 3.51668924e-01
3.02553087e-01 7.27936149e-01 6.96485102e-01 6.28966466e-02
-8.54125082e-01 -2.40239706e-02 -9.04375851e-01 6.54075086e-01
1.42716324e+00 -3.28503102e-01 -4.70561981e-01 -5.84683239e-01
-5.25061011e-01 -2.23791841e-02 1.68243840e-01 -1.41760349e+00
1.23631306e-01 3.30092400e-01 4.59965140e-01 -1.15746722e-01
9.56002057e-01 -3.79485548e-01 -2.04727829e-01 5.12019753e-01
-5.78996301e-01 3.90390784e-01 8.16525161e-01 3.77206095e-02
2.78585672e-01 -1.57846302e-01 8.15655887e-01 -4.83482704e-03
7.25914091e-02 -3.03755313e-01 -1.12157428e+00 3.14713359e-01
2.56936938e-01 -1.75030097e-01 -4.39083844e-01 -8.37935507e-01
-1.01020181e+00 -1.62053749e-01 7.73031935e-02 6.72831833e-01
3.96624207e-01 -1.29326093e+00 -9.29715991e-01 3.65717560e-01
8.52607638e-02 -6.45818651e-01 6.05433956e-02 1.06588221e+00
-8.29049870e-02 6.66409373e-01 -2.06549883e-01 -7.37814158e-02
-1.38868308e+00 -9.64193940e-02 5.56928217e-01 2.67349154e-01
-8.34306836e-01 8.38424087e-01 -1.71151698e-01 -2.61655569e-01
6.97388530e-01 -5.50013959e-01 -1.31636679e-01 1.33518517e-01
9.38531756e-01 7.08525717e-01 4.31229621e-01 -7.02411294e-01
-2.08328828e-01 -1.02260381e-01 -5.33439279e-01 -9.79835212e-01
9.38994169e-01 -9.14527699e-02 4.00551587e-01 5.08968949e-01
6.77474201e-01 -2.87936535e-02 -8.23132038e-01 -1.52211457e-01
-3.13690275e-01 -1.63765088e-01 4.24549222e-01 -1.50499976e+00
-6.08229756e-01 9.38241541e-01 1.11214793e+00 1.44370139e-01
1.05731058e+00 -2.37919122e-01 8.30660343e-01 4.76717621e-01
1.15471005e-01 -1.22206211e+00 5.71932271e-02 6.17496669e-01
1.38851833e+00 -8.01077247e-01 -5.04427195e-01 -7.12525472e-02
-9.65566754e-01 6.34091616e-01 7.33907521e-01 2.84857243e-01
4.88817632e-01 2.43300647e-01 1.20690979e-01 4.47432488e-01
-4.94536459e-01 -1.26855105e-01 1.78463057e-01 7.74051428e-01
4.80274796e-01 8.55419114e-02 -2.80159742e-01 1.05199015e+00
-1.03999615e+00 -1.31969735e-01 5.14525414e-01 7.75294721e-01
-5.53118706e-01 -7.32438862e-01 -5.42507827e-01 1.01529062e+00
-1.15416303e-01 -5.67614436e-01 -6.90795898e-01 6.25660479e-01
-2.95510381e-01 1.50211048e+00 3.29048872e-01 -4.94319350e-01
6.93874359e-01 7.80330002e-01 2.78642207e-01 -7.88752079e-01
-9.11686599e-01 4.65525612e-02 5.43850899e-01 -2.29362443e-01
2.93010831e-01 -1.06042540e+00 -1.30456007e+00 -2.69032925e-01
-2.29502887e-01 5.87337539e-02 4.58343208e-01 9.77670491e-01
6.58010125e-01 6.33712769e-01 2.42521480e-01 -3.73243898e-01
-4.93985057e-01 -2.03980350e+00 -8.48942026e-02 1.63316861e-01
5.35586774e-01 -3.81177485e-01 -6.92822039e-01 1.15262911e-01] | [14.456958770751953, 6.373517036437988] |
0258a8ff-087e-4f08-85ef-b2a46874daa6 | decorrelated-adversarial-learning-for-age | 1904.04972 | null | http://arxiv.org/abs/1904.04972v1 | http://arxiv.org/pdf/1904.04972v1.pdf | Decorrelated Adversarial Learning for Age-Invariant Face Recognition | There has been an increasing research interest in age-invariant face
recognition. However, matching faces with big age gaps remains a challenging
problem, primarily due to the significant discrepancy of face appearances
caused by aging. To reduce such a discrepancy, in this paper we propose a novel
algorithm to remove age-related components from features mixed with both
identity and age information. Specifically, we factorize a mixed face feature
into two uncorrelated components: identity-dependent component and
age-dependent component, where the identity-dependent component includes
information that is useful for face recognition. To implement this idea, we
propose the Decorrelated Adversarial Learning (DAL) algorithm, where a
Canonical Mapping Module (CMM) is introduced to find the maximum correlation
between the paired features generated by a backbone network, while the backbone
network and the factorization module are trained to generate features reducing
the correlation. Thus, the proposed model learns the decomposed features of age
and identity whose correlation is significantly reduced. Simultaneously, the
identity-dependent feature and the age-dependent feature are respectively
supervised by ID and age preserving signals to ensure that they both contain
the correct information. Extensive experiments are conducted on popular
public-domain face aging datasets (FG-NET, MORPH Album 2, and CACD-VS) to
demonstrate the effectiveness of the proposed approach. | ['Wei Liu', 'Hao Wang', 'Zhifeng Li', 'Dihong Gong'] | 2019-04-10 | decorrelated-adversarial-learning-for-age-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Wang_Decorrelated_Adversarial_Learning_for_Age-Invariant_Face_Recognition_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Wang_Decorrelated_Adversarial_Learning_for_Age-Invariant_Face_Recognition_CVPR_2019_paper.pdf | cvpr-2019-6 | ['age-invariant-face-recognition'] | ['computer-vision'] | [ 6.18759245e-02 -1.94298625e-01 1.54894054e-01 -5.00984192e-01
-1.03100859e-01 -1.23713866e-01 2.76345998e-01 -3.44036460e-01
-2.25694686e-01 6.49412513e-01 1.63702890e-01 2.33204678e-01
-3.13879997e-02 -8.22457314e-01 -4.63347673e-01 -9.95479405e-01
-7.74141699e-02 -9.17070135e-02 -2.99548507e-01 -2.31522992e-01
4.14176062e-02 4.11341697e-01 -1.67771637e+00 1.27638057e-02
1.16020656e+00 1.39337540e+00 -2.87668735e-01 1.19459992e-02
-1.10809445e-01 4.72221285e-01 -4.17550832e-01 -4.84234631e-01
6.03638291e-01 -5.21582067e-01 -1.10903472e-01 -2.16278364e-03
6.01322830e-01 -3.86723071e-01 -5.33946276e-01 1.29665399e+00
7.18219459e-01 -3.24678048e-02 5.33563018e-01 -1.75013459e+00
-7.57797301e-01 1.02172032e-01 -9.10308480e-01 -6.69904891e-03
1.43796712e-01 -9.52445902e-03 3.45868349e-01 -1.06616294e+00
3.22724968e-01 1.68615723e+00 6.55704916e-01 9.30454552e-01
-1.01445901e+00 -1.36658955e+00 1.23205736e-01 4.78283882e-01
-1.45673323e+00 -5.85147917e-01 1.05267930e+00 -2.92436093e-01
5.98100238e-02 9.32244956e-02 5.70558190e-01 9.87597942e-01
1.49205387e-01 3.03220034e-01 1.37569916e+00 -3.32672328e-01
-2.37348061e-02 1.66850239e-01 -3.01543977e-02 7.97434509e-01
3.17788005e-01 1.62481219e-01 -5.41887403e-01 -5.36787510e-02
6.46394789e-01 3.02936375e-01 -3.24774086e-01 -3.03371161e-01
-6.27949476e-01 5.27050853e-01 3.05903345e-01 1.18347473e-01
-3.18315148e-01 -2.24261507e-01 2.75361449e-01 5.67647576e-01
6.32995903e-01 -2.31272265e-01 -4.23364788e-01 4.29582864e-01
-6.77338839e-01 2.59241581e-01 3.96431297e-01 6.46517336e-01
9.77768600e-01 7.47886598e-02 -9.74716693e-02 1.01023197e+00
4.21445310e-01 7.08258033e-01 6.23222351e-01 -7.46251941e-01
3.69190425e-01 7.65807688e-01 -1.26894861e-01 -1.40033901e+00
-2.80291855e-01 -3.08924824e-01 -1.07836998e+00 3.90570700e-01
4.03628081e-01 -1.53352514e-01 -8.13057065e-01 2.37851238e+00
5.64676344e-01 5.67402780e-01 -1.47253033e-02 7.23386168e-01
6.93556607e-01 4.09625322e-01 2.91245103e-01 -6.07777596e-01
1.51905072e+00 -6.22788846e-01 -7.25276828e-01 -2.73793906e-01
9.77625251e-02 -8.87218595e-01 7.05638826e-01 -1.07816339e-01
-8.08301389e-01 -8.17545354e-01 -1.20014751e+00 6.64258003e-02
-2.35203087e-01 3.39269578e-01 4.22016412e-01 7.04978466e-01
-8.25556874e-01 6.26140356e-01 -3.91688019e-01 -1.98752686e-01
5.56660891e-01 5.52812219e-01 -7.21879601e-01 -3.39234293e-01
-1.27404404e+00 6.00071251e-01 6.36436939e-02 3.30907643e-01
-4.93844211e-01 -8.75973880e-01 -7.44276524e-01 -8.46252814e-02
6.05284385e-02 -6.58056259e-01 5.68113863e-01 -1.24556077e+00
-1.08379388e+00 8.23365331e-01 -1.41300917e-01 -7.82605633e-03
3.55915844e-01 -1.26066625e-01 -7.92030454e-01 6.20597377e-02
2.18938529e-01 5.06188393e-01 1.40227759e+00 -1.10471511e+00
-5.25408685e-01 -1.08023345e+00 -2.65694171e-01 1.18764080e-01
-8.12676311e-01 1.11159265e-01 -3.44506204e-01 -8.27218831e-01
1.72497511e-01 -7.95098305e-01 2.09165528e-01 5.27384818e-01
3.31018615e-04 -6.92719221e-02 1.07854915e+00 -1.29452407e+00
1.15359652e+00 -2.41016698e+00 1.98750928e-01 1.65732458e-01
2.47784004e-01 1.70761541e-01 -4.88091081e-01 -2.82678455e-02
-5.98099351e-01 -1.81795105e-01 -2.19197392e-01 -4.12469387e-01
-4.70460087e-01 -6.08597100e-02 -1.44019470e-01 5.56652904e-01
3.43752921e-01 4.66350526e-01 -7.42222488e-01 -5.55255294e-01
-9.39989313e-02 5.76041520e-01 -4.39029038e-01 3.72079760e-01
1.78161919e-01 6.66046381e-01 -4.57777768e-01 7.68140435e-01
1.43758821e+00 4.73392427e-01 9.22706574e-02 -6.38987362e-01
5.11802174e-02 -4.85489875e-01 -1.30697560e+00 1.39935422e+00
-3.03485572e-01 -7.64339492e-02 4.00768131e-01 -9.39255893e-01
1.12420189e+00 1.66774869e-01 6.82230055e-01 -7.17543364e-01
3.53733271e-01 1.66140452e-01 3.42172049e-02 -3.64423782e-01
-1.19346105e-01 -2.27540851e-01 3.29903990e-01 3.28653902e-01
1.16396524e-01 4.76024747e-01 -4.70756218e-02 -4.97374721e-02
9.93088305e-01 -2.49496922e-02 -4.32693511e-02 -2.13741317e-01
1.12297630e+00 -7.75769413e-01 1.13839495e+00 -1.01426922e-01
-5.00875354e-01 4.10002083e-01 4.15741831e-01 -4.40347224e-01
-9.19596136e-01 -9.32837844e-01 2.46897805e-02 7.52023101e-01
2.34102502e-01 -2.67896891e-01 -8.77422452e-01 -9.35722589e-01
3.17945212e-01 9.83679667e-02 -1.01072121e+00 -6.59271121e-01
-5.93447983e-01 -6.59118533e-01 2.25739062e-01 4.16426599e-01
8.31515014e-01 -9.09186363e-01 3.04722518e-01 -1.24145724e-01
-7.09941890e-03 -7.86282480e-01 -8.90485168e-01 -4.85446006e-01
-7.07597792e-01 -1.29776287e+00 -7.78401315e-01 -7.76668727e-01
1.15903747e+00 4.40295756e-01 4.72784728e-01 2.50498682e-01
-3.12736005e-01 2.02479482e-01 -1.18494123e-01 -1.91947922e-01
-1.98524073e-01 -2.35080197e-01 4.79368120e-01 7.66266763e-01
3.54731619e-01 -9.53074634e-01 -7.23636866e-01 2.45106265e-01
-7.41250217e-01 -1.52204961e-01 7.04374671e-01 9.16941226e-01
1.44121632e-01 1.06406607e-01 8.34879577e-01 -5.82618296e-01
4.79152441e-01 -4.20678258e-01 -2.26626068e-01 2.21907452e-01
-7.24610984e-01 2.62497831e-02 7.36828566e-01 -4.85720009e-01
-1.20783949e+00 8.67452249e-02 -7.82443359e-02 -7.82138646e-01
1.31147981e-01 9.60245132e-02 -8.83530200e-01 -2.85016984e-01
3.03733677e-01 2.99601406e-01 5.06623924e-01 -5.21581471e-01
2.31665567e-01 5.24791360e-01 6.31347895e-01 -5.42937398e-01
1.50548339e+00 4.50825781e-01 2.05128431e-01 -4.74294007e-01
-6.00703418e-01 -1.11751162e-01 -4.17281419e-01 -2.09894896e-01
6.11021221e-01 -1.05591846e+00 -6.03178263e-01 9.74886179e-01
-1.07445216e+00 5.66909671e-01 -7.91639760e-02 4.05265480e-01
-1.52798072e-01 5.31015992e-01 -4.98452544e-01 -7.43062079e-01
-6.45509601e-01 -8.40405345e-01 6.85602188e-01 7.60518432e-01
2.24865332e-01 -5.31160057e-01 -5.95190674e-02 1.85735792e-01
2.98380107e-01 3.71710211e-01 9.83796239e-01 -2.86708742e-01
-1.21173650e-01 -2.42855608e-01 -4.74952519e-01 7.33719289e-01
5.96214771e-01 -4.72363494e-02 -9.83186424e-01 -5.70848882e-01
1.47426128e-01 -2.34475702e-01 7.46860027e-01 3.38613428e-02
1.05452216e+00 -3.05621147e-01 -2.41487220e-01 7.05955982e-01
1.16967368e+00 2.52208173e-01 7.37156272e-01 -1.26643432e-02
7.12983966e-01 7.24558890e-01 5.55918097e-01 4.69301701e-01
1.51596323e-01 5.31195998e-01 3.16275090e-01 -1.31846696e-01
-2.35289440e-01 -3.48397225e-01 3.63837749e-01 8.27584267e-01
-1.17939159e-01 4.16992724e-01 -2.83545494e-01 1.45374224e-01
-1.67896724e+00 -1.01762414e+00 2.71359891e-01 2.27583623e+00
8.70719492e-01 -1.87967971e-01 8.44408572e-03 2.26306155e-01
1.09064090e+00 1.81652546e-01 -6.73583388e-01 1.04342245e-01
-2.20987320e-01 4.64569956e-01 8.60901996e-02 4.92054224e-02
-8.88605058e-01 4.85043585e-01 4.74712181e+00 8.55365515e-01
-1.14447975e+00 8.75714198e-02 6.72171831e-01 9.22751054e-02
-1.15809925e-01 -1.21378198e-01 -5.73703706e-01 8.33284438e-01
4.48081732e-01 -3.59187335e-01 6.53780639e-01 1.01307023e+00
1.62422761e-01 2.76045650e-01 -8.03037226e-01 1.11763692e+00
3.25620711e-01 -6.29687071e-01 7.79743567e-02 -4.45247181e-02
5.00129879e-01 -8.96230757e-01 3.35622013e-01 3.74572724e-01
-2.93565422e-01 -7.94947743e-01 5.29057384e-01 6.68509305e-01
1.06319892e+00 -9.62750971e-01 5.76212347e-01 -5.17178811e-02
-1.58366752e+00 -3.21747363e-01 -4.01847929e-01 1.78009749e-03
-3.89410496e-01 6.90219939e-01 -1.18521333e-01 7.25596905e-01
7.27914631e-01 8.07770610e-01 -8.18941891e-01 7.26316631e-01
-1.39466897e-01 2.39073619e-01 -3.18031833e-02 6.20369017e-01
-5.44318140e-01 -5.24336100e-01 3.66151601e-01 3.24881077e-01
5.46073556e-01 1.42980501e-01 -9.82625410e-02 6.45345390e-01
-4.51669216e-01 2.19657987e-01 -4.22470152e-01 1.30070031e-01
7.36465037e-01 1.55824673e+00 -4.36617613e-01 -1.33556679e-01
-4.68437463e-01 1.12826693e+00 3.11350524e-01 1.36677146e-01
-8.26125145e-01 -3.55511129e-01 1.05201638e+00 6.94579408e-02
3.12753655e-02 4.95390259e-02 3.36313732e-02 -1.24854946e+00
3.38929236e-01 -1.11216724e+00 2.59072185e-01 -4.52147245e-01
-1.57812357e+00 3.41452748e-01 -4.71457392e-01 -1.24610031e+00
2.03679964e-01 -3.38070512e-01 -8.62656415e-01 1.18282485e+00
-1.44814134e+00 -1.41797030e+00 -7.11083651e-01 7.26186693e-01
1.49044782e-01 -3.95188868e-01 7.45621026e-01 8.61848891e-01
-9.63405073e-01 9.54386055e-01 6.16696430e-03 2.36811101e-01
1.30538106e+00 -7.09133327e-01 4.69911983e-03 8.50185096e-01
-5.02703607e-01 7.97817528e-01 4.34487462e-01 -7.36194015e-01
-1.42198479e+00 -1.31681430e+00 7.13817596e-01 -3.90462615e-02
1.65419161e-01 -2.38747865e-01 -1.03750467e+00 3.38517100e-01
-1.67284012e-01 3.79790485e-01 4.93462920e-01 -1.56926647e-01
-7.65380263e-01 -7.83199310e-01 -1.39799213e+00 4.34757262e-01
1.21244860e+00 -4.56169218e-01 -4.54192758e-01 -1.06746398e-01
6.45218909e-01 4.98116715e-03 -7.77655602e-01 6.51239991e-01
9.03055608e-01 -9.56638157e-01 8.99306417e-01 -4.60720569e-01
4.13677335e-01 -4.77335662e-01 5.37089221e-02 -1.32339621e+00
-3.44484925e-01 -2.85485923e-01 -1.48344263e-01 1.82722688e+00
-2.53137529e-01 -7.08053827e-01 7.69978881e-01 4.20666099e-01
8.38139206e-02 -6.84634387e-01 -9.77803409e-01 -7.86319613e-01
-1.20790727e-01 2.57773399e-01 8.32812190e-01 1.00442231e+00
-5.00359893e-01 2.52747983e-01 -5.22614241e-01 2.04820365e-01
8.42398405e-01 -5.03049158e-02 7.08940744e-01 -1.40521479e+00
6.40912503e-02 -1.77606687e-01 -4.57959980e-01 -3.35551023e-01
4.89400476e-01 -6.54657900e-01 -2.49702781e-01 -8.39768827e-01
3.38738203e-01 -3.87174010e-01 -5.20268261e-01 4.16678697e-01
-5.23588896e-01 3.10736626e-01 6.22500144e-02 6.61481246e-02
-9.28895199e-04 9.26454306e-01 1.20348275e+00 -2.98454851e-01
-4.07240726e-02 -7.53834322e-02 -8.36426616e-01 5.61634183e-01
6.60752714e-01 -3.86286825e-01 -4.77921069e-01 -7.68634602e-02
-3.21154833e-01 -1.42189652e-01 2.98061401e-01 -1.33467650e+00
8.21462870e-02 -6.65104911e-02 8.59716415e-01 -2.39717707e-01
3.32140803e-01 -7.84891367e-01 1.74281329e-01 6.49423540e-01
8.92375633e-02 7.37219751e-02 1.22724600e-01 5.16659379e-01
-2.23480493e-01 8.86581838e-02 1.05178785e+00 1.87860191e-01
-4.56299305e-01 8.87001216e-01 3.79861206e-01 -1.09793715e-01
1.07574177e+00 2.44789990e-03 -3.27362537e-01 -1.69442713e-01
-4.72862422e-01 1.67158812e-01 4.66390729e-01 7.27188706e-01
6.83856606e-01 -1.74264550e+00 -8.75393629e-01 6.67587101e-01
7.87261575e-02 -4.62386727e-01 7.35732734e-01 6.16805732e-01
3.77815799e-03 -1.17638908e-01 -6.93988800e-01 1.87713671e-02
-1.42337453e+00 8.00280690e-01 2.41105914e-01 -6.08591661e-02
4.51737433e-04 7.41863847e-01 2.54338384e-01 -2.43427575e-01
4.07401472e-03 4.04191643e-01 -3.11180651e-01 3.79448086e-01
7.37738192e-01 4.87981081e-01 -7.12535009e-02 -8.32356274e-01
-4.96161669e-01 8.64022553e-01 -1.48871690e-01 1.95574582e-01
1.22424042e+00 -1.80250928e-01 -5.63696682e-01 -1.25751421e-01
1.38671839e+00 1.97861478e-01 -1.14903200e+00 -2.37405330e-01
-4.45833087e-01 -6.71033263e-01 -1.52819678e-01 -3.95645797e-01
-1.56884599e+00 7.55153835e-01 1.18105590e+00 -2.59256631e-01
1.55293453e+00 -3.46897125e-01 8.07520330e-01 -3.38743180e-01
2.33995751e-01 -9.36935902e-01 2.51527309e-01 1.02012806e-01
9.22386885e-01 -1.02516425e+00 5.64922094e-02 -6.34085238e-01
2.68799029e-02 9.93801773e-01 1.10891676e+00 -9.35142934e-02
8.62098217e-01 -1.47546843e-01 -4.11660485e-02 3.09437424e-01
-4.01521474e-01 -7.94156734e-03 2.38327265e-01 7.58730590e-01
1.10890083e-01 -2.43594214e-01 -6.67260528e-01 1.01372588e+00
-1.10411555e-01 8.47924724e-02 -7.66074806e-02 7.36163497e-01
-1.81945249e-01 -1.41862059e+00 -5.19240797e-01 4.67380404e-01
-3.12675416e-01 6.11132607e-02 -1.75804302e-01 5.06197929e-01
8.24951053e-01 7.34384656e-01 -5.50701050e-03 -7.90521204e-01
2.33916685e-01 1.55347064e-01 5.22657156e-01 -3.03997517e-01
-2.24597648e-01 -3.31386805e-01 -3.18010807e-01 -5.58396280e-01
-2.54457414e-01 -5.90574741e-01 -9.97177541e-01 -5.15369058e-01
-1.62864223e-01 1.04601309e-01 4.92673784e-01 6.34492815e-01
4.42836493e-01 3.23733270e-01 1.29456806e+00 -6.35465503e-01
-6.67432427e-01 -1.00374496e+00 -8.17115963e-01 7.65473008e-01
1.53925195e-01 -9.47357595e-01 -5.39744377e-01 5.57597540e-02] | [13.223690032958984, 0.5382445454597473] |
db3be938-06ea-4df2-bb1c-e15445cb2484 | occlusions-motion-and-depth-boundaries-with-a | 1808.01838 | null | http://arxiv.org/abs/1808.01838v2 | http://arxiv.org/pdf/1808.01838v2.pdf | Occlusions, Motion and Depth Boundaries with a Generic Network for Disparity, Optical Flow or Scene Flow Estimation | Occlusions play an important role in disparity and optical flow estimation,
since matching costs are not available in occluded areas and occlusions
indicate depth or motion boundaries. Moreover, occlusions are relevant for
motion segmentation and scene flow estimation. In this paper, we present an
efficient learning-based approach to estimate occlusion areas jointly with
disparities or optical flow. The estimated occlusions and motion boundaries
clearly improve over the state-of-the-art. Moreover, we present networks with
state-of-the-art performance on the popular KITTI benchmark and good generic
performance. Making use of the estimated occlusions, we also show improved
results on motion segmentation and scene flow estimation. | ['Thomas Brox', 'Margret Keuper', 'Eddy Ilg', 'Tonmoy Saikia'] | 2018-08-06 | occlusions-motion-and-depth-boundaries-with-a-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Eddy_Ilg_Occlusions_Motion_and_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Eddy_Ilg_Occlusions_Motion_and_ECCV_2018_paper.pdf | eccv-2018-9 | ['scene-flow-estimation'] | ['computer-vision'] | [-6.98799640e-02 -4.24055934e-01 -5.69988012e-01 -3.15025359e-01
-3.70301336e-01 -4.91196573e-01 2.68276662e-01 -2.03694910e-01
-5.85911751e-01 9.36613381e-01 1.27009556e-01 -8.36626515e-02
2.93428063e-01 -7.05564797e-01 -4.19040769e-01 -5.52366734e-01
-1.93211511e-01 1.03055328e-01 8.17902088e-01 1.24647751e-01
3.40717018e-01 6.24548137e-01 -1.59993923e+00 1.80357933e-01
8.95742714e-01 9.66500759e-01 5.27641699e-02 1.00519621e+00
-4.06747848e-01 1.06508684e+00 -4.26097125e-01 -2.61208028e-01
3.67062569e-01 -1.69907793e-01 -9.40600038e-01 7.84252733e-02
1.25893283e+00 -8.66127968e-01 -7.72179067e-01 7.00776398e-01
4.22096252e-01 3.47775310e-01 5.02160430e-01 -1.22454727e+00
1.79526117e-02 -4.66950610e-02 -8.83848965e-01 6.73043787e-01
3.23719889e-01 3.02744627e-01 7.89890289e-01 -7.87478507e-01
9.28999364e-01 1.23790014e+00 3.68429005e-01 4.61569905e-01
-9.50872600e-01 -4.09194142e-01 4.31203008e-01 4.96448427e-01
-1.09917879e+00 -5.46157181e-01 7.86249340e-01 -5.95389068e-01
9.70642328e-01 7.01043606e-02 7.04405069e-01 5.23126662e-01
1.50369883e-01 1.19719303e+00 6.69807732e-01 -2.56326586e-01
9.67182443e-02 -2.03352764e-01 1.52298152e-01 8.58355701e-01
2.74946094e-01 3.36726159e-01 -5.30889094e-01 2.97031432e-01
1.15905964e+00 -3.73686999e-01 -4.81818020e-01 -4.74524617e-01
-1.16255665e+00 7.55411506e-01 5.45457721e-01 7.67338872e-02
-9.20890421e-02 6.04145765e-01 4.14652914e-01 -2.57104337e-01
6.26478374e-01 -4.67972569e-02 -2.43411139e-01 -2.83617854e-01
-1.24430406e+00 2.25922823e-01 6.14969194e-01 7.81594694e-01
1.19673502e+00 2.55897552e-01 -3.57386053e-01 7.13362277e-01
1.49802193e-01 4.34846252e-01 -2.53463052e-02 -1.84318399e+00
7.06094444e-01 2.53324509e-01 2.36445174e-01 -1.08967304e+00
-4.69461501e-01 -3.48449588e-01 -8.66716087e-01 4.62901682e-01
9.44770515e-01 -9.19139832e-02 -1.08087111e+00 1.41836071e+00
5.33123910e-01 8.14590991e-01 -5.11465520e-02 1.02345014e+00
9.59640861e-01 6.26258552e-01 -8.15140083e-02 -1.79923147e-01
9.74660695e-01 -1.41433012e+00 -8.63719583e-01 -5.97239554e-01
6.12792611e-01 -9.18668211e-01 5.77958584e-01 2.80766189e-01
-1.27046919e+00 -8.09001982e-01 -8.41545701e-01 -5.33713222e-01
3.94336507e-02 -5.64150959e-02 1.06741667e+00 6.77190840e-01
-1.16667283e+00 5.84122181e-01 -8.79101813e-01 6.77351356e-02
7.57071495e-01 3.39593560e-01 -1.97892994e-01 -3.83384466e-01
-9.34043050e-01 7.34948516e-01 1.98063135e-01 4.38602775e-01
-7.14846849e-01 -8.01997721e-01 -1.21682584e+00 -1.89139634e-01
-1.31487371e-02 -1.08036518e+00 8.55398893e-01 -4.66565967e-01
-1.39041376e+00 8.84103954e-01 -6.49062932e-01 -4.60752130e-01
9.74569798e-01 -2.91307986e-01 4.48879376e-02 6.69194341e-01
1.05151519e-01 1.39493394e+00 4.47150379e-01 -1.04424989e+00
-1.13235247e+00 -5.49765639e-02 2.60854185e-01 1.76898882e-01
-9.80889890e-03 -2.88372070e-01 -8.30050230e-01 -2.47054368e-01
1.00957118e-01 -6.48205936e-01 -3.41559738e-01 5.78592479e-01
-3.78545225e-01 -4.04778868e-02 9.46804464e-01 -6.41156852e-01
1.34890628e+00 -1.78960097e+00 -8.07419345e-02 -1.65168449e-01
2.87283897e-01 3.94419789e-01 -1.77050531e-01 -3.22401702e-01
2.75548011e-01 -1.87299326e-01 -3.27346116e-01 -5.67877293e-01
-2.07734495e-01 3.56304526e-01 -1.08181886e-01 6.76901579e-01
2.21568748e-01 9.59089041e-01 -8.98810089e-01 -8.64857435e-01
1.04687774e+00 7.38551974e-01 -6.90465987e-01 3.09261885e-02
-1.32345762e-02 9.21041131e-01 -3.97906482e-01 5.97432315e-01
1.06803751e+00 -1.93910161e-03 -3.56555462e-01 -1.86148286e-01
-2.60642052e-01 2.20311135e-01 -1.36383200e+00 1.97427440e+00
-3.73457462e-01 1.40621841e+00 -6.11994639e-02 -6.28775358e-01
5.28202534e-01 -2.22492721e-02 7.23567128e-01 -6.31500840e-01
-2.86009721e-03 1.33868694e-01 -1.19835183e-01 -5.54785967e-01
5.84327996e-01 3.91660005e-01 6.72349811e-01 1.00818863e-02
-2.62153000e-01 -2.29114905e-01 6.98756993e-01 2.73161884e-02
8.31360281e-01 3.59020442e-01 1.09253980e-01 -2.09330127e-01
9.90076363e-01 2.96175592e-02 7.72235572e-01 6.75314903e-01
-6.41665399e-01 8.91442895e-01 5.15364170e-01 -7.20822215e-01
-7.77979553e-01 -1.01879478e+00 -3.58561426e-01 5.44035316e-01
6.57907069e-01 -3.77412736e-02 -6.08195961e-01 -3.94406348e-01
-9.60574225e-02 2.91833371e-01 -4.98981744e-01 4.73878473e-01
-1.12333572e+00 -5.19733429e-01 3.08647394e-01 7.59921908e-01
8.43528688e-01 -8.64439011e-01 -7.33943939e-01 3.62629145e-01
-6.11531258e-01 -1.79354286e+00 -4.49352801e-01 -3.88928950e-01
-1.07127798e+00 -1.22297776e+00 -1.04761624e+00 -8.19374919e-01
4.63522673e-01 4.89127666e-01 1.39813590e+00 7.47376978e-02
-4.84320760e-01 7.75109306e-02 1.22401133e-01 1.47976473e-01
4.96215038e-02 1.88948259e-01 -7.13027179e-01 -1.73032835e-01
6.89963549e-02 -4.75768298e-01 -1.29638457e+00 5.58339000e-01
-7.04442918e-01 9.11379531e-02 -6.10980503e-02 5.07786274e-01
4.58368063e-01 -2.97129631e-01 3.00222333e-03 -7.23831177e-01
-2.32755259e-01 1.03106625e-01 -1.06551111e+00 -6.70539439e-02
-1.37151226e-01 -1.14049539e-01 1.56019107e-01 -1.87408134e-01
-1.21629500e+00 3.02202463e-01 -2.28192404e-01 -6.35790706e-01
-2.89062262e-01 -2.73706555e-01 -8.28911085e-03 -3.73952568e-01
4.00594562e-01 -2.41394669e-01 -3.17595929e-01 -1.03613652e-01
4.08112258e-01 2.76169449e-01 7.20045090e-01 -2.81426311e-01
5.56146085e-01 1.30085111e+00 3.50219786e-01 -7.91987300e-01
-8.02076459e-01 -9.80259895e-01 -8.77475619e-01 -3.40754837e-01
1.15083230e+00 -1.07144010e+00 -7.57084489e-01 7.97534168e-01
-1.57106316e+00 -5.10281682e-01 -1.81484476e-01 7.43488550e-01
-5.99966049e-01 6.00963116e-01 -9.82775986e-01 -7.92575359e-01
-1.06947623e-01 -1.38801551e+00 1.10385311e+00 3.98931086e-01
4.64442484e-02 -1.52275574e+00 8.13003555e-02 4.76483762e-01
3.07691723e-01 4.35480982e-01 3.63674968e-01 4.47449833e-01
-1.30988252e+00 1.29336745e-01 -6.23815894e-01 2.21659452e-01
9.41982716e-02 2.76395112e-01 -1.33836150e+00 -4.29815706e-03
-3.37626994e-01 1.91485986e-01 1.33402097e+00 1.12605762e+00
9.92743909e-01 3.09993029e-01 -3.76918733e-01 1.18104386e+00
1.54248428e+00 5.80122098e-02 1.10013855e+00 3.20940316e-01
1.17927408e+00 9.65636551e-01 7.05401659e-01 3.11003208e-01
4.66522872e-01 7.35435426e-01 4.34331834e-01 -4.32642192e-01
-7.09872842e-01 3.04517224e-02 -1.04636420e-02 3.49697769e-01
-1.42006958e-02 -3.58316422e-01 -7.29994118e-01 7.17041016e-01
-1.93404138e+00 -8.27497363e-01 -9.52844262e-01 2.07733107e+00
5.47418058e-01 1.21795937e-01 3.77576351e-02 3.95686999e-02
8.27297151e-01 5.81174076e-01 -5.52113175e-01 -2.08793223e-01
-4.46392894e-01 1.30431533e-01 7.61514425e-01 1.27149653e+00
-1.34778821e+00 1.36834300e+00 7.06087875e+00 7.52796829e-01
-9.23861563e-01 1.01373658e-01 8.94575357e-01 -1.21116720e-01
-2.27978289e-01 5.60069270e-02 -1.11112654e+00 3.36940408e-01
2.79435694e-01 3.05615872e-01 7.28685558e-02 5.35430670e-01
4.72950697e-01 -8.06615293e-01 -1.02588165e+00 1.22489917e+00
-9.19782966e-02 -1.61194575e+00 -2.96786427e-01 1.08648680e-01
1.16643667e+00 1.41091079e-01 -4.85968329e-02 -2.62770474e-01
5.92859127e-02 -8.06841969e-01 3.81803036e-01 1.01572543e-01
6.24520957e-01 -5.15891194e-01 6.94874465e-01 -1.28596783e-01
-1.49866116e+00 2.77306676e-01 -6.11532152e-01 -2.64783889e-01
8.20496976e-01 9.49547231e-01 -1.86292514e-01 3.32004011e-01
4.03090119e-01 1.18794572e+00 -4.58878011e-01 1.52983046e+00
-6.28362358e-01 1.59422576e-01 -3.34616989e-01 5.06504178e-01
4.64600742e-01 -3.39166582e-01 3.53291988e-01 1.42892957e+00
-7.13656247e-02 -2.49320164e-01 1.18546538e-01 8.90720904e-01
4.69247214e-02 -6.50836304e-02 -5.27307570e-01 7.63882637e-01
1.42546564e-01 1.11944675e+00 -9.61803496e-01 -5.68060637e-01
-4.91381824e-01 9.26774740e-01 7.62073621e-02 8.01499546e-01
-8.34929883e-01 -1.15858860e-01 1.19137418e+00 7.51783624e-02
4.36781138e-01 -3.08109611e-01 -3.53824645e-01 -1.35066342e+00
7.62443542e-02 3.47568616e-02 1.91369891e-01 -5.60626805e-01
-9.35471475e-01 4.20890123e-01 2.33272277e-02 -1.37118077e+00
-5.00944436e-01 -5.97922444e-01 -5.67498982e-01 8.51043165e-01
-2.38010979e+00 -5.79372466e-01 -7.29835510e-01 5.13850868e-01
7.72735894e-01 4.54482675e-01 1.49237394e-01 7.65479863e-01
-6.18621588e-01 1.89173147e-01 -1.17485933e-01 4.06973958e-01
6.53604627e-01 -1.02918267e+00 6.57433987e-01 1.17711115e+00
1.17620297e-01 -2.35930420e-02 3.44697177e-01 -4.15075660e-01
-7.79957175e-01 -9.96429443e-01 7.96526313e-01 -3.06151092e-01
4.06142503e-01 -1.30454943e-01 -8.14539731e-01 4.19768482e-01
1.76422998e-01 6.04554176e-01 1.53195783e-01 -3.71277750e-01
-2.93152817e-02 -1.60017192e-01 -9.49569345e-01 4.09341365e-01
1.27693379e+00 -3.17786098e-01 -3.96734811e-02 2.66507596e-01
4.87134695e-01 -8.51336658e-01 -3.58638227e-01 4.70083624e-01
7.01202393e-01 -1.46878254e+00 1.43691695e+00 -2.60191470e-01
5.15563309e-01 -4.46620077e-01 1.63193628e-01 -7.47465014e-01
-7.94134736e-02 -6.31029963e-01 -1.91298753e-01 1.02303469e+00
5.89999482e-02 -4.71879512e-01 1.36945319e+00 7.28383183e-01
-7.41485208e-02 -4.24915612e-01 -1.05419946e+00 -6.92570269e-01
-1.36318116e-03 -6.53517008e-01 2.66233869e-02 9.31828499e-01
-5.94086289e-01 6.07065260e-02 -3.44070196e-01 -1.90814510e-02
9.11782742e-01 1.43712103e-01 8.38894427e-01 -9.64096844e-01
1.68577321e-02 -6.23897731e-01 -6.98844552e-01 -1.74956739e+00
4.69770491e-01 -3.89613122e-01 6.37924522e-02 -1.83304989e+00
-2.60122955e-01 -2.93514192e-01 -4.67405878e-02 2.17413325e-02
-4.46375877e-01 6.23211205e-01 2.43135065e-01 1.45229295e-01
-5.96514106e-01 3.12681884e-01 1.71951425e+00 -1.72373682e-01
-4.72762108e-01 -1.06604099e-01 6.30835295e-02 9.78085637e-01
7.16173112e-01 -2.41561353e-01 -4.83090550e-01 -8.18535507e-01
-2.58716941e-01 1.93531483e-01 3.91876042e-01 -1.06716156e+00
3.72420460e-01 -9.23884381e-03 4.58706051e-01 -9.71121550e-01
6.36052132e-01 -4.39056695e-01 -2.17902079e-01 5.81624329e-01
-4.54645902e-02 -2.18750849e-01 2.36942485e-01 3.71247888e-01
-4.59376484e-01 -2.60540217e-01 8.83184314e-01 -6.44645840e-02
-1.18829644e+00 5.23654163e-01 -2.39699692e-01 5.02135396e-01
8.22024465e-01 -7.11401284e-01 -5.83040595e-01 -4.12115604e-01
-5.53559244e-01 3.69781464e-01 3.69385511e-01 2.48964891e-01
7.17364490e-01 -1.06781304e+00 -7.68499672e-01 1.33281022e-01
-1.70334995e-01 2.93318957e-01 3.99957716e-01 9.37596142e-01
-1.33253622e+00 5.69881916e-01 -2.40519941e-01 -1.08422959e+00
-1.05194366e+00 2.67482460e-01 4.13971066e-01 -1.94397599e-01
-5.87451875e-01 9.59198654e-01 7.97343791e-01 1.67933792e-01
5.11838913e-01 -6.82274580e-01 -2.40872473e-01 -5.24992906e-02
6.41672373e-01 7.73505092e-01 -3.08739811e-01 -5.81340551e-01
-4.58849907e-01 1.18114054e+00 2.47450829e-01 -1.05142675e-01
6.99126542e-01 -4.66156125e-01 -1.09228361e-02 1.88083336e-01
1.40430725e+00 -2.45852947e-01 -1.86255944e+00 -2.41273046e-02
-2.84965515e-01 -1.24874043e+00 2.81688839e-01 -2.28931621e-01
-1.70933974e+00 1.30474985e+00 6.75421536e-01 -2.37465858e-01
1.04107571e+00 -2.82357246e-01 1.04593623e+00 -6.65419698e-02
2.01323599e-01 -9.85665321e-01 -7.74054080e-02 4.54934686e-01
1.96417600e-01 -1.53207505e+00 9.81823429e-02 -1.25918055e+00
-2.68549770e-01 1.12659872e+00 7.72078574e-01 -5.43530621e-02
4.80694473e-01 3.14668745e-01 3.07426184e-01 2.09549487e-01
-4.80562329e-01 -6.21350884e-01 3.94071221e-01 8.84161830e-01
4.41311210e-01 -3.92868310e-01 -2.86336333e-01 -6.93925977e-01
2.00478852e-01 3.01497113e-02 5.54745853e-01 5.34415603e-01
-4.73256946e-01 -1.09656692e+00 -3.07111382e-01 3.67122218e-02
-4.28752154e-01 -1.85732275e-01 8.25615227e-02 9.15101588e-01
1.31537579e-02 1.11328876e+00 4.41069931e-01 2.76864469e-01
1.61134824e-01 -5.36861539e-01 7.10074842e-01 -3.03004295e-01
-3.45792592e-01 3.14120531e-01 2.70110160e-01 -1.04804337e+00
-9.14392889e-01 -3.71791065e-01 -1.25851679e+00 -5.09370685e-01
-2.60284573e-01 -2.71848977e-01 5.12973666e-01 7.12297678e-01
1.10167034e-01 4.75263804e-01 5.31473815e-01 -1.04417741e+00
5.19824922e-01 -5.64759314e-01 -3.54450643e-01 4.24711496e-01
5.40033400e-01 -5.57958722e-01 -5.69119930e-01 1.03912026e-01] | [8.757762908935547, -1.8818602561950684] |
7224a21c-ee63-4746-b2d7-2b1cfc0443e1 | hopfield-networks-is-all-you-need | 2008.02217 | null | https://arxiv.org/abs/2008.02217v3 | https://arxiv.org/pdf/2008.02217v3.pdf | Hopfield Networks is All You Need | We introduce a modern Hopfield network with continuous states and a corresponding update rule. The new Hopfield network can store exponentially (with the dimension of the associative space) many patterns, retrieves the pattern with one update, and has exponentially small retrieval errors. It has three types of energy minima (fixed points of the update): (1) global fixed point averaging over all patterns, (2) metastable states averaging over a subset of patterns, and (3) fixed points which store a single pattern. The new update rule is equivalent to the attention mechanism used in transformers. This equivalence enables a characterization of the heads of transformer models. These heads perform in the first layers preferably global averaging and in higher layers partial averaging via metastable states. The new modern Hopfield network can be integrated into deep learning architectures as layers to allow the storage of and access to raw input data, intermediate results, or learned prototypes. These Hopfield layers enable new ways of deep learning, beyond fully-connected, convolutional, or recurrent networks, and provide pooling, memory, association, and attention mechanisms. We demonstrate the broad applicability of the Hopfield layers across various domains. Hopfield layers improved state-of-the-art on three out of four considered multiple instance learning problems as well as on immune repertoire classification with several hundreds of thousands of instances. On the UCI benchmark collections of small classification tasks, where deep learning methods typically struggle, Hopfield layers yielded a new state-of-the-art when compared to different machine learning methods. Finally, Hopfield layers achieved state-of-the-art on two drug design datasets. The implementation is available at: https://github.com/ml-jku/hopfield-layers | ['Thomas Adler', 'Sepp Hochreiter', 'Günter Klambauer', 'Milena Pavlović', 'Markus Holzleitner', 'Bernhard Schäfl', 'Victor Greiff', 'Michael Widrich', 'Michael Kopp', 'Johannes Brandstetter', 'Hubert Ramsauer', 'Lukas Gruber', 'Philipp Seidl', 'Johannes Lehner', 'Geir Kjetil Sandve', 'David Kreil'] | 2020-07-16 | null | https://openreview.net/forum?id=tL89RnzIiCd | https://openreview.net/pdf?id=tL89RnzIiCd | iclr-2021-1 | ['immune-repertoire-classification'] | ['medical'] | [-1.00704193e-01 -1.24212034e-01 -1.74330696e-02 -7.29412064e-02
-2.46898830e-01 -4.70274240e-01 7.01700628e-01 2.51313329e-01
-6.95414186e-01 8.16184580e-01 9.03387964e-02 3.17505449e-02
-3.88958603e-01 -9.72419560e-01 -9.82375979e-01 -1.05832887e+00
-5.82911849e-01 7.38969386e-01 4.70876634e-01 -2.49984488e-01
3.04026127e-01 6.46187305e-01 -1.58859360e+00 5.85824370e-01
3.60270113e-01 1.34455216e+00 1.15690701e-01 5.19548893e-01
4.82983363e-04 7.80457854e-01 -4.28461879e-01 -2.48634622e-01
4.01970521e-02 -3.39064980e-03 -1.00629270e+00 -7.67425954e-01
2.95799464e-01 -9.97662172e-02 -6.45662546e-01 8.31985176e-01
8.22926044e-01 2.87657827e-01 6.10792220e-01 -8.15112293e-01
-1.10794687e+00 5.23894250e-01 1.91944927e-01 7.00538456e-01
2.92469617e-02 4.55312699e-01 1.02436256e+00 -1.16990304e+00
7.52224267e-01 1.05365372e+00 8.70108187e-01 5.57400167e-01
-1.25474966e+00 -6.50701761e-01 -2.52341684e-02 4.15089130e-01
-1.40103257e+00 -4.72512752e-01 4.08804156e-02 -2.93574661e-01
2.00427842e+00 -2.40222868e-02 8.81286979e-01 1.19317400e+00
7.47994125e-01 7.10020423e-01 9.45026696e-01 -7.06446692e-02
3.14816296e-01 -2.67953038e-01 5.14109254e-01 9.55374122e-01
1.82161286e-01 2.38021180e-01 -8.57852399e-01 -2.07832322e-01
7.44434893e-01 4.92271483e-01 -4.47681546e-01 1.08544193e-01
-1.19193017e+00 5.96985996e-01 9.96812344e-01 5.14877856e-01
-5.26178181e-01 4.07577425e-01 1.46341905e-01 4.75574881e-01
2.31807202e-01 8.45399857e-01 -5.41819870e-01 3.48066390e-01
-6.91791952e-01 2.10411489e-01 8.10064375e-01 6.55856371e-01
6.97212219e-01 -1.43056795e-01 -3.94391447e-01 5.61421752e-01
-2.22396990e-03 4.09483969e-01 8.64638269e-01 -5.18492460e-01
5.79201095e-02 5.39314032e-01 -6.17675036e-02 -5.04841149e-01
-8.33431065e-01 -7.60987878e-01 -1.11498845e+00 6.05074167e-02
3.51596832e-01 1.64363265e-01 -1.18958867e+00 1.85932672e+00
-3.27666104e-01 9.20242220e-02 1.88368797e-01 7.97596157e-01
8.53148818e-01 6.87698305e-01 1.69675946e-01 -1.29854366e-01
1.46672821e+00 -8.04760098e-01 -6.20455205e-01 -2.00767279e-01
4.29462850e-01 -4.10407662e-01 8.21781158e-01 3.62995058e-01
-1.42510450e+00 -3.24848682e-01 -1.12040591e+00 -2.34890342e-01
-1.09323668e+00 -3.29732716e-01 6.79471970e-01 -4.75781523e-02
-1.49332106e+00 1.17933178e+00 -7.71329999e-01 -2.74062663e-01
5.85924804e-01 9.50092435e-01 -3.93066019e-01 5.38395904e-02
-1.70152664e+00 1.08224738e+00 3.13499779e-01 2.33604282e-01
-1.06370151e+00 -9.81342614e-01 -4.92699385e-01 4.12861079e-01
-2.08511487e-01 -9.53742325e-01 9.91066575e-01 -7.65384316e-01
-1.27147257e+00 6.98548138e-01 -1.53201073e-01 -8.21487308e-01
4.40848097e-02 -1.48364186e-01 -2.90920436e-01 -2.92247478e-02
-3.37790936e-01 9.40913022e-01 8.12966585e-01 -3.34761024e-01
-2.34805956e-01 -4.85264540e-01 -2.99813122e-01 -3.46360207e-02
-2.62845278e-01 -3.01373631e-01 -1.41131431e-01 -5.32126129e-01
1.79804280e-01 -8.93446445e-01 -3.98593955e-02 -1.72898069e-01
-2.93200333e-02 -5.89979887e-01 4.52601880e-01 -3.76609057e-01
1.18018794e+00 -2.15537119e+00 4.92202371e-01 1.63019523e-01
3.45884472e-01 1.41005516e-01 -2.30219334e-01 3.65534276e-01
-3.20370972e-01 1.79711774e-01 -2.20808029e-01 -2.51831561e-02
-1.28148431e-02 7.02809691e-02 -4.16033775e-01 4.62697923e-01
5.14115334e-01 1.27775061e+00 -9.48440492e-01 1.44383594e-01
-3.24735314e-01 7.75192320e-01 -5.86805820e-01 1.52601087e-02
-1.58329725e-01 -1.38925612e-01 -6.50581196e-02 5.00863910e-01
3.68345231e-01 -7.39110827e-01 9.49048158e-03 -2.18031839e-01
-8.55071917e-02 5.69400847e-01 -7.14496553e-01 1.68308926e+00
-7.29087815e-02 4.79669541e-01 -5.25968134e-01 -6.53670609e-01
6.08762026e-01 4.28972900e-01 1.61112145e-01 -8.38451445e-01
1.11574017e-01 4.38053221e-01 2.31535152e-01 -1.27974615e-01
2.36953795e-01 -3.60516608e-01 -8.54006559e-02 5.70002675e-01
8.97724628e-01 3.55711371e-01 7.83933997e-02 9.59808528e-02
1.42782509e+00 -3.49934489e-01 7.43549094e-02 -5.73199809e-01
2.71441579e-01 -3.31623107e-01 2.95327753e-01 8.30743968e-01
-6.48624972e-02 4.02173430e-01 5.12090743e-01 -9.65159416e-01
-9.02335823e-01 -1.20356941e+00 -3.33147228e-01 1.06494331e+00
-1.78258792e-01 -3.19469959e-01 -5.46036661e-01 -1.34056434e-01
7.05067813e-02 1.80538595e-01 -1.05116189e+00 -7.02599704e-01
-3.65439057e-01 -1.05238914e+00 8.54541659e-01 5.77857316e-01
7.47207761e-01 -1.54518533e+00 -5.98342836e-01 3.40599209e-01
6.37067705e-02 -5.05434990e-01 -3.35047692e-01 8.58029485e-01
-9.38683331e-01 -9.86297667e-01 -7.66100645e-01 -7.95573533e-01
4.33943540e-01 -4.47240889e-01 1.15180826e+00 1.47217304e-01
-3.32438707e-01 -8.65123868e-02 1.80894569e-01 -3.44799668e-01
2.03795526e-02 5.20791292e-01 3.99154693e-01 -2.20837176e-01
3.83738130e-01 -6.32841766e-01 -9.24270749e-01 1.30416512e-01
-8.15436423e-01 -4.60303634e-01 8.78306627e-01 1.16751802e+00
7.10106194e-01 -2.85173416e-01 6.27356887e-01 -5.56307733e-01
6.33446991e-01 -4.74718511e-01 -4.83950227e-01 2.20930025e-01
-6.16699517e-01 7.16053665e-01 5.30824661e-01 -5.61349392e-01
-5.15792608e-01 -2.40339369e-01 -4.46238741e-02 -5.65224946e-01
2.46767908e-01 3.93664300e-01 3.49114478e-01 -5.60935140e-02
7.71251738e-01 4.12823468e-01 4.44314294e-02 -4.16787475e-01
8.73636156e-02 2.77539074e-01 3.29044074e-01 -3.69462758e-01
-1.21806063e-01 2.06287280e-01 -4.63565476e-02 -4.27560419e-01
-6.66809142e-01 -1.43858314e-01 -5.41495204e-01 9.61613376e-03
9.27105725e-01 -6.80334568e-01 -1.06222332e+00 9.04342532e-01
-1.12466538e+00 -6.30963206e-01 -3.69120389e-01 2.59352028e-01
-4.66888785e-01 -4.00023460e-01 -1.21586990e+00 -3.29523921e-01
-7.47179866e-01 -1.11355686e+00 7.09019661e-01 2.46802568e-01
-1.44053951e-01 -9.27842081e-01 2.14593172e-01 -4.37125742e-01
8.02972436e-01 -2.64873177e-01 1.21205080e+00 -7.57736087e-01
-7.27084577e-01 -1.66472897e-01 -2.03670375e-02 3.67543474e-02
-2.64221072e-01 -3.92881393e-01 -1.17882574e+00 -4.11317289e-01
-2.91262209e-01 -4.04820085e-01 1.65154147e+00 3.36556613e-01
1.09345412e+00 -5.31353593e-01 -4.94740158e-01 6.51697934e-01
1.39199615e+00 3.37985568e-02 9.55491722e-01 1.14671148e-01
2.12336227e-01 2.16556489e-01 -3.08649480e-01 1.74859211e-01
-4.95259650e-02 3.47377002e-01 2.72765368e-01 2.46713609e-01
-1.22387439e-01 -4.47102031e-03 3.56805861e-01 8.46700609e-01
-3.71007085e-01 -4.80554439e-02 -1.03011572e+00 5.25215030e-01
-1.77428162e+00 -1.31336927e+00 3.70419472e-01 2.32828856e+00
9.88419414e-01 1.79778472e-01 -1.23567283e-01 9.06499859e-04
5.14422178e-01 3.12857516e-02 -1.00848782e+00 -5.38408160e-01
-3.49230230e-01 8.06891322e-01 5.79208314e-01 4.47249711e-01
-1.07108212e+00 9.23902810e-01 6.85840130e+00 6.43304229e-01
-1.36767006e+00 2.50312179e-01 6.48255885e-01 -3.89532983e-01
-4.23974097e-02 -5.33980012e-01 -9.08336103e-01 3.39599729e-01
1.40724111e+00 1.45655155e-01 8.36938798e-01 4.18787152e-01
-4.99305844e-01 2.13351354e-01 -1.17121661e+00 8.99915814e-01
-3.08938295e-01 -1.84170043e+00 2.44070053e-01 9.85265002e-02
6.38640821e-01 8.27287436e-01 3.86058778e-01 4.81625646e-01
2.91946113e-01 -1.38335538e+00 6.22594535e-01 9.69570935e-01
8.46478581e-01 -6.18438900e-01 8.12495768e-01 1.08476672e-02
-9.17052627e-01 -3.94302577e-01 -4.87425506e-01 -4.01216745e-02
-4.33919489e-01 4.46764708e-01 -4.49845701e-01 9.36704203e-02
1.11811495e+00 9.08935964e-01 -5.74271739e-01 8.93182397e-01
3.94131877e-02 2.19307572e-01 -4.55803066e-01 -3.02047044e-01
3.33574951e-01 2.38016993e-01 3.33963633e-01 1.23139799e+00
1.09844908e-01 2.75460899e-01 -4.00307775e-01 1.01370227e+00
-4.80637670e-01 -2.24155068e-01 -5.87183177e-01 -1.15821257e-01
4.02453423e-01 1.04800129e+00 -6.13618731e-01 -4.04398292e-01
1.26637265e-01 9.71464097e-01 7.06655025e-01 6.64497197e-01
-5.47442138e-01 -6.95129931e-01 8.17875266e-01 1.85654074e-01
5.51861227e-01 -3.10603622e-03 -5.14162071e-02 -9.71368790e-01
-2.39530519e-01 -6.35715067e-01 5.62228203e-01 -7.16787875e-01
-1.27426827e+00 9.16911244e-01 -4.05875444e-01 -6.04455352e-01
-8.39511752e-02 -1.00475383e+00 -4.63775814e-01 9.00487602e-01
-1.44926739e+00 -5.65107226e-01 1.12526596e-01 7.59145439e-01
9.76434946e-02 -1.57867014e-01 1.22002959e+00 3.24295551e-01
-5.82197309e-01 5.95806599e-01 1.32912129e-01 1.40313715e-01
5.59753656e-01 -1.12280905e+00 2.89734811e-01 2.76668876e-01
-1.27777874e-01 1.03084683e+00 3.23035538e-01 -2.91434050e-01
-1.40392339e+00 -1.01649058e+00 1.04567814e+00 -3.41749161e-01
7.08803892e-01 -3.74616593e-01 -1.31489933e+00 7.18802214e-01
2.01215237e-01 2.78721243e-01 5.61261117e-01 1.50947124e-01
-5.21956086e-01 -1.10866427e-01 -1.09113967e+00 3.42416912e-01
1.13386738e+00 -7.57480323e-01 -5.66377819e-01 3.46327960e-01
6.78504586e-01 -3.05715233e-01 -9.97080863e-01 3.92437696e-01
8.06488812e-01 -8.89990985e-01 8.60171139e-01 -1.06921685e+00
3.11405182e-01 -1.32774591e-01 5.13833612e-02 -1.25237203e+00
-7.88080752e-01 -6.24132752e-01 -4.90863055e-01 3.28741431e-01
8.71754229e-01 -1.10261440e+00 2.37472773e-01 3.88706326e-01
-1.96809962e-01 -1.32284141e+00 -9.64988112e-01 -6.13264143e-01
2.84957528e-01 -2.66801771e-02 6.27487659e-01 6.48220360e-01
1.77982762e-01 4.13995057e-01 1.80752054e-01 8.05572197e-02
3.66042048e-01 -3.50310616e-02 -2.80663192e-01 -1.29865181e+00
-2.56162852e-01 -9.23989952e-01 -6.16049111e-01 -7.57671237e-01
5.97347543e-02 -1.28339553e+00 -2.55433023e-01 -1.35623121e+00
2.42060527e-01 -3.69373411e-01 -1.04611516e+00 9.25848842e-01
7.45305493e-02 3.72368664e-01 -8.00163578e-03 5.76534390e-01
-6.87297702e-01 3.31976473e-01 9.81726348e-01 -3.02499563e-01
-8.30819681e-02 -3.52194875e-01 -4.92167473e-01 2.90439785e-01
8.36231411e-01 -5.24239421e-01 -2.58423351e-02 -4.31930661e-01
5.89186966e-01 -3.77404064e-01 5.15930891e-01 -1.09207523e+00
4.58639681e-01 5.65658629e-01 8.91349733e-01 -2.17305899e-01
3.54899108e-01 -3.36105853e-01 2.06460133e-01 9.62220252e-01
-4.67517138e-01 3.04152876e-01 4.16923642e-01 4.76059258e-01
9.23075899e-03 9.20250639e-02 8.03292274e-01 -3.21929038e-01
-5.55715263e-01 4.84861463e-01 -5.25069296e-01 -4.26798798e-02
5.46963871e-01 1.16297036e-01 -7.30616093e-01 1.77245513e-02
-1.20738506e+00 1.62688777e-01 1.90266818e-01 4.16864634e-01
6.62039936e-01 -1.30503154e+00 -3.40535462e-01 5.49385846e-01
-1.72542349e-01 -1.13559693e-01 8.46185982e-02 1.05440891e+00
-3.45815241e-01 8.02535772e-01 -4.62457240e-01 -5.93571246e-01
-5.75122654e-01 7.41700411e-01 9.39363658e-01 -5.37191272e-01
-3.38756323e-01 9.56568897e-01 1.23349324e-01 -2.81790853e-01
2.98586369e-01 -6.14184320e-01 -5.22535779e-02 2.01035514e-01
7.81336606e-01 1.22393787e-01 4.67131227e-01 -1.64305478e-01
-6.12189233e-01 5.14232993e-01 -2.51620471e-01 1.09016269e-01
1.58237016e+00 4.66673821e-01 -3.96643877e-01 6.51876986e-01
1.18672621e+00 -8.16239893e-01 -1.02039683e+00 -2.49150828e-01
9.74890310e-03 4.09190148e-01 2.15763822e-01 -1.06342196e+00
-9.96849000e-01 1.11547792e+00 8.54902565e-01 1.27482578e-01
9.77192461e-01 1.51869968e-01 7.27772832e-01 8.43111575e-01
4.01824445e-01 -7.55690813e-01 1.14600778e-01 9.82788444e-01
1.10067904e+00 -8.42366397e-01 -2.64006019e-01 4.67067301e-01
-2.01110348e-01 1.09722567e+00 3.63039285e-01 -3.30071419e-01
1.06729090e+00 3.52979094e-01 -4.25109088e-01 -6.17381215e-01
-1.37404907e+00 -2.74598897e-01 4.69385862e-01 1.09408036e-01
5.23312390e-01 -9.79838818e-02 6.92791790e-02 8.34919274e-01
1.11303672e-01 1.23230144e-01 -2.92525422e-02 7.76060700e-01
-5.37669599e-01 -6.12736046e-01 1.16647929e-01 6.12217009e-01
-5.65404892e-01 -6.62871599e-01 -3.48824531e-01 5.64182639e-01
-1.18431961e-02 2.59917974e-01 4.07088101e-01 -3.72479141e-01
3.41014355e-01 7.05537975e-01 7.06770957e-01 -3.27457309e-01
-1.03886533e+00 -2.47433841e-01 -3.81083071e-01 -7.33889759e-01
-9.27718952e-02 -4.26523864e-01 -1.49334490e+00 -5.29732049e-01
-2.83086061e-01 1.47270024e-01 2.42279664e-01 8.50282133e-01
9.11872864e-01 5.83782196e-01 9.25518498e-02 -1.01452625e+00
-6.32363021e-01 -1.13967383e+00 -5.27455509e-01 2.00876668e-01
6.47776186e-01 -6.27360463e-01 -4.04730529e-01 -1.76869988e-01] | [8.7337007522583, 3.2767155170440674] |
f11054ed-40ef-4014-8903-547373bc2f9f | recursive-tree-attention-improving-semantic | null | null | https://openreview.net/forum?id=NH-gIZJ8mOD | https://openreview.net/pdf?id=NH-gIZJ8mOD | Recursive Tree Attention: Improving Semantic Representations with Syntactic Tree Structured Attention Mechanism | Attention mechanism has shown its effectiveness in state-of-the-art methods on various tasks in natural language processing (NLP). However, these methods are still using attention mechanism in plain, linear topological structures while the syntactic structures of natural languages are known to be hierarchical. Modeling in such syntactic structures like syntactic trees is proved to be superior in semantic representation learning. In this paper, to improve semantic representation learning by modelling attention mechanism in syntactic structures, we propose recursive tree attention, a syntactic tree structured attention mechanism which recursively calculate attention weights and summarize information through constituency parsing trees. The information of the children is cursively summarized to their parent through the constituency parsing tree. Then the summarized information from the root node is used as the semantic representations of the current sentence. Experimental results on the text classification tasks demonstrate the effectiveness of our proposed attention mechanism. The approaches with recursive tree attention outperform conventional attention mechanism and other syntactic tree based approaches. | ['Anonymous'] | 2021-06-04 | null | null | null | null | ['constituency-parsing'] | ['natural-language-processing'] | [-6.73711896e-02 6.40363038e-01 -1.55801192e-01 -4.19018775e-01
-2.76117772e-01 -2.20209673e-01 2.07987368e-01 5.46180665e-01
-2.15688318e-01 5.50569594e-01 8.89250875e-01 -3.91554892e-01
-7.62148798e-02 -1.03553212e+00 -5.20228863e-01 -4.42120790e-01
-1.06061764e-01 4.43253040e-01 3.64155412e-01 -2.90207714e-01
6.76265240e-01 7.93859437e-02 -1.21259880e+00 4.51671779e-01
8.97454858e-01 7.02750802e-01 6.58897817e-01 6.67607605e-01
-1.22569716e+00 1.04162383e+00 -5.54710507e-01 -2.63367116e-01
-4.37754065e-01 -2.42124125e-01 -1.45372820e+00 -3.72150362e-01
2.18448654e-01 1.23530040e-02 3.68507989e-02 1.11750388e+00
1.00423671e-01 3.04406881e-01 5.39969206e-01 -6.89951658e-01
-1.27190065e+00 9.84766960e-01 -7.00471401e-01 5.21441102e-01
3.24188262e-01 -7.27703571e-01 1.36583793e+00 -1.01507735e+00
5.76750696e-01 1.86985445e+00 5.78204930e-01 4.46691036e-01
-6.70790195e-01 -3.41165990e-01 8.39492381e-01 5.04329801e-01
-6.97660148e-01 2.40595758e-01 1.00313997e+00 -3.89655590e-01
1.27014530e+00 -1.59440696e-01 1.71694413e-01 3.72202784e-01
5.85277617e-01 8.16971898e-01 7.29309082e-01 -5.83251774e-01
6.21497743e-02 -1.60968617e-01 1.24266577e+00 8.13470602e-01
2.18892664e-01 -6.17860854e-01 -1.90054640e-01 -1.71528216e-02
4.19450313e-01 -1.63670033e-01 -5.30370884e-02 -6.18931800e-02
-6.96393430e-01 1.04402626e+00 9.83388722e-01 7.59714365e-01
-6.00979388e-01 4.26165879e-01 6.98859215e-01 -9.57424119e-02
6.48591340e-01 1.70252264e-01 -6.95429087e-01 2.14428499e-01
-1.34180993e-01 -1.14231082e-02 4.74232823e-01 9.04574096e-01
6.40818298e-01 -3.00671719e-02 -1.69219077e-01 9.12554204e-01
4.40944046e-01 -1.42760560e-01 7.85645187e-01 -5.26538432e-01
9.12458658e-01 1.12692332e+00 -2.33196884e-01 -1.24090660e+00
-6.25339329e-01 -1.95579290e-01 -7.51137733e-01 -8.05622116e-02
-6.12657219e-02 -2.02430934e-01 -1.01837325e+00 1.70067394e+00
3.37012321e-01 -2.32037142e-01 2.98160195e-01 7.28301167e-01
1.31628585e+00 1.05855095e+00 8.32483113e-01 -3.12574580e-02
1.80548644e+00 -1.07940209e+00 -9.05412316e-01 -3.16144153e-02
8.04556489e-01 -4.16327447e-01 1.06991804e+00 -1.59921467e-01
-9.73476648e-01 -6.70434177e-01 -7.56787956e-01 -6.08180940e-01
-6.25086963e-01 -7.47439340e-02 8.39556038e-01 2.09122211e-01
-8.16796899e-01 5.40078044e-01 -6.71580732e-01 -5.77734292e-01
5.88101804e-01 4.81517762e-01 -4.29007798e-01 7.59807602e-02
-1.15044057e+00 8.97459507e-01 7.17285156e-01 -5.21074682e-02
-4.57109123e-01 -4.69020933e-01 -1.10271943e+00 6.28068328e-01
1.90909892e-01 -7.33183682e-01 1.26044834e+00 -9.61995840e-01
-1.11706781e+00 6.58901691e-01 -4.27295089e-01 -5.39015412e-01
-3.15494329e-01 -4.58322108e-01 3.17330897e-01 1.52773008e-01
3.77966493e-01 6.81865931e-01 3.13778430e-01 -1.07475293e+00
-5.88245988e-01 -5.49151242e-01 2.07283318e-01 5.38743973e-01
-2.07415104e-01 3.94094557e-01 -2.70010270e-02 -6.44365489e-01
4.36411172e-01 -3.22178721e-01 -4.77980733e-01 -4.27024007e-01
-3.24539989e-01 -9.87606943e-01 1.12155759e+00 -7.97939897e-01
1.20806503e+00 -1.89992571e+00 2.85518557e-01 -3.94382775e-01
1.15560181e-01 9.82682854e-02 1.38360634e-01 2.55068243e-01
-1.84505120e-01 5.50627649e-01 -2.66634583e-01 -1.93571523e-01
-2.95682251e-01 2.49835834e-01 -2.83880353e-01 -7.43161440e-02
2.58797288e-01 9.86133873e-01 -9.18114245e-01 -7.89099932e-01
1.63166597e-01 2.27313831e-01 -6.59918189e-01 1.36761740e-01
-2.54476935e-01 3.88158798e-01 -9.72742260e-01 3.91126931e-01
5.51865578e-01 -5.28527573e-02 3.96630168e-02 -1.18733220e-01
-1.41207706e-02 7.32336283e-01 -6.70231879e-01 1.51100445e+00
-6.06096566e-01 4.42367911e-01 8.82415250e-02 -1.23636723e+00
1.04581881e+00 3.47434670e-01 4.86993194e-02 -2.93465108e-01
5.03398776e-01 -3.55918348e-01 2.33542949e-01 -5.80747366e-01
5.38523734e-01 -1.59669891e-01 -2.78241277e-01 4.08323020e-01
2.20353380e-01 1.01500459e-01 2.01894324e-02 3.89895797e-01
7.77813137e-01 1.27110541e-01 5.10219395e-01 -6.77118957e-01
8.98978412e-01 -2.58918665e-02 5.33752978e-01 4.80385244e-01
-1.88258141e-01 2.57423878e-01 7.45429277e-01 -6.48562014e-01
-9.97153699e-01 -6.45256341e-01 -6.32840097e-02 1.72836089e+00
-3.50873056e-03 -3.76035333e-01 -1.06780672e+00 -9.44644153e-01
-3.05414945e-01 7.89592624e-01 -9.63742077e-01 1.87264740e-01
-6.84547842e-01 -5.25974333e-01 1.75478414e-01 9.50978100e-01
8.40604246e-01 -1.75647330e+00 -6.64736688e-01 3.82927537e-01
-4.57985401e-02 -9.02019203e-01 -2.04060331e-01 1.96204901e-01
-1.11547363e+00 -9.92762744e-01 -5.44894397e-01 -1.34743845e+00
8.17622900e-01 1.38492078e-01 9.41809058e-01 2.81060725e-01
-1.71378225e-01 4.32866067e-02 -5.37409365e-01 -6.18969440e-01
-1.35946572e-01 2.00913608e-01 -2.45139629e-01 -1.67665511e-01
3.75789613e-01 -4.59057808e-01 -3.27325881e-01 -2.34485567e-01
-4.96244967e-01 1.15018703e-01 2.97665775e-01 7.70684063e-01
4.09686983e-01 1.97728947e-02 7.36215115e-01 -1.16532052e+00
8.83034170e-01 -6.83488786e-01 -3.08573157e-01 1.92028254e-01
-7.53764063e-02 5.15606999e-01 7.54900277e-01 7.23039955e-02
-1.59437776e+00 -1.88668102e-01 -2.04026118e-01 1.01573519e-01
-2.55389452e-01 7.52114773e-01 -1.02717362e-01 3.43489498e-01
8.05752799e-02 3.67868058e-02 -4.95576978e-01 -7.39021003e-01
4.24494028e-01 5.18703938e-01 4.76684481e-01 -9.37493682e-01
5.03494628e-02 1.86919197e-01 1.36885252e-02 -7.31561005e-01
-1.04761565e+00 -3.59788358e-01 -8.50815237e-01 3.08096677e-01
1.47439945e+00 -3.57530653e-01 -6.24770284e-01 1.57297343e-01
-1.64991963e+00 2.41432637e-01 1.52907252e-01 2.36248344e-01
-2.97429979e-01 5.24714828e-01 -7.84146369e-01 -8.31498861e-01
-7.97777593e-01 -7.93976307e-01 1.22723448e+00 3.22353184e-01
-1.84915558e-01 -1.38121974e+00 -6.91435784e-02 1.49491698e-01
2.64127672e-01 -1.52202155e-02 1.64038754e+00 -8.27128649e-01
-3.03742349e-01 8.93777311e-02 -4.79683936e-01 3.19571868e-02
1.31366834e-01 -8.92313123e-02 -7.98383117e-01 2.61746645e-01
9.23387110e-02 -2.48827152e-02 1.02955484e+00 4.54311520e-01
1.25517964e+00 -3.15145552e-01 -4.28206712e-01 2.20038354e-01
1.35146582e+00 3.15751731e-01 5.14796615e-01 3.43807399e-01
8.46591115e-01 1.13149595e+00 4.75427091e-01 2.22170316e-02
4.45724338e-01 1.49496913e-01 4.88581330e-01 1.33384570e-01
-3.50609533e-02 -2.69068539e-01 1.13425098e-01 9.79645431e-01
-3.33418772e-02 -1.59856796e-01 -1.06226373e+00 6.64766788e-01
-1.81262600e+00 -8.26644182e-01 -4.57516551e-01 1.61451340e+00
4.08518225e-01 2.18346491e-01 -5.29439330e-01 9.21304300e-02
1.03176701e+00 1.09186709e-01 -4.86504436e-02 -1.11992466e+00
1.26946166e-01 2.83764601e-01 1.46696478e-01 6.70307398e-01
-1.15922809e+00 1.44119751e+00 6.00469637e+00 4.30298626e-01
-6.44690752e-01 1.86896876e-01 5.97829819e-01 6.51346922e-01
-2.44696781e-01 1.76337123e-01 -9.31854248e-01 1.81550249e-01
6.72820508e-01 -4.26018536e-01 -2.68117607e-01 9.81694281e-01
1.29206643e-01 -2.37068728e-01 -8.00725698e-01 4.56417799e-01
-8.84236842e-02 -1.20369256e+00 5.36092818e-01 -3.51738513e-01
5.50791025e-01 -2.10134879e-01 -2.18615904e-01 4.39413369e-01
3.74959469e-01 -1.00465655e+00 2.13037685e-01 1.25688344e-01
1.86400875e-01 -8.08664382e-01 1.12394595e+00 4.41757083e-01
-1.55847907e+00 -2.63897419e-01 -7.47503102e-01 -4.11788791e-01
2.26994947e-01 1.24410562e-01 -6.65380299e-01 6.40588582e-01
8.55705142e-01 5.74779332e-01 -5.34445167e-01 5.04894257e-01
-4.97746468e-01 6.65684521e-01 -1.22007251e-01 -3.74202996e-01
7.69179761e-01 -7.98962861e-02 4.78561193e-01 1.36914706e+00
4.23064493e-02 4.59806561e-01 1.33082703e-01 7.03451276e-01
-1.64878324e-01 7.35113561e-01 -8.10364902e-01 1.67280376e-01
2.20677570e-01 1.08361268e+00 -8.51287305e-01 -6.97947502e-01
-4.26476806e-01 5.68051457e-01 7.21875727e-01 8.84608403e-02
-6.40121877e-01 -5.49853384e-01 2.11368665e-01 -3.53009365e-02
3.74288112e-01 -2.04772487e-01 -5.98980904e-01 -7.77869344e-01
-4.37924713e-02 -9.56089869e-02 7.61028528e-01 -1.11068654e+00
-1.02264690e+00 7.17038512e-01 1.93929344e-01 -4.67611611e-01
2.35293657e-01 -7.83402205e-01 -1.22980130e+00 9.87630367e-01
-1.17534196e+00 -1.15843272e+00 -2.39676923e-01 3.11146140e-01
1.16549087e+00 -4.64272015e-02 9.15464640e-01 -1.59165502e-01
-4.58294630e-01 1.18679509e-01 -2.21264303e-01 2.92956144e-01
-1.10458426e-01 -1.36748898e+00 4.71978575e-01 5.05510926e-01
-5.74697256e-02 9.48301435e-01 3.48254055e-01 -7.74742186e-01
-1.00114131e+00 -9.05096292e-01 1.22460437e+00 -2.49995917e-01
7.45522797e-01 -2.40433082e-01 -1.35233426e+00 9.34007108e-01
6.53738320e-01 -2.72021890e-01 3.99842024e-01 3.38786304e-01
-2.32577771e-01 3.27383637e-01 -1.01169538e+00 3.99530232e-01
8.63918424e-01 -1.50662974e-01 -1.51834464e+00 1.24117553e-01
1.04259920e+00 -1.23088323e-01 -5.90566456e-01 3.89855683e-01
1.66355178e-01 -5.24458826e-01 5.88729620e-01 -1.10924625e+00
6.14485741e-01 -1.64584592e-01 -1.36337206e-02 -1.14514530e+00
-6.80213749e-01 -1.09806582e-01 2.60630727e-01 1.53133190e+00
3.09163302e-01 -6.55048609e-01 8.08302522e-01 4.83182788e-01
-3.90221089e-01 -5.76052606e-01 -8.96866679e-01 -3.42423379e-01
5.02965689e-01 -2.04340056e-01 3.85608643e-01 8.00324917e-01
2.33485758e-01 1.02973044e+00 2.09542856e-01 2.23246273e-02
3.93646002e-01 6.39220476e-02 2.91265458e-01 -1.45484483e+00
1.98547885e-01 -5.46592653e-01 -4.32545185e-01 -9.47303712e-01
6.07519329e-01 -1.03693318e+00 9.44755599e-03 -2.07113123e+00
3.94536734e-01 -3.00925970e-01 -3.31793189e-01 5.14607608e-01
-5.40894866e-01 -6.32198989e-01 2.03494757e-01 -3.85783203e-02
-5.98121285e-01 6.35761559e-01 1.26949263e+00 -2.67391115e-01
-5.03259674e-02 -3.28561217e-01 -7.24837065e-01 1.13946092e+00
1.07823670e+00 -5.33101261e-01 -3.94466311e-01 -6.15909100e-01
-1.94269836e-01 -4.43141758e-02 -1.13488950e-01 -7.79727161e-01
2.78025240e-01 -5.01100384e-02 2.23502755e-01 -8.92673373e-01
-3.26613449e-02 -7.45830238e-01 -5.13179779e-01 4.74531591e-01
-5.42711556e-01 3.00148606e-01 3.51477563e-01 4.40866619e-01
-3.61877859e-01 -6.25976443e-01 6.81459486e-01 -5.56549370e-01
-8.65668535e-01 1.22984909e-01 -5.09470046e-01 1.33133635e-01
9.81108725e-01 -3.26307788e-02 -2.52519190e-01 -9.32369754e-02
-1.03084469e+00 3.97805870e-01 -1.30659342e-01 6.26176000e-01
5.62749147e-01 -1.15678084e+00 -6.55483007e-01 -1.56803876e-01
-1.07295081e-01 4.17242825e-01 3.04173082e-01 2.65770555e-01
-6.89427435e-01 8.26962590e-01 -2.17588842e-01 -3.97828937e-01
-1.47797823e+00 8.83582294e-01 1.62561238e-01 -5.08055985e-01
-7.45625496e-01 8.44030797e-01 9.77903664e-01 -3.34222704e-01
1.79248706e-01 -7.98820317e-01 -9.53535020e-01 -1.38972089e-01
4.60138083e-01 1.67716429e-01 -2.25877464e-01 -8.21394980e-01
-4.33058709e-01 1.13710082e+00 -3.06838721e-01 1.26304224e-01
1.24971497e+00 -9.87682268e-02 -5.08689582e-01 3.56543958e-01
1.13464808e+00 -2.17987120e-01 -6.09105647e-01 -1.37727559e-01
5.42973578e-01 -8.29003379e-02 -8.38523209e-02 -3.89902830e-01
-7.69674778e-01 1.34251475e+00 1.63615733e-01 3.42391819e-01
7.61912584e-01 3.48611265e-01 6.83161497e-01 5.96920669e-01
3.23300302e-01 -8.52512181e-01 -1.68094605e-01 1.03339815e+00
1.13536036e+00 -9.32032168e-01 -9.80952010e-02 -7.73013294e-01
-5.24846256e-01 1.21777618e+00 9.04926419e-01 -4.46585447e-01
6.85519099e-01 2.26060897e-01 -1.95000276e-01 -3.69928747e-01
-9.83576357e-01 -1.85000703e-01 3.27177159e-02 6.81905985e-01
8.88825536e-01 -3.60048972e-02 -5.33672333e-01 8.54092121e-01
-2.04561949e-01 -3.52714807e-01 2.95333087e-01 1.01258790e+00
-1.02859592e+00 -8.62167895e-01 -5.28714955e-01 1.22352108e-01
-8.76056194e-01 -3.98479164e-01 -5.62354863e-01 6.83649600e-01
4.91135195e-03 7.51278937e-01 3.96488696e-01 1.97180212e-01
3.02922159e-01 3.10143113e-01 2.56918609e-01 -1.03895390e+00
-8.84698033e-01 -3.32712919e-01 9.58860591e-02 -1.96643278e-01
-1.82926014e-01 -3.76345605e-01 -2.02608180e+00 2.99361954e-03
-2.14084044e-01 6.20566368e-01 6.87897027e-01 8.66742015e-01
3.79594415e-01 8.92267287e-01 2.56442875e-01 -7.50380099e-01
-2.31111825e-01 -1.26702571e+00 -3.33060622e-02 4.84506071e-01
6.63159788e-02 -8.78013015e-01 -2.16242038e-02 1.54776801e-03] | [10.652423858642578, 9.22823429107666] |
6a363c1e-b334-48e0-aafc-101152a9e980 | a-sketch-based-3d-shape-retrieval-approach | 1903.00117 | null | http://arxiv.org/abs/1903.00117v2 | http://arxiv.org/pdf/1903.00117v2.pdf | A Sketch Based 3D Shape Retrieval Approach Based on Efficient Deep Point-to-Subspace Metric Learning | A sketch based 3D shape retrieval | ['Lingqiao Liu', 'Pingping Zhang', 'Zijun Ma', 'Yinjie Lei', 'Ziqin Zhou', 'Yulan Guo'] | 2019-03-01 | null | null | null | null | ['3d-shape-retrieval'] | ['computer-vision'] | [-4.98136103e-01 -4.64488864e-01 -1.02306657e-01 -3.32421422e-01
-6.04498982e-01 -1.06826818e+00 8.76795828e-01 -4.52027321e-01
5.32107115e-01 -3.03859189e-02 2.60196388e-01 -1.75194085e-01
-1.96995437e-01 -1.22423029e+00 -3.53060305e-01 -1.10791497e-01
9.63878557e-02 1.04821622e+00 4.22664970e-01 -2.55159736e-01
1.11628962e+00 1.99342597e+00 -1.10399699e+00 4.10042614e-01
-8.46985638e-01 1.40541720e+00 -3.54798883e-01 9.16941941e-01
-1.18808782e+00 -4.99264836e-01 -4.14728761e-01 -7.85210371e-01
6.55368805e-01 9.42915604e-02 -1.88972503e-01 3.81925181e-02
1.46742630e+00 -8.82717907e-01 -6.61524773e-01 7.03903735e-01
9.84911799e-01 -9.38147977e-02 1.63450575e+00 -8.48019123e-01
-1.33786631e+00 -3.87629360e-01 -1.64692000e-01 -7.00998485e-01
1.14929700e+00 -5.56113899e-01 1.00499582e+00 -2.21951580e+00
1.52385664e+00 2.28450298e+00 1.08120596e+00 6.61046207e-01
-1.45171368e+00 -2.29329616e-01 -5.16876757e-01 -5.38701773e-01
-1.46252942e+00 -3.67232114e-01 1.49737191e+00 -1.60034001e-01
1.07837403e+00 5.82537770e-01 1.31098557e+00 8.93852949e-01
3.03667426e-01 1.31471908e+00 7.04235077e-01 -4.09083247e-01
1.48823783e-01 -4.54858989e-01 -1.75926387e-02 1.25467956e+00
2.41380051e-01 1.20178834e-01 -6.64614260e-01 -1.11896980e+00
1.91534424e+00 2.76803881e-01 5.10351419e-01 -1.25804186e+00
-2.57242739e-01 6.51969671e-01 2.12629557e-01 4.75126244e-02
-8.42040479e-02 6.65741682e-01 3.51922601e-01 6.12395942e-01
6.81656480e-01 5.40818632e-01 -1.01053558e-01 -7.14207739e-02
-9.43867803e-01 1.00586498e+00 7.02865660e-01 1.63215017e+00
9.81282592e-02 2.17079461e-01 -3.82796347e-01 9.48188126e-01
6.24570191e-01 1.31993413e+00 -1.19360447e-01 -6.89655066e-01
3.01030744e-02 7.94513404e-01 3.17094117e-01 -1.65937281e+00
3.77022952e-01 1.10968733e+00 -2.14243010e-01 5.62952936e-01
-1.30663514e-01 1.19352770e+00 -5.93369126e-01 1.89856201e-01
2.79094756e-01 -4.96036738e-01 -3.62343580e-01 1.20632982e+00
1.74352753e+00 4.06818271e-01 -2.07409784e-01 7.40627348e-01
1.36391115e+00 -2.17077523e-01 -6.30970120e-01 8.96453023e-01
-8.51053372e-02 -1.54552794e+00 1.04479539e+00 3.57655615e-01
-1.89830995e+00 -3.26932430e-01 -1.33947933e+00 -5.62563598e-01
-7.26942062e-01 1.89306065e-01 8.00436020e-01 9.01504934e-01
-1.20646048e+00 7.00322568e-01 -1.94786713e-01 -6.28547668e-02
8.96321416e-01 -4.84012775e-02 -7.31443465e-01 -5.04146993e-01
-6.44060671e-01 5.68337858e-01 -3.45583737e-01 -7.94744790e-01
-4.46251333e-01 -5.75145721e-01 -6.44365072e-01 -3.15802991e-01
-1.18023552e-01 -8.77052724e-01 6.18057489e-01 4.24821258e-01
-1.97071266e+00 1.56709456e+00 -1.82416290e-01 6.01610005e-01
4.66912806e-01 1.83596581e-01 2.64614522e-01 6.59477055e-01
-6.22061312e-01 3.75891745e-01 1.70944071e+00 -1.14490759e+00
5.33419371e-01 -5.65573215e-01 -3.09573077e-02 -2.47393418e-02
2.18171984e-01 4.01884839e-02 -4.64537621e-01 -1.25015175e+00
8.93400729e-01 -6.41006351e-01 5.33926606e-01 1.35221553e+00
3.13520908e-01 -7.36138999e-01 1.46217513e+00 -4.19013679e-01
3.44999194e-01 -1.68733358e+00 -4.99935970e-02 5.80513299e-01
2.02835873e-01 5.81981957e-01 -6.38657153e-01 1.33025956e+00
2.29909107e-01 7.31991291e-01 6.77942336e-01 -6.87384367e-01
5.26797593e-01 1.04075141e-01 -8.04473996e-01 1.05764963e-01
5.78304946e-01 1.64110422e+00 -1.04702461e+00 -4.84645098e-01
4.24797773e-01 7.25714147e-01 -1.97289810e-01 5.63100278e-01
1.72786772e-01 -3.54591846e-01 -1.03869438e+00 1.80989242e+00
1.12398255e+00 4.40138072e-01 -4.49028969e-01 -2.02385396e-01
4.19990152e-01 -1.06692865e-01 -1.06186450e+00 2.15308166e+00
-2.35323399e-01 6.22206569e-01 1.09445140e-01 -4.76310849e-01
1.78653669e+00 9.80267763e-01 1.67095929e-01 -4.59397942e-01
-1.74332559e-01 6.07780695e-01 -1.54453242e+00 -8.53638817e-03
4.44869310e-01 -2.63657242e-01 8.15268829e-02 8.86524558e-01
-1.62512094e-01 -1.70118392e+00 -9.81874704e-01 4.79709953e-01
8.65498543e-01 6.20641828e-01 1.05253220e-01 -5.68823099e-01
7.78173387e-01 -3.59647930e-01 -7.91275382e-01 5.12539446e-01
1.50595814e-01 1.24819314e+00 -6.00609630e-02 -1.03243220e+00
-1.46548057e+00 -1.78425550e+00 -4.04342890e-01 6.98863685e-01
9.85269397e-02 -3.69421154e-01 -2.31186375e-02 -4.22889769e-01
1.25902021e+00 8.19688588e-02 -4.52850133e-01 4.99515891e-01
-6.93783104e-01 6.23564661e-01 1.08152497e+00 5.19127011e-01
-2.42499113e-01 -1.42132711e+00 -4.58872095e-02 3.15772258e-02
8.73340249e-01 -2.59224743e-01 -1.10030210e+00 -8.86248291e-01
-1.58550835e+00 -8.55030298e-01 -1.41087234e+00 -8.24357212e-01
1.10964596e+00 8.65136445e-01 1.43031490e+00 4.74321514e-01
-9.14171338e-01 1.54059112e+00 -1.27584577e-01 -4.17036474e-01
-3.29846501e-01 -6.23531997e-01 2.90062755e-01 -5.85818708e-01
1.01884268e-01 -7.38158166e-01 -5.04754484e-01 3.58333379e-01
-1.00941873e+00 -4.01483804e-01 9.17149931e-02 8.08394492e-01
8.91028643e-01 -1.09631288e+00 4.02820915e-01 2.14947253e-01
1.58428383e+00 5.14158964e-01 -5.45858264e-01 5.94642639e-01
-3.42882186e-01 1.81043267e-01 -1.09856896e-01 -1.08012903e+00
-7.10232019e-01 -7.20842779e-02 -1.97637364e-01 -1.35198355e+00
-3.66985530e-01 -1.79440260e-01 1.64878160e-01 -8.14618349e-01
3.92460287e-01 6.38539016e-01 6.21891141e-01 -1.21816254e+00
8.98232460e-01 3.25478911e-01 -1.77815259e-01 -1.12223685e+00
1.12208462e+00 5.76629758e-01 3.36642385e-01 -1.32211626e+00
4.25720550e-02 -5.20768762e-01 -1.10174382e+00 -3.42092276e-01
-1.74944237e-01 -3.66833806e-01 -7.62864470e-01 1.05113968e-01
-1.54920876e+00 2.51404166e-01 -6.24568403e-01 -4.48463976e-01
-1.17005551e+00 6.99977875e-01 -7.01532125e-01 -1.56980014e+00
-1.41169167e+00 -5.70136368e-01 2.29210925e+00 -4.38610733e-01
-3.98102164e-01 -6.65351629e-01 5.75575292e-01 -3.71772468e-01
4.17593241e-01 3.62913042e-01 1.26862919e+00 1.47283956e-01
-8.09646308e-01 -1.30317557e+00 -4.26447362e-01 -4.00552541e-01
3.68078686e-02 1.17917947e-01 -4.51939791e-01 -2.03311071e-01
-6.01835489e-01 -6.02793813e-01 9.69643742e-02 1.47643730e-01
8.99144053e-01 -1.64831236e-01 -4.42119956e-01 3.87822032e-01
1.65003884e+00 -8.03190619e-02 4.62158084e-01 -1.01504469e+00
3.25439125e-01 1.93379164e-01 2.43434593e-01 2.13619590e-01
-4.66120034e-01 9.49326396e-01 -1.21319450e-01 3.85696411e-01
-1.08503532e+00 -9.02757227e-01 -2.46768683e-01 9.81351852e-01
-5.54127693e-01 4.52231735e-01 -1.27482802e-01 4.30618167e-01
-1.19529200e+00 -9.46260691e-01 -1.00493617e-01 1.91599989e+00
6.20434701e-01 -2.41618082e-01 -2.41733789e-02 1.41135290e-01
2.94957042e-01 4.18662094e-02 -1.20624915e-01 -1.02242815e+00
-5.36708772e-01 1.20712590e+00 -7.77621865e-02 6.95199311e-01
-3.86819065e-01 1.30151951e+00 7.77295589e+00 1.62316597e+00
-6.42502606e-01 -3.13017011e-01 5.23638874e-02 2.34211028e-01
-7.18812346e-01 -2.74690986e-01 -3.79086614e-01 -6.87932730e-01
-4.51167315e-01 -1.68323830e-01 7.10608304e-01 1.15999520e+00
-2.19800860e-01 2.09991500e-01 -1.26229644e+00 1.86903441e+00
4.98606324e-01 -2.01428866e+00 9.75020528e-01 1.79428890e-01
4.68927532e-01 -6.62087679e-01 -1.17810577e-01 -9.10884961e-02
-4.09830958e-01 -1.11089265e+00 1.03279650e+00 1.16937280e+00
1.33329093e+00 -7.18433261e-01 -2.02174127e-01 -5.77748334e-03
-1.37280309e+00 8.26105833e-01 -7.71782517e-01 2.53228068e-01
-8.15169513e-02 -5.31256981e-02 -7.20677137e-01 3.88147324e-01
2.46081322e-01 3.57698917e-01 -4.80984002e-01 1.24857652e+00
2.32227162e-01 -2.56370217e-01 -3.12033594e-01 -1.19943178e+00
1.21292777e-01 -5.23466587e-01 1.29853690e+00 1.11057794e+00
3.25170934e-01 8.24097335e-01 -3.08786601e-01 1.43730342e+00
-4.85547520e-02 7.91504502e-01 -1.80736208e+00 -5.01376987e-01
2.35129386e-01 9.05129850e-01 -9.11821485e-01 -4.90826339e-01
3.34463865e-01 1.40191019e+00 5.93141466e-02 2.73504052e-02
6.60690844e-01 -7.21069574e-01 6.27628565e-01 2.34424487e-01
5.30861199e-01 -1.05220616e+00 -6.42970622e-01 -7.87146270e-01
2.86803246e-02 -1.09587841e-01 -6.75280631e-01 -1.29979563e+00
-2.55550766e+00 2.10628092e-01 -1.24327861e-01 -1.52439988e+00
2.22646296e-01 -1.13751209e+00 -5.08652568e-01 9.88682449e-01
-7.00656891e-01 -1.47931886e+00 2.96056509e-01 6.55195236e-01
5.14171302e-01 -5.29340982e-01 1.71367943e+00 -1.80161759e-01
1.17722571e+00 6.78816080e-01 -4.06286679e-02 -1.55666173e-02
9.76231456e-01 -9.70675170e-01 1.19021153e+00 -6.41151667e-01
6.76144361e-01 7.29221165e-01 -1.76275969e-01 -1.01640844e+00
-2.88770700e+00 -1.18736163e-01 8.11668754e-01 -1.29821205e+00
-8.29848871e-02 -7.29967415e-01 -8.85208368e-01 -4.24016088e-01
-5.23522615e-01 2.99107824e-02 1.82714254e-01 -3.62775028e-01
-1.00339961e+00 4.79865968e-01 -2.00766444e+00 1.11158121e+00
2.03509736e+00 -1.07404518e+00 -1.08638906e+00 2.64895558e-01
1.28018409e-01 -2.87196577e-01 -1.48292220e+00 1.60483181e-01
1.39461732e+00 -3.20943415e-01 2.19000411e+00 -1.15513647e+00
1.67026341e-01 3.01225930e-01 -2.72405416e-01 -7.35499620e-01
-5.87420344e-01 -7.78008461e-01 -4.72797930e-01 4.95357841e-01
-1.84950128e-01 -2.52116561e-01 7.62718499e-01 6.20108843e-01
1.41907632e-01 -7.22980917e-01 -9.71895754e-01 -1.14224589e+00
2.80767620e-01 -4.96634632e-01 7.57026315e-01 7.41516173e-01
-1.86233385e-03 -2.98187658e-02 -2.43645571e-02 -6.95069492e-01
7.99906254e-01 9.40145195e-01 1.08573174e+00 -1.80629778e+00
7.90524542e-01 -1.39154005e+00 -6.11231267e-01 -1.39502585e+00
9.28640366e-05 -1.07844257e+00 -7.07513332e-01 -1.16099775e+00
-4.34398025e-01 -5.78304827e-01 4.39999312e-01 -1.41338676e-01
7.10923314e-01 7.16451406e-01 2.16920957e-01 3.37432325e-01
-2.15784177e-01 8.50133836e-01 2.13929963e+00 -6.53260648e-01
1.34752139e-01 -6.02844283e-02 3.00883651e-01 2.67949164e-01
-2.65660912e-01 -2.84068674e-01 -2.24359304e-01 -3.55854556e-02
6.75382316e-02 5.89324355e-01 3.09352934e-01 -5.60899526e-02
-1.68198943e-01 -2.69318253e-01 9.63539541e-01 -1.60645700e+00
1.07998145e+00 -9.65109169e-01 -1.37226284e-01 1.20159954e-01
-8.24714601e-02 9.95661616e-02 1.45243719e-01 7.59644747e-01
2.37004936e-01 -4.20661300e-01 1.73600845e-03 -7.40252078e-01
6.26829639e-02 8.32157254e-01 -2.42433012e-01 -2.08768606e-01
-2.15996969e-02 -9.38702047e-01 5.07274210e-01 -5.26070714e-01
-8.19318414e-01 -8.46122503e-01 8.56332779e-01 9.95559037e-01
1.76386404e+00 -2.65553379e+00 -7.48624682e-01 6.83416486e-01
3.96487266e-01 -6.36900365e-01 -4.02274966e-01 -4.37514544e-01
-9.79614973e-01 1.02378190e+00 -4.64311868e-01 -3.80544692e-01
-1.38523972e+00 5.75601697e-01 -5.14711887e-02 5.71843863e-01
-1.18064034e+00 6.47764444e-01 -3.88507158e-01 -7.16867566e-01
2.85222083e-01 -8.84852707e-02 2.75861800e-01 -2.93243676e-01
6.32777631e-01 4.96629596e-01 8.99957567e-02 -2.17972800e-01
-2.17465445e-01 1.49721289e+00 7.39874184e-01 -4.56713468e-01
1.15093565e+00 8.36378694e-01 -3.66709441e-01 3.78025115e-01
1.23356962e+00 2.73000836e-01 -5.50508559e-01 -2.95202702e-01
-5.02249487e-02 -1.34291029e+00 -1.49551421e-01 -6.27143264e-01
-5.03067255e-01 8.51371408e-01 4.58342433e-01 4.77625102e-01
3.22333336e-01 5.91449291e-02 1.07616866e+00 9.87227142e-01
1.19244623e+00 -1.05179358e+00 4.17289674e-01 6.09994292e-01
2.51411057e+00 -1.05724418e+00 4.52994972e-01 -4.70685363e-01
1.75609663e-01 1.89850593e+00 -6.58868194e-01 -1.23267341e+00
1.53081071e+00 1.00270003e-01 -3.03423196e-01 -9.44610059e-01
-4.85181957e-02 3.89008671e-01 1.55004644e+00 7.98712194e-01
2.47582167e-01 2.80622184e-01 -5.30356348e-01 2.87903070e-01
3.07494760e-01 -2.92059928e-01 -3.97001624e-01 1.01032901e+00
3.20846826e-01 -1.69606555e+00 -6.02381170e-01 -7.30597079e-02
2.25833252e-01 -3.58433604e-01 -1.40453947e+00 3.89473706e-01
-1.09687424e+00 1.86155483e-01 -1.35147288e-01 2.12653324e-01
1.09863067e+00 7.62567520e-01 1.42640591e+00 -3.30334663e-01
-3.17926049e-01 2.34698832e-01 -2.02834144e-01 -9.83072400e-01
-2.41861612e-01 -3.84076953e-01 -8.89282644e-01 -3.57742846e-01
-1.90888301e-01 -5.14919460e-01 1.16210043e+00 8.25518072e-02
5.62876940e-01 -1.04295015e+00 6.78187668e-01 -1.79064453e+00
-2.56208003e-01 -5.52787483e-01 -9.44687426e-01 7.06995249e-01
1.08858896e-02 -1.21255624e+00 -2.31834784e-01 -1.54044226e-01] | [8.44394588470459, -3.688795566558838] |
97ce9a1e-0376-472c-913c-b0f2a7306bb9 | assessing-differentially-private-deep | 1912.11328 | null | https://arxiv.org/abs/1912.11328v4 | https://arxiv.org/pdf/1912.11328v4.pdf | Assessing differentially private deep learning with Membership Inference | Attacks that aim to identify the training data of public neural networks represent a severe threat to the privacy of individuals participating in the training data set. A possible protection is offered by anonymization of the training data or training function with differential privacy. However, data scientists can choose between local and central differential privacy and need to select meaningful privacy parameters $\epsilon$ which is challenging for non-privacy experts. We empirically compare local and central differential privacy mechanisms under white- and black-box membership inference to evaluate their relative privacy-accuracy trade-offs. We experiment with several datasets and show that this trade-off is similar for both types of mechanisms. This suggests that local differential privacy is a sound alternative to central differential privacy for differentially private deep learning, since small $\epsilon$ in central differential privacy and large $\epsilon$ in local differential privacy result in similar membership inference attack risk. | ['Philip-William Grassal', 'Florian Kerschbaum', 'Jonas Robl', 'Daniel Bernau'] | 2019-12-24 | null | null | null | null | ['membership-inference-attack'] | ['computer-vision'] | [ 9.30254534e-02 2.41130710e-01 7.26186903e-03 -6.76585495e-01
-6.44572377e-01 -1.22404087e+00 3.70322406e-01 2.02565044e-01
-9.39389229e-01 7.81522691e-01 -2.13870645e-01 -9.13523912e-01
-3.26695919e-01 -8.86655152e-01 -8.32448602e-01 -8.91069710e-01
-2.84067988e-01 1.57226861e-01 -4.16359901e-01 3.15220833e-01
3.70443389e-02 7.71668136e-01 -1.20678067e+00 2.23246068e-01
6.88492179e-01 9.83364940e-01 -6.68191969e-01 3.71692210e-01
1.20938390e-01 -4.15538885e-02 -8.79272938e-01 -9.79935288e-01
1.09862506e+00 -1.93839043e-01 -7.09278524e-01 -7.47641087e-01
6.43684745e-01 -4.35527653e-01 -2.95555264e-01 1.55597568e+00
5.50445080e-01 -8.08421709e-03 3.11895460e-01 -1.60225487e+00
-4.85493958e-01 7.75244951e-01 -4.89476293e-01 1.65173814e-01
-1.14244990e-01 3.15683067e-01 8.72445762e-01 -7.06992894e-02
3.07675391e-01 9.74603176e-01 9.47797716e-01 8.67111444e-01
-1.77927065e+00 -1.32466352e+00 -4.98899557e-02 -4.94378179e-01
-1.54921448e+00 -4.67178762e-01 4.17604148e-01 -4.70324963e-01
3.14934760e-01 6.59805357e-01 4.17809248e-01 1.18422508e+00
1.72053039e-01 3.47636282e-01 1.19450259e+00 1.60416141e-01
5.13990879e-01 4.91193771e-01 8.21957365e-02 1.92281038e-01
1.02268445e+00 1.43138394e-01 -2.18305826e-01 -1.06773305e+00
6.67993009e-01 2.84383278e-02 -2.84440875e-01 -5.64880013e-01
-4.57515150e-01 7.90646911e-01 1.60463884e-01 -3.68186049e-02
9.88895819e-02 2.52116710e-01 5.88646293e-01 7.98203111e-01
3.44872773e-01 7.60779202e-01 -6.42942369e-01 2.13970408e-01
-5.96698225e-01 5.30740380e-01 9.45524633e-01 9.75584269e-01
7.50087678e-01 -2.53074199e-01 4.85643446e-02 2.73555458e-01
1.09602943e-01 -3.88925411e-02 3.13968956e-01 -1.08396924e+00
6.18924797e-01 4.02652234e-01 1.19936123e-01 -1.09194958e+00
-7.36485869e-02 -2.68143088e-01 -8.41661751e-01 6.66114509e-01
9.46373165e-01 -8.96511436e-01 -4.89791334e-01 2.10576606e+00
3.15700352e-01 -2.66929060e-01 1.96422458e-01 4.52037424e-01
2.73706824e-01 3.64152521e-01 3.28835070e-01 6.65991530e-02
1.38633108e+00 -5.12546524e-02 -2.67566741e-01 -2.60483205e-01
8.38933885e-01 -1.13033831e-01 8.92158806e-01 -4.40666601e-02
-8.62509668e-01 1.15900449e-01 -1.00957823e+00 -1.03758134e-01
-4.18107271e-01 -2.79761493e-01 8.77817214e-01 1.46750879e+00
-1.10211647e+00 8.55470002e-01 -7.86378503e-01 -5.59656769e-02
9.77279007e-01 9.23935354e-01 -6.55397236e-01 2.82723010e-01
-1.25826073e+00 3.47687930e-01 4.75136250e-01 -1.21542238e-01
-4.88361448e-01 -1.22006571e+00 -6.19706213e-01 2.23627403e-01
3.39469351e-02 -4.94225383e-01 8.34658384e-01 -1.01233566e+00
-1.02000225e+00 1.22035766e+00 4.35001671e-01 -8.77356768e-01
9.33281064e-01 1.40842870e-01 -2.25544930e-01 -2.50044197e-01
-2.07621232e-01 4.65198964e-01 5.03186584e-01 -1.10608232e+00
-5.43146729e-01 -8.51220787e-01 1.08631223e-01 8.54534060e-02
-4.81199205e-01 1.60117179e-01 2.09719583e-01 -4.83310401e-01
-2.28396859e-02 -8.65288496e-01 -3.92914712e-01 5.04016817e-01
-4.35220987e-01 2.03847110e-01 9.51676726e-01 -5.90072572e-01
8.66107345e-01 -2.34193110e+00 -6.74258947e-01 5.32815814e-01
3.78319412e-01 3.19915861e-01 1.71243429e-01 2.46147700e-02
-1.56911269e-01 6.37319922e-01 -1.86667249e-01 -4.37663257e-01
2.66200304e-01 7.35981092e-02 -3.61439914e-01 9.13681030e-01
-1.07350469e-01 5.10419667e-01 -4.65280473e-01 1.48947639e-02
-3.73141855e-01 4.15482819e-01 -7.78878391e-01 -3.03386021e-02
7.82576017e-03 2.92637646e-01 -6.25721395e-01 3.49276692e-01
1.28438008e+00 1.74478114e-01 3.88329744e-01 3.38349700e-01
5.08188037e-03 2.44539291e-01 -1.03378856e+00 1.05682886e+00
5.21623977e-02 6.74403787e-01 5.37530124e-01 -6.27679944e-01
8.64940584e-01 3.44675988e-01 2.95506954e-01 -2.25539833e-01
4.20796245e-01 1.90514792e-02 4.56776954e-02 1.05224445e-01
3.30101937e-01 -3.32061261e-01 -2.32699558e-01 7.67455876e-01
-2.51054168e-01 4.81947035e-01 -5.83770096e-01 -1.66054770e-01
1.19433701e+00 -3.82848263e-01 3.49991918e-02 -9.20564532e-01
1.12198085e-01 -4.10004914e-01 1.06290567e+00 8.39263976e-01
-6.21491075e-01 5.62368333e-01 1.15343189e+00 -4.61352527e-01
-9.46914852e-01 -7.40013659e-01 -3.14740539e-01 8.49571288e-01
2.39378549e-02 -1.94305465e-01 -1.12498355e+00 -1.06526470e+00
4.12437439e-01 6.21290386e-01 -6.48360252e-01 -3.94187093e-01
-1.63640991e-01 -7.77871788e-01 1.35186255e+00 3.55682701e-01
5.63806593e-01 -4.82108712e-01 -7.79838741e-01 -5.10247529e-01
4.82449889e-01 -7.64934123e-01 -7.07191288e-01 4.17242348e-01
-8.69547248e-01 -7.32481241e-01 -5.29923022e-01 -3.82055044e-01
9.66527760e-01 -3.45350020e-02 5.94383657e-01 -1.68283284e-01
-7.85511360e-02 1.80681735e-01 1.33054182e-01 -6.62371278e-01
-4.82199430e-01 3.41748670e-02 8.23980570e-02 1.02882408e-01
7.15973556e-01 -8.70218635e-01 -7.13292539e-01 2.37025365e-01
-8.82487357e-01 -4.81502056e-01 1.40507460e-01 3.74219626e-01
4.17912215e-01 2.90959120e-01 2.23383993e-01 -1.48996282e+00
7.80898333e-01 -5.49121439e-01 -1.01827145e+00 1.65621728e-01
-7.31338441e-01 3.29228848e-01 8.94367099e-01 -5.51385403e-01
-9.16210771e-01 2.90190816e-01 -1.60005122e-01 -5.76861739e-01
-3.10201913e-01 -1.94954090e-02 -7.69295275e-01 -4.52618450e-01
7.82204449e-01 -2.22294405e-01 2.04179585e-01 -7.19314516e-01
1.83779687e-01 4.79667306e-01 3.07669431e-01 -8.33559930e-01
7.16437042e-01 5.96475422e-01 5.33321984e-02 -5.34575582e-01
-9.02209133e-02 2.38579616e-01 -2.06439421e-01 5.50909579e-01
8.04349303e-01 -8.09691906e-01 -1.18040133e+00 5.89203298e-01
-7.00667262e-01 -1.77614942e-01 -6.41951025e-01 1.68678075e-01
-3.45757425e-01 3.95465225e-01 -4.64857072e-01 -8.20280075e-01
-3.85635197e-01 -1.13534296e+00 4.50423628e-01 1.83685139e-01
-3.17703873e-01 -1.01329362e+00 -2.69285381e-01 2.68986374e-01
4.65276361e-01 5.73413432e-01 9.33570802e-01 -1.18628669e+00
-3.02355558e-01 -6.31925285e-01 -1.04665034e-01 4.27078366e-01
7.79364407e-02 -1.65513262e-01 -1.26705432e+00 -6.51486397e-01
3.41590643e-01 -2.18346432e-01 5.82799852e-01 2.17979267e-01
1.55831742e+00 -9.58796144e-01 -1.96309149e-01 1.18949616e+00
1.27889872e+00 3.05918306e-01 5.73594391e-01 3.05638045e-01
5.40080488e-01 7.67044544e-01 -7.22686946e-02 5.80443621e-01
-4.74900473e-03 1.53420761e-01 2.06777230e-01 -1.33777246e-01
7.84144461e-01 -4.75731224e-01 2.87057936e-01 -4.54809487e-01
5.00489712e-01 -4.05209661e-02 -8.30100536e-01 1.94762558e-01
-1.44059849e+00 -8.95526648e-01 2.41921753e-01 2.65989423e+00
1.06559312e+00 9.77762491e-02 2.59798974e-01 -2.11465001e-01
7.91896999e-01 1.69123590e-01 -7.71651626e-01 -6.75829470e-01
-2.48401150e-01 1.05247222e-01 1.30161309e+00 2.39051729e-01
-1.13745451e+00 5.77577829e-01 6.59960270e+00 6.85231864e-01
-9.66072679e-01 3.11921034e-02 1.23272371e+00 -3.96686703e-01
-5.36735833e-01 1.82245970e-01 -8.24298918e-01 5.30055225e-01
1.11883152e+00 -6.76508904e-01 1.87682182e-01 9.35371339e-01
-3.14191997e-01 2.72755384e-01 -1.44306612e+00 9.11578536e-01
-5.22983193e-01 -1.26565671e+00 -1.68655008e-01 7.85189509e-01
5.97379625e-01 -1.49941131e-01 5.20682275e-01 -1.80847079e-01
8.16942513e-01 -9.94174123e-01 5.30553460e-01 5.96356168e-02
9.83010292e-01 -1.16393018e+00 3.89255494e-01 4.10147041e-01
-6.90234482e-01 -2.45197088e-01 -5.49856067e-01 1.65972441e-01
-4.53709900e-01 3.42664361e-01 -3.24531466e-01 1.13622576e-01
8.52084100e-01 -1.14868827e-01 -3.61970812e-01 6.29858077e-01
1.37654275e-01 2.66038507e-01 -6.09793067e-01 3.38731974e-01
-1.65816043e-02 -4.13347274e-01 3.89007479e-01 9.52548981e-01
2.35508129e-01 2.58978773e-02 -3.40115041e-01 1.37932670e+00
-6.02893174e-01 -1.93923160e-01 -8.46000731e-01 -3.12192351e-01
7.70167589e-01 8.65356088e-01 -3.54704320e-01 2.30974331e-01
-1.02656692e-01 9.14188385e-01 1.83785602e-01 3.20457995e-01
-4.68571603e-01 -5.11918545e-01 1.66032183e+00 8.38095620e-02
-2.92993858e-02 4.66903672e-02 -6.58866286e-01 -9.45113242e-01
3.05903815e-02 -9.79611278e-01 7.02768862e-01 1.19113326e-01
-1.42584836e+00 1.38970971e-01 -4.16161157e-02 -9.51966643e-01
8.43939409e-02 -5.45832515e-01 -6.06789947e-01 1.21608257e+00
-7.77180672e-01 -7.03104079e-01 3.75364095e-01 5.76057076e-01
-5.50728559e-01 -1.75150529e-01 7.98754871e-01 2.33386338e-01
-7.44330168e-01 1.64648986e+00 4.84623253e-01 4.00287569e-01
3.62324238e-01 -9.72580671e-01 8.61730278e-01 8.30341339e-01
-7.59109631e-02 1.24644864e+00 7.33335495e-01 -5.37198663e-01
-1.39810002e+00 -1.12925041e+00 6.94598734e-01 -5.87581635e-01
2.13947728e-01 -7.19554365e-01 -1.02140236e+00 9.34127212e-01
-1.06944904e-01 1.77287146e-01 1.08613396e+00 1.24471992e-01
-6.86185420e-01 -2.54317671e-01 -2.03891850e+00 8.07102263e-01
8.03441823e-01 -9.24058557e-01 1.11636199e-01 1.30888239e-01
8.73024583e-01 -1.88773423e-01 -9.38017011e-01 4.45870683e-02
6.23210013e-01 -9.85062659e-01 8.16710532e-01 -7.87116945e-01
-2.46985763e-01 6.18226873e-03 -2.13969961e-01 -8.49927068e-01
3.86177376e-02 -1.07555938e+00 2.67392248e-01 1.46549785e+00
6.46922171e-01 -1.26108098e+00 1.43335032e+00 1.75316560e+00
5.22342920e-01 -4.62570280e-01 -1.15138876e+00 -8.46780658e-01
7.71543086e-01 -2.54355401e-01 9.55215871e-01 1.31520581e+00
-1.10552832e-01 -4.60069001e-01 -2.02614650e-01 3.59364033e-01
8.77048790e-01 -3.71770173e-01 6.91823423e-01 -1.15111971e+00
-3.34573001e-01 -4.63861823e-01 -5.84514260e-01 -6.29880726e-01
5.20846963e-01 -8.67684603e-01 -4.06816214e-01 -5.29914320e-01
-3.08860037e-02 -7.05266833e-01 -2.60435700e-01 7.03610003e-01
9.84010845e-02 -6.79908097e-02 -3.00235059e-02 -5.65233082e-02
2.42424294e-01 2.72101194e-01 5.21222889e-01 2.63651200e-02
-3.54373246e-01 3.42462540e-01 -1.39505768e+00 6.23356819e-01
9.57236052e-01 -7.62144029e-01 -5.85908949e-01 -4.21610534e-01
3.19152951e-01 -2.91131109e-01 4.45571989e-01 -8.76862466e-01
3.31673414e-01 -1.94000155e-01 2.43565038e-01 1.25316933e-01
1.08050682e-01 -1.06464279e+00 5.47466397e-01 4.47087914e-01
-5.96485019e-01 1.19028352e-01 3.36472332e-01 5.52116156e-01
2.87171066e-01 -1.30071655e-01 1.02049959e+00 -1.07565738e-01
-5.00144847e-02 5.00048637e-01 -2.61073768e-01 2.52811134e-01
1.16930509e+00 -6.12876117e-01 -4.99242544e-01 -3.29176277e-01
-4.39789116e-01 2.21544981e-01 9.46497142e-01 1.12359114e-01
2.66652048e-01 -9.14409161e-01 -2.98507422e-01 6.65611506e-01
-3.91782820e-02 -1.32214829e-01 2.40647808e-01 4.39672887e-01
-6.16877794e-01 5.92962317e-02 -3.74995202e-01 -2.58400608e-02
-1.43573391e+00 4.84493762e-01 7.21574545e-01 1.95510685e-01
-5.25063455e-01 1.27012062e+00 4.22490418e-01 -4.86554623e-01
4.78953809e-01 -1.33877084e-01 4.30553377e-01 -1.07414022e-01
5.22109091e-01 3.98063630e-01 -3.62541825e-02 -3.91041428e-01
-3.24848086e-01 -3.72846164e-02 -2.95224756e-01 -1.90541387e-01
9.37164307e-01 1.12282336e-01 -1.73743770e-01 -8.84053484e-02
1.56772399e+00 -3.60074267e-02 -1.53530633e+00 -2.95301899e-02
-1.61095217e-01 -9.13439929e-01 -7.97384009e-02 -2.72724867e-01
-1.18171632e+00 7.11351275e-01 7.90442705e-01 2.54414469e-01
9.21841681e-01 -2.07759574e-01 4.36818719e-01 4.53160137e-01
5.27253926e-01 -9.05314684e-01 -9.55821931e-01 4.23772410e-02
3.87329102e-01 -8.20953965e-01 9.73928943e-02 -2.21370593e-01
-5.65095782e-01 4.73201364e-01 7.04838455e-01 -1.29498377e-01
8.10714722e-01 5.08578479e-01 1.05748504e-01 -1.03457859e-02
-6.50071323e-01 6.95214212e-01 -1.39171109e-01 6.45607054e-01
8.13559368e-02 1.67326331e-01 -4.25494790e-01 8.00834179e-01
-4.63896394e-01 -5.67160964e-01 3.38577092e-01 1.24293566e+00
5.06535806e-02 -1.35111392e+00 -2.62048066e-01 5.26796937e-01
-7.60311306e-01 3.17521617e-02 -8.76213610e-01 7.56380498e-01
2.80061632e-01 5.52976966e-01 2.64544100e-01 -4.12746191e-01
1.19238436e-01 1.78667799e-01 1.29909664e-01 -2.29314446e-01
-1.09893429e+00 -4.25105751e-01 -2.44606268e-02 -5.88260472e-01
3.15883905e-01 -7.70391285e-01 -1.12907660e+00 -9.50147808e-01
-1.82431564e-01 2.99163729e-01 5.02402067e-01 5.57391822e-01
8.33188057e-01 -2.19532818e-01 5.94028890e-01 -3.65275711e-01
-1.13072968e+00 -2.25007161e-01 -1.14278293e+00 4.72923696e-01
4.69402820e-01 4.47799824e-02 -6.01711452e-01 -4.94401336e-01] | [5.920676231384277, 6.9641432762146] |
3c9f5f39-ae84-4dcf-854e-b8c248c07d91 | intra-and-inter-epoch-temporal-context | 1902.06562 | null | https://arxiv.org/abs/1902.06562v2 | https://arxiv.org/pdf/1902.06562v2.pdf | Intra- and Inter-epoch Temporal Context Network (IITNet) Using Sub-epoch Features for Automatic Sleep Scoring on Raw Single-channel EEG | A deep learning model, named IITNet, is proposed to learn intra- and inter-epoch temporal contexts from raw single-channel EEG for automatic sleep scoring. To classify the sleep stage from half-minute EEG, called an epoch, sleep experts investigate sleep-related events and consider the transition rules between the found events. Similarly, IITNet extracts representative features at a sub-epoch level by a residual neural network and captures intra- and inter-epoch temporal contexts from the sequence of the features via bidirectional LSTM. The performance was investigated for three datasets as the sequence length (L) increased from one to ten. IITNet achieved the comparable performance with other state-of-the-art results. The best accuracy, MF1, and Cohen's kappa ($\kappa$) were 83.9%, 77.6%, 0.78 for SleepEDF (L=10), 86.5%, 80.7%, 0.80 for MASS (L=9), and 86.7%, 79.8%, 0.81 for SHHS (L=10), respectively. Even though using four epochs, the performance was still comparable. Compared to using a single epoch, on average, accuracy and MF1 increased by 2.48%p and 4.90%p and F1 of N1, N2, and REM increased by 16.1%p, 1.50%p, and 6.42%p, respectively. Above four epochs, the performance improvement was not significant. The results support that considering the latest two-minute raw single-channel EEG can be a reasonable choice for sleep scoring via deep neural networks with efficiency and reliability. Furthermore, the experiments with the baselines showed that introducing intra-epoch temporal context learning with a deep residual network contributes to the improvement in the overall performance and has the positive synergy effect with the inter-epoch temporal context learning. | ['Kyoobin Lee', 'Seongju Lee', 'Seunghyeok Back', 'Deokhwan Park', 'Tae Kim', 'Hogeon Seo'] | 2019-02-18 | null | null | null | null | ['sleep-stage-detection'] | ['medical'] | [-1.41383614e-02 -2.56164849e-01 -1.52421361e-02 -4.29117590e-01
-6.05266273e-01 -2.00206056e-01 1.87848017e-01 -6.74867332e-02
-8.68810058e-01 1.13744283e+00 5.27452603e-02 -2.60660984e-02
-3.41273159e-01 -4.00099963e-01 -5.25394380e-01 -7.26005435e-01
-4.06717688e-01 -3.13740999e-01 2.22879216e-01 -6.77956790e-02
2.82086641e-01 -2.44016480e-02 -1.73868048e+00 4.91979450e-01
8.82338107e-01 1.50379622e+00 2.82391399e-01 3.01822603e-01
9.78947133e-02 4.82309997e-01 -9.18014824e-01 2.68793516e-02
1.27842911e-02 -5.08269012e-01 -6.04193807e-01 -3.91071826e-01
-1.72127739e-01 6.43744245e-02 -9.58590135e-02 9.63043451e-01
6.53895199e-01 2.01510653e-01 3.17938000e-01 -1.20360708e+00
-1.99025214e-01 6.71739638e-01 -5.83379745e-01 6.94713533e-01
3.83040726e-01 1.45462498e-01 6.14569962e-01 -6.11847281e-01
3.15865837e-02 3.72218043e-01 9.08106267e-01 4.72954541e-01
-1.09051716e+00 -1.25985909e+00 -8.25847387e-02 6.01090789e-01
-1.62842226e+00 -3.00808221e-01 4.67943877e-01 -2.70557910e-01
1.24600422e+00 3.77026916e-01 1.06947172e+00 1.20885754e+00
9.19552743e-01 3.57467264e-01 1.44089043e+00 -8.01481605e-02
2.42169172e-01 4.39858250e-02 4.19575363e-01 1.72972739e-01
-1.34991392e-01 1.78087652e-02 -6.67684913e-01 2.05954790e-01
2.19991133e-01 2.45908022e-01 -1.36229411e-01 7.29389429e-01
-1.01528621e+00 4.44815725e-01 5.75981617e-01 7.28548467e-01
-5.73858738e-01 -1.64119303e-01 5.61094820e-01 1.83644742e-01
5.54659784e-01 4.14700747e-01 -5.66601276e-01 -6.50056183e-01
-1.49493730e+00 -1.41922265e-01 3.99426818e-01 6.98765218e-01
6.96066320e-01 1.57161340e-01 -5.16219974e-01 8.47059488e-01
-1.31106839e-01 5.17396331e-01 1.00642502e+00 -8.43341708e-01
2.80513167e-01 8.14339101e-01 -3.12601216e-02 -5.58170438e-01
-1.06384349e+00 -8.11035991e-01 -1.10067272e+00 -2.78088421e-01
1.90628115e-02 3.91221466e-03 -8.57843339e-01 1.84676409e+00
-4.64192152e-01 3.17105561e-01 -6.27097338e-02 6.20086551e-01
8.95825446e-01 7.83316851e-01 9.04489830e-02 -7.68878937e-01
1.48428440e+00 -7.16148138e-01 -1.00376749e+00 -4.28001761e-01
7.78153986e-02 -5.30850232e-01 1.20607638e+00 5.17472744e-01
-1.33365798e+00 -8.11329126e-01 -1.06435561e+00 3.28408301e-01
-2.84557551e-01 3.52727890e-01 2.17229187e-01 3.70201051e-01
-1.31581938e+00 5.40323377e-01 -8.54354262e-01 -2.98463196e-01
2.25143954e-01 8.33776414e-01 -1.36493668e-01 4.52752948e-01
-1.39811516e+00 8.51913154e-01 3.23973328e-01 2.39290759e-01
-9.29219663e-01 -6.15947187e-01 -3.41338038e-01 2.51225352e-01
-1.73551202e-01 -3.64453793e-01 1.01279032e+00 -9.45956051e-01
-1.24725223e+00 6.35862648e-01 -5.35234988e-01 -6.93643510e-01
-1.17135108e-01 -1.66865289e-01 -8.39770257e-01 2.93924045e-02
2.94367582e-01 5.13981044e-01 5.18285990e-01 -6.42507911e-01
-7.77152002e-01 -5.62113941e-01 1.09371692e-02 4.57044765e-02
-3.39915752e-01 3.09666127e-01 -1.97897151e-01 -3.50663513e-01
-1.67670950e-01 -9.08534706e-01 1.24477550e-01 -6.67977631e-01
-1.86347961e-01 -3.04153413e-01 3.00712407e-01 -8.42153132e-01
1.65863621e+00 -2.29346967e+00 -1.95272967e-01 2.15644599e-03
1.74280524e-01 3.27172726e-01 1.82944849e-01 7.57783130e-02
-9.09702480e-02 1.53725334e-02 -3.70716423e-01 -2.93217361e-01
-1.94728702e-01 3.97701524e-02 9.41822454e-02 3.30763102e-01
-1.55376568e-01 7.93729365e-01 -7.82579899e-01 -4.39594314e-02
7.62315616e-02 5.44005573e-01 -8.80660713e-02 2.40926087e-01
5.60484886e-01 2.81034589e-01 1.52516887e-01 3.27114403e-01
4.01796997e-01 -2.29423508e-01 -1.57902688e-01 -2.08591148e-01
-5.08678973e-01 5.16060412e-01 -7.44478703e-01 1.72538137e+00
-6.51204228e-01 9.29488778e-01 -3.75534832e-01 -9.45954144e-01
9.85223472e-01 4.48953748e-01 4.11844343e-01 -1.44358873e+00
2.23956287e-01 1.89973205e-01 3.60816598e-01 -5.71782053e-01
3.28452885e-01 -3.22829902e-01 -4.22334708e-02 3.24121714e-01
6.73182756e-02 3.86953264e-01 2.15497941e-01 -2.28101630e-02
1.28591025e+00 -3.62087518e-01 3.98901880e-01 -4.85909760e-01
6.42964482e-01 -7.61553705e-01 7.41848111e-01 5.18280625e-01
-2.16812834e-01 3.61113071e-01 2.63875932e-01 -4.49275196e-01
-5.31258464e-01 -1.09679556e+00 -3.17370713e-01 8.38102698e-01
2.23037839e-01 -5.90409338e-01 -7.52178788e-01 -3.52025658e-01
-4.34138775e-01 8.51358771e-01 -8.72875810e-01 -7.08179355e-01
-2.36546069e-01 -1.12501740e+00 4.99439329e-01 7.03174591e-01
8.07719350e-01 -1.48600328e+00 -8.58449101e-01 2.03972444e-01
-4.02836561e-01 -1.06779218e+00 -3.38052392e-01 5.52960932e-01
-9.57943201e-01 -6.72930121e-01 -5.21903336e-01 -4.73355949e-01
3.39783937e-01 -4.27922122e-02 9.94585931e-01 -8.37988779e-02
-1.03597656e-01 -4.47562486e-02 -4.09039915e-01 -3.38206708e-01
2.25034550e-01 2.01550692e-01 5.00008404e-01 1.06376119e-01
4.32844728e-01 -1.04592621e+00 -9.38960433e-01 4.64002401e-01
-6.28019512e-01 -1.29100069e-01 6.61506414e-01 7.07062244e-01
6.81079805e-01 1.05238572e-01 8.55707884e-01 -2.70363480e-01
6.25101626e-01 -5.08490980e-01 -8.12321305e-02 -6.29741997e-02
-1.05619979e+00 -1.93637639e-01 8.10767710e-01 -4.00589019e-01
-7.53760397e-01 -5.28750241e-01 -1.40207425e-01 -2.17319965e-01
-1.31092057e-01 2.86733091e-01 1.77159414e-01 1.89807802e-01
5.68828225e-01 6.12396061e-01 -4.89095420e-01 -2.19385266e-01
-3.22243482e-01 8.53930414e-01 4.37502474e-01 -4.77312803e-02
2.22905755e-01 2.45904073e-01 -4.65561956e-01 -6.29774392e-01
-7.83952713e-01 -4.39760387e-01 -3.40482295e-01 -1.37888610e-01
9.88621354e-01 -9.93247032e-01 -7.95006990e-01 4.77430135e-01
-8.48796189e-01 -4.90058005e-01 -2.49837130e-01 7.01110065e-01
-2.86221355e-01 -3.58278379e-02 -4.12121505e-01 -8.31465840e-01
-1.06770682e+00 -1.13267088e+00 7.01499581e-01 4.43955809e-01
-5.27415216e-01 -5.29439151e-01 -1.13740698e-01 1.94788069e-01
5.17238617e-01 -1.20152086e-01 5.34118712e-01 -6.45820081e-01
5.49310036e-02 9.19783562e-02 -8.43733996e-02 5.07971704e-01
4.51310903e-01 -2.81452954e-01 -1.16609848e+00 -2.97881126e-01
2.76002556e-01 -2.27577183e-02 3.81746769e-01 5.89142919e-01
1.47951007e+00 -2.95775354e-01 -2.02434972e-01 7.63826311e-01
1.08978879e+00 8.94760072e-01 8.73914421e-01 5.03369391e-01
1.22323230e-01 1.05434425e-01 2.86242068e-01 5.32653749e-01
4.12528962e-01 6.47017956e-01 3.62450987e-01 4.18923981e-02
-1.70466471e-02 3.43162894e-01 7.14492679e-01 1.21819556e+00
-2.51628608e-01 1.09663777e-01 -8.22441518e-01 7.36763239e-01
-1.43697774e+00 -9.91790414e-01 -8.32548663e-02 2.31774879e+00
8.90743732e-01 4.93845195e-01 2.33004972e-01 4.26250786e-01
6.39093220e-01 -1.23601891e-02 -6.78794503e-01 -6.46439135e-01
5.63834570e-02 6.23150766e-01 2.54089653e-01 -1.03155151e-01
-6.16388738e-01 4.39107388e-01 5.79462528e+00 7.97606170e-01
-1.35079134e+00 4.38187212e-01 5.59923172e-01 -8.29198301e-01
1.92531839e-01 -4.34829533e-01 -7.62428164e-01 1.45402086e+00
1.79509521e+00 -1.89401865e-01 8.51363659e-01 3.54061246e-01
6.59537733e-01 -4.02876377e-01 -8.89080644e-01 1.43049288e+00
-4.35195602e-02 -9.57275331e-01 -5.31064093e-01 -6.84070541e-03
7.42349148e-01 3.84920210e-01 -4.54856157e-02 5.73173523e-01
-5.18820941e-01 -1.00479615e+00 7.86716640e-01 7.30256438e-01
9.98526990e-01 -9.36433434e-01 1.13353825e+00 3.05268168e-01
-1.38057256e+00 -3.02543581e-01 -1.55090764e-01 -3.62884611e-01
-8.25072825e-03 6.90368176e-01 -3.79948527e-01 5.52687883e-01
1.48733437e+00 8.62080812e-01 -7.39625096e-01 1.07061648e+00
-3.33489686e-01 9.75464821e-01 -2.75360584e-01 -9.20443758e-02
1.51848108e-01 -9.19466745e-03 1.48021609e-01 1.21388304e+00
4.94460374e-01 1.62440553e-01 -4.13422793e-01 7.99351037e-01
-1.06059149e-01 -2.53639281e-01 -1.50970623e-01 3.31619024e-01
5.31328022e-01 1.34776628e+00 -6.71469152e-01 -2.57687479e-01
-2.67984539e-01 7.65980780e-01 6.37754872e-02 3.72840196e-01
-1.20467257e+00 -7.34914422e-01 4.69971925e-01 -8.76097456e-02
1.27186328e-02 1.96014404e-01 -5.05339503e-01 -8.44006836e-01
3.02048534e-01 -6.44097388e-01 2.04655901e-01 -1.04093027e+00
-9.98235166e-01 1.00615776e+00 -7.49545498e-03 -1.22384894e+00
-1.92021932e-02 -1.98814213e-01 -9.57132757e-01 7.88903058e-01
-1.27381563e+00 -5.85986078e-01 -3.64857346e-01 7.91757405e-01
5.99690974e-01 -2.05036979e-02 6.19434416e-01 5.39687395e-01
-8.03614676e-01 8.33733261e-01 6.58031106e-02 -1.92593768e-01
5.28001785e-01 -1.01366210e+00 9.89648178e-02 7.76361525e-01
-2.22876400e-01 8.71189117e-01 2.93965399e-01 -2.08980247e-01
-8.82268608e-01 -9.99291360e-01 1.03120649e+00 -8.38753507e-02
5.18290281e-01 -3.59828889e-01 -8.98034990e-01 4.93746668e-01
4.01855201e-01 -2.02888235e-01 7.60885417e-01 1.54459745e-01
2.75646597e-01 -7.68543899e-01 -1.09014273e+00 3.68147343e-01
1.00725520e+00 -7.09356308e-01 -7.68057883e-01 -8.54520798e-02
4.95336682e-01 -1.55111313e-01 -1.14822102e+00 5.39839566e-01
7.94118941e-01 -1.20535707e+00 7.37702549e-01 1.59083847e-02
1.74110264e-01 -1.22624427e-01 -9.33808535e-02 -1.18252385e+00
-4.25695151e-01 -5.00500977e-01 -1.56442583e-01 1.13874054e+00
4.87453759e-01 -7.22002149e-01 3.60187799e-01 4.41425800e-01
-5.84512293e-01 -1.04626667e+00 -1.18392694e+00 -8.42488706e-01
-1.85473517e-01 -6.92116261e-01 7.63183296e-01 4.65191424e-01
6.80603683e-02 3.80858839e-01 -3.33405942e-01 -1.99374673e-03
9.40141380e-02 -1.22364171e-01 6.04379475e-02 -1.19674015e+00
1.64534017e-01 -3.49514812e-01 -3.21854264e-01 -4.90858436e-01
1.11073539e-01 -8.85273218e-01 1.45948902e-02 -1.68435121e+00
3.97373557e-01 -1.24548264e-01 -1.28672445e+00 7.27566779e-01
-1.97188959e-01 4.70161140e-01 1.05047990e-02 1.43786609e-01
-6.33257806e-01 6.32894576e-01 7.13639557e-01 -1.02757439e-01
-4.47658062e-01 1.57631427e-01 -6.13321066e-01 6.49325728e-01
1.27255869e+00 -6.62201464e-01 -5.49898505e-01 -2.85017073e-01
2.62777239e-01 -9.84247327e-02 7.84172490e-02 -1.52322423e+00
2.88953424e-01 2.80189812e-01 5.81167877e-01 -6.01454556e-01
5.12420416e-01 -5.85801184e-01 2.74430841e-01 4.86788660e-01
-1.59167826e-01 3.13299030e-01 5.72436512e-01 1.67002112e-01
-6.64033741e-02 4.19912040e-02 6.40253961e-01 5.19355934e-04
-7.41787136e-01 1.38243660e-01 -5.90733945e-01 1.57252982e-01
8.02987218e-01 -5.65004349e-01 -2.59746373e-01 -1.49703726e-01
-8.80455911e-01 1.02205127e-01 -6.95763971e-04 3.99591833e-01
6.78333163e-01 -1.24653101e+00 -2.87789494e-01 4.78417039e-01
6.97603375e-02 -3.66700530e-01 5.51299870e-01 1.41814184e+00
2.15751976e-01 3.30973268e-01 -5.27496040e-01 -6.73979700e-01
-1.14019525e+00 1.65341556e-01 3.95066172e-01 -2.94030041e-01
-3.69494408e-01 8.28059614e-01 1.47589594e-01 1.40417054e-01
1.89423993e-01 -7.29547203e-01 -3.55256915e-01 2.30348855e-01
6.81705236e-01 6.81060553e-01 5.84255040e-01 -4.62438732e-01
-7.35861778e-01 6.77281141e-01 9.22484845e-02 8.72889087e-02
1.44165838e+00 -2.48193160e-01 -3.69635463e-01 9.38646853e-01
1.19711137e+00 -2.71708071e-01 -1.05892932e+00 2.67486960e-01
-1.20674431e-01 1.20841846e-01 4.84421179e-02 -1.23598564e+00
-1.25884426e+00 9.56438184e-01 1.14903855e+00 2.62566447e-01
1.64137053e+00 -2.27385372e-01 9.90438342e-01 3.78086492e-02
5.40755272e-01 -9.44493711e-01 -8.19451436e-02 5.35620749e-01
7.37759411e-01 -8.21672440e-01 2.63208337e-03 4.88637656e-01
-5.17869473e-01 9.11978006e-01 5.96671641e-01 -2.97914147e-02
5.49730062e-01 7.40410984e-02 -2.36151516e-01 -2.39753544e-01
-6.86805964e-01 8.47233459e-02 4.82485354e-01 1.13505408e-01
4.32480425e-01 2.75151819e-01 -6.30419791e-01 1.23153245e+00
-5.06519258e-01 2.44883627e-01 2.79856503e-01 5.03160119e-01
-2.44804099e-01 -7.90392280e-01 -9.69302580e-02 8.41936886e-01
-7.97820032e-01 -3.27431202e-01 1.06953911e-01 6.97619796e-01
6.34667635e-01 1.22503865e+00 2.16103971e-01 -1.03524637e+00
3.79391223e-01 2.97202289e-01 2.13605419e-01 -4.83095586e-01
-1.05064809e+00 -8.27437714e-02 -1.33227170e-01 -8.33936155e-01
-6.31709874e-01 -4.62743133e-01 -1.41400456e+00 -9.87943783e-02
-1.67786494e-01 4.79032725e-01 5.61024427e-01 1.04344702e+00
5.87280154e-01 9.09452319e-01 6.20580852e-01 -8.02754164e-01
9.22077298e-02 -1.22603333e+00 -8.16144407e-01 8.47604424e-02
4.43920940e-01 -6.96383953e-01 -6.18823469e-01 -5.44055849e-02] | [13.490825653076172, 3.507603883743286] |
a2d29af7-e57c-4b17-a7ef-6c028934eaa5 | hierarchical-gnns-for-large-graph-generation | 2306.11412 | null | https://arxiv.org/abs/2306.11412v1 | https://arxiv.org/pdf/2306.11412v1.pdf | Hierarchical GNNs for Large Graph Generation | Large graphs are present in a variety of domains, including social networks, civil infrastructure, and the physical sciences to name a few. Graph generation is similarly widespread, with applications in drug discovery, network analysis and synthetic datasets among others. While GNN (Graph Neural Network) models have been applied in these domains their high in-memory costs restrict them to small graphs. Conversely less costly rule-based methods struggle to reproduce complex structures. We propose HIGGS (Hierarchical Generation of Graphs) as a model-agnostic framework of producing large graphs with realistic local structures. HIGGS uses GNN models with conditional generation capabilities to sample graphs in hierarchies of resolution. As a result HIGGS has the capacity to extend the scale of generated graphs from a given GNN model by quadratic order. As a demonstration we implement HIGGS using DiGress, a recent graph-diffusion model, including a novel edge-predictive-diffusion variant edge-DiGress. We use this implementation to generate categorically attributed graphs with tens of thousands of nodes. These HIGGS generated graphs are far larger than any previously produced using GNNs. Despite this jump in scale we demonstrate that the graphs produced by HIGGS are, on the local scale, more realistic than those from the rule-based model BTER. | ['Telmo M. Silva Filho', 'Nirav S. Ajmeri', 'Alex O. Davies'] | 2023-06-20 | null | null | null | null | ['drug-discovery'] | ['medical'] | [ 1.54216975e-01 1.00558019e+00 -8.93560946e-02 2.80620847e-02
-3.28941762e-01 -6.14938796e-01 9.45841670e-01 3.21217120e-01
1.67267427e-01 1.13241708e+00 9.51202214e-02 -5.33198237e-01
-3.88667405e-01 -1.60230470e+00 -8.25668812e-01 -2.98570186e-01
-7.13568747e-01 1.02069867e+00 4.17350680e-01 -4.02659357e-01
-5.81469871e-02 5.20337224e-01 -1.00487125e+00 -8.70914757e-02
7.82859087e-01 2.02852622e-01 -2.57150382e-01 7.85978854e-01
6.59421980e-02 7.23402321e-01 -4.36476529e-01 -5.96765161e-01
2.92525887e-01 -4.47555006e-01 -7.30836511e-01 8.60449076e-02
4.45085794e-01 1.41324222e-01 -6.40975237e-01 9.93355453e-01
5.46715975e-01 2.37242617e-02 7.69647479e-01 -1.56818068e+00
-8.26652467e-01 1.22703850e+00 -4.16427642e-01 -2.14046780e-02
3.99777353e-01 3.96674693e-01 1.07637501e+00 -4.60437179e-01
1.22062135e+00 1.69694960e+00 1.03929579e+00 4.44178402e-01
-1.93891990e+00 -3.80900890e-01 -1.23025328e-01 -3.46761584e-01
-1.45997679e+00 1.65115539e-02 6.78223312e-01 -4.55178708e-01
8.92328084e-01 1.64533585e-01 9.74400103e-01 1.34186482e+00
2.99407214e-01 1.36715785e-01 1.18989229e+00 -3.95022601e-01
4.30324107e-01 -1.11487083e-01 -4.93877642e-02 9.75569308e-01
8.89136374e-01 1.43896028e-01 -2.82109052e-01 -6.06589496e-01
9.33001101e-01 -8.49281698e-02 -7.06684440e-02 -4.34499890e-01
-1.04212582e+00 1.07774198e+00 7.92129755e-01 1.62527427e-01
-4.97384280e-01 6.88820541e-01 2.13097259e-01 2.41723329e-01
6.82348311e-01 6.64261699e-01 -1.32654816e-01 4.32369292e-01
-7.76700437e-01 5.12036741e-01 1.24498534e+00 1.14918017e+00
8.91577899e-01 3.52193892e-01 -9.32980105e-02 4.61862892e-01
3.77647318e-02 2.79128909e-01 2.47191656e-02 -9.01699483e-01
1.97712988e-01 7.92688489e-01 -3.24751973e-01 -1.28780568e+00
-5.46130121e-01 -4.92228895e-01 -1.28641248e+00 2.22857948e-02
4.24261659e-01 -3.12800169e-01 -1.01239836e+00 1.72441816e+00
3.84710193e-01 3.84173766e-02 -1.34888679e-01 2.69254804e-01
6.45381093e-01 5.24739683e-01 9.00419056e-02 8.94276723e-02
9.20961082e-01 -7.54615605e-01 -1.63807064e-01 -9.97685939e-02
6.37683690e-01 -2.41620034e-01 7.68618047e-01 3.80960703e-01
-1.07166767e+00 -7.23666921e-02 -7.23912537e-01 3.09617072e-01
-6.23980343e-01 -8.59083891e-01 9.41533089e-01 7.99837232e-01
-1.71810210e+00 1.14007127e+00 -4.94707823e-01 -6.45429611e-01
5.84127069e-01 2.40562215e-01 -3.67412627e-01 -2.58980066e-01
-1.25414598e+00 6.85095727e-01 6.05034173e-01 -2.79272228e-01
-8.49680483e-01 -5.77715158e-01 -9.96863723e-01 3.39271389e-02
3.43901366e-01 -1.12840915e+00 8.66661489e-01 -8.05656672e-01
-9.41052735e-01 5.19938111e-01 3.59537601e-01 -7.71612644e-01
6.46000504e-01 8.03547263e-01 -3.52435499e-01 4.63791639e-02
1.97377980e-01 8.88800144e-01 6.46120727e-01 -1.14446521e+00
1.02970116e-01 -2.50232458e-01 -2.62240246e-02 -1.45451725e-01
-1.45865232e-01 -4.49278474e-01 -2.32974496e-02 -4.87154156e-01
-2.19642699e-01 -1.04017115e+00 -7.88550198e-01 -1.43007755e-01
-9.91711497e-01 -2.53613204e-01 4.32444394e-01 -3.87974054e-01
1.15339446e+00 -1.50561821e+00 1.30891144e-01 9.12858784e-01
9.00467277e-01 -2.74394214e-01 -4.34581101e-01 1.04027295e+00
-1.83895156e-02 6.45327628e-01 -3.47541839e-01 -1.30928643e-02
1.46305755e-01 2.73062885e-01 3.83251697e-01 2.55157441e-01
1.85265824e-01 1.15952086e+00 -1.13827705e+00 -5.14150202e-01
-2.52275854e-01 3.42508852e-01 -7.80867457e-01 -2.30202585e-01
-6.12522364e-01 -8.46486632e-03 -3.30898315e-01 3.97704154e-01
4.49547857e-01 -6.68386817e-01 7.59435236e-01 2.73889989e-01
5.80224276e-01 -9.26256999e-02 -9.74719942e-01 1.34955621e+00
-2.16419309e-01 4.22170281e-01 -1.33302271e-01 -6.66090906e-01
1.01995349e+00 -7.60330632e-02 2.13370770e-01 -2.90769845e-01
-8.31333473e-02 1.53978318e-01 4.63208199e-01 -2.25470541e-03
4.70534056e-01 -2.18912080e-01 -8.51202384e-02 6.40436053e-01
-3.89724337e-02 -5.24272680e-01 7.20422387e-01 9.46435571e-01
1.87233675e+00 -1.88023597e-01 3.62702340e-01 -4.58673328e-01
-2.21886724e-01 3.64445567e-01 -8.17513019e-02 8.44062865e-01
2.94203311e-01 4.58940119e-01 9.72286701e-01 -4.97635186e-01
-1.45935285e+00 -1.06009519e+00 2.40680560e-01 6.19435728e-01
-3.36092055e-01 -7.12525487e-01 -8.90733004e-01 -7.35012114e-01
2.84660161e-01 7.71184802e-01 -7.76204288e-01 -1.91624090e-01
-3.98250639e-01 -7.77618647e-01 5.89818895e-01 4.38360572e-02
-8.06315802e-03 -1.35320985e+00 2.27126226e-01 3.91683608e-01
3.29147696e-01 -9.52683985e-01 -3.65471750e-01 -1.65322781e-01
-8.77844036e-01 -1.41504002e+00 -6.43538058e-01 -6.90063655e-01
1.03890157e+00 -1.69134825e-01 1.67170405e+00 2.96337575e-01
-4.02397245e-01 2.82815039e-01 -1.39193252e-01 -1.94889829e-01
-1.20880497e+00 4.82464105e-01 -1.15867689e-01 -3.13555539e-01
-8.55616257e-02 -1.10590470e+00 -5.56938231e-01 -1.05519861e-01
-1.12886202e+00 1.45627648e-01 5.56959331e-01 7.21335411e-01
2.28354096e-01 5.45539744e-02 7.96012938e-01 -1.57978511e+00
1.40715778e+00 -9.37598705e-01 -5.36257982e-01 -3.48081440e-02
-1.00601745e+00 2.77174741e-01 8.52142930e-01 -4.73188668e-01
-2.56050855e-01 -2.79123098e-01 4.89784852e-02 -2.41883159e-01
1.20794132e-01 7.79152274e-01 3.36259604e-01 -3.38563547e-02
1.18201053e+00 1.29562207e-02 4.17798162e-01 -8.71952921e-02
7.62395501e-01 3.93970683e-03 1.98137090e-02 -5.51615953e-01
8.65853310e-01 1.31528988e-01 6.14614606e-01 -6.91974223e-01
-3.28424990e-01 2.50912756e-01 -3.25044036e-01 -1.72738150e-01
4.64840382e-01 -5.96476436e-01 -6.34833515e-01 1.40782341e-01
-1.02050579e+00 -7.45192826e-01 -5.88386714e-01 -8.78375992e-02
-4.26065236e-01 3.84727180e-01 -8.35349500e-01 -4.98743027e-01
-3.64355713e-01 -7.93815076e-01 8.33948970e-01 -8.03807154e-02
-4.52084482e-01 -1.44466472e+00 2.40387782e-01 -4.23813224e-01
6.05955064e-01 9.14239764e-01 1.20729089e+00 -8.15743148e-01
-7.52395928e-01 -1.76061332e-01 -3.58567506e-01 5.58885634e-02
4.25941385e-02 2.09411383e-01 -3.46943587e-01 -3.86407346e-01
-1.00532997e+00 -1.71887606e-01 6.43165588e-01 3.46062899e-01
8.73213589e-01 -8.02656233e-01 -6.84034288e-01 1.29591554e-01
1.55584824e+00 -2.47629151e-01 7.58299947e-01 -1.02267273e-01
8.60299587e-01 5.92380285e-01 -3.21975112e-01 1.70115963e-01
4.00464177e-01 4.35977608e-01 4.38000053e-01 -2.62257934e-01
-3.72217059e-01 -7.65584767e-01 1.41611770e-01 8.71955097e-01
-7.76999965e-02 -4.97559577e-01 -1.21542454e+00 5.73318541e-01
-1.71320391e+00 -1.04395282e+00 -5.39238811e-01 1.98457372e+00
8.00312519e-01 4.36093688e-01 5.18384099e-01 -1.70706049e-01
1.01895761e+00 1.03784710e-01 -5.21182120e-01 -4.51227099e-01
-1.88776180e-01 5.86310685e-01 6.83287799e-01 7.23332584e-01
-5.27173221e-01 9.06167090e-01 6.82789516e+00 8.54749978e-01
-5.49268186e-01 -4.72523496e-02 8.57140183e-01 1.29870877e-01
-7.14636624e-01 3.42836112e-01 -3.90988231e-01 4.05306518e-01
1.41637456e+00 -6.84028804e-01 6.71315551e-01 8.59650075e-01
3.09745148e-02 2.18943641e-01 -1.04000449e+00 6.05950058e-01
-2.54542947e-01 -1.89509928e+00 4.61329639e-01 5.08988559e-01
9.86921191e-01 1.11306079e-01 -3.93063068e-01 3.89674336e-01
1.33676457e+00 -1.30402780e+00 4.14753228e-01 5.62856972e-01
9.40226972e-01 -8.72292340e-01 3.10403973e-01 4.06164199e-01
-9.81611788e-01 1.41652524e-01 -3.11739355e-01 9.98849049e-03
1.87155783e-01 9.95033801e-01 -1.46232426e+00 5.66484749e-01
2.00224787e-01 5.23700833e-01 -7.56245136e-01 8.65724564e-01
-7.43359476e-02 7.59404421e-01 -4.28304166e-01 -3.92686099e-01
2.07958922e-01 -3.64376903e-01 5.74440539e-01 1.06801462e+00
4.98058647e-01 -2.53515542e-01 2.67889678e-01 1.24974656e+00
-5.23470283e-01 2.46974677e-02 -1.39677989e+00 -2.46050283e-01
5.44522703e-01 1.31222320e+00 -8.91201794e-01 -5.02082884e-01
4.37491387e-02 6.75617993e-01 6.05469942e-01 4.31796521e-01
-6.04449570e-01 -5.07131338e-01 8.03568289e-02 7.79497325e-01
5.79226799e-02 -2.26576179e-01 3.31462830e-01 -8.04241061e-01
-3.92443955e-01 -1.11910665e+00 1.90782130e-01 -9.21808422e-01
-1.65361214e+00 9.40211713e-01 1.17870562e-01 -7.49978423e-01
-3.87861878e-01 -4.74145323e-01 -5.27443886e-01 7.89734602e-01
-8.47101331e-01 -1.01563632e+00 -2.52504617e-01 3.78219336e-01
8.33174586e-02 7.02891946e-02 7.62114286e-01 3.27707902e-02
-3.04888189e-01 1.29184186e-01 4.50954288e-02 6.29961044e-02
2.22661108e-01 -1.53066790e+00 1.19040155e+00 6.49443030e-01
1.58399999e-01 5.42238295e-01 7.45998085e-01 -1.15849102e+00
-1.23980594e+00 -1.46815825e+00 6.51570678e-01 -3.04915398e-01
9.42971587e-01 -6.22088015e-01 -6.29798472e-01 8.30857933e-01
3.27753752e-01 1.80830970e-01 4.37781103e-02 6.04666434e-02
-3.78967047e-01 2.04935104e-01 -1.27340066e+00 9.76194859e-01
1.56759346e+00 -2.32097358e-01 1.75537780e-01 9.04913962e-01
9.46223259e-01 -7.61455391e-03 -1.16665554e+00 1.16427720e-01
1.49109945e-01 -1.03791356e+00 7.50029325e-01 -7.76780725e-01
4.59329307e-01 2.61261370e-02 1.53696597e-01 -1.79676545e+00
-5.39005339e-01 -1.14829409e+00 -7.17899799e-02 1.00330436e+00
8.41647029e-01 -1.15921772e+00 9.94051576e-01 3.84644032e-01
2.72428870e-01 -7.71003544e-01 -6.31668985e-01 -9.02471423e-01
1.54172003e-01 -2.60788277e-02 7.63138413e-01 9.71503854e-01
-2.79212743e-02 9.60292876e-01 -1.23151518e-01 -2.92296410e-01
9.86034930e-01 -1.65901929e-01 1.07243502e+00 -1.68581033e+00
-4.90264863e-01 -5.85052907e-01 -5.58590233e-01 -4.74813372e-01
4.28399332e-02 -1.41169620e+00 -3.95417809e-01 -2.03572249e+00
3.59679937e-01 -6.85007334e-01 4.38921809e-01 3.92952442e-01
5.42499088e-02 2.50568897e-01 4.11117263e-02 1.52491080e-02
-2.94161558e-01 3.12280327e-01 1.47263670e+00 -2.16719672e-01
-4.28734049e-02 -2.18241185e-01 -7.79716969e-01 4.97825891e-01
8.82668138e-01 -7.14747965e-01 -6.06524289e-01 2.11296365e-01
7.07820058e-01 1.79785743e-01 4.65949953e-01 -8.16089213e-01
1.99036021e-02 5.63350953e-02 9.99436900e-02 -2.93551199e-02
6.28340244e-02 -3.58078182e-01 8.07270646e-01 6.38777554e-01
-2.87595779e-01 5.56690335e-01 5.43685667e-02 8.61581028e-01
2.10955858e-01 8.90446156e-02 4.90049511e-01 -5.73247671e-01
-2.26531371e-01 6.08133316e-01 -2.04643741e-01 3.76947075e-01
9.20995712e-01 -2.19951019e-01 -7.80170262e-01 -8.12158704e-01
-1.02288568e+00 -1.67776253e-02 6.40927672e-01 9.64381453e-03
6.01890862e-01 -1.19385648e+00 -1.01099479e+00 -5.34550771e-02
-1.07284226e-01 1.66887507e-01 -8.14027414e-02 7.58305311e-01
-8.43267322e-01 -1.27500787e-01 -2.81248894e-02 -4.24740672e-01
-9.00281489e-01 6.18038833e-01 3.33499700e-01 -7.86558568e-01
-6.67258918e-01 4.75656748e-01 -4.54644207e-04 -9.15389895e-01
-3.13263208e-01 -1.20596588e-01 3.67774785e-01 -3.91930640e-01
-4.21463251e-02 4.19562548e-01 -2.02724338e-01 -1.24371454e-01
-6.13700971e-03 -4.77094799e-02 4.18957099e-02 1.78835224e-02
1.51462138e+00 3.13846350e-01 -4.08940077e-01 1.22658521e-01
7.75172234e-01 2.77779490e-01 -9.24810886e-01 1.21006154e-01
7.33231083e-02 -3.10911760e-02 -4.91073519e-01 -6.83010757e-01
-9.33633327e-01 4.35457766e-01 -1.51627481e-01 1.21373558e+00
6.12898409e-01 1.79217324e-01 4.29595828e-01 3.10320258e-01
6.84547007e-01 -5.47576189e-01 -6.26478950e-03 1.69262752e-01
1.02098918e+00 -7.27840722e-01 2.74202883e-01 -5.86229503e-01
-3.06318313e-01 8.97898912e-01 3.48086625e-01 -6.03168905e-01
7.18472064e-01 1.44543707e-01 -5.42278945e-01 -6.73809350e-01
-9.28688109e-01 -8.11603665e-02 -2.06709597e-02 9.93214965e-01
1.00714728e-01 3.41016412e-01 -2.71649268e-02 -1.56072289e-01
-2.98567086e-01 -1.22502752e-01 9.55004394e-01 4.45318639e-01
-2.30461761e-01 -1.30535984e+00 -1.22207813e-01 9.52239335e-01
-1.20805360e-01 -5.35160422e-01 -8.28875124e-01 1.25662696e+00
-3.21620375e-01 7.30546534e-01 -1.87808141e-01 -2.80100584e-01
2.11680070e-01 -1.92120761e-01 5.85520744e-01 -7.65276253e-01
-5.65340221e-01 -4.30857599e-01 7.93466806e-01 -4.13391173e-01
-1.68002456e-01 -2.34392554e-01 -8.94796133e-01 -9.77132499e-01
-1.93162695e-01 1.90850884e-01 4.19874936e-01 2.41740763e-01
8.61576974e-01 5.46485603e-01 3.16704452e-01 -1.01056528e+00
-4.18276787e-01 -1.12241304e+00 -8.94517601e-01 5.95203340e-01
-1.81613490e-01 -3.42743456e-01 -4.23631847e-01 -3.49849612e-01] | [6.932538986206055, 6.012482166290283] |
6a45b812-6e83-4370-b300-f520b30a154e | a-stepwise-label-based-approach-for-improving | null | null | https://dl.acm.org/doi/abs/10.1145/3347449.3357482 | https://dl.acm.org/doi/abs/10.1145/3347449.3357482 | A Stepwise, Label-based Approach for Improving the Adversarial Training in Unsupervised Video Summarization | In this paper we present our work on improving the efficiency of adversarial training for unsupervised video summarization. Our starting point is the SUM-GAN model, which creates a representative summary based on the intuition that such a summary should make it possible to reconstruct a video that is indistinguishable from the original one. We build on a publicly available implementation of a variation of this model, that includes a linear compression layer to reduce the number of learned parameters and applies an incremental approach for training the different components of the architecture. After assessing the impact of these changes to the model's performance, we propose a stepwise, label-based learning process to improve the training efficiency of the adversarial part of the model. Before evaluating our model's efficiency, we perform a thorough study with respect to the used evaluation protocols and we examine the possible performance on two benchmarking datasets, namely SumMe and TVSum. Experimental evaluations and comparisons with the state of the art highlight the competitiveness of the proposed method. An ablation study indicates the benefit of each applied change on the model's performance, and points out the advantageous role of the introduced stepwise, label-based training strategy on the learning efficiency of the adversarial part of the architecture. | ['Ioannis Patras', 'Vasileios Mezaris', 'Eleni Adamantidou', 'Alexandros I. Metsai', 'Evlampios Apostolidis'] | 2019-10-21 | null | null | null | ai4tv-2019-10 | ['unsupervised-video-summarization'] | ['computer-vision'] | [ 4.89090085e-01 4.56079274e-01 1.37782514e-01 -4.68138307e-02
-7.73106039e-01 -6.11979902e-01 8.51836085e-01 7.68805156e-04
-5.55524468e-01 6.92845225e-01 3.60709667e-01 -2.85429627e-01
7.29177892e-02 -6.12806380e-01 -1.20182049e+00 -6.77902579e-01
-2.57959783e-01 3.98790598e-01 3.24439585e-01 -2.90859073e-01
7.75111765e-02 4.22440141e-01 -1.52853549e+00 4.63189393e-01
6.66830719e-01 8.87862027e-01 -1.57649800e-01 8.15979242e-01
1.63584411e-01 1.17863941e+00 -1.03523588e+00 -7.42282271e-01
4.43298787e-01 -6.36163533e-01 -9.65250850e-01 8.08466226e-02
5.71472764e-01 -5.81195831e-01 -3.20205569e-01 7.61855960e-01
4.94233936e-01 2.89747268e-02 7.55440772e-01 -1.02862608e+00
5.94380172e-03 8.73843610e-01 -1.79441303e-01 1.53828919e-01
2.86539495e-01 3.04065526e-01 8.85925412e-01 -3.65842462e-01
9.14541602e-01 7.21924305e-01 7.88172483e-01 5.37761748e-01
-1.14240277e+00 -4.04361963e-01 7.57483095e-02 2.33128533e-01
-1.22511578e+00 -7.72818267e-01 7.78357387e-01 -3.25554401e-01
8.22154284e-01 4.31945503e-01 4.98045623e-01 1.29157043e+00
-9.14399922e-02 6.29261136e-01 1.06739902e+00 -6.02875650e-01
4.86316681e-01 3.48852754e-01 -2.57873029e-01 5.33396959e-01
1.31311536e-01 1.14696234e-01 -3.62797856e-01 4.61521707e-02
2.54499227e-01 -5.25564730e-01 -3.95646393e-01 -5.84761143e-01
-8.17551136e-01 6.95053458e-01 5.02566159e-01 5.29180944e-01
-3.83923620e-01 3.51482362e-01 7.35226750e-01 4.54688758e-01
5.80081642e-01 4.96419728e-01 -7.66248479e-02 -1.67094409e-01
-1.52115571e+00 2.04007566e-01 9.22859430e-01 7.11520135e-01
5.14748991e-01 7.42535070e-02 -4.86427248e-01 4.87768203e-01
-1.44556955e-01 -1.02202728e-01 4.81841654e-01 -1.09181535e+00
3.39764565e-01 3.74112219e-01 -1.81518212e-01 -7.90455699e-01
-7.44369552e-02 -7.53624499e-01 -6.56299949e-01 4.04948562e-01
2.49708682e-01 -2.88181216e-01 -7.89832532e-01 1.76044953e+00
5.15929200e-02 4.21602130e-01 1.35958225e-01 6.28991425e-01
6.12669170e-01 5.25232255e-01 4.48354110e-02 -2.28034377e-01
8.77492309e-01 -1.06259489e+00 -3.81555408e-01 1.14424266e-02
6.30949795e-01 -6.18280053e-01 8.55231762e-01 3.87056530e-01
-1.45561540e+00 -6.16330385e-01 -1.30906487e+00 1.31799027e-01
-1.22834019e-01 2.28468761e-01 2.05847248e-01 7.88242638e-01
-1.20313871e+00 9.48228955e-01 -7.26057708e-01 -3.56760949e-01
4.44442064e-01 3.48662823e-01 -3.26914817e-01 1.15060151e-01
-1.09938252e+00 8.81939113e-01 6.45148098e-01 -1.05382636e-01
-1.00094748e+00 -5.26938498e-01 -5.07306457e-01 3.53701353e-01
5.03387153e-01 -1.12505233e+00 1.18180287e+00 -1.41000879e+00
-1.66629362e+00 7.60101378e-01 1.62066817e-01 -9.61430728e-01
1.18010175e+00 -6.26905486e-02 3.17503203e-04 5.14843404e-01
-4.42869544e-01 6.72930002e-01 1.15911961e+00 -1.59361291e+00
-4.04545903e-01 9.82459784e-02 3.51491809e-01 -3.87402251e-03
-3.35283071e-01 -2.52479881e-01 -4.06969815e-01 -8.67704809e-01
-5.52860796e-01 -1.09583104e+00 -1.42376209e-02 -5.43156683e-01
-2.66903222e-01 2.54327148e-01 5.33043027e-01 -7.72336006e-01
1.28822422e+00 -2.11536789e+00 4.71819192e-01 2.75108576e-01
2.45212354e-02 5.08117795e-01 -1.58277914e-01 7.03530252e-01
-2.66868412e-01 2.33874053e-01 -5.62121809e-01 -8.51463437e-01
-1.17140956e-01 1.19154148e-01 -2.86739588e-01 3.73557419e-01
2.16156468e-01 6.92187071e-01 -6.06450438e-01 -2.82449305e-01
1.93469569e-01 4.86874759e-01 -8.55601728e-01 5.22190928e-01
-3.08809936e-01 4.54979628e-01 -7.10732117e-02 -3.49625312e-02
5.30644178e-01 1.10911682e-01 2.43506536e-01 -3.80873233e-01
9.39362794e-02 3.22745442e-01 -1.04585695e+00 1.82654381e+00
-5.50743997e-01 4.57480669e-01 -1.84855670e-01 -1.18876493e+00
5.59705079e-01 3.97289872e-01 2.92644799e-01 -5.52520156e-01
2.39532560e-01 1.77478895e-01 3.41863930e-02 -4.75429147e-01
6.00485384e-01 -1.24635242e-01 9.79939625e-02 4.86403972e-01
3.14697087e-01 8.09090585e-02 2.95044333e-01 4.48566407e-01
1.16056013e+00 3.47256273e-01 3.08124602e-01 -1.63776964e-01
9.06479120e-01 -1.51177824e-01 4.53701392e-02 7.39292562e-01
1.10240474e-01 7.21956015e-01 7.47534633e-01 -2.09850058e-01
-1.24440372e+00 -7.21649408e-01 3.62962872e-01 9.00744081e-01
-2.96713591e-01 -6.30990088e-01 -1.23905301e+00 -9.93786037e-01
-4.32069957e-01 9.91691709e-01 -8.28359842e-01 -4.82827485e-01
-7.65107572e-01 -7.16027677e-01 7.95171261e-01 2.72026092e-01
5.28830111e-01 -1.05441368e+00 -9.94543612e-01 -1.16395839e-01
-1.85963199e-01 -1.07382405e+00 -8.09495524e-02 6.61320835e-02
-8.42009902e-01 -8.70949209e-01 -5.61338365e-01 -4.49728191e-01
4.53027815e-01 -2.07559735e-01 1.31590700e+00 2.60464191e-01
2.02887014e-01 5.20511568e-01 -7.38218904e-01 -2.70124376e-01
-1.22794259e+00 6.48213744e-01 -3.01242620e-01 -7.25976750e-03
-3.65593344e-01 -8.77968311e-01 -5.78716099e-01 -1.47991866e-01
-1.36514533e+00 4.72203232e-02 6.55822754e-01 6.64642096e-01
2.52336293e-01 1.37816980e-01 4.12000239e-01 -1.30999851e+00
4.79155809e-01 -4.01956141e-01 -2.38063842e-01 1.42837614e-01
-6.07051134e-01 4.24200773e-01 9.08563852e-01 -1.94647431e-01
-8.52030337e-01 -4.77292500e-02 -4.11815882e-01 -5.27719080e-01
-6.04801662e-02 3.55779111e-01 -3.66173647e-02 -1.27154917e-01
5.27649760e-01 3.69412273e-01 5.92187941e-02 -4.37849015e-01
4.12706852e-01 3.74478370e-01 7.00682461e-01 -3.83367956e-01
9.70825374e-01 4.34135675e-01 6.57976493e-02 -5.43215871e-01
-4.10916001e-01 -1.26560181e-01 -5.03352940e-01 -2.19738513e-01
6.30786061e-01 -7.67392874e-01 -4.26653206e-01 5.18976331e-01
-1.11868858e+00 -3.79269868e-01 -7.29163170e-01 1.04322433e-01
-8.23327124e-01 5.20023048e-01 -3.35713953e-01 -4.72686052e-01
-5.60286522e-01 -1.09325337e+00 8.63916218e-01 -1.21350102e-01
-5.79675511e-02 -1.03588772e+00 2.75480807e-01 4.44635242e-01
6.92810833e-01 4.71149564e-01 8.90050054e-01 -1.00400591e+00
-4.16103601e-01 -1.93660229e-01 2.36505091e-01 1.02590322e+00
-2.42390901e-01 3.43550034e-02 -1.27977943e+00 -5.27023077e-01
1.80792451e-01 -2.35269085e-01 1.22834158e+00 -7.44241755e-03
9.13691103e-01 -5.88984132e-01 1.11931503e-01 7.37233937e-01
1.46098602e+00 -2.55440950e-01 1.04909348e+00 5.97977817e-01
6.12136841e-01 4.78898317e-01 3.22947621e-01 3.19776326e-01
1.03845552e-01 8.48836005e-01 7.64569521e-01 -9.50179622e-02
-4.38576311e-01 -2.58507431e-01 7.15861857e-01 8.55314553e-01
-3.53869647e-01 -3.87812853e-01 -4.74887282e-01 2.58046180e-01
-1.54586864e+00 -1.08022869e+00 3.02379131e-01 2.50938082e+00
5.53986371e-01 4.19246703e-01 4.01937693e-01 3.55875522e-01
3.53792489e-01 4.38287884e-01 -1.38778314e-01 -7.92106867e-01
-1.18757449e-02 4.62982357e-01 5.34137785e-01 4.27480102e-01
-1.06289124e+00 6.91727936e-01 6.44896269e+00 9.91765499e-01
-1.33898747e+00 1.89506575e-01 5.18534064e-01 -1.71450913e-01
-1.90377757e-01 -1.22848935e-02 -2.10444003e-01 5.50104439e-01
1.27749610e+00 -1.33093819e-01 6.37027204e-01 6.36752605e-01
4.95758541e-02 -4.94733304e-02 -1.31162429e+00 3.16139191e-01
4.25381362e-01 -1.38019705e+00 3.57248276e-01 -4.58901236e-03
6.32946491e-01 2.86404155e-02 -3.20379399e-02 4.33737338e-01
-4.71741334e-02 -8.80970001e-01 1.05894458e+00 5.21062970e-01
7.32650638e-01 -8.91103923e-01 1.00806534e+00 3.21269631e-01
-6.83555603e-01 -7.12291449e-02 -1.14503011e-01 -1.46217011e-02
-1.13097150e-02 2.49957100e-01 -9.98214960e-01 9.24834073e-01
3.31934720e-01 3.44176739e-01 -1.01328027e+00 7.42686152e-01
-2.96205908e-01 8.93678427e-01 -1.80804178e-01 3.99259597e-01
2.31413826e-01 -8.66858214e-02 7.51984954e-01 1.49428618e+00
2.16579691e-01 -3.45804572e-01 -1.70546338e-01 5.85650086e-01
-3.84690553e-01 1.66830927e-01 -5.24221241e-01 1.04561292e-01
2.58888841e-01 1.08961058e+00 -5.12792110e-01 -4.85708654e-01
-1.85213178e-01 1.19665253e+00 3.33217263e-01 6.38833791e-02
-1.12799799e+00 -4.78165876e-03 8.69672075e-02 3.47033709e-01
7.58196175e-01 1.34019658e-01 -7.12912679e-02 -1.02268624e+00
2.59440720e-01 -1.30477989e+00 3.81272674e-01 -5.66545188e-01
-6.59539998e-01 8.17635775e-01 4.67514656e-02 -1.24327159e+00
-6.80585027e-01 2.61464026e-02 -7.42716789e-01 7.81917810e-01
-1.38870645e+00 -1.21427548e+00 -2.58457631e-01 3.62437487e-01
3.45853537e-01 -1.97978228e-01 8.13055456e-01 1.96664721e-01
-5.28681397e-01 9.41271007e-01 7.58601874e-02 -2.19323263e-01
6.10403478e-01 -1.19783223e+00 3.46236587e-01 1.21291840e+00
1.60930857e-01 3.97697061e-01 1.05649972e+00 -2.30688408e-01
-1.08872139e+00 -1.07515705e+00 5.95178962e-01 -3.90015095e-01
4.63151038e-01 -2.16491207e-01 -7.99519479e-01 7.09446669e-01
5.25472701e-01 -3.67448658e-01 4.38183904e-01 -2.88117558e-01
-2.95160264e-01 -2.36659765e-01 -1.18127429e+00 4.26960886e-01
9.02798653e-01 -2.56915420e-01 -6.32114530e-01 7.20142797e-02
6.56211853e-01 -2.39349172e-01 -8.49648893e-01 5.07878304e-01
5.24572194e-01 -1.59319639e+00 8.97123098e-01 -3.99652749e-01
8.93128753e-01 -1.32604584e-01 4.54584211e-02 -1.40460527e+00
-1.61753982e-01 -5.97941101e-01 -2.84828246e-01 1.39402950e+00
1.87153071e-01 -4.05584872e-01 7.65093923e-01 1.38367340e-01
-2.01873019e-01 -8.06647360e-01 -1.03632045e+00 -6.99483156e-01
1.73886359e-01 -1.24848977e-01 3.99502188e-01 5.78469992e-01
-4.36523676e-01 4.13965970e-01 -6.40575230e-01 8.36980529e-03
2.67370701e-01 -1.23258367e-01 1.11415350e+00 -7.80345619e-01
-7.84620106e-01 -3.86618555e-01 -4.97963250e-01 -5.17824411e-01
1.04047209e-01 -8.75232577e-01 -2.71039486e-01 -1.31021559e+00
1.88032433e-01 -1.11699440e-01 -2.73497343e-01 2.30206892e-01
-1.22814462e-01 4.18524176e-01 7.31812835e-01 4.10515547e-01
-5.99741518e-01 4.12953556e-01 7.07176566e-01 -7.62896761e-02
-3.46464664e-02 -2.25801440e-03 -7.09359109e-01 5.40678918e-01
6.69415593e-01 -6.19662523e-01 -4.43354756e-01 -4.52496022e-01
2.09881186e-01 -2.06643805e-01 3.72348487e-01 -1.37622106e+00
7.29531050e-02 5.64845979e-01 9.83681809e-03 -3.79972421e-02
1.89515665e-01 -1.04398227e+00 4.07468855e-01 8.04144919e-01
-4.51320857e-01 -1.70838833e-02 2.80244172e-01 3.08175057e-01
-2.13091195e-01 -5.02927840e-01 8.37830961e-01 4.53718603e-02
-4.55229819e-01 -2.04546034e-01 -1.15035601e-01 -1.23599410e-01
1.09883118e+00 -1.26793325e-01 -1.66389033e-01 -4.97845173e-01
-5.52914202e-01 -1.86740503e-01 8.79121661e-01 1.47015661e-01
9.27152485e-02 -8.52257192e-01 -9.20603752e-01 1.10837281e-01
-1.45720214e-01 -3.03724736e-01 3.25929850e-01 8.45428467e-01
-8.11843276e-01 2.28838772e-01 -3.93918097e-01 -3.00450653e-01
-1.25177026e+00 9.20476198e-01 4.37099546e-01 -7.46670187e-01
-4.78425831e-01 4.66083676e-01 -1.30182877e-01 -3.63381505e-02
2.88768977e-01 -1.23933539e-01 -3.13891023e-01 4.67959382e-02
2.20450163e-01 5.83049893e-01 4.10026312e-01 -5.67105532e-01
-2.47265950e-01 2.85490572e-01 -9.85556245e-02 -1.66758165e-01
1.38345253e+00 -3.02279815e-02 7.78266191e-02 3.59783858e-01
1.11521316e+00 3.02051604e-01 -1.23870790e+00 2.02669755e-01
-3.30442190e-01 -1.14735879e-01 1.20847318e-02 -7.90129840e-01
-1.13726544e+00 7.00564325e-01 5.61477602e-01 2.97458023e-01
1.59052825e+00 -2.24385038e-01 6.23155892e-01 1.06366687e-01
2.47673988e-01 -7.37852812e-01 -1.84747994e-01 1.21274166e-01
7.72924721e-01 -8.30684245e-01 3.12550157e-01 -4.79481816e-01
-6.24389052e-01 1.04552913e+00 1.82085354e-02 -4.74360555e-01
1.67580068e-01 1.75463900e-01 -1.56699851e-01 -2.22333651e-02
-7.66103387e-01 8.82929340e-02 1.69098616e-01 3.38045657e-01
2.98005521e-01 -2.03065306e-01 -6.03375256e-01 3.35203230e-01
-3.34835678e-01 2.10168213e-01 7.82999396e-01 8.36902022e-01
-8.50063264e-02 -1.28795230e+00 -9.02097449e-02 1.75131410e-01
-6.97446942e-01 -8.98964107e-02 -4.91550982e-01 9.36227083e-01
2.50313491e-01 6.40429020e-01 -1.05578437e-01 -4.46182489e-01
4.08430874e-01 2.10942060e-01 6.61227345e-01 -3.43815535e-01
-1.06153917e+00 -1.80822119e-01 2.52988756e-01 -6.10596299e-01
-7.98975348e-01 -7.48018146e-01 -7.18341291e-01 -3.95468861e-01
5.47248386e-02 2.76014239e-01 7.78412044e-01 9.85374749e-01
4.03115779e-01 9.11226630e-01 7.61600792e-01 -8.63302171e-01
-8.68165553e-01 -9.77358639e-01 -1.54714555e-01 7.30720639e-01
1.41307727e-01 -3.00832063e-01 -3.81503433e-01 3.81631374e-01] | [10.498174667358398, 0.33130115270614624] |
5bbe047d-b6b5-4bff-bb66-7f909facbfb7 | real-world-font-recognition-using-deep | 1504.00028 | null | http://arxiv.org/abs/1504.00028v1 | http://arxiv.org/pdf/1504.00028v1.pdf | Real-World Font Recognition Using Deep Network and Domain Adaptation | We address a challenging fine-grain classification problem: recognizing a
font style from an image of text. In this task, it is very easy to generate
lots of rendered font examples but very hard to obtain real-world labeled
images. This real-to-synthetic domain gap caused poor generalization to new
real data in previous methods (Chen et al. (2014)). In this paper, we refer to
Convolutional Neural Networks, and use an adaptation technique based on a
Stacked Convolutional Auto-Encoder that exploits unlabeled real-world images
combined with synthetic data. The proposed method achieves an accuracy of
higher than 80% (top-5) on a real-world dataset. | ['Jianchao Yang', 'Aseem Agarwala', 'Zhangyang Wang', 'Thomas S. Huang', 'Hailin Jin', 'Eli Shechtman', 'Jonathan Brandt'] | 2015-03-31 | null | null | null | null | ['font-recognition'] | ['computer-vision'] | [ 6.35572851e-01 -9.82040092e-02 3.04806978e-01 -5.64223528e-01
-5.77583730e-01 -8.49481046e-01 6.42284274e-01 -4.03394133e-01
-2.23715305e-01 1.00661731e+00 -1.93983659e-01 -3.08960825e-01
3.92668933e-01 -9.02171314e-01 -1.13751042e+00 -2.12486610e-01
5.47455728e-01 4.46700543e-01 4.58329171e-02 -2.58468121e-01
3.34252030e-01 4.95952964e-01 -1.54709852e+00 7.31177449e-01
1.03710032e+00 1.09701514e+00 1.69966951e-01 8.28736365e-01
-4.16702569e-01 8.35798562e-01 -1.14968085e+00 -7.44107306e-01
5.20322561e-01 -2.82547265e-01 -7.32873499e-01 7.26713479e-01
7.96316564e-01 -3.88423800e-01 -4.21305634e-02 1.24628353e+00
3.14659894e-01 -1.27360404e-01 8.96182835e-01 -1.08880210e+00
-1.45281208e+00 5.29746711e-01 -6.86625302e-01 -1.88165024e-01
2.33716667e-01 2.16821879e-01 4.98164654e-01 -9.82195497e-01
6.64059758e-01 1.06392026e+00 4.58148748e-01 4.73723292e-01
-1.15325367e+00 -8.04917097e-01 -5.08480296e-02 -2.18158185e-01
-1.18691742e+00 -2.86741666e-02 8.40708077e-01 -6.79414511e-01
3.52504641e-01 -2.33860333e-02 4.80832934e-01 1.71647990e+00
-4.43276539e-02 8.11653972e-01 1.72135603e+00 -6.40991449e-01
1.88051224e-01 3.06183308e-01 2.41382439e-02 5.13757706e-01
1.91785350e-01 1.36801228e-02 -2.96633512e-01 2.97050476e-01
8.67349803e-01 -1.08686291e-01 -1.60430118e-01 -2.66228616e-01
-1.33294034e+00 7.91108668e-01 4.93056923e-01 2.25922585e-01
-1.13210455e-01 -3.39297891e-01 4.36005831e-01 4.72989976e-01
5.22945046e-01 6.96197271e-01 -3.94336224e-01 -1.82167307e-01
-8.13613892e-01 2.75204182e-01 6.73264921e-01 1.17835450e+00
5.81036687e-01 3.97709757e-01 -1.42712042e-01 1.00166774e+00
-3.94119889e-01 6.80019379e-01 8.92362893e-01 -5.80792308e-01
7.40699589e-01 7.29506671e-01 2.03778163e-01 -5.44064462e-01
-8.27878863e-02 -5.11118233e-01 -1.12424743e+00 4.13291216e-01
6.84603035e-01 -2.58156389e-01 -1.00952005e+00 1.40629816e+00
-1.55242518e-01 -7.51981661e-02 1.67862058e-01 8.43823671e-01
6.81793511e-01 6.93625212e-01 -1.43007234e-01 1.31657049e-01
1.11303413e+00 -1.33655310e+00 -4.96740431e-01 -4.61025059e-01
9.70573798e-02 -8.63360763e-01 1.78001308e+00 7.23092854e-01
-7.22503662e-01 -1.12539852e+00 -1.29229105e+00 1.02683149e-01
-8.96359801e-01 5.39042652e-01 4.94967192e-01 7.78886080e-01
-7.01314509e-01 5.26073456e-01 1.18827395e-01 -1.85267746e-01
4.72954780e-01 1.22400385e-03 -5.18305123e-01 -2.25337297e-01
-1.14006639e+00 5.05696297e-01 6.32192969e-01 -1.58441588e-01
-4.84908223e-01 -4.20747995e-01 -6.45292819e-01 -8.38797614e-02
2.86822736e-01 -2.97791600e-01 1.38706338e+00 -1.63672376e+00
-1.50340819e+00 9.44278359e-01 3.57472330e-01 -4.53805983e-01
7.75526583e-01 -2.09431887e-01 -8.53082359e-01 -1.38713092e-01
5.16787823e-03 6.49671078e-01 1.39885950e+00 -1.43665445e+00
-4.51245964e-01 -3.01344454e-01 6.44075647e-02 3.98295484e-02
-7.74673998e-01 -2.44361848e-01 -3.36805172e-02 -1.05044210e+00
-3.46439958e-01 -8.62924695e-01 2.12617010e-01 -7.37784654e-02
-5.07292509e-01 -7.31947869e-02 8.34403455e-01 -6.52365208e-01
8.53504717e-01 -2.04432034e+00 -1.77795380e-01 -2.16868654e-01
1.55984968e-01 6.06923938e-01 -3.28273177e-01 1.80653393e-01
-3.64835590e-01 1.19909495e-01 -2.34095842e-01 4.89350855e-02
2.12241150e-03 -9.43878070e-02 -4.46496367e-01 -9.84659418e-02
4.74746257e-01 8.46145749e-01 -7.99565196e-01 -2.65564859e-01
2.00663865e-01 1.08060218e-01 1.15445159e-01 2.73224145e-01
-4.56395745e-01 2.64582098e-01 -4.03698385e-02 6.38081849e-01
8.63854766e-01 -6.01094604e-01 -9.45888609e-02 -1.48773074e-01
1.39410704e-01 -4.57703680e-01 -1.19149756e+00 1.51190054e+00
-6.15628362e-01 8.06574404e-01 -5.72172582e-01 -6.64614201e-01
1.37056112e+00 -4.65535000e-02 -2.23707110e-01 -5.92577457e-01
2.70137817e-01 3.13352317e-01 -4.64610569e-02 -3.21371406e-01
7.26678133e-01 1.35409951e-01 -2.32344821e-01 4.73164111e-01
4.64266865e-03 -4.25526381e-01 5.76233923e-01 -2.77446173e-02
8.11445773e-01 2.78048337e-01 2.42547721e-01 -2.50831366e-01
5.56072116e-01 -2.10958514e-02 2.20378935e-01 6.70792460e-01
-8.33704174e-02 9.51548040e-01 3.88752699e-01 -8.64089012e-01
-1.62100875e+00 -9.33309913e-01 -2.47317851e-02 8.72456551e-01
-2.09752232e-01 -1.21032096e-01 -1.09342492e+00 -9.00334597e-01
2.18510330e-02 6.48552001e-01 -9.38424051e-01 -2.27351300e-02
-4.42627132e-01 -5.02678752e-01 5.31241953e-01 7.81194270e-01
9.53486800e-01 -1.31304932e+00 -3.91241580e-01 1.35554537e-01
2.26350105e-03 -1.34411657e+00 -4.95720983e-01 2.05747649e-01
-5.27273476e-01 -9.80834067e-01 -1.16881692e+00 -9.81422365e-01
6.88244164e-01 1.11053385e-01 1.28781879e+00 -3.71177793e-01
-3.41089278e-01 -3.15587580e-01 -5.56560397e-01 -6.97581172e-01
-7.38498926e-01 8.84957388e-02 -1.03005588e-01 4.26582873e-01
4.88474637e-01 -1.83727220e-02 -3.12140912e-01 2.76243567e-01
-1.16427255e+00 4.60550874e-01 6.70668423e-01 1.21705890e+00
4.06163692e-01 5.08555658e-02 5.05283892e-01 -1.13016474e+00
9.46412742e-01 5.24714217e-02 -7.97916889e-01 4.78417963e-01
-5.74720621e-01 1.46298185e-01 1.31822717e+00 -8.69888067e-01
-1.05496156e+00 3.00223619e-01 1.12147219e-01 -5.19967079e-01
-4.08344686e-01 -6.75685033e-02 -7.06868768e-02 7.00632557e-02
9.97887850e-01 3.61141920e-01 -3.08922499e-01 -4.78294134e-01
3.94943982e-01 1.24615288e+00 5.96425712e-01 -6.78971350e-01
8.94604445e-01 1.73192725e-01 -1.66522995e-01 -6.94251597e-01
-1.04677463e+00 3.58605325e-01 -9.62252557e-01 -1.28653288e-01
6.99073613e-01 -8.86073112e-01 -2.73973703e-01 9.08223391e-01
-1.00377965e+00 -5.26493371e-01 -3.95151675e-01 1.14772893e-01
-6.20216668e-01 2.94005334e-01 -4.87223983e-01 -5.42900741e-01
-2.00675830e-01 -1.19704449e+00 1.29532540e+00 3.37950140e-01
-5.52627221e-02 -7.20195293e-01 -2.13983178e-01 2.94736952e-01
4.19014603e-01 4.22864974e-01 7.52591670e-01 -4.16834295e-01
-3.72420698e-01 -2.48536855e-01 -7.01450646e-01 7.18075931e-01
3.23423892e-01 2.09600076e-01 -1.16777861e+00 -2.23678634e-01
-3.30451518e-01 -9.41190839e-01 6.21024370e-01 -1.36206672e-01
1.83750606e+00 -1.38353989e-01 3.22587430e-01 5.08958638e-01
1.41942525e+00 1.97693720e-01 5.96005261e-01 4.87774312e-01
6.98510826e-01 4.74597335e-01 5.95395267e-01 4.13229078e-01
-7.99470618e-02 4.39417869e-01 -1.87658332e-02 -3.48051190e-01
-3.84848446e-01 -3.80589932e-01 -2.79884301e-02 4.85806942e-01
2.19864875e-01 -5.03226519e-01 -8.27385783e-01 3.24976206e-01
-1.42636704e+00 -6.64971590e-01 2.50226259e-02 2.03316450e+00
9.83785093e-01 5.45841455e-01 2.76080035e-02 2.37427130e-01
1.09544599e+00 9.70518067e-02 -7.30949700e-01 -7.47318089e-01
-5.45953274e-01 3.69295299e-01 4.33027804e-01 6.70377258e-03
-1.24149919e+00 1.04664981e+00 6.15054798e+00 8.63696277e-01
-1.21654260e+00 -2.87051052e-01 1.18617213e+00 4.48686838e-01
8.96611065e-02 -4.87682194e-01 -6.63000822e-01 8.01414013e-01
9.09130871e-01 -7.32151559e-03 6.47053897e-01 9.96026516e-01
-3.50048304e-01 2.63671458e-01 -1.03054607e+00 1.28777313e+00
5.00734091e-01 -1.21795130e+00 2.90994108e-01 -1.81186259e-01
1.08273923e+00 -3.89874548e-01 3.05256188e-01 3.97786558e-01
6.03005767e-01 -1.26477158e+00 5.95716178e-01 4.74215597e-01
1.49298263e+00 -5.52575767e-01 4.87034202e-01 3.93864006e-01
-9.67793345e-01 -3.00430119e-01 -5.44047534e-01 7.29808817e-03
-2.41875455e-01 6.71847761e-01 -9.34922040e-01 3.43481094e-01
7.78792500e-01 6.02619648e-01 -1.12934053e+00 5.73777676e-01
-3.16203952e-01 3.83516163e-01 6.24390766e-02 -3.27961773e-01
2.38482043e-01 -6.08005524e-02 -1.54508576e-01 1.04746723e+00
3.54437053e-01 -4.21516418e-01 3.44313160e-02 9.41203356e-01
-6.32097781e-01 1.79252312e-01 -7.99729526e-01 -2.85340548e-01
6.19354807e-02 1.33422840e+00 -6.95149064e-01 -7.14020431e-01
-5.40153086e-01 1.38796365e+00 5.41486323e-01 2.57122636e-01
-6.45552218e-01 -7.40141213e-01 -1.98726077e-02 -1.96265891e-01
2.58191943e-01 8.75449926e-02 -6.09183431e-01 -1.33816099e+00
2.25872844e-01 -1.24133193e+00 9.87181142e-02 -1.32006502e+00
-1.70048881e+00 1.02436841e+00 -3.85693252e-01 -1.70040834e+00
-2.93901235e-01 -1.21550930e+00 -4.49030310e-01 1.06226313e+00
-1.18381631e+00 -1.30044925e+00 -7.75023103e-01 4.64414597e-01
1.05131221e+00 -6.61539614e-01 7.85156846e-01 -1.12987965e-01
-2.97390521e-01 6.55089974e-01 4.90058154e-01 5.63287556e-01
1.09320867e+00 -1.59001350e+00 9.11773264e-01 8.05439830e-01
1.18419133e-01 2.81945527e-01 3.89923275e-01 -6.02697492e-01
-9.41658080e-01 -1.42636287e+00 5.84425867e-01 -4.10410047e-01
6.55730307e-01 -6.67664826e-01 -8.63208771e-01 5.11851192e-01
5.77114403e-01 1.30224317e-01 5.31399071e-01 -2.98084229e-01
-8.15310419e-01 -2.54446387e-01 -1.21585178e+00 7.09244311e-01
9.18821275e-01 -2.87533730e-01 -5.04627287e-01 5.13767958e-01
6.25517786e-01 -4.12458420e-01 -8.74387622e-01 3.37912321e-01
4.69380587e-01 -9.81725752e-01 6.80064678e-01 -6.19071424e-01
1.01428854e+00 -1.58383429e-01 -1.65522233e-01 -1.63174915e+00
-2.66688108e-01 -3.00594598e-01 -2.73115467e-02 1.01587272e+00
4.81581062e-01 -4.69028562e-01 7.88032234e-01 2.81633347e-01
1.95593357e-01 -5.43790340e-01 -2.57053643e-01 -1.05688155e+00
4.77335393e-01 4.70095910e-02 1.12452912e+00 9.49355066e-01
-4.09195751e-01 5.21951795e-01 -5.39487243e-01 -2.59776354e-01
5.83922863e-01 3.88168812e-01 1.08459401e+00 -1.37277877e+00
-4.02162075e-01 -3.27864766e-01 -2.73863524e-01 -7.88041770e-01
2.95524478e-01 -6.07586384e-01 -2.27863312e-01 -1.27337694e+00
2.32123762e-01 -8.40180591e-02 -1.25736088e-01 1.98983714e-01
-1.98194042e-01 6.34829402e-01 3.44245285e-01 -2.54277233e-02
-3.90156090e-01 4.66802359e-01 1.66790092e+00 -5.50170422e-01
2.30991483e-01 -2.18739714e-02 -6.99508369e-01 4.35349882e-01
1.05906785e+00 1.28875911e-01 -3.41571301e-01 -5.51528275e-01
1.06010154e-01 -5.79235982e-03 1.96126193e-01 -1.26059008e+00
-3.30577582e-01 -1.38381228e-01 1.20732737e+00 -5.96655786e-01
2.52291024e-01 -8.04046094e-01 -1.96221188e-01 3.15629333e-01
-5.47805309e-01 1.99484155e-01 1.42008156e-01 3.44173759e-01
-3.05588663e-01 -4.58413064e-01 9.11908388e-01 -2.57311106e-01
-8.29214215e-01 1.81282237e-01 -5.87276667e-02 4.61302772e-02
1.07576752e+00 -2.99909949e-01 -6.71419501e-01 -3.52632493e-01
-5.08351088e-01 -2.60150433e-01 6.33884966e-01 8.25719893e-01
5.12767375e-01 -1.48460484e+00 -8.31412911e-01 4.92588460e-01
5.60140431e-01 -1.27543673e-01 1.42154932e-01 -8.96752719e-03
-7.01146305e-01 3.54336768e-01 -6.66497350e-01 -4.25917059e-01
-1.01861215e+00 1.01705635e+00 1.35863617e-01 -1.92368478e-01
-4.98869956e-01 6.48985445e-01 1.37563691e-01 -6.06080592e-01
1.02903759e-02 -2.26357669e-01 -8.17321688e-02 -2.00948209e-01
6.39505982e-01 3.76358479e-02 2.23059669e-01 -4.29595888e-01
3.31190765e-01 5.37771821e-01 -2.33199239e-01 -5.68669140e-02
9.81794596e-01 2.87258506e-01 3.18093866e-01 7.01935828e-01
1.26681578e+00 -2.32368812e-01 -1.47966385e+00 -3.82028133e-01
-2.02513024e-01 -8.09002459e-01 -4.50840801e-01 -1.29405320e+00
-9.20738757e-01 1.14733732e+00 1.00227630e+00 2.53132582e-01
1.12385333e+00 -5.40800869e-01 7.05996037e-01 6.85081244e-01
3.25019449e-01 -1.50122499e+00 3.98130804e-01 5.33112228e-01
9.07278359e-01 -1.68499112e+00 -2.88729072e-01 -3.14737707e-01
-9.61418986e-01 1.52602386e+00 1.11668909e+00 -9.42844823e-02
3.15976650e-01 3.26403201e-01 2.85964638e-01 3.39722425e-01
-5.22814631e-01 7.27295354e-02 1.42836660e-01 6.94747686e-01
4.71768856e-01 2.50956625e-01 1.44689651e-02 4.90073025e-01
-4.82071042e-01 2.63887823e-01 8.22800517e-01 8.36778104e-01
-2.37065300e-01 -1.17812586e+00 -5.44484496e-01 7.83258498e-01
-1.80538893e-01 -9.42395031e-02 -7.01559663e-01 7.80516684e-01
1.09903455e-01 7.29347289e-01 2.02527508e-01 -4.84424502e-01
2.71799743e-01 3.59741062e-01 5.00801623e-01 -4.31911916e-01
-3.50658894e-01 -1.44643098e-01 -1.03892893e-01 4.83790375e-02
-1.67117164e-01 -4.53520983e-01 -5.01186848e-01 -1.43373162e-01
-6.92091882e-02 -1.39122427e-01 7.54693389e-01 5.62859654e-01
7.10343421e-02 8.04020405e-01 1.04673135e+00 -6.49517238e-01
-7.51024485e-01 -1.26254594e+00 -7.45454490e-01 8.82996380e-01
9.19694304e-02 -5.53874731e-01 -4.76538017e-02 2.95064569e-01] | [11.901555061340332, 1.8166049718856812] |
ea3acaec-2e1a-435e-9dce-b523a8e25905 | spectral-reconstruction-from-dispersive-blur | null | null | http://openaccess.thecvf.com/content_CVPR_2019/html/Zhao_Spectral_Reconstruction_From_Dispersive_Blur_A_Novel_Light_Efficient_Spectral_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Zhao_Spectral_Reconstruction_From_Dispersive_Blur_A_Novel_Light_Efficient_Spectral_CVPR_2019_paper.pdf | Spectral Reconstruction From Dispersive Blur: A Novel Light Efficient Spectral Imager | Developing high light efficiency imaging techniques to retrieve high dimensional optical signal is a long-term goal in computational photography. Multispectral imaging, which captures images of different wavelengths and boosting the abilities for revealing scene properties, has developed rapidly in the last few decades. From scanning method to snapshot imaging, the limit of light collection efficiency is kept being pushed which enables wider applications especially under the light-starved scenes. In this work, we propose a novel multispectral imaging technique, that could capture the multispectral images with a high light efficiency. Through investigating the dispersive blur caused by spectral dispersers and introducing the difference of blur (DoB) constraints, we propose a basic theory for capturing multispectral information from a single dispersive-blurred image and an additional spectrum of an arbitrary point in the scene. Based on the theory, we design a prototype system and develop an optimization algorithm to realize snapshot multispectral imaging. The effectiveness of the proposed method is verified on both the synthetic data and real captured images.
