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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
053f2478-4d2d-47df-9af8-522f04e4ea74 | curriculum-sampling-for-dense-retrieval-with | 2212.09114 | null | https://arxiv.org/abs/2212.09114v1 | https://arxiv.org/pdf/2212.09114v1.pdf | Curriculum Sampling for Dense Retrieval with Document Expansion | The dual-encoder has become the de facto architecture for dense retrieval. Typically, it computes the latent representations of the query and document independently, thus failing to fully capture the interactions between the query and document. To alleviate this, recent work expects to get query-informed representations of documents. During training, it expands the document with a real query, while replacing the real query with a generated pseudo query at inference. This discrepancy between training and inference makes the dense retrieval model pay more attention to the query information but ignore the document when computing the document representation. As a result, it even performs worse than the vanilla dense retrieval model, since its performance depends heavily on the relevance between the generated queries and the real query. In this paper, we propose a curriculum sampling strategy, which also resorts to the pseudo query at training and gradually increases the relevance of the generated query to the real query. In this way, the retrieval model can learn to extend its attention from the document only to both the document and query, hence getting high-quality query-informed document representations. Experimental results on several passage retrieval datasets show that our approach outperforms the previous dense retrieval methods1. | ['Nan Duan', 'Siu Ming Yiu', 'Jian Jiao', 'Anlei Dong', 'Hang Zhang', 'A-Long Jin', 'Yeyun Gong', 'Xingwei He'] | 2022-12-18 | null | null | null | null | ['passage-retrieval'] | ['natural-language-processing'] | [-5.23712253e-03 4.12466489e-02 -3.91615272e-01 -2.13855371e-01
-1.14002621e+00 -6.54070199e-01 9.67642307e-01 4.66259532e-02
-5.05717278e-01 6.50675237e-01 4.78933364e-01 -9.59908292e-02
-4.20523956e-02 -1.04304397e+00 -8.15920174e-01 -7.55647540e-01
3.13585311e-01 1.03443372e+00 2.20959485e-01 -1.83921367e-01
1.15551837e-01 1.04729339e-01 -1.52705133e+00 2.02194318e-01
6.71654463e-01 6.45370960e-01 6.39243543e-01 6.02009892e-01
-2.20178410e-01 5.69931865e-01 -7.74274707e-01 -3.36603791e-01
8.81662369e-02 -3.52050394e-01 -8.39005470e-01 -1.13063201e-01
4.16128904e-01 -8.62412095e-01 -7.21642494e-01 8.24629724e-01
3.63975555e-01 3.03656608e-01 5.79824030e-01 -7.38908291e-01
-9.60327327e-01 7.18208194e-01 -3.42259735e-01 1.85675859e-01
3.97304684e-01 -2.69361705e-01 1.46368492e+00 -9.25791085e-01
8.43916059e-01 1.19800055e+00 -1.04509965e-01 5.15290141e-01
-1.20984685e+00 -5.12005329e-01 4.57112610e-01 2.12635830e-01
-1.43549967e+00 -3.48404825e-01 6.31196082e-01 -1.29662991e-01
9.89211917e-01 2.62898117e-01 7.03831434e-01 1.18699169e+00
-1.49132922e-01 1.12126577e+00 3.71164411e-01 -5.60475349e-01
3.15803856e-01 3.22317600e-01 2.23569810e-01 3.88504952e-01
7.00128824e-02 2.58756220e-01 -3.15643251e-01 -3.87091070e-01
7.43789911e-01 3.27577293e-01 -5.82279623e-01 -3.36852312e-01
-8.85082424e-01 1.01578546e+00 4.60593551e-01 4.57679480e-01
-5.88018119e-01 3.16437930e-01 2.55030364e-01 3.92609656e-01
2.83253223e-01 3.66058946e-01 -3.90026927e-01 1.05426824e-02
-1.00607634e+00 5.65592885e-01 7.55670428e-01 1.02804363e+00
9.08640623e-01 -2.34166861e-01 -5.72556078e-01 8.77407789e-01
4.97933209e-01 4.67020005e-01 5.42534053e-01 -6.50341213e-01
5.62078774e-01 3.32307577e-01 2.37677008e-01 -7.53385603e-01
3.03991616e-01 -7.75198817e-01 -6.19412303e-01 -4.04546887e-01
1.03884771e-01 3.92171890e-02 -1.02370632e+00 1.89266658e+00
2.05074713e-01 8.45646709e-02 1.95330173e-01 9.39329267e-01
5.03871262e-01 9.52379346e-01 8.63061845e-02 -1.57076582e-01
1.17592323e+00 -9.75069582e-01 -7.41047084e-01 -2.68693238e-01
8.56229842e-01 -7.35462666e-01 9.37487483e-01 1.62285954e-01
-1.19412112e+00 -7.05829084e-01 -8.96037102e-01 -3.69923174e-01
-2.17032477e-01 1.36832699e-01 5.87319493e-01 5.34025617e-02
-1.03799736e+00 2.93189943e-01 -7.89566576e-01 -1.02759004e-01
1.37485251e-01 1.78632140e-01 -1.09163538e-01 -8.01276743e-01
-1.60320401e+00 6.77157044e-01 4.69038248e-01 3.41230221e-02
-1.08785701e+00 -6.97298706e-01 -8.36267292e-01 6.07976735e-01
4.42502677e-01 -9.67087090e-01 1.40266180e+00 -7.31216729e-01
-1.17718804e+00 4.24389541e-01 -4.30808961e-01 -3.34170520e-01
1.46180287e-01 -4.09961909e-01 5.03968389e-04 3.03453356e-01
3.88423279e-02 8.02868187e-01 8.78973484e-01 -1.25965881e+00
-4.21351552e-01 -3.16945016e-01 4.85361487e-01 3.68881673e-01
-1.19661979e-01 -3.15173507e-01 -1.07945395e+00 -4.79588181e-01
4.00379263e-02 -8.40678871e-01 -1.72519274e-02 -1.48924455e-01
-2.78627932e-01 -5.11792183e-01 6.96002722e-01 -1.75782651e-01
1.19465327e+00 -2.34017777e+00 3.31610858e-01 1.08897291e-01
2.48084292e-01 2.03799799e-01 -4.53830332e-01 7.65804887e-01
3.94881107e-02 -8.87353048e-02 8.44129175e-02 -6.48528934e-01
7.48757794e-02 5.90183437e-01 -8.02123904e-01 2.93206155e-01
1.73506737e-02 1.14649785e+00 -1.10258150e+00 -3.86856228e-01
-1.19600475e-01 7.37459719e-01 -9.63603497e-01 4.74446982e-01
-5.28121889e-01 1.50460955e-02 -8.15507293e-01 8.97245705e-02
5.30922115e-01 -5.38235128e-01 8.32376704e-02 -9.43111032e-02
3.45587313e-01 5.28705418e-01 -9.18821275e-01 1.92741787e+00
-6.09740019e-01 5.45046270e-01 -7.59989470e-02 -8.02659631e-01
7.44305909e-01 6.44236863e-01 2.49960542e-01 -9.76104915e-01
-2.52519846e-01 -2.02465504e-02 -1.95330247e-01 -4.20914471e-01
6.53177023e-01 -1.88825309e-01 4.56334054e-02 6.47967696e-01
4.87522669e-02 -8.96967128e-02 3.52248639e-01 8.25426161e-01
7.89212763e-01 1.33819908e-01 -3.21059585e-01 1.91077694e-01
3.67560714e-01 -3.57968003e-01 3.51932257e-01 1.07346916e+00
3.69978875e-01 6.10083520e-01 4.88155305e-01 -1.23756684e-01
-8.94971192e-01 -1.05029917e+00 9.88208428e-02 1.25223482e+00
1.10394269e-01 -5.49575150e-01 -5.04724503e-01 -6.06644928e-01
-9.18600522e-03 7.83340573e-01 -6.82692647e-01 -4.84184712e-01
-6.35661244e-01 -2.62660116e-01 1.23627573e-01 4.96203870e-01
2.59583652e-01 -9.18740511e-01 -3.66964489e-01 3.31889123e-01
-3.08205098e-01 -7.79570222e-01 -7.72164285e-01 1.58002123e-01
-7.99018741e-01 -8.62028599e-01 -1.03139663e+00 -6.82013690e-01
6.65696442e-01 2.72679538e-01 1.15442359e+00 4.70931977e-01
4.08011628e-03 5.56580245e-01 -4.82696086e-01 3.66019718e-02
-2.19008297e-01 1.90997466e-01 -4.67181772e-01 -9.58475173e-02
4.12637800e-01 -3.10209960e-01 -7.03602552e-01 -3.21566164e-02
-1.40588820e+00 -9.99089926e-02 6.95770681e-01 1.10820055e+00
5.04659891e-01 2.57115532e-02 6.07126892e-01 -9.21121240e-01
7.28045106e-01 -6.53711081e-01 -5.73145986e-01 3.10567588e-01
-4.94662017e-01 4.41705197e-01 4.59717542e-01 -6.32957697e-01
-1.00309932e+00 -3.92993987e-01 -1.48037925e-01 -6.34781003e-01
1.16251856e-01 7.64447212e-01 -6.26643151e-02 7.33473003e-01
3.20229560e-01 4.87657309e-01 -4.21734124e-01 -7.32182801e-01
3.95300567e-01 3.95806968e-01 2.67263263e-01 -7.05185056e-01
6.53995812e-01 1.91276193e-01 -4.45288390e-01 -6.74043834e-01
-8.80721629e-01 -6.46021962e-01 -2.91369885e-01 2.36145794e-01
6.43747091e-01 -8.94981265e-01 -2.37345129e-01 4.23762910e-02
-1.33592427e+00 -1.90271795e-01 -5.15974939e-01 6.93619192e-01
-1.78591698e-01 3.07839423e-01 -6.66390002e-01 -6.32865846e-01
-1.32476851e-01 -1.36210167e+00 1.55127609e+00 -1.19919900e-03
-6.58336058e-02 -1.04284382e+00 2.61107951e-01 4.79519479e-02
5.45814455e-01 -4.24795508e-01 1.28990936e+00 -7.25649834e-01
-9.96681213e-01 -6.50677383e-01 -3.24914694e-01 2.12409586e-01
-5.56515716e-02 -2.03958660e-01 -9.63297486e-01 -5.24910033e-01
4.45825942e-02 -5.48933923e-01 1.15891218e+00 1.50868624e-01
9.82377768e-01 -3.40829104e-01 -3.75992030e-01 3.76325279e-01
1.43739140e+00 7.35579580e-02 6.62879407e-01 1.57158710e-02
3.98312211e-01 3.36748868e-01 3.57907146e-01 2.83897281e-01
3.45743150e-01 6.43418610e-01 3.82548869e-01 9.37365219e-02
-4.33694571e-02 -5.79873323e-01 1.73712999e-01 7.57344782e-01
4.81859624e-01 -6.14183605e-01 -4.59262013e-01 5.92469156e-01
-1.65547836e+00 -9.21619356e-01 4.04266596e-01 2.24423742e+00
9.61656630e-01 1.02239966e-01 -4.85964566e-01 -3.30781192e-02
2.51080543e-01 2.81620324e-01 -4.92169470e-01 -8.51220787e-02
1.79891601e-01 2.74067432e-01 -1.13964580e-01 8.20589125e-01
-6.13765895e-01 1.04116392e+00 6.35036087e+00 6.99735463e-01
-1.11313903e+00 -1.46365806e-01 2.50995010e-01 -3.86963278e-01
-9.20243621e-01 1.31546661e-01 -9.90635097e-01 2.62000471e-01
9.01161492e-01 -2.41196036e-01 3.19670022e-01 6.00035012e-01
-1.19949847e-01 -1.47244200e-01 -1.56640232e+00 5.89974761e-01
-3.18874419e-02 -1.06203210e+00 6.56767786e-01 2.65319467e-01
5.11183977e-01 -5.43228500e-02 8.78967047e-02 9.26464796e-01
7.55353794e-02 -7.74630129e-01 4.81867641e-01 7.03975201e-01
6.08173966e-01 -5.42184889e-01 7.05793560e-01 6.31949365e-01
-8.98107886e-01 9.98812392e-02 -5.27106822e-01 6.44650310e-02
2.38622986e-02 4.42628115e-01 -7.46923566e-01 3.36880833e-01
1.97336495e-01 4.47657317e-01 -3.26352000e-01 9.09158230e-01
-2.80513972e-01 4.67011750e-01 -2.72582233e-01 4.40597460e-02
4.74905759e-01 -6.31236061e-02 3.11209857e-01 1.03724039e+00
1.75759837e-01 6.73875958e-02 5.16514242e-01 9.17532444e-01
-3.16099912e-01 7.79572353e-02 -5.70090353e-01 -2.48790085e-01
5.03621101e-01 8.19320381e-01 -1.25928223e-01 -7.97898412e-01
-4.16726321e-01 9.59549904e-01 5.53175449e-01 1.05630267e+00
-4.82554734e-01 -3.71881455e-01 3.95739317e-01 3.97977903e-02
6.34260714e-01 -3.38779986e-02 4.74666566e-01 -1.14390147e+00
1.61499515e-01 -6.96861148e-01 3.22570503e-01 -8.73652160e-01
-9.36193407e-01 5.49869478e-01 2.78658301e-01 -8.04260433e-01
-7.99960196e-01 -5.22400476e-02 -2.55783409e-01 1.38622785e+00
-1.88888466e+00 -8.08746755e-01 1.81831792e-01 4.97780561e-01
6.15238607e-01 1.83140174e-01 1.02774489e+00 4.05204684e-01
-2.24989235e-01 5.22026837e-01 1.39161006e-01 2.85614282e-01
6.61139727e-01 -1.19659662e+00 3.91603559e-02 3.56164426e-01
2.54256159e-01 1.09181082e+00 5.89091778e-01 -4.89188522e-01
-1.55220342e+00 -7.45481312e-01 9.93121982e-01 -2.34651878e-01
4.59262401e-01 -3.67974848e-01 -1.40213048e+00 7.50545979e-01
3.18505973e-01 -7.47929811e-02 7.88867891e-01 2.02234760e-01
-5.84594727e-01 -5.07299453e-02 -7.13809788e-01 5.57452977e-01
3.72986883e-01 -8.74052763e-01 -9.18408096e-01 2.88865298e-01
1.03634322e+00 -1.94993168e-01 -5.46711862e-01 9.24687237e-02
5.58048427e-01 -6.78594947e-01 9.90057468e-01 -5.92360020e-01
5.02827406e-01 -7.59000257e-02 -2.12665975e-01 -1.17030239e+00
-3.14334363e-01 -9.82104018e-02 -6.29844248e-01 8.00090671e-01
4.02238131e-01 -3.74924690e-01 7.24233747e-01 5.81671715e-01
-6.29559113e-03 -8.75935376e-01 -6.77314162e-01 -4.22313243e-01
3.37345898e-01 -3.63682568e-01 6.18711948e-01 6.00720406e-01
-2.30831221e-01 7.09503829e-01 -2.02176362e-01 1.72360152e-01
4.09864008e-01 4.76103574e-01 5.92344761e-01 -1.08484507e+00
-7.58051097e-01 -3.27905208e-01 9.09809023e-02 -1.90333676e+00
2.30033293e-01 -9.22486007e-01 6.92565814e-02 -1.66209126e+00
2.69295812e-01 -3.43296766e-01 -2.86720783e-01 2.87525296e-01
-4.57901239e-01 -1.00991555e-01 -7.58582447e-03 2.82231480e-01
-6.18331254e-01 8.07234168e-01 1.72643018e+00 -3.40415448e-01
-1.48309499e-01 2.16911617e-03 -7.39172935e-01 1.96054950e-01
4.34435964e-01 -5.78894496e-01 -9.71101165e-01 -9.42242920e-01
4.39281255e-01 3.16787630e-01 2.89361984e-01 -4.02013779e-01
4.94224280e-01 1.02171831e-01 4.65065598e-01 -7.99968421e-01
6.05068386e-01 -8.14630687e-01 -2.02628955e-01 2.42914140e-01
-8.41617644e-01 -7.01563060e-02 5.25377318e-02 8.14930141e-01
-4.64141905e-01 -5.30671120e-01 4.07199472e-01 -3.36601883e-01
-1.69737771e-01 4.56226021e-01 -3.26426238e-01 7.70978779e-02
3.77261996e-01 1.50210947e-01 -1.12845115e-01 -5.94872296e-01
-7.55393267e-01 5.21419168e-01 2.25299239e-01 3.28273326e-01
6.82059228e-01 -1.17553151e+00 -6.35193408e-01 4.33262020e-01
1.07022405e-01 3.55406404e-01 1.65767163e-01 3.85324806e-01
1.18398458e-01 9.17608380e-01 5.73227823e-01 -4.96421367e-01
-8.35842371e-01 8.34691465e-01 1.28565371e-01 -6.53319359e-01
-6.28119409e-01 7.39639282e-01 4.68072355e-01 -1.02324270e-01
5.97118318e-01 -3.36646914e-01 -1.98630214e-01 1.21864736e-01
8.02279115e-01 -1.12584889e-01 -6.12764210e-02 -2.58832425e-01
2.41577253e-01 3.99539053e-01 -7.70291448e-01 -3.10696095e-01
1.16995263e+00 -1.29566014e-01 -9.12902318e-03 4.89778757e-01
1.75780141e+00 -1.61603037e-02 -9.31402028e-01 -6.79360867e-01
-3.42889018e-02 -6.31147623e-01 3.20202887e-01 -5.80016971e-01
-1.04363000e+00 9.35969055e-01 2.03438714e-01 2.93401182e-01
9.20323133e-01 1.98721096e-01 8.29170287e-01 8.03987026e-01
2.59859651e-01 -7.81098723e-01 1.94066942e-01 6.34082615e-01
8.36383224e-01 -1.03639567e+00 -8.10335279e-02 -3.09680775e-03
-3.20594937e-01 8.94652009e-01 4.56637174e-01 -2.35962942e-01
6.27351701e-01 -1.26394063e-01 -7.91485161e-02 -3.35228443e-01
-1.22744644e+00 -9.66861546e-02 3.49933743e-01 2.29213193e-01
4.97053504e-01 -2.43057653e-01 5.29116914e-02 1.22379296e-01
-3.79932374e-02 1.17787207e-02 1.66018426e-01 1.04412627e+00
-4.12595779e-01 -1.37088573e+00 -1.77184477e-01 4.44567084e-01
-2.88184732e-01 -3.82751435e-01 -2.51310468e-01 6.92820489e-01
-3.36073786e-01 6.15823209e-01 2.28934795e-01 1.23349242e-01
3.24737161e-01 2.93022215e-01 3.40383589e-01 -9.53072906e-01
-2.97863096e-01 4.93085712e-01 -2.01409101e-01 -4.55020905e-01
-8.94572660e-02 -3.00492465e-01 -1.16378832e+00 1.30609553e-02
-5.39072812e-01 8.47235203e-01 5.66355705e-01 1.02527106e+00
3.04600298e-01 5.66483080e-01 5.93819916e-01 -6.40125453e-01
-9.34295833e-01 -9.91407633e-01 -4.66532499e-01 3.68293524e-01
6.85748637e-01 -4.59255636e-01 -3.57263803e-01 -2.73234814e-01] | [11.43510627746582, 7.667717456817627] |
d6769cf3-eddc-4532-bef5-50fd0cf281c2 | pose-estimation-of-specific-rigid-objects | 2112.15075 | null | https://arxiv.org/abs/2112.15075v1 | https://arxiv.org/pdf/2112.15075v1.pdf | Pose Estimation of Specific Rigid Objects | In this thesis, we address the problem of estimating the 6D pose of rigid objects from a single RGB or RGB-D input image, assuming that 3D models of the objects are available. This problem is of great importance to many application fields such as robotic manipulation, augmented reality, and autonomous driving. First, we propose EPOS, a method for 6D object pose estimation from an RGB image. The key idea is to represent an object by compact surface fragments and predict the probability distribution of corresponding fragments at each pixel of the input image by a neural network. Each pixel is linked with a data-dependent number of fragments, which allows systematic handling of symmetries, and the 6D poses are estimated from the links by a RANSAC-based fitting method. EPOS outperformed all RGB and most RGB-D and D methods on several standard datasets. Second, we present HashMatch, an RGB-D method that slides a window over the input image and searches for a match against templates, which are pre-generated by rendering 3D object models in different orientations. The method applies a cascade of evaluation stages to each window location, which avoids exhaustive matching against all templates. Third, we propose ObjectSynth, an approach to synthesize photorealistic images of 3D object models for training methods based on neural networks. The images yield substantial improvements compared to commonly used images of objects rendered on top of random photographs. Fourth, we introduce T-LESS, the first dataset for 6D object pose estimation that includes 3D models and RGB-D images of industry-relevant objects. Fifth, we define BOP, a benchmark that captures the status quo in the field. BOP comprises eleven datasets in a unified format, an evaluation methodology, an online evaluation system, and public challenges held at international workshops organized at the ICCV and ECCV conferences. | ['Tomas Hodan'] | 2021-12-30 | null | null | null | null | ['6d-pose-estimation'] | ['computer-vision'] | [ 3.60940039e-01 1.13058552e-01 8.26821551e-02 -4.09130484e-01
-6.27113998e-01 -6.11947000e-01 5.03523052e-01 -4.83834267e-01
-9.42188799e-02 1.04274906e-01 -3.14833164e-01 -1.63299888e-01
-2.46004105e-01 -8.50991249e-01 -1.24793303e+00 -4.49472785e-01
1.05943643e-01 1.01341772e+00 4.09690261e-01 -2.43325368e-01
3.41694504e-01 1.38277590e+00 -1.89228702e+00 2.05613866e-01
9.13223177e-02 1.46953261e+00 2.48112097e-01 5.77004552e-01
2.21814238e-03 1.68365687e-01 -4.82058108e-01 -5.16846299e-01
7.71653712e-01 -2.44997907e-02 -5.91888130e-01 3.13150644e-01
9.19641435e-01 -3.12520862e-01 -5.00291288e-01 6.88847542e-01
4.51914907e-01 -2.36148685e-02 7.91480482e-01 -1.59017205e+00
-2.91545749e-01 -4.46117446e-02 -4.85860944e-01 -4.01779979e-01
6.14627898e-01 1.04713909e-01 5.27613342e-01 -1.12279630e+00
9.75713134e-01 1.48241651e+00 7.08193779e-01 5.35515726e-01
-1.18486965e+00 -4.62286085e-01 -2.17108324e-01 1.21964082e-01
-1.43973577e+00 -2.11380422e-02 1.06434464e+00 -4.07175094e-01
8.84962797e-01 5.33966541e-01 7.17633724e-01 8.24555814e-01
3.36640358e-01 6.00403070e-01 1.10590219e+00 -6.07059658e-01
1.09788626e-01 -2.27798205e-02 -1.41581520e-01 6.79181576e-01
-7.74665624e-02 2.16041282e-01 -6.82878733e-01 -2.46469170e-01
1.08000469e+00 -1.71398651e-02 -6.16055503e-02 -1.02470374e+00
-1.22430563e+00 4.00991827e-01 6.10491157e-01 -3.02509815e-01
-4.17443484e-01 1.43030673e-01 -3.52288969e-02 -3.58298346e-02
2.34009936e-01 2.57342875e-01 -4.54115093e-01 1.67034268e-01
-5.64564347e-01 3.58140796e-01 7.40287542e-01 1.17848468e+00
8.95902336e-01 -2.22834900e-01 2.30188638e-01 5.49587369e-01
5.59042156e-01 6.57608747e-01 1.24811687e-01 -1.38440466e+00
5.11312723e-01 8.50302339e-01 2.64773011e-01 -1.27460337e+00
-3.82509291e-01 7.70356730e-02 -6.86556995e-01 8.34789455e-01
1.90160394e-01 4.22393441e-01 -1.08664608e+00 1.17201328e+00
8.42450917e-01 -1.84889227e-01 6.52990341e-02 9.16049659e-01
9.05009687e-01 6.53313756e-01 -6.29556537e-01 1.65192217e-01
1.24946189e+00 -6.50763631e-01 -2.94294953e-01 -1.23456202e-01
3.18889581e-02 -1.00044882e+00 8.74459684e-01 6.78786457e-01
-1.29968047e+00 -7.99423277e-01 -1.08231544e+00 -2.04939321e-01
-5.19345343e-01 2.56719559e-01 4.08327579e-01 6.19045615e-01
-9.58760798e-01 7.04924405e-01 -7.35737264e-01 -8.10767859e-02
2.01949626e-01 7.01974332e-01 -6.39541805e-01 -1.07273899e-01
-6.60917819e-01 1.08850229e+00 4.93316323e-01 4.63233382e-01
-6.25789225e-01 -4.93636757e-01 -8.91439080e-01 -5.12691617e-01
2.24772930e-01 -5.25216937e-01 1.04039991e+00 -5.93694329e-01
-1.67008996e+00 1.20032048e+00 -3.66608761e-02 -1.61201298e-01
6.13831460e-01 -1.10524110e-01 -1.11784644e-01 1.64680630e-01
-2.73665369e-01 9.84154403e-01 9.22052681e-01 -1.69798779e+00
-4.50280398e-01 -6.94813013e-01 1.56883642e-01 3.71157646e-01
3.24296087e-01 -1.24422908e-01 -9.45807099e-01 -2.72670299e-01
8.37830842e-01 -1.20889735e+00 -3.03798765e-01 4.59068149e-01
-5.58113098e-01 -1.79262102e-01 1.22778130e+00 -5.51099300e-01
2.87084788e-01 -2.18514228e+00 2.64169931e-01 5.33628345e-01
-1.29550859e-01 1.05587639e-01 -1.32341877e-01 2.19643205e-01
-1.98952600e-01 -1.16915591e-01 1.20710768e-01 -5.58106363e-01
2.78907716e-01 2.69385546e-01 -3.01999211e-01 5.99424243e-01
6.84402734e-02 6.97821140e-01 -2.60155737e-01 -4.87253428e-01
6.27904534e-01 7.07620680e-01 -2.40549207e-01 2.81397760e-01
-4.16983604e-01 1.83549523e-01 -2.50279069e-01 6.83605850e-01
1.13613486e+00 -1.03306755e-01 -1.63208485e-01 -6.90359473e-01
-2.99413484e-02 -3.33553515e-02 -1.70042527e+00 1.77748740e+00
-2.71902859e-01 4.09527600e-01 -2.41249189e-01 -6.76943302e-01
1.35962021e+00 1.40122309e-01 5.20639837e-01 -5.26103973e-01
1.35239467e-01 2.95449197e-01 -5.47645986e-01 -3.46435755e-01
7.34005511e-01 2.95996547e-01 -9.84032899e-02 3.70752603e-01
-1.13440201e-01 -1.04883718e+00 -1.07651025e-01 -8.07398334e-02
6.59009278e-01 7.08480299e-01 -6.64368793e-02 2.92984903e-01
3.84245604e-01 2.50200689e-01 2.34863386e-01 5.01213312e-01
2.54156083e-01 1.23501956e+00 2.92779446e-01 -8.25197935e-01
-1.27044368e+00 -1.10656381e+00 -9.60507095e-02 3.36176395e-01
5.79477072e-01 -8.63904953e-02 -6.22005939e-01 -3.62381637e-01
2.88617283e-01 5.27257085e-01 -6.55743003e-01 -3.56615125e-03
-7.79338956e-01 -4.47262138e-01 1.21993855e-01 4.55448180e-01
4.36231345e-01 -1.06999993e+00 -9.30579543e-01 -2.68296808e-01
4.31766808e-02 -1.17628849e+00 -1.73901711e-02 3.11869442e-01
-1.01247811e+00 -1.08933604e+00 -5.14184952e-01 -7.34111786e-01
8.42625856e-01 3.86219263e-01 1.19133592e+00 -1.35121644e-01
-3.57012063e-01 5.41962624e-01 -1.15105107e-01 -4.81854409e-01
-3.92886043e-01 -2.12428465e-01 9.54576433e-02 -1.63959816e-01
1.36057809e-01 -4.04012203e-01 -5.04125416e-01 9.90403712e-01
-7.46110201e-01 1.86571702e-01 5.05248964e-01 3.59816402e-01
1.11429703e+00 -3.77648473e-02 -3.99612874e-01 -1.56829104e-01
-8.84412825e-02 9.38869268e-02 -9.94744182e-01 1.69787377e-01
-1.84081383e-02 -9.02871564e-02 6.64184764e-02 -5.12863100e-01
-9.30730343e-01 7.85846949e-01 6.39041071e-04 -8.20887208e-01
-3.45028669e-01 -1.05132654e-01 -4.00901496e-01 -4.26577121e-01
7.62004018e-01 4.10403498e-02 1.43207446e-01 -4.70588773e-01
2.13787779e-01 6.66704535e-01 7.18097746e-01 -6.05520606e-01
1.16010869e+00 5.86359799e-01 4.33938652e-01 -7.49866128e-01
-5.59719145e-01 -3.81162345e-01 -9.80362654e-01 -5.10138035e-01
7.06979394e-01 -7.21114099e-01 -9.22209799e-01 6.23395681e-01
-1.61078370e+00 -2.57652342e-01 -3.92008513e-01 5.77121735e-01
-9.27017570e-01 6.31044284e-02 -2.27232203e-01 -7.64541566e-01
-4.57175039e-02 -1.42948818e+00 1.65469253e+00 2.08511472e-01
-3.21510546e-02 -3.32319736e-01 -9.86456648e-02 5.62743008e-01
-3.16518173e-02 4.83078629e-01 6.32372558e-01 -2.44834095e-01
-1.05138779e+00 -4.81571317e-01 -1.49819359e-01 4.06348377e-01
-6.17809668e-02 2.25681901e-01 -8.99789095e-01 7.58119822e-02
-1.38195783e-01 -4.53032702e-01 1.10479891e-01 1.90929294e-01
1.29026127e+00 9.15711448e-02 -4.05633926e-01 6.85499609e-01
1.43413246e+00 2.39708498e-01 8.39458406e-01 4.45390671e-01
7.56597221e-01 7.17307270e-01 7.49173403e-01 1.65707365e-01
1.89384356e-01 1.07198763e+00 1.06268370e+00 -5.64078689e-02
-7.34332427e-02 -1.55194744e-01 1.42496794e-01 5.18810868e-01
-4.41090673e-01 -2.98909452e-02 -9.26365554e-01 -3.54666226e-02
-1.43495619e+00 -4.37281728e-01 -3.83368522e-01 2.33627009e+00
4.68639314e-01 2.35175699e-01 6.74716383e-03 3.48105431e-01
6.84029281e-01 -1.14047930e-01 -6.30577266e-01 -2.77846515e-01
-9.87626165e-02 2.61278450e-01 5.74649096e-01 2.63877600e-01
-8.59925151e-01 6.44666433e-01 6.26979303e+00 7.79680729e-01
-1.10821807e+00 -2.28712499e-01 5.28736413e-01 1.49514601e-01
-2.84672249e-02 -1.48743272e-01 -1.08353698e+00 4.23608981e-02
5.23943186e-01 3.16274732e-01 2.35108033e-01 1.14479017e+00
-1.01575866e-01 -2.74454445e-01 -1.18992043e+00 1.10771441e+00
4.55587119e-01 -1.34435093e+00 2.39448491e-02 1.96923628e-01
7.75819659e-01 1.17315361e-02 1.33517206e-01 -2.33475104e-01
1.13426074e-01 -8.43241990e-01 1.18444073e+00 7.28651106e-01
6.32932425e-01 -6.68689251e-01 6.24093771e-01 3.97638798e-01
-9.47592497e-01 2.27992281e-01 -5.74815691e-01 2.49526113e-01
1.93909958e-01 2.77444959e-01 -9.01267052e-01 6.81604326e-01
9.80636239e-01 1.39963701e-01 -4.58049417e-01 1.13069463e+00
-1.57619119e-01 -8.59522372e-02 -7.23575234e-01 4.79048565e-02
-1.54380947e-02 -1.43039227e-01 4.26732928e-01 6.47778332e-01
5.86775303e-01 -4.77595888e-02 -2.52117552e-02 7.96065152e-01
-1.97176114e-02 -1.60871759e-01 -6.88062668e-01 6.30246222e-01
2.89822549e-01 1.20084453e+00 -9.40623581e-01 -1.02862619e-01
-1.27959000e-02 9.89264846e-01 -1.23821266e-01 1.93401545e-01
-8.80560935e-01 -3.00283372e-01 1.99189678e-01 1.97287738e-01
4.10251409e-01 -4.53487933e-01 -5.84182777e-02 -4.51916993e-01
3.46138954e-01 -8.24992180e-01 -1.03243828e-01 -1.52041197e+00
-9.70202386e-01 6.91459596e-01 3.64234656e-01 -1.62568712e+00
-2.60426730e-01 -1.24110889e+00 -1.12439871e-01 9.11809862e-01
-1.18603373e+00 -1.27418137e+00 -7.28797376e-01 4.05288786e-01
4.77549732e-01 3.86360474e-02 7.71072984e-01 -1.76292062e-02
-1.41447425e-01 -1.85111386e-03 -8.01458880e-02 -1.25064611e-01
4.81123596e-01 -1.04018044e+00 6.56659901e-01 2.60402083e-01
3.26184034e-01 2.11108208e-01 5.82537055e-01 -4.69385803e-01
-1.80095017e+00 -9.30387020e-01 5.88610351e-01 -7.93028176e-01
1.43587574e-01 -4.90010113e-01 -6.44176304e-01 7.02864230e-01
-1.73907951e-01 5.03427267e-01 3.96302380e-02 -5.80753803e-01
-8.30984190e-02 -2.18176216e-01 -1.33128059e+00 3.17010820e-01
1.12027180e+00 -4.07153994e-01 -7.22933292e-01 5.28418303e-01
6.85467899e-01 -1.31222773e+00 -9.62343931e-01 5.56944966e-01
8.51806819e-01 -1.18379807e+00 1.48017466e+00 -1.70938954e-01
4.03049588e-01 -6.49337649e-01 -5.64624667e-01 -9.88583148e-01
8.81861746e-02 -5.59754670e-01 -1.13099307e-01 8.61888409e-01
2.12683216e-01 -2.52307504e-01 1.24644375e+00 8.13140690e-01
-1.32246152e-01 -6.85005009e-01 -1.08741415e+00 -8.54499578e-01
-3.14865619e-01 -6.49233222e-01 9.07109082e-01 4.79877084e-01
-8.45060349e-01 -2.94625998e-01 -1.41215041e-01 4.34406847e-01
6.42989576e-01 2.19347179e-01 1.46275020e+00 -1.26403511e+00
-7.89657906e-02 -1.23912543e-01 -9.59543824e-01 -1.15903687e+00
-2.21757079e-03 -4.97526944e-01 1.58266693e-01 -1.47356379e+00
-2.25708112e-01 -6.34780347e-01 2.72338361e-01 3.02795112e-01
4.48282897e-01 7.44406998e-01 1.87269971e-01 2.06557497e-01
-2.02447757e-01 3.07364225e-01 1.34770417e+00 -1.62145942e-01
-1.47032991e-01 2.25341186e-01 1.31209463e-01 1.09207928e+00
5.96070647e-01 -3.79727185e-01 -7.61468709e-02 -3.74696016e-01
3.96573514e-01 -4.13321443e-02 8.00397515e-01 -1.15323222e+00
2.17101276e-01 -1.96451917e-01 7.65052259e-01 -1.43950129e+00
1.03082609e+00 -1.28259027e+00 7.49703646e-01 2.84728229e-01
1.22495040e-01 3.41849804e-01 1.58089280e-01 1.86426789e-01
4.50451765e-03 -4.37293917e-01 5.38395584e-01 -3.44480902e-01
-6.48456752e-01 2.83066630e-01 1.98899835e-01 -5.56379318e-01
1.12247014e+00 -7.92357147e-01 4.38268408e-02 -6.27432019e-02
-7.06456542e-01 -2.12759316e-01 6.34697378e-01 4.19092268e-01
9.06496584e-01 -1.54692113e+00 -3.85871381e-01 4.06383276e-01
1.65357329e-02 7.30149806e-01 1.16418801e-01 3.75576735e-01
-8.97932827e-01 1.22124158e-01 -3.25873315e-01 -1.16154623e+00
-1.45300651e+00 4.29692090e-01 4.31336880e-01 1.79287240e-01
-5.08513987e-01 6.95487738e-01 -8.43252912e-02 -8.81480694e-01
4.90098983e-01 -4.34831887e-01 6.51097000e-02 -2.06341520e-01
1.66999146e-01 3.94927084e-01 4.31890905e-01 -1.07407773e+00
-2.30859041e-01 1.17506981e+00 3.50426227e-01 -1.57658711e-01
1.43794036e+00 1.34207726e-01 -4.55417894e-02 4.13930476e-01
1.21980131e+00 -1.45834297e-01 -1.41222489e+00 -9.58972052e-02
-3.38418543e-01 -6.89756572e-01 -1.70574173e-01 -6.31743908e-01
-1.12996626e+00 7.91647077e-01 7.65784562e-01 6.55456111e-02
8.48354280e-01 1.36187017e-01 4.24563915e-01 6.15160406e-01
6.00425661e-01 -8.70997548e-01 3.19085240e-01 4.06543463e-01
1.33452892e+00 -9.86381114e-01 2.28490993e-01 -6.34915650e-01
-2.94879258e-01 1.43937373e+00 7.03355372e-01 -1.17453426e-01
4.42233324e-01 2.68317610e-01 5.38864657e-02 -3.44655901e-01
-6.58648908e-02 2.01628163e-01 5.79635441e-01 7.56145954e-01
-2.55116522e-01 -1.95770964e-01 5.11165857e-01 3.02888956e-02
-4.81502593e-01 -1.50357515e-01 2.59358704e-01 1.04117465e+00
-7.99725801e-02 -1.17782319e+00 -1.09433770e+00 3.79201621e-02
-3.94851938e-02 4.94331092e-01 -3.94493997e-01 1.04525101e+00
4.30339098e-01 2.94689149e-01 2.72390932e-01 -4.75529641e-01
8.33123147e-01 -1.21616371e-01 9.40821648e-01 -2.78868437e-01
-2.80247003e-01 -1.45157218e-01 -1.08279228e-01 -7.52894819e-01
-7.30687678e-01 -5.76437235e-01 -9.64240015e-01 1.41095677e-02
-4.58827496e-01 -2.21761018e-01 1.33802819e+00 6.21920288e-01
3.52104455e-01 1.06192380e-01 7.62120247e-01 -1.84330463e+00
-4.66445506e-01 -5.30959785e-01 -3.46947134e-01 3.12818557e-01
1.51001871e-01 -9.61735070e-01 -3.38881195e-01 2.01976925e-01] | [7.527201175689697, -2.6418511867523193] |
d4cd935d-3409-4c1d-9ddc-0494492de411 | nonlinear-controllability-and-function | 2212.00896 | null | https://arxiv.org/abs/2212.00896v1 | https://arxiv.org/pdf/2212.00896v1.pdf | Nonlinear controllability and function representation by neural stochastic differential equations | There has been a great deal of recent interest in learning and approximation of functions that can be expressed as expectations of a given nonlinearity with respect to its random internal parameters. Examples of such representations include "infinitely wide" neural nets, where the underlying nonlinearity is given by the activation function of an individual neuron. In this paper, we bring this perspective to function representation by neural stochastic differential equations (SDEs). A neural SDE is an It\^o diffusion process whose drift and diffusion matrix are elements of some parametric families. We show that the ability of a neural SDE to realize nonlinear functions of its initial condition can be related to the problem of optimally steering a certain deterministic dynamical system between two given points in finite time. This auxiliary system is obtained by formally replacing the Brownian motion in the SDE by a deterministic control input. We derive upper and lower bounds on the minimum control effort needed to accomplish this steering; these bounds may be of independent interest in the context of motion planning and deterministic optimal control. | ['Maxim Raginsky', 'Tanya Veeravalli'] | 2022-12-01 | null | null | null | null | ['motion-planning'] | ['robots'] | [ 4.25845414e-01 5.08876801e-01 -7.34533370e-02 -2.95122355e-01
-9.78427753e-03 -5.10405958e-01 7.62766480e-01 -4.36215401e-01
-7.24077344e-01 7.99029052e-01 1.05759002e-01 -1.30878106e-01
-4.98269737e-01 -8.47006202e-01 -9.71319020e-01 -1.24257743e+00
-1.43846959e-01 4.72152263e-01 1.25096232e-01 -3.07032824e-01
1.88235968e-01 7.55544960e-01 -1.17705989e+00 -7.54981562e-02
7.56703436e-01 8.12975407e-01 3.82216662e-01 1.04934847e+00
-1.56092331e-01 7.57167101e-01 -6.02781296e-01 1.88922927e-01
2.78276443e-01 -5.46992958e-01 -6.59347475e-01 2.00318694e-01
-9.10686627e-02 1.90334231e-01 -5.86221933e-01 1.30482876e+00
4.02669132e-01 5.72362304e-01 1.24093878e+00 -1.00013697e+00
-9.94626224e-01 6.44835234e-01 6.33968413e-02 3.23805362e-01
-3.17062914e-01 2.25067452e-01 5.41430593e-01 -4.28941548e-01
4.56393450e-01 1.21511328e+00 6.10577941e-01 1.23245943e+00
-1.60798860e+00 -3.45993228e-02 1.50295019e-01 -4.67619807e-01
-1.25742996e+00 -4.87626106e-01 5.01696825e-01 -7.44003057e-01
7.44064987e-01 6.42152354e-02 4.12437469e-01 7.91548014e-01
7.08257079e-01 7.51521051e-01 5.90383947e-01 -3.24268281e-01
7.64299512e-01 -1.11416668e-01 1.06176309e-01 9.37729061e-01
2.56935298e-01 3.45544815e-02 6.39425498e-03 -5.50748408e-02
1.22934449e+00 -2.59939786e-02 -5.01158893e-01 -5.25304615e-01
-9.37499762e-01 1.01826692e+00 4.53605175e-01 4.28987324e-01
-6.22669399e-01 8.53382826e-01 -2.87644807e-02 3.93219084e-01
3.50303948e-01 4.28892612e-01 -2.23448366e-01 8.07752386e-02
-4.23030615e-01 3.44371706e-01 9.31560874e-01 7.56479025e-01
3.62599254e-01 4.08489585e-01 -5.30195832e-02 5.09303987e-01
1.64054796e-01 8.40310395e-01 5.12927890e-01 -1.39121187e+00
2.88854897e-01 8.56203809e-02 4.56689030e-01 -6.04402363e-01
-3.75511736e-01 -2.61132032e-01 -9.59302306e-01 4.18831140e-01
7.36875474e-01 -5.24310410e-01 -8.23349833e-01 2.27143502e+00
-1.17464975e-01 5.81586584e-02 1.87735379e-01 5.28879285e-01
-2.88846433e-01 1.04748237e+00 -3.20016116e-01 -3.20093542e-01
4.17999506e-01 -5.01324236e-01 -7.49471128e-01 -2.08424721e-02
6.03723705e-01 -7.55235627e-02 9.37908530e-01 1.99178681e-01
-1.47243559e+00 -1.88719541e-01 -9.16231155e-01 3.49852085e-01
-1.99829534e-01 -1.31622896e-01 1.02658443e-01 3.56489331e-01
-1.38334346e+00 1.07499254e+00 -1.25665236e+00 -2.06842437e-01
2.84950733e-01 5.87988257e-01 1.06569216e-01 3.66740197e-01
-1.00333869e+00 8.83098722e-01 1.44642264e-01 4.32823211e-01
-9.76051748e-01 -4.71143126e-01 -4.35421437e-01 -1.28104821e-01
1.15313381e-01 -8.72469425e-01 1.45643091e+00 -8.05196524e-01
-1.63478768e+00 3.29848468e-01 -2.41279021e-01 -7.28035867e-01
6.11384034e-01 1.26097007e-02 -1.18674953e-02 -1.95107937e-01
8.41589645e-02 5.31319320e-01 1.07058322e+00 -9.62860286e-01
-3.32916409e-01 -4.10499692e-01 -4.22712602e-02 1.54390454e-01
-2.01123729e-01 -5.59775412e-01 2.25977704e-01 -5.78620255e-01
8.42356905e-02 -1.32208061e+00 -6.69242918e-01 1.71736136e-01
-2.53370106e-01 -3.70408058e-01 4.88847166e-01 -9.30004194e-02
1.24680471e+00 -1.85723865e+00 7.52841771e-01 1.70781419e-01
-8.57648998e-02 -1.00574614e-02 7.00958446e-02 1.91397160e-01
2.09764719e-01 1.77882865e-01 -6.50853336e-01 -8.00473243e-02
7.36041516e-02 6.27665579e-01 -6.69031203e-01 7.95464098e-01
1.60398513e-01 9.11766231e-01 -8.40648055e-01 1.20228358e-01
-1.80483103e-01 5.09113133e-01 -4.36836362e-01 -1.94839254e-01
-4.48411673e-01 2.13687807e-01 -8.10159743e-01 -6.36989400e-02
3.51443142e-02 -1.42382095e-02 -3.55677426e-01 4.41054761e-01
-8.80501196e-02 -4.05150115e-01 -1.12564695e+00 1.18878961e+00
-4.23249155e-01 7.96167254e-01 3.08904648e-01 -1.04587471e+00
1.02310908e+00 1.72208816e-01 3.16219598e-01 1.92191854e-01
2.29599372e-01 3.06804627e-01 2.73000181e-01 -4.22552764e-01
2.76225924e-01 -4.83698577e-01 -3.10068317e-02 4.48039681e-01
-3.43587637e-01 -4.01567459e-01 2.01629493e-02 -1.77020714e-01
1.22141695e+00 -2.71194309e-01 -1.00210004e-01 -7.88953662e-01
7.18531251e-01 -1.44195363e-01 2.60300308e-01 9.60750043e-01
-1.19889505e-01 2.73361295e-01 4.33336914e-01 -4.58242595e-01
-1.30807185e+00 -1.12715197e+00 -7.08547384e-02 5.91250181e-01
-1.57887459e-01 4.52670783e-01 -1.07829356e+00 6.03741826e-03
1.84136614e-01 8.85451794e-01 -9.70768452e-01 -4.99478340e-01
-8.63551795e-01 -5.86414039e-01 4.48229998e-01 5.31144381e-01
4.18886393e-01 -1.19471395e+00 -7.13242412e-01 4.64927226e-01
3.77593547e-01 -5.09753585e-01 -7.99545646e-01 3.47042471e-01
-1.25416934e+00 -5.89961112e-01 -1.10852933e+00 -7.56345689e-01
9.55152750e-01 -1.84828803e-01 5.83087265e-01 -4.99466926e-01
9.84178260e-02 6.14975631e-01 5.06757975e-01 -3.90898615e-01
-6.45519257e-01 2.24214748e-01 5.24201334e-01 1.43828660e-01
7.19524594e-03 -3.66653144e-01 -4.78167385e-01 4.19691116e-01
-1.27695239e+00 -2.78720468e-01 1.53548881e-01 6.34740770e-01
6.54328763e-01 2.83430010e-01 4.45518911e-01 -5.10986149e-01
1.23242474e+00 -4.01730895e-01 -7.55922496e-01 8.17828625e-02
-2.70268887e-01 8.99777889e-01 7.25440323e-01 -8.59875143e-01
-8.62584531e-01 1.93109319e-01 1.95948303e-01 -3.92718285e-01
2.32066587e-01 3.12706172e-01 2.07784679e-02 -6.81864172e-02
7.08303750e-01 4.21960980e-01 3.98133337e-01 -1.89340904e-01
4.16624546e-01 3.73498827e-01 6.35638654e-01 -7.57467091e-01
4.08717930e-01 6.36175454e-01 3.89906853e-01 -1.00154006e+00
-7.94546247e-01 1.42146870e-01 -4.60252106e-01 -2.76280314e-01
8.74581337e-01 -1.12739682e-01 -9.57141280e-01 5.29105961e-01
-1.28785443e+00 -8.38085711e-01 -9.63778198e-01 2.51557261e-01
-1.27445686e+00 -3.17244887e-01 -4.72579360e-01 -1.23847926e+00
5.77806234e-02 -1.01652849e+00 5.09979188e-01 2.59269208e-01
-1.33930475e-01 -1.31836450e+00 2.42168516e-01 -5.03434777e-01
6.07558668e-01 2.03705788e-01 9.46352780e-01 -3.82203788e-01
-4.43413198e-01 -3.69461328e-01 4.39410418e-01 6.75102890e-01
-1.06215924e-02 -9.32595283e-02 -3.75087947e-01 -6.76523745e-02
6.75198436e-01 2.33130202e-01 1.02369392e+00 1.09163904e+00
7.94615507e-01 -5.32006860e-01 -3.59178960e-01 4.03714567e-01
1.34816313e+00 5.68720639e-01 1.80849344e-01 2.40549773e-01
4.68415052e-01 7.22991049e-01 -9.30918455e-02 2.22247422e-01
-5.35055362e-02 2.37097025e-01 2.41750747e-01 5.34639657e-01
3.38402420e-01 -7.89842084e-02 5.62388301e-01 6.30863070e-01
-2.05208734e-01 -3.37676018e-01 -9.78142679e-01 5.96745968e-01
-2.02993751e+00 -1.21951139e+00 1.86331384e-02 2.26286316e+00
6.52990282e-01 1.97707906e-01 -4.48453277e-02 3.01512033e-02
1.10230315e+00 5.15199378e-02 -1.10182917e+00 -5.95424831e-01
-3.15334767e-01 -1.79281041e-01 8.68630171e-01 1.03135657e+00
-6.61015987e-01 5.71566820e-01 7.15729046e+00 6.48154497e-01
-9.18656230e-01 -9.88894105e-02 6.95653319e-01 -9.50906798e-02
-3.22841972e-01 -3.75550956e-01 -8.95710468e-01 5.42834103e-01
1.18451929e+00 -4.69706535e-01 6.95978165e-01 7.25065589e-01
4.71791506e-01 4.14350480e-02 -1.06084657e+00 5.00872016e-01
-4.60773945e-01 -1.53318977e+00 -2.36239694e-02 5.04539073e-01
1.12001371e+00 2.98087560e-02 5.00333130e-01 -5.56063727e-02
6.73791885e-01 -1.01012874e+00 6.28464878e-01 1.22307098e+00
4.50460047e-01 -9.05873537e-01 3.66391569e-01 8.32271039e-01
-8.09967577e-01 -2.96830475e-01 -7.12511480e-01 -1.09630018e-01
4.70645815e-01 3.25072378e-01 -3.45421940e-01 -5.49451411e-01
-7.65916482e-02 2.07462788e-01 2.55059689e-01 8.27761769e-01
1.75546348e-01 3.88327956e-01 -5.95188200e-01 -6.08126998e-01
5.80290675e-01 -4.67923015e-01 7.42701471e-01 9.16173458e-01
5.51740348e-01 2.58948177e-01 -1.91338718e-01 1.06905198e+00
8.74891356e-02 -3.18542898e-01 -9.91307139e-01 -1.31981626e-01
1.46625757e-01 6.35870576e-01 -7.66387284e-01 -2.15574324e-01
7.59286582e-02 9.62218940e-01 1.91974476e-01 6.28664970e-01
-6.93379045e-01 -5.25557756e-01 9.62298334e-01 2.31388226e-01
3.56434315e-01 -5.66802800e-01 -3.12311977e-01 -8.90673101e-01
-1.68583542e-01 4.52749198e-03 -1.83918923e-01 -5.25423706e-01
-1.04544568e+00 6.85949624e-01 -6.71088621e-02 -6.35291338e-01
-6.72198653e-01 -7.33483374e-01 -5.53483248e-01 9.59148943e-01
-6.51795924e-01 -1.76535919e-01 3.32343727e-01 6.51038170e-01
5.64451337e-01 -3.69571775e-01 4.84520555e-01 -4.23796058e-01
-4.42600757e-01 -3.31429504e-02 6.95232391e-01 -8.88321996e-02
-2.61381149e-01 -1.39161873e+00 4.61726069e-01 7.79811621e-01
-8.90783221e-02 5.56141078e-01 1.14105988e+00 -7.13291585e-01
-1.32559848e+00 -1.01561201e+00 8.12111259e-01 -5.85664511e-01
1.01922619e+00 -1.81497768e-01 -9.24533427e-01 7.01635957e-01
-2.50370681e-01 2.46302545e-01 -1.97857972e-02 -6.53552711e-01
4.51042354e-01 1.08524136e-01 -1.18292880e+00 8.27497363e-01
1.13563216e+00 -3.20329040e-01 -3.91521186e-01 1.85892478e-01
6.54064834e-01 -1.23843782e-01 -6.21734798e-01 1.64909288e-01
3.60550761e-01 -4.76194173e-01 7.70382047e-01 -8.64505172e-01
1.44389048e-01 -3.11980158e-01 -2.62132823e-01 -1.55416071e+00
-3.36037725e-01 -1.00839007e+00 -2.39110038e-01 5.36213160e-01
3.10961187e-01 -6.42875493e-01 1.04225242e+00 9.01063383e-01
-4.77129370e-02 -1.00489557e+00 -1.02343273e+00 -1.02279699e+00
6.37460172e-01 -3.94566268e-01 2.42272601e-01 2.53966123e-01
-2.36871168e-01 1.28019899e-01 -1.10831089e-01 2.81796157e-02
5.21564126e-01 -4.47588474e-01 1.04658961e-01 -1.18944156e+00
-2.97674447e-01 -8.17763805e-01 -3.14934731e-01 -1.47296679e+00
3.28917205e-01 -8.44992638e-01 4.66159165e-01 -1.41537702e+00
-4.16529626e-01 -2.97351956e-01 -5.65964263e-03 -3.05786163e-01
3.99451226e-01 -5.19972920e-01 6.70581013e-02 5.09805918e-01
-1.46612078e-02 7.07087338e-01 1.29157972e+00 8.63737762e-02
-4.94336575e-01 7.66879320e-01 -3.14076543e-01 9.72436070e-01
8.39727998e-01 -4.80499685e-01 -7.38698065e-01 -5.85692883e-01
1.40296310e-01 3.57726306e-01 8.51570666e-02 -9.72180009e-01
6.63307071e-01 -2.55992264e-01 -5.00514843e-02 -2.44331017e-01
2.35336185e-01 -5.99854648e-01 9.95879918e-02 9.72310543e-01
-9.29786861e-01 3.63654524e-01 -1.98639199e-01 1.02760458e+00
8.08490887e-02 -7.81244576e-01 9.41208243e-01 -1.73800543e-01
-2.23507017e-01 3.60106379e-01 -9.86438036e-01 1.13308176e-01
9.48494434e-01 -2.34162882e-01 1.21795520e-01 -6.39912546e-01
-1.15550613e+00 5.02792597e-02 3.35639179e-01 -8.48676339e-02
5.06835818e-01 -1.35369396e+00 -3.34966570e-01 4.25332636e-02
-6.32152915e-01 9.11908448e-02 -5.06641530e-02 5.50848007e-01
-4.20693487e-01 6.41921222e-01 -3.24524716e-02 -4.28851873e-01
-5.70406854e-01 5.35220385e-01 8.86814296e-01 7.23286122e-02
-3.25874448e-01 9.22823608e-01 2.09542707e-01 -1.63474917e-01
2.83562422e-01 -7.75298297e-01 1.06871970e-01 -2.01177925e-01
4.81774330e-01 5.48523724e-01 -5.41061223e-01 -5.28552711e-01
-2.41432749e-02 5.44590712e-01 2.44358659e-01 -8.17554772e-01
1.28690100e+00 -2.61833161e-01 -4.64244224e-02 8.83382499e-01
1.21400690e+00 -3.64777058e-01 -1.73696685e+00 -3.87762815e-01
2.44670734e-01 1.88000828e-01 -5.86563023e-03 -2.13219866e-01
-9.54664052e-01 7.79368877e-01 3.44419241e-01 6.43985093e-01
6.30607128e-01 1.34964421e-01 5.43426991e-01 7.77047157e-01
2.05009311e-01 -1.14448977e+00 -1.02407239e-01 8.62309635e-01
9.62519825e-01 -6.51356518e-01 -6.85995102e-01 2.04392076e-01
-4.90317583e-01 1.43057418e+00 1.66686565e-01 -9.32943463e-01
8.79380822e-01 5.66788495e-01 -5.04104495e-01 8.31912905e-02
-8.53153646e-01 8.52455273e-02 2.95932561e-01 5.53255081e-01
2.62957692e-01 6.97172433e-02 -1.08712189e-01 3.13164771e-01
-8.02368969e-02 2.03319862e-01 6.23475790e-01 8.80306423e-01
-1.07478309e+00 -6.14482164e-01 -1.04425833e-01 2.48193130e-01
-2.64541358e-01 1.45804882e-01 -4.91929464e-02 6.63456082e-01
-2.96076655e-01 3.23683500e-01 3.10943276e-01 -6.71151727e-02
2.35547140e-01 7.27594122e-02 5.34343183e-01 -5.60078382e-01
3.69114764e-02 -1.35069594e-01 -3.83263260e-01 -3.63206923e-01
-1.33594126e-01 -1.06128383e+00 -1.36397970e+00 -3.28473926e-01
1.61444042e-02 7.07827210e-02 7.58806646e-01 9.93432403e-01
-1.70206428e-02 4.67707962e-01 3.64386678e-01 -8.18038404e-01
-1.27567184e+00 -6.30537570e-01 -6.71234906e-01 -1.50694311e-01
6.08993173e-01 -3.32198501e-01 -5.43925524e-01 -5.17862290e-03] | [6.960142135620117, 3.6128458976745605] |
5fcf1f2b-e877-4ff0-b937-0bdd3e536370 | mask2lesion-mask-constrained-adversarial-skin | 1906.05845 | null | https://arxiv.org/abs/1906.05845v2 | https://arxiv.org/pdf/1906.05845v2.pdf | Mask2Lesion: Mask-Constrained Adversarial Skin Lesion Image Synthesis | Skin lesion segmentation is a vital task in skin cancer diagnosis and further treatment. Although deep learning based approaches have significantly improved the segmentation accuracy, these algorithms are still reliant on having a large enough dataset in order to achieve adequate results. Inspired by the immense success of generative adversarial networks (GANs), we propose a GAN-based augmentation of the original dataset in order to improve the segmentation performance. In particular, we use the segmentation masks available in the training dataset to train the Mask2Lesion model, and use the model to generate new lesion images given any arbitrary mask, which are then used to augment the original training dataset. We test Mask2Lesion augmentation on the ISBI ISIC 2017 Skin Lesion Segmentation Challenge dataset and achieve an improvement of 5.17% in the mean Dice score as compared to a model trained with only classical data augmentation techniques. | ['Kumar Abhishek', 'Ghassan Hamarneh'] | 2019-06-13 | null | null | null | null | ['skin-lesion-segmentation'] | ['medical'] | [ 8.58922005e-01 3.27526301e-01 7.40544647e-02 -2.35287383e-01
-8.71586084e-01 -5.85664809e-01 5.96343935e-01 -1.99891943e-02
-5.41138530e-01 6.58674240e-01 -1.99856348e-02 -1.39997095e-01
4.54660624e-01 -8.90181661e-01 -5.67908466e-01 -9.72246051e-01
5.28303564e-01 3.69986743e-01 1.71453491e-01 -1.42627275e-02
-6.10337825e-03 4.26690698e-01 -7.76321471e-01 3.59427810e-01
1.41415918e+00 8.69371057e-01 1.90916229e-02 6.97910547e-01
-2.58821249e-01 3.89184713e-01 -6.24111593e-01 -5.93615532e-01
3.94632727e-01 -9.54029500e-01 -8.49611342e-01 1.72228262e-01
3.48665088e-01 -2.49649003e-01 -1.88006520e-01 1.09599495e+00
5.08204401e-01 -2.15870872e-01 7.35967934e-01 -8.39290559e-01
-2.78765172e-01 3.89202148e-01 -6.50136113e-01 -8.88452455e-02
-7.54495244e-03 3.33562523e-01 3.93730283e-01 -1.82138562e-01
7.71117449e-01 5.89479327e-01 7.13868499e-01 1.06976926e+00
-1.34188974e+00 -4.46430176e-01 -2.78330863e-01 -4.33910862e-02
-1.09585190e+00 -1.70494527e-01 6.49340153e-01 -2.94288516e-01
2.07918242e-01 6.04635656e-01 9.67835784e-01 1.16658354e+00
-5.28187081e-02 8.05351973e-01 1.37551963e+00 -4.35617834e-01
3.85431141e-01 7.92172272e-03 -3.81188720e-01 4.76424307e-01
7.87573159e-02 -1.67387798e-01 1.86800897e-01 2.69479811e-01
1.08616471e+00 -4.76670228e-02 -3.08373094e-01 -1.17112389e-02
-9.39105630e-01 8.99636149e-01 1.00820029e+00 4.35550839e-01
-7.16486573e-01 1.53420299e-01 2.94856638e-01 -6.82740584e-02
5.32414734e-01 5.75644553e-01 1.45638838e-01 1.03409760e-01
-1.25241792e+00 -5.82309999e-02 4.23189551e-01 1.36748224e-01
2.95838952e-01 -1.21268019e-01 -3.76164913e-01 9.65906978e-01
-2.16586575e-01 2.94195235e-01 6.05776966e-01 -7.51369059e-01
2.00681865e-01 8.74640703e-01 -2.50507236e-01 -4.90061492e-01
-2.08408311e-01 -5.31123817e-01 -1.09581566e+00 4.52171624e-01
6.71508670e-01 -1.17577329e-01 -1.68095064e+00 1.51848984e+00
4.63585585e-01 1.42018244e-01 -1.28130734e-01 7.93429077e-01
5.34770489e-01 2.73106545e-01 2.58990645e-01 5.63482717e-02
8.88144791e-01 -9.88177180e-01 -6.38447106e-01 -1.21102437e-01
7.22465932e-01 -6.57666147e-01 8.97230864e-01 5.08224249e-01
-1.18745732e+00 -2.08023757e-01 -9.38074827e-01 2.15189770e-01
-3.34815592e-01 -8.58901720e-03 5.98239124e-01 8.40121508e-01
-9.42648828e-01 5.01441419e-01 -1.03864634e+00 -4.11453843e-01
1.20971298e+00 3.37129295e-01 -5.05642295e-01 -4.72564101e-01
-7.21243620e-01 8.18443000e-01 4.12843853e-01 2.54256397e-01
-7.39421368e-01 -8.74709725e-01 -5.56760728e-01 -3.14030170e-01
2.00784117e-01 -7.14603424e-01 8.20221364e-01 -1.18371189e+00
-1.40613401e+00 9.37311351e-01 2.36037727e-02 -6.94776654e-01
1.06731224e+00 1.96813941e-01 -2.64811376e-03 4.53191817e-01
-1.09789446e-01 9.70321834e-01 7.53032207e-01 -1.31680286e+00
-3.49261612e-01 -2.80822545e-01 -1.60702512e-01 1.08240850e-01
-2.22051159e-01 -3.10825139e-01 -5.41551173e-01 -7.03608215e-01
-2.16939673e-02 -1.07067335e+00 -7.15822697e-01 -4.26478833e-02
-7.93187976e-01 3.94547015e-01 7.97039926e-01 -1.28131783e+00
7.87782252e-01 -1.87826574e+00 1.33510739e-01 4.52061027e-01
2.17623025e-01 6.70462668e-01 -3.58026087e-01 -4.89848740e-02
-2.18458816e-01 5.16030073e-01 -9.31906939e-01 -3.32464457e-01
-4.37205613e-01 1.24041818e-01 3.20929796e-01 3.30604345e-01
3.01061153e-01 1.26381052e+00 -7.75573909e-01 -4.30521250e-01
4.50109869e-01 6.43480122e-01 -4.68385935e-01 2.50476122e-01
-3.34531963e-01 9.21508789e-01 -2.16415837e-01 4.95743006e-01
8.25243592e-01 -1.05994292e-01 8.02351460e-02 -2.14684218e-01
4.89125282e-01 -2.21093029e-01 -6.72753870e-01 1.79126906e+00
-4.75400895e-01 4.81314868e-01 -6.69026794e-03 -8.79231870e-01
5.13779521e-01 4.17527109e-01 8.36153686e-01 -6.48040175e-01
3.31279963e-01 2.03657597e-01 4.39179331e-01 -3.31352592e-01
-1.36865407e-01 -4.26627040e-01 3.51037323e-01 2.07362786e-01
-2.61828959e-01 -5.62644482e-01 2.34367803e-01 1.86877102e-01
1.34477878e+00 -6.35313690e-02 -4.14403379e-02 1.70262873e-01
5.29319048e-01 7.33889118e-02 2.93785363e-01 2.95836210e-01
-2.16418892e-01 1.14480114e+00 6.09783292e-01 -1.51779696e-01
-1.24915349e+00 -9.93417859e-01 -1.00540504e-01 2.03042090e-01
-2.39781484e-01 4.32525836e-02 -1.49343204e+00 -1.42882991e+00
-3.45391989e-01 6.35255933e-01 -1.11116791e+00 -3.48676816e-02
-4.46654588e-01 -1.13179278e+00 5.94707727e-01 6.32697046e-01
7.86853492e-01 -1.15791488e+00 -2.53122777e-01 2.24284694e-01
-2.19075859e-01 -1.15516460e+00 -3.08083922e-01 -7.14524239e-02
-6.46884978e-01 -1.22632647e+00 -1.24075651e+00 -5.26058316e-01
1.26780343e+00 -3.69719028e-01 7.68910408e-01 4.27453160e-01
-8.41153741e-01 -2.17011832e-02 -4.21887875e-01 -3.12002569e-01
-8.67868543e-01 1.18414238e-01 -5.57939351e-01 1.56816527e-01
-3.09363869e-03 -4.24617887e-01 -8.88673961e-01 1.16472289e-01
-1.40817380e+00 2.31269270e-01 8.58659565e-01 1.03771615e+00
6.77801430e-01 1.47031456e-01 6.68665349e-01 -1.31601298e+00
4.22960192e-01 -1.67942941e-01 -3.56912315e-01 1.76753953e-01
-3.82825047e-01 -3.06707144e-01 6.21702492e-01 -3.73060703e-01
-9.67033267e-01 2.34719723e-01 -6.89178824e-01 -3.10311675e-01
-3.02605361e-01 3.70848835e-01 -6.11424036e-02 -4.34943855e-01
6.61942720e-01 1.86317831e-01 4.02862549e-01 -2.97311157e-01
2.85551906e-01 4.17646259e-01 5.22837579e-01 3.85651514e-02
8.65462840e-01 5.20434558e-01 1.32139802e-01 -5.73408544e-01
-6.89459443e-01 -1.12337910e-01 -9.25531566e-01 -1.03384025e-01
1.25623131e+00 -3.52635413e-01 -1.80458650e-01 7.50826895e-01
-8.84859324e-01 -8.15414608e-01 -4.91995811e-01 5.72647639e-02
-4.12600756e-01 3.12289149e-01 -4.56708759e-01 -4.24163431e-01
-5.32025158e-01 -1.21092391e+00 6.78747892e-01 5.08148313e-01
-2.61517651e-02 -1.34239841e+00 5.71831083e-03 6.13910139e-01
5.56338429e-01 9.37886357e-01 7.88416743e-01 -6.65524304e-01
-2.70209879e-01 -6.70026362e-01 -2.38135770e-01 7.41225839e-01
4.25058901e-01 -4.26586345e-02 -8.03511381e-01 -2.16820329e-01
-2.22492173e-01 -2.61671633e-01 9.74705398e-01 5.95837116e-01
1.58116162e+00 -5.76704629e-02 -4.00700122e-01 6.62106276e-01
1.53868008e+00 2.07707256e-01 1.17824626e+00 -1.30405780e-02
8.19580793e-01 5.53392291e-01 1.84479535e-01 -1.96938336e-01
-7.71864578e-02 4.20788050e-01 5.28880000e-01 -7.10248709e-01
-7.90158868e-01 -3.27749789e-01 -2.70846367e-01 3.63829434e-01
-1.84228495e-01 -2.50186265e-01 -8.52564752e-01 7.13768721e-01
-1.34023249e+00 -7.16175556e-01 -2.22322389e-01 2.18951035e+00
9.51328874e-01 5.67445308e-02 1.35460317e-01 2.21433997e-01
7.17737854e-01 -8.99081007e-02 -7.41408527e-01 -2.78047174e-01
-4.23670262e-02 7.36948133e-01 5.34581184e-01 3.80608022e-01
-1.06401277e+00 1.03318572e+00 6.26615620e+00 9.23723280e-01
-1.49934554e+00 1.14496253e-01 1.05783367e+00 6.45821691e-02
-2.52667397e-01 -2.55888700e-01 -1.08929604e-01 6.95947886e-01
6.75237477e-01 2.80748069e-01 4.18876261e-01 2.40759730e-01
1.23542272e-01 -3.64027023e-01 -7.62216091e-01 7.22080708e-01
9.75744203e-02 -1.40931404e+00 4.31684814e-02 4.02280599e-01
1.10862529e+00 -2.32842326e-01 2.84481406e-01 -8.03125650e-02
1.68471277e-01 -1.60447097e+00 -1.71266869e-01 6.31976306e-01
1.09203780e+00 -7.30448902e-01 1.08173180e+00 1.09594785e-01
-4.63264167e-01 3.44429553e-01 -3.59592475e-02 6.80430353e-01
2.46150717e-01 6.93870485e-01 -1.32914972e+00 5.55108905e-01
2.01395556e-01 3.31550628e-01 -8.91499162e-01 1.40912938e+00
-3.62873882e-01 9.49362099e-01 -3.72976303e-01 2.98051476e-01
2.22690433e-01 -3.96767646e-01 2.89910972e-01 8.69000971e-01
-6.16272306e-03 -9.89669338e-02 -2.46650904e-01 9.68445361e-01
-3.74875247e-01 1.70011476e-01 -3.33676904e-01 -2.61299521e-01
-1.21123507e-03 1.55873358e+00 -9.30120766e-01 -2.66333938e-01
-6.60696328e-02 1.57005167e+00 -4.50226152e-03 3.19874614e-01
-7.78926790e-01 -2.49383986e-01 2.49252304e-01 2.68408865e-01
2.19107419e-01 1.52397782e-01 -5.81506491e-01 -7.76012361e-01
-1.06446899e-01 -1.00368106e+00 2.44656533e-01 -4.63515967e-01
-1.34121633e+00 6.87514067e-01 -5.64917684e-01 -9.06727254e-01
-1.68443412e-01 -3.30015659e-01 -1.08236945e+00 1.07480240e+00
-1.47542882e+00 -1.55657721e+00 -7.01945066e-01 3.36830318e-01
3.41006219e-01 3.46937552e-02 9.51587141e-01 1.88713565e-01
-6.12112701e-01 7.76913404e-01 5.53965382e-02 4.17385250e-01
6.09259665e-01 -1.51279426e+00 3.96278501e-01 9.03609395e-01
9.36800018e-02 1.56299502e-01 4.00302738e-01 -7.25237668e-01
-8.46515179e-01 -1.19808257e+00 1.12252519e-01 -4.55687910e-01
3.39549094e-01 -1.46383509e-01 -9.18612301e-01 4.20840949e-01
3.52359056e-01 2.89480358e-01 7.86929548e-01 -4.72280502e-01
3.70554030e-02 8.89064744e-02 -1.77929330e+00 6.58983052e-01
5.97752213e-01 -8.94055590e-02 -1.24862432e-01 3.08764994e-01
3.84845704e-01 -4.51896876e-01 -9.38461065e-01 5.71662843e-01
3.69698703e-01 -7.88919151e-01 8.55413854e-01 -5.67334533e-01
6.85594857e-01 9.64550488e-03 2.11177960e-01 -1.63759708e+00
1.25919119e-01 -4.07767594e-01 3.46907020e-01 1.11277902e+00
4.11477476e-01 -6.03065431e-01 1.44134891e+00 6.94655538e-01
2.50219628e-02 -9.59176004e-01 -8.47804844e-01 -4.95705009e-01
5.77365279e-01 -6.15358204e-02 4.10757363e-01 9.10273731e-01
-4.63956505e-01 -1.61402002e-01 -3.55564095e-02 -2.99994379e-01
7.34441102e-01 -3.28231245e-01 7.11365998e-01 -8.40339243e-01
-6.31684586e-02 -5.46072960e-01 -3.20224434e-01 -4.94892150e-01
-6.09391229e-03 -1.07703745e+00 -2.22818162e-02 -1.92776227e+00
2.55822390e-01 -7.33957171e-01 -3.05232167e-01 7.06331670e-01
-4.72106129e-01 1.02785897e+00 2.39498943e-01 -6.66887537e-02
1.83024146e-02 2.99742997e-01 1.72246242e+00 -2.95776844e-01
-2.48199210e-01 7.37489909e-02 -6.78509057e-01 5.48337817e-01
8.88574481e-01 -1.92254111e-01 -1.33769840e-01 -9.56489667e-02
-3.30169976e-01 -1.18321605e-01 5.59296966e-01 -1.13563800e+00
-3.89365032e-02 -4.01201611e-03 8.25674295e-01 -3.17414969e-01
3.06599885e-01 -8.24119449e-01 1.01401575e-01 7.14854777e-01
-3.49787652e-01 -6.84347749e-01 3.05127859e-01 2.61348367e-01
-1.81253299e-01 -3.84892017e-01 1.10541975e+00 -1.38080075e-01
-1.95001110e-01 4.09377754e-01 -1.05295591e-01 -1.34077430e-01
1.20602250e+00 -1.91783234e-01 -1.07401296e-01 -2.83464909e-01
-9.24073756e-01 -9.82286483e-02 7.44489789e-01 -7.99845979e-02
4.14651394e-01 -1.14359403e+00 -8.17342579e-01 1.54569209e-01
-2.11788625e-01 1.98896423e-01 5.20266175e-01 1.15853405e+00
-8.15252542e-01 -1.64021719e-02 -4.82496887e-01 -5.08076489e-01
-1.07772064e+00 3.09643209e-01 6.36609554e-01 -5.38079023e-01
-5.37112117e-01 8.78878474e-01 7.63855651e-02 -2.52692103e-01
-1.48925379e-01 2.81084832e-02 1.55072694e-03 -2.94226080e-01
3.69838923e-01 8.44846815e-02 2.23046169e-01 -2.66540766e-01
-6.91514313e-02 4.15617079e-01 -1.76287696e-01 -1.07918262e-01
1.22633481e+00 3.50332111e-01 -8.75164121e-02 -2.43520051e-01
1.14227068e+00 1.96801007e-01 -1.28105831e+00 1.63227066e-01
-1.98894992e-01 -5.41063249e-01 2.09584042e-01 -1.55129778e+00
-1.39431310e+00 9.43033159e-01 8.81997168e-01 2.81761974e-01
1.29054737e+00 -2.20129922e-01 1.14476180e+00 -3.67960274e-01
-2.37671882e-02 -8.00074697e-01 1.01888388e-01 -1.10414043e-01
8.28225493e-01 -1.28224730e+00 -2.09807009e-01 -7.43822694e-01
-7.44598210e-01 9.23830450e-01 5.17028034e-01 -2.84747481e-01
1.63820565e-01 3.15915167e-01 2.83884436e-01 8.31165016e-02
-1.76922232e-02 -3.47819626e-01 3.42965484e-01 8.21216285e-01
2.81192124e-01 2.39662882e-02 -4.31453764e-01 2.95388609e-01
-6.75835609e-02 7.56422356e-02 4.32889342e-01 5.23979306e-01
5.18718921e-02 -1.53467929e+00 -1.38573423e-01 7.88398802e-01
-6.49000883e-01 4.22580838e-02 -6.05026662e-01 7.48369694e-01
2.05633864e-01 6.43666744e-01 -6.96543157e-02 -2.49018863e-01
7.19132414e-03 1.08501308e-01 7.34404027e-01 -5.71774304e-01
-7.72686541e-01 -7.15685776e-03 -1.33989036e-01 -3.57095420e-01
-1.92811832e-01 -5.21770000e-01 -1.27294350e+00 -2.90159825e-02
-1.69582754e-01 -2.30039060e-02 1.12758231e+00 8.65072668e-01
-8.70893802e-03 7.91370571e-01 6.11460268e-01 -5.33446133e-01
-2.95124769e-01 -1.12713039e+00 -3.49643081e-01 7.04827309e-01
-1.66251399e-02 -2.01787308e-01 -1.92488164e-01 2.36398593e-01] | [15.19552993774414, -2.6967155933380127] |
ea4a41ff-fcf0-40d9-8a73-adcc215082ac | learning-thematic-similarity-metric-from | null | null | https://aclanthology.org/P18-2009 | https://aclanthology.org/P18-2009.pdf | Learning Thematic Similarity Metric from Article Sections Using Triplet Networks | In this paper we suggest to leverage the partition of articles into sections, in order to learn thematic similarity metric between sentences. We assume that a sentence is thematically closer to sentences within its section than to sentences from other sections. Based on this assumption, we use Wikipedia articles to automatically create a large dataset of weakly labeled sentence triplets, composed of a pivot sentence, one sentence from the same section and one from another section. We train a triplet network to embed sentences from the same section closer. To test the performance of the learned embeddings, we create and release a sentence clustering benchmark. We show that the triplet network learns useful thematic metrics, that significantly outperform state-of-the-art semantic similarity methods and multipurpose embeddings on the task of thematic clustering of sentences. We also show that the learned embeddings perform well on the task of sentence semantic similarity prediction. | ['Noam Slonim', 'Yosi Mass', 'Alon Halfon', 'Ilya Shnayderman', 'Ranit Aharonov', 'Elad Venezian', 'Liat Ein Dor'] | 2018-07-01 | null | null | null | acl-2018-7 | ['text-clustering'] | ['natural-language-processing'] | [ 9.71637419e-05 7.53003582e-02 -1.64439768e-01 -6.41063690e-01
-9.69083250e-01 -6.45375967e-01 6.14472330e-01 8.99852812e-01
-4.72577244e-01 3.12755644e-01 9.07510817e-01 -7.05157071e-02
-1.37269512e-01 -7.08508849e-01 -7.11623192e-01 -4.67443049e-01
1.55604452e-01 3.69894147e-01 7.98300803e-02 -2.35909343e-01
5.89505851e-01 -1.03426479e-01 -1.31854260e+00 7.11212218e-01
8.18142235e-01 8.81872058e-01 2.17742488e-01 2.45629534e-01
-3.84773135e-01 7.57511437e-01 -3.06511819e-01 -4.64176983e-01
1.58236012e-01 -3.35002959e-01 -1.16438437e+00 1.77103584e-03
6.84197664e-01 1.06387548e-01 -3.52138728e-01 8.30014467e-01
1.59585461e-01 3.92571419e-01 8.76736343e-01 -1.07072759e+00
-9.11897600e-01 1.03952241e+00 -4.64963913e-01 3.92281622e-01
4.78300214e-01 -4.09310699e-01 1.72197080e+00 -9.22372162e-01
7.37264395e-01 1.14057565e+00 9.95068729e-01 3.64238739e-01
-1.22449398e+00 -2.10399047e-01 1.38600111e-01 1.94868475e-01
-1.01842880e+00 -3.44891369e-01 8.77080441e-01 -5.47315717e-01
1.00342846e+00 6.73203245e-02 2.73959368e-01 1.10677803e+00
1.60298929e-01 5.95489919e-01 7.01197088e-01 -4.04872745e-01
1.83193102e-01 3.70597452e-01 7.53900588e-01 3.75988811e-01
1.36535028e-02 -6.82249069e-01 -4.60227370e-01 -1.50237307e-01
-4.18770969e-01 1.23369314e-01 -2.70095497e-01 -5.27676940e-01
-1.13733590e+00 1.15520477e+00 6.08937502e-01 8.17050338e-01
-9.40239802e-02 -7.05891475e-02 8.15483749e-01 3.55099022e-01
8.07258844e-01 7.81588495e-01 -4.30101573e-01 -1.59470379e-01
-7.91213274e-01 3.38408172e-01 7.85678685e-01 7.39689410e-01
9.10782754e-01 -7.60259330e-01 -1.96390092e-01 1.17412698e+00
2.45770961e-01 1.04380688e-02 6.43257201e-01 -1.02014136e+00
8.84542584e-01 7.49792397e-01 -9.96620804e-02 -1.12408900e+00
-3.30766797e-01 -1.82931915e-01 -4.97122228e-01 -6.49392247e-01
1.85118124e-01 1.81157604e-01 -2.79773355e-01 1.72272325e+00
1.56848118e-01 5.65006584e-02 2.67065972e-01 6.07810378e-01
9.51947212e-01 6.45172477e-01 -3.39312963e-02 6.49527833e-02
1.30473232e+00 -1.07900202e+00 -3.69087785e-01 -1.83555499e-01
1.23055172e+00 -6.26992106e-01 1.24650300e+00 -2.97870874e-01
-8.36529016e-01 -6.20615244e-01 -9.35376585e-01 -3.41125518e-01
-4.54217255e-01 -1.49152115e-01 3.24692637e-01 1.46785691e-01
-1.00939393e+00 9.65386689e-01 -5.07098854e-01 -7.26298869e-01
3.68743360e-01 -1.18951440e-01 -5.17660677e-01 -6.87584057e-02
-1.29712844e+00 9.81427372e-01 2.59857178e-01 -4.21009362e-01
-3.68389547e-01 -1.24293685e+00 -1.20328224e+00 1.99020892e-01
-2.23323032e-01 -7.58559465e-01 1.07862055e+00 -8.73916686e-01
-7.31028080e-01 1.35633159e+00 -3.62388790e-01 -6.03591979e-01
-4.29448076e-02 -7.75768831e-02 -2.05879390e-01 4.68870729e-01
7.57111311e-01 5.97078919e-01 3.70475739e-01 -1.10418952e+00
-4.96868908e-01 -6.25654221e-01 2.57884771e-01 3.00493449e-01
-1.02413428e+00 2.28610441e-01 -3.26352045e-02 -5.35565436e-01
-1.04775585e-01 -7.02595413e-01 6.94315135e-02 -1.69382438e-01
-4.87319678e-01 -8.04588974e-01 6.55427575e-01 -5.46816766e-01
1.26088965e+00 -2.35603929e+00 1.15736522e-01 -8.71358514e-02
2.58171767e-01 -2.97223985e-01 -3.69613528e-01 7.59055376e-01
-3.06013793e-01 3.21238548e-01 -3.50792378e-01 -8.71714115e-01
2.99747437e-01 1.11599632e-01 -5.20520985e-01 2.79161066e-01
1.10920779e-01 7.39330173e-01 -1.04725838e+00 -6.02333426e-01
-1.53461128e-01 6.12052046e-02 -4.97717291e-01 9.27410647e-02
-5.00247367e-02 -9.02502462e-02 -3.97864550e-01 -3.31960805e-02
5.16738892e-01 -2.60874271e-01 1.68721065e-01 -4.86664772e-01
-7.28057604e-03 9.99127746e-01 -5.34196138e-01 1.94076264e+00
-6.89092457e-01 8.46141696e-01 -4.34628427e-01 -1.33194327e+00
8.71710122e-01 1.73609048e-01 6.03897214e-01 -5.75979710e-01
6.93196729e-02 -5.77213764e-02 -1.69455692e-01 -6.83826149e-01
6.76259279e-01 -2.10424453e-01 -4.44402456e-01 9.28446054e-01
7.20227063e-02 -1.11069106e-01 3.96003515e-01 6.45872116e-01
1.23475957e+00 -3.21552962e-01 2.00721528e-03 -5.02116680e-01
4.15179193e-01 -1.03835799e-01 3.99393409e-01 5.43202758e-01
-1.17417037e-01 7.14918196e-01 7.93980122e-01 -2.93540776e-01
-1.38692498e+00 -1.35193968e+00 -3.89346927e-01 1.15487039e+00
1.02658667e-01 -8.57735932e-01 -7.38014936e-01 -1.05980480e+00
3.01265031e-01 7.30716527e-01 -1.09311914e+00 -1.99693576e-01
-4.29037005e-01 -3.44999552e-01 2.58827984e-01 7.97896743e-01
2.49169141e-01 -7.04317391e-01 -1.99818574e-02 -8.89279768e-02
-4.06566858e-01 -1.19030583e+00 -7.47278035e-01 7.57286474e-02
-5.23118496e-01 -1.23867524e+00 -2.54721820e-01 -1.21110368e+00
7.06501186e-01 5.16305745e-01 1.47786808e+00 -5.23806438e-02
1.09918147e-01 4.71964628e-01 -6.68271422e-01 8.43214020e-02
-5.83470821e-01 2.84087360e-01 1.80261552e-01 -7.85704777e-02
6.90391183e-01 -4.86958414e-01 -3.07097256e-01 3.25952433e-02
-7.91009545e-01 -2.13177115e-01 -1.08596101e-01 8.16868961e-01
2.63366818e-01 -1.64285049e-01 5.89021325e-01 -9.14833307e-01
6.76937759e-01 -8.26224029e-01 5.77518344e-02 4.30271596e-01
-2.06480175e-01 2.19885737e-01 9.02419984e-01 1.12305460e-02
-5.48853755e-01 -2.64338821e-01 -8.81317332e-02 -2.03270629e-01
-1.28935948e-02 6.49503171e-01 4.82367016e-02 3.86822432e-01
5.98504484e-01 -1.89335961e-02 6.57144561e-02 -4.34091330e-01
4.09469783e-01 1.08990264e+00 3.30411524e-01 -5.62343776e-01
7.71846652e-01 3.41585845e-01 -4.71288800e-01 -6.70236290e-01
-1.40432847e+00 -7.03126252e-01 -8.32616985e-01 1.28520459e-01
1.03835845e+00 -8.91108215e-01 -3.56302887e-01 -1.82725877e-01
-1.31982660e+00 -4.59909774e-02 -2.77474523e-01 3.17407131e-01
-3.94094229e-01 7.43181586e-01 -5.68988144e-01 -7.64397904e-02
-2.75120795e-01 -7.26985455e-01 9.84311700e-01 -1.27718046e-01
-6.02546275e-01 -1.48853719e+00 3.74439001e-01 5.65930903e-01
3.09729651e-02 -4.62602600e-02 1.11187649e+00 -1.15721595e+00
6.87386934e-03 2.52967160e-02 -1.36293143e-01 6.39681578e-01
4.01708603e-01 1.00687362e-01 -8.79025519e-01 -2.44355798e-01
2.34963968e-01 -3.66189569e-01 1.24969661e+00 1.70321539e-01
9.79704857e-01 -2.45739594e-01 -4.93832320e-01 3.57098013e-01
1.19425082e+00 -3.66644561e-01 3.55813235e-01 6.41747773e-01
6.40264928e-01 9.21978951e-01 4.49291766e-01 3.10431987e-01
8.74265134e-01 5.72381496e-01 6.21548891e-02 2.88957685e-01
1.89743176e-01 -2.33212546e-01 3.33381623e-01 1.12892139e+00
7.29737699e-01 -7.90410042e-02 -9.45365429e-01 9.03506815e-01
-1.71621740e+00 -1.29494905e+00 -6.98326528e-02 1.83412671e+00
9.56954181e-01 6.41307086e-02 1.12824492e-01 1.08092435e-01
7.62007177e-01 3.42419446e-01 -6.87329397e-02 -6.91554189e-01
-1.68770373e-01 2.46590227e-02 3.97932418e-02 4.34381664e-01
-1.32794750e+00 7.27391899e-01 5.89196014e+00 4.88651305e-01
-5.93653977e-01 2.11701557e-01 4.86690581e-01 -1.64991207e-02
-9.69978034e-01 1.10470124e-01 -6.93720818e-01 8.65493536e-01
1.11319458e+00 -3.14928502e-01 3.28343064e-02 6.40840948e-01
2.20845461e-01 -2.34624650e-03 -1.53946531e+00 5.95356703e-01
5.28175771e-01 -1.57992029e+00 2.19007302e-02 -2.55157292e-01
9.53948081e-01 1.64993703e-01 5.66582493e-02 3.46851051e-01
9.06791165e-02 -8.37321699e-01 3.52556527e-01 2.57953435e-01
7.68480673e-02 -6.50798380e-01 7.84742594e-01 4.16902713e-02
-1.15569198e+00 -6.15975074e-02 -5.68488479e-01 -6.37414083e-02
2.68462747e-02 7.10236371e-01 -6.08101606e-01 5.74698508e-01
8.50331843e-01 1.42551422e+00 -8.15029800e-01 8.57965410e-01
-1.29485369e-01 4.28302526e-01 -5.96265774e-04 -5.33786938e-02
3.72127414e-01 -2.37751916e-01 3.07347327e-01 1.18189907e+00
1.65192679e-01 -3.73858213e-01 8.43347609e-02 8.27828348e-01
-4.80503708e-01 1.89002156e-01 -1.04591680e+00 1.94516778e-02
8.26426148e-01 1.23582113e+00 -3.45870763e-01 -3.79890472e-01
-7.45330989e-01 9.09505785e-01 7.41914868e-01 -1.97249036e-02
-7.48418510e-01 -6.91429377e-01 1.00597835e+00 -4.79845218e-02
4.81737405e-01 -9.93935578e-03 -4.03288931e-01 -1.14339352e+00
4.29452777e-01 -3.52256954e-01 3.63441736e-01 -7.20604002e-01
-2.04979277e+00 5.53016782e-01 -2.49597430e-01 -1.37807333e+00
-8.76456127e-02 -4.57180172e-01 -9.59457397e-01 5.56750059e-01
-1.46912265e+00 -1.03708386e+00 -1.45428285e-01 2.66310841e-01
4.01731223e-01 -9.16108489e-02 8.33668411e-01 1.30935565e-01
-4.81018364e-01 6.53327405e-01 4.82619286e-01 4.45858419e-01
8.58090997e-01 -1.34291255e+00 4.30198133e-01 4.51851189e-01
2.29274601e-01 8.24139774e-01 5.25955141e-01 -2.72455305e-01
-8.41302693e-01 -1.21155286e+00 1.64708066e+00 -8.55014980e-01
1.23031640e+00 -6.49825156e-01 -9.96342421e-01 6.89560294e-01
7.04466045e-01 -1.88818187e-01 1.19912446e+00 4.96210665e-01
-7.32518315e-01 -2.44406134e-01 -9.14188385e-01 3.59992921e-01
9.07280266e-01 -1.02756405e+00 -1.37731302e+00 7.12830663e-01
1.03059101e+00 3.91088799e-02 -1.18101895e+00 1.02956176e-01
2.73610860e-01 -9.28906620e-01 8.89884055e-01 -9.31648552e-01
1.13799870e+00 -7.42309727e-03 -3.73559445e-01 -1.56038713e+00
-1.87453181e-01 -1.05231009e-01 3.67725492e-01 1.59061277e+00
5.97042263e-01 -4.02799696e-01 6.61887884e-01 4.12645251e-01
-6.49004579e-01 -7.51357615e-01 -9.04050529e-01 -8.96624029e-01
6.75581813e-01 -1.69513240e-01 4.31447417e-01 1.26546824e+00
6.93717122e-01 6.90789461e-01 2.65954375e-01 -1.18996404e-01
4.12001878e-01 4.08303529e-01 3.67597997e-01 -1.34198499e+00
-9.09243375e-02 -4.09747005e-01 -4.32951301e-01 -8.43241394e-01
1.02783704e+00 -1.39092004e+00 1.01830944e-01 -1.88586080e+00
7.21601665e-01 -2.64959544e-01 -3.81342977e-01 3.46813232e-01
-3.83406460e-01 2.87126839e-01 2.67581344e-01 1.42375618e-01
-8.87079358e-01 7.58729875e-01 7.33879209e-01 -4.54761803e-01
2.10929602e-01 -3.52700204e-01 -9.27457631e-01 5.55717528e-01
8.53215575e-01 -6.06692791e-01 -2.69362420e-01 -7.35482991e-01
2.91356117e-01 -4.68297809e-01 2.48798996e-01 -7.82675982e-01
1.11705281e-01 2.86593467e-01 4.07198593e-02 -4.93556917e-01
2.00528085e-01 -6.36025667e-01 -5.47205091e-01 2.38664076e-02
-9.03013766e-01 3.26823652e-01 -1.15806527e-01 4.45256770e-01
-4.81434494e-01 -5.78008115e-01 5.86499751e-01 8.87467340e-02
-3.91326040e-01 -2.47483384e-02 -2.73727000e-01 6.15514696e-01
9.99410391e-01 -1.85004011e-01 -5.54462016e-01 -5.89457825e-02
-6.20953619e-01 4.14427280e-01 7.64036715e-01 7.27573574e-01
5.61770380e-01 -1.57327986e+00 -6.23718798e-01 -2.03542635e-01
4.78011489e-01 -4.04355258e-01 2.75505722e-01 9.17767882e-01
-1.13064818e-01 4.68147457e-01 1.04329705e-01 -5.51574707e-01
-1.21220410e+00 5.12457073e-01 1.26812652e-01 -1.52665585e-01
-4.61695045e-01 6.83461368e-01 1.45600393e-01 -8.83710504e-01
2.41635181e-02 -3.31531733e-01 -3.32696170e-01 3.81658316e-01
3.04437250e-01 9.62628275e-02 7.13840723e-02 -6.37569249e-01
-5.88420570e-01 8.39001834e-01 -2.38191962e-01 1.37685612e-01
1.51363730e+00 -1.57633230e-01 -5.11183023e-01 7.08735943e-01
2.11239624e+00 6.64581591e-03 -6.49367213e-01 -3.16061348e-01
2.93111444e-01 -4.37253326e-01 -2.68186659e-01 -2.41837114e-01
-7.13835597e-01 8.22978437e-01 -5.34841046e-02 3.71011078e-01
6.23998523e-01 3.82506639e-01 9.49176908e-01 4.53060925e-01
-1.26966774e-01 -1.33560264e+00 3.85549217e-01 8.58853936e-01
6.84220433e-01 -1.28211665e+00 -1.71728984e-01 -1.55724823e-01
-8.26016307e-01 1.12906897e+00 4.79526639e-01 -2.95486420e-01
6.37172937e-01 -4.53792177e-02 -9.92336031e-03 -1.64531454e-01
-8.42652261e-01 3.48272175e-02 3.05181146e-01 4.68187958e-01
7.41031766e-01 -2.85387993e-01 -3.73296559e-01 5.63700080e-01
-3.47745985e-01 -4.78413969e-01 5.19503772e-01 6.16925120e-01
-5.23381889e-01 -1.15439284e+00 1.60695717e-01 6.96229875e-01
-2.65297621e-01 -3.67798537e-01 -4.73971039e-01 4.89043057e-01
-7.96045586e-02 1.06617868e+00 7.22007632e-01 -4.64740247e-01
1.41647890e-01 1.14533253e-01 2.69257724e-01 -9.43530738e-01
-8.52922559e-01 -7.06203818e-01 3.17203641e-01 -3.02774042e-01
-6.06581926e-01 -9.19303000e-01 -1.28447402e+00 -3.15937459e-01
1.52410761e-01 7.07684994e-01 5.25152743e-01 1.06597364e+00
5.19167840e-01 2.75028706e-01 1.17728686e+00 -4.89327997e-01
-6.72209799e-01 -1.14807761e+00 -5.35837114e-01 1.05141675e+00
2.07959905e-01 -3.12927961e-01 -6.80547237e-01 -1.66405156e-01] | [11.033638954162598, 8.801326751708984] |
fd22ca6c-2630-4337-b707-7a448eee1382 | does-your-model-classify-entities-reasonably | 2205.12640 | null | https://arxiv.org/abs/2205.12640v2 | https://arxiv.org/pdf/2205.12640v2.pdf | Does Your Model Classify Entities Reasonably? Diagnosing and Mitigating Spurious Correlations in Entity Typing | Entity typing aims at predicting one or more words that describe the type(s) of a specific mention in a sentence. Due to shortcuts from surface patterns to annotated entity labels and biased training, existing entity typing models are subject to the problem of spurious correlations. To comprehensively investigate the faithfulness and reliability of entity typing methods, we first systematically define distinct kinds of model biases that are reflected mainly from spurious correlations. Particularly, we identify six types of existing model biases, including mention-context bias, lexical overlapping bias, named entity bias, pronoun bias, dependency bias, and overgeneralization bias. To mitigate model biases, we then introduce a counterfactual data augmentation method. By augmenting the original training set with their debiased counterparts, models are forced to fully comprehend sentences and discover the fundamental cues for entity typing, rather than relying on spurious correlations for shortcuts. Experimental results on the UFET dataset show our counterfactual data augmentation approach helps improve generalization of different entity typing models with consistently better performance on both the original and debiased test sets. | ['Muhao Chen', 'Mingtao Dong', 'Bangzheng Li', 'Fei Wang', 'Nan Xu'] | 2022-05-25 | null | null | null | null | ['entity-typing'] | ['natural-language-processing'] | [ 1.52335703e-01 4.08621073e-01 -6.45460546e-01 -7.97602534e-01
-3.87685001e-01 -5.62208593e-01 5.46222031e-01 1.70782432e-01
-6.17307603e-01 1.07825124e+00 2.92819887e-01 -3.76749247e-01
2.16061175e-01 -8.11651230e-01 -6.73773110e-01 -1.06597044e-01
-4.90557998e-02 5.16865790e-01 -9.07632634e-02 -2.57949471e-01
2.78215796e-01 -2.02243187e-04 -1.30878568e+00 4.19203401e-01
1.31954491e+00 5.94507873e-01 -7.52270743e-02 2.98766434e-01
-4.41517204e-01 5.86281538e-01 -8.19394469e-01 -1.04142737e+00
-1.80353925e-01 -1.21560171e-01 -9.20950711e-01 -3.67990166e-01
5.52680790e-01 -1.00695744e-01 -7.21275285e-02 1.13303423e+00
4.32720900e-01 3.85784283e-02 8.45401227e-01 -1.25285792e+00
-1.18963313e+00 1.09912741e+00 -4.26653028e-01 4.52913970e-01
5.05423844e-01 -4.26532440e-02 1.48836374e+00 -1.19382322e+00
8.40030849e-01 1.20724416e+00 1.07152605e+00 1.00237036e+00
-1.52175164e+00 -8.35056961e-01 6.22433245e-01 8.73414800e-02
-1.22348750e+00 -4.68788892e-01 6.56060636e-01 -4.27300513e-01
1.04370344e+00 4.63603199e-01 2.38532461e-02 1.81287479e+00
-2.48804182e-01 8.53663206e-01 1.25112236e+00 -4.57063913e-01
7.02689961e-02 6.11419201e-01 9.40592468e-01 4.31522727e-01
7.92934239e-01 1.53472707e-01 -6.67640746e-01 -3.69683743e-01
1.75169244e-01 -3.99043530e-01 -4.10825193e-01 -3.00156642e-02
-1.16061485e+00 5.51004112e-01 2.51510024e-01 2.72691131e-01
-2.41646886e-01 -2.50547558e-01 4.74547267e-01 2.17378154e-01
5.79588234e-01 1.06967330e+00 -1.02755868e+00 3.81466150e-02
-5.42533219e-01 4.85547423e-01 9.59140420e-01 1.35336792e+00
8.61404359e-01 -1.53442249e-01 -6.39118493e-01 1.20542145e+00
7.79369771e-02 6.89549088e-01 6.90879107e-01 -4.08125579e-01
6.68114841e-01 6.96952283e-01 3.45564425e-01 -8.80978644e-01
-5.38552284e-01 -6.21546686e-01 -6.93446219e-01 -5.26280701e-01
4.11733121e-01 -3.50263864e-01 -6.75453782e-01 2.22608876e+00
1.72272071e-01 -6.20227605e-02 5.75373210e-02 7.11496294e-01
9.61397886e-01 -5.84192090e-02 5.35421789e-01 -7.78685883e-02
1.44236076e+00 -6.31510675e-01 -8.92314017e-01 -5.38204372e-01
1.03368390e+00 -3.91079307e-01 1.51985610e+00 -1.95543572e-01
-6.87429488e-01 -3.72432202e-01 -9.88831639e-01 -5.54419644e-02
-6.04654133e-01 9.18498845e-04 7.11870790e-01 7.54806459e-01
-2.72481114e-01 6.65116727e-01 -2.92349964e-01 -1.69812158e-01
2.41288871e-01 -5.59081836e-03 -3.05577725e-01 6.89154565e-02
-1.80235350e+00 1.11175394e+00 4.10772800e-01 -9.37673524e-02
-2.48283759e-01 -1.07570314e+00 -1.01049590e+00 1.59906745e-02
4.48578238e-01 -7.73989558e-01 1.23764050e+00 -7.36847878e-01
-7.69423366e-01 1.05605328e+00 -6.19925916e-01 -3.52186531e-01
3.91849548e-01 -3.39383841e-01 -5.93912601e-01 -6.77028298e-01
5.10482848e-01 2.23399475e-01 3.88796479e-01 -1.36285758e+00
-5.74686646e-01 -2.11632565e-01 1.18983559e-01 1.21987969e-01
-3.15471143e-01 -5.37267653e-03 1.60239339e-01 -9.27873254e-01
-9.08943415e-02 -7.62336195e-01 5.50834388e-02 -6.89703286e-01
-8.67135823e-01 -5.81117034e-01 2.73394078e-01 -4.99417335e-01
1.62947643e+00 -1.99811721e+00 -1.95296690e-01 -1.26483180e-02
2.20486179e-01 9.33115333e-02 -6.12030476e-02 1.39815554e-01
-2.76587248e-01 6.63813651e-01 -2.42472187e-01 -4.29584235e-01
1.29071027e-01 3.33983570e-01 -4.93357211e-01 -7.90150613e-02
3.69993746e-01 8.96307111e-01 -1.11822486e+00 -3.97928655e-01
-5.24397910e-01 -2.10880965e-01 -6.07684195e-01 2.82628834e-01
-1.59476116e-01 1.09136015e-01 -2.31937602e-01 6.30176067e-01
6.40319884e-01 -1.63452730e-01 3.90597582e-01 -3.00973356e-01
5.94903119e-02 1.01833916e+00 -9.69838202e-01 1.15188217e+00
-5.77850461e-01 3.93947452e-01 -2.92566091e-01 -4.99546230e-01
7.27265060e-01 1.72325581e-01 -1.53319150e-01 -4.41408277e-01
-1.72998592e-01 3.69059056e-01 1.20921351e-01 -8.32525134e-01
7.83165097e-01 -4.75189894e-01 -3.23902458e-01 5.85938752e-01
2.81388145e-02 5.16402781e-01 1.61195740e-01 2.46708155e-01
7.98527777e-01 -8.42951313e-02 3.40482771e-01 -3.70693594e-01
1.39335692e-01 -5.86584993e-02 1.06263256e+00 1.07542813e+00
-1.39208421e-01 2.68011242e-01 5.70456743e-01 -2.45508626e-01
-9.06062186e-01 -8.23181152e-01 -5.69118321e-01 1.44983315e+00
2.55657315e-01 -4.70594227e-01 -4.65169400e-01 -1.31901991e+00
3.38221222e-01 1.28191137e+00 -1.05832088e+00 -3.08146983e-01
-5.98955870e-01 -1.17189002e+00 8.95698309e-01 9.44846630e-01
2.86875486e-01 -1.03002787e+00 -7.46562704e-02 7.58499056e-02
-4.97980893e-01 -1.00541663e+00 -4.91948575e-01 2.48975366e-01
-4.86359835e-01 -1.07539523e+00 -3.19021910e-01 -5.58104515e-01
7.95098305e-01 5.12580350e-02 1.53353953e+00 3.69716495e-01
1.80603445e-01 -1.43884078e-01 -4.18490201e-01 -6.22074187e-01
-4.35812920e-01 3.39137107e-01 4.37201679e-01 -2.72413492e-01
9.38045681e-01 -2.72847295e-01 -2.05931678e-01 2.61325210e-01
-4.92980391e-01 -8.47177804e-02 5.80129981e-01 1.23783600e+00
3.59106474e-02 -5.63641191e-01 8.11190128e-01 -1.86052918e+00
8.24128568e-01 -7.25805700e-01 -3.29538509e-02 4.86901551e-01
-1.10560668e+00 2.95668483e-01 4.54607904e-01 -6.79823220e-01
-1.46666050e+00 -5.67094684e-01 -1.30083680e-01 1.95433736e-01
-2.42491126e-01 6.64563656e-01 -4.34356481e-01 5.02802432e-01
9.38303769e-01 6.34746626e-02 -4.11615968e-01 -6.33514285e-01
3.26729774e-01 7.03184485e-01 4.61831570e-01 -1.10717428e+00
7.31058180e-01 2.66709249e-03 -5.48973620e-01 -5.82491040e-01
-1.37367237e+00 -2.60549009e-01 -7.64020264e-01 2.18685895e-01
4.46534425e-01 -6.81965768e-01 -4.68405724e-01 4.00361836e-01
-1.38920164e+00 -1.31828606e-01 -4.75472808e-02 2.53587455e-01
5.34867235e-02 3.49396557e-01 -5.95282137e-01 -7.74979353e-01
-1.51948795e-01 -5.13351560e-01 8.26790512e-01 3.84674780e-02
-8.02967250e-01 -1.12894642e+00 8.07503834e-02 1.34528160e-01
3.66840720e-01 -1.84598118e-02 1.39192390e+00 -1.32766545e+00
-1.26976222e-01 -2.10278675e-01 -1.80669904e-01 2.21494198e-01
1.00997396e-01 -6.89201280e-02 -1.04950392e+00 1.36510059e-02
-1.19804852e-01 -1.62779540e-01 7.95427740e-01 -1.20719358e-01
1.03233254e+00 -7.33604670e-01 -6.15901709e-01 5.37115514e-01
1.18313777e+00 1.06727649e-02 2.48206541e-01 3.88480335e-01
6.78005397e-01 7.32680559e-01 6.11881495e-01 9.37826782e-02
6.69901431e-01 5.44641614e-01 -1.11276440e-01 4.78141122e-02
5.40183075e-02 -6.77987635e-01 2.13962281e-03 5.53723633e-01
-2.45457590e-02 -1.19047627e-01 -9.53193843e-01 7.11545825e-01
-1.56857705e+00 -9.65035677e-01 -2.67704606e-01 2.18421125e+00
1.39956343e+00 2.73648858e-01 -1.09856308e-01 -6.84219748e-02
9.10084367e-01 1.15179487e-01 -5.86530566e-01 -2.45198816e-01
-2.97120631e-01 -1.31978333e-01 3.97745013e-01 5.95216930e-01
-9.83734667e-01 1.07349980e+00 6.66352510e+00 3.28942358e-01
-8.29649746e-01 1.86066516e-02 4.74493355e-01 1.41087785e-01
-7.02510476e-01 -1.25802413e-01 -1.15379560e+00 7.02221870e-01
5.86395979e-01 -4.56068724e-01 9.77647305e-02 9.90712583e-01
-2.91746557e-01 1.82610452e-01 -1.63727236e+00 4.98981357e-01
3.34743448e-02 -1.11973834e+00 2.02427626e-01 -1.25816405e-01
6.57843113e-01 -1.99772626e-01 -1.45126224e-01 8.05814266e-01
4.51604486e-01 -9.83454049e-01 7.43280709e-01 3.24042857e-01
8.97027254e-01 -3.56636435e-01 1.03283346e+00 4.80931044e-01
-5.77635169e-01 -4.97558936e-02 -3.97800744e-01 -2.41745487e-01
3.83169763e-02 5.89643180e-01 -9.94634151e-01 3.53375584e-01
4.78510529e-01 4.37509000e-01 -6.42313480e-01 6.11746550e-01
-5.64273417e-01 7.08082318e-01 9.18949619e-02 -3.51468861e-01
-3.88968408e-01 3.05941761e-01 6.59108639e-01 1.64127278e+00
-1.91852435e-01 1.21779256e-01 -2.44363397e-01 1.10104156e+00
-5.10597587e-01 -1.35654151e-01 -7.72736013e-01 1.37487113e-01
1.21570039e+00 1.07585156e+00 -1.05144382e-01 -5.76398194e-01
-5.04480064e-01 7.86838889e-01 8.91662061e-01 5.92833698e-01
-7.27109671e-01 -4.60310280e-01 8.91077578e-01 -1.30111985e-02
3.90851870e-04 1.27932638e-01 -9.72615957e-01 -1.57317841e+00
2.06607997e-01 -9.56219912e-01 4.84305114e-01 -4.87518191e-01
-1.89520442e+00 5.67025781e-01 -7.29802772e-02 -6.73454881e-01
-7.87478238e-02 -5.97254694e-01 -6.52208328e-01 1.06709683e+00
-1.51852643e+00 -8.51882458e-01 -2.42848888e-01 1.45286620e-02
4.44270581e-01 -1.44515941e-02 1.02128375e+00 3.67441386e-01
-8.68463397e-01 1.24176669e+00 -2.73241699e-01 6.94258749e-01
1.05346870e+00 -1.69813907e+00 7.15500176e-01 8.29097509e-01
-5.23509122e-02 1.47994113e+00 8.57630610e-01 -8.33752573e-01
-8.05408061e-01 -1.04291666e+00 1.60083449e+00 -1.08898568e+00
6.83941782e-01 -5.32558858e-01 -1.28768981e+00 1.11675501e+00
-1.09319173e-01 -3.80279392e-01 9.92320418e-01 1.17119515e+00
-9.85823750e-01 8.83927047e-02 -1.03619659e+00 6.59896076e-01
1.39863408e+00 -6.70675457e-01 -1.24819052e+00 1.37472540e-01
7.57400095e-01 -4.18430418e-01 -6.22797310e-01 5.43265104e-01
5.54707587e-01 -6.23685837e-01 6.77527428e-01 -1.54712319e+00
5.81561148e-01 7.23630637e-02 -6.17551059e-02 -1.61362875e+00
-5.40598392e-01 -2.25271285e-01 -1.49024025e-01 1.71418941e+00
1.14130247e+00 -7.34687209e-01 5.70338011e-01 1.26863146e+00
-1.09567881e-01 -7.21826911e-01 -7.10481107e-01 -9.66423810e-01
2.19480857e-01 -3.90338868e-01 9.47327197e-01 1.62781370e+00
5.12198031e-01 7.38706589e-01 -2.81841367e-01 2.22562358e-01
5.91744363e-01 1.57787770e-01 6.13698244e-01 -1.15797937e+00
-2.09993765e-01 -3.78745288e-01 1.37803286e-01 -1.11115050e+00
5.22718847e-01 -1.01557016e+00 9.81720164e-02 -1.03279555e+00
4.95070666e-01 -8.91614377e-01 -1.98977172e-01 4.44564223e-01
-1.07047045e+00 -1.36526302e-01 -1.67461202e-01 1.48162767e-01
-3.51751238e-01 3.01962942e-01 8.33502650e-01 -6.71505183e-03
9.98017043e-02 9.35217813e-02 -1.12110758e+00 8.14431727e-01
5.05894065e-01 -4.64797080e-01 -1.76256746e-01 -6.07769847e-01
6.03480220e-01 -3.22199374e-01 2.32121259e-01 -2.00533167e-01
5.49228229e-02 -2.35751480e-01 2.31701791e-01 -6.13971688e-02
-7.24671185e-02 -4.02862906e-01 -3.53317320e-01 1.73338905e-01
-9.19413090e-01 1.23645261e-01 -3.78085487e-02 6.11413121e-01
-2.59221867e-02 -4.78604257e-01 5.23760080e-01 -1.05398990e-01
-6.93006039e-01 -4.56426106e-02 -1.70830637e-01 6.88917160e-01
5.25630236e-01 -1.14513906e-02 -8.22992027e-01 1.20774984e-01
-6.74087822e-01 2.74888128e-01 4.00229812e-01 6.15914464e-01
2.27953643e-01 -1.28982472e+00 -7.29709208e-01 2.56603062e-01
4.44789022e-01 -5.26859388e-02 6.01241700e-02 6.18538976e-01
3.51975530e-01 4.02960718e-01 7.64824301e-02 -2.33421400e-01
-1.24291337e+00 5.83976388e-01 3.86627764e-01 -1.21789090e-01
-2.20463634e-01 1.10535073e+00 2.82205313e-01 -7.13454306e-01
2.76800841e-01 -4.81261224e-01 -1.79580376e-01 1.20617174e-01
4.11611795e-01 3.96439284e-01 7.46177137e-02 -3.17524821e-01
-4.70378369e-01 -3.76428030e-02 -4.02835518e-01 1.99527979e-01
9.71045315e-01 -1.48029670e-01 -6.15283772e-02 6.74217343e-01
7.91855812e-01 4.04550701e-01 -7.25906253e-01 -4.73634094e-01
6.93257630e-01 -5.07758915e-01 -5.34495413e-01 -1.27207744e+00
-4.07533646e-01 5.38226724e-01 -7.38860518e-02 3.88790458e-01
3.72623116e-01 2.41612308e-02 6.65372789e-01 4.81726766e-01
4.79119778e-01 -1.10848808e+00 -4.10551041e-01 7.41138458e-01
8.17452490e-01 -1.39217734e+00 -2.47332796e-01 -6.85439169e-01
-8.49405110e-01 5.81539750e-01 1.15383601e+00 2.69637614e-01
3.59877348e-01 3.44612360e-01 1.63273573e-01 -1.21941291e-01
-1.01921296e+00 -5.48082478e-02 3.73847991e-01 6.70964777e-01
9.49411750e-01 2.03084856e-01 -6.53950334e-01 1.32179499e+00
-3.45142663e-01 -4.12246287e-01 4.01664853e-01 4.81888920e-01
-1.00126795e-01 -9.87326920e-01 1.71496440e-02 7.03567564e-01
-5.03301799e-01 -5.91350138e-01 -7.38880932e-01 8.50063920e-01
9.65632796e-02 6.57336533e-01 -3.97873521e-02 -4.35249507e-01
6.80185139e-01 6.11308455e-01 2.65238881e-01 -9.32027876e-01
-5.97591579e-01 -8.47799659e-01 7.72425950e-01 -1.38139889e-01
-9.86314714e-02 -7.36910760e-01 -1.09166741e+00 -2.68044829e-01
-6.08766019e-01 3.43782932e-01 3.47916842e-01 1.08588398e+00
4.95465785e-01 3.37141842e-01 4.01361138e-01 -1.54799402e-01
-9.65126514e-01 -1.46841681e+00 -4.02234644e-01 9.94585574e-01
2.95937181e-01 -8.95110250e-01 -5.07833123e-01 -1.97844028e-01] | [9.78161907196045, 8.503780364990234] |
df0086a6-286b-4a1b-8860-3984ee582025 | automatic-acne-object-detection-and-acne | null | null | https://www.mdpi.com/2075-4418/12/8/1879 | https://www.mdpi.com/2075-4418/12/8/1879 | Automatic Acne Object Detection and Acne Severity Grading Using Smartphone Images and Artificial Intelligence | Skin image analysis using artificial intelligence (AI) has recently attracted significant research interest, particularly for analyzing skin images captured by mobile devices. Acne is one of the most common skin conditions with profound effects in severe cases. In this study, we developed an AI system called AcneDet for automatic acne object detection and acne severity grading using facial images captured by smartphones. AcneDet includes two models for two tasks: (1) a Faster R-CNN-based deep learning model for the detection of acne lesion objects of four types, including blackheads/whiteheads, papules/pustules, nodules/cysts, and acne scars; and (2) a LightGBM machine learning model for grading acne severity using the Investigator’s Global Assessment (IGA) scale. The output of the Faster R-CNN model, i.e., the counts of each acne type, were used as input for the LightGBM model for acne severity grading. A dataset consisting of 1572 labeled facial images captured by both iOS and Android smartphones was used for training. The results show that the Faster R-CNN model achieves a mAP of 0.54 for acne object detection. The mean accuracy of acne severity grading by the LightGBM model is 0.85. With this study, we hope to contribute to the development of artificial intelligent systems to help acne patients better understand their conditions and support doctors in acne diagnosis.
Keywords: deep learning; smartphone image; acne grading; acne object detection | ['Hoan Thanh Ngo', 'Trung Xuan Ngo', 'Tsuyoshi Ishii', 'Kazuhiro Tsuji', 'Kazuma Suda', 'Anh Tam Nguyen', 'Nga Thi Vu', 'Hoan Tam Nguyen', 'Mai Thi-Thanh Tran', 'Nhu-Thuy Trinh', 'Lua Thi Ngo', 'Hieu Xuan Le', 'Phuc Hoang Nguyen', 'Quan Thanh Huynh'] | 2022-08-03 | null | null | null | diagnostics-2022-8 | ['medical-object-detection', 'acne-severity-grading'] | ['computer-vision', 'medical'] | [ 3.75667155e-01 -3.16002965e-01 -3.11080813e-01 -6.79224059e-02
-6.66891575e-01 -4.30245608e-01 2.82630831e-01 4.21547405e-02
-4.05722708e-02 2.76410013e-01 -1.18310280e-01 4.64183986e-02
2.26917312e-01 -9.19881284e-01 -2.04895839e-01 -1.00399137e+00
4.21179086e-03 2.05830798e-01 -1.34773597e-01 -4.05510329e-02
-8.51183906e-02 7.26551354e-01 -1.53078771e+00 6.07084334e-01
9.92135644e-01 1.76750898e+00 -2.08308756e-01 1.13963735e+00
4.27197255e-02 6.71312809e-01 -7.02022314e-01 -3.41687977e-01
2.10385695e-02 -1.71469450e-01 -4.94478434e-01 4.79835980e-02
1.07163930e+00 -5.74538291e-01 -7.54151121e-02 9.41032767e-01
7.11520255e-01 -4.12327141e-01 7.72565961e-01 -1.37432432e+00
-4.58462864e-01 -3.07781994e-01 -7.87888646e-01 1.29559606e-01
1.71826139e-01 3.67312640e-01 4.75850254e-01 -7.46503711e-01
3.73539895e-01 1.01861632e+00 8.80699813e-01 8.93872321e-01
-6.69174671e-01 -8.48941267e-01 -4.79384810e-01 4.35237736e-01
-1.45645666e+00 -7.28842914e-02 2.65711725e-01 -4.79337364e-01
6.13556206e-01 5.09517074e-01 1.16117048e+00 8.38916421e-01
3.24923396e-01 9.36489642e-01 1.38228250e+00 -9.15456861e-02
3.64635915e-01 -1.06376139e-02 -1.98478341e-01 9.09447789e-01
-1.26960101e-02 -2.60007203e-01 -1.35386795e-01 -3.05006623e-01
1.00371993e+00 3.14050347e-01 2.02656373e-01 4.84097689e-01
-3.55304092e-01 8.47213089e-01 6.79317653e-01 2.04295274e-02
-6.57212436e-01 7.24357441e-02 2.58799762e-01 -5.52469678e-02
6.14144266e-01 1.39247313e-01 -2.62903571e-01 2.39343792e-01
-6.80653214e-01 1.73440441e-01 3.80853593e-01 3.30259889e-01
1.92421511e-01 -2.71369293e-02 -2.48868003e-01 1.35073256e+00
1.30216435e-01 8.06585312e-01 3.76857638e-01 -7.24885702e-01
-1.51059687e-01 1.12511861e+00 -3.35292369e-01 -6.95028007e-01
-3.78429562e-01 5.22024296e-02 -1.04028261e+00 4.81828123e-01
3.45877498e-01 -3.42277229e-01 -1.38173747e+00 1.14709640e+00
4.25580919e-01 1.73311323e-01 -4.42036986e-01 1.23318255e+00
1.07471168e+00 6.14870071e-01 5.56189895e-01 3.18987429e-01
1.70740712e+00 -7.63695657e-01 -7.82120347e-01 1.77079111e-01
6.88233674e-01 -7.50558794e-01 9.54483390e-01 7.10789502e-01
-1.01525414e+00 -1.67394981e-01 -8.41634512e-01 -1.28919363e-01
-7.13520348e-01 6.87457085e-01 9.91565347e-01 7.19582200e-01
-1.35809958e+00 4.71491553e-03 -4.32615846e-01 -7.03697741e-01
1.05538237e+00 5.90228319e-01 -4.90375340e-01 -2.11865097e-01
-9.89131093e-01 7.90469587e-01 -4.39909101e-01 1.46884218e-01
-9.33187187e-01 -8.37690532e-01 -6.45282507e-01 -1.44054502e-01
1.92868724e-01 -6.40248179e-01 1.20966744e+00 -1.27002168e+00
-1.25018930e+00 1.01829147e+00 -1.35457233e-01 -1.52891889e-01
1.20308153e-01 -1.51358381e-01 -7.16274023e-01 4.80631381e-01
-2.34007731e-01 8.53219032e-01 9.33829963e-01 -9.66060221e-01
-9.67353463e-01 -8.31230044e-01 -1.58381358e-01 2.89269560e-03
-3.58254343e-01 5.38676716e-02 -2.21168563e-01 -4.79468197e-01
-6.36999905e-01 -9.86073911e-01 -1.73967555e-01 7.29236126e-01
-2.92412102e-01 -5.98418534e-01 1.01866376e+00 -1.08593941e+00
1.23425174e+00 -1.79697120e+00 -4.95464772e-01 4.11748827e-01
6.10105395e-01 8.82810175e-01 -6.42554402e-01 -3.75374965e-02
-2.47847587e-01 2.71595359e-01 1.39509197e-02 6.41280487e-02
-4.61319536e-01 -2.52674937e-01 1.58379510e-01 2.20695302e-01
6.32405162e-01 9.34018552e-01 -7.55729020e-01 -4.75664884e-01
4.08368558e-01 8.88343930e-01 -1.25619054e-01 2.10280165e-01
-2.13687420e-01 -2.25945532e-01 -1.41593426e-01 1.35418010e+00
8.03792179e-01 -3.28443199e-01 -1.37863934e-01 -3.30869228e-01
1.45462975e-01 -2.80068010e-01 -8.00480008e-01 7.75141001e-01
-6.76955342e-01 7.74677634e-01 4.51326013e-01 5.03049120e-02
6.43184602e-01 3.93695742e-01 5.44538736e-01 -6.62375033e-01
5.07645130e-01 -4.84377928e-02 -1.79771706e-01 -7.76497185e-01
-5.91110885e-02 -1.60064101e-02 5.84670484e-01 5.73817082e-02
-1.12541534e-01 -6.42892197e-02 1.23248488e-01 -9.68623012e-02
9.97547388e-01 -5.80329001e-01 3.47584844e-01 2.26281196e-01
3.97528291e-01 1.00982048e-01 1.42159298e-01 3.07946652e-01
-3.04697990e-01 5.62230527e-01 4.81044531e-01 -8.56456876e-01
-8.74079049e-01 -1.10073686e+00 -2.66964912e-01 9.47299242e-01
-2.11134464e-01 -1.71083733e-01 -1.23535693e+00 -6.84629142e-01
7.09747449e-02 6.03217706e-02 -7.96739578e-01 7.56579265e-02
1.10144634e-02 -8.59372497e-01 5.75103462e-01 6.51643515e-01
6.23494685e-01 -1.08284628e+00 -2.64330232e-03 -3.06250662e-01
1.72128703e-03 -6.45658731e-01 -5.02724946e-01 -5.06230533e-01
-5.02152085e-01 -1.43471813e+00 -1.17032945e+00 -9.58936155e-01
8.90100002e-01 8.21137130e-02 5.29459834e-01 2.24009991e-01
-1.29483712e+00 4.97616440e-01 -3.10543887e-02 -8.01257670e-01
-8.52164775e-02 -2.78724611e-01 1.01498626e-01 3.35864186e-01
7.83710420e-01 4.94300434e-03 -1.07351184e+00 9.00102109e-02
-1.07416010e+00 -1.65350571e-01 8.04869950e-01 4.62919503e-01
7.53516793e-01 -8.40887874e-02 3.60300630e-01 -6.92073703e-01
7.83102691e-01 -5.77904344e-01 -4.70314890e-01 2.16545492e-01
-4.03901309e-01 -9.83391464e-01 4.51087475e-01 -5.10173678e-01
-8.73835206e-01 1.43190041e-01 -2.63761103e-01 -4.36122447e-01
-3.32667381e-01 6.22993186e-02 1.10856742e-01 -3.66385967e-01
5.93464375e-01 6.98976638e-03 3.95936549e-01 -1.11876316e-01
2.84085795e-02 1.24990702e+00 3.53324205e-01 1.00095026e-01
2.50461429e-01 5.55975974e-01 1.38917312e-01 -1.34935248e+00
-6.52983248e-01 -6.01163983e-01 7.61559187e-03 -5.69764614e-01
1.08195472e+00 -1.06785643e+00 -1.01584280e+00 1.15533030e+00
-9.58559811e-01 -5.20479202e-01 5.53874113e-02 -1.00865446e-01
-1.31782353e-01 9.99270678e-02 -8.53286922e-01 -9.49692369e-01
-1.03841352e+00 -7.95535147e-01 1.37128210e+00 8.98870528e-01
-3.19182903e-01 -1.01734495e+00 1.42818302e-01 8.84472311e-01
5.08821249e-01 2.52047867e-01 9.33164597e-01 -3.31651568e-01
-2.21222937e-01 -6.98864222e-01 -5.67659020e-01 7.30967283e-01
3.45922470e-01 4.00582939e-01 -1.19586432e+00 -4.01704535e-02
-5.55821359e-01 -3.91517639e-01 7.90359378e-01 7.18360543e-01
1.90031362e+00 -5.66048205e-01 -2.74157584e-01 5.42812526e-01
1.40240824e+00 4.60468441e-01 1.04734933e+00 -2.28376925e-01
6.67911291e-01 5.16222060e-01 5.45187593e-01 3.07049632e-01
2.85728157e-01 2.48600855e-01 5.99114418e-01 -9.75141704e-01
-3.73704642e-01 -1.65932439e-02 9.71010551e-02 1.20721452e-01
-5.36273301e-01 -2.99964219e-01 -7.75743246e-01 6.45508945e-01
-1.17118812e+00 -7.01693773e-01 -1.41931027e-01 1.88072300e+00
6.50411189e-01 -6.01090431e-01 3.54141623e-01 -2.06290595e-02
9.63087559e-01 -1.69049129e-01 -6.00065768e-01 -6.66871965e-01
5.27882110e-03 6.28963113e-01 3.89040947e-01 1.27752528e-01
-1.22562206e+00 8.45796049e-01 6.00760794e+00 1.00031269e+00
-1.74976623e+00 -1.13645554e-01 9.90208507e-01 -1.36556432e-01
1.04796797e-01 -6.93324447e-01 -6.68195724e-01 6.77916646e-01
4.07351524e-01 3.32694441e-01 6.13288999e-01 9.76666331e-01
1.73201799e-01 -3.14587742e-01 -4.96793061e-01 1.07958424e+00
2.62422681e-01 -1.32289994e+00 4.08986360e-01 8.56244192e-02
7.60799289e-01 1.43046984e-02 4.45003957e-01 -7.87678212e-02
3.67702805e-02 -1.37415111e+00 -3.96304935e-01 6.79270267e-01
1.64891648e+00 -6.73907280e-01 1.07314157e+00 -4.30848956e-01
-1.00704074e+00 -1.72682002e-01 -7.89722800e-03 2.31085479e-01
-3.74436498e-01 5.46149313e-01 -1.34640157e+00 -4.53802854e-01
8.26659083e-01 3.75805497e-01 -6.87020540e-01 1.27085793e+00
-6.60537556e-02 6.51469767e-01 -2.89615065e-01 -4.76942241e-01
7.22471178e-02 -9.37174335e-02 3.37644666e-01 1.20515513e+00
2.09944516e-01 2.95450836e-01 -3.70232463e-01 5.86576104e-01
-8.51795226e-02 1.39403492e-01 -3.76550972e-01 -4.50916350e-01
2.46894926e-01 1.96711552e+00 -3.72253329e-01 -1.83268845e-01
-1.76445141e-01 6.66509092e-01 -3.12818676e-01 2.58980364e-01
-4.92541164e-01 -5.76457381e-01 9.31610405e-01 7.47384191e-01
-1.76584780e-01 4.92588103e-01 -2.51985103e-01 -5.50547063e-01
-4.52382058e-01 -8.60769629e-01 7.37517059e-01 -1.21500564e+00
-1.66295362e+00 3.42453450e-01 -6.56107306e-01 -8.21806729e-01
1.13853745e-01 -1.12426329e+00 -1.08083415e+00 7.08625257e-01
-1.55990529e+00 -1.60626018e+00 -1.06081772e+00 6.91545725e-01
6.17318273e-01 -2.41987735e-01 9.14945960e-01 8.35891441e-02
-6.66120410e-01 5.29781461e-01 -4.29862410e-01 4.12972361e-01
7.86152005e-01 -1.46404219e+00 -1.76205620e-01 2.07543164e-01
-6.43853366e-01 4.45959359e-01 -1.37247741e-01 -7.93089926e-01
-1.22808993e+00 -1.42262733e+00 6.02069676e-01 -3.41713130e-01
5.09951174e-01 -4.50816564e-02 -4.68990147e-01 2.16916099e-01
1.27646834e-01 1.32441431e-01 1.14241219e+00 -1.27018318e-01
-2.27905735e-01 -4.58587408e-01 -1.51354074e+00 7.37904012e-01
4.24735546e-01 -3.55430424e-01 1.71509221e-01 7.11841464e-01
1.35401078e-02 -2.71404028e-01 -1.21060407e+00 5.27187288e-01
9.25900996e-01 -5.02866745e-01 7.20855355e-01 -4.67269570e-01
8.73579681e-01 8.43696520e-02 2.08189622e-01 -1.22344005e+00
-2.47617498e-01 -3.27283502e-01 -1.56154111e-01 7.57870436e-01
4.31212038e-02 -5.83300054e-01 8.73089373e-01 3.44361186e-01
2.13615656e-01 -1.25718629e+00 -6.83452368e-01 5.26893958e-02
-2.52025455e-01 7.78655186e-02 5.57723999e-01 8.32153499e-01
-3.65500808e-01 1.35987252e-01 2.10349560e-01 6.89241588e-02
5.36244810e-01 -4.12777245e-01 5.19227743e-01 -1.14015543e+00
5.67837596e-01 -5.01210093e-01 -5.11336684e-01 -1.28354177e-01
-5.01166046e-01 -5.73065937e-01 -5.32128274e-01 -1.64719403e+00
3.71736348e-01 -3.58057231e-01 -2.98190325e-01 9.15492475e-01
7.91302845e-02 7.53268480e-01 -1.37319521e-03 -2.80534565e-01
-3.42797875e-01 -9.51202884e-02 1.50780630e+00 -3.65639210e-01
-1.61180183e-01 1.89688146e-01 -6.27724648e-01 9.05675352e-01
9.67042148e-01 1.43504053e-01 -7.54782138e-03 1.02519058e-02
2.36239776e-01 -4.43295091e-02 7.50354648e-01 -1.00978398e+00
2.26343587e-01 -2.68649310e-01 1.08416927e+00 -6.12663388e-01
3.83036554e-01 -6.00452602e-01 -2.14906573e-01 5.51915109e-01
-2.22989753e-01 -4.99710381e-01 3.39706570e-01 1.59091726e-01
-1.14885978e-01 2.64830470e-01 1.09071386e+00 -9.85951275e-02
-8.13542306e-01 8.09171498e-01 -6.48851931e-01 -4.95927155e-01
1.23358047e+00 -2.81153858e-01 -8.24491501e-01 -5.41416824e-01
-7.52853572e-01 2.92384028e-01 2.23197013e-01 5.78930318e-01
9.48081791e-01 -1.29335904e+00 -6.77637994e-01 4.13507104e-01
3.33585083e-01 2.46112552e-02 7.51021326e-01 8.26275587e-01
-8.15235198e-01 3.85474682e-01 -5.25830448e-01 -6.17352426e-01
-1.85890257e+00 -1.37285858e-01 6.29985809e-01 2.04138190e-01
1.75672382e-01 8.35222483e-01 2.01051697e-01 -1.70063302e-01
3.03018987e-01 -1.25502810e-01 -6.20958984e-01 1.40670672e-01
9.49327707e-01 7.64793873e-01 1.10251732e-01 -2.23163560e-01
-1.73089996e-01 7.79261172e-01 -2.74885118e-01 4.91600722e-01
1.15267348e+00 3.92882258e-01 -4.58944112e-01 -1.88811004e-01
9.32370901e-01 -1.40851378e-01 -9.89129186e-01 2.68876046e-01
-8.02911580e-01 -1.95458889e-01 4.71271753e-01 -1.56302786e+00
-1.29996109e+00 8.53172600e-01 1.46817362e+00 8.46851915e-02
1.57600439e+00 -6.51009753e-02 1.21387303e+00 -1.96400117e-02
-4.98797372e-02 -1.21248245e+00 2.39820942e-01 -1.05827138e-01
9.41942513e-01 -1.30317414e+00 -2.74112910e-01 -5.61695993e-01
-7.51464963e-01 1.19462967e+00 9.65261340e-01 -1.79652363e-01
3.62922758e-01 3.77411366e-01 4.94856060e-01 -3.63086522e-01
-5.31430364e-01 -3.76176089e-01 4.68435168e-01 8.99958253e-01
8.69457126e-02 4.51354980e-01 -7.39612952e-02 7.58923769e-01
1.32228032e-01 3.88273984e-01 3.27026993e-01 3.71829838e-01
-3.75564218e-01 -5.13395190e-01 -3.70765775e-01 1.15745473e+00
-4.98197913e-01 7.22346455e-02 -1.14478695e+00 7.65016854e-01
5.57315171e-01 7.76879013e-01 4.04553920e-01 -4.95959401e-01
-2.07845736e-02 -2.77474493e-01 4.63858873e-01 -4.37066585e-01
-6.21137440e-01 -4.94541824e-02 -8.64835605e-02 -6.18759334e-01
3.40927839e-02 -1.90729812e-01 -8.89811814e-01 -2.90250123e-01
-2.36557782e-01 -5.24105191e-01 1.10155272e+00 5.63111782e-01
3.37852925e-01 9.27066803e-02 7.19256461e-01 -3.49987537e-01
-2.63602939e-02 -9.86708939e-01 -8.46802533e-01 8.64818990e-02
4.42401886e-01 -2.44718626e-01 -2.63255000e-01 7.11703300e-02] | [15.677382469177246, -2.9852828979492188] |
b1fb6419-a2ed-4699-ba7e-ebdfa8c07f18 | two-level-graph-network-for-few-shot-class | 2303.13862 | null | https://arxiv.org/abs/2303.13862v1 | https://arxiv.org/pdf/2303.13862v1.pdf | Two-level Graph Network for Few-Shot Class-Incremental Learning | Few-shot class-incremental learning (FSCIL) aims to design machine learning algorithms that can continually learn new concepts from a few data points, without forgetting knowledge of old classes. The difficulty lies in that limited data from new classes not only lead to significant overfitting issues but also exacerbates the notorious catastrophic forgetting problems. However, existing FSCIL methods ignore the semantic relationships between sample-level and class-level. % Using the advantage that graph neural network (GNN) can mine rich information among few samples, In this paper, we designed a two-level graph network for FSCIL named Sample-level and Class-level Graph Neural Network (SCGN). Specifically, a pseudo incremental learning paradigm is designed in SCGN, which synthesizes virtual few-shot tasks as new tasks to optimize SCGN model parameters in advance. Sample-level graph network uses the relationship of a few samples to aggregate similar samples and obtains refined class-level features. Class-level graph network aims to mitigate the semantic conflict between prototype features of new classes and old classes. SCGN builds two-level graph networks to guarantee the latent semantic of each few-shot class can be effectively represented in FSCIL. Experiments on three popular benchmark datasets show that our method significantly outperforms the baselines and sets new state-of-the-art results with remarkable advantages. | ['Fenglei Xu', 'Zhenping Xia', 'Fuyuan Hu', 'Fan Lyu', 'Linyan Li', 'Hao Chen'] | 2023-03-24 | null | null | null | null | ['class-incremental-learning', 'few-shot-class-incremental-learning'] | ['computer-vision', 'methodology'] | [ 2.71846056e-01 2.64004290e-01 -5.05793333e-01 -3.89611959e-01
-2.42501706e-01 6.44498616e-02 2.67399520e-01 2.74725586e-01
-9.10103098e-02 8.02338719e-01 -1.02417797e-01 6.71211258e-02
-3.85746658e-01 -1.24346280e+00 -6.79383337e-01 -6.31380439e-01
-1.34633809e-01 4.74653661e-01 6.89633071e-01 -2.11828128e-01
-4.52857278e-02 2.95660108e-01 -1.90663278e+00 3.71126205e-01
1.01981783e+00 1.03195786e+00 3.34740251e-01 2.82189906e-01
-6.20639265e-01 9.51947391e-01 -5.31758070e-01 -3.07749093e-01
1.48327559e-01 -4.58592474e-01 -4.56641912e-01 2.30167672e-01
4.70382154e-01 -9.37676132e-02 -5.12595415e-01 1.16117263e+00
3.14183652e-01 5.80998600e-01 2.62532622e-01 -1.50385332e+00
-1.07803369e+00 8.61823857e-01 -4.79988247e-01 1.46652609e-01
-5.18330112e-02 3.06400329e-01 8.47208083e-01 -1.32723618e+00
7.56730795e-01 1.33278108e+00 8.25391054e-01 7.73444176e-01
-9.50844824e-01 -7.53394783e-01 7.68625557e-01 4.50849563e-01
-1.40010452e+00 -1.64493814e-01 9.80565429e-01 -9.80178565e-02
9.65700448e-01 -9.83560532e-02 8.09895635e-01 9.45174038e-01
1.95376769e-01 1.05379319e+00 6.29367530e-01 -2.22662628e-01
5.49237609e-01 1.15528315e-01 5.93169928e-01 8.86525869e-01
5.86917698e-01 -4.55038287e-02 -6.91911042e-01 1.76450759e-02
3.91696453e-01 6.83800757e-01 -1.25088543e-01 -6.99551821e-01
-5.82605720e-01 8.77824187e-01 6.54673278e-01 3.49672973e-01
-7.45410174e-02 -7.16777816e-02 4.42817003e-01 6.07856989e-01
7.18958259e-01 1.47547275e-01 -5.72787583e-01 2.22874239e-01
-7.62945950e-01 -2.81718578e-02 4.60302860e-01 1.29938769e+00
1.13906288e+00 3.65447819e-01 -2.70800442e-01 9.90270257e-01
-1.18774608e-01 2.47457311e-01 8.82954240e-01 -3.82488638e-01
2.75104046e-01 1.10096121e+00 -6.90360308e-01 -9.86633539e-01
-2.20459476e-01 -9.51300085e-01 -8.83705020e-01 -4.27621417e-02
-3.02159846e-01 4.25380021e-02 -1.28913736e+00 1.76395345e+00
3.80114317e-01 6.08091950e-01 1.79493305e-04 2.20482111e-01
1.05962980e+00 6.67954743e-01 -1.04930415e-03 -6.39034569e-01
7.44073927e-01 -1.18087149e+00 -5.58510482e-01 -5.53255796e-01
4.93482798e-01 8.60430580e-03 1.38977385e+00 3.93480152e-01
-6.88246489e-01 -8.82913113e-01 -1.31671238e+00 2.19349653e-01
-8.28602970e-01 -3.72079760e-01 8.20803940e-01 4.74810869e-01
-7.85814047e-01 8.57336164e-01 -5.24929583e-01 -3.67311835e-01
1.00586510e+00 1.96690679e-01 -8.49009678e-02 -5.41844964e-01
-1.18483460e+00 4.11211938e-01 9.50527191e-01 -2.03185022e-01
-9.74219739e-01 -9.51252937e-01 -9.54458356e-01 3.98574263e-01
9.76133704e-01 -4.81919527e-01 1.03965569e+00 -1.00232923e+00
-1.16538382e+00 3.23738128e-01 -2.57657841e-03 -4.60459232e-01
2.53440231e-01 7.09649250e-02 -7.45603740e-01 -1.57120407e-01
1.16804376e-01 4.80198413e-01 1.19343662e+00 -1.25007915e+00
-9.69574392e-01 -3.41360062e-01 -1.77074060e-01 1.11088701e-01
-7.83857226e-01 -8.48917127e-01 -3.88712972e-01 -7.03421235e-01
1.77387476e-01 -3.50604981e-01 -1.38117507e-01 -6.15827516e-02
1.03529111e-01 -6.46505415e-01 1.27839661e+00 -5.27701515e-04
1.50375211e+00 -2.13729453e+00 -1.05458044e-01 -2.46338382e-01
6.01350069e-01 6.27749860e-01 -4.34175014e-01 1.51545539e-01
-1.35530129e-01 -2.07193196e-01 -1.54371127e-01 -2.66826779e-01
-2.79625893e-01 3.74234498e-01 -4.24296767e-01 -5.82178961e-03
1.15882985e-01 1.10093391e+00 -1.56202388e+00 -2.63268232e-01
1.81866422e-01 7.02494308e-02 -4.51746285e-01 4.11244147e-02
-4.97677118e-01 -3.16273302e-01 -2.61421949e-01 9.12074924e-01
7.37911403e-01 -5.37869096e-01 -9.32853073e-02 -1.65242478e-01
3.61253530e-01 -2.55466312e-01 -1.06487072e+00 1.82864451e+00
-1.68732271e-01 2.00345933e-01 -6.61005974e-01 -1.19063008e+00
1.03459704e+00 -7.90058821e-02 2.62011886e-01 -6.51981413e-01
5.11516184e-02 8.70003253e-02 -2.11140513e-01 -2.90050507e-01
2.82613605e-01 -4.30403501e-01 -1.05350621e-01 3.13077867e-01
7.54150033e-01 1.35785891e-02 2.99484253e-01 5.37734449e-01
1.00278652e+00 -7.75130317e-02 5.50090432e-01 -1.06322296e-01
3.38923186e-01 -1.65172637e-01 1.01413727e+00 1.14203465e+00
-3.02916527e-01 3.25303882e-01 2.81252414e-01 -8.42720509e-01
-5.48412681e-01 -1.35553920e+00 4.07904238e-01 1.36647046e+00
1.78252250e-01 -5.20870328e-01 -3.43326896e-01 -1.23068166e+00
1.83716938e-01 1.08886051e+00 -7.86913991e-01 -1.09355271e+00
-2.55151987e-01 -7.14793861e-01 -6.14040457e-02 5.65270126e-01
4.29475307e-01 -1.11393809e+00 -1.94492817e-01 4.12871808e-01
3.80259782e-01 -7.26194441e-01 -5.74919105e-01 2.74937183e-01
-1.18911982e+00 -1.19440520e+00 -4.27192658e-01 -1.04331815e+00
7.79863894e-01 7.43561208e-01 1.16509044e+00 1.11953489e-01
-5.73114991e-01 2.49793097e-01 -6.04455352e-01 -5.73985577e-01
-5.52263856e-02 2.30424210e-01 1.42788082e-01 1.94412604e-01
6.22789383e-01 -8.89374733e-01 -3.15562546e-01 -9.46042091e-02
-1.00321782e+00 1.23729512e-01 5.87248981e-01 9.75338936e-01
6.30272269e-01 3.96853447e-01 1.20170653e+00 -1.46229768e+00
6.10933244e-01 -6.34585500e-01 -3.92260909e-01 6.39120877e-01
-1.20100749e+00 4.39421758e-02 1.13420868e+00 -7.37856328e-01
-1.03812563e+00 -1.84863299e-01 4.26522702e-01 -7.98621595e-01
2.24693343e-01 5.99494457e-01 -2.73996353e-01 5.35146669e-02
6.84745133e-01 5.32878041e-01 -6.82270601e-02 -4.17005986e-01
6.76912904e-01 3.23699594e-01 5.40651381e-01 -3.72580439e-01
8.24189007e-01 3.23287129e-01 -1.08482458e-01 -7.00564325e-01
-1.53615952e+00 -3.20339084e-01 -7.66476870e-01 -2.67629862e-01
2.34869629e-01 -8.67909431e-01 -2.20339805e-01 6.04799390e-01
-6.21106207e-01 -3.49266618e-01 -1.10302913e+00 4.85982001e-02
-3.29948932e-01 1.43717214e-01 -3.16223979e-01 -5.55594921e-01
-3.98574799e-01 -6.04979396e-01 5.97025573e-01 6.23826623e-01
2.87391424e-01 -1.03558850e+00 -5.65748997e-02 -2.06950247e-01
4.06656176e-01 2.33626887e-01 1.12000585e+00 -7.06599534e-01
-6.10325038e-01 -2.59282440e-01 -2.44948551e-01 4.64552134e-01
6.60097659e-01 -4.68478829e-01 -9.86823738e-01 -8.09907734e-01
9.94119793e-02 -5.78200579e-01 1.30110466e+00 1.05073757e-01
1.46234596e+00 -4.06378955e-01 -4.19510871e-01 6.49306476e-01
1.67245841e+00 2.87749648e-01 2.58065552e-01 -2.02319846e-01
9.74600613e-01 1.71727285e-01 6.30341887e-01 5.21333396e-01
2.76541919e-01 2.38797590e-02 2.47329608e-01 2.52431124e-01
-7.39296615e-01 -6.55465901e-01 1.19747721e-01 1.20614088e+00
3.00926030e-01 -2.10885957e-01 -5.68633020e-01 5.54368019e-01
-2.06424069e+00 -7.98452020e-01 5.52551687e-01 2.22283864e+00
7.48728871e-01 4.48011786e-01 -2.05553055e-01 -7.15201199e-02
8.53490770e-01 3.00543576e-01 -1.12970102e+00 1.72036365e-01
-1.11573875e-01 3.54824930e-01 2.37855285e-01 4.08737808e-01
-8.08941066e-01 1.25265431e+00 5.32665539e+00 1.24368823e+00
-9.00743961e-01 2.09588587e-01 6.30515397e-01 -4.02385294e-01
-3.10492188e-01 9.29006934e-02 -1.05960846e+00 3.42144310e-01
6.09698057e-01 -6.72004044e-01 5.75976551e-01 1.14958680e+00
-3.38366210e-01 2.18723789e-01 -1.07254565e+00 1.12926590e+00
5.95638931e-01 -1.43003631e+00 6.67962611e-01 -5.28399289e-01
1.00991809e+00 -3.15497257e-02 7.48295244e-03 1.06286776e+00
5.45166552e-01 -6.30662739e-01 4.09149438e-01 6.97109997e-01
9.83282268e-01 -9.65140522e-01 4.81683046e-01 4.72698241e-01
-1.56043470e+00 -4.10823107e-01 -7.82453120e-01 -2.63803322e-02
-1.19456179e-01 6.50563896e-01 -8.80573153e-01 6.25228524e-01
6.82347238e-01 1.26613820e+00 -1.01629031e+00 1.00948608e+00
-2.59123564e-01 5.25947571e-01 1.06235743e-01 -1.11976936e-01
2.35298410e-01 1.03890233e-01 4.31330234e-01 7.28384376e-01
3.50579649e-01 1.46607280e-01 4.88593102e-01 8.40342343e-01
-3.93902540e-01 -9.75105241e-02 -5.84271193e-01 -2.02681601e-01
6.48810387e-01 1.02981246e+00 -9.38167095e-01 -7.60733008e-01
-4.88822788e-01 1.00402987e+00 7.73562610e-01 2.57043421e-01
-4.67029750e-01 -7.97216535e-01 3.33252341e-01 5.56620732e-02
4.00891155e-01 2.00682342e-01 -9.37195681e-03 -1.38152194e+00
-1.28477976e-01 -6.66709542e-01 8.48338902e-01 -5.24845958e-01
-1.69520903e+00 5.09704173e-01 -1.58632487e-01 -1.15637743e+00
-1.00311562e-02 -3.86510372e-01 -7.28256226e-01 2.49847844e-01
-1.38279593e+00 -1.14296579e+00 -7.02255309e-01 8.38297725e-01
1.18401420e+00 -5.89213789e-01 5.38446665e-01 2.09228054e-01
-5.05471647e-01 7.95853972e-01 9.58862305e-02 -6.83646426e-02
5.02420962e-01 -1.05531180e+00 6.03897810e-01 7.83010304e-01
1.27074987e-01 4.51425791e-01 4.05702531e-01 -1.06961644e+00
-1.17387426e+00 -1.69870448e+00 7.47398555e-01 -1.74150601e-01
6.11223459e-01 -5.24858832e-01 -1.31546390e+00 6.66039765e-01
-4.22728956e-01 5.25622964e-01 3.53752375e-01 4.31139804e-02
-4.03284371e-01 -5.35051048e-01 -1.06102490e+00 5.39310038e-01
1.48070753e+00 -4.34418708e-01 -8.30018103e-01 4.54057723e-01
1.55227983e+00 5.57330810e-02 -3.26661408e-01 4.92164701e-01
1.96482420e-01 -7.40641117e-01 9.06422019e-01 -8.90743375e-01
-5.61762489e-02 -1.56764045e-01 3.98850068e-02 -1.48731768e+00
-6.16463065e-01 -3.77885759e-01 -6.98872924e-01 1.15999675e+00
1.81977138e-01 -8.05827916e-01 9.73485172e-01 8.52328464e-02
-3.29928100e-01 -8.56710911e-01 -6.54832959e-01 -1.30164289e+00
-1.22096017e-01 -2.31278226e-01 8.37754428e-01 1.14545369e+00
-2.07078278e-01 6.98792756e-01 -4.86925453e-01 -2.23638117e-01
8.84168506e-01 2.18722224e-01 5.34349680e-01 -1.69181859e+00
-2.48144969e-01 -3.95060062e-01 -3.74191016e-01 -6.30548835e-01
6.43225759e-02 -1.24272501e+00 -9.55769718e-02 -1.45245147e+00
4.47825700e-01 -3.63898724e-01 -9.13415551e-01 7.47924566e-01
-5.12776434e-01 -2.08460763e-01 1.57952338e-01 -7.58493617e-02
-1.11718309e+00 9.47618067e-01 1.29497445e+00 -4.47220296e-01
-5.21928787e-01 3.59024317e-03 -8.69954526e-01 6.92925096e-01
5.02053440e-01 -6.82115793e-01 -1.16552889e+00 -1.03416689e-01
2.01413870e-01 -3.83514017e-01 -5.49303889e-02 -1.33791578e+00
5.55904746e-01 -2.96301335e-01 4.17467773e-01 -7.56317914e-01
7.81571418e-02 -6.88679278e-01 -1.49586752e-01 6.70209527e-01
-2.12770328e-01 -4.15689021e-01 7.56537095e-02 1.07097840e+00
-2.78747082e-02 -2.65022129e-01 8.61590505e-01 -4.90160912e-01
-1.24098885e+00 1.04548800e+00 1.29517898e-01 2.94248879e-01
1.09162641e+00 -2.24880233e-01 -4.27511275e-01 -3.35197784e-02
-8.69635940e-01 3.39727044e-01 2.10586369e-01 8.68986726e-01
1.09425843e+00 -1.54822934e+00 -3.75821531e-01 5.40881634e-01
5.71771324e-01 2.23430291e-01 4.66985792e-01 1.45340323e-01
-4.37611341e-02 5.68895191e-02 -1.56388015e-01 -3.31389457e-01
-8.94809723e-01 1.18998623e+00 4.27174047e-02 -2.08199620e-01
-8.26981544e-01 1.23825693e+00 2.30520651e-01 -5.26955783e-01
4.36400384e-01 -2.13396817e-01 -1.16915770e-01 2.07519993e-01
8.21395040e-01 4.60426241e-01 -1.71210077e-02 -1.05863558e-02
-5.85358813e-02 1.84429258e-01 -6.04906738e-01 6.23813093e-01
1.62957263e+00 -1.61881521e-01 -4.48266529e-02 1.01710665e+00
1.17120755e+00 -6.50506437e-01 -1.40787768e+00 -7.84682214e-01
1.95786431e-02 -4.89706546e-01 -5.22590429e-02 -7.28715479e-01
-1.15931952e+00 7.72548795e-01 7.64353573e-01 -3.28775570e-02
1.12704277e+00 -4.31381464e-02 1.00264049e+00 6.44858837e-01
6.46590829e-01 -1.47105598e+00 6.09522343e-01 6.69280231e-01
5.58443666e-01 -1.12285256e+00 1.03090212e-01 -3.16749930e-01
-2.46513277e-01 1.05891144e+00 1.00059855e+00 -2.93507457e-01
8.31889331e-01 -1.93358287e-01 -5.36647439e-01 -3.74470830e-01
-1.02555597e+00 -9.00419578e-02 3.04896265e-01 6.06598079e-01
-1.76839352e-01 6.21974617e-02 1.01671249e-01 9.96475697e-01
4.32830192e-02 3.47687393e-01 3.55398208e-01 1.11479723e+00
-1.05510426e+00 -8.09329748e-01 3.20096046e-01 9.75872934e-01
2.83680856e-01 -2.08713293e-01 -1.22236602e-01 6.89471722e-01
5.20323038e-01 6.00312054e-01 -2.34944164e-03 -6.34083807e-01
3.16222966e-01 2.63704687e-01 3.69687587e-01 -1.13898015e+00
-5.70942089e-02 -3.92237067e-01 -4.61969405e-01 -5.46534717e-01
4.47682887e-02 -2.83874393e-01 -1.17769897e+00 -1.11818211e-02
-3.68646830e-01 2.03003764e-01 -1.51447700e-02 9.60810840e-01
4.07423735e-01 8.48334193e-01 6.88424468e-01 -3.79442364e-01
-4.64444935e-01 -7.92609453e-01 -9.08532798e-01 3.96898836e-01
3.21144849e-01 -7.68115103e-01 -4.88604575e-01 -1.01077884e-01] | [9.816107749938965, 3.425081968307495] |
5fb233d7-a0f3-4412-875d-38f6bf1181a8 | it-s-all-in-the-embedding-fake-news-detection | 2304.07781 | null | https://arxiv.org/abs/2304.07781v1 | https://arxiv.org/pdf/2304.07781v1.pdf | It's All in the Embedding! Fake News Detection Using Document Embeddings | With the current shift in the mass media landscape from journalistic rigor to social media, personalized social media is becoming the new norm. Although the digitalization progress of the media brings many advantages, it also increases the risk of spreading disinformation, misinformation, and malformation through the use of fake news. The emergence of this harmful phenomenon has managed to polarize society and manipulate public opinion on particular topics, e.g., elections, vaccinations, etc. Such information propagated on social media can distort public perceptions and generate social unrest while lacking the rigor of traditional journalism. Natural Language Processing and Machine Learning techniques are essential for developing efficient tools that can detect fake news. Models that use the context of textual data are essential for resolving the fake news detection problem, as they manage to encode linguistic features within the vector representation of words. In this paper, we propose a new approach that uses document embeddings to build multiple models that accurately label news articles as reliable or fake. We also present a benchmark on different architectures that detect fake news using binary or multi-labeled classification. We evaluated the models on five large news corpora using accuracy, precision, and recall. We obtained better results than more complex state-of-the-art Deep Neural Network models. We observe that the most important factor for obtaining high accuracy is the document encoding, not the classification model's complexity. | ['Elena-Simona Apostol', 'Ciprian-Octavian Truică'] | 2023-04-16 | null | null | null | null | ['misinformation', 'fake-news-detection'] | ['miscellaneous', 'natural-language-processing'] | [-2.19227329e-01 -4.33094576e-02 -4.15703207e-01 -1.70928407e-02
-2.79638737e-01 -5.12174726e-01 1.06869793e+00 7.41520226e-01
-5.12052417e-01 7.12794840e-01 5.74551225e-01 -4.92853314e-01
4.57888961e-01 -1.10611093e+00 -7.32698798e-01 -2.98750937e-01
3.38749826e-01 3.02669793e-01 1.69163689e-01 -7.66533136e-01
6.23384714e-01 2.15985671e-01 -1.43208504e+00 6.67028487e-01
7.20952690e-01 8.17059517e-01 -3.47690672e-01 2.67015308e-01
-4.72694904e-01 1.08184505e+00 -9.70104992e-01 -6.08462811e-01
-1.68714851e-01 -3.65074277e-01 -5.73493421e-01 -2.04800680e-01
3.09679687e-01 -2.10720345e-01 -4.17413980e-01 1.38760746e+00
1.93936706e-01 -3.51622313e-01 6.63232982e-01 -1.00522995e+00
-1.27976179e+00 8.82134974e-01 -5.36422670e-01 4.48132426e-01
3.94814849e-01 -2.29831681e-01 7.08440125e-01 -7.67439604e-01
9.22939777e-01 1.47915387e+00 6.79708838e-01 1.20319217e-01
-8.42195988e-01 -6.71322107e-01 -1.71289250e-01 3.26362103e-01
-1.07177246e+00 -2.20141575e-01 8.23842466e-01 -8.69826376e-01
6.59526348e-01 3.51854295e-01 7.93925822e-01 1.48495185e+00
8.92723441e-01 5.51037908e-01 1.39179516e+00 -4.45447475e-01
-2.29279995e-02 6.20858431e-01 2.39269301e-01 8.02549601e-01
7.96642303e-01 5.25331534e-02 -4.68902260e-01 -4.67985839e-01
1.01920284e-01 1.31586522e-01 -2.69691765e-01 3.54233831e-01
-1.08876014e+00 1.47468781e+00 4.92855996e-01 9.47104990e-01
-3.83093148e-01 3.66012082e-02 6.53889298e-01 7.65970588e-01
1.07527792e+00 9.52505231e-01 1.67566277e-02 3.70383598e-02
-8.12205195e-01 2.58969188e-01 7.64772534e-01 4.87040691e-02
4.93441492e-01 -1.36554003e-01 -4.24792767e-02 7.73181975e-01
3.01600367e-01 7.35624373e-01 9.16966677e-01 -1.55530378e-01
4.25712079e-01 8.16064477e-01 2.33130798e-01 -2.28713346e+00
-4.49687660e-01 -5.27603209e-01 -7.23907709e-01 -1.53705969e-01
2.71922797e-01 -7.93873295e-02 -6.17766440e-01 1.19445753e+00
1.84012964e-01 -2.05197036e-01 -1.23869367e-01 9.20666635e-01
8.02999854e-01 1.04323447e+00 -2.37774655e-01 -1.44359488e-02
1.37078810e+00 -8.32914948e-01 -1.21801889e+00 -3.12987030e-01
7.63185740e-01 -1.02987075e+00 6.88885093e-01 2.25275040e-01
-4.99090791e-01 -9.60165411e-02 -1.20954680e+00 -1.20940413e-02
-1.08676815e+00 -6.44544289e-02 5.27754247e-01 7.74452925e-01
-5.77058494e-01 5.83444417e-01 -4.03875887e-01 -2.00037450e-01
5.33327579e-01 -2.55237937e-01 -3.91135931e-01 4.02926430e-02
-1.67928183e+00 1.39314759e+00 3.16681027e-01 -8.21103826e-02
-4.22684461e-01 -1.48551196e-01 -6.62893891e-01 -1.91295400e-01
2.00703025e-01 -9.94751751e-02 8.17212582e-01 -1.33437800e+00
-1.23244452e+00 9.06931996e-01 3.54538023e-01 -6.01498604e-01
6.18976235e-01 -2.02969745e-01 -9.81784642e-01 5.72109297e-02
1.45606369e-01 2.21560016e-01 1.03261435e+00 -1.11113465e+00
-3.15674633e-01 -4.02518779e-01 -1.42498851e-01 -2.41212457e-01
-7.23456442e-01 1.97758585e-01 3.79024267e-01 -7.35598624e-01
5.63379936e-02 -9.62962151e-01 1.91909507e-01 -2.93386370e-01
-5.57826519e-01 -1.14765488e-01 1.04546142e+00 -7.94351697e-01
1.25277722e+00 -2.00379610e+00 -1.10349797e-01 -1.26744471e-02
5.58857679e-01 4.42821443e-01 1.50758520e-01 6.89008534e-01
2.10978001e-01 4.87416148e-01 6.76659793e-02 -8.68407264e-02
-1.84985831e-01 -4.17942740e-02 -6.15577519e-01 8.89911115e-01
5.06038740e-02 7.47595966e-01 -9.24190044e-01 -2.03358918e-01
-1.37742102e-01 5.07901430e-01 -5.20603240e-01 -3.11706930e-01
-2.14211851e-01 1.69479832e-01 -4.43481624e-01 4.53762859e-01
5.64643919e-01 -3.46651256e-01 5.12539931e-02 -1.07077241e-01
-3.06109041e-01 5.02538681e-01 -6.07481301e-01 9.07030404e-01
-3.41061026e-01 1.10973716e+00 -3.85840058e-01 -1.03351569e+00
9.28700030e-01 1.23926744e-01 1.18181691e-01 -7.67955184e-01
7.90175259e-01 4.00943786e-01 -1.37414277e-01 -6.14725947e-01
8.53273749e-01 -1.42083704e-01 -3.61059636e-01 6.53688967e-01
-2.27984428e-01 -4.15445864e-03 -1.95491239e-01 1.85623243e-01
7.71096826e-01 -5.05131125e-01 4.30236399e-01 -1.79335192e-01
3.08751255e-01 3.00005168e-01 6.54795840e-02 6.01734459e-01
-7.02866837e-02 8.00306872e-02 5.83076119e-01 -8.30418289e-01
-1.00982487e+00 -3.60648066e-01 -1.01135321e-01 9.45373297e-01
1.28615454e-01 -7.55035058e-02 -4.62802708e-01 -6.50180936e-01
3.35296124e-01 8.46767724e-01 -7.55018711e-01 -4.66811538e-01
-2.62858272e-01 -1.01642907e+00 7.04037011e-01 -3.37315381e-01
4.70849782e-01 -7.59829998e-01 -4.54479724e-01 3.17731708e-01
-3.55199397e-01 -9.89707410e-01 -1.37984157e-01 -2.07652301e-01
-4.68911260e-01 -1.06221902e+00 -3.76455039e-01 -5.94758570e-01
5.41553438e-01 3.45414609e-01 7.74285018e-01 2.09416255e-01
3.35042737e-02 -2.32013315e-01 -5.95727921e-01 -5.82669199e-01
-1.10058022e+00 -5.98877296e-02 2.01603085e-01 2.49665245e-01
4.10365015e-01 -1.08874306e-01 -1.64455295e-01 4.21257690e-02
-1.13624132e+00 1.09678283e-02 3.00884247e-01 8.28175247e-01
-1.79550186e-01 5.23877330e-02 7.07519591e-01 -1.18556309e+00
1.03616405e+00 -9.08677280e-01 -4.65155989e-01 -1.42401129e-01
-6.03031874e-01 6.90999255e-02 7.37921119e-01 -5.57459116e-01
-5.65560579e-01 -7.45873213e-01 -3.24260257e-02 6.01438284e-02
2.32956871e-01 8.97139311e-01 6.77252829e-01 -1.12936899e-01
1.11507559e+00 7.83592835e-02 2.22442955e-01 -4.44537491e-01
2.93302804e-01 1.23146868e+00 -1.47600889e-01 1.76097855e-01
6.57244980e-01 7.18486786e-01 -4.88286436e-01 -9.70725179e-01
-1.10765862e+00 -2.60182321e-01 1.88485794e-02 -2.26431593e-01
5.54036021e-01 -7.64377594e-01 -3.87191266e-01 6.60278618e-01
-1.53291738e+00 4.96019959e-01 2.70007044e-01 3.77522081e-01
3.28164101e-01 3.34185362e-01 -7.79673278e-01 -6.16264582e-01
-1.74445778e-01 -9.38961446e-01 5.33503056e-01 -3.10947537e-01
-1.85994521e-01 -9.87949967e-01 3.00773412e-01 6.29831970e-01
8.04063141e-01 5.65408885e-01 7.68400371e-01 -1.03530514e+00
-1.19254768e-01 -5.88561833e-01 -2.64818639e-01 5.52735209e-01
2.13583291e-01 -8.04498941e-02 -9.03482974e-01 -7.65938610e-02
2.44600579e-01 -3.53810906e-01 1.01367080e+00 -1.06779620e-01
5.95166743e-01 -1.06191015e+00 -3.07374179e-01 -8.48172158e-02
1.31282294e+00 9.55783650e-02 5.77302635e-01 6.19390905e-01
6.36766851e-01 6.38650179e-01 2.70583987e-01 2.55710453e-01
3.64023179e-01 5.26744843e-01 3.87562215e-01 2.70864666e-01
9.08299088e-02 -3.39573443e-01 5.62042475e-01 1.12890887e+00
1.84955716e-01 -3.98967475e-01 -1.03928840e+00 4.29199219e-01
-1.48534596e+00 -1.22256732e+00 -3.89381021e-01 1.85118473e+00
7.82328725e-01 2.74810314e-01 -1.65002361e-01 3.37064415e-01
8.03574026e-01 4.99641299e-01 1.13329999e-01 -7.89427340e-01
-2.54143178e-01 -3.35067272e-01 6.71131492e-01 6.96531653e-01
-1.16696560e+00 9.81011868e-01 5.68365097e+00 5.95686615e-01
-1.80198777e+00 6.34639859e-01 5.50228834e-01 2.49845430e-01
-5.30840158e-01 -5.22819638e-01 -6.15583718e-01 7.82833219e-01
9.94021356e-01 -3.52582592e-03 2.98082888e-01 7.65198290e-01
2.75555164e-01 -1.16798937e-01 -4.67356473e-01 8.87542963e-01
6.81398630e-01 -1.84342945e+00 1.52836531e-01 3.46683636e-02
8.42479825e-01 3.86369675e-01 1.42827511e-01 1.76905051e-01
-4.44419645e-02 -9.47013557e-01 8.92010510e-01 2.97669202e-01
3.92706513e-01 -5.37872970e-01 1.15926814e+00 4.15715516e-01
-8.42178762e-02 -6.35821298e-02 -3.20382923e-01 -4.73598868e-01
1.93599146e-03 1.03233123e+00 -1.00675845e+00 7.44936941e-03
1.89759687e-01 6.01069987e-01 -7.08732843e-01 5.63248932e-01
-1.71221986e-01 5.52623510e-01 -2.11461455e-01 -7.02713072e-01
5.31328499e-01 -3.09479702e-03 7.41143167e-01 1.19537759e+00
2.57930458e-01 -4.81182098e-01 -1.84234176e-02 6.54447377e-01
-1.76047340e-01 3.67499292e-01 -1.08817518e+00 -7.07810938e-01
1.81565836e-01 8.31816316e-01 -7.21176267e-01 -4.18179691e-01
-2.97323197e-01 9.06591713e-01 2.91771591e-01 -1.24494128e-01
-9.81955349e-01 -1.57672063e-01 3.13033581e-01 4.96083260e-01
-2.53583044e-02 -2.14149013e-01 -1.36890486e-01 -1.35952532e+00
-1.48274392e-01 -1.10459483e+00 -5.01372404e-02 -2.92082399e-01
-1.23465097e+00 7.18770742e-01 -3.28750491e-01 -9.63328600e-01
3.97895575e-02 -5.63056588e-01 -1.38840467e-01 3.07029933e-01
-1.53792644e+00 -8.31040323e-01 4.38435525e-02 1.41483352e-01
1.78920880e-01 -3.59973967e-01 7.97483206e-01 4.94851202e-01
-2.94479817e-01 3.11313361e-01 3.89539123e-01 3.21931750e-01
6.87850237e-01 -6.98403478e-01 1.99745849e-01 4.98182982e-01
2.43459761e-01 3.15442771e-01 9.53692496e-01 -6.94896519e-01
-1.21588695e+00 -1.00631535e+00 1.50229406e+00 -4.98943061e-01
1.15420794e+00 -3.69197845e-01 -6.80917919e-01 4.05648112e-01
2.00217098e-01 -3.45002890e-01 7.43041039e-01 6.60912097e-02
-8.08658183e-01 1.57231838e-01 -1.31203806e+00 4.32331681e-01
3.92822325e-01 -6.56166255e-01 -8.14848542e-01 7.43445575e-01
7.32043803e-01 -1.10216632e-01 -3.25146645e-01 -2.21643135e-01
5.55706441e-01 -9.21239495e-01 5.52580714e-01 -7.91829288e-01
9.15716767e-01 -7.32595054e-03 -4.43768464e-02 -1.75157559e+00
-2.69439757e-01 -2.15435132e-01 1.32260412e-01 7.12877214e-01
7.06080556e-01 -1.13384748e+00 2.90137023e-01 -2.46425003e-01
1.21271662e-01 -4.39647675e-01 -9.07817841e-01 -5.14751792e-01
7.22108036e-02 -2.76650459e-01 3.33361030e-01 1.67776883e+00
1.63069591e-01 4.64984089e-01 -6.15285397e-01 1.11536652e-01
2.61815339e-01 2.05399673e-02 4.66527790e-01 -1.22793698e+00
-7.54061267e-02 -3.98911893e-01 -7.37271607e-01 -4.80542153e-01
2.07418352e-02 -8.64941835e-01 -3.64347160e-01 -1.30306053e+00
-9.92495120e-02 -9.58684310e-02 -2.54676957e-02 1.74929574e-01
2.59291381e-01 5.73121965e-01 7.53770247e-02 2.66621530e-01
-3.22828829e-01 3.99218917e-01 1.43236911e+00 -5.13955712e-01
-7.50670284e-02 -3.37449938e-01 -8.82570326e-01 9.01621580e-01
8.40899527e-01 -9.03555870e-01 1.31182045e-01 -5.11475325e-01
1.02795148e+00 -1.44676417e-01 3.39799047e-01 -8.76054287e-01
8.82470682e-02 -5.46167642e-02 5.61777800e-02 -1.13496721e-01
2.64431864e-01 -6.96250677e-01 -2.18651488e-01 8.66569102e-01
-3.70361477e-01 2.02252641e-01 3.39528657e-02 6.74584508e-01
-4.43267047e-01 2.56843809e-02 9.43938434e-01 -5.06941117e-02
-2.14076281e-01 -2.77109236e-01 -9.00384724e-01 -3.02873012e-02
8.91069829e-01 3.11199367e-01 -8.40694606e-01 -5.69652855e-01
-4.69340324e-01 -3.54746938e-01 3.38064909e-01 9.87893403e-01
5.27470827e-01 -1.26685083e+00 -9.99175727e-01 6.17959537e-02
2.04702824e-01 -8.01748157e-01 -1.55966818e-01 7.79543102e-01
-9.65461731e-01 4.61648494e-01 -2.19336480e-01 -1.79784775e-01
-9.75497484e-01 5.22669673e-01 1.94566116e-01 -1.51591465e-01
-3.91566992e-01 6.54941559e-01 -5.30718029e-01 -3.63511860e-01
-1.95895389e-01 -4.65716988e-01 -5.50108135e-01 6.41356468e-01
8.32922459e-01 3.02967846e-01 1.83799788e-01 -1.05130577e+00
-3.30638796e-01 -8.41629133e-03 -2.63649076e-01 1.33882985e-01
1.30610788e+00 1.53753921e-01 -5.57859302e-01 7.34717965e-01
1.30274105e+00 3.41513842e-01 8.72704238e-02 3.33603099e-02
2.20122421e-03 -6.88133478e-01 6.05169058e-01 -8.96358430e-01
-8.47260296e-01 5.49838543e-01 4.13080543e-01 9.39180076e-01
3.00035179e-01 -1.01832390e-01 8.75902772e-01 4.74890202e-01
3.93001169e-01 -1.14010394e+00 1.27378017e-01 6.37834847e-01
1.02481067e+00 -1.37656212e+00 2.01480538e-02 -3.18670243e-01
-5.55484712e-01 9.69949305e-01 -1.74752064e-02 -3.25018287e-01
7.80085862e-01 -2.15104390e-02 1.84277967e-01 -5.21200597e-01
-3.89032006e-01 4.18400228e-01 2.77960747e-01 1.56521276e-02
5.49090266e-01 4.17969584e-01 -9.64981437e-01 3.65689903e-01
-3.18524927e-01 -1.02670699e-01 9.66757596e-01 6.79967284e-01
-7.72221088e-01 -7.46214867e-01 -7.38197863e-01 5.65059006e-01
-8.52456033e-01 -9.18213055e-02 -7.16463327e-01 5.46997607e-01
3.06136787e-01 1.15172100e+00 3.35302651e-02 -5.47282219e-01
-1.10712126e-01 9.42253545e-02 9.43567753e-02 -4.81199682e-01
-7.65643954e-01 -4.96067941e-01 5.36497176e-01 -4.33871895e-01
-3.60968649e-01 -3.29087377e-01 -6.45083368e-01 -8.24508071e-01
-5.59770882e-01 3.46575379e-01 1.12343383e+00 1.09523749e+00
4.04061198e-01 2.40294710e-01 7.25492477e-01 -4.52522367e-01
-7.41691947e-01 -1.22463703e+00 -2.58988708e-01 5.35350144e-01
5.71900904e-01 -7.72991061e-01 -7.14460611e-01 -3.76717001e-01] | [8.20324993133545, 10.271595001220703] |
d5cf9636-9980-4dca-ac64-f8e18f6906b4 | predictive-business-process-monitoring-via | 2003.11268 | null | https://arxiv.org/abs/2003.11268v2 | https://arxiv.org/pdf/2003.11268v2.pdf | Predictive Business Process Monitoring via Generative Adversarial Nets: The Case of Next Event Prediction | Predictive process monitoring aims to predict future characteristics of an ongoing process case, such as case outcome or remaining timestamp. Recently, several predictive process monitoring methods based on deep learning such as Long Short-Term Memory or Convolutional Neural Network have been proposed to address the problem of next event prediction. However, due to insufficient training data or sub-optimal network configuration and architecture, these approaches do not generalize well the problem at hand. This paper proposes a novel adversarial training framework to address this shortcoming, based on an adaptation of Generative Adversarial Networks (GANs) to the realm of sequential temporal data. The training works by putting one neural network against the other in a two-player game (hence the adversarial nature) which leads to predictions that are indistinguishable from the ground truth. We formally show that the worst-case accuracy of the proposed approach is at least equal to the accuracy achieved in non-adversarial settings. From the experimental evaluation it emerges that the approach systematically outperforms all baselines both in terms of accuracy and earliness of the prediction, despite using a simple network architecture and a naive feature encoding. Moreover, the approach is more robust, as its accuracy is not affected by fluctuations over the case length. | ['Marcello La Rosa', 'Sarah Erfani', 'Ilya Verenich', 'Zahra Dasht Bozorgi', 'Farbod Taymouri'] | 2020-03-25 | null | null | null | null | ['predictive-process-monitoring'] | ['time-series'] | [ 4.56735104e-01 4.37577397e-01 3.08226734e-01 -1.34545833e-01
-6.55686617e-01 -4.29761440e-01 1.18067563e+00 3.67259443e-01
-2.84959853e-01 8.75013530e-01 1.39710261e-02 -1.98302925e-01
-4.07583982e-01 -1.11192775e+00 -7.05219507e-01 -7.25107372e-01
-2.20552519e-01 6.73163235e-01 3.02370906e-01 -2.81450544e-02
9.40679684e-02 4.72487390e-01 -1.26162505e+00 1.43440282e-02
4.40999240e-01 1.29541910e+00 -4.48450923e-01 8.34259987e-01
1.90896541e-01 1.40662050e+00 -8.23576152e-01 -6.26782775e-01
3.74605089e-01 -5.44070244e-01 -6.79243863e-01 -2.51855463e-01
-4.71533202e-02 -3.61693889e-01 -5.20010412e-01 5.97879827e-01
3.74985188e-01 1.04159117e-01 4.69340205e-01 -1.37183988e+00
-3.18313539e-01 5.64485371e-01 -1.85027104e-02 2.41108447e-01
4.73440707e-01 3.55190039e-01 7.74646878e-01 -2.96613544e-01
5.47002733e-01 9.97917235e-01 9.77934539e-01 5.88889837e-01
-1.16648114e+00 -4.12147552e-01 1.43912241e-01 9.30569619e-02
-1.15496457e+00 -3.00649911e-01 5.59220254e-01 -4.03511286e-01
8.43886375e-01 2.20619768e-01 5.07828891e-01 1.55392826e+00
6.83343291e-01 5.77430069e-01 1.14301109e+00 -1.49806306e-01
6.38988674e-01 -2.87001491e-01 -2.70896435e-01 8.88302550e-02
5.81984073e-02 4.97873187e-01 -4.94302183e-01 -3.54505450e-01
7.63638794e-01 3.50890607e-01 -7.09390789e-02 -1.91775858e-01
-1.25277221e+00 5.82169533e-01 2.06734613e-01 4.82146919e-01
-8.01542699e-01 3.79636884e-01 4.83941525e-01 4.38413590e-01
4.01856989e-01 4.17508543e-01 -3.87197793e-01 -4.02831107e-01
-1.17706573e+00 4.51263309e-01 1.16742110e+00 5.87284267e-01
2.47294649e-01 2.67852455e-01 -6.40326917e-01 1.83428042e-02
1.00668836e-02 1.97115839e-01 5.99027395e-01 -7.72559941e-01
3.10298175e-01 3.54899943e-01 2.62698859e-01 -8.89803767e-01
-3.00184220e-01 -4.43317056e-01 -1.01383758e+00 3.37465376e-01
6.51476920e-01 -4.05133545e-01 -7.74593055e-01 1.83781350e+00
-6.12671748e-02 5.79131424e-01 1.70123518e-01 4.94265079e-01
1.85001701e-01 6.05394006e-01 3.01923305e-01 -3.66753340e-01
1.02222037e+00 -7.23974049e-01 -8.04034650e-01 -2.59975106e-01
-1.29425824e-01 -3.76472801e-01 6.28565669e-01 4.78025526e-01
-1.18691278e+00 -5.37525415e-01 -9.63336825e-01 5.55000007e-01
-3.77751321e-01 -5.38176000e-01 5.01526833e-01 7.78666615e-01
-1.08736253e+00 1.10434556e+00 -1.12983239e+00 -4.16409433e-01
4.32563275e-01 3.41343433e-01 -2.31798023e-01 1.75875917e-01
-1.32084644e+00 7.42563009e-01 5.10546744e-01 8.67579728e-02
-1.29332542e+00 -5.62507033e-01 -5.30236781e-01 3.54452193e-01
3.82884681e-01 -7.64778078e-01 1.28342509e+00 -1.10825801e+00
-1.70858431e+00 2.87956417e-01 1.08423114e-01 -9.96663451e-01
1.09579158e+00 -2.83233196e-01 -5.04611433e-01 -2.11830921e-02
-2.66920805e-01 5.78892678e-02 9.19583321e-01 -1.11430705e+00
-5.50815284e-01 -2.02880889e-01 9.37282518e-02 -2.16722235e-01
-6.88770711e-02 2.30402071e-02 -4.83877696e-02 -8.33752513e-01
-2.10583925e-01 -8.22366297e-01 -5.09249210e-01 -2.60267138e-01
-3.79821777e-01 7.27405353e-03 6.16243184e-01 -6.20415628e-01
1.04629755e+00 -1.91121197e+00 -1.43436790e-01 2.94792056e-01
1.24576233e-01 1.77337930e-01 2.10776001e-01 8.57969284e-01
-1.77726522e-01 1.22143850e-01 -2.84241229e-01 -5.32065153e-01
1.05283335e-01 2.68756479e-01 -4.69519883e-01 3.14866096e-01
4.08423364e-01 1.03044713e+00 -9.78310227e-01 -2.19262019e-01
2.21477926e-01 3.40945631e-01 -6.11938797e-02 6.32894874e-01
-2.39607215e-01 8.52390707e-01 -3.24554682e-01 3.85947764e-01
3.53527963e-01 -1.10182472e-01 1.74387217e-01 5.04235327e-01
2.85050631e-01 5.26539162e-02 -1.13705063e+00 1.45852900e+00
-5.35406828e-01 4.41866875e-01 -4.11387801e-01 -8.81986201e-01
9.93167162e-01 7.35170424e-01 6.41224265e-01 -5.31573176e-01
-5.18073812e-02 -1.32852763e-01 -8.62143040e-02 -1.26685768e-01
3.63536388e-01 -5.30933678e-01 -1.51202232e-01 6.29599452e-01
-3.52572650e-02 3.16713661e-01 -9.54248160e-02 -1.45300791e-01
1.81135368e+00 3.06365043e-01 5.21695435e-01 1.68867305e-01
5.17984450e-01 -4.29149002e-01 7.09725738e-01 1.01201749e+00
-2.87212998e-01 6.12712979e-01 6.99082494e-01 -5.63200772e-01
-1.03606749e+00 -1.25106347e+00 2.61308312e-01 8.13975573e-01
-2.90001687e-02 -3.26322585e-01 -6.89001322e-01 -7.01990843e-01
-1.39614269e-01 8.75304818e-01 -9.26254034e-01 -4.57844138e-01
-7.10392416e-01 -6.05141461e-01 8.79176557e-01 6.72186136e-01
4.11372900e-01 -1.47903931e+00 -7.50745475e-01 6.46108627e-01
2.49284387e-01 -1.30616379e+00 1.66248575e-01 1.72448918e-01
-9.22128558e-01 -1.00548804e+00 -4.23484176e-01 -3.63086313e-02
6.00441247e-02 -7.34015882e-01 1.33544183e+00 -3.31178695e-01
2.13523373e-01 3.13386679e-01 -2.13797465e-01 -6.48632169e-01
-8.97699714e-01 6.28236830e-02 -1.03476018e-01 3.37328166e-01
2.81228065e-01 -9.65454221e-01 -6.93086863e-01 6.51295632e-02
-1.08925629e+00 -3.21418852e-01 7.59887934e-01 6.46228373e-01
4.97012287e-01 2.97510028e-01 8.29720259e-01 -8.99078190e-01
7.49327123e-01 -6.54927015e-01 -2.93865770e-01 2.14201763e-01
-8.93415868e-01 1.06883146e-01 1.08476508e+00 -4.00770903e-01
-1.21855807e+00 -1.09941252e-01 -7.69996941e-02 -5.41721165e-01
-5.43869495e-01 2.42458478e-01 -1.64439112e-01 3.45631123e-01
5.23047566e-01 4.29067403e-01 -1.61937460e-01 -1.98751047e-01
2.28966791e-02 2.07757846e-01 6.48649752e-01 -3.38614017e-01
8.96703303e-01 6.61242366e-01 1.97667554e-01 -6.70372099e-02
-6.31871402e-01 2.98231784e-02 -4.73869354e-01 -3.54602695e-01
7.18295634e-01 -6.17679715e-01 -7.92724252e-01 6.12672627e-01
-1.02020776e+00 -4.24197644e-01 -8.32420528e-01 2.24786341e-01
-9.74735558e-01 2.01642066e-01 -9.14020061e-01 -1.04786897e+00
-5.89008391e-01 -7.49047816e-01 9.66490448e-01 5.52147478e-02
-3.63447458e-01 -1.21141160e+00 2.57644325e-01 -4.48508374e-02
6.27308190e-01 9.21792686e-01 6.60691798e-01 -1.25986803e+00
-4.82754827e-01 -7.19290495e-01 1.85704157e-01 3.66164476e-01
-2.83348025e-03 -1.71068788e-01 -1.10200465e+00 -2.71648973e-01
3.15619409e-01 3.42675089e-03 5.07155061e-01 9.09985900e-02
1.09685338e+00 -2.96208799e-01 -1.73331738e-01 2.99860924e-01
1.50340652e+00 1.97123840e-01 1.01371813e+00 4.58348960e-01
4.37726557e-01 3.88855010e-01 4.77494568e-01 7.03108788e-01
3.13868634e-02 4.37487096e-01 8.19699883e-01 1.73482463e-01
1.66408256e-01 -3.89470637e-01 5.08380771e-01 1.59624681e-01
-4.25437778e-01 -4.62882400e-01 -1.05543351e+00 4.02753115e-01
-1.95313370e+00 -1.41680086e+00 8.37280601e-02 2.36974192e+00
4.63819653e-01 6.41026974e-01 4.50458042e-02 4.13654536e-01
5.83284974e-01 2.48754412e-01 -4.53041613e-01 -5.14627635e-01
2.94170789e-02 5.29656053e-01 5.02800524e-01 1.45595362e-02
-1.15860438e+00 4.60454762e-01 6.67765665e+00 7.60658145e-01
-9.56725776e-01 4.01223063e-01 5.88548064e-01 -5.77513129e-03
9.49918628e-02 -1.51894242e-01 -4.11247313e-01 6.43060625e-01
1.61645627e+00 -2.51408041e-01 3.56707960e-01 6.49336576e-01
4.19414014e-01 1.29109919e-01 -1.47239470e+00 6.41853571e-01
-1.52946725e-01 -1.18036449e+00 -6.57149823e-03 1.10804826e-01
7.76520371e-01 -2.04742894e-01 -1.78256109e-02 5.05415082e-01
4.90382433e-01 -1.37781072e+00 8.30075622e-01 1.11256027e+00
4.44511682e-01 -9.48152006e-01 1.20262063e+00 5.09344041e-01
-9.83641207e-01 -3.13955128e-01 2.15265915e-01 -4.19081509e-01
2.86948502e-01 5.73918343e-01 -8.27965796e-01 7.51948059e-01
5.24709642e-01 4.37441528e-01 -4.20675874e-01 8.08569193e-01
-1.96710199e-01 9.13974166e-01 -9.75996777e-02 2.87311614e-01
2.35622510e-01 6.66952282e-02 6.98488712e-01 1.10348868e+00
3.53938282e-01 -3.74385953e-01 8.45932811e-02 8.61273825e-01
6.95120776e-03 -2.20700353e-01 -7.32988358e-01 7.99641758e-02
2.41793513e-01 9.51848865e-01 -6.71413600e-01 -2.38677144e-01
-3.33253980e-01 1.11478639e+00 1.48341149e-01 2.54002810e-01
-1.01437390e+00 6.66645542e-02 4.89565969e-01 1.54128492e-01
4.37805772e-01 8.47896039e-02 -2.91557461e-01 -7.45553613e-01
2.05336526e-01 -8.15593541e-01 4.51189250e-01 -5.12017488e-01
-1.60968840e+00 8.13437819e-01 -2.53068775e-01 -1.30558038e+00
-8.79592359e-01 -1.92882985e-01 -9.58099961e-01 7.72589386e-01
-1.27043259e+00 -1.33957446e+00 -2.54416108e-01 5.99498153e-01
4.15236652e-01 -1.62574455e-01 9.69760239e-01 8.70757774e-02
-4.80644882e-01 5.10241866e-01 2.29637489e-01 3.95734608e-02
6.21919215e-01 -1.47593892e+00 5.28836668e-01 1.11332273e+00
-1.16653144e-01 2.30394557e-01 9.95088160e-01 -6.19217396e-01
-9.93419409e-01 -1.44589686e+00 7.43638754e-01 -7.92879939e-01
8.54385316e-01 -2.13679761e-01 -9.57404792e-01 8.92509222e-01
2.12986737e-01 1.31753415e-01 6.34230494e-01 -2.02691719e-01
-1.27679827e-02 -1.09544590e-01 -1.28655136e+00 4.18910593e-01
8.85557413e-01 -4.44782317e-01 -6.42883062e-01 1.66803852e-01
5.22132158e-01 -2.48675585e-01 -1.27568233e+00 3.26200336e-01
3.44547451e-01 -1.25256443e+00 7.49345899e-01 -6.88447177e-01
6.91643655e-01 -6.85169846e-02 3.42617966e-02 -9.60554600e-01
-1.29355311e-01 -9.19396579e-01 -7.31163085e-01 1.33390021e+00
1.27161294e-01 -6.31764412e-01 8.34427238e-01 6.53094172e-01
1.37709454e-01 -8.79030466e-01 -1.14596856e+00 -9.53746915e-01
1.51400104e-01 -6.73252106e-01 9.29889798e-01 7.24260151e-01
-5.05123317e-01 1.96750998e-01 -5.69997609e-01 3.07452738e-01
4.46228147e-01 4.71252836e-02 6.48113012e-01 -1.36250639e+00
-5.50626457e-01 -2.92612314e-01 -6.81971967e-01 -4.90529180e-01
1.58928737e-01 -3.83691490e-01 -1.47318821e-02 -1.35155618e+00
4.82150018e-02 -3.04750144e-01 -6.68838263e-01 3.62410933e-01
-2.47214243e-01 6.55989647e-02 2.46561423e-01 2.94542342e-01
-6.14649296e-01 6.12606287e-01 8.81548882e-01 -4.77296393e-03
-9.09223333e-02 4.50626165e-01 -3.44267368e-01 6.86533093e-01
9.67486501e-01 -5.50737917e-01 -2.43561909e-01 1.71398371e-01
3.34852874e-01 4.31924880e-01 7.85485983e-01 -1.51657140e+00
3.12774360e-01 -6.04736470e-02 3.79379570e-01 -1.41041845e-01
2.60662913e-01 -1.14969909e+00 6.44686103e-01 5.82696021e-01
-4.49362487e-01 1.45362571e-01 5.18867821e-02 1.19074643e+00
-3.13908845e-01 -1.32931188e-01 4.37238157e-01 -5.90308048e-02
-5.34356296e-01 4.12748903e-01 -4.90665674e-01 -1.17055871e-01
1.36438441e+00 -3.40848833e-01 5.29125184e-02 -7.78353214e-01
-9.16863143e-01 -1.12218179e-01 4.37537819e-01 2.71073371e-01
1.76840439e-01 -1.18879819e+00 -7.06942499e-01 -7.77443871e-02
5.85869001e-03 -8.04263502e-02 2.00806707e-01 8.02978098e-01
-2.82496601e-01 1.34859115e-01 -2.50739396e-01 -3.61135125e-01
-7.37612486e-01 8.46822023e-01 5.34299314e-01 -1.24503720e+00
-6.43365085e-01 2.20766872e-01 -1.78448424e-01 -1.76804498e-01
2.12155357e-02 -8.35781097e-02 -1.96054187e-02 -1.08752735e-01
4.77556080e-01 3.95928651e-01 2.71045923e-01 -2.67921597e-01
-1.18458770e-01 -4.13968526e-02 2.25103050e-01 -6.62933588e-02
1.32106936e+00 6.04144931e-02 8.94972235e-02 7.00819492e-01
5.60934126e-01 -7.31392503e-02 -1.48800194e+00 -1.31357566e-01
1.53501555e-01 -1.81190088e-01 -1.49997622e-01 -8.58405828e-01
-1.04849923e+00 7.03306913e-01 6.89264715e-01 7.51321375e-01
1.25853312e+00 -1.96829259e-01 7.12238610e-01 2.18417823e-01
3.52050722e-01 -8.23468149e-01 -2.39616446e-02 5.01466393e-01
7.21961856e-01 -1.04587471e+00 -2.74165899e-01 -3.43417786e-02
-4.89161879e-01 1.09530938e+00 4.35701847e-01 -2.99192727e-01
5.19299269e-01 3.57895195e-01 -1.25669673e-01 -2.36555915e-02
-9.26793218e-01 2.85176169e-02 -1.99048132e-01 7.98394978e-01
1.16036184e-01 -1.80908926e-02 -6.59984052e-02 8.09169054e-01
-1.81276098e-01 2.35923558e-01 4.09902394e-01 1.02603757e+00
1.62622139e-01 -9.38017666e-01 -4.70133513e-01 5.43859184e-01
-8.22547793e-01 1.14960901e-01 -2.47180969e-01 8.69893670e-01
1.43899977e-01 9.01375234e-01 2.52509683e-01 -2.52113193e-01
4.56362486e-01 2.92088479e-01 1.64538875e-01 -3.05100173e-01
-1.03304052e+00 -4.64476615e-01 -4.97014225e-02 -7.39248574e-01
-4.64240491e-01 -7.62577593e-01 -9.14704502e-01 -5.26800036e-01
2.63121165e-02 -1.26644984e-01 2.33071625e-01 1.13157392e+00
3.22588533e-01 1.01444852e+00 6.22604370e-01 -7.49735832e-01
-8.22836578e-01 -1.12887084e+00 -5.35746694e-01 7.16851056e-01
6.15873933e-02 -3.93544585e-01 -1.52585864e-01 7.54417777e-02] | [8.55825424194336, 5.801504135131836] |
2b56422f-ed39-4635-8664-25f29f270247 | deep-learning-based-ecg-classification-on | 2209.00989 | null | https://arxiv.org/abs/2209.00989v1 | https://arxiv.org/pdf/2209.00989v1.pdf | Deep Learning-based ECG Classification on Raspberry PI using a Tensorflow Lite Model based on PTB-XL Dataset | The number of IoT devices in healthcare is expected to rise sharply due to increased demand since the COVID-19 pandemic. Deep learning and IoT devices are being employed to monitor body vitals and automate anomaly detection in clinical and non-clinical settings. Most of the current technology requires the transmission of raw data to a remote server, which is not efficient for resource-constrained IoT devices and embedded systems. Additionally, it is challenging to develop a machine learning model for ECG classification due to the lack of an extensive open public database. To an extent, to overcome this challenge PTB-XL dataset has been used. In this work, we have developed machine learning models to be deployed on Raspberry Pi. We present an evaluation of our TensorFlow Model with two classification classes. We also present the evaluation of the corresponding TensorFlow Lite FlatBuffers to demonstrate their minimal run-time requirements while maintaining acceptable accuracy. | ['Rasit Eskicioglu', 'Kushagra Sharma'] | 2022-08-25 | null | null | null | null | ['ecg-classification'] | ['medical'] | [ 1.12658809e-03 -3.69918168e-01 2.57012695e-01 -3.71725827e-01
-2.07353935e-01 -2.24872246e-01 -2.78758258e-01 5.84581614e-01
-5.41725874e-01 6.66187823e-01 4.83195484e-02 -6.20150030e-01
-1.13940701e-01 -5.55532217e-01 -1.90636262e-01 -3.22267920e-01