| [' Xun Cao', ' Tao Yue', ' Zhan Ma', ' Hui Guo', ' Xuemei Hu', 'Yuanyuan Zhao'] | 2019-06-01 | null | null | null | cvpr-2019-6 | ['spectral-reconstruction'] | ['computer-vision'] | [ 5.76788723e-01 -1.04959345e+00 5.00955462e-01 -1.04564041e-01
-2.56934434e-01 -5.75699508e-01 3.15397620e-01 -8.70209813e-01
-2.62905687e-01 7.90724337e-01 1.60367355e-01 -3.89846824e-02
-6.81661606e-01 -4.63695139e-01 -2.65099287e-01 -1.14396012e+00
2.12546825e-01 -3.93783033e-01 2.50518769e-01 3.35593149e-02
3.95056576e-01 5.44704974e-01 -1.61219358e+00 -1.81773435e-02
1.17490280e+00 8.36331785e-01 7.52030551e-01 6.38611853e-01
1.34997234e-01 6.06470883e-01 -3.95900607e-01 6.09162785e-02
5.63764989e-01 -5.20267189e-01 -3.41460764e-01 4.37041819e-01
2.46580973e-01 -8.27202499e-01 -4.28989530e-01 1.48314250e+00
6.28286183e-01 5.02391495e-02 4.48530018e-01 -7.85505891e-01
-9.51650500e-01 -8.54927003e-02 -1.11766934e+00 3.73683572e-01
7.16429204e-02 4.08735216e-01 4.24703449e-01 -8.27006936e-01
1.83729827e-01 7.41618216e-01 5.11942983e-01 2.10599259e-01
-7.22926974e-01 -2.80948818e-01 -5.92016935e-01 3.83979678e-01
-1.24794471e+00 -5.22915363e-01 1.12350535e+00 -2.70922482e-01
2.71335840e-01 4.76952940e-01 7.41082013e-01 2.64854729e-01
2.59051800e-01 2.17598423e-01 1.80337143e+00 -5.93134105e-01
-1.46845698e-01 9.70771462e-02 1.33324251e-01 5.79428315e-01
5.70285738e-01 4.73491788e-01 -2.08736569e-01 -7.97258243e-02
8.31190705e-01 3.42413783e-01 -8.79204273e-01 -8.80283341e-02
-9.77178752e-01 2.04075485e-01 3.14946592e-01 3.53868484e-01
-4.15943325e-01 -1.64174005e-01 -2.85160035e-01 -2.26333067e-02
3.93530309e-01 3.89212638e-01 1.14126336e-02 1.27618343e-01
-8.88494551e-01 -8.88245329e-02 3.23057741e-01 5.64976156e-01
7.69861937e-01 3.43946107e-02 9.98781249e-03 1.05864704e+00
2.82401323e-01 9.77462530e-01 4.93683159e-01 -1.07468522e+00
4.30195266e-03 3.20680261e-01 4.29830968e-01 -7.64347672e-01
-2.88352787e-01 -5.85796773e-01 -9.55573618e-01 2.34855533e-01
1.51378557e-01 -3.47474009e-01 -5.28354645e-01 1.24584734e+00
3.46254557e-01 3.15986335e-01 5.27836233e-02 1.41578746e+00
3.81452709e-01 9.92384493e-01 -6.21341169e-01 -9.75515187e-01
1.27138257e+00 -7.34514058e-01 -7.42449701e-01 2.92810285e-03
-5.32770827e-02 -1.20925283e+00 8.49979103e-01 4.93385762e-01
-1.11003757e+00 -3.82537276e-01 -8.80717576e-01 1.22682989e-01
1.28658444e-01 4.60007846e-01 5.17129302e-01 5.48293054e-01
-8.92585993e-01 2.71591902e-01 -3.22396129e-01 -2.55918086e-01
6.99759349e-02 -1.27841383e-01 2.91986912e-02 -4.25255090e-01
-8.44981015e-01 7.64388263e-01 6.19160868e-02 2.37052485e-01
-5.76256275e-01 -7.95283914e-01 -8.46662447e-02 -1.81769684e-01
3.13458025e-01 -7.94984281e-01 8.55983675e-01 -7.60606229e-01
-1.59748816e+00 6.58119977e-01 -2.11711004e-01 1.08013041e-01
2.84668893e-01 -8.05634707e-02 -7.59940684e-01 4.83874112e-01
5.48643991e-02 -4.03434724e-01 8.51593673e-01 -1.18548203e+00
-4.17488515e-01 -4.97311950e-01 4.29089516e-02 5.23605108e-01
-2.71808177e-01 2.29491279e-01 -3.49090621e-02 -1.54084623e-01
1.86456293e-01 -7.74525404e-01 -7.31123090e-02 4.35160771e-02
-4.00140256e-01 5.68671048e-01 1.17553031e+00 -3.39203268e-01
9.44631100e-01 -2.19092989e+00 -1.59121007e-01 -3.54557544e-01
7.43041327e-03 6.14184678e-01 1.40531445e-02 6.05641365e-01
9.27903876e-02 -4.40700442e-01 -4.18296993e-01 2.29381368e-01
-6.98433340e-01 -2.32099578e-01 -1.62966296e-01 7.07944274e-01
-2.97873706e-01 5.18514514e-01 -7.70628273e-01 -2.80280113e-01
4.07253295e-01 3.55573744e-01 -2.09967181e-01 2.91330189e-01
2.00210422e-01 4.19915736e-01 -5.28278649e-01 7.86270499e-01
1.35618103e+00 -4.07988846e-01 -2.26458877e-01 -6.40351236e-01
-8.55454266e-01 -6.23367846e-01 -1.17670584e+00 1.53481591e+00
-4.72306550e-01 5.23687005e-01 3.81323904e-01 -5.38171113e-01
8.16594064e-01 -8.05148110e-03 5.38694739e-01 -6.54406488e-01
-5.21183573e-02 3.22881132e-01 -1.08784348e-01 -1.29479074e+00
5.36967516e-01 -4.53451425e-01 5.63082635e-01 6.16785347e-01
-4.09595877e-01 -4.03132498e-01 -3.62140164e-02 -2.76656270e-01
6.33807719e-01 -5.88376261e-02 3.86141121e-01 -3.90802860e-01
8.09513807e-01 7.84454346e-02 2.65869945e-01 4.13213700e-01
-1.34847239e-01 5.07708728e-01 -2.98881561e-01 -2.67184913e-01
-1.15212786e+00 -8.43134880e-01 -3.37257564e-01 3.55643392e-01
7.98194528e-01 3.25040966e-01 -4.72446620e-01 1.54911622e-01
-1.15804024e-01 3.12335134e-01 1.25507250e-01 5.67223132e-02
-2.65764356e-01 -1.32014430e+00 1.77009851e-01 -2.77860433e-01
1.36502635e+00 -4.42727774e-01 -8.20127547e-01 -1.18681751e-02
-2.10867152e-01 -1.06497216e+00 -2.55538672e-01 -5.71433246e-01
-7.25278258e-01 -1.29943037e+00 -1.12733817e+00 -5.53740442e-01
4.82027352e-01 1.45679688e+00 4.65584755e-01 -1.30468309e-01
-7.01783001e-01 3.65549415e-01 -1.92179620e-01 -2.64405102e-01
-6.76722229e-02 -5.68834841e-01 1.84992738e-02 4.94991750e-01
7.44760223e-03 -6.95056140e-01 -1.03844583e+00 2.84007847e-01
-1.04457080e+00 4.42082882e-01 9.90498006e-01 7.78487623e-01
2.23573223e-01 5.13944209e-01 2.41215587e-01 -4.29097921e-01
6.28998458e-01 -2.62314707e-01 -9.71845388e-01 3.77345532e-01
-5.49796879e-01 -3.46748531e-01 7.07023263e-01 -3.25826526e-01
-1.61211646e+00 -2.78124899e-01 2.94763088e-01 -2.09425837e-01
-2.02503875e-01 4.57809061e-01 -1.28103243e-02 -5.69925249e-01
6.00296021e-01 7.81335771e-01 2.58093506e-01 -5.74088275e-01
2.72680491e-01 1.14098716e+00 6.44616604e-01 -2.42027134e-01
8.31102490e-01 9.41202581e-01 4.61396396e-01 -1.50136733e+00
-7.94484973e-01 -7.31201530e-01 -2.31155589e-01 -5.98667800e-01
5.89754462e-01 -9.29726660e-01 -9.25104141e-01 9.33449566e-01
-1.20308387e+00 9.10520703e-02 1.76461905e-01 8.63723278e-01
-1.99737132e-01 6.57683551e-01 -5.09615719e-01 -9.86540675e-01
-2.30563149e-01 -1.05741501e+00 9.62852538e-01 6.69176638e-01
8.04759383e-01 -8.51513922e-01 9.59519893e-02 3.53212029e-01
6.65601671e-01 -6.64697960e-03 6.76746130e-01 6.67484522e-01
-1.14153588e+00 1.18578419e-01 -7.58774817e-01 2.83647388e-01
5.20560682e-01 3.39253731e-02 -1.04334605e+00 -1.49568781e-01
8.64032626e-01 -1.37944855e-02 6.91390991e-01 7.44059980e-01
1.07191634e+00 -1.52015328e-01 -2.06417590e-01 1.11025882e+00
2.21473503e+00 3.23802441e-01 6.19492710e-01 3.39001000e-01
5.83599865e-01 6.10871434e-01 5.91160357e-01 4.33570415e-01
-1.76018681e-02 5.07831037e-01 5.04980206e-01 -3.06718946e-01
-1.86199725e-01 3.78738165e-01 1.87160838e-02 7.67644107e-01
-4.48286742e-01 -1.91746086e-01 -4.43583935e-01 3.39249402e-01
-1.46936214e+00 -1.30855274e+00 -7.87198603e-01 2.06351566e+00
7.31296003e-01 -7.37435460e-01 -2.24442422e-01 -8.56717080e-02
9.28198695e-01 3.32976013e-01 -5.96209347e-01 3.91555607e-01
-3.90676290e-01 -9.65961441e-02 6.87023878e-01 6.61671102e-01
-7.48542011e-01 3.83289158e-01 6.16316795e+00 6.00020707e-01
-1.49631643e+00 6.75115064e-02 1.33608758e-01 -8.69098231e-02
-3.97132337e-01 9.42523852e-02 -5.00450313e-01 7.52429724e-01
4.29053545e-01 -4.79304940e-01 6.38507962e-01 2.29964301e-01
8.24630976e-01 -7.09169447e-01 -1.76978767e-01 1.32239044e+00
1.37610987e-01 -1.14987648e+00 1.47746384e-01 1.59763128e-01
8.63866925e-01 -1.66958362e-01 2.56473184e-01 -8.49356890e-01
-1.63072228e-01 -4.36609745e-01 2.10305214e-01 1.07094717e+00
1.00106883e+00 -4.17879999e-01 3.66813630e-01 7.13746011e-01
-8.50304246e-01 -3.02417397e-01 -6.23642504e-01 -2.58774102e-01
3.85706156e-01 1.15181708e+00 -2.37246513e-01 9.52094078e-01
4.20930177e-01 8.05826724e-01 -3.88356537e-01 1.47052038e+00
3.46930213e-02 3.73574883e-01 -2.24768817e-02 -6.06221333e-03
1.42780840e-01 -1.15933752e+00 7.35707462e-01 7.08559990e-01
8.28983903e-01 6.45984411e-01 -1.95019051e-01 1.02012229e+00
4.08154547e-01 -2.36049488e-01 -7.12224424e-01 1.18688472e-01
3.86003703e-01 1.54388928e+00 -2.33496144e-01 -5.01275696e-02
-6.29059613e-01 1.01637053e+00 -4.17208314e-01 6.41339600e-01
-7.65457511e-01 -4.48775232e-01 5.24848163e-01 2.94657741e-02
-2.30462149e-01 -4.17009205e-01 -1.71177194e-01 -1.29635715e+00
7.91541487e-02 -4.99883115e-01 -9.70059559e-02 -1.51677489e+00
-1.07487226e+00 3.18760425e-01 -6.93701394e-03 -1.52585042e+00
4.66619223e-01 -6.03641033e-01 -5.92491567e-01 1.33481848e+00
-1.99545670e+00 -1.22051764e+00 -7.77513802e-01 8.20657015e-01
3.73664975e-01 6.90295473e-02 4.59048450e-01 4.16749448e-01
-4.26210284e-01 -3.82504582e-01 7.66874254e-01 -4.82211977e-01
7.07454681e-01 -7.18046725e-01 -4.84678000e-01 1.22808731e+00
-4.18202072e-01 6.88906133e-01 6.87936425e-01 -3.75820190e-01
-1.61609375e+00 -8.30042720e-01 2.85690516e-01 2.76528209e-01
5.90549529e-01 3.04348528e-01 -6.44312918e-01 3.21585615e-03
4.40162659e-01 3.07226870e-02 5.05095541e-01 -6.82888210e-01
1.59551933e-01 -5.33734679e-01 -1.06439579e+00 2.61396408e-01
9.83578384e-01 -6.17827237e-01 -4.69071418e-01 5.25415778e-01
4.23994213e-01 -2.73523778e-01 -4.89816278e-01 3.91917288e-01
7.32680917e-01 -1.42579615e+00 1.13625813e+00 1.07641362e-01
4.49631304e-01 -6.74311399e-01 -6.53092116e-02 -1.45768046e+00
-4.58139211e-01 -6.60423458e-01 3.53532255e-01 1.10516179e+00
-1.16275862e-01 -8.86639476e-01 4.73136395e-01 1.85002565e-01
-2.46214792e-01 -3.17927986e-01 -4.38472718e-01 -7.58727133e-01
-5.52071273e-01 2.89704323e-01 5.72843194e-01 8.62121820e-01
-4.33348089e-01 1.57448277e-01 -7.43972301e-01 7.12132692e-01
1.28057134e+00 5.94329059e-01 4.49835867e-01 -1.14622188e+00
-2.82423496e-01 -2.00371251e-01 1.04634194e-02 -1.14938736e+00
-2.51141310e-01 -5.44373333e-01 -1.67559326e-01 -1.55131149e+00
5.47591388e-01 -2.33050033e-01 1.03716113e-01 -4.59179223e-01
-1.60893753e-01 1.93372622e-01 -2.42155623e-02 7.89691985e-01
-1.09306261e-01 3.15355182e-01 1.70649934e+00 -5.44737615e-02
-9.78942439e-02 9.61852446e-02 -6.37446165e-01 6.26086593e-01
5.20417631e-01 7.07821846e-02 -6.78652167e-01 -8.16896021e-01
3.75978500e-02 2.78671145e-01 6.52705610e-01 -1.19435024e+00
5.65527916e-01 -4.35995668e-01 2.92435139e-01 -3.53877962e-01
5.23526788e-01 -9.44510400e-01 5.65463841e-01 4.41448808e-01
4.60222252e-02 -4.39399153e-01 -2.62675703e-01 5.05923986e-01
-1.83608472e-01 -4.77958232e-01 1.06599748e+00 -4.61017102e-01
-1.00737691e+00 3.01017582e-01 -4.79064956e-02 -3.29619229e-01
1.16040552e+00 -2.78052956e-01 -1.05246639e+00 -1.59895256e-01
-1.34238077e-03 -1.30248815e-01 6.41419649e-01 -3.74712616e-01
7.12206960e-01 -1.18767452e+00 -4.64801610e-01 3.09866339e-01
-6.95475936e-02 -4.17452693e-01 8.44660461e-01 9.77585375e-01
-9.52524185e-01 3.10488939e-01 -3.61312896e-01 -5.10784864e-01
-1.41021764e+00 3.57620835e-01 6.33820355e-01 3.72610211e-01
-4.25704449e-01 6.15061820e-01 4.87073734e-02 1.61793157e-01
-5.69308996e-01 -1.27654672e-01 -1.69723257e-01 -2.95189232e-01
8.05011213e-01 8.23434830e-01 -1.34425059e-01 -6.05221272e-01
2.32860446e-02 1.23910594e+00 3.86409312e-01 -1.11273579e-01
1.35959899e+00 -7.59219170e-01 -5.34436524e-01 3.73137683e-01
1.18710983e+00 3.06535721e-01 -1.18760455e+00 -2.27542624e-01
-5.68518162e-01 -1.16100037e+00 3.20941180e-01 -7.41165996e-01
-7.24206448e-01 8.98481965e-01 5.16045809e-01 6.34279609e-01
1.44519198e+00 -2.92393804e-01 7.08161175e-01 8.85738134e-02
6.23142302e-01 -8.16252589e-01 -8.61582384e-02 8.90173391e-02
6.17645025e-01 -1.25371993e+00 2.22971618e-01 -5.86314857e-01
-2.99351931e-01 1.26315928e+00 2.03452229e-01 2.05763593e-01
5.42250335e-01 8.85432586e-02 7.81282708e-02 -2.65366495e-01
-2.65395582e-01 -4.43030179e-01 -1.56252570e-02 6.20415449e-01
1.57883525e-01 -1.52020454e-01 -4.16184902e-01 -8.25978816e-02
2.65139818e-01 3.15411568e-01 1.02344596e+00 5.37473023e-01
-8.08107793e-01 -6.90710425e-01 -5.96208215e-01 4.29903090e-01
-3.97262484e-01 -9.71983820e-02 2.33765319e-02 4.91092294e-01
5.38068935e-02 1.14161003e+00 -1.95735037e-01 -5.26035093e-02
2.33420148e-01 -2.69498020e-01 4.88892198e-01 -3.49287957e-01
2.83403903e-01 1.88795060e-01 -2.45013520e-01 -2.58414775e-01
-1.05627429e+00 -5.78397334e-01 -7.31298864e-01 -2.41541386e-01
-6.21373534e-01 2.48746976e-01 7.30912268e-01 7.16937780e-01
2.91267842e-01 1.64682224e-01 1.06542158e+00 -6.62661850e-01
-5.48326433e-01 -8.77418578e-01 -1.31308722e+00 1.24099389e-01
6.31689429e-01 -4.53172147e-01 -7.43710041e-01 3.11903600e-02] | [10.857457160949707, -2.6165482997894287] |
319d9fed-a099-45c6-8619-61e2838c2ba6 | steering-distortions-to-preserve-classes-and | null | null | http://proceedings.neurips.cc/paper/2020/hash/99607461cdb9c26e2bd5f31b12dcf27a-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/99607461cdb9c26e2bd5f31b12dcf27a-Paper.pdf | Steering Distortions to Preserve Classes and Neighbors in Supervised Dimensionality Reduction | Nonlinear dimensionality reduction of high-dimensional data is challenging as the low-dimensional embedding will necessarily contain distortions, and it can be hard to determine which distortions are the most important to avoid. When annotation of data into known relevant classes is available, it can be used to guide the embedding to avoid distortions that worsen class separation. The supervised mapping method introduced in the present paper, called ClassNeRV, proposes an original stress function that takes class annotation into account and evaluates embedding quality both in terms of false neighbors and missed neighbors. ClassNeRV shares the theoretical framework of a family of methods descended from Stochastic Neighbor Embedding (SNE). Our approach has a key advantage over previous ones: in the literature supervised methods often emphasize class separation at the price of distorting the data neighbors' structure; conversely, unsupervised methods provide better preservation of structure at the price of often mixing classes. Experiments show that ClassNeRV can preserve both neighbor structure and class separation, outperforming nine state of the art alternatives. | ['Sylvain Lespinats', 'Denys Dutykh', 'Michael Aupetit', 'Jaakko Peltonen', 'Benoît Colange'] | 2020-12-01 | null | null | null | neurips-2020-12 | ['supervised-dimensionality-reduction'] | ['computer-vision'] | [ 1.25655636e-01 3.39074343e-01 -2.65504599e-01 -2.93370545e-01
-4.31622356e-01 -8.03744614e-01 8.01307142e-01 5.25028586e-01
-3.86906475e-01 6.31933749e-01 4.07546610e-01 1.01010136e-01
-5.92319429e-01 -9.86244678e-01 -2.29040846e-01 -9.18240786e-01
-2.15070602e-02 6.15041673e-01 4.36939508e-01 -1.45585716e-01
3.70305419e-01 7.11049139e-01 -1.81047738e+00 1.24328464e-01
6.93805814e-01 6.48327649e-01 -9.29934606e-02 3.45695198e-01
-3.00264031e-01 3.98901016e-01 -4.47251946e-01 -3.94188672e-01
3.49356085e-01 -5.29213548e-01 -8.04637671e-01 -1.25474051e-01
4.20579046e-01 8.59238356e-02 -3.34710777e-01 1.34137738e+00
5.45904517e-01 1.31813034e-01 9.82165039e-01 -1.20071507e+00
-6.86704874e-01 6.05176806e-01 -1.96098879e-01 6.51301891e-02
8.06817412e-02 -4.08152282e-01 1.09595287e+00 -1.09586883e+00
7.35148370e-01 1.12669384e+00 7.10232735e-01 5.98096251e-01
-1.75572395e+00 -2.05706343e-01 -5.31611685e-03 3.83147955e-01
-1.45710623e+00 -4.06263590e-01 1.16025031e+00 -5.48382461e-01
6.00145936e-01 5.76276362e-01 3.93782020e-01 8.45397711e-01
-1.09216876e-01 4.33324695e-01 1.01823163e+00 -6.39267445e-01
5.70203125e-01 4.81640548e-01 4.35982019e-01 3.55172157e-01
4.71450716e-01 2.37743303e-01 -2.82912761e-01 -2.60220170e-01
2.68769205e-01 -5.42654190e-03 -3.16522628e-01 -1.15481794e+00
-1.08049762e+00 9.25171375e-01 3.90882790e-01 5.30185103e-01
3.18783708e-02 -2.63176382e-01 3.68432522e-01 3.48682523e-01
5.50308526e-01 5.38445115e-01 -6.28417730e-02 -8.97379965e-02
-9.12774563e-01 2.31873572e-01 7.09308863e-01 6.38037980e-01
9.38640654e-01 -2.80422300e-01 3.98902594e-05 8.14393580e-01
8.16782191e-02 4.15320918e-02 7.11168945e-01 -8.07358682e-01
2.31710806e-01 7.76024997e-01 -7.38497376e-02 -1.38979983e+00
-2.19691589e-01 -3.23217779e-01 -9.84120905e-01 5.24963081e-01
2.52622664e-01 4.87923801e-01 -5.10197103e-01 1.65093017e+00
3.34165663e-01 -2.86017603e-04 2.37877578e-01 6.51587307e-01
5.44451773e-01 4.58122998e-01 -4.30131376e-01 -3.45942765e-01
8.85687411e-01 -6.46455646e-01 -8.83386612e-01 8.50547925e-02
7.25411832e-01 -6.76643133e-01 9.33925509e-01 3.01208884e-01
-8.66459548e-01 -4.91697639e-01 -1.21588290e+00 -7.79255033e-02
-8.53274643e-01 -1.03748612e-01 1.71537340e-01 9.12424386e-01
-1.03640330e+00 9.30998087e-01 -7.82182276e-01 -3.01735491e-01
2.47106835e-01 4.55795765e-01 -6.17195427e-01 3.39405313e-02
-1.11291349e+00 8.99123967e-01 4.96620029e-01 -5.97858429e-02
-3.33230317e-01 -4.93784904e-01 -8.45701933e-01 1.57808922e-02
2.55772024e-01 -1.30074665e-01 4.67028409e-01 -5.88404059e-01
-1.20014572e+00 7.73226440e-01 -5.87823801e-02 -4.91292596e-01
5.71947396e-01 2.90420316e-02 -3.87258768e-01 6.60883635e-02
-4.96045090e-02 4.81249809e-01 9.20468450e-01 -1.25245965e+00
-3.34137261e-01 -5.20096064e-01 -1.12740092e-01 8.16111341e-02
-1.09740508e+00 -5.41141033e-01 -9.75969359e-02 -8.06247473e-01
3.13797235e-01 -8.89138401e-01 2.70997845e-02 2.14687362e-01
-3.71870816e-01 -2.67164141e-01 1.16558290e+00 -1.35346502e-01
1.43275952e+00 -2.30415034e+00 6.09366059e-01 3.91286790e-01
5.41627586e-01 4.14878845e-01 -7.70125687e-02 5.72813511e-01
-4.36832994e-01 1.93532929e-01 -5.77599823e-01 -4.59365398e-01
3.01865518e-01 4.78469670e-01 -3.58789802e-01 7.25360811e-01
2.02273671e-02 5.38514793e-01 -8.62286389e-01 -4.30510610e-01
3.83253872e-01 7.65695453e-01 -6.69756889e-01 -8.90834257e-02
3.09716851e-01 -1.52690485e-01 5.92754297e-02 1.51333049e-01
6.94254518e-01 8.78653601e-02 1.36020511e-01 -3.48294765e-01
-3.67946029e-02 1.72242284e-01 -1.65780807e+00 1.33841491e+00
-4.04448174e-02 7.15935528e-01 -2.97675401e-01 -1.45383537e+00
1.06639671e+00 8.69615600e-02 4.47330236e-01 -2.63992310e-01
-1.76310435e-01 2.04515740e-01 -1.07128650e-01 -2.35273868e-01
6.73909128e-01 1.08055986e-01 4.20240648e-02 4.00410503e-01
-3.97164449e-02 6.38013035e-02 6.46988153e-02 1.14370465e-01
1.00943124e+00 -2.22793981e-01 3.76031965e-01 -5.79289377e-01
5.90978026e-01 -1.41120329e-01 6.17435753e-01 4.54320282e-01
-2.32363477e-01 7.62716591e-01 7.36645699e-01 -4.28799510e-01
-1.19808996e+00 -1.21069336e+00 -5.18103838e-01 5.80590308e-01
2.99375504e-01 -4.83099520e-01 -7.12180197e-01 -9.96220648e-01
1.11642197e-01 5.56828082e-01 -9.53338027e-01 -6.03469074e-01
-3.53641927e-01 -6.67491972e-01 3.59307170e-01 2.32809395e-01
1.29340112e-01 -6.52745545e-01 -2.08679348e-01 -2.81123351e-03
8.84584934e-02 -6.13382936e-01 -4.34117168e-01 3.21326166e-01
-1.02315378e+00 -1.04854238e+00 -6.81311131e-01 -7.64519393e-01
8.26026559e-01 1.61465690e-01 9.05380189e-01 -1.81979194e-01
-2.97413498e-01 1.58250093e-01 -4.86827314e-01 -7.15608522e-02
-5.23211360e-01 2.16926068e-01 3.41496497e-01 2.00805619e-01
7.42705226e-01 -7.21384943e-01 -2.96234310e-01 3.28658223e-01
-1.21546912e+00 -5.36385715e-01 4.02487606e-01 9.64028776e-01
7.38433361e-01 5.12132764e-01 4.34672475e-01 -1.07205141e+00
4.90694284e-01 -3.22762698e-01 -5.43890238e-01 1.41543433e-01
-1.19131029e+00 3.70377064e-01 7.28819907e-01 -3.39787185e-01
-4.33625251e-01 3.87641788e-02 2.25846022e-01 -3.68084610e-01
-1.17331564e-01 6.38728365e-02 -4.34770793e-01 -2.04333574e-01
8.77881706e-01 3.25439721e-01 1.12778597e-01 -7.17827082e-01
4.66506541e-01 7.00597167e-01 3.55431885e-01 -1.89631015e-01
9.52086926e-01 5.78352988e-01 7.81583115e-02 -9.51471984e-01
-3.57468545e-01 -3.78650576e-01 -1.12575281e+00 1.40191898e-01
5.97770333e-01 -2.89940894e-01 -1.57211542e-01 5.32816537e-02
-9.13038194e-01 2.94248760e-01 -8.33000422e-01 3.45341146e-01
-4.59009796e-01 6.48380637e-01 -1.21624596e-01 -5.49728096e-01
3.15243930e-01 -1.11132061e+00 6.68946624e-01 -2.60129035e-01
-2.85673112e-01 -1.22706962e+00 3.92030835e-01 -8.04230347e-02
2.04095259e-01 1.90898880e-01 1.20652139e+00 -9.47411418e-01
-1.44000664e-01 -3.94939601e-01 4.79043610e-02 6.29124343e-01
4.46658432e-01 -1.72527730e-01 -9.96870279e-01 -4.68412042e-01
-1.66722313e-02 1.92433983e-01 9.58970904e-01 1.79791257e-01
9.72054660e-01 -3.85950208e-01 -3.25879693e-01 5.05383134e-01
1.41003871e+00 -2.45494489e-02 7.32045114e-01 2.98230648e-01
6.64753377e-01 9.26323295e-01 3.73071194e-01 1.28531158e-01
1.07196204e-01 9.79485631e-01 4.04427946e-01 1.05958176e-03
-2.69631803e-01 -1.08424015e-01 2.44387642e-01 8.66692722e-01
3.61063451e-01 -7.79370144e-02 -7.90110528e-01 7.07936823e-01
-1.78650749e+00 -1.00402558e+00 -3.80416028e-02 2.48613238e+00
7.26409853e-01 1.45635679e-01 2.85355270e-01 9.51067984e-01
8.17862511e-01 2.21367791e-01 -3.07701290e-01 -3.41547817e-01
-3.05364430e-01 -1.60672739e-01 3.16781670e-01 7.97303438e-01
-1.09850585e+00 5.75724363e-01 6.12848330e+00 7.96544313e-01
-6.44023478e-01 -8.22284594e-02 2.84496725e-01 -5.62336072e-02
-4.65263456e-01 7.73161352e-02 -8.34502935e-01 5.52541375e-01
5.70865452e-01 -2.29685828e-01 2.49697089e-01 8.42726529e-01
-1.00582525e-01 2.79334962e-01 -1.41330051e+00 1.00386620e+00
2.30488062e-01 -1.22532201e+00 2.23115161e-01 3.32550287e-01
6.59190536e-01 -4.53307062e-01 2.21200645e-01 -1.36735782e-01
-1.37605714e-02 -7.85362303e-01 3.24791521e-01 4.43401217e-01
6.14398241e-01 -1.19795382e+00 7.67297268e-01 1.68296129e-01
-9.66908693e-01 -4.66987975e-02 -6.54150486e-01 1.25790104e-01
-2.19920203e-01 9.54892278e-01 -4.92541403e-01 3.46025914e-01
4.33994055e-01 9.23461914e-01 -6.70947075e-01 8.16413045e-01
6.97924569e-02 3.58817488e-01 -1.75878093e-01 3.40567529e-02
6.94028661e-02 -4.13768888e-01 9.63394523e-01 1.01606965e+00
1.40471995e-01 -5.46960868e-02 -3.77941946e-03 7.30615735e-01
1.27948523e-01 2.23145366e-01 -9.40260410e-01 -5.38421087e-02
6.72101796e-01 9.19228256e-01 -7.81931758e-01 -2.54028201e-01
-1.12891510e-01 1.22821534e+00 2.22516432e-01 3.88255343e-02
-2.61119187e-01 -8.41462851e-01 8.28619301e-01 2.02082008e-01
4.27322984e-01 -2.08729506e-01 -3.20102721e-01 -1.09289718e+00
2.51388460e-01 -6.26116395e-01 3.78324211e-01 -7.90812895e-02
-1.28847027e+00 7.48190403e-01 -1.64153632e-02 -1.60728514e+00
-2.34258220e-01 -6.42117321e-01 -1.25151291e-01 5.51157176e-01
-1.33472347e+00 -7.06318259e-01 1.37386536e-02 3.90933573e-01
3.38526309e-01 -2.68233508e-01 1.03058350e+00 4.79226530e-01
-4.44792479e-01 8.04787338e-01 5.84112287e-01 1.07035920e-01
7.04927504e-01 -1.51303935e+00 4.27946635e-02 7.76387155e-01
4.06330794e-01 3.38865787e-01 7.51124322e-01 -3.56038451e-01
-1.03341317e+00 -1.07745481e+00 1.28952968e+00 -5.94939888e-01
5.45027316e-01 -4.99398708e-01 -1.14843929e+00 2.73225784e-01
-2.66646773e-01 2.35244408e-01 8.66479993e-01 -8.00152216e-03
-5.38836658e-01 -3.36262614e-01 -1.26109898e+00 4.56466377e-01
9.60659623e-01 -7.15558648e-01 -7.94298053e-01 2.48963609e-01
5.66500485e-01 2.05477118e-01 -8.30991864e-01 1.98533311e-01
4.25693095e-01 -1.18383586e+00 1.13596869e+00 -4.96603251e-01
2.13414937e-01 -4.22844946e-01 -4.02766526e-01 -1.31818080e+00
-5.03710151e-01 -3.74144375e-01 -1.91901103e-01 1.48286343e+00
3.14484864e-01 -7.59887993e-01 9.47043538e-01 3.87722880e-01
1.48599744e-01 -4.39991891e-01 -1.12250471e+00 -1.14909399e+00
2.30098680e-01 -1.98052883e-01 6.75788820e-01 1.30885375e+00
3.22815403e-02 1.52604088e-01 -3.63327742e-01 1.74649835e-01
9.75038290e-01 1.22467126e-03 4.61642057e-01 -1.56308329e+00
-5.36887348e-02 -6.59711778e-01 -1.16801333e+00 -5.74545622e-01
2.51869768e-01 -1.11561108e+00 -1.49395719e-01 -1.18040454e+00
2.90052244e-03 -4.27030891e-01 -4.49610859e-01 3.74206245e-01
7.94670954e-02 5.92636466e-01 -4.54966277e-02 4.44069594e-01
-3.31239998e-01 7.67013848e-01 5.87008297e-01 -1.99776143e-01
-4.24906939e-01 -1.48683846e-01 -5.35690129e-01 6.73744857e-01
3.72402728e-01 -7.50626385e-01 -6.25361741e-01 -3.59447211e-01
2.11430073e-01 -6.42529011e-01 2.49661282e-01 -1.11529469e+00
3.21017116e-01 1.36234298e-01 2.10307598e-01 -4.39295560e-01
3.22971791e-01 -1.15188897e+00 1.00068443e-01 4.90502536e-01
-7.03588784e-01 -3.12555581e-02 -2.13131562e-01 8.34492087e-01
-3.71402413e-01 -3.80384237e-01 1.06225348e+00 2.17005745e-01
-5.01966774e-01 1.71411753e-01 -2.27874354e-01 -6.72731623e-02
1.00079238e+00 -3.91321629e-01 -4.29461934e-02 -1.76514924e-01
-8.61755848e-01 -1.58066317e-01 7.54488528e-01 4.90085185e-01
6.81279957e-01 -1.79595041e+00 -6.07154846e-01 4.54157263e-01
3.82605523e-01 -2.53907144e-01 -2.12275628e-02 5.41267872e-01
-2.63944447e-01 2.37690657e-01 3.05746235e-02 -6.30175829e-01
-1.32869363e+00 7.63087034e-01 1.79871589e-01 -3.73242013e-02
-8.02745402e-01 6.57325804e-01 4.33995649e-02 -6.99335515e-01
2.52223253e-01 -2.91833538e-03 -4.20752794e-01 4.78179544e-01
6.33965135e-01 7.68818498e-01 3.49279046e-01 -8.47671866e-01
-4.28395718e-01 7.30713487e-01 -1.62320092e-01 -1.13554455e-01
1.38290918e+00 -2.12742046e-01 -3.05319875e-01 5.99994779e-01
1.57566202e+00 3.06196132e-04 -9.18748915e-01 -4.65624154e-01
2.13623106e-01 -7.62199759e-01 3.19281846e-01 -3.41984928e-01
-7.43567824e-01 9.33858454e-01 9.78664577e-01 6.04969978e-01
1.13865364e+00 -4.01920527e-02 4.57539171e-01 4.70724970e-01
4.76569831e-02 -1.18006575e+00 -5.15855923e-02 2.95054644e-01
7.01212108e-01 -1.05760598e+00 -1.04424106e-02 -3.21877092e-01
-1.61408156e-01 1.12420714e+00 1.39719054e-01 -3.73926699e-01
1.01068664e+00 8.33703279e-02 -1.11627974e-01 -9.13038030e-02
-4.05727476e-01 2.79862583e-02 4.30539727e-01 7.95270383e-01
-2.41085235e-03 -5.96060529e-02 -5.59374571e-01 3.51856798e-01
-2.08391771e-01 -5.78516960e-01 4.80056882e-01 7.19775379e-01
-3.97211879e-01 -1.53951705e+00 -3.82074744e-01 3.82956713e-01
-1.14689820e-01 2.85324478e-03 -6.88060105e-01 7.24153697e-01
2.55728543e-01 6.15749538e-01 1.80738091e-01 -4.77651536e-01
3.97387296e-01 3.36439647e-02 2.24584341e-01 -5.85305035e-01
-9.40739214e-02 -1.20671034e-01 -3.06772053e-01 -4.73802477e-01
-4.86218214e-01 -7.01884985e-01 -9.02283847e-01 -2.28746712e-01
-3.11539948e-01 5.60878396e-01 5.44265091e-01 7.70105183e-01
5.20705104e-01 1.58303097e-01 9.71063137e-01 -7.16890335e-01
-6.39362752e-01 -5.56971729e-01 -8.28049898e-01 5.40976048e-01
6.50862753e-01 -8.27360570e-01 -7.46800840e-01 7.89674893e-02] | [7.991508483886719, 4.15956449508667] |
16825b43-9e89-4f1c-a864-245be1f20723 | densehybrid-hybrid-anomaly-detection-for | 2207.02606 | null | https://arxiv.org/abs/2207.02606v1 | https://arxiv.org/pdf/2207.02606v1.pdf | DenseHybrid: Hybrid Anomaly Detection for Dense Open-set Recognition | Anomaly detection can be conceived either through generative modelling of regular training data or by discriminating with respect to negative training data. These two approaches exhibit different failure modes. Consequently, hybrid algorithms present an attractive research goal. Unfortunately, dense anomaly detection requires translational equivariance and very large input resolutions. These requirements disqualify all previous hybrid approaches to the best of our knowledge. We therefore design a novel hybrid algorithm based on reinterpreting discriminative logits as a logarithm of the unnormalized joint distribution $\hat{p}(\mathbf{x}, \mathbf{y})$. Our model builds on a shared convolutional representation from which we recover three dense predictions: i) the closed-set class posterior $P(\mathbf{y}|\mathbf{x})$, ii) the dataset posterior $P(d_{in}|\mathbf{x})$, iii) unnormalized data likelihood $\hat{p}(\mathbf{x})$. The latter two predictions are trained both on the standard training data and on a generic negative dataset. We blend these two predictions into a hybrid anomaly score which allows dense open-set recognition on large natural images. We carefully design a custom loss for the data likelihood in order to avoid backpropagation through the untractable normalizing constant $Z(\theta)$. Experiments evaluate our contributions on standard dense anomaly detection benchmarks as well as in terms of open-mIoU - a novel metric for dense open-set performance. Our submissions achieve state-of-the-art performance despite neglectable computational overhead over the standard semantic segmentation baseline. | ['Siniša Šegvić', 'Petra Bevandić', 'Matej Grcić'] | 2022-07-06 | null | null | null | null | ['scene-segmentation', 'open-set-learning'] | ['computer-vision', 'miscellaneous'] | [ 3.44065696e-01 3.16706806e-01 1.92571431e-01 -5.32876194e-01
-1.04596698e+00 -7.24877059e-01 7.05537140e-01 1.50431722e-01
-5.84626675e-01 4.93894368e-01 -3.74209940e-01 -2.91724473e-01
-8.78021866e-02 -9.55500841e-01 -9.92737710e-01 -9.68926787e-01
-1.61594942e-01 8.42737317e-01 4.73879039e-01 -1.16349466e-01
2.50616521e-01 5.75843394e-01 -1.81452143e+00 -1.63289830e-02
9.66670752e-01 1.38291943e+00 -2.31564939e-01 6.53568327e-01
-3.31483692e-01 8.10970962e-02 -5.81463218e-01 -5.81882536e-01
5.98620832e-01 -2.37612143e-01 -7.91905880e-01 -2.61040986e-01
9.98189867e-01 -2.90224046e-01 -1.36964336e-01 1.45467412e+00
2.81391472e-01 2.16626629e-01 1.08541453e+00 -1.28322005e+00
-6.15425706e-01 2.17663512e-01 -7.81523943e-01 6.34843707e-01
1.38053849e-01 2.52830446e-01 1.21642673e+00 -8.93630683e-01
5.73947191e-01 9.21773434e-01 7.43733764e-01 5.59904039e-01
-1.50008094e+00 -6.30037844e-01 4.66339827e-01 -4.43402342e-02
-1.32839429e+00 -1.50857210e-01 5.46252370e-01 -5.02646387e-01
1.05945706e+00 4.67315137e-01 4.18275476e-01 1.21656740e+00
1.43628865e-01 7.05072999e-01 1.03142655e+00 -4.02538002e-01
3.01866710e-01 -1.94201559e-01 3.48280430e-01 7.91968286e-01
3.59716117e-01 -2.25811917e-03 -2.54812062e-01 -2.82683939e-01
7.48962641e-01 3.88889126e-02 -1.13030724e-01 -4.30739552e-01
-8.64430964e-01 8.46944869e-01 3.13651502e-01 1.15067430e-01
4.66401018e-02 2.15859860e-01 2.90789485e-01 2.62630254e-01
3.82634908e-01 3.36943179e-01 -5.14142692e-01 -5.39810918e-02
-8.74618709e-01 4.43818688e-01 7.44994462e-01 8.74897599e-01
9.79967296e-01 2.74685118e-02 -2.43814886e-02 7.96339750e-01
4.77468848e-01 4.12988156e-01 4.47657883e-01 -9.13788557e-01
2.45069250e-01 5.41074336e-01 -2.57696301e-01 -6.24873638e-01
-3.12178522e-01 -3.43905896e-01 -7.93258131e-01 5.43058515e-01
8.21857512e-01 1.20829023e-01 -1.41406870e+00 1.87476790e+00
1.43368676e-01 7.16575012e-02 -1.89554200e-01 6.96385324e-01
5.52303970e-01 5.14115930e-01 3.53496149e-02 1.57642439e-01
1.01227427e+00 -5.85672319e-01 -1.86721504e-01 -3.38838786e-01
7.09571481e-01 -6.41105950e-01 1.26989913e+00 5.08439362e-01
-1.06871784e+00 -2.14487836e-01 -9.57972586e-01 -1.07780613e-01
-5.72528660e-01 -8.78150389e-02 5.41477919e-01 6.17945671e-01
-1.05173504e+00 6.51815951e-01 -1.11074579e+00 -2.88430125e-01
6.85164034e-01 6.78746939e-01 -3.69432628e-01 2.49775097e-01
-6.13591373e-01 6.39812291e-01 2.51469344e-01 -1.11484572e-01
-7.17128575e-01 -7.03865469e-01 -9.33227003e-01 -1.61150843e-01
2.52592266e-01 -5.40937543e-01 9.99494374e-01 -9.45735037e-01
-1.12572455e+00 1.45750821e+00 1.97029114e-01 -5.98174870e-01
5.91209888e-01 -3.86254072e-01 -1.88113824e-01 -1.42670274e-02
3.59259516e-01 7.46263921e-01 8.52583826e-01 -1.17203665e+00
-4.16581929e-01 -7.93874145e-01 -1.07900307e-01 7.70825893e-03
-1.48848638e-01 -1.61993369e-01 -3.40568453e-01 -9.67151165e-01
6.58595085e-01 -8.14259648e-01 -1.73089609e-01 1.41300559e-01
-6.85574174e-01 -1.59648225e-01 8.54112625e-01 -5.01522958e-01
1.09993064e+00 -2.08018756e+00 6.44816756e-02 7.80977666e-01
4.58725870e-01 3.03613096e-01 1.78866431e-01 -1.84920698e-01
-4.03714240e-01 1.78224593e-01 -7.78507471e-01 -2.64163643e-01
2.10483938e-01 5.01292408e-01 -4.92703229e-01 6.44146442e-01
3.48444819e-01 6.16754353e-01 -6.31881535e-01 -2.70325810e-01
1.79903060e-01 7.95170516e-02 -9.17017937e-01 1.01665996e-01
-5.00851691e-01 2.71833301e-01 -3.46643358e-01 7.74546266e-01
8.59275699e-01 -1.00678965e-01 -3.84926379e-01 1.36590019e-01
-1.57581959e-02 7.79425874e-02 -1.20244706e+00 1.75128555e+00
2.81778187e-01 2.47058481e-01 1.42233729e-01 -1.33069015e+00
9.97763216e-01 -1.96724296e-01 6.06491446e-01 -6.45159543e-01
1.95224971e-01 4.54758108e-01 9.28413123e-02 -2.38283034e-02
2.68500924e-01 -3.15091968e-01 -1.77303270e-01 2.83935368e-01
4.63178277e-01 -1.89642712e-01 8.96948203e-03 1.19914569e-01
1.37765753e+00 3.92755777e-01 4.96962182e-02 -5.14258742e-01
4.03119564e-01 -1.20145880e-01 4.39225465e-01 9.65422392e-01
-3.76619279e-01 9.63179648e-01 8.99341583e-01 -4.08449620e-01
-1.05857944e+00 -1.70940006e+00 -5.82212567e-01 1.11372125e+00
4.01778519e-02 -2.12141007e-01 -1.03352237e+00 -1.01467729e+00
3.57788764e-02 7.07714617e-01 -6.60483003e-01 -2.43730888e-01
-6.65994883e-01 -1.06480682e+00 8.61005247e-01 7.82763064e-01
3.48725557e-01 -9.72004354e-01 -3.08559179e-01 -1.75932094e-01
2.16099352e-01 -8.05389762e-01 -2.05786571e-01 7.02259839e-01
-7.35543132e-01 -9.43113863e-01 -5.13444543e-01 -5.89178145e-01
6.87523723e-01 -4.69937831e-01 1.20980263e+00 -4.11922634e-02
-4.86791790e-01 5.38781047e-01 -1.05623797e-01 -6.35073721e-01
-1.62538886e-01 -2.17851624e-02 2.18136355e-01 -1.67622849e-01
6.78673029e-01 -8.43317270e-01 -5.30294359e-01 3.29223394e-01
-1.05791640e+00 -5.24239421e-01 3.74973267e-01 8.75533938e-01
1.26746821e+00 -4.57651615e-01 2.73822188e-01 -7.86818564e-01
2.60446197e-03 -3.20849150e-01 -7.71288991e-01 -1.61780477e-01
-6.07571125e-01 8.10353458e-02 4.31515813e-01 -2.89213747e-01
-7.90972292e-01 -8.79954621e-02 -6.95190787e-01 -7.61098981e-01
-6.77574813e-01 -1.06498733e-01 -2.87134081e-01 -1.24662779e-02
1.02423441e+00 6.52039871e-02 -6.59491047e-02 -4.78355646e-01
3.55699986e-01 2.31886357e-01 9.94177282e-01 -9.30697083e-01
7.66655922e-01 6.65906429e-01 -9.23161488e-03 -7.17505217e-01
-8.18896294e-01 -4.06985581e-01 -7.69823313e-01 2.38599226e-01
1.09929097e+00 -6.26885593e-01 -5.71157992e-01 4.79654521e-01
-8.43413770e-01 -1.46088511e-01 -7.99586833e-01 2.64185995e-01
-8.14531505e-01 5.54225206e-01 -4.55606580e-01 -5.04026353e-01
-3.08618754e-01 -1.18321025e+00 1.30220830e+00 9.27550942e-02
-3.51420969e-01 -8.62928748e-01 2.18409494e-01 1.41147688e-01
1.34045929e-01 4.19671416e-01 9.79211867e-01 -1.24043083e+00
-5.87275743e-01 -3.43705475e-01 -3.85141402e-01 7.27009773e-01
-4.23373580e-01 2.11919025e-02 -1.20655882e+00 -3.87409814e-02
-2.36787573e-02 -1.50822997e-01 1.13568068e+00 3.48671913e-01
1.67497015e+00 -2.37695500e-01 -2.94252932e-01 9.61422324e-01
1.22030604e+00 -9.76596251e-02 8.70727837e-01 4.26768661e-01
8.29604983e-01 1.01329759e-01 3.42897326e-01 2.98472762e-01
-2.18140129e-02 4.76446718e-01 7.68631339e-01 9.15446058e-02
1.39411375e-01 1.02701731e-01 4.12683159e-01 1.48479685e-01
-1.17717497e-01 -2.83114880e-01 -1.16508222e+00 4.83034402e-01
-1.54352903e+00 -9.54738557e-01 -2.31591761e-01 2.32998276e+00
5.98766685e-01 4.73805279e-01 1.32830755e-03 1.21541403e-01
3.68705004e-01 1.41126812e-01 -5.77958524e-01 -5.41730106e-01
-2.32248113e-01 9.28413153e-01 3.89926791e-01 2.74997473e-01
-1.34005392e+00 8.18301678e-01 5.61150694e+00 9.48837936e-01
-9.57890451e-01 8.66865553e-03 7.44623482e-01 -2.43113458e-01
-3.97416890e-01 -6.98873103e-02 -1.03339255e+00 5.72881043e-01
7.11911559e-01 2.90952384e-01 -3.23426947e-02 1.07088602e+00
-6.13778651e-01 -1.14726231e-01 -1.27824533e+00 8.98106933e-01
1.79710835e-01 -1.01887155e+00 2.51921058e-01 3.13601226e-01
4.49714899e-01 3.74815047e-01 2.56683290e-01 3.92164558e-01
4.07447577e-01 -1.14473152e+00 7.47105658e-01 5.06868601e-01
8.42332184e-01 -4.32513446e-01 5.62440574e-01 2.32655585e-01
-8.95187020e-01 1.59275651e-01 -1.72968298e-01 1.10113226e-01
-1.26404211e-01 5.80750763e-01 -3.76104534e-01 3.46301466e-01
1.02699637e+00 4.00452554e-01 -4.25822616e-01 7.87907183e-01
4.45529744e-02 5.84552228e-01 -8.95670414e-01 6.35428727e-01
3.45450282e-01 -4.01478320e-01 9.20491695e-01 1.11485803e+00
2.89047092e-01 3.02004330e-02 2.35736057e-01 1.26303697e+00
-6.79928511e-02 6.31989390e-02 -6.08030438e-01 2.99413893e-02
-1.94623396e-01 1.05151629e+00 -9.07651663e-01 1.37603153e-02
-3.55063140e-01 9.94065881e-01 4.56031650e-01 2.63733804e-01
-8.75324070e-01 -2.69266397e-01 1.03644574e+00 3.05697918e-01
3.31995815e-01 -1.28456503e-02 -6.20796740e-01 -1.17180204e+00
2.83953190e-01 -5.14976442e-01 6.65888786e-01 -4.41315681e-01
-1.61079109e+00 3.99789065e-01 1.25986278e-01 -1.10187292e+00
-2.48881102e-01 -1.15894353e+00 -5.59249222e-01 1.00068724e+00
-1.15090871e+00 -9.59169090e-01 -1.73389435e-01 5.80828846e-01
1.99797019e-01 -2.21896082e-01 1.13406575e+00 2.91170806e-01
-6.73588812e-01 8.92099619e-01 3.29456814e-02 3.53110015e-01
6.42828941e-01 -1.67232609e+00 4.24602896e-01 1.02010155e+00
2.28496984e-01 4.55883533e-01 5.42720973e-01 -5.85327446e-01
-7.68720448e-01 -9.41188931e-01 3.56343180e-01 -9.03428853e-01
6.04999423e-01 -3.67340505e-01 -1.22991920e+00 1.05721176e+00
-3.46592873e-01 5.54324746e-01 7.59665072e-01 7.89753944e-02
-5.44036746e-01 7.78359622e-02 -1.38111603e+00 4.31826621e-01
1.13144410e+00 -3.54778588e-01 -6.68499887e-01 1.37455910e-01
4.14437741e-01 -7.07693219e-01 -8.18022072e-01 6.10013247e-01
3.81000191e-01 -1.32935929e+00 1.07433999e+00 -5.81930101e-01
3.57067049e-01 -1.75588995e-01 -4.64917511e-01 -7.48993099e-01
2.41558701e-02 -4.01064992e-01 -2.69213110e-01 1.04674089e+00
4.24054384e-01 -7.50898063e-01 9.64027166e-01 6.65361941e-01
-6.13633513e-01 -9.58370924e-01 -1.28840685e+00 -5.93505859e-01
4.57715452e-01 -7.61795342e-01 2.62598455e-01 8.30325723e-01
-5.09246528e-01 -7.90185556e-02 3.41728151e-01 2.86606997e-01
5.08234084e-01 -1.61144733e-01 5.43222308e-01 -1.47650707e+00
-3.60070765e-01 -7.26545870e-01 -9.39887524e-01 -9.04953480e-01
7.80845731e-02 -1.12468588e+00 -1.35069564e-01 -1.15023744e+00
2.79391706e-02 -6.56288624e-01 -3.09846371e-01 6.26163840e-01
1.62704885e-01 4.95602578e-01 -4.00299966e-01 1.43151790e-01
-5.65333128e-01 4.11098868e-01 7.95681477e-01 1.06498219e-01
-3.59159335e-02 -1.36676654e-01 -5.24282873e-01 1.38695657e+00
5.28199315e-01 -3.72173935e-01 -1.20760113e-01 -4.52930421e-01
1.43569082e-01 -5.14831126e-01 7.41413653e-01 -1.06957626e+00
-1.82846218e-01 1.38505116e-01 5.53731322e-01 -5.43483138e-01
3.71888191e-01 -6.67167783e-01 -4.06212181e-01 -7.26837367e-02
5.70948655e-03 1.69394746e-01 3.25520486e-01 6.60180748e-01
-7.05421865e-02 -2.12038159e-01 1.03763235e+00 -1.29347086e-01
-5.55685937e-01 4.68243331e-01 6.05409220e-02 3.64956528e-01
1.12185788e+00 -5.54484606e-01 -2.70142645e-01 7.18218982e-02
-1.04986608e+00 -1.56090813e-04 6.60847843e-01 1.62468567e-01
3.48703295e-01 -1.08729184e+00 -2.55728126e-01 7.02687621e-01
2.37327486e-01 6.05907977e-01 1.38788432e-01 9.11307514e-01
-7.03659534e-01 -7.88673609e-02 -1.85866311e-01 -9.73090649e-01
-8.09336722e-01 1.81314573e-01 5.68731427e-01 -2.85470337e-01
-6.61912560e-01 1.26028824e+00 3.68692636e-01 -6.14721298e-01
3.72250259e-01 -3.47700953e-01 2.18684316e-01 -6.81849942e-02
2.23106965e-01 3.17502886e-01 3.01032215e-01 -7.41617441e-01
-2.95685053e-01 4.30818915e-01 -3.66607666e-01 -3.80230024e-02
1.14019358e+00 1.40573636e-01 -2.18072698e-01 3.44574064e-01
1.21189260e+00 -1.07448265e-01 -1.07255113e+00 -1.48294121e-01
-1.08460531e-01 -4.02389079e-01 -2.38082305e-01 -7.24398911e-01
-8.31040263e-01 8.59212101e-01 9.11436260e-01 1.34009838e-01
9.80207086e-01 3.30043763e-01 6.44081533e-01 3.17842603e-01
7.98365921e-02 -1.24138176e+00 1.42052189e-01 6.17701232e-01
6.46353066e-01 -1.29697049e+00 -1.52279720e-01 -4.00820404e-01
-2.07105309e-01 8.62785816e-01 9.35333431e-01 -5.20673871e-01
7.97626734e-01 4.04170096e-01 7.37015232e-02 -4.05053526e-01
-3.91949266e-02 -2.21703365e-01 4.88245726e-01 5.12390077e-01
3.82393926e-01 -1.04850732e-01 -4.33916412e-02 8.59111011e-01
-4.61064726e-01 -5.93835592e-01 8.29449072e-02 9.18240547e-01
-4.35585439e-01 -1.09037673e+00 -3.97058427e-01 1.01084948e+00
-7.10561633e-01 -2.02645492e-02 -2.76982248e-01 8.07671428e-01
4.82112080e-01 1.55078366e-01 6.41445160e-01 -1.26325205e-01
2.42251903e-01 5.37651420e-01 4.78135049e-01 -7.07442701e-01
-3.76133651e-01 7.68138021e-02 -4.04514909e-01 -7.68887103e-01
-1.13333156e-02 -7.87774801e-01 -1.65282476e+00 -1.60622876e-02
-2.34951228e-01 -1.92023665e-01 3.85918617e-01 1.07214415e+00
1.48658082e-01 3.64880770e-01 -7.11440518e-02 -8.07919502e-01
-6.57999992e-01 -7.67690778e-01 -6.91559732e-01 6.32539988e-01
2.01739267e-01 -7.87335098e-01 -5.66895366e-01 -9.52957645e-02] | [7.6944899559021, 2.170311450958252] |
682cee23-4ed9-4790-90c5-f97860770c59 | finding-optimal-combination-of-kernels-using | 1604.02376 | null | http://arxiv.org/abs/1604.02376v2 | http://arxiv.org/pdf/1604.02376v2.pdf | Finding Optimal Combination of Kernels using Genetic Programming | In Computer Vision, problem of identifying or classifying the objects present
in an image is called Object Categorization. It is a challenging problem,
especially when the images have clutter background, occlusions or different
lighting conditions. Many vision features have been proposed which aid object
categorization even in such adverse conditions. Past research has shown that,
employing multiple features rather than any single features leads to better
recognition. Multiple Kernel Learning (MKL) framework has been developed for
learning an optimal combination of features for object categorization. Existing
MKL methods use linear combination of base kernels which may not be optimal for
object categorization. Real-world object categorization may need to consider
complex combination of kernels(non-linear) and not only linear combination.
Evolving non-linear functions of base kernels using Genetic Programming is
proposed in this report. Experiment results show that non-kernel generated
using genetic programming gives good accuracy as compared to linear combination
of kernels. | ['Jyothi Korra'] | 2016-04-08 | null | null | null | null | ['object-categorization'] | ['computer-vision'] | [ 1.7088161e-01 -7.7588302e-01 4.3847892e-02 -6.5394366e-01
-2.3726138e-01 -7.0498741e-01 4.4198465e-01 4.0633157e-01
-4.0655929e-01 5.4157388e-01 -3.4098497e-01 -1.3104512e-01
-6.4565367e-01 -6.8003160e-01 -2.3922956e-01 -6.8609303e-01
-2.3351288e-01 1.7246567e-01 4.9912140e-01 -1.8809801e-02
8.0673367e-01 6.4953125e-01 -2.0851557e+00 5.0399745e-01
8.7933350e-01 6.1184263e-01 3.0926195e-01 1.1605785e+00
-4.1929832e-01 3.9295122e-01 -6.1900175e-01 7.2002381e-02
5.3427070e-01 -2.2881185e-01 -6.2259269e-01 1.3014533e-01
4.4518963e-01 1.1607281e-01 3.5987949e-01 1.0514542e+00
2.3716536e-01 2.5489944e-01 1.3764677e+00 -1.7025373e+00
-1.0672001e+00 1.0263252e-01 -7.0118338e-01 4.6710655e-01
3.5385191e-02 -1.2888023e-01 3.1412649e-01 -5.7561821e-01
-1.2244733e-01 1.4600312e+00 6.7353255e-01 1.9172224e-01
-1.3076371e+00 -5.0985509e-01 -1.0291969e-01 6.7279255e-01
-1.5211420e+00 1.2106592e-01 7.5748861e-01 -5.6102115e-01
1.0134063e+00 3.5505384e-01 3.2859010e-01 7.2632305e-02
5.2294314e-01 3.5830986e-01 1.8030590e+00 -9.3023127e-01
2.7955987e-04 7.7009499e-01 9.1749239e-01 6.0873204e-01
4.5805255e-01 8.9861013e-02 -4.8473045e-02 -4.1415924e-01
3.1563523e-01 3.5533842e-02 1.3646294e-01 -2.1552153e-01
-9.9055254e-01 8.8530135e-01 3.1377211e-01 4.2723250e-01
-4.0426046e-01 4.3538152e-03 3.3710518e-01 7.3216110e-01
-5.1958725e-02 4.7062573e-01 -5.6600934e-01 2.3190458e-01
-5.3982753e-01 1.5916800e-01 5.8562469e-01 5.8501083e-01
1.0015271e+00 3.0879399e-02 -3.7238825e-02 1.1817032e+00
2.9819915e-01 3.5159788e-01 5.6136405e-01 -3.8207895e-01
-1.5939169e-01 9.8033345e-01 7.9349115e-02 -1.0398394e+00
-5.5717325e-01 2.3366541e-02 -4.7005185e-01 8.6797458e-01
3.3002412e-01 4.9075887e-02 -1.2414290e+00 1.3014375e+00
5.4172683e-01 3.9385593e-01 5.0398356e-01 6.8559432e-01
8.0371428e-01 8.3019346e-01 2.1802332e-01 -1.1429410e-01
1.3473319e+00 -7.1705383e-01 -4.3507212e-01 1.8081968e-01
3.9840305e-01 -1.3453258e+00 7.1221453e-01 5.1560104e-01
-2.9915798e-01 -8.9091593e-01 -1.0902920e+00 4.8352528e-01
-9.9042052e-01 3.5756072e-01 8.1928295e-01 1.1257492e+00
-8.9975381e-01 2.0915809e-01 -1.8313804e-01 -6.4394146e-01
1.1416447e-02 8.7042606e-01 -5.6562382e-01 -5.4973967e-02
-7.0854908e-01 1.1592213e+00 9.3584263e-01 -1.5371384e-01
-2.8954297e-01 -4.2405456e-01 -5.3752917e-01 -3.2154533e-01
1.1435013e-01 -6.4361759e-02 8.2405645e-01 -1.3663085e+00
-1.1915282e+00 6.7140955e-01 2.0312449e-02 -1.8394405e-01
1.0226355e-01 8.0881566e-02 -6.2159693e-01 7.3461331e-02
-2.2819538e-01 6.6289461e-01 1.1755447e+00 -1.1892145e+00
-7.2584397e-01 -2.3806903e-01 6.9160432e-02 2.3205483e-01
-2.8549710e-01 2.9547313e-01 2.5668290e-01 -5.2580029e-01
1.7356223e-01 -1.1678544e+00 -1.0546252e-01 -1.8686274e-01
-2.4262710e-02 -6.0835057e-01 1.3504854e+00 -5.4666841e-01
9.1145307e-01 -1.8853215e+00 -2.3766549e-01 3.1729943e-01
-4.3354517e-01 6.4010477e-01 -6.8958953e-02 2.8170642e-01
-3.2459101e-01 3.4145135e-02 -5.4104826e-03 5.8205056e-01
-3.1620744e-01 1.4309093e-01 3.2456329e-01 3.8583961e-01
4.5675600e-01 4.2068946e-01 -4.8780298e-01 -6.3730103e-01
4.5487431e-01 4.0441149e-01 -5.9207719e-02 -2.9640725e-02
2.7671957e-01 -8.2305886e-02 -3.8628009e-01 8.2045102e-01
9.4609827e-01 1.1603299e-01 -4.6499643e-01 -5.0415802e-01
-1.7535167e-02 -8.2403767e-01 -1.6614243e+00 9.7259796e-01
-3.2545090e-01 6.0824484e-01 -5.0788301e-01 -1.3705320e+00
1.1980222e+00 2.5704542e-01 2.7002311e-01 -2.8646204e-01
1.7935893e-01 8.2569972e-02 5.2470970e-01 -7.8570068e-01
7.9769544e-02 -1.4479463e-02 3.4361872e-01 4.3900701e-01
1.2988620e-01 1.4167352e-01 3.0132902e-01 -2.9972816e-01
8.3283317e-01 -9.1561496e-02 5.3242230e-01 -4.4406542e-01
8.7834460e-01 2.6507691e-01 3.1721172e-01 7.7685672e-01
-2.9869869e-01 2.8410867e-01 2.3701997e-02 -5.4824728e-01
-1.1603895e+00 -9.5554531e-01 -3.9179423e-01 9.7277075e-01
5.5999860e-02 2.6012155e-01 -4.0193650e-01 -4.7193477e-01
2.3196055e-01 3.9264691e-01 -5.4162490e-01 -4.1532278e-01
-3.0430034e-01 -1.3063548e+00 3.9790881e-01 2.9194310e-01
6.8919861e-01 -9.3774563e-01 -7.5609821e-01 2.2187722e-01
6.2263733e-01 -6.7776018e-01 -5.2245665e-02 2.2935142e-01
-8.3413023e-01 -1.1828412e+00 -5.2611554e-01 -8.7315965e-01
9.3094289e-01 6.8888420e-01 6.1112827e-01 2.5827888e-01
-1.0876111e+00 5.8824897e-01 -7.3348016e-01 -8.9716047e-01
-4.2752427e-01 -3.8046989e-01 1.8510863e-02 1.3649544e-01
6.9807631e-01 -1.8668571e-01 -3.2529435e-01 2.2041990e-01
-1.1064391e+00 -2.4470948e-01 7.7785087e-01 9.0721565e-01
4.1007482e-02 7.4398178e-01 8.7899816e-01 -5.5341589e-01
7.7787548e-01 -5.5339098e-01 -6.9600326e-01 8.7505215e-01
-6.1999798e-01 2.1936694e-01 5.0613970e-01 -8.5822558e-01
-9.7740382e-01 2.3133764e-01 5.1311767e-01 -8.4696241e-02
-5.1057625e-01 1.9885556e-01 1.1948856e-01 -6.7791885e-01
8.4917539e-01 2.7538225e-01 4.1119017e-02 -2.1636972e-01
3.6630221e-02 1.0272468e+00 4.8407726e-02 -5.8690286e-01
7.4416173e-01 9.1985255e-02 2.5700247e-01 -9.9531138e-01
-1.4846833e-01 -8.2239473e-01 -8.7313485e-01 -4.3775454e-01
8.7239730e-01 -3.6390749e-01 -6.6243869e-01 6.3734657e-01
-7.5261629e-01 2.5369823e-01 4.3918711e-01 8.1702429e-01
-3.5972655e-01 1.2332254e-01 7.8729466e-03 -1.2115502e+00
-3.0028856e-01 -1.1209716e+00 3.3920586e-01 7.9313910e-01
4.0756758e-02 -8.2050633e-01 -2.7067012e-01 2.0322582e-01
4.2093390e-01 1.4395128e-01 1.3795261e+00 -7.6593888e-01
-2.8745148e-01 -4.2598480e-01 -4.1171768e-01 5.5870426e-01
8.1160861e-01 3.6309153e-01 -7.7432698e-01 -1.6758077e-01
-3.0207819e-01 -3.7027547e-01 6.9161475e-01 4.1779134e-01
6.8298960e-01 -5.6423333e-02 -2.8267869e-01 2.3632997e-01
1.9790081e+00 9.5768648e-01 3.6365929e-01 4.9913734e-01
4.6881062e-01 6.5831488e-01 7.0586318e-01 1.2594424e-01
-2.9314769e-02 3.9484048e-01 -2.6784973e-02 4.7394842e-02
3.7599020e-02 6.4372700e-01 -1.3615906e-01 2.7999452e-01
1.0296942e-01 2.4251070e-02 -1.1188258e+00 4.2875466e-01
-1.8106024e+00 -1.0005721e+00 -3.2207537e-01 2.2166672e+00
5.3297782e-01 -1.2354292e-01 2.0386732e-01 4.8175788e-01
9.5938194e-01 -6.9656569e-01 -5.7649702e-01 -9.8040539e-01
-5.9556197e-02 2.3316975e-01 6.1853290e-01 4.9767414e-01
-1.2031140e+00 7.8021204e-01 6.6677694e+00 7.2805768e-01
-1.4085495e+00 -1.3257945e-01 4.3619135e-01 4.0190175e-01
2.5193435e-01 2.1710196e-01 -9.5785910e-01 5.2086127e-01
5.6684887e-01 -4.5356655e-01 2.8055060e-01 8.3299839e-01
-1.7167261e-02 -6.7857403e-01 -6.3107389e-01 1.0353471e+00
2.5231853e-01 -8.0830574e-01 2.0765381e-01 -1.7805423e-01
6.8920738e-01 -6.0838467e-01 -1.4538027e-02 2.4010313e-01
3.4034005e-01 -9.5838636e-01 1.4905764e-01 5.8404708e-01
-6.6432245e-03 -1.0540411e+00 7.3342365e-01 3.4525105e-01
-1.1269164e+00 -3.3256325e-01 -6.7371053e-01 4.6178639e-02
-3.9978883e-01 2.0151180e-01 -1.2016482e+00 4.1620308e-01
7.6653665e-01 2.0304334e-01 -1.1093796e+00 1.6511060e+00
6.4045841e-01 4.2748085e-01 -4.7730774e-01 -3.3977145e-01
2.4115719e-01 -2.0085983e-01 3.4846446e-01 1.4045364e+00
4.7514141e-01 2.1210308e-01 3.6345279e-01 4.0503293e-01
7.5363111e-01 6.7088240e-01 -7.4877226e-01 -5.3768799e-02
1.6361493e-01 1.3327439e+00 -1.3353796e+00 -4.5308283e-01
-5.0614721e-01 9.8145032e-01 2.5624987e-01 2.5849798e-01
-5.1877201e-01 -6.4515918e-01 4.7712460e-01 -2.3544674e-01
1.4132415e-01 -2.8605291e-01 -2.0685177e-02 -5.1857442e-01
-1.9764842e-01 -6.0989892e-01 5.5152047e-01 -6.9752401e-01
-1.4206501e+00 4.4528717e-01 6.2800866e-01 -1.1258835e+00
6.7234889e-02 -1.0102204e+00 -7.5192195e-01 1.0526710e+00
-1.1746547e+00 -1.4729494e+00 -2.7862346e-01 5.4297066e-01
7.7959377e-01 -6.7723405e-01 7.5958610e-01 6.3748807e-03
-1.7563127e-01 3.1551325e-01 2.0932715e-01 -1.5746382e-01
8.9130998e-01 -1.2802318e+00 -5.3594846e-01 6.9242138e-01
-8.0144532e-02 5.3764236e-01 7.4140030e-01 -6.3450456e-01
-1.3508517e+00 -7.5450480e-01 2.2382824e-01 -4.4055011e-02
2.1575014e-01 1.5524149e-01 -1.0359064e+00 1.5034044e-01
1.8458375e-01 2.8725043e-02 1.0740129e+00 -1.2065678e-01
-2.6852188e-01 -3.7953311e-01 -1.4748683e+00 3.3319142e-01
1.3533680e-01 -1.2621486e-01 -5.7994705e-01 2.9666540e-01
6.4911477e-02 2.2965164e-01 -5.7790816e-01 5.0967509e-01
6.6335386e-01 -9.8827130e-01 1.0511477e+00 -7.4713153e-01
-3.6414286e-01 -9.0857202e-01 -2.8872120e-01 -1.2392409e+00
-6.3345563e-01 1.3134563e-01 4.3694180e-01 1.3436637e+00
3.2397923e-01 -5.6705844e-01 5.1260424e-01 5.3825849e-01
2.0402671e-01 -4.8257023e-01 -5.3052354e-01 -1.0927117e+00
-7.3250391e-02 -1.5522705e-01 4.2571720e-01 9.7872221e-01
-4.9614796e-01 6.1893154e-02 -2.8728157e-01 4.0052146e-01
8.0951840e-01 1.3528281e-01 6.0869348e-01 -1.4782195e+00
-2.2809772e-01 -4.0437442e-01 -1.1953520e+00 3.3908364e-01
1.8491885e-01 -7.6604235e-01 -2.6913893e-01 -1.4720435e+00
3.9013198e-01 -8.7262022e-01 -5.4545391e-01 6.3148946e-01
-4.0282586e-01 3.1845638e-01 1.5771464e-01 3.0015947e-02
8.1883311e-02 -2.2359081e-01 8.0024743e-01 -2.8404847e-01
-4.1363832e-01 9.9837594e-02 -4.6456105e-01 6.3431954e-01
1.1088113e+00 -5.2425265e-01 -7.8008968e-01 -5.0299052e-02
-1.7741920e-01 -3.4933975e-01 3.1718087e-01 -1.3520050e+00
2.2382607e-01 -6.2888557e-01 7.2133964e-01 -6.3305938e-01
3.1595239e-01 -1.0493563e+00 4.8749778e-01 4.5802945e-01
-1.9718489e-01 3.4352940e-01 4.3817130e-01 4.8687091e-01
-1.2448793e-01 -9.6517289e-01 1.0733923e+00 -3.8098970e-01
-1.3610984e+00 -1.0939537e-01 -4.3200448e-01 -7.8039676e-01
1.7605751e+00 -1.0067708e+00 -4.7723036e-02 5.0926786e-03
-6.5887564e-01 -4.8296396e-02 2.6350188e-01 7.7158046e-01
5.5006057e-01 -1.3498930e+00 -7.4165380e-01 3.3344829e-03
3.2110900e-01 -8.0855727e-01 1.3559134e-01 4.5972425e-01
-7.9312503e-01 3.6103234e-01 -8.1652868e-01 -5.5374664e-01
-2.1160254e+00 7.5372332e-01 1.5366185e-01 2.6015329e-01
-9.2825264e-02 8.3505398e-01 -9.5966600e-02 -2.3491019e-01
-1.2829441e-01 -1.6480255e-01 -6.9892925e-01 8.9269720e-02
5.8491904e-01 5.1826376e-01 -4.4161115e-02 -7.6463610e-01
-3.0226442e-01 1.1066744e+00 -2.8625652e-01 5.0457273e-02
1.0335795e+00 1.6946177e-01 -2.8599331e-01 6.0438466e-01
1.3581996e+00 -5.2545524e-01 -6.3435441e-01 -2.2016005e-01
2.8521147e-01 -7.6904136e-01 -3.8055398e-02 -8.6474276e-01
-5.1826453e-01 6.1057192e-01 1.4779445e+00 2.2709006e-01
1.4014555e+00 -4.3146843e-01 -2.5329582e-02 7.5699657e-01
3.4204805e-01 -1.2490324e+00 1.5995872e-01 2.5474992e-01
7.1456689e-01 -1.8273178e+00 1.4040521e-01 -4.4181213e-01
-6.3821083e-01 1.6330513e+00 8.6622554e-01 -3.1311202e-01
1.1986942e+00 1.7344300e-01 2.4956867e-01 1.4839950e-01
-6.7368126e-01 -3.4420764e-01 4.5612988e-01 9.6501422e-01
3.7924346e-01 8.9807741e-02 -6.1443973e-01 -2.4235107e-01
1.9487758e-01 -4.8427183e-02 4.5332658e-01 1.6089015e+00
-8.3099681e-01 -1.3265314e+00 -9.9310350e-01 8.6109161e-01
-4.0080562e-01 1.5719041e-01 -2.8057718e-01 6.1354029e-01
5.8069038e-01 1.0818616e+00 -7.7597715e-02 -3.4405941e-01
8.3310053e-02 3.1105080e-01 6.6500521e-01 -5.1634985e-01
-4.9858794e-01 -1.9968575e-01 -4.8978132e-01 2.6117313e-01
-6.1074865e-01 -5.3521538e-01 -1.2107506e+00 -1.1027029e-01
-6.5067804e-01 2.3916657e-01 1.1097907e+00 8.1599307e-01
-1.6478106e-02 1.9571437e-01 5.3668708e-01 -5.7237220e-01
-3.4197894e-01 -8.2877761e-01 -6.1080503e-01 3.7495661e-01
2.6907545e-01 -9.5123553e-01 -2.2915635e-01 5.3135276e-01] | [8.214740753173828, 3.930480718612671] |
68ec89d7-e547-48ab-881f-5c5c8225cd1d | improving-cross-modal-alignment-in-vision | 2104.09580 | null | https://arxiv.org/abs/2104.09580v1 | https://arxiv.org/pdf/2104.09580v1.pdf | Improving Cross-Modal Alignment in Vision Language Navigation via Syntactic Information | Vision language navigation is the task that requires an agent to navigate through a 3D environment based on natural language instructions. One key challenge in this task is to ground instructions with the current visual information that the agent perceives. Most of the existing work employs soft attention over individual words to locate the instruction required for the next action. However, different words have different functions in a sentence (e.g., modifiers convey attributes, verbs convey actions). Syntax information like dependencies and phrase structures can aid the agent to locate important parts of the instruction. Hence, in this paper, we propose a navigation agent that utilizes syntax information derived from a dependency tree to enhance alignment between the instruction and the current visual scenes. Empirically, our agent outperforms the baseline model that does not use syntax information on the Room-to-Room dataset, especially in the unseen environment. Besides, our agent achieves the new state-of-the-art on Room-Across-Room dataset, which contains instructions in 3 languages (English, Hindi, and Telugu). We also show that our agent is better at aligning instructions with the current visual information via qualitative visualizations. Code and models: https://github.com/jialuli-luka/SyntaxVLN | ['Mohit Bansal', 'Hao Tan', 'Jialu Li'] | 2021-04-19 | null | https://aclanthology.org/2021.naacl-main.82 | https://aclanthology.org/2021.naacl-main.82.pdf | naacl-2021-4 | ['vision-language-navigation'] | ['computer-vision'] | [-9.73544121e-02 -1.18056491e-01 7.86610246e-02 -6.93320334e-01
-2.88330734e-01 -6.99247956e-01 5.18639147e-01 1.62913680e-01
-6.21702552e-01 3.36019307e-01 8.25179696e-01 -6.93262994e-01
3.68676990e-01 -4.77953762e-01 -6.30919099e-01 -6.13697350e-01
3.66149366e-01 2.72434294e-01 1.87785476e-02 -5.97766101e-01
4.48801905e-01 9.64963362e-02 -1.54100180e+00 2.61592627e-01
9.33240950e-01 4.64143455e-01 1.03987157e+00 6.35151386e-01
-3.09840411e-01 1.07538593e+00 -5.12261510e-01 -4.21962328e-03
6.66367486e-02 -5.13049543e-01 -7.47398973e-01 -9.22174901e-02
7.20136464e-01 -4.70512897e-01 -1.78893283e-01 1.07843935e+00
4.34655905e-01 4.38123882e-01 6.34482682e-01 -1.07362008e+00
-1.14231980e+00 4.66263890e-01 -3.67263258e-01 1.18507914e-01
8.25605631e-01 3.69803280e-01 1.02278113e+00 -1.11113477e+00
5.65511763e-01 1.40517223e+00 -6.80193305e-02 6.31886303e-01
-8.21434379e-01 -4.23911124e-01 8.24180722e-01 5.12930095e-01
-8.93823624e-01 -2.52193272e-01 6.87572420e-01 -4.55419511e-01
1.22246790e+00 1.27847105e-01 4.61754829e-01 1.14020693e+00
3.40992689e-01 8.92043352e-01 1.07879984e+00 -6.11509681e-01
1.87718019e-01 -8.62893462e-02 4.23544884e-01 1.10270286e+00
-1.71566475e-02 2.06758857e-01 -6.85343981e-01 5.44495583e-01
4.52639729e-01 8.31995755e-02 -6.55384898e-01 -5.80401719e-01
-1.40654957e+00 8.40318680e-01 7.01086462e-01 1.92644507e-01
-3.91356587e-01 4.25656773e-02 1.81740239e-01 9.31869000e-02
-1.40866309e-01 4.09922093e-01 -3.19348305e-01 -2.23387673e-01
-1.44333169e-01 6.96364716e-02 4.72548962e-01 1.10795128e+00
8.85258913e-01 -8.16689897e-03 -2.05025777e-01 6.35994792e-01
7.45701730e-01 7.07177937e-01 4.99333322e-01 -1.05544269e+00
7.74256229e-01 5.96444249e-01 1.20172188e-01 -9.36822474e-01
-5.44472933e-01 2.16363538e-02 -4.65762824e-01 5.07446289e-01
3.17282379e-01 -5.40026464e-02 -1.19575548e+00 1.81813240e+00
3.57637793e-01 -2.40937576e-01 3.57602507e-01 1.09460700e+00
1.49683142e+00 7.96329916e-01 5.42589240e-02 5.68126254e-02
1.55728459e+00 -1.61098552e+00 -1.05361593e+00 -1.03195703e+00
7.18415856e-01 -7.39974976e-01 1.86258197e+00 5.67614622e-02
-7.16161430e-01 -8.42738688e-01 -1.07474196e+00 -4.43222493e-01
-5.32752633e-01 2.20389768e-01 3.58518571e-01 9.08985138e-02
-1.09215271e+00 -3.16582412e-01 -7.56174505e-01 -5.02325177e-01
-2.17762925e-02 5.95110189e-03 -5.39636493e-01 -2.62732863e-01
-9.33935523e-01 1.07347834e+00 1.94296792e-01 3.16854447e-01
-9.05960202e-01 -1.44178689e-01 -1.57886767e+00 -1.60208330e-01
4.90531147e-01 -6.76934004e-01 1.54045415e+00 -7.06418216e-01
-1.51390243e+00 7.10406244e-01 -6.73043311e-01 -1.00326911e-01
2.41601374e-02 -4.39170331e-01 4.34866473e-02 -1.20495573e-01
4.27464813e-01 8.19726467e-01 4.35904950e-01 -1.48537505e+00
-8.15047383e-01 -6.58819795e-01 5.11281669e-01 7.16808021e-01
1.60354808e-01 -1.93518266e-01 -7.06871629e-01 -3.55349362e-01
2.74554901e-02 -9.74961758e-01 -1.93977147e-01 -3.58007401e-01
-3.84256363e-01 -3.14000517e-01 5.68595052e-01 -7.68373728e-01
1.07559705e+00 -2.08147597e+00 3.26315761e-01 -2.31859878e-01
1.80809960e-01 -3.10290344e-02 -2.25030780e-01 3.45349103e-01
1.69953004e-01 -1.27723426e-01 -1.13549717e-01 -5.12521386e-01
1.36254594e-01 3.97547662e-01 -2.18151107e-01 1.46335915e-01
-3.21848959e-01 1.10380280e+00 -1.03682375e+00 -4.36083078e-01
5.88016570e-01 4.79261130e-01 -4.65456396e-01 5.37543535e-01
-1.22818463e-01 6.42402887e-01 -3.83113891e-01 5.15961349e-01
2.84285963e-01 -1.33072361e-01 -1.42736793e-01 -1.99375853e-01
-3.54102015e-01 5.03655732e-01 -8.68182480e-01 2.09749532e+00
-7.70256698e-01 8.73068213e-01 8.16523954e-02 -5.88805914e-01
8.56829524e-01 -1.56848401e-01 -3.12751472e-01 -1.19309878e+00
6.05007559e-02 -2.68504154e-02 1.01668112e-01 -8.46331358e-01
4.91645008e-01 4.94521976e-01 -1.88873678e-01 3.85345310e-01
-3.34745973e-01 -9.88085195e-02 3.63016933e-01 2.19863892e-01
8.83839250e-01 4.07657295e-01 4.99596745e-01 -4.91264276e-02
6.05282605e-01 9.66546759e-02 2.20435128e-01 9.08854306e-01
-4.89416122e-01 3.57702166e-01 -3.78181897e-02 -4.96371120e-01
-5.07659554e-01 -9.42451060e-01 4.25356358e-01 1.52580214e+00
5.32869875e-01 -6.07344091e-01 -6.28784239e-01 -8.05931211e-01
-4.35679376e-01 1.21146989e+00 -7.98609197e-01 1.13772094e-01
-7.74411738e-01 -1.74210966e-02 -1.83131725e-01 6.78071856e-01
6.84333801e-01 -1.54385519e+00 -1.28803945e+00 -1.48444220e-01
-6.14133477e-01 -1.21602464e+00 -9.14375722e-01 3.97452086e-01
-3.81807953e-01 -1.07644951e+00 -2.39751458e-01 -1.27312040e+00
9.25818741e-01 6.23347580e-01 1.42135358e+00 9.28266495e-02
-2.79270690e-02 7.47352481e-01 -5.39624929e-01 -4.49211478e-01
-3.06205124e-01 -2.09782854e-01 -2.66321033e-01 -4.61032331e-01
5.38044214e-01 4.16586809e-02 -7.36094475e-01 2.70833641e-01
-2.66354084e-01 6.03863597e-01 5.22105813e-01 8.32820058e-01
5.31089425e-01 -4.36088324e-01 -1.63837641e-01 -4.87497628e-01
8.11217129e-01 1.12867527e-01 -5.94779849e-01 3.97356451e-01
-3.67994010e-01 4.29544270e-01 5.17668605e-01 -5.02932742e-02
-1.24318445e+00 2.60045916e-01 -9.06230323e-03 2.29375586e-02
-5.78434408e-01 5.29914081e-01 -3.47599059e-01 3.88634890e-01
4.86881554e-01 2.73368716e-01 -1.54942095e-01 -1.06775872e-01
6.91373527e-01 5.87097168e-01 6.66126788e-01 -4.84805137e-01
5.01402855e-01 3.86341393e-01 -1.88332185e-01 -5.67894280e-01
-8.51437449e-01 -4.48360145e-01 -8.63721311e-01 -2.89436907e-01
1.33340085e+00 -7.30491996e-01 -9.38485742e-01 1.55613393e-01
-1.32993174e+00 -8.27274323e-01 8.20979923e-02 5.84805787e-01
-5.97988963e-01 1.87567905e-01 -3.55018556e-01 -5.87628186e-01
-1.57116935e-01 -1.71760273e+00 1.14763856e+00 4.87778127e-01
-3.07055324e-01 -1.01028633e+00 1.50273398e-01 4.39453036e-01
1.65791109e-01 -9.98915583e-02 8.72086942e-01 -3.88011426e-01
-5.81915379e-01 3.33241820e-01 -9.86742005e-02 -4.47269157e-02
5.69290340e-01 -9.24315527e-02 -6.48304701e-01 -5.73112257e-02
1.05699912e-01 -2.59882540e-01 7.55922377e-01 5.35750151e-01
7.82687247e-01 -2.84844935e-01 -1.80305064e-01 5.50773561e-01
1.11304605e+00 7.13753819e-01 4.43658024e-01 7.81031251e-01
1.04099643e+00 8.83344710e-01 1.03259838e+00 9.62480437e-03
9.81178522e-01 7.68410087e-01 7.79854894e-01 -1.38994008e-01
-1.91459164e-01 -3.38183522e-01 6.81774259e-01 6.71412289e-01
8.74394178e-02 -3.40952665e-01 -1.11406958e+00 3.32362592e-01
-2.05707908e+00 -8.31557572e-01 -2.26643104e-02 1.99490917e+00
6.73369169e-01 2.16098800e-02 -3.65623832e-01 -4.60945904e-01
3.78236443e-01 3.08834374e-01 -4.90725219e-01 -6.06547296e-01
8.70251358e-02 -2.70356476e-01 1.89303741e-01 1.11006403e+00
-1.09904695e+00 1.20480359e+00 5.09122324e+00 1.06822200e-01
-9.53142822e-01 -2.55066097e-01 4.19941664e-01 1.05405837e-01
-1.82191044e-01 -1.39857396e-01 -9.88551557e-01 2.38320976e-01
5.76289237e-01 1.82770029e-01 5.50342977e-01 6.99242234e-01
6.03901565e-01 -4.93097931e-01 -1.38478673e+00 1.20442140e+00
5.62786758e-01 -1.00299418e+00 3.59129570e-02 -1.61010832e-01
4.05847341e-01 1.16942503e-01 2.61751682e-01 3.72385263e-01
5.27873337e-01 -1.22742975e+00 7.98976839e-01 2.64649212e-01
2.69619942e-01 -5.74062526e-01 6.59861028e-01 3.48493487e-01
-1.37501550e+00 -7.81589299e-02 -1.24246903e-01 -2.19745457e-01
3.21349233e-01 -2.86537588e-01 -1.02746797e+00 1.93487167e-01
9.22656834e-01 7.29349136e-01 -8.84787917e-01 6.37149274e-01
-9.13632929e-01 1.49694353e-01 7.78052434e-02 -3.24136168e-01
5.00610292e-01 -4.29845959e-01 3.57072324e-01 1.07753420e+00
3.91423166e-01 7.95234367e-02 5.30352235e-01 6.02538884e-01
2.16139182e-01 2.19729438e-01 -9.39592004e-01 3.63363415e-01
4.01073247e-01 1.06590903e+00 -6.21911407e-01 -1.87060788e-01
-7.61881232e-01 1.02624357e+00 4.54156190e-01 7.34201849e-01
-8.27834010e-01 -2.94235796e-01 7.42129803e-01 -2.76784092e-01
2.85782158e-01 -6.07760608e-01 -1.72976434e-01 -8.55339408e-01
6.89770281e-02 -1.07118928e+00 3.67207021e-01 -1.41770518e+00
-7.78187215e-01 7.55564868e-01 -9.57414508e-02 -1.12058961e+00
-3.78186375e-01 -9.64408278e-01 -5.03153145e-01 8.97702277e-01
-1.60920930e+00 -1.02120233e+00 -9.18468535e-01 6.31847799e-01
1.32750571e+00 -9.57549363e-02 8.78364503e-01 -2.24918425e-01
-5.16200602e-01 2.09477782e-01 -3.14734876e-01 3.27692628e-01
9.95326340e-01 -1.57764637e+00 4.73488897e-01 9.55462515e-01
5.86050510e-01 8.25844824e-01 8.88018191e-01 -6.69808745e-01
-1.42664421e+00 -6.01532340e-01 7.97394156e-01 -8.61534834e-01
3.94850016e-01 -4.15047318e-01 -7.44399309e-01 9.77726579e-01
8.72288942e-01 -2.54762679e-01 6.03038609e-01 1.49348928e-02
-3.55703741e-01 1.55350000e-01 -5.97290039e-01 1.22418523e+00
1.09398091e+00 -4.68170404e-01 -9.63669002e-01 1.15661860e-01
9.32518363e-01 -8.00574481e-01 6.21343628e-02 1.50966421e-01
4.34353262e-01 -1.14311194e+00 8.77553701e-01 -4.92144138e-01
4.57053602e-01 -6.91082060e-01 -4.32910472e-01 -1.68986845e+00
-3.73796016e-01 -3.10680687e-01 5.05278222e-02 8.58483195e-01
5.26349247e-01 -3.44248772e-01 4.78812546e-01 6.97172165e-01
-4.99856204e-01 -4.87184554e-01 -6.19870543e-01 -3.34692150e-01
-3.04035574e-01 -2.61372268e-01 4.52900320e-01 4.93915498e-01
2.49130875e-02 8.62978518e-01 -1.06069773e-01 2.67786354e-01
2.98121929e-01 3.14589560e-01 1.01108587e+00 -8.60080063e-01
1.90523770e-02 -4.86146927e-01 -3.47548202e-02 -1.58685303e+00
4.53670442e-01 -8.30410063e-01 6.08221591e-01 -2.33537507e+00
1.16672061e-01 9.95582342e-02 -7.99760818e-02 6.88741267e-01
-4.67017591e-01 -3.26626986e-01 5.15973508e-01 -3.14981826e-02
-7.57804155e-01 5.90412557e-01 1.55915976e+00 -4.46424693e-01
-5.30203998e-01 -3.61464411e-01 -7.16524243e-01 9.04079735e-01
8.17289948e-01 -1.91770704e-03 -6.43135130e-01 -8.97324681e-01
1.42576456e-01 -2.48827398e-01 3.50715876e-01 -7.25512862e-01
4.65690434e-01 -4.51953590e-01 2.58050293e-01 -9.00291204e-01
3.52139235e-01 -9.15592909e-01 -4.76221055e-01 4.26292658e-01
-4.47738975e-01 8.44521999e-01 3.14427674e-01 4.27503347e-01
-1.27210319e-01 -2.13998631e-01 2.47839645e-01 -2.97445923e-01
-1.36590552e+00 -7.68656135e-02 -5.15041769e-01 1.14053354e-01
8.65425229e-01 -1.72417134e-01 -7.56093323e-01 -6.23155892e-01
-5.73646188e-01 6.69271946e-01 5.09949028e-01 7.75634408e-01
1.02597582e+00 -1.23609102e+00 -5.43182433e-01 2.18315482e-01
3.64823490e-01 2.50706583e-01 1.70030013e-01 5.48471630e-01
-5.75508058e-01 7.41336644e-01 -3.05583596e-01 -8.07497323e-01
-1.68382978e+00 6.98805749e-01 2.99216449e-01 -1.13185216e-02
-5.70243597e-01 9.31385458e-01 9.26785052e-01 -6.00509346e-01
5.37788391e-01 -6.95314765e-01 -6.78770661e-01 -3.53531651e-02
6.55634701e-01 -7.79343620e-02 -1.44014806e-01 -8.72311771e-01
-5.00386298e-01 8.88114691e-01 -1.16584450e-01 -2.43747666e-01
1.01946366e+00 -5.94484806e-01 -8.78385305e-02 7.38115191e-01
8.69975269e-01 2.20524833e-01 -1.48170984e+00 -1.62392005e-01
-2.42331013e-01 -5.20293355e-01 -7.92145431e-02 -1.01824689e+00
-8.05180132e-01 1.10807586e+00 7.11982250e-01 -1.19110577e-01
9.14050162e-01 1.87976688e-01 3.25908542e-01 8.58736753e-01
3.47325116e-01 -9.31467116e-01 4.65464592e-01 9.49889421e-01
1.26048207e+00 -1.72995198e+00 -2.38464832e-01 -2.03724533e-01
-9.76986945e-01 1.10066092e+00 1.17974401e+00 5.13282180e-01
2.97449887e-01 1.06046118e-01 7.23208904e-01 -3.68853688e-01
-6.30520403e-01 -5.94857812e-01 3.12077373e-01 9.05664742e-01
5.67274988e-01 6.98686689e-02 1.43690735e-01 3.06077272e-01
-4.89096195e-01 -6.29443228e-01 5.13572931e-01 1.02152777e+00
-5.64233840e-01 -9.18521047e-01 -3.84351104e-01 -3.79730277e-02
1.80281207e-01 -3.93083841e-01 -4.92958069e-01 7.65782177e-01
-5.24346121e-02 1.39322209e+00 2.59758867e-02 -1.28578186e-01
6.15163267e-01 -1.93905905e-02 3.73414069e-01 -8.44764471e-01
-4.29578513e-01 9.82097466e-04 -1.83577016e-02 -7.85924733e-01
-4.97709662e-01 -4.96769845e-01 -1.81258452e+00 6.40342012e-02
2.14127570e-01 3.32385600e-02 6.82343960e-01 8.29956591e-01
3.05853963e-01 8.69631886e-01 2.29407191e-01 -8.74607742e-01
-8.30278359e-03 -7.66873360e-01 1.61303520e-01 4.41155821e-01
6.20842814e-01 -7.96785355e-01 -3.65735710e-01 8.30043331e-02] | [4.4437055587768555, 0.44788965582847595] |
1689216b-f329-42e4-9751-9b1c5277dc37 | focusing-on-context-is-nice-improving | 2210.06164 | null | https://arxiv.org/abs/2210.06164v1 | https://arxiv.org/pdf/2210.06164v1.pdf | Focusing on Context is NICE: Improving Overshadowed Entity Disambiguation | Entity disambiguation (ED) is the task of mapping an ambiguous entity mention to the corresponding entry in a structured knowledge base. Previous research showed that entity overshadowing is a significant challenge for existing ED models: when presented with an ambiguous entity mention, the models are much more likely to rank a more frequent yet less contextually relevant entity at the top. Here, we present NICE, an iterative approach that uses entity type information to leverage context and avoid over-relying on the frequency-based prior. Our experiments show that NICE achieves the best performance results on the overshadowed entities while still performing competitively on the frequent entities. | ['Evangelos Kanoulas', 'Roberto Navigli', 'Svitlana Vakulenko', 'Simone Tedeschi', 'Vera Provatorova'] | 2022-10-12 | null | null | null | null | ['entity-disambiguation'] | ['natural-language-processing'] | [-4.28294957e-01 5.51218987e-01 -5.19813895e-01 -2.00257018e-01
-8.83639991e-01 -7.96613455e-01 4.78151530e-01 7.56387949e-01
-8.32477629e-01 1.09076273e+00 4.32275116e-01 -2.37943128e-01
-9.24713835e-02 -8.21192384e-01 -6.43583775e-01 -9.16233659e-02
-3.70780438e-01 8.42985392e-01 6.89618647e-01 -2.35264331e-01
7.05989376e-02 2.15977892e-01 -1.27447927e+00 3.01967382e-01
1.03100610e+00 5.71671009e-01 2.38785148e-01 1.00162141e-01
-4.51838553e-01 4.98967856e-01 -6.63633049e-01 -8.15978646e-01
3.50525714e-02 4.42476779e-01 -1.18393600e+00 -5.67679763e-01
6.61533237e-01 1.22935310e-01 -1.27127841e-01 1.13435340e+00
3.14765066e-01 3.61856133e-01 5.00026047e-01 -8.95348191e-01
-7.69792616e-01 1.11232245e+00 -4.70814049e-01 6.63948953e-01
5.98746300e-01 -6.75131202e-01 1.54368854e+00 -1.45289230e+00
1.17699504e+00 1.32662630e+00 7.37348258e-01 1.07247658e-01
-1.04866421e+00 -5.85143924e-01 5.33429265e-01 3.21833462e-01
-1.69365966e+00 -1.85422808e-01 7.12462664e-02 -2.49753833e-01
1.41854203e+00 1.56960681e-01 6.13961294e-02 5.23141861e-01
-2.85044342e-01 5.52483201e-01 6.37474060e-01 -5.06566584e-01
5.12340330e-02 1.27791643e-01 5.91202617e-01 3.13288689e-01
1.04236054e+00 -1.89484060e-01 -4.26202953e-01 -4.96377081e-01
9.21457559e-02 -2.45696187e-01 -1.93427548e-01 -9.19829607e-02
-9.82592523e-01 4.13637489e-01 5.93170643e-01 5.75730085e-01
-6.69870794e-01 -1.20440543e-01 1.90318912e-01 -8.17568973e-02
2.34867394e-01 1.01667154e+00 -8.91128540e-01 1.51608974e-01
-7.60835052e-01 6.03319705e-01 1.00226569e+00 1.17068660e+00
1.13254821e+00 -5.99231064e-01 -3.05472702e-01 6.99862301e-01
2.03479394e-01 3.08395177e-01 1.56870484e-01 -7.57594645e-01
6.47975862e-01 8.70258391e-01 7.54771411e-01 -9.31435108e-01
-5.32613814e-01 -4.78611708e-01 -9.40592885e-02 -4.94259536e-01
1.03649698e-01 -2.36675754e-01 -1.03723073e+00 1.83332145e+00
4.91047949e-01 2.53234386e-01 2.27865919e-01 6.12616241e-01
1.13729358e+00 3.38425934e-01 8.89878333e-01 -1.44726768e-01
1.72827339e+00 -6.71760440e-01 -9.28647697e-01 -5.94135165e-01
7.92530537e-01 -7.86586165e-01 7.48818338e-01 -3.17530572e-01
-8.93318295e-01 -5.64115644e-02 -1.02381945e+00 -1.76109374e-01
-8.25231612e-01 -1.61001347e-02 7.26011395e-01 3.41889918e-01
-7.72773683e-01 5.52350104e-01 -5.98280489e-01 -4.97726142e-01
-8.62620845e-02 3.45410913e-01 -5.82021236e-01 -1.42187789e-01
-1.88501883e+00 1.32254457e+00 9.25088227e-01 -3.53080004e-01
-5.64992540e-02 -1.01465929e+00 -1.16217291e+00 1.61936954e-01
9.77787852e-01 -6.98419034e-01 1.20325816e+00 -1.81310102e-01
-4.27121967e-01 8.29625845e-01 -4.35782075e-01 -6.67092443e-01
-1.00171521e-01 -5.30085266e-01 -9.84648466e-01 -6.43700436e-02
7.50228405e-01 4.29737806e-01 -1.58928905e-03 -1.24915743e+00
-1.15986180e+00 -1.87350139e-01 4.06132072e-01 3.97658795e-01
5.50994016e-02 7.74845704e-02 -7.22452104e-01 -6.05577826e-01
2.40977764e-01 -8.20134223e-01 -2.72048771e-01 -1.00010705e+00
-6.07357562e-01 -5.49496710e-01 5.09108782e-01 -6.19654596e-01
1.94611013e+00 -1.91536093e+00 -2.79019237e-01 2.22415656e-01
2.10042611e-01 2.21866131e-01 3.29423130e-01 6.03924096e-01
-1.59552738e-01 5.43880880e-01 1.27943739e-01 -2.02556534e-04
9.98580381e-02 1.97324365e-01 -3.49425435e-01 -1.51568681e-01
1.12279430e-01 8.87090802e-01 -1.28490388e+00 -6.48516178e-01
-4.59976017e-01 1.09133296e-01 -4.41492260e-01 -2.27713078e-01
-1.30248606e-01 -2.89178818e-01 -5.95132828e-01 7.35439956e-01
5.32464743e-01 -3.84772420e-01 5.93872011e-01 -4.51593429e-01
-1.84435993e-01 8.14895868e-01 -1.47326970e+00 1.23251903e+00
-3.06807756e-01 1.64422110e-01 2.31166016e-02 -2.20782265e-01
5.23460150e-01 3.65486145e-01 3.89994621e-01 -5.46709418e-01
-3.84849787e-01 5.46785653e-01 -2.45979398e-01 -3.05283695e-01
1.22237575e+00 5.82883246e-02 -4.56203431e-01 6.33772463e-02
1.40351236e-01 5.80898464e-01 5.57686687e-01 7.05115020e-01
1.07772279e+00 -1.22775175e-01 8.36211085e-01 -4.67722893e-01
3.41154069e-01 3.66680562e-01 1.03047419e+00 7.82773554e-01
4.89697866e-02 -1.31528890e-02 7.20110014e-02 -2.37094864e-01
-6.72474980e-01 -1.09433031e+00 -2.09139302e-01 1.25826085e+00
5.49139738e-01 -9.42595601e-01 -4.58566010e-01 -1.12820649e+00
5.01973927e-01 9.59485352e-01 -5.68160474e-01 1.18650563e-01
-6.75272048e-01 -7.28369653e-01 3.85167003e-01 7.04533517e-01
1.88112274e-01 -6.91895664e-01 -1.27572507e-01 4.46835458e-01
-4.83453661e-01 -1.22495866e+00 -4.57162619e-01 3.56865674e-01
-3.56455475e-01 -9.64051187e-01 -4.85891134e-01 -8.61133516e-01
6.78620160e-01 1.67116314e-01 1.66199040e+00 1.84783354e-01
1.20660841e-01 1.38462424e-01 -2.66874999e-01 -5.19511521e-01
-3.40763964e-02 4.63980079e-01 2.02988639e-01 -5.59760511e-01
8.82977545e-01 -3.22336644e-01 -4.33695704e-01 3.58833909e-01
-6.67517066e-01 -6.08951509e-01 5.90507865e-01 6.23531401e-01
6.80875421e-01 5.36539480e-02 8.93008173e-01 -1.52080858e+00
4.13238347e-01 -8.32979441e-01 -2.91176498e-01 6.25090718e-01
-1.03236604e+00 3.58925581e-01 1.50204107e-01 -1.53976068e-01
-1.20143819e+00 -1.95993446e-02 1.48489382e-02 -6.86698686e-03
-1.34905027e-02 1.01749718e+00 -4.00477648e-01 2.29413092e-01
8.44675124e-01 -4.39368546e-01 -1.22373605e+00 -7.53905118e-01
5.26808739e-01 3.12309772e-01 8.37507963e-01 -9.98478472e-01
7.15524912e-01 2.41365850e-01 -2.55517572e-01 -2.24224806e-01
-1.18340874e+00 -9.59448993e-01 -7.54934311e-01 2.77240515e-01
5.58281422e-01 -1.29163229e+00 -1.93257242e-01 -3.23456347e-01
-9.38614130e-01 4.02123302e-01 -1.70818359e-01 3.66839319e-01
1.55675322e-01 4.72759515e-01 -4.17771667e-01 -3.95743966e-01
-2.17355907e-01 -5.20607769e-01 8.90828550e-01 3.97605717e-01
-4.53261733e-01 -9.67329264e-01 2.51818031e-01 -4.09202576e-02
2.11155415e-01 1.23099372e-01 1.07936728e+00 -1.45879805e+00
-5.26634276e-01 -3.63147616e-01 -1.87838122e-01 -5.28343797e-01
2.11108357e-01 -1.15086317e-01 -5.80590308e-01 -4.21520583e-02
-7.53756940e-01 2.07191497e-01 9.12273049e-01 -3.91435549e-02
4.13766026e-01 -4.43391800e-01 -1.00492394e+00 2.26556629e-01
1.54208684e+00 1.60610318e-01 3.94492656e-01 8.15600336e-01
5.99089503e-01 4.61814851e-01 1.16495132e+00 1.99047700e-01
8.94069731e-01 7.65727699e-01 -4.60736570e-04 6.63675889e-02
6.86661899e-02 -6.00570560e-01 -2.21116483e-01 2.93269396e-01
4.56894077e-02 -2.76735753e-01 -1.15123284e+00 1.13647795e+00
-1.85260558e+00 -1.05494380e+00 -5.16487546e-02 2.25759482e+00
1.13193238e+00 2.71099865e-01 5.14339143e-03 -3.90568554e-01
1.02693582e+00 -1.05072856e-01 -3.17557693e-01 4.75020669e-02
-3.75923604e-01 6.37562647e-02 7.95507729e-01 7.35566497e-01
-1.28432810e+00 1.34777272e+00 7.15792036e+00 6.36759818e-01
-5.73161900e-01 -6.08232245e-02 8.40186179e-02 1.66461527e-01
-5.50300062e-01 3.19413364e-01 -1.51056588e+00 4.28797722e-01
6.91399992e-01 -8.73902917e-01 -2.30094656e-01 1.06427383e+00
-3.65517050e-01 -3.42765152e-01 -1.17919362e+00 4.84370142e-01
-3.20927382e-01 -1.29800951e+00 2.22671121e-01 2.22147200e-02
8.68123353e-01 1.86462663e-02 -2.90238470e-01 6.80349052e-01
8.47268045e-01 -7.43031442e-01 6.08482242e-01 2.78806597e-01
5.52392364e-01 -6.86345279e-01 8.52834523e-01 1.34750992e-01
-1.52693760e+00 1.32439151e-01 -3.25199842e-01 2.87607849e-01
3.17234278e-01 6.60919547e-01 -1.34857893e+00 8.52692842e-01
8.36810052e-01 4.45512086e-01 -4.18452054e-01 1.17239594e+00
-3.36864173e-01 3.00254941e-01 -3.61142874e-01 1.54881015e-01
1.23358078e-01 4.72348958e-01 8.57752621e-01 1.68889141e+00
3.03838938e-01 3.57732177e-01 4.13290530e-01 4.62558687e-01
-4.57024395e-01 2.37355947e-01 -5.49481511e-01 1.71317235e-01
1.20229244e+00 1.16859567e+00 -5.84711432e-01 -7.72736073e-01
-4.68098342e-01 5.47918558e-01 6.01784229e-01 3.22707236e-01
-5.78740537e-01 -5.24310529e-01 7.14811265e-01 2.36857273e-02
6.50259018e-01 4.93569002e-02 -1.81285247e-01 -1.11766684e+00
7.30223209e-02 -3.29393297e-01 1.16661632e+00 -5.24695396e-01
-1.29642236e+00 6.45376325e-01 2.12435856e-01 -9.12920892e-01
-3.33799481e-01 -2.84878105e-01 -3.55747014e-01 1.00263441e+00
-1.69748557e+00 -6.92237973e-01 -3.75069082e-02 1.85066491e-01
1.26213804e-01 3.29221636e-01 8.44699681e-01 7.51977146e-01
-4.55016136e-01 7.25606918e-01 -1.00232869e-01 2.06168056e-01
1.08494306e+00 -1.84444129e+00 5.36715329e-01 1.08602810e+00
1.42932907e-01 1.28152859e+00 1.02429199e+00 -1.20308769e+00
-7.14365363e-01 -1.14754713e+00 1.65213692e+00 -7.47464776e-01
6.38928831e-01 1.52820408e-01 -1.23428667e+00 1.02240419e+00
1.49994805e-01 -1.16858236e-01 8.31901550e-01 8.44054937e-01
-6.40949786e-01 2.85363823e-01 -1.21937251e+00 4.82612073e-01
1.11247146e+00 -4.69866306e-01 -1.38405323e+00 2.42645256e-02
7.66479611e-01 -5.59549689e-01 -1.06663859e+00 6.43134713e-01
2.37313226e-01 -2.13367000e-01 1.20474005e+00 -7.22359300e-01
-2.49465238e-02 -6.37411475e-01 -2.41776988e-01 -1.32918406e+00
-6.82356477e-01 -5.03535092e-01 -5.87943792e-01 1.48555768e+00
9.94383931e-01 -3.36644381e-01 6.46811843e-01 1.02727759e+00
-6.46838695e-02 -4.99794602e-01 -5.78983963e-01 -1.00451636e+00
-2.14987755e-01 1.67309314e-01 9.63408411e-01 1.29431319e+00
4.96467888e-01 4.58604842e-01 4.87669595e-02 7.53881156e-01
4.22242761e-01 2.29766831e-01 2.66353011e-01 -1.52723312e+00
2.18500093e-01 -2.05919743e-01 -2.41253346e-01 -9.81320679e-01
6.23834841e-02 -9.50453818e-01 1.34996474e-01 -1.73550272e+00
1.71697766e-01 -7.22984314e-01 -6.85412943e-01 6.11478269e-01
-9.34175789e-01 -1.38550058e-01 -5.15167415e-03 2.89851397e-01
-1.12437677e+00 -1.39032342e-02 6.65577888e-01 1.07929580e-01
-2.45786175e-01 -2.81099528e-01 -1.21099806e+00 7.95771122e-01
2.00934619e-01 -6.99679315e-01 -4.25613411e-02 -4.53177691e-01
5.82897365e-01 -2.06453606e-01 -1.47863328e-01 -6.60843492e-01
5.86035609e-01 -1.28309190e-01 9.68008637e-02 -8.34939361e-01
-1.42090339e-02 -7.05533385e-01 2.19532520e-01 2.67684031e-02
-3.27224076e-01 3.85346621e-01 3.22054088e-01 6.53063178e-01
-3.49631667e-01 -3.54289234e-01 2.85215080e-01 -2.34110653e-01
-1.47824430e+00 2.14740440e-01 -3.71774323e-02 4.75673318e-01
8.36125493e-01 5.92415892e-02 -5.60893297e-01 2.00089708e-01
-1.06057501e+00 4.92107153e-01 5.37428439e-01 4.96485382e-01
9.11176652e-02 -1.35413063e+00 -4.69137669e-01 -5.16385257e-01
5.03883243e-01 8.19218829e-02 1.19228043e-01 3.54629487e-01
7.89800286e-03 7.13075101e-01 1.68114707e-01 -8.48036408e-02
-1.16373444e+00 6.61558747e-01 2.57649183e-01 -7.20373392e-01
-4.42369252e-01 9.44509983e-01 1.37498528e-01 -3.00604671e-01
2.39757001e-01 -2.22568840e-01 -5.68158209e-01 3.17826092e-01
6.63223267e-01 3.18626881e-01 5.21290302e-01 -6.48831367e-01
-8.09144974e-01 1.41233250e-01 -5.51142871e-01 -8.17886069e-02
1.07512355e+00 -3.66657764e-01 -7.32794106e-02 2.33386055e-01
6.35159969e-01 6.56046093e-01 -6.08425677e-01 -8.28213274e-01
1.06479895e+00 -4.41471279e-01 -4.55584526e-02 -1.08797371e+00
-6.31409526e-01 -7.56867379e-02 -2.07403358e-02 2.90479809e-01
6.40386283e-01 2.31172979e-01 7.93767810e-01 7.65969098e-01
7.30785549e-01 -1.19401383e+00 -6.94700181e-01 8.07534814e-01
5.49997687e-01 -1.08859253e+00 1.92913219e-01 -9.83537734e-01
-6.94640577e-01 5.92857361e-01 9.48090792e-01 2.03075200e-01
7.74095058e-01 4.09506917e-01 -8.68798047e-03 -4.24536347e-01
-8.67804050e-01 -6.74639404e-01 5.19601762e-01 4.07550961e-01
5.82769334e-01 1.42308250e-01 -5.97633421e-01 1.02126491e+00
-1.72103867e-01 -2.80022204e-01 2.60715663e-01 1.06193984e+00
-6.94259584e-01 -1.17973518e+00 -2.14159071e-01 3.30403537e-01
-1.00157452e+00 -5.42784691e-01 -5.50461709e-01 8.73884976e-01
1.66768745e-01 6.75172865e-01 -4.79297452e-02 -2.43637055e-01
7.34709740e-01 3.23353291e-01 -2.61138510e-02 -1.01531303e+00
-7.19052136e-01 -2.39819482e-01 6.84644222e-01 -3.70054126e-01
-3.43700618e-01 -7.65505493e-01 -1.66937017e+00 -1.92249060e-01
-5.27592778e-01 6.77252650e-01 1.96889967e-01 1.05711603e+00
7.75765896e-01 2.27571175e-01 2.79621750e-01 -1.55846551e-01
-9.39484015e-02 -7.12162614e-01 -4.90964085e-01 5.53459942e-01
3.10326666e-01 -1.08927965e+00 -1.41687736e-01 -2.42274404e-01] | [9.423666000366211, 8.843281745910645] |
b2ef01e4-f749-46a3-833f-3128c0071d68 | reaction-network-analysis-of-metabolic | 2204.04614 | null | https://arxiv.org/abs/2204.04614v3 | https://arxiv.org/pdf/2204.04614v3.pdf | Reaction Network Analysis of Metabolic Insulin Signaling | Absolute concentration robustness (ACR) and concordance are novel concepts in the theory of robustness and stability within Chemical Reaction Network Theory. In this paper, we have extended Shinar and Feinberg's reaction network analysis approach to the insulin signaling system based on recent advances in decomposing reaction networks. We have shown that the network with 20 species, 35 complexes, and 35 reactions is concordant, implying at most one positive equilibrium in each of its stoichiometric compatibility class. We have obtained the system's finest independent decomposition consisting of 10 subnetworks, a coarsening of which reveals three subnetworks which are not only functionally but also structurally important. Utilizing the network's deficiency-oriented coarsening, we have developed a method to determine positive equilibria for the entire network. Our analysis has also shown that the system has ACR in 8 species all coming from a deficiency zero subnetwork. Interestingly, we have shown that, for a set of rate constants, the insulin-regulated glucose transporter GLUT4 (important in glucose energy metabolism), has stable ACR. | ['Angelyn R. Lao', 'Eduardo R. Mendoza', 'Patrick Vincent N. Lubenia'] | 2022-04-10 | null | null | null | null | ['novel-concepts'] | ['reasoning'] | [ 4.36538517e-01 1.41196504e-01 -9.81364325e-02 2.95115858e-01
2.82008410e-01 -1.09092009e+00 3.52885127e-01 3.17142874e-01
-1.23433210e-01 1.15685976e+00 4.33158614e-02 -4.52588707e-01
-7.66242445e-01 -6.12456977e-01 -5.28844357e-01 -9.57430243e-01
-6.20452285e-01 4.55917567e-01 1.80460885e-01 -6.23162866e-01
2.41350420e-02 6.36129498e-01 -9.73003924e-01 -1.94020271e-01
6.97393358e-01 5.60713649e-01 -6.23224199e-01 1.00215852e+00
1.42277598e-01 8.13364804e-01 -3.44808221e-01 -9.02884007e-02
4.35360432e-01 -8.40576291e-01 -8.20074558e-01 -9.21648964e-02
-5.21160781e-01 5.79023957e-01 -1.74172685e-01 8.16990852e-01
3.78149927e-01 5.58113046e-02 9.00082469e-01 -1.61578441e+00
-2.43082836e-01 8.94354939e-01 -5.39115846e-01 1.73313335e-01
2.42082268e-01 -7.09873799e-04 1.09905279e+00 -4.89757746e-01
8.24609995e-01 1.14648283e+00 6.46336615e-01 6.16644681e-01
-1.87437701e+00 -5.33541203e-01 8.52154568e-02 -3.42614472e-01
-1.54620838e+00 -4.24527347e-01 4.30484354e-01 -5.68550050e-01
9.79961097e-01 6.71972990e-01 1.12909400e+00 6.45607471e-01
6.30161822e-01 2.59206593e-02 1.04657662e+00 -2.18788758e-01
4.24206167e-01 -2.42506310e-01 2.99471647e-01 5.73782980e-01
8.73029172e-01 2.40142699e-02 -1.51445359e-01 -2.02141210e-01
8.77256930e-01 -1.56014815e-01 -4.18410569e-01 -5.94201207e-01
-1.18095899e+00 5.36385059e-01 1.52810127e-01 4.57357049e-01
-6.71981424e-02 1.84337631e-01 4.09923524e-01 4.26793873e-01
3.20725858e-01 5.68708181e-01 -4.68345970e-01 3.89760882e-01
-1.84660271e-01 -1.10349387e-01 1.31925464e+00 8.16998363e-01
5.41325152e-01 -1.82960585e-01 5.24363160e-01 4.09158051e-01
3.09644580e-01 2.87054516e-02 -1.39737539e-02 -8.86226058e-01
3.52514118e-01 9.26533937e-01 1.30915329e-01 -8.80048692e-01
-1.09375775e+00 -6.56584799e-01 -1.32509315e+00 1.33678630e-01
7.90216684e-01 -2.92859942e-01 -4.18446630e-01 2.04106450e+00
2.17871442e-01 -1.35862783e-01 4.01914984e-01 7.39984751e-01
2.24789947e-01 5.82116008e-01 2.33811006e-01 -8.34635079e-01
1.11331713e+00 -4.35914040e-01 -4.71999437e-01 6.08570397e-01
4.91392314e-01 -4.36046153e-01 1.45664051e-01 5.60516834e-01
-1.19295347e+00 2.77364589e-02 -1.38977647e+00 3.41862470e-01
-6.36217475e-01 -5.06735444e-01 7.18714297e-01 5.74871361e-01
-1.20610631e+00 8.54529619e-01 -5.66803336e-01 -7.86924362e-01
-1.08374991e-01 4.96882737e-01 -4.66879576e-01 1.58340350e-01
-1.46912956e+00 9.61201727e-01 2.88451761e-01 2.40575090e-01
-7.34835386e-01 -5.67328095e-01 -5.79519570e-01 3.53776142e-02
2.17438400e-01 -1.14928365e+00 5.22976995e-01 -8.47530246e-01
-1.37408423e+00 3.88974458e-01 2.65585691e-01 -6.93430156e-02
5.36507308e-01 9.29936469e-01 -3.95159036e-01 2.21778050e-01
-7.82594755e-02 1.19206235e-01 4.47291322e-02 -1.15507364e+00
-8.09418503e-03 -2.76066184e-01 1.40753850e-01 1.69704363e-01
1.43434882e-01 1.42937630e-01 3.41326535e-01 -2.55738109e-01
6.12427056e-01 -7.72647202e-01 -6.41626179e-01 -1.05407193e-01
-8.09536338e-01 -7.63124749e-02 -1.55469865e-01 -9.32604074e-03
9.83424306e-01 -1.65123582e+00 7.65583932e-01 9.05373514e-01
6.73704386e-01 -4.36737061e-01 -1.63481236e-01 8.48266900e-01
-1.03156114e+00 8.37524652e-01 -2.94284195e-01 3.50124121e-01
6.31499887e-02 6.89548552e-02 2.64032364e-01 9.83263910e-01
2.05880463e-01 4.93169695e-01 -9.27825630e-01 -1.80532262e-01
-4.84126657e-02 4.00249749e-01 -2.47964531e-01 -4.14202034e-01
9.67785865e-02 3.58740032e-01 -2.50161082e-01 6.29682839e-01
5.36867857e-01 -2.94409186e-01 8.32012713e-01 -2.02190250e-01
-3.31791967e-01 -2.88744837e-01 -1.34388375e+00 1.26923883e+00
3.04511130e-01 1.07632145e-01 1.93164855e-01 -7.87065923e-01
6.39687181e-01 6.19907320e-01 1.05756760e+00 -1.96325243e-01
3.79291922e-01 7.38725215e-02 5.88963866e-01 -7.89187104e-02
-1.06302775e-01 -4.74981040e-01 3.73735349e-03 5.94276845e-01
2.24176273e-02 2.72021472e-01 7.61853278e-01 5.52100182e-01
1.43679154e+00 -3.17586154e-01 5.56511283e-01 -9.90954340e-01
7.27847338e-01 -7.70637915e-02 5.83991885e-01 3.79235119e-01
-5.34604311e-01 1.77719697e-01 1.44881988e+00 1.46880420e-02
-1.23803890e+00 -1.08614075e+00 -8.88792649e-02 5.14837623e-01
2.25435331e-01 -4.61351901e-01 -6.87026739e-01 2.08202273e-01
1.26237571e-01 -1.33314535e-01 -8.21864784e-01 -2.35619426e-01
-1.70234099e-01 -1.30382156e+00 7.43379831e-01 1.21363200e-01
1.09740801e-03 -1.49513662e-01 1.37297645e-01 4.16390628e-01
-3.05299293e-02 -6.05468333e-01 -2.95323461e-01 5.51215172e-01
-7.61725605e-01 -1.76284802e+00 -4.44738597e-01 -2.09450215e-01
9.34984028e-01 1.67762637e-01 7.59738922e-01 1.30670266e-02
-4.27809864e-01 3.17363650e-01 8.96729752e-02 -9.98344719e-02
-6.50857627e-01 -2.67430484e-01 5.13036728e-01 -1.46549329e-01
-4.54650998e-01 -8.53548765e-01 -6.52369916e-01 6.44206703e-01
-9.27809000e-01 -2.83295631e-01 1.92005821e-02 4.40861017e-01
5.07300973e-01 4.51229900e-01 7.46481657e-01 -4.30720031e-01
7.17470169e-01 -8.58369648e-01 -5.68262279e-01 1.78031072e-01
-7.59456992e-01 -5.10734646e-03 6.38243020e-01 -1.86083674e-01
-5.40497959e-01 1.92197174e-01 2.59761393e-01 2.99388677e-01
2.76888818e-01 5.46656549e-01 -3.83399010e-01 -4.00036484e-01
7.43026614e-01 -5.81043512e-02 3.47825050e-01 -5.32849133e-02
3.76563638e-01 -2.52543360e-01 -1.09769451e-02 -6.22060478e-01
5.92400730e-01 3.56737107e-01 8.70553017e-01 -6.78750694e-01
1.65087692e-02 -4.32684124e-01 -4.48790789e-01 -5.34043789e-01
6.95468545e-01 -7.61531472e-01 -1.55045903e+00 3.53345960e-01
-8.57824087e-01 -5.15522659e-02 -2.26982847e-01 3.76447767e-01
-7.17339158e-01 5.24206758e-01 -1.06759739e+00 -8.39247763e-01
5.54456748e-02 -8.16587031e-01 9.72479209e-02 -3.81433107e-02
-2.47306749e-01 -1.16829014e+00 5.48017979e-01 -1.81287751e-01
3.18018138e-01 8.26548517e-01 9.28268194e-01 -6.90630317e-01
-4.43461657e-01 -5.39132506e-02 -3.37520763e-02 -1.21923149e-01
1.34828538e-01 7.09002674e-01 -4.01188850e-01 -2.57180214e-01
-1.46123365e-01 3.01502526e-01 7.26674497e-01 3.35646063e-01
2.33014151e-01 -2.04163119e-01 -3.83514583e-01 2.42334306e-01
1.79080796e+00 3.10183138e-01 6.33841157e-01 1.31036371e-01
4.48522836e-01 7.25614548e-01 8.36821180e-03 1.83479428e-01
-1.01008825e-01 6.58544675e-02 3.77818674e-01 -3.28881174e-01
3.44091564e-01 6.61773980e-02 2.99305737e-01 7.28099644e-01
-6.73903942e-01 -5.74661374e-01 -5.17658055e-01 1.28628155e-02
-1.94879127e+00 -1.06233025e+00 -6.29010081e-01 1.87754130e+00
9.42733109e-01 1.93566352e-01 5.23730576e-01 4.06992733e-01
9.14491773e-01 -4.34331149e-01 -5.53355038e-01 -3.78297567e-01
-4.91470814e-01 6.13579080e-02 9.13053751e-01 8.29379857e-01
-6.42701864e-01 2.84777015e-01 8.24394131e+00 9.70231667e-02
-4.53123868e-01 -1.31863728e-01 4.80553210e-01 -4.98511037e-03
-2.54498392e-01 1.76451549e-01 -2.40006939e-01 9.73371267e-02
1.06693888e+00 -6.09963238e-01 3.66095185e-01 2.54752249e-01
6.19888663e-01 -3.10104877e-01 -1.08095229e+00 3.52875471e-01
-4.27980095e-01 -1.12867820e+00 -1.61306709e-01 3.72161359e-01
7.34869123e-01 -4.61721480e-01 -4.65939820e-01 -3.82561862e-01
6.92341745e-01 -7.61977255e-01 4.23275143e-01 6.61852241e-01
4.91843700e-01 -9.83699858e-01 4.53178376e-01 4.99878936e-02
-1.38758790e+00 -6.16437197e-02 -2.02829182e-01 -3.38551998e-01
-4.77648638e-02 8.93386900e-01 -4.99059439e-01 6.42633677e-01
1.22261807e-01 7.78408408e-01 -5.61219826e-02 8.34544957e-01
7.32902810e-02 9.37966630e-02 -3.35665882e-01 -5.91092044e-03
-2.55080581e-01 -6.68841481e-01 5.54427207e-01 9.70200241e-01
-5.50429821e-02 5.73440552e-01 -2.01390639e-01 9.05195236e-01
6.07612841e-02 1.78078562e-01 -6.70545101e-01 -2.06078395e-01
1.45655140e-01 1.33100808e+00 -1.40069449e+00 -3.14101666e-01
1.82106674e-01 6.12743855e-01 -1.73775986e-01 5.38663149e-01
-7.00939476e-01 -3.87636960e-01 1.29447353e+00 -8.97767469e-02
-2.91123331e-01 -2.11456969e-01 -1.89505413e-01 -1.10909843e+00
-5.79494596e-01 -4.59925801e-01 3.77876669e-01 -6.08763933e-01
-1.20904541e+00 5.00014164e-02 -2.62054056e-01 -9.15031612e-01
1.42061695e-01 -7.03740120e-01 -3.43595326e-01 8.76027465e-01
-1.00572288e+00 -6.29275620e-01 3.49201858e-01 6.78330958e-01
-1.39013529e-01 3.35403234e-01 6.39959395e-01 4.10532862e-01
-1.28451240e+00 1.95181370e-01 2.68409073e-01 -2.88965017e-01
4.68410552e-01 -1.25789618e+00 4.34259288e-02 9.39107120e-01
-9.15064275e-01 9.13260520e-01 9.65841413e-01 -7.71233141e-01
-1.68421948e+00 -9.29675758e-01 8.44428420e-01 -3.14142823e-01
1.10599339e+00 -5.59446752e-01 -2.74483413e-01 6.42627180e-01
7.67732337e-02 -4.54344064e-01 8.47531259e-01 -2.00404376e-01
-2.98289210e-01 -2.23650962e-01 -1.41165566e+00 7.35571384e-01
1.29605258e+00 -1.20072395e-01 1.68216437e-01 6.30543888e-01
6.47611797e-01 -2.67454982e-02 -1.39977360e+00 -1.11674834e-02
6.03722453e-01 -8.95137250e-01 1.06250763e+00 -8.92047584e-01
1.65413991e-01 -7.27242053e-01 4.02031578e-02 -1.18908036e+00
-8.36709976e-01 -7.84941912e-01 3.04970920e-01 9.93496358e-01
6.68467820e-01 -1.22353721e+00 2.80950189e-01 6.83563828e-01
-1.23843625e-01 -3.44490945e-01 -7.91239202e-01 -9.97645319e-01
6.34963363e-02 1.47848308e-01 2.94423878e-01 1.09766233e+00
1.07616007e+00 5.09227633e-01 -4.54903282e-02 -1.27941355e-01
7.11518943e-01 -4.54324961e-01 1.31487831e-01 -1.50114405e+00
-4.57412042e-02 -5.71332812e-01 -5.73576450e-01 -2.07834989e-01
-1.32734045e-01 -9.15390372e-01 -9.49585512e-02 -1.36815536e+00
4.87733006e-01 -3.44785511e-01 -5.90697169e-01 4.51198131e-01
4.93567854e-01 3.05473387e-01 -1.00541532e-01 2.06102341e-01
-4.68024880e-01 -2.74522185e-01 1.28717792e+00 -5.14185987e-02
-1.56749845e-01 -2.44085088e-01 -1.17766118e+00 4.05301303e-01
1.01027620e+00 -4.32897925e-01 -5.36400139e-01 7.51841843e-01
9.69407797e-01 3.14581633e-01 1.04701892e-02 -7.09181786e-01
1.99167073e-01 -3.81717622e-01 4.98649240e-01 -7.23602623e-02
7.86032379e-02 -8.78613532e-01 1.08242846e+00 1.11477208e+00
-4.46486741e-01 1.15098067e-01 3.80633119e-03 7.39906788e-01
4.39760447e-01 1.19147763e-01 7.72843361e-01 -3.51147324e-01
-1.82044450e-02 2.10574254e-01 -1.19552231e+00 -3.39976966e-01
1.23650444e+00 -1.99642420e-01 -8.01280379e-01 -2.47844178e-02
-1.32003915e+00 3.13489437e-01 6.44753098e-01 -2.48817727e-01
2.41239250e-01 -1.26962781e+00 -4.94927496e-01 -2.10346401e-01
-1.33942768e-01 -5.79078257e-01 1.23398520e-01 1.19313765e+00
-6.40165150e-01 4.48081851e-01 -6.18259430e-01 -4.12781984e-01
-1.06832683e+00 9.07322943e-01 7.50193596e-01 -1.02389470e-01
1.55964255e-01 4.64275151e-01 5.93960099e-02 -7.82746524e-02
-2.26043776e-01 -6.12735689e-01 -1.61897764e-01 3.18233520e-01
1.96979836e-01 8.18501651e-01 -1.90311104e-01 -5.39715767e-01
-4.65865016e-01 6.79528058e-01 3.99519801e-01 3.26402895e-02
1.08208835e+00 -4.20066446e-01 -9.75075364e-01 3.88729721e-01
1.03579235e+00 -1.71918422e-01 -6.89040363e-01 5.07321239e-01
-1.85000703e-01 2.97027022e-01 -6.11592233e-01 -7.59292841e-01
-9.22313094e-01 6.53072298e-02 1.04902089e-01 7.84139931e-01
1.09126186e+00 -1.70042962e-01 -6.29166961e-02 1.80502653e-01
1.95566565e-01 -7.84421742e-01 -2.01301247e-01 3.00505579e-01
8.30179513e-01 -3.64806622e-01 1.27911776e-01 -9.12637651e-01
5.20845652e-02 1.19863260e+00 2.72932082e-01 -2.03131765e-01
7.33477056e-01 2.64216155e-01 -4.95317221e-01 -1.14161991e-01
-1.04820859e+00 -1.60368100e-01 -3.75899971e-01 5.73114216e-01
3.55689138e-01 2.18035609e-01 -9.40510809e-01 4.40821052e-01
3.12683046e-01 -2.68797159e-01 1.12130594e+00 6.68366075e-01
-5.58390617e-01 -1.08009005e+00 -3.62946928e-01 -1.44558875e-02
-1.27486706e-01 -1.72800594e-03 -9.06520486e-01 1.09527874e+00
8.11871886e-02 1.01102209e+00 -1.65926129e-01 -3.55978578e-01
4.71923947e-01 1.04809359e-01 7.89089501e-01 -2.06067085e-01
-5.32789707e-01 2.46299416e-01 3.05465549e-01 -6.05650902e-01
-5.03883541e-01 -5.60693562e-01 -1.44660211e+00 -8.30447793e-01
-4.07070130e-01 4.47208434e-01 6.06211722e-01 6.46323085e-01
1.34477168e-01 8.26378763e-01 8.42341125e-01 -6.02709174e-01
-1.08530253e-01 -5.94830275e-01 -1.26631629e+00 -1.96570665e-01
2.72036701e-01 -5.96622050e-01 -6.93937182e-01 -1.20775685e-01] | [6.351328372955322, 4.755594253540039] |
c9a91991-c02d-4fd2-800e-91fe5651e089 | efficient-estimation-of-weighted-cumulative | 2305.02373 | null | https://arxiv.org/abs/2305.02373v1 | https://arxiv.org/pdf/2305.02373v1.pdf | Efficient estimation of weighted cumulative treatment effects by double/debiased machine learning | In empirical studies with time-to-event outcomes, investigators often leverage observational data to conduct causal inference on the effect of exposure when randomized controlled trial data is unavailable. Model misspecification and lack of overlap are common issues in observational studies, and they often lead to inconsistent and inefficient estimators of the average treatment effect. Estimators targeting overlap weighted effects have been proposed to address the challenge of poor overlap, and methods enabling flexible machine learning for nuisance models address model misspecification. However, the approaches that allow machine learning for nuisance models have not been extended to the setting of weighted average treatment effects for time-to-event outcomes when there is poor overlap. In this work, we propose a class of one-step cross-fitted double/debiased machine learning estimators for the weighted cumulative causal effect as a function of restriction time. We prove that the proposed estimators are consistent, asymptotically linear, and reach semiparametric efficiency bounds under regularity conditions. Our simulations show that the proposed estimators using nonparametric machine learning nuisance models perform as well as established methods that require correctly-specified parametric nuisance models, illustrating that our estimators mitigate the need for oracle parametric nuisance models. We apply the proposed methods to real-world observational data from a UK primary care database to compare the effects of anti-diabetic drugs on cancer clinical outcomes. | ['Zach Shahn', 'Ioanna Tzoulaki', 'Kenney Ng', 'Roy Welsch', 'Stan Finkelstein', 'Bowen Su', 'Bang Zheng', 'Shenbo Xu'] | 2023-05-03 | null | null | null | null | ['causal-inference', 'causal-inference'] | ['knowledge-base', 'miscellaneous'] | [ 2.74933606e-01 -6.26225844e-02 -1.42120337e+00 -3.38172942e-01
-1.02951658e+00 -2.76288569e-01 1.38806805e-01 4.88620043e-01
-3.95808399e-01 1.11233187e+00 3.00296009e-01 -9.24466252e-01
-7.71235585e-01 -5.00839174e-01 -9.04538393e-01 -5.76912224e-01
-5.59088528e-01 4.62156713e-01 -5.10419011e-01 6.61263287e-01
2.24067494e-01 3.39848578e-01 -7.81289697e-01 -2.30650455e-01
1.17186749e+00 4.57052946e-01 -2.52124399e-01 5.56768179e-01
3.50937903e-01 2.79570758e-01 -2.51121074e-01 -8.51725489e-02
4.82827052e-02 -1.31672814e-01 -2.15073213e-01 -4.88706082e-02
5.44402063e-01 -2.46826783e-01 -5.29772677e-02 6.90369844e-01
6.42687798e-01 -7.03624561e-02 1.33846831e+00 -1.63777542e+00
-6.23045444e-01 6.03504062e-01 -1.08538163e+00 9.80759338e-02
-7.98633695e-02 -3.18378508e-02 7.80883551e-01 -5.67470431e-01
3.45854051e-02 1.34193563e+00 9.00016487e-01 3.87519300e-01
-1.45052302e+00 -9.64153409e-01 -5.56385219e-02 -2.05107823e-01
-1.08275819e+00 -3.04765016e-01 2.71271616e-01 -6.80497408e-01
4.03937638e-01 3.03430498e-01 1.94200315e-02 1.08566833e+00
7.35343933e-01 3.09252083e-01 1.19892645e+00 -5.91954947e-01
4.07991916e-01 1.35658309e-02 3.37217420e-01 4.25726116e-01
7.62990117e-01 7.51275480e-01 4.72619422e-02 -9.23867702e-01
1.09799576e+00 5.72758615e-01 -9.74044129e-02 -5.02471268e-01
-9.54476357e-01 1.38821507e+00 -4.56325747e-02 -1.39461607e-01
-5.05541861e-01 1.86476871e-01 6.38476849e-01 1.39873773e-01
5.72403431e-01 1.13211207e-01 -8.47413123e-01 2.35246509e-01
-8.47226441e-01 1.39647081e-01 7.38321066e-01 7.15761662e-01
-2.54502706e-02 -1.21749684e-01 -3.52812558e-01 7.62393296e-01
1.12736709e-01 6.40990853e-01 3.18471819e-01 -9.00484741e-01
3.81776005e-01 2.98001021e-01 5.31174898e-01 -4.87460196e-01
-6.93257451e-01 -3.25771004e-01 -1.41148293e+00 1.13029964e-02
6.61836565e-01 -4.14028674e-01 -7.16273665e-01 2.00448155e+00
5.02290606e-01 3.78023416e-01 -8.06095973e-02 5.02004445e-01
2.02702284e-01 1.85236856e-01 7.78493583e-01 -9.44500327e-01
1.40859473e+00 -3.63837898e-01 -9.30531740e-01 2.31045187e-01
1.09369183e+00 -3.85979116e-01 1.00719512e+00 1.18817985e-01
-1.15074599e+00 -1.59899279e-01 -5.96240699e-01 3.32459599e-01
-3.81027395e-03 8.39291960e-02 8.23144495e-01 8.72622907e-01
-2.92265087e-01 4.85601872e-01 -7.12269843e-01 -6.65332526e-02
5.70968330e-01 4.53197867e-01 -1.73466071e-01 -1.75840538e-02
-1.13822293e+00 6.10738456e-01 1.67005733e-01 -4.13597040e-02
-7.72955835e-01 -1.72930980e+00 -8.20661843e-01 2.18971685e-01
6.83881164e-01 -1.01026928e+00 1.09153771e+00 -7.63367593e-01
-8.89082670e-01 4.59508002e-01 -1.33556277e-01 -5.02399027e-01
4.85361189e-01 6.96407035e-02 -4.10043567e-01 -2.94237316e-01
2.94863582e-01 -5.75897880e-02 5.68565309e-01 -7.35601544e-01
-8.02925766e-01 -5.48866153e-01 -3.93157303e-01 -1.37566939e-01
-7.39111081e-02 1.32847175e-01 3.48442048e-01 -1.00707674e+00
-4.24752712e-01 -8.15138519e-01 -5.12027144e-01 -8.40372816e-02
-1.36107579e-01 -3.02959442e-01 3.34495038e-01 -5.15823185e-01
1.50525701e+00 -2.07086897e+00 -3.65817666e-01 6.54918253e-02
-6.00288883e-02 -2.11660057e-01 3.51304226e-02 4.39294040e-01
-5.44668257e-01 3.66551012e-01 -4.49531347e-01 2.18871340e-01
8.24713986e-03 6.27021566e-02 -1.00292012e-01 9.88716424e-01
-2.31071822e-02 6.14444911e-01 -6.14196539e-01 -6.24142230e-01
2.17646986e-01 -1.79892592e-02 -5.40798247e-01 3.76050733e-02
1.64764494e-01 1.31047919e-01 -5.29146552e-01 1.92219451e-01
8.39669406e-01 -3.89103562e-01 3.30152541e-01 1.67471185e-01
-2.84368917e-02 -3.42762880e-02 -1.36084855e+00 9.29325700e-01
-6.33869588e-01 3.45696136e-02 1.78648792e-02 -1.58288133e+00
2.51459599e-01 6.31928384e-01 6.17562115e-01 -2.35603631e-01
-1.60350129e-01 3.11611652e-01 7.78251141e-02 -7.16987908e-01
-4.61110532e-01 -9.64539349e-01 -4.02609557e-01 3.08821946e-01
-2.22637191e-01 2.98866451e-01 -2.19079748e-01 -1.90472543e-01
9.79725420e-01 -4.11269933e-01 1.02049720e+00 -5.31525195e-01
1.48111716e-01 -2.52948135e-01 7.46515334e-01 1.14122987e+00
2.76914872e-02 3.60354155e-01 7.36908734e-01 -3.10794264e-01
-1.07554233e+00 -1.13432145e+00 -7.91054964e-01 8.97602975e-01
-4.65002120e-01 4.36676145e-01 -3.23201656e-01 -6.69193029e-01
4.53589857e-01 8.85393202e-01 -1.03030062e+00 -2.60636091e-01
-1.84383243e-01 -1.41704202e+00 4.63562399e-01 8.14020455e-01
-1.73359945e-01 -4.67092782e-01 -4.43188101e-01 1.77785218e-01
3.19122195e-01 -5.23782313e-01 -8.19499075e-01 1.90602183e-01
-1.08831656e+00 -1.55967128e+00 -9.96341646e-01 -3.99222791e-01
4.55946475e-01 -1.52655423e-01 9.39391315e-01 -2.27506846e-01
-8.03621411e-02 3.26025426e-01 1.16374858e-01 -7.98381686e-01
-4.21143800e-01 -3.15123558e-01 2.28236854e-01 -1.68134362e-01
3.82575631e-01 -2.96440870e-01 -8.88927519e-01 4.49271888e-01
-1.00779068e+00 -3.35331023e-01 6.96726203e-01 1.18799484e+00
4.83943343e-01 -1.45728350e-01 1.39496624e+00 -1.16036975e+00
3.77417803e-01 -9.00171220e-01 -8.66130829e-01 3.49882156e-01
-1.03111100e+00 1.35090072e-02 3.75739008e-01 -1.01998830e+00
-1.15315199e+00 -4.33277786e-01 5.57561457e-01 -2.60223448e-01
-2.04873040e-01 7.99577057e-01 -1.10185467e-01 3.67236078e-01
5.71712375e-01 -3.62462223e-01 1.66029587e-01 -3.66842538e-01
-5.70491813e-02 8.78818333e-01 1.73981294e-01 -6.82669938e-01
1.98155090e-01 4.46016639e-01 5.88798523e-01 -5.12114406e-01
-7.72529244e-01 -7.17109799e-01 -1.04454637e-01 5.56901872e-01
5.86419284e-01 -9.84687030e-01 -1.15435767e+00 4.89000827e-02
-5.80608904e-01 -5.37531495e-01 -2.09414542e-01 1.18399906e+00
-8.98891091e-01 2.10909411e-01 -3.81284475e-01 -1.07449532e+00
-2.46851012e-01 -8.34790170e-01 8.74091446e-01 9.47528426e-03
-3.28005403e-01 -1.58541656e+00 3.24273586e-01 -6.82977494e-03
-1.62541419e-02 4.35645729e-01 1.40438044e+00 -8.12897265e-01
2.65638709e-01 -4.61755842e-01 -2.76624531e-01 -1.24450929e-01
4.82234240e-01 -1.33513376e-01 -2.93506145e-01 -3.88164699e-01
-3.49742696e-02 -1.90406926e-02 5.37817478e-01 1.66094422e+00
1.53918636e+00 -7.68452883e-01 -6.50902569e-01 3.68507802e-01
1.30893803e+00 3.39951456e-01 2.17299834e-01 -4.07338291e-02
3.57324123e-01 6.51248217e-01 5.80146074e-01 8.10589135e-01
2.36749664e-01 5.68134487e-01 1.27191201e-01 -3.31626713e-01
5.76717913e-01 -1.15780555e-01 -1.04682684e-01 -2.21435189e-01
1.11930341e-01 -1.02963619e-01 -5.64404905e-01 7.12160766e-01
-1.90965211e+00 -9.66274858e-01 -7.03043997e-01 2.91594744e+00
1.09820104e+00 -6.68160841e-02 5.26130497e-01 -2.54182845e-01
9.08163428e-01 -5.32136202e-01 -6.71038449e-01 -7.21404612e-01
5.72188348e-02 1.16877012e-01 1.06177950e+00 4.08874393e-01
-1.08874249e+00 -9.93172824e-03 6.49709034e+00 9.84738350e-01
-7.05455065e-01 3.48104179e-01 8.74382615e-01 -5.54084182e-02
-5.14098331e-02 5.01804538e-02 -4.78903860e-01 4.67326313e-01
1.32419252e+00 -2.01344386e-01 -3.12196046e-01 4.73192900e-01
1.15125716e+00 -2.45458335e-01 -1.34115219e+00 4.80703086e-01
-5.78096092e-01 -1.11450005e+00 -4.48353738e-01 4.33539867e-01
9.94281411e-01 -5.28305113e-01 2.39129275e-01 2.33099461e-01
5.43907344e-01 -1.27430582e+00 1.83670431e-01 6.03927851e-01
1.11551619e+00 -8.16764653e-01 8.65025163e-01 4.54179078e-01
-4.18628901e-01 -3.84483159e-01 -4.47382748e-01 -9.85407755e-02
1.36231244e-01 1.04822898e+00 -7.50743926e-01 3.69642913e-01
2.85402954e-01 2.89130211e-01 1.04360439e-01 1.32666290e+00
3.64531755e-01 1.04017198e+00 -2.57486612e-01 2.45843187e-01
1.29940465e-01 -1.50017673e-02 2.21120939e-01 1.04732227e+00
5.66206038e-01 1.90407380e-01 -2.07887024e-01 6.83094025e-01
4.23896052e-02 2.42591605e-01 -5.80606818e-01 2.14056045e-01
5.18151104e-01 6.63291872e-01 -3.22784156e-01 -2.60448098e-01
-8.47515583e-01 2.59912580e-01 -1.56161174e-01 5.09906411e-01
-8.72970223e-01 1.80384651e-01 3.31317246e-01 3.05635363e-01
8.72585550e-02 3.50038439e-01 -6.24208450e-01 -8.11914563e-01
-3.22342515e-01 -9.95215237e-01 1.33073294e+00 -2.48516008e-01
-1.54986465e+00 -7.88503826e-01 5.38575172e-01 -7.87774205e-01
-2.52477705e-01 -4.96058762e-01 -6.33294642e-01 9.98465657e-01
-1.11593533e+00 -9.68117177e-01 5.52063167e-01 3.31019580e-01
5.84798098e-01 1.74176320e-01 7.23038316e-01 5.95646538e-02
-6.21504068e-01 6.58545017e-01 5.99977434e-01 -3.32575887e-01
1.08696961e+00 -1.33343709e+00 -4.48485434e-01 1.48334697e-01
-3.58080268e-01 6.42657340e-01 8.87666106e-01 -9.14378107e-01
-1.06704736e+00 -1.09451258e+00 6.62373304e-01 -2.72530347e-01
7.42533565e-01 7.40063041e-02 -9.03038144e-01 8.10363710e-01
-1.27438843e-01 1.31463990e-01 9.58404422e-01 6.76210403e-01
-2.05036893e-01 -3.43584456e-03 -1.20903063e+00 5.42876244e-01
5.39506137e-01 -2.86344141e-02 -4.31389660e-01 3.57548118e-01
5.34438372e-01 2.93241497e-02 -1.18464518e+00 8.36758614e-01
7.90140271e-01 -4.36038256e-01 9.16088104e-01 -1.32293916e+00
2.92409360e-01 2.64698625e-01 8.25272724e-02 -1.23539364e+00
-2.00819194e-01 -5.04999638e-01 2.20187772e-02 9.48678255e-01
6.19731247e-01 -6.38685882e-01 4.49794352e-01 7.50600457e-01
1.86352476e-01 -5.79441428e-01 -1.14901412e+00 -8.76049995e-01
7.27926075e-01 -2.65367955e-01 3.52442175e-01 1.19330764e+00
1.62851855e-01 1.63453445e-01 -5.08397639e-01 3.25568676e-01
9.32135522e-01 2.46344835e-01 4.29266334e-01 -1.29116523e+00
-2.71270365e-01 -4.40587670e-01 1.80921674e-01 -2.44531602e-01
4.44564342e-01 -3.02854240e-01 -3.58275652e-01 -1.00888693e+00
9.61659253e-01 -5.10533333e-01 -3.11945885e-01 2.30738401e-01
-6.05704606e-01 -5.77780128e-01 -5.85071206e-01 -1.95781305e-01
-1.14482470e-01 4.35465485e-01 8.46302569e-01 2.30543893e-02
-4.98474836e-01 7.50356257e-01 -6.68446243e-01 7.95686841e-01
6.82292938e-01 -8.27019393e-01 -5.43479383e-01 4.02368158e-01
9.82082859e-02 7.58256435e-01 7.07187116e-01 -1.55906156e-01
-1.31595597e-01 -8.87179255e-01 3.06583822e-01 -2.11753234e-01
-3.86858284e-01 -1.14201403e+00 5.29367507e-01 6.95098758e-01
-7.34560668e-01 -5.72562218e-02 3.03641021e-01 9.54842150e-01
1.41500488e-01 -2.95677364e-01 6.22215569e-01 2.38346532e-01
2.48204455e-01 3.60514134e-01 -3.60675007e-01 2.69933105e-01
8.84015799e-01 2.10513454e-02 -2.32418686e-01 -4.17549223e-01
-7.41476238e-01 3.61182928e-01 2.61593144e-02 1.05645493e-01
2.55676180e-01 -1.36065114e+00 -1.01101553e+00 -2.97584444e-01
1.53017044e-01 -4.87420589e-01 6.10743046e-01 1.35870624e+00
2.79407501e-01 7.26442516e-01 2.97543287e-01 -3.01913053e-01
-1.16168392e+00 1.19259107e+00 3.50013256e-01 -4.43197221e-01
-2.78738141e-01 1.67363305e-02 9.95552897e-01 -3.67070079e-01
2.91720051e-02 -4.29629296e-01 1.53622985e-01 -5.71525507e-02
3.45236212e-01 8.49367797e-01 -1.76794797e-01 -1.24796396e-02
-2.26248130e-01 2.93089062e-01 2.02332154e-01 -2.85237972e-02
1.15408313e+00 -3.84782612e-01 1.39657438e-01 5.85060060e-01
1.22666037e+00 -1.49500728e-01 -1.17833352e+00 -2.47860193e-01
1.52549103e-01 -1.30119532e-01 -3.30825779e-03 -7.89678454e-01
-4.41979796e-01 7.02466309e-01 7.52906740e-01 2.16731071e-01
1.06776774e+00 -6.28412887e-02 -9.36756507e-02 -2.38968804e-01
-6.79952949e-02 -9.52244699e-01 -4.84796137e-01 -3.43152761e-01
7.66411245e-01 -1.35590637e+00 1.00964129e-01 -3.89793396e-01
-2.91802645e-01 6.59580588e-01 1.47164315e-01 -6.62838817e-02
8.51974785e-01 2.39630058e-01 -4.94195670e-01 2.15833988e-02
-8.71892989e-01 3.28391403e-01 4.83814180e-01 6.96166635e-01
4.93445218e-01 5.23383737e-01 -1.04843462e+00 8.97227526e-01
3.60440224e-01 3.58427405e-01 5.48492849e-01 4.99095440e-01
7.32133240e-02 -7.05988050e-01 -7.60845721e-01 8.75131190e-01
-1.14257312e+00 -1.50379494e-01 4.31079179e-01 1.28162980e+00
-2.38748163e-01 8.65019441e-01 1.96673930e-01 8.75860095e-01
4.82992679e-01 8.55657384e-02 3.43874991e-01 -3.81381959e-01
8.72037467e-03 5.31840861e-01 9.10514221e-02 -1.52589515e-01
-5.60069621e-01 -7.51413643e-01 -9.34858084e-01 -3.38642687e-01
-6.43739104e-01 2.59092033e-01 1.70459941e-01 1.07150900e+00
2.11385727e-01 3.48039776e-01 8.43076169e-01 -1.13175757e-01
-1.16580808e+00 -9.60892081e-01 -8.21871996e-01 1.85533792e-01
7.13801563e-01 -8.11999798e-01 -6.45969331e-01 -3.20141576e-02] | [8.007328987121582, 5.268938064575195] |
fbffca20-895c-4fdd-9af6-07a9db3fa513 | a-deep-learning-based-bayesian-approach-to | 2001.04567 | null | https://arxiv.org/abs/2001.04567v2 | https://arxiv.org/pdf/2001.04567v2.pdf | A deep-learning based Bayesian approach to seismic imaging and uncertainty quantification | Uncertainty quantification is essential when dealing with ill-conditioned inverse problems due to the inherent nonuniqueness of the solution. Bayesian approaches allow us to determine how likely an estimation of the unknown parameters is via formulating the posterior distribution. Unfortunately, it is often not possible to formulate a prior distribution that precisely encodes our prior knowledge about the unknown. Furthermore, adherence to handcrafted priors may greatly bias the outcome of the Bayesian analysis. To address this issue, we propose to use the functional form of a randomly initialized convolutional neural network as an implicit structured prior, which is shown to promote natural images and excludes images with unnatural noise. In order to incorporate the model uncertainty into the final estimate, we sample the posterior distribution using stochastic gradient Langevin dynamics and perform Bayesian model averaging on the obtained samples. Our synthetic numerical experiment verifies that deep priors combined with Bayesian model averaging are able to partially circumvent imaging artifacts and reduce the risk of overfitting in the presence of extreme noise. Finally, we present pointwise variance of the estimates as a measure of uncertainty, which coincides with regions that are more difficult to image. | ['Felix J. Herrmann', 'Ali Siahkoohi', 'Gabrio Rizzuti'] | 2020-01-13 | null | null | null | null | ['seismic-imaging'] | ['miscellaneous'] | [ 3.97902459e-01 1.92772731e-01 4.37586039e-01 -3.37346107e-01
-6.43393576e-01 -2.91026086e-01 5.51327586e-01 -2.98623234e-01
-6.29231215e-01 1.03810740e+00 2.05927327e-01 1.50044352e-01
-4.77471918e-01 -6.18703902e-01 -7.44941533e-01 -9.72813368e-01
3.64669532e-01 3.91421437e-01 -5.84094860e-02 2.55726933e-01
2.84721822e-01 4.23859596e-01 -1.28435004e+00 -2.92153299e-01
1.07703114e+00 8.89621198e-01 3.17540973e-01 2.25717783e-01
4.01525311e-02 5.25296867e-01 -4.31783587e-01 -1.85243249e-01
1.35330975e-01 -3.01158041e-01 -4.79717761e-01 8.50019529e-02
1.17468379e-01 -5.13970256e-01 7.39622712e-02 1.40606809e+00
3.71676862e-01 2.12885246e-01 1.09717119e+00 -5.49669921e-01
-2.45617971e-01 3.85859847e-01 -5.19862711e-01 1.31823996e-04
2.25679073e-02 2.28311136e-01 5.64282238e-01 -7.03021884e-01
6.30369008e-01 1.01045656e+00 7.26899564e-01 1.38370246e-01
-1.63334990e+00 -2.07331493e-01 -9.68226716e-02 -1.15485989e-01
-1.65686393e+00 -5.26518881e-01 9.50410008e-01 -5.98906875e-01
1.10395963e-03 -5.84228858e-02 3.59301448e-01 1.21013951e+00
3.90818954e-01 2.66030520e-01 1.36810744e+00 -5.48628867e-01
5.16558766e-01 2.92396903e-01 1.42259866e-01 4.23444569e-01
4.89217192e-01 1.96416214e-01 -2.99683690e-01 -2.73574740e-01
8.79706442e-01 -3.83908808e-01 -6.50161445e-01 -6.08067155e-01
-1.01203716e+00 5.96938193e-01 2.94380933e-01 1.91466123e-01
-5.94117045e-01 2.40159675e-01 -7.28552938e-02 -2.39313722e-01
3.56085986e-01 4.67631608e-01 1.82082932e-02 6.89580441e-02
-9.80274975e-01 1.33240253e-01 7.30632424e-01 4.55624908e-01
7.55068600e-01 9.23166275e-02 -1.70191556e-01 6.04251206e-01
4.75766480e-01 6.60142899e-01 -2.84467898e-02 -1.34747779e+00
-1.98541299e-01 -8.21674466e-02 6.81081533e-01 -1.08758509e+00
-1.93691462e-01 -7.53284693e-01 -9.53978002e-01 5.58931231e-01
8.65346432e-01 -1.85236961e-01 -8.04565966e-01 1.78594899e+00
1.70416117e-01 1.21181473e-01 -3.67381461e-02 1.17609990e+00
2.64936864e-01 4.94314253e-01 -3.25556435e-02 -3.10265034e-01
1.06163168e+00 -1.24746457e-01 -8.93267632e-01 -1.59488380e-01
-9.49045271e-02 -6.26849592e-01 9.10807908e-01 4.46331680e-01
-8.00146818e-01 -2.05455706e-01 -1.09107006e+00 2.77081192e-01
9.29940343e-02 2.12027922e-01 2.03635857e-01 7.22886622e-01
-7.14406192e-01 8.42754304e-01 -9.21323955e-01 1.39881656e-01
2.23340869e-01 1.55713782e-01 -6.84735328e-02 -2.29836013e-02
-9.87435222e-01 1.19719958e+00 3.15908074e-01 4.89084393e-01
-7.41198421e-01 -6.82007670e-01 -6.72382534e-01 2.05214292e-01
4.50332493e-01 -7.35207379e-01 8.50299716e-01 -8.53038073e-01
-1.72871554e+00 3.25952470e-01 -6.61184490e-02 -3.33367586e-01
8.03269506e-01 -2.68768609e-01 7.19704628e-02 1.63139105e-01
-1.42930090e-01 4.89822745e-01 1.39522445e+00 -1.33595431e+00
3.10436100e-01 -1.62193090e-01 -9.30274501e-02 -8.31558630e-02
1.00173093e-01 -4.82500225e-01 -2.38804385e-01 -3.38863760e-01
3.71139675e-01 -7.87854433e-01 -3.56527299e-01 1.65872887e-01
-3.72974217e-01 5.33111215e-01 3.65993410e-01 -7.11712956e-01
8.25355947e-01 -2.15641141e+00 1.75135195e-01 5.08754432e-01
1.97460756e-01 -1.59016013e-01 1.67619735e-01 8.60676076e-03
1.72295988e-01 -8.78372043e-02 -7.90402889e-01 -1.06423192e-01
5.33156283e-02 2.62482613e-01 -2.40227908e-01 7.65320897e-01
2.18669638e-01 3.95430684e-01 -6.74782157e-01 -3.83472979e-01
4.41215992e-01 8.96363914e-01 -5.42931974e-01 3.50266881e-02
-3.06414694e-01 9.15072322e-01 -5.23708701e-01 -3.57781462e-02
9.15591002e-01 -3.26265603e-01 1.82020158e-01 -5.74418783e-01
-6.91995770e-02 3.85309197e-02 -1.58555067e+00 1.46886396e+00
-4.42851216e-01 5.19976258e-01 4.26561624e-01 -9.12155509e-01
7.58903980e-01 1.82731658e-01 3.46009254e-01 -2.69552082e-01
3.76011491e-01 1.60299450e-01 5.55925556e-02 -3.79047930e-01
3.50109994e-01 -3.83186102e-01 3.42107207e-01 4.34126765e-01
5.85660934e-02 -5.21193087e-01 -8.65218695e-03 5.90187591e-03
6.52078569e-01 3.97043765e-01 2.30740651e-01 -7.16642320e-01
4.06288415e-01 -4.31090951e-01 5.58418393e-01 8.77798378e-01
3.46255116e-02 1.07946682e+00 6.34214997e-01 -1.96244657e-01
-1.00580180e+00 -1.10747886e+00 -5.94652474e-01 -7.83994943e-02
1.72895044e-02 1.48827597e-01 -9.28532302e-01 -1.96125239e-01
-2.88783938e-01 8.79858196e-01 -4.76125628e-01 -3.57133821e-02
-1.85233697e-01 -1.06283057e+00 1.26692593e-01 -1.91166662e-02
5.70774496e-01 -5.48166215e-01 -7.95978427e-01 3.15074116e-01
-4.06134933e-01 -1.17082691e+00 -5.75182922e-02 6.93278462e-02
-7.31508672e-01 -8.80706489e-01 -9.03168499e-01 1.82223171e-01
7.88539827e-01 -3.66172314e-01 9.49619353e-01 -3.56034070e-01
-1.19963445e-01 4.80325669e-01 1.26723513e-01 -1.06061801e-01
-5.09427428e-01 -3.70152324e-01 1.58837307e-02 2.91693777e-01
-7.26332590e-02 -7.69100428e-01 -5.65414906e-01 1.38188213e-01
-9.43441570e-01 -6.01400770e-02 3.80070359e-01 8.77828240e-01
4.80402142e-01 1.42118603e-01 2.61723816e-01 -6.18221521e-01
6.29835665e-01 -2.04007626e-01 -1.04762948e+00 2.13099763e-01
-3.95150870e-01 6.21843576e-01 3.07716370e-01 -4.61443722e-01
-1.49051595e+00 2.22401217e-01 2.08523236e-02 -3.54539484e-01
-1.40034825e-01 5.56164503e-01 -2.43675467e-02 -1.80864587e-01
7.22359419e-01 -1.84045266e-02 2.83420801e-01 -4.43935424e-01
1.59666836e-01 2.89148688e-01 4.61559325e-01 -9.19923246e-01
4.33953851e-01 6.17905080e-01 3.18954080e-01 -1.07366943e+00
-8.15280795e-01 7.06049204e-02 -5.04024565e-01 -3.52494329e-01
8.54414225e-01 -6.67810738e-01 -7.46704876e-01 4.15617377e-01
-1.21751690e+00 -2.04407707e-01 -2.74127454e-01 8.49324942e-01
-6.87553942e-01 5.98602891e-01 -2.49973953e-01 -1.03256583e+00
-1.12331780e-02 -1.42558944e+00 6.19255543e-01 1.49068683e-01
-3.31203282e-01 -9.05412555e-01 -2.02995971e-01 1.21008148e-02
6.42140508e-01 2.34931633e-01 7.55314112e-01 7.10773095e-02
-6.81240261e-01 -1.10067144e-01 -2.94974804e-01 4.97861326e-01
-1.60947934e-01 8.96132737e-02 -1.22531509e+00 3.82621624e-02
4.76881951e-01 3.46011035e-02 8.48864675e-01 9.73194003e-01
9.85249221e-01 2.45259143e-02 2.98187714e-02 3.95223320e-01
1.49715805e+00 -1.26188397e-01 8.46546233e-01 -1.46216318e-01
2.64884472e-01 7.00385153e-01 1.16939597e-01 5.48838496e-01
-1.25707284e-01 5.84966302e-01 4.93595153e-01 3.78787935e-01
3.28170741e-03 3.81717794e-02 5.85168712e-02 3.66123140e-01
-1.32331491e-01 -1.42675638e-01 -9.27740276e-01 3.10293525e-01
-1.46343982e+00 -7.54704833e-01 -6.68816939e-02 2.34892225e+00
7.21029580e-01 2.42545336e-01 -5.04667103e-01 -6.42721951e-02
7.21201599e-01 -8.36877078e-02 -3.41155350e-01 2.02283084e-01
5.41980192e-03 -1.31298583e-02 5.28263152e-01 8.53566468e-01
-9.32267070e-01 4.37367171e-01 6.58646059e+00 8.40675473e-01
-1.09487247e+00 4.14332673e-02 5.93890429e-01 1.35301098e-01
-5.14852703e-01 1.56756908e-01 -4.15596008e-01 6.64904058e-01
6.59180224e-01 1.22159280e-01 4.37044650e-01 3.77702594e-01
3.67758811e-01 -6.85348511e-01 -8.45023632e-01 8.26083481e-01
-2.03385070e-01 -1.13521588e+00 -1.54210284e-01 1.65547967e-01
6.47582591e-01 -1.42172262e-01 2.72596255e-02 -3.12728465e-01
3.08495939e-01 -9.97942388e-01 8.55474174e-01 1.21674383e+00
5.00362992e-01 -5.84614217e-01 7.56801844e-01 4.98389930e-01
-3.84838194e-01 2.85705060e-01 -4.27646697e-01 -1.45971075e-01
4.55629826e-01 1.27769732e+00 -5.37268102e-01 2.14024410e-01
5.47977507e-01 2.93019742e-01 -1.04360484e-01 1.04858065e+00
-2.72910923e-01 4.90905821e-01 -9.06723022e-01 8.49317312e-02
-3.21930461e-02 -7.57124245e-01 8.64251375e-01 6.91839278e-01
5.83827674e-01 2.36882225e-01 -2.32211351e-01 1.65039682e+00
2.63646662e-01 -2.53900141e-01 -5.36845863e-01 -1.35891885e-02
2.13388756e-01 9.90316868e-01 -9.38686609e-01 -1.80129781e-02
-1.26516774e-01 6.07793093e-01 8.32685456e-03 8.40091527e-01
-5.51325679e-01 -1.13689471e-02 2.73995847e-01 4.67235297e-02
1.72490031e-01 -3.40755969e-01 -4.55323577e-01 -1.16431689e+00
5.64585216e-02 -5.89414179e-01 -1.39755443e-01 -1.00643146e+00
-1.06333768e+00 3.71740639e-01 2.34470576e-01 -9.88460004e-01
-4.20989364e-01 -6.11564040e-01 -3.78268719e-01 1.08034456e+00
-1.20792401e+00 -5.56986153e-01 -1.19244248e-01 1.24429412e-01
-6.55394942e-02 2.39668474e-01 7.15635896e-01 7.96020702e-02
-3.73217136e-01 8.15443322e-03 2.59724557e-01 -1.39321402e-01
5.72751045e-01 -9.41897690e-01 -4.73070234e-01 9.58861887e-01
-2.70436883e-01 7.89188683e-01 1.25629628e+00 -5.85290730e-01
-1.08029532e+00 -4.45809990e-01 2.50068575e-01 -2.19670311e-01
5.16348064e-01 -1.03819355e-01 -9.95676816e-01 4.16465133e-01
6.31291594e-04 1.65843666e-01 1.01176441e-01 -1.04251452e-01
-5.50495042e-03 -6.74107596e-02 -1.32053697e+00 5.19412994e-01
3.43838841e-01 -4.76530164e-01 -3.65377635e-01 -1.00545399e-01
3.18830818e-01 -2.27553010e-01 -8.28532040e-01 5.44129252e-01
6.19497776e-01 -1.12822032e+00 9.40441906e-01 1.52175531e-01
3.48660409e-01 -3.91049713e-01 -2.51183152e-01 -1.31301880e+00
-3.17959823e-02 -4.28835273e-01 5.64792529e-02 1.13653588e+00
2.84001380e-01 -6.31988704e-01 6.10795736e-01 9.49117899e-01
1.95136368e-01 -1.97955370e-01 -1.03860748e+00 -6.77843809e-01
9.57680792e-02 -6.93833649e-01 2.87458569e-01 6.56100869e-01
-4.81752485e-01 -7.51579106e-02 -3.93157929e-01 4.47794169e-01
1.16892624e+00 -1.25596866e-01 3.85947496e-01 -1.26585639e+00
-3.53597045e-01 -3.04107696e-01 -2.10230902e-01 -1.03179395e+00
1.34955883e-01 -3.28556210e-01 3.81175578e-01 -1.14560878e+00
1.79538205e-01 -4.06696796e-01 -7.15142265e-02 -1.60448655e-01
1.70845821e-01 8.04539993e-02 -7.16644153e-02 9.80029628e-03
-2.83933636e-02 7.18169689e-01 1.40229261e+00 -7.06088915e-02
9.51160863e-02 3.70572656e-02 -2.47510716e-01 1.06946790e+00
5.40477335e-01 -5.91162324e-01 -3.78822803e-01 -3.52955043e-01
4.40390587e-01 1.55806392e-01 6.88478529e-01 -1.03440440e+00
1.74587935e-01 -1.00119725e-01 4.60283637e-01 -3.98788542e-01
5.02592385e-01 -1.03138399e+00 5.55927813e-01 2.54291892e-01
-2.52547234e-01 -7.80969024e-01 6.03313893e-02 5.56036055e-01
-1.16570324e-01 -9.33880210e-01 9.77612078e-01 -1.59536496e-01
-3.96544784e-01 -1.09657891e-01 -5.55112481e-01 -8.11040550e-02
4.09269869e-01 2.99108345e-02 1.56432450e-01 -6.64776504e-01
-9.62247252e-01 -1.10187434e-01 4.92165864e-01 -2.86368579e-01
4.56210166e-01 -9.33355629e-01 -5.57035387e-01 2.32089981e-01
-3.19144845e-01 -3.78067009e-02 3.47047627e-01 1.03422260e+00
-5.82971513e-01 1.15438133e-01 -4.69500497e-02 -8.35783839e-01