-3.44761223e-01 6.31498992e-01 1.36552110e-01 1.59752265e-01
-2.18784615e-01 5.14920413e-01 -1.12603664e+00 7.26798415e-01
4.64960575e-01 1.42005563e+00 -2.74887353e-01 1.05913341e+00
1.13515526e-01 8.07901502e-01 -7.04497933e-01 -2.47581109e-01
2.53425926e-01 -3.55931967e-01 -7.68023968e-01 -2.87435621e-01
1.28210545e-01 -6.52349472e-01 -4.58148643e-02 6.97111726e-01
1.09858572e+00 -3.61461014e-01 9.04812478e-03 -1.18821335e+00
2.51664430e-01 4.43347722e-01 -1.18874215e-01 9.09022748e-01
2.03162059e-01 2.89503336e-01 5.38608909e-01 -3.00889641e-01
2.45264068e-01 5.91731191e-01 1.00650263e+00 3.91178995e-01
-9.95313585e-01 -4.12602305e-01 -3.87300640e-01 4.30617422e-01
-1.30353141e+00 -2.40821585e-01 6.17312729e-01 -3.33516449e-01
1.19288528e+00 3.36644769e-01 9.49575603e-01 1.20411038e+00
5.01561403e-01 3.25756371e-01 9.02655721e-01 -9.54919606e-02
4.08415258e-01 -1.19128145e-01 1.82052001e-01 4.15811032e-01
4.60147649e-01 -2.91506469e-01 -4.30571675e-01 -5.26547790e-01
4.66215730e-01 4.48906898e-01 -7.63488412e-02 -9.81811434e-02
-1.33604312e+00 4.46435004e-01 4.24593836e-02 5.76315284e-01
-7.45228410e-01 1.07900731e-01 9.60982144e-01 2.84961343e-01
3.38593423e-01 2.96908110e-01 -7.68019319e-01 -8.35224271e-01
-8.80276918e-01 -4.70658019e-02 1.02316189e+00 5.75533807e-01
-4.75275815e-02 8.89869183e-02 1.57355726e-01 3.37777615e-01
3.24594289e-01 2.27996826e-01 7.09040582e-01 -7.18988895e-01
3.87083560e-01 6.27256751e-01 -3.33210900e-02 -9.72666085e-01
-9.80026186e-01 -4.62969661e-01 -1.14909363e+00 -1.58723127e-02
5.69255650e-01 -2.31427968e-01 -6.50864363e-01 9.96249437e-01
2.98061401e-01 4.67341751e-01 4.27616611e-02 7.30355859e-01
7.23205328e-01 2.09927678e-01 1.33974224e-01 -1.96884856e-01
1.49755991e+00 -2.60165930e-01 -8.63946080e-01 2.47174904e-01
9.74835575e-01 -5.46117365e-01 1.03779173e+00 7.35832155e-01
-9.39936221e-01 -3.94049644e-01 -1.09285939e+00 6.03614561e-02
-3.00332665e-01 -2.19326511e-01 6.85220599e-01 1.12640715e+00
-6.72539115e-01 8.00696492e-01 -1.53031385e+00 -4.24933076e-01
8.54291856e-01 4.89105046e-01 -1.80571347e-01 9.83483940e-02
-1.03421056e+00 7.48402476e-01 1.27045229e-01 2.87348717e-01
-6.88029826e-01 -8.16252410e-01 -4.36678886e-01 1.45801872e-01
-1.67217597e-01 -7.03246415e-01 9.72274423e-01 -5.17704189e-01
-1.28030348e+00 7.09954023e-01 4.11637008e-01 -7.92890966e-01
5.28759480e-01 -3.24038029e-01 -7.76792109e-01 1.59387574e-01
-2.43208647e-01 4.39432561e-02 5.00177503e-01 -2.69569784e-01
-5.57561338e-01 -5.70199847e-01 3.21680680e-02 -2.43610859e-01
-5.70513606e-01 8.71164501e-02 -5.11497818e-02 -4.42913741e-01
-1.11454092e-01 -1.04698145e+00 -2.55655885e-01 -1.47977918e-01
-2.80711681e-01 2.00132743e-01 1.00351179e+00 -5.51195443e-01
1.18648803e+00 -2.16920447e+00 -3.83605987e-01 1.50248021e-01
5.13098836e-01 6.29486084e-01 4.46530670e-01 8.21498856e-02
-5.47784157e-02 -3.31293494e-02 -1.71754751e-02 -2.03841012e-02
-4.68985528e-01 4.34601337e-01 -2.13974901e-02 6.15391195e-01
-1.83638811e-01 6.57514334e-01 -6.11456573e-01 -5.94863832e-01
3.73027831e-01 6.45965815e-01 -6.21084988e-01 1.35155097e-01
1.59226924e-01 7.57939160e-01 -5.40861309e-01 6.54432058e-01
4.58291829e-01 -5.20721793e-01 2.94921666e-01 -3.91470462e-01
2.71686297e-02 2.79192716e-01 -1.13862002e+00 1.96927106e+00
-3.42088699e-01 2.67704427e-01 -8.13518316e-02 -1.06175971e+00
6.74822628e-01 6.39434099e-01 1.18276680e+00 -7.08146214e-01
4.23099637e-01 1.77149326e-01 4.06658679e-01 -9.11549211e-01
1.81843806e-02 5.91173535e-03 9.65939835e-02 5.28967559e-01
-2.29798958e-01 3.38527560e-01 7.00052083e-02 4.69453223e-02
1.56724191e+00 -7.70699903e-02 1.83724925e-01 -2.76212782e-01
4.07787859e-01 -1.11977365e-02 5.94283342e-01 6.46017849e-01
-5.60339093e-01 5.32594085e-01 3.22029889e-01 -1.16256678e+00
-8.62855494e-01 -8.43264520e-01 -4.16740984e-01 6.49853885e-01
-3.52324903e-01 -5.55167675e-01 -5.81943631e-01 -5.69787323e-01
-2.09545463e-01 2.17631161e-01 -1.31795242e-01 -4.49503735e-02
-6.54092848e-01 -1.21545660e+00 8.63815486e-01 7.32411802e-01
4.35072422e-01 -1.11557007e+00 -1.74258256e+00 7.19219387e-01
-1.60543725e-01 -1.25864148e+00 -9.95685253e-03 3.92424278e-02
-1.26983476e+00 -1.12537026e+00 -3.27661484e-01 -2.82952666e-01
1.17996752e-01 -3.89086664e-01 1.16167212e+00 1.68464169e-01
-8.93144429e-01 3.67286235e-01 -2.73452401e-01 -8.64204884e-01
-2.67782688e-01 2.43680298e-01 2.62132734e-01 3.39828618e-02
5.94005644e-01 -7.16711879e-01 -1.07578814e+00 1.47048011e-01
-8.60963404e-01 -1.72882780e-01 2.55105287e-01 5.23917556e-01
6.56708777e-01 2.44298317e-02 5.53824604e-01 -7.36927509e-01
5.92444420e-01 -5.33467829e-01 -4.51764435e-01 -1.55088663e-01
-8.57762218e-01 -1.95777610e-01 5.03106415e-01 -3.31778318e-01
-4.61303025e-01 1.47073239e-01 -5.37935615e-01 -3.83929640e-01
-1.66244894e-01 4.58219051e-01 7.48605281e-02 3.94972682e-01
5.86248815e-01 -3.40289325e-01 2.34393984e-01 -5.19503176e-01
-2.72891432e-01 9.26135004e-01 3.70971233e-01 -2.98079967e-01
6.41002506e-02 6.56157374e-01 1.95065156e-01 -9.29222345e-01
-6.45661473e-01 -4.25050795e-01 -5.54673076e-01 -6.03326969e-02
1.08044255e+00 -8.13958108e-01 -1.18732071e+00 4.49437529e-01
-9.67126489e-01 -2.37627476e-01 -6.02164090e-01 6.67724133e-01
-2.21314803e-01 1.71090245e-01 -6.90738201e-01 -4.88115937e-01
-8.21846247e-01 -1.12551284e+00 6.38444781e-01 -1.08362712e-01
-5.36361635e-01 -7.85570860e-01 2.06243366e-01 3.11497062e-01
9.46031451e-01 6.62122726e-01 4.83287930e-01 -8.61574471e-01
-3.53449106e-01 -4.04904693e-01 -3.50017175e-02 4.44065332e-01
2.46955827e-01 -2.71535933e-01 -9.51707006e-01 -3.75480533e-01
3.18503588e-01 -1.56143099e-01 2.50035405e-01 4.22617525e-01
1.56647396e+00 -1.39267311e-01 -2.87406415e-01 7.29022682e-01
1.07482040e+00 3.87646258e-01 6.79722846e-01 2.12573424e-01
7.21578479e-01 5.47312899e-03 2.04832345e-01 7.79592872e-01
3.36324304e-01 4.29205686e-01 4.52250242e-01 -2.29884118e-01
1.01246014e-01 4.47824448e-01 -1.49970248e-01 9.46873665e-01
-3.23166251e-01 -2.45799311e-04 -1.28836823e+00 3.88409883e-01
-1.65702868e+00 -8.43080282e-01 -3.69456142e-01 2.05486917e+00
5.83055019e-01 2.91704684e-01 2.03603521e-01 6.01582229e-01
1.78644136e-01 -3.89764547e-01 -5.88106871e-01 -5.26621580e-01
3.22708905e-01 4.96382445e-01 2.20143035e-01 -5.12783565e-02
-9.36759353e-01 1.53392315e-01 6.39969635e+00 -1.68926239e-01
-1.47801483e+00 4.62717563e-01 6.93128824e-01 -3.53577733e-01
5.05938768e-01 -4.92032737e-01 -3.98635954e-01 7.62015045e-01
1.55309713e+00 1.21588901e-01 9.92626175e-02 8.60927105e-01
3.50141048e-01 1.12927839e-01 -1.03883481e+00 1.37794197e+00
-2.48404518e-01 -1.34793139e+00 -5.51188529e-01 1.75432190e-01
1.10226706e-01 5.09190679e-01 -3.04263622e-01 -5.83848022e-02
-2.25884452e-01 -8.91127527e-01 6.30266443e-02 4.93724823e-01
8.23238432e-01 -5.99760056e-01 1.10174763e+00 -8.00209343e-02
-9.24412370e-01 -1.38408497e-01 5.30020148e-03 -3.27331364e-01
2.09574059e-01 7.65176594e-01 -9.96273220e-01 1.92089692e-01
1.29038572e+00 5.03971279e-01 -4.62727040e-01 1.07804906e+00
4.43103760e-01 8.08072388e-01 -5.76843679e-01 2.11461797e-01
-2.28942111e-01 2.15188324e-01 2.87511796e-01 1.09331143e+00
3.30826700e-01 2.60393433e-02 3.73867244e-01 3.00238937e-01
5.24570197e-02 1.59879148e-01 -4.32338804e-01 8.09341446e-02
1.97701883e-02 1.34901047e+00 -8.65625679e-01 -4.45410520e-01
-6.54155791e-01 8.12208354e-01 -3.74250710e-01 -2.21883103e-01
-1.11317062e+00 -3.48392606e-01 8.32688510e-01 4.36891317e-01
-3.86532024e-02 -9.23470706e-02 -4.68273252e-01 -1.17640305e+00
7.42665026e-03 -1.00981307e+00 6.74296737e-01 -3.70834976e-01
-1.26755202e+00 8.59953582e-01 -2.32842639e-01 -1.20180583e+00
-2.18521267e-01 -4.60742503e-01 -3.95024627e-01 6.15998447e-01
-1.13728666e+00 -6.74500883e-01 -5.73060453e-01 9.50826406e-01
1.17854945e-01 -9.58822221e-02 1.40073419e+00 9.39838767e-01
-5.85198820e-01 4.16346967e-01 -1.59296736e-01 1.77138746e-01
3.75584632e-01 -1.03834629e+00 3.21638793e-01 6.67210102e-01
-1.09206058e-01 4.80487317e-01 5.86395741e-01 -3.97888422e-01
-1.49074137e+00 -9.81193960e-01 6.01038694e-01 -5.14257252e-01
5.69441855e-01 -1.77874655e-01 -9.90581214e-01 6.51615739e-01
-5.57179078e-02 7.96699286e-01 1.02251232e+00 -2.30847951e-02
-3.17188650e-02 -5.96368134e-01 -1.35474730e+00 6.32662326e-02
7.15275407e-01 -2.66676486e-01 -3.73522490e-01 3.71009529e-01
3.09968442e-01 -6.06862128e-01 -1.33253992e+00 3.58029097e-01
6.86911225e-01 -1.02908289e+00 8.56044471e-01 -6.98261440e-01
6.37583584e-02 -3.35564166e-01 -9.56732221e-03 -9.53195632e-01
1.05685264e-01 -7.12275445e-01 -3.60095084e-01 9.90047812e-01
2.59736240e-01 -7.71943927e-01 1.02462339e+00 8.99973869e-01
-1.22658066e-01 -6.98706567e-01 -1.10577035e+00 -4.49443519e-01
-3.48736674e-01 -7.39181936e-01 6.88267469e-01 9.22002256e-01
1.71864554e-01 2.77255327e-01 -2.85613954e-01 5.89000024e-02
4.30715740e-01 -2.98807621e-02 5.67850888e-01 -1.43897367e+00
-4.13579673e-01 1.38373390e-01 -8.52964520e-01 -2.02984735e-01
-6.34018183e-01 -7.92763650e-01 -5.70184290e-01 -1.47984827e+00
-3.92301008e-02 -7.17627466e-01 -8.02481890e-01 6.23392820e-01
2.98936099e-01 6.75843000e-01 -1.86885241e-03 7.59798139e-02
-5.64470172e-01 -1.25653386e-01 7.53414035e-01 -4.46942858e-02
-4.61453497e-01 1.06846072e-01 -3.30246150e-01 7.24502206e-01
1.07800329e+00 -6.90337777e-01 -3.62163514e-01 -3.40852052e-01
4.64277565e-01 8.71979669e-02 3.39943439e-01 -1.55494547e+00
3.57606374e-02 4.21330899e-01 6.88578308e-01 -5.09089351e-01
2.68131882e-01 -1.15484548e+00 3.64594698e-01 9.42572415e-01
-1.96913481e-02 5.98675251e-01 2.81155616e-01 2.11979821e-01
5.75030856e-02 3.35143328e-01 6.61655307e-01 -5.48880175e-02
-1.83668405e-01 4.55683887e-01 -4.75738078e-01 1.53101459e-01
1.08075571e+00 -2.79339571e-02 -1.44274712e-01 4.74481061e-02
-9.21274006e-01 -2.38980399e-03 1.25308156e-01 3.85396719e-01
5.57760239e-01 -9.75445569e-01 -5.10243475e-01 3.74399573e-01
1.33660346e-01 -2.00431664e-02 3.21141481e-01 1.21284485e+00
-1.09844041e+00 2.84800649e-01 -6.50547147e-01 -9.00757313e-01
-1.47878408e+00 7.25483894e-01 4.16113734e-01 -4.47562724e-01
-1.09148049e+00 2.05552667e-01 -5.72470903e-01 7.54346773e-02
2.82239288e-01 -6.69599652e-01 -1.01540409e-01 -1.20001271e-01
8.81175935e-01 5.91355026e-01 7.34088719e-01 -2.13763595e-01
-6.89307809e-01 2.38195751e-02 6.70829415e-02 3.08953822e-01
1.40118575e+00 -1.57689396e-02 -9.45965946e-02 3.63550097e-01
9.65369701e-01 -4.12900865e-01 -7.69561589e-01 2.94608504e-01
8.93538967e-02 -1.21260963e-01 1.53957352e-01 -7.01368451e-01
-1.28185201e+00 1.05060148e+00 1.41989446e+00 4.12552178e-01
1.20054138e+00 -4.10590112e-01 1.08802640e+00 3.72258902e-01
4.31163490e-01 -8.66232455e-01 -6.39605671e-02 1.59597293e-01
2.94611156e-01 -1.08023238e+00 3.39030288e-02 4.34016362e-02
-5.48038900e-01 1.03481448e+00 1.32436469e-01 1.99771181e-01
1.11781538e+00 7.01247334e-01 4.09328789e-01 -4.01269168e-01
-5.10506153e-01 2.06528023e-01 -3.16966146e-01 5.63956678e-01
6.98721766e-01 9.88005996e-02 -2.22141147e-01 4.29266244e-01
-7.54601061e-02 5.99162519e-01 6.64129794e-01 1.19851303e+00
1.91048279e-01 -1.17097545e+00 -2.81360209e-01 8.38189483e-01
-1.50264513e+00 2.13656835e-02 3.55180174e-01 3.68447900e-01
2.84704775e-01 1.01735616e+00 2.27930322e-01 -1.28672600e-01
3.52425605e-01 3.45882624e-01 3.26839864e-01 -3.55315536e-01
-8.15667331e-01 -7.76824430e-02 2.57115830e-02 -9.02990997e-01
-4.14539427e-01 -5.31449497e-01 -1.31994593e+00 -3.38297606e-01
1.15837976e-01 -1.79918200e-01 1.05377924e+00 8.21537495e-01
8.64906907e-01 9.24038112e-01 2.69622337e-02 -4.67986196e-01
-4.64440942e-01 -1.01447308e+00 -5.55742264e-01 5.44296026e-01
2.23749995e-01 -2.79894382e-01 6.82891235e-02 2.30871186e-01] | [13.972831726074219, 3.270650625228882] |
ce3067a9-51d2-42b7-bdcf-caef0fee7d93 | stop-a-dataset-for-spoken-task-oriented | 2207.10643 | null | https://arxiv.org/abs/2207.10643v3 | https://arxiv.org/pdf/2207.10643v3.pdf | STOP: A dataset for Spoken Task Oriented Semantic Parsing | End-to-end spoken language understanding (SLU) predicts intent directly from audio using a single model. It promises to improve the performance of assistant systems by leveraging acoustic information lost in the intermediate textual representation and preventing cascading errors from Automatic Speech Recognition (ASR). Further, having one unified model has efficiency advantages when deploying assistant systems on-device. However, the limited number of public audio datasets with semantic parse labels hinders the research progress in this area. In this paper, we release the Spoken Task-Oriented semantic Parsing (STOP) dataset, the largest and most complex SLU dataset to be publicly available. Additionally, we define low-resource splits to establish a benchmark for improving SLU when limited labeled data is available. Furthermore, in addition to the human-recorded audio, we are releasing a TTS-generated version to benchmark the performance for low-resource domain adaptation of end-to-end SLU systems. Initial experimentation show end-to-end SLU models performing slightly worse than their cascaded counterparts, which we hope encourages future work in this direction. | ['Yossi Adi', 'Abdelrahman Mohamed', 'Luke Zettlemoyer', 'Emmanuel Dupoux', 'Tu Ahn Nguyen', 'Robin Algayres', 'Wei-Ning Hsu', 'Jade Copet', 'Ali Elkahky', 'Adithya Sagar', 'Duc Le', 'Daniel Lazar', 'Akshat Shrivastava', 'Po-chun Hsu', 'Paden Tomasello'] | 2022-06-29 | null | null | null | null | ['spoken-language-understanding', 'spoken-language-understanding'] | ['natural-language-processing', 'speech'] | [ 3.25142205e-01 4.98317778e-01 -5.74903004e-02 -9.07572687e-01
-1.60016763e+00 -4.84178364e-01 2.59108782e-01 -3.10372144e-01
-5.21738410e-01 6.37321055e-01 8.69787753e-01 -4.45798486e-01
4.70113099e-01 -2.54320264e-01 -6.48368239e-01 -6.29017055e-02
2.34321300e-02 7.49023318e-01 1.67781845e-01 -3.13984305e-01
-3.36913645e-01 -1.76370278e-01 -1.38127732e+00 6.85846269e-01
6.83142722e-01 9.53938961e-01 5.07469296e-01 1.12931764e+00
-1.28747849e-02 9.16019499e-01 -6.80599034e-01 -4.51161899e-02
-6.70335591e-02 -6.28454268e-01 -1.23224890e+00 -9.06364024e-02
3.28432798e-01 -6.50310755e-01 -2.65232176e-01 5.37177205e-01
8.26269627e-01 2.23464787e-01 1.74326539e-01 -1.12690341e+00
-5.08853316e-01 8.75127554e-01 2.87299216e-01 -7.55024403e-02
7.74242342e-01 7.73094445e-02 1.33420813e+00 -8.61148715e-01
6.22964144e-01 1.52708220e+00 6.37152612e-01 1.00802648e+00
-1.02709901e+00 -5.47090352e-01 4.41479027e-01 2.51819074e-01
-1.03903043e+00 -1.35519028e+00 6.61942363e-01 2.45815575e-01
1.55779064e+00 3.65100682e-01 -2.72098724e-02 1.75048256e+00
-4.04090822e-01 1.55504489e+00 8.28540981e-01 -6.29196107e-01
3.13116044e-01 1.43887714e-01 4.82821554e-01 6.10121608e-01
-5.77808201e-01 -5.23292273e-02 -1.07439172e+00 -1.34876758e-01
-2.77290344e-02 -6.37138546e-01 -4.05980796e-01 -1.58649646e-02
-1.07196331e+00 6.01192415e-01 6.55788109e-02 1.07418582e-01
-1.25521138e-01 -1.01661064e-01 5.85407138e-01 6.04319513e-01
6.45411193e-01 3.08949202e-01 -9.53559816e-01 -8.42430472e-01
-7.04388618e-01 -3.31836566e-02 1.11192477e+00 1.24873459e+00
2.76332676e-01 1.18409067e-01 2.89983917e-02 1.31564403e+00
3.03470671e-01 4.68582898e-01 6.08409584e-01 -1.17253852e+00
7.67230928e-01 -2.17420273e-02 1.84801862e-01 -4.02422771e-02
-5.32562852e-01 -1.86756790e-01 -2.25839883e-01 -1.98019817e-01
6.16885841e-01 -3.53212088e-01 -7.57294059e-01 1.97838974e+00
-5.84787391e-02 3.04155707e-01 7.01092005e-01 9.05989885e-01
9.54693794e-01 8.69199872e-01 1.96910441e-01 -1.56261086e-01
1.32157719e+00 -1.25228775e+00 -9.93702650e-01 -7.57678866e-01
1.09534967e+00 -7.90159762e-01 1.68741453e+00 5.60249090e-01
-1.04413211e+00 -5.11198342e-01 -1.03476548e+00 -3.59867603e-01
-6.27130270e-02 2.55241901e-01 5.04902840e-01 5.37186027e-01
-1.18777382e+00 1.81350291e-01 -1.06778681e+00 -6.88283682e-01
-1.00268070e-02 1.02394737e-01 -3.84119332e-01 -1.76435798e-01
-1.50301051e+00 6.93815053e-01 2.46087179e-01 -3.45175080e-02
-8.39057744e-01 -5.93717992e-01 -9.83062923e-01 -7.73689374e-02
6.27702773e-01 -3.77999604e-01 2.13558984e+00 -7.33134210e-01
-2.02973104e+00 7.29808509e-01 -6.65556610e-01 -7.23280191e-01
3.90621096e-01 -6.54150665e-01 -4.16414678e-01 -1.23467200e-01
-2.56976336e-02 7.34987497e-01 4.20293599e-01 -1.02141619e+00
-8.78179610e-01 -5.64924717e-01 -2.39039660e-01 3.78460437e-01
-3.63230675e-01 2.30429456e-01 -1.83117568e-01 -4.08417374e-01
4.85891551e-02 -7.58688450e-01 1.46210492e-01 -5.73315024e-01
-3.81352156e-01 -6.23821080e-01 9.63671505e-01 -1.11970615e+00
1.18202055e+00 -2.28700852e+00 1.50754407e-01 -3.66927058e-01
-3.85305256e-01 1.98244110e-01 -4.30484384e-01 4.62080956e-01
2.12086245e-01 -1.15039445e-01 -1.64380491e-01 -9.56221938e-01
1.45659760e-01 3.68986607e-01 -6.42176926e-01 -7.23842755e-02
2.53173798e-01 6.52564287e-01 -9.11014438e-01 -2.50407994e-01
3.70786265e-02 1.58693612e-01 -7.73345828e-01 5.98176956e-01
-3.90321255e-01 4.43749428e-01 -4.30602968e-01 6.18073940e-01
6.77493587e-02 3.17261785e-01 8.36821720e-02 1.60408840e-01
1.15067586e-01 1.14165854e+00 -9.48323488e-01 2.18572307e+00
-8.62800658e-01 6.39324844e-01 5.19990504e-01 -8.85346949e-01
9.32201922e-01 6.08014166e-01 4.86337990e-02 -6.25591934e-01
-6.63635954e-02 2.90168554e-01 -2.65507042e-01 -2.90248811e-01
7.40679204e-01 -9.87542607e-03 -5.54199219e-01 4.23532844e-01
5.41683793e-01 -2.42956176e-01 -2.87313193e-01 2.94866353e-01
1.36484873e+00 1.50961084e-02 1.10906467e-01 1.01889394e-01
2.61195660e-01 1.67058825e-01 2.88235635e-01 9.94055450e-01
-6.71896636e-01 6.37343764e-01 1.60465509e-01 -1.10344790e-01
-8.76235127e-01 -1.14619744e+00 -4.74645682e-02 1.71349776e+00
-3.52687240e-01 -5.73166072e-01 -1.01417792e+00 -6.72396600e-01
-3.07435006e-01 1.33633864e+00 1.12052336e-01 -3.31779420e-02
-6.37257874e-01 -2.03714326e-01 9.56754088e-01 7.80838013e-01
4.36358601e-01 -1.10123050e+00 1.07324496e-01 5.30187249e-01
-7.84543693e-01 -1.59637189e+00 -3.87921691e-01 4.27347898e-01
-7.70111263e-01 -6.62124634e-01 -3.76193613e-01 -8.99217904e-01
-1.02788582e-01 2.82476217e-01 1.06355476e+00 -4.41310376e-01
1.40584052e-01 7.07114577e-01 -6.88175261e-01 -7.66667545e-01
-9.15488839e-01 5.07153749e-01 3.17798913e-01 -2.73921460e-01
7.60784268e-01 -2.64884502e-01 6.52675256e-02 3.19712609e-01
-3.06186765e-01 1.03882276e-01 2.39547968e-01 9.27725792e-01
2.67942339e-01 -3.56160551e-01 1.01229584e+00 -5.54767430e-01
7.62387097e-01 -1.72898784e-01 -1.31596744e-01 2.07292363e-01
-1.89458847e-01 7.24959671e-02 6.22398019e-01 -1.67764857e-01
-1.46095800e+00 1.51864827e-01 -6.05315864e-01 1.78897157e-02
-5.66457331e-01 4.02068734e-01 -5.19858897e-01 7.07727790e-01
6.05497360e-01 3.45389158e-01 7.43356794e-02 -7.04880178e-01
5.44727683e-01 1.45725989e+00 9.87773359e-01 -5.55837154e-01
9.90313441e-02 -1.47448005e-02 -1.00223780e+00 -1.33130491e+00
-1.01566660e+00 -6.74902260e-01 -4.00231987e-01 4.83005159e-02
7.65490532e-01 -1.23910499e+00 -4.72128838e-01 5.47286928e-01
-1.31503403e+00 -8.55930150e-01 -4.85589989e-02 5.72465897e-01
-9.19720709e-01 1.99304789e-01 -8.05593252e-01 -1.14346933e+00
-2.72668988e-01 -1.05775464e+00 1.27840936e+00 -2.42518798e-01
-7.67469227e-01 -6.43129110e-01 -1.90455228e-01 8.53330135e-01
3.35001171e-01 -7.17563212e-01 5.28273582e-01 -1.11738944e+00
-2.62967348e-01 -2.72018343e-01 1.81228518e-01 6.26795650e-01
-9.72717032e-02 -5.59316218e-01 -1.48691094e+00 -2.87464291e-01
3.51001956e-02 -9.09200490e-01 7.19119072e-01 2.74023980e-01
8.71573091e-01 -2.12587133e-01 -2.07561404e-01 2.01565117e-01
4.80585665e-01 2.63005435e-01 4.09786254e-01 1.79792002e-01
3.27503175e-01 7.56300688e-01 7.59359002e-01 3.32198143e-01
6.06756032e-01 8.06484818e-01 -8.87443647e-02 1.98518842e-01
-4.61508453e-01 -5.91140568e-01 9.56157029e-01 1.13354504e+00
4.13707763e-01 -4.93917137e-01 -9.61264670e-01 5.47629595e-01
-1.77531946e+00 -6.56851232e-01 1.69627175e-01 2.01670098e+00
9.97695327e-01 1.98066980e-01 7.43030943e-03 -6.16976768e-02
3.43830377e-01 2.44662240e-01 -2.89289057e-01 -4.39123094e-01
-1.08065559e-02 3.49083431e-02 2.02995032e-01 1.10246825e+00
-1.13467157e+00 1.54336607e+00 6.32466364e+00 7.00157642e-01
-7.86155879e-01 3.84177417e-01 4.88645047e-01 -2.26279750e-01
-7.25188032e-02 -3.33585382e-01 -1.22114933e+00 2.54470408e-01
1.68145895e+00 4.21025790e-02 4.81381118e-01 1.15233278e+00
4.46185201e-01 -9.17325243e-02 -1.32556593e+00 7.76225626e-01
8.38957727e-02 -7.05450714e-01 -3.06076258e-01 -4.16957825e-01
9.16183889e-02 5.27054846e-01 -2.06796497e-01 8.32006633e-01
6.41848981e-01 -7.63044059e-01 7.86944032e-01 6.66969223e-03
6.71096921e-01 -6.09238207e-01 8.04958940e-01 4.90319192e-01
-8.62266362e-01 -5.59432991e-02 -2.16656625e-01 -2.32425645e-01
5.02768576e-01 3.19184996e-02 -1.51582539e+00 3.05574715e-01
6.74966514e-01 3.99700463e-01 -2.03500599e-01 2.73994058e-01
-3.89591455e-01 1.26303875e+00 -5.91151476e-01 -1.54009297e-01
3.95623982e-01 2.23599955e-01 6.99432850e-01 1.38984692e+00
2.29757160e-01 2.81594634e-01 3.67490351e-01 4.39632237e-01
-1.62409931e-01 4.59995493e-02 -8.26052129e-01 -1.24054700e-01
6.81422353e-01 5.49686849e-01 -1.49146155e-01 -2.96807230e-01
-5.17589033e-01 1.41983771e+00 3.54309708e-01 4.35464203e-01
-4.06091869e-01 -1.71306789e-01 1.03032565e+00 -2.60751337e-01
-2.58269548e-01 -5.90548813e-01 -2.64621466e-01 -1.07572281e+00
6.67986572e-02 -9.80973780e-01 4.45828557e-01 -9.52962399e-01
-1.09525347e+00 8.30908239e-01 -1.65951595e-01 -7.77360380e-01
-8.88300359e-01 -6.88009441e-01 -2.16870219e-01 8.49321544e-01
-1.48364460e+00 -1.16467285e+00 -1.00357503e-01 3.86952788e-01
1.38952720e+00 -3.18949223e-01 1.49603534e+00 6.69219792e-02
-4.28057015e-01 8.28985393e-01 -5.70901996e-03 2.24859431e-01
1.19789410e+00 -1.09813714e+00 6.90908909e-01 7.51953602e-01
2.70484418e-01 2.18312234e-01 9.13756728e-01 -5.39573550e-01
-1.28910053e+00 -8.86637628e-01 1.22748530e+00 -9.56721425e-01
7.02694595e-01 -7.51081407e-01 -9.88560438e-01 1.12217951e+00
2.87913412e-01 -2.90232331e-01 7.11035371e-01 6.25406981e-01
-2.02633023e-01 1.16650462e-01 -8.63687575e-01 5.81379116e-01
1.29329264e+00 -8.48044038e-01 -9.02203262e-01 2.26569772e-01
1.26874304e+00 -5.22356689e-01 -5.98353505e-01 2.44405895e-01
5.30109704e-01 -7.10310400e-01 7.41524339e-01 -9.80361104e-01
2.48831257e-01 1.78349525e-01 -5.43122649e-01 -1.71268654e+00
1.49352238e-01 -7.37003267e-01 -8.36004689e-03 1.41172469e+00
7.57562816e-01 -5.36383152e-01 7.89520085e-01 6.13235176e-01
-7.67565489e-01 -2.91421920e-01 -1.03497660e+00 -9.07280445e-01
-1.16724618e-01 -1.21801329e+00 1.99079871e-01 6.25740409e-01
3.02680701e-01 7.24787593e-01 -4.14692432e-01 3.48489970e-01
3.19405794e-01 -5.51542580e-01 9.35206115e-01 -8.33041370e-01
-3.11237305e-01 -8.40990543e-02 -1.81188792e-01 -1.51020527e+00
6.98525250e-01 -9.08646822e-01 5.40303469e-01 -1.34949803e+00
-3.12408179e-01 -3.66613090e-01 -1.54158190e-01 7.32351303e-01
-1.75728962e-01 -2.09986135e-01 2.39141211e-01 2.27698125e-03
-6.46259665e-01 8.54781926e-01 6.36921763e-01 -1.87640458e-01
-5.71496010e-01 1.53423309e-01 -5.11341512e-01 7.16084540e-01
8.11306715e-01 -2.42842913e-01 -4.47024494e-01 -7.17617333e-01
-6.61166131e-01 2.72690654e-01 -4.93718386e-02 -1.02683270e+00
2.62311727e-01 3.21393907e-01 -1.66400641e-01 -6.14627779e-01
6.16056979e-01 -5.96902668e-01 -7.12918937e-01 5.37676029e-02
-6.95401490e-01 -3.85809034e-01 3.15698802e-01 4.01911706e-01
-4.16788936e-01 -1.62434772e-01 5.09307325e-01 -4.99594584e-02
-1.10365045e+00 -2.50649095e-01 -8.53307247e-01 1.96187004e-01
4.57361162e-01 -5.92348315e-02 -5.15077747e-02 -1.01678836e+00
-9.28652287e-01 4.24669653e-01 -4.80669737e-02 9.79603887e-01
5.30122936e-01 -9.95630205e-01 -8.45350683e-01 3.99830461e-01
3.99981052e-01 3.70931327e-02 2.04610169e-01 4.87593889e-01
5.65380193e-02 8.32146943e-01 2.20256671e-01 -6.23915553e-01
-1.52922404e+00 4.23072465e-02 1.74299911e-01 -5.92076927e-02
-6.08511865e-01 1.24975777e+00 5.71517274e-03 -9.72480595e-01
9.30283785e-01 -3.68915588e-01 2.12533697e-01 -1.45234019e-01
9.12666857e-01 2.42702693e-01 2.77948499e-01 -3.56796890e-01
-2.98742503e-01 -3.38348031e-01 -1.55773088e-01 -6.72121048e-01
1.27513385e+00 -4.42514777e-01 4.82270539e-01 6.94601893e-01
1.13908637e+00 -1.05999224e-01 -1.27802038e+00 -5.24262249e-01
2.54499644e-01 6.85006678e-02 2.29195610e-01 -1.13397253e+00
-2.19215348e-01 9.82192039e-01 4.76114571e-01 2.94999480e-02
8.35018098e-01 2.60626942e-01 1.20205355e+00 9.48591232e-01
6.89899862e-01 -1.35107148e+00 4.37101573e-02 1.07124686e+00
1.10620332e+00 -1.42206979e+00 -7.43683577e-01 -3.66535366e-01
-9.78958130e-01 8.48144352e-01 7.27917016e-01 4.37425196e-01
2.80929923e-01 5.19580245e-01 5.08192539e-01 3.64527613e-01
-8.73891413e-01 -2.22160250e-01 -1.06337473e-01 7.83374250e-01
6.14301920e-01 3.35199088e-01 8.09798315e-02 1.18891513e+00
-6.44083917e-01 -3.35643603e-03 4.27882463e-01 8.66214216e-01
-6.33837521e-01 -1.23301280e+00 -4.11039740e-01 1.33527949e-01
-3.08593392e-01 -2.06587404e-01 -7.71030664e-01 6.34557486e-01
-5.89595497e-01 1.46981204e+00 5.11595281e-03 -4.09489483e-01
7.16250181e-01 8.83750081e-01 2.38286823e-01 -7.29979157e-01
-8.00102577e-02 3.53152715e-02 9.11179364e-01 -8.24945927e-01
9.43499878e-02 -6.16237760e-01 -1.46653986e+00 1.24338225e-01
-1.38542175e-01 1.52288035e-01 6.91242814e-01 9.56446290e-01
4.17549491e-01 5.50530851e-01 4.85255808e-01 -7.44599581e-01
-9.38984931e-01 -1.37733340e+00 -3.60683590e-01 1.57715216e-01
2.75704384e-01 -3.53499502e-01 -2.88698018e-01 6.02589659e-02] | [14.061359405517578, 6.982806205749512] |
9fa8c9de-2824-4748-8363-0e372b7df88f | light-source-separation-and-intrinsic-image | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Yoshida_Light_Source_Separation_and_Intrinsic_Image_Decomposition_Under_AC_Illumination_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Yoshida_Light_Source_Separation_and_Intrinsic_Image_Decomposition_Under_AC_Illumination_CVPR_2023_paper.pdf | Light Source Separation and Intrinsic Image Decomposition Under AC Illumination | Artificial light sources are often powered by an electric grid, and then their intensities rapidly oscillate in response to the grid's alternating current (AC). Interestingly, the flickers of scene radiance values due to AC illumination are useful for extracting rich information on a scene of interest. In this paper, we show that the flickers due to AC illumination is useful for intrinsic image decomposition (IID). Our proposed method conducts the light source separation (LSS) followed by the IID under AC illumination. In particular, we reveal the ambiguity in the blind LSS via matrix factorization and the ambiguity in the IID assuming the Lambert model, and then show why and how those ambiguities can be resolved. We experimentally confirmed that our method can recover the colors of the light sources, the diffuse reflectance values, and the diffuse and specular intensities (shadings) under each of the light sources, and that the IID under AC illumination is effective for application to auto white balancing. | ['Takahiro Okabe', 'Ryo Kawahara', 'Yusaku Yoshida'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['intrinsic-image-decomposition'] | ['computer-vision'] | [ 6.44034505e-01 -7.51897097e-01 5.05068541e-01 2.25633066e-02
-2.07046553e-01 -8.61925840e-01 4.82683569e-01 -7.25086272e-01
1.95460796e-01 7.02979684e-01 1.49356410e-01 -1.54541552e-01
-3.74939404e-02 -4.17452306e-01 -3.76275003e-01 -1.42927265e+00
5.30576110e-01 -3.61874521e-01 -1.85914293e-01 -8.66991654e-02
2.62315243e-01 6.26898468e-01 -1.94654751e+00 2.24638626e-01
1.17415547e+00 1.07045054e+00 3.05953383e-01 7.91667342e-01
-5.18538535e-01 8.66916716e-01 -9.49859262e-01 3.02975066e-02
4.56950545e-01 -6.73284709e-01 -3.12180042e-01 3.29472572e-01
2.57195264e-01 -2.29188710e-01 -7.08168074e-02 1.37347245e+00
2.80381680e-01 -6.50326908e-02 6.56512320e-01 -1.37823391e+00
-6.79414630e-01 -2.75072008e-01 -1.09859061e+00 -1.98642805e-01
5.55738986e-01 2.02814519e-01 5.72992325e-01 -8.46156478e-01
6.02625199e-02 9.45812047e-01 3.31441373e-01 3.14678371e-01
-1.08604586e+00 -5.80683291e-01 -2.32834727e-01 3.54282498e-01
-1.14565611e+00 -6.49295092e-01 1.40393651e+00 -3.26363176e-01
3.90820026e-01 7.17875838e-01 7.23566651e-01 7.15854883e-01
1.90870807e-01 5.96218050e-01 1.83848441e+00 -8.62520754e-01
3.84722948e-01 2.60541618e-01 1.44839451e-01 3.03099185e-01
4.79456246e-01 9.15580541e-02 -4.17670071e-01 -3.13766748e-01
5.80478430e-01 1.05934329e-01 -9.31302667e-01 3.05054598e-02
-9.38685596e-01 8.31336901e-02 1.31992862e-01 2.44462267e-01
-2.26428077e-01 -2.30049640e-01 -3.21050823e-01 1.33632854e-01
2.21544489e-01 4.63283211e-01 -2.26343021e-01 3.78213942e-01
-6.09419942e-01 -3.13161314e-01 7.39299655e-01 7.44349957e-01
9.09539878e-01 2.53469676e-01 1.48207486e-01 9.83081341e-01
3.75234574e-01 1.39781868e+00 3.87778491e-01 -9.67141390e-01
1.27321379e-02 3.63911331e-01 8.02850842e-01 -9.45240915e-01
-3.10414374e-01 -2.49537975e-01 -8.79159153e-01 3.88196141e-01
5.36092877e-01 -3.65930468e-01 -7.56002307e-01 1.45719647e+00
3.26935768e-01 3.30516845e-01 2.67556638e-01 1.31695044e+00
7.67283082e-01 4.93227899e-01 -6.48795128e-01 -8.63312542e-01
1.34163284e+00 -6.33017123e-01 -1.26909733e+00 -2.19668224e-01
2.57573314e-02 -1.22195888e+00 1.02851713e+00 5.52854955e-01
-9.20876920e-01 -5.86605251e-01 -9.51097250e-01 1.25658363e-01
6.82358891e-02 4.63102371e-01 5.85872293e-01 8.46257448e-01
-9.99418497e-01 -7.91028365e-02 -4.58943874e-01 -7.19204172e-02
-1.86991960e-01 -1.43150002e-01 1.08373262e-01 -2.56612748e-01
-7.26940036e-01 5.16571343e-01 -6.54217124e-01 6.29796147e-01
-2.86776751e-01 -4.91808742e-01 -3.53304118e-01 -1.59410059e-01
2.48946741e-01 -2.69197971e-01 7.53059149e-01 -1.32621574e+00
-1.83165658e+00 6.53562367e-01 -7.93523431e-01 2.37893537e-01
1.41495273e-01 -1.30382091e-01 -8.86026204e-01 3.88339877e-01
-4.55034636e-02 -2.97416598e-01 1.15803432e+00 -1.97194433e+00
-2.15433404e-01 -4.38819766e-01 -1.88539073e-01 2.41367340e-01
5.07614622e-03 -1.50557347e-02 -3.54011089e-01 -2.26172894e-01
4.18776095e-01 -9.61938679e-01 2.00117543e-01 -1.84347063e-01
-5.82127690e-01 3.32621574e-01 8.50702345e-01 -6.04733467e-01
7.49578774e-01 -2.36873794e+00 -2.79994100e-01 2.10848793e-01
7.02811107e-02 8.49608779e-02 -1.32847056e-01 3.93122673e-01
-2.87283033e-01 -3.94516528e-01 -2.92770267e-01 2.22099498e-01
-2.67907143e-01 5.97019680e-02 -5.97350299e-01 6.28292799e-01
-1.41989589e-01 4.67244744e-01 -7.49695182e-01 -4.92927805e-03
1.88562796e-01 5.20463824e-01 -1.68465916e-02 4.89193380e-01
2.35938624e-01 4.61469203e-01 -1.26389325e-01 6.61519885e-01
1.37747765e+00 2.86728702e-02 1.92775741e-01 -7.95255482e-01
-3.25460941e-01 -7.69880190e-02 -1.42553926e+00 1.05208230e+00
-3.65762234e-01 7.45954156e-01 6.24371231e-01 -5.20457804e-01
9.94382203e-01 1.23652518e-01 5.03264427e-01 -9.84373331e-01
-7.34848976e-02 2.12895662e-01 -1.90189064e-01 -5.45022190e-01
2.47204438e-01 -2.11627558e-01 6.79774344e-01 5.91440916e-01
-4.21719313e-01 -4.24695939e-01 -1.01001814e-01 5.22975298e-03
5.77560067e-01 5.88466041e-02 6.25983030e-02 -4.39204782e-01
6.77778482e-01 -3.15968722e-01 4.67753828e-01 5.41816950e-01
1.97404847e-01 5.40760696e-01 2.40637183e-01 -3.30453753e-01
-5.54139495e-01 -8.90740275e-01 -1.81291744e-01 3.65759641e-01
6.24977350e-01 8.40806868e-03 -7.52234757e-01 -3.42883766e-02
-1.27912194e-01 8.91675651e-01 -1.18404806e-01 1.10277072e-01
-1.28280237e-01 -1.16678298e+00 -1.55646894e-02 9.64599289e-03
6.69617176e-01 -5.73683619e-01 -4.28417176e-01 -1.77314878e-01
-5.41395903e-01 -1.20762026e+00 -2.82182515e-01 -2.94390339e-02
-3.86806071e-01 -1.55194128e+00 -6.65044367e-01 -2.33279198e-01
9.42211926e-01 1.05158079e+00 9.25701022e-01 -4.59557250e-02
-6.40245497e-01 9.08959210e-01 -2.03688323e-01 -5.50154746e-01
-1.13809474e-01 -1.02751064e+00 2.39613637e-01 7.12019086e-01
3.29994351e-01 -6.53859258e-01 -8.57906103e-01 6.68517172e-01
-8.21527600e-01 1.79101020e-01 1.15213796e-01 4.22376692e-01
4.07161891e-01 4.84926134e-01 2.88805757e-02 -7.18865275e-01
4.89062548e-01 -6.79011270e-02 -1.01336849e+00 2.05335334e-01
-4.04436082e-01 -2.72055775e-01 7.17505515e-01 -2.23993644e-01
-1.74963033e+00 -2.30841830e-01 3.93099368e-01 -9.59760845e-02
-4.48304802e-01 -2.27472439e-01 -5.19861281e-01 -4.43159044e-01
5.78057468e-01 3.52126032e-01 -3.18471968e-01 -6.44778728e-01
2.28136763e-01 7.61220574e-01 5.19672871e-01 -5.54575562e-01
9.62767422e-01 9.51131105e-01 2.49836758e-01 -1.33778095e+00
-7.88095593e-01 -4.47836936e-01 -3.09782982e-01 -3.48681241e-01
7.30466962e-01 -9.50459242e-01 -1.17923784e+00 9.77073610e-01
-1.04493463e+00 -1.41795710e-01 -1.59690708e-01 5.62966347e-01
-1.33080259e-01 6.22532189e-01 -5.07832289e-01 -1.42161429e+00
1.13537638e-02 -1.10736346e+00 7.66697049e-01 5.68011343e-01
3.34470004e-01 -6.85142219e-01 -3.12379468e-02 4.57103431e-01
3.15096110e-01 -1.08345091e-01 9.98702407e-01 4.62875366e-01
-7.26366878e-01 1.65029302e-01 -5.08728325e-01 3.48238736e-01
6.84526026e-01 7.85666257e-02 -1.48563504e+00 -2.70185828e-01
7.17834830e-01 2.22215444e-01 5.60076952e-01 4.99710172e-01
9.77184355e-01 -3.35521966e-01 -7.98629317e-03 9.40741301e-01
1.80142570e+00 4.02667969e-01 8.56333315e-01 9.17112455e-02
6.96890056e-01 5.24981976e-01 1.67818055e-01 3.13227654e-01
-6.66179731e-02 5.85377872e-01 5.28765261e-01 -7.03975439e-01
-3.10236514e-01 3.39627475e-01 3.08914989e-01 8.17432344e-01
-2.81866401e-01 -3.71547729e-01 -4.24719781e-01 3.50293756e-01
-1.27508426e+00 -8.15370262e-01 -9.16295171e-01 2.29558682e+00
8.14164102e-01 -5.80394983e-01 -3.76603633e-01 2.00990513e-01
7.21732855e-01 9.43931043e-02 -5.01854181e-01 -1.78667903e-01
-7.03406751e-01 4.86824140e-02 5.86195827e-01 6.54561996e-01
-4.07307237e-01 2.43789032e-01 6.96884727e+00 3.48002702e-01
-1.13697326e+00 -2.90583670e-01 2.46790335e-01 2.61076063e-01
-7.98993886e-01 1.25654429e-01 -4.34068590e-01 5.94919682e-01
1.96503446e-01 2.45782226e-01 8.29619288e-01 7.34922737e-02
2.78072327e-01 -7.08926916e-01 -5.44416070e-01 1.22269964e+00
2.44675398e-01 -5.34851134e-01 8.72963518e-02 -6.48152977e-02
8.57189775e-01 -3.35173011e-01 6.62509874e-02 -6.34050429e-01
2.72643298e-01 -4.71066922e-01 1.98011443e-01 9.01769519e-01
9.57178533e-01 -3.01115543e-01 4.22452539e-01 1.89383194e-01
-8.19243908e-01 1.99724417e-02 -4.77372497e-01 -4.35328297e-02
-1.12308962e-02 1.09643900e+00 -4.20834839e-01 5.85722268e-01
6.18018329e-01 4.29978460e-01 -3.21045578e-01 7.76168227e-01
-6.48917675e-01 7.47719407e-01 -3.32415491e-01 2.76271552e-01
-3.15101087e-01 -1.09513795e+00 6.95692599e-01 7.93697596e-01
3.89229864e-01 4.93572235e-01 -2.01046750e-01 1.14729071e+00
2.46281639e-01 -1.51478544e-01 -3.19907993e-01 6.91548139e-02
4.99397097e-03 1.39259446e+00 -6.31139517e-01 -6.28406852e-02
-5.89352727e-01 1.15216255e+00 -5.39903760e-01 1.31315029e+00
-5.47893703e-01 -5.80545843e-01 8.79259229e-01 -2.88782418e-02
-1.57819182e-01 -5.26570529e-02 -3.58490378e-01 -1.45403600e+00
2.03562543e-01 -7.57367194e-01 -1.51782766e-01 -1.59261787e+00
-1.31346691e+00 2.65186369e-01 -4.02018368e-01 -1.31584847e+00
3.99536878e-01 -7.26825476e-01 -6.89548612e-01 1.30122328e+00
-2.05311489e+00 -7.02492118e-01 -7.72506654e-01 1.09703362e+00
2.48340085e-01 6.12390526e-02 7.72696614e-01 7.38227507e-03
-3.44745159e-01 -2.82772537e-02 5.06428182e-01 -1.02323644e-01
8.80132854e-01 -1.17440474e+00 -2.94206589e-01 1.04453421e+00
7.71539435e-02 4.90482002e-01 9.03000832e-01 -3.90988052e-01
-1.97436976e+00 -5.51807642e-01 1.82962358e-01 -1.40705526e-01
4.85237807e-01 -2.60338873e-01 -6.43640518e-01 1.65364653e-01
3.46391708e-01 2.86790375e-02 8.26943099e-01 -3.67087185e-01
-4.65708822e-01 -4.84164327e-01 -1.11708367e+00 4.84634250e-01
7.26268351e-01 -6.42631054e-01 -2.70858109e-01 5.67105055e-01
4.19705540e-01 -1.04474433e-01 -2.88468122e-01 1.91742808e-01
5.08114278e-01 -1.30267048e+00 9.35226440e-01 -1.71218231e-01
1.37626246e-01 -8.73764932e-01 -3.87202382e-01 -1.36142313e+00
-4.27833408e-01 -7.92091429e-01 3.42367560e-01 1.25294328e+00
6.75620139e-02 -1.04933774e+00 9.14303213e-02 5.79545856e-01
1.20484151e-01 2.35617727e-01 -2.85481513e-01 -3.98124427e-01
-7.77096152e-01 -2.43713602e-01 5.89198828e-01 1.08241844e+00
-2.54241198e-01 1.39633328e-01 -4.90316004e-01 7.75494516e-01
1.12097573e+00 7.18348801e-01 8.00382733e-01 -1.22258830e+00
-2.30114326e-01 -2.25614458e-02 1.43154785e-01 -8.38908553e-01
-6.02225587e-02 -4.75717664e-01 -1.25026731e-02 -1.51446688e+00
5.87516688e-02 -2.67398357e-01 -2.42078230e-01 1.84933916e-01
-2.67766148e-01 3.43778789e-01 1.51797742e-01 4.77447748e-01
-9.62885190e-03 3.24831218e-01 1.29968548e+00 -1.02622397e-01
-3.42589021e-01 1.96674988e-02 -7.00498819e-01 8.71470392e-01
4.52172607e-01 -1.94775797e-02 -4.26713288e-01 -5.98698318e-01
6.01313472e-01 -1.20517105e-01 4.88782227e-01 -8.47979307e-01
1.67032152e-01 -4.09723461e-01 5.37978709e-01 -4.21135992e-01
3.25352728e-01 -1.23083508e+00 3.73133570e-01 4.26903255e-02
2.24238172e-01 -5.94416738e-01 1.26032352e-01 5.70311964e-01
-9.59212258e-02 -6.88295290e-02 6.48942411e-01 -2.60687679e-01
-3.55946600e-01 -2.71296233e-01 -4.15704757e-01 -4.84398156e-02
6.42911732e-01 -7.63451010e-02 -7.04348564e-01 -5.45871496e-01
-1.63574561e-01 -2.76530087e-01 6.37596190e-01 -9.29684937e-02
4.11558032e-01 -1.20044160e+00 -3.38742763e-01 7.84566343e-01
-1.59188122e-01 -4.28403109e-01 4.18899447e-01 9.77512479e-01
-3.97208214e-01 -7.86938332e-03 -1.41018452e-02 -6.33548558e-01
-1.32071829e+00 4.89199191e-01 4.44381177e-01 4.89616305e-01
-4.15254503e-01 4.85544026e-01 7.36132681e-01 1.05363585e-01
-1.45426482e-01 -1.21722497e-01 -2.45287836e-01 -5.14001437e-02
8.15489531e-01 6.28509820e-01 6.47502020e-02 -6.10811114e-01
-2.35169291e-01 1.24494803e+00 6.26278818e-01 1.50717705e-01
1.21185112e+00 -6.14692926e-01 -7.54293680e-01 5.50969422e-01
9.87426102e-01 7.97155380e-01 -9.81835902e-01 -1.79789022e-01
-7.37521768e-01 -8.74207020e-01 1.73491806e-01 -8.72220874e-01
-1.23675179e+00 9.08278823e-01 4.61884975e-01 5.77998757e-01
1.92094016e+00 -4.15864319e-01 5.40760756e-01 2.81736434e-01
3.61520976e-01 -1.05189776e+00 -3.28300387e-01 -8.53495523e-02
6.85651600e-01 -7.20457911e-01 1.24783618e-02 -7.57473946e-01
-5.53945363e-01 1.31375217e+00 2.68589735e-01 2.54802883e-01
5.19562840e-01 6.72777712e-01 5.29944420e-01 -1.05805457e-01
-3.92240763e-01 -2.87191927e-01 3.14383596e-01 8.27563167e-01
1.78744927e-01 4.29803208e-02 7.29867145e-02 3.80071521e-01
1.50820017e-01 -2.10609809e-01 7.95478046e-01 4.42072153e-01
-3.11626583e-01 -5.92717171e-01 -9.77752209e-01 7.76383653e-02
-2.00884700e-01 -6.22945372e-03 -4.74299669e-01 6.53726235e-02
1.70624301e-01 1.40215790e+00 -1.59807906e-01 8.93900543e-02
2.88462698e-01 1.19188353e-01 3.56276035e-01 7.67773204e-03
1.77000090e-01 5.69733024e-01 -3.15822184e-01 -6.96573257e-01
-8.71307194e-01 -3.23108077e-01 -9.08562541e-01 1.04250930e-01
-5.78244507e-01 1.97010934e-01 8.30856025e-01 6.75218165e-01
7.65520707e-02 2.92991966e-01 1.07009995e+00 -5.33453822e-01
1.77504137e-01 -5.85322142e-01 -1.16896117e+00 5.47788203e-01
7.83247292e-01 -5.61122000e-01 -1.26700234e+00 1.90540761e-01] | [10.380537986755371, -2.717160224914551] |
ecf9a800-8f67-486a-9a2e-78c8e0dbcd3b | video-specific-query-key-attention-modeling | 2305.04186 | null | https://arxiv.org/abs/2305.04186v1 | https://arxiv.org/pdf/2305.04186v1.pdf | Video-Specific Query-Key Attention Modeling for Weakly-Supervised Temporal Action Localization | Weakly-supervised temporal action localization aims to identify and localize the action instances in the untrimmed videos with only video-level action labels. When humans watch videos, we can adapt our abstract-level knowledge about actions in different video scenarios and detect whether some actions are occurring. In this paper, we mimic how humans do and bring a new perspective for locating and identifying multiple actions in a video. We propose a network named VQK-Net with a video-specific query-key attention modeling that learns a unique query for each action category of each input video. The learned queries not only contain the actions' knowledge features at the abstract level but also have the ability to fit this knowledge into the target video scenario, and they will be used to detect the presence of the corresponding action along the temporal dimension. To better learn these action category queries, we exploit not only the features of the current input video but also the correlation between different videos through a novel video-specific action category query learner worked with a query similarity loss. Finally, we conduct extensive experiments on three commonly used datasets (THUMOS14, ActivityNet1.2, and ActivityNet1.3) and achieve state-of-the-art performance. | ['Aggelos K. Katsaggelos', 'Xijun Wang'] | 2023-05-07 | null | null | null | null | ['weakly-supervised-temporal-action', 'action-localization', 'action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.94399625e-01 -3.61912668e-01 -7.40738153e-01 -2.16137335e-01
-6.48124278e-01 -4.03288811e-01 3.25257033e-01 -2.25122958e-01
-5.37638843e-01 3.81106257e-01 4.36741352e-01 3.54764134e-01
-5.85066453e-02 -1.98635131e-01 -8.26432884e-01 -6.94557726e-01
-4.51429725e-01 -5.22601791e-02 6.45286500e-01 3.61565679e-01
1.76218003e-01 2.78654844e-01 -1.40316570e+00 7.13962078e-01
2.88879037e-01 1.23060048e+00 3.57463419e-01 5.35556614e-01
2.18616337e-01 1.62079108e+00 -4.81487989e-01 2.43619867e-02
1.36475876e-01 -6.28462911e-01 -8.63002121e-01 4.33723360e-01
5.89807510e-01 -6.78127468e-01 -8.58914673e-01 8.98387134e-01
6.84628114e-02 5.56279719e-01 3.34533572e-01 -1.58190310e+00
-4.73392487e-01 5.54249704e-01 -3.67364377e-01 7.18247056e-01
7.17641592e-01 5.80264568e-01 8.57737184e-01 -8.40992093e-01
8.52705777e-01 1.23503983e+00 1.37181640e-01 5.59254766e-01
-5.90148449e-01 -5.51881552e-01 5.87846458e-01 9.10840154e-01
-1.41613567e+00 -3.32165629e-01 8.56790841e-01 -4.66688514e-01
8.29606950e-01 1.55511424e-02 7.37522006e-01 1.23652649e+00
1.03527017e-01 1.31616783e+00 4.82317865e-01 1.39688492e-01
9.92204845e-02 -1.59125358e-01 -2.52046436e-01 8.37242782e-01
-4.88833755e-01 -9.79524702e-02 -7.92338729e-01 1.55482888e-01
8.06877792e-01 5.63331664e-01 -4.46782887e-01 -5.65917611e-01
-1.75502253e+00 5.21584749e-01 5.33297777e-01 4.96950775e-01
-6.20532453e-01 4.59332764e-01 5.81123829e-01 3.17577958e-01
1.79190397e-01 2.72880733e-01 -6.93082750e-01 -4.37787503e-01
-5.74854314e-01 -1.14326307e-03 5.21766722e-01 9.68571246e-01
8.62054288e-01 -1.45335183e-01 -6.62492454e-01 3.07719529e-01
-5.52324206e-02 3.43623221e-01 6.99277937e-01 -1.38879836e+00
5.50944090e-01 6.98609650e-01 1.51240215e-01 -1.08434451e+00
-4.55266424e-02 -1.25439242e-01 -3.80980641e-01 -3.38165998e-01
1.02888823e-01 7.18244240e-02 -7.77009666e-01 1.69222701e+00
3.33934307e-01 8.25490057e-01 1.40100777e-01 9.74115372e-01
6.77262604e-01 6.35391116e-01 2.78410643e-01 -3.24971080e-01
1.16622436e+00 -1.35244441e+00 -6.66070342e-01 -2.07040429e-01
8.54635835e-01 -3.20320576e-01 1.04962409e+00 2.01051727e-01
-7.97013342e-01 -8.94626915e-01 -6.31017029e-01 6.82037696e-02
-2.80534327e-01 3.65409493e-01 3.57636452e-01 -2.20937580e-01
-6.69891059e-01 5.06884694e-01 -8.48429382e-01 -4.81900275e-01
5.90311229e-01 7.27626635e-03 -5.49946487e-01 -3.09103280e-01
-1.19896472e+00 4.69967961e-01 6.23144329e-01 -1.57634273e-01
-1.53227890e+00 -4.74181175e-01 -1.00498247e+00 5.94690032e-02
1.14401770e+00 -3.30207616e-01 1.13547850e+00 -1.47784126e+00
-1.02251828e+00 4.95527208e-01 -1.84021100e-01 -5.26834607e-01
3.93675119e-01 -3.89735430e-01 -5.16499698e-01 7.76302755e-01
4.01399136e-01 7.97513127e-01 9.76251125e-01 -7.14161336e-01
-1.00359726e+00 -1.45753082e-02 5.46791732e-01 3.50478023e-01
-3.28770578e-01 1.04333535e-01 -7.83707500e-01 -6.25626683e-01
-7.03585520e-02 -8.10410261e-01 2.72157956e-02 3.27712983e-01
-5.78507185e-02 -4.92377251e-01 1.00742042e+00 -5.72686791e-01
1.18343759e+00 -2.56183052e+00 3.11382383e-01 -2.28062361e-01
7.98483342e-02 1.15567431e-01 -3.50059569e-01 4.41597104e-01
3.55779454e-02 -1.94648117e-01 1.17685802e-01 3.02918591e-02
-2.96599865e-01 3.04918498e-01 -2.07161009e-01 4.52751964e-01
8.89704898e-02 9.76198256e-01 -1.34162605e+00 -6.35575473e-01
2.84870625e-01 1.68059155e-01 -5.28353691e-01 3.99291366e-01
-4.07195479e-01 5.77979684e-01 -7.20447719e-01 7.06871867e-01
-6.52062818e-02 -2.39842474e-01 -1.30903542e-01 -4.80201036e-01
1.14888526e-01 -1.22060195e-01 -1.14548051e+00 1.94969285e+00
-2.61500359e-01 5.63792646e-01 -3.06878507e-01 -1.12276697e+00
3.39023679e-01 3.67686242e-01 9.21246052e-01 -6.02346420e-01
-2.24853694e-01 -2.05294728e-01 1.13154603e-02 -1.19808817e+00
5.33605218e-02 2.99803495e-01 -7.16156065e-02 3.22478205e-01
5.29915273e-01 6.28284097e-01 5.14119685e-01 3.79290700e-01
1.27015400e+00 4.16898489e-01 2.78057128e-01 1.99388489e-01
7.58962154e-01 -2.08617479e-01 6.57883883e-01 7.68897414e-01
-5.76307356e-01 3.08075577e-01 5.13009608e-01 -6.54652536e-01
-3.94049287e-01 -7.24274635e-01 5.41979790e-01 1.64054155e+00
3.29062223e-01 -5.07115364e-01 -4.11437690e-01 -1.15292978e+00
-1.12459511e-01 3.72853041e-01 -8.47532451e-01 -5.79555333e-01
-5.14086366e-01 1.16917528e-01 2.28758439e-01 8.51871550e-01
8.25034559e-01 -1.50452375e+00 -7.64350593e-01 1.28795013e-01
-4.35814768e-01 -1.31560731e+00 -1.05040729e+00 -7.39392191e-02
-5.75674415e-01 -1.55055368e+00 -6.20223284e-01 -7.88255155e-01
5.17123818e-01 4.21358138e-01 8.06421280e-01 -1.96295744e-03
-9.51429233e-02 8.86352062e-01 -6.50092244e-01 1.87036648e-01
-1.81707889e-02 -2.40465298e-01 1.85384303e-01 7.26748824e-01
5.93611360e-01 -2.22063601e-01 -7.41095304e-01 4.87099886e-01
-9.00778770e-01 -1.62560597e-01 6.64985538e-01 5.08666277e-01
5.72392881e-01 1.27872184e-01 3.10281307e-01 -4.10223216e-01
4.77964208e-02 -5.98518372e-01 -7.39547238e-02 3.88593644e-01
9.80435684e-02 -2.27663830e-01 4.25707191e-01 -9.07142699e-01
-8.05971324e-01 3.68755013e-01 3.80030870e-01 -1.13720131e+00
-3.10550034e-01 5.43806255e-01 -3.84596527e-01 2.06329897e-01
3.37085634e-01 5.98495722e-01 -2.84765154e-01 -2.73087233e-01
3.59914392e-01 3.22051018e-01 5.99198282e-01 -2.46533349e-01
5.37654042e-01 4.51999456e-01 -2.40988195e-01 -6.04815006e-01
-1.09776616e+00 -8.68175387e-01 -7.93414354e-01 -5.03631890e-01
1.17593265e+00 -1.10206199e+00 -6.90189242e-01 3.81696790e-01
-9.30089116e-01 -4.05735672e-01 -3.97041440e-01 7.74606049e-01
-7.89192438e-01 4.72601593e-01 -4.90850300e-01 -4.41980243e-01
2.04152584e-01 -1.06837153e+00 1.06129861e+00 1.12258308e-01
-9.66391712e-02 -7.75522709e-01 -1.05993621e-01 2.16929570e-01
-5.51562272e-02 1.15608785e-03 5.15795946e-01 -8.00701380e-01
-8.06768179e-01 -1.51598528e-01 -4.30233330e-02 4.82381493e-01
3.47835422e-01 -2.22604960e-01 -4.71068144e-01 -3.57906163e-01
4.83175591e-02 -4.08104062e-01 9.83356714e-01 2.96793014e-01
1.43398404e+00 -5.67072749e-01 -4.82379943e-01 5.83867788e-01
1.02704120e+00 3.34660739e-01 4.20195609e-01 2.15943512e-02
9.16743577e-01 3.97292018e-01 1.02971005e+00 3.32261384e-01
1.99866623e-01 8.18562925e-01 4.51793164e-01 1.83596328e-01
-3.77403833e-02 -5.53161204e-01 8.27397168e-01 4.84055907e-01
-8.40728059e-02 -1.79718986e-01 -4.86080796e-01 5.76792002e-01
-2.16311479e+00 -1.43837643e+00 4.67894167e-01 1.91240788e+00
4.66880739e-01 1.24751940e-01 2.50469118e-01 -2.72605836e-01
6.23547554e-01 4.69303280e-01 -9.16286170e-01 3.53257328e-01
1.00015111e-01 -3.58695179e-01 3.21166813e-01 2.67978292e-02
-1.53459334e+00 9.67483044e-01 5.40275717e+00 6.95591509e-01
-9.29129422e-01 1.93727478e-01 3.00022393e-01 -3.90129626e-01
3.30267370e-01 -9.31940153e-02 -3.53163511e-01 7.35744178e-01
6.50737584e-01 -9.18585435e-02 2.81435281e-01 1.03268850e+00
2.84768760e-01 -1.94343671e-01 -1.47826898e+00 1.06821561e+00
4.50263739e-01 -1.07312882e+00 2.81968206e-01 -2.33134225e-01
5.07789552e-01 -8.59457701e-02 -6.36015236e-02 6.88151062e-01
-2.37842306e-01 -6.83726072e-01 5.04451156e-01 8.41666222e-01
5.89045167e-01 -4.45997596e-01 5.82376361e-01 3.98891121e-01
-1.48257351e+00 -5.57999551e-01 -2.59146720e-01 -1.18133672e-01
9.37415883e-02 -2.15835273e-01 -5.13882637e-01 3.69898468e-01
9.22118008e-01 1.66299069e+00 -5.95062733e-01 1.02507079e+00
-3.54777247e-01 5.06222785e-01 7.06627667e-02 1.33203983e-01
4.68547165e-01 9.49204490e-02 4.43892241e-01 8.98159385e-01
1.66194856e-01 3.57721925e-01 7.78927803e-01 4.17845875e-01
-1.29960969e-01 -3.14937718e-02 -6.02954507e-01 -2.04918832e-01
1.72857553e-01 9.25381243e-01 -5.49663961e-01 -6.09288514e-01
-7.41387904e-01 1.29787874e+00 1.65998340e-01 6.47810042e-01
-8.73648226e-01 -1.13408588e-01 7.73258150e-01 1.28792748e-02
5.95969558e-01 8.32841322e-02 9.73614097e-01 -1.42432475e+00
1.77155465e-01 -8.43826115e-01 7.66072989e-01 -1.08828735e+00
-1.08186746e+00 1.94666073e-01 2.64637738e-01 -1.56872547e+00
-3.47093344e-01 -4.42555308e-01 -5.84968328e-01 1.89512074e-01
-1.05800152e+00 -1.07553792e+00 -4.29878026e-01 9.42694485e-01
1.09028268e+00 -2.74617195e-01 3.50362569e-01 2.56747127e-01
-3.30703914e-01 3.57899874e-01 -1.99149922e-01 5.80442250e-01
7.61376500e-01 -8.82774651e-01 -1.78202614e-02 8.14762354e-01
5.12386978e-01 2.79907972e-01 3.38705868e-01 -6.47666872e-01
-1.41009176e+00 -1.29086280e+00 4.21774805e-01 -5.80128610e-01
7.00676024e-01 -1.26766533e-01 -9.42030370e-01 1.05660701e+00
-3.07717714e-02 3.70504588e-01 5.20104289e-01 -3.60873669e-01
-2.93854654e-01 -2.11996719e-01 -8.06179643e-01 5.16695857e-01
1.49290848e+00 -8.49901915e-01 -8.04636121e-01 5.36657572e-01
9.91114974e-01 -1.97005436e-01 -6.19750559e-01 3.52086365e-01
5.57238460e-01 -7.82861114e-01 9.92695928e-01 -1.06789339e+00
3.53684098e-01 -4.28577334e-01 -2.10715935e-01 -1.03247702e+00
-4.29885864e-01 -3.52998048e-01 -4.44317132e-01 9.09782648e-01
-1.57835364e-01 -3.68788727e-02 6.89858317e-01 3.22558820e-01
-7.34467283e-02 -7.86882222e-01 -9.81587529e-01 -8.10818076e-01
-6.04086518e-01 -3.26544225e-01 3.22324723e-01 8.52459967e-01
4.73285243e-02 1.61310524e-01 -6.99064612e-01 1.52475625e-01
3.17558944e-01 1.58987164e-01 7.31443226e-01 -7.11143434e-01
-3.28875333e-01 -2.18371153e-01 -9.36959565e-01 -1.36652625e+00
4.10282910e-01 -5.78189135e-01 1.01321042e-01 -1.19304001e+00
5.24727046e-01 4.02249664e-01 -8.24146986e-01 6.70022905e-01
-1.84141323e-01 1.51530668e-01 3.70895028e-01 2.58226156e-01
-1.42555261e+00 5.89796722e-01 1.21628296e+00 -4.07193899e-01
-2.03604326e-01 5.72710671e-02 -2.29430720e-01 7.50641704e-01
4.33802098e-01 -4.97677684e-01 -6.73285365e-01 -3.02035183e-01
-2.11370140e-01 2.43440300e-01 6.71029687e-01 -1.31111574e+00
3.74437839e-01 -4.72031623e-01 4.33290631e-01 -5.74416876e-01
3.66665304e-01 -9.80335116e-01 -1.31570578e-01 4.80180770e-01
-7.15886474e-01 -1.22744374e-01 -1.69579238e-01 1.03496110e+00
-4.43379134e-01 -5.35733588e-02 4.90693241e-01 -3.59275758e-01
-1.52002323e+00 7.98399389e-01 -3.68398905e-01 1.08280033e-01
1.46400774e+00 -1.72650471e-01 -7.99671113e-02 -6.05191410e-01
-9.16314602e-01 5.26964903e-01 3.45130622e-01 7.83097446e-01
8.33526969e-01 -1.38601768e+00 -4.48993146e-01 1.26830265e-01
5.92281222e-01 -4.32273507e-01 6.05344176e-01 8.81158352e-01
8.17676261e-03 2.98687845e-01 -2.76598126e-01 -7.12095141e-01
-1.21311092e+00 1.15724909e+00 4.40773070e-01 1.42454594e-01
-6.07064366e-01 8.04499090e-01 6.16251230e-01 1.47621095e-01
5.99995315e-01 -4.17461991e-01 -3.97130996e-01 3.28939557e-02
7.46677041e-01 2.19495371e-01 -5.88370025e-01 -9.55197752e-01
-5.47710896e-01 4.99035537e-01 -2.48218570e-02 1.73798695e-01
9.17733550e-01 -1.48406178e-01 1.45739779e-01 6.39707744e-01
1.51154673e+00 -4.60761398e-01 -1.76180756e+00 -6.07111871e-01
-1.32208258e-01 -6.58500671e-01 -2.77204901e-01 -6.93347752e-01
-1.30990326e+00 7.58237243e-01 6.35483742e-01 -3.15380603e-01
1.17668426e+00 4.28345233e-01 5.97674489e-01 5.95939100e-01
3.76179010e-01 -1.05319536e+00 8.08560014e-01 3.39817911e-01
7.82495737e-01 -1.25256383e+00 -2.45784633e-02 1.82103142e-02
-9.92976785e-01 8.57427418e-01 9.34008121e-01 -7.77384406e-03
6.41636431e-01 -3.47252846e-01 -3.23176607e-02 -3.27820539e-01
-8.37096155e-01 -4.10022289e-01 3.66519868e-01 6.51527941e-01
-1.09704286e-01 -1.82681724e-01 4.71439958e-02 4.81871158e-01
6.86878741e-01 1.44371867e-01 3.81216794e-01 1.05363977e+00
-5.44517159e-01 -5.54803491e-01 1.32883176e-01 4.98639733e-01
-2.75526553e-01 2.22022027e-01 -3.94063979e-01 6.56312048e-01
2.76821226e-01 7.60174751e-01 1.34918079e-01 -5.63223839e-01
3.67620260e-01 1.84910446e-01 2.81041712e-01 -7.70539522e-01
-2.56740659e-01 -3.20172124e-03 -2.84515887e-01 -1.19794965e+00
-9.80091631e-01 -7.92344987e-01 -1.16135681e+00 2.39487126e-01
1.60209194e-01 1.97711915e-01 -1.33043990e-01 1.05920291e+00
4.77917999e-01 4.77191657e-01 6.91320419e-01 -6.73458040e-01
-3.94118816e-01 -9.84735906e-01 -5.93979597e-01 7.81241655e-01
4.65162843e-01 -8.58130336e-01 -3.59102577e-01 4.90620673e-01] | [8.522329330444336, 0.5838790535926819] |
167de426-1b10-4afb-8e82-e98b92369a27 | transferring-convnet-features-from-passive-to | 2204.10497 | null | https://arxiv.org/abs/2204.10497v2 | https://arxiv.org/pdf/2204.10497v2.pdf | Active Domain-Invariant Self-Localization Using Ego-Centric and World-Centric Maps | The training of a next-best-view (NBV) planner for visual place recognition (VPR) is a fundamentally important task in autonomous robot navigation, for which a typical approach is the use of visual experiences that are collected in the target domain as training data. However, the collection of a wide variety of visual experiences in everyday navigation is costly and prohibitive for real-time robotic applications. We address this issue by employing a novel {\it domain-invariant} NBV planner. A standard VPR subsystem based on a convolutional neural network (CNN) is assumed to be available, and its domain-invariant state recognition ability is proposed to be transferred to train the domain-invariant NBV planner. Specifically, we divide the visual cues that are available from the CNN model into two types: the output layer cue (OLC) and intermediate layer cue (ILC). The OLC is available at the output layer of the CNN model and aims to estimate the state of the robot (e.g., the robot viewpoint) with respect to the world-centric view coordinate system. The ILC is available within the middle layers of the CNN model as a high-level description of the visual content (e.g., a saliency image) with respect to the ego-centric view. In our framework, the ILC and OLC are mapped to a state vector and subsequently used to train a multiview NBV planner via deep reinforcement learning. Experiments using the public NCLT dataset validate the effectiveness of the proposed method. | ['Mitsuki Yoshida', 'Ryogo Yamamoto', 'Kanji Tanaka', 'Kanya Kurauchi'] | 2022-04-22 | null | null | null | null | ['visual-place-recognition'] | ['computer-vision'] | [-4.11265269e-02 1.62529483e-01 -2.93702245e-01 -2.02773139e-01
-1.85345784e-01 -4.52492505e-01 6.75436795e-01 3.36870588e-02
-4.66452688e-01 4.16747063e-01 1.20234020e-01 -1.05563693e-01
1.23452283e-01 -8.32576692e-01 -9.74398196e-01 -6.05787992e-01
3.23113054e-01 2.50186920e-01 3.12047362e-01 -3.23080629e-01
4.03603822e-01 5.82343757e-01 -1.66035032e+00 -1.05382740e-01
7.26487398e-01 1.14586973e+00 1.02171683e+00 2.85546750e-01
2.03680664e-01 8.13969076e-01 -1.21238388e-01 4.27973777e-01
1.79961324e-01 -1.25721961e-01 -8.05820107e-01 7.44486898e-02
-8.46164450e-02 -3.92208457e-01 -3.74119967e-01 1.18085909e+00
1.49573296e-01 4.97406095e-01 5.45198798e-01 -1.36354208e+00
-3.88473809e-01 3.06572709e-02 -8.41862410e-02 -9.44935996e-03
3.33414912e-01 3.62155855e-01 9.13052797e-01 -9.68702793e-01
1.24876893e+00 1.15533328e+00 -9.74299312e-02 4.32066500e-01
-9.64951694e-01 -1.96140856e-01 5.07298648e-01 3.47692996e-01
-1.25068402e+00 -2.98574239e-01 1.01154494e+00 -5.02851963e-01
8.52893472e-01 -2.67998487e-01 9.02188361e-01 1.07616413e+00
4.02058661e-01 9.49200988e-01 8.62912714e-01 -2.06932649e-02
7.55018830e-01 1.34266868e-01 -1.89806119e-01 7.86942840e-01
-1.71124365e-03 1.65791005e-01 -4.79130030e-01 2.41515771e-01
1.02047873e+00 2.51391441e-01 -3.59293163e-01 -1.23604345e+00
-1.20906651e+00 7.31555223e-01 1.15534067e+00 6.47421852e-02
-7.52155244e-01 7.15087578e-02 2.39806786e-01 2.91399215e-03
-1.12530351e-01 4.56553102e-01 -2.64617682e-01 1.81598395e-01
-2.87119627e-01 6.12699874e-02 4.13267672e-01 1.12628424e+00
1.12388027e+00 1.77632764e-01 -1.27693549e-01 5.91762722e-01
5.00165164e-01 5.59222221e-01 4.53568637e-01 -1.13223684e+00
6.55214429e-01 8.08801889e-01 4.57669884e-01 -1.06542659e+00
-4.63975817e-01 -2.80471891e-01 -6.00265920e-01 5.88361561e-01
1.42110020e-01 -1.22465510e-02 -9.47356880e-01 1.78708208e+00
3.71776283e-01 -7.86108896e-03 3.89995575e-01 1.32739389e+00
6.21457160e-01 7.45597363e-01 -3.75487581e-02 1.91571862e-01
1.09970975e+00 -1.04933691e+00 -4.46900576e-01 -7.61422634e-01
4.93114978e-01 -1.99065551e-01 9.24913287e-01 -5.89789972e-02
-7.37546086e-01 -5.56353092e-01 -1.24671268e+00 -2.17265800e-01
-5.19509077e-01 3.80230367e-01 1.79561511e-01 -3.20042789e-01
-1.05214798e+00 2.81646311e-01 -8.53414237e-01 -6.33361578e-01
1.20613076e-01 1.36915192e-01 -7.83036768e-01 -3.05355996e-01
-1.06860971e+00 1.15177846e+00 7.17705309e-01 2.73110002e-01
-1.41842389e+00 -9.52966511e-02 -1.42596161e+00 5.91549017e-02
4.94841397e-01 -6.14740968e-01 1.07573020e+00 -9.74528253e-01
-1.54970658e+00 5.01718938e-01 -2.30024293e-01 -2.68614441e-01
3.56677115e-01 -9.33689438e-03 2.27618665e-02 2.59323865e-01
3.58929068e-01 8.65110338e-01 8.40540588e-01 -1.62135231e+00
-8.06683958e-01 -2.98209280e-01 4.06575590e-01 7.40172923e-01
1.28303483e-01 -5.66054165e-01 -7.20404148e-01 -1.02673963e-01
5.97340941e-01 -1.07203674e+00 -3.25657934e-01 -1.28589501e-03
-4.03139681e-01 -4.50068451e-02 7.95809209e-01 -6.24166965e-01
3.91773492e-01 -2.07593274e+00 6.21931791e-01 1.52031958e-01
3.09553668e-02 1.30236428e-02 -9.13003236e-02 3.74206096e-01
-2.86962986e-02 -4.09776986e-01 7.95629050e-04 -2.63726413e-01
-1.71447873e-01 1.51908487e-01 -3.25830251e-01 5.63239634e-01
1.01796992e-01 7.73849845e-01 -1.32288396e+00 -1.50630459e-01
6.38402343e-01 3.68377715e-01 -5.39510250e-01 4.01469648e-01
-4.32672471e-01 8.10075760e-01 -6.90188527e-01 3.97228479e-01
2.62951195e-01 -9.97292027e-02 1.17839217e-01 -2.63845533e-01
-5.79342604e-01 3.91902745e-01 -1.03047180e+00 1.92143083e+00
-6.80455804e-01 5.73154330e-01 9.69179794e-02 -8.24570417e-01
1.01755059e+00 1.04979277e-01 1.95278153e-02 -9.05503154e-01
2.67545998e-01 1.38310656e-01 -2.62403041e-01 -3.37188244e-01
5.39998114e-01 2.36061931e-01 -1.80800825e-01 5.34760915e-02
3.23266327e-01 2.96472735e-03 -1.55864984e-01 1.26423866e-01
8.81645441e-01 3.92375469e-01 5.79217851e-01 -9.96891931e-02
6.29952550e-01 2.81986207e-01 6.85297966e-01 5.19969583e-01
-2.89017737e-01 4.38671380e-01 4.87806410e-01 -3.90656114e-01
-1.13208675e+00 -9.27724421e-01 3.23682070e-01 6.83496833e-01
7.21301794e-01 2.13090200e-02 -5.04115224e-01 -5.38282394e-01
-2.38901272e-01 8.90585423e-01 -6.82223737e-01 -3.49092513e-01
-4.18612510e-01 1.76424503e-01 -2.44106814e-01 5.75465441e-01
7.87062049e-01 -1.46910143e+00 -1.27837336e+00 9.45468768e-02
-2.57653415e-01 -1.12223709e+00 -1.57730669e-01 3.20847839e-01
-7.17054784e-01 -1.04467952e+00 -5.00511408e-01 -9.45489168e-01
9.91373479e-01 4.64045018e-01 5.56454182e-01 -2.56297529e-01
2.33775914e-01 4.76932377e-01 -3.57789636e-01 -8.06986690e-02
-1.95788786e-01 -3.16530354e-02 7.68303052e-02 1.59330875e-01
-7.52055421e-02 -3.13226610e-01 -7.56632745e-01 1.87793911e-01
-5.82384229e-01 4.51801330e-01 6.86712384e-01 8.31651390e-01
6.99988604e-01 -3.09528381e-01 3.77924383e-01 -4.44503933e-01
1.52161062e-01 -3.96054655e-01 -9.83107686e-01 4.68718484e-02
-2.30651274e-01 1.88138694e-01 7.03268290e-01 -1.42851815e-01
-9.99081910e-01 5.16648829e-01 -2.78548757e-03 -7.78928518e-01
-3.25768083e-01 7.76017606e-01 -4.86777782e-01 2.20637262e-01
3.50910276e-01 4.61206734e-01 -1.04136944e-01 -9.41597074e-02
4.32021648e-01 3.89425606e-01 6.74869061e-01 -1.62189648e-01
7.24912584e-01 5.61752379e-01 -8.81857723e-02 -6.69549823e-01
-5.32330990e-01 -5.99499285e-01 -8.45556319e-01 -4.74649936e-01
1.10081315e+00 -1.07753527e+00 -6.45259559e-01 3.36971045e-01
-1.18371618e+00 -5.48918247e-01 -2.26909164e-02 5.74518263e-01
-9.86085057e-01 -6.92006722e-02 -1.41177312e-01 -6.11159384e-01
-1.88704785e-02 -1.49144518e+00 9.92540538e-01 3.88690889e-01
1.23305067e-01 -8.41819286e-01 -1.15088098e-01 -5.85275032e-02
-3.60671766e-02 1.49679348e-01 9.95816469e-01 -4.88026202e-01
-7.82251179e-01 -1.55404583e-01 -2.53914446e-01 1.09539419e-01
6.24367502e-03 -3.55823636e-01 -9.75480855e-01 -3.19765031e-01
9.78050102e-03 -2.43162379e-01 7.11354971e-01 3.31312418e-01
6.74495876e-01 -9.80374962e-02 -6.03244781e-01 5.34507871e-01
1.72283804e+00 4.65426713e-01 5.19442618e-01 7.69322991e-01
7.12619424e-01 6.33704603e-01 1.02236271e+00 2.84483463e-01
7.28045106e-01 8.28113019e-01 1.19318998e+00 -1.38974674e-02
1.33767948e-01 -5.34341037e-01 4.29800600e-01 4.38601315e-01
9.65721086e-02 -1.83725301e-02 -9.53922331e-01 7.26461112e-01
-2.14562178e+00 -7.34865069e-01 4.24849331e-01 2.28731799e+00
2.67399460e-01 8.65918249e-02 -3.85328948e-01 -3.05517465e-01
5.78886926e-01 3.11942875e-01 -9.21208203e-01 -2.96531439e-01
3.16470176e-01 -6.20205045e-01 4.09316689e-01 4.58549082e-01
-1.06984270e+00 1.15270030e+00 4.18040180e+00 2.02853948e-01
-1.40320003e+00 -1.97571531e-01 1.78188905e-02 2.94369251e-01
-6.20856732e-02 7.23183230e-02 -7.44524658e-01 2.02167690e-01
4.67619568e-01 7.30517358e-02 5.48645198e-01 1.35163951e+00
3.06061238e-01 -4.06861752e-01 -1.32198107e+00 9.52805519e-01
5.41713201e-02 -1.24966943e+00 -2.48398278e-02 2.11097412e-02
6.17161155e-01 2.18927279e-01 2.03970924e-01 4.25154179e-01
2.68253326e-01 -6.59557939e-01 9.76445138e-01 4.94108260e-01
5.95544994e-01 -7.63116241e-01 6.61900997e-01 7.17089832e-01
-1.25240290e+00 -4.13506448e-01 -5.82468271e-01 4.81902882e-02
1.90899462e-01 -4.23620529e-02 -1.11244738e+00 6.30951762e-01
5.73553503e-01 9.65808213e-01 -3.83075476e-01 9.89503682e-01
-5.10486543e-01 -1.29090041e-01 4.48814733e-03 -2.76367497e-02
6.14147902e-01 -2.22973153e-01 6.59859478e-01 3.88952911e-01
3.75904858e-01 -4.07598801e-02 3.33033174e-01 8.27719569e-01
6.26877248e-02 -1.39339522e-01 -1.06062150e+00 4.24681932e-01
4.64158624e-01 1.25479627e+00 -7.36750782e-01 -3.79970580e-01
-3.18787098e-01 9.52806950e-01 7.04990685e-01 6.48295820e-01
-5.34645557e-01 -2.21008539e-01 6.89376354e-01 -2.82342374e-01
5.61181903e-01 -2.11617857e-01 -5.38238548e-02 -1.13217616e+00
-1.04600184e-01 -5.52496731e-01 -9.07808691e-02 -1.25808477e+00
-5.93086004e-01 6.34454608e-01 -9.68637690e-02 -1.66798747e+00
-5.45083880e-01 -6.84523225e-01 -5.37549913e-01 1.15785313e+00
-1.73485005e+00 -9.93134439e-01 -6.22428298e-01 6.40407860e-01
6.74071431e-01 -1.83135450e-01 7.61265814e-01 -2.84463614e-01
-4.56891626e-01 -2.92325057e-02 1.98329650e-02 1.95412114e-01
3.53899568e-01 -1.02996671e+00 1.93568528e-01 8.28887284e-01
-2.22934440e-01 6.11912251e-01 7.16171384e-01 -6.61796689e-01
-1.61475658e+00 -1.44827497e+00 4.98194993e-01 -3.09709191e-01
3.88227552e-01 -2.66697258e-01 -6.66193545e-01 9.11576629e-01
-6.93806335e-02 7.47257844e-02 -1.52736604e-01 -3.29577684e-01
-1.88697711e-01 4.12289388e-02 -1.00664592e+00 9.77753937e-01
7.17623353e-01 -7.25459039e-01 -7.34874368e-01 -6.11249469e-02
7.49922752e-01 -6.76657975e-01 -3.96800429e-01 2.25609213e-01
4.80390757e-01 -7.77921319e-01 9.81232941e-01 -3.25368047e-01
4.52417731e-01 -6.39918923e-01 -2.79840320e-01 -1.62890947e+00
-2.30626255e-01 1.84054896e-01 1.46824509e-01 8.06013286e-01
2.02803761e-01 -6.19043410e-01 5.23737967e-01 4.01359975e-01
-3.44293088e-01 -5.67984462e-01 -9.48287725e-01 -5.81817508e-01
-4.92090791e-01 -3.44058633e-01 2.89520472e-01 6.81390703e-01
3.28715779e-02 5.13681412e-01 -7.97137842e-02 7.24578977e-01
3.44454378e-01 3.24004352e-01 7.35541403e-01 -9.83740449e-01
2.61447877e-01 -9.95065272e-02 -5.17446041e-01 -1.22014713e+00
3.39762002e-01 -9.68731284e-01 6.11931920e-01 -2.03752899e+00
-8.15755650e-02 -2.97030002e-01 -2.23693267e-01 5.02703488e-01
2.34572068e-01 -3.40678155e-01 5.08308470e-01 2.37213373e-01
-6.21064425e-01 8.72664690e-01 1.33839452e+00 -2.21369594e-01
-4.84222651e-01 2.55947728e-02 -2.18763351e-01 8.53674531e-01
7.71498442e-01 -1.48911744e-01 -5.74177086e-01 -4.72706467e-01
9.77799520e-02 5.12251973e-01 5.80782652e-01 -1.13646626e+00
5.10183513e-01 -1.71425283e-01 3.65632415e-01 -8.01777124e-01
6.53383493e-01 -1.06104553e+00 -1.48417160e-01 5.33892453e-01
-2.44020134e-01 -2.03730948e-02 -8.11027214e-02 7.56580174e-01
-2.39001751e-01 -3.10276121e-01 5.90366721e-01 -3.60582024e-01
-1.34096694e+00 2.37526506e-01 -3.24205250e-01 -2.99116760e-01
9.76670027e-01 -5.99925071e-02 -3.72255772e-01 -4.49302077e-01
-6.50037348e-01 4.77802396e-01 7.47738600e-01 6.41195774e-01
9.49960113e-01 -1.30768371e+00 -2.67949584e-03 4.31572556e-01
5.24734259e-01 4.05591965e-01 2.49387205e-01 6.95676923e-01
-3.79029959e-01 7.37439454e-01 -5.19258559e-01 -7.46132672e-01
-6.35640621e-01 8.73991430e-01 4.01480585e-01 1.09948426e-01
-7.92254746e-01 3.94312710e-01 7.14831471e-01 -6.14923418e-01
1.56589836e-01 -3.65452826e-01 -7.50382781e-01 3.88892088e-03
3.13577741e-01 8.89073834e-02 -1.42627552e-01 -1.08628047e+00
-3.41163367e-01 4.24462765e-01 2.76263833e-01 -2.96352595e-01
1.27849126e+00 -3.13705534e-01 -2.96256188e-02 6.30135357e-01
1.20362651e+00 -5.64264953e-01 -1.76449144e+00 -2.37905145e-01
-5.22314087e-02 -8.51619616e-02 6.83712885e-02 -4.43559438e-01
-8.05699646e-01 9.27120626e-01 5.22450387e-01 -2.72436559e-01
7.83954263e-01 1.09581016e-02 2.55938113e-01 7.10949481e-01
8.52807939e-01 -1.11672568e+00 1.46797463e-01 7.79260039e-01
1.16631114e+00 -1.23573148e+00 -3.36510032e-01 -6.67489246e-02
-9.49366748e-01 8.95623863e-01 9.69571710e-01 -2.40639836e-01
4.51799393e-01 -5.01955748e-01 1.91628546e-01 -1.74223244e-01
-6.86824918e-01 -4.14381027e-01 2.11140051e-01 7.67029464e-01
-4.03083533e-01 -1.00844815e-01 2.86410630e-01 3.99241000e-01
3.75232995e-02 -2.45985210e-01 5.79909623e-01 1.04906809e+00
-6.43063962e-01 -4.49333012e-01 -7.10450038e-02 3.67299877e-02
3.18292141e-01 2.29495108e-01 4.33864333e-02 7.51559436e-01
-5.51094450e-02 7.86421180e-01 2.43445486e-02 -3.92779917e-01
5.31824470e-01 -2.00983305e-02 6.88070804e-02 -8.76632929e-01
-8.27684924e-02 -9.96766239e-02 -2.26568550e-01 -8.71206045e-01
-2.61484534e-01 -6.73951387e-01 -1.60546505e+00 3.26344132e-01
7.92220011e-02 -1.88109040e-01 9.03473198e-01 1.10874116e+00
2.17896551e-01 5.54308414e-01 6.81499362e-01 -1.37835288e+00
-1.96179792e-01 -5.84842980e-01 -3.70583534e-01 2.36073434e-01
6.48077667e-01 -1.04832733e+00 -1.08796045e-01 -3.07462532e-02] | [7.389827728271484, -1.9726543426513672] |
9176cd23-0650-4ef6-8132-fd04db9567ab | selecting-better-samples-from-pre-trained | 2209.11000 | null | https://arxiv.org/abs/2209.11000v1 | https://arxiv.org/pdf/2209.11000v1.pdf | Selecting Better Samples from Pre-trained LLMs: A Case Study on Question Generation | Large Language Models (LLMs) have in recent years demonstrated impressive prowess in natural language generation. A common practice to improve generation diversity is to sample multiple outputs from the model. However, there lacks a simple and robust way of selecting the best output from these stochastic samples. As a case study framed in the context of question generation, we propose two prompt-based approaches to selecting high-quality questions from a set of LLM-generated candidates. Our method works under the constraints of 1) a black-box (non-modifiable) question generation model and 2) lack of access to human-annotated references -- both of which are realistic limitations for real-world deployment of LLMs. With automatic as well as human evaluations, we empirically demonstrate that our approach can effectively select questions of higher qualities than greedy generation. | ['Pierre-Yves Oudeyer', 'Hélène Sauzéon', 'Pauline Lucas', 'Rania Abdelghani', 'Emery Fine', 'Yen-Hsiang Wang', 'Tong Wang', 'Xingdi Yuan'] | 2022-09-22 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [ 2.58307517e-01 4.82617110e-01 -7.77426139e-02 -1.76893577e-01
-1.72394800e+00 -8.15570533e-01 9.41715896e-01 1.69297650e-01
-3.76851708e-01 9.91793633e-01 4.45379943e-01 -4.37420785e-01
-5.77148199e-02 -7.69546509e-01 -5.09423852e-01 1.77273199e-01
3.59670311e-01 8.57377350e-01 2.36784413e-01 -3.89658570e-01
4.52531159e-01 1.29473999e-01 -1.63989723e+00 4.60777372e-01
1.48375642e+00 3.42193693e-01 3.95677984e-01 8.51752043e-01
-5.62102377e-01 6.90248489e-01 -1.10625875e+00 -5.00884831e-01
8.06982815e-02 -9.36277330e-01 -9.61528361e-01 1.77015141e-01
4.31914747e-01 -2.68355787e-01 3.64682853e-01 5.69011927e-01
5.59393287e-01 2.20108241e-01 5.72601795e-01 -1.11766124e+00
-5.58311343e-01 8.66559029e-01 3.03175133e-02 7.55575299e-02
9.04294014e-01 4.73052800e-01 1.24870682e+00 -8.31187606e-01
8.62378836e-01 1.40738344e+00 2.86681682e-01 6.79238737e-01
-1.39603055e+00 -3.64108980e-01 5.69891371e-02 -4.17929888e-01
-1.17518270e+00 -7.56540358e-01 2.41645068e-01 -2.38823503e-01
8.79495978e-01 5.38239062e-01 4.13838744e-01 8.58798027e-01
-4.67372835e-02 8.55751216e-01 9.59443688e-01 -8.79033506e-01
3.73167545e-01 5.96913040e-01 -8.30766037e-02 4.88258094e-01
3.29117537e-01 -2.25097701e-01 -7.51069903e-01 -8.25766861e-01
4.09437656e-01 -7.87836075e-01 -3.21748883e-01 4.49715778e-02
-1.11888111e+00 9.93193150e-01 -1.85024694e-01 3.49886298e-01
-2.33419761e-01 5.08131552e-03 -1.01959474e-01 2.62868851e-01
2.63981044e-01 1.36931467e+00 -3.80097598e-01 -3.27294827e-01
-1.15420914e+00 8.84767413e-01 1.18453526e+00 1.03776908e+00
5.62593877e-01 -6.84449598e-02 -6.85645998e-01 8.74990463e-01
2.92114943e-01 4.64748889e-01 7.23261654e-01 -9.70110357e-01
6.25524521e-01 5.03396511e-01 7.81845808e-01 -6.32282257e-01
-9.75546986e-03 -2.07680583e-01 1.24246152e-02 7.73581415e-02
7.03084588e-01 -3.71955633e-01 -6.73416972e-01 1.84285462e+00
2.20091522e-01 -5.16953588e-01 -2.14040820e-02 5.05001247e-01
6.27869010e-01 7.68419147e-01 1.47881135e-01 -2.04205140e-01
1.11574447e+00 -6.24933124e-01 -4.05567676e-01 -3.38435203e-01
6.19321108e-01 -8.72338355e-01 1.48888707e+00 3.42581838e-01
-1.48562872e+00 -3.79281968e-01 -7.34817028e-01 1.12075761e-01
-4.23820838e-02 2.36824602e-01 4.88970667e-01 7.86227584e-01
-1.13359141e+00 3.38760287e-01 -3.12710464e-01 -5.21874309e-01
-1.39332473e-01 1.44203275e-01 -2.62850598e-02 -3.35673280e-02
-1.06679296e+00 8.00353825e-01 7.84728006e-02 -3.47238600e-01
-6.70316637e-01 -4.81854826e-01 -6.52059019e-01 -4.92134988e-02
3.97317111e-01 -9.48164165e-01 1.75876105e+00 -1.02981222e+00
-1.39042807e+00 5.72278619e-01 -3.58917087e-01 -2.59184271e-01
5.19962192e-01 -2.95084476e-01 -1.63511157e-01 1.07631154e-01
4.22821671e-01 8.99886608e-01 6.78124249e-01 -1.40417671e+00
-6.33772075e-01 2.20878020e-01 2.71910846e-01 2.89486557e-01
-2.90553957e-01 2.35663936e-01 -9.20703039e-02 -5.11440516e-01
-1.85411245e-01 -7.18576431e-01 -4.80210871e-01 -3.15815568e-01
-3.78022850e-01 -4.86213773e-01 1.90591276e-01 -5.39069235e-01
1.43708622e+00 -1.31304967e+00 -2.41221353e-01 1.06505908e-01
-1.51875392e-01 3.16270143e-01 -4.56948936e-01 9.37729597e-01
3.39488924e-01 6.41027808e-01 8.50759745e-02 -2.12185040e-01
2.31583670e-01 -2.59055376e-01 -6.05651796e-01 -3.60428005e-01
2.72129655e-01 8.38353038e-01 -1.23160219e+00 -6.68387294e-01
-2.94824749e-01 -7.22906590e-02 -6.27223969e-01 5.20667911e-01
-9.35566723e-01 8.99878070e-02 -4.89560753e-01 4.27523524e-01
7.84545168e-02 -2.56753683e-01 1.48011759e-01 3.21232408e-01
1.11433275e-01 7.58804500e-01 -1.32307160e+00 1.37549388e+00
-6.08103395e-01 2.75340527e-01 -2.17536569e-01 4.71352451e-02
8.66288960e-01 3.46933305e-01 -1.00383401e-01 -3.66464823e-01
-2.46366158e-01 4.76904392e-01 8.17629173e-02 -4.91927832e-01
9.38388050e-01 -1.53500393e-01 -1.75841630e-01 1.02049637e+00
1.02764308e-01 -6.05060220e-01 8.62779140e-01 4.24516350e-01
1.09177196e+00 1.65173471e-01 1.72069147e-01 -1.55152112e-01
3.45251352e-01 2.63566941e-01 3.24779600e-01 1.34125352e+00
1.14463806e-01 7.71293342e-01 4.10124749e-01 1.40359446e-01
-9.09445286e-01 -9.65414882e-01 2.93541819e-01 1.00262880e+00
-4.09832180e-01 -6.20700419e-01 -9.32012022e-01 -6.81640625e-01
-9.36126933e-02 1.31979239e+00 -1.50643826e-01 1.78272963e-01
-6.11694217e-01 -4.63979036e-01 5.87298930e-01 1.63449362e-01
-5.83516583e-02 -1.37463522e+00 -7.44446158e-01 3.92715275e-01
-4.86921251e-01 -7.91262925e-01 -6.44213080e-01 -2.68375725e-01
-6.20931029e-01 -8.68606746e-01 -6.40502632e-01 -5.51079214e-01
6.93166316e-01 8.52573588e-02 1.64035785e+00 2.90121377e-01
-7.61989877e-02 6.08499527e-01 -3.93262625e-01 -3.85355771e-01
-1.02384758e+00 4.11489964e-01 -3.35932702e-01 -3.26580465e-01
1.80589974e-01 -9.81079414e-02 -3.94096434e-01 1.02340363e-01
-1.08422470e+00 2.10623592e-02 5.03734529e-01 6.26861393e-01
2.71533042e-01 -3.79461139e-01 1.18312204e+00 -9.21365738e-01
1.52837312e+00 -4.19037670e-01 -5.94375312e-01 5.43692410e-01
-7.62389779e-01 2.22451061e-01 6.42585516e-01 -4.05230075e-01
-1.09011185e+00 -2.75193870e-01 -2.89643314e-02 2.37225160e-01
-2.73315102e-01 4.81375873e-01 -1.16821252e-01 2.46955797e-01
1.01677668e+00 1.50611043e-01 -8.10760185e-02 -2.97499716e-01
5.53039670e-01 6.44862175e-01 1.06985271e-01 -9.97039080e-01
8.42176795e-01 -1.54644579e-01 -5.19768298e-01 -6.32646501e-01
-6.44626319e-01 -2.09035352e-01 -9.12606046e-02 -1.33073047e-01
4.68216658e-01 -7.32329547e-01 -2.08861232e-01 -7.04528391e-03
-1.17955816e+00 -5.28216064e-01 -4.85019267e-01 2.17075482e-01
-5.51780701e-01 2.76127279e-01 -4.58576709e-01 -1.14906073e+00
-5.25563955e-01 -9.52560961e-01 1.07141709e+00 4.79202360e-01
-9.59501565e-01 -7.37986386e-01 7.97759518e-02 5.60840189e-01
5.67382753e-01 2.17059031e-02 1.12186861e+00 -7.85584748e-01
-8.41246307e-01 -4.30347979e-01 3.82276416e-01 7.84296170e-02
1.14459254e-01 1.42547429e-01 -6.66251481e-01 -1.17724746e-01
-3.03957850e-01 -5.70566237e-01 3.49944264e-01 1.33351758e-01
6.11992776e-01 -4.59367901e-01 -3.15850288e-01 -1.32014662e-01
1.23086059e+00 9.57506150e-02 3.81312817e-01 2.26757884e-01
1.37623817e-01 7.94875026e-01 6.87232316e-01 4.32201892e-01
5.70404589e-01 5.28400600e-01 -1.77100167e-01 2.12993398e-01
-3.19073573e-02 -6.64439917e-01 3.83205771e-01 6.22328877e-01
5.16849935e-01 -8.07599425e-01 -9.40086067e-01 9.91709292e-01
-1.64273739e+00 -9.69956815e-01 5.04957326e-02 2.38353777e+00
1.06872487e+00 2.40420356e-01 3.31163883e-01 -1.81468919e-01
5.60902536e-01 6.71517849e-03 -2.40650758e-01 -6.00064635e-01
3.46256886e-03 3.48602086e-01 -5.97725660e-02 6.20784104e-01
-5.28158367e-01 8.01267028e-01 7.12679958e+00 7.00887322e-01
-5.61524451e-01 -1.02063820e-01 7.33768940e-01 -3.05220693e-01
-1.11779392e+00 2.96466410e-01 -1.01405466e+00 4.51255351e-01
1.12171161e+00 -6.44009471e-01 3.03916484e-01 5.47980189e-01
3.34624916e-01 -3.39936167e-01 -1.03164840e+00 5.13359964e-01
1.94893464e-01 -1.31340659e+00 4.76003498e-01 -4.06218059e-02
9.17960584e-01 -5.15141070e-01 -2.25217327e-01 4.54086125e-01
7.87950933e-01 -1.16093016e+00 9.33486760e-01 5.01732886e-01
6.31562471e-01 -7.57170081e-01 4.01506841e-01 8.27039838e-01
-8.17502022e-01 -9.67817530e-02 -9.76437479e-02 1.70169696e-01
4.59503114e-01 4.96428102e-01 -1.11351442e+00 3.30883443e-01
4.98739257e-02 -4.32015508e-01 -6.85996592e-01 1.05211520e+00
-3.10467124e-01 9.49010313e-01 -2.66483456e-01 -3.99950624e-01
2.19945207e-01 3.33318934e-02 4.31170344e-01 1.20252645e+00
5.05837023e-01 -1.60491075e-02 3.08487535e-01 1.14584255e+00
-4.89313230e-02 3.40604126e-01 -5.11760592e-01 -3.11471194e-01
7.84074783e-01 1.27651668e+00 -5.53510785e-01 -3.35355073e-01
-8.42336938e-02 4.90665764e-01 1.90216988e-01 3.56199026e-01
-4.38610762e-01 -4.57128555e-01 3.07542473e-01 5.19342959e-01
-8.90289396e-02 -1.25831991e-01 -3.29570204e-01 -9.94285583e-01
2.69488723e-04 -1.53945780e+00 3.54547828e-01 -8.38596761e-01
-1.13044465e+00 6.78656280e-01 1.20694079e-01 -8.63613188e-01
-1.05046570e+00 -3.15820649e-02 -7.67412007e-01 1.31628871e+00
-1.05144799e+00 -1.00467002e+00 -5.99169917e-02 -5.97226024e-02
7.01181471e-01 -1.48466527e-01 7.84794331e-01 -5.96833788e-02
-1.14111878e-01 6.39152467e-01 -3.13627750e-01 -3.31794977e-01
6.46613240e-01 -1.36228764e+00 7.57281661e-01 9.02608335e-01
2.88940847e-01 9.59963739e-01 8.20180178e-01 -7.94781327e-01
-1.18625998e+00 -9.77071583e-01 1.52002513e+00 -5.98616183e-01
3.87955785e-01 -4.25459832e-01 -8.48305047e-01 3.23340118e-01
3.74927670e-01 -7.77151585e-01 7.55879283e-01 1.96332838e-02
1.05652772e-01 2.39910483e-01 -1.19006443e+00 9.21186388e-01
6.70319974e-01 -4.44006115e-01 -6.62519276e-01 5.42758107e-01
6.62033260e-01 -3.90131146e-01 -5.16517282e-01 2.08053999e-02
3.64381045e-01 -8.54324818e-01 5.90941012e-01 -5.14298856e-01
4.25002426e-01 -3.37551564e-01 5.38841039e-02 -1.26618576e+00
-3.73934731e-02 -1.24123943e+00 2.13762999e-01 1.58560741e+00
7.89139986e-01 -4.92000937e-01 6.67800009e-01 1.04097426e+00
1.22856334e-01 -9.32420254e-01 -5.28207660e-01 -6.97849154e-01
1.52612656e-01 -2.12381363e-01 7.87337065e-01 3.85659695e-01
-2.25492362e-02 5.00141203e-01 -4.11743112e-02 -2.07888454e-01
2.80633658e-01 2.24019915e-01 1.11169779e+00 -9.25695598e-01
-5.78892648e-01 -4.73366141e-01 2.94440895e-01 -1.17360091e+00
-2.03042142e-02 -6.63097322e-01 4.53013062e-01 -1.95632219e+00
7.99150616e-02 -5.13235331e-01 1.59951672e-01 5.20902872e-02
-5.72243214e-01 -2.69624293e-01 2.97430366e-01 -6.15707133e-03
-5.69855273e-01 3.48922133e-01 9.33673024e-01 2.75425553e-01
-3.78082186e-01 1.29163370e-01 -1.08755028e+00 4.37170416e-01
8.07733536e-01 -5.53659558e-01 -8.14386725e-01 -3.84671211e-01
3.94432783e-01 3.49841833e-01 9.80049819e-02 -9.09836709e-01
4.20983545e-02 -4.57609564e-01 9.95441005e-02 -5.30516028e-01
1.87449887e-01 1.05860263e-01 9.71487164e-02 2.39921257e-01
-9.26186621e-01 3.11693311e-01 8.60933736e-02 2.80737877e-01
-1.14938933e-02 -9.29557621e-01 4.66849297e-01 -6.04762435e-01
-2.65308619e-01 -1.62004381e-02 -6.12586319e-01 6.55400455e-01
6.32674456e-01 -1.80656716e-01 -3.46978098e-01 -8.08876872e-01
-4.51889522e-02 1.91521138e-01 5.61604023e-01 5.15232921e-01
4.08089370e-01 -1.00737858e+00 -9.22867000e-01 -1.86187699e-01
1.46399245e-01 -2.18037888e-01 -3.33480388e-01 2.51270318e-03
-3.47950041e-01 6.22611642e-01 3.13479125e-01 -1.37596786e-01
-9.60290432e-01 5.96164204e-02 3.27788264e-01 -5.91567695e-01
-1.52668476e-01 9.04795945e-01 -1.65132478e-01 -5.74578881e-01
6.82993457e-02 -1.86273962e-01 -5.29597327e-02 -2.38135122e-02
4.61673051e-01 2.03222856e-01 3.41673158e-02 -2.47760773e-01
-1.09053634e-01 -1.90822214e-01 -3.74450386e-02 -9.36052918e-01
9.48487937e-01 -7.62097985e-02 -2.81371847e-02 4.08615023e-01
5.22546053e-01 4.32004690e-01 -8.03420722e-01 3.61038893e-02
4.54526246e-01 -5.46375513e-01 -5.29347539e-01 -9.56325829e-01
-1.07314795e-01 3.81525815e-01 5.73027208e-02 4.99496371e-01
7.37430692e-01 -9.79772434e-02 1.06142068e+00 4.59334075e-01
5.16084313e-01 -1.19184637e+00 3.40455234e-01 2.76809514e-01
1.00689423e+00 -9.94197965e-01 -2.50840575e-01 -1.18054330e-01
-7.32872307e-01 9.16237056e-01 9.47537959e-01 2.53776371e-01
-2.37925537e-02 3.51657234e-02 2.98090190e-01 1.21049352e-01
-1.38800311e+00 -1.72767386e-01 2.07437992e-01 5.64279556e-01
8.40060413e-01 -9.18930620e-02 -6.85608923e-01 4.41630691e-01
-4.82029110e-01 -2.17075627e-02 8.30049574e-01 1.01564252e+00
-7.16222763e-01 -1.42282355e+00 -5.12093365e-01 8.23649585e-01
-4.60148811e-01 -2.56403863e-01 -5.92407346e-01 6.25748098e-01
3.26720364e-02 1.61362076e+00 -2.68551975e-01 4.76653725e-02
3.93759191e-01 3.73921305e-01 4.05236155e-01 -1.11142778e+00
-1.01630342e+00 -1.12035505e-01 5.71979463e-01 -4.65596281e-02
1.91624627e-01 -7.41079986e-01 -1.01265407e+00 1.09822869e-01
-6.08610809e-01 4.81326461e-01 4.69332069e-01 8.74837697e-01
5.35938740e-01 3.60941328e-02 4.43019718e-01 -3.34834278e-01
-1.12759292e+00 -9.10013378e-01 -2.50727415e-01 4.95282710e-01
-5.68193868e-02 -1.80310860e-01 -1.82612747e-01 4.07515988e-02] | [11.713275909423828, 8.388513565063477] |
7b8e465e-82ea-445a-a908-314680fde612 | multi-agent-feedback-enabled-neural-networks | 2205.10750 | null | https://arxiv.org/abs/2205.10750v1 | https://arxiv.org/pdf/2205.10750v1.pdf | Multi-Agent Feedback Enabled Neural Networks for Intelligent Communications | In the intelligent communication field, deep learning (DL) has attracted much attention due to its strong fitting ability and data-driven learning capability. Compared with the typical DL feedforward network structures, an enhancement structure with direct data feedback have been studied and proved to have better performance than the feedfoward networks. However, due to the above simple feedback methods lack sufficient analysis and learning ability on the feedback data, it is inadequate to deal with more complicated nonlinear systems and therefore the performance is limited for further improvement. In this paper, a novel multi-agent feedback enabled neural network (MAFENN) framework is proposed, which make the framework have stronger feedback learning capabilities and more intelligence on feature abstraction, denoising or generation, etc. Furthermore, the MAFENN framework is theoretically formulated into a three-player Feedback Stackelberg game, and the game is proved to converge to the Feedback Stackelberg equilibrium. The design of MAFENN framework and algorithm are dedicated to enhance the learning capability of the feedfoward DL networks or their variations with the simple data feedback. To verify the MAFENN framework's feasibility in wireless communications, a multi-agent MAFENN based equalizer (MAFENN-E) is developed for wireless fading channels with inter-symbol interference (ISI). Experimental results show that when the quadrature phase-shift keying (QPSK) modulation scheme is adopted, the SER performance of our proposed method outperforms that of the traditional equalizers by about 2 dB in linear channels. When in nonlinear channels, the SER performance of our proposed method outperforms that of either traditional or DL based equalizers more significantly, which shows the effectiveness and robustness of our proposal in the complex channel environment. | ['Kai Li', 'Yang Yang', 'Jun Wang', 'Jingchen Hu', 'Ying Wen', 'Yang Li', 'Fanglei Sun'] | 2022-05-22 | null | null | null | null | ['intelligent-communication'] | ['time-series'] | [-3.07069123e-01 4.43049427e-03 -1.90163881e-01 1.23170717e-03
-3.94400284e-02 1.87120870e-01 1.94267422e-01 -2.95366228e-01
-5.35467207e-01 1.18965757e+00 -1.28308907e-01 -5.95457137e-01
-7.44249165e-01 -8.68210137e-01 -4.86099213e-01 -1.10291529e+00
-5.85789740e-01 -1.33579060e-01 -7.52029717e-02 -7.04113603e-01
2.12847907e-02 1.19871788e-01 -1.11687088e+00 -2.51258582e-01
1.18476450e+00 1.14617229e+00 4.73112404e-01 7.47560203e-01
1.33543294e-02 8.21649313e-01 -6.32921994e-01 -9.42368582e-02
3.18168640e-01 -5.99297166e-01 -1.49285659e-01 -2.19891384e-01
-3.18989575e-01 -5.42297602e-01 -6.34101689e-01 1.14024568e+00
1.08161044e+00 -1.37530133e-01 4.61317241e-01 -1.30571067e+00
-5.32168627e-01 6.99224412e-01 -4.33447689e-01 1.02428049e-01
-8.12084228e-02 -2.86516070e-01 6.38394892e-01 -9.18227196e-01
7.03468546e-02 1.21539438e+00 9.02234137e-01 5.92613578e-01
-5.63393891e-01 -1.23152554e+00 2.15873808e-01 4.02488738e-01
-1.15173745e+00 -1.15115620e-01 5.74755013e-01 -2.40329430e-02
5.40515125e-01 1.28805682e-01 9.01402950e-01 5.23445189e-01
3.74991149e-01 9.92523789e-01 1.05097091e+00 -6.55018628e-01
1.05321057e-01 3.59144539e-01 9.89268869e-02 6.99429035e-01
3.79605740e-01 5.85827768e-01 -2.53670782e-01 1.38164446e-01
9.78907645e-01 -9.99619514e-02 -6.07531965e-01 -5.56268170e-02
-5.96967161e-01 7.47400939e-01 8.80564153e-01 3.02061647e-01
-4.20087755e-01 2.68470466e-01 1.14983372e-01 9.88121271e-01
3.45094651e-01 2.38620192e-01 -4.23853248e-01 1.37238309e-01
-6.44945264e-01 1.89689696e-01 8.73095095e-01 8.34298015e-01
6.24625802e-01 6.23054147e-01 1.07496500e-01 4.73853827e-01
6.72990024e-01 7.85638988e-01 5.58152258e-01 -4.88042265e-01
4.58316863e-01 4.33958173e-01 3.05876639e-02 -1.13550770e+00
-8.58859420e-01 -1.23503184e+00 -1.50951278e+00 4.45576847e-01
1.67215914e-01 -9.15946662e-01 -4.51554805e-01 1.54991412e+00
-5.19022942e-02 1.36532351e-01 6.68347418e-01 1.03502977e+00
1.07585609e+00 9.49344635e-01 -3.93014699e-01 -7.51082957e-01
7.69185126e-01 -6.42961442e-01 -1.11050189e+00 -2.25203410e-02
5.24581671e-01 -5.31332016e-01 4.72559035e-01 6.83506668e-01
-7.96356976e-01 -6.47814453e-01 -1.47885716e+00 7.24365115e-01
-2.77945008e-02 4.83249947e-02 7.15266287e-01 1.05953097e+00
-1.06042957e+00 3.08471113e-01 -2.77478456e-01 1.18839890e-02
1.22545823e-01 8.32598209e-01 3.21242772e-02 1.90640688e-01
-1.85718071e+00 6.57395065e-01 5.54185510e-01 3.99580091e-01
-5.34700990e-01 -4.22401756e-01 -4.68368441e-01 5.31756841e-02
3.31739008e-01 -5.71950793e-01 1.30902708e+00 -1.54987657e+00
-1.83453953e+00 -2.32923791e-01 3.19704026e-01 -6.58503413e-01
5.62798023e-01 -1.48636192e-01 -5.57393551e-01 -1.07237905e-01
-4.11861956e-01 1.62819460e-01 6.82592869e-01 -1.04343939e+00
-8.94034982e-01 -2.48833504e-02 2.69269139e-01 5.23345530e-01
-7.09177136e-01 -2.29418650e-01 1.05117485e-01 -6.39410198e-01
-8.35442021e-02 -4.58573222e-01 -4.34291422e-01 -2.03801394e-02
-1.38155207e-01 -6.07632212e-02 1.07386529e+00 -4.01821613e-01
1.62543476e+00 -1.99047971e+00 -1.79253921e-01 4.77369845e-01
2.33851314e-01 7.30829537e-01 7.93882906e-02 5.14245391e-01
7.01724440e-02 2.90833544e-02 2.48073474e-01 3.77615303e-01
-1.19416527e-01 1.22511528e-01 2.06117362e-01 2.99368054e-01
-2.86688209e-01 6.35376871e-01 -8.14405084e-01 -2.66695440e-01
-1.98951233e-02 1.73890635e-01 -6.24107897e-01 1.43289000e-01
6.43310025e-02 2.02722445e-01 -6.53210521e-01 4.22995329e-01
8.63904059e-01 -7.23563433e-02 -3.10753472e-02 -6.09482080e-02
-3.07483763e-01 -3.36641639e-01 -1.52042556e+00 8.87118578e-01
-6.99279487e-01 6.14107847e-01 6.11578643e-01 -1.20492148e+00
1.28010118e+00 6.49575353e-01 4.03238803e-01 -1.10113513e+00
4.21939582e-01 5.18580496e-01 4.05468911e-01 -6.42335117e-01
1.35390654e-01 -1.95087641e-01 2.41783425e-01 2.46608526e-01
1.73905864e-01 1.96611986e-01 -1.11286208e-01 1.98068514e-01
7.31130779e-01 -3.53498220e-01 2.16895789e-01 -4.77065116e-01
9.84846890e-01 -5.58122993e-01 7.27334082e-01 7.89760053e-01
-9.80811417e-02 -1.74961388e-01 1.72758862e-01 -2.67773628e-01
-7.35793114e-01 -5.61406910e-01 -3.12594473e-02 8.05230916e-01
5.82173765e-01 -6.92910179e-02 -4.94905621e-01 -3.35954458e-01
8.62378478e-02 3.17996830e-01 -9.51225758e-02 -3.74371022e-01
-8.02596211e-02 -1.11761296e+00 7.62855530e-01 1.62762970e-01
1.51537323e+00 -7.72432089e-01 7.80828344e-03 5.67393601e-01
9.76192355e-02 -4.14503664e-01 -9.40772742e-02 2.41303787e-01
-5.63459456e-01 -8.95596266e-01 -1.10654294e+00 -1.01418281e+00
2.98378170e-01 1.39331818e-01 4.80592370e-01 2.10801080e-01
3.28213781e-01 2.53688157e-01 -5.63865185e-01 -8.85348320e-01
-4.39934820e-01 2.13772082e-03 3.32103133e-01 1.24457315e-01
-8.62933397e-02 -4.38595444e-01 -5.64411640e-01 3.30647618e-01
-5.66799164e-01 -3.00796656e-03 1.03796685e+00 1.14228106e+00
-2.39042938e-01 4.81104553e-01 1.47353470e+00 -6.06480479e-01
1.18523300e+00 -3.33972722e-01 -6.35076821e-01 8.50086380e-03
-9.20024753e-01 -1.59649756e-02 1.00702488e+00 -2.50449568e-01
-1.29709411e+00 -4.58742142e-01 -3.81522208e-01 1.34645760e-01
6.71850443e-01 9.00941253e-01 -2.39562809e-01 -6.65984035e-01
7.20162034e-01 3.10157925e-01 4.92107600e-01 1.05207658e-03
3.47809750e-03 1.24601471e+00 2.06757888e-01 -2.80082405e-01
7.29690611e-01 -1.91910461e-01 1.08158410e-01 -9.25004601e-01
-2.10309178e-01 -1.02156438e-01 1.07020251e-01 -6.58620656e-01
3.35528225e-01 -1.02195740e+00 -1.19716597e+00 9.69350219e-01
-9.75650012e-01 -2.52361774e-01 2.60020554e-01 1.03703332e+00
-4.08413947e-01 2.23763376e-01 -8.10505927e-01 -1.17510331e+00
-5.79644680e-01 -9.45894480e-01 -6.24676906e-02 8.03096652e-01
5.26010036e-01 -1.21080363e+00 -4.03190583e-01 -2.15665638e-01
7.95740962e-01 -6.64503947e-02 7.33056784e-01 -3.75559360e-01
-4.37821954e-01 -1.15760416e-01 -3.03109020e-01 6.08552635e-01
-5.91448285e-02 -2.72015303e-01 -6.59374177e-01 -6.21413887e-01
1.20611399e-01 -3.41594964e-01 6.82162106e-01 6.21937811e-01
6.84353411e-01 -4.99035180e-01 -1.74325243e-01 6.77055538e-01
1.72365808e+00 6.06213868e-01 6.70247734e-01 5.72628796e-01
6.58682659e-02 1.65129736e-01 6.50822341e-01 6.58791363e-01
2.94877380e-01 1.63624868e-01 8.41378689e-01 -5.23295045e-01
1.84298575e-01 4.33983430e-02 5.64947546e-01 1.09643745e+00
-3.23933542e-01 -4.79224473e-01 -3.07817042e-01 -1.54484315e-02
-1.92088914e+00 -9.39311028e-01 -4.58344847e-01 2.07211876e+00
6.37263477e-01 2.13402972e-01 -1.37785003e-01 3.60664845e-01
6.98124468e-01 -2.58323669e-01 -6.89396977e-01 -5.76704204e-01
-3.28530729e-01 1.40473351e-01 8.58602881e-01 4.85494554e-01
-1.05799127e+00 2.68492371e-01 5.67941189e+00 1.13807046e+00
-1.40570974e+00 -1.49037644e-01 4.38827068e-01 4.18637961e-01
-1.24687418e-01 -2.86103398e-01 -6.58903539e-01 4.89441901e-01
6.31123185e-01 -5.15346885e-01 3.44249666e-01 5.92028439e-01
6.73423409e-01 -4.32530940e-02 -3.04584175e-01 1.17291331e+00
-3.74811649e-01 -1.18896985e+00 -1.80797160e-01 1.21731147e-01
7.14626849e-01 -2.45286286e-01 1.65802971e-01 6.16208911e-01
3.07046294e-01 -8.33542347e-01 3.71366858e-01 8.11337113e-01
5.98371744e-01 -1.07783616e+00 1.39703512e+00 7.97415853e-01
-9.23806310e-01 -7.53516376e-01 -3.41392726e-01 -7.58627474e-01
-1.31353602e-01 6.88371122e-01 -5.99529028e-01 1.14219379e+00
4.30158824e-01 6.27288282e-01 -1.10879935e-01 1.44783771e+00
-1.38233945e-01 6.34234846e-01 -2.01892644e-01 -8.56868565e-01
4.34708834e-01 -1.90939233e-01 4.98398095e-01 9.17901337e-01
5.58352709e-01 2.54362315e-01 2.22462401e-01 3.10907841e-01
6.68843687e-02 3.57608229e-01 -3.03055167e-01 3.66466284e-01
5.68499565e-01 1.18390441e+00 -1.87512171e-02 -2.22504914e-01
-4.94079202e-01 4.44136500e-01 -2.25597635e-01 4.68272090e-01
-6.09762669e-01 -1.00048602e+00 3.42260063e-01 -6.71755821e-02
-1.52140846e-02 -6.66920766e-02 -7.90563822e-02 -8.21924150e-01
-1.54715151e-01 -8.41396332e-01 1.21958084e-01 -6.37404799e-01
-1.22622716e+00 4.59275216e-01 -4.26385999e-01 -1.71822894e+00
-2.55010903e-01 -6.73457623e-01 -6.47025704e-01 8.68095756e-01
-1.72370470e+00 -7.63820112e-01 -4.19249058e-01 7.89280117e-01
1.30496845e-01 -7.42643476e-01 6.81192398e-01 6.71112955e-01
-4.52763677e-01 8.93015623e-01 8.10872853e-01 3.29230540e-02
5.83097041e-01 -1.01035035e+00 -6.39547825e-01 6.69317842e-01
-6.35729671e-01 3.34049314e-01 6.33282542e-01 -2.94669390e-01
-1.37796104e+00 -9.42851007e-01 3.43062222e-01 8.84468377e-01
4.96684700e-01 -8.23194981e-02 -5.14526844e-01 8.62243995e-02
4.66850758e-01 -1.59987941e-01 3.73454362e-01 -1.91158444e-01
5.37243426e-01 -6.02564573e-01 -1.03546131e+00 3.88495922e-01
6.34373546e-01 1.66545704e-01 -4.33287956e-02 6.89218715e-02
4.38202113e-01 -4.08866495e-01 -6.87264442e-01 4.15206909e-01
6.01234555e-01 -9.49854553e-01 6.44725978e-01 -1.88209310e-01
9.75068212e-02 -3.58506858e-01 -9.52310767e-03 -1.72863507e+00
-4.49199080e-01 -8.13004851e-01 1.37688488e-01 1.07965529e+00
4.05773908e-01 -1.05355763e+00 7.95621455e-01 -9.41739753e-02
-2.18962073e-01 -9.33351755e-01 -8.98336232e-01 -7.84094274e-01
1.17209569e-01 -2.25961849e-01 3.71401340e-01 5.17377734e-01
2.92230904e-01 4.82399732e-01 -6.49358749e-01 2.25711882e-01
4.79640365e-01 -5.32790005e-01 7.29126573e-01 -1.35103345e+00
-4.56318319e-01 -4.74921882e-01 -5.12108624e-01 -1.50310552e+00
-1.59220576e-01 -8.01964164e-01 -2.13557124e-01 -1.49117756e+00
-2.93832451e-01 -9.27889526e-01 -3.95304650e-01 4.36705463e-02
1.54982936e-02 -7.28128254e-02 2.18261123e-01 9.51325074e-02
-3.04698050e-01 8.04838598e-01 1.61290538e+00 -3.05338323e-01
-3.28928649e-01 7.28397608e-01 -7.44245172e-01 5.67912936e-01
8.14444065e-01 1.52858004e-01 -6.64491773e-01 -1.19834207e-01
3.63898724e-01 5.96940160e-01 1.71416953e-01 -1.26140928e+00
5.54482698e-01 1.01558760e-01 4.91144031e-01 -2.23459557e-01
4.09365632e-02 -1.00419021e+00 -1.95916042e-01 9.68598366e-01
-6.89079314e-02 -4.09629703e-01 -4.75938199e-03 5.42526543e-01
-4.73793060e-01 -3.02520394e-01 5.51967084e-01 2.27186512e-02
-7.84177184e-01 4.38484699e-01 -7.34247684e-01 -3.33457619e-01
9.94084954e-01 -3.87320727e-01 -1.82086483e-01 -1.08546245e+00
-7.43766963e-01 6.84172690e-01 -3.82872909e-01 6.96846843e-02
7.58487880e-01 -1.51099467e+00 -7.37943709e-01 4.64533716e-01
-4.67273146e-01 -2.16155916e-01 1.49550661e-01 9.61230397e-01
-4.99140292e-01 3.55732679e-01 -2.18435660e-01 -3.61561418e-01
-1.10077870e+00 3.90954912e-02 8.41181517e-01 -3.36949199e-01
-1.47648782e-01 8.03616285e-01 2.56366562e-02 -5.31196535e-01
3.41870993e-01 -1.38871491e-01 -5.31949043e-01 -2.92583033e-02
3.57918978e-01 4.57648367e-01 7.46892169e-02 -1.63472995e-01
-1.54121891e-02 3.13872814e-01 3.61213945e-02 2.08189636e-02
1.20336580e+00 -1.81292430e-01 -7.74195120e-02 -1.06810136e-02
1.01011801e+00 -8.00977647e-02 -1.14467657e+00 -3.46173972e-01
-3.62744093e-01 -7.49014318e-02 3.61484706e-01 -7.29239881e-01
-1.20393157e+00 6.87201798e-01 9.27984715e-01 4.97913063e-01
1.42164612e+00 -7.84191608e-01 6.86240733e-01 5.46565354e-01
6.56732202e-01 -1.18016684e+00 7.97384456e-02 7.42289305e-01
7.02973187e-01 -9.32877541e-01 -7.61844069e-02 -3.50084692e-01
-1.28529847e-01 1.71057880e+00 7.92262197e-01 -3.08213532e-01
1.14368010e+00 5.32470047e-01 1.97063655e-01 2.62808710e-01
-6.65436804e-01 -1.81822300e-01 -1.83629096e-01 7.79554188e-01
3.64607424e-01 -7.73654506e-02 -8.03812385e-01 9.42364573e-01
-2.74447858e-01 3.02275419e-01 8.28025639e-01 6.60878897e-01
-9.71159637e-01 -1.10384762e+00 -5.42481899e-01 6.47232473e-01
-3.50439608e-01 1.07236587e-01 1.40125751e-01 1.09229815e+00
2.17575759e-01 1.13677335e+00 -1.75675020e-01 -6.35556519e-01
3.87575567e-01 -7.04786837e-01 3.33984256e-01 3.27995010e-02
-3.67856175e-01 9.17866901e-02 2.36355700e-02 8.62120092e-03
-5.41420162e-01 -1.61679655e-01 -1.55559289e+00 -3.66271675e-01
-8.90931070e-01 4.74859744e-01 4.12060440e-01 1.12227738e+00
-1.51662966e-02 8.22301567e-01 1.04506969e+00 -5.85986018e-01
-6.88346148e-01 -9.58444476e-01 -9.00057435e-01 -3.18491757e-01
5.47127903e-01 -7.55547345e-01 -2.86366582e-01 -6.14082396e-01] | [6.292973041534424, 1.5160455703735352] |
c4098202-c7ac-4315-a02c-2a0f4ec997f6 | local-and-global-context-based-pairwise | 2110.04291 | null | https://arxiv.org/abs/2110.04291v2 | https://arxiv.org/pdf/2110.04291v2.pdf | Local and Global Context-Based Pairwise Models for Sentence Ordering | Sentence Ordering refers to the task of rearranging a set of sentences into the appropriate coherent order. For this task, most previous approaches have explored global context-based end-to-end methods using Sequence Generation techniques. In this paper, we put forward a set of robust local and global context-based pairwise ordering strategies, leveraging which our prediction strategies outperform all previous works in this domain. Our proposed encoding method utilizes the paragraph's rich global contextual information to predict the pairwise order using novel transformer architectures. Analysis of the two proposed decoding strategies helps better explain error propagation in pairwise models. This approach is the most accurate pure pairwise model and our encoding strategy also significantly improves the performance of other recent approaches that use pairwise models, including the previous state-of-the-art, demonstrating the research novelty and generalizability of this work. Additionally, we show how the pre-training task for ALBERT helps it to significantly outperform BERT, despite having considerably lesser parameters. The extensive experimental results, architectural analysis and ablation studies demonstrate the effectiveness and superiority of the proposed models compared to the previous state-of-the-art, besides providing a much better understanding of the functioning of pairwise models. | ['Aditya Jyoti Paul', 'Ruskin Raj Manku'] | 2021-10-08 | null | null | null | null | ['sentence-ordering'] | ['natural-language-processing'] | [ 4.41387743e-01 1.55326560e-01 -7.90724531e-03 -6.15737975e-01
-1.01823199e+00 -5.49515009e-01 6.61458731e-01 1.76600248e-01
-1.74669847e-01 1.01509917e+00 6.91334546e-01 -3.72607261e-01
-1.70346931e-01 -4.64704424e-01 -7.10884571e-01 -4.95086133e-01
-1.07222438e-01 5.31914771e-01 1.99743479e-01 -6.87366545e-01
7.88024962e-01 -1.68874916e-02 -1.46044970e+00 5.66942692e-01
9.53306377e-01 5.95479071e-01 3.40807885e-01 6.91517234e-01
-4.51867431e-02 8.14704835e-01 -6.66310012e-01 -6.68709993e-01
4.06043716e-02 -7.41089344e-01 -9.14534271e-01 -3.99797797e-01
6.42772138e-01 -2.87019819e-01 -3.48748527e-02 6.44828081e-01
6.98177338e-01 1.72288105e-01 4.71328348e-01 -9.08493698e-01
-8.66054237e-01 1.05731320e+00 -2.91217715e-01 1.07025385e-01
7.92061329e-01 -1.44178942e-01 1.52416968e+00 -8.88416350e-01
6.02648258e-01 9.99035120e-01 7.79881477e-01 5.05321503e-01
-1.06231308e+00 -6.14390671e-01 3.80354196e-01 7.44500637e-01
-1.32379866e+00 -5.32663345e-01 8.25811982e-01 -8.47885683e-02
1.38778532e+00 5.74523211e-01 5.62846780e-01 1.41455674e+00
1.15833186e-01 8.34952712e-01 1.01640058e+00 -8.70248795e-01
1.29369840e-01 -1.49868146e-01 3.70267063e-01 5.56119263e-01
-1.76140554e-02 -4.39354330e-02 -8.73755336e-01 7.71529153e-02
6.60580024e-02 -4.63197321e-01 -3.36352021e-01 -1.66802272e-01
-1.24739301e+00 5.46583414e-01 2.00528592e-01 6.30309999e-01
-1.24387704e-01 -2.43473630e-02 5.38497984e-01 2.24239752e-01
4.77966338e-01 5.22042155e-01 -4.39445347e-01 -5.24781764e-01
-1.41148782e+00 3.19840223e-01 8.97154629e-01 1.18368828e+00
3.99216264e-01 -2.90981501e-01 -6.45675063e-01 6.92582011e-01
3.98056209e-01 -7.31280521e-02 3.23439837e-01 -7.13697076e-01
8.02543521e-01 3.73114228e-01 7.33209550e-02 -9.15174186e-01
-2.11856782e-01 -7.56269217e-01 -7.10492015e-01 -5.59533894e-01
1.31407052e-01 2.32661888e-01 -6.37277722e-01 1.92262495e+00
-9.74305719e-02 4.41017061e-01 -6.94680065e-02 6.67123318e-01
6.11671388e-01 6.08187735e-01 -1.19644500e-01 -2.82882124e-01
1.20269763e+00 -1.31784093e+00 -7.64134765e-01 -3.86144608e-01
6.76967919e-01 -8.97466004e-01 8.89987350e-01 4.08544689e-01
-1.23731124e+00 -5.07266343e-01 -1.25263357e+00 -1.62722528e-01
-9.36902612e-02 2.41765127e-01 5.28122127e-01 7.40974724e-01
-1.55550981e+00 6.37098908e-01 -5.26749194e-01 -6.90690160e-01
1.05896965e-01 2.07800895e-01 -2.41778508e-01 -2.13104710e-02
-1.48188770e+00 1.11998558e+00 2.72972465e-01 3.09160769e-01
-5.81063449e-01 -7.29154825e-01 -7.34059572e-01 2.61855215e-01
6.96712583e-02 -1.04539943e+00 1.33643353e+00 -5.76250792e-01
-1.76463115e+00 4.97217745e-01 -7.38685548e-01 -7.24039495e-01
3.72749448e-01 -3.49657029e-01 -8.00981820e-02 1.28229082e-01
2.31202375e-02 6.94911242e-01 4.99870449e-01 -1.16848326e+00
-4.21337843e-01 1.14744641e-02 -5.24191894e-02 3.55450958e-01
-4.41827208e-01 1.06263692e-02 -3.38860363e-01 -7.24768281e-01
-3.09195947e-02 -8.34654868e-01 -1.65041342e-01 -7.41508961e-01
-6.13774836e-01 -1.86786577e-01 4.43486154e-01 -8.33344221e-01
1.76929462e+00 -1.78310204e+00 4.28451985e-01 -7.05912127e-04
-1.00550041e-01 7.66885802e-02 -2.59168237e-01 1.20255649e+00
-1.55521244e-01 3.86074364e-01 -4.40031528e-01 -8.69160593e-01
1.91953212e-01 2.24086065e-02 -3.50356370e-01 1.11604922e-01
1.51072830e-01 8.68169487e-01 -9.39277112e-01 -4.53884214e-01
1.46663547e-01 4.84388828e-01 -6.95729613e-01 3.20186943e-01
-1.01636216e-01 5.18660903e-01 -5.28162606e-02 2.13905409e-01
5.68925917e-01 -6.08562725e-03 5.43631494e-01 -2.16054678e-01
1.45005351e-02 9.28429604e-01 -7.46942163e-01 2.24102592e+00
-5.35756111e-01 6.92434847e-01 -4.91682649e-01 -1.12174356e+00
8.76556396e-01 2.75399476e-01 6.14232160e-02 -7.52025127e-01
-5.97447306e-02 1.18771486e-01 7.98312426e-02 -4.53592986e-01
9.68991041e-01 -2.95839906e-02 -2.69295275e-02 3.00566703e-01
8.02836008e-03 2.26820603e-01 5.26247859e-01 6.45304441e-01
1.14568937e+00 4.93687391e-01 3.71985525e-01 -2.51082927e-01
6.49088264e-01 -2.69375384e-01 4.69816685e-01 1.05614293e+00
-1.25310749e-01 8.11122477e-01 3.47681999e-01 -7.65867671e-03
-1.10936987e+00 -7.58061528e-01 -7.02470588e-03 9.63651121e-01
9.66344550e-02 -9.65633094e-01 -8.75285625e-01 -7.12385833e-01
-5.52737355e-01 1.34897530e+00 -6.58899844e-01 -1.23341486e-01
-8.26439321e-01 -5.99555433e-01 6.85088098e-01 5.77366352e-01
4.51655179e-01 -7.20081568e-01 -3.02661866e-01 3.06707740e-01
-7.89971054e-01 -1.15890324e+00 -4.93049145e-01 -1.04528151e-01
-8.40201616e-01 -6.41713679e-01 -4.57918286e-01 -8.50165009e-01
4.18081522e-01 1.43951297e-01 1.31021726e+00 1.18306711e-01
6.26320988e-02 1.40025824e-01 -8.94345522e-01 -1.69816315e-02
-4.61892843e-01 3.87121737e-01 -2.46018887e-01 -9.76278409e-02
3.04145277e-01 -8.67115378e-01 -6.58985674e-01 1.14281662e-01
-5.88423252e-01 3.40933681e-01 7.54142225e-01 1.05657041e+00
1.73175797e-01 -3.00102472e-01 6.21537268e-01 -9.16633964e-01
7.46379733e-01 -3.47298384e-01 9.00754631e-02 5.67152083e-01
-7.29376912e-01 2.68575311e-01 6.71601236e-01 1.52932599e-01
-1.19455779e+00 -4.99920785e-01 -2.98579663e-01 1.93457007e-01
-1.84021160e-01 5.97187996e-01 -1.36092395e-01 2.96908349e-01
3.37296307e-01 5.56115746e-01 -2.13170901e-01 -3.69575381e-01
4.35885072e-01 5.14745295e-01 2.54982263e-01 -5.56613147e-01
5.06496489e-01 5.89666478e-02 -3.63954790e-02 -4.65675086e-01
-7.77769625e-01 -4.31241602e-01 -7.71447957e-01 -1.70776412e-01
5.75635850e-01 -7.28725851e-01 -4.46114391e-01 2.64506847e-01
-1.42443514e+00 -9.24611406e-04 1.01850726e-01 1.24500550e-01
-5.48864186e-01 8.70224059e-01 -6.25931382e-01 -8.44847679e-01
-5.48750997e-01 -1.24453390e+00 1.20305204e+00 -1.84172720e-01
-6.69698954e-01 -1.03151703e+00 1.57706633e-01 5.58161497e-01
6.10487819e-01 -1.55712724e-01 8.62105966e-01 -9.14977014e-01
-5.50886750e-01 1.64145119e-02 5.83438314e-02 2.02529013e-01
-2.21602410e-01 2.56749447e-02 -7.87981749e-01 -2.91866541e-01
-4.15450335e-02 2.51140185e-02 9.38846231e-01 1.68223098e-01
9.54168677e-01 -3.86110634e-01 -4.11427587e-01 2.98718482e-01
1.40009212e+00 7.22604245e-02 1.01417315e+00 6.44251466e-01
5.05670130e-01 6.45351470e-01 7.73165941e-01 4.70414817e-01
8.09840739e-01 9.40419018e-01 3.31376463e-01 2.97113270e-01
-2.14067593e-01 -3.71307552e-01 6.00129843e-01 1.33217156e+00
1.30878463e-01 -6.24910057e-01 -7.47225523e-01 6.11946464e-01
-2.12511444e+00 -1.33892798e+00 -1.61394790e-01 2.10206056e+00
9.31025267e-01 1.50071800e-01 -1.45264670e-01 1.75185829e-01
5.66234708e-01 2.89807022e-01 1.21659100e-01 -7.68118024e-01
-1.19898751e-01 4.34586078e-01 2.60063022e-01 6.61330819e-01
-8.90177786e-01 8.53053868e-01 6.91273975e+00 8.28527927e-01
-6.13273919e-01 -1.91467665e-02 5.26593745e-01 -2.81813473e-01
-5.35374403e-01 4.19154108e-01 -9.25612569e-01 6.25949800e-01
1.23525000e+00 1.06718652e-01 4.22059000e-01 4.08649236e-01
4.47314501e-01 -3.58089283e-02 -1.41137874e+00 6.31798267e-01
5.66769600e-01 -1.39309323e+00 2.13759288e-01 -3.02847266e-01
4.98100042e-01 -2.84873307e-01 -1.19032398e-01 3.40259135e-01
1.03857674e-01 -9.84910846e-01 1.04719150e+00 7.26147294e-01
2.90633202e-01 -6.88520133e-01 8.27836990e-01 2.86556721e-01
-1.06237721e+00 -6.41734377e-02 -9.60173607e-02 -3.65012407e-01
5.26414394e-01 7.40155160e-01 -9.62114334e-01 1.21243763e+00
5.44496953e-01 9.78985429e-01 -7.66094506e-01 9.17124689e-01
-4.63602126e-01 9.98100460e-01 -1.83585688e-01 -3.42571616e-01
2.51550287e-01 -1.84953272e-01 6.85943604e-01 1.67461729e+00
4.87213194e-01 -8.72270018e-02 -8.29212442e-02 6.50026977e-01
1.72786426e-03 2.32855022e-01 -4.27139163e-01 3.21539044e-01
7.48034775e-01 9.00748610e-01 -2.72770643e-01 -3.51480812e-01
-2.77271777e-01 1.18655062e+00 6.02863014e-01 2.15317711e-01
-1.01469123e+00 -3.95014316e-01 3.52190763e-01 -2.91324943e-01
6.11154437e-01 -3.06209266e-01 -6.34272158e-01 -1.20039415e+00
4.10611361e-01 -1.03164458e+00 2.06461266e-01 -7.32609451e-01
-1.20052445e+00 6.89436615e-01 1.02798574e-01 -1.33460462e+00
-4.28491533e-01 -1.09886713e-01 -7.56702483e-01 7.82416344e-01
-1.53202856e+00 -1.40540850e+00 5.37891462e-02 1.21545643e-02
8.00706208e-01 -2.86796559e-02 8.62474680e-01 2.55167127e-01
-6.64824009e-01 1.08422983e+00 1.54700682e-01 -1.84030905e-01
9.64139640e-01 -1.26901555e+00 5.06439030e-01 1.32722235e+00
2.05357358e-01 1.07395899e+00 1.02713621e+00 -4.65648413e-01
-1.28234386e+00 -7.19469905e-01 1.57606447e+00 -5.25807261e-01
6.76915526e-01 -4.78278667e-01 -6.91414356e-01 6.90870583e-01
9.00112510e-01 -7.48431623e-01 8.86732101e-01 4.10423815e-01
-3.67476046e-01 -8.80184025e-02 -7.10028291e-01 8.15888882e-01
1.41187155e+00 -3.33115518e-01 -9.40922081e-01 1.38347223e-01
6.87973857e-01 -3.43012094e-01 -6.61517799e-01 2.62632787e-01
5.96350491e-01 -1.41525328e+00 9.10035074e-01 -4.21867669e-01
8.67914200e-01 -2.01420754e-01 -2.87541360e-01 -1.49037445e+00
-5.66110194e-01 -8.12094986e-01 -3.12689580e-02 1.62964249e+00
5.30493200e-01 -4.96179104e-01 5.57962835e-01 2.92866707e-01
-5.24519861e-01 -8.17560196e-01 -1.06465900e+00 -6.71051621e-01
2.14850575e-01 -4.16721880e-01 7.17424870e-01 7.58481205e-01
2.67609984e-01 6.69677556e-01 -5.99328637e-01 3.30101959e-02
5.18585682e-01 1.40671611e-01 5.92918575e-01 -6.22848868e-01
-4.78928953e-01 -5.10450721e-01 -4.25984055e-01 -1.39677036e+00
3.22260290e-01 -1.11108816e+00 1.67470366e-01 -1.85051131e+00
3.35021615e-01 -2.58191019e-01 -2.68012494e-01 3.55391026e-01
-5.26399076e-01 1.57754421e-01 3.73179793e-01 1.03071928e-01
-7.90660441e-01 7.90226936e-01 8.52879524e-01 -1.75867379e-01
5.55125475e-02 -4.54366744e-01 -1.00684655e+00 3.68830413e-02
7.28132188e-01 -3.83978724e-01 -5.00740826e-01 -7.39993691e-01
4.45870280e-01 -2.43467689e-01 1.58523709e-01 -9.35443223e-01
4.05475914e-01 2.91225761e-01 6.03825003e-02 -6.04272008e-01
3.16776961e-01 -5.82025945e-01 4.73094955e-02 2.72301435e-01
-7.41588414e-01 2.58104533e-01 1.54580548e-01 6.15973353e-01
-3.80517900e-01 -3.27468872e-01 3.08733195e-01 -3.02323117e-03
-6.83083713e-01 -2.66353875e-01 -4.58511055e-01 -6.12388737e-02
9.70976353e-01 -2.92329162e-01 -3.03037047e-01 -5.28749168e-01
-5.89157283e-01 1.44999295e-01 2.78835088e-01 4.33297545e-01
6.27572894e-01 -1.06967318e+00 -1.09233463e+00 5.18933982e-02
1.33582130e-01 -4.05344844e-01 2.66708821e-01 1.15723479e+00
-4.45079952e-01 6.49994195e-01 6.34088442e-02 -6.45293653e-01
-1.27538550e+00 3.19761634e-01 6.03727885e-02 -7.26583123e-01
-4.86708313e-01 9.54826951e-01 -1.51162937e-01 -3.79123479e-01
3.41207609e-02 -1.58690885e-01 -2.30840176e-01 -5.58768921e-02
4.05015409e-01 2.33590156e-01 4.62284714e-01 -4.30662781e-01
-5.40288985e-01 3.61810744e-01 -5.03459156e-01 -2.15904027e-01
1.32945991e+00 -4.26658392e-01 -4.36597735e-01 2.70005912e-01
1.09663761e+00 7.45294765e-02 -6.71442151e-01 7.64741153e-02
7.55290538e-02 -5.22954404e-01 -1.75096735e-01 -1.20186746e+00
-4.56612706e-01 1.02535343e+00 2.52752513e-01 4.18972149e-02
1.22908914e+00 -2.75109082e-01 8.41601253e-01 7.84294009e-02
4.78734583e-01 -7.35888958e-01 -1.44242078e-01 8.03199530e-01
8.62449944e-01 -9.63252008e-01 1.66887324e-02 -5.42498708e-01
-7.15663075e-01 1.10214400e+00 4.68468010e-01 1.52094528e-01
2.31924772e-01 1.65003851e-01 -2.24104434e-01 2.40086287e-01
-1.08564544e+00 3.50284055e-02 1.97769642e-01 4.68451828e-01
8.28444660e-01 -1.19227082e-01 -8.54995668e-01 6.60784364e-01
-4.96641457e-01 -7.55819827e-02 4.66453135e-01 9.27405894e-01
-1.64232180e-01 -1.76881826e+00 -1.55556202e-01 3.10840726e-01
-3.31044823e-01 -8.30202520e-01 -5.37185490e-01 6.04754210e-01
-1.96156129e-01 1.20536745e+00 -1.67813942e-01 -5.25384545e-01
3.09150159e-01 3.45946550e-01 6.97057307e-01 -4.71935511e-01
-9.67382967e-01 -2.32528016e-01 5.09108365e-01 -4.60716724e-01
-4.84853804e-01 -8.71566594e-01 -8.12828600e-01 -3.10667455e-01
-3.36003095e-01 2.04976857e-01 4.69996870e-01 1.17407501e+00
7.55532801e-01 7.01269865e-01 5.70552707e-01 -8.48434269e-01
-5.88812053e-01 -1.24086678e+00 3.18060480e-02 3.94099534e-01
1.07326925e-01 -4.74786162e-01 -2.46232212e-01 -9.25675686e-03] | [11.680402755737305, 9.247743606567383] |
19ccc995-f156-4051-bd6a-abd9b605ff15 | multi-task-recurrent-convolutional-network | 1907.06099 | null | https://arxiv.org/abs/1907.06099v1 | https://arxiv.org/pdf/1907.06099v1.pdf | Multi-Task Recurrent Convolutional Network with Correlation Loss for Surgical Video Analysis | Surgical tool presence detection and surgical phase recognition are two fundamental yet challenging tasks in surgical video analysis and also very essential components in various applications in modern operating rooms. While these two analysis tasks are highly correlated in clinical practice as the surgical process is well-defined, most previous methods tackled them separately, without making full use of their relatedness. In this paper, we present a novel method by developing a multi-task recurrent convolutional network with correlation loss (MTRCNet-CL) to exploit their relatedness to simultaneously boost the performance of both tasks. Specifically, our proposed MTRCNet-CL model has an end-to-end architecture with two branches, which share earlier feature encoders to extract general visual features while holding respective higher layers targeting for specific tasks. Given that temporal information is crucial for phase recognition, long-short term memory (LSTM) is explored to model the sequential dependencies in the phase recognition branch. More importantly, a novel and effective correlation loss is designed to model the relatedness between tool presence and phase identification of each video frame, by minimizing the divergence of predictions from the two branches. Mutually leveraging both low-level feature sharing and high-level prediction correlating, our MTRCNet-CL method can encourage the interactions between the two tasks to a large extent, and hence can bring about benefits to each other. Extensive experiments on a large surgical video dataset (Cholec80) demonstrate outstanding performance of our proposed method, consistently exceeding the state-of-the-art methods by a large margin (e.g., 89.1% v.s. 81.0% for the mAP in tool presence detection and 87.4% v.s. 84.5% for F1 score in phase recognition). The code can be found on our project website. | ['Pheng-Ann Heng', 'Chi-Wing Fu', 'Qi Dou', 'Hao Chen', 'Huaxia Li', 'Yueming Jin', 'Jing Qin'] | 2019-07-13 | null | null | null | null | ['surgical-tool-detection', 'surgical-phase-recognition'] | ['computer-vision', 'computer-vision'] | [ 3.96551609e-01 -2.31560841e-02 -4.56814647e-01 -1.82072192e-01
-8.64527047e-01 -2.40039691e-01 4.05304879e-01 4.91930917e-02
-6.42314076e-01 1.62194744e-01 1.47790059e-01 -2.54807234e-01
-3.95214766e-01 -2.42769197e-01 -5.84592164e-01 -8.06612730e-01
-3.47468108e-01 -2.24339336e-01 2.34101608e-01 -4.96115461e-02
2.92347491e-01 2.76445866e-01 -1.16595793e+00 5.55551112e-01
6.92363501e-01 1.20953381e+00 5.30740440e-01 7.13620305e-01
1.75863937e-01 1.19532382e+00 -1.89549446e-01 -1.49245426e-01
6.91063032e-02 -3.57111782e-01 -6.60113752e-01 -1.17256947e-01
-2.13552592e-03 -2.01412410e-01 -6.57638133e-01 8.61531079e-01
6.02671802e-01 -9.40974653e-02 3.35198224e-01 -8.11057210e-01
-1.12675972e-01 5.52878976e-01 -7.21999586e-01 4.34655249e-01
4.51757386e-02 2.20818713e-01 8.35804939e-01 -6.37444019e-01
3.69082481e-01 5.55137098e-01 6.08576655e-01 3.96272629e-01
-8.70084286e-01 -7.76598096e-01 3.52332890e-01 1.71106294e-01
-1.17414570e+00 -2.88411587e-01 8.26687872e-01 -5.87713718e-01
7.85892904e-01 1.01053342e-01 7.77706921e-01 1.25242376e+00
8.66007507e-01 9.97216225e-01 8.33344936e-01 -8.67085084e-02
-2.25981474e-01 -6.93881065e-02 4.56344709e-02 9.56692815e-01
-1.56159207e-01 2.95486003e-01 -6.92023754e-01 2.51490891e-01
9.12445247e-01 4.96668935e-01 -4.86571491e-01 -4.12543982e-01
-1.61628044e+00 5.42870104e-01 7.63423085e-01 5.19810081e-01
-4.62690026e-01 3.26863527e-01 7.51617372e-01 1.53357238e-01
9.41610113e-02 4.18008536e-01 -2.56933182e-01 -3.24219733e-01
-7.36785531e-01 -2.83089459e-01 5.35110533e-01 9.24017847e-01
2.90267527e-01 -1.75732046e-01 -5.62292337e-01 7.74564445e-01
2.16711536e-01 1.71341039e-02 7.62461007e-01 -5.41000307e-01
3.34309697e-01 5.47006965e-01 -2.58295476e-01 -8.86385322e-01
-6.89468622e-01 -9.94075358e-01 -1.00199842e+00 -9.47471634e-02
3.43198106e-02 3.23102251e-02 -9.04236436e-01 1.80438161e+00
-7.41283670e-02 3.07385594e-01 -7.51012936e-02 7.90461898e-01
8.13559830e-01 1.70497954e-01 2.14195251e-01 -2.94756144e-01
1.42216182e+00 -1.14435339e+00 -8.18892360e-01 -5.57545364e-01
9.63337183e-01 -7.15107143e-01 6.52257025e-01 2.69587636e-01
-9.04782414e-01 -5.36523998e-01 -1.19174051e+00 3.65376659e-02
1.68257043e-01 5.08292198e-01 7.23496020e-01 1.50888100e-01
-8.51082087e-01 7.13520885e-01 -1.22061253e+00 -4.88347001e-02
5.10963380e-01 7.07513571e-01 -3.35675389e-01 -5.16259819e-02
-9.49893534e-01 7.97961771e-01 1.56760663e-01 6.09819889e-01
-1.04967833e+00 -5.71079493e-01 -8.73061597e-01 -1.11491688e-01
4.59097594e-01 -6.31367743e-01 1.36635530e+00 -9.10094202e-01
-1.38288760e+00 8.69300604e-01 -2.21527182e-02 -4.27198291e-01
5.09230018e-01 -4.19468850e-01 -2.39933565e-01 1.50910631e-01
-2.00783499e-02 5.25776863e-01 7.13888407e-01 -8.84385169e-01
-7.26405144e-01 -3.19416642e-01 -4.65511754e-02 3.35002840e-01
-4.03960109e-01 -1.02876835e-01 -7.24814355e-01 -7.23784029e-01
1.50992826e-01 -1.03380191e+00 -4.41721678e-01 2.28164285e-01
-2.03787148e-01 4.39828709e-02 4.72931445e-01 -7.51811922e-01
1.35924518e+00 -2.35211587e+00 2.15086088e-01 -4.73187640e-02
4.92402315e-01 3.53929937e-01 -7.07224309e-02 2.42624238e-01
-1.01519465e-01 -1.53595582e-01 -1.14869922e-01 -3.10321480e-01
-4.46532398e-01 2.03911122e-02 4.58250679e-02 7.42145956e-01
2.61153400e-01 8.74779105e-01 -1.00790441e+00 -4.23635215e-01
2.70088494e-01 5.33092439e-01 -4.37152356e-01 3.74549747e-01
3.10883164e-01 5.41016996e-01 -4.83641207e-01 6.14719629e-01
1.70975223e-01 -3.37965041e-01 3.72145087e-01 -3.38688880e-01
-3.40936519e-02 1.45136431e-01 -5.77654123e-01 2.20402145e+00
-8.17815125e-01 6.49282694e-01 -4.27677259e-02 -1.05824041e+00
7.27722526e-01 5.17189682e-01 8.28739703e-01 -8.13094258e-01
4.47455943e-01 4.38728541e-01 5.75330615e-01 -5.35308003e-01
1.35982215e-01 -1.41656727e-01 -7.75571689e-02 -2.59758737e-02
1.74652964e-01 4.75107312e-01 -4.59487922e-02 6.27685711e-03
1.30927980e+00 9.90730524e-02 4.56984252e-01 -1.77363411e-01
5.09038508e-01 -2.77618438e-01 7.34903276e-01 6.42260313e-01
-5.90195477e-01 5.53133309e-01 5.59233546e-01 -3.56784195e-01
-5.15021920e-01 -8.32172334e-01 7.12876841e-02 8.22991073e-01
3.50137681e-01 -2.52825767e-01 -3.53003204e-01 -9.36309814e-01
-2.83347160e-01 2.41185382e-01 -9.86248612e-01 -7.95681417e-01
-8.70530605e-01 -4.03716356e-01 3.35099220e-01 8.68318498e-01
1.68057144e-01 -1.06787419e+00 -8.61891627e-01 3.75802279e-01
-1.10947765e-01 -1.37946796e+00 -7.14242399e-01 5.50608337e-01
-8.75447929e-01 -1.20314515e+00 -8.35333765e-01 -1.06284153e+00
5.50659001e-01 5.09835303e-01 8.52323771e-01 1.27629191e-01
-5.03960192e-01 1.09571502e-01 -4.01275009e-01 -1.93998501e-01
-1.67513430e-01 1.69585556e-01 -2.17954531e-01 8.71825293e-02
2.23183215e-01 -4.97848839e-01 -8.88085485e-01 3.24737519e-01
-8.12379956e-01 4.64935213e-01 1.16486228e+00 1.24322045e+00
4.11935508e-01 -4.76113379e-01 4.55983579e-01 -6.84629560e-01
9.99440625e-02 -4.52590913e-01 -3.16884607e-01 1.44254819e-01
-3.91440719e-01 4.16524671e-02 6.39483452e-01 -5.15727580e-01
-7.04767168e-01 2.43800372e-01 -1.35985285e-01 -8.16507459e-01
2.21619472e-01 6.08777583e-01 1.18322819e-01 -1.06683120e-01
2.01084092e-01 4.01586801e-01 2.07970381e-01 -7.90785402e-02
-1.92693144e-01 5.15608490e-01 5.08443594e-01 -4.44081612e-02
5.20310044e-01 3.12137842e-01 -1.07902423e-01 -4.59009469e-01
-1.00740707e+00 -8.30697536e-01 -4.35786128e-01 -2.93864667e-01
9.63378906e-01 -1.08157301e+00 -1.02026415e+00 4.77997750e-01
-9.98110712e-01 -2.55513608e-01 -2.62198430e-02 9.12055790e-01
-6.14667356e-01 2.00944811e-01 -6.32455468e-01 -5.25638402e-01
-6.26868844e-01 -1.58154809e+00 1.07207298e+00 2.36045748e-01
-2.83360537e-02 -9.68324125e-01 -1.09271049e-01 2.71995157e-01
3.30367833e-01 3.11670125e-01 6.34050727e-01 -7.16673434e-01
-4.70104814e-01 -3.33513677e-01 -2.30446324e-01 3.47397178e-01
3.21635127e-01 -2.71495074e-01 -8.62099767e-01 -4.92391497e-01
1.87363625e-01 -1.21778004e-01 1.01188946e+00 5.10027766e-01
1.36603713e+00 4.60747294e-02 -6.30423546e-01 9.11622107e-01
1.31940305e+00 4.33489263e-01 4.94373024e-01 3.31937045e-01
7.80820191e-01 4.51251566e-01 8.05376351e-01 4.29787099e-01
2.18660370e-01 7.49728501e-01 5.74462414e-01 -2.98109978e-01
-1.75049976e-01 -9.29116160e-02 4.61721450e-01 1.09935212e+00
7.54519328e-02 9.17405337e-02 -1.01757193e+00 5.01575887e-01
-2.00471401e+00 -7.74092615e-01 1.00527659e-01 2.16265440e+00
7.67347872e-01 2.70266712e-01 -2.63575107e-01 2.54673380e-02
5.84485710e-01 3.84116411e-01 -5.03566682e-01 -1.85336858e-01
3.77009511e-01 9.46604833e-02 5.94006777e-01 1.19871311e-01
-1.38960052e+00 6.20730817e-01 5.01503086e+00 8.10483277e-01
-1.26462412e+00 2.03024335e-02 5.41992247e-01 -4.61953133e-01
2.66771495e-01 -1.62313715e-01 -5.24603426e-01 4.02871251e-01
8.18780005e-01 5.01195714e-02 9.38615277e-02 6.47645891e-01
2.45435074e-01 -1.11827189e-02 -1.36206508e+00 1.12829649e+00
5.09778224e-02 -1.28376710e+00 -3.37151855e-01 2.74209362e-02
3.51881951e-01 1.33868698e-02 1.47619233e-01 3.67738605e-01
-7.51234144e-02 -1.12418616e+00 4.96075183e-01 5.44922292e-01
8.84652853e-01 -7.21631348e-01 9.97311831e-01 3.30397993e-01
-1.39205098e+00 -2.41746053e-01 9.52650383e-02 2.37139851e-01
5.86712882e-02 3.81263226e-01 -6.86798930e-01 8.10332298e-01
5.59366226e-01 1.13737822e+00 -3.47502470e-01 1.06608784e+00
-2.02717230e-01 3.08597088e-01 2.55772322e-02 1.06178351e-01
4.65824008e-01 1.50087923e-01 4.23413515e-01 1.34492862e+00
2.18415856e-01 -2.12016046e-01 2.57830650e-01 4.27921683e-01
-4.51688357e-02 -7.61162564e-02 -4.33610618e-01 -2.42300238e-02
5.97447939e-02 1.33467829e+00 -5.66536844e-01 -3.59376706e-02
-5.49686372e-01 9.48777258e-01 4.47894216e-01 5.87152280e-02
-1.10650384e+00 -5.24780571e-01 6.35772288e-01 -2.30360478e-02
3.15773785e-01 -9.61152390e-02 -2.16395870e-01 -1.06116521e+00
1.94912776e-01 -9.07144666e-01 3.31623465e-01 -2.68269390e-01
-8.70739937e-01 8.54490161e-01 -4.94285762e-01 -1.88826060e+00
-1.82706431e-01 -6.81416810e-01 -4.95896608e-01 7.54259646e-01
-1.78731382e+00 -1.25798488e+00 -5.67694545e-01 5.51648140e-01
7.42144942e-01 2.26202011e-02 8.39272201e-01 3.61547053e-01
-8.06041360e-01 8.00041020e-01 -2.09236909e-02 3.86600912e-01
8.36460054e-01 -9.58270967e-01 1.03341853e-02 9.22257125e-01
-1.39382347e-01 5.68052649e-01 2.30597481e-01 -3.88114303e-01
-1.54554677e+00 -1.11238563e+00 4.65308666e-01 -8.56062323e-02
6.66458249e-01 -3.12214196e-01 -7.30156958e-01 6.04700029e-01
5.76749854e-02 2.82813549e-01 8.69784236e-01 1.81426480e-02
-3.04258049e-01 -1.58596277e-01 -5.03764272e-01 5.72434425e-01
1.12004030e+00 -5.36583960e-01 -3.31398845e-01 3.49325895e-01
5.82518995e-01 -6.79283917e-01 -1.07014978e+00 8.42670381e-01
6.63958848e-01 -9.31563497e-01 7.17620015e-01 -3.67430508e-01
8.38521242e-01 -1.33397896e-02 4.93854005e-03 -9.92375076e-01
-4.14388031e-01 -5.84339976e-01 -2.12537214e-01 7.03571498e-01
2.84585565e-01 -3.73809397e-01 6.17928624e-01 2.17230439e-01
-6.54628694e-01 -1.39900410e+00 -8.72887909e-01 -6.80708051e-01
-1.46414116e-01 -4.11005497e-01 -1.85837954e-01 6.31184280e-01
1.07766069e-01 3.08174372e-01 -5.65353751e-01 2.11410388e-01
2.92297840e-01 1.33947819e-01 3.72781634e-01 -8.29024673e-01
-4.61299717e-01 -6.78195655e-01 -5.31761646e-01 -1.11811399e+00
-4.05170955e-02 -9.85103846e-01 2.64957547e-01 -1.55523849e+00
4.73309189e-01 -3.24522942e-01 -8.80405426e-01 5.27411222e-01
-4.23257858e-01 1.11916149e-02 1.61128432e-01 4.29063171e-01
-6.36629403e-01 6.75762236e-01 1.37387884e+00 -3.52337025e-02
-2.35307738e-01 1.62018761e-01 -7.00616121e-01 7.26186097e-01
5.37923753e-01 -5.92815936e-01 -3.15312415e-01 -5.01565635e-01
-1.80211306e-01 3.88247520e-01 2.14234009e-01 -1.09804666e+00
5.17159343e-01 8.96216482e-02 3.40703338e-01 -4.12372261e-01
3.61402512e-01 -7.14630485e-01 -4.72893864e-02 9.85939682e-01
-3.89319241e-01 -1.11387677e-01 2.46732980e-01 6.77649438e-01
-5.65661013e-01 1.08416826e-01 9.59843040e-01 -8.39311257e-03
-7.63316870e-01 6.05931997e-01 -3.53027523e-01 -1.36012852e-01
1.30307412e+00 -2.44067460e-01 -1.38184920e-01 -7.25208670e-02
-5.62367141e-01 2.93930948e-01 1.54389769e-01 6.72424912e-01
8.84861946e-01 -1.10133612e+00 -6.36068225e-01 1.92728803e-01
3.03106219e-01 -1.91911664e-02 6.94064140e-01 1.61646867e+00
-4.39783812e-01 4.75816429e-01 -1.59859464e-01 -7.71504402e-01
-1.19075811e+00 4.95622963e-01 4.91831005e-01 -8.28720629e-01
-8.05637598e-01 8.95745873e-01 6.63634539e-01 -9.02166814e-02
5.37832797e-01 -2.43903011e-01 -3.32451671e-01 -3.57907414e-02
5.47018409e-01 -1.98140085e-01 1.52110115e-01 -2.86332548e-01
-6.52097046e-01 5.92378974e-01 -6.14385545e-01 3.88792127e-01
1.24591613e+00 1.84595853e-01 1.74990997e-01 3.61907870e-01
1.49959958e+00 -4.02013242e-01 -1.45634294e+00 -3.84871960e-01
-3.57042551e-02 -2.82594085e-01 2.91132718e-01 -7.79968917e-01
-1.41504550e+00 1.02547324e+00 7.88182914e-01 -3.87099802e-01
1.41173244e+00 -1.53871849e-01 8.60600293e-01 4.53236699e-02
2.29833856e-01 -6.62313282e-01 3.46357554e-01 4.59017307e-01
7.55371749e-01 -1.36539257e+00 -1.14790067e-01 -4.11670864e-01
-8.14793229e-01 1.08076262e+00 7.44471252e-01 -1.73259601e-01
5.36231399e-01 3.96903187e-01 8.34932998e-02 -2.11364031e-01
-9.59904790e-01 -3.63021232e-02 4.41159517e-01 2.68718898e-01
7.35539019e-01 -1.43189192e-01 -3.09778899e-01 9.03906763e-01
2.92745262e-01 1.00519665e-01 3.42952460e-01 1.16530740e+00
-1.63507268e-01 -7.09702373e-01 2.64304489e-01 6.55542314e-01
-6.98285341e-01 -1.62745282e-01 1.89416334e-01 7.70247757e-01
-2.08550945e-01 7.10972011e-01 -5.59350699e-02 -7.66230106e-01
3.52010518e-01 -4.30486262e-01 2.59266526e-01 -5.44620335e-01
-9.90359306e-01 2.38497436e-01 3.77511829e-02 -1.00291014e+00
-3.61132205e-01 -4.03822929e-01 -1.17243266e+00 1.00174181e-01
-3.49926859e-01 -1.27720416e-01 4.98036206e-01 8.05383801e-01
3.14687580e-01 1.25736892e+00 7.80598760e-01 -7.57631242e-01
-6.41460061e-01 -8.38010311e-01 -4.57971901e-01 1.19456112e-01
5.33138156e-01 -7.25016356e-01 -1.93558201e-01 -2.35391065e-01] | [14.166297912597656, -3.2553646564483643] |
d1926d88-d142-45be-8533-ba5d8319f587 | the-effect-of-changing-training-data-on-a | null | null | https://iopscience.iop.org/article/10.1088/1757-899X/568/1/012087 | https://iopscience.iop.org/article/10.1088/1757-899X/568/1/012087/pdf | The effect of changing training data on a fixed deep learning detection model | Within the lack of accurate data, for some computer vision applications, researchers usually use other pictures collected from different sources for the training. To know the effect of these added data, we compare the detection results of a customized dataset of objects, using the same detection model, while changing the training data fed into the network. For our work, we run the detection on images captured by the Microsoft Kinect sensor after training the network on different combinations of training data. The first part of the training data is captured by the Kinect itself, and the second is collected from several sources from the internet, referred to as collected images. We then change the distribution of these images between training and validation to feed them into the fixed training model. The results prove that this distribution of data can considerably affect training and detection results under the same model parameters. In addition, mixing the captured images with other collected ones can improve these results. | ['Ammar Alsabbagh1 and Husi Géza', 'Aram Nasser'] | 2019-09-17 | null | null | null | annual-session-of-scientific-papers-imt | ['object-detection-in-indoor-scenes'] | ['computer-vision'] | [ 5.55315316e-02 -3.39942425e-01 -2.45129943e-01 -3.58470261e-01
1.84975415e-01 -4.73879278e-01 2.11677223e-01 -2.16351122e-01
-9.90420640e-01 3.13849449e-01 -4.50605065e-01 2.57924646e-02
2.89707154e-01 -1.03938603e+00 -7.18258142e-01 -9.16577876e-01
1.43038362e-01 2.50565916e-01 8.07572722e-01 2.55278260e-01
2.44724810e-01 7.13619232e-01 -1.89326441e+00 9.93856117e-02
4.09391105e-01 8.89613032e-01 6.15113735e-01 5.13761163e-01
-3.71647120e-01 3.78641158e-01 -7.92430520e-01 -1.52863622e-01
8.53930533e-01 -1.22236937e-01 -1.51206449e-01 3.11356038e-01
2.44111821e-01 -7.89108753e-01 -2.82626957e-01 1.15805864e+00
5.32892168e-01 -6.39176965e-02 1.36762396e-01 -1.29005098e+00
-2.99035579e-01 4.49645042e-01 -6.07292354e-01 1.71357617e-01
2.38395408e-01 4.17435855e-01 1.53791919e-01 -7.23697841e-01
6.35832608e-01 1.06933820e+00 6.34045720e-01 8.83908331e-01
-8.43995452e-01 -8.83327961e-01 -2.06019670e-01 2.92526424e-01
-1.30640233e+00 -1.03178360e-01 1.09731138e+00 -4.04593378e-01
4.93591130e-02 8.93052891e-02 8.66107762e-01 1.33014941e+00
-1.09525867e-01 4.36303258e-01 1.23119760e+00 -5.19005954e-01
2.45653912e-01 4.79661226e-01 2.54229337e-01 3.97873670e-01
7.66705990e-01 2.71896213e-01 -1.37315124e-01 5.82102500e-02
9.38028336e-01 4.69531775e-01 -4.72494692e-01 -3.94561261e-01
-8.26631904e-01 7.69873679e-01 3.52441579e-01 5.34433126e-01
-3.03114861e-01 -2.32981220e-01 1.22828402e-01 2.11000443e-01
-1.63221464e-01 -6.59708157e-02 -5.81278682e-01 1.53087229e-01
-7.18645275e-01 -2.28686690e-01 7.14195907e-01 6.64105892e-01
1.12904131e+00 -2.72845834e-01 3.09788734e-01 7.86705017e-01
5.40166020e-01 8.58357370e-01 9.59891796e-01 -5.78467548e-01
5.34803152e-01 9.28204179e-01 1.44302934e-01 -1.28418338e+00
-4.36829329e-01 -8.78551006e-02 -5.07720172e-01 4.16085213e-01
8.96718740e-01 -4.22232419e-01 -9.33113456e-01 1.28950381e+00
5.35895467e-01 2.02446774e-01 -1.79584995e-01 1.29849768e+00
1.06463361e+00 4.63355392e-01 -1.72448367e-01 -2.30225563e-01
1.00722551e+00 -7.91055083e-01 -6.35470986e-01 -2.81268895e-01
3.67947280e-01 -6.82889879e-01 1.06620824e+00 6.15432322e-01
-3.25611830e-01 -8.06094646e-01 -1.16311681e+00 5.12826264e-01
-5.93538642e-01 7.63344824e-01 3.66653264e-01 8.56294692e-01
-6.68797672e-01 7.02679038e-01 -9.19778049e-01 -6.89335883e-01
1.48108572e-01 2.99832582e-01 -4.00760561e-01 -2.08689064e-01
-9.37751412e-01 7.73601532e-01 5.83548188e-01 1.21130571e-01
-7.76867568e-01 -1.77631333e-01 -3.49642545e-01 -4.66984957e-01
3.80618006e-01 -1.31056949e-01 8.38449895e-01 -1.11497843e+00
-1.40476406e+00 9.33665395e-01 4.00036842e-01 -1.37982413e-01
7.84357846e-01 -2.06768408e-01 -3.12196285e-01 1.10258020e-01
-1.73898623e-03 4.82314080e-01 8.96220028e-01 -1.51809227e+00
-5.21751881e-01 -6.42264485e-01 1.87243134e-01 -1.29255980e-01
-5.23442745e-01 -2.47247629e-02 -8.06008935e-01 -1.26790658e-01
2.09604472e-01 -9.42945063e-01 -1.07435830e-01 1.57591254e-01
-2.86792368e-01 1.66973740e-01 1.22643530e+00 -3.88748378e-01
8.72300327e-01 -2.16231275e+00 -1.54733494e-01 4.08044845e-01
-6.77025691e-02 4.57697302e-01 -8.50793570e-02 2.54890889e-01
-8.24913606e-02 1.06547542e-01 1.25685692e-01 -3.63295138e-01
-6.20218098e-01 7.42235959e-01 2.48584747e-01 3.73862356e-01
-2.09567562e-01 1.81958139e-01 -7.67965257e-01 -6.60662055e-01
5.55461943e-01 1.92691520e-01 -1.71086401e-01 5.32235265e-01
1.10621259e-01 6.72292531e-01 -6.05369329e-01 5.30007243e-01
9.70066011e-01 5.27739935e-02 3.81202102e-02 -4.10582513e-01
-4.60584983e-02 -3.19749475e-01 -1.49247110e+00 1.40478718e+00
-2.25624919e-01 5.87190986e-01 7.29509965e-02 -1.00709403e+00
1.05551732e+00 6.61457404e-02 7.24618673e-01 -5.16480684e-01
6.42097294e-01 -3.10791750e-02 1.07810553e-02 -1.18602705e+00
1.78179577e-01 4.63644154e-02 5.52050531e-01 4.38667685e-01
4.07839119e-02 2.14221090e-01 2.14108318e-01 -2.62161881e-01
8.46515000e-01 -3.33194174e-02 -9.03773308e-02 4.56601977e-01
3.84450495e-01 1.27505213e-01 5.50160289e-01 9.87286747e-01
-3.17650646e-01 7.39087045e-01 1.64696686e-02 -5.41974843e-01
-1.11288822e+00 -8.03247213e-01 -2.67469049e-01 8.03878307e-01
4.60059941e-01 -6.91764653e-02 -7.22765565e-01 -7.24569798e-01
5.29045537e-02 1.80205569e-01 -6.50424480e-01 -2.14747429e-01
-2.89461225e-01 -1.04010427e+00 7.42816567e-01 3.20928812e-01
8.88754129e-01 -1.03536916e+00 -1.22937393e+00 -1.64693892e-01
-1.87723506e-02 -1.10116863e+00 8.43886212e-02 8.41395482e-02
-1.00092769e+00 -1.57935381e+00 -5.59818268e-01 -4.76943463e-01
7.70416439e-01 5.05038977e-01 6.55419827e-01 3.39620054e-01
-2.42346227e-01 5.01327574e-01 -7.59961426e-01 -6.26420617e-01
-4.42243546e-01 -1.29257008e-01 1.20333150e-01 3.35128397e-01
4.79396641e-01 -5.55155218e-01 -4.36946303e-01 5.32067478e-01
-1.17454314e+00 -1.88592449e-01 5.62789977e-01 4.91919935e-01
2.16533497e-01 3.15032691e-01 6.39745072e-02 -5.49355984e-01
3.80278468e-01 -3.64237577e-01 -7.30321705e-01 1.12233564e-01
-4.72604662e-01 -4.18370664e-02 4.95983154e-01 -9.51219618e-01
-6.78959906e-01 4.20192063e-01 1.71265244e-01 -1.09008670e+00
-4.43201989e-01 2.64359295e-01 -1.99696586e-01 -2.46368527e-01
8.59121978e-01 1.91571653e-01 3.49090755e-01 -6.90588892e-01
-5.38176857e-02 7.95741677e-01 2.63245136e-01 -1.62252486e-01
8.54169607e-01 5.80993831e-01 -4.34300214e-01 -7.23136425e-01
-4.46566790e-01 -5.43096304e-01 -6.72518194e-01 -6.47447944e-01
9.56445098e-01 -4.98218328e-01 -4.01432693e-01 9.63848948e-01
-1.22375512e+00 -9.59450230e-02 -1.66307643e-01 7.67151535e-01
1.47928223e-01 3.06281120e-01 -3.64983678e-01 -8.51849258e-01
-3.73062268e-02 -1.15160382e+00 7.74859309e-01 5.53735256e-01
5.99118829e-01 -8.10569406e-01 1.75188899e-01 2.07031742e-01
1.97531804e-01 2.02192590e-01 1.31185338e-01 -7.56264746e-01
-4.15337175e-01 -5.76633811e-01 -9.95905921e-02 6.65339649e-01
5.31496882e-01 3.57691109e-01 -1.05505323e+00 -2.67233904e-02
5.65961421e-01 2.51918044e-02 6.72178447e-01 1.41029999e-01
1.01893568e+00 -2.69758463e-01 -4.66773242e-01 3.97652328e-01
1.64575028e+00 2.94223458e-01 6.75042868e-01 6.43226266e-01
7.42736995e-01 4.93323028e-01 6.06715143e-01 3.97645116e-01
-9.52941775e-02 5.06870329e-01 8.15081060e-01 -1.18131422e-01
1.78354725e-01 -5.64451106e-02 4.69529182e-01 5.80769062e-01
-3.83425951e-01 1.91269413e-01 -9.63313580e-01 3.08671892e-01
-1.67500615e+00 -1.06619656e+00 -2.48399302e-01 2.46047568e+00
7.08105564e-01 1.07612506e-01 2.29927436e-01 4.32675987e-01
1.13584423e+00 -7.24871233e-02 -5.71089268e-01 1.81620538e-01
-2.38827914e-02 -2.24070568e-02 9.83265400e-01 7.40222707e-02
-1.06285143e+00 4.19359535e-01 6.47944593e+00 5.83899856e-01
-1.88334775e+00 1.46383762e-01 2.70553052e-01 -1.69219717e-01
2.15974972e-01 -1.88335299e-01 -5.44264078e-01 9.33415473e-01
5.96255898e-01 1.44728884e-01 4.83554006e-01 1.25767100e+00
4.97326463e-01 -5.78944802e-01 -9.32371914e-01 1.25578535e+00
2.64890373e-01 -7.33232439e-01 -1.06346741e-01 -1.40278239e-03
4.32440639e-01 1.75351143e-01 -4.60582495e-01 -8.49627033e-02
1.34085920e-02 -6.42284572e-01 4.29683685e-01 5.47270417e-01
2.95343429e-01 -2.20132992e-01 1.10134280e+00 7.44299769e-01
-8.14594269e-01 -1.17589951e-01 -6.05348408e-01 -1.20914564e-01
-2.96795458e-01 7.38041103e-01 -9.73668635e-01 5.62567532e-01
8.95772457e-01 5.79926431e-01 -1.01432538e+00 1.29433620e+00
-1.86780766e-01 6.87134564e-01 -6.00796163e-01 -1.21676877e-01
-2.33985841e-01 -4.55155373e-01 4.29687709e-01 7.75518358e-01
2.97111511e-01 -3.04567277e-01 2.04023108e-01 9.15210783e-01
1.68360919e-01 -1.20819904e-01 -9.63811636e-01 3.56269956e-01
4.71008807e-01 1.42527306e+00 -8.22014332e-01 -3.94711465e-01
-4.78409052e-01 6.29251897e-01 -1.40706599e-01 2.80275583e-01
-1.01949990e+00 8.01402610e-03 2.80331429e-02 2.65262812e-01
1.46561459e-01 -1.95664078e-01 -2.82111645e-01 -1.16828716e+00
1.72712266e-01 -5.58578432e-01 9.47904587e-02 -1.04570353e+00
-1.35882330e+00 2.35826373e-01 2.04175666e-01 -1.68238783e+00
-4.50609401e-02 -9.45259154e-01 -7.14933753e-01 5.60745001e-01
-1.06095803e+00 -7.56597757e-01 -8.33108127e-01 7.26076961e-01
1.72403902e-01 -1.04486167e-01 2.87193418e-01 7.67739356e-01
-8.09755921e-01 3.66859972e-01 1.03728017e-02 6.78669035e-01
7.02973545e-01 -7.69254208e-01 -3.71612579e-01 8.51456106e-01
1.27306819e-01 4.77204353e-01 5.62121987e-01 -5.28762221e-01
-1.30410993e+00 -1.11473393e+00 2.22433936e-02 -4.11374480e-01
3.59547228e-01 1.69797558e-02 -8.63838971e-01 6.49948418e-01
-1.13818623e-01 1.47229895e-01 3.25228781e-01 -3.66297841e-01
1.33449407e-02 -3.39705050e-01 -1.44109082e+00 1.67156145e-01
7.30978608e-01 -8.70112553e-02 -7.89367676e-01 -1.93512309e-02
2.72598982e-01 -4.92530644e-01 -6.19287074e-01 2.55360186e-01
7.53548682e-01 -1.15813577e+00 8.84431779e-01 -2.18496367e-01
3.93411934e-01 -4.50719625e-01 -1.21518955e-01 -1.29364455e+00
1.39877111e-01 3.85324776e-01 3.80475372e-01 1.28913653e+00
1.43026290e-02 -8.39350224e-01 1.18001771e+00 4.98770326e-01
3.61622274e-01 -4.50050592e-01 -8.60000610e-01 -9.02688086e-01
-2.14863986e-01 -4.45980847e-01 4.28585470e-01 9.15533721e-01
-3.20844084e-01 6.69114590e-02 -2.97945529e-01 3.90907943e-01
5.20045817e-01 -3.74612808e-02 1.13723230e+00 -1.36302972e+00
-1.41634166e-01 -4.64989506e-02 -6.07172191e-01 -7.78578937e-01
-4.80583608e-01 -3.84296596e-01 -5.80801964e-02 -1.34455287e+00
7.03177452e-02 -6.59809589e-01 1.18032873e-01 6.38584554e-01
-1.08192764e-01 3.50613981e-01 4.86663491e-01 5.43883204e-01
-2.07104146e-01 2.98157364e-01 1.22394216e+00 -3.15827318e-02
-2.55439579e-01 1.75725460e-01 -5.45394011e-02 1.02799857e+00
8.29285324e-01 -6.39202893e-01 -2.72421271e-01 -4.89251256e-01
-1.87191069e-01 -1.05124414e-01 6.12375617e-01 -1.46642554e+00
3.25101286e-01 -2.31842950e-01 5.96065640e-01 -5.58509231e-01
2.38774449e-01 -1.52556348e+00 3.05002779e-01 6.77417576e-01
2.02378944e-01 -1.13122866e-01 1.08813055e-01 3.79980087e-01
-9.57223400e-02 -7.33591139e-01 7.17147470e-01 -4.04335856e-01
-7.70260334e-01 5.56257414e-03 -1.78908557e-01 -2.94728488e-01
1.20722759e+00 -8.04584682e-01 -2.03347102e-01 -2.02068865e-01
-8.25718582e-01 -7.23282397e-02 6.63652897e-01 4.69931841e-01
5.14249623e-01 -1.30439436e+00 -2.13743329e-01 4.28290248e-01
1.13002993e-02 2.63406605e-01 -9.01476219e-02 6.67848766e-01
-4.84159887e-01 -2.31529489e-01 -5.41916370e-01 -1.00276279e+00
-1.29372835e+00 5.67157507e-01 6.16835952e-01 -1.94029622e-02
-2.67577708e-01 1.75868139e-01 -2.57487595e-01 -6.13585174e-01
2.92499840e-01 -4.96812612e-01 -3.75789315e-01 1.89243093e-01
4.22991723e-01 4.54672307e-01 1.47816837e-01 -3.40393454e-01
-3.80640537e-01 8.48949611e-01 3.35882694e-01 -6.29701093e-02
1.20049310e+00 -7.90285703e-04 5.13779819e-02 7.03932405e-01
8.86748850e-01 1.97241038e-01 -1.19747317e+00 -2.68225316e-02
-2.98067868e-01 -7.10402846e-01 -3.11550409e-01 -5.17250836e-01
-1.79661489e+00 6.45282567e-01 1.20528412e+00 4.76842970e-01
1.10194302e+00 -1.45376071e-01 3.13457996e-01 4.27920550e-01
8.16689789e-01 -1.04461598e+00 1.42908454e-01 1.62010774e-01
4.46379960e-01 -1.44435191e+00 -1.42311491e-02 -4.69273180e-01
-3.59796554e-01 1.18587637e+00 8.83838654e-01 -4.28240478e-01
5.33512592e-01 2.99375862e-01 3.82803380e-01 2.61925091e-03
-1.87372774e-01 -1.88648343e-01 -3.51012021e-01 8.44471812e-01
-1.89646780e-01 -3.00515741e-01 -5.35160422e-01 4.97083038e-01
-1.54579226e-02 4.78835225e-01 7.66227543e-01 8.02857876e-01
-5.29223084e-01 -1.14745760e+00 -1.07748759e+00 5.03118753e-01
-2.46660441e-01 3.62072229e-01 -4.97714609e-01 9.97496545e-01
7.75798619e-01 1.04973149e+00 8.10352415e-02 -7.36151099e-01
4.13318098e-01 -3.59909497e-02 2.29802832e-01 -5.26054382e-01
-4.55465287e-01 -3.19967300e-01 -2.61017859e-01 -4.28468376e-01
-8.54678094e-01 -5.75525343e-01 -1.25677836e+00 -2.52501547e-01
-5.72160959e-01 2.43124831e-02 8.75233352e-01 8.20941925e-01
1.24968946e-01 4.64680046e-01 8.27057540e-01 -1.29367650e+00
-5.62266052e-01 -1.19794285e+00 -4.42065418e-01 7.77118266e-01
1.86720312e-01 -9.45287943e-01 -7.16590047e-01 3.35587323e-01] | [8.193036079406738, -0.8541222214698792] |
1caa044e-3378-46a0-9a91-bfc39fffb353 | practical-and-ethical-challenges-of-large | 2303.13379 | null | https://arxiv.org/abs/2303.13379v1 | https://arxiv.org/pdf/2303.13379v1.pdf | Practical and Ethical Challenges of Large Language Models in Education: A Systematic Literature Review | Educational technology innovations that have been developed based on large language models (LLMs) have shown the potential to automate the laborious process of generating and analysing textual content. While various innovations have been developed to automate a range of educational tasks (e.g., question generation, feedback provision, and essay grading), there are concerns regarding the practicality and ethicality of these innovations. Such concerns may hinder future research and the adoption of LLMs-based innovations in authentic educational contexts. To address this, we conducted a systematic literature review of 118 peer-reviewed papers published since 2017 to pinpoint the current state of research on using LLMs to automate and support educational tasks. The practical and ethical challenges of LLMs-based innovations were also identified by assessing their technological readiness, model performance, replicability, system transparency, privacy, equality, and beneficence. The findings were summarised into three recommendations for future studies, including updating existing innovations with state-of-the-art models (e.g., GPT-3), embracing the initiative of open-sourcing models/systems, and adopting a human-centred approach throughout the developmental process. These recommendations could support future research to develop practical and ethical innovations for supporting diverse educational tasks and benefiting students, teachers, and institutions. | ['Dragan Gašević', 'Yueqiao Jin', 'Xinyu Li', 'Guanliang Chen', 'Roberto Martinez-Maldonado', 'Yuheng Li', 'Linxuan Zhao', 'Lele Sha', 'Lixiang Yan'] | 2023-03-17 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [-8.80224332e-02 4.73661155e-01 -2.74921954e-01 2.41527215e-01
-5.48951864e-01 -1.04403913e+00 6.84522152e-01 7.16068625e-01
-4.96417046e-01 3.49439651e-01 7.70197511e-01 -1.03414762e+00
-4.17190373e-01 -3.27395380e-01 -6.36560142e-01 -8.16780031e-02
9.21474040e-01 -3.85321319e-01 5.97712658e-02 1.99448075e-02
9.88958776e-01 4.02403533e-01 -1.77892244e+00 1.27266407e-01
1.49497998e+00 3.69724393e-01 2.41323691e-02 5.10895193e-01
-8.21130753e-01 1.10360587e+00 -8.81660581e-01 -7.37381577e-01
-9.46060419e-02 -4.18417186e-01 -8.20448101e-01 -1.69022486e-01
6.08139932e-01 -4.97863889e-01 4.18990105e-01 9.32491362e-01
7.72565603e-01 -1.31333336e-01 1.86525717e-01 -1.08209097e+00
-1.41509497e+00 2.97497332e-01 1.09190673e-01 -1.62003443e-01
8.06432903e-01 2.46816888e-01 4.69695777e-01 -6.30052865e-01
6.52498305e-01 9.11491752e-01 5.78065336e-01 3.62477273e-01
-7.08362639e-01 -1.04405332e+00 -1.11134656e-01 6.35894164e-02
-7.73409545e-01 -6.72673464e-01 2.24581927e-01 -1.02943552e+00
8.28617513e-01 5.14881551e-01 1.26739883e+00 7.61435866e-01
7.63556778e-01 1.87134907e-01 1.75678694e+00 -9.01505470e-01
3.66735280e-01 1.02525401e+00 3.65099534e-02 2.85064816e-01
8.34530294e-01 -4.16376680e-01 -5.17632186e-01 1.24803603e-01
5.10854483e-01 -1.73815757e-01 1.84245929e-02 7.42341056e-02
-8.23557258e-01 7.45192230e-01 -2.60964423e-01 5.73589623e-01
-2.79565543e-01 -2.23943576e-01 3.10383141e-01 4.09278333e-01
5.89905620e-01 7.36053705e-01 -3.31192583e-01 -9.45686579e-01
-7.44087994e-01 -8.21273401e-02 1.06457746e+00 8.50784421e-01