-6.23811483e-01 2.40787312e-01 6.69026494e-01 -7.97506645e-02
-6.02610290e-01 4.97819573e-01 5.79832830e-02 -3.45604777e-01
2.41617531e-01 -3.62923831e-01 -7.27193132e-02 -5.31772673e-02
3.56280416e-01 3.34693283e-01 -1.14981227e-01 -3.97237957e-01
-2.11737350e-01 6.15856111e-01 2.09843099e-01 -4.58557546e-01
1.21324587e+00 -4.00766969e-01 -1.61502838e-01 3.85696620e-01
8.85582387e-01 5.70065007e-02 -1.82386303e+00 -1.64690018e-01
-1.04484729e-01 -2.77943075e-01 4.31831211e-01 -6.82571948e-01
-8.47702563e-01 9.53906894e-01 6.22173607e-01 -6.01934791e-02
8.58329713e-01 -3.62314045e-01 1.14705645e-01 4.42670763e-01
3.59358221e-01 -1.04548132e+00 -2.36885384e-01 4.68636096e-01
9.62431133e-01 -1.22100532e+00 2.80869007e-01 -4.04896498e-01
-2.64403194e-01 1.17980385e+00 1.46544471e-01 -1.48444310e-01
9.69394326e-01 4.17936713e-01 -1.85169447e-02 -7.42051750e-02
-9.22873914e-02 -2.06716452e-02 2.94097841e-01 4.94780868e-01
1.55355126e-01 -1.17991678e-01 -3.52753192e-01 4.27396685e-01
8.75891093e-03 2.75265098e-01 7.42478371e-01 6.00473881e-01
-2.56084532e-01 -6.67217076e-01 -7.34937072e-01 1.21525086e-01
-4.93869841e-01 -2.90145036e-02 1.92544907e-01 5.67211449e-01
-6.90098330e-02 6.05686903e-01 -4.69905604e-03 2.64363319e-01
-1.52435526e-01 5.57786040e-02 7.81839609e-01 -1.83541700e-01
2.40051135e-01 3.89130265e-01 -9.82651766e-03 -3.23962480e-01
-4.82875586e-01 -5.61930656e-01 -1.05371821e+00 -8.49253684e-02
-3.83987963e-01 1.07012488e-01 7.97723234e-01 1.09854567e+00
8.10434073e-02 5.11807561e-01 -5.57497069e-02 -1.06937003e+00
-8.28926563e-01 -9.47601438e-01 -6.14401340e-01 1.71443492e-01
3.80196989e-01 -8.30673933e-01 -6.90736115e-01 -3.52518843e-03] | [6.840970993041992, 3.548114776611328] |
6c284a1d-979f-4966-8a8c-fad9b506910b | pose-randomization-for-weakly-paired-image | 2011.00301 | null | https://arxiv.org/abs/2011.00301v2 | https://arxiv.org/pdf/2011.00301v2.pdf | PREGAN: Pose Randomization and Estimation for Weakly Paired Image Style Translation | Utilizing the trained model under different conditions without data annotation is attractive for robot applications. Towards this goal, one class of methods is to translate the image style from another environment to the one on which models are trained. In this paper, we propose a weakly-paired setting for the style translation, where the content in the two images is aligned with errors in poses. These images could be acquired by different sensors in different conditions that share an overlapping region, e.g. with LiDAR or stereo cameras, from sunny days or foggy nights. We consider this setting to be more practical with: (i) easier labeling than the paired data; (ii) better interpretability and detail retrieval than the unpaired data. To translate across such images, we propose PREGAN to train a style translator by intentionally transforming the two images with a random pose, and to estimate the given random pose by differentiable non-trainable pose estimator given that the more aligned in style, the better the estimated result is. Such adversarial training enforces the network to learn the style translation, avoiding being entangled with other variations. Finally, PREGAN is validated on both simulated and real-world collected data to show the effectiveness. Results on down-stream tasks, classification, road segmentation, object detection, and feature matching show its potential for real applications. https://github.com/wrld/PRoGAN | ['Rong Xiong', 'Yue Wang', 'Yunkai Wang', 'Xuecheng Xu', 'Jiaxin Guo', 'Zexi Chen'] | 2020-10-31 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 4.47333694e-01 3.55975956e-01 3.04785520e-02 -6.02123976e-01
-6.61144078e-01 -8.46795380e-01 5.66986144e-01 -6.93962932e-01
-3.33898038e-01 7.31086969e-01 -2.88056999e-01 -2.92192977e-02
2.74370164e-01 -7.93046832e-01 -1.33540213e+00 -8.01356256e-01
4.28544641e-01 8.11308742e-01 -4.58979085e-02 -1.67887002e-01
-1.37643278e-01 4.63518500e-01 -1.47840190e+00 -2.03997225e-01
8.78087461e-01 8.24235797e-01 5.70006847e-01 6.33936167e-01
2.17810124e-01 3.24807942e-01 -5.45361400e-01 -4.63122189e-01
7.90538609e-01 -2.56783366e-01 -5.45365453e-01 4.37756658e-01
7.30413973e-01 -3.99398953e-01 -3.90463531e-01 1.34996545e+00
1.91033646e-01 -3.77595536e-02 4.32947308e-01 -1.36040163e+00
-5.74629784e-01 1.32758915e-01 -5.63911021e-01 -5.82814634e-01
3.89780611e-01 3.56385350e-01 4.81435061e-01 -6.75266623e-01
7.35561967e-01 1.37936640e+00 5.91995656e-01 5.94478369e-01
-1.28586650e+00 -6.89932942e-01 -4.33138385e-02 6.04059659e-02
-1.36957049e+00 -3.62110615e-01 7.82597184e-01 -2.34819546e-01
1.47454321e-01 1.49755791e-01 5.91927409e-01 1.49489474e+00
1.71361715e-01 5.25647402e-01 1.38749123e+00 -2.59803921e-01
1.63875237e-01 3.21473688e-01 -5.24089456e-01 5.96455634e-01
2.29719937e-01 2.79810727e-01 -4.72608656e-01 2.36399859e-01
5.75071335e-01 8.70541781e-02 -5.47496974e-01 -5.40406108e-01
-1.22782075e+00 5.84955156e-01 7.53845513e-01 -2.20834672e-01
-2.12071627e-01 2.03986451e-01 1.82242338e-02 4.12240982e-01
4.55235124e-01 3.94437551e-01 -4.23200846e-01 2.56276459e-01
-6.04971111e-01 1.69672221e-01 6.35479152e-01 1.40377212e+00
1.10671508e+00 7.01554120e-02 3.05137843e-01 5.91175735e-01
4.97599870e-01 1.28393388e+00 3.30233753e-01 -1.14043880e+00
5.60849190e-01 1.49583414e-01 1.79175526e-01 -8.22097242e-01
-2.93120116e-01 -3.31792563e-01 -7.39491522e-01 4.88017380e-01
6.07902527e-01 -1.44236475e-01 -1.23724735e+00 1.86232042e+00
4.12875503e-01 2.82667667e-01 1.55556381e-01 1.18456256e+00
4.11396950e-01 7.08277822e-01 -3.63431901e-01 2.65544593e-01
1.29433894e+00 -9.58804488e-01 -6.07314825e-01 -8.10826778e-01
3.37209076e-01 -9.68860924e-01 1.18768096e+00 3.58304679e-01
-6.55958772e-01 -6.26257479e-01 -1.19179785e+00 -1.38027132e-01
-3.40551525e-01 2.83865601e-01 9.94547009e-02 4.65127915e-01
-8.53262484e-01 5.27521253e-01 -7.91780174e-01 -5.22463918e-01
1.66729271e-01 1.48525938e-01 -4.69971657e-01 -2.67274171e-01
-1.19158363e+00 1.10963500e+00 2.89802164e-01 2.80619502e-01
-8.96820486e-01 -4.87793386e-01 -1.03174758e+00 -4.14817512e-01
4.08174187e-01 -7.88789093e-01 1.11384141e+00 -1.37846601e+00
-1.71521258e+00 9.69577610e-01 3.89339146e-03 -4.48173583e-01
8.36881101e-01 -4.77074206e-01 -1.49352565e-01 -3.62507044e-03
3.51399571e-01 1.08062661e+00 1.22364020e+00 -1.59703445e+00
-4.84044313e-01 -5.62368572e-01 9.23214704e-02 5.04186034e-01
-5.09760007e-02 -5.42199492e-01 -5.45754015e-01 -4.51966465e-01
2.17923358e-01 -1.52913809e+00 -1.43176585e-01 3.79172921e-01
-5.97325206e-01 5.07902205e-01 1.27408099e+00 -6.84621632e-01
5.66340163e-02 -2.03712773e+00 3.20868820e-01 1.09850541e-01
-1.12342633e-01 1.45515874e-01 -2.72780597e-01 7.75528923e-02
1.13760427e-01 -1.51865751e-01 -4.60057765e-01 -6.84170842e-01
-1.60816938e-01 6.47885084e-01 -5.40787160e-01 7.88598299e-01
2.64358670e-01 8.54734898e-01 -9.89238024e-01 -2.23512173e-01
4.18122023e-01 4.43434000e-01 -3.04534793e-01 4.47393388e-01
-2.53737211e-01 1.19336092e+00 -2.59655535e-01 4.84133542e-01
8.96613240e-01 1.48385733e-01 1.09567128e-01 -5.12308002e-01
8.84057879e-02 1.12252638e-01 -1.07633805e+00 1.97905397e+00
-7.09456027e-01 8.64186704e-01 2.01822355e-01 -7.82218993e-01
8.77928615e-01 3.73699665e-02 1.60231426e-01 -6.53686404e-01
2.19225511e-01 2.47791454e-01 -1.59507185e-01 -4.86532629e-01
5.22497296e-01 -1.33752972e-01 -2.53415912e-01 8.57529268e-02
1.94226913e-02 -8.84116888e-01 -1.16991416e-01 -1.53851226e-01
6.20818317e-01 5.62482357e-01 4.03093770e-02 -5.21482341e-02
2.59880066e-01 -9.93981659e-02 6.96152687e-01 5.13329148e-01
-1.24432683e-01 1.00911212e+00 1.15694730e-02 -2.55066186e-01
-1.35052359e+00 -1.06161106e+00 -7.09736347e-02 4.41298217e-01
5.60495973e-01 2.73589224e-01 -8.16247106e-01 -7.00509965e-01
-4.15451415e-02 6.01427555e-01 -5.27456880e-01 -1.85338438e-01
-6.92945480e-01 -5.49249165e-02 6.68008566e-01 3.01373750e-01
9.02752638e-01 -7.71780550e-01 -6.07286036e-01 -2.04346359e-01
-3.77892226e-01 -1.49690294e+00 -6.27698839e-01 1.85132280e-01
-5.28967142e-01 -9.02945280e-01 -6.50173903e-01 -6.66952610e-01
7.59463668e-01 3.22835773e-01 9.38333869e-01 -2.55295485e-01
1.66682735e-01 3.80048066e-01 -3.72911841e-01 -4.99165595e-01
-5.24030745e-01 1.01685792e-01 2.12443069e-01 2.38920242e-01
-1.81406796e-01 -5.32917440e-01 -4.24132913e-01 5.08030772e-01
-1.12693954e+00 2.82638520e-01 6.65026665e-01 8.92462730e-01
5.96513331e-01 -2.95244575e-01 8.67434517e-02 -8.11834753e-01
-7.19902962e-02 -2.67017305e-01 -7.18346477e-01 7.17430264e-02
-4.83762026e-01 9.16715562e-02 7.13720918e-01 -4.61957216e-01
-1.15374053e+00 4.91095155e-01 4.57431562e-03 -7.65661240e-01
-2.75114298e-01 1.61075398e-01 -5.74291170e-01 -1.53443575e-01
7.37382472e-01 6.07898757e-02 2.20242038e-01 -1.71197593e-01
5.94882786e-01 6.06608927e-01 8.65305185e-01 -5.69135964e-01
1.44615078e+00 7.99697578e-01 1.00576900e-01 -8.99992764e-01
-8.00122619e-01 -1.65176794e-01 -6.65744960e-01 -1.62155986e-01
1.01475930e+00 -1.02832890e+00 -2.26084262e-01 7.09762037e-01
-1.32920718e+00 -5.61364412e-01 -2.04549208e-01 4.85063136e-01
-6.79676890e-01 2.61450410e-01 -2.34467730e-01 -5.60394526e-01
-1.48082033e-01 -1.29698312e+00 1.40426481e+00 4.02472496e-01
-7.07094148e-02 -9.10710752e-01 -1.86495945e-01 4.51069564e-01
1.56631231e-01 3.28832209e-01 2.79167831e-01 -1.05305873e-01
-6.60177290e-01 -5.75699620e-02 -1.39015630e-01 5.31426311e-01
4.02306825e-01 -2.27633566e-02 -1.17777967e+00 -3.36862922e-01
8.49442035e-02 -4.38056558e-01 5.01355588e-01 6.03518672e-02
9.06844318e-01 -2.30054110e-01 -1.93078086e-01 9.86557424e-01
1.23806500e+00 1.24342822e-01 7.39612579e-01 4.70810473e-01
1.12196195e+00 7.70727158e-01 9.13023949e-01 -6.74173888e-03
3.25492442e-01 9.78957176e-01 8.33922267e-01 -2.09809422e-01
-2.64643669e-01 -3.84062946e-01 5.93604207e-01 4.15957093e-01
2.43577227e-01 -3.93952817e-01 -7.47126698e-01 4.20937121e-01
-1.72144353e+00 -5.91839492e-01 -7.44363293e-02 2.19190049e+00
6.83798850e-01 5.82153350e-02 -3.30912799e-01 -2.29395911e-01
7.86112189e-01 1.61320210e-01 -7.68614113e-01 -6.16331510e-02
-2.50556707e-01 4.58048098e-02 9.86777961e-01 5.60738862e-01
-1.02246475e+00 9.59666729e-01 4.38729286e+00 5.52905977e-01
-1.48472524e+00 1.41394094e-01 6.25216961e-01 1.52732819e-01
-2.79817790e-01 1.28762454e-01 -5.51065981e-01 5.41342080e-01
7.36260712e-01 2.17169344e-01 5.78562677e-01 6.96804285e-01
1.98829174e-01 -5.18129729e-02 -1.13836074e+00 8.99441123e-01
1.07216373e-01 -1.03023398e+00 -1.42105103e-01 3.81580219e-02
8.44950080e-01 2.73797095e-01 2.54693568e-01 4.93571199e-02
2.76592523e-01 -7.76528060e-01 1.30131507e+00 5.00921130e-01
1.00676703e+00 -4.23800677e-01 6.17375374e-01 3.93025011e-01
-9.60498512e-01 2.12398961e-01 -1.70296684e-01 9.55206528e-02
3.13364774e-01 5.54819465e-01 -9.10445511e-01 7.93683648e-01
7.27930188e-01 7.73300469e-01 -5.28946340e-01 6.86373889e-01
-7.10747778e-01 3.88089478e-01 -5.01524448e-01 3.29852402e-01
-4.79683727e-02 -6.09616697e-01 7.43543506e-01 7.93121815e-01
5.07202029e-01 -3.86103153e-01 2.29536116e-01 9.57492471e-01
-1.35111466e-01 -5.44944227e-01 -1.03439534e+00 3.88288349e-01
5.69165111e-01 1.29756761e+00 -5.80537677e-01 -2.22138494e-01
-1.71157762e-01 1.33852828e+00 1.16529897e-01 4.01313186e-01
-1.14341462e+00 -1.83251321e-01 8.46750855e-01 5.89899831e-02
1.56058911e-02 -3.44035983e-01 -5.22596896e-01 -1.22329783e+00
3.02617669e-01 -7.23767877e-01 -1.98439494e-01 -1.16362560e+00
-1.13532150e+00 6.76620722e-01 -2.85805878e-03 -1.66613364e+00
-2.31281787e-01 -5.95985711e-01 -4.44752395e-01 9.33072507e-01
-1.53905475e+00 -1.49769270e+00 -6.22340858e-01 2.67896801e-01
6.03026271e-01 2.08248168e-01 5.27609050e-01 4.25772816e-01
-4.91265476e-01 3.85629207e-01 2.03454494e-02 1.09613195e-01
1.08256185e+00 -1.22709870e+00 4.49672818e-01 1.02871537e+00
1.06713772e-01 1.90473288e-01 1.06434214e+00 -5.25784850e-01
-1.42139852e+00 -1.61220348e+00 2.54038036e-01 -4.43731993e-01
3.49361569e-01 -3.87474030e-01 -7.25169063e-01 8.29840124e-01
1.93183988e-01 1.95389122e-01 -1.51725143e-01 -6.02549911e-01
-3.18086326e-01 -3.66168797e-01 -1.28879321e+00 6.67713344e-01
1.05301392e+00 -5.37229061e-01 -3.01201195e-01 4.37218517e-01
8.54167044e-01 -1.02663708e+00 -5.30225217e-01 4.63516116e-01
5.97624898e-01 -7.74760246e-01 8.52520525e-01 -1.26937956e-01
4.24903452e-01 -6.66618466e-01 -3.37808788e-01 -1.72468185e+00
1.41500875e-01 -5.63790560e-01 3.51734579e-01 1.26711047e+00
3.45720619e-01 -9.43507850e-01 8.18406105e-01 5.25513172e-01
-3.79530042e-01 -4.40881938e-01 -8.44939530e-01 -9.90353644e-01
3.79258543e-02 -3.25911522e-01 7.05090404e-01 7.94539452e-01
-7.84748077e-01 4.24236506e-01 -6.47702634e-01 5.79311609e-01
6.49088621e-01 1.27506092e-01 1.27019560e+00 -8.47804904e-01
-3.65021437e-01 6.84407949e-02 -4.78869975e-01 -1.24421883e+00
4.57025886e-01 -6.29802108e-01 4.81323242e-01 -1.12016344e+00
-2.90575027e-01 -6.05765998e-01 4.51936632e-01 4.24145699e-01
5.16009517e-02 5.26338220e-01 3.04803997e-01 2.76211858e-01
-1.53376326e-01 7.22933888e-01 1.33724713e+00 -3.91044348e-01
8.00822452e-02 1.27875824e-02 -2.85250485e-01 7.96616256e-01
9.49502409e-01 -5.74981272e-01 -4.69596744e-01 -7.79527426e-01
1.11064032e-01 -3.49505246e-02 4.49265540e-01 -1.06722963e+00
-4.32645753e-02 -1.23261184e-01 2.61062205e-01 -2.54635155e-01
5.72362959e-01 -1.08034790e+00 4.12966222e-01 4.41286594e-01
-9.55329984e-02 -2.76691169e-02 -3.76743563e-02 5.46302378e-01
-5.40574789e-02 -2.05607489e-01 9.58045304e-01 8.25291649e-02
-6.37312233e-01 3.05853665e-01 -4.82410099e-03 -1.38994113e-01
1.00005746e+00 -2.41980001e-01 -3.40422630e-01 -5.89511096e-01
-4.46447939e-01 1.73995599e-01 8.85684669e-01 5.30182958e-01
5.12117982e-01 -1.40435481e+00 -5.14586329e-01 3.37207943e-01
2.89843857e-01 6.57873869e-01 -3.53356376e-02 7.55599320e-01
-7.23103702e-01 1.19442053e-01 -2.37330317e-01 -9.47620213e-01
-1.08753514e+00 3.17409694e-01 4.92492884e-01 8.24075367e-04
-5.60873866e-01 5.15194416e-01 3.07554394e-01 -9.94931936e-01
5.51214442e-03 -3.54192585e-01 3.15590441e-01 -2.34888420e-01
1.65922523e-01 2.09585354e-01 2.15846241e-01 -9.15004671e-01
-1.14316702e-01 6.66265547e-01 2.49021992e-01 -3.03635776e-01
9.64235425e-01 -4.70264822e-01 2.11474244e-02 4.45303798e-01
1.34981620e+00 2.25052945e-02 -1.84823239e+00 -4.10154253e-01
-2.72494793e-01 -5.95953822e-01 -1.87572375e-01 -4.37841862e-01
-1.32820559e+00 6.75317228e-01 8.32986116e-01 -4.28577960e-02
9.99303281e-01 -5.85300140e-02 8.25235724e-01 3.84749383e-01
6.25364363e-01 -9.27292645e-01 3.83730680e-02 5.58649123e-01
9.77979958e-01 -1.37215793e+00 -2.34213546e-01 -4.72282946e-01
-6.00577831e-01 1.14285171e+00 7.36760557e-01 -1.96103841e-01
3.32599640e-01 1.06835984e-01 4.70657766e-01 1.95980836e-02
-8.26335400e-02 -2.45165601e-01 1.33062094e-01 7.80978262e-01
-1.18541986e-01 3.01573634e-01 2.68992126e-01 -1.70194600e-02
-4.94398624e-01 -3.42150301e-01 7.11232901e-01 8.27845097e-01
-7.64888152e-02 -1.04838073e+00 -7.57948756e-01 1.46619782e-01
6.87058643e-02 2.48821199e-01 -3.62468570e-01 7.08417058e-01
3.39959741e-01 9.26505089e-01 8.38776752e-02 -5.00499010e-01
3.90409559e-01 -2.26489693e-01 5.99944890e-01 -5.76823056e-01
-3.25682311e-04 -1.38104662e-01 4.12337743e-02 -6.71601295e-01
-4.97946292e-01 -7.14143574e-01 -1.25161147e+00 -1.37948900e-01
-2.48439312e-01 -1.91298112e-01 9.33515728e-01 1.03187597e+00
2.30720386e-01 4.17013228e-01 8.97151470e-01 -1.45842028e+00
-5.69646299e-01 -9.41948056e-01 -4.35811639e-01 5.05805135e-01
4.63216871e-01 -7.47415245e-01 -4.05383557e-01 2.60832816e-01] | [8.389904022216797, -2.2223429679870605] |
ac2251b7-e130-4499-b548-22b2f6b99aac | fine-tuning-self-organizing-maps-for-sentinel | null | null | https://www.mdpi.com/2072-4292/12/12/1923 | https://www.mdpi.com/2072-4292/12/12/1923/htm | Fine-Tuning Self-Organizing Maps for Sentinel-2 Imagery: Separating Clouds from Bright Surfaces | Removal of cloud interference is a crucial step for the exploitation of the spectral information stored in optical satellite images. Several cloud masking approaches have been developed through time, based on direct interpretation of the spectral and temporal properties of clouds through thresholds. The problem has also been tackled by machine learning methods with artificial neural networks being among the most recent ones. Detection of bright non-cloud objects is one of the most difficult tasks in cloud masking applications since spectral information alone often proves inadequate for their separation from clouds. Scientific attention has recently been redrawn on self-organizing maps (SOMs) because of their unique ability to preserve topologic relations, added to the advantage of faster training time and more interpretative behavior compared to other types of artificial neural networks. This study evaluated a SOM for cloud masking Sentinel-2 images and proposed a fine-tuning methodology to separate clouds from bright land areas. The fine-tuning process which is based on the output of the non-fine-tuned network, at first directly locates the neurons that correspond to the misclassified pixels. Then, the incorrect labels of the neurons are altered without applying further training. The fine-tuning method follows a general procedure, thus its applicability is broad and not confined only in the field of cloud-masking. The network was trained on the largest publicly available spectral database for Sentinel-2 cloud masking applications and was tested on a truly independent database of Sentinel-2 cloud masks. It was evaluated both qualitatively and quantitatively with the interpretation of its behavior through multiple visualization techniques being a main part of the evaluation. It was shown that the fine-tuned SOM successfully recognized the bright non-cloud areas and outperformed the state-of-the-art algorithms: Sen2Cor and Fmask, as well as the version that was not fine-tuned. | ['Vassilia Karathanassi', 'Viktoria Kristollari'] | 2020-06-14 | null | null | null | null | ['cloud-detection'] | ['computer-vision'] | [ 3.25951606e-01 -3.81224394e-01 4.54227060e-01 -6.17412962e-02
-1.49581820e-01 -8.43787253e-01 5.61448216e-01 3.22118253e-01
-5.78711510e-01 6.00605607e-01 -3.74373555e-01 -4.53192145e-01
-4.84340549e-01 -9.48167741e-01 -2.03758955e-01 -1.19790196e+00
-2.39885956e-01 5.98152876e-01 4.55911219e-01 -2.28378251e-01
2.66108423e-01 1.08625889e+00 -2.17394781e+00 5.57585895e-01
1.04893041e+00 7.78084755e-01 4.47367013e-01 4.36297983e-01
-2.59723395e-01 2.95059532e-01 -6.19410992e-01 2.72678345e-01
3.21536362e-01 -3.09634805e-01 -2.99214363e-01 1.48587942e-01
3.95593971e-01 2.97623426e-01 3.61054450e-01 1.16731048e+00
2.40932778e-01 -2.96313558e-02 7.16664612e-01 -8.50999892e-01
-2.04708368e-01 2.79701091e-02 -2.69041866e-01 4.12964344e-01
-3.14856261e-01 1.77629694e-01 4.81079936e-01 -1.03901136e+00
2.81803340e-01 7.71225631e-01 5.51371098e-01 2.12807685e-01
-1.39150274e+00 -7.20339417e-01 -1.59651697e-01 3.59924495e-01
-1.51669431e+00 2.42155381e-02 4.33460891e-01 -1.07259285e+00
8.45902264e-01 7.02667534e-01 8.65912020e-01 2.30364785e-01
-1.18674010e-01 -7.38805830e-02 1.83423507e+00 -5.65015495e-01
4.22587603e-01 5.97080290e-01 1.92891121e-01 5.45964949e-02
4.99637991e-01 4.80497569e-01 -1.47073865e-01 -7.46187195e-02
2.55758345e-01 -7.11107801e-04 -2.35142410e-01 -2.78277367e-01
-7.36707509e-01 7.44321942e-01 5.82646072e-01 7.52269626e-01
-4.89444673e-01 -4.25937891e-01 8.81562456e-02 1.31229699e-01
7.24040329e-01 5.49067140e-01 -1.83098078e-01 4.34065372e-01
-1.14425397e+00 1.97723478e-01 3.36845636e-01 2.62044352e-02
9.14090633e-01 3.58518869e-01 -1.38071343e-01 5.20211339e-01
5.95663078e-02 8.23962808e-01 2.93196529e-01 -3.26266915e-01
6.01103082e-02 8.67601931e-01 1.99271023e-01 -1.07567263e+00
-3.61429840e-01 -6.28608882e-01 -8.13717246e-01 1.05259657e+00
1.93809777e-01 1.22806676e-01 -1.14159989e+00 1.06123888e+00
2.34254003e-01 -9.48273242e-02 5.36504649e-02 9.60264742e-01
6.04366064e-01 5.89028418e-01 2.42371317e-02 -2.28539437e-01
1.23848355e+00 -3.91717911e-01 -4.56194788e-01 -1.19394280e-01
1.50576350e-03 -6.00604653e-01 7.99172163e-01 3.05762351e-01
-4.34769899e-01 -6.42058074e-01 -1.06253695e+00 7.88982570e-01
-1.10386848e+00 -3.50193940e-02 3.96174729e-01 8.43916833e-01
-1.08444631e+00 5.86506307e-01 -6.23416901e-01 -4.08656389e-01
1.33156255e-01 3.35668147e-01 -2.62820452e-01 4.30525720e-01
-1.06217480e+00 1.02022994e+00 6.79102659e-01 6.18667424e-01
-6.21247709e-01 -3.32761347e-01 -3.74698013e-01 2.62060910e-01
5.88133521e-02 -1.92943066e-01 5.30665755e-01 -1.33287239e+00
-1.09585965e+00 1.04003894e+00 -1.28096685e-01 -5.55921137e-01
3.62395734e-01 3.01141948e-01 -7.25858092e-01 3.83942544e-01
2.30112094e-02 2.98295170e-01 1.02307415e+00 -1.73526764e+00
-6.66287065e-01 -4.68875021e-01 -2.39444748e-01 6.49203435e-02
-4.17624384e-01 1.09092332e-01 2.32363328e-01 -4.95375156e-01
2.68555999e-01 -9.31138337e-01 -4.88403440e-02 -4.15289313e-01
3.05951107e-02 2.30838716e-01 1.03018105e+00 -5.10295808e-01
1.08269930e+00 -2.19039345e+00 -1.53803468e-01 6.04905009e-01
1.21375345e-01 8.66373062e-01 1.13748543e-01 5.01860499e-01
-4.15965885e-01 1.88448593e-01 -6.41898632e-01 1.09056227e-01
-3.19459498e-01 1.72325835e-01 -3.69160056e-01 3.85717869e-01
3.10908616e-01 1.49325550e-01 -7.41047740e-01 -1.72955036e-01
4.89229828e-01 4.96348083e-01 4.96067069e-02 4.78558205e-02
-1.87598839e-01 5.13356149e-01 -4.32448089e-02 4.86734539e-01
1.14620793e+00 2.01126903e-01 1.52114198e-01 7.13443309e-02
-8.46416414e-01 -2.08403677e-01 -1.11312747e+00 6.11101747e-01
-1.16730042e-01 7.65220881e-01 8.47771764e-02 -7.12179601e-01
1.10566831e+00 4.91280705e-01 2.34711066e-01 -7.22953439e-01
-2.64913410e-01 4.77559924e-01 1.90641675e-02 -7.45288312e-01
3.79598796e-01 -4.31998163e-01 6.73111796e-01 9.49787945e-02
-3.37514251e-01 -1.94143623e-01 1.52420655e-01 -3.44818860e-01
4.55786020e-01 -4.97023351e-02 1.17899291e-01 -4.40979332e-01
7.02879608e-01 4.60158676e-01 2.30222329e-01 6.91881478e-01
-1.40609005e-02 5.82423747e-01 1.22858956e-01 -5.07387936e-01
-7.91513622e-01 -1.02095270e+00 -5.02569139e-01 8.03079844e-01
-9.42147002e-02 1.50474206e-01 -8.24501097e-01 -3.66856158e-01
1.76831454e-01 6.97956204e-01 -7.81593204e-01 1.85325578e-01
-2.29187354e-01 -1.16498041e+00 4.02320325e-01 1.52236493e-02
4.28743899e-01 -1.35147822e+00 -8.76328290e-01 -3.09876464e-02
7.64308944e-02 -8.72359455e-01 5.98706126e-01 4.89083320e-01
-1.09488404e+00 -1.07912028e+00 -2.91520208e-01 -4.10952181e-01
7.99134791e-01 6.12113476e-01 8.74911487e-01 3.93753797e-01
-2.87156105e-01 -1.48094445e-01 -3.73353243e-01 -8.10280740e-01
-5.37271142e-01 -9.34140533e-02 -6.95397779e-02 3.91733825e-01
4.84424412e-01 -7.04360545e-01 -3.59084874e-01 4.85899955e-01
-1.23347533e+00 -2.04318300e-01 7.04784691e-01 5.67250788e-01
5.48196852e-01 4.20027852e-01 2.08328038e-01 -9.03949022e-01
2.42229208e-01 -2.06279188e-01 -1.07395577e+00 3.57674398e-02
-7.40611970e-01 -1.40575200e-01 5.37020147e-01 1.60345342e-02
-8.50700498e-01 2.40725517e-01 2.60386914e-01 -3.77772093e-01
-5.85484147e-01 5.20172179e-01 -1.03068523e-01 -5.54545105e-01
9.43463802e-01 3.86191577e-01 -1.37183726e-01 -4.14575934e-01
-1.89340249e-01 7.64468729e-01 5.93026578e-01 -1.05478995e-01
1.30331123e+00 8.16237628e-01 7.10966513e-02 -1.24353778e+00
-3.81815195e-01 -8.60192597e-01 -8.22395802e-01 -6.06594026e-01
9.38511133e-01 -5.83988667e-01 -4.21994358e-01 5.84922373e-01
-1.00864530e+00 1.62481871e-02 -1.90023437e-01 4.13207829e-01
1.14451982e-01 2.51187474e-01 1.59877941e-01 -1.24921024e+00
-3.00757557e-01 -8.81907403e-01 6.44996345e-01 8.21060613e-02
3.22115660e-01 -7.77587652e-01 2.66448647e-01 2.18535155e-01
6.53514981e-01 5.25698423e-01 1.02117133e+00 -4.49529648e-01
-5.80722392e-01 -1.39358878e-01 -2.65856653e-01 5.22992730e-01
3.24463904e-01 3.17529023e-01 -1.56193960e+00 -2.57912517e-01
2.19398290e-01 4.04040396e-01 1.05015182e+00 1.98377281e-01
8.33178937e-01 -1.57959163e-02 -1.20560370e-01 4.92803961e-01
1.77493608e+00 4.21238482e-01 7.27146804e-01 8.05383146e-01
2.98122883e-01 7.85480618e-01 4.99673367e-01 1.11391254e-01
-3.33053201e-01 6.62281334e-01 9.79480028e-01 -5.17760873e-01
1.66544244e-01 2.98033088e-01 2.13000085e-02 4.21871871e-01
-6.37512386e-01 -4.81401803e-03 -1.11902702e+00 6.08982861e-01
-1.70605350e+00 -1.25041640e+00 -4.43672568e-01 2.57731152e+00
2.69597054e-01 2.36584231e-01 1.94303971e-02 3.96324962e-01
9.00411189e-01 1.29737929e-01 -7.07901195e-02 -4.06762719e-01
-5.85719109e-01 4.98804212e-01 6.60338998e-01 4.88450408e-01
-1.34074748e+00 7.13548720e-01 5.76474762e+00 5.08531094e-01
-1.34113038e+00 1.82157829e-01 8.59787017e-02 1.12005182e-01
-4.66436744e-02 1.25144079e-01 -4.55217063e-01 4.70799237e-01
6.91384017e-01 2.18426615e-01 4.36253339e-01 4.66065288e-01
7.21836329e-01 -4.48550403e-01 -3.50500613e-01 6.37477756e-01
-5.98994680e-02 -1.22749996e+00 1.52914107e-01 9.86785665e-02
7.38071501e-01 1.71484333e-02 1.18926186e-02 -1.61462411e-01
-3.56501698e-01 -1.00462377e+00 6.90916002e-01 5.89179456e-01
5.47025263e-01 -6.33292854e-01 1.06383419e+00 1.91929221e-01
-1.14192820e+00 -2.47236177e-01 -3.84927392e-01 -3.55555445e-01
-3.51611972e-01 7.32648730e-01 -9.55021501e-01 6.23692691e-01
9.34501529e-01 7.07264543e-02 -8.69437397e-01 1.38459790e+00
-2.34577879e-01 6.30682409e-01 -3.16154808e-01 1.40413553e-01
3.88197422e-01 -4.57464635e-01 7.30678082e-01 1.45346117e+00
2.41646305e-01 2.07233243e-02 -1.79561809e-01 8.43293071e-01
6.65695965e-01 1.81315884e-01 -5.23423314e-01 -8.07681531e-02
3.94767374e-01 1.33122790e+00 -1.03618288e+00 -3.39698434e-01
-6.97307289e-02 5.10389626e-01 -2.21641883e-01 3.41622651e-01
-4.87060785e-01 -2.74231404e-01 6.91245437e-01 4.75431502e-01
4.48029488e-01 -1.64197683e-01 -3.59435737e-01 -6.02796078e-01
8.87687951e-02 -9.11659300e-01 2.66655654e-01 -9.15468812e-01
-9.03944552e-01 1.05320787e+00 9.56142098e-02 -1.51375484e+00
-6.83299974e-02 -8.06714833e-01 -5.80655992e-01 1.29906523e+00
-1.61899483e+00 -9.59623575e-01 -8.20484161e-01 4.73570287e-01
-5.47728362e-03 -1.67528167e-01 1.12321413e+00 2.59199560e-01
-2.04256654e-01 -1.08356558e-01 4.87525433e-01 -1.01145908e-01
4.07467008e-01 -1.30799520e+00 6.14016727e-02 1.16731334e+00
-1.55595969e-02 2.85748035e-01 9.16804373e-01 -6.53804183e-01
-4.82384831e-01 -1.06038988e+00 9.06013250e-01 -8.77835006e-02
5.75383246e-01 -2.82365590e-01 -1.23830295e+00 1.84825920e-02
1.27776861e-01 -1.42643958e-01 4.53688741e-01 -2.93699741e-01
-1.77393511e-01 -3.87163699e-01 -1.05997670e+00 1.80275992e-01
3.52687001e-01 -6.42620444e-01 -5.26234865e-01 4.65950042e-01
1.34873614e-01 -7.49420226e-02 -3.58651340e-01 5.50312936e-01
3.14457625e-01 -1.55565464e+00 8.17056894e-01 -4.34640884e-01
4.96473201e-02 -8.27403545e-01 -3.10937185e-02 -1.15488636e+00
-5.04828930e-01 1.50875626e-02 5.65977573e-01 1.06179690e+00
3.82228017e-01 -7.77851343e-01 7.54702568e-01 -6.97923228e-02
-1.07055455e-01 -7.65307322e-02 -8.70388746e-01 -9.95740354e-01
-2.15155959e-01 -9.79591161e-02 4.59519982e-01 9.40698922e-01
-6.40767455e-01 -2.13033333e-01 6.39209822e-02 9.20844138e-01
5.80074608e-01 4.82835799e-01 5.07888734e-01 -1.67427289e+00
-2.07847700e-01 -4.77271140e-01 -7.47345567e-01 4.17046785e-01
-2.07635537e-01 -7.49438286e-01 4.82292026e-02 -1.28888607e+00
9.81368707e-04 -3.55757862e-01 -4.72314477e-01 5.62002897e-01
-1.61244139e-01 5.59434831e-01 3.91709983e-01 5.59931040e-01
2.81393260e-01 2.24603098e-02 6.87839270e-01 -2.86209047e-01
-4.35438454e-01 2.62801737e-01 -7.10208192e-02 6.18981302e-01
8.20443869e-01 -7.78064311e-01 -3.82181965e-02 -2.31581986e-01
1.63428292e-01 -4.42205220e-01 8.15958917e-01 -1.46095836e+00
1.02035403e-01 -6.32597953e-02 4.01447117e-01 -7.87131786e-01
2.06951246e-01 -1.15486062e+00 7.23432839e-01 6.51489258e-01
2.24697605e-01 5.99565618e-02 5.61882794e-01 2.56071955e-01
-4.84203815e-01 -4.68128085e-01 9.75208282e-01 -2.92852104e-01
-9.08691168e-01 -1.94888622e-01 -7.84609437e-01 -5.23606420e-01
9.68185008e-01 -6.49891198e-01 -2.23442674e-01 -2.73064952e-02
-1.04157281e+00 -1.99143335e-01 5.37801623e-01 8.32342580e-02
3.69012743e-01 -7.12131739e-01 -6.67511046e-01 3.54821801e-01
3.10920477e-01 -3.79733562e-01 3.63138825e-01 8.35294247e-01
-7.87017643e-01 4.82131630e-01 -4.74007338e-01 -8.77964914e-01
-1.68609416e+00 7.73635924e-01 7.53738165e-01 -1.07737206e-01
-2.02707827e-01 2.77956724e-01 1.75675511e-01 -3.12827468e-01
-4.69096154e-02 -1.69449598e-01 -6.40243769e-01 3.57761204e-01
4.39903468e-01 2.79741943e-01 7.30184793e-01 -6.32310927e-01
-5.40000677e-01 7.09080040e-01 4.98486847e-01 -6.59351274e-02
1.34027064e+00 2.05310971e-01 -5.78431964e-01 4.98970360e-01
4.99465942e-01 1.98320106e-01 -7.59623468e-01 -1.34702278e-02
1.53619960e-01 -6.66014135e-01 1.31097108e-01 -1.07059503e+00
-1.04203737e+00 9.88094926e-01 1.24067688e+00 6.74354017e-01
1.45650768e+00 -5.47828853e-01 -2.48255774e-01 3.28430176e-01
1.67226598e-01 -9.80352342e-01 -4.58644480e-01 3.19558203e-01
6.98916018e-01 -1.10492873e+00 1.64110754e-02 -5.38701832e-01
-2.68514425e-01 1.27378690e+00 2.40479469e-01 9.79111493e-02
4.24694836e-01 2.30996951e-01 4.22442198e-01 -4.46858943e-01
-1.14853382e-01 -6.16167903e-01 1.89000294e-01 7.24486649e-01
6.45808801e-02 3.87279183e-01 -1.96356088e-01 -5.33845276e-02
3.54042947e-02 -2.14118580e-03 3.02429229e-01 6.75659001e-01
-7.80730009e-01 -8.53423059e-01 -1.16224754e+00 2.74993896e-01
-1.43701687e-01 -2.72729814e-01 -7.45231152e-01 1.00626302e+00
6.39754832e-01 1.03153455e+00 5.96617907e-02 -4.50673729e-01
3.61777037e-01 5.40868975e-02 9.99558270e-02 -5.33745289e-01
-1.15584445e+00 -7.08819032e-02 -1.32640347e-01 -1.44962624e-01
-8.14167023e-01 -6.63652480e-01 -8.89678955e-01 -1.85967848e-01
-3.61303836e-01 6.32850587e-01 9.68587697e-01 8.28694522e-01
1.48168698e-01 4.13381010e-01 8.46517682e-01 -1.09449625e+00
-1.45909473e-01 -8.52614105e-01 -7.74163842e-01 1.87983483e-01
4.07743633e-01 -7.24871039e-01 -6.73544884e-01 -5.24787158e-02] | [9.720248222351074, -1.751252293586731] |
e1638a77-a1af-42e5-adc9-73b1594e4563 | improving-transformer-based-end-to-end | 2303.01192 | null | https://arxiv.org/abs/2303.01192v1 | https://arxiv.org/pdf/2303.01192v1.pdf | Improving Transformer-based End-to-End Speaker Diarization by Assigning Auxiliary Losses to Attention Heads | Transformer-based end-to-end neural speaker diarization (EEND) models utilize the multi-head self-attention (SA) mechanism to enable accurate speaker label prediction in overlapped speech regions. In this study, to enhance the training effectiveness of SA-EEND models, we propose the use of auxiliary losses for the SA heads of the transformer layers. Specifically, we assume that the attention weight matrices of an SA layer are redundant if their patterns are similar to those of the identity matrix. We then explicitly constrain such matrices to exhibit specific speaker activity patterns relevant to voice activity detection or overlapped speech detection tasks. Consequently, we expect the proposed auxiliary losses to guide the transformer layers to exhibit more diverse patterns in the attention weights, thereby reducing the assumed redundancies in the SA heads. The effectiveness of the proposed method is demonstrated using the simulated and CALLHOME datasets for two-speaker diarization tasks, reducing the diarization error rate of the conventional SA-EEND model by 32.58% and 17.11%, respectively. | ['Joon-Hyuk Chang', 'Jeong-Hwan Choi', 'Joon-Young Yang', 'Ye-Rin Jeoung'] | 2023-03-02 | null | null | null | null | ['activity-detection'] | ['computer-vision'] | [ 2.19327435e-01 3.06462735e-01 1.33745968e-01 -5.29623687e-01
-6.42718136e-01 -3.06622654e-01 2.48968899e-01 -3.20448816e-01
-2.41721958e-01 3.27463657e-01 3.63185644e-01 -2.92683452e-01
-2.49950718e-02 -1.93791270e-01 -3.99849176e-01 -9.08965051e-01
-7.38765392e-03 2.93068826e-01 -2.29505301e-02 -6.91801012e-02
1.40891818e-04 4.22691584e-01 -1.61084664e+00 1.13087744e-01
1.10041809e+00 1.05278301e+00 4.39491451e-01 4.55511928e-01
-5.34244403e-02 6.01261556e-01 -8.24264050e-01 -1.46144047e-01
2.34083422e-02 -4.22065049e-01 -3.79400849e-01 2.38939717e-01
4.11887884e-01 -1.64216459e-01 -3.83502871e-01 1.01963758e+00
9.92459357e-01 4.16653126e-01 7.38243222e-01 -1.16943276e+00
-4.62387472e-01 8.02219987e-01 -6.60734057e-01 5.25828183e-01
-9.12494585e-03 -4.26446684e-02 1.08172286e+00 -9.96494293e-01
1.01913959e-01 1.37537956e+00 4.83459592e-01 7.06052780e-01
-1.17739737e+00 -9.62623656e-01 6.25618160e-01 3.80613416e-01
-1.56722236e+00 -1.06773150e+00 1.11762011e+00 -4.09641147e-01
8.97811115e-01 4.11302477e-01 1.59668207e-01 9.53015924e-01
-1.69913486e-01 8.52528632e-01 5.35019219e-01 -3.95655096e-01
2.90067792e-02 3.20330143e-01 3.35617006e-01 5.72348654e-01
-1.42652944e-01 -1.12652116e-01 -6.85000539e-01 1.17989399e-01
3.66121590e-01 -2.57545441e-01 -3.70001197e-01 -7.11311027e-02
-9.32907522e-01 7.65808463e-01 1.69461906e-01 3.09538782e-01
-5.01057327e-01 -4.06959832e-01 4.67005879e-01 -1.35971131e-02
6.42979324e-01 5.17856441e-02 -1.85681209e-01 1.26493886e-01
-8.58845592e-01 -8.62384513e-02 2.98235625e-01 1.15448248e+00
4.73130494e-01 5.30228436e-01 -4.80566949e-01 1.39934611e+00
5.08315742e-01 3.20432246e-01 7.62757599e-01 -7.20329463e-01
7.24092305e-01 2.63022482e-01 -1.27989843e-01 -7.33776987e-01
-2.74972886e-01 -8.76935065e-01 -9.67007816e-01 -1.95409551e-01
5.39224371e-02 -2.54990637e-01 -7.65037060e-01 2.04907250e+00
3.46740395e-01 1.63481891e-01 1.18725128e-01 8.97157490e-01
7.12928891e-01 7.83398390e-01 -1.02511689e-01 -2.69605219e-01
1.35762584e+00 -9.11857188e-01 -9.74950433e-01 -4.80979711e-01
3.44219059e-01 -7.25788414e-01 1.36169553e+00 1.58788875e-01
-1.09751391e+00 -7.33829021e-01 -1.11446202e+00 8.04651603e-02
2.34468400e-01 5.15505254e-01 1.17730834e-01 5.62524617e-01
-9.01548564e-01 1.05841652e-01 -6.56599581e-01 -9.39293802e-02
3.15879852e-01 4.01262194e-01 1.25669420e-01 1.87030911e-01
-1.33628833e+00 5.46656609e-01 5.00777215e-02 3.52973700e-01
-1.13104153e+00 -7.36431420e-01 -7.52943933e-01 2.94096172e-01
-7.34312832e-03 -3.02441180e-01 1.34259903e+00 -8.87418509e-01
-1.65402198e+00 7.75440395e-01 -4.75584090e-01 -3.90375763e-01
1.70228064e-01 -1.06573090e-01 -4.47721988e-01 -3.77347097e-02
-1.05539419e-01 6.19479597e-01 9.50912237e-01 -1.19154012e+00
-6.20113254e-01 -4.23953712e-01 -2.72847474e-01 5.45072198e-01
-6.64141774e-01 6.13055080e-02 -3.79349709e-01 -8.53382409e-01
1.98385939e-01 -8.93835902e-01 2.56111681e-01 -3.04568708e-01
-7.26128459e-01 -4.28639889e-01 7.90268719e-01 -9.82228220e-01
1.48472846e+00 -2.60012317e+00 2.14572653e-01 1.41466618e-01
1.70140862e-01 2.58206993e-01 -2.63508838e-02 -1.36614054e-01
-3.65844592e-02 -2.07666799e-01 -9.28585753e-02 -8.13507974e-01
2.07933441e-01 1.16997257e-01 -1.70567766e-01 3.25959295e-01
1.84941366e-01 2.13409378e-03 -3.25881302e-01 -4.87744033e-01
-2.99954670e-03 5.68085074e-01 -7.08751976e-01 4.93219227e-01
1.97025061e-01 2.69318491e-01 -1.63360626e-01 3.46104532e-01
6.76304698e-01 2.89766669e-01 1.19672172e-01 -4.01907384e-01
-2.58976758e-01 8.98964107e-01 -1.07368875e+00 1.16549945e+00
-6.28269732e-01 6.56521916e-01 5.03820479e-01 -8.52887928e-01
9.03578281e-01 5.25699973e-01 1.82629839e-01 -6.49421155e-01
1.78780615e-01 -1.54412528e-02 5.00401139e-01 -1.70388609e-01
1.81345522e-01 -1.74811155e-01 2.99376756e-01 2.76462317e-01
9.31327790e-02 2.68930733e-01 -1.53517336e-01 -4.21790183e-02
5.88894665e-01 -5.99799514e-01 -1.69438437e-01 -4.66693163e-01
8.34812164e-01 -7.81592786e-01 9.46351707e-01 4.75697339e-01
-4.20216411e-01 3.76752764e-01 1.73059344e-01 1.40719280e-01
-7.59111285e-01 -1.01291728e+00 -1.29941806e-01 1.36086476e+00
-1.19669639e-01 4.46607843e-02 -1.02320623e+00 -3.89839023e-01
-2.86628574e-01 1.14210486e+00 -3.41898829e-01 -5.50028205e-01
-6.24926984e-01 -5.71276903e-01 7.67808259e-01 4.88412142e-01
4.88250613e-01 -9.29482043e-01 -1.47903383e-01 3.38209391e-01
-3.76128346e-01 -9.84199405e-01 -1.16498435e+00 4.80527610e-01
-6.03422821e-01 -4.29268062e-01 -7.51081109e-01 -1.17505622e+00
5.21720171e-01 3.63698393e-01 5.47417819e-01 -3.03792864e-01
1.58448249e-01 -7.91872144e-02 -4.96224463e-02 -3.59995544e-01
-4.80073929e-01 3.06635410e-01 4.59423691e-01 4.46680665e-01
2.75640696e-01 -6.81055605e-01 -4.62011099e-01 5.82356632e-01
-4.56046522e-01 -2.60839425e-02 2.86863893e-01 9.21983480e-01
4.27690119e-01 1.36082023e-01 9.88047957e-01 -3.47883493e-01
5.92091441e-01 -3.61081243e-01 -4.54929590e-01 7.48432651e-02
-4.41781282e-01 9.94038135e-02 6.42558515e-01 -7.23878384e-01
-1.35048902e+00 -1.45471051e-01 -4.67576236e-01 -7.03851938e-01
-3.50078195e-02 1.81519657e-01 -8.92262101e-01 3.27959388e-01
3.35407645e-01 4.55893099e-01 -7.24626631e-02 -6.63906872e-01
-8.38592649e-02 8.98668170e-01 3.14687997e-01 -2.73283571e-01
5.00203490e-01 -9.84662697e-02 -6.41777158e-01 -1.09451997e+00
-7.39749551e-01 -4.20840859e-01 -2.63997406e-01 -9.19152889e-03
7.75359690e-01 -1.03214800e+00 -6.78354561e-01 6.73133016e-01
-1.06917322e+00 -2.09575996e-01 -8.93587619e-02 6.48533404e-01
-3.44305903e-01 3.21155429e-01 -6.05907083e-01 -1.09647810e+00
-6.28396571e-01 -1.32871568e+00 9.00363624e-01 1.28067851e-01
-1.95339888e-01 -7.19522059e-01 -1.73870713e-01 3.11659962e-01
4.23368692e-01 -6.38072729e-01 1.11698401e+00 -8.51199448e-01
-2.53916695e-03 1.08494319e-01 1.11821957e-01 5.96972883e-01
4.12475616e-01 -1.73340872e-01 -1.47111857e+00 -1.97456032e-01
1.87794149e-01 1.63527712e-01 6.34380162e-01 5.87460577e-01
1.12562811e+00 -5.54131031e-01 -2.08713442e-01 4.74153787e-01
6.15081310e-01 7.19891965e-01 3.12746406e-01 -6.90655485e-02
6.09421492e-01 8.05235445e-01 4.98316497e-01 5.61972439e-01
3.41255218e-01 9.09179807e-01 2.45576873e-01 -7.95876235e-02
-2.83125579e-01 -1.76647529e-01 5.88800788e-01 1.33666456e+00
3.65111977e-01 -5.21490574e-01 -6.63922191e-01 7.92770982e-01
-1.33119977e+00 -8.53901088e-01 1.62476107e-01 2.33881664e+00
1.00758994e+00 3.93829554e-01 1.92352518e-01 3.28387558e-01
1.20952296e+00 1.63218126e-01 -7.87324727e-01 -4.84901190e-01
-1.30741104e-01 -1.58824503e-01 6.89733773e-02 6.22612238e-01
-1.05695319e+00 8.04278612e-01 5.30724859e+00 9.23190296e-01
-1.02700257e+00 2.64609218e-01 4.71923530e-01 -2.11803153e-01
-2.65602082e-01 -5.68367124e-01 -1.29060626e+00 5.90032876e-01
1.09946239e+00 -1.89373612e-01 4.21930373e-01 7.71968365e-01
5.45777142e-01 3.93219680e-01 -1.14231765e+00 8.98406982e-01
1.31501600e-01 -7.25079119e-01 1.23722471e-01 1.56223020e-02
3.88354391e-01 -1.50272369e-01 3.77269715e-01 3.66268635e-01
1.96039546e-02 -6.21594369e-01 9.79787588e-01 2.04726085e-01
6.85139894e-01 -9.13348973e-01 5.25518477e-01 3.13915014e-01
-1.14880216e+00 -1.73982486e-01 -2.63925672e-01 3.37541521e-01
1.38628766e-01 5.72218239e-01 -1.19961727e+00 1.48131698e-01
5.22743344e-01 3.08090925e-01 -1.79915503e-01 9.28449452e-01
8.32304135e-02 1.03226197e+00 -3.75146478e-01 2.24325567e-01
1.20642640e-01 -2.62017325e-02 8.17287505e-01 1.12829769e+00
3.67713392e-01 -1.96376279e-01 -1.60694465e-01 7.18525410e-01
-2.22905129e-01 9.36863273e-02 -2.68179864e-01 9.84123722e-02
8.27913284e-01 7.92044520e-01 3.49768139e-02 -1.27020612e-01
-2.31051281e-01 9.81956422e-01 1.74425140e-01 4.30930227e-01
-1.05886209e+00 -6.92592502e-01 1.24392045e+00 4.68784720e-02
6.19323015e-01 -4.08896282e-02 -1.29918352e-01 -8.48688781e-01
-1.21804729e-01 -9.57075894e-01 2.14211121e-01 -4.69437659e-01
-1.30785501e+00 9.58266973e-01 -3.07963520e-01 -1.11695325e+00
-1.51252210e-01 -1.52020693e-01 -8.66314113e-01 1.16721082e+00
-1.43669927e+00 -9.70064402e-01 -8.93357918e-02 7.43915260e-01
9.91794109e-01 -2.85167634e-01 6.37991071e-01 7.92105317e-01
-1.22816730e+00 1.29756761e+00 1.83355361e-01 -1.59608554e-02
6.73248172e-01 -9.35877562e-01 2.49305621e-01 8.81029308e-01
-4.76719365e-02 5.49311757e-01 7.95602202e-01 -1.67413071e-01
-7.96694636e-01 -1.26320076e+00 8.78027737e-01 2.84865081e-01
2.73076773e-01 -7.04113424e-01 -1.06329250e+00 6.55282855e-01
1.54263750e-01 -3.65665376e-01 7.25383580e-01 1.38543427e-01
-9.39705893e-02 -4.91768271e-01 -8.52664113e-01 5.85317314e-01
9.07303572e-01 -6.65186465e-01 -5.99331915e-01 1.31853238e-01
8.63271117e-01 -3.67942788e-02 -5.62187672e-01 8.74100104e-02
4.12501067e-01 -5.69188297e-01 9.39081132e-01 -3.49273533e-01
-2.31470972e-01 -3.76872778e-01 -3.07649165e-01 -1.44694078e+00
-5.69446266e-01 -4.03392047e-01 -1.79530323e-01 1.77333391e+00
5.00565469e-01 -4.95510906e-01 4.86534446e-01 2.85213619e-01
-7.45959878e-01 -3.83549303e-01 -1.30752897e+00 -6.87450945e-01
-1.83442935e-01 -2.16475770e-01 6.06369078e-01 8.65953505e-01
-1.67753324e-01 6.69686139e-01 -4.06785786e-01 4.85406309e-01
5.08991659e-01 -2.60379642e-01 2.12827742e-01 -1.04570568e+00
-2.82167464e-01 -5.01068532e-01 -9.17065889e-02 -1.38018060e+00
4.39178973e-01 -7.64739990e-01 3.68888736e-01 -1.11024320e+00
-2.76754171e-01 -4.65563089e-01 -5.20867348e-01 4.29517180e-01
-2.48755500e-01 -4.04985994e-02 -6.26460984e-02 5.17067201e-02
-2.34290823e-01 1.12007093e+00 9.09964323e-01 -3.44777554e-01
-4.88627344e-01 4.88869846e-01 -6.90467298e-01 7.40978420e-01
8.58869433e-01 -3.92584622e-01 -7.13813722e-01 -4.65479821e-01
-5.76257408e-01 -1.78717859e-02 -6.86521232e-02 -9.56838608e-01
2.53507882e-01 1.20501615e-01 4.25510146e-02 -7.57131159e-01
5.92399478e-01 -6.48114026e-01 -7.42262602e-02 2.66972840e-01
-5.13546884e-01 -2.62156010e-01 4.47545648e-01 3.69656682e-01
-2.30646431e-01 -1.01610601e-01 9.90078509e-01 3.86348635e-01
-2.36868933e-01 1.05792500e-01 -8.39699924e-01 -5.97244427e-02
7.18755543e-01 -2.44942725e-01 1.33534685e-01 -4.62081462e-01
-7.89500713e-01 2.11771160e-01 -2.01375738e-01 3.87721717e-01
5.79709113e-01 -1.30190110e+00 -6.77236199e-01 4.92412329e-01
8.63809232e-03 -9.13039818e-02 6.41900420e-01 9.60145116e-01
5.64663224e-02 5.47083497e-01 -1.17582276e-01 -5.23522317e-01
-1.42610252e+00 4.20314729e-01 6.74846172e-01 -9.51224100e-03
-4.49783981e-01 1.16754615e+00 9.32829320e-01 -3.86019796e-01
7.44919896e-01 -3.89244825e-01 -3.91011596e-01 1.13552347e-01
4.58555579e-01 3.84615570e-01 2.26946056e-01 -7.62633383e-01
-6.44452572e-01 1.87431335e-01 -3.74786139e-01 -6.01152666e-02
1.14756477e+00 -6.47814274e-01 4.64882404e-01 4.70528275e-01
1.11589611e+00 2.58966446e-01 -1.46304846e+00 -2.92476445e-01
-2.35115737e-01 -5.12113646e-02 4.35766578e-01 -7.13813841e-01
-1.13201237e+00 1.27626598e+00 8.53171587e-01 5.47018349e-02
1.22560883e+00 -1.61538236e-02 7.77993560e-01 5.38094416e-02
-1.74729809e-01 -9.09996092e-01 -1.03042789e-01 4.99369770e-01
1.02257407e+00 -7.54976809e-01 -6.08997941e-01 -3.19571584e-01
-8.73701692e-01 6.30408466e-01 8.53492677e-01 3.23767215e-01
5.60773551e-01 -3.65315825e-02 1.24733141e-02 1.55764088e-01
-7.13235378e-01 -3.96759063e-02 4.41010535e-01 5.33878863e-01
5.49748361e-01 -1.01138309e-01 5.52005060e-02 9.84853864e-01
-1.40478328e-01 -7.96000421e-01 2.34964669e-01 3.87180984e-01
-5.17009854e-01 -6.74571157e-01 -4.85865921e-01 2.52697647e-01
-3.29807371e-01 -3.67289007e-01 -2.34637037e-01 2.48744726e-01
-1.10910833e-02 1.14258850e+00 2.80463129e-01 -4.85681802e-01
6.18767202e-01 5.20297527e-01 1.01319805e-01 -5.45773268e-01
-5.62945604e-01 4.82140541e-01 1.71059236e-01 1.04838312e-01
-4.65821028e-02 -7.20307887e-01 -1.24857914e+00 -1.50340214e-01
-5.83665609e-01 2.87490994e-01 5.52642345e-01 7.54022121e-01
7.11909771e-01 1.02670753e+00 9.57134247e-01 -7.23481238e-01
-7.11769402e-01 -1.34116864e+00 -8.70972693e-01 3.79023552e-02
4.98471230e-01 -9.72253859e-01 -6.01321459e-01 -5.23336008e-02] | [14.5946044921875, 6.10465145111084] |
ba01dd81-3453-4751-8367-ba86f29fea71 | malaria-likelihood-prediction-by-effectively | 1711.09223 | null | http://arxiv.org/abs/1711.09223v1 | http://arxiv.org/pdf/1711.09223v1.pdf | Malaria Likelihood Prediction By Effectively Surveying Households Using Deep Reinforcement Learning | We build a deep reinforcement learning (RL) agent that can predict the
likelihood of an individual testing positive for malaria by asking questions
about their household. The RL agent learns to determine which survey question
to ask next and when to stop to make a prediction about their likelihood of
malaria based on their responses hitherto. The agent incurs a small penalty for
each question asked, and a large reward/penalty for making the correct/wrong
prediction; it thus has to learn to balance the length of the survey with the
accuracy of its final predictions. Our RL agent is a Deep Q-network that learns
a policy directly from the responses to the questions, with an action defined
for each possible survey question and for each possible prediction class. We
focus on Kenya, where malaria is a massive health burden, and train the RL
agent on a dataset of 6481 households from the Kenya Malaria Indicator Survey
2015. To investigate the importance of having survey questions be adaptive to
responses, we compare our RL agent to a supervised learning (SL) baseline that
fixes its set of survey questions a priori. We evaluate on prediction accuracy
and on the number of survey questions asked on a holdout set and find that the
RL agent is able to predict with 80% accuracy, using only 2.5 questions on
average. In addition, the RL agent learns to survey adaptively to responses and
is able to match the SL baseline in prediction accuracy while significantly
reducing survey length. | ['Vinaya Polamreddi', 'Anusha Balakrishnan', 'Pranav Rajpurkar'] | 2017-11-25 | null | null | null | null | ['holdout-set'] | ['computer-vision'] | [ 3.57847333e-01 4.63568836e-01 -2.13695079e-01 -8.23090315e-01
-7.66401052e-01 -4.15434837e-01 3.22836697e-01 3.22187930e-01
-8.30066442e-01 9.83027577e-01 1.79239467e-01 -6.64771855e-01
-1.25742853e-01 -1.36739945e+00 -7.38150656e-01 -5.60867310e-01
-2.23160803e-01 1.05967891e+00 -3.91290598e-02 -1.44674582e-02
-5.56946471e-02 2.12759361e-01 -9.94905889e-01 2.41892397e-01
7.64329016e-01 6.37453914e-01 4.15677249e-01 1.15575922e+00
1.68496802e-01 1.31866634e+00 -7.74365187e-01 -2.45805845e-01
1.85830131e-01 -7.61779666e-01 -9.01570439e-01 2.75665298e-02
1.88272417e-01 -1.23004413e+00 -1.73405722e-01 4.36395794e-01
5.93919337e-01 -1.11429542e-01 5.25223672e-01 -1.06828642e+00
-2.91589618e-01 4.13496912e-01 -1.84667528e-01 2.71541923e-01
7.03414857e-01 6.55327678e-01 9.66856837e-01 -1.30581260e-01
5.20576179e-01 1.24940145e+00 7.03933835e-01 6.91115379e-01
-1.47626519e+00 -6.03969514e-01 -1.77894682e-02 2.60825634e-01
-5.27878642e-01 -3.67998898e-01 2.26090640e-01 -4.32256401e-01
1.19870210e+00 -4.54194732e-02 6.04682624e-01 9.42122757e-01
3.36005241e-01 5.24435043e-01 1.22844899e+00 8.47007707e-02
8.53857338e-01 4.20654565e-02 -1.83599442e-01 5.36813855e-01
-8.96113887e-02 3.59349251e-01 -8.59659314e-02 -5.35961449e-01
6.44516706e-01 2.79157251e-01 6.66681901e-02 1.87858026e-02
-5.91645241e-01 1.43906069e+00 5.01206040e-01 -3.04820061e-01
-9.85051274e-01 1.74411163e-02 3.45786028e-02 5.99875212e-01
1.99762389e-01 5.10716975e-01 -9.49619055e-01 1.46852985e-01
-5.84467769e-01 6.51635468e-01 1.07960427e+00 2.88117025e-02
1.10735893e+00 -8.18344057e-02 -4.31639403e-01 6.04853094e-01
1.60922319e-01 1.09831750e+00 1.25758752e-01 -1.14985943e+00
2.54559577e-01 7.29067028e-01 3.81190509e-01 -6.22955978e-01
-7.56170988e-01 7.06657469e-02 -4.68344361e-01 1.12187199e-01
5.19853115e-01 -1.03067160e+00 -6.98356330e-01 1.89638305e+00
5.19722641e-01 -1.74239874e-01 6.18176907e-02 5.40680528e-01
4.29612726e-01 7.64977694e-01 3.76713485e-01 -1.33310392e-01
1.19874489e+00 -3.08803290e-01 -3.26248050e-01 -5.94138384e-01
5.69249153e-01 -3.56019646e-01 8.51792395e-01 1.86060533e-01
-8.22637856e-01 -2.93094158e-01 -4.33519632e-01 5.00616848e-01
-1.07474513e-01 -8.40395167e-02 4.29445356e-01 4.40644145e-01
-1.28684425e+00 5.37469804e-01 -8.84164333e-01 -5.89731812e-01
2.00017989e-01 7.80728281e-01 -1.61708519e-01 -2.34858140e-01
-1.28438735e+00 9.36214864e-01 6.71454817e-02 -3.36746633e-01
-1.11672747e+00 -5.05144358e-01 -8.05421591e-01 2.70703435e-01
3.45074087e-01 -4.28249985e-01 1.46386814e+00 -1.11089277e+00
-1.43626487e+00 4.97643143e-01 -2.42776811e-01 -8.38281035e-01
3.81293148e-01 7.46875629e-02 -1.26944751e-01 4.46012139e-01
4.36142325e-01 1.19196081e+00 5.19966364e-01 -5.64171731e-01
-9.92954493e-01 -4.27185684e-01 3.88751924e-01 2.38987520e-01
1.13743683e-02 -8.66434872e-02 3.80419880e-01 -1.33131355e-01
-5.78175247e-01 -8.80839586e-01 -5.16052961e-01 -2.39485919e-01
1.47374094e-01 -6.96047068e-01 3.90271932e-01 -8.16256166e-01
7.24908412e-01 -1.64404547e+00 -4.92198318e-01 -1.09736234e-01
-1.05342746e-01 1.80381164e-01 -6.66167200e-01 6.83917761e-01
4.14176464e-01 -1.48375958e-01 -3.43372285e-01 7.64925480e-02
-1.09957173e-01 4.78544027e-01 -2.74207503e-01 3.07254642e-01
5.98534346e-01 8.05261612e-01 -1.15480936e+00 -1.06638163e-01
2.00742751e-01 3.84607166e-01 -1.09421432e+00 1.10006118e+00
-7.52705574e-01 4.13300455e-01 -4.48019505e-01 2.13230610e-01
6.67121887e-01 -5.09058714e-01 6.49371326e-01 5.85596025e-01
2.96050876e-01 7.10728228e-01 -7.74720848e-01 4.92322922e-01
-3.96936953e-01 3.05464566e-01 8.54086205e-02 -1.03285313e+00
9.51921165e-01 3.62043828e-01 4.65639472e-01 -9.30716276e-01
-3.75485837e-01 4.62871976e-02 1.25814945e-01 -7.06225932e-01
-3.18646021e-02 -1.35744631e-01 1.35180624e-02 9.17201757e-01
-1.26174629e-01 -2.54734755e-02 1.20004572e-01 9.37717631e-02
1.63466609e+00 -2.95806359e-02 4.01764870e-01 -1.53416023e-01
4.55472410e-01 1.72500476e-01 5.53885698e-01 1.09895241e+00
-2.07038268e-01 -7.70664215e-02 8.22434425e-01 -9.20245051e-01
-9.42745507e-01 -1.14989853e+00 1.55387297e-01 1.57062614e+00
-5.66986620e-01 3.71611238e-01 -7.31925070e-01 -9.58860397e-01
1.25996530e-01 7.05953836e-01 -8.76507521e-01 1.06681570e-01
-6.06230199e-01 -6.87049985e-01 2.96966016e-01 2.57843167e-01
5.86394727e-01 -1.84276283e+00 -1.11254203e+00 5.36637247e-01
-2.78477417e-03 -5.52092791e-01 -2.46502712e-01 4.46840912e-01
-6.11285150e-01 -1.48522782e+00 -5.31701148e-01 -6.92919433e-01
6.82215929e-01 -1.25642195e-01 1.25738537e+00 1.04837961e-01
2.81279564e-01 3.68970275e-01 -2.05118835e-01 -1.54172689e-01
-5.82054436e-01 1.77849129e-01 -8.98876488e-02 -3.35240155e-01
8.95873487e-01 -2.10752040e-01 -1.04293203e+00 1.25318214e-01
-9.75088954e-01 -3.11933696e-01 4.57895696e-01 8.86223257e-01
1.86657801e-01 -2.95189232e-01 1.12266433e+00 -1.09825468e+00
5.97831309e-01 -1.08750367e+00 -8.72878611e-01 2.35318869e-01
-3.66191357e-01 1.54183939e-01 1.04954004e+00 -4.62683052e-01
-7.75863707e-01 2.31460303e-01 -5.03060341e-01 5.02883673e-01
-4.92054403e-01 3.30866694e-01 3.19032401e-01 3.16188782e-01
4.52435166e-01 2.94666827e-01 1.85234845e-01 -2.71095991e-01
-4.47203591e-02 5.26650846e-01 2.04932213e-01 -2.57520527e-01
3.75456303e-01 6.31508157e-02 -1.84150591e-01 -6.24585450e-01
-7.35622287e-01 -8.98308605e-02 2.84028891e-03 2.69691013e-02
8.73056591e-01 -8.46722901e-01 -1.38304758e+00 3.56626928e-01
-1.08297265e+00 -1.20029092e+00 -1.38850629e-01 3.30191344e-01
-4.21018213e-01 -1.51179790e-01 -7.83752203e-01 -7.87767410e-01
-4.11063462e-01 -1.08957803e+00 6.23671114e-01 3.14420044e-01
-3.98889899e-01 -1.20660102e+00 5.84851265e-01 1.84020102e-01
7.37959981e-01 1.51124552e-01 9.81059790e-01 -1.04438770e+00
-2.61076272e-01 3.31654251e-01 -1.39039621e-01 8.53517503e-02
2.63896346e-01 -4.10483301e-01 -6.86783135e-01 -6.24595344e-01
1.27967447e-01 -8.77089202e-01 6.29670262e-01 5.29327691e-01
6.94608927e-01 -1.07031655e+00 1.92349136e-01 2.05622017e-01
1.48286963e+00 4.51517701e-01 2.35103294e-01 1.92106068e-01
-7.11898133e-02 9.80208099e-01 3.09325218e-01 6.37098193e-01
9.95766938e-01 3.50225806e-01 5.11076510e-01 -6.89616948e-02
4.27624345e-01 -4.70582634e-01 7.24509537e-01 -8.69256910e-03
7.60368347e-01 -1.74044386e-01 -9.66679692e-01 5.56198776e-01
-1.49111021e+00 -8.21417272e-01 3.77405703e-01 2.23525500e+00
9.97430086e-01 -1.49599984e-01 4.84277606e-01 -1.62171185e-01
2.21407235e-01 7.39638694e-03 -1.05617416e+00 -6.42669916e-01
1.13247223e-01 5.98958313e-01 4.91011471e-01 1.06221604e+00
-8.83266389e-01 8.04517031e-01 6.85067749e+00 2.14099750e-01
-1.23016739e+00 -2.48178929e-01 1.10292161e+00 -9.64943727e-04
-2.63151318e-01 -2.92120844e-01 -6.28328204e-01 3.42689782e-01
1.39889443e+00 9.33004990e-02 9.03322458e-01 5.91728687e-01
4.14524823e-01 -5.61423182e-01 -1.20905471e+00 1.83030233e-01
-2.80065745e-01 -9.98739243e-01 -2.21794881e-02 -1.79911777e-02
8.27769697e-01 5.06501734e-01 -8.50092396e-02 4.98151720e-01
1.31978250e+00 -1.33315825e+00 -7.26304725e-02 3.18394452e-01
3.76647055e-01 -7.60813594e-01 9.29792643e-01 7.43042231e-01
-7.36069918e-01 -1.97516620e-01 -2.75256366e-01 -6.61826372e-01
-2.61474818e-01 -2.21144259e-02 -1.70499337e+00 -2.89698541e-01
5.69255590e-01 1.34109333e-01 -5.36561131e-01 5.15784144e-01
-4.46761668e-01 9.75123703e-01 -4.46559757e-01 -4.90761340e-01
5.75023890e-01 4.60565612e-02 -2.41611898e-01 1.02957118e+00
2.50604264e-02 6.07076466e-01 4.34012771e-01 6.45962000e-01
7.72254681e-03 -1.33500680e-01 -5.41087389e-01 2.12957397e-01
6.09563947e-01 8.37395310e-01 -2.32169092e-01 -4.75178868e-01
-2.54848361e-01 6.63029552e-01 5.40476203e-01 4.78428006e-01
-2.61667788e-01 -2.31156930e-01 4.41649765e-01 1.39681712e-01
3.17127436e-01 4.98612553e-01 8.65553990e-02 -6.42673790e-01
-2.92105705e-01 -1.13477087e+00 6.02405727e-01 -4.58986491e-01
-1.23343420e+00 5.42731822e-01 -2.25546628e-01 -2.39818037e-01
-1.05398345e+00 -3.21241856e-01 -7.17137396e-01 9.84430850e-01
-1.80947709e+00 -4.10152495e-01 1.52530730e-01 5.11806488e-01
4.22667474e-01 6.54293522e-02 8.87916386e-01 -6.42109960e-02
-2.42071450e-01 4.04534876e-01 3.95422056e-02 3.47172856e-01
2.82685101e-01 -1.34482551e+00 4.48057801e-01 1.45132110e-01
-6.04577303e-01 2.91408122e-01 6.28701270e-01 -6.77948236e-01
-1.26509523e+00 -1.16081595e+00 1.37901092e+00 -3.87183189e-01
4.17049110e-01 -1.69164926e-01 -1.07279503e+00 7.10414112e-01
1.18040830e-01 -2.78604925e-01 6.39693797e-01 -2.75712162e-01
2.04088017e-02 -2.26978466e-01 -1.73501658e+00 2.87521899e-01
9.95537415e-02 -5.08518219e-01 -5.67658722e-01 4.76387441e-01
6.64079428e-01 1.58074200e-01 -7.75582850e-01 1.63150311e-01
4.60881591e-01 -1.02141809e+00 5.96176505e-01 -5.93971968e-01
5.26965082e-01 1.17563397e-01 -1.80709183e-01 -1.29776704e+00
-3.01037669e-01 -3.10194075e-01 2.26036847e-01 5.93027353e-01
3.23435426e-01 -7.83853531e-01 1.27668655e+00 7.42635608e-01
8.77447963e-01 -1.04228449e+00 -9.48002398e-01 -1.97848380e-01
3.27550113e-01 1.86113730e-01 8.74833345e-01 5.78169763e-01
-2.29379550e-01 4.42145050e-01 -4.72370267e-01 3.33812654e-01
5.95104694e-01 5.57941608e-02 8.00612032e-01 -7.45912850e-01
-6.11400068e-01 -1.07577071e-01 4.21070576e-01 -9.84521866e-01
-5.90020604e-02 -3.27031523e-01 9.91934165e-02 -1.60914171e+00
4.19648439e-01 -3.14929515e-01 -8.58185813e-02 7.73178875e-01
-5.69844842e-01 -4.58868332e-02 2.41013765e-01 -1.91795319e-01
-6.77678943e-01 2.36083090e-01 7.35267878e-01 -1.19101420e-01
-4.13624167e-01 2.95471311e-01 -3.62233788e-01 4.33208168e-01
1.07868493e+00 -6.35122836e-01 -2.43505478e-01 -5.84202528e-01
2.80539304e-01 9.56364095e-01 2.71914423e-01 -7.30065167e-01
1.51060745e-01 -6.43287778e-01 6.76075697e-01 -5.31330943e-01
2.57333852e-02 -5.18676877e-01 1.49489060e-01 1.38606238e+00
-7.18031287e-01 2.62123942e-01 3.74253877e-02 1.38169110e-01
3.65753323e-01 -4.88326736e-02 9.59504426e-01 -3.36026400e-01
-2.01430857e-01 3.78348529e-01 -7.44237185e-01 1.47193745e-01
6.23216867e-01 2.91042387e-01 -1.49182871e-01 -7.23621726e-01
-3.89666677e-01 8.40810239e-01 3.75208795e-01 6.18293732e-02
5.46515882e-01 -9.10379827e-01 -1.13934028e+00 2.89483428e-01
-3.29185903e-01 -2.41204321e-01 -2.69141421e-02 1.08104482e-01
-5.07569969e-01 6.47280455e-01 -3.47845435e-01 -3.08242559e-01
-7.77160406e-01 3.49756867e-01 5.10891914e-01 -9.84342158e-01
-3.26853395e-02 4.99014527e-01 3.05578202e-01 -1.12491143e+00
2.71292806e-01 -5.40425777e-02 -3.16878021e-01 -1.13782816e-01
7.50571966e-01 4.71415788e-01 -3.72424543e-01 -1.32514417e-01
-3.96604091e-01 8.92020315e-02 3.23022413e-03 -2.31486380e-01
1.58453310e+00 2.57759869e-01 -1.16145443e-02 -8.29490125e-02
1.07606590e+00 -3.74480903e-01 -1.88938093e+00 -2.54040599e-01
-2.01081228e-03 -8.44070390e-02 -3.67434561e-01 -1.42502546e+00
-8.42466652e-01 8.15831780e-01 6.14054561e-01 4.26343918e-01
1.00449014e+00 -2.44118527e-01 6.13451183e-01 8.40407312e-01
2.74375498e-01 -7.15371132e-01 2.41204098e-01 5.27592540e-01
6.39168501e-01 -1.50712669e+00 -3.91277336e-02 6.33871436e-01
-5.71468234e-01 8.06884766e-01 6.50858641e-01 -4.22450125e-01
2.13329688e-01 -1.14877708e-02 1.95534229e-01 -6.95402101e-02
-1.24260986e+00 1.29545897e-01 -3.42064559e-01 9.13456261e-01
2.10110396e-01 5.46131074e-01 -3.97636518e-02 3.97527367e-02
-2.41597623e-01 1.53518498e-01 5.13824761e-01 5.39922655e-01
-8.73458624e-01 -1.11303401e+00 -4.06757057e-01 9.31081533e-01
-5.94958007e-01 -7.06910901e-03 -4.05012935e-01 3.61388862e-01
8.17274749e-02 1.33950007e+00 4.32283401e-01 -1.65076062e-01
-5.03554717e-02 -2.42135048e-01 6.79922253e-02 -5.64141273e-01
-9.46231604e-01 -2.89455295e-01 -1.15694404e-01 -4.63659286e-01
-1.00944079e-01 -6.88637078e-01 -1.20121145e+00 -6.17725492e-01
2.21337363e-01 4.59795833e-01 2.00716451e-01 9.13952112e-01
1.94307208e-01 -8.45231265e-02 1.23497736e+00 -4.03085917e-01
-1.04888153e+00 -9.67888117e-01 -1.69125944e-01 3.57074261e-01
7.82923937e-01 3.04315938e-03 -3.29330117e-01 -4.66317058e-01] | [4.06635856628418, 1.8331241607666016] |
150f1ea7-0cab-42f5-9d7f-ea2170920f29 | improving-solvability-for-procedurally | 1810.01926 | null | https://arxiv.org/abs/1810.01926v3 | https://arxiv.org/pdf/1810.01926v3.pdf | Improving Solvability for Procedurally Generated Challenges in Physical Solitaire Games Through Entangled Components | Challenges for physical solitaire puzzle games are typically designed in advance by humans and limited in number. Alternatively, some games incorporate rules for stochastic setup, where the human solver randomly sets up the game board before solving the challenge. These setup rules greatly increase the number of possible challenges, but can often generate unsolvable or uninteresting challenges. To better understand the compromises involved in minimizing undesirable challenges, we examine three games where component design choices can influence the stochastic nature of the resulting challenge generation algorithms. We evaluate the effect of these components and algorithms on challenge solvability and challenge engagement. We find that algorithms which control randomness through entangling components based on sub-elements of the puzzle mechanics can generate interesting challenges with a high probability of being solvable. | ['Mark Goadrich', 'James Droscha'] | 2018-10-03 | null | null | null | null | ['solitaire'] | ['playing-games'] | [ 2.29201138e-01 -3.15231644e-02 3.24522346e-01 2.69610703e-01
-8.34153235e-01 -1.25915718e+00 3.66762578e-01 -5.21629266e-02
-4.21045572e-01 8.92190695e-01 2.27972969e-01 -4.52582419e-01
-6.38064265e-01 -1.15460563e+00 -4.24416840e-01 -3.83897185e-01
1.18326485e-01 5.91642022e-01 1.42558604e-01 -3.16221565e-01
4.87140268e-01 1.15181282e-01 -1.36151671e+00 3.21859300e-01
9.33534980e-01 2.81023979e-01 2.94215798e-01 8.99507821e-01
2.24107113e-02 8.51769567e-01 -1.04156172e+00 -1.75886482e-01
6.05280697e-01 -7.49533236e-01 -6.64282441e-01 -1.58189878e-01
-3.81315559e-01 -1.54302180e-01 -1.99827746e-01 7.17229187e-01
4.70464736e-01 2.36326367e-01 7.97608614e-01 -1.25704980e+00
-2.62446076e-01 7.98914850e-01 -3.31023544e-01 2.71851629e-01
8.64336491e-01 6.91076994e-01 9.77894962e-01 -1.08256653e-01
7.76999712e-01 7.92501152e-01 6.41029537e-01 4.46215451e-01
-1.47348118e+00 -6.05382621e-01 7.04770908e-03 -4.22739796e-02
-1.53734314e+00 -1.95741460e-01 5.57078898e-01 -4.17587370e-01
6.92098558e-01 6.07433438e-01 7.03958988e-01 8.48223925e-01
1.59792423e-01 -2.76826322e-02 1.26690459e+00 -3.48280072e-01
6.62828147e-01 -2.55918562e-01 -1.37243524e-01 8.56963843e-02
8.57694924e-01 1.85334533e-01 -4.40029830e-01 -6.84964240e-01
8.56456041e-01 -5.41856885e-01 -2.25391090e-01 -3.02882046e-01
-7.16618419e-01 7.38941133e-01 -1.15618758e-01 4.17879522e-02
-3.33680302e-01 1.35929585e-01 1.24334786e-02 4.01613384e-01
-3.18569630e-01 1.60063565e+00 -4.60440457e-01 -6.35901093e-01
-6.77593946e-01 9.82623398e-01 1.09100211e+00 5.19915879e-01
5.41811347e-01 -1.85650781e-01 -2.77116269e-01 4.11701709e-01
-2.59513676e-01 2.20768247e-02 4.49726731e-02 -1.15057218e+00
5.13301373e-01 5.80158293e-01 5.67027092e-01 -7.91363776e-01
-4.23322797e-01 -2.26853848e-01 1.46162644e-01 6.21053994e-01
1.00134337e+00 -6.85820937e-01 -7.37543583e-01 1.81076014e+00
2.23542079e-01 -1.90745890e-01 -2.65313834e-01 8.41805398e-01
5.10542095e-01 5.17891824e-01 8.98182094e-02 -3.89482193e-02
1.20314169e+00 -5.03750503e-01 -3.95710975e-01 -3.27049166e-01
5.42656064e-01 -8.10841024e-01 1.34162676e+00 3.65925670e-01
-1.59719229e+00 -3.57094333e-02 -9.69106317e-01 3.86449784e-01
1.63325056e-01 -4.37024713e-01 5.51707864e-01 1.04785967e+00
-8.79649401e-01 7.00681984e-01 -6.98944092e-01 -3.69284488e-02
7.87608400e-02 5.68144977e-01 -5.06031774e-02 -1.24717221e-01
-8.13459218e-01 8.67760599e-01 6.57827035e-02 -1.68896735e-01
-7.22953498e-01 -6.99540496e-01 -5.56387603e-01 3.46838385e-01
1.04857922e+00 -6.90221727e-01 1.12472045e+00 -4.77389008e-01
-1.39993525e+00 2.94466525e-01 2.03928098e-01 2.23450407e-01
5.75385571e-01 4.45431411e-01 4.02873158e-01 -2.27596477e-01
8.29126835e-02 3.99354279e-01 2.71067828e-01 -1.09769380e+00
-2.52477139e-01 9.36734080e-02 9.29738820e-01 4.97288197e-01
1.45491198e-01 4.99818847e-02 5.82184792e-02 -2.80664921e-01
4.92926762e-02 -1.16734469e+00 -6.31936908e-01 -7.28073239e-01
-2.72742212e-01 5.72681166e-02 -7.66614079e-02 -1.30426526e-01
1.21582592e+00 -1.82517314e+00 3.10640395e-01 3.29288691e-01
1.93618253e-01 -2.10236341e-01 -6.19952917e-01 8.11374784e-01
1.17997020e-01 6.08842552e-01 2.79113710e-01 2.97798961e-01
3.00087720e-01 9.16557908e-02 2.83872802e-02 -9.07662045e-03
-4.16818783e-02 9.13198292e-01 -7.04392910e-01 8.47423002e-02
-1.63916841e-01 -1.20876744e-01 -1.06374443e+00 3.04787040e-01
-4.10121322e-01 4.27340031e-01 -5.67687869e-01 2.87666529e-01
6.31002128e-01 -2.32271969e-01 5.96507132e-01 5.87180734e-01
-1.73055843e-01 8.95186782e-01 -1.52377617e+00 9.73924994e-01
-2.76598543e-01 1.77344382e-01 1.11816034e-01 -3.10103238e-01
5.94858646e-01 7.56591707e-02 1.09446876e-01 -3.85192513e-01
2.13859364e-01 1.93879977e-01 6.75711572e-01 -2.06189439e-01
7.15605080e-01 -2.64402181e-01 -3.50766629e-01 1.18305337e+00
-6.01891637e-01 -9.15287077e-01 4.84157771e-01 1.28692716e-01
1.72999597e+00 -1.79418832e-01 1.66369870e-01 -3.93342227e-01
-5.99865243e-02 5.21522045e-01 8.26875150e-01 1.37019193e+00
-7.02160075e-02 8.73898208e-01 1.14085376e+00 -4.14021969e-01
-1.24991000e+00 -1.11925590e+00 4.24802750e-01 6.72038019e-01
2.91954279e-01 -7.51811624e-01 -8.96880686e-01 -3.67526650e-01
-1.69796258e-01 6.95821464e-01 -5.34716427e-01 -1.73133850e-01
-6.52447999e-01 -5.93919873e-01 5.00043392e-01 3.26726496e-01
-5.39706945e-02 -1.12636423e+00 -8.11949313e-01 2.17017829e-01
-4.20185387e-01 -1.00891805e+00 -6.30112290e-01 1.35944843e-01
-3.48332316e-01 -1.37963712e+00 -9.06471312e-02 -3.68095726e-01
7.64330447e-01 4.90817517e-01 1.31190932e+00 5.10535717e-01
-5.24560034e-01 3.89350563e-01 -4.55928952e-01 -2.82919884e-01
-3.49488735e-01 3.22719485e-01 3.62054296e-02 -9.02149260e-01
1.77028388e-01 -8.24130654e-01 -4.78087783e-01 4.10255224e-01
-7.71667361e-01 -9.85719785e-02 7.54072890e-02 6.89436853e-01
1.60329983e-01 4.86224234e-01 1.81949496e-01 -6.89945102e-01
1.43732429e+00 -5.44453382e-01 -6.78323805e-01 1.40606299e-01
-1.05051823e-01 2.16980606e-01 7.05023348e-01 -8.78411710e-01
-5.00883043e-01 -1.08116977e-01 2.69290537e-01 5.06874695e-02
-4.58236001e-02 4.76766467e-01 -7.43709598e-03 -4.09221381e-01
8.65399361e-01 1.18580749e-02 -2.03644648e-01 -6.14734599e-03
-5.15740141e-02 -6.31584004e-02 -1.67201668e-01 -1.33422148e+00
9.80051219e-01 -3.08876997e-03 -9.52305719e-02 -3.49739969e-01
-2.10660160e-01 3.04368466e-01 2.68096030e-02 -1.98926508e-01
6.65849984e-01 -6.41249895e-01 -1.17989564e+00 3.48806024e-01
-9.94213164e-01 -7.28196800e-01 -4.68834817e-01 2.28230387e-01
-5.77238798e-01 6.99774772e-02 -5.14919043e-01 -9.42269444e-01
2.63710231e-01 -1.52634895e+00 5.36745727e-01 6.91263855e-01
-1.00595629e+00 -4.59594637e-01 2.57606089e-01 4.97722089e-01
6.65314496e-01 3.53676796e-01 1.04169607e+00 -4.36661750e-01
-9.43677247e-01 2.82320827e-02 9.96381566e-02 -5.61938822e-01
1.31822482e-01 -1.69895887e-01 -3.74193817e-01 -8.94001275e-02
-1.44161284e-02 -4.14638132e-01 2.19343863e-02 5.04858732e-01
5.68625808e-01 -2.30724007e-01 -1.05722032e-01 3.12301874e-01
1.25217354e+00 3.24896544e-01 8.82380486e-01 5.51763237e-01
1.59318805e-01 6.85315609e-01 2.97294259e-01 7.48993814e-01
3.35965753e-01 7.07879484e-01 -1.43443700e-02 4.48666871e-01
3.59499693e-01 -2.75314599e-01 9.16396827e-02 1.98958576e-01
-2.86333948e-01 -3.79168004e-01 -1.04118288e+00 4.79517847e-01
-1.48791778e+00 -1.00427222e+00 -4.60515581e-02 2.08966613e+00
8.98184717e-01 4.75816935e-01 3.90159816e-01 2.61535704e-01
5.99004924e-01 1.40147686e-01 -2.16109365e-01 -5.50448120e-01
1.25409197e-02 5.67759871e-01 3.96500915e-01 8.34034503e-01
-2.00223848e-01 7.68318474e-01 7.56108427e+00 8.02304804e-01
-4.65849996e-01 -1.49113551e-01 5.52477479e-01 -6.74410760e-01
-7.54263639e-01 3.25235873e-01 -4.41493392e-01 4.77930576e-01
6.04755938e-01 -7.45113492e-01 9.67493415e-01 5.30391037e-01
4.69941825e-01 -6.15862787e-01 -9.89781976e-01 3.97428989e-01
-4.61175501e-01 -1.46734977e+00 -3.64643067e-01 3.21792573e-01
8.02792013e-01 -5.58765471e-01 1.67451501e-01 1.91463903e-01
1.13928568e+00 -1.37786710e+00 7.01512933e-01 7.10614696e-02
6.05542541e-01 -8.18611264e-01 2.44622320e-01 5.25882900e-01
-1.01374316e+00 -2.95450270e-01 -2.16064751e-01 -1.10532701e+00
2.94836551e-01 2.04373121e-01 -6.74156368e-01 6.15119934e-02
3.08320493e-01 -7.28876293e-01 -1.09727763e-01 1.06005359e+00
-3.76931936e-01 5.93432546e-01 -5.33752799e-01 -4.09507245e-01
2.88797021e-01 -2.96313792e-01 6.68040633e-01 2.77681231e-01
2.99235016e-01 9.65614021e-01 1.38222724e-01 1.29453516e+00
5.06458282e-01 -2.69173175e-01 -6.30951941e-01 -2.97921717e-01
9.42015827e-01 9.51131463e-01 -1.15008152e+00 3.38299125e-01
3.08785141e-01 5.36376238e-01 2.57517904e-01 4.36564356e-01
-9.02660012e-01 -2.45000884e-01 1.14924467e+00 3.82474363e-01
2.08144441e-01 -7.27281749e-01 -1.07783234e+00 -8.65797162e-01
1.64387390e-01 -1.37301981e+00 1.73780546e-01 -7.45835543e-01
-1.05901921e+00 5.35763800e-01 -9.06547233e-02 -5.84950745e-01
-2.92913079e-01 -2.87252814e-01 -1.20599043e+00 1.13028646e+00
-7.88403869e-01 -3.67100745e-01 -2.27567255e-01 1.86535209e-01
3.91371846e-01 -2.47213077e-02 5.63592911e-01 -2.76788235e-01
-4.91272777e-01 6.99387252e-01 -2.54566491e-01 -2.54407048e-01
1.70586452e-01 -9.01350558e-01 8.31272483e-01 7.35364616e-01
-3.67532894e-02 8.91410112e-01 1.03628516e+00 -9.36367095e-01
-1.47505450e+00 -2.89102525e-01 6.84468925e-01 -8.29407811e-01
4.59951609e-01 -5.71842670e-01 -4.69553947e-01 8.89084935e-02
-1.01959884e-01 -6.72197878e-01 6.06867790e-01 4.62226421e-01
-2.18713194e-01 6.35388494e-01 -1.26111519e+00 1.24174094e+00
1.22962725e+00 -3.47623557e-01 -2.89626151e-01 1.44803464e-01
7.41044283e-01 -7.01013863e-01 -3.72817338e-01 1.70549959e-01
4.03124124e-01 -9.15098190e-01 1.00695562e+00 -6.94274783e-01
7.91419268e-01 -1.67712912e-01 -3.99004407e-02 -1.21412790e+00
-8.51765990e-01 -1.08186603e+00 5.89936376e-01 7.96629310e-01
5.58695793e-01 -4.48217601e-01 1.30749977e+00 1.57293439e+00
3.04072231e-01 -7.47958481e-01 -7.59330750e-01 -9.47221458e-01
3.31024855e-01 -3.64062756e-01 9.42081392e-01 7.03503907e-01
4.11095768e-01 -8.93270224e-02 -2.08432883e-01 8.23764279e-02
3.76699239e-01 -8.30503404e-02 8.62353027e-01 -7.04185069e-01
-7.74746358e-01 -5.18977463e-01 -8.45424384e-02 -6.12682998e-01
-2.79318184e-01 -3.30834091e-01 2.41597295e-02 -1.32914650e+00
1.64793685e-01 -9.55052733e-01 2.20812410e-01 2.64829844e-01
-5.28251290e-01 1.32849440e-01 6.08542383e-01 -5.01494445e-02
-5.43607712e-01 7.77330324e-02 1.10436571e+00 2.42054731e-01
-6.56462610e-01 -1.18881039e-01 -1.28393257e+00 4.45945531e-01
7.95729578e-01 -5.84827840e-01 -5.51830828e-01 -5.01645625e-01
7.06388474e-01 3.43524724e-01 -1.60204366e-01 -9.46354926e-01
3.54615003e-01 -8.78783584e-01 2.70404872e-02 7.52123296e-02
1.79166645e-01 -2.98112124e-01 8.28003407e-01 6.04800761e-01
-2.10659429e-01 4.00602341e-01 3.35603148e-01 1.28829092e-01
3.51390630e-01 -6.08389437e-01 4.81661111e-01 -3.27554345e-01
1.25192598e-01 -1.35762796e-01 -1.05134213e+00 4.79072273e-01
1.18207300e+00 -6.13147438e-01 -3.76953512e-01 -6.43107951e-01
-4.34559762e-01 1.87879190e-01 8.96021485e-01 4.34231274e-02
1.48903891e-01 -1.05598092e+00 -6.48368955e-01 -8.96664187e-02
-3.33060026e-01 -1.55009016e-01 4.31036204e-01 2.43168250e-01
-8.89145494e-01 8.61301422e-02 -3.48248124e-01 1.17284827e-01
-1.13808179e+00 2.17151344e-01 3.12276810e-01 -5.35833597e-01
-3.38354468e-01 1.02606916e+00 2.17662513e-01 -5.76245666e-01
-1.32661194e-01 -3.83294970e-02 1.79877028e-01 -1.73209652e-01
4.93861020e-01 3.29051614e-01 -1.76957309e-01 1.95251271e-01
-3.39851469e-01 3.46024692e-01 -1.84471279e-01 -6.07495964e-01
1.30466652e+00 2.31289238e-01 1.23613458e-02 -1.72102615e-01
1.98355258e-01 3.51374000e-01 -1.33248985e+00 3.91412377e-01
-3.07965726e-01 -7.94334531e-01 -4.83008504e-01 -9.26450372e-01
-5.51056921e-01 2.06127569e-01 -4.16066855e-01 3.97007942e-01
8.95840526e-01 -3.56976449e-01 6.50672436e-01 2.64498025e-01
1.02605844e+00 -9.44748104e-01 1.94321841e-01 6.69463813e-01
4.56150144e-01 -5.81257701e-01 -9.60557896e-04 -7.08849192e-01
-6.79591835e-01 8.45732629e-01 9.98884022e-01 -3.45372945e-01
2.59958487e-02 8.97817850e-01 3.84504274e-02 -1.23002671e-01
-1.14492524e+00 -6.04062900e-02 -3.70886177e-01 7.95111299e-01
1.15753524e-01 1.32511467e-01 -9.58203495e-01 1.18168557e+00
-7.01631486e-01 -1.36067286e-01 1.36521292e+00 9.82864797e-01
-3.82006466e-01 -1.61774302e+00 -9.16159809e-01 4.07491893e-01
-3.82991135e-01 -7.93957859e-02 -8.62776637e-01 5.78278661e-01
2.40708366e-01 1.26862204e+00 -1.85989559e-01 -5.35507083e-01
3.11357558e-01 -2.30015650e-01 7.60652602e-01 -8.74029160e-01
-9.98809397e-01 -2.87669241e-01 4.66568589e-01 -4.30994838e-01
5.96609414e-01 -8.64084601e-01 -9.81963336e-01 -8.57910573e-01
-3.38016450e-01 5.13757169e-01 1.08397521e-01 1.00083089e+00
4.58206475e-01 4.57467705e-01 3.79889518e-01 -7.00924039e-01
-5.99793851e-01 -4.48539883e-01 -5.28426886e-01 1.79494798e-01
-4.11811769e-01 -7.12391615e-01 -4.49203491e-01 -4.54550266e-01] | [3.521829843521118, 1.5244169235229492] |
21f6a1b0-639f-45de-93ae-cb956021286e | robots-enact-malignant-stereotypes | 2207.11569 | null | https://arxiv.org/abs/2207.11569v1 | https://arxiv.org/pdf/2207.11569v1.pdf | Robots Enact Malignant Stereotypes | Stereotypes, bias, and discrimination have been extensively documented in Machine Learning (ML) methods such as Computer Vision (CV) [18, 80], Natural Language Processing (NLP) [6], or both, in the case of large image and caption models such as OpenAI CLIP [14]. In this paper, we evaluate how ML bias manifests in robots that physically and autonomously act within the world. We audit one of several recently published CLIP-powered robotic manipulation methods, presenting it with objects that have pictures of human faces on the surface which vary across race and gender, alongside task descriptions that contain terms associated with common stereotypes. Our experiments definitively show robots acting out toxic stereotypes with respect to gender, race, and scientifically-discredited physiognomy, at scale. Furthermore, the audited methods are less likely to recognize Women and People of Color. Our interdisciplinary sociotechnical analysis synthesizes across fields and applications such as Science Technology and Society (STS), Critical Studies, History, Safety, Robotics, and AI. We find that robots powered by large datasets and Dissolution Models (sometimes called "foundation models", e.g. CLIP) that contain humans risk physically amplifying malignant stereotypes in general; and that merely correcting disparities will be insufficient for the complexity and scale of the problem. Instead, we recommend that robot learning methods that physically manifest stereotypes or other harmful outcomes be paused, reworked, or even wound down when appropriate, until outcomes can be proven safe, effective, and just. Finally, we discuss comprehensive policy changes and the potential of new interdisciplinary research on topics like Identity Safety Assessment Frameworks and Design Justice to better understand and address these harms. | ['Matthew Gombolay', 'Severin Kacianka', 'Vicky Zeng', 'William Agnew', 'Andrew Hundt'] | 2022-07-23 | null | null | null | null | ['gender-bias-detection', 'gender-bias-detection', 'stereotypical-bias-analysis', 'robotic-grasping'] | ['miscellaneous', 'natural-language-processing', 'natural-language-processing', 'robots'] | [ 2.82945991e-01 7.02204645e-01 -4.02457118e-01 -2.27510426e-02
-2.22563148e-01 -6.64885223e-01 7.02604234e-01 9.95023474e-02
-4.90804553e-01 6.33653879e-01 5.40484667e-01 -3.58404398e-01
-3.36656690e-01 -6.11351669e-01 -9.16276455e-01 -5.37136078e-01
1.40985260e-02 3.51373643e-01 -5.10119677e-01 -3.36511075e-01
4.86347556e-01 5.23285806e-01 -1.72518945e+00 -5.05372137e-03
8.71552467e-01 7.77903736e-01 -1.10746093e-01 1.35640085e-01
3.36375177e-01 8.79471362e-01 -3.78221571e-01 -6.08453751e-01
4.69046772e-01 -1.15144759e-01 -6.68890297e-01 -4.12666440e-01
8.96231830e-01 -4.35044497e-01 -2.12087870e-01 1.05983353e+00
5.31012058e-01 -3.67973685e-01 1.03768396e+00 -1.63710475e+00
-1.14986777e+00 7.07881808e-01 -6.53415084e-01 -6.53371155e-01
4.82149571e-01 6.24818385e-01 4.62340176e-01 -6.28606856e-01
9.10114110e-01 1.96758473e+00 9.28976536e-01 7.81907201e-01
-1.21323442e+00 -1.26683497e+00 5.14834970e-02 7.09982738e-02
-1.06192672e+00 -4.89787728e-01 5.68975091e-01 -1.02459526e+00
3.04648221e-01 1.79766804e-01 7.98618674e-01 1.60601687e+00
4.52185899e-01 1.56677753e-01 1.21481025e+00 -2.92923331e-01
1.79928049e-01 1.38376921e-01 -2.71326035e-01 8.24500024e-01
9.79253590e-01 3.79496485e-01 -5.01120746e-01 -3.01100045e-01
7.71531522e-01 -3.30456257e-01 3.73766012e-02 -5.37899673e-01
-1.29698813e+00 7.04646289e-01 4.18989956e-01 3.18603553e-02
-4.16536808e-01 5.54834843e-01 5.27856052e-01 2.11286210e-02
2.43899196e-01 1.10936928e+00 -2.10130915e-01 -5.17141670e-02
-2.53021475e-02 3.99448067e-01 6.65354252e-01 8.78096342e-01
6.36069834e-01 -3.37174743e-01 7.14809150e-02 8.91754389e-01
1.18806191e-01 6.79990947e-01 2.41834849e-01 -1.85734606e+00
-2.39523612e-02 5.59471488e-01 3.08952779e-01 -1.43839002e+00
-7.04219282e-01 3.32782179e-01 -4.80723292e-01 8.16559613e-01
4.73180056e-01 -4.37940508e-01 -5.95224380e-01 1.91897106e+00
2.22381815e-01 -4.70781386e-01 -5.88601157e-02 1.02150905e+00
6.37677491e-01 2.24945158e-01 7.79862821e-01 1.52739793e-01
1.39786279e+00 -4.10567671e-01 -4.11808759e-01 -3.89063895e-01
8.62994134e-01 -4.82717067e-01 1.23457289e+00 4.63774681e-01
-8.16525042e-01 -2.51519024e-01 -1.12555456e+00 -1.96213514e-01
-4.59476531e-01 -1.07454106e-01 7.84677088e-01 9.38297808e-01
-7.10851014e-01 6.04529142e-01 -1.84316263e-01 -8.11279714e-01
8.12656343e-01 2.44077370e-01 -4.28788930e-01 -1.56787992e-01
-9.70026433e-01 1.40764236e+00 -2.44415011e-02 -1.25380963e-01
-9.52716172e-01 -8.40528190e-01 -8.61898124e-01 -4.54759657e-01
2.40786687e-01 -7.07501352e-01 9.08024013e-01 -1.37134588e+00
-1.18633437e+00 1.15054560e+00 6.25246167e-01 -1.81267425e-01
6.34979546e-01 -4.39280808e-01 -1.69079214e-01 1.23990728e-02
2.62292236e-01 1.44237173e+00 5.08371055e-01 -1.81781197e+00
-4.27844018e-01 -6.89488053e-01 4.64684308e-01 3.30987781e-01
-4.51391757e-01 3.34204316e-01 6.11674845e-01 -2.86343217e-01
-1.98980004e-01 -1.23970532e+00 -6.38245717e-02 7.02997029e-01
-3.25895101e-01 -1.90824330e-01 5.09472787e-01 -5.56646347e-01
4.52911735e-01 -2.36114454e+00 6.68665320e-02 -1.73249871e-01
3.02522659e-01 -1.48228735e-01 -1.95310876e-01 2.82135963e-01
1.04265764e-01 4.72084552e-01 9.17909816e-02 2.87940651e-01
2.30423316e-01 7.89908469e-02 -1.55272745e-02 1.01536644e+00
1.39648244e-01 6.05961144e-01 -1.16686106e+00 -6.09447062e-01
6.08073436e-02 1.29284650e-01 -6.49301052e-01 -3.33011895e-01
-3.00228801e-02 4.61674005e-01 -1.26274168e-01 7.70142674e-01
5.20789862e-01 4.07798231e-01 1.87062427e-01 -2.06009939e-01
-4.14637804e-01 -1.96741283e-01 -4.84452367e-01 1.44671631e+00
-4.49253410e-01 7.02641487e-01 3.51941437e-01 -5.02170563e-01
7.10731328e-01 5.03475033e-02 4.14218485e-01 -7.32324660e-01
3.39003861e-01 3.86286288e-01 3.99610996e-01 -9.65584099e-01
4.37210560e-01 -1.99342340e-01 -2.99595863e-01 1.56938776e-01
-4.16674614e-01 -6.50454164e-01 -3.40054274e-01 -1.51193470e-01
9.16552305e-01 3.13367635e-01 -9.32670906e-02 -5.37715912e-01
-1.51785180e-01 5.52937508e-01 4.97692436e-01 5.84957063e-01
-6.15035951e-01 3.86750221e-01 6.31025016e-01 -2.05333441e-01
-1.31663573e+00 -1.01083624e+00 -3.05894583e-01 1.27955580e+00
4.82504696e-01 1.69254929e-01 -6.48843646e-01 -3.41332525e-01
4.96440560e-01 8.32795560e-01 -7.70280361e-01 -6.89318657e-01
-3.57942581e-01 -3.19646925e-01 9.03099120e-01 2.62203127e-01
2.46265203e-01 -1.09304976e+00 -1.09912932e+00 -3.74221712e-01
1.43149853e-01 -8.14380467e-01 2.23632231e-01 -2.57771254e-01
-3.95893574e-01 -1.18370748e+00 -7.40102410e-01 -8.03827763e-01
7.07767904e-01 3.80446911e-02 5.78889132e-01 -8.78505856e-02
-4.25049603e-01 6.67688727e-01 -2.68686682e-01 -9.81530666e-01
-7.11414635e-01 -4.23284411e-01 5.12726903e-01 -4.75299031e-01
8.13401043e-02 -3.50528568e-01 -7.24705279e-01 4.15458739e-01
-5.58768332e-01 6.79561719e-02 8.71501982e-01 4.48651612e-01
-1.77909583e-01 -6.62595034e-01 8.87535512e-01 -8.25045586e-01
5.17872453e-01 -7.57603645e-01 -1.03189766e-01 1.52607143e-01
-6.26046300e-01 -3.93218398e-01 1.99000731e-01 -7.57316411e-01
-1.10405767e+00 -1.02667652e-01 5.54318309e-01 -3.55207622e-01
-4.29355204e-01 1.92345694e-01 -5.33656701e-02 -1.55917615e-01
1.37124848e+00 -6.76637232e-01 4.17318404e-01 -4.34855856e-02
5.70793092e-01 9.44291055e-01 4.72095340e-01 -9.69094992e-01
5.31386554e-01 7.37985134e-01 -4.02013026e-02 -7.42637694e-01
-2.79863954e-01 2.15542614e-01 -2.93465227e-01 -7.81559706e-01
9.52213049e-01 -1.04260516e+00 -9.52243149e-01 4.74957764e-01
-9.67868924e-01 -6.82567179e-01 -2.16036633e-01 6.20472729e-01
-8.61266553e-01 7.33093917e-02 -4.64524359e-01 -9.80600238e-01
5.98700298e-03 -8.34632635e-01 1.17528570e+00 2.33670637e-01
-7.67562211e-01 -4.42305267e-01 -1.27621517e-01 4.62547362e-01
2.85301417e-01 6.36183143e-01 1.39150608e+00 -3.66969138e-01
-9.23836082e-02 6.16145134e-02 -5.74032187e-01 2.08010554e-01
2.14214906e-01 5.79487011e-02 -9.43296313e-01 9.68510956e-02
-3.23704600e-01 -9.09144402e-01 3.29575896e-01 1.40363559e-01
1.03185284e+00 -5.51000774e-01 -5.08981228e-01 1.65838189e-02
1.13500786e+00 4.18199837e-01 5.63764572e-01 3.95034105e-01
7.27856159e-01 1.52850950e+00 8.17913711e-01 2.58417845e-01
4.10601139e-01 1.59142807e-01 5.04188418e-01 5.69989569e-02
-7.64142126e-02 -2.93887407e-01 6.09748542e-01 3.34377997e-02
-3.76046151e-02 2.69784719e-01 -1.07444739e+00 5.80778241e-01
-1.66651607e+00 -9.26424444e-01 -1.54965743e-01 2.00032091e+00
6.99312627e-01 -6.19895116e-04 1.19281627e-01 -2.89353579e-01
8.20462584e-01 -2.46707529e-01 -7.80415833e-01 -6.95313215e-01
-6.37038499e-02 -3.98416370e-01 8.39435101e-01 6.15896061e-02
-7.56558776e-01 7.84877062e-01 6.69909286e+00 4.68491286e-01
-1.04744601e+00 1.34408483e-02 9.52592552e-01 -1.36908114e-01
-4.29350108e-01 -1.19528681e-01 1.18787680e-02 1.62854657e-01
6.21808887e-01 -1.43219039e-01 5.19802332e-01 8.89266729e-01
2.48481706e-01 -3.78297031e-01 -1.38422942e+00 8.81788194e-01
2.48185113e-01 -8.06551099e-01 -1.82733342e-01 -1.46837290e-02
5.93977749e-01 -1.61333069e-01 4.44438607e-01 2.98058510e-01
5.28815508e-01 -1.31205642e+00 1.40627956e+00 2.45073408e-01
1.13227725e+00 -2.48743430e-01 2.82744974e-01 -6.89147562e-02
-2.26036012e-01 -7.31669307e-01 -3.76494139e-01 -4.66820002e-01
-1.81799740e-01 1.73374549e-01 -4.76737529e-01 2.60539874e-02
8.85258794e-01 3.86958331e-01 -3.79314363e-01 7.15812564e-01
1.02458365e-01 9.76205617e-02 -1.50320157e-01 -2.03085795e-01
-1.28813982e-01 -4.26748060e-02 4.52210158e-01 7.96899199e-01
1.29678667e-01 1.13478921e-01 -1.62856564e-01 9.27661359e-01
1.17928483e-01 5.46313357e-03 -1.19945157e+00 -1.97125554e-01
7.04268992e-01 1.13698912e+00 -5.27333558e-01 -8.60062079e-04
-2.98591703e-01 5.29167414e-01 -4.65951562e-02 2.77556419e-01
-7.51915276e-01 -6.49425611e-02 9.58917975e-01 7.73757622e-02
-6.14020765e-01 -2.75759734e-02 -8.11478555e-01 -4.32963997e-01
-3.14861238e-01 -9.16393161e-01 1.04599353e-02 -1.04223680e+00
-1.46592915e+00 -1.88778132e-01 3.87419522e-01 -9.55614209e-01
8.12707096e-02 -1.03182197e+00 -1.38916969e-01 3.15543324e-01
-8.91843617e-01 -1.46002316e+00 -1.64021388e-01 6.44115582e-02
2.12851912e-01 1.07915908e-01 5.77566266e-01 1.24182537e-01
-4.40888137e-01 4.39785093e-01 -1.21903315e-01 -1.56549029e-02
1.08606589e+00 -7.29066789e-01 -2.20647082e-01 -1.63712248e-01
-7.52419591e-01 5.50065756e-01 7.58736372e-01 -9.58503842e-01
-1.53184927e+00 -8.15107286e-01 2.38012910e-01 -6.66307211e-01
7.10848987e-01 -5.50694346e-01 -1.61243081e-01 6.89665616e-01
1.02554681e-02 -5.82182348e-01 4.81634498e-01 2.01307580e-01
-6.15148544e-01 1.51873042e-03 -1.54672229e+00 1.18251264e+00
1.67224848e+00 -4.04324919e-01 -4.41830158e-01 4.07510579e-01
8.21055472e-01 -9.75792632e-02 -7.35599518e-01 5.08287251e-01
1.21977472e+00 -7.20234752e-01 8.30479980e-01 -5.82258642e-01
8.62080395e-01 2.51895338e-01 -1.90472811e-01 -1.17167783e+00
-4.32671428e-01 -3.07177246e-01 7.33425200e-01 1.21910107e+00
4.99310523e-01 -7.67330706e-01 4.45526868e-01 1.07748818e+00
-4.03340906e-01 -5.71877599e-01 -8.57654035e-01 -8.40258121e-01
6.91075981e-01 -1.62631795e-01 3.64170462e-01 1.21819842e+00
5.64481497e-01 -6.92375675e-02 -3.66097651e-02 -3.43999971e-04
4.46545452e-01 -4.06064898e-01 7.28698015e-01 -1.50049102e+00
4.03508902e-01 -7.56269515e-01 -4.23437089e-01 1.21070229e-01
3.10977906e-01 -6.00124896e-01 3.19667041e-01 -1.41411436e+00
3.20906758e-01 -8.39274466e-01 3.13373953e-01 4.55106825e-01
2.96010584e-01 5.28244935e-02 3.20867509e-01 2.57786125e-01
-1.44099683e-01 3.58486265e-01 1.31767321e+00 -3.41614664e-01
1.43474229e-02 -9.77526963e-01 -1.13562417e+00 1.09298992e+00
8.16437960e-01 -3.76557440e-01 -1.38743624e-01 -2.59029478e-01
5.38253546e-01 -2.88341701e-01 8.82504642e-01 -1.10107970e+00
-7.33701587e-02 -7.43066847e-01 3.86860281e-01 4.94751453e-01
4.15301234e-01 -9.88644242e-01 2.31964931e-01 8.04056466e-01
-4.07260805e-01 -1.42030716e-01 2.37267300e-01 2.52671540e-01
7.28030264e-01 -9.39056873e-02 7.24911749e-01 -1.92369908e-01
-4.18808013e-01 -2.44805530e-01 -5.68007052e-01 -5.28157316e-02
1.28337979e+00 -1.08463921e-01 -1.06025922e+00 -4.04765695e-01
-1.55785665e-01 3.06668282e-01 1.11752403e+00 7.51882553e-01
2.12471873e-01 -1.17219281e+00 -6.92348003e-01 -3.84862125e-01
3.97695094e-01 -2.35110030e-01 1.24529257e-01 7.41087496e-01
-6.64490283e-01 -2.40810812e-02 -7.11535931e-01 -1.42689377e-01
-6.95643663e-01 8.17659199e-01 2.28850797e-01 8.62022281e-01
-3.33137780e-01 6.31642461e-01 7.35221684e-01 -5.11120021e-01
2.51526326e-01 -2.21425772e-01 -7.18878731e-02 1.85111716e-01
2.44174786e-02 8.37368190e-01 -7.21548140e-01 -7.82118201e-01
-2.90094286e-01 6.72327220e-01 3.75991702e-01 -2.38171354e-01
9.74586785e-01 -6.56347647e-02 -3.17932993e-01 4.90552336e-01
7.61395335e-01 3.67455408e-02 -1.23676801e+00 7.04997420e-01
-9.21234190e-02 -3.60341966e-01 -5.73568642e-01 -1.06559813e+00
-5.35165787e-01 4.74971294e-01 8.37696135e-01 -2.73110047e-02
6.36997163e-01 3.17546368e-01 3.99433881e-01 3.02375525e-01
7.20514119e-01 -1.55587494e+00 1.88328370e-01 2.40434334e-02
1.42520893e+00 -9.71787393e-01 1.54186441e-02 -5.70973516e-01
-7.47317433e-01 5.11723459e-01 1.10767007e+00 -1.84452921e-01
4.47103202e-01 -2.11475678e-02 1.23022012e-01 -3.22427094e-01
-3.50987613e-01 -3.86243723e-02 -3.49025875e-01 1.06211698e+00
1.53735563e-01 3.96697521e-01 -6.87399387e-01 4.77295965e-01
-4.38719481e-01 -2.32140664e-02 6.70514166e-01 8.23143065e-01
-4.37136620e-01 -5.91023386e-01 -8.03585410e-01 4.27969396e-01
-2.32185185e-01 3.87720495e-01 -9.91221726e-01 7.67351747e-01
8.02508950e-01 1.05992854e+00 2.00406805e-01 -5.36561787e-01
3.39036912e-01 9.00970995e-02 5.98731458e-01 -4.09647286e-01
-4.98942316e-01 -6.11005902e-01 6.26508117e-01 -5.02941012e-01
-3.02755624e-01 -8.72835934e-01 -1.24324358e+00 -2.60286480e-01
-7.41086751e-02 -5.08052111e-01 9.41875935e-01 5.90964496e-01
5.32899439e-01 -4.73016500e-02 2.67752111e-01 -1.21821046e+00
-3.83713067e-01 -8.69714737e-01 -4.27122712e-01 5.62920809e-01
4.67941202e-02 -1.10831499e+00 -3.56819779e-01 -1.05433561e-01] | [12.888812065124512, 1.4021923542022705] |
0ab1e27d-dedd-43e5-86d5-849091a023fe | design-and-analysis-of-optimized-portfolios | 2210.03943 | null | https://arxiv.org/abs/2210.03943v1 | https://arxiv.org/pdf/2210.03943v1.pdf | Design and Analysis of Optimized Portfolios for Selected Sectors of the Indian Stock Market | Portfolio optimization is a challenging problem that has attracted considerable attention and effort from researchers. The optimization of stock portfolios is a particularly hard problem since the stock prices are volatile and estimation of their future volatilities and values, in most cases, is very difficult, if not impossible. This work uses three ratios, the Sharpe ratio, the Sortino ratio, and the Calmar ratio, for designing the mean-variance optimized portfolios for six important sectors listed in the National Stock Exchange (NSE) of India. Three portfolios are designed for each sector maximizing the ratios based on the historical prices of the ten most important stocks of each sector from Jan 1, 2017, to Dec 31, 2020. The evaluation of the portfolios is done based on their cumulative returns over the test period from Jan 1, 2021, to Dec 31, 2021. The ratio that yields the maximum cumulative returns for both the training and the test periods for the majority of the sectors is identified. The sectors that exhibit the maximum cumulative returns for the same ratio are also identified. The results provide useful insights for investors in the stock market in making their investment decisions based on the current return and risks associated with the six sectors and their stocks. | ['Abhishek Dutta', 'Jaydip Sen'] | 2022-10-08 | null | null | null | null | ['portfolio-optimization'] | ['time-series'] | [-6.06650770e-01 -3.03197414e-01 -1.23647675e-01 6.85141385e-02
-4.30173725e-01 -1.01355255e+00 4.68119085e-01 -7.58536458e-02
-1.75957814e-01 9.15241659e-01 2.06436187e-01 -6.47686541e-01
-8.77077222e-01 -8.22599053e-01 -2.18086123e-01 -8.76052380e-01
-1.62349239e-01 3.02218199e-01 1.54502928e-01 -8.29363018e-02
7.60892272e-01 6.74743235e-01 -1.00569427e+00 -2.21363187e-01
6.19079351e-01 1.54481196e+00 9.00482535e-02 2.00522721e-01
3.76885640e-03 6.23883843e-01 -4.96988505e-01 -7.39142895e-01
7.94143617e-01 -2.07638621e-01 3.32048386e-02 -2.75938481e-01
-3.54710788e-01 -2.27403596e-01 1.29107079e-02 1.22205555e+00
1.09953664e-01 2.68675774e-01 5.16121507e-01 -1.01470339e+00
-2.34324336e-01 7.19486177e-01 -5.14222622e-01 6.41391218e-01
-3.96490067e-01 2.89749242e-02 1.31394005e+00 -8.59588504e-01
1.00713916e-01 6.50442898e-01 1.96109980e-01 -1.13008872e-01
-5.52697480e-01 -8.23799253e-01 7.68164992e-02 -1.66395716e-02
-9.01443839e-01 -5.34939431e-02 6.52233005e-01 -6.77191436e-01
7.11593866e-01 2.11989224e-01 7.81768322e-01 7.93007836e-02
7.39050925e-01 1.86796069e-01 9.85288918e-01 8.00438374e-02
2.61049032e-01 3.31891716e-01 -2.63311028e-01 -2.93946236e-01
8.51955831e-01 1.79460987e-01 -7.71227032e-02 -4.55455296e-02
4.74625170e-01 1.01057336e-01 -3.54586579e-02 1.52989477e-01
-9.84770119e-01 8.60296428e-01 7.06751198e-02 3.58859390e-01
-8.50633740e-01 -2.47954786e-01 7.66153559e-02 4.06813949e-01
2.83092111e-01 5.54318190e-01 -4.28078324e-01 -3.42373475e-02
-1.07390451e+00 3.88670504e-01 7.49721348e-01 5.20911276e-01
1.45354137e-01 4.28741843e-01 -1.92148671e-01 1.26430869e-01
4.16004807e-01 7.63650000e-01 4.09280181e-01 -9.72459137e-01
7.76070178e-01 4.27674890e-01 4.80256200e-01 -1.08171368e+00
3.12513970e-02 -8.34925294e-01 -4.64856863e-01 2.46152818e-01
3.00487280e-01 -4.65846777e-01 -2.24138200e-01 1.18451881e+00
-1.28003865e-01 1.15483426e-01 4.56805259e-01 4.29673672e-01
-1.44291297e-01 9.70562518e-01 -3.70066732e-01 -7.22438872e-01
1.01998627e+00 -7.72451580e-01 -4.25689250e-01 -1.07986227e-01
-1.08598739e-01 -7.00870574e-01 1.60063624e-01 4.06852007e-01
-1.09778249e+00 -8.10899585e-02 -9.80727017e-01 1.26267207e+00
-1.45862922e-01 -1.07178845e-01 3.54190201e-01 4.17442024e-01
-5.49909830e-01 8.87633801e-01 -6.11531019e-01 6.77860141e-01
-1.03314735e-01 3.50941271e-01 2.63013631e-01 6.86513722e-01
-1.18822634e+00 1.01907706e+00 5.37388802e-01 3.55587751e-01
-5.64520299e-01 -6.75667584e-01 -1.40226349e-01 4.60085601e-01
3.94110620e-01 1.71085581e-01 1.02015007e+00 -7.13208318e-01
-1.06006098e+00 -3.01365312e-02 6.84618711e-01 -7.87186265e-01
6.25011623e-01 5.82803460e-03 -4.15183574e-01 -8.15036669e-02
2.22059824e-02 -2.79910803e-01 2.63293356e-01 -5.37612975e-01
-9.91692185e-01 -2.93610424e-01 -1.97042197e-01 1.43119439e-01
-4.61418390e-01 1.96820930e-01 2.26628080e-01 -1.06594908e+00
7.18882009e-02 -9.68821943e-01 -1.73562348e-01 -9.31534410e-01
-1.77601412e-01 1.77154452e-01 4.77353960e-01 -9.66324091e-01
1.38043678e+00 -1.91652179e+00 -3.38685572e-01 7.28743672e-01
-4.83953029e-01 -9.16783139e-03 4.88175452e-01 4.57405895e-01
-3.99111450e-01 1.80624887e-01 -1.49610013e-01 4.90596741e-01
1.78745896e-01 -2.09293097e-01 -8.63680422e-01 2.20410720e-01
2.48850837e-01 5.41233838e-01 -5.24971545e-01 2.62490392e-01
4.67750207e-02 -2.20492586e-01 -7.07169175e-02 3.55360806e-01
-8.04476961e-02 6.79938495e-02 -5.38142920e-01 5.82243502e-01
5.83852530e-01 -7.40860328e-02 9.64230746e-02 8.77825841e-02
-6.35693431e-01 1.98548079e-01 -1.42413330e+00 3.77600454e-02
1.09355161e-02 4.39656138e-01 -2.96105444e-01 -8.24382961e-01
1.24978602e+00 4.21531171e-01 6.24492645e-01 -5.26337802e-01
-6.62782714e-02 6.52974129e-01 2.82581031e-01 -2.01676674e-02
5.01686513e-01 -6.51590466e-01 -7.11013377e-02 7.55354166e-01
-4.53185737e-01 -2.69920155e-02 6.74907744e-01 -2.05466047e-01
5.05504251e-01 -2.58771509e-01 2.64914215e-01 -5.06219685e-01
3.92210066e-01 -4.56036866e-01 7.76727498e-01 9.59003344e-02
6.96986318e-02 2.81819761e-01 1.02255452e+00 -9.27665178e-03
-8.94441903e-01 -9.94571865e-01 -5.88697419e-02 2.92522281e-01
-1.84054077e-01 6.36122763e-01 -2.46713072e-01 -2.36049786e-01
4.55592871e-01 9.50749993e-01 -4.44568694e-01 -1.64833423e-02
-2.11941808e-01 -8.80405009e-01 8.71835947e-02 3.30576926e-01
6.34578764e-01 -1.19575596e+00 -8.76712501e-01 3.45236242e-01
3.14235240e-01 -5.52937925e-01 -4.88926768e-01 1.99482679e-01
-9.18728650e-01 -1.02816927e+00 -1.15684807e+00 -1.55581772e-01
5.59960186e-01 -9.20319743e-03 6.90386415e-01 -3.85602504e-01
4.79190379e-01 -9.86687690e-02 -7.85169471e-03 -9.24705744e-01
-7.41101801e-02 -1.65711701e-01 3.32313627e-02 4.20048803e-01
6.70164824e-03 -2.72233695e-01 -5.13043940e-01 4.42828625e-01
-5.77075005e-01 -4.23136920e-01 5.29724181e-01 4.34144914e-01
5.56358814e-01 7.89848089e-01 9.43192899e-01 -4.02955234e-01
6.49807990e-01 -7.16587067e-01 -1.62351048e+00 4.77627158e-01
-1.02623630e+00 1.47956222e-01 3.05540055e-01 -3.79963219e-01
-1.18054998e+00 -4.03693765e-01 3.62594157e-01 -2.21757859e-01
6.45852208e-01 8.69698644e-01 -2.24880856e-02 1.44825518e-01
-2.22094223e-01 2.72098899e-01 -2.10450605e-01 -5.11363804e-01
-4.21367586e-01 4.25145328e-01 5.91772318e-01 -2.82918215e-01
1.08881688e+00 -4.06570658e-02 1.12290725e-01 -1.65818170e-01
-9.40234840e-01 -2.73344159e-01 -2.58337587e-01 -3.90105039e-01
3.90667170e-01 -6.57797158e-01 -5.89751363e-01 5.92266917e-01
-4.59817827e-01 1.49133682e-01 -4.53880519e-01 9.37694728e-01
-3.22123230e-01 -1.32645279e-01 -2.12688491e-01 -1.26414192e+00
-6.34320855e-01 -1.23432958e+00 4.62800972e-02 5.81923187e-01
-1.83024764e-01 -9.08006012e-01 7.24055395e-02 7.59536633e-03
6.35147333e-01 7.11408496e-01 1.00767159e+00 -9.39521909e-01
-7.99681842e-01 -4.27322984e-01 -1.19761787e-01 7.10446715e-01
3.79193604e-01 3.67770344e-01 -1.46940142e-01 -3.78458112e-01
5.22045672e-01 2.95690387e-01 3.90830874e-01 6.03780508e-01
3.94009560e-01 -6.75597072e-01 3.01434577e-01 3.71586680e-01
1.70391190e+00 1.20683467e+00 5.99441767e-01 9.71498787e-01
-1.41695336e-01 8.22594702e-01 9.54780340e-01 7.70463705e-01
-5.05471230e-02 2.27104723e-01 4.53850865e-01 6.06603026e-01
7.46890008e-01 2.13177949e-01 5.54856598e-01 9.85829115e-01
-6.16330281e-02 -5.82411233e-03 -7.96220005e-01 6.93219423e-01
-1.33496082e+00 -1.14781761e+00 2.42632061e-01 2.46975064e+00
4.61307973e-01 7.10017204e-01 1.91076532e-01 3.37790936e-01
6.29872561e-01 2.34671354e-01 -6.84826732e-01 -2.91546047e-01
-3.02479804e-01 8.09485614e-02 9.31255400e-01 1.92573577e-01
-6.91937327e-01 1.96728200e-01 5.73442793e+00 3.78181010e-01
-1.25499582e+00 -6.43752337e-01 1.07095063e+00 -1.41103655e-01
-5.64207971e-01 3.08742523e-01 -1.02965641e+00 9.64509666e-01
1.22538388e+00 -1.08841395e+00 2.61617243e-01 9.21788037e-01
4.40242976e-01 -2.67293811e-01 -4.09376353e-01 3.32213759e-01
-4.28296298e-01 -1.47568953e+00 -3.43715474e-02 4.64158982e-01
1.14351439e+00 -2.13231117e-01 4.47123319e-01 -2.08650567e-02
6.84343278e-02 -6.47951663e-01 1.05281210e+00 1.06682241e+00
1.36718139e-01 -1.33818877e+00 1.17354345e+00 1.64947972e-01
-1.20163393e+00 -6.43737078e-01 -3.25151831e-01 1.91529244e-01
2.73293734e-01 7.62045324e-01 -3.38353276e-01 4.73233610e-01
6.03451312e-01 2.34434754e-01 -1.97629496e-01 1.27782381e+00
6.75355690e-03 5.56295812e-01 -2.56502688e-01 -2.68037260e-01
4.74210650e-01 -9.32328701e-01 4.99543548e-01 5.74714065e-01
9.76161540e-01 -4.13829880e-03 -4.57494885e-01 7.00112283e-01
-2.81924829e-02 1.08333178e-01 -4.25027311e-01 -4.62327689e-01
6.33561194e-01 8.16157162e-01 -6.29712582e-01 3.22915278e-02
-2.80931056e-01 -7.42233172e-02 -4.80719268e-01 1.58754230e-01
-5.30387938e-01 -7.03897834e-01 4.94056165e-01 5.70089966e-02
4.56777990e-01 -1.33776113e-01 -5.84289670e-01 -6.22830927e-01
2.06423238e-01 -9.28020298e-01 3.10242921e-01 -3.70994836e-01
-8.57865393e-01 2.29649782e-01 2.73800164e-01 -1.27464843e+00
-5.99816084e-01 -5.37310541e-01 -1.10483789e+00 1.22935557e+00
-1.48157954e+00 1.33863222e-02 3.88208002e-01 9.43564158e-03
2.06254900e-01 -8.93919826e-01 1.95979133e-01 6.42502159e-02
-6.67966425e-01 2.16283932e-01 8.78668606e-01 1.25124037e-01
-1.55095473e-01 -1.22895110e+00 4.16926086e-01 8.98776889e-01
-2.07253471e-01 5.92175245e-01 6.07305408e-01 -9.63571608e-01
-9.87071037e-01 -8.14591169e-01 9.06710863e-01 8.54866728e-02
1.15471458e+00 4.06873941e-01 -7.47172356e-01 5.73251545e-01
2.91823521e-02 -7.31481612e-01 4.22111928e-01 -6.01906717e-01
4.52466488e-01 -6.05664074e-01 -1.06116652e+00 4.72543299e-01
-2.61916459e-01 -6.75521567e-02 -5.21523476e-01 2.56236121e-02
3.25781494e-01 -8.79980475e-02 -1.20880997e+00 3.97797734e-01
7.75562704e-01 -8.58698845e-01 7.17966557e-01 -3.87937695e-01
3.72320041e-02 9.78483073e-03 -1.86712965e-01 -1.26576328e+00
-2.45619237e-01 -6.87511027e-01 -6.06851876e-02 1.40065968e+00
6.60944223e-01 -1.02491307e+00 8.88064921e-01 8.94086421e-01
3.91538143e-01 -9.53439415e-01 -1.13535202e+00 -1.03600705e+00
2.70730653e-03 -1.74660981e-02 1.10898411e+00 4.82855111e-01
-6.47796094e-01 -3.28176707e-01 -3.17352861e-01 6.59085885e-02
8.83241594e-01 4.84987527e-01 3.51942658e-01 -1.18544281e+00
-1.30652547e-01 -8.78023028e-01 9.15548280e-02 -1.48233771e-01
-2.15502456e-01 -4.89979714e-01 -3.29450428e-01 -1.14538884e+00
-4.17519286e-02 -2.81205356e-01 -7.63607979e-01 6.14794120e-02
5.48666678e-02 -4.94706869e-01 5.74205518e-01 4.16209996e-01
3.42135191e-01 4.68486428e-01 9.26226616e-01 -9.00685713e-02
-1.02368416e-02 7.87672818e-01 -9.74780500e-01 6.30028248e-01
9.88291264e-01 -4.46425766e-01 -3.64732891e-01 1.42004147e-01
4.78900552e-01 3.97660494e-01 -1.30947635e-01 -8.44255090e-01
2.39429306e-02 -9.46220160e-01 1.58459663e-01 -1.05678105e+00
1.78431720e-02 -8.10037971e-01 9.57701027e-01 6.16084039e-01
-2.08765671e-01 7.86347210e-01 9.43158716e-02 2.28338644e-01
-4.91916716e-01 -8.38496983e-01 7.62766480e-01 -2.16264367e-01
-2.04242855e-01 2.40033820e-01 -1.64054409e-01 1.06272824e-01
1.34704757e+00 -2.30978951e-01 9.81026590e-02 -5.44895113e-01
-3.85495216e-01 5.07981420e-01 7.38001317e-02 1.74382165e-01
4.99610364e-01 -1.37582552e+00 -8.70549142e-01 -1.48399502e-01
-4.88433838e-01 -3.58518869e-01 -1.32192209e-01 5.62433600e-01
-6.50661409e-01 7.46011913e-01 -2.99389690e-01 2.58336961e-01
-6.17483675e-01 1.05252780e-01 5.89034140e-01 -8.29735935e-01
-9.59450006e-02 4.69290972e-01 2.11962685e-01 3.78424227e-01
7.06094056e-02 -2.01843768e-01 -4.80261505e-01 7.24500179e-01
6.66283071e-01 7.19354451e-01 1.91383302e-01 -5.60727835e-01
-2.07043856e-01 5.50662279e-01 1.94171861e-01 -3.56911063e-01
1.75486004e+00 1.44084588e-01 -2.33740956e-01 5.07950425e-01
9.86158907e-01 6.52827546e-02 -1.45690250e+00 3.84417474e-02
7.89988041e-01 -5.36401987e-01 -3.36371176e-02 -6.10769331e-01
-1.57095587e+00 4.60389733e-01 3.05137634e-01 2.61464745e-01
9.63399768e-01 -6.17977977e-01 6.89964771e-01 1.54699013e-01
1.37245864e-01 -1.15377283e+00 -5.83144315e-02 5.02226412e-01
1.10806155e+00 -5.57520270e-01 1.43515855e-01 3.51334870e-01
-7.41267443e-01 1.16248512e+00 1.52827486e-01 -2.37003326e-01
9.68772411e-01 1.19824044e-01 -1.02905244e-01 1.25371724e-01
-6.72690094e-01 4.17007595e-01 5.37476957e-01 3.47401388e-02
-2.89342739e-02 2.09213823e-01 -4.99447584e-01 8.17061186e-01
-4.07544732e-01 -4.21115160e-01 7.34829724e-01 8.53862464e-01
-5.54343760e-01 -7.26798534e-01 -6.27211630e-01 8.63289654e-01
-1.01499164e+00 -8.82971361e-02 -1.10025451e-01 7.18385279e-01
-4.18816745e-01 4.50220287e-01 2.02803478e-01 1.12383161e-02
6.31584525e-01 -1.72518000e-01 -2.33099893e-01 -2.71614850e-01
-6.43778384e-01 6.63185418e-02 -2.30322942e-01 3.03565234e-01
5.86696379e-02 -1.17927027e+00 -1.22740757e+00 -4.78693902e-01
-6.34759605e-01 6.06288433e-01 6.41486168e-01 8.81246865e-01
-2.37406641e-01 1.98592857e-01 1.33681285e+00 -5.70903659e-01
-1.36838293e+00 -7.27545202e-01 -1.13426363e+00 -1.74347684e-01
1.39628619e-01 -6.35597706e-01 -7.98038661e-01 -4.35944974e-01] | [4.632765293121338, 4.09143590927124] |
6b466ee9-b3f3-4dc8-9480-7537973d9232 | high-resolution-boundary-detection-for | 2211.02419 | null | https://arxiv.org/abs/2211.02419v1 | https://arxiv.org/pdf/2211.02419v1.pdf | High-Resolution Boundary Detection for Medical Image Segmentation with Piece-Wise Two-Sample T-Test Augmented Loss | Deep learning methods have contributed substantially to the rapid advancement of medical image segmentation, the quality of which relies on the suitable design of loss functions. Popular loss functions, including the cross-entropy and dice losses, often fall short of boundary detection, thereby limiting high-resolution downstream applications such as automated diagnoses and procedures. We developed a novel loss function that is tailored to reflect the boundary information to enhance the boundary detection. As the contrast between segmentation and background regions along the classification boundary naturally induces heterogeneity over the pixels, we propose the piece-wise two-sample t-test augmented (PTA) loss that is infused with the statistical test for such heterogeneity. We demonstrate the improved boundary detection power of the PTA loss compared to benchmark losses without a t-test component. | ['Jian Yang', 'Jue Hou', 'Feifei Wang', 'Deqiang Xiao', 'Tianyu Fu', 'Jingfan Fan', 'Hong Song', 'Danni Ai', 'Jiaan Luo', 'Saining Zhang', 'Hanchao Yan', 'Yuhao Wei', 'Yuhang Li', 'Jinhua Su', 'Yucong Lin'] | 2022-11-04 | null | null | null | null | ['boundary-detection'] | ['computer-vision'] | [ 5.68615079e-01 2.09309459e-01 -1.82443738e-01 -6.09992921e-01
-9.24557030e-01 -3.76680166e-01 1.79400414e-01 3.98663193e-01
-4.23519820e-01 7.29595184e-01 -1.48548096e-01 -3.71008515e-01
-1.74362704e-01 -6.35420561e-01 -3.85666281e-01 -1.02763331e+00
-6.61081225e-02 2.66560130e-02 3.33422452e-01 3.64894420e-01
7.15451837e-02 3.49310607e-01 -1.07894766e+00 1.92357704e-01
1.50667465e+00 1.32205224e+00 8.80317912e-02 2.96232611e-01
-9.15517434e-02 4.00175840e-01 -4.70279783e-01 -4.57573026e-01
4.40477669e-01 -6.36633873e-01 -5.22482216e-01 -3.26415561e-02
3.58805507e-01 -3.78976434e-01 -1.82152912e-01 1.12946761e+00
7.30953038e-01 -9.39241517e-03 7.68685877e-01 -1.05377281e+00
-2.84219891e-01 4.63273942e-01 -7.87405670e-01 2.94839829e-01
-2.32761830e-01 3.08953911e-01 9.46657062e-01 -6.00571275e-01
5.22220969e-01 6.40984893e-01 1.06895316e+00 3.65658253e-01
-1.17727232e+00 -4.56127465e-01 -2.50622630e-01 -4.95978929e-02
-1.29032767e+00 7.17659295e-02 8.51728201e-01 -6.01506472e-01
1.70218363e-01 3.65349978e-01 5.71341634e-01 7.40777314e-01
5.06362259e-01 9.52879786e-01 1.14047945e+00 -1.42259315e-01
1.34615675e-01 -8.18769112e-02 -1.22891277e-01 7.29100049e-01
3.75880539e-01 -1.30837545e-01 1.41325638e-01 -4.28947806e-02
1.06517208e+00 -5.86861707e-02 -4.54948574e-01 -4.28550869e-01
-9.68570590e-01 7.21128702e-01 4.88975853e-01 2.31659085e-01
-2.41988748e-01 -1.03883989e-01 6.12509549e-01 -5.40387444e-02
6.67175949e-01 4.75332022e-01 -2.99663872e-01 3.47387046e-02
-1.26012480e+00 8.43825378e-03 3.51526707e-01 4.06454176e-01
5.10439157e-01 -2.89951473e-01 -8.80234957e-01 8.43647122e-01
7.50231743e-02 1.87551379e-01 4.94149238e-01 -1.06472969e+00
2.00318962e-01 7.58134186e-01 -1.29727963e-02 -8.77745628e-01
-4.92476821e-01 -7.69660950e-01 -9.42255259e-01 2.53896564e-01
6.90633237e-01 -2.30922461e-01 -1.07118416e+00 1.69850516e+00
6.16504252e-01 4.67205569e-02 -4.13923591e-01 9.18843269e-01
6.80611312e-01 1.07944950e-01 -1.80996284e-02 -1.51770830e-01
1.09046686e+00 -7.07315564e-01 -8.57809782e-01 2.90050879e-02
7.85813868e-01 -4.22865838e-01 1.16216242e+00 7.81813785e-02
-1.35013354e+00 -3.08188915e-01 -1.07279050e+00 -5.07055260e-02
-2.67474782e-02 -1.59894451e-01 4.36907291e-01 6.39296055e-01
-8.28103960e-01 9.09162760e-01 -8.34517777e-01 1.04480058e-01
1.04144263e+00 3.26989710e-01 -1.50787354e-01 7.06931949e-02
-1.06742525e+00 6.17890894e-01 -9.05039608e-02 2.59557784e-01
-5.16614795e-01 -1.16876733e+00 -8.08388531e-01 -2.35763360e-02
2.60853857e-01 -6.23662591e-01 9.37243521e-01 -9.41092789e-01
-1.45633948e+00 1.21537173e+00 1.98641479e-01 -4.01665688e-01
1.27328134e+00 5.69379330e-02 4.03652973e-02 3.19230109e-01
3.18717927e-01 5.55359602e-01 5.99691927e-01 -1.14814425e+00
-5.05775332e-01 -5.18400788e-01 -3.16542834e-01 2.60378304e-03
-6.02062121e-02 -2.75820941e-01 -2.94578165e-01 -9.67299759e-01
-9.45176056e-04 -6.25515938e-01 -3.29270840e-01 5.94810307e-01
-5.65615296e-01 1.58192053e-01 8.79881263e-01 -7.56936967e-01
1.22062278e+00 -2.16203284e+00 -4.25175011e-01 2.92715460e-01
4.04142559e-01 2.12308690e-01 1.26269802e-01 -3.13026190e-01
1.84952453e-01 1.42922938e-01 -8.92368793e-01 -1.85697749e-01
-1.53004348e-01 -3.41406651e-02 1.84644699e-01 7.11028874e-01
3.71843427e-01 9.78194773e-01 -9.21070218e-01 -7.40334511e-01
1.42248034e-01 6.50914788e-01 -5.19702613e-01 1.44963667e-01
3.80669683e-02 7.40757465e-01 -4.51719075e-01 4.02254403e-01
8.30240130e-01 -3.63271326e-01 -1.99993819e-01 -1.96574584e-01
-5.29329851e-02 1.11248307e-01 -6.75601482e-01 1.50542808e+00
-1.60214081e-01 5.76426506e-01 4.97925490e-01 -1.02080834e+00
8.17331195e-01 1.74891427e-01 8.72661114e-01 -7.31766224e-01
3.24956268e-01 3.34863663e-01 2.04038396e-01 -4.74728376e-01
-7.55100623e-02 -6.10435545e-01 2.65781611e-01 9.20445397e-02
-4.11364079e-01 -3.18121046e-01 -6.46357387e-02 -1.91852078e-01
1.06752479e+00 -9.46651697e-02 6.79931492e-02 -5.14928699e-01
4.43026483e-01 -1.76633254e-01 9.23575580e-01 6.65987492e-01
-8.68721962e-01 1.06802177e+00 7.49133468e-01 -1.06015228e-01
-9.56406891e-01 -1.12775791e+00 -8.26554894e-01 3.23973238e-01
4.41966802e-01 3.28662813e-01 -9.06548440e-01 -9.84592736e-01
1.65933043e-01 3.30697119e-01 -6.23052955e-01 -2.80926526e-01
-5.10953724e-01 -9.94515359e-01 7.20521450e-01 6.07729495e-01
8.22508812e-01 -7.82875359e-01 -8.14798594e-01 2.32173339e-01
-1.17342420e-01 -1.10184026e+00 -7.79016078e-01 2.89824307e-01
-9.99475360e-01 -1.09055448e+00 -1.04376221e+00 -8.16037178e-01
7.77782619e-01 -1.37745500e-01 1.01918638e+00 2.91106943e-03
-6.87061727e-01 1.31385684e-01 -1.21738590e-01 -2.15413466e-01
-3.85535866e-01 -2.26298288e-01 -5.45232236e-01 1.30776286e-01
-8.43018889e-02 -4.29612815e-01 -1.14319205e+00 4.47774500e-01
-9.66060340e-01 -1.06360316e-02 6.29175186e-01 1.22517264e+00
8.23975027e-01 1.72002479e-01 5.55432856e-01 -8.82784307e-01
3.51627797e-01 -1.73618823e-01 -4.69060361e-01 1.99158832e-01
-6.01983249e-01 -3.34184468e-02 3.39689106e-01 -2.40412176e-01
-1.09169984e+00 -2.98688412e-01 -2.62437671e-01 -2.70354301e-01
1.17944792e-01 4.62575614e-01 -1.16946489e-01 -2.79001981e-01
2.69340307e-01 -7.06429407e-02 3.40129644e-01 -1.70884505e-01
-8.46440718e-02 4.05892760e-01 5.55755377e-01 -3.73715073e-01
2.70096391e-01 7.71895230e-01 2.50478625e-01 -6.77510202e-01
-8.53014410e-01 -4.01152611e-01 -4.55924779e-01 -2.40612477e-01
1.10645926e+00 -4.00530607e-01 -4.45319653e-01 6.50491297e-01
-9.94935036e-01 -7.85210311e-01 -6.11273348e-01 5.17343283e-01
-5.60992062e-01 5.09206891e-01 -7.32569158e-01 -6.30067408e-01
-4.82473046e-01 -1.28965032e+00 1.01255715e+00 2.41625369e-01
-2.42206268e-02 -1.20851541e+00 5.24524786e-02 3.97444665e-01
4.93719727e-01 7.74315298e-01 1.15913355e+00 -3.80856246e-01
-2.25289568e-01 -3.26441407e-01 -3.52605194e-01 6.06781900e-01
2.41990358e-01 4.73732539e-02 -9.56883371e-01 -1.86222628e-01
1.74269006e-01 -1.21175058e-01 9.83435392e-01 1.22739375e+00
1.70054507e+00 8.50324556e-02 -3.58220726e-01 9.49393570e-01
1.23757112e+00 2.80559838e-01 7.83290863e-01 9.45429578e-02
7.03918040e-01 7.09342301e-01 4.26361978e-01 4.46883768e-01
1.12647206e-01 5.75424314e-01 4.00183618e-01 -6.78184628e-01
-2.58792192e-01 -9.83474031e-02 -1.45100176e-01 4.59312916e-01
2.02578247e-01 -1.53925449e-01 -8.49653959e-01 5.69249272e-01
-1.58336127e+00 -7.10414767e-01 -1.35314897e-01 2.47091246e+00
1.09136128e+00 1.85927391e-01 -8.78389850e-02 1.22890934e-01
9.67473626e-01 1.12237096e-01 -9.85129118e-01 -1.60187840e-01
-2.36069903e-01 1.31539792e-01 5.76186061e-01 5.98154485e-01
-1.24811387e+00 4.52584088e-01 6.61853409e+00 1.02564108e+00
-1.28152585e+00 8.26390982e-02 1.30465448e+00 1.93564370e-01
-4.52122509e-01 -3.59598607e-01 -4.64933246e-01 6.21769488e-01
4.43235844e-01 1.84898134e-02 -2.07757145e-01 3.94076794e-01
5.16177535e-01 -1.28432438e-01 -8.91252398e-01 7.73534894e-01
-2.58518606e-01 -1.17804968e+00 -1.49050862e-01 1.59390107e-01
9.21209097e-01 -9.70342159e-02 4.14673448e-01 -1.32830113e-01
-4.97812815e-02 -1.00653303e+00 4.47536379e-01 1.05105765e-01
9.43310857e-01 -5.26960492e-01 7.73783743e-01 -1.03115201e-01
-8.70669663e-01 4.79870774e-02 -5.67915626e-02 2.45622680e-01
1.92900971e-01 1.26820183e+00 -7.35887647e-01 2.52372503e-01
5.36271334e-01 5.74481308e-01 -2.50581384e-01 1.43421113e+00
1.30657861e-02 4.85982746e-01 -1.39387488e-01 4.61271733e-01
3.62206727e-01 -6.61620498e-01 7.54507720e-01 1.09634840e+00
3.63213032e-01 -1.84734479e-01 -5.25379442e-02 1.31067753e+00
-2.62332976e-01 1.33348376e-01 -2.56254792e-01 1.80493951e-01
2.55651802e-01 1.17235005e+00 -1.07905769e+00 -1.59542203e-01
-4.30835277e-01 9.49686289e-01 -3.89279351e-02 2.13192567e-01
-8.82373512e-01 -4.64914471e-01 6.88699186e-01 5.03882945e-01
2.44867623e-01 2.10954517e-01 -1.02988160e+00 -8.49210620e-01
3.09617162e-01 -4.29946125e-01 4.38920647e-01 -2.53289431e-01
-1.53003693e+00 2.45906457e-01 -2.20778272e-01 -1.01124489e+00
4.59949940e-01 -5.86775780e-01 -9.26522195e-01 7.38999248e-01
-1.66959774e+00 -7.05360651e-01 -4.54469591e-01 2.49093696e-01
1.23267055e-01 4.20002669e-01 2.79903948e-01 6.89740121e-01
-7.45754838e-01 8.70975614e-01 2.85860896e-01 2.92253554e-01
7.32537508e-01 -1.46195769e+00 -1.41034842e-01 7.13489234e-01
-6.14123821e-01 2.38002196e-01 5.91048777e-01 -5.64661086e-01
-6.66087151e-01 -1.11149406e+00 4.03488219e-01 -1.30742610e-01
5.56452990e-01 -8.55521932e-02 -1.09052813e+00 3.15994620e-01
-1.62786990e-01 5.37395477e-01 6.88520610e-01 -4.24115598e-01
5.97573183e-02 -2.09403485e-01 -1.74125922e+00 5.52883267e-01
7.29694545e-01 -2.87998885e-01 -1.39553562e-01 4.32015397e-02
5.07581472e-01 -3.45468849e-01 -1.05493462e+00 9.84789431e-01
5.82456529e-01 -1.09287632e+00 7.93552637e-01 -2.51002878e-01
6.92078710e-01 -5.88771738e-02 3.29679877e-01 -9.97712970e-01
-1.70782536e-01 -6.16586089e-01 3.66537988e-01 1.07672596e+00
4.60062116e-01 -7.20068932e-01 1.10331869e+00 8.74885082e-01
-4.27349240e-01 -1.08996499e+00 -1.16953850e+00 -6.17367089e-01
4.83897835e-01 -1.52539670e-01 3.42574000e-01 8.96769404e-01
-1.08942695e-01 -2.94732958e-01 2.61206150e-01 -2.46743768e-01
8.28132451e-01 -7.22553432e-02 2.42389098e-01 -1.25124967e+00
-2.21410379e-01 -1.04572916e+00 -5.82890511e-01 -8.73197675e-01
-1.14557184e-01 -8.68658483e-01 2.62684345e-01 -1.50095952e+00
4.34133083e-01 -6.80973649e-01 -5.27086377e-01 2.38924295e-01
-5.47380745e-01 3.87152821e-01 -3.94109279e-01 3.80465505e-03
-2.15363428e-01 6.89524353e-01 1.87810159e+00 -2.65667856e-01
-2.84322381e-01 1.17825055e-02 -6.05147481e-01 6.82791710e-01
6.18239105e-01 -3.62948745e-01 -2.76630163e-01 -1.02842756e-01
-6.36472926e-02 -1.18685946e-01 4.06296790e-01 -8.51737857e-01
-9.83375162e-02 1.64192662e-01 3.61413360e-01 -3.65447521e-01
-1.29147291e-01 -6.33952141e-01 -2.85204828e-01 6.65407419e-01
-5.47820270e-01 -3.26022089e-01 1.01605602e-01 4.14570630e-01
-3.97202551e-01 -1.61901806e-02 1.34072375e+00 4.16270681e-02
-1.69423491e-01 5.61617613e-01 7.37719759e-02 5.38791060e-01
1.15569890e+00 -4.52916801e-01 -3.26499879e-01 -1.93625987e-01
-4.61639345e-01 2.45741338e-01 5.05125701e-01 -2.03953639e-01
5.88194549e-01 -1.19260299e+00 -6.63899660e-01 1.66775256e-01
-9.99426991e-02 4.20666516e-01 4.36734945e-01 1.60649788e+00
-7.34944344e-01 -3.25995050e-02 -1.80243641e-01 -7.00234234e-01
-9.45578873e-01 2.48668447e-01 7.19142675e-01 -5.62335670e-01
-1.09919655e+00 9.43594635e-01 7.69875705e-01 -2.44296044e-01
1.88447982e-01 -5.80184162e-01 8.23121443e-02 -1.75508469e-01
4.27891254e-01 5.87908924e-01 2.41316911e-02 -3.51331174e-01
-3.69939059e-01 4.99025047e-01 -1.35250995e-02 3.23626585e-02
1.07289803e+00 -1.23461664e-01 -6.74837828e-02 3.26783150e-01
1.36086285e+00 -2.84722418e-01 -1.93877041e+00 2.77104229e-02
-4.98151481e-02 -4.94058669e-01 4.51410443e-01 -8.28105032e-01
-1.41969061e+00 8.92444491e-01 7.65716493e-01 2.31804579e-01
9.61254239e-01 -1.97758928e-01 1.20221555e+00 -4.69352424e-01
-1.84038691e-02 -1.20157874e+00 -4.82590869e-02 5.14984131e-02
4.06202644e-01 -1.49496841e+00 -6.47661462e-02 -5.39497793e-01
-5.97052276e-01 8.12231123e-01 3.37668628e-01 3.46766151e-02
6.88085139e-01 6.51917040e-01 2.21046150e-01 -2.02653036e-01
-1.04394689e-01 -7.50756785e-02 4.55701530e-01 5.76498032e-01
4.47234899e-01 1.14531763e-01 -6.47978485e-01 5.61015546e-01
2.44043797e-01 -4.01699841e-02 2.67153606e-02 7.38313854e-01
-2.35778809e-01 -6.35227740e-01 -7.91972429e-02 7.84141362e-01
-7.93408811e-01 3.19202915e-02 -1.92305192e-01 5.90980947e-01
2.82832086e-01 5.83964407e-01 2.49258488e-01 2.94584804e-03
1.50882944e-01 -3.12216550e-01 4.82045889e-01 -3.25475603e-01
-4.90873724e-01 7.50884861e-02 -3.47964674e-01 -5.99774957e-01
-3.45891625e-01 -5.91902077e-01 -1.41377175e+00 4.36761118e-02
-3.27735811e-01 2.54662931e-02 3.83659482e-01 8.99192631e-01
1.88515559e-01 4.88122195e-01 6.38081014e-01 -7.39498079e-01
-5.31493366e-01 -7.13295162e-01 -8.69239986e-01 7.64867842e-01
4.95078295e-01 -7.26991236e-01 -6.59979582e-01 -2.53957689e-01] | [14.60712718963623, -2.265944004058838] |
76f54655-2c5f-4f12-9e75-1b81bd5b4ba9 | patch-based-knowledge-distillation-for | null | null | https://dl.acm.org/doi/abs/10.1145/3503161.3548179 | http://www.muyadong.com/paper/acmmm22_sunzc.pdf | Patch-based Knowledge Distillation for Lifelong Person Re-Identification | The task of lifelong person re-identification aims to match a person across multiple cameras given continuous data streams. Similar to other lifelong learning tasks, it severely suffers from the so-called catastrophic forgetting problem, which refers to the notable performance degradation on previously-seen data after adapting the model to some newly incoming data. To alleviate it, a few existing methods have utilized knowledge distillation to enforce consistency between the original and adapted models. However, the effectiveness of such a strategy can be largely reduced facing the data distribution discrepancy between seen and new data. The hallmark of our work is using adaptively-chosen patches (rather than whole images as in other works) to pilot the forgetting-resistant distillation. Specifically, the technical contributions of our patch-based new solution are two-fold: first, a novel patch sampler is proposed. It is fully differentiable and trained to select a diverse set of image patches that stay crucial and discriminative under streaming data. Secondly, with those patches we curate a novel knowledge distillation framework. Valuable patch-level knowledge within individual patch features and mutual relations is well preserved by the two newly introduced distillation modules, further mitigating catastrophic forgetting. Extensive experiments on twelve person re-identification datasets clearly validate the superiority of our method over state-of-the-art competitors by large performance margins. | ['Yadong Mu', 'Zhicheng Sun'] | 2022-10-10 | null | null | null | acm-multimedia-2022-10 | ['person-re-identification'] | ['computer-vision'] | [ 2.53170311e-01 -1.20550774e-01 -6.69035017e-02 -2.71621615e-01
-6.53422892e-01 -2.59788752e-01 6.42594874e-01 1.35202315e-02
-5.56349933e-01 7.71816015e-01 1.91141814e-01 5.15559912e-01
-2.66020536e-01 -4.83913869e-01 -7.21995294e-01 -7.79059947e-01
1.10024661e-01 3.01187247e-01 2.31499746e-01 -6.23926148e-02
-1.14027083e-01 8.90652686e-02 -1.78192472e+00 -8.76362808e-03
1.04972982e+00 7.95420349e-01 1.27197623e-01 4.06701565e-01
3.11497450e-01 4.36898261e-01 -3.65081847e-01 -7.51275897e-01
3.38355541e-01 -3.62806708e-01 -4.56014395e-01 2.82245368e-01
8.28038573e-01 -3.06878388e-01 -7.26910353e-01 1.03700471e+00
7.20552325e-01 2.99100310e-01 3.35739136e-01 -1.06623268e+00
-7.48139560e-01 3.86526525e-01 -5.58241606e-01 3.65704983e-01
4.58151758e-01 3.62411827e-01 6.90005302e-01 -1.26833451e+00
1.98678598e-01 1.12772524e+00 1.25740230e+00 7.36061454e-01
-1.31607783e+00 -7.17845201e-01 5.43687284e-01 4.26250905e-01
-1.76107419e+00 -5.99601805e-01 7.72847295e-01 -3.80506814e-01
7.65106738e-01 1.24907695e-01 9.84419286e-01 1.30216706e+00
-1.38185471e-01 7.86815405e-01 9.27268565e-01 -2.36873060e-01
-1.92065127e-02 4.38422918e-01 1.64706409e-01 5.64545333e-01
4.16217119e-01 1.60897925e-01 -8.54178846e-01 -1.41590238e-01
5.83737671e-01 5.11419713e-01 -4.46866661e-01 -5.71873844e-01
-1.18046999e+00 4.21286166e-01 2.57412314e-01 2.12876350e-01
-3.22870672e-01 -1.28523678e-01 2.65346527e-01 3.02413255e-01
5.24135411e-01 4.44162935e-02 -3.45711768e-01 -5.50708771e-02
-1.00506079e+00 4.11842287e-01 5.39123416e-01 9.55951631e-01
9.55162168e-01 -1.72219023e-01 -4.60455924e-01 1.03862393e+00
-6.26953617e-02 5.83825588e-01 6.90706372e-01 -4.65212405e-01
3.49055588e-01 5.34426093e-01 7.23160245e-03 -1.02051771e+00
-1.85946643e-01 -7.59339511e-01 -1.12490678e+00 -3.20191324e-01
3.51196676e-01 -6.65842816e-02 -7.35241592e-01 2.00412273e+00
3.89258772e-01 7.75252402e-01 -1.31962836e-01 7.25993872e-01
6.44723713e-01 2.53578335e-01 1.51757523e-01 -3.01431537e-01
1.13728750e+00 -1.06981611e+00 -5.27524531e-01 -3.01383227e-01
-4.25682925e-02 -2.65176237e-01 9.91964281e-01 2.45835364e-01
-9.45359230e-01 -1.01099992e+00 -1.15379798e+00 2.14079067e-01
-2.54174322e-01 -1.59569547e-01 4.67045456e-01 7.49171793e-01
-9.77145493e-01 6.06131554e-01 -4.69678283e-01 -6.00983799e-01
5.62561750e-01 3.03500384e-01 -4.45770383e-01 -1.96897209e-01
-1.27770662e+00 6.02406681e-01 3.01844537e-01 9.88997668e-02
-9.11957860e-01 -1.06268668e+00 -7.48827338e-01 -3.00666783e-04
4.97116804e-01 -1.13098931e+00 9.81339574e-01 -1.16143811e+00
-1.39810824e+00 8.32949877e-01 -3.06745499e-01 -6.44472241e-01
7.81652689e-01 -5.19648373e-01 -6.52188480e-01 9.45800692e-02
1.52206391e-01 5.41768372e-01 1.54861796e+00 -1.21241796e+00
-1.00369859e+00 -4.78696167e-01 -2.20576629e-01 4.32125330e-01
-8.88734162e-01 -6.43778503e-01 -9.22101021e-01 -9.90538657e-01
-1.34251773e-01 -9.21721578e-01 -1.25762805e-01 -5.17450459e-02
1.98777132e-02 -2.06242353e-01 6.56049132e-01 -6.52755022e-01
1.21680570e+00 -2.26838613e+00 1.66842416e-01 7.39626121e-03
3.05684209e-01 4.17877465e-01 -1.42346665e-01 2.28613660e-01
-5.36467955e-02 -4.21837091e-01 -3.34849000e-01 -7.34928012e-01
-5.30193523e-02 1.03221700e-01 -3.96802187e-01 5.91618538e-01
1.71673819e-01 9.54849184e-01 -1.09960306e+00 -2.61156350e-01
3.01162481e-01 7.58231401e-01 -4.09155816e-01 2.03662530e-01
1.97620600e-01 5.81395030e-01 2.00650450e-02 5.70757389e-01
8.18228304e-01 -1.53975457e-01 -9.63694900e-02 -2.10939512e-01
2.22538546e-01 -2.72353142e-01 -1.25302219e+00 1.91177988e+00
-1.56978935e-01 1.00278914e-01 -3.31989676e-01 -9.39366400e-01
8.08375955e-01 2.34301418e-01 5.77083886e-01 -6.32772267e-01
-3.14995617e-01 -5.07202893e-02 -6.32218659e-01 -3.32308710e-01
5.15807569e-01 -1.81194305e-01 -9.33701321e-02 1.28345728e-01
2.25976303e-01 4.58229989e-01 -6.13318905e-02 1.75438628e-01
8.73937964e-01 1.23764768e-01 1.93992585e-01 -1.47667989e-01
7.60576606e-01 -3.69770169e-01 1.04834437e+00 1.13684952e+00
-6.32497489e-01 8.50869715e-01 -1.88756391e-01 -6.60967827e-01
-1.00466037e+00 -1.14229751e+00 -6.03443049e-02 1.09589744e+00
4.30689216e-01 -4.20892954e-01 -7.73707986e-01 -8.75588894e-01
3.52182686e-01 3.14290017e-01 -7.80224264e-01 -5.09912848e-01
-6.16333306e-01 -9.12332773e-01 4.54324245e-01 4.09518659e-01
8.08285356e-01 -6.75083160e-01 -4.11867112e-01 1.56672940e-01
-2.72830606e-01 -1.00706756e+00 -9.39347029e-01 -2.62321681e-01
-7.55943418e-01 -1.08386898e+00 -1.22738159e+00 -8.17799687e-01
7.48638213e-01 7.17060685e-01 1.09819365e+00 -5.89201860e-02
-2.91138679e-01 8.77935886e-01 -3.14124137e-01 -9.18364227e-02
9.13144872e-02 9.61307511e-02 4.84766126e-01 6.64222002e-01
6.37308657e-01 -7.32048750e-01 -8.50378811e-01 2.31746361e-01
-6.53438509e-01 -2.56600995e-02 6.45387650e-01 1.09615529e+00
7.13890731e-01 4.01056767e-01 8.27405751e-01 -8.71457279e-01
4.35552210e-01 -4.60082501e-01 -7.59504288e-02 3.87075186e-01
-1.00242281e+00 -1.84336111e-01 5.99352777e-01 -8.41596603e-01
-1.16058767e+00 1.72741920e-01 1.83252603e-01 -6.39488578e-01
-8.27580914e-02 -2.73886528e-02 -2.58940786e-01 -2.67967954e-02
3.53728086e-01 6.27970576e-01 -5.59418909e-02 -5.56543231e-01
5.22195756e-01 3.83008003e-01 1.04506874e+00 -4.33301717e-01
1.01692057e+00 7.06742108e-01 -5.61413884e-01 -6.98122025e-01
-8.74504328e-01 -6.13328397e-01 -7.82514930e-01 -2.72061288e-01
3.65790188e-01 -1.24186552e+00 -6.70205235e-01 8.26439381e-01
-6.94164336e-01 -5.02516143e-03 -8.71959209e-01 1.86365083e-01
-4.30276453e-01 6.37757123e-01 -2.66925812e-01 -8.23156357e-01
-5.02690136e-01 -6.28430068e-01 9.26753342e-01 5.63578248e-01
1.94832310e-02 -8.35487723e-01 3.21205974e-01 3.18620592e-01
3.69650662e-01 9.73025635e-02 3.91042769e-01 -2.76155472e-01
-2.28514344e-01 -3.33283067e-01 -1.97001979e-01 4.13892388e-01
3.33836377e-01 -6.76285267e-01 -1.13642359e+00 -9.79205370e-01
3.13994251e-02 -1.21492997e-01 1.38617754e+00 3.69229093e-02
9.26128209e-01 -3.24715793e-01 -4.56330419e-01 8.60183001e-01
1.29261875e+00 -2.25228831e-01 3.95272255e-01 3.04397464e-01
8.03503036e-01 4.45843875e-01 4.75836128e-01 6.54871702e-01
6.56522810e-01 7.02156663e-01 9.82633755e-02 6.54953793e-02
-4.61996257e-01 -7.67289460e-01 4.29450393e-01 7.42498517e-01
8.19592625e-02 2.09912643e-01 -5.75055361e-01 8.23475718e-01
-2.26810455e+00 -1.07697022e+00 5.33006489e-01 2.53655243e+00
8.94022703e-01 5.04581481e-02 4.70843405e-01 1.02348797e-01
8.70269835e-01 1.29736617e-01 -1.06406987e+00 6.48951352e-01
-5.02846777e-01 -2.97646020e-02 4.29743171e-01 3.41094047e-01
-1.34652078e+00 8.48269880e-01 5.81111765e+00 8.45698833e-01
-6.98848307e-01 3.20840061e-01 4.31350470e-01 -2.61899829e-01
-8.83142557e-03 -9.26295072e-02 -1.05938041e+00 6.76571786e-01
6.58781946e-01 -3.68192643e-01 4.79548752e-01 6.83979273e-01
-8.64986610e-03 8.75662342e-02 -1.21167016e+00 1.56390226e+00
5.02620816e-01 -1.00522041e+00 2.25301161e-01 -3.19839478e-01
8.27203453e-01 -3.49580765e-01 4.29261088e-01 5.02383530e-01
4.52803224e-02 -7.38792896e-01 7.96819568e-01 1.05714893e+00
8.14957857e-01 -8.58946919e-01 6.37749612e-01 1.93351015e-01
-1.38471997e+00 -4.27346379e-01 -4.84379858e-01 6.54893592e-02
8.73433277e-02 7.96861708e-01 -5.57794094e-01 6.69533610e-01
1.04321873e+00 1.25735867e+00 -9.71180439e-01 1.18368649e+00
3.84690426e-02 4.44570035e-01 -1.01596691e-01 5.26255548e-01
-2.84272879e-01 8.20263997e-02 7.25513399e-01 1.33540606e+00
2.21033007e-01 -7.95012861e-02 1.41773090e-01 5.45734346e-01
4.06174697e-02 -1.16576709e-01 -5.14441431e-01 4.84349281e-01
4.78334099e-01 1.04086578e+00 -1.63679481e-01 -4.76485252e-01
-4.81781006e-01 1.53031051e+00 5.08232832e-01 4.65221375e-01
-6.86780572e-01 -1.36267275e-01 6.96563005e-01 2.24857673e-01
5.85662961e-01 3.14638466e-02 -5.24971485e-02 -1.55991387e+00
3.38315874e-01 -9.05744076e-01 7.58660793e-01 -2.14764699e-01
-1.95502448e+00 4.62915510e-01 -5.72407693e-02 -1.24494338e+00
4.54375632e-02 4.33016531e-02 -2.84627110e-01 6.94552898e-01
-1.68498433e+00 -1.35273266e+00 -5.68756461e-01 1.09820318e+00
6.81128561e-01 -4.68091875e-01 7.38897026e-01 6.00167453e-01
-7.45365798e-01 1.13550234e+00 1.43661529e-01 -3.67126465e-02
1.08407390e+00 -1.20458651e+00 3.70290041e-01 1.02746773e+00
2.00419515e-01 6.77169502e-01 4.96037424e-01 -7.77905583e-01
-1.38001072e+00 -1.39668691e+00 6.84295475e-01 -5.92690706e-01
2.65860170e-01 -3.65344107e-01 -1.09119320e+00 5.25493801e-01
-8.09179095e-04 1.41465262e-01 6.03789806e-01 3.86596769e-02
-6.54512644e-01 -7.65823364e-01 -1.05712092e+00 4.73232061e-01
1.32110596e+00 -5.07224202e-01 -6.47164166e-01 1.97850958e-01
6.37833059e-01 -3.41073312e-02 -6.93333864e-01 4.85268831e-01
6.52713656e-01 -1.06214201e+00 1.23547494e+00 -4.69918162e-01
-3.92056018e-01 -3.20283085e-01 6.30082563e-02 -1.16277587e+00
-6.78320885e-01 -9.99523580e-01 -6.53308272e-01 1.52609932e+00
-9.82926562e-02 -7.90054917e-01 8.26394677e-01 5.38623989e-01
3.18393916e-01 -4.40952599e-01 -1.00328183e+00 -1.00395298e+00
-1.23934008e-01 -2.36892983e-01 7.67742336e-01 7.74395049e-01
-9.19428840e-02 3.38351846e-01 -9.61465001e-01 2.73752391e-01
1.08401990e+00 -2.08029598e-02 9.56325650e-01 -1.24705470e+00
-5.93445003e-01 -2.73895323e-01 -3.52688581e-01 -1.30527890e+00
-1.15200169e-01 -6.65809095e-01 -9.18917824e-03 -9.19511497e-01
6.86178446e-01 -4.25984472e-01 -7.79283941e-01 3.64256978e-01
-6.99928939e-01 3.69673193e-01 3.31947356e-01 5.54821193e-01
-8.76438737e-01 8.65245402e-01 9.51726019e-01 -3.87000084e-01
-2.96956331e-01 1.86054200e-01 -9.04467821e-01 5.43787360e-01
4.48140621e-01 -3.76743376e-01 -4.74620342e-01 -3.11156988e-01
5.07307462e-02 -3.51160079e-01 7.11741567e-01 -1.29905033e+00
5.48931956e-01 2.66126126e-01 4.67133731e-01 -4.56275702e-01
3.48960400e-01 -8.16783547e-01 2.87322372e-01 3.46269161e-01
-1.76538736e-01 -4.05705720e-02 -1.90409902e-03 1.20699728e+00
-1.70218408e-01 1.23397298e-01 9.58862782e-01 -1.24667898e-01
-9.16895628e-01 6.84692383e-01 1.46647841e-01 6.91124350e-02
1.14461696e+00 -3.74015063e-01 8.58017281e-02 -3.84831399e-01
-7.81296968e-01 2.30715856e-01 4.77003723e-01 6.40631855e-01
8.01685393e-01 -1.50714529e+00 -8.33507597e-01 5.64745963e-01
3.05028230e-01 -1.36362240e-01 6.40808642e-01 8.18385541e-01
2.11491510e-01 1.99948594e-01 -1.14137545e-01 -6.55651569e-01
-1.02275229e+00 8.46585870e-01 3.95744771e-01 -2.37746403e-01
-1.07178295e+00 9.14342523e-01 3.75487864e-01 -1.85345903e-01
5.10383010e-01 2.02188790e-01 -1.65880412e-01 -1.47148333e-02
9.26303089e-01 4.55141157e-01 -1.03536181e-01 -6.29376292e-01
-3.87337267e-01 6.76650286e-01 -4.32976961e-01 5.50181530e-02
1.29215229e+00 -6.93600774e-01 2.35442534e-01 5.38417220e-01
8.44870150e-01 -2.00437531e-01 -1.89583373e+00 -8.46036911e-01
-2.28586689e-01 -6.51984036e-01 -3.10563773e-01 -5.72412193e-01
-1.01910651e+00 5.85501969e-01 9.71399903e-01 3.66212502e-02
1.23587942e+00 -1.93794310e-01 1.09452963e+00 1.25906199e-01
4.94163424e-01 -1.24558079e+00 3.07513565e-01 3.12661469e-01
6.39223099e-01 -1.26689327e+00 -7.20318779e-02 -1.13213688e-01
-5.32660842e-01 7.65556335e-01 5.56216061e-01 -1.22604996e-01
7.17323720e-01 -1.36062369e-01 -2.55594999e-01 1.38636276e-01
-5.08714736e-01 -3.45539033e-01 2.92320788e-01 8.55539620e-01
-2.88177669e-01 -9.45424363e-02 3.27962674e-02 9.33211982e-01
-2.76431479e-02 1.52719080e-01 -3.27642485e-02 6.69790983e-01
-2.68009841e-01 -9.43931937e-01 -3.15765947e-01 2.26748705e-01
-7.04825670e-02 -1.50289796e-02 -1.31835535e-01 4.49014217e-01
5.11368454e-01 9.15954053e-01 -1.36222228e-01 -5.74343503e-01
6.18426800e-01 3.56902443e-02 4.35929865e-01 -4.28085536e-01
-5.88249207e-01 -2.48091191e-01 -4.93842721e-01 -5.28911531e-01
-4.98102874e-01 -1.00958204e+00 -5.96963882e-01 -2.19599649e-01
-1.70867056e-01 1.13714831e-02 -9.50692892e-02 8.21636021e-01
6.17574275e-01 3.13480914e-01 6.87581480e-01 -7.11802304e-01
-6.30658269e-01 -7.83495605e-01 -5.59971273e-01 8.87295544e-01
5.82735658e-01 -7.08902419e-01 -2.17407018e-01 3.89757395e-01] | [14.714506149291992, 1.0506998300552368] |
5dd036a8-78ae-4403-8948-2638d8bee95c | regularization-strategy-for-point-cloud-via | 2102.01929 | null | https://arxiv.org/abs/2102.01929v3 | https://arxiv.org/pdf/2102.01929v3.pdf | Regularization Strategy for Point Cloud via Rigidly Mixed Sample | Data augmentation is an effective regularization strategy to alleviate the overfitting, which is an inherent drawback of the deep neural networks. However, data augmentation is rarely considered for point cloud processing despite many studies proposing various augmentation methods for image data. Actually, regularization is essential for point clouds since lack of generality is more likely to occur in point cloud due to small datasets. This paper proposes a Rigid Subset Mix (RSMix), a novel data augmentation method for point clouds that generates a virtual mixed sample by replacing part of the sample with shape-preserved subsets from another sample. RSMix preserves structural information of the point cloud sample by extracting subsets from each sample without deformation using a neighboring function. The neighboring function was carefully designed considering unique properties of point cloud, unordered structure and non-grid. Experiments verified that RSMix successfully regularized the deep neural networks with remarkable improvement for shape classification. We also analyzed various combinations of data augmentations including RSMix with single and multi-view evaluations, based on abundant ablation studies. | ['Sangyoun Lee', 'Sungmin Woo', 'Minhyeok Lee', 'Hyeongmin Lee', 'Junhyeop Lee', 'Jaeha Lee', 'Dogyoon Lee'] | 2021-02-03 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Lee_Regularization_Strategy_for_Point_Cloud_via_Rigidly_Mixed_Sample_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Lee_Regularization_Strategy_for_Point_Cloud_via_Rigidly_Mixed_Sample_CVPR_2021_paper.pdf | cvpr-2021-1 | ['3d-object-classification'] | ['computer-vision'] | [ 8.98732319e-02 4.60056262e-03 1.40015706e-01 -3.03839505e-01
-3.84828538e-01 -4.59498554e-01 5.05475998e-01 3.28209922e-02
-2.65778303e-01 5.08519292e-01 -3.80411632e-02 -2.77166124e-02
-9.20152217e-02 -1.02377999e+00 -1.06373036e+00 -8.95203173e-01
2.24061638e-01 6.35122120e-01 -1.83732167e-01 -3.61379027e-01
1.80924401e-01 1.34851372e+00 -1.64870346e+00 2.54479110e-01
9.79320049e-01 8.91609907e-01 3.26233774e-01 -5.87117597e-02
-6.02755010e-01 -9.16985143e-03 -1.60593718e-01 -1.57324180e-01
7.02757955e-01 2.22597510e-01 -4.53608781e-01 4.00006682e-01
7.05095351e-01 -2.22262949e-01 -9.36329067e-02 9.96062696e-01
4.42543209e-01 1.21216215e-01 7.31757045e-01 -1.31997609e+00
-6.68717086e-01 2.64280736e-01 -1.01088667e+00 -1.74482077e-01
-4.42480147e-01 -5.50224893e-02 5.69983840e-01 -1.20357573e+00
5.19879401e-01 9.67574596e-01 9.56036508e-01 3.67699862e-01
-1.09728384e+00 -7.50581264e-01 6.82145357e-02 -3.78590167e-01
-1.29041064e+00 -1.60714716e-01 1.35539436e+00 -4.42301095e-01
6.77722335e-01 4.08143401e-01 7.88611174e-01 6.52579725e-01
-6.26532808e-02 3.89661819e-01 1.02979827e+00 -2.07651943e-01
-1.00367695e-01 1.85416386e-01 1.43536091e-01 3.48223299e-01
4.66004401e-01 1.12041593e-01 -2.65123285e-02 -2.64577061e-01
1.10770810e+00 3.00260067e-01 -2.83387423e-01 -7.06505716e-01
-1.09069324e+00 5.91674268e-01 6.98832035e-01 2.44858578e-01
-5.10925651e-01 -1.50221884e-01 3.54195505e-01 -6.25026971e-02
6.84559345e-01 2.55452991e-01 -4.77524281e-01 4.91601139e-01
-8.32259595e-01 3.52007180e-01 -2.43588146e-02 1.09283614e+00
9.86121893e-01 5.35590649e-01 5.03748171e-02 9.27308798e-01
3.03496212e-01 6.21925652e-01 3.63955230e-01 -7.26269782e-01
4.95822400e-01 1.17089581e+00 -1.10172089e-02 -1.00417173e+00
-5.71060181e-01 -7.04447627e-01 -1.42355299e+00 5.68693697e-01
2.04522745e-03 1.20587632e-01 -9.38494444e-01 1.56246781e+00
4.94022906e-01 2.60419548e-01 1.57666374e-02 9.02205527e-01
1.16750860e+00 4.97523874e-01 -1.83484361e-01 -2.48000026e-01
1.21044564e+00 -5.36888421e-01 -5.08867741e-01 -9.18056071e-02
5.37651837e-01 -7.70079672e-01 1.20679963e+00 3.15469384e-01
-9.48310733e-01 -6.49482787e-01 -1.00581038e+00 -5.64468466e-02
-2.20423505e-01 4.03916538e-01 6.68323398e-01 3.48924994e-01
-7.76691020e-01 6.97302878e-01 -6.38353765e-01 -1.96743757e-02
7.91618168e-01 4.93099719e-01 -7.99070299e-01 2.35528320e-01
-6.16139174e-01 4.12972540e-01 1.74412936e-01 2.99457163e-01
-3.54245484e-01 -9.47248459e-01 -7.00568259e-01 -8.55585784e-02
-3.57114077e-02 -6.59584939e-01 7.12037921e-01 -1.00782132e+00
-1.19991505e+00 1.04044306e+00 -1.14294402e-01 -2.51377970e-01
4.83009070e-01 -1.84174463e-01 -1.29254699e-01 -3.04713935e-01
-7.65388310e-02 5.87699056e-01 1.09984326e+00 -1.80670917e+00
-1.53101102e-01 -6.91401899e-01 -3.72915626e-01 3.84553015e-01
-2.42066786e-01 -3.94241035e-01 -1.82272807e-01 -7.26118445e-01
6.70356929e-01 -9.32683051e-01 -2.16008484e-01 -2.71929279e-02
-3.60242933e-01 -9.93082225e-02 1.26159263e+00 -4.30746138e-01
7.42435396e-01 -2.16734529e+00 -1.33412883e-01 2.12382987e-01
2.87988454e-01 4.45499897e-01 -2.89007813e-01 1.45759061e-01
-6.40561581e-01 3.15038025e-01 -6.32401764e-01 -5.65636635e-01
-4.05304074e-01 1.27152801e-01 -5.39446533e-01 5.44073105e-01
2.80926287e-01 6.83735490e-01 -3.84417862e-01 -2.03786314e-01
4.70141858e-01 6.59093678e-01 -5.99115312e-01 1.27858341e-01
-3.73403847e-01 5.70613563e-01 -2.89707929e-01 6.95709884e-01
1.59161854e+00 4.61494885e-02 -4.03109193e-01 -5.90089619e-01
-4.18896675e-01 -3.27426881e-01 -1.21514142e+00 1.54662383e+00
-3.52620691e-01 3.70522648e-01 2.91799486e-01 -8.48968685e-01
1.36806273e+00 1.99435264e-01 7.66461015e-01 -2.26423755e-01
1.62991598e-01 1.67536989e-01 -8.12402740e-02 -4.54651326e-01
6.07119322e-01 -2.97594637e-01 4.83136982e-01 1.41280025e-01
-3.46670032e-01 -3.84059787e-01 -3.78556341e-01 -1.42003328e-01
3.29028338e-01 3.09905201e-01 -2.96570025e-02 -2.59881198e-01
5.99144459e-01 5.57189323e-02 6.55934453e-01 2.91416913e-01
1.60748109e-01 1.19407988e+00 1.41891301e-01 -8.03859890e-01
-1.45155072e+00 -8.37159872e-01 -3.70183557e-01 4.42606419e-01
2.82563325e-02 3.31367068e-02 -5.40814936e-01 -4.61493284e-01
1.19877450e-01 4.91382152e-01 -4.07130688e-01 -1.06323324e-01
-7.34105229e-01 -9.63243067e-01 3.06428313e-01 4.89218116e-01
8.93039823e-01 -1.16196311e+00 -9.16188359e-02 -1.84027165e-01
-6.81327134e-02 -9.80269611e-01 -2.23431587e-01 -5.11609297e-03
-1.37339509e+00 -1.02017856e+00 -4.75986123e-01 -8.48643243e-01
1.06238210e+00 5.19883990e-01 1.05230749e+00 3.19522113e-01
2.00565964e-01 -1.94795839e-02 -2.07068041e-01 -6.35390699e-01
-2.58639514e-01 -7.69805014e-02 2.90044516e-01 1.11702494e-01
3.28400701e-01 -8.87844384e-01 -3.48857552e-01 8.69523957e-02
-1.14446712e+00 7.21589848e-02 5.56627274e-01 7.48132050e-01
9.69647765e-01 1.27786538e-02 3.95979017e-01 -8.95673990e-01
4.95008528e-01 -2.28358939e-01 -6.90183461e-01 -7.99821764e-02
-2.74136990e-01 -6.84945360e-02 5.70577145e-01 -1.85366437e-01
-9.39926386e-01 2.90348381e-01 -2.94522882e-01 -9.84971821e-01
-4.40423816e-01 3.42120469e-01 -3.96720618e-01 -4.46120083e-01
5.04389107e-01 4.04162765e-01 4.15452331e-01 -6.16281986e-01
1.78764820e-01 4.71804202e-01 3.21464092e-01 -6.06115758e-01
1.00934172e+00 8.95693302e-01 3.09292406e-01 -1.29833674e+00
-4.04869765e-01 -2.63843626e-01 -9.52494204e-01 -6.91626966e-02
7.06795096e-01 -8.90265226e-01 -7.35012352e-01 6.18383884e-01
-1.48152745e+00 1.41960099e-01 -4.82050776e-01 3.82593393e-01
-2.52879560e-01 4.95227069e-01 -1.27562776e-01 -7.27040350e-01
-5.92623711e-01 -1.12544775e+00 9.45333481e-01 8.98074880e-02
2.77337670e-01 -5.83203554e-01 -4.54012863e-02 4.43312287e-01
2.48268470e-01 4.96189058e-01 9.73560870e-01 -5.73998332e-01
-6.43533170e-01 -2.11615875e-01 -2.52318263e-01 7.49269962e-01
3.04453015e-01 3.67637098e-01 -1.10593128e+00 -3.62241536e-01
2.70657539e-01 -3.98592837e-02 6.97326005e-01 5.82615077e-01
1.56372571e+00 -1.61584318e-01 -2.57734209e-01 1.08064997e+00
1.48874152e+00 1.41068131e-01 7.96463966e-01 3.99554402e-01
1.18583667e+00 4.76607293e-01 3.59737664e-01 3.07092577e-01
2.22943164e-02 4.33864415e-01 9.48686242e-01 -4.08845335e-01
5.36293872e-02 8.57364833e-02 -3.93074378e-02 7.93713331e-01
-5.13805330e-01 4.26171198e-02 -9.52768683e-01 3.59204262e-01
-1.38766825e+00 -6.74806952e-01 -6.48803651e-01 2.36352539e+00
3.10247421e-01 4.86218780e-02 -1.48070619e-01 1.60962403e-01
8.28059316e-01 1.44323289e-01 -6.17308557e-01 -1.21806592e-01
-4.24417943e-01 1.98887542e-01 4.84966099e-01 2.44009495e-01
-1.21267450e+00 7.75848567e-01 5.49706841e+00 7.51502097e-01
-1.31844568e+00 -9.00054127e-02 4.28965002e-01 1.01030380e-01
-4.82483000e-01 -1.98918909e-01 -6.64937139e-01 4.17277455e-01
1.64126962e-01 2.76204646e-01 9.02653188e-02 8.62151504e-01
3.83652180e-01 3.39216650e-01 -7.65444458e-01 1.07039237e+00
-1.20769963e-01 -1.55751586e+00 5.66893280e-01 1.67448491e-01
7.50559151e-01 2.87395000e-01 4.09165472e-02 2.10134573e-02
-1.99352905e-01 -9.33958054e-01 3.55966151e-01 6.77562892e-01
7.88351297e-01 -8.84608984e-01 8.16989779e-01 4.29857522e-01
-1.21433258e+00 2.19948947e-01 -5.87220490e-01 8.88667069e-04
-4.33766544e-02 7.45329976e-01 -5.84890723e-01 8.21619213e-01
6.81934893e-01 7.52578080e-01 -4.68458742e-01 9.23964322e-01
2.25509837e-01 2.85131305e-01 -6.19822860e-01 5.42159498e-01
9.50232223e-02 -1.05780506e+00 9.30867136e-01 7.03592300e-01
4.99455601e-01 8.50528665e-03 -4.03559171e-02 1.10803354e+00
-1.68839142e-01 1.30480617e-01 -1.05598092e+00 1.70013130e-01
5.73366463e-01 1.44747674e+00 -6.24710441e-01 -9.31612551e-02
-3.38420242e-01 3.79904807e-01 2.62799352e-01 3.19163203e-01
-5.18055081e-01 2.22908426e-02 7.27664769e-01 5.72042227e-01
-9.21226572e-03 -3.29298437e-01 -7.79113948e-01 -9.60204601e-01
2.11458325e-01 -7.15766549e-01 -3.39767039e-02 -9.21966910e-01
-1.43145132e+00 8.26763451e-01 -3.78433987e-02 -1.87138224e+00
6.06815219e-01 -4.60673660e-01 -8.55718613e-01 1.07273257e+00
-1.35876739e+00 -1.48067021e+00 -8.12117398e-01 3.20825100e-01
3.27243000e-01 -4.68570888e-01 6.12161517e-01 3.14078748e-01
-6.20971501e-01 2.99832523e-01 8.21982101e-02 -7.64326006e-02
4.19435918e-01 -9.49204624e-01 3.58337581e-01 7.15997696e-01
-1.32958844e-01 6.22340143e-01 4.23628777e-01 -8.32375050e-01
-1.22480762e+00 -1.29752147e+00 3.21469694e-01 -1.81009531e-01
2.56684069e-02 -2.01148987e-01 -1.46095884e+00 7.42590964e-01
1.82584986e-01 3.06018949e-01 3.38559151e-01 -1.07415766e-01
1.13547090e-02 -2.78737903e-01 -1.29671800e+00 4.10612196e-01
9.29151237e-01 -8.49103108e-02 -4.81392473e-01 2.37105191e-01
7.85207510e-01 -5.80217659e-01 -8.10737133e-01 1.01754129e+00
1.22801021e-01 -1.06156945e+00 1.15755463e+00 -4.50931281e-01
4.36500460e-01 -5.73938966e-01 -1.13634065e-01 -1.32154405e+00
-4.43855375e-01 -1.44828930e-01 1.75479159e-01 1.45284724e+00
5.77079132e-02 -6.92001998e-01 1.31057775e+00 3.78681928e-01
-6.50999069e-01 -7.48811066e-01 -9.11995709e-01 -6.63134813e-01
3.79103273e-01 -3.09654802e-01 9.64212656e-01 1.31570303e+00
-8.82737458e-01 3.42922993e-02 -2.59989083e-01 5.19640923e-01
6.56272113e-01 2.14971632e-01 8.82292211e-01 -1.61504269e+00
4.19100583e-01 -3.36523503e-01 -1.73665807e-01 -5.00499904e-01
1.58367380e-01 -1.07174587e+00 -3.74688506e-01 -1.62386227e+00
-1.89893603e-01 -9.86390412e-01 7.54911378e-02 3.97160947e-01
1.48109049e-01 2.91528553e-01 1.23636127e-02 4.98676777e-01
4.37028468e-01 9.70583975e-01 1.65429676e+00 -9.72561613e-02
-4.37962115e-01 1.83886752e-01 -6.01270497e-01 1.04995978e+00
9.82120931e-01 -3.03881049e-01 -4.25553054e-01 -6.68058395e-01
3.04713070e-01 -2.60831565e-01 4.39239413e-01 -1.11586857e+00
-9.02093202e-02 -1.21360421e-01 4.50327903e-01 -1.29597390e+00
4.79623646e-01 -1.34480441e+00 4.74349797e-01 9.46694165e-02
1.07270151e-01 1.86444327e-01 6.82285786e-01 4.17366326e-01
-2.17931181e-01 -2.19199866e-01 9.16763842e-01 -2.86605835e-01
-4.74181235e-01 6.23956382e-01 1.73584983e-01 -3.42047930e-01
9.61587846e-01 -4.52103704e-01 -1.53807133e-01 1.83656096e-01
-6.34486556e-01 -8.53194296e-03 6.55750215e-01 2.61031866e-01
8.59536469e-01 -1.73744011e+00 -7.03848779e-01 6.55608654e-01
-3.18502411e-02 9.66926932e-01 5.16164005e-01 6.17597759e-01
-7.87269175e-01 4.82966527e-02 -4.55782801e-01 -8.33846390e-01
-1.32024002e+00 4.55472142e-01 5.14969349e-01 7.79249072e-02
-9.67778385e-01 7.82893717e-01 3.44020516e-01 -9.80474770e-01
5.31378314e-02 -5.01362920e-01 -5.22940636e-01 6.60881996e-02
1.24410829e-02 2.55672991e-01 4.95689213e-01 -8.08196068e-01
-2.75238067e-01 9.01101708e-01 -7.48496056e-02 4.50427502e-01
1.70594537e+00 2.47425437e-01 -5.96881747e-01 2.04628408e-01
1.04219997e+00 4.53170203e-02 -1.28644431e+00 -2.53185362e-01
-4.63974833e-01 -4.65468526e-01 1.83005586e-01 -2.93532908e-01
-1.42725122e+00 7.87059188e-01 6.38529956e-01 -1.32124340e-02
1.04950488e+00 -4.03959781e-01 6.49986148e-01 3.11428994e-01
1.23805739e-01 -8.21773410e-01 -1.89134717e-01 4.98510092e-01
1.51892209e+00 -1.43105316e+00 2.39728734e-01 -6.12575769e-01
-5.01838923e-01 1.10832489e+00 9.56215262e-01 -4.33414787e-01
8.26880991e-01 2.24399015e-01 -2.66258027e-02 -4.75577354e-01
-1.82442769e-01 2.04473600e-01 3.28653246e-01 5.79858243e-01
3.57767642e-01 -1.10171445e-01 1.53796570e-02 5.13745844e-01
-3.02489251e-01 -1.68253496e-01 5.42698503e-01 8.60221982e-01
-2.20315710e-01 -1.05259740e+00 -7.43937910e-01 6.81406021e-01
8.97150189e-02 -5.30072823e-02 -2.53862739e-01 9.91177142e-01
5.06882071e-01 2.41792142e-01 5.76361835e-01 -4.74617571e-01
5.09102762e-01 -1.25645474e-03 2.85751224e-01 -4.93540466e-01
-6.53774321e-01 1.53318182e-01 -3.74708891e-01 -1.47850260e-01
-5.33337176e-01 -7.22666979e-01 -1.29883802e+00 -2.89951861e-01
-2.33416095e-01 3.37499864e-02 7.82322526e-01 7.36659586e-01
5.61822057e-01 6.24463022e-01 6.67399168e-01 -1.23028660e+00
-2.48666927e-01 -1.10388947e+00 -6.91784203e-01 4.78724122e-01
5.34316003e-01 -8.57276201e-01 -4.74104941e-01 -1.54441427e-02] | [8.211981773376465, -3.566131114959717] |
c08e4101-60d0-4271-91f9-fa18fc94aa30 | movie101-a-new-movie-understanding-benchmark | 2305.12140 | null | https://arxiv.org/abs/2305.12140v2 | https://arxiv.org/pdf/2305.12140v2.pdf | Movie101: A New Movie Understanding Benchmark | To help the visually impaired enjoy movies, automatic movie narrating systems are expected to narrate accurate, coherent, and role-aware plots when there are no speaking lines of actors. Existing works benchmark this challenge as a normal video captioning task via some simplifications, such as removing role names and evaluating narrations with ngram-based metrics, which makes it difficult for automatic systems to meet the needs of real application scenarios. To narrow this gap, we construct a large-scale Chinese movie benchmark, named Movie101. Closer to real scenarios, the Movie Clip Narrating (MCN) task in our benchmark asks models to generate role-aware narration paragraphs for complete movie clips where no actors are speaking. External knowledge, such as role information and movie genres, is also provided for better movie understanding. Besides, we propose a new metric called Movie Narration Score (MNScore) for movie narrating evaluation, which achieves the best correlation with human evaluation. Our benchmark also supports the Temporal Narration Grounding (TNG) task to investigate clip localization given text descriptions. For both two tasks, our proposed methods well leverage external knowledge and outperform carefully designed baselines. The dataset and codes are released at https://github.com/yuezih/Movie101. | ['Qin Jin', 'Ziheng Wang', 'Liang Zhang', 'Anwen Hu', 'Qi Zhang', 'Zihao Yue'] | 2023-05-20 | null | null | null | null | ['video-captioning'] | ['computer-vision'] | [ 3.33875537e-01 -1.52426824e-01 -3.34877908e-01 -5.15430033e-01
-9.60964859e-01 -6.66355133e-01 7.26590216e-01 -3.26806642e-02
-3.03840309e-01 6.47231996e-01 1.19811141e+00 2.46415809e-01
3.39386672e-01 -4.97750431e-01 -5.52458763e-01 -5.03682733e-01
2.92322427e-01 -2.54244924e-01 1.58010185e-01 -3.03294569e-01
1.94945708e-01 -7.76404738e-02 -1.49229431e+00 1.08282685e+00
7.12914467e-01 8.68074834e-01 5.79657197e-01 9.16793227e-01
1.25535518e-01 1.40386569e+00 -8.61873150e-01 -7.58710444e-01
-2.74240136e-01 -9.27513897e-01 -5.97036302e-01 1.26402706e-01
6.07200742e-01 -6.28363371e-01 -6.29864395e-01 1.02469659e+00
5.57987511e-01 3.27828497e-01 5.31224489e-01 -1.16033113e+00
-1.09497893e+00 1.23399174e+00 -9.34230164e-02 9.72470865e-02
7.52393484e-01 1.39393747e-01 1.28861940e+00 -8.42544794e-01
8.26556981e-01 9.81694221e-01 4.11115170e-01 5.78609228e-01
-6.52120769e-01 -3.90500247e-01 3.42493862e-01 6.14776969e-01
-1.23191774e+00 -6.52958214e-01 8.39675665e-01 -4.32442933e-01
7.17931747e-01 7.29578614e-01 5.74370861e-01 1.79292095e+00
-3.85141343e-01 1.12008274e+00 4.54187870e-01 -1.33616909e-01
5.47729097e-02 2.48491183e-01 -7.11724907e-02 2.11036921e-01
-4.36798215e-01 -5.54464221e-01 -8.62863600e-01 3.30461413e-01
6.99724853e-01 -1.82720184e-01 -7.68709123e-01 3.61853868e-01
-1.69315112e+00 7.19864249e-01 2.15845898e-01 3.22518319e-01
-3.22400272e-01 1.30944982e-01 6.19333148e-01 1.96298817e-03
4.55709934e-01 6.31305099e-01 1.81521803e-01 -5.21931887e-01
-9.42430019e-01 3.63645792e-01 5.38783848e-01 1.23103607e+00
-2.87372798e-01 -6.14057072e-02 -8.57503772e-01 9.64804947e-01
-2.42226124e-01 4.80047226e-01 3.32397252e-01 -1.23820877e+00
8.27571273e-01 8.77681896e-02 3.37274224e-01 -1.13985765e+00
-1.74875334e-01 -5.13642848e-01 -7.43027687e-01 -6.21203065e-01
2.44471014e-01 -7.63431266e-02 -3.03081423e-01 1.89624989e+00
-1.97030708e-01 2.08595455e-01 -2.86603137e-03 1.25925362e+00
1.34006906e+00 9.63319659e-01 -4.73935939e-02 -6.83299541e-01
1.45835519e+00 -1.49993455e+00 -1.19526732e+00 -6.03138767e-02
4.21255201e-01 -7.78126359e-01 1.67063665e+00 3.58305007e-01
-1.20865035e+00 -6.93219185e-01 -9.33814466e-01 -2.65518278e-01
7.77373016e-02 3.35507303e-01 3.84360641e-01 1.46196216e-01
-7.63080955e-01 2.77483821e-01 -4.25550938e-01 -4.07133549e-01
4.00131553e-01 -5.03679812e-01 -3.27383965e-01 -1.23658977e-01
-1.46693110e+00 5.97424865e-01 1.34771034e-01 2.99938917e-02
-1.13017285e+00 -6.82803929e-01 -8.74776065e-01 -1.63953289e-01
3.19605321e-01 -6.11290991e-01 1.60200369e+00 -1.04783547e+00
-1.22563505e+00 8.21493030e-01 -3.95269632e-01 -4.30129826e-01
6.13887072e-01 -3.97645533e-01 -6.25756979e-01 6.53813958e-01
1.95866883e-01 5.59625149e-01 8.03712130e-01 -1.23322368e+00
-5.84806204e-01 7.17609897e-02 5.24549603e-01 4.61800128e-01
-8.88105989e-01 4.90648180e-01 -6.56390607e-01 -1.07255328e+00
-2.24266261e-01 -4.53836471e-01 1.02531508e-01 -2.45079353e-01
-6.05212748e-01 -1.49856269e-01 3.30329835e-01 -9.60719228e-01
1.73463011e+00 -2.16637754e+00 2.97798842e-01 -6.34962916e-01
3.73770148e-01 -2.01981038e-01 -2.36549690e-01 7.30811298e-01
1.86587244e-01 7.77834207e-02 -1.63726464e-01 -6.35451674e-01
5.15043698e-02 -1.04973927e-01 -4.23460215e-01 1.25822440e-01
4.68888842e-02 8.31797183e-01 -1.02949286e+00 -5.45460880e-01
-1.37266368e-01 3.89036328e-01 -3.31566244e-01 4.70770448e-01
-4.52033371e-01 6.11508012e-01 -2.61578053e-01 5.20562947e-01
2.81913340e-01 -4.21134263e-01 -1.12582698e-01 -3.90204906e-01
-9.01494771e-02 2.37514406e-01 -8.58390868e-01 1.84019780e+00
-4.05702829e-01 8.83657157e-01 -2.82776266e-01 -5.92921019e-01
7.27871597e-01 6.07038021e-01 1.57474548e-01 -6.00403607e-01
9.83362570e-02 -3.21194649e-01 -4.88677055e-01 -1.28946698e+00
7.23308146e-01 -1.29348719e-02 -4.09494966e-01 2.39355743e-01
-1.49163470e-01 8.48047361e-02 4.00511861e-01 5.57772934e-01
1.16330004e+00 5.10324240e-02 3.44482541e-01 4.05582398e-01
3.91044796e-01 4.66247089e-02 4.00201708e-01 8.26333046e-01
-2.12914333e-01 1.38603425e+00 7.39648044e-01 -6.31858706e-02
-1.09294295e+00 -1.03864455e+00 1.01088375e-01 1.26282334e+00
4.32667971e-01 -8.54279160e-01 -8.54921877e-01 -5.71313202e-01
-6.23531103e-01 9.48800862e-01 -6.85889661e-01 9.81444642e-02
-5.52937508e-01 -3.71028990e-01 4.72542226e-01 5.99889636e-01
4.44722801e-01 -1.29693401e+00 -4.69942003e-01 1.62596524e-01
-9.10012305e-01 -1.62360573e+00 -9.58464324e-01 -6.23214543e-01
-2.26884604e-01 -7.22036362e-01 -9.80384171e-01 -9.16427135e-01
3.22375536e-01 2.94733614e-01 1.05830884e+00 -3.15875202e-01
1.96667761e-01 2.45794609e-01 -9.68802392e-01 -4.34013009e-02
-2.61320770e-01 -1.12691008e-01 -4.67622392e-02 2.19131008e-01
1.01978339e-01 -6.74863398e-01 -5.73335826e-01 2.42437303e-01
-7.85385370e-01 7.43653417e-01 2.31659874e-01 6.23651326e-01
5.05605519e-01 -1.17839076e-01 6.86672688e-01 -6.98233128e-01
9.16765094e-01 -6.12113535e-01 1.55170232e-01 2.97863811e-01
1.15198947e-01 -5.44685304e-01 1.09299219e+00 -8.89222980e-01
-1.10163724e+00 -1.50887087e-01 -2.00397804e-01 -4.71498102e-01
-2.67068416e-01 5.09617150e-01 -5.12274921e-01 6.79977298e-01
6.72714412e-01 4.09099013e-01 -4.00073051e-01 -4.36391979e-01
4.68774348e-01 7.65460193e-01 1.14610028e+00 -1.67895749e-01
4.57736880e-01 3.35669309e-01 -5.61314166e-01 -6.94981456e-01
-1.36680400e+00 -4.71734256e-01 -2.47447670e-01 -7.30284274e-01
1.07855737e+00 -1.14619398e+00 -5.80340683e-01 3.93411487e-01
-1.53428257e+00 -1.66324645e-01 -2.14739367e-02 6.46373570e-01
-7.38671005e-01 2.34575331e-01 -6.62987649e-01 -7.72363722e-01
-1.66701958e-01 -1.01589561e+00 8.87532473e-01 2.79653102e-01
-3.66305292e-01 -6.74442053e-01 -2.12017789e-01 7.75727630e-01
3.34711611e-01 5.56988776e-01 5.53307950e-01 -5.31326234e-01
-2.45844081e-01 -1.66718483e-01 -1.90224737e-01 3.84748399e-01
-1.65476501e-01 -1.82938591e-01 -9.96676147e-01 1.28293827e-01
6.17852993e-02 -5.37353158e-01 8.45637321e-01 3.36752295e-01
1.24336553e+00 -7.19036758e-01 2.10881159e-01 5.30504167e-01
1.03634524e+00 1.17638752e-01 7.06393778e-01 1.77359581e-01
7.82042086e-01 7.99353182e-01 8.75310063e-01 7.62864113e-01
8.84489119e-01 7.93108284e-01 4.26596940e-01 1.03664309e-01
-4.04383123e-01 -6.51957154e-01 7.27243543e-01 1.03886008e+00
-1.43347174e-01 -7.66553998e-01 -7.36533463e-01 6.99202359e-01
-2.12038970e+00 -1.36316550e+00 -3.88620436e-01 1.76018846e+00
9.80386257e-01 1.28382489e-01 2.71210462e-01 -5.89798354e-02
8.54446828e-01 4.92195427e-01 -4.88785475e-01 2.45329048e-02
-4.90615040e-01 -4.18451160e-01 -2.09912762e-01 3.97455692e-01
-1.24612617e+00 9.18687344e-01 4.86537981e+00 9.80232179e-01
-8.09751332e-01 3.82340491e-01 5.24505496e-01 -5.89340329e-01
-4.21469420e-01 -3.21519911e-01 -6.61072254e-01 6.69509053e-01
7.95401216e-01 -2.05296800e-01 3.35968137e-01 6.36796892e-01
8.44741344e-01 5.82509823e-02 -1.17073238e+00 1.31327295e+00
8.26779425e-01 -1.51262522e+00 1.39286160e-01 -6.46644413e-01
5.72696269e-01 -4.19844955e-01 8.16067457e-02 2.28936553e-01
-3.87299031e-01 -1.06175911e+00 1.20619869e+00 7.30046690e-01
9.02518928e-01 -5.12755394e-01 7.39695311e-01 2.30252028e-01
-1.17675781e+00 -6.23232424e-02 -7.29505494e-02 -2.49234229e-01
9.02067482e-01 1.69577733e-01 -5.05992651e-01 3.55732828e-01
6.20668352e-01 1.02996218e+00 -6.19095802e-01 1.01228130e+00
-6.84468985e-01 7.67463863e-01 2.11677119e-01 -2.63684571e-01
7.35356733e-02 1.91136673e-01 7.88340986e-01 1.28944862e+00
2.93718755e-01 4.24551576e-01 -8.06163624e-02 7.50637531e-01
-2.15038523e-01 4.48779792e-01 -3.71670604e-01 -3.19812566e-01
6.32057071e-01 1.12811959e+00 -4.67528015e-01 -2.96138257e-01
-4.47773963e-01 1.17497110e+00 2.60757059e-01 4.11841184e-01
-1.13080156e+00 -4.16329414e-01 4.31246519e-01 1.73182756e-01
3.51032205e-02 -1.39180452e-01 -1.38847619e-01 -1.32541943e+00
3.39977860e-01 -9.21979129e-01 2.89325565e-01 -1.29872823e+00
-1.32180107e+00 9.25969601e-01 5.99129125e-03 -1.57425976e+00
-5.83426990e-02 -9.44152176e-02 -7.40660131e-01 3.13656449e-01
-1.12031472e+00 -1.40704107e+00 -6.86431110e-01 5.87458849e-01
1.10461891e+00 -4.26665284e-02 5.16398907e-01 3.29255790e-01
-6.86608493e-01 8.71740460e-01 -2.59880036e-01 1.40634090e-01
1.15838671e+00 -1.00122821e+00 1.42162919e-01 1.00946951e+00
3.29631090e-01 3.54231417e-01 1.21542037e+00 -5.38343012e-01
-1.00203192e+00 -1.25182247e+00 1.11374605e+00 -6.04440928e-01
8.64876807e-01 -6.15711093e-01 -7.83766270e-01 3.85512769e-01
3.59975278e-01 -2.46045113e-01 8.15362453e-01 -2.13167354e-01
-2.89778978e-01 2.81201210e-02 -5.53799808e-01 7.01747596e-01
1.58396232e+00 -6.09561801e-01 -5.52803934e-01 7.89001703e-01
1.45231843e+00 -2.59088069e-01 -9.20565546e-01 2.00660244e-01
6.65311337e-01 -7.74271131e-01 6.98472440e-01 -7.70553946e-01
1.30253291e+00 -2.31804848e-01 -4.54226255e-01 -1.01369405e+00
-1.55530974e-01 -6.23142004e-01 -3.00786674e-01 1.59176219e+00
6.53588533e-01 1.20309629e-01 3.77490371e-01 3.87725770e-01
-3.89341563e-01 -6.47692144e-01 -3.42240423e-01 -7.25910425e-01
-4.56425160e-01 -6.17166281e-01 5.21652699e-01 1.04295897e+00
2.80550420e-01 5.35895944e-01 -1.03566623e+00 2.05794066e-01
3.22690785e-01 1.86445132e-01 6.06302619e-01 -4.62743700e-01
-4.53294605e-01 -4.19385761e-01 -1.51072010e-01 -1.12593770e+00
8.75219628e-02 -7.15457022e-01 6.16236478e-02 -1.88682151e+00
6.10357761e-01 2.70455748e-01 4.28373320e-03 2.14914516e-01
-2.49693811e-01 4.45044547e-01 3.21052998e-01 3.29565346e-01
-1.32141054e+00 6.79247677e-01 1.62395597e+00 -5.66139109e-02
-1.52583495e-01 -6.50582388e-02 -8.03046882e-01 7.29850948e-01
6.93944573e-01 -2.26979434e-01 -6.21232152e-01 -6.66492105e-01
4.54943538e-01 6.15818262e-01 3.43041271e-01 -9.23756182e-01
4.24961656e-01 -3.39368045e-01 1.30755559e-01 -5.16742527e-01
4.49900985e-01 -2.05773354e-01 1.52173951e-01 -2.35916167e-01
-9.59529161e-01 8.52533206e-02 -3.04051757e-01 4.38786030e-01
-5.48921406e-01 -2.40865663e-01 4.25781608e-01 -7.67153352e-02
-7.22184002e-01 3.40661854e-01 -3.47585320e-01 2.77868897e-01
1.10183966e+00 -1.27249047e-01 -7.02278972e-01 -1.08042991e+00
-9.16213274e-01 3.37210059e-01 3.72277945e-01 7.41444945e-01
8.47296536e-01 -1.52481079e+00 -1.18954670e+00 -4.90187675e-01
5.30500352e-01 -2.69410104e-01 7.39945710e-01 8.52467597e-01
-3.75753492e-01 9.05131996e-02 -1.73066538e-02 -2.64292032e-01
-1.39615679e+00 5.29548228e-01 -1.39624178e-01 2.13274464e-01
-6.91284359e-01 1.10845983e+00 4.09305632e-01 3.57507437e-01
6.70763791e-01 -2.98556406e-02 -7.72693992e-01 3.70523244e-01
1.23844695e+00 2.00891756e-02 -4.27303046e-01 -7.20116198e-01
3.98833714e-02 1.46205083e-01 5.58593944e-02 -1.90746307e-01
1.31455815e+00 -5.11390626e-01 2.09618788e-02 6.42350435e-01
1.12597239e+00 5.48671708e-02 -1.20227635e+00 -1.75873876e-01
-2.39919260e-01 -4.74493653e-01 -1.80247918e-01 -8.46780002e-01
-7.11284339e-01 8.41053665e-01 -1.80859659e-02 3.40943843e-01
1.30950725e+00 3.53321463e-01 1.06795096e+00 1.26419380e-01
1.34205455e-02 -9.87922370e-01 5.45988798e-01 2.91982532e-01
1.52676713e+00 -1.13857281e+00 -4.41239029e-01 -3.66246969e-01
-1.37278247e+00 9.75197554e-01 8.35511327e-01 2.90584773e-01
5.78424148e-02 9.48694274e-02 3.35017336e-03 6.28266782e-02
-9.37312901e-01 -1.95541471e-01 3.25864553e-01 5.19436955e-01
2.57321775e-01 2.45223880e-01 -3.91843408e-01 1.53268075e+00
-6.51217163e-01 -2.96736568e-01 7.52240360e-01 3.12174767e-01
-4.20893788e-01 -5.78938603e-01 -8.66335481e-02 2.40640506e-01
-4.51125085e-01 -2.28193760e-01 -4.00018781e-01 3.39728326e-01
1.11031443e-01 1.16732490e+00 -8.60114768e-02 -5.97467303e-01
4.25956339e-01 -3.13353330e-01 3.97973686e-01 -6.40469253e-01
-4.24024940e-01 4.21891324e-02 6.42664075e-01 -6.31264031e-01
-4.74421740e-01 -6.70678258e-01 -9.78479981e-01 -9.81788859e-02
8.17581266e-02 2.76749954e-02 4.11729664e-01 8.09488893e-01
7.81609192e-02 6.55787587e-01 7.47451305e-01 -5.94551980e-01
-1.91632256e-01 -1.18480372e+00 -4.00279701e-01 6.80785120e-01
2.55988032e-01 -2.63586253e-01 -4.34087753e-01 5.19709826e-01] | [10.608951568603516, 0.7275850176811218] |
fefa6bce-a0c4-45c2-a117-32a50be92c8d | combating-the-curse-of-multilinguality-in | null | null | https://aclanthology.org/2022.naacl-main.176 | https://aclanthology.org/2022.naacl-main.176.pdf | Combating the Curse of Multilinguality in Cross-Lingual WSD by Aligning Sparse Contextualized Word Representations | In this paper, we advocate for using large pre-trained monolingual language models in cross lingual zero-shot word sense disambiguation (WSD) coupled with a contextualized mapping mechanism. We also report rigorous experiments that illustrate the effectiveness of employing sparse contextualized word representations obtained via a dictionary learning procedure. Our experimental results demonstrate that the above modifications yield a significant improvement of nearly 6.5 points of increase in the average F-score (from 62.0 to 68.5) over a collection of 17 typologically diverse set of target languages. We release our source code for replicating our experiments at https://github.com/begab/sparsity_makes_sense. | ['Gábor Berend'] | null | null | null | null | naacl-2022-7 | ['word-sense-disambiguation'] | ['natural-language-processing'] | [-1.99482098e-01 -1.57550573e-01 -6.29941404e-01 -2.79071093e-01
-1.23135829e+00 -6.31362796e-01 8.22263598e-01 3.80853355e-01
-7.11670578e-01 9.55323458e-01 5.94273627e-01 -5.10467291e-01
8.19614902e-02 -4.76906091e-01 -3.09257358e-01 -3.23910594e-01
-1.52337337e-02 3.62544149e-01 -2.82626003e-01 -8.19445968e-01
4.12297726e-01 6.00439832e-02 -1.22691154e+00 -2.38878369e-01
1.19519699e+00 1.85505986e-01 4.58843589e-01 1.46012679e-01
-4.44481224e-01 2.03728959e-01 -5.59802473e-01 -4.66026843e-01
3.23488474e-01 -7.73319453e-02 -8.83306921e-01 -2.89146513e-01
5.22931516e-01 2.03754008e-01 -1.54657632e-01 1.31909740e+00
6.78759158e-01 5.04434705e-01 3.37864935e-01 -6.75851762e-01
-1.25712621e+00 8.95693898e-01 -6.39992416e-01 6.20039940e-01
5.69725931e-01 -1.81000695e-01 1.36685240e+00 -1.27494645e+00
9.29412067e-01 1.16691852e+00 5.70680261e-01 5.80959857e-01
-1.25056899e+00 -8.93513858e-01 2.90741146e-01 1.96971491e-01
-1.73367465e+00 -8.32808375e-01 5.58450937e-01 -2.77677387e-01
1.55991971e+00 -2.11292375e-02 3.59194577e-01 1.34988856e+00
2.58755740e-02 5.13845265e-01 1.23378491e+00 -8.91250670e-01
1.26381114e-01 3.49528641e-02 4.59062636e-01 5.44211984e-01
8.21337461e-01 2.14222874e-02 -6.77321851e-01 -3.43879610e-01
5.87016821e-01 -2.75276691e-01 -1.39450550e-01 -1.14211679e-01
-1.31118727e+00 9.99650776e-01 1.04830042e-01 7.93973565e-01
-1.87541187e-01 -2.41802081e-01 5.07150531e-01 5.45130372e-01
7.35368192e-01 8.70202422e-01 -5.80465674e-01 -1.92885548e-01
-8.27103138e-01 1.20203882e-01 5.87152958e-01 1.04066241e+00
7.51628518e-01 2.75302380e-01 5.91991283e-02 1.36715829e+00
1.15697689e-01 8.09721351e-01 7.40985572e-01 -7.12221026e-01
2.59418577e-01 2.85689235e-01 1.88331142e-01 -8.52319479e-01
2.02407781e-02 -4.23448622e-01 -3.70064735e-01 -5.18834710e-01
8.90604279e-04 -2.41497278e-01 -8.96733642e-01 1.90959454e+00
1.42545000e-01 2.35686302e-01 2.03873381e-01 8.29150677e-01
6.83086514e-01 4.65963185e-01 5.27246773e-01 -1.49416238e-01
1.61652553e+00 -6.90299273e-01 -9.00355637e-01 -6.58046067e-01
7.90133834e-01 -1.25819921e+00 1.53182626e+00 4.70295735e-02
-8.44816446e-01 -3.40226650e-01 -1.11631703e+00 -2.71216422e-01
-5.62708497e-01 -1.19847573e-01 7.12913454e-01 4.50222641e-01
-9.89457965e-01 1.16171174e-01 -6.83490336e-01 -9.09521341e-01
4.90786135e-03 -6.70797825e-02 -4.06209499e-01 -3.19450766e-01
-1.67281151e+00 1.03230274e+00 3.91212076e-01 -4.84377414e-01
-5.69926739e-01 -7.45319843e-01 -1.00175107e+00 -2.31111333e-01
4.21998739e-01 -3.69813830e-01 1.12805259e+00 -5.71334541e-01
-9.72404420e-01 1.40481949e+00 -5.00598371e-01 -2.59125024e-01
-1.35140881e-01 -5.74244678e-01 -8.78070295e-01 -2.55122155e-01
7.38542974e-01 4.77264732e-01 2.69334823e-01 -1.06378186e+00
-6.07090414e-01 -1.41751885e-01 1.14661187e-01 2.95439482e-01
-2.98892856e-01 3.11615109e-01 -5.06902337e-01 -1.17175782e+00
-1.47114828e-01 -9.07709837e-01 -3.53925765e-01 -7.22559869e-01
-1.29804224e-01 -4.04728860e-01 1.35895953e-01 -8.27388585e-01
1.44157016e+00 -2.06084609e+00 -3.99327651e-02 6.55089552e-03
-8.72875080e-02 3.08072567e-01 -3.74368936e-01 7.38179862e-01
-4.40286808e-02 3.81374389e-01 -2.34654933e-01 -3.03645790e-01
6.88752308e-02 2.54701674e-01 -3.35573256e-01 3.45559686e-01
8.73505697e-02 6.58292890e-01 -1.34242010e+00 -3.01349074e-01
9.97136310e-02 4.92099047e-01 -3.90800744e-01 -3.00919786e-02
5.79315275e-02 1.63534552e-01 -4.04930621e-01 8.91557336e-01
5.13601601e-01 -2.14952901e-02 7.53587365e-01 2.28439763e-01
-7.25680217e-02 8.81119609e-01 -9.87044871e-01 2.41286945e+00
-7.09881425e-01 5.02988100e-01 -1.30440339e-01 -8.45274270e-01
1.02477562e+00 2.32667878e-01 2.09399775e-01 -9.00242627e-01
1.12938859e-01 3.29537928e-01 -1.84036732e-01 -3.34213525e-02
9.09336090e-01 -2.96640873e-01 -5.51631272e-01 5.72399437e-01
5.51135421e-01 6.80986792e-02 3.01863283e-01 3.43074352e-01
7.90800214e-01 -1.34408921e-02 8.46257269e-01 -1.01202989e+00
4.03247386e-01 3.94429654e-01 7.14968920e-01 5.80824018e-01
-3.75375152e-01 1.42609522e-01 9.96152125e-03 -2.56742418e-01
-1.08062291e+00 -1.11350942e+00 -4.29297805e-01 1.40768707e+00
1.12333976e-01 -7.07721472e-01 -3.80466074e-01 -4.26925331e-01
5.35684489e-02 1.03467667e+00 -3.96330029e-01 -3.68667021e-02
-6.67336643e-01 -6.78827584e-01 5.65430105e-01 3.45646262e-01
7.62914568e-02 -7.29529023e-01 5.75386658e-02 1.51966497e-01
-2.50764608e-01 -1.06267738e+00 -5.84265471e-01 1.05510488e-01
-5.14029682e-01 -7.69261837e-01 -4.10512507e-01 -1.11017811e+00
3.21353137e-01 6.94183111e-01 1.28109550e+00 6.61825910e-02
-2.54746228e-01 2.40938634e-01 -6.48336053e-01 -2.62918681e-01
-6.41575828e-02 4.70143467e-01 6.71818435e-01 -6.35758758e-01
1.16919863e+00 -4.53863829e-01 -3.31274450e-01 -1.29618213e-01
-6.89083874e-01 -4.30997044e-01 1.30733475e-01 8.48472476e-01
7.03971803e-01 -6.68861866e-01 7.90759981e-01 -8.96085501e-01
9.99475360e-01 -8.62331748e-01 -5.19481003e-01 3.47731054e-01
-1.03028083e+00 1.26365885e-01 1.81241646e-01 -3.42456043e-01
-9.70801532e-01 -3.67347747e-01 -1.66893348e-01 -1.37706220e-01
-1.09622501e-01 6.35788858e-01 2.56419003e-01 1.08379431e-01
8.32397938e-01 -8.59110355e-02 -3.43355328e-01 -6.22389913e-01
7.44221091e-01 8.54247928e-01 3.73793721e-01 -1.05385160e+00
5.48606098e-01 2.11098328e-01 -7.96455026e-01 -5.32038927e-01
-9.78829741e-01 -7.58588612e-01 -5.99767327e-01 2.35868126e-01
6.50245249e-01 -1.35759854e+00 9.56376866e-02 -1.79456696e-01
-7.76593983e-01 -2.09193781e-01 -5.14766313e-02 6.26091242e-01
-9.00619775e-02 3.20011050e-01 -4.45590347e-01 -3.89138699e-01
-6.16934121e-01 -8.77722263e-01 9.07386422e-01 9.72615182e-02
-6.09954596e-01 -1.36720562e+00 4.61146176e-01 3.39394987e-01
4.61991638e-01 1.33595709e-02 7.69315302e-01 -9.25328612e-01
-1.21915322e-02 5.37090003e-02 6.25983253e-02 4.59495299e-02
4.13063794e-01 -4.41500187e-01 -9.05228853e-01 -4.94635403e-01
-2.68962979e-01 -2.81078696e-01 6.30633175e-01 2.20238894e-01
4.26252663e-01 9.92352795e-03 -2.61428237e-01 4.75503713e-01
1.78390312e+00 4.71904613e-02 1.68342814e-01 6.29208267e-01
5.95528901e-01 2.09404796e-01 8.70128393e-01 3.17320108e-01
5.19865453e-01 5.75564086e-01 -4.47043210e-01 1.71008587e-01
-3.86081576e-01 -3.46396327e-01 2.68914938e-01 1.42153275e+00
-4.16835919e-02 2.38404237e-03 -1.42566764e+00 1.32024789e+00
-1.63333488e+00 -8.47527683e-01 2.59170502e-01 2.15446544e+00
1.20970476e+00 -8.90547633e-02 -2.36388043e-01 -4.00018603e-01
8.72547269e-01 3.55312347e-01 -2.96175450e-01 -7.54751086e-01
-3.04778576e-01 8.57761323e-01 5.31112909e-01 1.10047793e+00
-8.35884392e-01 1.75396347e+00 6.34910727e+00 1.05907261e+00
-1.04304528e+00 5.70115387e-01 1.72994182e-01 -1.99563280e-01
-7.18223631e-01 3.17141116e-02 -8.15401316e-01 1.12148233e-01
1.09382510e+00 -9.17666256e-01 6.70201480e-01 8.05872440e-01
6.47155344e-02 1.06904030e-01 -5.97581446e-01 8.59544992e-01
1.70790717e-01 -1.18333769e+00 1.42729133e-01 -5.44664003e-02
1.10731542e+00 5.75900018e-01 -1.70221522e-01 5.33415437e-01
7.78981805e-01 -8.93653691e-01 4.27348047e-01 1.19618960e-01
1.07837164e+00 -7.85883963e-01 4.52211916e-01 -9.50655490e-02
-1.17560387e+00 4.15813148e-01 -5.42772412e-01 -8.99238586e-02
1.17154233e-01 6.87182784e-01 -8.37659061e-01 6.63684070e-01
5.41247070e-01 7.66407013e-01 -3.50570083e-01 4.70829904e-01
-4.35260385e-01 7.45922446e-01 4.26596124e-03 7.67736584e-02
3.31597894e-01 -7.87550807e-02 6.83611631e-01 1.54121006e+00
3.12413394e-01 4.31182124e-02 2.99462497e-01 4.52796668e-01
-1.86322808e-01 6.87652826e-01 -7.83074677e-01 -1.99356481e-01
1.11272252e+00 9.82161641e-01 -3.98284614e-01 -5.62852740e-01
-6.37937427e-01 7.40966618e-01 7.02915549e-01 4.61601943e-01
-4.31306273e-01 -5.83448350e-01 1.33377588e+00 -2.43024766e-01
3.40997070e-01 -6.11769795e-01 -5.17209470e-01 -1.66780925e+00
-1.40609547e-01 -9.31562841e-01 5.26606202e-01 -4.12384927e-01
-1.66775668e+00 6.36816502e-01 -8.86618346e-02 -9.42877591e-01
-2.42444679e-01 -3.99177015e-01 -2.33026087e-01 1.10789073e+00
-1.64109826e+00 -8.46110702e-01 2.47851014e-01 5.07745981e-01
7.66259491e-01 -4.08885449e-01 1.23694730e+00 4.55333680e-01
-4.32636738e-01 7.79398382e-01 1.56995997e-01 1.89152136e-01
1.13422418e+00 -1.11485744e+00 7.81112611e-01 1.01750159e+00
4.26355422e-01 1.36072087e+00 6.79167032e-01 -8.11134100e-01
-1.32601976e+00 -8.93044233e-01 1.63487697e+00 -4.68805164e-01
1.20086324e+00 -4.16827172e-01 -7.91251063e-01 9.16665554e-01
8.18427205e-01 -3.29711497e-01 1.10792565e+00 8.70888531e-01
-5.97196639e-01 1.37911305e-01 -1.08069897e+00 7.91489184e-01
1.31039155e+00 -8.97206783e-01 -1.25629580e+00 2.43481055e-01
8.35192680e-01 -1.31295472e-01 -1.16505516e+00 1.58217072e-01
2.84088016e-01 -2.83856153e-01 9.63464797e-01 -8.18721652e-01
1.14248715e-01 -1.42459974e-01 -6.29844129e-01 -1.48696971e+00
-5.57006180e-01 -4.24280882e-01 2.39682704e-01 1.22220910e+00
4.58234191e-01 -7.51600504e-01 9.27219540e-02 7.46196434e-02
-1.43862933e-01 -5.23670852e-01 -9.69417751e-01 -8.51862431e-01
4.96423781e-01 -5.89510500e-01 5.77687025e-01 1.87301588e+00
3.70962650e-01 5.07941306e-01 -2.56469309e-01 1.21933520e-01
5.05917251e-01 7.87860900e-02 1.97246924e-01 -8.92952800e-01
1.50980912e-02 -2.55011439e-01 -2.12961078e-01 -6.90141022e-01
5.32587409e-01 -1.25054777e+00 -1.87492713e-01 -1.38976455e+00
2.01561913e-01 -4.79837060e-01 -8.21415067e-01 5.69033444e-01
-5.12447119e-01 1.91690728e-01 6.43076524e-02 1.48115918e-01
-5.29599190e-01 2.99255908e-01 8.82385135e-01 -7.49148056e-02
-1.60841141e-02 -7.91288316e-01 -1.24489689e+00 3.06284457e-01
1.19211268e+00 -6.72316968e-01 -3.93979102e-01 -9.92412090e-01
1.94400474e-01 -3.75563741e-01 -1.81555063e-01 -5.38379252e-01
-1.96071491e-01 -5.07992625e-01 -1.90150708e-01 2.00225323e-01
1.25572160e-01 -1.24119684e-01 -3.07458222e-01 2.23971382e-01
-3.53110433e-01 3.70799214e-01 5.01685917e-01 2.78396487e-01
-2.86338300e-01 -3.65737639e-02 5.16827047e-01 -3.14997911e-01
-1.28741562e+00 5.00697829e-02 -3.28625917e-01 5.52576125e-01
6.85346842e-01 2.72671312e-01 -2.86315292e-01 1.31067131e-02
-6.47580206e-01 2.14396149e-01 5.72856784e-01 9.83852804e-01
1.43713713e-01 -1.64506483e+00 -8.11641812e-01 7.37464726e-02
5.26441693e-01 -6.45338237e-01 -2.49333470e-03 4.19940829e-01
-2.45429516e-01 6.74089849e-01 -2.42897093e-01 -1.19808435e-01
-9.07057345e-01 4.45216298e-01 7.41401017e-02 -8.78482759e-02
-5.22027314e-01 8.11097443e-01 -2.49259099e-01 -7.43829906e-01
-4.13664654e-02 -1.11493960e-01 -6.23962749e-03 1.29006896e-02
5.10638058e-01 2.20918268e-01 1.65970847e-01 -7.55373895e-01
-7.48704135e-01 3.85800987e-01 -1.12613171e-01 -3.74712646e-01
1.40540910e+00 -3.50724816e-01 -2.73383781e-02 5.66670179e-01
8.63288403e-01 5.79223633e-01 -3.74589980e-01 -6.00986421e-01
4.61439639e-01 -5.99560261e-01 1.88749343e-01 -9.18968976e-01
-6.46739185e-01 4.62808520e-01 5.55822730e-01 -9.57469270e-02
8.60182524e-01 2.03982487e-01 8.77902806e-01 4.05792981e-01
7.23701417e-01 -1.30407262e+00 -5.77387273e-01 7.97849596e-01
6.72438323e-01 -1.34648168e+00 -4.30087186e-02 -1.13302670e-01
-7.82781720e-01 5.65310538e-01 3.63599867e-01 -5.11867464e-01
6.89709008e-01 2.36290321e-01 5.76998830e-01 -2.58063585e-01
-8.18174481e-01 -5.63670754e-01 1.91108420e-01 5.60034394e-01
1.01734698e+00 5.78935504e-01 -9.97124076e-01 6.32359922e-01
-4.23755407e-01 -2.94811964e-01 2.41235271e-01 1.01418710e+00
-5.31113803e-01 -1.56252265e+00 -2.71579832e-01 6.83107898e-02
-5.34915864e-01 -7.86222577e-01 -3.57162714e-01 8.19467604e-01
2.04581823e-02 9.94546115e-01 1.28681185e-02 -2.76565492e-01
1.06009036e-01 4.23236728e-01 2.42480561e-01 -9.00865793e-01
-2.89894342e-01 1.57600045e-01 4.93912637e-01 -4.59463149e-01
-4.66072470e-01 -8.46396387e-01 -1.19528615e+00 -6.13490939e-01
7.88382515e-02 1.95608139e-01 5.12938619e-01 8.89154971e-01
5.64137816e-01 1.16697140e-01 3.28121483e-01 -2.31716722e-01
-2.75509298e-01 -1.16306329e+00 -5.25098205e-01 4.34156150e-01
1.78687483e-01 -8.37161481e-01 -1.62492722e-01 -2.94255540e-02] | [10.749703407287598, 9.730742454528809] |
a2482f28-3c35-47f5-99da-1677e01332d5 | raising-the-limit-of-image-rescaling-using | 2303.06747 | null | https://arxiv.org/abs/2303.06747v1 | https://arxiv.org/pdf/2303.06747v1.pdf | Raising The Limit Of Image Rescaling Using Auxiliary Encoding | Normalizing flow models using invertible neural networks (INN) have been widely investigated for successful generative image super-resolution (SR) by learning the transformation between the normal distribution of latent variable $z$ and the conditional distribution of high-resolution (HR) images gave a low-resolution (LR) input. Recently, image rescaling models like IRN utilize the bidirectional nature of INN to push the performance limit of image upscaling by optimizing the downscaling and upscaling steps jointly. While the random sampling of latent variable $z$ is useful in generating diverse photo-realistic images, it is not desirable for image rescaling when accurate restoration of the HR image is more important. Hence, in places of random sampling of $z$, we propose auxiliary encoding modules to further push the limit of image rescaling performance. Two options to store the encoded latent variables in downscaled LR images, both readily supported in existing image file format, are proposed. One is saved as the alpha-channel, the other is saved as meta-data in the image header, and the corresponding modules are denoted as suffixes -A and -M respectively. Optimal network architectural changes are investigated for both options to demonstrate their effectiveness in raising the rescaling performance limit on different baseline models including IRN and DLV-IRN. | ['Paul Bogdan', 'Le Kang', 'Xin Zhou', 'Zhihong Pan', 'Chenzhong Yin'] | 2023-03-12 | null | null | null | null | ['image-super-resolution'] | ['computer-vision'] | [ 6.65294349e-01 2.11681902e-01 -2.63698041e-01 -3.59217077e-01
-5.86697698e-01 -2.57832617e-01 5.51088274e-01 -6.51921868e-01
-3.21053058e-01 1.03078973e+00 5.14706016e-01 -2.31117755e-01
-5.64739034e-02 -1.09196734e+00 -7.54325390e-01 -9.36095297e-01
1.83754399e-01 -1.94505960e-01 6.80723488e-02 -2.70474076e-01
2.65441269e-01 5.89847982e-01 -1.77662873e+00 2.74614602e-01
8.08378816e-01 8.01343858e-01 5.62671363e-01 8.09557855e-01
-2.88756579e-01 8.12942386e-01 -4.76849496e-01 -5.38910031e-02
5.82542956e-01 -6.94704831e-01 -6.33591831e-01 -1.87799007e-01
7.23843932e-01 -6.12484455e-01 -6.19618237e-01 1.08848369e+00
6.18952036e-01 2.19127104e-01 4.38975364e-01 -1.07241809e+00
-1.21535325e+00 6.74925983e-01 -7.58388996e-01 4.12097603e-01
1.90569367e-02 2.89762676e-01 6.75051332e-01 -6.06784046e-01
1.00323808e+00 1.42106271e+00 3.78844529e-01 6.26874864e-01
-1.43263018e+00 -7.78727233e-01 -1.63752392e-01 1.10122032e-01
-1.35111868e+00 -6.18187189e-01 7.21853197e-01 -1.92217901e-01
7.64480591e-01 2.61967242e-01 2.15980411e-01 1.15069258e+00
2.66074121e-01 6.16618767e-02 1.21429265e+00 -3.44325066e-01
6.69325367e-02 2.26616278e-01 -3.33829880e-01 3.49218935e-01
1.29835516e-01 2.33609051e-01 -6.87491894e-01 1.74031779e-01
1.53635406e+00 -3.12756300e-01 -4.26815003e-01 -9.73628759e-02
-1.14620304e+00 6.97679341e-01 5.90992212e-01 1.00496300e-01
-3.98454875e-01 3.27380151e-01 1.69996530e-01 2.73138732e-01
3.46741319e-01 2.76639313e-01 -7.75780827e-02 6.82510212e-02
-1.10139823e+00 8.35682601e-02 7.11360425e-02 9.86409783e-01
8.00886691e-01 5.49726605e-01 -4.29255128e-01 9.96570766e-01
-1.08089976e-01 3.83270621e-01 5.32370210e-01 -1.30800259e+00
5.32756805e-01 3.17397922e-01 2.33412147e-01 -9.05189812e-01
-1.81282237e-01 -5.79588056e-01 -1.31216276e+00 3.69814634e-01
2.67823398e-01 2.34650448e-02 -1.06344926e+00 2.01811647e+00
1.32185593e-01 1.72841862e-01 2.15781361e-01 8.63018036e-01
9.23754036e-01 8.40638459e-01 1.56784609e-01 -5.34889877e-01
1.26373041e+00 -7.15588808e-01 -8.72873306e-01 -9.46189910e-02
1.54569283e-01 -9.27591324e-01 1.39031065e+00 5.50697073e-02
-1.36436617e+00 -1.08489394e+00 -1.13787961e+00 -5.04275322e-01
-2.19816446e-01 2.24125087e-01 3.99889857e-01 3.91106904e-01
-1.40909612e+00 6.46079063e-01 -4.91530329e-01 -1.67462692e-01
4.43190277e-01 2.19485119e-01 -4.79686886e-01 -1.59398112e-02
-1.30229509e+00 6.04997218e-01 2.98162252e-01 1.78392276e-01
-6.64006591e-01 -7.33889937e-01 -6.70173407e-01 6.13527931e-02
-2.87660658e-02 -7.81228542e-01 5.41540921e-01 -8.84574473e-01
-1.66606128e+00 6.47695303e-01 -1.21036559e-01 -3.43302816e-01
5.81465662e-01 4.75862576e-03 -4.78937984e-01 2.53122091e-01
-7.87782017e-03 1.27510011e+00 1.27605081e+00 -1.45800149e+00
-4.73983884e-01 -1.89902663e-01 4.11114469e-02 3.36537182e-01
-2.78684139e-01 -1.78244058e-02 -2.24247694e-01 -6.81157410e-01
2.51905411e-01 -6.47504270e-01 -1.30554780e-01 5.24902493e-02
-4.17139977e-01 2.49103516e-01 1.08232391e+00 -7.97883928e-01
1.20950592e+00 -2.09182429e+00 -3.08652855e-02 -1.12502556e-03
5.63100204e-02 2.01297641e-01 -3.03780645e-01 -1.48190407e-03
-4.59736764e-01 1.63638443e-01 -1.29612520e-01 -1.56053036e-01
-4.09597009e-01 3.02529007e-01 -5.25553644e-01 3.14037710e-01
2.34748647e-01 8.41687560e-01 -4.94915485e-01 -5.66976249e-01
2.55312741e-01 9.26191509e-01 -4.22296524e-01 1.84292585e-01
-5.21142855e-02 6.62543297e-01 3.40187810e-02 1.65913180e-01
1.05235982e+00 -3.06438804e-01 4.72268872e-02 -7.37119973e-01
-2.89869487e-01 2.03034580e-02 -1.50308442e+00 1.40053833e+00
-4.13101405e-01 5.56964338e-01 3.18369106e-03 -5.16974747e-01
1.15506887e+00 1.90254182e-01 3.01666170e-01 -1.01343942e+00
-1.66737646e-01 -2.36176923e-01 -2.00966746e-01 -4.09529269e-01
9.85933244e-01 -2.79114932e-01 3.15150559e-01 4.50520396e-01
-4.48375046e-02 3.48473825e-02 2.03042910e-01 1.85915947e-01
5.72256684e-01 5.32124102e-01 1.46718621e-01 -2.45430917e-01
7.40744770e-01 -3.25936735e-01 5.75135887e-01 7.73039043e-01
2.23503653e-02 9.69135463e-01 5.90488076e-01 -4.81561244e-01
-1.63701952e+00 -1.20055926e+00 -2.02919289e-01 1.02895820e+00
5.49133196e-02 -1.90422624e-01 -7.80142725e-01 -2.14779258e-01
-5.15911102e-01 7.82925129e-01 -5.59015095e-01 -8.89223367e-02
-8.36508155e-01 -8.98106337e-01 4.94378388e-01 4.94631380e-01
1.02170217e+00 -1.14702082e+00 -5.90945721e-01 -2.97321714e-02
-4.67181504e-01 -1.12363160e+00 -4.10626233e-01 7.06931353e-02
-8.20559144e-01 -5.33667028e-01 -8.16465795e-01 -4.50234205e-01
6.84380531e-01 3.97232026e-01 1.02769542e+00 -1.21911615e-01
-3.14571649e-01 -1.86547369e-01 -6.80627748e-02 1.62722155e-01
-5.94382703e-01 3.80453677e-03 3.55564132e-02 2.91189644e-02
-2.95999885e-01 -8.55916440e-01 -8.97211909e-01 3.27234715e-01
-1.24494123e+00 4.68637198e-01 4.72093850e-01 6.86235607e-01
7.08178222e-01 1.88095838e-01 4.92540061e-01 -7.18345225e-01
4.33729678e-01 -1.72369182e-01 -5.64844370e-01 3.20016108e-02
-6.55628145e-01 3.33919853e-01 7.65632987e-01 -4.51169342e-01
-1.46932256e+00 -2.29155317e-01 -1.77390635e-01 -5.19559562e-01
-8.22230577e-02 -4.49252911e-02 -1.67074144e-01 6.16757087e-02
7.54405797e-01 3.55914891e-01 1.05773583e-01 -3.82520169e-01
6.70316339e-01 5.93813121e-01 8.84745657e-01 -3.63318443e-01
9.33460116e-01 6.40062094e-01 1.81808293e-01 -7.40312934e-01
-6.21151567e-01 6.42074049e-02 -5.20313084e-01 -1.21077634e-01
1.05300117e+00 -1.00562894e+00 -4.37995851e-01 3.43969673e-01
-8.75134349e-01 -1.59280136e-01 -4.84398067e-01 2.98045456e-01
-6.66007698e-01 2.13222355e-01 -5.87852240e-01 -5.34978867e-01
-3.36773813e-01 -1.35053754e+00 9.13283288e-01 5.78296661e-01
7.45538324e-02 -6.52832031e-01 -1.16873085e-01 3.69323105e-01
1.00071335e+00 1.89272374e-01 1.24274504e+00 2.19403505e-01
-7.45737314e-01 3.52323592e-01 -8.09615374e-01 5.32439530e-01
1.15667626e-01 -3.54813784e-02 -1.04685235e+00 -3.33951443e-01
-1.22552395e-01 -6.23226725e-02 7.32889175e-01 6.68470323e-01
1.27701926e+00 -3.64928186e-01 3.16152126e-02 9.56853807e-01
1.71299100e+00 4.61407192e-02 1.40659785e+00 4.19952273e-01
7.01368749e-01 4.44527119e-01 1.69327304e-01 2.91084349e-01
2.06236154e-01 8.31883669e-01 4.04376745e-01 -3.55645597e-01
-6.92449808e-01 -2.93265194e-01 3.93612832e-01 4.94757295e-01
-4.28485036e-01 -1.36550412e-01 -4.56967443e-01 1.34734422e-01
-1.34448421e+00 -1.22331905e+00 9.13840160e-02 2.38971090e+00
9.80936706e-01 2.35729478e-03 -1.81954563e-01 -1.85266614e-01
9.78603601e-01 5.87098598e-01 -6.34472370e-01 -3.26698810e-01
-5.60897827e-01 2.18074545e-01 6.95607364e-01 7.08540380e-01
-7.80471981e-01 9.11467910e-01 6.46833849e+00 9.48225021e-01
-1.28270638e+00 1.65806204e-01 9.95029986e-01 -2.48181283e-01
-4.67750400e-01 1.71973202e-02 -1.14032876e+00 4.25283700e-01
1.04309034e+00 1.22306682e-01 5.44086516e-01 5.81288099e-01
5.97459972e-01 -1.01771563e-01 -7.05182314e-01 1.12183535e+00
-8.50526392e-02 -1.59452903e+00 5.49819469e-01 1.27374545e-01
7.03745127e-01 -2.38104030e-01 2.92954385e-01 9.28866938e-02
1.08852230e-01 -1.24514258e+00 5.16394615e-01 5.72239459e-01
1.40305459e+00 -7.35939562e-01 5.49017072e-01 -1.03131920e-01
-1.10527956e+00 -1.50764957e-01 -6.18249476e-01 2.15124235e-01
4.57117021e-01 4.08779174e-01 -3.85443598e-01 4.36486840e-01
8.40430856e-01 4.44847494e-01 -6.76948488e-01 1.94643646e-01
-4.34489921e-02 2.67378837e-01 -8.86227712e-02 8.80183280e-01
8.42738152e-03 -4.64078605e-01 6.05035186e-01 1.04342306e+00
5.01424432e-01 -2.65975706e-02 -3.39090347e-01 1.20468855e+00
-1.09219059e-01 -1.39071479e-01 -4.36607450e-01 1.72330543e-01
3.84431392e-01 1.35938716e+00 -6.80539250e-01 -2.70191401e-01
-9.26863328e-02 9.12954926e-01 -2.62424466e-03 6.31539047e-01
-9.58879471e-01 -2.93187857e-01 6.40586913e-01 4.18595105e-01
2.72300899e-01 -1.78573683e-01 -2.28266582e-01 -9.07362342e-01
-1.11104161e-01 -7.51204312e-01 1.55410245e-01 -1.26316524e+00
-8.73120070e-01 8.54839563e-01 1.54443473e-01 -1.23100913e+00
-2.08818570e-01 -1.91860676e-01 -2.59470105e-01 1.25712824e+00
-1.69610775e+00 -1.17236185e+00 -6.64659321e-01 7.88178205e-01
4.28394914e-01 -1.61258295e-01 5.60347557e-01 3.60846221e-01
-5.71122527e-01 5.76766193e-01 7.71630332e-02 -8.61291960e-02
8.69688153e-01 -9.63759780e-01 2.72977293e-01 1.04402733e+00
-1.58623293e-01 7.06939340e-01 9.06910479e-01 -6.72549367e-01
-1.17268038e+00 -1.07075381e+00 5.06838560e-01 -1.64325698e-03
5.16885184e-02 -2.86674142e-01 -1.04956329e+00 6.28590226e-01
3.60800624e-01 1.91323012e-01 3.02490830e-01 -4.61378098e-01
-3.65171939e-01 -3.30524385e-01 -1.26320517e+00 7.24205673e-01
9.90417480e-01 -4.04369086e-01 4.10278477e-02 5.87708578e-02
9.29967880e-01 -3.91311288e-01 -1.08106649e+00 3.98917943e-01
3.33373547e-01 -1.13835061e+00 1.43458176e+00 -2.71176696e-01
8.90967071e-01 -6.04548514e-01 -3.84233594e-01 -1.02426374e+00
-5.29512346e-01 -7.05655634e-01 2.06912551e-02 1.35418868e+00
1.08609222e-01 -5.33263147e-01 7.58748889e-01 4.65375006e-01
1.91786677e-01 -4.01950240e-01 -8.95581245e-01 -3.57277542e-01
-5.81006482e-02 -4.15736511e-02 6.39372408e-01 8.53820801e-01
-6.75247610e-01 2.74107844e-01 -7.02399135e-01 1.93895757e-01
7.91309536e-01 -1.33320913e-01 6.70639634e-01 -7.46928394e-01
-1.75166860e-01 -2.38910094e-01 -1.09487571e-01 -9.89553094e-01
-1.62646383e-01 -6.29153728e-01 -2.80393392e-01 -1.46366918e+00
2.09152475e-01 -5.35251081e-01 -1.75848737e-01 2.43137449e-01
-5.48788533e-02 6.25256658e-01 2.83756673e-01 4.11287636e-01
2.42084377e-02 4.83735383e-01 1.64461410e+00 2.39464760e-01
-3.39722544e-01 -4.39755380e-01 -7.83837378e-01 5.10815620e-01
6.66202307e-01 -2.41723999e-01 -6.82350814e-01 -3.69464219e-01
3.02864283e-01 3.57317924e-01 4.74704921e-01 -8.92120361e-01
5.76613657e-02 -2.77354985e-01 7.02961206e-01 -6.42151654e-01
4.11110312e-01 -4.66187090e-01 4.75743681e-01 1.04928557e-02
-5.26082814e-01 2.36133307e-01 -1.92992054e-02 3.75550359e-01
-1.90880209e-01 -3.04073002e-02 1.33087075e+00 -2.93118894e-01
-6.14501536e-01 3.21665674e-01 1.41434316e-02 -1.36094570e-01
6.52335227e-01 -5.83909035e-01 -6.35408282e-01 -4.85108405e-01
-5.40070713e-01 -3.98336709e-01 4.58183467e-01 5.83151996e-01
6.74131393e-01 -1.39030433e+00 -7.55268872e-01 5.23796797e-01
-1.75793767e-01 -3.36407833e-02 6.78512931e-01 6.36808813e-01
-6.00393653e-01 2.82200605e-01 -7.15374231e-01 -5.04967272e-01
-1.06797981e+00 5.37034690e-01 3.39171857e-01 -2.20877334e-01
-8.69884431e-01 6.56062365e-01 3.40097070e-01 -4.18268070e-02
-7.01450258e-02 -7.25871176e-02 -4.76813823e-01 -8.51426423e-02
7.64500499e-01 4.24799740e-01 -3.08737576e-01 -7.84785032e-01
1.65648982e-01 5.89238465e-01 -1.22726180e-01 -1.97926044e-01
1.44226444e+00 -7.22590744e-01 -3.28542650e-01 5.43268099e-02
1.13448226e+00 -1.15644582e-01 -1.64339840e+00 -2.82066196e-01
-6.18661165e-01 -6.56300843e-01 3.41093540e-01 -6.38004959e-01
-1.38438773e+00 6.60488069e-01 9.19467568e-01 -6.41637519e-02
1.38493502e+00 -2.78319955e-01 6.44911826e-01 -2.24928156e-01
1.75964952e-01 -1.13785148e+00 -8.65185261e-03 1.53774783e-01
9.29254591e-01 -9.45801377e-01 9.75841433e-02 -3.18140119e-01
-5.91295600e-01 1.06611550e+00 6.83413506e-01 -4.51318175e-02
2.40703776e-01 4.47555840e-01 1.84475386e-03 1.18673228e-01
-5.49062848e-01 1.18863419e-01 -1.48064131e-02 7.15194225e-01
3.94625425e-01 -2.02630147e-01 -1.30320445e-01 -4.57521975e-02
-3.92177582e-01 1.86804891e-01 8.19769323e-01 4.20031965e-01
-1.61227405e-01 -1.02062833e+00 -5.87462187e-01 3.12718779e-01
-4.35681701e-01 -3.48538578e-01 5.12237668e-01 6.78552151e-01
2.81692028e-01 5.68971753e-01 2.25454301e-01 -6.63532466e-02
3.32638398e-02 -6.83072209e-02 4.26862210e-01 -5.32028675e-02
-2.38337114e-01 5.88434711e-02 -1.38838425e-01 -7.50136197e-01
-5.21488845e-01 -3.11659276e-01 -9.84624863e-01 -4.19678986e-01
1.10510085e-02 -3.17621380e-01 6.81547821e-01 6.34475529e-01
5.04815340e-01 6.33772373e-01 3.75375509e-01 -9.36677814e-01
-3.49529296e-01 -9.26797092e-01 -5.91925502e-01 4.51293796e-01
2.99154639e-01 -5.76831818e-01 -4.51203704e-01 4.24480498e-01] | [11.139507293701172, -1.9931883811950684] |
40284940-4461-41ed-b837-ff896f60bba0 | zero-shot-clarifying-question-generation-for | 2301.12660 | null | https://arxiv.org/abs/2301.12660v2 | https://arxiv.org/pdf/2301.12660v2.pdf | Zero-shot Clarifying Question Generation for Conversational Search | A long-standing challenge for search and conversational assistants is query intention detection in ambiguous queries. Asking clarifying questions in conversational search has been widely studied and considered an effective solution to resolve query ambiguity. Existing work have explored various approaches for clarifying question ranking and generation. However, due to the lack of real conversational search data, they have to use artificial datasets for training, which limits their generalizability to real-world search scenarios. As a result, the industry has shown reluctance to implement them in reality, further suspending the availability of real conversational search interaction data. The above dilemma can be formulated as a cold start problem of clarifying question generation and conversational search in general. Furthermore, even if we do have large-scale conversational logs, it is not realistic to gather training data that can comprehensively cover all possible queries and topics in open-domain search scenarios. The risk of fitting bias when training a clarifying question retrieval/generation model on incomprehensive dataset is thus another important challenge. In this work, we innovatively explore generating clarifying questions in a zero-shot setting to overcome the cold start problem and we propose a constrained clarifying question generation system which uses both question templates and query facets to guide the effective and precise question generation. The experiment results show that our method outperforms existing state-of-the-art zero-shot baselines by a large margin. Human annotations to our model outputs also indicate our method generates 25.2\% more natural questions, 18.1\% more useful questions, 6.1\% less unnatural and 4\% less useless questions. | ['Qingyao Ai', 'Ming Wu', 'Nick Craswell', 'Corby Rosset', 'Yuancheng Tu', 'Zhenduo Wang'] | 2023-01-30 | null | null | null | null | ['natural-questions', 'conversational-search', 'question-generation'] | ['miscellaneous', 'natural-language-processing', 'natural-language-processing'] | [ 2.76810050e-01 4.46744055e-01 -9.13882330e-02 -2.86599517e-01
-1.19446874e+00 -7.83010781e-01 8.29196692e-01 -2.48177871e-01
-5.85980892e-01 8.68215680e-01 5.67023337e-01 -5.76515794e-01
-1.88728750e-01 -4.68008041e-01 -2.55509377e-01 -1.19585022e-01
6.65474892e-01 9.30492580e-01 2.38795668e-01 -8.41711521e-01
3.93053681e-01 -2.04262406e-01 -1.68149900e+00 4.54079807e-01
1.45021427e+00 6.71979249e-01 5.96285284e-01 8.11572552e-01
-5.65262556e-01 6.19792163e-01 -1.05781150e+00 -5.09143293e-01
2.07476486e-02 -5.28056502e-01 -1.43715096e+00 -2.70292193e-01
4.90714669e-01 -6.16530955e-01 -1.71386614e-01 6.88192427e-01
7.19896734e-01 4.04612839e-01 3.10762733e-01 -1.18722224e+00
-7.18823493e-01 4.77462769e-01 9.06396359e-02 3.51460159e-01
9.11720395e-01 3.24376196e-01 1.28859067e+00 -5.80830276e-01
5.80374658e-01 1.32257080e+00 1.67839006e-01 8.83452654e-01
-9.74716961e-01 -5.71862698e-01 1.75604522e-01 1.22693084e-01
-9.51131701e-01 -4.96172041e-01 5.58370054e-01 -1.37525395e-01
1.11695409e+00 9.32767630e-01 4.00870889e-01 1.37524593e+00
-3.24055254e-01 8.94846737e-01 9.70511198e-01 -3.92634004e-01
-1.63534880e-02 4.30409133e-01 3.16576302e-01 2.71014512e-01
3.72557864e-02 -9.86273587e-02 -4.67511594e-01 -6.27090991e-01
3.04016978e-01 -2.27666751e-01 -5.54623902e-01 8.05492550e-02
-9.80127275e-01 9.02397156e-01 3.08711290e-01 3.41150671e-01
-2.11590588e-01 -2.68712759e-01 4.90337946e-02 4.21344250e-01
2.62691736e-01 1.23129046e+00 -4.11737263e-01 -5.84113538e-01
-5.44479430e-01 8.74004543e-01 1.36878216e+00 1.04819453e+00
6.15896225e-01 -5.82249582e-01 -5.87307394e-01 1.33037519e+00
1.16689876e-01 5.32050431e-01 6.54117167e-01 -1.21872711e+00
6.74058735e-01 8.97907078e-01 5.49635947e-01 -8.61283123e-01
-1.75851420e-01 -3.00402969e-01 -3.22878152e-01 -5.60827971e-01
5.81606686e-01 -1.68693438e-01 -5.50827205e-01 1.76997209e+00
1.31545842e-01 -3.43331933e-01 1.38751760e-01 1.15230036e+00
1.17301655e+00 5.58884382e-01 -3.88113521e-02 -1.45367444e-01
1.78543437e+00 -1.03223789e+00 -8.87510478e-01 -5.13629556e-01
4.56709117e-01 -1.04507244e+00 1.73211741e+00 -4.18739319e-02
-9.55680549e-01 -3.19553077e-01 -5.64468205e-01 -3.65699947e-01
-2.92469412e-01 -2.10902646e-01 6.64699435e-01 5.22471428e-01
-8.79438996e-01 -1.05651766e-01 -1.09702513e-01 -6.71230078e-01
-2.58672059e-01 8.02573413e-02 1.39087036e-01 -1.71089962e-01
-1.61683607e+00 9.53094125e-01 -1.21797591e-01 -4.91126962e-02
-4.28869069e-01 -5.63206255e-01 -5.50384641e-01 3.43205005e-01
8.48826587e-01 -9.08447146e-01 1.99012482e+00 -3.12390953e-01
-1.27118957e+00 7.27458775e-01 -5.77678502e-01 -1.82130605e-01
3.27133834e-01 -2.97981143e-01 -2.28574499e-01 -3.29682902e-02
4.44357336e-01 8.28037620e-01 5.22969544e-01 -1.24042273e+00
-6.71123564e-01 -1.29502729e-01 7.52536595e-01 7.16518044e-01
-9.95180383e-02 -1.79389834e-01 -5.25480628e-01 -4.38881576e-01
-4.05525006e-02 -9.79409575e-01 -1.45496443e-01 -4.04821634e-01
-3.20560783e-01 -8.42753530e-01 7.52250314e-01 -5.79099596e-01
1.50596297e+00 -1.60154521e+00 -2.49929100e-01 -3.18150789e-01
3.31014097e-01 4.57620859e-01 -2.88837701e-01 7.68010437e-01
4.53440219e-01 3.51703256e-01 -6.69324771e-02 -1.18501827e-01
2.62096673e-01 2.24843532e-01 -5.83103120e-01 -5.56585789e-01
-1.53891258e-02 1.24565029e+00 -1.15183389e+00 -6.17986679e-01
-5.92394583e-02 -4.48922403e-02 -6.26726031e-01 6.59065902e-01
-6.07396364e-01 1.41731963e-01 -5.91172814e-01 5.64445674e-01
3.09618860e-01 -4.25390750e-01 -1.01247691e-01 1.54864430e-01
2.04690367e-01 7.89117098e-01 -8.36435199e-01 1.54760742e+00
-6.65931642e-01 6.62064373e-01 1.78734601e-01 -5.05846143e-01
7.78464615e-01 4.18730944e-01 1.74915969e-01 -1.02801251e+00
-1.24727279e-01 1.88829526e-01 -2.71437597e-03 -9.72622514e-01
1.02193868e+00 5.67847081e-02 -3.37197989e-01 8.72375667e-01
-2.77009428e-01 -6.00690186e-01 2.53381759e-01 4.59208190e-01
1.23711216e+00 -3.28332841e-01 -9.35358927e-02 -1.45135149e-01
4.94945973e-01 2.16093495e-01 1.97424456e-01 1.27733767e+00
-3.71495932e-01 4.22036350e-01 3.24338526e-01 -2.14867637e-01
-5.07417619e-01 -8.67074251e-01 2.33993962e-01 1.40976179e+00
4.20598298e-01 -4.07927275e-01 -8.60358953e-01 -6.83312714e-01
-9.00926962e-02 9.38287377e-01 -1.02023780e-01 -1.27172232e-01
-5.51947117e-01 -3.52208823e-01 5.60152054e-01 -2.66184080e-02
6.82273209e-01 -1.31106114e+00 -6.85399055e-01 1.85185000e-01
-1.17053688e+00 -1.18152809e+00 -8.36198509e-01 -2.01830357e-01
-5.28259158e-01 -1.21231186e+00 -7.89803982e-01 -8.53364348e-01
2.89173663e-01 8.82690966e-01 1.54122174e+00 3.60154897e-01
-1.85275421e-01 6.68841898e-01 -4.11683977e-01 -3.11865717e-01
-4.45989609e-01 4.96155441e-01 -3.79759461e-01 -5.23461580e-01
6.71812832e-01 -3.25058430e-01 -1.08402836e+00 7.82752812e-01
-8.80849242e-01 -6.73791617e-02 4.30819333e-01 9.85053778e-01
-2.09918663e-01 -5.87936461e-01 1.02500880e+00 -7.29693234e-01
1.89360857e+00 -4.07467008e-01 -2.80932784e-01 6.70453370e-01
-8.90897512e-01 2.04398394e-01 1.97130114e-01 -4.50514972e-01
-1.23276472e+00 -5.37764430e-01 -1.11587755e-01 1.84224218e-01
-1.09812453e-01 2.82115668e-01 9.54399705e-02 3.21884513e-01
1.04861140e+00 2.38884002e-01 3.91780660e-02 -3.60401303e-01
4.57622528e-01 1.16569972e+00 3.21585208e-01 -6.90486014e-01
5.73220134e-01 -9.25662182e-03 -8.67782533e-01 -9.57443953e-01
-7.37022996e-01 -8.79798174e-01 2.99292743e-01 -2.67828584e-01
6.93088412e-01 -6.06194794e-01 -1.11826551e+00 -8.63517728e-03
-1.38932669e+00 -1.93978325e-01 -4.68060896e-02 1.58376060e-02
-3.96373183e-01 6.01063550e-01 -3.64718229e-01 -1.13526809e+00
-6.73741162e-01 -1.25472784e+00 1.19392645e+00 4.53555346e-01
-8.86474967e-01 -5.21168590e-01 9.37138796e-02 1.29418397e+00
8.16277623e-01 -4.35191453e-01 9.70833838e-01 -8.18687439e-01
-7.65158355e-01 -8.11444409e-03 -2.06230551e-01 -2.14048605e-02
2.15899646e-01 -6.32214844e-01 -9.60155845e-01 -3.90970819e-02
1.58794835e-01 -5.96941233e-01 5.08662343e-01 -1.19560033e-01
8.25873435e-01 -7.63397694e-01 -2.70940781e-01 -1.82804599e-01
7.08353758e-01 3.24861497e-01 4.51236606e-01 1.53382868e-01
2.44521978e-03 1.00234640e+00 7.98266828e-01 2.11700767e-01
8.53746772e-01 8.78995299e-01 1.78587779e-01 1.89605027e-01
-2.69343555e-02 -3.94218177e-01 -1.08975947e-01 4.46175992e-01
3.42833012e-01 -6.19556546e-01 -8.70886862e-01 7.36750305e-01
-1.75253153e+00 -1.14112043e+00 1.17342010e-01 1.97278130e+00
1.21023417e+00 -1.48694381e-01 -4.11667489e-02 -3.66729468e-01
5.98155499e-01 2.25768402e-01 -5.18744826e-01 -3.85012269e-01
1.70646206e-01 9.53854471e-02 -2.66139477e-01 8.20337117e-01
-5.66516638e-01 1.02751887e+00 5.85811472e+00 5.13378620e-01
-7.84512162e-01 -8.99078622e-02 5.02377033e-01 3.14379744e-02
-7.46948361e-01 1.13946877e-01 -6.99458480e-01 5.36620378e-01
5.43101609e-01 -4.79788959e-01 6.29568517e-01 8.83160472e-01
1.34190053e-01 -4.30443525e-01 -1.02714515e+00 1.15548515e+00
8.15573260e-02 -1.10643852e+00 9.71343890e-02 -1.14513159e-01
3.95959496e-01 -1.68651119e-01 -1.63000569e-01 9.11038876e-01
3.23386371e-01 -1.00030577e+00 1.57759804e-02 3.69314641e-01
3.84960175e-01 -1.44284278e-01 6.76047444e-01 9.40353155e-01
-6.96280539e-01 -9.82441753e-03 -1.25924200e-01 -1.54532462e-01
3.19557011e-01 7.32159391e-02 -1.47132790e+00 8.04120675e-02
6.06696546e-01 -3.32519799e-01 -5.22344232e-01 7.18908489e-01
-2.14648787e-02 4.12588149e-01 -3.81433815e-01 -7.07580447e-01
3.03534180e-01 -8.88513699e-02 6.35655880e-01 8.77290547e-01
1.57242879e-01 4.83512759e-01 1.41572850e-02 8.83093059e-01
-2.30366871e-01 4.16762605e-02 -6.42229259e-01 -2.84219921e-01
8.14328313e-01 1.12226570e+00 -1.17815308e-01 -3.01488131e-01
-3.22341233e-01 1.04512525e+00 2.10832343e-01 5.19935548e-01
-4.78441834e-01 -4.39604968e-01 6.80911124e-01 7.15720877e-02
-2.27462471e-01 9.32706818e-02 1.41825806e-02 -1.18742108e+00
4.99221802e-01 -1.44691432e+00 4.55347329e-01 -9.25214410e-01
-1.21695876e+00 8.02825093e-01 1.48796529e-01 -8.46823156e-01
-1.05485606e+00 -4.44139391e-02 -5.54536045e-01 1.09789312e+00
-1.60607171e+00 -6.45642340e-01 -7.19547629e-01 2.86773145e-01
1.05924726e+00 3.70511301e-02 9.40546513e-01 1.10133007e-01
-1.26239419e-01 6.28985703e-01 -3.79582316e-01 -1.73469409e-01
9.50246215e-01 -1.08228552e+00 4.38651055e-01 3.88716787e-01
1.19224384e-01 1.15608513e+00 8.42193604e-01 -5.30365527e-01
-1.42108870e+00 -3.57207775e-01 1.44306970e+00 -7.97836244e-01
3.90054524e-01 -4.79725093e-01 -1.19131792e+00 2.25360721e-01
6.30262673e-01 -6.78487420e-01 7.29568779e-01 3.42291296e-01
-8.00568536e-02 -2.35655457e-02 -1.03733194e+00 1.01006651e+00
1.09472680e+00 -7.57035017e-01 -1.14092064e+00 5.89468360e-01
1.03575075e+00 -4.37113136e-01 -3.69870007e-01 3.73702943e-01
5.31537056e-01 -8.54669750e-01 8.45620990e-01 -5.90081990e-01
1.16428971e-01 9.82963368e-02 8.69864747e-02 -1.06770039e+00
1.50099009e-01 -1.01778769e+00 1.75327063e-02 1.15128827e+00
6.65300608e-01 -8.04052651e-01 7.74508834e-01 1.31413460e+00
-6.03322387e-02 -7.87041306e-01 -1.03927183e+00 -4.50881004e-01
-2.22038984e-01 -1.09761238e-01 7.49439895e-01 7.12803125e-01
3.74840945e-01 9.96972322e-01 -1.30903378e-01 -1.72001556e-01
7.52177984e-02 3.13524932e-01 1.09415591e+00 -1.02167118e+00
-2.33542427e-01 -5.60504079e-01 4.20397520e-01 -1.85768580e+00
1.18126802e-01 -5.65538466e-01 3.74668211e-01 -1.85853434e+00
1.57885000e-01 -4.47561502e-01 3.31611603e-01 3.43351364e-02
-6.94657087e-01 -4.03767854e-01 9.66250449e-02 2.65981555e-01
-6.12691641e-01 6.98058546e-01 1.48941875e+00 -2.33258113e-01
-3.33261281e-01 2.87180096e-01 -1.19703996e+00 2.47322917e-01
6.78493619e-01 -2.78711114e-02 -8.99309158e-01 -5.39785981e-01
2.79242903e-01 3.83091658e-01 3.00348252e-01 -5.03837764e-01
5.00565469e-01 -2.25431740e-01 -4.21715677e-01 -6.34972036e-01
5.77810168e-01 -6.10283673e-01 -2.80715525e-01 1.94468215e-01
-7.58350253e-01 2.69348055e-01 -4.50450107e-02 5.34961402e-01
-3.19073617e-01 -4.35795903e-01 8.00605044e-02 -4.00174230e-01
-5.34079611e-01 -5.74616082e-02 -4.75145876e-01 6.71695888e-01
5.40425062e-01 -1.18913069e-01 -6.87086105e-01 -1.20438421e+00
-1.69965938e-01 8.80514383e-01 1.29105613e-01 9.65123475e-01
5.07666171e-01 -9.22175586e-01 -6.10324919e-01 -1.36695698e-01
4.61016148e-01 4.21383232e-02 1.60309285e-01 2.39084497e-01
-4.99314703e-02 1.01921427e+00 4.59071904e-01 -5.28489172e-01
-1.25614476e+00 2.51804054e-01 2.09014550e-01 -4.18827951e-01
-1.07650571e-02 8.24802160e-01 2.16612205e-01 -7.36688733e-01
4.09070373e-01 -4.04494405e-01 -1.93893343e-01 -1.30162369e-02
4.49335843e-01 1.81636184e-01 1.01035018e-03 -9.83715355e-02
-1.56290472e-01 8.32327157e-02 -2.90010899e-01 -3.88959914e-01
7.35375166e-01 -3.88721108e-01 -3.80842537e-02 4.55397815e-02
9.35208678e-01 -6.03016280e-02 -6.18290246e-01 -3.80015194e-01
3.06560606e-01 -4.38561887e-01 -4.61254656e-01 -1.19119716e+00
-1.92903116e-01 6.41567111e-01 3.26849490e-01 7.26435065e-01
8.57237935e-01 3.65067363e-01 1.09286809e+00 1.01354337e+00
3.37020040e-01 -1.20597172e+00 3.65397632e-01 7.37335384e-01
1.33205009e+00 -1.62176824e+00 -4.14266437e-01 -3.77260029e-01
-6.51212394e-01 6.65780127e-01 1.06774545e+00 5.97805023e-01
-3.42564657e-02 -3.47957373e-01 4.96923894e-01 -3.20086837e-01
-1.04936492e+00 -3.23110759e-01 2.95097381e-01 3.48189533e-01
4.99555737e-01 -1.70952857e-01 -6.90410376e-01 6.69122875e-01
-5.90584815e-01 -2.62400120e-01 1.97722420e-01 8.05730283e-01
-4.27716494e-01 -1.25782692e+00 -2.58029729e-01 6.17767811e-01
-1.98820502e-01 -2.87768275e-01 -8.70170951e-01 6.68851137e-01
-5.27141213e-01 1.58441591e+00 -2.21614763e-01 -1.99822977e-01
5.67281783e-01 3.45848858e-01 -4.09907550e-02 -7.08066344e-01
-8.41779351e-01 -3.87344778e-01 6.17050469e-01 -5.12003899e-01
-4.98370565e-02 -2.26891503e-01 -7.87744999e-01 -1.02013052e-01
-7.24387467e-01 8.28281283e-01 4.64905947e-01 8.17910433e-01
8.61896336e-01 2.64986884e-02 4.87019897e-01 -1.07308887e-01
-1.07248473e+00 -1.33805621e+00 2.51194805e-01 6.00869298e-01
2.68461823e-01 -5.27289450e-01 -5.39770365e-01 -3.33256006e-01] | [12.1127290725708, 7.81272554397583] |
3cd677ff-8fff-4dc6-939c-46e2429fd013 | deep-simbad-active-landmark-based-self | 2109.02786 | null | https://arxiv.org/abs/2109.02786v1 | https://arxiv.org/pdf/2109.02786v1.pdf | Deep SIMBAD: Active Landmark-based Self-localization Using Ranking -based Scene Descriptor | Landmark-based robot self-localization has recently garnered interest as a highly-compressive domain-invariant approach for performing visual place recognition (VPR) across domains (e.g., time of day, weather, and season). However, landmark-based self-localization can be an ill-posed problem for a passive observer (e.g., manual robot control), as many viewpoints may not provide an effective landmark view. In this study, we consider an active self-localization task by an active observer and present a novel reinforcement learning (RL)-based next-best-view (NBV) planner. Our contributions are as follows. (1) SIMBAD-based VPR: We formulate the problem of landmark-based compact scene description as SIMBAD (similarity-based pattern recognition) and further present its deep learning extension. (2) VPR-to-NBV knowledge transfer: We address the challenge of RL under uncertainty (i.e., active self-localization) by transferring the state recognition ability of VPR to the NBV. (3) NNQL-based NBV: We regard the available VPR as the experience database by adapting nearest-neighbor approximation of Q-learning (NNQL). The result shows an extremely compact data structure that compresses both the VPR and NBV into a single incremental inverted index. Experiments using the public NCLT dataset validated the effectiveness of the proposed approach. | ['Tanaka Kanji'] | 2021-09-06 | null | null | null | null | ['visual-place-recognition'] | ['computer-vision'] | [-1.73249960e-01 -6.71945810e-02 -6.02998555e-01 -4.44287270e-01
-1.11935210e+00 -7.16512799e-01 4.81360912e-01 1.83796391e-01
-5.21618366e-01 6.81675553e-01 5.04934415e-02 -1.40050367e-01
-3.75912011e-01 -7.09649086e-01 -1.14582372e+00 -6.63666904e-01
-1.98028564e-01 4.36514378e-01 3.00306618e-01 -3.34148675e-01
4.82591331e-01 7.05900788e-01 -1.41507506e+00 -2.80010104e-01
8.49600434e-01 1.19127476e+00 7.39762664e-01 4.01006490e-01
3.23672175e-01 1.02869284e+00 -3.51499230e-01 4.12473977e-01
4.77970749e-01 -8.01131725e-02 -7.18939006e-01 6.37723655e-02
2.10270718e-01 -3.59603465e-01 -4.09943670e-01 9.18633640e-01
5.03294766e-01 6.60583198e-01 4.68743026e-01 -1.66326714e+00
-6.39030039e-01 -4.44168858e-02 -4.05743390e-01 -7.64852390e-02
6.04184926e-01 2.09929496e-01 8.46912801e-01 -7.79453814e-01
8.43235195e-01 1.09895289e+00 3.74599725e-01 3.63440603e-01
-1.12002897e+00 -2.34669447e-01 3.44041824e-01 8.27474773e-01
-1.73736942e+00 -3.75014484e-01 8.57517123e-01 -2.29951963e-01
1.03307760e+00 -3.01234536e-02 6.84176266e-01 1.05736375e+00
2.66069472e-01 9.09199297e-01 1.10725486e+00 -2.05318049e-01
9.92893934e-01 -6.41535521e-02 -4.41113651e-01 6.16338849e-01
2.13066842e-02 3.27729404e-01 -7.11984217e-01 -1.10305939e-02
7.90124118e-01 5.86576574e-02 -1.03350252e-01 -1.23047245e+00
-1.44410110e+00 9.00900960e-01 9.05455291e-01 -4.44335341e-02
-4.72955048e-01 1.80917099e-01 1.86751604e-01 3.67052168e-01
-1.03616994e-02 5.76937497e-01 -4.52904999e-01 -3.58556621e-02
-5.67032158e-01 7.63204172e-02 6.23129487e-01 1.22221994e+00
1.03244495e+00 9.83899981e-02 2.20120661e-02 6.96916401e-01
4.42379415e-01 9.17756557e-01 3.64744693e-01 -1.18579972e+00
3.98788929e-01 3.59584332e-01 6.12495422e-01 -1.07562315e+00
-5.43952763e-01 -1.12154983e-01 -8.09176445e-01 4.60748613e-01
-4.91810217e-02 -1.39366193e-02 -8.11214268e-01 1.85715592e+00
4.00034070e-01 1.01064757e-01 5.22944570e-01 1.00177777e+00
5.40785015e-01 8.34369361e-01 -1.88067779e-01 -2.70106405e-01
9.00545895e-01 -1.08237219e+00 -5.39846539e-01 -6.18586183e-01
3.69520336e-01 -1.09208725e-01 6.86390102e-01 2.97015429e-01
-5.26850224e-01 -5.78988492e-01 -1.23356462e+00 3.38096619e-02
-5.36707461e-01 6.37529641e-02 5.02944946e-01 -7.68291531e-03
-1.34420288e+00 2.72981554e-01 -8.29190791e-01 -6.12814546e-01
3.50818902e-01 4.60896373e-01 -9.37716126e-01 -3.91273975e-01
-1.04944360e+00 1.03305626e+00 5.72464705e-01 -8.46382454e-02
-1.35416913e+00 -2.08184004e-01 -1.38106465e+00 -2.92439550e-01
8.10006380e-01 -2.95707136e-01 1.16043282e+00 -7.78445721e-01
-1.67737246e+00 5.82000077e-01 -5.02321899e-01 -6.00765049e-01
2.80807912e-01 -7.94924982e-03 -3.25899452e-01 2.03147605e-01
4.38565224e-01 8.24032784e-01 8.21325898e-01 -1.56199217e+00
-8.38557959e-01 -5.42339385e-01 2.41794169e-01 8.10247779e-01
2.80120522e-01 -5.42955399e-01 -3.20268214e-01 -1.26893461e-01
6.47365570e-01 -9.70393717e-01 -4.50898558e-01 2.26003751e-01
-1.03135943e-01 -3.64419281e-01 6.35599613e-01 -4.59945977e-01
5.07388115e-01 -2.14598227e+00 1.90395176e-01 9.53619555e-02
-1.05383083e-01 -1.16334282e-01 -2.42347866e-01 6.62405193e-01
2.19504431e-01 -3.61786216e-01 -4.01584506e-02 -4.18144137e-01
5.81819043e-02 6.59381688e-01 -5.25639594e-01 8.72242570e-01
-1.46184534e-01 9.14000034e-01 -1.30338752e+00 -4.99183387e-01
4.07387108e-01 1.06749117e-01 -3.51864845e-01 3.39255661e-01
-1.51405290e-01 6.73464596e-01 -4.02184963e-01 8.32257330e-01
6.50383055e-01 -1.34047002e-01 5.14852665e-02 -1.65901139e-01
-3.20302844e-01 -1.28635526e-01 -1.40765715e+00 2.25528646e+00
-5.17933726e-01 5.69514871e-01 -5.87305576e-02 -9.82361853e-01
1.09705603e+00 4.62208409e-03 5.48502028e-01 -1.16750586e+00
-3.00851703e-01 2.14549705e-01 -4.42837894e-01 -2.71665096e-01
6.87393010e-01 1.01852298e-01 -4.72542107e-01 1.27278706e-02
3.02544773e-01 -1.46694615e-01 -1.98558077e-01 -8.29182267e-02
1.12390566e+00 4.36960697e-01 9.44584429e-01 1.07467510e-01
3.63906354e-01 3.47694695e-01 8.11469734e-01 1.01110530e+00
-6.77121878e-01 4.05550838e-01 1.04260780e-01 -4.04045343e-01
-8.06667924e-01 -1.24329281e+00 1.90716043e-01 9.24207330e-01
8.56433690e-01 3.31380740e-02 -9.93447825e-02 -7.71254897e-01
1.38209462e-02 8.46162975e-01 -4.53762323e-01 -2.83804148e-01
-4.01618004e-01 -7.63832256e-02 1.39082983e-01 5.35442054e-01
8.10124576e-01 -1.06737435e+00 -1.09549403e+00 1.22481368e-01
-3.00811112e-01 -1.05240440e+00 -3.07150602e-01 6.14687264e-01
-5.27605236e-01 -7.90168047e-01 -6.02963328e-01 -8.66708279e-01
6.03845716e-01 5.09074867e-01 7.45323181e-01 -6.43925250e-01
2.69828528e-01 9.58823025e-01 -5.53444445e-01 -2.49854267e-01
-1.08936414e-01 -2.32694834e-01 5.07487237e-01 -4.48524654e-02
1.28575400e-01 -3.18419278e-01 -5.97592592e-01 5.20488679e-01
-4.10581887e-01 -2.81371772e-01 6.58285081e-01 8.55811357e-01
1.12321842e+00 2.59564873e-02 5.06203234e-01 -1.66213289e-01
1.84599146e-01 -4.59826708e-01 -1.11047280e+00 3.20242763e-01
-5.08546710e-01 9.71378759e-02 4.52310145e-01 -4.23583359e-01
-8.94257009e-01 5.62238514e-01 9.34556276e-02 -6.64532006e-01
-4.12949592e-01 4.17590052e-01 -3.45320672e-01 -4.40635324e-01
5.40395081e-01 6.94246709e-01 -3.19831930e-02 4.31682691e-02
5.28922379e-01 6.38752282e-01 7.14977384e-01 -2.82169312e-01
8.79204631e-01 6.16173446e-01 2.03761250e-01 -6.77682817e-01
-6.27446473e-01 -9.08989966e-01 -7.61149645e-01 -1.73666671e-01
9.11679029e-01 -1.23866272e+00 -1.00177145e+00 1.92031950e-01
-1.19204724e+00 -5.94006300e-01 -5.76849163e-01 5.43498695e-01
-1.33220208e+00 2.46572644e-01 1.53885558e-01 -9.50434685e-01
1.32929623e-01 -1.04691482e+00 1.15052235e+00 2.64321983e-01
1.59276590e-01 -8.28636706e-01 4.48052645e-01 1.47414386e-01
1.35019422e-01 2.30360702e-01 3.41700882e-01 -7.42351830e-01
-9.52957988e-01 -4.29075299e-04 -5.31280087e-03 7.87055939e-02
2.95726676e-02 -9.02148068e-01 -8.40374231e-01 -5.54588437e-01
-3.71121578e-02 -6.17399991e-01 5.34173310e-01 3.33575487e-01
6.38502300e-01 -2.86352754e-01 -4.70281690e-01 7.18468666e-01
1.75405741e+00 7.22224832e-01 4.60937291e-01 7.26658285e-01
4.86512572e-01 3.51048082e-01 1.23130596e+00 5.51568389e-01
9.31357384e-01 8.44939232e-01 9.28123891e-01 1.67792775e-02
1.91248298e-01 -5.91316342e-01 4.87503469e-01 3.92070681e-01
2.97274768e-01 -3.91018093e-01 -9.93547797e-01 6.64100826e-01
-2.23973203e+00 -9.45845842e-01 6.00765646e-01 2.23531771e+00
2.45525360e-01 -3.42702031e-01 -2.78878987e-01 -2.91307777e-01
4.84967321e-01 3.78653109e-01 -1.08561110e+00 -8.88491496e-02
-1.46528482e-01 -4.80729789e-01 7.73129404e-01 4.94121283e-01
-1.40499914e+00 1.06201768e+00 5.18396807e+00 6.42989516e-01
-1.01336229e+00 1.82714343e-01 1.26717642e-01 3.98224175e-01
2.19595626e-01 5.56564219e-02 -6.69551611e-01 1.30492687e-01
5.04971147e-01 3.69639061e-02 8.09684157e-01 1.28396893e+00
3.65818217e-02 -5.11653483e-01 -1.41143739e+00 1.51202583e+00
3.48689616e-01 -1.07661545e+00 -2.48619720e-01 -1.22475781e-01
6.53463125e-01 4.45111871e-01 1.56825811e-01 6.67917848e-01
3.89527291e-01 -7.83249497e-01 1.03336442e+00 5.45484245e-01
8.67729187e-01 -6.06095731e-01 6.36510253e-01 7.41157591e-01
-1.33566344e+00 -6.27615869e-01 -5.05671918e-01 1.75021887e-01
1.35059878e-01 -2.09288806e-01 -1.02023125e+00 9.77663636e-01
6.45163417e-01 8.95205498e-01 -4.16035622e-01 1.20371139e+00
-1.92180276e-01 -1.12123139e-01 -2.71762133e-01 -5.95226027e-02
3.67477417e-01 -9.81878191e-02 7.51421094e-01 4.72313344e-01
3.70343029e-01 1.46812394e-01 4.38270062e-01 7.99784660e-01
2.69214988e-01 -2.07563192e-01 -9.40269291e-01 4.29303467e-01
5.82069218e-01 9.23699677e-01 -5.66307902e-01 -5.68443239e-02
-3.64929110e-01 1.31601322e+00 5.26969492e-01 5.25379300e-01
-4.95938122e-01 -2.39824727e-01 5.92546642e-01 -3.85298014e-01
5.88524818e-01 -3.55577320e-01 2.08970696e-01 -1.09681892e+00
-9.07062739e-02 -7.23539472e-01 3.47171456e-01 -1.07485592e+00
-1.07967007e+00 4.37169969e-01 8.05754140e-02 -1.63178587e+00
-3.68419737e-01 -3.80088031e-01 -8.54233727e-02 6.02702558e-01
-1.76214743e+00 -1.18979049e+00 -3.80336106e-01 7.59535193e-01
6.63954258e-01 -3.29149812e-01 9.30128753e-01 -1.41854897e-01
2.63049342e-02 2.58602768e-01 3.87489229e-01 2.05341987e-02
6.75840974e-01 -1.30816424e+00 1.77920908e-01 8.08204770e-01
3.07172269e-01 3.50064605e-01 5.02331078e-01 -7.12848306e-01
-1.93555880e+00 -1.43013287e+00 4.87449735e-01 -5.16927660e-01
3.33951384e-01 -3.62368673e-01 -4.21371222e-01 8.05163801e-01
-1.30284637e-01 4.42059278e-01 1.69637576e-01 -2.28042707e-01
-1.95428327e-01 -3.54001909e-01 -1.34056056e+00 6.06299520e-01
1.06654108e+00 -6.37246013e-01 -6.63800597e-01 3.02874833e-01
9.42852974e-01 -6.71284556e-01 -4.88917261e-01 2.93356925e-01
2.71639317e-01 -8.01704109e-01 1.09908557e+00 -1.91954225e-01
-3.07182670e-01 -6.84224904e-01 -7.91021168e-01 -1.52642357e+00
-3.84007961e-01 -3.69049639e-01 -1.32118329e-01 8.27254772e-01
2.96130218e-03 -7.37179816e-01 5.86795866e-01 1.02007158e-01
-5.30158393e-02 -5.61083674e-01 -1.47282732e+00 -1.11299121e+00
-5.19937992e-01 -5.16063511e-01 4.29681122e-01 6.42149150e-01
-8.54409337e-02 1.74102828e-01 -4.07537162e-01 7.46679246e-01
7.39668310e-01 3.11573833e-01 8.00428391e-01 -1.00630772e+00
-1.75319344e-01 2.43840963e-01 -7.26674438e-01 -1.39411759e+00
2.89926618e-01 -7.66385257e-01 7.01342404e-01 -1.93836558e+00
-5.38736470e-02 -5.08206725e-01 -4.71397758e-01 5.72278082e-01
5.06255746e-01 -2.88739443e-01 2.37596273e-01 3.02118599e-01
-1.48041034e+00 8.79230201e-01 1.04168987e+00 -4.15962309e-01
-4.50169951e-01 3.11008114e-02 -2.82026589e-01 5.57159722e-01
6.74738526e-01 -3.63971651e-01 -6.90877318e-01 -2.96122879e-01
2.63532043e-01 6.12461269e-01 5.54939449e-01 -1.14605176e+00
7.09969223e-01 -3.92500490e-01 2.68784046e-01 -8.11458170e-01
6.24238372e-01 -1.22164941e+00 -2.35881031e-01 4.39719737e-01
-2.91917294e-01 -2.59579998e-03 2.19429545e-02 1.08661854e+00
-2.02611983e-01 -7.50436708e-02 5.67172408e-01 -2.70112902e-01
-1.58984113e+00 3.82280439e-01 -2.55515635e-01 -2.11899225e-02
1.29990554e+00 -2.99167365e-01 -2.16804460e-01 -5.91329634e-01
-5.27859211e-01 4.90257114e-01 5.66005170e-01 5.21381676e-01
1.04478145e+00 -1.31618357e+00 -2.02259913e-01 2.51868218e-01
5.68982005e-01 3.24195236e-01 3.28073919e-01 6.06443524e-01
-2.19155744e-01 5.92595279e-01 -3.21821183e-01 -8.40098083e-01
-7.98993826e-01 9.44002509e-01 2.98275083e-01 -1.41943274e-02
-5.23473203e-01 5.46398759e-01 1.91570222e-01 -8.20639253e-01
4.23882872e-01 -1.65486157e-01 -4.07355249e-01 -3.90376709e-02
3.79335523e-01 3.77556860e-01 -1.47917494e-01 -9.33152080e-01
-6.64928198e-01 6.52295768e-01 3.27704012e-01 -2.63833702e-01
1.42822611e+00 -5.60548663e-01 1.10881828e-01 7.46170461e-01
1.10961723e+00 -4.30399984e-01 -1.58482897e+00 -3.84635568e-01
7.42290169e-03 -2.97823608e-01 -6.17406005e-03 -7.74181485e-01
-4.16211218e-01 5.04097879e-01 1.04465568e+00 -2.73069948e-01
8.96031439e-01 2.02266216e-01 3.51177275e-01 1.14351749e+00
1.28304207e+00 -1.25666499e+00 3.10919851e-01 7.69764245e-01
1.06920540e+00 -1.70523798e+00 -1.62562221e-01 1.69925317e-01
-9.12836552e-01 8.11185479e-01 7.32036591e-01 9.21473652e-03
3.76651496e-01 -2.29188606e-01 6.35478273e-02 3.12779727e-03
-5.80235779e-01 -3.71643990e-01 2.08517119e-01 1.16406333e+00
-7.03415751e-01 1.35899514e-01 5.21508992e-01 3.02319288e-01
8.63787383e-02 -6.77291229e-02 3.33017975e-01 1.20201337e+00
-4.58340257e-01 -4.82097119e-01 -3.00827801e-01 -3.79263833e-02
4.60623145e-01 2.67337084e-01 9.03156549e-02 6.49319530e-01
1.52445555e-01 1.13293231e+00 9.86579210e-02 -3.28908265e-01
4.46842492e-01 -3.07570547e-01 3.47270817e-01 -5.54879963e-01
1.69739887e-01 -2.98532397e-01 -3.58501375e-01 -1.07031453e+00
-3.80882263e-01 -6.35339379e-01 -1.39149177e+00 2.71769613e-01
1.64010935e-02 -6.47037476e-02 7.57867515e-01 8.86438549e-01
4.12621468e-01 2.33847439e-01 8.67534101e-01 -1.00348806e+00
-7.46064067e-01 -4.56259906e-01 -6.91154063e-01 4.16942500e-02
8.52514803e-01 -8.83687913e-01 -2.47766703e-01 -2.75339305e-01] | [7.417845726013184, -1.98819100856781] |
c62d6c28-8b8f-4ab2-b532-fcb013416b24 | planning-for-complex-non-prehensile | 2303.13352 | null | https://arxiv.org/abs/2303.13352v1 | https://arxiv.org/pdf/2303.13352v1.pdf | Planning for Complex Non-prehensile Manipulation Among Movable Objects by Interleaving Multi-Agent Pathfinding and Physics-Based Simulation | Real-world manipulation problems in heavy clutter require robots to reason about potential contacts with objects in the environment. We focus on pick-and-place style tasks to retrieve a target object from a shelf where some `movable' objects must be rearranged in order to solve the task. In particular, our motivation is to allow the robot to reason over and consider non-prehensile rearrangement actions that lead to complex robot-object and object-object interactions where multiple objects might be moved by the robot simultaneously, and objects might tilt, lean on each other, or topple. To support this, we query a physics-based simulator to forward simulate these interaction dynamics which makes action evaluation during planning computationally very expensive. To make the planner tractable, we establish a connection between the domain of Manipulation Among Movable Objects and Multi-Agent Pathfinding that lets us decompose the problem into two phases our M4M algorithm iterates over. First we solve a multi-agent planning problem that reasons about the configurations of movable objects but does not forward simulate a physics model. Next, an arm motion planning problem is solved that uses a physics-based simulator but does not search over possible configurations of movable objects. We run simulated and real-world experiments with the PR2 robot and compare against relevant baseline algorithms. Our results highlight that M4M generates complex 3D interactions, and solves at least twice as many problems as the baselines with competitive performance. | ['Maxim Likhachev', 'Dhruv Mauria Saxena'] | 2023-03-23 | null | null | null | null | ['motion-planning'] | ['robots'] | [ 1.52439505e-01 2.78050601e-01 -3.23640071e-02 3.59377787e-02
-5.70914328e-01 -1.13975799e+00 4.81122583e-01 2.19550043e-01
-4.26437318e-01 7.58908510e-01 -2.75414675e-01 -3.72512192e-01
-5.37635744e-01 -8.10104609e-01 -1.02238882e+00 -3.75074744e-01
-5.16588569e-01 1.41953838e+00 6.71341538e-01 -6.60577714e-01
4.19125021e-01 8.20431590e-01 -1.53919959e+00 6.17961437e-02
3.64584893e-01 4.37269241e-01 9.59347248e-01 9.46844757e-01
2.01965421e-01 5.46186030e-01 -4.27336723e-01 2.64883190e-01
5.34837961e-01 -5.04871346e-02 -1.38733852e+00 2.59511381e-01
-1.75840244e-01 -4.59762156e-01 6.30267411e-02 7.27762163e-01
1.03588000e-01 5.58259666e-01 7.24691033e-01 -1.77091420e+00
-6.55879155e-02 7.40770340e-01 -4.96871173e-01 -1.15244463e-01
8.38137090e-01 6.12413347e-01 6.51654959e-01 -4.11360115e-01
9.25917685e-01 1.73198497e+00 4.43070568e-02 2.18145266e-01
-1.11174583e+00 -2.85358950e-02 4.56255615e-01 1.25404224e-01
-1.06594300e+00 -8.95112976e-02 2.37188637e-01 -1.75350741e-01
1.27022481e+00 5.15005469e-01 6.43052995e-01 7.15862274e-01
6.50329590e-01 6.24940217e-01 5.76805949e-01 -3.70418400e-01
3.61115307e-01 -3.69526446e-01 -1.77075431e-01 6.90998554e-01
5.00900865e-01 2.81895231e-02 -3.90972607e-02 -2.83442438e-01
8.04685235e-01 -1.25228316e-01 -2.79535931e-02 -1.01431680e+00
-1.51531029e+00 5.85377276e-01 4.79629785e-01 3.85089219e-02
-5.52981257e-01 4.82276171e-01 -2.18956638e-02 3.16228479e-01
-5.96072257e-01 1.00046289e+00 -6.15390062e-01 -1.86963722e-01
6.54039606e-02 1.01753581e+00 1.16868925e+00 1.48828185e+00
7.13277459e-01 -4.98580158e-01 2.65479535e-01 2.78193176e-01
2.71837503e-01 4.49428827e-01 -1.50970846e-01 -1.60872960e+00
9.13413823e-01 4.01169300e-01 8.00575018e-01 -8.69585812e-01
-9.03029740e-01 3.21118832e-01 -1.33758098e-01 8.26197386e-01
3.58887345e-01 -9.30333808e-02 -8.56261611e-01 1.34788442e+00
7.36076951e-01 -4.72830802e-01 1.11725867e-01 1.09978640e+00
2.32772201e-01 6.15602314e-01 -1.50428772e-01 -3.98803316e-03
1.41775072e+00 -1.29921329e+00 -2.81291664e-01 -5.17547071e-01
8.49337459e-01 -6.65908456e-01 6.94353580e-01 4.06741917e-01
-1.52865028e+00 -2.23632351e-01 -8.62308621e-01 -2.26789750e-02
-1.61183164e-01 -3.67577285e-01 7.43975759e-01 -2.22479418e-01
-8.96957219e-01 7.78886020e-01 -1.10946155e+00 -5.26705682e-01
-1.08571887e-01 6.53051496e-01 -4.20675904e-01 -2.40208864e-01
-5.46547413e-01 1.55209064e+00 6.23979926e-01 2.14612305e-01
-1.26630962e+00 6.05455860e-02 -8.82602930e-01 -7.40323635e-03
1.00964677e+00 -8.92221272e-01 1.70900428e+00 -3.69564593e-01
-1.60011578e+00 4.19718146e-01 1.02474950e-01 -3.54782604e-02
4.20621544e-01 -1.09762721e-01 2.93170840e-01 1.41800940e-01
3.66913140e-01 8.53024423e-01 3.86638969e-01 -1.65420628e+00
-8.01801264e-01 -2.13003814e-01 6.58503175e-01 7.15069652e-01
8.39217901e-01 -2.24005535e-01 -5.35076082e-01 -7.95316622e-02
4.84700322e-01 -1.76081264e+00 -5.99830508e-01 -3.94284911e-02
-6.15403533e-01 -2.07783505e-01 4.87728715e-01 1.22707911e-01
1.74366638e-01 -1.77804852e+00 6.46385133e-01 2.78113186e-01
-2.00155333e-01 -4.52976882e-01 -5.53381801e-01 9.06216443e-01
1.57574669e-01 -1.48724496e-01 2.27487218e-02 -8.40599611e-02
1.59713283e-01 6.07668936e-01 -2.21225992e-01 3.51026148e-01
-1.71233982e-01 9.54728901e-01 -1.07597435e+00 -3.77141714e-01
1.62321091e-01 -1.53685957e-01 -7.39966094e-01 5.47232628e-02
-8.08464110e-01 4.60021704e-01 -8.98516476e-01 6.13287628e-01
5.58498740e-01 3.42939235e-02 4.57068384e-01 2.50732929e-01
-3.53608161e-01 2.46973053e-01 -1.49457014e+00 1.79400027e+00
-3.20126325e-01 5.34102097e-02 4.90832120e-01 -4.52519745e-01
2.85367936e-01 -1.74059063e-01 4.03979421e-01 -1.79765284e-01
1.12513557e-01 2.31647745e-01 1.57428548e-01 -6.06589317e-01
8.11413229e-01 5.56075759e-02 -4.29854423e-01 5.20693302e-01
-5.27122855e-01 -1.12587559e+00 1.40521213e-01 4.17450890e-02
1.51095557e+00 5.13253331e-01 2.29685143e-01 -1.32586926e-01
-1.59177762e-02 9.39404607e-01 1.52523860e-01 1.18837476e+00
1.97000027e-01 1.97239131e-01 2.44484887e-01 -2.92598248e-01
-7.40643978e-01 -1.16889453e+00 5.06397486e-01 1.14575672e+00
1.17182148e+00 -7.79209435e-02 -3.95050764e-01 -2.78802335e-01
1.44297227e-01 1.03803027e+00 -2.68523186e-01 3.00268661e-02
-1.06252015e+00 -2.44549021e-01 -1.22463658e-01 4.18238223e-01
1.70209765e-01 -1.32419205e+00 -1.58199954e+00 3.93611073e-01
-2.53560454e-01 -8.85254323e-01 -3.04425299e-01 4.91747767e-01
-8.11924160e-01 -1.39950931e+00 -2.15032831e-01 -9.79238689e-01
9.77341056e-01 6.71371698e-01 8.57769489e-01 1.37236953e-01
-3.41265887e-01 7.15243578e-01 -4.67609555e-01 -3.90511990e-01
-5.08313060e-01 8.24619532e-02 1.56125044e-02 -1.04186785e+00
-4.28286970e-01 -5.27840674e-01 -3.86960298e-01 4.86582875e-01
-5.83272874e-01 2.75747448e-01 8.62423003e-01 2.72981048e-01
4.94927049e-01 6.60108626e-01 -2.34391347e-01 -2.11943880e-01
6.56230330e-01 -4.20729935e-01 -7.58293152e-01 2.56286144e-01
1.42582372e-01 1.97445378e-01 2.39560276e-01 -6.47118807e-01
-8.44883442e-01 4.69024390e-01 6.01724744e-01 -1.23379156e-01
-1.78210601e-01 5.48748255e-01 -2.01697573e-01 -2.58586686e-02
4.45397347e-01 -2.04307109e-01 -6.78865314e-02 -2.50036955e-01
4.65440035e-01 -5.58130555e-02 4.37657446e-01 -1.04926741e+00
8.74651730e-01 3.89964432e-01 4.52899516e-01 -3.47458631e-01
9.79246479e-03 -1.29767612e-01 -6.04340553e-01 -5.26423082e-02
7.58236229e-01 -4.06147897e-01 -1.61691129e+00 2.03983888e-01
-1.47342455e+00 -9.63926852e-01 -2.06859872e-01 3.49240661e-01
-1.18324614e+00 8.90612975e-02 -3.74731332e-01 -8.54295254e-01
3.86001945e-01 -1.54287934e+00 1.23615456e+00 -8.98933932e-02
-5.58990300e-01 -1.67874485e-01 -1.69800505e-01 -8.24062228e-02
2.27473438e-01 3.28219354e-01 1.06990194e+00 -4.78920490e-01
-1.27262628e+00 1.44816504e-03 1.28507331e-01 -9.97938037e-01
1.10848255e-01 -3.93201649e-01 2.83033907e-01 -4.19864446e-01
-2.48478442e-01 -2.08037883e-01 1.13116682e-01 1.23520888e-01
8.47941518e-01 -3.32854450e-01 -1.10703242e+00 -1.39398262e-01
1.13431799e+00 5.71284056e-01 3.51477653e-01 9.55129802e-01
3.52670103e-01 8.80004168e-01 1.20936084e+00 2.84815013e-01
7.46641278e-01 1.02615809e+00 1.02435505e+00 5.33514857e-01
3.59514922e-01 -3.10576055e-02 1.88578367e-01 -7.58478837e-03
-6.75129145e-02 -5.53868055e-01 -1.07178545e+00 4.04991865e-01
-2.20648766e+00 -7.11937785e-01 -2.94963062e-01 1.89649463e+00
5.03788888e-01 3.51431876e-01 1.27203345e-01 -2.13169843e-01
4.41047430e-01 -3.42237949e-01 -7.69535661e-01 -3.12342107e-01
6.32221401e-01 -2.60739595e-01 8.01432550e-01 9.68863130e-01
-7.49445617e-01 1.06387806e+00 6.03614759e+00 1.85787156e-01
-4.85626936e-01 -1.70539945e-01 -1.90566674e-01 -3.68702173e-01
-8.25016424e-02 4.07340616e-01 -6.88281357e-01 -5.66510148e-02
3.97469431e-01 6.38860837e-02 8.77531409e-01 1.13469100e+00
1.40375733e-01 -8.94292831e-01 -1.62381017e+00 4.69404399e-01
-3.63889217e-01 -8.67200613e-01 -9.24167857e-02 1.29555002e-01
2.10841238e-01 -2.44654473e-02 -1.90025672e-01 4.37954009e-01
1.14101088e+00 -8.27447653e-01 1.22192276e+00 2.75788695e-01
1.32920444e-02 -5.23715556e-01 2.00657472e-01 9.35255885e-01
-1.23174024e+00 -3.28695714e-01 -1.57251030e-01 -4.75781530e-01
8.03879559e-01 -2.00530291e-01 -1.23900080e+00 6.21569514e-01
6.12177968e-01 -2.48695716e-01 2.90729344e-01 1.15105593e+00
-2.54893333e-01 -5.47219574e-01 -6.51468575e-01 -4.45717990e-01
3.18148732e-01 -1.61888406e-01 8.59747469e-01 5.18789351e-01
2.63856739e-01 6.65528297e-01 7.08865821e-01 8.54125261e-01
5.88610530e-01 -4.79696959e-01 -6.54797316e-01 2.90487915e-01
5.64537764e-01 9.34465110e-01 -1.19683707e+00 -2.79977649e-01
2.61234909e-01 9.98158097e-01 2.28569448e-01 3.94748896e-01
-9.93486762e-01 -9.41027254e-02 6.24365807e-01 1.44853070e-01
2.70900875e-01 -8.15144837e-01 1.47506399e-02 -6.51658833e-01
1.25870064e-01 -8.67405176e-01 4.37997095e-02 -1.21735096e+00
-6.24906659e-01 3.28674078e-01 6.81086957e-01 -1.01352870e+00
-3.99979949e-01 -4.89055216e-01 -4.52345908e-01 6.62934721e-01
-1.21974826e+00 -9.68764782e-01 -3.65024030e-01 2.12244600e-01
9.60448682e-01 4.59647417e-01 7.77117848e-01 -5.34291565e-01
-4.12933826e-02 -3.85942459e-01 -4.68431383e-01 -5.83519399e-01
1.62039801e-01 -9.61486220e-01 5.22712409e-01 3.28784257e-01
-5.50876379e-01 7.03723252e-01 9.75888431e-01 -1.08398926e+00
-2.13296771e+00 -6.46209657e-01 3.07006270e-01 -7.20084131e-01
4.74500060e-01 -2.88611621e-01 -4.82903898e-01 1.13566101e+00
-1.38982192e-01 -3.13301504e-01 -1.35099933e-01 -1.06330432e-01
2.26392418e-01 5.64339995e-01 -1.07170105e+00 1.07279968e+00
1.42687571e+00 2.90735871e-01 -9.23369825e-01 5.95666587e-01
8.06532800e-01 -1.08801413e+00 -2.95819223e-01 4.74119157e-01
5.18791020e-01 -4.31112796e-01 1.08211327e+00 -6.92886293e-01
2.33441219e-01 -6.82272613e-01 -2.53178149e-01 -1.41237020e+00
-4.53602552e-01 -7.71552801e-01 9.45601091e-02 5.55277467e-01
4.50417280e-01 -6.39541447e-01 5.93707800e-01 9.68562603e-01
-2.44860336e-01 -5.51902473e-01 -9.14839506e-01 -9.92642581e-01
-1.86800748e-01 -3.38841319e-01 6.99757993e-01 5.23353457e-01
3.44301373e-01 1.68335676e-01 6.99253455e-02 7.26847947e-01
3.47598612e-01 4.98624533e-01 1.11231124e+00 -9.25465047e-01
-4.28337663e-01 -4.76759523e-01 1.19410597e-01 -1.43670487e+00
2.27980167e-01 -6.11391306e-01 7.77417243e-01 -1.95399451e+00
2.37357779e-03 -9.87664223e-01 5.04257262e-01 7.43314505e-01
3.06098431e-01 -5.66359580e-01 7.31388807e-01 4.73164320e-01
-8.89305472e-01 2.35368222e-01 1.64116454e+00 -1.66445717e-01
-6.24428928e-01 9.78774726e-02 -3.60292763e-01 8.37626278e-01
6.23701334e-01 -4.20454293e-01 -4.67701823e-01 -8.03042829e-01
4.35301393e-01 8.13847125e-01 4.08443183e-01 -8.87238503e-01
4.39894527e-01 -8.80652130e-01 6.42556697e-02 -6.11480713e-01
8.35828722e-01 -1.24403644e+00 4.57078338e-01 8.80607724e-01
-2.48897612e-01 4.36562270e-01 3.62450778e-01 5.25168478e-01
5.60936034e-01 -5.03197730e-01 1.90844104e-01 -5.41654825e-01
-6.51232600e-01 1.02591077e-02 -9.63154018e-01 -4.84976143e-01
1.60426879e+00 -3.06471139e-01 -4.31774378e-01 -3.46869141e-01
-9.34041798e-01 8.40620518e-01 7.31200457e-01 6.53521597e-01
5.52625716e-01 -9.98549521e-01 -1.81035042e-01 -3.10273767e-01
-1.55049652e-01 7.03122616e-01 -8.48763362e-02 6.27933741e-01
-7.72044301e-01 3.05647045e-01 -2.79249966e-01 -4.27440226e-01
-1.10736895e+00 9.01498020e-01 9.89622548e-02 -2.93950085e-02
-7.41437495e-01 6.78841114e-01 2.50013232e-01 -6.39445305e-01
5.94709665e-02 -7.27918684e-01 -3.72741707e-02 -5.19606709e-01
1.58443853e-01 4.60895538e-01 -4.03528363e-01 -2.64500916e-01
-6.04801416e-01 6.29059553e-01 5.79051636e-02 -5.31624973e-01
1.38449800e+00 -2.32683897e-01 -1.96445420e-01 2.41507262e-01
4.29878503e-01 -1.44171864e-01 -1.11565185e+00 2.94113994e-01
4.54339832e-02 -4.87661839e-01 -6.55176878e-01 -8.99007201e-01
-2.80485839e-01 1.47529110e-01 -9.98391733e-02 2.86259443e-01
5.27476132e-01 4.07742649e-01 5.93595922e-01 1.17469466e+00
1.28628922e+00 -8.97137225e-01 4.26585674e-01 7.76511312e-01
1.35284257e+00 -8.16842258e-01 1.99727729e-01 -8.04990709e-01
-5.78895986e-01 1.07307684e+00 8.90834987e-01 4.81362222e-03
5.73144481e-02 3.99662793e-01 -3.15501720e-01 -4.01895672e-01
-5.68872213e-01 -1.63276091e-01 -2.89254248e-01 5.41563570e-01
-5.29534876e-01 4.52289991e-02 9.77436453e-02 5.76852001e-02
-5.07623196e-01 -1.94316491e-01 6.88316464e-01 1.76710963e+00
-9.46400821e-01 -1.08921075e+00 -7.31514931e-01 1.30355284e-01
3.61252785e-01 5.05641997e-01 -4.66608047e-01 1.12932599e+00
-8.95563439e-02 1.09625041e+00 8.12876597e-02 4.29055430e-02
6.33800030e-01 -3.23489040e-01 9.32154536e-01 -8.93490314e-01
-3.06419611e-01 -1.17575012e-01 2.75263876e-01 -9.72911417e-01
-1.58593848e-01 -7.55070746e-01 -2.10604191e+00 -9.31608081e-02
-3.69023532e-01 7.23379850e-02 7.29910553e-01 1.09514987e+00
4.35349405e-01 4.89787489e-01 1.78397566e-01 -1.70619965e+00
-5.91367662e-01 -5.21863639e-01 -6.14900701e-02 8.17147046e-02
2.99482644e-01 -1.04001153e+00 -1.11129582e-01 -3.68553966e-01] | [4.721461772918701, 0.9402682781219482] |
a860398a-bc47-485a-bdd7-e3af58437bf1 | span-based-discontinuous-constituency-parsing-1 | null | null | https://aclanthology.org/2020.emnlp-main.219 | https://aclanthology.org/2020.emnlp-main.219.pdf | Span-based discontinuous constituency parsing: a family of exact chart-based algorithms with time complexities from O(n\^6) down to O(n\^3) | We introduce a novel chart-based algorithm for span-based parsing of discontinuous constituency trees of block degree two, including ill-nested structures. In particular, we show that we can build variants of our parser with smaller search spaces and time complexities ranging from O(n{\^{}}6) down to O(n{\^{}}3). The cubic time variant covers 98{\%} of constituents observed in linguistic treebanks while having the same complexity as continuous constituency parsers. We evaluate our approach on German and English treebanks (Negra, Tiger, and DPTB) and report state-of-the-art results in the fully supervised setting. We also experiment with pre-trained word embeddings and Bert-based neural networks. | ['Caio Corro'] | null | null | null | null | emnlp-2020-11 | ['constituency-parsing'] | ['natural-language-processing'] | [ 1.15024403e-01 5.46911657e-01 -3.06673758e-02 -3.91847670e-01
-1.36686599e+00 -1.09970522e+00 -7.65354633e-02 5.92024744e-01
-7.32061207e-01 6.58633649e-01 3.48361820e-01 -1.20660949e+00
1.73227221e-01 -9.42985058e-01 -5.41120470e-01 -3.14285994e-01
-5.24945498e-01 5.07815599e-01 5.86775541e-01 -4.90408540e-01
-3.60749736e-02 3.28958154e-01 -8.83934200e-01 2.91383445e-01
6.73475027e-01 6.84320629e-01 1.61566198e-01 1.19542253e+00
-5.56159139e-01 4.70646530e-01 -4.88420457e-01 -9.20167446e-01
1.54204249e-01 -2.71365587e-02 -1.06998479e+00 -3.25217366e-01
7.43343890e-01 2.68469807e-02 -2.43139431e-01 1.02381420e+00
2.93739617e-01 -1.20843589e-01 1.99631210e-02 -3.81233931e-01
-4.70937371e-01 1.20140779e+00 -2.87839025e-01 7.29868531e-01
2.17752486e-01 -2.81124026e-01 1.86273944e+00 -5.70440590e-01
6.38457716e-01 1.27487028e+00 7.45054781e-01 5.57572901e-01
-1.35801589e+00 -4.47432905e-01 6.61637783e-01 -2.74038017e-01
-6.65014267e-01 -4.84050483e-01 5.44053435e-01 -4.52359952e-02
1.47631097e+00 1.04140304e-01 3.67863446e-01 6.57582760e-01
1.71488285e-01 7.43623495e-01 1.06587422e+00 -9.46467638e-01
1.22427836e-01 -5.04544079e-01 8.49111319e-01 1.17533088e+00
5.24363816e-01 -6.90007284e-02 -2.80387312e-01 -2.59964615e-01
6.99048102e-01 -6.76722705e-01 1.52240753e-01 3.03678334e-01
-8.84570301e-01 1.21628952e+00 -4.55423258e-03 5.67762196e-01
2.91362125e-02 3.01355004e-01 7.12594450e-01 4.09309328e-01
6.13492906e-01 1.82809964e-01 -1.00710773e+00 -1.77877590e-01
-5.65390050e-01 1.49670929e-01 1.06716192e+00 1.27262664e+00
6.05245531e-01 -1.47833386e-02 1.94513708e-01 8.03288519e-01
-1.58863232e-01 2.26966202e-01 1.40884623e-01 -8.43027472e-01
1.17751908e+00 3.44892949e-01 -6.87844530e-02 2.92355637e-03
-7.40582526e-01 -1.76733047e-01 -4.13296163e-01 -1.43271759e-01
9.06073093e-01 -5.93148530e-01 -8.49280834e-01 1.88300955e+00
3.73694569e-01 -3.53933364e-01 1.51387155e-01 1.76361561e-01
5.45982182e-01 7.14302361e-01 3.34480405e-01 -3.72527242e-01
2.02382469e+00 -7.82888830e-01 -6.16611123e-01 -6.59179807e-01
1.07256722e+00 -6.86134219e-01 9.89092708e-01 2.16455489e-01
-1.49594605e+00 -3.34611863e-01 -8.95857632e-01 -3.78528476e-01
-1.86429814e-01 -6.58195615e-02 1.01738822e+00 1.03143692e+00
-1.23315549e+00 7.11034715e-01 -1.20290136e+00 1.74292102e-02
1.08743176e-01 3.38062555e-01 -4.79155838e-01 2.37383228e-03
-1.02552974e+00 6.81360602e-01 4.42093164e-01 1.30486310e-01
-1.96559101e-01 -5.72713315e-01 -1.21737230e+00 8.30607563e-02
1.49172306e-01 -1.28651187e-01 1.54693937e+00 -2.15631455e-01
-1.34160697e+00 1.09245169e+00 -3.76755744e-01 -5.04624426e-01
-6.61697909e-02 -3.56214643e-01 -3.81234527e-01 4.86826673e-02
2.09064350e-01 2.64765799e-01 -1.02120332e-01 -5.88271320e-01
-1.19769442e+00 -4.61080849e-01 4.42628920e-01 -1.79419771e-01
-1.97161317e-01 5.68286777e-01 -2.43945032e-01 -5.69734812e-01
5.75904489e-01 -7.98604786e-01 -4.85043287e-01 -4.82293218e-01
-4.48106080e-01 -7.44187236e-01 1.73136704e-02 -9.94706810e-01
1.52555358e+00 -1.99293423e+00 -5.17401919e-02 -1.65523350e-01
-5.18989703e-03 2.92958766e-01 -2.80345708e-01 5.93245685e-01
-6.33377815e-03 2.66824663e-01 -4.29283142e-01 -3.05844218e-01
2.18194067e-01 5.15909255e-01 -8.63235369e-02 2.21425891e-01
4.31297958e-01 6.82008445e-01 -9.02366519e-01 -5.92016518e-01
-1.18783057e-01 -1.79859489e-01 -6.95514798e-01 -4.71565500e-02
-4.27288234e-01 -2.11864989e-02 -4.56729978e-01 6.86413825e-01
6.62603736e-01 2.59739637e-01 9.74138737e-01 2.26163134e-01
-4.08679783e-01 1.14991713e+00 -1.08217931e+00 1.63227522e+00
-7.86913574e-01 3.74536723e-01 4.94903237e-01 -8.49541724e-01
7.56842256e-01 3.47613782e-01 6.10489324e-02 -4.56936717e-01
7.14056268e-02 3.70545268e-01 2.25009590e-01 -1.59815148e-01
5.42160809e-01 -2.94796735e-01 -8.43752265e-01 2.94649631e-01
2.40112305e-01 -3.51907965e-03 5.75751007e-01 -1.04274526e-02
1.64990795e+00 9.95672569e-02 4.14060146e-01 -4.71512794e-01
2.64153928e-01 1.03911966e-01 9.13661420e-01 6.31504416e-01
-3.30969393e-01 6.58555776e-02 1.03475392e+00 -7.67948210e-01
-1.04879785e+00 -1.28562605e+00 -3.46275598e-01 1.84471190e+00
-3.74803156e-01 -5.53391278e-01 -7.76455283e-01 -8.06328773e-01
-4.31506723e-01 7.98313260e-01 -4.58462894e-01 6.23902798e-01
-1.64616370e+00 -7.03631043e-01 7.90706992e-01 9.68292058e-01
1.63328737e-01 -1.27833641e+00 -6.51329219e-01 9.64539826e-01
2.44231187e-02 -1.58947372e+00 -4.79775697e-01 9.67308164e-01
-1.27961731e+00 -1.09577990e+00 7.22635612e-02 -1.54549253e+00
3.24398756e-01 -3.35394412e-01 1.58857083e+00 -1.15106374e-01
-8.44482705e-02 -3.03046823e-01 -5.54858267e-01 -2.09692731e-01
-5.89681327e-01 4.19547379e-01 -3.32460195e-01 -1.00202608e+00
5.30543685e-01 -6.72687113e-01 -1.43573716e-01 -2.30701864e-01
-5.77303767e-01 -2.68735439e-01 4.40561473e-01 1.07167506e+00
6.24774396e-01 -2.74577618e-01 2.11196005e-01 -1.36253166e+00
5.21399915e-01 -1.01152375e-01 -1.06363070e+00 3.08633000e-01
-7.44165331e-02 3.84713382e-01 9.75650072e-01 -9.25493687e-02
-1.00874698e+00 2.71843046e-01 -8.21245670e-01 4.29042011e-01
-2.47835115e-01 3.68695706e-01 -1.99669018e-01 4.36371893e-01
4.64951128e-01 -1.16231151e-01 -7.42232978e-01 -7.98002064e-01
7.39003658e-01 2.64606446e-01 7.15193927e-01 -9.92844164e-01
5.31730950e-01 5.89641221e-02 -1.10974275e-01 -7.41227627e-01
-9.38543558e-01 -2.94306874e-01 -7.15763390e-01 7.24972308e-01
1.03223217e+00 -8.70071948e-01 -4.72021550e-01 -9.93290395e-02
-1.58289409e+00 -3.76268506e-01 -1.87134817e-01 2.85891563e-01
-4.03990656e-01 4.94164824e-01 -1.32424867e+00 -7.98800170e-01
-5.64681709e-01 -7.61296570e-01 1.07344830e+00 -1.01108126e-01
-7.07830340e-02 -1.04125094e+00 3.15140575e-01 5.78067340e-02
-1.86114565e-01 2.47306630e-01 1.54870903e+00 -8.80818009e-01
-3.73879820e-01 -8.65197927e-02 -1.26640841e-01 2.86890209e-01
7.48841092e-02 -2.92882085e-01 -6.76159561e-01 -6.90831989e-02
-2.33682141e-01 -6.45916313e-02 9.39237058e-01 3.73907715e-01
5.58759034e-01 -4.41773027e-01 -1.92199469e-01 4.34268892e-01
1.37145960e+00 8.80665630e-02 3.13654214e-01 2.95440823e-01
3.75407666e-01 6.69103146e-01 5.02598405e-01 1.24499194e-01
5.40449023e-01 1.63056016e-01 2.27916792e-01 3.00839603e-01
5.51766641e-02 -2.70273417e-01 5.19617617e-01 1.10776401e+00
-3.90495621e-02 -1.39619753e-01 -8.42005432e-01 1.05438757e+00
-1.40552902e+00 -6.22902274e-01 -2.85300076e-01 1.73166800e+00
1.17688894e+00 5.95512271e-01 1.12467341e-01 3.72842401e-01
8.78313780e-01 3.70166063e-01 1.91044994e-02 -1.38565087e+00
7.71015733e-02 1.40800142e+00 7.42379904e-01 7.24394321e-01
-1.42983639e+00 1.69132411e+00 6.48728800e+00 5.26700795e-01
-5.27612031e-01 3.55731875e-01 4.46933925e-01 1.28514037e-01
-3.78056765e-01 2.96919465e-01 -1.16913331e+00 -7.41399359e-03
1.29010999e+00 1.79243803e-01 8.66994485e-02 9.24172878e-01
-3.66932184e-01 7.54507780e-02 -1.19812274e+00 3.60836565e-01
-6.12174392e-01 -1.38696134e+00 -6.89958274e-01 -9.58617851e-02
3.84511799e-01 3.44531536e-01 -4.09608543e-01 4.51314867e-01
1.10106504e+00 -5.91491818e-01 7.50769794e-01 -6.26884162e-01
9.91036773e-01 -6.81973636e-01 5.98874390e-01 2.92934000e-01
-1.71089113e+00 -2.20963247e-02 -5.45287848e-01 -2.76223719e-01
5.95541239e-01 5.97436607e-01 -5.11978984e-01 5.28948486e-01
6.36154354e-01 -1.25967160e-01 -1.84084803e-01 4.65097040e-01
-7.13842988e-01 1.03026402e+00 -7.12787211e-01 -4.32794958e-01
5.56306601e-01 -1.03198476e-01 3.67361516e-01 1.62093472e+00
-2.33997894e-03 5.58171868e-01 1.92104951e-01 1.29813895e-01
-2.43968770e-01 1.79762006e-01 -2.65999854e-01 5.38405776e-02
7.41852880e-01 1.14988446e+00 -8.53914499e-01 -2.39702985e-01
-5.40052652e-01 6.62440896e-01 9.75898266e-01 -1.58439681e-01
-5.20535886e-01 -6.34531319e-01 6.91709518e-01 -1.73448354e-01
1.00768757e+00 -6.90857708e-01 -4.02139604e-01 -9.37574387e-01
3.74468356e-01 -6.26034021e-01 8.33704650e-01 1.30141273e-01
-1.24219072e+00 1.00545025e+00 8.05732608e-02 -4.51726794e-01
-1.87762812e-01 -1.14615285e+00 -7.59833217e-01 7.79127121e-01
-1.53363216e+00 -9.32985663e-01 5.02006829e-01 7.32576028e-02
5.22952497e-01 1.11769311e-01 1.35671186e+00 7.66199678e-02
-4.98820037e-01 8.24252546e-01 -1.02450676e-01 6.26112103e-01
3.91852520e-02 -1.62170148e+00 1.44899738e+00 1.06969881e+00
4.96898979e-01 4.43259716e-01 5.41897833e-01 -2.88540065e-01
-1.43868697e+00 -1.03948665e+00 1.63479102e+00 -3.54863405e-01
1.09073615e+00 -7.69026101e-01 -6.04170740e-01 1.04854655e+00
2.82233268e-01 2.19640315e-01 6.50220990e-01 7.45269656e-01
-6.41997457e-01 3.69480625e-02 -1.03058970e+00 5.04275084e-01
1.52966535e+00 -4.37149704e-01 -9.99792337e-01 3.78316075e-01
1.07022572e+00 -6.87047064e-01 -9.75245893e-01 -1.41434791e-02
4.14895296e-01 -8.48544657e-01 5.59604824e-01 -8.23380947e-01
2.11029008e-01 3.77025425e-01 -3.41990858e-01 -9.95052397e-01
-4.37608451e-01 -8.94612908e-01 6.67422637e-02 1.13165081e+00
8.33827853e-01 -7.46344924e-01 9.61510241e-01 3.00723046e-01
-5.13591170e-01 -7.04340875e-01 -1.50566387e+00 -7.52861321e-01
9.51539755e-01 -6.74581826e-01 4.63370949e-01 4.10036594e-01
1.04552798e-01 3.69424880e-01 1.03624791e-01 3.94469798e-01
4.03340429e-01 2.34858379e-01 1.44307956e-01 -9.97512162e-01
-4.80368376e-01 -2.85844594e-01 -1.85959697e-01 -1.24384069e+00
3.23816866e-01 -9.00901139e-01 1.34081915e-01 -1.42109203e+00
-3.36558163e-01 -8.67108643e-01 -2.46725500e-01 6.98460221e-01
-2.02036679e-01 -1.13814056e-01 1.32014260e-01 -5.45832336e-01
-2.71699399e-01 -2.50388570e-02 7.60130882e-01 6.63500205e-02
-1.28569841e-01 -1.82087585e-01 -7.92002082e-01 8.43441904e-01
8.47479641e-01 -7.81727731e-01 2.25803941e-01 -1.05325770e+00
2.95063525e-01 5.06590486e-01 -1.69715166e-01 -6.96346343e-01
-3.76244485e-02 2.12918054e-02 -2.26239011e-01 -6.89319372e-01
2.95556467e-02 -2.53106534e-01 -7.06111252e-01 5.33021450e-01
-2.61249423e-01 8.38849664e-01 4.87799287e-01 3.11734080e-01
-1.09507501e-01 -6.00572586e-01 4.63962704e-01 -4.49021608e-01
-3.89042497e-01 3.05494606e-01 -4.33068693e-01 6.24156833e-01
4.84622121e-01 3.53273973e-02 -2.72621781e-01 2.66765982e-01
-8.63158822e-01 1.49551108e-01 -1.53976560e-01 3.19135413e-02
1.81549683e-01 -8.45779121e-01 -8.22312236e-01 -8.29860047e-02
-1.51605859e-01 3.20212126e-01 -2.19183471e-02 2.20949486e-01
-9.81273115e-01 5.24448633e-01 1.92816705e-01 -8.11521634e-02
-1.13274860e+00 2.50125021e-01 -7.38314390e-02 -1.01409519e+00
-7.19635069e-01 1.29936051e+00 -8.61913711e-02 -5.71794093e-01
3.96993905e-02 -9.33833838e-01 1.05309464e-01 -7.68854469e-02
2.75341213e-01 7.95055777e-02 2.27263793e-01 -3.31479311e-01
-6.28201962e-01 3.66337210e-01 -1.86314240e-01 -1.84353247e-01
1.50671041e+00 3.25426966e-01 -2.15264872e-01 9.80962738e-02
1.09071350e+00 4.01068121e-01 -9.69240904e-01 -2.55134493e-01
6.38201654e-01 2.67486393e-01 -3.03252518e-01 -4.57065910e-01
-6.82256520e-01 9.35536325e-01 2.15836748e-01 4.65303719e-01
9.97123420e-01 2.78409779e-01 1.26650989e+00 6.18921995e-01
5.82621515e-01 -1.14739406e+00 -4.90972072e-01 9.57070231e-01
2.69689173e-01 -8.26308370e-01 -2.47988388e-01 -6.78191483e-01
-2.05636203e-01 1.33802366e+00 1.40327781e-01 -6.66916072e-01
7.13019311e-01 7.41645634e-01 3.13826767e-03 2.43280996e-02
-9.71611142e-01 -2.57701218e-01 -3.57865214e-01 3.80795240e-01
8.58266950e-01 4.29904550e-01 -7.82214522e-01 9.82053876e-01
-7.07857370e-01 -5.73404551e-01 3.25564831e-01 1.36184919e+00
-6.77662253e-01 -1.79358327e+00 -3.61214876e-02 9.22722071e-02
-1.00602591e+00 -6.93648636e-01 2.16411445e-02 1.02635527e+00
1.28774837e-01 1.11457133e+00 2.08810449e-01 -1.64466910e-02
5.49679518e-01 4.37035888e-01 8.87426853e-01 -9.40102875e-01
-1.03252769e+00 9.02832076e-02 1.03324723e+00 -4.39956754e-01
-7.54378065e-02 -6.83781266e-01 -1.48666036e+00 -3.54102433e-01
-4.47034568e-01 1.99863195e-01 4.73128140e-01 8.91662240e-01
-1.07598737e-01 3.96013141e-01 4.76999342e-01 -3.32860261e-01
-8.30664754e-01 -1.07673669e+00 -4.15678054e-01 2.22555120e-02
3.44682559e-02 -9.69392583e-02 -8.79053399e-02 -1.29477471e-01] | [10.319816589355469, 9.662688255310059] |
ac57006e-b4fc-4983-a700-ebe65023b7af | learning-with-noisy-labels-by-adaptive | 2306.04502 | null | https://arxiv.org/abs/2306.04502v2 | https://arxiv.org/pdf/2306.04502v2.pdf | Learning with Noisy Labels by Adaptive Gradient-Based Outlier Removal | An accurate and substantial dataset is essential for training a reliable and well-performing model. However, even manually annotated datasets contain label errors, not to mention automatically labeled ones. Previous methods for label denoising have primarily focused on detecting outliers and their permanent removal - a process that is likely to over- or underfilter the dataset. In this work, we propose AGRA: a new method for learning with noisy labels by using Adaptive GRAdient-based outlier removal. Instead of cleaning the dataset prior to model training, the dataset is dynamically adjusted during the training process. By comparing the aggregated gradient of a batch of samples and an individual example gradient, our method dynamically decides whether a corresponding example is helpful for the model at this point or is counter-productive and should be left out for the current update. Extensive evaluation on several datasets demonstrates AGRA's effectiveness, while a comprehensive results analysis supports our initial hypothesis: permanent hard outlier removal is not always what model benefits the most from. | ['Benjamin Roth', 'Lena Zellinger', 'Anastasiia Sedova'] | 2023-06-07 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [ 2.59168595e-01 8.64034221e-02 7.21003637e-02 -6.55134559e-01
-9.69560683e-01 -3.68528783e-01 2.02837631e-01 7.94892132e-01
-5.52744329e-01 5.92606664e-01 9.92172360e-02 -7.33087510e-02
-8.93384293e-02 -4.54334736e-01 -5.80622733e-01 -8.11124027e-01
1.52315587e-01 5.28847337e-01 1.22372828e-01 2.53385454e-01
3.67691189e-01 4.25099105e-01 -1.67432678e+00 1.63326219e-01
8.95920634e-01 8.16590309e-01 -9.44166407e-02 4.82438296e-01
-1.47688329e-01 8.83858681e-01 -8.21786106e-01 -2.41908774e-01
3.09663326e-01 -6.48464561e-01 -5.98624945e-01 3.76003921e-01
6.14359200e-01 -6.53967559e-02 2.95452535e-01 1.10263753e+00
4.37214375e-01 2.09879026e-01 4.41406131e-01 -1.20354128e+00
-1.89184710e-01 8.54228020e-01 -2.76732504e-01 1.94076791e-01
1.30985081e-01 2.37951115e-01 8.36751640e-01 -8.73486638e-01
6.41576767e-01 9.26999629e-01 1.13410294e+00 4.31286514e-01