7.89475590e-02 -4.07555625e-02 -2.87925899e-01 5.52173436e-01
8.59207153e-01 2.85449237e-01 5.36854208e-01 -1.03393149e+00
4.07541310e-03 9.71062064e-01 1.19007891e-02 -5.90706766e-01
8.03945512e-02 -2.21070960e-01 1.85891747e-01 1.35220706e-01
6.43189698e-02 -4.47020292e-01 -5.67007124e-01 8.99957240e-01
4.78916734e-01 -5.07609397e-02 -1.07279199e-03 3.33655298e-01
1.50742161e+00 4.18289006e-01 5.04032671e-01 -1.31370407e-02
1.20604169e+00 -6.48794413e-01 -1.06867290e+00 -4.74375486e-02
1.01887333e+00 -1.25481451e+00 1.09372079e+00 2.54405111e-01
-1.45646679e+00 -2.65865386e-01 -6.37229919e-01 -1.40651464e-01
-4.82593715e-01 -1.08976729e-01 2.04653323e-01 1.64614713e+00
-1.23276687e+00 1.81985218e-02 -4.30724055e-01 -6.35100543e-01
4.26887333e-01 -1.40830532e-01 -1.48601443e-01 3.01070753e-02
-8.21273565e-01 1.06708467e+00 -5.77026568e-02 -6.93328902e-02
-1.96229994e-01 -1.05651987e+00 -6.40645623e-01 -7.87569433e-02
-5.60560077e-02 -6.46102548e-01 1.38386059e+00 -8.48570645e-01
-1.81185579e+00 4.16291714e-01 3.22004288e-01 4.27747965e-02
2.55488515e-01 -3.92214149e-01 -3.22316945e-01 7.27943704e-03
3.56762320e-01 2.42334515e-01 8.38207304e-02 -9.04968739e-01
-8.03813100e-01 3.58058624e-02 -2.52869591e-04 2.52695233e-02
-9.72023487e-01 7.93887436e-01 7.82070979e-02 -1.53139025e-01
-3.10591251e-01 -6.92431271e-01 -5.91360964e-02 1.93749350e-02
3.67440343e-01 -2.73641706e-01 6.50848389e-01 -8.19247365e-01
1.44462764e+00 -1.79440987e+00 -9.01342630e-01 -2.09910013e-02
2.40431264e-01 7.91650355e-01 -3.66179273e-02 1.32949162e+00
3.89069915e-01 8.79415274e-01 8.87142122e-01 1.39051512e-01
3.86253804e-01 -2.54436433e-01 -5.39073860e-03 2.49924168e-01
-2.88272470e-01 8.98239791e-01 -1.01844192e+00 -3.88826609e-01
3.97550702e-01 6.88816011e-01 -2.55513936e-01 6.23541251e-02
3.47039938e-01 3.53987008e-01 -2.90340126e-01 2.66407520e-01
3.67974639e-01 9.83380526e-03 -1.37321934e-01 6.90764606e-01
-9.80899155e-01 8.34031880e-01 -1.14780402e+00 1.19216096e+00
-8.53992641e-01 9.69240665e-01 1.27437964e-01 -4.22972403e-02
9.34110284e-01 8.32241356e-01 4.63610023e-01 -8.56137633e-01
7.27245510e-02 2.60233968e-01 1.62155688e-01 -1.10757828e+00
3.81292254e-01 2.13065401e-01 3.93132448e-01 7.69675970e-01
-4.72456403e-02 -4.88293946e-01 -2.11892769e-01 2.13251725e-01
8.45702231e-01 4.11553115e-01 8.49234313e-02 -2.00444818e-01
2.39215165e-01 -1.21806726e-01 1.77841723e-01 5.21775603e-01
-3.16606432e-01 -8.32995698e-02 3.02236468e-01 1.83914110e-01
-6.00091636e-01 -3.16876262e-01 -2.17916429e-01 1.04938161e+00
-5.93116105e-01 -8.08216572e-01 -1.01986003e+00 -3.52594554e-01
-2.53269851e-01 1.16433465e+00 -1.30748432e-02 5.18266810e-04
4.82540935e-01 2.49002233e-01 4.07479525e-01 -1.21748159e-02
2.61592716e-01 -9.02075112e-01 -1.28677797e+00 4.94564116e-01
1.42220140e-01 -9.37595844e-01 1.90868881e-02 -6.33847356e-01
-6.20085478e-01 -7.38352537e-01 -4.38270956e-01 -6.99368954e-01
6.19622767e-01 4.31802213e-01 5.11937797e-01 9.90245342e-02
4.25808839e-02 1.14285409e+00 -5.60770690e-01 -1.20148885e+00
-6.52206063e-01 -3.12695980e-01 -3.85903269e-01 -7.39067852e-01
4.84639525e-01 -1.00452431e-01 -5.29328883e-01 -3.20921481e-01
-1.20789373e+00 1.53266877e-01 6.77491844e-01 3.71808469e-01
-1.24768272e-01 -4.77762930e-02 1.09809291e+00 -7.31267393e-01
1.33812249e+00 -6.18711591e-01 -3.24250251e-01 3.45142394e-01
-1.19436419e+00 -6.65934682e-01 2.82698005e-01 -4.47401494e-01
-1.39541900e+00 -7.18252957e-01 -4.04559582e-01 2.42298320e-01
-3.48877579e-01 9.93393838e-01 1.64821312e-01 -7.75113761e-01
4.33842391e-01 -2.76383132e-01 2.06268519e-01 -1.56528249e-01
3.33624244e-01 1.21819365e+00 -1.84728146e-01 -4.31012958e-01
4.31081295e-01 -2.90871322e-01 -3.95660520e-01 -7.61828303e-01
-2.32453689e-01 -4.39414680e-01 -2.21243888e-01 -8.81913841e-01
4.83685732e-01 -9.97742951e-01 -8.66655290e-01 6.19520992e-02
-6.35700762e-01 -3.54401082e-01 -5.85504889e-01 8.85155201e-01
-4.47458066e-02 1.75464675e-01 -3.90665799e-01 -1.11335731e+00
-5.64781845e-01 -1.13292658e+00 4.76878762e-01 8.47918808e-01
-6.24248087e-01 -1.30260777e+00 1.82103544e-01 1.14244950e+00
7.79783189e-01 1.58970028e-01 9.10777867e-01 -8.10742736e-01
-4.39601719e-01 -4.39220726e-01 3.42905939e-01 3.92120808e-01
1.45097852e-01 6.40983224e-01 -6.46418810e-01 -1.23752221e-01
7.15695471e-02 -2.86040395e-01 -4.20415610e-01 8.04509670e-02
4.06683296e-01 -9.32394505e-01 2.30929613e-01 -1.92443892e-01
1.33996451e+00 3.95679086e-01 3.64926428e-01 8.35556269e-01
3.13126892e-01 1.11778593e+00 8.51892292e-01 4.09945369e-01
8.81326258e-01 -2.11256668e-01 2.56172419e-02 3.03125352e-01
-1.86360538e-01 -4.96286809e-01 4.12268877e-01 1.51555252e+00
3.31110209e-01 2.65907198e-01 -1.32438087e+00 9.28582966e-01
-1.48709929e+00 -5.52485883e-01 -6.59132183e-01 2.27115417e+00
7.43491471e-01 1.52164117e-01 -1.79884076e-01 -1.96945086e-01
1.08902432e-01 -2.35172614e-01 -9.76435244e-02 -1.29058504e+00
6.02971673e-01 5.99734128e-01 1.93854496e-01 8.62115696e-02
-1.65426191e-02 6.39958084e-01 6.09642649e+00 3.32267702e-01
-1.20559847e+00 3.38214904e-01 2.38888413e-01 6.75589442e-02
-1.22647822e+00 3.32313031e-01 -4.55360085e-01 3.40380609e-01
1.58874667e+00 -6.82743669e-01 -1.21902384e-01 4.10068393e-01
6.49946272e-01 -2.25751996e-01 -2.40010843e-01 2.23462909e-01
-5.55122308e-02 -1.49207270e+00 -5.06365418e-01 1.34065062e-01
1.21356273e+00 -2.20964834e-01 2.94401497e-01 1.35961562e-01
3.86735708e-01 -8.14826190e-01 9.46364939e-01 5.66259801e-01
6.02167428e-01 -4.98130620e-01 6.34850204e-01 3.16737264e-01
-6.68603659e-01 -2.09215164e-01 -1.08170055e-01 -6.42855763e-01
-3.45734566e-01 6.66273460e-02 -8.70151103e-01 4.33238655e-01
7.01906085e-01 3.65162194e-01 -6.98771596e-01 1.43397200e+00
-3.90965432e-01 1.09456801e+00 3.03051800e-01 -4.30549383e-01
2.10154369e-01 -6.33561432e-01 1.21956810e-01 1.04114604e+00
5.22766352e-01 1.61457621e-02 -6.34323120e-01 5.63775122e-01
3.36211473e-01 7.70120025e-01 -7.30957925e-01 -5.11329532e-01
1.14761734e+00 1.39546835e+00 -6.46088302e-01 8.16208422e-02
-1.03599858e+00 1.18544206e-01 -3.42882991e-01 3.86463910e-01
-8.66445601e-02 -3.83756280e-01 3.98447365e-01 5.05401373e-01
-8.62748474e-02 -2.22239811e-02 -6.98972225e-01 -3.64577115e-01
-2.19374985e-01 -1.19846117e+00 -1.75108433e-01 -8.34023654e-01
-8.79000306e-01 -1.97291285e-01 -7.27085024e-02 -9.16135073e-01
-2.21960545e-01 -1.34973630e-01 -7.06847489e-01 9.72135186e-01
-1.11505198e+00 -1.41314197e+00 -2.92438013e-03 5.75664006e-02
2.26000190e-01 2.65450150e-01 7.12045610e-01 1.47136882e-01
-4.69390243e-01 4.54257011e-01 6.41243517e-01 -6.01023972e-01
9.55269873e-01 -8.55426013e-01 -4.38825004e-02 8.76645565e-01
-3.66815388e-01 9.88116562e-01 3.80462646e-01 -9.17140186e-01
-1.67150056e+00 -7.60646522e-01 1.32021689e+00 -4.29512978e-01
9.23333347e-01 -1.37089118e-01 -7.27001250e-01 7.37767220e-01
7.58449018e-01 -7.86289454e-01 1.52914798e+00 -1.92824841e-01
3.58996272e-01 2.14664601e-02 -1.33352280e+00 8.61218154e-01
5.72897851e-01 -7.88576782e-01 -5.41337192e-01 1.14280939e-01
5.76054156e-01 -3.46661985e-01 -1.56239367e+00 -1.50907144e-01
8.53459716e-01 -7.99519300e-01 5.34107268e-01 -3.02677065e-01
4.03978765e-01 1.82074502e-01 5.99900901e-01 -1.02748191e+00
-1.69157490e-01 -1.03302634e+00 1.57606125e-01 2.08860588e+00
-6.91481307e-02 -1.23035574e+00 3.20097327e-01 1.62376201e+00
-3.83347064e-01 -7.90697396e-01 -9.82350588e-01 -2.21821785e-01
5.15272319e-01 -5.61617196e-01 7.04532385e-01 1.02259016e+00
3.54606092e-01 -1.25025660e-01 1.90972179e-01 -3.60059179e-02
-7.00748852e-03 -5.52993476e-01 8.54614496e-01 -1.27082956e+00
6.02228880e-01 -6.85410023e-01 -3.44176322e-01 -1.72512174e-01
-6.76685870e-02 -4.28910792e-01 -3.53520423e-01 -2.47602081e+00
-1.07104750e-02 -1.09557413e-01 2.06457391e-01 1.82601273e-01
-2.17800841e-01 -3.39428753e-01 3.83395135e-01 -4.02284600e-02
-3.25425923e-01 1.47002921e-01 1.46230280e+00 5.33111751e-01
-4.26199317e-01 -7.83226416e-02 -1.55726421e+00 4.81230319e-01
7.40540802e-01 -3.34050983e-01 -7.06628740e-01 -1.06644988e-01
6.75793350e-01 7.64916884e-03 8.87939557e-02 -7.67010391e-01
6.13116801e-01 -6.20318830e-01 5.28723672e-02 -1.17628053e-01
-4.01827127e-01 -1.19183493e+00 5.47162652e-01 4.13037211e-01
-4.23646152e-01 2.40453184e-01 5.32012820e-01 -3.98024559e-01
-4.99524642e-03 -8.89034152e-01 -1.69207498e-01 2.11101755e-01
-2.03551725e-01 -3.18773925e-01 -9.79804873e-01 -1.44927591e-01
1.30349338e+00 -6.54431999e-01 -6.78957105e-01 -6.47828996e-01
2.02678323e-01 6.59801662e-02 7.01144993e-01 6.30571663e-01
5.10849297e-01 -8.15268815e-01 -4.30086464e-01 2.06720486e-01
4.59789857e-03 -1.16350800e-01 4.09760684e-01 7.14843392e-01
-4.96028602e-01 8.84011686e-01 -4.17728633e-01 1.57590523e-01
-1.26106811e+00 4.29515950e-02 -1.54426947e-01 1.79893911e-01
-3.02179247e-01 5.88791370e-01 -3.30348581e-01 -3.92717630e-01
2.56707281e-01 -2.63763130e-01 -5.91067076e-01 2.14565739e-01
2.62386829e-01 8.43078852e-01 -4.29423014e-03 -4.95502919e-01
-3.40376608e-02 2.82824814e-01 1.18157253e-01 -4.26991910e-01
1.22774124e+00 -4.85685617e-01 -3.27240288e-01 5.56697547e-01
5.83087921e-01 7.09496081e-01 -9.52593744e-01 2.57202953e-01
-6.18975889e-03 -7.31650293e-01 7.41326958e-02 -1.08936751e+00
-3.69870812e-01 7.21484065e-01 6.71732187e-01 3.20527405e-01
1.12689710e+00 -4.44103688e-01 3.73167545e-01 -3.97770822e-01
-1.27310961e-01 -1.46379364e+00 -5.02318256e-02 1.12430118e-01
7.62536526e-01 -6.22896135e-01 -1.55442404e-02 -9.54353064e-02
-7.18652666e-01 1.18231857e+00 5.93119919e-01 8.08272898e-01
7.70511150e-01 1.86214410e-02 3.69500011e-01 -1.06803283e-01
-7.59962320e-01 1.32385969e-01 4.74247962e-01 7.19038427e-01
1.02524734e+00 9.11513120e-02 -1.15821886e+00 5.25748968e-01
-2.98357487e-01 5.60307562e-01 9.90791321e-01 1.36451268e+00
-1.78212270e-01 -1.28862488e+00 -7.81981826e-01 6.08363688e-01
-1.00415099e+00 9.43042617e-03 -7.45032489e-01 6.72413051e-01
1.89670563e-01 1.65235949e+00 -2.93343663e-01 -2.17626244e-01
3.20925534e-01 1.69433132e-01 4.80427444e-02 -6.19192123e-01
-1.45181143e+00 -7.71257877e-02 3.43580842e-01 9.02769044e-02
-2.94243842e-01 -8.00635695e-01 -6.47018194e-01 -4.72827405e-01
-4.66610014e-01 2.44904697e-01 1.43359828e+00 8.06547582e-01
9.54136193e-01 5.57707310e-01 2.68013686e-01 2.27128744e-01
-4.01376009e-01 -9.82801735e-01 8.82340502e-03 -1.69667229e-01
1.06353439e-01 -2.27551702e-02 1.89931780e-01 -3.12599927e-01] | [10.249577522277832, 7.304266452789307] |
9c8507ac-141d-4373-89dc-1782ce8c2d4c | large-scale-entity-alignment-via-knowledge | 2208.11125 | null | https://arxiv.org/abs/2208.11125v1 | https://arxiv.org/pdf/2208.11125v1.pdf | Large-scale Entity Alignment via Knowledge Graph Merging, Partitioning and Embedding | Entity alignment is a crucial task in knowledge graph fusion. However, most entity alignment approaches have the scalability problem. Recent methods address this issue by dividing large KGs into small blocks for embedding and alignment learning in each. However, such a partitioning and learning process results in an excessive loss of structure and alignment. Therefore, in this work, we propose a scalable GNN-based entity alignment approach to reduce the structure and alignment loss from three perspectives. First, we propose a centrality-based subgraph generation algorithm to recall some landmark entities serving as the bridges between different subgraphs. Second, we introduce self-supervised entity reconstruction to recover entity representations from incomplete neighborhood subgraphs, and design cross-subgraph negative sampling to incorporate entities from other subgraphs in alignment learning. Third, during the inference process, we merge the embeddings of subgraphs to make a single space for alignment search. Experimental results on the benchmark OpenEA dataset and the proposed large DBpedia1M dataset verify the effectiveness of our approach. | ['Xiaofang Zhou', 'Jianfeng Qu', 'Wei Hu', 'Wen Hua', 'Zequn Sun', 'Kexuan Xin'] | 2022-08-23 | null | null | null | null | ['entity-alignment', 'entity-alignment'] | ['knowledge-base', 'natural-language-processing'] | [-5.38553931e-02 3.86526585e-01 -4.15517151e-01 -4.61764127e-01
-5.45171261e-01 -3.49718124e-01 3.13018829e-01 7.96709239e-01
-1.30970240e-01 7.55252600e-01 3.58332813e-01 8.14344957e-02
-3.47583383e-01 -1.30825412e+00 -8.39807153e-01 -5.53789735e-01
-8.37661400e-02 6.33406401e-01 4.54088956e-01 4.94564958e-02
8.46969485e-02 1.34702295e-01 -1.17101490e+00 -2.81236172e-02
1.30193889e+00 5.18130302e-01 6.96497485e-02 -1.33087933e-01
-5.22558868e-01 6.48730397e-01 -1.77793935e-01 -7.30905414e-01
2.15204045e-01 -2.90197462e-01 -9.32813704e-01 -6.59917593e-02
3.45455319e-01 -6.96299002e-02 -3.31464410e-01 1.20852804e+00
5.30586720e-01 4.15419936e-02 4.83066648e-01 -1.44666457e+00
-3.33224475e-01 1.13908803e+00 -8.75698924e-01 -2.43004784e-02
3.54945421e-01 -3.10865521e-01 1.42319036e+00 -7.51803935e-01
7.26248085e-01 1.08790779e+00 8.16235304e-01 -8.67851917e-03
-1.12158716e+00 -7.45152593e-01 3.95327747e-01 4.00344253e-01
-1.68501782e+00 -2.36141011e-01 9.52113628e-01 -3.08681101e-01
7.56492317e-01 1.97037444e-01 7.77252913e-01 5.50848663e-01
-3.53060275e-01 6.53594673e-01 5.29937565e-01 -2.99463660e-01
3.47699411e-02 2.26237893e-01 3.97134066e-01 8.84198785e-01
8.57338548e-01 -5.18373668e-01 -2.76595205e-01 -3.51624459e-01
4.08255756e-01 -3.47853787e-02 -3.01628411e-01 -7.89452374e-01
-1.09591341e+00 6.99438751e-01 8.49824071e-01 1.95594236e-01
-3.68793845e-01 1.67912245e-03 2.92006791e-01 1.28784344e-01
4.70219672e-01 4.08613861e-01 -3.37747335e-01 4.19316590e-01
-8.14623475e-01 1.87888920e-01 7.81123877e-01 1.08713222e+00
1.31610811e+00 -4.85746980e-01 8.87593627e-02 6.84751093e-01
4.63492304e-01 1.75522089e-01 1.22220203e-01 -3.96364510e-01
9.74432945e-01 1.24768090e+00 -1.13556869e-01 -1.69682169e+00
-4.47156429e-01 -4.02046233e-01 -9.29021418e-01 -5.54761827e-01
2.38636229e-02 7.14812800e-02 -6.62179291e-01 1.82947826e+00
9.63652253e-01 4.64727879e-01 9.99629572e-02 7.36975014e-01
8.56773138e-01 4.12452191e-01 1.96485713e-01 -1.08355083e-01
1.19345200e+00 -9.65545118e-01 -5.94596088e-01 -1.13492385e-01
1.00363195e+00 -4.65107203e-01 4.60781008e-01 -2.78616816e-01
-8.05174589e-01 -3.15968484e-01 -1.08886123e+00 -3.61014307e-02
-3.58699679e-01 1.89427137e-02 6.64771497e-01 3.53826791e-01
-6.30598485e-01 6.00871086e-01 -8.17200303e-01 -3.44054550e-01
3.05642098e-01 4.76117551e-01 -6.15817428e-01 -2.29856640e-01
-1.36221910e+00 8.35935891e-01 9.41272795e-01 9.93833542e-02
-1.43472731e-01 -6.49286389e-01 -1.07008064e+00 1.96524918e-01
5.29173493e-01 -8.37948024e-01 4.65616256e-01 -5.87825596e-01
-5.03258705e-01 6.60821855e-01 -2.26573363e-01 -5.71995974e-01
2.04567537e-01 -1.76894292e-01 -5.56092799e-01 -1.82527956e-02
3.44246686e-01 5.08364141e-01 2.84553289e-01 -1.34154117e+00
-7.49844909e-01 -6.14740551e-01 2.88152605e-01 4.49447334e-01
-5.20705879e-01 -5.23223042e-01 -6.18538320e-01 -3.89329553e-01
6.26189351e-01 -7.57122815e-01 -4.47528094e-01 -5.85486770e-01
-6.63967431e-01 -2.55848706e-01 3.74161631e-01 -7.34301209e-01
1.82249975e+00 -1.94066143e+00 3.84716690e-01 7.22869277e-01
5.29039204e-01 -8.99067521e-02 -2.04249434e-02 5.06193519e-01
-1.19156733e-01 1.64133072e-01 -2.13832811e-01 -1.03267789e-01
-6.10680394e-02 6.82128966e-02 -9.32006240e-02 3.45516443e-01
1.54189065e-01 8.98416042e-01 -8.48041058e-01 -9.79489684e-01
-6.50940686e-02 6.77114129e-02 -7.76096761e-01 5.30935973e-02
-8.33799243e-02 1.72077119e-01 -6.96422815e-01 6.55103803e-01
9.00531530e-01 -4.38060790e-01 6.89965069e-01 -7.56397307e-01
1.45520002e-01 3.90019327e-01 -1.50297201e+00 1.79183185e+00
-2.42861807e-01 3.17568593e-02 -2.41402164e-01 -1.26906145e+00
8.20916057e-01 2.28708517e-02 6.93880975e-01 -3.68144214e-01
-1.27332166e-01 3.39060575e-01 -1.44505829e-01 -3.60146105e-01
5.51323116e-01 3.23950857e-01 -1.45555139e-01 3.14842820e-01
3.43428738e-02 4.94729877e-01 2.71264285e-01 4.96985584e-01
1.06532145e+00 1.50028348e-01 5.39687395e-01 -5.95362820e-02
6.58262730e-01 2.47962326e-01 9.50870693e-01 2.19792947e-01
1.89276055e-01 3.10020685e-01 4.41086233e-01 -4.60096449e-01
-8.70872796e-01 -9.13644612e-01 2.20859885e-01 5.76148033e-01
6.66077614e-01 -8.57517421e-01 -7.19039977e-01 -1.01264155e+00
1.97077155e-01 4.43320185e-01 -3.36422890e-01 -3.92146081e-01
-6.91422224e-01 -9.13216114e-01 3.63768369e-01 4.67188597e-01
3.34907562e-01 -7.48182774e-01 1.73359066e-01 3.10746044e-01
-6.46803737e-01 -1.11003327e+00 -3.48035455e-01 -4.00337689e-02
-8.85420203e-01 -1.45163059e+00 -1.39365792e-01 -1.07803285e+00
1.17203867e+00 3.03407311e-01 1.06517971e+00 2.11117432e-01
-6.10380061e-02 2.12327018e-02 -2.04142004e-01 1.33672625e-01
-7.08382353e-02 5.48841536e-01 2.61278097e-02 -7.36066848e-02
6.89772010e-01 -8.20251465e-01 -5.71129918e-01 3.46922338e-01
-4.81548816e-01 1.95236057e-01 7.04304218e-01 6.61465406e-01
7.47351050e-01 1.47802681e-01 6.55620992e-01 -1.13534582e+00
3.87760967e-01 -7.48766303e-01 -5.53623617e-01 6.77042723e-01
-8.41182411e-01 2.29670659e-01 4.08212870e-01 -7.63661638e-02
-8.63621056e-01 2.28446439e-01 2.39770338e-01 -2.11711183e-01
1.74091727e-01 7.45390296e-01 -7.02111840e-01 9.15459543e-02
2.65679210e-01 2.05177501e-01 -2.80036151e-01 -3.84987652e-01
5.40992022e-01 5.40166795e-01 3.27427030e-01 -5.66637099e-01
1.14284837e+00 2.93975413e-01 4.00024764e-02 -3.79719377e-01
-8.11267078e-01 -7.53759265e-01 -7.28107393e-01 1.28768340e-01
7.34326601e-01 -1.26112354e+00 -4.64556128e-01 7.64034390e-02
-9.70191419e-01 4.69887316e-01 -1.36164963e-01 4.38742489e-01
-1.73701420e-01 6.94968343e-01 -1.75975844e-01 -4.93230581e-01
-4.59788114e-01 -9.11180139e-01 8.00696075e-01 3.41650397e-01
-1.83549449e-01 -8.15149009e-01 3.78893256e-01 5.07105529e-01
-6.31905645e-02 3.11142266e-01 9.81841266e-01 -1.13363862e+00
-9.63463306e-01 -1.42802775e-01 -3.60165656e-01 -1.93371534e-01
2.76087224e-01 -3.13270032e-01 -4.90000039e-01 -3.16147208e-01
-5.58100820e-01 -1.26679733e-01 8.36079895e-01 -6.26464188e-02
7.02529609e-01 -4.18710411e-01 -9.38966274e-01 7.01845646e-01
1.50592852e+00 -2.02649474e-01 3.41053665e-01 4.14188921e-01
1.10772896e+00 5.77353120e-01 9.66585755e-01 2.81157523e-01
8.91214073e-01 5.49541473e-01 2.32547179e-01 -2.11399570e-02
1.39447033e-01 -6.87165916e-01 4.43352014e-02 1.25688541e+00
-2.61411760e-02 -2.52535462e-01 -8.48697603e-01 7.04078376e-01
-2.13214111e+00 -7.59975016e-01 -2.75126278e-01 2.25199866e+00
8.18011045e-01 1.74501948e-02 -3.89382876e-02 -3.19561064e-02
1.08363903e+00 6.45727441e-02 -2.67436624e-01 3.35320234e-01
-4.59638312e-02 -4.50059921e-02 4.73080069e-01 3.36465836e-01
-1.15661085e+00 9.84517574e-01 4.21686745e+00 7.18982995e-01
-5.09644568e-01 3.42490524e-02 2.82860398e-01 3.01919848e-01
-7.22689092e-01 5.37363529e-01 -1.02545607e+00 4.83263284e-01
5.74693501e-01 -4.45399523e-01 1.36224538e-01 7.75426149e-01
-2.37368256e-01 8.26298743e-02 -8.96161139e-01 7.45260596e-01
-6.16612881e-02 -1.17823422e+00 2.29656145e-01 5.10464981e-02
7.11963832e-01 -1.71876132e-01 -7.22770274e-01 3.82784426e-01
3.76870245e-01 -6.07461870e-01 1.23039797e-01 3.82417649e-01
3.60114038e-01 -1.04303169e+00 7.94898212e-01 2.10254163e-01
-1.70057559e+00 1.05573341e-01 -6.14685833e-01 5.14129639e-01
2.21784234e-01 7.47716248e-01 -8.69357765e-01 1.30943739e+00
4.71831143e-01 7.72334576e-01 -6.15824461e-01 1.13145661e+00
-1.79135770e-01 2.60448605e-01 -4.64157969e-01 2.37366974e-01
-1.18214041e-02 -5.97910583e-01 4.75558132e-01 9.16878521e-01
2.38342285e-01 6.00024164e-02 4.63936776e-01 6.81821942e-01
-3.66196841e-01 4.74237233e-01 -7.93747425e-01 -3.52485687e-03
1.07036269e+00 1.29706061e+00 -6.46830678e-01 -1.75503835e-01
-6.02642953e-01 9.35727775e-01 8.47326159e-01 1.47207156e-01
-1.10973251e+00 -5.48561573e-01 6.10882342e-01 2.19737515e-01
3.13716561e-01 6.45091608e-02 5.31935990e-02 -1.32441556e+00
2.68673718e-01 -7.34061003e-01 6.71224952e-01 -1.97755694e-01
-1.01309323e+00 2.15429753e-01 -2.83647552e-02 -1.29492962e+00
1.03475690e-01 1.66351303e-01 -5.46177566e-01 5.20069540e-01
-1.49385488e+00 -1.36153293e+00 -5.06951094e-01 2.72780508e-01
-1.01218045e-01 -6.39297888e-02 6.74228370e-01 6.92075253e-01
-6.86773181e-01 7.17228472e-01 -4.95316237e-02 3.90000701e-01
7.16052294e-01 -1.15736949e+00 6.30332112e-01 7.76931942e-01
3.64464104e-01 9.61189449e-01 3.68282735e-01 -1.01043367e+00
-1.15791786e+00 -1.28869033e+00 1.04555368e+00 -1.68567032e-01
4.65908796e-01 -2.83991486e-01 -1.04636753e+00 8.88848126e-01
-6.98595867e-02 -1.15500197e-01 8.86097491e-01 4.57641095e-01
-3.71138215e-01 -3.47124577e-01 -1.05747700e+00 5.44646204e-01
1.29044604e+00 -3.05659950e-01 -7.33115554e-01 2.90862143e-01
8.73461187e-01 -2.34030336e-01 -1.23626792e+00 7.26875782e-01
4.70667928e-01 -7.01605320e-01 1.07285345e+00 -7.49019444e-01
3.49018246e-01 -6.49074137e-01 -1.14312246e-01 -1.20066726e+00
-3.32829535e-01 -2.99618244e-01 -3.02175939e-01 1.82772815e+00
5.75103700e-01 -6.52481616e-01 9.22818542e-01 3.49518031e-01
1.90912947e-01 -7.39154577e-01 -7.42172241e-01 -4.95856464e-01
-3.37520629e-01 2.02481523e-01 9.49095607e-01 1.36661172e+00
1.51262268e-01 5.65409541e-01 -2.52294779e-01 6.97865605e-01
8.81468296e-01 6.24543011e-01 1.00787330e+00 -1.42139649e+00
-3.84751186e-02 -2.31250022e-02 -8.14038575e-01 -9.15696144e-01
1.21051811e-01 -1.18832564e+00 -3.33975196e-01 -1.82287073e+00
5.94963849e-01 -7.25009203e-01 -4.45182025e-01 3.35651249e-01
-5.75314105e-01 -3.06420058e-01 -1.85368404e-01 2.64781207e-01
-8.95317554e-01 6.88781917e-01 7.20390618e-01 -2.25470096e-01
-1.73598915e-01 -1.92951903e-01 -7.56019771e-01 5.48845470e-01
7.28185534e-01 -6.81880474e-01 -5.77849984e-01 -2.91394711e-01
3.62800628e-01 -4.57661413e-02 1.85107052e-01 -9.28558171e-01
5.45886934e-01 3.76264192e-02 2.11498141e-01 -8.79189610e-01
-4.25327644e-02 -1.04987729e+00 4.69099075e-01 3.14722091e-01
-2.41484586e-02 -2.98654009e-02 -1.42539620e-01 9.56894219e-01
-4.34723318e-01 -1.01921104e-01 4.79974627e-01 -4.09652740e-02
-7.02216148e-01 4.55749273e-01 5.40955484e-01 3.84107471e-01
1.23311925e+00 -1.73252806e-01 -3.28810722e-01 1.08193628e-01
-7.32073247e-01 8.32415342e-01 6.05873585e-01 3.73129874e-01
4.18924034e-01 -1.56933939e+00 -5.91128469e-01 -7.16385711e-03
2.70948410e-01 5.38334966e-01 2.21864581e-01 9.57963347e-01
-3.89330715e-01 1.54479101e-01 7.05591738e-02 -4.26947743e-01
-1.37933099e+00 5.03737569e-01 1.06759675e-01 -6.13824129e-01
-5.13079524e-01 6.85412169e-01 2.03180641e-01 -7.45018065e-01
1.23849459e-01 9.42902490e-02 -4.43820655e-01 3.03647906e-01
-1.03011299e-02 3.49553406e-01 -5.39902560e-02 -4.92886573e-01
-5.28265595e-01 5.15914202e-01 -3.87764245e-01 3.60029250e-01
1.37131250e+00 -2.57756203e-01 -3.58437359e-01 7.22736716e-02
1.07686269e+00 2.68320531e-01 -4.91937995e-01 -3.77965927e-01
4.93601739e-01 -3.88149887e-01 -3.08254480e-01 -1.48943678e-01
-1.10078621e+00 3.96148801e-01 1.59636453e-01 -7.65596032e-02
9.55017269e-01 5.93695901e-02 9.45837200e-01 5.41275382e-01
3.00448298e-01 -9.59291041e-01 -4.09759998e-01 3.20957378e-02
2.97690570e-01 -1.19975591e+00 3.61957252e-01 -8.29804540e-01
-4.76333231e-01 7.49038637e-01 1.10088873e+00 1.61139429e-01
6.00741327e-01 -1.45388737e-01 -3.32960010e-01 -5.16408861e-01
-5.45151949e-01 -2.68625677e-01 2.73145944e-01 3.00322860e-01
2.70436108e-01 5.83147407e-02 -5.73290527e-01 7.00601101e-01
1.69597361e-02 -2.84946233e-01 1.29621431e-01 8.80383313e-01
-5.80073535e-01 -1.23847711e+00 2.16806993e-01 4.69605207e-01
-2.57584780e-01 -2.46886075e-01 -5.06345272e-01 9.17283833e-01
1.89265430e-01 5.27327955e-01 -8.99541602e-02 -5.18630803e-01
3.78489733e-01 6.16957471e-02 2.90104538e-01 -4.93028671e-01
-3.84537995e-01 -2.56298095e-01 3.03152412e-01 -3.94439042e-01
-3.98764074e-01 -5.01105785e-01 -1.43459630e+00 -1.48347706e-01
-8.01408947e-01 5.83939552e-01 2.48443782e-01 7.71599412e-01
6.39895141e-01 3.39545488e-01 7.30004072e-01 -1.27296269e-01
-4.92535293e-01 -8.81534576e-01 -4.43259895e-01 5.92513978e-01
-1.34845108e-01 -6.31521285e-01 -3.36758077e-01 -3.42780977e-01] | [8.71433162689209, 8.0066499710083] |
986148ce-7c49-484c-9e9a-9870f4849ecb | a-comprehensive-review-of-modern-object | 2301.07499 | null | https://arxiv.org/abs/2301.07499v1 | https://arxiv.org/pdf/2301.07499v1.pdf | A Comprehensive Review of Modern Object Segmentation Approaches | Image segmentation is the task of associating pixels in an image with their respective object class labels. It has a wide range of applications in many industries including healthcare, transportation, robotics, fashion, home improvement, and tourism. Many deep learning-based approaches have been developed for image-level object recognition and pixel-level scene understanding-with the latter requiring a much denser annotation of scenes with a large set of objects. Extensions of image segmentation tasks include 3D and video segmentation, where units of voxels, point clouds, and video frames are classified into different objects. We use "Object Segmentation" to refer to the union of these segmentation tasks. In this monograph, we investigate both traditional and modern object segmentation approaches, comparing their strengths, weaknesses, and utilities. We examine in detail the wide range of deep learning-based segmentation techniques developed in recent years, provide a review of the widely used datasets and evaluation metrics, and discuss potential future research directions. | ['Matthew Hagen', 'Hanyan Li', 'Unaiza Ahsan', 'Yuanbo Wang'] | 2023-01-13 | null | null | null | null | ['video-semantic-segmentation'] | ['computer-vision'] | [ 4.74225283e-01 4.67740893e-02 -2.86651611e-01 -6.07809961e-01
-5.76345026e-01 -4.79273140e-01 4.38327938e-01 3.47659439e-01
-4.60110366e-01 3.28850508e-01 -3.26641947e-01 -8.94112587e-02
-1.68877877e-02 -7.69225419e-01 -6.10275567e-01 -5.77116489e-01
-3.37986685e-02 5.36998391e-01 5.18218756e-01 3.20146203e-01
3.65812600e-01 7.95640290e-01 -1.52066267e+00 2.03869879e-01
6.44822657e-01 1.36371851e+00 1.93073884e-01 4.72891212e-01
-7.13053584e-01 4.59585994e-01 -6.45709932e-01 -1.47833258e-01
2.64258325e-01 -1.97395101e-01 -1.16055572e+00 8.10605407e-01
5.72832644e-01 -2.86187530e-01 -5.80334142e-02 1.08219528e+00
6.46309480e-02 2.37124488e-01 7.59541750e-01 -1.32703781e+00
-3.75065416e-01 2.28214756e-01 -7.95337796e-01 2.10799634e-01
6.02421872e-02 -1.13097308e-02 7.34786212e-01 -6.74858153e-01
5.52708566e-01 1.09275389e+00 7.14496851e-01 3.76387835e-01
-1.05805969e+00 -3.16808075e-01 2.88750201e-01 1.31882936e-01
-1.20260489e+00 1.39217675e-01 8.30856740e-01 -7.59854198e-01
8.54265928e-01 2.57248014e-01 8.39138627e-01 3.72749329e-01
-1.18536621e-01 1.27113056e+00 9.70080256e-01 -2.81995237e-01
3.71747583e-01 -8.40653628e-02 5.80367029e-01 5.48150897e-01
1.47071719e-01 -4.06966567e-01 -4.80255000e-02 2.20719919e-01
1.09266376e+00 1.67753622e-01 6.50356784e-02 -5.80630660e-01
-1.28994322e+00 7.66663373e-01 6.73320711e-01 3.55234921e-01
-4.35537696e-01 3.68201345e-01 3.92435223e-01 -1.71674162e-01
5.85275412e-01 2.70689368e-01 -5.86533725e-01 2.92265624e-01
-1.07844841e+00 2.02281728e-01 5.20468354e-01 1.09781945e+00
9.61795926e-01 -3.85464579e-02 -1.03937676e-02 8.58401299e-01
3.12600881e-01 1.20613649e-01 3.05328131e-01 -1.44071996e+00
5.42201884e-02 7.17873633e-01 -3.03648710e-02 -7.47112751e-01
-5.63416898e-01 -1.16050482e-01 -6.84489369e-01 3.10502738e-01
2.36537471e-01 2.42331505e-01 -1.46121323e+00 9.62646067e-01
5.26288390e-01 3.34207177e-01 -1.94509551e-01 8.79445910e-01
1.21528983e+00 6.46362484e-01 4.01721776e-01 8.71374682e-02
1.33422124e+00 -1.08088458e+00 -4.99643028e-01 -3.43255401e-01
3.04297388e-01 -6.08325720e-01 8.64900291e-01 3.29567969e-01
-1.08254397e+00 -8.28877211e-01 -7.44240701e-01 -3.76207173e-01
-7.59860992e-01 -2.54002035e-01 8.60386848e-01 5.87789178e-01
-9.42872524e-01 5.26657581e-01 -9.70577598e-01 -5.89988887e-01
9.87360358e-01 5.81896365e-01 -1.76539570e-01 5.61845973e-02
-7.10544229e-01 5.11004567e-01 5.42767823e-01 3.48630138e-02
-8.55018020e-01 -4.24939722e-01 -9.57033753e-01 -1.71317220e-01
1.61656618e-01 -5.28949618e-01 1.39412642e+00 -1.04285634e+00
-1.06414974e+00 1.42943239e+00 3.57431732e-02 -5.03834546e-01
3.53843540e-01 -3.02405924e-01 2.74906158e-02 1.90205485e-01
1.60129309e-01 1.39004719e+00 6.16301835e-01 -1.57848966e+00
-1.13783836e+00 -3.97907645e-01 1.92797169e-01 4.07569617e-01
7.01056570e-02 -3.51234786e-02 -9.12164629e-01 -4.24598247e-01
4.78102177e-01 -6.86937392e-01 -6.24445140e-01 9.19089615e-02
-5.47699749e-01 -4.06687289e-01 1.15096843e+00 -2.84999996e-01
6.72669172e-01 -2.10492468e+00 8.40431899e-02 4.09025624e-02
2.21129626e-01 1.49032235e-01 -5.96194193e-02 -7.16341361e-02
1.42794728e-01 2.66626477e-01 -7.19805241e-01 -3.57658595e-01
-1.34070098e-01 3.75010580e-01 -1.97359584e-02 4.98059481e-01
5.07485196e-02 1.00462353e+00 -7.53120065e-01 -7.86534846e-01
7.77751684e-01 3.14859807e-01 -1.29591480e-01 3.35317515e-02
-5.27922392e-01 5.73773503e-01 -5.61110795e-01 9.01025772e-01
6.15839422e-01 -1.79620713e-01 -2.36544415e-01 -2.15019703e-01
-1.55730218e-01 -1.53986719e-02 -1.10323584e+00 1.82390106e+00
-5.50409965e-03 9.24645245e-01 1.90130547e-01 -1.22414160e+00
6.88415766e-01 7.93167502e-02 1.09447098e+00 -5.99093437e-01
2.88193494e-01 1.70198828e-01 -2.82864928e-01 -6.11672282e-01
5.24220705e-01 9.11685899e-02 -1.71087012e-01 2.04491064e-01
6.15126342e-02 -6.30432963e-01 3.02697808e-01 -1.72867239e-01
5.43830574e-01 2.06272960e-01 2.50708342e-01 -1.46350890e-01
5.01423120e-01 5.22413790e-01 4.66854602e-01 7.65809715e-01
-5.43231606e-01 8.41896832e-01 1.47521317e-01 -6.04842365e-01
-8.62633586e-01 -1.01521933e+00 -4.40532982e-01 8.96882832e-01
6.66890502e-01 1.85217351e-01 -1.09698927e+00 -5.99231601e-01
1.35752887e-01 3.31035197e-01 -5.01443744e-01 3.18892926e-01
-6.71259642e-01 -7.88495958e-01 3.01944792e-01 7.92634487e-01
7.41945386e-01 -1.34015465e+00 -7.55696535e-01 5.54999597e-02
-7.08864704e-02 -1.28164911e+00 -1.59014519e-02 2.91531533e-01
-1.22704864e+00 -1.10362017e+00 -8.90565217e-01 -1.28217471e+00
6.77643120e-01 4.24094766e-01 1.21605468e+00 1.51667178e-01
-4.04661059e-01 5.99942744e-01 -3.41635436e-01 -4.11026478e-01
-4.13042344e-02 8.19477215e-02 -2.09661812e-01 -2.00640067e-01
5.78587532e-01 -8.18844661e-02 -5.16655266e-01 2.14828312e-01
-1.14358222e+00 7.50258341e-02 5.40148973e-01 2.89990127e-01
1.24014962e+00 1.60801098e-01 1.43247321e-01 -1.06401038e+00
2.29289904e-01 -3.34846169e-01 -5.85534036e-01 9.42966864e-02
-1.40040353e-01 -5.04968405e-01 -5.29315881e-02 -7.89641738e-02
-7.63706505e-01 3.35753769e-01 -2.88976729e-01 -3.28045219e-01
-8.73390973e-01 3.79673392e-01 -3.01840633e-01 -1.37282759e-01
2.41490722e-01 -3.25863212e-02 -1.79329515e-01 -5.49129784e-01
6.09957457e-01 6.48905635e-01 6.50768995e-01 -4.20506686e-01
3.95714551e-01 7.18364000e-01 -7.61878341e-02 -1.02108455e+00
-9.42787945e-01 -9.22107935e-01 -1.30543697e+00 -4.31584716e-01
1.44180334e+00 -5.51538229e-01 -2.62170970e-01 7.32659698e-01
-1.27593982e+00 -6.26017094e-01 -6.05877042e-01 3.45289052e-01
-8.06917727e-01 3.41028094e-01 -6.53358400e-01 -4.58865434e-01
-1.70580789e-01 -1.46407497e+00 1.37105012e+00 5.32520294e-01
-1.08975820e-01 -1.11232567e+00 -1.32928506e-01 5.88908553e-01
-2.20043480e-01 4.63757128e-01 8.93916011e-01 -6.36262834e-01
-7.02418029e-01 -1.10228762e-01 -4.55211103e-01 5.13099670e-01
1.82753310e-01 -9.51457471e-02 -8.72242332e-01 1.21605687e-01
-1.07224181e-01 -1.08558327e-01 7.94875562e-01 9.38229978e-01
1.63063788e+00 2.21548483e-01 -6.55270338e-01 6.68559194e-01
1.52065372e+00 5.20832062e-01 5.52868485e-01 3.24213952e-01
1.01244521e+00 8.25177848e-01 7.78039515e-01 -2.77571362e-02
1.82840064e-01 4.31969136e-01 5.54397881e-01 -5.88992715e-01
-1.49974197e-01 3.88007522e-01 -2.46884614e-01 4.43165362e-01
7.31724426e-02 -3.73136699e-01 -1.03684342e+00 6.16007030e-01
-1.76769733e+00 -6.19788945e-01 -4.77319628e-01 1.86086810e+00
4.92307991e-01 1.76845670e-01 3.26292336e-01 1.86561182e-01
8.33436668e-01 2.66249981e-02 -8.51113141e-01 -2.36309230e-01
3.51723209e-02 1.73054278e-01 7.06163168e-01 2.90336490e-01
-1.68317342e+00 1.18890226e+00 7.25741386e+00 6.64425373e-01
-9.46434379e-01 2.26636305e-02 1.11674774e+00 3.23110163e-01
1.50241718e-01 -3.09568346e-01 -5.78429818e-01 1.96223408e-01
3.52162302e-01 4.80243593e-01 -2.14056056e-02 9.44210708e-01
1.77918434e-01 -3.82281184e-01 -1.19549084e+00 9.22457814e-01
2.89324373e-02 -1.38215220e+00 8.94456729e-03 6.24782080e-03
9.24724758e-01 2.61930346e-01 -2.78488994e-02 5.27771488e-02
3.09844613e-02 -9.87742126e-01 7.73119807e-01 4.02860492e-01
4.99457628e-01 -4.94308740e-01 6.14011168e-01 3.68203782e-02
-1.20500505e+00 4.63576578e-02 -1.83651775e-01 1.35893002e-01
3.85015368e-01 5.28024852e-01 -3.91553342e-01 3.20210427e-01
1.06644678e+00 7.67075598e-01 -4.02372658e-01 1.36140871e+00
1.10557757e-01 4.12036031e-01 -3.53066206e-01 3.28530878e-01
5.64920425e-01 -5.91040254e-01 2.53175259e-01 1.30750299e+00
-7.55110234e-02 2.28105620e-01 6.75145268e-01 8.92174304e-01
-1.19743384e-01 -1.48133770e-01 -2.71686107e-01 3.92027237e-02
2.81127274e-01 1.34092689e+00 -1.59995866e+00 -6.26696587e-01
-5.74586153e-01 1.06866300e+00 -2.01476380e-01 4.17843133e-01
-8.36354077e-01 -4.94750887e-01 8.50941479e-01 1.21875964e-01
1.01202346e-01 -5.37933767e-01 -8.15091610e-01 -5.57434618e-01
-3.80277246e-01 -3.88762325e-01 3.68078202e-01 -8.04131210e-01
-1.07807314e+00 3.60411137e-01 3.84012580e-01 -1.13405442e+00
9.94229093e-02 -8.03145826e-01 -4.83491927e-01 4.70959842e-01
-1.47956181e+00 -1.08210742e+00 -6.17027581e-01 3.35659057e-01
1.06998420e+00 2.02552870e-01 3.33082765e-01 3.56221735e-01
-6.06137395e-01 -1.63110197e-01 1.36268109e-01 5.28399825e-01
2.54675373e-02 -1.35841823e+00 5.98056436e-01 6.51345670e-01
2.53288954e-01 9.38016102e-02 3.76768798e-01 -5.51263273e-01
-8.87401819e-01 -1.33231020e+00 3.92670780e-01 -3.54919463e-01
2.37988472e-01 -2.33175173e-01 -8.34885597e-01 7.95599103e-01
9.74109992e-02 5.47557883e-02 5.58185697e-01 -3.61759156e-01
2.36223683e-01 -5.97118214e-02 -1.24510932e+00 4.36418295e-01
8.73031855e-01 -2.01946557e-01 -3.75527501e-01 5.77332020e-01
7.79904842e-01 -5.25432348e-01 -8.56951296e-01 5.48095524e-01
2.85834640e-01 -9.82640266e-01 1.22806382e+00 -3.55615854e-01
1.41841546e-01 -4.00512338e-01 -1.65994894e-02 -7.65485406e-01
-2.74213344e-01 2.07489487e-02 5.60673326e-02 1.12242496e+00
-2.08543763e-02 -1.66266665e-01 1.18492043e+00 6.76587045e-01
-5.56023777e-01 -8.81559789e-01 -7.18812644e-01 -6.64178252e-01
5.85598312e-02 -6.44867718e-01 5.62085688e-01 7.93167830e-01
-5.40823162e-01 1.69864595e-02 2.91918665e-01 1.05098546e-01
7.07290232e-01 3.25161725e-01 5.88018358e-01 -1.43810368e+00
3.46803308e-01 -8.60919714e-01 -6.64928377e-01 -1.27072644e+00
1.05974823e-01 -7.28922367e-01 3.09151918e-01 -2.30646229e+00
5.57596162e-02 -6.32265687e-01 -1.97537839e-01 3.65838587e-01
1.32265687e-01 7.38977313e-01 2.41959803e-02 4.87955719e-01
-8.11043203e-01 8.05586651e-02 1.14455438e+00 -5.23569286e-01
-4.35484111e-01 1.14978671e-01 -2.72194475e-01 1.07206511e+00
7.93695331e-01 -1.65036768e-01 -3.06590617e-01 -7.37372994e-01
-3.77588421e-01 -3.83785903e-01 3.35140973e-01 -1.19649088e+00
6.41019419e-02 -3.55104476e-01 4.81313765e-01 -9.07223105e-01
4.21711624e-01 -1.00619614e+00 4.69396785e-02 3.91899198e-01
-2.94854283e-01 -3.01357627e-01 1.65682316e-01 4.09677148e-01
-4.23579097e-01 -3.64511371e-01 1.00634956e+00 -6.10893607e-01
-1.53531837e+00 5.80089927e-01 -5.76778293e-01 -2.25273192e-01
1.40757442e+00 -8.04576755e-01 2.60318011e-01 2.15775575e-02
-9.00069296e-01 3.24299753e-01 3.89541239e-01 4.04297113e-01
6.38973415e-01 -1.04286516e+00 -2.05152005e-01 9.36130062e-03
2.68283170e-02 6.96932614e-01 1.90362230e-01 7.04022050e-01
-9.11373377e-01 4.74824578e-01 -1.11814044e-01 -1.03523636e+00
-1.09185565e+00 4.26835090e-01 5.27390063e-01 2.05832601e-01
-5.92323661e-01 9.32621002e-01 5.32268226e-01 -3.91787589e-01
5.61832368e-01 -6.86173379e-01 -3.78170162e-01 3.57900746e-02
4.70294654e-02 4.37742472e-01 4.33974601e-02 -8.68366539e-01
-3.69433284e-01 8.63952875e-01 1.76636040e-01 5.35982475e-02
9.75449622e-01 -2.12421075e-01 -2.75548846e-01 5.25163114e-01
1.21204007e+00 -8.10535192e-01 -1.35686493e+00 -1.59903020e-01
2.88299650e-01 -2.11667970e-01 1.83508918e-01 -4.80753064e-01
-1.42937207e+00 1.03859258e+00 7.90462494e-01 4.45473969e-01
1.18867671e+00 4.00034696e-01 1.06853485e+00 1.15496777e-01
4.43212092e-01 -1.28959203e+00 5.63264973e-02 5.45783937e-01
4.10018206e-01 -1.35873902e+00 9.30151939e-02 -6.33782625e-01
-5.18829882e-01 1.00828981e+00 6.83105290e-01 -1.38180479e-01
8.41785073e-01 1.40259027e-01 1.96490943e-01 -3.77601177e-01
2.32589483e-01 -6.09439254e-01 3.98848772e-01 8.34511399e-01
3.32097411e-01 1.15001630e-02 -6.73342720e-02 2.44791716e-01
1.82897106e-01 -1.60589844e-01 1.72206208e-01 1.06176746e+00
-8.32612038e-01 -9.37918723e-01 -4.22119379e-01 6.94281399e-01
-3.21110457e-01 1.58546284e-01 -4.58953977e-01 8.85572493e-01
4.75864679e-01 7.90347219e-01 7.48723149e-01 -2.16761947e-01
9.83069241e-02 -1.52241468e-01 4.31028694e-01 -8.28118384e-01
-5.21745682e-01 2.29124635e-01 -2.38125652e-01 -4.44573134e-01
-1.00338256e+00 -7.40585864e-01 -1.67953587e+00 1.88063979e-01
-1.59095421e-01 -1.00960992e-01 9.10977840e-01 1.10901737e+00
1.16123296e-01 3.91609460e-01 3.01864386e-01 -1.39349520e+00
4.03176695e-01 -5.95660508e-01 -7.30964065e-01 3.80305737e-01
1.53425321e-01 -5.24277627e-01 9.90684256e-02 5.08650243e-01] | [9.586910247802734, 0.34783217310905457] |
e7beb69f-7471-4719-923c-082a469f6145 | caunlp-at-nlp4if-2019-shared-task-context | null | null | https://aclanthology.org/D19-5010 | https://aclanthology.org/D19-5010.pdf | CAUnLP at NLP4IF 2019 Shared Task: Context-Dependent BERT for Sentence-Level Propaganda Detection | The goal of fine-grained propaganda detection is to determine whether a given sentence uses propaganda techniques (sentence-level) or to recognize which techniques are used (fragment-level). This paper presents the sys- tem of our participation in the sentence-level subtask of the propaganda detection shared task. In order to better utilize the document information, we construct context-dependent input pairs (sentence-title pair and sentence- context pair) to fine-tune the pretrained BERT, and we also use the undersampling method to tackle the problem of imbalanced data. | ['Ying Chen', 'Wenjun Hou'] | 2019-11-01 | null | null | null | ws-2019-11 | ['propaganda-detection'] | ['natural-language-processing'] | [ 4.08834726e-01 -3.31881881e-01 -2.66955048e-01 -6.19229496e-01
-8.13425422e-01 -4.53705221e-01 9.77561355e-01 8.00902426e-01
-3.78973782e-01 5.02347887e-01 7.77272880e-01 -6.86490953e-01
2.02692688e-01 -8.36919308e-01 -2.18778700e-01 -3.77691776e-01
1.82536080e-01 2.52135783e-01 -2.95364093e-02 -4.83552277e-01
1.20218062e+00 3.66986871e-01 -1.24238718e+00 9.31342363e-01
9.72515345e-01 2.95129478e-01 -3.82528007e-01 8.95236611e-01
-2.61557877e-01 1.34825146e+00 -1.34295475e+00 -3.26126926e-02
-1.31800562e-01 -7.06068993e-01 -1.16165686e+00 -9.14816186e-02
6.60439074e-01 -1.71547711e-01 9.94495526e-02 9.65302408e-01
4.20204461e-01 -3.95725034e-02 7.72293985e-01 -6.43729389e-01
-3.03185254e-01 7.74137616e-01 -7.53051102e-01 1.01055884e+00
5.05759239e-01 -4.03677672e-02 8.50420117e-01 -4.46288109e-01
9.28637087e-01 1.62385654e+00 2.71661848e-01 3.24615091e-01
-1.04509580e+00 -5.41100681e-01 3.36230993e-02 4.17243212e-01
-7.23938525e-01 -4.83621299e-01 1.00593364e+00 -8.08557212e-01
1.13730168e+00 2.67612308e-01 4.47488397e-01 1.37782943e+00
5.05548596e-01 7.93968201e-01 1.45242167e+00 -7.28334963e-01
2.08500981e-01 -4.22639288e-02 8.99423540e-01 5.91291606e-01
1.28616840e-01 -2.06148267e-01 -7.39303470e-01 -6.25381589e-01
1.10246085e-01 -4.68967587e-01 4.87994738e-02 4.13515836e-01
-8.55045676e-01 1.38653743e+00 3.19577873e-01 8.02522063e-01
-2.96529055e-01 -8.84173438e-02 9.55144703e-01 5.56196272e-01
7.78832793e-01 6.96793377e-01 -1.95665494e-01 -3.66603822e-01
-1.13242137e+00 6.76074862e-01 8.26079249e-01 1.23014152e-01
3.30305398e-01 6.17008097e-03 -7.78948247e-01 6.40524447e-01
-1.19378932e-01 3.09327126e-01 5.45945883e-01 -2.47293234e-01
1.03808117e+00 5.21809578e-01 -5.99021241e-02 -1.16061223e+00
-6.25104785e-01 -2.89915800e-01 -6.45222366e-01 -8.97168219e-02
3.55333209e-01 -7.17681423e-02 -9.45649862e-01 1.42056406e+00
4.08952147e-01 -3.35539490e-01 -4.04668480e-01 7.00816751e-01
6.54927075e-01 7.47876227e-01 3.25264961e-01 -4.85297382e-01
1.71862376e+00 -8.33851874e-01 -8.53691816e-01 -4.59722489e-01
1.26697600e+00 -8.31921935e-01 1.07546258e+00 1.91633210e-01
-7.27587402e-01 -2.14969635e-01 -1.03336525e+00 -7.53978081e-03
-2.90819079e-01 -1.44328196e-02 9.16797519e-02 8.02520037e-01
-3.08938175e-01 4.11266208e-01 -5.39274693e-01 -1.78050965e-01
4.02727813e-01 -4.81274247e-01 -1.85288772e-01 1.08660355e-01
-1.48314655e+00 1.35325277e+00 2.73322284e-01 -1.88651338e-01
-7.64530480e-01 -5.70574105e-01 -9.57734168e-01 -1.11077875e-01
2.23271996e-01 -4.44722593e-01 1.02724254e+00 -8.87390375e-01
-1.01203060e+00 1.14342630e+00 -1.11721255e-01 -4.48509604e-01
2.19017610e-01 -2.75446653e-01 -2.79318511e-01 1.89429849e-01
2.92267859e-01 -2.83017904e-01 1.32354236e+00 -6.46297276e-01
-2.78922528e-01 -6.25332117e-01 -1.35203293e-02 1.35800108e-01
-3.99606138e-01 9.68167722e-01 6.17541671e-01 -9.61556077e-01
-1.64915785e-01 -4.02015120e-01 5.27515449e-02 -8.28917384e-01
-7.45461464e-01 -5.42942166e-01 8.09792757e-01 -9.76753175e-01
1.63397038e+00 -1.95791447e+00 6.84238374e-02 1.33098155e-01
2.27613643e-01 4.66776550e-01 -2.82869577e-01 7.21215308e-01
-2.40153939e-01 3.67015749e-01 1.46994628e-02 -4.67894748e-02
-3.32267791e-01 -1.11079976e-01 -3.56603235e-01 6.33338988e-01
2.64613390e-01 3.09143424e-01 -9.28303003e-01 -6.06758058e-01
-8.90968516e-02 -7.92064816e-02 -4.73362356e-01 1.13031767e-01
-8.30763951e-02 8.16078037e-02 -6.15900695e-01 3.27403098e-01
5.11342525e-01 1.12903891e-02 1.06156208e-01 -4.46702652e-02
-3.45202416e-01 1.17136955e+00 -8.04084003e-01 1.05307412e+00
-3.89404893e-01 6.64789021e-01 1.70769215e-01 -1.03876770e+00
7.41486907e-01 -1.95732918e-02 -5.84320016e-02 -7.75526941e-01
3.09660643e-01 -1.28774658e-01 2.24360041e-02 -7.76018679e-01
6.79833829e-01 -3.81978452e-01 -4.51854229e-01 9.67940390e-01
-2.71054745e-01 -5.59988394e-02 5.64134538e-01 6.48228347e-01
1.55392325e+00 -5.50029695e-01 7.23975539e-01 -3.35051745e-01
6.00978434e-01 3.21327478e-01 3.67605925e-01 9.58664179e-01
-4.28094506e-01 1.53456911e-01 9.44532752e-01 -6.77478194e-01
-8.14726770e-01 -7.34777272e-01 -1.25294432e-01 1.49133229e+00
-3.42134237e-01 -7.54684687e-01 -7.07093179e-01 -1.15548372e+00
-2.49189973e-01 1.09751034e+00 -9.72995579e-01 -1.02370940e-01
-9.84000444e-01 -8.46704066e-01 5.86287558e-01 2.11103737e-01
2.94440120e-01 -1.20636284e+00 -9.95575786e-01 9.76030976e-02
-4.28863406e-01 -5.38703978e-01 -3.38530838e-01 4.16237831e-01
-6.35999858e-01 -1.23435318e+00 -5.26464283e-02 -3.68457705e-01
3.62044007e-01 8.46251771e-02 1.25655913e+00 1.77722290e-01
-4.80756909e-01 -3.94836158e-01 -6.98855400e-01 -2.50506192e-01
-8.75547290e-01 2.16443896e-01 -1.29288390e-01 -4.02862161e-01
3.21575105e-01 -1.09889589e-01 2.88782232e-02 -3.04122031e-01
-8.90563250e-01 -3.47255021e-02 2.41151020e-01 9.56897378e-01
-2.36083195e-01 -3.51464339e-02 6.18187606e-01 -1.18565345e+00
1.31487799e+00 -4.10038263e-01 -1.94374979e-01 1.82655588e-01
-1.63358390e-01 1.07096590e-01 5.72036982e-01 -1.10082977e-01
-8.76088321e-01 -3.89390379e-01 -3.96610081e-01 3.89488161e-01
-3.23090643e-01 7.61263788e-01 1.66946828e-01 4.47242230e-01
1.21631813e+00 -4.73348238e-02 -1.87125131e-01 -5.12325943e-01
3.02079529e-01 9.22609508e-01 1.55068174e-01 -4.45969343e-01
5.65793097e-01 2.58037518e-03 -5.48727214e-01 -6.91051602e-01
-1.34318388e+00 -6.35763705e-01 -4.91251469e-01 6.93063140e-02
8.25029731e-01 -6.24218345e-01 -1.09501757e-01 4.05830950e-01
-1.52496362e+00 -1.55825540e-01 1.99004710e-02 -4.40338301e-03
-2.14481711e-01 3.39974046e-01 -9.29045320e-01 -9.52824414e-01
-7.02689111e-01 -8.40070426e-01 9.23259974e-01 -1.78733408e-01
-5.13365567e-01 -8.43881190e-01 4.77019519e-01 7.50739336e-01
2.30566025e-01 2.10693091e-01 1.50184619e+00 -1.04615009e+00
3.26496273e-01 -2.75350392e-01 -2.77256876e-01 2.84607768e-01
9.92235988e-02 1.02302849e-01 -7.00870395e-01 -1.63700968e-01
3.48915279e-01 -5.28485179e-01 1.00784087e+00 2.90560097e-01
9.47915494e-01 -7.75875151e-01 -8.57986733e-02 -1.08212359e-01
9.53197777e-01 -3.08995008e-01 5.24945796e-01 5.80816329e-01
4.25523579e-01 7.17269599e-01 6.42315626e-01 6.02447629e-01
2.58398890e-01 7.23081291e-01 1.95877835e-01 2.85572350e-01
-6.95230737e-02 -8.84773061e-02 5.69079697e-01 5.69356382e-01
3.04866195e-01 -3.45258772e-01 -1.08195889e+00 4.97041970e-01
-1.66832829e+00 -1.48949623e+00 -4.61455405e-01 1.74748981e+00
1.05897474e+00 2.98126876e-01 3.53127301e-01 3.56121033e-01
6.93701088e-01 6.75776660e-01 1.92844525e-01 -1.29730535e+00
6.91725165e-02 7.60799125e-02 -6.08997233e-02 6.48748696e-01
-1.47181714e+00 8.95536780e-01 7.30559349e+00 1.10866332e+00
-9.27241087e-01 2.02775389e-01 5.51707804e-01 -1.15578227e-01
-1.81660846e-01 -1.87398121e-01 -8.01538467e-01 5.71785212e-01
1.17932129e+00 2.06305794e-02 -3.96605060e-02 6.17363572e-01
4.49655563e-01 -4.99143332e-01 -8.84857059e-01 4.24346477e-01
3.17088783e-01 -1.50394821e+00 1.89520642e-01 -1.76967427e-01
4.88626242e-01 -5.54662421e-02 -3.04806411e-01 4.73773658e-01
1.99939698e-01 -8.64539385e-01 8.14553499e-01 -1.78802796e-02
4.63682711e-02 -5.40818572e-01 7.61565149e-01 7.77719975e-01
-5.34826279e-01 -6.83699697e-02 -3.20295128e-03 -6.04570448e-01
3.78475726e-01 1.03173327e+00 -9.86268222e-01 4.24411505e-01
4.37569827e-01 4.84852999e-01 -7.73968756e-01 5.21684349e-01
-6.17808878e-01 1.08566952e+00 -9.17008370e-02 -3.54359567e-01
2.45192572e-01 3.22117537e-01 6.93743169e-01 1.67269862e+00
-2.72336751e-01 -1.14671081e-01 2.95426995e-01 5.77517807e-01
1.04125310e-02 2.68404037e-01 -2.93752968e-01 -1.61144078e-01
3.38667721e-01 1.21162975e+00 -4.69493330e-01 -5.09693503e-01
3.07559997e-01 6.95595086e-01 6.12589896e-01 -3.75693925e-02
-5.22954643e-01 -4.39551711e-01 4.65996891e-01 1.88405365e-01
-6.94810459e-03 -1.50079712e-01 -6.86292350e-01 -1.09165609e+00
-1.25049844e-01 -1.13231468e+00 7.80123293e-01 -4.99779493e-01
-1.38755786e+00 4.84812021e-01 2.46373285e-02 -5.46361566e-01
-5.75710475e-01 -4.69810098e-01 -9.57742274e-01 8.31721961e-01
-1.01432765e+00 -7.87560821e-01 -1.03989653e-02 2.69836843e-01
7.17862308e-01 1.01973958e-01 7.29913652e-01 4.38892804e-02
-5.12507439e-01 2.97187448e-01 -2.95747936e-01 4.35918480e-01
8.35672021e-01 -9.81301188e-01 2.96501428e-01 9.23373282e-01
3.28210212e-04 8.12397480e-01 1.18204618e+00 -9.30243671e-01
-9.58357811e-01 -8.84551823e-01 1.59070873e+00 -6.71656668e-01
1.02790785e+00 -5.09002030e-01 -9.05632555e-01 5.36994860e-02
2.17973143e-01 -4.67903614e-01 7.37276018e-01 7.35213220e-01
-1.04347062e+00 3.73583466e-01 -1.31136942e+00 2.56440789e-01
5.92773318e-01 -6.75222576e-01 -1.28952885e+00 7.87802219e-01
3.34812224e-01 -2.28898436e-01 -3.31467688e-01 1.37812793e-01
1.95506960e-01 -8.22644651e-01 6.41632736e-01 -1.14809871e+00
1.06592512e+00 7.95953870e-02 1.51381418e-01 -1.59568894e+00
-4.27179694e-01 -1.82999879e-01 -1.25939935e-01 1.14148307e+00
2.25817204e-01 -2.38934338e-01 5.05777836e-01 -2.31494933e-01
1.86660010e-02 -6.24672651e-01 -1.09551620e+00 -6.53354168e-01
3.82001191e-01 -1.85803607e-01 1.37677332e-02 1.06984448e+00
4.81537431e-01 8.17494392e-01 -4.44718808e-01 -1.81871846e-01
4.54639196e-01 4.18028086e-01 5.02318561e-01 -8.87226760e-01
-1.91986337e-01 -6.24418616e-01 -2.55914539e-01 -6.10202670e-01
1.99364603e-01 -8.25460851e-01 1.70769364e-01 -1.61138022e+00
5.37855685e-01 -8.84577334e-02 7.94791058e-02 5.75456023e-01
-5.91461360e-01 9.82318148e-02 4.06080872e-01 8.46615657e-02
-5.19428611e-01 3.12636018e-01 7.34988689e-01 -2.82231569e-01
-2.10842937e-01 -2.07605988e-01 -6.99561059e-01 5.30645192e-01
9.32564974e-01 -9.06548738e-01 5.52586243e-02 -4.92671162e-01
3.58374655e-01 -8.92244726e-02 3.44174832e-01 -6.34267986e-01
-8.35533440e-02 -4.78797734e-01 6.87813237e-02 -9.10347581e-01
-7.11130202e-02 9.70836580e-02 -6.98961794e-01 7.85347939e-01
-6.99303389e-01 4.05641720e-02 5.93399480e-02 3.51887554e-01
-1.17963597e-01 -6.08733356e-01 8.37214649e-01 -2.05445867e-02
-2.50879559e-03 -3.99412721e-01 -1.05076325e+00 3.55289102e-01
7.49234676e-01 2.30090156e-01 -1.15262222e+00 -8.11923295e-02
-3.80941451e-01 -3.01558543e-02 2.41832286e-01 4.56407815e-01
3.27932298e-01 -9.76910055e-01 -1.06258261e+00 6.73790425e-02
2.31495991e-01 -7.45922446e-01 2.05054730e-01 8.89926672e-01
-6.04536295e-01 2.49240085e-01 -1.01770930e-01 -2.76406288e-01
-1.57331729e+00 4.79939342e-01 -1.69560820e-01 -8.67751539e-01
-4.37303871e-01 9.25825357e-01 -2.89146781e-01 -3.21487874e-01
-1.60135671e-01 5.80475181e-02 -3.53557289e-01 2.52023011e-01
1.12472177e+00 3.78215641e-01 3.58489662e-01 -4.58357990e-01
-7.44663477e-01 -1.38220087e-01 -4.67394024e-01 2.27325615e-02
1.29222476e+00 3.50443035e-01 -5.29580176e-01 3.93160731e-01
1.09247160e+00 2.78257817e-01 -2.19409078e-01 5.29218465e-02
2.43184403e-01 -4.69448149e-01 1.90648988e-01 -1.14657927e+00
-2.33430430e-01 9.96294916e-01 2.57614434e-01 5.85564196e-01
7.74861336e-01 5.69568612e-02 4.57039654e-01 3.27205062e-01
3.76861244e-02 -1.37123120e+00 1.32253289e-01 1.00757992e+00
1.01800144e+00 -9.10506427e-01 3.47644091e-01 -5.32602787e-01
-3.23217034e-01 1.00251448e+00 4.56724107e-01 -1.48337319e-01
2.86023170e-01 3.45928311e-01 2.05204740e-01 -5.55172205e-01
-9.35283244e-01 1.65644139e-01 3.50316703e-01 3.88592690e-01
9.10264730e-01 1.51464090e-01 -9.92412329e-01 3.04578453e-01
-1.75039262e-01 -4.59966034e-01 5.95643163e-01 1.25747216e+00
-9.45352018e-01 -1.03262174e+00 -7.37235546e-01 8.00652266e-01
-4.82203573e-01 -3.07906002e-01 -9.99504030e-01 2.41166309e-01
-5.33780083e-02 1.22953677e+00 3.60886268e-02 -4.08749968e-01
2.40714222e-01 5.36302850e-02 5.44071376e-01 -9.84952033e-01
-1.48141301e+00 -1.02257229e-01 7.78888762e-01 -4.87803161e-01
-3.43512714e-01 -6.72478378e-01 -9.57578719e-01 -5.47769189e-01
-3.18885565e-01 3.06460202e-01 6.69990480e-01 1.22247708e+00
2.10674644e-01 5.67735076e-01 6.57625020e-01 -6.42823040e-01
-1.10382080e+00 -1.39978874e+00 -2.02990338e-01 6.47448242e-01
3.42192620e-01 -4.01097178e-01 -5.22220492e-01 -1.83864728e-01] | [8.498794555664062, 10.620155334472656] |
6f37ba16-cadd-4220-9fd4-079c1b5aea91 | monocular-3d-human-pose-estimation-in-the | 1611.09813 | null | http://arxiv.org/abs/1611.09813v5 | http://arxiv.org/pdf/1611.09813v5.pdf | Monocular 3D Human Pose Estimation In The Wild Using Improved CNN Supervision | We propose a CNN-based approach for 3D human body pose estimation from single
RGB images that addresses the issue of limited generalizability of models
trained solely on the starkly limited publicly available 3D pose data. Using
only the existing 3D pose data and 2D pose data, we show state-of-the-art
performance on established benchmarks through transfer of learned features,
while also generalizing to in-the-wild scenes. We further introduce a new
training set for human body pose estimation from monocular images of real
humans that has the ground truth captured with a multi-camera marker-less
motion capture system. It complements existing corpora with greater diversity
in pose, human appearance, clothing, occlusion, and viewpoints, and enables an
increased scope of augmentation. We also contribute a new benchmark that covers
outdoor and indoor scenes, and demonstrate that our 3D pose dataset shows
better in-the-wild performance than existing annotated data, which is further
improved in conjunction with transfer learning from 2D pose data. All in all,
we argue that the use of transfer learning of representations in tandem with
algorithmic and data contributions is crucial for general 3D body pose
estimation. | ['Weipeng Xu', 'Pascal Fua', 'Oleksandr Sotnychenko', 'Helge Rhodin', 'Dushyant Mehta', 'Dan Casas', 'Christian Theobalt'] | 2016-11-29 | null | null | null | null | ['monocular-3d-human-pose-estimation'] | ['computer-vision'] | [ 4.81337123e-02 5.19349203e-02 -1.98440969e-01 -4.02838856e-01
-8.81760418e-01 -4.19038653e-01 2.75425524e-01 -3.04291606e-01
-5.48598170e-01 6.14568830e-01 5.69183528e-01 3.61271858e-01
1.60953134e-01 -3.78518909e-01 -8.72235954e-01 -2.84159601e-01
-1.74666479e-01 8.52174520e-01 4.12007533e-02 -5.43827415e-01
-4.54819113e-01 4.01865095e-01 -1.39641964e+00 -1.71827152e-01
3.58400606e-02 1.02430427e+00 -3.93453211e-01 7.35051632e-01
8.61626923e-01 3.30069840e-01 -2.87230253e-01 -4.64286536e-01
6.35968506e-01 -3.76664728e-01 -7.94342518e-01 4.02839750e-01
1.35188830e+00 -7.31529653e-01 -4.83799458e-01 3.17208737e-01
9.44622219e-01 -2.60015409e-02 3.07848722e-01 -1.11425519e+00
-3.78014147e-01 -2.14652091e-01 -4.36090559e-01 -8.00835260e-04
9.59253430e-01 3.71732652e-01 8.21213067e-01 -6.56592369e-01
8.88105154e-01 1.24549878e+00 1.30136073e+00 8.43660235e-01
-1.27104223e+00 -3.62741739e-01 3.00949551e-02 -2.63132483e-01
-1.11559224e+00 -2.37383127e-01 6.13054872e-01 -5.87284923e-01
9.11359370e-01 1.29177824e-01 1.34685755e+00 1.40059161e+00
4.24649902e-02 9.16015625e-01 1.06025779e+00 -4.18962002e-01
-3.01584989e-01 -5.80697954e-01 -3.36407721e-01 9.60788667e-01
4.63254541e-01 3.29281181e-01 -6.67054713e-01 -1.19790889e-01
1.10081470e+00 -8.77055973e-02 -2.72113264e-01 -1.14421785e+00
-1.44672012e+00 5.58860719e-01 7.22738624e-01 -2.43570372e-01
-1.94965467e-01 5.14606118e-01 4.94541258e-01 -1.21015068e-02
5.85132539e-01 4.20050800e-01 -8.89094055e-01 -2.37734929e-01
-9.57410514e-01 7.78914511e-01 7.22126365e-01 1.03512120e+00
4.92365658e-01 -1.44318536e-01 1.21516950e-01 5.50200999e-01
3.98205161e-01 8.37357521e-01 2.45579392e-01 -1.10117590e+00
5.30131340e-01 5.90664208e-01 1.17218524e-01 -8.23475063e-01
-8.70257497e-01 -4.88681495e-01 -3.52781564e-01 2.10454389e-01
7.09801257e-01 -1.59921348e-01 -9.30950820e-01 1.83978176e+00
6.77286744e-01 -2.58961052e-01 -3.79751593e-01 1.36838162e+00
9.30481076e-01 -1.38486683e-01 -1.45674152e-02 3.96329701e-01
1.32163835e+00 -1.04054546e+00 -2.29527161e-01 -6.25393867e-01
4.32769716e-01 -5.02430201e-01 8.79127026e-01 3.86177987e-01
-1.13257957e+00 -6.02943301e-01 -1.09298599e+00 -5.20154834e-01
-1.14524394e-01 3.21790844e-01 8.63455772e-01 7.28839755e-01
-9.31438327e-01 6.47054791e-01 -1.17899418e+00 -8.93009603e-01
2.76336968e-01 6.41073763e-01 -9.02291238e-01 7.93050751e-02
-8.92125189e-01 1.11283588e+00 -1.08653912e-02 2.69067883e-01
-7.61145711e-01 -5.82785845e-01 -1.39856756e+00 -7.62994945e-01
3.62890005e-01 -1.35283172e+00 1.36346567e+00 -5.44041812e-01
-1.37801945e+00 1.56050539e+00 4.05406266e-01 -3.83600742e-01
9.82251346e-01 -9.35802579e-01 5.12090921e-02 1.66512609e-01
8.81177410e-02 8.58233094e-01 7.19201744e-01 -1.16525793e+00
-1.12116709e-01 -8.13177109e-01 1.47530055e-02 3.93658876e-01
4.64008078e-02 -2.32129484e-01 -7.54256964e-01 -4.72418398e-01
2.80353963e-01 -1.42062521e+00 -1.87450469e-01 4.40801680e-01
-2.41983294e-01 2.95071781e-01 3.30218256e-01 -8.68350923e-01
5.96100390e-01 -1.70219564e+00 6.18181288e-01 -1.14895135e-01
2.01281697e-01 -7.65506402e-02 1.91547096e-01 3.88724357e-01
3.99768688e-02 -3.44211876e-01 2.51176930e-03 -6.18911088e-01
8.80315751e-02 2.46048376e-01 3.45085144e-01 9.34519172e-01
1.86073124e-01 1.07189643e+00 -9.28810596e-01 -6.81200206e-01
4.67891604e-01 4.30447817e-01 -8.44391644e-01 2.44833201e-01
3.57306488e-02 8.97865534e-01 -5.67981265e-02 9.60004866e-01
2.51834303e-01 -3.33381176e-01 1.53487176e-01 -4.75400209e-01
4.07979727e-01 1.15924753e-01 -9.85985637e-01 2.58577347e+00
-8.95207003e-02 2.96627313e-01 5.99132031e-02 -3.94688725e-01
5.92830420e-01 2.64010072e-01 8.83059502e-01 -2.41616547e-01
2.95069367e-01 1.67961910e-01 -7.09738955e-02 -2.92509526e-01
4.13589865e-01 -2.20496714e-01 -3.61407995e-01 1.26962557e-01
5.78049123e-01 -4.21942145e-01 -1.70698538e-01 -1.62887841e-01
9.72184837e-01 1.10727632e+00 4.60607588e-01 -2.08136532e-02
5.70320971e-02 1.17742226e-01 3.81796271e-01 3.41035098e-01
-4.61670041e-01 9.83937562e-01 -4.10585403e-02 -7.69742668e-01
-1.21928728e+00 -1.27310681e+00 -1.20381907e-01 1.12102199e+00
4.22636978e-03 -5.49256921e-01 -6.22431457e-01 -5.48693419e-01
4.85622168e-01 -2.60084033e-01 -1.05719507e+00 3.54283340e-02
-8.14970076e-01 -5.24807572e-01 7.95931160e-01 9.74656224e-01
4.82988417e-01 -6.38097525e-01 -1.09417474e+00 2.97018401e-02
-6.12597018e-02 -1.25326800e+00 -3.77288222e-01 9.93268341e-02
-1.06961620e+00 -1.33454382e+00 -9.70184565e-01 -5.67590296e-01
4.86856908e-01 -2.34123811e-01 1.44836164e+00 -1.26037791e-01
-5.09927988e-01 9.48074579e-01 -3.48895580e-01 -2.22346112e-01
1.61243737e-01 7.87800178e-02 2.60877162e-01 -5.94692290e-01
1.59493774e-01 -3.38138163e-01 -8.62826347e-01 4.78527069e-01
-6.36426210e-02 8.09414908e-02 4.69668210e-01 9.72568393e-01
4.84989405e-01 -9.51622128e-01 -4.08599898e-02 -4.51581001e-01
2.69675013e-02 -5.34351543e-02 -1.56863615e-01 -1.58935606e-01
-1.72862217e-01 -2.20657676e-01 -1.17907859e-01 -2.42648140e-01
-7.87672579e-01 4.52966154e-01 -2.86033541e-01 -3.91864300e-01
-3.50369513e-01 6.65283874e-02 -6.45287782e-02 -2.43913159e-01
8.97985160e-01 -2.63186961e-01 4.31159288e-01 -6.21050835e-01
4.91429389e-01 1.30213991e-01 1.00752902e+00 -7.57814467e-01
6.62410796e-01 7.47174799e-01 2.36620426e-01 -4.11705494e-01
-8.61540973e-01 -5.19708276e-01 -1.34684801e+00 -2.96009868e-01
9.44288254e-01 -1.42798555e+00 -6.78568661e-01 5.51358879e-01
-6.96893156e-01 -5.03832400e-01 -3.63212854e-01 6.93368375e-01
-1.12920964e+00 3.75576258e-01 -9.36545551e-01 -6.29187584e-01
-4.14507627e-01 -9.19956982e-01 1.93464077e+00 -2.89413720e-01
-8.39426816e-01 -7.94995844e-01 3.07364970e-01 6.57092631e-01
-8.46860185e-02 9.75498080e-01 2.34168321e-01 -2.72505760e-01
-1.52937949e-01 -6.80526435e-01 2.16829151e-01 2.68714935e-01
5.04347607e-02 -4.81186718e-01 -8.79106283e-01 -6.09565914e-01
-4.05079931e-01 -9.46242034e-01 5.81613004e-01 4.88955170e-01
4.83739644e-01 2.08128452e-01 -2.17720583e-01 8.62778604e-01
1.03658998e+00 -7.03130662e-01 3.53641301e-01 5.42737305e-01
9.70294654e-01 4.93883371e-01 7.52295613e-01 4.31123406e-01
6.29648447e-01 1.05321050e+00 4.83414948e-01 -4.35582221e-01
-4.32846695e-01 -5.52182436e-01 1.03639416e-01 6.05062187e-01
-6.38222933e-01 3.43916804e-01 -8.86877000e-01 3.91408354e-01
-1.75679803e+00 -5.48204184e-01 1.20526403e-01 2.21249413e+00
8.69747698e-01 -2.61822231e-02 8.03634465e-01 1.99316964e-01
3.30953091e-01 4.41721119e-02 -5.52801669e-01 1.14890549e-03
1.67595580e-01 3.36587042e-01 7.35608578e-01 1.78463802e-01
-1.46685922e+00 7.90592492e-01 7.24362707e+00 2.30572615e-02
-9.28964972e-01 1.45785078e-01 6.50352389e-02 -5.19069612e-01
2.82602847e-01 -4.01059300e-01 -7.59425879e-01 5.74745871e-02
5.50730228e-01 5.41509867e-01 1.07722819e-01 9.75733042e-01
-2.67241180e-01 3.02079022e-02 -1.51798332e+00 1.34289658e+00
4.98223513e-01 -7.36301959e-01 -4.96428549e-01 3.56976271e-01
6.87673450e-01 1.60292462e-01 -1.23429634e-01 1.98035091e-01
1.30825683e-01 -9.60796416e-01 1.01817262e+00 3.85049403e-01
9.51526761e-01 -3.48299325e-01 6.60248518e-01 2.18633324e-01
-1.12535059e+00 1.51109695e-01 -5.14283217e-02 -5.29950261e-01
1.96450263e-01 -2.52315123e-02 -6.95556045e-01 6.60145700e-01
9.15802240e-01 8.82629395e-01 -7.63246298e-01 9.68674839e-01
-2.17272013e-01 1.17692553e-01 -4.69543785e-01 1.60797969e-01
-1.58528298e-01 2.13293642e-01 2.45546132e-01 9.86517787e-01
1.82601679e-02 -2.12986410e-01 4.96378750e-01 4.97732967e-01
-6.93595558e-02 4.04952988e-02 -6.88272119e-01 3.60789210e-01
-3.71357463e-02 1.12555802e+00 -4.16855663e-01 -6.31869584e-02
-3.50136071e-01 1.05178022e+00 3.02642167e-01 -6.05043396e-02
-8.23125064e-01 2.89714843e-01 6.13325536e-01 3.81454140e-01
2.32046366e-01 -6.48448169e-01 -1.88419700e-01 -1.21910787e+00
7.77253434e-02 -9.14385438e-01 3.48812729e-01 -8.57016325e-01
-1.38231063e+00 3.06607634e-01 3.46901476e-01 -1.35995579e+00
-6.09857917e-01 -9.93354201e-01 1.58878356e-01 6.19189799e-01
-9.89766657e-01 -1.75235403e+00 -5.86147189e-01 6.69902682e-01
2.68966049e-01 2.89798349e-01 9.98083472e-01 2.69508958e-01
-1.84897304e-01 6.49357736e-01 -5.35745621e-01 4.32485461e-01
1.11685646e+00 -1.38110352e+00 7.64813542e-01 3.94011468e-01
1.20639935e-01 5.76360583e-01 6.42633557e-01 -6.54967010e-01
-1.73077238e+00 -7.27670133e-01 4.60359484e-01 -1.39501929e+00
2.00131372e-01 -6.11716926e-01 -9.35627222e-02 1.12272584e+00
-2.08797842e-01 3.45266163e-01 8.43452513e-01 4.91911739e-01
-3.29115123e-01 7.77075514e-02 -1.18329322e+00 3.75481963e-01
1.61130047e+00 -4.26957995e-01 -9.13516045e-01 2.88283825e-01
4.71354902e-01 -1.29556394e+00 -1.19019568e+00 7.70159900e-01
1.27844596e+00 -7.84987986e-01 1.34285355e+00 -7.38782465e-01
4.35547143e-01 -1.74952060e-01 -4.40992296e-01 -1.06210053e+00
-3.07792127e-01 -3.08163255e-01 -1.35756940e-01 6.24522805e-01
-1.13175120e-02 -1.42285019e-01 1.19341028e+00 6.26138866e-01
-1.52029023e-01 -7.24720478e-01 -9.06087995e-01 -8.15872729e-01
1.30023062e-01 -4.53245312e-01 3.42821360e-01 5.01080453e-01
-1.83743283e-01 9.58616436e-02 -8.27224493e-01 -1.04166545e-01
6.68035626e-01 1.53342739e-01 1.55930233e+00 -1.27251601e+00
-4.82427537e-01 1.50654688e-01 -1.06582034e+00 -1.26697230e+00
-1.88532826e-02 -4.80296433e-01 2.44422287e-01 -1.38973057e+00
2.02341616e-01 8.55004564e-02 1.30911008e-01 5.96502602e-01
2.07414627e-02 9.81144130e-01 3.92435849e-01 1.89756453e-01
-7.86099792e-01 4.25702870e-01 1.45277107e+00 1.37807548e-01
2.70047188e-01 -1.72788084e-01 -3.48546028e-01 9.66539860e-01
2.69333661e-01 -1.35631055e-01 -4.89118472e-02 -5.04057050e-01
2.31597915e-01 -1.63997248e-01 9.01472449e-01 -1.38036776e+00
-5.97930066e-02 3.26716006e-01 1.08204806e+00 -6.36283457e-01
9.95655417e-01 -7.47990906e-01 1.43558487e-01 6.08901441e-01
-1.01951048e-01 1.17366299e-01 2.87464380e-01 6.06328309e-01
2.18732625e-01 4.91015404e-01 4.81943041e-01 -7.04469442e-01
-7.53366470e-01 4.12220746e-01 2.55977780e-01 3.17026407e-01
7.19186008e-01 -5.19418061e-01 1.41967274e-02 -3.62127334e-01
-1.07352352e+00 1.06780706e-02 1.01677406e+00 4.48564917e-01
3.89852405e-01 -1.56054389e+00 -5.72032809e-01 4.05604362e-01
3.55248749e-01 9.60279629e-02 -8.08178112e-02 8.68003786e-01
-7.68316805e-01 4.28400636e-01 -6.05698049e-01 -1.02799428e+00
-1.27850807e+00 1.73162445e-01 4.00757790e-01 -1.03462726e-01
-6.87651277e-01 8.51548553e-01 -3.11883837e-01 -1.00628066e+00
1.34526014e-01 -5.04480422e-01 4.86216038e-01 -2.87778050e-01
-4.42257822e-02 3.50572318e-01 1.55169204e-01 -1.09041822e+00
-6.17819607e-01 8.87071192e-01 4.74927485e-01 -1.28737628e-01
1.21440995e+00 -5.89760281e-02 2.44665861e-01 5.21830857e-01
1.10708725e+00 7.61226937e-02 -1.62261498e+00 -5.93549982e-02
-3.14563662e-01 -7.14319825e-01 -3.89914095e-01 -9.40002441e-01
-8.81390810e-01 8.00815046e-01 7.55995512e-01 -5.34058750e-01
8.08853090e-01 2.56932825e-01 8.33160162e-01 1.98116720e-01
8.85904670e-01 -1.02863348e+00 2.79692113e-01 3.67314517e-01
1.00455117e+00 -1.41014349e+00 5.26478946e-01 -3.98989290e-01
-5.63040853e-01 9.16365325e-01 7.41067946e-01 -4.78756100e-01
4.11497533e-01 6.20562062e-02 8.92130062e-02 -3.89807194e-01
-1.85428038e-01 -4.35355037e-01 6.60099268e-01 9.86344218e-01
7.53679216e-01 1.28893003e-01 -1.45771682e-01 5.14423847e-01
-5.96977830e-01 1.31135136e-01 -1.32396296e-01 1.21669364e+00
3.98717634e-02 -1.12522030e+00 -5.96370876e-01 1.62073925e-01
-4.79758888e-01 3.15206230e-01 -6.00137711e-01 1.34061980e+00
4.07207966e-01 5.77197075e-01 -7.92363286e-02 -4.63737160e-01
6.12560689e-01 4.06306759e-02 1.36462533e+00 -8.98586035e-01
-5.82049489e-01 2.38842502e-01 3.93751174e-01 -9.31856871e-01
-6.46450996e-01 -7.36760139e-01 -8.25312316e-01 -1.46597594e-01
-2.21995562e-01 -5.04325449e-01 5.95230877e-01 8.90406668e-01
2.40603089e-01 3.46624404e-01 -1.88345954e-01 -1.69685090e+00
-5.79787314e-01 -1.03679180e+00 -6.14626944e-01 8.24716687e-01
3.18197519e-01 -1.12062013e+00 6.47564754e-02 8.99079219e-02] | [6.9926581382751465, -0.9354548454284668] |
ca409013-f954-4e6a-a4b4-09a48de674aa | condensed-prototype-replay-for-class | 2305.16143 | null | https://arxiv.org/abs/2305.16143v1 | https://arxiv.org/pdf/2305.16143v1.pdf | Condensed Prototype Replay for Class Incremental Learning | Incremental learning (IL) suffers from catastrophic forgetting of old tasks when learning new tasks. This can be addressed by replaying previous tasks' data stored in a memory, which however is usually prone to size limits and privacy leakage. Recent studies store only class centroids as prototypes and augment them with Gaussian noises to create synthetic data for replay. However, they cannot effectively avoid class interference near their margins that leads to forgetting. Moreover, the injected noises distort the rich structure between real data and prototypes, hence even detrimental to IL. In this paper, we propose YONO that You Only Need to replay One condensed prototype per class, which for the first time can even outperform memory-costly exemplar-replay methods. To this end, we develop a novel prototype learning method that (1) searches for more representative prototypes in high-density regions by an attentional mean-shift algorithm and (2) moves samples in each class to their prototype to form a compact cluster distant from other classes. Thereby, the class margins are maximized, which effectively reduces interference causing future forgetting. In addition, we extend YONO to YONO+, which creates synthetic replay data by random sampling in the neighborhood of each prototype in the representation space. We show that the synthetic data can further improve YONO. Extensive experiments on IL benchmarks demonstrate the advantages of YONO/YONO+ over existing IL methods in terms of both accuracy and forgetting. | ['Huajie Shao', 'Tianyi Zhou', 'Zhenyu Zong', 'Jiangtao Kong'] | 2023-05-25 | null | null | null | null | ['class-incremental-learning', 'incremental-learning'] | ['computer-vision', 'methodology'] | [ 2.06008807e-01 -6.15767390e-02 -2.56685346e-01 -1.36680514e-01
-4.32509333e-01 -3.51477742e-01 4.98809874e-01 3.36166382e-01
-6.26318097e-01 1.07151008e+00 -3.80525477e-02 9.54409037e-03
-1.80237576e-01 -6.54043853e-01 -9.80983257e-01 -8.90925169e-01
6.63089454e-02 4.88949418e-01 3.21779162e-01 2.77792335e-01
2.90798992e-01 3.45561147e-01 -1.80878460e+00 2.42043033e-01
1.21184337e+00 7.30631649e-01 4.61993545e-01 1.29258320e-01
-1.51679710e-01 3.98687452e-01 -9.65019166e-01 -2.88698763e-01
2.55894452e-01 -1.83711633e-01 -4.35377240e-01 9.48596150e-02
3.56477737e-01 -1.69406056e-01 -4.99599427e-01 1.10432613e+00
4.56953645e-01 6.06786847e-01 4.79647487e-01 -1.31959414e+00
-9.68115389e-01 7.32855856e-01 -6.37948394e-01 5.55147678e-02
-3.16130333e-02 1.48255017e-03 3.19626182e-01 -1.34672189e+00
6.20641470e-01 1.08832681e+00 7.87432909e-01 7.30880797e-01
-1.46180427e+00 -1.04399931e+00 4.39641237e-01 1.92398041e-01
-1.76763582e+00 -6.00630283e-01 8.60733688e-01 -9.85770673e-02
6.51707232e-01 3.44697177e-01 6.34794950e-01 1.32763815e+00
1.37045771e-01 1.03427124e+00 1.01950037e+00 -3.87845308e-01
6.69813037e-01 5.20935476e-01 4.19451147e-01 3.89002860e-01
7.38170087e-01 -6.05128370e-02 -7.10720062e-01 -3.06688964e-01
4.16170865e-01 5.24490356e-01 -5.59149504e-01 -8.25234890e-01
-1.19345665e+00 6.84274912e-01 5.61203241e-01 2.25798443e-01
-1.57561794e-01 -1.48235589e-01 4.10034895e-01 4.48726863e-01
4.38704133e-01 4.71715301e-01 -2.33910054e-01 1.97090745e-01
-9.33290422e-01 2.51405239e-01 4.27530140e-01 1.11223817e+00
8.68761003e-01 -1.49643987e-01 -2.17283756e-01 9.79014754e-01
-2.02699974e-01 4.24462646e-01 1.03641617e+00 -7.05639958e-01
3.79890680e-01 5.82178414e-01 2.49201700e-01 -8.60627651e-01
-1.91732854e-01 -7.21100986e-01 -1.05776608e+00 -1.03679700e-02
2.51675725e-01 1.94452465e-01 -9.01534677e-01 1.78524816e+00
3.77583623e-01 4.26820129e-01 2.23258901e-02 6.53322935e-01
2.52354711e-01 5.35143435e-01 -1.59330428e-01 -6.30308390e-01
9.36646342e-01 -9.61707830e-01 -8.47618103e-01 -3.65394980e-01
5.84347486e-01 -3.72734964e-01 1.44328904e+00 4.05776262e-01
-7.09422052e-01 -7.58447051e-01 -1.20487428e+00 1.23201780e-01
-4.09857154e-01 -7.29061067e-02 4.93470341e-01 6.66693747e-01
-6.58709526e-01 5.46822309e-01 -8.29024851e-01 8.83940235e-02
9.30581808e-01 1.64291427e-01 -2.44302168e-01 -3.68119448e-01
-1.02029097e+00 4.87767458e-01 6.09084070e-01 -1.87104911e-01
-7.50632465e-01 -1.01562190e+00 -7.15052485e-01 -2.06505414e-02
6.42735064e-01 -7.30141401e-01 9.89858806e-01 -7.40271628e-01
-1.00236213e+00 2.64496565e-01 -4.95266438e-01 -7.59112895e-01
7.04943538e-01 -4.63676065e-01 -4.58715141e-01 -2.28895530e-01
7.13532344e-02 7.13173509e-01 1.34615350e+00 -1.46433175e+00
-4.84836310e-01 -4.57220137e-01 -5.52661657e-01 2.38910303e-01
-6.63678348e-01 -8.59503865e-01 -3.42142045e-01 -9.00645196e-01
1.06742673e-01 -8.98773193e-01 -4.65198187e-03 9.34371576e-02
-2.25265756e-01 -2.19410092e-01 1.16672230e+00 -5.31594120e-02
1.52557182e+00 -2.50410962e+00 -1.65415511e-01 8.40643644e-02
4.96786177e-01 5.56447029e-01 -1.24296043e-02 -1.09997444e-01
7.19954297e-02 -6.16033003e-02 -2.43338555e-01 -6.83127642e-01
-2.05119625e-01 4.12973225e-01 -8.03065777e-01 4.24329072e-01
-2.29777828e-01 1.02149189e+00 -1.09305584e+00 -2.62403220e-01
-8.87210947e-03 4.05407995e-01 -5.02402544e-01 -1.21710271e-01
-1.46090865e-01 7.93430507e-02 -4.24545258e-02 5.06716013e-01
9.60006535e-01 -4.41669486e-02 -3.94459665e-02 1.44052565e-01
2.51614630e-01 -8.87794271e-02 -1.04870021e+00 1.66885161e+00
-3.21827024e-01 4.46881592e-01 -3.75297517e-01 -8.34833205e-01
1.22054589e+00 -1.04073338e-01 7.26083014e-03 -5.94427764e-01
-2.50076652e-01 1.63798913e-01 -3.72209519e-01 -1.24503285e-01
6.73768401e-01 -1.02831893e-01 -6.84167519e-02 4.96528715e-01
-1.75224513e-01 1.49299309e-01 -7.67341852e-02 2.76543796e-01
8.86427522e-01 -3.02261978e-01 1.77329466e-01 -3.51570360e-02
1.82049781e-01 -2.63528019e-01 8.14285934e-01 1.19991136e+00
-2.70361781e-01 5.65037012e-01 7.34267086e-02 -5.28822064e-01
-9.39840555e-01 -1.40848327e+00 -2.69406199e-01 9.50105309e-01
9.61696506e-02 -2.05173045e-01 -5.68429351e-01 -8.57060373e-01
4.30835545e-01 1.14976668e+00 -7.76049852e-01 -8.93571973e-01
-4.40232933e-01 -6.64378464e-01 2.78599858e-01 4.48094904e-01
4.51579541e-01 -9.45250034e-01 -5.96642435e-01 1.73409835e-01
-1.69522703e-01 -4.26666319e-01 -8.80176127e-01 3.08593303e-01
-1.13385403e+00 -7.97488153e-01 -8.58496368e-01 -6.20589018e-01
9.80850399e-01 7.93766975e-01 7.11046576e-01 -2.07931519e-01
-2.72515327e-01 7.75168240e-02 -3.02958518e-01 -6.44851744e-01
-1.46850720e-01 2.27540359e-01 5.52320361e-01 6.11744374e-02
5.81888139e-01 -5.76873779e-01 -5.50606608e-01 3.21881652e-01
-9.34484303e-01 -2.04984233e-01 6.29623711e-01 1.11701334e+00
7.57120311e-01 2.83799469e-01 1.05303335e+00 -1.16020334e+00
8.45079601e-01 -5.56829870e-01 -2.89238244e-01 4.07866329e-01
-1.01647174e+00 1.31599372e-02 6.66141331e-01 -1.07257700e+00
-1.06331968e+00 6.65786043e-02 5.97374678e-01 -9.24549639e-01
9.66318920e-02 1.92093119e-01 -1.46553487e-01 1.68596804e-01
9.58246231e-01 5.38724005e-01 3.07545781e-01 -4.32884216e-01
5.06906271e-01 7.90383935e-01 7.75433421e-01 -4.64224964e-01
6.16917670e-01 4.54521149e-01 -3.68599117e-01 -7.87415802e-01
-8.34858239e-01 -3.29768896e-01 -7.08997071e-01 1.52220381e-02
-1.25755779e-02 -9.00297105e-01 -4.30871636e-01 4.72222030e-01
-6.26507461e-01 -2.92375147e-01 -8.23673308e-01 4.20806170e-01
-3.71648580e-01 3.78393263e-01 -2.77719021e-01 -9.58782077e-01
-3.04920465e-01 -7.58488774e-01 6.72307611e-01 3.22794318e-01
-2.04425901e-01 -5.99542618e-01 -1.09196231e-01 1.15969114e-01
3.86818647e-01 -1.25631258e-01 9.15956795e-01 -6.73147559e-01
-3.61741215e-01 -2.63054162e-01 -4.87837046e-02 2.91722387e-01
8.72615501e-02 -6.30473673e-01 -9.77328241e-01 -8.72211576e-01
2.26015896e-01 -2.55707920e-01 1.21605742e+00 5.80159537e-02
1.34912789e+00 -4.51442420e-01 -7.06769943e-01 4.23270673e-01
1.06122434e+00 2.42249787e-01 4.87163872e-01 2.27041557e-01
4.54426527e-01 2.22347349e-01 7.66075253e-01 6.17683411e-01
8.01230967e-02 3.76057982e-01 1.40744805e-01 4.23464209e-01
-2.43652701e-01 -4.93778259e-01 3.11043620e-01 8.96110058e-01
5.99392474e-01 -2.84535997e-02 -6.66113615e-01 7.40224719e-01
-2.00952601e+00 -7.44019270e-01 3.03314418e-01 2.69905138e+00
1.01868510e+00 3.23873848e-01 -9.20017511e-02 2.02746525e-01
8.31567526e-01 -1.07904859e-01 -1.04225588e+00 1.50831088e-01
-1.33334756e-01 7.22947642e-02 3.32411081e-01 2.05450147e-01
-9.59837914e-01 6.88463211e-01 5.43327522e+00 9.94798899e-01
-7.27106571e-01 2.69986570e-01 6.25550270e-01 -6.72512829e-01
-1.72485620e-01 -2.28782296e-01 -9.46946681e-01 6.32925808e-01
7.78916001e-01 -4.85802978e-01 6.15652204e-01 1.10676694e+00
-3.86480600e-01 -2.59226292e-01 -1.04650903e+00 1.16687715e+00
3.50100726e-01 -1.19380307e+00 3.22264791e-01 -2.36243650e-01
9.04507577e-01 -2.97648609e-01 5.02547145e-01 6.69684529e-01
1.05699562e-01 -7.30167866e-01 7.67960489e-01 8.39301944e-01
7.40294874e-01 -1.16812146e+00 5.91180682e-01 5.93005002e-01
-9.24534917e-01 -5.05380690e-01 -1.03293860e+00 2.11661264e-01
-2.22227708e-01 9.56048965e-01 -8.88024330e-01 3.72816861e-01
6.55982971e-01 4.73594964e-01 -1.01812470e+00 1.25455284e+00
1.29929766e-01 2.80522048e-01 -2.60096520e-01 -9.72022265e-02
-1.79946795e-01 1.52882785e-02 5.13755322e-01 8.13812673e-01
5.01897216e-01 -2.75094151e-01 1.63530678e-01 7.70040393e-01
-3.39808792e-01 -1.50339335e-01 -9.35195863e-01 2.77042210e-01
1.35194027e+00 6.96597397e-01 -6.10102713e-01 -4.81991917e-01
-1.43912673e-01 1.13019955e+00 6.94046676e-01 3.82294834e-01
-5.85327208e-01 -7.75720239e-01 6.23171568e-01 3.00451636e-01
2.82407641e-01 -9.37576145e-02 -2.43347317e-01 -1.13147378e+00
2.49609098e-01 -6.31933093e-01 4.52898830e-01 -5.16101480e-01
-1.42645383e+00 3.40526640e-01 2.62129810e-02 -1.08513558e+00
1.33185878e-01 -4.80731018e-03 -3.50182235e-01 4.93242949e-01
-1.01996124e+00 -5.20566165e-01 -2.54085869e-01 6.59155071e-01
6.47809029e-01 -2.28561252e-01 6.63600922e-01 2.37118199e-01
-6.45568311e-01 1.09970295e+00 5.38630903e-01 -2.66820252e-01
1.08688235e+00 -9.42207217e-01 3.61002594e-01 5.64005256e-01
1.60009861e-01 1.11317050e+00 4.51083124e-01 -9.52409148e-01
-1.20192719e+00 -1.63217449e+00 7.46276557e-01 -5.27685702e-01
2.68181503e-01 -7.15973318e-01 -1.50678933e+00 7.62080550e-01
-3.23707968e-01 1.98626205e-01 5.29136479e-01 8.87878835e-02
-5.47201753e-01 -3.06039423e-01 -1.26578379e+00 7.76164293e-01
1.28565431e+00 -6.36756122e-01 -9.06198263e-01 3.25112760e-01
9.66471374e-01 -9.48865339e-02 -4.41326559e-01 2.19266087e-01
3.17380786e-01 -8.78901064e-01 1.07270646e+00 -3.89581442e-01
-4.25324053e-01 -4.49171871e-01 3.91606204e-02 -1.48704565e+00
-3.76775682e-01 -5.92800975e-01 -5.69470882e-01 1.29422140e+00
2.17327654e-01 -8.13926220e-01 7.90542543e-01 5.39953470e-01
-5.54491691e-02 -6.34259403e-01 -1.17322123e+00 -1.21300316e+00
7.20803663e-02 -3.55227977e-01 8.24088573e-01 9.58156943e-01
-8.47256370e-03 3.08698803e-01 -4.15407002e-01 -1.81432098e-01
7.98805535e-01 4.67166267e-02 7.48476803e-01 -1.35185349e+00
9.39511806e-02 -3.08710873e-01 -1.37220204e-01 -1.00368023e+00
1.63256109e-01 -1.04211533e+00 7.85699561e-02 -8.60198140e-01
2.81662911e-01 -7.21332848e-01 -4.41720754e-01 3.76088798e-01
-5.31968415e-01 3.03093903e-02 7.50109404e-02 5.84704578e-01
-8.48652244e-01 8.56962085e-01 1.05878139e+00 -1.70037076e-01
-5.07232666e-01 2.41506230e-02 -6.73793972e-01 4.55283582e-01
8.22929740e-01 -6.25210643e-01 -7.36521363e-01 -1.67160332e-01
-8.14027712e-03 -3.68490130e-01 2.21520171e-01 -1.24698901e+00
5.19656837e-01 2.69518226e-01 7.77458191e-01 -6.92035913e-01
2.25712016e-01 -7.72038817e-01 2.74996698e-01 6.40819132e-01
-3.26887906e-01 -9.49578285e-02 7.08680153e-02 1.20158768e+00
1.27914190e-01 -4.22662348e-01 7.55823314e-01 -1.54852107e-01
-4.99024779e-01 1.82578623e-01 -2.39383340e-01 -3.51221859e-02
1.32153404e+00 -2.27291927e-01 -5.40294230e-01 7.34132379e-02
-7.72912145e-01 2.41347298e-01 6.87366009e-01 5.21266758e-01
9.13022816e-01 -1.54630113e+00 -2.58586884e-01 6.65056884e-01
2.08681330e-01 2.26594269e-01 3.79220873e-01 6.06155992e-01
7.90476650e-02 2.77672440e-01 -4.47460748e-02 -4.96168107e-01
-9.65575993e-01 1.41323471e+00 -5.69940433e-02 9.31407139e-02
-8.75580251e-01 8.48737359e-01 1.58640429e-01 -4.50245202e-01
6.03061199e-01 -1.20047167e-01 1.76265672e-01 2.88672417e-01
9.18901145e-01 5.20703018e-01 5.00321314e-02 -3.62732187e-02
-1.88160002e-01 4.67566550e-02 -7.31676638e-01 1.99204266e-01
1.06730819e+00 -3.91313463e-01 9.51744020e-02 8.55598450e-01
1.10855556e+00 -1.15343273e-01 -1.48050880e+00 -7.28927612e-01
7.15276450e-02 -8.38689804e-01 -2.65440226e-01 -4.82617676e-01
-7.95972109e-01 7.16968238e-01 7.42532074e-01 -1.13341384e-01
1.01731026e+00 -1.44970238e-01 9.10466731e-01 6.95863605e-01
6.73641860e-01 -1.29753196e+00 3.29643667e-01 2.83425599e-01
7.40308285e-01 -9.80139911e-01 5.03535010e-02 4.45747608e-03
-7.14923203e-01 8.09216261e-01 6.64477587e-01 4.87700216e-02
7.00702846e-01 1.24557339e-01 -2.72003800e-01 2.29458988e-01
-8.50778759e-01 2.67436326e-01 -1.42412754e-02 7.90442467e-01
-2.31751814e-01 1.82795420e-01 -2.86450442e-02 8.48824382e-01
-1.19454950e-01 6.63705543e-02 3.32616001e-01 1.13510370e+00
-6.89985812e-01 -8.48539829e-01 -4.99170154e-01 7.68099666e-01
8.22096765e-02 2.36917976e-02 -2.92531163e-01 6.11200452e-01
1.36330456e-01 6.35701418e-01 2.24160001e-01 -4.29323018e-01
3.94424528e-01 5.32671332e-01 2.59767771e-01 -4.75675404e-01
-2.18897432e-01 -1.24405339e-01 -5.63388765e-01 -2.91878462e-01
3.11131358e-01 -8.21452737e-01 -1.02086568e+00 -3.73586029e-01
-4.44603920e-01 2.56612241e-01 2.37517968e-01 4.94636655e-01
7.07268238e-01 5.13685286e-01 5.48011243e-01 -6.80881441e-01
-8.54793847e-01 -8.96069109e-01 -6.27002716e-01 3.80334020e-01
2.90994912e-01 -9.86758888e-01 -4.37932700e-01 -2.24317998e-01] | [9.809309959411621, 3.4126760959625244] |
5a7919f5-bd18-4f42-9e37-0d46cae7c3ce | janus-parallel-tempered-genetic-algorithm | 2106.04011 | null | https://arxiv.org/abs/2106.04011v2 | https://arxiv.org/pdf/2106.04011v2.pdf | JANUS: Parallel Tempered Genetic Algorithm Guided by Deep Neural Networks for Inverse Molecular Design | Inverse molecular design, i.e., designing molecules with specific target properties, can be posed as an optimization problem. High-dimensional optimization tasks in the natural sciences are commonly tackled via population-based metaheuristic optimization algorithms such as evolutionary algorithms. However, expensive property evaluation, which is often required, can limit the widespread use of such approaches as the associated cost can become prohibitive. Herein, we present JANUS, a genetic algorithm that is inspired by parallel tempering. It propagates two populations, one for exploration and another for exploitation, improving optimization by reducing expensive property evaluations. Additionally, JANUS is augmented by a deep neural network that approximates molecular properties via active learning for enhanced sampling of the chemical space. Our method uses the SELFIES molecular representation and the STONED algorithm for the efficient generation of structures, and outperforms other generative models in common inverse molecular design tasks achieving state-of-the-art performance. | ['Alan Aspuru-Guzik', 'Robert Pollice', 'AkshatKumar Nigam'] | 2021-06-07 | null | null | null | null | ['metaheuristic-optimization'] | ['methodology'] | [ 5.24573147e-01 -1.64040610e-01 -2.61862248e-01 1.50164127e-01
-6.74556136e-01 -5.23446858e-01 5.49927473e-01 4.48872566e-01
-5.19940913e-01 1.40997350e+00 -2.66235381e-01 -2.79972970e-01
-3.68503541e-01 -1.14085126e+00 -8.95615637e-01 -1.38951075e+00
1.45113602e-01 8.35446596e-01 -2.89054751e-01 -2.38097668e-01
6.23493314e-01 5.88677466e-01 -1.75205874e+00 -1.10620573e-01
1.57701230e+00 8.30778062e-01 1.26757741e-01 3.52437168e-01
-6.00090660e-02 3.52150470e-01 -4.87787426e-01 -3.89089108e-01
-1.11647204e-01 -7.45943129e-01 -5.45771718e-01 -3.74131560e-01
-2.31759533e-01 4.40542579e-01 2.48717561e-01 8.02014232e-01
8.44279706e-01 5.99652290e-01 9.67166185e-01 -8.35861683e-01
-8.96445513e-01 4.58475828e-01 -3.09351563e-01 -2.22395122e-01
3.02114427e-01 4.24406976e-01 7.62670398e-01 -7.11887658e-01
5.51904798e-01 1.06845951e+00 5.11571765e-01 6.36054635e-01
-1.72291315e+00 -6.47503614e-01 -1.90637708e-01 3.12782377e-01
-1.40232861e+00 -2.96642184e-01 9.16154623e-01 -2.02058315e-01
1.18830955e+00 3.56667012e-01 1.07634568e+00 1.28664303e+00
3.96288216e-01 7.35959947e-01 8.40542793e-01 -3.21693242e-01
1.00937533e+00 -1.66218415e-01 -4.52805459e-01 5.95540881e-01
4.34246421e-01 6.25768542e-01 -5.97507656e-01 -6.56095445e-01
2.69791722e-01 -1.76025778e-01 -3.66358280e-01 -7.94260502e-01
-8.60072494e-01 1.25939679e+00 5.18364608e-01 -8.86462927e-02
-9.72736955e-01 6.48577288e-02 7.32378056e-03 -1.84827715e-01
4.52529937e-01 1.28469479e+00 -2.49695003e-01 -1.89596400e-01
-1.02789927e+00 5.58025301e-01 8.24942887e-01 4.11710769e-01
7.23659039e-01 1.21881373e-01 -5.44375256e-02 5.72053552e-01
5.56781709e-01 3.67673159e-01 2.11993068e-01 -6.82514369e-01
-7.87064731e-02 8.65223050e-01 3.11724469e-02 -5.82196236e-01
-3.67285579e-01 -7.82244444e-01 -1.00985932e+00 3.63277972e-01
3.68498228e-02 -2.15284958e-01 -8.28682184e-01 1.66507912e+00
7.38301694e-01 6.05439097e-02 1.68042302e-01 5.34840941e-01
6.23746991e-01 8.77391636e-01 4.02848840e-01 -6.88784540e-01
8.91637087e-01 -9.66187000e-01 -6.26554728e-01 7.34008551e-02
4.27630752e-01 -5.76709926e-01 4.63937193e-01 4.63711798e-01
-1.21129549e+00 -2.13945702e-01 -1.30691016e+00 3.09618533e-01
-7.08462119e-01 -2.54954427e-01 1.00114930e+00 1.02437341e+00
-6.35595560e-01 1.10588348e+00 -7.74242103e-01 1.42961845e-01
7.18207240e-01 7.88940787e-01 -6.59036115e-02 2.47326329e-01
-1.15039706e+00 8.06300163e-01 6.90482080e-01 2.05720186e-01
-7.95945048e-01 -9.71761048e-01 -7.79704332e-01 1.50832105e-02
4.96899396e-01 -1.29082477e+00 6.10147238e-01 -5.61340332e-01
-2.18005133e+00 1.67661980e-01 -2.25074515e-02 -2.78674901e-01
2.41243035e-01 7.38888532e-02 -1.31677181e-01 -8.29567760e-02
-3.13700527e-01 4.75193769e-01 9.32252288e-01 -1.12674284e+00
9.70069021e-02 -5.05304635e-01 -3.47746819e-01 4.11013365e-01
-1.45828694e-01 -5.01584947e-01 2.91109197e-02 -5.81449628e-01
-1.54717058e-01 -1.06609070e+00 -6.76342845e-01 -4.10205752e-01
-4.05091554e-01 -1.93173885e-01 5.13026416e-01 -1.81091815e-01
1.28850484e+00 -1.56612599e+00 9.30986166e-01 5.96681297e-01
1.37562469e-01 3.70161891e-01 -1.35270134e-01 6.57987714e-01
-1.59851789e-01 9.61671472e-02 -5.83737135e-01 -8.09306055e-02
-3.63250598e-02 -9.97452065e-02 -2.95022000e-02 3.05225730e-01
2.70364642e-01 1.19197679e+00 -1.10403156e+00 -1.51647091e-01
8.89185932e-04 8.71026814e-01 -8.55001032e-01 2.32772604e-01
-6.78545415e-01 7.71219313e-01 -7.70603061e-01 6.35874987e-01
5.97651243e-01 -3.13430816e-01 2.17846945e-01 -1.17762454e-01
-3.18668932e-01 -8.91478583e-02 -9.67573047e-01 1.70612228e+00
-3.88889015e-01 -4.27821279e-02 -3.36798757e-01 -1.05979347e+00
1.01072776e+00 2.68484384e-01 5.56961119e-01 -7.58735657e-01
2.50224590e-01 4.84330177e-01 8.79024491e-02 -2.51236767e-01
4.55998629e-01 -2.05878720e-01 3.71773183e-01 5.62113881e-01
-1.94755971e-01 -5.37973285e-01 4.32284206e-01 -3.62943739e-01
6.92477465e-01 8.52556646e-01 4.43181366e-01 -2.55554885e-01
6.97273493e-01 1.30707607e-01 4.23815370e-01 5.39593637e-01
3.78114343e-01 2.32313827e-01 2.09904492e-01 -3.79249066e-01
-1.13641572e+00 -7.05175519e-01 -2.26178452e-01 8.23570311e-01
2.65899599e-02 -3.63743275e-01 -8.63190830e-01 -3.92514259e-01
-4.15588766e-02 8.03709149e-01 -7.27925122e-01 -6.90605879e-01
-4.83803481e-01 -1.57979953e+00 2.15764239e-01 9.33237076e-02
2.16794536e-01 -1.36076033e+00 -6.69033170e-01 6.77476466e-01
1.84370577e-01 -1.80072784e-01 -7.62931481e-02 4.37751114e-01
-9.27670956e-01 -9.61409032e-01 -9.45123017e-01 -3.07052702e-01
5.68688214e-01 -3.50427538e-01 1.28867877e+00 -9.76663083e-03
-6.43603325e-01 -6.78500384e-02 -2.61296153e-01 -4.05780166e-01
-5.23415208e-01 4.55101579e-01 -2.77602896e-02 1.73928658e-03
1.06500156e-01 -8.65915060e-01 -9.82014954e-01 1.62561208e-01
-8.69834185e-01 -4.48406078e-02 8.02356124e-01 1.17864156e+00
9.95891273e-01 1.15099654e-01 6.90614104e-01 -7.26838827e-01
7.81965435e-01 -3.38956088e-01 -6.58648074e-01 3.88518959e-01
-7.80324459e-01 6.67140841e-01 5.62981009e-01 -5.84153116e-01
-9.64982986e-01 5.93159497e-02 -2.19374895e-01 -1.85032208e-02
1.23897798e-01 4.57428843e-01 -3.82936448e-01 -4.87047583e-01
8.14458430e-01 5.15413046e-01 4.32365537e-02 -2.21216545e-01
3.74502033e-01 3.14114183e-01 -7.09439069e-02 -8.73061121e-01
5.77818513e-01 1.11508124e-01 5.34188867e-01 -9.31857228e-01
-5.98772049e-01 9.50751528e-02 -2.31401876e-01 -1.37824705e-02
7.77050853e-01 -3.58560592e-01 -1.35864770e+00 3.16926271e-01
-8.33824217e-01 -2.15422317e-01 -4.26186502e-01 5.17350316e-01
-9.02627587e-01 2.49663308e-01 -6.04665419e-03 -1.03431690e+00
-7.26694226e-01 -1.50670135e+00 9.69931960e-01 4.81789619e-01
-4.32747632e-01 -9.33184564e-01 5.69456518e-01 2.98723578e-01
5.55473387e-01 6.36361420e-01 1.25633383e+00 -3.76042038e-01
-7.40036190e-01 -8.00702721e-02 4.80024397e-01 -3.14970434e-01
4.03221101e-02 1.19108334e-01 -6.61475837e-01 -3.73687625e-01
-2.36688793e-01 -4.37150002e-01 8.33155215e-01 6.68129742e-01
1.16290677e+00 -2.11747319e-01 -5.38122118e-01 9.35496926e-01
1.26588166e+00 7.30129182e-01 7.12775290e-01 5.81300557e-01
3.92695367e-01 4.72401619e-01 4.08417374e-01 6.85522258e-01
-8.41660276e-02 6.07744098e-01 5.53112924e-01 -1.44127607e-01
3.27903032e-01 -1.83893219e-01 -1.66806236e-01 1.72665358e-01
-6.45202398e-01 -5.36826611e-01 -6.67159021e-01 -6.51510507e-02
-1.87414145e+00 -1.09437168e+00 1.73000619e-01 2.27465796e+00
1.00477517e+00 -2.26976737e-01 1.35692582e-01 1.33056894e-01
5.49189687e-01 -9.39104054e-03 -1.17220140e+00 -3.11179370e-01
-3.24945360e-01 6.53246403e-01 1.14809886e-01 2.68312424e-01
-8.04208875e-01 7.33320475e-01 6.00094700e+00 9.49709415e-01
-8.86974037e-01 -3.67161483e-01 8.87048066e-01 -2.67073691e-01
-4.45467383e-01 -1.73660200e-02 -6.26848757e-01 4.35088485e-01
8.57715428e-01 -1.56369418e-01 7.48072088e-01 5.56063831e-01
1.60494894e-01 -1.32689595e-01 -9.94403958e-01 9.78844583e-01
-1.18385747e-01 -1.81633306e+00 2.38216683e-01 2.95445085e-01
1.08570850e+00 -5.00472307e-01 4.39468801e-01 1.12804540e-01
-2.08185110e-02 -1.30715239e+00 5.35868406e-01 6.33975267e-01
4.50538069e-01 -1.44915080e+00 4.63796347e-01 2.66768813e-01
-8.88741374e-01 1.28202988e-02 -1.97878093e-01 4.75675851e-01
2.90058702e-01 3.72271955e-01 -2.23919004e-01 6.42237842e-01
2.78602809e-01 4.44685012e-01 -2.36920059e-01 1.32503009e+00
-9.11411047e-02 2.34722391e-01 -3.45405698e-01 -6.26824021e-01
3.22160214e-01 -9.59595442e-01 7.30270743e-01 6.36302769e-01
5.29642105e-01 1.99750811e-02 -1.00043401e-01 1.29146039e+00
-4.43776660e-02 2.83112735e-01 -3.89112830e-01 -5.02173543e-01
3.95823240e-01 9.99445200e-01 -7.25309312e-01 -9.31960568e-02
3.95235330e-01 8.52238178e-01 2.49592543e-01 5.43022156e-01
-8.89019132e-01 -3.93830895e-01 4.74328071e-01 -6.67892247e-02
4.22609299e-01 -6.37892112e-02 -8.89448225e-02 -6.38359249e-01
-5.36242604e-01 -9.41574693e-01 1.17033385e-01 -3.48487496e-01
-9.92858827e-01 6.75832331e-01 -3.19010735e-01 -8.25680137e-01
-2.81306237e-01 -5.43683529e-01 -5.90538263e-01 8.50550354e-01
-1.43967152e+00 -7.69119918e-01 -1.27626937e-02 3.54017079e-01
2.02349901e-01 -3.20274800e-01 9.76042211e-01 1.64928734e-02
-9.67285335e-01 5.22356451e-01 5.97944200e-01 -8.84824336e-01
2.87431747e-01 -1.15614212e+00 1.38541862e-01 1.84011236e-01
-1.98934838e-01 8.63837600e-01 8.67688894e-01 -7.30283618e-01
-1.80780423e+00 -8.45741212e-01 4.07126844e-01 -4.08903435e-02
4.42510366e-01 -2.06391856e-01 -8.15767348e-01 -3.52228016e-01
3.78461666e-02 -4.21715111e-01 1.12897146e+00 5.32307513e-02
1.61220729e-01 3.74273717e-01 -1.20549631e+00 7.72376120e-01
1.16175795e+00 -4.92449887e-02 -2.68801190e-02 4.09153312e-01
4.17202771e-01 -2.90887445e-01 -1.06334627e+00 3.94103974e-01
6.17968380e-01 -8.22871685e-01 1.35383451e+00 -7.06529915e-01
3.61701667e-01 -3.31707299e-01 2.07505524e-01 -1.50999522e+00
-4.81562138e-01 -1.02717185e+00 -6.43825173e-01 8.17685783e-01
7.11031675e-01 -7.65371025e-01 1.00588953e+00 4.26644117e-01
1.10611133e-01 -1.17338419e+00 -9.56950665e-01 -8.09970737e-01
1.20190211e-01 1.21639028e-01 1.08535838e+00 6.25790775e-01
-2.33126134e-01 6.57366335e-01 -1.61799476e-01 -2.08252937e-01
8.82430196e-01 3.56978476e-01 3.22632730e-01 -1.28644407e+00
-4.75002199e-01 -7.91618407e-01 -4.12412547e-02 -5.95734656e-01
2.77202010e-01 -7.55204141e-01 -1.62977055e-01 -1.27469754e+00
1.22849025e-01 -4.91576374e-01 -4.44821864e-02 7.39275888e-02
-3.93545181e-01 2.58080006e-01 -2.34177575e-01 -5.03875874e-02
-2.88830698e-01 1.45057142e+00 1.44679129e+00 -3.73870760e-01
-7.26409733e-01 4.93125357e-02 -6.40547693e-01 3.34489763e-01
7.34108388e-01 -4.95833278e-01 -4.20034349e-01 1.88052356e-01
8.74483347e-01 1.29108094e-02 -6.38433918e-02 -9.69566107e-01
1.33683741e-01 -2.76640683e-01 5.02237201e-01 -5.15231490e-01
5.55787385e-01 -5.07466197e-01 6.60143375e-01 7.32198715e-01
-1.67818710e-01 -1.03999272e-01 1.41020864e-01 7.89704919e-01
-6.77963272e-02 -3.72438937e-01 9.17085052e-01 -6.70874715e-02
-4.39481847e-02 6.48106217e-01 -3.45738411e-01 -2.23235220e-01
1.09684622e+00 -6.11618102e-01 -5.15458547e-02 1.31069958e-01
-5.24065316e-01 1.00144364e-01 5.55308044e-01 -1.35592418e-02
6.28070533e-01 -1.27013671e+00 -6.29047453e-01 1.42351493e-01
2.30783448e-02 8.59020576e-02 3.59681845e-01 8.40241492e-01
-6.04803920e-01 2.10382313e-01 4.99901921e-02 -3.91789496e-01
-9.64605987e-01 7.44670391e-01 4.95409518e-01 -3.84191543e-01
-2.46193916e-01 8.68203223e-01 -3.82794067e-02 -4.14149165e-01
-1.62928715e-01 4.06275690e-01 -4.02221769e-01 3.95404935e-01
3.76605302e-01 5.81345797e-01 1.56095549e-01 -3.33324522e-01
-3.85986775e-01 7.58540154e-01 -4.03690636e-02 1.81904450e-01
1.68522513e+00 4.25029665e-01 -2.78400987e-01 -1.89969867e-01
9.90112782e-01 -3.20884019e-01 -1.12864602e+00 1.58034772e-01
-1.40321448e-01 -6.61739558e-02 3.54535013e-01 -8.97954285e-01
-7.61467159e-01 5.22301853e-01 5.37888169e-01 -4.91209850e-02
9.92635071e-01 -3.94291371e-01 6.85985804e-01 7.24720597e-01
2.46745363e-01 -1.13666308e+00 9.15378034e-02 4.58758950e-01
8.81602407e-01 -1.10624564e+00 3.26213300e-01 -9.91746113e-02
-3.03923279e-01 1.23529232e+00 1.90963417e-01 1.19111069e-01
2.77322799e-01 1.41030131e-02 -4.32854205e-01 -2.61686802e-01
-6.94293439e-01 -1.27288491e-01 3.40202481e-01 5.49857914e-01
4.08595830e-01 -6.52451962e-02 -4.08987790e-01 1.00928932e-01
-2.28335842e-01 -2.39327401e-01 -1.05061933e-01 1.11882484e+00
-3.04173589e-01 -1.47240424e+00 -4.82974976e-01 1.76801085e-01
-1.26270577e-01 -2.97945380e-01 -4.16113585e-01 4.18676645e-01
2.27974027e-01 7.31668711e-01 -1.84991241e-01 1.01066463e-01
1.38609007e-01 -1.00042962e-01 8.89740884e-01 -2.80282825e-01
-8.29701900e-01 -7.79314083e-04 -6.59691393e-02 -4.04760301e-01
-3.79264832e-01 -4.36689079e-01 -9.93325174e-01 -3.12015235e-01
-5.78755558e-01 7.01159060e-01 7.42645085e-01 7.79207289e-01
5.23004651e-01 6.98265970e-01 6.02874160e-01 -1.05465317e+00
-4.36085522e-01 -4.33295786e-01 -4.13980663e-01 -2.08533164e-02
1.59669042e-01 -1.00780368e+00 -1.32827967e-01 -3.06228667e-01] | [5.056747913360596, 5.414766311645508] |
b5a00351-fb72-47c2-8faf-a9ca11f38603 | eac-net-a-region-based-deep-enhancing-and | 1702.02925 | null | http://arxiv.org/abs/1702.02925v1 | http://arxiv.org/pdf/1702.02925v1.pdf | EAC-Net: A Region-based Deep Enhancing and Cropping Approach for Facial Action Unit Detection | In this paper, we propose a deep learning based approach for facial action
unit detection by enhancing and cropping the regions of interest. The approach
is implemented by adding two novel nets (layers): the enhancing layers and the
cropping layers, to a pretrained CNN model. For the enhancing layers, we
designed an attention map based on facial landmark features and applied it to a
pretrained neural network to conduct enhanced learning (The E-Net). For the
cropping layers, we crop facial regions around the detected landmarks and
design convolutional layers to learn deeper features for each facial region
(C-Net). We then fuse the E-Net and the C-Net to obtain our Enhancing and
Cropping (EAC) Net, which can learn both feature enhancing and region cropping
functions. Our approach shows significant improvement in performance compared
to the state-of-the-art methods applied to BP4D and DISFA AU datasets. | ['Lijun Yin', 'Zhigang Zhu', 'Wei Li', 'Farnaz Abtahi'] | 2017-02-09 | null | null | null | null | ['action-unit-detection', 'facial-action-unit-detection'] | ['computer-vision', 'computer-vision'] | [ 1.43618420e-01 4.41396534e-01 1.43478751e-01 -5.47427058e-01
-4.28382158e-01 -9.36478078e-02 4.44783986e-01 -4.98499185e-01
-7.28341699e-01 3.45644832e-01 2.77253151e-01 2.14381605e-01
4.18357462e-01 -8.72982800e-01 -8.41041386e-01 -5.95970392e-01
-1.40219852e-01 -2.35020846e-01 2.94755220e-01 -1.17016278e-01
1.30482659e-01 1.08524024e+00 -1.59088862e+00 5.95985353e-01
4.60047901e-01 1.35015368e+00 -7.13254660e-02 4.29416686e-01
5.10663874e-02 9.02824938e-01 -3.73766154e-01 -2.29827195e-01
4.26012933e-01 -1.66803345e-01 -7.72478998e-01 8.26373622e-02
3.94664109e-01 -8.88793766e-01 -4.00423080e-01 1.01397753e+00
5.52279770e-01 1.04796492e-01 4.34834927e-01 -1.30626166e+00
-7.97335088e-01 3.39999884e-01 -1.00713313e+00 8.85920003e-02
1.94171742e-01 -1.18329339e-01 5.60157120e-01 -1.42594600e+00
6.43434048e-01 1.48744941e+00 7.30304122e-01 8.53611827e-01
-6.24001622e-01 -8.83916855e-01 9.13614184e-02 1.14920318e-01
-1.48371017e+00 -6.10564947e-01 8.06693554e-01 -1.36459947e-01
9.84637380e-01 -2.99759209e-01 6.73461974e-01 8.00756097e-01
2.24771798e-02 9.60023046e-01 8.33352685e-01 -4.38088059e-01
1.70614477e-02 -3.59225608e-02 -2.17257529e-01 1.22188199e+00
-2.19175279e-01 3.37745279e-01 -2.05786780e-01 1.55842215e-01
1.13816011e+00 5.37046790e-02 -8.98041204e-02 7.55833685e-02
-6.92604780e-01 7.17713535e-01 1.06914043e+00 2.50567615e-01
-8.54559660e-01 4.03196931e-01 2.68547326e-01 -1.04024246e-01
4.75024253e-01 2.57850196e-02 -3.61587912e-01 3.46251786e-01
-1.00982392e+00 1.19478315e-01 2.34967127e-01 7.47623324e-01
1.17424417e+00 -9.37218219e-02 -5.54940820e-01 8.00475597e-01
4.82926786e-01 3.73961478e-02 2.97637612e-01 -8.83921504e-01
1.92332417e-01 1.04831362e+00 -1.64991736e-01 -7.76165426e-01
-6.96759164e-01 -2.26074159e-01 -5.77204466e-01 7.87615299e-01
2.54198015e-01 -6.63278937e-01 -1.35371339e+00 1.67912817e+00
3.65545124e-01 4.31770474e-01 3.11264824e-02 9.02663529e-01
1.08026850e+00 5.79889536e-01 2.31265754e-01 3.33279967e-01
1.27405441e+00 -1.39881372e+00 -4.74288374e-01 -1.23351425e-01
6.82340026e-01 -5.39025664e-01 6.73519194e-01 -1.33718997e-01
-1.23167753e+00 -8.36509764e-01 -8.71033907e-01 -4.24831599e-01
-6.77181184e-01 6.31466746e-01 4.65790987e-01 2.58408487e-01
-1.72333276e+00 5.06349206e-01 -7.24287987e-01 -4.29651201e-01
1.09530938e+00 7.84602463e-01 -8.46738100e-01 -2.04793680e-02
-1.07186186e+00 9.50487971e-01 3.43111694e-01 3.61915350e-01
-1.26730955e+00 -5.48259199e-01 -1.13737154e+00 3.65093678e-01
1.18303627e-01 -3.39013606e-01 1.09052444e+00 -1.34528780e+00
-1.51975429e+00 9.97245729e-01 -8.16291943e-02 -1.52687579e-01
9.02476087e-02 -1.80183053e-01 -8.75792652e-02 4.27984834e-01
-9.28765535e-02 1.47405851e+00 9.00345147e-01 -1.09746706e+00
-7.99924970e-01 -3.64617348e-01 2.57048756e-01 8.90002921e-02
-2.75494784e-01 3.97753984e-01 -7.05261528e-01 -6.36310518e-01
-9.87338647e-02 -6.91677034e-01 -1.64216995e-01 5.09705365e-01
-2.90013283e-01 -3.93913120e-01 1.27077401e+00 -8.54917824e-01
1.04096007e+00 -2.22814822e+00 -6.09603189e-02 3.45189601e-01
3.39908957e-01 5.89911759e-01 -6.16033375e-01 -1.17670499e-01
-4.19594020e-01 -8.44049156e-02 7.04524219e-02 -5.33133090e-01
-2.64295340e-01 -2.62684729e-02 2.70014077e-01 4.65126395e-01
8.09588492e-01 1.21941090e+00 -8.39437842e-01 -4.36664939e-01
3.79203558e-01 8.35221946e-01 -8.35995138e-01 1.31822914e-01
-6.35327771e-02 1.37365656e-02 -3.49993289e-01 1.02055788e+00
9.44214165e-01 1.63286567e-01 -3.00367743e-01 -3.58225048e-01
-2.59731859e-01 -3.65244061e-01 -8.00235331e-01 1.51794863e+00
-3.33683342e-01 6.74416661e-01 1.95583612e-01 -7.28927314e-01
1.10493600e+00 2.88857341e-01 4.68340904e-01 -6.86920941e-01
6.69316590e-01 -5.93636259e-02 -1.58763379e-01 -7.19040215e-01
3.46144259e-01 -3.59645374e-02 3.73799205e-01 3.89660567e-01
6.57894909e-01 5.58585763e-01 -1.93377942e-01 -1.10104762e-01
9.28416312e-01 5.31553149e-01 1.78473502e-01 -1.89387307e-01
8.31072986e-01 -5.24889410e-01 3.58889729e-01 7.89411068e-02
-5.25562584e-01 5.41378796e-01 7.68455267e-01 -6.50054991e-01
-8.46916497e-01 -5.13698280e-01 1.11401498e-01 1.37437308e+00
-2.00171277e-01 -2.12293565e-01 -1.32204139e+00 -1.07291579e+00
1.59046307e-01 6.38894662e-02 -1.43952382e+00 -4.05566096e-01
-4.27375138e-01 -2.97252208e-01 6.96574509e-01 1.07026613e+00
9.84236538e-01 -1.52081859e+00 -5.09340703e-01 5.91101199e-02
2.42863730e-01 -9.64224398e-01 -6.10546827e-01 2.18163565e-01
-3.92206848e-01 -1.00583923e+00 -1.09957790e+00 -1.18578207e+00
1.02708018e+00 -5.83882742e-02 6.00485384e-01 2.41033599e-01
-1.91410527e-01 -6.98740734e-03 -3.37618738e-01 -7.09479332e-01
5.21392636e-02 -1.84664726e-02 -2.49057233e-01 4.93245900e-01
6.46867037e-01 -2.41334274e-01 -7.33959138e-01 2.47043848e-01
-6.77655995e-01 -1.75559267e-01 8.35255980e-01 6.78355992e-01
4.54836488e-01 -2.52665550e-01 5.05589366e-01 -5.38778126e-01
3.82499784e-01 -2.39876717e-01 -6.13472402e-01 5.21761142e-02
-1.01301949e-02 2.01573651e-02 1.84406742e-01 -1.82798207e-01
-9.98432994e-01 5.68899155e-01 -5.90901256e-01 -6.86430216e-01
-2.22328424e-01 2.46818542e-01 -3.33460242e-01 -6.41096234e-01
4.52014089e-01 -1.04690023e-01 1.14364125e-01 -3.59872192e-01
4.32779402e-01 7.19256580e-01 6.04982018e-01 -1.56036288e-01
5.24569929e-01 5.27132034e-01 -1.17886856e-01 -5.68025112e-01
-8.12904537e-01 -2.17104018e-01 -1.17260802e+00 -5.80613136e-01
1.27396452e+00 -8.14820766e-01 -6.74338102e-01 6.98728859e-01
-1.25607073e+00 -4.50783908e-01 -2.09617451e-01 2.55294889e-01
-4.72390026e-01 -1.24103606e-01 -5.61981738e-01 -6.39268696e-01
-4.19655770e-01 -1.12484503e+00 1.49595559e+00 6.95558250e-01
9.57434624e-02 -1.00634348e+00 -4.86414470e-02 -2.16544002e-01
4.65254128e-01 6.57710135e-01 6.25390649e-01 -2.77628124e-01
-2.20237896e-01 -3.17890763e-01 -7.79105902e-01 5.39826989e-01
6.12170361e-02 9.43435058e-02 -1.51323295e+00 -5.49698696e-02
-4.51209307e-01 -4.73417580e-01 1.18523324e+00 6.04591846e-01
1.44583905e+00 -1.77621245e-01 -4.30833399e-01 8.81471813e-01
1.23032212e+00 2.34789878e-01 1.17057657e+00 2.64119655e-01
4.73212391e-01 5.82671463e-01 4.12952155e-01 4.37664866e-01
7.45915473e-02 4.93365556e-01 6.88092172e-01 -8.71511102e-01
-2.28650883e-01 -2.82869101e-01 3.30469042e-01 -2.86982536e-01
-5.42583287e-01 2.68659651e-01 -5.52645326e-01 6.10193908e-01
-1.59072769e+00 -9.23586249e-01 5.34048378e-01 1.64161360e+00
6.61777854e-01 -2.46616811e-01 2.43094221e-01 2.07620743e-03
7.44513392e-01 4.20612469e-02 -5.11937976e-01 -5.93903005e-01
6.97466284e-02 7.20429122e-01 2.87058324e-01 3.95976663e-01
-1.33253694e+00 1.41430438e+00 6.33612013e+00 5.35060346e-01
-1.38347650e+00 1.07460000e-01 7.86544502e-01 9.97982100e-02
4.03727919e-01 -5.10731041e-01 -9.30448055e-01 7.79138282e-02
5.88367283e-01 2.76340127e-01 2.33609572e-01 1.10671031e+00
1.67757154e-01 1.07917778e-01 -9.84986663e-01 9.17397082e-01
2.77085662e-01 -1.22614443e+00 -1.64951179e-02 1.50705636e-01
7.47169435e-01 2.35730466e-02 -3.25658210e-02 4.78337824e-01
1.77747905e-01 -1.29659235e+00 4.37963158e-01 6.11429572e-01
9.19045210e-01 -1.04170537e+00 1.01231301e+00 -2.32818887e-01
-1.33266830e+00 -1.73393592e-01 -7.59981334e-01 1.08795203e-01
-2.28344530e-01 1.80610884e-02 -4.76674855e-01 1.29537106e-01
7.48332560e-01 8.05521667e-01 -5.33363879e-01 9.54487681e-01
-3.84395987e-01 2.26854131e-01 -1.42101079e-01 2.64184296e-01
7.83278704e-01 2.99303308e-02 -5.81355765e-02 1.21252942e+00
3.54481071e-01 1.58494264e-01 -2.05405101e-01 8.99536073e-01
-5.20668030e-01 -2.60788575e-02 -4.66680706e-01 2.07395822e-01
1.07972287e-01 1.87350559e+00 -6.19080186e-01 -4.00716811e-01
-6.80671632e-01 1.23756301e+00 7.14843750e-01 4.02684420e-01
-8.91349316e-01 -6.95295453e-01 7.25904763e-01 7.95327220e-03
4.60895747e-01 3.65281999e-01 9.59892794e-02 -6.28189325e-01
-2.32144803e-01 -5.08430183e-01 3.62215668e-01 -8.89001131e-01
-9.13513124e-01 1.02338505e+00 -2.39056483e-01 -8.24388027e-01
2.45441839e-01 -1.03673363e+00 -8.87458146e-01 1.01304317e+00
-1.78368664e+00 -1.39060736e+00 -6.67278945e-01 1.04873478e+00
2.03842819e-01 -2.92880416e-01 5.94958842e-01 3.41507316e-01
-6.64117813e-01 8.29009771e-01 -4.70602661e-01 6.86282575e-01
5.63735723e-01 -1.01928186e+00 3.58144790e-01 6.38950348e-01
-1.39185891e-01 4.45912182e-01 -2.99316615e-01 -4.18359697e-01
-7.62669683e-01 -1.64605105e+00 9.55145597e-01 -1.17452919e-01
2.93376952e-01 -4.12844718e-01 -6.86607301e-01 9.26398754e-01
3.16753596e-01 6.07106149e-01 4.15033132e-01 -3.04603875e-01
-1.33240014e-01 -5.59136793e-02 -1.48789477e+00 4.54712808e-01
1.00425720e+00 -4.54715282e-01 -4.71260905e-01 1.31675139e-01
4.49002326e-01 -3.32996100e-01 -6.52222753e-01 3.45298380e-01
7.92521417e-01 -8.14734876e-01 8.41718674e-01 -6.89671338e-01
4.72112238e-01 -1.88141853e-01 3.10465004e-02 -1.27191508e+00
-6.07019484e-01 -2.30387673e-01 -9.21087805e-03 1.09790325e+00
4.88365322e-01 -3.51038873e-01 9.74259555e-01 3.92759711e-01
-2.89366603e-01 -1.17982447e+00 -6.43715382e-01 -2.86708742e-01
4.59493101e-02 -6.19016364e-02 8.74518156e-01 7.14242578e-01
1.94130198e-03 1.10274022e-02 -2.25972429e-01 1.82683885e-01
2.61456132e-01 -4.54723507e-01 4.32408541e-01 -9.29166436e-01
3.54933292e-01 -5.19333780e-01 -6.23756111e-01 -8.13741565e-01
2.71442443e-01 -1.02078450e+00 -7.54045835e-03 -1.50155210e+00
2.12781250e-01 -3.34194191e-02 -5.19762933e-01 1.22447395e+00
-2.64156163e-01 9.50275064e-01 1.54263124e-01 -4.65886623e-01
-4.02555287e-01 5.77989519e-01 1.37287211e+00 5.00806756e-02
-3.10931385e-01 -1.80852786e-01 -8.22773814e-01 8.70556414e-01
6.58498108e-01 -9.20125842e-02 -3.71355889e-03 -4.46104378e-01
-3.84493470e-01 -4.05637801e-01 5.19309998e-01 -1.08060515e+00
2.73478419e-01 4.21263933e-01 1.01921749e+00 -7.59558916e-01
2.33238623e-01 -8.82308304e-01 -6.72858357e-01 4.56363082e-01
-3.31128776e-01 -6.10633083e-02 5.20309329e-01 -2.70839687e-02
-2.75335491e-01 1.80267803e-02 1.07177615e+00 -1.20300718e-01
-9.68660533e-01 7.03899205e-01 -2.66176671e-01 -5.91297269e-01
1.45513439e+00 -2.68307805e-01 -1.55320153e-01 -2.07445875e-01
-1.03102326e+00 3.16889793e-01 7.47405216e-02 4.30357188e-01
1.04025722e+00 -1.52212524e+00 -6.88742638e-01 7.53109574e-01
-4.47767712e-02 -1.98058486e-01 1.86308041e-01 8.05533230e-01
-6.42378330e-01 3.08695555e-01 -8.07792425e-01 -4.16918099e-01
-1.36582708e+00 6.82555199e-01 7.10814238e-01 -6.72906684e-03
-3.10112983e-01 1.14938307e+00 3.75315845e-01 -3.47689062e-01
4.49660778e-01 -4.67561513e-01 -6.84184015e-01 1.18058346e-01
8.87960792e-01 2.90978551e-01 -4.80148662e-03 -8.33628535e-01
-5.86361349e-01 7.71336257e-01 -6.25861958e-02 1.78973973e-01
1.42293143e+00 1.06687143e-01 -2.43767142e-01 -5.76648653e-01
1.57752466e+00 -3.23517650e-01 -1.53651798e+00 -2.95730472e-01
-1.12997457e-01 -1.59757242e-01 3.90890568e-01 -7.26046443e-01
-1.76625526e+00 9.32718635e-01 7.80935109e-01 -5.56781352e-01
1.54589880e+00 9.03663337e-02 4.74007428e-01 5.37794195e-02
7.41853118e-02 -9.58901346e-01 1.56669155e-01 4.96281475e-01
1.09292519e+00 -1.14936686e+00 -3.14057261e-01 -2.95621783e-01
-5.52297354e-01 1.25590038e+00 9.50211346e-01 -5.10400712e-01
1.02339220e+00 6.52268887e-01 2.54381508e-01 -3.46711934e-01
-4.57627594e-01 -6.60406351e-01 4.41713005e-01 7.31319249e-01
3.89624596e-01 -3.20359111e-01 -1.60012245e-02 5.51816463e-01
2.49619540e-02 4.51169640e-01 1.73123777e-02 9.12583470e-01
-5.64966440e-01 -8.37882042e-01 -1.71380609e-01 3.73685122e-01
-3.76039684e-01 -9.27586108e-02 -7.79069781e-01 1.10318244e+00
7.23828554e-01 5.75126946e-01 3.72471631e-01 -5.99797368e-01
3.90075833e-01 4.34715264e-02 5.04293203e-01 -6.17578268e-01
-7.13332593e-01 1.69835523e-01 -2.12271899e-01 -9.20616686e-01
-4.31779534e-01 -2.68915683e-01 -1.39060295e+00 -6.07136860e-02
-3.20662171e-01 -2.48753399e-01 4.26992118e-01 8.66212845e-01
3.90567988e-01 6.77939177e-01 7.40851402e-01 -1.32628369e+00
2.33244114e-02 -1.16041458e+00 -6.45172060e-01 1.47914097e-01
4.19501781e-01 -8.02459061e-01 1.08009763e-01 -8.21021199e-02] | [13.574636459350586, 1.4641364812850952] |
1fa57bca-f4f0-4b00-8276-b423a8d76933 | empathetic-bert2bert-conversational-model | 2103.04353 | null | https://arxiv.org/abs/2103.04353v1 | https://arxiv.org/pdf/2103.04353v1.pdf | Empathetic BERT2BERT Conversational Model: Learning Arabic Language Generation with Little Data | Enabling empathetic behavior in Arabic dialogue agents is an important aspect of building human-like conversational models. While Arabic Natural Language Processing has seen significant advances in Natural Language Understanding (NLU) with language models such as AraBERT, Natural Language Generation (NLG) remains a challenge. The shortcomings of NLG encoder-decoder models are primarily due to the lack of Arabic datasets suitable to train NLG models such as conversational agents. To overcome this issue, we propose a transformer-based encoder-decoder initialized with AraBERT parameters. By initializing the weights of the encoder and decoder with AraBERT pre-trained weights, our model was able to leverage knowledge transfer and boost performance in response generation. To enable empathy in our conversational model, we train it using the ArabicEmpatheticDialogues dataset and achieve high performance in empathetic response generation. Specifically, our model achieved a low perplexity value of 17.0 and an increase in 5 BLEU points compared to the previous state-of-the-art model. Also, our proposed model was rated highly by 85 human evaluators, validating its high capability in exhibiting empathy while generating relevant and fluent responses in open-domain settings. | ['Hazem Hajj', 'Reem A. Mahmoud', 'Wissam Antoun', 'Tarek Naous'] | 2021-03-07 | null | https://aclanthology.org/2021.wanlp-1.17 | https://aclanthology.org/2021.wanlp-1.17.pdf | eacl-wanlp-2021-4 | ['empathetic-response-generation'] | ['natural-language-processing'] | [-2.97270775e-01 7.65725434e-01 2.10674465e-01 -4.45220411e-01
-8.54542553e-01 -4.41893578e-01 9.06143010e-01 -2.03144819e-01
-4.63973969e-01 1.17851162e+00 7.47790039e-01 4.37997952e-02
4.01037186e-01 -7.65338898e-01 -2.58129030e-01 -2.52525628e-01
3.62681359e-01 1.10389340e+00 -4.36108947e-01 -1.08547378e+00
8.89948159e-02 9.13363323e-02 -8.94812524e-01 7.69094646e-01
1.20376277e+00 5.40739954e-01 -9.01292562e-02 8.37413669e-01
-1.93368904e-02 1.55889738e+00 -1.06145692e+00 -1.14089930e+00
-1.90090418e-01 -9.07307327e-01 -1.43600190e+00 -2.43266791e-01
-2.66850233e-01 -8.92299771e-01 -2.38706067e-01 2.54470050e-01
7.18703985e-01 2.28032276e-01 8.38516891e-01 -1.36100030e+00
-1.17836344e+00 1.02158046e+00 -6.52360693e-02 -4.36263680e-01
7.18088031e-01 3.25327426e-01 1.14428735e+00 -6.94918334e-01
7.63609350e-01 1.44467866e+00 5.45575798e-01 1.23726630e+00
-9.33628440e-01 -3.62771630e-01 -3.64980578e-01 1.75540552e-01
-6.47502363e-01 -4.82303321e-01 4.57914650e-01 -1.78012505e-01
1.36164010e+00 -1.82412758e-01 4.56825525e-01 1.73772490e+00
7.57352412e-02 9.52271760e-01 9.81165767e-01 -3.98239136e-01
-3.65271457e-02 3.44700485e-01 -2.16544569e-01 4.18860465e-01
-3.66630197e-01 -1.26294047e-01 -7.15763748e-01 -3.96482907e-02
5.43468475e-01 -6.35269940e-01 -5.69937974e-02 3.77090037e-01
-1.32439029e+00 1.41741037e+00 4.92911458e-01 1.69511642e-02
-7.01916993e-01 5.66384718e-02 6.37506545e-01 5.38483560e-01
3.54173124e-01 1.08176327e+00 1.29397232e-02 -6.81160152e-01
-3.59105825e-01 4.31827188e-01 1.30003178e+00 8.33697677e-01
2.01619565e-01 1.37098119e-01 -4.36808765e-01 1.31361210e+00
7.39277378e-02 5.69186747e-01 5.95272243e-01 -1.09822822e+00
5.82527637e-01 5.94235539e-01 3.02145749e-01 -6.54041708e-01
-3.48792940e-01 2.37064883e-02 -5.00622749e-01 7.06982017e-02
6.45824313e-01 -6.91374302e-01 -2.75881499e-01 1.88804674e+00
-2.30510328e-02 -5.93531847e-01 7.99087346e-01 1.05962801e+00
9.62355196e-01 9.50465918e-01 2.12942004e-01 7.69291222e-02
1.30345559e+00 -1.23907471e+00 -6.50813401e-01 -4.70042974e-01
8.53201687e-01 -9.36956346e-01 1.23186755e+00 2.41336629e-01
-1.44128072e+00 -1.49503872e-01 -6.67649150e-01 -2.98725516e-01
-2.31055785e-02 1.11363709e-01 6.97325647e-01 4.29481000e-01
-1.02359664e+00 1.55467153e-01 -3.48282665e-01 -6.14540875e-01
2.06026528e-03 1.29978061e-01 -5.32849193e-01 5.28779067e-02
-1.65308607e+00 1.78358388e+00 2.53739029e-01 -2.09770929e-02
-7.46855199e-01 -3.17248583e-01 -7.36506283e-01 8.11452344e-02
-2.63688624e-01 -7.93235719e-01 1.77999032e+00 -1.00753558e+00
-2.12450409e+00 7.65511334e-01 1.81808308e-01 -7.72201240e-01
5.24480581e-01 -1.99839234e-01 -2.88943559e-01 3.57504696e-01
-3.91982123e-02 1.22412372e+00 3.24168324e-01 -1.01582968e+00
-3.20788294e-01 1.45567000e-01 4.31510895e-01 6.22473240e-01
-4.33846027e-01 2.98193663e-01 2.15430453e-01 -3.91424894e-01
-5.92776477e-01 -9.44132566e-01 -4.34823474e-03 -3.46201539e-01
-5.31051960e-03 -2.75990099e-01 3.46588790e-01 -7.72868335e-01
1.01298809e+00 -1.55658388e+00 2.06268821e-02 -2.56515265e-01
-5.63673601e-02 3.39739084e-01 -5.46949625e-01 1.14663470e+00
2.56144553e-01 -2.86653817e-01 -4.94272821e-02 -2.78975785e-01
3.10260653e-01 1.52967796e-01 -4.00745034e-01 -8.83352831e-02
5.74988842e-01 1.07401681e+00 -1.12691224e+00 -3.67370039e-01
-9.70858335e-02 5.43034911e-01 -7.76025832e-01 7.89207935e-01
-2.59626061e-01 4.21388716e-01 -2.22617403e-01 2.54843354e-01
2.30130240e-01 -1.68361738e-01 8.06212947e-02 2.86386311e-01
1.74144059e-01 6.65240407e-01 -3.18170935e-01 1.45660055e+00
-9.18034315e-01 6.12430155e-01 -1.47336006e-01 -3.84235293e-01
1.29885066e+00 6.68372512e-01 1.84680402e-01 -9.37363505e-01
3.17267686e-01 2.91922301e-01 3.44076782e-01 -5.02529860e-01
8.48463714e-01 -4.11245465e-01 -3.68843138e-01 1.30390477e+00
1.11894205e-01 -4.64875281e-01 1.46604344e-01 5.59440434e-01
8.58657360e-01 -1.86497886e-02 1.20851479e-01 1.03884570e-01
5.69611907e-01 3.47863108e-01 -1.23157769e-01 4.14991170e-01
-2.59780973e-01 4.72965121e-01 6.62834942e-01 -2.22034961e-01
-1.00786591e+00 -7.54817247e-01 2.42772207e-01 1.33429682e+00
-4.25889611e-01 -2.31578141e-01 -1.18315935e+00 -4.87638563e-01
-3.87096107e-01 1.16569436e+00 -4.66696650e-01 -4.22067136e-01
-5.66358507e-01 -7.15455830e-01 1.25785637e+00 4.97714937e-01
5.59183002e-01 -1.63490498e+00 -6.32362068e-01 6.48045659e-01
-9.12724912e-01 -1.16365707e+00 -3.16802740e-01 -2.39998132e-01
-4.29427654e-01 -7.40839005e-01 -8.41697752e-01 -7.84178913e-01
4.48971152e-01 -1.01692811e-01 1.34147465e+00 -1.94081977e-01
1.86871186e-01 1.46726355e-01 -7.42300868e-01 -3.95341069e-01
-1.17177725e+00 3.27275336e-01 -1.21878013e-01 -4.37711716e-01
4.76966649e-01 -2.07066849e-01 -5.26275098e-01 3.98342013e-01
-5.94581664e-01 3.60604852e-01 5.49044013e-01 1.28480506e+00
-5.36408722e-01 -9.47394848e-01 1.39836776e+00 -6.84461832e-01
1.54156327e+00 -6.46483183e-01 1.14336185e-01 1.96368501e-01
-4.78798836e-01 3.42640169e-02 8.72012854e-01 -4.33102101e-01
-1.54142451e+00 -3.39501798e-01 -5.08671284e-01 3.77700925e-01
5.03700450e-02 4.65471357e-01 2.68168747e-01 3.65808129e-01
1.07733583e+00 -9.59306583e-02 5.99677205e-01 9.17933583e-02
7.18428791e-01 1.23913300e+00 4.94620353e-01 -9.54648495e-01
1.63564444e-01 -1.42610744e-01 -6.14642441e-01 -4.72009093e-01
-6.35975420e-01 5.00410004e-03 -2.11098835e-01 -4.12053466e-01
8.27393949e-01 -1.13541615e+00 -1.13549149e+00 5.40131390e-01
-1.49477267e+00 -7.03951478e-01 -6.14416562e-02 3.68406117e-01
-8.83135378e-01 1.12453662e-01 -1.21243536e+00 -8.38880956e-01
-8.77389252e-01 -1.04636729e+00 7.72407293e-01 1.21930234e-01
-1.12862217e+00 -1.12924242e+00 3.42151105e-01 7.87333608e-01
7.57403970e-01 1.49913162e-01 9.24579501e-01 -8.55610132e-01
1.47749126e-01 -1.92357406e-01 -2.78157681e-01 2.38936901e-01
-1.40704334e-01 -2.87466645e-01 -9.66413319e-01 -2.84320563e-02
-1.70385048e-01 -1.30434811e+00 1.49284333e-01 -2.32789263e-01
1.58077702e-01 -6.94758654e-01 3.11921209e-01 -8.16466659e-03
6.57952845e-01 9.53166783e-02 6.74414635e-01 3.78915727e-01
2.49706089e-01 9.85974729e-01 7.36698568e-01 6.73630297e-01
9.45179999e-01 6.71202421e-01 2.63069302e-01 -3.99215221e-02
3.05102412e-02 -4.76280868e-01 8.24913263e-01 8.27555716e-01
-1.14329539e-01 -4.98366296e-01 -9.62325633e-01 6.38099492e-01
-1.88863873e+00 -1.12303567e+00 -2.54289031e-01 1.67454088e+00
1.37591004e+00 -3.26918453e-01 2.85960943e-01 -4.09282416e-01
3.56488377e-01 2.44020149e-02 -2.41111353e-01 -1.34895873e+00
-2.01497003e-01 2.07578331e-01 -1.60820484e-01 7.59240568e-01
-3.61943811e-01 1.36610019e+00 6.07748079e+00 2.34587207e-01
-9.16322052e-01 -1.30821555e-03 5.54378510e-01 -3.49185690e-02
-3.37784588e-01 -2.84378380e-01 -5.32449245e-01 3.16613227e-01
1.31553960e+00 -3.35895061e-01 5.62361896e-01 7.54320443e-01
4.13840413e-01 1.78380176e-01 -1.13784754e+00 6.25206590e-01
3.19196165e-01 -1.02059734e+00 2.28893384e-01 -1.80341646e-01
7.44670331e-01 -2.22706586e-01 -1.01636268e-01 7.98192143e-01
8.70723903e-01 -1.40334320e+00 4.78005171e-01 2.85603642e-01
6.40188336e-01 -8.52628767e-01 1.07919979e+00 3.99397731e-01
-2.80536234e-01 -4.21976931e-02 -3.40679109e-01 -3.28085750e-01
4.71600145e-01 -1.52065963e-01 -1.63977563e+00 1.06837891e-01
1.51475951e-01 3.61979157e-01 -1.78159550e-01 3.50212574e-01
-5.01989663e-01 4.55016226e-01 3.26711833e-02 -3.85348916e-01
6.25057876e-01 -3.35264027e-01 7.18811825e-02 1.36874831e+00
2.97574908e-01 2.52867669e-01 -2.47482643e-01 8.24943066e-01
-2.99545676e-01 3.20103496e-01 -5.49583793e-01 -2.49401733e-01
5.13674974e-01 1.24459326e+00 1.17988020e-01 -1.62356764e-01
-2.45143279e-01 1.22811627e+00 6.50431693e-01 1.37350127e-01
-9.48701441e-01 -3.89370382e-01 5.24499416e-01 -2.30940208e-01
-3.43304574e-01 4.83242944e-02 -2.73714930e-01 -8.89159381e-01
-3.89737695e-01 -1.40564561e+00 2.54727304e-01 -1.13278091e+00
-1.43606627e+00 1.02181530e+00 -2.02943996e-01 -1.12331295e+00
-1.12948811e+00 -5.84878623e-01 -7.33439207e-01 1.11034203e+00
-1.14556217e+00 -1.57296109e+00 -2.62823373e-01 6.22486055e-01
6.63541675e-01 -3.86996299e-01 1.25811398e+00 5.47349639e-02
-2.77024299e-01 8.99038792e-01 -2.18205303e-01 2.25095376e-01
1.27097678e+00 -1.08500159e+00 4.45286810e-01 4.08736229e-01
-4.88689899e-01 7.70544350e-01 6.86099529e-01 -3.63477707e-01
-9.74230766e-01 -7.66836405e-01 1.38753903e+00 -6.24510586e-01
8.66376281e-01 -2.08420143e-01 -6.55393124e-01 6.39207244e-01
8.76808643e-01 -8.71640027e-01 8.67899895e-01 -2.91865990e-02
-2.75185496e-01 3.37747395e-01 -1.18513536e+00 8.82715523e-01
6.21209562e-01 -4.53789741e-01 -8.28144014e-01 4.94120896e-01
5.20785511e-01 -5.04001260e-01 -1.08180165e+00 -1.66996643e-01
6.24555290e-01 -9.23528790e-01 7.61673987e-01 -8.95554304e-01
1.34352326e+00 3.28734636e-01 2.46029213e-01 -1.90347302e+00
8.03868994e-02 -8.91164303e-01 -2.28641722e-02 1.24767137e+00
7.37696648e-01 -6.86445653e-01 6.00328982e-01 1.02746499e+00
-2.50613421e-01 -5.61435759e-01 -4.70110476e-01 -2.51622379e-01
5.95124304e-01 4.24505286e-02 5.80039084e-01 9.45551932e-01
9.62917626e-01 9.13380384e-01 -7.59165585e-01 -5.49116552e-01
1.06292134e-02 -1.75730780e-01 1.04614806e+00 -9.10159230e-01
-2.00139835e-01 -5.27368844e-01 2.41231069e-01 -9.74461973e-01
6.05086088e-01 -8.63163173e-01 2.41258308e-01 -1.85162926e+00
9.20943394e-02 -2.45594561e-01 5.54235816e-01 7.02983201e-01
-2.73395658e-01 1.59181759e-01 2.74327040e-01 7.26366863e-02
-2.89251715e-01 8.28400552e-01 1.39831412e+00 2.26186529e-01
-1.23026870e-01 -1.86467692e-01 -9.79506195e-01 3.19062203e-01
9.70071137e-01 -3.07590753e-01 -5.33317506e-01 -6.51827037e-01
2.46979117e-01 4.13750410e-01 1.03524774e-01 -4.97250140e-01
2.57546514e-01 -1.83903873e-01 5.80097958e-02 -8.30589607e-03
6.80070043e-01 -3.14349711e-01 -2.60172576e-01 3.23054761e-01
-7.41405666e-01 3.61190051e-01 9.23999399e-02 -1.65486053e-01
-4.04661238e-01 -2.95859575e-01 7.61831403e-01 -1.25764981e-01
-3.77211094e-01 -2.44132742e-01 -8.06005180e-01 2.99426258e-01
8.55040491e-01 -1.15392856e-01 -9.29505825e-01 -1.20636761e+00
-2.25337565e-01 4.64049727e-01 4.48932260e-01 4.81075197e-01
6.38452828e-01 -1.21373630e+00 -1.20759141e+00 -2.25937009e-01
3.52883518e-01 -3.55126709e-01 1.65719032e-01 7.31103420e-01
-8.60964358e-01 5.17466664e-01 -8.08407605e-01 6.43774495e-02
-8.90386105e-01 -1.58530295e-01 3.49107414e-01 -4.68347430e-01
-3.61912847e-01 7.39636779e-01 -1.27246052e-01 -8.59724462e-01
-7.50761777e-02 3.44590932e-01 -4.30286050e-01 1.28601104e-01
5.46910465e-01 3.15755576e-01 -9.49189886e-02 -7.73979425e-01
-1.63563481e-03 -1.98972136e-01 -2.47701988e-01 -6.64294004e-01
1.13205242e+00 7.74433613e-02 -7.44646266e-02 2.53778184e-03
9.63671505e-01 -7.32366294e-02 -1.01219332e+00 1.09019615e-01
-1.81271195e-01 -2.74874885e-02 -5.16212821e-01 -1.27152157e+00
-4.82746571e-01 9.07939494e-01 -1.91783667e-01 -1.38391539e-01
6.39032066e-01 -2.39894956e-01 1.25623608e+00 7.75562942e-01
3.65151465e-01 -1.20419776e+00 4.64403868e-01 1.22384834e+00
1.20830858e+00 -1.41479361e+00 -4.73897964e-01 1.64359778e-01
-1.73775232e+00 1.20335686e+00 1.05495834e+00 4.56461199e-02
-2.84927547e-01 1.55139148e-01 7.82214165e-01 -3.57033056e-03
-1.15605736e+00 7.63048744e-03 -9.98680070e-02 7.95902848e-01
1.09699154e+00 4.15722765e-02 -5.22799253e-01 7.34145403e-01
-9.46084082e-01 -5.42808138e-02 9.23515201e-01 5.32366097e-01
-1.94596246e-01 -1.21898139e+00 -1.48649052e-01 -6.13556728e-02
-3.09530407e-01 -2.30816111e-01 -8.92929792e-01 8.63712728e-01
-7.28193521e-01 1.40497577e+00 1.75356735e-02 -3.10654938e-01
4.10623401e-01 2.14043945e-01 2.84222096e-01 -6.10343397e-01
-1.33192098e+00 -5.92709899e-01 8.95936131e-01 -2.60657519e-01
-1.19299956e-01 -3.30270380e-01 -1.43786001e+00 -7.30880916e-01
-7.48920366e-02 4.52244431e-01 4.66724575e-01 8.88711333e-01
2.22076103e-01 1.52574420e-01 6.33149266e-01 -5.60078800e-01
-8.85543942e-01 -1.41906202e+00 -9.55367386e-02 7.26018250e-01
-2.14072555e-01 -1.92789823e-01 -1.31818622e-01 -7.65156671e-02] | [13.0009765625, 7.827927589416504] |
0e77fed9-5338-4ed3-ab5e-b2e7af4243bf | anomaly-crossing-a-new-method-for-video | 2112.06320 | null | https://arxiv.org/abs/2112.06320v3 | https://arxiv.org/pdf/2112.06320v3.pdf | Anomaly Crossing: New Horizons for Video Anomaly Detection as Cross-domain Few-shot Learning | Video anomaly detection aims to identify abnormal events that occurred in videos. Since anomalous events are relatively rare, it is not feasible to collect a balanced dataset and train a binary classifier to solve the task. Thus, most previous approaches learn only from normal videos using unsupervised or semi-supervised methods. Obviously, they are limited in capturing and utilizing discriminative abnormal characteristics, which leads to compromised anomaly detection performance. In this paper, to address this issue, we propose a new learning paradigm by making full use of both normal and abnormal videos for video anomaly detection. In particular, we formulate a new learning task: cross-domain few-shot anomaly detection, which can transfer knowledge learned from numerous videos in the source domain to help solve few-shot abnormality detection in the target domain. Concretely, we leverage self-supervised training on the target normal videos to reduce the domain gap and devise a meta context perception module to explore the video context of the event in the few-shot setting. Our experiments show that our method significantly outperforms baseline methods on DoTA and UCF-Crime datasets, and the new task contributes to a more practical training paradigm for anomaly detection. | ['Chenliang Xu', 'Jing Shi', 'Lianggong Wen', 'Zhang Liu', 'Guangyu Sun'] | 2021-12-12 | null | null | null | null | ['cross-domain-few-shot', 'cross-domain-few-shot-learning'] | ['computer-vision', 'computer-vision'] | [ 2.69104034e-01 -4.15069491e-01 -2.49117404e-01 -3.65807921e-01
-5.14314473e-01 -3.40607196e-01 4.04307693e-01 2.50184596e-01
-2.59342164e-01 3.30570012e-01 2.27063328e-01 4.75244597e-02
1.37626827e-01 -4.89129573e-01 -6.57993078e-01 -6.11524463e-01
-1.35658339e-01 -8.49033818e-02 3.03128570e-01 -1.49171120e-02
2.84046531e-01 2.90359586e-01 -1.65962446e+00 5.29794097e-01
1.00652730e+00 1.14595282e+00 -9.81468111e-02 3.05304915e-01
-1.84747279e-01 1.07684243e+00 -5.34096360e-01 -2.45887786e-02
2.55040258e-01 -7.15012312e-01 -5.46065867e-01 5.25053680e-01
5.19309819e-01 -7.71696746e-01 -5.82847655e-01 1.17772675e+00
2.23959982e-01 4.32045072e-01 6.69071198e-01 -1.58200586e+00
-4.24850136e-01 4.95264307e-02 -6.34681046e-01 1.10645056e+00
6.33620620e-01 3.72542739e-01 9.05731738e-01 -8.62327456e-01
5.00876665e-01 9.30606186e-01 4.08657968e-01 5.87634981e-01
-8.25698256e-01 -5.97213268e-01 6.34359121e-01 7.39910960e-01
-1.13009810e+00 -5.22490919e-01 9.91566300e-01 -5.93002319e-01
7.63325095e-01 1.42577156e-01 8.01798880e-01 1.62655854e+00
-1.47808537e-01 1.07690895e+00 5.67245841e-01 -2.50575542e-01
3.99365664e-01 -3.82096171e-01 -8.62578023e-03 5.17620802e-01
2.70575017e-01 -2.06411898e-01 -5.93979716e-01 -2.69471973e-01
4.02953178e-01 6.70928478e-01 -3.39584589e-01 -3.39004636e-01
-1.15464473e+00 7.67763853e-01 1.70367565e-02 4.61164176e-01
-5.73620021e-01 -3.04849654e-01 9.59403276e-01 6.35350585e-01
6.61632299e-01 3.82876962e-01 -1.89955458e-01 -4.58158374e-01
-7.26699233e-01 2.87264645e-01 5.24283826e-01 6.36322021e-01
6.39218211e-01 2.11274356e-01 -2.96737999e-01 6.96690023e-01
1.02078475e-01 2.62032181e-01 8.10554922e-01 -7.25602865e-01
6.38302565e-01 7.29349852e-01 -5.78982532e-02 -1.43790269e+00
-1.00500695e-01 6.99415132e-02 -6.33850336e-01 -1.49467915e-01
5.58216035e-01 -1.33300368e-02 -9.37592268e-01 1.53778243e+00
3.97236943e-01 9.28560853e-01 8.14131869e-04 9.60466981e-01
4.07723606e-01 7.07731724e-01 2.29619294e-01 -4.44628805e-01
1.04471028e+00 -8.75593662e-01 -8.51967275e-01 -4.84651178e-01
9.66332376e-01 -3.94411683e-01 1.21536589e+00 4.91231889e-01
-7.18223035e-01 -3.99148554e-01 -9.58791733e-01 3.21060121e-01
-3.18786085e-01 -2.29494542e-01 3.52137417e-01 2.48568505e-01
-3.61393511e-01 3.88018906e-01 -9.69751000e-01 -4.90646631e-01
7.75591195e-01 -1.45843029e-01 -4.98842984e-01 -5.25846064e-01
-1.02530634e+00 4.16846067e-01 7.00747669e-01 -2.03281678e-02
-9.83564496e-01 -4.13197219e-01 -1.13160694e+00 -1.76867723e-01
6.91202164e-01 -1.25558734e-01 9.79476333e-01 -1.45305073e+00
-8.56972694e-01 8.12248290e-01 -1.66697279e-01 -5.29264033e-01
2.60566413e-01 -4.28232133e-01 -8.49669993e-01 4.73671138e-01
2.67521024e-01 7.40328506e-02 1.16643417e+00 -8.10282528e-01
-8.10263336e-01 -4.97364163e-01 6.42488105e-03 1.57743573e-01
-8.66323948e-01 -2.36339550e-02 -3.95228982e-01 -1.07852328e+00
7.25395679e-02 -5.55974007e-01 -5.00744991e-02 -2.10810646e-01
-1.60247639e-01 -1.89956337e-01 1.32437944e+00 -8.16275120e-01
1.50777972e+00 -2.65346122e+00 1.97921358e-02 1.66776836e-01
3.08352053e-01 4.97038394e-01 -1.92010999e-01 2.42305040e-01
-2.37636879e-01 -2.74657667e-01 -2.78239131e-01 -5.47074936e-02
-3.14572215e-01 4.61809754e-01 -5.27357817e-01 4.87429798e-01
4.62305576e-01 5.61088502e-01 -1.21112454e+00 -5.77469707e-01
1.01399869e-01 -6.80181608e-02 -7.93115377e-01 5.45615375e-01
-1.28077209e-01 7.38087595e-01 -7.33366191e-01 9.73702013e-01
4.75176305e-01 -1.80934891e-02 -2.06103623e-01 5.80157302e-02
2.71518528e-01 -1.00025617e-01 -1.08169842e+00 1.81401742e+00
-1.66233792e-03 5.53921402e-01 -2.30852932e-01 -1.71648192e+00
5.82826376e-01 2.73845285e-01 7.87531912e-01 -8.03034186e-01
-1.31278172e-01 2.35354334e-01 -2.20149644e-02 -1.26328707e+00
1.39216363e-01 1.25706270e-01 4.58130287e-03 3.25853139e-01
1.59859598e-01 5.54081738e-01 5.17937362e-01 2.96555728e-01
1.48769486e+00 -3.23752724e-02 5.01209199e-01 2.82244772e-01
6.81756496e-01 4.08284999e-02 9.35701549e-01 7.71408856e-01
-6.24440670e-01 6.77929759e-01 7.27980971e-01 -7.76270330e-01
-8.34662259e-01 -9.51869965e-01 1.86204404e-01 1.32176673e+00
2.62641102e-01 -5.95658720e-01 -6.84577048e-01 -1.28083503e+00
-7.52453133e-02 5.76911032e-01 -6.18375957e-01 -5.75705409e-01
-7.27324426e-01 -7.50255346e-01 4.01352555e-01 5.52434623e-01
4.58281368e-01 -1.15711570e+00 -4.33376551e-01 1.25123471e-01
-5.07241845e-01 -1.36499429e+00 -5.47385275e-01 -3.06149781e-01
-8.78756642e-01 -1.28459668e+00 -3.98219705e-01 -5.80790043e-01
6.57497108e-01 4.77169484e-01 7.79090941e-01 1.66618496e-01
-4.55287784e-01 6.53336525e-01 -8.57303917e-01 -3.23303282e-01
-2.23106489e-01 -2.99264014e-01 3.33965421e-01 5.11132717e-01
1.03278482e+00 -5.97511351e-01 -6.53242588e-01 2.84111440e-01
-1.12772727e+00 -4.58340347e-01 5.06223559e-01 8.39556754e-01
5.52497447e-01 1.47618890e-01 6.74949706e-01 -6.54915690e-01
3.31955224e-01 -1.01555562e+00 -1.64916530e-01 -3.76442187e-02
-2.09453002e-01 -4.27012533e-01 8.16092134e-01 -5.23978055e-01
-9.49827850e-01 3.30509134e-02 -2.15266161e-02 -8.98163438e-01
-6.12110853e-01 4.55281109e-01 -3.24321926e-01 2.05282271e-01
6.37036502e-01 4.34330612e-01 6.48107156e-02 -3.76336008e-01
-1.47437781e-01 5.51208675e-01 5.33942699e-01 -3.71315241e-01
7.83865750e-01 7.18883157e-01 -3.40706706e-01 -8.88456821e-01
-1.09591579e+00 -9.33669686e-01 -6.12346947e-01 -3.04935008e-01
8.46846581e-01 -1.06957781e+00 -2.56254449e-02 5.38881898e-01
-8.08587730e-01 -8.14511552e-02 -4.84952688e-01 5.82713068e-01
-5.58846056e-01 8.76501083e-01 -3.05975616e-01 -6.06911421e-01
1.42645001e-01 -9.27327394e-01 9.36388433e-01 8.82899296e-03
-2.27474123e-01 -7.44283855e-01 1.82044864e-01 2.83915222e-01
1.38497069e-01 3.90887886e-01 5.43449640e-01 -1.28405499e+00
-3.68787110e-01 -4.94605184e-01 -1.81765541e-01 5.25678337e-01
3.06478202e-01 -3.12562168e-01 -8.55675519e-01 -4.01364028e-01
2.15359509e-01 -4.22264397e-01 1.00339091e+00 7.37935379e-02
1.83965850e+00 -3.91797513e-01 -2.97687024e-01 5.86664200e-01
9.60651875e-01 1.75577566e-01 6.22717261e-01 4.60990638e-01
9.84137833e-01 5.39323032e-01 1.00798059e+00 7.15457380e-01
2.35595495e-01 5.46961725e-01 4.31023657e-01 2.70914108e-01
2.47584060e-01 -1.77338958e-01 6.87842786e-01 6.24176562e-01
-1.79839566e-01 -1.07223630e-01 -9.20308650e-01 7.15095818e-01
-2.11446285e+00 -1.41851783e+00 2.33166680e-01 2.01354384e+00
4.24980521e-01 1.09567255e-01 3.78954381e-01 2.50706792e-01
6.67454779e-01 3.76585871e-01 -6.07598007e-01 2.06331518e-02
4.14144434e-02 -3.13991010e-01 -1.45532683e-01 -2.26027057e-01
-1.55289924e+00 6.31257117e-01 5.58054161e+00 9.39763427e-01
-1.08038187e+00 1.27734661e-01 4.60978687e-01 -4.13022935e-01
1.71022192e-01 -2.62914866e-01 -3.70895386e-01 9.65238214e-01
7.71986783e-01 -6.65052459e-02 1.21344045e-01 1.03189588e+00
3.33162487e-01 1.56291053e-01 -1.19905078e+00 1.14853716e+00
4.81757641e-01 -8.37142110e-01 2.41345584e-01 -6.50383011e-02
5.93809724e-01 -1.25891387e-01 -8.22813362e-02 5.20916045e-01
-4.82147425e-01 -7.10126281e-01 2.03438938e-01 3.18779558e-01
4.21539694e-01 -6.84904158e-01 6.30904555e-01 3.54635447e-01
-9.93613124e-01 -5.45982718e-01 -3.48186165e-01 -1.48759916e-01
9.53977704e-02 5.29476941e-01 -4.43665892e-01 4.44807410e-01
1.00719571e+00 1.38336384e+00 -6.41356468e-01 1.07120121e+00
1.81058839e-01 7.47615993e-01 -6.12591989e-02 5.91732800e-01
4.25639004e-01 -1.52350560e-01 9.02001143e-01 1.17836988e+00
4.98712599e-01 1.85383454e-01 5.90632439e-01 3.33435714e-01
9.14751515e-02 1.83735743e-01 -1.06089199e+00 -2.50338942e-01
2.26307765e-01 9.93362784e-01 -5.58843732e-01 -3.39019656e-01
-9.33013439e-01 1.18021584e+00 2.35317141e-01 4.31927174e-01
-8.87167990e-01 -3.52811813e-01 8.86828959e-01 7.34399557e-02
3.45872760e-01 7.91747496e-02 2.47195989e-01 -1.82613623e+00
3.66102993e-01 -1.19577229e+00 9.26080465e-01 -2.71685064e-01
-1.60398090e+00 2.63498306e-01 1.41136870e-01 -1.77658319e+00
-4.93171722e-01 -5.25382340e-01 -9.96175885e-01 1.23861879e-01
-1.44777322e+00 -8.79978597e-01 -5.72156131e-01 9.87345934e-01
8.91979814e-01 -5.89745224e-01 4.73841906e-01 6.25670195e-01
-8.94642115e-01 6.25616431e-01 -4.12904881e-02 4.26911294e-01
9.15972531e-01 -1.09152257e+00 1.56294495e-01 1.40655553e+00
1.56847358e-01 2.80878007e-01 4.94141340e-01 -6.88483357e-01
-1.23720980e+00 -1.27635288e+00 2.14226484e-01 -3.81806612e-01
7.87878454e-01 -1.16106704e-01 -1.49895525e+00 8.51270974e-01
-1.52783692e-01 5.43952405e-01 8.57258677e-01 -7.94543326e-02
-3.93027842e-01 -4.78626303e-02 -1.06270826e+00 5.29918611e-01
1.31711781e+00 -5.74515581e-01 -8.94253552e-01 3.43956351e-01
4.82758760e-01 -2.83282399e-01 -4.33941990e-01 4.93803173e-01
2.31634125e-01 -9.85838234e-01 9.47298944e-01 -1.03948641e+00
5.72175384e-01 -2.79974282e-01 -2.43678495e-01 -1.34126580e+00
-5.26001714e-02 -4.30572540e-01 -8.61497104e-01 1.12797606e+00
-1.43177822e-01 -5.69443762e-01 7.44482040e-01 2.43647128e-01
-3.69414717e-01 -5.58249176e-01 -9.03662920e-01 -8.77968371e-01
-4.73072559e-01 -5.73071480e-01 4.91122544e-01 1.24298227e+00
1.45708218e-01 -1.90506175e-01 -5.44893980e-01 2.81572759e-01
5.21064520e-01 1.00843590e-02 7.58657992e-01 -1.10876071e+00
-1.30689964e-01 -2.83675343e-01 -9.60295677e-01 -8.73960495e-01
3.43875498e-01 -6.28254294e-01 6.96143880e-02 -8.59400451e-01
2.40071878e-01 4.61331643e-02 -6.45546138e-01 3.61079305e-01
-4.20760423e-01 3.92713547e-01 -1.92408144e-01 3.10506254e-01
-1.12733233e+00 5.29056489e-01 9.01294827e-01 -1.08734772e-01
-1.90883577e-01 -5.35311922e-02 -5.28335214e-01 1.14059114e+00
7.03706622e-01 -2.46586710e-01 -4.21453953e-01 -3.89452130e-01
-8.72024298e-02 -3.55748683e-01 5.16907215e-01 -1.16456962e+00
8.51777047e-02 -2.92547911e-01 4.66275394e-01 -4.98043418e-01
1.28061086e-01 -8.48424494e-01 -6.41934216e-01 2.16492400e-01
-1.80486515e-01 9.35794935e-02 8.38634744e-02 1.21672821e+00
-6.43218994e-01 -1.49938315e-01 4.81187046e-01 -1.15846723e-01
-1.43335116e+00 6.03936076e-01 -5.43253541e-01 4.25594985e-01
1.48809814e+00 -4.14910853e-01 -1.34652227e-01 -3.27251762e-01
-8.89185607e-01 3.41742754e-01 4.45626527e-01 6.49281979e-01
8.45931828e-01 -1.53564715e+00 -5.87355912e-01 4.96948570e-01
7.75925159e-01 -5.01063205e-02 6.49638116e-01 1.00421405e+00
-3.60426068e-01 4.28125635e-02 -4.60651904e-01 -7.31896877e-01
-1.06652319e+00 8.54268312e-01 3.16784419e-02 1.35960467e-02
-9.56256926e-01 4.62645292e-01 3.37818176e-01 5.01873866e-02
2.85066813e-01 6.65580183e-02 -3.92615378e-01 2.09089950e-01
9.78346109e-01 4.56004590e-01 -1.13599405e-01 -7.25267291e-01
-2.67825425e-01 2.28276342e-01 -3.62134784e-01 4.19606000e-01
1.23223901e+00 -2.76760012e-01 2.14028992e-02 4.89470214e-01
1.15508997e+00 -1.76684514e-01 -1.33077991e+00 -3.79109859e-01
2.02684030e-01 -1.02363598e+00 -1.41366407e-01 -1.69953272e-01
-1.08685017e+00 7.56077528e-01 5.69459021e-01 3.56545597e-01
1.43175876e+00 -1.65987741e-02 1.11399996e+00 5.87008238e-01
-6.64345026e-02 -1.26638448e+00 6.72856867e-01 3.65364820e-01
6.08082950e-01 -1.69339466e+00 -2.12034106e-01 -2.15308115e-01
-8.81253302e-01 9.45026278e-01 1.10225177e+00 -1.41644657e-01
3.85605663e-01 -2.17030838e-01 -9.14343074e-02 -3.00686270e-01
-4.74502921e-01 -1.99342474e-01 4.27614629e-01 6.65042818e-01
3.13368998e-02 -3.68868083e-01 -2.57469937e-02 5.11996090e-01
4.87967044e-01 -1.09531403e-01 5.10542154e-01 1.17356157e+00
-4.09263790e-01 -8.26364636e-01 -3.27391386e-01 9.00097549e-01
-8.95225465e-01 2.87561297e-01 -8.32879618e-02 5.58564067e-01
2.40966439e-01 8.00966740e-01 3.63294989e-01 -3.11723322e-01
3.62839460e-01 3.64596158e-01 1.64163157e-01 -7.32532859e-01
7.05668330e-02 -4.82658558e-02 -1.93363607e-01 -9.88721490e-01
-4.20199454e-01 -9.12238061e-01 -7.44991839e-01 -2.39793267e-02
-2.13933550e-02 5.38369305e-02 -7.53331482e-02 1.25044405e+00
4.50605780e-01 3.42466533e-01 7.70453751e-01 -7.27690339e-01
-4.18648869e-01 -7.95176983e-01 -6.05302274e-01 9.78567004e-01
7.21041799e-01 -8.39906812e-01 -5.52299857e-01 1.25613108e-01] | [7.861595153808594, 1.5908324718475342] |
c61d9dcb-e160-4f19-950e-fe049dff4606 | ranking-facts-for-explaining-answers-to | 2110.09036 | null | https://arxiv.org/abs/2110.09036v1 | https://arxiv.org/pdf/2110.09036v1.pdf | Ranking Facts for Explaining Answers to Elementary Science Questions | In multiple-choice exams, students select one answer from among typically four choices and can explain why they made that particular choice. Students are good at understanding natural language questions and based on their domain knowledge can easily infer the question's answer by 'connecting the dots' across various pertinent facts. Considering automated reasoning for elementary science question answering, we address the novel task of generating explanations for answers from human-authored facts. For this, we examine the practically scalable framework of feature-rich support vector machines leveraging domain-targeted, hand-crafted features. Explanations are created from a human-annotated set of nearly 5,000 candidate facts in the WorldTree corpus. Our aim is to obtain better matches for valid facts of an explanation for the correct answer of a question over the available fact candidates. To this end, our features offer a comprehensive linguistic and semantic unification paradigm. The machine learning problem is the preference ordering of facts, for which we test pointwise regression versus pairwise learning-to-rank. Our contributions are: (1) a case study in which two preference ordering approaches are systematically compared; (2) it is a practically competent approach that can outperform some variants of BERT-based reranking models; and (3) the human-engineered features make it an interpretable machine learning model for the task. | ['Soeren Auer', "Isaiah Onando Mulang'", "Jennifer D'Souza"] | 2021-10-18 | null | null | null | null | ['science-question-answering'] | ['miscellaneous'] | [ 2.54934818e-01 7.44936883e-01 -4.88619089e-01 -5.96207440e-01
-1.11742759e+00 -9.17974532e-01 6.32264495e-01 8.48247886e-01
-2.17737898e-01 1.09830439e+00 5.42097449e-01 -9.60930824e-01
-8.68961811e-01 -9.35374856e-01 -7.06077874e-01 3.70549671e-02
3.14338923e-01 8.09697032e-01 3.76223177e-01 -6.32107794e-01
6.68697953e-01 1.85146019e-01 -2.13680601e+00 7.88326204e-01
1.39502656e+00 8.99923742e-01 -2.45170351e-02 7.02126443e-01
-7.24998176e-01 1.21060228e+00 -5.28852463e-01 -1.11967719e+00
-1.63535848e-01 -5.45087457e-01 -1.61603630e+00 -2.99437642e-01
1.16623914e+00 7.50718117e-02 1.35164887e-01 7.87630200e-01
-5.30126914e-02 4.61217940e-01 8.61255705e-01 -1.17535830e+00
-8.75735700e-01 1.00273156e+00 1.11885034e-01 4.64446545e-01
1.13405085e+00 -3.62468451e-01 1.87742186e+00 -7.65397012e-01
6.49106383e-01 1.20273459e+00 2.71810532e-01 6.31201029e-01
-1.04347372e+00 -3.00993443e-01 3.81023467e-01 7.93891132e-01
-6.76308513e-01 3.15356217e-02 6.30035877e-01 -5.64874828e-01
7.48887658e-01 7.81450748e-01 5.46789467e-01 7.80588388e-01
5.95981302e-03 7.67020822e-01 1.32664061e+00 -8.10160756e-01
1.75664768e-01 3.68232340e-01 6.63464487e-01 7.62462914e-01
4.87999231e-01 -1.70706108e-01 -7.94634402e-01 -2.10179999e-01
2.74869144e-01 -2.29779243e-01 -2.35831305e-01 -4.26762849e-01
-1.23885775e+00 1.06159437e+00 3.11347842e-01 3.48225296e-01
-6.39779717e-02 -1.70728073e-01 -1.26975879e-01 8.17230403e-01
-1.07048126e-02 1.31404209e+00 -1.01333034e+00 7.45168850e-02
-7.74105906e-01 7.32018471e-01 1.14001894e+00 7.21607506e-01
6.73622072e-01 -5.20125091e-01 -2.30705604e-01 5.49377441e-01
2.62373924e-01 1.22374952e-01 6.09409034e-01 -1.14457321e+00
5.72543025e-01 9.86211121e-01 1.74436167e-01 -1.07985592e+00
-1.77535325e-01 -3.26633304e-01 -1.39648765e-01 4.04550582e-02
8.21422756e-01 7.97826350e-02 -3.32704544e-01 1.59658062e+00
5.16382277e-01 9.42953601e-02 3.40362161e-01 7.84106493e-01
1.50449002e+00 8.64461139e-02 3.41405906e-02 1.93280429e-02
1.55451798e+00 -8.49834561e-01 -4.60653305e-01 -1.48731470e-01
7.04762876e-01 -7.83054829e-01 1.12926471e+00 5.14627278e-01
-1.09189141e+00 -6.30073845e-01 -9.00678575e-01 -5.48582613e-01
-5.83501399e-01 -1.33157715e-01 9.23261702e-01 4.87870842e-01
-7.12295294e-01 6.22705281e-01 1.79974109e-01 -1.13054700e-02
3.48145664e-02 3.89296353e-01 -4.19018626e-01 -2.28661716e-01
-1.30921221e+00 1.10851598e+00 3.87973696e-01 -6.17030263e-01
-1.47404298e-01 -1.03606355e+00 -1.08472383e+00 3.97244275e-01
7.03722715e-01 -1.05729997e+00 1.44425309e+00 -7.54897535e-01
-1.34729230e+00 1.22574830e+00 -3.11723083e-01 -3.83258134e-01
4.61325169e-01 -2.46546030e-01 -4.12403077e-01 9.12113860e-02
5.86566806e-01 4.56209093e-01 3.88838977e-01 -1.16599476e+00
-8.60845864e-01 3.15513574e-02 5.16044796e-01 1.45087242e-01
-7.51251578e-02 -2.87351042e-01 2.59133190e-01 -2.88702369e-01
4.84901875e-01 -5.27313530e-01 -2.28638962e-01 -3.84639889e-01
-4.54198092e-01 -8.48053873e-01 1.83049068e-01 -1.76329046e-01
1.17204475e+00 -1.40698314e+00 1.48366289e-02 3.69743764e-01
4.91135091e-01 7.25729540e-02 -4.79943566e-02 3.91092896e-01
-4.24120069e-01 2.49581605e-01 6.55626133e-02 2.96723723e-01
3.68543386e-01 2.07966283e-01 -9.22523379e-01 -1.76888913e-01
-3.84924375e-02 8.72899175e-01 -1.23175323e+00 -6.45151854e-01
2.06168815e-01 -3.51364940e-01 -8.82836580e-01 2.17152059e-01
-5.04588127e-01 1.37416974e-01 -7.08547533e-01 3.11407417e-01
1.47030666e-01 -3.19753915e-01 1.02781788e-01 1.67598441e-01
3.73888901e-03 9.12012398e-01 -1.32904303e+00 1.27685618e+00
-4.68693435e-01 5.68518043e-01 -6.76657975e-01 -9.92452323e-01
9.73963559e-01 3.48164290e-01 -1.90261565e-02 -4.55394894e-01
-3.53432782e-02 6.08031571e-01 1.02642559e-01 -8.71472955e-01
5.36625743e-01 -1.90997884e-01 -2.55021334e-01 3.56106728e-01
1.81357741e-01 -6.33489907e-01 5.73593915e-01 4.37752277e-01
1.06522095e+00 2.06448492e-02 8.13235164e-01 -4.70665812e-01
9.52155650e-01 3.28741670e-01 1.90616101e-01 9.83396769e-01
1.76873192e-01 3.86143982e-01 6.38472736e-01 -8.85342956e-01
-4.63505447e-01 -1.02906585e+00 -8.30270946e-02 1.32732177e+00
2.67413221e-02 -4.80687976e-01 -1.94308832e-01 -1.08775318e+00
4.51140046e-01 1.26010191e+00 -7.34751046e-01 2.36122742e-01
-5.26587248e-01 9.92417932e-02 4.58178557e-02 3.00925940e-01
2.27026388e-01 -7.86650538e-01 -5.28893352e-01 -5.29885963e-02
-4.43093747e-01 -8.68522882e-01 -7.11040944e-02 2.25423932e-01
-6.14731252e-01 -1.57689655e+00 -1.56828195e-01 -8.47937703e-01
5.37136257e-01 2.06836924e-01 1.83065188e+00 6.55229986e-01
8.27032328e-02 6.75289452e-01 -3.24022025e-01 -5.86920023e-01
-1.72427729e-01 4.97923903e-02 -5.99813238e-02 -4.65287089e-01
7.91583598e-01 -5.74248731e-01 -2.67164141e-01 6.62157731e-03
-7.26858079e-01 -1.74861670e-01 3.07310671e-01 8.61328602e-01
5.07578790e-01 -1.51378006e-01 5.83189785e-01 -1.40942228e+00
8.10658216e-01 -6.45903289e-01 -4.49407846e-01 5.94954908e-01
-5.83984733e-01 5.42164803e-01 7.85245419e-01 -1.63683370e-01
-1.04155946e+00 -2.96284825e-01 -1.45640224e-01 1.84225142e-01
-3.06608140e-01 6.66439950e-01 -1.23318527e-02 -4.13500331e-02
1.03179812e+00 -1.31588057e-01 -6.08896136e-01 -2.50927776e-01
6.41584516e-01 9.32000577e-02 5.60426831e-01 -1.22063506e+00
1.04486167e+00 -1.30111679e-01 3.42156321e-01 -4.44612771e-01