-1.42753971e+00 -5.73278129e-01 3.38815451e-01 -9.90396366e-02
-1.32358754e+00 -5.05222440e-01 7.81677008e-01 -5.44209778e-01
5.37097454e-01 3.12982440e-01 5.55482745e-01 8.58911693e-01
9.17281210e-03 3.92654002e-01 1.19860649e+00 -3.36071312e-01
5.11249721e-01 1.55441821e-01 4.08034027e-01 6.09804809e-01
6.24329865e-01 -2.35601142e-01 -7.33806133e-01 -4.23085988e-01
2.03060389e-01 -1.01836249e-01 -1.56652346e-01 -2.01318011e-01
-9.29916322e-01 5.52261829e-01 1.03318863e-01 2.37741947e-01
-6.15124583e-01 1.32661611e-01 4.65014577e-01 4.80571091e-01
7.18830884e-01 7.42035389e-01 -5.95188737e-01 -3.43847573e-01
-1.44110644e+00 2.49658570e-01 5.53240418e-01 5.24723053e-01
9.56804454e-01 -4.61085401e-02 -2.35834092e-01 9.30904984e-01
2.51694202e-01 6.13471074e-03 4.40060586e-01 -1.01281750e+00
2.11508051e-02 1.02351248e+00 1.72898635e-01 -9.02928889e-01
-5.16950250e-01 -5.16979575e-01 -5.37509263e-01 2.76288748e-01
7.28234529e-01 -4.43929173e-02 -1.06084991e+00 1.48679948e+00
5.93390942e-01 3.23943943e-01 -1.24943249e-01 6.41551614e-01
4.97713774e-01 2.97199339e-01 3.16213816e-01 -5.15173852e-01
8.88227284e-01 -5.82831621e-01 -7.95714378e-01 -2.53239751e-01
9.05194104e-01 -7.88385868e-01 1.24527478e+00 7.63075292e-01
-8.08490753e-01 -4.33236450e-01 -9.69612837e-01 1.49058804e-01
-2.27900147e-01 6.85923174e-02 2.62313873e-01 5.28164744e-01
-7.00701773e-01 9.68185008e-01 -8.07746530e-01 -3.92049044e-01
5.35031199e-01 2.40029514e-01 -3.10852528e-01 -1.72106430e-01
-6.39316440e-01 5.97671092e-01 4.86313969e-01 5.59080765e-02
-6.74411237e-01 -6.75024867e-01 -6.03533208e-01 6.34597498e-04
5.92416048e-01 -3.24772775e-01 1.22461736e+00 -9.04420972e-01
-1.06844950e+00 8.37614477e-01 -2.68670678e-01 -5.59665799e-01
8.45437407e-01 -4.48072553e-01 -2.44756058e-01 -1.99690133e-01
1.54615164e-01 4.32304382e-01 1.09583187e+00 -1.71977949e+00
-8.69861126e-01 -4.64232922e-01 -3.58566523e-01 2.81157289e-02
-3.77866000e-01 7.03208745e-02 -3.32543552e-01 -8.15441966e-01
6.39247179e-01 -8.98224354e-01 -2.87548780e-01 -2.19387740e-01
-4.98925626e-01 -2.19312221e-01 9.17751074e-01 -6.75677657e-01
1.54488683e+00 -2.25984979e+00 -2.88763255e-01 4.84697282e-01
2.28709295e-01 5.94482422e-02 1.77681714e-01 1.65431365e-01
-1.03589945e-01 4.41865623e-01 -4.46999341e-01 -6.21842146e-01
-2.26810813e-01 2.32650116e-01 -2.64709562e-01 4.75316286e-01
1.42812282e-01 2.15955287e-01 -1.08047223e+00 -5.05096614e-01
2.74824966e-02 1.27304658e-01 -3.60154927e-01 2.40785450e-01
-2.16057241e-01 6.26600564e-01 -7.78877065e-02 7.60719001e-01
5.33881247e-01 -2.16580089e-02 -4.45858054e-02 -1.04987822e-01
4.76284027e-02 1.30912051e-01 -1.54836154e+00 1.20881987e+00
-1.97790354e-01 4.65765238e-01 5.51411808e-02 -6.26741767e-01
9.74182069e-01 1.31008506e-01 6.34872019e-01 -4.48121488e-01
-1.75703652e-02 4.18101788e-01 -1.75248086e-01 -6.78439558e-01
5.98928690e-01 -9.89184529e-03 1.97130162e-02 4.97058094e-01
-1.96918696e-01 -2.05118489e-02 5.57483375e-01 2.15652078e-01
1.30543590e+00 1.47780165e-01 1.17202051e-01 -1.32371649e-01
3.19597930e-01 1.09893523e-01 1.02166831e+00 1.05030298e+00
-2.69758791e-01 7.97023058e-01 5.60201347e-01 -4.16590452e-01
-8.79007638e-01 -7.31786668e-01 -1.99955627e-01 9.97228384e-01
-1.45054176e-01 -5.31520188e-01 -7.74450243e-01 -1.11979365e+00
1.48401797e-01 9.26695287e-01 -5.06007791e-01 -3.18133831e-01
-4.61085588e-01 -8.78635764e-01 3.27053666e-01 2.61597782e-01
2.78409094e-01 -1.11929655e+00 -6.34018362e-01 3.88505697e-01
-2.26276785e-01 -6.79112375e-01 -3.39033514e-01 5.84966362e-01
-1.05430710e+00 -1.29273605e+00 -8.83783959e-03 -4.38106090e-01
1.09888983e+00 -4.52739373e-02 1.15245891e+00 4.77392316e-01
-4.11465392e-02 2.02752143e-01 -4.09107476e-01 -5.40922463e-01
-7.93243408e-01 -7.89733138e-03 1.35979280e-01 1.33244768e-01
3.74424458e-01 -2.89720505e-01 -2.76154846e-01 2.33666137e-01
-1.00442564e+00 -4.11510289e-01 2.91451603e-01 7.38371789e-01
8.43681931e-01 3.64757210e-01 7.40361691e-01 -1.47517133e+00
8.00571680e-01 -3.96155715e-01 -5.47337770e-01 1.33734196e-01
-1.06857610e+00 2.34598313e-02 7.13893235e-01 -3.99165630e-01
-9.44087386e-01 4.10689592e-01 -6.10158481e-02 -2.97062993e-01
-4.18067038e-01 5.55498302e-01 -1.43235683e-01 2.76214629e-01
9.62711692e-01 -2.94412196e-01 -2.88022548e-01 -6.24128938e-01
-4.76178303e-02 5.22738099e-01 5.19054294e-01 -4.08389270e-01
6.48134232e-01 3.66710722e-01 -2.21022055e-01 -7.08208203e-01
-1.16061652e+00 -6.84651673e-01 -6.95993841e-01 -5.03675759e-01
4.81355011e-01 -7.02750742e-01 -2.18206823e-01 6.65931344e-01
-6.12196803e-01 -4.88113135e-01 -6.26854897e-01 2.60082036e-01
-1.53924674e-01 2.64158785e-01 -3.37051809e-01 -9.22359228e-01
-2.58392375e-02 -9.95536149e-01 8.42425227e-01 2.21510157e-01
-5.41949213e-01 -8.80077481e-01 1.04795560e-01 4.55118984e-01
1.95565760e-01 3.40072632e-01 7.90707767e-01 -9.64189410e-01
-2.58318275e-01 -5.60269058e-01 1.70845732e-01 5.93490660e-01
2.51619875e-01 2.99181491e-01 -1.15674174e+00 -3.06165487e-01
1.38946205e-01 -1.86955422e-01 9.01523709e-01 3.43780369e-01
1.30651414e+00 -2.72230327e-01 -1.29855409e-01 3.17284048e-01
1.45999503e+00 -5.52813783e-02 5.16709924e-01 4.86021608e-01
7.50548005e-01 4.02479291e-01 9.40714538e-01 5.01618624e-01
1.96708634e-01 3.23665202e-01 5.01572073e-01 -1.18312776e-01
3.38356569e-02 -1.56796291e-01 4.15756464e-01 7.13297844e-01
3.18818003e-01 -3.48382443e-02 -8.88537109e-01 5.79057217e-01
-1.84796774e+00 -7.58242965e-01 -4.19744313e-01 2.58751631e+00
9.34555054e-01 5.26546359e-01 3.41590829e-02 4.96594310e-01
5.99872231e-01 -8.97297189e-02 -6.21329427e-01 -2.02722937e-01
-2.66377367e-02 -2.04888880e-01 5.04771054e-01 3.81484032e-01
-1.20171046e+00 7.71013618e-01 6.24087334e+00 4.79152113e-01
-9.19351459e-01 -2.46674046e-02 6.93371475e-01 -2.11731121e-01
-5.85332811e-02 2.12875322e-01 -7.11900771e-01 6.56871796e-01
9.52575147e-01 5.68076111e-02 2.77805179e-01 9.89646614e-01
6.24012172e-01 -5.49857795e-01 -1.19988096e+00 7.10744321e-01
8.58902112e-02 -1.04209101e+00 -1.61393523e-01 -2.85784644e-03
7.93942034e-01 -2.23029643e-01 -3.05269033e-01 2.56994665e-01
4.99666899e-01 -6.97281301e-01 8.68914783e-01 7.57181883e-01
1.27727613e-01 -8.53028655e-01 9.68392551e-01 3.71725827e-01
-6.65415585e-01 -3.53502333e-01 -2.53039211e-01 2.72485781e-02
2.23429967e-02 1.33809686e+00 -8.60018134e-01 2.60056198e-01
9.52973187e-01 5.67870140e-01 -1.11736560e+00 1.28368723e+00
-3.06115866e-01 1.23488307e+00 -4.52497035e-01 4.70024526e-01
-9.19496790e-02 -2.54843146e-01 5.52783608e-01 1.31390762e+00
2.68933624e-01 -2.86446095e-01 5.92746258e-01 4.54444051e-01
-2.98527718e-01 3.81685436e-01 -5.85289001e-01 1.82607934e-01
6.80031180e-01 1.26626337e+00 -1.11236131e+00 -5.85731089e-01
-1.69016868e-01 7.11080849e-01 2.74326533e-01 1.89692691e-01
-6.12982810e-01 -2.05615744e-01 3.12857360e-01 3.00827980e-01
8.32044557e-02 7.65013695e-02 -8.18107188e-01 -8.83999109e-01
5.19143417e-02 -9.98616993e-01 5.82784712e-01 -5.92711329e-01
-1.31476045e+00 2.33950078e-01 -2.59634078e-01 -1.24253035e+00
4.61609848e-02 -1.07381918e-01 -5.12801826e-01 4.55070764e-01
-1.12222552e+00 -6.83483899e-01 -5.48701465e-01 1.72998160e-01
4.09440249e-01 2.76264280e-01 5.77223778e-01 3.65477890e-01
-8.13799739e-01 5.19597352e-01 1.36188686e-01 -9.28999484e-02
1.11431539e+00 -1.53534126e+00 2.87270304e-02 1.23108494e+00
1.72978818e-01 5.09093821e-01 1.13571429e+00 -1.03882897e+00
-7.60140836e-01 -1.31642711e+00 8.83313775e-01 -5.10439754e-01
3.89685541e-01 -7.47803450e-02 -1.47063303e+00 5.37341416e-01
-2.51097560e-01 -1.20271973e-01 6.61089122e-01 1.32793352e-01
-1.42543867e-01 -3.50242466e-01 -1.40074527e+00 4.17940259e-01
7.90461898e-01 -1.07540376e-01 -3.02962542e-01 4.61093843e-01
3.50318432e-01 -2.63053536e-01 -8.25144768e-01 4.31313425e-01
1.15692571e-01 -1.08031297e+00 4.53429550e-01 -3.30935091e-01
7.38832206e-02 -5.62978745e-01 3.05194850e-03 -1.37173617e+00
-8.46553594e-02 -4.73485619e-01 -1.10697269e-01 1.56185210e+00
4.22664851e-01 -4.30608660e-01 8.07859778e-01 7.82519042e-01
-2.86516637e-01 -5.73991597e-01 -7.64960051e-01 -7.05076158e-01
-3.29880089e-01 -6.58872604e-01 4.56769079e-01 1.16243124e+00
-3.52132678e-01 -1.69605354e-03 -3.56126457e-01 3.31026912e-01
7.69550502e-01 -3.39142174e-01 9.36191261e-01 -1.61361647e+00
1.05470695e-01 -3.39437127e-01 -2.08102286e-01 -2.95448214e-01
-9.00213569e-02 -7.37588286e-01 4.83999014e-01 -1.56260300e+00
-9.67417005e-03 -5.77669919e-01 -3.96146446e-01 7.90969610e-01
-4.97502029e-01 2.77657121e-01 -1.14860078e-02 3.20754111e-01
-6.71426952e-01 1.27172500e-01 7.07826734e-01 3.82393226e-02
-5.85133135e-01 2.01027334e-01 -6.36563480e-01 9.27350342e-01
8.68167400e-01 -1.05460846e+00 -2.47070298e-01 2.08170852e-03
3.74563903e-01 -5.78055799e-01 2.68803239e-01 -1.23544121e+00
1.44961432e-01 -1.34799123e-01 3.94492447e-01 -4.71070349e-01
-1.11173980e-01 -9.53235984e-01 1.84613094e-01 4.18486297e-01
-5.18416166e-01 2.27332443e-01 2.54320484e-02 6.52741790e-01
-1.04355723e-01 -7.00941801e-01 1.00964296e+00 -2.13017240e-01
-5.99153280e-01 3.75472531e-02 -4.83653218e-01 1.45381302e-01
9.88177001e-01 -3.00209701e-01 1.54631464e-02 -1.58196524e-01
-1.09134412e+00 3.30857277e-01 8.78968716e-01 2.07403392e-01
3.94247741e-01 -9.83045042e-01 -6.32424116e-01 1.63092449e-01
2.51614451e-01 2.81104296e-01 7.05537871e-02 6.60924852e-01
-2.75462389e-01 -3.91000211e-01 3.72654527e-01 -6.05504155e-01
-1.52529049e+00 3.79229039e-01 2.38433957e-01 -4.06417727e-01
-6.81961358e-01 6.87449634e-01 -6.39292061e-01 -3.79987031e-01
6.52214408e-01 -4.97006416e-01 -1.02685466e-01 4.74012911e-01
2.76880771e-01 5.70585966e-01 5.47024608e-01 -5.22417426e-01
-7.75123388e-02 1.84179664e-01 -2.15132862e-01 1.12340607e-01
1.40229416e+00 -1.80057153e-01 -3.50516558e-01 9.51337934e-01
6.91405058e-01 2.09315524e-01 -1.40234208e+00 -1.70788080e-01
4.76576239e-01 -6.35765851e-01 2.02243384e-02 -8.80736351e-01
-9.99282300e-01 3.44737560e-01 6.34713054e-01 4.26864266e-01
1.15236926e+00 -1.47978932e-01 5.18652081e-01 3.51098895e-01
5.38145602e-02 -1.67236519e+00 2.09244519e-01 2.90691704e-01
5.77855885e-01 -1.29673982e+00 3.46749634e-01 -1.70538217e-01
-6.49869740e-01 7.25760043e-01 7.67660618e-01 1.07263532e-02
6.27958894e-01 1.56646982e-01 4.29964364e-01 -3.40574443e-01
-6.72553062e-01 -1.56305998e-03 -4.67512868e-02 3.36220264e-01
2.15444043e-01 -2.59865761e-01 -2.38768384e-01 2.09436461e-01
-1.20101325e-01 -1.23677133e-02 7.54370987e-01 1.20488393e+00
-6.81399524e-01 -1.01030779e+00 -6.75345063e-01 9.87014592e-01
-5.09751320e-01 2.46776477e-01 -4.75198835e-01 4.85049993e-01
4.56215173e-01 1.13467741e+00 -7.93165565e-02 -2.02344447e-01
6.28913224e-01 6.97429657e-01 3.91932279e-02 -7.68889964e-01
-8.49004686e-01 1.46676570e-01 2.47174930e-02 -5.30319870e-01
-3.56340617e-01 -1.06613326e+00 -1.46301007e+00 -8.50539189e-03
-5.77952504e-01 1.00122891e-01 7.92469800e-01 9.04224634e-01
3.09574425e-01 5.52325845e-01 5.48534632e-01 -6.51076138e-01
-6.37452602e-01 -9.99007165e-01 -4.11000580e-01 8.06641579e-01
4.13655311e-01 -5.57373822e-01 -7.45803893e-01 2.88036704e-01] | [9.403743743896484, 4.046433448791504] |
71e42050-6b2c-4a4e-8c09-4e4c580d9ae3 | diffuseir-diffusion-models-for-isotropic | 2306.12109 | null | https://arxiv.org/abs/2306.12109v1 | https://arxiv.org/pdf/2306.12109v1.pdf | DiffuseIR:Diffusion Models For Isotropic Reconstruction of 3D Microscopic Images | Three-dimensional microscopy is often limited by anisotropic spatial resolution, resulting in lower axial resolution than lateral resolution. Current State-of-The-Art (SoTA) isotropic reconstruction methods utilizing deep neural networks can achieve impressive super-resolution performance in fixed imaging settings. However, their generality in practical use is limited by degraded performance caused by artifacts and blurring when facing unseen anisotropic factors. To address these issues, we propose DiffuseIR, an unsupervised method for isotropic reconstruction based on diffusion models. First, we pre-train a diffusion model to learn the structural distribution of biological tissue from lateral microscopic images, resulting in generating naturally high-resolution images. Then we use low-axial-resolution microscopy images to condition the generation process of the diffusion model and generate high-axial-resolution reconstruction results. Since the diffusion model learns the universal structural distribution of biological tissues, which is independent of the axial resolution, DiffuseIR can reconstruct authentic images with unseen low-axial resolutions into a high-axial resolution without requiring re-training. The proposed DiffuseIR achieves SoTA performance in experiments on EM data and can even compete with supervised methods. | ['Dawei Li', 'Shanghang Zhang', 'Aimin Wang', 'Jiaming Liu', 'Fangxu Zhou', 'Yulu Gan', 'Mingjie Pan'] | 2023-06-21 | null | null | null | null | ['super-resolution'] | ['computer-vision'] | [ 3.99364650e-01 -7.35265063e-03 3.74890655e-01 -2.77462870e-01
-7.30792522e-01 -4.00771558e-01 4.69188005e-01 -5.26241720e-01
-7.33818412e-01 9.49922800e-01 2.51868457e-01 6.28354326e-02
-3.29254627e-01 -6.65407896e-01 -6.03768766e-01 -1.40700984e+00
3.02775919e-01 6.84408426e-01 4.64111388e-01 3.10061932e-01
1.36284575e-01 6.62039697e-01 -1.03424942e+00 2.56323248e-01
6.29465342e-01 5.88096440e-01 6.40305042e-01 9.08985615e-01
-1.14652678e-01 7.02721596e-01 -3.01263690e-01 1.68289050e-01
1.69657707e-01 -4.64562178e-01 -9.21186268e-01 2.16816366e-02
4.14856911e-01 -8.13278675e-01 -3.50149006e-01 1.08385193e+00
8.47370565e-01 -1.62270367e-01 8.75459790e-01 -1.50439769e-01
-1.14158297e+00 3.80248189e-01 -7.42182732e-01 7.24940240e-01
-1.30530775e-01 1.47493988e-01 3.68241847e-01 -8.43742609e-01
1.18349290e+00 1.06439412e+00 4.19136763e-01 1.10234654e+00
-1.81683743e+00 -3.51181328e-01 -1.40129969e-01 -9.19431001e-02
-1.00115657e+00 -5.89023352e-01 6.97778404e-01 -4.69208658e-01
6.32441044e-01 -1.14351906e-01 2.58366853e-01 1.25395656e+00
4.56008971e-01 4.13103640e-01 1.53723168e+00 -9.79173742e-03
2.22094223e-01 -3.51202816e-01 -1.79459840e-01 4.82500076e-01
1.70002759e-01 6.11577965e-02 -3.55094820e-01 -6.37756065e-02
1.73482156e+00 -7.52308890e-02 -6.74215555e-01 -2.22288951e-01
-1.63977671e+00 3.22283357e-01 4.64864582e-01 6.40713036e-01
-5.53843558e-01 -1.76877663e-01 8.93402100e-02 6.19867258e-02
5.22949159e-01 6.06554747e-01 -1.94763318e-01 1.59280956e-01
-9.17891264e-01 -1.15843855e-01 1.99198812e-01 5.01419663e-01
4.68654156e-01 2.65118219e-02 2.87206098e-02 8.48664463e-01
1.26386657e-02 3.39515179e-01 7.10530281e-01 -1.52495790e+00
-2.02706426e-01 1.24345206e-01 2.11643189e-01 -5.98917305e-01
-5.47035098e-01 -4.21447754e-01 -1.50872207e+00 2.98718750e-01
6.93127334e-01 3.76059636e-02 -9.71306145e-01 1.70420372e+00
4.90875334e-01 -2.20815931e-03 6.80398047e-02 1.34704804e+00
6.35539174e-01 5.21043122e-01 -2.77570695e-01 -5.29806376e-01
9.00685191e-01 -7.10850894e-01 -8.77384424e-01 4.80308011e-02
3.09627712e-01 -6.30361974e-01 9.12632704e-01 3.55095744e-01
-1.24630189e+00 -2.60564566e-01 -9.28756118e-01 -2.53498673e-01
1.18835248e-01 -1.08028330e-01 3.19726557e-01 -2.61060819e-02
-1.27418280e+00 7.47987330e-01 -1.21187854e+00 -2.35760927e-01
4.81478304e-01 3.11083794e-01 -6.80447102e-01 -1.99605092e-01
-7.33416975e-01 5.28327584e-01 1.10212266e-01 -2.88216006e-02
-1.04345703e+00 -1.08175945e+00 -4.80710626e-01 -3.14070016e-01
-1.54174432e-01 -1.03079939e+00 8.33455265e-01 -3.47203523e-01
-1.79839456e+00 9.74879861e-01 -3.65045965e-01 -2.42213577e-01
7.40176022e-01 1.12797171e-01 5.71496598e-03 7.15032339e-01
1.43788874e-01 6.67142510e-01 6.39161229e-01 -1.51842308e+00
-1.30992204e-01 -6.50657237e-01 -3.82271767e-01 -9.68508422e-03
-1.59937516e-01 -1.36170983e-01 -1.33187100e-01 -2.96393722e-01
5.55101156e-01 -6.11665189e-01 -3.35871905e-01 2.99782425e-01
-3.76783013e-01 4.23154175e-01 8.85820746e-01 -5.73605597e-01
6.12376213e-01 -1.90433824e+00 4.23555344e-01 -2.97781169e-01
7.27327883e-01 8.76464173e-02 -2.31556520e-01 -1.26469716e-01
2.60233339e-02 -3.08758859e-02 -2.71575063e-01 -3.58359814e-01
-4.97948587e-01 1.89163044e-01 -2.96875685e-02 7.63105214e-01
4.59132046e-02 8.21924627e-01 -9.60682213e-01 -5.98562479e-01
1.88798875e-01 1.01042509e+00 -3.35822821e-01 4.40516353e-01
1.38388515e-01 1.36695874e+00 -3.19990039e-01 3.58303040e-01
9.08015072e-01 -8.22112083e-01 3.23883772e-01 -5.67383885e-01
-1.94946215e-01 -1.00073263e-01 -6.53544009e-01 1.59912491e+00
-3.77788514e-01 5.12341619e-01 5.40083826e-01 -8.05669010e-01
6.33498251e-01 2.28192672e-01 7.09529400e-01 -7.46596634e-01
-6.98033571e-02 3.96148056e-01 1.01268804e-02 -4.56716537e-01
-3.14781256e-02 -5.56874454e-01 6.58860922e-01 6.98791325e-01
1.27335489e-01 -6.59282506e-02 2.12294124e-02 9.96087566e-02
1.10869122e+00 1.41538948e-01 -2.13886335e-01 -5.90026557e-01
4.15207863e-01 -4.81620848e-01 5.50632834e-01 5.99583149e-01
-1.96063459e-01 1.03441763e+00 3.14538330e-01 -7.45844483e-01
-1.81752801e+00 -1.26727855e+00 -6.44982040e-01 5.87623119e-01
2.12050557e-01 2.14256585e-01 -9.21723008e-01 -3.54224861e-01
-5.61429918e-01 -4.17588465e-02 -8.27703178e-01 1.19431108e-01
-6.50451779e-01 -1.28400862e+00 3.61937135e-01 3.16999704e-01
7.82941937e-01 -8.89289081e-01 -4.17429715e-01 2.13280603e-01
-3.25812250e-01 -1.48595870e+00 -4.06373858e-01 1.30373582e-01
-9.43506122e-01 -8.15253794e-01 -1.36074221e+00 -7.01363087e-01
1.04736781e+00 3.07742655e-01 1.01490319e+00 -1.07038610e-01
-4.26393628e-01 8.49360228e-03 2.48357043e-01 3.36382598e-01
-5.14616013e-01 -1.00892715e-01 3.31913143e-01 -1.06640734e-01
7.76031688e-02 -1.09511054e+00 -1.18812919e+00 5.08312404e-01
-1.22384453e+00 2.20033348e-01 6.75913692e-01 1.07301319e+00
1.22857106e+00 2.17777491e-01 3.99427682e-01 -9.96254146e-01
4.14981455e-01 -6.42919987e-02 -5.96917927e-01 -1.28565639e-01
-4.52326626e-01 3.37860912e-01 8.62518489e-01 -7.26634979e-01
-1.32069993e+00 -2.13969514e-01 -2.02065423e-01 -4.67593044e-01
-3.79315138e-01 1.50392488e-01 1.03262395e-01 -2.53990829e-01
8.95925760e-01 6.97037458e-01 1.83038309e-01 -4.22582477e-01
-1.63945004e-01 3.51963609e-01 6.70045972e-01 -6.93448722e-01
6.51874542e-01 1.22006941e+00 2.09459126e-01 -1.02534223e+00
-1.17015421e+00 -2.28598043e-01 -9.14488673e-01 1.60319172e-02
1.12978292e+00 -7.36640096e-01 -6.21730268e-01 8.00356448e-01
-1.00546193e+00 -6.82968676e-01 -1.04312703e-01 6.07177436e-01
-6.73163235e-01 5.62336564e-01 -1.34787345e+00 -5.08076549e-01
-2.96219409e-01 -1.31707120e+00 1.23053586e+00 1.99141979e-01
9.03973207e-02 -1.06818581e+00 9.64074135e-02 4.53278601e-01
7.70528674e-01 1.45611718e-01 1.05118871e+00 2.02559724e-01
-8.01684499e-01 3.39194357e-01 -5.62451065e-01 2.03134239e-01
3.06620717e-01 -1.91816509e-01 -1.06352973e+00 -3.67916733e-01
3.92751813e-01 -4.29766506e-01 8.10483754e-01 8.74760985e-01
1.32121968e+00 -2.46913597e-01 -1.37162283e-01 1.16122854e+00
1.28541100e+00 -2.16012076e-01 6.96189165e-01 3.58935803e-01
6.66914642e-01 5.58613360e-01 1.58771779e-02 1.86061561e-01
9.74997506e-03 5.17408073e-01 2.56692052e-01 -3.10174286e-01
-4.38801378e-01 1.30673684e-02 5.03103994e-03 8.58665705e-01
-3.56471688e-01 1.92190781e-01 -6.35634959e-01 5.34955025e-01
-1.46548808e+00 -1.01206493e+00 -1.89177264e-02 2.01014948e+00
1.23661470e+00 -3.20616364e-02 -3.37616242e-02 -2.35412151e-01
5.68549335e-01 -2.31806803e-02 -9.88506675e-01 3.26136172e-01
-4.52174664e-01 -2.04836339e-01 4.12684560e-01 6.79252505e-01
-8.42348099e-01 7.52782941e-01 6.87307501e+00 6.33513927e-01
-1.44711173e+00 1.49135202e-01 9.56357121e-01 -2.84752309e-01
-4.55856740e-01 -5.88190854e-01 -7.89002120e-01 4.44349974e-01
6.94655359e-01 6.46915585e-02 4.71340060e-01 3.61992240e-01
3.62823486e-01 1.61828220e-01 -9.24422562e-01 9.73264813e-01
-2.75354147e-01 -1.58273518e+00 3.07701141e-01 4.77804929e-01
9.67248023e-01 2.38882691e-01 3.61947238e-01 -4.98902082e-01
4.08442348e-01 -1.12576985e+00 5.20655513e-02 6.39792264e-01
1.05972254e+00 -5.52119732e-01 8.03345919e-01 5.09276927e-01
-6.06094718e-01 3.00226241e-01 -1.06780994e+00 5.26761949e-01
2.38055781e-01 1.13112307e+00 -4.14778084e-01 1.38372257e-01
8.96419823e-01 6.02112114e-01 1.81088839e-02 5.60989201e-01
1.04425214e-01 1.85355097e-01 -2.92000234e-01 4.82443601e-01
6.05949014e-02 -4.21552658e-01 5.52168965e-01 1.09783340e+00
4.42970842e-01 1.58704266e-01 -2.83102095e-01 1.34529185e+00
-1.36611536e-01 -2.98489958e-01 -5.70817173e-01 -5.11905625e-02
1.69396415e-01 1.46840775e+00 -7.04926193e-01 -1.03326201e-01
-1.77597944e-02 1.14942920e+00 5.20905018e-01 7.18589127e-01
-4.51919466e-01 1.34052575e-01 4.69619215e-01 5.37148178e-01
1.21039592e-01 -4.00508940e-01 1.93776116e-02 -1.36130583e+00
-8.58118683e-02 -5.67861795e-01 -4.52855639e-02 -9.18030977e-01
-1.56816363e+00 8.28914165e-01 -4.05124426e-01 -8.28224361e-01
5.73497303e-02 -7.77558863e-01 -3.12776655e-01 1.01834667e+00
-1.68593729e+00 -9.63892043e-01 -4.31055188e-01 4.62694287e-01
3.08206558e-01 1.55769005e-01 8.05481434e-01 3.03792387e-01
-3.69915456e-01 9.13775787e-02 4.19573039e-01 1.09823845e-01
9.02834356e-01 -1.46877837e+00 2.38294434e-02 5.59744537e-01
-4.24109966e-01 8.92532706e-01 7.88178444e-01 -5.26242375e-01
-1.10056508e+00 -9.69169974e-01 4.28377777e-01 -4.38331157e-01
8.11138451e-01 -1.71589419e-01 -1.21174979e+00 4.05059487e-01
4.90822643e-02 5.40047467e-01 8.05664062e-01 -4.06379670e-01
-2.41703123e-01 2.17628896e-01 -1.45092475e+00 6.81152999e-01
1.05432355e+00 -6.58711851e-01 -1.37656569e-01 2.86144346e-01
5.55682778e-01 -3.78649861e-01 -1.25040746e+00 3.04027945e-01
5.74756026e-01 -1.20608509e+00 9.78167415e-01 -4.14154559e-01
9.12090302e-01 -3.65938306e-01 1.05084702e-01 -1.28003275e+00
-6.99021637e-01 -5.81337929e-01 -1.26856387e-01 7.18960106e-01
3.16722542e-01 -5.25337696e-01 9.06408608e-01 3.76514584e-01
5.37307262e-02 -6.20336592e-01 -9.58698511e-01 -6.02984428e-01
4.89412755e-01 3.86660516e-01 2.68904895e-01 9.59157646e-01
-4.28209603e-01 4.76956278e-01 -2.72519350e-01 2.03526065e-01
1.31757820e+00 1.71555117e-01 3.56892526e-01 -1.03561532e+00
-3.85156810e-01 -3.20364565e-01 -2.05689877e-01 -1.47953546e+00
1.79082498e-01 -5.48592567e-01 5.10950983e-02 -1.38797033e+00
7.39023864e-01 -1.80072919e-01 -1.90579802e-01 -1.95015103e-01
-1.73579995e-02 6.15386426e-01 -5.24329126e-01 6.76949561e-01
-5.62580228e-01 5.26298761e-01 2.16911483e+00 -1.71047762e-01
1.11041889e-02 -4.29037869e-01 -5.38844883e-01 7.03817368e-01
5.85687160e-01 -2.32957393e-01 -4.53317493e-01 -8.26226592e-01
4.10542898e-02 1.00856490e-01 2.53178298e-01 -8.77341509e-01
2.92294681e-01 -1.58820942e-01 8.87092888e-01 -2.00299665e-01
3.31264943e-01 -6.62911236e-01 1.11131817e-01 1.32858053e-01
-5.01286864e-01 -2.82584846e-01 -1.61806285e-01 5.30603468e-01
-1.57791078e-02 1.12303495e-01 1.40798807e+00 -4.27389503e-01
1.04558626e-02 6.57712281e-01 -5.01767874e-01 -1.20845728e-01
5.56667089e-01 -2.75305033e-01 -7.25351989e-01 -2.85167813e-01
-1.03311729e+00 -2.71839768e-01 9.29844379e-01 -2.48014987e-01
7.44193912e-01 -1.25199664e+00 -6.69251978e-01 2.28926107e-01
-2.29224652e-01 5.39105713e-01 7.37202168e-01 1.14251089e+00
-7.37115383e-01 2.80410677e-01 -4.57818061e-01 -8.05042684e-01
-8.65794420e-01 6.25193059e-01 7.57980287e-01 -4.87664700e-01
-9.96029496e-01 8.31879795e-01 9.42137241e-01 -5.89724958e-01
-2.92732835e-01 6.66723549e-02 -6.12341836e-02 -8.12519848e-01
9.60729659e-01 3.44003849e-02 -2.05259860e-01 -5.84302664e-01
-7.38151297e-02 9.77807701e-01 -4.00322497e-01 -4.24354151e-02
1.60429227e+00 -4.90418494e-01 -2.81801969e-01 3.95208150e-01
1.22665262e+00 -1.00309595e-01 -1.81198025e+00 -3.40882361e-01
-4.08778459e-01 -4.26521152e-01 4.14287210e-01 -6.14890277e-01
-1.27858019e+00 1.08120203e+00 5.76925993e-01 3.02413404e-02
8.92034709e-01 9.31671932e-02 9.24470484e-01 3.37558508e-01
3.38674128e-01 -8.05016875e-01 3.02013636e-01 3.67083311e-01
6.01310849e-01 -1.28808093e+00 -1.41101584e-01 -2.73696989e-01
-1.79298669e-01 1.16597068e+00 6.45829141e-01 -7.76861608e-02
5.37814319e-01 6.93456769e-01 3.50208759e-01 -3.17550033e-01
-8.16259325e-01 3.64111423e-01 -3.84519175e-02 9.46999848e-01
5.93657613e-01 -4.08570915e-01 1.28474444e-01 2.09185645e-01
1.34902060e-01 8.63421783e-02 6.91841424e-01 5.14879107e-01
-3.81887436e-01 -7.27447927e-01 -9.79191810e-02 3.15732986e-01
-8.67439866e-01 1.17333628e-01 -2.22176552e-01 2.77361810e-01
-3.08804154e-01 3.47535521e-01 1.93672746e-01 4.79185618e-02
-1.08729884e-01 -4.44139093e-01 8.52408409e-01 -3.05321068e-01
2.49107227e-01 5.52722812e-01 -4.60919678e-01 -5.46874762e-01
-6.24722362e-01 -3.18663925e-01 -1.31179011e+00 -5.58378458e-01
6.90990388e-02 1.47004515e-01 3.73719931e-01 8.46869111e-01
4.75600243e-01 5.81558883e-01 2.45961577e-01 -1.13079476e+00
-3.37855965e-01 -9.32347775e-01 -9.33394253e-01 4.30648446e-01
7.54870415e-01 -3.87580693e-01 -6.66355729e-01 2.53810555e-01] | [12.870826721191406, -2.7210185527801514] |
66463241-f60f-4d38-a0e3-f0362b074f59 | speaker-recognition-in-realistic-scenario | 2302.13033 | null | https://arxiv.org/abs/2302.13033v1 | https://arxiv.org/pdf/2302.13033v1.pdf | Speaker Recognition in Realistic Scenario Using Multimodal Data | In recent years, an association is established between faces and voices of celebrities leveraging large scale audio-visual information from YouTube. The availability of large scale audio-visual datasets is instrumental in developing speaker recognition methods based on standard Convolutional Neural Networks. Thus, the aim of this paper is to leverage large scale audio-visual information to improve speaker recognition task. To achieve this task, we proposed a two-branch network to learn joint representations of faces and voices in a multimodal system. Afterwards, features are extracted from the two-branch network to train a classifier for speaker recognition. We evaluated our proposed framework on a large scale audio-visual dataset named VoxCeleb$1$. Our results show that addition of facial information improved the performance of speaker recognition. Moreover, our results indicate that there is an overlap between face and voice. | ['Muhammad Haroon Yousaf', 'Shah Nawaz', 'Muhammad Saad Saeed', 'Saqlain Hussain Shah'] | 2023-02-25 | null | null | null | null | ['speaker-recognition'] | ['speech'] | [-2.19474569e-01 -2.29809731e-01 6.94793314e-02 -5.54648340e-01
-8.55373144e-01 -4.84344125e-01 5.48557162e-01 -2.51238137e-01
-2.96477884e-01 4.37846661e-01 3.25981617e-01 2.24668160e-01
2.04101905e-01 -5.33169866e-01 -6.83109462e-01 -5.83451569e-01
-2.49693487e-02 -1.52836055e-01 -1.33169264e-01 -8.45073955e-04
1.64327770e-01 7.68033564e-01 -1.78119433e+00 7.78556406e-01
1.84630156e-01 1.35823619e+00 -1.64011762e-01 6.48438990e-01
-2.51506418e-01 5.86981654e-01 -4.10997391e-01 -4.89259034e-01
1.24996252e-01 -7.68302158e-02 -6.19883716e-01 1.21469215e-01
7.59363770e-01 -3.82731885e-01 -2.77228624e-01 7.74569988e-01
7.37842917e-01 -3.99656631e-02 4.92804050e-01 -1.55033922e+00
-4.14096802e-01 6.36613548e-01 -4.42350954e-01 2.60372579e-01
4.23703432e-01 1.63856730e-01 1.14279032e+00 -1.17805612e+00
4.90624160e-01 1.46750474e+00 6.35834217e-01 4.94217277e-01
-1.00388324e+00 -1.00117922e+00 -4.69588786e-02 3.73076409e-01
-1.72323227e+00 -1.09749675e+00 1.10876298e+00 -3.29310298e-01
8.69953215e-01 2.02914521e-01 6.65891230e-01 1.20389593e+00
-2.94008195e-01 8.34267855e-01 8.85812700e-01 -3.04265499e-01
-2.23643370e-02 4.15003031e-01 -1.87073126e-01 6.35794163e-01
-4.01967257e-01 9.92876962e-02 -1.06996560e+00 -1.38211578e-01
4.92835671e-01 -1.18192777e-01 -3.02536357e-02 -7.20634460e-02
-7.86251783e-01 8.24736893e-01 4.87208724e-01 4.47831005e-01
-2.90828824e-01 1.82142541e-01 4.63074625e-01 2.52502382e-01
5.02327740e-01 -9.20748711e-02 -1.28031477e-01 -1.96120769e-01
-1.08327591e+00 -1.31733239e-01 7.76621282e-01 3.55594665e-01
6.86360657e-01 2.03143016e-01 -6.65408149e-02 1.05778491e+00
7.64720500e-01 4.67256457e-01 4.97562468e-01 -6.75793529e-01
4.27063882e-01 7.05349863e-01 -2.74669349e-01 -9.86253381e-01
-3.33646685e-01 -2.84675062e-01 -5.53980291e-01 1.63106009e-01
3.42941135e-01 7.69758597e-02 -6.08097613e-01 1.51190805e+00
4.63877022e-01 4.31234688e-01 -4.96703312e-02 9.94500995e-01
1.28175175e+00 4.72386390e-01 9.28984359e-02 1.41172245e-01
1.39996004e+00 -6.25371635e-01 -5.12450993e-01 1.77670360e-01
2.28274301e-01 -8.01724613e-01 9.15620148e-01 2.54571915e-01
-8.24150681e-01 -7.59843469e-01 -8.66057515e-01 1.37350187e-01
-3.98793906e-01 5.96899927e-01 3.82205933e-01 1.04632890e+00
-1.16561961e+00 1.60695627e-01 -4.94226545e-01 -3.77717823e-01
8.61492455e-01 4.70878541e-01 -7.22886324e-01 -6.09605201e-02
-1.03325796e+00 3.38152647e-01 7.78585151e-02 3.33398849e-01
-1.39820778e+00 -4.27016884e-01 -8.55210721e-01 1.76505998e-01
1.58503186e-02 -2.50960886e-01 9.97097850e-01 -1.31707561e+00
-1.55193949e+00 8.63397896e-01 -2.77959466e-01 -2.94954389e-01
3.68163735e-01 -2.40720771e-02 -6.55797303e-01 3.46691102e-01
-1.86168537e-01 8.52829278e-01 1.28428650e+00 -1.12991512e+00
-5.59605479e-01 -5.31173766e-01 -6.99882731e-02 -1.93235695e-01
-7.27566540e-01 3.04355174e-01 -3.62539291e-01 -3.65620315e-01
-2.87432432e-01 -6.51483059e-01 5.35522580e-01 7.79694319e-02
-3.80730927e-01 -4.95875061e-01 1.02451658e+00 -6.93968832e-01
8.65884364e-01 -2.53231478e+00 -2.20583588e-01 2.57687986e-01
2.73827851e-01 3.63223225e-01 -5.45720339e-01 3.68979603e-01
-1.64947942e-01 1.26294017e-01 4.13168818e-01 -4.92069840e-01
-7.68730929e-03 -1.73000678e-01 -3.25375527e-01 4.98388082e-01
5.08811831e-01 8.70837331e-01 -3.43241751e-01 -5.41226923e-01
3.47636081e-02 9.86348510e-01 -5.71634114e-01 2.68860817e-01
1.82976916e-01 2.12459311e-01 -1.76150620e-01 1.00686407e+00
6.64333820e-01 4.56849597e-02 -5.40604442e-02 -4.12880659e-01
-5.83330579e-02 3.88764888e-02 -9.97105360e-01 1.50354135e+00
-3.30118984e-01 8.63877535e-01 5.03190219e-01 -9.98797178e-01
1.04036641e+00 6.49533808e-01 4.50789690e-01 -6.33939683e-01
3.84063989e-01 6.80584833e-02 -8.50276574e-02 -5.59268832e-01
1.27559096e-01 -2.93029577e-01 2.71932036e-01 1.79652691e-01
5.20659924e-01 2.65877932e-01 -1.06069230e-01 2.04128936e-01
7.69244671e-01 -2.02881381e-01 -1.12998486e-01 1.94372952e-01
8.11376810e-01 -5.31968176e-01 2.68647373e-01 3.27476859e-01
-4.58684266e-01 4.49249536e-01 3.82613331e-01 -3.73998076e-01
-6.74848020e-01 -8.99358153e-01 -1.61001578e-01 1.36843812e+00
-5.11983395e-01 -5.07043719e-01 -6.30563915e-01 -6.79315329e-01
1.62259802e-01 8.86153802e-02 -7.08035529e-01 4.17637005e-02
-2.01084256e-01 -2.35637054e-01 9.58803236e-01 5.41222095e-01
4.95683670e-01 -1.04670143e+00 -2.05457896e-01 -4.53758948e-02
-4.13908251e-02 -1.24313378e+00 -3.77946049e-01 -3.41775000e-01
-5.12720942e-01 -1.17944121e+00 -8.23131323e-01 -6.86240196e-01
3.34228367e-01 4.69025336e-02 6.93598866e-01 -1.88668996e-01
-5.80045164e-01 7.97982872e-01 -2.03542292e-01 -6.29916012e-01
-1.95788503e-01 -1.06914625e-01 2.22927704e-01 7.24623561e-01
4.29830372e-01 -4.61246192e-01 -3.93445164e-01 5.09158432e-01
-7.76554346e-01 -4.66356099e-01 5.62652409e-01 5.80470324e-01
1.25268087e-01 -2.42339522e-01 9.11097765e-01 -6.49382547e-02
5.78442097e-01 -3.34045410e-01 -5.36978304e-01 1.70354709e-01
-3.36445495e-02 -2.83615261e-01 2.34270170e-01 -5.78969479e-01
-8.45408618e-01 3.34921837e-01 -3.40296596e-01 -6.50783241e-01
-4.58267927e-01 6.03919685e-01 -4.37945306e-01 -1.44072294e-01
3.14822942e-01 1.17451482e-01 2.77260900e-01 -5.15998483e-01
2.93815464e-01 1.20192981e+00 2.85602152e-01 -2.55538672e-01
5.18235028e-01 5.20822763e-01 -2.62897074e-01 -1.27627993e+00
-4.27882940e-01 -5.08209229e-01 -6.37297571e-01 -6.15637720e-01
8.94417465e-01 -1.21645176e+00 -1.13863599e+00 4.72834706e-01
-1.08387589e+00 1.24815971e-01 -5.90256229e-02 5.82436442e-01
-1.67955458e-01 2.23841652e-01 -4.00807947e-01 -1.15343916e+00
-3.38786125e-01 -1.11276412e+00 1.16345549e+00 1.65742338e-01
1.19191175e-02 -6.49239182e-01 2.56351143e-01 7.08639681e-01
5.68480670e-01 -1.89072952e-01 3.08677286e-01 -9.36050951e-01
-4.44925606e-01 -4.26087558e-01 -4.99124974e-01 4.60162133e-01
1.53251588e-01 2.80986100e-01 -1.62464666e+00 -3.05682272e-01
-4.21027064e-01 -6.72682285e-01 9.90706027e-01 1.89038649e-01
9.64165211e-01 -2.38682836e-01 -1.64662555e-01 4.02450204e-01
8.74099255e-01 -6.03121594e-02 4.21484232e-01 -1.10826552e-01
6.89048290e-01 6.74169302e-01 1.06484935e-01 6.35225177e-01
3.30258518e-01 7.49549389e-01 5.50239325e-01 1.06958851e-01
-2.73380876e-01 -2.03957140e-01 7.80993044e-01 5.94624460e-01
-1.14059843e-01 1.34565368e-01 -9.18284774e-01 5.93857825e-01
-1.45964098e+00 -1.07534397e+00 3.97576630e-01 1.84122336e+00
5.94876647e-01 -2.71477968e-01 5.79307675e-01 1.15969621e-01
7.21473753e-01 1.02308288e-01 -2.87921220e-01 -1.66020453e-01
1.22207580e-02 1.61510363e-01 -2.58529246e-01 2.29931757e-01
-1.23836231e+00 8.13166082e-01 6.39545155e+00 7.08658397e-01
-1.69542646e+00 -7.43761845e-03 4.63211060e-01 -5.75871408e-01
1.91975627e-02 -5.95413685e-01 -9.53314900e-01 1.74908876e-01
1.23326647e+00 1.05621025e-01 2.92157471e-01 7.74173737e-01
2.75585592e-01 4.52350497e-01 -1.10123098e+00 1.29039061e+00
4.12818730e-01 -1.28423977e+00 -3.30891907e-02 2.23608524e-01
1.59923553e-01 2.68055439e-01 2.38690928e-01 3.61140847e-01
3.28920074e-02 -1.33916771e+00 5.64367235e-01 3.51090789e-01
8.34087670e-01 -8.15028012e-01 7.41693676e-01 3.56822088e-02
-1.50613678e+00 -2.09886849e-01 -1.88784063e-01 1.88100412e-01
1.44891553e-02 2.90442199e-01 -1.15291953e+00 3.30323488e-01
9.22232628e-01 8.67322862e-01 -6.19551599e-01 1.09950805e+00
6.55541718e-02 6.48493886e-01 -2.81164318e-01 1.42052710e-01
-4.80317064e-02 2.59016126e-01 4.08923954e-01 1.19506061e+00
3.78769428e-01 -3.12827617e-01 -2.01665983e-02 8.67465496e-01
-4.46181893e-01 3.59830856e-01 -7.53830254e-01 -5.69603443e-01
2.72052288e-01 1.56868732e+00 -4.11070347e-01 -6.08149432e-02
-6.04081094e-01 5.64111173e-01 2.99762845e-01 3.06171745e-01
-8.43526721e-01 -1.63087249e-01 6.21963978e-01 7.74218664e-02
5.15433252e-01 -7.23251253e-02 2.20845640e-01 -1.13658297e+00
-1.12469308e-01 -8.27271760e-01 3.44407856e-01 -6.29266024e-01
-1.37175667e+00 7.50174582e-01 -3.59998852e-01 -1.05755258e+00
-2.04223637e-02 -6.46429479e-01 -6.31663382e-01 7.78524935e-01
-1.53937387e+00 -1.61377656e+00 -4.09033000e-01 1.03981113e+00
4.05294299e-01 -6.66594088e-01 8.48124325e-01 4.45052117e-01
-7.26150870e-01 8.42172325e-01 -2.46374294e-01 6.12709165e-01
8.51369441e-01 -6.65796399e-01 -1.17078915e-01 3.58671635e-01
6.29893064e-01 4.93628055e-01 2.84828424e-01 -3.98079157e-01
-1.41591644e+00 -1.02765357e+00 6.58981502e-01 -1.93436950e-01
5.97655416e-01 -4.53188747e-01 -7.01045036e-01 4.10909623e-01
4.29079443e-01 3.46506417e-01 1.21460676e+00 1.87448949e-01
-7.92336166e-01 -5.26143074e-01 -1.19008279e+00 7.52402171e-02
4.88477707e-01 -1.14694393e+00 -3.86956483e-01 -1.43913895e-01
2.53044039e-01 2.31972083e-01 -7.89051235e-01 1.50174335e-01
9.86172080e-01 -8.38263869e-01 1.15837967e+00 -7.54852772e-01
5.01507163e-01 -5.50998896e-02 -4.97845173e-01 -1.14784813e+00
1.74855068e-01 -1.91770807e-01 -2.02552113e-03 1.61497974e+00
3.25103611e-01 -4.21469003e-01 7.08844900e-01 3.91859800e-01
2.02094033e-01 -2.99504966e-01 -1.35038066e+00 -5.55849612e-01
-1.91514298e-01 -6.08216405e-01 4.15026873e-01 8.94391298e-01
6.13345914e-02 3.73651236e-01 -3.68805170e-01 2.44017273e-01
5.49202383e-01 3.06599941e-02 9.09483612e-01 -1.41236603e+00
-1.20517440e-01 -3.43938917e-01 -6.86269701e-01 -6.40068531e-01
5.71159780e-01 -9.29560900e-01 -2.56593972e-01 -1.09253657e+00
2.43439347e-01 5.39905950e-02 -5.90643644e-01 6.43641591e-01
2.52200752e-01 6.91161990e-01 3.04392964e-01 6.76431367e-03
-6.22328758e-01 7.42041230e-01 8.68391573e-01 -4.61600304e-01
-1.58265233e-01 -3.68753113e-02 -5.39277196e-01 5.11488795e-01
7.21069455e-01 -2.22272545e-01 -2.77850609e-02 -1.25053152e-01
-2.87315045e-02 8.47954005e-02 6.77869201e-01 -8.73084784e-01
1.54400587e-01 2.55732059e-01 5.45196891e-01 -5.20972848e-01
7.98609793e-01 -8.85603428e-01 -2.27311671e-01 1.42860606e-01
-3.53518456e-01 -3.45719069e-01 5.27591288e-01 5.49374342e-01
-6.66041791e-01 2.38313004e-01 6.92318916e-01 1.21407866e-01
-4.92943317e-01 1.81412801e-01 -3.09322417e-01 -4.01962608e-01
9.65530574e-01 -5.97038679e-02 -1.19380772e-01 -5.48241198e-01
-9.18236017e-01 -7.10062087e-02 -6.66659623e-02 6.24192894e-01
8.82918239e-01 -1.54647100e+00 -7.21875966e-01 5.98728538e-01
2.06765711e-01 -5.90201676e-01 3.35495532e-01 9.24318671e-01
-6.99133798e-02 4.96737927e-01 -5.23458779e-01 -8.14337492e-01
-2.02876210e+00 2.88286299e-01 3.26732218e-01 2.85463989e-01
-8.49502236e-02 1.20762777e+00 2.70028263e-02 -4.00858104e-01
5.96468389e-01 -6.60073012e-02 -5.81615746e-01 5.26148081e-01
8.02621067e-01 2.26831019e-01 -4.28560823e-02 -1.05706608e+00
-6.51092291e-01 3.92000049e-01 -1.00658789e-01 -3.46863985e-01
1.62114108e+00 -5.80502190e-02 -6.22837022e-02 4.74403471e-01
1.42447615e+00 1.37006447e-01 -1.08821797e+00 -2.45016873e-01
-2.13934571e-01 -5.63405216e-01 2.04539433e-01 -6.26291335e-01
-1.48358047e+00 1.39057446e+00 9.38663721e-01 1.27939239e-01
1.01342320e+00 1.20015591e-01 4.38078582e-01 2.89121538e-01
1.45654619e-01 -9.07316506e-01 3.26675922e-01 2.34748140e-01
1.05879223e+00 -1.47298169e+00 -3.96829098e-01 -1.58697933e-01
-6.63347960e-01 1.26408422e+00 5.17097950e-01 2.42366299e-01
8.36759508e-01 7.26043805e-02 2.71321088e-01 -2.53626764e-01
-6.37042105e-01 -1.67917818e-01 6.10096693e-01 4.82719272e-01
6.46089256e-01 -1.75420478e-01 4.48953718e-01 7.88622975e-01
-1.20973103e-01 1.71568617e-01 5.35072125e-02 6.98743463e-01
-4.46249068e-01 -1.08243239e+00 -5.56751311e-01 1.67840838e-01
-5.64359426e-01 2.58474071e-02 -7.30282366e-01 5.00431776e-01
-7.99263120e-02 1.18473697e+00 7.48369247e-02 -5.11978626e-01
1.37003958e-01 4.46489245e-01 2.92826861e-01 -3.65527302e-01
-6.46449268e-01 4.16600257e-01 3.28149050e-02 -7.68656909e-01
-5.68687737e-01 -5.14550388e-01 -8.49647641e-01 -1.22514293e-01
-2.12122858e-01 4.27062288e-02 9.55869496e-01 7.75433779e-01
5.33098221e-01 3.98713470e-01 6.65729165e-01 -9.90200400e-01
-4.11798000e-01 -1.03934157e+00 -7.24718869e-01 2.90752500e-01
6.24515891e-01 -6.39942408e-01 -4.10687417e-01 5.13189808e-02] | [14.527854919433594, 4.929131507873535] |
06cf9073-fdc5-4079-a9db-fe91785b5a6c | graph-based-knowledge-distillation-a-survey | 2302.14643 | null | https://arxiv.org/abs/2302.14643v1 | https://arxiv.org/pdf/2302.14643v1.pdf | Graph-based Knowledge Distillation: A survey and experimental evaluation | Graph, such as citation networks, social networks, and transportation networks, are prevalent in the real world. Graph Neural Networks (GNNs) have gained widespread attention for their robust expressiveness and exceptional performance in various graph applications. However, the efficacy of GNNs is heavily reliant on sufficient data labels and complex network models, with the former obtaining hardly and the latter computing costly. To address the labeled data scarcity and high complexity of GNNs, Knowledge Distillation (KD) has been introduced to enhance existing GNNs. This technique involves transferring the soft-label supervision of the large teacher model to the small student model while maintaining prediction performance. This survey offers a comprehensive overview of Graph-based Knowledge Distillation methods, systematically categorizing and summarizing them while discussing their limitations and future directions. This paper first introduces the background of graph and KD. It then provides a comprehensive summary of three types of Graph-based Knowledge Distillation methods, namely Graph-based Knowledge Distillation for deep neural networks (DKD), Graph-based Knowledge Distillation for GNNs (GKD), and Self-Knowledge Distillation based Graph-based Knowledge Distillation (SKD). Each type is further divided into knowledge distillation methods based on the output layer, middle layer, and constructed graph. Subsequently, various algorithms' ideas are analyzed and compared, concluding with the advantages and disadvantages of each algorithm supported by experimental results. In addition, the applications of graph-based knowledge distillation in CV, NLP, RS, and other fields are listed. Finally, the graph-based knowledge distillation is summarized and prospectively discussed. We have also released related resources at https://github.com/liujing1023/Graph-based-Knowledge-Distillation. | ['Qinfen Hao', 'Guanzheng Zhang', 'Tongya Zheng', 'Jing Liu'] | 2023-02-27 | null | null | null | null | ['self-knowledge-distillation'] | ['computer-vision'] | [-2.35950917e-01 6.98908925e-01 -6.32775664e-01 -2.05697790e-01
2.34973133e-01 -5.64192235e-01 3.48916203e-01 1.40736222e-01
-1.77899629e-01 1.02227998e+00 -2.67323673e-01 -6.99098766e-01
-5.97617686e-01 -1.23789096e+00 -5.15483618e-01 -6.43607378e-01
-9.55623239e-02 6.36364579e-01 1.90260515e-01 -1.67269096e-01
-1.91687599e-01 6.02167606e-01 -1.10359693e+00 -4.78934228e-01
1.27753115e+00 7.60644555e-01 -3.32425791e-03 3.97163153e-01
-5.03175557e-01 1.05892158e+00 -5.97826064e-01 -9.11023140e-01
1.80707216e-01 -4.84773636e-01 -9.59571302e-01 -3.55604887e-01
3.54617238e-01 3.15991640e-02 -9.63534594e-01 1.30008888e+00
4.68337715e-01 4.32459205e-01 6.98374093e-01 -1.77166128e+00
-1.23142898e+00 1.06309474e+00 -2.20083445e-01 1.24339841e-01
-4.19792309e-02 -1.21992685e-01 9.98629272e-01 -5.38847625e-01
6.50391281e-01 9.85288918e-01 7.18629241e-01 4.99589384e-01
-8.54065955e-01 -9.39186215e-01 3.51435691e-01 5.69992244e-01
-1.59925842e+00 2.43264258e-01 1.01864648e+00 -4.14702445e-01
1.02520335e+00 -1.64591596e-01 1.15958643e+00 8.67736161e-01
-3.73072684e-01 8.28552186e-01 7.12579310e-01 -5.08373022e-01
-2.05077212e-02 -2.77967211e-02 4.26424116e-01 1.24288321e+00
6.88746810e-01 5.63083738e-02 -3.35194051e-01 9.98249650e-02
8.91902387e-01 1.57749951e-02 -3.64987314e-01 -3.89392078e-01
-7.07742691e-01 8.59440088e-01 8.94413412e-01 4.06115860e-01
-1.79200485e-01 1.11027703e-01 3.24554265e-01 3.52199554e-01
6.48991644e-01 3.23785841e-01 -3.89223844e-01 1.52126744e-01
-8.41996193e-01 -1.47088528e-01 1.06752861e+00 1.26470649e+00
8.62904906e-01 5.31644762e-01 -8.74832347e-02 8.38944614e-01
2.92861819e-01 3.32266629e-01 2.01203063e-01 -5.71633160e-01
3.26179266e-01 1.08336997e+00 -8.00629139e-01 -1.27304292e+00
-5.93816578e-01 -6.00522220e-01 -1.29757154e+00 -1.87644541e-01
6.48558289e-02 -3.35809171e-01 -1.23613524e+00 1.56283557e+00
3.26262295e-01 6.78430855e-01 -1.16254576e-02 4.38370198e-01
1.75678849e+00 3.42081547e-01 1.63860932e-01 -9.21703652e-02
1.01616430e+00 -1.26896977e+00 -7.67046213e-01 7.89022446e-02
9.23903942e-01 -5.86490780e-02 6.98999286e-01 1.87509581e-01
-7.01196671e-01 -4.23712134e-01 -9.13309634e-01 -1.20838769e-01
-1.04469752e+00 -1.09601699e-01 9.95328248e-01 6.36083424e-01
-1.40541327e+00 6.33082688e-01 -4.87480551e-01 -3.24197948e-01
8.15817714e-01 4.68850106e-01 -3.82207811e-01 -1.72866151e-01
-1.70503342e+00 9.07905579e-01 1.01237571e+00 1.69639945e-01
-5.33889592e-01 -6.83880270e-01 -1.04449642e+00 2.12511569e-01
6.40836954e-01 -9.76034880e-01 1.01095855e+00 -6.59259737e-01
-1.53654528e+00 7.11184144e-01 4.67250437e-01 -3.40960026e-01
2.34876052e-01 3.05994093e-01 -6.04593575e-01 2.10535616e-01
-1.93661347e-01 5.21507919e-01 3.02409917e-01 -1.01555860e+00
-2.94786185e-01 -1.23100385e-01 2.36044466e-01 3.78194034e-01
-4.38171148e-01 -5.07665098e-01 -7.83876121e-01 -6.65232658e-01
-1.93299726e-01 -6.82772756e-01 -1.40330926e-01 -2.30814636e-01
-6.95848227e-01 -7.49721706e-01 8.63624275e-01 -4.03452456e-01
1.70707381e+00 -1.74595726e+00 4.66710143e-02 4.53101844e-01
1.01849473e+00 8.44315767e-01 -3.38256538e-01 3.93474936e-01
-2.56641865e-01 2.80276895e-01 -1.40174121e-01 5.06729484e-02
5.87062631e-03 6.72368944e-01 1.13041185e-01 1.64505363e-01
-2.58338489e-02 1.66308427e+00 -1.34235644e+00 -6.58712506e-01
1.93105802e-01 5.14786780e-01 -2.03915045e-01 2.45309696e-02
-1.94507271e-01 2.41844475e-01 -5.26777208e-01 8.04985464e-01
4.59526628e-01 -7.42699802e-01 4.29218292e-01 -3.24289113e-01
4.01494920e-01 5.48780896e-02 -9.97729719e-01 1.30607903e+00
-7.12602809e-02 6.50028050e-01 -1.66090667e-01 -1.40180409e+00
1.10613418e+00 2.58265167e-01 4.33742166e-01 -4.76033360e-01
3.40786129e-01 1.78073391e-01 -9.82835796e-03 -3.59187990e-01
4.22099322e-01 -6.28500879e-02 3.01920086e-01 2.99169481e-01
5.86920440e-01 -1.91978335e-01 4.00031149e-01 6.49010122e-01
1.06713200e+00 -4.80781682e-02 5.07038057e-01 1.63932294e-02
4.10382152e-01 -2.19217986e-01 4.35376316e-01 7.77576566e-01
-3.74267399e-01 1.38379987e-02 5.79613805e-01 -3.15134376e-01
-4.53024924e-01 -9.73934889e-01 1.94922209e-01 9.53599632e-01
1.30015671e-01 -5.03161848e-01 -5.77462912e-01 -1.27491319e+00
2.07922116e-01 5.14302254e-01 -6.23897851e-01 -4.66967553e-01
-2.47775212e-01 -5.76369703e-01 9.11116838e-01 5.81770122e-01
9.12942231e-01 -1.09224105e+00 5.69558263e-01 -1.02470025e-01
-1.59017986e-03 -1.04761696e+00 -1.04676977e-01 9.00324881e-02
-8.74681830e-01 -1.49355960e+00 -6.49110079e-01 -1.09932494e+00
8.44598830e-01 3.09201121e-01 1.32933080e+00 4.20302957e-01
7.03818426e-02 6.32238269e-01 -3.75129044e-01 -4.15791869e-01
-1.48084611e-01 3.75630379e-01 9.06971693e-02 -4.63530958e-01
6.62494659e-01 -7.71219254e-01 -1.82205319e-01 -1.06694654e-01
-8.73437345e-01 9.99901816e-02 5.92445552e-01 8.39154482e-01
5.50497591e-01 4.19476599e-01 7.90366948e-01 -1.29809070e+00
9.76130009e-01 -6.41314924e-01 -5.57887793e-01 5.86106300e-01
-1.18238306e+00 -1.56066418e-01 6.79230928e-01 -4.03345138e-01
-8.14204931e-01 -4.59469944e-01 3.51237915e-02 -7.12968647e-01
9.15265381e-02 1.08498490e+00 -1.02121376e-01 -5.44676304e-01
6.48888111e-01 6.17276914e-02 -3.52228805e-02 -3.54967415e-01
8.06305826e-01 4.29027051e-01 3.95863265e-01 -5.67589283e-01
8.03465307e-01 -1.13662116e-01 1.81761056e-01 -7.72350907e-01
-1.06691468e+00 -2.32120559e-01 -7.06501544e-01 -4.83477920e-01
6.20577931e-01 -7.37005651e-01 -8.15804541e-01 6.82618380e-01
-9.77167726e-01 -6.07745349e-01 -5.14436901e-01 5.78003109e-01
-1.69026971e-01 4.45928037e-01 -7.87090421e-01 -3.15357208e-01
-4.24996018e-01 -8.25054705e-01 1.99465305e-01 5.95012784e-01
4.11824822e-01 -1.78357494e+00 3.28108519e-02 5.58638513e-01
4.36662227e-01 3.73985678e-01 1.08413732e+00 -1.12769556e+00
-3.59537989e-01 -2.27896973e-01 -6.95257366e-01 5.41439235e-01
1.49628684e-01 -5.99237345e-02 -7.82491684e-01 -1.23609722e-01
-8.30353022e-01 -3.70260924e-01 8.01025450e-01 3.79235625e-01
1.23084795e+00 -3.80733639e-01 -5.40721953e-01 6.33226097e-01
1.27206874e+00 6.60318360e-02 3.24529320e-01 6.24512509e-02
1.38633859e+00 2.97822922e-01 9.52072144e-02 -2.58935481e-01
8.22351098e-01 7.89819583e-02 4.12600219e-01 -2.09144503e-01
-5.11385858e-01 -4.33784008e-01 1.14362948e-02 1.58846641e+00
-5.29132783e-01 -6.96490228e-01 -1.14518964e+00 4.53871876e-01
-1.96205473e+00 -5.67710042e-01 -2.62250632e-01 1.65187597e+00
8.36687386e-01 -1.76299252e-02 9.98970270e-02 -8.79243240e-02
1.07909119e+00 1.99047491e-01 -7.76897848e-01 -3.09948146e-01
-2.32433021e-01 3.37357819e-01 5.21840572e-01 6.34258330e-01
-8.24850559e-01 1.35494375e+00 5.88747454e+00 1.10450995e+00
-7.91382849e-01 5.79455122e-02 3.06756228e-01 3.45814914e-01
-2.44449869e-01 -7.83888027e-02 -6.06581986e-01 1.78479597e-01
9.27375138e-01 -4.54793990e-01 4.49122250e-01 9.11293626e-01
-3.60694617e-01 1.78083837e-01 -7.19988585e-01 1.05168629e+00
-6.16221456e-03 -1.50284684e+00 4.63352412e-01 -9.92534980e-02
1.01957536e+00 3.89610201e-01 -3.11244905e-01 9.04417515e-01
8.38424325e-01 -1.05474639e+00 -2.27704886e-02 6.04161978e-01
9.80964363e-01 -6.40903354e-01 8.32023621e-01 1.55523092e-01
-1.39521873e+00 2.24414796e-01 -2.87310570e-01 -1.00000896e-01
-5.70713654e-02 1.05574059e+00 -8.58868599e-01 1.27029848e+00
5.38864076e-01 1.29201448e+00 -5.47892213e-01 9.66323733e-01
-8.56871367e-01 9.12899494e-01 -1.16667539e-01 -1.49574816e-01
3.15435082e-01 -5.70437253e-01 3.95535350e-01 1.19786119e+00
5.86721748e-02 2.55656302e-01 2.66387433e-01 8.65422308e-01
-5.50187945e-01 -2.68773846e-02 -8.46795619e-01 -5.50347626e-01
7.81642497e-01 1.23993456e+00 -6.64340258e-01 -6.31728232e-01
-5.61605096e-01 5.41953921e-01 7.52828419e-01 7.16028512e-01
-6.97082996e-01 -9.07215059e-01 2.30634436e-01 -2.41333023e-01
1.59521550e-02 -2.69494921e-01 1.12383455e-01 -1.19453096e+00
-3.41860622e-01 -4.89610314e-01 7.88137674e-01 -8.68613124e-01
-1.58190489e+00 6.01523280e-01 2.48980731e-01 -8.29820454e-01
4.76486646e-02 -7.48186231e-01 -5.17509699e-01 8.21580112e-01
-1.84793961e+00 -1.28743041e+00 -6.60575271e-01 7.43876338e-01
-2.63780683e-01 -3.40726316e-01 6.44532502e-01 5.06202400e-01
-8.72033894e-01 7.69815445e-01 -6.09089434e-02 6.49862051e-01
3.52816135e-01 -1.32661438e+00 1.65775016e-01 5.98534107e-01
4.99689281e-02 4.62626934e-01 3.16042267e-02 -9.02947426e-01
-9.64798272e-01 -1.37614655e+00 8.89870763e-01 -1.70117795e-01
9.60986257e-01 -1.55363351e-01 -1.07760358e+00 9.91589427e-01
8.72254223e-02 3.12026471e-01 8.38791370e-01 2.92262495e-01
-4.10650671e-01 -6.70001935e-03 -1.10944283e+00 6.24678195e-01
1.43658900e+00 -5.17541766e-01 -4.45326269e-01 5.26490152e-01
9.70080554e-01 -4.82915819e-01 -1.24663424e+00 4.17587996e-01
1.49017662e-01 -6.17271721e-01 8.53383422e-01 -7.62270212e-01
6.19238010e-03 -2.03418374e-01 4.23956305e-01 -1.48720443e+00
-6.84128344e-01 -4.88845855e-01 -8.10033977e-01 1.17760277e+00
3.06450039e-01 -1.09579396e+00 1.06766987e+00 4.52584237e-01
-2.46791378e-01 -9.72077012e-01 -6.15008175e-01 -1.18843186e+00
1.12962954e-01 -2.95416087e-01 7.25190222e-01 1.71212256e+00
1.11549854e-01 6.76968098e-01 1.97346564e-02 -3.01042031e-02
5.04314184e-01 -1.00215219e-01 6.01440489e-01 -1.51374507e+00
-7.43404543e-03 -7.47636199e-01 -7.03673959e-01 -1.02063072e+00
4.52688932e-01 -1.67051828e+00 -6.23202622e-01 -2.19261575e+00
9.78099629e-02 -5.70478439e-01 -4.94940221e-01 1.13790584e+00
-2.16121286e-01 7.62872770e-02 8.59691799e-02 -3.32085490e-02
-8.11807990e-01 6.97604835e-01 1.71481884e+00 -2.60081410e-01
-2.71295249e-01 -1.56235829e-01 -7.68075407e-01 5.44441223e-01
1.01485121e+00 -2.98191071e-01 -9.69423532e-01 -3.10553014e-01
5.16884267e-01 -3.49179298e-01 3.16299051e-01 -5.69915295e-01
6.04519188e-01 -2.28179827e-01 -9.53652710e-02 -3.79171580e-01
-8.51424597e-03 -6.38724566e-01 -1.47595378e-02 4.26768214e-01
5.41466800e-03 -1.54686093e-01 1.79997340e-01 7.02722549e-01
-3.56816351e-01 -1.53003544e-01 5.04791737e-01 -4.71266545e-02
-9.86388683e-01 8.73855889e-01 1.22293480e-01 4.30380076e-01
1.03754580e+00 -4.22117323e-01 -6.82819605e-01 -3.86835039e-01
-7.63697505e-01 5.88222563e-01 -1.93928093e-01 4.13330108e-01
6.18305981e-01 -1.45101225e+00 -4.96116698e-01 5.13866208e-02
-1.31406579e-02 5.34498215e-01 2.10247859e-01 9.51655507e-01
-5.23357809e-01 5.24003506e-01 4.49557118e-02 -1.99060202e-01
-9.17554975e-01 6.42364442e-01 4.58562583e-01 -4.16861564e-01
-5.46425581e-01 1.05653322e+00 1.75226126e-02 -1.03608596e+00
4.37371701e-01 -3.55555058e-01 -5.07353544e-01 -1.28740385e-01
-7.06396252e-02 6.61407471e-01 1.16945460e-01 -2.68186510e-01
-1.89504892e-01 3.77694070e-01 -1.45818144e-02 7.77752638e-01
1.20460188e+00 5.41012026e-02 -3.55906069e-01 1.83689088e-01
9.32341039e-01 -4.66227263e-01 -7.51960635e-01 -5.10898411e-01
-1.80146694e-01 1.60814837e-01 2.74626762e-01 -9.70450819e-01
-1.68980241e+00 6.61784887e-01 -1.49087846e-01 4.92755264e-01
1.14018393e+00 1.04918167e-01 6.64993227e-01 6.00934744e-01
3.02018136e-01 -9.82177258e-01 -4.50925753e-02 8.40620577e-01
4.74871725e-01 -1.03951538e+00 2.41716206e-01 -6.24303818e-01
-3.52481633e-01 1.08534265e+00 8.91054213e-01 1.14589646e-01
1.26867414e+00 -1.80233538e-01 -1.96757019e-01 -7.34729588e-01
-3.64891648e-01 -3.98014694e-01 5.19039631e-01 9.67679501e-01
1.33395150e-01 3.54564518e-01 -1.31037384e-01 7.64583051e-01
-2.99077094e-01 8.15086812e-02 3.21211755e-01 7.70752966e-01
-2.83632249e-01 -9.89661932e-01 2.06513777e-01 8.38999629e-01
1.50066122e-01 -3.92137170e-01 -7.72138774e-01 1.07863390e+00
2.45111719e-01 8.66331995e-01 -2.90330768e-01 -7.07730055e-01
2.64447421e-01 1.27681065e-02 4.99335170e-01 -6.74813509e-01
-4.46093023e-01 -7.10721314e-01 1.66401029e-01 -2.07172200e-01
-5.71606815e-01 2.11952150e-01 -1.37070358e+00 -6.85401082e-01
-7.35830188e-01 3.93771648e-01 3.55288029e-01 8.25217128e-01
5.84876537e-01 7.41433024e-01 6.70865476e-02 -3.30208093e-01
1.04116732e-02 -7.91012645e-01 -9.14677083e-01 -6.95960373e-02
-3.34523953e-02 -9.80282962e-01 -4.70594436e-01 -3.47687751e-01] | [7.310359477996826, 6.320670127868652] |
53378ef4-0a0f-4a57-ac80-7a57b655ca95 | ulip-learning-unified-representation-of | 2212.05171 | null | https://arxiv.org/abs/2212.05171v4 | https://arxiv.org/pdf/2212.05171v4.pdf | ULIP: Learning a Unified Representation of Language, Images, and Point Clouds for 3D Understanding | The recognition capabilities of current state-of-the-art 3D models are limited by datasets with a small number of annotated data and a pre-defined set of categories. In its 2D counterpart, recent advances have shown that similar problems can be significantly alleviated by employing knowledge from other modalities, such as language. Inspired by this, leveraging multimodal information for 3D modality could be promising to improve 3D understanding under the restricted data regime, but this line of research is not well studied. Therefore, we introduce ULIP to learn a unified representation of images, texts, and 3D point clouds by pre-training with object triplets from the three modalities. To overcome the shortage of training triplets, ULIP leverages a pre-trained vision-language model that has already learned a common visual and textual space by training with massive image-text pairs. Then, ULIP learns a 3D representation space aligned with the common image-text space, using a small number of automatically synthesized triplets. ULIP is agnostic to 3D backbone networks and can easily be integrated into any 3D architecture. Experiments show that ULIP effectively improves the performance of multiple recent 3D backbones by simply pre-training them on ShapeNet55 using our framework, achieving state-of-the-art performance in both standard 3D classification and zero-shot 3D classification on ModelNet40 and ScanObjectNN. ULIP also improves the performance of PointMLP by around 3% in 3D classification on ScanObjectNN, and outperforms PointCLIP by 28.8% on top-1 accuracy for zero-shot 3D classification on ModelNet40. Our code and pre-trained models are released at https://github.com/salesforce/ULIP. | ['Silvio Savarese', 'Juan Carlos Niebles', 'ran Xu', 'Caiming Xiong', 'Jiajun Wu', 'Roberto Martín-Martín', 'Chen Xing', 'Mingfei Gao', 'Le Xue'] | 2022-12-10 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Gao_ULIP_Learning_a_Unified_Representation_of_Language_Images_and_Point_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Gao_ULIP_Learning_a_Unified_Representation_of_Language_Images_and_Point_CVPR_2023_paper.pdf | cvpr-2023-1 | ['training-free-3d-point-cloud-classification', '3d-point-cloud-classification', '3d-classification', 'zero-shot-transfer-3d-point-cloud'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [-1.53782800e-01 -7.06996322e-02 -4.60327029e-01 -3.37736934e-01
-8.77781212e-01 -8.45612407e-01 9.90499616e-01 -2.34694764e-01
-7.84844086e-02 -1.52067155e-01 7.48270229e-02 -4.31943417e-01
2.41910607e-01 -5.78519940e-01 -9.29411769e-01 -2.70040840e-01
3.29064339e-01 8.17902386e-01 1.77228585e-01 -1.10932931e-01
-4.87336045e-04 7.71095037e-01 -1.57765055e+00 3.29057485e-01
4.56393629e-01 1.34774196e+00 1.13655828e-01 3.38050187e-01
-6.28695250e-01 3.52976322e-01 -5.22417016e-02 -2.47406349e-01
6.59894884e-01 7.46055394e-02 -6.14237368e-01 5.19976199e-01
1.00915098e+00 -4.21894252e-01 -5.05460382e-01 7.17815578e-01
3.98305327e-01 -1.64046511e-01 7.94945180e-01 -1.29942632e+00
-9.58697617e-01 1.21409059e-01 -7.34556079e-01 -3.14286739e-01
3.04487914e-01 3.30901295e-01 1.03655350e+00 -1.56786072e+00
6.50658727e-01 1.34034526e+00 7.49652267e-01 4.58665550e-01
-1.22674060e+00 -6.59806073e-01 1.59441531e-01 -3.57882641e-02
-1.40849757e+00 -3.67987633e-01 8.26904476e-01 -5.44801593e-01
1.34927142e+00 -7.44854212e-02 7.00552523e-01 1.20580399e+00
-2.43187353e-01 9.28705990e-01 8.51874709e-01 -3.57181102e-01
-1.25626191e-01 -2.60006357e-02 -2.41303146e-02 8.30286622e-01
-8.02491419e-03 8.37240517e-02 -7.99330890e-01 -7.20556127e-03
1.01511216e+00 9.19960812e-02 4.44690771e-02 -1.09249818e+00
-1.25420201e+00 6.88639224e-01 6.94574893e-01 2.06691399e-01
-3.01709380e-02 9.12981480e-02 2.80569017e-01 3.16257030e-01
6.79840028e-01 1.15237504e-01 -5.46497583e-01 2.66733486e-02
-7.29609489e-01 1.27039865e-01 7.27596998e-01 1.27013874e+00
8.90863776e-01 -2.20635943e-02 2.02145010e-01 9.98171389e-01
4.67261493e-01 1.16634810e+00 3.19470540e-02 -8.34074020e-01
7.19608009e-01 9.85762358e-01 -1.28291070e-01 -8.30877960e-01
-4.16449726e-01 -3.83988976e-01 -7.38458991e-01 2.60815501e-01
2.94915766e-01 3.61082673e-01 -1.28225946e+00 1.46335506e+00
3.73361379e-01 -7.05093592e-02 -9.66874883e-02 9.58046794e-01
1.03969526e+00 4.29991007e-01 -2.44330883e-01 4.91862774e-01
1.15223289e+00 -9.08326030e-01 4.56851758e-02 -3.85298371e-01
7.32768059e-01 -8.65084052e-01 1.26989555e+00 1.16918139e-01
-8.89415741e-01 -5.45674026e-01 -8.21726799e-01 -4.43482757e-01
-3.93021792e-01 1.78419068e-01 4.14208442e-01 5.05647421e-01
-1.02999330e+00 8.09947252e-02 -8.65590572e-01 -7.64802814e-01
7.75449634e-01 1.77733660e-01 -7.32034206e-01 -4.99597937e-01
-6.97394729e-01 1.03907430e+00 1.66275024e-01 -1.74188465e-01
-9.54141438e-01 -1.01920819e+00 -9.09636438e-01 -3.32334757e-01
4.12412494e-01 -9.90598083e-01 1.20989299e+00 -6.13318563e-01
-1.26499414e+00 1.42673028e+00 5.82997017e-02 -1.67458266e-01
5.80922425e-01 -1.92351118e-01 -1.45174429e-01 3.12516272e-01
2.05811664e-01 9.69032645e-01 9.38023329e-01 -1.50435901e+00
-4.03385729e-01 -6.68776989e-01 1.13124870e-01 3.89941543e-01
-2.44295359e-01 -3.82552713e-01 -9.54000413e-01 -3.20229024e-01
4.73631650e-01 -1.12410939e+00 5.05931862e-02 6.15683138e-01
-3.68617117e-01 -2.46631205e-01 1.06788647e+00 -1.52856052e-01
2.54286051e-01 -2.17035532e+00 1.57047570e-01 1.95702627e-01
2.52881408e-01 1.93921119e-01 -5.76296151e-01 3.29783887e-01
1.15214344e-02 1.17466055e-01 -4.46015559e-02 -7.01546907e-01
3.45513433e-01 3.71828347e-01 -3.62088203e-01 5.67410707e-01
2.59556949e-01 1.19450665e+00 -6.40417993e-01 -2.92177677e-01
6.35067582e-01 6.03832245e-01 -5.57017326e-01 2.04269364e-01
-4.91655678e-01 3.49437147e-01 -3.88271898e-01 1.01290071e+00
7.95157015e-01 -6.44692421e-01 -2.03851834e-02 -5.95890522e-01
-5.36449105e-02 1.97024763e-01 -8.55356455e-01 2.37880230e+00
-4.87364560e-01 4.49450940e-01 -1.25561252e-01 -1.13457179e+00
1.02340758e+00 7.66185746e-02 6.92068875e-01 -6.53625667e-01
1.67959750e-01 3.54561210e-01 -4.65842515e-01 -4.02451783e-01
2.53762335e-01 -6.08660802e-02 -1.33544877e-01 6.07765555e-01
3.84282380e-01 -7.53364146e-01 -2.63978124e-01 2.33526453e-01
9.39813673e-01 4.55720603e-01 -6.90820515e-02 9.84072760e-02
1.54292688e-01 1.46577775e-01 -5.39788641e-02 8.44595015e-01
3.33837117e-03 7.58976519e-01 1.42378628e-01 -4.44781601e-01
-1.28099430e+00 -1.28349972e+00 -2.62137830e-01 8.56265068e-01
2.20622227e-01 -3.72385412e-01 -1.54027224e-01 -9.15119767e-01
5.09972751e-01 5.09189487e-01 -5.13777494e-01 -7.66290305e-03
-2.51110315e-01 -2.01205000e-01 6.80326819e-01 5.52710235e-01
5.54896891e-01 -4.86800313e-01 -1.19899191e-01 -2.12589845e-01
3.47560570e-02 -1.52402973e+00 -2.50937462e-01 1.92788199e-01
-9.31017399e-01 -1.02237809e+00 -6.58090949e-01 -7.09679961e-01
6.56484544e-01 9.07071352e-01 1.21781898e+00 3.25027592e-02
8.51890352e-03 1.01070523e+00 -4.27447021e-01 -4.38654780e-01
-2.45317653e-01 2.87033290e-01 9.71393138e-02 -2.97355890e-01
5.49760699e-01 -6.65247619e-01 -3.70369315e-01 4.73258793e-01
-7.33308375e-01 1.94956124e-01 5.70041478e-01 5.77894032e-01
7.01685011e-01 -6.33448780e-01 1.55449092e-01 -3.87315780e-01
-7.57124946e-02 -3.83378536e-01 -4.59601611e-01 1.18040234e-01
-3.36849630e-01 -6.20561764e-02 2.79175222e-01 -3.79941255e-01
-7.30187058e-01 1.34247288e-01 -7.91410357e-02 -1.21771884e+00
-3.83607686e-01 4.64770228e-01 -1.75560012e-01 -3.69626015e-01
6.02967262e-01 2.16811076e-01 2.11224034e-01 -7.65175402e-01
7.78913319e-01 6.29254401e-01 3.52739811e-01 -6.72351658e-01
1.20028007e+00 7.50931978e-01 6.90841973e-02 -7.92516708e-01
-1.23507309e+00 -5.42327046e-01 -9.49300110e-01 -1.63481012e-01
6.91990316e-01 -1.18220687e+00 -4.93022501e-01 5.30456781e-01
-1.13278747e+00 -3.16919506e-01 -2.26339608e-01 3.56170654e-01
-6.76889956e-01 3.49609822e-01 -2.77652144e-01 -3.53534043e-01
-1.57949239e-01 -1.00459492e+00 1.57507896e+00 -2.16522828e-01
8.79051387e-02 -9.05829787e-01 -1.79169536e-01 6.30597234e-01
1.22509599e-01 -4.55315523e-02 9.91023719e-01 -6.33264124e-01
-8.91148567e-01 -2.79063553e-01 -5.44814289e-01 5.12031853e-01
1.08894818e-01 -2.48107851e-01 -9.44958210e-01 -1.91039935e-01
-9.41599086e-02 -6.73343062e-01 8.43876600e-01 1.32996872e-01
9.89193559e-01 2.61312097e-01 -3.93129617e-01 8.00993681e-01
1.27042234e+00 -3.50791544e-01 1.91012219e-01 2.36379728e-01
1.14708388e+00 3.94713670e-01 4.70389962e-01 2.48394638e-01
6.74448788e-01 7.23341227e-01 7.58948445e-01 -1.70814037e-01
-4.07469869e-01 -4.01150912e-01 2.22116441e-01 7.64263034e-01
8.68199542e-02 -1.70084506e-01 -1.36839974e+00 4.51321572e-01
-1.81584466e+00 -7.06018388e-01 5.45163117e-02 1.84609890e+00
5.62329948e-01 1.90321043e-01 1.68017298e-02 -2.41749853e-01
4.11739707e-01 3.14094126e-01 -7.78367043e-01 1.89603463e-01
-4.11483139e-01 7.73571283e-02 4.99174684e-01 1.73420295e-01
-1.10384536e+00 1.07780671e+00 5.71069145e+00 7.75072992e-01
-1.33811378e+00 5.02028018e-02 3.63799125e-01 -3.95481825e-01
-4.07968909e-01 -3.48675214e-02 -8.98666680e-01 -2.35482305e-02
5.45917511e-01 2.71042697e-02 2.86921680e-01 8.81728053e-01
3.46472524e-02 2.23787561e-01 -1.40452945e+00 1.41003883e+00
5.00208259e-01 -1.69581878e+00 4.27130073e-01 2.07780510e-01
7.32394993e-01 8.21728349e-01 2.16512188e-01 2.89192021e-01
2.33559564e-01 -1.04271245e+00 9.43267405e-01 4.47185427e-01
1.10351694e+00 -3.18135738e-01 2.97623813e-01 4.82666016e-01
-1.11634767e+00 1.75006360e-01 -4.58033711e-01 1.79976031e-01
1.12456962e-01 4.68469113e-01 -8.41915250e-01 6.69735968e-01
8.02837849e-01 1.19025910e+00 -5.41949153e-01 8.20615470e-01
-1.54055327e-01 2.69315302e-01 -6.79143012e-01 2.65478015e-01
5.08610129e-01 -4.74817902e-02 6.97748601e-01 7.90448129e-01
4.76551741e-01 1.87280178e-02 4.76001412e-01 9.20403659e-01
-3.59704882e-01 -2.62279082e-02 -9.70322311e-01 2.26693749e-02
4.65165854e-01 1.10924387e+00 -3.85961026e-01 -1.74432591e-01
-8.63909662e-01 6.89101934e-01 3.29002410e-01 2.73614228e-01
-7.03082263e-01 8.63442346e-02 6.39268041e-01 1.51544392e-01
6.32906616e-01 -6.81884468e-01 -3.98255497e-01 -1.34635937e+00
2.61627533e-03 -7.79113293e-01 1.61404639e-01 -1.12269568e+00
-1.69161284e+00 4.57894564e-01 -1.18413009e-02 -1.51743793e+00
8.94652456e-02 -9.71142113e-01 -5.48925251e-02 7.61279404e-01
-1.51498151e+00 -1.89117670e+00 -4.72927868e-01 8.35691988e-01
5.17301142e-01 -3.45752418e-01 7.38429308e-01 1.92199335e-01
-1.64240181e-01 3.96320283e-01 -2.84799673e-02 1.74328640e-01
8.29062045e-01 -9.67764854e-01 5.94483256e-01 4.27435756e-01
5.76143265e-01 2.05865040e-01 1.54579297e-01 -5.12833416e-01
-1.95623505e+00 -1.14071596e+00 5.66932976e-01 -9.56423938e-01
8.65062475e-01 -5.69091380e-01 -8.26323271e-01 7.63275385e-01
-1.09624267e-01 2.84423470e-01 5.56117058e-01 2.37062559e-01
-1.04656100e+00 -8.59923381e-03 -8.35960686e-01 5.65620184e-01
1.55816162e+00 -9.91953850e-01 -7.09320009e-01 5.36552608e-01
7.40125775e-01 -8.71067941e-01 -1.04730296e+00 4.88862634e-01
4.77451921e-01 -7.89497256e-01 1.26135039e+00 -5.00509441e-01
4.80526507e-01 -2.24824905e-01 -6.71987832e-01 -1.13322008e+00
-5.50957918e-02 -8.03552195e-02 -1.15506165e-01 8.49287450e-01
3.64880145e-01 -4.43987876e-01 8.96551132e-01 4.13320124e-01
-5.05663395e-01 -7.33491063e-01 -9.99268770e-01 -9.38792765e-01
2.89346874e-01 -9.78760123e-01 5.74812293e-01 9.54614460e-01
-4.28635597e-01 5.09849668e-01 -2.45416746e-01 1.50604710e-01
6.94311142e-01 4.38235521e-01 1.31876576e+00 -1.19971263e+00
5.45087084e-02 -4.11704153e-01 -4.91950750e-01 -1.61354554e+00
4.47885096e-01 -1.44841456e+00 -2.74742067e-01 -1.52611840e+00
1.40256941e-01 -7.00962007e-01 5.61913252e-02 9.05178010e-01
4.53055978e-01 6.40223742e-01 5.18255830e-01 5.06704509e-01
-5.44237852e-01 8.09163153e-01 1.49127698e+00 -4.42767680e-01
-1.19925849e-02 -3.81359220e-01 -5.06811500e-01 7.13438928e-01
4.53090459e-01 -2.29947388e-01 -3.69320691e-01 -9.71130729e-01
6.60046861e-02 -1.51273847e-01 7.96153009e-01 -7.31963754e-01
1.66250974e-01 -8.97757560e-02 5.45125663e-01 -1.10227048e+00
6.77455485e-01 -1.03840840e+00 -5.16101643e-02 -9.72079337e-02
-9.17857662e-02 -2.03590378e-01 4.30960119e-01 5.79743207e-01
-1.10590257e-01 1.39770582e-01 5.42852044e-01 -1.85340106e-01
-8.16466212e-01 7.12934196e-01 1.78527251e-01 -4.66318354e-02
7.67752230e-01 -4.20076847e-01 -5.73062122e-01 -1.68112800e-01
-6.70435369e-01 2.12402806e-01 8.78081322e-01 7.97876298e-01
5.82262576e-01 -1.57149792e+00 -5.36859274e-01 3.98853987e-01
5.52879214e-01 2.30400905e-01 2.96407938e-01 8.15486491e-01
-2.88976759e-01 5.85830152e-01 -1.43045038e-01 -1.39659286e+00
-9.70732212e-01 3.25722903e-01 4.57383454e-01 1.14723645e-01
-8.20848465e-01 6.71052933e-01 2.44060084e-01 -9.51980472e-01
1.72991902e-01 -2.80910790e-01 3.32907647e-01 -1.02871858e-01
5.89316748e-02 -1.41331062e-01 2.72036076e-01 -8.69142890e-01
-6.11269891e-01 1.08425915e+00 -6.61688596e-02 2.79221889e-02
1.49989009e+00 -1.43852860e-01 7.34439269e-02 5.51256597e-01
1.42568445e+00 -2.91023135e-01 -1.34719145e+00 -6.10382318e-01
-2.53668219e-01 -5.87566853e-01 1.33477464e-01 -8.30332756e-01
-1.02104104e+00 1.04834795e+00 4.17479217e-01 -1.01581523e-02
8.31718504e-01 6.25338197e-01 6.89115345e-01 5.83805621e-01
4.68844831e-01 -6.29139602e-01 4.54168558e-01 1.00340044e+00
1.04563618e+00 -1.52755880e+00 -6.21512234e-02 -3.58369261e-01
-5.20103037e-01 1.11208725e+00 6.43798411e-01 -8.49414766e-02
8.86006176e-01 -2.18228642e-02 2.35792652e-01 -5.06516278e-01
-7.47898996e-01 -2.48459026e-01 6.55092001e-01 6.97871745e-01
1.98723286e-01 -9.91525128e-02 5.76187611e-01 3.40365142e-01
-1.13846771e-01 -1.39962584e-01 5.64978421e-02 8.61205578e-01
-2.58973211e-01 -1.15603793e+00 -3.16487372e-01 3.75071198e-01
1.61592230e-01 -5.63851818e-02 -4.65232164e-01 1.01648700e+00
-3.31084393e-02 5.72745860e-01 2.93329030e-01 -4.17243063e-01
5.89049280e-01 6.67315498e-02 8.22651565e-01 -7.75245368e-01
-1.60354629e-01 1.45584628e-01 3.19310762e-02 -7.01770365e-01
-7.48559952e-01 -6.61808014e-01 -1.04071736e+00 -3.52776885e-01
-1.85205653e-01 -4.37937081e-01 7.96133101e-01 1.03544235e+00
5.99518836e-01 -1.08386204e-01 5.47952354e-01 -1.24344230e+00
-5.72563887e-01 -6.44925773e-01 -3.57421339e-01 3.82777393e-01
2.39305794e-01 -9.73381877e-01 -3.45844746e-01 -8.77258107e-02] | [8.160085678100586, -3.319859027862549] |
b982c856-7fa3-4c88-ae7e-78cfa7a844a7 | improved-soccer-action-spotting-using-both | 2011.04258 | null | https://arxiv.org/abs/2011.04258v1 | https://arxiv.org/pdf/2011.04258v1.pdf | Improved Soccer Action Spotting using both Audio and Video Streams | In this paper, we propose a study on multi-modal (audio and video) action spotting and classification in soccer videos. Action spotting and classification are the tasks that consist in finding the temporal anchors of events in a video and determine which event they are. This is an important application of general activity understanding. Here, we propose an experimental study on combining audio and video information at different stages of deep neural network architectures. We used the SoccerNet benchmark dataset, which contains annotated events for 500 soccer game videos from the Big Five European leagues. Through this work, we evaluated several ways to integrate audio stream into video-only-based architectures. We observed an average absolute improvement of the mean Average Precision (mAP) metric of $7.43\%$ for the action classification task and of $4.19\%$ for the action spotting task. | ['Stéphane Dupont', 'Bastien Vanderplaetse'] | 2020-11-09 | null | null | null | null | ['action-spotting'] | ['computer-vision'] | [ 2.12907955e-01 -4.33623433e-01 -2.06608519e-01 -2.53617078e-01
-1.05243099e+00 -3.62092763e-01 3.15235883e-01 1.75787598e-01
-8.47223461e-01 6.32586598e-01 2.73579180e-01 3.71230572e-01
-1.73282757e-01 -5.45528352e-01 -9.66162264e-01 -5.58851361e-01
-4.30496961e-01 3.57530192e-02 6.41712904e-01 -2.21135467e-01
3.05026054e-01 2.95104444e-01 -1.92124987e+00 1.11335242e+00
-1.83619589e-01 1.54532719e+00 -1.68853492e-01 9.92543101e-01
4.32827681e-01 1.36736453e+00 -7.99000800e-01 -2.43924066e-01
1.74247652e-01 -4.30107147e-01 -7.66402185e-01 -1.88058734e-01
4.65954274e-01 -2.64634311e-01 -5.89305997e-01 7.53390729e-01
4.93935853e-01 5.59451520e-01 8.21191445e-02 -1.58914232e+00
2.90250480e-01 7.69482851e-01 -4.50739145e-01 9.01095212e-01
5.92477858e-01 6.51839525e-02 1.17731607e+00 -7.33199418e-01
7.20738471e-01 9.47213352e-01 7.11303949e-01 1.15599312e-01
-7.58028388e-01 -9.44073856e-01 1.67222202e-01 9.46167827e-01
-1.41391313e+00 -5.84007859e-01 5.14367640e-01 -5.65190077e-01
1.03501832e+00 9.12294686e-02 9.44106281e-01 1.03534913e+00
2.81040430e-01 1.27202868e+00 5.61511934e-01 -2.79815912e-01
1.60464615e-01 -3.00882936e-01 9.85106379e-02 4.05533344e-01
-2.90964246e-01 9.76633728e-02 -1.31445956e+00 1.25061557e-01
6.12314820e-01 -8.17588791e-02 -2.05471084e-01 2.50729751e-02
-1.49067545e+00 7.46639550e-01 1.75221875e-01 3.42581987e-01
-5.59003353e-01 6.73106372e-01 1.04832947e+00 3.64413440e-01
2.67107069e-01 3.58958453e-01 -3.43829870e-01 -8.80858004e-01
-1.01406300e+00 4.91031766e-01 4.87270027e-01 5.87001085e-01
3.23869824e-01 -7.38599747e-02 -3.69040430e-01 7.81038582e-01
-8.80072042e-02 -8.37097690e-03 5.09113014e-01 -1.37088251e+00
6.08448148e-01 4.38830227e-01 1.82453141e-01 -1.10214829e+00
-5.09064138e-01 -1.93680674e-01 -1.70566320e-01 1.19271530e-02
6.96988165e-01 -2.02924699e-01 -4.20036793e-01 1.50786948e+00
4.96128947e-02 4.90994692e-01 -1.75723717e-01 1.08226228e+00
7.78356910e-01 5.84329545e-01 1.57209352e-01 -9.86375958e-02
1.58495128e+00 -1.02753389e+00 -7.14040339e-01 -2.04922378e-01
6.68462873e-01 -6.73573017e-01 7.75896966e-01 8.03863764e-01
-1.00524199e+00 -8.36514950e-01 -1.25907540e+00 2.30967954e-01
-1.70595810e-01 4.39554781e-01 5.36115170e-01 2.27479964e-01
-6.94305837e-01 6.31475627e-01 -1.02029371e+00 -2.15510368e-01
1.60795122e-01 1.72979340e-01 -6.06652200e-01 2.06230000e-01
-1.29876781e+00 4.07839179e-01 7.09410131e-01 2.67160274e-02
-1.34456336e+00 -6.00679696e-01 -7.75042772e-01 7.77559578e-02
7.96071410e-01 -1.23331919e-02 1.53009391e+00 -1.39201128e+00
-1.27141345e+00 7.91071236e-01 1.95965722e-01 -1.00160360e+00
3.91459256e-01 -5.84730804e-01 -6.32170141e-01 4.65477407e-01
1.15033306e-01 6.67330682e-01 5.26554167e-01 -4.12450880e-01
-1.43378460e+00 -2.55105585e-01 4.48950499e-01 2.04325050e-01
-1.00069329e-01 3.04267853e-01 -4.49351877e-01 -7.79353678e-01
4.45185378e-02 -9.28045511e-01 1.12876005e-01 -1.66881472e-01
9.92912706e-03 -3.14653277e-01 6.27783477e-01 -7.48823881e-01
1.31251264e+00 -2.30832863e+00 7.03527406e-02 -7.40377158e-02
-2.85548158e-02 -2.20335629e-02 2.19850801e-02 3.86020541e-01
-1.86544597e-01 -3.45649242e-01 4.52381015e-01 -5.66107184e-02
-4.50324789e-02 -3.36013511e-02 -2.21035853e-01 4.36241180e-01
-5.90123124e-02 5.68874121e-01 -7.66289651e-01 -4.55359370e-01
6.10097237e-02 3.88358682e-01 -5.63626647e-01 -4.02136259e-02
-9.21618119e-02 1.69737235e-01 -5.59122711e-02 6.53991759e-01
-5.36189303e-02 9.79573801e-02 8.54059160e-02 -4.05194491e-01
-7.46012330e-02 3.50391150e-01 -1.63797355e+00 2.12452507e+00
-2.46368900e-01 1.17238450e+00 -6.77792132e-02 -1.15226674e+00
5.66766798e-01 5.90336084e-01 9.21431720e-01 -7.63106346e-01
2.30796933e-01 3.25536020e-02 3.76138389e-02 -4.81262296e-01
8.59488785e-01 1.60288922e-02 -4.33417231e-01 5.12205362e-01
5.33626437e-01 6.07358992e-01 8.70306432e-01 4.25317064e-02
1.45941854e+00 3.31294715e-01 2.27439821e-01 9.29119661e-02
4.95377600e-01 8.05981681e-02 7.06351340e-01 7.79412687e-01
-4.64149475e-01 4.81479883e-01 6.12744331e-01 -7.45263636e-01
-6.37266576e-01 -7.24090755e-01 4.22816336e-01 1.78400028e+00
1.46334777e-02 -7.60856986e-01 -6.32317662e-01 -6.37539029e-01
-2.96375841e-01 3.34953606e-01 -7.07274199e-01 -2.13100478e-01
-9.26125467e-01 -5.85299432e-01 8.36051047e-01 9.86982822e-01
6.66956127e-01 -1.16710043e+00 -9.34227884e-01 4.37579155e-01
-5.95053375e-01 -1.33322442e+00 -4.84943420e-01 1.60920605e-01
-5.31894565e-01 -1.40145683e+00 -5.76285541e-01 -5.56477487e-01
-1.64552882e-01 1.92018375e-01 1.02263570e+00 -3.66773278e-01
-1.72865659e-01 4.28583652e-01 -6.26956403e-01 -4.03502911e-01
-7.81793892e-02 1.89368322e-01 1.43228412e-01 3.86961639e-01
7.02033937e-01 -3.98748815e-01 -7.25422800e-01 7.45963812e-01
-7.58157551e-01 -1.07771143e-01 4.42312002e-01 5.72206318e-01
7.31271088e-01 -9.63031687e-03 3.12027574e-01 -1.50725469e-01
2.85818875e-01 -2.63266981e-01 -4.69685227e-01 4.97256964e-02
1.06914692e-01 -4.04869258e-01 1.24821626e-01 -4.72074479e-01
-6.12951577e-01 4.17685062e-01 -2.11066574e-01 -4.81828123e-01
-3.07881594e-01 5.27781606e-01 3.57206129e-02 8.75885487e-02
6.42360866e-01 2.97871411e-01 -3.15131873e-01 -4.00991797e-01
-9.41336751e-02 5.50672591e-01 6.92453563e-01 -4.26921874e-01
-1.56960301e-02 5.39800942e-01 -2.16791019e-01 -7.73644388e-01
-8.29799592e-01 -7.12202370e-01 -6.72864020e-01 -9.15464878e-01
1.06156266e+00 -1.23076689e+00 -1.18945181e+00 5.10689139e-01
-9.00157630e-01 -1.86548099e-01 -1.79822832e-01 9.36804056e-01
-8.93738806e-01 2.11001381e-01 -6.40244305e-01 -5.10520458e-01
-9.65356380e-02 -1.12621534e+00 1.11668289e+00 8.02911296e-02
-4.08691496e-01 -3.95840138e-01 1.61635146e-01 6.86712801e-01
-8.61859918e-02 4.02260870e-01 1.16760336e-01 -8.61375690e-01
-4.12377715e-01 -5.07331729e-01 2.40114913e-03 3.46604556e-01
-3.02380085e-01 -2.60802388e-01 -9.80499268e-01 -2.48035699e-01
-1.63033918e-01 -4.43130404e-01 9.28946137e-01 4.66948271e-01
1.20569038e+00 3.13425809e-02 -1.79300502e-01 3.48835915e-01
1.10962534e+00 6.29379153e-01 7.16197670e-01 4.91255283e-01
3.53511333e-01 3.56988132e-01 1.16254318e+00 5.85603595e-01
1.67388897e-02 1.08179557e+00 3.97176236e-01 4.50286120e-01
-4.15585674e-02 -1.90298662e-01 8.21768403e-01 3.76728803e-01
-4.80060518e-01 -2.76691884e-01 -8.32188368e-01 7.31613100e-01
-2.09580183e+00 -1.37887907e+00 1.58412933e-01 1.99407840e+00
4.80072826e-01 4.35025662e-01 6.43134832e-01 5.91042936e-01
7.18997657e-01 1.60833299e-01 -2.25630358e-01 -1.82884768e-01
5.55580407e-02 2.60704666e-01 3.85587752e-01 -1.05645722e-02
-1.64150321e+00 6.11778498e-01 5.83798456e+00 1.15842140e+00
-1.04218030e+00 2.53757000e-01 3.08955461e-01 -8.33661079e-01
7.83942699e-01 -4.17178333e-01 -8.89028728e-01 4.47732031e-01
1.32534170e+00 5.49317524e-02 -2.87343282e-02 7.71527886e-01
2.89948910e-01 -5.74168026e-01 -1.36791980e+00 1.11511385e+00
2.27572188e-01 -1.51978135e+00 -2.69411683e-01 -2.37147123e-01
3.09802264e-01 8.11406970e-02 -3.13340992e-01 5.20726085e-01
-5.44043258e-02 -8.69437397e-01 1.15777946e+00 4.39934313e-01
3.91883373e-01 -9.28192735e-01 8.11168611e-01 1.02257036e-01
-1.47608280e+00 -3.89763504e-01 2.34969422e-01 -3.37908238e-01
4.19724017e-01 6.20169230e-02 -5.20633340e-01 3.23066711e-01
1.16928828e+00 1.05879653e+00 -3.24483007e-01 1.28474271e+00
-1.99499354e-01 6.50206685e-01 -2.75393248e-01 7.03246891e-02
4.19673443e-01 3.55037510e-01 5.65142274e-01 1.17549431e+00
3.57150793e-01 1.07143149e-01 3.04112285e-01 1.39646947e-01
8.20447877e-02 -1.63478199e-02 -2.47650340e-01 -4.56097275e-02
3.06172445e-02 1.05935776e+00 -1.09511685e+00 -3.23299289e-01
-3.54292810e-01 7.01811671e-01 -1.52192995e-01 -4.60700393e-02
-1.43195796e+00 -6.44728124e-01 8.52199852e-01 2.84495950e-01
6.00553870e-01 -1.20821171e-01 2.91642129e-01 -9.72051442e-01
5.40738925e-02 -8.44733834e-01 8.11206818e-01 -8.49539995e-01
-6.50244951e-01 3.41622293e-01 6.21287562e-02 -1.53214765e+00
-2.95626044e-01 -6.49529874e-01 -3.30972016e-01 1.76330693e-02
-6.65178955e-01 -7.75697052e-01 -2.66640872e-01 5.85059166e-01
8.77331614e-01 -2.57285386e-01 4.23975557e-01 8.17326844e-01
-5.21970570e-01 5.87163866e-01 -1.80895597e-01 5.69106460e-01
6.66049123e-01 -7.47270703e-01 1.41178602e-02 8.12752664e-01
7.27221251e-01 -2.13468354e-02 8.08856428e-01 -3.39391142e-01
-1.06692374e+00 -8.84983718e-01 6.42080128e-01 -2.75956601e-01
7.60036886e-01 -2.84643769e-01 -4.23563063e-01 8.63028169e-01
1.08515114e-01 -2.58299232e-01 7.37932146e-01 -8.29496309e-02
-5.89411631e-02 -3.26360166e-01 -7.55715132e-01 2.75572002e-01
1.11852777e+00 -5.96779227e-01 -5.17819643e-01 1.93345472e-01
5.02186298e-01 -5.37173212e-01 -1.11707735e+00 2.52731800e-01
8.75731945e-01 -1.19112706e+00 9.97323751e-01 -6.49060845e-01
5.87741137e-01 -3.14907551e-01 -4.10854459e-01 -1.09751344e+00
4.54526506e-02 -2.99425811e-01 -4.26481590e-02 8.13281000e-01
2.84408540e-01 8.51339549e-02 8.51524055e-01 -7.47506507e-03
-2.89262265e-01 -3.68982136e-01 -1.32429552e+00 -5.81130862e-01
-5.47498226e-01 -8.51140916e-01 2.06524298e-01 6.46976471e-01
1.42782122e-01 6.45035133e-02 -5.16459703e-01 -1.44332752e-01
2.10070580e-01 1.12029295e-02 7.11449742e-01 -9.39895034e-01
-4.68527347e-01 -3.57401431e-01 -1.02380240e+00 -9.38000917e-01
2.82709044e-03 -4.57364529e-01 1.11578465e-01 -9.81453061e-01
2.75771711e-02 4.05238926e-01 -8.26349258e-01 5.04594982e-01
4.14610714e-01 6.81319535e-01 2.85730690e-01 -8.61154422e-02
-1.23194432e+00 1.19889170e-01 6.91429794e-01 -1.99471444e-01
-2.66765989e-02 2.01035351e-01 -1.39640421e-01 8.18494737e-01
7.49230564e-01 -5.21102548e-01 -1.76899120e-01 -4.37467724e-01
5.06771147e-01 4.23144758e-01 6.42878890e-01 -1.60408986e+00
2.95829177e-01 -6.35679364e-02 2.67095000e-01 -7.70395100e-01
8.23784232e-01 -5.55817723e-01 2.17943132e-01 6.18854225e-01
-6.44763947e-01 9.25696269e-02 4.77250546e-01 7.37568736e-01
-7.06243098e-01 1.59350887e-01 5.62833905e-01 -2.00069159e-01
-1.34092867e+00 -1.08236680e-02 -7.65379429e-01 7.07551613e-02
1.36258495e+00 -2.18641445e-01 -1.14575528e-01 -3.94464731e-01
-1.24246299e+00 1.58761516e-02 -2.33016700e-01 7.39655912e-01
5.31009555e-01 -1.47217906e+00 -5.38780868e-01 8.59649479e-02
2.80817747e-01 -6.88231826e-01 4.02282536e-01 1.17139030e+00
-6.28226280e-01 5.62562287e-01 -6.49095893e-01 -6.83960378e-01
-1.55564618e+00 1.53345436e-01 4.54794943e-01 -1.56099290e-01
-3.75821412e-01 1.06357038e+00 -2.10471764e-01 1.36813387e-01
5.60608804e-01 -4.97835070e-01 -5.16465902e-01 5.79286277e-01
9.42463934e-01 8.03793728e-01 2.79534817e-01 -8.13174367e-01
-5.93548179e-01 2.35637411e-01 9.44709703e-02 -2.86015660e-01
1.32317030e+00 2.09779963e-01 1.72884658e-01 7.22656906e-01
1.02776408e+00 -5.66065729e-01 -1.16714275e+00 -4.30445895e-02
2.08030686e-01 -3.74176562e-01 1.12488359e-01 -6.78436518e-01
-1.34812498e+00 8.42645884e-01 8.13464582e-01 1.47808433e-01
1.11295736e+00 -5.96061312e-02 9.14223671e-01 4.29234415e-01
5.63360512e-01 -1.53804898e+00 2.61790544e-01 4.99011159e-01
7.85109878e-01 -1.06741738e+00 -1.81169987e-01 1.16173681e-02
-7.24221051e-01 1.02938426e+00 6.74711823e-01 -2.47131184e-01
6.41700089e-01 1.37299359e-01 -1.12788267e-01 -4.01878387e-01
-9.43646193e-01 -2.46019781e-01 2.80205160e-01 1.70631677e-01
3.34959179e-01 -1.60848290e-01 -3.68821204e-01 8.44196796e-01
-1.49259558e-02 4.86305743e-01 4.88897920e-01 1.03956056e+00
-5.90367436e-01 -6.01879120e-01 -3.61043185e-01 3.43741626e-01
-8.84319186e-01 3.15751135e-01 -3.01383018e-01 7.91078031e-01
3.81571621e-01 9.52506304e-01 3.46619070e-01 -7.46464610e-01
4.90307003e-01 1.76526353e-01 3.80687058e-01 -3.34803879e-01
-8.97707880e-01 5.14174178e-02 4.26092088e-01 -1.17135894e+00
-7.74377286e-01 -8.04568231e-01 -1.01085889e+00 -1.96314529e-01
1.15664423e-01 2.97908217e-01 4.66219395e-01 8.63838851e-01
2.44616076e-01 7.65594244e-01 1.84578910e-01 -9.58878577e-01
-1.23274932e-02 -9.99717653e-01 -6.33276522e-01 3.34696501e-01
1.16994688e-02 -8.73144925e-01 -1.94307193e-01 2.29606941e-01] | [7.994211673736572, 0.18361078202724457] |
70cbbf19-a9d9-491d-9236-adbee81f628f | memory-augmented-neural-networks-for | 1802.00938 | null | https://arxiv.org/abs/1802.00938v2 | https://arxiv.org/pdf/1802.00938v2.pdf | DeepProcess: Supporting business process execution using a MANN-based recommender system | Process-aware Recommender systems can provide critical decision support functionality to aid business process execution by recommending what actions to take next. Based on recent advances in the field of deep learning, we present a novel memory-augmented neural network (MANN) based approach for constructing a process-aware recommender system. We propose a novel network architecture, namely Write-Protected Dual Controller Memory-Augmented Neural Network (DCw-MANN), for building prescriptive models. To evaluate the feasibility and usefulness of our approach, we consider three real-world datasets and show that our approach leads to better performance on several baselines for the task of suffix recommendation and next task prediction. | ['Hoa Dam', 'Aditya Ghose', 'Truyen Tran', 'Asjad Khan', 'Renuka Sindhgatta', 'Kien Do', 'Hung Le'] | 2018-02-03 | null | null | null | null | ['activity-prediction', 'activity-prediction'] | ['computer-vision', 'time-series'] | [ 5.06236315e-01 -7.42267370e-02 -5.45266628e-01 -4.06760126e-01
-2.64860749e-01 -4.52777475e-01 9.08032060e-01 -2.19855160e-02
-8.54819715e-02 1.12207204e-01 6.69231296e-01 -1.10742438e+00
-4.52659965e-01 -8.74279916e-01 -5.99491119e-01 -1.71386585e-01
1.18032888e-01 7.48778701e-01 1.84551433e-01 -3.35489601e-01
7.31721938e-01 5.91399312e-01 -1.39449775e+00 7.53555954e-01
5.26218057e-01 1.12220335e+00 -4.42941021e-03 7.21256316e-01
-3.46753716e-01 1.47584450e+00 -2.35036641e-01 -3.33267331e-01
2.14741349e-01 1.31879106e-01 -8.83523107e-01 -4.06576514e-01
8.59328359e-02 -5.45129061e-01 -6.42253757e-01 2.42070511e-01
3.33485723e-01 5.61897278e-01 7.02537835e-01 -8.66151035e-01
-9.23532009e-01 1.20092475e+00 -2.17034817e-01 3.62943679e-01
1.44454539e-01 3.79804373e-01 1.21517444e+00 -7.20996022e-01
5.71398735e-01 1.10898721e+00 6.03842199e-01 7.00602591e-01
-1.14434433e+00 -5.62890649e-01 5.78527451e-01 2.97155619e-01
-6.04572177e-01 -4.29125786e-01 4.81674433e-01 -3.38077813e-01
1.42674327e+00 -2.96806563e-02 4.65039760e-01 1.46196520e+00
7.24361241e-01 1.13847017e+00 6.79983377e-01 -2.62305945e-01
3.87850225e-01 -6.18193567e-01 7.33829379e-01 3.23023021e-01
2.39138946e-01 2.24665448e-01 -8.22887599e-01 -2.27407902e-01
6.81925297e-01 5.39376378e-01 -1.56631432e-02 -3.21114630e-01
-1.42195082e+00 6.87756300e-01 -5.27446009e-02 3.00113171e-01
-6.11333311e-01 6.37083411e-01 4.39247936e-01 4.95593995e-01
2.24354252e-01 8.83268774e-01 -5.71568191e-01 -4.82174993e-01
-8.81454349e-01 2.28916541e-01 1.33922994e+00 8.29513073e-01
1.22538373e-01 3.40292513e-01 -7.92790234e-01 3.53229821e-01
2.77745068e-01 3.78029317e-01 5.91069818e-01 -1.14149344e+00
3.82323474e-01 3.87095183e-01 2.06911236e-01 -4.35861170e-01
-3.79744321e-01 -3.47018242e-01 -5.94834864e-01 -3.28175992e-01
1.32450864e-01 -1.40413135e-01 -7.92603552e-01 9.34875727e-01
-4.91609275e-01 3.73674184e-01 1.00314289e-01 4.71241146e-01
5.06418467e-01 7.51040280e-01 3.36596370e-01 -1.02247059e-01
1.01905167e+00 -1.49773717e+00 -6.96776748e-01 -1.58073366e-01
3.52101535e-01 -3.04717779e-01 1.12245417e+00 8.06919754e-01
-1.02658856e+00 -7.20804214e-01 -1.22587550e+00 4.17866141e-01
5.06142862e-02 -2.08873376e-01 1.17755508e+00 7.10786283e-01
-1.07226038e+00 1.17284060e+00 -9.74923015e-01 -2.47993380e-01
2.55297780e-01 6.50633514e-01 7.85140991e-02 9.57684033e-03
-8.35534990e-01 6.67961538e-01 4.03347403e-01 -1.76221188e-02
-1.13429058e+00 -5.30847788e-01 -9.84410271e-02 4.84725952e-01
8.89631629e-01 -7.96648860e-01 1.85524261e+00 -6.50254667e-01
-1.89890790e+00 -8.84313807e-02 7.22573400e-02 -8.87224793e-01
-1.28203005e-01 -6.97952509e-01 -8.71020436e-01 -1.62609652e-01
-7.88007975e-01 8.77954140e-02 8.58821630e-01 -9.38527286e-01
-9.71190631e-01 -1.69302642e-01 -7.26133212e-02 -9.12437066e-02
-4.23916638e-01 1.37022346e-01 -4.02946770e-01 -5.53522587e-01
-2.11552605e-01 -9.10552979e-01 -5.21545827e-01 -6.76642537e-01
-1.78559825e-01 -4.35009211e-01 5.73691905e-01 -4.58163202e-01
1.46877444e+00 -1.88373375e+00 -1.22426726e-01 3.72849256e-01
3.32129866e-01 4.05163705e-01 -4.84319359e-01 5.28755426e-01
9.44531113e-02 7.64661357e-02 3.96861911e-01 4.52254489e-02
2.91927099e-01 3.16098541e-01 -6.82127178e-01 -2.70587534e-01
1.25295401e-01 9.93735850e-01 -7.58322358e-01 1.82987258e-01
4.97015454e-02 1.31710321e-01 -5.37526727e-01 4.32046890e-01
-6.39711738e-01 1.46975070e-01 -4.56208616e-01 6.83422923e-01
7.68544078e-02 -3.18884403e-01 6.85769916e-01 1.35289416e-01
9.13337022e-02 6.83451951e-01 -1.06896317e+00 1.48350561e+00
-8.18259358e-01 4.29617971e-01 -2.65266031e-01 -6.05571985e-01
1.11470115e+00 2.47103900e-01 5.12286901e-01 -8.88827085e-01
5.66719212e-02 -2.18106747e-01 2.88190067e-01 -7.33195692e-02
9.05921400e-01 2.26629332e-01 4.91818339e-02 1.22808588e+00
5.43454252e-02 4.29515630e-01 -1.37392087e-02 -5.01773413e-03
1.68632615e+00 5.11936486e-01 2.64684170e-01 -1.52832597e-01
5.45331359e-01 -3.31118703e-01 5.46111763e-01 1.05342436e+00
4.72863168e-02 3.17058526e-02 4.44065899e-01 -7.70525992e-01
-7.81682611e-01 -8.63266230e-01 4.91343975e-01 2.04523802e+00
-3.70186180e-01 -6.31282389e-01 -2.49948204e-01 -8.04674625e-01
5.37676215e-02 1.30167687e+00 -3.27194840e-01 -2.10654259e-01
-1.04613364e+00 -4.01101530e-01 3.46142024e-01 1.28632975e+00
1.21801421e-01 -1.31432855e+00 -2.82262802e-01 5.98920763e-01
3.41140866e-01 -8.38507891e-01 -6.34661496e-01 6.98633492e-01
-1.14830375e+00 -8.22286725e-01 5.01089916e-02 -4.56846297e-01
-1.39881223e-01 2.40767166e-01 1.40459144e+00 -2.35687330e-01
6.54816926e-01 3.72288734e-01 -2.47282133e-01 -4.11357999e-01
-7.83645928e-01 4.64647382e-01 9.46397409e-02 -2.39146799e-01
5.72428823e-01 -6.30483389e-01 -5.18082082e-01 3.18095952e-01
-4.91136819e-01 -2.61195928e-01 9.30084467e-01 7.46049345e-01
3.82167995e-01 -9.91176441e-02 4.52866465e-01 -1.38441002e+00
1.12608790e+00 -2.92890340e-01 -3.41587096e-01 2.38007024e-01
-1.40562952e+00 4.26232070e-01 7.49650896e-01 -6.75816596e-01
-1.30988705e+00 -1.44995004e-01 -2.47251779e-01 -2.72089571e-01
-1.73857301e-01 4.94488150e-01 -2.73927897e-01 4.06770587e-01
6.35299325e-01 2.21548140e-01 -8.50838274e-02 -4.12012488e-01
7.34342992e-01 6.90693736e-01 7.04012513e-01 -8.14600766e-01
3.98877770e-01 9.03606117e-02 -6.56579211e-02 1.19340613e-01
-6.56768978e-01 -4.29101169e-01 -4.04015839e-01 1.30230844e-01
4.64673370e-01 -7.53103673e-01 -1.16569686e+00 1.99570693e-03
-7.42468357e-01 -7.10347116e-01 -2.30782047e-01 1.81984514e-01
-9.76526022e-01 1.34831339e-01 -1.03083205e+00 -6.46398127e-01
-1.13879025e+00 -9.92269099e-01 4.97951984e-01 -1.85048863e-01
-5.48596442e-01 -9.81772721e-01 -6.76152483e-02 5.01040995e-01
8.01614642e-01 -6.19109571e-01 1.41573727e+00 -1.58043540e+00
-8.12932372e-01 -4.58466634e-02 8.25380627e-03 3.35259914e-01
-1.74594924e-01 -5.13686500e-02 -7.90874958e-01 3.84512618e-02
-2.24508345e-01 2.57111341e-01 7.95070350e-01 8.23399499e-02
1.52537155e+00 -2.40634948e-01 -4.47723687e-01 4.68479007e-01
1.14576912e+00 6.65589631e-01 7.37082243e-01 5.14543891e-01
8.22178602e-01 2.76838779e-01 6.95243776e-01 4.27368164e-01
2.78808147e-01 3.56983364e-01 2.85414159e-01 5.96399903e-01
5.48152551e-02 -3.48330736e-01 5.85359633e-01 8.91851723e-01
-4.58933353e-01 -5.60339808e-01 -1.15485382e+00 2.41299048e-01
-2.22479630e+00 -1.11875665e+00 -6.13483898e-02 1.98109102e+00
4.93668973e-01 6.10957682e-01 5.66686988e-02 2.72196438e-02
3.57858658e-01 8.24770927e-02 -5.92864156e-01 -1.04101038e+00
4.08904165e-01 6.46742105e-01 5.54838717e-01 -1.76433071e-01
-1.19116700e+00 8.87531161e-01 7.33715916e+00 6.73632860e-01
-1.07288885e+00 2.35818014e-01 3.69741797e-01 -2.84570187e-01
-3.99010390e-01 -7.35335499e-02 -9.57286835e-01 2.33143643e-01
1.88098156e+00 -7.72474930e-02 4.71927643e-01 1.12141085e+00
2.35590339e-01 3.46398592e-01 -1.71234953e+00 7.57012248e-01
-1.57246962e-01 -1.85709417e+00 3.67150784e-01 1.89487562e-01
8.34837258e-01 1.03258401e-01 3.05595845e-01 8.37798595e-01
1.10720634e+00 -1.02852571e+00 4.12505180e-01 8.32899749e-01
3.88862371e-01 -1.10398638e+00 8.61049056e-01 1.80187359e-01
-8.02151620e-01 -7.28242278e-01 -1.58595711e-01 -2.16032237e-01
-1.75514191e-01 2.86139399e-01 -1.03775048e+00 2.98298568e-01
3.29572171e-01 5.95207393e-01 -2.06259683e-01 5.39436817e-01
-1.66799039e-01 1.12783122e+00 2.44240612e-01 -1.69671059e-01
1.43179610e-01 1.11183241e-01 9.74476188e-02 1.23413408e+00
2.58265585e-01 -2.43154824e-01 2.54289389e-01 5.08786023e-01
-8.29280913e-02 1.63188532e-01 -3.38141650e-01 -1.99160293e-01
4.66880649e-01 1.10553467e+00 -6.11759126e-01 -1.38423145e-01
-5.04772246e-01 7.48664320e-01 1.36137947e-01 1.64090917e-01
-4.05390382e-01 9.65596810e-02 6.66189492e-01 2.53248364e-01
8.30762506e-01 -3.00221711e-01 -4.38653976e-01 -7.32942760e-01
-5.29518545e-01 -1.32317042e+00 3.65961879e-01 -5.06283998e-01
-1.19670320e+00 4.54196900e-01 -4.56374139e-01 -9.11303997e-01
-5.89790046e-01 -6.56855166e-01 -8.34446132e-01 6.89061761e-01
-1.13024116e+00 -1.38702381e+00 1.20406179e-02 2.43623465e-01
7.24840879e-01 -7.75786877e-01 8.08387816e-01 -4.06915620e-02
-4.31804746e-01 5.42160511e-01 3.86015207e-01 -1.54292241e-01
8.65985930e-01 -1.26410675e+00 9.11397040e-01 6.61984801e-01
6.01122417e-02 1.14878511e+00 5.03669381e-01 -6.89756811e-01
-2.01613474e+00 -1.29694974e+00 5.34232318e-01 -7.37503767e-01
9.40142751e-01 -7.30536059e-02 -6.72317088e-01 1.00286639e+00
3.21465582e-01 -5.36905646e-01 1.08803499e+00 8.06590140e-01
-4.09937292e-01 -3.39297354e-01 -6.21026933e-01 7.01662540e-01
1.16866922e+00 -4.35933828e-01 -6.75991774e-01 -7.59984627e-02
9.44968343e-01 -4.41433804e-04 -1.17666495e+00 2.31501758e-01
6.48454070e-01 -7.11537719e-01 7.81092703e-01 -7.23439276e-01
7.23690748e-01 -5.35611622e-03 -3.62327576e-01 -1.20259678e+00
-7.65186906e-01 -8.69986355e-01 -1.14414740e+00 9.54380035e-01
3.84566456e-01 -2.24526331e-01 1.07061315e+00 4.93058503e-01
-3.25673223e-01 -9.21658099e-01 -2.43206143e-01 -5.90134382e-01
6.44474104e-02 -8.13571215e-01 1.06151032e+00 5.25594890e-01
-8.96601081e-02 7.15910673e-01 -4.20256615e-01 -2.19173312e-01
1.75167471e-02 3.23380560e-01 7.25409210e-01 -1.52576697e+00
-8.46544981e-01 -5.76611161e-01 1.16503492e-01 -1.31490409e+00
1.86847970e-01 -7.95459151e-01 -1.19530968e-01 -1.63382947e+00
1.78712174e-01 -2.88003206e-01 -8.88215482e-01 4.71536815e-01
2.31094174e-02 -3.19271922e-01 3.03877115e-01 3.59112799e-01
-9.77235019e-01 3.23626786e-01 1.16242206e+00 -2.27346525e-01
-3.77503872e-01 5.49081564e-01 -1.24355137e+00 5.98079681e-01
7.92782843e-01 -8.48294050e-02 -4.66483384e-01 -1.36287659e-01
4.43734765e-01 2.88725853e-01 -3.19419801e-01 -1.09364283e+00
4.09571588e-01 -2.28450373e-01 1.71347886e-01 -9.89854038e-01
7.62198195e-02 -7.76895165e-01 -4.54195179e-02 4.50082034e-01
-8.29936922e-01 1.68014050e-01 -8.17530155e-02 9.66386616e-01
3.12586099e-01 -2.40646213e-01 -5.74151762e-02 5.38044199e-02
-1.23761559e+00 3.43194783e-01 -9.41912472e-01 -6.22100055e-01
6.92466378e-01 -6.20356984e-02 -5.24057806e-01 -5.12558818e-01
-8.76024604e-01 3.78123485e-02 1.07849076e-01 6.92589581e-01
4.95502442e-01 -8.49806309e-01 -3.05382907e-01 4.28627878e-02
3.21515769e-01 -5.93865991e-01 -1.15038998e-01 5.18813133e-01
-2.87339449e-01 7.87006497e-01 -2.72067338e-01 -2.04525180e-02
-1.01204848e+00 9.78600979e-01 1.75252408e-01 -8.61631215e-01
-5.01852572e-01 4.23012346e-01 -2.31519684e-01 -5.17285705e-01
3.58038872e-01 -6.34146869e-01 -4.73888606e-01 -2.12865859e-01
6.50585949e-01 6.29305482e-01 2.94272244e-01 2.88119674e-01
1.23527616e-01 -3.34671885e-01 -5.66079319e-01 9.28756595e-02
1.32699347e+00 1.63708940e-01 -6.05495125e-02 5.40582299e-01
2.51354605e-01 -1.59076780e-01 -1.11530066e+00 -4.49274719e-01
5.85999429e-01 2.28972770e-02 3.73848468e-01 -1.14949596e+00
-9.37699914e-01 7.89315581e-01 5.75909495e-01 -1.97686136e-01
9.32282627e-01 -3.71161312e-01 7.58377194e-01 1.17569816e+00
2.55698383e-01 -1.38162994e+00 3.79872710e-01 9.31054592e-01
3.94007862e-01 -7.00383306e-01 -1.84084531e-02 -2.47800812e-01
-7.34313607e-01 1.34615910e+00 8.45876038e-01 2.24430766e-02
8.48472059e-01 4.50427681e-01 -1.55713439e-01 -4.78167785e-03
-1.56017733e+00 -2.18172260e-02 1.05960034e-01 8.14051569e-01
6.27801418e-01 2.08438709e-01 -7.25558028e-02 1.21118987e+00
1.47339985e-01 3.56788725e-01 6.26358628e-01 1.17697370e+00
-5.67715704e-01 -1.31954217e+00 -1.48207471e-01 1.22791183e+00
-6.13363504e-01 -3.59993249e-01 -5.12067564e-02 3.86432976e-01
-1.71951130e-01 1.07184827e+00 3.97105217e-01 -8.96031618e-01
4.21050429e-01 4.97721225e-01 3.53025407e-01 -7.79172838e-01
-1.12783325e+00 4.93445285e-02 5.08074462e-01 -7.07195818e-01
-4.94712107e-02 -5.85494757e-01 -1.12068534e+00 -7.83202291e-01
-1.43468797e-01 -2.20156744e-01 5.25911868e-01 8.57323766e-01
6.61528945e-01 1.21638703e+00 3.69228691e-01 -4.72185075e-01
-1.10011089e+00 -1.11723530e+00 -4.12954420e-01 3.81970368e-02
-3.64060551e-01 -3.96765053e-01 4.46806997e-01 -1.25282660e-01] | [8.613099098205566, 5.933106422424316] |
2dc41d06-04e5-4fba-be94-09c6db651bb5 | openldn-learning-to-discover-novel-classes | 2207.02261 | null | https://arxiv.org/abs/2207.02261v2 | https://arxiv.org/pdf/2207.02261v2.pdf | OpenLDN: Learning to Discover Novel Classes for Open-World Semi-Supervised Learning | Semi-supervised learning (SSL) is one of the dominant approaches to address the annotation bottleneck of supervised learning. Recent SSL methods can effectively leverage a large repository of unlabeled data to improve performance while relying on a small set of labeled data. One common assumption in most SSL methods is that the labeled and unlabeled data are from the same data distribution. However, this is hardly the case in many real-world scenarios, which limits their applicability. In this work, instead, we attempt to solve the challenging open-world SSL problem that does not make such an assumption. In the open-world SSL problem, the objective is to recognize samples of known classes, and simultaneously detect and cluster samples belonging to novel classes present in unlabeled data. This work introduces OpenLDN that utilizes a pairwise similarity loss to discover novel classes. Using a bi-level optimization rule this pairwise similarity loss exploits the information available in the labeled set to implicitly cluster novel class samples, while simultaneously recognizing samples from known classes. After discovering novel classes, OpenLDN transforms the open-world SSL problem into a standard SSL problem to achieve additional performance gains using existing SSL methods. Our extensive experiments demonstrate that OpenLDN outperforms the current state-of-the-art methods on multiple popular classification benchmarks while providing a better accuracy/training time trade-off. | ['Mubarak Shah', 'Fahad Shahbaz Khan', 'Salman Khan', 'Navid Kardan', 'Mamshad Nayeem Rizve'] | 2022-07-05 | null | null | null | null | ['open-world-semi-supervised-learning'] | ['computer-vision'] | [ 2.94987738e-01 1.62275031e-01 -7.33104706e-01 -5.58924854e-01
-1.07626700e+00 -7.84972668e-01 3.24136555e-01 3.71626168e-01
-3.99274647e-01 8.39156985e-01 -2.44019002e-01 -6.38238620e-04
-1.52842104e-02 -4.70455468e-01 -6.30508900e-01 -8.27355683e-01
2.76062727e-01 6.63407385e-01 1.83299929e-01 3.51785541e-01
7.65344948e-02 4.35613632e-01 -1.70342386e+00 2.13208824e-01
8.61090481e-01 1.00805032e+00 7.35230744e-02 6.41176626e-02
-3.87213975e-01 5.73490143e-01 -2.82381892e-01 -1.48181275e-01
5.58399856e-01 -3.78008455e-01 -1.09343469e+00 3.81826758e-01
6.88854992e-01 2.49402612e-01 2.26489231e-02 1.18683064e+00
4.10589457e-01 2.88605571e-01 6.88628197e-01 -1.42568135e+00
-3.66972297e-01 5.58677316e-01 -8.16794574e-01 3.89295258e-02
7.25080445e-02 -7.60233775e-02 1.15044653e+00 -1.12897956e+00
6.80937350e-01 1.05412471e+00 7.74797440e-01 4.85040277e-01
-1.49311674e+00 -7.54466534e-01 4.71204430e-01 1.88130945e-01
-1.64700413e+00 -4.28845406e-01 6.71216488e-01 -2.14354381e-01
5.09235561e-01 1.31943241e-01 2.36296237e-01 7.95500278e-01
-4.01678860e-01 1.20288360e+00 1.26384866e+00 -5.70704997e-01
4.21817750e-01 3.40718389e-01 6.73447788e-01 6.46737039e-01
3.44334275e-01 -1.24993607e-01 -6.24663651e-01 -2.44542345e-01
1.79815918e-01 2.91696429e-01 -2.17121348e-01 -8.40019822e-01
-1.13704205e+00 8.37950528e-01 3.28642726e-01 1.60813555e-01
2.86250152e-02 -5.53337216e-01 5.65057695e-01 2.97584087e-01
6.90397799e-01 4.71460670e-01 -7.81934977e-01 1.94210142e-01
-1.09267557e+00 2.49708313e-02 8.57158184e-01 9.70373273e-01
1.14469397e+00 -4.44268912e-01 -6.17544502e-02 1.07017028e+00
3.84555757e-01 3.62881124e-01 5.80925345e-01 -7.37629473e-01
4.34738427e-01 1.01571894e+00 -1.00525409e-01 -5.45359790e-01
-2.59702593e-01 -5.91521561e-01 -5.84609509e-01 -6.81243911e-02
5.13549984e-01 1.38804689e-01 -9.14414346e-01 1.60627210e+00
6.79165483e-01 4.83863384e-01 1.59815267e-01 7.23610222e-01
6.82586432e-01 5.33524811e-01 -2.53664911e-01 -5.47009170e-01
8.63727331e-01 -1.24192441e+00 -5.48522830e-01 -2.98895448e-01
9.53706861e-01 -6.06987715e-01 1.11755276e+00 4.05144155e-01
-5.26954710e-01 -4.23074067e-01 -1.01528597e+00 2.59776227e-02
-4.62373286e-01 1.64641261e-01 5.38797617e-01 5.47342777e-01
-4.99177605e-01 4.01307523e-01 -6.55234814e-01 -3.40337604e-01
6.96839929e-01 3.19707334e-01 -3.94422442e-01 -5.02872109e-01
-7.97533691e-01 4.38721389e-01 6.73995912e-01 5.48328552e-03
-7.68159509e-01 -7.53380299e-01 -9.52332020e-01 -3.02553289e-02
9.28398967e-01 -7.94944540e-02 1.39485347e+00 -9.66156483e-01
-9.64430332e-01 1.19567180e+00 -2.30197772e-01 -3.39212686e-01
3.24732572e-01 -4.09919992e-02 -4.83036995e-01 -3.33168842e-02
5.04194498e-01 5.42085528e-01 6.22773886e-01 -1.35027027e+00
-8.39881420e-01 -5.99697769e-01 -1.56126365e-01 1.14344947e-01
-4.40242708e-01 -3.37696105e-01 -3.80266935e-01 -7.20493972e-01
3.26004714e-01 -1.07036054e+00 -2.05479279e-01 1.47194698e-01
-4.79603767e-01 -5.65002620e-01 1.12432134e+00 6.65075257e-02
1.11381996e+00 -2.20294380e+00 -1.59115478e-01 2.48750567e-01
4.08805132e-01 6.91730201e-01 -1.13065705e-01 2.81376123e-01
-5.03505655e-02 -2.00764284e-01 -4.39056456e-01 -3.79507512e-01
6.83943974e-03 4.15531904e-01 -3.77112329e-01 5.91620505e-01
5.66053428e-02 7.28583515e-01 -1.20901227e+00 -6.38846457e-01
-3.87452979e-04 -5.14930747e-02 -2.96566814e-01 2.12575659e-01
-4.53524023e-01 3.82841736e-01 -3.02627861e-01 7.53335714e-01
7.47141540e-01 -6.23858273e-01 2.47600392e-01 -1.22760743e-01
1.44710958e-01 2.35431585e-02 -1.46175265e+00 1.76463437e+00
-3.26484650e-01 3.99210393e-01 -1.48447350e-01 -1.41611993e+00
9.12299991e-01 1.93216741e-01 5.35881817e-01 -2.51400113e-01
-1.81800462e-02 5.60461402e-01 -1.68753058e-01 -4.01098490e-01
-2.52246261e-02 -3.04894954e-01 -5.13465889e-02 7.27988005e-01
3.96638632e-01 2.48651311e-01 3.89252126e-01 2.92762756e-01
8.98533642e-01 -9.16463044e-03 5.02139390e-01 -2.08521426e-01
4.56192464e-01 1.47986058e-02 9.88189936e-01 8.72433960e-01
-2.64410824e-01 5.10697961e-01 2.37717345e-01 -5.13429046e-01
-6.54964209e-01 -1.19414651e+00 -3.21254969e-01 1.08599782e+00
2.23497167e-01 -2.84725398e-01 -5.40393472e-01 -1.45678818e+00
1.04965851e-01 4.77095991e-01 -4.87487227e-01 -2.30737597e-01
-2.90717129e-02 -7.00675845e-01 4.14724857e-01 4.21054333e-01
3.73568475e-01 -7.39260316e-01 -1.24422804e-01 1.18293114e-01
-1.72971889e-01 -1.10463142e+00 -5.53962708e-01 4.62552160e-01
-8.58608007e-01 -1.36497939e+00 -4.78176117e-01 -1.14511168e+00
9.99277592e-01 5.06508887e-01 1.06115329e+00 -1.60805285e-01
-4.50989544e-01 9.63020846e-02 -5.52756310e-01 -3.37627918e-01
-2.89153397e-01 2.74601579e-01 2.63956696e-01 4.02241468e-01
7.60014951e-01 -2.42117569e-01 -2.03726515e-01 6.30744457e-01
-9.54467535e-01 -2.96284799e-02 4.84083503e-01 1.06643605e+00
9.55076873e-01 2.35410810e-01 8.45438123e-01 -1.38975883e+00
-6.09751306e-02 -6.28942251e-01 -4.55653816e-01 5.14383376e-01
-9.05818462e-01 1.42196566e-01 7.54296839e-01 -5.76847970e-01
-8.97047997e-01 3.27075839e-01 2.11208627e-01 -6.46633506e-01
-2.66464680e-01 6.32844150e-01 -4.55183208e-01 -2.27530450e-02
5.96705377e-01 2.03172922e-01 4.89011072e-02 -7.07110345e-01
2.56716967e-01 9.40552771e-01 4.24651861e-01 -6.04217649e-01
7.20211983e-01 5.68308532e-01 -1.05946474e-01 -6.38022542e-01
-1.84184849e+00 -9.90679681e-01 -1.00561190e+00 3.36954258e-02
2.67922729e-01 -1.01921487e+00 -3.40279400e-01 4.32792962e-01
-4.74524736e-01 -3.31322610e-01 -7.74094403e-01 4.39069629e-01
-3.33016902e-01 4.51945454e-01 -2.79133439e-01 -6.01305962e-01
-1.89632326e-02 -1.04468989e+00 1.05432105e+00 2.05659688e-01
-4.68762815e-02 -1.02180052e+00 9.87420827e-02 6.11384273e-01
-1.17989823e-01 -2.52883732e-02 6.00024700e-01 -1.30159199e+00
-3.35060000e-01 -4.66547847e-01 -1.82466671e-01 4.83890355e-01
4.17770326e-01 -4.12370801e-01 -1.10376048e+00 -6.52544141e-01
-1.03851490e-01 -8.83261204e-01 9.49875891e-01 -1.26169343e-02
1.34876990e+00 -1.67480558e-01 -5.00136256e-01 4.98384088e-01
1.14948761e+00 -6.86800778e-02 2.64465325e-02 -3.34679596e-02
8.10715199e-01 6.22919619e-01 1.08449256e+00 4.35038775e-01
2.97521323e-01 5.01575708e-01 1.01017646e-01 -1.97248429e-01
1.27746731e-01 -3.51922154e-01 1.56550050e-01 7.85038590e-01
6.09010816e-01 -6.80134892e-02 -9.84879553e-01 5.22704601e-01
-2.04728055e+00 -7.66843915e-01 -2.01304071e-02 2.39784598e+00
1.01185274e+00 1.03619561e-01 3.93722765e-02 3.25833142e-01
8.19675446e-01 -9.89545602e-03 -1.02848434e+00 1.73645586e-01
-1.67557910e-01 7.83905014e-02 4.02147740e-01 1.58601567e-01
-1.44801128e+00 7.05925107e-01 5.80599451e+00 1.16001058e+00
-1.01635337e+00 7.72463456e-02 7.02290475e-01 -3.18530649e-01
2.44765375e-02 -2.16016844e-02 -1.27388096e+00 4.28037107e-01
7.42867947e-01 -1.35146111e-01 1.08049035e-01 8.09726954e-01
3.71385738e-02 -1.33664131e-01 -1.44265294e+00 1.14475012e+00
3.86478066e-01 -1.24446666e+00 -5.50705194e-02 -4.59754132e-02
1.06109488e+00 1.38371512e-01 -1.05826505e-01 5.16658604e-01
3.25654864e-01 -7.69983768e-01 4.78198558e-01 1.13907255e-01
8.64611804e-01 -6.46400988e-01 7.33493328e-01 6.77459478e-01
-1.29397714e+00 -1.65706888e-01 -2.03946874e-01 7.46996794e-03
-1.33545294e-01 8.97149086e-01 -7.73072541e-01 4.20467198e-01
6.21647954e-01 1.22825325e+00 -6.78262234e-01 1.18858433e+00
-1.46015927e-01 8.36772561e-01 -3.14271420e-01 3.43972385e-01
3.93967390e-01 -1.50804564e-01 2.26320341e-01 9.73083496e-01
-1.27321750e-01 -6.24599792e-02 9.40801322e-01 5.53444803e-01
-4.12326872e-01 1.84134930e-01 -6.10105634e-01 -8.75877291e-02
6.05740070e-01 1.12384188e+00 -9.49834466e-01 -4.01641369e-01
-6.62464142e-01 7.99214423e-01 5.58368921e-01 1.63053870e-01
-5.31246841e-01 -3.16823244e-01 4.22444820e-01 7.56188855e-03
2.38155484e-01 2.28687063e-01 -5.92159368e-02 -1.38588250e+00
2.36050293e-01 -9.32585895e-01 8.52810800e-01 -2.83108473e-01
-1.84658921e+00 1.50978029e-01 -1.12008169e-01 -1.47675157e+00
6.46239668e-02 -5.56751788e-01 -1.63746104e-01 5.72072625e-01
-1.57416439e+00 -1.26785791e+00 -4.47180355e-03 5.31100810e-01
8.79191518e-01 -3.08390170e-01 7.39805758e-01 3.74140233e-01
-8.14465582e-01 7.29240417e-01 6.20064080e-01 1.77939638e-01
1.02189851e+00 -1.26771820e+00 3.00417971e-02 8.31046522e-01
4.68017519e-01 5.10228813e-01 2.67000139e-01 -5.41584849e-01
-1.14439166e+00 -1.47017825e+00 8.34832489e-01 -2.75915146e-01
5.46968997e-01 -4.48795259e-01 -1.19566023e+00 7.17782438e-01
-4.12926227e-01 8.40003192e-01 1.19685137e+00 1.99823588e-01
-7.66092002e-01 -1.20050728e-01 -1.18519282e+00 2.15022653e-01
1.04259753e+00 -6.52504027e-01 -5.50717831e-01 4.49015856e-01
6.14087582e-01 -2.48237997e-01 -7.86971509e-01 3.68240416e-01
2.52214521e-01 -5.60339808e-01 8.34991455e-01 -7.18664289e-01
1.29543275e-01 -5.96238852e-01 -1.86733887e-01 -1.34136617e+00
1.05366960e-01 -3.80468011e-01 -1.94425493e-01 1.31972027e+00
4.64830071e-01 -7.60389149e-01 9.36798275e-01 6.13024116e-01
-6.27644807e-02 -8.16581607e-01 -8.36922348e-01 -1.17402875e+00
-2.20368952e-02 -2.14982852e-01 3.11915636e-01 1.33431053e+00
3.51677574e-02 4.17634785e-01 -1.33922696e-01 1.17859110e-01
9.90607738e-01 6.64437234e-01 6.58023119e-01 -1.63007748e+00
-2.49841690e-01 -1.36254147e-01 -4.03877378e-01 -1.02126408e+00
7.24773347e-01 -1.37775373e+00 1.86635420e-01 -1.17480111e+00
6.12526417e-01 -1.00762558e+00 -4.51584816e-01 8.05023372e-01
-3.75968218e-01 4.20632094e-01 1.95933014e-01 3.70657057e-01
-1.19141114e+00 5.48181057e-01 8.54422867e-01 -2.38264591e-01
-1.97704822e-01 1.58243492e-01 -7.94619679e-01 8.94458652e-01
6.18199468e-01 -6.86527014e-01 -5.62533855e-01 -2.04657853e-01
-1.19124442e-01 -2.81883955e-01 1.48558810e-01 -8.87157083e-01
3.06389570e-01 -2.52498180e-01 2.03416958e-01 -6.71149015e-01
1.81690902e-02 -8.76868427e-01 -1.65119007e-01 2.90241450e-01
-6.74699247e-01 -6.06615603e-01 -2.13215593e-02 8.93236518e-01
-3.91528368e-01 -2.83570111e-01 1.02025127e+00 7.15037882e-02
-8.42915773e-01 5.12272775e-01 -8.06413889e-02 5.80266416e-01
1.29003751e+00 -1.52760252e-01 -1.19702116e-01 1.85068071e-01
-7.59635687e-01 5.39701700e-01 6.07911706e-01 3.90014827e-01
3.07649374e-01 -1.19827199e+00 -5.23869038e-01 2.68162489e-01
7.22750366e-01 4.12558317e-01 3.02522462e-02 6.48378074e-01
-7.20704868e-02 2.51170576e-01 1.86855063e-01 -7.44259000e-01
-1.29918742e+00 8.10271680e-01 2.01740623e-01 -4.41285431e-01
-4.50898945e-01 8.71987700e-01 2.67968118e-01 -1.12410438e+00
5.44752598e-01 1.52418956e-01 -1.97066255e-02 1.36942968e-01
5.64943433e-01 2.89143980e-01 2.86672831e-01 -5.09933412e-01
-3.34561586e-01 3.38209152e-01 -4.01809692e-01 3.65747720e-01
1.35997641e+00 -2.08962470e-01 -5.29989228e-02 8.73997748e-01
1.50193369e+00 -2.33201295e-01 -1.20494795e+00 -1.04825807e+00
2.73874968e-01 -6.49820387e-01 -4.38169092e-02 -7.82725513e-01
-1.06017327e+00 8.23131382e-01 5.10952652e-01 -1.17124736e-01
1.02296877e+00 2.82333463e-01 7.42098987e-01 5.27729332e-01
5.42601407e-01 -1.25680697e+00 5.71766011e-02 5.02392054e-01
2.63247430e-01 -1.78055739e+00 6.06956072e-02 -6.58356607e-01
-5.17231524e-01 8.85870993e-01 7.64664114e-01 1.91261321e-01
7.65031338e-01 1.06130064e-01 1.31009802e-01 -1.46293506e-01
-6.61637306e-01 -2.50741780e-01 2.37181783e-01 3.47368568e-01
2.87419140e-01 5.25251701e-02 -4.56100851e-02 5.39974928e-01
3.56778741e-01 -4.95226160e-02 2.78324246e-01 1.16812336e+00
-3.66985857e-01 -1.28812265e+00 -3.68846864e-01 9.24835503e-01
-2.68894315e-01 1.24815159e-01 -3.88583690e-01 3.44237387e-01
2.04324633e-01 8.58083963e-01 -1.29550218e-03 -4.84379604e-02
1.75975710e-01 2.00039148e-01 2.01493546e-01 -1.08000410e+00
-3.18302512e-01 -1.31620076e-02 -2.27097690e-01 -3.96375179e-01
-7.03085184e-01 -8.01028430e-01 -1.31794715e+00 1.57861874e-01
-6.66005373e-01 3.19184840e-01 2.82615215e-01 1.14720738e+00
3.13416988e-01 3.83348241e-02 9.86674249e-01 -5.62195361e-01
-8.04926693e-01 -5.31082332e-01 -9.05489326e-01 5.84483206e-01
3.06416273e-01 -8.93289626e-01 -4.37772840e-01 1.84733361e-01] | [9.52745246887207, 3.4814610481262207] |
9228c627-2499-4553-b016-65005dcaf844 | recall-and-learn-a-memory-augmented-solver | 2109.13112 | null | https://arxiv.org/abs/2109.13112v1 | https://arxiv.org/pdf/2109.13112v1.pdf | Recall and Learn: A Memory-augmented Solver for Math Word Problems | In this article, we tackle the math word problem, namely, automatically answering a mathematical problem according to its textual description. Although recent methods have demonstrated their promising results, most of these methods are based on template-based generation scheme which results in limited generalization capability. To this end, we propose a novel human-like analogical learning method in a recall and learn manner. Our proposed framework is composed of modules of memory, representation, analogy, and reasoning, which are designed to make a new exercise by referring to the exercises learned in the past. Specifically, given a math word problem, the model first retrieves similar questions by a memory module and then encodes the unsolved problem and each retrieved question using a representation module. Moreover, to solve the problem in a way of analogy, an analogy module and a reasoning module with a copy mechanism are proposed to model the interrelationship between the problem and each retrieved question. Extensive experiments on two well-known datasets show the superiority of our proposed algorithm as compared to other state-of-the-art competitors from both overall performance comparison and micro-scope studies. | ['Ming Yang', 'Da Cao', 'Jiao Xu', 'Jiawei Wang', 'Shifeng Huang'] | 2021-09-27 | null | https://aclanthology.org/2021.findings-emnlp.68 | https://aclanthology.org/2021.findings-emnlp.68.pdf | findings-emnlp-2021-11 | ['math-word-problem-solving', 'math-word-problem-solving', 'math-word-problem-solving'] | ['knowledge-base', 'reasoning', 'time-series'] | [ 2.54059523e-01 -3.23174242e-03 8.55070353e-02 -2.19158292e-01
-5.88721156e-01 -3.69259268e-01 7.50390172e-01 5.11727393e-01
-2.86092311e-01 6.13862813e-01 9.38549638e-02 -2.41173789e-01
-5.15904009e-01 -1.35527742e+00 -6.14170849e-01 -2.16074154e-01
5.54649711e-01 4.28675264e-01 3.73310000e-01 -3.88301015e-01
9.05075371e-01 1.79400697e-01 -1.44059229e+00 2.41360590e-01
1.33719707e+00 9.10969198e-01 4.54484940e-01 1.13598928e-01
-5.18728912e-01 1.14277577e+00 -6.41909182e-01 -6.86609268e-01
-4.28749137e-02 -9.00992930e-01 -1.15134966e+00 -1.81941181e-01
2.69401044e-01 -7.75296539e-02 -4.93258476e-01 9.33583021e-01
3.22503984e-01 6.12528980e-01 7.10401654e-01 -7.05219090e-01
-1.12024355e+00 5.05449057e-01 -2.81200409e-01 3.35992694e-01
7.08333611e-01 -1.09271549e-01 8.56687009e-01 -8.50327134e-01
3.61723900e-01 1.07763338e+00 3.81674618e-01 5.07296741e-01
-9.48289990e-01 -5.57472348e-01 8.02915618e-02 8.16589057e-01
-1.27950668e+00 -4.36403081e-02 9.78333056e-01 -1.46961182e-01
7.48146355e-01 1.51727617e-01 7.53368378e-01 4.98471379e-01
2.77825356e-01 6.63188934e-01 1.17763698e+00 -4.68406767e-01
3.56114745e-01 4.36810195e-01 3.71311754e-01 5.36880672e-01
3.03509403e-02 -1.39608085e-01 -4.19710070e-01 3.11943591e-02
6.21021271e-01 1.42612115e-01 -3.02393794e-01 -1.55908391e-01
-8.83350253e-01 7.22072124e-01 7.65765548e-01 5.95843971e-01
-5.53177774e-01 -1.78583741e-01 9.53205600e-02 2.60943979e-01
-1.68942176e-02 7.16939330e-01 -1.92563966e-01 4.58321124e-01
-1.03678811e+00 4.38993663e-01 7.80781507e-01 7.94268906e-01
6.76583350e-01 -3.30207855e-01 -4.13741738e-01 5.49632668e-01
3.06107938e-01 9.19380412e-02 8.52876306e-01 -5.99158049e-01
6.38150871e-01 9.69014585e-01 -1.89173564e-01 -1.37015593e+00
-1.79808035e-01 -7.71353245e-01 -6.69140339e-01 -2.50140935e-01
1.42591432e-01 2.64119685e-01 -3.73690665e-01 1.81228650e+00
2.68837720e-01 6.43900871e-01 4.69129324e-01 7.92779207e-01
1.25301886e+00 8.41069520e-01 2.08353698e-01 -4.08317178e-01
1.31750476e+00 -1.16237819e+00 -7.52448261e-01 -3.52181911e-01
3.39194149e-01 -8.07738245e-01 9.52377677e-01 3.51085991e-01
-1.34189129e+00 -9.13074672e-01 -1.05908060e+00 -1.63329914e-01
-5.47461092e-01 -1.06653543e-02 3.47478837e-01 3.68324369e-01
-6.49499714e-01 6.55297101e-01 5.85457869e-02 -4.83889729e-01
2.14431182e-01 -1.37023941e-01 5.96865267e-02 -2.71938890e-01
-1.32010889e+00 1.13388658e+00 7.94879377e-01 6.31643012e-02
-4.65874106e-01 -7.06746042e-01 -6.44464314e-01 4.35354024e-01
5.55905104e-01 -9.88794565e-01 9.16510642e-01 -6.20305538e-01
-1.38598275e+00 6.94271624e-01 5.41412272e-02 -4.44866359e-01
1.65861752e-02 -1.73546627e-01 -4.62337196e-01 1.24075882e-01
-8.09209645e-02 3.37529689e-01 4.37792361e-01 -1.06771290e+00
-4.08411562e-01 -5.32982171e-01 4.71131235e-01 4.06322122e-01
-4.43079829e-01 -4.88560081e-01 -3.68549079e-01 -8.21271062e-01
4.77514356e-01 -5.94746947e-01 -9.80239511e-02 -2.75788128e-01
9.81580094e-02 -4.83477980e-01 3.51916611e-01 -8.25450838e-01
1.67634380e+00 -1.82757735e+00 4.63837117e-01 2.21001908e-01
7.13344738e-02 3.96803498e-01 -2.37729818e-01 6.05100870e-01
-4.95428406e-02 -6.03364334e-02 2.34472081e-02 3.27755094e-01
-7.52189010e-02 2.73616482e-02 -5.54340303e-01 -1.60962656e-01
-9.80983675e-02 8.91861558e-01 -1.06039917e+00 -6.57314956e-01
1.25692070e-01 -2.20050756e-02 -6.25565171e-01 5.80495298e-01
-4.46998537e-01 2.21707195e-01 -6.05157018e-01 3.54166359e-01
6.86120629e-01 -1.45082697e-01 3.35032582e-01 -3.25229555e-01
1.55105233e-01 4.18998033e-01 -1.25686526e+00 2.11259389e+00
-7.56974459e-01 2.91099120e-02 -6.09012544e-01 -1.24547148e+00
1.25526118e+00 1.51868433e-01 -2.92536050e-01 -9.66880500e-01
5.53275719e-02 2.12222591e-01 -1.98500846e-02 -7.59946287e-01
5.27047753e-01 -2.95086920e-01 2.23918915e-01 3.99864703e-01
1.87800899e-01 -2.99634248e-01 2.25285113e-01 3.85594875e-01
1.06750751e+00 4.05407220e-01 4.68359798e-01 -8.61256346e-02
1.20831692e+00 -1.18263632e-01 1.86888173e-01 7.11035907e-01
1.68312982e-01 1.22681655e-01 1.21902362e-01 -3.27686369e-01
-5.32287300e-01 -1.13714123e+00 1.64480433e-01 7.52460837e-01
4.39332128e-01 -5.28558791e-01 -6.15783274e-01 -4.43169415e-01
-3.34312707e-01 1.04767191e+00 -4.90744829e-01 -4.45818096e-01
-5.34449756e-01 -3.55557561e-01 2.25770578e-01 4.77016866e-01
1.03094518e+00 -1.26116145e+00 -6.22745395e-01 3.71956587e-01
-2.78357387e-01 -7.30795622e-01 -1.80209309e-01 -2.71092981e-01
-1.12146771e+00 -1.02317047e+00 -6.16565228e-01 -1.02394247e+00
5.08892357e-01 3.86715055e-01 1.17484140e+00 4.69622701e-01
-1.11618698e-01 3.22067738e-01 -3.92168730e-01 -1.19698562e-01
-2.30258241e-01 2.95661706e-02 -4.00524318e-01 -1.00118801e-01
4.23737466e-01 -7.77931571e-01 -6.45586133e-01 6.84597623e-03
-1.04526365e+00 2.44411975e-02 9.69181359e-01 8.22423160e-01
4.80839103e-01 2.88260758e-01 9.17394519e-01 -7.32139707e-01
9.94913697e-01 -8.11629415e-01 -4.01452363e-01 6.75054431e-01
-7.77702689e-01 2.88309544e-01 6.50595188e-01 -3.50703925e-01
-1.47538078e+00 -2.16242671e-01 5.88840209e-02 -2.65683770e-01
-5.39866351e-02 8.96992624e-01 -1.69996321e-01 1.53494596e-01
6.74495161e-01 9.44584191e-01 -2.29919314e-01 -4.36015517e-01
5.50048172e-01 4.11817014e-01 6.29453123e-01 -9.51162577e-01
8.44233632e-01 -8.00973475e-02 2.75576562e-02 -2.33253554e-01
-1.09111857e+00 -2.54768968e-01 -3.82521063e-01 -2.32427925e-01
6.70776904e-01 -6.09957695e-01 -9.03759956e-01 3.69966850e-02
-1.39999950e+00 3.85662347e-01 -2.00835526e-01 3.99486363e-01
-4.41933602e-01 3.50881457e-01 -3.92169297e-01 -8.29742491e-01
-4.28943366e-01 -7.38829076e-01 4.62658882e-01 7.89251804e-01
-2.02315822e-01 -9.23191786e-01 1.89951941e-01 6.58095658e-01
5.89128375e-01 -1.09951615e-01 1.51443148e+00 -8.96380186e-01
-7.51231611e-01 -5.73643073e-02 -2.52698064e-01 1.20996326e-01
-1.54767781e-02 -8.02378535e-01 -6.55015945e-01 7.08459392e-02
3.32267731e-01 -3.94503742e-01 7.41780937e-01 -2.60692805e-01
1.24891770e+00 -4.04368281e-01 -1.71617210e-01 -1.93566438e-02
1.57163060e+00 3.90115619e-01 9.86834049e-01 4.12944436e-01
1.02527902e-01 6.72243953e-01 7.14858234e-01 3.06660295e-01
3.88032109e-01 4.81226414e-01 2.04795063e-01 4.27028120e-01
-3.36209685e-02 -4.44193751e-01 -1.17832430e-01 8.74420166e-01
9.18244012e-03 -4.79987785e-02 -7.62534201e-01 5.02311766e-01
-1.86334801e+00 -1.19732094e+00 -6.03164099e-02 2.21850467e+00
1.03403664e+00 1.20134227e-01 -2.39451796e-01 2.75216252e-01
4.55986947e-01 8.19435045e-02 -3.46701384e-01 -3.31608713e-01
1.36250436e-01 6.76517427e-01 -3.63916576e-01 1.98883027e-01
-4.52667654e-01 9.57411766e-01 5.27309799e+00 1.21811044e+00
-5.98316848e-01 -8.23197439e-02 3.54098737e-01 1.80862874e-01
-3.79201025e-01 2.18958542e-01 -3.97895396e-01 1.40704855e-01
6.07734144e-01 -5.76019943e-01 6.20144486e-01 5.21773338e-01
-1.55500144e-01 -1.66673049e-01 -1.16534126e+00 9.43008125e-01
3.73379081e-01 -1.35386789e+00 8.37789059e-01 -4.98340696e-01
5.33637106e-01 -1.01793945e+00 6.62333295e-02 6.63622856e-01
-3.93759727e-01 -1.01524425e+00 4.85117793e-01 1.14370549e+00
-6.52997345e-02 -8.77126932e-01 7.72131264e-01 6.14201307e-01
-1.31933379e+00 -1.76580861e-01 -4.78660494e-01 -4.16310728e-01
-2.74839878e-01 1.74283952e-01 -3.41249317e-01 1.15176976e+00
4.26989615e-01 4.40719187e-01 -7.74186611e-01 1.07195044e+00
-5.79278767e-01 3.37714881e-01 2.78060228e-01 -2.51180679e-01
1.42590344e-01 -3.89154077e-01 2.25871459e-01 7.74721384e-01
4.75325465e-01 8.68645489e-01 -5.42458035e-02 1.33464015e+00
-1.27643719e-01 6.73452735e-01 -4.70201284e-01 2.80201491e-02
7.16903150e-01 1.21070313e+00 -4.48032290e-01 -5.69492579e-01
-3.23112726e-01 9.06261027e-01 5.61513424e-01 1.50405914e-01
-8.06977272e-01 -5.94821155e-01 -1.29918680e-01 1.55715212e-01
4.21822332e-02 1.28421992e-01 -2.30871171e-01 -1.02159655e+00
2.35015154e-01 -9.34006989e-01 5.82852960e-01 -1.14633679e+00
-1.34699035e+00 4.06380147e-01 6.62541715e-03 -1.20950663e+00
-1.14746317e-01 -3.34124476e-01 -1.04423809e+00 1.04658425e+00
-1.40064216e+00 -9.17545021e-01 -5.68240881e-01 5.71150541e-01
7.44588017e-01 -3.43257695e-01 9.07263637e-01 1.29851654e-01
-3.82748604e-01 3.96390140e-01 -3.01143944e-01 -1.31871790e-01
5.27772605e-01 -8.96099567e-01 -1.92007586e-01 6.35602176e-01
2.58803159e-01 9.70439374e-01 5.38016915e-01 -6.11096799e-01
-1.49788630e+00 -8.42918515e-01 1.06536973e+00 -1.01145379e-01
5.69878399e-01 2.26342678e-01 -1.23897600e+00 3.81072313e-01
5.16070426e-01 -4.03501779e-01 8.18031907e-01 -1.45645097e-01
-3.09702873e-01 -1.72378525e-01 -1.13231969e+00 7.41075575e-01
9.29358900e-01 -2.92362779e-01 -1.59502399e+00 3.12250972e-01
5.21993876e-01 -4.10508096e-01 -7.28278756e-01 1.83468938e-01
5.08662641e-01 -8.11903000e-01 9.87787068e-01 -7.48215139e-01
8.49861860e-01 -3.42680186e-01 -2.24443823e-01 -1.29590774e+00
-4.96063977e-01 -1.26430184e-01 -1.44624382e-01 1.41943884e+00
2.13415444e-01 -4.36442256e-01 5.61290503e-01 4.24212813e-01
9.24005210e-02 -9.05921578e-01 -6.79238498e-01 -8.47079158e-01
9.72269177e-02 -4.80983518e-02 8.11090589e-01 9.92048740e-01
3.01801592e-01 8.26312423e-01 -1.28599361e-01 1.04060069e-01
4.32777494e-01 5.95468998e-01 5.38799286e-01 -1.23563147e+00
-5.55015862e-01 -3.85274291e-01 -2.38287389e-01 -1.21189666e+00
3.10193390e-01 -1.26487947e+00 -2.59933710e-01 -1.66743505e+00
5.16622186e-01 -1.73641413e-01 -4.72815484e-01 1.19750217e-01
-4.25769061e-01 -1.35342032e-01 3.90622258e-01 6.42188489e-02
-6.26772404e-01 6.21705472e-01 1.21387053e+00 -2.91949958e-01
-2.60368615e-01 -1.65386930e-01 -8.76338482e-01 6.56436682e-01
8.16751361e-01 -4.77299303e-01 -6.31502688e-01 -4.58845913e-01
2.27396414e-01 5.36261141e-01 5.00048041e-01 -1.33667004e+00
6.62440419e-01 -2.11727679e-01 5.60511231e-01 -6.02006853e-01