-1.77166641e+00 -5.64759314e-01 -6.29169106e-01 -1.11196645e-01
4.92423683e-01 -6.00340188e-01 -1.20551348e+00 -6.67975962e-01
-1.15041530e+00 7.63999075e-02 -4.75397646e-01 6.62897050e-01
-5.47304153e-01 1.32179782e-01 -2.46319011e-01 -6.05013132e-01
3.29105198e-01 -8.44018579e-01 5.86904824e-01 5.08618951e-01
-8.89266014e-01 -1.30531001e+00 1.29033342e-01 1.06575096e+00
7.94836953e-02 7.64133483e-02 1.56658900e+00 -1.49383843e+00
-5.82727611e-01 -1.52915968e-02 9.24743488e-02 -8.84156227e-02
-1.78744793e-01 1.24298427e-02 -8.89900923e-01 2.46975645e-01
-1.09631997e-02 -5.22485733e-01 9.87272918e-01 8.66299644e-02
1.22081673e+00 -5.90863705e-01 -1.72172412e-01 -8.81532058e-02
1.22262478e+00 -4.01164651e-01 1.36021882e-01 5.67496955e-01
3.86077672e-01 1.18810940e+00 5.70890963e-01 -9.46367756e-02
7.87991941e-01 3.62063825e-01 2.21613839e-01 4.44652081e-01
-1.04609184e-01 -6.02820635e-01 -1.84328765e-01 7.30337977e-01
-1.37289107e-01 -7.29851201e-02 -9.11366045e-01 8.01691532e-01
-1.68401551e+00 -1.22776639e+00 -7.06110299e-01 2.08227897e+00
1.01655173e+00 1.27529621e-01 -1.35717154e-01 3.69787127e-01
3.41969371e-01 -1.89173952e-01 -4.02550120e-03 -6.12117589e-01
8.55023861e-02 6.32413566e-01 -4.58172113e-02 1.07723880e+00
-8.27930331e-01 7.01171994e-01 6.17102194e+00 8.23558569e-01
-4.64771956e-01 -3.04426581e-01 3.96800250e-01 1.81282386e-01
-1.35766101e+00 1.47605255e-01 -8.73857975e-01 2.56872922e-02
7.05630481e-01 -4.13322449e-01 6.61976859e-02 8.73166025e-01
-4.01333451e-01 -8.56043249e-02 -1.48587239e+00 4.77888435e-01
2.65651166e-01 -1.59390688e+00 4.51351851e-01 -2.61403501e-01
1.18227541e+00 -8.15733731e-01 1.88578695e-01 5.34706354e-01
7.98914313e-01 -1.27387679e+00 6.59025371e-01 5.04561901e-01
3.63695711e-01 -6.94988787e-01 7.30776489e-01 3.98407221e-01
-7.92653024e-01 -3.62032920e-01 -2.84057617e-01 -6.50365293e-01
-3.61108512e-01 7.21682966e-01 -9.01955903e-01 6.02630138e-01
4.03289914e-01 3.10778916e-01 -7.77718365e-01 9.91893291e-01
-9.44760680e-01 8.00471306e-01 8.82924423e-02 -4.75277215e-01
2.85156310e-01 2.31359839e-01 3.62536341e-01 8.64563823e-01
4.17640358e-01 5.96978962e-01 -2.87300292e-02 8.37258935e-01
-1.28465548e-01 2.68690020e-01 -4.75524455e-01 3.90416175e-01
5.73432088e-01 1.07993317e+00 -6.49696350e-01 -3.72769743e-01
-3.97074938e-01 4.48982060e-01 5.29554129e-01 1.02174111e-01
-3.91288370e-01 -3.16883296e-01 5.14967918e-01 -8.71144086e-02
8.24399740e-02 3.49982530e-01 -6.26103699e-01 -1.22318375e+00
-9.25632007e-03 -1.07149994e+00 8.80740106e-01 -8.21674645e-01
-1.60104704e+00 3.99522096e-01 -2.53714994e-02 -1.02502394e+00
-4.34320718e-01 -9.24572408e-01 -8.09837162e-01 8.97603214e-01
-1.74097371e+00 -6.88241720e-01 -1.37481183e-01 5.53350270e-01
4.80623722e-01 -2.30183184e-01 9.53208327e-01 -7.98602626e-02
-1.77311078e-02 4.30520117e-01 -3.69399905e-01 -1.65516093e-01
7.32980013e-01 -1.94241083e+00 3.46955955e-02 6.94536805e-01
7.02771246e-01 9.66876209e-01 1.20461833e+00 -4.33015883e-01
-8.76907289e-01 -4.96468782e-01 1.93101013e+00 -1.11768556e+00
7.62981176e-01 4.30474252e-01 -1.10448015e+00 7.55875409e-01
2.88317084e-01 -1.95411548e-01 1.26154459e+00 6.31285071e-01
-6.27967000e-01 1.63274243e-01 -1.14390028e+00 6.70754969e-01
7.40235150e-01 -3.81437570e-01 -1.73123431e+00 5.02027571e-01
5.85616827e-01 -6.25191927e-01 -7.40508258e-01 5.47391549e-02
3.50768149e-01 -1.13167071e+00 1.03936720e+00 -1.31492472e+00
1.18738699e+00 -2.83807367e-01 -1.45390958e-01 -1.28165329e+00
-2.68463135e-01 -5.10614932e-01 -5.42314090e-02 1.11632419e+00
8.62393916e-01 -5.05341113e-01 8.61104906e-01 1.01712012e+00
-6.79680184e-02 -1.03132153e+00 -9.97037053e-01 -2.80476034e-01
3.80445272e-01 -3.17877531e-01 9.01228726e-01 1.16937912e+00
5.22203863e-01 5.56568503e-01 2.09252879e-01 1.90490633e-01
4.30317461e-01 7.31055021e-01 6.46723568e-01 -1.69952357e+00
-2.42648125e-01 -6.36943340e-01 -1.84106782e-01 -1.13013875e+00
4.05923724e-01 -1.13668931e+00 -1.49435177e-01 -1.67251134e+00
6.36955500e-02 -2.68595964e-01 -1.52522698e-01 2.66311139e-01
-7.38515139e-01 -2.39354908e-01 8.94135535e-02 -2.93124467e-01
-8.42634082e-01 1.26593977e-01 1.53401494e+00 6.44038990e-02
2.61496842e-01 1.65112928e-01 -1.25327289e+00 1.09171891e+00
5.42000413e-01 -4.35055494e-01 -6.05676651e-01 -3.55248094e-01
7.80008852e-01 2.56171733e-01 5.11150420e-01 -6.85920656e-01
4.73219007e-01 -7.02032089e-01 4.75484490e-01 -1.08643003e-01
3.42248827e-02 -7.56185830e-01 -4.93144065e-01 2.18514219e-01
-1.10800695e+00 1.13547429e-01 -1.48025667e-02 3.93602431e-01
-4.18477237e-01 -6.60863280e-01 1.40613452e-01 -2.90764481e-01
-7.28855193e-01 -2.51672924e-01 -1.94813892e-01 6.72306716e-01
7.27796316e-01 -1.21180825e-01 -5.26545107e-01 -3.93659592e-01
-4.83047813e-01 2.09068388e-01 -1.67125955e-01 3.36148351e-01
6.98608935e-01 -1.34099984e+00 -8.88444543e-01 -2.14035407e-01
2.78929114e-01 -2.04311699e-01 2.76025623e-01 3.62817913e-01
-2.88474709e-01 9.32466328e-01 3.34412120e-02 -7.69615471e-02
-1.39026439e+00 4.60468501e-01 6.47412017e-02 -6.25370800e-01
-8.19640085e-02 1.35911238e+00 -1.94117069e-01 -7.96252310e-01
-9.59600955e-02 -6.75110102e-01 -8.70759189e-01 1.87353507e-01
5.24413109e-01 3.51504743e-01 1.32503912e-01 -2.55899847e-01
-1.58545539e-01 4.70550179e-01 8.49977583e-02 6.97918534e-02
1.12903655e+00 9.20642540e-02 -1.09232493e-01 4.24067676e-01
6.70347035e-01 7.32947528e-01 -4.07764286e-01 -3.06072086e-01
3.03507805e-01 -6.63617074e-01 -5.10684788e-01 -1.02615893e+00
-3.57799888e-01 7.36668348e-01 -2.97745079e-01 6.42607629e-01
5.91616690e-01 2.63196021e-01 5.30577004e-01 8.11578453e-01
2.38513857e-01 -6.61273301e-01 2.46708110e-01 6.67064667e-01
8.77239048e-01 -1.31153786e+00 1.12204291e-01 -7.70042896e-01
-6.67375565e-01 1.43445456e+00 7.35457838e-01 6.33590594e-02
1.33910447e-01 -4.15055096e-01 8.39564130e-02 -4.49206769e-01
-1.04275382e+00 -2.53444195e-01 9.53274965e-01 3.31295013e-01
8.66453052e-01 1.53443500e-01 -5.46663165e-01 1.13465190e+00
-1.19789457e+00 -3.51561695e-01 6.72475517e-01 3.94311756e-01
-9.07269955e-01 -1.15785933e+00 -5.33736825e-01 7.35941410e-01
-4.95125562e-01 -2.54021674e-01 -6.96873248e-01 5.77441156e-01
2.93533504e-01 1.34284604e+00 -5.04500091e-01 -1.08735874e-01
3.67970109e-01 2.68913299e-01 5.23869812e-01 -9.41905737e-01
-7.69349813e-01 -1.17745984e+00 2.22559765e-01 -2.73756862e-01
-4.09263343e-01 -5.79006910e-01 -1.04326391e+00 -3.93150121e-01
-2.57207185e-01 8.67217183e-01 1.26892000e-01 1.68337977e+00
-1.18386611e-01 6.76581502e-01 2.10579649e-01 -2.40371693e-02
-7.12662876e-01 -4.18966085e-01 -2.00004116e-01 6.38468146e-01
4.57841873e-01 -8.61600280e-01 -6.27649486e-01 -1.67005658e-01] | [10.9735746383667, 7.793949127197266] |
2a03b1c7-e9f3-4b3e-be3e-f245ede206e7 | improving-neural-language-processing-with | null | null | https://aclanthology.org/2021.ranlp-main.107 | https://aclanthology.org/2021.ranlp-main.107.pdf | Improving Neural Language Processing with Named Entities | Pretraining-based neural network models have demonstrated state-of-the-art (SOTA) performances on natural language processing (NLP) tasks. The most frequently used sentence representation for neural-based NLP methods is a sequence of subwords that is different from the sentence representation of non-neural methods that are created using basic NLP technologies, such as part-of-speech (POS) tagging, named entity (NE) recognition, and parsing. Most neural-based NLP models receive only vectors encoded from a sequence of subwords obtained from an input text. However, basic NLP information, such as POS tags, NEs, parsing results, etc, cannot be obtained explicitly from only the large unlabeled text used in pretraining-based models. This paper explores use of NEs on two Japanese tasks; document classification and headline generation using Transformer-based models, to reveal the effectiveness of basic NLP information. The experimental results with eight basic NEs and approximately 200 extended NEs show that NEs improve accuracy although a large pretraining-based model trained using 70 GB text data was used. | ['Tomoya Iwakura', 'Takuya Makino', 'Kyoumoto Matsushita'] | null | null | https://aclanthology.org/2021.ranlp-1.107 | https://aclanthology.org/2021.ranlp-1.107.pdf | ranlp-2021-9 | ['headline-generation'] | ['natural-language-processing'] | [ 3.56132895e-01 2.28474662e-01 -1.02103010e-01 -7.35857606e-01
-7.75350630e-01 -5.32849371e-01 2.68927842e-01 1.67772695e-01
-8.73313367e-01 1.23700154e+00 5.24732172e-01 -6.33493662e-01
3.72127295e-01 -9.29262042e-01 -8.71685266e-01 -3.36095989e-01
1.46102637e-01 6.47616506e-01 6.86617121e-02 -3.55708271e-01
2.86668658e-01 5.81273258e-01 -1.31676269e+00 7.22023308e-01
7.58215547e-01 8.79312038e-01 4.13905680e-01 7.05128670e-01
-1.15709126e+00 8.61153543e-01 -1.08790433e+00 -4.44211245e-01
-1.02522708e-02 -3.91564459e-01 -7.35137999e-01 -4.23624516e-01
3.12519044e-01 -1.12431705e-01 -2.29569614e-01 9.92460489e-01
4.94886130e-01 2.82577902e-01 6.74897015e-01 -6.36907399e-01
-1.06878567e+00 1.09781814e+00 -9.08908993e-02 2.87205696e-01
1.86728612e-01 -2.91006178e-01 9.58819091e-01 -9.52331126e-01
7.90741682e-01 1.07071471e+00 7.75271177e-01 9.39942181e-01
-8.18589509e-01 -5.02851605e-01 -1.63728759e-01 -4.66693416e-02
-9.97332036e-01 -3.70232999e-01 6.36324883e-01 -4.57869135e-02
1.74056923e+00 2.12196149e-02 3.74110252e-01 1.09998763e+00
3.89234513e-01 1.05997896e+00 6.87159717e-01 -1.01302290e+00
1.95417076e-01 2.59623051e-01 7.82924354e-01 7.62229562e-01
-1.84628516e-02 -5.08738915e-03 -4.82993335e-01 -5.85093163e-02
5.91195047e-01 -4.48479056e-01 -1.15563072e-01 2.11298928e-01
-1.16743231e+00 1.00209391e+00 2.04819128e-01 7.09523022e-01
-6.20605052e-01 -1.08107574e-01 7.44676054e-01 1.63330197e-01
4.17144507e-01 6.26332462e-01 -1.19249606e+00 -4.19491261e-01
-9.60269332e-01 -1.56593949e-01 1.12791848e+00 1.20476365e+00
5.18755794e-01 6.60535574e-01 -4.53333855e-01 1.13700247e+00
-1.68689582e-02 6.29672647e-01 1.09185183e+00 -4.95697647e-01
1.00821245e+00 4.14781898e-01 -1.93115100e-01 -8.68308663e-01
-3.05236131e-01 -2.74493545e-01 -9.22013700e-01 -4.39126790e-01
1.90258801e-01 -7.98374116e-01 -1.35781419e+00 1.61695516e+00
-1.58483699e-01 -4.68681544e-01 6.50152326e-01 2.00961009e-01
1.08741915e+00 1.57911968e+00 6.39095083e-02 -1.85944572e-01
1.38411343e+00 -1.19154203e+00 -8.16785038e-01 -3.36755663e-01
9.53211546e-01 -4.30756032e-01 7.39549637e-01 1.28238633e-01
-9.79080915e-01 -6.64055347e-01 -8.25891316e-01 -1.13909714e-01
-8.90237331e-01 7.16348052e-01 5.61959207e-01 8.95907402e-01
-1.22667658e+00 5.04225492e-01 -4.14438605e-01 -3.24321985e-01
1.80849120e-01 3.57925862e-01 -4.81674045e-01 -1.09790049e-01
-1.58877015e+00 8.83024395e-01 1.02023852e+00 2.98047900e-01
-4.76267695e-01 -4.89121795e-01 -1.02803183e+00 3.88008147e-01
4.04509194e-02 -4.95613694e-01 1.33690345e+00 -9.90965426e-01
-1.61786067e+00 4.85264480e-01 -3.36499184e-01 -7.69269228e-01
-3.30370575e-01 -1.98216930e-01 -5.98890364e-01 2.31670648e-01
-1.19201094e-01 1.06051719e+00 5.58029175e-01 -8.05395722e-01
-5.85149407e-01 -8.03989545e-02 -3.49599302e-01 7.24190623e-02
-5.34379542e-01 6.02768898e-01 -8.03660005e-02 -6.01929605e-01
-2.56724387e-01 -4.75457162e-01 -3.85427296e-01 -6.42584741e-01
-6.98856950e-01 -5.25774419e-01 3.87160361e-01 -1.07832110e+00
1.11446226e+00 -1.87492919e+00 -3.97974312e-01 -7.05753714e-02
-2.87953138e-01 6.65565312e-01 -5.55396855e-01 5.02951205e-01
-1.48774877e-01 5.09047866e-01 -9.70696136e-02 -2.14314729e-01
1.50389150e-01 4.41098601e-01 -4.18298423e-01 -4.95159149e-01
4.42946762e-01 9.14760530e-01 -5.52849114e-01 -6.35580540e-01
-1.22573458e-01 3.35501194e-01 -1.51061445e-01 1.63984597e-01
-2.86221653e-01 -1.79124743e-01 -4.13637727e-01 4.12249267e-01
3.40501040e-01 2.01702081e-02 1.11009881e-01 3.39702517e-02
-4.96484041e-02 8.15380752e-01 -6.43302858e-01 1.46655738e+00
-4.88843590e-01 8.84604514e-01 -3.16597939e-01 -1.08669341e+00
1.26251161e+00 6.93309546e-01 -3.06869559e-02 -5.38388371e-01
2.19678923e-01 2.07219154e-01 -2.39607431e-02 -4.43380833e-01
1.01703119e+00 -1.86248019e-01 -2.96288908e-01 3.40840757e-01
6.10975325e-01 1.29307747e-01 6.71027124e-01 9.53794271e-02
1.19799721e+00 -1.18201524e-01 4.08355653e-01 1.10483579e-02
5.07429063e-01 3.73193890e-01 6.58874989e-01 9.96683359e-01
7.64059946e-02 5.87323070e-01 3.56280029e-01 -4.69316274e-01
-1.36382449e+00 -6.44972801e-01 -1.85105816e-01 1.11096251e+00
-8.01949859e-01 -2.57451832e-01 -9.56401050e-01 -9.93838608e-01
-4.65656668e-01 1.15675890e+00 -4.57214892e-01 2.13955909e-01
-8.79271030e-01 -9.30229545e-01 1.09767842e+00 8.35304260e-01
4.55987126e-01 -1.80330193e+00 -9.35510546e-02 7.28767335e-01
-2.16710180e-01 -1.16448176e+00 -2.70345211e-01 6.72647834e-01
-1.09408236e+00 -4.02408004e-01 -9.85989928e-01 -1.26493454e+00
8.33294570e-01 -3.44002694e-01 1.13699543e+00 -2.58215308e-01
1.31759122e-01 -3.73554125e-04 -6.58860326e-01 -5.20647585e-01
-7.96769619e-01 3.16199362e-01 -4.75205630e-02 -3.84221464e-01
6.85954750e-01 -2.40934908e-01 2.92347312e-01 -2.98131377e-01
-7.53347516e-01 1.52834713e-01 1.05415010e+00 1.16596913e+00
3.91142070e-01 1.96270511e-01 7.75816083e-01 -1.01573896e+00
9.57096279e-01 -1.88503474e-01 -1.81720436e-01 5.33027351e-01
-2.31553629e-01 3.99355650e-01 9.36784208e-01 -5.57094872e-01
-1.50013745e+00 -6.63190559e-02 -8.16762030e-01 1.81845203e-01
-7.45085478e-01 8.80704463e-01 -2.20245406e-01 3.48147154e-01
5.73805869e-01 8.56399119e-01 -2.83132970e-01 -6.98940575e-01
3.91689122e-01 9.66044247e-01 3.65429431e-01 -6.91518068e-01
2.40392834e-01 -4.32625294e-01 -4.06365961e-01 -9.58146334e-01
-9.61264074e-01 -2.36005858e-01 -6.98657036e-01 2.42754832e-01
8.56766045e-01 -5.04632235e-01 -1.83617011e-01 4.75110352e-01
-1.87310553e+00 -5.89495264e-02 -5.85750759e-01 6.19719565e-01
-7.82543570e-02 2.77901947e-01 -1.20748448e+00 -6.63713038e-01
-9.12871897e-01 -6.65323913e-01 6.73132896e-01 3.20220679e-01
-5.95746599e-02 -1.02156985e+00 1.86971232e-01 2.24122703e-01
5.06677866e-01 -4.70880181e-01 1.43004322e+00 -1.45583367e+00
4.06488627e-02 -6.02454603e-01 -2.04511866e-01 9.61926043e-01
-1.64665863e-01 -1.44815654e-01 -8.01370680e-01 1.75401658e-01
5.78710698e-02 -4.49237317e-01 7.25152075e-01 4.19422388e-01
8.41776967e-01 -6.66002452e-01 -2.89769828e-01 8.83292183e-02
1.40591240e+00 6.50804400e-01 7.23542690e-01 1.33783653e-01
5.71803272e-01 8.77041221e-01 2.12919652e-01 5.34413829e-02
1.49109066e-01 -3.02903764e-02 -1.25130489e-01 -2.34933142e-02
-1.33438250e-02 -3.44914407e-01 5.85595906e-01 1.54261029e+00
3.24008346e-01 -8.81955504e-01 -1.19031179e+00 4.95869160e-01
-1.35579205e+00 -7.81921804e-01 -3.35987285e-02 1.71558940e+00
1.16447842e+00 3.36513698e-01 -5.45469940e-01 -8.42479407e-04
9.99742806e-01 3.57271060e-02 -2.24969685e-01 -7.93914914e-01
-2.49153256e-01 4.60489005e-01 4.08851773e-01 1.35545254e-01
-8.71566892e-01 1.10742474e+00 6.64628553e+00 1.02189684e+00
-1.02468705e+00 1.15908636e-02 6.26179576e-01 3.60633641e-01
-1.68827757e-01 -3.40099841e-01 -1.69734704e+00 3.83913010e-01
1.63200557e+00 -5.72726838e-02 -1.08348571e-01 1.04750979e+00
-6.05420247e-02 1.31060630e-01 -1.00301814e+00 6.72065258e-01
5.43191969e-01 -1.63403428e+00 6.42821312e-01 -2.18488365e-01
5.37206411e-01 3.07757467e-01 -4.40259218e-01 8.62287939e-01
5.07123411e-01 -7.12591290e-01 5.28654158e-01 3.29007119e-01
5.94455838e-01 -6.47450507e-01 1.25576091e+00 5.49269378e-01
-8.33260536e-01 6.97443858e-02 -8.24161828e-01 9.35413912e-02
5.46375096e-01 7.89698839e-01 -1.07322371e+00 5.14344096e-01
3.89926016e-01 4.12867785e-01 -4.77086484e-01 9.08100307e-01
-4.35495704e-01 1.00845885e+00 -3.39042455e-01 -9.84548628e-01
5.31977355e-01 1.63744822e-01 4.24441129e-01 1.63332808e+00
4.50102806e-01 8.96611288e-02 -1.80855677e-01 7.27780044e-01
-2.98546582e-01 4.82207388e-01 -6.54356658e-01 -5.51231027e-01
4.10690904e-01 1.06429458e+00 -6.30419970e-01 -8.18552852e-01
-3.63820046e-01 7.26336598e-01 5.14353037e-01 3.17896396e-01
-4.49818939e-01 -9.46511805e-01 -2.64346935e-02 -3.60599816e-01
6.34236813e-01 -3.00110132e-01 -1.90741017e-01 -1.26527941e+00
4.15496938e-02 -6.75047517e-01 4.77797948e-02 -1.15562201e+00
-1.45857227e+00 1.22725344e+00 -1.49121419e-01 -7.22885251e-01
-5.76758921e-01 -1.08078969e+00 -4.78494734e-01 1.07995009e+00
-1.20932221e+00 -8.90097857e-01 5.21449685e-01 1.54224038e-01
7.94988275e-01 -5.81260502e-01 1.28292239e+00 7.60549456e-02
-6.88470125e-01 3.57772976e-01 3.73281449e-01 9.54842389e-01
1.94268242e-01 -1.18254662e+00 6.35341108e-01 9.62312639e-01
4.63132709e-01 6.91353440e-01 3.14062536e-01 -6.48362875e-01
-1.01475346e+00 -9.41485167e-01 1.79105771e+00 -1.83840841e-01
5.41637957e-01 -3.66586655e-01 -8.38320374e-01 6.95574224e-01
4.23173308e-01 -3.27975810e-01 9.21348512e-01 1.55366331e-01
-3.18363495e-02 -2.17869040e-02 -1.03080714e+00 4.75347161e-01
6.34458125e-01 -4.23398197e-01 -1.32875097e+00 2.41951257e-01
1.13364530e+00 -1.65629134e-01 -5.97513437e-01 2.83302575e-01
2.11770579e-01 -3.38455468e-01 6.31256044e-01 -1.03388894e+00
5.24893820e-01 1.92876503e-01 -1.67932689e-01 -1.35367954e+00
-2.37554848e-01 -5.27735129e-02 1.07821904e-01 1.56427264e+00
1.08602214e+00 -6.22028112e-01 9.65150833e-01 6.70357287e-01
-4.14700836e-01 -6.70256734e-01 -7.79901743e-01 -9.11408961e-01
2.71549970e-01 -5.48051119e-01 5.60666561e-01 8.42556179e-01
3.33158895e-02 9.49765861e-01 -1.17810518e-01 -1.84033081e-01
1.60379812e-01 -3.04034203e-01 2.44448468e-01 -1.12802482e+00
-1.73624530e-01 -2.03553170e-01 -1.26101986e-01 -1.39697874e+00
5.61065495e-01 -8.91858697e-01 6.25918210e-01 -1.78699803e+00
1.89182721e-02 -3.70073885e-01 -2.12000236e-01 1.00317121e+00
1.52325734e-01 -2.75519282e-01 1.64961055e-01 -5.22506721e-02
-2.45076641e-01 6.17383003e-01 8.92716169e-01 -2.72622794e-01
-1.93955272e-01 -2.56934911e-01 -2.54357189e-01 6.22699201e-01
1.03872299e+00 -1.01145518e+00 -1.63788736e-01 -6.80019617e-01
3.03641289e-01 4.01399791e-01 -3.31428260e-01 -9.51503217e-01
3.42848003e-01 5.86312786e-02 5.16767800e-01 -1.00642276e+00
3.15887868e-01 -5.82997382e-01 -4.76485133e-01 3.42963964e-01
-6.30866647e-01 2.22676635e-01 3.28191042e-01 2.98260331e-01
-4.75433886e-01 -1.16373610e+00 2.26199627e-01 -5.23147821e-01
-8.55246902e-01 -2.85138264e-02 -9.73453462e-01 5.50377145e-02
4.50294971e-01 -1.75728634e-01 -3.84331018e-01 -7.40261301e-02
-6.54843450e-01 -2.26692095e-01 -3.30555618e-01 2.96956420e-01
8.07059824e-01 -1.07090509e+00 -7.38262296e-01 2.12382838e-01
-1.72265753e-01 -1.50841802e-01 4.22836170e-02 9.78173688e-02
-7.25820839e-01 1.09668493e+00 -3.54284853e-01 -5.09864800e-02
-1.11674225e+00 5.13512373e-01 5.45651652e-02 -7.39281893e-01
-5.47630966e-01 9.28884089e-01 -8.73470977e-02 -1.00837004e+00
2.56307572e-01 -4.85485673e-01 -6.72782898e-01 -1.50810793e-01
7.12855160e-01 -6.98721185e-02 1.54902592e-01 -4.78147089e-01
-9.00103822e-02 -2.52847802e-02 -3.86563003e-01 -3.60505462e-01
1.31506538e+00 2.17868581e-01 -1.76307812e-01 4.91875529e-01
1.18665385e+00 -1.56321377e-01 -5.21035016e-01 -2.78418958e-01
2.28549272e-01 2.78403312e-01 -7.87467211e-02 -1.05409420e+00
-8.74730408e-01 1.18838894e+00 3.15236785e-02 9.96501148e-02
8.45546007e-01 -2.69414663e-01 1.30144477e+00 1.17110026e+00
5.09241283e-01 -1.21149254e+00 -1.58986688e-01 1.21408451e+00
7.64252901e-01 -8.81277919e-01 -6.41304076e-01 -2.17767373e-01
-5.26978433e-01 1.39865935e+00 5.88696182e-01 2.07712159e-01
6.02443874e-01 5.54351807e-01 1.11254893e-01 1.40999019e-01
-8.75127435e-01 2.01171517e-01 1.07992187e-01 6.90225661e-01
4.82971549e-01 -1.11423701e-01 -3.20734054e-01 8.82534385e-01
-4.16813940e-01 5.09363376e-02 6.20438039e-01 1.19043767e+00
-5.70963800e-01 -1.17332911e+00 -2.04311222e-01 8.07083249e-01
-6.99465394e-01 -8.71520579e-01 -2.67043680e-01 5.52903414e-01
-3.78095768e-02 7.23374784e-01 2.89367616e-01 -2.46680275e-01
1.11781619e-01 6.36480272e-01 1.66425452e-01 -1.10858142e+00
-8.53402555e-01 -5.38420856e-01 7.23858654e-01 1.10231914e-01
-3.55808288e-01 -3.58175367e-01 -1.34673858e+00 -6.97490796e-02
-4.95882601e-01 7.12750018e-01 9.99632597e-01 1.00225484e+00
3.00389200e-01 4.48474616e-01 3.24696541e-01 -6.75778687e-01
-5.73185086e-01 -1.45323884e+00 -6.15364194e-01 -6.62251655e-03
-9.92765352e-02 9.28965285e-02 -2.24642843e-01 2.71286428e-01] | [10.023185729980469, 9.749314308166504] |
0878bf52-78ca-46ee-ae97-2f5302d1e2e6 | survival-kernets-scalable-and-interpretable | 2206.10477 | null | https://arxiv.org/abs/2206.10477v4 | https://arxiv.org/pdf/2206.10477v4.pdf | Survival Kernets: Scalable and Interpretable Deep Kernel Survival Analysis with an Accuracy Guarantee | Kernel survival analysis models estimate individual survival distributions with the help of a kernel function, which measures the similarity between any two data points. Such a kernel function can be learned using deep kernel survival models. In this paper, we present a new deep kernel survival model called a survival kernet, which scales to large datasets in a manner that is amenable to model interpretation and also theoretical analysis. Specifically, the training data are partitioned into clusters based on a recently developed training set compression scheme for classification and regression called kernel netting that we extend to the survival analysis setting. At test time, each data point is represented as a weighted combination of these clusters, and each such cluster can be visualized. For a special case of survival kernets, we establish a finite-sample error bound on predicted survival distributions that is, up to a log factor, optimal. Whereas scalability at test time is achieved using the aforementioned kernel netting compression strategy, scalability during training is achieved by a warm-start procedure based on tree ensembles such as XGBoost and a heuristic approach to accelerating neural architecture search. On four standard survival analysis datasets of varying sizes (up to roughly 3 million data points), we show that survival kernets are highly competitive compared to various baselines tested in terms of time-dependent concordance index. Our code is available at: https://github.com/georgehc/survival-kernets | ['George H. Chen'] | 2022-06-21 | null | null | null | null | ['survival-analysis'] | ['miscellaneous'] | [-0.2860886 -0.12880889 -0.40515488 -0.5449456 -1.1241026 -0.43475142
0.24224183 0.81410974 -0.42613348 0.55169255 0.2723748 -0.60261464
-0.6462781 -0.8172001 -0.51580685 -0.95438206 -0.5162084 0.7996524
-0.10707052 0.14708133 -0.16093428 0.43908268 -1.064891 0.19475313
0.7122879 1.0646583 -0.40857032 0.9681229 0.22125198 0.5828364
-0.4156426 -0.40930763 0.03838519 -0.25811118 -0.7615049 -0.41756618
0.17068838 -0.31191215 -0.4706515 0.5117513 0.48117656 0.10258949
0.96121216 -1.2820605 -0.3098719 0.5384704 -0.27065256 0.1642534
-0.23757935 -0.14451648 0.7828023 -0.59039706 0.19561282 0.7167795
1.2188596 0.31285468 -1.7073429 -0.2887897 -0.6849641 0.1753306
-1.5216058 -0.26676735 0.34975633 -0.48538435 0.7509596 0.66988236
0.79160047 0.8456236 0.6035402 0.57763594 0.75936794 -0.46758968
0.745654 0.0828927 0.6611108 0.70525676 0.10130349 0.44630006
-0.32051367 -0.7982627 0.47496986 0.31648952 -0.23653916 -0.5196544
-0.8897777 1.2546537 0.5953567 0.02345782 -0.2761818 0.5064675
0.83359474 0.35872817 0.7292785 -0.11885704 -0.5499259 0.04059348
-1.2139453 0.14436059 0.76214164 0.40660116 0.38124245 -0.31778553
-0.33280894 0.802152 -0.02227666 0.03611184 0.7783607 -0.6894833
-0.152448 0.6036986 -0.22007823 -0.6374936 -0.9110707 -0.8209761
-1.2745906 0.17288335 0.6074403 0.11359842 -0.8433979 1.7579619
0.5639166 0.17321734 0.17271164 0.6300804 0.5267426 0.31076208
0.23242675 -0.18025559 1.5825078 -0.6512731 -0.09103397 0.15195073
1.3822119 -0.14073178 1.0030853 0.25076634 -0.7164799 -0.05493636
-0.9143901 -0.02190926 -0.29383552 0.23157658 0.9722922 0.7465892
-1.344527 0.8118044 -1.1775576 -0.59848195 0.5571151 0.4195673
-0.62534016 -0.08922692 -1.0369059 0.61003697 0.6503331 -0.18342979
-0.64304733 -0.9146234 -0.6888808 0.3316597 -0.17947684 -1.2057904
1.0406437 -0.7754075 -1.0187255 0.7730529 0.08568987 -0.8660541
0.5296801 0.07200371 -0.11177539 0.09215142 -0.12644216 0.23009351
0.4257855 -0.59645915 -0.56073475 -0.6059324 -0.48045594 -0.10097758
-0.7185815 0.03955942 -0.3876087 -0.7317891 -0.13562322 -0.9919925
-0.45329946 0.00698001 -0.36170506 -0.30920735 0.29411682 -0.8567316
1.4046214 -2.1795466 0.16218464 0.4893944 0.5395259 -0.4035228
0.1854055 0.41270247 -0.34167323 -0.10131874 -0.48085314 -0.58779776
-0.12369441 0.01099238 -0.0970782 0.7674103 -0.463043 0.8989543
-0.48774573 -0.51379406 0.00630198 0.52382517 -0.45982414 0.17103355
0.25308275 -0.03080259 -0.13272263 0.7331899 0.41571262 -0.6293264
0.09324276 0.12499287 0.40538013 -0.2740658 -0.4982737 1.5302777
-0.35890886 0.42861542 -0.4609824 -1.1767087 0.7834086 0.24001409
0.6298385 -0.02394579 0.23626344 0.28832376 -0.32152104 0.15549012
0.15087605 -0.3324112 -0.27699712 0.41010615 0.13712303 0.61195874
0.01137987 0.39830062 1.6147232 -0.4664633 0.6719838 -0.44346666
0.07622041 0.2647402 0.3057735 0.8504653 -0.01209951 0.38044673
0.6758108 -0.7747953 -0.9498028 -1.2381186 -0.77275884 1.0333657
-0.4147348 -0.7122615 -0.73712945 -0.83865666 0.35839045 0.8416257
-1.1807253 -0.52486414 -0.14543892 -1.2805969 0.8111018 0.8271388
-0.01178896 -0.673804 -0.7700465 0.16930684 0.14417496 -0.44427687
-0.31020546 0.7644389 -1.3086222 -1.1039712 -0.77508396 -0.49644688
0.65476555 -0.15608034 1.1035981 0.2560033 -0.5008778 0.19112702
-0.33359927 -0.09369814 -0.519024 0.14829035 0.09059598 -0.21779725
0.43489215 -0.5493208 -0.99307954 0.11960297 -0.93687636 0.12378044
0.65022355 1.153587 0.60852796 0.15787384 0.54004145 -0.6417174
0.3429194 -0.9417786 -0.429346 0.32161042 -0.8735066 0.04498276
0.8147764 -0.3278127 -0.33621725 0.11915188 0.17280015 -0.54483896
-0.19605085 0.9087969 0.38162836 -0.00696988 0.9197459 0.32469016
0.26857778 -0.4394905 0.16992071 0.6110816 0.50823176 -0.39600047
0.44009706 0.5273205 0.3095288 -0.34538925 -0.40525296 -0.4443919
-0.4151401 0.01427402 0.5629661 -0.61481273 -0.7465501 0.37327784
-0.49527156 -0.5536242 -0.35155237 0.56223124 -0.68919927 0.25974068
-0.8778204 -0.522342 -0.8077745 -0.6912359 0.9329416 -0.11774292
-0.29327172 -1.3816749 0.31022602 0.08729792 0.39085138 0.49314576
1.337011 -1.044464 0.07747292 -0.8846417 -0.05103761 0.12541968
-0.04909398 -0.2616509 -0.58670497 -0.69514847 -0.21874242 -0.40534717
1.1576347 0.79241276 1.4881968 -0.3679216 -0.77700955 0.99786896
1.46647 -0.05799498 0.45268947 0.45998356 0.48082015 0.5809099
0.1396731 0.4374996 0.37383524 0.8290611 0.07370094 -0.18006025
0.26046622 -0.04275258 0.01730354 0.3277936 0.11472178 -0.10350131
-1.1579939 0.4911788 -1.9835463 -0.57960933 -0.0863595 2.6822965
0.886186 -0.05600225 0.36791453 0.17694728 0.43159088 -0.44916856
-0.5901156 -0.19131151 -0.06033772 0.13616592 0.6701644 0.20448163
-1.1543162 0.5645133 6.0906444 1.1771258 -0.9733169 0.13797987
1.1786251 -0.46925342 0.13917096 0.19722311 -0.48316178 0.541135
1.4857941 -0.47605094 0.11613299 1.028247 -0.14546874 -0.13134103
-1.1780038 0.88903725 -0.23324569 -1.4017347 -0.3060321 0.38289186
0.1176753 -0.10341076 0.08481631 0.4163954 0.5931086 -1.2225171
0.41575927 0.43305358 1.0421892 -1.1008631 0.8582887 0.3048133
-0.88987744 -0.33291525 -0.4054426 0.4063494 -0.05925305 0.8920731
-0.9567899 0.6286006 0.96147645 0.2635476 -0.8655889 1.2219411
0.40814835 1.0841482 -0.5902142 0.25811875 -0.12984735 0.10216691
0.14188288 1.2041554 0.6903719 0.06191568 -0.10613802 0.39320123
0.19548607 0.35080254 -0.3019914 0.39106596 0.41437125 1.3034573
-0.8935718 -0.43863097 -0.14819309 0.97436494 0.76813966 0.1249238
-0.78763276 -0.28403148 0.38483527 0.21639465 0.15089078 0.19849838
-0.5288619 -0.97922426 -0.27367628 -0.4171979 1.1305944 -0.4069776
-1.4135467 0.5909051 0.15586475 -1.0983593 -0.4314495 -0.3777873
-0.62709963 0.6594693 -0.7129055 -1.1021037 -0.30427006 0.56751657
0.01858853 -0.19885276 1.2791628 -0.0622871 -0.6636281 1.072933
0.73308766 0.08132404 0.64826554 -1.3998907 0.29549307 0.28093404
-0.02267784 0.6210888 0.65033704 -0.5665047 -1.0894096 -1.2297461
0.6045057 -0.44321197 0.5136107 -0.35480028 -1.1396567 0.6466001
-0.20950112 0.26664218 1.4008478 0.55772084 -0.32967037 -0.19292901
-1.2189395 0.35701704 0.61018354 -0.07714834 -0.07380667 0.49011946
0.34018525 -0.28417638 -1.2681409 0.47901213 0.652584 -1.069084
0.95236593 -0.9053139 0.40114427 -0.07453249 -0.1304493 -1.5076149
-0.6096302 -0.1836419 -0.4597919 0.71368504 0.27018878 -0.7389558
1.0272903 0.7516485 0.04278541 -1.2382694 -1.3394525 -0.85567015
0.3533656 -0.3768335 0.49266356 0.92606133 0.21747297 0.00860865
-0.28808153 0.04786572 0.9964309 0.08487589 0.5827247 -1.2998365
-0.6532961 -0.4811917 -0.6922962 -0.38767907 0.13484138 -1.1746416
-0.36860758 -1.2837414 0.57930493 -0.58142567 -0.6905606 0.6940211
-0.22360626 0.24558707 -0.31111845 0.50151885 -0.31761944 0.46154043
0.28841114 0.22791296 -0.2219666 0.09227424 -0.6226102 0.4243363
1.0030209 -0.62805676 -0.13162561 0.08597223 -0.05610541 0.75261134
0.79056644 -1.107346 0.44147274 0.10826028 0.8289868 -0.3864745
0.11110537 -0.6037153 0.5489229 0.68448496 -0.47637543 0.05812406
0.22370112 0.8145468 0.02120123 0.11867107 0.7740939 0.34596342
-0.2770201 0.36927488 -0.25573385 -0.41281295 1.1863315 -0.2996308
-0.31485438 -0.28277668 -0.8290245 0.21248789 0.5274736 -0.05868751
0.63439876 -1.5179149 -0.8976329 0.05351599 0.4160786 -0.13908093
0.60404664 1.2913849 -0.5080456 0.34981602 0.06347766 -0.464627
-1.3935635 0.66760874 0.52506405 -0.48625025 -0.86528623 0.8265841
0.45154122 -0.32459876 0.15510662 0.02014894 0.09657886 -0.28380802
0.6724555 0.44996205 0.19950233 -0.23811378 -0.33015746 0.12472664
-0.27903023 0.0722448 1.1700715 0.31862304 -0.21034281 0.36284575
1.45023 -0.4424368 -0.9839746 -0.15332662 0.07182764 -0.33363417
0.3034109 -1.0061905 -0.9865477 0.35364786 0.908494 0.1803503
1.3920203 0.37050596 0.7158028 0.20302501 0.09412593 -0.51212233
-0.36617762 0.06452478 0.6416807 -1.1039909 -0.09657574 0.09942064
-0.33783236 1.1794956 0.20378222 -0.06204715 0.7513941 0.38424283
-0.08275984 -0.19194892 -1.0611185 0.24581398 0.19166687 0.31616753
0.32624626 0.38658428 -0.31864837 1.0136139 -0.43199527 0.14122498
0.06036624 0.5257737 -0.40082026 -0.91364974 -0.36548233 0.9980397
-0.44702727 -0.2612257 -0.14519371 0.70971453 -0.5384657 0.5703784
0.24509652 -0.56917304 -0.0359961 0.19563372 0.09552159 -0.08350442
-0.46080947 -0.19113186 -0.12959832 -0.4084082 0.3843062 -0.6434311
-1.1772906 -0.67982817 -0.23002021 0.4587529 0.48227942 0.64491975
0.34116253 0.2835286 0.49101436 -0.4157123 -0.85406804 -0.9641675
-0.8835358 0.13319021 0.26294944 -0.46810135 -0.6447941 -0.14582533] | [7.792775630950928, 5.622522830963135] |
917f9e02-e3e7-4228-93fa-ad097f8efc90 | homogeneous-network-embedding-for-massive | 1906.06826 | null | https://arxiv.org/abs/1906.06826v6 | https://arxiv.org/pdf/1906.06826v6.pdf | Homogeneous Network Embedding for Massive Graphs via Reweighted Personalized PageRank | Given an input graph G and a node v in G, homogeneous network embedding (HNE) maps the graph structure in the vicinity of v to a compact, fixed-dimensional feature vector. This paper focuses on HNE for massive graphs, e.g., with billions of edges. On this scale, most existing approaches fail, as they incur either prohibitively high costs, or severely compromised result utility. Our proposed solution, called Node-Reweighted PageRank (NRP), is based on a classic idea of deriving embedding vectors from pairwise personalized PageRank (PPR) values. Our contributions are twofold: first, we design a simple and efficient baseline HNE method based on PPR that is capable of handling billion-edge graphs on commodity hardware; second and more importantly, we identify an inherent drawback of vanilla PPR, and address it in our main proposal NRP. Specifically, PPR was designed for a very different purpose, i.e., ranking nodes in G based on their relative importance from a source node's perspective. In contrast, HNE aims to build node embeddings considering the whole graph. Consequently, node embeddings derived directly from PPR are of suboptimal utility. The proposed NRP approach overcomes the above deficiency through an effective and efficient node reweighting algorithm, which augments PPR values with node degree information, and iteratively adjusts embedding vectors accordingly. Overall, NRP takes O(mlogn) time and O(m) space to compute all node embeddings for a graph with m edges and n nodes. Our extensive experiments that compare NRP against 18 existing solutions over 7 real graphs demonstrate that NRP achieves higher result utility than all the solutions for link prediction, graph reconstruction and node classification, while being up to orders of magnitude faster. In particular, on a billion-edge Twitter graph, NRP terminates within 4 hours, using a single CPU core. | ['Jieming Shi', 'Xiaokui Xiao', 'Renchi Yang', 'Yin Yang', 'Sourav S. Bhowmick'] | 2019-06-17 | null | null | null | null | ['graph-reconstruction'] | ['graphs'] | [-1.46302879e-01 3.16043496e-01 -4.27032232e-01 1.37709612e-02
-3.00480515e-01 -4.43091959e-01 3.57555777e-01 6.26391411e-01
-3.69572163e-01 2.14787886e-01 1.27992526e-01 -6.66066647e-01
-3.45351666e-01 -1.28596079e+00 -3.65659207e-01 -4.60947514e-01
-6.37069762e-01 4.08931673e-01 3.45121503e-01 -2.80903220e-01
9.64361951e-02 5.91173828e-01 -1.19793069e+00 -5.92500329e-01
5.19619465e-01 8.06971014e-01 -3.22128870e-02 9.35312510e-01
-1.86592266e-01 3.84377092e-01 -1.75066933e-01 -7.33435333e-01
3.51287276e-01 1.53013572e-01 -8.00517380e-01 -3.07403933e-02
2.46622011e-01 -3.38083744e-01 -1.00788224e+00 9.23678935e-01
4.18931186e-01 -3.51282395e-02 5.02152264e-01 -1.68623638e+00
-6.33928299e-01 8.77096236e-01 -7.09135950e-01 4.02520411e-02
2.85513639e-01 -1.38531908e-01 1.70933998e+00 -9.16535139e-01
4.47345436e-01 9.74540949e-01 9.03283894e-01 2.98127681e-01
-1.31907654e+00 -3.82844269e-01 1.29004866e-01 4.24911641e-02
-1.56466341e+00 -5.92451878e-02 8.33082736e-01 -2.23378107e-01
7.85418510e-01 7.00992644e-01 7.72571802e-01 6.34672821e-01
1.30585935e-02 5.39938569e-01 6.26818061e-01 -1.06479831e-01
1.20426819e-01 1.22199096e-01 5.22701919e-01 8.85736823e-01
7.86471367e-01 -2.05313161e-01 -2.62596816e-01 -6.38241887e-01
5.30759931e-01 2.60128647e-01 -3.97261500e-01 -5.13198912e-01
-1.18837559e+00 9.50977087e-01 8.65810096e-01 2.18552977e-01
-3.18542749e-01 4.02839214e-01 4.68228191e-01 3.38591933e-01
3.70462507e-01 2.65071422e-01 -4.18822527e-01 7.98202008e-02
-6.38955891e-01 -7.17841759e-02 9.97357965e-01 8.62697244e-01
1.10172546e+00 -2.35992506e-01 4.57191579e-02 6.94503129e-01
5.70093811e-01 2.94432104e-01 2.02889085e-01 -5.77550352e-01
4.70268220e-01 8.13071966e-01 -1.42082393e-01 -1.80364895e+00
-4.39790726e-01 -4.21319127e-01 -1.10579896e+00 -2.07466796e-01
-1.88930947e-02 3.70550044e-02 -6.21982157e-01 1.60380697e+00
5.76554716e-01 2.56233066e-01 -3.14393789e-02 7.94620812e-01
9.60559011e-01 7.78103948e-01 -1.07730724e-01 6.43473715e-02
1.42349887e+00 -1.11523736e+00 -2.98414588e-01 -1.87262535e-01
7.67957568e-01 -3.85796845e-01 1.03396928e+00 -1.50464654e-01
-7.03177452e-01 -5.62657490e-02 -1.12280524e+00 -3.73234004e-02
-4.15405840e-01 -1.65602461e-01 9.74958777e-01 7.78712630e-01
-1.55481493e+00 6.86165571e-01 -5.48208892e-01 -5.02420485e-01
1.19752772e-01 5.53695798e-01 -6.53504789e-01 -1.23020150e-01
-1.19838023e+00 3.46029907e-01 2.67856121e-01 7.68540651e-02
-2.25403339e-01 -8.52810979e-01 -1.02031755e+00 5.01310110e-01
3.78375471e-01 -4.91215944e-01 6.69676602e-01 -4.28188592e-01
-1.20652843e+00 4.51527953e-01 -7.98182413e-02 -3.79162520e-01
1.21859893e-01 1.98181346e-01 -5.15731514e-01 2.14642391e-01
-1.53607622e-01 4.00579095e-01 5.71624279e-01 -1.26440883e+00
-2.37039223e-01 -2.10869923e-01 2.54321188e-01 5.97706670e-03
-1.22804832e+00 -3.07114661e-01 -9.50908005e-01 -5.14940619e-01
2.32616380e-01 -9.72986996e-01 -6.49351478e-01 1.15134239e-01
-4.93578702e-01 -2.87866831e-01 5.50687432e-01 -5.13059735e-01
1.71215248e+00 -2.03492904e+00 3.48713398e-02 9.17685986e-01
1.21053386e+00 2.79271990e-01 -3.97902697e-01 6.95956111e-01
4.20500971e-02 5.21339417e-01 -5.65717146e-02 -1.56781375e-01
1.69516593e-01 1.99881777e-01 -5.29311635e-02 5.24947345e-01
-1.34635955e-01 1.00667119e+00 -1.19493413e+00 -6.24312341e-01
3.27675119e-02 5.84640205e-01 -6.98475301e-01 6.13931217e-04
3.31562966e-01 -4.27601576e-01 -3.64151359e-01 4.60153967e-01
8.02573442e-01 -6.56437457e-01 7.79405713e-01 -4.34474617e-01
1.51331931e-01 8.20322484e-02 -1.24352217e+00 9.89051700e-01
-4.77861106e-01 4.92982417e-01 8.12787488e-02 -7.63975799e-01
9.38165724e-01 7.38169476e-02 7.73552775e-01 -2.42547184e-01
4.11245376e-02 3.41960043e-01 -1.13038078e-01 2.54799649e-02
7.81482339e-01 4.70395267e-01 -1.21875770e-01 6.49829149e-01
-1.40505329e-01 6.01097345e-01 1.50954291e-01 7.99048781e-01
1.86984468e+00 -6.51962936e-01 4.39193845e-01 -2.05124944e-01
4.16033953e-01 -3.31855208e-01 5.53885996e-01 5.03456712e-01
-2.06552044e-01 2.56487310e-01 8.00576866e-01 -5.10878503e-01
-9.15234447e-01 -1.04392946e+00 1.64554492e-01 9.26709235e-01
3.47910076e-01 -1.10672987e+00 -3.29131097e-01 -8.71135712e-01
3.03191811e-01 1.37243241e-01 -5.80335140e-01 1.12332013e-02
-5.18130243e-01 -8.17315698e-01 2.97028780e-01 3.62378329e-01
1.26833215e-01 -5.80644369e-01 2.41984241e-02 2.36445069e-01
1.14752084e-01 -9.00383770e-01 -8.18078339e-01 -2.49739185e-01
-9.21252728e-01 -1.08446622e+00 -5.10676444e-01 -8.96664143e-01
1.13085985e+00 8.46944392e-01 1.19941592e+00 6.56645179e-01
-1.39228597e-01 4.77812022e-01 -5.05509496e-01 4.58740592e-01
-7.15260208e-02 4.27749544e-01 1.39857084e-01 1.52397871e-01
2.35518187e-01 -8.98961127e-01 -8.31552684e-01 2.96129614e-01
-8.32528651e-01 -8.78918767e-02 8.91878963e-01 8.19535673e-01
4.15338337e-01 2.09705040e-01 4.16693240e-01 -1.27533019e+00
6.49064541e-01 -8.19736660e-01 -5.05328596e-01 2.34208658e-01
-1.12684083e+00 8.76699612e-02 7.26357758e-01 -2.75159210e-01
-2.33862430e-01 -8.64982903e-02 -2.75185138e-01 -3.08032453e-01
6.80195034e-01 5.53359568e-01 -9.62706804e-02 -4.20017093e-01
3.54601949e-01 9.97441038e-02 -4.98908982e-02 -3.28090519e-01
5.02794087e-01 6.49484575e-01 1.44761220e-01 -3.47349703e-01
1.43801022e+00 1.94547042e-01 2.32431829e-01 -8.10303390e-01
-1.55212179e-01 -6.12958252e-01 -2.37854794e-01 -1.21153660e-01
1.40087947e-01 -7.59797394e-01 -1.00911069e+00 1.87936611e-02
-9.16835010e-01 -1.50412515e-01 3.75328250e-02 3.64410877e-01
1.05786152e-01 7.51232982e-01 -7.71587074e-01 -5.63139975e-01
-6.20177686e-01 -7.20869303e-01 7.07555473e-01 -7.60993436e-02
-8.88217241e-02 -1.19335711e+00 1.36394158e-01 4.01583761e-02
5.34155309e-01 1.98875427e-01 1.00666237e+00 -5.60937822e-01
-6.11937344e-01 -5.50053120e-01 -6.61027431e-01 -5.24533503e-02
1.40149310e-01 8.35017040e-02 -4.79943573e-01 -6.75209165e-01
-8.09790015e-01 1.07414819e-01 6.39883280e-01 -2.24189758e-01
1.15391004e+00 -7.11532414e-01 -5.31991601e-01 6.83431685e-01
1.73930514e+00 -6.14509583e-01 5.16787648e-01 1.59552351e-01
1.09524107e+00 3.97548437e-01 5.06968975e-01 4.35602039e-01
6.65298760e-01 6.42360806e-01 5.79218090e-01 -1.91543892e-01
-1.27265573e-01 -5.15012860e-01 3.43619019e-01 1.34674573e+00
-3.83919813e-02 -4.34544951e-01 -7.37239301e-01 7.08815753e-01
-1.71583414e+00 -5.81687927e-01 -2.89857507e-01 2.24031234e+00
5.29858112e-01 1.29617706e-01 1.37474388e-01 2.45410174e-01
8.37577403e-01 5.43922603e-01 -4.06378388e-01 -2.55707741e-01
2.54874319e-01 3.64132077e-02 1.06284726e+00 6.34037673e-01
-7.84362018e-01 6.54277861e-01 5.48527718e+00 8.24038088e-01
-7.59341836e-01 2.35215455e-01 3.12981278e-01 1.97069809e-01
-8.26234698e-01 1.91173106e-01 -5.92007756e-01 4.46363032e-01
9.98923421e-01 -5.25910318e-01 5.86315870e-01 9.81837869e-01
-2.62840807e-01 5.93912542e-01 -9.06728566e-01 1.08725262e+00
2.97626015e-02 -1.37605166e+00 1.09096795e-01 4.23624367e-01
5.34374297e-01 1.05035715e-01 -9.24896821e-03 2.86222965e-01
4.84235883e-01 -8.01298738e-01 9.85015556e-02 7.39439623e-03
7.75906324e-01 -7.22892165e-01 7.84735441e-01 -3.13744023e-02
-1.80363047e+00 1.58134159e-02 -5.76849341e-01 2.08461910e-01
6.49207979e-02 8.56328607e-01 -7.62372136e-01 6.65007353e-01
5.42512000e-01 6.25712693e-01 -6.64687514e-01 8.97913575e-01
-1.17567867e-01 6.15303278e-01 -4.39087242e-01 -2.91247338e-01
2.58402824e-01 -1.52455583e-01 5.68825722e-01 1.04766071e+00
3.91826153e-01 -7.60991871e-02 2.08887190e-01 2.09536701e-01
-5.64811289e-01 4.00061160e-01 -6.00661993e-01 -1.49822325e-01
9.65052068e-01 1.86227286e+00 -8.50496292e-01 -1.43009052e-01
-5.60964048e-01 9.09777761e-01 7.75380075e-01 7.69847706e-02
-8.90476584e-01 -8.06485832e-01 8.48867178e-01 2.80471295e-01
3.93542200e-01 -2.56012499e-01 1.35590211e-01 -1.01613128e+00
2.03348517e-01 -5.16190231e-01 2.97318757e-01 -8.38685706e-02
-1.30642962e+00 8.09010923e-01 -4.78533119e-01 -1.16370833e+00
2.02176556e-01 -6.13732338e-01 -6.11221373e-01 5.39796710e-01
-1.70283258e+00 -9.59051549e-01 -5.43031216e-01 3.25850099e-01
2.23587882e-02 1.74820229e-01 9.22673583e-01 6.49214685e-01
-6.97117150e-01 1.10259163e+00 2.24732429e-01 2.49164924e-01
2.92432994e-01 -1.44624305e+00 8.81209314e-01 6.27966404e-01
2.37504497e-01 7.45698810e-01 5.26669919e-01 -5.28151929e-01
-1.99533594e+00 -1.42725348e+00 1.12944913e+00 -1.11397162e-01
1.01516318e+00 -5.52468836e-01 -8.19201410e-01 5.90907693e-01
-2.35048592e-01 5.63169718e-01 8.19628656e-01 3.55292737e-01
-5.72642028e-01 -3.45895857e-01 -1.15937638e+00 1.00932014e+00
1.35291636e+00 -3.35535735e-01 -1.24227041e-02 5.08311093e-01
9.28714275e-01 -3.17976102e-02 -1.44174790e+00 2.20250428e-01
6.22502625e-01 -5.77665627e-01 1.12924182e+00 -3.90843511e-01
3.13664116e-02 -3.66582662e-01 -1.75202593e-01 -1.15865207e+00
-7.43365288e-01 -9.01393712e-01 -4.78159934e-01 1.32771099e+00
5.34847975e-01 -1.05130041e+00 9.65433419e-01 3.99726182e-01
2.22307429e-01 -1.09241712e+00 -7.12100148e-01 -6.52632773e-01
-4.53338474e-01 -2.51057535e-01 9.62771297e-01 9.17033672e-01
2.91290917e-02 4.13975388e-01 -3.95887434e-01 5.62589228e-01
8.20853770e-01 1.16672546e-01 1.01832020e+00 -1.45368361e+00
-4.04794544e-01 -2.35128179e-01 -8.28095734e-01 -1.16748869e+00
1.33019611e-01 -1.21507728e+00 -2.35152766e-01 -1.66906548e+00
2.49867007e-01 -9.60678518e-01 -4.93100971e-01 5.48015833e-01
-2.81226456e-01 5.32948792e-01 1.70224115e-01 3.29614609e-01
-6.18175566e-01 4.18038726e-01 7.85212159e-01 -2.07487077e-01
-1.09145589e-01 -1.47017285e-01 -9.21859860e-01 6.38408780e-01
8.03023160e-01 -3.88671845e-01 -4.10770863e-01 -3.79297584e-01
5.88048637e-01 -1.41007304e-01 2.01634139e-01 -7.32504427e-01
3.13362807e-01 1.72526062e-01 -4.19698805e-02 -4.11710829e-01
1.68276876e-01 -9.31196034e-01 2.45987296e-01 5.22875905e-01
-1.00883227e-02 3.04038465e-01 -3.39934200e-01 9.26388144e-01
5.86653017e-02 -5.66945635e-02 3.38511556e-01 2.81581879e-01
-7.20194101e-01 8.15237224e-01 3.07474304e-02 -2.03533128e-01
1.02084637e+00 -1.14979453e-01 -4.98747766e-01 -3.09547693e-01
-3.65099996e-01 4.79649037e-01 6.19506776e-01 2.84639060e-01
7.77786672e-01 -1.67399871e+00 -5.87067485e-01 2.60048881e-02
4.22762841e-01 -3.78758460e-01 2.45091379e-01 8.25210154e-01
-7.03271449e-01 1.85613751e-01 2.94256628e-01 -1.69470161e-01
-1.47433305e+00 6.57363236e-01 -2.36065939e-01 -6.46927893e-01
-8.53162766e-01 6.59384072e-01 -2.05631569e-01 -6.96884930e-01
1.23755604e-01 8.72088894e-02 -1.73451394e-01 -4.57364060e-02
4.26851273e-01 5.13570845e-01 -2.86468794e-03 -6.74137414e-01
-4.53266740e-01 6.07957304e-01 -1.53187886e-01 2.12318003e-01
1.37167490e+00 -5.08878231e-02 -4.09723252e-01 8.23552832e-02
1.63553965e+00 4.35735941e-01 -8.03403258e-01 -4.02380943e-01
-7.05863908e-02 -7.42157280e-01 1.83538735e-01 -7.08345184e-03
-1.40964699e+00 3.53640229e-01 2.40313902e-01 5.57273507e-01
9.82949734e-01 2.37269439e-02 1.24653256e+00 5.23094714e-01
5.46126008e-01 -6.56746626e-01 -1.53000265e-01 1.40931964e-01
3.75127822e-01 -1.05560434e+00 3.14333230e-01 -7.67491043e-01
-2.83586949e-01 1.04812849e+00 4.88668233e-01 -2.12307200e-01
8.81638587e-01 -9.17381793e-02 -4.02607590e-01 -3.65510702e-01
-6.75032258e-01 -1.41595170e-01 9.08709764e-02 3.36466908e-01
1.48638040e-01 3.47919405e-01 -5.64184487e-01 4.31642741e-01
-1.91281289e-01 -4.96831536e-01 5.02847672e-01 6.43127382e-01
-4.72918063e-01 -1.32685113e+00 2.41150986e-02 8.45384538e-01
-1.30208150e-01 -2.74335086e-01 -2.21955299e-01 7.31956005e-01
-3.77767175e-01 7.71714628e-01 5.52381165e-02 -9.53645289e-01
6.52962252e-02 -5.55006564e-01 -8.30020979e-02 -5.53664207e-01
-3.83803666e-01 -5.75477540e-01 1.46770418e-01 -6.31635427e-01
2.01629419e-02 -3.21541935e-01 -1.10675955e+00 -9.91817176e-01
-4.17214096e-01 4.24917996e-01 6.21166408e-01 2.29284778e-01
6.10836029e-01 3.22565526e-01 1.26304483e+00 -7.73977578e-01
-5.93588173e-01 -6.12436473e-01 -8.07314694e-01 2.58135945e-01
1.67981386e-01 -4.36743617e-01 -7.12184370e-01 -7.98596740e-01] | [7.167760848999023, 6.120738506317139] |
bb74bd8b-b46b-4661-a551-9c14a7db4054 | computing-extremely-accurate-quantiles-using | 1902.04023 | null | http://arxiv.org/abs/1902.04023v1 | http://arxiv.org/pdf/1902.04023v1.pdf | Computing Extremely Accurate Quantiles Using t-Digests | We present on-line algorithms for computing approximations of rank-based
statistics that give high accuracy, particularly near the tails of a
distribution, with very small sketches. Notably, the method allows a quantile
$q$ to be computed with an accuracy relative to $\max(q, 1-q)$ rather than
absolute accuracy as with most other methods. This new algorithm is robust with
respect to skewed distributions or ordered datasets and allows separately
computed summaries to be combined with no loss in accuracy.
An open-source Java implementation of this algorithm is available from the
author. Independent implementations in Go and Python are also available. | ['Otmar Ertl', 'Ted Dunning'] | 2019-02-11 | null | null | null | null | ['sequential-quantile-estimation'] | ['miscellaneous'] | [-3.20209861e-01 -1.44120738e-01 -4.06826496e-01 -4.59701300e-01
-1.48371887e+00 -1.10602212e+00 5.01716554e-01 5.60330808e-01
-3.37307632e-01 9.82889712e-01 3.37548465e-01 -5.75475514e-01
-6.15545869e-01 -9.48720872e-01 -3.08004946e-01 -2.65684098e-01
-4.48908687e-01 7.07280576e-01 2.62066066e-01 3.41653004e-02
7.87308276e-01 5.71383059e-01 -1.60295033e+00 1.03244223e-01
7.03024030e-01 1.23201978e+00 -6.36352599e-01 9.97933269e-01
-3.07891488e-01 6.02370381e-01 -7.17990279e-01 -6.71658337e-01
2.08830431e-01 -1.76034227e-01 -5.31699777e-01 -5.03786743e-01
6.64368391e-01 -7.35934675e-01 -1.62988394e-01 9.13396001e-01
6.16782069e-01 2.19405159e-01 9.62571859e-01 -1.25704157e+00
-2.99954802e-01 7.57500887e-01 -4.69142646e-01 -3.15273292e-02
8.88109982e-01 -7.18116909e-02 1.24554300e+00 -9.76476908e-01
4.13958460e-01 1.15725970e+00 8.79233360e-01 -1.51770577e-01
-1.42779076e+00 -7.21739054e-01 -2.13248521e-01 -4.52862263e-01
-1.74709797e+00 -4.60531741e-01 4.42834407e-01 -2.55601197e-01
5.76094866e-01 6.02889001e-01 5.95542610e-01 4.18894202e-01
8.11502412e-02 4.73607928e-01 1.16286147e+00 2.68445145e-02
4.98828351e-01 1.22789508e-02 3.15738916e-01 6.76449835e-01
6.95778847e-01 7.19554052e-02 -6.07979774e-01 -9.53231454e-01
5.86427510e-01 6.79788888e-02 3.24116528e-01 -1.62655637e-01
-9.58461463e-01 7.62547970e-01 -6.74265325e-02 -6.03494495e-02
-1.95531145e-01 5.61772168e-01 5.90795219e-01 2.28016809e-01
7.18747437e-01 2.26894528e-01 -3.01113546e-01 -7.03715146e-01
-1.57142377e+00 8.58636916e-01 1.15413952e+00 1.20857608e+00
7.45311499e-01 -1.04257107e-01 -3.56511563e-01 7.43474066e-01
-2.53173202e-01 6.48554027e-01 -8.32740366e-02 -1.24604404e+00
2.28396982e-01 2.36904100e-01 5.79503059e-01 -9.59552109e-01
-3.37462366e-01 -3.26474845e-01 -6.18555963e-01 2.09335729e-01
7.67291248e-01 1.02725802e-02 -5.74127018e-01 1.29722023e+00
1.29219014e-02 -5.17315328e-01 -3.38785142e-01 4.54110354e-01
8.93650830e-01 4.15476620e-01 -6.56901672e-02 -2.55160242e-01
1.17544138e+00 3.49791050e-02 -6.59176350e-01 3.27589482e-01
2.91364402e-01 -8.73227894e-01 1.13552809e+00 6.33256078e-01
-1.38590860e+00 -3.15897912e-02 -7.58597374e-01 -1.26825482e-01
-6.08492672e-01 -1.21948183e-01 7.71855712e-01 8.44363034e-01
-1.10742772e+00 1.00153947e+00 -4.80304062e-01 -2.84283400e-01
6.36833668e-01 4.02941495e-01 -1.27621785e-01 3.45789418e-02
-7.70015001e-01 5.68327785e-01 2.34261483e-01 -4.45385754e-01
-5.05954385e-01 -8.29810381e-01 -4.57097828e-01 6.34357557e-02
2.27323383e-01 -3.98235589e-01 1.19772148e+00 -2.97130525e-01
-1.20127535e+00 6.21708453e-01 -1.45116523e-01 -9.85842794e-02
8.05601954e-01 -2.78415501e-01 -6.79466426e-02 1.22920670e-01
1.15211621e-01 2.10284933e-01 3.81014317e-01 -1.00540912e+00
-5.81045747e-01 -3.15297008e-01 -1.56643972e-01 -1.33589253e-01
-9.09694210e-02 2.04509899e-01 -1.59378007e-01 -6.75914466e-01
-2.93356087e-02 -4.68730450e-01 -2.73906887e-01 2.05786392e-01
-4.45706636e-01 -3.36002201e-01 2.83429235e-01 -5.56537807e-01
1.75978398e+00 -1.87481153e+00 -5.55621445e-01 7.33384192e-01
2.74881214e-01 -1.86215639e-01 1.51899695e-01 1.05816579e+00
3.70674968e-01 3.80493194e-01 -4.03140694e-01 2.29503009e-02
4.73474413e-01 -5.71494456e-03 -4.88408655e-01 6.26695812e-01
-1.59776509e-01 5.18533885e-01 -9.98443961e-01 -6.69536173e-01
-1.68954343e-01 -5.54293022e-02 -5.86417556e-01 1.69547126e-01
-1.14335194e-01 -3.42158437e-01 -2.15033993e-01 1.04212666e+00
9.87514615e-01 -1.60211608e-01 5.64422682e-02 6.39738068e-02
-4.52300519e-01 2.59551406e-01 -1.49280822e+00 1.31887674e+00
-1.17799468e-01 4.88594830e-01 2.12113306e-01 -6.63770497e-01
1.18486679e+00 -5.61515242e-02 4.11139458e-01 -7.70448148e-02
-3.56736891e-02 6.55280888e-01 -4.94473159e-01 8.61811712e-02
1.01256502e+00 -2.94663638e-01 -3.48243833e-01 7.70018518e-01
-1.95633844e-01 -6.38129711e-01 7.29029357e-01 3.30249608e-01
1.18677270e+00 -9.13546160e-02 4.45737511e-01 -5.52076161e-01
1.22998301e-02 1.51267186e-01 2.01546878e-01 1.10020351e+00
2.24591959e-02 6.61211312e-01 8.71943653e-01 -3.05816770e-01
-1.18649626e+00 -1.55705571e+00 -6.28316700e-01 1.45682955e+00
-1.76644400e-01 -8.57160509e-01 -3.76939386e-01 -4.72455859e-01
6.24952912e-01 6.24407709e-01 -4.71346885e-01 4.41152960e-01
-9.73158479e-02 -4.66251671e-01 7.60965407e-01 6.58455372e-01
1.17805704e-01 -6.63191080e-01 -3.39996338e-01 8.00464600e-02
1.08720988e-01 -5.52879453e-01 -6.16239250e-01 3.32955003e-01
-1.16003680e+00 -1.02790821e+00 -6.83544040e-01 -2.83660531e-01
4.96616602e-01 -1.98542342e-01 1.21149898e+00 4.72619152e-03
-2.33466282e-01 4.02879775e-01 -4.59579639e-02 -4.05453831e-01
-5.59579171e-02 -2.23284448e-03 -2.42984835e-02 -6.59884512e-01
3.69423896e-01 -5.38589716e-01 -6.08682156e-01 1.68972388e-01
-9.77687955e-01 -7.01531231e-01 3.78496796e-01 7.54191041e-01
5.86733222e-01 1.08012840e-01 5.80707848e-01 -1.03324771e+00
1.08034790e+00 -6.52288675e-01 -6.47043169e-01 -1.69828534e-01
-7.96935856e-01 1.45621866e-01 8.86435211e-01 -1.72549963e-01
-5.38771391e-01 -1.57026619e-01 -2.95667619e-01 -2.34535597e-02
7.96380937e-02 6.79447174e-01 3.70418727e-01 3.42739284e-01
7.57294655e-01 1.90143645e-01 1.13940910e-01 -6.67410135e-01
5.71774006e-01 7.16916382e-01 6.32804036e-01 -9.48533416e-01
6.73828602e-01 5.16200542e-01 1.50540739e-01 -6.01817667e-01
-5.00747263e-01 -4.05593872e-01 -2.90968150e-01 4.81292717e-02
-1.21143736e-01 -6.25229120e-01 -8.26300979e-01 9.48184803e-02
-6.21327698e-01 -1.71053082e-01 -5.48666537e-01 3.29155445e-01
-7.49671221e-01 1.80356801e-01 -4.75154996e-01 -1.17265999e+00
-4.38146859e-01 -4.64871556e-01 8.65141988e-01 1.44229159e-01
-6.37101710e-01 -1.02282906e+00 9.75262970e-02 -3.08341742e-01
5.34850538e-01 4.72128659e-01 8.86239707e-01 -9.64136302e-01
-2.97153294e-01 -7.50625968e-01 -4.77699697e-01 1.85663015e-01
-8.49737450e-02 6.75119102e-01 -6.83245063e-01 -3.63483697e-01
-8.18431973e-01 -3.92148226e-01 7.38651931e-01 5.19953966e-01
1.30120981e+00 -7.90469527e-01 -3.24722081e-01 4.44168180e-01
1.38686883e+00 -3.54540616e-01 5.56611955e-01 -2.14217734e-02
-2.07567979e-02 6.61431849e-01 7.76647151e-01 1.26956344e+00
3.15915495e-01 1.00405134e-01 -1.52544500e-02 2.03210980e-01
1.67527243e-01 -4.36332434e-01 1.45147756e-01 5.71934760e-01
3.21733840e-02 -2.94701047e-02 -8.93025339e-01 6.57052517e-01
-1.65673018e+00 -1.27726400e+00 -2.94330001e-01 2.63849068e+00
8.73878241e-01 3.72572869e-01 7.63768792e-01 3.16446811e-01
4.91008848e-01 2.86815226e-01 -3.66537750e-01 -8.81055176e-01
6.92179725e-02 7.03800261e-01 7.54592061e-01 5.56034446e-01
-7.34530032e-01 5.04959583e-01 8.79800892e+00 1.14035153e+00
-6.03990674e-01 -4.38947946e-01 6.12444162e-01 -4.11082000e-01
-8.26018035e-01 1.68008417e-01 -6.26693070e-01 7.24442601e-01
9.85084534e-01 -6.93470120e-01 3.24486583e-01 1.08306134e+00
2.07499430e-01 -5.92718959e-01 -1.07412410e+00 8.67327690e-01
-4.89222616e-01 -1.33203578e+00 4.17519547e-02 1.84384778e-01
4.39298838e-01 -3.67770016e-01 2.83222407e-01 1.27661049e-01
7.39431322e-01 -1.23953819e+00 6.81157768e-01 7.53052175e-01
1.47915852e+00 -1.35316110e+00 6.82647526e-01 2.33024850e-01
-1.26917100e+00 -6.93711713e-02 -4.81961817e-01 -3.53200585e-01
-2.23232165e-01 1.08029568e+00 -5.26061833e-01 2.79894501e-01
7.85930514e-01 3.96999210e-01 -4.07555044e-01 1.03141594e+00
1.97709262e-01 5.99000335e-01 -7.07920849e-01 -5.34384131e-01
-5.02508543e-02 -2.12999836e-01 3.69036615e-01 1.59428585e+00
6.55114174e-01 1.90538615e-01 9.39037502e-02 7.24035203e-01
1.19043153e-03 2.73585945e-01 -7.69156933e-01 -1.39850870e-01
9.75564480e-01 1.10923016e+00 -6.25261903e-01 -6.73437953e-01
-1.12385370e-01 4.05089229e-01 4.24833745e-01 3.12826753e-01
-3.68008912e-01 -1.14989150e+00 6.81643665e-01 5.28476894e-01
3.03097010e-01 -4.55536932e-01 -6.84130728e-01 -6.74530804e-01
-7.25456551e-02 -7.70414472e-01 8.24247718e-01 -2.58837968e-01
-1.58716989e+00 1.90526694e-01 3.30152273e-01 -1.34658217e+00
-4.30022478e-01 -6.29573166e-01 -6.11176014e-01 8.53857398e-01
-8.51914585e-01 -4.38454777e-01 -7.11786300e-02 1.93141803e-01
-2.74651498e-01 8.96756798e-02 8.63826334e-01 -7.64986849e-04
-1.19566046e-01 9.53204751e-01 5.43437421e-01 5.13414219e-02
8.58184040e-01 -1.60989475e+00 2.08244994e-02 4.23436671e-01
-1.96715847e-01 9.11760390e-01 1.03777993e+00 -6.67330444e-01
-1.31379712e+00 -6.92660511e-01 8.96299183e-01 -6.44529462e-01
6.36838377e-01 -1.69522360e-01 -6.40621364e-01 3.25638801e-01
-6.10440783e-02 1.28660604e-01 1.01001763e+00 4.99151647e-01
-4.30794120e-01 -4.61776495e-01 -1.45792472e+00 4.23345476e-01
9.15001750e-01 -6.57690108e-01 -3.04285675e-01 1.46326587e-01
-2.97250487e-02 -4.36043859e-01 -1.26134598e+00 1.09542362e-01
1.17049718e+00 -1.25281656e+00 8.19359660e-01 -3.24724168e-01
4.09915328e-01 -2.56295592e-01 -3.59372079e-01 -1.00750947e+00
-2.10913152e-01 -9.00822043e-01 -2.51915306e-01 1.21914375e+00
2.69529492e-01 -6.61278248e-01 8.06250215e-01 7.74734795e-01
3.76550943e-01 -7.64715552e-01 -8.91777396e-01 -1.03013778e+00
3.28641862e-01 -6.41427279e-01 8.09484243e-01 3.60856026e-01
3.23139310e-01 -2.05671877e-01 -1.82049438e-01 -2.46899977e-01
9.34102714e-01 4.73476171e-01 1.02912891e+00 -1.22716510e+00
-7.32673407e-02 -7.45197058e-01 -2.75133759e-01 -1.05568218e+00
-2.73028165e-01 -9.64258730e-01 -1.41073838e-01 -1.58500338e+00
2.98081279e-01 -6.26809597e-01 -3.39409173e-01 3.41816485e-01
-1.00624584e-01 6.75081909e-01 -1.26116082e-01 8.81183222e-02
-6.34112418e-01 3.38303864e-01 7.79526114e-01 1.83100328e-01
-9.32087526e-02 -1.07581876e-02 -9.51830089e-01 5.13601005e-01
6.67087555e-01 -6.88574016e-01 2.67568836e-03 2.38086909e-01
2.51694560e-01 2.18896016e-01 2.47994825e-01 -8.59981120e-01
1.49581850e-01 -2.71211714e-01 6.07562125e-01 -1.04493237e+00
1.78050026e-02 -3.94207090e-01 1.27862751e-01 2.79678762e-01
-4.91421163e-01 3.96926314e-01 2.13608786e-01 4.45520312e-01
-1.17168956e-01 -2.64284045e-01 6.04383349e-01 -1.65831968e-01
2.23065298e-02 3.25703502e-01 -4.52950209e-01 2.58765548e-01
5.70634723e-01 -2.39460707e-01 -5.27960420e-01 -8.20540547e-01
-3.25129777e-01 1.38987256e-02 7.25675642e-01 -3.88785124e-01
4.87340003e-01 -1.46746540e+00 -8.11438560e-01 -2.19650090e-01
1.76281378e-01 -3.23290616e-01 -2.75633663e-01 6.23318374e-01
-7.44500637e-01 2.23742753e-01 -3.95420566e-02 -2.77712882e-01
-8.78673494e-01 4.84245718e-01 -9.16784629e-02 -2.31649369e-01
-1.40782490e-01 5.96977055e-01 -4.62230325e-01 -3.44983310e-01
2.57507205e-01 -2.39812329e-01 3.35199386e-01 3.05536538e-01
7.30077982e-01 1.09420121e+00 -1.44640744e-01 -1.38551161e-01
-7.23646998e-01 5.18109441e-01 1.79313302e-01 -4.41191822e-01
1.22343194e+00 3.37477624e-02 -1.65620774e-01 7.28838146e-01
1.32355189e+00 7.15551734e-01 -1.17410088e+00 -1.21014463e-02
6.83004707e-02 -8.45350206e-01 -2.66060919e-01 -7.81634271e-01
-4.66573685e-01 5.39162517e-01 1.20540604e-01 5.35179913e-01
9.77655351e-01 -5.90996929e-02 5.03261805e-01 1.17535725e-01
3.35916936e-01 -1.38912129e+00 2.03520339e-03 3.71342480e-01
7.59866774e-01 -9.21900451e-01 7.55743384e-01 -1.34119704e-01
-4.14283872e-01 1.38031888e+00 -4.55041975e-02 -4.52952266e-01
7.33880877e-01 6.79212451e-01 -1.09738782e-01 -1.15199052e-01
-7.19254375e-01 -2.23616943e-01 3.55007768e-01 6.51263773e-01
8.34737599e-01 1.77789286e-01 -8.11057568e-01 7.39895999e-01
-5.07992327e-01 -2.11050343e-02 5.23674428e-01 1.11310673e+00
-5.94073117e-01 -9.91842449e-01 -5.42531610e-01 1.14393353e+00
-6.85394406e-01 -2.31845334e-01 -2.37830296e-01 7.66732752e-01
-3.40421677e-01 1.10003221e+00 3.06125313e-01 -3.30689937e-01
3.00621182e-01 1.46642461e-01 3.25899988e-01 -2.88137585e-01
-5.87556183e-01 -1.17230965e-02 3.91016692e-01 -6.96508646e-01
2.23134622e-01 -7.90544033e-01 -1.17419016e+00 -1.19807708e+00
3.62402014e-02 4.30164546e-01 4.21522528e-01 2.78245419e-01
3.25762987e-01 -1.92550227e-01 7.76576936e-01 -5.90299070e-01
-8.09800446e-01 -7.03531444e-01 -1.20462584e+00 2.41524741e-01
2.53750503e-01 -4.88498867e-01 -6.01212025e-01 -3.58923107e-01] | [7.235615253448486, 4.525822162628174] |
83335df8-1f4c-4bdd-b5cf-8e49173cb7c8 | towards-cnn-map-representation-and | 1709.05972 | null | http://arxiv.org/abs/1709.05972v2 | http://arxiv.org/pdf/1709.05972v2.pdf | Towards CNN map representation and compression for camera relocalisation | This paper presents a study on the use of Convolutional Neural Networks for
camera relocalisation and its application to map compression. We follow state
of the art visual relocalisation results and evaluate the response to different
data inputs. We use a CNN map representation and introduce the notion of map
compression under this paradigm by using smaller CNN architectures without
sacrificing relocalisation performance. We evaluate this approach in a series
of publicly available datasets over a number of CNN architectures with
different sizes, both in complexity and number of layers. This formulation
allows us to improve relocalisation accuracy by increasing the number of
training trajectories while maintaining a constant-size CNN. | ['Walterio Mayol-Cuevas', 'Luis Contreras'] | 2017-09-15 | null | null | null | null | ['camera-relocalization'] | ['computer-vision'] | [ 2.95533955e-01 -4.21135277e-02 -1.98632792e-01 -2.24335745e-01
-4.55961470e-03 -5.38495541e-01 9.33902740e-01 4.17619556e-01
-1.11420345e+00 3.28121573e-01 3.88605714e-01 -2.60406196e-01
-4.38606232e-01 -8.65404308e-01 -9.29854989e-01 -2.81262726e-01
-2.99581718e-02 2.06184089e-01 6.89660370e-01 -2.35255569e-01
4.28344756e-01 8.62133503e-01 -1.76571846e+00 6.24134600e-01
1.01925015e-01 1.08787584e+00 1.82531923e-01 1.13052440e+00
4.78711396e-01 1.02837145e+00 -7.88057864e-01 -3.43425781e-01
4.76811111e-01 -1.82884857e-02 -9.79432583e-01 -1.54048592e-01
9.45584416e-01 -2.76445568e-01 -8.24124336e-01 8.61179352e-01
5.42655706e-01 2.39654079e-01 4.36452895e-01 -1.03873360e+00
-6.77823603e-01 5.24734735e-01 1.19477957e-01 7.78679013e-01
2.19252422e-01 9.41120982e-02 6.82258904e-01 -6.03475869e-01
7.93662965e-01 1.02372956e+00 8.80233169e-01 2.37976357e-01
-1.48309922e+00 -4.54796374e-01 -3.26064765e-01 4.21618789e-01
-1.67540276e+00 -5.02294362e-01 2.02696323e-01 -4.80760127e-01
1.73435271e+00 2.78337955e-01 8.24453294e-01 9.06073749e-01
2.44591370e-01 3.05681795e-01 6.54121280e-01 -3.29483330e-01
2.88217843e-01 -1.63032106e-04 3.21581550e-02 4.33869600e-01
2.49538004e-01 5.72095104e-02 -5.66582859e-01 1.86551824e-01
8.01816463e-01 1.30325004e-01 -2.01942429e-01 -4.42175031e-01
-1.50470376e+00 7.00780451e-01 1.01103222e+00 3.88016969e-01
-2.13845879e-01 7.08801270e-01 8.23727369e-01 4.85526621e-01
3.93535286e-01 6.52760327e-01 -1.43195152e-01 -7.09247738e-02
-1.34946132e+00 3.74271482e-01 5.43051839e-01 9.80661333e-01
9.03151870e-01 -9.13276821e-02 -2.98917323e-01 4.97045815e-01
-2.57878929e-01 2.68164694e-01 8.24564338e-01 -1.06238508e+00
5.55376053e-01 5.48246145e-01 -2.11802900e-01 -1.36073291e+00
-4.93376851e-01 -1.99638680e-01 -9.78416741e-01 3.33988130e-01
2.78191958e-02 2.61966795e-01 -1.13746488e+00 1.30224490e+00
-4.30131912e-01 1.65980399e-01 1.72808729e-02 7.76183486e-01
6.21108651e-01 5.84028959e-01 -1.25320941e-01 3.31229955e-01
1.15272295e+00 -1.04457593e+00 -4.27326083e-01 -4.99889394e-03
8.22962821e-01 -4.91235554e-01 9.36576903e-01 3.47586930e-01
-1.22288251e+00 -7.85309553e-01 -1.44100714e+00 -3.73204380e-01
-8.67175877e-01 3.05087358e-01 3.15697253e-01 3.89150530e-01
-1.83807111e+00 9.21258628e-01 -7.57390976e-01 -6.97673380e-01
5.22749603e-01 6.13727450e-01 -7.00221241e-01 9.45152044e-02
-9.25194204e-01 9.29145992e-01 1.08979487e+00 -2.58856803e-01
-9.21063602e-01 -5.65616906e-01 -7.14071691e-01 4.29395080e-01
-1.56693399e-01 -4.43316787e-01 1.15060461e+00 -8.86019349e-01
-9.68264818e-01 7.65083492e-01 1.95297331e-01 -1.11040461e+00
5.07448435e-01 -7.48397559e-02 -2.54414380e-01 2.98349023e-01
-3.15450937e-01 1.19327700e+00 9.83914018e-01 -5.71549416e-01
-6.30293131e-01 1.67063430e-01 3.92961770e-01 5.28785959e-02
-6.32484138e-01 -4.86052185e-02 -4.55671161e-01 -6.10646725e-01
-1.57162249e-01 -1.09284186e+00 -1.71583980e-01 1.71086550e-01
-2.84407258e-01 4.19131145e-02 9.80440557e-01 -2.35209599e-01
1.28057933e+00 -2.17574334e+00 4.53913540e-01 2.02012345e-01
5.68357170e-01 6.26832843e-01 -8.76478106e-02 4.14515346e-01
-3.49298239e-01 3.35561872e-01 -1.03474997e-01 -6.67741895e-01
-1.95581928e-01 3.00682127e-01 -4.51243669e-01 4.54397589e-01
1.43941283e-01 1.17030847e+00 -6.83903396e-01 -3.20334822e-01
5.94196200e-01 6.89831257e-01 -5.44355690e-01 -2.08376441e-02
1.00665748e-01 -1.20358892e-01 4.14309710e-01 2.30762377e-01
4.77724463e-01 -2.84509242e-01 -1.59025565e-01 -3.19504082e-01
-2.27276772e-01 1.88122720e-01 -9.11776543e-01 1.80475378e+00
-3.36719602e-01 1.61623693e+00 -5.64769745e-01 -8.65618229e-01
7.98538804e-01 9.05059651e-02 2.02130929e-01 -9.36885536e-01
1.29669338e-01 -3.48640122e-02 -2.73033738e-01 -1.77701429e-01
1.13373041e+00 5.45231044e-01 8.79785120e-02 5.53960443e-01
3.52503508e-01 2.45583847e-01 3.27431887e-01 2.38395333e-01
1.17397499e+00 -3.85981053e-01 3.06276351e-01 -4.54105884e-01
2.58513808e-01 8.08459148e-02 -4.04564619e-01 9.24629867e-01
-1.56536669e-01 8.26064110e-01 4.69048381e-01 -1.07008922e+00
-1.74699199e+00 -6.94807589e-01 -1.60518866e-02 1.09496677e+00
-1.57733203e-03 -7.49147654e-01 -8.88084769e-01 -3.33039105e-01
-1.27585709e-01 3.02407920e-01 -1.07952833e+00 -3.41944456e-01
-8.13933790e-01 -6.60482347e-01 1.07211328e+00 6.45605862e-01
7.98693538e-01 -1.14442313e+00 -1.40680695e+00 -1.18927732e-01
1.41126871e-01 -9.88688052e-01 -1.16785802e-01 4.80118781e-01
-7.40211546e-01 -9.20642436e-01 -7.30200291e-01 -5.30223846e-01
5.56646883e-01 4.62244093e-01 1.18049920e+00 8.65546837e-02
-3.28136712e-01 3.17262650e-01 -4.28459197e-01 -3.29807311e-01
-2.94383854e-01 8.86603653e-01 -1.34217858e-01 -3.91447425e-01
3.10594648e-01 -4.68245804e-01 -7.09548473e-01 1.39760435e-01
-1.46520352e+00 4.36508358e-02 4.79946047e-01 3.90598178e-01
3.57575655e-01 -4.33229841e-02 -1.30168200e-01 -4.69728559e-01
9.19880271e-01 -3.45701337e-01 -6.18067205e-01 1.64904576e-02
-8.76464546e-01 3.43485624e-01 4.44274545e-01 -6.10877156e-01
-2.26512074e-01 8.38584602e-02 2.14243218e-01 -9.73915637e-01
-1.23273887e-01 2.59697020e-01 6.97712362e-01 -4.42376971e-01
1.14513087e+00 3.36224586e-01 -1.90614387e-01 -1.84776336e-01
5.74648499e-01 4.53382045e-01 8.71787846e-01 1.32334858e-01
4.54465479e-01 6.65845573e-01 1.17116854e-01 -5.33151507e-01
-2.35509515e-01 -4.72719759e-01 -1.00487542e+00 -4.10733938e-01
9.37764049e-01 -8.00373077e-01 -8.12737763e-01 2.10163534e-01
-1.41045868e+00 -4.26501036e-01 -5.04341841e-01 2.61277080e-01
-5.85649014e-01 5.46427593e-02 -3.94969851e-01 -1.43330963e-03
-3.13785344e-01 -1.15846586e+00 1.00757182e+00 -9.63434353e-02
-1.31550565e-01 -9.81052041e-01 4.31516647e-01 -5.58381319e-01
9.84268367e-01 1.13478936e-01 4.65426862e-01 -5.55766940e-01
-7.45287001e-01 -3.18057418e-01 -6.52002990e-01 3.00689101e-01
-3.69827479e-01 -1.67469829e-01 -1.14379525e+00 -6.05534315e-01
-4.80222136e-01 -3.47594261e-01 1.34817791e+00 1.63408592e-01
1.13120759e+00 -5.29862583e-01 -2.91575700e-01 9.26844656e-01
1.64563394e+00 -7.44269192e-02 1.14188945e+00 8.55172813e-01
8.27374935e-01 3.40440035e-01 9.91657674e-02 4.42799419e-01
2.55967081e-02 8.33684087e-01 7.78132498e-01 -1.37782127e-01
-2.25147724e-01 -1.95343390e-01 8.29314440e-02 5.24527729e-01
-2.82420784e-01 -5.94344914e-01 -1.17418063e+00 8.20889652e-01
-2.03553796e+00 -1.00405359e+00 2.62117505e-01 2.10967731e+00
3.47402006e-01 4.24701720e-02 1.58948943e-01 1.61075935e-01
5.44785023e-01 4.41780120e-01 -1.49298102e-01 -7.33279467e-01
-1.39979050e-01 1.88921794e-01 1.39948440e+00 4.38978344e-01
-1.13361168e+00 8.74515414e-01 7.80070114e+00 5.91113210e-01
-1.53350592e+00 1.69192076e-01 3.69241059e-01 -3.25539529e-01
1.76700294e-01 -8.47209990e-02 -7.34769940e-01 3.80625904e-01
1.44759560e+00 -5.54273650e-02 6.58951521e-01 8.84054601e-01
-2.84619957e-01 -4.05320479e-03 -1.21886706e+00 1.21955431e+00
4.23079669e-01 -1.91778398e+00 3.12875956e-01 2.69071814e-02
5.80897510e-01 6.42422438e-01 1.95413992e-01 2.13378714e-03
1.14444420e-01 -1.40391040e+00 1.14694667e+00 7.56279945e-01
1.05966711e+00 -7.73376405e-01 8.22302520e-01 -4.85684089e-02
-1.29108942e+00 -2.62368143e-01 -8.12287807e-01 -2.00793803e-01
-8.36116299e-02 -9.66455135e-03 -1.01264703e+00 2.28637740e-01
1.15810275e+00 9.31208313e-01 -1.28055501e+00 1.10699487e+00
2.00049952e-01 1.61113739e-01 -3.33895952e-01 1.30222455e-01
4.81694162e-01 4.43380415e-01 4.14119989e-01 1.69105279e+00
3.39448154e-01 -5.86327136e-01 -2.33955175e-01 6.44781590e-01
-1.55586988e-01 -2.71438092e-01 -8.86688232e-01 3.18467557e-01
4.43670928e-01 8.47518086e-01 -8.50182593e-01 -5.62576771e-01
-2.69178063e-01 1.21216774e+00 6.17322981e-01 2.40760669e-01
-6.58537149e-01 -5.79325199e-01 5.25050342e-01 4.16812986e-01
5.73892653e-01 -3.44564587e-01 -1.70505811e-02 -7.80194819e-01
-1.17582113e-01 -2.86956131e-01 2.81489313e-01 -1.09392715e+00
-4.56291527e-01 8.74481380e-01 6.45416856e-01 -1.33213556e+00
-3.87729049e-01 -5.82972229e-01 -3.52980256e-01 6.68778062e-01
-1.48425603e+00 -9.53530669e-01 -7.75333822e-01 6.49330795e-01
3.19413364e-01 -2.12650269e-01 6.84716880e-01 4.53762054e-01
-1.12310886e-01 7.42583632e-01 -6.82929950e-03 3.78218479e-02
6.43708169e-01 -1.01012897e+00 9.15645540e-01 7.89857924e-01
1.94848865e-01 5.46563804e-01 5.91159999e-01 -1.70296028e-01
-8.34165633e-01 -1.51148880e+00 8.86094034e-01 -5.30136287e-01
5.28120637e-01 -5.82162380e-01 -7.60503769e-01 1.09589660e+00
5.68990648e-01 2.46452421e-01 4.51803952e-02 -2.51070738e-01
-5.53172231e-01 -8.20874274e-02 -9.18152690e-01 5.20316422e-01
1.00238681e+00 -8.67179871e-01 -1.53136849e-01 1.70928538e-01
1.05562031e+00 -4.51962382e-01 -8.74983788e-01 2.10998982e-01
4.00274217e-01 -1.10764277e+00 1.18169391e+00 -3.39453876e-01
3.74604642e-01 -3.36423874e-01 -1.40272334e-01 -1.09974372e+00
-6.18882656e-01 -3.02119762e-01 -8.08223262e-02 5.06037176e-01
3.56058121e-01 -3.95568669e-01 7.96531141e-01 1.54092893e-01
-1.35933623e-01 -4.52765584e-01 -1.09124792e+00 -8.55103552e-01
-2.59851098e-01 -5.07980943e-01 5.10794997e-01 8.12414289e-01
-9.35212448e-02 9.97944698e-02 -6.74464941e-01 -1.05328970e-01
3.93560268e-02 -5.81370056e-01 7.97009945e-01 -1.00264406e+00
1.37841227e-02 -3.93280804e-01 -1.25388145e+00 -8.06583822e-01
-1.67466417e-01 -1.03504741e+00 -3.40689361e-01 -1.20548034e+00
1.91387743e-01 -1.23261712e-01 -5.14498234e-01 7.03425944e-01
6.34835064e-01 8.62703800e-01 4.76755291e-01 7.96162963e-01
-8.62513840e-01 7.28743672e-02 5.82472682e-01 -3.06084633e-01
-1.77365825e-01 -6.61108196e-01 -6.92872480e-02 2.24567324e-01
9.45971370e-01 -6.16796792e-01 -7.52078533e-01 -7.98677146e-01
7.06606269e-01 -4.39823717e-01 8.16383243e-01 -1.71486425e+00
6.48746789e-01 3.40469658e-01 3.46008658e-01 -6.11930013e-01
2.71753579e-01 -9.43053126e-01 4.44842368e-01 6.03442967e-01
-7.89395213e-01 9.23133314e-01 5.24063468e-01 6.09717429e-01
-2.79807776e-01 -1.26510948e-01 7.96938539e-01 -9.90281105e-02
-8.88271153e-01 1.56543657e-01 -3.43226850e-01 -4.84063476e-01
1.10576594e+00 -3.92637342e-01 -6.46159410e-01 -3.03725630e-01
-6.75217688e-01 -3.68024886e-01 8.61599684e-01 5.35710990e-01
7.91595757e-01 -1.47684693e+00 -4.81990129e-01 4.23562467e-01
4.84364301e-01 -2.03199014e-01 -9.78697538e-02 7.14470088e-01
-1.19350731e+00 9.64981616e-01 -7.09917545e-01 -9.40524399e-01
-1.30240297e+00 9.01491106e-01 5.38906157e-01 -1.58531964e-01
-6.21472180e-01 4.92263973e-01 -2.40957573e-01 -3.02358959e-02
2.83604175e-01 -5.73476970e-01 -3.17439824e-01 6.05893023e-02
7.35314250e-01 4.56985474e-01 4.19979036e-01 -7.16094673e-01
-1.94803968e-01 3.81514579e-01 -2.27928087e-01 -1.93437189e-01
9.87145126e-01 -3.64346169e-02 7.63322264e-02 2.41976812e-01
1.51408422e+00 -7.21961021e-01 -1.26123476e+00 -1.46979362e-01
-1.17533013e-01 -4.98020142e-01 2.34038994e-01 -2.82776028e-01
-1.02778709e+00 8.64206374e-01 1.09494138e+00 1.04914203e-01
9.88095760e-01 5.06221224e-03 4.65094924e-01 7.51475930e-01
1.53575182e-01 -1.05382609e+00 2.40298271e-01 5.68251908e-01
8.24958265e-01 -9.68662024e-01 1.76548958e-01 1.87246487e-01
-2.84605503e-01 1.18807733e+00 2.49665722e-01 -5.30578732e-01
5.55923820e-01 2.22453892e-01 -3.46194297e-01 -4.53434020e-01
-7.39312947e-01 -9.31324959e-02 1.84249580e-01 6.32921815e-01
-4.71146517e-02 -8.74330923e-02 5.04534654e-02 -3.92522126e-01
-1.87091663e-01 -5.90059943e-02 5.72176635e-01 1.08524013e+00
-4.45712477e-01 -6.54339910e-01 -8.77115354e-02 2.44608000e-01
-2.80074894e-01 -3.15483391e-01 -4.94940758e-01 9.99848127e-01
1.18585035e-01 4.20441955e-01 6.79143190e-01 -8.97605717e-01
4.32124525e-01 -3.89788270e-01 1.53460711e-01 -2.69849569e-01
-6.59069300e-01 -5.56202412e-01 -2.59248078e-01 -9.57480550e-01
-7.00657189e-01 -3.68446589e-01 -6.26697600e-01 -7.94538915e-01
-6.34611174e-02 -4.55154061e-01 8.27465415e-01 5.20310402e-01
5.22899866e-01 6.48617864e-01 1.97247967e-01 -1.32820308e+00
1.07405782e-01 -1.02671218e+00 -3.55364643e-02 4.76220131e-01
6.30927384e-01 -3.46862137e-01 -1.71090454e-01 2.50536114e-01] | [7.812493801116943, -1.7015706300735474] |
534f17e4-a447-49c5-914f-302feb5202ee | controllable-abstractive-sentence | null | null | https://aclanthology.org/2020.coling-main.497 | https://aclanthology.org/2020.coling-main.497.pdf | Controllable Abstractive Sentence Summarization with Guiding Entities | Entities are the major proportion and build up the topic of text summaries. Although existing text summarization models can produce promising results of automatic metrics, for example, ROUGE, it is difficult to guarantee that an entity is contained in generated summaries. In this paper, we propose a controllable abstractive sentence summarization model which generates summaries with guiding entities. Instead of generating summaries from left to right, we start with a selected entity, generate the left part first, then the right part of a complete summary. Compared to previous entity-based text summarization models, our method can ensure that entities appear in final output summaries rather than generating the complete sentence with implicit entity and article representations. Our model can also generate more novel entities with them incorporated into outputs directly. To evaluate the informativeness of the proposed model, we develop a fine-grained informativeness metrics in the relevance, extraness and omission perspectives. We conduct experiments in two widely-used sentence summarization datasets and experimental results show that our model outperforms the state-of-the-art methods in both automatic evaluation scores and informativeness metrics. | ['Qing Li', 'Guanjie Zhang', 'Yi Cai', 'Changmeng Zheng'] | 2020-12-01 | null | null | null | coling-2020-8 | ['abstractive-sentence-summarization'] | ['natural-language-processing'] | [ 3.60095024e-01 6.60187304e-01 -1.56438410e-01 -2.98990667e-01
-1.02752197e+00 -5.55243969e-01 6.51535571e-01 6.57716155e-01
-2.91808397e-01 1.38146329e+00 1.08854043e+00 1.20781817e-01
4.83077988e-02 -8.96661520e-01 -3.72434795e-01 -2.49905199e-01
2.71101117e-01 4.13448513e-01 2.75405914e-01 -1.21986181e-01
7.55444705e-01 3.00081018e-02 -1.42342615e+00 4.04362112e-01
1.85628629e+00 3.25926483e-01 1.03780210e-01 8.79982233e-01
-3.86842161e-01 7.26919711e-01 -1.30974913e+00 -4.48997587e-01
-2.61755198e-01 -6.51929080e-01 -7.79535174e-01 6.59233555e-02
4.14514184e-01 -4.81380105e-01 -6.27441555e-02 1.00179017e+00
6.69852018e-01 2.19305903e-01 1.05152941e+00 -1.03410912e+00
-7.21889377e-01 1.26129615e+00 -6.52146161e-01 8.49414989e-02
7.46772051e-01 -2.00034127e-01 1.30741894e+00 -7.66279936e-01
8.24571311e-01 1.21542287e+00 2.73680329e-01 3.81199658e-01
-8.89278650e-01 -3.60544205e-01 3.49146307e-01 -3.26703519e-01
-8.61022949e-01 -5.10577500e-01 6.15664661e-01 -1.43423423e-01
9.09519792e-01 6.87896848e-01 4.34826434e-01 7.04271257e-01
2.54016668e-01 1.12867820e+00 5.24888694e-01 -3.87593478e-01
1.63968429e-01 1.30165249e-01 6.68670654e-01 2.89743662e-01
9.69431758e-01 -8.31447244e-01 -4.87025976e-01 -3.01406831e-01
1.20542310e-01 -1.65156037e-01 -3.41355681e-01 1.78225741e-01
-1.51145339e+00 7.11057365e-01 9.93057266e-02 1.79648697e-01
-8.30664098e-01 8.34390521e-03 5.59734225e-01 -1.30864099e-01
6.81414068e-01 6.85473025e-01 -1.03036247e-01 -2.24122167e-01
-1.43656874e+00 6.22621000e-01 1.00095880e+00 1.30375433e+00
4.86073911e-01 -2.48171836e-02 -1.17269313e+00 6.49462342e-01
-3.53989564e-02 5.30831695e-01 5.78191459e-01 -7.67295718e-01
8.74690294e-01 9.79539394e-01 4.25741762e-01 -1.09838367e+00
-1.61947057e-01 -5.32013655e-01 -9.84606564e-01 -6.11373961e-01
-3.74664128e-01 -6.25432909e-01 -5.54450393e-01 1.40887189e+00
1.61512882e-01 -4.32879150e-01 3.73132974e-01 4.90857214e-01
1.42674685e+00 1.11882007e+00 -4.27409112e-02 -7.63336062e-01
1.19171953e+00 -1.27419019e+00 -1.18276894e+00 -4.78787906e-02
5.35657048e-01 -6.91448450e-01 7.31928766e-01 1.80544823e-01
-1.41207778e+00 -3.79550457e-01 -1.07650483e+00 -1.99853420e-01
-1.26995325e-01 6.39138460e-01 3.63609910e-01 2.71271825e-01
-8.23505759e-01 7.34935701e-01 -5.48180103e-01 -3.98467869e-01
2.51322359e-01 -2.15108767e-02 -2.43177697e-01 2.59566933e-01
-1.14287400e+00 6.92422152e-01 7.03335285e-01 -2.87691563e-01
-3.07132691e-01 -5.65428317e-01 -8.76016676e-01 4.31886911e-01
2.79428452e-01 -1.18456447e+00 1.36511612e+00 -3.89826685e-01
-1.21577692e+00 2.39268601e-01 -5.52255154e-01 -4.68170911e-01
4.97966081e-01 -5.94993234e-01 -1.12828285e-01 2.60568351e-01
4.77728903e-01 6.50562227e-01 3.21435481e-01 -1.14829254e+00
-9.47961330e-01 -1.14390235e-02 -6.20504515e-03 5.09637892e-01
-4.40765530e-01 -2.28862166e-02 -2.51562506e-01 -6.83964670e-01
-1.05959274e-01 -4.27515745e-01 -1.78683832e-01 -7.15362608e-01
-1.11072707e+00 -5.84466159e-01 5.29679716e-01 -8.24248135e-01
2.11797762e+00 -1.64648533e+00 1.40391350e-01 -2.10063890e-01
1.76760197e-01 3.22241575e-01 1.07600547e-01 9.81691182e-01
3.06427747e-01 5.36719084e-01 -3.41974378e-01 -3.56306314e-01
2.17556641e-01 -2.64047861e-01 -6.49034202e-01 -1.71336725e-01
2.88980782e-01 8.50799382e-01 -1.13705158e+00 -8.89814436e-01
-2.18265757e-01 -3.81366089e-02 -5.07950008e-01 2.50469744e-01
-2.36762449e-01 -5.73378205e-02 -8.33836675e-01 2.34121725e-01
5.75459719e-01 -9.31726545e-02 -2.55429566e-01 -1.55391291e-01
-3.54353696e-01 5.99765956e-01 -1.17755020e+00 1.39285743e+00
-1.22889206e-01 5.97429633e-01 -4.71972853e-01 -5.72576165e-01
9.57144856e-01 2.35273242e-01 1.23814242e-02 -1.96292251e-01
-1.22354634e-01 1.62247688e-01 -3.98978770e-01 -4.37825739e-01
1.62229729e+00 3.07626426e-01 -4.31585133e-01 7.14286447e-01
-1.05897784e-01 -2.85160959e-01 9.96430397e-01 9.13704991e-01
1.07105696e+00 -7.23299831e-02 7.38354623e-01 -1.21511623e-01
4.89911675e-01 2.36900121e-01 4.71066892e-01 9.71491873e-01
2.30761394e-01 8.50902021e-01 8.42981279e-01 1.68355674e-01
-1.13737631e+00 -7.50097990e-01 2.22408876e-01 7.17769980e-01
1.35004774e-01 -8.50799084e-01 -9.97275651e-01 -8.74732673e-01
-1.03649795e-01 1.39704943e+00 -4.40707803e-01 -1.21744655e-01
-4.94079113e-01 -5.99681318e-01 5.43186724e-01 3.73209268e-01
5.81636727e-01 -1.17270994e+00 -4.81311709e-01 2.64719307e-01
-6.98175073e-01 -6.66253805e-01 -8.75228465e-01 -3.59293610e-01
-8.60892534e-01 -6.74933553e-01 -8.58210087e-01 -6.08439982e-01
8.20587277e-01 3.39206874e-01 1.08510065e+00 -9.91625860e-02
3.07794213e-01 1.19794093e-01 -5.66074967e-01 -6.67975247e-01
-5.86916089e-01 6.24118865e-01 -8.97172689e-02 -2.59232074e-01
-2.09336523e-02 -2.83694923e-01 -6.18081152e-01 -2.43778199e-01
-1.09873211e+00 2.74656057e-01 7.36718297e-01 6.85248375e-01
3.51958871e-01 -1.33900046e-01 1.28600383e+00 -1.17923319e+00
1.52555168e+00 -4.74136412e-01 4.57424894e-02 5.54515481e-01
-5.83445668e-01 3.61213326e-01 7.52468586e-01 -1.84886783e-01
-1.25925636e+00 -4.60733056e-01 1.38661936e-01 3.15322220e-01
1.00897141e-01 8.05982471e-01 -7.69305006e-02 1.01406896e+00
6.58458233e-01 5.64165235e-01 -3.65062952e-01 -2.26892769e-01
5.18907011e-01 1.02216530e+00 4.86638904e-01 -3.19677711e-01
6.57442272e-01 6.35600183e-03 -3.75863671e-01 -8.59834254e-01
-9.20893312e-01 -5.24789810e-01 -4.39243853e-01 -1.33235082e-01
5.92474461e-01 -9.10083354e-01 -1.72473937e-01 5.52083626e-02
-1.61407924e+00 5.96945524e-01 -5.54322004e-01 2.69188523e-01
-2.44581133e-01 7.24898756e-01 -2.22936243e-01 -8.47637177e-01
-1.16416800e+00 -6.75209701e-01 1.32934582e+00 5.94260335e-01
-7.55541384e-01 -7.02596784e-01 2.02417731e-01 1.06237233e-01
3.13767731e-01 4.10158128e-01 7.81622291e-01 -1.15986907e+00
-3.25177372e-01 -4.98140365e-01 -1.72447339e-01 1.89084321e-01
2.97114432e-01 4.46280479e-01 -5.62537313e-01 1.61114533e-03
-2.59697944e-01 -1.27667591e-01 1.19636905e+00 5.11508703e-01
7.29043186e-01 -9.68207836e-01 -5.30759573e-01 -2.93135922e-02
9.02634263e-01 -6.39604926e-02 7.35883415e-01 1.36740237e-01
4.22167808e-01 6.84618235e-01 1.01108158e+00 6.91428542e-01
5.91541529e-01 1.50808692e-01 -5.54463230e-02 2.14850754e-01
-1.36691004e-01 -6.82189345e-01 4.30733383e-01 1.37047672e+00
1.10864095e-01 -8.46022308e-01 -2.38387063e-01 7.72997260e-01
-2.03644919e+00 -1.18902528e+00 -3.53727430e-01 2.01048708e+00
1.13668883e+00 1.90646574e-01 -4.61650304e-02 7.74536505e-02
9.39436257e-01 3.87196809e-01 -2.96041697e-01 -5.40089071e-01
-2.08781749e-01 -2.69692034e-01 1.02719784e-01 4.19634581e-01
-9.12130475e-01 9.01041210e-01 5.99766922e+00 8.87592435e-01
-6.01391673e-01 -3.69889051e-01 4.50518161e-01 -1.66596860e-01
-9.05820131e-01 6.06280379e-02 -1.10266316e+00 4.93609667e-01
7.31022298e-01 -1.06407058e+00 -3.64141047e-01 8.36900413e-01
5.62077284e-01 -3.25821668e-01 -1.02187073e+00 4.96717483e-01
3.24230701e-01 -1.44704652e+00 7.73090422e-01 -2.18929201e-01
1.10292232e+00 -6.44262731e-01 -4.98071373e-01 3.73354524e-01
2.97062188e-01 -7.05751121e-01 7.48148918e-01 8.37831676e-01
6.00989103e-01 -8.46737802e-01 8.61723065e-01 5.53334534e-01
-8.60523462e-01 3.12400848e-01 -5.61168432e-01 8.91768411e-02
4.49712843e-01 9.58571911e-01 -1.02808285e+00 1.12585950e+00
9.62531194e-03 8.36584449e-01 -5.95997930e-01 1.09961092e+00
-4.69117105e-01 6.85173750e-01 8.97447858e-03 -6.60095513e-01
1.09947518e-01 -2.81264096e-01 9.47010756e-01 1.52378249e+00
6.07737899e-01 1.14589624e-01 6.79991767e-02 8.77012253e-01
-4.84802336e-01 3.87492388e-01 -6.38853610e-01 -2.85986632e-01
7.91713953e-01 1.34416950e+00 -4.90212888e-01 -8.14041615e-01
1.29776865e-01 8.97836566e-01 2.01374918e-01 2.72945762e-01
-6.65370405e-01 -1.09829330e+00 -3.99799794e-02 -1.83408000e-02
2.27812812e-01 1.03130512e-01 -4.70019311e-01 -1.28499508e+00
2.98435569e-01 -7.25226521e-01 1.43725470e-01 -8.63312602e-01
-8.85415375e-01 6.16440415e-01 1.50963113e-01 -1.31648409e+00
-4.84104753e-01 3.04734945e-01 -1.17864609e+00 8.11158597e-01
-1.21418643e+00 -8.79173398e-01 -3.50331604e-01 -2.45306149e-01
8.90340388e-01 -8.36169198e-02 6.91033423e-01 -1.90277830e-01
-6.73426151e-01 4.99525309e-01 2.35553876e-01 -6.01642653e-02
9.10371542e-01 -1.43303049e+00 5.24134517e-01 1.11673689e+00
-2.67564982e-01 8.84555042e-01 1.02231169e+00 -1.07127523e+00
-1.04808187e+00 -1.24980557e+00 1.45678377e+00 -2.25325465e-01
3.45591009e-01 5.23435511e-02 -7.14886844e-01 5.21808982e-01
7.56419599e-01 -9.36245441e-01 5.24898887e-01 -1.76592663e-01
2.54561007e-01 7.26418272e-02 -9.22013938e-01 7.89663076e-01
8.12269628e-01 2.59211242e-01 -1.15737975e+00 3.78901154e-01
1.15054917e+00 -3.26981485e-01 -6.28627241e-01 3.53375524e-01
3.65153879e-01 -6.81831658e-01 4.29892272e-01 -5.92314422e-01
1.04551578e+00 -3.96088958e-01 2.77819306e-01 -1.74089181e+00
-2.65130401e-01 -7.62382448e-01 -3.16914082e-01 1.84750128e+00
7.97957778e-01 -4.35991734e-01 3.91833484e-01 4.80694324e-01
-5.32541454e-01 -6.89576149e-01 -4.48164850e-01 -4.59956765e-01
-2.31504533e-02 2.89329946e-01 8.34655702e-01 4.29855645e-01
4.58843946e-01 9.13895965e-01 -2.42736608e-01 -1.23049758e-01
4.03453916e-01 3.30075502e-01 9.50474679e-01 -1.19023943e+00
2.24355429e-01 -5.71690321e-01 1.54466495e-01 -1.17133069e+00
-1.45787057e-02 -6.98448300e-01 4.00588550e-02 -2.61929178e+00
7.40993261e-01 1.36160240e-01 2.60431468e-01 1.78309694e-01
-8.11703205e-01 -5.30678570e-01 6.47951616e-03 2.04208598e-01
-1.08605075e+00 9.24211621e-01 1.32215238e+00 -2.50274122e-01
-4.29211706e-01 1.35149593e-02 -1.33362782e+00 4.57882285e-01
7.15888500e-01 -4.22804683e-01 -4.24590886e-01 -2.67876148e-01
2.53930748e-01 2.04833865e-01 -2.58935899e-01 -8.46687376e-01
4.29741174e-01 -1.50008097e-01 1.78858623e-01 -1.25570714e+00
-2.38689959e-01 -3.87338549e-02 -6.72538877e-02 2.41727620e-01
-7.74962366e-01 1.78633109e-01 4.59851557e-03 5.03211677e-01
-3.88716906e-01 -5.65113306e-01 2.47606233e-01 -3.63801084e-02
-1.19017400e-02 -3.09895780e-02 -3.84610146e-01 3.68187815e-01
7.66795993e-01 -2.35252395e-01 -8.06640625e-01 -5.56796610e-01
-1.16809972e-01 5.91181695e-01 4.46890712e-01 2.00112283e-01
6.15139365e-01 -1.21223223e+00 -1.25161147e+00 -4.11302119e-01
1.26254752e-01 3.04374725e-01 1.87296182e-01 6.30435765e-01
-4.84936327e-01 7.50395596e-01 1.66088566e-01 -2.24740058e-01
-1.19165218e+00 2.91451901e-01 -3.41469526e-01 -7.45978177e-01
-4.82844502e-01 3.16835016e-01 1.92018360e-01 -2.82218367e-01
8.32248256e-02 -5.45415342e-01 -7.57151186e-01 4.25501287e-01
7.87634432e-01 6.88993275e-01 -5.74132130e-02 -2.98138857e-01
-4.05967608e-02 1.24151520e-01 -3.22597474e-01 -1.40111819e-01
1.26278532e+00 -1.77752897e-01 -3.97282124e-01 3.87162954e-01
7.26375520e-01 5.44847727e-01 -6.90997362e-01 1.19985685e-01
5.59477732e-02 -2.45162714e-02 -2.38020688e-01 -7.31366038e-01
-4.53410536e-01 4.81012464e-01 -3.67740124e-01 5.89344978e-01
9.57823932e-01 -2.18291506e-01 9.32156920e-01 5.07912517e-01
-8.64218101e-02 -1.18408954e+00 -1.95137560e-02 5.77413917e-01
1.21502602e+00 -9.23529923e-01 5.35365701e-01 -4.38629806e-01
-1.08076441e+00 9.41485107e-01 5.49588919e-01 -1.06382146e-01
-6.41013980e-02 -1.71417907e-01 -4.99345601e-01 -1.50787169e-02
-9.77569997e-01 1.03337713e-01 4.68251646e-01 2.78001338e-01
7.15744138e-01 1.17296591e-01 -1.09341598e+00 9.49103475e-01
-6.44235730e-01 -1.73329577e-01 1.15564489e+00 8.24537873e-01
-9.80926991e-01 -8.71202171e-01 -1.38289124e-01 9.09723461e-01
-6.03695691e-01 -2.09667936e-01 -7.85248637e-01 5.24952292e-01
-6.24285340e-01 1.19576955e+00 -1.20694406e-01 -1.03546523e-01
6.67084694e-01 6.54266179e-02 -2.39826068e-02 -9.94427204e-01
-6.34004951e-01 -7.09776133e-02 6.61617577e-01 1.26085013e-01
-3.42371315e-01 -8.38940501e-01 -1.37579286e+00 -2.59357363e-01
-6.32862210e-01 7.45650291e-01 4.78817761e-01 6.95759892e-01
7.20996439e-01 9.02458072e-01 7.16163695e-01 -6.54021680e-01
-7.82852352e-01 -1.65305960e+00 -3.28935295e-01 3.64964694e-01
2.28820801e-01 -1.67164356e-01 -4.13155675e-01 1.46928996e-01] | [12.538616180419922, 9.493229866027832] |
145d5fbc-8439-4600-a748-8a54306d2a60 | mmgcn-multimodal-fusion-via-deep-graph | 2107.06779 | null | https://arxiv.org/abs/2107.06779v1 | https://arxiv.org/pdf/2107.06779v1.pdf | MMGCN: Multimodal Fusion via Deep Graph Convolution Network for Emotion Recognition in Conversation | Emotion recognition in conversation (ERC) is a crucial component in affective dialogue systems, which helps the system understand users' emotions and generate empathetic responses. However, most works focus on modeling speaker and contextual information primarily on the textual modality or simply leveraging multimodal information through feature concatenation. In order to explore a more effective way of utilizing both multimodal and long-distance contextual information, we propose a new model based on multimodal fused graph convolutional network, MMGCN, in this work. MMGCN can not only make use of multimodal dependencies effectively, but also leverage speaker information to model inter-speaker and intra-speaker dependency. We evaluate our proposed model on two public benchmark datasets, IEMOCAP and MELD, and the results prove the effectiveness of MMGCN, which outperforms other SOTA methods by a significant margin under the multimodal conversation setting. | ['Qin Jin', 'Jinming Zhao', 'Yuchen Liu', 'Jingwen Hu'] | 2021-07-14 | null | https://aclanthology.org/2021.acl-long.440 | https://aclanthology.org/2021.acl-long.440.pdf | acl-2021-5 | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [-2.19931126e-01 -1.00078888e-01 5.33041880e-02 -5.43048024e-01
-7.39086926e-01 -3.87599796e-01 6.50929093e-01 -1.34255111e-01
-3.69213432e-01 4.65158731e-01 7.80620754e-01 -5.33791212e-03
3.97315264e-01 -4.17136848e-01 -2.88120862e-02 -5.39240777e-01
1.86653003e-01 1.81849793e-01 -4.23468888e-01 -8.68404329e-01
-1.30677730e-01 1.40250713e-01 -1.10747731e+00 5.15021503e-01
9.16495979e-01 9.67032909e-01 -3.78941089e-01 8.22336555e-01
-5.14774323e-01 1.17903447e+00 -4.96745110e-01 -8.64334524e-01
-4.82685328e-01 -6.92529917e-01 -7.93105125e-01 1.19497897e-02
-3.87696981e-01 -3.86203900e-02 -2.38190576e-01 6.99391186e-01
8.89234841e-01 4.23150092e-01 4.62968409e-01 -1.32289660e+00
-2.55447030e-01 9.59260106e-01 -4.94351238e-01 -3.60656619e-01
7.35697806e-01 1.26969308e-01 1.04086912e+00 -8.22553277e-01
2.54941016e-01 1.48373234e+00 5.00776112e-01 8.69772315e-01
-7.94984162e-01 -6.43843830e-01 3.88516903e-01 2.17605591e-01
-1.07258320e+00 -6.04026079e-01 1.27121007e+00 2.42935587e-02
9.51838315e-01 5.34130752e-01 4.57200587e-01 1.55986559e+00
-2.06266373e-01 1.08743489e+00 9.70080972e-01 -4.00152475e-01
-1.93380505e-01 2.33994588e-01 3.37222576e-01 5.91721714e-01
-8.80548179e-01 -3.15681577e-01 -8.25409472e-01 -2.33824939e-01
4.10738140e-02 -8.76213834e-02 -3.79115283e-01 3.58557582e-01
-9.10313845e-01 1.08871162e+00 3.87638688e-01 4.50981021e-01
-3.10814440e-01 4.06154990e-02 6.94752395e-01 3.47099781e-01
7.11285591e-01 3.70370448e-01 -1.46718949e-01 -6.66083336e-01
-1.70349181e-01 -2.67860413e-01 1.00227857e+00 6.02341294e-01
5.77059031e-01 -1.21303603e-01 -2.88152277e-01 1.33546662e+00
4.43228930e-01 5.63524604e-01 4.69449043e-01 -6.17330611e-01
6.45586193e-01 9.03669238e-01 -2.29288802e-01 -1.11509979e+00
-8.24483871e-01 1.28036523e-02 -8.62797797e-01 -5.07072866e-01
-8.24581608e-02 -7.31527269e-01 -4.73543108e-01 2.06995010e+00
3.75468612e-01 1.27841949e-01 4.59938526e-01 9.71618831e-01
1.53164780e+00 6.96323991e-01 2.83508927e-01 -2.24120229e-01
1.25093913e+00 -1.22628558e+00 -1.15373588e+00 -1.87217072e-01
9.26594973e-01 -7.78535187e-01 1.06977320e+00 1.72250286e-01
-8.75148416e-01 -2.53369242e-01 -7.04756618e-01 -1.32682890e-01
-4.97215480e-01 9.54451561e-02 8.06558073e-01 5.67264080e-01
-8.51069093e-01 -1.98680647e-02 -4.48712081e-01 -2.54178673e-01
-5.69861419e-02 2.94548661e-01 -5.24435759e-01 8.08240753e-03
-1.63367212e+00 9.36610043e-01 9.58201475e-03 6.93930745e-01
-3.06154042e-01 -9.22471657e-02 -1.04073024e+00 8.12812895e-02
2.36469716e-01 -4.07066226e-01 1.13854229e+00 -1.10428619e+00
-2.21210694e+00 4.32437122e-01 -3.71157080e-01 -8.69518220e-02
3.03238034e-01 -2.30390340e-01 -7.81401694e-01 5.41365072e-02
-7.10980892e-01 5.83686888e-01 5.13522685e-01 -1.18948817e+00
-1.72950432e-01 -3.85520101e-01 1.35675266e-01 6.09626293e-01
-7.75917530e-01 1.65556356e-01 -6.47371471e-01 -2.54625231e-01
-2.82783478e-01 -1.05311441e+00 -1.64396673e-01 -9.44318473e-01
-7.82842755e-01 -4.73670214e-01 8.06800961e-01 -6.35602415e-01
1.44457114e+00 -2.04022050e+00 5.21971703e-01 1.67939112e-01
2.62657672e-01 2.48729572e-01 -4.53628272e-01 9.05088127e-01
1.40668496e-01 5.46143465e-02 -1.68779060e-01 -9.94097233e-01
3.64245534e-01 8.20919126e-02 -1.87021270e-01 3.71562392e-02
2.16770276e-01 1.22667015e+00 -6.59053028e-01 -5.13545215e-01
1.66771442e-01 7.41568089e-01 -3.59741241e-01 5.88635743e-01
2.44425368e-02 6.71275377e-01 -5.39453506e-01 4.68145549e-01
6.33643091e-01 -1.28884956e-01 3.51350158e-01 -1.88601166e-01
1.20410666e-01 1.22052014e-01 -6.98973358e-01 1.73538578e+00
-1.00536084e+00 5.30241609e-01 2.04327837e-01 -6.27390981e-01
9.73522365e-01 5.86199701e-01 4.01944250e-01 -8.42023730e-01
4.91288543e-01 -1.24614052e-01 -1.92302853e-01 -7.57084072e-01
5.73439419e-01 -1.63431868e-01 -6.05621040e-01 5.69508791e-01
8.01092982e-02 1.70792535e-01 -2.60666877e-01 6.50536239e-01
8.22041750e-01 -1.56066120e-01 -4.59337607e-02 3.44887078e-01
8.20548594e-01 -5.59928656e-01 3.14606577e-01 3.01358163e-01
-2.31827661e-01 2.65624434e-01 7.48503923e-01 -4.87440526e-02
-2.10436031e-01 -4.10832226e-01 4.08154219e-01 1.50149024e+00
1.54327899e-01 -5.73083639e-01 -7.26891756e-01 -9.80962336e-01
-3.18814784e-01 6.77473903e-01 -7.39319384e-01 -3.40970159e-01
-2.99068749e-01 -8.27243686e-01 6.69854164e-01 4.44636524e-01
5.90348184e-01 -1.19772017e+00 3.10353246e-02 2.70975322e-01
-9.21465814e-01 -1.34889066e+00 -5.93505919e-01 -1.18722156e-01
-1.65070787e-01 -7.03922212e-01 -5.07798076e-01 -2.46609107e-01
1.79401949e-01 2.01129079e-01 1.05872977e+00 5.70736267e-02
1.38998125e-02 7.56043971e-01 -7.56883025e-01 -2.76816428e-01
-3.33105981e-01 3.95024776e-01 -3.11619043e-01 5.66214204e-01
4.93382573e-01 -4.77923751e-01 -4.10768181e-01 2.35312760e-01
-7.66258478e-01 2.54623622e-01 4.12860572e-01 8.36920738e-01
-2.62753069e-01 -5.80342770e-01 1.04517639e+00 -1.00019848e+00
1.10646272e+00 -6.80696607e-01 2.09197044e-01 4.58999932e-01
-1.30321816e-01 -1.58639565e-01 5.03440917e-01 -3.71157706e-01
-1.54382324e+00 -3.60390320e-02 -5.57139516e-01 -2.95159876e-01
-6.68884367e-02 8.45287085e-01 -5.78442574e-01 5.37643917e-02
9.28655416e-02 -6.99873567e-02 -3.99578847e-02 -2.67052889e-01
7.32671320e-01 1.12850118e+00 2.86921531e-01 -5.64655185e-01
1.10897221e-01 2.37100169e-01 -2.90361702e-01 -8.34661067e-01
-6.61172271e-01 -6.22299075e-01 -3.67002815e-01 -5.77864647e-01
1.08894432e+00 -8.43157470e-01 -1.33340299e+00 4.58158195e-01
-1.36426628e+00 -2.41980270e-01 5.39053261e-01 5.46220660e-01
-2.18158305e-01 3.18583250e-01 -9.02572393e-01 -1.27542639e+00
-5.31179786e-01 -1.09173191e+00 1.08516061e+00 4.16432381e-01
-2.52865821e-01 -1.39195824e+00 2.01713160e-01 7.76885033e-01
4.78548527e-01 3.52039546e-01 6.41085386e-01 -1.04109704e+00
8.24656710e-02 -3.47448945e-01 -1.72014773e-01 3.25467169e-01
1.23705640e-02 -1.82605349e-02 -1.32248259e+00 6.11206517e-02
-2.39801720e-01 -8.49680841e-01 9.15108204e-01 -2.49814421e-01
7.46873081e-01 -2.99612045e-01 -1.34358658e-02 3.78968000e-01
9.20382261e-01 -8.47467557e-02 6.81801021e-01 -1.36915132e-01
1.04099989e+00 9.23156917e-01 3.65835428e-01 6.69644654e-01
9.54860866e-01 6.43039227e-01 5.47542751e-01 -3.63554060e-01
3.00336421e-01 -4.77597490e-02 4.72490579e-01 1.36216116e+00
1.23190256e-02 -6.20496452e-01 -6.91311061e-01 2.03876674e-01
-2.28177285e+00 -8.87807071e-01 -2.30609879e-01 1.61082506e+00
7.63591588e-01 -4.00550812e-01 1.57551780e-01 -3.15468460e-01
5.41872144e-01 3.09309989e-01 -2.80275345e-01 -8.37423325e-01
-4.02766585e-01 1.02032542e-01 -1.52098835e-01 7.47687876e-01
-9.94598746e-01 1.04572666e+00 5.25803757e+00 6.34001791e-01
-1.24752355e+00 -1.12407357e-02 6.40846729e-01 -6.39098808e-02
-4.72484648e-01 -2.94166893e-01 -5.67208588e-01 3.99827123e-01
1.13641644e+00 1.88456193e-01 5.22035897e-01 4.79163051e-01
1.61794588e-01 1.46441072e-01 -8.20358276e-01 1.13277316e+00
3.85354638e-01 -8.45702946e-01 -2.60557443e-01 -1.46656811e-01
4.83921647e-01 -2.06979439e-01 7.04991668e-02 8.86305630e-01
4.03419375e-01 -1.20864820e+00 -1.43166911e-03 7.32720971e-01
3.53966475e-01 -1.16608536e+00 1.07241333e+00 2.06410185e-01
-1.07404745e+00 1.33657306e-01 1.85243770e-01 7.61389956e-02
2.94004172e-01 1.78652197e-01 -7.24221051e-01 7.72878110e-01
3.83913040e-01 7.35434532e-01 -4.93103057e-01 4.18335229e-01
-2.47507259e-01 7.36244440e-01 -1.55149281e-01 -3.83167058e-01
3.91949415e-01 -2.26006836e-01 2.65863746e-01 1.63279510e+00
8.71442407e-02 2.52815008e-01 1.69912234e-01 4.55752611e-01
-3.93371671e-01 6.20166659e-01 -4.56039697e-01 -4.03478056e-01
3.49783868e-01 1.86408138e+00 -1.60716027e-01 -1.89943746e-01
-5.18056750e-01 1.19911742e+00 5.62307000e-01 5.14388144e-01
-1.09642887e+00 -5.74540079e-01 4.99018669e-01 -9.00649607e-01
-1.39292270e-01 -6.65161908e-02 -2.62960675e-03 -1.29244268e+00
-1.86057121e-01 -9.86981332e-01 5.78171849e-01 -6.95413888e-01
-1.44220412e+00 9.51659799e-01 -5.41651189e-01 -9.09594953e-01
-4.73500639e-01 -3.27951014e-01 -9.43315983e-01 9.86443639e-01
-1.49071348e+00 -1.70733571e+00 -4.72318709e-01 8.81566823e-01
2.56111264e-01 -2.89607551e-02 1.11707675e+00 2.82493562e-01
-9.23846126e-01 9.86747861e-01 -2.60513365e-01 2.80092835e-01
1.10298765e+00 -1.00932562e+00 -1.09290868e-01 2.48459607e-01
-6.74011484e-02 5.51999688e-01 4.53902394e-01 -2.21098483e-01
-1.54295599e+00 -8.24065268e-01 8.64117384e-01 -4.28974837e-01
6.19762242e-01 -6.02676570e-01 -9.95468080e-01 5.11166453e-01
8.65770638e-01 -4.48634237e-01 1.30706275e+00 6.23209238e-01
-5.30966938e-01 7.20458403e-02 -8.34872007e-01 7.43724644e-01
6.58303261e-01 -7.45321929e-01 -2.46293962e-01 1.01453803e-01
1.07807231e+00 -2.70024985e-01 -9.97569025e-01 4.09526974e-01
5.01607835e-01 -1.00833464e+00 5.39580286e-01 -7.96621621e-01
5.77401638e-01 2.18319774e-01 -2.03409001e-01 -1.60167503e+00
2.73340821e-01 -9.78068531e-01 -2.71949321e-02 1.71243024e+00
6.39976561e-01 -6.67558193e-01 3.32832813e-01 8.64684880e-01
-2.70577401e-01 -7.47465730e-01 -7.09572792e-01 -3.79535630e-02
-1.33168265e-01 -5.78870296e-01 7.66368866e-01 1.26064432e+00
7.59580731e-01 1.01766026e+00 -8.43882859e-01 -9.10009295e-02
3.21237072e-02 8.78576934e-02 1.02361667e+00 -8.87662768e-01
-1.85874656e-01 -6.78168833e-01 7.04906043e-03 -7.82409906e-01
6.96729362e-01 -6.82303905e-01 1.09546334e-01 -1.44404995e+00
1.45862848e-01 -8.89025927e-02 -3.33272159e-01 4.16063517e-01
-5.31250775e-01 2.07617432e-01 1.75763726e-01 -3.08177561e-01
-1.05351818e+00 1.02905631e+00 1.12907386e+00 -6.35205060e-02
-4.80412960e-01 -1.89367354e-01 -8.79832089e-01 6.58979058e-01
7.38374472e-01 1.49798438e-01 -3.22287500e-01 -3.12343031e-01
2.90490210e-01 5.00289917e-01 8.72024223e-02 -4.24131274e-01
3.15738410e-01 -1.72684327e-01 1.00484639e-01 -4.72912729e-01
8.90914857e-01 -6.67971969e-01 -3.56575131e-01 -2.56270438e-01
-7.19392776e-01 5.09407409e-02 2.77194113e-01 5.34529388e-01
-5.29068530e-01 2.05783844e-01 3.99255663e-01 1.82549953e-01
-5.89884877e-01 2.10770190e-01 -2.99531609e-01 -2.46504080e-02
6.90291464e-01 4.99866277e-01 -5.52597284e-01 -1.05536950e+00
-6.83885276e-01 6.77467108e-01 -2.84482427e-02 6.89469576e-01
6.95365787e-01 -1.56260896e+00 -7.28120565e-01 -1.21374980e-01
4.69471753e-01 -5.82158685e-01 8.48970473e-01 1.16280937e+00
1.11439191e-01 3.20256859e-01 2.31706262e-01 -3.11104029e-01
-1.47965324e+00 3.15752864e-01 4.17247623e-01 -3.08953464e-01
-4.21894118e-02 9.55378830e-01 1.08078644e-01 -8.72436881e-01
3.61497909e-01 3.13585997e-01 -6.20915830e-01 2.36162931e-01
4.66501802e-01 7.54598305e-02 -7.81065226e-02 -1.01606584e+00
-3.82249057e-01 2.99838275e-01 -2.62714201e-03 -5.40137827e-01
1.07523537e+00 -4.68948662e-01 -2.99041003e-01 6.75764203e-01
1.56630802e+00 2.94318348e-01 -7.40590692e-01 -4.69059765e-01
-2.13649690e-01 -1.08171180e-01 2.14186218e-02 -1.01205254e+00
-9.91848648e-01 1.30102754e+00 2.56604373e-01 2.82577783e-01
1.22404611e+00 -7.30199516e-02 1.05713618e+00 4.47943419e-01
-1.17808469e-01 -1.18478751e+00 5.09117067e-01 7.21721649e-01
9.79783952e-01 -1.43905556e+00 -5.10793209e-01 -2.41784647e-01
-1.55408812e+00 1.18266833e+00 8.40867758e-01 4.37834501e-01
4.73090649e-01 -4.18192148e-02 6.60857797e-01 -1.42109528e-01
-9.69101906e-01 -4.12245303e-01 4.29958463e-01 2.38404423e-01
8.40389550e-01 2.19485328e-01 -2.03985468e-01 1.11838436e+00
-4.83297557e-02 -4.36351836e-01 3.04161459e-01 6.02817416e-01
-1.38260990e-01 -1.22400308e+00 1.01850629e-02 -1.47200838e-01
-3.97691399e-01 -2.18213886e-01 -9.59254086e-01 5.71577311e-01
-3.23389441e-01 1.62418747e+00 -2.43137673e-01 -9.72658992e-01
3.18528771e-01 3.35638672e-01 2.88927704e-02 -1.50232151e-01
-9.46141064e-01 2.63056576e-01 6.74144089e-01 -5.42988241e-01
-6.37807608e-01 -2.50481367e-01 -1.34658968e+00 -4.41184163e-01
-4.21513647e-01 4.04668480e-01 8.84449363e-01 1.09751189e+00
4.77501214e-01 6.86217129e-01 1.19752765e+00 -6.35930836e-01
-1.71775550e-01 -1.32875788e+00 -3.29046577e-01 6.07269526e-01
1.52363285e-01 -3.77344549e-01 -4.68640149e-01 -5.10038137e-01] | [13.044110298156738, 5.965666770935059] |
da7ee345-c6ff-41ea-8c58-16c3b7416453 | incoder-a-generative-model-for-code-infilling | 2204.05999 | null | https://arxiv.org/abs/2204.05999v3 | https://arxiv.org/pdf/2204.05999v3.pdf | InCoder: A Generative Model for Code Infilling and Synthesis | Code is seldom written in a single left-to-right pass and is instead repeatedly edited and refined. We introduce InCoder, a unified generative model that can perform program synthesis (via left-to-right generation) as well as editing (via infilling). InCoder is trained to generate code files from a large corpus of permissively licensed code, where regions of code have been randomly masked and moved to the end of each file, allowing code infilling with bidirectional context. Our model is the first generative model that is able to directly perform zero-shot code infilling, which we evaluate on challenging tasks such as type inference, comment generation, and variable re-naming. We find that the ability to condition on bidirectional context substantially improves performance on these tasks, while still performing comparably on standard program synthesis benchmarks in comparison to left-to-right only models pretrained at similar scale. The InCoder models and code are publicly released. https://sites.google.com/view/incoder-code-models | ['Mike Lewis', 'Luke Zettlemoyer', 'Wen-tau Yih', 'Ruiqi Zhong', 'Freda Shi', 'Eric Wallace', 'Sida Wang', 'Jessy Lin', 'Armen Aghajanyan', 'Daniel Fried'] | 2022-04-12 | null | null | null | null | ['program-synthesis', 'comment-generation'] | ['computer-code', 'natural-language-processing'] | [ 3.24487001e-01 2.23185301e-01 -3.16648930e-01 -3.43292594e-01
-9.90384519e-01 -9.47063148e-01 8.92341793e-01 2.71657482e-02
2.10012849e-02 6.46542788e-01 3.82612646e-01 -9.24484909e-01
5.74016631e-01 -6.68939292e-01 -9.92811024e-01 -8.13179016e-02
2.60902464e-01 2.13308319e-01 3.87338921e-02 -1.98036879e-01
3.62742573e-01 -2.44110331e-01 -1.41624904e+00 4.73798335e-01
1.11420381e+00 -5.07673137e-02 2.82437682e-01 9.79592443e-01
-1.77228019e-01 1.01218331e+00 -5.19262135e-01 -6.43873155e-01
4.22578342e-02 -5.86811960e-01 -7.81925142e-01 -3.43077242e-01
3.93510431e-01 -3.01201761e-01 -4.33009258e-03 9.54144955e-01
2.49180570e-01 -1.17877871e-01 6.12875998e-01 -1.15364909e+00
-1.31015110e+00 1.22824502e+00 -5.26834071e-01 6.27869740e-02
5.92203915e-01 3.21174622e-01 1.19910777e+00 -1.15681493e+00
1.04143524e+00 9.12507772e-01 7.72118032e-01 8.79211545e-01
-1.69726884e+00 -5.86426258e-01 -4.35890369e-02 -5.18600881e-01
-1.07806420e+00 -4.36337829e-01 3.71208727e-01 -1.14368284e+00
1.29267943e+00 3.70986640e-01 2.71022528e-01 1.11864841e+00
1.66942418e-01 3.90403628e-01 6.70214474e-01 -3.56587768e-01
3.81158106e-02 1.24154445e-02 7.32687712e-02 9.44665074e-01
9.77827311e-02 -5.81938997e-02 -1.96247980e-01 -5.68591714e-01
3.46804947e-01 -1.49648100e-01 -1.52972981e-01 -1.71877965e-01
-1.37581789e+00 8.71169388e-01 2.58102864e-01 9.52302665e-02
6.32375032e-02 6.60636842e-01 5.95606506e-01 3.10853183e-01
3.89235973e-01 8.09028506e-01 -3.54188502e-01 -5.87907493e-01
-1.23202598e+00 4.98948157e-01 8.47797275e-01 1.53886437e+00
8.33819151e-01 1.25342488e-01 -8.15423608e-01 8.35487068e-01
1.32533923e-01 4.30737704e-01 2.88050532e-01 -7.12949276e-01
8.34073007e-01 5.47959149e-01 6.14734590e-02 -4.71434057e-01
6.67141005e-02 -3.39064419e-01 -3.95233333e-01 3.73598099e-01
1.94643036e-01 -3.66217822e-01 -8.69866729e-01 1.81044865e+00
-1.20202884e-01 -2.17227340e-01 -2.03935355e-01 1.29559293e-01
8.96305203e-01 6.90750182e-01 -2.64017563e-02 2.28444323e-01
1.04388547e+00 -1.26497066e+00 -3.14681292e-01 -3.71541202e-01
9.65331614e-01 -9.58206236e-01 1.45979011e+00 1.19262315e-01
-1.08024406e+00 -4.32699710e-01 -8.31989527e-01 -4.80554581e-01
-2.18846321e-01 2.41128534e-01 5.73648870e-01 6.25134587e-01
-1.29944074e+00 3.63826603e-01 -7.83444822e-01 -6.82711601e-02
5.47939718e-01 -1.89648360e-01 -4.11522627e-01 9.59901884e-02
-6.12660110e-01 5.89879870e-01 1.89565688e-01 -4.05996501e-01
-1.17956448e+00 -1.14878118e+00 -1.02129710e+00 3.16123404e-02
1.61315233e-01 -8.65669727e-01 1.58945858e+00 -8.96249950e-01
-1.23317206e+00 9.69515085e-01 -4.83868927e-01 -3.76506418e-01
7.27923930e-01 -1.76275864e-01 -1.59005821e-01 -7.11056232e-01
3.83484870e-01 5.84981143e-01 8.15057397e-01 -1.28632486e+00
-2.78546631e-01 3.21038932e-01 1.39671609e-01 -3.11651617e-01
-6.15989193e-02 3.79531622e-01 -4.20987040e-01 -1.09259176e+00
-5.66213548e-01 -1.07038593e+00 1.42917614e-02 -1.63449973e-01
-8.27408850e-01 -1.31124005e-01 6.58851683e-01 -9.44313765e-01
1.85555422e+00 -2.14893365e+00 4.49945927e-01 -3.86179425e-02
2.15075463e-01 7.50598684e-02 -2.98962116e-01 6.23443305e-01
-3.87386233e-01 7.19954491e-01 -8.08695018e-01 -6.33707047e-01
2.14637935e-01 -1.20880939e-01 -5.38659573e-01 5.98782301e-02
1.95574343e-01 1.12945676e+00 -9.65629697e-01 -1.19822957e-01
-2.96452165e-01 2.68418908e-01 -1.40492213e+00 2.71021247e-01
-6.81786120e-01 1.72196597e-01 -6.19130954e-02 5.69503665e-01
4.87325668e-01 -3.69105846e-01 3.17807570e-02 3.43171328e-01
-2.78011113e-01 6.74421310e-01 -6.85930133e-01 1.80550051e+00
-9.40992892e-01 9.31493759e-01 -1.93651333e-01 -3.84491324e-01
8.62504125e-01 1.40538439e-01 -3.11886430e-01 -2.90443480e-01
-2.89216548e-01 3.11404943e-01 -1.07033193e-01 -5.10585606e-01
8.90329421e-01 2.82768190e-01 -5.68525612e-01 8.52607906e-01
-9.60911065e-02 -4.53029335e-01 5.72834313e-01 6.80967867e-01
1.32590401e+00 3.14796269e-01 2.72480309e-01 -2.85227299e-01
3.53946060e-01 -4.41089552e-03 3.78875077e-01 9.60941970e-01
5.37532210e-01 7.30959356e-01 1.06160760e+00 -9.90622044e-02
-1.44666517e+00 -9.86034214e-01 1.29199326e-01 1.36680973e+00
-4.61803377e-01 -1.00120926e+00 -9.11402464e-01 -8.99756551e-01
2.19824091e-02 1.26608729e+00 -8.48277867e-01 -8.01621750e-02
-6.27681494e-01 -4.41684455e-01 8.71317089e-01 6.00105643e-01
9.41933915e-02 -1.03185380e+00 -2.76658326e-01 1.06563114e-01
-4.41363081e-02 -4.15728033e-01 -8.52816164e-01 1.56556979e-01
-4.91981328e-01 -9.66209590e-01 -5.79474926e-01 -7.80974746e-01
8.98379803e-01 -3.18891197e-01 1.64769959e+00 6.69441819e-01
-1.63038820e-01 4.68417332e-02 -3.17319125e-01 -1.18456431e-01
-1.25821924e+00 5.27952313e-01 -6.98053420e-01 -7.00585663e-01
-8.10524449e-02 -5.33109486e-01 -1.69210255e-01 7.62555450e-02
-9.28114474e-01 3.48892421e-01 2.69804180e-01 9.65303481e-01
1.59717157e-01 -6.28763616e-01 4.30370778e-01 -1.57430410e+00
8.22544038e-01 -7.62946665e-01 -7.60382771e-01 1.68483153e-01
-5.03080904e-01 2.06052423e-01 7.14546025e-01 -1.09254465e-01
-1.09820867e+00 -1.33522704e-01 -3.60625714e-01 5.02283275e-02
-6.22828864e-02 6.61545575e-01 1.36833459e-01 2.31185585e-01
1.04463768e+00 2.61599779e-01 -1.66173965e-01 -4.41136211e-01
7.05016077e-01 5.95517337e-01 5.69090784e-01 -7.76263058e-01
1.14870822e+00 1.88218318e-02 -4.70082372e-01 -1.91500738e-01
-3.97947133e-01 1.38337702e-01 -4.57357973e-01 2.60906041e-01
7.79103816e-01 -9.97188330e-01 -1.04130059e-02 4.52056736e-01
-1.24699616e+00 -9.93328094e-01 -2.66917139e-01 -2.28450686e-01
-4.67022210e-01 1.99773341e-01 -6.74775064e-01 -2.05614522e-01
-3.71231645e-01 -1.47182727e+00 9.93912995e-01 -3.58814187e-02
-6.44228399e-01 -8.61883938e-01 2.86236823e-01 9.63112637e-02
7.14343309e-01 1.33124664e-01 1.33964562e+00 -4.18994278e-01
-6.34683013e-01 6.24654368e-02 -1.14543028e-01 3.16370308e-01
1.86443925e-01 7.62285113e-01 -6.68428123e-01 -1.54962778e-01
-5.69591820e-01 -2.20387533e-01 9.44283366e-01 -6.99575171e-02
1.29350448e+00 -4.54306692e-01 -3.62013549e-01 8.89097512e-01
1.19553018e+00 -6.42336393e-03 6.18782163e-01 1.72287047e-01
8.07110131e-01 -1.83815882e-01 1.31625116e-01 4.84324306e-01
5.05376041e-01 5.42358696e-01 3.84273082e-01 6.28565997e-02
-3.07109296e-01 -4.79019970e-01 5.27866840e-01 7.47466445e-01
2.27024779e-01 -3.12646151e-01 -1.11053896e+00 8.21598709e-01
-1.62031066e+00 -1.04465818e+00 -2.87816942e-01 2.06166863e+00
1.28228128e+00 8.71123746e-02 9.92910191e-02 -4.00257051e-01
7.05823064e-01 2.41788134e-01 -3.60763162e-01 -8.05921376e-01
1.67950526e-01 2.93826967e-01 4.05504525e-01 7.02377558e-01
-1.01598108e+00 8.79275262e-01 6.81390142e+00 7.29410410e-01
-9.59315121e-01 4.01404381e-01 5.67529500e-01 -6.27086163e-02
-1.22450912e+00 4.23944116e-01 -8.71077120e-01 8.48481297e-01
7.18174815e-01 -3.11294407e-01 8.36507380e-01 1.13349867e+00
-1.10337369e-01 2.99416035e-02 -1.41043770e+00 5.72974861e-01
8.10058042e-02 -1.64069593e+00 4.37075049e-02 -9.35970768e-02
1.41284847e+00 1.80939510e-01 1.19305573e-01 6.98793113e-01
8.91093314e-01 -1.05385184e+00 1.21881771e+00 5.03180683e-01
1.18506825e+00 -4.71994072e-01 2.17095152e-01 1.89676404e-01
-1.02734137e+00 -9.23111886e-02 1.96101498e-02 -1.10258004e-02
5.46859466e-02 7.24849164e-01 -8.44300628e-01 2.29777753e-01
3.31829548e-01 9.46707845e-01 -9.71214533e-01 8.97954166e-01
-5.77973843e-01 7.74100959e-01 9.74429250e-02 3.28886509e-02
-1.33740246e-01 4.68661711e-02 6.62191749e-01 1.57330954e+00
6.99580133e-01 -4.87068504e-01 -1.00944757e-01 1.79063761e+00
-5.78388989e-01 -2.00073466e-01 -6.30465269e-01 -1.81586221e-01
6.73698843e-01 1.13256884e+00 -5.03768206e-01 -6.07551157e-01
-5.52892148e-01 8.64990294e-01 3.42628628e-01 4.19015318e-01
-1.20913851e+00 -7.45448411e-01 5.71061552e-01 5.56495897e-02
5.33174992e-01 -2.07995206e-01 -4.03933346e-01 -1.30152118e+00
-8.36048741e-03 -1.15635991e+00 2.21626028e-01 -9.41873908e-01
-9.33592618e-01 5.87694943e-01 -6.52599633e-02 -9.42291498e-01
-5.19959390e-01 -2.37040550e-01 -9.77391660e-01 1.02007186e+00
-1.10769653e+00 -9.57687140e-01 -2.70535290e-01 1.76616013e-01
6.14774168e-01 -1.57808170e-01 6.99993551e-01 4.07797754e-01
-5.71821392e-01 8.90261531e-01 7.39777908e-02 3.40242624e-01
7.55972326e-01 -1.45697045e+00 1.08888984e+00 1.34350097e+00
-1.47297993e-01 1.11025143e+00 7.41331041e-01 -8.82101059e-01
-1.11611080e+00 -1.31566370e+00 9.67325211e-01 -8.44926834e-01
7.35849738e-01 -8.41587007e-01 -8.83536875e-01 1.12282598e+00
5.27126491e-01 -3.02533656e-02 5.65257192e-01 1.28770679e-01
-8.08021426e-01 4.18283612e-01 -7.15424180e-01 6.75509632e-01
1.07836115e+00 -8.65950942e-01 -5.91571808e-01 2.33044162e-01
8.18005621e-01 -7.09220827e-01 -7.30581462e-01 -4.10220213e-02
3.29231381e-01 -9.47282135e-01 6.21034682e-01 -4.61174011e-01
1.25649977e+00 -5.41849017e-01 -3.28432545e-02 -1.48018789e+00
-2.83790797e-01 -9.20813799e-01 4.38546203e-02 1.67825329e+00
9.75623369e-01 -6.19336009e-01 2.75977284e-01 4.95974183e-01
-5.14350474e-01 -3.54210407e-01 -4.27902728e-01 -6.73167586e-01
4.53681558e-01 -2.42615908e-01 8.85908723e-01 9.90568697e-01
1.40655547e-01 8.97396505e-02 -2.84827948e-01 -3.37753862e-01
8.00812468e-02 2.37215310e-01 1.17702377e+00 -7.75415421e-01
-7.05058336e-01 -7.59016454e-01 2.23591208e-01 -8.44579697e-01
3.02720726e-01 -1.46932602e+00 2.19092980e-01 -1.65856338e+00
4.71501738e-01 -5.23615599e-01 4.50901568e-01 8.31636846e-01
-4.83267695e-01 1.44823149e-01 1.12870537e-01 4.85322513e-02
-2.06426412e-01 3.05906534e-01 8.74215961e-01 -1.86699390e-01
-7.51491114e-02 -1.70192912e-01 -9.30722654e-01 4.13143069e-01
8.08851004e-01 -8.26552868e-01 -3.11828256e-01 -7.02871621e-01
9.46871042e-01 -1.17293224e-01 4.26798820e-01 -7.65676439e-01
-1.18188635e-01 -1.19986132e-01 -8.88917223e-02 -2.16070041e-01
-2.00850680e-01 -1.14699632e-01 4.24842089e-01 2.89333552e-01
-7.14305401e-01 4.03019756e-01 2.94098139e-01 2.31543839e-01
7.15865791e-02 -5.77740073e-01 6.11562550e-01 -2.39448965e-01
-4.78093624e-01 -2.46734023e-02 -5.67789793e-01 4.84823734e-01
8.29720140e-01 1.36643738e-01 -8.52149010e-01 -1.02386653e-01
-4.28140134e-01 -5.37422337e-02 1.06304610e+00 6.68016315e-01
9.39050093e-02 -1.15517044e+00 -6.90631211e-01 3.64114255e-01
3.83650422e-01 -1.26436204e-01 -3.03010326e-02 5.89456201e-01
-7.25798130e-01 9.71279740e-02 8.37843567e-02 -3.59342456e-01
-1.22657239e+00 6.36644244e-01 2.22969294e-01 -2.86857963e-01
-4.80674297e-01 1.01755297e+00 1.16033390e-01 -8.73012543e-01
-2.42506266e-01 -5.96725583e-01 3.28492254e-01 -2.27772504e-01
3.48334283e-01 1.51968449e-01 6.82080686e-02 -2.96581298e-01
-2.23626778e-01 2.78622746e-01 -5.07984534e-02 -4.62695286e-02
1.30057251e+00 1.85037941e-01 -5.49939096e-01 3.46797973e-01
1.09097970e+00 5.98466098e-01 -1.14621735e+00 1.11124210e-01
-1.75887242e-01 -5.68400681e-01 -3.66156787e-01 -9.62168455e-01
-9.77339566e-01 7.25680590e-01 -1.46307722e-01 1.39045224e-01
5.68128049e-01 -7.09169405e-03 4.71335918e-01 1.73597544e-01
2.34527007e-01 -7.33646870e-01 6.19469173e-02 7.58813679e-01
9.77331758e-01 -9.60833013e-01 -1.98808730e-01 -2.77548552e-01
-5.88619053e-01 8.63707483e-01 6.00073814e-01 -1.07831821e-01
4.50591713e-01 8.21382940e-01 -3.02505851e-01 7.53191933e-02
-1.09454489e+00 2.59366542e-01 2.04521030e-01 4.93655920e-01
9.73412395e-01 8.97568017e-02 -6.92128688e-02 4.04076666e-01
-3.88239771e-01 1.22185856e-01 9.38001156e-01 1.00996518e+00
-1.11370254e-02 -1.41117358e+00 -1.92974180e-01 8.03958416e-01
-4.17998195e-01 -6.89521194e-01 -2.20625222e-01 5.67189932e-01
-1.09681217e-02 6.58525467e-01 3.35008278e-02 -2.01622695e-01
3.13045755e-02 2.77749687e-01 1.29674494e-01 -1.18368721e+00
-1.10997260e+00 -3.07912588e-01 2.12895438e-01 -3.27883601e-01
2.06679732e-01 -7.17959285e-01 -1.11688256e+00 -3.96738350e-01
-3.14249960e-03 1.19680073e-02 3.85111004e-01 4.43647534e-01
7.76730299e-01 8.31713915e-01 1.40206352e-01 -6.24842644e-01
-2.67965525e-01 -8.41786265e-01 -1.00656515e-02 5.12338877e-01
4.73442376e-01 -3.04845661e-01 -4.34033483e-01 5.45059144e-01] | [7.730855464935303, 7.8608717918396] |
4877ea0b-21c2-435b-b63e-272456992bb5 | visual-heart-rate-estimation-from-rgb-facial | 2208.04947 | null | https://arxiv.org/abs/2208.04947v1 | https://arxiv.org/pdf/2208.04947v1.pdf | Visual Heart Rate Estimation from RGB Facial Video using Spectral Reflectance | Estimation of the Heart rate from the facial video has a number of applications in the medical and fitness industries. Additionally, it has become useful in the field of gaming as well. Several approaches have been proposed to seamlessly obtain the Heart rate from the facial video, but these approaches have had issues in dealing with motion and illumination artifacts. In this work, we propose a reliable HR estimation framework using the spectral reflectance of the user, which makes it robust to motion and illumination disturbances. We employ deep learning-based frameworks such as Faster RCNNs to perform face detection as opposed to the Viola Jones algorithm employed by previous approaches. We evaluate our method on the MAHNOB HCI dataset and found that the proposed method is able to outperform previous approaches.Estimation of the Heart rate from facial video has a number of applications in the medical and the fitness industries. Additionally, it has become useful in the field of gaming as well. Several approaches have been proposed to seamlessly obtain the Heart rate from the facial video, but these approaches have had issues in dealing with motion and illumination artifacts. In this work, we propose a reliable HR estimation framework using the spectral reflectance of the user, which makes it robust to motion and illumination disturbances. We employ deep learning-based frameworks such as Faster RCNNs to perform face detection as opposed to the Viola-Jones algorithm employed by previous approaches. We evaluate our method on the MAHNOB HCI dataset and found that the proposed method is able to outperform previous approaches. | ['Hassan Ali', 'Ruijia Deng', 'Bharath Ramakrishnan'] | 2022-08-09 | null | null | null | null | ['face-detection', 'heart-rate-estimation'] | ['computer-vision', 'medical'] | [-4.20998968e-02 -2.41310045e-01 2.99141079e-01 -1.87450796e-01
-2.14663178e-01 -1.93024471e-01 -4.00113268e-03 -2.64806479e-01
-6.56030715e-01 4.18378711e-01 -3.13146383e-01 -1.84137523e-02
1.41082972e-01 -7.34310389e-01 -3.64183098e-01 -8.74898016e-01
-1.41262650e-01 -4.19683069e-01 1.10774122e-01 -8.70053098e-02
2.29848891e-01 5.27100027e-01 -1.77479935e+00 -1.09653816e-01
2.93673515e-01 1.28880620e+00 -2.48074934e-01 8.28292787e-01
3.16390783e-01 4.32233542e-01 -6.76951349e-01 -7.68851191e-02
3.90464664e-01 -5.25424957e-01 -3.76803935e-01 6.57297373e-02
6.29101932e-01 -4.70768631e-01 -3.08966458e-01 7.85968125e-01
1.13051820e+00 1.18971147e-01 1.47333905e-01 -1.28253841e+00
-1.45648137e-01 -3.41666341e-01 -6.98359609e-01 2.17860788e-01
3.36581618e-01 8.40317979e-02 2.53314763e-01 -5.58554351e-01
3.80680531e-01 1.35802448e+00 1.03493762e+00 7.69863605e-01
-1.00590360e+00 -8.54729354e-01 -3.65508616e-01 3.85520846e-01
-1.66226435e+00 -4.73323166e-01 8.04080784e-01 -1.07216716e-01
6.16829991e-01 3.43759000e-01 7.90982246e-01 8.41215611e-01
1.71842963e-01 5.17293274e-01 1.15555203e+00 -4.89701688e-01
2.95908868e-01 1.21018089e-01 -3.77507865e-01 7.56855786e-01
-1.03027798e-01 1.94454253e-01 -2.59445041e-01 2.24453509e-02
1.40354443e+00 -1.12250738e-01 -3.69087517e-01 1.77232385e-01
-7.22799957e-01 8.35088789e-01 8.07861909e-02 3.16936672e-01
-2.81212717e-01 2.63811052e-01 4.49072689e-01 1.63180158e-01
3.12949121e-01 2.66980737e-01 -2.34381840e-01 -3.60165685e-01
-1.13042378e+00 1.65433660e-02 6.09455049e-01 6.56121671e-01
3.87644827e-01 1.85993686e-01 -1.11399427e-01 9.69380200e-01
2.54989862e-01 5.17731547e-01 5.38519681e-01 -1.57140613e+00
-1.52767286e-01 3.25941265e-01 9.16436911e-02 -1.31167090e+00
-6.57726586e-01 3.57117467e-02 -9.39216673e-01 6.72988236e-01
4.89675671e-01 -1.87367991e-01 -7.13956892e-01 1.63927388e+00
4.24680084e-01 2.49763414e-01 7.76741747e-03 1.16793418e+00
1.14074671e+00 3.96108180e-01 -2.19333559e-01 -5.28399885e-01
1.48740590e+00 -4.20542300e-01 -6.92286015e-01 1.95446476e-01
1.06163755e-01 -9.64778066e-01 9.60282743e-01 8.75276685e-01
-1.14664555e+00 -1.17069697e+00 -1.00187314e+00 -5.70363663e-02
1.57930925e-01 3.63570660e-01 1.17709666e-01 1.35312665e+00
-1.25212705e+00 6.28859639e-01 -7.77892470e-01 -7.31844306e-01
8.70262086e-02 4.34671491e-01 -3.90692741e-01 1.33569106e-01
-8.47144008e-01 6.86893225e-01 1.70983508e-01 4.32796150e-01
-3.83886844e-01 -2.06965536e-01 -8.67643237e-01 -1.20889276e-01
2.84484327e-01 -3.04681122e-01 1.08670056e+00 -1.35613871e+00
-1.95051289e+00 6.29507899e-01 -1.93657309e-01 -1.63005576e-01
5.85169911e-01 -5.20167388e-02 -4.30593461e-01 4.11961764e-01
-4.31695372e-01 7.09495366e-01 7.91094482e-01 -6.19844556e-01
-4.21700776e-01 -2.16323823e-01 -5.84638678e-02 6.45238832e-02
-3.02793533e-01 -7.74207860e-02 -5.46334684e-01 -4.63444591e-01
-9.87195820e-02 -1.28855467e+00 2.85609066e-01 3.66918176e-01
1.38318911e-02 -7.27689117e-02 9.32197452e-01 -6.70215189e-01
1.18616700e+00 -2.12420273e+00 -5.19089341e-01 8.91086757e-02
1.57625407e-01 6.86850905e-01 -3.23535874e-02 3.62610817e-02
-6.26231544e-03 9.86771006e-03 1.90824062e-01 -1.18865818e-01
-5.07924914e-01 1.47868663e-01 3.01482230e-01 6.64140880e-01
7.92159233e-03 3.05004478e-01 -4.73737031e-01 -3.77801150e-01
4.09548283e-01 1.18520582e+00 -2.67741829e-01 2.30096251e-01
2.86690801e-01 4.00825620e-01 3.38732786e-02 6.59823120e-01
7.88038909e-01 4.20531780e-01 -6.38849586e-02 -7.17077911e-01
-1.11788243e-01 -2.85999805e-01 -1.23833072e+00 1.30349362e+00
-4.85223413e-01 9.32907581e-01 1.41397491e-01 -9.54918563e-01
1.28269351e+00 6.87334776e-01 6.19908512e-01 -9.73530829e-01
2.91938156e-01 5.30406125e-02 3.82661521e-02 -6.42522395e-01
1.88753739e-01 -4.51963514e-01 4.24364626e-01 2.32353672e-01
-1.52109310e-01 1.76280007e-01 5.89864478e-02 -1.70453340e-01
1.13218772e+00 5.10568738e-01 2.64457941e-01 -8.01398382e-02
8.27958345e-01 -5.48096240e-01 6.85404420e-01 2.71283388e-01
-5.83836377e-01 8.25897396e-01 2.72003144e-01 -7.99361348e-01
-8.35910738e-01 -7.75536597e-01 -1.29373610e-01 6.30471766e-01
2.53585055e-02 -3.62688720e-01 -7.33932614e-01 -3.01929533e-01
-2.02453852e-01 -8.82215500e-02 -5.18859744e-01 -2.55069137e-01
-2.72995561e-01 -5.05685270e-01 7.13600755e-01 5.69254220e-01
8.28438461e-01 -1.28498685e+00 -1.05499935e+00 3.55049103e-01
-3.43873322e-01 -1.37468088e+00 -4.18405265e-01 -2.68459648e-01
-8.82173777e-01 -1.03849614e+00 -8.05765271e-01 -2.88403690e-01
2.94652045e-01 3.48573297e-01 8.18938494e-01 2.90684789e-01
-9.76281643e-01 7.10811615e-01 -2.72985935e-01 -5.62249541e-01
-3.75318497e-01 -3.86172235e-01 1.81857273e-01 2.81574726e-01
2.47003093e-01 -3.50730211e-01 -1.12634635e+00 6.10366285e-01
-9.34327185e-01 -2.49531135e-01 1.51746973e-01 2.99685150e-01
5.69573104e-01 6.70287013e-02 6.89723611e-01 -4.56948310e-01
3.80536467e-01 1.82834510e-02 -7.12773502e-01 -1.78834230e-01
-5.59450805e-01 -4.45102870e-01 4.21628833e-01 -3.59289348e-01
-8.94563913e-01 -4.38036695e-02 -2.48642072e-01 -4.99667943e-01
-2.23707840e-01 -3.73723246e-02 -1.31798759e-02 -1.93142831e-01
6.71830893e-01 -3.53428662e-01 3.24559957e-01 -2.91278869e-01
-1.48509860e-01 9.54349339e-01 6.14155531e-01 -1.75261900e-01
3.59269708e-01 4.13580894e-01 4.59296584e-01 -1.23981750e+00
-4.64144737e-01 -6.79827869e-01 -4.44014132e-01 -7.22625077e-01
9.57782745e-01 -1.02660215e+00 -1.17082632e+00 6.76948190e-01
-8.86816502e-01 -2.20286734e-02 2.49209389e-01 5.50571084e-01
-7.36992896e-01 4.03088540e-01 -7.40230560e-01 -1.16460717e+00
-7.97265768e-01 -1.11960495e+00 8.99872124e-01 7.03914285e-01
-3.85978252e-01 -8.95286918e-01 -7.78965205e-02 5.49117982e-01
4.81199741e-01 6.69151425e-01 3.21400434e-01 8.79171267e-02
-4.69843782e-02 -1.54902458e-01 -2.37813115e-01 5.42718351e-01
5.15073836e-01 3.45097125e-01 -1.18072224e+00 -4.37632829e-01
-2.49437951e-02 1.73669029e-02 4.70197201e-01 4.90940690e-01
1.13238263e+00 3.12031060e-02 1.48820639e-01 5.62650383e-01
1.33778596e+00 4.67764705e-01 1.17078972e+00 3.60451072e-01
5.12658298e-01 5.85889161e-01 6.89545810e-01 3.62432003e-01
7.21002296e-02 7.82567203e-01 4.24514174e-01 -6.04884803e-01
-1.39615938e-01 4.18278903e-01 5.04251659e-01 2.52897203e-01
-6.19432449e-01 2.55008966e-01 -4.47575986e-01 2.84718752e-01
-1.65557086e+00 -1.06141520e+00 -2.96020061e-01 2.44702435e+00
6.41042829e-01 -1.94884062e-01 7.61970758e-01 5.16875386e-01
6.12100482e-01 -1.96686208e-01 -4.94412661e-01 -1.13941300e+00
2.78142214e-01 6.03307307e-01 1.45861179e-01 -9.04172435e-02
-1.00464654e+00 3.18866462e-01 4.64276075e+00 5.03429353e-01
-1.60627472e+00 6.86371401e-02 6.07927978e-01 -3.41419399e-01
5.59105873e-01 -3.45869005e-01 -5.26013792e-01 4.94921058e-01
9.77849722e-01 3.14084470e-01 3.24704409e-01 7.03209758e-01
5.27936578e-01 -7.04357564e-01 -9.32641089e-01 1.45289588e+00
1.58760399e-01 -7.71867692e-01 -8.18291247e-01 5.90734892e-02
3.70368987e-01 -5.10645449e-01 -3.79634112e-01 1.49038553e-01
-6.73677385e-01 -1.08426762e+00 2.31662840e-01 5.80610096e-01
8.88843238e-01 -9.39880490e-01 9.65358615e-01 1.31492876e-02
-1.32409561e+00 1.71428695e-01 -2.75796592e-01 -1.22074047e-02
-3.01839232e-01 5.35895407e-01 -8.81678700e-01 1.66779295e-01
9.28036690e-01 2.38546044e-01 -2.98803121e-01 1.24584413e+00
-1.89090881e-03 4.53319609e-01 -1.84279561e-01 2.63259292e-01
-2.10266009e-01 -1.43144771e-01 1.45196989e-01 1.13630068e+00
6.49930418e-01 -1.61207914e-02 2.02160716e-01 5.94520211e-01
2.95259561e-02 3.07829052e-01 -4.42789942e-01 1.95703059e-01
-1.31762654e-01 1.68114567e+00 -8.34397554e-01 -1.75997794e-01
-4.75938469e-01 9.43950653e-01 -1.95380375e-01 1.96801871e-01
-1.02017462e+00 -6.04811072e-01 5.45396090e-01 6.09851062e-01
1.13150030e-01 -3.23587619e-02 1.84633255e-01 -6.43699527e-01
-1.00076720e-01 -8.63815248e-01 4.18828547e-01 -1.00615072e+00
-6.96303666e-01 5.33007622e-01 -1.34506911e-01 -1.47821212e+00
-1.27785340e-01 -2.77116299e-01 -7.67152429e-01 1.00572312e+00
-1.45782769e+00 -8.91456962e-01 -8.43713045e-01 7.85217583e-01
5.11255920e-01 9.83565301e-02 8.25507581e-01 5.17120898e-01
-6.63599491e-01 6.18387103e-01 -2.89299697e-01 9.84035581e-02
1.08176374e+00 -8.14303100e-01 -7.85228144e-03 5.98535657e-01
-1.46201685e-01 4.75861460e-01 4.95508939e-01 -2.52350450e-01
-1.39333034e+00 -7.71840632e-01 3.08715463e-01 2.93082774e-01
-2.33649477e-01 -1.13244928e-01 -9.85915601e-01 2.97924221e-01
1.69239774e-01 1.81257606e-01 7.43759692e-01 -4.58473302e-02
-1.56567678e-01 -6.88958526e-01 -1.37474036e+00 2.79538661e-01
3.52843612e-01 -6.07193291e-01 -6.83915764e-02 -9.57164392e-02
-1.70779139e-01 -3.52300584e-01 -9.96486723e-01 4.60521311e-01
1.19267595e+00 -1.19336915e+00 6.73364878e-01 1.97436064e-01
1.88608006e-01 -5.14832377e-01 1.76254705e-01 -1.18273556e+00
-1.69630527e-01 -6.05700731e-01 -2.89445296e-02 1.24069428e+00
-7.22351745e-02 -7.82369554e-01 6.30189717e-01 8.90013456e-01
4.63195294e-01 -6.92393601e-01 -9.78664756e-01 -7.44051814e-01
-2.98762500e-01 -2.74724573e-01 -1.51988104e-01 6.67392313e-01
-4.34628688e-02 -2.71508675e-02 -6.16203010e-01 -2.90763704e-03
4.46792513e-01 -9.30964798e-02 1.10326910e+00 -1.28400266e+00
-1.58584684e-01 -8.92366394e-02 -5.72315693e-01 -2.15896472e-01
-3.00634831e-01 -7.54047185e-02 5.26361838e-02 -1.37060511e+00
1.25044748e-01 -2.19613522e-01 -4.80511129e-01 5.99391997e-01
4.37347740e-02 9.79222894e-01 3.15113217e-01 -7.84598663e-02
-1.58217415e-01 -6.78078085e-02 1.03719676e+00 2.83267140e-01
-4.08063948e-01 1.46911949e-01 -5.36962986e-01 7.83749163e-01
9.49445128e-01 -3.42503786e-01 -3.19860250e-01 1.43369108e-01
2.44080067e-01 4.55064237e-01 5.14159024e-01 -1.30256939e+00
6.07877076e-02 1.34446055e-01 7.57284462e-01 -1.44765466e-01
5.39630651e-01 -7.21395254e-01 3.07836235e-01 5.39371848e-01
2.65962064e-01 9.55072269e-02 6.27944410e-01 1.53265566e-01
-7.53106773e-02 7.56942630e-02 1.10712445e+00 -3.38114083e-01
-4.49470401e-01 9.94978547e-02 -3.97081256e-01 -4.22871649e-01
9.39819515e-01 -8.38573039e-01 -1.21871293e-01 -7.68749416e-01
-7.20079780e-01 -4.92479429e-02 3.88620526e-01 5.38294673e-01
5.92018783e-01 -1.20288491e+00 -4.21498179e-01 6.37848675e-01
-1.62990257e-01 -2.61009872e-01 2.60043621e-01 1.27445173e+00
-6.50456548e-01 1.38021752e-01 -6.28222287e-01 -6.42628610e-01
-2.02944708e+00 3.72226894e-01 5.69716930e-01 3.95715743e-01
-3.70972604e-01 3.38575482e-01 6.52377531e-02 1.87522829e-01
2.62678683e-01 -2.58061260e-01 -3.11848313e-01 9.37903449e-02
5.94838023e-01 8.61346185e-01 2.63412625e-01 -6.98572636e-01
-3.26851517e-01 9.57747877e-01 3.03939819e-01 6.87063187e-02
9.59240377e-01 -3.47646505e-01 -1.89346708e-02 3.77642214e-01
1.35273468e+00 -1.90034192e-02 -1.00775838e+00 1.01589181e-01
-4.47922200e-01 -5.53138316e-01 2.72638977e-01 -8.11403513e-01
-1.50851202e+00 1.04989958e+00 1.46349669e+00 8.23446438e-02
1.70864213e+00 -6.24666631e-01 9.47178245e-01 -7.66639933e-02
4.26648974e-01 -1.39252281e+00 7.98946992e-02 2.69333622e-03
5.43889642e-01 -1.31078923e+00 1.08314537e-01 -1.95540220e-01
-6.32657349e-01 1.69125533e+00 7.23087072e-01 4.93503176e-02
4.51797843e-01 1.29086643e-01 6.34800315e-01 1.25378236e-01
-1.78738922e-01 -3.11645150e-01 3.57807040e-01 5.78660607e-01
9.15178180e-01 -2.39315107e-02 -4.08074319e-01 7.64832571e-02
6.53332993e-02 4.88218784e-01 6.22472227e-01 7.56721854e-01
-1.88556120e-01 -9.64227021e-01 -6.33049428e-01 1.52564704e-01
-7.12767601e-01 3.09795201e-01 -1.45244092e-01 8.16763520e-01
4.62294608e-01 1.28452384e+00 5.10869399e-02 -1.84680015e-01
5.35474777e-01 9.19988453e-02 6.31421626e-01 -1.24294750e-01
-8.07600617e-01 6.43906534e-01 3.99077274e-02 -8.18915784e-01
-6.20267987e-01 -4.48709279e-01 -1.16901243e+00 -3.97451699e-01
-1.97517112e-01 -2.49162033e-01 8.45812142e-01 7.03250229e-01
4.18818861e-01 2.64849663e-01 6.93996787e-01 -8.89615774e-01
1.55235901e-01 -8.39707971e-01 -7.85000443e-01 3.59223932e-01
3.27824146e-01 -5.76369405e-01 -1.16025172e-01 4.20803875e-02] | [13.867256164550781, 2.698054552078247] |
b6d4bb6b-d7f2-495c-bda3-8c543c66bacb | cooperative-multi-agent-path-finding-beyond | 2105.10993 | null | https://arxiv.org/abs/2105.10993v1 | https://arxiv.org/pdf/2105.10993v1.pdf | Cooperative Multi-Agent Path Finding: Beyond Path Planning and Collision Avoidance | We introduce the Cooperative Multi-Agent Path Finding (Co-MAPF) problem, an extension to the classical MAPF problem, where cooperative behavior is incorporated. In this setting, a group of autonomous agents operate in a shared environment and have to complete cooperative tasks while avoiding collisions with the other agents in the group. This extension naturally models many real-world applications, where groups of agents are required to collaborate in order to complete a given task. To this end, we formalize the Co-MAPF problem and introduce Cooperative Conflict-Based Search (Co-CBS), a CBS-based algorithm for solving the problem optimally for a wide set of Co-MAPF problems. Co-CBS uses a cooperation-planning module integrated into CBS such that cooperation planning is decoupled from path planning. Finally, we present empirical results on several MAPF benchmarks demonstrating our algorithm's properties. | ['Nahum Shimkin', 'Oren Salzman', 'Ofir Gordon', 'Nir Greshler'] | 2021-05-23 | null | null | null | null | ['multi-agent-path-finding'] | ['playing-games'] | [-9.92631838e-02 5.85706770e-01 -9.64659378e-02 -4.67485674e-02
-4.30306435e-01 -7.50840187e-01 5.60144305e-01 4.70200777e-01
-5.19785583e-01 1.03503764e+00 -1.21780194e-01 -1.43899366e-01
-7.51824200e-01 -7.86133945e-01 -4.23690140e-01 -7.04051733e-01
-9.07588303e-01 1.05896974e+00 8.59302282e-01 -6.74234569e-01
3.34045976e-01 2.83690870e-01 -9.02205586e-01 -2.63050735e-01
8.26939285e-01 3.85207951e-01 7.62826383e-01 7.47205436e-01
3.67466539e-01 6.99095666e-01 -5.49748302e-01 1.71457529e-01
4.84716266e-01 -1.04551896e-01 -1.38640416e+00 2.47097611e-01
-6.62410498e-01 -1.66299805e-01 -1.29581437e-01 6.40860379e-01
-3.53280939e-02 4.09262419e-01 6.91108763e-01 -2.22405457e+00
2.84018517e-01 7.62228012e-01 -6.80315316e-01 -3.59354243e-02
7.24707425e-01 1.20420754e-01 8.87489736e-01 -2.10683644e-01
1.03771663e+00 1.19611979e+00 3.16445440e-01 2.84314841e-01
-1.16463912e+00 1.88514944e-02 5.04788160e-01 4.71206456e-01
-1.28841054e+00 -1.12259962e-01 1.92473978e-01 -1.58623993e-01
1.50598454e+00 1.88446000e-01 6.15677714e-01 3.33937615e-01
7.57761240e-01 3.88479084e-01 9.13598120e-01 -4.12285954e-01
5.46800017e-01 -4.42084849e-01 1.61456928e-01 3.83757889e-01
4.13528174e-01 1.37725234e-01 -4.51342821e-01 -5.98003268e-01
5.17260075e-01 -2.49387220e-01 -1.49079114e-01 -8.62444222e-01
-1.66333389e+00 8.62834990e-01 3.77844781e-01 2.75554746e-01
-6.90605521e-01 5.36714971e-01 2.27523446e-01 8.04493427e-01
-5.59518375e-02 6.00440145e-01 -2.54911959e-01 1.10457122e-01
-1.15778580e-01 7.69443274e-01 1.07933283e+00 1.13476908e+00
1.00382960e+00 -5.74424982e-01 2.07795367e-01 2.24982277e-01
3.41564864e-01 1.62177980e-01 -3.94931972e-01 -1.83181047e+00
3.74514282e-01 5.46002567e-01 5.91994226e-01 -9.96168673e-01
-9.33737934e-01 3.89687791e-02 -4.98327613e-01 8.48286688e-01
2.90124089e-01 4.02849354e-02 -3.75538580e-02 1.69612432e+00
6.39674008e-01 -1.64914131e-02 4.71087337e-01 8.75799000e-01
-2.13089474e-02 7.40955114e-01 -2.79923588e-01 -6.26251757e-01
9.70647395e-01 -1.64261854e+00 -4.39963996e-01 -2.28809834e-01
9.75286245e-01 -3.89812469e-01 3.63171607e-01 4.20245826e-01
-1.34107351e+00 2.05807060e-01 -9.82320726e-01 3.96028548e-01
-1.49085790e-01 -8.59690547e-01 5.77115119e-01 1.65156767e-01
-1.60689926e+00 3.38526338e-01 -8.12312961e-01 -5.35759032e-01
-8.92369375e-02 6.50229752e-01 -6.66383386e-01 -3.00629914e-01
-6.59946799e-01 7.53603816e-01 5.06424665e-01 -4.03776288e-01
-1.12584090e+00 -1.13574393e-01 -4.59761709e-01 1.19364224e-01
1.38476050e+00 -9.55226243e-01 1.59016812e+00 -4.13225740e-01
-1.26730418e+00 3.93595994e-01 5.10225035e-02 -4.29202080e-01
3.60189319e-01 2.53828615e-01 7.30639920e-02 5.57591282e-02
6.57254994e-01 6.35106921e-01 1.72866046e-01 -1.49030828e+00
-1.09843493e+00 -1.62976831e-01 6.22291803e-01 5.56599498e-01
2.96361983e-01 4.69770655e-02 -2.50689954e-01 -3.47640179e-02
6.12942390e-02 -1.37224877e+00 -9.78825927e-01 -2.79100984e-01
-3.33412081e-01 -7.54333735e-01 6.11939430e-01 3.01293522e-01
9.07414138e-01 -1.76078951e+00 6.33526385e-01 5.39623737e-01
6.98744357e-01 -5.66403449e-01 -4.26118493e-01 1.24750805e+00
5.15565574e-01 -4.52241302e-02 -2.41581216e-01 -2.86049455e-01
1.64457560e-01 5.90426385e-01 4.62366760e-01 5.44707119e-01
-4.46971178e-01 6.82811975e-01 -1.19950247e+00 -3.48277569e-01
-2.31550023e-01 -5.80304563e-01 -6.61774278e-01 -1.30331099e-01
-4.63461667e-01 2.51597196e-01 -6.89425111e-01 3.18901777e-01
6.26998186e-01 -1.37284532e-01 6.16947532e-01 9.05897319e-01
-5.73522091e-01 5.44160120e-02 -1.42739153e+00 1.86387169e+00
-8.62616450e-02 2.24334866e-01 8.00750017e-01 -7.99933374e-01
4.98605609e-01 1.82595626e-01 8.80655527e-01 -5.16709983e-01
-3.34047377e-02 4.59996313e-01 2.84286976e-01 5.22265211e-02
6.88168287e-01 4.40181851e-01 -4.85174537e-01 1.28028727e+00
-4.36366469e-01 -9.14380178e-02 5.30963659e-01 7.78992712e-01
1.88446915e+00 -4.43055242e-01 7.52920091e-01 -5.48659325e-01
5.60525417e-01 8.96812499e-01 6.53323710e-01 1.25562084e+00
-5.03807783e-01 -3.33512247e-01 5.00404418e-01 -3.98764104e-01
-5.31660795e-01 -9.42250729e-01 7.77935863e-01 1.03473175e+00
9.36422944e-01 -7.36255348e-01 -6.38983250e-01 -7.41966963e-01
-1.40712773e-02 4.04425979e-01 -4.35321838e-01 1.90675452e-01
-8.47376704e-01 -1.52618766e-01 -1.09417476e-01 2.71507222e-02
3.83293569e-01 -1.02549815e+00 -1.19986677e+00 7.10046828e-01
-3.08322102e-01 -9.73429620e-01 -6.36287928e-01 2.05030799e-01
-1.69063672e-01 -1.59201872e+00 -1.63759172e-01 -9.89468336e-01
5.95304489e-01 1.25472999e+00 1.02721202e+00 4.20858234e-01
4.86979857e-02 7.12734461e-01 -7.45327890e-01 -1.95947126e-01
-5.09899616e-01 2.45715454e-01 8.76408145e-02 -4.86570388e-01
-2.35451967e-01 -5.45137525e-01 -4.81433392e-01 7.39355206e-01
-4.32977498e-01 5.85940816e-02 2.51853466e-01 5.21711051e-01
4.86792237e-01 6.56370521e-01 5.92023373e-01 -4.14155692e-01
9.44485605e-01 -7.73495078e-01 -6.00347996e-01 2.13176936e-01
-4.76749301e-01 -4.81528133e-01 3.59757274e-01 -1.21489234e-01
-6.66306734e-01 1.33353561e-01 7.06722319e-01 1.27263948e-01
-2.90820003e-02 6.35971606e-01 -2.46755574e-02 -4.06618357e-01
3.34922075e-01 6.17061928e-03 1.72931999e-01 1.08527280e-01
1.78678215e-01 3.10132593e-01 5.59377909e-01 -6.45353615e-01
5.03271818e-01 5.08807480e-01 3.99497479e-01 -2.77662784e-01
2.68569291e-01 -6.30588233e-01 -4.74304497e-01 -4.58063006e-01
5.26571631e-01 -6.10083401e-01 -1.35953355e+00 4.23600525e-01
-1.50627017e+00 -9.65620995e-01 -4.63735834e-02 1.11634262e-01
-1.10730326e+00 3.64860237e-01 -2.88533777e-01 -9.69741404e-01
1.88448295e-01 -1.23914778e+00 6.82677984e-01 7.70066604e-02
-2.17158243e-01 -9.03856337e-01 4.83948261e-01 6.12062886e-02
4.29665893e-01 3.38945299e-01 6.60766482e-01 -6.67901039e-01
-9.26440060e-01 2.85783708e-01 8.95833299e-02 -7.81815290e-01
-8.20883177e-03 -3.91302288e-01 1.43402830e-01 -5.24968445e-01
-2.08143592e-01 -8.78418609e-02 1.23806410e-01 3.89976025e-01
6.61573783e-02 -3.38061601e-01 -1.09680092e+00 -2.72621363e-01
1.31237042e+00 7.42391586e-01 3.98095220e-01 8.64011943e-01
6.03892561e-03 1.09199464e+00 1.26655078e+00 6.25444472e-01
1.06602859e+00 1.15692306e+00 8.19256723e-01 3.24901849e-01
3.85953575e-01 4.56100076e-01 3.83369803e-01 2.91490585e-01
-1.93412140e-01 -6.83814764e-01 -1.19509888e+00 5.90779364e-01
-2.65849853e+00 -7.66801298e-01 -5.20692050e-01 1.78703868e+00
3.30083519e-01 3.03242039e-02 6.93972409e-01 2.41413102e-01
7.35344768e-01 -1.93957984e-01 -3.31495970e-01 -4.35319036e-01
7.95564502e-02 -4.42877859e-01 6.63537800e-01 8.56156528e-01
-8.09128344e-01 1.03550279e+00 6.37527037e+00 5.57454646e-01
-3.79654706e-01 4.46550429e-01 5.19191809e-02 -1.26308380e-02
-9.33383033e-03 2.53747165e-01 -1.88315794e-01 -2.34672450e-03
6.84892058e-01 -7.20248818e-01 9.07129705e-01 6.86632514e-01
5.26610374e-01 -8.05676818e-01 -1.00739920e+00 4.45594192e-01
-4.68674988e-01 -1.30189407e+00 -7.77795255e-01 5.04188240e-01
7.69957483e-01 5.08708134e-02 -4.90639508e-01 -2.68595777e-02
1.15526450e+00 -7.58672476e-01 8.83755088e-01 -6.37308955e-02
1.08675949e-01 -8.66752863e-01 4.88651842e-01 8.58377337e-01
-1.49865878e+00 -5.05417407e-01 -1.56880975e-01 -5.79497635e-01
8.37826669e-01 1.13111988e-01 -9.79045808e-01 1.05011809e+00
7.37859666e-01 3.71584356e-01 2.03051284e-01 1.19314647e+00
-1.00548705e-02 -3.20952326e-01 -3.48439723e-01 -1.64755359e-02
6.49613261e-01 -3.60932767e-01 9.87479150e-01 6.03930950e-01
1.59669787e-01 4.07864958e-01 9.37310696e-01 5.98261058e-01
5.37200451e-01 -3.80054004e-02 -5.75040817e-01 3.57309312e-01
9.46690619e-01 1.14099550e+00 -1.09562230e+00 -3.38980444e-02
-2.88831294e-01 9.41863656e-01 3.89793307e-01 2.53367007e-01
-7.68634915e-01 -2.30098963e-01 8.35848451e-01 4.29397486e-02
-9.61586386e-02 -7.59002447e-01 -1.08758077e-01 -3.09413075e-01
-1.33240834e-01 -7.97385156e-01 3.99275780e-01 -7.77552843e-01
-7.21190214e-01 5.72314680e-01 1.02042720e-01 -9.77956712e-01
-4.24107552e-01 1.11168928e-01 -8.35891485e-01 3.73688400e-01
-1.50972688e+00 -8.78233314e-01 -4.54124719e-01 8.83349299e-01
6.66918993e-01 -1.03246197e-01 6.77162945e-01 -2.46556789e-01
-5.10845721e-01 -1.57376051e-01 -1.28035575e-01 -6.90684497e-01
4.02292281e-01 -1.11678553e+00 4.40613180e-01 7.49553561e-01
-5.06258190e-01 3.16039920e-01 5.58972538e-01 -9.41559196e-01
-1.66683817e+00 -8.34152222e-01 8.73985708e-01 8.08722898e-02
6.58670068e-01 -2.65765041e-01 -3.39898199e-01 8.71046960e-01
4.37884122e-01 -1.64074972e-01 2.52924234e-01 -1.61985144e-01
1.53856605e-01 2.21844465e-01 -1.20937276e+00 6.49208784e-01
1.60769796e+00 2.48994783e-01 -2.11917043e-01 4.30838436e-01
9.26367760e-01 -2.59126067e-01 -5.48595369e-01 1.94721464e-02
8.12962055e-02 -1.16439521e+00 7.22178698e-01 -3.32111567e-01
-3.40748578e-02 -6.87439322e-01 -1.95596516e-01 -1.84051323e+00
-8.57209086e-01 -1.16796827e+00 3.33987087e-01 6.62588835e-01
1.88215747e-01 -1.15478778e+00 7.58880436e-01 4.43243146e-01
-5.74876308e-01 -4.69617218e-01 -1.29445660e+00 -1.20556378e+00
-2.17801612e-02 -1.21519551e-01 7.15066731e-01 7.73007751e-01
7.00168312e-01 1.07170418e-01 -1.39899001e-01 4.62670237e-01
6.69902682e-01 2.49213129e-02 1.14892697e+00 -1.20610690e+00
-4.94301200e-01 -3.88455689e-01 5.86400852e-02 -9.48339283e-01
3.03383470e-01 -7.37777948e-01 3.77145588e-01 -1.99919689e+00
8.60651359e-02 -1.14501727e+00 3.64060640e-01 7.47780085e-01
3.76604259e-01 -2.71311730e-01 6.58763051e-01 7.05067933e-01
-1.29557681e+00 1.47099465e-01 1.32920277e+00 -4.98748198e-02
-4.92199838e-01 -3.71123701e-02 -5.96962750e-01 3.77342135e-01
8.71940196e-01 -6.23544693e-01 -6.08458936e-01 -2.70354867e-01
2.48929232e-01 9.03436363e-01 1.04311623e-01 -7.38385558e-01
1.01683354e+00 -1.00171089e+00 -1.00642478e+00 -3.61881286e-01
2.58781791e-01 -1.12020648e+00 6.93569899e-01 1.04091084e+00
-9.15272832e-02 4.12542909e-01 -3.42319965e-01 7.97319055e-01
-1.32096842e-01 -1.83284760e-01 2.97414899e-01 -2.04741016e-01
-9.64596152e-01 5.68974242e-02 -1.20069802e+00 -3.18345755e-01
2.06869650e+00 -4.08711046e-01 -8.39830220e-01 -5.95629394e-01
-7.45189548e-01 1.06586254e+00 7.37344801e-01 -9.16812792e-02
4.54667896e-01 -9.83454645e-01 -7.81825185e-01 -3.31385761e-01
2.40147457e-01 1.77708298e-01 2.15507641e-01 1.28431773e+00
-5.76829553e-01 4.62376714e-01 -3.63472342e-01 -2.56150931e-01
-1.29122496e+00 7.33341277e-01 2.28498608e-01 -6.98938131e-01
-5.35267651e-01 3.14912081e-01 3.43016177e-01 -2.85275072e-01
6.16110936e-02 6.26330227e-02 -8.53098631e-02 -4.54874873e-01
3.94970834e-01 8.98696721e-01 -1.96266159e-01 -3.57712537e-01
-7.19229937e-01 1.65644541e-01 4.52273674e-02 -4.18367893e-01
1.33855426e+00 -7.28998780e-01 -4.42919642e-01 -3.24774981e-01
3.77281070e-01 -1.23069048e-01 -9.21234667e-01 -1.91378862e-01
2.32762173e-01 -4.14163560e-01 -1.52919963e-01 -7.71744609e-01
-8.12972248e-01 -1.90530494e-01 -3.98346364e-01 5.82399189e-01
1.00764775e+00 2.97612429e-01 4.40636456e-01 6.66611731e-01
1.45800233e+00 -1.02711713e+00 3.39328289e-01 8.06347966e-01
9.27530944e-01 -6.47422254e-01 -2.30973184e-01 -1.23234975e+00
-7.09480047e-01 9.08713520e-01 7.21792996e-01 -1.84964001e-01
3.20775509e-01 5.37797987e-01 -2.50728637e-01 -3.84255052e-01
-1.17759550e+00 -2.62159318e-01 -9.65366125e-01 1.06851947e+00
-6.97161615e-01 2.12175697e-01 -6.17137015e-01 5.66014051e-01
-6.19721748e-02 -1.70300245e-01 1.20107186e+00 1.62117350e+00
-8.68410170e-01 -1.47082722e+00 -6.46908224e-01 4.05907109e-02
4.20922428e-01 5.38268149e-01 -6.68753445e-01 8.59684229e-01
-3.09667885e-02 1.53576767e+00 2.08577558e-01 -3.42323095e-01
2.48840123e-01 -4.75052536e-01 3.99913400e-01 -6.84558868e-01
-6.67581081e-01 1.51314333e-01 4.73749667e-01 -9.75589633e-01
-6.23629689e-01 -8.31297457e-01 -1.65254128e+00 -7.61043906e-01
9.36070457e-03 3.90767306e-01 3.03362697e-01 7.97356784e-01
3.08564126e-01 4.42684740e-01 8.93361628e-01 -9.29245949e-01
-1.99775904e-01 -2.98087180e-01 -7.84666538e-01 -3.01001519e-01
7.22377300e-02 -8.93258452e-01 -1.82131499e-01 -4.84399766e-01] | [4.95358419418335, 1.7350457906723022] |
17253a12-c5ac-4174-9a6e-1e94549d3bcc | projective-view-at-optimization-problem-for | 1912.00197 | null | http://arxiv.org/abs/1912.00197v2 | http://arxiv.org/pdf/1912.00197v2.pdf | Projective view at Optimization Problem for Multiband Filter | The best uniform rational approximation of the \emph{sign} function on two
intervals separated by zero was explicitly solved by E.I. Zolotar\"ev in 1877.
This optimization problem is the initial step in the staircase of the so called
approximation problems for multiband filters which are of great importance for
electrical engineering. We show that known in the literature optimality
criterion for this problem may be contradictory since it does not take into
account the projective invariance of the problem. We propose a new consistently
projective formulation of this problem and give a constructive optimality
criterion for it. | [] | 2020-01-21 | null | null | null | null | ['electrical-engineering'] | ['miscellaneous'] | [ 2.20924601e-01 2.70281732e-01 1.26745120e-01 -2.81942278e-01
-5.31293809e-01 -6.88281178e-01 9.73264799e-02 -3.15098137e-01
-4.28880453e-01 1.34880018e+00 -2.15312704e-01 -4.10191834e-01
-6.18800044e-01 -5.16660511e-01 -5.45458794e-01 -9.05622303e-01
2.12184146e-01 2.02638134e-01 -1.80557191e-01 -6.69933796e-01
2.34728813e-01 6.91983998e-01 -1.13236511e+00 -2.95391008e-02
8.62596929e-01 1.19029903e+00 -1.47163659e-01 8.82459879e-01
1.34117931e-01 1.36272311e-01 -8.26131165e-01 -3.27594548e-01
4.56755370e-01 -7.04358876e-01 -1.00845993e+00 3.30483168e-01
1.09790482e-01 8.75882283e-02 -8.47077966e-02 1.59146225e+00
3.43724310e-01 1.12562869e-02 9.69804645e-01 -1.06559634e+00
-5.37893414e-01 4.15153801e-01 -4.73104149e-01 3.22803557e-01
2.76969165e-01 -5.64247787e-01 1.21083331e+00 -7.95294464e-01
4.21610892e-01 6.83940649e-01 8.73605669e-01 2.47935951e-01
-1.49514508e+00 -4.25873585e-02 -4.37189639e-01 4.13705677e-01
-1.66632283e+00 -2.77080745e-01 9.49999213e-01 -3.26903373e-01
3.74218702e-01 8.08817446e-01 7.77043223e-01 4.60368842e-01
3.49519968e-01 3.45704168e-01 1.34166694e+00 -6.91842437e-01
5.28223515e-02 1.70041129e-01 2.91356564e-01 7.29495049e-01
4.98861074e-01 -1.30603373e-01 2.61559784e-02 -1.05667472e-01
1.05011463e+00 -6.10973418e-01 -6.08750105e-01 -7.66809732e-02
-6.99639976e-01 7.04430521e-01 2.25309670e-01 6.52350605e-01
-2.48640433e-01 1.71134770e-01 -1.41599461e-01 5.53045154e-01
1.87487841e-01 6.87940180e-01 -1.22350499e-01 1.67789102e-01
-7.29507864e-01 1.07813843e-01 8.87995660e-01 1.01252234e+00
4.00689870e-01 4.83890027e-01 2.68206894e-01 6.20777249e-01
1.53965250e-01 5.44382930e-01 -1.43446162e-01 -9.91217256e-01
-7.42717739e-03 7.80870169e-02 3.51239771e-01 -7.51997828e-01
-6.75151527e-01 -7.34395921e-01 -6.85551226e-01 2.58208752e-01
8.34122300e-01 -3.30051154e-01 -2.46080264e-01 1.43605387e+00
1.67672426e-01 -1.80200562e-02 9.39053670e-02 8.26653719e-01
6.25727654e-01 7.25858450e-01 -7.32392371e-01 -8.59515846e-01
1.19547510e+00 -3.37791562e-01 -9.91926312e-01 6.17508948e-01
7.62866288e-02 -1.19726074e+00 6.57757044e-01 9.01490927e-01
-1.28175342e+00 -3.83322567e-01 -1.49959111e+00 1.94835857e-01
1.32984715e-02 3.06332141e-01 4.25445467e-01 1.02289093e+00
-1.12286043e+00 8.19221795e-01 -5.15975207e-02 9.85806659e-02
-1.99670270e-01 4.34614301e-01 -1.68605968e-01 5.46936035e-01
-1.00369453e+00 8.94745648e-01 2.50252366e-01 6.35406911e-01
1.81055069e-02 -3.05755883e-01 -3.12046140e-01 -6.36666566e-02
2.00897634e-01 -3.03161919e-01 1.11076951e+00 -1.16951561e+00
-1.81274152e+00 8.25587153e-01 -2.38307953e-01 -2.37842143e-01
7.58836269e-01 1.77316219e-02 -6.33185804e-01 1.75889000e-01
-3.05967122e-01 -2.07745552e-01 1.12208164e+00 -1.21080339e+00
-2.64479697e-01 -1.73674181e-01 7.06352592e-02 -8.02834406e-02
1.58395380e-01 -1.61036700e-01 1.82149872e-01 -7.10996389e-01
4.53189254e-01 -6.82149470e-01 -2.73825943e-01 -3.02906334e-01
-4.54529583e-01 -5.75708896e-02 3.39718789e-01 -6.95503891e-01
1.25416100e+00 -1.96214259e+00 2.94259489e-01 6.14724636e-01
3.54068577e-02 1.14899054e-01 3.41913670e-01 3.18595767e-01
-5.17922342e-01 1.13295995e-01 -1.73930243e-01 4.33963329e-01
-7.22842617e-03 5.71539402e-02 -2.33313143e-01 1.20626962e+00
-1.25714004e-01 2.59930491e-01 -4.32180941e-01 -3.12847435e-01
-4.20670249e-02 4.01413083e-01 -7.64891207e-01 -2.44105726e-01
4.84678783e-02 3.72355521e-01 -2.91077107e-01 2.02300638e-01
8.54702413e-01 -8.57414901e-02 1.33061200e-01 -4.91324514e-01
-6.28279626e-01 -5.38064688e-02 -1.59998882e+00 9.69355822e-01
-8.00126716e-02 8.46135736e-01 3.77142757e-01 -1.39143884e+00
9.24819827e-01 7.37046003e-01 9.07133460e-01 -3.11942458e-01
4.42907721e-01 3.69550735e-01 8.08321387e-02 -3.77602428e-01
6.04710221e-01 -5.69824100e-01 -5.68746217e-02 -1.38989523e-01
-1.23592168e-01 -5.34585953e-01 3.03985983e-01 -4.16092634e-01
5.43577135e-01 -2.68995643e-01 6.19972348e-01 -1.00917149e+00
1.01775885e+00 7.91861489e-02 5.54395914e-01 5.09266913e-01
-4.11177017e-02 7.67207503e-01 7.38582611e-01 -2.11345345e-01
-1.15442598e+00 -9.96613085e-01 -6.91020489e-01 2.14495197e-01
1.92082137e-01 -2.24549860e-01 -7.79050946e-01 1.24577411e-01
-3.09478492e-01 6.73653305e-01 -3.74088317e-01 2.57155925e-01
-7.71402240e-01 -8.13352704e-01 4.11019117e-01 6.13781512e-02
6.84535921e-01 -1.94552660e-01 -3.49401891e-01 2.80049980e-01
-1.62615135e-01 -9.36425507e-01 -1.97480544e-01 2.66688347e-01
-6.82693243e-01 -1.23178542e+00 -1.07350719e+00 -6.88782632e-01
7.09549069e-01 -1.78894594e-01 7.45446146e-01 -2.64629662e-01
-3.35099369e-01 3.77838194e-01 -3.23969796e-02 -3.60194564e-01
-4.36934769e-01 -3.03500473e-01 4.72394489e-02 7.61434203e-03
-3.34552318e-01 -4.92416888e-01 -4.46393669e-01 6.49527848e-01
-5.10497630e-01 -3.02355945e-01 1.14013940e-01 9.08060372e-01
7.15404332e-01 3.72135997e-01 6.21028841e-01 -5.42299390e-01
6.16628349e-01 1.17720619e-01 -1.19146132e+00 2.45106995e-01
-2.69606680e-01 3.40626746e-01 1.06460667e+00 5.57225607e-02
-8.47800016e-01 -2.73844432e-02 -7.07790434e-01 1.48389071e-01
2.47656062e-01 1.94720283e-01 -1.28749266e-01 -6.88692153e-01
6.34003699e-01 1.39553010e-01 -3.51761520e-01 -4.14920390e-01
1.38651147e-01 4.75435048e-01 7.82568097e-01 -6.13609970e-01
6.35417640e-01 3.26694995e-01 7.23361015e-01 -1.61915410e+00
-6.17102146e-01 -4.32820916e-01 -2.97143877e-01 -4.83240128e-01
7.59216845e-01 -1.92759365e-01 -1.09558535e+00 1.32617041e-01
-1.40631747e+00 2.24026278e-01 -6.69467449e-01 5.81770182e-01
-1.03458309e+00 3.87796700e-01 -3.01465362e-01 -8.48627508e-01
5.44384122e-02 -9.90977049e-01 3.37061465e-01 2.20569938e-01
-7.18033537e-02 -9.53449428e-01 1.69954058e-02 -1.07592516e-01
1.85534120e-01 3.30769181e-01 9.75497365e-01 -2.60472924e-01
-2.98065245e-01 8.09370168e-03 -5.76377548e-02 5.40635407e-01
-9.10137966e-02 2.52177641e-02 -7.30967402e-01 4.14751330e-03
6.28917754e-01 5.17235816e-01 5.20375669e-01 5.99311054e-01
8.60153139e-01 -4.26114231e-01 1.61331341e-01 7.69125700e-01
1.94558668e+00 2.90278167e-01 4.54609036e-01 1.05588436e-02
1.75859615e-01 1.32445797e-01 3.93168122e-01 4.22421813e-01
-3.26982528e-01 8.70531321e-01 2.46549264e-01 1.52470704e-04
1.55966997e-01 3.67497236e-01 8.09236243e-02 7.68255174e-01
-6.00072503e-01 -7.32010752e-02 -5.00040650e-01 1.96214259e-01
-1.58613074e+00 -1.11338699e+00 -7.09960759e-01 2.34113026e+00
6.01747632e-01 2.23897725e-01 1.30079716e-01 7.74725676e-01
8.42120707e-01 -1.19351950e-02 -3.64505723e-02 -9.32533205e-01
-4.77247149e-01 4.42103684e-01 9.11534190e-01 9.54456508e-01
-9.79477346e-01 2.98909783e-01 7.89235163e+00 8.34263682e-01
-9.52283680e-01 -3.44807729e-02 1.12539209e-01 3.05321038e-01
-3.70582253e-01 -5.00587039e-02 -7.29678631e-01 1.80900604e-01
5.82132936e-01 -6.25747263e-01 3.68847221e-01 5.27019799e-01
1.19649522e-01 -1.69308603e-01 -7.21288800e-01 1.07575798e+00
-6.77068764e-03 -1.19162273e+00 -3.30576420e-01 -8.33212119e-03
1.03117847e+00 -7.47024715e-01 2.17078671e-01 -2.92793870e-01
-2.77138680e-01 -1.00043213e+00 7.03317165e-01 5.44380844e-01
5.12728870e-01 -9.80829179e-01 4.33548242e-01 1.82360142e-01
-1.27104986e+00 1.42654419e-01 -3.46729487e-01 -1.15915783e-01
3.71936083e-01 6.80272639e-01 -6.24898255e-01 8.00161302e-01
-1.12080663e-01 2.07563803e-01 9.79548171e-02 1.50781131e+00
-3.04013163e-01 5.56468964e-01 -6.16138756e-01 -3.60751987e-01
3.83360535e-01 -9.08751607e-01 1.03329933e+00 8.66505384e-01
5.52241027e-01 5.56098819e-01 -1.64811298e-01 6.05021834e-01
1.90794379e-01 3.42044324e-01 -3.38225037e-01 1.49758160e-01
-1.43970132e-01 6.27838016e-01 -7.93697417e-01 -2.14443039e-02
-3.34605277e-01 8.13895524e-01 -4.54020023e-01 5.25701821e-01
-9.43144202e-01 -6.79715037e-01 5.62259436e-01 1.87438250e-01
2.46537834e-01 -3.74548376e-01 -6.53390229e-01 -9.33310568e-01
2.58000549e-02 -3.91469985e-01 2.97839075e-01 -4.21117544e-01
-6.66132450e-01 6.44716144e-01 2.29269415e-02 -1.26689923e+00
-2.10735098e-01 -9.64364409e-01 -2.30401829e-01 1.03624022e+00
-1.15686548e+00 -5.01925170e-01 3.74973089e-01 6.14094615e-01
3.55493754e-01 7.22474083e-02 6.94889247e-01 3.81341070e-01
-4.38583344e-01 4.32609528e-01 7.47666836e-01 -2.84355134e-01
2.03244850e-01 -1.51855516e+00 -2.78856307e-01 1.02299535e+00
-8.16008449e-02 3.41218442e-01 1.62196147e+00 -1.86589688e-01
-1.36933792e+00 -3.17397743e-01 1.10575867e+00 2.46283352e-01
6.77977562e-01 1.64006025e-01 -5.02496660e-01 4.87957716e-01
3.38527471e-01 -1.88632756e-01 5.15433192e-01 -1.76038802e-01
1.80303514e-01 -2.32764259e-01 -1.08194053e+00 4.43804681e-01
7.32787430e-01 -1.57099545e-01 -5.76808870e-01 4.93816644e-01
2.38466263e-01 -4.30654049e-01 -9.58716869e-01 2.43185580e-01
6.05668545e-01 -1.04264355e+00 9.99799490e-01 -4.45586622e-01
2.16556996e-01 -3.64670604e-01 -3.31830859e-01 -1.24167275e+00
-3.02640229e-01 -1.19060969e+00 5.23384750e-01 6.60307586e-01
4.46477205e-01 -9.38260734e-01 6.82470262e-01 6.99408650e-02
-3.83389503e-01 -9.28999841e-01 -1.38290012e+00 -1.03436744e+00
8.40962529e-02 -2.57269114e-01 2.28457004e-01 7.33254135e-01
3.91180515e-01 7.08490312e-02 -5.29151618e-01 3.09212148e-01
7.61751771e-01 1.46086231e-01 2.14782029e-01 -1.18827665e+00
-6.20537698e-01 -8.60050380e-01 -4.79742348e-01 -1.24104941e+00
-1.19111635e-01 -7.89544821e-01 -2.72582322e-02 -1.42079651e+00
-5.77084184e-01 -4.02952135e-01 -1.57657847e-01 -3.54078054e-01
2.86706120e-01 2.59253085e-01 -3.37650739e-02 -1.38859093e-01
-9.51123238e-03 3.67461354e-01 1.23390996e+00 -3.69806029e-02
-1.51978076e-01 5.79117775e-01 -5.17605960e-01 1.06244957e+00
7.63238788e-01 -2.09037974e-01 -2.74462491e-01 -1.35561645e-01
5.18088281e-01 3.82436395e-01 1.31579503e-01 -1.03883827e+00
1.36050403e-01 -5.22478633e-02 8.26196969e-02 -6.14679754e-01
5.88339031e-01 -1.14549685e+00 4.73977536e-01 4.50332582e-01
-4.32064161e-02 2.26032764e-01 -6.58958256e-02 2.17346221e-01
-1.49777785e-01 -7.10278749e-01 1.18095362e+00 6.75606579e-02
-5.20624697e-01 -2.83335090e-01 -3.83013844e-01 -7.68689215e-02
1.00884509e+00 -1.90316662e-01 -2.75976449e-01 -4.51311409e-01
-9.96320426e-01 -8.54932517e-02 1.66834876e-01 -4.99323994e-01
4.38571990e-01 -1.36691833e+00 -5.92319608e-01 1.21941186e-01
-4.53898728e-01 -5.67838132e-01 2.17202067e-01 1.14465725e+00
-1.04515600e+00 5.36048234e-01 -3.82314712e-01 -5.80195308e-01
-1.42955208e+00 3.61321688e-01 6.37030244e-01 -3.89302373e-02
-3.75201851e-01 7.86198616e-01 -2.24478140e-01 4.09553200e-01
-1.50109336e-01 -4.56045777e-01 -3.01599503e-01 5.89196794e-02
3.70609492e-01 7.40072846e-01 -8.76063183e-02 -8.66078496e-01
-1.86228290e-01 9.17978048e-01 4.94208962e-01 -5.34331203e-01
1.09262502e+00 -2.95835510e-02 -3.49797070e-01 5.12928784e-01
1.36459351e+00 2.70811051e-01 -7.35322714e-01 3.65821598e-03
-3.01236868e-01 -3.32666934e-01 -1.84238032e-01 -3.16539556e-01
-6.86369181e-01 4.71675038e-01 2.10304588e-01 7.35315919e-01
1.32691288e+00 -3.90740454e-01 4.76041526e-01 5.10926425e-01
3.57488632e-01 -1.33128774e+00 -3.51110607e-01 4.74715531e-01
1.25222218e+00 -4.49896395e-01 3.17004174e-01 -9.86631334e-01
-1.10567242e-01 1.51655626e+00 -9.72766355e-02 -5.77983201e-01
9.37676489e-01 2.04072505e-01 -3.29075217e-01 2.29295537e-01
-1.11627668e-01 -2.49572054e-01 4.10577148e-01 3.12823206e-01
5.43073475e-01 6.19031601e-02 -1.34604454e+00 5.63778996e-01
-5.10193825e-01 -2.13447571e-01 8.01963568e-01 7.59883404e-01
-6.94401205e-01 -1.07205725e+00 -8.89501929e-01 8.95072743e-02
-5.29546082e-01 1.61799729e-01 -1.26183331e-01 7.26975381e-01
1.72097743e-01 8.20551217e-01 -4.25185889e-01 -1.16225608e-01
6.08444333e-01 -1.87409203e-02 1.09277105e+00 9.97381881e-02
-3.60228837e-01 3.75758231e-01 2.48380959e-01 -2.55776942e-01
-2.67193943e-01 -6.72977388e-01 -1.07066381e+00 -4.20138717e-01
-2.98938334e-01 7.25253463e-01 6.86531007e-01 8.35683525e-01
-4.61866140e-01 6.85980499e-01 7.71330535e-01 -4.55090731e-01
-7.37383127e-01 -4.25014526e-01 -1.01754868e+00 -1.71272144e-01
5.37836492e-01 -3.20025176e-01 -3.13946486e-01 -2.14694981e-02] | [6.301736831665039, 3.338390350341797] |
d0fdd500-33c3-41ca-a44b-2a538dfaf80e | faster-segment-anything-towards-lightweight | 2306.14289 | null | https://arxiv.org/abs/2306.14289v2 | https://arxiv.org/pdf/2306.14289v2.pdf | Faster Segment Anything: Towards Lightweight SAM for Mobile Applications | Segment Anything Model (SAM) has attracted significant attention due to its impressive zero-shot transfer performance and high versatility for numerous vision applications (like image editing with fine-grained control). Many of such applications need to be run on resource-constraint edge devices, like mobile phones. In this work, we aim to make SAM mobile-friendly by replacing the heavyweight image encoder with a lightweight one. A naive way to train such a new SAM as in the original SAM paper leads to unsatisfactory performance, especially when limited training sources are available. We find that this is mainly caused by the coupled optimization of the image encoder and mask decoder, motivated by which we propose decoupled distillation. Concretely, we distill the knowledge from the heavy image encoder (ViT-H in the original SAM) to a lightweight image encoder, which can be automatically compatible with the mask decoder in the original SAM. The training can be completed on a single GPU within less than one day, and the resulting lightweight SAM is termed MobileSAM which is more than 60 times smaller yet performs on par with the original SAM. For inference speed, With a single GPU, MobileSAM runs around 10ms per image: 8ms on the image encoder and 4ms on the mask decoder. With superior performance, our MobileSAM is around 5 times faster than the concurrent FastSAM and 7 times smaller, making it more suitable for mobile applications. Moreover, we show that MobileSAM can run relatively smoothly on CPU. The code for our project is provided at \href{https://github.com/ChaoningZhang/MobileSAM}{\textcolor{red}{MobileSAM}}), with a demo showing that MobileSAM can run relatively smoothly on CPU. | ['Choong Seon Hong', 'Seungkyu Lee', 'Sung-Ho Bae', 'Jung Uk Kim', 'Yu Qiao', 'Dongshen Han', 'Chaoning Zhang'] | 2023-06-25 | null | null | null | null | ['panoptic-segmentation', 'instance-segmentation'] | ['computer-vision', 'computer-vision'] | [ 3.81659687e-01 1.20127663e-01 -1.42710894e-01 6.12707436e-03
-9.06329513e-01 -6.05545521e-01 4.13358301e-01 -3.30438763e-01
-6.43315136e-01 4.75133538e-01 -1.57541960e-01 -6.32361472e-01
3.35526139e-01 -6.80955946e-01 -1.02562833e+00 -7.58152485e-01
3.66633147e-01 4.06796157e-01 3.04133207e-01 9.47160125e-02
2.37534824e-03 9.18755159e-02 -1.49448156e+00 5.59757277e-02
8.81440222e-01 8.47771823e-01 8.13407660e-01 1.05931091e+00
2.18839664e-02 5.80921292e-01 -2.76688993e-01 -4.68332648e-01
2.40453258e-01 -4.33522224e-01 -8.05903912e-01 2.23018110e-01
3.73765737e-01 -4.56096798e-01 -4.37221110e-01 1.10155237e+00
4.78172719e-01 1.92385539e-02 9.25039575e-02 -1.16962612e+00
-2.41875201e-01 4.78240788e-01 -7.77379572e-01 9.87276062e-03
-1.45871192e-01 1.70311943e-01 8.68144393e-01 -8.84799719e-01
5.96591532e-01 7.85173476e-01 5.48286378e-01 4.22695309e-01
-1.06070745e+00 -5.29018223e-01 1.20133936e-01 1.24074794e-01
-1.37833905e+00 -6.43289149e-01 1.81488723e-01 -2.30299830e-01
7.65763938e-01 4.83332038e-01 6.68085098e-01 8.21365118e-01
-1.30736023e-01 9.60781336e-01 9.43938613e-01 -3.52778435e-01
1.42853439e-01 1.13006741e-01 2.35335361e-02 1.07238042e+00
1.00653492e-01 -2.02003717e-01 -3.44571233e-01 3.87011208e-02
9.13863003e-01 -7.85697103e-02 -3.82396668e-01 -8.52891430e-02
-1.42694950e+00 7.02572882e-01 4.71701592e-01 1.39350027e-01
-9.40952227e-02 3.61719012e-01 3.55805248e-01 2.15867475e-01
3.44259828e-01 1.45224086e-03 -4.33782339e-01 -4.50027734e-01
-1.06959271e+00 2.21597869e-02 8.16971958e-01 1.11745775e+00
9.19035435e-01 -1.12620927e-01 -6.11014338e-03 7.53329575e-01
-5.34181856e-02 6.08865857e-01 2.91875690e-01 -1.15173113e+00
3.31925839e-01 7.87067413e-02 -3.07813310e-03 -6.77449465e-01
-1.99699998e-01 -3.71151805e-01 -1.05632555e+00 2.35724524e-02
3.68700206e-01 -3.06711197e-01 -8.19703519e-01 1.71301305e+00
3.75400484e-01 5.95657825e-01 -2.33851552e-01 8.75508249e-01
5.20446002e-01 1.08304513e+00 -2.79478222e-01 -8.60081539e-02
1.55674481e+00 -1.42259383e+00 -2.31598511e-01 -5.97776890e-01
8.51255476e-01 -7.81737745e-01 1.30009520e+00 3.69007736e-01
-1.21745324e+00 -5.39827824e-01 -1.01915431e+00 -3.22576940e-01
-6.26716530e-03 4.03034329e-01 7.29189575e-01 4.48366225e-01
-1.24770236e+00 5.45182467e-01 -1.09786713e+00 -2.92207390e-01
4.06327665e-01 4.72685575e-01 -3.61803532e-01 -2.54403442e-01
-7.40097106e-01 7.34219491e-01 3.80200386e-01 7.30291754e-02
-5.87077916e-01 -5.02888918e-01 -9.01711583e-01 2.33603716e-02
6.12246335e-01 -9.91542339e-01 1.17066860e+00 -1.03442752e+00
-1.71217692e+00 9.00207758e-01 -3.26211333e-01 -3.41130376e-01
6.74237907e-01 -2.27255329e-01 1.04249366e-01 1.00415237e-01
2.52276827e-02 8.44243467e-01 1.03695452e+00 -9.07447755e-01
-5.96522093e-01 -1.59397766e-01 2.52873480e-01 2.94605404e-01
-3.98813933e-01 -3.53428796e-02 -1.37988865e+00 -5.99414051e-01
-1.67163000e-01 -1.44558406e+00 -1.84948713e-01 1.08277284e-01
-3.63067299e-01 1.12484962e-01 7.23950624e-01 -7.47535765e-01
1.07739973e+00 -2.36076450e+00 3.12625021e-01 -6.43904135e-03
2.34051153e-01 5.60122848e-01 -1.05915606e-01 9.98378918e-02
1.26932174e-01 -2.03092307e-01 -5.52865207e-01 -5.83458602e-01
1.67617276e-02 4.18168396e-01 -8.89387429e-02 6.13682806e-01
-1.50967449e-01 1.04443491e+00 -8.14864337e-01 -4.80414361e-01
3.04252386e-01 5.57486773e-01 -7.26452053e-01 2.57925212e-01
-1.18621036e-01 4.79879200e-01 -2.19489709e-01 3.89483690e-01
8.35939527e-01 -6.70458317e-01 2.28705287e-01 -1.87882915e-01
-1.90777183e-01 1.16773702e-01 -1.10365689e+00 1.91395473e+00
-7.48876333e-01 7.41453528e-01 3.50010693e-01 -9.67212021e-01
3.27844918e-01 1.09675840e-01 1.71660677e-01 -4.94260609e-01
1.81654051e-01 1.98332697e-01 -1.67459249e-01 -2.34885588e-01
4.92715210e-01 1.37607723e-01 3.72208208e-02 6.18436694e-01
-6.10556193e-02 -1.09580800e-01 2.60087371e-01 3.78398448e-01
9.88996387e-01 2.73649663e-01 5.32475673e-02 -1.68640882e-01
3.64985108e-01 -1.33833155e-01 4.88479286e-01 6.87533379e-01
5.26369549e-02 6.14177525e-01 4.09783304e-01 -1.93384945e-01
-1.13097751e+00 -9.74201918e-01 -5.04496321e-02 1.12921703e+00
3.18226784e-01 -6.54207885e-01 -1.13903296e+00 -4.74982142e-01
-4.47733164e-01 3.89547646e-01 -2.63265699e-01 -3.07766926e-02
-5.65463185e-01 -8.09543371e-01 5.50502956e-01 5.42733550e-01
8.38692248e-01 -7.90265501e-01 -5.84371924e-01 7.50607252e-02
-1.20334774e-01 -1.21343040e+00 -6.49986386e-01 7.32673034e-02
-6.84865534e-01 -8.58877599e-01 -9.41995978e-01 -7.90870011e-01
8.35957944e-01 4.95982170e-01 9.68478262e-01 4.49177116e-01
-2.45568544e-01 1.50489351e-02 -8.23163167e-02 -3.40595767e-02
-2.12615758e-01 3.43243808e-01 -1.39083251e-01 -6.14438877e-02
-1.72734722e-01 -4.84426737e-01 -6.76887035e-01 2.58674234e-01
-1.03129876e+00 8.29316556e-01 6.28993750e-01 9.85705554e-01
6.87117100e-01 -4.67027649e-02 1.39341831e-01 -1.08029366e+00
-9.52379219e-03 -4.25187767e-01 -8.65351498e-01 1.35328308e-01
-3.60680133e-01 -2.50229100e-03 6.16048574e-01 -2.98389971e-01
-9.45637941e-01 2.49366164e-01 -5.41262686e-01 -3.22155893e-01
1.33718610e-01 3.05678755e-01 -1.65256009e-01 -8.63915831e-02
2.21668690e-01 3.31985056e-01 1.82513557e-02 -4.21422184e-01
5.83505809e-01 7.32509851e-01 6.95052087e-01 -5.16349912e-01
7.39541352e-01 4.44851577e-01 -2.97464669e-01 -8.75119150e-01
-7.69415319e-01 -2.65598476e-01 -3.71396542e-01 -1.98489521e-02
8.29157412e-01 -1.05392396e+00 -8.07002246e-01 6.80923641e-01
-1.05461526e+00 -8.54102790e-01 4.60661529e-03 3.38053823e-01
-6.16275609e-01 3.01482588e-01 -8.80905688e-01 -3.52915436e-01
-4.56076115e-01 -1.44060838e+00 1.22387207e+00 2.52981454e-01
-1.05315074e-01 -9.23790038e-01 -3.41084152e-01 5.48158169e-01
3.47377747e-01 -1.85436264e-01 6.59912288e-01 -3.54092009e-02
-7.45019078e-01 -2.65202969e-02 -5.78370035e-01 3.28751862e-01
-1.26168817e-01 -1.53771535e-01 -1.04408360e+00 -6.16346717e-01
-5.98442182e-02 -1.56757906e-01 9.64386523e-01 2.84280479e-01
1.48246348e+00 -2.00596303e-01 -4.02783364e-01 1.00192940e+00
1.48184931e+00 -5.98850772e-02 7.21843898e-01 2.54159182e-01
1.00030744e+00 9.02323574e-02 4.61299866e-01 2.56069243e-01
4.95677501e-01 8.32360148e-01 2.42242306e-01 -3.74500036e-01
-1.56698093e-01 -6.50264472e-02 6.17819011e-01 1.16755879e+00
-1.67745635e-01 -2.88534731e-01 -7.54799128e-01 3.23632687e-01
-2.02362657e+00 -6.70194089e-01 -1.52107731e-01 2.33744121e+00
8.10165465e-01 2.84935515e-02 -5.13357036e-02 -1.35017082e-01
6.42698467e-01 2.28211626e-01 -5.40657282e-01 -3.28692734e-01
7.94113949e-02 3.28557581e-01 8.28341961e-01 7.31646061e-01
-1.07134008e+00 1.03043354e+00 5.43259764e+00 1.06984758e+00
-1.12916696e+00 4.06281143e-01 8.28650773e-01 -1.98171526e-01
-1.74820706e-01 1.61648184e-01 -7.06934214e-01 8.01574349e-01
1.04830050e+00 -1.07150316e-01 6.17739797e-01 8.07061136e-01
1.06456660e-01 -2.95459926e-01 -9.99317884e-01 1.21867621e+00
1.16801364e-02 -1.44929075e+00 -2.48927623e-01 2.27793172e-01
6.59637272e-01 2.25013897e-01 -4.22732756e-02 2.31988072e-01
1.14695579e-01 -8.05539012e-01 6.75956845e-01 1.72791287e-01
1.09108829e+00 -7.39469290e-01 6.08193457e-01 5.73756218e-01
-1.25988960e+00 2.94248998e-01 -4.89340246e-01 -1.13180587e-02
2.36535400e-01 6.46809816e-01 -4.89303321e-01 4.21581924e-01
6.18219912e-01 5.91488600e-01 -3.58754843e-01 7.50206649e-01
-2.11462006e-01 5.16837597e-01 -3.63845497e-01 3.67858708e-01
2.00051546e-01 -3.81460041e-01 4.20434207e-01 1.20057511e+00
6.41635299e-01 2.18798984e-02 2.70263940e-01 5.24148583e-01
-3.64402056e-01 -1.31280869e-01 -5.26504099e-01 -4.10954766e-02
3.32404524e-01 1.42943549e+00 -9.32490706e-01 -6.63909018e-01
-6.19455040e-01 1.66234291e+00 4.16457921e-01 1.88073009e-01
-1.24291754e+00 -3.36062819e-01 6.65028214e-01 -8.89448449e-02
4.73730415e-01 -3.91185224e-01 -2.03508258e-01 -1.41020966e+00
1.01159833e-01 -8.60462189e-01 1.65649876e-01 -7.93430924e-01
-7.76331365e-01 6.29341960e-01 -2.97558188e-01 -9.83139813e-01
-9.40906182e-02 -5.40873706e-01 -4.71337587e-01 8.27461481e-01
-1.27858770e+00 -9.91488516e-01 -4.31077242e-01 6.25239253e-01
6.46794081e-01 1.84230864e-01 8.80368769e-01 5.89433193e-01
-8.08104694e-01 6.81474805e-01 1.25835106e-01 3.15846154e-03
6.25532091e-01 -1.10689080e+00 8.56412649e-01 1.05019855e+00
3.00243795e-01 5.26186824e-01 4.84130770e-01 -3.87798637e-01
-1.82154715e+00 -1.19792068e+00 7.40983427e-01 -2.24005669e-01
6.04442656e-01 -4.70543802e-01 -9.08746362e-01 7.47989416e-01
2.56143630e-01 1.83873743e-01 4.22741860e-01 -2.25926697e-01
-1.50512725e-01 1.92198381e-02 -7.12954998e-01 7.65742660e-01
1.13924694e+00 -5.56590199e-01 -1.11396872e-01 5.23367763e-01
6.63640440e-01 -8.09492588e-01 -4.71174270e-01 3.80319133e-02
4.24491674e-01 -8.95786583e-01 8.36650252e-01 8.82935300e-02
3.25984240e-01 -6.07138872e-01 -2.76584458e-02 -1.14875114e+00
-1.79730073e-01 -7.94193864e-01 -3.62331957e-01 9.58485842e-01
2.70560950e-01 -6.71271205e-01 8.17688525e-01 3.84485692e-01
-1.18388854e-01 -8.30890834e-01 -7.04386473e-01 -6.56882346e-01
-1.90143734e-01 -6.06738567e-01 3.47269833e-01 7.19923675e-01
-2.97370911e-01 4.84284729e-01 -5.45338154e-01 1.61649778e-01
4.34181392e-01 2.39724696e-01 8.82260621e-01 -7.95439839e-01
-8.53100121e-01 -2.22071871e-01 -1.11198746e-01 -1.61658144e+00
1.21217771e-02 -1.01494849e+00 1.47479519e-01 -1.46825588e+00
5.09782195e-01 -7.03635812e-01 3.48029658e-02 7.37769783e-01
-3.43040377e-01 7.45557845e-01 4.06538695e-01 2.12149337e-01
-5.68130136e-01 2.58755058e-01 1.16958988e+00 -4.64650281e-02
7.45278150e-02 -1.47171011e-02 -7.50987053e-01 7.86633372e-01
6.64213359e-01 -2.55434781e-01 -4.38671649e-01 -7.94941664e-01
3.35547179e-01 1.30398288e-01 3.71271908e-01 -9.58194911e-01
3.10172528e-01 2.36048177e-01 7.98209235e-02 -1.81517020e-01
4.67702597e-01 -5.28137267e-01 3.42671365e-01 4.37199682e-01
1.10335171e-01 -2.14622561e-02 2.43718415e-01 2.76065856e-01
-4.46626879e-02 -2.57951677e-01 7.98716307e-01 -1.27805158e-01
-7.86008775e-01 5.34575284e-01 -3.17217022e-01 -7.78091624e-02
9.97191370e-01 -9.73416939e-02 -3.60060126e-01 -3.41627657e-01
-6.53347909e-01 1.38513252e-01 7.83960104e-01 1.56641811e-01
3.31204772e-01 -1.09602070e+00 -5.02971351e-01 2.03014389e-01
-1.56670362e-01 2.85795808e-01 4.48555559e-01 1.18398404e+00
-7.47255802e-01 1.28987059e-01 8.20292607e-02 -6.55763447e-01
-1.37881100e+00 4.02165323e-01 1.76743925e-01 -8.15689489e-02
-1.01139331e+00 1.06007671e+00 4.97301012e-01 -1.49868071e-01
6.87578991e-02 -2.16765210e-01 4.56653357e-01 -1.60311520e-01
7.96909809e-01 3.70328635e-01 1.24777913e-01 -5.20520151e-01
-2.37296447e-01 4.99821663e-01 -1.32921904e-01 -1.25994936e-01
1.30847800e+00 -1.64910689e-01 -2.60222465e-01 1.78587511e-01
1.35664856e+00 6.62169158e-02 -1.58122516e+00 -5.13695776e-02
-3.08908015e-01 -5.11721551e-01 2.28940442e-01 -3.45347881e-01
-1.25605810e+00 8.69526088e-01 4.63875592e-01 -1.13437707e-02
1.20861745e+00 -2.76670102e-02 1.25301397e+00 2.93528795e-01
5.05699456e-01 -9.19400632e-01 -2.93630153e-01 5.83693087e-01
4.81380224e-01 -1.23335969e+00 -7.49632120e-02 -4.55979377e-01
-6.19007766e-01 8.98007035e-01 3.06863099e-01 -4.99664657e-02
3.37027192e-01 5.64197719e-01 -1.14863388e-01 1.21897765e-01
-6.35747910e-01 -1.19909763e-01 5.80382124e-02 2.74396330e-01
2.39005879e-01 1.28238127e-01 -9.36986506e-02 3.82798433e-01
-2.80104131e-01 7.38880038e-02 4.55099255e-01 8.95598352e-01
-3.00469488e-01 -1.39897966e+00 -1.76393256e-01 4.59168941e-01
-3.59780639e-01 -3.85451794e-01 1.26688212e-01 3.86057049e-01
2.95139343e-01 7.32191622e-01 2.46596932e-01 -3.15322161e-01
-4.71776128e-02 -1.91893548e-01 5.71866274e-01 -5.67235589e-01
-3.37697238e-01 9.67244506e-02 8.98077115e-02 -7.70748556e-01
-2.05692545e-01 -4.53997016e-01 -1.33008635e+00 -6.66254163e-01
-1.79337159e-01 -2.46646553e-02 6.86533093e-01 9.90573287e-01
4.21863317e-01 4.25757170e-01 4.10519898e-01 -1.13032675e+00
1.27471229e-02 -5.44912994e-01 -3.80517066e-01 -2.61096843e-02
1.16880611e-01 -3.56979966e-01 -8.56582522e-02 2.22786099e-01] | [9.769163131713867, 0.10285787284374237] |
b79c13c9-238f-4825-be8f-2fa2f2b0f7c2 | architecture-agnostic-masked-image-modeling | 2205.13943 | null | https://arxiv.org/abs/2205.13943v4 | https://arxiv.org/pdf/2205.13943v4.pdf | Architecture-Agnostic Masked Image Modeling -- From ViT back to CNN | Masked image modeling, an emerging self-supervised pre-training method, has shown impressive success across numerous downstream vision tasks with Vision transformers. Its underlying idea is simple: a portion of the input image is masked out and then reconstructed via a pre-text task. However, the working principle behind MIM is not well explained, and previous studies insist that MIM primarily works for the Transformer family but is incompatible with CNNs. In this work, we observe that MIM essentially teaches the model to learn better middle-order interactions among patches for more generalized feature extraction. We then propose an Architecture-Agnostic Masked Image Modeling framework (A$^2$MIM), which is compatible with both Transformers and CNNs in a unified way. Extensive experiments on popular benchmarks show that A$^2$MIM learns better representations without explicit design and endows the backbone model with the stronger capability to transfer to various downstream tasks. | ['Stan. Z. Li', 'Zelin Zang', 'Fang Wu', 'Di wu', 'Siyuan Li'] | 2022-05-27 | null | null | null | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 5.01820147e-01 5.60061753e-01 -2.33370647e-01 -3.42780650e-01
-3.75791818e-01 -4.09568042e-01 7.28094220e-01 -5.26081860e-01
-6.73411461e-03 2.40228966e-01 2.70465761e-01 -4.00768787e-01
5.27413338e-02 -6.98135614e-01 -1.16066730e+00 -7.99471796e-01
2.39558756e-01 5.48411235e-02 3.20591420e-01 -4.06963795e-01
-4.16267253e-02 1.03776239e-01 -1.58772635e+00 6.90753281e-01
6.87106907e-01 1.02237892e+00 3.64546984e-01 2.94473112e-01
-5.64996265e-02 1.33403516e+00 -2.98511416e-01 -6.59486353e-01
5.59030950e-01 -3.77797306e-01 -8.23122978e-01 3.60549301e-01
8.34786057e-01 -2.86733836e-01 -7.08270252e-01 9.27044451e-01
1.22519501e-01 -3.22705954e-01 5.27388096e-01 -1.04438198e+00
-1.19500983e+00 9.66067016e-01 -5.33996165e-01 9.42478403e-02
-4.55939084e-01 5.40601134e-01 1.15554261e+00 -1.07098162e+00
3.33682477e-01 9.94627416e-01 7.73083150e-01 8.15718472e-01
-1.49379301e+00 -4.93186742e-01 6.11579716e-01 1.21907391e-01
-1.13660061e+00 -3.75643760e-01 6.08792245e-01 -5.61188340e-01
1.01727152e+00 6.82837740e-02 5.06005824e-01 1.24282598e+00
4.43351530e-02 9.74654019e-01 1.19798899e+00 -2.90793091e-01
-6.00187369e-02 1.91325694e-01 2.05675036e-01 9.18375194e-01
2.00725645e-01 3.21685493e-01 -8.59496176e-01 3.51752788e-01
8.53504717e-01 1.70924559e-01 -3.25952649e-01 -4.18831319e-01
-1.18020296e+00 6.94437861e-01 9.72785771e-01 1.45001680e-01
-1.59967870e-01 2.53391325e-01 9.01009440e-02 5.22032976e-01
4.20459896e-01 2.84167469e-01 -4.49160755e-01 5.58803916e-01
-9.44263697e-01 6.79935366e-02 4.06629086e-01 9.83760357e-01
1.09234440e+00 3.42177570e-01 -3.27959478e-01 5.31429708e-01
2.15496421e-01 1.63093194e-01 4.92039204e-01 -9.17443991e-01
2.23307312e-01 7.90721178e-01 -2.03404397e-01 -6.30809963e-01
-2.03373924e-01 -7.76417196e-01 -1.07780254e+00 5.84191382e-01
3.99912566e-01 1.77410483e-01 -1.33749354e+00 2.06505775e+00
-1.84726492e-01 1.50845870e-01 6.62747324e-02 6.83912098e-01
1.01005685e+00 3.25746417e-01 3.13561000e-02 2.86164820e-01
1.44279623e+00 -1.55316365e+00 -2.12986141e-01 -7.64874935e-01
8.70494321e-02 -4.36569989e-01 1.24452364e+00 4.79110360e-01
-1.38366365e+00 -9.90755916e-01 -1.21859717e+00 -3.55480790e-01
-3.67345423e-01 1.37402996e-01 9.51194465e-01 4.99023527e-01
-1.65755022e+00 5.51852584e-01 -6.41716480e-01 -2.30940849e-01
8.87843132e-01 5.11236250e-01 -3.90646458e-01 -1.83249846e-01
-7.03851998e-01 9.33610260e-01 -2.16387212e-03 1.06753208e-01
-1.35972536e+00 -8.09183300e-01 -8.70239496e-01 6.27743974e-02
6.53482080e-02 -1.24603486e+00 1.33305740e+00 -1.37712991e+00
-1.42851269e+00 1.23790908e+00 -3.72183263e-01 -8.01068604e-01
4.03186888e-01 -7.59285912e-02 -9.08491481e-03 6.31013885e-02
5.60025610e-02 1.01183403e+00 1.49448788e+00 -1.43822718e+00
-4.17596698e-01 -3.59584510e-01 3.67100775e-01 4.14500348e-02
-3.30198497e-01 -2.23951519e-01 -5.09234011e-01 -7.17146397e-01
2.45692939e-01 -7.06394315e-01 -4.85566288e-01 1.15872934e-01
-5.77739835e-01 4.40459587e-02 4.87768024e-01 -2.27916002e-01
6.88983738e-01 -2.20910406e+00 1.38198748e-01 -1.77151039e-01
9.61356997e-01 2.33881190e-01 -3.42340469e-01 4.16557312e-01
-3.44686300e-01 -1.65722847e-01 -3.90186578e-01 -7.35333443e-01
7.68708289e-02 8.67522657e-02 -8.14066947e-01 1.90497905e-01
5.04796028e-01 1.33924937e+00 -5.65410733e-01 -1.37204304e-01
4.67970520e-02 4.19324964e-01 -6.47313952e-01 2.37051055e-01
-3.98047119e-01 5.29125869e-01 -8.31112489e-02 6.18901014e-01
8.30361128e-01 -7.65482962e-01 1.01024233e-01 -4.43864971e-01
-3.32193166e-01 4.29360062e-01 -4.52272981e-01 1.75514936e+00
-1.02059118e-01 7.05881178e-01 2.42060810e-01 -1.26231599e+00
6.69535637e-01 1.55543312e-01 3.19653541e-01 -8.30068409e-01
-3.79657634e-02 -1.40969872e-01 -2.46151080e-06 -4.38938379e-01
3.52456659e-01 9.11915489e-03 3.61661583e-01 3.52062464e-01
2.80391634e-01 1.21680118e-01 -2.12415889e-01 2.19629467e-01
1.30926144e+00 1.06045544e-01 -4.97291088e-02 -2.32921317e-01
2.76284844e-01 2.58746713e-01 3.51845741e-01 1.03618944e+00
-1.61990359e-01 8.80563796e-01 2.02145219e-01 -5.54719269e-01
-8.60901117e-01 -1.41038489e+00 1.12768471e-01 1.14933681e+00
2.90522119e-03 -3.65193635e-01 -8.83321881e-01 -7.53830969e-01
-7.63310939e-02 1.53306335e-01 -8.87458920e-01 -3.51634264e-01
-5.20015955e-01 -7.48048842e-01 5.33599138e-01 7.42728233e-01
7.76029825e-01 -1.10472381e+00 -2.94511050e-01 2.84457691e-02
3.18227187e-02 -1.06974936e+00 -3.75918657e-01 5.40439606e-01
-9.11906004e-01 -9.59507287e-01 -5.76230705e-01 -1.30975735e+00
9.53019619e-01 8.80209625e-01 1.34836364e+00 3.15463364e-01
-2.52678454e-01 2.79704988e-01 -3.73404361e-02 -4.21312064e-01
-1.52502105e-01 -3.02359704e-02 -2.27858409e-01 2.62526095e-01
3.22470695e-01 -8.68637681e-01 -9.71205592e-01 1.59723565e-01
-9.39198613e-01 3.40892434e-01 1.03227174e+00 1.03480279e+00
4.61236417e-01 -2.25448221e-01 3.96116346e-01 -1.01768637e+00
4.28510122e-02 -2.49769494e-01 -3.21178913e-01 1.15661442e-01
-6.52920842e-01 -1.54969357e-02 5.64197540e-01 -4.78378087e-01
-8.63550246e-01 2.04896852e-01 -1.31106943e-01 -5.11972606e-01
-1.53566554e-01 2.60502428e-01 -3.12968940e-01 -3.19465220e-01
6.71457171e-01 5.80349088e-01 1.35581493e-01 -5.57215750e-01
6.76104307e-01 7.87708312e-02 8.13718557e-01 -3.87479782e-01
1.14152038e+00 1.02238476e+00 -2.12133825e-01 -6.26864076e-01
-1.11151338e+00 -1.48379160e-02 -6.07396901e-01 2.77691096e-01
9.85376656e-01 -1.39558959e+00 -7.86834359e-01 5.75342000e-01
-1.05049455e+00 -7.40426064e-01 -5.30246735e-01 1.53451730e-02
-6.16150141e-01 2.03838736e-01 -7.06478775e-01 -4.08170134e-01
-2.50872135e-01 -1.19418836e+00 8.77472639e-01 6.42807335e-02
5.02702259e-02 -7.25694656e-01 -2.96348184e-01 6.36011422e-01
6.87387109e-01 -2.72098035e-01 1.06442058e+00 -2.14592695e-01
-1.14728987e+00 1.98981419e-01 -5.51661789e-01 6.23246551e-01
5.16228043e-02 -3.74445319e-01 -1.52615833e+00 -2.52000749e-01
3.23705196e-01 -3.87630522e-01 1.60198462e+00 4.67586577e-01
1.34509993e+00 -3.65121961e-01 -3.36399376e-01 9.55588698e-01
1.34415984e+00 -2.57164448e-01 8.82206202e-01 1.62911326e-01
8.27638388e-01 5.73678613e-01 -1.36611387e-01 -1.13400467e-01
8.06143463e-01 3.61493081e-01 7.88063645e-01 -5.18380225e-01
-4.73690867e-01 -4.23313290e-01 5.69172561e-01 6.12908959e-01
5.74535178e-03 -1.10809866e-03 -5.39351344e-01 5.75391054e-01
-1.84560513e+00 -9.59085643e-01 -3.28260884e-02 1.86130440e+00
8.93950701e-01 2.27885604e-01 1.43708870e-01 -3.89152057e-02
4.37698603e-01 2.25226939e-01 -7.13208854e-01 -8.61553883e-04
-4.04605597e-01 5.49707472e-01 4.72882986e-01 2.79841214e-01
-1.12358522e+00 1.23152280e+00 7.31090736e+00 5.50779164e-01
-1.25860298e+00 3.17867517e-01 7.89362669e-01 2.08066419e-01
-3.22098494e-01 2.70861179e-01 -6.89806223e-01 2.21611395e-01
5.22254765e-01 2.18211219e-01 4.60279524e-01 6.53341174e-01
-1.23465799e-01 2.71538138e-01 -1.30547106e+00 8.54044557e-01
-1.84653755e-02 -1.85614097e+00 2.95677722e-01 7.39640519e-02
7.81081438e-01 5.31008959e-01 7.13261783e-01 2.64306694e-01
5.03314018e-01 -1.30079961e+00 1.01342070e+00 3.85112703e-01
7.22788036e-01 -7.26656094e-02 9.65728760e-02 3.83303910e-01
-1.05011880e+00 -2.19761953e-01 -5.56642950e-01 -2.99464494e-01
-3.63099724e-01 3.79060328e-01 -5.35528958e-01 3.83618951e-01
8.11308146e-01 8.34659040e-01 -8.58578682e-01 7.86179721e-01
-2.54559040e-01 5.87433577e-01 2.69451290e-01 5.14694035e-01
2.73838520e-01 -8.80621970e-02 1.79835960e-01 1.09550655e+00
-9.67555307e-03 -1.57398224e-01 2.00601444e-02 1.20903277e+00
-2.45902002e-01 -2.46628180e-01 -8.09169114e-01 1.07194707e-01
1.45764083e-01 1.15010512e+00 -5.42912066e-01 -3.88777852e-01
-8.49037290e-01 1.19227326e+00 6.51896894e-01 4.72692609e-01
-6.44586325e-01 1.98688939e-01 8.49822938e-01 2.18791917e-01
7.64058530e-01 -1.82627976e-01 -6.39713705e-01 -1.38057172e+00
4.29607965e-02 -1.16533017e+00 1.63085803e-01 -8.13248694e-01
-1.59788072e+00 7.84119666e-01 -4.61417854e-01 -1.27345181e+00
2.11728051e-01 -9.91976261e-01 -9.42042112e-01 6.92431986e-01
-1.91638112e+00 -1.59966123e+00 -4.27549452e-01 1.00115287e+00
5.68968415e-01 -9.26653072e-02 7.46314645e-01 1.22622289e-01
-7.51323164e-01 7.33701110e-01 -2.49080807e-01 2.38432854e-01
5.01029372e-01 -1.25540113e+00 7.08903551e-01 9.48616564e-01
3.67180586e-01 9.60797668e-01 5.43852866e-01 -2.80230194e-01
-1.52235520e+00 -1.30019283e+00 6.95311487e-01 -7.30506539e-01
5.91325760e-01 -4.00501758e-01 -7.52287149e-01 9.46299613e-01
7.41148412e-01 1.58865795e-01 4.74345118e-01 -1.52683094e-01
-1.06140077e+00 -3.13604385e-01 -8.61488819e-01 7.34339774e-01
1.49360847e+00 -7.81281233e-01 -4.57770199e-01 3.84665132e-01
9.01328921e-01 -1.04243755e-01 -4.38531399e-01 4.47467506e-01
2.68506527e-01 -1.39411557e+00 1.25991106e+00 -6.05714798e-01
6.02067471e-01 -2.53306717e-01 -3.23080808e-01 -1.02987683e+00
-5.51441967e-01 -7.90468931e-01 -1.56594813e-01 9.23742652e-01
5.28286397e-01 -7.94347167e-01 1.01832652e+00 3.47404927e-01
-4.88840878e-01 -7.51544297e-01 -5.37130237e-01 -6.85530901e-01
2.81178087e-01 -3.43198568e-01 5.15778244e-01 7.94058323e-01
-3.07337523e-01 4.44660425e-01 -3.57582211e-01 4.11877275e-01
8.71272564e-01 3.92306358e-01 7.98685610e-01 -1.10767508e+00
-6.41314626e-01 -7.38484144e-01 -1.29576176e-01 -1.62964988e+00
-4.29243706e-02 -1.02565706e+00 1.04456700e-01 -1.46779919e+00
3.38670999e-01 -2.41721481e-01 -4.50258166e-01 7.37872064e-01
-9.21459422e-02 6.78421497e-01 5.99196665e-02 4.68804359e-01
-4.02545691e-01 3.71682525e-01 1.29393625e+00 -4.91662174e-01
1.76106364e-01 4.15534861e-02 -1.37401009e+00 7.63389647e-01
6.28005385e-01 -2.08133370e-01 -5.79844356e-01 -9.66500044e-01
1.33727536e-01 -5.15608788e-01 8.58486533e-01 -9.10077870e-01
3.00195485e-01 1.36295408e-01 4.18044537e-01 -4.04410690e-01
5.11465549e-01 -6.07916892e-01 -1.03135988e-01 2.94562280e-01
-2.49230504e-01 4.33616526e-02 7.35936165e-02 4.67172116e-01
-2.39472792e-01 2.27606818e-01 8.59676123e-01 -5.53220451e-01
-7.17657983e-01 4.96267796e-01 -1.72185808e-01 -1.60747305e-01
6.25358224e-01 -4.15299326e-01 -6.31967664e-01 -2.92216152e-01
-8.64186883e-01 -8.12627822e-02 5.59270442e-01 2.38390744e-01
7.94780493e-01 -9.81439471e-01 -5.52871525e-01 6.10899985e-01
1.44872174e-01 2.09674649e-02 2.80512869e-01 9.81620848e-01
-1.21545814e-01 2.52773523e-01 -1.67774692e-01 -6.55810893e-01
-7.86347449e-01 8.78042161e-01 3.92684400e-01 -3.54526937e-02
-8.93295705e-01 1.34806132e+00 1.01025844e+00 -1.14680044e-01
4.72651511e-01 -4.28365350e-01 1.05351888e-01 -2.13407695e-01
6.53308988e-01 -3.04770440e-01 9.73010510e-02 -4.85097855e-01
-1.21125430e-01 5.73049247e-01 -2.91952759e-01 6.96424097e-02
1.47138214e+00 -2.00597271e-01 -3.35294634e-01 5.91977909e-02
8.91727567e-01 -1.17673047e-01 -1.67871952e+00 -6.18682444e-01
-1.48770839e-01 -1.35854200e-01 -1.62459968e-03 -7.10665643e-01
-1.44420409e+00 1.01502204e+00 3.20718676e-01 2.66118556e-01
1.35958087e+00 2.29440227e-01 6.55079424e-01 3.30343962e-01
3.09712112e-01 -5.01520216e-01 4.15418178e-01 4.09192652e-01
8.75988245e-01 -1.39043593e+00 -3.19348752e-01 -5.00436902e-01
-7.05671787e-01 8.00980866e-01 9.93319631e-01 -3.22238952e-01
8.59005988e-01 3.40817124e-01 3.72818768e-01 -5.26095569e-01
-9.96340215e-01 -6.96162045e-01 2.98123330e-01 7.98108459e-01
3.21001381e-01 -2.91201025e-01 4.00302947e-01 7.44820833e-01
-1.93671063e-01 -2.48702183e-01 1.97855875e-01 7.85226941e-01
-4.13804263e-01 -1.14090931e+00 -5.45877554e-02 2.14245692e-01
-2.33509868e-01 -4.32624936e-01 -6.11795843e-01 5.89254260e-01
5.53255796e-01 9.64897454e-01 4.21629101e-02 -7.19882965e-01
1.74142212e-01 2.44224761e-02 6.81509376e-01 -8.83468688e-01
-8.95138979e-01 -1.01757392e-01 -4.17955399e-01 -7.62080371e-01
-5.10457337e-01 -2.02156231e-01 -8.51290166e-01 -2.53172219e-01
8.34874138e-02 -2.87070423e-01 3.66425872e-01 8.34356666e-01
6.18769109e-01 5.84448516e-01 6.67830586e-01 -1.01634800e+00
-5.08582354e-01 -7.62955368e-01 -2.49058053e-01 2.37522036e-01
7.52603412e-01 -4.17381108e-01 -1.61556840e-01 3.47965717e-01] | [9.608291625976562, 1.2169467210769653] |
4dca25c0-6567-405d-9320-224354114356 | pi-pe-a-pipeline-for-pulmonary-embolism | 1910.02175 | null | https://arxiv.org/abs/1910.02175v3 | https://arxiv.org/pdf/1910.02175v3.pdf | Pi-PE: A Pipeline for Pulmonary Embolism Detection using Sparsely Annotated 3D CT Images | Pulmonary embolisms (PE) are known to be one of the leading causes for cardiac-related mortality. Due to inherent variabilities in how PE manifests and the cumbersome nature of manual diagnosis, there is growing interest in leveraging AI tools for detecting PE. In this paper, we build a two-stage detection pipeline that is accurate, computationally efficient, robust to variations in PE types and kernels used for CT reconstruction, and most importantly, does not require dense annotations. Given the challenges in acquiring expert annotations in large-scale datasets, our approach produces state-of-the-art results with very sparse emboli contours (at 10mm slice spacing), while using models with significantly lower number of parameters. We achieve AUC scores of 0.94 on the validation set and 0.85 on the test set of highly severe PEs. Using a large, real-world dataset characterized by complex PE types and patients from multiple hospitals, we present an elaborate empirical study and provide guidelines for designing highly generalizable pipelines. | ['Ehsan Dehghan', 'David Beymer', 'Shafiqul Abedin', 'Deepta Rajan'] | 2019-10-05 | null | null | null | null | ['pulmonary-embolism-detection'] | ['medical'] | [-1.72749877e-01 -1.67585909e-01 4.89559136e-02 1.83270231e-01
-1.02280891e+00 -8.81893516e-01 1.77211866e-01 5.56826174e-01
-3.41976523e-01 6.00993037e-01 1.02088507e-02 -5.65786481e-01
-2.26223603e-01 -3.84398013e-01 -1.79334000e-01 -3.92335445e-01
-3.87512565e-01 9.24270749e-01 1.00336921e+00 4.45489705e-01
3.78960371e-02 9.00254428e-01 -7.31400073e-01 2.85390228e-01
8.54112566e-01 9.65181947e-01 1.30669758e-01 1.07998943e+00
2.01756209e-01 9.70675707e-01 -6.18099689e-01 -4.21940744e-01
3.55792791e-01 -5.71588397e-01 -6.40556872e-01 -1.44204021e-01
2.41774142e-01 -1.50295198e-01 -3.38466465e-01 3.20258260e-01
8.44489336e-01 -2.50744194e-01 9.61251676e-01 -6.76320195e-01
-8.38249642e-03 1.66851521e-01 -3.08183938e-01 9.99285877e-01
-3.62235717e-02 4.18447524e-01 6.13921404e-01 -7.36432135e-01
3.40681165e-01 5.86793005e-01 1.06954384e+00 3.51898491e-01
-9.22916293e-01 -3.65594983e-01 -5.56446552e-01 7.91292787e-02
-1.28792286e+00 -1.76420122e-01 9.25203040e-02 -6.40672684e-01
7.50131071e-01 4.21840608e-01 8.58574390e-01 7.63920903e-01
5.09498715e-01 2.68747181e-01 9.99645114e-01 -1.83910817e-01
1.63141582e-02 1.32964745e-01 -1.59970567e-01 8.14177454e-01
6.04772151e-01 1.11687057e-01 -2.33332545e-01 -5.66815376e-01
1.09634340e+00 1.11358501e-01 -3.93160790e-01 -4.68784928e-01
-1.38929331e+00 5.55866539e-01 1.58247024e-01 1.76242217e-01
-6.13355458e-01 -3.77532304e-03 4.14309949e-01 -2.81918615e-01
6.66251630e-02 6.12712741e-01 -3.81619155e-01 -4.50643897e-01
-1.05834949e+00 1.97465178e-02 1.00990963e+00 7.15079188e-01
-1.51299104e-01 -1.96560398e-01 -5.53544283e-01 6.31324828e-01
9.84647050e-02 4.32330370e-01 4.39199537e-01 -8.29734445e-01
3.33805621e-01 2.69698203e-01 1.61333382e-01 -7.77611792e-01
-5.20053864e-01 -8.66614461e-01 -7.75592923e-01 2.29865536e-01
5.71083009e-01 -3.58712822e-01 -7.79719353e-01 1.04611719e+00
1.71199664e-01 5.14985502e-01 -2.89169252e-01 8.17730248e-01
7.55487621e-01 3.88561226e-02 4.43039387e-01 -1.00155942e-01
1.73890698e+00 -9.07971740e-01 -1.86730191e-01 -1.63110256e-01
6.41458690e-01 -6.68943167e-01 8.04525495e-01 4.45093632e-01
-1.07904077e+00 -1.93552315e-01 -7.15729475e-01 4.09294695e-01
2.17545509e-01 3.57491821e-01 4.16266441e-01 8.70500565e-01
-7.21142232e-01 6.93331599e-01 -1.26187575e+00 -4.10004556e-01
1.05523610e+00 1.60894454e-01 -1.69944435e-01 2.42380366e-01
-5.99033475e-01 1.18694961e+00 2.16499999e-01 -2.65626878e-01
-1.01550901e+00 -1.33685303e+00 -2.06914380e-01 2.86730707e-01
4.24600422e-01 -1.00487709e+00 1.12164319e+00 -1.13978453e-01
-1.26498866e+00 7.55085230e-01 -1.72010958e-02 -5.59522092e-01
1.05120933e+00 -3.58561456e-01 -2.11393237e-01 6.15656912e-01
-1.11719256e-03 2.34561741e-01 7.58182824e-01 -8.86946797e-01
-7.78106093e-01 6.66380227e-02 -5.03861666e-01 3.44012454e-02
-1.29800392e-02 2.96812564e-01 -3.40990514e-01 -6.81601107e-01
-2.99816221e-01 -1.00514233e+00 -6.44341946e-01 3.47808689e-01
-4.78795543e-02 1.03271656e-01 6.73464775e-01 -7.44899869e-01
1.41897726e+00 -1.96114445e+00 -1.89528942e-01 1.95165813e-01
8.81812274e-01 6.07169390e-01 4.67400163e-01 3.39839533e-02
1.16493031e-02 2.78271407e-01 -5.09831250e-01 1.21381044e-01
-3.83331090e-01 1.30301282e-01 -7.05735832e-02 2.91240185e-01
3.12297225e-01 1.22037351e+00 -9.56782222e-01 -8.45570982e-01
5.41898429e-01 3.77876103e-01 -4.53578383e-01 4.13189948e-01
3.74845117e-01 8.62432063e-01 -5.40000558e-01 3.98731589e-01
2.66288012e-01 -8.59573662e-01 -1.06752838e-03 -5.70503809e-02
-6.56269025e-03 5.95489889e-03 -1.12925363e+00 1.44020283e+00
-2.32045650e-01 5.89298010e-01 -1.05246849e-01 -6.52812421e-01
4.11399066e-01 7.59503305e-01 8.22130263e-01 -1.51482271e-02
2.14760303e-01 3.65868092e-01 3.83281022e-01 -6.17871463e-01
-2.06191987e-01 -3.30863327e-01 3.23872924e-01 3.43696952e-01
-1.80872098e-01 -2.25889638e-01 2.17107132e-01 7.84198567e-02
1.69263554e+00 -3.49289209e-01 5.85427463e-01 -4.32492524e-01
4.21216428e-01 2.82931924e-01 4.22280490e-01 9.54296887e-01
-5.95135868e-01 1.11219668e+00 4.93672013e-01 -5.41839957e-01
-8.81779552e-01 -1.36650240e+00 -4.40697163e-01 2.56784201e-01
-2.02819124e-01 -4.58643287e-01 -4.72002268e-01 -8.85674417e-01
-2.50236690e-01 7.00682774e-02 -3.22027087e-01 1.10832199e-01
-8.45129192e-01 -9.81591403e-01 9.05658364e-01 8.23232174e-01
3.97949010e-01 -9.41527605e-01 -1.43832648e+00 5.33447027e-01
-1.76607221e-01 -1.32938194e+00 -1.55393496e-01 -7.25466609e-02
-1.10529149e+00 -1.56345212e+00 -1.10770512e+00 -6.02770388e-01
4.45036471e-01 -4.07559052e-03 1.61294985e+00 2.24930689e-01
-9.96620059e-01 2.86719918e-01 -2.96952248e-01 -5.83509326e-01
-2.89073676e-01 4.06614132e-03 -3.92935276e-01 -4.61003423e-01
-7.36066476e-02 -3.09063822e-01 -1.07169306e+00 2.57429153e-01
-5.22107244e-01 -2.68007606e-01 8.90857637e-01 7.31351078e-01
6.82861745e-01 2.35520974e-02 5.10009229e-01 -1.22130573e+00
6.37197495e-01 -6.06019974e-01 -4.15214211e-01 2.50378996e-01
-5.01451492e-01 -3.38965237e-01 5.82909405e-01 -4.27727878e-01
-9.20043528e-01 9.13066939e-02 8.03719386e-02 -4.99870181e-01
-4.41368878e-01 1.74678624e-01 4.86383200e-01 -2.31405511e-01
9.07416761e-01 -2.12811351e-01 -1.79122984e-01 -2.35519901e-01
1.71190556e-02 3.85202587e-01 6.25866055e-01 -5.08279860e-01
7.54390240e-01 5.28586566e-01 5.07539749e-01 -6.96951032e-01
-7.18021512e-01 -5.23943365e-01 -8.49728107e-01 -2.43256930e-02
9.63527083e-01 -7.48729169e-01 -1.96118593e-01 6.40431792e-02
-8.13483655e-01 -2.86048919e-01 -5.37292123e-01 5.84601820e-01
-2.66352594e-01 7.74972975e-01 -5.56974947e-01 -4.32926744e-01
-7.93765545e-01 -1.15738261e+00 6.50955915e-01 7.77954683e-02
-5.64014912e-01 -1.11199677e+00 1.95799366e-01 1.81898519e-01
6.91897273e-01 4.08285677e-01 9.09129918e-01 -6.65589511e-01
-6.10219300e-01 -2.73777127e-01 -4.54297274e-01 1.61176667e-01
1.92060307e-01 -6.36071637e-02 -6.20809436e-01 -5.11285253e-02
-1.13101408e-01 1.77389741e-01 4.87786323e-01 6.54668391e-01
1.34692490e+00 2.08372697e-01 -5.43107390e-01 6.17216825e-01
1.02437973e+00 4.53864560e-02 5.11272132e-01 1.25672325e-01
7.12881207e-01 2.55116165e-01 4.72871691e-01 5.37067771e-01
1.57025903e-02 2.50524759e-01 1.53947428e-01 -2.19971403e-01
-3.97280931e-01 1.98134303e-01 -3.86624902e-01 3.98051769e-01
-2.89085120e-01 -2.38574579e-01 -1.34822679e+00 5.85217595e-01
-1.57797933e+00 -7.78752744e-01 -6.81220472e-01 2.14434171e+00
5.19329786e-01 1.15884669e-01 3.16360295e-01 -1.55568585e-01
3.70578855e-01 -2.50672311e-01 -2.39865541e-01 1.17739011e-02
2.40080521e-01 8.77323151e-01 5.24518788e-01 1.12855285e-01
-1.32501709e+00 5.70936501e-01 7.15011454e+00 5.32392263e-01
-1.09685278e+00 2.48580292e-01 5.23665369e-01 -2.89983824e-02
2.95278370e-01 -1.71192586e-01 -3.18374664e-01 6.46759033e-01
1.06303811e+00 4.34072223e-03 -1.23098902e-02 4.49736297e-01
2.19182834e-01 -9.91538540e-02 -1.01827848e+00 9.37577248e-01
-1.79820299e-01 -1.49620008e+00 -3.45617920e-01 -2.29473338e-01
4.01450068e-01 2.48896837e-01 -1.46547228e-01 -8.67366977e-03
2.01405346e-01 -1.24642813e+00 3.04194361e-01 2.97493964e-01
1.07420683e+00 -3.88915092e-01 9.00654376e-01 2.45382667e-01
-1.35959291e+00 2.00275078e-01 -1.84274688e-01 2.52314359e-01
4.16225940e-01 5.59287488e-01 -1.03870308e+00 4.09139782e-01
9.86588359e-01 6.85511470e-01 -6.89946651e-01 1.74654663e+00
-3.25202107e-01 8.88034046e-01 -5.71732163e-01 3.28628391e-01
-2.57032722e-01 1.45660803e-01 5.79412997e-01 1.52291894e+00
3.89329523e-01 4.80929315e-01 1.94492623e-01 6.95383191e-01
1.54996276e-01 1.36684701e-01 -2.96243638e-01 1.19642802e-01
5.85602224e-01 1.52939391e+00 -9.19801354e-01 -5.18869638e-01
-3.73451829e-01 7.63795257e-01 2.44214579e-01 6.12403862e-02
-1.00094712e+00 -1.06789686e-01 8.72223526e-02 4.79158103e-01
2.65316308e-01 -4.84028235e-02 -5.65805435e-01 -1.03494215e+00
-3.59320678e-02 -5.93738437e-01 7.54910409e-01 -3.25302571e-01
-1.36328888e+00 7.42491961e-01 -1.13619655e-01 -1.46093690e+00
-7.95168057e-02 -7.53886580e-01 -1.03771472e+00 1.00839627e+00
-1.67624474e+00 -7.48984218e-01 -7.26015985e-01 3.21772695e-01
4.14026022e-01 -2.17465665e-02 1.07639420e+00 4.49414343e-01
-4.59665775e-01 4.09959346e-01 -3.59652758e-01 1.61188081e-01
8.35884273e-01 -1.29618239e+00 1.48781016e-01 7.58746982e-01
-6.11796267e-02 4.16651040e-01 3.71825367e-01 -6.18280411e-01
-8.54056001e-01 -1.09709084e+00 5.75862885e-01 -1.04569054e+00
6.64604247e-01 3.54832083e-01 -8.56428564e-01 5.44552207e-01
-1.88799784e-01 5.51212132e-01 9.66007233e-01 -4.11270291e-01
1.01392560e-01 3.52518052e-01 -1.22937131e+00 2.23097652e-01
8.25939596e-01 -4.50779013e-02 -7.33846188e-01 2.17634737e-01
-1.33089945e-01 -8.36739004e-01 -1.31424522e+00 6.85521245e-01
5.25254369e-01 -8.65169346e-01 1.17376280e+00 -5.32549858e-01
3.42459112e-01 -1.55834764e-01 5.56403697e-01 -9.70512807e-01
-5.95703959e-01 -3.66553426e-01 -2.64268458e-01 6.14518344e-01
3.81735176e-01 -6.50656581e-01 1.02813828e+00 5.33956409e-01
-2.27263033e-01 -1.10954046e+00 -8.82120609e-01 -4.77830738e-01
2.87223905e-01 -6.05802573e-02 3.04421186e-02 7.53947079e-01
-2.43203983e-01 3.61062251e-02 -1.07734408e-02 4.04667079e-01
7.24911690e-01 -7.63096213e-02 5.53690732e-01 -1.66356623e+00
-6.06563985e-01 -4.80100393e-01 -2.91975975e-01 -6.82920277e-01
-3.47489953e-01 -8.57300043e-01 -1.82698905e-01 -1.56352067e+00
5.06283164e-01 -8.30507755e-01 -4.36766893e-01 3.35140288e-01
-5.26938319e-01 3.42573971e-01 3.29447500e-02 6.53392255e-01
-3.87891084e-01 -2.38768943e-02 1.22696507e+00 3.85464787e-01
-2.38553844e-02 -1.29730906e-02 -4.80486006e-01 9.39731300e-01
9.30681705e-01 -7.10535049e-01 -3.07318360e-01 -2.77307361e-01
-2.86866665e-01 2.82422185e-01 5.73923051e-01 -1.27967381e+00
7.12472051e-02 -7.49632018e-03 6.62058651e-01 -4.95570242e-01
5.52923381e-02 -9.74056542e-01 2.49521479e-01 8.70049715e-01
9.59228054e-02 1.61622748e-01 3.66028726e-01 4.16265339e-01
1.07381515e-01 -1.90104023e-01 1.18916547e+00 -4.36171144e-01
-3.30307603e-01 5.91815472e-01 -4.93568689e-01 6.63120389e-01
1.28259408e+00 -1.67327702e-01 -2.07763925e-01 4.89121340e-02
-7.30234265e-01 2.25148350e-01 3.33277762e-01 7.28207529e-02
4.10654396e-01 -7.42702305e-01 -9.07692969e-01 5.58169112e-02
-8.54742602e-02 2.88385302e-01 1.98877901e-01 1.36636126e+00
-1.24752617e+00 4.02933389e-01 -1.23696111e-01 -1.00917065e+00
-1.40218389e+00 2.24788323e-01 7.44408071e-01 -7.42045820e-01
-1.21377993e+00 8.32320333e-01 2.09418908e-01 1.31383315e-01
5.90735162e-03 -1.69104829e-01 -4.97448668e-02 -2.66376376e-01
3.56724918e-01 6.41933858e-01 2.09691510e-01 -2.93121129e-01
-5.33666790e-01 5.56518912e-01 4.31150720e-02 2.00472176e-01
1.05911863e+00 3.10336828e-01 1.47930384e-01 -8.18415806e-02
4.64245588e-01 1.49462596e-01 -1.17255247e+00 1.02817304e-01
8.22964832e-02 -5.53641438e-01 -4.61755656e-02 -9.72807527e-01
-1.10373008e+00 1.08909214e+00 8.22479844e-01 -2.07106367e-01
8.20511103e-01 2.00705484e-01 8.71542692e-01 -6.21091686e-02
2.21435964e-01 -3.89850825e-01 5.73661737e-02 1.99408621e-01
3.29587758e-01 -1.09106815e+00 1.18960388e-01 -7.26282179e-01
-7.97665238e-01 1.02915907e+00 4.44495738e-01 -2.54866421e-01
7.17764318e-01 4.85716343e-01 3.60392034e-01 -3.74347508e-01
-4.91815984e-01 1.39907286e-01 3.70451003e-01 6.31377220e-01
4.77781028e-01 2.39907190e-01 -3.34277362e-01 4.78723764e-01
-2.18125768e-02 1.45944348e-02 4.44435179e-01 1.08347917e+00
-5.54700911e-01 -1.15021682e+00 -2.01119721e-01 8.90723467e-01
-1.02007413e+00 -8.95849466e-02 -2.31579766e-01 6.55442536e-01
-7.73852989e-02 7.74880230e-01 -2.82823056e-01 2.73259103e-01
5.38848817e-01 -6.87857717e-02 5.20790935e-01 -8.03171873e-01
-8.72091711e-01 5.95879219e-02 8.34878013e-02 -4.82124299e-01
-3.08707003e-02 -7.23163247e-01 -1.59296405e+00 6.94652945e-02
-2.61029750e-01 1.93474859e-01 4.86553349e-02 8.61215949e-01
6.63741231e-01 8.73785734e-01 1.28702551e-01 -4.74191219e-01
-5.18497348e-01 -8.03822339e-01 -3.77997875e-01 6.15429938e-01
1.44462407e-01 -7.01511085e-01 -3.23635489e-01 6.82192892e-02] | [15.180645942687988, -2.062096118927002] |
4e3c1618-5e8b-4731-a070-a7feb1e7e5cf | paris-lille-3d-a-large-and-high-quality | 1712.00032 | null | http://arxiv.org/abs/1712.00032v2 | http://arxiv.org/pdf/1712.00032v2.pdf | Paris-Lille-3D: a large and high-quality ground truth urban point cloud dataset for automatic segmentation and classification | This paper introduces a new Urban Point Cloud Dataset for Automatic
Segmentation and Classification acquired by Mobile Laser Scanning (MLS). We
describe how the dataset is obtained from acquisition to post-processing and
labeling. This dataset can be used to learn classification algorithm, however,
given that a great attention has been paid to the split between the different
objects, this dataset can also be used to learn the segmentation. The dataset
consists of around 2km of MLS point cloud acquired in two cities. The number of
points and range of classes make us consider that it can be used to train
Deep-Learning methods. Besides we show some results of automatic segmentation
and classification. The dataset is available at:
http://caor-mines-paristech.fr/fr/paris-lille-3d-dataset/ | ['François Goulette', 'Jean-Emmanuel Deschaud', 'Xavier Roynard'] | 2017-11-30 | null | null | null | null | ['lidar-semantic-segmentation'] | ['computer-vision'] | [-1.85818709e-02 1.55199245e-01 -1.05545670e-01 -7.03863919e-01
-7.65775740e-01 -2.75163502e-01 6.50948346e-01 2.72980005e-01
-4.91517723e-01 5.82578063e-01 -5.40298343e-01 -4.17209178e-01
5.06934188e-02 -1.24462724e+00 -8.27703655e-01 -5.83397865e-01
-1.42974779e-01 1.17566156e+00 4.59114760e-01 -3.01541984e-01
4.45208959e-02 1.02726936e+00 -1.69104302e+00 1.65064573e-01
7.64686704e-01 9.88661587e-01 7.30803072e-01 2.51649350e-01
-4.28718716e-01 2.68543642e-02 -5.38890958e-02 5.23667550e-03
7.95589387e-01 -9.10415724e-02 -1.00903749e+00 2.23827392e-01
3.69515717e-01 -2.40661297e-03 1.08317569e-01 8.85561109e-01
8.48986879e-02 1.79359347e-01 9.08636928e-01 -1.16213393e+00
2.91876495e-01 3.12115371e-01 -5.34170210e-01 1.35402083e-01
-1.25658855e-01 2.04461645e-02 8.90685201e-01 -6.94511414e-01
5.25531292e-01 1.03425956e+00 6.49540544e-01 1.73787028e-02
-8.10779631e-01 -4.55291748e-01 -1.76514834e-01 4.60162759e-01
-1.61810565e+00 -1.49387985e-01 6.78495646e-01 -7.77677059e-01
4.23703283e-01 2.22401381e-01 1.07066905e+00 4.83955473e-01
-4.95191544e-01 1.04525268e+00 1.16791260e+00 -4.48788136e-01
4.27273929e-01 1.66037589e-01 3.87815326e-01 6.01616740e-01
2.35294104e-01 8.89172684e-03 3.22932631e-01 4.18786705e-01
8.91680658e-01 1.47823006e-01 8.15557614e-02 -3.83284926e-01
-7.12134361e-01 9.90991235e-01 9.29307342e-01 6.23148382e-01
-3.32587689e-01 -4.15726975e-02 -7.00785452e-03 1.69170126e-01
4.50116992e-01 5.66136464e-03 -8.25767577e-01 1.82178188e-02
-1.23903143e+00 1.78127453e-01 2.98702776e-01 9.40951645e-01
1.50172436e+00 -2.67069638e-01 5.67556977e-01 7.74985850e-01
3.98335129e-01 6.88025355e-01 1.62187338e-01 -8.58022213e-01
5.97149611e-01 7.71046996e-01 7.68888891e-02 -8.32159221e-01
-7.06271172e-01 -2.25743338e-01 -5.03353417e-01 4.18399036e-01
6.30008698e-01 9.31117125e-03 -1.11132777e+00 9.07493353e-01
2.99371213e-01 1.85360551e-01 -2.08728611e-01 7.52148807e-01
6.30034745e-01 4.11352217e-01 -2.22619772e-01 1.61901549e-01
1.08767056e+00 -7.04672635e-01 -1.37180910e-01 -3.96237463e-01
8.12952936e-01 -4.69218433e-01 1.13469088e+00 2.26516545e-01
-6.82602525e-01 -5.89965463e-01 -8.58037233e-01 8.02475885e-02
-8.47443163e-01 2.79585719e-01 4.98670369e-01 4.98945296e-01
-8.01291704e-01 6.11268699e-01 -1.02984250e+00 -6.29099727e-01
1.01763177e+00 3.76421183e-01 -8.46082345e-02 -3.91956158e-02
-8.57306302e-01 8.24124038e-01 7.01841414e-01 -3.47518995e-02
-4.37996387e-01 -2.61113077e-01 -6.29454851e-01 -1.14979759e-01
8.55707079e-02 -3.72082293e-01 1.23681974e+00 -8.66098881e-01
-1.15374291e+00 1.08945131e+00 -1.37852570e-02 -5.81373930e-01
7.42103517e-01 -8.17643702e-02 -1.85633510e-01 2.31459305e-01
2.39117384e-01 9.48291183e-01 5.54498136e-01 -1.40756476e+00
-1.04945230e+00 -6.11231923e-01 9.35310349e-02 1.46095073e-02
1.78221673e-01 -3.26172322e-01 -4.44028586e-01 -1.87361464e-01
5.50762832e-01 -1.27815807e+00 -3.67862761e-01 -3.06824654e-01
-4.43722755e-01 -1.16734661e-01 1.07040989e+00 -5.15340328e-01
3.77072841e-01 -1.78330314e+00 -2.90594518e-01 3.82931381e-01
-3.13530654e-01 4.19126928e-01 -1.26178488e-01 3.68584424e-01
-1.61575392e-01 1.63547367e-01 -9.53847647e-01 -3.64549398e-01
-1.58072442e-01 3.93241197e-01 8.98826942e-02 4.91347402e-01
2.67992541e-02 1.03100753e+00 -6.88237071e-01 -5.70030153e-01
5.82543135e-01 3.51917475e-01 -7.03307390e-02 -1.67584509e-01
-4.09961969e-01 9.57087755e-01 -6.47570908e-01 5.15405476e-01
1.02805114e+00 1.38091609e-01 -4.22318697e-01 1.16797984e-02
-2.38283008e-01 1.24321587e-01 -1.31634331e+00 1.50893438e+00
-4.66429561e-01 7.38934159e-01 -4.34734374e-02 -1.32299638e+00
1.00708234e+00 -1.74226627e-01 8.31322491e-01 -7.84140408e-01
3.62608582e-01 3.47887635e-01 -3.28696251e-01 -4.27076578e-01
3.97003263e-01 -4.37135436e-02 1.55496985e-01 2.26529390e-01
-2.93322951e-01 -6.60812199e-01 4.96465415e-01 -3.51450741e-01
5.49976349e-01 1.34063333e-01 -3.20269689e-02 -2.89161921e-01
6.79114342e-01 6.58449948e-01 2.26501286e-01 3.65478933e-01
1.15823768e-01 6.84610605e-01 -2.53717182e-03 -6.12223148e-01
-8.90983462e-01 -9.40719903e-01 -5.01331747e-01 7.31733024e-01
1.72185421e-01 9.17644352e-02 -8.58282924e-01 -6.89420521e-01
-2.85132676e-02 7.48804212e-01 -3.71979237e-01 4.92662638e-01
-5.37483811e-01 -5.23010552e-01 3.73779863e-01 3.30425471e-01
9.77660418e-01 -1.00191188e+00 -6.55959785e-01 -1.37500823e-01
-2.29858235e-01 -1.12761974e+00 4.32853460e-01 1.42551810e-01
-1.28022385e+00 -1.46966720e+00 -3.97116274e-01 -4.56685007e-01
6.30705535e-01 4.68062252e-01 1.15147209e+00 1.33572131e-01
-7.23394081e-02 5.29821277e-01 -4.62465137e-01 -6.81062341e-01
-3.47479731e-01 5.66381454e-01 -3.25361341e-01 5.59917986e-02
4.98337269e-01 -6.66590869e-01 -4.62138504e-01 3.65151733e-01
-7.56999254e-01 -6.69796765e-02 4.95508105e-01 1.16339438e-01
8.05838943e-01 3.38073075e-01 -3.41472067e-02 -1.08935821e+00
-1.28598407e-01 -3.79712015e-01 -9.41132903e-01 -1.64640248e-01
-2.29789406e-01 -3.45044434e-01 1.81694269e-01 3.59058768e-01
-8.00751030e-01 6.22236252e-01 -5.74430108e-01 -3.20311725e-01
-1.08828878e+00 5.44504702e-01 -3.75312090e-01 3.62347849e-02
4.03114557e-01 -4.02944870e-02 -1.88268825e-01 -7.79718041e-01
5.67085564e-01 9.42390561e-01 1.94344327e-01 -2.49039963e-01
1.08688915e+00 7.32206285e-01 7.93997124e-02 -1.36762917e+00
-6.67655230e-01 -7.77930081e-01 -1.70304990e+00 -2.93831140e-01
8.71471047e-01 -8.22961330e-01 -1.83377475e-01 3.82562310e-01
-1.04097497e+00 -7.05639005e-01 -5.97132027e-01 3.91267776e-01
-8.52071524e-01 4.97441143e-02 -2.32446983e-01 -7.33020663e-01
1.63390443e-01 -1.11955309e+00 1.19996655e+00 9.26276147e-02
1.01651311e-01 -1.10999680e+00 -1.11739993e-01 6.61197364e-01
1.91507954e-02 3.34912390e-01 6.61813974e-01 -5.69696367e-01
-9.71490085e-01 -3.70222598e-01 -1.90969296e-02 2.44941339e-01
1.68683559e-01 -1.14429332e-01 -1.02388275e+00 -1.01468347e-01
-2.01393105e-02 1.04016505e-01 9.93298233e-01 4.23033595e-01
1.18141222e+00 2.18868807e-01 -5.70105433e-01 4.87594247e-01
1.65952778e+00 1.45078981e-02 9.28752542e-01 7.10976362e-01
7.51847982e-01 7.90905118e-01 1.01734650e+00 8.06052312e-02
4.81850147e-01 9.11297798e-01 7.76881158e-01 -7.42479265e-02
-2.71153189e-02 1.54936567e-01 1.09776380e-02 6.73995376e-01
-2.79542625e-01 1.96045756e-01 -1.56563795e+00 6.70022130e-01
-1.82919407e+00 -7.95872927e-01 -9.85800624e-01 2.08459854e+00
3.44415665e-01 1.12956807e-01 4.82371658e-01 4.95395631e-01
6.39740288e-01 5.78004420e-02 -2.50256896e-01 -4.76438925e-03
-4.57067378e-02 3.04171860e-01 8.33318233e-01 8.44045401e-01
-1.38492787e+00 1.18438351e+00 4.97519827e+00 1.08412993e+00
-1.36564577e+00 1.69444367e-01 5.10560811e-01 2.84400851e-01
-1.07201226e-02 -2.81290978e-01 -8.57870758e-01 5.60073972e-01
1.12596262e+00 3.67730886e-01 2.24825330e-02 9.35807288e-01
4.87464756e-01 -4.53876764e-01 -6.90330446e-01 9.29147780e-01
-5.59761405e-01 -1.29648721e+00 -2.89696068e-01 3.72352064e-01
6.31178021e-01 8.69831443e-01 -1.95348203e-01 2.33172059e-01
2.43984923e-01 -8.57193828e-01 6.52362645e-01 4.96695459e-01
6.41334653e-01 -8.47336650e-01 7.00648844e-01 8.37058246e-01
-1.25638163e+00 -4.35732156e-02 -5.58435142e-01 -1.22125871e-01
1.15065187e-01 9.81482863e-01 -1.15145648e+00 8.57997298e-01
7.02127874e-01 7.91496396e-01 -8.85417581e-01 1.33911538e+00
-4.62034881e-01 7.34978199e-01 -6.80281460e-01 2.09712550e-01
4.48143333e-01 -9.33233857e-01 3.67014587e-01 1.15842819e+00
4.95776922e-01 -4.44987863e-02 4.73388165e-01 6.91484809e-01
1.92381814e-01 2.21535906e-01 -7.67301500e-01 2.14070365e-01
1.92907810e-01 1.26079679e+00 -1.40398121e+00 -1.73705146e-01
-1.12885721e-01 6.10122979e-01 1.26236692e-01 1.11246660e-01
-6.24901593e-01 -8.39066431e-02 5.16627431e-01 6.56301320e-01
4.42178518e-01 -7.23609865e-01 -5.96710563e-01 -7.69407570e-01
-2.91465342e-01 -1.28264219e-01 1.15044795e-01 -6.72360837e-01
-7.94199467e-01 3.07722807e-01 3.16499889e-01 -1.32200873e+00
-1.08524933e-01 -5.38233876e-01 -5.72112560e-01 5.75598598e-01
-1.79335308e+00 -1.36925399e+00 -4.35961634e-01 4.74437147e-01
7.78331578e-01 -1.26636893e-01 5.08049726e-01 5.06364882e-01
-4.25420552e-01 -2.95577407e-01 2.90612668e-01 3.36860001e-01
-1.06926404e-01 -1.34470963e+00 4.52946514e-01 6.63239479e-01
4.35693592e-01 -1.32922649e-01 4.45490897e-01 -5.46069801e-01
-6.01591051e-01 -1.67735374e+00 6.37023628e-01 -4.97679591e-01
2.22462609e-01 -2.15531558e-01 -8.01637888e-01 9.02179241e-01
-5.05261980e-02 6.22056685e-02 3.82606000e-01 -1.84227750e-01
2.21053332e-01 -2.97368318e-01 -1.28075731e+00 2.46663317e-01
1.00692725e+00 -2.67698377e-01 -4.89933461e-01 6.67863667e-01
2.14279428e-01 -5.81341624e-01 -4.73554939e-01 4.34254766e-01
-6.54620156e-02 -1.16438842e+00 9.18757975e-01 -1.85220480e-01
2.59541929e-01 -2.99330890e-01 -1.50968790e-01 -1.30569112e+00
5.27689382e-02 1.46183729e-01 3.20254236e-01 1.08837605e+00
3.53880554e-01 -5.91144145e-01 1.25133145e+00 1.77030787e-01
-2.70289719e-01 -4.37488437e-01 -1.15753138e+00 -9.34991241e-01
4.56530064e-01 -1.16958821e+00 8.58003020e-01 9.62286830e-01
-8.46342623e-01 2.72394210e-01 3.06306422e-01 4.27711427e-01
5.48318207e-01 3.86739641e-01 1.10057545e+00 -1.86252403e+00
2.80283600e-01 -4.74982619e-01 -4.99053627e-01 -1.12900782e+00
3.10983926e-01 -1.26374674e+00 -1.42016202e-01 -1.89128971e+00
-7.03351498e-01 -1.08066404e+00 2.52471268e-01 4.52990830e-01
4.03949440e-01 5.46020508e-01 3.06176543e-01 6.03262007e-01
-2.91869134e-01 4.12023008e-01 9.66564655e-01 -3.22438985e-01
-1.88034460e-01 6.98485613e-01 -3.18660699e-02 9.12399769e-01
1.37538910e+00 -6.14319801e-01 -1.05277449e-01 -2.66722202e-01
-5.69682047e-02 -3.79508823e-01 2.29412749e-01 -1.42108822e+00
2.80716065e-02 -2.38709971e-01 2.04190850e-01 -1.37296164e+00
6.38444006e-01 -1.44977379e+00 1.55216679e-01 5.95282972e-01
2.14069992e-01 -4.37635124e-01 8.52648765e-02 2.93447495e-01
-1.29809156e-01 -6.04178488e-01 1.01225567e+00 -4.93351191e-01
-8.87196481e-01 5.15316010e-01 -3.98855031e-01 -1.29788041e-01
1.10761869e+00 -5.65853655e-01 3.06810290e-01 -8.68229792e-02
-9.38143730e-01 2.93648601e-01 5.24161637e-01 1.05371647e-01
3.62386405e-01 -1.28705609e+00 -6.67799890e-01 2.32236803e-01
1.07468180e-01 5.94108462e-01 7.96889514e-02 6.49120629e-01
-9.98620808e-01 5.62342525e-01 -2.66792864e-01 -1.30213726e+00
-1.25449038e+00 2.30493739e-01 4.21424538e-01 8.24025199e-02
-7.72406220e-01 5.37012577e-01 -3.90666068e-01 -7.86976099e-01
-2.48988152e-01 -3.86882931e-01 -3.77944380e-01 5.25878847e-01
-4.60371226e-02 5.53673744e-01 4.15788144e-01 -1.02265155e+00
-2.57144660e-01 9.61713314e-01 5.53706527e-01 -3.33704464e-02
1.71803904e+00 -3.45089376e-01 -1.78093672e-01 7.83840954e-01
1.21986544e+00 -1.76201150e-01 -1.06439602e+00 -6.13217317e-02
5.95914066e-01 -5.51958799e-01 1.02175720e-01 -3.47148895e-01
-1.44187665e+00 1.14931321e+00 9.64539111e-01 4.10641432e-01
6.19892418e-01 1.75585181e-01 7.50417352e-01 6.57111526e-01
6.21054053e-01 -1.10163200e+00 -3.62384260e-01 8.62205386e-01
7.63512909e-01 -1.44531083e+00 -1.93464711e-01 -7.42491663e-01
-3.49605262e-01 1.20527637e+00 6.66729659e-02 -2.36285105e-01
1.01048672e+00 6.71989983e-03 3.28637987e-01 -3.59834582e-01
3.10425490e-01 -9.82961118e-01 1.11601837e-01 1.02289641e+00
-2.70605087e-05 1.65849507e-01 -1.14848882e-01 6.68863207e-02
-7.69328475e-01 1.66578233e-01 6.46327257e-01 7.36498356e-01
-8.33331048e-01 -1.26186907e+00 -5.84947288e-01 4.72821683e-01
1.65739089e-01 1.72182143e-01 -3.29558849e-01 1.08996844e+00
5.78032255e-01 7.33609200e-01 3.89112443e-01 -1.82522297e-01
3.68845314e-01 1.52083546e-01 4.14471984e-01 -6.44339502e-01
-2.26842016e-01 -1.71330705e-01 -6.42579943e-02 -2.69399136e-01
-7.27839887e-01 -1.03294086e+00 -1.54522026e+00 -1.50904953e-01
-1.31434098e-01 4.21943158e-01 9.95472312e-01 1.11617613e+00
1.18832059e-01 2.16567025e-01 7.06073225e-01 -1.43951535e+00
2.69004911e-01 -8.74110341e-01 -8.59299421e-01 3.95309180e-01
8.64460021e-02 -6.84054255e-01 -7.74993077e-02 5.21883368e-02] | [8.437350273132324, -2.453350067138672] |
84d99308-77ee-42b0-ac0d-8f29c41fdef7 | a-transformer-based-framework-for-1 | 2010.02803 | null | https://arxiv.org/abs/2010.02803v3 | https://arxiv.org/pdf/2010.02803v3.pdf | A Transformer-based Framework for Multivariate Time Series Representation Learning | In this work we propose for the first time a transformer-based framework for unsupervised representation learning of multivariate time series. Pre-trained models can be potentially used for downstream tasks such as regression and classification, forecasting and missing value imputation. By evaluating our models on several benchmark datasets for multivariate time series regression and classification, we show that not only does our modeling approach represent the most successful method employing unsupervised learning of multivariate time series presented to date, but also that it exceeds the current state-of-the-art performance of supervised methods; it does so even when the number of training samples is very limited, while offering computational efficiency. Finally, we demonstrate that unsupervised pre-training of our transformer models offers a substantial performance benefit over fully supervised learning, even without leveraging additional unlabeled data, i.e., by reusing the same data samples through the unsupervised objective. | ['Carsten Eickhoff', 'Anuradha Bhamidipaty', 'Dhaval Patel', 'Srideepika Jayaraman', 'George Zerveas'] | 2020-10-06 | a-transformer-based-framework-for | https://openreview.net/forum?id=lE1AB4stmX | https://openreview.net/pdf?id=lE1AB4stmX | null | ['time-series-regression'] | ['time-series'] | [ 4.80514914e-01 7.86858723e-02 -5.09533405e-01 -2.95259893e-01
-9.48974192e-01 -5.39975464e-01 7.65464842e-01 3.46576929e-01
-2.30737835e-01 7.87729502e-01 1.88587531e-01 -3.92799348e-01
-4.27341431e-01 -6.08950973e-01 -6.12273037e-01 -7.68104136e-01
-4.15906012e-01 6.31091595e-01 -3.15076381e-01 -9.33586583e-02
9.77864303e-03 4.63013291e-01 -1.69318187e+00 9.56354961e-02
6.48410201e-01 1.27578723e+00 -3.33360076e-01 3.39983881e-01
2.83516124e-02 1.21521938e+00 -4.42323267e-01 -4.51012217e-02
2.26067424e-01 -2.28295878e-01 -4.19506371e-01 1.11236811e-01
-2.89447010e-01 -2.13767499e-01 -2.43921876e-01 2.79176950e-01
1.50708526e-01 2.09678829e-01 8.76680374e-01 -1.30331755e+00
-2.29491115e-01 5.89108467e-01 -3.59430730e-01 2.31252551e-01
3.22242416e-02 -2.69675478e-02 9.35625076e-01 -6.15641236e-01
4.74687546e-01 6.19185686e-01 8.11488390e-01 2.37506151e-01
-1.64397788e+00 -6.37641907e-01 9.33537707e-02 -3.10834646e-02
-9.82608378e-01 -6.08840525e-01 9.70532119e-01 -5.31475127e-01
1.23872781e+00 1.25405878e-01 3.49046290e-01 1.04236078e+00
-1.33819759e-01 6.93437397e-01 1.12910318e+00 -4.66871172e-01
3.67312312e-01 -8.15276131e-02 -2.72225738e-02 3.31174850e-01
-1.70072645e-01 4.05412018e-01 -4.35559332e-01 -2.08738923e-01
5.91436327e-01 3.65757287e-01 1.48070395e-01 -4.16602522e-01
-1.43043530e+00 6.51991248e-01 -6.84742928e-02 5.44881880e-01
-5.06804287e-01 1.45106107e-01 5.39782166e-01 7.70544589e-01
1.03312111e+00 2.73039430e-01 -9.84123468e-01 -5.19341230e-01
-1.33600712e+00 -1.81393832e-01 7.39913225e-01 7.80818045e-01
6.80667281e-01 4.78914142e-01 -1.72296464e-02 6.30772769e-01
-7.92356133e-02 4.03562278e-01 6.71400368e-01 -7.14904070e-01
5.66166401e-01 7.35167742e-01 4.01711166e-02 -4.79834169e-01
-4.10707831e-01 -6.54374659e-01 -8.80426526e-01 5.12043126e-02
6.73097730e-01 -3.77752692e-01 -8.73507619e-01 1.73762119e+00
-4.30923402e-02 4.84489650e-01 2.19702482e-01 3.19027215e-01
3.93322766e-01 5.46074986e-01 3.72370370e-02 -7.62192130e-01
7.33791769e-01 -5.99726617e-01 -6.26250446e-01 2.83304323e-02
7.26585746e-01 -5.41280866e-01 4.71417189e-01 5.08885145e-01
-8.18192899e-01 -2.82042116e-01 -7.95530736e-01 3.03197384e-01
-4.45838332e-01 1.51351407e-01 1.01898789e+00 5.63106120e-01
-6.75920427e-01 9.36751366e-01 -1.16800499e+00 -3.79675657e-01
4.22528565e-01 5.20471931e-01 -5.13172328e-01 1.98894933e-01
-7.50810921e-01 7.48921692e-01 1.73969418e-01 2.31852755e-02
-1.12478518e+00 -1.00375473e+00 -7.41858840e-01 6.41353950e-02
2.26982161e-01 -2.58693933e-01 1.12169933e+00 -1.08278489e+00
-1.42109561e+00 5.72026849e-01 -3.79290253e-01 -8.23547184e-01
5.15805244e-01 -5.61560504e-02 -5.44048965e-01 -6.97329491e-02
-2.15090513e-01 1.60706848e-01 1.10329211e+00 -9.77210283e-01
-3.80850405e-01 -3.16993505e-01 -2.11835638e-01 -4.18604642e-01
-6.51542485e-01 6.81759138e-03 9.22985971e-02 -9.07550216e-01
1.85339525e-01 -7.25946844e-01 -5.08369505e-01 -4.02877688e-01
-1.84941411e-01 -3.09843600e-01 7.84108520e-01 -6.96821928e-01
1.24476850e+00 -2.11056089e+00 2.23630190e-01 4.35898513e-01
1.04585037e-01 -2.11426005e-01 -3.32928784e-02 8.88523281e-01
-4.52736467e-01 -4.77437600e-02 -6.51419640e-01 -8.05220068e-01
-2.00623162e-02 3.01051795e-01 -6.94358408e-01 6.15108669e-01
3.63510221e-01 9.16175365e-01 -8.21713150e-01 -1.15382254e-01
4.66811866e-01 2.66916960e-01 -2.41958752e-01 2.26308420e-01
-2.79682726e-01 1.03846014e+00 -2.50237197e-01 7.43705511e-01
3.26901913e-01 -4.00036961e-01 3.16121817e-01 -1.39503628e-01
-1.77129760e-01 3.54715258e-01 -9.54549909e-01 1.73176742e+00
-5.69651246e-01 6.67215526e-01 -6.12931550e-01 -1.77040088e+00
1.04604506e+00 5.57789147e-01 1.22765625e+00 -7.08740294e-01
-4.83906679e-02 2.59779483e-01 -2.55561352e-01 -4.40699369e-01
1.06413342e-01 -2.75972813e-01 -2.57978857e-01 6.87491000e-01
3.38638335e-01 9.35157984e-02 1.86788380e-01 -1.35424197e-01
1.05680275e+00 3.78375947e-01 1.69509336e-01 1.10141914e-02
3.89575988e-01 3.98882292e-02 4.50029165e-01 6.00405812e-01
1.71333164e-01 3.61690968e-01 5.81858277e-01 -3.32335591e-01
-1.18215585e+00 -1.06351948e+00 -2.19296172e-01 1.15784514e+00
-6.37011707e-01 -4.08138514e-01 -1.83165282e-01 -6.20359123e-01
1.35641575e-01 5.22732079e-01 -6.08868599e-01 1.26042768e-01
-4.92388397e-01 -8.19714367e-01 5.56695580e-01 7.43893147e-01
-2.70724654e-01 -9.78285611e-01 -3.50318819e-01 5.48680961e-01
1.89674124e-02 -1.20092928e+00 2.82583833e-01 7.47931600e-01
-1.56037927e+00 -8.88881981e-01 -5.32749712e-01 -5.25196970e-01
5.91904879e-01 -1.93248078e-01 1.19744563e+00 -1.04594685e-01
-7.06971809e-02 6.23825490e-01 -3.44724596e-01 -4.90555167e-01
-1.56218126e-01 2.79261202e-01 1.35139599e-01 1.03076786e-01
2.84651250e-01 -1.27942622e+00 -2.86101103e-01 6.08991012e-02
-7.99222410e-01 -1.26202017e-01 3.24707329e-01 9.01838541e-01
5.82145452e-01 2.11232677e-01 1.14079618e+00 -8.38493586e-01
2.10500255e-01 -7.58587122e-01 -6.32946193e-01 3.94333899e-02
-8.60036254e-01 2.11362690e-01 9.36878681e-01 -4.45649594e-01
-8.21968317e-01 1.31171346e-01 4.81843986e-02 -5.22638977e-01
-2.31998637e-01 7.93520868e-01 4.76684302e-01 2.24826142e-01
3.54155600e-01 4.02570009e-01 2.27782816e-01 -8.84981513e-01
2.98862129e-01 4.57363576e-01 4.96389061e-01 -5.94781339e-01
9.85148728e-01 8.02739143e-01 3.37620348e-01 -6.15287840e-01
-6.68648183e-01 -5.44188976e-01 -8.26956630e-01 -1.23888291e-01
4.43886727e-01 -1.00019407e+00 -6.77010775e-01 3.22580963e-01
-5.89643538e-01 -4.43964332e-01 -3.82352591e-01 6.84118748e-01
-8.08839083e-01 1.74963266e-01 -4.48505372e-01 -1.18464804e+00
-5.78849204e-02 -6.23988807e-01 9.65643466e-01 -3.33419710e-01
-1.50019184e-01 -1.34264457e+00 6.84419356e-04 2.50913560e-01
4.41237360e-01 7.16847897e-01 1.04539287e+00 -8.78207982e-01
-4.17807788e-01 -3.45952988e-01 2.09239528e-01 3.35976899e-01
4.73928213e-01 -1.37487561e-01 -1.11738980e+00 -3.60172778e-01
-4.05254737e-02 -3.63738328e-01 8.97143900e-01 2.09916189e-01
1.29426169e+00 -1.35378554e-01 -1.52326122e-01 6.63255155e-01
1.29569304e+00 4.72863764e-02 4.43143219e-01 3.19447458e-01
4.77506787e-01 7.49445975e-01 6.28355742e-01 7.86223233e-01
3.96407425e-01 3.35997283e-01 4.01824564e-01 -2.29038656e-01
4.17271018e-01 -2.36963212e-01 2.32880160e-01 8.91049385e-01
-4.31370407e-01 2.00278163e-01 -8.22687745e-01 7.74907947e-01
-2.12718129e+00 -1.16417456e+00 -3.24529000e-02 2.55851007e+00
6.45711303e-01 1.96684837e-01 3.14646244e-01 7.50244141e-01
1.16897561e-01 1.63293332e-01 -6.26909137e-01 -1.03314452e-01
-8.51543173e-02 6.98368371e-01 5.17984986e-01 -1.11081496e-01
-1.25924277e+00 5.46918929e-01 7.02832222e+00 4.63794738e-01
-1.22058749e+00 1.03232771e-01 4.25115258e-01 -1.98601931e-01
-3.50451380e-01 1.15178019e-01 -1.81301460e-01 4.16179359e-01
1.45182753e+00 -1.24597996e-01 5.90555668e-01 5.66418111e-01
5.68877578e-01 2.35475078e-01 -1.36103284e+00 9.26024854e-01
-2.16973752e-01 -1.15597081e+00 -3.46822828e-01 -2.67899260e-02
7.37640142e-01 2.09824488e-01 7.22777024e-02 4.57132399e-01
2.18448637e-04 -1.22258985e+00 4.68633324e-01 5.34702122e-01
6.11082196e-01 -7.36763477e-01 5.69951594e-01 6.35711014e-01
-1.18209934e+00 -3.50574672e-01 5.69317155e-02 -3.85435253e-01
1.84830129e-02 8.18667352e-01 -4.46865112e-01 8.86357605e-01
4.22375053e-01 1.28062963e+00 -4.48665500e-01 8.58946562e-01
-3.49145271e-02 1.14426279e+00 -4.63456035e-01 4.77202773e-01
4.19528633e-02 -1.20421231e-01 3.52207989e-01 1.05982125e+00
4.34957594e-01 -2.60856301e-01 1.71064734e-01 4.13166583e-01
4.15273570e-02 1.08438022e-01 -7.09437251e-01 -4.97029692e-01
3.98725033e-01 9.89284992e-01 -4.69328970e-01 -4.41677302e-01
-6.58661604e-01 5.00474215e-01 3.16021860e-01 5.58346629e-01
-6.19683564e-01 2.15209890e-02 4.04128730e-01 -5.51421307e-02
5.00009000e-01 -4.91917998e-01 -3.70589167e-01 -1.30803549e+00
1.17863007e-01 -8.53889704e-01 6.00161195e-01 -5.03525019e-01
-1.51197827e+00 2.46186048e-01 9.29170102e-02 -1.76668191e+00
-9.28510368e-01 -4.40101147e-01 -3.94002467e-01 6.80735409e-01
-1.83065116e+00 -1.15704393e+00 4.02899347e-02 7.90612280e-01
3.65761966e-01 -3.99601877e-01 1.08513784e+00 3.85150582e-01
-4.79032278e-01 3.88666332e-01 6.58405960e-01 1.38029203e-01
5.52835941e-01 -1.11495531e+00 1.09583050e-01 7.43075132e-01
2.29477584e-01 5.81786573e-01 6.48037910e-01 -2.72578657e-01
-1.81933808e+00 -1.09435415e+00 8.75769556e-01 -4.76548284e-01
1.02124298e+00 -5.22479594e-01 -8.00689936e-01 8.93669963e-01
-1.58337921e-01 1.32668778e-01 9.74157691e-01 4.44340080e-01
-4.88695413e-01 -4.11625803e-01 -9.27510202e-01 1.33380979e-01
8.00338686e-01 -7.48043358e-01 -6.23312414e-01 3.14354151e-01
4.39657956e-01 -8.94212574e-02 -1.38742471e+00 6.39611840e-01
7.04992652e-01 -8.11331093e-01 1.04935718e+00 -8.33262086e-01
5.88667870e-01 -2.02324986e-01 -1.69482023e-01 -1.05288196e+00
6.31964877e-02 -7.32606173e-01 -7.17732549e-01 1.52363873e+00
4.93615896e-01 -7.91671693e-01 7.46342003e-01 4.45698410e-01
8.09142664e-02 -6.22447491e-01 -1.04512370e+00 -9.48052108e-01
4.76690009e-02 -1.05264699e+00 7.90902615e-01 1.10309064e+00
4.49123494e-02 -1.05193637e-01 -6.65214241e-01 -8.45873170e-03
9.10299778e-01 3.70119721e-01 7.62856245e-01 -1.50217116e+00
-4.27481472e-01 -2.22108856e-01 -4.82089221e-01 -7.25421250e-01
4.58180279e-01 -9.24489558e-01 -3.06892812e-01 -1.21438670e+00
-1.36276662e-01 -5.89764953e-01 -8.45378876e-01 6.86631501e-01
3.17889094e-01 3.42935801e-01 2.80502383e-02 5.61894596e-01
-2.50111639e-01 4.44088668e-01 8.12681198e-01 -1.79316521e-01
-3.17208499e-01 1.58143610e-01 -4.94953513e-01 4.91965979e-01
7.97234774e-01 -6.63864374e-01 -5.01765251e-01 -2.01363415e-01
-1.50438249e-01 3.75841498e-01 5.01856029e-01 -8.15453410e-01
5.40101677e-02 -1.75859585e-01 4.24221367e-01 -5.89115620e-01
3.11413020e-01 -1.11136782e+00 1.95381165e-01 2.25089371e-01
-2.97083259e-01 -8.11877847e-02 1.88098207e-01 7.21976817e-01
-3.16087574e-01 6.47948533e-02 2.74667084e-01 1.24555349e-01
-7.37756670e-01 2.23956719e-01 -5.26785910e-01 -2.03237265e-01
7.63201356e-01 -1.15115859e-01 -1.24333933e-01 -4.81372327e-01
-9.49774504e-01 6.19835220e-02 1.22709408e-01 4.58680421e-01
3.02002579e-01 -1.22154284e+00 -6.52963161e-01 4.30279106e-01
3.48110229e-01 -4.07077640e-01 -2.99317278e-02 1.13469827e+00
2.31042087e-01 3.64733785e-01 -3.79311815e-02 -7.82496870e-01
-8.45777154e-01 7.08752275e-01 1.24414358e-02 -5.93482852e-01
-7.22357333e-01 8.18407442e-03 -2.92800367e-01 -4.91778135e-01
3.34024698e-01 -3.88661593e-01 -2.85122216e-01 1.71180546e-01
2.47018382e-01 3.62532884e-01 3.21381152e-01 -4.06986266e-01
-3.80427688e-01 4.33379740e-01 3.06011319e-01 -5.24109006e-02
1.87881923e+00 8.28509554e-02 -1.49120882e-01 9.02822077e-01
1.11816978e+00 -2.09593505e-01 -1.06948948e+00 -2.00351700e-01
3.85955602e-01 -2.49284282e-01 -1.14542432e-01 -6.36714756e-01
-9.99251008e-01 8.04894507e-01 3.88140172e-01 3.85244131e-01
1.46364403e+00 -1.97788060e-01 4.26940888e-01 5.30681610e-01
6.52458966e-01 -8.52661729e-01 -1.79889500e-01 3.96323055e-01
5.32047272e-01 -1.31387901e+00 4.73258980e-02 -1.25199407e-01
-3.86795908e-01 1.14436221e+00 -5.76646961e-02 -2.90352672e-01
7.59224355e-01 4.21257228e-01 5.91191975e-03 1.16280615e-01
-1.18039322e+00 -4.45122927e-01 3.16785634e-01 7.75980413e-01
7.14749575e-01 -4.24998105e-02 -6.64958581e-02 6.32489085e-01
-1.07415989e-02 3.86192203e-01 2.37256736e-01 1.10009205e+00
7.44489506e-02 -1.37054312e+00 -2.29120150e-01 7.74338305e-01
-6.43042624e-01 -8.53343010e-02 8.92754346e-02 6.65721178e-01
-2.08944261e-01 1.24444544e+00 4.16113399e-02 -2.09023476e-01
2.78106898e-01 3.45744163e-01 3.94976884e-01 -3.07203799e-01
-8.39775562e-01 2.69433856e-01 1.70938522e-01 -3.36665988e-01
-8.10153484e-01 -7.46786237e-01 -1.05090129e+00 -1.21371023e-01
-5.48871607e-02 1.36786118e-01 5.77570975e-01 1.23362780e+00
2.34131426e-01 6.16413713e-01 1.09982193e+00 -9.47614551e-01
-6.42923236e-01 -9.79325473e-01 -4.39119786e-01 4.85624552e-01
6.97211504e-01 -7.39536047e-01 -5.37956595e-01 4.59409386e-01] | [7.1559014320373535, 3.0583484172821045] |
bb1ecc38-4612-4df1-b27e-257d560b6dec | diversity-matters-robustness-of-bias | 2302.14027 | null | https://arxiv.org/abs/2302.14027v1 | https://arxiv.org/pdf/2302.14027v1.pdf | Diversity matters: Robustness of bias measurements in Wikidata | With the widespread use of knowledge graphs (KG) in various automated AI systems and applications, it is very important to ensure that information retrieval algorithms leveraging them are free from societal biases. Previous works have depicted biases that persist in KGs, as well as employed several metrics for measuring the biases. However, such studies lack the systematic exploration of the sensitivity of the bias measurements, through varying sources of data, or the embedding algorithms used. To address this research gap, in this work, we present a holistic analysis of bias measurement on the knowledge graph. First, we attempt to reveal data biases that surface in Wikidata for thirteen different demographics selected from seven continents. Next, we attempt to unfold the variance in the detection of biases by two different knowledge graph embedding algorithms - TransE and ComplEx. We conduct our extensive experiments on a large number of occupations sampled from the thirteen demographics with respect to the sensitive attribute, i.e., gender. Our results show that the inherent data bias that persists in KG can be altered by specific algorithm bias as incorporated by KG embedding learning algorithms. Further, we show that the choice of the state-of-the-art KG embedding algorithm has a strong impact on the ranking of biased occupations irrespective of gender. We observe that the similarity of the biased occupations across demographics is minimal which reflects the socio-cultural differences around the globe. We believe that this full-scale audit of the bias measurement pipeline will raise awareness among the community while deriving insights related to design choices of data and algorithms both and refrain from the popular dogma of ``one-size-fits-all''. | ['Animesh Mukherjee', 'Soumya Sarkar', 'Bhanu Prakash Reddy Guda', 'Anirban Panda', 'Sai Keerthana Karnam', 'Paramita Das'] | 2023-02-27 | null | null | null | null | ['knowledge-graph-embedding'] | ['graphs'] | [ 3.00429165e-02 2.81610638e-01 -3.94967973e-01 -2.91977614e-01
-1.67885453e-01 -5.86136341e-01 7.99368203e-01 4.34574783e-01
-6.15696609e-01 4.56145138e-01 9.02877510e-01 -4.35449779e-01
-6.66005731e-01 -9.84933436e-01 -5.37723601e-01 -3.19611698e-01
9.59807411e-02 4.44413245e-01 -8.38009417e-02 -4.63598996e-01
5.13397574e-01 2.75426209e-01 -1.36200738e+00 -1.50066346e-01
6.86437011e-01 4.49131489e-01 -4.68926728e-01 4.31564897e-01
1.99614778e-01 6.65976167e-01 -6.36286676e-01 -9.99618709e-01
1.67272344e-01 -1.99405119e-01 -8.30041170e-01 -5.01956582e-01
9.21201766e-01 -1.92869648e-01 -5.29610038e-01 1.05502951e+00
8.27305794e-01 -2.42772311e-01 9.14518297e-01 -1.46024001e+00
-9.36876535e-01 8.05308998e-01 -3.86326611e-01 3.17732453e-01
1.61209658e-01 2.75354862e-01 1.06779218e+00 -4.43712741e-01
1.07921553e+00 1.42129445e+00 9.90741014e-01 1.47071302e-01
-1.45335495e+00 -8.87251496e-01 1.15425713e-01 1.79747909e-01
-1.29370904e+00 -5.16503751e-01 9.77579415e-01 -9.30350959e-01
6.20291710e-01 2.39182010e-01 7.94603944e-01 1.47308981e+00
2.17190579e-01 3.71804051e-02 1.51492965e+00 -5.44325948e-01
-1.98799353e-02 5.66203952e-01 5.51697969e-01 8.08698535e-01
1.17360747e+00 1.65844694e-01 -8.89234006e-01 -6.20423734e-01
3.95570993e-01 -2.85377502e-01 -7.24259615e-02 -6.90872490e-01
-1.32637596e+00 8.42840075e-01 4.67048287e-01 3.36871594e-01
-1.84274971e-01 3.67814124e-01 4.47797179e-01 3.92512262e-01
4.64935035e-01 9.10449028e-01 -3.15960556e-01 -3.98872465e-01
-6.91193581e-01 5.03036857e-01 9.47828174e-01 5.85898340e-01
7.69234776e-01 -2.42926896e-01 -1.53312579e-01 6.30525827e-01
1.87237665e-01 5.01933396e-01 2.27831677e-01 -7.45285571e-01
2.23342866e-01 8.36397886e-01 1.14171296e-01 -1.88791037e+00
-4.38579202e-01 -4.72147495e-01 -2.71751076e-01 1.03149995e-01
8.69866908e-01 -2.92875767e-01 -6.44788265e-01 1.80997145e+00
3.10854375e-01 -5.00031173e-01 -2.40972474e-01 8.15611660e-01
5.29574811e-01 -2.96331257e-01 2.87927270e-01 4.02041376e-01
1.65905023e+00 -1.22880936e-01 -7.03333855e-01 -5.17237008e-01
9.01252031e-01 -5.49896955e-01 1.23543942e+00 3.69754769e-02
-7.89079428e-01 -2.46785834e-01 -1.12065613e+00 -1.34661540e-01
-9.63052988e-01 -1.55909553e-01 8.01459551e-01 1.17399919e+00
-6.90568686e-01 5.89271724e-01 -3.98889691e-01 -8.94708753e-01
4.53566074e-01 2.04647765e-01 -2.50017673e-01 -5.22070639e-02
-1.50444496e+00 1.16674709e+00 1.71617344e-01 -2.48715609e-01
-2.95793384e-01 -9.90877450e-01 -4.36651140e-01 -2.30610594e-01
4.76351470e-01 -8.43191028e-01 6.55773759e-01 -9.88036633e-01
-7.37295151e-01 1.06501627e+00 1.22366384e-01 -3.02486327e-02
5.42611361e-01 -1.43687174e-01 -5.74531913e-01 -2.90474266e-01
1.99153736e-01 3.06693256e-01 6.89322293e-01 -1.28573668e+00
-2.81832129e-01 -7.19173193e-01 1.26354247e-01 3.74438129e-02
-7.68689215e-01 -4.83195186e-02 9.12387893e-02 -5.16758740e-01
-1.86450049e-01 -1.05064166e+00 1.48785532e-01 -2.62925446e-01
-1.81531459e-01 5.80204390e-02 4.17269081e-01 -7.68037081e-01
1.46397114e+00 -2.18782091e+00 1.79705843e-01 4.78468865e-01
4.14504200e-01 -2.26802006e-01 2.84419864e-01 8.21173787e-01
2.67146379e-01 6.07830167e-01 2.68092275e-01 -3.44829517e-03
4.46500182e-01 1.10401108e-03 -1.62869394e-01 7.27704823e-01
-1.34755799e-03 9.29082632e-01 -1.03360009e+00 -5.33973157e-01
-1.74047962e-01 3.26289207e-01 -5.92393041e-01 -3.75888079e-01
3.51223677e-01 -2.27332979e-01 -1.62388772e-01 8.11658084e-01
3.05564851e-01 1.30314101e-02 4.39982355e-01 -3.29053402e-01
-1.12903617e-01 4.69779611e-01 -8.73060167e-01 1.33581150e+00
-8.17461014e-02 7.76440918e-01 5.02843335e-02 -5.42555153e-01
7.55612791e-01 -2.28431880e-01 5.76829575e-02 -5.31950295e-01
1.21450424e-01 3.78491104e-01 4.84216601e-01 -1.74793333e-01
9.77952063e-01 -9.24934670e-02 -3.20769459e-01 5.61789691e-01
-1.16752587e-01 -3.53895798e-02 -3.24714393e-03 2.98334718e-01
1.18332994e+00 -2.78478384e-01 9.94536951e-02 -6.97187364e-01
-2.81962752e-01 4.54985261e-01 3.87554526e-01 8.48232448e-01
-4.73636985e-01 2.00398490e-01 8.49569559e-01 -2.05337524e-01
-8.88257265e-01 -8.26535583e-01 -3.76423150e-01 1.23135316e+00
4.95689828e-03 -4.09469664e-01 -4.25709873e-01 -9.06319976e-01
7.33292222e-01 7.57680774e-01 -1.01696515e+00 -6.26103461e-01
-1.25254869e-01 -8.28998029e-01 9.72409308e-01 3.25077474e-01
2.23442361e-01 -6.54159427e-01 -7.40953386e-01 -2.24169180e-01
-1.13371694e-02 -8.08493733e-01 -7.87218884e-02 -7.75479525e-03
-6.53296888e-01 -1.16925740e+00 -2.64101058e-01 -3.71515512e-01
6.96967900e-01 2.09680349e-02 1.41961730e+00 1.34549692e-01
-1.25547484e-01 7.64604092e-01 -2.75541276e-01 -9.53273118e-01
-2.97648817e-01 5.39239764e-01 1.41729921e-01 -3.76766056e-01
9.80917394e-01 -6.10328257e-01 -5.82740009e-01 1.74775779e-01
-6.14030480e-01 -3.32229942e-01 8.25544477e-01 5.24302602e-01
-9.51749310e-02 1.84320379e-02 6.52132809e-01 -1.28886056e+00
1.09786153e+00 -7.13265479e-01 2.06339569e-03 8.39286298e-02
-1.32655001e+00 3.26756164e-02 -1.53379530e-01 -4.27211344e-01
-6.97506607e-01 -7.47194231e-01 4.74083960e-01 2.26055190e-01
7.07704350e-02 5.70850551e-01 1.06126927e-01 -2.16271371e-01
1.02484894e+00 -3.38029921e-01 -2.69253310e-02 -2.14691713e-01
4.07646924e-01 8.88840199e-01 1.55889034e-01 -7.77585030e-01
8.30767751e-01 5.52172661e-01 -1.60401151e-01 -7.95916796e-01
-5.07999957e-01 -1.58364624e-01 -5.65192044e-01 -3.01051617e-01
6.06452167e-01 -7.12831140e-01 -5.69335639e-01 2.35732988e-01
-5.98057449e-01 -3.28917116e-01 -1.59862697e-01 4.49754804e-01
-8.75437930e-02 9.42672193e-02 -2.66260296e-01 -6.22611225e-01
-1.49190322e-01 -8.19883943e-01 7.41042554e-01 -9.35217738e-02
-1.05164504e+00 -1.07494354e+00 1.94841832e-01 6.01562619e-01
6.01780713e-01 4.12642628e-01 1.20950341e+00 -6.04144394e-01
-2.56424457e-01 -1.80594310e-01 -3.02313119e-01 -8.63602534e-02
3.24655056e-01 7.79088913e-03 -9.55576897e-01 -3.42864186e-01
-6.27980471e-01 -4.21134591e-01 6.62369728e-01 4.67597395e-02
6.87847316e-01 -1.32748857e-01 -4.62483287e-01 2.58336782e-01
1.38125932e+00 -2.25750744e-01 4.58366573e-01 7.75752485e-01
8.33618879e-01 9.02297795e-01 3.89357418e-01 1.19426008e-02
5.83243370e-01 4.35724407e-01 1.36069521e-01 8.65993127e-02
-9.53926146e-02 -4.45229799e-01 3.65735412e-01 5.52724242e-01
-2.14365631e-01 1.58125550e-01 -1.20726943e+00 8.89249563e-01
-1.57998955e+00 -7.81028569e-01 7.40524158e-02 2.21240878e+00
9.33415174e-01 3.32282096e-01 3.08803678e-01 2.37754211e-01
2.68085957e-01 3.55009556e-01 -3.52609456e-01 -5.97289681e-01
1.10433556e-01 -8.06734934e-02 9.85347927e-01 3.06034207e-01
-6.28069878e-01 6.43281400e-01 6.85809422e+00 2.55104065e-01
-9.59365547e-01 -1.38639584e-01 2.54913688e-01 -3.06199610e-01
-5.52857399e-01 8.30949098e-02 -6.62979484e-01 4.64467168e-01
9.49268043e-01 -3.81809115e-01 5.30259311e-01 6.99459732e-01
-1.78606942e-01 -3.90547067e-02 -1.12199974e+00 6.54301047e-01
1.73254132e-01 -9.73992348e-01 -1.97912499e-01 4.12350684e-01
6.37345850e-01 -1.44803256e-01 1.77140936e-01 4.23090786e-01
6.61101103e-01 -1.06758904e+00 8.23961973e-01 4.11300927e-01
6.24439955e-01 -7.54136443e-01 6.52447164e-01 -1.57279015e-01
-4.13358748e-01 -2.25040734e-01 -2.08633676e-01 -4.92873132e-01
-3.08241874e-01 7.71300435e-01 -9.34161127e-01 3.95692497e-01
8.99305403e-01 3.74385029e-01 -1.00677323e+00 1.90466434e-01
5.59572577e-02 7.86279142e-01 -1.20260501e-02 -1.75432205e-01
-1.73019111e-01 -7.96296224e-02 4.66694444e-01 1.36237705e+00
2.94831060e-02 -2.90981025e-01 -3.08326453e-01 8.35338950e-01
-2.16204315e-01 5.34768514e-02 -1.06206417e+00 -7.02660441e-01
7.72042930e-01 1.26630425e+00 -5.46770632e-01 -4.75185253e-02
-5.27485251e-01 4.76343751e-01 5.63850105e-01 4.35160249e-01
-3.56051475e-01 -4.95956481e-01 1.06856644e+00 4.63245809e-01
2.89009977e-02 -3.17682832e-01 -5.54894865e-01 -7.27724552e-01
-5.93522005e-02 -1.17151153e+00 3.54885429e-01 -5.46737373e-01
-1.46849036e+00 -1.92142963e-01 2.50860721e-01 -1.70534879e-01
6.10618703e-02 -6.79683685e-01 -1.41170219e-01 8.89255524e-01
-1.21059179e+00 -1.22329938e+00 -3.87632281e-01 1.28171176e-01
-4.34904784e-01 1.31432572e-02 4.79357332e-01 1.77584365e-01
-5.45106292e-01 7.29330361e-01 -9.93205979e-02 2.13238418e-01
1.27348113e+00 -1.27923834e+00 2.52010196e-01 5.04674613e-01
-3.54143903e-02 1.27362382e+00 8.26268613e-01 -9.84521806e-01
-1.52458680e+00 -6.63177371e-01 1.05632424e+00 -1.14367151e+00
9.30610001e-01 -4.95748192e-01 -6.18415654e-01 8.99012864e-01
2.07687885e-01 -4.58841175e-01 8.96373451e-01 1.09466171e+00
-5.10336995e-01 -1.16892233e-01 -1.13764715e+00 8.18479598e-01
1.51285744e+00 -7.89258718e-01 -7.10757136e-01 -8.82331505e-02
6.08257771e-01 -7.19588175e-02 -1.23400712e+00 2.41859645e-01
1.00389338e+00 -9.71173525e-01 9.41912591e-01 -8.61414492e-01
6.50672913e-01 2.60257900e-01 -1.79324329e-01 -1.51832688e+00
-8.16286027e-01 -1.21616013e-01 9.28765535e-02 1.60115576e+00
3.92261088e-01 -1.10162795e+00 7.78615415e-01 9.91070807e-01
1.87364757e-01 -7.43089497e-01 -5.50190330e-01 -4.59948003e-01
3.71410131e-01 -2.11292267e-01 8.06495667e-01 1.45602286e+00
9.68583822e-02 2.35528901e-01 -9.42940935e-02 8.88566077e-02
4.72912848e-01 -3.84234637e-02 1.14570940e+00 -1.47870457e+00
-5.89922592e-02 -4.43628371e-01 -6.78595126e-01 6.98309252e-03
3.97964520e-03 -9.29830253e-01 -5.67178786e-01 -1.44518065e+00
4.32440400e-01 -5.61734319e-01 -3.56705517e-01 2.99191773e-01
-3.08338642e-01 1.99605972e-02 3.30960937e-02 9.83460322e-02
-2.80243363e-02 9.33946893e-02 9.25198436e-01 -6.58273920e-02
-1.52128279e-01 -7.48535454e-01 -1.47613716e+00 6.59311295e-01
8.26000333e-01 -5.90884864e-01 -4.96994287e-01 -4.54883009e-01
9.97078657e-01 -8.87859225e-01 6.22977078e-01 -7.32604146e-01
-6.24374747e-02 -2.17205837e-01 3.55499685e-01 5.21472543e-02
1.40783340e-01 -7.67935634e-01 1.91376671e-01 4.41121489e-01
-3.20396245e-01 1.48373708e-01 3.52680683e-01 5.90312779e-01
3.00434470e-01 4.99414222e-04 7.37844408e-02 -7.46818483e-02
-5.72534144e-01 -2.97261328e-01 -1.70441866e-01 4.59370971e-01
6.07695222e-01 -2.54589766e-01 -8.33500624e-01 -2.55072236e-01
-1.61511879e-02 7.80429244e-02 9.74130571e-01 5.40051997e-01
-7.50770196e-02 -1.45798433e+00 -5.44221044e-01 -4.65571648e-03
4.47930783e-01 -6.60452843e-01 -2.17003748e-01 8.62108648e-01
-2.07729295e-01 3.14075947e-01 -3.33517581e-01 -5.33965677e-02
-1.32012308e+00 4.03247833e-01 4.22257595e-02 -9.41109732e-02
-1.24608271e-01 4.89711165e-01 -3.24982107e-01 -5.34006417e-01
-3.91455255e-02 -2.13695347e-01 3.77586111e-02 5.78191996e-01
-2.36607473e-02 9.69655931e-01 -6.58367053e-02 -5.26310325e-01
-5.19650936e-01 3.44425410e-01 -7.30686709e-02 -2.48442724e-01
1.11268926e+00 -2.33078562e-02 -9.42223445e-02 8.11116457e-01
8.87178659e-01 5.65548599e-01 -5.54645717e-01 6.30982071e-02
1.20898169e-02 -7.76291132e-01 9.81634632e-02 -9.49278116e-01
-6.16189957e-01 4.15366024e-01 5.74910700e-01 3.47636104e-01
5.98088026e-01 -1.27244607e-01 3.72283846e-01 3.29680681e-01
4.22391713e-01 -1.52016556e+00 -9.21676829e-02 7.03378320e-02
8.23930979e-01 -1.19535851e+00 5.68455637e-01 -2.81968951e-01
-5.90961635e-01 5.34926057e-01 4.70552117e-01 2.33191282e-01
5.65315962e-01 -1.32709339e-01 2.80171186e-01 -7.31500387e-01
-4.85827714e-01 -9.47674662e-02 2.95698762e-01 6.36175036e-01
6.42505109e-01 2.14120910e-01 -7.21551001e-01 4.74145383e-01
-7.72008479e-01 -6.79385662e-02 3.60492468e-01 9.63482022e-01
-6.28409609e-02 -1.12982666e+00 -4.45869356e-01 7.63617516e-01
-4.72134382e-01 1.22293718e-01 -1.31598198e+00 1.31034458e+00
2.10272580e-01 8.45545709e-01 -1.33529842e-01 -7.76944757e-01
5.03686190e-01 4.76309806e-01 5.98096967e-01 -4.34298903e-01
-6.89219832e-01 -8.79858971e-01 6.99542701e-01 -3.88398945e-01
-2.59742796e-01 -8.36647034e-01 -6.58681512e-01 -7.44183004e-01
-2.58288413e-01 -1.95226997e-01 6.00837290e-01 5.53806186e-01
7.29206383e-01 3.61660957e-01 -5.75989634e-02 -4.01253313e-01
-6.77804053e-01 -1.02625680e+00 -6.93906665e-01 8.19779932e-01
3.55904400e-02 -1.04038298e+00 -4.82787430e-01 -2.98346907e-01] | [9.271663665771484, 10.155893325805664] |
0868139a-dabb-456e-bd86-6b2d70a1ae99 | swift-markov-logic-for-probabilistic | 2210.00283 | null | https://arxiv.org/abs/2210.00283v1 | https://arxiv.org/pdf/2210.00283v1.pdf | Swift Markov Logic for Probabilistic Reasoning on Knowledge Graphs | We provide a framework for probabilistic reasoning in Vadalog-based Knowledge Graphs (KGs), satisfying the requirements of ontological reasoning: full recursion, powerful existential quantification, expression of inductive definitions. Vadalog is a Knowledge Representation and Reasoning (KRR) language based on Warded Datalog+/-, a logical core language of existential rules, with a good balance between computational complexity and expressive power. Handling uncertainty is essential for reasoning with KGs. Yet Vadalog and Warded Datalog+/- are not covered by the existing probabilistic logic programming and statistical relational learning approaches for several reasons, including insufficient support for recursion with existential quantification, and the impossibility to express inductive definitions. In this work, we introduce Soft Vadalog, a probabilistic extension to Vadalog, satisfying these desiderata. A Soft Vadalog program induces what we call a Probabilistic Knowledge Graph (PKG), which consists of a probability distribution on a network of chase instances, structures obtained by grounding the rules over a database using the chase procedure. We exploit PKGs for probabilistic marginal inference. We discuss the theory and present MCMC-chase, a Monte Carlo method to use Soft Vadalog in practice. We apply our framework to solve data management and industrial problems, and experimentally evaluate it in the Vadalog system. Under consideration in Theory and Practice of Logic Programming (TPLP). | ['Evgeny Sherkhonov', 'Emanuel Sallinger', 'Eleonora Laurenza', 'Luigi Bellomarini'] | 2022-10-01 | null | null | null | null | ['relational-reasoning'] | ['natural-language-processing'] | [-3.09764177e-01 5.51256001e-01 -3.26492697e-01 -3.95220101e-01
-3.40720087e-01 -4.78987366e-01 5.41396916e-01 3.33379768e-02
-5.29772863e-02 1.00699735e+00 -1.84808761e-01 -6.59850121e-01
-9.77293313e-01 -1.64866042e+00 -8.19023132e-01 -6.49024129e-01
-3.42892081e-01 1.31497371e+00 6.73728824e-01 -1.01384513e-01
-5.80358617e-02 6.80488884e-01 -1.94188774e+00 4.24214989e-01
4.42488611e-01 9.49405491e-01 -2.42100522e-01 3.80056471e-01
-3.94608080e-01 1.43729985e+00 -8.66591260e-02 -5.75233996e-01
-1.84948161e-01 8.73299539e-02 -1.09488022e+00 -7.70162880e-01
-2.18304042e-02 -6.95373043e-02 -3.05174828e-01 1.27055824e+00
-1.73940156e-02 5.50797358e-02 9.15949821e-01 -1.79995418e+00
-5.10761797e-01 1.48550856e+00 -2.67180324e-01 -9.59302038e-02
7.02063024e-01 -4.09457862e-01 1.22818041e+00 -4.57293034e-01
5.38625777e-01 1.75904751e+00 5.79686761e-01 2.89806992e-01
-1.15411818e+00 -3.39242458e-01 -2.94318020e-01 5.55601358e-01
-1.95400417e+00 1.60815880e-01 3.75201672e-01 -4.61650461e-01
1.07052946e+00 4.77889538e-01 3.55270028e-01 6.70105517e-01
4.66341972e-01 6.80406094e-01 1.09050727e+00 -9.15540516e-01
7.34695673e-01 3.64206880e-01 8.19569111e-01 8.89787018e-01
7.38135219e-01 2.71117896e-01 -6.48011208e-01 -7.13578165e-01
4.95653629e-01 -3.65698993e-01 1.68451265e-01 -5.87293565e-01
-5.95348477e-01 9.49285507e-01 -3.31659228e-01 5.83909415e-02
-2.71583557e-01 4.55324233e-01 3.08429152e-01 3.71148624e-02
-2.97025919e-01 -2.63054490e-01 -6.45189047e-01 1.50544971e-01
-5.39295495e-01 6.64688349e-01 1.37970722e+00 1.18226647e+00
7.47715831e-01 -1.81990385e-01 -2.80808434e-02 2.11425781e-01
1.05829120e+00 6.98085964e-01 8.99926350e-02 -1.07191730e+00
-1.58572599e-01 5.49952388e-01 -5.29981218e-02 -6.93071961e-01
-2.35781595e-01 3.83378744e-01 -4.62477088e-01 4.34547782e-01
2.20878735e-01 3.09333444e-01 -6.53725028e-01 1.78963292e+00
5.06352305e-01 -2.59460323e-02 6.21230125e-01 1.96928486e-01
7.69642174e-01 6.72483206e-01 2.35598892e-01 -3.22077036e-01
1.67703497e+00 8.04146305e-02 -8.62848639e-01 4.87099677e-01
7.24075556e-01 -9.38462019e-02 8.62800956e-01 9.34910238e-01
-9.73977149e-01 8.23472217e-02 -1.01267195e+00 -7.19655352e-03
-6.79957569e-01 -4.96351451e-01 1.07837594e+00 1.18912876e+00
-7.37849653e-01 2.92689145e-01 -9.33447301e-01 -5.86605668e-02
7.50627145e-02 2.92826235e-01 -2.33636513e-01 -1.82758376e-01
-1.57382071e+00 1.04439378e+00 1.43866193e+00 -1.32952407e-01
-8.88769686e-01 -5.87460279e-01 -9.72264528e-01 7.14783296e-02
8.00568759e-01 -6.40074968e-01 9.86783862e-01 1.25386938e-01
-1.47700584e+00 8.86300802e-01 2.28120431e-01 -7.25214005e-01
3.73106509e-01 1.04349554e-01 -7.04607308e-01 -2.64245957e-01
-2.56566912e-01 -7.42927417e-02 2.13726223e-01 -1.18897831e+00
-6.48097336e-01 -6.17472768e-01 5.53202033e-01 -2.47198597e-01
4.97654229e-01 -5.09183668e-02 -1.09993830e-01 8.64251181e-02
2.54855603e-01 -6.84334338e-01 4.74347919e-02 -6.91326916e-01
-5.89884520e-01 -6.26391351e-01 5.87317169e-01 -1.10105999e-01
1.27216280e+00 -1.93710744e+00 1.40117094e-01 9.32936609e-01
-6.29526451e-02 -1.60876527e-01 7.94385731e-01 4.79202479e-01
4.74903286e-02 1.41390339e-01 -2.17423439e-01 3.61470670e-01
7.10789680e-01 9.01461065e-01 -7.31072962e-01 3.57525945e-01
-3.62421811e-01 6.22089565e-01 -7.37983644e-01 -1.17359507e+00
3.75203967e-01 1.09156236e-01 -6.05569243e-01 -8.11336264e-02
-9.90325809e-01 -6.71429157e-01 -4.43253517e-01 8.74326885e-01
8.55339229e-01 9.91135463e-02 8.91517222e-01 -2.66942054e-01
1.13544025e-01 2.29064114e-02 -1.98683345e+00 1.14674079e+00
-2.24114448e-01 -3.95940602e-01 -2.64814049e-01 -5.79509079e-01
7.71975875e-01 1.97988614e-01 -3.20598930e-02 -6.93821758e-02
2.43768960e-01 1.79975614e-01 -4.27307755e-01 -4.83383387e-01
4.37446237e-01 -5.75584233e-01 -5.47001183e-01 3.91017109e-01
2.57475942e-01 -5.30130446e-01 2.88368136e-01 5.88243127e-01
8.85811687e-01 4.25498605e-01 6.88354373e-01 -6.06887639e-01
5.77166140e-01 8.01117569e-02 7.09753990e-01 8.65724623e-01
2.16510385e-01 -1.84103116e-01 8.99020016e-01 -2.57025868e-01
-4.89385724e-01 -1.63182068e+00 -6.25488698e-01 1.00793588e+00
4.30151895e-02 -7.26393461e-01 -5.19130170e-01 -4.87832218e-01
3.10938567e-01 1.40281272e+00 -4.20840979e-01 -1.66642562e-01
9.26660523e-02 -1.01527667e+00 1.00942612e+00 3.05034637e-01
2.45782301e-01 -9.81563568e-01 -4.14526284e-01 7.66470358e-02
1.12383053e-01 -1.00957823e+00 7.34138966e-01 4.97924268e-01
-6.94070220e-01 -1.29739416e+00 7.51610219e-01 -1.98844329e-01
3.67860124e-02 -6.49843574e-01 1.34539580e+00 -2.59531617e-01
-1.90356374e-01 5.19688845e-01 -1.97170809e-01 -6.48022294e-01
-7.04434514e-01 -5.15246272e-01 4.17809129e-01 -4.80903655e-01
7.64479637e-01 -8.12535703e-01 4.06657845e-01 1.70929044e-01
-1.17696834e+00 -3.02441925e-01 2.10477948e-01 5.30569375e-01
9.24367845e-01 1.12066734e+00 1.83372155e-01 -1.38671565e+00
2.30497465e-01 -5.73543727e-01 -1.22436202e+00 7.38600254e-01
-5.78997552e-01 5.11242509e-01 2.11349413e-01 1.31787136e-01
-1.24819112e+00 -1.82725012e-01 2.74770316e-02 -2.39795640e-01
-2.82228086e-02 1.03113091e+00 -8.37111712e-01 1.82949036e-01
7.67943263e-01 -8.30811933e-02 -3.70385677e-01 -6.28892407e-02
6.57354832e-01 5.63181818e-01 6.39387608e-01 -1.57938445e+00
5.77406883e-01 5.89319110e-01 5.98984540e-01 -5.06532669e-01
-8.82321000e-01 -4.07531820e-02 -3.77989888e-01 8.63548145e-02
6.21505797e-01 -6.72233582e-01 -1.49694419e+00 1.70824632e-01
-8.00765932e-01 -1.66513562e-01 -5.25367320e-01 5.70152760e-01
-9.59145606e-01 4.42972243e-01 -3.94772410e-01 -1.51636863e+00
3.10557261e-02 -7.15840816e-01 5.89730799e-01 -1.61687091e-01
-3.48353907e-02 -9.65595186e-01 2.56808132e-01 2.33298093e-01
-3.61979723e-01 4.42642272e-01 1.38458550e+00 -6.73282743e-01
-7.24165440e-01 -2.77903706e-01 -3.89749371e-02 3.17986459e-01
-3.55550140e-01 4.99628097e-01 -9.15051699e-01 3.56350392e-01
-1.88838005e-01 -3.57518971e-01 3.43583733e-01 1.09680004e-01
1.11912966e+00 -1.25348479e-01 -2.55083025e-01 1.34818986e-01
1.91892517e+00 1.40270114e-01 9.79722083e-01 1.25769675e-01
3.25904101e-01 3.97584379e-01 7.28921950e-01 5.93397081e-01
6.86500072e-01 4.49192584e-01 5.30656099e-01 1.05056489e+00
2.66622156e-01 -4.51107234e-01 2.62921005e-01 6.13170862e-01
-4.02866632e-01 7.00378940e-02 -1.12221062e+00 3.37848544e-01
-2.03690553e+00 -1.22982740e+00 -4.28891182e-01 2.28382707e+00
1.19333804e+00 4.68728960e-01 -6.96913972e-02 4.08795565e-01
6.17950201e-01 -3.02105874e-01 -4.47181314e-02 -5.24625003e-01
-3.03767830e-01 4.03354138e-01 5.49460888e-01 9.59850669e-01
-7.52427101e-01 8.22369516e-01 6.03414488e+00 1.17652833e+00
-1.32854849e-01 2.00135320e-01 -1.63007379e-01 2.86116511e-01
-7.33921647e-01 3.82118195e-01 -1.16973448e+00 2.85978258e-01
1.27882779e+00 -3.81597906e-01 6.14436626e-01 1.25030077e+00
-4.31018323e-01 -5.24832666e-01 -1.27076244e+00 8.91096473e-01
-8.82349461e-02 -1.32008433e+00 1.84125200e-01 2.65264511e-02
4.40853924e-01 -1.77377462e-01 -4.44725513e-01 6.39926255e-01
1.34562385e+00 -1.00332737e+00 9.89167452e-01 8.56077611e-01
5.49188852e-01 -1.08740592e+00 1.09598899e+00 2.92635381e-01
-8.76794577e-01 -8.08680803e-03 -4.20424044e-01 9.06267166e-02
1.83837563e-01 1.06970704e+00 -5.51705360e-01 1.23797691e+00
7.63396680e-01 -2.16764078e-01 -6.33512661e-02 4.55816507e-01
-3.72128904e-01 3.76408815e-01 -8.83899331e-01 1.78074539e-02
-2.32357576e-01 -2.46646419e-01 1.82572082e-01 1.03096735e+00
1.58711895e-01 2.99738795e-01 1.48940999e-02 1.01488507e+00
2.65839040e-01 -2.69659549e-01 -6.76107645e-01 8.54322910e-02
7.51468778e-01 8.58698726e-01 -4.54084814e-01 -4.00102049e-01
-6.15514554e-02 4.75681350e-02 1.36951163e-01 6.03581853e-02
-1.05482912e+00 -3.52293819e-01 2.38157213e-01 -1.54613303e-02
-6.59120679e-02 -1.28404796e-02 -2.67135501e-01 -9.87730324e-01
-2.19389901e-01 -7.20826387e-01 1.13811326e+00 -8.63759935e-01
-1.42108905e+00 2.92307615e-01 9.38432992e-01 -3.96739990e-01
-3.69539171e-01 -1.03775048e+00 -2.24547163e-02 6.93283319e-01
-1.20002401e+00 -1.29672587e+00 3.04327309e-02 9.46726501e-01
-6.33223593e-01 -3.65715884e-02 1.17780566e+00 6.86950237e-02
-3.96738231e-01 2.85988957e-01 -2.66501874e-01 -4.80246931e-01
8.88177976e-02 -1.61532366e+00 -4.85318869e-01 5.58320343e-01
4.19008806e-02 7.86041319e-01 1.06812584e+00 -7.66867697e-01
-1.82222223e+00 -9.49720860e-01 7.96995580e-01 -8.45921874e-01
8.85633230e-01 -2.80087441e-01 -8.29794884e-01 1.27034843e+00
-4.92618054e-01 6.00486435e-02 7.76542842e-01 6.18379116e-01
-7.56697714e-01 -2.69064426e-01 -1.75110185e+00 4.42157537e-01
9.34913576e-01 -8.12219918e-01 -1.27704310e+00 3.73110741e-01
7.88036466e-01 -3.43625844e-01 -1.29910982e+00 5.49044967e-01
6.38475478e-01 -1.02202690e+00 9.72166240e-01 -4.74492937e-01
-1.63744226e-01 -8.48598063e-01 -1.04725826e+00 -3.13211620e-01
-9.76620466e-02 -3.34007382e-01 -7.01842129e-01 1.20812762e+00
2.22737655e-01 -9.13259447e-01 6.72435880e-01 8.45294833e-01
1.74226910e-01 -1.88697621e-01 -1.15974832e+00 -9.22786474e-01
-1.51191372e-02 -1.14319217e+00 9.53168154e-01 8.30786705e-01
5.50248265e-01 -4.71232384e-02 -2.79891118e-02 7.55402267e-01
1.07500184e+00 1.57142758e-01 4.77246851e-01 -1.55938673e+00
-6.65270627e-01 1.04698949e-01 -8.44917357e-01 1.09547243e-01
4.06517059e-01 -9.67131376e-01 -5.95004447e-02 -1.40676284e+00
2.87492186e-01 -7.22752631e-01 -2.02777877e-01 6.43033564e-01
5.03352821e-01 -1.20655686e-01 -1.86534762e-01 -1.39957756e-01
-6.60027266e-01 2.88112938e-01 3.35622013e-01 1.71726704e-01
5.86924562e-03 -1.66925579e-01 -4.09839660e-01 1.00631404e+00
4.74036396e-01 -4.67903584e-01 -7.21298397e-01 5.08492589e-01
1.09287441e+00 1.22223616e-01 5.41356981e-01 -9.95497525e-01
3.19136739e-01 -5.66419780e-01 -3.52214009e-01 -1.04183424e+00
2.26744801e-01 -9.54122841e-01 9.35600460e-01 2.52817273e-01
1.46266133e-01 -3.91661495e-01 1.78974763e-01 6.02374792e-01
-9.87531617e-02 -8.25960159e-01 6.64181173e-01 -9.21615139e-02
-9.02622581e-01 -5.30585907e-02 -3.49008173e-01 -1.49599403e-01
1.19740808e+00 1.54173017e-01 -2.89841861e-01 2.31250718e-01
-1.07835388e+00 4.52338867e-02 1.68122992e-01 -4.01898116e-01
3.81745785e-01 -1.23692107e+00 -3.20131123e-01 -4.38758247e-02
1.44108579e-01 2.11297214e-01 3.98552865e-01 4.55347508e-01
-7.20082223e-01 4.23244774e-01 -3.46314944e-02 -2.34812990e-01
-8.32997203e-01 9.77726221e-01 2.03197390e-01 -4.26228613e-01
-4.05085444e-01 6.08455300e-01 -1.96832065e-02 -8.66688371e-01
3.39195073e-01 -5.61626792e-01 -9.62954834e-02 -1.00560434e-01
5.37449419e-01 5.54441035e-01 7.64917135e-02 -1.10365413e-01
-6.55284286e-01 1.14681952e-01 2.31583491e-01 -2.49857783e-01
9.41212952e-01 4.12175152e-03 -8.24926853e-01 9.33863819e-01
2.08465204e-01 1.04760848e-01 -2.46879518e-01 -1.65629297e-01
2.65245378e-01 -9.64498371e-02 6.89642876e-02 -1.03881025e+00
-3.63696307e-01 2.52669811e-01 3.64975840e-01 5.29902577e-01
6.40407562e-01 4.17242140e-01 -3.50864194e-02 7.98959017e-01
1.16531324e+00 -1.00961590e+00 -7.51831055e-01 5.50284564e-01
7.40530610e-01 -6.15779042e-01 2.55651504e-01 -8.06977153e-01
-3.88017327e-01 1.04279566e+00 2.70986289e-01 5.97567856e-02
9.88380671e-01 7.39515007e-01 -5.74640691e-01 -6.25770509e-01
-7.43610203e-01 -1.41915038e-01 -3.84605139e-01 6.70593083e-01
-1.22856714e-01 6.36804223e-01 -1.70956716e-01 1.21777201e+00
-4.20725048e-01 6.84486628e-01 6.35239959e-01 1.18865800e+00
-4.62914765e-01 -1.15306830e+00 -7.23785698e-01 2.34482929e-01
-4.33871001e-01 -7.31016174e-02 2.29031712e-01 1.06127501e+00
3.78601730e-01 6.84511304e-01 -1.33815616e-01 -1.80043399e-01
2.78597295e-01 3.45868796e-01 8.65672171e-01 -6.26628578e-01
1.74769774e-01 -6.52716458e-01 5.60042441e-01 -2.93768436e-01
-3.37538928e-01 -6.12387419e-01 -1.66554224e+00 -7.63635397e-01
-3.53187680e-01 4.79904324e-01 5.59792578e-01 9.68764544e-01
-7.45171234e-02 4.19357210e-01 -1.12040929e-01 1.03937869e-03
-7.62588084e-01 -5.48366308e-01 -1.46939826e+00 -3.10453773e-01
-5.20284534e-01 -1.03490293e+00 -4.60620522e-01 -4.69883196e-02] | [8.593924522399902, 6.742289066314697] |
f16cde17-8b04-49a0-b74c-cf4007b192a7 | motiontrack-learning-robust-short-term-and | 2303.10404 | null | https://arxiv.org/abs/2303.10404v2 | https://arxiv.org/pdf/2303.10404v2.pdf | MotionTrack: Learning Robust Short-term and Long-term Motions for Multi-Object Tracking | The main challenge of Multi-Object Tracking~(MOT) lies in maintaining a continuous trajectory for each target. Existing methods often learn reliable motion patterns to match the same target between adjacent frames and discriminative appearance features to re-identify the lost targets after a long period. However, the reliability of motion prediction and the discriminability of appearances can be easily hurt by dense crowds and extreme occlusions in the tracking process. In this paper, we propose a simple yet effective multi-object tracker, i.e., MotionTrack, which learns robust short-term and long-term motions in a unified framework to associate trajectories from a short to long range. For dense crowds, we design a novel Interaction Module to learn interaction-aware motions from short-term trajectories, which can estimate the complex movement of each target. For extreme occlusions, we build a novel Refind Module to learn reliable long-term motions from the target's history trajectory, which can link the interrupted trajectory with its corresponding detection. Our Interaction Module and Refind Module are embedded in the well-known tracking-by-detection paradigm, which can work in tandem to maintain superior performance. Extensive experimental results on MOT17 and MOT20 datasets demonstrate the superiority of our approach in challenging scenarios, and it achieves state-of-the-art performances at various MOT metrics. | ['Wei Tang', 'Gang Hua', 'Jinghai Duan', 'Le Wang', 'Sanping Zhou', 'Zheng Qin'] | 2023-03-18 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Qin_MotionTrack_Learning_Robust_Short-Term_and_Long-Term_Motions_for_Multi-Object_Tracking_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Qin_MotionTrack_Learning_Robust_Short-Term_and_Long-Term_Motions_for_Multi-Object_Tracking_CVPR_2023_paper.pdf | cvpr-2023-1 | ['motion-prediction'] | ['computer-vision'] | [-3.54670018e-01 -7.17820764e-01 -5.53836748e-02 -8.00252259e-02
-7.24974811e-01 -5.73565185e-01 4.36951250e-01 -1.95765615e-01
-2.83662796e-01 4.53239083e-01 -1.34627447e-01 3.23109001e-01
-5.03729992e-02 -4.04549152e-01 -7.82226682e-01 -9.36898708e-01
-2.01740488e-01 5.23953795e-01 1.19262719e+00 3.31644737e-03
-1.10409960e-01 4.87016290e-01 -1.55482948e+00 5.45623749e-02
6.95738792e-01 8.62308621e-01 3.73800993e-01 8.90871763e-01
1.42625228e-01 8.02570939e-01 -5.76059222e-01 -1.82300553e-01
1.86194301e-01 -1.92778215e-01 -2.52704352e-01 1.78151906e-01
7.06252337e-01 -5.06341457e-01 -7.19768763e-01 7.68024504e-01
4.54252869e-01 1.98783755e-01 4.85176295e-01 -1.43039048e+00
-4.43914503e-01 1.18144780e-01 -8.47221375e-01 4.79184091e-01
5.10699272e-01 6.28918648e-01 5.35379589e-01 -8.19410384e-01
5.41397810e-01 1.66665649e+00 9.66766834e-01 6.72908306e-01
-9.69986916e-01 -8.95186186e-01 5.11348844e-01 2.30326459e-01
-1.50803769e+00 -4.91262734e-01 3.42476368e-01 -6.24474406e-01
2.87219971e-01 2.24515885e-01 7.58369386e-01 9.70635355e-01
4.45762038e-01 9.28234458e-01 5.85850000e-01 2.05371976e-01
-1.85986653e-01 -2.64378220e-01 -1.32213250e-01 7.55137324e-01
3.65600348e-01 2.49432415e-01 -4.73512530e-01 -2.75624871e-01
6.41491055e-01 5.07451952e-01 -2.08312184e-01 -3.81172091e-01
-1.52310550e+00 4.82091814e-01 5.88728070e-01 2.33721614e-01
-2.24135920e-01 4.77799863e-01 1.56150594e-01 1.97055697e-01
3.10128301e-01 -3.70312750e-01 -1.35796621e-01 -7.10099610e-03
-8.69227469e-01 4.07189727e-01 4.63335931e-01 1.27889347e+00
8.33002985e-01 -9.99762490e-02 -7.22996473e-01 2.98707098e-01
6.24666452e-01 1.06399333e+00 2.64514536e-01 -7.80166030e-01
3.61202687e-01 4.91252720e-01 6.42169893e-01 -1.15119648e+00
-3.40812594e-01 -2.52722532e-01 -7.04644263e-01 9.66343284e-02
5.35608590e-01 -1.58608705e-01 -7.00210512e-01 1.60681951e+00
9.96211112e-01 9.44374025e-01 -1.17654599e-01 9.45832431e-01
7.19255507e-01 6.22409582e-01 6.12578951e-02 -3.14697087e-01
1.03469682e+00 -1.20042479e+00 -6.19434476e-01 -2.21850067e-01
5.17938256e-01 -6.33917689e-01 3.89316142e-01 -4.04014699e-02
-9.30343449e-01 -8.79957378e-01 -6.67451739e-01 3.72488141e-01
1.17008336e-01 1.46893829e-01 2.41772309e-01 2.34100282e-01
-9.09480214e-01 6.58087432e-01 -1.14682579e+00 -2.78023332e-01
4.06427175e-01 5.20862699e-01 -2.58687437e-01 -2.50592500e-01
-6.33644223e-01 5.96489489e-01 4.03297544e-01 2.28409916e-01
-1.30609083e+00 -5.47599733e-01 -7.18006790e-01 -4.93035406e-01
3.50745171e-01 -7.16774344e-01 1.14368176e+00 -6.17684960e-01
-1.15788269e+00 4.22141433e-01 -5.39146841e-01 -1.65234238e-01
8.13262284e-01 -3.89304459e-01 -6.36845887e-01 7.90478475e-03
4.58294690e-01 6.09136701e-01 1.03043151e+00 -1.31361067e+00
-1.27121782e+00 -3.04881424e-01 -3.83093148e-01 1.06770337e-01
-1.19835719e-01 1.59679681e-01 -1.02188671e+00 -5.91256440e-01
4.46527191e-02 -1.27670395e+00 -2.23921046e-01 4.31583405e-01
-9.70350131e-02 -4.38477308e-01 1.32766557e+00 -4.14493531e-01
1.20936108e+00 -2.09107351e+00 2.58975595e-01 -2.74558574e-01
3.91794860e-01 3.78677517e-01 -2.46622294e-01 1.09850794e-01
6.86125398e-01 -5.09547710e-01 2.16673911e-01 -6.72088563e-01
-1.76072806e-01 2.48898670e-01 -2.16427669e-01 9.27843630e-01
1.09160684e-01 1.12611306e+00 -1.42822492e+00 -7.57612586e-01
2.70588666e-01 4.66944128e-01 -1.97203606e-01 4.66453493e-01
-2.94945568e-01 1.05677676e+00 -7.61700928e-01 9.22406495e-01
8.58756006e-01 -3.99909824e-01 -1.66671365e-01 -4.66507822e-02
-3.06793511e-01 -4.40093666e-01 -1.27483296e+00 1.63116968e+00
-1.09562045e-03 5.63493848e-01 2.52516586e-02 -3.52992088e-01
8.06028426e-01 1.38510704e-01 7.74919152e-01 -2.68605351e-01
-5.70045188e-02 2.33020291e-01 -3.17436814e-01 -5.37960231e-01
5.72906256e-01 2.42430687e-01 1.16506480e-02 9.65429917e-02
-2.25404561e-01 5.03359377e-01 9.95679647e-02 3.21857408e-02
1.26789927e+00 4.29118425e-01 -1.21815123e-01 1.71894282e-01
5.21298945e-01 1.32483080e-01 1.07129109e+00 8.30508947e-01
-5.95287561e-01 4.33065712e-01 -4.47430164e-01 -6.25406921e-01
-6.94793582e-01 -1.07196367e+00 1.31679252e-01 1.29131162e+00
9.87035334e-01 -1.09143667e-01 -6.27059862e-03 -8.17871213e-01
1.89886093e-01 -1.56856209e-01 -4.03676063e-01 8.87977332e-03
-1.09287620e+00 -6.66049719e-01 4.31377798e-01 6.17662251e-01
4.13839966e-01 -8.88495266e-01 -5.25157094e-01 4.89798278e-01
-3.81505787e-01 -1.27201533e+00 -8.62864137e-01 -4.30259109e-01
-6.98174477e-01 -1.02940094e+00 -9.52979028e-01 -7.94070065e-01
4.47895348e-01 1.02782834e+00 8.39763820e-01 4.14503872e-01
-4.99913320e-02 4.24201012e-01 -3.55607420e-01 -2.44331390e-01
-3.29768628e-01 -3.68072808e-01 3.31370592e-01 2.68243313e-01
1.80215046e-01 -3.64455223e-01 -8.63018751e-01 9.56532836e-01
-5.92438698e-01 -2.61546046e-01 5.96986771e-01 5.49057305e-01
7.48150468e-01 -1.00081220e-01 3.32290143e-01 -1.06684782e-01
-2.79303730e-01 -7.37771034e-01 -7.37949848e-01 3.27561051e-01
-6.34180829e-02 -2.25584120e-01 2.69004434e-01 -1.00149763e+00
-7.96533704e-01 4.99533057e-01 2.49474779e-01 -9.38629031e-01
-1.20287850e-01 -2.51302958e-01 -1.47439957e-01 -5.67193329e-01
2.58222908e-01 4.74102974e-01 -1.70197234e-01 -5.40861130e-01
3.63954961e-01 3.65390331e-01 8.48678887e-01 -2.76072383e-01
1.35312355e+00 1.00584078e+00 8.63119289e-02 -5.89491308e-01
-9.26313221e-01 -1.09148991e+00 -7.98732519e-01 -7.06330240e-01
8.26947451e-01 -1.49453878e+00 -8.96046519e-01 6.53711557e-01
-1.24238527e+00 -2.87468314e-01 -2.28355192e-02 5.14282882e-01
-3.35324943e-01 6.83951080e-01 -3.96000892e-01 -8.33436608e-01
-1.80652693e-01 -1.00028872e+00 1.62173975e+00 5.59296906e-01
2.14983746e-01 -9.13918257e-01 4.04351562e-01 6.52308064e-03
7.47283250e-02 5.14536738e-01 -2.51038790e-01 -3.14067543e-01
-1.14506233e+00 -2.52641052e-01 -1.45426452e-01 -2.93401837e-01
-2.22910363e-02 -3.38550508e-02 -7.62581766e-01 -9.12806630e-01
-2.42743298e-01 -1.07863620e-01 1.02121603e+00 4.58338588e-01
6.02802038e-01 -2.26190731e-01 -1.18531621e+00 7.91885555e-01
1.29632890e+00 2.78556836e-03 3.91398668e-01 2.57797778e-01
1.11671972e+00 2.80796528e-01 1.26640737e+00 4.70701069e-01
6.98682666e-01 1.12143028e+00 5.67257047e-01 3.52485627e-01
-3.18378299e-01 -1.87716439e-01 8.75920713e-01 7.94274151e-01
-1.10640913e-01 -2.64404565e-01 -5.82126617e-01 7.18930066e-01
-2.44228101e+00 -1.22371495e+00 -5.72635889e-01 2.17655110e+00
3.24379534e-01 -4.92714755e-02 5.49507618e-01 -4.68334347e-01
1.04061961e+00 1.99828938e-01 -5.63284159e-01 8.22997332e-01
-9.88231003e-02 -7.29774475e-01 4.94022369e-01 3.45767945e-01
-1.36299312e+00 9.49019194e-01 5.86505795e+00 1.03072679e+00
-8.06919336e-01 3.73061866e-01 -1.88092679e-01 -2.14792475e-01
2.43010908e-01 4.58338149e-02 -1.54887998e+00 8.64870727e-01
7.60618627e-01 9.73606855e-02 3.57152745e-02 6.81102633e-01
2.24600762e-01 8.05892423e-03 -9.90935624e-01 1.02284658e+00
-5.37883304e-02 -1.27119887e+00 -1.78488940e-01 4.05188054e-02
7.71024227e-01 2.89077759e-01 -1.71554953e-01 3.89286965e-01
4.17392135e-01 -4.94528621e-01 1.01936638e+00 7.84178197e-01
5.95901549e-01 -4.44228977e-01 4.24175262e-01 5.02739251e-01
-2.27379680e+00 -2.51785517e-01 -6.26070619e-01 1.42573431e-01
4.18808758e-01 3.18308353e-01 -3.78949434e-01 7.76688993e-01
8.45319688e-01 1.26525903e+00 -7.05845773e-01 1.65803230e+00
1.92098022e-01 1.61755517e-01 -3.98308456e-01 -1.01416400e-02
2.44887576e-01 2.59616543e-02 1.04703808e+00 1.23467946e+00
5.03636003e-01 1.98919438e-02 9.29728270e-01 4.30435151e-01
3.08191240e-01 -3.32293957e-01 -5.92876434e-01 3.92737776e-01
7.23856032e-01 1.30821586e+00 -6.05639338e-01 -3.54958504e-01
-4.59808558e-01 1.00861156e+00 3.22163939e-01 1.70618623e-01
-1.26160789e+00 6.53724521e-02 8.36793661e-01 6.19988963e-02
7.24425912e-01 -2.54993200e-01 6.39854193e-01 -1.24532068e+00
4.16681409e-01 -3.81298512e-01 5.40305257e-01 -4.80215818e-01
-1.35210145e+00 5.03191650e-01 -1.21413283e-01 -1.92617536e+00
-2.33973414e-02 -1.59271359e-01 -8.08938682e-01 4.42705452e-01
-1.47479379e+00 -1.67495406e+00 -7.95785010e-01 7.86724150e-01
5.45395434e-01 -7.81896710e-02 2.68793166e-01 4.65411127e-01
-6.39850438e-01 4.24856365e-01 4.77774024e-01 1.38991162e-01
8.11246872e-01 -8.66800845e-01 3.33279371e-01 9.19148445e-01
5.99314049e-02 5.27514368e-02 6.31075919e-01 -1.08961439e+00
-1.75603080e+00 -1.85061753e+00 5.07840514e-01 -8.94901156e-01
6.53073370e-01 -1.47221163e-01 -9.37994242e-01 8.41901124e-01
-2.06868589e-01 6.34119332e-01 1.66030407e-01 -4.79163915e-01
-5.71308620e-02 -1.02626905e-01 -7.04027593e-01 3.81018430e-01
1.29089415e+00 -7.67021775e-02 -3.90716612e-01 4.70408201e-01
8.97280931e-01 -5.91752410e-01 -8.54679346e-01 4.45497692e-01
6.52367949e-01 -7.72844791e-01 1.24504709e+00 -3.49949002e-01
-3.88510376e-01 -9.31429684e-01 1.23004792e-02 -9.57746446e-01
-5.94877005e-01 -1.00832474e+00 -7.72139847e-01 1.31312561e+00
-3.04756999e-01 -2.35466212e-01 7.29633391e-01 7.06463829e-02
-1.66913733e-01 -5.74557841e-01 -1.09169531e+00 -1.32866645e+00
-3.01731169e-01 -1.94762111e-01 5.11698604e-01 5.53967476e-01
-5.98144233e-01 3.14965881e-02 -8.97851288e-01 8.12091231e-01
1.07243955e+00 4.39939052e-01 1.35587466e+00 -1.36324918e+00
-2.97784835e-01 -1.16483524e-01 -7.43309200e-01 -1.64439750e+00
8.77101719e-02 -5.51006794e-01 5.18824160e-01 -1.29007483e+00
5.31767905e-01 -5.92146277e-01 -1.12963550e-01 2.19500780e-01
-5.88743865e-01 2.42678508e-01 3.90212387e-01 7.91200519e-01
-1.56027865e+00 7.68611789e-01 1.26683080e+00 -3.39851528e-01
-3.51934344e-01 3.82350743e-01 -7.49760717e-02 7.87410498e-01
2.33203337e-01 -9.44340467e-01 7.96351358e-02 -5.99961460e-01
-3.42515558e-01 1.60969377e-01 7.34068692e-01 -1.55638278e+00
6.81295991e-01 -2.79627174e-01 4.90339100e-01 -1.10740268e+00
5.63163161e-01 -9.27119911e-01 5.50895631e-01 6.61360979e-01
2.27083191e-01 1.89770743e-01 2.90306211e-01 1.36528933e+00
7.88345784e-02 1.64461091e-01 7.33584225e-01 1.32543012e-01
-1.07396019e+00 8.89768004e-01 -6.42142072e-02 2.29115747e-02
1.40805602e+00 -3.33103865e-01 -3.33620787e-01 -1.12434603e-01
-6.15033448e-01 7.40565717e-01 6.98119462e-01 7.34678388e-01
6.58862114e-01 -1.72711766e+00 -8.79068494e-01 -2.02354118e-01
2.81176478e-01 1.06542550e-01 4.14212853e-01 1.25882292e+00
-8.82772133e-02 8.86332393e-02 -3.79429646e-02 -1.27596545e+00
-1.50791156e+00 6.34095609e-01 2.68421590e-01 -1.40066922e-01
-9.52706218e-01 7.02889860e-01 3.39646250e-01 9.27259699e-02
3.88271332e-01 -1.15504913e-01 -3.13956700e-02 -1.61403626e-01
1.04461920e+00 5.95957458e-01 -6.14737213e-01 -1.21642733e+00
-5.97694695e-01 1.04324257e+00 7.31501915e-03 2.88293749e-01
1.08731604e+00 -6.21592045e-01 1.53618485e-01 4.25856709e-01
8.58987927e-01 -1.70293357e-02 -1.94388819e+00 -4.79260206e-01
-1.16547078e-01 -9.22790408e-01 -3.62792432e-01 -6.39339164e-02
-1.09212542e+00 4.58412051e-01 8.08298230e-01 -6.99116141e-02
7.75919139e-01 1.86543196e-01 1.13210559e+00 4.03555781e-01
6.96633995e-01 -6.29858613e-01 4.35856432e-01 4.62222010e-01
4.97138828e-01 -1.46418047e+00 -1.53488606e-01 -1.87148854e-01
-5.37217736e-01 8.55355680e-01 8.46031964e-01 3.79249267e-02
3.87501508e-01 8.18326026e-02 -1.98038910e-02 -2.56434113e-01
-6.12234116e-01 -4.92742658e-01 3.11459541e-01 8.47311199e-01
-1.56804577e-01 -6.74344525e-02 3.31432283e-01 1.97405159e-01