2.20428631e-01 -7.76184678e-01 1.25266807e-02 6.84100628e-01
-6.37878776e-01 2.47135729e-01 2.69969612e-01 6.15503967e-01
-3.51592839e-01 -7.26696908e-01 6.18718326e-01 -4.18050855e-01
-6.41009271e-01 -8.60110000e-02 -1.02214240e-01 1.66694567e-01
9.67179716e-01 9.97255966e-02 -2.68419296e-01 -2.60190457e-01
-6.95315659e-01 1.94316477e-01 -1.38769969e-01 4.11178261e-01
9.03684199e-01 -1.57804716e+00 -6.53856397e-01 -1.93096384e-01
1.66872770e-01 -2.94055611e-01 4.69760060e-01 5.20330071e-01
-2.91611433e-01 4.98184681e-01 -3.32454711e-01 -3.29428054e-02
-1.04665995e+00 1.01569963e+00 2.78920203e-01 -6.23470843e-01
-1.68581888e-01 4.89980310e-01 1.08926058e-01 -3.18684191e-01
2.13560328e-01 -6.40472919e-02 -7.54124761e-01 2.82864831e-02
6.22725964e-01 4.14527029e-01 -4.38365079e-02 -2.05799878e-01
-2.50674993e-01 7.45604157e-01 -1.02120548e-01 -9.94853750e-02
1.07335544e+00 1.35171607e-01 -2.50960082e-01 3.09824914e-01
7.56035566e-01 -1.57022893e-01 -1.44787207e-01 -6.93002760e-01
5.30710146e-02 -4.40157741e-01 -1.77803233e-01 -1.00531518e+00
-8.81597221e-01 1.01028204e+00 5.36735654e-01 -8.75111371e-02
1.31854546e+00 -3.96274440e-02 8.39067400e-01 7.71566153e-01
2.59483546e-01 -9.76893544e-01 5.44416666e-01 4.26457971e-01
1.13275957e+00 -8.06618452e-01 1.18671708e-01 -4.53932762e-01
-1.01960950e-01 1.30359173e+00 9.83413875e-01 -1.91849187e-01
3.85154903e-01 -4.75901723e-01 -3.68547857e-01 -2.12523282e-01
-7.19501853e-01 -2.70172805e-02 4.89388406e-01 1.44347638e-01
5.58664858e-01 -3.66129637e-01 -1.06478298e+00 1.16081357e+00
-1.45550340e-01 3.63397539e-01 3.05838197e-01 9.05760884e-01
-7.05128789e-01 -1.10878229e+00 -3.44192594e-01 2.37140357e-01
4.72735008e-03 -3.11778396e-01 -3.40836257e-01 6.30862296e-01
2.54438549e-01 1.06817412e+00 -1.90910891e-01 -3.68458360e-01
5.95289588e-01 3.34637195e-01 7.52346277e-01 -6.65680766e-01
-8.46537352e-01 -4.71489906e-01 -3.04871112e-01 -3.17405730e-01
-3.43333542e-01 -2.16616154e-01 -1.40328252e+00 -1.57303065e-01
-2.24940524e-01 5.15439451e-01 2.88419336e-01 1.30559659e+00
1.63942352e-01 7.45923400e-01 4.51619983e-01 -2.88362294e-01
-8.06722343e-01 -9.87628877e-01 -3.72726023e-01 5.30964673e-01
-4.99410689e-01 -5.42220533e-01 -4.67610806e-02 -1.08349204e-01] | [10.755148887634277, 8.050178527832031] |
d3851193-e3a6-455e-b9ef-e8366926520b | implicit-neural-deformation-for-multi-view | 2112.02494 | null | https://arxiv.org/abs/2112.02494v2 | https://arxiv.org/pdf/2112.02494v2.pdf | Implicit Neural Deformation for Sparse-View Face Reconstruction | In this work, we present a new method for 3D face reconstruction from sparse-view RGB images. Unlike previous methods which are built upon 3D morphable models (3DMMs) with limited details, we leverage an implicit representation to encode rich geometric features. Our overall pipeline consists of two major components, including a geometry network, which learns a deformable neural signed distance function (SDF) as the 3D face representation, and a rendering network, which learns to render on-surface points of the neural SDF to match the input images via self-supervised optimization. To handle in-the-wild sparse-view input of the same target with different expressions at test time, we propose residual latent code to effectively expand the shape space of the learned implicit face representation as well as a novel view-switch loss to enforce consistency among different views. Our experimental results on several benchmark datasets demonstrate that our approach outperforms alternative baselines and achieves superior face reconstruction results compared to state-of-the-art methods. | ['Chongyang Ma', 'Nong Sang', 'Mengtian Li', 'Yi Zheng', 'Haibin Huang', 'Moran Li'] | 2021-12-05 | null | null | null | null | ['3d-face-reconstruction', 'face-reconstruction'] | ['computer-vision', 'computer-vision'] | [ 1.49630249e-01 2.33790115e-01 -5.16049610e-03 -7.90060580e-01
-7.06037104e-01 -4.36422259e-01 4.34214413e-01 -8.15898776e-01
3.16792130e-01 2.02042297e-01 1.94145203e-01 1.29261002e-01
3.46067518e-01 -7.50063896e-01 -9.69018519e-01 -4.00771052e-01
1.38195664e-01 6.69761896e-01 -2.40203664e-01 6.71886094e-03
1.29356161e-01 9.48155284e-01 -1.60892820e+00 4.41643655e-01
3.78985822e-01 1.23207664e+00 -3.40404272e-01 1.93993539e-01
-1.26924127e-01 6.07594609e-01 1.10261492e-01 -5.05665898e-01
6.80366933e-01 -2.32400373e-01 -6.61857724e-01 5.08600831e-01
1.24979079e+00 -7.32797623e-01 -4.56028312e-01 8.00886929e-01
5.67184687e-01 5.19441627e-02 7.01153815e-01 -1.10155523e+00
-7.99661100e-01 -3.46672088e-01 -7.97433615e-01 -4.46174443e-01
7.64729261e-01 8.78754854e-02 6.67981207e-01 -1.45865750e+00
9.59860742e-01 1.75076222e+00 8.21285844e-01 9.39764261e-01
-1.48531723e+00 -7.63447881e-01 2.47991666e-01 -3.05820942e-01
-1.37872267e+00 -9.59982812e-01 1.06589830e+00 -5.22559464e-01
1.03299463e+00 -4.59560044e-02 7.38241434e-01 1.09099579e+00
-6.11961819e-02 4.43423569e-01 1.15184391e+00 -1.57681927e-01
1.97340874e-03 -2.27073938e-01 -5.94042897e-01 1.38653719e+00
-2.06689358e-01 2.53860503e-01 -8.75945807e-01 -2.77424872e-01
1.39013112e+00 1.10334083e-01 -2.75687337e-01 -1.06092083e+00
-5.64854741e-01 7.81277537e-01 4.01533872e-01 -3.25372189e-01
-1.64144546e-01 2.72245914e-01 -6.17725030e-02 1.14249527e-01
8.33976686e-01 -2.48483539e-01 -4.80015874e-01 2.41813719e-01
-9.49373543e-01 3.74008179e-01 8.19388688e-01 8.05861831e-01
1.05814123e+00 2.87458241e-01 1.46450549e-01 7.67095268e-01
7.59775341e-01 6.97390258e-01 5.56513807e-03 -1.30831695e+00
4.08608437e-01 7.57250845e-01 -3.04555833e-01 -9.99315917e-01
-3.03414073e-02 1.15429889e-02 -6.01993501e-01 5.29930770e-01
1.81358028e-02 3.51758391e-01 -1.11955392e+00 1.71974421e+00
7.15181530e-01 6.14782989e-01 -3.20512593e-01 8.64456415e-01
1.06887257e+00 3.21744382e-01 -4.89656955e-01 3.01728882e-02
9.15079474e-01 -8.52949679e-01 -4.34830993e-01 -1.91141039e-01
1.20450437e-01 -6.19697332e-01 9.00227785e-01 1.48103118e-01
-1.52047384e+00 -3.26701343e-01 -7.76320577e-01 -4.74345326e-01
2.80177668e-02 -1.03760779e-01 5.90582490e-01 5.15690565e-01
-1.40000677e+00 6.53978765e-01 -9.92894828e-01 -1.30477205e-01
9.08630431e-01 4.86439288e-01 -9.11561191e-01 -3.75381708e-01
-3.73696685e-01 4.57386494e-01 -4.36792225e-01 4.86005098e-02
-1.21469772e+00 -1.02702761e+00 -1.34880590e+00 -2.32758194e-01
8.77784193e-02 -1.08083117e+00 8.14865470e-01 -9.09306347e-01
-1.91310704e+00 1.67199075e+00 -3.48094761e-01 4.04605746e-01
4.42863345e-01 -7.15773478e-02 9.17424336e-02 2.94252813e-01
1.48278885e-02 6.56181514e-01 1.25323355e+00 -1.69845366e+00
9.70050022e-02 -8.97599518e-01 -6.09208420e-02 2.13039756e-01
1.03301674e-01 -1.59201488e-01 -8.38559330e-01 -6.70404196e-01
5.00582457e-01 -9.18986082e-01 1.05711386e-01 7.92242944e-01
-2.08372027e-01 1.95044160e-01 9.39451039e-01 -7.07930684e-01
4.51781452e-01 -2.14391851e+00 6.97294652e-01 3.23551118e-01
2.57445544e-01 -1.64459348e-01 -3.39418203e-01 -5.57732508e-02
-2.20733851e-01 -4.80043553e-02 -3.72063935e-01 -1.18359363e+00
-3.86934094e-02 3.32859963e-01 -3.41690361e-01 7.84725726e-01
4.08011198e-01 9.34268296e-01 -5.87232113e-01 -2.91150987e-01
1.48105919e-01 1.04721951e+00 -1.00774646e+00 7.40684927e-01
-1.47607341e-01 7.37145483e-01 -3.68898749e-01 9.84224498e-01
9.98177826e-01 -3.42662990e-01 4.90928665e-02 -3.75264525e-01
2.69906253e-01 2.13981777e-01 -1.07765472e+00 2.53236675e+00
-4.05068219e-01 1.44027686e-02 5.09307504e-01 -8.06522250e-01
8.73792529e-01 2.10454375e-01 5.85280538e-01 -5.29791534e-01
7.40904212e-02 3.95508334e-02 -8.21783125e-01 -2.92646468e-01
-6.23637810e-02 -2.05509499e-01 5.33250630e-01 5.30927300e-01
2.88636029e-01 -5.32150149e-01 -6.53308153e-01 9.54878703e-02
8.79747808e-01 8.91588330e-01 -2.02254161e-01 -2.30899584e-02
4.95841861e-01 -7.14074552e-01 6.67557299e-01 -1.59130335e-01
1.83799624e-01 1.03997779e+00 4.38367635e-01 -7.07098961e-01
-7.66948700e-01 -1.35981488e+00 -1.42722279e-01 6.51712477e-01
1.55726885e-02 -4.46293414e-01 -7.68881679e-01 -8.92077684e-01
3.62254649e-01 1.03306636e-01 -9.24528062e-01 9.89345759e-02
-5.86088061e-01 -9.33333114e-02 1.96740776e-01 5.01323581e-01
3.65587503e-01 -6.93321168e-01 -3.27983171e-01 -3.64453316e-01
5.88874817e-02 -1.24358428e+00 -6.36496961e-01 -2.27749035e-01
-1.08629751e+00 -1.14987671e+00 -4.67185110e-01 -7.29211688e-01
1.21884620e+00 3.02055925e-01 1.22304249e+00 3.03813308e-01
-4.14219975e-01 8.37592006e-01 5.72352000e-02 5.02267368e-02
-2.99822856e-02 -4.28373277e-01 6.67160675e-02 2.62380332e-01
-1.47068994e-02 -1.03374052e+00 -7.17384219e-01 2.51061141e-01
-7.69983828e-01 1.11560427e-01 1.00189470e-01 7.38985956e-01
1.02707088e+00 -7.03833163e-01 -5.17971702e-02 -9.45980668e-01
-1.89398855e-01 -2.80277342e-01 -6.45444512e-01 2.44585127e-01
-3.77638042e-01 1.02696933e-01 2.03458667e-01 -1.66758940e-01
-9.63128507e-01 4.30498391e-01 -1.21176496e-01 -1.16391027e+00
2.07978711e-02 -4.61308286e-02 -6.27334058e-01 -5.82406878e-01
2.23477826e-01 1.75637156e-01 3.97006482e-01 -6.31784081e-01
4.45757627e-01 1.27848670e-01 3.41614038e-01 -7.84522891e-01
1.09492850e+00 9.67869580e-01 2.67446429e-01 -5.30713379e-01
-8.04637313e-01 -2.52469555e-02 -9.64608908e-01 -1.78682938e-01
7.93368638e-01 -1.22953594e+00 -6.00593090e-01 4.57577348e-01
-1.20944905e+00 -5.03370702e-01 -2.63423026e-01 5.03103621e-02
-9.34779644e-01 2.27945745e-01 -6.31555855e-01 -5.67642152e-01
-3.02209258e-01 -1.22070694e+00 1.86030257e+00 -4.62026969e-02
9.46823210e-02 -8.95085692e-01 1.66686416e-01 6.10089064e-01
2.16730922e-01 6.58401549e-01 7.57481754e-01 1.52554646e-01
-8.93528759e-01 1.00360930e-01 -1.23147182e-01 2.77850062e-01
2.41038650e-01 -7.37764910e-02 -1.26668084e+00 -5.75703442e-01
9.04440582e-02 -6.56815410e-01 6.79180324e-01 2.80862153e-01
1.37086642e+00 -3.86311769e-01 -2.66174346e-01 1.72442567e+00
1.33347261e+00 -2.65510529e-01 6.66090012e-01 -2.39136726e-01
1.15666533e+00 7.40277767e-01 4.20733280e-02 4.48395908e-01
6.01139486e-01 8.17506790e-01 8.42651129e-01 -1.36401400e-01
-2.90776402e-01 -6.07009709e-01 5.54643512e-01 7.30691850e-01
-2.56819665e-01 2.81514108e-01 -4.76167411e-01 5.79697415e-02
-1.44689941e+00 -8.22336257e-01 4.62084234e-01 2.18569112e+00
6.69496596e-01 -4.30986971e-01 -9.54306871e-02 -1.89495966e-01
2.22558945e-01 3.53277057e-01 -7.60703564e-01 -2.86124587e-01
-1.87237021e-02 5.98751962e-01 -8.65002498e-02 7.07694829e-01
-7.92767584e-01 1.08009160e+00 6.08449650e+00 3.56358200e-01
-1.22593296e+00 1.96013585e-01 6.87897384e-01 -3.62935781e-01
-6.45754933e-01 -5.95530681e-02 -6.30451620e-01 5.17143719e-02
2.74873048e-01 3.35402995e-01 7.86427617e-01 6.90983951e-01
-1.00405321e-01 3.97056848e-01 -1.35594475e+00 1.43156672e+00
6.90012276e-01 -1.29000866e+00 3.34174633e-01 3.73049766e-01
9.63071227e-01 -1.77615806e-01 3.48997265e-01 -1.43357560e-01
1.26706600e-01 -1.39624965e+00 6.80131078e-01 7.39116967e-01
1.34130621e+00 -6.75969362e-01 2.47786604e-02 -3.31096388e-02
-1.25419486e+00 4.54867274e-01 -3.44855100e-01 1.84526637e-01
3.70114557e-02 2.98192412e-01 -2.89035827e-01 5.72495162e-01
8.25469077e-01 1.17556655e+00 -2.92054325e-01 2.57102132e-01
-2.96103239e-01 1.03429973e-01 -1.62364036e-01 8.64810228e-01
-2.51985222e-01 -4.24172401e-01 3.89149100e-01 5.89223683e-01
3.78425866e-01 3.13869864e-01 1.72461584e-01 1.07930553e+00
-4.77806985e-01 7.01018050e-02 -1.00155199e+00 1.94769502e-01
2.11324319e-01 1.32714081e+00 -4.20946658e-01 5.27469367e-02
-5.71349025e-01 1.39791942e+00 7.72786617e-01 3.85740876e-01
-5.91433883e-01 3.98377955e-01 9.89470065e-01 4.80717629e-01
4.94122148e-01 -3.32067788e-01 -1.98118128e-02 -1.43115759e+00
1.71844795e-01 -8.01777303e-01 2.05109015e-01 -8.63199711e-01
-1.43375576e+00 6.64588809e-01 -3.01035285e-01 -1.15248895e+00
-2.15213537e-01 -7.66954839e-01 -3.74890894e-01 9.94242549e-01
-1.70832443e+00 -1.73730969e+00 -5.16853869e-01 9.13826346e-01
3.52669448e-01 -3.43542844e-01 1.04201496e+00 2.07778439e-01
-3.87455583e-01 7.35633135e-01 -3.54003012e-01 1.42779320e-01
7.81635523e-01 -8.84494126e-01 4.60445374e-01 4.25295472e-01
2.70253837e-01 7.05617666e-01 -1.06855147e-01 -6.06940746e-01
-2.09834647e+00 -1.05248117e+00 3.56627852e-01 -7.67803907e-01
4.17351387e-02 -5.06878912e-01 -8.31676066e-01 9.81679678e-01
-5.96887805e-02 7.76893973e-01 8.32940698e-01 -8.56109187e-02
-1.04778934e+00 -1.17409661e-01 -1.42056918e+00 3.00688893e-01
1.70680690e+00 -8.68764043e-01 -2.65766352e-01 1.58283174e-01
4.20162112e-01 -9.04180229e-01 -8.41526508e-01 4.83484209e-01
8.81160438e-01 -1.28894639e+00 1.19621086e+00 -7.22137392e-01
5.67056358e-01 -1.98269099e-01 -5.03740609e-01 -1.23670816e+00
-2.14477688e-01 -7.19489515e-01 -4.98911619e-01 1.07741559e+00
-1.46538377e-01 -3.86291176e-01 1.00647795e+00 7.16305435e-01
-1.10525317e-01 -1.10938609e+00 -1.04622018e+00 -4.66578126e-01
1.13000870e-02 -2.29551151e-01 7.96037316e-01 9.91302490e-01
-5.88972807e-01 1.30715594e-01 -2.49495402e-01 2.49389097e-01
1.00656474e+00 2.82192767e-01 1.03929615e+00 -1.21555591e+00
-2.61141777e-01 -2.25483030e-01 -5.47380924e-01 -1.24424243e+00
8.75975788e-01 -1.23741543e+00 -2.21345678e-01 -1.23366380e+00
2.54897505e-01 -3.55855554e-01 2.06883609e-01 6.00872040e-01
1.15622327e-01 5.35303354e-01 -3.00772656e-02 2.24372782e-02
-2.85998493e-01 9.46284175e-01 1.43980134e+00 -3.31709236e-02
8.95123184e-03 -3.71271998e-01 -5.18412232e-01 8.60695958e-01
4.91172336e-02 -3.76940936e-01 -4.69565779e-01 -8.86609972e-01
1.44661471e-01 1.62808627e-01 5.50522506e-01 -5.35370946e-01
-7.51434639e-02 -2.23195940e-01 8.18061352e-01 -4.13271010e-01
7.83184767e-01 -9.15064335e-01 3.65547359e-01 3.06431856e-02
-4.38361503e-02 1.42307907e-01 2.11305588e-01 5.33757269e-01
9.10698175e-02 3.66325945e-01 8.49835038e-01 -1.58608645e-01
-2.11630896e-01 1.18480444e+00 5.41543722e-01 1.65208980e-01
7.50869691e-01 -3.06905091e-01 4.29911762e-02 -3.03764880e-01
-5.61463535e-01 2.71405578e-02 1.13652313e+00 4.79603738e-01
1.18074000e+00 -1.62814260e+00 -7.41645038e-01 8.61909986e-01
5.73717393e-02 3.85991126e-01 2.84811288e-01 4.00404155e-01
-6.33092105e-01 -7.13909343e-02 -3.20273072e-01 -6.99442267e-01
-1.30228519e+00 3.13290775e-01 6.12879634e-01 7.47206435e-02
-7.46930897e-01 1.16544139e+00 5.33150494e-01 -8.95423174e-01
3.19882393e-01 -3.72561067e-02 2.89296836e-01 -3.04790109e-01
4.60877061e-01 6.40140101e-02 6.90109357e-02 -1.11897564e+00
-4.92362261e-01 1.29879546e+00 2.28076652e-01 -2.10109800e-01
1.59839773e+00 9.81215835e-02 -4.66035277e-01 2.64644861e-01
1.50222254e+00 1.15809590e-01 -1.70026457e+00 -2.67051458e-01
-6.62542641e-01 -1.06588984e+00 6.98118880e-02 -4.77860481e-01
-1.62069356e+00 1.03129494e+00 5.55558503e-01 -6.50121987e-01
1.05341566e+00 1.13381147e-01 6.25241339e-01 4.03030030e-03
6.05412722e-01 -5.37223339e-01 4.32102233e-01 4.58666116e-01
1.30161667e+00 -1.28477108e+00 2.52117455e-01 -7.64636695e-01
-2.49573857e-01 1.14474177e+00 7.45342433e-01 -3.15214187e-01
9.47808921e-01 3.11187476e-01 -2.27992740e-02 -5.40271103e-01
-6.15243316e-01 8.67443159e-02 5.31242132e-01 7.59001911e-01
4.42011178e-01 -2.72666961e-01 4.41611767e-01 2.52954483e-01
-1.38560236e-01 1.96509026e-02 -1.02542445e-01 8.62186551e-01
2.36125931e-01 -1.24946666e+00 -2.02088252e-01 2.21703559e-01
-2.45338306e-01 9.04532149e-02 -5.25198400e-01 4.91887212e-01
7.30720013e-02 5.10242760e-01 3.11742127e-01 -3.77756715e-01
4.56595808e-01 8.57811570e-02 1.23279738e+00 -9.36268866e-01
-2.64779568e-01 6.63974062e-02 -3.30054522e-01 -1.24429357e+00
-6.21604741e-01 -7.52499938e-01 -1.20254874e+00 -3.43058467e-01
5.37758805e-02 -4.34677243e-01 5.43656945e-01 7.05607772e-01
7.14456141e-01 4.47245874e-02 1.07120037e+00 -1.47265685e+00
-3.47093433e-01 -4.38426316e-01 -6.85565829e-01 6.50245547e-01
4.21127230e-01 -9.92082953e-01 -3.70394409e-01 2.95891762e-01] | [13.075584411621094, -0.0381205715239048] |
24f19759-753f-483d-90e7-b89da22b51f7 | explaining-deep-neural-networks-using | 1908.02374 | null | https://arxiv.org/abs/1908.02374v2 | https://arxiv.org/pdf/1908.02374v2.pdf | Explaining Image Classifiers using Statistical Fault Localization | The black-box nature of deep neural networks (DNNs) makes it impossible to understand why a particular output is produced, creating demand for "Explainable AI". In this paper, we show that statistical fault localization (SFL) techniques from software engineering deliver high quality explanations of the outputs of DNNs, where we define an explanation as a minimal subset of features sufficient for making the same decision as for the original input. We present an algorithm and a tool called DeepCover, which synthesizes a ranking of the features of the inputs using SFL and constructs explanations for the decisions of the DNN based on this ranking. We compare explanations produced by DeepCover with those of the state-of-the-art tools GradCAM, LIME, SHAP, RISE and Extremal and show that explanations generated by DeepCover are consistently better across a broad set of experiments. On a benchmark set with known ground truth, DeepCover achieves 76.7% accuracy, which is 6% better than the second best Extremal. | ['Xiaowei Huang', 'Youcheng Sun', 'Daniel Kroening', 'Hana Chockler'] | 2019-08-06 | null | null | null | null | ['fault-localization'] | ['computer-code'] | [ 2.01632068e-01 7.34738171e-01 -8.82918164e-02 -6.57413304e-01
-3.16379368e-01 -4.58576173e-01 5.12685955e-01 -1.56670317e-01
4.06207323e-01 7.66443372e-01 1.36257425e-01 -5.68735242e-01
-4.92066890e-01 -6.14530504e-01 -1.01911390e+00 -3.83108348e-01
1.05537362e-02 3.83103162e-01 2.51162708e-01 4.11634780e-02
4.74444151e-01 5.23096144e-01 -1.93064427e+00 8.83405447e-01
7.99057424e-01 1.00202680e+00 -3.51962484e-02 7.98109174e-01
-1.86298683e-01 1.08650982e+00 -9.80968773e-01 -2.22280949e-01
-6.37626573e-02 -6.30160689e-01 -9.43456888e-01 -1.02611341e-01
6.40186191e-01 -3.41583610e-01 -2.87595570e-01 1.00426877e+00
-4.94172648e-02 -4.94599640e-01 6.54021502e-01 -2.01601648e+00
-9.37618375e-01 1.07702255e+00 1.87780902e-01 9.63922739e-02
4.40227762e-02 5.84801324e-02 1.14178455e+00 -8.95066500e-01
4.93093640e-01 1.22493672e+00 7.63806760e-01 9.60841060e-01
-1.18713963e+00 -5.33164620e-01 1.85789075e-02 3.21817845e-01
-8.18555892e-01 -3.25800568e-01 3.74503523e-01 -4.25371468e-01
1.54928195e+00 3.61662567e-01 4.41654623e-01 1.14652705e+00
8.04391325e-01 6.18798673e-01 7.00668037e-01 -5.44799566e-01
6.56816125e-01 -1.13307744e-01 5.30444026e-01 1.03383827e+00
8.40645611e-01 1.57429039e-01 -7.59187162e-01 -8.24950859e-02
7.16133595e-01 2.32991025e-01 -2.16285929e-01 -1.81124598e-01
-1.00118232e+00 5.81633031e-01 4.37014371e-01 3.50914478e-01
-5.14949620e-01 8.40402663e-01 1.32502437e-01 4.13570166e-01
2.11158887e-01 7.95498252e-01 -1.08856654e+00 4.76772524e-02
-6.16873324e-01 2.05268130e-01 9.64993238e-01 1.08528054e+00
8.70011032e-01 5.73997438e-01 -8.25585499e-02 8.79938230e-02
3.67007703e-01 2.74689823e-01 4.30547655e-01 -1.48198628e+00
7.31079057e-02 1.14402354e+00 -1.46126747e-01 -7.16898620e-01
-2.98124492e-01 -6.98612392e-01 -5.93936563e-01 6.52413785e-01
1.49327889e-01 -2.46168226e-01 -1.13672101e+00 1.54630053e+00
-4.56228882e-01 -2.20269617e-03 3.91435444e-01 6.16829813e-01
7.96764255e-01 3.88994813e-01 -3.95111024e-01 3.11046422e-01
7.24217951e-01 -9.34940934e-01 -3.89628679e-01 -6.69058561e-01
6.93127751e-01 7.85849094e-02 9.13326383e-01 9.19717968e-01
-6.73781991e-01 -5.46380579e-01 -1.49567401e+00 2.62325466e-01
-2.12226450e-01 3.05883586e-01 8.09337258e-01 2.37116233e-01
-1.37435484e+00 1.09578168e+00 -8.56779695e-01 -1.81173608e-01
4.84030992e-01 5.54386914e-01 -5.06708145e-01 -4.78200689e-02
-7.05627739e-01 9.14457023e-01 3.04822236e-01 -1.44435227e-01
-1.48231518e+00 -5.78051090e-01 -7.15513051e-01 4.61080939e-01
2.47322798e-01 -5.96250296e-01 1.55720794e+00 -1.21491826e+00
-7.23764479e-01 4.03270185e-01 6.26383210e-03 -8.00067544e-01
1.91398397e-01 -3.69607389e-01 -4.37938005e-01 -5.96672148e-02
1.94108486e-01 7.07902133e-01 4.67287660e-01 -1.46966565e+00
-7.62698472e-01 -2.05033198e-01 2.47792751e-01 -6.86442733e-01
-1.67565420e-01 -3.89702588e-01 8.06391239e-02 7.11024404e-02
3.62964451e-01 -7.04175711e-01 -2.64590293e-01 3.66858346e-03
-9.69269037e-01 -3.70881587e-01 7.80086935e-01 -4.13866729e-01
9.23778236e-01 -1.89942157e+00 -5.51099330e-02 1.26722082e-01
7.30552912e-01 -1.60709247e-01 -8.30424130e-02 2.06302807e-01
-6.53318703e-01 4.30151433e-01 -1.89977139e-01 -2.47448608e-01
4.02223408e-01 6.66537821e-01 -4.25559878e-01 1.85834467e-01
5.45145988e-01 6.08185232e-01 -7.71854937e-01 -1.26929238e-01
-5.48396558e-02 2.57981181e-01 -6.11929655e-01 1.64676517e-01
-5.45744061e-01 -2.90216655e-01 -2.53548771e-01 7.73631752e-01
1.10148333e-01 -3.72383058e-01 1.61850788e-02 2.24274993e-01
8.23667794e-02 4.69163984e-01 -8.91361237e-01 1.30319238e+00
-1.46863952e-01 1.27932250e+00 -6.66629016e-01 -7.18084633e-01
1.27238953e+00 5.17973065e-01 -1.48656413e-01 -3.43003124e-01
1.56500220e-01 4.44474876e-01 2.97959328e-01 -4.87915784e-01
1.00473806e-01 3.53589319e-02 -5.48474453e-02 5.91330349e-01
3.26279342e-01 1.76453039e-01 3.21136462e-03 1.47972822e-01
1.97382295e+00 -1.72782950e-02 2.53807873e-01 -4.04263496e-01
7.62376711e-02 2.97107488e-01 7.78064549e-01 1.06105244e+00
1.15112506e-01 8.61912668e-01 1.25939763e+00 -1.01846433e+00
-1.05628431e+00 -9.36836720e-01 3.04447830e-01 6.40842140e-01
-2.29123741e-01 -3.36922675e-01 -1.16989815e+00 -1.21660173e+00
1.20603301e-01 1.44865036e+00 -1.01691175e+00 -4.37504023e-01
7.94432685e-03 -3.83117013e-02 6.23584926e-01 9.87560868e-01
1.78717449e-01 -1.42181134e+00 -9.93548572e-01 8.04803818e-02
2.90522873e-01 -8.40507150e-01 3.29476774e-01 1.02492309e+00
-1.05062258e+00 -1.21565223e+00 3.52933221e-02 -5.33496380e-01
1.11682236e+00 -4.14250754e-02 1.59407902e+00 7.67119646e-01
-1.52545482e-01 -8.80480781e-02 -2.71846712e-01 -5.65520108e-01
-8.49468410e-01 8.13591108e-03 2.28490487e-01 -6.52809918e-01
4.77384329e-01 -6.61860764e-01 -1.08956315e-01 1.82461649e-01
-9.50476885e-01 2.66238451e-01 8.59998465e-01 6.74358428e-01
3.43050241e-01 3.43904585e-01 6.24770880e-01 -1.04290700e+00
4.83499944e-01 -5.75483322e-01 -3.90014172e-01 3.17159384e-01
-9.64211822e-01 7.68247724e-01 8.82524431e-01 -1.25904694e-01
-5.99837184e-01 8.29664767e-02 1.40733287e-01 -4.41565543e-01
-5.44503987e-01 3.66332233e-01 -3.19531858e-01 2.99035370e-01
1.25418758e+00 -3.46401334e-02 -3.38398725e-01 -3.72425526e-01
3.54893506e-02 6.01450443e-01 8.72719347e-01 -3.68534267e-01
5.91707408e-01 4.18332517e-02 -4.09288630e-02 4.13847975e-02
-9.34496880e-01 4.59444314e-01 -5.06030142e-01 -1.92020029e-01
4.61412668e-01 -3.02803606e-01 -4.40055877e-01 6.87563233e-03
-1.63226962e+00 -3.29697758e-01 -4.44036573e-01 1.87641472e-01
-5.44435441e-01 -3.93593162e-01 -3.37265372e-01 -6.79866433e-01
-2.67829299e-01 -1.27854824e+00 7.43135929e-01 2.55947351e-01
-7.99744368e-01 -6.43124640e-01 -1.91116437e-01 -2.02060357e-01
2.41484597e-01 5.18675923e-01 1.31982362e+00 -1.35424542e+00
-6.33609951e-01 -3.23727310e-01 -1.41989797e-01 7.44216383e-01
4.12304886e-02 4.18391049e-01 -1.26202452e+00 3.54064196e-01
-1.11901097e-01 2.53296681e-02 7.11474001e-01 3.15213829e-01
1.26820922e+00 -4.40337181e-01 -3.71765792e-01 2.63329595e-01
1.45641351e+00 3.02306771e-01 6.23522818e-01 5.10059476e-01
5.44525981e-01 4.76970434e-01 2.65489966e-01 3.55355650e-01
2.15294808e-02 -1.38659999e-01 1.22814369e+00 3.57071258e-04
-7.93797672e-02 -1.25036404e-01 4.92148131e-01 2.78325349e-01
6.66350126e-02 -4.14749444e-01 -1.22679710e+00 7.73799658e-01
-2.06643128e+00 -6.53586566e-01 -4.59969938e-01 1.62343013e+00
3.85233402e-01 6.70575142e-01 -5.46494186e-01 6.67453349e-01
4.22788054e-01 -4.02009606e-01 -8.28368247e-01 -7.85788834e-01
-9.10031497e-02 1.13115557e-01 3.48165900e-01 3.80268365e-01
-5.02047241e-01 6.51170373e-01 6.99249601e+00 1.74845457e-01
-6.40269697e-01 -1.50191024e-01 5.88370383e-01 2.96955615e-01
-7.00043023e-01 5.92886917e-02 -5.68603873e-01 5.82623333e-02
1.32073152e+00 -1.76994324e-01 3.80689621e-01 1.41347599e+00
-1.06157072e-01 1.07092202e-01 -1.94657815e+00 3.49550337e-01
-1.16639994e-02 -1.70921004e+00 1.53301684e-02 -4.38224375e-02
9.74240065e-01 2.80767474e-02 -9.10636038e-02 6.32942319e-02
7.00127482e-01 -1.40692413e+00 1.08468974e+00 6.29910648e-01
5.21368444e-01 -8.75015795e-01 1.19735217e+00 3.74853224e-01
-3.48555326e-01 -3.61423880e-01 -4.93768930e-01 -3.53564203e-01
-7.19076991e-01 6.94687724e-01 -1.31889069e+00 2.33594984e-01
7.47761726e-01 7.05193281e-01 -6.84409320e-01 7.97474444e-01
-9.30594862e-01 7.56124794e-01 1.13396309e-01 -1.37988165e-01
1.48233980e-01 7.82585979e-01 3.16623867e-01 1.11620522e+00
6.04252100e-01 -2.58934796e-01 -4.70353872e-01 1.49420142e+00
-1.30468860e-01 -8.13085794e-01 -8.20337355e-01 -2.69480944e-01
5.65554798e-01 1.07968271e+00 -7.27813363e-01 -4.92539793e-01
-7.69920647e-02 8.04038227e-01 6.37402833e-02 1.43949687e-01
-7.58886099e-01 -3.92073750e-01 7.84641564e-01 6.21596798e-02
1.16977252e-01 1.83397800e-01 -8.35664332e-01 -6.01279795e-01
3.64482240e-03 -1.14841688e+00 2.38171276e-02 -1.33365321e+00
-1.05389106e+00 1.18948567e+00 -1.56697661e-01 -1.11092079e+00
-7.33965278e-01 -9.47629869e-01 -8.15437973e-01 7.06583202e-01
-8.97206724e-01 -5.41718841e-01 -5.53025663e-01 -2.44193189e-02
5.69083750e-01 -4.89084005e-01 1.07992446e+00 -2.61341035e-01
-4.57214922e-01 2.78191477e-01 -1.95956841e-01 -4.39313315e-02
7.62351826e-02 -1.71416247e+00 9.88532364e-01 1.02028155e+00
3.52933884e-01 5.43152452e-01 1.32651591e+00 -5.68578064e-01
-1.27981746e+00 -1.04124117e+00 1.10219038e+00 -6.10336661e-01
5.15581250e-01 -1.01568662e-01 -9.13830876e-01 1.02509177e+00
2.80123532e-01 -1.54395336e-02 3.09619039e-01 1.63704250e-02
-4.35202420e-01 -8.63789096e-02 -1.17088091e+00 4.63464707e-01
1.08854401e+00 -4.19778973e-01 -7.16696620e-01 2.07398921e-01
1.05201674e+00 -2.64975458e-01 -3.68871868e-01 2.59467244e-01
4.92807150e-01 -1.53770518e+00 4.99024570e-01 -8.89418185e-01
1.21380043e+00 -5.21108150e-01 -3.99672061e-01 -1.62369728e+00
-3.05510670e-01 -2.94015408e-01 -2.09008366e-01 9.75077748e-01
9.08316433e-01 -6.11553490e-01 9.67799544e-01 8.02082419e-01
-8.39912891e-01 -9.25278246e-01 -6.59720659e-01 -7.02445626e-01
-3.42148572e-01 -8.02345693e-01 9.49362516e-01 6.03466153e-01
2.83582192e-02 8.83864164e-02 1.07135601e-01 4.89888549e-01
6.29972458e-01 -4.54429490e-03 5.46774209e-01 -1.64831603e+00
-2.85782993e-01 -2.67095715e-01 -7.29406536e-01 -4.31362957e-01
4.41553444e-01 -7.52628326e-01 5.19733429e-01 -2.03501749e+00
2.38707423e-01 4.85920161e-02 -4.72245663e-01 1.22744191e+00
2.86772251e-01 -1.54863179e-01 -1.55062005e-01 -5.65150529e-02
-5.40723443e-01 1.34979725e-01 7.27529168e-01 -9.75732133e-02
1.60737813e-01 -2.75867403e-01 -1.22640502e+00 1.07727194e+00
7.25728810e-01 -1.07641768e+00 -3.08711678e-01 -8.26639652e-01
3.59636366e-01 -2.47235671e-02 6.65948927e-01 -1.43173587e+00
1.59541622e-01 -8.08877349e-02 6.08047128e-01 -4.70077038e-01
-1.60122529e-01 -8.28916848e-01 4.19241399e-01 7.88137257e-01
-7.03615665e-01 4.03164297e-01 2.50673473e-01 2.38865882e-01
-1.02830678e-01 -5.40534437e-01 2.15327978e-01 1.22417971e-01
-8.57259214e-01 -1.92444891e-01 -4.44806963e-01 -3.41579825e-01
6.08955503e-01 -3.01735252e-01 -8.17059338e-01 -3.66314203e-01
-5.14813900e-01 -4.73348536e-02 4.41729128e-01 4.48009133e-01
9.13329244e-01 -1.23277199e+00 -4.65287626e-01 2.08815098e-01
1.24275193e-01 5.61346747e-02 -2.50757545e-01 2.78006345e-01
-5.40085554e-01 5.04660666e-01 -3.69237989e-01 -4.77134585e-01
-1.12420416e+00 1.17728151e-01 6.55652523e-01 1.60500512e-01
-6.80691779e-01 9.48226571e-01 1.25408038e-01 -4.12730783e-01
2.80346513e-01 -9.07638788e-01 1.49837285e-01 -6.48433685e-01
4.95889217e-01 2.93759227e-01 2.12948859e-01 4.30332124e-02
-4.85807180e-01 -1.69444412e-01 2.35738978e-01 5.48145100e-02
1.84329975e+00 5.96102595e-01 -9.38381031e-02 6.69424295e-01
6.81300163e-01 -5.72649300e-01 -1.50039196e+00 1.55115262e-01
4.28919375e-01 -1.18003920e-01 7.39577860e-02 -1.16725397e+00
-1.23588431e+00 1.13266718e+00 3.57995152e-01 6.82427287e-01
1.05836880e+00 1.86209366e-01 3.15838814e-01 6.44801438e-01
4.26340878e-01 -5.53779066e-01 1.18749671e-01 5.18424809e-01
1.01905441e+00 -7.35700727e-01 -3.64880800e-01 -3.89075987e-02
-4.66766566e-01 1.58098328e+00 8.76991034e-01 -4.28620487e-01
7.75968358e-02 5.96100330e-01 -7.28228912e-02 -5.74934483e-01
-1.44911671e+00 3.29845399e-01 2.19275042e-01 5.41528165e-01
2.75196582e-01 8.67399499e-02 4.50319439e-01 1.24495769e+00
-3.74723345e-01 1.98689297e-01 8.22742939e-01 7.86377668e-01
-8.91198993e-01 -6.95342958e-01 -3.11722398e-01 6.09052241e-01
-3.45740795e-01 -1.00875802e-01 -9.31061387e-01 9.82404768e-01
4.52001244e-01 9.84295249e-01 2.31222194e-02 -1.02664554e+00
4.58598971e-01 1.81512296e-01 1.49349347e-01 -9.03250277e-01
-5.04144371e-01 -6.24006689e-01 3.44896406e-01 -6.94080710e-01
1.14465080e-01 -4.22349721e-01 -1.98533869e+00 -3.24235052e-01
-2.54085332e-01 1.43391609e-01 8.22680473e-01 1.41661561e+00
3.84777576e-01 1.06951427e+00 3.29021782e-01 -5.88623166e-01
-5.26528120e-01 -7.94045985e-01 -4.85726923e-01 -8.15932751e-02
5.02056718e-01 -5.25573134e-01 -8.37965667e-01 1.19682997e-01] | [8.931354522705078, 5.75799560546875] |
fe77dd06-0831-40ab-a679-6458bf97298c | robustness-of-on-device-models-adversarial | 2101.04401 | null | https://arxiv.org/abs/2101.04401v2 | https://arxiv.org/pdf/2101.04401v2.pdf | Robustness of on-device Models: Adversarial Attack to Deep Learning Models on Android Apps | Deep learning has shown its power in many applications, including object detection in images, natural-language understanding, and speech recognition. To make it more accessible to end users, many deep learning models are now embedded in mobile apps. Compared to offloading deep learning from smartphones to the cloud, performing machine learning on-device can help improve latency, connectivity, and power consumption. However, most deep learning models within Android apps can easily be obtained via mature reverse engineering, while the models' exposure may invite adversarial attacks. In this study, we propose a simple but effective approach to hacking deep learning models using adversarial attacks by identifying highly similar pre-trained models from TensorFlow Hub. All 10 real-world Android apps in the experiment are successfully attacked by our approach. Apart from the feasibility of the model attack, we also carry out an empirical study that investigates the characteristics of deep learning models used by hundreds of Android apps on Google Play. The results show that many of them are similar to each other and widely use fine-tuning techniques to pre-trained models on the Internet. | ['Chunyang Chen', 'Han Hu', 'Yujin Huang'] | 2021-01-12 | null | null | null | null | ['mobile-security'] | ['miscellaneous'] | [-3.01263072e-02 -5.02108708e-02 -4.27692115e-01 -6.81108050e-03
-5.08467019e-01 -9.71510947e-01 3.04253697e-01 -6.92516983e-01
-9.54543203e-02 4.34160471e-01 -4.77457106e-01 -9.51099813e-01
2.37716600e-01 -5.22707224e-01 -1.23589540e+00 -1.21438861e-01
-1.41958028e-01 4.26046811e-02 3.94628733e-01 -2.23134439e-02
3.96650322e-02 3.74216497e-01 -9.79080319e-01 4.68275934e-01
8.20137084e-01 9.81377244e-01 -2.64067799e-01 8.22212100e-01
2.04077080e-01 9.97943163e-01 -9.43936944e-01 -1.20407796e+00
3.59584928e-01 3.71421501e-02 -8.14758301e-01 -5.46889722e-01
4.25203830e-01 -6.08678222e-01 -4.47278410e-01 1.29737628e+00
5.14961362e-01 -5.10241091e-01 3.15458737e-02 -1.56195700e+00
-4.48559105e-01 6.24313831e-01 -4.61044401e-01 1.56681374e-01
1.94467291e-01 4.36137259e-01 4.58602279e-01 -2.77835578e-01
-5.27233332e-02 1.06078255e+00 1.03431141e+00 7.97481418e-01
-7.61392236e-01 -1.37291670e+00 -1.99821681e-01 1.71642557e-01
-1.16252732e+00 -5.73680282e-01 6.57630324e-01 -2.75213748e-01
9.15574253e-01 3.09855610e-01 4.86761689e-01 1.91058052e+00
6.51260912e-01 5.93724668e-01 1.07248521e+00 -3.94551679e-02
3.76687914e-01 4.32406336e-01 9.46008228e-03 6.01595700e-01
3.50858033e-01 1.02045171e-01 -2.53621489e-01 -7.89003074e-01
3.80033821e-01 2.15779886e-01 1.37889609e-01 2.91884448e-02
-5.59053421e-01 7.33551562e-01 1.87719226e-01 -4.57753241e-02
-7.86274299e-02 4.69205886e-01 7.64395833e-01 5.41838035e-02
2.05959573e-01 4.86905545e-01 -1.06093848e+00 -6.63766921e-01
-6.96337402e-01 -2.53175497e-01 1.05012226e+00 8.10551047e-01
6.87543750e-01 1.04606591e-01 6.13543868e-01 2.69155920e-01
2.30049908e-01 5.86672187e-01 1.01760185e+00 -9.77848649e-01
5.37479341e-01 4.33340192e-01 -3.50082785e-01 -1.02836812e+00
2.07258910e-01 -2.96860039e-01 -4.49777693e-01 3.23353847e-03
3.01798433e-01 -6.51768804e-01 -5.58297038e-01 1.30971432e+00
6.09823614e-02 9.49202478e-01 -3.81942429e-02 1.57046720e-01
3.58914226e-01 3.98177445e-01 1.01462938e-01 2.53083229e-01
1.30665231e+00 -1.08478987e+00 -3.39202374e-01 -3.76172572e-01
8.10233772e-01 -6.54040575e-01 1.72039235e+00 5.71859777e-01
-7.52856612e-01 -6.47092402e-01 -1.19611037e+00 7.36363903e-02
-5.74939847e-01 -4.50662374e-02 9.39167261e-01 1.61753535e+00
-7.65518010e-01 4.71978366e-01 -1.23549318e+00 2.51247128e-03
1.04087842e+00 8.19445789e-01 -1.19075172e-01 2.92466789e-01
-1.07026541e+00 2.11136192e-01 1.57766659e-02 -2.90798098e-01
-1.41195250e+00 -7.46890545e-01 -4.07656848e-01 1.96098998e-01
3.38187456e-01 -3.61786157e-01 1.45319057e+00 -1.33402467e+00
-1.83943462e+00 5.67406654e-01 8.78460482e-02 -7.15607643e-01
2.81327903e-01 -6.69203818e-01 -6.93257332e-01 -6.65837899e-02
-1.34219170e-01 7.97230154e-02 1.39769602e+00 -9.52270091e-01
-2.69043803e-01 -2.40149006e-01 5.50079465e-01 -5.03415227e-01
-1.05522299e+00 2.14657620e-01 -3.99605811e-01 -5.50286591e-01
-7.78621197e-01 -1.31263113e+00 -8.71219113e-02 -3.45885336e-01
-7.40557611e-01 3.38628292e-01 1.43055522e+00 -7.63542950e-01
1.11342311e+00 -2.36502624e+00 -5.70780396e-01 2.97454596e-01
4.05879617e-01 8.81573141e-01 1.09808259e-02 -8.80726874e-02
1.12630725e-02 8.11228395e-01 1.92991093e-01 -2.52213189e-03
-1.16769947e-01 2.20957860e-01 -6.70492470e-01 1.56604454e-01
-2.03218043e-01 1.18866944e+00 -5.97062767e-01 -2.27752224e-01
-2.11003572e-01 5.36871612e-01 -8.73953402e-01 1.34260431e-01
-1.21024221e-01 1.09963052e-01 -5.59717238e-01 6.59553885e-01
6.66385472e-01 -3.49535882e-01 3.00088763e-01 -9.49546173e-02
7.33101964e-01 5.34335077e-01 -5.64469576e-01 1.13100171e+00
-7.52013206e-01 7.89567173e-01 -4.45861220e-02 -7.06409276e-01
3.01440597e-01 2.92547476e-02 1.02269799e-01 -5.67749500e-01
2.06248671e-01 1.56060264e-01 2.39553049e-01 -5.31109989e-01
1.69573575e-01 5.68466008e-01 -6.76922873e-02 6.24168038e-01
-3.89956795e-02 2.43080512e-01 -5.80475867e-01 2.12664500e-01
1.36113453e+00 -2.89220303e-01 -4.25779335e-02 6.89965487e-02
4.42867428e-01 -5.14716089e-01 1.08230710e-01 9.12868857e-01
-3.15696061e-01 -4.66796383e-02 5.99395037e-01 -4.46169198e-01
-4.38454390e-01 -1.06940806e+00 2.94487178e-01 1.20009172e+00
-4.83848602e-02 -9.21235085e-01 -1.51166129e+00 -1.54801095e+00
-3.76015991e-01 3.74297082e-01 -5.98992527e-01 -5.92527926e-01
-5.26637673e-01 -6.09349728e-01 1.44865668e+00 5.10946214e-01
9.57426488e-01 -8.55981886e-01 -3.33124250e-01 -1.46784514e-01
1.69303983e-01 -1.35578465e+00 -5.93237698e-01 8.31635073e-02
-9.31202054e-01 -1.22725952e+00 -1.24494925e-01 -5.00303090e-01
4.11433846e-01 2.48233944e-01 1.05754483e+00 3.32063079e-01
-1.11112639e-01 3.86347353e-01 -9.31294113e-02 -5.79554021e-01
-6.67405009e-01 6.98717237e-01 3.39183569e-01 -1.25365483e-03
4.96371984e-01 -5.82279265e-01 -3.30016196e-01 6.48309946e-01
-8.77578914e-01 -7.67247081e-01 3.60822350e-01 4.67788041e-01
1.63626015e-01 6.43196046e-01 3.57804179e-01 -1.16490757e+00
7.03575730e-01 -6.08274877e-01 -5.02120137e-01 4.74166684e-02
-4.60402191e-01 -3.93310398e-01 8.87290835e-01 -1.16016567e+00
-7.25088418e-01 -5.60492761e-02 -5.86804450e-01 -4.12530810e-01
-8.84390026e-02 8.04105923e-02 -6.72911584e-01 -6.81200027e-01
7.95585275e-01 -1.21890351e-01 -5.19904234e-02 -4.35539424e-01
2.99413830e-01 7.63461590e-01 2.93821990e-01 -4.48565036e-01
1.00595856e+00 3.52145016e-01 -4.13153261e-01 -8.60199809e-01
-5.78717411e-01 5.05758405e-01 -1.89816896e-02 1.52499095e-01
6.53825343e-01 -8.89592946e-01 -1.03244603e+00 7.71192372e-01
-1.02241266e+00 -7.31204212e-01 -3.77179645e-02 1.45456806e-01
-1.16859041e-01 4.70078200e-01 -7.97548592e-01 -2.56050348e-01
-3.90599012e-01 -1.57026255e+00 7.21458316e-01 1.52928561e-01
-2.41918027e-01 -1.14114141e+00 -1.52635306e-01 6.23367786e-01
5.81233561e-01 -3.36885482e-01 8.34740520e-01 -8.89240086e-01
-4.58582550e-01 -3.75584066e-01 5.86224236e-02 8.31047237e-01
1.33968547e-01 2.88714975e-01 -1.31745541e+00 -3.51775736e-01
4.83246207e-01 -1.73280507e-01 1.82144791e-01 1.92944661e-01
1.89720285e+00 -9.23613012e-01 -5.32874525e-01 9.61112618e-01
9.72960472e-01 2.69176126e-01 1.02659488e+00 3.76395255e-01
1.11460507e+00 1.12680480e-01 8.52804556e-02 -3.70042883e-02
9.81017426e-02 6.63446546e-01 6.82215512e-01 -1.09177396e-01
3.70185822e-01 -3.42733890e-01 8.97053003e-01 4.90783334e-01
1.66150048e-01 6.38407916e-02 -8.31007242e-01 -9.81926173e-02
-1.14422369e+00 -6.59453869e-01 -9.53324810e-02 1.93939328e+00
7.20389962e-01 7.36462951e-01 2.21171230e-01 9.23556313e-02
4.61555392e-01 -1.85309201e-01 -6.27966821e-01 -6.71139359e-01
3.44923556e-01 7.38759995e-01 7.27733552e-01 2.54348010e-01
-1.00836730e+00 1.17127216e+00 6.79015160e+00 1.13262498e+00
-1.52805340e+00 6.78832114e-01 8.71689618e-01 -2.03831941e-01
-3.44536863e-02 -8.60585496e-02 -8.43990505e-01 1.00859058e+00
1.49564898e+00 1.87296689e-01 6.67989612e-01 1.40707517e+00
-6.10169433e-02 3.71867657e-01 -1.00819671e+00 1.03006589e+00
-2.87830889e-01 -1.23532367e+00 -1.12702951e-01 4.93282139e-01
6.68622732e-01 2.18680918e-01 6.53339863e-01 6.65493250e-01
2.08124936e-01 -1.04811907e+00 2.74051905e-01 -2.65168883e-02
8.44505191e-01 -1.03943145e+00 6.87248290e-01 3.50941658e-01
-6.72542632e-01 -3.23475808e-01 -1.75576985e-01 -1.98476091e-01
-3.94524217e-01 3.30583185e-01 -7.66763687e-01 -2.00052246e-01
1.14229572e+00 4.92400169e-01 -1.26251054e+00 3.75951558e-01
-1.96322933e-01 1.42640293e+00 -2.06198663e-01 1.69950090e-02
-2.86902133e-02 3.14769417e-01 1.29042834e-01 7.93345988e-01
1.56938106e-01 -5.57509542e-01 -2.89120108e-01 5.98270535e-01
-5.80018878e-01 -1.33122727e-01 -8.79010916e-01 -3.72174978e-01
4.79510754e-01 1.18301380e+00 -5.28313577e-01 -2.57152498e-01
-4.26474690e-01 1.21672475e+00 -3.81643884e-02 3.16759169e-01
-1.40653145e+00 -3.87582242e-01 1.18963289e+00 2.94146985e-01
-2.96315667e-03 -2.13503525e-01 -3.54197443e-01 -1.16700351e+00
1.33764520e-01 -1.62214828e+00 -1.34770676e-01 -4.94178951e-01
-1.16762578e+00 6.93979442e-01 -3.64406615e-01 -8.51740241e-01
-2.76446819e-01 -7.13868499e-01 -9.02207732e-01 4.49763864e-01
-9.50550318e-01 -1.01753020e+00 -1.12322643e-02 8.21067274e-01
2.80085295e-01 -7.40428388e-01 7.38489270e-01 5.83898306e-01
-8.47582638e-01 1.31118679e+00 8.06381777e-02 3.07506710e-01
4.73179132e-01 -7.48439431e-01 1.05863631e+00 8.72940361e-01
3.67225796e-01 1.01035833e+00 1.28988966e-01 -7.98547029e-01
-1.34640157e+00 -1.14179444e+00 1.76736653e-01 -8.00179482e-01
9.49602842e-01 -7.36260533e-01 -9.63607728e-01 9.75561798e-01
2.27010190e-01 1.30307481e-01 9.80835438e-01 -9.32682082e-02
-5.78937292e-01 -2.77474612e-01 -1.16297102e+00 8.50947201e-01
7.83251584e-01 -1.03543174e+00 2.55957425e-01 3.29875410e-01
9.27625239e-01 -3.28293920e-01 -6.56134725e-01 1.50518835e-01
7.84132063e-01 -9.48463261e-01 1.09557390e+00 -1.00452220e+00
3.27182084e-01 1.41547218e-01 -1.64823662e-02 -8.64538431e-01
2.29149744e-01 -1.29244351e+00 -8.59257162e-01 1.27803254e+00
4.28682297e-01 -7.78399944e-01 1.29612935e+00 6.16384268e-01
3.16379666e-01 -6.48189366e-01 -7.65564978e-01 -8.72687578e-01
7.54561797e-02 -7.65486062e-01 9.73544061e-01 1.08672214e+00
-5.44300795e-01 3.11570793e-01 -4.74653602e-01 3.44088346e-01
3.73315215e-01 -9.05956566e-01 1.20472574e+00 -9.05981362e-01
-9.28422511e-01 -1.61441714e-01 -3.78899276e-01 -9.68932569e-01
5.90185583e-01 -5.35502791e-01 -6.10760570e-01 -3.79472762e-01
-1.81322750e-02 -3.32363695e-01 -5.83661981e-02 6.85262799e-01
2.44212728e-02 7.03893483e-01 1.43292487e-01 1.63492292e-01
-4.58088875e-01 -4.56761895e-03 7.12042332e-01 -3.56939614e-01
-3.43701661e-01 5.05589187e-01 -9.96433616e-01 1.20554137e+00
1.05802786e+00 -7.40391910e-01 -6.61337435e-01 -4.31818545e-01
5.05747080e-01 -5.19466996e-01 4.48888570e-01 -1.20688963e+00
-7.35436678e-02 2.21360967e-01 1.43214956e-01 2.81114399e-01
1.15979917e-01 -1.17029428e+00 1.09740570e-01 5.46307802e-01
-2.42966563e-02 1.38173223e-01 4.98490930e-01 3.10172796e-01
2.85907954e-01 -2.95122862e-01 5.97026587e-01 9.77898538e-02
-4.77718502e-01 2.15730861e-01 -6.02565646e-01 4.26605269e-02
1.11732101e+00 -2.53283024e-01 -3.95969778e-01 -4.61501271e-01
-6.01963520e-01 -4.81423378e-01 5.19610226e-01 6.59827948e-01
4.29561555e-01 -9.91628170e-01 1.18713327e-01 5.54984868e-01
-2.76194781e-01 -4.91037905e-01 3.66774611e-02 6.32669389e-01
-5.93698919e-01 -4.76826318e-02 -2.28923157e-01 -4.16853189e-01
-1.44823551e+00 8.29137921e-01 6.56561673e-01 -2.63383329e-01
-9.40526873e-02 6.20442092e-01 2.39272743e-01 -4.09793615e-01
2.75105417e-01 -1.78142503e-01 3.39337379e-01 -5.19429505e-01
6.07774258e-01 5.78020751e-01 2.20718011e-01 -3.09781939e-01
-4.14524436e-01 2.15798154e-01 -3.83572251e-01 2.98671752e-01
6.67549849e-01 2.05462977e-01 -1.50499806e-01 -1.31353483e-01
1.54851425e+00 5.92288077e-01 -1.16759062e+00 2.66466230e-01
-2.09867209e-01 -2.94584692e-01 -1.70962825e-01 -7.55213797e-01
-1.52636957e+00 1.12074602e+00 8.95764768e-01 5.37272096e-01
1.08019793e+00 -1.47668630e-01 1.31106591e+00 6.27764523e-01
4.22323614e-01 -6.11183107e-01 5.60205638e-01 4.02342439e-01
1.34875417e-01 -1.09508026e+00 -3.25351596e-01 -4.63258088e-01
-3.44826251e-01 8.02282572e-01 8.08806181e-01 -8.03957060e-02
1.06185448e+00 7.62231946e-01 1.55657560e-01 3.15665337e-03
-2.63554513e-01 5.84885001e-01 -4.38335389e-02 9.40121889e-01
9.03651565e-02 3.12129199e-03 4.02143449e-01 9.51485693e-01
-2.62998343e-01 -2.59570684e-02 6.18017197e-01 6.55945539e-01
1.68037862e-01 -1.27807796e+00 -1.54744834e-01 5.79719603e-01
-1.33900964e+00 -2.20433131e-01 -5.57407081e-01 6.11710191e-01
1.97009578e-01 9.55843151e-01 -2.75916100e-01 -1.13227296e+00
1.01169357e-02 -1.36557311e-01 8.04914907e-02 -5.29140711e-01
-9.70411777e-01 -2.98592478e-01 2.18929611e-02 -8.27080607e-01
1.42189026e-01 -3.75237092e-02 -7.32760608e-01 -5.63609958e-01
-3.18828315e-01 1.06447767e-02 8.23477149e-01 9.59465325e-01
7.42293775e-01 4.32751715e-01 7.90234029e-01 -8.28036368e-01
-6.72372460e-01 -7.06360400e-01 -3.23340148e-01 1.77186757e-01
1.44577771e-01 -3.95891458e-01 -4.48325008e-01 1.00543305e-01] | [14.419920921325684, 9.679676055908203] |
147f3dac-3512-466d-bc50-df4ecb0ac21b | a-short-survey-of-systematic-generalization | 2211.11956 | null | https://arxiv.org/abs/2211.11956v1 | https://arxiv.org/pdf/2211.11956v1.pdf | A Short Survey of Systematic Generalization | This survey includes systematic generalization and a history of how machine learning addresses it. We aim to summarize and organize the related information of both conventional and recent improvements. We first look at the definition of systematic generalization, then introduce Classicist and Connectionist. We then discuss different types of Connectionists and how they approach the generalization. Two crucial problems of variable binding and causality are discussed. We look into systematic generalization in language, vision, and VQA fields. Recent improvements from different aspects are discussed. Systematic generalization has a long history in artificial intelligence. We could cover only a small portion of many contributions. We hope this paper provides a background and is beneficial for discoveries in future work. | ['Yuanpeng Li'] | 2022-11-22 | null | null | null | null | ['systematic-generalization'] | ['reasoning'] | [ 1.63289979e-01 4.17661704e-02 -8.02839935e-01 -3.73854518e-01
-1.37565523e-01 -6.18901491e-01 5.86434722e-01 -3.91302519e-02
-2.77905554e-01 9.96285081e-01 7.91574121e-02 -4.83177006e-01
-5.68515897e-01 -8.04526806e-01 -3.62733990e-01 -6.71590149e-01
-3.95266235e-01 1.87127918e-01 1.47748343e-03 -6.92472219e-01
6.62287593e-01 6.81149244e-01 -1.26794231e+00 2.35025942e-01
6.97083056e-01 7.92912483e-01 3.07209104e-01 6.80046022e-01
-3.84065777e-01 6.67842925e-01 -7.82226562e-01 -4.74246234e-01
-2.50278190e-02 -6.89055502e-01 -1.29249990e+00 -1.76665604e-01
1.84802189e-01 2.61786014e-01 -4.25600141e-01 1.09842575e+00
4.20371890e-01 2.09388405e-01 8.62104893e-01 -1.64563859e+00
-1.35997438e+00 9.23323810e-01 -4.56637293e-01 1.06329155e+00
3.68374735e-01 -1.78197563e-01 1.48336077e+00 -7.38214314e-01
4.85287398e-01 1.47274518e+00 7.29400098e-01 1.15979004e+00
-9.00258958e-01 -4.15446341e-01 8.79222989e-01 9.14516985e-01
-1.18024993e+00 2.27942467e-01 8.60726535e-01 -4.38273251e-01
1.42072666e+00 3.00533980e-01 8.91666293e-01 1.03361845e+00
4.61050868e-01 7.93186307e-01 1.16791821e+00 -6.80952191e-01
9.21680480e-02 -1.09992914e-01 7.74528861e-01 6.22069240e-01
2.23959431e-01 4.21795577e-01 -7.90382922e-01 2.61645466e-02
6.19802117e-01 -3.61618340e-01 -2.15652451e-01 -3.19996566e-01
-1.06067431e+00 1.03076291e+00 5.92633784e-01 5.47103643e-01
-9.74932909e-02 3.54299456e-01 2.05946401e-01 4.34003443e-01
-1.32299647e-01 7.41614878e-01 -8.53287697e-01 3.26373070e-01
-3.92128855e-01 4.22108769e-01 5.56459725e-01 1.14659035e+00
6.84890747e-01 3.84856373e-01 1.53096989e-01 7.14785516e-01
-1.66408857e-03 4.63647038e-01 6.24986172e-01 -8.24605346e-01
1.38452455e-01 2.49494314e-01 -4.03340936e-01 -7.63976634e-01
-9.24523413e-01 -5.20293772e-01 -7.02184975e-01 1.41085843e-02
-5.13805412e-02 -8.14747885e-02 -6.82399988e-01 1.82491922e+00
-1.61684781e-01 -8.25436264e-02 1.40885845e-01 6.22173071e-01
1.22271895e+00 4.51905459e-01 3.98811549e-01 -6.96797550e-01
1.14930844e+00 -9.40461218e-01 -1.06815529e+00 -1.58096597e-01
5.29082596e-01 -2.42253497e-01 8.35982025e-01 3.84625107e-01
-9.75688636e-01 -4.22203362e-01 -1.10760272e+00 -2.04852641e-01
-7.23696172e-01 -4.09492493e-01 1.41430962e+00 7.94789433e-01
-1.17497647e+00 6.20218396e-01 -6.49759769e-01 -7.14661896e-01
4.36179340e-01 5.62564611e-01 1.38771564e-01 3.05541605e-01
-1.49949813e+00 1.32884920e+00 2.62871027e-01 -1.17693990e-01
-6.91859186e-01 -4.75324929e-01 -8.35939765e-01 -4.46635514e-01
4.35464472e-01 -1.24491119e+00 1.50393581e+00 -9.43126380e-01
-1.16886270e+00 8.05772305e-01 -4.26963598e-01 -6.23574018e-01
-2.41885215e-01 -5.70216440e-02 -6.79259002e-01 -8.87129456e-02
-4.07163352e-02 5.19105256e-01 2.17053264e-01 -1.21594834e+00
-8.72524261e-01 -5.47439396e-01 4.24623907e-01 3.39910239e-01
-2.08024353e-01 2.00861990e-01 -1.11794069e-01 -7.60468006e-01
3.15680236e-01 -7.07784295e-01 -2.11483270e-01 -2.75891513e-01
-2.37505838e-01 -7.69202352e-01 4.20797974e-01 -9.91977900e-02
1.54494548e+00 -1.60285962e+00 4.37704057e-01 -1.73028901e-01
3.42766702e-01 -1.03308961e-01 -8.82776380e-02 7.81877697e-01
-7.04367876e-01 3.12799007e-01 -3.62464160e-01 -1.00752242e-01
-1.71802133e-01 5.17115474e-01 -7.73295879e-01 3.34300667e-01
7.92610273e-02 1.53446436e+00 -1.17881513e+00 -2.76992917e-01
1.57484576e-01 1.28718451e-01 -1.78899005e-01 -2.10383773e-01
-2.28253722e-01 4.29842621e-01 -4.74037558e-01 8.76544595e-01
2.92722255e-01 -1.30024269e-01 3.82932201e-02 -1.66087374e-01
-2.19745189e-01 7.22000420e-01 -7.37707973e-01 1.59516656e+00
-2.33743712e-01 8.43750536e-01 -6.14642262e-01 -1.51930320e+00
7.76253104e-01 3.41138512e-01 5.03718108e-02 -7.40510046e-01
3.90132586e-03 1.37289032e-01 3.72098804e-01 -7.32845306e-01
3.38779807e-01 -5.20317912e-01 -9.03052837e-02 2.51237035e-01
2.69387756e-03 -4.66375798e-01 1.66299492e-01 4.16669428e-01
7.54833817e-01 1.90126076e-02 1.14401174e+00 -3.68021488e-01
8.35908830e-01 1.34497181e-01 5.19256175e-01 9.58468020e-01
-3.97978306e-01 2.67745815e-02 5.33470869e-01 -7.03154743e-01
-4.85411137e-01 -1.20913804e+00 -1.91686466e-01 1.43521810e+00
3.36498767e-01 -6.75692022e-01 -3.05161715e-01 -6.95411205e-01
-2.25634366e-01 9.50706542e-01 -1.07736540e+00 -1.44463509e-01
-4.46773916e-01 -1.25878477e+00 4.34342563e-01 1.21568131e+00
4.33504462e-01 -1.20606518e+00 -2.07816690e-01 -3.60006452e-01
-7.75722116e-02 -5.94780982e-01 2.95208275e-01 3.42511207e-01
-1.38632929e+00 -7.56298721e-01 -2.25689888e-01 -9.25255299e-01
2.24612236e-01 2.66236186e-01 1.26272476e+00 1.80176929e-01
-1.57420695e-01 6.27824783e-01 -2.29610875e-01 -6.34928584e-01
-6.48380369e-02 -1.27526253e-01 3.42891604e-01 -7.88414776e-01
9.59268808e-01 -7.92893171e-01 -1.94003344e-01 5.56675047e-02
-4.44602996e-01 -5.70946932e-01 2.53648132e-01 8.13168585e-01
4.26462650e-01 -1.42816558e-01 8.66497278e-01 -8.73610735e-01
8.49796116e-01 -4.42657471e-01 -4.39672053e-01 2.77683496e-01
-9.97072339e-01 1.80834159e-02 3.10694665e-01 -3.35895002e-01
-9.45878327e-01 -2.86858767e-01 -1.03099056e-01 2.98134506e-01
7.79079366e-03 7.28465974e-01 -1.74506828e-01 -2.40642577e-01
9.94858325e-01 1.49386242e-01 -5.91896951e-01 -5.38716912e-01
7.86396563e-01 4.63908285e-01 5.45579135e-01 -5.23601890e-01
4.28276241e-01 5.02261519e-01 4.85676795e-01 -8.86147618e-01
-1.00457811e+00 -3.30654085e-01 -8.22857857e-01 8.41146186e-02
5.06930709e-01 -3.20090145e-01 -6.32043123e-01 3.17844748e-01
-1.28578818e+00 -2.68007874e-01 -5.03838122e-01 4.16709393e-01
-8.63256633e-01 4.76181924e-01 -7.18884706e-01 -7.69526899e-01
-1.37216136e-01 -7.83633173e-01 3.80933732e-01 4.35810834e-01
-3.91299009e-01 -1.55396128e+00 2.98038483e-01 -8.98324847e-02
2.67273128e-01 -2.43730232e-01 1.01676011e+00 -7.69246101e-01
-3.39708477e-01 3.16281408e-01 5.72164617e-02 5.97893400e-03
2.93926477e-01 -1.12804160e-01 -9.98942912e-01 1.78644195e-01
3.02713722e-01 -1.28134295e-01 1.33525681e+00 5.64886749e-01
1.27452302e+00 6.63006231e-02 -7.91916132e-01 7.00624228e-01
1.37958968e+00 5.46030521e-01 4.90870357e-01 4.28156137e-01
5.96884668e-01 8.58489811e-01 6.12732649e-01 6.47098571e-03
4.39431697e-01 4.61603761e-01 4.23945367e-01 2.10541889e-01
-3.95764440e-01 1.29330367e-01 2.41550338e-02 9.63619053e-01
-6.45519197e-01 -4.39270943e-01 -1.01008403e+00 6.91730559e-01
-1.90659750e+00 -9.93737102e-01 -2.61817425e-01 1.64203453e+00
9.07168210e-01 -2.29337811e-01 2.47459546e-01 2.24804789e-01
8.40894759e-01 -9.03872773e-02 -5.14421880e-01 -8.28439713e-01
-7.67804265e-01 2.74724871e-01 3.96931320e-01 9.72234368e-01
-9.74135816e-01 1.41870940e+00 8.50904179e+00 7.29559422e-01
-1.02075148e+00 7.65566900e-02 4.32392918e-02 6.73604757e-02
-4.79519486e-01 5.29670864e-02 -9.12936747e-01 -2.99330175e-01
7.10898280e-01 -2.66414434e-01 3.77553791e-01 7.34555185e-01
-1.08846806e-01 5.59025630e-02 -1.33848822e+00 1.19601595e+00
3.44062746e-01 -1.50026035e+00 5.84949911e-01 -2.77114898e-01
7.85786152e-01 2.32736906e-03 2.71055847e-01 3.18748146e-01
2.80722678e-01 -1.24527669e+00 1.91763252e-01 3.84982854e-01
3.22191179e-01 -5.02196789e-01 4.30960119e-01 8.00821632e-02
-1.13931322e+00 -3.83415282e-01 -5.83925009e-01 -8.42626810e-01
1.73535541e-01 2.75618553e-01 -1.94419548e-01 6.57096028e-01
6.43325806e-01 1.11907029e+00 -4.03463364e-01 1.18876219e+00
-1.00644791e+00 4.31049079e-01 1.86700940e-01 -4.27708328e-01
8.49573687e-02 3.55461575e-02 7.62719333e-01 1.44415331e+00
-2.80887783e-01 5.36325634e-01 -1.79195940e-01 7.45219886e-01
2.03075916e-01 1.59961488e-02 -5.40078163e-01 -1.55913606e-02
4.69809383e-01 8.43659282e-01 -6.54628456e-01 -4.41552818e-01
-6.84124708e-01 6.56954050e-01 3.80394489e-01 5.12786210e-01
-7.52869546e-01 -4.30249631e-01 5.59605300e-01 -4.70104456e-01
2.29829907e-01 -2.89141119e-01 -8.63159359e-01 -9.95883226e-01
-3.66131693e-01 -6.10401869e-01 9.24211979e-01 -7.65593886e-01
-1.56061637e+00 5.16586602e-01 4.11478043e-01 -5.43545783e-01
-2.64187694e-01 -1.04307199e+00 -4.11202312e-01 8.43156517e-01
-1.42771590e+00 -9.37532425e-01 -9.37890932e-02 5.48013389e-01
5.36681890e-01 -2.51285344e-01 1.07332110e+00 -1.73127607e-01
-4.49580550e-01 5.30976593e-01 -4.13038373e-01 -7.63785765e-02
3.75888020e-01 -1.43111515e+00 6.87512755e-01 6.10908985e-01
3.27473313e-01 1.09616888e+00 8.83388758e-01 -7.24353671e-01
-1.12235773e+00 -8.05047452e-01 1.26137280e+00 -7.02889025e-01
8.19107175e-01 -1.28891869e-02 -8.05823982e-01 1.22772813e+00
5.67109823e-01 -2.81175166e-01 7.13428497e-01 5.30466795e-01
-4.64559525e-01 -7.44812191e-02 -8.99704218e-01 8.44080567e-01
1.62348461e+00 -3.63340825e-01 -1.37990808e+00 4.02881414e-01
7.22169995e-01 -2.78976653e-02 -4.16493982e-01 4.40585494e-01
5.57400763e-01 -9.06356812e-01 8.87825131e-01 -1.19934416e+00
-8.12873319e-02 -2.02843040e-01 -3.10345322e-01 -1.30709326e+00
-7.52873898e-01 -8.06189120e-01 -2.55528569e-01 8.31198573e-01
5.33332407e-01 -9.55237925e-01 3.93432498e-01 3.98974180e-01
-2.15290383e-01 -1.05185187e+00 -8.25405478e-01 -1.20007896e+00
6.81126714e-01 -8.68750751e-01 3.13727796e-01 1.02610838e+00
8.03157270e-01 9.05699193e-01 5.60970642e-02 3.17917243e-02
4.21953738e-01 2.32960984e-01 2.61559308e-01 -1.03577626e+00
-1.13711998e-01 -8.64869416e-01 -7.77549028e-01 -1.56822944e+00
1.26540139e-01 -9.90059972e-01 -4.95678991e-01 -1.73669147e+00
4.21115309e-01 6.78767276e-04 -4.22095060e-01 4.01399374e-01
-1.35404095e-02 4.23262894e-01 -2.27392912e-01 -1.55986017e-02
-4.03866440e-01 3.88198614e-01 1.25904524e+00 -2.74006963e-01
-3.15545768e-01 1.17128231e-01 -1.03200448e+00 9.18876052e-01
1.31948984e+00 -3.34564000e-01 -5.94278574e-01 -4.62418705e-01
7.61898637e-01 -4.70994711e-01 3.58170003e-01 -5.15490472e-01
2.57276654e-01 -5.78226447e-01 2.17645526e-01 -9.20180500e-01
3.62423807e-01 -2.66605675e-01 -6.85398102e-01 6.32721961e-01
-3.77629310e-01 6.13157392e-01 3.66157293e-01 5.03149807e-01
-4.26366068e-02 -4.63047445e-01 6.69771016e-01 -3.80864382e-01
-1.28332984e+00 1.00456506e-01 -5.41673481e-01 8.17132890e-02
1.02589095e+00 -2.25360729e-02 -3.28765124e-01 -4.22703743e-01
-9.95577872e-01 2.53003538e-01 -1.27075553e-01 5.41746736e-01
6.33202553e-01 -1.08189225e+00 -4.15951818e-01 -3.17466676e-01
1.05494380e-01 -6.39511764e-01 1.60125606e-02 8.34514976e-01
-1.00002125e-01 1.14979565e+00 -9.63102207e-02 -4.47752476e-01
-1.09049165e+00 1.45877051e+00 5.85803151e-01 4.48471397e-01
-3.57265174e-01 1.31275153e+00 6.79145932e-01 -8.62068012e-02
5.25623620e-01 -5.98128796e-01 -7.50914872e-01 -1.11145660e-01
7.70728111e-01 5.85605025e-01 -1.87748834e-01 -4.99977618e-01
-8.44076395e-01 9.67531085e-01 3.17044854e-02 -2.33445302e-01
9.81117845e-01 -4.78269160e-01 -5.28279305e-01 1.23980558e+00
7.98560262e-01 -2.25870628e-02 -3.54780197e-01 -3.00667197e-01
-5.09975664e-03 3.22389454e-02 -2.35238686e-01 -9.69732106e-01
-8.13920200e-01 1.13240170e+00 4.46535200e-01 4.40402001e-01
1.33297324e+00 6.71337247e-01 3.69948119e-01 7.30987072e-01
3.32744747e-01 -9.11156178e-01 -1.51363358e-01 7.76551187e-01
1.02322769e+00 -8.63616049e-01 3.81834239e-01 -7.50266433e-01
-5.08463264e-01 1.32929695e+00 5.73516071e-01 -4.33806628e-01
8.79186809e-01 2.75198400e-01 1.77330954e-03 -4.81985211e-01
-8.73240352e-01 -5.18208086e-01 4.35006648e-01 1.21712768e+00
6.88845098e-01 -4.33857739e-02 -9.07698870e-01 7.36935019e-01
-5.45622766e-01 -1.93170875e-01 4.61501628e-01 6.47014916e-01
-5.04703343e-01 -1.33221900e+00 -1.44877851e-01 -8.19606632e-02
-3.21714610e-01 -5.90168059e-01 -9.71575856e-01 1.08784246e+00
3.77799273e-01 1.14406168e+00 1.14131859e-02 -4.12366211e-01
1.74024120e-01 2.87739336e-02 1.21646786e+00 -7.82121062e-01
-3.79345804e-01 -3.59108716e-01 -5.10643283e-03 -5.11248171e-01
-1.03437054e+00 -7.65027344e-01 -1.53583622e+00 -1.67611703e-01
-3.65067631e-01 1.26652732e-01 5.18792987e-01 1.24193323e+00
-1.71501383e-01 4.87593651e-01 5.15397452e-03 -2.48620003e-01
-3.41906041e-01 -8.68432343e-01 -5.25503278e-01 -1.51834384e-01
5.81032097e-01 -9.50516522e-01 -4.83502924e-01 5.68928905e-02] | [9.358468055725098, 6.471621990203857] |
aa06379c-0e90-4ceb-8018-72100e92508d | protein-structure-prediction-until-casp15 | 2212.07702 | null | https://arxiv.org/abs/2212.07702v1 | https://arxiv.org/pdf/2212.07702v1.pdf | Protein Structure Prediction until CASP15 | In Dec 2020, the results of AlphaFold2 were presented at CASP14, sparking a revolution in the field of protein structure predictions. For the first time, a purely computational method could challenge experimental accuracy for structure prediction of single protein domains. The code of AlphaFold2 was released in the summer of 2021, and since then, it has been shown that it can be used to accurately predict the structure of most (ordered) proteins and many protein-protein interactions. It has also sparked an explosion of development in the field, improving AI-based methods to predict protein complexes, disordered regions, and protein design. Here I will review some of the inventions sparked by the release of AlphaFold. | ['Arne Elofsson'] | 2022-12-15 | null | null | null | null | ['protein-design'] | ['medical'] | [ 2.25370646e-01 4.33970720e-01 -1.30708799e-01 -2.94010013e-01
-3.82377267e-01 -6.65133059e-01 3.17046076e-01 5.42127490e-01
-1.20890290e-01 1.25396609e+00 2.14345098e-01 -5.77253997e-01
1.05150163e-01 -3.40104073e-01 -7.19233811e-01 -7.51014352e-01
-8.14448074e-02 7.49189377e-01 3.84309828e-01 -3.59478533e-01
3.92307162e-01 5.82085729e-01 -1.30121398e+00 5.28882861e-01
9.86008942e-01 6.02191031e-01 5.25259435e-01 3.05938095e-01
-1.54185042e-01 4.68948156e-01 -3.07807624e-01 -3.31480563e-01
-1.32060051e-01 -7.87135422e-01 -9.55200613e-01 -4.73957956e-01
-1.30964145e-01 2.40169689e-01 6.85719848e-02 4.83784705e-01
3.55833858e-01 -2.07647324e-01 6.49982095e-01 -5.86914301e-01
-3.82292479e-01 -4.78037447e-02 -2.91114271e-01 -2.59694010e-01
6.46823764e-01 4.24959898e-01 1.20418739e+00 -8.85113657e-01
1.15531206e+00 9.62884486e-01 6.68404460e-01 5.67351639e-01
-1.41922963e+00 -3.08770418e-01 -2.92666852e-01 4.29379821e-01
-9.17461872e-01 -1.35972992e-01 2.43795678e-01 -5.34328282e-01
1.71849036e+00 1.26339272e-01 8.64678502e-01 7.94816971e-01
6.60468400e-01 4.99743730e-01 9.40940976e-01 -3.04909080e-01
3.81470144e-01 -4.35621589e-01 -7.13911876e-02 6.27023995e-01
1.74811140e-01 1.05560243e-01 -5.88016152e-01 -8.86284173e-01
6.47037551e-02 -2.25366071e-01 -1.21416889e-01 -5.42494774e-01
-1.05054247e+00 8.52849483e-01 7.83956125e-02 2.35674977e-01
-5.09559870e-01 -4.97054219e-01 3.80503923e-01 3.36112231e-02
1.72662914e-01 9.45652902e-01 -7.86536217e-01 -4.61777478e-01
-6.09376371e-01 7.62465477e-01 8.62268567e-01 3.21143985e-01
5.22945464e-01 -5.30464947e-01 5.91134965e-01 5.66857100e-01
9.66992080e-02 2.59099603e-02 3.02079707e-01 -7.38977790e-01
-1.29084334e-01 5.86839259e-01 4.95767146e-01 -5.35260558e-01
-7.43652046e-01 3.05054605e-01 -4.73701805e-01 2.43804663e-01
4.87943769e-01 -3.80128101e-02 -6.94233537e-01 1.34104538e+00
2.62662083e-01 -4.46883202e-01 -2.50593033e-02 7.57506847e-01
3.81639093e-01 6.17326379e-01 1.28278628e-01 -5.81392944e-01
9.52181041e-01 -4.84516203e-01 -4.96181637e-01 3.04750085e-01
6.57235861e-01 -8.20797324e-01 3.75493139e-01 9.00652945e-01
-1.00946677e+00 -2.89265156e-01 -1.17484081e+00 4.90598530e-02
-3.10330391e-01 -5.19638181e-01 8.19769025e-01 3.52612078e-01
-5.86978853e-01 7.31073201e-01 -8.63801956e-01 -5.79827845e-01
1.31992966e-01 4.95386302e-01 -5.77763915e-01 1.42496929e-01
-1.20010257e+00 1.40845835e+00 5.77716351e-01 -3.15086693e-01
-6.39485121e-01 -7.30384588e-01 -3.58355902e-02 -2.22502306e-01
1.54367775e-01 -4.83394623e-01 8.37294042e-01 -7.12461293e-01
-1.13484323e+00 1.07605779e+00 -4.71005380e-01 -6.35940135e-01
2.92413570e-02 -1.77724898e-01 -3.68301064e-01 2.54941452e-02
-1.63743213e-01 6.01607442e-01 4.43520471e-02 -6.43860519e-01
-4.82516140e-01 -5.42765498e-01 -3.28287363e-01 5.36381602e-02
5.56657791e-01 2.92636603e-01 3.10173452e-01 -3.51658493e-01
1.38797238e-01 -8.78876805e-01 -5.87013245e-01 -8.19422677e-02
-2.32641712e-01 -4.27824348e-01 3.47154528e-01 -5.97092628e-01
7.23546982e-01 -1.89320099e+00 4.37744141e-01 1.89242631e-01
4.56319690e-01 4.86577690e-01 2.16212913e-01 1.10508740e+00
-5.57300150e-01 9.45458412e-02 -3.83870184e-01 6.37229264e-01
-4.85730469e-01 6.51706457e-02 -2.55790383e-01 5.26048958e-01
2.24665761e-01 7.71837831e-01 -7.60350287e-01 1.56853944e-02
9.96640474e-02 4.68391001e-01 -5.95984042e-01 3.01761568e-01
-6.42464817e-01 5.79932749e-01 -3.92837405e-01 3.96228313e-01
7.31632173e-01 -5.25146067e-01 9.96376991e-01 -7.80223459e-02
-4.70168442e-01 5.48392594e-01 -3.87045771e-01 1.26152873e+00
4.54254508e-01 2.73989439e-01 -1.44719228e-01 -8.90304327e-01
9.83631909e-01 4.05360013e-01 1.13378012e+00 -3.26094687e-01
-2.76534140e-01 1.64357945e-01 4.43453640e-01 -1.36912718e-01
-7.47101940e-03 -3.58189702e-01 2.60528207e-01 3.71777326e-01
-1.14349693e-01 -7.69232884e-02 5.81339225e-02 1.66031510e-01
1.23551130e+00 5.70696712e-01 7.47375071e-01 -2.57819414e-01
4.16605502e-01 7.55637467e-01 6.19465709e-01 1.23043820e-01
-1.68433592e-01 4.79398847e-01 5.06786346e-01 -1.17256165e+00
-1.39641619e+00 -8.13785315e-01 -3.55424076e-01 8.29111934e-01
8.60103443e-02 -1.00632346e+00 -8.86540890e-01 -3.44490290e-01
8.19913596e-02 2.01644555e-01 -3.33345592e-01 -2.68109322e-01
-5.60927033e-01 -1.10158658e+00 4.06912893e-01 -1.65345699e-01
8.95558298e-02 -1.14267182e+00 -5.83683252e-01 6.03952706e-01
-7.50284120e-02 -7.51817167e-01 -1.24203665e-02 7.55231142e-01
-8.26601982e-01 -1.65441179e+00 -4.90010500e-01 -6.54125690e-01
2.48683557e-01 -6.17527105e-02 9.30951893e-01 8.47531706e-02
-6.58790886e-01 -4.27179456e-01 -1.84353590e-01 -4.81125712e-01
-5.66583395e-01 -1.07361928e-01 5.24052322e-01 -4.87249494e-01
8.34094644e-01 -7.85279691e-01 -4.70078051e-01 3.14696282e-01
-5.40817440e-01 3.63675535e-01 3.26001674e-01 9.55407321e-01
6.31329000e-01 -2.09252313e-01 5.55810690e-01 -7.91742980e-01
4.67118293e-01 -2.78817385e-01 -5.96120358e-01 4.00299639e-01
-9.46719050e-01 3.49220604e-01 5.94330490e-01 -6.02611899e-02
-8.04452062e-01 3.85993004e-01 -4.41639155e-01 4.58958209e-01
-1.91951856e-01 4.96619195e-01 -1.26918063e-01 -1.49631113e-01
6.72614217e-01 4.42424089e-01 3.53336006e-01 -9.45070326e-01
1.41036019e-01 4.84387547e-01 2.89923161e-01 -5.83460987e-01
3.12709630e-01 8.88132080e-02 1.75902620e-01 -1.03932142e+00
-4.36945707e-01 -2.08086357e-01 -5.57238877e-01 2.06400648e-01
8.72586370e-01 -5.58292150e-01 -1.29825389e+00 4.23546344e-01
-1.12531412e+00 6.79507665e-03 1.56394929e-01 2.01614007e-01
-7.57125735e-01 6.82280004e-01 -3.50880325e-01 -4.60540414e-01
-1.82766080e-01 -1.20628667e+00 5.45318842e-01 5.78704067e-02
-6.00357831e-01 -4.76847202e-01 5.32251060e-01 3.44270170e-01
8.84538665e-02 3.77967179e-01 1.53818560e+00 -7.44604707e-01
-2.70919532e-01 1.46017000e-01 5.81037775e-02 1.07087620e-01
1.55141681e-01 2.10807413e-01 -4.13948357e-01 7.17606675e-03
-3.65144819e-01 -5.50959051e-01 6.72301590e-01 2.60208279e-01
6.42894745e-01 -1.72485024e-01 -7.31787920e-01 3.73099953e-01
1.18517125e+00 5.42565644e-01 6.94902897e-01 3.34924906e-01
3.38968813e-01 5.29483438e-01 7.91610301e-01 2.70609081e-01
-1.90623477e-01 9.02451694e-01 4.06751513e-01 3.71074826e-02
2.90446639e-01 -2.77135998e-01 3.02293096e-02 3.37639600e-01
-5.08813739e-01 2.06459444e-02 -1.24941719e+00 4.46312092e-02
-1.84141207e+00 -9.13144886e-01 -2.26039603e-01 2.00207806e+00
1.27613652e+00 2.24219695e-01 4.24649835e-01 -3.76930863e-01
4.78367120e-01 -3.27874810e-01 -1.04041862e+00 -5.19491196e-01
-5.90563238e-01 3.98051828e-01 3.02379280e-01 5.83555102e-01
-7.63370752e-01 9.80866849e-01 8.65740871e+00 5.08066177e-01
-8.27920377e-01 -3.97887349e-01 7.09584773e-01 1.70490652e-01
-1.11992300e-01 3.73541832e-01 -7.82835305e-01 4.90797013e-01
1.13461220e+00 -3.61822635e-01 2.22365111e-01 7.25473940e-01
4.09658551e-01 -1.98602989e-01 -7.15375602e-01 6.73999846e-01
-4.89093006e-01 -1.82155931e+00 -9.91137419e-03 3.54629666e-01
5.07489979e-01 3.26989800e-01 -4.53151137e-01 -2.27904066e-01
3.20892662e-01 -1.26452982e+00 1.27289426e-02 5.48713148e-01
4.77879614e-01 -9.59067166e-01 4.76859480e-01 4.47311133e-01
-6.47855043e-01 1.73735946e-01 -5.18824339e-01 -2.00747997e-01
1.60840124e-01 7.69390225e-01 -1.03743100e+00 2.77832985e-01
4.27687079e-01 6.69413865e-01 -2.10749984e-01 9.13980365e-01
8.20104256e-02 6.26418412e-01 -2.46447727e-01 5.29781030e-03
5.61151095e-02 -5.13912320e-01 4.58058774e-01 7.38435805e-01
-4.22649235e-01 6.04966283e-01 2.18352884e-01 6.98428810e-01
2.37411529e-01 9.30606350e-02 -2.28460282e-01 -4.25244659e-01
1.84504986e-01 7.62691796e-01 -2.61225820e-01 -1.00702576e-01
-3.27451527e-01 7.59636581e-01 2.60073513e-01 2.01755434e-01
-7.54480302e-01 -1.79089397e-01 1.15159547e+00 4.65144962e-01
1.54102132e-01 -3.60112876e-01 1.74172133e-01 -8.04532468e-01
-2.81191617e-01 -1.29957712e+00 1.10005969e-02 -7.67825902e-01
-1.04236329e+00 3.06897193e-01 -3.33081037e-01 -4.14197713e-01
-1.71779871e-01 -1.03025651e+00 -1.62703678e-01 9.16085660e-01
-7.35737085e-01 -5.67468226e-01 5.48554301e-01 -1.88358307e-01
1.60271838e-01 -2.83461988e-01 1.25612175e+00 -3.27544510e-01
-2.31598526e-01 7.88809657e-02 8.13104630e-01 -5.31172693e-01
9.82956588e-01 -1.10236454e+00 7.97477305e-01 5.58941253e-02
-3.48391235e-01 8.05181444e-01 1.15535736e+00 -9.32616651e-01
-1.42254889e+00 -4.29982513e-01 1.00768030e+00 -3.74835432e-01
3.82330775e-01 -3.32034081e-01 -9.10880387e-01 3.14311206e-01
-1.04377456e-01 -6.80347621e-01 8.76890540e-01 -9.04992223e-03
-2.30849460e-01 4.48513448e-01 -1.18017137e+00 2.61697531e-01
8.32926691e-01 -2.94079155e-01 -6.48282766e-01 6.49003863e-01
7.18866527e-01 -2.59093344e-01 -1.03091002e+00 5.05047262e-01
8.78862560e-01 -1.16209865e+00 9.65163946e-01 -1.17464483e+00
9.30873826e-02 -3.23976099e-01 3.10216378e-03 -7.48490095e-01
-5.42564273e-01 -1.03966403e+00 -2.99155414e-01 3.19745898e-01
6.26844883e-01 -5.60845137e-01 1.15625262e+00 5.93445122e-01
-5.94112575e-02 -1.17243421e+00 -9.70841944e-01 -4.91195321e-01
3.08459014e-01 3.44395161e-01 3.50991219e-01 6.20481610e-01
6.70681953e-01 4.02059644e-01 -1.65254250e-01 -5.61559260e-01
2.60841250e-01 1.10756002e-01 5.98118722e-01 -1.55753028e+00
-4.99537230e-01 -1.43714026e-01 -5.37114084e-01 -1.09013176e+00
-2.49328405e-01 -5.82841337e-01 -2.88212985e-01 -1.15896487e+00
5.29413640e-01 1.99089795e-01 -8.97391662e-02 6.96689010e-01
-9.83100682e-02 4.96231169e-02 -2.08824709e-01 4.33927208e-01
-2.61216044e-01 2.12801501e-01 9.70708132e-01 4.10442650e-02
2.68312991e-02 -1.22308969e-01 -7.24216938e-01 4.51265514e-01
9.05911148e-01 -3.62133265e-01 4.32865247e-02 6.28774226e-01
7.02732325e-01 -2.43896589e-01 -6.53034681e-03 -9.50294375e-01
-4.18639928e-01 -5.49850285e-01 4.67196792e-01 -7.37363458e-01
3.48399848e-01 -6.39262676e-01 5.69802463e-01 8.37686121e-01
-1.12062126e-01 5.41228503e-02 1.94716886e-01 4.48514819e-01
-2.90399715e-02 9.68145207e-02 9.98897910e-01 -2.81372458e-01
-4.00019258e-01 -3.58452788e-03 -8.22384238e-01 -1.22458562e-01
1.08323228e+00 -3.43635470e-01 -2.36284524e-01 -3.68418545e-02
-1.04272330e+00 -1.30387619e-02 9.89284158e-01 1.89762190e-01
4.28949922e-01 -6.03169143e-01 -4.51952577e-01 7.91543052e-02
5.29728131e-03 -5.19955575e-01 -1.16528861e-01 7.28248000e-01
-9.35283065e-01 1.02018130e+00 -4.60920781e-01 -4.68020350e-01
-1.57312942e+00 7.83723354e-01 4.76268202e-01 -3.42304617e-01
-5.48301995e-01 4.20467138e-01 2.62859762e-01 -1.47087261e-01
-2.61902750e-01 2.89634347e-01 1.09707519e-01 -4.69100803e-01
6.60823584e-01 1.36010394e-01 1.78284645e-02 -5.67921281e-01
-6.94782257e-01 3.43603492e-01 -4.34862047e-01 2.53018260e-01
1.69054914e+00 2.08154678e-01 -3.25530231e-01 2.75822252e-01
8.75666261e-01 -3.43177229e-01 -1.16026771e+00 1.94249526e-01
3.70545000e-01 2.63286263e-01 -4.84493315e-01 -1.39918685e+00
-9.90618318e-02 5.87930977e-01 5.79648912e-01 -1.35624424e-01
8.25581908e-01 8.57469365e-02 9.39185381e-01 6.17447615e-01
7.01840341e-01 -8.08146000e-01 -2.80834556e-01 6.13008916e-01
5.96705079e-01 -9.99210417e-01 3.09100837e-01 -2.16574296e-01
-5.02095997e-01 1.17125130e+00 3.61797601e-01 1.37271136e-01
3.66258293e-01 2.37099528e-01 -3.16436708e-01 -3.74923468e-01
-9.82129455e-01 2.28177965e-01 6.31042346e-02 8.77365112e-01
1.10056746e+00 -4.04204540e-02 -7.84133255e-01 5.87004840e-01
1.32451028e-01 1.12384714e-01 2.21529484e-01 1.06308293e+00
-1.02248299e+00 -1.89693654e+00 -2.72276938e-01 1.81337327e-01
-5.41028678e-01 -2.62281090e-01 -1.33163929e+00 4.91589218e-01
1.52411520e-01 7.16306210e-01 -5.26293814e-01 -1.08949356e-01
1.56112164e-01 6.36016667e-01 7.80962288e-01 -4.08342063e-01
-6.22705519e-01 -2.33317912e-01 8.87553170e-02 -6.24318779e-01
-2.37821862e-01 -5.08773565e-01 -1.65896833e+00 -5.35667837e-01
-3.67331564e-01 1.00037658e+00 4.62283343e-01 8.09850812e-01
9.28102791e-01 -4.72637787e-02 3.01045239e-01 -6.31095588e-01
-2.97377348e-01 -8.52747560e-01 -6.13834858e-01 -7.91222416e-03
1.59692809e-01 -5.59630930e-01 5.01337759e-02 9.96535718e-02] | [4.720815181732178, 5.364953994750977] |
252505ce-b4c0-47a9-b523-c37f6b27fc58 | flipnerf-flipped-reflection-rays-for-few-shot | 2306.17723 | null | https://arxiv.org/abs/2306.17723v1 | https://arxiv.org/pdf/2306.17723v1.pdf | FlipNeRF: Flipped Reflection Rays for Few-shot Novel View Synthesis | Neural Radiance Field (NeRF) has been a mainstream in novel view synthesis with its remarkable quality of rendered images and simple architecture. Although NeRF has been developed in various directions improving continuously its performance, the necessity of a dense set of multi-view images still exists as a stumbling block to progress for practical application. In this work, we propose FlipNeRF, a novel regularization method for few-shot novel view synthesis by utilizing our proposed flipped reflection rays. The flipped reflection rays are explicitly derived from the input ray directions and estimated normal vectors, and play a role of effective additional training rays while enabling to estimate more accurate surface normals and learn the 3D geometry effectively. Since the surface normal and the scene depth are both derived from the estimated densities along a ray, the accurate surface normal leads to more exact depth estimation, which is a key factor for few-shot novel view synthesis. Furthermore, with our proposed Uncertainty-aware Emptiness Loss and Bottleneck Feature Consistency Loss, FlipNeRF is able to estimate more reliable outputs with reducing floating artifacts effectively across the different scene structures, and enhance the feature-level consistency between the pair of the rays cast toward the photo-consistent pixels without any additional feature extractor, respectively. Our FlipNeRF achieves the SOTA performance on the multiple benchmarks across all the scenarios. | ['Nojun Kwak', 'Yeonjin Chang', 'Seunghyeon Seo'] | 2023-06-30 | null | null | null | null | ['depth-estimation', 'novel-view-synthesis'] | ['computer-vision', 'computer-vision'] | [ 6.58392832e-02 -2.77886748e-01 1.23797692e-01 -5.04642963e-01
-8.66104484e-01 -2.12024823e-01 5.02383113e-01 -4.18730319e-01
-1.65574417e-01 7.01814413e-01 2.66761929e-01 2.12279588e-01
-1.90751731e-01 -1.04890668e+00 -1.01694536e+00 -9.45259154e-01
4.47675228e-01 3.50423336e-01 2.59615034e-01 -2.15623140e-01
3.48894238e-01 5.16115427e-01 -1.71677196e+00 2.05074608e-01
9.75870311e-01 1.31744695e+00 3.82352203e-01 5.54417789e-01
-1.31992623e-01 8.15083146e-01 -5.77307381e-02 -3.02512616e-01
5.15354991e-01 -2.95719683e-01 -1.86849847e-01 2.95439586e-02
9.26470220e-01 -8.30279887e-01 -4.17251408e-01 9.79713500e-01
5.32943189e-01 2.81129658e-01 5.53167701e-01 -8.33727121e-01
-2.81161129e-01 3.36659397e-03 -9.68799293e-01 -1.35245025e-01
1.93621427e-01 3.55639845e-01 8.29279482e-01 -1.10744035e+00
5.89592457e-01 1.21374738e+00 4.71693814e-01 3.95112723e-01
-8.40270221e-01 -7.46557295e-01 2.60371745e-01 2.56396472e-01
-1.29178882e+00 -4.93278831e-01 9.49521542e-01 -1.32932961e-01
7.76321411e-01 2.13345900e-01 7.13601470e-01 8.14558268e-01
4.30893868e-01 4.89998192e-01 1.01924133e+00 -1.04511566e-01
1.71054676e-01 1.91853359e-01 -1.39216453e-01 8.83511722e-01
7.18958452e-02 1.69738829e-01 -6.90228701e-01 1.81207791e-01
1.09469962e+00 3.25197577e-01 -8.82161260e-01 -3.04079652e-01
-1.07991314e+00 5.48935831e-01 6.61711633e-01 -2.62409776e-01
-2.87370145e-01 3.13288152e-01 2.04989374e-01 -1.26634642e-01
6.42487109e-01 1.77793235e-01 -5.02020299e-01 1.12089105e-02
-6.53741777e-01 1.73752815e-01 3.77499670e-01 9.04652297e-01
1.08720505e+00 1.57696307e-01 -1.64240748e-01 7.83133209e-01
5.01047015e-01 9.23443973e-01 1.55411065e-01 -1.16173112e+00
6.87703431e-01 7.86352754e-01 1.48293108e-01 -1.18239355e+00
-2.29422003e-01 -5.37644982e-01 -1.02183628e+00 3.48809749e-01
2.42285490e-01 1.70707017e-01 -9.83242452e-01 1.53425062e+00
6.29359365e-01 2.60385782e-01 -3.99022773e-02 1.05363274e+00
7.53566206e-01 1.05215180e+00 -5.93586445e-01 -2.55278349e-01
1.20054996e+00 -1.02450037e+00 -4.70566541e-01 -2.07162619e-01
2.08672419e-01 -8.04136693e-01 1.04507589e+00 7.39931524e-01
-1.20305693e+00 -4.44125295e-01 -1.12526762e+00 -4.83163744e-01
1.50863528e-01 1.99284982e-02 8.32839847e-01 3.10113013e-01
-7.73396611e-01 5.70780218e-01 -6.26726449e-01 2.22394764e-01
6.08341396e-01 1.47230253e-01 -3.75237703e-01 -7.94111371e-01
-8.76414120e-01 4.42343861e-01 1.32839885e-02 3.64667267e-01
-8.18539262e-01 -1.08530140e+00 -7.76476622e-01 1.73296556e-01
5.11960506e-01 -9.97343838e-01 7.93962955e-01 -6.71315134e-01
-1.51906073e+00 3.64983529e-01 -1.27838969e-01 9.17074382e-02
6.08714938e-01 -3.09409410e-01 -1.03092939e-01 3.13441992e-01
7.12689832e-02 6.40158892e-01 9.30602551e-01 -1.51336622e+00
-6.95070207e-01 -5.87411702e-01 4.11510617e-02 6.51015103e-01
-2.15522632e-01 -6.66994572e-01 -5.85832298e-01 -3.74923259e-01
4.62423712e-01 -5.32233894e-01 -1.51808769e-01 4.43619043e-01
-1.99918330e-01 1.40562072e-01 6.81346953e-01 -6.56125546e-01
7.87427008e-01 -2.11290646e+00 1.86115578e-01 8.19348991e-02
3.14308852e-01 -1.70292646e-01 1.30377531e-01 7.03684241e-02
2.23683596e-01 -3.92946780e-01 -4.78836864e-01 -3.68529439e-01
-3.77894729e-01 8.25583562e-02 -3.53976399e-01 4.61681217e-01
-6.80279210e-02 6.59478068e-01 -9.07096207e-01 -3.18586409e-01
5.51108360e-01 8.18616867e-01 -7.06489027e-01 3.39652747e-01
-3.65223736e-01 5.58855593e-01 -5.10350943e-01 3.99737656e-01
1.29446769e+00 -2.84813493e-01 -2.87651330e-01 -5.62697172e-01
-1.79767326e-01 4.93939593e-02 -1.24883223e+00 2.05403471e+00
-9.07814085e-01 3.00213009e-01 -3.08920890e-02 -5.09141028e-01
8.48551869e-01 -8.72160494e-02 3.25967640e-01 -1.05583477e+00
-5.24504529e-03 1.32172897e-01 -4.31749821e-01 -3.59217197e-01
4.78623241e-01 -4.14635688e-02 3.12063873e-01 2.30597809e-01
-2.74789482e-01 -4.53628689e-01 -2.42510110e-01 2.61287302e-01
6.95874691e-01 4.59738553e-01 7.51770288e-02 -7.67192766e-02
6.80058956e-01 -2.46352196e-01 7.99638748e-01 3.01714808e-01
4.47176546e-01 1.03177118e+00 6.99938163e-02 -5.04848123e-01
-1.15685380e+00 -1.27033508e+00 -2.31876329e-01 5.06170988e-01
4.34650183e-01 -2.67728597e-01 -5.28520286e-01 -6.34864450e-01
-2.42508590e-01 7.27618396e-01 -3.16886991e-01 -7.28081986e-02
-6.95111990e-01 -6.34471416e-01 -9.23698023e-02 4.51358050e-01
9.75484431e-01 -5.48841774e-01 -6.64188325e-01 -1.02802254e-02
-3.19227606e-01 -9.84503090e-01 -5.06111085e-01 1.98376179e-02
-9.47892070e-01 -1.03222597e+00 -8.75182569e-01 -1.63254678e-01
8.32262456e-01 6.35617852e-01 1.16696322e+00 1.35998977e-02
-2.22554475e-01 2.28368491e-01 -9.92074609e-02 -5.52093647e-02
7.24142045e-03 -2.94369608e-01 -3.33859921e-01 1.19976409e-01
-2.33095080e-01 -7.53435671e-01 -1.22975612e+00 4.25976574e-01
-8.99920762e-01 6.10075235e-01 4.63433146e-01 9.20764625e-01
7.60279477e-01 3.77853103e-02 2.45269403e-01 -9.49387610e-01
-2.68086761e-01 -3.47862810e-01 -8.57535899e-01 6.66824207e-02
-7.12068617e-01 1.47383735e-01 9.24978971e-01 1.95466787e-01
-1.64446449e+00 -1.31128982e-01 -4.03721660e-01 -6.45664990e-01
1.83518931e-01 -8.37677121e-02 -3.52587163e-01 -9.30264145e-02
4.14186567e-01 4.14814711e-01 -2.78151274e-01 -3.36458892e-01
3.19602072e-01 3.74740988e-01 1.60846964e-01 -5.01466870e-01
7.17024565e-01 9.92188752e-01 3.81725520e-01 -7.84845889e-01
-1.04766786e+00 -4.34145719e-01 -2.99344301e-01 -4.07878071e-01
7.44803965e-01 -1.19875240e+00 -7.91508496e-01 6.82687700e-01
-1.22249734e+00 -9.01675746e-02 -2.35013247e-01 5.14675140e-01
-3.68641704e-01 4.80479598e-01 -4.11333472e-01 -7.61772752e-01
-5.44087172e-01 -1.35177386e+00 1.31617975e+00 3.79011512e-01
5.90973377e-01 -8.21588695e-01 -1.16974160e-01 5.24240434e-01
1.67569429e-01 9.89781246e-02 9.52116549e-01 5.06341040e-01
-1.34295499e+00 1.41375810e-01 -5.51427960e-01 6.70399189e-01
3.69283855e-02 -5.26053831e-03 -1.31485939e+00 -2.95558333e-01
4.57863182e-01 -2.40188435e-01 9.40125406e-01 6.01381540e-01
1.16613448e+00 -4.40527871e-02 -2.04191431e-01 1.29867387e+00
1.87851000e+00 6.75657168e-02 8.85185957e-01 4.41654697e-02
1.14952457e+00 5.16133845e-01 6.98920071e-01 5.76137722e-01
5.10010898e-01 5.78209341e-01 7.43491530e-01 1.25670567e-01
-2.67807066e-01 -3.55164826e-01 2.16947421e-01 1.09291410e+00
-2.41889104e-01 -5.11152804e-01 -4.56829876e-01 1.09200865e-01
-1.67814100e+00 -7.59916842e-01 -3.93566608e-01 2.35994124e+00
3.99790287e-01 -1.61720049e-02 -5.43718219e-01 -8.10237452e-02
3.62071186e-01 4.09985363e-01 -8.80246520e-01 -2.27515772e-02
-9.72189531e-02 1.59746096e-01 5.07095993e-01 4.58599359e-01
-3.54434818e-01 5.35723209e-01 4.88773775e+00 1.05397820e+00
-1.09665215e+00 -9.98601019e-02 8.98019314e-01 -1.80695042e-01
-9.20154691e-01 9.18938518e-02 -1.02153480e+00 3.95200729e-01
2.68983275e-01 3.49130899e-01 4.48540419e-01 7.61896789e-01
2.78204054e-01 -5.56115508e-01 -9.15801585e-01 1.25788116e+00
3.10473859e-01 -1.46716821e+00 3.25814039e-01 -1.96966361e-02
1.03530908e+00 3.30575146e-02 1.56284839e-01 -7.52290785e-02
-1.57828689e-01 -8.26791883e-01 6.14405990e-01 7.14670002e-01
1.05403805e+00 -9.89619136e-01 5.48376083e-01 3.01931202e-01
-1.18031752e+00 -4.54418175e-02 -5.39321423e-01 4.56333607e-02
3.30894202e-01 1.07919633e+00 -2.81973898e-01 1.07863545e+00
8.05621386e-01 9.72025573e-01 -2.96709597e-01 6.82502508e-01
-1.84193552e-01 1.98388726e-01 -3.82477582e-01 1.96069255e-01
1.14463724e-01 -6.08972549e-01 6.04500234e-01 4.27462488e-01
4.52530682e-01 1.57364830e-01 -9.93998647e-02 8.65145862e-01
-6.28690496e-02 7.35088959e-02 -4.16020542e-01 5.33848405e-01
2.20230564e-01 1.46506977e+00 -6.14021838e-01 -1.01439565e-01
-4.59445417e-01 1.03488684e+00 3.42121333e-01 3.59221667e-01
-8.78364682e-01 -1.36184171e-01 6.19616926e-01 4.58924711e-01
9.91506651e-02 -4.83185463e-02 -2.61484325e-01 -1.39450216e+00
2.95242786e-01 -5.59304535e-01 9.75123569e-02 -1.13897860e+00
-1.21427226e+00 4.98483896e-01 -1.07414991e-01 -1.40721619e+00
-1.32173290e-02 -5.30121505e-01 -5.09336472e-01 9.03298259e-01
-1.96706522e+00 -9.89854813e-01 -8.00701320e-01 5.43260694e-01
7.67935634e-01 1.80818453e-01 4.78750169e-01 4.40487802e-01
-4.07002568e-01 4.04927522e-01 2.81262159e-01 -3.83577913e-01
7.22668886e-01 -8.22190702e-01 2.40366846e-01 6.80512369e-01
-6.04734346e-02 3.41268480e-01 3.44408065e-01 -4.78107363e-01
-1.64121902e+00 -1.02975416e+00 -1.11199230e-01 -2.32677996e-01
-3.34326662e-02 -3.16116601e-01 -7.36026227e-01 3.75613242e-01
6.88720644e-02 1.79134279e-01 1.25490963e-01 -3.72526318e-01
-2.91652501e-01 -5.27014077e-01 -1.14986205e+00 4.83384997e-01
1.26421368e+00 -2.28513420e-01 -2.41153147e-02 2.43101507e-01
7.88914323e-01 -6.95279479e-01 -5.53006768e-01 5.29773474e-01
5.62973082e-01 -1.66188085e+00 1.04570770e+00 2.38794148e-01
9.57469940e-01 -4.44364697e-01 -3.23566109e-01 -1.21204567e+00
-5.62207028e-02 -3.85365576e-01 -5.06753698e-02 1.28420150e+00
-6.03069551e-03 -7.50005603e-01 7.97652304e-01 2.96172291e-01
-4.68032688e-01 -1.09318125e+00 -8.37558150e-01 -3.02501440e-01
-3.22565079e-01 -5.06839395e-01 6.03430510e-01 6.55921519e-01
-8.38544726e-01 3.73822004e-01 -4.96188015e-01 2.52091825e-01
8.76582980e-01 2.26418823e-01 9.47255552e-01 -9.63839829e-01
-4.85627115e-01 -2.51942407e-02 -2.03036472e-01 -1.57213092e+00
-2.56816387e-01 -7.72765875e-01 8.04684684e-02 -1.68942487e+00
2.62125582e-01 -6.42298102e-01 1.02537423e-01 -2.79304236e-01
-6.75275698e-02 1.88270986e-01 -6.64680228e-02 -3.53193171e-02
-1.69441953e-01 8.97149026e-01 1.76428366e+00 -9.54129547e-03
1.28756184e-02 -6.06056340e-02 -5.51930130e-01 9.52894628e-01
3.18215787e-01 -3.62525702e-01 -7.92074382e-01 -6.86388552e-01
7.30404675e-01 2.57739484e-01 3.93024892e-01 -1.07288730e+00
1.73205853e-01 -2.12721527e-01 6.85116947e-01 -8.74380827e-01
7.87590623e-01 -9.54160333e-01 1.99716076e-01 6.23598322e-02
1.66292652e-01 -3.93271558e-02 -1.64320841e-02 8.65632653e-01
-1.23751657e-02 -1.38190776e-01 9.31997895e-01 -3.58983755e-01
-7.47877598e-01 6.40657187e-01 4.47111666e-01 1.13842919e-01
9.35851574e-01 -4.07536626e-01 -2.35204831e-01 -2.59135365e-01
8.88890401e-02 2.11835876e-01 7.38065958e-01 1.45741418e-01
1.03709805e+00 -1.14785433e+00 -6.59279525e-01 5.20256996e-01
1.19820856e-01 8.65025043e-01 9.76051033e-01 5.40712953e-01
-8.28781247e-01 -7.37479255e-02 -6.25997409e-02 -8.93338501e-01
-9.02347744e-01 2.03521103e-01 2.77610272e-01 -1.66103929e-01
-1.04836428e+00 9.67236102e-01 8.13971281e-01 -3.76802117e-01
-7.37357372e-03 -3.04601610e-01 3.28888856e-02 -2.78826177e-01
6.47065699e-01 6.08912408e-01 1.02706060e-01 -3.84929389e-01
-1.38550669e-01 1.07270241e+00 -1.73103586e-01 2.61489097e-02
1.51059127e+00 -3.64541739e-01 -2.97186188e-02 5.14377594e-01
1.28046644e+00 3.23715448e-01 -1.76627743e+00 5.22635356e-02
-8.81836295e-01 -1.03342247e+00 5.14823198e-01 -6.42205894e-01
-1.38567424e+00 9.57509041e-01 5.72120488e-01 -5.22620857e-01
1.23086584e+00 -3.57921064e-01 1.27532339e+00 1.82014748e-01
6.44598305e-01 -8.68962288e-01 1.07243247e-01 3.34517688e-01
7.03435063e-01 -1.31561553e+00 4.54723448e-01 -6.54870927e-01
-4.76313561e-01 1.15260732e+00 7.29857266e-01 -2.06114799e-01
4.33892667e-01 1.98553190e-01 -1.57555848e-01 -2.70544946e-01
-6.06548905e-01 3.20206016e-01 2.63622582e-01 4.62219328e-01
1.12819828e-01 -3.79723310e-01 1.82546631e-01 1.05099805e-01
1.86320752e-01 -1.66693479e-01 4.97578025e-01 3.97175938e-01
-3.10973883e-01 -6.84691906e-01 -2.48247698e-01 6.39347136e-01
1.60620101e-02 -1.96617305e-01 3.68774563e-01 6.61393344e-01
1.25535131e-01 4.68869030e-01 1.02381475e-01 -1.74079552e-01
3.98579180e-01 -4.69717801e-01 7.40429044e-01 -5.29892087e-01
-2.56582469e-01 -4.10834067e-02 -2.44130060e-01 -7.40666151e-01
-2.81771511e-01 -3.42813015e-01 -1.37964642e+00 -3.36581737e-01
-5.43046355e-01 -1.53843611e-01 7.68706501e-01 8.00398648e-01
3.21718842e-01 5.73911667e-01 1.14543366e+00 -9.01065350e-01
-3.97564113e-01 -5.29411793e-01 -5.02160728e-01 2.48504385e-01
2.25713819e-01 -7.82718420e-01 -7.07150042e-01 -3.82771105e-01] | [9.28121566772461, -2.749241352081299] |
3f972feb-a227-4909-943a-8085e159ff10 | towards-accurate-and-reliable-change | 2305.19513 | null | https://arxiv.org/abs/2305.19513v1 | https://arxiv.org/pdf/2305.19513v1.pdf | Towards Accurate and Reliable Change Detection of Remote Sensing Images via Knowledge Review and Online Uncertainty Estimation | Change detection (CD) is an essential task for various real-world applications, such as urban management and disaster assessment. However, previous methods primarily focus on improving the accuracy of CD, while neglecting the reliability of detection results. In this paper, we propose a novel change detection network, called AR-CDNet, which is able to provide accurate change maps and generate pixel-wise uncertainty. Specifically, an online uncertainty estimation branch is constructed to model the pixel-wise uncertainty, which is supervised by the difference between predicted change maps and corresponding ground truth during the training process. Furthermore, we introduce a knowledge review strategy to distill temporal change knowledge from low-level features to high-level ones, thereby enhancing the discriminability of temporal difference features. Finally, we aggregate the uncertainty-aware features extracted from the online uncertainty estimation branch with multi-level temporal difference features to improve the accuracy of CD. Once trained, our AR-CDNet can provide accurate change maps and evaluate pixel-wise uncertainty without ground truth. Experimental results on two benchmark datasets demonstrate the superior performance of AR-CDNet in the CD task. The demo code for our work will be publicly available at \url{https://github.com/guanyuezhen/AR-CDNet}. | ['Xinzhong Zhu', 'Kun Sun', 'Weiying Xie', 'Xianju Li', 'Chang Tang', 'Zhenglai Li'] | 2023-05-31 | null | null | null | null | ['change-detection'] | ['computer-vision'] | [-1.05938047e-01 -2.29512319e-01 2.58819722e-02 -4.86504734e-01
-9.75786924e-01 -3.38140279e-01 7.35942364e-01 1.91599563e-01
-2.40685388e-01 9.72133815e-01 8.59494880e-02 -2.14297637e-01
-1.33037835e-01 -1.26049638e+00 -6.19388878e-01 -7.60004103e-01
-1.20659374e-01 -2.17793174e-02 4.74732727e-01 -1.24938890e-01
-8.18561688e-02 2.93788105e-01 -1.35838485e+00 -3.11318427e-01
1.37864161e+00 1.47914624e+00 8.36791098e-02 2.84371108e-01
2.48751804e-01 4.30279016e-01 -3.55319083e-01 7.32058883e-02
1.04771599e-01 -1.53980523e-01 -3.76480877e-01 -1.57722741e-01
1.09511711e-01 -5.89436710e-01 -5.70357442e-01 1.19213283e+00
4.57535595e-01 2.71607369e-01 5.40976048e-01 -1.23248076e+00
-3.78426939e-01 5.25534093e-01 -5.90202689e-01 5.75881004e-01
7.82956555e-02 4.33969110e-01 8.23570132e-01 -9.70858991e-01
4.63934958e-01 1.15018952e+00 6.02196813e-01 -3.82978879e-02
-8.94397318e-01 -8.94056976e-01 5.86718500e-01 4.12763178e-01
-1.62977755e+00 -3.50177318e-01 8.22049201e-01 -6.04393125e-01
5.08465469e-01 7.29032457e-02 5.00831246e-01 6.52055740e-01
2.84606904e-01 8.52376580e-01 9.82148111e-01 1.21060222e-01
3.30672562e-01 -3.42678845e-01 5.26927784e-03 4.93008196e-01
3.53512764e-01 3.82401913e-01 -3.73178095e-01 1.63023844e-01
7.12396801e-01 6.16042726e-02 -5.17635882e-01 -6.03340939e-02
-1.16633368e+00 4.97604132e-01 1.08150303e+00 2.85251498e-01
-4.87707347e-01 3.99174392e-01 2.48852134e-01 2.34021130e-03
9.20424402e-01 -9.33387280e-02 -3.33747208e-01 -1.76338881e-01
-8.82893860e-01 1.22514382e-01 7.86667317e-02 8.63155425e-01
1.04511690e+00 -7.47003034e-02 -5.34542084e-01 5.28722107e-01
5.09005249e-01 8.06131124e-01 2.70271808e-01 -6.26554072e-01
5.65925539e-01 5.59754074e-01 3.03401947e-01 -1.32384431e+00
-4.00707483e-01 -6.22654855e-01 -1.15577018e+00 7.49266446e-02
-5.39928861e-02 -2.66418189e-01 -8.72980833e-01 1.77006209e+00
6.19992793e-01 4.92097974e-01 -1.13120303e-01 8.35438788e-01
6.56318665e-01 8.67432237e-01 -1.40104562e-01 -2.46109232e-01
9.98653352e-01 -5.86467803e-01 -6.93256438e-01 -1.75547943e-01
4.23971444e-01 -7.25826249e-02 6.87937677e-01 1.28534427e-02
-5.97951531e-01 -5.18659949e-01 -1.09147191e+00 2.96174616e-01
-2.59420574e-01 2.73399234e-01 3.21739793e-01 7.35217556e-02
-8.61955166e-01 6.70665741e-01 -1.00942659e+00 -1.18055463e-01
6.01800084e-01 -7.65879601e-02 -6.97890520e-02 -1.81588605e-01
-1.69975865e+00 6.62853777e-01 6.26815557e-01 6.10469818e-01
-9.57042098e-01 -7.01001227e-01 -1.00869310e+00 -1.09769687e-01
4.48731422e-01 -4.53556538e-01 1.10773528e+00 -5.17414272e-01
-1.05672860e+00 2.77937561e-01 -1.16869651e-01 -4.49529767e-01
8.21464658e-01 -1.40621439e-01 -5.58680832e-01 -4.78602909e-02
4.37864035e-01 6.55474544e-01 7.76279747e-01 -1.38451481e+00
-1.05080807e+00 -2.59576052e-01 -2.51712631e-02 1.10531636e-01
3.29801887e-02 -6.28280699e-01 -5.82029343e-01 -7.08843589e-01
4.50774908e-01 -6.96978509e-01 -1.94647774e-01 1.93968102e-01
-3.02567482e-01 -2.42213935e-01 7.18543649e-01 -8.10308933e-01
1.59145784e+00 -2.19385004e+00 -3.31609696e-01 4.40955400e-01
2.14477479e-01 1.59950808e-01 3.95655632e-02 2.14669943e-01
2.53656328e-01 1.77858531e-01 -8.34367037e-01 -5.61328381e-02
5.44780120e-03 6.17880449e-02 -3.50159824e-01 4.74556565e-01
4.28464621e-01 8.94094944e-01 -1.19398463e+00 -3.51331204e-01
3.32941204e-01 2.99996138e-01 -1.24684408e-01 -7.18787983e-02
-4.12994474e-01 6.88128650e-01 -6.95448875e-01 5.85072577e-01
1.00939727e+00 -1.07200064e-01 -2.79157937e-01 -1.14799410e-01
-2.41825178e-01 2.25845605e-01 -1.38561797e+00 1.51950562e+00
-5.47815621e-01 6.83567226e-01 -3.33439052e-01 -4.68990982e-01
9.70266998e-01 -2.42894720e-02 3.33120674e-01 -7.81121731e-01
7.11962953e-02 3.42551172e-01 -2.45412618e-01 -1.86800420e-01
5.75321555e-01 2.65230685e-01 -1.67394921e-01 1.83667213e-01
-6.00629807e-01 -2.15917349e-01 5.80723286e-02 2.21097142e-01
1.03368640e+00 2.11423218e-01 2.46357143e-01 -1.41395673e-01
6.26944542e-01 -7.63053223e-02 1.10492992e+00 5.48362732e-01
-4.86508191e-01 5.54800391e-01 2.31563509e-01 -2.42303208e-01
-4.47195381e-01 -1.00160456e+00 -4.13818300e-01 2.87400901e-01
5.76654196e-01 -1.87257349e-01 -3.81063014e-01 -7.17128336e-01
2.58741081e-01 7.82123268e-01 -5.61271191e-01 -3.59673858e-01
-2.76486397e-01 -7.81816244e-01 3.88994902e-01 6.24631286e-01
1.15581834e+00 -7.18521178e-01 -3.20575178e-01 3.78463745e-01
-6.30718768e-01 -8.12793493e-01 -4.57367569e-01 -2.91138381e-01
-7.76917458e-01 -9.22771811e-01 -3.11018288e-01 -2.61948943e-01
6.54221773e-01 3.81605178e-01 7.70924449e-01 6.52011782e-02
-7.43394122e-02 2.38773227e-01 -3.45588684e-01 -3.02597404e-01
5.45229912e-02 -1.36628613e-01 2.10782234e-02 6.84242770e-02
8.37456658e-02 -7.34680891e-01 -8.18157017e-01 3.56464922e-01
-9.04959619e-01 -3.67567502e-02 5.36045074e-01 7.22733855e-01
8.89477015e-01 5.39339006e-01 7.75327206e-01 -4.18889254e-01
3.32420588e-01 -8.02967131e-01 -7.26343930e-01 2.22917888e-02
-8.48508358e-01 3.49174589e-02 3.34938355e-02 -6.02167659e-02
-1.35112834e+00 -1.64918631e-01 -1.71539292e-01 -1.75317019e-01
6.16312176e-02 9.61629868e-01 -4.17241573e-01 2.12158278e-01
6.34227931e-01 2.88069099e-01 -4.30250168e-01 -2.87549436e-01
3.21316600e-01 6.74228251e-01 7.78758049e-01 -5.35062015e-01
9.38895822e-01 6.12413704e-01 -1.41094863e-01 -4.44848418e-01
-8.98454309e-01 -5.08554578e-01 -4.73448783e-01 -5.45870662e-01
3.81433696e-01 -1.34589303e+00 -7.53277093e-02 9.42537189e-01
-9.93543983e-01 -3.89603078e-01 -2.85914272e-01 4.18552399e-01
-3.31151515e-01 3.14771205e-01 -2.37276584e-01 -6.95028305e-01
-4.03531462e-01 -7.14357197e-01 1.17937744e+00 4.22773123e-01
3.79909098e-01 -9.17009592e-01 1.73959985e-01 -2.52190847e-02
3.88498336e-01 7.32602179e-01 5.25690377e-01 -1.26995388e-02
-9.66371298e-01 -1.92791432e-01 -4.95709419e-01 2.80257046e-01
3.62669140e-01 1.12692781e-01 -9.85014796e-01 -1.80747986e-01
-3.33544910e-01 1.05735697e-01 1.33830643e+00 5.53114176e-01
1.08288217e+00 -1.64562181e-01 -5.92807114e-01 4.02861714e-01
1.52138674e+00 5.75351864e-02 7.03329682e-01 3.61272752e-01
4.95695025e-01 2.43409172e-01 1.11209416e+00 7.62018859e-01
8.02191138e-01 5.56396067e-01 6.60404444e-01 2.32028380e-01
-2.47692019e-01 -3.72028887e-01 3.57823431e-01 5.85631132e-01
2.13465765e-01 -3.32644761e-01 -1.13296199e+00 8.35255146e-01
-2.24320674e+00 -8.95037830e-01 -3.64594966e-01 2.17987585e+00
8.94497454e-01 3.17477256e-01 -3.52160126e-01 8.87711272e-02
9.45378780e-01 4.02561784e-01 -9.37628508e-01 5.18839657e-01
-4.00758088e-02 -3.41524333e-01 4.40304697e-01 5.50254822e-01
-1.25822830e+00 8.41422319e-01 4.72753191e+00 9.81157064e-01
-1.04863167e+00 2.61967003e-01 5.37404478e-01 1.31574750e-01
-6.42957091e-01 6.79409578e-02 -7.58002937e-01 7.18533456e-01
4.86519396e-01 -3.74804169e-01 7.43392929e-02 6.41743839e-01
6.72765613e-01 -6.05445743e-01 -8.08628321e-01 8.19618165e-01
-2.99010783e-01 -1.25412858e+00 -2.22701862e-01 -1.84865013e-01
1.09068394e+00 2.16279075e-01 -1.41273558e-01 3.86196375e-01
4.54503387e-01 -5.26222527e-01 7.26189613e-01 9.66361046e-01
8.22218120e-01 -6.84172869e-01 8.11414242e-01 3.35440814e-01
-1.60232651e+00 2.16444228e-02 -1.95664600e-01 5.01853079e-02
2.56287158e-01 1.46708858e+00 -5.66936076e-01 1.02552402e+00
8.35769176e-01 1.25557280e+00 -6.18360519e-01 1.28436446e+00
-6.13727510e-01 5.47192514e-01 -5.32021344e-01 2.71006584e-01
1.67934403e-01 -1.75870478e-01 7.40213573e-01 9.79234695e-01
5.68817496e-01 2.07540050e-01 9.45756137e-02 1.00215590e+00
-1.28353149e-01 -2.45043740e-01 -3.85670871e-01 3.92325401e-01
8.72478068e-01 1.17136228e+00 -4.57445592e-01 -2.57766515e-01
-1.15756840e-02 7.43607402e-01 1.56232879e-01 2.76883215e-01
-9.28959787e-01 -4.31110382e-01 8.84994686e-01 -1.00729965e-01
3.27404261e-01 -2.60896057e-01 -2.24602833e-01 -1.13231170e+00
3.99315357e-01 -3.00039083e-01 2.88758665e-01 -9.80764210e-01
-1.19399345e+00 4.59765047e-01 1.57013893e-01 -1.52455962e+00
-1.51275873e-01 -2.80403458e-02 -9.31863785e-01 8.88825178e-01
-1.88093901e+00 -1.10348749e+00 -8.55664313e-01 3.60704988e-01
3.55190009e-01 4.45581049e-01 7.23729506e-02 2.38560989e-01
-8.79473090e-01 2.38673687e-01 4.62287903e-01 6.66138008e-02
6.86934710e-01 -1.16404557e+00 5.06495655e-01 1.33611321e+00
-3.80197465e-01 9.32236835e-02 5.53522646e-01 -9.49756920e-01
-7.03386724e-01 -1.80248642e+00 7.03423560e-01 -1.57255873e-01
6.98297799e-01 1.45937130e-01 -9.31671321e-01 4.33741421e-01
-4.21597302e-01 3.26078564e-01 -4.21399623e-02 -1.71310693e-01
-1.64610848e-01 -5.62813222e-01 -1.16494524e+00 3.87133539e-01
1.13719988e+00 -5.63246250e-01 -3.13381493e-01 1.83035374e-01
9.57711637e-01 -5.74372530e-01 -8.71288300e-01 7.41627514e-01
2.23488182e-01 -7.57996798e-01 5.17739415e-01 1.53073326e-01
3.54199111e-01 -9.03369725e-01 -8.77544954e-02 -1.55181897e+00
-3.74646276e-01 -4.40422334e-02 -2.32290864e-01 1.55765843e+00
2.11393803e-01 -9.17803347e-01 2.15290710e-01 3.74125272e-01
-2.72377968e-01 -5.95143259e-01 -1.13733685e+00 -9.61209834e-01
-7.45744407e-02 -7.77393043e-01 9.23749626e-01 6.45597219e-01
-4.34668452e-01 -1.03896849e-01 -9.46031809e-02 7.47805953e-01
4.49686885e-01 2.06523165e-01 5.28294623e-01 -1.31026423e+00
1.10863172e-01 -3.95443439e-01 -5.06839454e-01 -1.06131470e+00
-6.37823865e-02 -6.86199903e-01 5.42781830e-01 -1.91892540e+00
6.85233332e-04 -8.40160787e-01 -4.37693328e-01 5.07764578e-01
-5.27213275e-01 -1.17081188e-01 -7.45441839e-02 4.35468048e-01
-4.40983206e-01 1.11619651e+00 1.17433190e+00 -3.79530162e-01
-2.22666532e-01 1.20069087e-01 -4.38130140e-01 4.05068666e-01
8.64528239e-01 -5.94601631e-01 -5.03917217e-01 -3.28497499e-01
2.63882786e-01 -1.60652995e-01 6.46053851e-01 -1.41112316e+00
2.42758840e-01 -3.30706328e-01 2.63678670e-01 -9.52569723e-01
-1.30242363e-01 -6.26504898e-01 3.58753502e-01 3.72117162e-01
4.45630439e-02 -2.96971709e-01 2.65184641e-01 9.82670963e-01
-4.20695543e-01 3.77835520e-02 5.91229796e-01 1.38620436e-01
-1.15967631e+00 8.51375699e-01 -1.18083619e-01 -5.34312017e-02
1.12814283e+00 2.52664596e-01 -5.63230515e-01 -4.90327448e-01
-4.71309453e-01 8.92296076e-01 4.74723101e-01 3.74352455e-01
5.89368165e-01 -1.66629529e+00 -7.76525199e-01 -8.30256194e-02
4.93926823e-01 6.02059126e-01 5.79500616e-01 9.34670448e-01
-2.07349807e-01 2.08661363e-01 1.88288629e-01 -8.59123528e-01
-6.60306931e-01 3.03721279e-01 5.94534814e-01 -2.07067668e-01
-5.77077687e-01 5.85737586e-01 5.24083376e-02 -3.33812654e-01
2.19056066e-02 -6.45911813e-01 -4.10494432e-02 1.76801741e-01
4.99631077e-01 4.02894527e-01 1.63965672e-01 -4.20130402e-01
-5.70392847e-01 4.03387487e-01 3.06732357e-01 -1.76073804e-01
1.20640779e+00 -5.84336162e-01 -4.83628623e-02 4.98071879e-01
8.68386805e-01 -2.73164243e-01 -1.77966046e+00 -7.05600441e-01
-9.92806181e-02 -5.26976466e-01 5.14611125e-01 -8.68999183e-01
-1.17130041e+00 7.15919495e-01 8.70112836e-01 -1.27185151e-01
1.35677087e+00 -1.42346993e-01 7.27432013e-01 3.69658709e-01
5.57808340e-01 -1.13032293e+00 -7.00731426e-02 4.68679041e-01
1.08468950e+00 -1.65212893e+00 7.08276778e-02 -2.92531401e-01
-4.79047298e-01 7.27882624e-01 7.26706386e-01 9.10115913e-02
9.50449288e-01 -1.60574943e-01 -2.01759160e-01 -5.98319992e-02
-5.67904651e-01 -5.57295859e-01 2.57717252e-01 3.98919493e-01
-1.71851143e-01 2.93066651e-01 -3.08178365e-01 6.38246059e-01
2.86762953e-01 1.75016910e-01 3.78429949e-01 9.07409787e-01
-6.96294844e-01 -6.41946256e-01 -2.30703682e-01 5.33899486e-01
3.01108569e-01 -1.11602740e-02 1.09855928e-01 4.40487564e-01
2.78537273e-01 1.08305895e+00 1.71551540e-01 -5.51637173e-01
2.03962803e-01 -3.77409339e-01 -1.08381443e-01 -3.55535686e-01
-4.23650024e-03 -3.29305381e-01 1.22723885e-01 -5.92225313e-01
-4.18640465e-01 -9.80335414e-01 -1.34503937e+00 -3.23075861e-01
-3.59695762e-01 -3.99328209e-02 5.19958496e-01 9.82332647e-01
5.80905735e-01 6.30189121e-01 9.58786249e-01 -6.64413869e-01
-3.64709944e-01 -1.11547565e+00 -4.62562114e-01 -3.93835977e-02
5.42578638e-01 -9.39153850e-01 -3.99648160e-01 -1.46884575e-01] | [9.672961235046387, -1.2651631832122803] |
3dd15130-801c-416c-9d3d-b15a542a5ea8 | intra-inter-interaction-network-with-latent | null | null | https://aclanthology.org/2020.coling-main.437 | https://aclanthology.org/2020.coling-main.437.pdf | Intra-/Inter-Interaction Network with Latent Interaction Modeling for Multi-turn Response Selection | Multi-turn response selection has been extensively studied and applied to many real-world applications in recent years. However, current methods typically model the interactions between multi-turn utterances and candidate responses with iterative approaches, which is not practical as the turns of conversations vary. Besides, some latent features, such as user intent and conversation topic, are under-discovered in existing works. In this work, we propose Intra-/Inter-Interaction Network (I$^3$) with latent interaction modeling to comprehensively model multi-level interactions between the utterance context and the response. In specific, we first encode the intra- and inter-utterance interaction with the given response from both individual utterance and the overall utterance context. Then we develop a latent multi-view subspace clustering module to model the latent interaction between the utterance and response. Experimental results show that the proposed method substantially and consistently outperforms existing state-of-the-art methods on three multi-turn response selection benchmark datasets. | ['Wai Lam', 'Wenxuan Zhang', 'Yang Deng'] | 2020-12-01 | null | null | null | coling-2020-8 | ['multi-view-subspace-clustering'] | ['computer-vision'] | [ 2.48317257e-01 -3.26257437e-01 -1.85089424e-01 -9.55638945e-01
-1.09535015e+00 -4.93179440e-01 5.18219471e-01 -2.55951226e-01
4.99425232e-02 2.39120632e-01 8.39574099e-01 5.32712787e-02
-2.06745695e-02 -1.92257524e-01 4.97890525e-02 -7.20347524e-01
4.68986213e-01 5.76135933e-01 -8.56538117e-02 -3.57908845e-01
2.54411161e-01 -3.09290707e-01 -1.24137461e+00 9.10593450e-01
7.07853913e-01 6.79185808e-01 2.31244132e-01 5.73829949e-01
-2.66414165e-01 7.50540078e-01 -3.08628947e-01 -6.53158128e-02
-2.11281478e-01 -8.53518128e-01 -9.42193151e-01 4.79960769e-01
-8.66956487e-02 -1.25043169e-01 -2.13632464e-01 6.25147760e-01
6.10837102e-01 5.14806569e-01 4.83701497e-01 -9.73399043e-01
-3.13726813e-01 6.61383152e-01 -5.30742705e-01 -1.03434935e-01
9.71822619e-01 4.76108976e-02 1.06021440e+00 -1.11027825e+00
5.54671288e-01 1.83074033e+00 2.64981985e-01 5.40955365e-01
-1.22857606e+00 -6.99171126e-01 6.46659076e-01 3.65047216e-01
-1.04134107e+00 -7.15892076e-01 1.14910877e+00 -3.33419025e-01
7.56159902e-01 6.53744221e-01 3.49824995e-01 1.19731939e+00
-3.82741005e-03 1.37022460e+00 1.25668097e+00 -3.33670795e-01
-1.01306096e-01 4.44880456e-01 6.67703152e-01 5.05481334e-03
-1.14746082e+00 -5.74990988e-01 -6.68139994e-01 -5.58835924e-01
1.93275765e-01 1.72596171e-01 -2.11552680e-01 -1.85616300e-01
-1.11625171e+00 8.31287444e-01 -1.56785220e-01 1.45100147e-01
-4.17844296e-01 -3.85543108e-01 6.09457612e-01 4.85542595e-01
7.57114649e-01 -1.50289640e-01 -4.90870714e-01 -4.63790178e-01
-5.59008658e-01 2.00732172e-01 9.40663099e-01 1.00897622e+00
8.18884552e-01 -1.93645716e-01 -1.87672049e-01 1.48286641e+00
2.87539274e-01 1.41967878e-01 2.63960361e-01 -7.49416173e-01
7.06645131e-01 8.62304926e-01 1.38271853e-01 -1.13761091e+00
-4.72570121e-01 -1.08753882e-01 -8.83675158e-01 -9.36393142e-01
3.54110636e-02 -1.28715351e-01 -2.38376871e-01 1.54857230e+00
6.85457170e-01 1.08914450e-01 2.65704215e-01 9.25194740e-01
1.13906240e+00 7.27191865e-01 -1.11128420e-01 -9.72382665e-01
1.25696015e+00 -1.22593296e+00 -1.02610493e+00 -2.85058230e-01
6.46916032e-01 -1.11887836e+00 1.32777059e+00 1.98489204e-01
-9.78464603e-01 -6.80279016e-01 -4.06634778e-01 1.02497511e-01
3.63474250e-01 3.60614687e-01 5.16909599e-01 2.24157289e-01
-7.28721619e-01 -4.04465109e-01 -3.72613490e-01 -3.76918793e-01
-4.76433933e-01 5.43370485e-01 -2.26183057e-01 -2.68214315e-01
-1.32293582e+00 4.60634232e-01 -1.59686372e-01 1.83663234e-01
-6.67713225e-01 -1.56810433e-01 -5.43766081e-01 -6.83726072e-02
6.53426826e-01 -5.11543214e-01 1.42589021e+00 -1.03113914e+00
-2.01312566e+00 4.76405174e-01 -8.63172948e-01 2.38075569e-01
2.75257956e-02 -1.90671965e-01 -5.38153708e-01 -4.05225493e-02
-1.78439006e-01 2.00484842e-01 6.01027668e-01 -1.28530765e+00
-5.55937171e-01 -6.14105999e-01 2.36739710e-01 8.63033831e-01
-3.76344651e-01 4.75093335e-01 -8.45303059e-01 -2.74480700e-01
5.65227568e-01 -1.43763065e+00 -1.32416695e-01 -1.15306234e+00
-5.35494804e-01 -7.75874138e-01 1.05435300e+00 -4.36730534e-01
1.41653788e+00 -2.28964877e+00 4.37476099e-01 -1.24405161e-01
5.02094924e-02 -2.35859111e-01 -2.08976455e-02 9.30865049e-01
-1.38026103e-01 -1.95726320e-01 1.56515017e-01 -6.89481795e-01
1.00433221e-03 -2.16500074e-01 -5.09525001e-01 4.60297376e-01
-5.85523307e-01 5.91435671e-01 -7.26174653e-01 -3.30614299e-01
1.16154321e-01 2.97491103e-01 -5.61630249e-01 6.19906962e-01
-1.36907147e-02 8.86202157e-01 -6.58754945e-01 3.44029009e-01
5.15167534e-01 -8.62536132e-02 6.56991303e-01 -1.42249167e-01
-3.62936081e-03 6.25242472e-01 -1.03281081e+00 1.69224179e+00
-5.75679898e-01 4.68690604e-01 2.97099769e-01 -9.17569160e-01
9.61673498e-01 7.37633169e-01 6.62707508e-01 -2.94693023e-01
9.57505777e-03 -9.26170275e-02 6.88157454e-02 -6.17846072e-01
5.40278554e-01 -3.62356706e-03 -6.58736587e-01 9.58794117e-01
-9.21835005e-02 -1.05029613e-01 -2.11665958e-01 4.78987366e-01
6.19728386e-01 -2.39822954e-01 3.56960386e-01 -1.46846585e-02
8.45372677e-01 -4.34482425e-01 8.35174799e-01 4.55213070e-01
-1.83440641e-01 5.02074778e-01 5.36819696e-01 -3.66345376e-01
-4.06838357e-01 -4.29356933e-01 2.69868851e-01 1.62070727e+00
3.36604536e-01 -4.37897325e-01 -8.03431630e-01 -6.76335454e-01
-5.50902307e-01 7.47579992e-01 -4.47347671e-01 3.34420539e-02
-4.96046871e-01 -7.06948102e-01 1.08524792e-01 1.93661883e-01
2.88116485e-01 -1.05550623e+00 2.23933205e-01 4.43682194e-01
-9.72773314e-01 -1.22356820e+00 -1.00464845e+00 -1.86481357e-01
-4.98865157e-01 -7.69249797e-01 -1.38281584e-01 -7.06968725e-01
4.30057228e-01 7.89744318e-01 9.22120690e-01 -1.64791375e-01
2.79094726e-01 6.62615895e-01 -6.36336029e-01 1.35283154e-02
-2.18285933e-01 2.45015044e-02 3.82573336e-01 4.85108912e-01
7.06659377e-01 -4.88040149e-01 -6.80507064e-01 9.10784662e-01
-6.31743789e-01 1.19180977e-01 1.78585172e-01 9.50957954e-01
3.84377778e-01 -1.58147514e-01 8.60106409e-01 -1.11414707e+00
1.01649153e+00 -5.86851478e-01 1.09692767e-01 4.54809695e-01
-3.31687629e-01 -5.13512731e-01 5.53967476e-01 -4.68932569e-01
-1.48381197e+00 7.83873126e-02 -2.45311819e-02 -4.69367839e-02
-3.95684451e-01 5.59677303e-01 -5.23355782e-01 4.15386945e-01
5.10084964e-02 4.57886606e-01 -2.97111630e-01 -3.32774729e-01
2.80817062e-01 1.22465301e+00 -9.17194784e-02 -5.11688709e-01
1.22977771e-01 3.66537988e-01 -7.24020183e-01 -8.21215272e-01
-8.69546533e-01 -1.17515552e+00 -6.55239701e-01 -4.63531256e-01
6.75919354e-01 -9.62164104e-01 -8.69508862e-01 4.54244584e-01
-1.27776599e+00 2.04020366e-02 5.28219461e-01 6.44653022e-01
-5.60639203e-01 4.66031641e-01 -7.27711916e-01 -1.21674299e+00
-2.84577072e-01 -1.64999199e+00 1.08602762e+00 1.46577045e-01
-4.84938592e-01 -9.37113464e-01 6.05673492e-02 9.08738256e-01
3.55983153e-02 -3.89460087e-01 8.28977704e-01 -7.45600104e-01
-3.98338318e-01 -9.53619629e-02 1.53262272e-01 3.30850668e-02
2.93060064e-01 -1.78841323e-01 -8.88188362e-01 -3.17812353e-01
5.56274712e-01 -5.81465960e-01 5.75029016e-01 3.77206445e-01
8.68853569e-01 -2.26765931e-01 -3.75408888e-01 1.68202043e-01
6.95397973e-01 6.03996038e-01 2.84489721e-01 -2.67274737e-01
7.15476513e-01 1.28752542e+00 9.21969354e-01 5.23890436e-01
8.64842176e-01 1.01587081e+00 -1.02412187e-01 5.60534671e-02
6.14934623e-01 9.13756341e-03 5.54591656e-01 1.67585754e+00
4.05690253e-01 -4.74934399e-01 -7.10401595e-01 4.03409541e-01
-2.29716587e+00 -1.03665960e+00 -3.15091103e-01 1.88439906e+00
7.07303107e-01 -9.52917635e-02 2.46886820e-01 -2.93764085e-01
6.36471033e-01 3.15635920e-01 -5.40673137e-01 -5.53088665e-01
2.47863978e-02 -5.00416517e-01 -3.47239822e-01 5.97818792e-01
-9.33381200e-01 9.97892857e-01 6.03666735e+00 7.78430939e-01
-8.22576404e-01 2.20252216e-01 9.27272975e-01 -2.34103665e-01
-5.02026677e-01 2.12921217e-01 -9.56309080e-01 2.46953726e-01
7.51615226e-01 -6.21075444e-02 4.84280884e-01 7.50957429e-01
6.91794574e-01 -4.42782119e-02 -1.22022462e+00 1.00976801e+00
3.71167004e-01 -8.09625447e-01 -3.15359145e-01 -1.31126702e-01
6.83156073e-01 -4.05281782e-01 1.97608232e-01 6.20561242e-01
2.75108278e-01 -6.62675023e-01 1.01529144e-01 3.81901950e-01
6.29838467e-01 -9.70722258e-01 6.36985183e-01 8.80757391e-01
-1.36709785e+00 -8.89579877e-02 -1.34136707e-01 -6.60037175e-02
1.76441044e-01 2.75402308e-01 -8.29208374e-01 5.72040856e-01
4.95770872e-01 6.27243698e-01 -1.85231511e-02 7.38773644e-02
6.61563536e-04 8.16749096e-01 -5.33409743e-03 -2.84910109e-02
3.41558427e-01 -5.19410133e-01 5.97391129e-01 1.15776873e+00
2.01815944e-02 5.90936661e-01 7.93113708e-01 1.93877533e-01
1.63217247e-01 6.16679847e-01 -4.85534161e-01 9.04358625e-02
6.61980450e-01 1.23271704e+00 -4.84154671e-01 -3.75144809e-01
-5.45566499e-01 7.94497907e-01 1.72912955e-01 5.82369447e-01
-6.23211861e-01 1.30017221e-01 7.20990717e-01 -2.45604753e-01
-1.08766280e-01 4.93448712e-02 -9.97864008e-02 -1.33553720e+00
2.77872309e-02 -1.37844038e+00 3.08035642e-01 -4.32707638e-01
-1.09384918e+00 5.91911376e-01 -6.81564352e-03 -1.41388035e+00
-6.06833398e-01 2.37449244e-01 -6.27793968e-01 9.66607690e-01
-6.57607853e-01 -1.18257797e+00 6.76609296e-03 5.32291710e-01
1.41596663e+00 -3.13661724e-01 1.05163443e+00 7.03274980e-02
-8.63480031e-01 7.04295397e-01 2.16476634e-01 -1.16960935e-01
1.05313742e+00 -7.89221466e-01 8.34947303e-02 4.59529877e-01
-8.15493464e-02 1.24027371e+00 6.60074949e-01 -4.19071347e-01
-1.62547398e+00 -6.22724116e-01 9.09912050e-01 -4.16985303e-01
3.56825262e-01 -6.34666800e-01 -7.77044415e-01 6.90311849e-01
5.70313632e-01 -5.98431587e-01 1.33695686e+00 7.73483872e-01
-1.18662812e-01 -1.36481270e-01 -5.88409662e-01 7.26094484e-01
7.13379860e-01 -1.03914976e+00 -4.03485149e-01 4.94336188e-01
9.09808576e-01 -4.89275903e-01 -5.95263779e-01 4.36752707e-01
7.36460567e-01 -1.12245214e+00 9.11101043e-01 -4.99744833e-01
2.59145021e-01 3.73743884e-02 -3.41686785e-01 -1.46476400e+00
-2.66672492e-01 -8.31097424e-01 7.02371448e-02 1.46033800e+00
1.03698134e-01 -2.07747787e-01 6.46197855e-01 8.74985158e-01
-1.66916460e-01 -8.60226750e-01 -6.68188691e-01 2.24200636e-02
-3.81483316e-01 -4.89229470e-01 5.47650576e-01 1.08107877e+00
5.45552731e-01 1.01344848e+00 -1.00116563e+00 -1.76509008e-01
2.45396122e-01 7.75591552e-01 1.27111804e+00 -7.84619689e-01
-4.53065336e-01 -2.00984761e-01 1.66701898e-01 -1.74650109e+00
5.36126733e-01 -4.38229144e-01 2.65428782e-01 -1.32915664e+00
6.32444739e-01 -2.00762644e-01 -1.84357792e-01 -1.48445591e-01
-6.74558759e-01 -4.22827393e-01 -6.10288745e-03 5.12802303e-01
-1.05724287e+00 7.48241961e-01 1.00310242e+00 -6.83237687e-02
-6.46448672e-01 4.20620352e-01 -7.38879442e-01 7.70331144e-01
5.28037131e-01 -4.45363581e-01 -7.98574805e-01 -2.88108021e-01
-1.18311808e-01 8.88377309e-01 -3.28963310e-01 -1.67098448e-01
3.45132560e-01 -5.22511363e-01 -2.19390512e-01 -9.98193622e-01
6.82945073e-01 -6.08267426e-01 4.99812663e-02 -2.51418054e-01
-9.34815168e-01 -3.37234922e-02 -3.29918116e-01 9.52720165e-01
-5.92962742e-01 1.84557274e-01 3.79085779e-01 -1.57603532e-01
-4.52272445e-01 2.33480304e-01 -5.87922454e-01 -1.82907417e-01
7.65086353e-01 -1.83021836e-02 5.63515071e-03 -1.05327797e+00
-8.12680900e-01 5.36962748e-01 2.38257423e-01 6.92689776e-01
7.73995519e-01 -1.22858763e+00 -7.06159532e-01 -5.47796041e-02
2.58127004e-01 -8.65748078e-02 9.66110706e-01 1.06023109e+00
4.38368112e-01 6.18029237e-01 4.19923663e-01 -8.29259932e-01
-1.86945999e+00 3.02495927e-01 -1.60356194e-01 -4.74611551e-01
-1.34873033e-01 1.02150345e+00 8.30891371e-01 -9.30193663e-01
2.80609310e-01 1.27766237e-01 -6.34589553e-01 2.57861674e-01
4.97934699e-01 1.38958275e-01 -3.92747372e-01 -1.04662371e+00
-3.80476892e-01 3.40392083e-01 -4.51042712e-01 -2.47780845e-01
1.18055367e+00 -8.39708984e-01 -3.66542339e-01 1.05948651e+00
1.46074224e+00 1.99831668e-02 -9.15288329e-01 -7.39971697e-01
-5.53508475e-02 -4.23855633e-01 -2.15925813e-01 -2.86302328e-01
-6.14362538e-01 8.47019613e-01 1.54363602e-01 2.49988869e-01
1.11276090e+00 -1.05551176e-01 9.48125482e-01 5.61156332e-01
3.77857983e-01 -1.43889415e+00 4.58501577e-01 6.69732869e-01
9.41230536e-01 -1.41611564e+00 -9.38342139e-02 -6.71811640e-01
-1.19746876e+00 8.87696862e-01 8.79207790e-01 2.70162791e-01
6.21172071e-01 -9.38490480e-02 4.39368725e-01 -8.54880661e-02
-1.38274014e+00 2.27338195e-01 3.50421488e-01 4.24551927e-02
9.60695922e-01 3.29458505e-01 -4.49361175e-01 7.53669262e-01
6.29793257e-02 -4.81054723e-01 3.08315605e-01 5.94050765e-01
-1.57264918e-01 -1.43886364e+00 -3.28077078e-01 3.20708185e-01
-3.16846192e-01 -3.56515050e-02 -8.58307362e-01 1.98638767e-01
-2.58868665e-01 1.67052138e+00 -3.64332557e-01 -9.31336045e-01
2.50179678e-01 3.84169668e-01 -2.21923679e-01 -9.12785888e-01
-8.01312506e-01 8.81582201e-01 2.91691810e-01 -4.57516968e-01
-7.24427223e-01 -1.11308765e+00 -9.21330929e-01 -9.16540176e-02
-5.92036366e-01 4.25468594e-01 3.77461195e-01 1.18348837e+00
3.71410370e-01 3.91776264e-01 1.30090535e+00 -5.93932390e-01
-4.98164862e-01 -1.32296872e+00 -3.84233207e-01 4.86469001e-01
4.80297506e-02 -5.20383120e-01 -3.28160256e-01 -2.60440074e-03] | [12.612022399902344, 7.583490371704102] |
eb19b8d9-3906-42c5-8ecc-24b4cdee82d5 | stem-seg-spatio-temporal-embeddings-for | 2003.08429 | null | https://arxiv.org/abs/2003.08429v3 | https://arxiv.org/pdf/2003.08429v3.pdf | STEm-Seg: Spatio-temporal Embeddings for Instance Segmentation in Videos | Existing methods for instance segmentation in videos typi-cally involve multi-stage pipelines that follow the tracking-by-detectionparadigm and model a video clip as a sequence of images. Multiple net-works are used to detect objects in individual frames, and then associatethese detections over time. Hence, these methods are often non-end-to-end trainable and highly tailored to specific tasks. In this paper, we pro-pose a different approach that is well-suited to a variety of tasks involvinginstance segmentation in videos. In particular, we model a video clip asa single 3D spatio-temporal volume, and propose a novel approach thatsegments and tracks instances across space and time in a single stage. Ourproblem formulation is centered around the idea of spatio-temporal em-beddings which are trained to cluster pixels belonging to a specific objectinstance over an entire video clip. To this end, we introduce (i) novel mix-ing functions that enhance the feature representation of spatio-temporalembeddings, and (ii) a single-stage, proposal-free network that can rea-son about temporal context. Our network is trained end-to-end to learnspatio-temporal embeddings as well as parameters required to clusterthese embeddings, thus simplifying inference. Our method achieves state-of-the-art results across multiple datasets and tasks. Code and modelsare available at https://github.com/sabarim/STEm-Seg. | ['Laura Leal-Taixé', 'Aljoša Ošep', 'Sabarinath Mahadevan', 'Bastian Leibe', 'Ali Athar'] | 2020-03-18 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/1299_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123560154.pdf | eccv-2020-8 | ['video-instance-segmentation', 'unsupervised-video-object-segmentation'] | ['computer-vision', 'computer-vision'] | [-1.72219917e-01 -2.55964369e-01 -4.05273825e-01 -2.62712985e-01
-9.23099697e-01 -7.15201616e-01 8.24099004e-01 1.49151489e-01
-6.85927570e-01 2.37517208e-01 -7.66035020e-02 -4.28880826e-02
6.68233782e-02 -5.46140373e-01 -1.06216168e+00 -6.29825234e-01
-2.38039777e-01 3.80869299e-01 6.82414830e-01 3.30618501e-01
-4.55343388e-02 5.49814701e-01 -1.41850662e+00 4.40123558e-01
5.56107581e-01 1.12594640e+00 2.34071031e-01 8.91482174e-01
8.94934759e-02 7.40246415e-01 -2.34820738e-01 -2.28775427e-01
1.22647159e-01 -3.58550251e-01 -6.92970753e-01 4.85563666e-01
5.82935750e-01 -2.93056160e-01 -6.02532387e-01 7.44745970e-01
3.89503967e-03 3.42039168e-01 6.58477366e-01 -1.13189828e+00
-2.56937236e-01 3.32207859e-01 -9.13686037e-01 6.26107097e-01
-1.23232706e-02 4.33355123e-02 8.67651701e-01 -1.23822582e+00
6.42433763e-01 1.12788546e+00 6.66116714e-01 4.33056176e-01
-1.40527320e+00 -4.71855253e-01 5.17287314e-01 2.58961439e-01
-1.30113471e+00 -3.72720540e-01 6.43174231e-01 -6.93993628e-01
7.41628528e-01 8.73599201e-02 8.38017285e-01 9.34074104e-01
3.09507132e-01 1.23376024e+00 6.39824808e-01 -9.02878419e-02
9.52186808e-02 -2.46547669e-01 8.77885818e-02 7.00045586e-01
-4.46774848e-02 -5.91963977e-02 -4.78027940e-01 1.72413319e-01
8.73810410e-01 3.66977781e-01 -2.54176706e-02 -6.37272656e-01
-1.23415422e+00 6.51478648e-01 4.18294519e-01 3.96315664e-01
-5.10217011e-01 4.48125035e-01 3.56083363e-01 -7.68224970e-02
6.91149354e-01 2.70596687e-02 -4.60951656e-01 -6.92675007e-04
-1.41195977e+00 5.36787629e-01 3.98586035e-01 7.45490313e-01
8.33850384e-01 -2.05656052e-01 -3.16691160e-01 6.19617641e-01
3.67201984e-01 1.16968811e-01 3.50785464e-01 -1.06057739e+00
3.51291448e-01 1.23331249e-01 1.44430235e-01 -8.87842894e-01
-4.60750967e-01 -2.63427526e-01 -4.22377229e-01 2.33533934e-01
4.24392998e-01 -1.67957053e-01 -1.20981658e+00 1.51926208e+00
6.19215727e-01 8.07039261e-01 -4.05526459e-01 9.93851304e-01
5.74303806e-01 8.39532793e-01 9.76154283e-02 -5.13142347e-02
1.58176136e+00 -1.29951823e+00 -6.01141453e-01 -2.99013376e-01
5.64434052e-01 -6.40735030e-01 6.97482884e-01 3.73998970e-01
-1.27966082e+00 -7.40032077e-01 -7.71985650e-01 -1.83503166e-01
-3.48458856e-01 -1.60992090e-02 3.70805353e-01 -9.30082202e-02
-1.01336610e+00 6.40968800e-01 -1.42666531e+00 -3.47496927e-01
7.10513294e-01 6.47669733e-02 -2.48642623e-01 2.34880596e-02
-6.59684837e-01 5.16015828e-01 5.92169821e-01 4.63932976e-02
-1.25980127e+00 -8.04733098e-01 -9.14546967e-01 -1.90966427e-01
6.37279034e-01 -4.91732210e-01 1.46222246e+00 -1.08466578e+00
-9.94056880e-01 9.83253598e-01 -3.08096319e-01 -5.72454870e-01
4.95910943e-01 -5.33387244e-01 -2.34139219e-01 4.78349686e-01
2.41447195e-01 8.18166554e-01 8.45443010e-01 -1.26355422e+00
-1.00627220e+00 -2.47266412e-01 1.95712045e-01 1.79499924e-01
-1.54615819e-01 3.85921061e-01 -1.20134068e+00 -1.01055276e+00
1.40564933e-01 -9.15558040e-01 -2.10077107e-01 3.47205609e-01
-3.52274328e-01 -2.37177089e-01 9.40597951e-01 -6.60587728e-01
1.23642886e+00 -2.24758792e+00 3.75976115e-01 -2.26472154e-01
3.63415569e-01 2.78057963e-01 -1.57563552e-01 2.29187518e-01
-1.45328656e-01 5.35374247e-02 -2.41611660e-01 -8.20803285e-01
-4.36084978e-02 1.90732598e-01 -7.58847073e-02 6.85720563e-01
4.75486189e-01 9.21995640e-01 -1.16376877e+00 -7.04411447e-01
5.22194028e-01 5.38902462e-01 -5.28439164e-01 4.75512356e-01
-4.60456491e-01 4.09046948e-01 -3.56779546e-01 6.91564858e-01
5.20043194e-01 -1.85161203e-01 -1.43769950e-01 -1.81024298e-01
-3.80465508e-01 5.61337471e-02 -1.14748085e+00 1.93536222e+00
-2.30095059e-01 8.01745653e-01 2.39240140e-01 -1.11373615e+00
1.99826673e-01 3.60011399e-01 7.96613872e-01 -2.13345170e-01
3.23925018e-01 2.33482663e-02 -2.43959859e-01 -7.76959360e-01
3.27091664e-01 1.69394389e-01 6.39594859e-03 2.56551147e-01
2.69167453e-01 8.98617208e-02 5.67874134e-01 1.32289842e-01
9.49564755e-01 5.71274281e-01 9.09679532e-02 2.75783916e-03
2.75368452e-01 3.24385762e-02 7.25841343e-01 6.53490067e-01
-1.83725670e-01 9.53621626e-01 3.69005263e-01 -5.46355426e-01
-9.87468123e-01 -1.11751366e+00 -1.90457210e-01 1.22939181e+00
1.66678652e-01 -4.15050000e-01 -7.42176771e-01 -7.29752064e-01
-8.21664277e-03 3.58452499e-01 -8.70753646e-01 2.67488807e-01
-7.85675645e-01 -5.45006692e-01 2.62657464e-01 7.37604916e-01
2.71848083e-01 -9.01460528e-01 -8.69895339e-01 4.26175207e-01
-1.36967689e-01 -1.25796127e+00 -7.83458531e-01 2.21627191e-01
-8.09632838e-01 -1.17591321e+00 -8.64992440e-01 -8.98691893e-01
4.67696249e-01 3.99076104e-01 1.05027127e+00 -1.91086069e-01
-4.01867449e-01 3.64897668e-01 -3.64272058e-01 -1.79649666e-01
-2.84872088e-03 2.27633435e-02 -1.98787212e-01 2.74388105e-01
4.70446438e-01 -3.74661118e-01 -7.89539158e-01 1.72284290e-01
-1.03177488e+00 -2.09468044e-03 4.08857107e-01 4.36686426e-01
1.05863988e+00 -2.48894721e-01 -9.42661613e-03 -5.79676151e-01
1.17791802e-01 -6.96125507e-01 -7.11141407e-01 1.70631364e-01
2.56643016e-02 -3.07775587e-01 4.91928965e-01 -5.23504078e-01
-6.59056187e-01 2.91147619e-01 -6.10240065e-02 -1.00468397e+00
-3.75364959e-01 3.96146834e-01 2.80178338e-02 2.37773299e-01
3.75180155e-01 1.55142739e-01 -1.47864804e-01 -5.84525585e-01
5.96177518e-01 3.84369820e-01 6.87535524e-01 -3.92496198e-01
7.64737666e-01 6.88365698e-01 -2.69414544e-01 -6.90472841e-01
-1.04196382e+00 -8.10365558e-01 -9.01212573e-01 -3.60906512e-01
1.41328228e+00 -1.09175384e+00 -4.08381373e-01 5.15365422e-01
-1.20203745e+00 -6.23192012e-01 -1.73218608e-01 6.30254924e-01
-5.78126192e-01 2.54149795e-01 -8.15153301e-01 -5.65430939e-01
3.58704827e-03 -1.00274992e+00 1.21680677e+00 3.54088873e-01
-5.06315492e-02 -1.15348613e+00 1.04028441e-01 1.12067886e-01
-1.33110628e-01 4.47843760e-01 3.10591310e-01 -5.41065454e-01
-8.14749599e-01 -1.96195498e-01 -2.79967219e-01 3.39871854e-01
-5.89908063e-02 2.17357740e-01 -1.01182628e+00 -2.27749467e-01
-1.10381305e-01 -1.80132627e-01 1.10719764e+00 8.00516129e-01
1.53850996e+00 -9.49926898e-02 -5.91087818e-01 8.52671802e-01
1.35816610e+00 1.51331842e-01 4.26172465e-01 4.40861315e-01
8.90722394e-01 5.53271472e-01 7.80590236e-01 3.20245951e-01
5.59983730e-01 9.19709265e-01 4.48980719e-01 -3.57784212e-01
-1.50941551e-01 -3.55423056e-02 3.63414466e-01 3.35298926e-01
-2.03954782e-02 -4.20824409e-01 -7.67211914e-01 9.22730923e-01
-2.13752937e+00 -1.15828669e+00 -1.89012036e-01 1.95626128e+00
7.23971426e-01 1.28714338e-01 5.61790824e-01 -1.72491208e-01
6.57953382e-01 5.19900918e-01 -5.95581472e-01 4.13326882e-02
1.01885356e-01 -1.56553239e-01 4.56486464e-01 4.50246960e-01
-1.65835202e+00 1.03667605e+00 5.38471746e+00 5.26675403e-01
-1.09359276e+00 3.35417241e-01 6.13858581e-01 -3.45807463e-01
2.07339838e-01 -9.89812836e-02 -7.27059722e-01 5.39669454e-01
7.85267770e-01 1.22453511e-01 2.82560557e-01 6.68301642e-01
5.12830138e-01 -1.19261429e-01 -1.15209424e+00 8.95471752e-01
2.59541363e-01 -1.36823273e+00 -2.82596827e-01 -1.25234202e-01