5.52264690e-01 -1.06578432e-01 1.14887394e-01 1.08436584e+00
-8.18847537e-01 -7.99021482e-01 -8.18665445e-01 1.43450409e-01
-1.31833494e-01 4.84058350e-01 -1.06950819e-01 6.90260708e-01
3.91398638e-01 1.10240197e+00 -1.76686645e-02 -6.14305615e-01
3.19738507e-01 -4.42830980e-01 3.82897526e-01 -3.89474571e-01
-5.03682971e-01 2.46524766e-01 -2.40113258e-01 -8.65450859e-01
-8.32419038e-01 -9.62284088e-01 -1.17110658e+00 -5.04029334e-01
-6.05060399e-01 -1.15229562e-01 8.73970985e-03 9.84873176e-01
4.83414203e-01 3.99410635e-01 6.49249673e-01 -1.36250103e+00
-4.77932572e-01 -8.32518756e-01 -4.36207145e-01 5.65532625e-01
7.46965468e-01 -1.05731285e+00 -4.15118970e-02 1.61556542e-01] | [6.4028754234313965, -2.096855401992798] |
0ed9abb8-27b5-4c14-9d86-58860a1d9d89 | adaptive-r-peak-detection-on-wearable-ecg | 2112.04369 | null | https://arxiv.org/abs/2112.04369v1 | https://arxiv.org/pdf/2112.04369v1.pdf | Adaptive R-Peak Detection on Wearable ECG Sensors for High-Intensity Exercise | Objective: Continuous monitoring of biosignals via wearable sensors has quickly expanded in the medical and wellness fields. At rest, automatic detection of vital parameters is generally accurate. However, in conditions such as high-intensity exercise, sudden physiological changes occur to the signals, compromising the robustness of standard algorithms. Methods: Our method, called BayeSlope, is based on unsupervised learning, Bayesian filtering, and non-linear normalization to enhance and correctly detect the R peaks according to their expected positions in the ECG. Furthermore, as BayeSlope is computationally heavy and can drain the device battery quickly, we propose an online design that adapts its robustness to sudden physiological changes, and its complexity to the heterogeneous resources of modern embedded platforms. This method combines BayeSlope with a lightweight algorithm, executed in cores with different capabilities, to reduce the energy consumption while preserving the accuracy. Results: BayeSlope achieves an F1 score of 99.3% in experiments during intense cycling exercise with 20 subjects. Additionally, the online adaptive process achieves an F1 score of 99% across five different exercise intensities, with a total energy consumption of 1.55+-0.54~mJ. Conclusion: We propose a highly accurate and robust method, and a complete energy-efficient implementation in a modern ultra-low-power embedded platform to improve R peak detection in challenging conditions, such as during high-intensity exercise. Significance: The experiments show that BayeSlope outperforms a state-of-the-art algorithm up to 8.4% in F1 score, while our online adaptive method can reach energy savings up to 38.7% on modern heterogeneous wearable platforms. | ['David Atienza', 'Grégoire P. Millet', 'Tomas Teijeiro', 'Elisabetta De Giovanni'] | 2021-12-08 | null | null | null | null | ['total-energy'] | ['miscellaneous'] | [ 5.59945762e-01 -3.38904500e-01 -3.18900168e-01 -1.65035307e-01
-5.90608895e-01 -3.57640713e-01 -5.39534688e-01 3.92762512e-01
-4.81218040e-01 5.72189033e-01 -1.31936789e-01 -6.24279901e-02
-2.19609104e-02 -5.96656144e-01 -3.84656757e-01 -6.30521774e-01
-3.89356792e-01 -2.10196376e-01 2.78892219e-01 1.20598249e-01
-1.48642927e-01 2.70485461e-01 -1.48059976e+00 1.51532650e-01
8.43729079e-01 1.16728497e+00 2.09182799e-02 8.45834792e-01
6.80776000e-01 2.28536338e-01 -7.16415584e-01 4.26378176e-02
-2.86016259e-02 -6.05687320e-01 -2.95757174e-01 -7.47248530e-01
-1.78801045e-01 -2.57005423e-01 9.50034261e-02 6.53285325e-01
1.25023425e+00 -1.78246126e-01 1.68673843e-01 -9.81595814e-01
3.27620566e-01 5.84128201e-01 -6.16122782e-01 4.36074942e-01
3.83738220e-01 2.26180986e-01 3.33974481e-01 -4.23612446e-01
-3.97772230e-02 3.47630233e-01 1.35060191e+00 4.62327302e-01
-1.43145716e+00 -8.57450366e-01 -4.59157944e-01 2.71263272e-01
-1.79658508e+00 -6.93915009e-01 8.98448884e-01 -8.41902867e-02
1.04203856e+00 7.30189264e-01 1.11435032e+00 8.19687128e-01
8.35973859e-01 1.74017355e-01 1.28327286e+00 -2.73769796e-01
5.52143991e-01 -3.12162101e-01 2.17884947e-02 4.31055069e-01
6.10235810e-01 -7.33089745e-02 -9.09468174e-01 -3.95546585e-01
2.62564600e-01 1.29690915e-01 -1.33118570e-01 3.33956003e-01
-1.00479460e+00 7.97343776e-02 7.44524747e-02 2.80468553e-01
-5.00157893e-01 5.10590732e-01 5.50145209e-01 -3.06079894e-01
2.33999714e-01 3.57101291e-01 -5.65810978e-01 -7.50331044e-01
-1.16086423e+00 -7.61898831e-02 7.07620978e-01 5.02043426e-01
1.88328668e-01 1.46926418e-01 -3.96934628e-01 5.95240474e-01
3.17309827e-01 9.28430259e-01 6.68486059e-01 -7.94906735e-01
-4.36002435e-03 5.85411847e-01 8.18054527e-02 -1.03359115e+00
-1.05005491e+00 -5.88120759e-01 -1.06777060e+00 -3.22086096e-01
1.09341942e-01 -4.51752812e-01 -3.72443825e-01 1.67991090e+00
5.40275872e-01 1.37058735e-01 -2.63806283e-01 8.23719382e-01
7.64634609e-01 3.68669450e-01 3.78329962e-01 -6.37230933e-01
1.70839417e+00 -2.45396569e-01 -1.10537028e+00 -2.57168293e-01
2.63008505e-01 -4.15471822e-01 9.93944585e-01 4.62145329e-01
-1.18822706e+00 -7.32086658e-01 -1.39380741e+00 2.41556764e-01
2.28054106e-01 3.31773520e-01 3.12113136e-01 1.50846112e+00
-7.57827520e-01 8.88345599e-01 -1.37620735e+00 -2.41206259e-01
3.69587272e-01 5.68572104e-01 3.37051749e-01 2.64945567e-01
-1.15412068e+00 5.50809503e-01 1.71090290e-01 2.18565330e-01
-4.15827572e-01 -9.52219367e-01 -5.15359521e-01 1.22363493e-01
2.85793185e-01 -6.74429834e-01 8.15528870e-01 -6.90648139e-01
-1.81042516e+00 5.18968880e-01 -1.17234260e-01 -3.18868458e-01
2.82075197e-01 -4.89588469e-01 -6.23869002e-01 4.02274609e-01
-2.61533707e-01 -1.37693202e-02 6.70507967e-01 -3.92249703e-01
-2.59042648e-03 -5.43763936e-01 -5.71132243e-01 4.35527340e-02
-5.50639510e-01 -7.32570663e-02 -1.37274459e-01 -4.73404199e-01
1.30467772e-01 -1.04230738e+00 -1.64743751e-01 -5.02914228e-02
-6.13613874e-02 2.11596280e-01 3.30931693e-01 -6.64311945e-01
1.69302011e+00 -2.22266960e+00 -4.13535833e-01 3.30310017e-01
1.86317369e-01 1.62103891e-01 6.07877731e-01 6.61484003e-02
1.89972728e-01 6.00988194e-02 -1.96245387e-01 1.54175863e-01
-7.42766485e-02 1.48946941e-02 3.36964875e-01 8.90051603e-01
-2.64815539e-01 9.71986532e-01 -8.04633617e-01 -4.72936213e-01
1.68628797e-01 5.19212484e-01 -2.75025398e-01 1.90342963e-01
5.25393486e-01 1.62776724e-01 -1.61880583e-01 7.48020053e-01
4.15769100e-01 -7.15808645e-02 8.15133929e-01 -8.13740492e-01
-1.63824856e-02 8.65465477e-02 -1.38098681e+00 1.78292787e+00
-1.95898473e-01 1.61523283e-01 1.19065113e-01 -8.15682709e-01
9.83116329e-01 4.19287056e-01 9.36101794e-01 -9.08372641e-01
4.05934960e-01 2.82705337e-01 4.55753654e-02 -7.47557461e-01
2.58577317e-01 -2.38852471e-01 -2.92334616e-01 3.88674527e-01
-2.25568026e-01 2.37372965e-01 -3.13497446e-02 -1.04245529e-01
1.37106514e+00 3.77680063e-01 4.73916441e-01 -7.70814478e-01
1.75022900e-01 -4.05175388e-01 8.42493951e-01 7.40443528e-01
-5.42376161e-01 2.24744990e-01 -2.62481533e-02 -3.34267646e-01
-3.83511275e-01 -1.08643639e+00 -2.51769096e-01 8.75360906e-01
1.82542846e-01 -7.20627785e-01 -7.85830200e-01 -1.46679766e-03
-1.22320160e-01 2.45685697e-01 -1.98875844e-01 -6.80860639e-01
-6.07460916e-01 -1.27438676e+00 8.95299256e-01 7.62944877e-01
6.99013591e-01 -6.66928649e-01 -1.71149087e+00 6.31160438e-01
-4.20851529e-01 -1.01627183e+00 -4.55781490e-01 3.20622087e-01
-1.18917155e+00 -8.78360212e-01 -2.18311578e-01 -1.72458038e-01
3.69881660e-01 -2.64460891e-01 1.17595899e+00 -1.79936849e-02
-8.85122657e-01 5.42819440e-01 -2.19179839e-01 -6.67325020e-01
1.42938584e-01 -1.21862097e-02 3.99692833e-01 3.66594531e-02
1.67719156e-01 -7.54266441e-01 -1.46070087e+00 2.94807643e-01
-4.62171137e-01 -2.56297380e-01 5.38003087e-01 4.82474625e-01
8.63391042e-01 3.36098298e-02 8.36894989e-01 -5.40085733e-01
3.64436477e-01 -3.07493657e-01 -1.48493424e-01 -1.33079022e-01
-1.07666910e+00 -2.52768159e-01 7.05395758e-01 -6.46818876e-01
-6.71765029e-01 4.17490542e-01 -2.66724199e-01 1.71732083e-01
2.86192477e-01 3.85116190e-02 -2.77329504e-01 8.61351471e-03
9.76405442e-01 1.79552615e-01 1.97315011e-02 -1.79973483e-01
-1.17178440e-01 5.76678216e-01 7.60538042e-01 -5.45780838e-01
3.41392457e-01 3.59310806e-01 3.55468065e-01 -7.14918971e-01
-4.03479874e-01 -3.80416274e-01 -1.94760919e-01 -5.67822635e-01
7.80422330e-01 -1.12458766e+00 -1.11004257e+00 4.09914494e-01
-4.06734705e-01 -5.02328038e-01 -3.81885886e-01 6.41938508e-01
-2.23335475e-01 3.64213049e-01 -5.25471151e-01 -1.02578712e+00
-1.41974413e+00 -7.25296021e-01 7.82615483e-01 4.10375863e-01
-7.32526720e-01 -5.87660134e-01 4.44505103e-02 3.55034232e-01
5.99855602e-01 7.32669592e-01 2.86543012e-01 -7.53192753e-02
1.76943779e-01 -1.80492818e-01 4.14318323e-01 9.92251337e-02
2.78867066e-01 -3.36087227e-01 -1.18382466e+00 -5.21673918e-01
1.84377611e-01 3.96218151e-03 2.30996534e-01 5.49565077e-01
1.37772465e+00 1.53413638e-02 -6.61020756e-01 5.71392953e-01
1.46204150e+00 1.41470999e-01 8.39847147e-01 -1.53992340e-01
3.69783223e-01 9.02440678e-03 5.58419883e-01 7.23847330e-01
1.97123170e-01 6.08519197e-01 2.19681263e-01 -3.56206536e-01
3.13091837e-02 2.17487410e-01 5.65269291e-01 9.27731335e-01
-3.24526310e-01 1.17657587e-01 -6.81313753e-01 4.37807322e-01
-1.42361760e+00 -7.37766385e-01 -3.36916089e-01 2.56803942e+00
1.21684718e+00 2.23285660e-01 3.47268462e-01 6.02194190e-01
6.10184669e-01 -8.95697251e-02 -8.50127876e-01 -4.73081946e-01
1.95775971e-01 7.78354764e-01 6.47608876e-01 -2.56466687e-01
-8.25265050e-01 -2.27918491e-01 6.36867857e+00 4.70742166e-01
-1.15234554e+00 5.27882040e-01 6.08009160e-01 -5.81764460e-01
2.09343940e-01 -4.21129942e-01 -5.35248995e-01 8.72949004e-01
1.67225480e+00 -1.81246147e-01 1.01773933e-01 6.84183776e-01
4.51812059e-01 -3.66837204e-01 -8.99242699e-01 1.30891013e+00
7.01606423e-02 -8.74187469e-01 -9.95104015e-01 -3.26674283e-01
2.52324045e-01 -2.27403089e-01 -4.19526339e-01 9.68998894e-02
-6.33091390e-01 -7.91900992e-01 4.79394495e-01 5.50759375e-01
1.23416233e+00 -8.57355177e-01 7.00863779e-01 1.96476862e-01
-1.57665062e+00 -1.22424588e-02 5.33480830e-02 -1.90757915e-01
2.90279444e-02 1.33793759e+00 -3.70001405e-01 4.89087164e-01
1.03779709e+00 3.52752030e-01 -4.20226306e-01 8.64224076e-01
-7.50383511e-02 9.79130447e-01 -7.43358374e-01 -2.43015096e-01
-7.55527556e-01 3.27496976e-01 2.66834736e-01 1.27239811e+00
4.14315015e-01 5.68305850e-01 1.91099659e-01 4.42951888e-01
-3.89031693e-02 2.30599523e-01 5.27556241e-02 4.12209183e-01
6.47321165e-01 1.58232284e+00 -8.51937592e-01 -4.07835871e-01
7.05768168e-02 6.90891922e-01 -4.44316775e-01 -2.58743197e-01
-1.27204025e+00 -7.74374008e-01 3.39230776e-01 5.32255709e-01
-2.65873402e-01 -2.16386821e-02 -6.28550529e-01 -6.68781638e-01
4.30010349e-01 -6.53460145e-01 4.33919460e-01 -5.78854382e-01
-7.45992243e-01 3.52796763e-01 -1.47174537e-01 -1.16194522e+00
1.93879291e-01 -1.63167432e-01 -5.54401636e-01 5.76178551e-01
-1.21762085e+00 -4.02688891e-01 -8.35740745e-01 6.23629272e-01
1.79817572e-01 6.56626046e-01 9.78596985e-01 6.51148260e-01
-7.66950846e-01 8.57060313e-01 -4.68656048e-02 -3.66819441e-01
7.46131122e-01 -9.37685966e-01 -1.12486690e-01 8.85461211e-01
-5.86717486e-01 6.14431560e-01 5.07784069e-01 -7.81900048e-01
-2.06648040e+00 -1.13567162e+00 4.51885372e-01 9.26861260e-03
1.95359334e-01 -4.52314168e-01 -6.86076105e-01 -1.76637784e-01
-8.31978917e-02 3.67955565e-01 1.09693170e+00 -1.78351834e-01
2.13912413e-01 -9.45756972e-01 -1.42237020e+00 4.57228094e-01
1.03881252e+00 -2.54125834e-01 -1.27019182e-01 3.12551856e-01
5.66343702e-02 -5.80477536e-01 -1.64247334e+00 7.09008455e-01
1.17428625e+00 -5.76105773e-01 9.27560151e-01 1.58624142e-01
-2.04708025e-01 -4.25721854e-01 2.38036737e-01 -8.34421039e-01
-3.00626278e-01 -1.22513986e+00 -5.35964251e-01 1.21039736e+00
5.73494956e-02 -6.74358010e-01 5.92981040e-01 7.45051503e-01
-5.03604300e-02 -8.36645544e-01 -1.13580799e+00 -7.07976758e-01
-7.27182865e-01 -4.38158542e-01 2.58099109e-01 5.76305926e-01
4.36325401e-01 1.67806432e-01 -3.93641740e-01 2.53355671e-02
7.55064309e-01 6.30954355e-02 2.91167974e-01 -1.14979279e+00
-4.57550585e-01 1.39309034e-01 -3.82960469e-01 -5.08579314e-01
-6.19712949e-01 -5.77511191e-01 2.18851820e-01 -1.10683239e+00
2.76453555e-01 -3.22182149e-01 -8.22852910e-01 7.28010058e-01
-6.32307768e-01 9.07168031e-01 -2.84704357e-01 -8.83141086e-02
-6.00515246e-01 1.19672567e-01 5.57667315e-01 -7.12462738e-02
-6.42481863e-01 -6.77232295e-02 -5.98417640e-01 6.02669775e-01
9.43749070e-01 -8.78103554e-01 -3.62631679e-01 1.96609125e-01
4.37423557e-01 2.61844378e-02 1.67295232e-01 -1.53373516e+00
5.76888062e-02 9.41754282e-02 7.62525558e-01 -3.18637997e-01
1.64205253e-01 -8.89705420e-01 7.82803178e-01 1.18048131e+00
2.34680817e-01 1.35329023e-01 4.81244147e-01 2.79892415e-01
5.87459922e-01 1.90165639e-01 7.36755133e-01 1.02629445e-01
-2.25564927e-01 -1.44840419e-01 -5.27040482e-01 6.94539845e-02
1.11744452e+00 -4.50749397e-01 -1.65029600e-01 2.22046316e-01
-7.28059590e-01 8.48128460e-03 5.96432909e-02 4.57451725e-03
4.89347667e-01 -1.25189424e+00 -4.45224285e-01 1.49911046e-01
-1.69569273e-02 -4.21421915e-01 7.18654871e-01 1.50886476e+00
-5.10351896e-01 -1.63946405e-01 -3.38606358e-01 -8.55782747e-01
-1.47143388e+00 1.53115556e-01 4.67871964e-01 -1.49921507e-01
-7.52461374e-01 2.59802550e-01 -6.85499549e-01 5.37655175e-01
-1.37214184e-01 -3.76437634e-01 -1.20726144e-02 2.36884430e-02
6.04812205e-01 9.53973472e-01 4.48109448e-01 9.25352201e-02
-9.30012286e-01 6.91501677e-01 6.56222045e-01 4.09405679e-01
1.05030107e+00 -1.66989699e-01 6.29899502e-02 5.24046779e-01
7.13604927e-01 2.32201740e-01 -9.79604363e-01 3.12458813e-01
-3.39435965e-01 -9.09920856e-02 1.86143726e-01 -8.30327213e-01
-1.14784443e+00 3.90174925e-01 1.51758444e+00 -1.68398079e-02
1.96230209e+00 -5.19443274e-01 1.06484008e+00 2.52830293e-02
4.10266131e-01 -1.29713750e+00 8.98246616e-02 -3.67589682e-01
1.90404296e-01 -6.23322368e-01 7.74414957e-01 -2.19902560e-01
-1.69430196e-01 9.67361391e-01 3.61323506e-01 1.17792366e-02
6.59136236e-01 7.56793618e-01 -1.88300967e-01 -4.04391773e-02
-3.09476703e-01 1.74800962e-01 3.48110870e-02 4.70906794e-01
4.45743054e-01 2.70591795e-01 -8.92700732e-01 1.00496233e+00
1.34046867e-01 4.15874809e-01 2.79036164e-01 1.22510815e+00
-2.34923720e-01 -7.02094972e-01 -4.75854456e-01 6.94152951e-01
-1.01497853e+00 6.90592304e-02 2.03620732e-01 2.66862541e-01
3.58860224e-01 1.27076006e+00 -1.34901196e-01 -4.20831621e-01
6.62805796e-01 1.94765657e-01 4.42645699e-01 -2.57499844e-01
-1.12044430e+00 2.22096011e-01 6.00853190e-02 -9.91672277e-01
-6.32853031e-01 -4.82818425e-01 -1.44006312e+00 -8.52585435e-02
-3.08995456e-01 -1.01668008e-01 9.39791381e-01 6.30759895e-01
9.06386554e-01 1.10297835e+00 6.68718338e-01 -5.96358418e-01
-2.67094791e-01 -9.20507014e-01 -6.67682707e-01 2.23722950e-01
-7.98716694e-02 -5.25302649e-01 -8.59244838e-02 2.16431767e-01] | [13.945591926574707, 3.0895798206329346] |
c43df57e-9d4f-40f7-b8e2-397811fcc278 | towards-document-level-paraphrase-generation | 2109.07095 | null | https://arxiv.org/abs/2109.07095v1 | https://arxiv.org/pdf/2109.07095v1.pdf | Towards Document-Level Paraphrase Generation with Sentence Rewriting and Reordering | Paraphrase generation is an important task in natural language processing. Previous works focus on sentence-level paraphrase generation, while ignoring document-level paraphrase generation, which is a more challenging and valuable task. In this paper, we explore the task of document-level paraphrase generation for the first time and focus on the inter-sentence diversity by considering sentence rewriting and reordering. We propose CoRPG (Coherence Relationship guided Paraphrase Generation), which leverages graph GRU to encode the coherence relationship graph and get the coherence-aware representation for each sentence, which can be used for re-arranging the multiple (possibly modified) input sentences. We create a pseudo document-level paraphrase dataset for training CoRPG. Automatic evaluation results show CoRPG outperforms several strong baseline models on the BERTScore and diversity scores. Human evaluation also shows our model can generate document paraphrase with more diversity and semantic preservation. | ['Xiaojun Wan', 'Yitao Cai', 'Zhe Lin'] | 2021-09-15 | null | https://aclanthology.org/2021.findings-emnlp.89 | https://aclanthology.org/2021.findings-emnlp.89.pdf | findings-emnlp-2021-11 | ['paraphrase-generation', 'paraphrase-generation'] | ['computer-code', 'natural-language-processing'] | [ 4.28670645e-01 1.09248213e-01 -2.73280293e-01 -3.22427124e-01
-8.03618252e-01 -7.89080262e-01 6.89148664e-01 4.56466824e-01
1.29147619e-01 7.36749530e-01 1.04238224e+00 -4.58361149e-01
6.67066053e-02 -8.19934309e-01 -6.86083436e-01 -1.05602697e-01
5.39685369e-01 3.13893527e-01 3.11806370e-02 -6.60517573e-01
8.16649139e-01 1.26579747e-01 -1.20704436e+00 8.44171107e-01
1.31219947e+00 1.52693197e-01 4.94584560e-01 7.03257203e-01
-3.82724911e-01 8.43660593e-01 -8.74589205e-01 -6.27933919e-01
-6.33561611e-03 -1.12299705e+00 -1.18839204e+00 -1.80805624e-02
5.43273926e-01 9.21398178e-02 -1.40352994e-01 1.04041100e+00
5.51685929e-01 1.93514913e-01 5.82304239e-01 -9.24589574e-01
-1.12832749e+00 1.06327403e+00 -4.60972250e-01 5.23072302e-01
9.96973813e-01 4.77012433e-02 1.40599334e+00 -1.06468225e+00
8.84736300e-01 1.32815552e+00 3.44668478e-01 5.55794895e-01
-1.24245512e+00 -3.08437020e-01 -4.02266607e-02 4.59682196e-01
-1.01051319e+00 -4.46451515e-01 9.21612561e-01 9.29874927e-02
1.48185813e+00 5.84373415e-01 6.66144788e-01 1.15666473e+00
4.86093998e-01 8.56465280e-01 8.40938866e-01 -5.93978703e-01
9.39076990e-02 -2.83950120e-01 4.59745347e-01 4.85682935e-01
4.14522946e-01 -4.26965654e-01 -5.80554605e-01 6.46023303e-02
3.12735617e-01 -3.78995277e-02 -3.69866431e-01 -1.20814696e-01
-1.18229055e+00 9.08463538e-01 5.66048145e-01 3.46010149e-01
-2.14721441e-01 -1.80901736e-01 6.92601860e-01 8.49819779e-01
3.00185025e-01 1.20221651e+00 9.73533317e-02 -7.50043616e-02
-1.22405589e+00 4.99101371e-01 8.69204521e-01 1.28383970e+00
4.60197568e-01 -2.69712955e-01 -9.14071143e-01 1.16517961e+00
-1.31763354e-01 4.55755740e-01 9.92626786e-01 -8.35580051e-01
1.01552129e+00 8.42167377e-01 -2.70023793e-01 -9.72878814e-01
-8.63612518e-02 -5.40620923e-01 -1.01946497e+00 -4.79809016e-01
-4.03233886e-01 1.17121033e-01 -6.72418535e-01 1.48234487e+00
-1.94378003e-01 -2.35618547e-01 2.05951706e-01 6.50518835e-01
1.05662966e+00 7.65892863e-01 -3.61745775e-01 -3.34438533e-01
1.16371381e+00 -1.63990438e+00 -4.77901548e-01 -5.72990000e-01
7.86146224e-01 -8.89015377e-01 1.49451780e+00 -1.09614953e-02
-1.43890929e+00 -6.11451089e-01 -1.11743295e+00 -3.98853987e-01
-6.92810584e-03 -7.66194165e-02 2.97868341e-01 1.90081686e-01
-1.21323752e+00 6.93631053e-01 -3.65776569e-01 -5.98141730e-01
1.53773695e-01 -2.79456168e-01 -1.72700971e-01 -3.90141338e-01
-1.37651968e+00 1.06680334e+00 5.03329992e-01 -3.57670397e-01
-2.22995982e-01 -7.67631352e-01 -1.02904058e+00 2.61815310e-01
1.40764073e-01 -1.39061999e+00 1.27805936e+00 -5.44872880e-01
-1.30204618e+00 7.92366803e-01 -5.31423688e-01 -6.19111121e-01
1.60562634e-01 7.32922405e-02 -6.79008290e-02 6.72504678e-02
4.20733333e-01 5.12241721e-01 6.47430599e-01 -8.62005889e-01
-2.14223698e-01 -4.79384512e-02 1.65519461e-01 7.65633225e-01
-2.13180140e-01 -3.91453598e-03 -2.50785708e-01 -1.04612792e+00
8.79068822e-02 -7.73438752e-01 -2.25969061e-01 -6.82627201e-01
-7.40683019e-01 -2.69252509e-01 2.51143545e-01 -8.64477992e-01
1.86060929e+00 -1.72964859e+00 6.84041858e-01 -1.55448169e-01
-6.93466514e-02 1.83497325e-01 -5.26516557e-01 1.15952122e+00
-2.23378688e-01 2.91340560e-01 -3.65474045e-01 -5.08668244e-01
-5.87441996e-02 3.64618637e-02 -7.05690622e-01 -4.88518268e-01
8.96266922e-02 1.53434587e+00 -1.22592938e+00 -5.40462911e-01
-1.84736326e-01 -3.74570847e-01 -6.59991026e-01 2.25973323e-01
-1.49280563e-01 3.92160239e-03 -2.21939400e-01 3.80014062e-01
4.54782307e-01 -2.88505286e-01 5.97326979e-02 -1.35954976e-01
3.96496803e-01 6.85799778e-01 -3.51680487e-01 2.15192556e+00
-8.73035610e-01 4.17375624e-01 -6.67445004e-01 -5.51969588e-01
9.60653901e-01 -2.36936405e-01 -1.90676406e-01 -9.44259346e-01
-3.58080328e-01 2.21712619e-01 -7.78010339e-02 -3.70801270e-01
1.15254629e+00 -1.78568780e-01 -5.05151868e-01 8.94201458e-01
-2.49515787e-01 -7.74812520e-01 7.09842086e-01 9.41381991e-01
1.28040314e+00 -2.11055562e-01 6.18842304e-01 -2.22348601e-01
6.04153514e-01 1.90054804e-01 -8.43033791e-02 9.46308374e-01
3.03896666e-01 9.24285114e-01 5.43611050e-01 3.84095311e-02
-1.24588108e+00 -1.11094689e+00 5.06552994e-01 7.82951295e-01
3.08919787e-01 -8.61957550e-01 -6.26322746e-01 -6.77666724e-01
-4.08201851e-02 1.23529315e+00 -4.03632671e-01 -6.74551129e-01
-8.39170218e-01 -4.15970564e-01 5.22819638e-01 4.83289123e-01
3.76322240e-01 -1.26367664e+00 1.22692771e-02 2.38472551e-01
-6.82978928e-01 -8.85117114e-01 -1.01739049e+00 -4.08917099e-01
-9.40461934e-01 -7.39300132e-01 -7.48352587e-01 -1.08059287e+00
7.64924347e-01 7.69328356e-01 1.55586123e+00 2.06189260e-01
-1.38679221e-02 -1.13396347e-01 -8.00688028e-01 -9.19580087e-02
-9.60070312e-01 5.43098092e-01 -2.81775683e-01 -6.46635950e-01
9.12479758e-02 -8.77483606e-01 -6.49531245e-01 -1.36352107e-01
-9.32522178e-01 5.03098726e-01 6.40442550e-01 9.55848277e-01
3.84648114e-01 -4.68422621e-01 8.77202690e-01 -9.40521777e-01
1.74230349e+00 -4.21178699e-01 2.49774694e-01 7.09472716e-01
-6.20696127e-01 2.60782599e-01 1.15559375e+00 3.92675363e-02
-8.25624824e-01 -5.53173542e-01 -1.30721256e-01 -1.92019597e-01
2.63460457e-01 8.04018855e-01 1.17649578e-01 4.35815990e-01
8.63447249e-01 7.97317922e-01 -4.03213017e-02 -3.60149592e-01
7.27998734e-01 7.23682523e-01 5.64732790e-01 -4.42025036e-01
6.97487056e-01 -2.50991583e-02 -3.00107393e-02 -5.51317215e-01
-1.19551492e+00 -3.50327790e-01 -4.84735608e-01 2.44890839e-01
3.29776198e-01 -8.43892097e-01 2.48419195e-01 4.13161479e-02
-1.43109071e+00 3.31369485e-03 -6.40340567e-01 -2.33405605e-01
-4.41059917e-01 1.06659794e+00 -6.12925649e-01 2.38240529e-02
-1.01317501e+00 -8.21854115e-01 1.22071779e+00 3.75656746e-02
-6.53444231e-01 -8.96305203e-01 4.58707690e-01 5.03051281e-01
3.67606372e-01 7.04167113e-02 1.14341533e+00 -5.85164189e-01
-3.16977382e-01 -9.85541791e-02 -2.42992908e-01 3.61014366e-01
3.66786271e-01 -3.07239115e-01 -1.17978603e-01 -4.07085031e-01
-1.55544067e-02 -4.33567077e-01 1.10971856e+00 -4.79004160e-02
9.99632537e-01 -6.91277206e-01 -8.92906189e-02 5.58286011e-01
1.09317219e+00 -3.21265161e-01 7.31098831e-01 1.76230505e-01
6.95501029e-01 4.28878158e-01 6.65400088e-01 3.24184388e-01
4.88310903e-01 6.91588223e-01 -8.37714523e-02 2.06315368e-01
-7.14699566e-01 -7.95211077e-01 5.02400279e-01 1.44862020e+00
4.63697255e-01 -5.71766734e-01 -5.27443707e-01 6.05129778e-01
-1.98422611e+00 -1.28970110e+00 -1.68924540e-01 1.69069064e+00
1.10676789e+00 1.56831704e-02 4.24686528e-04 7.30986670e-02
6.50318086e-01 4.94500130e-01 -4.01000410e-01 -1.08751130e+00
-2.98201442e-01 4.84252095e-01 -8.06331336e-02 6.99072719e-01
-3.94198626e-01 1.25104547e+00 5.95574045e+00 9.71336305e-01
-6.72826827e-01 -2.01413497e-01 2.65794605e-01 -2.96075940e-01
-9.04184222e-01 1.87739551e-01 -5.35160780e-01 7.04089284e-01
5.69958329e-01 -1.06505072e+00 5.99616826e-01 5.99172235e-01
3.02418560e-01 1.25781953e-01 -1.16216540e+00 8.88480484e-01
6.53544724e-01 -1.53108287e+00 1.01102495e+00 -4.95797008e-01
1.01525068e+00 -3.11282009e-01 -2.56996065e-01 5.24549007e-01
2.24523693e-01 -9.17754948e-01 4.01620746e-01 5.32378018e-01
6.35339916e-01 -7.81221747e-01 6.36252105e-01 4.65243995e-01
-1.05617249e+00 8.16676691e-02 -6.52490020e-01 -1.68075785e-01
5.09876907e-01 6.14650965e-01 -9.29381073e-01 9.33316350e-01
1.38755273e-02 1.11210215e+00 -1.26800060e+00 8.18046391e-01
-7.04112411e-01 3.79317284e-01 2.18616664e-01 -4.80804056e-01
1.48444131e-01 -2.39394829e-01 7.47293115e-01 1.48326933e+00
5.81629336e-01 -1.28152013e-01 -1.21574625e-02 8.33427727e-01
-2.93104708e-01 2.36818910e-01 -8.18161070e-01 -4.82898485e-03
7.31718242e-01 1.10006058e+00 -2.96711177e-01 -4.65151936e-01
-3.61471847e-02 1.77258337e+00 5.96317530e-01 1.70952484e-01
-6.67124569e-01 -8.68858159e-01 3.20277870e-01 -1.30734116e-01
1.34329647e-01 -1.34507179e-01 -5.36129534e-01 -1.63398683e+00
3.41300935e-01 -9.90741849e-01 3.54201138e-01 -1.07939160e+00
-1.41738749e+00 6.49681747e-01 1.38807893e-01 -1.17210495e+00
-7.32288301e-01 1.68171711e-02 -1.12225330e+00 9.46097672e-01
-1.34720194e+00 -1.21244681e+00 -3.10835123e-01 2.51481980e-01
1.24819255e+00 -2.25380346e-01 6.03502393e-01 -1.88320994e-01
-3.98432612e-01 8.55937362e-01 1.36829237e-03 -1.95591867e-01
5.24691403e-01 -1.24826944e+00 1.23136592e+00 1.23849356e+00
2.79057056e-01 1.03558528e+00 7.93944776e-01 -7.59580791e-01
-1.13189077e+00 -1.30810201e+00 1.66587758e+00 -5.03385961e-01
7.65844047e-01 -2.51095355e-01 -8.25899839e-01 2.72026092e-01
5.34063697e-01 -8.18506181e-01 4.70603824e-01 -9.49945971e-02
-4.12670344e-01 1.90282434e-01 -8.31211984e-01 1.10572493e+00
1.73760796e+00 -7.08186209e-01 -1.33743501e+00 6.28745437e-01
1.27951133e+00 -3.38205040e-01 -4.75748658e-01 2.16580689e-01
9.37541127e-02 -1.02135909e+00 9.76389647e-01 -8.16986382e-01
1.21598983e+00 -1.17505640e-01 1.18534110e-01 -1.86320603e+00
-7.06780732e-01 -6.16440296e-01 -2.34598294e-01 1.28368258e+00
5.32944858e-01 -3.17824364e-01 7.02361166e-01 1.49835661e-01
-4.99434352e-01 -8.38406861e-01 -6.39703572e-01 -1.11963499e+00
1.67552814e-01 3.15758139e-01 8.22011113e-01 8.38655472e-01
6.35278106e-01 1.09421086e+00 -3.24724406e-01 -4.94550228e-01
3.68175536e-01 8.08602393e-01 7.85099387e-01 -4.42161202e-01
-6.59961700e-01 -6.53975666e-01 -6.84700981e-02 -1.32279909e+00
4.03453022e-01 -1.73311114e+00 -1.22966327e-01 -2.20046043e+00
6.09826922e-01 3.73366505e-01 2.36608937e-01 2.03038618e-01
-5.47913492e-01 7.70261511e-02 3.98613334e-01 3.68043721e-01
-6.55105829e-01 8.00616205e-01 1.52399385e+00 -2.16302335e-01
-2.51539469e-01 -1.12158075e-01 -1.26682520e+00 2.01700494e-01
9.68365073e-01 -2.74032623e-01 -8.22082222e-01 -7.67118454e-01
3.87644976e-01 1.53051183e-01 1.24823026e-01 -8.15047324e-01
2.82913744e-01 -3.35801423e-01 -1.96050435e-01 -4.41462606e-01
-6.62830845e-02 -3.27385888e-02 9.02753770e-02 5.25071681e-01
-9.03791308e-01 7.52864242e-01 -1.09310038e-01 4.46354121e-01
-3.79614711e-01 -6.46505415e-01 5.29509604e-01 -2.91445583e-01
-4.59009379e-01 -9.84476283e-02 -3.90838496e-02 5.44964910e-01
7.27655351e-01 -4.36469764e-01 -6.38340950e-01 -5.35849750e-01
-4.11575556e-01 3.32835108e-01 7.72558391e-01 7.44373083e-01
9.93739724e-01 -1.44668961e+00 -1.05059731e+00 1.27281733e-02
4.61082935e-01 -1.94083810e-01 1.42591268e-01 2.29518831e-01
-5.85121632e-01 4.45753008e-01 -1.66271791e-01 -1.37538493e-01
-1.42816174e+00 5.09840190e-01 -6.38000891e-02 -8.29214513e-01
-5.64548731e-01 8.56994152e-01 -1.46653056e-01 -3.50396931e-01
-3.40797693e-01 -3.90612453e-01 -1.73359111e-01 -7.44767115e-02
5.00870764e-01 3.28522533e-01 1.70163304e-01 -2.57859439e-01
-1.99535146e-01 4.47939605e-01 -5.73056459e-01 -5.57743795e-02
1.06962490e+00 -3.07343155e-01 -4.55231994e-01 -6.72286469e-03
1.28494155e+00 4.10282277e-02 -2.71188468e-01 -1.19185485e-01
1.67085141e-01 -5.25985897e-01 -5.72218597e-01 -8.04876506e-01
-5.03385723e-01 7.18206942e-01 -4.25857574e-01 1.98662639e-01
1.05490398e+00 -2.39844024e-02 1.27490735e+00 7.49932826e-01
4.59051639e-01 -8.23189914e-01 5.81556022e-01 9.38273072e-01
1.57629669e+00 -7.13468373e-01 2.20468074e-01 -4.75301385e-01
-9.95711744e-01 8.55707765e-01 6.72442675e-01 -3.61154497e-01
-1.18976273e-02 -3.73580426e-01 -2.70486444e-01 -2.50901356e-02
-1.01280570e+00 5.19184135e-02 4.03774351e-01 3.39473456e-01
5.48043609e-01 -2.16928408e-01 -9.74051416e-01 4.85044479e-01
-8.68586183e-01 -1.04804337e-01 7.99769878e-01 9.33377862e-01
-5.71196258e-01 -1.49871695e+00 2.03043655e-01 7.24688470e-01
3.25527526e-02 -6.79651856e-01 -1.05274594e+00 1.11316182e-01
-5.61423123e-01 1.02519774e+00 -2.13069856e-01 -5.37654042e-01
6.01855218e-01 -4.58251312e-02 6.64619803e-01 -1.22614884e+00
-8.91539276e-01 -6.89072132e-01 4.05679315e-01 -3.53482068e-01
8.43425766e-02 -4.08420503e-01 -1.11289549e+00 -4.11973476e-01
-5.81777729e-02 3.42642814e-01 1.74467534e-01 6.64240241e-01
9.28644061e-01 4.63166833e-01 8.85203421e-01 -6.04838669e-01
-9.28712368e-01 -1.01691365e+00 -3.15368205e-01 9.84329045e-01
1.98640078e-02 1.53673038e-01 -3.67664665e-01 -1.11802764e-01] | [11.739031791687012, 9.2855863571167] |
28623a5d-b173-44f0-ae0f-154960584ec4 | unisurf-unifying-neural-implicit-surfaces-and | 2104.10078 | null | https://arxiv.org/abs/2104.10078v2 | https://arxiv.org/pdf/2104.10078v2.pdf | UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction | Neural implicit 3D representations have emerged as a powerful paradigm for reconstructing surfaces from multi-view images and synthesizing novel views. Unfortunately, existing methods such as DVR or IDR require accurate per-pixel object masks as supervision. At the same time, neural radiance fields have revolutionized novel view synthesis. However, NeRF's estimated volume density does not admit accurate surface reconstruction. Our key insight is that implicit surface models and radiance fields can be formulated in a unified way, enabling both surface and volume rendering using the same model. This unified perspective enables novel, more efficient sampling procedures and the ability to reconstruct accurate surfaces without input masks. We compare our method on the DTU, BlendedMVS, and a synthetic indoor dataset. Our experiments demonstrate that we outperform NeRF in terms of reconstruction quality while performing on par with IDR without requiring masks. | ['Andreas Geiger', 'Songyou Peng', 'Michael Oechsle'] | 2021-04-20 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Oechsle_UNISURF_Unifying_Neural_Implicit_Surfaces_and_Radiance_Fields_for_Multi-View_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Oechsle_UNISURF_Unifying_Neural_Implicit_Surfaces_and_Radiance_Fields_for_Multi-View_ICCV_2021_paper.pdf | iccv-2021-1 | ['3d-object-reconstruction'] | ['computer-vision'] | [ 5.76041639e-01 1.02428310e-01 1.39696568e-01 -2.87164330e-01
-7.13886440e-01 -6.92927778e-01 6.61986947e-01 -2.64228135e-01
2.57692367e-01 7.35490203e-01 8.22114348e-02 -1.73899848e-02
1.44078478e-01 -1.20217168e+00 -9.24139977e-01 -4.94788200e-01
4.72525775e-01 2.84920901e-01 2.03847489e-03 -1.66052476e-01
6.29233047e-02 8.85195136e-01 -1.86268377e+00 3.45595449e-01
9.54835892e-01 1.16429400e+00 2.90761083e-01 6.81201220e-01
-2.57968545e-01 3.56185734e-01 -3.38333517e-01 -2.60443181e-01
5.45060992e-01 -7.99601004e-02 -4.13181901e-01 1.55398399e-01
1.01176059e+00 -6.58387601e-01 -2.99638331e-01 7.10139990e-01
2.75614649e-01 7.72121623e-02 8.32277119e-01 -5.96722603e-01
-6.68559313e-01 -3.33380491e-01 -5.71474135e-01 -4.19736177e-01
4.28980052e-01 -7.19685629e-02 8.73457015e-01 -1.02640927e+00
8.45629275e-01 1.14200056e+00 6.78091586e-01 6.02221906e-01
-1.68978381e+00 -3.34386140e-01 2.68946081e-01 -2.76230365e-01
-1.36918080e+00 -5.51238418e-01 1.02447450e+00 -2.95352340e-01
7.85998285e-01 7.03495800e-01 7.27399111e-01 9.32642698e-01
2.73016989e-01 6.35250986e-01 1.19229960e+00 -1.17798358e-01
2.53392547e-01 1.80840671e-01 -4.12723184e-01 6.40666008e-01
1.95911929e-01 5.00106923e-02 -5.91965556e-01 -2.23494813e-01
1.47148311e+00 1.43297136e-01 -6.48161113e-01 -6.65264130e-01
-1.17871273e+00 6.33092880e-01 3.93472403e-01 2.81831734e-02
-3.19302976e-01 3.78393352e-01 -6.74321949e-02 1.37161210e-01
1.05027211e+00 4.49574739e-01 -5.81777573e-01 2.69007832e-01
-8.12477112e-01 3.62484783e-01 5.26291788e-01 8.71914923e-01
8.97857070e-01 3.96452457e-01 9.26251188e-02 6.97844863e-01
1.85647190e-01 8.63258362e-01 -2.97247201e-01 -1.53875387e+00
1.79442033e-01 5.09362459e-01 4.25791353e-01 -9.99851048e-01
-3.10585722e-02 -5.00115275e-01 -8.79503191e-01 7.82530963e-01
1.46613657e-01 3.73246849e-01 -1.08044541e+00 1.46794665e+00
4.16199088e-01 1.22161128e-01 -1.60703078e-01 7.93274641e-01
6.84858859e-01 8.09337556e-01 -5.76671541e-01 -1.35810584e-01
1.01590347e+00 -5.55556834e-01 -6.91381693e-01 -1.20377308e-02
2.41004765e-01 -5.50481796e-01 1.13754416e+00 7.58937836e-01
-1.32209444e+00 -4.60966706e-01 -9.97846603e-01 -4.73946303e-01
-1.92944314e-02 -1.13013595e-01 9.00193989e-01 5.23266971e-01
-1.20300460e+00 7.49028027e-01 -6.45744860e-01 3.19795728e-01
6.54260337e-01 1.21174812e-01 -3.50358307e-01 -1.94314331e-01
-5.69611609e-01 5.07854521e-01 -4.69231635e-01 -8.81167874e-03
-8.53576124e-01 -1.13000476e+00 -9.69992280e-01 5.84615245e-02
2.55752802e-01 -1.18349636e+00 1.03683400e+00 -9.89713073e-01
-1.75540268e+00 9.01542902e-01 -3.70602041e-01 -1.39791608e-01
6.08147264e-01 -2.25671634e-01 7.77025381e-03 1.98368981e-01
-9.57564116e-02 6.15913451e-01 9.07386601e-01 -1.84211767e+00
-1.87911034e-01 -6.23131812e-01 1.90864101e-01 3.71622533e-01
1.90519437e-01 -8.41773033e-01 -8.58642161e-02 -6.66634917e-01
4.93965983e-01 -3.06690991e-01 -3.04461926e-01 7.83567786e-01
-2.53332436e-01 1.99933782e-01 7.60957479e-01 -8.09913039e-01
6.40024364e-01 -2.01594639e+00 4.52688485e-01 1.71912611e-01
5.78438640e-01 -3.06709975e-01 7.30807856e-02 2.60599740e-02
1.31910935e-01 1.29589990e-01 -5.03633916e-01 -6.38282835e-01
-2.17603743e-01 2.23525718e-01 -5.39722681e-01 5.07697821e-01
-2.12806612e-02 6.84297919e-01 -6.74809992e-01 -1.10837981e-01
5.79916716e-01 1.08832359e+00 -8.38232577e-01 2.34027162e-01
-5.52914739e-01 8.80078316e-01 -2.97213405e-01 5.64097166e-01
8.32417428e-01 -4.04280931e-01 1.31793201e-01 -3.95503014e-01
-2.11988196e-01 3.09846371e-01 -1.11471319e+00 2.05838490e+00
-1.03420782e+00 6.70823932e-01 4.49492961e-01 -5.73263228e-01
1.01795268e+00 3.16540778e-01 5.16127825e-01 -8.65363181e-01
-1.33302629e-01 1.76102027e-01 -7.36581802e-01 8.04384723e-02
5.68084121e-01 -3.35275143e-01 4.43293124e-01 1.68606028e-01
-3.78005952e-01 -7.16047049e-01 -5.40618718e-01 -8.54264945e-02
7.54136086e-01 5.85439146e-01 1.79661945e-01 -2.04657316e-01
3.01186174e-01 -4.14113909e-01 4.95399863e-01 2.62088120e-01
5.49204111e-01 1.11551058e+00 1.39640316e-01 -5.72096467e-01
-1.18432438e+00 -1.64299476e+00 -3.98649901e-01 3.72652143e-01
1.70939296e-01 -1.88834533e-01 -6.20188117e-01 -2.96173573e-01
3.99903394e-02 8.13368082e-01 -7.18059301e-01 4.23412502e-01
-6.16965890e-01 -1.34439543e-01 -1.15894467e-01 3.52992475e-01
2.97230631e-01 -5.73143065e-01 -7.32494175e-01 2.55768816e-03
-1.08033068e-01 -1.07674110e+00 -2.41161913e-01 -1.83119535e-01
-1.22408664e+00 -8.65261137e-01 -9.46471810e-01 -2.53024310e-01
8.41787159e-01 5.92659354e-01 1.45060062e+00 4.51760553e-02
-3.65656167e-01 5.66541493e-01 1.27438396e-01 -1.11644529e-01
-3.89407068e-01 -3.42662722e-01 -1.46672741e-01 -2.10777931e-02
-6.33716643e-01 -1.23423588e+00 -8.25662196e-01 2.77684003e-01
-9.63893175e-01 7.40779936e-01 1.21263057e-01 5.60362577e-01
1.11995780e+00 -4.49259728e-01 2.94834733e-01 -8.87224913e-01
-1.13111690e-01 -1.94529191e-01 -9.75683928e-01 4.52802926e-02
-5.37711322e-01 8.05259570e-02 6.55495107e-01 -8.77761617e-02
-1.34772861e+00 -9.22585800e-02 -2.80789554e-01 -6.94337249e-01
-1.40356526e-01 -8.42404515e-02 -3.34217280e-01 -7.61023238e-02
7.08031952e-01 2.75578141e-01 -1.73943177e-01 -7.80401409e-01
4.80610847e-01 2.25104809e-01 2.14708045e-01 -4.81105596e-01
7.50830293e-01 1.24094212e+00 1.61865279e-01 -1.06603372e+00
-6.48721099e-01 -1.22062508e-02 -5.59986830e-01 -2.39761963e-01
8.98918211e-01 -9.91361260e-01 -6.61372781e-01 2.89830804e-01
-1.33937430e+00 -4.25038606e-01 -6.44748509e-01 6.74115568e-02
-7.70031154e-01 3.41052473e-01 -3.97714496e-01 -9.81817424e-01
-2.56255984e-01 -1.10035694e+00 1.49080706e+00 2.74031330e-02
-4.29069996e-02 -9.70233023e-01 -9.18267444e-02 6.03734791e-01
4.38987046e-01 6.43731356e-01 8.87969792e-01 6.00081325e-01
-1.24849808e+00 3.19162548e-01 -3.39723319e-01 3.56713802e-01
5.17043531e-01 -1.43865854e-01 -1.55816233e+00 -1.23370595e-01
1.98937505e-01 7.73039535e-02 7.45048583e-01 5.89889765e-01
1.46290982e+00 -2.39018813e-01 -1.95513681e-01 1.22775269e+00
1.65372324e+00 -6.92096632e-03 7.02633798e-01 -1.58635467e-01
1.04784918e+00 6.21720552e-01 5.21021970e-02 3.68271142e-01
3.59608233e-01 8.37949038e-01 6.49676561e-01 -1.31870374e-01
-3.47311765e-01 -2.35959575e-01 3.30614001e-02 7.95165479e-01
-2.97234982e-01 -2.00285181e-01 -5.67792237e-01 2.85712659e-01
-1.36176109e+00 -7.71244228e-01 -2.14626610e-01 2.38205695e+00
6.41109824e-01 -1.31343871e-01 -3.76978099e-01 1.14585228e-01
9.13548991e-02 4.87979025e-01 -7.31511056e-01 -3.61937463e-01
-2.94059247e-01 4.89497364e-01 4.42364722e-01 8.76392424e-01
-6.22566164e-01 6.25583351e-01 6.10128736e+00 5.62915564e-01
-1.05244899e+00 1.59482136e-01 9.24766362e-01 -2.14890480e-01
-1.43175709e+00 -2.28365570e-01 -4.36196774e-01 -6.39675558e-02
4.57973003e-01 1.41256258e-01 7.56301641e-01 6.20933771e-01
6.43124804e-02 -1.52642712e-01 -1.07030988e+00 1.29743207e+00
1.69235408e-01 -1.92899632e+00 3.55205446e-01 2.51856834e-01
9.43970680e-01 -1.01091295e-01 1.26826495e-01 -1.71901226e-01
1.98923964e-02 -1.22530556e+00 7.62622058e-01 7.97403157e-01
1.30861533e+00 -5.59809625e-01 -2.08822731e-02 3.28922212e-01
-1.14405811e+00 5.34311175e-01 -2.08157122e-01 -8.20849743e-03
3.22998345e-01 8.42141926e-01 -3.33985865e-01 8.33795607e-01
6.26292646e-01 7.23736227e-01 -1.29556119e-01 4.53191847e-01
-3.19306403e-02 1.39816061e-01 -3.32258135e-01 3.97333890e-01
-3.44194978e-01 -5.37561953e-01 6.64385259e-01 4.89717603e-01
4.40302968e-01 8.61691907e-02 -2.46012211e-01 1.46481287e+00
-9.97749865e-02 4.76842839e-03 -9.85061467e-01 3.23900789e-01
6.36030808e-02 1.07093799e+00 -6.77851260e-01 -1.63971573e-01
-3.17095637e-01 1.18709838e+00 3.41502458e-01 6.06206596e-01
-6.17659271e-01 2.19080687e-01 9.66715813e-01 6.01981223e-01
2.68081635e-01 -5.18432021e-01 -7.19718456e-01 -1.37934947e+00
1.85238749e-01 -5.32935381e-01 -4.09924984e-01 -1.16217458e+00
-1.06802261e+00 5.17400324e-01 -8.26066434e-02 -1.14217627e+00
-3.15568708e-02 -6.72316492e-01 -1.48486182e-01 1.01846051e+00
-1.70811570e+00 -1.03624260e+00 -4.61745083e-01 3.49830657e-01
6.92305505e-01 3.85651499e-01 1.06574881e+00 2.25846142e-01
-2.62951683e-02 1.08352207e-01 3.08666807e-02 -4.44600463e-01
2.85262555e-01 -1.15119004e+00 7.16927886e-01 4.45629388e-01
2.16827050e-01 3.64634961e-01 4.66142267e-01 -6.00663960e-01
-1.66156781e+00 -9.09474313e-01 1.74718779e-02 -6.91277206e-01
-1.38832062e-01 -5.71580112e-01 -1.06475270e+00 5.67490995e-01
-1.96508672e-02 1.63581774e-01 4.04008001e-01 -1.01068757e-01
-4.35369670e-01 -9.45987776e-02 -1.28734529e+00 7.30354309e-01
1.40566671e+00 -5.86840093e-01 -1.29684791e-01 1.33134171e-01
9.72555220e-01 -6.62041068e-01 -9.25718486e-01 5.28241873e-01
6.76720321e-01 -1.37584865e+00 1.25485659e+00 -9.60311219e-02
5.69418550e-01 -3.92045945e-01 -7.21135199e-01 -1.23644900e+00
-1.72774464e-01 -5.91805935e-01 -4.18498486e-01 8.11836004e-01
2.29451463e-01 -7.93932378e-01 8.47784758e-01 5.03088236e-01
-3.09386075e-01 -1.06292641e+00 -9.40036237e-01 -5.76210022e-01
-9.85636935e-02 -6.21440828e-01 6.48739755e-01 9.48948801e-01
-9.86920714e-01 2.70502623e-02 -2.82332808e-01 1.76027596e-01
9.81858969e-01 3.68263930e-01 8.03097129e-01 -1.44936848e+00
-4.46861386e-01 -3.28599304e-01 -1.24303393e-01 -1.41593599e+00
1.74879003e-02 -7.96782434e-01 -1.68478832e-01 -1.86967957e+00
-2.21905619e-01 -6.50776088e-01 2.41273493e-01 -1.19754724e-01
1.72385022e-01 4.18720752e-01 -6.35692403e-02 2.17253366e-03
1.08113654e-01 8.75489056e-01 1.67865551e+00 8.11085850e-03
-1.85496673e-01 -9.30127576e-02 -5.75117886e-01 9.08772886e-01
6.70826316e-01 2.85211094e-02 -5.99961996e-01 -8.43403876e-01
6.37453973e-01 2.73568273e-01 5.60839474e-01 -8.22807372e-01
-4.34556097e-01 -1.84000432e-01 7.09195733e-01 -6.47663295e-01
9.86202419e-01 -8.42908263e-01 5.33342004e-01 -4.84594107e-02
1.20783448e-01 -5.58941253e-02 2.28269234e-01 6.30800188e-01
1.17498681e-01 3.14873397e-01 7.81830549e-01 -2.67761320e-01
-3.39982480e-01 4.93891895e-01 1.40326008e-01 -4.05603088e-02
6.73336565e-01 -6.25312328e-01 -5.89822084e-02 -4.80334103e-01
-5.37218094e-01 -2.64834106e-01 1.11533904e+00 2.79005319e-01
1.07454228e+00 -1.29625118e+00 -5.88033736e-01 5.70943415e-01
-1.19784661e-01 7.03511775e-01 5.97433031e-01 3.87853831e-01
-8.36588800e-01 8.15549195e-02 3.01518664e-02 -7.92149007e-01
-9.88870561e-01 1.46164715e-01 5.67828596e-01 -1.41253555e-02
-1.03845966e+00 7.63516366e-01 1.05168843e+00 -6.95133567e-01
-4.69426662e-02 -6.48993731e-01 2.64986306e-01 -3.70372653e-01
3.95448238e-01 3.91401887e-01 -3.44064496e-02 -3.84534806e-01
-5.01776189e-02 8.74440193e-01 3.97366405e-01 -1.62129030e-01
1.55399144e+00 -1.17313698e-01 9.58140753e-03 7.88768947e-01
1.10094976e+00 4.28543687e-01 -1.62166524e+00 1.96610279e-02
-7.09267735e-01 -8.03123653e-01 3.46424043e-01 -5.85898101e-01
-1.12437475e+00 9.81495261e-01 4.67975259e-01 5.42332046e-02
9.65817213e-01 -1.27685875e-01 9.14351046e-01 1.21826060e-01
8.38355839e-01 -6.15678728e-01 -2.03426883e-01 1.97133929e-01
1.24849379e+00 -9.63894427e-01 2.88438529e-01 -8.34804356e-01
-6.38737008e-02 9.02480721e-01 4.69277024e-01 -2.06995264e-01
5.77989638e-01 4.10875410e-01 -1.54425517e-01 -2.50394911e-01
-5.89574397e-01 3.19265008e-01 5.59081852e-01 6.62840784e-01
4.81146216e-01 1.66678816e-01 3.84326577e-01 -9.07570794e-02
-1.99113071e-01 -1.93587959e-01 4.49013919e-01 5.38556159e-01
-2.02068374e-01 -8.60996604e-01 -3.43526125e-01 5.84089458e-01
-1.48253530e-01 -1.32374555e-01 -1.10205384e-02 5.99170685e-01
-7.41707683e-02 3.30363393e-01 3.29161078e-01 -1.42035618e-01
4.47112441e-01 -2.21442521e-01 1.01333845e+00 -7.49155045e-01
-2.13640437e-01 5.61594404e-03 8.10510293e-03 -8.96293581e-01
-2.67773509e-01 -4.77939636e-01 -1.06013763e+00 -2.62754530e-01
-1.62027732e-01 -2.57929683e-01 7.36196160e-01 6.03663146e-01
5.07334709e-01 6.67659581e-01 7.40166306e-01 -1.39721310e+00
-7.80899376e-02 -5.19060314e-01 -5.51521659e-01 7.36797825e-02
6.46999598e-01 -7.53993988e-01 -4.05026466e-01 2.12665573e-02] | [9.218076705932617, -3.149090528488159] |
c9a660c9-6d9c-42fa-a955-54237953a07e | road-segmentation-in-sar-satellite-images | 1802.01445 | null | http://arxiv.org/abs/1802.01445v2 | http://arxiv.org/pdf/1802.01445v2.pdf | Road Segmentation in SAR Satellite Images with Deep Fully-Convolutional Neural Networks | Remote sensing is extensively used in cartography. As transportation networks
grow and change, extracting roads automatically from satellite images is
crucial to keep maps up-to-date. Synthetic Aperture Radar satellites can
provide high resolution topographical maps. However roads are difficult to
identify in these data as they look visually similar to targets such as rivers
and railways. Most road extraction methods on Synthetic Aperture Radar images
still rely on a prior segmentation performed by classical computer vision
algorithms. Few works study the potential of deep learning techniques, despite
their successful applications to optical imagery. This letter presents an
evaluation of Fully-Convolutional Neural Networks for road segmentation in SAR
images. We study the relative performance of early and state-of-the-art
networks after carefully enhancing their sensitivity towards thin objects by
adding spatial tolerance rules. Our models shows promising results,
successfully extracting most of the roads in our test dataset. This shows that,
although Fully-Convolutional Neural Networks natively lack efficiency for road
segmentation, they are capable of good results if properly tuned. As the
segmentation quality does not scale well with the increasing depth of the
networks, the design of specialized architectures for roads extraction should
yield better performances. | ['Seyed Majid Azimi', 'Corentin Henry', 'Nina Merkle'] | 2018-02-05 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 4.36642826e-01 2.01838940e-01 -4.18913886e-02 -5.75425923e-01
-3.80419433e-01 -4.97151732e-01 6.48145616e-01 -3.62232864e-01
-6.85838521e-01 8.34649563e-01 -3.30137640e-01 -7.19345987e-01
-1.03103027e-01 -1.41680324e+00 -6.79437160e-01 -5.26363730e-01
-4.26789969e-01 5.90290129e-01 5.88996232e-01 -4.93685186e-01
-5.94872199e-02 1.18547189e+00 -1.61127627e+00 -9.41841155e-02
1.12215841e+00 8.03513288e-01 2.33458474e-01 5.62119365e-01
-2.22165897e-01 3.61369312e-01 -1.22798793e-01 -1.16446465e-01
6.02639139e-01 -4.02912721e-02 -9.02270436e-01 8.81612226e-02
7.04430997e-01 -5.18047571e-01 -3.51136774e-01 9.13442731e-01
3.21816415e-01 -2.54868865e-01 8.34015548e-01 -8.28247190e-01
-2.14315102e-01 6.43520117e-01 -5.43430328e-01 4.02783602e-01
-4.61558849e-01 6.84206784e-02 8.37349355e-01 -8.12392116e-01
5.74712694e-01 8.18310678e-01 8.73570859e-01 6.03098832e-02
-1.09074235e+00 -5.76766729e-01 8.30759704e-02 9.99079049e-02
-1.56370544e+00 -5.07676065e-01 5.59462845e-01 -4.98037070e-01
9.52505171e-01 2.57343411e-01 5.89714468e-01 2.08738357e-01
2.66627409e-02 6.32638097e-01 1.23828018e+00 -2.45310873e-01
-4.16709855e-02 1.14444636e-01 9.02922228e-02 4.08288896e-01
5.24716675e-01 1.49642557e-01 1.23451993e-01 5.73508978e-01
8.16194057e-01 -3.11575472e-01 -1.69328004e-01 -2.58325368e-01
-8.36474836e-01 9.41958606e-01 1.00777328e+00 4.64384526e-01
-7.05110371e-01 1.60360113e-02 -4.16449942e-02 1.98156625e-01
6.41715229e-01 4.90752101e-01 -4.48622793e-01 4.68716085e-01
-1.75295258e+00 1.53682694e-01 2.97344208e-01 5.56080282e-01
1.11276114e+00 2.87713349e-01 3.15966517e-01 8.13922942e-01
-4.57131900e-02 7.50501454e-01 -2.45643586e-01 -6.74051881e-01
4.89478737e-01 7.24256694e-01 1.23897893e-02 -1.02770615e+00
-8.59261274e-01 -7.84810424e-01 -8.92633379e-01 5.05404055e-01
5.22628784e-01 -2.44281307e-01 -1.43678010e+00 8.78772318e-01
8.75326805e-03 -2.70061374e-01 1.77288968e-02 8.79722953e-01
8.18616927e-01 5.31683207e-01 3.93438749e-02 2.56389946e-01
1.26186478e+00 -5.64645410e-01 -3.96025479e-01 -8.77737761e-01
3.41106087e-01 -5.84817767e-01 8.47592473e-01 2.14105338e-01
-6.06436491e-01 -4.58477408e-01 -1.13672400e+00 1.03470214e-01
-6.71659648e-01 4.43451196e-01 9.19449210e-01 9.32620585e-01
-9.77724969e-01 7.09791243e-01 -7.70692229e-01 -5.37205040e-01
1.01744390e+00 4.20327067e-01 -2.15512633e-01 3.05291396e-02
-1.17840552e+00 1.10058355e+00 4.39005882e-01 7.02217519e-01
-4.72958982e-01 -3.19277018e-01 -7.66134143e-01 -7.97229633e-03
2.76152343e-01 -1.75800130e-01 7.69944966e-01 -9.39456224e-01
-9.88305449e-01 1.07421803e+00 2.47056767e-01 -7.29362071e-01
5.76110125e-01 -2.32297629e-01 -4.79393750e-01 3.01921129e-01
6.62233680e-02 1.18109429e+00 6.99246883e-01 -1.29652178e+00
-9.21532929e-01 -3.74615252e-01 6.62395880e-02 2.40142532e-02
-1.85509715e-02 -5.36261760e-02 -8.44472572e-02 -2.56192923e-01
3.85076314e-01 -9.74401712e-01 -5.27289808e-01 1.22798212e-01
-4.44371283e-01 3.90996486e-01 1.08329272e+00 -7.63410985e-01
9.90516126e-01 -1.65141189e+00 -3.63151610e-01 3.66328448e-01
-4.96656746e-02 6.19050443e-01 -3.59900221e-02 4.34938036e-02
-6.68810681e-02 2.95549005e-01 -7.06414521e-01 4.26761627e-01
-2.42079645e-01 3.93125147e-01 -2.06266180e-01 5.37316918e-01
5.42099416e-01 1.04553699e+00 -3.68314087e-01 -5.45540392e-01
3.40697974e-01 5.64496219e-01 -7.14406520e-02 -1.92127213e-01
-4.31660935e-02 2.41479516e-01 -4.93525952e-01 7.77802944e-01
1.16484797e+00 -8.50419328e-03 -2.59327162e-02 -3.79137248e-01
-5.08942306e-01 1.72337040e-01 -1.03267193e+00 8.84464681e-01
-3.71516734e-01 1.17151093e+00 3.49514216e-01 -1.19461906e+00
1.11586881e+00 -2.10947260e-01 2.99517125e-01 -1.13336658e+00
-1.09394267e-01 3.37615013e-01 1.93738118e-01 -4.27262723e-01
7.66182780e-01 -1.93500310e-01 1.96927994e-01 5.62836230e-02
-4.64840621e-01 -4.16274279e-01 2.45788381e-01 -1.19792297e-01
5.71250677e-01 1.42178699e-01 -4.70007285e-02 -3.48515004e-01
3.90651762e-01 6.13199174e-01 2.35271603e-01 5.35319328e-01
1.44301057e-01 7.26591587e-01 2.84534454e-01 -8.16380322e-01
-1.26737130e+00 -8.55098307e-01 -7.26937890e-01 7.48901606e-01
-7.30204806e-02 4.75558817e-01 -8.02858531e-01 -4.40349311e-01
-9.94200185e-02 4.54100609e-01 -4.41437453e-01 5.46856105e-01
-6.55211508e-01 -1.43769753e+00 7.12202549e-01 5.79537511e-01
1.04032910e+00 -1.01661098e+00 -1.02310550e+00 4.37312663e-01
-1.27262846e-01 -1.45454514e+00 4.55185384e-01 2.08581790e-01
-1.11658978e+00 -1.05085254e+00 -8.95577550e-01 -5.37875831e-01
5.96647739e-01 3.78625005e-01 1.08308542e+00 9.70429331e-02
-4.08120990e-01 -2.77656287e-01 -2.53486931e-01 -3.32171202e-01
-1.59625337e-01 6.64968014e-01 -4.95478213e-01 -1.19336747e-01
3.09443384e-01 -5.67074358e-01 -5.88483334e-01 4.73561168e-01
-7.37803519e-01 5.47442771e-02 1.10197949e+00 4.37190920e-01
3.78351986e-01 3.60143065e-01 3.61309916e-01 -9.03297484e-01
1.46138683e-01 -2.55314529e-01 -7.67148495e-01 -5.47579750e-02
-6.39708281e-01 -4.28841040e-02 2.47580007e-01 2.01478571e-01
-8.65196645e-01 4.13745850e-01 -3.91766757e-01 2.49688998e-01
-5.59693158e-01 6.68329775e-01 3.36968340e-02 -3.52542609e-01
8.89962554e-01 1.05770379e-01 -2.58884635e-02 -3.78658623e-01
3.43562871e-01 7.71042407e-01 4.39911872e-01 3.22588384e-02
1.05639243e+00 8.64049792e-01 1.12201445e-01 -1.50896037e+00
-7.67163217e-01 -3.50767910e-01 -1.07802165e+00 -3.11125934e-01
9.08509731e-01 -8.42285097e-01 -1.89487755e-01 6.06504977e-01
-7.95140505e-01 -6.07652843e-01 -3.51622552e-02 4.38635409e-01
-2.21942648e-01 1.52892411e-01 -1.77473828e-01 -7.56732404e-01
-3.47717524e-01 -1.02914155e+00 8.14240158e-01 3.59673470e-01
1.43451259e-01 -8.20352316e-01 -3.83426666e-01 3.49341214e-01
7.61513531e-01 3.49542886e-01 7.43429542e-01 -2.11632550e-01
-9.03553009e-01 -3.06918938e-02 -7.98756957e-01 4.16872412e-01
-6.17006458e-02 2.25804552e-01 -1.04270673e+00 9.31463167e-02
-7.37646818e-01 5.27111664e-02 1.51732230e+00 7.14625895e-01
7.08413303e-01 -1.35446548e-01 -5.29078543e-01 6.67309999e-01
1.63443661e+00 1.58430804e-02 1.21021807e+00 8.48734081e-01
7.68636227e-01 9.59532380e-01 5.17826140e-01 -1.02057718e-02
2.55299538e-01 3.81902635e-01 7.45887637e-01 -6.77838802e-01
-2.38449305e-01 2.95572251e-01 7.47346431e-02 -1.32923655e-03
-4.00070071e-01 7.55156502e-02 -1.40856171e+00 8.99024785e-01
-1.34171855e+00 -9.34101582e-01 -8.52508605e-01 1.95258963e+00
3.95869762e-01 4.26284254e-01 8.48938450e-02 2.95328766e-01
5.26321173e-01 2.22989053e-01 -4.54254419e-01 -2.99775094e-01
-5.77039778e-01 4.85271007e-01 1.33754945e+00 4.62414503e-01
-1.54853916e+00 1.26019597e+00 6.42310953e+00 5.64211369e-01
-1.42643023e+00 -1.34418055e-01 6.19438291e-01 2.86959201e-01
-1.05389021e-01 -1.63796376e-02 -1.00686610e+00 6.49767742e-02
8.44715893e-01 6.35716379e-01 -5.55166118e-02 5.71873605e-01
3.74678493e-01 -5.14155567e-01 -2.04913259e-01 4.16622102e-01
-4.07229632e-01 -1.58950067e+00 4.80873547e-02 2.55788982e-01
6.96163833e-01 6.44356966e-01 -4.58391011e-02 6.12192415e-02
1.57348335e-01 -1.45481157e+00 5.02543211e-01 6.22265756e-01
7.48117805e-01 -8.49759340e-01 9.18711364e-01 1.69919640e-01
-1.21123683e+00 -6.23134002e-02 -5.10894477e-01 -1.21913441e-02
1.93771273e-01 8.10230374e-01 -9.04121459e-01 4.29503858e-01
7.25507319e-01 5.48400223e-01 -9.35336411e-01 1.20884871e+00
-3.11579049e-01 7.38232851e-01 -6.41737580e-01 2.19546959e-01
8.24877560e-01 -4.05712575e-01 2.03274250e-01 1.39807546e+00
2.38244370e-01 -1.12862848e-01 -3.85608710e-02 7.95542419e-01
2.80418634e-01 -7.63073144e-03 -7.31110692e-01 -1.77202076e-02
2.09125787e-01 1.43227732e+00 -1.28139365e+00 -5.30450195e-02
-2.93929785e-01 4.43072081e-01 -8.12428743e-02 3.47484052e-01
-6.81458294e-01 -5.01911998e-01 5.04680753e-01 7.51905024e-01
5.90889454e-01 -6.27577186e-01 -6.71657979e-01 -6.23557389e-01
-8.42904747e-02 -5.37952900e-01 -1.13763688e-02 -7.28989899e-01
-7.22517192e-01 7.92460203e-01 4.77520339e-02 -1.15697324e+00
6.80038631e-02 -7.52702653e-01 -5.11568904e-01 7.78093934e-01
-2.23605776e+00 -1.60316002e+00 -5.65661073e-01 1.82050273e-01
2.92873353e-01 -1.61748603e-02 5.18584192e-01 3.11111093e-01
-3.13780278e-01 5.23075908e-02 2.46620193e-01 5.12408435e-01
2.73161501e-01 -9.87853110e-01 7.56462812e-01 1.07583463e+00
1.58189405e-02 1.02256164e-01 6.00488842e-01 -5.72567403e-01
-8.04515421e-01 -1.26285565e+00 6.82192981e-01 1.68999046e-01
4.97552603e-01 -9.86288264e-02 -1.09196115e+00 5.43121159e-01
9.22615081e-02 -1.02644540e-01 1.18439563e-01 6.12005293e-02
2.17605382e-02 -3.08168203e-01 -9.50124025e-01 4.77235794e-01
8.10501456e-01 -2.31559530e-01 -3.60531241e-01 2.23369762e-01
5.57779297e-02 -1.45221710e-01 -5.51112235e-01 7.08333910e-01
5.28695941e-01 -1.18031156e+00 1.02179086e+00 -1.76772878e-01
5.03607213e-01 -3.27577978e-01 7.26041794e-02 -1.15606308e+00
-1.90488547e-01 -5.00053875e-02 6.65498018e-01 8.92595768e-01
8.27289820e-01 -5.11711776e-01 1.21777821e+00 1.03158079e-01
-1.15485080e-01 -3.29278082e-01 -7.63727069e-01 -9.30751979e-01
3.52370441e-01 -3.73871773e-01 3.85319382e-01 9.90958989e-01
-9.26467240e-01 4.37772304e-01 -3.15717310e-02 4.25756454e-01
7.03326166e-01 1.18516274e-01 7.27122068e-01 -1.74649060e+00
4.83827680e-01 -7.08263159e-01 -4.20444489e-01 -7.81765342e-01
-2.84818076e-02 -7.84097195e-01 -3.12914723e-03 -2.05497980e+00
-3.51493567e-01 -8.20862532e-01 4.66511548e-01 7.31131077e-01
7.93114603e-02 7.50095248e-01 -1.56742200e-01 1.19797476e-01
1.04916945e-01 2.96308249e-01 1.13400054e+00 -3.37703258e-01
-1.81928337e-01 1.25149861e-01 -3.69127154e-01 8.53384674e-01
1.21169269e+00 -4.05915260e-01 -1.36346996e-01 -4.95254964e-01
4.36155021e-01 -3.81745994e-01 5.05939424e-01 -1.29446530e+00
2.22154975e-01 2.96651907e-02 4.13109452e-01 -1.11460769e+00
1.94301471e-01 -8.53975654e-01 1.35880291e-01 4.14491445e-01
2.47076884e-01 -3.44278991e-01 5.31595647e-01 1.26410097e-01
-2.81133622e-01 -2.57496923e-01 1.09021890e+00 -1.81634277e-01
-1.16902614e+00 2.31833652e-01 -7.99852490e-01 -2.32584134e-01
7.41001666e-01 -8.13863814e-01 -3.06770116e-01 -1.58773586e-01
-7.22567439e-01 2.97323883e-01 4.00076360e-01 1.97118983e-01
5.69560766e-01 -7.12251902e-01 -1.05823183e+00 2.06094414e-01
-2.90589333e-02 2.44206950e-01 1.95157379e-01 8.04717481e-01
-1.20463109e+00 5.23595691e-01 -6.10702753e-01 -7.40871608e-01
-1.08597565e+00 8.29025730e-02 6.89687729e-01 -1.60217106e-01
-6.64360583e-01 6.03597462e-01 -1.74994737e-01 -5.00700235e-01
-2.78913677e-01 -4.74816054e-01 -5.86139679e-01 5.60163379e-01
2.91912466e-01 3.23624849e-01 2.89941192e-01 -7.92278826e-01
-4.68811423e-01 8.64809871e-01 -2.92397141e-02 -1.05407745e-01
1.77999139e+00 -1.56990156e-01 -4.01433259e-02 -7.92863518e-02
7.71749139e-01 -2.86823720e-01 -1.30952454e+00 -2.27004409e-01
1.38757572e-01 -2.30416715e-01 6.22190237e-01 -7.90065527e-01
-1.54719174e+00 1.21860635e+00 7.03127086e-01 2.30357990e-01
1.06157732e+00 -1.68082908e-01 6.01110399e-01 5.61693013e-01
1.68451190e-01 -1.21775854e+00 -5.67481339e-01 6.87199533e-01
6.66350424e-01 -1.44131184e+00 3.81377637e-01 -5.61133265e-01
-5.02546370e-01 1.31077492e+00 2.39163935e-01 -1.74142882e-01
7.19522297e-01 4.39078838e-01 1.86543405e-01 -4.12943214e-01
8.93981606e-02 -1.05800724e+00 2.57323503e-01 8.13676238e-01
2.38614604e-01 2.35487998e-01 -2.11150825e-01 -1.16312616e-01
-3.94214272e-01 -1.67535484e-01 7.43948340e-01 7.22875774e-01
-9.65120256e-01 -8.03993642e-01 -5.41814387e-01 6.32584989e-01
-2.94284374e-01 -2.92010099e-01 -4.66521502e-01 1.22705674e+00
1.11651473e-01 8.37614596e-01 4.31155533e-01 -8.94412100e-02
3.51954699e-01 -2.50989497e-01 3.22279900e-01 -4.02765483e-01
-4.27287728e-01 -3.75683755e-02 5.68468809e-01 -2.06612766e-01
-8.32570791e-01 -5.86364686e-01 -1.20838785e+00 -3.16302180e-01
-3.08569402e-01 -7.90437162e-02 9.65951085e-01 1.00053203e+00
-3.53385918e-02 4.13878232e-01 4.14079338e-01 -9.61280823e-01
5.26566096e-02 -8.33957493e-01 -7.42645264e-01 -1.93964437e-01
1.92976117e-01 -5.68879724e-01 -4.17373292e-02 6.64320067e-02] | [9.219640731811523, -1.5230505466461182] |
e14e0b21-869b-4739-8324-5816d6a44438 | 3d-human-shape-style-transfer | 2109.01587 | null | https://arxiv.org/abs/2109.01587v1 | https://arxiv.org/pdf/2109.01587v1.pdf | 3D Human Shape Style Transfer | We consider the problem of modifying/replacing the shape style of a real moving character with those of an arbitrary static real source character. Traditional solutions follow a pose transfer strategy, from the moving character to the source character shape, that relies on skeletal pose parametrization. In this paper, we explore an alternative approach that transfers the source shape style onto the moving character. The expected benefit is to avoid the inherently difficult pose to shape conversion required with skeletal parametrization applied on real characters. To this purpose, we consider image style transfer techniques and investigate how to adapt them to 3D human shapes. Adaptive Instance Normalisation (AdaIN) and SPADE architectures have been demonstrated to efficiently and accurately transfer the style of an image onto another while preserving the original image structure. Where AdaIN contributes with a module to perform style transfer through the statistics of the subjects and SPADE contribute with a residual block architecture to refine the quality of the style transfer. We demonstrate that these approaches are extendable to the 3D shape domain by proposing a convolutional neural network that applies the same principle of preserving the shape structure (shape pose) while transferring the style of a new subject shape. The generated results are supervised through a discriminator module to evaluate the realism of the shape, whilst enforcing the decoder to synthesise plausible shapes and improve the style transfer for unseen subjects. Our experiments demonstrate an average of $\approx 56\%$ qualitative and quantitative improvements over the baseline in shape transfer through optimization-based and learning-based methods. | ['Edmond Boyer', 'Joao Regateiro'] | 2021-09-03 | null | null | null | null | ['pose-transfer'] | ['computer-vision'] | [ 6.85449660e-01 3.81228864e-01 4.44531590e-01 -4.56773728e-01
-5.50011337e-01 -8.43171179e-01 9.07366335e-01 -3.79309833e-01
-5.95793903e-01 4.75010395e-01 -2.25472003e-02 1.59204662e-01
1.96662173e-01 -8.41551304e-01 -9.74730015e-01 -7.41091251e-01
3.79165709e-01 7.49060869e-01 4.13189620e-01 -4.60252017e-01
2.38666952e-01 8.04053605e-01 -1.27595258e+00 4.95267153e-01
5.58081686e-01 8.42236280e-01 8.79637823e-02 8.40747118e-01
2.17438899e-02 8.50026757e-02 -7.67293155e-01 -7.99707472e-01
6.50485694e-01 -6.99705005e-01 -7.20370352e-01 2.99792439e-01
8.23427379e-01 -2.53848433e-01 -1.39842749e-01 8.19287121e-01
6.98468506e-01 8.99789855e-02 8.75103951e-01 -9.32888329e-01
-6.62119925e-01 1.20107390e-01 -3.29037189e-01 -3.20780873e-01
3.35522026e-01 2.93959916e-01 6.11060321e-01 -5.97877443e-01
9.34656739e-01 1.40329075e+00 6.99654996e-01 9.39918041e-01
-1.60814321e+00 -2.88612396e-01 -1.05452336e-01 -1.11103117e-01
-1.13417327e+00 -3.76947135e-01 8.67310166e-01 -4.04143304e-01
4.92873281e-01 4.05532241e-01 7.40585268e-01 1.27190197e+00
7.54185915e-02 6.88130975e-01 1.11042523e+00 -5.14345288e-01
1.28480956e-01 1.86521426e-01 -5.65295994e-01 4.79821324e-01
-1.79338098e-01 2.23205715e-01 -3.25980455e-01 6.39916360e-02
1.02649868e+00 -5.03700137e-01 -2.98826814e-01 -8.02625477e-01
-1.08191049e+00 6.31270230e-01 5.27079582e-01 2.55952805e-01
-3.03002954e-01 2.82138616e-01 2.56742835e-01 3.01084578e-01
4.51616287e-01 5.11070311e-01 -2.85213292e-01 -5.06009944e-02
-1.10352397e+00 5.64612865e-01 8.14607859e-01 1.14979661e+00
6.44789577e-01 3.27494502e-01 -4.08290446e-01 7.19989955e-01
2.47850269e-01 5.63388169e-01 3.40976179e-01 -1.05941236e+00
1.73053488e-01 6.10520661e-01 -9.02407989e-02 -8.23519051e-01
-1.01467788e-01 -2.87656814e-01 -6.24378502e-01 8.07519436e-01
6.64300799e-01 3.05145383e-02 -1.12023306e+00 1.74756908e+00
4.76679176e-01 -2.29028553e-01 -7.70342350e-02 8.16454172e-01
6.51391745e-01 4.46930021e-01 4.92252931e-02 3.36739779e-01
1.25198746e+00 -6.73601985e-01 -2.41668522e-01 -1.57186732e-01
1.93286940e-01 -9.39321458e-01 1.19548714e+00 2.41213098e-01
-1.61173880e+00 -7.02150166e-01 -9.68325198e-01 -3.18532169e-01
-1.37955189e-01 6.78655654e-02 -4.25546877e-02 7.43571639e-01
-1.25070405e+00 9.52108026e-01 -6.60851002e-01 -4.18027014e-01
5.27892113e-01 4.91692871e-01 -4.36231703e-01 2.39521667e-01
-8.59283447e-01 9.75940883e-01 3.33662480e-01 2.16309726e-02
-6.98899031e-01 -8.89145255e-01 -9.31551814e-01 -1.52713776e-01
-4.13106708e-03 -9.83622253e-01 9.80898798e-01 -1.62504601e+00
-1.98771083e+00 1.23706794e+00 1.57017395e-01 -1.76607907e-01
1.12358797e+00 -5.65600991e-02 9.66946855e-02 5.23379594e-02
-9.36259255e-02 1.09610152e+00 1.13378358e+00 -1.47613132e+00
-2.87033409e-01 -1.56957790e-01 5.17605431e-02 2.37976506e-01
-3.28793488e-02 -1.55599877e-01 -7.10937679e-01 -1.13747191e+00
-1.54319152e-01 -1.12914836e+00 -6.38491008e-03 5.42668939e-01
-3.36064517e-01 2.08863050e-01 8.60812247e-01 -8.19991767e-01
5.58883727e-01 -2.07237196e+00 6.98915482e-01 3.61938030e-01
-2.01608315e-01 3.80535513e-01 -5.82728744e-01 2.79663593e-01
-1.27173766e-01 -1.46058828e-01 -7.49854684e-01 -5.97880840e-01
-1.68435555e-02 2.78202266e-01 -9.51585546e-02 3.72888029e-01
5.75430691e-01 1.10175109e+00 -6.78331435e-01 -2.18184724e-01
1.65166147e-02 9.52249646e-01 -9.13585186e-01 4.46377844e-01
-3.47625166e-01 8.88750315e-01 -1.67093620e-01 1.69966295e-01
6.46084249e-01 2.67465085e-01 3.34139243e-02 -3.90356094e-01
2.08062485e-01 -8.02099407e-02 -1.17703927e+00 1.90527403e+00
-3.72391313e-01 3.65265995e-01 3.05387199e-01 -7.32254505e-01
1.17112708e+00 2.51518160e-01 1.81038946e-01 -5.94577074e-01
2.35211968e-01 3.39094520e-01 1.16436720e-01 -2.18531698e-01
4.13740486e-01 -5.58774650e-01 -5.83240390e-02 3.67094994e-01
1.42290369e-01 -6.54858232e-01 -1.15000509e-01 -1.71447143e-01
6.57340288e-01 9.84872758e-01 -1.72758531e-02 -4.70661730e-01
6.93506002e-01 -1.19202770e-01 1.87024668e-01 2.79726297e-01
1.86156020e-01 1.23936903e+00 4.70652312e-01 -3.91410440e-01
-1.64406502e+00 -1.23803544e+00 9.70194712e-02 1.01380861e+00
-1.95666417e-01 1.69514552e-01 -1.16154540e+00 -7.41182089e-01
-6.74146637e-02 7.99876213e-01 -7.98714995e-01 -1.84677839e-01
-1.16644955e+00 -2.83451140e-01 7.15228081e-01 5.92031658e-01
4.61296260e-01 -1.24798417e+00 -6.52739644e-01 1.21352144e-01
1.33767454e-02 -9.20242131e-01 -8.36375117e-01 -1.31614566e-01
-7.14308441e-01 -7.00350940e-01 -1.29670131e+00 -7.60526478e-01
8.98690820e-01 -5.64980507e-01 1.01725769e+00 5.85746132e-02
-2.06715405e-01 3.98344487e-01 -2.25009143e-01 -2.66628593e-01
-9.22088027e-01 3.22082527e-02 -1.53200403e-01 3.13793987e-01
-2.63969958e-01 -7.03071117e-01 -6.51481271e-01 2.94572234e-01
-1.35205364e+00 1.84600577e-01 4.80288118e-01 7.24224567e-01
5.29549658e-01 -5.96565366e-01 1.94922253e-01 -8.58325303e-01
3.68453413e-01 6.86967671e-02 -3.85155022e-01 5.99442907e-02
-2.36624837e-01 3.44205230e-01 5.88906646e-01 -7.26489961e-01
-1.04711628e+00 3.10297608e-01 -3.90008748e-01 -4.40923095e-01
-3.18656534e-01 -1.50117278e-01 -6.26325011e-01 -1.58773467e-01
8.05237651e-01 3.26615512e-01 3.38578880e-01 -5.54150105e-01
6.69740558e-01 3.23370308e-01 7.28589356e-01 -8.57203126e-01
1.20303321e+00 6.56167030e-01 2.11048931e-01 -6.12519920e-01
-3.05143952e-01 8.29143599e-02 -1.11178744e+00 -1.49799094e-01
9.31105196e-01 -4.75458324e-01 -4.30440724e-01 5.80501020e-01
-1.37328517e+00 -5.87224483e-01 -7.31226683e-01 8.44500586e-02
-1.04315805e+00 4.66482431e-01 -3.87179881e-01 -2.77194232e-01
-3.61445129e-01 -1.09061384e+00 1.37364352e+00 -3.34687568e-02
-4.93310004e-01 -1.09231412e+00 2.15081796e-01 1.92281216e-01
5.70645809e-01 5.68170249e-01 1.02193439e+00 -3.70935529e-01
-3.15059394e-01 -9.66178328e-02 -3.05577870e-02 5.98370314e-01
1.28176048e-01 -2.66806334e-01 -9.95133698e-01 -5.43031871e-01
-1.24588616e-01 -8.48071128e-02 5.65861642e-01 1.03467166e-01
6.23890460e-01 -3.31450671e-01 1.17817998e-01 8.77015650e-01
1.20964825e+00 1.69490993e-01 9.28896666e-01 2.29654118e-01
9.70458806e-01 1.01904607e+00 3.35639976e-02 5.38534299e-02
-5.09086922e-02 1.03768349e+00 1.54811576e-01 -2.14648783e-01
-7.43804812e-01 -3.23017895e-01 5.85596859e-01 4.46343541e-01
-5.05307496e-01 -2.05445111e-01 -6.23219848e-01 4.58744824e-01
-1.47444010e+00 -8.37008357e-01 -1.68107823e-02 2.20439506e+00
1.11982155e+00 4.80027236e-02 3.58600944e-01 6.11488484e-02
6.22952163e-01 -1.44219398e-01 -3.94164562e-01 -7.39462912e-01
-2.27572054e-01 5.64683735e-01 2.04000294e-01 6.45761907e-01
-8.65002036e-01 9.51844752e-01 5.78688192e+00 8.29669118e-01
-1.30371904e+00 -1.24547981e-01 5.34369528e-01 5.20486347e-02
-4.07795668e-01 -1.24364510e-01 -4.77386534e-01 1.95125535e-01
6.59493446e-01 1.98783726e-03 2.72878408e-01 5.36181688e-01
8.08359757e-02 2.38740116e-01 -1.39831579e+00 5.56698740e-01
3.45447987e-01 -9.98040617e-01 4.79565322e-01 2.92625390e-02
5.80246568e-01 -5.61315656e-01 2.31409565e-01 3.28507014e-02
-9.83634889e-02 -1.12497163e+00 1.16314697e+00 9.11563993e-01
1.22631872e+00 -6.26832843e-01 4.01486486e-01 1.88740402e-01
-9.18670535e-01 4.57808226e-01 -9.10923928e-02 1.90597072e-01
2.17096567e-01 -5.49113452e-02 -8.50732505e-01 5.53510904e-01
3.66065443e-01 2.98794001e-01 -6.94636822e-01 8.97216201e-01
-2.35964537e-01 1.27827674e-01 -2.47252584e-01 2.84495234e-01
1.04554199e-01 -2.36988053e-01 9.51638222e-01 1.31534827e+00
3.68081123e-01 -2.00120062e-01 -2.39568070e-01 1.28163171e+00
1.35490045e-01 1.73629344e-01 -5.18095911e-01 2.97080457e-01
-1.66799054e-01 1.10318136e+00 -6.54922545e-01 -2.31902167e-01
-3.33629213e-02 1.57920432e+00 7.35574514e-02 2.17827782e-01
-5.93445837e-01 -2.61250198e-01 3.43723953e-01 5.88939190e-01
5.45850933e-01 -5.08107990e-02 -4.75441605e-01 -8.99123311e-01
1.46347269e-01 -1.03606045e+00 5.25734164e-02 -7.76556015e-01
-1.33103120e+00 9.15536582e-01 1.12811714e-01 -1.25799119e+00
-3.25052887e-01 -6.34027421e-01 -6.81390166e-01 1.24579930e+00
-1.13043308e+00 -1.67571902e+00 -2.32372239e-01 4.84012425e-01
5.33518553e-01 -1.84979320e-01 7.20386505e-01 1.13517866e-01
9.49648097e-02 8.04735363e-01 -2.94148475e-01 1.70475274e-01
7.74105787e-01 -1.37610543e+00 6.41833484e-01 7.42361128e-01
-5.79241812e-02 3.57260525e-01 7.87316322e-01 -7.32165217e-01
-9.86835957e-01 -1.18451095e+00 6.78522170e-01 -6.80893421e-01
1.62883967e-01 -4.32841480e-01 -1.04841137e+00 4.49223995e-01
3.47537547e-01 -1.86560769e-02 3.49543542e-01 -5.37600458e-01
-3.52857143e-01 1.73137203e-01 -1.35445118e+00 7.12896049e-01
1.10145211e+00 -2.92339593e-01 -7.21951723e-01 -3.07639509e-01
2.86027253e-01 -5.19264221e-01 -9.66601968e-01 4.42474872e-01
7.41505623e-01 -8.06975007e-01 1.05357575e+00 -6.36302173e-01
5.35913229e-01 -4.22018915e-01 -1.04327165e-01 -1.47482264e+00
-2.74534225e-01 -6.50072515e-01 1.85380384e-01 1.21890044e+00
3.07673186e-01 -1.82714298e-01 9.57760274e-01 6.40130043e-01
-8.44125748e-02 -5.54450333e-01 -1.06217694e+00 -7.69826233e-01
6.98325634e-01 -1.00930527e-01 4.74486560e-01 6.53249741e-01
-5.96784711e-01 9.70453098e-02 -3.35103214e-01 -1.01173185e-01
5.63161910e-01 -5.91763668e-02 8.97864521e-01 -8.81537199e-01
-6.50913954e-01 -5.61690450e-01 -5.42217135e-01 -9.50831831e-01
1.56689584e-01 -1.04055452e+00 1.34890620e-02 -1.28321099e+00
-3.60332355e-02 -3.36360812e-01 3.31171960e-01 3.56319547e-01
-1.71555430e-01 6.25798643e-01 6.05031788e-01 1.63592681e-01
-3.99860926e-02 6.77569389e-01 1.75284231e+00 -6.42577186e-02
-1.97061747e-01 1.26011252e-01 -3.90884817e-01 7.12342322e-01
5.39041758e-01 -3.81435037e-01 -2.84526497e-01 -6.15241051e-01
4.63491902e-02 -1.97338670e-01 6.84474707e-01 -8.97284508e-01
-3.98235358e-02 1.02316767e-01 5.56771576e-01 -1.16773397e-01
4.91007537e-01 -8.94948125e-01 3.73420298e-01 5.43155313e-01
-4.12731826e-01 -8.85761809e-03 4.13591683e-01 3.37740749e-01
7.14467987e-02 -1.91141516e-01 1.26230192e+00 -6.90388083e-02
-3.25074732e-01 1.22853242e-01 -2.61665322e-02 1.29602090e-01
8.81662428e-01 -5.88553667e-01 2.01704323e-01 -3.21213663e-01
-1.04396069e+00 -3.45747203e-01 8.31529737e-01 5.00687778e-01
6.06421828e-01 -1.52130151e+00 -1.01573277e+00 3.91839951e-01
-8.63108337e-02 1.02509614e-02 1.05116680e-01 6.05148315e-01
-7.54024684e-01 -9.14763063e-02 -6.44878447e-01 -5.53930223e-01
-1.22289228e+00 3.66546541e-01 7.11777270e-01 1.67530682e-02
-6.31262839e-01 7.50385821e-01 4.21073109e-01 -6.35012984e-01
6.19391836e-02 -1.73602745e-01 -4.76461053e-02 -4.29818667e-02
2.68144071e-01 2.84584284e-01 1.60499699e-02 -9.97866511e-01
-7.26875141e-02 1.01454103e+00 2.68824786e-01 -4.25928861e-01
1.37539458e+00 1.24432929e-01 2.56128572e-02 1.90253258e-01
1.26046002e+00 1.64345458e-01 -1.75469065e+00 -2.18748435e-01
-1.85959622e-01 -4.49271619e-01 -3.57779831e-01 -8.93475711e-01
-1.01076090e+00 9.04766798e-01 6.95585549e-01 -1.45242199e-01
9.58021283e-01 -4.88475300e-02 7.12420881e-01 -1.21150836e-02
2.70485818e-01 -8.24660897e-01 1.15578733e-01 4.27987427e-01
1.45108533e+00 -7.62085438e-01 -2.74472296e-01 -4.19530928e-01
-6.37604237e-01 1.28284299e+00 4.16879028e-01 -3.84593695e-01
3.85359168e-01 2.65942395e-01 1.58574253e-01 -8.94363597e-03
-1.52757287e-01 -1.02476835e-01 8.77930582e-01 8.55706632e-01
4.24250245e-01 -1.67893484e-01 -1.63920864e-01 2.44744956e-01
-4.03181106e-01 -2.04912722e-01 3.83410662e-01 8.02069843e-01
-1.53743271e-02 -1.68560338e+00 -5.59869528e-01 -6.77317977e-02
-3.24507207e-01 -2.96002701e-02 -6.47340894e-01 8.12354505e-01
4.00440454e-01 3.32655698e-01 1.71205595e-01 -4.56542671e-02
8.80202293e-01 1.55972779e-01 9.06147480e-01 -7.00094700e-01
-9.61354375e-01 1.67578563e-01 -4.94873077e-02 -4.46572393e-01
-4.44148809e-01 -6.91958249e-01 -1.01728559e+00 -1.77614555e-01
2.01815158e-01 -2.47586772e-01 4.38223362e-01 7.32332766e-01
2.30676979e-01 4.28515732e-01 3.72539043e-01 -1.28997135e+00
-4.67461199e-01 -7.96319544e-01 -2.90825456e-01 9.54575241e-01
1.12258524e-01 -4.71603155e-01 -1.17722265e-01 4.84307826e-01] | [11.65660572052002, -0.5197592973709106] |
75101422-7661-4a6c-b99d-8ffaeacf4b26 | investigating-pretrained-language-models-for | 2007.08426 | null | https://arxiv.org/abs/2007.08426v3 | https://arxiv.org/pdf/2007.08426v3.pdf | Investigating Pretrained Language Models for Graph-to-Text Generation | Graph-to-text generation aims to generate fluent texts from graph-based data. In this paper, we investigate two recently proposed pretrained language models (PLMs) and analyze the impact of different task-adaptive pretraining strategies for PLMs in graph-to-text generation. We present a study across three graph domains: meaning representations, Wikipedia knowledge graphs (KGs) and scientific KGs. We show that the PLMs BART and T5 achieve new state-of-the-art results and that task-adaptive pretraining strategies improve their performance even further. In particular, we report new state-of-the-art BLEU scores of 49.72 on LDC2017T10, 59.70 on WebNLG, and 25.66 on AGENDA datasets - a relative improvement of 31.8%, 4.5%, and 42.4%, respectively. In an extensive analysis, we identify possible reasons for the PLMs' success on graph-to-text tasks. We find evidence that their knowledge about true facts helps them perform well even when the input graph representation is reduced to a simple bag of node and edge labels. | ['Hinrich Schütze', 'Leonardo F. R. Ribeiro', 'Iryna Gurevych', 'Martin Schmitt'] | 2020-07-16 | null | https://aclanthology.org/2021.nlp4convai-1.20 | https://aclanthology.org/2021.nlp4convai-1.20.pdf | emnlp-nlp4convai-2021-11 | ['kb-to-language-generation', 'kg-to-text'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.79512227e-01 1.08891368e+00 -2.74339437e-01 -9.66407508e-02
-9.28707302e-01 -5.41465819e-01 1.18498850e+00 4.15287882e-01
-2.83390194e-01 1.24298263e+00 5.46139896e-01 -5.83100796e-01
6.11727461e-02 -1.21690094e+00 -9.21265543e-01 -1.82550550e-01
-5.97653091e-02 9.61258173e-01 -9.76254698e-03 -5.37374496e-01
6.70273229e-02 5.08403219e-02 -9.24760520e-01 5.37881374e-01
1.27713931e+00 2.37579018e-01 3.95179279e-02 9.14511681e-01
-5.36905885e-01 1.03919172e+00 -9.19661939e-01 -1.12935448e+00
-1.43540010e-01 -6.92062914e-01 -1.05264580e+00 -2.08482727e-01
6.00743771e-01 1.63509652e-01 -4.12161022e-01 9.14035439e-01
6.37569368e-01 1.56152576e-01 9.86514032e-01 -1.31229079e+00
-9.31159496e-01 1.49091828e+00 -4.45948750e-01 5.79766519e-02
5.64292014e-01 5.82428984e-02 1.23336172e+00 -5.67378461e-01
1.15236938e+00 1.37435400e+00 6.13633871e-01 8.57695222e-01
-1.11114454e+00 -3.80429626e-01 2.27458864e-01 8.55087787e-02
-1.31364393e+00 -4.06194866e-01 4.98190671e-01 -3.10727566e-01
1.53179014e+00 2.04640165e-01 5.54063618e-01 1.29261756e+00
2.25815713e-01 8.40791464e-01 8.92717302e-01 -8.84968579e-01
-1.26210392e-01 -6.05865270e-02 -1.50714163e-02 1.05733991e+00
9.26150739e-01 -2.05197945e-01 -7.04900563e-01 -1.53621316e-01
5.61118960e-01 -9.20241296e-01 -4.08076435e-01 2.17917919e-01
-1.27655303e+00 9.97098386e-01 4.47413266e-01 3.97562176e-01
-3.76615465e-01 3.44544023e-01 3.98888320e-01 3.18914771e-01
7.60280311e-01 7.99226522e-01 -3.19815367e-01 5.31057939e-02
-6.56505167e-01 3.56472015e-01 1.11488283e+00 1.43984854e+00
5.16836524e-01 3.89658272e-01 -6.06212556e-01 5.92874050e-01
1.18525334e-01 4.21497732e-01 5.38483977e-01 -2.98047930e-01
1.03729904e+00 5.02724349e-01 -1.68126941e-01 -9.39738154e-01
-4.30501401e-01 -6.34855568e-01 -1.01949906e+00 -5.48184812e-01
4.10550028e-01 -5.33888280e-01 -1.19821799e+00 1.82184970e+00
-1.68725237e-01 -9.63328108e-02 3.48493278e-01 3.55815649e-01
1.33516169e+00 8.11308265e-01 4.55958515e-01 -1.00443318e-01
1.18584967e+00 -1.00447714e+00 -7.90026546e-01 -6.51868641e-01
1.15349138e+00 -4.86020774e-01 1.18976176e+00 1.79968119e-01
-1.15352976e+00 -5.41554987e-01 -7.81528711e-01 -7.47892782e-02
-6.25283480e-01 2.20770374e-01 7.27133512e-01 5.21987855e-01
-1.47952044e+00 7.15127468e-01 -3.10688794e-01 -5.70813119e-01
6.98051900e-02 -1.00856960e-01 -2.52933323e-01 -2.61360139e-01
-1.59881330e+00 9.86767352e-01 8.93520951e-01 -3.25675964e-01
-7.54457414e-01 -5.20254433e-01 -9.31591749e-01 2.27190237e-02
4.57829773e-01 -1.18661761e+00 1.14253759e+00 -4.42914665e-01
-1.19856775e+00 9.14122820e-01 -1.45915866e-01 -7.35772014e-01
5.06269157e-01 -7.95648992e-02 -5.55187345e-01 1.20730456e-02
3.04026306e-01 8.51561368e-01 4.90043551e-01 -1.15560162e+00
-3.13475698e-01 5.36452383e-02 6.06947169e-02 4.07766849e-01
-1.63161784e-01 -4.05610144e-01 -3.87975663e-01 -7.72169232e-01
-3.36269408e-01 -7.50364304e-01 -2.68910408e-01 -9.54188406e-01
-8.55968356e-01 -7.26831615e-01 -9.09426599e-05 -8.64275992e-01
1.40771592e+00 -1.38279819e+00 2.35930979e-01 1.04607612e-01
3.46877277e-01 1.49001375e-01 -4.87280995e-01 8.69992793e-01
-7.61977071e-03 6.55421019e-01 -1.11493561e-02 -2.67476678e-01
6.57226369e-02 1.20773271e-01 -1.02362253e-01 -2.43280590e-01
3.24051917e-01 1.50821495e+00 -1.14709592e+00 -6.06099963e-01
-1.40835509e-01 1.68902650e-01 -2.29206443e-01 4.62192334e-02
-5.89502335e-01 6.91090617e-03 -4.08552736e-01 2.09929153e-01
1.73433676e-01 -4.16718185e-01 4.38636452e-01 -2.08060890e-02
3.39724690e-01 6.49549603e-01 -5.50335705e-01 1.70577049e+00
-4.36089277e-01 6.10158145e-01 -4.77181077e-01 -6.91398680e-01
8.36406171e-01 3.38215679e-01 -2.52933562e-01 -6.54814005e-01
-1.47669956e-01 1.19742677e-01 3.15715373e-01 -1.65667936e-01
9.07136917e-01 -2.01237530e-01 -1.56381652e-01 4.26715612e-01
3.82790834e-01 -4.55181301e-01 8.89696002e-01 9.78022397e-01
1.16459382e+00 -3.53543647e-02 3.84059340e-01 -3.95923972e-01
4.01273668e-01 1.01201527e-01 -7.09123686e-02 1.12061083e+00
4.86612350e-01 3.56276512e-01 7.64501691e-01 -1.74118593e-01
-7.82352626e-01 -7.95118451e-01 6.24615312e-01 1.04668009e+00
-5.41335821e-01 -1.05661356e+00 -8.59028578e-01 -1.12013245e+00
-7.90136978e-02 1.63240314e+00 -6.72496378e-01 -3.45779598e-01
-6.43736601e-01 -8.36235940e-01 8.29373002e-01 4.56633151e-01
2.50176817e-01 -1.25328016e+00 1.99147493e-01 3.01741004e-01
-4.33735639e-01 -1.28893602e+00 -3.52347493e-01 -1.59185573e-01
-8.40451539e-01 -8.83158088e-01 -7.53705800e-01 -7.73697257e-01
8.61469030e-01 -2.24049971e-03 1.85572112e+00 2.16397718e-01
8.54583532e-02 3.19334537e-01 -6.35171533e-01 -4.33001548e-01
-9.57006097e-01 6.27982855e-01 -3.10014814e-01 -5.92672169e-01
-1.49214551e-01 -2.10193351e-01 -3.82560827e-02 -2.42828563e-01
-6.72253251e-01 5.57751954e-01 7.99163163e-01 5.89204669e-01
3.41178238e-01 4.29173419e-03 8.56618822e-01 -1.43810940e+00
1.29880476e+00 -3.77407968e-01 -1.88239768e-01 5.80495834e-01
-9.55146253e-01 3.52479547e-01 7.13170052e-01 -5.01739271e-02
-1.19005227e+00 -4.29562211e-01 -1.59290984e-01 1.49023712e-01
9.67193544e-02 1.00315380e+00 -1.30667342e-02 2.61391878e-01
9.67300534e-01 2.67805636e-01 -4.28690344e-01 -8.27891380e-02
1.00571728e+00 2.96444356e-01 3.54916811e-01 -7.30199516e-01
7.35775590e-01 -1.86505035e-01 -7.96349049e-02 -7.65205622e-01
-1.04945505e+00 -9.70245674e-02 -3.10209394e-01 -2.04535171e-01
8.50591481e-01 -1.03807735e+00 -1.20918289e-01 2.81130165e-01
-1.37743497e+00 -8.78568292e-01 -3.93781543e-01 1.23972543e-01
-3.71251613e-01 3.48476171e-01 -8.23513091e-01 -5.21424472e-01
-9.20239508e-01 -5.78023434e-01 9.48384225e-01 7.05651343e-02
-3.82791013e-01 -1.59979188e+00 -1.25628412e-02 2.50398755e-01
3.66584212e-01 4.36836213e-01 1.26736975e+00 -8.71346354e-01
-2.43393838e-01 -1.14339992e-01 -3.15243810e-01 -1.08873080e-02
6.34372979e-02 -1.88079104e-01 -5.81910491e-01 -3.80559713e-01
-7.33507454e-01 -4.11248893e-01 9.67359066e-01 1.90425143e-01
9.10900712e-01 -7.37462640e-01 -5.38562536e-01 1.26541764e-01
1.33818698e+00 -1.62663847e-01 7.09991813e-01 -1.56898648e-01
1.05431986e+00 6.10085130e-01 2.39595383e-01 4.29269150e-02
6.47591770e-01 4.34240580e-01 1.47701338e-01 1.43466890e-01
-7.89245367e-01 -8.97448838e-01 4.88361359e-01 1.18595207e+00
-2.64909148e-01 -1.08065677e+00 -1.05807698e+00 6.89161479e-01
-1.63525903e+00 -7.07880378e-01 -7.29241908e-01 1.97786224e+00
1.07995605e+00 4.02308166e-01 -2.92883511e-03 -2.46300071e-01
7.80074239e-01 3.65857333e-01 -2.56284811e-02 -3.88706952e-01
-2.62461424e-01 4.48638737e-01 5.15451849e-01 7.85481930e-01
-6.69099927e-01 1.50105631e+00 6.07777262e+00 1.18692935e+00
-5.82398236e-01 -1.76100992e-02 5.90644777e-01 3.15692544e-01
-7.16645837e-01 -1.07875228e-01 -8.33317101e-01 1.00718006e-01
1.33168209e+00 -7.47884750e-01 3.69946063e-01 6.25006437e-01
-1.21330351e-01 1.30128920e-01 -1.01792586e+00 7.28334606e-01
3.83339435e-01 -1.55130577e+00 7.35178828e-01 -7.28135370e-03
1.11923444e+00 9.30372626e-02 -3.86473864e-01 7.88703978e-01
8.82530987e-01 -1.24152410e+00 5.80685675e-01 4.97122943e-01
9.27871585e-01 -6.52619302e-01 7.69719362e-01 3.42144579e-01
-1.05846441e+00 5.44053257e-01 -4.62015092e-01 -2.89315078e-02
3.34290117e-01 8.11802447e-01 -1.44910049e+00 1.13135171e+00
1.27011510e-02 6.29530668e-01 -9.81017768e-01 5.14871418e-01
-8.84641469e-01 8.94048989e-01 1.11448064e-01 -4.40601707e-01
3.06890547e-01 1.12372726e-01 4.44805533e-01 1.46609509e+00
3.64971906e-01 2.60669435e-03 1.54316753e-01 8.19969237e-01
-5.55387199e-01 3.18788201e-01 -8.50085795e-01 -5.75574756e-01
2.19344452e-01 1.32841814e+00 -7.47338295e-01 -8.16454470e-01
-1.58126220e-01 9.51848567e-01 7.30725706e-01 3.51047367e-01
-6.04344368e-01 -6.87330842e-01 -1.21312901e-01 1.78325504e-01
-4.76296805e-02 -1.92779496e-01 3.93221359e-04 -1.13108611e+00
-2.88416177e-01 -8.75628531e-01 5.94147801e-01 -9.91789997e-01
-1.51693010e+00 8.09228599e-01 2.40595549e-01 -4.49542075e-01
-7.74273038e-01 -5.49413979e-01 -8.53106201e-01 9.43244576e-01
-1.39397061e+00 -1.18112719e+00 -2.31052727e-01 3.84104431e-01
5.10213852e-01 -2.68306613e-01 8.14886987e-01 -8.72081444e-02
-2.38703400e-01 6.75328255e-01 -2.79541790e-01 3.65026623e-01
5.39846599e-01 -1.62712467e+00 1.34007406e+00 9.47169185e-01
5.71229339e-01 4.44656640e-01 6.85830593e-01 -1.19273949e+00
-1.20906472e+00 -1.36598110e+00 1.46899128e+00 -4.41100776e-01
7.52361655e-01 -4.67191368e-01 -7.84083784e-01 8.60726595e-01
6.36854291e-01 -5.57764947e-01 3.18047464e-01 2.64668494e-01
-2.46807218e-01 4.34859961e-01 -8.69527221e-01 7.88822889e-01
1.51631021e+00 -2.54512399e-01 -5.56658983e-01 9.29254472e-01
9.78856504e-01 -5.56559980e-01 -7.77305007e-01 1.83420897e-01
-2.32627410e-02 -6.10077918e-01 6.21531248e-01 -1.00250614e+00
5.74300706e-01 1.40109375e-01 2.62680888e-01 -2.03316736e+00
-5.44738412e-01 -8.72773588e-01 -2.35815078e-01 1.22373831e+00
1.02020574e+00 -7.28415966e-01 8.51442158e-01 1.98002622e-01
-4.77916300e-01 -5.42545199e-01 -5.66427052e-01 -8.42014372e-01
3.18928450e-01 -2.52878100e-01 3.72088403e-01 1.04368496e+00
1.35428905e-01 1.15425324e+00 -1.86660275e-01 -3.19796056e-01
4.81190294e-01 -1.69573471e-01 8.27450573e-01 -1.15373349e+00
-2.53385901e-01 -3.99558574e-01 1.21754676e-01 -6.94180429e-01
5.85952520e-01 -1.51770067e+00 -1.41975731e-01 -2.54714656e+00
7.58326203e-02 -1.32465303e-01 1.88671529e-01 7.36522496e-01
-5.80239475e-01 -1.13181524e-01 2.29893863e-01 -2.71291435e-01
-5.19444048e-01 4.35163170e-01 1.48615170e+00 -2.38270506e-01
-2.37374287e-02 -2.29768947e-01 -9.37392652e-01 3.90855461e-01
1.22136056e+00 -4.20000792e-01 -7.93605328e-01 -7.02373266e-01
6.39723778e-01 5.06497063e-02 9.06702802e-02 -7.36445248e-01
-1.71828996e-02 -2.66349781e-02 2.16029391e-01 -2.68978894e-01
-1.44537449e-01 1.09452628e-01 1.00992717e-01 6.60915971e-01
-5.07942259e-01 2.46311858e-01 4.10715461e-01 5.84706903e-01
2.70492807e-02 -3.79164606e-01 2.13795274e-01 -5.16464353e-01
-5.63013732e-01 6.01075357e-03 -3.28124225e-01 7.94565856e-01
5.76535165e-01 2.25583091e-01 -7.81444013e-01 -9.13348496e-01
-5.20155668e-01 2.14859530e-01 2.60222163e-02 5.35223961e-01
5.63670933e-01 -1.26495755e+00 -1.38420856e+00 -2.44755775e-01
2.14340657e-01 -1.08220726e-01 6.45051375e-02 4.47578818e-01
-4.67927009e-01 7.30205894e-01 1.75931931e-01 8.81929602e-03
-9.02482748e-01 2.71578282e-01 5.65849505e-02 -9.98341382e-01
-4.81753886e-01 9.55449402e-01 -2.40454227e-02 -4.32735682e-01
-2.10093975e-01 -3.97681713e-01 -2.02722281e-01 1.77301578e-02
1.63087636e-01 4.35449451e-01 2.65001118e-01 -3.50005060e-01
-2.78535429e-02 1.60587981e-01 -2.32816234e-01 -1.32855669e-01
1.01117682e+00 2.52874672e-01 -1.22175999e-01 1.42150417e-01
7.54428923e-01 2.52278179e-01 -3.61610442e-01 -1.24009266e-01
1.36920840e-01 9.80487559e-03 -1.63225666e-01 -1.29738605e+00
-9.95955765e-01 6.33693874e-01 -3.79445314e-01 5.17530024e-01
6.46030307e-01 2.13052198e-01 6.81238472e-01 4.92097944e-01
4.23564941e-01 -8.65597367e-01 2.48559527e-02 7.82830417e-01
1.26860631e+00 -9.43728328e-01 2.28040203e-01 -7.52288759e-01
-8.68623137e-01 8.00099492e-01 6.39158368e-01 5.04877158e-02
6.99746087e-02 -1.01424113e-01 -3.76344413e-01 -3.17669928e-01
-1.09111428e+00 -3.73572975e-01 6.66743755e-01 8.25259447e-01
8.05683196e-01 2.98347384e-01 -5.66930056e-01 4.90538448e-01
-6.95025504e-01 -3.49168837e-01 8.17898870e-01 4.79572862e-01
-2.43262157e-01 -1.11312628e+00 1.96647882e-01 7.43048191e-01
-3.52511585e-01 -6.48443401e-01 -9.52481329e-01 1.15240192e+00
-3.46691400e-01 1.06277573e+00 -2.67915964e-01 -2.22174823e-01
3.66821855e-01 4.43970114e-01 6.90819979e-01 -9.86672401e-01
-5.74727833e-01 -3.52287471e-01 1.06600463e+00 -6.64212927e-02
-1.80966750e-01 -2.56666541e-01 -1.28812361e+00 -3.55471313e-01
-3.27717364e-01 5.04997730e-01 4.13513452e-01 5.62559009e-01
4.98470366e-01 8.02989304e-01 -2.15391591e-01 -3.32743049e-01
-2.47496828e-01 -1.40494347e+00 -4.23794985e-01 4.17634845e-01
-4.31059420e-01 -1.74446315e-01 -3.12499970e-01 2.38467231e-02] | [10.278118133544922, 8.322210311889648] |
0322dbb0-800a-486d-bb96-ee62e99dcb8f | etop-early-termination-of-pipelines-for | 2304.08597 | null | https://arxiv.org/abs/2304.08597v1 | https://arxiv.org/pdf/2304.08597v1.pdf | eTOP: Early Termination of Pipelines for Faster Training of AutoML Systems | Recent advancements in software and hardware technologies have enabled the use of AI/ML models in everyday applications has significantly improved the quality of service rendered. However, for a given application, finding the right AI/ML model is a complex and costly process, that involves the generation, training, and evaluation of multiple interlinked steps (called pipelines), such as data pre-processing, feature engineering, selection, and model tuning. These pipelines are complex (in structure) and costly (both in compute resource and time) to execute end-to-end, with a hyper-parameter associated with each step. AutoML systems automate the search of these hyper-parameters but are slow, as they rely on optimizing the pipeline's end output. We propose the eTOP Framework which works on top of any AutoML system and decides whether or not to execute the pipeline to the end or terminate at an intermediate step. Experimental evaluation on 26 benchmark datasets and integration of eTOPwith MLBox4 reduces the training time of the AutoML system upto 40x than baseline MLBox. | ['Yash Garg', 'Juliana Freire', 'Haoxiang Zhang'] | 2023-04-17 | null | null | null | null | ['feature-engineering', 'automl'] | ['methodology', 'methodology'] | [ 5.75554930e-02 -2.75350779e-01 -2.10056314e-03 -4.26538855e-01
-9.53064024e-01 -7.41360188e-01 4.57160056e-01 4.48564351e-01
-5.24322867e-01 1.23795517e-01 -2.28155449e-01 -5.34181058e-01
-5.92582859e-04 -5.38666904e-01 -6.10988438e-01 -4.23939645e-01
-1.26868919e-01 9.97673213e-01 5.33811927e-01 2.30748981e-01
4.76285666e-01 5.17110229e-01 -1.69787955e+00 2.84086138e-01
6.60627186e-01 1.17437601e+00 1.53983086e-01 9.94302809e-01
-2.26802915e-01 5.48612356e-01 -3.74469906e-01 -2.09531173e-01
2.40868613e-01 -1.06082864e-01 -6.21636808e-01 -8.28338563e-02
2.17399642e-01 -5.89376129e-02 1.65936768e-01 7.29317844e-01
6.79783762e-01 -8.44720304e-02 2.73228496e-01 -1.52184057e+00
4.81638648e-02 5.26029885e-01 -3.44465911e-01 -1.46448135e-01
5.87489046e-02 6.67336226e-01 9.53777194e-01 -1.29962254e+00
3.20590407e-01 1.24517763e+00 5.49153268e-01 5.11101447e-02
-1.25437558e+00 -6.06588125e-01 -1.77132726e-01 2.60323565e-02
-1.65605569e+00 -8.62262547e-01 3.10994834e-01 -6.07225060e-01
1.30810440e+00 1.25124514e-01 4.93901551e-01 4.73477274e-01
1.91322207e-01 7.74568677e-01 5.60266674e-01 -4.49969202e-01
5.22063911e-01 3.85492928e-02 2.81659871e-01 8.31397891e-01
1.68353796e-01 -3.15762043e-01 -6.07188404e-01 -4.53292131e-01
3.26088816e-01 -3.27594608e-01 2.43166924e-01 -2.57271618e-01
-8.37972999e-01 6.13303602e-01 4.13901173e-02 -2.46497374e-02
-6.41029239e-01 1.56285241e-01 5.00838041e-01 4.13972139e-01
1.05594859e-01 7.16351569e-01 -6.85085773e-01 -2.77276695e-01
-1.25389719e+00 4.67003703e-01 1.04994452e+00 1.02219820e+00
1.00021017e+00 -4.89267349e-01 -2.68556744e-01 7.35370100e-01
3.19387347e-01 3.07719767e-01 5.21240354e-01 -7.20940948e-01
4.69146997e-01 1.26942515e+00 1.05007276e-01 -5.92344403e-01
-7.11429775e-01 -3.56410265e-01 -5.14649212e-01 1.82544783e-01
2.79328495e-01 -2.36920580e-01 -9.24872577e-01 9.86711442e-01
6.79280460e-01 -1.75284967e-02 -1.29308134e-01 5.97730696e-01
5.44032574e-01 6.16809726e-01 3.37565809e-01 7.67791197e-02
1.25114107e+00 -1.14522946e+00 -3.35946798e-01 -6.65742755e-01
9.56813753e-01 -1.09756875e+00 1.29349124e+00 5.83565593e-01
-1.33981586e+00 -4.93210256e-01 -1.00782382e+00 -2.70000547e-01
-3.27333212e-01 3.92239183e-01 5.59368610e-01 2.55284488e-01
-9.89046812e-01 6.85844004e-01 -1.08512843e+00 -1.71298459e-01
4.15972263e-01 7.87974179e-01 -2.18465105e-01 -4.46884595e-02
-7.56395459e-01 6.73306704e-01 5.70459604e-01 2.62080491e-01
-7.03201771e-01 -8.00986230e-01 -6.95750833e-01 1.07514545e-01
7.38304317e-01 -6.30725920e-01 1.43280470e+00 -8.25678170e-01
-1.45307779e+00 4.89544481e-01 -1.51647460e-02 -4.26875085e-01
4.07877028e-01 -3.92731220e-01 -1.18013106e-01 -2.66628504e-01
-5.59415892e-02 5.57385385e-01 6.53764784e-01 -6.41440809e-01
-9.14070070e-01 -2.38035351e-01 -1.04441211e-01 2.06074625e-01
-2.85817496e-03 3.69987190e-01 -1.24496233e+00 -5.14143929e-02
4.29345667e-02 -1.16207123e+00 -4.14990366e-01 -2.73237266e-02
-3.52019846e-01 -3.95807862e-01 6.87833130e-01 -5.91065764e-01
1.47158563e+00 -2.19732141e+00 -1.74955830e-01 3.32032472e-01
1.30739994e-02 4.87870663e-01 -1.19840153e-01 6.30703568e-01
2.18200266e-01 7.39415586e-02 -1.97080061e-01 -4.56129313e-01
2.33001783e-01 -7.88945630e-02 8.86658430e-02 3.78465891e-01
4.17007655e-01 8.01231205e-01 -5.71429908e-01 -7.99421906e-01
1.64560214e-01 4.05780435e-01 -7.72802532e-01 4.31933582e-01
-5.94920754e-01 3.46886023e-04 -5.03226161e-01 6.61984742e-01
5.31828046e-01 -4.76481885e-01 1.15236923e-01 -1.70832843e-01
-4.24219638e-01 5.03729165e-01 -1.42628622e+00 1.67562723e+00
-6.84251130e-01 4.74227190e-01 -5.93930930e-02 -4.55319494e-01
7.65057743e-01 2.07154989e-01 3.29498857e-01 -4.83922303e-01
1.71176493e-01 4.36392725e-01 1.59905285e-01 -5.15721738e-01
4.16090578e-01 5.59588075e-01 -2.05604196e-01 5.89266777e-01
2.54003480e-02 -1.25553429e-01 7.05300212e-01 -4.83092405e-02
1.33790064e+00 3.46834123e-01 3.44578594e-01 -7.24154338e-02
6.61190808e-01 2.78154671e-01 5.74711382e-01 5.98850727e-01
2.55983591e-01 3.16522092e-01 6.94896102e-01 -4.33694184e-01
-1.11315191e+00 -4.48594600e-01 4.50544991e-02 1.34413970e+00
-2.66861349e-01 -7.54316032e-01 -8.51785719e-01 -6.00151122e-01
1.02686033e-01 9.31311727e-01 -1.91425472e-01 -1.25108525e-01
-7.55307674e-01 -7.73113191e-01 3.81741881e-01 4.51382548e-01
2.30392888e-01 -1.30179608e+00 -9.96867180e-01 5.54478645e-01
2.09396645e-01 -1.14342952e+00 -5.03222048e-01 3.44445288e-01
-6.59675300e-01 -1.00191092e+00 6.94514439e-02 -4.11870092e-01
7.26920128e-01 -1.82580531e-01 1.44894493e+00 4.46209222e-01
-6.44200206e-01 -3.15281898e-01 -2.19713867e-01 -5.04842520e-01
-4.80993569e-01 4.48989391e-01 -3.94092321e-01 -8.64553973e-02
3.73583496e-01 -2.88151503e-01 -7.10515261e-01 4.54673350e-01
-8.94280434e-01 3.93227428e-01 8.85288060e-01 5.43281555e-01
7.98295796e-01 1.08673580e-01 2.61268109e-01 -9.57381248e-01
4.89677012e-01 -3.07313263e-01 -1.27890134e+00 4.71237391e-01
-7.72736847e-01 3.56343418e-01 7.45753229e-01 -1.75969899e-01
-6.25869751e-01 4.93650883e-01 -2.36967295e-01 -2.46531263e-01
4.69948985e-02 6.56444132e-01 -3.63005549e-01 9.09101218e-02
4.78430927e-01 -3.33341323e-02 -6.03672815e-03 -7.75381148e-01
2.43195564e-01 8.36660087e-01 3.02843958e-01 -4.08858597e-01
6.25928402e-01 -9.22138244e-02 -9.33135152e-02 -5.72624743e-01
-6.57497585e-01 -4.50487047e-01 -8.35835040e-01 9.45408940e-02
4.24635798e-01 -7.23674357e-01 -8.74176085e-01 4.14289117e-01
-1.03717351e+00 -5.26163340e-01 -8.04390386e-02 6.06771931e-02
-3.20241362e-01 -1.98408306e-01 -1.95337906e-01 -5.15592396e-01
-8.56949806e-01 -1.51682127e+00 1.26408112e+00 1.80608660e-01
-3.40855271e-01 -6.07297719e-01 -1.75967708e-01 2.52313465e-01
5.36971092e-01 -9.85994712e-02 1.00720119e+00 -7.78898776e-01
-6.79832160e-01 -6.96267545e-01 -2.21056402e-01 1.11448027e-01
-4.40180063e-01 5.45682430e-01 -9.37713623e-01 -2.42641106e-01
-4.09075290e-01 -2.93616563e-01 3.62770468e-01 1.87306687e-01
1.36536849e+00 -2.36615583e-01 -3.87011975e-01 7.05742598e-01
1.41350925e+00 1.78218916e-01 3.14364523e-01 2.30628967e-01
4.36632156e-01 4.63483572e-01 9.12417889e-01 4.86002415e-01
3.25575143e-01 7.26816654e-01 3.19178700e-01 -1.59282014e-01
9.68613010e-03 -9.67480689e-02 3.87854069e-01 5.73822975e-01
3.99270713e-01 -3.21482152e-01 -1.32193863e+00 4.10951018e-01
-1.94472051e+00 -3.64876062e-01 -5.77043071e-02 2.55742288e+00
9.86315727e-01 4.04657304e-01 7.51537308e-02 8.42341110e-02
1.77850723e-01 -3.63627791e-01 -8.06565344e-01 -3.25282484e-01
3.90145600e-01 2.67853022e-01 6.68346584e-01 5.32237828e-01
-1.01945019e+00 1.37627673e+00 5.56677485e+00 7.35618412e-01
-1.22177291e+00 -2.91665643e-01 5.16758263e-01 -3.10685694e-01
1.84823498e-01 2.53672898e-01 -1.13609552e+00 2.67322898e-01
1.35520732e+00 -3.26930374e-01 5.93045831e-01 9.32088971e-01
2.89954931e-01 -4.46739137e-01 -1.44857264e+00 7.55299091e-01
-2.74415463e-01 -1.28372836e+00 -1.63839340e-01 -2.66024321e-01
1.23732440e-01 3.08867216e-01 -3.57125044e-01 3.44993800e-01
4.86535877e-01 -9.18914080e-01 6.74974084e-01 3.76871198e-01
5.49968243e-01 -1.03096068e+00 5.78570247e-01 4.65660274e-01
-1.13581443e+00 -1.55205224e-02 -2.67139703e-01 4.88589168e-01
2.02997886e-02 5.64574838e-01 -1.48216081e+00 2.71189004e-01
7.07230985e-01 1.41556635e-01 -9.15806532e-01 1.20360804e+00
-3.31940919e-01 5.99566102e-01 -3.71179610e-01 -2.34235823e-02
1.58563569e-01 5.61330542e-02 1.46761239e-01 1.30584693e+00
1.71705678e-01 -2.40622267e-01 4.76411641e-01 6.61371291e-01
4.94131260e-02 2.42352396e-01 -9.50715318e-03 -2.36296251e-01
6.55527830e-01 1.51289356e+00 -6.28332019e-01 -4.54940706e-01
-2.59979904e-01 6.97714925e-01 2.71191776e-01 -2.35978216e-02
-7.88613319e-01 -5.44744670e-01 5.61934531e-01 2.55649626e-01
1.23931229e-01 -1.70524701e-01 -2.89808005e-01 -5.32359600e-01
1.30907640e-01 -1.22439885e+00 2.67456621e-01 -5.88567257e-01
-7.48381317e-01 8.02077115e-01 -2.35559955e-01 -8.44638467e-01
-4.90648836e-01 -4.34576750e-01 -3.89921784e-01 1.11713648e+00
-1.28410447e+00 -9.26171482e-01 -4.58341211e-01 3.37475121e-01
6.14226580e-01 9.92661268e-02 8.38273048e-01 3.76060277e-01
-9.72430408e-01 4.59493279e-01 -8.51818621e-02 -1.64763361e-01
7.26649880e-01 -1.19149601e+00 9.20452476e-01 7.40190983e-01
-2.15898171e-01 5.94988346e-01 7.46025264e-01 -6.09208465e-01
-1.94011545e+00 -1.14724040e+00 1.22363102e+00 -2.01046139e-01
6.99493408e-01 -6.58267021e-01 -8.77681851e-01 4.88355815e-01
8.51191282e-02 6.81856126e-02 5.86477816e-01 1.19804159e-01
-3.24445963e-02 -2.18147740e-01 -9.30658400e-01 6.32980287e-01
5.85219920e-01 -1.44496143e-01 -5.91690429e-02 5.35890043e-01
5.66351235e-01 -4.44632709e-01 -9.61456656e-01 2.31357396e-01
5.81949532e-01 -6.93472564e-01 7.56670535e-01 -4.00999933e-01
1.44378692e-01 -5.58131456e-01 1.38404608e-01 -9.13898587e-01
-8.70942920e-02 -9.42041159e-01 1.04746297e-02 1.16194129e+00
8.18181813e-01 -3.80208582e-01 4.80256528e-01 9.52749193e-01
-1.88399449e-01 -9.62652564e-01 -2.97800839e-01 -3.48716438e-01
-5.54299414e-01 -5.87695539e-01 9.26874757e-01 2.97110736e-01
-3.40945572e-01 6.30689144e-01 9.67609137e-02 3.88948619e-01
5.22352099e-01 3.22835356e-01 1.27224672e+00 -1.24399626e+00
-4.04255718e-01 -4.56400573e-01 3.94399336e-04 -9.04244065e-01
-2.72726744e-01 -7.11463988e-01 3.62868488e-01 -1.48328221e+00
-2.10497215e-01 -9.26760435e-01 -1.15092054e-01 9.19707716e-01
-3.01897079e-01 -1.37063190e-01 1.32528558e-01 3.09758693e-01
-7.73176968e-01 1.01756141e-01 7.83792973e-01 1.75817877e-01
-4.83774245e-01 2.11162433e-01 -3.96211326e-01 8.22426617e-01
4.51743096e-01 -6.01968646e-01 -2.88646638e-01 -4.64272469e-01
5.39155722e-01 9.93962511e-02 9.28481445e-02 -1.16392016e+00
6.20039344e-01 -7.10330158e-02 2.71346360e-01 -9.73365784e-01
3.50634515e-01 -9.15859282e-01 2.30403304e-01 3.86134267e-01
-3.09834599e-01 4.64229643e-01 2.96684831e-01 -3.14985402e-02
-4.29684892e-02 -4.79248255e-01 7.20951140e-01 8.46134275e-02
-6.09291375e-01 4.43629235e-01 -1.16620131e-01 3.04319244e-02
1.14753819e+00 1.23088628e-01 -2.89718732e-02 2.15476856e-01
-5.80479145e-01 6.54462636e-01 4.47575331e-01 3.32067937e-01
1.01617053e-01 -6.46090984e-01 -6.53085470e-01 4.78477031e-01
1.49411216e-01 4.50589746e-01 -2.08174914e-01 8.91001821e-01
-9.20434713e-01 3.66124332e-01 1.64964274e-01 -5.88948071e-01
-1.45058811e+00 3.64677846e-01 2.41748467e-01 -5.90494215e-01
-4.33682889e-01 8.22402358e-01 -2.36808985e-01 -2.82118857e-01
4.36404467e-01 -1.50569737e-01 -7.21802860e-02 6.94207028e-02
6.10088408e-01 4.25479919e-01 5.82181454e-01 -2.33841076e-01
-5.55490017e-01 2.58213401e-01 -2.78739423e-01 4.95095551e-02
1.59450960e+00 2.60205925e-01 -3.17631960e-01 3.29672515e-01
9.40908253e-01 -3.38538140e-01 -1.25510311e+00 -1.83067530e-01
5.81561029e-01 -1.61143214e-01 3.20283860e-01 -9.80255604e-01
-7.68689632e-01 8.08450758e-01 4.07095969e-01 2.00239196e-02
1.14807737e+00 -1.87429145e-01 6.90945983e-01 4.55459118e-01
2.37652063e-01 -1.29864609e+00 -3.05882543e-01 4.22232360e-01
8.57192039e-01 -9.93377268e-01 1.95094526e-01 -3.29934627e-01
-5.35543919e-01 1.06343663e+00 7.01498926e-01 -2.26857625e-02
6.22973502e-01 6.45608544e-01 -1.35356011e-02 -2.11793095e-01
-1.47406876e+00 -1.39519516e-02 4.33378458e-01 -1.81644797e-01
6.01723552e-01 -6.99309930e-02 -3.30147177e-01 3.93393576e-01
-1.32181466e-01 2.55151302e-01 2.37654857e-02 1.20233440e+00
-4.52436179e-01 -1.40861702e+00 -1.95354953e-01 4.67826754e-01
-3.56128097e-01 -2.11645871e-01 -3.70598435e-01 7.71156847e-01
1.37294486e-01 9.03308034e-01 1.28247336e-01 -3.38193059e-01
6.34067714e-01 2.83610821e-01 1.36792203e-02 -7.34018922e-01
-1.08405519e+00 2.62937784e-01 2.39002779e-01 -7.91917562e-01
3.64773422e-01 -7.63151407e-01 -1.47347069e+00 -1.37546524e-01
-3.58076006e-01 1.06144696e-01 1.22706723e+00 1.00108278e+00
8.27194929e-01 4.74737674e-01 5.98360538e-01 -8.40082467e-01
-6.47808254e-01 -9.35444772e-01 9.16302279e-02 -2.38189436e-02
-2.18518265e-02 -2.76522040e-01 -7.40878657e-02 1.11392766e-01] | [8.526512145996094, 4.105472564697266] |
3d15012d-9748-4660-8016-86787271f62f | user-guided-domain-adaptation-for-rapid | 2009.02455 | null | https://arxiv.org/abs/2009.02455v1 | https://arxiv.org/pdf/2009.02455v1.pdf | User-Guided Domain Adaptation for Rapid Annotation from User Interactions: A Study on Pathological Liver Segmentation | Mask-based annotation of medical images, especially for 3D data, is a bottleneck in developing reliable machine learning models. Using minimal-labor user interactions (UIs) to guide the annotation is promising, but challenges remain on best harmonizing the mask prediction with the UIs. To address this, we propose the user-guided domain adaptation (UGDA) framework, which uses prediction-based adversarial domain adaptation (PADA) to model the combined distribution of UIs and mask predictions. The UIs are then used as anchors to guide and align the mask prediction. Importantly, UGDA can both learn from unlabelled data and also model the high-level semantic meaning behind different UIs. We test UGDA on annotating pathological livers using a clinically comprehensive dataset of 927 patient studies. Using only extreme-point UIs, we achieve a mean (worst-case) performance of 96.1%(94.9%), compared to 93.0% (87.0%) for deep extreme points (DEXTR). Furthermore, we also show UGDA can retain this state-of-the-art performance even when only seeing a fraction of available UIs, demonstrating an ability for robust and reliable UI-guided segmentation with extremely minimal labor demands. | ['Chien-Hung Liao', 'Jing Xiao', 'Chi Tung Cheng', 'Le Lu', 'Jinzheng Cai', 'Junzhou Huang', 'Zhanghexuan Ji', 'Ashwin Raju', 'Adam P. Harrison'] | 2020-09-05 | null | null | null | null | ['liver-segmentation'] | ['medical'] | [ 3.09662521e-01 5.75421035e-01 -5.42513616e-02 -3.06272179e-01
-1.19102156e+00 -7.91722000e-01 3.25496405e-01 9.43974480e-02
-1.23002678e-01 4.66607064e-01 2.79417057e-02 -2.78296620e-01
2.94080168e-01 -2.56638080e-01 -7.84439564e-01 -8.01695108e-01
-8.45231041e-02 6.03462577e-01 2.58095145e-01 1.67937055e-01
-2.48108268e-01 2.56739050e-01 -5.09065270e-01 6.01804554e-01
9.74833727e-01 1.05889344e+00 2.56681412e-01 7.72274435e-01
1.15761228e-01 3.24477255e-01 -7.78952897e-01 -4.08109546e-01
5.92581868e-01 -5.12505412e-01 -7.56648600e-01 6.07712679e-02
2.69074708e-01 -3.49051535e-01 -5.73958233e-02 8.29799473e-01
7.99884677e-01 -2.60997027e-01 9.84721482e-01 -1.01317823e+00
-5.96854687e-01 4.05504376e-01 -5.76101720e-01 5.70319779e-02
1.90986842e-01 2.95632094e-01 5.24978518e-01 -5.25148273e-01
9.27127779e-01 5.03033102e-01 1.03140020e+00 9.41756427e-01
-1.56169951e+00 -5.48645556e-01 -1.03488654e-01 -3.59125704e-01
-1.51034677e+00 -2.88752526e-01 6.41554713e-01 -7.36300290e-01
6.32275760e-01 3.24145287e-01 6.51577294e-01 1.08411181e+00
5.00750780e-01 7.91568339e-01 1.00456512e+00 -4.18710530e-01
1.57855555e-01 1.85148239e-01 -3.22458565e-01 7.27419198e-01
-2.78610038e-03 7.38531351e-02 -1.69922605e-01 -3.18412185e-01
8.24833810e-01 -2.80761182e-01 -4.60641831e-01 -5.72824299e-01
-1.37757325e+00 5.99296451e-01 5.35129249e-01 3.09308022e-02
-3.23621958e-01 -1.05529577e-01 3.93485248e-01 -7.73578957e-02
5.47319531e-01 5.51494539e-01 -6.91391230e-01 3.05436134e-01
-1.01707220e+00 -8.84200335e-02 8.70065749e-01 1.04251754e+00
2.96227217e-01 -9.13646892e-02 -3.53405416e-01 6.83983624e-01
2.58821458e-01 2.07755417e-01 4.90988851e-01 -9.93658006e-01
7.12314295e-03 3.37471247e-01 3.40346023e-02 -7.53035605e-01
-5.84167600e-01 -6.69939756e-01 -1.01295853e+00 5.51132083e-01
7.76693046e-01 -2.62665570e-01 -1.35505438e+00 1.71127057e+00
4.55678433e-01 2.82670766e-01 3.06198895e-02 8.20125341e-01
9.79881346e-01 2.77467132e-01 3.32247525e-01 -1.38845518e-02
1.31883919e+00 -7.48956561e-01 -3.95233005e-01 -2.44355962e-01
7.89906681e-01 -6.73788011e-01 1.10544205e+00 3.06904018e-01
-9.38924134e-01 -3.17862570e-01 -1.03499854e+00 1.11444190e-01
-1.42147318e-01 9.16000530e-02 1.28169253e-01 6.24725640e-01
-1.13031471e+00 5.38621187e-01 -1.21650863e+00 -1.22218348e-01
8.72555971e-01 6.55973911e-01 -4.40773964e-01 1.19236760e-01
-9.24086750e-01 9.04501379e-01 1.55835807e-01 -2.76610911e-01
-9.95185018e-01 -1.31012928e+00 -9.33866560e-01 -2.94845104e-01
2.41027504e-01 -8.96844625e-01 1.26321554e+00 -9.10844862e-01
-1.27781415e+00 1.33982992e+00 8.19602683e-02 -7.22355068e-01
9.20024216e-01 1.31110787e-01 -1.14606619e-01 2.16671273e-01
4.32059951e-02 1.08622313e+00 5.90903878e-01 -1.58771408e+00
-1.57538846e-01 -1.57538518e-01 -3.86839688e-01 1.74507454e-01
8.41159374e-02 -1.80450052e-01 -4.77718711e-01 -8.67094755e-01
1.87698323e-02 -9.94334877e-01 -5.07414460e-01 4.77881074e-01
-5.87729812e-01 2.32887164e-01 7.65499055e-01 -1.14565766e+00
7.73339629e-01 -2.10805345e+00 4.20686183e-03 2.10567713e-01
3.66271973e-01 2.85915703e-01 3.87561470e-02 -3.37031364e-01
-5.88915646e-02 2.38042951e-01 -6.76051855e-01 -5.34866333e-01
-1.06776252e-01 3.64657104e-01 -1.02329273e-02 4.49439824e-01
1.67528003e-01 1.18068993e+00 -8.69601071e-01 -7.11846471e-01
4.02106524e-01 4.81826186e-01 -7.54616559e-01 4.45966661e-01
-2.55369574e-01 1.11719358e+00 -1.56030983e-01 7.55107343e-01
4.82445419e-01 -3.35777938e-01 3.56263191e-01 -5.08337617e-01
4.51217175e-01 -2.67949790e-01 -7.07025588e-01 2.11124134e+00
-3.06710988e-01 3.69771570e-01 1.33054212e-01 -8.26745152e-01
7.95479894e-01 3.90329272e-01 8.74833584e-01 -4.86495674e-01
4.99350876e-02 2.38780975e-01 7.01399744e-02 -3.79335791e-01
-1.89520910e-01 -3.53935122e-01 -2.85828590e-01 1.21098593e-01
1.92685530e-01 -4.15162235e-01 -3.55887115e-01 1.89812128e-02
1.06311262e+00 3.17814946e-01 5.63679039e-01 -4.29952592e-01
2.62138605e-01 2.53389746e-01 7.13887274e-01 6.43520892e-01
-6.70925915e-01 1.27648568e+00 6.19036257e-01 -4.58864629e-01
-1.02135086e+00 -1.15706789e+00 -3.31770152e-01 6.05410039e-01
2.21236125e-01 -1.12944663e-01 -1.04747868e+00 -1.40420091e+00
7.83351157e-03 7.62003005e-01 -9.36784804e-01 -1.96908846e-01
-4.56935644e-01 -8.46821487e-01 6.84631467e-01 5.43785870e-01
1.81202576e-01 -7.42889464e-01 -5.34287095e-01 1.82464167e-01
-1.33860424e-01 -1.16646290e+00 -6.78072274e-01 3.19455892e-01
-5.76903820e-01 -1.06895494e+00 -1.02333581e+00 -7.46208906e-01
9.75555420e-01 -4.13741916e-01 1.48618281e+00 -6.09360859e-02
-3.79884839e-01 3.09375882e-01 -2.10892722e-01 -5.63690662e-01
-8.76477182e-01 6.19192868e-02 -1.19527288e-01 -4.04967189e-01
-8.68307948e-02 -3.11315864e-01 -1.03391612e+00 5.84026754e-01
-7.84947217e-01 3.12616736e-01 6.58632815e-01 1.01766729e+00
1.10677516e+00 -4.96163517e-01 5.84479094e-01 -1.15213692e+00
-1.75280159e-03 -5.29529393e-01 -3.61563414e-01 3.40474010e-01
-5.20163715e-01 -2.81576872e-01 5.68713427e-01 -5.12160540e-01
-8.20937991e-01 3.97129297e-01 -3.25600803e-01 -6.39862239e-01
-4.81214374e-01 2.04940155e-01 -1.08007789e-01 -3.80223036e-01
9.70301509e-01 -4.03239876e-02 3.06126803e-01 -2.54960984e-01
2.47351483e-01 5.47206461e-01 8.67254496e-01 -4.87099856e-01
4.89872158e-01 4.74173993e-01 -8.39335695e-02 -2.40825862e-01
-7.86355615e-01 -2.78813154e-01 -9.43311691e-01 -9.41048861e-02
1.10070050e+00 -7.71646261e-01 -1.67493060e-01 2.93514520e-01
-9.08224940e-01 -9.21613395e-01 -5.38378179e-01 1.45967603e-01
-8.24237108e-01 3.44085187e-01 -6.62882209e-01 -2.52199799e-01
-6.48431480e-01 -1.51895618e+00 1.18299878e+00 5.65345120e-03
-6.87527299e-01 -1.23487663e+00 -1.77074492e-01 4.36846286e-01
2.43750557e-01 8.55692685e-01 8.69826257e-01 -1.20270288e+00
-2.84331381e-01 -2.63531953e-01 -8.04868862e-02 4.71455395e-01
2.56947219e-01 -3.83663148e-01 -9.15781021e-01 -1.39174193e-01
-5.42520508e-02 -1.00187309e-01 3.78952235e-01 7.63252735e-01
1.51412201e+00 -1.67742833e-01 -5.63378394e-01 8.92703235e-01
1.22193849e+00 2.12769642e-01 5.08199811e-01 1.63726166e-01
5.71867406e-01 1.56739369e-01 5.96240997e-01 2.87544072e-01
1.94739997e-01 7.51109660e-01 5.19889772e-01 -6.36027753e-01
-5.15773714e-01 -2.08374500e-01 -2.82986369e-02 2.98508972e-01
2.74813741e-01 -2.28132099e-01 -1.34272492e+00 4.49830770e-01
-1.54616582e+00 -2.94455558e-01 1.12384357e-01 2.03180313e+00
1.11813736e+00 7.38737732e-02 4.17759083e-02 -3.31454664e-01
5.42289853e-01 -1.78078488e-01 -6.88147187e-01 -1.06343679e-01
1.84043676e-01 1.27072841e-01 6.18333101e-01 3.75874788e-01
-1.39743769e+00 6.93289042e-01 6.74701309e+00 8.19482863e-01
-9.86060381e-01 2.62020558e-01 1.24292183e+00 7.55770579e-02
-4.99530248e-02 -5.02800465e-01 -2.48845562e-01 5.91604292e-01
7.63111055e-01 1.01488680e-01 2.02408269e-01 8.32466841e-01
5.05445711e-02 -8.89794007e-02 -1.15980053e+00 7.26793230e-01
1.56374887e-01 -1.47967160e+00 -1.58987567e-01 -2.13012546e-02
1.07106256e+00 -5.96488230e-02 1.92271709e-01 5.54115996e-02
3.63043368e-01 -1.20880353e+00 5.88085711e-01 3.15932870e-01
1.39970589e+00 -4.84607607e-01 1.08861721e+00 2.91385978e-01
-7.40737319e-01 3.38825941e-01 -1.32683143e-01 6.63598597e-01
1.08198859e-01 5.68987787e-01 -1.46762657e+00 6.15696490e-01
4.82869238e-01 3.73431563e-01 -2.65432328e-01 1.04372716e+00
-9.90822837e-02 5.62954128e-01 -3.97872388e-01 6.30427718e-01
-4.84050252e-03 8.35811198e-02 4.98207867e-01 1.42668533e+00
2.82481253e-01 -1.10369623e-02 3.43658209e-01 7.89831161e-01
-2.29490399e-01 3.35924365e-02 -4.46224451e-01 2.86934704e-01
2.70589501e-01 1.24408972e+00 -9.37443912e-01 -2.44345292e-01
-2.54638433e-01 1.31885064e+00 -9.09950286e-02 1.45522609e-01
-1.01947331e+00 1.01725899e-01 6.34646475e-01 2.50237405e-01
3.14551219e-02 -4.09371108e-02 -8.50891471e-01 -7.15427935e-01
-3.39515358e-01 -8.52194071e-01 7.87306607e-01 -8.57906342e-01
-1.35558426e+00 6.33627534e-01 -1.81071863e-01 -1.45766532e+00
-2.12200522e-01 -6.64622664e-01 -5.28578401e-01 9.31629181e-01
-1.39124453e+00 -1.52811658e+00 -4.10184741e-01 3.04706395e-01
4.63092208e-01 7.40302652e-02 1.22328794e+00 1.20947428e-01
-3.58530819e-01 9.75798905e-01 -3.13252769e-02 4.08763885e-01
9.85436201e-01 -1.55822790e+00 2.00275853e-01 5.74947417e-01
1.64711978e-02 2.01178238e-01 6.12188876e-01 -8.05139601e-01
-9.96518195e-01 -1.39263153e+00 2.22803488e-01 -8.44087481e-01
2.81050861e-01 -1.67811558e-01 -8.41474771e-01 7.90145338e-01
9.51611474e-02 6.70732141e-01 1.16184866e+00 -4.04018223e-01
-4.63136025e-02 1.18502736e-01 -1.83003724e+00 5.71354866e-01
9.09369588e-01 -1.69679061e-01 -4.24980640e-01 3.11932862e-01
6.83719099e-01 -1.26397467e+00 -1.27791584e+00 6.87934279e-01
5.09012222e-01 -7.13914514e-01 1.22650278e+00 -4.94927108e-01
3.97889704e-01 -3.63853037e-01 -1.26110718e-01 -1.44206572e+00
-2.15297669e-01 -5.39617479e-01 -1.69897079e-01 1.02013612e+00
5.15988648e-01 -3.69157225e-01 1.12129962e+00 1.08548188e+00
-5.37837088e-01 -1.03141689e+00 -9.65258420e-01 -5.22446215e-01
3.65577519e-01 -4.21982288e-01 4.39956248e-01 1.04269361e+00
-1.46865353e-01 -3.90411377e-01 -1.78404793e-01 4.93111879e-01
6.12408459e-01 -6.59955218e-02 5.42916298e-01 -8.82032692e-01
-2.66449571e-01 -2.46690989e-01 -5.05763590e-01 -5.95446169e-01
6.92183971e-02 -1.27860165e+00 1.14137016e-01 -1.51683676e+00
2.00639024e-01 -7.05613017e-01 -4.32881355e-01 6.43803716e-01
-2.62926012e-01 6.81516349e-01 1.03312418e-01 2.17614695e-01
-4.49113458e-01 4.24429663e-02 1.45619035e+00 -9.23687741e-02
-2.41894320e-01 -4.17267680e-02 -6.70917928e-01 9.30731177e-01
9.45302606e-01 -4.07884359e-01 -1.46970272e-01 -3.20890754e-01
-4.22710538e-01 7.70904347e-02 5.84964216e-01 -9.54079807e-01
1.24519594e-01 1.27730239e-02 9.84634638e-01 -2.65704036e-01
4.55911532e-02 -1.00473845e+00 3.98083299e-01 6.34934664e-01
-3.56866151e-01 -4.61146355e-01 4.19588268e-01 4.95928019e-01
1.61356255e-01 -7.12272003e-02 1.06647778e+00 -3.03316951e-01
-5.26148081e-01 4.86479551e-01 -9.43176150e-02 2.84332752e-01
1.33979142e+00 -1.99633837e-01 2.62539499e-02 -9.40753967e-02
-1.23294425e+00 2.22567543e-01 6.58809185e-01 -9.67478529e-02
4.84018594e-01 -1.08079624e+00 -6.79513514e-01 3.80448163e-01
5.03200516e-02 6.83134913e-01 3.69704604e-01 8.57319057e-01
-5.88801444e-01 -3.11933085e-02 -3.23170610e-02 -9.29638326e-01
-1.12141442e+00 3.24714273e-01 7.37592041e-01 -5.31667709e-01
-8.32703114e-01 9.81303036e-01 4.98554051e-01 -5.74078262e-01
8.46767947e-02 -2.19758540e-01 2.94170678e-01 -1.98059037e-01
3.14970881e-01 -8.42202082e-02 1.50778428e-01 -2.56817758e-01
-3.72022659e-01 3.40639323e-01 -1.30975042e-02 1.40873864e-01
1.08844137e+00 -1.89375617e-02 3.23378742e-01 7.83433616e-02
9.45654869e-01 -5.58767132e-02 -1.69043112e+00 -3.14115845e-02
-1.23440668e-01 -3.49854797e-01 -6.57269433e-02 -1.51338792e+00
-1.12118316e+00 7.46224523e-01 7.14071870e-01 -7.89469555e-02
1.10424447e+00 3.01271886e-01 9.74979639e-01 -3.80788386e-01
3.36239129e-01 -6.45137966e-01 -6.00522049e-02 4.37242649e-02
8.84048462e-01 -1.28076506e+00 -2.42128193e-01 -6.53843343e-01
-9.98847008e-01 7.91361868e-01 7.15829194e-01 7.75935948e-02
6.05633676e-01 7.81438231e-01 5.34665048e-01 1.32669926e-01
-3.32152724e-01 1.30304143e-01 5.15826225e-01 1.19162774e+00
3.15737337e-01 2.28500888e-01 1.05928116e-01 1.13080347e+00
-1.09153334e-02 -6.02270141e-02 1.66341394e-01 6.17940545e-01
2.81881299e-02 -1.02798855e+00 -2.70515144e-01 6.05134487e-01
-7.93663621e-01 -4.77971062e-02 -7.99937323e-02 6.57306731e-01
2.70454019e-01 4.46406901e-01 -6.10403866e-02 -3.17977697e-01
2.43105486e-01 2.78458625e-01 4.24364835e-01 -6.91677153e-01
-6.16241455e-01 8.76681656e-02 -6.43673688e-02 -5.04656553e-01
-8.83445144e-03 -6.15275443e-01 -1.39607024e+00 5.11871651e-02
-1.94578320e-02 -2.19365209e-01 4.75525290e-01 7.10758924e-01
5.22596300e-01 6.44007683e-01 3.33987176e-01 -9.36906040e-01
-5.38416684e-01 -6.21474147e-01 -2.87151337e-01 7.23953903e-01
2.24129140e-01 -4.69396085e-01 -1.66379139e-01 4.49035823e-01] | [14.537158966064453, -2.0712192058563232] |
72deac14-1e05-4995-b6d2-2a337172b1dc | block-matching-in-fpga | 2006.14105 | null | https://arxiv.org/abs/2006.14105v1 | https://arxiv.org/pdf/2006.14105v1.pdf | Block-matching in FPGA | Block-matching and 3D filtering (BM3D) is an image denoising algorithm that works in two similar steps. Both of these steps need to perform grouping by block-matching. We implement the block-matching in an FPGA, leveraging its ability to perform parallel computations. Our goal is to enable other researchers to use our solution in the future for real-time video denoising in video cameras that use FPGAs (such as the AXIOM Beta). | ['Michal Pleskowicz', 'Rafael Pizarro Solar'] | 2020-06-24 | null | null | null | null | ['video-denoising'] | ['computer-vision'] | [ 2.10756630e-01 -4.86902714e-01 5.62600076e-01 -3.48499656e-01
-1.69647068e-01 -2.74163127e-01 2.27059051e-01 -3.85161400e-01
-1.96920380e-01 -1.41600996e-01 2.55816668e-01 -7.41777182e-01
1.95931643e-01 -9.24891472e-01 -5.93833923e-01 -3.66743326e-01
-2.51215488e-01 -5.13072431e-01 6.56786025e-01 -3.26618493e-01
5.72683334e-01 6.62991822e-01 -1.52603936e+00 7.30008304e-01
3.54938477e-01 1.14163649e+00 2.73029983e-01 1.08278561e+00
1.01868257e-01 6.68207467e-01 -4.58669692e-01 -3.00605565e-01
1.24036789e+00 -4.44049299e-01 -3.00116003e-01 1.10435627e-01
9.59660113e-01 -7.66081452e-01 -4.97703671e-01 1.13398504e+00
6.40643954e-01 3.58363055e-03 1.63118020e-01 -7.72801876e-01
1.31748810e-01 -4.35581319e-02 -6.59628153e-01 2.11555257e-01
6.43409908e-01 1.20540164e-01 5.19790910e-02 -6.76949620e-01
4.77030486e-01 1.53537345e+00 1.33387196e+00 4.14626777e-01
-1.03344190e+00 -8.17193151e-01 -5.48726976e-01 2.84476310e-01
-1.24850214e+00 -3.81985784e-01 6.50813878e-01 -3.52970749e-01
1.20854259e+00 3.46054256e-01 9.50583696e-01 3.10357571e-01
7.67010033e-01 2.42794901e-01 1.22898436e+00 -6.86237395e-01
1.63650677e-01 -5.77099323e-01 2.72473156e-01 7.13830292e-01
1.35161310e-01 4.51853037e-01 -7.34909713e-01 -1.77011061e-02
1.20709455e+00 3.14883254e-02 -1.52424023e-01 6.44566193e-02
-1.07582104e+00 6.71802998e-01 1.28681496e-01 1.44149318e-01
-4.14037555e-01 6.02264166e-01 6.86524451e-01 8.85032475e-01
4.57768798e-01 -1.76873766e-02 -3.52086306e-01 -1.36929862e-02
-1.40885973e+00 1.81775942e-01 6.26860023e-01 1.16758490e+00
7.91117370e-01 -8.00847337e-02 1.26427859e-01 3.02527279e-01
5.46328664e-01 2.25669771e-01 3.91458541e-01 -1.41584396e+00
1.10464454e-01 3.88028383e-01 -1.99469417e-01 -9.75391507e-01
-2.20075265e-01 1.37307942e-01 -9.96298730e-01 1.09640753e+00
9.25264359e-02 4.52279337e-02 -1.16694629e+00 7.79179394e-01
2.55227208e-01 6.86354995e-01 -2.36834943e-01 7.91330636e-01
9.70494509e-01 6.15066528e-01 -2.52690792e-01 5.26547395e-02
1.42022634e+00 -8.39957774e-01 -7.47047842e-01 2.81556427e-01
3.77129257e-01 -1.41247034e+00 4.52063203e-01 8.97455692e-01
-1.34508717e+00 -1.20219338e+00 -1.56799018e+00 -2.47278318e-01
-1.22367255e-01 -2.89363623e-01 6.57614291e-01 1.29370570e+00
-1.44439447e+00 1.00288367e+00 -9.50708389e-01 -4.43133980e-01
3.08417588e-01 7.50861406e-01 -5.32638073e-01 -9.28318799e-02
-7.07385004e-01 8.99968147e-01 9.36867669e-02 2.02212289e-01
-7.39777923e-01 -8.16426098e-01 -6.69256270e-01 -1.57137126e-01
-3.01474363e-01 -1.03911781e+00 9.01222646e-01 -9.58458662e-01
-1.52345848e+00 1.18304658e+00 -2.64490396e-01 -7.09812462e-01
4.10647899e-01 -2.35167295e-01 -3.53856862e-01 1.91254109e-01
-2.71817118e-01 6.57594323e-01 1.36566091e+00 -7.01046109e-01
-7.36803591e-01 -4.86300111e-01 1.35817632e-01 -3.17878217e-01
-8.06850791e-02 1.84688359e-01 -5.62280536e-01 -9.83187675e-01
5.79074025e-01 -4.73431170e-01 -3.13852578e-01 3.27963561e-01
1.78966105e-01 1.80176452e-01 1.16975737e+00 -9.42286074e-01
1.17692101e+00 -2.33142018e+00 -1.23015054e-01 3.70577663e-01
1.67335793e-01 5.68123937e-01 9.33514535e-02 6.22019589e-01
-1.97491929e-01 4.22606943e-03 2.78622121e-01 -2.50380129e-01
-3.22992623e-01 -3.26472931e-02 -7.18754381e-02 6.96982503e-01
-5.97807840e-02 1.46466628e-01 -3.44691217e-01 -4.25114661e-01
7.15424061e-01 4.68853623e-01 -7.72736609e-01 4.83198576e-02
4.94387299e-01 1.14087738e-01 -2.64934320e-02 7.07357764e-01
1.48062742e+00 7.75776684e-01 -1.39425054e-01 -9.59847748e-01
-5.66267312e-01 -9.85520855e-02 -1.71735048e+00 1.84221792e+00
-3.93576950e-01 9.50065613e-01 7.33138680e-01 -8.44522715e-01
8.90452087e-01 4.94527847e-01 3.95177811e-01 -4.98765200e-01
2.64187366e-01 2.03147292e-01 -1.97192818e-01 -6.16798341e-01
2.71362424e-01 8.08146223e-03 5.05054593e-01 1.62534446e-01
1.40239432e-01 -5.36820054e-01 3.13279301e-01 5.79923578e-03
1.53851807e+00 2.04426020e-01 1.82088181e-01 -7.55863130e-01
6.51575863e-01 1.46320611e-01 4.95476663e-01 8.88168991e-01
-2.99652129e-01 7.23269045e-01 3.48185673e-02 -7.16541171e-01
-1.17578983e+00 -7.99211562e-01 -3.48756686e-02 2.74924040e-01
2.72277921e-01 -7.28205860e-01 -7.57150948e-01 -2.67983794e-01
5.97375445e-02 2.41750907e-02 -3.22633922e-01 -2.60309246e-03
-6.65488482e-01 -2.46538833e-01 2.77008653e-01 3.26772898e-01
8.97807777e-01 -4.39943045e-01 -9.26447332e-01 4.50339437e-01
6.83602393e-01 -7.07874656e-01 -4.36933249e-01 3.02152872e-01
-1.34627938e+00 -1.17844725e+00 -4.30156469e-01 -9.96298671e-01
5.67568243e-01 7.39425242e-01 1.12800503e+00 5.88896036e-01
-5.02565920e-01 4.22966301e-01 -5.44515789e-01 -4.01534259e-01
-4.76896226e-01 -6.74355924e-01 -2.01319456e-01 -4.96825725e-01
5.79621375e-01 -8.49354625e-01 -8.33055556e-01 5.64401388e-01
-1.06823635e+00 1.10040560e-01 1.93055600e-01 6.42617941e-01
3.55688065e-01 5.68423927e-01 -2.18540773e-01 -5.28595805e-01
4.49729919e-01 2.14145079e-01 -9.42154408e-01 -1.79637536e-01
-3.83888483e-02 -5.30409932e-01 2.86404490e-01 -1.47121489e-01
-7.38676608e-01 5.02429605e-01 -6.35336995e-01 -3.19697440e-01
-2.45766900e-02 -8.20058025e-03 -1.25311077e-01 -8.45702231e-01
4.30645108e-01 -3.38333040e-01 3.32666993e-01 -6.14168704e-01
1.34178717e-02 7.60573030e-01 5.80253959e-01 5.91253564e-02
8.98042262e-01 7.88516223e-01 4.15439576e-01 -9.24157441e-01
-8.66437405e-02 -7.45423019e-01 -5.53065240e-01 -5.27934134e-01
8.29292238e-01 -1.26165891e+00 -9.02423382e-01 8.71983945e-01
-1.26445103e+00 -1.45816818e-01 4.18681949e-02 5.07366240e-01
-4.01558757e-01 5.85531414e-01 -6.92940474e-01 -3.38375419e-01
-4.89552826e-01 -1.18332195e+00 8.52368832e-01 2.83203870e-01
-1.62751123e-01 -6.07152879e-01 -7.92533085e-02 2.81830728e-01
3.98169160e-01 1.79459885e-01 5.31357169e-01 1.42578259e-01
-5.75747848e-01 -1.23496130e-01 -6.71277940e-02 9.68903840e-01
5.17733283e-02 3.50673884e-01 -8.49320292e-01 -2.85569906e-01
6.32653952e-01 4.49838281e-01 9.89121318e-01 7.63690412e-01
9.12873566e-01 3.53636950e-01 -1.71818092e-01 9.63744104e-01
1.66643143e+00 3.75207335e-01 1.17096174e+00 5.14936268e-01
2.13581860e-01 3.23785424e-01 7.32840121e-01 3.27651083e-01
-3.76886338e-01 6.24740124e-01 5.58942795e-01 -4.40886289e-01
-6.57773674e-01 1.69292361e-01 3.88188750e-01 8.37306023e-01
-2.12287694e-01 3.85819823e-01 -5.79374313e-01 1.63861901e-01
-1.63491392e+00 -1.19237125e+00 -8.43073368e-01 2.02035856e+00
5.17950892e-01 2.24527046e-01 -1.63786530e-01 5.59913397e-01
6.83926761e-01 -3.60026151e-01 1.55684814e-01 -8.63238454e-01
9.09272656e-02 8.33367109e-01 8.65801573e-01 4.54659671e-01
-1.42391455e+00 5.64769685e-01 7.24957752e+00 8.88279676e-01
-9.97899711e-01 3.16866972e-02 3.93402994e-01 2.85871089e-01
2.43969008e-01 2.17117593e-01 -7.46830761e-01 4.39664423e-01
5.95671356e-01 3.38762462e-01 2.63810962e-01 5.27349114e-01
7.69789159e-01 -5.23812652e-01 -8.04863334e-01 1.09988582e+00
-9.07788053e-02 -1.41931713e+00 -2.06056118e-01 -1.53442338e-01
6.77999139e-01 -3.77851576e-01 -4.14391458e-01 -1.79206505e-01
-1.81185141e-01 -7.06344008e-01 8.20255399e-01 4.13641006e-01
3.03712726e-01 -7.70183206e-01 9.61063325e-01 1.53340707e-02
-1.07870483e+00 7.67649338e-02 -6.77772164e-01 -4.53238040e-01
2.31571808e-01 9.37110007e-01 -4.63104755e-01 4.00742412e-01
1.20801473e+00 7.32194126e-01 -3.29580456e-01 1.55830038e+00
6.66620880e-02 5.99181652e-01 -1.85978383e-01 4.94951397e-01
2.21116953e-02 -6.48269713e-01 3.89463514e-01 1.27117705e+00
6.32278681e-01 2.36460820e-01 -2.91610479e-01 1.93783507e-01
1.66873097e-01 -2.30566114e-01 -5.00108540e-01 5.65936565e-01
3.32928151e-02 9.90308344e-01 -8.14690232e-01 -4.06820565e-01
-7.97180414e-01 1.08354139e+00 -7.41826057e-01 -2.10843116e-01
-7.56299078e-01 -8.16913545e-01 9.33165908e-01 2.76187062e-01
5.47365069e-01 -6.69439971e-01 -6.49054945e-01 -6.96782947e-01
-3.03762257e-01 -1.07032001e+00 1.60482489e-02 -1.01380837e+00
-1.24162912e+00 2.20036164e-01 -2.16680199e-01 -1.54057837e+00
1.44460618e-01 -9.17824864e-01 -5.44124007e-01 8.76393139e-01
-1.20713925e+00 -1.08850563e+00 -6.01188242e-01 1.00022852e+00
5.27685881e-01 -7.57147744e-02 6.28880560e-01 8.62387061e-01
1.02069996e-01 6.80324342e-03 1.74750745e-01 -7.51304254e-02
8.14312041e-01 -8.97300482e-01 7.41371334e-01 1.08353627e+00
-6.93461439e-03 7.13689506e-01 8.95408869e-01 -7.23736942e-01
-1.77477467e+00 -7.29739368e-01 5.42616546e-01 1.68022200e-01
9.46656540e-02 -4.09160495e-01 -3.51474464e-01 5.39654076e-01
6.52209759e-01 -1.36769086e-01 3.95448655e-01 -3.45124245e-01
1.50732696e-02 -6.59003198e-01 -1.49381268e+00 5.56046903e-01
1.16145074e+00 -1.94508210e-01 -5.31943262e-01 1.59001142e-01
2.68068790e-01 -6.67665362e-01 -8.79870832e-01 2.93359488e-01
4.95862007e-01 -1.56394625e+00 1.33810782e+00 7.89032876e-02
1.37202680e-01 -7.58038342e-01 -3.04356903e-01 -8.56722414e-01
-2.87093788e-01 -9.61398304e-01 2.95453250e-01 1.12277555e+00
-4.30182010e-01 -4.61321056e-01 6.91645324e-01 1.27620280e-01
-4.74980086e-01 -1.13907792e-01 -1.02293050e+00 -6.47744596e-01
-5.15077829e-01 -5.69696188e-01 5.18989623e-01 3.36033642e-01
-3.73765111e-01 -2.48093724e-01 -4.09253448e-01 2.31338888e-01
7.46492624e-01 -1.02445595e-01 8.28181028e-01 -9.35372651e-01
-6.60291612e-02 -1.15853943e-01 -1.13700569e+00 -1.19855273e+00
-3.73827130e-01 -3.79820347e-01 -3.00577044e-01 -1.24342513e+00
-1.90467611e-01 -4.14654873e-02 2.42778569e-01 1.68172121e-01
2.70841539e-01 8.36949825e-01 2.08535150e-01 -2.02398896e-01
-4.46937867e-02 -4.64140065e-02 1.08646035e+00 -1.21611111e-01
6.06431030e-02 1.85691088e-01 -3.74544114e-02 8.05621624e-01
5.91063976e-01 -4.54178214e-01 1.90138351e-02 -6.41374409e-01
-8.08267444e-02 -1.94283441e-01 5.12344599e-01 -1.58572781e+00
5.02335966e-01 4.80173171e-01 6.83329046e-01 -9.01616395e-01
4.19660956e-01 -1.48478031e+00 6.31703258e-01 9.06050801e-01
4.24655378e-01 4.44677532e-01 3.15836847e-01 5.88119365e-02
-4.59473729e-01 -6.58730626e-01 9.03684735e-01 -3.82451713e-01
-9.92495120e-01 -2.19140142e-01 -9.03858006e-01 -5.73627770e-01
1.24012923e+00 -8.42975259e-01 -3.04737724e-02 -2.29490876e-01
-5.49385071e-01 -2.90542275e-01 6.97380066e-01 -1.55058742e-01
7.30387688e-01 -1.15022421e+00 -6.02144361e-01 6.08980715e-01
-6.92960858e-01 -4.41011131e-01 9.95745882e-02 7.36954808e-01
-1.62264442e+00 1.36402667e-01 -6.10125601e-01 -7.05930710e-01
-1.97897851e+00 2.88559407e-01 3.14419627e-01 1.00824907e-01
-7.10387766e-01 9.26145434e-01 -5.93140900e-01 9.75157022e-02
2.39419922e-01 -3.05296481e-01 2.48757377e-01 -1.17778838e-01
6.99493229e-01 7.11221814e-01 7.37781703e-01 -9.69826505e-02
-3.37497503e-01 9.33748662e-01 3.51206422e-01 5.04030474e-02
1.46187818e+00 -2.90578548e-02 -5.23625076e-01 -5.71431458e-01
1.17341733e+00 1.50007963e-01 -1.12324846e+00 6.20862246e-01
-1.76583275e-01 -1.04475760e+00 6.40251756e-01 -3.15163612e-01
-1.25673306e+00 6.89333856e-01 1.14244747e+00 1.50548488e-01
1.79970396e+00 -6.59946799e-01 8.57949317e-01 8.50533545e-02
5.07574558e-01 -1.01602042e+00 -4.15217131e-01 3.47566307e-01
4.67583239e-01 -7.18464375e-01 4.92085665e-01 -7.90081799e-01
2.86407471e-01 1.69354939e+00 1.65593237e-01 -6.29953265e-01
1.09752393e+00 8.07577670e-01 2.55485207e-01 -1.32301480e-01
-3.67219448e-01 -1.24719806e-01 2.14958955e-02 9.43881869e-01
5.43670356e-01 -5.18364906e-01 -8.13948691e-01 4.70338091e-02
2.49407783e-01 1.70517191e-01 5.04165590e-01 1.31627071e+00
-4.56739724e-01 -1.58375466e+00 -8.99209142e-01 4.20971364e-01
-5.30162513e-01 -1.94310173e-01 9.91687998e-02 7.41301656e-01
6.48346007e-01 1.34017479e+00 1.42647028e-01 -5.34048200e-01
6.32990301e-01 -5.48998475e-01 7.96877623e-01 -2.41740480e-01
-1.45840418e+00 3.01853240e-01 -2.75697690e-02 -1.11268938e+00
-8.33073795e-01 -5.71320951e-01 -7.97555029e-01 -7.97542632e-01
-1.09876640e-01 -3.33392203e-01 1.02929533e+00 1.52494013e-01
3.98181707e-01 3.97732973e-01 4.91113961e-01 -1.14531088e+00
-2.38750890e-01 -6.14478886e-01 -7.54810512e-01 2.91048080e-01
9.25573036e-02 -3.63332659e-01 -2.86306202e-01 4.47585225e-01] | [11.357461929321289, -2.359069585800171] |
42a2862d-88eb-490a-a3e0-0c73716c223e | membership-inference-attacks-from-first | 2112.03570 | null | https://arxiv.org/abs/2112.03570v2 | https://arxiv.org/pdf/2112.03570v2.pdf | Membership Inference Attacks From First Principles | A membership inference attack allows an adversary to query a trained machine learning model to predict whether or not a particular example was contained in the model's training dataset. These attacks are currently evaluated using average-case "accuracy" metrics that fail to characterize whether the attack can confidently identify any members of the training set. We argue that attacks should instead be evaluated by computing their true-positive rate at low (e.g., <0.1%) false-positive rates, and find most prior attacks perform poorly when evaluated in this way. To address this we develop a Likelihood Ratio Attack (LiRA) that carefully combines multiple ideas from the literature. Our attack is 10x more powerful at low false-positive rates, and also strictly dominates prior attacks on existing metrics. | ['Florian Tramer', 'Andreas Terzis', 'Shuang Song', 'Milad Nasr', 'Steve Chien', 'Nicholas Carlini'] | 2021-12-07 | null | null | null | null | ['membership-inference-attack'] | ['computer-vision'] | [ 2.06540361e-01 8.14812109e-02 -4.57979798e-01 -1.17729537e-01
-1.23204160e+00 -1.29966295e+00 6.24760866e-01 4.72069949e-01
-5.00587583e-01 8.19737136e-01 -7.14077115e-01 -1.03812993e+00
5.84039986e-02 -1.20263529e+00 -8.14866483e-01 -4.47835892e-01
-2.43194550e-01 6.53126419e-01 4.83733445e-01 3.58614296e-01
3.61848652e-01 7.13813126e-01 -8.89187753e-01 2.16966987e-01
3.96836132e-01 6.90001905e-01 -8.46332431e-01 9.91460502e-01
5.10975718e-01 7.19957709e-01 -1.00242579e+00 -6.34232640e-01
3.50897819e-01 -1.92077354e-01 -9.73230660e-01 -5.33885300e-01
3.80092442e-01 -5.66219330e-01 -4.31396544e-01 1.39743388e+00
-8.95411298e-02 -3.35817635e-01 6.17225170e-01 -1.71951735e+00
3.39166969e-02 6.98515117e-01 -3.21254730e-01 5.89969099e-01
4.16670054e-01 2.99951196e-01 1.10310864e+00 -4.25667197e-01
1.95883781e-01 8.99448276e-01 4.35795754e-01 3.70949775e-01
-1.33509719e+00 -1.14722216e+00 -1.48978576e-01 -9.23922062e-02
-1.52466750e+00 -2.09244162e-01 3.46228033e-01 -2.39841074e-01
6.63256407e-01 6.86866164e-01 1.04680620e-02 1.11807752e+00
2.03837112e-01 2.64148921e-01 1.27690804e+00 -2.71571040e-01
4.00864571e-01 3.43925446e-01 3.20954882e-02 5.29829860e-01
9.39305723e-01 4.81112033e-01 -1.59523472e-01 -1.11218667e+00
6.21957719e-01 -3.21765959e-01 -1.48110673e-01 -1.48632661e-01
-8.21846306e-01 1.16517460e+00 3.23171029e-03 2.71536887e-01
9.93624181e-02 2.82742172e-01 2.04939395e-01 4.04043376e-01
1.52598098e-01 7.91198730e-01 -6.05352998e-01 1.92265306e-02
-8.78760993e-01 3.51269603e-01 1.20570719e+00 5.44124067e-01
6.91752553e-01 -6.56002834e-02 1.88068166e-01 3.23070548e-02
3.25354069e-01 5.97400904e-01 -3.93613242e-02 -7.93117762e-01
2.85556942e-01 2.52483368e-01 3.34393293e-01 -9.52063859e-01
6.69893250e-02 -5.98456085e-01 -2.66312391e-01 3.27572584e-01
7.06042349e-01 -3.69880587e-01 -5.78477025e-01 1.93639755e+00
1.65082604e-01 3.65418881e-01 4.58964705e-02 4.30696309e-01
1.80722505e-01 5.78325987e-01 3.31417859e-01 -2.46042162e-01
1.20099056e+00 -8.53684172e-03 -1.60079345e-01 -8.00014362e-02
8.63971591e-01 -6.25058889e-01 9.18635905e-01 6.25453711e-01
-7.99315512e-01 5.02132811e-02 -1.46100438e+00 7.48088479e-01
-4.08267111e-01 -4.30268884e-01 7.30752468e-01 1.48598886e+00
-4.86068428e-01 3.71632844e-01 -7.68785298e-01 2.07436383e-01
5.11945188e-01 3.64547908e-01 -2.80540317e-01 1.34951621e-01
-1.41201007e+00 8.02561402e-01 4.02973086e-01 -6.31935000e-01
-1.48405313e+00 -6.97138786e-01 -4.26393926e-01 4.65064459e-02
5.49261451e-01 -2.88387001e-01 1.32916856e+00 -4.03761327e-01
-7.85270452e-01 8.16117287e-01 -2.79750042e-02 -8.82937253e-01
6.12718642e-01 -1.09219804e-01 -5.82542658e-01 3.77361238e-01
-2.35786825e-01 -5.99219464e-03 7.05512881e-01 -1.42313361e+00
-7.37090945e-01 -1.74476102e-01 5.88565528e-01 -4.14071292e-01
-2.15378314e-01 3.06939900e-01 1.36183947e-01 -5.65737665e-01
-1.29283428e-01 -1.03461540e+00 -4.06113058e-01 -2.85571367e-01
-5.32727659e-01 -3.24805938e-02 1.19933021e+00 -5.71061336e-02
1.28870630e+00 -1.74551821e+00 -9.25348520e-01 9.01998401e-01
1.61620110e-01 7.51277685e-01 1.08507551e-01 3.12008858e-01
-6.05430044e-02 8.13134611e-01 -2.78766125e-01 1.75125539e-01
-1.12049729e-01 1.14400655e-01 -9.96164858e-01 8.06103766e-01
1.58064172e-01 4.53375697e-01 -7.22310781e-01 -2.43511125e-01
2.42482629e-02 1.27927169e-01 -7.28418887e-01 2.12029561e-01
-3.08265060e-01 4.61261310e-02 -5.55776834e-01 4.48414683e-01
6.33971214e-01 -4.05548811e-01 1.92543507e-01 -4.09864821e-02
3.57949466e-01 5.19596756e-01 -1.27288401e+00 6.17893875e-01
-2.72556633e-01 4.58241373e-01 -1.00021437e-01 -7.32140541e-01
6.46274328e-01 3.32878888e-01 3.78401317e-02 3.05950269e-03
1.83796287e-01 3.11078459e-01 2.30782539e-01 -1.07082445e-02
2.51371646e-04 -2.97631085e-01 -2.92477667e-01 7.90064931e-01
-3.09190929e-01 1.82448421e-02 -1.06578648e-01 3.34295243e-01
1.47844458e+00 -6.18403733e-01 3.12533438e-01 -2.29755715e-01
4.88416076e-01 -5.69449179e-02 4.90187913e-01 1.26933992e+00
-3.47105801e-01 2.78775364e-01 7.56977022e-01 -1.62611976e-01
-8.74157846e-01 -1.44862556e+00 -4.46075261e-01 7.09355772e-01
2.43276909e-01 -7.04614639e-01 -6.22736394e-01 -1.29386914e+00
3.79871689e-02 8.31736684e-01 -4.77109104e-01 -2.66633838e-01
-1.86526522e-01 -1.04962063e+00 1.34522295e+00 3.49305809e-01
4.76674527e-01 -3.80063564e-01 -4.59735692e-01 -1.78218901e-01
9.29464549e-02 -1.09012699e+00 -3.44377123e-02 2.37857908e-01
-4.32469249e-01 -1.50708520e+00 2.55485207e-01 -3.36217791e-01
7.13593662e-01 -2.10795239e-01 9.85589921e-01 2.45539755e-01
-1.66142941e-01 2.99490035e-01 -2.72522330e-01 -4.02878970e-01
-1.00323701e+00 -1.94327030e-02 1.08092912e-01 -1.42700449e-01
6.68055654e-01 -5.72775841e-01 -2.26609319e-01 4.48581368e-01
-9.96587813e-01 -9.52511191e-01 4.61693376e-01 5.03598571e-01
1.97304383e-01 4.44957912e-01 7.67800868e-01 -1.40712655e+00
5.38028657e-01 -6.13990963e-01 -9.27023172e-01 3.32584113e-01
-5.90397596e-01 -7.01416731e-02 8.34206998e-01 -7.83959568e-01
-3.78039122e-01 -2.10797727e-01 -5.13246693e-02 -4.11928684e-01
-4.30520833e-01 3.46280634e-01 -3.20330352e-01 -3.50998044e-01
9.80107725e-01 -3.30449045e-02 -2.69710302e-01 -2.02212498e-01
4.93728258e-02 5.06163001e-01 5.67216337e-01 -9.55653906e-01
1.40690839e+00 2.44733781e-01 2.36135945e-01 -4.30618852e-01
-8.33989978e-01 -1.20555103e-01 -6.92733824e-02 1.48862451e-01
1.82689384e-01 -5.84384143e-01 -9.77835178e-01 1.70409173e-01
-8.02814245e-01 -1.57357335e-01 6.25480488e-02 5.84997356e-01
-1.51504189e-01 5.44203341e-01 -6.12188756e-01 -1.03658724e+00
-1.29384249e-01 -9.68749821e-01 2.94141948e-01 9.40655079e-03
-4.80054170e-01 -9.97722089e-01 -5.75020090e-02 2.14122772e-01
3.59200865e-01 4.60059196e-01 1.00406182e+00 -1.65855908e+00
-3.84783745e-01 -9.63363051e-01 -1.17551215e-01 4.43159610e-01
3.07472330e-02 3.36598396e-01 -1.14693105e+00 -4.90954638e-01
2.28709817e-01 -3.52007627e-01 5.67789972e-01 -1.95741326e-01
1.39540040e+00 -9.47248995e-01 -3.78824204e-01 2.29270861e-01
1.41303861e+00 3.29728514e-01 6.00677013e-01 1.47905231e-01
8.11075643e-02 2.29096442e-01 4.12046373e-01 5.56934118e-01
-2.34483257e-01 3.08633506e-01 4.58470315e-01 2.02644691e-01
5.51030338e-01 -5.50817728e-01 2.91117162e-01 -2.68683672e-01
5.13987124e-01 -3.50526989e-01 -9.39342618e-01 1.86309621e-01
-1.05050588e+00 -1.13574958e+00 6.87089860e-02 2.58070636e+00
1.18753958e+00 9.92122769e-01 2.95885086e-01 4.01577652e-01
7.09950984e-01 4.96415235e-02 -4.81533140e-01 -4.22350824e-01
-1.18555285e-01 6.48783326e-01 8.82459521e-01 7.45657623e-01
-1.45090365e+00 7.40130663e-01 7.48719263e+00 9.90657747e-01
-1.04297650e+00 -6.67217225e-02 1.02857745e+00 -2.99830399e-02
-5.29033124e-01 5.95860243e-01 -7.94377267e-01 4.69795614e-01
1.34025168e+00 -5.10071933e-01 1.49109364e-01 1.06476176e+00
-4.98697013e-01 -1.25767380e-01 -1.26959443e+00 5.13451576e-01
-6.99512959e-02 -1.33414996e+00 1.57514602e-01 4.15833980e-01
4.18872893e-01 -1.47252619e-01 4.24848050e-01 4.22793239e-01
8.54539335e-01 -1.34438908e+00 2.75869310e-01 -1.00656720e-02
8.66370678e-01 -1.16887772e+00 6.03719175e-01 6.78841770e-01
-6.87847257e-01 -5.44597395e-02 -1.04119182e-01 8.63563493e-02
-3.56443495e-01 4.80430782e-01 -1.22217560e+00 2.24048570e-01
2.43601739e-01 -1.94446623e-01 -5.56116045e-01 1.09102201e+00
-2.13916078e-01 1.43625164e+00 -8.95911217e-01 1.92772914e-02
2.60723144e-01 4.12611187e-01 6.68214738e-01 1.19237864e+00
-4.61429022e-02 1.33552507e-01 4.05395210e-01 8.21986437e-01
-2.87479699e-01 -3.73857677e-01 -5.67137957e-01 -1.48765430e-01
9.98001456e-01 7.89150357e-01 -4.97573078e-01 -2.88411200e-01
-1.40794069e-01 1.59177944e-01 -2.18619451e-01 1.43054485e-01
-8.43434691e-01 -5.30947685e-01 5.66195965e-01 2.24945739e-01
3.10648847e-02 -1.38510555e-01 -1.76117986e-01 -1.01224577e+00
-1.50289699e-01 -1.10310519e+00 7.37977505e-01 -2.06931725e-01
-1.33355534e+00 3.61659437e-01 2.37442136e-01 -1.04422271e+00
-5.73094249e-01 -5.25886655e-01 -7.75403142e-01 8.63571286e-01
-1.01887846e+00 -8.09517801e-01 5.60356140e-01 4.74227041e-01
-3.83637726e-01 -1.53543696e-01 1.20102870e+00 3.91196506e-03
-5.49832284e-01 1.37492120e+00 -2.66595155e-01 5.16451776e-01
4.43540901e-01 -1.07916844e+00 3.70314538e-01 1.29135239e+00
4.47372496e-01 9.27958250e-01 1.08903599e+00 -6.90970182e-01
-1.09930253e+00 -9.48569477e-01 5.80621958e-01 -9.00393903e-01
7.08985150e-01 -1.21842466e-01 -8.59613359e-01 8.47309947e-01
-4.87781197e-01 2.38790393e-01 9.61928964e-01 2.39364848e-01
-1.22804630e+00 1.81051176e-02 -1.68285978e+00 4.70707625e-01
3.49672288e-01 -8.27453613e-01 -5.29956996e-01 1.63621694e-01
5.11302114e-01 -5.17646410e-02 -9.39490020e-01 7.97819555e-01
4.44267362e-01 -6.89161241e-01 8.83086979e-01 -1.03685617e+00
-2.30634540e-01 -3.98447692e-01 -4.08433467e-01 -6.22718990e-01
1.18877172e-01 -7.65333831e-01 -4.12571490e-01 1.11559093e+00
7.26320267e-01 -9.97182429e-01 1.08018064e+00 6.51786327e-01
6.28209233e-01 -5.26811123e-01 -1.18404162e+00 -8.70219469e-01
5.14048815e-01 -6.15447700e-01 5.77340066e-01 1.24124181e+00
6.18935190e-02 2.46965159e-02 -1.74537942e-01 8.58733833e-01
8.12353194e-01 -5.64753972e-02 8.22056532e-01 -1.24106228e+00
-4.52708185e-01 -4.57984835e-01 -5.44318020e-01 -5.15944839e-01
1.91463828e-02 -7.09724784e-01 -4.33664918e-01 -5.47328651e-01
1.32139042e-01 -6.87087893e-01 -4.09659952e-01 6.28869951e-01
-7.82698169e-02 4.30497617e-01 -9.85637531e-02 1.85977712e-01
-3.51823121e-01 -1.48644984e-01 4.37410414e-01 1.08184395e-02
2.90024132e-01 4.81692642e-01 -8.81776452e-01 8.06994498e-01
1.05660748e+00 -7.39762723e-01 -3.59470189e-01 4.73753780e-01
3.22537273e-01 -6.24756739e-02 6.68801010e-01 -1.24687505e+00
1.94804952e-01 -4.13391680e-01 3.59895527e-01 -3.16649854e-01
6.12541810e-02 -6.36233151e-01 1.36852100e-01 8.43953013e-01
-7.15403497e-01 -4.64995904e-03 1.45336062e-01 5.55839658e-01
1.47187009e-01 -3.99675667e-01 1.03591096e+00 1.12118505e-01
6.56551030e-03 2.97243327e-01 -3.32841039e-01 2.52128273e-01
9.81309891e-01 1.29467562e-01 -6.56284034e-01 -3.85323524e-01
-5.36670029e-01 -2.35980488e-02 6.06426895e-01 -1.73050672e-01
4.93369490e-01 -8.20549786e-01 -8.02813828e-01 2.96675950e-01
2.11264640e-01 -4.18850958e-01 -3.40239555e-01 2.53888816e-01
-4.40548182e-01 2.15265974e-01 2.32635543e-01 -2.84405529e-01
-1.32718730e+00 5.93271971e-01 4.68703926e-01 -4.33898062e-01
-2.25363672e-01 8.15119028e-01 7.27556422e-02 -8.34979117e-02
3.83182973e-01 1.36939734e-01 2.95638472e-01 -5.22177100e-01
7.79490769e-01 1.54753044e-01 2.07846500e-02 -4.05048698e-01
-5.31712413e-01 -6.92410618e-02 -3.93891066e-01 -3.17769200e-01
6.66225493e-01 5.90430856e-01 2.00574964e-01 1.16771050e-01
1.28043628e+00 5.21027803e-01 -6.91387296e-01 -2.13202491e-01
-2.40858104e-02 -7.83848166e-01 -1.96931422e-01 -9.71739590e-01
-6.46098077e-01 6.88899875e-01 4.26896065e-01 6.70299113e-01
7.04712272e-01 -6.00600336e-03 5.82054436e-01 6.71388924e-01
4.81273592e-01 -5.81961632e-01 -5.56370690e-02 1.63635164e-01
3.96463513e-01 -8.91898990e-01 2.55438298e-01 -3.91248584e-01
-2.94037968e-01 7.53180861e-01 6.27638042e-01 -3.54682475e-01
7.51584470e-01 6.93604171e-01 -2.71197557e-01 1.78218678e-01
-1.01903105e+00 2.87011623e-01 8.85732621e-02 3.93216312e-01
3.10424287e-02 2.02061966e-01 -1.45336390e-01 7.13097095e-01
-3.93316716e-01 -3.55211288e-01 6.45885646e-01 8.40885222e-01
-5.88468671e-01 -1.19206667e+00 -4.94616747e-01 6.19730413e-01
-1.19495344e+00 7.02692196e-02 -5.06474257e-01 7.97178328e-01
-1.71126306e-01 1.20366693e+00 -1.03760444e-01 -7.59658933e-01
-7.78157935e-02 1.86764973e-03 1.51822537e-01 -5.85148454e-01
-6.45574570e-01 -2.41770029e-01 3.45314384e-01 -4.17702496e-01
-1.22794233e-01 -4.81455833e-01 -1.02731442e+00 -6.63203061e-01
-5.22035658e-01 6.32497549e-01 3.50241721e-01 8.38255286e-01
-7.54320025e-02 -3.16307396e-01 1.07188952e+00 1.11556947e-01
-1.15707898e+00 -5.80240428e-01 -3.42768848e-01 1.92101151e-01
2.34156892e-01 -3.79259467e-01 -1.05959189e+00 -4.00945783e-01] | [5.905200481414795, 7.344105243682861] |
aca585eb-8566-4389-bcbe-4890a5a68d2a | a-survey-on-measuring-and-mitigating | 2209.01824 | null | https://arxiv.org/abs/2209.01824v1 | https://arxiv.org/pdf/2209.01824v1.pdf | A Survey on Measuring and Mitigating Reasoning Shortcuts in Machine Reading Comprehension | The issue of shortcut learning is widely known in NLP and has been an important research focus in recent years. Unintended correlations in the data enable models to easily solve tasks that were meant to exhibit advanced language understanding and reasoning capabilities. In this survey paper, we focus on the field of machine reading comprehension (MRC), an important task for showcasing high-level language understanding that also suffers from a range of shortcuts. We summarize the available techniques for measuring and mitigating shortcuts and conclude with suggestions for further progress in shortcut research. Most importantly, we highlight two main concerns for shortcut mitigation in MRC: the lack of public challenge sets, a necessary component for effective and reusable evaluation, and the lack of certain mitigation techniques that are prominent in other areas. | ['Akiko Aizawa', 'Saku Sugawara', 'Johannes Mario Meissner', 'Xanh Ho'] | 2022-09-05 | null | null | null | null | ['machine-reading-comprehension'] | ['natural-language-processing'] | [ 5.45530021e-01 5.69310725e-01 -3.74406010e-01 -6.96023524e-01
-1.04213572e+00 -9.09473002e-01 6.28973901e-01 6.33453250e-01
-4.52353150e-01 6.55765235e-01 6.27560198e-01 -8.81223679e-01
-5.68988621e-01 -6.86705232e-01 -9.74896729e-01 -8.35663229e-02
9.76352319e-02 4.20216918e-01 -3.25628556e-02 -4.20170844e-01
4.40485418e-01 2.55079567e-01 -1.64399898e+00 8.16093504e-01
1.08708620e+00 5.25266171e-01 1.12663105e-01 8.62858474e-01
-4.24470901e-01 1.20222688e+00 -7.29588866e-01 -6.58741951e-01
-1.00110762e-01 -5.02517819e-02 -1.43620837e+00 -5.85867822e-01
8.63412619e-01 -9.00641605e-02 -6.49582967e-02 7.93924093e-01
3.41506332e-01 1.89742431e-01 4.72844839e-01 -1.25848532e+00
-3.93129945e-01 1.10512030e+00 -3.58272433e-01 4.98975813e-01
6.29459381e-01 1.62931904e-01 1.58798659e+00 -5.93214512e-01
4.96264815e-01 1.42751634e+00 5.20262659e-01 7.36377299e-01
-1.03025866e+00 -6.49183452e-01 3.38635236e-01 3.70361924e-01
-8.83861840e-01 -2.88986027e-01 2.39454523e-01 -4.15096909e-01
1.37300122e+00 5.89459538e-01 2.66068429e-01 1.16182697e+00
1.50888652e-01 1.17510211e+00 1.20709491e+00 -7.25754619e-01
-1.66136295e-01 3.37756611e-02 9.51239169e-01 6.45626903e-01
2.59397447e-01 1.86651483e-01 -7.67953932e-01 -4.61753719e-02
1.73629209e-01 -5.12165308e-01 -3.84827524e-01 -2.44637534e-01
-1.04207337e+00 6.60151064e-01 1.24294020e-01 3.35965335e-01
1.89764619e-01 1.24139540e-01 3.62757951e-01 7.08215058e-01
1.90068051e-01 1.10770452e+00 -1.18029618e+00 -3.49486887e-01
-7.81863034e-01 4.36554521e-01 1.00323105e+00 1.04549456e+00
5.24273992e-01 -3.81695181e-01 -4.26540315e-01 5.16305387e-01
-5.18324003e-02 2.32108712e-01 7.80350296e-04 -8.46615970e-01
8.60149503e-01 7.01206088e-01 -1.35279968e-01 -7.14450717e-01
-5.28130293e-01 -5.55044591e-01 -4.59722400e-01 -4.81033931e-03
6.67208254e-01 -1.18300371e-01 -5.77843547e-01 2.01716590e+00
-8.80199373e-02 -7.77285323e-02 1.23395897e-01 4.29594100e-01
8.78368616e-01 4.04846281e-01 2.66264468e-01 -1.54594267e-02
1.31026125e+00 -8.83402586e-01 -4.63130504e-01 -5.05043864e-01
9.89821553e-01 -6.23316228e-01 1.72872853e+00 5.55541039e-01
-1.06195366e+00 -1.91683307e-01 -1.07303607e+00 -5.40406048e-01
-4.63118374e-01 -4.63368088e-01 9.33774412e-01 6.46296918e-01
-7.25853801e-01 5.00026822e-01 -5.61297596e-01 -2.17578128e-01
3.58029395e-01 1.48766175e-01 -1.06855668e-01 -3.60885322e-01
-1.37131369e+00 1.33251286e+00 2.78682470e-01 -2.61586159e-01
-8.25212359e-01 -1.22228837e+00 -8.29403996e-01 2.35113159e-01
7.22485304e-01 -6.49979234e-01 1.64492476e+00 -8.67589176e-01
-9.77298737e-01 9.71091211e-01 -1.84358507e-01 -4.31496948e-01
4.91967231e-01 -8.18988562e-01 -2.76923805e-01 -1.93595722e-01
-7.39429817e-02 4.75832552e-01 3.78001630e-01 -7.78569579e-01
-5.40295899e-01 -2.35456690e-01 5.54063678e-01 3.86917263e-01
-1.37286767e-01 2.45655060e-01 7.89834112e-02 -5.59175551e-01
-2.11997032e-01 -2.89460897e-01 -1.03479763e-02 -4.49346125e-01
-4.89720941e-01 -4.43443149e-01 3.06427538e-01 -5.92380583e-01
1.37172258e+00 -1.68700302e+00 1.88110396e-01 2.51234472e-02
3.66998702e-01 1.55895323e-01 -3.49534661e-01 6.19556069e-01
-2.26372957e-01 6.11045957e-01 -1.90782472e-01 1.01386324e-01
9.02099162e-02 1.89436272e-01 -7.11777866e-01 -4.05484848e-02
2.17726856e-01 1.06746113e+00 -1.13601065e+00 -1.61673024e-01
-1.67567991e-02 -4.76090498e-02 -4.63482112e-01 2.58108974e-01
-8.45826328e-01 1.58034757e-01 -3.98932517e-01 5.57353675e-01
3.64866346e-01 -1.72356948e-01 -2.67698541e-02 1.94003299e-01
-2.13262901e-01 1.06140912e+00 -8.68434191e-01 1.55148876e+00
-5.26929796e-01 8.01896274e-01 -3.08507774e-02 -8.43881190e-01
3.72849762e-01 1.59102470e-01 -3.83051299e-02 -5.25116503e-01
-1.22736255e-02 1.22206613e-01 2.22915769e-01 -6.43282473e-01
5.26330709e-01 -4.51717190e-02 -1.60264939e-01 8.21063280e-01
-6.07666699e-03 -2.84618944e-01 3.72777879e-01 4.80870038e-01
1.35765612e+00 -1.68456316e-01 3.49257052e-01 -5.25084615e-01
5.47495782e-01 1.42617211e-01 1.89357087e-01 1.25477397e+00
1.37837559e-01 3.34384173e-01 7.66003430e-01 -3.08845997e-01
-6.95004702e-01 -9.35966909e-01 -1.43415198e-01 1.53442359e+00
-1.97527945e-01 -7.46813178e-01 -7.32537031e-01 -8.50784302e-01
-2.47698780e-02 1.18025696e+00 -5.24449766e-01 -2.86625773e-01
-5.30367076e-01 -5.87046683e-01 1.09632921e+00 7.10139573e-01
2.35225827e-01 -9.79057729e-01 -6.71745479e-01 -4.44533303e-02
-4.35968250e-01 -8.80784869e-01 -4.48820740e-02 4.28092241e-01
-8.41218770e-01 -1.41127765e+00 -3.55127335e-01 -6.61522925e-01
4.02297497e-01 2.95761824e-01 1.94657874e+00 5.11506677e-01
-2.52002031e-01 7.11814463e-01 -4.70277429e-01 -8.72678101e-01
-6.20120049e-01 4.33653831e-01 -2.03639016e-01 -8.80732179e-01
7.85383165e-01 -1.98347807e-01 -3.28928307e-02 1.15560390e-01
-9.16507781e-01 2.51688451e-01 5.21847486e-01 7.90197909e-01
2.58296847e-01 -3.68741900e-02 5.48338354e-01 -1.35470831e+00
9.89225566e-01 -4.33837026e-01 -5.09241223e-01 8.19901645e-01
-7.84450054e-01 2.22567707e-01 3.96318197e-01 3.39391939e-02
-1.18504989e+00 -5.45919061e-01 -1.87766597e-01 4.06859726e-01
-3.36180568e-01 7.76825547e-01 -3.86716574e-01 7.54057616e-02
8.22431087e-01 -1.34907722e-01 -2.30653092e-01 -6.48487747e-01
7.03345358e-01 4.94020492e-01 2.40730852e-01 -1.12720418e+00
6.14730239e-01 -7.69882649e-02 -3.50713879e-01 -8.87725651e-01
-1.55331647e+00 -5.63432097e-01 -4.45986927e-01 1.98391840e-01
6.57867789e-01 -7.93444633e-01 -3.92815411e-01 3.43664646e-01
-1.26734304e+00 -6.00666583e-01 -3.13467145e-01 1.75905898e-01
-5.62786579e-01 1.72037587e-01 -6.36874497e-01 -4.56394941e-01
-3.48587066e-01 -9.63515759e-01 6.97298586e-01 3.20205420e-01
-7.07220495e-01 -1.13847244e+00 -9.52989236e-03 6.39377892e-01
3.24637383e-01 -2.46522993e-01 1.74722648e+00 -8.96785676e-01
-6.83889568e-01 2.84204274e-01 -2.01856673e-01 5.04723191e-01
-2.76857018e-01 -1.66294977e-01 -1.18700099e+00 -1.34243563e-01
3.76820862e-02 -7.11432338e-01 9.37586606e-01 2.35640347e-01
1.36919057e+00 -2.46842518e-01 -3.68722007e-02 3.69188309e-01
1.23188174e+00 -2.63234526e-01 4.98061955e-01 3.65849733e-01
5.74126184e-01 9.70378220e-01 7.06898630e-01 -1.17800077e-02
3.60778034e-01 1.99339584e-01 1.19658664e-01 -6.19435534e-02
-1.41751483e-01 -5.48559606e-01 3.92950296e-01 8.08712125e-01
1.12914495e-01 -4.87217784e-01 -1.34373426e+00 5.01285255e-01
-1.57811892e+00 -9.17351365e-01 -3.53449225e-01 2.03050447e+00
1.13156354e+00 2.15378061e-01 -3.69087160e-01 8.98080617e-02
3.46123517e-01 2.03717560e-01 -4.01679099e-01 -7.50272810e-01
-3.67926091e-01 7.39945650e-01 4.07592297e-01 8.74958098e-01
-8.04783046e-01 1.16175234e+00 7.54154205e+00 6.22681737e-01
-5.27116954e-01 -7.96181858e-02 4.95828331e-01 8.36370736e-02
-7.86728919e-01 1.61210343e-01 -8.63706231e-01 -2.29708344e-01
6.94849014e-01 -4.41680044e-01 3.94929886e-01 6.24031007e-01
-1.27920464e-01 -2.24258378e-01 -1.57935417e+00 3.59524310e-01
4.45729382e-02 -1.25283945e+00 1.88508496e-01 -4.23712879e-01
7.68162847e-01 2.92929739e-01 4.09938395e-03 4.77349550e-01
4.83325660e-01 -1.42138064e+00 4.13950801e-01 2.92302161e-01
6.27199233e-01 -8.27145994e-01 7.09782302e-01 6.39961302e-01
-7.03765035e-01 -1.41969144e-01 -3.92928779e-01 -6.85318589e-01
-3.57078724e-02 6.01625741e-01 -6.45571470e-01 2.70428509e-01
5.23835063e-01 5.42803049e-01 -6.13666296e-01 1.03617656e+00
-9.04693007e-01 8.86129081e-01 -2.96748504e-02 -1.03226356e-01
1.54079005e-01 3.04315567e-01 7.02013493e-01 1.50543702e+00
-1.62493676e-01 4.40696850e-02 4.28952985e-02 8.00604284e-01
-2.67665505e-01 8.75496864e-02 -6.06528521e-01 -2.54532099e-01
4.87611562e-01 7.90920556e-01 -3.06897312e-01 -3.01839650e-01
-5.51264405e-01 4.46678489e-01 5.59247196e-01 2.87959933e-01
-6.03054523e-01 -3.18470776e-01 7.50159621e-01 -4.55860347e-02
-2.48251259e-01 -1.32007405e-01 -7.14869976e-01 -1.28486967e+00
1.00975014e-01 -1.55177855e+00 7.11922586e-01 -5.94927132e-01
-1.43698955e+00 1.00025654e-01 3.12075555e-01 -3.74129266e-01
-1.30774841e-01 -6.92348659e-01 -5.82706451e-01 8.87789547e-01
-1.54991055e+00 -8.65250289e-01 -2.87287116e-01 3.24984640e-01
6.92892969e-01 5.36311977e-02 9.61282015e-01 4.94901910e-02
-2.84563243e-01 7.04803705e-01 -2.59466290e-01 -5.15697896e-02
6.27791703e-01 -1.56207252e+00 6.75455689e-01 9.44322109e-01
5.01023829e-01 1.06697583e+00 7.97323883e-01 -4.45533901e-01
-1.21420729e+00 -7.23151386e-01 1.46178210e+00 -1.06172538e+00
7.31669664e-01 -4.28948551e-01 -1.13347769e+00 1.02544498e+00
1.67027801e-01 -7.57449031e-01 7.61852503e-01 7.68192530e-01
-7.91196048e-01 1.34903699e-01 -7.88326144e-01 6.17477596e-01
1.24021208e+00 -8.19348931e-01 -1.23497796e+00 4.06130344e-01
8.96664619e-01 -6.08926594e-01 -4.65450495e-01 4.45899576e-01
5.43580413e-01 -8.15186501e-01 7.99353957e-01 -1.23813796e+00
6.97180033e-01 2.26328313e-01 -7.19470829e-02 -1.36718869e+00
-9.60765108e-02 -7.25607336e-01 -1.85870882e-02 1.25549030e+00
8.58621895e-01 -4.02362883e-01 7.14171469e-01 1.01584077e+00
-1.54553965e-01 -7.71993041e-01 -5.79820991e-01 -7.27671444e-01
6.95236087e-01 -9.31595147e-01 5.37606955e-01 8.58181119e-01
2.88001537e-01 6.70314848e-01 -1.39003694e-01 3.98774669e-02
4.34963107e-01 -7.03366473e-02 7.30534077e-01 -1.14513123e+00
-3.20488781e-01 -5.42809784e-01 -1.93023831e-02 -1.35894859e+00
3.12410235e-01 -1.15291178e+00 1.44638382e-02 -1.63920105e+00
3.80852073e-01 -5.06491125e-01 -4.34004992e-01 4.31887746e-01
-2.56768525e-01 -6.00577056e-01 1.19080089e-01 -3.15503299e-01
-5.17720103e-01 -5.77202775e-02 1.23191619e+00 -2.44772062e-01
1.25280470e-01 2.06265062e-01 -1.26436293e+00 8.95998001e-01
8.94899547e-01 -5.68235695e-01 -8.46762121e-01 -1.06525302e+00
8.58506262e-01 -3.30730379e-01 3.29195827e-01 -6.92946911e-01
2.82803923e-01 -2.69203931e-01 -1.45280778e-01 -5.50933480e-01
-1.16608493e-01 -5.30721247e-01 -4.77600902e-01 3.75463724e-01
-1.11590409e+00 3.85172576e-01 3.05817693e-01 3.97013098e-01
-2.17540532e-01 -4.50418502e-01 4.24255848e-01 -2.76264131e-01
-8.73218060e-01 -9.97510031e-02 -2.47774437e-01 9.63556528e-01
6.44542813e-01 2.06431732e-01 -7.30700970e-01 -3.02404881e-01
-2.73209006e-01 6.21404111e-01 1.48090616e-01 8.11704636e-01
4.49184954e-01 -5.04967928e-01 -8.33378792e-01 3.01590208e-02
3.10881704e-01 9.84321684e-02 1.03615493e-01 5.42363822e-01
-2.77955323e-01 7.63705134e-01 -5.11433231e-03 -2.78683454e-01
-1.36221027e+00 3.60132396e-01 2.82519042e-01 -5.76196313e-01
-5.66675484e-01 1.28714013e+00 1.30455688e-01 -6.45894706e-01
7.30036855e-01 -6.40407205e-01 -2.44148612e-01 -5.49733415e-02
8.63899827e-01 4.24265891e-01 3.48434150e-01 4.67158528e-03
-3.43330681e-01 2.29717806e-01 -3.96486044e-01 6.11019544e-02
1.21746457e+00 5.32540753e-02 -2.59049743e-01 6.20738745e-01
8.74343574e-01 1.39932945e-01 -7.32284129e-01 -1.18508667e-01
5.76236725e-01 -2.10499808e-01 9.54482611e-03 -1.48419166e+00
-3.83259118e-01 1.29084659e+00 5.03283143e-02 -7.62172267e-02
8.12982976e-01 1.29086033e-01 7.47046351e-01 7.93862283e-01
4.48529631e-01 -1.13755548e+00 -1.91135511e-01 1.03734255e+00
1.02373564e+00 -1.11485696e+00 6.85100406e-02 -6.33973181e-01
-3.57587993e-01 9.69109774e-01 8.59649062e-01 3.53373408e-01
3.51394564e-01 4.61446732e-01 1.60540447e-01 -4.13223565e-01
-1.23679900e+00 -7.46758655e-02 3.32105488e-01 6.18977547e-01
7.55683005e-01 1.08145699e-01 -3.58608216e-01 7.59784460e-01
-5.26929557e-01 -7.97902048e-02 5.30567348e-01 9.37128007e-01
-4.85714793e-01 -1.37211728e+00 -9.06777009e-02 6.10888243e-01
-5.97060978e-01 -5.36536396e-01 -7.68370092e-01 9.24414217e-01
-6.26099408e-02 1.02627754e+00 -5.78158721e-02 -3.43066990e-01
3.97026360e-01 2.18013734e-01 6.93633735e-01 -9.90714908e-01
-9.23439622e-01 -9.14619029e-01 5.39615214e-01 -6.33974135e-01
-8.32281262e-02 -4.44583118e-01 -1.03443301e+00 -2.85528660e-01
-4.16496426e-01 1.25426993e-01 3.49173307e-01 1.20513821e+00
1.44819573e-01 4.68084484e-01 -1.12694487e-01 1.38612419e-01
-9.75400567e-01 -8.19719017e-01 -2.17868567e-01 2.75191069e-01
3.35631430e-01 -3.79137546e-01 -4.49520409e-01 -2.90571362e-01] | [10.829687118530273, 8.147252082824707] |
cf65d83b-9166-408e-9ee8-429a23b8614d | compression-phase-is-not-necessary-for | 2102.07402 | null | https://arxiv.org/abs/2102.07402v2 | https://arxiv.org/pdf/2102.07402v2.pdf | Information flows of diverse autoencoders | The outstanding performance of deep learning in various fields has been a fundamental query, which can be potentially examined using information theory that interprets the learning process as the transmission and compression of information. Information plane analyses of the mutual information between the input-hidden-output layers demonstrated two distinct learning phases of fitting and compression. It is debatable if the compression phase is necessary to generalize the input-output relations extracted from training data. In this study, we investigated this through experiments with various species of autoencoders and evaluated their information processing phase with an accurate kernel-based estimator of mutual information. Given sufficient training data, vanilla autoencoders demonstrated the compression phase, which was amplified after imposing sparsity regularization for hidden activities. However, we found that the compression phase is not universally observed in different species of autoencoders, including variational autoencoders, that have special constraints on network weights or manifold of hidden space. These types of autoencoders exhibited perfect generalization ability for test data without requiring the compression phase. Thus, we conclude that the compression phase is not necessary for generalization in representation learning. | ['Junghyo Jo', 'Sungyeop Lee'] | 2021-02-15 | null | null | null | null | ['information-plane'] | ['methodology'] | [ 1.99244186e-01 4.28812593e-01 2.42264643e-01 -1.92934170e-01
3.23725194e-02 -3.35218817e-01 6.55016243e-01 6.90386519e-02
-5.73630512e-01 4.98305768e-01 -1.02119200e-01 -4.81693000e-01
-5.67608833e-01 -8.59360695e-01 -7.41376579e-01 -1.00823522e+00
-3.23029369e-01 3.41158926e-01 1.32670328e-01 -9.47466940e-02
1.39521569e-01 4.89690393e-01 -1.67738068e+00 2.69067019e-01
6.22250915e-01 9.27033603e-01 4.14531499e-01 8.10843408e-01
-3.27327624e-02 8.20061326e-01 -4.18551356e-01 -4.19538200e-01
2.37982124e-01 -7.26078928e-01 -7.49204338e-01 -1.44477800e-01
2.67770682e-02 -3.29280853e-01 -6.91860080e-01 9.52322245e-01
1.17781989e-01 1.86857224e-01 1.18260109e+00 -1.14237237e+00
-7.82670736e-01 7.23162293e-01 3.59765477e-02 4.35077101e-01
-2.27986142e-01 -1.46165356e-01 1.35557437e+00 -7.23306417e-01
4.27200943e-01 7.78052509e-01 4.03268427e-01 5.58670342e-01
-1.48955524e+00 -4.54075336e-01 -4.08008426e-01 2.31750384e-01
-1.39101481e+00 -2.22883925e-01 9.33214843e-01 -6.00307405e-01
9.35012937e-01 2.41550371e-01 1.00397921e+00 9.24794555e-01
3.28410059e-01 6.39481962e-01 8.76680076e-01 -5.43391764e-01
4.12680328e-01 6.42729700e-01 7.91882426e-02 6.19739711e-01
5.41288614e-01 3.27166408e-01 -5.83760679e-01 2.28078842e-01
9.66975033e-01 -1.44472271e-01 -3.53794515e-01 -3.38762194e-01
-7.20011353e-01 1.07913458e+00 4.54648048e-01 6.45341337e-01
-2.30663761e-01 -3.06611024e-02 1.38398007e-01 7.63110518e-01
1.84937805e-01 4.77553278e-01 -4.06277388e-01 8.96159410e-02
-1.02206218e+00 -8.66314173e-02 9.20495093e-01 7.82070160e-01
8.04154873e-01 2.39726946e-01 4.43080664e-01 4.21413928e-01
4.44777369e-01 3.11530501e-01 7.38213956e-01 -1.04364610e+00
2.12730318e-01 6.54292643e-01 -4.54467714e-01 -9.54542935e-01
-3.69211286e-01 -7.90588677e-01 -1.04579198e+00 2.66515553e-01
5.08642018e-01 -1.53556570e-01 -4.29865748e-01 1.75911868e+00
-7.23893940e-02 -1.99617773e-01 5.14944077e-01 7.45894670e-01
5.79042733e-01 4.57116574e-01 -1.56167477e-01 -4.34678584e-01
1.12071586e+00 -2.16756269e-01 -9.04666007e-01 -2.35151220e-02
8.09218764e-01 -3.98495704e-01 7.34028339e-01 3.72033983e-01
-1.30544376e+00 -5.68094015e-01 -1.51893306e+00 -7.54711330e-02
-4.47974950e-01 -2.60582864e-01 7.09788561e-01 6.99602723e-01
-1.07278764e+00 1.25060344e+00 -1.13008392e+00 -8.74491930e-02
3.53401423e-01 3.73454481e-01 -5.29060543e-01 3.07715893e-01
-1.16739535e+00 7.85812914e-01 6.58726335e-01 -8.76693130e-02
-6.57902122e-01 -5.38731277e-01 -6.84803843e-01 3.70208532e-01
-3.37757885e-01 -5.93256474e-01 8.78686368e-01 -1.07884741e+00
-1.27429008e+00 7.09455013e-01 -4.73484471e-02 -6.39622450e-01
1.10992484e-01 -5.25233634e-02 -2.65308887e-01 4.24995899e-01
-5.61013699e-01 4.83857512e-01 8.79538596e-01 -1.20871770e+00
-6.49985299e-02 -5.81491053e-01 -3.20346475e-01 1.10496745e-01
-5.91542244e-01 -7.06610918e-01 2.41708651e-01 -3.38931799e-01
6.20899200e-01 -6.85691118e-01 1.90272585e-01 -8.19253325e-02
3.02437730e-02 -1.86351895e-01 6.94661140e-01 -4.54762131e-01
1.05027628e+00 -2.23232269e+00 3.44351321e-01 4.05288965e-01
3.76146525e-01 3.22267367e-03 6.93745688e-02 5.45972347e-01
-2.58883715e-01 1.88473240e-01 -1.69228002e-01 -1.87890291e-01
-1.76680600e-03 5.40090442e-01 -4.86568809e-01 6.36938870e-01
2.48334572e-01 7.09471405e-01 -5.32568455e-01 -4.23592657e-01
-1.42803499e-02 7.87710726e-01 -9.94145870e-01 2.77966410e-01
6.31847279e-03 3.14640820e-01 -1.82839796e-01 3.38174179e-02
3.16842973e-01 -4.57415134e-01 2.84429520e-01 -1.74387366e-01
1.11949891e-02 2.75707424e-01 -9.29885328e-01 1.51487398e+00
-1.71685442e-01 1.34089339e+00 -1.32743150e-01 -1.48554337e+00
7.79454648e-01 5.67872763e-01 5.68464577e-01 -7.05216825e-01
3.54222953e-01 7.19687194e-02 6.00299120e-01 -4.30140227e-01
4.05792296e-01 -2.32128695e-01 3.75799775e-01 5.84727407e-01
8.22365522e-01 -6.81338459e-02 2.39435676e-02 2.48853654e-01
9.07163441e-01 -1.38761386e-01 -6.40988350e-04 -4.70351398e-01
1.14210494e-01 -3.22228551e-01 1.52276859e-01 6.33128107e-01
4.86151874e-02 5.67492545e-01 5.87832928e-01 -1.38189450e-01
-1.42261541e+00 -1.15309966e+00 -6.12584054e-01 9.38362658e-01
-1.21413387e-01 -5.39675653e-01 -8.36586118e-01 -5.27480990e-02
-3.03412616e-01 8.71706307e-01 -6.31326854e-01 -5.62778533e-01
-1.43013075e-01 -6.05238497e-01 4.97892350e-01 4.24435735e-01
4.93205607e-01 -7.67036855e-01 -7.84848273e-01 8.98945034e-02
8.68415460e-02 -9.00797188e-01 2.18436286e-01 6.75008416e-01
-1.19157040e+00 -1.08884716e+00 -5.90793133e-01 -6.02740347e-01
6.40300930e-01 -1.02847651e-01 8.26750755e-01 1.28637820e-01
-1.67376727e-01 5.58602095e-01 -1.46079049e-01 -3.43372494e-01
-6.09145939e-01 -3.78048345e-02 2.11129397e-01 -2.88210839e-01
3.63449723e-01 -1.12410271e+00 -7.07117975e-01 3.92158031e-01
-1.13310671e+00 -1.52236342e-01 7.08961189e-01 9.08066213e-01
3.51250798e-01 3.47988397e-01 2.26905227e-01 -4.06926811e-01
4.99295324e-01 -5.03176093e-01 -4.19150323e-01 -3.34538184e-02
-7.08838820e-01 6.37332976e-01 5.63935757e-01 -3.36621016e-01
-9.04711485e-01 -1.84123456e-01 -2.58470923e-01 -3.85524571e-01
-9.85724851e-02 4.56326425e-01 6.28293119e-03 1.03893541e-01
8.35252941e-01 5.50432801e-01 1.69586107e-01 -3.60316247e-01
1.17407382e-01 4.50325310e-01 2.96925694e-01 -3.27679157e-01
5.71601808e-01 4.45644438e-01 1.60768315e-01 -1.27900982e+00
-5.56161106e-01 -2.96504110e-01 -8.04375410e-01 -2.05823556e-01
1.02169514e+00 -6.42234683e-01 -9.47380841e-01 1.38805389e-01
-1.08267033e+00 -2.78034389e-01 -5.79782367e-01 8.93778324e-01
-8.38858187e-01 1.02744639e-01 -6.56070650e-01 -8.45591187e-01
-1.25685439e-01 -8.98179233e-01 4.62265491e-01 1.29947886e-01
-1.29375786e-01 -1.39473438e+00 1.23828217e-01 3.95374605e-03
4.38880205e-01 -2.48578683e-01 1.06585193e+00 -1.06518865e+00
-4.46931988e-01 -4.38863523e-02 1.45310059e-01 5.63267767e-01
-1.00258224e-01 6.34657964e-03 -1.30686009e+00 -2.57550359e-01
4.63915139e-01 -3.08142990e-01 1.13452482e+00 3.77450466e-01
1.09420598e+00 -6.37211621e-01 -7.50612002e-03 8.16470981e-01
1.35836053e+00 2.79719353e-01 7.65896857e-01 -3.00517343e-02
2.49951348e-01 6.67871237e-01 -3.87834966e-01 2.29743123e-01
-7.91063234e-02 1.93681628e-01 2.90852517e-01 2.08225206e-01
1.92080930e-01 -4.59856689e-01 3.00684541e-01 1.44296348e+00
-5.28928697e-01 -2.67594159e-01 -5.89364767e-01 2.32284531e-01
-1.32233751e+00 -1.20888400e+00 2.43388936e-02 2.17656350e+00
8.38532805e-01 2.53123939e-01 -2.08455890e-01 5.65445721e-01
2.71283090e-01 -9.10502952e-03 -5.03829777e-01 -2.72504956e-01
-2.09396973e-01 2.57078260e-01 3.57012212e-01 4.83427972e-01
-6.89669967e-01 4.38230038e-01 7.15601397e+00 5.47397316e-01
-1.01559603e+00 -3.43704075e-02 2.68444121e-01 3.92463841e-02
-4.40778971e-01 -9.49450135e-02 -6.10859394e-01 2.61741638e-01
1.25101101e+00 -1.67194735e-02 4.29327428e-01 7.43090272e-01
-3.54871362e-01 -4.14131172e-02 -1.36605978e+00 7.32612789e-01
-2.06145689e-01 -1.17871869e+00 1.48078250e-02 3.67695719e-01
5.24977803e-01 6.76198006e-02 3.17895234e-01 1.77893400e-01
1.76195219e-01 -1.02909601e+00 2.47316763e-01 6.33248091e-01
4.22288597e-01 -5.69727957e-01 5.18338442e-01 5.81334352e-01
-7.24669635e-01 8.07577558e-03 -7.98419595e-01 -1.45918742e-01
-2.02094585e-01 4.38412011e-01 -7.36421883e-01 1.83318377e-01
4.55889195e-01 6.21191502e-01 -4.47574198e-01 7.27236927e-01
-3.13572809e-02 8.13642561e-01 -4.81943309e-01 3.65117155e-02
-6.25998601e-02 -4.29052949e-01 5.65132380e-01 8.68434310e-01
3.77613783e-01 1.44685954e-01 -6.42008901e-01 1.19242215e+00
1.24987759e-01 -1.10880263e-01 -8.00117314e-01 -5.17365634e-01
1.57981753e-01 9.65986431e-01 -6.63170159e-01 -1.17268309e-01
-3.46479803e-01 8.70935321e-01 3.25482100e-01 5.52758217e-01
-3.91261905e-01 -2.05917686e-01 4.44986314e-01 8.98211747e-02
4.42824900e-01 -3.79793584e-01 -3.25002909e-01 -1.15541601e+00
-2.17712432e-01 -2.87179381e-01 2.58255601e-01 -5.00440240e-01
-8.09380293e-01 5.31271875e-01 1.43304691e-01 -9.77961898e-01
-5.68487465e-01 -7.90412903e-01 -5.76386034e-01 6.70474112e-01
-9.45706546e-01 -4.89059448e-01 2.14317869e-02 7.50896394e-01
7.20294639e-02 -4.93735284e-01 9.25225139e-01 5.07882535e-02
-4.57670599e-01 7.09987462e-01 3.91440153e-01 2.44460836e-01
-1.47191677e-02 -1.28524280e+00 -2.74958670e-01 5.27414143e-01
6.40867591e-01 9.07077193e-01 1.04036951e+00 -1.77444041e-01
-1.33226025e+00 -5.67623079e-01 7.16832399e-01 -2.31862262e-01
6.01863444e-01 -3.96202713e-01 -1.39311779e+00 8.12603772e-01
4.11980599e-01 -2.91866660e-01 9.98438895e-01 1.65870175e-01
-3.59658688e-01 -5.58023714e-02 -7.81738698e-01 3.17853451e-01
8.56041074e-01 -8.06745350e-01 -9.19178784e-01 2.36691326e-01
6.36210084e-01 1.12413675e-01 -1.11827052e+00 1.38331458e-01
6.42278373e-01 -1.21539402e+00 7.60317087e-01 -7.21470296e-01
6.50928140e-01 2.80438900e-01 -4.17375952e-01 -1.23425436e+00
-3.68905723e-01 -1.54780075e-01 -3.38050663e-01 8.70460451e-01
6.21717989e-01 -5.82505524e-01 6.75250947e-01 6.30083799e-01
-1.77044749e-01 -6.48603380e-01 -8.57451022e-01 -6.58776462e-01
4.17758733e-01 -4.50127274e-01 -1.57735527e-01 8.67321491e-01
4.63776529e-01 3.66986990e-01 -4.37800512e-02 -5.35211107e-03
6.24621034e-01 -2.66521633e-01 3.38380098e-01 -1.37627757e+00
-6.68557882e-01 -5.95322430e-01 -6.15705013e-01 -9.98135388e-01
1.56633332e-01 -1.20805478e+00 -3.91756862e-01 -1.12589478e+00
1.39032630e-02 -8.00195932e-02 -3.35061997e-01 1.54298872e-01
3.85599524e-01 -1.18713863e-01 -6.76396117e-02 4.74120110e-01
-1.36557845e-02 6.34462833e-01 1.25915945e+00 1.04605727e-01
-2.59121031e-01 1.44235089e-01 -4.64836359e-01 7.72635162e-01
9.16346788e-01 -3.17018718e-01 -7.90055156e-01 -2.94872731e-01
5.79077780e-01 1.13327190e-01 4.06586260e-01 -1.23007703e+00
4.56982374e-01 1.76594958e-01 6.78536415e-01 -4.58522588e-01
6.14298880e-01 -1.06412816e+00 2.36529693e-01 5.95546722e-01
-6.72592998e-01 -1.36355430e-01 8.23695809e-02 5.39979815e-01
-1.68165594e-01 -4.55859751e-01 7.21342146e-01 -1.18541963e-01
-2.77461678e-01 1.45448864e-01 -5.62085629e-01 -2.18214933e-02
7.97792315e-01 -1.69570476e-01 -1.04021318e-01 -3.66431087e-01
-9.40807402e-01 -3.42605680e-01 1.67117327e-01 -1.74227022e-02
8.34119081e-01 -1.21710265e+00 -4.78671044e-01 7.64242768e-01
-3.09983015e-01 -1.53844938e-01 1.80687338e-01 8.65574181e-01
-4.27453578e-01 5.14577985e-01 -3.83977234e-01 -7.20095336e-01
-8.93790841e-01 6.25118077e-01 5.00960886e-01 6.18323460e-02
-6.22530878e-01 7.10278451e-01 1.90972671e-01 1.74838513e-01
2.33527139e-01 -2.80016392e-01 -2.50273168e-01 2.31504545e-01
4.41728592e-01 1.49917305e-01 -1.39106400e-04 -4.46458429e-01
4.29368094e-02 2.17145115e-01 -1.74720302e-01 -3.07187647e-01
1.22851038e+00 1.15790054e-01 1.71152502e-01 8.40414941e-01
1.71940184e+00 -4.16562319e-01 -1.28952277e+00 -1.81890517e-01
-1.19129717e-01 2.38585658e-02 2.69785762e-01 -8.78121406e-02
-1.24567068e+00 1.19056714e+00 4.98263031e-01 6.30151212e-01
9.36877847e-01 2.82949716e-01 3.36724102e-01 7.00548649e-01
7.07752034e-02 -1.17117333e+00 1.33513853e-01 3.64651918e-01
9.26331937e-01 -1.11997545e+00 -8.93881023e-02 1.16261914e-01
-4.00912762e-01 1.13028836e+00 3.35464209e-01 -3.67410392e-01
1.17436719e+00 2.13703036e-01 -4.50898230e-01 -3.20583135e-01
-1.13542092e+00 -6.19765222e-02 5.23866475e-01 3.52527142e-01
5.97792208e-01 -6.76751807e-02 -2.08545327e-01 5.24298131e-01
-6.49776518e-01 -3.82221371e-01 2.67361164e-01 7.55463660e-01
-7.28768349e-01 -6.33537829e-01 4.24794182e-02 4.03196096e-01
-3.45092922e-01 -4.25490700e-02 -2.51832515e-01 9.62574720e-01
-1.57542944e-01 6.83942318e-01 5.98607361e-01 -6.57377839e-01
-1.31642088e-01 4.05828208e-01 6.69367850e-01 -1.34661630e-01
-9.36341211e-02 -1.74095988e-01 -4.05635566e-01 -3.76175553e-01
-4.59919661e-01 -5.57532787e-01 -1.20221591e+00 -3.39257956e-01
-4.88236278e-01 3.75741065e-01 8.67717266e-01 1.14920127e+00
-4.65872250e-02 6.60231173e-01 4.34945405e-01 -5.82555234e-01
-5.87363183e-01 -1.10098338e+00 -9.75006878e-01 4.16225910e-01
4.91495132e-01 -5.68036079e-01 -8.97329569e-01 3.08013976e-01] | [7.933018207550049, 3.5540614128112793] |
bbaa21e3-438f-4d88-9ee4-abc6060454c6 | discriminative-density-ratio-estimation | 1311.4486 | null | http://arxiv.org/abs/1311.4486v2 | http://arxiv.org/pdf/1311.4486v2.pdf | Discriminative Density-ratio Estimation | The covariate shift is a challenging problem in supervised learning that
results from the discrepancy between the training and test distributions. An
effective approach which recently drew a considerable attention in the research
community is to reweight the training samples to minimize that discrepancy. In
specific, many methods are based on developing Density-ratio (DR) estimation
techniques that apply to both regression and classification problems. Although
these methods work well for regression problems, their performance on
classification problems is not satisfactory. This is due to a key observation
that these methods focus on matching the sample marginal distributions without
paying attention to preserving the separation between classes in the reweighted
space. In this paper, we propose a novel method for Discriminative
Density-ratio (DDR) estimation that addresses the aforementioned problem and
aims at estimating the density-ratio of joint distributions in a class-wise
manner. The proposed algorithm is an iterative procedure that alternates
between estimating the class information for the test data and estimating new
density ratio for each class. To incorporate the estimated class information of
the test data, a soft matching technique is proposed. In addition, we employ an
effective criterion which adopts mutual information as an indicator to stop the
iterative procedure while resulting in a decision boundary that lies in a
sparse region. Experiments on synthetic and benchmark datasets demonstrate the
superiority of the proposed method in terms of both accuracy and robustness. | ['Yun-Qian Miao', 'Mohamed S. Kamel', 'Ahmed K. Farahat'] | 2013-11-18 | null | null | null | null | ['density-ratio-estimation'] | ['methodology'] | [ 4.45597887e-01 -2.74948865e-01 -3.45542550e-01 -4.98075336e-01
-7.44132876e-01 -3.40705849e-02 4.61177409e-01 2.51150697e-01
-4.10986394e-01 8.86808574e-01 -1.07542090e-01 -2.64360663e-02
-4.11939591e-01 -7.78166354e-01 -2.68787801e-01 -9.85077679e-01
3.87048811e-01 3.25322598e-01 2.10570514e-01 2.90597349e-01
4.26402897e-01 4.58915353e-01 -1.57976556e+00 -3.00144047e-01
1.29721117e+00 9.60768223e-01 2.20172256e-01 1.81051031e-01
-1.18445471e-01 5.14262557e-01 -5.60708225e-01 -1.57937616e-01
2.10874557e-01 -6.22298479e-01 -2.17950031e-01 3.03013980e-01
1.13769151e-01 -6.80991709e-02 -1.91966835e-02 1.24124908e+00
3.91837716e-01 3.15933526e-01 1.22213864e+00 -1.19209015e+00
-1.75381154e-01 1.85551077e-01 -1.18779218e+00 2.05972359e-01
8.48646164e-02 -4.27804470e-01 7.57860065e-01 -8.81619573e-01
-8.70637894e-02 9.39500928e-01 5.23478329e-01 7.09322542e-02
-1.31089795e+00 -8.09150934e-01 -6.78215027e-02 3.57573152e-01
-1.80282187e+00 -1.89891353e-01 1.03230155e+00 -5.58472812e-01
2.41487488e-01 1.49862796e-01 3.60521466e-01 4.30978090e-01
2.06293821e-01 6.55691087e-01 1.18281877e+00 -5.14897168e-01
3.82093042e-01 5.12731850e-01 1.32405356e-01 3.12355220e-01
4.28356439e-01 -1.87876090e-01 -2.49040812e-01 -2.85249233e-01
5.70990384e-01 1.49926975e-01 -4.71655846e-01 -6.07572973e-01
-6.98483765e-01 8.18179905e-01 1.46064013e-01 5.51567256e-01
-3.08458537e-01 -4.30370271e-01 2.00391069e-01 -1.48269087e-02
6.83973193e-01 -1.49264604e-01 -1.47316918e-01 1.41416520e-01
-1.24198568e+00 1.96791083e-01 6.16870522e-01 5.46699107e-01
8.61447930e-01 -6.50150031e-02 -1.16900772e-01 1.20065713e+00
5.37611842e-01 3.94731373e-01 6.47869647e-01 -3.06210697e-01
4.86918598e-01 7.03097463e-01 1.94574278e-02 -1.36334348e+00
-4.93553653e-02 -6.45462751e-01 -1.10308349e+00 2.17132136e-01
5.86379409e-01 3.63284275e-02 -6.71674490e-01 1.69818318e+00
6.41794562e-01 3.64796638e-01 -2.71748193e-02 6.46449387e-01
3.37591350e-01 6.46963418e-01 1.19855873e-01 -4.47478950e-01
1.07008898e+00 -4.76903468e-01 -6.98861182e-01 -1.08904667e-01
3.96264285e-01 -8.23902130e-01 7.08214700e-01 3.62271518e-01
-7.64486313e-01 -6.71175182e-01 -1.25664294e+00 4.06816602e-01
-9.86493826e-02 4.17304754e-01 1.55590817e-01 5.94904959e-01
-4.64767963e-01 5.33981144e-01 -5.86227596e-01 -2.00097322e-01
2.65665174e-01 3.48645031e-01 -2.84797251e-01 -1.28521591e-01
-9.09739494e-01 6.99948549e-01 4.62041259e-01 1.17935330e-01
-2.49644309e-01 -5.84744513e-01 -7.84827173e-01 1.44057065e-01
2.01273665e-01 -3.04782301e-01 8.59663725e-01 -1.00707209e+00
-1.40447724e+00 5.22665739e-01 -1.84696570e-01 -1.81269303e-01
5.53678453e-01 1.38097435e-01 -2.80838519e-01 -4.09199335e-02
1.36744529e-01 1.75841913e-01 1.01648748e+00 -1.05770838e+00
-8.93433511e-01 -6.78436756e-01 -5.84070563e-01 3.77407908e-01
-4.36758101e-01 -2.89645046e-01 -3.46300900e-01 -7.76094258e-01
4.07943368e-01 -6.71165347e-01 4.31416323e-03 -2.40268230e-01
-2.67655641e-01 -3.74141723e-01 9.33212399e-01 -6.46254957e-01
1.35640526e+00 -2.32018566e+00 3.86862531e-02 4.68686849e-01
1.22204356e-01 3.03772122e-01 3.23950499e-01 1.02392778e-01
-1.37537718e-01 -3.88552099e-01 -7.49945164e-01 -1.51375607e-01
-2.46363029e-01 -1.48058146e-01 -1.37787089e-01 8.84721756e-01
2.05991581e-01 5.31427190e-02 -6.56731308e-01 -6.31291330e-01
3.69315833e-01 5.70449054e-01 -5.17648280e-01 4.83350247e-01
2.43563622e-01 5.52575052e-01 -3.98271769e-01 4.86237884e-01
1.21767294e+00 -1.66814737e-02 1.87817901e-01 -3.47288907e-01
-5.65012023e-02 -1.25196710e-01 -1.55321121e+00 1.04542291e+00
-3.31732661e-01 3.48425806e-01 1.27393883e-02 -1.51918685e+00
1.35875845e+00 2.52351046e-01 5.93328834e-01 -4.11236703e-01
2.53500074e-01 4.48549032e-01 -5.14916405e-02 -2.38499373e-01
3.77606899e-01 -5.77276170e-01 5.98927177e-02 2.39312053e-01
-1.65081441e-01 -1.29775137e-01 1.51748225e-01 -2.77004063e-01
5.52443027e-01 6.15734868e-02 8.27788413e-01 -2.66313434e-01
9.79142189e-01 -3.95840436e-01 7.21059442e-01 3.70716214e-01
-8.84420276e-02 5.29671371e-01 3.38094592e-01 1.14048719e-01
-8.50494802e-01 -1.01859915e+00 -5.87415636e-01 4.70190376e-01
2.58852333e-01 5.76251298e-02 -6.24665976e-01 -7.09334254e-01
8.38084966e-02 7.26097047e-01 -5.09389281e-01 -3.13284874e-01
-3.66335362e-01 -1.02087522e+00 1.19329207e-01 3.32518727e-01
7.71806359e-01 -4.25422013e-01 -2.27785334e-01 8.25006068e-02
-1.28509715e-01 -7.03921974e-01 -4.64958608e-01 1.00044616e-01
-1.01000476e+00 -1.18760884e+00 -1.13899863e+00 -8.74004364e-01
9.30369198e-01 4.45288360e-01 4.68764186e-01 -2.00250223e-01
-1.22118644e-01 1.20963231e-02 -2.70756185e-01 -1.28938600e-01
-2.89531171e-01 8.22009146e-02 -7.71252811e-02 4.22283858e-01
5.66364467e-01 -5.35502791e-01 -4.69717383e-01 7.20797658e-01
-9.99926865e-01 -1.44321173e-01 7.07803488e-01 7.34124362e-01
5.78207493e-01 3.93783092e-01 9.75116372e-01 -8.44172120e-01
5.91160238e-01 -8.50917697e-01 -7.08327711e-01 2.68728018e-01
-7.60596573e-01 5.48584796e-02 6.81330860e-01 -3.45484734e-01
-1.27328777e+00 -1.14626056e-02 -1.11565739e-01 -2.72330195e-01
-1.49167359e-01 5.03217697e-01 -3.42893213e-01 1.18492134e-01
3.04592252e-01 4.06618357e-01 2.35383019e-01 -4.90174711e-01
-1.20707497e-01 1.10944021e+00 4.13318008e-01 -3.77429336e-01
7.52858520e-01 2.20921353e-01 1.56405538e-01 -7.69912601e-01
-6.88019156e-01 -7.82880664e-01 -5.24969578e-01 -1.47902235e-01
6.67172313e-01 -7.17599928e-01 -5.11082888e-01 6.00115418e-01
-6.03437424e-01 1.85397670e-01 -9.29646716e-02 9.08997834e-01
-3.69627088e-01 7.16184556e-01 1.54062817e-02 -1.00823915e+00
-1.10030644e-01 -1.05040193e+00 6.84808493e-01 4.65772420e-01
-7.17507070e-03 -1.13596833e+00 1.79587409e-01 2.11021021e-01
3.45537722e-01 1.11627385e-01 9.06287193e-01 -7.39446998e-01
-9.42429304e-02 -4.66442168e-01 -4.97746259e-01 6.45512521e-01
6.55606508e-01 -1.12432092e-01 -7.05247879e-01 -3.55277359e-01
2.71093309e-01 -5.89198992e-02 7.43580461e-01 5.00924885e-01
1.07210946e+00 -3.08059305e-02 -3.72636676e-01 2.69387394e-01
1.50342822e+00 4.15897548e-01 5.76245785e-01 7.18696341e-02
4.87554997e-01 5.59830606e-01 9.83332396e-01 7.00776517e-01
1.34921327e-01 7.37824321e-01 1.50941208e-01 -2.23938916e-02
6.42040148e-02 -2.34987080e-01 1.58919260e-01 7.64003932e-01
2.72158325e-01 -1.56097069e-01 -5.80010891e-01 4.49175388e-01
-1.81991482e+00 -8.34340990e-01 -8.40518996e-02 2.74916434e+00
9.44186449e-01 2.25306489e-02 7.73273110e-02 5.40688157e-01
1.28249729e+00 -3.30405496e-02 -4.52789605e-01 7.53079448e-03
1.84138864e-01 6.85523376e-02 3.19073737e-01 4.27494347e-01
-9.45102155e-01 3.37367296e-01 5.33781385e+00 1.17521322e+00
-1.11517763e+00 -2.90500432e-01 7.49012232e-01 3.55824292e-01
-9.63235497e-02 -1.48083949e-02 -8.86752009e-01 7.22043574e-01
5.59041440e-01 -2.55145580e-01 1.45888487e-02 6.48571670e-01
2.30357125e-01 -6.01717174e-01 -8.06144774e-01 1.04170442e+00
1.71063080e-01 -6.04032993e-01 -6.27764761e-02 7.42070451e-02
6.50324583e-01 -6.63048148e-01 1.90206334e-01 3.13846767e-01
-3.21854681e-01 -7.33318269e-01 4.72535402e-01 6.50595427e-01
6.31203294e-01 -1.02925158e+00 9.98537660e-01 5.71385741e-01
-1.16152608e+00 5.59997633e-02 -4.42015737e-01 3.68010029e-02
-1.66977510e-01 1.01933849e+00 -1.03021371e+00 5.88861525e-01
3.92072439e-01 8.01243484e-01 -4.46812958e-01 1.44887245e+00
-1.94775034e-02 5.20618439e-01 -2.39694834e-01 7.15325475e-02
-1.53515920e-01 -7.02538431e-01 4.58160907e-01 9.48754013e-01
6.05916500e-01 -2.10289955e-01 2.39604451e-02 7.64733851e-01
4.09498811e-02 4.74578977e-01 -5.00390232e-01 2.37911731e-01
4.61040258e-01 1.13528407e+00 -8.06699574e-01 -2.07312852e-01
-4.86732960e-01 6.45299613e-01 2.22172916e-01 1.51162744e-01
-9.25481915e-01 -5.16393423e-01 2.31185272e-01 2.48218328e-01
4.11719054e-01 -5.96371591e-02 -2.14491457e-01 -9.60700691e-01
1.30719226e-02 -6.26976013e-01 4.19779152e-01 -1.92308456e-01
-1.42971659e+00 2.78474271e-01 3.74212325e-01 -1.45450985e+00
-2.32270837e-01 -2.72484183e-01 -5.30246675e-01 9.11855996e-01
-1.64138997e+00 -6.09300077e-01 -4.24691707e-01 4.90516007e-01
4.50259835e-01 -1.25464901e-01 4.43880141e-01 6.60627127e-01
-7.89747357e-01 7.12529004e-01 4.53412920e-01 -1.41744494e-01
9.33346510e-01 -1.11405885e+00 -4.35770035e-01 6.80727005e-01
-2.86911190e-01 4.69738930e-01 8.07349145e-01 -6.72862828e-01
-6.67316020e-01 -9.61495280e-01 7.36581862e-01 3.17969918e-01
3.75291288e-01 6.63768724e-02 -1.08094680e+00 2.66227961e-01
-1.56537637e-01 -3.15310881e-02 9.08863902e-01 -1.37981951e-01
2.15355563e-03 -4.41380829e-01 -1.48265851e+00 2.80497342e-01
2.83126563e-01 -1.42205268e-01 -5.70113480e-01 -1.40519980e-02
-7.24029690e-02 -1.07327066e-01 -8.05930495e-01 6.61506474e-01
4.50606734e-01 -9.88206387e-01 6.18498385e-01 1.47417679e-01
1.28028899e-01 -6.79202139e-01 -1.88283861e-01 -1.23389411e+00
-1.09396935e-01 2.61644442e-02 3.80446054e-02 1.51798224e+00
1.62271112e-01 -8.08605552e-01 8.11994791e-01 5.11941969e-01
2.20314756e-01 -8.22133005e-01 -9.98828769e-01 -5.92045844e-01
-2.36275762e-01 -1.33687750e-01 3.14569443e-01 7.16317177e-01
-1.44555345e-01 3.67884248e-01 -3.55749518e-01 -3.07495110e-02
8.14790070e-01 1.17279366e-01 6.94136858e-01 -1.34538329e+00
-2.76202708e-01 -2.65863359e-01 -5.41064382e-01 -9.36730683e-01
2.30980903e-01 -7.77574718e-01 1.46842673e-01 -1.27862012e+00
4.83784139e-01 -6.33717895e-01 -4.10837293e-01 3.55698727e-02
-4.17382061e-01 2.90612102e-01 -1.97882891e-01 3.54902655e-01
-2.72371203e-01 8.24154615e-01 9.95522738e-01 -3.21346372e-02
-3.12197655e-01 5.01574457e-01 -7.54259288e-01 6.42224550e-01
8.99536610e-01 -5.51953375e-01 -6.19739115e-01 8.47339109e-02
-2.48475373e-01 8.87095854e-02 1.98116779e-01 -1.23790598e+00
1.99541047e-01 -9.36885774e-02 4.41241205e-01 -7.63229072e-01
9.98128802e-02 -9.82611179e-01 7.33137131e-02 2.22987711e-01
-2.13592216e-01 -2.56610990e-01 1.38214894e-03 8.15849245e-01
-5.19731104e-01 -5.40518999e-01 1.31469810e+00 2.91392237e-01
-3.30031723e-01 9.85479206e-02 -3.01691741e-01 -1.15949586e-01
1.24680138e+00 -4.28177625e-01 1.36631727e-01 -1.71441406e-01
-4.22691882e-01 -2.14078072e-02 2.89010495e-01 2.87164543e-02
6.08935595e-01 -1.30458856e+00 -8.06499422e-01 2.75258511e-01
1.23953886e-01 -6.86571151e-02 2.51376271e-01 9.54038560e-01
-2.71349460e-01 3.48825723e-01 -7.97333270e-02 -6.18220270e-01
-1.18195271e+00 2.40279406e-01 2.00722188e-01 -3.85171026e-01
-2.02896178e-01 5.04300892e-01 2.41102904e-01 -2.09731475e-01
1.78558469e-01 -1.02675162e-01 -5.78861713e-01 2.84928560e-01
4.95453060e-01 5.71595490e-01 1.07487515e-01 -6.55844331e-01
-3.35104674e-01 6.86267495e-01 -2.17287242e-01 6.35477677e-02
1.22667778e+00 -1.52236938e-01 -9.41063091e-02 5.08126795e-01
1.44409275e+00 1.82725564e-01 -1.22846532e+00 -4.14149493e-01
2.41363719e-02 -8.18160474e-01 1.51364133e-01 -3.81642371e-01
-9.44087088e-01 6.42398417e-01 7.50622690e-01 1.89179420e-01
1.26726472e+00 -2.02424929e-01 5.80718875e-01 -1.82836041e-01
3.92222740e-02 -1.16975200e+00 3.42391757e-03 1.38624266e-01
5.68138719e-01 -1.18502057e+00 2.53914654e-01 -4.59449917e-01
-3.52288693e-01 1.09183156e+00 6.20711923e-01 -2.39234403e-01
8.63529146e-01 -4.56507653e-02 -3.81305844e-01 2.53535688e-01
-1.81802735e-01 -7.32725486e-02 4.47558969e-01 3.33333135e-01
6.29502058e-01 -8.19429383e-02 -7.46312737e-01 4.35030192e-01
2.27089837e-01 5.37218750e-02 2.28081495e-01 8.71060073e-01
-5.99214137e-01 -1.21746230e+00 -5.92114806e-01 6.03784919e-01
-4.36588556e-01 2.66534388e-01 9.78219602e-03 6.47572041e-01
1.38061315e-01 9.79835987e-01 1.10641614e-01 -2.12882608e-01
1.73830554e-01 -8.14621374e-02 4.39297497e-01 -5.28896034e-01
6.27733255e-03 1.61766976e-01 -3.70679349e-01 1.08392037e-01
-5.77055275e-01 -7.67515361e-01 -1.08861506e+00 -6.15934730e-02
-6.89141989e-01 4.94992614e-01 6.67732894e-01 9.50861871e-01
-9.58349556e-02 3.82206202e-01 1.05009258e+00 -6.28413260e-01
-7.19695807e-01 -9.52375114e-01 -9.64344740e-01 2.42200151e-01
1.64532945e-01 -9.63343680e-01 -7.20689118e-01 -3.01878452e-01] | [8.775723457336426, 3.994135856628418] |
7d8321fb-78a3-47ee-bfa7-60a89e63fe34 | contrastive-learning-mri-reconstruction | 2306.00530 | null | https://arxiv.org/abs/2306.00530v1 | https://arxiv.org/pdf/2306.00530v1.pdf | Contrastive Learning MRI Reconstruction | Purpose: We propose a novel contrastive learning latent space representation for MRI datasets with partially acquired scans. We show that this latent space can be utilized for accelerated MR image reconstruction. Theory and Methods: Our novel framework, referred to as COLADA (stands for Contrastive Learning for highly accelerated MR image reconstruction), maximizes the mutual information between differently accelerated images of an MRI scan by using self-supervised contrastive learning. In other words, it attempts to "pull" the latent representations of the same scan together and "push" the latent representations of other scans away. The generated MRI latent space is subsequently utilized for MR image reconstruction and the performance was assessed in comparison to several baseline deep learning reconstruction methods. Furthermore, the quality of the proposed latent space representation was analyzed using Alignment and Uniformity. Results: COLADA comprehensively outperformed other reconstruction methods with robustness to variations in undersampling patterns, pathological abnormalities, and noise in k-space during inference. COLADA proved the high quality of reconstruction on unseen data with minimal fine-tuning. The analysis of representation quality suggests that the contrastive features produced by COLADA are optimally distributed in latent space. Conclusion: To the best of our knowledge, this is the first attempt to utilize contrastive learning on differently accelerated images for MR image reconstruction. The proposed latent space representation has practical usage due to a large number of existing partially sampled datasets. This implies the possibility of exploring self-supervised contrastive learning further to enhance the latent space of MRI for image reconstruction. | ['Zhaolin Chen', 'Mehrtash Harandi', 'Gary Egan', 'Zhifeng Chen', 'Mevan Ekanayake'] | 2023-06-01 | null | null | null | null | ['image-reconstruction', 'mri-reconstruction'] | ['computer-vision', 'computer-vision'] | [ 4.94443983e-01 1.02058955e-01 -2.25791886e-01 -2.70000428e-01
-1.02757394e+00 -1.66370928e-01 5.94379425e-01 -1.02327831e-01
-4.30090994e-01 5.30772507e-01 6.90577626e-01 -7.16295987e-02
-6.08980298e-01 -5.15639782e-01 -6.43977761e-01 -1.31316221e+00
-4.71610665e-01 5.20187438e-01 -9.58645865e-02 1.94820464e-01
1.45741269e-01 7.37423837e-01 -1.11676049e+00 3.98292631e-01
4.81797069e-01 5.03022850e-01 6.72323525e-01 4.12977904e-01
2.49402985e-01 9.20168161e-01 -1.77015841e-01 4.77661937e-01
1.65094256e-01 -3.59979689e-01 -9.56669271e-01 1.13486931e-01
1.70287982e-01 -4.67800111e-01 -4.64850515e-01 8.08056712e-01
7.28956580e-01 1.90379657e-02 7.97498763e-01 -7.03595579e-01
-3.42068791e-01 7.13651299e-01 -4.29075032e-01 7.03942716e-01
5.59819974e-02 2.62441430e-02 4.72457290e-01 -9.05183077e-01
8.88909757e-01 7.46947467e-01 5.38653910e-01 4.88933235e-01
-1.52010345e+00 -6.70192599e-01 -2.60892749e-01 1.61719844e-01
-1.02444243e+00 -3.46863419e-01 9.86055315e-01 -6.42734885e-01
7.44047225e-01 3.36439341e-01 6.05563939e-01 1.16954076e+00
6.85928047e-01 1.05748034e+00 1.68617451e+00 -4.76083755e-01
2.04617172e-01 -2.19730496e-01 2.29136586e-01 8.78776491e-01
-1.80183902e-01 2.48031974e-01 -4.30395454e-01 -3.41555566e-01
8.97639990e-01 3.12601291e-02 -2.23321125e-01 -7.85133839e-01
-1.61853290e+00 8.86953890e-01 3.69995862e-01 5.38689494e-01
-7.52253354e-01 7.22351298e-02 4.03824419e-01 1.12582766e-01
5.86073995e-01 5.51250517e-01 -7.48799667e-02 6.00867420e-02
-1.29375577e+00 -1.25069275e-01 1.22100726e-01 3.93274158e-01
1.86271846e-01 3.48586112e-01 -1.34442613e-01 9.06718493e-01
2.87070304e-01 4.27942634e-01 1.22159648e+00 -8.36779535e-01
2.42491737e-01 9.34751779e-02 -4.48810011e-01 -9.66401875e-01
-6.47808909e-01 -8.26575935e-01 -9.85119641e-01 2.29596168e-01
4.30410318e-02 3.00284415e-01 -9.59605157e-01 1.70341945e+00
9.20483172e-02 4.06398892e-01 1.43553481e-01 7.88836956e-01
5.94568133e-01 5.87103546e-01 2.35499125e-02 -6.98469937e-01
1.14048052e+00 -1.00815511e+00 -8.81583512e-01 6.30746037e-02
4.74976182e-01 -7.87454724e-01 9.71677840e-01 4.06528324e-01
-1.28649116e+00 -4.68979627e-01 -1.13319242e+00 3.96449655e-01
1.62235275e-01 6.03280924e-02 8.43004644e-01 4.99717265e-01
-1.13623548e+00 7.11957097e-01 -1.42421889e+00 1.54971793e-01
4.37647521e-01 4.37607825e-01 -5.75082719e-01 -1.56600997e-01
-1.04973400e+00 8.22278559e-01 1.68820873e-01 8.74131322e-02
-1.22656369e+00 -8.07440877e-01 -8.18042874e-01 -3.95657450e-01
-2.89844144e-02 -6.17983401e-01 7.62123048e-01 -6.43759310e-01
-1.40993464e+00 9.76002395e-01 9.74278431e-03 -5.90190768e-01
7.55532146e-01 -1.59264341e-01 -2.88193703e-01 6.11949682e-01
2.97961146e-01 6.80198193e-01 1.10458875e+00 -1.07567692e+00
3.09502572e-01 -3.90593916e-01 -4.52002764e-01 2.06708834e-01
-1.10863615e-02 -1.42261758e-01 -1.02807917e-01 -1.08587730e+00
8.83207381e-01 -1.17677069e+00 -4.51121867e-01 -2.27576956e-01
4.20872346e-02 1.84118614e-01 5.97463965e-01 -8.07741702e-01
8.64803910e-01 -2.01486516e+00 3.76044244e-01 2.14983255e-01
5.87792337e-01 -1.01545028e-01 -1.55767754e-01 8.20393339e-02
-7.87000358e-01 -2.21295193e-01 -2.98906386e-01 -1.19222783e-01
-5.42346597e-01 2.67618477e-01 -3.59572947e-01 9.89818037e-01
-2.76777864e-01 1.02861500e+00 -1.06949961e+00 -5.80817044e-01
4.18697000e-01 6.11956656e-01 -5.74308991e-01 1.84576988e-01
5.54990530e-01 8.89013648e-01 -1.57163218e-01 2.58521676e-01
6.18283093e-01 -4.33645993e-01 2.55324036e-01 -5.35353959e-01
6.24547489e-02 9.58999842e-02 -8.75516653e-01 1.93119597e+00
-5.57169616e-01 5.28216660e-01 -4.62630421e-01 -1.08858764e+00
6.33855462e-01 4.76677090e-01 1.10121107e+00 -8.66448522e-01
-1.90940157e-01 2.71477222e-01 -9.08411890e-02 -5.50180018e-01
1.42221302e-01 -4.62687910e-01 1.30571172e-01 7.11126685e-01
2.33747177e-02 2.54377779e-02 -4.14747238e-01 3.13077092e-01
9.04221475e-01 5.64441942e-02 3.10667276e-01 -4.14564312e-01
3.35173458e-01 -4.04276878e-01 1.24426723e-01 8.46398115e-01
-1.64749429e-01 6.35630250e-01 1.47092536e-01 -4.45601314e-01
-1.24438429e+00 -1.47589314e+00 -5.89922488e-01 5.72586179e-01
-1.73845604e-01 -3.08577903e-03 -6.90279782e-01 -5.87963760e-01
-5.09173512e-01 5.52496254e-01 -7.65712023e-01 -2.79398531e-01
-8.24227333e-01 -1.04171956e+00 3.95240992e-01 4.88052160e-01
2.84508169e-01 -1.01273417e+00 -6.67693019e-01 3.44547510e-01
-5.14410079e-01 -7.90044367e-01 -3.04324687e-01 2.71633714e-01
-1.54923785e+00 -7.29013801e-01 -9.82913077e-01 -5.61148226e-01
9.06448960e-01 1.89790830e-01 8.66539776e-01 -2.23660052e-01
-4.27076727e-01 6.38977885e-01 -1.94337398e-01 3.21560144e-01
-6.26514435e-01 -2.27894709e-01 3.78925711e-01 -2.15752482e-01
-3.39138776e-01 -8.14426064e-01 -8.35655391e-01 1.86049134e-01
-1.08514822e+00 3.72435808e-01 8.03043783e-01 1.26360786e+00
9.58054066e-01 1.44406930e-01 3.60437214e-01 -9.16932285e-01
3.60958636e-01 -6.01970255e-01 -2.62628615e-01 6.47892952e-02
-6.85947359e-01 5.80532074e-01 1.80988193e-01 -5.20912528e-01
-7.41313577e-01 -8.60808641e-02 -1.76590219e-01 -4.82611954e-01
9.60670859e-02 4.19993162e-01 2.92430699e-01 2.83890720e-02
5.78548729e-01 5.61123192e-01 3.45070988e-01 -4.50795949e-01
1.89445585e-01 2.77358204e-01 5.17317593e-01 -4.82736051e-01
3.97674471e-01 8.44982147e-01 2.34269783e-01 -7.07827032e-01
-6.63844883e-01 -5.25182486e-01 -8.38667035e-01 -3.81800979e-01
9.31976616e-01 -7.18027055e-01 -1.73459813e-01 4.82877910e-01
-6.08022273e-01 -2.93879032e-01 -4.84476447e-01 1.04837394e+00
-9.63933289e-01 7.47459292e-01 -7.29103625e-01 -4.54933226e-01
-4.48347479e-01 -1.74346435e+00 9.24969018e-01 -5.22272348e-01
-1.91978812e-01 -1.15061617e+00 1.91228434e-01 5.80555856e-01
3.32443267e-01 3.33918631e-01 9.42390501e-01 -3.55031163e-01
-4.78748143e-01 3.14109540e-03 1.78059757e-01 3.70842606e-01
1.74767002e-01 -6.27501309e-01 -8.81504714e-01 -6.04102373e-01
5.49497187e-01 -1.89654097e-01 8.77435625e-01 8.70175183e-01
1.26419163e+00 -3.82950366e-01 -1.24174811e-01 7.38366306e-01
1.26509857e+00 -7.12138936e-02 7.28460312e-01 4.66040820e-01
4.75644648e-01 4.81729865e-01 3.10577303e-01 1.92610994e-01
-6.73184246e-02 7.62683868e-01 1.40431404e-01 -3.99048507e-01
-4.50833142e-01 -1.80768520e-01 2.32793689e-01 1.49627852e+00
-3.56878377e-02 4.13009137e-01 -1.04424810e+00 4.31632847e-01
-1.40316999e+00 -9.39087749e-01 3.18770260e-02 2.00469398e+00
6.95892036e-01 1.40781878e-02 -2.36139759e-01 1.77128345e-01
3.89062583e-01 2.85436302e-01 -4.19093519e-01 5.04606739e-02
-9.55659598e-02 3.97354573e-01 5.83968043e-01 5.96921384e-01
-9.17977750e-01 4.52358931e-01 6.96767330e+00 9.12524700e-01
-1.32199478e+00 6.42723620e-01 6.41649365e-01 -7.10282251e-02
-4.90782470e-01 -9.22434852e-02 -2.48451233e-01 4.39424634e-01
8.80810678e-01 1.73617586e-01 2.97041446e-01 6.90981507e-01
3.30297351e-01 -2.54099723e-02 -8.49406660e-01 9.92666602e-01
2.18232766e-01 -1.57319307e+00 1.22892193e-01 1.39142841e-01
8.54377806e-01 2.68392920e-01 5.57127893e-01 7.89222643e-02
-8.10102299e-02 -1.03698611e+00 5.88229060e-01 7.56352425e-01
9.20443177e-01 -5.73324680e-01 7.55932510e-01 3.64137977e-01
-7.50814736e-01 8.29623863e-02 -1.81652576e-01 4.37995493e-01
1.52510166e-01 5.83823919e-01 -1.07111728e+00 4.59879637e-01
4.39953923e-01 4.82537895e-01 -5.16032577e-01 6.52117789e-01
-4.26200926e-02 8.06020677e-01 -2.92733461e-02 6.45360053e-01
4.57534373e-01 7.60396123e-02 8.71252298e-01 1.08209610e+00
9.62296724e-02 -1.76845002e-04 2.02003047e-01 5.65443635e-01
4.40640688e-01 3.01556915e-01 -5.78176439e-01 1.61035299e-01
-3.40712219e-02 9.76113737e-01 -1.01495922e+00 -4.54237640e-01
4.22900915e-02 9.96966243e-01 1.36138692e-01 1.72231466e-01
-7.25211918e-01 4.63231534e-01 -1.85626775e-01 3.42595994e-01
-1.23208359e-01 -3.78206581e-01 -2.68618226e-01 -1.23176014e+00
-2.05668941e-01 -7.23380148e-01 5.58946431e-01 -9.11390483e-01
-1.25679529e+00 7.75881350e-01 5.32081902e-01 -1.31886923e+00
-3.73773783e-01 -2.29549691e-01 -4.34012748e-02 6.66876376e-01
-1.39650571e+00 -1.16526258e+00 -3.89534794e-02 5.87822914e-01
7.10981965e-01 -5.63116312e-01 8.79485905e-01 1.63869277e-01
2.70797946e-02 5.22229135e-01 4.13075536e-01 -1.47955269e-01
3.88490170e-01 -1.11831903e+00 -1.60301760e-01 7.13295341e-01
1.04246929e-01 9.56078112e-01 5.94261169e-01 -8.69716644e-01
-1.16276348e+00 -7.02492774e-01 4.13518369e-01 -2.77506173e-01
5.39507806e-01 -7.55373538e-02 -9.52828526e-01 5.89876294e-01
-4.98097986e-02 1.27625778e-01 1.01926494e+00 -1.90101951e-01
-1.46134272e-01 2.99220551e-02 -1.14313698e+00 4.91692334e-01
6.83776021e-01 -6.28696740e-01 -8.52449119e-01 5.73653877e-01
5.12119174e-01 -4.67524201e-01 -1.11000526e+00 5.57289004e-01
6.45205259e-01 -6.31563306e-01 1.39745128e+00 -4.27839786e-01
5.22713304e-01 1.91135444e-02 -2.02590257e-01 -1.08738935e+00
-5.34083307e-01 -2.59229809e-01 -1.21022739e-01 4.82023835e-01
-4.70815115e-02 -5.95148861e-01 7.72294700e-01 2.72252023e-01
-2.34360620e-01 -9.09091771e-01 -1.00654614e+00 -6.49240375e-01
2.11327210e-01 -5.37965059e-01 2.57294446e-01 1.26237082e+00
-2.38584250e-01 -3.25125381e-02 -5.58891535e-01 3.19574744e-01
1.09516740e+00 6.24423996e-02 1.77984089e-01 -6.34744883e-01
-5.80666065e-01 -2.16823027e-01 -5.51463723e-01 -7.67256021e-01
2.15990886e-01 -1.60512757e+00 -2.93834388e-01 -1.38138819e+00
5.23357034e-01 -7.30997980e-01 -7.29295194e-01 3.12231958e-01
-3.44379172e-02 4.54955339e-01 5.04395142e-02 8.30810785e-01
-2.39463300e-01 3.85644287e-01 1.57156718e+00 -6.83309957e-02
-1.15219697e-01 -7.60843679e-02 -3.48526746e-01 5.89099705e-01
7.71843731e-01 -7.50140250e-01 -6.11722231e-01 -2.79576838e-01
8.23269784e-02 4.75393116e-01 4.09322321e-01 -9.89121616e-01
1.10403314e-01 1.54222280e-01 5.41214228e-01 -7.77523696e-01
1.89810574e-01 -6.65450692e-01 5.20881534e-01 8.50147188e-01
-6.58729732e-01 4.59132791e-02 2.85628177e-02 5.11071086e-01
-1.22886471e-01 -3.80185515e-01 1.10694313e+00 -2.91878998e-01
-4.42944050e-01 2.50550896e-01 -6.35592401e-01 -3.56271267e-01
8.71762335e-01 -3.15872133e-01 3.25299144e-01 -1.65383846e-01
-1.16245615e+00 -3.52517754e-01 8.75434205e-02 1.24298856e-01
9.00647998e-01 -1.49702382e+00 -7.42844045e-01 4.60103512e-01
-5.30381612e-02 -3.56199622e-01 8.60103965e-01 1.22129726e+00
-6.28553510e-01 4.22727585e-01 -5.19431591e-01 -1.13599539e+00
-1.25913751e+00 5.44552207e-01 3.14315915e-01 -8.71325970e-01
-1.32477963e+00 5.73783875e-01 3.63548249e-01 -5.06380737e-01
-1.66013479e-01 -1.43220350e-01 -3.02175015e-01 2.38440391e-02
4.79362398e-01 2.40671903e-01 7.12419376e-02 -6.82008326e-01
-2.49114037e-01 4.29926902e-01 -3.81512463e-01 -3.64270508e-01
1.76313829e+00 -1.91216290e-01 -1.63269669e-01 5.78951895e-01
1.47428155e+00 1.48683220e-01 -1.12775218e+00 -2.69549072e-01
-1.49909571e-01 -4.57669079e-01 4.42336291e-01 -6.45686746e-01
-1.08425474e+00 8.28420997e-01 1.28703928e+00 -5.07163405e-01
1.09087908e+00 7.52455071e-02 6.00794077e-01 7.61113763e-02
3.81528229e-01 -8.15418601e-01 5.98704457e-01 6.22787476e-02
9.92712975e-01 -9.29388463e-01 4.63231176e-01 9.12307724e-02
-7.10778058e-01 1.11738217e+00 -1.01128638e-01 -1.74387142e-01
6.14390016e-01 3.45455438e-01 1.79180596e-02 -5.49064457e-01
-4.14426267e-01 4.78116274e-01 3.84057909e-01 4.52494591e-01
4.27789241e-01 3.17327738e-01 -3.03955883e-01 1.51434317e-01
-2.29132667e-01 7.00967684e-02 4.67743427e-01 8.23681772e-01
-5.97691946e-02 -9.81202066e-01 -4.17297721e-01 4.52316791e-01
-4.85746384e-01 -8.51958096e-02 5.96746802e-01 5.66117942e-01
-1.41455233e-01 2.59967148e-01 -1.70057923e-01 9.22970027e-02
-1.46793962e-01 -2.28910893e-02 7.79759169e-01 -4.99232411e-01
-2.41835818e-01 4.56159264e-01 -3.12108576e-01 -5.58265805e-01
-6.48089349e-01 -8.21836174e-01 -1.28132129e+00 2.22008377e-01
-2.84869105e-01 9.51013267e-02 8.83802891e-01 8.45704734e-01
-1.11472514e-02 7.70891130e-01 7.28401005e-01 -7.69517243e-01
-5.26310802e-01 -7.15400636e-01 -5.04402518e-01 5.55835724e-01
3.05619389e-01 -8.79505396e-01 -2.95099467e-01 5.64561412e-02] | [13.536686897277832, -2.3677353858947754] |
6144c12d-6fba-4e7d-834c-904664351c3d | feedback-stability-analysis-via-dissipativity | 2209.08322 | null | https://arxiv.org/abs/2209.08322v1 | https://arxiv.org/pdf/2209.08322v1.pdf | Feedback Stability Analysis via Dissipativity with Dynamic Supply Rates | In this paper, we propose a notion of dissipativity with dynamic supply rates for nonlinear differential input-state-output equations via the use of auxiliary systems. This extends the classical dissipativity with static supply rates and miscellaneous dynamic quadratic forms. The main results of this paper concern Lyapunov (asymptotic/exponential) stability analysis for nonlinear feedback dissipative systems that are characterised by dissipation inequalities with respect to compatible dynamic supply rates. Importantly, dissipativity conditions guaranteeing partial stability of the state of the feedback systems without concerning that of the state of the auxiliary systems are provided. They are shown to recover several existing results in the literature. Comparison with the input-output approach to feedback stability analysis based on integral quadratic constraints is also made. | ['Chao Chen', 'Sei Zhen Khong'] | 2022-09-17 | null | null | null | null | ['miscellaneous'] | ['miscellaneous'] | [-2.04519048e-01 4.50364500e-01 1.42571777e-01 4.25098211e-01
4.05463483e-03 -8.76371861e-01 2.36240417e-01 5.28287366e-02
-3.14552099e-01 1.11358380e+00 -1.74548551e-01 -2.68491954e-01
-5.39001048e-01 -2.69755751e-01 -4.32650805e-01 -1.04172122e+00
1.02272816e-01 -3.89778972e-01 1.86645433e-01 -9.83114898e-01
-2.80689180e-01 6.16257668e-01 -9.01486874e-01 -6.23342276e-01
7.82326818e-01 1.00278842e+00 -2.97059059e-01 9.90182638e-01
1.02908921e+00 5.02277672e-01 -2.13719338e-01 3.40586722e-01
3.42504084e-01 -5.22441626e-01 -4.21523750e-01 8.11917335e-02
-9.45049748e-02 -2.98217982e-01 -5.24367332e-01 1.09227943e+00
5.59446514e-01 4.57291365e-01 7.78355122e-01 -1.33811033e+00
-9.86258507e-01 1.68973118e-01 -3.62557620e-02 3.22486401e-01
-1.83213856e-02 2.21166760e-01 5.53136170e-01 -8.35870206e-01
5.63800931e-01 1.05995762e+00 6.79838121e-01 7.43728280e-01
-1.58722794e+00 -2.21361861e-01 1.13978699e-01 -1.13637201e-01
-1.48786509e+00 -3.80917817e-01 4.36933368e-01 -8.13897729e-01
7.74938583e-01 7.42936194e-01 6.96233988e-01 3.45347822e-01
7.75806546e-01 2.08432645e-01 8.62334132e-01 -4.19856682e-02
1.54206172e-01 4.20061707e-01 5.27052701e-01 8.25857043e-01
6.18271053e-01 2.31665581e-01 1.22286901e-01 -2.82974571e-01
8.70876193e-01 -4.65191722e-01 -6.29187226e-01 -5.57508707e-01
-3.86036873e-01 6.96220219e-01 4.57486331e-01 3.78557831e-01
-2.33381227e-01 2.53349096e-01 4.01536107e-01 8.10220122e-01
5.66357493e-01 1.63945556e-01 -1.28523692e-01 4.40613270e-01
-1.22255236e-01 4.22890544e-01 1.04549861e+00 1.18671238e+00
4.78772491e-01 6.63524449e-01 -1.85351282e-01 -1.32442487e-03
5.55248439e-01 9.08963799e-01 1.91072002e-02 -6.92003489e-01
2.85237357e-02 6.82376981e-01 4.50730294e-01 -8.40772569e-01
-5.22808373e-01 -1.98183760e-01 -1.03735697e+00 2.75144249e-01
2.42990911e-01 -7.12553918e-01 -2.36526206e-01 1.78769350e+00
3.46662223e-01 -4.05227005e-01 1.58763006e-01 1.12339807e+00
-5.93326911e-02 7.93171227e-01 -3.62233609e-01 -1.00310469e+00
4.95645881e-01 -3.40417586e-02 -1.31007087e+00 4.62389261e-01
4.40728873e-01 -4.25461203e-01 7.11767972e-01 -3.29169072e-02
-1.23081219e+00 -3.48879769e-02 -1.03366590e+00 4.27037105e-02
-1.81616828e-01 1.65953159e-01 -3.04890424e-01 3.94855320e-01
-1.45658326e+00 4.61221486e-01 -8.29447448e-01 -6.08862331e-03
-6.31630778e-01 5.60070336e-01 -2.02137455e-01 1.00534880e+00
-1.35967827e+00 1.21347392e+00 1.40322790e-01 8.23530316e-01
-4.42926109e-01 -1.06182015e+00 -6.70420408e-01 -2.69236684e-01
3.24625134e-01 -5.84118903e-01 9.75551307e-01 -1.25375533e+00
-1.73948956e+00 2.74804235e-01 2.55739063e-01 -3.87344182e-01
1.01979148e+00 -2.71312654e-01 1.07721716e-01 -7.05250502e-02
-5.04358590e-01 -5.28916180e-01 6.06931567e-01 -8.80107999e-01
1.11402012e-01 -3.17235589e-02 8.39893073e-02 -6.75738975e-02
-3.18769127e-01 -1.36466876e-01 4.95738477e-01 -3.03106487e-01
-2.22440973e-01 -1.00449657e+00 -2.91719079e-01 4.97928709e-01
-3.91246647e-01 -2.59290896e-02 1.21235740e+00 -3.90108198e-01
1.45877039e+00 -1.75742006e+00 8.43358457e-01 2.27135271e-01
-5.02452254e-03 3.69096756e-01 3.63718390e-01 9.05397892e-01
-1.68670654e-01 -3.94459218e-02 -2.83425122e-01 2.10054934e-01
6.61163703e-02 -1.62813649e-01 -5.21464646e-01 1.22879159e+00
5.49794793e-01 5.86595178e-01 -5.98792553e-01 -8.05714056e-02
1.75605878e-01 5.28905988e-01 -4.34619993e-01 -1.96735766e-02
1.84184402e-01 1.74867615e-01 -5.01453996e-01 2.15853065e-01
4.58591044e-01 5.08212864e-01 1.26771882e-01 2.82535292e-02
-7.82983184e-01 -3.85391742e-01 -1.41675210e+00 4.76914078e-01
-4.45201874e-01 5.19372940e-01 1.06356740e+00 -1.11977959e+00
6.66444838e-01 6.43003523e-01 4.39774334e-01 -5.67507111e-02
6.37302577e-01 4.82115656e-01 2.64937758e-01 -4.10739034e-01
1.81663543e-01 -6.68962300e-01 -1.88627884e-01 4.09678407e-02
-7.28868544e-02 -5.46841979e-01 1.01179615e-01 2.12793335e-01
7.51801789e-01 -4.01698977e-01 4.35879886e-01 -1.34299839e+00
1.16762507e+00 -1.77875847e-01 6.02975786e-01 -1.66282892e-01
-5.27227998e-01 -1.54552102e-01 9.72766280e-01 3.67915124e-01
-1.28620183e+00 -7.29707956e-01 -2.18974069e-01 3.97928178e-01
1.78457782e-01 9.70051065e-02 -6.23043716e-01 2.87758052e-01
3.56452525e-01 1.54026553e-01 -1.01696920e+00 -7.21114159e-01
-4.70765650e-01 -3.24040115e-01 4.21887696e-01 3.80158782e-01
5.69060668e-02 -1.20310046e-01 -6.71675682e-01 2.84096450e-01
3.45921844e-01 -4.68573272e-01 -6.87914908e-01 4.69798058e-01
-7.51396835e-01 -1.09169710e+00 -6.95495188e-01 -5.29412866e-01
8.83105814e-01 -3.07481378e-01 1.80202723e-01 4.53228354e-02
-3.48164558e-01 7.74321496e-01 1.80690527e-01 -4.11100328e-01
-6.31537437e-01 -4.40404594e-01 7.64975369e-01 1.87376961e-02
-8.24993670e-01 -1.40706807e-01 -3.60556960e-01 4.44280058e-01
-1.19325650e+00 -3.96204531e-01 -1.95869014e-01 9.46287215e-01
5.96675992e-01 -4.26248699e-01 8.20469856e-01 -5.25191605e-01
1.04852796e+00 -3.20485771e-01 -1.29645860e+00 2.23593712e-02
-6.94244981e-01 1.86287984e-01 1.02112699e+00 -7.18037903e-01
-8.75253320e-01 2.05002740e-01 3.41515779e-01 -4.39189285e-01
9.66170013e-01 1.62532136e-01 -2.11961150e-01 -4.45628196e-01
7.06241429e-01 -6.43968582e-02 4.42012280e-01 -3.05653680e-02
3.74982893e-01 4.02700722e-01 1.91630587e-01 -4.24719334e-01
9.67841566e-01 2.66892910e-01 6.56089664e-01 -8.15537095e-01
9.32367295e-02 -4.02302086e-01 -8.09075534e-01 -4.56593186e-01
1.63063332e-01 -3.81809533e-01 -1.05321836e+00 6.50804758e-01
-9.42955434e-01 -5.79310358e-01 -6.62117422e-01 2.00957596e-01
-1.08103383e+00 -7.78153762e-02 -9.20914531e-01 -1.75987804e+00
-2.56571233e-01 -6.84501052e-01 4.60011095e-01 1.13757133e-01
-1.13985084e-01 -1.44579983e+00 4.33869958e-01 -6.90804482e-01
6.30515456e-01 6.77172422e-01 4.31278646e-01 3.48278135e-01
-2.88337708e-01 -2.97178566e-01 2.71023840e-01 5.88947177e-01
-1.89259604e-01 4.98322248e-01 -4.70977843e-01 -5.42670906e-01
3.52568418e-01 -7.49431271e-03 6.77436292e-01 5.47855079e-01
-2.81813651e-01 -8.84245574e-01 -1.79196894e-01 9.35780630e-02
2.07683182e+00 1.53675660e-01 1.09679751e-01 -1.30277202e-01
3.29645425e-01 7.76014209e-01 4.30655867e-01 4.18760180e-01
-1.60379812e-01 3.93567890e-01 4.12104994e-01 -3.62371281e-03
5.30019581e-01 4.29089189e-01 6.68541014e-01 6.82353675e-01
-5.74442185e-03 -1.79516032e-01 -5.63960850e-01 8.41211915e-01
-2.01377940e+00 -6.51581407e-01 -7.26174116e-01 2.11114192e+00
9.09660935e-01 -3.51857483e-01 4.54770982e-01 3.70611519e-01
7.73989379e-01 -4.83686715e-01 -7.94006765e-01 -1.17281747e+00
-4.29678053e-01 -1.65139064e-01 1.01434565e+00 1.13353896e+00
-8.21657538e-01 1.23038135e-01 6.83610582e+00 1.25650585e-01
-1.32303202e+00 6.49078637e-02 1.22524284e-01 -2.01959476e-01
-2.16852710e-01 -1.77294776e-01 -4.65969294e-01 2.13010862e-01
1.10157347e+00 -1.12929344e+00 6.07440285e-02 6.62087321e-01
9.74882603e-01 -3.52885395e-01 -6.18406117e-01 1.74821526e-01
-4.34676200e-01 -7.67029107e-01 -6.75690234e-01 1.11566171e-01
9.30354118e-01 -6.75896525e-01 4.85410541e-01 -1.15591511e-01
-7.05783069e-02 -3.92121553e-01 8.59500647e-01 1.01129663e+00
4.88906443e-01 -7.78662086e-01 6.80929184e-01 5.06169856e-01
-1.17842638e+00 -3.47896308e-01 -2.65899360e-01 -5.72070420e-01
2.74430454e-01 5.25467038e-01 -7.77798742e-02 4.90713388e-01
-5.01485765e-02 4.88874912e-01 -1.14660468e-02 8.33935320e-01
5.58144934e-02 2.09408224e-01 -5.42164385e-01 -6.49578273e-01
6.71923757e-02 -4.59194690e-01 8.11342418e-01 1.23739314e+00
5.14409207e-02 3.78252953e-01 -4.58477996e-02 9.66716349e-01
3.63800704e-01 3.93788099e-01 -7.43004978e-01 2.08319262e-01
-1.51888177e-01 1.01220739e+00 -1.91852853e-01 -1.90969840e-01
1.38890907e-01 8.35644066e-01 -1.11434020e-01 4.13308322e-01
-7.32162118e-01 -7.47442305e-01 1.13214684e+00 4.01810348e-01
-3.37841123e-01 -5.26135802e-01 -1.66186541e-01 -9.21087742e-01
2.60471612e-01 3.44173521e-01 1.09734073e-01 -2.31413841e-01
-9.20271575e-01 1.93708032e-01 1.76039457e-01 -1.31668591e+00
-4.59319353e-01 -4.63544071e-01 -6.20225310e-01 1.26061904e+00
-1.10581672e+00 -5.61820388e-01 5.04916370e-01 4.73882765e-01
-1.84305638e-01 5.28278947e-01 4.57476586e-01 2.15891406e-01
-8.63875449e-01 1.30297542e-01 7.93370068e-01 -3.92101765e-01
4.77679580e-01 -1.39531195e+00 -4.95302856e-01 1.13553333e+00
-1.33431590e+00 7.60379612e-01 1.43993747e+00 -3.96778882e-01
-1.89719963e+00 -1.10331380e+00 7.83283532e-01 -6.12732768e-02
1.29750860e+00 -1.70678988e-01 -1.12217069e+00 4.48342621e-01
3.65945488e-01 1.84978917e-01 -5.89624047e-02 -7.86181629e-01
2.87731260e-01 -2.22218692e-01 -1.23631728e+00 3.17016125e-01
4.92655098e-01 -4.57232028e-01 -5.99659793e-02 -5.15350066e-02
5.56026936e-01 -4.43713069e-01 -9.80334103e-01 2.11341247e-01
5.62307477e-01 -1.74221501e-01 4.96824831e-01 -6.11652553e-01
-1.59726903e-01 -5.02430975e-01 1.08651392e-01 -1.40797853e+00
-2.71662265e-01 -1.16739678e+00 -3.14063847e-01 1.08163857e+00
2.64743239e-01 -8.25361133e-01 -7.17206579e-03 8.91667962e-01
-1.82744548e-01 -9.53596175e-01 -1.20071638e+00 -1.29213119e+00
7.15065241e-01 2.21762419e-01 -8.00132632e-01 5.73053539e-01
9.05982375e-01 2.40019262e-01 -3.08754265e-01 2.06660539e-01
1.80953622e-01 -5.51098228e-01 1.47057608e-01 -1.00156593e+00
5.31025454e-02 -5.06518006e-01 -4.14782047e-01 -2.84128696e-01
2.84540087e-01 -5.55531144e-01 5.22496521e-01 -1.35305023e+00
-3.44763100e-01 -3.35289985e-02 -2.66889930e-01 1.05247118e-01
-5.64424545e-02 1.33265674e-01 5.61879016e-02 4.14788816e-03
1.06442697e-01 6.85472548e-01 1.13873351e+00 -8.20217878e-02
-7.14095592e-01 2.96310574e-01 -2.62394577e-01 2.77506709e-01
8.15290511e-01 5.38889542e-02 -8.39321494e-01 1.77037790e-01
3.00404668e-01 3.25272262e-01 3.37982833e-01 -7.70871043e-01
2.58101791e-01 -4.82925624e-01 -4.64667708e-01 -9.63278338e-02
2.14712143e-01 -7.70333469e-01 5.71884155e-01 1.25128436e+00
-3.75148118e-01 1.60499603e-01 2.22539932e-01 6.81813002e-01
-1.75872017e-02 -1.41600788e-01 1.32909369e+00 6.63522661e-01
-8.79820436e-02 -2.40713656e-02 -1.26876187e+00 -7.10858479e-02
1.30824602e+00 2.46551279e-02 -2.93036431e-01 -4.70659614e-01
-9.40328956e-01 3.39653552e-01 2.75119931e-01 1.58355720e-02
2.34807193e-01 -1.11476052e+00 -3.70109588e-01 -5.69197051e-02
-3.48513573e-01 -7.47590840e-01 3.58878881e-01 1.51521170e+00
-2.80737281e-01 8.68907392e-01 -3.25857043e-01 -1.35730773e-01
-1.36854756e+00 6.24519408e-01 9.13744569e-01 1.72174603e-01
-8.07646960e-02 5.05069911e-01 -5.58056571e-02 1.83115140e-01
-2.13096023e-01 -8.17219734e-01 3.47392201e-01 1.80525452e-01
1.36490285e-01 6.08051419e-01 4.92932573e-02 -7.48019278e-01
-3.64395499e-01 5.29827952e-01 7.95060337e-01 -5.10663152e-01
9.83430266e-01 -4.91639525e-01 -1.81487650e-01 8.84685814e-01
1.22404051e+00 -1.86352342e-01 -1.37002945e+00 -7.28748143e-02
-2.35297665e-01 3.07233572e-01 -2.72099003e-02 -2.34759256e-01
-8.48406315e-01 6.10881507e-01 5.71879327e-01 6.28256202e-01
1.19629169e+00 -5.02644122e-01 2.35573635e-01 3.39624494e-01
-1.07864991e-01 -1.38719547e+00 -4.34008420e-01 7.40706086e-01
1.22992277e+00 -3.40986282e-01 -1.42719775e-01 -7.45753646e-01
-1.94174692e-01 1.29158843e+00 6.17543578e-01 -9.45954800e-01
1.08285916e+00 7.06865013e-01 -2.27722839e-01 5.68028033e-01
-1.10027361e+00 -2.74856359e-01 3.08605701e-01 4.32466149e-01
7.43146002e-01 -5.59716951e-03 -1.39710426e+00 4.52994585e-01
6.13216043e-01 -9.25210491e-02 9.51016426e-01 1.00915945e+00
-5.79594076e-01 -7.46081114e-01 -4.29221869e-01 -1.47144780e-01
-1.74809724e-01 4.11550045e-01 -7.83633351e-01 1.08277154e+00
-3.56557608e-01 8.30903947e-01 -6.38151914e-02 5.40826693e-02
1.04802227e+00 5.44655845e-02 3.83445799e-01 -3.52463961e-01
-8.67452443e-01 2.62732595e-01 -1.65480182e-01 -2.68203795e-01
-3.69938135e-01 -5.06830692e-01 -1.24519467e+00 -3.84109974e-01
-8.27114880e-01 3.27159047e-01 3.46655369e-01 5.72275162e-01
2.46666837e-02 6.96635246e-01 6.57047749e-01 -5.15801787e-01
-9.64786649e-01 -1.01569712e+00 -8.48337233e-01 -2.86035746e-01
9.51979995e-01 -6.23045981e-01 -6.03958309e-01 2.82255262e-01] | [5.453096866607666, 2.665771722793579] |
d0a8fbe2-d5ae-4f20-bf66-b69eba7e5653 | exponential-family-embeddings | 1608.00778 | null | http://arxiv.org/abs/1608.00778v2 | http://arxiv.org/pdf/1608.00778v2.pdf | Exponential Family Embeddings | Word embeddings are a powerful approach for capturing semantic similarity
among terms in a vocabulary. In this paper, we develop exponential family
embeddings, a class of methods that extends the idea of word embeddings to
other types of high-dimensional data. As examples, we studied neural data with
real-valued observations, count data from a market basket analysis, and ratings
data from a movie recommendation system. The main idea is to model each
observation conditioned on a set of other observations. This set is called the
context, and the way the context is defined is a modeling choice that depends
on the problem. In language the context is the surrounding words; in
neuroscience the context is close-by neurons; in market basket data the context
is other items in the shopping cart. Each type of embedding model defines the
context, the exponential family of conditional distributions, and how the
latent embedding vectors are shared across data. We infer the embeddings with a
scalable algorithm based on stochastic gradient descent. On all three
applications - neural activity of zebrafish, users' shopping behavior, and
movie ratings - we found exponential family embedding models to be more
effective than other types of dimension reduction. They better reconstruct
held-out data and find interesting qualitative structure. | ['Maja R. Rudolph', 'Francisco J. R. Ruiz', 'David M. Blei', 'Stephan Mandt'] | 2016-08-02 | exponential-family-embeddings-1 | http://papers.nips.cc/paper/6571-exponential-family-embeddings | http://papers.nips.cc/paper/6571-exponential-family-embeddings.pdf | neurips-2016-12 | ['movie-recommendation'] | ['miscellaneous'] | [-3.89639169e-01 -3.91342163e-01 -3.84052753e-01 -4.58887368e-01
1.37217388e-01 -6.37556136e-01 5.79377413e-01 4.98155147e-01
-9.47191417e-01 3.30425620e-01 5.46950400e-01 -1.22929424e-01
-2.07124397e-01 -8.65481555e-01 -5.74668646e-01 -8.87353063e-01
-1.86737597e-01 4.23057139e-01 2.47529354e-02 -4.88669634e-01
6.25275671e-01 1.82990491e-01 -1.34352636e+00 2.04297394e-01
-3.65289226e-02 7.79686451e-01 2.88478106e-01 4.00312483e-01
-4.04806346e-01 2.90864021e-01 -1.57920763e-01 -3.48373652e-01
1.44412965e-01 -1.92192361e-01 -3.20603341e-01 -1.83312029e-01
-6.88854456e-02 3.59162991e-03 -2.87450373e-01 1.07048047e+00
4.67480421e-01 3.34505260e-01 1.11457467e+00 -1.13057804e+00
-1.41234541e+00 4.59740490e-01 -3.09771091e-01 4.89258528e-01
7.07714856e-02 -3.54054868e-01 1.49539375e+00 -1.08025491e+00
7.08597422e-01 1.18154418e+00 4.97336596e-01 6.27212584e-01
-1.67167699e+00 -1.90731123e-01 3.19289327e-01 2.04946429e-01
-1.30324721e+00 -2.85878126e-03 4.46571261e-01 -9.85625327e-01
8.34518850e-01 -5.83927184e-02 8.70898128e-01 1.20965195e+00
6.52639329e-01 4.91703987e-01 5.86378932e-01 -3.77069265e-01
8.03997517e-01 4.10621613e-01 5.92922866e-01 3.45882773e-01
2.18859211e-01 2.12430246e-02 -5.87498248e-01 -4.89733934e-01
7.04240918e-01 6.84282422e-01 -4.49003354e-02 -6.11516178e-01
-1.11235011e+00 1.36980307e+00 1.82710692e-01 4.75987107e-01
-2.32019112e-01 2.33724222e-01 5.09947598e-01 3.54839712e-01
4.62733656e-01 4.97227788e-01 -7.34997392e-01 2.30338886e-01
-4.20234144e-01 4.10758048e-01 6.87377810e-01 5.44922531e-01
6.57498419e-01 -2.59942144e-01 1.76520765e-01 9.40881908e-01
6.28529787e-01 1.62420571e-02 1.00499451e+00 -6.67579114e-01
-1.11339346e-01 3.84895831e-01 2.26573080e-01 -1.04802501e+00
-4.98475671e-01 -6.93752691e-02 -3.68663222e-01 1.12644508e-01
6.16834223e-01 6.26804456e-02 -6.52737796e-01 2.10204101e+00
2.50274956e-01 -5.25650457e-02 -2.12781981e-01 9.85383451e-01
3.97216558e-01 7.18138218e-01 3.41498218e-02 -6.13172129e-02
1.73386717e+00 -5.44369638e-01 -8.20871294e-01 -8.36360008e-02
7.02057481e-01 -1.98153958e-01 9.98226881e-01 4.94339615e-01
-6.09646380e-01 -3.50269496e-01 -9.25155342e-01 -2.90333062e-01
-9.47412133e-01 -2.82994449e-01 7.22318232e-01 3.36645216e-01
-8.49681675e-01 5.52709877e-01 -6.15915775e-01 -5.47892034e-01
2.52745241e-01 2.47871295e-01 -5.48382699e-01 -1.45311542e-02
-1.27550173e+00 8.39384615e-01 -6.40655980e-02 -3.74130309e-01
-6.89703286e-01 -6.12113118e-01 -9.12775576e-01 2.89656907e-01
-2.18969882e-02 -5.46337187e-01 7.43248522e-01 -5.32807767e-01
-1.01938772e+00 9.79591250e-01 -5.62321059e-02 -3.78068477e-01
-4.02030259e-01 1.57005876e-01 -3.61030251e-01 -2.99799412e-01
3.69619392e-02 5.39612591e-01 8.72415125e-01 -6.88809395e-01
-3.25851679e-01 -9.57743287e-01 4.74412218e-02 2.40899306e-02
-7.42239535e-01 -1.78693160e-02 -8.18247870e-02 -7.32036531e-01
1.60557609e-02 -9.63737547e-01 -3.70202839e-01 4.68678862e-01
4.09241259e-01 -5.85777521e-01 3.75323176e-01 -2.85190552e-01
1.25145221e+00 -2.53324223e+00 4.57711935e-01 -1.78832375e-02
5.14773250e-01 -1.57613397e-01 -2.56213993e-01 7.10525036e-01
-1.64884329e-01 2.96288788e-01 -1.28417805e-01 -7.84811080e-02
2.74756908e-01 3.56886119e-01 -4.14869398e-01 7.28475630e-01
8.71887803e-02 7.71056056e-01 -8.33873034e-01 -5.49282990e-02
-2.64989045e-02 5.73264420e-01 -9.08504009e-01 -3.24189365e-02
-2.55040139e-01 -3.98702584e-02 -4.19223189e-01 1.48958698e-01
2.04893410e-01 -2.63155192e-01 1.69612139e-01 -8.49592760e-02
-1.42827332e-01 1.22702405e-01 -1.23205316e+00 1.94906497e+00
-4.16179597e-01 7.40610838e-01 -1.59514084e-01 -1.24778986e+00
8.83945704e-01 2.94361681e-01 4.15166587e-01 -5.52641153e-01
2.20000595e-01 -2.87787914e-01 3.23098451e-01 -7.89044499e-01
4.08851594e-01 -3.40757072e-01 -2.54186571e-01 6.25252366e-01
4.44259971e-01 3.17682087e-01 8.89183730e-02 2.05211937e-01
1.01744998e+00 -2.48282164e-01 4.34899628e-01 -6.07290387e-01
-1.18225209e-01 -2.90148228e-01 6.62630320e-01 3.70391905e-01
-1.80890918e-01 4.89232212e-01 8.59216571e-01 -6.47922635e-01
-1.20320678e+00 -1.17398715e+00 -5.61612666e-01 1.34290588e+00
3.50202210e-02 -6.70199454e-01 -4.18286979e-01 -6.16632700e-02
3.21986407e-01 6.49099112e-01 -1.35446501e+00 -3.17246199e-01
5.42076714e-02 -8.37395370e-01 -2.69368161e-02 6.59762621e-01
-5.10878861e-01 -7.93344200e-01 -4.98664796e-01 3.36696237e-01
3.19226563e-01 -7.64338493e-01 -7.93937922e-01 5.25690854e-01
-8.78625393e-01 -8.35095882e-01 -5.45967817e-01 -7.89859653e-01
5.32442451e-01 -6.37975857e-02 9.39027011e-01 -2.79869378e-01
-5.75769186e-01 3.43271494e-01 -2.80968636e-01 -4.60159689e-01
2.37766087e-01 -4.56310332e-01 6.66006029e-01 3.38596642e-01
8.18137467e-01 -6.25368476e-01 -6.81650758e-01 1.43579930e-01
-1.28542924e+00 -5.51560640e-01 5.01178876e-02 1.02997756e+00
6.87462270e-01 -1.99130714e-01 5.03924608e-01 -8.18492174e-01
8.74844074e-01 -9.84303951e-01 -3.55846375e-01 6.51371256e-02
-5.91019750e-01 4.02340889e-01 5.82810819e-01 -8.16302180e-01
-3.67711246e-01 -1.76498920e-01 1.11384504e-01 -2.39590630e-01
2.22457703e-02 6.25902534e-01 1.38727203e-01 5.58146417e-01
6.51899755e-01 5.97327128e-02 4.60093282e-03 -6.58306181e-01
5.14981270e-01 5.84016919e-01 -5.97802587e-02 -3.42640251e-01
3.45313177e-02 5.16692400e-01 -2.64848977e-01 -1.08073282e+00
-3.96357238e-01 -4.84425753e-01 -5.92391551e-01 3.28383237e-01
1.25937641e+00 -7.52131224e-01 -7.15800822e-01 -1.12969249e-01
-9.78177428e-01 -1.41294047e-01 -5.30004203e-01 8.12643588e-01
-4.26888913e-01 1.60384998e-02 -7.75249302e-01 -8.11269403e-01
1.71147093e-01 -9.87407386e-01 9.06015873e-01 -1.10411523e-02
-5.32924175e-01 -1.34157193e+00 7.24011004e-01 -1.88586190e-01
2.24256277e-01 -1.92318693e-01 1.14659643e+00 -9.75042820e-01
1.49619937e-01 -3.35782588e-01 1.13376245e-01 2.62860745e-01
9.79924276e-02 2.09511612e-02 -7.58324325e-01 8.67807772e-03
2.32606992e-01 -1.61870927e-01 9.10500526e-01 6.61489546e-01
1.16616464e+00 -3.91532816e-02 -2.97336251e-01 4.50033665e-01
1.34394908e+00 2.41739005e-01 3.96584034e-01 -2.13549323e-02
3.92545462e-01 7.56758451e-01 2.27925658e-01 5.40852547e-01
1.94412887e-01 7.09214211e-01 2.72220194e-01 3.00690889e-01
5.61791360e-01 -3.36963326e-01 4.65937883e-01 9.20256317e-01
3.87946129e-01 -1.74002171e-01 -6.06024384e-01 7.06489146e-01
-1.78233862e+00 -1.07231224e+00 7.26399049e-02 2.10441947e+00
7.12551653e-01 4.85874712e-03 2.50586540e-01 -1.02381669e-02
6.47447407e-01 -8.41451623e-03 -6.35819316e-01 -7.47994721e-01
2.30727959e-02 7.17838630e-02 1.71165153e-01 5.93958259e-01
-6.86089277e-01 5.63591301e-01 6.26646996e+00 7.35203505e-01
-8.97495508e-01 2.92783946e-01 3.10649753e-01 -3.96469802e-01
-6.66465938e-01 -8.29140022e-02 -8.65382135e-01 6.91072702e-01
1.04762554e+00 -5.94890937e-02 4.96664405e-01 6.94559634e-01
1.37993395e-01 4.99880640e-03 -1.49284077e+00 9.61824477e-01
8.40367079e-02 -1.26511621e+00 -1.38764635e-01 5.05145431e-01
3.43458563e-01 5.90011887e-02 2.09919840e-01 1.47874519e-01
2.61632979e-01 -1.11656809e+00 5.52358687e-01 6.53866172e-01
5.48341453e-01 -3.42185348e-01 4.29177970e-01 4.64765817e-01
-8.74911249e-01 -2.73905069e-01 -9.38870013e-01 -3.61341029e-01
1.56786814e-01 6.74334347e-01 -2.84308791e-01 -3.92073631e-01
8.41370642e-01 9.37867343e-01 -1.83642536e-01 6.53381407e-01
7.04794228e-02 4.99282271e-01 -2.44088173e-01 -4.70765054e-01
8.16400573e-02 -5.90530336e-01 2.42241383e-01 1.17753005e+00
5.56612201e-02 2.27322787e-01 -2.23711640e-01 1.16428912e+00
-2.39014048e-02 2.20361814e-01 -1.01053846e+00 -3.17011654e-01
3.15534115e-01 1.27835476e+00 -7.57931054e-01 -9.86942723e-02
-7.08812475e-01 8.64848554e-01 4.63892460e-01 3.62678170e-01
-6.08314574e-01 -2.64755785e-01 1.29873645e+00 8.76777396e-02
4.97015864e-01 -3.92129302e-01 1.45056574e-02 -1.38595498e+00
-1.44973010e-01 -2.31594697e-01 2.65313357e-01 -6.14642978e-01
-1.77602351e+00 2.29973510e-01 -1.03562295e-01 -1.05243778e+00
4.44009565e-02 -1.09307694e+00 -4.89394516e-01 7.76888907e-01
-9.69425917e-01 -3.17881286e-01 4.70503122e-01 5.18345773e-01
5.79430580e-01 -2.38983750e-01 1.24626017e+00 3.05613041e-01
-5.22096574e-01 2.48480052e-01 6.41022980e-01 1.96846917e-01
5.54745078e-01 -1.21122253e+00 2.43787542e-01 5.44626638e-02
8.16733360e-01 1.03395045e+00 6.91691458e-01 -2.94680804e-01
-1.47536159e+00 -5.06337345e-01 9.65737462e-01 -7.78222322e-01
1.01388252e+00 -1.00146770e+00 -8.39560091e-01 7.41595209e-01
4.89531010e-02 3.08572710e-01 1.48938227e+00 3.76386166e-01
-5.81875801e-01 5.95442206e-02 -1.07760251e+00 7.19435692e-01
8.34444046e-01 -7.41234601e-01 -7.42912412e-01 3.64889383e-01
8.96610975e-01 3.34224433e-01 -1.00471079e+00 -2.47848451e-01
8.97132397e-01 -6.95277691e-01 8.49467337e-01 -1.31087160e+00
3.97073239e-01 -1.62824035e-01 -7.19399869e-01 -1.59644783e+00
-7.87041485e-01 1.37376383e-01 -1.01621017e-01 9.69003260e-01
5.48574388e-01 -5.99044383e-01 5.39102912e-01 8.73503268e-01
2.78134644e-01 -1.07094872e+00 -8.84654820e-01 -4.65576231e-01
3.05888325e-01 -4.53101963e-01 3.32409233e-01 1.13192606e+00
4.83074039e-01 6.57056391e-01 -3.78124528e-02 -1.00387245e-01
3.45474869e-01 -3.27172056e-02 1.56434119e-01 -1.18363154e+00
-5.14291108e-01 -3.70676637e-01 -8.43644023e-01 -9.24169540e-01
1.03195250e-01 -1.04188490e+00 -3.20106089e-01 -1.20743179e+00
1.48897827e-01 -1.90073192e-01 -5.52034378e-01 -3.16977408e-03
2.02481806e-01 1.08822457e-01 1.59861110e-02 -7.66056478e-02
-4.77963626e-01 7.00071037e-01 8.06257725e-01 -7.91529492e-02
-1.66645665e-02 -3.28651905e-01 -8.33905458e-01 6.23744726e-01
6.48707271e-01 -6.73763573e-01 -4.93412107e-01 -6.33628666e-01
7.65709043e-01 -1.58445612e-02 8.53776708e-02 -2.74446994e-01
1.94376871e-01 -1.25065640e-01 3.91315520e-01 -1.84929911e-02
4.38528150e-01 -9.75942850e-01 -6.89859465e-02 2.40077183e-01
-8.54632854e-01 2.83574760e-01 -7.09118024e-02 1.02344239e+00
-2.91141942e-02 -4.49971497e-01 5.37736416e-01 -6.70768023e-02
-5.84602237e-01 5.00091314e-01 -7.53320754e-01 1.75774321e-01
8.04805875e-01 -6.24899045e-02 1.02125980e-01 -2.87274182e-01
-1.14897823e+00 6.32059202e-02 3.79149526e-01 5.44634163e-01
6.86753213e-01 -1.58267128e+00 -4.74559158e-01 3.77661645e-01
3.32907647e-01 -5.03002167e-01 1.19169459e-01 5.71625769e-01
1.33384362e-01 2.79487401e-01 -1.23201273e-01 -3.90684843e-01
-7.76666403e-01 9.55115080e-01 1.79798737e-01 1.23555802e-01
-2.65595913e-01 1.02469015e+00 7.07245886e-01 -4.95538175e-01
1.08300596e-01 -5.85198998e-01 -6.41261041e-01 5.46015620e-01
6.90531254e-01 6.02584779e-02 -1.59583136e-01 -4.42931443e-01
-3.36795360e-01 6.02691650e-01 -9.94097367e-02 -2.46003792e-01
1.65954888e+00 -1.58333272e-01 -1.35572851e-01 1.34723783e+00
1.54456913e+00 -1.99552506e-01 -1.08565021e+00 -3.90185237e-01
-9.95904431e-02 -3.77583265e-01 9.68983099e-02 -3.07332426e-01
-9.98788059e-01 1.05006599e+00 6.64994717e-01 6.18185282e-01
5.25705934e-01 2.72867531e-01 4.24235821e-01 2.56299824e-01
3.48865479e-01 -1.19933629e+00 6.86987257e-03 4.14825112e-01
6.95913315e-01 -1.17995584e+00 -2.10401416e-01 2.81214416e-01
-6.08197570e-01 1.04363370e+00 2.55099148e-01 -6.35155559e-01
1.35336971e+00 9.87783223e-02 -3.19567442e-01 -4.50914323e-01
-1.07975936e+00 4.95900027e-02 4.22237664e-02 4.08825696e-01
5.75901270e-01 1.13298185e-01 -5.98345637e-01 1.20276380e+00
-3.52600007e-03 -1.84796065e-01 4.73239988e-01 5.68458855e-01
-3.67505074e-01 -1.00828218e+00 -3.40742692e-02 6.62065387e-01
-4.30513501e-01 -2.43971497e-01 -1.06239669e-01 3.83425385e-01
1.74880207e-01 7.21043110e-01 4.31239665e-01 -4.91835415e-01
2.20628291e-01 3.42509300e-01 4.82950807e-01 -1.03803921e+00
-2.14312032e-01 -2.16023713e-01 -4.02655482e-01 -4.19147670e-01
-1.88773543e-01 -8.74631882e-01 -1.34869230e+00 -1.10824995e-01
-5.11462390e-01 2.26427361e-01 9.60114658e-01 8.88993561e-01
3.46450806e-01 1.58233315e-01 5.71476519e-01 -7.71046042e-01
-5.74602425e-01 -8.77583206e-01 -1.32183743e+00 6.20254755e-01
4.34016615e-01 -8.72639060e-01 -6.08560979e-01 1.17353365e-01] | [10.405820846557617, 8.67973804473877] |
93159f62-aa07-4ccb-b962-ba52f1994567 | solving-the-side-chain-packing-arrangement-of | 2212.03320 | null | https://arxiv.org/abs/2212.03320v1 | https://arxiv.org/pdf/2212.03320v1.pdf | Solving the Side-Chain Packing Arrangement of Proteins from Reinforcement Learned Stochastic Decision Making | Protein structure prediction is a fundamental problem in computational molecular biology. Classical algorithms such as ab-initio or threading as well as many learning methods have been proposed to solve this challenging problem. However, most reinforcement learning methods tend to model the state-action pairs as discrete objects. In this paper, we develop a reinforcement learning (RL) framework in a continuous setting and based on a stochastic parametrized Hamiltonian version of the Pontryagin maximum principle (PMP) to solve the side-chain packing and protein-folding problem. For special cases our formulation can be reduced to previous work where the optimal folding trajectories are trained using an explicit use of Langevin dynamics. Optimal continuous stochastic Hamiltonian dynamics folding pathways can be derived with use of different models of molecular energetics and force fields. In our RL implementation we adopt a soft actor-critic methodology however we can replace this other RL training based on A2C, A3C or PPO. | ['Minh Nguyen', 'Conrad Li', 'Chandrajit Bajaj'] | 2022-12-06 | null | null | null | null | ['protein-folding'] | ['natural-language-processing'] | [ 4.16218117e-02 3.00132513e-01 -5.38851954e-02 -2.83035815e-01
-6.98306262e-01 -3.88875872e-01 6.36334300e-01 2.17821181e-01
-6.51467204e-01 1.52924478e+00 -6.00294620e-02 -5.33984363e-01
2.48241425e-02 -6.55318677e-01 -1.05526865e+00 -1.14854145e+00
-1.20729946e-01 7.19849825e-01 2.10597426e-01 -5.09371519e-01
4.52682406e-01 5.99412084e-01 -1.11301184e+00 -5.34377545e-02
9.85723078e-01 2.91127950e-01 3.78284693e-01 9.50611413e-01
-6.72961697e-02 5.71397781e-01 -1.55390322e-01 -1.54499352e-01
2.33629748e-01 -9.56840515e-01 -1.00281703e+00 -6.47791699e-02
-1.28909141e-01 2.35825315e-01 1.70439303e-01 7.61824429e-01
6.54157996e-01 4.63298202e-01 6.86440408e-01 -5.78304231e-01
-1.25853837e-01 2.38881081e-01 -2.04724386e-01 2.75079869e-02
3.69659752e-01 5.53299367e-01 1.02193165e+00 -5.44348359e-01
1.02480519e+00 1.16092372e+00 6.32370055e-01 6.21126413e-01
-1.90584743e+00 -7.64663965e-02 5.47365472e-02 4.30289328e-01
-8.50758851e-01 -2.11090390e-02 6.06536865e-01 -3.40811551e-01
1.44436741e+00 -6.55456334e-02 8.67280722e-01 1.06039953e+00
6.58145547e-01 5.35389245e-01 1.34465575e+00 -7.10910976e-01
7.86385000e-01 -2.68447638e-01 8.33035856e-02 7.06990600e-01
-1.13795415e-01 3.94264787e-01 -9.22501832e-02 -5.55974364e-01
4.69751388e-01 -2.79285926e-02 -3.42651904e-02 -7.82683134e-01
-1.01005006e+00 1.27910185e+00 2.58259863e-01 -2.07916498e-01
-6.67563856e-01 3.00785661e-01 3.78234535e-01 2.68776298e-01
7.77196363e-02 7.74758577e-01 -6.42863810e-01 -4.94857654e-02
-7.34240294e-01 9.03220892e-01 1.02561641e+00 3.32117796e-01
8.62438560e-01 -1.30646497e-01 -5.87950349e-02 3.30369622e-01
5.10634482e-01 7.90963322e-02 2.22974896e-01 -1.07773626e+00
-4.17182334e-02 1.32013708e-01 5.62844276e-01 -2.78239489e-01
-3.78073931e-01 2.21221343e-01 -4.03509408e-01 6.39716148e-01
6.00136697e-01 -4.80065286e-01 -8.10871065e-01 1.63259196e+00
6.50062740e-01 9.14233327e-02 1.52262613e-01 6.86227918e-01
-3.12585533e-02 9.52798963e-01 3.80015016e-01 -8.21363091e-01
1.10065401e+00 -8.86594594e-01 -8.43657553e-01 3.57414722e-01
6.24463201e-01 -5.74595153e-01 7.39371359e-01 6.21222734e-01
-1.16790867e+00 -3.42369407e-01 -1.06660080e+00 -7.32502118e-02
-3.37123334e-01 -1.86096877e-01 5.07163882e-01 4.27740991e-01
-1.01185298e+00 1.50733292e+00 -1.24595511e+00 -3.50682616e-01
4.85137738e-02 6.84233665e-01 -2.51973253e-02 5.35219073e-01
-1.12331116e+00 1.36659098e+00 7.09241986e-01 -1.20169550e-01
-1.12597513e+00 -2.13229209e-01 -4.51387793e-01 -3.72390658e-01
4.10531789e-01 -8.08556318e-01 1.24586082e+00 -1.16831362e+00
-2.28921580e+00 5.46954811e-01 -1.12010576e-01 -8.32851529e-01
5.44889867e-01 1.70393498e-03 3.96970123e-01 8.49763677e-02
-4.08576488e-01 7.40705371e-01 6.31568253e-01 -1.01261890e+00
4.07263674e-02 -2.36399353e-01 5.97065389e-02 2.61877924e-01
5.55497468e-01 1.58749018e-02 5.38711727e-01 -3.47634405e-01
-2.83828348e-01 -1.26458204e+00 -9.47367311e-01 -1.81657031e-01
-3.11330467e-01 -3.22955936e-01 4.82282221e-01 -5.71307123e-01
8.20085227e-01 -1.28084445e+00 9.77518499e-01 6.69796616e-02
-3.53338778e-01 5.13441801e-01 2.96249717e-01 9.63898003e-01
-1.71011612e-01 -7.92028159e-02 -4.31411952e-01 -8.21432546e-02
-8.33022818e-02 4.86284524e-01 -2.49100611e-01 4.23020065e-01
2.03550503e-01 8.35439682e-01 -9.42420959e-01 -4.32742059e-01
2.01330811e-01 5.41297257e-01 -8.88015091e-01 3.59902263e-01
-8.46207380e-01 8.11774075e-01 -5.48927844e-01 -3.39324959e-02
4.12789673e-01 -1.64345264e-01 8.60127747e-01 1.65022701e-01
-4.91502553e-01 2.78308749e-01 -8.98150802e-01 1.71697640e+00
-2.35254467e-01 2.65131146e-02 -5.26888408e-02 -1.32773614e+00
9.03266668e-01 4.18238133e-01 4.49999005e-01 -8.71174335e-02
8.82047042e-02 1.75932005e-01 3.17848146e-01 -5.93746305e-01
1.43656418e-01 -6.86452448e-01 4.89496887e-01 3.30591291e-01
1.58226758e-01 -1.01397753e-01 6.39897287e-02 -1.75089166e-01
8.99965405e-01 1.07923651e+00 6.94344759e-01 -4.94670600e-01
7.00347424e-01 3.47261518e-01 6.03968024e-01 6.13783181e-01
-1.86263517e-01 3.69809508e-01 7.62638509e-01 -6.66520715e-01
-1.60533094e+00 -6.40507042e-01 -1.93367749e-01 8.12451065e-01
-1.69577003e-01 -4.08195257e-01 -1.21323240e+00 -5.50867498e-01
-1.30054414e-01 5.91771007e-01 -4.59669739e-01 -8.47381055e-02
-1.02530563e+00 -1.16469514e+00 -8.82874057e-02 1.81586444e-02
-9.18367207e-02 -1.70631647e+00 -9.12957668e-01 9.01522994e-01
2.48165533e-01 -5.21863282e-01 -8.66920874e-02 8.25584829e-01
-1.16610634e+00 -9.20296133e-01 -7.91407049e-01 -5.54702699e-01
3.21617633e-01 -4.01083142e-01 8.88425887e-01 -7.33421817e-02
-4.75743413e-01 -1.63075924e-01 -2.94791371e-01 -1.81830481e-01
-7.78438032e-01 1.24412142e-01 6.79613873e-02 -1.72356531e-01
2.41578475e-01 -8.11859608e-01 -8.03672731e-01 -5.32691143e-02
-6.39441311e-01 1.67440236e-01 2.92054862e-01 1.23090947e+00
8.95435512e-01 -4.96820182e-01 5.68733454e-01 -1.01012409e+00
6.48514926e-01 -3.60741943e-01 -8.07915688e-01 2.97830909e-01
-9.49223816e-01 8.44637096e-01 9.93140757e-01 -4.25485104e-01
-1.03945875e+00 6.09601557e-01 -4.82503474e-01 -1.55019403e-01
-1.96700647e-01 1.92687139e-01 -9.80903655e-02 -2.13317946e-01
5.17085910e-01 4.02473360e-01 2.69118160e-01 -5.12666941e-01
2.41018683e-01 1.95250899e-01 -1.99900612e-01 -9.84420300e-01
3.26547801e-01 2.51820445e-01 3.56745183e-01 -6.35952771e-01
-4.82600480e-01 -3.00359596e-02 -7.47300506e-01 6.88558146e-02
1.17815554e+00 -2.90653467e-01 -1.45572960e+00 7.45153278e-02
-1.23831284e+00 -5.06450355e-01 -3.18176299e-01 4.81254190e-01
-1.28993499e+00 7.22129107e-01 -5.26224375e-01 -1.05134404e+00
-2.82691956e-01 -1.66083217e+00 9.28988516e-01 2.36411691e-01
-6.56743422e-02 -9.77287114e-01 7.14763105e-01 1.59719605e-02
1.71222270e-01 7.06695080e-01 1.11621571e+00 -3.46852034e-01
-5.46566308e-01 2.59694129e-01 4.71483499e-01 1.67970732e-01
-3.56884450e-01 1.89907938e-01 -5.21155536e-01 -2.93255448e-01
-1.27342105e-01 -5.23733377e-01 9.07815576e-01 7.19975173e-01
7.93219149e-01 -4.27646875e-01 -3.02326471e-01 4.31366682e-01
1.70249629e+00 4.54270542e-01 6.34809017e-01 5.38722992e-01
3.49730790e-01 8.04764211e-01 4.67509627e-01 4.75466371e-01
-3.72402780e-02 9.47450995e-01 4.63878244e-01 3.57740462e-01
3.84026885e-01 -3.33498418e-01 5.30486703e-01 3.64217579e-01
-6.18371725e-01 9.56911370e-02 -9.54632580e-01 2.95696352e-02
-2.13293242e+00 -1.17291772e+00 -5.95969968e-02 2.20910215e+00
1.23314047e+00 1.11406468e-01 4.37987119e-01 -2.11444721e-01
4.04299378e-01 -1.39029741e-01 -7.89428532e-01 -8.45685601e-01
1.87358987e-02 3.97146165e-01 5.97413063e-01 8.78889382e-01
-7.66797662e-01 9.73692477e-01 7.07449245e+00 5.48833609e-01
-9.46188509e-01 -1.68360509e-02 5.91235578e-01 6.32805228e-02
-1.38751253e-01 6.72691166e-01 -9.60046411e-01 5.57988644e-01
1.30025411e+00 1.72182500e-01 7.89865255e-01 7.40128875e-01
6.70887470e-01 -2.74436325e-01 -8.48201215e-01 5.15246332e-01
-6.33935630e-01 -1.46619487e+00 -1.40461385e-01 2.59951353e-01
7.11915493e-01 -2.75934100e-01 -2.81191826e-01 1.23868041e-01
4.76721019e-01 -1.02579010e+00 4.22452867e-01 6.14224017e-01
1.82107508e-01 -8.54374111e-01 4.09152448e-01 5.70775092e-01
-6.25789583e-01 1.21858716e-01 -4.86171842e-01 -9.92522016e-02
3.19927722e-01 1.04703240e-01 -6.96798444e-01 1.92426145e-01
8.73435959e-02 5.51967323e-01 2.54464503e-02 6.28197193e-01
-2.15204358e-02 6.66069388e-01 -2.22530603e-01 -5.01749873e-01
4.56952631e-01 -9.75298643e-01 4.80954856e-01 1.09770644e+00
-1.37911752e-01 2.17724830e-01 2.42945820e-01 9.28725243e-01
2.85015196e-01 2.58832723e-01 -3.03804427e-01 5.14186323e-02
-1.61498278e-01 9.57142770e-01 -8.80139172e-01 -2.56518155e-01
-1.31372005e-01 8.25461268e-01 5.86241901e-01 4.12119925e-01
-8.05260181e-01 -2.42598683e-01 7.51172543e-01 4.35544699e-02
5.24356067e-01 -3.10169309e-01 4.65099603e-01 -1.27810025e+00
-4.06649530e-01 -9.61557627e-01 -1.53137505e-01 -2.98341304e-01
-8.75161648e-01 2.86642939e-01 -1.39311515e-02 -6.27010643e-01
-3.78605843e-01 -7.22511947e-01 -6.19356871e-01 9.17025328e-01
-1.58569098e+00 -5.80801845e-01 6.26004577e-01 3.35255444e-01
5.87094545e-01 1.79309193e-02 9.62228715e-01 -3.25438201e-01
-6.19752526e-01 6.35900721e-02 7.19021738e-01 -5.41364908e-01
3.70822340e-01 -1.73305964e+00 2.24367827e-01 3.86160284e-01
-3.34785610e-01 5.66047966e-01 1.48612034e+00 -7.22733676e-01
-1.52369905e+00 -5.98691285e-01 6.74677014e-01 -2.17334792e-01
5.89018703e-01 -2.86928028e-01 -1.06030679e+00 4.79592651e-01
4.81379062e-01 9.79681537e-02 3.73569369e-01 -6.08574599e-02
2.40765780e-01 1.94076136e-01 -1.13368034e+00 4.35751915e-01
7.94052124e-01 -1.15421824e-01 -3.76249045e-01 7.62693644e-01
4.96076286e-01 -2.22077176e-01 -8.95108104e-01 -2.95052622e-02
4.10971791e-01 -9.61565018e-01 9.78270590e-01 -1.36208880e+00
2.73541003e-01 -1.81833148e-01 1.64588705e-01 -1.25960755e+00
-1.67813510e-01 -1.26758611e+00 -3.10217291e-01 5.28195500e-01
2.97237396e-01 -4.91544217e-01 8.66145670e-01 3.43488842e-01
-2.56749913e-02 -1.20042932e+00 -8.42706442e-01 -6.65366232e-01
5.33199608e-01 2.80263573e-01 1.25792548e-01 5.41627824e-01
2.75134981e-01 3.06700349e-01 -6.61606729e-01 -2.63144553e-01
7.86033392e-01 2.59744138e-01 4.00594771e-01 -9.26344872e-01
-8.24824631e-01 -1.96414888e-01 -2.98925191e-02 -8.40891957e-01
4.54790622e-01 -8.63520980e-01 7.49302059e-02 -1.15822077e+00
2.50055760e-01 -6.01828657e-02 -1.05950452e-01 2.58855462e-01
-1.03643760e-01 -5.01717806e-01 1.22460816e-02 2.27033481e-01
-4.77854878e-01 8.06760609e-01 1.43055177e+00 2.99185514e-01
-3.76238286e-01 2.19039544e-02 5.89386374e-02 4.37117010e-01
8.37161243e-01 -7.03261554e-01 -1.88180655e-01 3.68141562e-01
4.96695817e-01 5.27252614e-01 2.43273050e-01 -4.98242021e-01
-7.42852166e-02 -4.75235909e-01 1.39329091e-01 -1.38503313e-01
2.51536638e-01 -3.84365112e-01 2.96430022e-01 9.65432763e-01
-5.39029896e-01 1.32692322e-01 -1.56473473e-01 7.98248827e-01
7.89689422e-02 -4.79450583e-01 1.25032449e+00 -6.46929443e-01
-1.89344347e-01 1.17642410e-01 -7.20163286e-01 -1.43324897e-01
1.25362575e+00 -1.21085256e-01 2.51209587e-01 -3.43264863e-02
-1.26063120e+00 3.35001498e-02 6.88502073e-01 -3.10271025e-01
3.88469011e-01 -9.89841521e-01 -4.94048864e-01 -1.31573603e-01
-3.76836270e-01 -3.04987878e-01 2.75018699e-02 8.90372694e-01
-1.08404720e+00 5.54012716e-01 -5.34889519e-01 -1.96594432e-01
-1.06550324e+00 8.74310434e-01 7.18971193e-01 -6.09083354e-01
-5.08003056e-01 2.98598468e-01 -2.54434288e-01 -3.97116244e-01
-1.41012460e-01 -8.69584307e-02 -4.73800451e-02 -1.44227788e-01
1.98014453e-01 3.55991483e-01 -1.86920792e-01 -4.81197447e-01
-1.40682697e-01 5.66556275e-01 -2.90578693e-01 -1.50466457e-01
1.60033214e+00 8.69665816e-02 8.52766111e-02 2.40836978e-01
9.67120528e-01 -5.52850783e-01 -1.66901219e+00 1.27462894e-01
3.34757358e-01 2.22591400e-01 -2.51281280e-02 -6.14375532e-01
-2.56775707e-01 9.16166306e-01 6.69190049e-01 1.54466163e-02
6.57730758e-01 -3.25450510e-01 6.73274696e-01 7.30255127e-01
4.51797932e-01 -1.19840980e+00 -1.44192219e-01 4.22667176e-01
6.15634263e-01 -1.36227357e+00 5.29808439e-02 2.03734174e-01
-5.36766827e-01 1.55609548e+00 2.18698829e-01 -6.05241716e-01
5.00372708e-01 1.07710928e-01 -1.65835574e-01 1.13831945e-01
-9.45885599e-01 -1.63884223e-01 -2.30677933e-01 3.76561403e-01
7.33215988e-01 -2.70439256e-02 -1.15803993e+00 -6.02463372e-02
2.94680208e-01 5.94452918e-02 4.68843430e-01 1.11245859e+00
-5.55114508e-01 -1.90332818e+00 -2.69974947e-01 -6.48919940e-02
-5.61338902e-01 2.99724229e-02 -3.36131811e-01 6.00761831e-01
1.10402647e-02 3.78154218e-01 -4.82558936e-01 1.30186126e-01
-1.09980293e-01 4.59851831e-01 8.28021824e-01 -6.01038337e-01
-8.94985557e-01 1.79961115e-01 -1.64285466e-01 -4.86294061e-01
-7.42698610e-01 -9.28523660e-01 -1.55037260e+00 -3.01324874e-01
-4.28102046e-01 4.99699891e-01 7.94996262e-01 9.56689298e-01
4.47983503e-01 1.62368670e-01 4.53854859e-01 -1.08322775e+00
-1.08231854e+00 -9.54617083e-01 -5.38939059e-01 3.42460066e-01
4.58703458e-01 -7.30328202e-01 -8.08190331e-02 1.09252647e-01] | [4.758528232574463, 5.496603012084961] |
38bce51c-a8e9-4b40-8b23-7cfc8064c85c | condition-number-analysis-of-kernel-based | 0912.2800 | null | https://arxiv.org/abs/0912.2800v1 | https://arxiv.org/pdf/0912.2800v1.pdf | Condition Number Analysis of Kernel-based Density Ratio Estimation | The ratio of two probability densities can be used for solving various machine learning tasks such as covariate shift adaptation (importance sampling), outlier detection (likelihood-ratio test), and feature selection (mutual information). Recently, several methods of directly estimating the density ratio have been developed, e.g., kernel mean matching, maximum likelihood density ratio estimation, and least-squares density ratio fitting. In this paper, we consider a kernelized variant of the least-squares method and investigate its theoretical properties from the viewpoint of the condition number using smoothed analysis techniques--the condition number of the Hessian matrix determines the convergence rate of optimization and the numerical stability. We show that the kernel least-squares method has a smaller condition number than a version of kernel mean matching and other M-estimators, implying that the kernel least-squares method has preferable numerical properties. We further give an alternative formulation of the kernel least-squares estimator which is shown to possess an even smaller condition number. We show that numerical studies meet our theoretical analysis. | ['Taiji Suzuki', 'Masashi Sugiyama', 'Takafumi Kanamori'] | 2009-12-15 | null | null | null | null | ['density-ratio-estimation'] | ['methodology'] | [ 1.25692666e-01 8.32597688e-02 -3.85257483e-01 -3.37724328e-01
-9.21147168e-01 -1.25137389e-01 1.58190891e-01 2.12339833e-01
-5.73958397e-01 9.69481051e-01 -1.77826405e-01 -4.09673840e-01
-5.97702265e-01 -4.18523461e-01 -5.09448886e-01 -8.45190525e-01
-2.51944542e-01 3.49360377e-01 1.06972180e-01 1.15747407e-01
4.43507642e-01 5.28691411e-01 -1.29580820e+00 -8.00354719e-01
1.39162576e+00 1.05140042e+00 -1.24955788e-01 4.74468291e-01
3.06752115e-01 2.24531382e-01 -1.05853580e-01 -5.36907390e-02
1.68063447e-01 -6.34676337e-01 -4.36282068e-01 -6.47007674e-02
3.59975129e-01 -1.85325563e-01 1.51799664e-01 1.44749117e+00
3.96383315e-01 4.38232899e-01 1.59046555e+00 -1.62553728e+00
-5.17749608e-01 1.76964790e-01 -9.84110832e-01 1.34240612e-01
3.52156222e-01 -5.49634635e-01 6.87956750e-01 -1.00200021e+00
8.92460197e-02 1.13944721e+00 8.77502322e-01 2.13890180e-01
-1.47835374e+00 -6.53895378e-01 -4.29079205e-01 1.64093319e-02
-1.90189826e+00 -3.81027281e-01 4.24548447e-01 -6.77448571e-01
5.79112649e-01 3.07791233e-01 1.33309901e-01 3.12338352e-01
3.86652917e-01 5.38743556e-01 1.08426642e+00 -5.57926655e-01
4.27430600e-01 5.44796467e-01 1.73778996e-01 8.65031064e-01
5.97929060e-01 -4.55103517e-02 -1.35090008e-01 -8.28534245e-01
1.17164350e+00 -2.41835371e-01 -5.83077908e-01 -9.03592408e-01
-1.07432938e+00 1.05074453e+00 -6.26819208e-02 3.14116091e-01
-2.27623791e-01 -3.22840214e-02 -8.83432403e-02 3.70039821e-01
6.73272491e-01 2.79679686e-01 -4.59334522e-01 2.96343490e-02
-8.90065670e-01 1.87568173e-01 1.14533770e+00 9.26352382e-01
7.85717130e-01 9.39603224e-02 2.18501434e-01 7.53346920e-01
5.86430490e-01 7.68735528e-01 4.54379946e-01 -8.28257561e-01
2.77646899e-01 2.35154495e-01 3.39924753e-01 -8.66560042e-01
-5.06594479e-01 -7.56703094e-02 -1.05089390e+00 7.24812597e-02
1.00306177e+00 -2.70136714e-01 -1.98415890e-01 1.87892282e+00
4.67070490e-01 1.53194636e-01 -2.28870407e-01 6.74185634e-01
-1.52415708e-01 4.21321511e-01 -8.40078890e-02 -7.40910828e-01
1.02563417e+00 -2.92867482e-01 -7.53158867e-01 7.55366609e-02
8.55793595e-01 -6.31491423e-01 8.29265118e-01 3.21208000e-01
-9.23792124e-01 -9.43936035e-02 -9.30526853e-01 4.30173784e-01
-7.55783692e-02 1.38072267e-01 4.15145516e-01 8.74328315e-01
-8.09884787e-01 8.53271723e-01 -6.92425430e-01 -3.79512191e-01
-6.58527762e-02 4.83721524e-01 -6.78284824e-01 4.73218530e-01
-9.41654086e-01 9.33756113e-01 1.83938444e-01 -1.14690676e-01
-2.85491510e-03 -8.59690964e-01 -1.06998098e+00 1.32228002e-01
-4.89797816e-03 -4.91908997e-01 9.92819846e-01 -7.28515029e-01
-1.46105897e+00 5.47036767e-01 -2.10875943e-01 -1.62542671e-01
5.45836031e-01 -2.25728795e-01 -1.40198022e-01 5.55058867e-02
1.95369378e-01 7.47614428e-02 9.98324394e-01 -4.73285913e-01
-3.32092643e-01 -5.91205180e-01 -7.90494740e-01 1.58024162e-01
-1.76625907e-01 6.62623048e-02 2.77461916e-01 -5.81974030e-01
2.68512756e-01 -7.63617814e-01 -4.00711417e-01 2.30442509e-01
-1.10104583e-01 -1.12227917e-01 1.71375513e-01 -5.84389567e-01
1.33530211e+00 -2.36247230e+00 -1.08390220e-01 6.46668315e-01
-7.22252345e-03 -2.85539180e-01 1.92628354e-01 1.81695834e-01
-5.17811716e-01 -1.00389235e-01 -6.28397465e-01 8.49473178e-02
-7.67310634e-02 -4.21285957e-01 4.92353514e-02 1.34524608e+00
2.42545933e-01 3.80117744e-01 -6.93758607e-01 -6.34566903e-01
6.46010265e-02 3.63686562e-01 -6.03307009e-01 -2.86650553e-04
5.10836124e-01 2.27415990e-02 -4.31223303e-01 3.68021131e-01
7.90221572e-01 -1.63844839e-01 -3.78437564e-02 -1.43038124e-01
-1.33523732e-01 -3.32460731e-01 -1.60952747e+00 8.70968878e-01
-2.20333442e-01 6.27765179e-01 1.50323689e-01 -1.04560053e+00
9.71338391e-01 2.14494526e-01 4.44814354e-01 -2.65619248e-01
1.88410014e-01 4.40629810e-01 -2.38435447e-01 -2.86616445e-01
2.58399755e-01 -4.70366925e-01 -1.15835525e-01 2.64190644e-01
-1.28708303e-01 1.49594918e-01 -1.06850386e-01 7.97928218e-03
6.94023073e-01 -1.74622610e-01 1.23990142e+00 -9.10383344e-01
5.55835664e-01 -4.06580925e-01 2.88072795e-01 7.81358778e-01
-4.33579892e-01 4.53670472e-01 7.36418784e-01 1.70762226e-01
-1.06216872e+00 -1.30003202e+00 -8.68517756e-01 5.18076956e-01
2.14950629e-02 4.79725264e-02 -8.05748522e-01 -2.57658720e-01
5.52878141e-01 7.20400393e-01 -4.52428490e-01 -3.77888560e-01
-2.87060052e-01 -1.09932828e+00 4.01363283e-01 4.16377038e-01
2.25855842e-01 -3.06692421e-01 -2.18618259e-01 5.91126606e-02
7.32525885e-02 -6.20062172e-01 -8.03159714e-01 2.27728695e-01
-1.13302290e+00 -1.00807321e+00 -9.39026475e-01 -6.57003224e-01
7.00912416e-01 3.32959136e-03 5.21136343e-01 -5.70366263e-01
-1.36917233e-01 5.09972811e-01 1.43564746e-01 -4.36438695e-02
-2.77439088e-01 -1.68228075e-01 5.97492278e-01 2.57789642e-01
4.66994137e-01 -6.53416753e-01 -4.43248719e-01 7.74160028e-01
-7.02462375e-01 -6.50728524e-01 6.86451733e-01 8.50500703e-01
6.06528699e-01 1.04726948e-01 1.00458312e+00 -7.61675060e-01
7.66016424e-01 -5.97186446e-01 -9.68608856e-01 1.35896638e-01
-8.87477279e-01 4.31074649e-01 3.93237710e-01 -7.93157756e-01
-1.02056766e+00 1.68314651e-01 3.28518867e-01 -1.70242921e-01
-7.84320459e-02 4.31208909e-01 -4.05631214e-01 -3.32683623e-01
7.60362267e-01 1.11399829e-01 3.54392588e-01 -5.57938159e-01
3.38191152e-01 8.59679341e-01 4.99751508e-01 -3.49001497e-01
7.20657587e-01 1.63322300e-01 3.24838400e-01 -1.14392543e+00
-5.53520977e-01 -7.53084064e-01 -4.47015256e-01 2.16472093e-02
6.13087952e-01 -6.76017106e-01 -8.85609329e-01 4.47869629e-01
-5.92283189e-01 -1.88283443e-01 -1.17249921e-01 1.00785899e+00
-1.02012646e+00 6.50215924e-01 -5.21417201e-01 -1.34785473e+00
-1.57959387e-01 -7.19579220e-01 7.89404392e-01 2.76785672e-01
-2.89948940e-01 -1.25127804e+00 2.20408082e-01 -2.36749426e-01
2.95907736e-01 1.44274473e-01 8.83743048e-01 -7.73522198e-01
7.66531453e-02 -4.27164376e-01 -4.60502744e-01 1.56783566e-01
1.25319630e-01 -4.55047488e-02 -4.91052806e-01 -3.95404249e-01
3.90863329e-01 2.86546260e-01 6.07608020e-01 9.35530782e-01
7.48939395e-01 -3.98525506e-01 -4.98890907e-01 6.57571077e-01
1.34854698e+00 8.32721889e-02 4.82668310e-01 2.10038379e-01
3.09294045e-01 3.67963016e-01 7.63950884e-01 6.47950590e-01
1.06023684e-01 4.68880147e-01 -2.09435537e-01 2.42155030e-01
6.70180023e-01 -5.56295216e-02 3.79609436e-01 7.54501283e-01
2.65876263e-01 2.97067493e-01 -5.32916248e-01 3.04491431e-01
-2.06094050e+00 -9.18783307e-01 -3.57289493e-01 2.97308898e+00
9.46185887e-01 -4.24037017e-02 3.04978549e-01 -4.59334394e-03
9.14534569e-01 -4.73002672e-01 -4.74986285e-01 -2.34203756e-01
6.71677887e-02 -6.53490573e-02 9.55593407e-01 7.60176063e-01
-1.16163898e+00 3.30916613e-01 7.25202370e+00 1.13661385e+00
-4.02442306e-01 -1.65524557e-01 3.83756548e-01 3.95916730e-01
7.76588172e-02 8.30871835e-02 -8.85395229e-01 4.98703301e-01
9.88124371e-01 -5.36986947e-01 1.33811414e-01 1.16314352e+00
1.56426698e-01 -7.53334284e-01 -9.64324653e-01 1.05609953e+00
-7.33233541e-02 -5.45687377e-01 -3.91336381e-01 4.74409908e-01
5.38122714e-01 -5.10812819e-01 5.33239655e-02 4.74234782e-02
-2.84628779e-01 -7.20919669e-01 3.01353127e-01 7.89997637e-01
9.57574010e-01 -9.52400327e-01 6.32262290e-01 4.13772732e-01
-1.26639843e+00 1.73766643e-01 -6.32440984e-01 -1.30367309e-01
2.48852462e-01 1.07324207e+00 -7.27378964e-01 2.00405359e-01
2.19846502e-01 3.86546105e-01 -4.50959951e-01 1.57298040e+00
3.76649946e-01 3.69429797e-01 -7.82347023e-01 -1.50395676e-01
-2.89472610e-01 -8.07115734e-01 7.22557485e-01 9.58167493e-01
5.19067824e-01 -9.03232321e-02 -1.99266389e-01 9.15463507e-01
4.97144163e-01 4.97416586e-01 -5.60525477e-01 1.70716941e-01
4.76337999e-01 9.33367610e-01 -6.80499732e-01 -1.22102141e-01
-4.39666092e-01 8.28542948e-01 2.97710091e-01 3.80435973e-01
-5.35973012e-01 -8.94098103e-01 7.21121848e-01 2.06655771e-01
2.09759489e-01 -1.83491424e-01 -1.36458084e-01 -1.24896598e+00
-1.30270883e-01 -4.92805451e-01 4.04803008e-01 -5.10744572e-01
-1.56717062e+00 2.29340553e-01 6.88763022e-01 -1.12953293e+00
-5.88554978e-01 -7.40141511e-01 -3.50042880e-01 1.03869987e+00
-9.02298093e-01 -8.16815495e-02 3.36972475e-01 5.60140252e-01
-1.76703632e-01 -9.11400765e-02 6.02451026e-01 2.73950726e-01
-6.83816612e-01 8.53728116e-01 5.68627656e-01 -1.65689379e-01
8.19102108e-01 -1.61352062e+00 1.30855516e-02 6.24229848e-01
-4.53932673e-01 7.68489718e-01 9.36026633e-01 -7.53727496e-01
-1.00636363e+00 -5.87368906e-01 9.58922505e-01 -9.60517377e-02
1.01411164e+00 7.65245929e-02 -1.00324583e+00 4.09604400e-01
-5.12590170e-01 -9.25235003e-02 8.83526325e-01 -2.13647857e-02
-1.34910882e-01 -3.06729525e-02 -1.41401017e+00 4.63940233e-01
4.62155998e-01 -4.82879937e-01 -3.92950505e-01 1.61337376e-01
2.63514817e-01 8.72203091e-05 -1.02594054e+00 3.34006339e-01
8.25052440e-01 -8.88811171e-01 8.32381368e-01 -4.80222702e-01
-3.78897995e-01 -2.38889977e-01 -1.12155229e-01 -1.02748346e+00
-4.50088650e-01 -7.81440318e-01 -1.22950934e-01 1.08280993e+00
4.28795040e-01 -1.20233953e+00 6.09290183e-01 8.73000562e-01
5.28584659e-01 -6.81437194e-01 -1.18378186e+00 -1.10692680e+00
1.40936792e-01 -2.61227638e-01 1.59754157e-02 9.12080646e-01
6.29701018e-01 3.22246134e-01 -3.72981280e-01 2.33632233e-02
1.02460682e+00 -2.78274208e-01 4.77190971e-01 -1.47441483e+00
-4.05614555e-01 -4.36327130e-01 -6.66805863e-01 -1.19490004e+00
2.52605975e-01 -5.42769849e-01 8.52063671e-02 -8.75641167e-01
1.51725680e-01 -1.55994400e-01 -1.49112567e-01 4.03218903e-03
-3.41493309e-01 -1.56892821e-01 -3.74983549e-01 3.66216041e-02
-1.45346254e-01 5.11966288e-01 7.66871035e-01 5.37145197e-01
-5.27389824e-01 5.14918089e-01 -4.87154543e-01 9.43985701e-01
7.55928218e-01 -6.42251551e-01 -2.34283447e-01 6.09389842e-01
3.47879589e-01 4.28456038e-01 1.91071406e-01 -7.95417249e-01
1.36362597e-01 -1.96483359e-01 4.09888387e-01 -3.08989048e-01
5.05340584e-02 -6.99875534e-01 1.11840069e-01 3.02616388e-01
-2.47967109e-01 9.78932381e-02 -1.08462207e-01 7.38345087e-01
1.10951595e-01 -7.56301939e-01 1.12719798e+00 3.18480641e-01
-1.28608242e-01 1.80447958e-02 -8.79997671e-01 6.16769232e-02
8.47270012e-01 -2.51619488e-01 1.31755769e-02 -6.30079806e-01