5.92376530e-01 4.47170705e-01 5.66183589e-03 4.78366725e-02
1.32883117e-02 -6.74728870e-01 1.04369581e+00 5.16089797e-01
5.46423376e-01 -5.24712026e-01 4.50570762e-01 1.67951673e-01
-1.44864309e+00 -1.29822344e-01 -1.30038187e-01 3.31663996e-01
5.50286472e-01 4.74624485e-01 -4.97657567e-01 4.36754018e-01
9.68064249e-01 1.06801784e+00 -3.64777654e-01 1.51510394e+00
-2.12534830e-01 4.56877708e-01 -2.12208301e-01 4.34416324e-01
5.48848629e-01 -1.36311248e-01 3.95813227e-01 1.57408249e+00
3.33108664e-01 1.45053891e-02 6.74964309e-01 6.87784910e-01
-6.83984300e-03 -2.56855816e-01 -3.53601336e-01 -1.13622099e-01
4.21599537e-01 1.41385877e+00 -1.08250642e+00 -5.25891483e-01
-7.21988976e-01 1.02752459e+00 2.61754692e-01 5.07467747e-01
-1.30810785e+00 -3.26417267e-01 7.50392139e-01 4.55115229e-01
8.04404199e-01 -4.06152666e-01 1.72713362e-02 -1.24994195e+00
1.47632331e-01 -4.72711712e-01 5.22484124e-01 -6.89054489e-01
-1.01200867e+00 5.17526507e-01 1.74134240e-01 -1.29583251e+00
-7.87849203e-02 -4.63990718e-01 -5.97938716e-01 5.11935592e-01
-1.83860171e+00 -1.17969012e+00 -3.27767372e-01 5.21009266e-01
9.64315891e-01 2.68711656e-01 4.03001010e-01 3.98129106e-01
-8.06746304e-01 2.73946434e-01 1.48925573e-01 4.26428944e-01
6.77293956e-01 -1.30500376e+00 4.26995993e-01 1.22010672e+00
2.78438598e-01 3.58760208e-01 6.72628880e-01 -5.91864228e-01
-1.07450449e+00 -1.47876322e+00 8.03266943e-01 -6.16453826e-01
8.42502892e-01 -6.42184317e-01 -1.07020915e+00 8.35992992e-01
2.44319960e-01 5.44553161e-01 4.20021594e-01 -5.38794615e-04
-1.57548711e-01 -5.64768314e-02 -7.07786262e-01 4.28195089e-01
9.61168587e-01 -4.88210171e-01 -6.02995813e-01 5.72537780e-01
6.26646757e-01 -7.20996320e-01 -7.68338323e-01 4.79689501e-02
3.98993433e-01 -8.97000074e-01 1.02695084e+00 -5.49340487e-01
4.27005053e-01 -5.56160927e-01 1.44494558e-03 -9.69090223e-01
-4.43684429e-01 -5.82355797e-01 -5.36541760e-01 9.59976792e-01
2.36795142e-01 -1.55421823e-01 8.96597385e-01 3.13398868e-01
-5.16524851e-01 -9.58542049e-01 -7.33714521e-01 -9.35296535e-01
4.18404341e-02 -5.31501949e-01 2.48412907e-01 7.45011985e-01
-3.88712287e-01 3.66053879e-02 -2.19591334e-01 3.82640988e-01
5.99890530e-01 -2.21776427e-03 7.10561812e-01 -1.05155659e+00
-2.66969383e-01 -4.20045048e-01 -4.47935969e-01 -1.39188588e+00
6.73392266e-02 -5.98712683e-01 4.30890828e-01 -1.64373684e+00
1.60860166e-01 -1.70717210e-01 -4.19455916e-01 5.38040578e-01
-2.66382426e-01 3.74341339e-01 2.57615387e-01 2.54194766e-01
-1.00286305e+00 3.63013178e-01 1.04333293e+00 5.81021383e-02
-1.86274171e-01 -1.62629381e-01 -1.50255054e-01 7.80937076e-01
6.29456282e-01 -6.85124099e-01 -4.40611482e-01 -7.05115199e-01
-2.87028372e-01 3.63913588e-02 7.87213981e-01 -1.04801345e+00
2.43959710e-01 -2.59078562e-01 4.87644434e-01 -8.87418866e-01
6.69808447e-01 -7.57435858e-01 -1.82834581e-01 2.32802868e-01
-1.10991776e-01 2.36208260e-01 1.48978904e-01 6.80458486e-01
-3.15599352e-01 -6.32283464e-02 8.09634864e-01 -1.84279591e-01
-8.71528208e-01 7.16141522e-01 -4.09917086e-01 1.96010783e-01
1.47511649e+00 -2.77549356e-01 9.53418463e-02 -1.08069800e-01
-8.57520044e-01 4.09864992e-01 5.87481081e-01 6.48917854e-01
3.90585601e-01 -1.18207014e+00 -6.28844202e-01 2.87030882e-04
4.81559932e-02 2.19571188e-01 4.48415160e-01 1.03962398e+00
-5.78150213e-01 3.09022456e-01 1.52474403e-01 -1.00208461e+00
-1.25718653e+00 6.97357357e-01 4.78799820e-01 5.60541973e-02
-7.26480603e-01 1.12411845e+00 4.86454248e-01 -1.39976695e-01
2.98152030e-01 -6.37872398e-01 -2.24716008e-01 3.39222401e-01
7.71198094e-01 5.56177087e-02 -1.84430480e-01 -7.99317420e-01
-3.47745150e-01 6.29102290e-01 -2.28591263e-01 -8.65825638e-02
1.38023376e+00 -9.59201604e-02 8.87160897e-02 6.60459042e-01
1.36547363e+00 -4.54854757e-01 -1.99577832e+00 -2.89913058e-01
-1.00670017e-01 -6.10899568e-01 4.35514525e-02 -2.67891884e-01
-1.20963967e+00 8.87690365e-01 6.03271067e-01 3.35496187e-01
9.91515934e-01 2.51051188e-01 7.68599153e-01 -1.74131751e-01
2.29382515e-02 -1.13499391e+00 2.91955829e-01 4.57288951e-01
6.20951474e-01 -1.25588989e+00 -5.40169403e-02 -2.57758617e-01
-5.29452801e-01 1.16491032e+00 5.83191574e-01 -3.06803554e-01
7.87184775e-01 1.21364772e-01 3.62716219e-03 -2.25837961e-01
-6.60954237e-01 -2.86855042e-01 4.59745616e-01 3.87807071e-01
4.66985494e-01 -1.21384420e-01 -5.33297621e-02 3.40980291e-01
3.64465117e-01 9.41659808e-02 2.34858915e-01 1.18132293e+00
-4.32312101e-01 -8.64801526e-01 -2.70849317e-01 3.61389011e-01
-5.61585665e-01 1.55277207e-01 4.00328524e-02 8.56135190e-01
4.53418344e-01 8.30986202e-01 4.13126677e-01 -2.78920889e-01
2.62755781e-01 -2.63128191e-01 3.32570374e-01 -8.24415088e-01
-2.48727500e-01 5.69698215e-01 -1.04994446e-01 -8.33306491e-01
-7.01597095e-01 -1.06023693e+00 -1.37965357e+00 -8.70867446e-03
-1.44863352e-01 -1.15416251e-01 4.42835957e-01 1.05375552e+00
3.76480103e-01 7.05900967e-01 4.56833035e-01 -1.32965505e+00
1.51545331e-01 -6.14921331e-01 -4.35963869e-01 4.02990431e-01
4.92477417e-01 -6.41591668e-01 -1.33547122e-02 3.61935139e-01] | [9.10243034362793, -0.1208840012550354] |
21b91d90-715d-4ee8-b40c-78acea8d877c | dense-captioning-with-joint-inference-and | 1611.06949 | null | http://arxiv.org/abs/1611.06949v2 | http://arxiv.org/pdf/1611.06949v2.pdf | Dense Captioning with Joint Inference and Visual Context | Dense captioning is a newly emerging computer vision topic for understanding
images with dense language descriptions. The goal is to densely detect visual
concepts (e.g., objects, object parts, and interactions between them) from
images, labeling each with a short descriptive phrase. We identify two key
challenges of dense captioning that need to be properly addressed when tackling
the problem. First, dense visual concept annotations in each image are
associated with highly overlapping target regions, making accurate localization
of each visual concept challenging. Second, the large amount of visual concepts
makes it hard to recognize each of them by appearance alone. We propose a new
model pipeline based on two novel ideas, joint inference and context fusion, to
alleviate these two challenges. We design our model architecture in a
methodical manner and thoroughly evaluate the variations in architecture. Our
final model, compact and efficient, achieves state-of-the-art accuracy on
Visual Genome for dense captioning with a relative gain of 73\% compared to the
previous best algorithm. Qualitative experiments also reveal the semantic
capabilities of our model in dense captioning. | ['Li-Jia Li', 'Kevin Tang', 'Linjie Yang', 'Jianchao Yang'] | 2016-11-21 | dense-captioning-with-joint-inference-and-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Yang_Dense_Captioning_With_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Yang_Dense_Captioning_With_CVPR_2017_paper.pdf | cvpr-2017-7 | ['dense-captioning'] | ['computer-vision'] | [ 1.79888129e-01 4.45485413e-01 -9.10429657e-02 -3.12480122e-01
-8.19177568e-01 -6.90821052e-01 6.26842439e-01 2.01130629e-01
-1.36478737e-01 5.23812354e-01 4.70461458e-01 2.52672639e-02
5.19145668e-01 -1.68135509e-01 -1.03298402e+00 -5.49599648e-01
2.41713032e-01 7.84247756e-01 3.01501364e-01 3.89821455e-03
9.66808423e-02 1.39770970e-01 -1.66655171e+00 6.25923693e-01
5.37117362e-01 1.02281868e+00 7.35291481e-01 4.89104748e-01
-3.92698079e-01 8.93728018e-01 -4.54136878e-01 -3.52415860e-01
-7.08994791e-02 -1.45567924e-01 -1.01627290e+00 4.53921854e-01
9.06764567e-01 -3.51167411e-01 -1.64530024e-01 1.09400856e+00
5.04026599e-02 -1.97971120e-01 6.99855745e-01 -1.27718937e+00
-1.00982225e+00 5.75322866e-01 -7.88037002e-01 2.32770845e-01
4.22405690e-01 1.05947360e-01 1.32050490e+00 -1.24721766e+00
6.58910811e-01 1.44231558e+00 3.70394677e-01 7.05538094e-01
-1.28200006e+00 -5.18722117e-01 6.86382830e-01 2.51066327e-01
-1.48979747e+00 -3.59878004e-01 5.52287400e-01 -6.37265861e-01
8.11837614e-01 1.85783595e-01 6.14264011e-01 1.12927294e+00
-3.30594450e-01 1.01356792e+00 9.22093451e-01 -3.30423564e-01
2.73386478e-01 4.03930038e-01 1.92516759e-01 6.32198572e-01
3.33051801e-01 -2.67387092e-01 -3.50905210e-01 1.41668245e-01
5.78376710e-01 4.83761765e-02 -2.42131248e-01 -7.12456644e-01
-1.41589928e+00 7.14381337e-01 7.24728346e-01 2.92993844e-01
-2.12159932e-01 4.94787365e-01 7.34229013e-02 -4.80852008e-01
4.23649073e-01 3.74705911e-01 -3.17604899e-01 1.85141832e-01
-8.88572574e-01 2.06980869e-01 5.33334196e-01 1.39269197e+00
9.52634215e-01 -3.83051842e-01 -4.09138471e-01 7.36777067e-01
5.83495200e-01 6.32778525e-01 -3.81881557e-02 -7.43668199e-01
4.66184795e-01 6.79353774e-01 1.76772267e-01 -1.02116156e+00
-1.85690165e-01 -4.02386248e-01 -7.97240794e-01 -1.81707233e-01
1.58395410e-01 3.76088530e-01 -1.29846454e+00 1.72213995e+00
1.81872293e-01 4.31023568e-01 -5.48176281e-02 1.05798411e+00
1.18871427e+00 7.81179607e-01 6.88564599e-01 9.29906890e-02
1.80543971e+00 -1.34561789e+00 -7.12129593e-01 -9.02860045e-01
1.51253000e-01 -6.92327499e-01 9.34359550e-01 -1.95125222e-01
-1.01562786e+00 -5.48609376e-01 -7.84864008e-01 -4.51040745e-01
-4.94215280e-01 8.91854241e-02 7.47651219e-01 1.29975080e-01
-1.28773928e+00 -2.18411610e-01 -5.07486343e-01 -4.79652256e-01
7.55338788e-01 1.15156978e-01 -4.97923762e-01 -4.50911283e-01
-7.69326210e-01 8.43351305e-01 5.96794069e-01 5.46533242e-02
-1.23248291e+00 -6.89455807e-01 -1.20263672e+00 2.44223177e-01
2.48221427e-01 -8.32713425e-01 1.23707914e+00 -1.01620722e+00
-5.77803791e-01 1.21383631e+00 -4.25490499e-01 -3.78102005e-01
2.81473193e-02 -2.47416526e-01 -1.21004611e-01 4.62082446e-01
4.36682671e-01 1.48028135e+00 8.05152833e-01 -2.04128194e+00
-6.41043723e-01 -1.97209969e-01 2.41576388e-01 3.17293525e-01
-1.67004555e-01 1.95874386e-02 -1.12153280e+00 -4.89275306e-01
-2.40041595e-02 -8.92666876e-01 -1.78612232e-01 2.45171651e-01
-3.49689305e-01 -1.50993481e-01 9.34245348e-01 -5.78968227e-01
8.34962726e-01 -2.15706825e+00 2.00665042e-01 -2.54344136e-01
6.79219782e-01 2.05362812e-01 -2.29757637e-01 5.02992496e-02
-1.95687957e-04 1.43905461e-01 -3.65440369e-01 -8.77370238e-01
-4.30590361e-02 4.60483640e-01 -5.46761811e-01 1.63710967e-01
5.27648747e-01 1.40305007e+00 -1.11449420e+00 -8.48485470e-01
3.25623721e-01 5.17656326e-01 -4.23611522e-01 5.14120102e-01
-6.68911815e-01 1.85897797e-01 -3.81960750e-01 9.89618540e-01
6.98480010e-01 -9.30493891e-01 9.68414545e-02 -5.53569257e-01
1.16142459e-01 -2.40494594e-01 -8.44571114e-01 1.80501330e+00
-2.42027029e-01 8.61385047e-01 2.52617151e-01 -8.92981589e-01
5.70682168e-01 2.26755366e-01 1.19782992e-01 -4.83821183e-01
-1.75565667e-02 -1.37265146e-01 -4.58126366e-01 -5.82707107e-01
4.62490946e-01 -8.50463212e-02 -1.70012236e-01 1.68238968e-01
2.76030898e-01 -4.85557951e-02 1.07422277e-01 6.08227968e-01
6.64364398e-01 5.98608926e-02 3.50143731e-01 -1.91075712e-01
2.46084735e-01 1.89082652e-01 1.86260402e-01 7.65174687e-01
-3.49656343e-01 1.15442848e+00 3.61541361e-01 -3.64151865e-01
-1.11632931e+00 -1.08675420e+00 2.41866428e-02 1.09163678e+00
6.26465201e-01 -4.00677681e-01 -5.97195208e-01 -6.88286602e-01
-6.54023588e-02 4.67273951e-01 -8.97856832e-01 1.79828212e-01
-4.12415385e-01 -4.18032855e-01 2.12839007e-01 7.10961580e-01
4.61938381e-01 -1.07413113e+00 -4.16739166e-01 -1.43749624e-01
-5.77793360e-01 -1.70426846e+00 -5.27960658e-01 7.32968096e-03
-4.54286933e-01 -9.79620695e-01 -9.10514534e-01 -1.21917689e+00
9.95080113e-01 6.42928243e-01 1.76671541e+00 1.54764980e-01
-3.01548302e-01 7.01298952e-01 -4.84603792e-01 -3.88123453e-01
-1.44592255e-01 -2.05453008e-01 -2.80677736e-01 1.30576253e-01
3.90434474e-01 -1.47035196e-01 -7.26842523e-01 1.57711774e-01
-8.43769610e-01 5.43720424e-01 1.06985664e+00 7.00623751e-01
8.56548011e-01 -5.74490786e-01 1.00547478e-01 -6.44120991e-01
1.70637459e-01 -7.16844440e-01 -2.90335566e-01 5.63574135e-01
-2.67050743e-01 1.17133141e-01 1.00408986e-01 -3.79116714e-01
-7.98374057e-01 3.89927000e-01 6.91204965e-02 -7.59212792e-01
-4.69296992e-01 1.00774609e-01 -1.90947965e-01 1.62272714e-02
3.74470621e-01 2.94515431e-01 -4.83693406e-02 -5.39995074e-01
9.06041741e-01 5.21563768e-01 7.68391371e-01 -5.57758629e-01
6.82900488e-01 6.69503868e-01 -3.10159206e-01 -6.94016039e-01
-1.29372847e+00 -9.26503062e-01 -5.40286601e-01 -1.34980097e-01
1.31861949e+00 -1.44354391e+00 -5.73361933e-01 6.23715743e-02
-1.52040100e+00 2.63074674e-02 -1.92043290e-01 7.63267726e-02
-4.64968532e-01 4.66573060e-01 -2.89956659e-01 -4.94367123e-01
-4.11333680e-01 -1.12620592e+00 1.74988842e+00 2.33373091e-01
-4.43315133e-02 -9.16228056e-01 -6.45166561e-02 7.03298628e-01
2.61301309e-01 2.83412218e-01 6.59783423e-01 -3.66696268e-01
-8.01489472e-01 1.72444373e-01 -9.65090752e-01 2.42330432e-01
-2.01300412e-01 -4.07654703e-01 -1.18054891e+00 -2.70772308e-01
-3.68398041e-01 -5.21291077e-01 1.04396629e+00 3.31626356e-01
1.26901054e+00 -2.09787250e-01 -7.03846633e-01 4.29102987e-01
1.44938946e+00 -2.61255801e-01 6.04345620e-01 -3.89461070e-02
1.17919707e+00 5.26101172e-01 5.18094599e-01 1.02655351e-01
6.08955562e-01 7.39713967e-01 8.74981999e-01 -5.51016152e-01
-5.07958770e-01 -2.71337658e-01 1.78747490e-01 3.84308994e-01
2.18588859e-01 -3.69615942e-01 -1.06291878e+00 8.75310659e-01
-1.99854898e+00 -8.49746108e-01 -1.46655887e-02 1.54445422e+00
7.16857374e-01 -6.31502792e-02 -1.29632920e-01 -4.96442765e-01
8.57561171e-01 1.62595674e-01 -3.76062959e-01 -2.49987599e-02
-1.71838090e-01 -1.32556573e-01 3.49688083e-01 4.54007477e-01
-1.19015145e+00 1.23534465e+00 6.78724432e+00 6.04012430e-01
-7.87750840e-01 1.40106410e-01 8.37667942e-01 1.97899602e-02
-4.47301507e-01 1.47443682e-01 -9.80000257e-01 3.57749104e-01
5.91012299e-01 1.42845005e-01 -1.02561228e-02 1.02513361e+00
-1.94847867e-01 -6.98547736e-02 -1.28777289e+00 1.41860342e+00
6.54126644e-01 -1.59088135e+00 5.20447254e-01 -1.18998148e-01
7.84152031e-01 1.06634326e-01 -4.69411444e-03 2.47962028e-01
8.07221979e-02 -1.23926330e+00 1.02512503e+00 3.96949291e-01
9.30006146e-01 -2.71188796e-01 6.75607026e-01 -3.61912772e-02
-1.34524810e+00 7.53060952e-02 -2.67385572e-01 3.01547721e-02
2.89487749e-01 5.05946159e-01 -9.39300716e-01 2.20098689e-01
7.46199250e-01 9.40036654e-01 -8.32033455e-01 1.01256144e+00
-2.34573036e-01 3.64161849e-01 -1.14803292e-01 7.59277195e-02
4.04745519e-01 2.71140307e-01 3.16760421e-01 1.36252189e+00
-3.82187590e-02 1.22768201e-01 4.12460595e-01 1.30576932e+00
-8.72358382e-02 -1.98379546e-01 -5.57097256e-01 -1.58836454e-01
2.98474550e-01 1.37030733e+00 -8.36166322e-01 -4.74911243e-01
-5.64299583e-01 1.11782539e+00 5.83543181e-01 3.93145561e-01
-8.62842023e-01 1.83353901e-01 7.79924572e-01 2.32017934e-01
5.11204362e-01 -2.76572227e-01 -2.16164321e-01 -1.23102105e+00
1.82094380e-01 -5.98321080e-01 2.85815746e-01 -1.27432323e+00
-1.31254852e+00 7.15234160e-01 2.17533603e-01 -1.10166276e+00
-8.84292498e-02 -6.75282121e-01 -1.76356077e-01 7.14078605e-01
-1.66921437e+00 -1.66623557e+00 -8.02801192e-01 5.74040592e-01
8.24976623e-01 2.48582050e-01 9.35575724e-01 2.67721355e-01
-3.55836898e-01 3.48577082e-01 -3.61524612e-01 9.52168554e-02
5.67385793e-01 -1.33927202e+00 4.71952140e-01 8.07186306e-01
4.52340692e-01 4.57304806e-01 7.64611006e-01 -4.91693795e-01
-1.15496075e+00 -1.27011621e+00 8.99669766e-01 -8.84332895e-01
3.98994356e-01 -9.37204242e-01 -8.61727059e-01 6.56033695e-01
3.16294014e-01 2.74087399e-01 4.70032007e-01 1.92675576e-01
-7.47978568e-01 1.96836695e-01 -8.67601693e-01 5.98673940e-01
1.00443482e+00 -6.34809911e-01 -9.01817679e-01 5.50916612e-01
1.19335723e+00 -3.20729434e-01 -3.67525458e-01 2.32504323e-01
2.67289847e-01 -7.17394292e-01 1.22594762e+00 -3.60656589e-01
5.53653359e-01 -6.34256959e-01 -1.85050517e-01 -9.66229439e-01
-3.82483274e-01 -2.00764075e-01 -3.95324916e-01 1.17150080e+00
3.95363301e-01 9.74530056e-02 6.24420285e-01 4.85953242e-01
-1.78218842e-01 -5.77053785e-01 -5.63065946e-01 -4.77208108e-01
-2.98172206e-01 -4.39779907e-01 2.15567544e-01 8.39594781e-01
-8.65273699e-02 8.19778919e-01 -5.18255651e-01 3.59159559e-01
7.14621484e-01 2.40240529e-01 5.26467085e-01 -1.04957426e+00
-9.16041657e-02 -2.15823159e-01 -6.40100837e-01 -1.37682641e+00
2.25245282e-01 -7.88614333e-01 3.63982797e-01 -2.09977722e+00
8.35474014e-01 -3.19723189e-01 -1.40149668e-01 8.07260454e-01
-2.96990484e-01 6.92395210e-01 2.45457485e-01 2.93928266e-01
-1.26983297e+00 5.83642244e-01 1.15244913e+00 -4.88596946e-01
2.86908537e-01 -6.29812002e-01 -9.96982813e-01 5.12132347e-01
3.83353978e-01 -2.54684478e-01 -4.57967818e-01 -6.99266374e-01
1.38415828e-01 -3.27309132e-01 7.84569263e-01 -9.61381674e-01
1.29253015e-01 -5.18596359e-02 3.98615927e-01 -7.18042314e-01
5.55960298e-01 -9.80282187e-01 -1.18490346e-01 6.91788644e-02
-2.62053549e-01 3.88582703e-03 5.05361617e-01 9.17863607e-01
-5.01537204e-01 1.67243302e-01 7.35723972e-01 -2.99659312e-01
-1.36772037e+00 3.02751005e-01 -3.07277758e-02 1.84683517e-01
1.22387969e+00 2.00299751e-02 -4.64262396e-01 -3.79740387e-01
-7.82085836e-01 5.28275728e-01 4.19755906e-01 8.52558255e-01
8.25823605e-01 -1.26764524e+00 -7.27258205e-01 8.76975283e-02
8.15941870e-01 2.52090007e-01 3.78190041e-01 4.15252924e-01
-4.42865312e-01 6.63312078e-01 -1.18227363e-01 -9.58104372e-01
-1.38920915e+00 9.70841825e-01 1.57350272e-01 1.75505672e-02
-7.52987504e-01 1.06641507e+00 9.33520555e-01 1.64531156e-01
3.81006181e-01 -4.86329377e-01 -3.76559943e-01 -2.66341157e-02
7.41010904e-01 -3.64798665e-01 -2.66899467e-01 -9.33540523e-01
-5.50927460e-01 8.34879279e-01 -3.70721430e-01 5.16516976e-02
1.05114317e+00 -4.95056689e-01 -2.28246644e-01 2.91240871e-01
1.21773219e+00 -4.56656456e-01 -1.39692676e+00 -2.38680512e-01
-2.02047154e-01 -3.60237181e-01 2.74638012e-02 -8.97836208e-01
-9.03420389e-01 8.58332455e-01 6.41032338e-01 4.38133366e-02
9.63611901e-01 8.40915859e-01 4.87637609e-01 1.59197316e-01
1.21861987e-01 -7.78548360e-01 3.74102503e-01 4.65415537e-01
1.04515803e+00 -1.64743912e+00 -7.05191493e-02 -7.24325478e-01
-8.14468741e-01 7.06685603e-01 7.67957509e-01 1.11616768e-01
4.67114687e-01 4.26489860e-02 9.38733295e-03 -5.19699693e-01
-8.22928548e-01 -6.67802393e-01 6.31687522e-01 7.11434722e-01
2.01559961e-01 9.36122909e-02 2.08167896e-01 3.06877792e-01
3.19158465e-01 -4.30771530e-01 2.26026490e-01 6.17118835e-01
-4.94630277e-01 -7.11552680e-01 -3.29920322e-01 8.72377604e-02
-8.97447243e-02 -4.75012302e-01 -5.19764960e-01 6.87869847e-01
2.93313205e-01 1.00836003e+00 5.09665251e-01 -2.40610912e-01
1.01232678e-01 -6.65545389e-02 3.36212456e-01 -9.28413391e-01
-1.60439566e-01 2.88267178e-03 -4.10386771e-02 -6.66486323e-01
-6.11568332e-01 -3.70099843e-01 -1.19313931e+00 2.76232392e-01
-4.93562110e-02 6.34690076e-02 7.15811193e-01 1.07346809e+00
5.05729973e-01 4.54963386e-01 1.00796968e-01 -9.15384531e-01
-2.72562522e-02 -8.60912025e-01 -2.79783845e-01 7.21791327e-01
5.83067358e-01 -7.57820070e-01 -3.73873115e-01 4.42615151e-01] | [10.466954231262207, 1.4386290311813354] |
644ac3f6-1813-4787-a401-8689269bc014 | persformer-3d-lane-detection-via-perspective-1 | 2203.11089 | null | https://arxiv.org/abs/2203.11089v3 | https://arxiv.org/pdf/2203.11089v3.pdf | PersFormer: 3D Lane Detection via Perspective Transformer and the OpenLane Benchmark | Methods for 3D lane detection have been recently proposed to address the issue of inaccurate lane layouts in many autonomous driving scenarios (uphill/downhill, bump, etc.). Previous work struggled in complex cases due to their simple designs of the spatial transformation between front view and bird's eye view (BEV) and the lack of a realistic dataset. Towards these issues, we present PersFormer: an end-to-end monocular 3D lane detector with a novel Transformer-based spatial feature transformation module. Our model generates BEV features by attending to related front-view local regions with camera parameters as a reference. PersFormer adopts a unified 2D/3D anchor design and an auxiliary task to detect 2D/3D lanes simultaneously, enhancing the feature consistency and sharing the benefits of multi-task learning. Moreover, we release one of the first large-scale real-world 3D lane datasets: OpenLane, with high-quality annotation and scenario diversity. OpenLane contains 200,000 frames, over 880,000 instance-level lanes, 14 lane categories, along with scene tags and the closed-in-path object annotations to encourage the development of lane detection and more industrial-related autonomous driving methods. We show that PersFormer significantly outperforms competitive baselines in the 3D lane detection task on our new OpenLane dataset as well as Apollo 3D Lane Synthetic dataset, and is also on par with state-of-the-art algorithms in the 2D task on OpenLane. The project page is available at https://github.com/OpenPerceptionX/PersFormer_3DLane and OpenLane dataset is provided at https://github.com/OpenPerceptionX/OpenLane. | ['Junchi Yan', 'Yu Qiao', 'Jianping Shi', 'Conghui He', 'Hongyang Li', 'Xiangwei Geng', 'Jiajie Xu', 'Zehan Zheng', 'Yang Li', 'Chonghao Sima', 'Li Chen'] | 2022-03-21 | persformer-3d-lane-detection-via-perspective | https://arxiv.org/abs/2203.11089 | https://arxiv.org/pdf/2203.11089.pdf | null | ['3d-lane-detection', 'lane-detection'] | ['computer-vision', 'computer-vision'] | [-3.99272770e-01 -2.20657140e-01 -1.84956536e-01 -6.16019607e-01
-1.07589257e+00 -8.40832651e-01 6.96997762e-01 -3.80639195e-01
-3.13464254e-01 3.71031821e-01 2.03620613e-01 -6.76563501e-01
2.48233989e-01 -4.63426650e-01 -7.60590255e-01 -4.82998520e-01
-6.81660101e-02 4.31113839e-01 8.12458813e-01 -5.38335681e-01
2.59038866e-01 7.09508359e-01 -1.82530892e+00 -2.33114306e-02
7.02982068e-01 7.27704823e-01 3.40068221e-01 8.84787142e-01
2.83618212e-01 4.20784026e-01 -3.05900946e-02 -2.51019269e-01
7.52802849e-01 2.61461705e-01 -8.61814842e-02 1.30268946e-01
1.22083378e+00 -4.63059217e-01 -6.54682279e-01 6.71760440e-01
6.06582522e-01 1.82908729e-01 4.75904435e-01 -1.78252733e+00
9.68860686e-02 -3.13168198e-01 -7.19500482e-01 2.29324073e-01
3.62778842e-01 7.79831469e-01 8.77048314e-01 -1.12436771e+00
8.22940350e-01 1.35223758e+00 8.23488533e-01 5.64987725e-03
-1.01120651e+00 -7.66846657e-01 1.71526060e-01 2.96074927e-01
-1.37447882e+00 -8.25790465e-01 7.85638452e-01 -5.53183079e-01
9.85909462e-01 -8.48368183e-03 2.78153032e-01 1.35129249e+00
4.54090625e-01 8.28520894e-01 1.14143491e+00 4.42982092e-02
-2.02769428e-01 8.35120305e-02 1.39318824e-01 8.12946558e-01
2.44429156e-01 6.69937074e-01 -2.62820810e-01 2.04631999e-01
4.92596030e-01 -3.66356790e-01 5.25304787e-02 -1.22425354e+00
-1.49467123e+00 7.96521723e-01 3.36878747e-01 -4.24199879e-01
1.76556781e-01 2.26346496e-03 5.84594786e-01 2.00476527e-01
1.89279348e-01 2.41956890e-01 -4.22564715e-01 -2.46090770e-01
-4.24827188e-01 7.47218788e-01 5.18707216e-01 1.59930670e+00
1.11292410e+00 3.49209346e-02 2.70648479e-01 6.55792177e-01
2.25189000e-01 8.97334993e-01 -2.90558308e-01 -1.48534954e+00
8.85066867e-01 2.04286426e-01 2.07266346e-01 -8.61679792e-01
-8.54443371e-01 -3.76019508e-01 -3.09273809e-01 7.14647532e-01
6.72296464e-01 -9.81519818e-02 -7.44206727e-01 1.42956781e+00
3.47849578e-01 -8.42638761e-02 1.12881519e-01 1.14629829e+00
7.08194971e-01 5.61802745e-01 -2.62773782e-01 4.45159793e-01
1.19707108e+00 -1.17257035e+00 -4.06038761e-01 -8.45707357e-01
9.98880744e-01 -9.80436146e-01 1.02334535e+00 2.03969568e-01
-5.25001466e-01 -9.26332772e-01 -1.06472933e+00 -4.06732351e-01
-7.33417094e-01 3.16386431e-01 5.87860286e-01 3.86091799e-01
-9.16098118e-01 -4.67247993e-01 -2.68435419e-01 -4.91536200e-01
2.18975976e-01 -2.06320345e-01 -6.93062961e-01 -3.57998937e-01
-1.15814722e+00 1.28723168e+00 2.36975968e-01 -3.35749276e-02
-9.93335724e-01 -7.77802110e-01 -1.32995152e+00 -6.59443319e-01
8.69067371e-01 -5.02165377e-01 1.18693709e+00 -1.76870465e-01
-9.85897958e-01 1.13777208e+00 -2.80945878e-02 -4.54580456e-01
8.56920481e-01 -3.28267038e-01 -7.03455210e-01 -3.10913414e-01
5.14219820e-01 1.06128883e+00 6.56245351e-01 -1.51029623e+00
-1.34568524e+00 -1.92356408e-01 -8.37564096e-02 2.70708621e-01
9.15802240e-01 -5.77158630e-01 -4.18876082e-01 -2.02151239e-01
8.45118314e-02 -1.08887327e+00 -2.48896286e-01 3.44219990e-02
-4.62336987e-01 -8.63363966e-02 1.13766205e+00 -2.91095287e-01
6.53301001e-01 -2.50553417e+00 -2.43334815e-01 -1.02263376e-01
2.38480940e-01 -6.84566945e-02 -2.47203559e-01 4.35530394e-01
1.52050838e-01 -3.61162156e-01 -2.58775353e-02 -3.22615236e-01
3.08910310e-01 2.60016501e-01 -5.00405312e-01 7.38481581e-01
7.43909404e-02 8.25781643e-01 -9.70267057e-01 -3.16058546e-01
8.74516249e-01 3.91541682e-02 -3.06094944e-01 8.24291259e-02
2.23019987e-01 2.97770172e-01 -1.62783489e-01 7.06734121e-01
1.02226269e+00 4.78979111e-01 -4.81367052e-01 -3.72324377e-01
-8.02998185e-01 2.58961439e-01 -1.18400669e+00 1.71007633e+00
-5.31528115e-01 1.10442507e+00 1.43138081e-01 -4.28004563e-01
1.10615516e+00 -3.92133147e-01 1.79375246e-01 -1.01695454e+00
-1.07104285e-02 4.68046993e-01 -9.85843688e-02 -3.50399166e-01
8.89856577e-01 3.94039363e-01 -7.61970699e-01 -1.26938120e-01
-1.38393104e-01 -6.07007027e-01 2.86282331e-01 1.65836602e-01
1.08073330e+00 3.63412470e-01 2.14607909e-01 -1.66020036e-01
4.49938864e-01 3.65511388e-01 5.88003218e-01 8.70648265e-01
-7.28956819e-01 5.17023087e-01 4.34494942e-01 -5.78022480e-01
-1.43716538e+00 -1.48624575e+00 -2.96519250e-01 8.72375011e-01
5.24268806e-01 -1.58953995e-01 -2.04321176e-01 -7.93212116e-01
6.26964211e-01 9.78950083e-01 -3.84474605e-01 6.58012182e-02
-6.54658616e-01 -1.41909480e-01 5.37554979e-01 3.48292172e-01
6.70829296e-01 -4.09129620e-01 -8.78528416e-01 1.27293080e-01
-6.95643574e-02 -1.65043330e+00 -5.29284000e-01 3.18375170e-01
-3.46396744e-01 -1.17614818e+00 -1.03207171e-01 -4.70998287e-01
8.83920640e-02 9.11106229e-01 1.13780546e+00 -6.75829232e-01
-1.67738795e-01 2.51970917e-01 -2.27834255e-01 -4.04076785e-01
-4.34326172e-01 1.99672684e-01 6.03281222e-02 -2.59952009e-01
4.56965178e-01 -1.61197826e-01 -4.44560826e-01 8.66961539e-01
-7.85146654e-02 4.40332890e-01 8.28186452e-01 5.71899056e-01
4.41949934e-01 -3.04287463e-01 4.01821643e-01 -5.63689411e-01
-1.19126085e-02 -5.29621482e-01 -1.09494734e+00 -5.16050279e-01
-5.03294408e-01 -2.37304375e-01 3.81179392e-01 2.20749632e-01
-1.09932911e+00 3.82274061e-01 -2.17613265e-01 -5.79900384e-01
-7.68718779e-01 -1.48996979e-01 -2.61786163e-01 2.87918895e-02
7.08461821e-01 7.56134912e-02 1.72360092e-01 -2.25160927e-01
6.89699709e-01 6.50747061e-01 9.49675739e-01 -1.02586888e-01
1.10272229e+00 7.30489433e-01 1.59839347e-01 -7.51788199e-01
-7.52071142e-01 -7.88189828e-01 -9.65249836e-01 -4.78190184e-01
8.65251720e-01 -1.39537704e+00 -3.15991819e-01 5.50058603e-01
-7.74315000e-01 -6.74307942e-01 -7.80408233e-02 3.86357009e-01
-9.97217715e-01 1.33154646e-01 1.02321319e-02 -5.41932106e-01
4.16926265e-01 -1.29493678e+00 1.47990024e+00 -2.52198040e-01
1.21738389e-02 -8.43590915e-01 -3.10784280e-02 6.43413603e-01
1.60928920e-01 4.11082625e-01 5.58334053e-01 -1.96976647e-01
-9.28790212e-01 -2.93015391e-01 -4.09434855e-01 1.67392701e-01
-1.20210551e-01 -2.40612566e-01 -1.03889763e+00 -2.34676778e-01
-6.82414711e-01 -2.53280461e-01 1.11603534e+00 2.95094043e-01
6.19386435e-01 4.00851786e-01 -6.02108181e-01 7.35223591e-01
1.08346820e+00 8.85985140e-03 5.28597772e-01 6.69338465e-01
8.75483394e-01 6.66119039e-01 1.23992515e+00 1.69962913e-01
1.01058435e+00 9.51047480e-01 6.49174809e-01 -3.59672397e-01
-3.81006241e-01 -4.20927286e-01 4.63124961e-01 2.66542614e-01
4.61417466e-01 -3.12600076e-01 -1.14508760e+00 7.13744640e-01
-1.84797859e+00 -9.07403946e-01 -6.44437611e-01 2.04065084e+00
4.37974464e-03 7.35354900e-01 3.72640431e-01 -1.15508743e-01
2.82672733e-01 5.45074344e-01 -5.85003376e-01 -3.85086298e-01
-2.72860199e-01 -8.68081510e-01 1.36881006e+00 7.61683106e-01
-1.54103374e+00 1.11551893e+00 4.96702909e+00 7.06064105e-01
-9.01137531e-01 1.31161496e-01 2.99159706e-01 5.03728315e-02
-1.33397043e-01 4.22135182e-02 -1.39869595e+00 3.67573917e-01
8.19511235e-01 1.42283395e-01 9.14270729e-02 9.26970363e-01
5.75201094e-01 -4.82641965e-01 -9.69655037e-01 1.01753795e+00
6.57621324e-02 -1.20529556e+00 -4.69399273e-01 3.94121915e-01
5.53543329e-01 7.09142268e-01 9.44139585e-02 5.48396409e-01
2.65458882e-01 -4.68224168e-01 1.20557058e+00 3.26132476e-01
8.65447223e-01 -7.66362786e-01 7.66139448e-01 4.95532840e-01
-1.40703475e+00 -3.22419167e-01 -1.45553306e-01 7.97732791e-04
5.42582929e-01 3.27476025e-01 -1.09057403e+00 5.01854062e-01
6.95689321e-01 1.08083200e+00 -8.69078577e-01 1.00149691e+00
-1.04296148e-01 1.46586239e-01 -3.54970038e-01 3.45904320e-01
7.50909030e-01 -6.36087731e-02 9.94767129e-01 1.35860622e+00
2.10443467e-01 -3.55462193e-01 4.82456774e-01 6.38286948e-01
3.40655029e-01 -3.64256650e-01 -1.40028596e+00 8.51061761e-01
5.82421422e-01 1.28912449e+00 -2.73009688e-01 5.53622702e-03
-6.85063183e-01 3.94281715e-01 3.38096917e-02 2.98930645e-01
-1.07001603e+00 -4.49089438e-01 1.15907073e+00 5.11784554e-01
3.54505122e-01 -7.06988633e-01 -1.07001208e-01 -7.62409031e-01
2.84896623e-02 -4.64962035e-01 7.61810318e-02 -9.95063484e-01
-1.08818507e+00 3.11050802e-01 3.73842061e-01 -1.74962878e+00
-1.82547867e-01 -7.50076532e-01 -1.91411585e-01 7.28127539e-01
-1.92999518e+00 -1.41323662e+00 -5.72934389e-01 2.84575731e-01
9.97142553e-01 -3.19376677e-01 1.79703951e-01 4.18751031e-01
-1.82387143e-01 3.58554393e-01 1.85047518e-02 -1.80856422e-01
1.19149792e+00 -1.24574625e+00 8.97721529e-01 9.65572655e-01
-3.83949965e-01 -2.48514220e-01 8.16926062e-01 -4.49116468e-01
-1.58996749e+00 -1.46556139e+00 8.04796934e-01 -1.12424910e+00
6.39127314e-01 -9.12086844e-01 -5.56316376e-01 8.24752510e-01
-4.69421670e-02 1.45396188e-01 -2.46145856e-02 -2.04413950e-01
-2.42739320e-01 -3.83657545e-01 -8.31857204e-01 7.39639401e-01
1.58095729e+00 -2.70016283e-01 -4.81703699e-01 2.86662668e-01
5.35475850e-01 -9.22770321e-01 -5.03801525e-01 4.73553777e-01
6.32809997e-01 -1.22407711e+00 1.04884493e+00 1.44091770e-01
-6.72436431e-02 -7.99100518e-01 -4.06163335e-01 -1.24705648e+00
-5.27999580e-01 -4.14541483e-01 1.42261967e-01 8.71118605e-01
4.71383601e-01 -8.52925003e-01 5.41557729e-01 -6.78364933e-02
-1.00839365e+00 -3.90700579e-01 -1.18719447e+00 -9.26833332e-01
-1.16856031e-01 -8.24220359e-01 3.17709744e-01 4.50251639e-01
-6.79825485e-01 4.22920913e-01 -3.60786915e-01 3.51147890e-01
9.38027203e-01 5.62973134e-02 1.65685868e+00 -1.11988974e+00
3.95268410e-01 -4.19050068e-01 -6.50954425e-01 -1.47165191e+00
3.77790034e-01 -8.87316406e-01 2.33931914e-01 -1.38645446e+00
-4.12688702e-01 -7.23200858e-01 2.62788832e-01 2.99019605e-01
2.95210600e-01 5.37510663e-02 9.85316932e-02 9.69133899e-02
-7.15229809e-01 3.64468157e-01 1.23679006e+00 -1.75588727e-02
-2.80725379e-02 -6.30829036e-02 -2.85357296e-01 6.16959572e-01
7.11251020e-01 2.53242441e-03 -3.98500472e-01 -3.48025709e-01
-2.97953282e-02 -1.36496559e-01 7.53879011e-01 -1.21347320e+00
2.22416863e-01 -1.60384476e-02 2.50410348e-01 -1.42896521e+00
5.34817338e-01 -8.80615413e-01 1.30627872e-02 1.43854678e-01
4.44534002e-03 2.81404108e-01 5.10214627e-01 5.23006022e-01
3.60313356e-02 3.61167967e-01 8.33798349e-01 -4.36113104e-02
-1.88603592e+00 1.13590851e-01 -5.32743812e-01 1.27203614e-01
1.35764539e+00 -6.89113855e-01 -7.06722021e-01 -3.39169979e-01
-2.31974408e-01 8.14660788e-01 6.93378806e-01 1.14609706e+00
6.59558117e-01 -1.26615083e+00 -7.41969407e-01 7.52322257e-01
7.58755624e-01 9.84222293e-02 3.34429622e-01 1.04203391e+00
-5.06997883e-01 7.46268272e-01 -4.18681294e-01 -1.03342199e+00
-9.38852966e-01 5.03262758e-01 4.61836815e-01 2.12251142e-01
-8.44837129e-01 2.38523379e-01 3.93859535e-01 -8.64863098e-01
-1.09831437e-01 -3.60227883e-01 1.07569523e-01 -2.57197879e-02
1.65615499e-01 5.76644182e-01 1.14899904e-01 -1.23177946e+00
-4.70991582e-01 7.53404617e-01 6.17276691e-02 2.20919624e-01
8.75536799e-01 -6.81780159e-01 6.05102360e-01 5.96552074e-01
1.22168398e+00 1.12832189e-01 -1.93759382e+00 6.60972595e-02
-1.75043091e-01 -7.71781802e-01 2.42257223e-01 -6.75556779e-01
-5.45235097e-01 7.25676894e-01 9.55385149e-01 -1.81683928e-01
5.35372794e-01 1.35437503e-01 7.02625453e-01 3.91720682e-01
7.69423544e-01 -1.12224627e+00 -3.37835401e-01 9.98132229e-01
8.94073427e-01 -1.78256691e+00 -1.92557812e-01 -6.00103080e-01
-6.88967466e-01 9.43172157e-01 9.24662173e-01 -6.71260580e-02
4.08404231e-01 3.85996848e-01 4.84184444e-01 -2.63729423e-01
-7.90312946e-01 -3.49935353e-01 9.32021961e-02 8.31700265e-01
-1.82786465e-01 1.63040146e-01 2.82366365e-01 -1.58249885e-01
-3.30544531e-01 -4.78107095e-01 5.23491621e-01 8.95153046e-01
-6.01380587e-01 -6.50356054e-01 -3.20383698e-01 3.29833776e-01
5.27266979e-01 3.74445468e-01 -5.45796566e-03 1.38571894e+00
4.71758127e-01 8.23018253e-01 2.14519948e-01 -4.95726645e-01
1.01579511e+00 -9.46996585e-02 1.35000929e-01 -3.42631280e-01
1.58230305e-01 -3.87247726e-02 5.99011779e-01 -9.77227926e-01
1.30454645e-01 -1.09579098e+00 -9.97010291e-01 -3.97024274e-01
1.36742771e-01 -4.05473351e-01 6.51201546e-01 5.84734082e-01
6.84924901e-01 3.16475630e-01 8.00619304e-01 -1.35923243e+00
-2.47862011e-01 -5.67992330e-01 -4.58167344e-01 2.55455047e-01
7.06588209e-01 -1.19075215e+00 -5.01900554e-01 -1.89329177e-01] | [7.89096736907959, -1.7657333612442017] |
c1b93a93-df8d-40d5-9e37-7733687ce483 | a-semi-supervised-approach-to-message-stance | 1902.03097 | null | http://arxiv.org/abs/1902.03097v1 | http://arxiv.org/pdf/1902.03097v1.pdf | A semi-supervised approach to message stance classification | Social media communications are becoming increasingly prevalent; some useful,
some false, whether unwittingly or maliciously. An increasing number of rumours
daily flood the social networks. Determining their veracity in an autonomous
way is a very active and challenging field of research, with a variety of
methods proposed. However, most of the models rely on determining the
constituent messages' stance towards the rumour, a feature known as the "wisdom
of the crowd". Although several supervised machine-learning approaches have
been proposed to tackle the message stance classification problem, these have
numerous shortcomings. In this paper we argue that semi-supervised learning is
more effective than supervised models and use two graph-based methods to
demonstrate it. This is not only in terms of classification accuracy, but
equally important, in terms of speed and scalability. We use the Label
Propagation and Label Spreading algorithms and run experiments on a dataset of
72 rumours and hundreds of thousands messages collected from Twitter. We
compare our results on two available datasets to the state-of-the-art to
demonstrate our algorithms' performance regarding accuracy, speed and
scalability for real-time applications. | ['Nikolaos Kaplis', 'Ioannis Agrafiotis', 'Jason R. C. Nurse', 'Georgios Giasemidis'] | 2019-01-29 | null | null | null | null | ['rumour-detection'] | ['natural-language-processing'] | [-2.07911972e-02 2.53491044e-01 -4.81631309e-01 -3.20818543e-01
-1.93722039e-01 -5.06947756e-01 1.08709693e+00 6.80133104e-01
-3.17135155e-01 1.04690170e+00 3.42351258e-01 -2.87049085e-01
1.06814438e-02 -1.02212119e+00 -1.57383144e-01 -5.33508718e-01
-4.21491474e-01 7.51511693e-01 7.12731898e-01 -6.51912928e-01
7.50573695e-01 1.55092642e-01 -1.44624293e+00 2.36439019e-01
6.05066478e-01 7.74540305e-01 -6.24119580e-01 4.85461354e-01
-4.03893679e-01 1.63234770e+00 -7.37656176e-01 -6.43533647e-01
-1.41526341e-01 -4.97531325e-01 -1.15859044e+00 2.52696335e-01
-1.27503291e-01 -1.48118287e-01 1.68033578e-02 1.02467132e+00
1.49800837e-01 -2.90158302e-01 6.74891353e-01 -1.61477661e+00
-2.98349559e-01 9.22325134e-01 -7.43503928e-01 3.73107344e-01
5.45256734e-01 -4.18178320e-01 1.01192141e+00 -5.25202036e-01
8.37906897e-01 1.09619141e+00 8.49052370e-01 7.92488232e-02
-1.01724279e+00 -7.03849614e-01 -9.79663059e-02 1.48421496e-01
-1.06722796e+00 -2.95631826e-01 7.70958304e-01 -7.85545290e-01
3.37001443e-01 3.44266504e-01 6.69805706e-01 8.73043239e-01
1.91103637e-01 5.62426805e-01 1.72131479e+00 -3.68201196e-01
3.33906591e-01 6.35089576e-01 4.27143872e-01 8.23817074e-01
2.93965489e-01 -2.82709152e-01 -6.73774898e-01 -9.32019413e-01
2.77701765e-01 -1.58751696e-01 -1.91234186e-01 -4.59832735e-02
-1.02276933e+00 1.53299475e+00 4.23696816e-01 4.25147057e-01
-3.89293253e-01 -2.26896435e-01 4.32454616e-01 6.84301436e-01
1.27964199e+00 4.53852415e-01 -2.67073754e-02 -2.26121798e-01
-1.17242813e+00 2.96997935e-01 1.52272511e+00 3.63946080e-01
7.76679814e-01 -3.14720094e-01 4.56552535e-01 8.50507796e-01
4.45143849e-01 2.37635687e-01 4.44498420e-01 -5.65034330e-01
1.83857203e-01 9.21943486e-01 1.66899592e-01 -1.88051534e+00
-7.69595683e-01 -4.38162923e-01 -8.54904473e-01 6.71106726e-02
4.88900989e-01 -2.50616044e-01 -3.07277858e-01 1.15686893e+00
4.99826282e-01 6.38523623e-02 -2.98784137e-01 9.39132690e-01
6.58619046e-01 6.06841326e-01 -8.84836167e-02 -6.32263124e-01
1.33727193e+00 -8.92280400e-01 -8.62129867e-01 -5.17311171e-02
6.61687613e-01 -9.99852538e-01 3.74910533e-01 5.06952941e-01
-8.45293701e-01 2.63865858e-01 -9.41778421e-01 3.89319032e-01
-5.31699419e-01 -6.35951221e-01 7.20070779e-01 8.32578242e-01
-9.17721748e-01 6.48027122e-01 -2.40135878e-01 -6.50335133e-01
3.08696985e-01 8.74472111e-02 -4.41482104e-03 2.10193157e-01
-1.48413765e+00 8.61549854e-01 3.22653502e-02 -4.62440848e-01
-4.70044374e-01 -4.72542085e-02 -2.23536223e-01 -4.05415326e-01
4.09300894e-01 -3.09588373e-01 1.23554707e+00 -9.20906425e-01
-1.28130567e+00 9.89070296e-01 2.84801349e-02 -7.31540084e-01
9.48998511e-01 1.95403218e-01 -4.79845107e-01 2.43790776e-01
2.25716874e-01 1.31958015e-02 8.88671577e-01 -1.32845426e+00
-8.02306473e-01 -2.91333944e-01 1.76608622e-01 -1.29402071e-01
-4.24937457e-01 4.17483509e-01 2.96014935e-01 -4.55616564e-01
3.74161512e-01 -1.14724326e+00 -1.51623264e-01 -3.72296065e-01
-5.11125982e-01 -4.08998549e-01 8.60842526e-01 -4.15758580e-01
1.34330475e+00 -1.53989458e+00 -2.74516463e-01 3.38697463e-01
6.86905503e-01 1.82828069e-01 4.83211458e-01 8.51994634e-01
4.59782213e-01 5.00208735e-01 -1.78072438e-01 -2.12081805e-01
-2.08876908e-01 5.44004776e-02 -4.17724192e-01 8.39184463e-01
-1.24604858e-01 5.78009844e-01 -1.22983491e+00 -6.52954042e-01
-2.54662514e-01 2.64303505e-01 -1.14014953e-01 5.02441172e-03
-2.45034173e-01 4.45512712e-01 -5.13428330e-01 4.05138135e-01
4.17484939e-01 -7.42031336e-01 3.68732154e-01 4.02020067e-01
-1.00788996e-01 5.25403917e-01 -7.65981436e-01 6.57679796e-01
-4.77547459e-02 7.32010603e-01 1.70309156e-01 -8.01797628e-01
1.13254404e+00 3.22161704e-01 5.62670827e-01 -3.67496669e-01
3.45826358e-01 3.90803784e-01 -2.32763693e-01 -5.98093569e-01
7.44085133e-01 -2.56002247e-01 4.32756431e-02 1.28623366e+00
-5.61658442e-01 1.47091687e-01 3.15155238e-01 5.72487593e-01
1.17903936e+00 -5.27979493e-01 5.94724894e-01 -4.21186030e-01
4.80150640e-01 4.05119985e-01 1.08833946e-01 6.52547061e-01
-2.93676794e-01 1.97959632e-01 8.77908885e-01 -4.41435754e-01
-8.27608228e-01 -4.73673105e-01 5.09100221e-02 1.39751852e+00
2.70739883e-01 -3.29983175e-01 -7.27327466e-01 -7.95645177e-01
9.73626003e-02 4.52098310e-01 -6.30391836e-01 4.58968043e-01
-3.56038302e-01 -1.07721138e+00 5.59480190e-01 -2.67418861e-01
4.75067854e-01 -9.31272805e-01 -2.86158204e-01 4.41188067e-01
-4.63627666e-01 -1.10388768e+00 9.18625519e-02 -3.37702543e-01
-7.17986763e-01 -1.31476176e+00 -3.01479906e-01 -4.10196334e-01
4.87710595e-01 5.75212955e-01 1.28108156e+00 7.03275263e-01
2.20556885e-01 3.19520608e-02 -6.11440897e-01 -4.16099906e-01
-9.19618309e-01 3.02808106e-01 9.36855972e-02 3.54330897e-01
3.59897882e-01 -7.94107378e-01 -5.18039584e-01 7.80416310e-01
-8.95568192e-01 -3.90919186e-02 2.46514484e-01 4.31706309e-01
-2.15971127e-01 1.56456828e-01 1.05377746e+00 -1.64050806e+00
1.14795041e+00 -1.20150256e+00 -3.54137629e-01 -2.21621528e-01
-1.01409638e+00 -3.18251222e-01 3.81626368e-01 -1.54654786e-01
-5.43497503e-01 -5.62781096e-01 4.84622689e-03 4.23946470e-01
7.44438171e-02 7.93689907e-01 8.45091403e-01 -1.66382641e-01
9.46918011e-01 6.66311085e-02 3.79928648e-01 -3.38788778e-01
2.15728551e-01 1.23335445e+00 -1.52528286e-01 -9.96568054e-02
7.71521986e-01 8.82401586e-01 -1.37179956e-01 -1.00146806e+00
-1.24647260e+00 -9.14522767e-01 -3.79247487e-01 -6.13965690e-01
3.92017931e-01 -5.33768177e-01 -7.14143574e-01 5.64295828e-01
-1.17431700e+00 1.66497286e-02 3.85701269e-01 9.03803781e-02
-2.13932410e-01 4.00591791e-01 -9.90593910e-01 -9.76545513e-01
-3.92186940e-01 -7.18580484e-01 3.79161894e-01 -2.06102461e-01
-5.30182600e-01 -1.32261920e+00 3.51504445e-01 6.33152664e-01
9.05122221e-01 4.78816330e-01 5.86444139e-01 -1.03059971e+00
-2.90509522e-01 -4.24325198e-01 -5.22828460e-01 -8.70696381e-02
2.59544373e-01 -1.65141135e-01 -8.32807243e-01 -2.32285649e-01
9.86938551e-02 -4.93806809e-01 6.81463659e-01 -2.43792459e-01
2.78010070e-01 -8.93956661e-01 -3.74954969e-01 -4.78716224e-01
1.19148028e+00 -4.68456179e-01 4.19903159e-01 5.53432763e-01
3.90193850e-01 9.81252730e-01 5.40987670e-01 6.83847368e-01
6.75380886e-01 4.72110838e-01 6.44639432e-01 1.75261676e-01
1.16917379e-01 -1.91032931e-01 2.43065223e-01 1.17824376e+00
-1.01335816e-01 -2.14195386e-01 -1.10419202e+00 3.88017446e-01
-1.99823904e+00 -1.13110149e+00 -7.42785037e-01 1.84298170e+00
8.77295732e-01 4.12548721e-01 6.87778413e-01 5.92382133e-01
8.93494844e-01 4.69534248e-01 1.43061206e-01 -4.41446155e-01
1.56606250e-02 -5.00565946e-01 4.07362700e-01 6.96217477e-01
-9.81330931e-01 5.77793956e-01 6.31566286e+00 4.42720324e-01
-1.27008867e+00 3.96520376e-01 6.17326438e-01 2.79718518e-01
-1.55976623e-01 -1.06848655e-02 -6.16249382e-01 5.14185905e-01
1.04654872e+00 -1.45139441e-01 3.70215505e-01 5.90063930e-01
2.89030284e-01 -4.19886440e-01 -6.54594958e-01 5.47529936e-01
3.32755178e-01 -1.47236240e+00 -2.58182585e-01 2.12644666e-01
8.34525406e-01 3.15337539e-01 -4.60509747e-01 8.33076239e-02
3.71241152e-01 -8.36821556e-01 7.23986328e-01 2.32578829e-01
1.03485398e-01 -5.04898012e-01 8.57738435e-01 8.26469839e-01
-6.31308198e-01 9.15313289e-02 -1.32570222e-01 -6.28428936e-01
3.55199605e-01 1.17381907e+00 -1.26809466e+00 2.38249823e-01
3.76060069e-01 7.01519907e-01 -6.63242280e-01 8.67866099e-01
-3.74610364e-01 1.10385132e+00 -2.79036254e-01 -6.46807909e-01
4.76137221e-01 -7.98091814e-02 6.66660964e-01 1.35523486e+00
-2.04000384e-01 -1.34895584e-02 3.89829725e-01 4.18539733e-01
-1.62091985e-01 3.96911055e-01 -6.13949001e-01 -1.64754853e-01
4.33223993e-01 1.37585104e+00 -1.11655259e+00 -4.17863488e-01
-2.47457489e-01 4.77006197e-01 3.97958100e-01 4.27230187e-02
-6.77594423e-01 2.57219411e-02 9.94146243e-02 6.97350144e-01
-2.39981771e-01 -2.41000473e-01 -1.22063167e-01 -9.41896141e-01
-3.38991404e-01 -9.01217818e-01 3.18455517e-01 -2.20017657e-01
-1.78012168e+00 7.50210941e-01 -2.81789333e-01 -1.10032856e+00
-3.04893702e-01 -1.55284896e-01 -3.48123610e-01 3.11917663e-01
-1.72894859e+00 -9.42202568e-01 -3.60079467e-01 3.05979639e-01
1.73321843e-01 5.35153262e-02 6.84877932e-01 3.03809285e-01
-8.02845508e-02 -4.51292237e-03 -2.36650519e-02 3.35772544e-01
6.32648289e-01 -1.00955784e+00 2.46121809e-01 1.65284246e-01
8.63139331e-02 3.23108524e-01 1.13281941e+00 -6.83009326e-01
-9.74811673e-01 -9.26416457e-01 1.31886780e+00 -4.89363253e-01
1.32564187e+00 -3.23318630e-01 -1.05841804e+00 4.64036405e-01
2.59966075e-01 -1.00102082e-01 7.13799953e-01 2.66007423e-01
-4.07032520e-01 4.36259449e-01 -1.20628428e+00 3.89238268e-01
7.52059340e-01 -2.75371671e-01 -5.60285628e-01 9.24277782e-01
2.70479739e-01 2.91137844e-02 -7.90990353e-01 3.88133782e-03
4.06648964e-01 -1.48909438e+00 3.19683850e-01 -5.05608916e-01
6.15606904e-01 -9.26556885e-02 1.33585259e-01 -1.27912426e+00
-5.70188016e-02 -7.82616258e-01 -6.23607337e-02 1.19376028e+00
6.66851640e-01 -1.13538098e+00 9.70562577e-01 1.64988220e-01
6.86405003e-01 -8.73261690e-01 -6.61969543e-01 -5.72306395e-01
-1.03801727e-01 -2.13756204e-01 3.44668180e-01 1.58013344e+00
4.83978182e-01 8.44937801e-01 -6.83708787e-01 -2.06548050e-01
7.20455050e-01 2.45149538e-01 8.69166851e-01 -1.66377354e+00
1.25770420e-01 -4.17257756e-01 -3.90906423e-01 -6.50594413e-01
-3.77198532e-02 -7.62158573e-01 -2.03242540e-01 -1.57137740e+00
2.24587202e-01 -8.21477473e-01 1.35862485e-01 2.38206476e-01
2.20143676e-01 6.24931753e-01 -5.32355942e-02 8.95357788e-01
-8.40213537e-01 6.30314788e-03 9.22762454e-01 -5.58913872e-02
-1.18089676e-01 1.85549781e-01 -4.21802819e-01 1.04330564e+00
1.02619946e+00 -8.93863082e-01 -1.02505103e-01 1.82937756e-01
8.81265581e-01 1.38307363e-02 2.35932037e-01 -7.21820891e-01
4.78626221e-01 -8.88283327e-02 -4.38000381e-01 -3.34769338e-01
2.33541518e-01 -3.76604557e-01 -1.14906713e-01 5.53848922e-01
-4.35916752e-01 -9.48114470e-02 -5.46116769e-01 9.35194433e-01
-2.46779725e-01 -2.84501672e-01 8.31746221e-01 -2.54314393e-01
-3.73344332e-01 5.43769507e-04 -5.64836800e-01 2.65010774e-01
9.36775684e-01 1.48730040e-01 -7.96199322e-01 -1.03265607e+00
-5.39869249e-01 8.68850276e-02 5.93088329e-01 4.24795538e-01
3.23832780e-01 -8.47233713e-01 -1.19942427e+00 -2.74932534e-01
1.10826492e-01 -7.38445401e-01 -4.85621214e-01 1.31830323e+00
-6.21167839e-01 1.94679797e-01 1.18051693e-01 -5.96803069e-01
-1.09913647e+00 3.67209464e-01 -2.63017304e-02 -4.42012310e-01
-3.62828285e-01 4.10343468e-01 -6.02936983e-01 -3.55511129e-01
7.21399393e-03 4.87779438e-01 -5.51540315e-01 6.30160630e-01
6.35533631e-01 6.81588531e-01 9.07827020e-02 -9.06024516e-01
-2.58855581e-01 1.05981655e-01 -1.32238075e-01 -4.76462059e-02
1.08198273e+00 -3.80989760e-01 -8.26157808e-01 7.04629660e-01
1.18214488e+00 3.17825586e-01 -3.19904625e-01 -5.13174593e-01
5.65662324e-01 -5.34295976e-01 2.11716276e-02 -6.19890153e-01
-7.73881495e-01 3.91624123e-01 -5.08362427e-02 1.51844156e+00
3.50886852e-01 5.78160845e-02 7.34060049e-01 1.86626166e-01
7.75181174e-01 -9.09087658e-01 1.85668513e-01 5.21528482e-01
7.08381057e-01 -1.54203129e+00 4.09278184e-01 -8.61130714e-01
-5.87117374e-01 9.65037525e-01 -7.09615275e-02 -2.05376610e-01
9.53694522e-01 2.12698821e-02 3.66023809e-01 -5.55945396e-01
-6.79070532e-01 6.97376132e-02 -1.68414563e-01 4.64023463e-02
7.14088380e-01 6.64868951e-02 -9.32736576e-01 3.16892304e-02
-3.12875986e-01 1.56601239e-02 1.00350487e+00 1.17736197e+00
-9.60752428e-01 -8.98640990e-01 -4.23048943e-01 6.91984057e-01
-7.78877735e-01 1.45385832e-01 -8.81735086e-01 6.79060757e-01
-3.02986473e-01 1.66182029e+00 -4.33451027e-01 -4.84053075e-01
-1.47864699e-01 -1.93970781e-02 -1.38798118e-01 -6.08832717e-01
-7.36937821e-01 -3.37532848e-01 7.52795219e-01 -2.50455260e-01
-9.59590137e-01 -5.23041070e-01 -1.01531303e+00 -1.04534817e+00
-6.35688305e-01 6.09925687e-01 8.73967409e-01 1.06267059e+00
3.62893343e-02 -1.07186303e-01 1.06286275e+00 -6.79091871e-01
-6.82647884e-01 -1.12049115e+00 -7.51901805e-01 5.74002087e-01
2.51633942e-01 -6.77549064e-01 -7.88981199e-01 -1.45361856e-01] | [8.258944511413574, 10.15349292755127] |
37d71184-b012-47f6-8621-6bc19997083d | 190503415 | 1905.03415 | null | https://arxiv.org/abs/1905.03415v2 | https://arxiv.org/pdf/1905.03415v2.pdf | PPGNet: Learning Point-Pair Graph for Line Segment Detection | In this paper, we present a novel framework to detect line segments in man-made environments. Specifically, we propose to describe junctions, line segments and relationships between them with a simple graph, which is more structured and informative than end-point representation used in existing line segment detection methods. In order to extract a line segment graph from an image, we further introduce the PPGNet, a convolutional neural network that directly infers a graph from an image. We evaluate our method on published benchmarks including York Urban and Wireframe datasets. The results demonstrate that our method achieves satisfactory performance and generalizes well on all the benchmarks. The source code of our work is available at \url{https://github.com/svip-lab/PPGNet}. | ['Zhengxin Li', 'Ziheng Zhang', 'Ning Bi', 'Kun Huang', 'Shenghua Gao', 'Jinlei Wang', 'Jia Zheng', 'Yanyu Xu', 'Weixin Luo'] | 2019-05-09 | ppgnet-learning-point-pair-graph-for-line | http://openaccess.thecvf.com/content_CVPR_2019/html/Zhang_PPGNet_Learning_Point-Pair_Graph_for_Line_Segment_Detection_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Zhang_PPGNet_Learning_Point-Pair_Graph_for_Line_Segment_Detection_CVPR_2019_paper.pdf | cvpr-2019-6 | ['line-segment-detection'] | ['computer-vision'] | [-9.39604566e-02 3.63997743e-02 -2.37014845e-01 -3.69145781e-01
-4.38893676e-01 -6.83903813e-01 4.12102997e-01 8.40308368e-02
-1.15379365e-02 4.50285137e-01 -1.87928334e-01 -5.81248522e-01
2.16627583e-01 -9.75999475e-01 -8.99400532e-01 6.92153582e-03
-2.49120548e-01 5.44641092e-02 7.42931306e-01 -3.80314767e-01
3.71743202e-01 8.67495775e-01 -7.67889917e-01 -1.01759560e-01
6.49102330e-01 1.02613425e+00 -2.43552536e-01 8.67207587e-01
-7.25772679e-02 5.66505671e-01 -3.70370895e-01 -5.27204752e-01
3.91568929e-01 -2.24780411e-01 -9.96241033e-01 2.38374367e-01
7.26908565e-01 -5.56583643e-01 -7.75676966e-01 8.50797415e-01
2.81615853e-01 -5.98259196e-02 6.20022833e-01 -1.27854073e+00
-3.18995476e-01 6.99477017e-01 -9.61431086e-01 4.02910173e-01
5.63567340e-01 3.24556753e-02 9.86848056e-01 -9.03572977e-01
7.10360169e-01 9.32463467e-01 9.77999747e-01 -8.42180699e-02
-8.69767487e-01 -4.38422382e-01 2.30412498e-01 1.91832762e-02
-1.68648577e+00 -3.79684120e-01 1.09362102e+00 -4.32952702e-01
7.06497550e-01 2.07560241e-01 8.49278331e-01 7.98358023e-01
1.01827793e-01 1.01601458e+00 5.70154488e-01 -4.12255585e-01
-2.13834569e-01 -4.71859127e-01 4.02581483e-01 1.19386721e+00
3.38891417e-01 -1.13848723e-01 -1.24563519e-02 1.50227711e-01
1.33863366e+00 -1.45975769e-01 -3.98418576e-01 -6.05357111e-01
-1.23166358e+00 5.84530532e-01 8.00547659e-01 1.06084391e-01
-8.28459039e-02 5.14221668e-01 2.67286688e-01 4.30679396e-02
3.52796555e-01 1.00822717e-01 -3.19073093e-03 -1.07945651e-01
-9.47998405e-01 8.63211453e-02 7.99070120e-01 1.40982890e+00
9.09625471e-01 -1.03370354e-01 1.50347920e-02 7.85264194e-01
2.56254584e-01 4.05679882e-01 -2.00163409e-01 -1.05419862e+00
6.10381186e-01 6.46738172e-01 -1.25187948e-01 -1.29837382e+00
-7.06554115e-01 -5.39855301e-01 -6.40962064e-01 5.01423478e-02
3.74820083e-01 -4.10735369e-01 -9.23152268e-01 9.86948311e-01
3.19524050e-01 3.90701473e-01 -3.43888581e-01 7.78842568e-01
1.07896674e+00 5.74675262e-01 -4.97990727e-01 4.46210653e-01
1.12141371e+00 -1.50113297e+00 -2.69266397e-01 -2.72144288e-01
6.93478942e-01 -7.82074928e-01 8.77973616e-01 7.17166588e-02
-9.76661563e-01 -5.65238655e-01 -1.32307148e+00 -4.39496011e-01
-2.09100649e-01 2.82027483e-01 6.71376288e-01 1.96840212e-01
-9.33450162e-01 4.91022438e-01 -9.14169669e-01 -6.22737169e-01
6.68710828e-01 1.51672419e-02 -3.21992606e-01 2.91503798e-02
-7.59640574e-01 2.33441263e-01 3.49217027e-01 2.47961044e-01
-4.39791292e-01 -2.58068383e-01 -1.10299993e+00 1.78670466e-01
4.25026178e-01 -8.10102344e-01 1.38446319e+00 -5.61201394e-01
-1.32427084e+00 9.01431799e-01 -8.38998854e-02 -5.61655045e-01
7.93484509e-01 -4.55519676e-01 -2.96089649e-01 4.41808373e-01
9.31074470e-02 6.33998513e-01 4.22276020e-01 -1.27510703e+00
-6.80476904e-01 3.33211757e-02 3.11092466e-01 4.32245024e-02
2.58651733e-01 -1.47045270e-01 -1.11523712e+00 -5.73982894e-01
3.50711882e-01 -9.55423653e-01 -3.37235123e-01 2.70312518e-01
-1.21062982e+00 -2.83464223e-01 8.83238494e-01 -3.98144215e-01
1.35430062e+00 -2.02361917e+00 -4.72350895e-01 5.95286310e-01
5.35604715e-01 2.28334039e-01 3.96408178e-02 6.99954510e-01
-5.53613305e-02 1.58064976e-01 -4.09990430e-01 -2.59134442e-01
-8.08040518e-03 -4.65578884e-02 -2.01172546e-01 6.56387150e-01
-3.10424319e-03 1.01118875e+00 -6.29893720e-01 -6.15551889e-01
4.34150457e-01 4.44421738e-01 -1.30126342e-01 -1.31314516e-01
2.03804416e-03 1.97837576e-01 -5.63044071e-01 7.73436844e-01
9.24412966e-01 -3.98141801e-01 -2.26275772e-01 -2.15339854e-01
-2.65193731e-01 -2.85130609e-02 -1.15043521e+00 1.87115109e+00
3.19234394e-02 1.14075267e+00 -4.74652916e-01 -8.02550852e-01
1.12030065e+00 -2.07533970e-01 5.51210821e-01 -5.95621288e-01
2.74111837e-01 -4.42964537e-03 -1.96420580e-01 -3.46283615e-01
4.62488443e-01 9.19961452e-01 -1.24327950e-01 4.96006645e-02
-1.38717771e-01 -2.90069908e-01 4.86899555e-01 3.41229826e-01
1.10681868e+00 2.93936044e-01 4.89881724e-01 -1.41743988e-01
5.37979662e-01 2.11438239e-01 5.13214767e-01 7.20749259e-01
-2.37096831e-01 9.10133243e-01 6.36084437e-01 -6.68820143e-01
-1.09722626e+00 -1.37702107e+00 -8.88151750e-02 5.50633252e-01
7.03124404e-01 -8.25356960e-01 -7.56535470e-01 -6.09602153e-01
-1.93925112e-01 3.43578190e-01 -3.29980880e-01 3.47251058e-01
-8.87304962e-01 3.08970809e-02 7.67577648e-01 7.81813323e-01
1.06218743e+00 -5.79237521e-01 -3.33036304e-01 4.29600989e-03
-2.08205596e-01 -1.44553185e+00 -7.06501961e-01 -3.38261604e-01
-7.05642879e-01 -1.27926171e+00 -6.84282899e-01 -1.03559601e+00
8.26271176e-01 5.00039041e-01 1.21976233e+00 3.74381274e-01
-2.98880339e-01 3.02054286e-01 -2.40916312e-01 -5.09671867e-01
2.37668708e-01 3.75229567e-01 -6.42154276e-01 -2.93031007e-01
2.36535758e-01 -4.74317342e-01 -1.04874778e+00 4.40695316e-01
-5.10197401e-01 3.43621910e-01 3.08488488e-01 2.92336643e-01
7.30672419e-01 -1.75075471e-01 -4.71428782e-02 -1.00071871e+00
4.70481932e-01 -3.27464819e-01 -8.55810344e-01 -4.22022976e-02
-2.62094051e-01 -3.16043347e-01 3.46664965e-01 3.06374818e-01
-5.57851791e-01 2.65986353e-01 -5.26966989e-01 -3.02599490e-01
-4.61005002e-01 4.01020765e-01 1.10602535e-01 -1.82237789e-01
5.76810241e-01 4.74009588e-02 -4.49752957e-01 -2.65978515e-01
4.81389970e-01 3.41229260e-01 8.08704197e-01 -5.29616177e-01
1.15019441e+00 8.44785690e-01 6.98716119e-02 -1.04617453e+00
-6.93075001e-01 -7.76544809e-01 -1.03808057e+00 -3.24425220e-01
5.67744672e-01 -7.74456203e-01 -5.04416347e-01 4.35859114e-01
-1.27902770e+00 -5.54655433e-01 -1.33688062e-01 4.73721772e-01
-8.38253140e-01 6.57842577e-01 -8.53105068e-01 -3.86611372e-01
-3.78249973e-01 -9.89456534e-01 1.16766894e+00 5.16578376e-01
-8.86835977e-02 -1.08426774e+00 1.91840708e-01 1.18241288e-01
7.76859447e-02 8.17732096e-01 4.24629390e-01 -4.12181407e-01
-9.21319723e-01 -3.11501175e-01 -5.10696292e-01 1.13682285e-01
1.33444313e-02 6.10651672e-01 -6.26631260e-01 -4.92742211e-02
-8.38547707e-01 1.09444655e-01 9.16106820e-01 4.53021765e-01
1.32006240e+00 -2.42037669e-01 -7.41064668e-01 1.26050544e+00
1.46663463e+00 1.63909286e-01 8.93625975e-01 6.83218658e-01
9.39838767e-01 1.56742170e-01 3.75137478e-01 4.32192504e-01
7.13210404e-01 4.94549483e-01 3.75262499e-01 -6.69373453e-01
-3.66514564e-01 -4.39809054e-01 -1.23404246e-03 5.30493975e-01
-3.68374705e-01 -5.73026896e-01 -1.14435554e+00 4.05335337e-01
-2.04121447e+00 -9.96108830e-01 -6.82730734e-01 1.67770350e+00
2.02219158e-01 5.10015070e-01 3.29933733e-01 7.45988190e-02
7.42537975e-01 3.62577260e-01 -4.35607851e-01 -4.01916206e-01
-1.93292707e-01 -9.98988561e-03 8.21874499e-01 4.98927444e-01
-1.51866066e+00 1.18046522e+00 6.59179449e+00 5.85627973e-01
-1.11676300e+00 -4.82312739e-01 5.94099224e-01 4.65790957e-01
-5.97216561e-02 1.47304172e-02 -6.41474485e-01 1.76629588e-01
3.63388240e-01 -2.98852503e-01 -4.42893431e-02 9.49763834e-01
2.89746046e-01 3.82843055e-02 -1.00496459e+00 9.99161899e-01
-4.54785079e-02 -1.62080979e+00 -3.07664663e-01 -3.78187038e-02
6.31036878e-01 4.54081088e-01 -8.23608115e-02 -3.99068147e-02
3.87603670e-01 -8.76664162e-01 6.38661981e-01 5.08913159e-01
7.61721194e-01 -6.50328636e-01 5.20766616e-01 1.04782470e-01
-1.66828430e+00 4.11122680e-01 -2.74862707e-01 -5.32995649e-02
3.31133902e-01 6.33979023e-01 -9.97826278e-01 7.01508284e-01
6.06854379e-01 1.00804675e+00 -8.37989926e-01 1.72113776e+00
-6.13835573e-01 7.69541562e-01 -5.20665407e-01 3.18256706e-01
3.56838226e-01 -4.61295396e-01 5.07930398e-01 1.64870501e+00
3.35639358e-01 -1.60656512e-01 4.04477417e-01 8.08465660e-01
-2.09079012e-01 1.94994569e-01 -1.07783079e+00 2.75584459e-01
4.88367766e-01 1.34324217e+00 -1.24965370e+00 -1.85030714e-01
-6.83988094e-01 1.10006595e+00 4.83226150e-01 6.31731153e-01
-1.11191320e+00 -9.49199915e-01 4.08360690e-01 3.12141955e-01
4.59205091e-01 -9.11818564e-01 -1.68236807e-01 -1.07345247e+00
1.65827259e-01 -1.85632199e-01 2.35503450e-01 -8.37184012e-01
-9.83850241e-01 7.22279012e-01 -9.77622867e-02 -1.38126087e+00
-1.43709809e-01 -4.20280546e-01 -1.02051246e+00 4.07579213e-01
-1.65431035e+00 -1.29069114e+00 -6.75737083e-01 7.32462347e-01
6.16608799e-01 2.92958379e-01 3.05828154e-01 2.72908986e-01
-7.20269680e-01 6.11179709e-01 1.26607805e-01 1.15854812e+00
4.69539791e-01 -1.28664517e+00 1.17377758e+00 1.15323436e+00
3.01109314e-01 5.47796011e-01 3.32937449e-01 -5.44121683e-01
-1.01012409e+00 -1.16652393e+00 3.76073271e-01 -2.16708481e-01
6.35621846e-01 -4.07289386e-01 -5.89477837e-01 1.18934596e+00
4.34819281e-01 2.88349241e-01 4.11218047e-01 -9.20957029e-02
-3.36385459e-01 3.83468345e-02 -6.75541103e-01 7.77139664e-01
1.52632535e+00 -1.70321435e-01 -3.45109493e-01 3.30939561e-01
7.19264865e-01 -8.02068651e-01 -8.25803936e-01 3.85375261e-01
4.38610137e-01 -1.21703398e+00 1.15748072e+00 -8.43774602e-02
3.99688274e-01 -5.42783976e-01 1.68132618e-01 -1.25371146e+00
-3.87689531e-01 -6.24461472e-01 -6.19967729e-02 1.03524685e+00
5.35953462e-01 -5.99905849e-01 9.58613455e-01 3.86654101e-02
-5.79898715e-01 -8.96906018e-01 -5.08985281e-01 -7.60604501e-01
-1.24778360e-01 -4.72739100e-01 6.47234201e-01 7.39348948e-01
-1.53349280e-01 3.42249811e-01 -3.91244888e-02 4.06304657e-01
7.23563731e-01 3.53493035e-01 1.28630912e+00 -1.05676675e+00
6.92432597e-02 -6.43191397e-01 -7.61049688e-01 -1.67037690e+00
-3.70050706e-02 -7.71867514e-01 -7.47347996e-02 -2.13096309e+00
-4.43298489e-01 -4.36221510e-01 1.47591874e-01 2.64851838e-01
1.36213616e-01 4.94245589e-01 2.49326736e-01 2.30162099e-01
-8.13785315e-01 1.95574537e-01 1.37162828e+00 -2.76467621e-01
-7.18153790e-02 2.07726002e-01 -4.50021684e-01 1.13290966e+00
1.14554870e+00 -4.35680058e-03 -3.72677952e-01 -6.12173080e-01
3.19529697e-02 5.12430165e-03 3.50567162e-01 -1.30295587e+00
4.35088813e-01 -2.09239759e-02 5.85018814e-01 -1.14597583e+00
2.69547284e-01 -6.11658871e-01 -1.03468679e-01 3.44661295e-01
-1.35083660e-01 2.91230977e-01 1.35040790e-01 4.47467655e-01
-2.04092100e-01 -1.12472557e-01 5.90290666e-01 9.58404690e-03
-1.14378321e+00 6.91910923e-01 -4.48096469e-02 -8.86994526e-02
1.23571372e+00 -5.24027288e-01 -7.30353236e-01 -4.56927150e-01
-6.36289358e-01 4.87320513e-01 5.56974173e-01 2.82542825e-01
8.97896111e-01 -1.39628291e+00 -7.17575312e-01 1.45844802e-01
3.94570410e-01 4.24387187e-01 -1.97529793e-03 7.73438752e-01
-1.60864353e+00 4.59362805e-01 -2.03952700e-01 -7.95426846e-01
-1.05326319e+00 4.22818810e-01 5.48416018e-01 1.86175629e-01
-1.17997456e+00 7.80594230e-01 2.09911630e-01 -1.35381147e-01
1.33470148e-01 -6.67679012e-01 -1.63429409e-01 -5.69895566e-01
1.99798718e-01 5.07673740e-01 -1.82997927e-01 -7.50822127e-01
-2.83577830e-01 1.03601348e+00 -2.70740595e-02 3.23318720e-01
1.11577809e+00 -4.13318574e-01 9.65645313e-02 2.98123747e-01
1.32394433e+00 3.28229100e-01 -1.37757277e+00 -2.60216832e-01
2.44306531e-02 -4.74758208e-01 -1.63796619e-01 -1.23872831e-01
-1.26679754e+00 6.57715380e-01 2.00780213e-01 4.08256382e-01
8.65346491e-01 6.13851063e-02 1.04558420e+00 4.47029024e-01
2.98772752e-01 -1.00893056e+00 -6.97552636e-02 5.19926131e-01
7.91808844e-01 -1.13934684e+00 8.40976462e-02 -1.05808306e+00
-3.15071732e-01 1.57241321e+00 7.14272976e-01 -5.45396030e-01
8.04798841e-01 4.01860029e-01 1.15296222e-01 -4.71167445e-01
-7.10215345e-02 -4.53022182e-01 2.05874860e-01 5.63081741e-01
4.24028903e-01 1.44074634e-01 -2.03262419e-01 -8.89627170e-03
-5.89249372e-01 1.22333065e-01 7.94489026e-01 8.96825075e-01
-5.27317941e-01 -8.25279593e-01 -2.87960380e-01 2.08903372e-01
-1.48919761e-01 5.66019341e-02 -4.77794826e-01 1.07363522e+00
-1.29763439e-01 8.00650954e-01 3.46894175e-01 -3.28452885e-01
5.66176772e-01 -5.79358280e-01 2.46650934e-01 -4.16900873e-01
9.03646722e-02 1.00236602e-01 3.43956143e-01 -8.54153037e-01
-3.95630747e-01 -3.59687030e-01 -1.55051661e+00 -4.78553206e-01
-6.22269586e-02 -9.35467556e-02 4.44319516e-01 5.15403152e-01
3.82446736e-01 4.86859322e-01 7.15479970e-01 -6.99037373e-01
3.96495640e-01 -5.56022167e-01 -5.13258517e-01 2.50543594e-01
3.80306125e-01 -5.00252426e-01 -1.18414056e-03 8.61961395e-02] | [8.26313591003418, -1.7037243843078613] |
7fff1a37-2230-4b50-b4f1-85a891065645 | modeling-complex-systems-a-case-study-of | 2110.02947 | null | https://arxiv.org/abs/2110.02947v2 | https://arxiv.org/pdf/2110.02947v2.pdf | Modeling complex systems: A case study of compartmental models in epidemiology | Compartmental epidemic models have been widely used for predicting the course of epidemics, from estimating the basic reproduction number to guiding intervention policies. Studies commonly acknowledge these models' assumptions but less often justify their validity in the specific context in which they are being used. Our purpose is not to argue for specific alternatives or modifications to compartmental models, but rather to show how assumptions can constrain model outcomes to a narrow portion of the wide landscape of potential epidemic behaviors. This concrete examination of well-known models also serves to illustrate general principles of modeling that can be applied in other contexts. | ['Alexander F. Siegenfeld', 'Yaneer Bar-Yam', 'Pratyush K. Kollepara'] | 2021-10-06 | null | null | null | null | ['epidemiology'] | ['medical'] | [-1.88206714e-02 -3.86933744e-01 -3.19760054e-01 2.01059073e-01
1.42620057e-01 -4.47710752e-01 7.32026458e-01 2.00296968e-01
-6.60026610e-01 8.52520645e-01 4.31551524e-02 -1.03212726e+00
-5.48895717e-01 -7.16234565e-01 -4.56350148e-02 -6.70459986e-01
-5.20669460e-01 6.65535629e-01 2.93275088e-01 -5.21979392e-01
1.27021983e-01 7.54712582e-01 -1.08674157e+00 -3.45788598e-01
6.79659486e-01 2.67314136e-01 3.39808688e-02 6.96893096e-01
-1.85049251e-01 7.63313711e-01 -9.40208077e-01 -2.85529852e-01
-9.27064493e-02 -6.09094977e-01 -3.71088564e-01 -1.71303302e-01
-1.08634222e+00 -5.14115393e-01 -4.72810030e-01 4.83548909e-01
2.74979055e-01 -3.12112749e-01 1.08145046e+00 -1.46851695e+00
-4.24999684e-01 2.78913766e-01 -3.85579646e-01 6.47009134e-01
1.95854157e-01 2.69417882e-01 2.55213290e-01 -3.54948379e-02
5.94349623e-01 9.72529173e-01 9.36079025e-01 4.62668449e-01
-1.31683135e+00 -5.35972357e-01 4.13740247e-01 -2.71666378e-01
-1.60412264e+00 -3.87873232e-01 3.90705228e-01 -5.53218484e-01
1.01535118e+00 4.60878104e-01 1.21373880e+00 7.85050511e-01
7.85041153e-01 1.08683370e-01 1.05902529e+00 -9.87160858e-03
2.86803037e-01 1.29086912e-01 -1.16473265e-01 -9.77232754e-02
1.01512647e+00 1.10168554e-01 1.88635409e-01 -7.54285336e-01
1.10365248e+00 2.75337964e-01 -1.51067674e-01 -2.01452404e-01
-6.10390425e-01 9.66116011e-01 -2.39907373e-02 4.78779167e-01
-6.79301083e-01 2.59523094e-01 2.06379786e-01 2.44583875e-01
6.21434450e-01 2.90421546e-01 -6.34311199e-01 -7.55839795e-02
-8.78149390e-01 6.68450415e-01 8.82309616e-01 5.04616201e-01
2.44919240e-01 4.39979956e-02 2.59600192e-01 3.88043106e-01
7.73742139e-01 7.71949232e-01 -1.15694948e-01 -8.06712568e-01
5.94270863e-02 4.16510344e-01 5.56027830e-01 -6.46374226e-01
-4.72243249e-01 -1.38814121e-01 -6.14287555e-01 -2.63657384e-02
5.33651114e-01 -6.15094483e-01 -5.84724307e-01 1.49146700e+00
5.45630381e-02 1.55954227e-01 -3.66693363e-02 2.61889160e-01
1.56777069e-01 6.91487491e-01 5.27881861e-01 -7.55200744e-01
1.03684270e+00 8.76331404e-02 -6.12743914e-01 -2.09957939e-02
6.70407414e-01 -6.49526954e-01 4.87599641e-01 -1.35194063e-01
-9.75483835e-01 5.29560685e-01 -2.54105926e-01 7.52619028e-01
-5.23548007e-01 -8.39277327e-01 7.34756708e-01 9.51249063e-01
-1.19224775e+00 2.46711358e-01 -8.49647880e-01 -9.34595585e-01
9.19057429e-02 3.63511264e-01 4.33430046e-01 2.46062338e-01
-1.13802993e+00 1.00277340e+00 -1.84082985e-01 2.37660527e-01
-5.20074725e-01 -6.98091626e-01 -4.03117478e-01 3.55136283e-02
1.63420737e-01 -6.95553184e-01 1.07275391e+00 -4.59813863e-01
-8.87620866e-01 5.31547666e-01 -1.51393265e-01 -4.13974494e-01
3.80891800e-01 3.13751847e-01 -7.03518152e-01 -1.36507265e-02
-4.67417650e-02 8.42240900e-02 1.63753062e-01 -1.53434968e+00
-3.45889717e-01 -1.61748976e-01 2.14656979e-01 3.69750470e-01
1.69839978e-01 7.90305734e-01 -1.19494379e-01 -6.14426732e-01
-2.65805602e-01 -7.74218440e-01 -7.79091120e-01 -8.02361891e-02
-2.18190640e-01 -7.66931400e-02 5.32229781e-01 -3.13438594e-01
1.59025359e+00 -1.54556358e+00 -2.79155791e-01 3.12300265e-01
1.42081305e-01 7.17857331e-02 1.03738375e-01 1.24090791e+00
5.00150681e-01 5.48633277e-01 -4.60384697e-01 -4.71923612e-02
-1.56511560e-01 3.38571668e-01 -3.93537164e-01 7.67657816e-01
3.99132855e-02 8.40951979e-01 -7.25000143e-01 -5.04990757e-01
2.64404684e-01 6.31273091e-01 -4.03476179e-01 -1.64162248e-01
2.20360920e-01 4.77378845e-01 -5.75456560e-01 4.27823693e-01
5.40201545e-01 -2.70131379e-01 4.79127496e-01 7.26701796e-01
-4.52803850e-01 1.28915519e-01 -7.03942657e-01 1.13424934e-01
-1.27382249e-01 5.63426256e-01 3.42754900e-01 -8.82361054e-01
3.91540080e-01 3.21371824e-01 8.78219366e-01 -1.08949363e-01
2.01289475e-01 5.10722548e-02 5.49991250e-01 -3.19159091e-01
1.47853807e-01 -4.46179360e-01 -2.57667229e-02 8.55430841e-01
-6.15179718e-01 -1.22621328e-01 1.15544580e-01 2.50308424e-01
8.69023860e-01 -2.39113390e-01 4.58347648e-01 -6.80708289e-01
8.88492819e-03 3.70812923e-01 3.84474725e-01 8.77800345e-01
-5.03804624e-01 -1.76969364e-01 6.80939913e-01 -2.76732475e-01
-1.01114786e+00 -1.11962438e+00 -4.63844508e-01 6.37804210e-01
3.95215780e-01 -2.06252217e-01 -5.42186916e-01 -1.66049570e-01
2.12234393e-01 6.08782530e-01 -8.25847268e-01 1.15437850e-01
-5.35188854e-01 -1.39295673e+00 5.76390862e-01 1.91597074e-01
-2.41520822e-01 -9.87414539e-01 -8.94124389e-01 2.32048929e-01
1.44042462e-01 -6.52230859e-01 -1.40074536e-01 5.57230823e-02
-9.57147777e-01 -1.50838447e+00 -1.06499469e+00 -2.99780101e-01
7.52406180e-01 4.59425479e-01 8.83643925e-01 5.27467847e-01
-1.09079607e-01 9.18015122e-01 -4.22605649e-02 -4.33988124e-01
-6.27544582e-01 -1.68143913e-01 3.31722319e-01 -4.66006249e-01
6.91586316e-01 -3.94704163e-01 -6.87028289e-01 4.04977500e-01
-9.30314362e-01 -4.72885430e-01 1.63311496e-01 2.08102137e-01
6.91657588e-02 1.08183678e-02 9.00270939e-01 -9.01337802e-01
1.24514544e+00 -9.45428967e-01 -5.19996464e-01 2.78352201e-01
-7.93775380e-01 -7.16440320e-01 4.41710174e-01 -4.65749651e-01
-8.76359761e-01 -5.77181756e-01 -1.83111325e-01 2.08649054e-01
-1.98059037e-01 6.49159610e-01 3.71973634e-01 2.23142328e-03
2.50871092e-01 4.52033997e-01 2.31861278e-01 -3.09139937e-01
-1.16724283e-01 6.95728838e-01 -3.02453697e-01 -2.77291209e-01
7.04788089e-01 5.39366007e-01 -3.97221968e-02 -1.33708990e+00
2.93007851e-01 -1.87668055e-01 -3.61046374e-01 -3.77263039e-01
4.01960582e-01 -6.33819342e-01 -9.31371510e-01 5.12317240e-01
-8.87588918e-01 -8.24159145e-01 -1.98471993e-01 4.81583089e-01
-6.86099529e-01 2.01686472e-01 -8.09619009e-01 -1.49577463e+00
1.76424757e-01 -1.11993492e+00 4.39997733e-01 9.63279679e-02
-3.37440580e-01 -1.69199133e+00 3.50139886e-01 -2.36553639e-01
9.20046329e-01 1.19461142e-01 7.96831727e-01 -5.54412305e-01
-1.94174498e-01 -1.33247986e-01 -2.61843260e-02 -5.25521338e-01
3.41920316e-01 5.30877590e-01 -4.05202180e-01 -3.59654099e-01
1.17772795e-01 4.03525919e-01 3.71656239e-01 1.01335108e+00
4.96897906e-01 -5.86421430e-01 -9.52836812e-01 2.50528276e-01
1.49152851e+00 6.40896022e-01 5.82923114e-01 2.85481840e-01
-1.99738607e-01 8.55162978e-01 2.42875218e-01 4.87529367e-01
5.00164390e-01 6.06894195e-01 1.20556071e-01 6.45810217e-02
4.75065589e-01 -4.72093858e-02 1.44181043e-01 6.79713309e-01
-2.71205336e-01 -4.15112257e-01 -1.10111606e+00 6.99003279e-01
-1.62305844e+00 -1.07840061e+00 -2.90432304e-01 2.21397352e+00
5.06203234e-01 2.68755496e-01 8.66354346e-01 -3.62784922e-01
8.27714324e-01 4.17259224e-02 -3.71751100e-01 -3.83029670e-01
-3.51693630e-02 -4.21251625e-01 8.03759873e-01 6.84306026e-01
-6.16666138e-01 7.01530039e-01 9.49072933e+00 3.85066748e-01
-1.12028098e+00 1.65383648e-02 7.37221897e-01 -5.15781939e-02
-3.68754089e-01 8.25794637e-02 -5.91270506e-01 5.84452450e-01
1.29001021e+00 -7.10636258e-01 4.64321584e-01 1.78312168e-01
9.40521002e-01 -2.80907631e-01 -4.42107737e-01 2.70998776e-01
-4.68116105e-01 -1.14050746e+00 -1.33824095e-01 5.77086091e-01
4.54606682e-01 1.20682582e-01 9.94369388e-02 1.46656200e-01
6.01515293e-01 -1.06441593e+00 2.85947591e-01 4.24492002e-01
7.56945431e-01 -5.59329748e-01 5.64520121e-01 4.52042878e-01
-1.09110343e+00 -5.29667549e-03 -3.21467131e-01 -4.59025025e-01
9.04887497e-01 5.22990245e-03 -5.08574367e-01 -1.27677411e-01
5.64817607e-01 2.07198769e-01 3.51586007e-03 1.42587197e+00
1.28273204e-01 8.80809128e-01 -6.50847197e-01 -2.05355972e-01
2.83322930e-01 -2.42172301e-01 6.39007449e-01 1.15595198e+00
1.65920675e-01 3.09293687e-01 -3.24798048e-01 7.61745214e-01
5.99328399e-01 1.14742085e-01 -1.00411904e+00 -3.58976096e-01
7.95357823e-01 5.97233474e-01 -1.07066417e+00 -4.51554582e-02
-4.62511569e-01 5.22243321e-01 -2.91692615e-01 8.07344317e-01
-8.33733797e-01 -2.07053155e-01 1.13429105e+00 7.47677505e-01
-4.38902862e-02 -3.97801578e-01 -1.80272684e-01 -9.29280758e-01
-7.08930910e-01 -4.93555725e-01 -8.03887192e-03 -3.51664811e-01
-1.09285080e+00 4.16400194e-01 7.26781785e-01 -7.92129576e-01
-3.29947501e-01 -4.44904566e-01 -1.01692057e+00 9.48175430e-01
-1.32465112e+00 -8.45200241e-01 3.75937134e-01 4.06715631e-01
3.18470716e-01 1.66316137e-01 6.73433125e-01 8.98579583e-02
-8.32852900e-01 1.05728582e-01 8.06188762e-01 -1.57652721e-01
8.24906230e-02 -7.83460140e-01 4.81744647e-01 4.73520190e-01
-6.07773483e-01 1.16001332e+00 1.16536474e+00 -1.17828155e+00
-1.00677228e+00 -6.95953250e-01 9.77897763e-01 -4.95329291e-01
8.78694475e-01 -2.99054146e-01 -5.92155039e-01 7.62165010e-01
7.51685649e-02 -6.40714049e-01 7.89406896e-01 2.92488337e-02
3.26138586e-01 2.40827098e-01 -1.29619622e+00 1.11912131e+00
8.16370904e-01 -1.20911427e-01 5.24149984e-02 2.32549548e-01
3.47745925e-01 2.98113257e-01 -9.84031022e-01 -5.44815175e-02
8.58417392e-01 -7.99158037e-01 9.24292445e-01 -9.46930587e-01
-1.01298384e-01 6.72110468e-02 3.88421901e-02 -1.06857729e+00
-3.56369078e-01 -8.28194737e-01 1.48058441e-02 9.11088705e-01
3.68089110e-01 -1.26834214e+00 5.77980220e-01 7.28680015e-01
5.80789387e-01 -1.13094854e+00 -1.02312744e+00 -9.41510081e-01
7.04676270e-01 -2.66933233e-01 7.77930140e-01 7.64478624e-01
1.09597534e-01 -2.52316952e-01 -2.77708113e-01 -7.43632317e-02
7.71867871e-01 -3.25561643e-01 6.26059294e-01 -1.42132282e+00
7.70594925e-02 -7.40096331e-01 -1.61331788e-01 -9.84013975e-01
-3.84151995e-01 -3.02931052e-02 -1.18656009e-01 -1.67493200e+00
2.62251139e-01 -1.00880730e+00 -2.88899988e-01 -1.17886126e-01
-3.11775245e-02 3.11661690e-01 1.69814959e-01 7.63063252e-01
-1.74204811e-01 2.57432193e-01 9.20352399e-01 2.76685447e-01
-3.95149469e-01 4.84430254e-01 -1.13201237e+00 7.38082886e-01
1.25081336e+00 -6.56173944e-01 -6.33166075e-01 8.22029251e-04
3.54399979e-01 2.12786108e-01 3.50262016e-01 -4.68628675e-01
1.10257097e-01 -1.04268658e+00 5.03060147e-02 -3.27878207e-01
3.35631549e-01 -9.60131884e-01 6.95754766e-01 9.97238934e-01
-3.54372640e-03 4.66265947e-01 4.56428617e-01 7.23423421e-01
5.96034765e-01 -2.88817316e-01 4.94850487e-01 1.05773434e-02
-7.59509280e-02 2.62816072e-01 -1.30843067e+00 1.74519390e-01
1.50103521e+00 -5.93538940e-01 -4.50570941e-01 -9.65623498e-01
-6.69394135e-01 9.65483710e-02 1.04228950e+00 6.11333884e-02
2.98638374e-01 -8.03635180e-01 -5.22069395e-01 -2.18312845e-01
-2.32551962e-01 -7.98147798e-01 2.56237805e-01 1.06999028e+00
-8.70121121e-01 7.89406896e-01 -8.90534669e-02 -1.99453339e-01
-9.72538233e-01 8.09101105e-01 6.67029023e-01 -2.73743242e-01
-6.04441583e-01 2.66818941e-01 4.39498901e-01 -1.49061501e-01
-1.93267986e-01 6.83696866e-02 -1.17642686e-01 -2.37059534e-01
4.45567429e-01 4.60475385e-01 -8.55249465e-01 -9.96814966e-01
-8.17987263e-01 2.99841851e-01 1.84918404e-01 -1.35176912e-01
1.39547276e+00 -6.01860702e-01 -2.11102620e-01 5.52482605e-01
5.18086672e-01 1.04423530e-01 -1.23771751e+00 3.49420637e-01
-2.73331672e-01 -1.58138767e-01 -2.71443546e-01 -6.14060640e-01
-7.72545516e-01 6.10542417e-01 3.08458388e-01 9.19561625e-01
9.78245676e-01 9.94571298e-03 3.04763198e-01 -8.95203128e-02
3.11410517e-01 -5.49360454e-01 -7.27408171e-01 1.28970802e-01
5.83544612e-01 -6.10934556e-01 2.30276391e-01 -4.25655305e-01
-5.35025001e-01 6.74980581e-01 1.20215252e-01 -2.17368424e-01
1.16769826e+00 6.51170552e-01 9.51112434e-02 -1.83735251e-01
-8.71513963e-01 -1.48512676e-01 -3.81237745e-01 1.25725138e+00
3.99011374e-01 5.29360659e-02 -9.87144053e-01 3.42289746e-01
3.25211555e-01 5.17933443e-02 7.56884634e-01 9.48275685e-01
-5.46434581e-01 -1.07420111e+00 -5.62878430e-01 8.60613346e-01
-5.68683803e-01 -1.46305025e-01 -5.85021079e-01 1.21205568e+00
-1.58050701e-01 1.24402928e+00 3.04206401e-01 1.29997313e-01
-5.22312783e-02 -1.75596222e-01 2.92593896e-01 -2.22394347e-01
-7.29537964e-01 2.38571823e-01 -9.72574111e-04 -4.58958633e-02
-3.19287002e-01 -8.46200943e-01 -7.51154304e-01 -1.17347360e+00
-4.49441165e-01 4.26411718e-01 3.56630087e-01 8.36534142e-01
2.51746505e-01 3.49745125e-01 2.89124787e-01 -5.76311469e-01
-3.47354889e-01 -7.94691026e-01 -1.05071497e+00 -2.33894363e-01
2.36152813e-01 -7.33395696e-01 -4.45334435e-01 -4.15913582e-01] | [5.968226432800293, 4.426397800445557] |
3d3499c6-1e0d-4fed-81c3-b2b4de99c158 | wbcatt-a-white-blood-cell-dataset-annotated | 2306.13531 | null | https://arxiv.org/abs/2306.13531v1 | https://arxiv.org/pdf/2306.13531v1.pdf | WBCAtt: A White Blood Cell Dataset Annotated with Detailed Morphological Attributes | The examination of blood samples at a microscopic level plays a fundamental role in clinical diagnostics, influencing a wide range of medical conditions. For instance, an in-depth study of White Blood Cells (WBCs), a crucial component of our blood, is essential for diagnosing blood-related diseases such as leukemia and anemia. While multiple datasets containing WBC images have been proposed, they mostly focus on cell categorization, often lacking the necessary morphological details to explain such categorizations, despite the importance of explainable artificial intelligence (XAI) in medical domains. This paper seeks to address this limitation by introducing comprehensive annotations for WBC images. Through collaboration with pathologists, a thorough literature review, and manual inspection of microscopic images, we have identified 11 morphological attributes associated with the cell and its components (nucleus, cytoplasm, and granules). We then annotated ten thousand WBC images with these attributes. Moreover, we conduct experiments to predict these attributes from images, providing insights beyond basic WBC classification. As the first public dataset to offer such extensive annotations, we also illustrate specific applications that can benefit from our attribute annotations. Overall, our dataset paves the way for interpreting WBC recognition models, further advancing XAI in the fields of pathology and hematology. | ['Bihan Wen', 'Winnie Pang', 'Satoshi Tsutsui'] | 2023-06-23 | null | null | null | null | ['explainable-artificial-intelligence'] | ['computer-vision'] | [ 1.71491206e-01 7.81639591e-02 -1.94144070e-01 -4.49583858e-01
-3.14929307e-01 -4.24467891e-01 3.04011762e-01 1.05429649e+00
-1.73830286e-01 8.49929154e-01 2.95586318e-01 -3.12329143e-01
-7.59185702e-02 -8.58989954e-01 1.52752372e-02 -1.05637801e+00
1.93782151e-01 1.05925548e+00 -2.12439477e-01 2.17372239e-01
2.02902302e-01 1.00775397e+00 -1.14396250e+00 5.09687603e-01
5.63145220e-01 1.08369434e+00 -4.91292268e-01 8.88315499e-01
-5.51431656e-01 9.36576545e-01 -6.90957367e-01 -5.75410068e-01
-3.40665311e-01 -3.89746636e-01 -8.76605392e-01 3.38073105e-01
-2.37177350e-02 -1.30377531e-01 -2.04154789e-01 6.44036412e-01
5.36158144e-01 -2.41080061e-01 1.15404999e+00 -1.19147813e+00
-6.09144151e-01 3.34547371e-01 -7.03347206e-01 4.54945952e-01
2.37958118e-01 4.79744315e-01 8.55824232e-01 -4.28969830e-01
4.94432092e-01 1.04270768e+00 3.98280621e-01 5.71251810e-01
-1.09259820e+00 -2.75464475e-01 -4.00533110e-01 2.11794630e-01
-1.16464806e+00 -4.16821420e-01 4.02203709e-01 -4.71061260e-01
8.00659060e-01 5.09185255e-01 7.09042072e-01 6.53591454e-01
1.45414144e-01 7.81898499e-01 1.11690593e+00 -4.11465019e-01
3.17705542e-01 1.30259976e-01 4.55871433e-01 6.86924279e-01
7.62278080e-01 -4.70427424e-01 -4.99816537e-01 -2.26402864e-01
5.07405043e-01 3.15037847e-01 -1.50672942e-01 -1.11747548e-01
-1.32475114e+00 5.03491640e-01 -2.92707775e-02 4.37918812e-01
-1.94814309e-01 -5.84740043e-02 4.68164295e-01 -1.53203920e-01
1.27721921e-01 6.00996614e-01 -4.60585773e-01 -1.14788432e-02
-5.36688507e-01 -3.79462957e-01 8.25618207e-01 5.84634066e-01
5.25526881e-01 -1.57279611e-01 -1.85090438e-01 7.24170208e-01
1.90791339e-02 4.77366328e-01 4.57629383e-01 -7.57926702e-01
-2.75072396e-01 1.15592897e+00 -1.62096158e-01 -8.97355020e-01
-8.16367447e-01 -2.96923190e-01 -1.14713836e+00 -1.23623408e-01
9.14544582e-01 4.10286456e-01 -6.52492285e-01 1.21690130e+00
4.77480322e-01 -3.66783366e-02 -1.80722430e-01 8.00965846e-01
1.06642830e+00 5.27764931e-02 4.00572628e-01 -8.22074860e-02
1.90277851e+00 -4.61621165e-01 -7.11407304e-01 4.33406919e-01
8.26876342e-01 -5.62584639e-01 8.57209980e-01 5.45129538e-01
-8.81027937e-01 -2.99188763e-01 -8.05713952e-01 -1.30952731e-01
-6.21110141e-01 1.00609884e-01 1.09298778e+00 6.61185026e-01
-7.99435973e-01 2.28100583e-01 -9.66135263e-01 -7.82609642e-01
9.01188672e-01 3.96926641e-01 -6.36165202e-01 1.57974258e-01
-6.18759513e-01 8.63843501e-01 6.12132363e-02 -2.17735246e-02
-1.89075962e-01 -9.19107556e-01 -7.12648451e-01 1.92689989e-03
4.25515585e-02 -9.89592016e-01 6.55314863e-01 -8.99424031e-03
-8.53554785e-01 1.38399720e+00 -4.03577447e-01 -2.78575450e-01
1.24237508e-01 3.48373085e-01 -3.90630394e-01 6.17537856e-01
-1.92563131e-01 5.68108439e-01 2.17352778e-01 -1.02588832e+00
-7.68924475e-01 -6.52064025e-01 -2.36940399e-01 -3.74359637e-01
-3.04392129e-01 -1.71502426e-01 -3.28620732e-01 -6.48467898e-01
-1.72872934e-02 -4.75327581e-01 -5.79616278e-02 1.51410297e-01
-5.60327411e-01 -2.50643730e-01 5.63687623e-01 -5.07350266e-01
1.08151972e+00 -1.94765115e+00 -1.00359775e-01 1.62290305e-01
8.62330616e-01 1.37369260e-01 1.96703836e-01 2.03055277e-01
1.43454537e-01 3.86804849e-01 -1.88714132e-01 -4.05676067e-01
9.94229466e-02 1.57929778e-01 -1.84354782e-01 7.20437884e-01
3.45471412e-01 1.14530396e+00 -8.39737296e-01 -8.77071857e-01
4.12374109e-01 5.73672295e-01 -1.58023655e-01 1.26473621e-01
-5.11511602e-02 8.65414381e-01 -3.51724863e-01 1.25033319e+00
3.60166788e-01 -7.75890827e-01 1.21942431e-01 -4.58502442e-01
4.66649383e-01 -1.98888108e-01 -6.40273988e-01 9.47040379e-01
-9.82921198e-02 6.12712383e-01 -5.89123443e-02 -9.25880849e-01
7.76858926e-01 1.22443162e-01 8.91809702e-01 -4.21787530e-01
5.06653011e-01 1.21139521e-02 1.02225065e-01 -5.11495531e-01
6.77445903e-02 -3.35182101e-01 1.43871993e-01 8.71372640e-01
-1.26434058e-01 -5.78982271e-02 4.32877541e-01 3.24459732e-01
1.08057213e+00 -3.33400190e-01 5.83541334e-01 -1.09253593e-01
8.65116537e-01 8.33617076e-02 4.41088349e-01 4.58082795e-01
-6.67430997e-01 6.34399951e-01 6.40394032e-01 -8.87572348e-01
-9.83028650e-01 -9.41746414e-01 -5.67360878e-01 7.01244295e-01
6.84328377e-02 3.78632657e-02 -5.90398371e-01 -5.40039301e-01
2.98352271e-01 2.61816412e-01 -9.33562040e-01 1.96139794e-02
-3.84949714e-01 -1.14422321e+00 8.42796564e-01 5.98452330e-01
2.03106180e-01 -1.13192677e+00 -7.06501067e-01 -4.07639742e-02
-2.99450815e-01 -1.20055330e+00 -1.08480781e-01 1.21943608e-01
-6.46381855e-01 -1.70115554e+00 -2.45578095e-01 -4.19984579e-01
8.22666347e-01 -1.09230459e-01 1.21682334e+00 9.21492815e-01
-1.22500789e+00 5.03806055e-01 -2.33188957e-01 -5.57445288e-01
-3.83567184e-01 -1.17317848e-01 9.22271758e-02 1.22304313e-01
8.77873957e-01 -2.44294345e-01 -7.46400058e-01 8.98402631e-02
-1.00941241e+00 -1.74140104e-03 6.06402934e-01 8.05723906e-01
7.98385561e-01 7.62887374e-02 5.03999710e-01 -1.23752034e+00
1.11312516e-01 -3.56064200e-01 9.13848728e-02 5.16273320e-01
-3.55572164e-01 -1.88222647e-01 6.70879960e-01 -4.85082753e-02
-8.73832881e-01 -4.44900215e-01 -3.84745225e-02 1.53932691e-01
-8.96827877e-01 9.05132890e-02 -1.77872762e-01 7.34905079e-02
2.90988266e-01 2.35923722e-01 1.43528908e-01 -2.96533734e-01
-8.61906633e-02 6.80200219e-01 6.72718227e-01 -5.87751925e-01
2.58444548e-01 1.06264377e+00 6.30540550e-01 -9.65270817e-01
-9.53730166e-01 -5.70353806e-01 -8.28199089e-01 2.69487314e-02
8.73671949e-01 -3.01067621e-01 -1.31199479e+00 4.18128699e-01
-6.56283021e-01 -1.97131798e-01 -3.23606610e-01 4.51149106e-01
-4.83709812e-01 7.52303958e-01 -1.30298030e+00 -5.98283887e-01
-4.37110811e-01 -9.25183177e-01 1.05805111e+00 4.39561605e-01
-7.19565451e-01 -1.49316788e+00 2.61310674e-02 8.08019221e-01
4.01769787e-01 4.84520435e-01 1.70377994e+00 -9.91535485e-01
-2.27663800e-01 -6.73168600e-02 -3.05525184e-01 -3.29185605e-01
5.26353478e-01 3.88741106e-01 -1.03998864e+00 -5.03274351e-02
2.41127480e-02 -3.51625293e-01 8.25856507e-01 4.84738559e-01
1.55135906e+00 -4.29575192e-03 -5.43560982e-01 8.31113338e-01
1.17533255e+00 1.69458836e-01 5.61176956e-01 2.82890379e-01
4.94449407e-01 9.65783477e-01 5.27974546e-01 7.01797605e-01
3.22819352e-01 1.09452330e-01 2.63319582e-01 -6.31683826e-01
-3.28676611e-01 3.52444232e-01 -5.68302989e-01 4.35624748e-01
-8.24024752e-02 -3.19138229e-01 -1.04446089e+00 4.56137359e-01
-1.32318258e+00 -7.14441538e-01 -3.17140043e-01 1.79856706e+00
8.92703772e-01 -6.86133802e-02 1.44390002e-01 2.79836595e-01
6.51868224e-01 -4.24972564e-01 -7.77110994e-01 -2.19264492e-01
-4.78355855e-01 4.50173676e-01 6.44362569e-02 3.95884365e-01
-9.63116646e-01 6.41914010e-01 6.98117399e+00 6.31720543e-01
-9.78832304e-01 -4.09942627e-01 1.37636411e+00 -1.05381347e-01
-1.18784115e-01 -3.06335658e-01 -8.03467691e-01 5.34345090e-01
4.63928223e-01 3.26824524e-02 2.97834072e-02 3.73339832e-01
7.08790570e-02 -3.79144996e-01 -1.46737504e+00 9.57565784e-01
5.45626543e-02 -1.36970139e+00 2.40765974e-01 2.64175057e-01
3.01977873e-01 -7.45123804e-01 7.82938823e-02 -3.36028971e-02
6.43994734e-02 -1.45317948e+00 1.31629324e-02 7.75344849e-01
1.05691755e+00 -4.69579250e-01 1.14988220e+00 5.26882112e-02
-9.62499380e-01 2.99321800e-01 -5.02190828e-01 -2.56706234e-02
-9.47128683e-02 9.57149565e-01 -9.87029731e-01 2.02997848e-01
5.14430821e-01 5.66461146e-01 -8.87680471e-01 9.96346474e-01
-7.18609476e-03 5.76272428e-01 -4.18387949e-02 -1.74663737e-02
-4.93263364e-01 -2.93915361e-01 -1.31141059e-02 1.54682195e+00
-1.07163591e-02 5.44478357e-01 -2.25683540e-01 6.02607369e-01
1.29747754e-02 1.40444562e-01 -2.09136531e-01 -1.45359114e-01
4.52705801e-01 1.58720946e+00 -1.42368639e+00 -6.46602869e-01
-1.97235897e-01 4.61925268e-01 2.08383486e-01 1.17323332e-01
-6.51643038e-01 -2.07429722e-01 7.37858713e-01 1.20909616e-01
-2.49415651e-01 3.23663920e-01 -9.61868703e-01 -9.64632213e-01
-4.59834456e-01 -6.47940755e-01 8.20724189e-01 -5.50014138e-01
-1.68015504e+00 3.24785203e-01 -3.97146881e-01 -9.19737399e-01
1.39733195e-01 -8.72545063e-01 -3.00356150e-01 4.76252973e-01
-1.53171837e+00 -1.24836946e+00 -6.40105188e-01 3.45776081e-01
1.25152141e-01 -5.05797006e-02 8.80923569e-01 4.69721928e-02
-7.87455976e-01 4.66573536e-01 -1.41794875e-01 3.35958302e-01
8.85409832e-01 -1.54258096e+00 -1.38658836e-01 1.82275876e-01
-1.07690886e-01 8.80827725e-01 6.40605867e-01 -4.52951938e-01
-1.36958718e+00 -8.88165414e-01 9.27218258e-01 -7.81647205e-01
5.75563431e-01 -1.03681952e-01 -9.38811004e-01 4.89492565e-01
8.44402704e-03 1.88678086e-01 1.66275012e+00 -1.19781382e-01
-2.26674646e-01 -3.65621336e-02 -1.40318108e+00 5.23478150e-01
7.18076646e-01 -2.00333551e-01 -3.92277122e-01 3.97573680e-01
1.41374558e-01 -3.13055754e-01 -1.25584435e+00 3.58020604e-01
5.95042050e-01 -1.14794362e+00 9.97563779e-01 -8.93000960e-01
4.52403992e-01 -4.02257591e-01 1.69530407e-01 -8.06237340e-01
-3.20791692e-01 3.26351039e-02 -4.49457616e-01 1.17554188e+00
-5.03657793e-04 -6.64094985e-01 1.21453214e+00 7.62084484e-01
2.85103135e-02 -1.19817317e+00 -5.83637536e-01 -3.27379495e-01
4.68055494e-02 -1.63678244e-01 1.01841235e+00 1.04425716e+00
4.92616206e-01 5.00545055e-02 3.91374677e-02 -2.88147658e-01
8.08538198e-01 2.76274383e-01 9.57120776e-01 -1.26024437e+00
-3.51937264e-01 -8.97318542e-01 -6.66137695e-01 -5.07436693e-01
-3.16121355e-02 -6.67785406e-01 -4.51054156e-01 -1.66021323e+00
8.73394370e-01 -6.55194342e-01 -4.10961568e-01 5.75214446e-01
-3.77405196e-01 7.33250201e-01 -1.88512594e-01 3.77420872e-01
-6.71436727e-01 -6.10421970e-02 1.21518493e+00 -3.77706259e-01
2.94980109e-01 -3.40994060e-01 -1.03273559e+00 6.62381411e-01
8.56900871e-01 -2.33488709e-01 2.97040474e-02 3.55342552e-02
8.45526811e-03 2.38036253e-02 7.28935152e-02 -7.03042746e-01
3.03951502e-01 -4.20074910e-01 9.44678724e-01 -7.86999047e-01
2.83994853e-01 -5.89232683e-01 1.45946532e-01 8.53534579e-01
-2.87690133e-01 -3.63715917e-01 1.72422882e-02 3.79386425e-01
-1.90435395e-01 -8.41727853e-03 9.18838859e-01 -2.25126430e-01
-3.20300788e-01 3.28973651e-01 -4.71557409e-01 -2.21008677e-02
1.26112044e+00 -6.52384460e-01 -7.99450874e-01 -2.24205479e-01
-8.78118634e-01 3.37017030e-01 9.79117930e-01 -3.33237648e-01
5.45682847e-01 -8.67094874e-01 -6.59426689e-01 2.92844683e-01
2.46507600e-01 8.78741592e-02 3.71340692e-01 1.21154118e+00
-9.57727551e-01 7.15130568e-01 -2.76633561e-01 -8.72439623e-01
-1.27064645e+00 7.02840567e-01 1.46406814e-01 -5.75908959e-01
-4.43824321e-01 6.23280227e-01 4.65738982e-01 -1.58073038e-01
2.00488538e-01 -3.16774637e-01 -4.71770525e-01 -8.79071653e-02
8.04929912e-01 6.11755311e-01 1.59373239e-01 -5.73607028e-01
-4.89295125e-01 5.23640335e-01 -7.31890276e-02 4.83876705e-01
1.28971374e+00 -1.76377609e-01 -5.70518672e-01 4.15627688e-01
8.49908352e-01 1.02469109e-01 -7.68424273e-01 -2.49438714e-02
5.38141578e-02 -5.18603086e-01 -6.60060346e-01 -7.14454591e-01
-1.16226780e+00 1.16809428e+00 3.16148661e-02 2.45368272e-01
1.16815925e+00 2.33915225e-01 8.08955014e-01 1.78535610e-01
3.60405684e-01 -7.29044378e-01 3.89591306e-01 7.10872337e-02
1.46973863e-01 -1.28305554e+00 3.67562264e-01 -7.03126729e-01
-6.66266501e-01 1.30930614e+00 6.80381954e-01 4.83814865e-01
4.47220266e-01 5.83057642e-01 2.20359981e-01 -3.55769396e-01
-8.84060979e-01 -8.78682658e-02 -8.57084617e-02 8.21721137e-01
7.01762974e-01 4.72029932e-02 -1.55675203e-01 5.73058963e-01
-1.55857638e-01 -1.85688704e-01 5.04571438e-01 8.20155621e-01
-6.15635276e-01 -1.03408563e+00 -5.70942163e-01 9.07089889e-01
-8.43503237e-01 1.59473792e-01 -5.87571800e-01 7.51606405e-01
1.77678224e-02 1.02183867e+00 3.08241963e-01 9.22535658e-02
-1.63776621e-01 -2.56988741e-02 6.85345948e-01 -5.73834717e-01
-4.03234243e-01 -4.05595340e-02 -3.31331164e-01 -2.19268128e-01
-3.32585275e-01 -5.49687743e-01 -1.65189934e+00 -6.30355954e-01
-8.96059498e-02 6.37441128e-02 1.82676896e-01 1.07449269e+00
9.26603749e-02 3.97739112e-01 2.42627591e-01 -2.63571590e-02
2.64092833e-02 -8.70573044e-01 -1.01410472e+00 8.36649716e-01
3.11844856e-01 -5.03470540e-01 -4.44248617e-01 4.73098695e-01] | [15.03522777557373, -3.0780580043792725] |
4728f9e2-6a66-45d4-b5fc-529b98a6daf8 | pareidolia-face-reenactment | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Song_Pareidolia_Face_Reenactment_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Song_Pareidolia_Face_Reenactment_CVPR_2021_paper.pdf | Pareidolia Face Reenactment | We present a new application direction named Pareidolia Face Reenactment, which is defined as animating a static illusory face to move in tandem with a human face in the video. For the large differences between pareidolia face reenactment and traditional human face reenactment, two main challenges are introduced, i.e., shape variance and texture variance. In this work, we propose a novel Parametric Unsupervised Reenactment Algorithm to tackle these two challenges. Specifically, we propose to decompose the reenactment into three catenate processes: shape modeling, motion transfer and texture synthesis. With the decomposition, we introduce three crucial components, i.e., Parametric Shape Modeling, Expansionary Motion Transfer and Unsupervised Texture Synthesizer, to overcome the problems brought by the remarkably variances on pareidolia faces. Extensive experiments show the superior performance of our method both qualitatively and quantitatively. Code, model and data are available on our project page. | ['Ran He', 'Chen Change Loy', 'Chen Qian', 'Chaoyou Fu', 'Wayne Wu', 'Linsen Song'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['texture-synthesis', 'face-reenactment'] | ['computer-vision', 'computer-vision'] | [ 2.05108017e-01 1.31747931e-01 1.86470687e-01 -4.08236049e-02
-3.14863712e-01 -4.81826156e-01 6.28062606e-01 -1.09579360e+00
2.32788578e-01 4.05873358e-01 3.70001167e-01 1.80298463e-01
2.32426301e-02 -5.88486075e-01 -5.64751267e-01 -8.57085407e-01
4.48143125e-01 4.97970343e-01 -2.76178587e-02 -1.36139140e-01
1.32847309e-01 6.16560876e-01 -1.67242289e+00 3.22803438e-01
8.15126956e-01 7.13506818e-01 -3.58475861e-03 6.54850721e-01
-2.02889144e-01 6.30583167e-01 -2.53741503e-01 -3.84158790e-01
1.65115848e-01 -6.43215477e-01 -6.49073839e-01 7.48336554e-01
3.32027256e-01 -2.88165689e-01 -2.23898694e-01 1.00005662e+00
4.01150733e-01 1.01248451e-01 8.66947293e-01 -1.52336633e+00
-4.95178223e-01 2.34886482e-01 -1.18720973e+00 -4.93628621e-01
3.09875101e-01 -4.82564569e-02 4.40202266e-01 -1.31655288e+00
9.38431799e-01 1.80348396e+00 6.78479135e-01 8.36301088e-01
-1.00530171e+00 -7.82232285e-01 -3.99181014e-03 -1.54216781e-01
-1.46254373e+00 -7.96201468e-01 1.11931551e+00 -6.50095582e-01
1.93241477e-01 4.04217660e-01 6.09214962e-01 1.08086157e+00
3.26148599e-01 8.12738419e-01 1.07439804e+00 -4.51846391e-01
-5.44994101e-02 1.19819216e-01 -4.75234002e-01 7.94661999e-01
-1.07487760e-01 1.22550316e-01 -4.16514575e-01 -2.74727941e-01
1.23481250e+00 -2.23392010e-01 -9.31169912e-02 -6.78530216e-01
-9.05260682e-01 3.96226466e-01 -2.24121407e-01 1.14329100e-01
-1.32851973e-01 8.66309926e-02 2.30881676e-01 1.77764580e-01
5.22068143e-01 -5.52152656e-02 -2.41710752e-01 -1.74992025e-01
-7.58866191e-01 2.25892469e-01 6.41171813e-01 1.33534944e+00
6.81302488e-01 1.65905550e-01 -6.85923323e-02 1.04581618e+00
6.83919787e-01 6.94741488e-01 4.77047265e-01 -1.06390524e+00
3.30018885e-02 4.13195878e-01 -4.11933148e-03 -1.28448606e+00
-2.88561378e-02 1.38773158e-01 -9.29510236e-01 2.08909288e-01
-4.38776910e-02 -2.63178200e-01 -7.88318813e-01 1.69460988e+00
6.71573818e-01 3.23133260e-01 4.54517826e-03 6.22390032e-01
9.41467345e-01 9.03012812e-01 -1.23005643e-01 -5.21832466e-01
1.26393628e+00 -1.12880540e+00 -1.05304861e+00 2.65791774e-01
3.01540196e-01 -1.17939734e+00 1.04559720e+00 3.42028737e-01
-1.24631178e+00 -6.88950717e-01 -8.71459663e-01 9.42644197e-03
7.42388815e-02 1.91755369e-01 2.81159371e-01 7.00265408e-01
-1.08360219e+00 4.39614385e-01 -6.70780599e-01 -4.11739260e-01
3.13655406e-01 3.95563275e-01 -4.27569747e-01 2.45326072e-01
-7.54157543e-01 5.38289189e-01 3.14760171e-02 2.26400897e-01
-6.12433791e-01 -6.93247139e-01 -7.72668064e-01 -3.04950029e-01
3.04519653e-01 -6.97119474e-01 1.42112863e+00 -1.35224235e+00
-2.21148181e+00 1.11589730e+00 -3.58049929e-01 2.82452345e-01
8.65726173e-01 -1.88370004e-01 -4.13834810e-01 -1.66625127e-01
-1.44756407e-01 6.29103720e-01 1.29033017e+00 -1.55290151e+00
-4.38450724e-01 -2.09478796e-01 -5.96199572e-01 2.06228405e-01
-2.68098086e-01 1.57114252e-01 -9.21487272e-01 -1.10107803e+00
1.00466773e-01 -1.16571045e+00 4.37523350e-02 2.11917743e-01
-2.28287816e-01 4.55115968e-03 1.68217671e+00 -6.44351661e-01
1.22108126e+00 -2.34404469e+00 4.94548053e-01 3.03120166e-01
2.67846942e-01 2.09701553e-01 -1.93531170e-01 2.90127516e-01
-2.41663963e-01 2.44603660e-02 -3.15773666e-01 -5.28036058e-01
-1.67750180e-01 1.53338462e-01 -2.19251409e-01 3.50644320e-01
-3.37075740e-02 9.38041508e-01 -7.26848900e-01 -5.89439690e-01
3.68387215e-02 5.19775093e-01 -6.91845596e-01 2.85262346e-01
-5.98307028e-02 8.09078515e-01 -3.89592856e-01 7.33156621e-01
1.13110340e+00 1.56106845e-01 1.44386634e-01 -4.53524500e-01
-1.20927170e-01 -4.84927952e-01 -1.18176663e+00 1.47862542e+00
-2.99922913e-01 3.43810886e-01 3.56887817e-01 -3.00440520e-01
1.12490475e+00 3.69577527e-01 7.01798797e-01 -1.44542217e-01
3.50313812e-01 3.53561521e-01 -1.96082219e-01 -6.65737927e-01
5.14528811e-01 -3.70951116e-01 1.82487771e-01 5.03220558e-01
-1.29967451e-01 -5.34551978e-01 -2.01820061e-01 -9.82926935e-02
3.15409750e-01 6.40522778e-01 1.55140772e-01 -6.64055765e-01
7.91358232e-01 -2.56177217e-01 7.76082098e-01 -4.31859419e-02
-7.74084777e-02 7.63311505e-01 5.78363419e-01 -2.86831707e-01
-1.09778309e+00 -9.58747506e-01 1.31999195e-01 7.48000622e-01
1.73246801e-01 -5.02671540e-01 -1.16929424e+00 -5.38555205e-01
-2.19319582e-01 1.99385181e-01 -7.81723857e-01 -1.17129199e-01
-8.68708253e-01 -5.62780321e-01 2.62844533e-01 4.04462665e-01
7.93385506e-01 -1.25100994e+00 -4.47702736e-01 -1.15756892e-01
-1.51594207e-01 -9.24324512e-01 -1.08077455e+00 -7.54339099e-01
-7.81600416e-01 -8.03602695e-01 -8.68839443e-01 -1.10384655e+00
7.60020792e-01 3.06556135e-01 8.79445970e-01 1.31342560e-01
-2.82705724e-01 5.92156291e-01 -2.40720242e-01 -3.09159577e-01
-5.55475652e-01 -2.19376400e-01 3.65875781e-01 5.14718354e-01
-1.91348139e-02 -7.83109426e-01 -5.30131757e-01 6.42735481e-01
-9.76004899e-01 6.29169583e-01 5.23412764e-01 7.12857068e-01
5.89197040e-01 -4.27111574e-02 2.91512609e-01 -1.09904051e+00
4.59337026e-01 -2.04723075e-01 -4.64406997e-01 1.04571827e-01
-1.87446684e-01 -7.20159188e-02 4.68191952e-01 -5.61994433e-01
-1.42656136e+00 2.15997353e-01 -4.16646861e-02 -7.62183845e-01
8.03797394e-02 9.06639695e-02 -7.04844475e-01 -2.41933852e-01
2.94090778e-01 3.52438569e-01 5.18569589e-01 -4.59998578e-01
5.81313252e-01 6.16892934e-01 6.89218879e-01 -7.13882744e-01
1.17509854e+00 7.51798213e-01 -6.09774366e-02 -1.14494669e+00
-1.40188694e-01 1.38792098e-02 -8.51113617e-01 -5.11442959e-01
7.52841353e-01 -6.04751885e-01 -6.34222269e-01 1.04533589e+00
-1.11508608e+00 -3.15292984e-01 -3.65643919e-01 3.23250778e-02
-1.14403260e+00 5.05541801e-01 -4.73204255e-01 -6.00602508e-01
-3.32302302e-01 -1.13475263e+00 1.02848876e+00 3.34946752e-01
-2.35400006e-01 -9.24760044e-01 2.37257630e-01 1.83189973e-01
3.50038469e-01 4.94308710e-01 9.25946116e-01 1.97645977e-01
-3.38769794e-01 2.64359087e-01 -1.21540517e-01 1.33076370e-01
6.71937704e-01 6.25722408e-01 -8.72093022e-01 -4.61087286e-01
2.71918982e-01 9.20135155e-03 3.50257158e-01 2.51892358e-01
9.62598324e-01 -2.65408307e-01 -4.26224530e-01 8.66448104e-01
8.64126384e-01 6.85742378e-01 8.75378489e-01 -7.74636120e-02
8.52157056e-01 9.17831182e-01 7.49274790e-01 4.88628358e-01
1.51919518e-02 8.35123777e-01 -1.20045722e-01 -1.09759174e-01
-3.11052322e-01 -5.57431042e-01 4.59779084e-01 1.34737933e+00
-3.00448418e-01 -7.15006003e-03 -6.70042336e-01 2.07237378e-01
-1.91920888e+00 -6.60485208e-01 1.98237617e-02 1.83250141e+00
5.21201372e-01 -3.86352152e-01 1.50237009e-01 -3.41764353e-02
9.12908673e-01 6.93322495e-02 -4.49026793e-01 -2.66965985e-01
-5.73383737e-03 -7.47389495e-02 -4.83390205e-02 5.93889832e-01
-9.95079517e-01 1.34100556e+00 6.59399986e+00 9.38692689e-01
-1.20335114e+00 6.58339486e-02 6.61789834e-01 3.62355530e-01
-3.54357898e-01 -9.49530583e-03 -5.61261535e-01 4.89120573e-01
3.67059946e-01 -3.64545345e-01 3.12862694e-01 5.15060306e-01
4.11427051e-01 2.51905411e-01 -8.97251248e-01 1.37724638e+00
3.74221087e-01 -1.05445147e+00 4.17934179e-01 -8.35490227e-02
8.28064561e-01 -8.65841687e-01 4.26937342e-01 1.12243004e-01
4.19763885e-02 -9.69517827e-01 7.79865205e-01 6.83628023e-01
1.09914720e+00 -8.98135066e-01 1.03518523e-01 1.12585329e-01
-1.51242292e+00 2.26250753e-01 -2.28293449e-01 2.88092136e-01
2.05417946e-01 1.58231691e-01 -2.20904976e-01 5.96615374e-01
4.18156564e-01 8.73250842e-01 -8.44798088e-02 8.24815631e-01
-1.55821275e-02 8.66138116e-02 -1.88129842e-02 2.56283373e-01
-9.89035144e-02 -6.11169994e-01 7.26119637e-01 9.74052072e-01
5.54123044e-01 3.17401439e-01 -2.92855259e-02 8.92479658e-01
-3.14961970e-02 3.56061041e-01 -6.44867539e-01 1.88849181e-01
2.73512304e-01 1.22625279e+00 -5.86893082e-01 -2.67925084e-01
-2.43601322e-01 1.14920831e+00 -1.50867388e-01 3.00732166e-01
-1.05436373e+00 -2.28185102e-01 5.68377554e-01 3.03926289e-01
7.79028609e-02 -2.82455891e-01 -1.32425144e-01 -1.27901113e+00
-1.92856103e-01 -1.03673613e+00 -4.71807607e-02 -9.47772443e-01
-1.00012660e+00 7.00073540e-01 1.95887581e-01 -1.50921488e+00
-1.22299977e-01 -5.51592767e-01 -6.27170146e-01 5.56667447e-01
-9.98514295e-01 -1.46346080e+00 -5.57235181e-01 6.70277953e-01
8.57355654e-01 -2.59865522e-01 4.98252362e-01 3.01697999e-01
-6.62859499e-01 6.45186543e-01 5.82348891e-02 -1.70341000e-01
7.79059470e-01 -5.65547943e-01 4.86479878e-01 6.14193857e-01
-1.76022962e-01 4.33094174e-01 6.48515940e-01 -6.99143291e-01
-1.57122374e+00 -1.20868421e+00 5.17636716e-01 -3.25442761e-01
4.75558221e-01 -5.44040143e-01 -7.41709888e-01 8.03915024e-01
9.49386135e-02 -3.14086229e-01 3.60482693e-01 -5.04717350e-01
-9.52408016e-02 -4.19052355e-02 -1.14395094e+00 1.12736428e+00
1.31709266e+00 -2.45414346e-01 -4.95538920e-01 8.54270607e-02
4.94173586e-01 -4.51741487e-01 -6.57579601e-01 5.18678248e-01
7.87967622e-01 -7.97835290e-01 6.93129897e-01 -2.53912270e-01
4.18723881e-01 -3.87749374e-01 3.59634571e-02 -1.17593348e+00
-3.97340238e-01 -1.32158744e+00 -1.47199050e-01 1.44729936e+00
1.56369343e-01 -7.18894780e-01 9.04511392e-01 3.98246050e-01
8.61762986e-02 -6.17563903e-01 -6.79917097e-01 -8.17507207e-01
1.77857742e-01 1.05262227e-01 6.93752646e-01 1.16096997e+00
-9.46481973e-02 1.71362266e-01 -8.78154457e-01 -2.02175736e-01
5.10854185e-01 1.09210424e-01 1.09227586e+00 -1.03731644e+00
-1.92726582e-01 -3.34727466e-01 -1.97913244e-01 -1.27727139e+00
1.79103225e-01 -6.53586984e-01 9.43320096e-02 -9.82803404e-01
4.49121147e-01 -2.90067613e-01 1.71878099e-01 2.46599182e-01
1.48868695e-01 1.90346181e-01 2.57624865e-01 4.22287762e-01
-3.24752694e-03 8.87858629e-01 1.67683959e+00 7.69067630e-02
-4.64154184e-01 -2.28965536e-01 -6.45511746e-01 9.82195973e-01
5.52934110e-01 -3.34956378e-01 -7.10229933e-01 -3.19511980e-01
-2.66753852e-01 2.44153216e-01 2.43638828e-02 -9.42673326e-01
7.95452148e-02 -4.60173935e-01 1.46106586e-01 -4.97551531e-01
3.36026341e-01 -7.99185455e-01 7.52212346e-01 4.53429639e-01
1.24710426e-01 3.25747222e-01 4.02578652e-01 3.50227922e-01
-1.68644682e-01 1.60574943e-01 1.25937736e+00 2.52270196e-02
-5.73148787e-01 5.64230561e-01 -5.84261835e-01 -1.57737419e-01
1.21864820e+00 -3.90283138e-01 -9.07926336e-02 -5.39456427e-01
-9.16089296e-01 -8.74525309e-02 6.00421488e-01 7.40846276e-01
7.75589049e-01 -1.70187581e+00 -8.38803530e-01 6.87149286e-01
-2.50660390e-01 -1.05971910e-01 4.06378865e-01 8.61210167e-01
-8.53185356e-01 -5.77938631e-02 -4.95438874e-01 -5.26872337e-01
-1.48291826e+00 4.68869895e-01 2.92762697e-01 8.01828578e-02
-6.10728800e-01 6.39440358e-01 8.75318825e-01 -4.91872638e-01
-1.65225461e-01 -5.84611110e-02 -7.23597109e-02 -6.47824854e-02
3.92435431e-01 3.91414195e-01 -3.99129629e-01 -1.21876013e+00
-1.57513142e-01 1.44114280e+00 -8.10551867e-02 -1.97931468e-01
9.66083944e-01 -2.28743985e-01 -4.28219587e-01 3.11219513e-01
1.10767078e+00 3.44387740e-01 -1.32558608e+00 5.64618371e-02
-2.05609724e-01 -6.89122438e-01 -4.30287570e-01 -2.77247816e-01
-1.31206381e+00 6.72828734e-01 5.63648641e-01 -4.17454630e-01
1.23788059e+00 -2.56740868e-01 8.39942813e-01 -5.44089898e-02
3.82124484e-01 -9.75324094e-01 3.48092556e-01 5.64562500e-01
1.21127796e+00 -6.41246557e-01 -1.43307552e-01 -1.03488100e+00
-6.17438793e-01 1.04569554e+00 7.50312388e-01 -1.58336639e-01
9.03215289e-01 4.63435948e-01 1.91550136e-01 -1.03313580e-01
-5.73525727e-01 5.02600849e-01 2.29140773e-01 5.47041237e-01
3.65748197e-01 -3.66110839e-02 -2.83241004e-01 5.56323230e-01
-3.44269723e-01 7.19770342e-02 5.57678521e-01 9.30859447e-01
-3.22313309e-01 -1.11875319e+00 -6.10138059e-01 -1.05529174e-01
-1.65234342e-01 2.54476309e-01 -4.36527669e-01 9.19071972e-01
1.81069538e-01 7.15852499e-01 1.25672713e-01 -6.28654301e-01
5.26607931e-01 -6.45755306e-02 5.88074565e-01 -3.70843977e-01
-1.85533360e-01 5.09220421e-01 -2.88603008e-01 -3.93621862e-01
-4.51564938e-01 -5.38776159e-01 -1.02745283e+00 -5.73833048e-01
-1.66701712e-02 1.48017397e-02 1.41400278e-01 5.57365060e-01
4.62440878e-01 3.30552459e-01 8.87295961e-01 -1.03294694e+00
-3.80285941e-02 -7.22595215e-01 -8.01167846e-01 4.17283624e-01
4.64690030e-02 -7.44378388e-01 -2.48322293e-01 4.63739902e-01] | [12.724838256835938, -0.3030398488044739] |
c6dc0d99-ce0c-4d93-a2f1-6b9028550c20 | medmcqa-a-large-scale-multi-subject-multi | 2203.14371 | null | https://arxiv.org/abs/2203.14371v1 | https://arxiv.org/pdf/2203.14371v1.pdf | MedMCQA : A Large-scale Multi-Subject Multi-Choice Dataset for Medical domain Question Answering | This paper introduces MedMCQA, a new large-scale, Multiple-Choice Question Answering (MCQA) dataset designed to address real-world medical entrance exam questions. More than 194k high-quality AIIMS \& NEET PG entrance exam MCQs covering 2.4k healthcare topics and 21 medical subjects are collected with an average token length of 12.77 and high topical diversity. Each sample contains a question, correct answer(s), and other options which requires a deeper language understanding as it tests the 10+ reasoning abilities of a model across a wide range of medical subjects \& topics. A detailed explanation of the solution, along with the above information, is provided in this study. | ['Malaikannan Sankarasubbu', 'Logesh Kumar Umapathi', 'Ankit Pal'] | 2022-03-27 | null | null | null | null | ['multiple-choice-qa'] | ['natural-language-processing'] | [-1.43039629e-01 3.83939475e-01 -4.00145471e-01 -4.47699606e-01
-1.44931519e+00 -7.70499051e-01 -1.82810381e-01 6.23792112e-01
-5.79736888e-01 9.80984032e-01 5.14383614e-01 -8.94556582e-01
-9.95214283e-01 -7.02140629e-01 -3.52616072e-01 -1.16033636e-01
2.16786250e-01 8.39253664e-01 2.63240397e-01 -4.63157862e-01
3.03498328e-01 -1.83104739e-01 -9.81419683e-01 8.45479131e-01
1.56283927e+00 9.24320698e-01 9.37731117e-02 1.16770196e+00
-4.38950479e-01 1.21847582e+00 -8.41392040e-01 -1.07475674e+00
-2.64810026e-01 -4.07584429e-01 -1.73443604e+00 -4.49863136e-01
8.26195478e-01 3.49789523e-02 -3.21378648e-01 7.85387039e-01
7.16278315e-01 1.73740789e-01 3.35509956e-01 -7.95911193e-01
-6.32279634e-01 4.77698386e-01 3.05994481e-01 7.62666583e-01
1.04407156e+00 1.07865125e-01 1.29576945e+00 -2.16426462e-01
9.09730256e-01 9.59826291e-01 3.77330959e-01 8.17396522e-01
-5.07659197e-01 -4.98586684e-01 -2.39083275e-01 4.97659206e-01
-1.04122627e+00 1.58195838e-01 3.20380241e-01 -2.20798984e-01
8.32904816e-01 8.23173702e-01 7.12827563e-01 9.24409449e-01
6.65863216e-01 1.11766386e+00 1.21449435e+00 -1.21057034e-01
2.64009118e-01 1.62013873e-01 6.28447056e-01 7.50639379e-01
1.10515222e-01 -5.56810737e-01 -4.84536767e-01 -7.74616420e-01
3.12694281e-01 -2.37121046e-01 -1.04334384e-01 4.89909887e-01
-1.12490034e+00 7.15228498e-01 3.68873119e-01 3.05740386e-01
-1.71926916e-01 -1.96218908e-01 6.18264861e-02 6.45536244e-01
-3.79412770e-01 1.34939015e+00 -1.05552340e+00 -4.22080576e-01
-5.16273022e-01 8.57925832e-01 1.04148769e+00 9.87301409e-01
1.91629499e-01 -8.50401223e-01 -6.91286206e-01 5.56718826e-01
1.83331653e-01 5.94528317e-01 7.07015693e-01 -1.24060822e+00
6.46089435e-01 1.02924132e+00 -1.47339806e-01 -9.85790312e-01
-4.76849109e-01 -5.77238858e-01 -4.69991982e-01 -9.14002240e-01
5.48719943e-01 -1.59799382e-01 -7.63833106e-01 1.31380069e+00
4.36129898e-01 3.21635418e-02 2.78344095e-01 6.34834409e-01
1.80913413e+00 2.59339452e-01 3.99670899e-01 1.76600337e-01
2.36454725e+00 -6.85389459e-01 -1.15212655e+00 -2.17248023e-01
8.05844307e-01 -9.31456029e-01 1.25607729e+00 4.46191341e-01
-1.35313976e+00 -2.08905250e-01 -3.33968639e-01 -5.40856719e-01
-3.49151880e-01 -2.62887478e-01 4.75465775e-01 7.26397514e-01
-1.01276970e+00 -8.45697969e-02 -1.70324191e-01 -1.66321974e-02
3.14530283e-01 1.68543011e-01 9.23585519e-02 -8.31279814e-01
-1.87317145e+00 7.57100165e-01 4.03015465e-02 -4.80452210e-01
-5.51770627e-01 -1.28544331e+00 -8.16240191e-01 2.91610938e-02
7.31895030e-01 -9.85308886e-01 1.64916360e+00 5.38099033e-04
-1.26623702e+00 8.33361745e-01 -2.02720076e-01 -1.21793091e-01
4.22580153e-01 -1.61641389e-01 -8.01966071e-01 6.04353487e-01
2.12423936e-01 5.17301917e-01 -6.40042424e-02 -4.33009267e-01
-5.59564888e-01 -1.79989580e-02 4.68327492e-01 4.64678138e-01
1.72113869e-02 -1.25741005e-01 -5.62959433e-01 -5.28736055e-01
-1.27704730e-02 -7.11977959e-01 -5.77782452e-01 -3.63786936e-01
-3.13285708e-01 -7.20837772e-01 -1.61397979e-01 -7.99105883e-01
1.81548429e+00 -1.81207228e+00 -1.83914363e-01 3.06197733e-01
4.45877701e-01 8.52819085e-02 -1.75127313e-01 4.91216153e-01
1.99503839e-01 3.57448459e-01 -2.30692774e-01 3.29648763e-01
-9.86544862e-02 3.20328414e-01 -1.16206191e-01 -1.94592401e-01
8.31558034e-02 1.37984979e+00 -1.19715428e+00 -1.13873076e+00
-2.48645246e-01 -3.84533495e-01 -8.36816549e-01 2.71243364e-01
-5.50503790e-01 2.31411755e-01 -1.05520904e+00 8.95258307e-01
2.10923463e-01 -6.15242898e-01 6.54576272e-02 4.10287976e-01
5.35340309e-01 7.18682289e-01 -1.01464462e+00 1.78704107e+00
-9.23155770e-02 1.20852821e-01 -2.54929602e-01 -3.77935886e-01
4.25704479e-01 5.48976243e-01 4.77066427e-01 -9.17928815e-01
-1.38466179e-01 4.78346169e-01 2.18150482e-01 -1.19852781e+00
5.57427466e-01 2.63120886e-02 -5.17935753e-01 2.40683019e-01
-3.80395614e-02 -2.33278096e-01 3.31504494e-01 7.06391990e-01
1.53677750e+00 -7.42774785e-01 4.64756042e-01 -6.88462079e-01
7.26980746e-01 6.03928626e-01 5.89762151e-01 1.13114965e+00
-3.17236573e-01 2.02716127e-01 5.54693103e-01 -3.69641453e-01
-2.55078495e-01 -1.08740854e+00 -5.61386406e-01 9.07966673e-01
-6.86710030e-02 -6.85518622e-01 -4.85154271e-01 -8.03497195e-01
8.65510106e-02 5.54251492e-01 -6.33434296e-01 1.01914249e-01
-6.30841017e-01 -3.82238150e-01 8.45600128e-01 2.80936003e-01
4.11043078e-01 -1.20173323e+00 -3.06133479e-01 2.84298986e-01
-1.03695953e+00 -1.14292526e+00 -5.21647096e-01 -4.02152151e-01
-9.31702614e-01 -1.62410581e+00 -6.91784501e-01 -9.47066486e-01
6.17397606e-01 -3.00324827e-01 1.86824596e+00 3.61038476e-01
-5.21732509e-01 8.28578472e-01 -3.47658962e-01 -3.62796992e-01
-1.55688420e-01 3.87338042e-01 -3.98539603e-01 -7.49115705e-01
7.11400628e-01 4.99962240e-01 -8.52327049e-01 1.95503697e-01
-1.08712089e+00 -3.93942475e-01 6.90491915e-01 7.59241998e-01
7.10585654e-01 -5.54671995e-02 7.38827705e-01 -1.20362842e+00
1.07485640e+00 -7.72713304e-01 -1.42854378e-02 8.01752269e-01
-5.19845426e-01 -6.24197237e-02 4.51451927e-01 -2.68993527e-01
-7.74937212e-01 -6.18589044e-01 -6.69794261e-01 4.04349476e-01
-3.36719483e-01 9.24816668e-01 4.70788889e-02 7.34392032e-02
8.84850383e-01 1.76701427e-01 -4.05143738e-01 -3.30101341e-01
1.63572311e-01 5.42253375e-01 3.46469462e-01 -6.90842748e-01
4.28801954e-01 -1.54963061e-01 -3.33950788e-01 -6.33892477e-01
-1.22487199e+00 -8.35122108e-01 6.02639988e-02 -1.56991795e-01
9.02531981e-01 -6.76258981e-01 -1.38858104e+00 5.95918521e-02
-6.53958499e-01 -1.87211215e-01 -3.97811264e-01 7.32265651e-01
-1.29719511e-01 2.66752064e-01 -1.07596028e+00 -4.32205141e-01
-2.39970282e-01 -1.11768341e+00 6.36356413e-01 4.33102548e-01
-5.43231070e-01 -1.05408204e+00 1.39717817e-01 1.23368585e+00
1.70029998e-01 -2.06551298e-01 1.30978727e+00 -7.83285618e-01
-6.35368824e-01 -2.33126983e-01 1.94533646e-01 -3.49332631e-01
-1.33937776e-01 -6.29805624e-01 -3.79343420e-01 -2.69053012e-01
1.42245337e-01 -5.65866351e-01 5.86722791e-01 5.55247009e-01
1.70576763e+00 -3.36432487e-01 -2.69561827e-01 7.21625313e-02
1.26065350e+00 7.22115487e-02 7.24949360e-01 1.72584429e-01
2.15045854e-01 6.65678442e-01 8.49262774e-01 1.86780110e-01
1.05325961e+00 8.47042724e-02 1.22815967e-01 9.46022645e-02
3.36591840e-01 -2.84670442e-02 -1.79649755e-01 1.38338673e+00
4.47634816e-01 -4.05685484e-01 -1.50302827e+00 6.90138638e-01
-1.14091945e+00 -6.17824316e-01 -2.03691795e-01 1.52946866e+00
1.37553418e+00 -2.50711385e-02 -1.76393732e-01 2.42402661e-03
-5.16306795e-02 -1.80315256e-01 -6.64533496e-01 -4.11747217e-01
-5.52891120e-02 7.18884051e-01 2.42465258e-01 7.17444599e-01
-5.12928486e-01 5.60082495e-01 7.25899029e+00 1.15274930e+00
-3.69091213e-01 -2.61222512e-01 8.40541959e-01 1.47133440e-01
-9.66478586e-01 -3.83241862e-01 -8.65729272e-01 3.11607242e-01
1.15101290e+00 -1.84593171e-01 -1.02521479e-01 5.12981951e-01
-4.30997312e-01 -4.24774557e-01 -6.13962233e-01 6.40585899e-01
5.29626571e-02 -1.56807172e+00 2.00005934e-01 -4.24534194e-02
7.02327192e-01 -3.54217261e-01 2.09133491e-01 6.13201678e-01
6.25327647e-01 -1.26929951e+00 -1.52374372e-01 8.91868114e-01
6.50707960e-01 -5.93779504e-01 1.12732089e+00 4.88988638e-01
-8.99660110e-01 -1.74853295e-01 -1.39504284e-01 4.40614261e-02
-3.12338233e-01 5.40931880e-01 -9.95125413e-01 1.01046598e+00
9.68897879e-01 1.89817175e-01 -8.01828444e-01 1.19517398e+00
-1.32346705e-01 8.89422596e-01 -1.42592475e-01 -4.51716542e-01
4.40692991e-01 3.45214278e-01 7.97159746e-02 1.14882219e+00
1.13476470e-01 1.00511837e+00 1.69428155e-01 2.32866064e-01
-7.84318745e-02 5.14760017e-01 1.01693280e-01 -2.33027875e-01
5.10497808e-01 9.33583915e-01 -4.54898179e-01 -5.45745552e-01
-1.96312293e-01 3.61648828e-01 -7.96262734e-03 2.51803845e-01
-5.22641659e-01 -4.58786607e-01 6.20421112e-01 -1.32820569e-03
-2.61512011e-01 2.44296953e-01 -3.97963345e-01 -1.15427351e+00
-5.66569231e-02 -1.49057233e+00 1.41809952e+00 -4.01751429e-01
-1.66887999e+00 3.85545373e-01 -1.62581086e-01 -1.06629717e+00
-1.40768453e-01 -7.07033157e-01 -2.11533666e-01 8.41318488e-01
-1.80721581e+00 -3.04923207e-01 -2.41104290e-01 9.09467816e-01
3.94595653e-01 -5.29534630e-02 1.11521971e+00 5.42211652e-01
-1.59265667e-01 1.01398921e+00 -1.04455709e-01 2.86216259e-01
8.99674237e-01 -1.57505620e+00 -1.35393059e-02 1.16401955e-01
-3.93931240e-01 8.09189737e-01 4.29498523e-01 -7.14359224e-01
-1.58151102e+00 -6.47927105e-01 1.58778489e+00 -1.02367508e+00
1.81329176e-01 1.40451118e-01 -1.10764205e+00 3.21329921e-01
3.55887771e-01 -2.70973593e-01 1.78486884e+00 1.70380592e-01
1.29156411e-01 8.33983123e-02 -1.46523166e+00 6.44798338e-01
6.66179895e-01 -5.42851388e-01 -1.03240454e+00 4.46696401e-01
9.59960818e-01 -1.08513105e+00 -1.49488676e+00 4.66202259e-01
2.95212001e-01 -2.53333151e-01 1.21003771e+00 -1.30465341e+00
7.17150807e-01 -5.78274578e-02 8.24666619e-02 -7.37643123e-01
-3.10894340e-01 -3.06878269e-01 -1.87947661e-01 3.82958144e-01
7.05184937e-01 -4.58273232e-01 9.17959273e-01 1.03962886e+00
-2.28916973e-01 -1.50086212e+00 -1.08159292e+00 -2.83155609e-02
4.46379334e-01 -5.56631982e-01 8.08734357e-01 1.47640240e+00
3.66433501e-01 -1.10962801e-01 2.10097730e-01 2.12534472e-01
2.35551268e-01 2.25033253e-01 1.12608679e-01 -1.06871605e+00
-3.20633203e-01 -3.09615731e-01 -1.36329699e-03 -1.29521549e+00
-3.39917958e-01 -9.93781030e-01 -3.29452097e-01 -1.84924316e+00
8.55064020e-02 -4.31970716e-01 -3.81955683e-01 4.18638796e-01
-8.41681659e-01 -2.18752667e-01 -2.61108071e-01 -2.77415574e-01
-9.33744013e-01 2.23955489e-04 1.97063887e+00 -6.91288114e-02
2.28875697e-01 1.21268585e-01 -9.77769971e-01 2.21763685e-01
7.22942173e-01 -5.16220868e-01 -5.18893480e-01 -2.77825266e-01
7.12130010e-01 7.89664209e-01 -1.31222188e-01 -6.73345804e-01
5.52104950e-01 -6.68966055e-01 4.91866529e-01 -6.51501417e-01
8.76255333e-02 -4.44596738e-01 -4.07949537e-01 9.69092071e-01
-1.05899262e+00 3.45910519e-01 3.89762878e-01 3.24123144e-01
-4.29289430e-01 -3.16057324e-01 2.74590552e-01 -6.34597242e-01
-5.91004193e-01 3.25146914e-01 -6.12013876e-01 9.86961305e-01
7.43632495e-01 1.71949580e-01 -4.79953527e-01 -3.04533213e-01
-9.32988942e-01 1.32755494e+00 -3.26132596e-01 3.93835068e-01
9.19522822e-01 -1.17812860e+00 -1.00917304e+00 -1.58454195e-01
3.35602999e-01 2.02848539e-01 1.01879120e+00 5.03521800e-01
-7.59386063e-01 8.61326218e-01 5.89444339e-02 -3.12265486e-01
-1.20961154e+00 1.96244493e-01 4.64961976e-01 -9.57168460e-01
-1.86914533e-01 1.21641850e+00 -3.22496325e-01 -9.50278103e-01
9.33520794e-02 -5.52924097e-01 -6.72192574e-01 -2.94355250e-05
6.55218780e-01 2.47563615e-01 2.60913819e-02 9.83599648e-02
-4.22716856e-01 2.51959115e-01 -1.89874277e-01 1.05208486e-01
6.39567971e-01 9.15553793e-02 -1.09320968e-01 1.76255614e-01
8.78410697e-01 6.75307810e-02 -1.44435242e-01 -6.12552464e-01
2.61922926e-01 -3.40756297e-01 -3.11561704e-01 -1.34980083e+00
-5.41144609e-01 4.46015090e-01 3.57636899e-01 1.08636096e-01
9.85696435e-01 2.80225754e-01 1.03570569e+00 8.07651520e-01
-7.37787187e-02 -1.10447729e+00 4.09373134e-01 6.66796088e-01
6.07512057e-01 -1.31990385e+00 -2.11759239e-01 -4.10512537e-01
-8.06542158e-01 8.09548199e-01 9.67981100e-01 2.01487496e-01
5.90228975e-01 -2.66962916e-01 5.09436309e-01 -6.47293389e-01
-9.22262549e-01 5.92475981e-02 7.73791432e-01 -3.77976596e-02
6.12310231e-01 1.92415670e-01 -6.59569919e-01 7.71761298e-01
-8.59804809e-01 4.77708131e-03 4.76583421e-01 1.05339062e+00
-1.60028607e-01 -1.05391192e+00 -3.70450109e-01 9.62029517e-01
-9.23946142e-01 -2.01944768e-01 -2.96070933e-01 3.90858322e-01
5.09097762e-02 1.13024652e+00 -2.19975725e-01 -8.24933350e-02
4.70463455e-01 2.17509612e-01 1.95611849e-01 -6.68194950e-01
-1.19617677e+00 -5.42575121e-01 -5.79317920e-02 -5.71464598e-01
-1.11525014e-01 -6.14762962e-01 -1.30495918e+00 -3.41682851e-01
-4.02274821e-03 8.41656446e-01 1.50748834e-01 8.88519406e-01
4.19274539e-01 9.16716039e-01 7.14795887e-02 1.06611490e+00
-4.12721694e-01 -8.18531036e-01 -3.44638050e-01 4.28795755e-01
5.33050776e-01 -1.04846604e-01 -1.04393857e-02 -4.12144631e-01] | [8.875441551208496, 8.513313293457031] |
11db914b-3e7c-4fc7-af53-dd4ea61e279e | self-supervised-auxiliary-learning-for-graph | 2103.00771 | null | https://arxiv.org/abs/2103.00771v2 | https://arxiv.org/pdf/2103.00771v2.pdf | Self-supervised Auxiliary Learning for Graph Neural Networks via Meta-Learning | In recent years, graph neural networks (GNNs) have been widely adopted in the representation learning of graph-structured data and provided state-of-the-art performance in various applications such as link prediction, node classification, and recommendation. Motivated by recent advances of self-supervision for representation learning in natural language processing and computer vision, self-supervised learning has been recently studied to leverage unlabeled graph-structured data. However, employing self-supervision tasks as auxiliary tasks to assist a primary task has been less explored in the literature on graphs. In this paper, we propose a novel self-supervised auxiliary learning framework to effectively learn graph neural networks. Moreover, this work is the first study showing that a meta-path prediction is beneficial as a self-supervised auxiliary task for heterogeneous graphs. Our method is learning to learn a primary task with various auxiliary tasks to improve generalization performance. The proposed method identifies an effective combination of auxiliary tasks and automatically balances them to improve the primary task. Our methods can be applied to any graph neural network in a plug-in manner without manual labeling or additional data. Also, it can be extended to any other auxiliary tasks. Our experiments demonstrate that the proposed method consistently improves the performance of node classification and link prediction. | ['Hyunwoo J. Kim', 'Jung-Woo Ha', 'Kyung-Min Kim', 'Sunyoung Kwon', 'Jinyoung Park', 'Dasol Hwang'] | 2021-03-01 | null | null | null | null | ['auxiliary-learning'] | ['methodology'] | [ 4.67448682e-01 5.43663502e-01 -7.68382192e-01 -3.41421902e-01
-1.61843628e-01 -1.88637882e-01 5.44600487e-01 5.83659053e-01
-1.51390687e-01 7.27971554e-01 -1.26319289e-01 -5.52955091e-01
-2.34192640e-01 -1.08520961e+00 -5.88430643e-01 -6.40028715e-01
-1.65907145e-01 5.35427988e-01 3.97724360e-01 -3.28696400e-01
-2.71776050e-01 4.84507799e-01 -1.29918826e+00 -1.43252194e-01
1.10785389e+00 7.84743667e-01 2.66190022e-01 2.92953491e-01
-2.64148206e-01 1.13045049e+00 -9.78990346e-02 -4.60244209e-01
1.49986878e-01 -3.07451695e-01 -8.27298701e-01 9.62990895e-02
2.03040957e-01 2.21199825e-01 -5.27224958e-01 1.00851703e+00
3.06373239e-01 1.15970790e-01 5.81977487e-01 -1.51004839e+00
-7.10382402e-01 8.36578727e-01 -6.12606704e-01 1.63580343e-01
2.11092755e-02 -3.23833555e-01 1.25499022e+00 -5.37994504e-01
6.49929821e-01 1.04668260e+00 6.18510246e-01 3.07400316e-01
-1.09855461e+00 -7.58121848e-01 3.38993907e-01 3.22991550e-01
-1.24769890e+00 -2.23844245e-01 1.33507311e+00 -1.99122772e-01
7.91775048e-01 -9.53934565e-02 6.03141129e-01 1.02534592e+00
-3.63588557e-02 8.30558181e-01 7.15574443e-01 -4.95723218e-01
1.30466921e-02 2.62815118e-01 4.41332489e-01 1.07435298e+00
4.46758389e-01 -1.39757171e-01 -1.58215612e-01 -5.49547970e-02
6.36472821e-01 1.18431754e-01 -1.91653162e-01 -6.94969714e-01
-8.71864378e-01 9.04364347e-01 8.25948238e-01 4.23748851e-01
-2.33045012e-01 6.44702837e-02 5.45089126e-01 5.47578514e-01
8.44522953e-01 1.64727166e-01 -3.61970454e-01 3.67066771e-01
-4.93235230e-01 -4.34296459e-01 8.81998062e-01 9.63489711e-01
9.97846067e-01 2.84242421e-01 2.63393987e-02 9.87789273e-01
2.89898068e-01 1.80726513e-01 2.97422647e-01 -3.27127367e-01
6.73295915e-01 1.10164058e+00 -5.65675855e-01 -1.20322788e+00
-6.55455947e-01 -9.97713268e-01 -1.37748480e+00 -2.57034898e-01
1.91511720e-01 2.58190818e-02 -7.83494353e-01 1.84187937e+00
2.94496745e-01 3.87474626e-01 7.26574799e-03 4.63557661e-01
1.04775918e+00 3.88566017e-01 1.72972366e-01 -3.30622733e-01
1.04567027e+00 -1.29655433e+00 -6.66839957e-01 -2.37553447e-01
1.09305131e+00 -3.08709264e-01 7.49686658e-01 1.17227301e-01
-5.45871019e-01 -5.75115502e-01 -1.00672364e+00 1.08826585e-01
-4.88737911e-01 1.28053904e-01 1.11718130e+00 4.94561881e-01
-1.16562152e+00 5.46702266e-01 -7.12210059e-01 -7.14765608e-01
5.79387963e-01 4.72245902e-01 -7.29256451e-01 -7.85659775e-02
-1.17197263e+00 6.81058109e-01 7.43787646e-01 -5.97200170e-02
-4.70347792e-01 -3.06456268e-01 -1.11598372e+00 2.93653429e-01
8.85334253e-01 -7.04527199e-01 5.80576301e-01 -1.05392015e+00
-1.30936182e+00 8.29643548e-01 1.47274643e-01 -6.57551885e-01
3.27602476e-02 1.62296861e-01 -6.79963946e-01 1.40816376e-01
1.89901628e-02 3.45603883e-01 8.60651016e-01 -1.16098797e+00
-2.89780259e-01 -6.04313314e-01 1.84885561e-01 1.36607051e-01
-9.32319641e-01 -4.78493273e-01 -5.04032254e-01 -8.17642450e-01
2.25566193e-01 -9.29071069e-01 -3.33555490e-01 -1.42012775e-01
-4.99517262e-01 -5.39003909e-01 8.40994656e-01 -3.76379281e-01
1.34358227e+00 -1.87366581e+00 1.89559266e-01 4.44447458e-01
5.43739855e-01 4.34676290e-01 -4.20868278e-01 4.28342730e-01
-2.30764031e-01 2.36847308e-02 -2.57422417e-01 -4.21369195e-01
-3.08001757e-01 2.78526098e-01 5.78375123e-02 3.66538584e-01
2.01869667e-01 1.18184125e+00 -9.84769344e-01 -7.63278544e-01
4.19492982e-02 3.99996251e-01 -3.64928305e-01 2.40671083e-01
-1.12343602e-01 4.89834607e-01 -4.80943382e-01 6.06301129e-01
4.17877078e-01 -7.60249376e-01 4.87748474e-01 -1.16142668e-01
4.61606503e-01 1.38140470e-01 -9.53023076e-01 1.73523581e+00
-4.81544465e-01 4.23767745e-01 1.89488471e-01 -1.80993557e+00
1.30298686e+00 1.33999944e-01 5.29215634e-01 -5.04467249e-01
8.11545253e-02 4.05894257e-02 2.13109612e-01 -3.83864880e-01
2.70311445e-01 1.44396394e-01 2.60936767e-01 4.31376636e-01
3.47895503e-01 2.39629790e-01 3.94005388e-01 4.85283881e-01
1.23944461e+00 1.46364449e-02 7.39241600e-01 -1.75434202e-01
8.56532931e-01 -1.95978388e-01 4.03283209e-01 6.74406052e-01
-1.96245965e-02 6.19691052e-02 3.79162788e-01 -3.67855608e-01
-6.10811114e-01 -6.48975611e-01 2.05988407e-01 1.32152379e+00
1.18713833e-01 -5.77174485e-01 -2.29682401e-01 -1.14728618e+00
-2.27344446e-02 3.31696808e-01 -4.13614959e-01 -3.46718878e-01
-4.71543521e-01 -7.26965427e-01 3.90718549e-01 6.84201539e-01
4.30642396e-01 -1.09531903e+00 1.95817560e-01 2.67940313e-01
2.54806757e-01 -1.26782715e+00 -1.81718916e-01 5.15697658e-01
-1.24223638e+00 -1.27485716e+00 -5.55529416e-01 -1.21708167e+00
1.02103603e+00 6.21618271e-01 1.19730127e+00 5.31133890e-01
4.83707711e-02 3.26174915e-01 -4.86795783e-01 -3.63219045e-02
-4.26833332e-01 6.10073566e-01 -9.63192433e-02 1.83264315e-01
1.03872888e-01 -9.74029899e-01 -1.99761838e-01 2.75160462e-01
-8.75097573e-01 1.24854729e-01 7.00203240e-01 1.04300416e+00
4.15689826e-01 1.81413397e-01 8.94954205e-01 -1.51249087e+00
7.84134448e-01 -7.42814600e-01 -5.31284809e-01 4.39920962e-01
-9.60744023e-01 4.13527578e-01 9.92907405e-01 -2.17925936e-01
-9.01122093e-01 3.40273753e-02 -1.21337265e-01 -3.56327474e-01
-6.66271076e-02 1.08348393e+00 -2.73205012e-01 -2.81990349e-01
4.81833637e-01 1.56267658e-01 2.63496578e-01 -5.21969199e-01
3.94251585e-01 4.26040947e-01 2.67363787e-01 -4.23030376e-01
1.10947406e+00 3.32410663e-01 4.96299297e-01 -7.08996773e-01
-9.01397467e-01 -6.10899925e-01 -8.04628074e-01 -6.95823953e-02
4.51931328e-01 -8.64237130e-01 -5.34552336e-01 2.18241826e-01
-7.68314302e-01 -3.24572265e-01 -2.43822243e-02 3.91235113e-01
-1.74433947e-01 7.94906378e-01 -4.54306692e-01 -6.25521123e-01
-4.90802050e-01 -9.55246210e-01 7.08153427e-01 1.57883823e-01
2.27877557e-01 -1.45155561e+00 1.48736537e-02 3.42387080e-01
3.28440011e-01 -1.13013228e-02 1.10335493e+00 -1.22549200e+00
-4.17162001e-01 -2.07150578e-01 -4.13185269e-01 2.09300131e-01
5.11692941e-01 -3.84767979e-01 -6.22781157e-01 -5.39354980e-01
-4.33323860e-01 -4.35147345e-01 1.13606179e+00 1.03768036e-01
1.18016481e+00 -2.88114756e-01 -6.81004465e-01 6.41268194e-01
1.24825835e+00 -1.92496832e-02 3.47891152e-01 2.09323317e-01
1.13344014e+00 5.79940677e-01 4.36740607e-01 4.24171612e-02
5.78340471e-01 6.07763231e-01 6.11920416e-01 -3.70904654e-01
-1.50272340e-01 -3.70298088e-01 1.03352949e-01 1.02876425e+00
-8.89544189e-02 -4.08243835e-01 -8.69414151e-01 4.45784599e-01
-2.14195800e+00 -7.93638051e-01 -1.52258053e-01 2.01849556e+00
5.14930189e-01 2.74675876e-01 1.72483355e-01 3.33879262e-01
9.09083486e-01 4.32360947e-01 -6.06490076e-01 1.19758435e-02
5.08398600e-02 1.97967291e-01 4.03996080e-01 2.82711118e-01
-1.30394661e+00 1.07276785e+00 4.84850359e+00 8.46699238e-01
-9.98155296e-01 1.77646294e-01 4.75271553e-01 6.45860076e-01
-3.38055730e-01 6.27228692e-02 -5.74124753e-01 1.50843449e-02
7.02836156e-01 -5.74533530e-02 3.36582214e-01 1.05056655e+00
-7.77096599e-02 3.46225411e-01 -1.13109756e+00 9.50707853e-01
2.31529817e-01 -1.19473875e+00 1.85209259e-01 1.08426675e-01
6.72366381e-01 -1.43380478e-01 -1.46419123e-01 6.16925657e-01
2.81272203e-01 -9.28448021e-01 -1.37030289e-01 1.44350007e-01
5.31181991e-01 -6.85750961e-01 9.44931030e-01 4.44403082e-01
-1.60473609e+00 -1.09936602e-01 -2.27276444e-01 -1.86901733e-01
-1.36119183e-02 6.90988541e-01 -1.03096914e+00 1.00736797e+00
2.24997595e-01 1.26528120e+00 -8.66654873e-01 8.04257631e-01
-4.08049732e-01 6.96638525e-01 -1.23323321e-01 -8.85561556e-02
1.38856918e-01 -3.71154577e-01 4.73990113e-01 8.16867709e-01
7.89600536e-02 -1.55826703e-01 6.12695992e-01 4.28532422e-01
-3.61475348e-01 5.89195728e-01 -9.83282804e-01 -3.16196144e-01
3.78899097e-01 1.39377022e+00 -1.00232172e+00 -2.32521757e-01
-5.50942004e-01 6.89760387e-01 9.15202856e-01 3.01267087e-01
-4.84845996e-01 -3.13533008e-01 1.54738188e-01 1.49542749e-01
2.02252537e-01 -1.52312353e-01 4.02473658e-03 -1.21988106e+00
-3.74048483e-03 -4.85763460e-01 8.03254128e-01 -3.73068601e-01
-1.51215231e+00 7.94942617e-01 -4.41441638e-03 -1.25418520e+00
-2.41358086e-01 -5.88581026e-01 -7.06677139e-01 3.77712369e-01
-1.86271143e+00 -1.45583379e+00 -5.15602648e-01 8.65563571e-01
2.46442661e-01 -6.83545232e-01 8.08377087e-01 3.60410154e-01
-7.52613783e-01 7.16866434e-01 -2.60497369e-02 3.70405942e-01
6.68129146e-01 -1.15567136e+00 1.53593838e-01 9.09655273e-01
3.99081200e-01 5.85194468e-01 2.84329355e-01 -8.12686205e-01
-1.51883757e+00 -1.38150501e+00 5.80883980e-01 1.75412774e-01
8.24904799e-01 -2.50423014e-01 -8.95049691e-01 7.62322366e-01
7.97663182e-02 4.63634223e-01 7.00685561e-01 5.43362439e-01
-3.76956224e-01 -4.55796570e-01 -8.39041352e-01 4.38921869e-01
1.32313025e+00 -6.02995336e-01 -2.95258909e-01 3.84316891e-01
7.94991016e-01 -9.28986818e-02 -8.70463073e-01 6.40225708e-01
8.85679722e-02 -6.07659578e-01 8.91505361e-01 -7.16210246e-01
2.14158997e-01 -1.61587462e-01 2.43972048e-01 -1.44400239e+00
-3.77498209e-01 -4.44629520e-01 -6.02934062e-01 1.29965055e+00
3.34882885e-01 -9.10407305e-01 1.07058203e+00 7.48129413e-02
-1.92818791e-01 -8.44413698e-01 -7.40058184e-01 -6.91680551e-01
-3.71757030e-01 -2.20028445e-01 3.33085507e-01 1.09980786e+00
3.02900467e-02 8.92322481e-01 -4.96622860e-01 1.02182537e-01
7.52935052e-01 2.09714860e-01 9.33725238e-01 -1.89966881e+00
-2.63680428e-01 -3.17593932e-01 -6.40024662e-01 -1.14759874e+00
7.50859618e-01 -1.63150835e+00 -3.79371107e-01 -1.69889081e+00
2.42334411e-01 -6.89211369e-01 -5.53956032e-01 7.55693018e-01
-3.65226537e-01 1.31916121e-01 3.12555656e-02 5.93813919e-02
-7.51385033e-01 7.76362717e-01 1.20184469e+00 -5.83948851e-01
-3.80500674e-01 3.48736107e-01 -8.46832633e-01 4.77291465e-01
8.13427508e-01 -4.33854491e-01 -9.36847627e-01 -3.03818323e-02
2.98638940e-01 9.80773345e-02 1.88570961e-01 -8.35987747e-01
3.42249870e-01 1.72825098e-01 -4.97510098e-02 -4.74287152e-01
1.61822379e-01 -1.10546446e+00 4.28696759e-02 4.80182886e-01
-2.66645730e-01 -8.67686123e-02 -8.99011642e-02 1.01788926e+00
-2.47793302e-01 -2.79414743e-01 3.97713810e-01 -1.77778736e-01
-8.01520824e-01 6.73757195e-01 1.83003098e-01 -2.05844101e-02
8.93478334e-01 -1.55229211e-01 -3.53174806e-01 -5.04008353e-01
-7.72555470e-01 3.33573818e-01 1.45061031e-01 5.08736193e-01
5.88756382e-01 -1.30229914e+00 -6.44300342e-01 9.25465021e-03
4.02393699e-01 -4.39846739e-02 -7.79819768e-03 9.28915918e-01
-2.02541962e-01 4.21425283e-01 -9.25245434e-02 -6.63333595e-01
-1.24893236e+00 8.43529046e-01 -1.03128761e-01 -9.03089106e-01
-7.02065825e-01 4.66017514e-01 6.53315932e-02 -6.22698426e-01
4.32478160e-01 -3.41551378e-02 -6.88189685e-01 1.23613179e-02
4.78303917e-02 1.98864058e-01 1.73031673e-01 -5.69086075e-01
-2.08632320e-01 3.41470659e-01 -1.72507554e-01 6.52005911e-01
1.57219636e+00 -9.84375179e-02 -2.65406460e-01 3.31300408e-01
1.24628520e+00 -2.06547052e-01 -7.06142724e-01 -6.19592905e-01
2.43851647e-01 2.10312195e-02 -1.06954959e-03 -2.38788113e-01
-1.56699276e+00 6.93261802e-01 8.99403840e-02 4.55327868e-01
1.20885420e+00 6.70905560e-02 4.93976921e-01 7.45840549e-01
5.88375151e-01 -7.89737463e-01 1.41721919e-01 4.71936226e-01
4.91411984e-01 -1.60578322e+00 1.49761021e-01 -8.99519563e-01
-4.59508300e-01 1.17933202e+00 8.46424043e-01 -5.93583062e-02
1.00823879e+00 -5.06490245e-02 -3.72626781e-01 -2.81025976e-01
-6.62614286e-01 -5.37228107e-01 4.99522686e-01 7.43557751e-01
5.55747926e-01 9.22208745e-03 -4.16663349e-01 5.10683596e-01
2.35094875e-01 -9.52802002e-02 2.08870590e-01 7.68221855e-01
-1.45296276e-01 -1.45452452e+00 1.67314753e-01 7.97095716e-01
-4.77756336e-02 -1.94313958e-01 -3.71485054e-01 7.67664611e-01
-2.77900845e-01 1.04190636e+00 -2.92023748e-01 -4.15595531e-01
2.35438556e-03 6.07196055e-02 2.21117243e-01 -8.55962634e-01
-4.87619430e-01 -2.76886046e-01 3.01756948e-01 -2.84582883e-01
-8.21288705e-01 9.43957567e-02 -1.08336747e+00 1.14695981e-01
-6.12442374e-01 4.07482296e-01 3.31175834e-01 9.70741868e-01
3.52782100e-01 6.97553992e-01 6.92534447e-01 -6.33603156e-01
-2.76346177e-01 -1.05000126e+00 -5.48761606e-01 4.44242358e-01
5.46081997e-02 -8.90367091e-01 -2.55479753e-01 -2.76344717e-01] | [7.35932731628418, 6.279865264892578] |
76ac5c08-943b-424c-b486-ebd3a3f62d54 | uncertainty-aware-cross-lingual-transfer-with | null | null | https://aclanthology.org/2022.findings-naacl.153 | https://aclanthology.org/2022.findings-naacl.153.pdf | Uncertainty-Aware Cross-Lingual Transfer with Pseudo Partial Labels | Large-scale multilingual pre-trained language models have achieved remarkable performance in zero-shot cross-lingual tasks. A recent study has demonstrated the effectiveness of self-learning-based approach on cross-lingual transfer, where only unlabeled data of target languages are required, without any efforts to annotate gold labels for target languages. However, it suffers from noisy training due to the incorrectly pseudo-labeled samples. In this work, we propose an uncertainty-aware Cross-Lingual Transfer framework with Pseudo-Partial-Label (CLTP)1 to maximize the utilization of unlabeled data by reducing the noise introduced in the training phase. To estimate pseudo-partial-label for each unlabeled data, we propose a novel estimation method, considering both prediction confidence and the limitation to the number of similar labels. Extensive experiments are conducted on two cross-lingual tasks, including Named Entity Recognition (NER) and Natural Language Inference (NLI) across 40 languages, which shows our method can outperform the baselines on both high-resource and low-resource languages, such as 6.9 on Kazakh (kk) and 5.2 Marathi (mr) for NER. | ['Chang-Tien Lu', 'Fanglan Chen', 'Jianfeng He', 'Xuchao Zhang', 'Shuo Lei'] | null | null | null | null | findings-naacl-2022-7 | ['self-learning'] | ['natural-language-processing'] | [-2.98171878e-01 -1.04290366e-01 -5.07672429e-01 -7.79974937e-01
-1.61909902e+00 -5.00623882e-01 6.14776909e-01 -2.31064186e-01
-9.37874138e-01 1.25914729e+00 1.65072605e-01 -2.17095166e-01
5.24647534e-01 -3.66531760e-01 -8.23585093e-01 -4.02154952e-01
2.10742384e-01 7.11358130e-01 6.94584772e-02 4.62146476e-02
-2.04960182e-01 -1.14998057e-01 -1.10528827e+00 1.63629264e-01
1.35555887e+00 6.00081742e-01 2.55979359e-01 -2.01256853e-02
-4.42017943e-01 6.13879502e-01 -3.48033428e-01 -7.04091907e-01
-1.99888702e-02 -2.98215777e-01 -7.55912542e-01 -5.07788509e-02
2.36625612e-01 9.41945687e-02 1.19652562e-01 1.11994994e+00
6.18626356e-01 2.41949767e-01 8.74171257e-01 -1.06208026e+00
-9.38119054e-01 8.49297225e-01 -6.51009142e-01 -1.83442056e-01
-2.30456650e-01 -2.17653751e-01 9.78616476e-01 -1.41164148e+00
8.09148848e-01 1.30319738e+00 7.22507536e-01 7.44241059e-01
-1.00159001e+00 -9.80045974e-01 9.78610069e-02 2.93711305e-01
-1.80188894e+00 -7.49732018e-01 3.75909865e-01 -2.62633204e-01
9.43557560e-01 -3.02911490e-01 -1.05139881e-01 1.19261420e+00
-1.16341956e-01 9.70926046e-01 1.39524412e+00 -8.01300287e-01
2.53349453e-01 5.46369374e-01 9.18480977e-02 5.47578394e-01
3.64230126e-02 -6.02512136e-02 -4.84819233e-01 -6.11159801e-02
2.29399756e-01 -4.13799703e-01 -1.49849266e-01 -1.23128161e-01
-1.20723939e+00 8.58969927e-01 5.37346676e-02 5.75772583e-01
-1.18462957e-01 -2.95795351e-01 6.30848050e-01 1.33646339e-01
1.02073252e+00 1.95798144e-01 -1.07856202e+00 6.78539202e-02
-8.57016802e-01 -4.74595100e-01 6.95669591e-01 1.26013827e+00
9.73474026e-01 1.73476562e-01 -2.28097618e-01 1.38028121e+00
4.41497415e-01 7.12729573e-01 7.13036001e-01 -4.40288305e-01
6.27854228e-01 3.56644720e-01 1.09783962e-01 -1.77044034e-01
-7.09496439e-02 -2.90332198e-01 -7.37769425e-01 -3.33040982e-01
1.72520146e-01 -5.76764345e-01 -1.08381641e+00 2.03728890e+00
4.75433230e-01 3.59468222e-01 5.38650990e-01 5.93733490e-01
5.99738955e-01 9.21729445e-01 6.55158579e-01 -3.52299511e-01
1.52003431e+00 -1.20058560e+00 -9.39354002e-01 -3.27346295e-01
1.23873472e+00 -9.29321587e-01 1.21915817e+00 -2.67275944e-02
-4.21016514e-01 -5.31863809e-01 -7.46676445e-01 -2.47635350e-01
-6.17726028e-01 7.11527169e-01 3.23655516e-01 5.74535489e-01
-6.80291772e-01 2.15666443e-01 -6.87980056e-01 -4.16336805e-01
1.23969056e-01 -4.37030271e-02 -4.83951151e-01 -4.99554217e-01
-1.68401277e+00 1.08982515e+00 7.96363950e-01 -2.85831057e-02
-9.41561282e-01 -7.29332626e-01 -1.13702834e+00 -1.01890884e-01
3.93360257e-01 -7.21104592e-02 1.11680949e+00 -6.79813385e-01
-1.35301328e+00 1.00334358e+00 -3.43533337e-01 -2.26708174e-01
4.01484460e-01 -4.26405907e-01 -7.56499469e-01 -5.33534110e-01
6.14870131e-01 8.16340864e-01 2.34375492e-01 -1.23885524e+00
-7.00117409e-01 -2.61060387e-01 -5.02739429e-01 4.06627536e-01
-4.60305631e-01 1.57857031e-01 -6.82382107e-01 -5.12466550e-01
-3.32555383e-01 -9.02656794e-01 -1.07893839e-01 -5.74960649e-01
-2.13802367e-01 -8.44048619e-01 5.75216532e-01 -8.69256735e-01
1.11659443e+00 -2.00154257e+00 -3.70889783e-01 -3.57533008e-01
-6.38003707e-01 5.51578403e-01 -2.32638240e-01 2.69115806e-01
2.64898628e-01 1.09158948e-01 -3.39064211e-01 -6.01676583e-01
5.71800489e-03 3.49677980e-01 -1.72149599e-01 3.12869608e-01
3.44027430e-01 7.17846453e-01 -1.10302579e+00 -9.66718972e-01
7.25513026e-02 6.54389322e-01 -5.82211874e-02 3.05490047e-01
-1.27232388e-01 6.75191998e-01 -4.02121633e-01 7.03166962e-01
6.78054333e-01 -3.32334191e-02 2.71959960e-01 -2.99116611e-01
-2.20113069e-01 4.81362820e-01 -9.49470818e-01 1.93917060e+00
-9.62889612e-01 8.01952854e-02 -2.03683391e-01 -7.51343071e-01
1.09481001e+00 6.48264945e-01 9.04562995e-02 -6.74052417e-01
1.77194793e-02 5.76224804e-01 -2.87461221e-01 -3.16238225e-01
4.37953055e-01 -2.70640969e-01 -4.38887328e-01 5.16711235e-01
5.30981839e-01 1.36417821e-01 2.18272880e-01 -2.35847253e-02
5.25593281e-01 6.25664651e-01 4.95420098e-01 -4.61473286e-01
6.29918277e-01 1.10992365e-01 1.02937412e+00 4.07866985e-01
-5.42535841e-01 1.75288171e-01 -1.82888344e-01 -5.23034930e-02
-1.05532038e+00 -8.77344549e-01 -4.62671816e-01 1.42943418e+00
-1.38804749e-01 -1.42636180e-01 -7.38420486e-01 -1.07508576e+00
-2.80781955e-01 1.22850275e+00 -3.68296236e-01 4.74893749e-02
-3.72634381e-01 -8.59066069e-01 7.78054535e-01 3.51859868e-01
5.37153482e-01 -1.05347431e+00 2.18551278e-01 3.47866952e-01
-5.80568314e-01 -1.39107275e+00 -8.53210628e-01 3.03609401e-01
-5.27984738e-01 -6.11474752e-01 -8.07155550e-01 -1.03653407e+00
6.20073199e-01 8.00778344e-02 1.10619187e+00 -7.21470892e-01
6.84759915e-02 8.58621374e-02 -2.94211090e-01 -3.56071562e-01
-3.80057544e-01 2.56742537e-01 3.53327245e-01 6.39985055e-02
1.02063179e+00 -1.00945354e-01 1.47717595e-02 4.60866034e-01
-4.44207072e-01 -6.38809055e-02 5.58396161e-01 1.07295609e+00
9.27797258e-01 -3.56715739e-01 9.78679299e-01 -1.05027151e+00
3.13092500e-01 -6.63884103e-01 -4.23249811e-01 9.56854641e-01
-6.74066722e-01 2.67863631e-01 6.78071916e-01 -5.05036473e-01
-1.75309980e+00 3.03185415e-02 9.88452602e-03 -3.42827797e-01
-2.15954557e-01 5.65069258e-01 -3.91338527e-01 2.85251290e-01
5.62406540e-01 2.77756155e-01 -6.98561132e-01 -7.91452944e-01
5.63290477e-01 1.19514608e+00 3.60244364e-01 -8.16517651e-01
3.56368542e-01 8.71267617e-02 -6.67277932e-01 -5.69011807e-01
-1.46972024e+00 -6.04994595e-01 -8.10016513e-01 5.87699004e-03
7.98432171e-01 -1.60438204e+00 -3.88107598e-02 3.22437435e-01
-1.25432885e+00 -1.26657754e-01 2.67186146e-02 1.06123900e+00
-1.62615001e-01 2.27668241e-01 -7.59411812e-01 -9.85665321e-01
-6.03794098e-01 -9.24762189e-01 1.00030708e+00 2.02895552e-01
6.84081167e-02 -1.20585930e+00 3.59641910e-01 3.58389825e-01
2.65670776e-01 -3.15777779e-01 7.02438295e-01 -1.12182260e+00
-6.74529672e-02 1.40298367e-03 -3.01275194e-01 6.16986036e-01
2.61644542e-01 -3.99272323e-01 -1.12791646e+00 -3.21418226e-01
-1.69996753e-01 -1.02016854e+00 5.45777977e-01 7.27028847e-02
5.80480278e-01 1.38131315e-02 -3.70418578e-01 2.32628912e-01
1.48168337e+00 1.51164541e-02 1.74733236e-01 1.29299387e-01
8.34680974e-01 7.36890078e-01 1.06456041e+00 1.47033438e-01
8.41981888e-01 6.07338846e-01 -4.13493425e-01 4.72964644e-02
-1.81621343e-01 -5.17814815e-01 5.77305794e-01 1.61539352e+00
1.63204730e-01 -1.88783094e-01 -1.07606089e+00 9.36566055e-01
-1.77477539e+00 -3.88414860e-01 7.22368807e-02 2.32544184e+00
1.42380869e+00 -1.52610064e-01 -4.98130381e-01 -7.61956453e-01
1.17711687e+00 -1.68742344e-01 -5.88586211e-01 -1.21435851e-01
-2.30578691e-01 9.42442045e-02 5.36497653e-01 6.34432495e-01
-1.27039874e+00 1.60195827e+00 5.39984560e+00 1.37377822e+00
-9.20875490e-01 8.03119719e-01 7.09157705e-01 4.69863385e-01
-2.97921717e-01 3.82988863e-02 -1.45555902e+00 4.14066255e-01
1.33609509e+00 -3.16783011e-01 -9.03317891e-03 1.10933340e+00
-6.82544336e-02 1.30515411e-01 -9.17837143e-01 6.77679002e-01
2.85224229e-01 -7.07187712e-01 -2.00766344e-02 -2.45953172e-01
1.03166819e+00 5.89025199e-01 -3.48857433e-01 1.07688856e+00
7.72134662e-01 -5.39960146e-01 4.56607223e-01 2.87338376e-01
1.11998689e+00 -7.51752257e-01 8.57851982e-01 5.96667945e-01
-1.28621161e+00 5.83465576e-01 -6.69721782e-01 4.90775079e-01
3.77647936e-01 5.47697842e-01 -8.36997926e-01 6.44605160e-01
5.73325872e-01 6.08448088e-01 -2.49694303e-01 5.75387418e-01
-5.47625721e-01 8.75325501e-01 -3.30449075e-01 5.84276617e-02
3.74155253e-01 -2.22779498e-01 1.35749616e-02 1.52121162e+00
4.64429498e-01 -1.12964995e-01 3.49454284e-01 7.00583458e-01
-4.79472458e-01 7.61855006e-01 -4.83480692e-01 -1.01000115e-01
7.10998595e-01 1.30944633e+00 -3.20930779e-01 -6.69885457e-01
-6.31479144e-01 1.04656565e+00 8.76999617e-01 3.25540543e-01
-8.47683311e-01 -3.48692566e-01 2.70890236e-01 -4.46474999e-01
2.41363883e-01 -2.49319464e-01 4.23860643e-03 -1.65042794e+00
-1.31229922e-01 -5.23766518e-01 5.63460052e-01 -5.01133502e-01
-1.73926711e+00 7.85405695e-01 -1.16709195e-01 -1.24620152e+00
-3.17786694e-01 -4.01682585e-01 -2.24474013e-01 1.09794807e+00
-1.95117319e+00 -1.47722161e+00 3.28119814e-01 4.99417722e-01
7.22738326e-01 -2.30786607e-01 1.07090890e+00 7.41448522e-01
-6.23115599e-01 8.63946974e-01 4.90520120e-01 2.90990680e-01
1.47986042e+00 -1.02462316e+00 3.41385424e-01 8.64820719e-01
2.46295214e-01 4.37603712e-01 2.06641227e-01 -7.69609392e-01
-8.75976503e-01 -1.44630730e+00 1.70082474e+00 -4.29201126e-01
7.26365089e-01 -4.35366511e-01 -1.14877665e+00 9.13446546e-01
3.20293546e-01 2.77316839e-01 1.01991165e+00 4.42082852e-01
-5.78873396e-01 1.91314761e-02 -1.04159427e+00 4.57090199e-01
7.92100251e-01 -6.66618049e-01 -7.44775832e-01 4.13892537e-01
9.46450591e-01 -5.72238080e-02 -1.09619963e+00 4.50854748e-01
2.08749384e-01 -2.75032490e-01 5.64737976e-01 -6.20132625e-01
7.82340169e-02 -3.39676291e-01 -2.76128381e-01 -1.51369894e+00
-6.56058341e-02 -9.90102962e-02 2.49311835e-01 1.88179135e+00
7.18304455e-01 -4.47440267e-01 2.43613735e-01 5.76477945e-01
-2.72427648e-01 -2.92310268e-01 -1.15830815e+00 -1.10228896e+00
1.65358409e-01 -4.00803864e-01 2.05594033e-01 1.61185873e+00
-7.61911422e-02 8.98962080e-01 -7.97401309e-01 1.56796992e-01
7.62032807e-01 -1.48038387e-01 4.68584180e-01 -9.67090189e-01
-1.72176212e-02 3.43469322e-01 1.52773216e-01 -8.41937244e-01
7.25175738e-01 -1.10754931e+00 5.06989717e-01 -1.19651127e+00
3.30876231e-01 -7.90072262e-01 -6.62050307e-01 8.49082708e-01
-4.75747466e-01 1.47011206e-01 -6.88206777e-02 4.01691556e-01
-9.27221298e-01 8.70224297e-01 8.72030318e-01 1.11536130e-01
-5.98729998e-02 -3.46577764e-01 -3.63278747e-01 6.87114775e-01
5.80324829e-01 -8.66703451e-01 -1.45562306e-01 -6.98363960e-01
-2.30440229e-01 -8.03037360e-02 -5.13374031e-01 -7.08464444e-01
7.82869235e-02 -1.06301717e-01 8.87849033e-02 -6.17886662e-01
1.43780604e-01 -6.19992852e-01 -1.21009603e-01 2.73001492e-01
-3.63794178e-01 -3.13226342e-01 1.63696408e-01 5.62761366e-01
-3.98072809e-01 -3.59338999e-01 8.91809106e-01 -1.41206682e-01
-1.06181014e+00 3.49929452e-01 8.14544968e-03 5.08367419e-01
9.70285773e-01 5.39232254e-01 -1.69761434e-01 3.68882529e-02
-5.49287498e-01 4.73314255e-01 2.16229230e-01 6.28670812e-01
-6.00246675e-02 -1.70752573e+00 -1.17046273e+00 -9.35703292e-02
6.85313165e-01 -2.32678041e-01 3.81693572e-01 6.27706945e-01
9.19285044e-02 5.90425253e-01 2.76407413e-03 -5.15936911e-01
-8.51620197e-01 5.16856611e-01 4.77107018e-02 -6.64725661e-01
-8.41597244e-02 8.13779771e-01 2.52068788e-01 -1.27645004e+00
7.99828023e-02 2.43924797e-01 -2.14136496e-01 1.01017371e-01
4.34902012e-01 8.53663385e-02 2.46614637e-03 -8.65659714e-01
-4.74288285e-01 4.54990745e-01 -4.98174280e-01 -2.65558153e-01
1.05328286e+00 -5.49238145e-01 6.71322048e-02 9.39729452e-01
1.24020255e+00 -1.59194805e-02 -1.01387215e+00 -1.00140977e+00
3.94812405e-01 -1.48341581e-01 4.82933186e-02 -9.50224519e-01
-6.60597026e-01 1.01613641e+00 5.04293203e-01 -6.20695114e-01
6.04549646e-01 4.07618359e-02 9.41253185e-01 4.73515511e-01
7.73776531e-01 -1.42238033e+00 -3.57574135e-01 8.30653310e-01
3.58005494e-01 -1.70593190e+00 -3.26169401e-01 -3.26094359e-01
-1.07780969e+00 6.66128993e-01 8.93022060e-01 2.65489966e-01
6.53673768e-01 9.81746912e-02 4.06896740e-01 3.35242838e-01
-8.50244522e-01 -3.14865112e-01 3.79700005e-01 4.60029840e-01
9.77619648e-01 3.63031387e-01 -6.23304784e-01 7.10534692e-01
2.97472358e-01 -1.33852381e-02 1.89128429e-01 8.36941898e-01
-4.10303891e-01 -1.33589470e+00 -1.44893169e-01 -2.11347509e-02
-5.60926497e-01 -6.98053122e-01 6.87724277e-02 7.19043851e-01
2.89772570e-01 8.15302730e-01 -1.38694629e-01 -2.88241282e-02
6.21997789e-02 6.45981014e-01 1.81972042e-01 -8.49061370e-01
-2.55439907e-01 2.77751803e-01 3.38864684e-01 -7.50877634e-02
-6.87207401e-01 -5.27984321e-01 -1.16464019e+00 1.64479151e-01
-6.54430926e-01 6.07263863e-01 7.76943922e-01 1.14369929e+00
5.17109632e-01 1.14385180e-01 5.30654609e-01 -3.43615443e-01
-7.11626530e-01 -1.47380841e+00 -6.80704713e-01 3.55233103e-01
-3.12545002e-01 -6.99056089e-01 -4.00917798e-01 2.66073555e-01] | [10.033404350280762, 9.713398933410645] |
c36f9985-8bac-43f7-9bcb-07067781d68c | dfraud3-multi-component-fraud-detection | 2006.05718 | null | https://arxiv.org/abs/2006.05718v2 | https://arxiv.org/pdf/2006.05718v2.pdf | DFraud3- Multi-Component Fraud Detection freeof Cold-start | Fraud review detection is a hot research topic inrecent years. The Cold-start is a particularly new but significant problem referring to the failure of a detection system to recognize the authenticity of a new user. State-of-the-art solutions employ a translational knowledge graph embedding approach (TransE) to model the interaction of the components of a review system. However, these approaches suffer from the limitation of TransEin handling N-1 relations and the narrow scope of a single classification task, i.e., detecting fraudsters only. In this paper, we model a review system as a Heterogeneous InformationNetwork (HIN) which enables a unique representation to every component and performs graph inductive learning on the review data through aggregating features of nearby nodes. HIN with graph induction helps to address the camouflage issue (fraudsterswith genuine reviews) which has shown to be more severe when it is coupled with cold-start, i.e., new fraudsters with genuine first reviews. In this research, instead of focusing only on one component, detecting either fraud reviews or fraud users (fraudsters), vector representations are learnt for each component, enabling multi-component classification. In other words, we are able to detect fraud reviews, fraudsters, and fraud-targeted items, thus the name of our approach DFraud3. DFraud3 demonstrates a significant accuracy increase of 13% over the state of the art on Yelp. | ['Mohammed Bennamoun', 'Wei Liu', 'Roberto Togneri', 'Saeedreza Shehnepoor'] | 2020-06-10 | null | null | null | null | ['component-classification'] | ['natural-language-processing'] | [-2.93988198e-01 2.59150356e-01 -3.59976143e-01 1.55802563e-01
-2.36631081e-01 -3.28116864e-01 6.41116619e-01 5.09428382e-01
-1.68956071e-01 5.49661279e-01 -1.48331314e-01 -3.34113181e-01
-1.85196966e-01 -1.00421190e+00 -5.20476818e-01 -2.96085984e-01
-1.73853606e-01 5.05608261e-01 3.00779771e-02 -4.18709427e-01
8.65631998e-02 3.98956507e-01 -1.15145564e+00 3.43437731e-01
7.28622258e-01 8.60491037e-01 -5.28362811e-01 6.11821711e-01
-2.95333803e-01 7.67129958e-01 -9.67603803e-01 -1.02196681e+00
2.06192374e-01 -4.29070801e-01 -4.93260384e-01 -1.24569662e-01
1.69662029e-01 -2.86316663e-01 -4.34609354e-01 9.37143087e-01
3.79524454e-02 -3.13913286e-01 6.98711634e-01 -1.58203149e+00
-6.33701026e-01 5.01089036e-01 -9.27091420e-01 2.21053630e-01
3.93068165e-01 -4.89295781e-01 1.47997820e+00 -9.09590781e-01
7.38043964e-01 9.29444194e-01 7.89462447e-01 2.49950141e-01
-1.14741147e+00 -7.50245392e-01 3.69818330e-01 5.02158284e-01
-9.56444085e-01 -1.31770633e-02 9.88344371e-01 -4.52153713e-01
9.98753667e-01 2.10768282e-01 7.89998531e-01 1.11108172e+00
3.18407625e-01 7.87203848e-01 6.41748369e-01 -4.90784526e-01
3.41639817e-01 4.59604502e-01 7.46694744e-01 5.37932992e-01
1.15124094e+00 -9.62021723e-02 -6.32302105e-01 -4.80755508e-01
1.75385654e-01 5.43289363e-01 -1.07650064e-01 -5.05939007e-01
-5.53531587e-01 1.23003829e+00 5.22974849e-01 5.66052556e-01
-4.18880314e-01 -4.14630510e-02 6.73225403e-01 7.67081380e-01
5.22437632e-01 5.36946893e-01 -3.81880462e-01 1.59160078e-01
-8.62823665e-01 1.40739590e-01 1.22832692e+00 5.77528358e-01
7.66106308e-01 -1.68689743e-01 4.43314523e-01 5.59337318e-01
2.56780803e-01 -1.50798298e-02 5.48949480e-01 4.09880280e-02
5.87460756e-01 1.30418861e+00 -2.45281368e-01 -1.41910613e+00
-3.49695802e-01 -3.44926089e-01 -8.40121806e-01 1.45444065e-01
1.50851101e-01 -1.98206484e-01 -7.24332690e-01 1.19301486e+00
1.12723708e-01 1.62374869e-01 7.46855289e-02 7.22131550e-01
5.73273897e-01 1.97327197e-01 -1.65997624e-01 -6.82915524e-02
1.35792899e+00 -8.30331802e-01 -6.83274686e-01 -3.30513030e-01
9.34894502e-01 -4.67225164e-01 4.69939888e-01 6.19413495e-01
-6.36630714e-01 -9.73017812e-02 -1.24633873e+00 2.78471529e-01
-9.47229028e-01 -1.09590895e-01 9.06653762e-01 7.39137590e-01
-8.66615653e-01 4.93763387e-01 -4.89004254e-01 -5.15411913e-01
4.28264022e-01 3.82359356e-01 -7.00305820e-01 -3.72263461e-01
-1.28540862e+00 9.82520342e-01 -8.67035910e-02 -2.34361999e-02
-2.36561656e-01 -4.53759819e-01 -7.17215300e-01 7.54624456e-02
5.92840612e-01 -5.02502561e-01 6.97147608e-01 -1.06702280e+00
-7.68742859e-01 7.29798198e-01 5.10857068e-02 -6.80613041e-01
6.32549644e-01 -2.66484141e-01 -7.48836219e-01 2.90881544e-01
1.29201561e-01 -2.30908826e-01 9.01901007e-01 -1.28985167e+00
-4.38959092e-01 -6.17635012e-01 3.94079536e-02 -1.00056484e-01
-6.13234401e-01 2.97277328e-02 -2.23286465e-01 -6.13858819e-01
4.83892597e-02 -8.66041660e-01 4.14612666e-02 -2.40062490e-01
-5.08822262e-01 -1.63834676e-01 1.22384405e+00 -7.19693780e-01
1.53466678e+00 -2.01685596e+00 4.53279689e-02 5.96157968e-01
7.64291704e-01 6.52338684e-01 3.91426347e-02 7.53863990e-01
-3.71105105e-01 4.73576009e-01 -1.32791683e-01 -4.96890098e-01
-3.83219011e-02 3.25395644e-01 -2.06460446e-01 4.27304417e-01
4.26020056e-01 9.37749505e-01 -9.58833039e-01 -1.24360345e-01
9.15462300e-02 3.64465475e-01 -3.25459719e-01 1.18474826e-01
2.85094053e-01 -4.42592353e-01 -3.09449375e-01 9.21756268e-01
6.25213385e-01 -1.97646499e-01 4.89820540e-01 -6.04064874e-02
3.55967283e-01 1.95749961e-02 -1.13221586e+00 9.55188274e-01
-3.88188630e-01 4.86695945e-01 7.10223988e-02 -1.30823541e+00
1.05458486e+00 3.09045374e-01 5.53687453e-01 -4.29833114e-01
4.75498252e-02 4.05941457e-01 -1.04695046e-02 -3.55514824e-01
4.89944816e-01 1.29625516e-03 -1.43637925e-01 6.04783297e-01
1.21190861e-01 3.63502055e-01 2.91103750e-01 5.20130277e-01
1.73786867e+00 -4.26509708e-01 4.72484320e-01 2.61470616e-01
3.09890389e-01 -1.01689056e-01 2.39388213e-01 8.16711128e-01
-3.63879502e-01 3.60097647e-01 1.08983052e+00 -3.62632722e-01
-7.66141415e-01 -8.63206267e-01 1.77675530e-01 6.54904127e-01
7.89108500e-02 -6.47342980e-01 -2.88927555e-01 -1.38753414e+00
7.26254523e-01 3.54357868e-01 -8.42658699e-01 -5.69273651e-01
-4.54734772e-01 -1.01138854e+00 4.95027184e-01 2.15505898e-01
3.21339041e-01 -8.50888968e-01 -3.04455519e-01 2.16860682e-01
8.19423944e-02 -1.00902104e+00 -7.35200271e-02 2.72880256e-01
-7.47833252e-01 -1.69091904e+00 -4.85768616e-01 -5.23986518e-01
6.45322442e-01 2.66071498e-01 1.17932940e+00 3.69592756e-01
-5.53981543e-01 3.86786282e-01 -6.18081033e-01 -3.20542991e-01
-2.62487918e-01 -2.48018019e-02 6.54425398e-02 2.96150893e-01
8.34688365e-01 -3.99260789e-01 -4.56033885e-01 1.96050555e-01
-9.03674126e-01 -7.14193106e-01 7.22842634e-01 9.65411901e-01
9.39415395e-02 1.31232902e-01 8.76450658e-01 -1.17597139e+00
9.79052722e-01 -9.75405097e-01 -2.39569947e-01 3.77892554e-01
-1.07136190e+00 -2.37812370e-01 6.92411542e-01 -5.00335574e-01
-4.71089154e-01 -3.08326781e-01 3.34263891e-01 -4.92424816e-01
8.75613932e-03 6.42127514e-01 2.35998124e-01 -2.19944015e-01
5.52644253e-01 -2.59163231e-01 1.00506365e-01 -3.45333815e-01
1.73018798e-01 6.89795256e-01 -6.72217533e-02 8.44557583e-02
7.33197689e-01 3.79820108e-01 1.38098942e-02 -8.47412109e-01
-4.47993279e-01 -8.59532535e-01 -4.48182195e-01 -6.25264570e-02
4.55938518e-01 -7.02978849e-01 -7.66124129e-01 4.28001702e-01
-1.05618894e+00 4.48820889e-01 -2.07397833e-01 3.64905804e-01
6.44176304e-02 5.78054965e-01 -7.37135231e-01 -9.85707939e-01
-4.66092497e-01 -6.34259880e-01 6.25400841e-01 -1.25459477e-01
-3.81553203e-01 -1.12119317e+00 1.86400101e-01 3.36715162e-01
2.54436612e-01 4.09994751e-01 8.21328759e-01 -1.08810222e+00
-3.75369519e-01 -9.36907470e-01 -3.90516639e-01 5.25241733e-01
-4.32825126e-02 -9.95057449e-02 -8.22799861e-01 -2.23883882e-01
8.73405859e-02 -6.52784482e-02 1.06922674e+00 -1.16196647e-01
4.70429957e-01 -3.22783798e-01 -4.92825896e-01 1.05949916e-01
1.51841879e+00 -1.49178663e-02 5.73020816e-01 3.73749971e-01
7.48867154e-01 5.93083620e-01 4.70777810e-01 2.83098549e-01
4.19485718e-01 5.73328972e-01 4.96300906e-01 -1.73686430e-01
1.18022993e-01 -1.34461775e-01 5.03560543e-01 6.65771902e-01
1.09951496e-01 -2.71532327e-01 -9.25430954e-01 5.94777048e-01
-2.10499048e+00 -7.72968888e-01 -3.17378372e-01 2.16380477e+00
2.45108083e-01 2.28713676e-01 1.75262585e-01 6.04541004e-01
7.34302044e-01 -3.92237566e-02 -4.11618561e-01 -6.46836460e-01
3.99519578e-02 2.34448627e-01 5.01746416e-01 3.03353935e-01
-8.88438880e-01 6.13703430e-01 5.51424026e+00 2.05226839e-01
-8.52987826e-01 7.67223909e-02 3.41228902e-01 -5.35635203e-02
-1.38439640e-01 3.36199701e-02 -7.51542509e-01 5.31505167e-01
7.84122288e-01 -1.98586769e-02 2.71690458e-01 7.65462697e-01
-9.76125151e-02 -3.21003646e-01 -8.47536922e-01 8.77580464e-01
5.97030282e-01 -1.00174177e+00 9.12830904e-02 2.22918525e-01
4.38748211e-01 4.28820513e-02 -3.32328141e-01 4.67480719e-01
1.58683255e-01 -7.58162498e-01 9.21308100e-02 6.67364776e-01
3.36054116e-01 -8.52447808e-01 1.26291943e+00 9.73340273e-02
-9.60864663e-01 -3.65734249e-01 -2.40685150e-01 -5.96956424e-02
3.87284756e-02 1.03245831e+00 -8.23629498e-01 8.23640168e-01
7.71425426e-01 9.74006891e-01 -6.74493074e-01 9.33050752e-01
-2.80754954e-01 4.69745129e-01 -2.12275445e-01 -1.03664413e-01
-1.06789306e-01 -3.94353092e-01 5.18194258e-01 1.13327217e+00
2.80853570e-01 -3.24040443e-01 1.30455554e-01 6.00059986e-01
-3.09307069e-01 1.26847267e-01 -1.00330842e+00 -3.40560198e-01
5.23911715e-02 1.37943602e+00 -5.34714401e-01 -1.88481629e-01
-5.97236216e-01 1.07767594e+00 5.15562952e-01 3.62277091e-01
-4.15412098e-01 -8.25046718e-01 5.90243816e-01 2.30073884e-01
4.67215538e-01 1.31859079e-01 -3.30360115e-01 -1.48022425e+00
3.46250594e-01 -7.59993553e-01 5.08033693e-01 -3.03212374e-01
-1.52214146e+00 4.42571491e-01 -4.09958333e-01 -1.00894511e+00
-2.75989860e-01 -7.31804371e-01 -8.19857001e-01 7.13126659e-01
-1.77901340e+00 -1.25886667e+00 -1.25127241e-01 5.69997847e-01
-9.53722522e-02 -3.72294277e-01 9.40329731e-01 4.57287490e-01
-7.29340196e-01 6.75731778e-01 -7.41312876e-02 1.90268308e-01
7.64020681e-01 -1.38246596e+00 2.96880275e-01 8.56788218e-01
1.30576596e-01 6.71330214e-01 4.31338280e-01 -1.00234675e+00
-1.23624170e+00 -7.83425033e-01 1.49842501e+00 -4.50033665e-01
1.15822935e+00 -6.07391536e-01 -1.07641637e+00 4.37209457e-01
-2.32846379e-01 1.87601745e-01 7.42486477e-01 5.05487084e-01
-6.40414298e-01 -1.62852645e-01 -1.18065107e+00 3.54820997e-01
8.00414264e-01 -7.05175340e-01 -5.92914343e-01 2.59660661e-01
3.33418339e-01 2.45140463e-01 -7.74929583e-01 2.75990188e-01
4.42462981e-01 -1.18895555e+00 7.44288146e-01 -5.55226982e-01
4.21533644e-01 -1.53532654e-01 2.62849659e-01 -1.20745254e+00
-3.11487298e-02 -3.82040918e-01 -9.30552781e-01 9.87819433e-01
3.33007783e-01 -8.94184709e-01 8.91512692e-01 3.16055983e-01
2.24701703e-01 -8.64504933e-01 -7.92722821e-01 -6.92706227e-01
-2.26979017e-01 -1.88709915e-01 3.79442513e-01 1.25667763e+00
4.92730558e-01 4.00900781e-01 -4.27734017e-01 1.32068736e-03
3.28142494e-01 5.21765556e-03 7.63523817e-01 -1.56895256e+00
-4.42664564e-01 -4.21989977e-01 -7.92998672e-01 -2.23551676e-01
6.43619969e-02 -8.91248167e-01 -5.11995852e-01 -1.55950367e+00
-2.94927992e-02 -2.80846655e-01 -4.70133483e-01 4.43013966e-01
-1.28538728e-01 2.89241463e-01 2.17426807e-01 3.01796764e-01
-4.71389502e-01 4.03948687e-02 6.47293389e-01 -1.71623185e-01
-2.50692993e-01 2.62745589e-01 -8.97044837e-01 5.44862807e-01
8.28663766e-01 -5.52280545e-01 -9.93258730e-02 9.26409513e-02
6.32689416e-01 -1.51069731e-01 5.69478095e-01 -6.60141289e-01
1.13418050e-01 5.41795254e-01 2.46128812e-01 -4.41846937e-01
2.02680796e-01 -9.43209350e-01 -6.96492270e-02 5.19816458e-01
1.70981094e-01 2.68872678e-01 -2.95475032e-02 1.02704680e+00
-4.13505644e-01 -4.17051286e-01 1.79895297e-01 -7.36552775e-02
-4.06032085e-01 9.32256356e-02 -1.86775342e-01 -2.37781480e-01
1.12383366e+00 -1.02555178e-01 -5.57603419e-01 -3.62176776e-01
-7.22887874e-01 1.18984364e-01 3.44382942e-01 4.97544944e-01
6.97790921e-01 -9.02184427e-01 -4.71279353e-01 3.50811839e-01
4.89492536e-01 -4.74297374e-01 -6.67897016e-02 7.86153734e-01
-1.39883682e-01 1.20225213e-01 1.49816096e-01 -1.53866425e-01
-1.37097478e+00 5.61467469e-01 2.70715177e-01 -8.56576800e-01
-4.34688032e-01 5.97728133e-01 -3.83732051e-01 -4.04415935e-01
1.56337783e-01 -6.67850822e-02 -5.05527556e-01 5.98000884e-01
2.72246778e-01 5.54869056e-01 4.57140893e-01 -4.09105539e-01
-4.90334213e-01 3.39055747e-01 -5.89243531e-01 4.36959386e-01
1.19106162e+00 2.16181248e-01 -3.15192878e-01 4.11409736e-01
1.26513338e+00 1.57597825e-01 -4.54774290e-01 -1.92666486e-01
3.07860434e-01 -3.16300154e-01 -1.47481272e-02 -8.80482495e-01
-1.08991778e+00 6.21921837e-01 3.95582825e-01 7.64586449e-01
7.18923926e-01 -1.29499108e-01 6.69311106e-01 5.15885174e-01
2.42768005e-01 -1.16997004e+00 5.83444498e-02 4.51676458e-01
6.74532533e-01 -1.48089433e+00 2.48580098e-01 -3.78810406e-01
-5.97477317e-01 1.04135704e+00 3.70261461e-01 -4.51213866e-01
9.70735192e-01 -2.47141514e-02 1.02577440e-01 -5.39758861e-01
-5.79161704e-01 -1.00677624e-01 1.43837810e-01 5.11830866e-01
1.92076966e-01 1.54060781e-01 -5.01796424e-01 6.78931177e-01
2.46595871e-02 -4.70754057e-02 6.87592208e-01 9.60022509e-01
3.99953453e-03 -1.29443085e+00 -1.60209090e-01 1.02267575e+00
-5.18912613e-01 -1.04012646e-01 -8.64494383e-01 1.06212842e+00
9.05050337e-02 1.14199519e+00 -8.94865543e-02 -6.32902026e-01
5.71681321e-01 1.31594285e-01 -1.97958350e-01 -7.32632577e-01
-1.01761544e+00 -3.26546669e-01 1.47130027e-01 -4.10705328e-01
-2.37566203e-01 -5.23492813e-01 -8.48159373e-01 -3.82800668e-01
-8.22788119e-01 2.17991605e-01 7.32335925e-01 8.39870274e-01
5.11234403e-01 5.70458412e-01 6.51317477e-01 -3.20619375e-01
-4.82097775e-01 -6.55608535e-01 -9.31040287e-01 6.98484361e-01
4.31848675e-01 -5.83164752e-01 -7.61079311e-01 -6.60827219e-01] | [6.888189315795898, 5.875773906707764] |
c7226876-cd0e-4dab-a1a0-6aecc142f172 | seeing-the-pose-in-the-pixels-learning-pose | 2306.09331 | null | https://arxiv.org/abs/2306.09331v1 | https://arxiv.org/pdf/2306.09331v1.pdf | Seeing the Pose in the Pixels: Learning Pose-Aware Representations in Vision Transformers | Human perception of surroundings is often guided by the various poses present within the environment. Many computer vision tasks, such as human action recognition and robot imitation learning, rely on pose-based entities like human skeletons or robotic arms. However, conventional Vision Transformer (ViT) models uniformly process all patches, neglecting valuable pose priors in input videos. We argue that incorporating poses into RGB data is advantageous for learning fine-grained and viewpoint-agnostic representations. Consequently, we introduce two strategies for learning pose-aware representations in ViTs. The first method, called Pose-aware Attention Block (PAAB), is a plug-and-play ViT block that performs localized attention on pose regions within videos. The second method, dubbed Pose-Aware Auxiliary Task (PAAT), presents an auxiliary pose prediction task optimized jointly with the primary ViT task. Although their functionalities differ, both methods succeed in learning pose-aware representations, enhancing performance in multiple diverse downstream tasks. Our experiments, conducted across seven datasets, reveal the efficacy of both pose-aware methods on three video analysis tasks, with PAAT holding a slight edge over PAAB. Both PAAT and PAAB surpass their respective backbone Transformers by up to 9.8% in real-world action recognition and 21.8% in multi-view robotic video alignment. Code is available at https://github.com/dominickrei/PoseAwareVT. | ['Srijan Das', 'Aman Chadha', 'Dominick Reilly'] | 2023-06-15 | null | null | null | null | ['pose-prediction', 'action-recognition-in-videos', 'video-alignment', 'action-recognition', 'imitation-learning'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'methodology'] | [ 2.50462055e-01 9.02287290e-02 -1.68446422e-01 -2.68446237e-01
-7.69437492e-01 -6.05570734e-01 8.36806774e-01 -3.65990996e-01
-3.67072582e-01 4.27222878e-01 3.44244242e-01 4.75945659e-02
-5.17407581e-02 -3.53649706e-01 -1.10746324e+00 -7.54688501e-01
2.59390414e-01 4.25675184e-01 3.63916427e-01 -2.54970223e-01
1.91471711e-01 5.65357685e-01 -1.39085698e+00 2.53465474e-01
4.43580091e-01 1.03203487e+00 5.29541433e-01 5.29391110e-01
2.37286583e-01 9.56301451e-01 -2.78561711e-01 -4.74892676e-01
4.87356573e-01 -5.31181321e-02 -7.50947595e-01 2.33350635e-01
3.99555027e-01 -4.12211925e-01 -5.72269917e-01 6.77410603e-01
4.45644408e-01 1.60996214e-01 5.69675922e-01 -1.31416810e+00
-2.46126875e-01 4.18428421e-01 -8.44416976e-01 1.37344331e-01
4.07122493e-01 4.44711924e-01 1.03986609e+00 -9.81021941e-01
5.99362612e-01 1.28007936e+00 6.52344644e-01 4.01921839e-01
-9.80226815e-01 -4.04919535e-01 4.61566836e-01 4.48228240e-01
-1.26768363e+00 -4.95339125e-01 9.47198510e-01 -5.20436585e-01
1.12930071e+00 2.70054210e-02 6.65040255e-01 1.36621702e+00
8.91424939e-02 1.15029299e+00 8.48589659e-01 -8.04475099e-02
-6.24395814e-03 -3.67921948e-01 -2.48921394e-01 6.72808230e-01
1.45700604e-01 -1.38959244e-01 -7.19378054e-01 1.70271069e-01
9.88252103e-01 1.76538944e-01 -5.10133326e-01 -1.06913280e+00
-1.55602992e+00 5.61306000e-01 5.45644701e-01 -8.01711082e-02
-6.68329775e-01 5.31461537e-01 3.68022144e-01 -4.49890941e-02
2.10684225e-01 4.65684295e-01 -6.75403833e-01 -2.79028952e-01
-4.30623859e-01 6.52215630e-02 3.28352630e-01 1.06932652e+00
8.35135818e-01 2.54825354e-02 -1.65809870e-01 7.05140471e-01
3.99574548e-01 4.97613728e-01 4.41902310e-01 -1.12487853e+00
5.77289224e-01 5.36359131e-01 1.01792723e-01 -9.24182117e-01
-5.32874048e-01 -4.13877726e-01 -3.69018137e-01 -8.26522056e-03
5.09302437e-01 3.17127146e-02 -1.07138395e+00 1.81727886e+00
4.89006311e-01 3.21333520e-02 -1.86237976e-01 9.76063907e-01
6.79239154e-01 2.94614583e-01 6.59724697e-02 1.09350480e-01
1.53016591e+00 -1.25695002e+00 -2.80926138e-01 -5.95801532e-01
4.17825013e-01 -6.84687197e-01 9.91338670e-01 4.54720646e-01
-9.06838596e-01 -5.15978098e-01 -9.07342851e-01 -3.04619521e-01
-1.06941365e-01 4.37095314e-01 6.60428286e-01 2.72632897e-01
-9.15894628e-01 3.03682178e-01 -1.02087295e+00 -5.48110008e-01
3.64102185e-01 4.14287746e-01 -6.61893547e-01 -1.13133714e-01
-7.25585103e-01 8.77094328e-01 4.22116183e-02 1.22263677e-01
-1.17193270e+00 -4.76287156e-01 -1.06958783e+00 -2.16677785e-01
7.07850575e-01 -8.20548594e-01 1.37879086e+00 -9.33674276e-01
-1.56830525e+00 8.32937837e-01 -1.57426834e-01 -3.45405757e-01
6.63380921e-01 -7.10616410e-01 1.53521806e-01 4.37116086e-01
1.93997577e-01 9.10355687e-01 1.22672355e+00 -1.31686044e+00
-4.40270185e-01 -4.91656482e-01 3.45126718e-01 4.32982028e-01
-1.14800103e-01 -2.69500464e-01 -9.97572482e-01 -6.10523224e-01
2.90218443e-01 -1.05092180e+00 -1.18018791e-01 2.82379925e-01
-1.95595622e-01 -1.63803518e-01 8.30353498e-01 -6.86595440e-01
6.00872457e-01 -1.92557502e+00 7.13082492e-01 -1.39766380e-01
1.97108671e-01 1.40024841e-01 -2.40239143e-01 3.69616568e-01
2.02155896e-02 -4.51477468e-01 5.80506735e-02 -5.47775924e-01
6.78867623e-02 4.55246031e-01 -1.53600886e-01 6.40490055e-01
4.12147373e-01 1.15844643e+00 -9.32375968e-01 -3.32889616e-01
5.14563680e-01 5.84745109e-01 -6.19476140e-01 1.22582547e-01
-2.19859391e-01 7.44335711e-01 -5.36507249e-01 8.92738700e-01
3.21144879e-01 -3.43533903e-01 1.36718959e-01 -6.12049460e-01
1.13555133e-01 2.31494620e-01 -8.48763108e-01 2.02978325e+00
-4.17043000e-01 4.90947843e-01 2.39558250e-01 -1.08157277e+00
6.35057569e-01 2.08940670e-01 7.40041256e-01 -6.47196472e-01
2.13213176e-01 3.60307544e-02 -2.32076600e-01 -5.54098487e-01
5.15046597e-01 1.91791564e-01 -1.10857643e-01 2.08732992e-01
2.69229412e-01 -7.92684257e-02 3.07390615e-02 8.68094638e-02
1.14412963e+00 7.61520505e-01 4.34568018e-01 1.07353278e-01
4.76600379e-01 -9.70540345e-02 6.90278232e-01 4.63605583e-01
-5.17969012e-01 8.39968264e-01 4.42517787e-01 -3.32068384e-01
-8.96405578e-01 -1.10354435e+00 1.93764791e-01 1.24587297e+00
3.88612747e-01 -4.60270256e-01 -5.24307251e-01 -6.99958622e-01
6.04260853e-03 2.60528713e-01 -6.23831928e-01 -1.07943691e-01
-7.04934359e-01 -3.87265474e-01 3.34254354e-01 9.04682636e-01
5.71901858e-01 -9.65310156e-01 -9.43681359e-01 -5.34714311e-02
-4.86037582e-01 -1.36532998e+00 -3.48449200e-01 1.44625217e-01
-6.28039360e-01 -1.22664964e+00 -8.85880053e-01 -3.36430728e-01
5.12246192e-01 6.25756681e-01 9.32392120e-01 -3.39405984e-01
-1.70401528e-01 9.78342175e-01 -5.91498196e-01 -2.59274304e-01
1.25252694e-01 6.88700080e-02 1.97889939e-01 2.13659346e-01
2.85286933e-01 -5.84781170e-01 -8.09817612e-01 5.16121030e-01
-5.10376096e-01 6.88199922e-02 9.14862692e-01 7.91720390e-01
6.79012656e-01 -6.55307174e-01 2.25632951e-01 -3.19000721e-01
-1.19492762e-01 -3.86056095e-01 -4.00721490e-01 6.29038289e-02
-8.73709098e-02 3.17860804e-02 2.94629753e-01 -4.08501267e-01
-8.02402496e-01 5.79925418e-01 -6.41078204e-02 -9.75150287e-01
-1.36818841e-01 2.24314570e-01 -5.25941193e-01 -6.34771809e-02
4.81372446e-01 3.05208653e-01 1.32767543e-01 -4.26650435e-01
3.35242838e-01 3.24105352e-01 6.92411482e-01 -6.69688463e-01
7.28595257e-01 5.84975481e-01 -1.17760628e-01 -6.92312837e-01
-7.87611544e-01 -6.30771101e-01 -7.72866666e-01 -3.51838171e-01
1.02372062e+00 -1.19052291e+00 -6.54944181e-01 6.29424214e-01
-1.03273404e+00 -5.39294720e-01 -2.01465696e-01 6.18384480e-01
-1.01853561e+00 4.34660614e-01 -3.01913232e-01 -6.25139475e-01
-3.59085500e-02 -1.39541221e+00 1.65532959e+00 -6.90090954e-02
-2.43164167e-01 -6.28682554e-01 -1.79102927e-01 8.20114374e-01
-8.12960602e-03 2.17690527e-01 4.41480547e-01 -4.05297816e-01
-8.01451981e-01 -8.80911574e-02 -1.73347279e-01 1.61157891e-01
1.30235955e-01 -2.71823406e-01 -1.11466932e+00 -3.79146934e-01
-2.57614404e-01 -5.33759892e-01 8.68732452e-01 3.63207906e-01
1.14484501e+00 -3.84003366e-03 -3.21919948e-01 7.54471064e-01
1.08402359e+00 8.64710659e-02 6.42954230e-01 4.33911920e-01
1.10614729e+00 5.25868058e-01 8.50961804e-01 4.83144313e-01
5.26049733e-01 1.11381888e+00 8.97215247e-01 1.47653461e-01
-2.30960324e-01 -2.96887219e-01 7.33537316e-01 4.82701570e-01
-4.27426696e-01 -1.42431229e-01 -8.99070978e-01 4.99738812e-01
-1.99835420e+00 -8.85077238e-01 1.27359271e-01 2.01221514e+00
5.58103263e-01 -1.86842531e-02 2.67909020e-01 3.81397679e-02
4.07575846e-01 3.32975388e-01 -8.30396056e-01 6.08763993e-02
1.27561167e-01 -2.87389755e-02 4.95551437e-01 1.84332520e-01
-1.33438516e+00 9.95506942e-01 4.96567869e+00 5.72732329e-01
-1.12246931e+00 1.43882170e-01 2.43361056e-01 -2.15597302e-01
8.14666599e-03 -1.10219784e-01 -6.46242797e-01 1.61681548e-01
4.46586132e-01 2.96350777e-01 2.43756995e-01 9.98587549e-01
1.93534687e-01 -4.35957909e-02 -1.23904312e+00 1.10522282e+00
1.57817319e-01 -1.00008464e+00 9.16099846e-02 1.29361600e-01
4.35129821e-01 2.73769259e-01 6.90962523e-02 3.70137751e-01
1.01148814e-01 -8.00257802e-01 1.02536535e+00 3.77448350e-01
6.91556811e-01 -5.24130046e-01 5.88375628e-01 1.15664408e-01
-1.35946012e+00 -2.36914724e-01 -1.80023372e-01 -3.75427939e-02
1.88099727e-01 5.60724586e-02 -5.73641062e-01 6.56163454e-01
8.65633190e-01 1.13320351e+00 -5.79788148e-01 7.94259250e-01
-3.75847697e-01 3.21094543e-01 -2.21437514e-01 3.49802583e-01
3.61529499e-01 4.97505665e-02 6.36296332e-01 7.96376824e-01
1.87390089e-01 3.69546115e-02 1.18044123e-01 5.15957654e-01
7.29238018e-02 -2.11258277e-01 -4.50739145e-01 1.25352051e-02
3.34802628e-01 1.26009846e+00 -6.87224209e-01 -5.74149154e-02
-4.95855659e-01 1.18274057e+00 3.18893313e-01 3.57557535e-01
-1.10479105e+00 2.78909970e-02 1.01043570e+00 8.13041180e-02
7.91674793e-01 -4.45502609e-01 -5.43034775e-03 -1.18041408e+00
2.26572186e-01 -1.07580400e+00 1.71204820e-01 -1.00420082e+00
-9.12581027e-01 3.23390335e-01 1.50143296e-01 -1.53892386e+00
-3.22936863e-01 -1.00502133e+00 -3.18752050e-01 4.54517245e-01
-1.29595006e+00 -1.57459319e+00 -6.25609279e-01 8.42112124e-01
9.23384666e-01 -1.43166147e-02 5.02403259e-01 1.54248238e-01
-5.85915685e-01 5.32172143e-01 -1.69540584e-01 1.28219098e-01
7.19999254e-01 -1.09402323e+00 2.64692843e-01 7.62880683e-01
2.98077613e-01 5.90171456e-01 5.76975763e-01 -4.33329165e-01
-1.79183900e+00 -1.09889686e+00 3.44662398e-01 -7.51828730e-01
5.57013273e-01 -2.92194724e-01 -5.72226346e-01 9.07787561e-01
-1.93461664e-02 1.80644542e-01 2.04212695e-01 -1.53576043e-02
-5.85466504e-01 -3.79169062e-02 -7.54763186e-01 6.02917373e-01
1.31936693e+00 -5.34869850e-01 -5.37942588e-01 3.51439893e-01
6.45348907e-01 -5.61585844e-01 -8.09618354e-01 5.92018306e-01
6.86350465e-01 -9.92362380e-01 1.33804989e+00 -4.05456543e-01
5.51250935e-01 -4.41704571e-01 -4.78369653e-01 -9.86944675e-01
-3.56035292e-01 -4.88389850e-01 -3.52782369e-01 8.91290724e-01
4.69220206e-02 -6.12928629e-01 7.73605168e-01 3.08356702e-01
-2.87667930e-01 -8.87898862e-01 -8.61541331e-01 -7.72837281e-01
-1.75246462e-01 -4.85894293e-01 4.14177477e-01 7.37127662e-01
-2.54559785e-01 3.75669032e-01 -4.75277930e-01 2.33067930e-01
4.01228517e-01 9.41130444e-02 1.16987407e+00 -9.15521443e-01
-6.04849637e-01 -4.04338270e-01 -6.99048162e-01 -1.42729700e+00
6.86770380e-02 -5.44048607e-01 1.20583609e-01 -1.58214891e+00
1.25625476e-01 -5.24038561e-02 -2.01536879e-01 7.49299288e-01
-1.02919191e-01 4.99385864e-01 3.22186798e-01 2.99253494e-01
-6.69662714e-01 8.04547310e-01 1.19630289e+00 -2.59011328e-01
6.44616038e-02 -2.14067567e-02 -3.82231355e-01 9.88345742e-01
7.46212363e-01 -6.83419257e-02 -6.96922719e-01 -5.64804971e-01
7.71926995e-03 -1.15057185e-01 8.18730116e-01 -1.05951333e+00
7.08930045e-02 -1.39142707e-01 4.27331597e-01 -6.63973987e-01
9.36893761e-01 -7.71852672e-01 1.39768034e-01 4.61390644e-01
-7.92522579e-02 7.00427145e-02 5.25363013e-02 8.10981274e-01
-7.38670304e-02 7.60934129e-02 5.34507632e-01 -1.97019592e-01
-1.27463889e+00 3.34570497e-01 -1.87778637e-01 4.45994101e-02
1.04806352e+00 -4.73693997e-01 -3.02487433e-01 -5.78430593e-01
-6.49126112e-01 1.64531544e-01 4.96857524e-01 6.82915866e-01
6.28447831e-01 -1.03559780e+00 -4.30521339e-01 1.75442442e-01
3.92133445e-01 1.83678642e-01 2.54478186e-01 1.22788310e+00
-4.12994027e-01 4.76423264e-01 -3.17029834e-01 -9.57736611e-01
-1.28025687e+00 5.23581028e-01 3.47691774e-01 -1.05316781e-01
-6.45760179e-01 1.10270691e+00 6.45099819e-01 -4.18320864e-01
2.11241022e-01 -2.59264410e-01 -1.44970238e-01 -3.51185985e-02
1.90901145e-01 3.40403736e-01 -8.60997736e-02 -1.02613163e+00
-5.42302370e-01 8.04195940e-01 -1.04051672e-01 6.08759793e-03
1.33041835e+00 -2.07058474e-01 1.42766759e-01 2.99711347e-01
1.02696216e+00 -1.60869524e-01 -1.88356173e+00 -2.38471404e-01
-2.57376164e-01 -5.04608274e-01 -1.08929656e-01 -5.75431526e-01
-1.22846031e+00 9.08468187e-01 3.32335174e-01 -3.84660482e-01
1.17045200e+00 1.56676188e-01 6.05107605e-01 4.88707453e-01
8.76959622e-01 -8.96748006e-01 5.65816402e-01 4.78988647e-01
1.19102836e+00 -1.30389357e+00 1.18675336e-01 -3.95093560e-01
-9.21714723e-01 7.66615629e-01 9.06298399e-01 3.61011922e-02
2.29912803e-01 -8.04917365e-02 3.56312469e-02 -1.93403944e-01
-6.07616186e-01 -4.41689551e-01 3.67855281e-01 6.65190816e-01
1.32564545e-01 -9.60382819e-02 5.61957918e-02 4.78830338e-01
-9.89569128e-02 -2.46265754e-01 2.34026313e-01 1.04445779e+00
-2.58719236e-01 -8.70176673e-01 -3.70588005e-01 1.44282892e-01
-1.81085974e-01 1.55842915e-01 -5.03620088e-01 9.21645999e-01
1.32824332e-01 8.19508731e-01 -4.76886109e-02 -6.52030885e-01
2.98464656e-01 -1.00954965e-01 7.79015899e-01 -5.00335038e-01
-3.02239478e-01 1.28829405e-01 9.88052860e-02 -1.21535408e+00
-6.47427201e-01 -9.25162494e-01 -1.11684704e+00 7.18551800e-02
-9.53591093e-02 -3.01633745e-01 4.70942885e-01 9.75502372e-01
3.84180695e-01 5.85523486e-01 3.27571899e-01 -1.50159287e+00
-5.94876230e-01 -8.66632402e-01 -2.89495051e-01 2.94937819e-01
3.21208239e-01 -1.27246785e+00 -1.37738615e-01 1.09678097e-01] | [7.905465126037598, 0.16718408465385437] |
1ae76daa-8278-49cb-853f-c7ce4115ad20 | from-chatgpt-to-threatgpt-impact-of | 2307.00691 | null | https://arxiv.org/abs/2307.00691v1 | https://arxiv.org/pdf/2307.00691v1.pdf | From ChatGPT to ThreatGPT: Impact of Generative AI in Cybersecurity and Privacy | Undoubtedly, the evolution of Generative AI (GenAI) models has been the highlight of digital transformation in the year 2022. As the different GenAI models like ChatGPT and Google Bard continue to foster their complexity and capability, it's critical to understand its consequences from a cybersecurity perspective. Several instances recently have demonstrated the use of GenAI tools in both the defensive and offensive side of cybersecurity, and focusing on the social, ethical and privacy implications this technology possesses. This research paper highlights the limitations, challenges, potential risks, and opportunities of GenAI in the domain of cybersecurity and privacy. The work presents the vulnerabilities of ChatGPT, which can be exploited by malicious users to exfiltrate malicious information bypassing the ethical constraints on the model. This paper demonstrates successful example attacks like Jailbreaks, reverse psychology, and prompt injection attacks on the ChatGPT. The paper also investigates how cyber offenders can use the GenAI tools in developing cyber attacks, and explore the scenarios where ChatGPT can be used by adversaries to create social engineering attacks, phishing attacks, automated hacking, attack payload generation, malware creation, and polymorphic malware. This paper then examines defense techniques and uses GenAI tools to improve security measures, including cyber defense automation, reporting, threat intelligence, secure code generation and detection, attack identification, developing ethical guidelines, incidence response plans, and malware detection. We will also discuss the social, legal, and ethical implications of ChatGPT. In conclusion, the paper highlights open challenges and future directions to make this GenAI secure, safe, trustworthy, and ethical as the community understands its cybersecurity impacts. | ['Lopamudra Praharaj', 'Eli Parker', 'Kshitiz Aryal', 'CharanKumar Akiri', 'Maanak Gupta'] | 2023-07-03 | null | null | null | null | ['code-generation', 'malware-detection'] | ['computer-code', 'miscellaneous'] | [ 3.89909409e-02 2.68197298e-01 1.65862422e-02 2.08821163e-01
-8.83343294e-02 -1.41166270e+00 7.23586440e-01 -2.55291294e-02
-1.84165016e-02 5.55433393e-01 1.04537919e-01 -1.05028236e+00
-2.74201095e-01 -8.97646546e-01 -3.40411603e-01 -3.85702670e-01
-1.28262147e-01 1.12310208e-01 -3.90884668e-01 -3.68585169e-01
7.72733569e-01 5.00968874e-01 -9.37170863e-01 -5.54655213e-03
8.61631513e-01 5.04191041e-01 -7.46230662e-01 8.05811882e-01
2.19509080e-01 6.62413776e-01 -1.23454165e+00 -1.20151818e+00
5.23612559e-01 -3.66645664e-01 -5.32369733e-01 -7.59274483e-01
2.69532371e-02 -6.27780020e-01 -5.74830337e-04 1.30226362e+00
4.31324363e-01 -4.26370054e-01 1.83088019e-01 -1.98386061e+00
-9.36617911e-01 6.31233633e-01 -1.44485384e-01 1.86570443e-03
6.80515885e-01 6.20115399e-01 4.21447545e-01 -2.47494169e-02
8.18352520e-01 1.26901507e+00 7.93992698e-01 9.28585231e-01
-9.24587965e-01 -1.26604509e+00 -1.40709281e-01 -3.10061276e-01
-8.99309576e-01 -5.85326217e-02 5.00635743e-01 -5.35495698e-01
1.07298017e+00 4.82127964e-01 1.08045805e+00 1.88931811e+00
8.38664830e-01 1.21448070e-01 9.89675581e-01 -2.70059496e-01
3.83661538e-01 5.90898871e-01 3.78823757e-01 3.29802901e-01
9.82345998e-01 6.00585878e-01 -1.44785032e-01 -9.92814720e-01
3.72115254e-01 -2.33991176e-01 5.63418306e-02 3.41380328e-01
-4.42914128e-01 1.13292301e+00 2.09157001e-02 4.32777137e-01
-2.56317317e-01 1.25213817e-01 6.56421840e-01 8.59519094e-02
3.27107996e-01 1.00364685e+00 -1.02531210e-01 -7.78955162e-01
-1.44753039e-01 2.78138816e-01 1.21815336e+00 6.13204002e-01
1.86932981e-02 2.58334398e-01 4.49720770e-01 8.17079842e-02
6.40143454e-01 6.32312059e-01 -1.95858888e-02 -6.07345343e-01
3.21017921e-01 6.40908241e-01 -3.27893049e-01 -1.39957440e+00
5.28141074e-02 -2.88965613e-01 9.52459350e-02 4.54455078e-01
1.29981712e-01 -6.86444759e-01 -7.60509491e-01 1.37006032e+00
1.88583195e-01 -9.91826132e-03 1.30048050e-02 2.95978099e-01
3.30257356e-01 7.38163948e-01 4.68008697e-01 1.59582347e-01
1.13343966e+00 4.83509675e-02 -5.42443573e-01 -3.07276219e-01
8.68223131e-01 -3.86091262e-01 6.94429159e-01 5.96599042e-01
-8.27984989e-01 3.62279207e-01 -9.95160997e-01 4.85183686e-01
-7.87469923e-01 -8.98930013e-01 6.94740295e-01 2.22163486e+00
-7.91770041e-01 2.75894642e-01 -5.88366568e-01 -4.45924968e-01
4.92349505e-01 3.27180296e-01 -1.32720366e-01 6.32892624e-02
-1.37589669e+00 1.07156324e+00 3.82798791e-01 -1.70457348e-01
-8.50661337e-01 -7.40539670e-01 -5.64008772e-01 4.42460924e-02
2.97581345e-01 -3.45039874e-01 5.74737191e-01 -6.56385660e-01
-1.15709579e+00 4.84935790e-01 7.73221910e-01 -6.08936787e-01
2.57935911e-01 -2.41959244e-01 -5.88931859e-01 -6.42873421e-02
1.28519475e-01 1.20683491e-01 6.35707140e-01 -1.07899976e+00
-1.47207500e-02 -3.63345027e-01 3.16883475e-01 -2.85160989e-01
-5.49335659e-01 7.86274731e-01 7.67114878e-01 -3.11518520e-01
-6.72081709e-01 -1.19028604e+00 -1.78266317e-02 -7.33894587e-01
-7.20978320e-01 2.05851734e-01 1.30521321e+00 -8.83150041e-01
1.21484339e+00 -2.11873007e+00 -3.78688097e-01 5.47630250e-01
2.30859578e-01 9.27612841e-01 2.19778702e-01 1.08460569e+00
3.84390354e-02 1.28372657e+00 2.23523512e-01 3.11214805e-01
1.88301459e-01 -1.39510781e-01 -6.27205133e-01 8.68748650e-02
-4.29687761e-02 9.22599554e-01 -7.92394876e-01 1.71405092e-01
-1.17713377e-01 6.32370889e-01 -6.44211709e-01 -5.96505105e-02
1.25230372e-01 3.68951529e-01 -5.17926633e-01 9.59045589e-01
6.89709961e-01 3.79720032e-01 2.88286090e-01 4.71637249e-01
-1.86777890e-01 4.32027102e-01 -4.37284768e-01 3.96743238e-01
-1.32190272e-01 6.18787706e-01 2.12712571e-01 -2.74374187e-01
9.02961373e-01 2.27915704e-01 -2.73568556e-02 -4.76323724e-01
6.85142100e-01 3.33446622e-01 5.48657596e-01 -4.87517506e-01
3.22397441e-01 3.53531390e-02 -3.42449218e-01 1.13186657e+00
-1.49940684e-01 -1.81711867e-01 -2.07817346e-01 3.36135566e-01
1.32527840e+00 -1.26781523e-01 1.52611107e-01 -8.02635849e-02
9.36643928e-02 1.60957336e-01 3.71968687e-01 6.29559875e-01
-3.88734907e-01 -3.27595800e-01 8.55141044e-01 -1.90769374e-01
-9.12655950e-01 -9.60421920e-01 3.08586717e-01 7.00694859e-01
2.58347206e-02 -6.55294895e-01 -1.27982819e+00 -1.14058852e+00
-3.01713999e-02 1.44261122e+00 -4.14618075e-01 -8.65907192e-01
-4.97813374e-01 -7.07834005e-01 1.30829704e+00 1.32597551e-01
6.77527785e-01 -8.21746826e-01 -6.33351982e-01 4.33932170e-02
2.02531099e-01 -1.00726497e+00 -3.88110392e-02 -3.58920753e-01
-2.73684025e-01 -1.12374377e+00 1.05655149e-01 -2.30146013e-02
5.07812083e-01 5.25923148e-02 2.63797402e-01 3.32419544e-01
-5.41075587e-01 6.99878097e-01 -5.31585097e-01 -1.03261948e+00
-1.03780615e+00 -9.95873213e-02 1.28875002e-01 -3.39968204e-01
6.18622959e-01 -5.19653082e-01 -1.00961141e-01 1.41728207e-01
-9.32192087e-01 -5.35039842e-01 1.55920357e-01 3.78838390e-01
-6.24990821e-01 3.34219426e-01 7.41180182e-01 -9.69899535e-01
1.32696664e+00 -9.16584790e-01 -2.15305284e-01 1.50985911e-01
-7.04644740e-01 -6.05508387e-01 5.73152602e-01 -6.39422596e-01
-1.00833118e+00 -8.63693178e-01 -2.05546156e-01 1.28501542e-02
-2.37790048e-01 4.99529511e-01 -4.58742052e-01 -6.15513265e-01
1.04673111e+00 -1.20558076e-01 3.17938298e-01 -1.78335737e-02
5.91478609e-02 6.28164768e-01 -3.49475235e-01 -5.24959147e-01
1.29705012e+00 1.47626474e-01 -2.64372706e-01 -8.36114943e-01
-2.48150289e-01 3.78836155e-01 1.08309291e-01 -4.71504241e-01
9.17130053e-01 -1.82409570e-01 -1.10088539e+00 6.32885396e-01
-1.07623971e+00 -1.60985038e-01 1.82959661e-01 2.26389706e-01
1.17094748e-01 3.45484436e-01 -6.55817270e-01 -1.28259826e+00
-6.52275681e-01 -9.88091946e-01 2.26738021e-01 4.06669259e-01
-6.61411822e-01 -1.15698206e+00 2.07098112e-01 9.07137334e-01
7.67589748e-01 7.31304407e-01 1.13902354e+00 -1.34268534e+00
-5.21346867e-01 -7.91338742e-01 7.85821024e-03 5.65018117e-01
-5.54794818e-02 4.67850268e-01 -5.59017062e-01 -2.07009867e-01
4.10026848e-01 -1.89116344e-01 -3.01686972e-01 -1.75588623e-01
1.85690075e-01 -9.56368446e-01 -3.78928423e-01 3.33005369e-01
1.12331545e+00 1.28339314e+00 9.13359165e-01 6.23000860e-01
5.93702972e-01 8.78004372e-01 5.31022489e-01 1.67584777e-01
-9.16979369e-03 -1.68133136e-02 6.62665188e-01 5.77182531e-01
5.94635189e-01 -4.69453245e-01 8.03751588e-01 1.29912689e-01
-1.92933261e-01 -4.19139653e-01 -1.23840868e+00 -9.58307758e-02
-1.21639967e+00 -1.14759088e+00 -5.07897399e-02 2.01368570e+00
4.32737246e-02 4.72552866e-01 2.55964011e-01 6.99922964e-02
8.37045014e-01 -1.46377727e-01 -3.02848190e-01 -1.32738614e+00
3.86518508e-01 3.14700693e-01 5.95439494e-01 3.55856240e-01
-6.45659983e-01 9.86246526e-01 6.63907433e+00 4.19767827e-01
-9.23090398e-01 2.01791272e-01 6.18517339e-01 1.04697630e-01
-3.87715101e-01 4.67089057e-01 -9.04210091e-01 6.44243300e-01
1.45512831e+00 -6.16971910e-01 6.67948008e-01 9.48873460e-01
-4.23115902e-02 1.04726791e-01 -4.78757858e-01 3.20852935e-01
4.99369115e-01 -1.19261432e+00 -1.23122014e-01 6.99034572e-01
2.43345410e-01 -4.97824371e-01 4.48761880e-01 2.63515830e-01
6.18170798e-01 -1.02334809e+00 5.42877793e-01 -3.02166436e-02
3.54055047e-01 -1.21395397e+00 6.80931926e-01 3.07212174e-01
-3.77599508e-01 -6.66950047e-01 2.08417669e-01 -2.34875575e-01
3.05088520e-01 -7.41024390e-02 -1.23725033e+00 2.85546780e-01
4.64561433e-01 1.03156619e-01 -4.03224319e-01 5.28758049e-01
-4.11195189e-01 9.15108621e-01 -1.24493800e-01 -4.45800334e-01
3.08627009e-01 -4.18327779e-01 9.84126270e-01 8.14212859e-01
3.43978167e-01 1.88647449e-01 -3.90870869e-01 1.18223095e+00
5.82593501e-01 -3.76180947e-01 -1.20060551e+00 -1.12773323e+00
6.16173029e-01 1.18202114e+00 -7.84116089e-01 2.06735969e-01
7.90345445e-02 4.86732662e-01 -5.32551944e-01 2.03752115e-01
-9.61320579e-01 -7.69852221e-01 9.96808589e-01 3.94487590e-01
-6.05251074e-01 -3.50394309e-01 -5.80418706e-01 -7.95688868e-01
-3.08050632e-01 -1.32266283e+00 7.15513676e-02 -4.46397394e-01
-1.12848747e+00 6.37404799e-01 3.33534032e-01 -8.22551489e-01
-4.64946002e-01 -5.97797215e-01 -1.12013257e+00 6.04600489e-01
-2.35904559e-01 -1.53826988e+00 2.65313476e-01 1.41900003e-01
-1.81613490e-01 -5.49825013e-01 6.07264698e-01 -7.77870864e-02
-5.99199533e-01 6.02849483e-01 -5.17819941e-01 2.53071994e-01
1.31141752e-01 -4.69146103e-01 9.79052544e-01 1.16652179e+00
-4.93025571e-01 1.38101482e+00 6.48341775e-01 -1.51815224e+00
-1.57878029e+00 -4.94300932e-01 5.52710652e-01 -8.11254025e-01
1.03197265e+00 -1.00147653e+00 -6.23336554e-01 8.68503630e-01
1.60458580e-01 -1.01139820e+00 1.16577101e+00 -1.67137031e-02
-5.00190496e-01 4.60454196e-01 -1.72300100e+00 1.03149331e+00
1.01883769e+00 -7.14031398e-01 -1.57315612e-01 4.00599331e-01
9.74029303e-01 1.47922710e-01 -5.27321637e-01 -2.70000398e-01
4.77385759e-01 -9.48334992e-01 8.04320395e-01 -7.91716933e-01
1.31741986e-01 4.04491782e-01 3.07472080e-01 -9.80272889e-01
-2.12447941e-02 -1.29389024e+00 1.93512276e-01 1.64520621e+00
2.92810678e-01 -1.44062948e+00 6.37452245e-01 1.36275887e+00
3.04818992e-02 -4.14256603e-01 -8.54073286e-01 -9.46237266e-01
2.65350670e-01 -5.42102039e-01 6.98658943e-01 1.28024304e+00
5.78581333e-01 -1.54709518e-01 -4.05136377e-01 1.88796327e-01
6.98202193e-01 -9.19235826e-01 8.08318675e-01 -1.09301972e+00
-2.43655011e-01 -2.51107782e-01 -7.28494108e-01 3.75231057e-01
1.06602898e-02 -7.61777222e-01 -7.69524336e-01 -9.88217473e-01
-1.61702588e-01 -2.86364794e-01 4.51708645e-01 4.29202288e-01
2.02431574e-01 -1.33278951e-01 7.94028521e-01 -1.68451115e-01
3.45522851e-01 -9.34322178e-02 7.70953059e-01 1.82600558e-01
-4.36361969e-01 -1.62419900e-01 -1.25221646e+00 6.19055808e-01
1.08844531e+00 -7.50842094e-01 -7.21806526e-01 1.83901101e-01
7.67044783e-01 -1.19984336e-01 5.54677010e-01 -8.17810476e-01
2.41266787e-02 -5.26275277e-01 6.69684783e-02 2.43241847e-01
2.72104919e-01 -7.96613276e-01 4.46153611e-01 8.48449647e-01
9.17357877e-02 2.74741471e-01 3.83328646e-01 2.28544921e-01
5.17499864e-01 -4.04584855e-01 4.64129150e-01 -4.88942722e-03
1.81327060e-01 7.98077062e-02 -1.00443733e+00 -4.33739573e-02
1.76122975e+00 -5.29909074e-01 -1.15316200e+00 -5.59951603e-01
-4.10825342e-01 -5.23546785e-02 6.35872662e-01 6.03844404e-01
7.28742123e-01 -7.83992052e-01 -3.79855186e-01 4.20200437e-01
-5.21868706e-01 -1.06793404e+00 2.35450208e-01 3.73253614e-01
-7.81636596e-01 5.31215109e-02 -8.12948108e-01 1.57142267e-01
-1.54949117e+00 7.46610045e-01 2.25388020e-01 4.33023609e-02
-3.44604433e-01 7.34239697e-01 -7.03757629e-02 -2.66727746e-01
-1.41528413e-01 4.93489951e-01 -5.93479611e-02 -1.49890453e-01
7.33609259e-01 6.96870983e-01 -4.80796278e-01 -4.98943776e-01
-5.16161501e-01 1.20380586e-02 -4.85168278e-01 -4.33155507e-01
8.99453461e-01 2.56845623e-01 -4.77072418e-01 -1.50686905e-01
8.28813493e-01 3.10072809e-01 -5.12157381e-01 8.70291233e-01
2.97499895e-02 -8.49197209e-01 -5.67973077e-01 -1.23570240e+00
-6.42492235e-01 8.10505271e-01 1.24927118e-01 9.89798605e-01
7.21842229e-01 -4.04909492e-01 9.44263518e-01 4.64927703e-02
6.09572291e-01 -8.85909855e-01 6.14456460e-02 4.03925568e-01
8.27157736e-01 -3.82627785e-01 -1.88515767e-01 -6.20072484e-01
-9.84978259e-01 1.02931201e+00 7.78488517e-01 -3.04290980e-01
6.92516387e-01 5.23352623e-01 -4.84019779e-02 -3.28475654e-01
-4.71654087e-01 6.94654524e-01 -3.76508325e-01 1.24106610e+00
-3.24843861e-02 1.44071028e-01 -4.20551300e-01 5.76052129e-01
-4.70377058e-01 -3.70777577e-01 1.04498661e+00 1.17029107e+00
-1.16274171e-01 -1.56675828e+00 -7.58241713e-01 2.68884957e-01
-8.00950170e-01 -1.39112234e-01 -1.51887619e+00 1.02589703e+00
2.00181901e-01 1.20077729e+00 -4.58537340e-01 -1.15101147e+00
-1.23165332e-01 2.62233078e-01 -1.81419000e-01 -5.43580890e-01
-1.53464663e+00 -3.89939427e-01 6.27223969e-01 -3.04314435e-01
4.34348524e-01 -7.32165873e-01 -9.74312067e-01 -1.01021695e+00
-3.58504713e-01 2.42945388e-01 1.06710815e+00 6.35287523e-01
6.63692057e-01 8.72307345e-02 4.69093412e-01 -4.10338372e-01
-3.65571022e-01 -6.67976618e-01 -2.25320548e-01 -3.11301649e-02
-2.41971210e-01 -4.68199700e-01 -7.09618688e-01 -6.54367507e-01] | [6.20848274230957, 7.782863616943359] |
fdb86f85-93c6-46d2-8abd-fd2310eb3d71 | cross-linked-variational-autoencoders-for | null | null | https://openreview.net/forum?id=BkghJoRNO4 | https://openreview.net/pdf?id=BkghJoRNO4 | Cross-Linked Variational Autoencoders for Generalized Zero-Shot Learning | Most approaches in generalized zero-shot learning rely on cross-modal mapping between an image feature space and a class embedding space or on generating artificial image features. However, learning a shared cross-modal embedding by aligning the latent spaces of modality-specific autoencoders is shown to be promising in (generalized) zero-shot learning. While following the same direction, we also take artificial feature generation one step further and propose a model where a shared latent space of image features and class embeddings is learned by aligned variational autoencoders, for the purpose of generating latent features to train a softmax classifier. We evaluate our learned latent features on conventional benchmark datasets and establish a new state of the art on generalized zero-shot as well as on few-shot learning. Moreover, our results on ImageNet with various zero-shot splits show that our latent features generalize well in large-scale settings. | ['Zeynep Akata', 'Trevor Darrell', 'Samarth Sinha', 'Sayna Ebrahimi', 'Edgar Schönfeld'] | 2019-03-24 | null | null | null | iclr-workshop-lld-2019 | ['generalized-zero-shot-learning', 'generalized-zero-shot-learning'] | ['computer-vision', 'methodology'] | [ 2.01065570e-01 2.50340879e-01 -5.63949823e-01 -4.42252100e-01
-9.43336308e-01 -1.36877477e-01 1.11455452e+00 -2.64024287e-01
-2.91623205e-01 4.62665975e-01 6.15846336e-01 3.92443001e-01
-1.00111291e-01 -8.80332112e-01 -8.84436011e-01 -8.67261350e-01
1.39618859e-01 2.56217539e-01 1.04075456e-02 -1.65597573e-01
-2.95064747e-02 -1.89617813e-01 -1.79264879e+00 3.71062547e-01
2.53662795e-01 7.99210131e-01 1.82201937e-02 5.56132913e-01
-8.31713751e-02 8.24601471e-01 -8.88340268e-03 -2.63333499e-01
2.87796497e-01 -4.67593312e-01 -1.01925576e+00 5.59751153e-01
6.03805602e-01 -3.06618303e-01 -6.33931458e-01 8.93754900e-01
2.06653684e-01 4.23830837e-01 9.96909022e-01 -1.68543398e+00
-1.27385283e+00 6.81911111e-01 -1.32970244e-01 -2.42247730e-01
2.39431597e-02 3.41672421e-01 1.50643909e+00 -1.01578605e+00
1.06864619e+00 1.15993249e+00 5.58180988e-01 6.79221153e-01
-1.73591137e+00 -3.12809825e-01 -1.28284290e-01 3.38616788e-01
-1.21266162e+00 -4.14987653e-01 8.19008052e-01 -8.71296704e-01
1.07321298e+00 -3.45628142e-01 4.91543412e-01 1.60232234e+00
2.09736928e-01 8.11583817e-01 8.57767403e-01 -5.84499121e-01
3.71485293e-01 3.64872247e-01 1.92600697e-01 8.59021008e-01
-2.88722575e-01 4.69506942e-02 -6.92225993e-01 -2.71353543e-01
6.02703750e-01 6.03490293e-01 -1.13239244e-01 -1.12770760e+00
-1.35627556e+00 1.67173767e+00 5.28443038e-01 4.52696174e-01
-3.69832188e-01 2.83506572e-01 4.22335058e-01 4.18177247e-01
5.06103754e-01 4.54390854e-01 -3.95411849e-01 2.20764369e-01
-8.93646777e-01 2.73132902e-02 5.49244344e-01 9.02429819e-01
1.23263836e+00 1.52508199e-01 -4.75981385e-01 9.27991211e-01
3.20073783e-01 2.96999484e-01 9.01051641e-01 -1.12386143e+00
1.05782643e-01 3.86991292e-01 -1.85829967e-01 -3.78689319e-01
-3.78176644e-02 7.88572128e-04 -6.45655572e-01 3.95039499e-01
4.93334867e-02 -3.55607793e-02 -1.19714987e+00 2.16023111e+00
-1.69665813e-01 5.93427658e-01 4.33861792e-01 5.96004009e-01
6.06843710e-01 8.24744582e-01 2.58402109e-01 3.08866017e-02
1.32968819e+00 -1.14251912e+00 -5.06998360e-01 -1.31247774e-01
6.65920079e-01 -2.49956176e-01 1.21818233e+00 -1.61848158e-01
-9.15168107e-01 -7.45981693e-01 -1.21412826e+00 -2.56285131e-01
-7.69333243e-01 -1.56739458e-01 4.18694764e-01 2.06815571e-01
-9.74119067e-01 6.34882510e-01 -7.34006763e-01 -5.55420697e-01
3.59114408e-01 1.60227567e-02 -7.78827667e-01 -1.44867286e-01
-1.21270704e+00 7.93640971e-01 4.30399150e-01 -7.94220328e-01
-1.27923453e+00 -8.62791777e-01 -1.51450551e+00 3.55576843e-01
2.07763582e-01 -1.11460066e+00 9.16562915e-01 -8.61887455e-01
-1.50560951e+00 9.80827153e-01 7.42412955e-02 -5.16730487e-01
-3.16086449e-02 7.42505416e-02 -2.49498516e-01 2.03143433e-01
2.41862997e-01 1.14248443e+00 1.28003645e+00 -9.99793649e-01
-3.40850502e-01 -3.46154898e-01 1.56813666e-01 -2.95201782e-02
-7.73466647e-01 -4.94495928e-01 1.03716426e-01 -4.45929468e-01
-2.74333566e-01 -1.02004373e+00 -1.22283943e-01 -2.41279863e-02
-2.46716857e-01 -3.60799998e-01 8.10633183e-01 -1.71969071e-01
7.48532951e-01 -2.15049648e+00 6.32264435e-01 -9.07423049e-02
3.07683975e-01 -3.10731709e-01 -5.77336609e-01 4.59890157e-01
-3.18747640e-01 2.77160807e-03 -3.07693869e-01 -6.24111354e-01
4.84602362e-01 2.87940800e-01 -5.29510081e-01 3.62890631e-01
3.69642049e-01 1.14974606e+00 -8.73745739e-01 -3.95093083e-01
5.18789947e-01 6.90399110e-01 -7.64333248e-01 1.82240635e-01
-8.08784515e-02 -8.56111944e-02 -1.33317366e-01 2.91819334e-01
6.14493676e-02 -6.95834041e-01 -7.16600418e-02 -1.99864492e-01
2.02479705e-01 -1.63341776e-01 -7.85902560e-01 2.11651516e+00
-6.25055075e-01 7.03868032e-01 -5.44741929e-01 -1.21242070e+00
4.76149619e-01 7.24933207e-01 5.52182317e-01 -2.56667495e-01
1.09757349e-01 -2.43159950e-01 -3.76768082e-01 -5.71060538e-01
3.50634634e-01 -5.55058777e-01 -2.67275184e-01 5.74840188e-01
1.35504580e+00 3.00021749e-02 4.88248765e-02 4.19744015e-01
9.06319439e-01 1.99160010e-01 2.87618548e-01 -1.39669552e-01
8.77361372e-02 -2.20819935e-01 1.65444791e-01 7.33270943e-01
-1.00445904e-01 7.95608580e-01 3.67125332e-01 -2.32863009e-01
-1.42415893e+00 -1.38750029e+00 -3.00145894e-01 1.30872917e+00
-1.31221160e-01 -5.81803739e-01 -5.48877358e-01 -7.23609567e-01
-5.12905009e-02 9.50286031e-01 -1.13992429e+00 -4.11994606e-01
2.80284613e-01 -3.85561287e-01 2.08278507e-01 7.06369460e-01
-9.95866135e-02 -1.10211897e+00 -4.73273218e-01 1.51803687e-01
2.30767988e-02 -1.16438699e+00 -3.27826411e-01 3.79701108e-01
-5.97668767e-01 -9.55795825e-01 -8.75560701e-01 -7.59482324e-01
4.66882825e-01 2.97063559e-01 1.10607517e+00 -4.45503443e-01
-6.31847382e-01 9.10808802e-01 -3.72831255e-01 -7.75424093e-02
-1.39604554e-01 7.16595575e-02 2.12054744e-01 2.20086083e-01
4.62463707e-01 -7.15032101e-01 -3.72605592e-01 -7.47479945e-02
-1.14409053e+00 4.04607877e-02 3.49898875e-01 1.25249493e+00
3.62811923e-01 -5.18398225e-01 5.21783412e-01 -6.90551758e-01
3.01630437e-01 -8.57735872e-01 -3.04798186e-01 3.19913954e-01
-4.39537257e-01 5.12632430e-01 3.85614812e-01 -4.52864021e-01
-9.81160223e-01 9.91549995e-03 1.69878662e-01 -1.03579426e+00
-3.50810468e-01 2.65406638e-01 1.03347458e-01 3.45435262e-01
7.20990181e-01 2.94026643e-01 1.22475050e-01 -2.10146040e-01
1.17588568e+00 3.66906136e-01 1.90186441e-01 -3.21064413e-01
9.85741675e-01 6.53688073e-01 -2.05066293e-01 -9.11886692e-01
-8.95854712e-01 -5.96670449e-01 -8.09762239e-01 2.35052675e-01
1.47430050e+00 -1.14050782e+00 -3.33070070e-01 1.24193765e-01
-8.23566377e-01 -3.36541861e-01 -7.62678206e-01 6.70489013e-01
-1.10874820e+00 1.35480493e-01 -7.14516222e-01 -2.80748636e-01
2.45580040e-02 -1.31211662e+00 1.22974157e+00 -1.11800097e-02
-2.89644271e-01 -1.31726635e+00 5.51091492e-01 2.63620406e-01
3.19428295e-01 3.51346768e-02 1.03353322e+00 -6.11139596e-01
-5.29656351e-01 -9.82009470e-02 -4.58711833e-02 3.62188965e-01
1.04250208e-01 -1.19687483e-01 -1.20814908e+00 -3.24874252e-01
-2.09660664e-01 -1.05336654e+00 1.22027910e+00 4.97019500e-01
8.29563558e-01 -4.40967306e-02 -2.00454757e-01 6.14194036e-01
1.63430750e+00 -4.32464331e-01 4.18121338e-01 2.44561881e-02
5.56373119e-01 4.86177862e-01 1.34791583e-01 5.23686349e-01
3.19042027e-01 6.45699143e-01 2.92275399e-01 1.44672990e-01
-7.09462836e-02 -3.54927897e-01 5.19254804e-01 7.38959730e-01
1.38273180e-01 7.14795068e-02 -4.78652149e-01 8.87009501e-01
-1.95535147e+00 -1.31503463e+00 5.19282699e-01 1.90856421e+00
6.97623372e-01 -1.45026430e-01 1.13478482e-01 -2.87560761e-01
5.41419446e-01 4.65390623e-01 -4.21836019e-01 -1.32612035e-01
-1.37872070e-01 3.38354886e-01 9.42965075e-02 6.15384936e-01
-1.24991369e+00 1.05230987e+00 6.51958132e+00 6.25644028e-01
-1.12741315e+00 6.35143042e-01 3.18492115e-01 -3.80164832e-01
-4.98229861e-01 1.96626410e-01 -5.49320161e-01 2.53622234e-01
8.65946889e-01 -2.00661927e-01 2.94660062e-01 1.05913055e+00
-4.93487179e-01 3.71164352e-01 -1.38948500e+00 1.02349436e+00
4.09539372e-01 -1.65516424e+00 4.87144999e-02 2.13380173e-01
1.11366200e+00 3.49165887e-01 2.12205932e-01 7.71553695e-01
6.00680292e-01 -9.75705147e-01 4.22124952e-01 7.07203746e-01
7.15110600e-01 -3.41754138e-01 3.79986882e-01 1.03219755e-01
-9.61298883e-01 -7.42574558e-02 -5.97985327e-01 3.88955958e-02
1.68161824e-01 1.36260614e-02 -5.31025589e-01 2.76002765e-01
5.12925148e-01 8.78071606e-01 -5.54641962e-01 5.86149454e-01
-1.22155257e-01 5.29764533e-01 -3.30597423e-02 4.63134348e-01
5.56212842e-01 6.59841821e-02 3.27321708e-01 8.67153823e-01
3.00447196e-01 -2.12411672e-01 3.06434155e-01 1.18262684e+00
-2.30543435e-01 6.53226376e-02 -1.01622677e+00 -2.29408517e-01
-3.14682908e-02 1.35409522e+00 -3.33398402e-01 -6.26303852e-01
-8.38635266e-01 1.18043709e+00 3.40537071e-01 6.67578280e-01
-7.69403338e-01 -3.06915790e-01 7.24925399e-01 -1.96323439e-01
5.64173877e-01 2.51271669e-03 1.44821271e-01 -1.78146398e+00
-3.69886369e-01 -2.31521949e-01 4.41805601e-01 -9.46437418e-01
-1.59469807e+00 5.19190848e-01 1.56541049e-01 -1.37622070e+00
-8.03428292e-01 -6.10161722e-01 -7.02480376e-01 6.15237653e-01
-1.30744243e+00 -1.52557409e+00 -1.33455470e-01 8.96546781e-01
8.58789980e-01 -5.35719812e-01 1.26457369e+00 -1.63128510e-01
-1.96615458e-01 5.07443845e-01 2.73612648e-01 8.17355812e-02
7.12426484e-01 -1.25634801e+00 7.38578290e-02 4.19896275e-01
7.04492450e-01 5.94106376e-01 5.87115288e-01 -2.23809004e-01
-1.20837092e+00 -9.35237944e-01 6.29328132e-01 -5.52695036e-01
9.80693877e-01 -4.66633677e-01 -7.89389491e-01 1.19904792e+00
7.81597376e-01 5.05962312e-01 1.05988121e+00 4.01346684e-01
-8.17850769e-01 2.91650057e-01 -7.69403696e-01 3.91861469e-01
8.79527986e-01 -1.11002481e+00 -9.46027100e-01 2.86582202e-01
9.32257235e-01 4.00387526e-01 -1.15363657e+00 1.35851860e-01
5.93634665e-01 -6.58370972e-01 1.05884862e+00 -1.18091273e+00
8.81015182e-01 7.08838180e-02 -6.51907325e-01 -1.56701899e+00
-6.82546854e-01 -9.57660154e-02 -2.40118667e-01 9.89769995e-01
3.45088005e-01 -3.71439487e-01 7.54430115e-01 4.59138900e-01
7.91110918e-02 -6.05174422e-01 -6.34429097e-01 -7.27214634e-01
1.49870574e-01 -2.60178387e-01 2.11841032e-01 1.24165058e+00
2.50486404e-01 7.80630171e-01 -6.52671754e-01 -2.24290043e-01
9.08105135e-01 1.75750569e-01 7.10754693e-01 -1.15796638e+00
-6.01933837e-01 -3.48467648e-01 -7.00943232e-01 -5.74611425e-01
7.42991626e-01 -1.21461821e+00 -1.39700726e-01 -1.17738390e+00
7.00902164e-01 2.80670762e-01 -7.16630399e-01 6.96193337e-01
2.02116538e-02 5.20631254e-01 4.31718200e-01 7.65709579e-02
-8.64814103e-01 1.09951258e+00 6.91482604e-01 -4.12562102e-01
1.17064722e-01 -4.18352544e-01 -4.47718561e-01 6.56119764e-01
3.97599667e-01 -4.96057987e-01 -7.29745984e-01 -2.07581162e-01
-3.62228677e-02 4.39064130e-02 6.52094781e-01 -8.02324653e-01
1.35315388e-01 -2.08658904e-01 4.17843550e-01 -2.17337370e-01
7.63724923e-01 -6.80989027e-01 -2.14219585e-01 3.15022916e-01
-7.20328271e-01 -4.97537524e-01 -3.28470200e-01 7.39612162e-01
-4.02023941e-01 -3.47260237e-01 7.75736511e-01 -1.82871878e-01
-1.00294507e+00 6.01302862e-01 -2.46509552e-01 3.81147526e-02
1.03992534e+00 -2.75050789e-01 -2.18478218e-01 -5.25512397e-01
-1.39425695e+00 1.76077217e-01 5.74650049e-01 8.18861365e-01
6.21634603e-01 -1.79595780e+00 -5.30400157e-01 3.65992486e-01
8.54635119e-01 -7.73738682e-01 6.18164420e-01 6.46475255e-01
2.42495850e-01 4.38450783e-01 -6.46417856e-01 -7.39436924e-01
-9.20435488e-01 8.93785954e-01 8.89993533e-02 -1.37441993e-01
-6.59153163e-01 7.17613399e-01 6.05235934e-01 -5.63742936e-01
4.43311706e-02 8.17135721e-02 -2.03924887e-02 2.55391151e-01
4.34946686e-01 4.70526218e-02 -4.51900065e-01 -8.47238004e-01
1.24922274e-02 5.13539076e-01 6.67261630e-02 -3.77833247e-01
1.71471286e+00 -8.58353078e-02 2.63696283e-01 9.78156984e-01
1.87329710e+00 -5.55283785e-01 -1.48884690e+00 -6.64090753e-01
-1.85550228e-01 -2.48226881e-01 5.40786646e-02 -1.64782614e-01
-9.03655171e-01 1.19177520e+00 6.02358818e-01 1.47635162e-01
6.05856717e-01 4.75331634e-01 6.47198558e-01 5.23043096e-01
1.47854567e-01 -1.04516327e+00 6.03275359e-01 3.42219383e-01
7.00338602e-01 -1.64956069e+00 -3.67821366e-01 1.53287455e-01
-8.04001868e-01 1.00799346e+00 4.89755094e-01 -5.04116476e-01
9.59863603e-01 -1.72310382e-01 -1.25742912e-01 -3.69507730e-01
-1.07315838e+00 -4.66380388e-01 3.50462139e-01 4.84055251e-01
3.11091393e-01 8.46590474e-02 1.61461323e-01 5.05063355e-01
8.29599574e-02 7.78584406e-02 4.03622061e-01 6.89448535e-01
-4.16560322e-01 -1.10302269e+00 3.24912965e-02 3.79384458e-01
-2.70608634e-01 -1.02941707e-01 6.31374121e-02 7.19404459e-01
2.50696130e-02 5.30451655e-01 2.55155772e-01 -2.87963867e-01
-2.33050406e-01 7.09613144e-01 7.15999603e-01 -9.24202323e-01
1.12419851e-01 -1.15036834e-02 -4.03028697e-01 -6.24287486e-01
-5.27808666e-01 -4.72555041e-01 -8.12450767e-01 1.74537495e-01
-1.73717365e-01 -4.58986498e-02 5.58454096e-01 1.12129986e+00
3.09053689e-01 4.37538564e-01 6.37459517e-01 -1.05919480e+00
-8.14154029e-01 -1.10603178e+00 -7.89061666e-01 1.16789842e+00
3.33938867e-01 -9.00299251e-01 -3.73231262e-01 3.99213582e-01] | [10.122066497802734, 2.3482532501220703] |
6ad43eae-cca4-4490-b5c3-576152233eb7 | classification-of-luminal-subtypes-in-full | 2301.09282 | null | https://arxiv.org/abs/2301.09282v1 | https://arxiv.org/pdf/2301.09282v1.pdf | Classification of Luminal Subtypes in Full Mammogram Images Using Transfer Learning | Automatic identification of patients with luminal and non-luminal subtypes during a routine mammography screening can support clinicians in streamlining breast cancer therapy planning. Recent machine learning techniques have shown promising results in molecular subtype classification in mammography; however, they are highly dependent on pixel-level annotations, handcrafted, and radiomic features. In this work, we provide initial insights into the luminal subtype classification in full mammogram images trained using only image-level labels. Transfer learning is applied from a breast abnormality classification task, to finetune a ResNet-18-based luminal versus non-luminal subtype classification task. We present and compare our results on the publicly available CMMD dataset and show that our approach significantly outperforms the baseline classifier by achieving a mean AUC score of 0.6688 and a mean F1 score of 0.6693 on the test dataset. The improvement over baseline is statistically significant, with a p-value of p<0.0001. | ['Andreas Maier', 'Prathmesh Madhu', 'Adarsh Bhandary Panambur'] | 2023-01-23 | null | null | null | null | ['breast-cancer-detection', 'breast-cancer-detection'] | ['knowledge-base', 'medical'] | [ 7.83175886e-01 6.23222113e-01 -1.01514423e+00 -8.30269575e-01
-1.42698038e+00 -2.95856804e-01 5.62963963e-01 6.47158265e-01
-4.24434304e-01 7.30877340e-01 2.72077739e-01 -9.14875627e-01
-2.14194655e-01 -7.54298389e-01 -6.90123737e-01 -8.79118800e-01
-9.36924666e-02 5.99700034e-01 1.73074976e-01 1.32558629e-01
5.18321581e-02 3.78178835e-01 -9.79912996e-01 1.09421062e+00
4.40526754e-01 1.33219016e+00 6.11721985e-02 9.92117167e-01
1.32877335e-01 1.06748724e+00 -8.37808624e-02 -3.57245445e-01
-1.19374953e-01 -3.45596641e-01 -9.33013380e-01 -1.87113345e-01
7.54617453e-01 -5.15009612e-02 -2.36488044e-01 9.45815563e-01
5.07124424e-01 -6.27148569e-01 9.74382699e-01 -3.52549523e-01
7.21787885e-02 6.17900968e-01 -6.62935734e-01 4.50934887e-01
-2.39528239e-01 -1.03330217e-01 9.70200658e-01 -4.66562808e-01
7.58913815e-01 5.75192153e-01 1.22884607e+00 4.78758574e-01
-1.29809773e+00 -6.13504708e-01 -5.27199566e-01 1.53867498e-01
-1.29778051e+00 -4.63747948e-01 1.69398397e-01 -5.75233996e-01
7.26062417e-01 7.38225937e-01 4.26063985e-01 7.45428681e-01
5.83235741e-01 5.96814334e-01 1.22525108e+00 -4.89670634e-01
1.85070202e-01 2.47968227e-01 1.05424486e-01 7.59753823e-01
2.60657668e-01 -2.49885041e-02 -2.01454878e-01 -3.76861930e-01
6.25488162e-01 -1.51478305e-01 1.50071293e-01 -3.14980090e-01
-8.06805313e-01 8.85498464e-01 6.66619539e-01 5.22324920e-01
-1.69531792e-01 3.92114557e-02 7.58380353e-01 5.55976890e-02
5.87587774e-01 2.34249383e-01 -1.91959739e-01 2.93577850e-01
-1.02528584e+00 -9.41237509e-02 3.72783303e-01 3.45914155e-01
2.72067070e-01 -6.07130170e-01 -3.95295382e-01 1.02089012e+00
-4.22975840e-03 3.88255417e-01 8.01902235e-01 -5.60771048e-01
-3.81975956e-02 7.22395837e-01 -2.51819223e-01 -7.62763083e-01
-1.06579113e+00 -6.80364132e-01 -1.02972555e+00 -1.64849922e-01
2.90076345e-01 2.35969901e-01 -1.34656489e+00 1.36969864e+00
1.80484161e-01 1.19919434e-01 9.78025645e-02 5.81002593e-01
1.05529189e+00 4.46656495e-02 5.82360506e-01 -1.54169798e-01
1.42241752e+00 -6.33813381e-01 -5.01152515e-01 -5.58237620e-02
1.45578265e+00 -3.41034889e-01 6.73119605e-01 1.28903762e-01
-8.65352809e-01 -2.77155727e-01 -9.84505594e-01 1.50232926e-01
-6.51411265e-02 6.30813301e-01 8.87954295e-01 9.79485869e-01
-9.96324539e-01 4.05107021e-01 -9.59409416e-01 -6.65619552e-01
9.05398250e-01 7.38170683e-01 -6.55373931e-01 -2.35641479e-01
-6.94545865e-01 8.24559093e-01 4.30633694e-01 -1.36869907e-01
-1.00653648e+00 -1.01033866e+00 -5.71580887e-01 -2.15137050e-01
9.25509483e-02 -6.23541832e-01 1.47987127e+00 -9.37303543e-01
-9.94139969e-01 1.36918938e+00 -1.89197287e-01 -7.00839877e-01
4.08067852e-01 6.25000238e-01 -3.79651308e-01 2.72356629e-01
2.10421592e-01 5.50803900e-01 2.69710422e-01 -9.07594502e-01
-1.25446784e+00 -5.34793496e-01 -3.91761750e-01 1.05536960e-01
-3.75779212e-01 -2.03364849e-01 -3.25188458e-01 -2.51086593e-01
2.71099895e-01 -1.00162232e+00 -5.11930585e-01 -4.17089695e-03
-3.12394854e-02 1.66123863e-02 4.09505069e-01 -7.28662074e-01
8.99917126e-01 -2.10722041e+00 -3.44268143e-01 3.49457502e-01
2.11109594e-01 1.30749971e-01 1.94271013e-01 -4.39477265e-01
-2.17599422e-01 3.40278536e-01 -7.78606385e-02 1.93463236e-01
-5.88406563e-01 7.75866508e-02 4.15069938e-01 7.16264844e-01
1.78250387e-01 1.21256864e+00 -7.93825507e-01 -5.81513405e-01
6.26683682e-02 2.38557868e-02 -5.60275078e-01 -5.49529567e-02
-6.89641107e-03 5.51517844e-01 -4.65803951e-01 7.59540200e-01
4.01438296e-01 -5.27335227e-01 7.61233926e-01 -5.18645287e-01
3.63894641e-01 1.06562808e-01 -4.34084266e-01 1.47623479e+00
-4.25768584e-01 6.16708636e-01 9.20081213e-02 -1.08919394e+00
7.15031028e-01 2.08997473e-01 7.36791849e-01 -9.40598011e-01
2.57829189e-01 4.73521739e-01 4.44839478e-01 -5.37706494e-01
1.60768464e-01 -4.32883382e-01 5.84095716e-02 9.37866420e-02
-4.76908423e-02 9.47740301e-02 5.94614521e-02 6.79503754e-03
1.91674411e+00 -4.54948962e-01 6.53686583e-01 -6.48928165e-01
5.18061996e-01 3.64818186e-01 4.10278559e-01 9.79727685e-01
-1.00877829e-01 5.26264012e-01 5.37314594e-01 -6.13509834e-01
-8.17791522e-01 -8.49329531e-01 -8.33524585e-01 9.44113731e-01
-1.52145773e-01 1.91663262e-02 -2.50521153e-01 -9.96839225e-01
5.38485236e-02 4.74942833e-01 -1.19913805e+00 -1.66233018e-01
-6.67617977e-01 -1.29541016e+00 7.12100029e-01 7.35380769e-01
2.57222682e-01 -7.30953872e-01 -3.25020045e-01 2.64345780e-02
-1.82310324e-02 -1.09304416e+00 2.28283674e-01 4.88381058e-01
-1.25864029e+00 -1.15399468e+00 -7.90052414e-01 -9.46479976e-01
1.03023624e+00 -2.16237471e-01 1.05085230e+00 1.47425504e-02
-6.76696599e-01 -4.51711975e-02 -2.72513688e-01 -4.72157270e-01
-6.70773864e-01 3.07140410e-01 -5.12147784e-01 4.29433137e-02
4.23389077e-01 4.04334776e-02 -6.81768835e-01 3.26586574e-01
-6.63535416e-01 2.16310576e-01 1.05422962e+00 1.17334139e+00
8.95288348e-01 -9.41510871e-02 5.40566266e-01 -1.48012018e+00
-1.14085168e-01 -5.51832139e-01 -1.63252831e-01 1.28474876e-01
-5.46370983e-01 -3.71477515e-01 9.47705805e-02 -2.15412945e-01
-1.03037560e+00 3.38572681e-01 -1.29582837e-01 3.32733154e-01
-1.58297032e-01 6.16618037e-01 2.03941658e-01 -3.77448708e-01
1.00340748e+00 -2.25983232e-01 -1.25269040e-01 -1.21111736e-01
-8.23820755e-02 7.84707546e-01 7.70203590e-01 -1.88194275e-01
1.94102079e-01 7.94366419e-01 3.16577524e-01 -6.76312566e-01
-1.10924816e+00 -7.66075850e-01 -4.96134758e-01 1.36746746e-02
8.99799526e-01 -8.01473618e-01 -4.90320802e-01 5.19284606e-02
-4.17200923e-01 -5.40256679e-01 -3.03253889e-01 4.86536950e-01
-6.67497993e-01 -2.48466119e-01 -7.82693923e-01 -2.68717825e-01
-3.00106317e-01 -1.01912618e+00 1.20972371e+00 2.15029433e-01
-4.45048928e-01 -1.01120877e+00 -3.08783203e-02 4.92357790e-01
5.08543372e-01 3.67016435e-01 1.24888265e+00 -9.32811856e-01
-1.27795666e-01 -6.67624712e-01 -5.42763770e-01 -2.07475469e-01
2.11157173e-01 -3.87935638e-01 -1.08128107e+00 -3.39291602e-01
-2.99527109e-01 -2.88575530e-01 1.14526772e+00 8.44183445e-01
1.61165559e+00 1.00893825e-01 -1.11500609e+00 7.15376377e-01
1.24845695e+00 2.81730264e-01 5.07591903e-01 5.25225103e-01
3.81079048e-01 3.92202705e-01 6.14240706e-01 8.95037800e-02
8.00406784e-02 4.18282330e-01 3.77103895e-01 -5.50350249e-01
-4.12260264e-01 -1.56361625e-01 -4.82549489e-01 -3.13070975e-02
6.04941584e-02 -1.33601546e-01 -1.31305254e+00 5.64713836e-01
-1.65868413e+00 -5.64831316e-01 -1.61043897e-01 1.93555117e+00
1.01519048e+00 3.10970157e-01 -2.94581234e-01 1.10085867e-01
5.36477625e-01 -3.29270035e-01 -4.46751624e-01 -2.83239275e-01
-7.76355416e-02 3.07405680e-01 8.73896182e-01 2.00343966e-01
-1.40265977e+00 6.83484256e-01 6.87516499e+00 9.15094674e-01
-1.23587513e+00 3.17980081e-01 1.39216530e+00 9.47527064e-04
1.00597881e-01 -2.75318980e-01 -8.02107811e-01 2.12647580e-02
1.22273219e+00 3.60017568e-02 -4.88114387e-01 5.79886854e-01
-5.98510774e-03 -4.95742291e-01 -1.26333010e+00 8.10720384e-01
1.93405107e-01 -1.74081826e+00 -1.89317048e-01 2.60862529e-01
8.47024798e-01 1.20430052e-01 3.47628593e-02 3.02690476e-01
8.83188471e-03 -1.37851441e+00 -1.13522910e-01 4.30661738e-01
1.21357000e+00 -5.76309919e-01 1.31891739e+00 1.17235094e-01
-7.05742836e-01 -3.20263319e-02 -1.91699475e-01 3.72506559e-01
-2.85595894e-01 5.93858778e-01 -1.51988554e+00 3.28670382e-01
7.21046448e-01 6.12593174e-01 -8.60311866e-01 1.05688632e+00
4.57223088e-01 9.97399688e-01 -1.22915171e-02 2.19670311e-01
-4.39047702e-02 6.20217562e-01 -7.36488402e-02 1.41832173e+00
2.89087981e-01 3.97073030e-01 3.44700962e-02 1.12777553e-01
-5.34213781e-02 2.56964386e-01 -2.10017473e-01 3.42924856e-02
-3.65084335e-02 1.63397312e+00 -1.04975951e+00 -4.40145522e-01
-2.40300104e-01 2.84622848e-01 -6.82364702e-02 -1.59523159e-01
-6.54087007e-01 3.56906019e-02 -1.57589495e-01 4.42240834e-01
1.67602420e-01 5.37575841e-01 -6.84257448e-01 -3.59289110e-01
-4.06435698e-01 -7.98685610e-01 7.79002070e-01 -2.77545482e-01
-1.12511945e+00 5.01458108e-01 -1.56487584e-01 -9.12583351e-01
-2.68850237e-01 -8.46898496e-01 -3.64893287e-01 4.68305409e-01
-1.51237059e+00 -1.34258533e+00 -4.32954699e-01 -1.07652381e-01
2.25235611e-01 -3.11772764e-01 1.15967178e+00 2.92224735e-01
-2.07148850e-01 1.08221674e+00 1.42913461e-01 1.76920906e-01
8.10597539e-01 -1.24434161e+00 -6.63708597e-02 -6.99423924e-02
-3.95459533e-01 -6.10221587e-02 3.38118792e-01 -7.69596934e-01
-1.14619946e+00 -1.50391603e+00 7.56980598e-01 -3.54395181e-01
6.73137248e-01 6.14249520e-02 -6.00859702e-01 7.85437822e-01
-3.34345311e-01 4.02182281e-01 1.05896044e+00 1.35820463e-01
-2.38785744e-02 -3.70909423e-02 -1.49003708e+00 1.96513742e-01
7.92908370e-01 -8.69774073e-02 -7.22676143e-02 5.82599998e-01
5.16453385e-02 -6.43652678e-01 -1.29679227e+00 1.31669211e+00
6.07076049e-01 -6.85907602e-01 8.66568506e-01 -5.69409788e-01
4.99323666e-01 4.60047312e-02 -2.27304593e-01 -7.05658257e-01
-5.53375423e-01 1.33138180e-01 3.61909449e-01 7.81748116e-01
8.40980291e-01 -3.82907718e-01 1.40115047e+00 3.20082515e-01
-1.55321717e-01 -1.04619217e+00 -1.01218140e+00 -3.51707935e-01
2.02817231e-01 -2.50835896e-01 9.48552191e-02 8.31146717e-01
-1.21743388e-01 8.19851756e-02 2.38546878e-01 1.72788888e-01
4.48384851e-01 6.67186873e-03 3.55122238e-01 -1.12531888e+00
-3.69260937e-01 -5.63984215e-01 -8.69997621e-01 -3.59066248e-01
-3.49656381e-02 -1.38276613e+00 -5.26058078e-02 -1.30849731e+00
1.10388589e+00 -9.06090498e-01 -6.12461150e-01 6.73180163e-01
-1.46091744e-01 7.93696165e-01 -4.56045002e-01 1.57504588e-01
-5.80362082e-01 -2.18593895e-01 1.05743301e+00 -6.35094106e-01
5.53772040e-02 1.23014025e-01 -7.81206131e-01 9.01792705e-01
8.37562621e-01 -6.64316893e-01 1.64826140e-02 -8.46310258e-02
-5.40560000e-02 3.14853549e-01 3.38657349e-01 -1.04177499e+00
-4.68012458e-03 -1.23048514e-01 8.87636662e-01 -6.94277108e-01
2.62144625e-01 -3.36736262e-01 8.13120231e-02 8.18191767e-01
-7.51060426e-01 -5.08093715e-01 4.65655088e-01 5.61621845e-01
-8.05844367e-02 -2.82126397e-01 7.89086461e-01 2.63306871e-02
-5.72212636e-01 1.41139001e-01 -3.61571491e-01 -2.70477802e-01
1.12101078e+00 -3.00656825e-01 -3.24202538e-01 3.92599367e-02
-9.68662620e-01 9.04899687e-02 2.66821861e-01 1.86521307e-01
3.25228809e-03 -1.21025777e+00 -9.53023851e-01 5.00097461e-02
4.76166874e-01 -1.81890935e-01 2.70887017e-01 1.08600068e+00
-4.66418356e-01 6.41971886e-01 -1.39453867e-02 -1.08015203e+00
-1.56159985e+00 -1.95194427e-02 7.43351579e-01 -5.86360276e-01
-4.18851703e-01 9.82693374e-01 4.38473612e-01 -4.02770460e-01
1.63955927e-01 -1.99005395e-01 -3.08698714e-01 -2.55737096e-01
5.51722348e-01 1.20461896e-01 6.73630238e-01 -5.02038181e-01
-2.71978289e-01 3.22838992e-01 -5.39664567e-01 2.07430050e-01
1.31394958e+00 4.60190266e-01 3.49980593e-02 2.43731633e-01
1.33934224e+00 -3.02021265e-01 -5.64825296e-01 -3.46241742e-01
2.86796331e-01 -1.20994724e-01 4.35589910e-01 -1.22606194e+00
-8.81212294e-01 4.10299003e-01 1.37833965e+00 -2.43087515e-01
1.14889407e+00 4.50018585e-01 2.86610305e-01 2.73869902e-01
1.24678917e-01 -4.69875693e-01 -1.60999045e-01 9.79475528e-02
5.23469031e-01 -1.54594946e+00 3.35012138e-01 -5.99584460e-01
-4.01992261e-01 9.14915442e-01 4.23924565e-01 -3.44988182e-02
5.65646768e-01 4.15796012e-01 2.22801968e-01 -2.93725938e-01
-7.09548652e-01 4.22837920e-02 3.78004164e-01 4.88059908e-01
8.76214564e-01 5.10169804e-01 -4.33110595e-01 7.74342000e-01
-1.94380224e-01 1.67507648e-01 1.46986142e-01 9.61081266e-01
-6.82886183e-01 -1.12060046e+00 -2.30156139e-01 1.30463052e+00
-8.12971950e-01 1.17829598e-01 -3.38995785e-01 8.23068500e-01
6.67848960e-02 6.06961429e-01 2.24861071e-01 -2.41249129e-01
1.94346711e-01 -2.84382433e-01 5.29484093e-01 -8.03367674e-01
-5.66140890e-01 1.02076814e-01 2.93249547e-01 -3.39825392e-01
-3.36202085e-01 -5.99488497e-01 -1.30385256e+00 2.33020619e-01
-3.47497255e-01 9.67868716e-02 7.39238918e-01 8.28520954e-01
1.26567572e-01 8.15476537e-01 3.70458871e-01 -5.98338783e-01
-5.23385763e-01 -9.95501339e-01 -3.80780846e-01 1.96556479e-01
2.06206053e-01 -4.11303163e-01 6.72803596e-02 1.35529220e-01] | [15.208754539489746, -2.606182336807251] |
e6835e5b-b2c1-4f8b-b1fe-d0a6febf3fa9 | sensitive-data-detection-with-high-throughput-1 | 2305.03169 | null | https://arxiv.org/abs/2305.03169v2 | https://arxiv.org/pdf/2305.03169v2.pdf | Sensitive Data Detection with High-Throughput Machine Learning Models in Electrical Health Records | In the era of big data, there is an increasing need for healthcare providers, communities, and researchers to share data and collaborate to improve health outcomes, generate valuable insights, and advance research. The Health Insurance Portability and Accountability Act of 1996 (HIPAA) is a federal law designed to protect sensitive health information by defining regulations for protected health information (PHI). However, it does not provide efficient tools for detecting or removing PHI before data sharing. One of the challenges in this area of research is the heterogeneous nature of PHI fields in data across different parties. This variability makes rule-based sensitive variable identification systems that work on one database fail on another. To address this issue, our paper explores the use of machine learning algorithms to identify sensitive variables in structured data, thus facilitating the de-identification process. We made a key observation that the distributions of metadata of PHI fields and non-PHI fields are very different. Based on this novel finding, we engineered over 30 features from the metadata of the original features and used machine learning to build classification models to automatically identify PHI fields in structured Electronic Health Record (EHR) data. We trained the model on a variety of large EHR databases from different data sources and found that our algorithm achieves 99% accuracy when detecting PHI-related fields for unseen datasets. The implications of our study are significant and can benefit industries that handle sensitive data. | ['Xiaoqian Jiang', 'Kai Zhang'] | 2023-04-30 | null | null | null | null | ['de-identification'] | ['natural-language-processing'] | [ 3.32157016e-01 -3.07480874e-03 -5.16762376e-01 -4.91072685e-01
-9.50310230e-01 -6.47928059e-01 -1.74090609e-01 9.43600535e-01
-2.03092530e-01 8.28054726e-01 6.50606036e-01 -4.60863352e-01
-1.88861340e-01 -1.00857997e+00 -5.25117636e-01 -4.97815669e-01
1.80073172e-01 2.90083021e-01 -5.88991530e-02 5.74254282e-02
2.53418088e-01 3.24308842e-01 -1.30140197e+00 8.86748552e-01
8.89802873e-01 9.43569362e-01 -1.86134234e-01 3.94495338e-01
-2.93928552e-02 9.90909278e-01 -5.87122619e-01 -3.44559699e-01
4.76443678e-01 -3.90768617e-01 -8.12416255e-01 -6.26449764e-01
7.01424405e-02 -9.21245292e-02 2.74847355e-02 1.07784319e+00
5.31233370e-01 -5.38259923e-01 3.21608275e-01 -1.11353624e+00
-8.27961504e-01 6.23580098e-01 -3.20761830e-01 1.47690788e-01
6.12918496e-01 -6.31363643e-03 9.69502032e-01 -2.02287093e-01
8.32161486e-01 5.90333819e-01 1.04208624e+00 2.24082813e-01
-1.11460710e+00 -6.54699922e-01 -5.70276260e-01 -3.39923166e-02
-1.46839976e+00 -1.55297682e-01 2.37782940e-01 -8.34971666e-01
7.41436362e-01 7.68084228e-01 4.36800092e-01 6.92513108e-01
6.08812332e-01 1.53539002e-01 9.89316106e-01 -3.86969119e-01
2.66857952e-01 5.26613414e-01 4.83393967e-01 3.36652189e-01
9.93204832e-01 1.09426275e-01 -2.04883263e-01 -9.84591842e-01
1.66510522e-01 5.98670363e-01 -2.70935595e-01 -1.54772997e-01
-1.15130079e+00 8.49015236e-01 4.51011546e-02 4.65365976e-01
-6.37069225e-01 -5.65508544e-01 3.93172383e-01 3.36876005e-01
7.91360065e-02 6.65670037e-01 -7.70592868e-01 -6.84112078e-03
-4.08313692e-01 2.60096788e-01 9.48363721e-01 8.03689897e-01
7.09089577e-01 -8.00258577e-01 -1.74040705e-01 4.13154811e-01
1.90703332e-01 4.98850465e-01 3.01390558e-01 -7.61418402e-01
5.24554014e-01 1.09059846e+00 1.45112932e-01 -1.35134065e+00
-5.04666448e-01 1.80233009e-02 -7.43865609e-01 -1.56008616e-01
3.76642048e-01 -3.15720826e-01 -6.36346579e-01 1.36429024e+00
2.95921475e-01 -4.45110083e-01 1.10600300e-01 3.62334073e-01
5.76984406e-01 2.06217289e-01 4.12904829e-01 -1.27508998e-01
1.63024020e+00 1.20405823e-01 -9.20163453e-01 3.62110227e-01
8.88215780e-01 -4.72639769e-01 7.18421221e-01 2.04437569e-01
-6.76105917e-01 -1.46230891e-01 -8.35525036e-01 5.17635047e-02
-6.74752474e-01 -6.80325985e-01 5.43064296e-01 1.00275600e+00
-6.33534253e-01 3.66373837e-01 -5.50769687e-01 -4.29157019e-01
7.47628868e-01 3.27753395e-01 -4.62052375e-01 -1.92168191e-01
-1.39513350e+00 7.19422758e-01 3.62335414e-01 -4.75071311e-01
-1.09785266e-01 -1.09261787e+00 -8.68721426e-01 2.04418242e-01
1.89284697e-01 -7.38015890e-01 7.66442776e-01 -6.01301908e-01
-3.59352380e-01 7.58284926e-01 4.83179130e-02 -3.13466161e-01
8.73396546e-02 -6.09665401e-02 -8.40239286e-01 -7.09828064e-02
3.08293849e-01 -1.41873851e-01 1.24854103e-01 -1.14655018e+00
-1.10454571e+00 -8.01791430e-01 -4.23684299e-01 -4.83871967e-01
-4.07979190e-01 2.68003017e-01 2.39162877e-01 -3.51336211e-01
-1.16654158e-01 -8.21151376e-01 -2.57769972e-01 -5.40409684e-01
-3.10376316e-01 1.64613381e-01 5.26903808e-01 -9.68256235e-01
1.57837343e+00 -2.24700689e+00 -5.99956155e-01 5.69462001e-01
5.05142272e-01 1.86398670e-01 3.99813443e-01 4.88446951e-01
6.21784143e-02 5.57910800e-01 -1.63397074e-01 5.45590281e-01
-4.20520633e-01 1.87282518e-01 -2.73307562e-01 3.73815149e-01
-6.25071749e-02 7.95719683e-01 -8.50923419e-01 -4.45743889e-01
-3.28128599e-03 5.47895432e-01 -5.66985011e-01 1.27422318e-01
1.11705892e-01 4.98924166e-01 -5.34704864e-01 9.96198416e-01
6.26042366e-01 -6.25696957e-01 5.84277987e-01 -2.99425751e-01
-1.65431738e-01 2.44817600e-01 -1.05456913e+00 1.08384168e+00
1.12556979e-01 6.89968392e-02 -2.71615654e-01 -6.01560295e-01
8.38221431e-01 5.50513506e-01 1.09756875e+00 -6.50090933e-01
1.34812415e-01 1.65358633e-01 -4.49361727e-02 -9.83761549e-01
2.72309661e-01 -9.93441716e-02 -3.10020030e-01 3.35899770e-01
-5.73796570e-01 7.53772199e-01 -2.87662506e-01 -3.03219445e-02
1.51538217e+00 -6.61978722e-01 6.88303471e-01 -3.22335005e-01
3.81079495e-01 2.02215225e-01 9.93702829e-01 5.98470569e-01
-3.44694346e-01 5.41507840e-01 4.57842708e-01 -7.13334441e-01
-9.05559361e-01 -8.27893972e-01 -6.77643478e-01 7.89729953e-01
-5.47839515e-02 -5.00912011e-01 -3.55723858e-01 -6.26348972e-01
6.84055150e-01 3.84117365e-01 -7.50582159e-01 4.29121442e-02
-3.54873478e-01 -8.60834360e-01 5.36808372e-01 4.46176797e-01
2.17422873e-01 -8.20900917e-01 -9.70243514e-01 1.97918206e-01
-2.88992733e-01 -7.15927005e-01 -3.44563395e-01 5.76187707e-02
-4.47394758e-01 -1.52221572e+00 -1.48382798e-01 -5.95467567e-01
5.96625507e-01 -1.25624677e-02 1.07949924e+00 1.57031491e-01
-5.69270909e-01 9.49866846e-02 -3.72853607e-01 -8.29210460e-01
-6.14228547e-01 5.90729713e-02 -1.56832978e-01 -5.98130897e-02
9.73381817e-01 -1.08324721e-01 -5.77931046e-01 3.16773564e-01
-9.88694370e-01 -5.05967975e-01 3.84688795e-01 6.05687022e-01
6.31781697e-01 1.29514650e-01 8.31888616e-01 -1.46477902e+00
4.90997255e-01 -9.37668145e-01 -4.59179908e-01 4.89958823e-01
-1.04094803e+00 8.41921493e-02 4.44857091e-01 -1.18896954e-01
-8.43835652e-01 8.19041207e-02 -2.89181359e-02 3.20846975e-01
-1.90025151e-01 6.03666663e-01 -4.75218743e-01 8.26304685e-03
8.43649685e-01 -4.02026266e-01 1.21019050e-01 -5.67550182e-01
-2.92351097e-01 1.30171168e+00 2.94089496e-01 -2.20435217e-01
3.63395214e-01 5.54031193e-01 -1.53836876e-01 -4.44499224e-01
-7.45108783e-01 -8.17058027e-01 -3.30674827e-01 3.45022768e-01
1.08785141e+00 -7.92955697e-01 -9.17159081e-01 1.21300943e-01
-6.58223629e-01 3.06250513e-01 -3.48564863e-01 4.01581347e-01
7.55293816e-02 4.12257642e-01 -5.60253024e-01 -5.49824536e-01
-4.32798296e-01 -7.44904339e-01 6.54583991e-01 1.08277403e-01
-7.15014100e-01 -8.10807288e-01 5.48589230e-01 7.30663002e-01
6.91907823e-01 7.85116374e-01 1.28577602e+00 -9.72790539e-01
-3.80411506e-01 -4.36767638e-01 -9.35407653e-02 5.87313473e-02
6.24610841e-01 -1.56093165e-01 -9.30626392e-01 -2.04273134e-01
2.68446684e-01 1.20426044e-01 4.18837756e-01 3.36986721e-01
1.07712615e+00 -6.31585598e-01 -6.53781474e-01 6.84811890e-01
1.63033342e+00 6.20888054e-01 9.53561842e-01 4.84921515e-01
6.71590090e-01 5.92011273e-01 4.49224263e-01 6.95940495e-01
4.75996435e-01 3.94826084e-01 4.32522371e-02 -2.32893795e-01
6.06669009e-01 -1.17375493e-01 -2.59271353e-01 1.85246021e-01
4.41608205e-02 5.98774776e-02 -1.26405585e+00 5.91233432e-01
-1.62924826e+00 -1.08400798e+00 -1.87010899e-01 2.47074366e+00
1.04351342e+00 -2.22951725e-01 1.76015407e-01 1.48748040e-01
8.05249751e-01 -3.44566911e-01 -3.68399739e-01 -2.06700072e-01
-7.93254897e-02 1.84825018e-01 1.05328143e+00 1.32299393e-01
-1.00851190e+00 6.24452047e-02 6.60714722e+00 -2.55653739e-01
-1.00668550e+00 9.59042385e-02 6.10136390e-01 1.48307353e-01
-5.07916808e-01 -2.33004630e-01 -7.66945779e-01 6.35888517e-01
1.15333557e+00 -4.24115181e-01 6.24174485e-03 7.91226506e-01
-5.15990704e-02 7.98663497e-02 -9.44422305e-01 8.00613463e-01
-8.47848952e-02 -1.48269439e+00 -1.68870509e-01 4.35193449e-01
8.10496390e-01 -8.27975199e-02 -4.15424332e-02 -8.37870538e-02
3.41387987e-01 -9.85186398e-01 -9.75512117e-02 6.71615303e-01
7.38852918e-01 -6.45613670e-01 1.07833123e+00 1.07680690e-02
-8.13622773e-01 -5.62633634e-01 -3.35817307e-01 -2.67072748e-02
5.14261387e-02 8.38718653e-01 -1.01634920e+00 3.99456382e-01
8.48151147e-01 2.90718377e-01 -4.23781425e-01 1.11875761e+00
4.87133086e-01 4.48232114e-01 1.99126363e-01 4.21506792e-01
-5.34613729e-01 1.90064847e-01 6.79082051e-02 1.05194747e+00
2.81318724e-01 1.81881219e-01 2.54990309e-02 5.63626885e-01
-9.44012776e-02 1.97545424e-01 -7.66064465e-01 -1.75619379e-01
6.00833952e-01 1.03315592e+00 -3.79507393e-01 -3.15106034e-01
-7.48310089e-01 2.56765991e-01 -1.84744671e-01 8.93483758e-02
-6.43751442e-01 -5.24025738e-01 1.09990454e+00 5.34467161e-01
2.12235034e-01 3.90899152e-01 -8.39742362e-01 -1.08909237e+00
-1.90115482e-01 -1.16506398e+00 1.07985771e+00 -7.09987581e-02
-1.42441821e+00 2.56737441e-01 -1.96030423e-01 -9.61724102e-01
-2.98427105e-01 -2.02712268e-01 2.20669374e-01 1.00584483e+00
-1.23064387e+00 -8.96560848e-01 -4.76175904e-01 6.63193464e-01
-4.69207317e-01 -3.62044238e-02 1.11793292e+00 5.40651619e-01
-1.95305333e-01 6.84815645e-01 2.58623809e-01 4.23362732e-01
8.61047328e-01 -9.55031216e-01 3.07023406e-01 4.50875521e-01
-2.88200557e-01 1.06807816e+00 4.77278501e-01 -1.23464704e+00
-1.50011683e+00 -1.12324452e+00 1.30070829e+00 -9.42275524e-01
2.03886300e-01 -2.05172315e-01 -1.27833509e+00 8.20613444e-01
-1.14991464e-01 -7.06017986e-02 1.61423135e+00 2.56475300e-01
-5.73881209e-01 -2.43650749e-01 -1.79292393e+00 -3.21428552e-02
6.62018657e-01 -4.81680185e-01 -5.29628813e-01 1.55660525e-01
7.27703631e-01 -9.23183486e-02 -1.59938884e+00 5.54269314e-01
7.63057888e-01 -8.63640606e-01 8.97048950e-01 -8.50084662e-01
8.73392299e-02 -2.45079622e-01 -3.63412142e-01 -7.04952180e-01
-5.48135638e-01 -3.38725984e-01 8.76501650e-02 1.07743871e+00
4.70857799e-01 -1.04136753e+00 6.82448030e-01 1.52634883e+00
2.55710244e-01 -4.64323312e-01 -5.34322441e-01 -4.04713631e-01
-1.03227295e-01 1.59941614e-02 1.11567879e+00 1.40816844e+00
4.78300691e-01 -3.80266048e-02 -3.20920646e-01 3.60347778e-01
5.53212643e-01 1.36201635e-01 5.82860947e-01 -1.67281532e+00
-2.63734192e-01 1.94485456e-01 -6.02065742e-01 1.46454930e-01
-3.50739509e-01 -7.01287031e-01 -3.60231847e-01 -1.47280288e+00
5.52562833e-01 -8.42868149e-01 -6.96936488e-01 8.51014972e-01
-3.74771833e-01 -4.79511842e-02 -8.02242607e-02 4.17462081e-01
-1.82083800e-01 -5.17562866e-01 6.43932700e-01 1.25789894e-02
-4.44540411e-01 -2.78417813e-03 -1.41432846e+00 1.94487751e-01
8.52496505e-01 -7.62114525e-01 -1.17082976e-01 -2.84466390e-02
5.61694443e-01 -2.28331331e-03 7.60688186e-02 -9.55699623e-01
7.15789646e-02 -4.55518812e-01 5.20113647e-01 -4.30677295e-01
-3.75991225e-01 -1.20554399e+00 8.77235770e-01 7.39416242e-01
-5.28357089e-01 1.25476405e-01 3.12628523e-02 3.02809656e-01
5.03211515e-03 1.13351136e-01 5.00380337e-01 -1.05632119e-01
-3.19701225e-01 1.26014650e-01 -1.32502288e-01 2.00386241e-01
1.29532254e+00 -2.27241181e-02 -7.20935404e-01 1.81761067e-02
-2.28157595e-01 4.79664169e-02 8.92035961e-01 2.40827516e-01
3.20438832e-01 -1.06106281e+00 -6.38552666e-01 6.59768224e-01
3.59728366e-01 -1.49840429e-01 2.40375042e-01 6.84396625e-01
-4.20107961e-01 5.35266519e-01 -3.62546086e-01 -3.82158190e-01
-1.48047864e+00 1.05465734e+00 -2.49441783e-03 -1.37675449e-01
-6.87328935e-01 3.92612033e-02 -3.53237167e-02 -1.47304311e-01
1.20517043e-02 -3.73743266e-01 -1.17677681e-01 1.09377049e-01
1.06959450e+00 4.22387064e-01 2.25251287e-01 -4.89466310e-01
-5.11738122e-01 1.23353921e-01 -2.25215092e-01 3.28573048e-01
1.42047739e+00 5.23723923e-02 -2.31901124e-01 3.23837698e-01
1.32921493e+00 3.50039303e-01 -5.48802257e-01 1.30319089e-01
2.75709659e-01 -8.69085133e-01 -4.80905533e-01 -8.68108630e-01
-9.00200784e-01 3.03555995e-01 7.35663474e-01 5.88893175e-01
1.03696048e+00 1.17344975e-01 8.79213214e-01 7.84412492e-03
5.28277695e-01 -8.92961740e-01 -6.76986873e-01 6.83181509e-02
1.90829113e-01 -1.22736466e+00 -6.72595724e-02 -2.69139856e-01
-7.43379653e-01 6.99689507e-01 5.32369465e-02 5.41613281e-01
7.51646519e-01 6.43035710e-01 3.49982530e-01 -3.27102929e-01
-5.91758013e-01 2.65343219e-01 -1.89533204e-01 8.76747072e-01
6.26490414e-01 3.16389263e-01 -5.58886826e-01 6.29041731e-01
1.05015570e-02 6.39825702e-01 4.15555477e-01 1.28316259e+00
-4.90419954e-01 -1.30369270e+00 -5.43998480e-01 1.07564044e+00
-1.05690014e+00 -1.39150962e-01 -2.66038865e-01 5.28221667e-01
3.79850298e-01 9.79060650e-01 -1.34853989e-01 -5.69475114e-01
3.67805123e-01 2.48880878e-01 -1.44578770e-01 -5.07327914e-01
-1.00532067e+00 -4.76749629e-01 7.79599845e-02 -4.61055070e-01
-1.70563668e-01 -8.34580421e-01 -1.35338795e+00 -3.68762672e-01
-1.10166082e-02 2.32913569e-01 5.95824718e-01 6.45077288e-01
9.98189092e-01 2.13323310e-01 6.03179991e-01 3.28212053e-01
-7.15229273e-01 -4.04452890e-01 -6.78418636e-01 1.02089858e+00
4.20764506e-01 -2.35177413e-01 -2.97078878e-01 1.76410988e-01] | [6.883928298950195, 6.843059539794922] |
ce76ea54-3355-446f-92a6-0595cdf47510 | unsupervised-view-invariant-human-posture | 2109.08730 | null | https://arxiv.org/abs/2109.08730v1 | https://arxiv.org/pdf/2109.08730v1.pdf | Unsupervised View-Invariant Human Posture Representation | Most recent view-invariant action recognition and performance assessment approaches rely on a large amount of annotated 3D skeleton data to extract view-invariant features. However, acquiring 3D skeleton data can be cumbersome, if not impractical, in in-the-wild scenarios. To overcome this problem, we present a novel unsupervised approach that learns to extract view-invariant 3D human pose representation from a 2D image without using 3D joint data. Our model is trained by exploiting the intrinsic view-invariant properties of human pose between simultaneous frames from different viewpoints and their equivariant properties between augmented frames from the same viewpoint. We evaluate the learned view-invariant pose representations for two downstream tasks. We perform comparative experiments that show improvements on the state-of-the-art unsupervised cross-view action classification accuracy on NTU RGB+D by a significant margin, on both RGB and depth images. We also show the efficiency of transferring the learned representations from NTU RGB+D to obtain the first ever unsupervised cross-view and cross-subject rank correlation results on the multi-view human movement quality dataset, QMAR, and marginally improve on the-state-of-the-art supervised results for this dataset. We also carry out ablation studies to examine the contributions of the different components of our proposed network. | ['Majid Mirmehdi', 'Björn Ommer', 'Faegheh Sardari'] | 2021-09-17 | null | null | null | null | ['3d-pose-estimation', 'action-analysis', 'action-assessment', '3d-human-action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 3.49526465e-01 -1.48473859e-01 -2.45683968e-01 -4.62400764e-01
-1.03065193e+00 -5.37429154e-01 4.78928745e-01 -3.99404049e-01
-5.93759060e-01 4.10618484e-01 5.76389790e-01 4.90004450e-01
-2.04142630e-01 -2.18725577e-01 -6.17285669e-01 -5.56873143e-01
-1.81866959e-01 5.77706993e-01 3.19681108e-01 -3.52992296e-01
-1.00554839e-01 5.16489923e-01 -1.70469463e+00 5.24586380e-01
8.79552588e-02 9.32100117e-01 -1.75395310e-01 7.01330006e-01
6.20297432e-01 5.89169741e-01 -3.09852511e-01 -1.48851275e-01
6.79839492e-01 -5.41994154e-01 -7.75221467e-01 5.26822031e-01
9.91491258e-01 -5.49990177e-01 -6.02317393e-01 5.81729531e-01
7.91592300e-01 1.16894066e-01 4.92883027e-01 -1.33492780e+00
-1.94384605e-01 2.47317236e-02 -7.01976597e-01 3.62192482e-01
8.19895029e-01 3.23537916e-01 1.04145122e+00 -7.86949396e-01
8.47577453e-01 1.16467035e+00 5.61023414e-01 6.11178339e-01
-1.22873843e+00 -4.05260563e-01 1.27703011e-01 3.41762930e-01
-1.03920364e+00 -4.22801346e-01 9.16852951e-01 -4.52052861e-01
1.11123943e+00 1.10729732e-01 9.70457792e-01 1.52459216e+00
1.48749277e-01 1.05817986e+00 1.32976508e+00 -2.43891612e-01
-6.37115091e-02 -6.26418233e-01 -1.31544143e-01 9.20716405e-01
1.62415996e-01 1.18681006e-01 -1.18393445e+00 1.69669703e-01
8.69159162e-01 1.14679001e-01 -3.76910418e-01 -1.24337292e+00
-1.44919252e+00 6.14350140e-01 3.96594554e-01 1.28044412e-01
-2.32818231e-01 2.48805031e-01 5.58527589e-01 3.23348314e-01
4.90415961e-01 3.02225500e-01 -8.11891079e-01 -5.82080305e-01
-6.46828413e-01 1.47325873e-01 3.14970195e-01 7.11144149e-01
2.86538213e-01 -6.51949272e-02 7.59289935e-02 4.74222839e-01
1.08887941e-01 5.56125104e-01 5.39765477e-01 -1.28672397e+00
7.58464217e-01 7.65146136e-01 -1.64974302e-01 -7.75306106e-01
-8.55155587e-01 -3.34109396e-01 -4.89392877e-01 3.81675094e-01
8.20557833e-01 2.37740427e-01 -8.61951470e-01 1.78170133e+00
3.50521058e-01 -3.83578122e-01 -2.11435989e-01 1.09905219e+00
5.25313795e-01 -2.35343009e-01 -2.14265764e-01 -6.30212650e-02
1.33316302e+00 -8.58670413e-01 -3.46002638e-01 -2.44550705e-01
7.62826324e-01 -3.84459555e-01 1.15196478e+00 5.79802752e-01
-9.85301554e-01 -6.49319232e-01 -1.06145811e+00 -1.69862583e-01
-1.03993751e-01 3.59119326e-01 6.64235294e-01 6.07945144e-01
-5.33578992e-01 7.93589115e-01 -1.29700136e+00 -6.14512682e-01
4.16315705e-01 3.18413377e-01 -1.16605830e+00 -1.20172612e-01
-8.82843733e-01 8.82996202e-01 -1.88978557e-02 -4.10646619e-03
-9.51484919e-01 -4.94522631e-01 -1.00629139e+00 -4.95624661e-01
6.35127664e-01 -8.87968242e-01 8.68077755e-01 -7.02911675e-01
-1.40469432e+00 1.22788405e+00 1.09526411e-01 -1.70909151e-01
8.38047743e-01 -7.59444773e-01 -1.10677756e-01 5.93921661e-01
2.19806656e-01 4.88817871e-01 7.29309916e-01 -1.01851201e+00
-4.91677940e-01 -1.05531704e+00 1.59434125e-01 4.37308162e-01
-1.72029704e-01 -3.83093774e-01 -5.63509405e-01 -6.46969080e-01
5.58119535e-01 -1.33560073e+00 -3.21906283e-02 3.92621070e-01
-1.09713249e-01 6.21233024e-02 4.95033979e-01 -7.68584013e-01
4.75490451e-01 -1.94344735e+00 8.51714253e-01 -3.55543382e-02
2.52692942e-02 -1.07679971e-01 -3.44281010e-02 1.13741994e-01
-2.73750484e-01 -3.91515970e-01 -7.60016665e-02 -3.27855051e-01
-1.39996171e-01 3.86566192e-01 2.91230172e-01 9.63618398e-01
5.71039841e-02 7.55124032e-01 -9.63267744e-01 -3.96665215e-01
3.59958410e-01 3.67688656e-01 -7.10006356e-01 1.30743042e-01
2.08193794e-01 9.25335526e-01 -3.62992972e-01 6.93611979e-01
3.49441692e-02 -2.63769597e-01 2.18567729e-01 -6.18548334e-01
4.16899711e-01 2.05220938e-01 -1.20617855e+00 2.73737383e+00
-2.92991668e-01 4.90469813e-01 -2.35010207e-01 -1.02284896e+00
3.90191227e-01 2.58510113e-01 1.12975335e+00 -5.98571122e-01
1.91081613e-01 2.55339183e-02 -1.45430332e-02 -5.33962667e-01
1.77196234e-01 -1.15543686e-01 -2.16391265e-01 4.15656507e-01
4.84416723e-01 7.58537650e-02 9.87863168e-02 5.12986332e-02
1.32382095e+00 8.78219783e-01 5.04110098e-01 1.12374604e-01
3.75341088e-01 -2.55769163e-01 4.75647151e-01 2.70325780e-01
-4.85024959e-01 8.34760666e-01 4.27084625e-01 -4.54748303e-01
-8.75129879e-01 -1.40801466e+00 1.78440109e-01 1.06269395e+00
-2.22800180e-01 -5.99540830e-01 -4.45809603e-01 -1.14136350e+00
8.93716961e-02 1.02497838e-01 -9.08027470e-01 -1.91366568e-01
-6.89669788e-01 -4.14541394e-01 5.58993518e-01 9.58930731e-01
6.09018862e-01 -5.10222733e-01 -1.12740982e+00 -3.49862516e-01
-3.99934620e-01 -1.51839423e+00 -2.37703681e-01 1.73001885e-01
-1.13675070e+00 -1.39475477e+00 -8.08951616e-01 -1.51700974e-01
5.56692243e-01 3.42655569e-01 1.02291322e+00 -4.33962852e-01
-3.79913330e-01 9.33529556e-01 -5.67451119e-01 1.92824304e-01
1.24123752e-01 -5.84492050e-02 5.24878025e-01 -9.78282988e-02
1.77906305e-01 -8.55037093e-01 -6.03121400e-01 6.18969202e-01
-5.42678654e-01 -1.06390066e-01 6.17815197e-01 8.22844207e-01
7.01104939e-01 -2.89166689e-01 2.57774908e-02 -4.02886540e-01
-1.58227846e-01 1.94325551e-01 -7.37326741e-02 7.56344348e-02
-1.98713824e-01 2.10797608e-01 8.10489058e-02 -2.60035157e-01
-5.99611223e-01 6.11690998e-01 1.04986154e-01 -7.76354074e-01
-2.21698582e-01 1.75879762e-01 -4.51844186e-01 -5.94153702e-02
8.51806223e-01 3.00143845e-02 2.20873803e-01 -4.96077210e-01
5.88491321e-01 1.96791038e-01 5.81813872e-01 -4.43399161e-01
8.02603245e-01 1.01241875e+00 4.56045121e-01 -5.74972510e-01
-1.13341749e+00 -7.00153768e-01 -1.47964489e+00 -4.59480554e-01
1.17753732e+00 -1.30168211e+00 -5.93073487e-01 4.49384451e-01
-7.13351905e-01 -1.78571716e-01 -4.81667876e-01 8.24021518e-01
-1.22993195e+00 5.82688868e-01 -3.24162692e-01 -3.76340389e-01
6.88647851e-02 -1.14257753e+00 1.50714719e+00 -4.54900861e-01
-4.60034847e-01 -5.68293214e-01 2.23496839e-01 9.97096419e-01
-3.68411869e-01 6.23632610e-01 4.72851545e-01 -5.47748208e-01
-2.89999574e-01 -3.94764811e-01 1.86133549e-01 4.88051206e-01
2.60878891e-01 -5.88734925e-01 -8.80736172e-01 -4.26772833e-01
-6.12096116e-02 -7.36829638e-01 9.79346454e-01 1.53665200e-01
8.69043112e-01 2.15226501e-01 -8.04537833e-02 6.16278946e-01
9.82529104e-01 -5.29423594e-01 4.01429743e-01 4.77289259e-01
1.03552079e+00 6.63030505e-01 7.18674600e-01 5.41590869e-01
2.30027437e-01 1.14038455e+00 3.60280007e-01 5.28081469e-02
-2.21644223e-01 -2.79713601e-01 5.86135149e-01 7.68051326e-01
-6.77463293e-01 2.58926630e-01 -7.59566247e-01 3.55644882e-01
-1.83264256e+00 -9.95116353e-01 1.03820086e-01 2.12731266e+00
5.70844173e-01 2.07272604e-01 5.72223842e-01 5.08133113e-01
1.70980796e-01 2.43861705e-01 -5.77459276e-01 3.07655543e-01
-1.98807102e-02 2.73491293e-01 5.66218197e-01 7.84856528e-02
-1.25311017e+00 7.41239846e-01 5.94206810e+00 4.01627272e-01
-7.81886458e-01 2.20482066e-01 4.71124761e-02 -8.13710749e-01
2.95722425e-01 -1.92819312e-01 -3.65738541e-01 -9.34361964e-02
6.44410312e-01 5.51427364e-01 8.70033354e-02 9.19402421e-01
1.06203340e-01 -2.25217193e-01 -1.58948171e+00 1.38775051e+00
5.67571640e-01 -7.16331780e-01 -2.81749517e-02 2.58914590e-01
5.85321724e-01 1.04382992e-01 -9.30960849e-02 9.65985656e-02
-7.95438662e-02 -7.47359037e-01 6.10236704e-01 5.13625741e-01
7.15636373e-01 -6.79629087e-01 4.13624495e-01 2.25649968e-01
-1.12675822e+00 -4.17641699e-02 -1.22159570e-01 -1.89387828e-01
3.24017793e-01 2.18294039e-01 -2.48513341e-01 8.70366573e-01
9.93581533e-01 1.30693531e+00 -9.39342201e-01 5.25785148e-01
-3.32501203e-01 4.42013890e-02 -2.19008580e-01 5.19152343e-01
-8.69884863e-02 1.01314038e-01 5.69988549e-01 5.82555175e-01
9.29711238e-02 1.36023745e-01 2.29148835e-01 2.36317009e-01
2.32531950e-01 -2.15261042e-01 -6.67364299e-01 1.91695526e-01
-4.81592119e-01 1.09343410e+00 -7.38408923e-01 -1.17423229e-01
-4.60162669e-01 1.41348076e+00 3.72902423e-01 3.98766398e-02
-9.16597664e-01 2.27558002e-01 7.56353438e-01 1.46351144e-01
4.98369455e-01 -7.07045734e-01 -3.97934131e-02 -1.45356774e+00
3.20694327e-01 -1.22554588e+00 6.02751076e-01 -8.42799842e-01
-1.25676382e+00 1.29651576e-01 3.07315737e-01 -1.70553827e+00
-4.91570532e-01 -1.03475177e+00 1.18120000e-01 2.91518182e-01
-9.96092260e-01 -1.25004578e+00 -4.32016730e-01 1.01261485e+00
5.11311471e-01 -2.77182460e-01 9.03453767e-01 2.01618299e-01
-2.12506518e-01 5.16444325e-01 -2.36795232e-01 3.28374237e-01
9.87815261e-01 -1.14650166e+00 1.89508587e-01 7.17467546e-01
6.54248416e-01 4.81533051e-01 5.10955095e-01 -4.27702844e-01
-1.65424323e+00 -6.48219824e-01 3.14269602e-01 -1.32520521e+00
4.35208470e-01 -4.22826231e-01 -2.30126187e-01 8.26532245e-01
-3.37943345e-01 3.46438915e-01 9.48769391e-01 3.05294216e-01
-6.75632536e-01 -1.42343953e-01 -8.46075177e-01 4.54792023e-01
1.74568832e+00 -6.75838709e-01 -7.89254010e-01 3.38319540e-01
3.40459853e-01 -5.27878881e-01 -1.14032960e+00 5.71543396e-01
1.15545213e+00 -1.23697519e+00 1.18934298e+00 -1.06873369e+00
5.29936254e-01 -3.30300391e-01 -7.14907169e-01 -1.16229451e+00
-1.39948651e-01 -2.46715188e-01 -9.64016691e-02 7.42871881e-01
8.93257186e-02 -9.40704942e-02 1.05513465e+00 2.61139780e-01
-4.30446155e-02 -6.70521498e-01 -1.32724798e+00 -1.05056548e+00
-1.73039302e-01 -6.51385546e-01 -1.57837838e-01 6.91007674e-01
9.56507474e-02 4.30969447e-01 -5.66792428e-01 2.64112093e-02
6.42890930e-01 1.12961270e-01 1.21071661e+00 -1.07907438e+00
-6.04620159e-01 -3.17247733e-02 -1.16229200e+00 -1.10285401e+00
1.51691452e-01 -8.05940151e-01 -9.80021581e-02 -1.33675861e+00
3.07551414e-01 3.07828724e-01 -3.16638887e-01 4.42493558e-01
8.72740299e-02 5.01877904e-01 4.23775166e-01 2.35283986e-01
-8.95732820e-01 7.30235875e-01 1.39050615e+00 -5.43924719e-02
1.49962649e-01 -6.32788911e-02 -2.75614530e-01 9.47131872e-01
4.16762769e-01 -4.14262444e-01 -5.90747774e-01 -3.60578001e-01
1.99140891e-01 2.04970967e-02 6.41858459e-01 -1.36923790e+00
-2.28479445e-01 2.07850430e-02 7.65187323e-01 -5.51768124e-01
7.89726734e-01 -9.16989982e-01 -1.01319417e-01 4.71487314e-01
-3.58561575e-01 2.22428024e-01 -1.47508336e-02 8.38174343e-01
6.65567890e-02 4.09852862e-01 5.82990110e-01 -5.45252800e-01
-6.77291632e-01 3.48782390e-01 -1.84789523e-02 4.31174397e-01
8.84827137e-01 -4.03546631e-01 -2.92457063e-02 -4.51191455e-01
-1.03349042e+00 -1.86789706e-01 7.07828164e-01 6.16322041e-01
5.66578627e-01 -1.53064656e+00 -4.42090988e-01 2.68981129e-01
4.34792429e-01 -2.46099666e-01 2.37325445e-01 1.16366470e+00
-1.97810382e-01 4.34175491e-01 -8.00311983e-01 -1.00187600e+00
-1.51873028e+00 4.61098731e-01 3.65296751e-01 -4.02337223e-01
-6.87976420e-01 6.31682277e-01 8.77858028e-02 -5.44607818e-01
1.46555915e-01 -3.54843318e-01 5.89421354e-02 1.58605099e-01
2.95289695e-01 6.04317307e-01 1.94922477e-01 -1.00488973e+00
-6.50165260e-01 1.04859400e+00 2.85848141e-01 -4.33013976e-01
1.33124208e+00 8.34549684e-03 3.52813095e-01 7.97357678e-01
1.42170739e+00 -2.07531512e-01 -1.45943570e+00 -1.20568439e-01
-3.40750009e-01 -7.37716794e-01 -1.87648341e-01 -6.06297255e-01
-1.19914377e+00 9.67017055e-01 9.62138176e-01 -5.38971305e-01
1.00816476e+00 1.04608491e-01 5.50253153e-01 5.26767194e-01
6.53037727e-01 -1.26992512e+00 7.07326174e-01 2.71007627e-01
1.08236825e+00 -1.47819996e+00 6.63281560e-01 -3.68321300e-01
-7.64782786e-01 9.53692019e-01 6.03596270e-01 -1.89172134e-01
5.91706812e-01 -3.76378119e-01 2.43423767e-02 -4.11502272e-01
-4.38598096e-01 -4.54202235e-01 5.57500124e-01 7.89703369e-01
4.55370963e-01 -1.04844637e-01 -1.11851573e-01 3.29705834e-01
-2.16481879e-01 -1.42317429e-01 1.91508889e-01 1.17282164e+00
-5.44647090e-02 -1.03085434e+00 -1.29332185e-01 1.83906004e-01
-4.05448258e-01 4.20918107e-01 -6.41008139e-01 1.17485392e+00
4.15748321e-02 7.57090151e-01 -1.85333386e-01 -6.97096109e-01
8.25194597e-01 2.72168130e-01 1.28083694e+00 -5.21597028e-01
-2.45891094e-01 1.35977462e-01 2.36389309e-01 -1.36115885e+00
-1.11824691e+00 -1.09406281e+00 -1.12031531e+00 8.76255408e-02
2.70367763e-03 -6.38513386e-01 3.02807242e-01 1.14824831e+00
2.65719891e-01 5.38528502e-01 3.28471065e-01 -1.18873465e+00
-6.50304615e-01 -8.52810800e-01 -6.67567492e-01 9.09444630e-01
2.16695234e-01 -1.28354526e+00 -2.73910403e-01 8.96658525e-02] | [7.1785149574279785, -0.6285148859024048] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.