-6.06720090e-01 -2.02613007e-02 4.37667042e-01 -4.05226111e-01
6.61848307e-01 -1.51686490e+00 -4.83308345e-01 3.59167039e-01
4.49629799e-02 -5.17301857e-01 -2.26409193e-02 1.50800443e+00
-4.77006555e-01 2.25816742e-01 1.10555060e-01 -3.94676030e-01
-8.56338143e-01 4.71392870e-01 5.10114372e-01 3.20877358e-02
-2.31977478e-01 5.80819964e-01 4.22106266e-01 -2.37286642e-01
1.42021835e-01 -1.52003899e-01 -2.39698756e-02 9.24194828e-02
7.09971905e-01 8.00136387e-01 -4.26775545e-01 -6.77898765e-01
-5.41063368e-01 8.19175005e-01 5.54307066e-02 -1.84765726e-01
8.17165375e-01 -2.56982416e-01 -2.66232908e-01 5.94607890e-01
1.59410942e+00 -5.71647398e-02 -1.05982792e+00 -2.38878891e-01
2.65890002e-01 -5.28520405e-01 4.36241403e-02 -6.49353564e-02
-5.06213367e-01 3.74673903e-01 6.05770111e-01 3.21443051e-01
9.86505866e-01 -6.02548048e-02 2.24545315e-01 4.70464736e-01
1.35026932e-01 -1.01922143e+00 -3.47322553e-01 1.31846815e-01
4.91405845e-01 -1.29175448e+00 2.28760406e-01 -5.78929126e-01
-3.60922873e-01 1.01142311e+00 4.50027019e-01 -3.59996766e-01
1.23536825e+00 2.76661724e-01 -3.79375070e-01 3.46024781e-01
-4.38661963e-01 -1.91261739e-01 7.98804522e-01 7.51807928e-01
4.32568043e-01 3.19209062e-02 -5.24623573e-01 6.26310766e-01
-3.11020520e-02 -2.16029704e-01 4.75529701e-01 4.50089753e-01
-6.59812450e-01 -6.19378448e-01 -4.32120591e-01 7.17177391e-01
-2.68714309e-01 -1.25951469e-01 -2.37756953e-01 7.86099136e-01
-6.52828217e-01 7.67052352e-01 6.72825873e-02 1.84651792e-01
2.89845794e-01 1.95851162e-01 5.99539995e-01 -1.38525352e-01
1.82584196e-01 3.56642783e-01 -2.09196746e-01 -4.67386961e-01
-2.37221390e-01 -7.92515576e-01 -7.99683332e-01 -3.88728857e-01
-1.05534947e+00 4.77510601e-01 4.31721240e-01 9.30035591e-01
-6.57315031e-02 -2.87013829e-01 5.75563252e-01 -7.18948305e-01
-1.07906783e+00 -9.07927871e-01 -1.35923171e+00 -6.96766451e-02
3.81262064e-01 -7.11838484e-01 -9.44132924e-01 -3.00575852e-01] | [7.367610454559326, 4.147401332855225] |
c418f946-6754-4e92-adbf-8c79565401f6 | opfython-a-python-inspired-optimum-path | 2001.10420 | null | https://arxiv.org/abs/2001.10420v3 | https://arxiv.org/pdf/2001.10420v3.pdf | OPFython: A Python-Inspired Optimum-Path Forest Classifier | Machine learning techniques have been paramount throughout the last years, being applied in a wide range of tasks, such as classification, object recognition, person identification, and image segmentation. Nevertheless, conventional classification algorithms, e.g., Logistic Regression, Decision Trees, and Bayesian classifiers, might lack complexity and diversity, not suitable when dealing with real-world data. A recent graph-inspired classifier, known as the Optimum-Path Forest, has proven to be a state-of-the-art technique, comparable to Support Vector Machines and even surpassing it in some tasks. This paper proposes a Python-based Optimum-Path Forest framework, denoted as OPFython, where all of its functions and classes are based upon the original C language implementation. Additionally, as OPFython is a Python-based library, it provides a more friendly environment and a faster prototyping workspace than the C language. | ['Alexandre Xavier Falcão', 'João Paulo Papa', 'Gustavo Henrique de Rosa'] | 2020-01-28 | null | null | null | null | ['person-identification'] | ['computer-vision'] | [ 1.79455161e-01 -1.05977394e-01 -5.13780475e-01 -5.16875148e-01
1.30857885e-01 -2.30294243e-01 7.29413629e-01 5.10563016e-01
-3.74969810e-01 8.17689359e-01 -3.27929795e-01 -8.11548531e-01
-3.88727188e-01 -1.10387838e+00 -1.35707259e-01 -4.91665155e-01
-1.10847503e-01 5.76752782e-01 6.22804105e-01 3.72148529e-02
1.69988796e-01 6.41764462e-01 -1.95589113e+00 -1.20441755e-02
1.00766182e+00 1.05121565e+00 -8.90788436e-02 3.32145363e-01
-3.68539512e-01 2.40621656e-01 -3.28603953e-01 -6.81921422e-01
-1.90393105e-01 -1.91744030e-01 -6.68654323e-01 -2.07021296e-01
1.53925382e-02 6.77054096e-03 5.39114513e-02 6.64466977e-01
1.40886590e-01 -2.70083360e-02 7.09578693e-01 -1.41124105e+00
-4.33504879e-02 3.03568453e-01 -4.16177571e-01 -9.00801420e-02
5.81158698e-01 -1.08048245e-01 6.33939862e-01 -5.33236682e-01
3.75669122e-01 1.21079803e+00 7.57335067e-01 4.62174386e-01
-1.26645994e+00 -7.31742740e-01 1.72932297e-01 5.88234305e-01
-1.34839082e+00 9.63873789e-02 6.98872328e-01 -5.14045954e-01
8.96283627e-01 4.82216179e-01 9.05361116e-01 9.38578486e-01
1.83334470e-01 7.18963802e-01 1.55517876e+00 -6.13046229e-01
3.77934486e-01 3.63399357e-01 4.20744419e-01 6.49638951e-01
4.83859777e-01 -2.95961015e-02 -5.00396669e-01 -4.55454230e-01
5.64210355e-01 1.47591308e-01 -8.09063762e-02 -4.73993123e-01
-9.33610260e-01 8.67843568e-01 4.14976597e-01 3.03928941e-01
-3.67419004e-01 -1.72144368e-01 3.25705558e-01 -1.08588375e-01
3.92659336e-01 -3.02946027e-02 -4.40147519e-01 -2.61794001e-01
-8.80779862e-01 3.39467824e-01 1.02130938e+00 7.00855732e-01
6.66015625e-01 -1.17913686e-01 1.20347947e-01 5.65562010e-01
4.18390274e-01 1.54470056e-01 3.23919058e-01 -3.12258601e-01
6.98143989e-02 1.09461224e+00 -1.74638286e-01 -9.13817883e-01
-7.01105416e-01 -7.15582252e-01 -1.07038128e+00 4.59920228e-01
6.53148532e-01 1.29089743e-01 -6.85134530e-01 1.26463091e+00
5.42555153e-01 9.23554227e-02 -2.42455959e-01 5.92361510e-01
7.54708469e-01 3.10650438e-01 3.25071871e-01 -1.75094113e-01
1.32612991e+00 -5.45144677e-01 -3.91164631e-01 -3.15452039e-01
3.22652251e-01 -6.76141262e-01 8.59318674e-01 7.33989716e-01
-3.11778635e-01 -4.79324520e-01 -7.95334637e-01 2.95120776e-01
-6.89636886e-01 -6.83121458e-02 1.10803497e+00 1.30203748e+00
-6.76547050e-01 6.60773277e-01 -7.93781519e-01 -6.79319024e-01
3.43859762e-01 3.82221967e-01 -4.53355610e-01 -8.17874148e-02
-8.49790990e-01 9.81648326e-01 5.72575569e-01 1.37896193e-02
-2.60702550e-01 -1.73680350e-01 -7.96041727e-01 6.22732611e-03
5.05150318e-01 -7.48857260e-01 9.28707004e-01 -6.39893055e-01
-1.41186225e+00 7.38627076e-01 -1.39650315e-01 -5.05193114e-01
6.15238905e-01 -2.11907953e-01 -4.99348015e-01 -2.02119127e-01
-1.69216916e-01 3.28910589e-01 7.48239756e-01 -8.81132007e-01
-5.56919694e-01 -6.18540108e-01 -1.42420411e-01 -1.16905674e-01
-2.51693755e-01 1.96236148e-01 -1.09551340e-01 -4.81760263e-01
6.58704638e-02 -7.84238219e-01 -4.22093928e-01 2.75255833e-03
-4.26393956e-01 -4.86470342e-01 9.19125617e-01 -6.18662298e-01
1.39678276e+00 -1.82397056e+00 -1.44136056e-01 5.45063496e-01
5.48994169e-02 3.61076891e-01 5.46675384e-01 4.43698436e-01
1.51949272e-01 -4.16705534e-02 -4.89171028e-01 -7.11561553e-03
-1.00987136e-01 2.40444198e-01 1.67729393e-01 5.37305057e-01
1.00081868e-01 4.27408546e-01 -7.87479818e-01 -6.23896658e-01
6.09466732e-01 4.18020099e-01 -7.65777305e-02 -3.24100226e-01
-2.00065866e-01 5.66498041e-01 -5.33466816e-01 7.01730967e-01
6.43700898e-01 -1.77427400e-02 3.95755708e-01 4.75048214e-01
-3.03615749e-01 5.74417971e-02 -1.35165596e+00 1.30448818e+00
-2.54608780e-01 3.78807545e-01 -1.06023021e-01 -1.08520496e+00
1.31549418e+00 4.20387834e-02 1.91659123e-01 -2.99449086e-01
5.02287932e-02 2.91871101e-01 -6.02105185e-02 -3.01544100e-01
3.54297757e-01 1.04300544e-01 5.82193360e-02 1.58807844e-01
-8.67543593e-02 -1.16218133e-02 3.13839316e-01 -6.04920872e-02
8.68868232e-01 4.53091949e-01 8.22690845e-01 -1.49130374e-01
7.44324386e-01 6.57569170e-02 3.02982301e-01 5.51059365e-01
-3.02444641e-02 1.36854336e-01 3.59432369e-01 -4.60869879e-01
-6.76451445e-01 -9.37183380e-01 -4.81556416e-01 7.06433177e-01
4.04114928e-03 -4.55258191e-01 -7.62449980e-01 -6.16145730e-01
1.44446015e-01 7.14430511e-01 -1.46767467e-01 1.57725722e-01
-2.95909137e-01 -7.99390435e-01 4.18782622e-01 4.41514999e-01
8.77893269e-01 -8.91093075e-01 -8.03866923e-01 4.30712312e-01
1.79264128e-01 -1.12555194e+00 4.23872203e-01 -5.27692586e-03
-1.23736727e+00 -1.22414339e+00 -5.41333139e-01 -5.14664710e-01
2.67362654e-01 8.82381201e-02 1.19536555e+00 1.57534048e-01
-3.47237676e-01 1.29523918e-01 -3.91459346e-01 -5.36830127e-01
-3.25305820e-01 3.57060313e-01 -8.62387940e-02 3.87367681e-02
6.66740835e-01 -5.83875716e-01 -4.47392672e-01 4.39845651e-01
-6.64010406e-01 8.28839093e-02 6.80769920e-01 8.33205402e-01
3.63513410e-01 4.74043459e-01 3.92682403e-01 -8.85066271e-01
5.03713846e-01 -2.73952127e-01 -5.23288846e-01 1.57302007e-01
-8.64152789e-01 -2.97766149e-01 6.28752589e-01 -1.69934645e-01
-8.74154031e-01 3.10328335e-01 -4.20119524e-01 2.40327641e-01
-6.75430894e-01 6.76433265e-01 -1.72573298e-01 -7.47451782e-02
6.03260279e-01 2.43128628e-01 3.87949171e-03 -6.50358737e-01
1.48862630e-01 9.87903178e-01 4.02744681e-01 -2.98440605e-01
5.81196129e-01 3.88068169e-01 3.12631607e-01 -1.01798499e+00
-3.06298494e-01 -7.42384076e-01 -8.16778719e-01 -5.33527136e-01
7.19579637e-01 -4.23232228e-01 -6.86546028e-01 8.98364604e-01
-9.86461401e-01 -6.06242456e-02 2.49117702e-01 5.91107190e-01
-2.08059952e-01 3.45469862e-01 -5.86876646e-02 -1.25907600e+00
-3.24236274e-01 -8.40375006e-01 6.32423937e-01 5.92178941e-01
-3.75340849e-01 -9.00482953e-01 -2.82496721e-01 4.12184983e-01
3.53802204e-01 5.23482025e-01 1.11426556e+00 -6.82402313e-01
-3.68021578e-01 -7.21704483e-01 -1.41751245e-01 7.47450292e-02
-5.25767244e-02 3.01314592e-01 -1.00287306e+00 -1.20751604e-01
-4.42538738e-01 2.31620550e-01 7.37425387e-01 2.73774087e-01
1.17531300e+00 -1.06014244e-01 -8.17122817e-01 2.97418088e-01
1.30356812e+00 1.16297588e-01 7.42447078e-01 6.91450894e-01
3.17465842e-01 7.25911319e-01 6.17221475e-01 3.66286606e-01
4.17175382e-01 8.94689560e-01 5.83915234e-01 -1.60208046e-01
1.90577298e-01 -2.82092929e-01 -7.74126174e-03 3.69526058e-01
-3.80131721e-01 1.70089230e-01 -1.10617113e+00 1.42230868e-01
-1.90663898e+00 -8.97057474e-01 -6.94994450e-01 2.59380436e+00
5.18822670e-01 4.51550633e-01 5.38264871e-01 7.56995440e-01
6.30554974e-01 -2.24276975e-01 -2.58482128e-01 -5.60220540e-01
1.07772492e-01 4.44075108e-01 4.14722085e-01 9.94652510e-02
-1.23160636e+00 7.39856362e-01 6.38144588e+00 8.70595694e-01
-1.16052020e+00 1.98968649e-02 4.14123863e-01 5.99124193e-01
1.23374835e-01 1.87503293e-01 -9.37753022e-01 4.59275544e-01
7.16223001e-01 4.36099395e-02 2.53131777e-01 1.27738416e+00
1.15673743e-01 -4.36640203e-01 -7.98677921e-01 7.97820568e-01
-8.03606436e-02 -1.02684879e+00 -1.39861405e-01 9.08517167e-02
8.23113322e-02 -3.67293268e-01 -4.29192573e-01 4.21379566e-01
4.72201817e-02 -1.13801622e+00 5.71888566e-01 1.60618708e-01
6.16346538e-01 -7.62287617e-01 9.71356869e-01 5.26222110e-01
-1.18159533e+00 -9.37354341e-02 -8.96231905e-02 -4.33203906e-01
5.79036884e-02 9.47729647e-01 -8.02107692e-01 9.70396101e-01
9.91194069e-01 4.78784084e-01 -7.10380435e-01 1.51865613e+00
-4.03962791e-01 7.80568004e-01 -4.79632050e-01 -3.81167084e-01
-1.14220634e-01 -3.46931547e-01 3.81932646e-01 1.38444066e+00
-8.57691932e-03 -3.09894174e-01 3.93957406e-01 3.83713186e-01
6.73404992e-01 3.73728544e-01 -6.20982766e-01 2.54302949e-01
3.11928540e-01 1.17753184e+00 -1.17129195e+00 -2.33740270e-01
-5.10811985e-01 5.69548607e-01 -1.13404483e-01 1.30668044e-01
-6.48569763e-01 -5.56755304e-01 4.28181946e-01 3.39968055e-01
1.70428995e-02 -5.38694620e-01 -4.68933225e-01 -7.78747976e-01
7.09936842e-02 -8.34629357e-01 4.49906021e-01 -3.64104301e-01
-1.12944627e+00 7.27300048e-01 4.52302039e-01 -1.12419772e+00
-3.92686546e-01 -9.77402806e-01 -5.13426781e-01 7.19617307e-01
-1.36574173e+00 -1.30133808e+00 -5.86631238e-01 5.86766601e-01
1.18059896e-01 -1.68941498e-01 1.07869053e+00 2.48918012e-01
-6.15853727e-01 2.39358097e-01 -6.05619932e-03 1.32192850e-01
3.08617562e-01 -1.25791299e+00 2.53880501e-01 8.07666898e-01
2.60038581e-02 5.72943449e-01 6.71623170e-01 -5.94796896e-01
-1.17659187e+00 -8.33528996e-01 8.41928482e-01 1.38215646e-01
5.81270278e-01 -4.42201763e-01 -8.93197596e-01 3.77359331e-01
-2.25534573e-01 -7.94589520e-02 7.90800154e-01 4.22093838e-01
-2.91537344e-01 -3.44119370e-01 -1.13476038e+00 4.71133858e-01
8.35132420e-01 -2.48430580e-01 -3.57422560e-01 2.64596581e-01
-1.07311994e-01 -2.55336136e-01 -8.92192543e-01 4.48610157e-01
8.58720660e-01 -1.14557219e+00 9.47501004e-01 -3.21468323e-01
-3.26071400e-03 -4.63162422e-01 1.44483328e-01 -9.73432481e-01
-2.01181024e-01 -4.64679599e-01 -2.36309052e-01 1.32221687e+00
3.84487182e-01 -1.22228658e+00 1.00836456e+00 3.99021596e-01
1.45892620e-01 -8.53180885e-01 -1.24560797e+00 -1.04653823e+00
-1.45982862e-01 -5.91937184e-01 7.38567472e-01 7.10100770e-01
3.37795876e-02 2.84138918e-01 -8.52274969e-02 -1.55268848e-01
7.58671284e-01 9.49968770e-02 1.02604461e+00 -1.83290637e+00
-4.87854540e-01 -6.33583128e-01 -1.03287625e+00 -7.33018398e-01
5.55798644e-03 -8.23803782e-01 -2.93393165e-01 -1.74575186e+00
9.72646624e-02 -9.33334410e-01 4.18703780e-02 5.94901323e-01
-2.18405828e-01 1.89219221e-01 4.33012769e-02 1.70471713e-01
-7.45707080e-02 1.34714320e-01 8.37128699e-01 -1.75781682e-01
-2.67683208e-01 6.17974281e-01 -3.13380152e-01 8.44833851e-01
9.12019849e-01 -5.51591337e-01 -5.91139775e-03 1.58296958e-01
-9.29808319e-02 -3.95330846e-01 4.64804858e-01 -1.21239221e+00
1.59779191e-01 -1.24200262e-01 4.40442920e-01 -5.42390764e-01
4.10934240e-01 -8.55523229e-01 4.14737880e-01 8.32070470e-01
2.62008727e-01 -1.69900000e-01 9.68000740e-02 4.66804534e-01
-2.24072322e-01 -4.49216843e-01 6.57895148e-01 4.79118712e-02
-8.95123720e-01 7.93845281e-02 -4.29032862e-01 -5.42885363e-01
1.33546984e+00 -6.32741570e-01 -2.67686546e-01 -2.31613412e-01
-4.76439744e-01 -2.68223621e-02 5.32418191e-01 2.34551534e-01
4.30425435e-01 -7.42874563e-01 -7.41011620e-01 2.25655451e-01
2.13974550e-01 -8.05474594e-02 -5.13770394e-02 8.13980758e-01
-6.27100825e-01 5.14021635e-01 -1.84308931e-01 -8.38570297e-01
-1.66277373e+00 5.01654923e-01 3.30400690e-02 -4.94318336e-01
-7.37905264e-01 4.70146924e-01 -3.81889939e-01 -4.76951927e-01
2.23539725e-01 -2.77493507e-01 -5.11574030e-01 -1.05563380e-01
5.24417281e-01 5.87613165e-01 2.41658211e-01 -3.10147792e-01
-7.05599368e-01 5.91217399e-01 1.82972357e-01 1.11005241e-02
1.17493582e+00 3.27260196e-01 -2.33138517e-01 3.39133918e-01
2.57285893e-01 -1.26325518e-01 -6.80853546e-01 -1.80206187e-02
5.64159155e-01 -5.33586204e-01 -3.80532909e-03 -8.25824320e-01
-6.10177696e-01 8.01195860e-01 5.53493083e-01 5.53133905e-01
1.12927330e+00 -1.50757506e-01 2.80392855e-01 2.07468763e-01
9.60273862e-01 -7.84356952e-01 -6.35354877e-01 1.84553280e-01
5.97267628e-01 -9.86121774e-01 3.80148947e-01 -9.93326783e-01
-4.07915264e-01 1.40926480e+00 2.59017110e-01 2.02713266e-01
7.96725452e-01 1.55053362e-01 -1.25698328e-01 3.50916445e-01
-5.05304337e-01 -4.50764090e-01 1.65201217e-01 1.00057578e+00
4.97285694e-01 4.07487988e-01 -5.85803866e-01 4.14354175e-01
-2.84384429e-01 4.93602097e-01 1.00357920e-01 9.85254765e-01
-3.54676217e-01 -1.34627068e+00 -6.76848710e-01 6.57617390e-01
-3.51250410e-01 1.31187066e-01 -3.78056407e-01 9.87200379e-01
2.58722186e-01 1.19202316e+00 -4.13636081e-02 -4.16078776e-01
2.75293499e-01 3.67694683e-02 5.00298500e-01 -4.90400106e-01
-7.00169265e-01 -3.50127816e-01 3.43695730e-01 -1.86133936e-01
-5.09183824e-01 -6.74530268e-01 -1.02815962e+00 -5.35476267e-01
-4.92391139e-01 1.94873154e-01 1.18816245e+00 1.06458712e+00
7.21238852e-02 3.14285308e-01 2.78371066e-01 -7.57259369e-01
-2.92277932e-01 -1.05538154e+00 -6.33280754e-01 1.39753185e-02
-3.40593368e-01 -9.55355167e-01 3.43543268e-03 -2.94507802e-01] | [8.406439781188965, 4.23612642288208] |
32b36910-caff-41f4-a398-91f2f10940e6 | learning-hierarchical-graph-neural-networks | 2107.01319 | null | https://arxiv.org/abs/2107.01319v2 | https://arxiv.org/pdf/2107.01319v2.pdf | Learning Hierarchical Graph Neural Networks for Image Clustering | We propose a hierarchical graph neural network (GNN) model that learns how to cluster a set of images into an unknown number of identities using a training set of images annotated with labels belonging to a disjoint set of identities. Our hierarchical GNN uses a novel approach to merge connected components predicted at each level of the hierarchy to form a new graph at the next level. Unlike fully unsupervised hierarchical clustering, the choice of grouping and complexity criteria stems naturally from supervision in the training set. The resulting method, Hi-LANDER, achieves an average of 54% improvement in F-score and 8% increase in Normalized Mutual Information (NMI) relative to current GNN-based clustering algorithms. Additionally, state-of-the-art GNN-based methods rely on separate models to predict linkage probabilities and node densities as intermediate steps of the clustering process. In contrast, our unified framework achieves a seven-fold decrease in computational cost. We release our training and inference code at https://github.com/dmlc/dgl/tree/master/examples/pytorch/hilander. | ['David Wipf', 'Stefano Soatto', 'Zheng Zhang', 'Wei Xia', 'Yuanjun Xiong', 'Yongxin Wang', 'Tianjun Xiao', 'Tong He', 'Yifan Xing'] | 2021-07-03 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Xing_Learning_Hierarchical_Graph_Neural_Networks_for_Image_Clustering_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Xing_Learning_Hierarchical_Graph_Neural_Networks_for_Image_Clustering_ICCV_2021_paper.pdf | iccv-2021-1 | ['image-clustering', 'face-clustering'] | ['computer-vision', 'computer-vision'] | [ 2.45163158e-01 4.75742787e-01 -2.07902834e-01 -5.22444487e-01
-4.26868618e-01 -5.74729681e-01 5.42343557e-01 2.97191501e-01
-1.85494602e-01 5.64810574e-01 -1.14202552e-01 -1.59839675e-01
-3.09948176e-01 -9.83053207e-01 -7.27197647e-01 -7.25238144e-01
-2.83264726e-01 1.11021614e+00 -2.82945260e-02 4.61553454e-01
-1.41858123e-02 4.53887463e-01 -1.51712537e+00 -2.01789513e-02
8.21580887e-01 6.08017862e-01 4.55823168e-02 7.67427862e-01
7.29117244e-02 6.96693122e-01 -2.68195778e-01 -5.39207101e-01
3.03240061e-01 -6.26601517e-01 -1.10167730e+00 7.99655169e-02
7.27326930e-01 -1.51069537e-01 -4.87243116e-01 1.16344714e+00
3.39562476e-01 -5.48965819e-02 8.57582867e-01 -1.48536050e+00
-8.10931385e-01 1.06310678e+00 -7.37262905e-01 -2.18066275e-01
-2.02083349e-01 -2.09720537e-01 1.14513588e+00 -4.96198058e-01
8.63283515e-01 1.08193123e+00 9.55259919e-01 4.89400148e-01
-1.64206517e+00 -7.02842772e-01 -7.24350885e-02 5.21170832e-02
-1.82734311e+00 -4.24165577e-01 3.71229649e-01 -5.44764757e-01
5.25627434e-01 -7.28883222e-02 6.09063029e-01 5.19685149e-01
-3.38458925e-01 5.57076097e-01 7.87681043e-01 -2.80204982e-01
1.55354023e-01 -2.22618029e-01 1.31355658e-01 1.03864324e+00
3.82379621e-01 -3.29478770e-01 -4.72303420e-01 -5.57877272e-02
9.87062335e-01 -1.69769496e-01 9.61432233e-03 -6.45883858e-01
-9.21074092e-01 8.16596270e-01 7.42153108e-01 3.36323172e-01
-9.09253806e-02 3.04274619e-01 3.23580168e-02 -1.31339371e-01
3.64247322e-01 4.85817373e-01 -2.69514352e-01 2.89270431e-01
-1.16347373e+00 -1.12030961e-01 9.62847769e-01 9.20665145e-01
9.46111262e-01 -2.84036547e-01 2.45048121e-01 7.62721837e-01
3.10131133e-01 8.59460048e-03 2.82812472e-02 -1.46578872e+00
-1.80786103e-01 8.54961693e-01 -4.17928189e-01 -1.09439933e+00
-4.73914325e-01 -5.77898026e-01 -1.05931544e+00 -7.88743496e-02
5.77032864e-01 -9.86497700e-02 -1.11679590e+00 1.92959476e+00
4.40652907e-01 2.76173711e-01 -2.89059043e-01 5.88486016e-01
8.48708630e-01 3.54123414e-01 -5.77775203e-02 4.16242741e-02
9.17410195e-01 -8.65418017e-01 -2.77553886e-01 1.32669047e-01
5.92656374e-01 -4.15429950e-01 4.01132792e-01 2.40637168e-01
-1.17907035e+00 -5.61276376e-01 -9.35564756e-01 -2.03270778e-01
-4.98977691e-01 1.72418445e-01 8.61640692e-01 4.67672527e-01
-1.72253609e+00 8.32903624e-01 -9.89888668e-01 -8.59414279e-01
6.56166494e-01 8.61846149e-01 -4.58398640e-01 -1.05047682e-02
-6.42282426e-01 4.34477955e-01 6.43119991e-01 1.15824059e-01
-8.82572830e-01 -5.05498707e-01 -7.88262308e-01 1.03779711e-01
2.16469437e-01 -7.70854354e-01 8.15544069e-01 -8.76085281e-01
-9.00346756e-01 1.19379127e+00 -3.06158274e-01 -4.67622876e-01
2.86553711e-01 1.04218319e-01 -1.66501552e-02 3.40023458e-01
3.40055943e-01 1.50918686e+00 2.32160136e-01 -1.62203038e+00
-6.03148639e-01 -3.66196483e-01 -1.18611433e-01 1.71589792e-01
-2.99099118e-01 -2.94384688e-01 -9.20777142e-01 -3.29695582e-01
3.77931386e-01 -9.37433124e-01 -2.12622717e-01 2.69867275e-02
-5.99231482e-01 -3.10821861e-01 3.70676339e-01 -6.61344111e-01
1.07858944e+00 -1.72428119e+00 3.12872112e-01 4.82117504e-01
7.01063693e-01 -1.20494582e-01 -1.42885298e-01 4.29344118e-01
-5.55450609e-03 3.47897589e-01 -4.13649917e-01 -3.99805099e-01
-3.83475833e-02 1.41834602e-01 3.19081753e-01 5.28649211e-01
6.02293015e-02 7.25432456e-01 -9.01984394e-01 -8.23753774e-01
2.03074649e-01 5.07451892e-01 -5.85247397e-01 8.59989002e-02
-1.68071911e-02 4.79993552e-01 1.65116608e-01 6.69502974e-01
5.15145183e-01 -9.82190490e-01 7.98080623e-01 -1.96396280e-02
2.24038318e-01 1.00167997e-01 -1.05156863e+00 1.66851997e+00
6.46909624e-02 6.57621443e-01 1.93725288e-01 -1.04934728e+00
8.06742489e-01 -4.09119576e-02 6.05479121e-01 -8.51717778e-03
2.12586224e-01 -1.79470718e-01 1.75212637e-01 4.99787740e-02
2.51764655e-01 3.39430273e-01 1.20771430e-01 3.27389121e-01
4.60127503e-01 3.85916650e-01 6.21369720e-01 6.41120732e-01
1.31201720e+00 2.66927660e-01 1.48751274e-01 -2.90172547e-01
1.58071682e-01 1.12805694e-01 7.85602748e-01 9.27606761e-01
-6.10598214e-02 7.01052129e-01 7.07988322e-01 -3.02637875e-01
-1.21458328e+00 -1.19012141e+00 -2.35197112e-01 1.21678984e+00
1.29417509e-01 -4.79467005e-01 -1.06201196e+00 -4.68436390e-01
4.18384299e-02 3.74518424e-01 -8.62073243e-01 2.11930722e-01
-3.83220583e-01 -8.19850266e-01 7.39397645e-01 3.66015494e-01
4.14231628e-01 -9.29634392e-01 1.05433613e-01 -9.83805135e-02
-1.43569708e-01 -9.47594821e-01 -1.61369070e-01 3.05325478e-01
-7.52201200e-01 -1.27374470e+00 -3.40336680e-01 -1.14143932e+00
1.31509328e+00 -3.35095227e-02 1.22583127e+00 5.40481567e-01
-3.65484595e-01 9.19943601e-02 -1.66457355e-01 5.66427261e-02
-4.37719285e-01 4.98063147e-01 6.75512999e-02 -2.03554019e-01
4.92346227e-01 -9.62614179e-01 -5.77505231e-01 2.71974176e-01
-6.64605856e-01 2.22331315e-01 4.79393572e-01 7.51934588e-01
6.10167086e-01 2.39085525e-01 2.86123186e-01 -1.08151865e+00
2.01015249e-01 -5.79944789e-01 -8.38088989e-01 3.15607905e-01
-6.54318333e-01 -6.05298765e-03 4.99147266e-01 -1.35163069e-01
-7.40829945e-01 3.68198037e-01 1.40960023e-01 -3.41760457e-01
-4.89130735e-01 4.38631922e-01 -6.73795119e-02 8.40004906e-02
4.57734913e-01 -2.93580219e-02 1.53703108e-01 -3.14531177e-01
5.89853942e-01 5.33530712e-01 8.55996251e-01 -4.85194772e-01
8.06726635e-01 5.34564674e-01 5.79186417e-02 -5.24463475e-01
-6.88861132e-01 -4.43071902e-01 -1.42769814e+00 -3.16604108e-01
9.71699417e-01 -9.79887307e-01 -1.02988696e+00 4.84866887e-01
-8.89299989e-01 -6.95659816e-01 5.01382239e-02 1.59493640e-01
-4.04279172e-01 4.40408587e-01 -1.01588953e+00 -5.22235632e-01
-2.25988120e-01 -7.59422064e-01 9.11959767e-01 4.00496840e-01
-3.90956908e-01 -9.98332024e-01 -9.23962519e-02 6.17052972e-01
-1.69956192e-01 3.27143729e-01 9.47848439e-01 -8.21715295e-01
-7.39763319e-01 -1.00603001e-02 -4.35897231e-01 7.82890841e-02
1.67118430e-01 1.98170066e-01 -6.71701670e-01 -3.62849861e-01
-6.24170184e-01 -3.38998139e-01 8.84753406e-01 6.53668165e-01
1.21539867e+00 -1.81879953e-01 -6.59024477e-01 9.15541828e-01
1.51148963e+00 5.02850190e-02 3.76083136e-01 5.85918352e-02
1.09623337e+00 7.04751015e-01 -5.83888032e-03 1.36950925e-01
6.68324053e-01 2.34653324e-01 2.97250748e-01 -2.36828685e-01
-9.24070254e-02 -3.47674936e-01 -9.47970897e-02 8.02593172e-01
-7.28869289e-02 -2.59855211e-01 -1.23548448e+00 7.47484267e-01
-2.00433898e+00 -7.43514836e-01 -1.83002457e-01 2.15645003e+00
9.03727829e-01 -8.61033518e-03 2.28866592e-01 -2.52963603e-01
1.29109156e+00 -1.95436686e-01 -7.63852715e-01 1.70446634e-01
-2.70609260e-02 -5.22314273e-02 7.32997358e-01 5.94594598e-01
-1.25251210e+00 1.23448110e+00 6.53736115e+00 7.85997093e-01
-6.46825790e-01 -2.35055983e-01 1.02255845e+00 -6.00123219e-02
5.12778722e-02 1.62713155e-01 -8.64208877e-01 3.57030541e-01
1.09478676e+00 -6.81818649e-02 6.61579430e-01 5.27400970e-01
-1.31019354e-01 -1.32768288e-01 -1.09554541e+00 5.85680187e-01
1.76653136e-02 -1.41293919e+00 7.09439963e-02 4.29888636e-01
8.48453343e-01 2.24404007e-01 -2.30479896e-01 -5.70092425e-02
1.33243167e+00 -1.21755207e+00 2.58379698e-01 3.55426878e-01
7.82674968e-01 -8.86017144e-01 6.23179793e-01 2.28053659e-01
-1.25536394e+00 1.45626187e-01 -4.47004050e-01 -9.00403112e-02
3.99504118e-02 5.76555431e-01 -8.66669357e-01 5.65627217e-01
1.00824916e+00 7.00345993e-01 -8.68928492e-01 8.62770736e-01
-2.83104479e-01 7.32490301e-01 -4.69483614e-01 3.01747203e-01
1.05373822e-01 -4.46799755e-01 2.65820548e-02 9.36868608e-01
1.74566478e-01 5.51940911e-02 2.78566420e-01 9.80678439e-01
-4.90793526e-01 -2.01684222e-01 -6.43944085e-01 -1.41721159e-01
9.04484093e-01 1.65949333e+00 -1.38918245e+00 -3.05261642e-01
-2.03246862e-01 8.95551205e-01 8.30137372e-01 3.46852899e-01
-8.21680248e-01 -3.78098428e-01 3.87620926e-01 1.93770334e-01
5.16629994e-01 -2.54014462e-01 -1.97309539e-01 -7.59782732e-01
-3.40010434e-01 -5.82467318e-01 6.00829244e-01 -8.01470459e-01
-1.51450109e+00 3.21108073e-01 -2.97057256e-02 -7.00939059e-01
-1.05718873e-01 -4.22993422e-01 -2.54029512e-01 5.68215072e-01
-7.71117568e-01 -1.46651745e+00 -2.78443635e-01 2.01933354e-01
-1.31757208e-03 -1.59012958e-01 7.90629387e-01 3.14771503e-01
-7.27574110e-01 5.06263793e-01 3.11413854e-01 8.40068758e-01
6.59654319e-01 -1.57933486e+00 5.00756860e-01 8.90152872e-01
2.05240875e-01 9.02338028e-01 4.38971162e-01 -8.05853844e-01
-8.43935847e-01 -1.22107410e+00 7.84727216e-01 -4.10139084e-01
6.82996273e-01 -6.67364657e-01 -7.60850787e-01 1.06613052e+00
3.37174475e-01 -1.83536008e-01 1.03242970e+00 4.96889502e-01
-4.01950479e-01 -1.37949372e-02 -1.03764462e+00 6.52728558e-01
1.39643550e+00 -2.64097065e-01 -5.94797358e-03 3.07020277e-01
6.59639478e-01 -6.60973191e-02 -1.43154013e+00 3.78252506e-01
5.52963436e-01 -1.02945960e+00 8.27259123e-01 -3.03795099e-01
6.84751272e-01 -5.77307940e-01 1.33204795e-02 -8.71822357e-01
-6.30733371e-01 -4.07073706e-01 6.49586320e-02 1.61034954e+00
4.19729948e-01 -3.66950601e-01 1.27310443e+00 4.27633047e-01
2.02068180e-01 -6.61305785e-01 -6.30617023e-01 -5.05704820e-01
1.49641678e-01 3.29299532e-02 3.09232920e-01 1.15644002e+00
-8.67832080e-02 6.01606250e-01 -2.43738607e-01 3.38070333e-01
1.14955592e+00 -1.59822255e-02 9.01745617e-01 -1.62778616e+00
-1.32362336e-01 -4.75123823e-01 -3.86994541e-01 -8.90822053e-01
2.81800866e-01 -1.31176579e+00 9.90055501e-02 -2.03701949e+00
5.48931301e-01 -4.17843908e-01 -2.10381135e-01 7.82660842e-01
-6.34188056e-02 5.67703903e-01 8.26947391e-02 3.52002770e-01
-9.95205700e-01 2.23332196e-01 5.89490891e-01 -1.00495834e-02
-1.91061497e-01 -4.16291445e-01 -8.54940772e-01 8.66170883e-01
8.15737247e-01 -5.19499540e-01 -2.90423453e-01 -4.12257701e-01
1.12679578e-01 -1.14962935e-01 3.06164950e-01 -1.29226315e+00
6.67206109e-01 2.48650864e-01 8.49400640e-01 -7.94866920e-01
1.94111109e-01 -5.89491487e-01 4.97259647e-01 3.46611202e-01
-5.15082836e-01 -7.52689168e-02 1.92807391e-02 3.89318228e-01
2.92982236e-02 4.74028774e-02 6.75081909e-01 -2.35825270e-01
-4.63553727e-01 3.10517520e-01 -4.04201716e-01 -2.29773611e-01
7.99277902e-01 -3.46815199e-01 -5.39586365e-01 -4.65375870e-01
-8.99672687e-01 5.53375185e-01 8.94231558e-01 1.10598661e-01
1.91771388e-01 -1.16217661e+00 -6.55301094e-01 -7.46100396e-02
-1.36371091e-01 3.42219293e-01 1.54016346e-01 6.67590320e-01
-7.08795309e-01 8.58319625e-02 -2.94890672e-01 -7.64713943e-01
-1.47882247e+00 3.53753448e-01 2.82665849e-01 -3.23030680e-01
-2.61657029e-01 1.04201984e+00 2.22194716e-01 -9.93239462e-01
2.97916830e-01 3.74761552e-01 -4.24260981e-02 4.93027419e-02
-9.75359380e-02 4.48718548e-01 -4.37028885e-01 -8.21849525e-01
-1.96309641e-01 4.68653709e-01 -2.84255415e-01 6.80576488e-02
1.50705338e+00 -1.70614213e-01 -7.48598278e-01 3.41757357e-01
1.15978193e+00 -4.40609783e-01 -1.24261689e+00 -1.50640726e-01
3.44398826e-01 -2.79866587e-02 -2.15063155e-01 -7.52070606e-01
-1.06935954e+00 4.39715296e-01 4.33322340e-01 2.07166821e-01
1.04682636e+00 4.18236971e-01 4.63971198e-01 4.35997784e-01
1.72821537e-01 -1.00483513e+00 -2.03504056e-01 2.39214376e-01
2.69930661e-01 -1.11119497e+00 2.75295079e-01 -6.24944687e-01
-2.27791548e-01 7.16763496e-01 7.55667567e-01 -9.21557769e-02
6.02644920e-01 3.57143432e-01 1.63275842e-02 -4.20367867e-01
-6.24646902e-01 -3.94636512e-01 2.96349525e-02 7.02875555e-01
5.48920155e-01 6.84342533e-02 -7.25742988e-03 1.13022782e-01
-4.16619599e-01 -1.66895717e-01 4.22021955e-01 5.39269745e-01
-3.17308635e-01 -1.08095157e+00 -1.36452347e-01 6.99515522e-01
-3.92358631e-01 -1.59391344e-01 -8.20035517e-01 8.47959280e-01
3.11300188e-01 9.74795580e-01 4.62764204e-01 -4.71702188e-01
-3.39167625e-01 9.59788822e-03 5.28243065e-01 -6.81430936e-01
-4.22283590e-01 2.17183992e-01 9.64313745e-02 -3.57747257e-01
-7.03836381e-01 -7.53387570e-01 -1.47796917e+00 -5.70848346e-01
-2.25866556e-01 1.62843227e-01 3.99324089e-01 5.59266686e-01
5.63385904e-01 3.34493667e-01 2.43792132e-01 -8.11451495e-01
3.05326611e-01 -9.05852079e-01 -7.07376420e-01 4.66963440e-01
-1.61394045e-01 -4.34240252e-01 -2.99281210e-01 3.38363171e-01] | [9.119307518005371, 3.3062386512756348] |
354f672a-9d33-430f-9692-56b5bdef359d | transforming-bells-inequalities-into-state | 1705.00813 | null | http://arxiv.org/abs/1705.00813v1 | http://arxiv.org/pdf/1705.00813v1.pdf | Transforming Bell's Inequalities into State Classifiers with Machine Learning | Quantum information science has profoundly changed the ways we understand,
store, and process information. A major challenge in this field is to look for
an efficient means for classifying quantum state. For instance, one may want to
determine if a given quantum state is entangled or not. However, the process of
a complete characterization of quantum states, known as quantum state
tomography, is a resource-consuming operation in general. An attractive
proposal would be the use of Bell's inequalities as an entanglement witness,
where only partial information of the quantum state is needed. The problem is
that entanglement is necessary but not sufficient for violating Bell's
inequalities, making it an unreliable state classifier. Here we aim at solving
this problem by the methods of machine learning. More precisely, given a family
of quantum states, we randomly picked a subset of it to construct a
quantum-state classifier, accepting only partial information of each quantum
state. Our results indicated that these transformed Bell-type inequalities can
perform significantly better than the original Bell's inequalities in
classifying entangled states. We further extended our analysis to three-qubit
and four-qubit systems, performing classification of quantum states into
multiple species. These results demonstrate how the tools in machine learning
can be applied to solving problems in quantum information science. | ['Man-Hong Yung', 'Yue-Chi Ma'] | 2017-05-02 | null | null | null | null | ['quantum-state-tomography'] | ['medical'] | [ 3.02640736e-01 -2.14491025e-01 -7.46859908e-02 -1.96294203e-01
-5.35576344e-01 -8.81385088e-01 3.66632640e-01 4.15120423e-01
-5.96933544e-01 7.90988207e-01 -4.15869504e-01 -7.95719028e-01
-8.49314779e-02 -1.22867954e+00 -2.56244689e-01 -1.01277435e+00
-1.69393539e-01 5.65123856e-01 6.25969991e-02 -2.91569918e-01
7.56649256e-01 6.89259589e-01 -1.36907649e+00 2.43519098e-01
6.57403708e-01 7.66508937e-01 -1.57035008e-01 9.56781268e-01
-2.35727355e-01 5.32243073e-01 -3.35811853e-01 -2.15325385e-01
2.40757138e-01 -8.53654385e-01 -1.34609377e+00 -3.35609585e-01
-5.68161458e-02 -1.30489126e-01 -3.78774971e-01 1.64800024e+00
-5.28580621e-02 -2.32486986e-02 7.30829597e-01 -1.08108664e+00
-3.06492358e-01 3.73549700e-01 3.05289716e-01 2.08828375e-01
6.95934594e-01 -6.03955127e-02 1.19357193e+00 -9.09662172e-02
4.15147960e-01 7.36489356e-01 8.32744837e-02 4.93642956e-01
-1.44389760e+00 -6.09476149e-01 -9.48656619e-01 7.19352722e-01
-1.60529935e+00 -3.86494130e-01 4.41698730e-01 -3.90838355e-01
8.95750403e-01 4.65394557e-01 7.68379271e-01 3.68893683e-01
6.10847235e-01 3.00190300e-01 1.61947238e+00 -8.71996343e-01
4.43409026e-01 3.43126595e-01 4.80012774e-01 8.66454363e-01
2.49016911e-01 4.98048663e-01 -3.40137422e-01 -5.54097034e-02
3.83038878e-01 -1.27679259e-01 -9.63462815e-02 -2.40120918e-01
-1.20370352e+00 9.14503157e-01 3.86413842e-01 6.00839674e-01
1.42131150e-02 -1.73003048e-01 1.90784082e-01 7.60058820e-01
-8.24246258e-02 8.79979312e-01 -1.93840638e-01 -2.47510538e-01
-5.94361901e-01 -6.76822737e-02 1.20195663e+00 5.03060520e-01
1.36225140e+00 -6.89661503e-01 7.91050792e-02 -1.19027913e-01
3.72921228e-02 6.27096474e-01 5.38642146e-02 -8.67076695e-01
-7.89788901e-04 3.88548553e-01 2.37130880e-01 -7.11354077e-01
-4.45765644e-01 8.60978365e-02 -9.81130183e-01 1.32152304e-01
7.86801815e-01 1.34434223e-01 -7.75091887e-01 1.52815413e+00
8.14583153e-02 -6.92630559e-02 3.52396816e-01 8.07334602e-01
4.09103930e-01 8.70812118e-01 -3.55567127e-01 -3.48237514e-01
1.20576191e+00 -2.38502234e-01 -6.77570760e-01 2.26337343e-01
1.15677512e+00 -7.60803699e-01 3.61945659e-01 3.70236039e-01
-7.54508674e-01 -3.11297059e-01 -1.05632091e+00 1.77387133e-01
-5.15338898e-01 -3.42108697e-01 1.06517315e+00 1.14580882e+00
-7.56762326e-01 1.05258083e+00 -7.35004961e-01 -1.66359231e-01
-3.17145139e-02 6.95413113e-01 -8.28462660e-01 -1.42078638e-01
-1.35555267e+00 1.18280935e+00 7.23286390e-01 1.94966540e-01
-4.22327459e-01 3.10294539e-01 -7.59326637e-01 1.81969404e-01
1.66976839e-01 -5.07938743e-01 9.50593472e-01 -4.94538009e-01
-1.58896720e+00 9.20955598e-01 -3.68550479e-01 -1.40435696e-01
-2.77714998e-01 6.67842984e-01 -4.23909038e-01 2.26659894e-01
-1.98562816e-02 -1.31184116e-01 3.62734437e-01 -6.17036819e-01
-3.21166575e-01 -4.76227641e-01 3.14306647e-01 -2.41830617e-01
2.28782833e-01 5.66669914e-04 1.92638308e-01 7.48750925e-01
9.47196066e-01 -1.37767553e+00 -1.75816402e-01 -6.37618601e-01
-6.03757560e-01 -3.60484511e-01 1.29563287e-01 5.62392697e-02
9.58720386e-01 -1.94723678e+00 2.56368130e-01 6.13213539e-01
1.12018496e-01 3.30978572e-01 -1.86941206e-01 8.21679115e-01
-1.49529502e-01 2.23312855e-01 -7.84395486e-02 -1.76241659e-02
1.07197016e-01 4.59735394e-01 -6.71629086e-02 7.42961347e-01
2.45890036e-01 6.53277576e-01 -9.31144059e-01 -4.56128716e-01
1.95938736e-01 -6.81021586e-02 -7.65073121e-01 1.24935031e-01
3.55439901e-01 8.67370188e-01 -5.50136209e-01 4.04113770e-01
7.51450479e-01 -2.03645080e-01 3.43257844e-01 -1.37840748e-01
-4.29181457e-01 7.63336658e-01 -1.44517267e+00 1.48345113e+00
-2.68524528e-01 5.92966557e-01 -2.57723987e-01 -1.42146075e+00
7.30920553e-01 3.40158731e-01 3.18118542e-01 -9.05141413e-01
4.74377960e-01 5.70238113e-01 5.94433784e-01 -6.40011907e-01
5.26001334e-01 -9.14309204e-01 -4.30074126e-01 6.72343433e-01
2.23141089e-01 -5.79760849e-01 6.06957018e-01 2.09556341e-01
1.38887489e+00 -3.56057346e-01 2.82222301e-01 -1.85044765e-01
6.09852672e-01 -5.38423769e-02 4.42717284e-01 9.36818898e-01
-4.68645096e-01 2.78176099e-01 6.83028817e-01 -3.44287574e-01
-1.06133890e+00 -1.14468086e+00 -6.53762877e-01 3.72250021e-01
4.90197808e-01 -5.83428860e-01 -5.16539931e-01 -2.00623482e-01
-2.62292176e-01 5.35506189e-01 -2.90884107e-01 -4.40439075e-01
-9.46564153e-02 -9.14107263e-01 4.36693609e-01 -1.32815108e-01
4.01858151e-01 -8.98232341e-01 -4.89481062e-01 1.31189913e-01
-2.27812335e-01 -8.82541955e-01 2.13874876e-01 6.75892234e-01
-7.75560439e-01 -1.14668977e+00 4.12961245e-02 -2.75507092e-01
6.70586705e-01 5.46774874e-03 8.31880212e-01 8.11179057e-02
-1.95041046e-01 -3.76857817e-02 -5.50243199e-01 6.14045449e-02
-9.88022387e-01 -2.89496197e-03 3.35134864e-01 4.50482918e-03
5.67932308e-01 -2.90724307e-01 -2.01273307e-01 3.82613204e-02
-1.12059665e+00 -2.43349418e-01 7.83163130e-01 8.59669268e-01
2.19580337e-01 4.54273224e-01 9.29360837e-02 -8.24832678e-01
4.42868769e-01 -1.92635626e-01 -6.80804670e-01 2.92108208e-01
-2.92816520e-01 7.92590320e-01 8.52243721e-01 1.36167973e-01
-5.39786756e-01 2.61245966e-01 -3.21684718e-01 3.32521200e-01
-2.45170921e-01 7.04730690e-01 9.04205218e-02 -4.02940691e-01
5.73817909e-01 4.03748274e-01 -1.66937843e-01 -5.95714338e-02
2.75004834e-01 8.86170268e-01 3.08710963e-01 -5.34098804e-01
8.82587194e-01 3.41646075e-01 8.79379570e-01 -1.05697751e+00
-9.12870944e-01 -8.82971406e-01 -1.15760660e+00 1.76461980e-01
9.21555758e-01 -2.65356719e-01 -1.27477181e+00 3.08859318e-01
-1.25364900e+00 2.43030801e-01 -6.93894923e-02 7.84505546e-01
-4.58767503e-01 8.48172784e-01 -7.21367478e-01 -1.23502910e+00
1.44144386e-01 -1.29529417e+00 7.97323287e-01 3.66911709e-01
6.38672635e-02 -9.55491424e-01 5.03404677e-01 2.73116887e-01
2.36942157e-01 -2.10149631e-01 9.80041981e-01 -5.96959352e-01
-8.04567695e-01 -7.16130733e-01 -2.52599508e-01 4.05420989e-01
4.20546271e-02 -1.07267402e-01 -7.93627560e-01 -2.81684428e-01
1.23511650e-01 -5.18553257e-01 7.75734127e-01 -2.85609365e-01
6.05482280e-01 7.50502571e-02 -1.57433182e-01 4.58951771e-01
1.41626608e+00 2.98292041e-01 8.69448781e-01 5.59315942e-02
3.04594398e-01 3.37160498e-01 2.92497635e-01 1.60002053e-01
1.03331380e-01 6.10536516e-01 1.57792717e-01 3.74350011e-01
5.83524704e-01 5.69965132e-03 1.76409662e-01 1.16821635e+00
-3.34558785e-01 5.87467104e-03 -7.37875760e-01 -4.00570519e-02
-1.40809941e+00 -1.41817689e+00 -5.44422746e-01 2.50608373e+00
9.03108180e-01 7.15909898e-02 -2.20006973e-01 6.12357259e-01
4.15646672e-01 -2.89699584e-01 -5.27109131e-02 -8.40601921e-01
-1.76996645e-02 8.29812109e-01 4.97611940e-01 6.92874908e-01
-1.10853267e+00 6.93453312e-01 6.44074774e+00 3.64742488e-01
-1.06533098e+00 3.32978852e-02 -7.57696927e-02 5.32278955e-01
-1.65534019e-01 7.24041402e-01 -5.88306010e-01 3.46750110e-01
1.29964817e+00 -3.13674733e-02 7.32565582e-01 2.19197527e-01
-1.82882085e-01 -7.81781733e-01 -1.30344355e+00 1.08899617e+00
-2.56712973e-01 -1.12469435e+00 -2.19657481e-01 2.84784019e-01
6.94635391e-01 -1.01650372e-01 -3.83355707e-01 4.11387742e-01
-7.96301067e-02 -9.71010923e-01 3.55621815e-01 5.35854578e-01
6.01868808e-01 -4.52998906e-01 1.02757764e+00 6.18948400e-01
-1.00452983e+00 3.12453732e-02 -6.04927480e-01 -6.24202967e-01
2.25901883e-02 5.58358967e-01 -6.40446126e-01 6.92466974e-01
4.19734903e-02 2.16953442e-01 -3.38836849e-01 1.19688916e+00
-2.15177178e-01 3.71983051e-01 -5.61207235e-01 -3.60481501e-01
2.20814198e-01 -7.81113207e-01 2.61187702e-01 9.86362696e-01
3.29277575e-01 5.04765272e-01 1.02109229e-02 1.08140814e+00
2.69922525e-01 -7.92976543e-02 -7.92218029e-01 -4.61287498e-01
2.40360707e-01 1.12267268e+00 -8.39950323e-01 -3.75834137e-01
-2.75101364e-01 1.19112980e+00 2.41059512e-01 -1.47329289e-02
-4.20337737e-01 -6.36619806e-01 3.25088441e-01 -3.81246358e-01
-6.77605867e-02 -5.57884216e-01 5.65290358e-03 -1.66362190e+00
-3.52094352e-01 -4.49342847e-01 8.23147669e-02 -3.12147468e-01
-1.04608154e+00 3.90038818e-01 -2.02296332e-01 -1.10468066e+00
-2.43468195e-01 -9.20991242e-01 -5.22413254e-01 1.15403616e+00
-1.44156337e+00 -5.16674221e-01 5.82954474e-02 4.93535846e-01
-5.99510193e-01 2.27341115e-01 1.40412295e+00 2.15605631e-01
-5.61086535e-01 1.12863958e-01 4.05415863e-01 1.41972825e-01
3.78399700e-01 -1.39895451e+00 -2.11874247e-01 7.76566565e-01
5.73845744e-01 7.53249943e-01 7.96514392e-01 -1.96637377e-01
-2.04908299e+00 -2.00098798e-01 1.22668207e+00 -4.46166992e-01
8.19716990e-01 -2.67787993e-01 -7.67458320e-01 3.55474561e-01
-1.59972891e-01 9.89304483e-02 8.09216082e-01 4.32616472e-01
-3.31939638e-01 1.28616542e-02 -9.85829353e-01 1.20363142e-02
6.16866469e-01 -1.28260255e+00 -6.04465902e-01 4.62892681e-01
-1.75434157e-01 -1.71045259e-01 -7.47277319e-01 8.23173076e-02
6.26594365e-01 -1.30419946e+00 4.50693041e-01 -8.29252303e-01
1.59434244e-01 -5.60415983e-01 -2.26083815e-01 -1.13760245e+00
-3.27491492e-01 -5.82387269e-01 4.91946578e-01 4.51994181e-01
3.64922851e-01 -8.04278314e-01 7.54147768e-01 6.46801472e-01
2.48546153e-02 -1.68584973e-01 -1.33351576e+00 -9.77975428e-01
3.05707186e-01 -4.09405529e-01 1.28350228e-01 8.81313503e-01
8.02487850e-01 5.38189948e-01 -1.17101371e-01 2.15850458e-01
6.02166414e-01 4.90909338e-01 6.38015509e-01 -1.24987555e+00
-3.58428299e-01 -3.31134856e-01 -9.68145549e-01 -1.00829697e+00
3.14337671e-01 -1.30480063e+00 1.44465059e-01 -1.24197352e+00
4.32301700e-01 -6.36226475e-01 -4.85811949e-01 2.14716062e-01
1.40051559e-01 3.93542051e-01 3.25455457e-01 3.12269002e-01
-6.80163920e-01 2.11721286e-01 1.29317021e+00 -1.67662859e-01
4.25497293e-02 2.56191313e-01 -9.21585187e-02 2.69344687e-01
6.50935769e-01 -6.51058257e-01 3.18103909e-01 1.52723968e-01
7.27704465e-01 4.69092935e-01 2.78566480e-01 -1.14934814e+00
5.76631248e-01 -1.67566180e-01 -2.52304435e-01 -3.78238946e-01
4.65985596e-01 -6.58123195e-01 1.76353574e-01 8.28358889e-01
1.57406349e-02 -3.92379105e-01 -2.29678646e-01 2.39167958e-01
-3.41446221e-01 -9.21054482e-01 9.19494510e-01 -2.19300032e-01
-7.08035946e-01 1.20203145e-01 -5.06937563e-01 -3.72081876e-01
8.35492313e-01 7.40153790e-02 -2.98006773e-01 -1.41578302e-01
-8.86201262e-01 -3.29042554e-01 4.94106978e-01 -3.23488891e-01
3.75646621e-01 -1.05592847e+00 -3.72317135e-01 6.07981563e-01
1.59191802e-01 -5.77456474e-01 1.33173540e-01 1.00354648e+00
-7.75161326e-01 9.42698896e-01 -3.76619637e-01 -5.88575125e-01
-9.16261256e-01 5.74743807e-01 4.49014127e-01 -2.25448877e-01
2.39474990e-04 5.16580820e-01 -2.40595773e-01 -3.44714165e-01
-5.09590209e-01 -5.25079012e-01 1.56660497e-01 -1.55096322e-01
3.85716110e-01 1.20739743e-01 1.12768836e-01 -8.95786464e-01
-4.83239442e-01 6.93687260e-01 2.69619226e-01 -1.16899379e-01
8.91500533e-01 5.95443211e-02 -6.45107031e-01 6.43742800e-01
1.55060196e+00 1.99783267e-03 -2.28276476e-01 -9.02956873e-02
-1.24749225e-02 -3.88354123e-01 2.05693264e-02 -3.67637813e-01
-4.72298384e-01 1.08961582e+00 5.09368479e-01 1.09491539e+00
8.77811313e-01 1.54798031e-01 7.44149506e-01 1.08410490e+00
9.65304494e-01 -8.09918642e-01 -3.51195753e-01 8.75511169e-01
-6.08700216e-02 -1.58524835e+00 -1.01251245e-01 -2.91724384e-01
-6.48572892e-02 1.57380104e+00 -2.32327983e-01 -5.39719388e-02
7.10466444e-01 -1.65649697e-01 -2.84473240e-01 -2.16241479e-01
-4.44277167e-01 -6.23178303e-01 3.37838322e-01 5.98709397e-02
4.46174234e-01 4.93538767e-01 -5.13172448e-01 -8.35852101e-02
-4.01980639e-01 -8.47615078e-02 9.36811149e-01 9.47376311e-01
-6.25540733e-01 -1.64373374e+00 -5.14415503e-01 3.65267247e-01
-2.62615293e-01 -3.63290273e-02 -1.66920975e-01 2.84770370e-01
2.34182656e-01 1.31634927e+00 -9.97013822e-02 -6.53865099e-01
7.05183223e-02 2.42916077e-01 1.12257099e+00 -9.62767959e-01
-7.64181241e-02 -5.51075220e-01 -2.26546396e-02 -5.34033060e-01
-7.12315798e-01 -5.48956573e-01 -1.31720042e+00 -7.97126830e-01
-7.90610015e-01 7.02744782e-01 9.60149825e-01 1.48606718e+00
-1.51272163e-01 3.87537330e-02 8.77899051e-01 -6.62579000e-01
-7.74400175e-01 -7.08246291e-01 -1.04878330e+00 5.09321690e-01
3.27973247e-01 -5.34676969e-01 -6.88783050e-01 -2.62506187e-01] | [5.63594388961792, 4.894177436828613] |
d59d0e14-5d4c-4c93-b035-f83a7175ddb4 | exploring-methods-for-generating-feedback | null | null | https://aclanthology.org/2021.emnlp-main.766 | https://aclanthology.org/2021.emnlp-main.766.pdf | Exploring Methods for Generating Feedback Comments for Writing Learning | The task of generating explanatory notes for language learners is known as feedback comment generation. Although various generation techniques are available, little is known about which methods are appropriate for this task. Nagata (2019) demonstrates the effectiveness of neural-retrieval-based methods in generating feedback comments for preposition use. Retrieval-based methods have limitations in that they can only output feedback comments existing in a given training data. Furthermore, feedback comments can be made on other grammatical and writing items than preposition use, which is still unaddressed. To shed light on these points, we investigate a wider range of methods for generating many feedback comments in this study. Our close analysis of the type of task leads us to investigate three different architectures for comment generation: (i) a neural-retrieval-based method as a baseline, (ii) a pointer-generator-based generation method as a neural seq2seq method, (iii) a retrieve-and-edit method, a hybrid of (i) and (ii). Intuitively, the pointer-generator should outperform neural-retrieval, and retrieve-and-edit should perform best. However, in our experiments, this expectation is completely overturned. We closely analyze the results to reveal the major causes of these counter-intuitive results and report on our findings from the experiments. | ['Kentaro Inui', 'Ryo Nagata', 'Kazuaki Hanawa'] | null | null | null | null | emnlp-2021-11 | ['comment-generation'] | ['natural-language-processing'] | [ 2.10779727e-01 2.07246825e-01 -7.54460469e-02 -1.97458401e-01
-9.50868130e-01 -6.86180592e-01 8.14779341e-01 3.05576593e-01
-3.68076682e-01 9.49295640e-01 6.32726133e-01 -8.06891561e-01
2.31740206e-01 -7.75976300e-01 -5.42374849e-01 -4.05309200e-01
2.43655443e-01 1.84884638e-01 1.59556314e-01 -4.61692512e-01
7.47362554e-01 -2.14141961e-02 -1.50174546e+00 5.21642029e-01
1.12951779e+00 5.48859417e-01 3.74822110e-01 8.65204036e-01
-3.89080375e-01 7.45398939e-01 -1.00773776e+00 -7.63490856e-01
5.54877147e-02 -9.37095106e-01 -8.11477542e-01 -3.30590278e-01
3.95405412e-01 -4.62768197e-01 9.69706625e-02 7.74694085e-01
6.44214272e-01 1.31706253e-01 1.06056750e+00 -9.71718788e-01
-1.17241442e+00 9.50627744e-01 -1.80925786e-01 1.96409717e-01
6.22417152e-01 2.64711350e-01 1.01848972e+00 -1.08149266e+00
6.81702673e-01 1.14136505e+00 5.30786276e-01 8.88744175e-01
-9.58219886e-01 -4.45416331e-01 1.85102299e-01 -1.17394373e-01
-1.01132643e+00 -3.74360383e-01 5.32396972e-01 -3.99519086e-01
9.56977785e-01 3.86089236e-01 6.46176934e-01 1.26744413e+00
3.65357287e-02 9.87548828e-01 1.24086177e+00 -6.38380706e-01
1.37066260e-01 5.58743596e-01 5.62517568e-02 3.34806055e-01
3.38233322e-01 3.10956359e-01 -6.87206268e-01 -1.49227470e-01
5.87161183e-01 -2.75199622e-01 -5.46303391e-01 2.36777395e-01
-1.02178884e+00 1.00516069e+00 4.01846647e-01 2.03214198e-01
-2.26818860e-01 8.41742158e-02 3.45354766e-01 5.59887707e-01
5.62219977e-01 9.48683918e-01 -4.27817285e-01 -3.24310064e-01
-1.02736461e+00 6.93185866e-01 1.09749746e+00 9.94550109e-01
5.50484955e-01 2.29663715e-01 -6.49977863e-01 8.46111119e-01
3.50184560e-01 3.34441066e-01 8.28071773e-01 -6.17770851e-01
7.44634628e-01 3.31890702e-01 3.28434467e-01 -7.31205642e-01
-1.93327248e-01 -2.35686913e-01 -4.91936475e-01 7.73573518e-02
6.46370351e-01 -5.16477764e-01 -6.37401223e-01 1.84685397e+00
-2.44693324e-01 -3.58262092e-01 1.86947867e-01 8.08866501e-01
9.86892700e-01 7.35237062e-01 1.57071967e-02 -2.66578496e-01
1.12589443e+00 -1.04402924e+00 -6.05792642e-01 9.80532728e-03
8.26061845e-01 -8.96485448e-01 1.41225731e+00 1.36888802e-01
-1.27681649e+00 -6.20647252e-01 -9.45984304e-01 -6.56489953e-02
-3.66378456e-01 1.31998092e-01 5.59201300e-01 7.37236559e-01
-1.24706495e+00 5.79041600e-01 -4.71674114e-01 -3.07340771e-01
-8.82946998e-02 -5.57745621e-02 9.61964577e-02 2.43215919e-01
-1.28115594e+00 9.21578765e-01 1.99941829e-01 -1.03177868e-01
-5.36607444e-01 -7.86744177e-01 -6.90909505e-01 1.13518789e-01
1.16530225e-01 -9.21222091e-01 1.78888416e+00 -1.18062389e+00
-1.54520988e+00 6.33048534e-01 -3.42995673e-01 -3.51146311e-01
5.18786609e-01 -2.28433385e-01 1.98000669e-02 -2.65148357e-02
-3.72219644e-02 7.59509683e-01 7.10750580e-01 -1.35269320e+00
-5.04341781e-01 -9.08636078e-02 2.57546335e-01 4.27738339e-01
-3.03540409e-01 1.45128235e-01 8.98505598e-02 -9.33642149e-01
-3.19434255e-01 -9.22905743e-01 1.17620379e-02 -2.06322014e-01
-5.36620378e-01 -5.83552122e-01 3.29311073e-01 -4.00530756e-01
1.59440243e+00 -1.88187099e+00 -1.76233366e-01 -2.29521662e-01
-7.14528337e-02 3.35548222e-01 -2.85009414e-01 1.00817299e+00
-1.40199497e-01 7.32132673e-01 -7.92368427e-02 -3.95440698e-01
9.13036019e-02 -2.76980162e-01 -6.30517602e-01 -4.34343427e-01
3.06419492e-01 1.10216069e+00 -1.03967845e+00 -2.84418255e-01
-4.01085407e-01 2.76924580e-01 -7.73479402e-01 3.71838987e-01
-4.26616281e-01 5.09718880e-02 -4.01325881e-01 4.44172859e-01
2.10873991e-01 -1.26744345e-01 -1.16457701e-01 2.02160984e-01
-3.88383746e-01 9.12624061e-01 -8.18963647e-01 1.42141080e+00
-3.78428906e-01 6.78049088e-01 -2.34935865e-01 -4.71918434e-01
9.09304261e-01 5.41393518e-01 -2.35436231e-01 -6.03307307e-01
-1.25930592e-01 5.17955363e-01 3.15869093e-01 -6.33112192e-01
1.10406971e+00 -2.96284825e-01 1.18711837e-01 1.08028352e+00
-1.32611021e-01 -3.46723109e-01 4.48242635e-01 3.81090850e-01
1.01315963e+00 2.83725142e-01 2.09033281e-01 -2.42449135e-01
4.51976433e-02 -9.85652953e-02 1.58865988e-01 1.06733811e+00
7.59286061e-02 8.90880585e-01 4.92640913e-01 -2.78532151e-02
-8.16073418e-01 -9.16787267e-01 2.42483214e-01 1.21409118e+00
-2.74489313e-01 -6.16692007e-01 -6.93072319e-01 -7.38079190e-01
-9.14984122e-02 1.06647873e+00 -5.46368301e-01 -1.67329386e-01
-5.27922571e-01 -4.36248034e-01 4.60866362e-01 7.19466329e-01
3.36064212e-02 -1.47357297e+00 -8.68557692e-01 4.98831943e-02
-4.30052042e-01 -2.69185096e-01 -6.43059611e-01 2.39010021e-01
-9.18185949e-01 -7.39423931e-01 -9.80357468e-01 -6.24483824e-01
6.35452867e-01 3.44981015e-01 1.29210925e+00 5.14031470e-01
3.78992110e-01 2.65710711e-01 -7.56849587e-01 -9.59016979e-01
-7.19364166e-01 3.22793990e-01 -3.39128733e-01 -5.13413489e-01
2.88886279e-01 -3.22005928e-01 -5.34579098e-01 -6.57971799e-02
-9.39360082e-01 9.07026082e-02 7.54638255e-01 8.91979635e-01
-3.91757414e-02 -4.96625900e-01 9.17140722e-01 -1.16808522e+00
1.49372625e+00 -4.98690337e-01 -1.24229632e-01 1.97576493e-01
-7.86894977e-01 2.55263895e-01 6.44997835e-01 -5.31875789e-01
-1.01137686e+00 -3.68768841e-01 -3.46830934e-01 -8.11979696e-02
4.40078266e-02 8.92710924e-01 1.83391422e-01 4.33699071e-01
8.48885715e-01 2.94811785e-01 -1.18785612e-01 -2.40662307e-01
3.17760706e-01 7.90300310e-01 2.57678896e-01 -7.80756652e-01
5.87049246e-01 -1.70996502e-01 -6.55501068e-01 -6.71730340e-01
-8.27144742e-01 -1.06703505e-01 -2.43618384e-01 -2.23229021e-01
6.32477999e-01 -6.70352876e-01 -2.66439766e-01 3.27381343e-01
-1.38426399e+00 -5.71755230e-01 -4.43216473e-01 2.79359162e-01
-6.06589317e-01 2.38144606e-01 -7.78074205e-01 -1.03556538e+00
-4.69134867e-01 -1.07448184e+00 8.52285087e-01 4.37196285e-01
-7.48829782e-01 -1.04201126e+00 3.72364372e-02 -4.64579090e-02
7.11792469e-01 -1.08005494e-01 1.11403656e+00 -8.75609577e-01
-5.01665056e-01 1.20825820e-01 -7.86511824e-02 2.41906390e-01
-1.87415835e-02 2.28949383e-01 -8.87462676e-01 -7.16948584e-02
-5.46623208e-02 -6.17651463e-01 8.72151792e-01 1.68581665e-01
1.05181551e+00 -7.37459958e-01 5.90551123e-02 1.58615619e-01
1.22749293e+00 2.83349574e-01 7.03143954e-01 1.75097048e-01
2.16268465e-01 7.24863291e-01 5.04847944e-01 3.51905584e-01
6.19316876e-01 5.97384930e-01 2.98621029e-01 1.52032644e-01
-1.09654844e-01 -6.65960431e-01 5.97100437e-01 1.06453228e+00
-1.14456467e-01 -7.29516745e-01 -6.12517178e-01 6.39352858e-01
-1.69331074e+00 -1.08760560e+00 -1.84230521e-01 2.32908821e+00
1.27136242e+00 1.87096640e-01 3.12291026e-01 1.89579293e-01
3.99799168e-01 1.09914519e-01 -2.01612055e-01 -8.18577945e-01
-3.34242657e-02 2.94439465e-01 1.14152755e-03 3.18991482e-01
-5.52257001e-01 6.63605988e-01 6.76991844e+00 4.15351123e-01
-1.20421410e+00 -1.20983154e-01 4.72402900e-01 1.23293884e-01
-8.79531682e-01 -3.34002003e-02 -8.52746964e-01 6.16910994e-01
1.00766683e+00 -4.01196212e-01 2.54479587e-01 5.62945008e-01
2.10836619e-01 -7.62728229e-02 -1.37962127e+00 5.57927430e-01
3.15434039e-01 -1.21894097e+00 4.07712996e-01 -2.00774804e-01
8.14231038e-01 -2.93460459e-01 6.44352362e-02 5.87783515e-01
3.25056642e-01 -9.37742293e-01 1.01424205e+00 4.92615789e-01
5.87788701e-01 -4.95987743e-01 6.01493597e-01 5.53874910e-01
-6.46939874e-01 -8.48107561e-02 -3.39498192e-01 -4.18605328e-01
3.21068704e-01 3.90946656e-01 -7.82566071e-01 4.97758001e-01
3.99053305e-01 2.77194530e-01 -5.79573452e-01 1.05983818e+00
-6.18986785e-01 9.60500240e-01 9.22934618e-03 -5.83817303e-01
2.80112416e-01 3.07981167e-02 4.57773775e-01 1.36444116e+00
5.85880578e-01 1.07076757e-01 -3.70548591e-02 1.06328464e+00
-3.20014022e-02 1.75096303e-01 -6.10610843e-01 -3.98239523e-01
7.05737650e-01 1.12067461e+00 -6.16173387e-01 -3.36870611e-01
-4.91444767e-01 7.95614421e-01 3.76467258e-01 4.47310925e-01
-4.55781549e-01 -5.95556676e-01 1.77626714e-01 3.71145338e-01
1.50038943e-01 -1.63297236e-01 -3.24394763e-01 -9.57469046e-01
9.48729925e-03 -1.04382741e+00 1.40142247e-01 -1.11956084e+00
-1.26198745e+00 4.75129187e-01 9.58257169e-02 -1.13415897e+00
-7.22861469e-01 -5.27950168e-01 -9.63906169e-01 1.05854440e+00
-1.40949500e+00 -6.50113106e-01 -1.45803958e-01 1.49182186e-01
8.73838186e-01 5.52036613e-02 9.14274514e-01 7.94386789e-02
-2.43096262e-01 6.93633556e-01 -2.18616351e-01 9.62712318e-02
9.43951607e-01 -1.53448987e+00 2.95903653e-01 7.12687731e-01
9.04428586e-02 1.04396617e+00 6.67503297e-01 -6.04290664e-01
-1.07350719e+00 -8.62017810e-01 1.29150236e+00 -5.34478426e-01
4.48261470e-01 -9.03903395e-02 -1.01630974e+00 5.58926463e-01
6.04006946e-01 -7.93561459e-01 7.31885493e-01 1.40499175e-01
-2.37044737e-01 3.07885259e-01 -6.02992237e-01 9.21836317e-01
8.11074257e-01 -4.14570987e-01 -8.00652266e-01 1.85705543e-01
9.45623457e-01 -5.07064819e-01 -4.74456251e-01 1.66282937e-01
6.14566326e-01 -1.20859683e+00 5.74081302e-01 -6.48763537e-01
1.17146564e+00 -5.79750873e-02 1.22067921e-01 -1.81185150e+00
-2.21359104e-01 -6.33070171e-01 2.62614004e-02 1.46493506e+00
9.49555337e-01 -6.95115447e-01 5.10331571e-01 5.95863760e-01
-5.40423751e-01 -1.11095357e+00 -3.13090473e-01 -6.34784698e-01
4.84206975e-01 -2.12158084e-01 6.29795969e-01 7.85491705e-01
2.65774012e-01 7.08325326e-01 -1.47916049e-01 -5.12098730e-01
-1.65988609e-01 8.45680237e-02 8.03693175e-01 -1.00327158e+00
-4.00667071e-01 -7.35919714e-01 3.43619198e-01 -1.33578789e+00
5.29810749e-02 -9.80676651e-01 1.95224315e-01 -1.69113922e+00
1.95974931e-01 -5.38684130e-01 5.08034192e-02 3.71592194e-01
-6.86282814e-01 -2.76657436e-02 4.93772358e-01 2.36947626e-01
-1.40008792e-01 4.33244854e-01 1.27941096e+00 3.10255382e-02
-3.06828558e-01 3.37490678e-01 -1.31410980e+00 4.26555246e-01
8.10640693e-01 -3.11598927e-01 -6.64155900e-01 -6.07733309e-01
7.02219367e-01 2.85423636e-01 1.37513682e-01 -5.85775614e-01
4.72323261e-02 -1.74260110e-01 1.38801858e-01 -5.81275225e-01
1.47520304e-01 -1.59740150e-01 -3.13772708e-01 4.00154531e-01
-8.47043514e-01 6.70610368e-01 4.55724373e-02 2.27208957e-01
-2.40064621e-01 -7.45143890e-01 3.70823294e-01 -3.61591369e-01
-2.53701925e-01 -1.09293818e-01 -6.45943344e-01 3.45330209e-01
4.16030258e-01 -3.40676367e-01 -6.12711489e-01 -8.52906883e-01
-2.25610524e-01 8.91224444e-02 3.73231113e-01 4.53145087e-01
5.59527755e-01 -1.19069552e+00 -1.02599466e+00 3.85113098e-02
1.78001806e-01 -5.00795916e-02 -3.27838689e-01 6.97487712e-01
-2.84080625e-01 4.91078019e-01 7.58189783e-02 -1.73063964e-01
-1.03880286e+00 3.82768303e-01 1.04968762e-02 -4.01301205e-01
-2.09865123e-01 1.00192070e+00 1.53552577e-01 -4.24843460e-01
1.96579084e-01 -4.48316872e-01 -3.27237576e-01 2.10471764e-01
7.01598883e-01 2.21674845e-01 7.58048594e-02 -2.28876323e-01
2.06544682e-01 3.64416949e-02 -5.83077483e-02 -4.20758069e-01
8.48782957e-01 -4.47940677e-02 1.45660847e-01 8.31942439e-01
7.88493216e-01 2.13777855e-01 -7.85193980e-01 5.42704575e-02
-9.96334385e-03 -1.95028603e-01 -4.41543937e-01 -1.08864689e+00
-5.86007714e-01 1.08117151e+00 -1.13752656e-01 6.36274517e-01
1.01004398e+00 -3.16116840e-01 5.94686091e-01 2.87458211e-01
9.57359374e-02 -9.22838926e-01 2.59793341e-01 7.89341390e-01
1.19853079e+00 -9.85538781e-01 -2.08491549e-01 -2.08974138e-01
-8.61879051e-01 1.25358772e+00 8.53290856e-01 7.18883872e-02
2.97341317e-01 1.10035799e-01 2.67781258e-01 -3.01294867e-02
-1.31021667e+00 -9.92779955e-02 1.97674140e-01 6.12885654e-01
1.21439171e+00 -1.20187528e-01 -7.70330369e-01 4.98642325e-01
-8.57780576e-01 -9.74331051e-02 8.16559672e-01 9.12610710e-01
-3.07667136e-01 -1.14093709e+00 -2.53700018e-01 6.78895950e-01
-5.11872470e-01 -4.84257460e-01 -7.68582642e-01 8.42015803e-01
-3.17667156e-01 1.12383044e+00 -5.17163984e-02 -1.92184955e-01
2.60241717e-01 3.86221945e-01 3.71359408e-01 -9.87170339e-01
-1.09003282e+00 -1.67590395e-01 1.88394353e-01 -5.77827096e-02
-2.61970907e-01 -7.05043256e-01 -1.00244033e+00 -1.38862044e-01
-6.12060428e-01 4.68234986e-01 6.64488196e-01 8.14351201e-01
2.76595086e-01 4.70401973e-01 4.58129078e-01 -8.06705177e-01
-1.05709565e+00 -1.36440444e+00 -2.22442433e-01 3.66500497e-01
1.66555926e-01 -2.53816485e-01 -3.84751320e-01 9.28824693e-02] | [11.753331184387207, 8.969487190246582] |
4e43703b-173a-444c-9e85-bb3a51eddb88 | tfr-texture-defect-detection-with-fourier | 2307.04574 | null | https://arxiv.org/abs/2307.04574v1 | https://arxiv.org/pdf/2307.04574v1.pdf | TFR: Texture Defect Detection with Fourier Transform using Normal Reconstructed Template of Simple Autoencoder | Texture is an essential information in image representation, capturing patterns and structures. As a result, texture plays a crucial role in the manufacturing industry and is extensively studied in the fields of computer vision and pattern recognition. However, real-world textures are susceptible to defects, which can degrade image quality and cause various issues. Therefore, there is a need for accurate and effective methods to detect texture defects. In this study, a simple autoencoder and Fourier transform are employed for texture defect detection. The proposed method combines Fourier transform analysis with the reconstructed template obtained from the simple autoencoder. Fourier transform is a powerful tool for analyzing the frequency domain of images and signals. Moreover, since texture defects often exhibit characteristic changes in specific frequency ranges, analyzing the frequency domain enables effective defect detection. The proposed method demonstrates effectiveness and accuracy in detecting texture defects. Experimental results are presented to evaluate its performance and compare it with existing approaches. | ['Sungyoung Kim', 'Jongwook Si'] | 2023-07-10 | null | null | null | null | ['defect-detection'] | ['computer-vision'] | [ 2.73071915e-01 -6.85796976e-01 1.22713335e-01 4.36662957e-02
-1.88044123e-02 1.29916862e-01 1.23291738e-01 7.71390349e-02
3.79868001e-02 2.86861539e-01 -1.02450751e-01 1.91679999e-01
-4.67445552e-01 -1.11402631e+00 -7.01491609e-02 -1.12992549e+00
1.45485550e-01 -6.55606464e-02 2.50458300e-01 -2.22914025e-01
4.42098409e-01 4.89460558e-01 -1.99671376e+00 2.74708420e-01
7.61171758e-01 1.40166354e+00 6.02081776e-01 1.08805440e-01
-2.57385492e-01 6.29107952e-01 -7.62835681e-01 1.44577965e-01
-1.39995515e-01 -5.38863957e-01 -3.96204740e-01 6.01105154e-01
-2.85635203e-01 -2.46390760e-01 -3.00739944e-01 1.21417570e+00
3.58150810e-01 2.50717819e-01 6.93312526e-01 -5.53664267e-01
-6.79356098e-01 6.16610348e-02 -4.22903240e-01 3.00228357e-01
1.60286635e-01 -2.09703118e-01 4.21349406e-01 -7.26306736e-01
2.00990632e-01 1.02768660e+00 5.42995453e-01 1.57346874e-01
-7.80149281e-01 -2.00830102e-01 -5.23072183e-01 6.91675246e-01
-9.73360419e-01 -1.53002068e-01 1.31060994e+00 -3.12710494e-01
6.71037793e-01 2.25213140e-01 6.86519682e-01 6.28901362e-01
9.48748052e-01 5.29006183e-01 1.42989910e+00 -6.89626694e-01
1.53983838e-03 -3.65996100e-02 -1.02298342e-01 6.26516938e-01
1.58462510e-01 2.57284343e-01 -9.29326043e-02 1.87026441e-01
1.05928755e+00 4.94228810e-01 -5.84363043e-01 6.34578913e-02
-1.08107245e+00 6.40978694e-01 2.40168199e-01 1.01923466e+00
-6.12512410e-01 -2.86063254e-01 3.89032394e-01 6.26192153e-01
4.23153996e-01 2.18386233e-01 2.59662028e-02 -2.60932207e-01
-5.15707493e-01 -3.03340167e-01 4.96067911e-01 1.48698941e-01
4.03448105e-01 4.12274987e-01 3.57309580e-01 1.30912089e+00
2.02529624e-01 7.42858410e-01 8.17953885e-01 -6.58315599e-01
-1.73096865e-01 4.40501839e-01 -2.40234151e-01 -1.62290394e+00
-1.31285578e-01 -1.62218556e-01 -1.08833778e+00 2.28189930e-01
1.68360069e-01 5.35615206e-01 -6.62340820e-01 6.65430725e-01
3.82841676e-01 -1.48806244e-01 -7.68765360e-02 9.58435297e-01
7.51675308e-01 7.76322067e-01 -3.69119227e-01 -2.56551117e-01
1.27769184e+00 -4.63413745e-01 -1.15302587e+00 1.61674276e-01
-9.42307711e-03 -1.27188301e+00 9.84305859e-01 7.19859719e-01
-6.03198826e-01 -8.02558541e-01 -9.85985816e-01 3.13917994e-01
-1.70158166e-02 5.99543393e-01 6.22651875e-01 4.10824776e-01
-5.12750983e-01 7.75080740e-01 -9.36648428e-01 -3.83202970e-01
5.65707162e-02 1.42796293e-01 -5.14199972e-01 -2.88031548e-01
-8.73406708e-01 8.40300858e-01 2.49355778e-01 3.34104091e-01
-1.72929779e-01 -9.35957357e-02 -7.48641789e-01 -2.51465291e-02
5.62658943e-02 -4.82549369e-02 9.79359150e-01 -8.90688717e-01
-1.60834372e+00 4.74954695e-01 -1.19955661e-02 -6.28919154e-02
5.51746413e-02 6.45291582e-02 -9.93851840e-01 4.90887552e-01
1.23562990e-02 -5.05070567e-01 1.30385745e+00 -9.18041885e-01
-6.00565612e-01 -3.18617404e-01 -3.82335573e-01 -2.58259326e-01
-4.04111117e-01 -2.72343308e-01 -2.65721660e-02 -7.96137631e-01
7.80240953e-01 -4.56000119e-01 3.34142536e-01 -1.88870713e-01
-6.78401217e-02 -2.45986477e-01 1.32528710e+00 -1.01353073e+00
1.11262381e+00 -2.51006198e+00 -3.05514455e-01 4.94211376e-01
2.92881243e-02 6.21930137e-02 1.03429109e-01 6.39800251e-01
-3.42302630e-03 -4.13700521e-01 -1.58773854e-01 3.97642136e-01
-3.13999563e-01 3.41062993e-01 1.38789192e-01 5.83431959e-01
2.91688204e-01 6.99344352e-02 -3.15814316e-01 -3.60750437e-01
6.45120859e-01 6.50415599e-01 -2.33205795e-01 1.27731636e-01
2.81256557e-01 4.95035887e-01 -8.62098634e-01 9.41908121e-01
6.16092265e-01 -6.11731634e-02 -1.56794325e-01 -6.56152368e-01
-2.13155165e-01 -2.40146652e-01 -1.02996731e+00 9.06390548e-01
-5.87868094e-01 6.87945306e-01 1.51367383e-02 -1.55049694e+00
1.43835187e+00 4.47665989e-01 6.30778790e-01 -1.14183354e+00
4.77049053e-01 5.22880077e-01 1.14959046e-01 -1.20648241e+00
7.92581290e-02 -3.03731352e-01 4.07953799e-01 1.73135370e-01
-9.13888887e-02 -3.04430485e-01 2.10928962e-01 -7.01111615e-01
9.07065451e-01 -1.49316266e-01 2.20818311e-01 -9.86938030e-02
7.55616307e-01 -2.38593686e-02 5.32056391e-01 1.63413331e-01
1.07486516e-01 3.75341177e-01 -4.10345085e-02 -8.91392171e-01
-1.16918421e+00 -6.20292306e-01 -4.43947822e-01 1.09379627e-01
3.55918646e-01 2.07556337e-01 -4.80726063e-01 -1.35893241e-01
1.73955634e-01 1.56576052e-01 -5.34390986e-01 -4.34161812e-01
-5.18656552e-01 -6.60559297e-01 -1.69302206e-02 2.54472077e-01
1.10890412e+00 -1.23357761e+00 -6.75070167e-01 5.85133731e-01
-3.78203154e-01 -7.49008656e-01 -2.77721360e-02 -2.71769613e-01
-1.02308428e+00 -1.40028751e+00 -6.59401953e-01 -1.11798930e+00
5.98260701e-01 5.80268204e-01 7.25328028e-01 4.43649322e-01
-7.22472012e-01 3.51967692e-01 -6.56469166e-01 -3.62925828e-01
-5.72445869e-01 -5.14259160e-01 -1.82738796e-01 3.07636976e-01
3.23932588e-01 -5.50252318e-01 -6.04522347e-01 5.52216291e-01
-9.31292474e-01 -1.93294808e-01 8.52781057e-01 1.22067130e+00
6.37131631e-01 1.32569826e+00 6.00699604e-01 -5.12664616e-01
7.38743663e-01 -7.16263652e-02 -6.38781607e-01 7.25403950e-02
-3.49518150e-01 -2.15635821e-01 7.24426925e-01 -2.66778857e-01
-1.31472921e+00 -5.33926606e-01 -2.92319834e-01 -3.07622224e-01
-3.04760337e-01 8.51560175e-01 -4.90753949e-02 -4.01838541e-01
3.88713092e-01 6.06024802e-01 7.38218367e-01 -6.98390126e-01
-7.78690279e-01 7.89261043e-01 4.92663383e-01 -3.08616251e-01
3.80413473e-01 4.86127287e-01 -5.44903018e-02 -1.30911541e+00
-1.71415240e-01 -5.26269734e-01 1.64573714e-02 -4.93180275e-01
6.91391945e-01 -4.27080810e-01 -8.80949795e-01 8.88354421e-01
-8.59249294e-01 2.16807142e-01 1.77793298e-02 9.05540884e-01
-4.33967769e-01 5.08881390e-01 -8.84918571e-01 -7.91390955e-01
-2.57148743e-01 -1.09611642e+00 7.90294230e-01 2.49985904e-01
2.08015442e-01 -1.04834437e+00 -3.19555819e-01 3.02569866e-01
7.19404578e-01 3.21883410e-01 1.16795933e+00 1.37515590e-01
-2.47574598e-01 -5.31002223e-01 -1.20426975e-02 6.35800421e-01
7.85808206e-01 1.03359468e-01 -6.97620451e-01 -1.33934826e-01
9.01006937e-01 7.20900670e-02 7.45943844e-01 4.90699381e-01
1.27563655e+00 -1.48338139e-01 -5.85531257e-02 1.62718549e-01
1.75365055e+00 7.93391943e-01 8.20999205e-01 4.66551989e-01
2.91693062e-01 5.73035419e-01 7.62891173e-01 4.35461253e-01
-2.36853033e-01 3.56205910e-01 1.84921041e-01 -1.96570188e-01
-1.28934771e-01 1.77453235e-01 2.35259277e-03 1.33218420e+00
-5.64786017e-01 -3.39335315e-02 -6.27145171e-01 3.97938848e-01
-1.41154015e+00 -1.19087350e+00 -1.11686736e-01 1.90913475e+00
4.95235771e-01 -6.55221343e-02 -3.27763349e-01 1.24917018e+00
8.62768114e-01 -2.55163133e-01 -1.88093141e-01 -2.36190692e-01
-2.07610726e-01 5.29833972e-01 1.56518072e-02 -2.41492074e-02
-1.05545020e+00 4.33570534e-01 5.71208525e+00 1.05187917e+00
-1.67653966e+00 -1.29009575e-01 1.78466335e-01 7.03519046e-01
-1.66161843e-02 -4.95755911e-01 1.83888227e-01 6.44631982e-01
1.88753650e-01 9.90044326e-02 3.96165013e-01 6.10721231e-01
4.00446713e-01 -6.07480407e-01 -4.00712311e-01 1.17442000e+00
-1.42578453e-01 -1.02709842e+00 -1.26633242e-01 -5.29699326e-02
4.25316513e-01 -6.68756783e-01 1.77646011e-01 -1.14868619e-01
-4.07325238e-01 -7.70512760e-01 2.52050638e-01 5.58836997e-01
6.32784545e-01 -8.89893889e-01 1.25356877e+00 9.20870602e-02
-1.07187641e+00 -1.76493853e-01 -7.27675378e-01 -1.64425477e-01
2.69472487e-02 1.27839732e+00 -5.18429935e-01 7.09027350e-01
9.18143094e-01 8.72491419e-01 9.44662839e-02 1.18643367e+00
-4.88394015e-02 7.02227950e-01 -2.22169727e-01 -1.87182948e-02
-8.41985419e-02 -5.33093810e-01 3.70379061e-01 5.97007513e-01
8.32800388e-01 1.88867182e-01 7.40860477e-02 5.54151297e-01
5.77834010e-01 2.13762119e-01 -5.88670969e-01 -2.99040675e-01
2.28481382e-01 9.03449178e-01 -9.29818213e-01 -7.31126145e-02
-4.39116389e-01 7.32024789e-01 -6.12756670e-01 2.74794698e-01
-4.70758647e-01 -6.65845931e-01 3.32183927e-01 1.19920680e-02
2.61696786e-01 -1.48695827e-01 4.82660383e-02 -8.87757659e-01
9.00837258e-02 -1.07174098e+00 6.79122955e-02 -5.02615273e-01
-1.35296428e+00 4.60318595e-01 -4.70675141e-01 -1.58936214e+00
1.15629882e-01 -7.40718186e-01 -5.60629189e-01 6.03145897e-01
-1.17373347e+00 -9.07774389e-01 -4.52161252e-01 7.18387663e-01
6.87863410e-01 -2.31204361e-01 9.25520718e-01 4.50719416e-01
-4.35977161e-01 5.76518923e-02 4.56218958e-01 -2.97188498e-02
3.02716434e-01 -8.17053676e-01 -3.27921331e-01 6.11090362e-01
-1.59173235e-01 1.60093144e-01 7.00020671e-01 -6.58935666e-01
-1.56967700e+00 -5.37872493e-01 5.27246237e-01 6.19487524e-01
3.16900134e-01 4.15218860e-01 -9.70076025e-01 -9.49068516e-02
-7.71725401e-02 -7.55695999e-03 4.67161506e-01 -4.00260836e-01
2.85754293e-01 -3.00272435e-01 -1.34096563e+00 1.17329538e-01
3.95482510e-01 -5.78622043e-01 -6.09740913e-01 4.39404808e-02
-2.28590388e-02 -2.80113518e-01 -1.16988111e+00 6.01221144e-01
7.41084576e-01 -1.24331284e+00 6.95467591e-01 3.92538160e-01
5.94154179e-01 -4.23071653e-01 -1.19514698e-02 -1.36057639e+00
-5.42936683e-01 2.02772260e-01 3.32615763e-01 9.72001135e-01
-1.39848903e-01 -8.73400629e-01 5.29631972e-01 -3.10816437e-01
-2.27263823e-01 -6.05526328e-01 -7.40287960e-01 -7.22553253e-01
-7.10762262e-01 -6.41042888e-02 4.92660761e-01 9.70879674e-01
1.05451522e-02 -2.38959849e-01 -2.31120273e-01 1.11581221e-01
6.62595987e-01 3.00958127e-01 2.60011822e-01 -1.39915860e+00
-2.72941798e-01 -3.83650482e-01 -8.83020878e-01 -3.89551491e-01
-2.51331925e-01 -3.38866949e-01 -5.51098324e-02 -1.56764591e+00
-2.38221750e-01 -4.39548314e-01 -3.00008535e-01 1.87808827e-01
1.67412430e-01 3.02288651e-01 -2.42435247e-01 4.48694617e-01
3.41855764e-01 7.18828440e-01 1.68824637e+00 -4.92999107e-01
1.71740308e-01 2.02352032e-01 -1.01890095e-01 5.88372171e-01
9.69121337e-01 -8.71261582e-02 -4.17149156e-01 -3.72902662e-01
-2.68474579e-01 1.56971723e-01 2.03851029e-01 -1.26558650e+00
7.59910122e-02 -9.54454243e-02 6.06003821e-01 -3.84511620e-01
1.43650696e-01 -1.20792949e+00 3.98740590e-01 6.68365121e-01
4.63855177e-01 -5.19149005e-02 1.95066988e-01 3.71369272e-01
-1.05643439e+00 -1.72141328e-01 7.35688806e-01 -1.79421961e-01
-1.06932843e+00 -7.15224966e-02 -6.61114216e-01 -8.30972731e-01
7.87779748e-01 -6.76862299e-01 -1.77413270e-01 -2.15967089e-01
-6.65733933e-01 -3.96679163e-01 4.15251523e-01 2.84041435e-01
1.10884488e+00 -1.35490406e+00 -3.38077575e-01 6.88993216e-01
-5.48887765e-03 -2.56770104e-01 7.71602988e-01 8.73722076e-01
-1.06990862e+00 -1.64230692e-03 -6.14287198e-01 -7.58825362e-01
-1.13287854e+00 2.57571459e-01 2.61046171e-01 9.26867351e-02
-8.16548944e-01 4.14022684e-01 -1.33786425e-02 1.74275175e-01
-2.23022982e-01 -2.68060267e-01 -6.12090886e-01 -3.29252481e-01
5.30495822e-01 5.18445313e-01 4.04880732e-01 -4.98337448e-01
4.93452027e-02 1.08203340e+00 2.51681030e-01 3.49278212e-01
1.36474073e+00 -9.56296995e-02 -4.96501267e-01 4.25436318e-01
1.05051184e+00 5.07614650e-02 -7.52981663e-01 -5.49114384e-02
-4.68154661e-02 -5.60770452e-01 4.11061734e-01 -4.30448145e-01
-1.32409573e+00 9.20175970e-01 8.79823625e-01 5.47826171e-01
1.55689359e+00 -3.27517956e-01 9.02308583e-01 2.91354060e-01
7.56295741e-01 -1.35416543e+00 2.21192032e-01 1.26546651e-01
9.40335453e-01 -1.09827042e+00 -1.51254609e-01 -7.96841323e-01
-1.03541344e-01 1.54205918e+00 3.99352372e-01 -2.99214602e-01
7.20782757e-01 1.73807159e-01 3.03982615e-01 -3.03411245e-01
-1.53427541e-01 3.05257048e-02 1.46504343e-01 5.16648769e-01
4.38080221e-01 -5.06169498e-02 -6.16943419e-01 2.29938596e-01
-1.04815304e-01 4.15448993e-02 1.44189954e-01 1.16364956e+00
-7.71603107e-01 -1.17760086e+00 -9.47095871e-01 6.76751494e-01
-7.86569536e-01 4.66220140e-01 2.88812101e-01 7.70581186e-01
5.66487610e-02 1.15155756e+00 1.14811212e-01 -5.49781799e-01
3.28805208e-01 -1.74526796e-01 6.97226465e-01 -4.14145887e-02
-3.95297147e-02 3.91765624e-01 -1.29253313e-01 -3.49014103e-01
-5.99639595e-01 -3.90037715e-01 -1.08419514e+00 -2.87865758e-01
-5.93516409e-01 2.97832251e-01 8.64981055e-01 9.10047531e-01
-6.35153055e-02 6.99789286e-01 9.62018073e-01 -5.52835763e-01
-2.56736428e-01 -1.04296482e+00 -1.16255569e+00 5.41984141e-01
2.94346422e-01 -1.10104525e+00 -3.77224952e-01 1.48149505e-01] | [7.486916542053223, 1.6883924007415771] |
59eb87d2-3ed2-4d29-9a93-9a011f547323 | cost-aware-asynchronous-multi-agent-active | 2210.02259 | null | https://arxiv.org/abs/2210.02259v1 | https://arxiv.org/pdf/2210.02259v1.pdf | Cost Aware Asynchronous Multi-Agent Active Search | Multi-agent active search requires autonomous agents to choose sensing actions that efficiently locate targets. In a realistic setting, agents also must consider the costs that their decisions incur. Previously proposed active search algorithms simplify the problem by ignoring uncertainty in the agent's environment, using myopic decision making, and/or overlooking costs. In this paper, we introduce an online active search algorithm to detect targets in an unknown environment by making adaptive cost-aware decisions regarding the agent's actions. Our algorithm combines principles from Thompson Sampling (for search space exploration and decentralized multi-agent decision making), Monte Carlo Tree Search (for long horizon planning) and pareto-optimal confidence bounds (for multi-objective optimization in an unknown environment) to propose an online lookahead planner that removes all the simplifications. We analyze the algorithm's performance in simulation to show its efficacy in cost aware active search. | ['Jeff Schneider', 'Ramina Ghods', 'Arundhati Banerjee'] | 2022-10-05 | null | null | null | null | ['thompson-sampling'] | ['methodology'] | [ 3.70737553e-01 6.11982644e-01 -7.15307832e-01 -1.01219870e-01
-1.17816842e+00 -8.72944236e-01 5.41854143e-01 2.43962333e-01
-8.08744133e-01 1.30169308e+00 1.01882860e-01 -3.42983037e-01
-6.42988503e-01 -8.15480828e-01 -2.96528220e-01 -8.30081522e-01
-3.67221266e-01 1.09233582e+00 2.86876857e-01 4.16929722e-02
4.50174004e-01 4.97109324e-01 -1.18401515e+00 -4.53689396e-01
1.05800867e+00 8.89812768e-01 5.39259851e-01 8.42362285e-01
2.28143707e-01 1.01777041e+00 -7.87168205e-01 1.32352009e-01
6.11496627e-01 -2.05360666e-01 -7.72466898e-01 1.57067075e-01
-6.27570748e-01 -7.38163114e-01 4.93245035e-01 1.14810705e+00
4.13056791e-01 4.70904678e-01 3.31515402e-01 -1.22375536e+00
-3.87590937e-02 9.46379006e-01 -5.95638216e-01 3.14082056e-01
2.72822499e-01 6.27256155e-01 8.95511210e-01 -9.18890908e-02
6.66781902e-01 1.41731811e+00 7.84276798e-02 3.50792497e-01
-1.14107299e+00 -2.61134267e-01 5.70341825e-01 3.46375942e-01
-1.01142144e+00 -5.68989754e-01 5.59857607e-01 1.74828190e-02
1.16759396e+00 5.10358870e-01 9.49662507e-01 7.17260778e-01
4.40407783e-01 8.48707139e-01 1.16365218e+00 -7.08282888e-01
1.07571805e+00 -2.34715551e-01 -3.75630856e-01 3.76590163e-01
7.44133294e-01 7.29748428e-01 -2.17225358e-01 -7.40001500e-01
4.97865915e-01 -1.80887878e-01 1.17568180e-01 -6.28519356e-01
-1.23090088e+00 9.58047390e-01 2.62410063e-02 -2.60753989e-01
-1.04423618e+00 3.69484574e-01 -6.18534721e-02 4.11155075e-01
2.43595570e-01 6.90875113e-01 -3.06924254e-01 -2.26779655e-01
-4.62212086e-01 2.20151916e-01 8.26368392e-01 7.70444989e-01
6.42742753e-01 1.54017761e-01 -2.21655056e-01 3.05466682e-01
5.85161269e-01 5.69738269e-01 2.68424720e-01 -1.62040186e+00
6.30029440e-01 3.42053622e-01 1.15597892e+00 -5.10222375e-01
-3.77500921e-01 -5.45486808e-01 5.90406545e-02 1.05327129e+00
9.44148749e-02 -8.28998327e-01 -7.03081548e-01 1.50420713e+00
6.18194818e-01 -4.67655301e-01 4.08201605e-01 6.85169518e-01
-4.86586750e-01 5.19726872e-01 -1.02804847e-01 -1.16768086e+00
6.46875143e-01 -1.08506477e+00 -6.54040635e-01 -4.96527046e-01
3.37748051e-01 -3.64174396e-01 3.79179209e-01 4.19123381e-01
-1.27256191e+00 1.95670530e-01 -1.07205141e+00 7.73848116e-01
1.17809689e-02 -4.45478827e-01 5.98641694e-01 5.32918334e-01
-1.17466426e+00 2.08154544e-01 -1.06863368e+00 -2.96080083e-01
1.40764698e-01 5.36847651e-01 5.64278185e-01 3.99279833e-01
-8.05687428e-01 1.06479311e+00 4.96477097e-01 4.53333696e-03
-1.62615323e+00 9.47943628e-02 -3.73356223e-01 2.31849715e-01
1.42585230e+00 -7.37954676e-01 1.81992245e+00 -1.12805307e+00
-1.86081958e+00 -4.27130163e-02 1.15369380e-01 -7.03326404e-01
7.07589090e-01 1.80201203e-01 1.47032514e-01 5.77727817e-02
3.46305728e-01 2.51885474e-01 4.84087050e-01 -1.20821714e+00
-1.09788156e+00 -4.37891096e-01 5.79276979e-01 1.10614049e+00
1.29157647e-01 -5.26230559e-02 2.41236791e-01 -1.37807071e-01
-2.12097079e-01 -1.06404650e+00 -1.22453773e+00 -4.80369955e-01
-4.23027903e-01 -1.08262666e-01 2.16756642e-01 2.48456776e-01
8.49308610e-01 -1.53923059e+00 -1.83613256e-01 3.56098533e-01
-5.13499863e-02 -3.43073308e-01 -2.68175304e-01 7.89496720e-01
6.31547928e-01 -1.29733250e-01 1.63889259e-01 -1.45128533e-01
1.95922211e-01 2.33656153e-01 -1.55013800e-01 3.49167794e-01
-5.88144064e-01 7.65158892e-01 -1.03530538e+00 -4.17020947e-01
-8.46239179e-02 -6.28023982e-01 -2.09508210e-01 -3.44749093e-02
-6.09496653e-01 1.19620584e-01 -1.25732744e+00 8.86441052e-01
2.35474229e-01 -2.40067571e-01 4.77046132e-01 8.53386223e-01
-8.02784324e-01 2.29861036e-01 -1.18775451e+00 1.14273059e+00
-2.97984898e-01 2.27737561e-01 7.58443892e-01 -7.16192305e-01
4.34644312e-01 8.11364576e-02 6.84769809e-01 -6.39426827e-01
4.17352989e-02 3.00211817e-01 1.06154777e-01 -1.05320007e-01
5.56974888e-01 3.64853501e-01 -2.13514566e-02 9.80235577e-01
-7.46319592e-01 9.57205966e-02 6.08126447e-02 1.77434925e-02
1.41424000e+00 -2.33231992e-01 8.44606936e-01 -3.93805474e-01
-8.01732298e-03 8.53212059e-01 7.56858468e-01 1.51211739e+00
-4.73076254e-01 -4.86696243e-01 1.25295699e-01 -3.30856442e-01
-4.00709778e-01 -7.29344726e-01 4.82378542e-01 1.20825446e+00
5.76626420e-01 1.93745598e-01 -4.60085571e-01 -5.07121861e-01
-4.54901904e-02 1.14892411e+00 -7.42721617e-01 3.51293266e-01
-4.61147964e-01 -9.14909065e-01 -3.25489968e-01 -3.88930775e-02
4.14730519e-01 -9.73110616e-01 -1.77479017e+00 6.47533119e-01
1.95534289e-01 -3.93209547e-01 -4.45265263e-01 5.17554104e-01
-6.93873525e-01 -1.02701664e+00 -4.46208239e-01 -4.48229797e-02
7.58449256e-01 4.32090491e-01 6.77253008e-01 -4.31778282e-01
-5.63523807e-02 8.29671264e-01 -3.43845248e-01 -8.50358248e-01
-2.69771159e-01 -1.02675818e-01 1.36945456e-01 -4.15718883e-01
-1.72078729e-01 -2.67348647e-01 -7.94991791e-01 2.61171967e-01
-3.19673061e-01 -7.52450451e-02 7.22628713e-01 6.52168572e-01
7.27373660e-01 4.58662242e-01 4.84637499e-01 -5.56099296e-01
1.01843989e+00 -3.67996544e-01 -1.47737312e+00 6.03117645e-01
-9.41845596e-01 8.36032350e-03 2.78992862e-01 -4.74633068e-01
-1.35389173e+00 2.11592048e-01 7.44009376e-01 5.01600727e-02
2.77592778e-01 3.50109458e-01 -7.79313445e-02 -2.94318467e-01
5.89174330e-01 1.51642933e-01 -1.05124727e-01 -1.96491703e-01
5.48260063e-02 3.05125058e-01 2.14410275e-02 -5.63121498e-01
6.26811683e-01 4.26640600e-01 9.73588973e-02 -2.70773858e-01
-4.70463276e-01 -7.93288201e-02 -1.12757504e-01 -2.62842268e-01
3.78358215e-01 -6.19074821e-01 -1.03403080e+00 1.04984809e-02
-9.35435891e-01 -6.23027325e-01 -7.78423965e-01 6.25212252e-01
-1.10353398e+00 1.99672788e-01 2.14046493e-01 -1.70234942e+00
-2.51265913e-01 -1.31895578e+00 5.15036583e-01 5.10104418e-01
-1.08548000e-01 -7.47543395e-01 5.24591625e-01 2.05469221e-01
5.48859596e-01 3.44362110e-01 2.90326178e-01 -8.07788014e-01
-1.45937681e+00 2.90962398e-01 5.57734907e-01 -7.23609805e-01
6.57248497e-02 -2.70743996e-01 -3.63123298e-01 -5.82535207e-01
1.32050768e-01 -2.37028539e-01 3.61111701e-01 8.66609991e-01
3.77033353e-01 -1.24565029e+00 -8.77223492e-01 9.70924422e-02
1.46369290e+00 1.28441107e+00 -4.63215709e-02 1.04871786e+00
-4.77569252e-01 5.29923081e-01 1.27311695e+00 1.05803728e+00
3.14604580e-01 7.53800690e-01 1.11386931e+00 3.58597428e-01
6.62597418e-01 3.67003381e-02 5.32791197e-01 -1.47835419e-01
-6.48564920e-02 -6.11304939e-01 -7.42496431e-01 7.27689207e-01
-2.10185599e+00 -1.14078045e+00 6.58075452e-01 2.45630693e+00
7.66658485e-01 3.72689277e-01 2.86370724e-01 -3.70892614e-01
6.96872234e-01 8.43296107e-03 -1.34193707e+00 -5.64575970e-01
2.74664965e-02 -4.96745348e-01 1.10738707e+00 9.95619297e-01
-7.42925644e-01 6.88273728e-01 6.86524105e+00 7.53968477e-01
-5.64260244e-01 3.03323358e-01 7.22049654e-01 -5.78163445e-01
-4.50971603e-01 3.60407740e-01 -7.51110077e-01 2.55772054e-01
7.03127027e-01 -6.68078184e-01 6.99555933e-01 1.10558546e+00
7.20537066e-01 -9.59647059e-01 -8.73646379e-01 4.39626426e-01
-6.18187606e-01 -1.39732099e+00 -5.94392598e-01 4.17783499e-01
9.78211820e-01 2.76100576e-01 -1.45055056e-01 -2.64781155e-02
1.58374012e+00 -6.48723245e-01 9.88153577e-01 4.76010591e-01
1.47878274e-01 -8.77002239e-01 4.56413746e-01 9.96429682e-01
-9.15359914e-01 -7.70678997e-01 -2.21402809e-01 -4.63996768e-01
3.75929922e-01 2.05253646e-01 -1.23017871e+00 1.43889293e-01
4.17830616e-01 -3.40440214e-01 -2.27683242e-02 1.30990553e+00
-7.91917965e-02 1.67983845e-01 -7.15281904e-01 -7.77550638e-01
6.58768654e-01 -2.84813195e-01 1.05983257e+00 4.33498114e-01
3.30548257e-01 4.41740423e-01 6.58996344e-01 7.86577284e-01
6.39991224e-01 -1.94072023e-01 -3.34336996e-01 1.03198644e-02
1.08991349e+00 7.58213162e-01 -1.00374460e+00 -2.80455023e-01
9.35912728e-02 5.38732529e-01 5.32833450e-02 5.18551171e-01
-4.97405440e-01 -1.95770770e-01 4.24237520e-01 -4.86971796e-01
1.51727587e-01 -2.01632276e-01 -2.82626927e-01 -5.38147688e-01
-1.45724639e-01 -5.77996790e-01 5.48005998e-01 -5.04094124e-01
-5.63544214e-01 3.79941463e-01 3.46701354e-01 -9.12849963e-01
-1.01061571e+00 8.50843415e-02 -7.94060886e-01 5.91309190e-01
-1.29730105e+00 -4.55188096e-01 2.97820151e-01 2.67479867e-01
1.08183420e+00 -3.10192436e-01 5.28860331e-01 -7.90549517e-01
-4.37293738e-01 4.82668430e-02 5.24310887e-01 -5.78222334e-01
-8.05501118e-02 -1.13266456e+00 1.15193449e-01 8.86250675e-01
-3.93307388e-01 3.54234755e-01 8.70450497e-01 -8.23486567e-01
-1.66945136e+00 -5.04292250e-01 2.34806791e-01 -2.89244298e-02
7.42448330e-01 2.30458796e-01 1.13635331e-01 6.54134274e-01
3.59551787e-01 -5.99726081e-01 3.27016234e-01 -1.22296363e-01
5.06364822e-01 -9.68127996e-02 -1.38028383e+00 7.97227561e-01
9.75287437e-01 3.49895298e-01 -3.40177566e-01 4.50720251e-01
6.67795181e-01 -2.23353103e-01 -2.16218784e-01 4.20631200e-01
4.45709437e-01 -9.14977610e-01 6.81858838e-01 -3.17929626e-01
-6.72865689e-01 -1.29548013e-01 -5.62414043e-02 -1.56672919e+00
-5.32117724e-01 -1.38142776e+00 5.04328124e-02 5.59314847e-01
6.95993423e-01 -9.84718561e-01 1.02292931e+00 7.83686221e-01
-2.60205623e-02 -6.78183913e-01 -1.43364561e+00 -7.96146989e-01
-3.10011655e-01 1.64149493e-01 6.13957644e-01 5.49320042e-01
8.82486850e-02 -3.16894725e-02 -1.41175151e-01 3.63093287e-01
1.12355494e+00 4.68760490e-01 5.09477973e-01 -8.85011733e-01
-5.68510473e-01 -4.32773113e-01 3.26819092e-01 -9.90150154e-01
-1.46307945e-01 1.45318896e-01 3.20779234e-01 -1.72830725e+00
1.98652178e-01 -7.43175328e-01 -7.82021135e-02 5.28527081e-01
3.14412296e-01 -8.48520637e-01 3.02416235e-01 5.68769395e-01
-1.22352159e+00 4.68969345e-01 1.21311843e+00 -4.14157927e-01
-5.05729437e-01 5.05683005e-01 -6.98175251e-01 7.37795770e-01
8.26681018e-01 -3.71819437e-01 -7.91877627e-01 -4.11828756e-01
3.95795852e-01 8.91217649e-01 -2.03856811e-01 -6.31656051e-01
7.36191809e-01 -1.35467148e+00 -3.08561120e-02 -7.09332407e-01
5.44733703e-01 -9.66503143e-01 5.19904792e-01 1.07731509e+00
-7.72284567e-01 2.14107960e-01 -2.62850374e-01 9.24808443e-01
2.05706254e-01 -6.98366046e-01 6.17854893e-01 -5.85322440e-01
-7.98362851e-01 8.07677358e-02 -1.01251030e+00 -1.85981110e-01
1.48407185e+00 -3.65710139e-01 -6.02020681e-01 -7.85805643e-01
-8.33309054e-01 8.23611617e-01 5.82862794e-01 -1.36516929e-01
3.13329250e-01 -7.27259815e-01 -5.04835486e-01 -4.12621468e-01
-2.76181340e-01 -2.25705300e-02 -1.72392100e-01 6.34358108e-01
-3.36075395e-01 6.64889097e-01 -3.30476984e-02 -1.26648903e-01
-1.13152039e+00 4.94958639e-01 4.92943704e-01 -6.02513492e-01
2.11869515e-02 6.17959619e-01 -4.67892922e-02 -3.03140674e-02
4.54047471e-01 -9.16039348e-02 -5.90315089e-02 6.40537292e-02
4.17826444e-01 9.16224837e-01 -3.89231890e-01 1.12248413e-01
-3.34186375e-01 2.65155762e-01 7.63856769e-02 -7.79956937e-01
1.20169389e+00 -6.74178600e-01 9.61970091e-02 1.87100038e-01
2.30897292e-01 2.01308995e-01 -1.61631525e+00 -2.30821922e-01
3.32627743e-01 -4.69819993e-01 3.00783157e-01 -1.24370039e+00
-5.03862202e-01 -1.72920480e-01 4.38553125e-01 5.13058662e-01
1.08424580e+00 -2.26654068e-01 -6.84058713e-03 9.38219607e-01
1.16385651e+00 -1.47191870e+00 -8.21360946e-02 3.22796047e-01
6.90097809e-01 -1.00695992e+00 2.87544578e-01 -1.28269121e-01
-6.99900091e-01 8.74230087e-01 4.14483577e-01 2.01161906e-01
1.88328966e-01 4.26985025e-01 1.26543334e-02 -7.40252063e-02
-1.60389686e+00 -4.41909581e-01 -6.37580633e-01 5.97207248e-01
-8.19799840e-01 3.86903465e-01 -3.04887652e-01 -1.13616865e-02
1.48816600e-01 -7.90451542e-02 8.11201036e-01 1.28902042e+00
-1.24858141e+00 -8.51841927e-01 -8.81191432e-01 3.80684495e-01
-1.75694615e-01 1.56594887e-01 -5.99544346e-01 6.99309647e-01
-1.81289569e-01 1.38303709e+00 7.38101974e-02 3.16525877e-01
-1.03906833e-01 -4.65129197e-01 3.10036480e-01 -5.53400517e-01
-1.84358865e-01 4.94471878e-01 5.78154027e-01 -7.37958431e-01
-4.74319965e-01 -9.74435151e-01 -1.06114995e+00 1.75397873e-01
-3.93002927e-01 7.14618504e-01 6.84702337e-01 7.66101003e-01
4.03132766e-01 1.50700450e-01 1.00916755e+00 -5.47912061e-01
-1.43410110e+00 -6.80759728e-01 -4.12524760e-01 -9.01057422e-01
4.79193747e-01 -8.04105222e-01 -5.83692074e-01 -7.59162188e-01] | [4.199602127075195, 2.0815181732177734] |
52c41f47-dc1c-468c-a997-e437928239b6 | the-active-filler-strategy-in-a-move-eager | null | null | https://aclanthology.org/W19-2901 | https://aclanthology.org/W19-2901.pdf | The Active-Filler Strategy in a Move-Eager Left-Corner Minimalist Grammar Parser | Recent psycholinguistic evidence suggests that human parsing of moved elements is {`}active{'}, and perhaps even {`}hyper-active{'}: it seems that a leftward-moved object is related to a verbal position rapidly, perhaps even before the transitivity information associated with the verb is available to the listener. This paper presents a formal, sound and complete parser for Minimalist Grammars whose search space contains branching points that we can identify as the locus of the decision to perform this kind of active gap-finding. This brings formal models of parsing into closer contact with recent psycholinguistic theorizing than was previously possible. | ["Milo{\\v{s}} Stanojevi{\\'c}", 'Tim Hunter', 'Edward Stabler'] | 2019-06-01 | null | null | null | ws-2019-6 | ['human-parsing'] | ['computer-vision'] | [ 3.45745713e-01 6.47829950e-01 8.70812461e-02 -5.18624067e-01
-7.50968337e-01 -9.53935921e-01 2.56979376e-01 6.47666454e-01
-7.29343534e-01 3.72559786e-01 4.25817370e-01 -9.83609200e-01
-2.06364319e-01 -7.82455444e-01 -5.69759548e-01 -3.53922188e-01
-2.40373388e-02 5.79609394e-01 6.50500953e-01 -4.56826627e-01
4.44570869e-01 2.96760738e-01 -1.29557633e+00 3.26666653e-01
4.91493225e-01 1.73752382e-01 8.00894916e-01 8.48076463e-01
-4.32305813e-01 6.01965606e-01 -5.16085565e-01 -6.40905559e-01
6.45803660e-02 -7.37567902e-01 -1.39865625e+00 -1.11651316e-01
1.89482987e-01 -8.02897811e-02 1.80231616e-01 1.27642179e+00
1.04253277e-01 2.53564000e-01 1.31984189e-01 -5.49753606e-01
-7.71835625e-01 1.16799998e+00 -3.60578030e-01 9.79180098e-01
6.21354103e-01 -4.35656756e-02 1.40131724e+00 -6.44915283e-01
9.36781824e-01 1.45627594e+00 3.00367534e-01 5.32865107e-01
-1.02940047e+00 -1.10469908e-02 4.24339056e-01 3.02865338e-02
-1.06554294e+00 -5.63438892e-01 7.36502588e-01 -3.00019175e-01
1.41998982e+00 3.98869187e-01 7.76768982e-01 5.35851359e-01
2.10923687e-01 5.79833806e-01 1.18024862e+00 -1.08201325e+00
6.27576560e-02 -2.16157854e-01 6.14733279e-01 7.68894494e-01
3.27744573e-01 3.18298399e-01 -9.01283622e-01 -1.80430621e-01
6.99286461e-01 -8.02737236e-01 -2.48579215e-02 4.30859774e-01
-1.00529826e+00 6.14669323e-01 -8.41151327e-02 8.53824496e-01
-4.81526673e-01 4.46013473e-02 1.72634900e-01 4.56774563e-01
5.12735248e-02 4.25582588e-01 -4.70925063e-01 -3.63038749e-01
-4.29429680e-01 8.27812180e-02 7.82330394e-01 9.39185679e-01
6.49757504e-01 -1.57803103e-01 4.95627999e-01 6.20961785e-01
5.67935824e-01 1.99746136e-02 5.76204062e-02 -1.15772796e+00
6.61045194e-01 5.75831652e-01 3.75371128e-02 -6.18459165e-01
-4.98983800e-01 1.44290686e-01 2.10729674e-01 1.01864278e-01
9.85790730e-01 4.00377065e-02 -3.46637100e-01 2.07011867e+00
4.68186915e-01 -6.13316536e-01 2.32909411e-01 7.29634285e-01
5.65894365e-01 8.20622325e-01 6.83392942e-01 -7.97479212e-01
1.84110379e+00 -1.91537142e-01 -6.62732363e-01 -6.79945350e-01
7.56995082e-01 -8.14977288e-01 1.36642599e+00 2.29180828e-01
-1.56571841e+00 -2.70562321e-01 -9.37269032e-01 -2.77079731e-01
7.24477172e-02 -5.67234337e-01 9.86001670e-01 6.59814477e-01
-9.88295376e-01 5.93007326e-01 -1.01468658e+00 -6.56727493e-01
-2.22295806e-01 1.56558931e-01 -4.58085716e-01 3.32679957e-01
-1.17993879e+00 9.49275136e-01 7.23035395e-01 5.79560161e-01
-3.09707671e-01 2.39331555e-02 -8.26847434e-01 -1.35379240e-01
4.75020438e-01 -2.59077191e-01 1.43054390e+00 -1.13301027e+00
-1.15440369e+00 1.46142828e+00 -4.19973135e-01 -1.38177559e-01
-1.12423621e-01 5.61277091e-04 -3.97569001e-01 3.87529075e-01
3.43161136e-01 1.94734216e-01 4.34003949e-01 -1.11495829e+00
-6.72220528e-01 -8.38040650e-01 3.09880584e-01 4.08548236e-01
3.50126266e-01 1.05274713e+00 -2.80923769e-02 -4.50014353e-01
1.16644096e+00 -7.73447037e-01 6.68104216e-02 -4.60556507e-01
-4.05898035e-01 -9.01042938e-01 -9.62881893e-02 -7.38162041e-01
1.43177021e+00 -2.14036012e+00 -1.76637545e-01 -7.13101309e-03
2.82601178e-01 -1.73200056e-01 1.94958255e-01 8.15148234e-01
-4.18688923e-01 3.97758514e-01 -2.98692603e-02 3.89373273e-01
1.04424998e-01 4.27189350e-01 3.00337654e-02 5.17849147e-01
-8.47594589e-02 7.27895916e-01 -6.57579124e-01 -6.78288460e-01
-1.95090264e-01 -2.84238636e-01 -5.41660905e-01 4.95328270e-02
3.74092683e-02 -1.15300499e-01 -6.74155176e-01 6.39069557e-01
3.41466188e-01 -3.85181420e-03 8.48695815e-01 2.75691926e-01
-6.66382551e-01 1.08838975e+00 -1.12008488e+00 1.38769281e+00
2.22152486e-01 3.73270959e-01 5.32644689e-01 -7.74091721e-01
6.26394093e-01 3.62036169e-01 -1.31378934e-01 -8.74112785e-01
2.28876665e-01 2.67472565e-01 8.13273966e-01 -5.88120818e-01
5.86671054e-01 -5.63509285e-01 -4.10617322e-01 5.34817398e-01
-1.77358374e-01 2.16062903e-01 2.26679668e-01 2.80561566e-01
1.06560147e+00 2.13976502e-01 6.02806330e-01 -8.09477448e-01
4.16098446e-01 2.41396874e-01 9.28114772e-01 7.79656708e-01
-3.66146028e-01 -7.40585104e-02 6.22078061e-01 -3.57806414e-01
-9.08231914e-01 -1.21611309e+00 -2.27789983e-01 1.75701761e+00
4.09452915e-01 -5.42288601e-01 -1.11627674e+00 -1.78671733e-01
-7.05532908e-01 1.10873485e+00 -4.99167711e-01 1.33240700e-01
-1.23062646e+00 -7.11021364e-01 4.30196941e-01 7.32235491e-01
1.41523238e-02 -1.68051052e+00 -1.22504365e+00 5.96326530e-01
-3.58598530e-01 -9.33357596e-01 -5.19932769e-02 7.04626858e-01
-9.41876709e-01 -1.09879911e+00 3.50427508e-01 -1.01242316e+00
5.68863750e-01 4.38922234e-02 1.08898187e+00 6.54626906e-01
-6.40243739e-02 3.80022734e-01 -5.86506486e-01 -4.80870277e-01
-7.67024398e-01 -3.89492601e-01 -2.89443791e-01 -6.62516832e-01
5.79802752e-01 -4.98244882e-01 -2.05095410e-01 -1.20581230e-02
-4.83380049e-01 2.27762964e-02 2.00592890e-01 3.97361517e-01
5.14119804e-01 -1.32245243e-01 2.79636323e-01 -8.89679432e-01
8.24027836e-01 -2.64163792e-01 -4.83823329e-01 3.71245801e-01
-1.21955700e-01 3.25653479e-02 2.33769700e-01 -1.15397945e-01
-1.30143094e+00 -2.38549411e-01 -3.90247822e-01 6.81047499e-01
-4.78043407e-01 4.97586638e-01 -4.76502806e-01 4.68333840e-01
8.29902887e-01 -4.28162925e-02 -2.20400929e-01 -5.25852442e-01
1.90184161e-01 1.40753835e-01 5.95262706e-01 -1.07480943e+00
4.20074761e-01 1.35284811e-01 -1.69510663e-01 -8.96478713e-01
-5.27220845e-01 -1.22081496e-01 -7.12393880e-01 -1.04980975e-01
1.03851390e+00 -4.47136521e-01 -9.00835693e-01 -4.50937776e-03
-1.49552453e+00 -2.45161861e-01 -4.35046792e-01 5.99608481e-01
-6.73799992e-01 5.77170312e-01 -7.82828212e-01 -9.77928638e-01
8.38088803e-03 -8.06279302e-01 5.14536440e-01 1.60543159e-01
-6.62904263e-01 -6.89227045e-01 -8.43354985e-02 2.48558477e-01
-2.72426665e-01 -2.21273601e-01 1.55152440e+00 -1.17236400e+00
-4.12925243e-01 -1.26703754e-01 1.70871541e-01 -1.31692961e-01
-1.84198469e-01 1.10980719e-01 -4.93825614e-01 2.24606842e-01
3.15647632e-01 -1.62630096e-01 2.47944981e-01 2.98348218e-01
2.64828593e-01 -1.59532309e-01 -2.44141728e-01 -3.57176401e-02
1.38638377e+00 8.32334220e-01 4.75996852e-01 3.52521390e-01
8.10403228e-02 1.05681407e+00 5.12499928e-01 -4.97901589e-02
6.31002367e-01 3.68984729e-01 1.38705164e-01 3.65954757e-01
4.82404083e-02 -3.60295832e-01 2.45159030e-01 8.97280037e-01
-3.29162121e-01 -4.34379816e-01 -1.09420705e+00 5.81806421e-01
-1.62526524e+00 -1.11697435e+00 -5.39311111e-01 1.97096527e+00
7.99767137e-01 5.51832795e-01 -5.18218614e-02 3.53148699e-01
1.01124287e+00 1.82548121e-01 3.83891053e-02 -1.11858010e+00
2.35946681e-02 3.04662108e-01 -1.46928243e-03 8.52916718e-01
-5.28809905e-01 1.68298066e+00 7.62387180e+00 1.64907590e-01
-4.73463625e-01 2.98947781e-01 2.77231429e-02 2.00573981e-01
-4.72763330e-01 4.46113646e-01 -7.21794844e-01 1.37905180e-01
9.87088263e-01 -1.58696502e-01 2.32151002e-01 5.71168005e-01
3.09008807e-01 -8.29156339e-01 -1.22505879e+00 3.75599265e-01
-2.86707252e-01 -8.79538536e-01 -1.04567051e-01 9.68816783e-03
-2.94527531e-01 -4.65212047e-01 -5.32690346e-01 -8.54773521e-02
4.13023055e-01 -5.72853982e-01 1.25461745e+00 8.14206749e-02
1.36786863e-01 -5.08460164e-01 1.35847643e-01 7.17438102e-01
-1.21923506e+00 1.32051017e-02 -4.53660607e-01 -4.09351349e-01
6.17491901e-01 3.76341082e-02 -4.97331113e-01 1.33032158e-01
6.94490373e-01 -4.35945511e-01 -3.77854735e-01 4.46603298e-01
-7.43442535e-01 9.86723363e-01 -4.41139460e-01 -3.71045649e-01
2.86442548e-01 -2.70366609e-01 9.91143405e-01 9.34035897e-01
4.10795175e-02 1.00650167e+00 1.09033778e-01 9.34830546e-01
4.69547957e-01 4.68281269e-01 -4.06901628e-01 -7.54090846e-02
5.82887232e-01 7.59824634e-01 -1.46825016e+00 -1.10340320e-01
-3.73072594e-01 7.17220128e-01 4.35772359e-01 -2.06090342e-02
-3.80503356e-01 -3.03845871e-02 1.89684585e-01 2.80994743e-01
-8.05388391e-03 -5.39369166e-01 -6.64607361e-02 -6.72381341e-01
1.64586589e-01 -3.96076024e-01 6.24749184e-01 -8.14523399e-01
-8.42053294e-01 6.10152960e-01 2.97743112e-01 -1.52005702e-01
-4.47774917e-01 -7.57905722e-01 -6.26243711e-01 1.08547306e+00
-6.11738741e-01 -8.47421885e-01 5.37527323e-01 5.59408665e-01
6.17026329e-01 2.16384396e-01 8.61844718e-01 -3.75883609e-01
-3.65369737e-01 2.54231244e-01 -8.83696556e-01 2.58193731e-01
-5.18980324e-02 -1.36213410e+00 5.99536657e-01 1.31425846e+00
4.24763113e-01 1.04409838e+00 9.58246469e-01 -8.23278725e-01
-1.31782413e+00 -3.55514325e-02 1.65192676e+00 -5.63681602e-01
6.58905983e-01 -3.08431745e-01 -1.06582391e+00 8.75994503e-01
4.13789809e-01 -3.61046702e-01 8.47111583e-01 4.61806923e-01
-1.52106345e-01 3.47106934e-01 -8.81632268e-01 5.56095600e-01
1.42878163e+00 -7.05114007e-01 -1.59869039e+00 1.44491076e-01
7.86487579e-01 -2.24937662e-01 -4.39110130e-01 1.42270863e-01
3.52107674e-01 -1.07891154e+00 5.98958313e-01 -8.03097486e-01
-1.82800565e-03 -1.21183008e-01 -2.41363555e-01 -6.28181458e-01
-7.33460367e-01 -8.87559235e-01 4.73210216e-01 1.15385926e+00
6.87904060e-01 -4.91674364e-01 5.82947731e-01 8.62996876e-01
-5.61958194e-01 -1.83022097e-01 -1.23686457e+00 -3.80029589e-01
2.22190574e-01 -9.41691399e-01 1.91596180e-01 8.48867118e-01
5.73731542e-01 4.67476934e-01 3.80741835e-01 2.52275348e-01
4.20404434e-01 1.11240372e-01 -1.48349270e-01 -1.39794993e+00
-3.55471075e-01 -3.94239366e-01 -7.07626194e-02 -9.67848480e-01
1.47453010e-01 -9.92636263e-01 4.19790626e-01 -1.50821424e+00
-8.21246058e-02 -4.66650784e-01 -1.17389552e-01 5.87535977e-01
-1.11947194e-01 -3.39134395e-01 3.29231888e-01 3.91253084e-02
-3.62323254e-01 -2.62522489e-01 1.23615873e+00 5.50881505e-01
-4.60337102e-01 -1.35654569e-01 -1.23806000e+00 1.25796676e+00
7.02443123e-01 -6.65447950e-01 -3.06701213e-01 -5.10637343e-01
7.91959226e-01 6.69152677e-01 2.62019992e-01 -1.91067696e-01
3.57021511e-01 -4.23515260e-01 -7.74566978e-02 -5.21770835e-01
-4.66099009e-02 -5.59256494e-01 -6.54744655e-02 3.07558566e-01
-3.92229766e-01 7.74833918e-01 3.20704013e-01 1.90109164e-01
2.13981599e-01 -8.23977053e-01 6.34726226e-01 -6.18255973e-01
-1.16095090e+00 -4.04819250e-01 -9.83786702e-01 4.38695669e-01
7.49693096e-01 -5.16745508e-01 -1.64152205e-01 2.49386817e-01
-1.42952168e+00 -2.89635479e-01 1.13769822e-01 1.56172141e-01
4.99434948e-01 -7.32758641e-01 -3.37346554e-01 -1.12733699e-03
-1.23332888e-01 -2.90050983e-01 2.28136517e-02 6.63368106e-01
-6.23303831e-01 1.52360305e-01 -1.30107179e-01 -1.96976930e-01
-1.28875220e+00 6.26695871e-01 2.06452683e-01 2.66504556e-01
-8.29298973e-01 1.21373308e+00 7.25035667e-02 -5.49774952e-02
-1.88987434e-01 -2.89672107e-01 -1.30711958e-01 1.73660710e-01
5.05227685e-01 2.25508034e-01 -2.44338408e-01 -9.76842463e-01
-6.28233135e-01 1.49908975e-01 -1.62940118e-02 -5.19787610e-01
9.75618541e-01 -4.65227872e-01 -7.65721381e-01 8.70144129e-01
6.08927429e-01 2.21043468e-01 -7.42110968e-01 -1.29974902e-01
3.32054198e-01 -2.46921197e-01 -3.77586991e-01 -8.38011742e-01
-2.28724658e-01 6.79460406e-01 3.94146591e-02 7.69746959e-01
7.66885102e-01 6.74728334e-01 1.94221586e-01 5.70226252e-01
5.80649853e-01 -1.46343017e+00 -2.42545277e-01 2.80439258e-01
8.11913192e-01 -6.60007715e-01 -1.75550252e-01 -8.37412417e-01
-4.28858161e-01 1.00505042e+00 6.75936460e-01 9.88463126e-03
4.56742555e-01 3.75234187e-01 -6.62573725e-02 -4.92021322e-01
-7.62020051e-01 -2.93316990e-01 -2.64766932e-01 4.96747404e-01
7.79410064e-01 2.08141297e-01 -1.12570119e+00 6.58852994e-01
-7.02071548e-01 -7.66297758e-01 4.42952543e-01 1.33202827e+00
-1.01549363e+00 -1.15813518e+00 -5.46330512e-01 -1.00225173e-01
-7.58916140e-01 -3.65892440e-01 -5.66088319e-01 1.26899219e+00
3.90747964e-01 1.19754100e+00 4.29228574e-01 1.68324307e-01
3.64109665e-01 3.93508196e-01 9.04446423e-01 -9.97032404e-01
-6.86492085e-01 5.22052467e-01 4.77913827e-01 -4.90773857e-01
-5.91116905e-01 -1.15510249e+00 -2.00156665e+00 -1.43010125e-01
-3.84690464e-01 5.56078494e-01 2.32080206e-01 1.21618164e+00
-4.06464845e-01 9.99200344e-02 4.14187051e-02 -2.90571183e-01
-2.42571592e-01 -5.46833873e-01 -6.34878933e-01 7.07582831e-02
-1.48746416e-01 -2.80853242e-01 -3.84180129e-01 4.28159609e-02] | [10.32627010345459, 9.347541809082031] |
2ee8419e-4472-4224-8256-cffa0b889502 | cascaded-lstms-based-deep-reinforcement | 1910.14229 | null | https://arxiv.org/abs/1910.14229v1 | https://arxiv.org/pdf/1910.14229v1.pdf | Cascaded LSTMs based Deep Reinforcement Learning for Goal-driven Dialogue | This paper proposes a deep neural network model for joint modeling Natural Language Understanding (NLU) and Dialogue Management (DM) in goal-driven dialogue systems. There are three parts in this model. A Long Short-Term Memory (LSTM) at the bottom of the network encodes utterances in each dialogue turn into a turn embedding. Dialogue embeddings are learned by a LSTM at the middle of the network, and updated by the feeding of all turn embeddings. The top part is a forward Deep Neural Network which converts dialogue embeddings into the Q-values of different dialogue actions. The cascaded LSTMs based reinforcement learning network is jointly optimized by making use of the rewards received at each dialogue turn as the only supervision information. There is no explicit NLU and dialogue states in the network. Experimental results show that our model outperforms both traditional Markov Decision Process (MDP) model and single LSTM with Deep Q-Network on meeting room booking tasks. Visualization of dialogue embeddings illustrates that the model can learn the representation of dialogue states. | ['Hong Chen', 'Yue Ma', 'Xiaojie Wang', 'Zhenjiang Dong'] | 2019-10-31 | null | null | null | null | ['dialogue-management'] | ['natural-language-processing'] | [-2.04327956e-01 1.01989758e+00 -5.85183464e-02 -6.73236847e-01
-2.91830242e-01 -1.86081588e-01 8.56630564e-01 7.13944063e-02
-5.92197180e-01 8.14912200e-01 8.76412392e-01 -3.82009387e-01
3.95505846e-01 -9.13242638e-01 -2.87503660e-01 -3.32130283e-01
4.73041162e-02 8.49802732e-01 7.26378784e-02 -9.00710523e-01
4.83034074e-01 -5.05250953e-02 -8.53617013e-01 5.00077844e-01
6.93920732e-01 8.59593511e-01 2.74105817e-01 1.32894278e+00
-7.61669219e-01 1.66031969e+00 -8.34374487e-01 -1.22125916e-01
-1.78631201e-01 -6.07361555e-01 -1.42616916e+00 2.74096313e-03
-5.61862588e-01 -1.12333608e+00 -7.56442308e-01 5.51880658e-01
6.42110705e-01 7.18425989e-01 5.84659219e-01 -1.16824794e+00
-7.99017549e-01 6.39686704e-01 5.71096465e-02 -1.29855305e-01
6.93965018e-01 3.96082252e-01 9.36006427e-01 -6.35407507e-01
2.57985890e-01 1.96362042e+00 1.69217855e-01 9.66619194e-01
-8.65965188e-01 -1.63525138e-02 2.74881244e-01 1.42632946e-01
-3.25737745e-01 -2.53259957e-01 5.21869779e-01 -3.16650659e-01
1.49307323e+00 -3.25047791e-01 5.97271323e-01 1.01873279e+00
5.04661977e-01 1.21565676e+00 8.33467484e-01 -6.54259026e-01
2.68349707e-01 3.73794615e-01 4.74352509e-01 9.38552797e-01
-1.02535808e+00 -3.35505567e-02 -5.80925047e-01 -3.68623376e-01
9.61166084e-01 1.85389265e-01 1.60684079e-01 -2.94025242e-01
-1.00006616e+00 1.26027381e+00 3.24123383e-01 1.45269558e-01
-4.07482594e-01 3.90525497e-02 7.70676076e-01 7.34701872e-01
3.72491091e-01 4.38069224e-01 -5.34716725e-01 -6.50349200e-01
-6.81395009e-02 3.21383327e-01 1.49442530e+00 7.95524240e-01
6.50421143e-01 3.77115011e-02 -7.94924676e-01 1.16446435e+00
6.05295122e-01 8.89438689e-02 7.37518907e-01 -1.19835770e+00
5.47329068e-01 6.57532573e-01 4.34807777e-01 -5.80555677e-01
-5.02077699e-01 6.12475038e-01 -6.79986358e-01 6.80744322e-03
1.88901827e-01 -9.79153633e-01 -6.26426697e-01 1.40108275e+00
1.90443590e-01 -2.52650440e-01 5.81035256e-01 6.88488245e-01
9.82963800e-01 1.11006021e+00 8.53898898e-02 -4.78587747e-02
1.36276448e+00 -1.42087877e+00 -1.35758924e+00 -7.49098957e-02
8.60364795e-01 -3.58008087e-01 8.47616911e-01 1.73913553e-01
-1.11135030e+00 -6.77231729e-01 -8.39922309e-01 -4.42497998e-01
-2.98935175e-01 -9.60031301e-02 3.17824870e-01 1.21283859e-01
-1.26444054e+00 5.00321507e-01 -6.06948018e-01 -3.94269347e-01
-2.33650178e-01 4.64860499e-01 3.65649313e-02 2.40481675e-01
-1.81527722e+00 1.35988426e+00 2.44817182e-01 4.26119447e-01
-1.17750037e+00 1.35422766e-01 -1.28064227e+00 2.00147554e-01
4.58083414e-02 -5.36550462e-01 2.03552723e+00 -8.72002542e-01
-2.70671892e+00 4.67555553e-01 -1.77476257e-01 -7.03041673e-01
3.24409813e-01 -3.51926863e-01 -1.71711259e-02 2.77487412e-02
-3.18984151e-01 1.08336937e+00 5.53709984e-01 -9.56028044e-01
-6.98301315e-01 -2.31795922e-01 5.40073395e-01 8.76423419e-01
-2.06168592e-01 -6.04830049e-02 -1.57407913e-02 1.34181097e-01
-4.07380402e-01 -7.53762543e-01 -5.68265796e-01 -3.96534413e-01
-4.41103309e-01 -9.02186334e-01 6.62237048e-01 -7.97458231e-01
1.32201505e+00 -1.74414587e+00 2.31701404e-01 -4.23464924e-01
1.06027760e-01 2.80570686e-01 -2.56530821e-01 1.14207661e+00
3.00535232e-01 -8.63004252e-02 6.15813881e-02 -5.64399302e-01
3.43431592e-01 5.91959000e-01 -3.97955060e-01 -1.53899640e-01
2.42968082e-01 9.76428270e-01 -1.03888118e+00 -3.14366490e-01
4.04716223e-01 3.95991020e-02 -3.60050410e-01 1.18545699e+00
-6.49120629e-01 5.00595510e-01 -5.12491226e-01 -2.22249880e-01
2.88299620e-01 -1.14554167e-02 4.70874995e-01 4.92393911e-01
-1.39058754e-01 8.59506309e-01 -8.80163491e-01 1.92430186e+00
-6.11171722e-01 5.15121043e-01 1.07209995e-01 -9.07166243e-01
1.23252547e+00 7.52728879e-01 -9.55373049e-02 -7.66378403e-01
1.02320910e-01 -3.38945448e-01 7.31964782e-02 -8.32519531e-01
9.06230628e-01 -3.46596539e-01 -3.66134018e-01 9.79065239e-01
4.59294766e-01 -1.35130331e-01 -1.95504457e-01 3.95975322e-01
7.17126966e-01 1.59600347e-01 2.19503343e-01 1.40169829e-01
6.22591376e-01 -1.10833898e-01 4.21908647e-01 6.88406885e-01
-4.94827598e-01 3.53227593e-02 1.16859174e+00 -7.10948050e-01
-8.37891519e-01 -7.40629196e-01 3.94693047e-01 1.84073770e+00
-5.51459715e-02 -2.07899660e-01 -7.56974638e-01 -6.84164047e-01
-8.45747367e-02 9.78735089e-01 -7.83925712e-01 -2.68118352e-01
-5.77754080e-01 -2.42383510e-01 3.42701256e-01 4.98614758e-01
6.68411970e-01 -1.68563414e+00 -3.81958485e-01 4.88652378e-01
-1.69781223e-01 -7.54349172e-01 -4.08058971e-01 2.59948224e-01
-7.89162993e-01 -7.34749436e-01 -6.59351587e-01 -1.03268981e+00
3.72477025e-01 -2.05999762e-01 9.35602605e-01 -2.59483635e-01
2.09968895e-01 4.97364044e-01 -2.11324066e-01 -3.93057704e-01
-1.00197709e+00 -1.38448616e-02 1.79172270e-02 -9.07585993e-02
6.34509146e-01 -8.52101743e-02 -5.59406340e-01 1.28589720e-01
-4.80537117e-01 1.62996978e-01 1.36304006e-01 1.24578559e+00
-2.57435352e-01 -6.65569007e-01 7.90719271e-01 -8.67224932e-01
1.60526669e+00 -4.02898282e-01 -2.36572832e-01 2.72832483e-01
-2.92789251e-01 4.68714088e-01 6.07825160e-01 -1.50797382e-01
-1.53620732e+00 -4.00336713e-01 -2.57955849e-01 1.45834893e-01
-2.30843991e-01 2.38599852e-01 1.12568654e-01 6.90656543e-01
3.51006806e-01 1.97841868e-01 6.31225824e-01 -3.90491635e-01
6.59841537e-01 1.18640625e+00 -4.29590885e-03 -4.35507268e-01
-4.58363146e-02 -1.25034168e-01 -6.98477805e-01 -7.76344240e-01
-7.46333361e-01 -3.06959808e-01 -8.64030600e-01 -2.90634036e-01
1.32522058e+00 -7.56583333e-01 -1.25965774e+00 4.92122799e-01
-1.33400989e+00 -7.79064775e-01 -2.83600688e-01 3.44201565e-01
-8.19378018e-01 3.18896234e-01 -1.26663005e+00 -1.32763422e+00
-3.47351879e-01 -1.11828029e+00 6.63756669e-01 5.55633307e-01
-3.46407473e-01 -1.41266036e+00 3.15171748e-01 3.97234976e-01
2.77880192e-01 -1.65560141e-01 1.04764175e+00 -1.09827960e+00
-8.36122781e-02 7.25990981e-02 2.13093050e-02 5.99265993e-01
2.12464735e-01 -1.86086550e-01 -1.03345942e+00 -7.03127235e-02
1.49465248e-01 -9.80279207e-01 6.23660684e-01 5.35408676e-01
5.61216474e-01 -4.98334885e-01 -5.23956418e-02 -2.84843951e-01
8.01193237e-01 5.70887208e-01 4.78920132e-01 1.65562868e-01
5.39297700e-01 1.06313324e+00 8.47149491e-01 7.05178320e-01
9.46350217e-01 2.09083378e-01 4.19487894e-01 -1.88494027e-01
5.54844141e-01 -4.60105270e-01 7.63303816e-01 1.09954953e+00
3.31553996e-01 -1.11183852e-01 -8.09273481e-01 3.92739117e-01
-2.28485656e+00 -8.15540373e-01 1.89523533e-01 1.89045393e+00
9.91815805e-01 1.59715906e-01 1.50372401e-01 -5.14007628e-01
5.88573694e-01 5.39215147e-01 -6.79454148e-01 -1.59404540e+00
4.57545429e-01 -1.95959896e-01 -6.84972480e-02 1.10096121e+00
-9.48395014e-01 1.08763707e+00 6.29382420e+00 7.83629268e-02
-6.22998178e-01 -3.33965160e-02 7.59943128e-01 1.60745144e-01
-9.51702222e-02 -7.61874020e-02 -7.97865629e-01 5.57976924e-02
1.35007966e+00 -1.15696393e-01 1.89672500e-01 7.17852533e-01
5.56149244e-01 -3.36987674e-01 -1.25883949e+00 4.65101242e-01
-8.88424739e-02 -1.11606300e+00 6.88108727e-02 -2.37249769e-02
5.36256194e-01 -1.28988549e-01 -1.55234188e-01 1.08303368e+00
1.19370842e+00 -1.12198961e+00 -6.01570308e-03 8.41391683e-01
2.80992627e-01 -1.07497871e+00 8.78776848e-01 8.80413473e-01
-6.84744477e-01 -4.65190440e-01 -6.16784453e-01 -4.95133758e-01
3.07885647e-01 -2.64067441e-01 -1.51186347e+00 2.22352609e-01
1.70000210e-01 5.86645722e-01 2.26936802e-01 1.80583254e-01
-3.98628026e-01 2.97049373e-01 2.52435803e-01 -4.62896287e-01
8.07371736e-01 -4.86803025e-01 1.73105747e-01 1.18939698e+00
-4.56047267e-01 2.10685343e-01 5.57284176e-01 6.28981054e-01
-1.51181310e-01 -6.46698698e-02 -7.39012837e-01 -1.22098356e-01
3.34605426e-01 1.08324993e+00 1.45680517e-01 -5.49087584e-01
-2.58498639e-01 1.15885341e+00 4.60524768e-01 4.36541557e-01
-3.50033760e-01 -4.64162618e-01 6.69633746e-01 -4.59002614e-01
-1.68023668e-02 -2.00359717e-01 1.29171044e-01 -7.61044562e-01
-5.37925124e-01 -7.17960000e-01 2.69684494e-01 -4.70444530e-01
-1.11628640e+00 5.61643720e-01 -2.47610271e-01 -7.98823059e-01
-9.51210320e-01 -4.15260971e-01 -8.25039625e-01 1.30783927e+00
-1.67892158e+00 -7.30174780e-01 1.58819169e-01 3.20595443e-01
1.05243897e+00 -2.73204505e-01 1.48957264e+00 -5.39941013e-01
-7.14432657e-01 1.64658561e-01 2.70240366e-01 4.66417670e-01
8.35752189e-01 -1.73498523e+00 5.49492955e-01 -1.44193396e-01
-6.54322207e-01 5.48664451e-01 4.95989054e-01 -3.76791358e-01
-1.06385696e+00 -7.30759263e-01 1.02788508e+00 -2.64108777e-01
5.23894548e-01 -4.41095531e-01 -8.61142159e-01 7.59966791e-01
1.01874363e+00 -6.53978229e-01 1.04413116e+00 2.02055961e-01
3.88719559e-01 3.04172069e-01 -1.00245810e+00 5.49444258e-01
8.21765810e-02 -7.64630556e-01 -1.14417017e+00 3.95026714e-01
1.12898767e+00 -4.42892015e-01 -1.08983374e+00 -2.78455794e-01
3.45919609e-01 -8.23072791e-01 5.59402704e-01 -1.05304348e+00
7.49930918e-01 3.27037305e-01 3.07242602e-01 -1.53642833e+00
6.32091612e-02 -8.86959314e-01 -2.69989669e-01 8.69839787e-01
4.48922813e-01 -5.07452548e-01 5.73183656e-01 1.23299897e+00
-1.19482055e-01 -1.04128182e+00 -6.10851228e-01 7.67368311e-03
2.47590989e-01 2.25712314e-01 3.83361220e-01 7.63724029e-01
6.42383277e-01 9.71013725e-01 -7.36651659e-01 -3.23850125e-01
-5.83841726e-02 -1.03227220e-01 9.05811250e-01 -9.56410706e-01
-2.53774524e-01 -7.68478960e-02 3.82565498e-01 -1.89429855e+00
2.34212399e-01 -4.11050647e-01 3.82409304e-01 -2.21511865e+00
-3.61438058e-02 -2.62965560e-02 -1.66843310e-01 3.09724540e-01
-3.45363468e-01 -1.11979020e+00 2.13520348e-01 -5.36339916e-02
-7.69436836e-01 1.29655015e+00 1.58965552e+00 -9.08813104e-02
-7.41750419e-01 1.26822799e-01 -3.48996758e-01 7.91441381e-01
8.32080722e-01 -2.37926453e-01 -4.66756672e-01 -3.53904247e-01
-1.97618484e-01 1.05386436e+00 -2.33175203e-01 -3.70928973e-01
4.70786065e-01 -1.43748105e-01 2.31761396e-01 -6.51835203e-01
6.74438357e-01 -2.52371609e-01 -1.07944369e+00 6.03047490e-01
-1.18329597e+00 9.29663554e-02 -1.26627544e-02 7.35353470e-01
-2.81963378e-01 -3.45916599e-01 6.10408247e-01 -5.01764536e-01
-6.00358725e-01 1.51557764e-02 -1.03292418e+00 -1.52198225e-01
8.29978645e-01 -1.41795367e-01 4.93640639e-03 -9.99247611e-01
-1.05429065e+00 1.13192999e+00 -1.63310349e-01 7.53549993e-01
8.14418495e-01 -9.59350169e-01 -5.45529723e-01 1.70364715e-02
-3.28550190e-01 2.19025359e-01 3.85472953e-01 2.73942649e-01
-3.26279432e-01 6.83414042e-01 -3.80375504e-01 -3.17972124e-01
-1.09648633e+00 2.19748229e-01 7.54587471e-01 -9.30574775e-01
-3.11150342e-01 1.04447746e+00 2.15723515e-01 -1.29098678e+00
6.38027132e-01 -3.54603022e-01 -1.02744460e+00 2.89425135e-01
5.46511531e-01 2.65662670e-01 -4.86625850e-01 -2.34825462e-01
1.81310803e-01 -2.51009196e-01 -5.12556911e-01 -6.25635147e-01
1.24212432e+00 -3.26890767e-01 -2.07700983e-01 1.22781384e+00
1.11030436e+00 -8.78729701e-01 -1.45272970e+00 -4.63683337e-01
6.47668019e-02 1.77913591e-01 -2.95368522e-01 -7.71269739e-01
-2.18003839e-01 1.42836082e+00 3.81157577e-01 3.80795062e-01
3.64580154e-01 -5.45704603e-01 1.07350135e+00 9.87458169e-01
2.90252063e-02 -1.80244994e+00 6.60984635e-01 1.24851179e+00
9.23495054e-01 -1.43097377e+00 -5.28656304e-01 5.25962114e-01
-1.39050376e+00 1.39373899e+00 9.57345366e-01 -2.54406959e-01
5.75126052e-01 -9.48604867e-02 3.86731833e-01 -1.81108057e-01
-1.40449536e+00 1.66976556e-01 -2.88577855e-01 4.36912119e-01
6.43703759e-01 1.30645052e-01 3.19472998e-02 4.22550529e-01
-1.13163134e-02 -1.20112211e-01 7.00434327e-01 1.09665024e+00
-8.31253648e-01 -1.34393823e+00 -1.41478136e-01 3.77870411e-01
-1.03534333e-01 3.12430542e-02 -6.52516425e-01 3.39723736e-01
-5.45716822e-01 1.32967234e+00 3.56766820e-01 -4.27163124e-01
4.15690273e-01 5.35230339e-01 2.15934385e-02 -1.15233719e+00
-1.19037485e+00 -2.58181989e-01 4.91145343e-01 -4.58206236e-01
-2.54308600e-02 -1.08516425e-01 -1.65496385e+00 -2.97170490e-01
-2.63566941e-01 5.29264510e-01 5.39214492e-01 9.83167946e-01
3.56782019e-01 8.51402164e-01 1.08584917e+00 -6.15726709e-01
-1.16083610e+00 -1.56656206e+00 -5.40026069e-01 3.26113887e-02
6.28608823e-01 -2.70117640e-01 3.55338715e-02 -2.69833595e-01] | [13.001531600952148, 7.948004245758057] |
81f90aa9-b811-491b-adb7-595ae03c13c5 | learning-the-best-pooling-strategy-for-visual | 2011.04305 | null | https://arxiv.org/abs/2011.04305v5 | https://arxiv.org/pdf/2011.04305v5.pdf | Learning the Best Pooling Strategy for Visual Semantic Embedding | Visual Semantic Embedding (VSE) is a dominant approach for vision-language retrieval, which aims at learning a deep embedding space such that visual data are embedded close to their semantic text labels or descriptions. Recent VSE models use complex methods to better contextualize and aggregate multi-modal features into holistic embeddings. However, we discover that surprisingly simple (but carefully selected) global pooling functions (e.g., max pooling) outperform those complex models, across different feature extractors. Despite its simplicity and effectiveness, seeking the best pooling function for different data modality and feature extractor is costly and tedious, especially when the size of features varies (e.g., text, video). Therefore, we propose a Generalized Pooling Operator (GPO), which learns to automatically adapt itself to the best pooling strategy for different features, requiring no manual tuning while staying effective and efficient. We extend the VSE model using this proposed GPO and denote it as VSE$\infty$. Without bells and whistles, VSE$\infty$ outperforms previous VSE methods significantly on image-text retrieval benchmarks across popular feature extractors. With a simple adaptation, variants of VSE$\infty$ further demonstrate its strength by achieving the new state of the art on two video-text retrieval datasets. Comprehensive experiments and visualizations confirm that GPO always discovers the best pooling strategy and can be a plug-and-play feature aggregation module for standard VSE models. Code and pre-trained models are available at https://vse-infty.github.io. | ['Changhu Wang', 'Yuning Jiang', 'Hao Wu', 'Hexiang Hu', 'Jiacheng Chen'] | 2020-11-09 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Chen_Learning_the_Best_Pooling_Strategy_for_Visual_Semantic_Embedding_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Chen_Learning_the_Best_Pooling_Strategy_for_Visual_Semantic_Embedding_CVPR_2021_paper.pdf | cvpr-2021-1 | ['video-text-retrieval', 'cross-modal-information-retrieval'] | ['computer-vision', 'miscellaneous'] | [-1.95216402e-01 -4.38202381e-01 -2.70376593e-01 -2.38455147e-01
-6.83156312e-01 -6.15139127e-01 8.49731445e-01 2.51617115e-02
-5.91814101e-01 1.30123615e-01 3.25386345e-01 -3.84193435e-02
-3.25304300e-01 -5.48537016e-01 -6.14378870e-01 -5.79141974e-01
-6.13864101e-02 -7.60594532e-02 3.12784493e-01 -1.85783565e-01
2.70951003e-01 3.48343223e-01 -1.59912813e+00 1.84292853e-01
3.52345884e-01 1.21294463e+00 4.29094315e-01 5.84481537e-01
-2.41912186e-01 2.13426635e-01 -4.89668190e-01 -3.39365125e-01
4.77492303e-01 -8.58647376e-03 -7.11205363e-01 -1.16610892e-01
5.87368369e-01 -3.99723142e-01 -6.33597493e-01 9.43866551e-01
7.05617428e-01 1.68241113e-01 6.50640905e-01 -1.52604830e+00
-1.18395603e+00 3.27770621e-01 -4.16649818e-01 1.51458681e-01
3.09102029e-01 2.62964904e-01 1.53018486e+00 -1.36240506e+00
6.53319776e-01 1.19884908e+00 5.15362740e-01 5.16938686e-01
-1.25544405e+00 -5.62950790e-01 2.62948006e-01 3.42331499e-01
-1.73005021e+00 -3.33858162e-01 5.81475973e-01 -4.38492894e-01
9.84809279e-01 3.78073454e-01 5.60159624e-01 1.16485560e+00
1.79623663e-01 1.03036630e+00 8.42394412e-01 -2.51125634e-01
9.51119885e-02 3.15765858e-01 3.05976450e-01 8.77682507e-01
2.98490494e-01 -2.98672110e-01 -8.50253463e-01 -1.99731290e-01
4.47712451e-01 3.65006000e-01 -6.16965175e-01 -5.52586675e-01
-1.42103875e+00 7.82703996e-01 7.05938935e-01 2.80846089e-01
-2.32584760e-01 4.48914558e-01 3.24788332e-01 4.38520730e-01
1.99748144e-01 5.81001639e-01 -3.71501982e-01 -6.82474971e-02
-9.62378621e-01 3.29760343e-01 4.50744271e-01 9.27469373e-01
1.00873899e+00 -4.04651970e-01 -5.90780258e-01 1.09716153e+00
3.40922683e-01 4.78681654e-01 5.22865534e-01 -7.62124360e-01
1.87446043e-01 8.43971670e-01 1.38696462e-01 -1.06205440e+00
-3.14253360e-01 -2.41799682e-01 -5.26257336e-01 1.07196540e-01
9.80361179e-02 9.11514983e-02 -1.01369929e+00 1.63411629e+00
8.63095149e-02 -1.15879267e-01 -1.49365559e-01 9.89575982e-01
1.07152271e+00 6.68689251e-01 1.76815819e-02 2.95438021e-01
1.57153177e+00 -1.19402182e+00 -5.76733410e-01 -2.62597293e-01
6.06927395e-01 -7.59752333e-01 1.52944791e+00 1.00013793e-01
-8.34449351e-01 -2.62261897e-01 -1.05773699e+00 -3.90791684e-01
-7.92836308e-01 2.23528206e-01 5.44840157e-01 3.50931138e-01
-1.41712737e+00 5.78444123e-01 -6.94514215e-01 -5.91713607e-01
6.50186837e-01 4.36098874e-01 -6.85322642e-01 -1.58463523e-01
-1.10993207e+00 6.85099304e-01 2.98325837e-01 6.71487954e-03
-5.95221698e-01 -7.70769238e-01 -8.22757065e-01 3.75886738e-01
5.23060620e-01 -9.08065259e-01 1.00546372e+00 -7.51454353e-01
-1.47732604e+00 7.99100995e-01 -2.10421905e-01 -3.09156001e-01
2.59696811e-01 -4.21282649e-01 -1.01346418e-01 4.94821012e-01
-1.01903029e-01 1.03940606e+00 1.13852012e+00 -1.13767374e+00
-2.46154666e-01 -5.65422885e-02 3.16161364e-01 7.98595846e-02
-9.39935863e-01 1.13597408e-01 -8.36021721e-01 -5.60803056e-01
-1.89524367e-02 -7.79736280e-01 1.35889336e-01 5.98200440e-01
-2.56493986e-01 -6.22705340e-01 8.65370572e-01 -4.39511657e-01
1.50908792e+00 -2.34554052e+00 2.73710907e-01 3.88051122e-02
5.30400515e-01 3.88955563e-01 -3.17821980e-01 6.08354926e-01
1.13953039e-01 3.42050672e-01 -3.01114749e-02 -5.83859801e-01
5.68264306e-01 1.49712056e-01 -1.08175851e-01 3.81482393e-01
3.57985973e-01 1.40953016e+00 -8.29473257e-01 -6.31560147e-01
3.31550866e-01 6.49243891e-01 -7.40087271e-01 1.10914111e-01
-1.53990209e-01 -1.73424765e-01 -4.95363921e-01 7.91453660e-01
5.08583903e-01 -6.60045803e-01 -2.64999777e-01 -3.17735970e-01
7.96385854e-03 2.55608372e-02 -1.10857260e+00 1.81034899e+00
-4.22562987e-01 8.69565845e-01 -2.57808547e-02 -1.02762163e+00
8.09755623e-01 6.94692805e-02 3.59373480e-01 -7.16891766e-01
2.38791078e-01 2.50005752e-01 -3.72013181e-01 -5.95085919e-01
7.54152477e-01 2.01047167e-01 -7.14551434e-02 3.61515313e-01
4.40382928e-01 1.29461735e-02 1.27198890e-01 4.48987007e-01
1.23938441e+00 -2.44080164e-02 2.75475532e-01 -2.40454450e-01
3.66195410e-01 -3.98137897e-01 2.12691978e-01 9.37840164e-01
-3.69247943e-01 8.81548762e-01 5.05150795e-01 -2.52548486e-01
-8.36212873e-01 -9.41415548e-01 -2.04701483e-01 1.32206130e+00
2.74749607e-01 -8.44520330e-01 -4.61872309e-01 -6.96492374e-01
5.95056266e-02 2.42965668e-01 -7.83707619e-01 -2.11038277e-01
-1.70274317e-01 -5.02903223e-01 2.43510187e-01 4.61475819e-01
4.40068066e-01 -1.14198649e+00 -7.12054312e-01 -9.81891379e-02
4.27214503e-02 -1.09447670e+00 -7.49742568e-01 -4.86211367e-02
-3.35334897e-01 -8.26406479e-01 -1.00464201e+00 -7.29766726e-01
4.85498577e-01 5.43255270e-01 9.28086877e-01 3.05673420e-01
-4.33132917e-01 9.22229171e-01 -7.54124522e-01 -2.67002702e-01
2.03714296e-01 9.13696438e-02 -2.55015582e-01 1.68597236e-01
5.38550198e-01 -3.17784309e-01 -1.11502182e+00 1.70728132e-01
-1.13766742e+00 -1.02919258e-01 4.67669278e-01 1.06563842e+00
5.14675736e-01 -4.02469903e-01 1.50169268e-01 -3.82819325e-01
7.36112952e-01 -3.48030239e-01 -3.75172913e-01 4.77612376e-01
-5.93517363e-01 2.40343511e-01 4.89580661e-01 -3.57609421e-01
-3.44886810e-01 -3.80650997e-01 3.41273136e-02 -7.78000593e-01
1.80697322e-01 4.74691570e-01 -7.67825022e-02 -1.46977738e-01
5.20259321e-01 3.70197773e-01 -2.19887998e-02 -5.16049206e-01
5.10023952e-01 7.08590627e-01 1.50333211e-01 -4.32161182e-01
6.68011248e-01 5.54538786e-01 -2.47211218e-01 -7.82945514e-01
-5.46186566e-01 -6.68801785e-01 -3.25518936e-01 7.17957690e-03
7.41337657e-01 -8.16627383e-01 -7.58718193e-01 4.52705264e-01
-9.72516298e-01 -1.91449866e-01 -2.21827731e-01 3.82941335e-01
-3.98100615e-01 4.42774773e-01 -2.70179182e-01 -3.84365112e-01
-4.68891889e-01 -1.20799887e+00 1.44588614e+00 2.24773332e-01
-1.69349760e-01 -9.07873094e-01 -1.64564550e-01 9.55753028e-02
6.66195691e-01 -2.91167237e-02 7.63979256e-01 -6.46117747e-01
-6.49238825e-01 -2.67730236e-01 -6.10044718e-01 4.79303807e-01
3.50546576e-02 1.40959263e-01 -9.54218268e-01 -5.24641812e-01
-4.69247311e-01 -3.13482046e-01 1.24505389e+00 2.64728725e-01
1.48042881e+00 -3.22671145e-01 -2.91444689e-01 8.12828302e-01
1.56233597e+00 -1.76700532e-01 5.36802113e-01 5.74369967e-01
5.80522478e-01 3.57289821e-01 2.97552675e-01 3.95107001e-01
3.93395603e-01 8.21099639e-01 4.24139053e-01 -7.80393630e-02
-2.55200416e-01 -1.71103299e-01 4.52584505e-01 5.72691917e-01
2.06837475e-01 -4.32906419e-01 -7.78213203e-01 5.80898464e-01
-1.78091383e+00 -6.82046235e-01 5.59932470e-01 2.09074283e+00
6.68067098e-01 -4.63922322e-02 -1.14894882e-01 -2.91307233e-02
3.35074693e-01 6.00327253e-01 -3.60285252e-01 -3.63356531e-01
-1.64667428e-01 5.22320688e-01 3.88655096e-01 3.68909508e-01
-1.08305907e+00 1.05161035e+00 5.33822441e+00 1.06046247e+00
-1.36886513e+00 2.20372155e-01 5.34107536e-02 -4.05722290e-01
-4.70420390e-01 -6.62236810e-02 -7.09646106e-01 4.71957326e-01
5.62149763e-01 -2.00053155e-01 5.23914993e-01 6.11790359e-01
-2.80799046e-02 1.05641410e-01 -1.19120753e+00 1.29519343e+00
1.95583865e-01 -1.50814342e+00 3.22185099e-01 -2.01321080e-01
4.23626006e-01 1.72529295e-01 2.76638001e-01 3.85706961e-01
-2.26194337e-02 -1.05727887e+00 7.62570202e-01 5.04015028e-01
7.78205872e-01 -2.02721447e-01 5.95744014e-01 -1.35281473e-01
-1.23138702e+00 -2.60715485e-01 -3.04749519e-01 4.13094848e-01
7.56277516e-02 2.71894157e-01 -3.92354429e-01 5.23454905e-01
1.00897443e+00 1.02938581e+00 -8.45845401e-01 1.01098526e+00
-9.52394009e-02 2.33966902e-01 -3.67652953e-01 -2.91620821e-01
5.04406273e-01 1.25812411e-01 6.45835221e-01 1.45591497e+00
3.91586989e-01 -2.00803757e-01 4.13859822e-02 8.62194479e-01
-3.13537121e-01 3.12700510e-01 -4.22919124e-01 -4.02330756e-01
3.43164504e-01 1.27731466e+00 -5.46749651e-01 -2.96161830e-01
-5.65965414e-01 1.09127319e+00 3.82493883e-01 7.52435207e-01
-7.08756268e-01 -6.31086886e-01 8.80485177e-01 1.01134172e-02
7.45644927e-01 -1.98496759e-01 4.94793542e-02 -1.34228742e+00
3.31228644e-01 -7.66282380e-01 2.68938005e-01 -8.36236954e-01
-1.20889330e+00 5.89461386e-01 9.71590504e-02 -1.26563966e+00
1.38957903e-01 -9.86257792e-01 -3.53898615e-01 6.30921364e-01
-1.69637370e+00 -1.12430120e+00 -4.16732967e-01 7.50230014e-01
5.71552694e-01 -1.01822630e-01 7.74074912e-01 1.37831524e-01
-5.28636158e-01 9.11218166e-01 1.35955870e-01 1.46179244e-01
8.03109884e-01 -1.00930178e+00 1.24293966e-02 6.01004183e-01
4.11465347e-01 7.99677253e-01 5.22881448e-01 -3.38153029e-03
-1.64347231e+00 -8.29953671e-01 8.53114367e-01 -5.04097700e-01
9.85978603e-01 -6.94820642e-01 -9.92746115e-01 6.12387061e-01
4.79367793e-01 4.06196862e-01 5.75662613e-01 1.21242953e-02
-6.22845709e-01 -2.29401797e-01 -8.08526754e-01 8.02156091e-01
1.08905458e+00 -7.19347715e-01 -6.19836092e-01 2.75775939e-01
9.55760062e-01 -1.53800890e-01 -8.35447013e-01 3.31361055e-01
7.08063364e-01 -9.98427570e-01 1.19271588e+00 -4.63034600e-01
4.52521801e-01 -2.44152471e-01 -3.24085742e-01 -1.08978105e+00
-2.65344054e-01 -7.11008251e-01 -1.56309262e-01 9.76697028e-01
3.28727424e-01 -9.12990034e-01 2.77468264e-01 4.11087960e-01
-6.61857948e-02 -1.11175704e+00 -9.61321890e-01 -8.26054513e-01
1.78549923e-02 -3.32680792e-01 4.95101362e-01 8.02908361e-01
-1.45967737e-01 7.32648699e-03 -2.29358613e-01 6.92962259e-02
3.05514127e-01 1.98055193e-01 7.04294801e-01 -9.96813834e-01
-2.44971141e-01 -9.08896863e-01 -7.02138782e-01 -1.22124684e+00
1.50745034e-01 -1.05493271e+00 -1.43089950e-01 -1.68724310e+00
3.80501926e-01 -2.97140688e-01 -8.34825516e-01 8.04877162e-01
-3.88461947e-01 4.43398148e-01 5.32392621e-01 1.60930485e-01
-8.65437210e-01 7.51719296e-01 1.21541917e+00 -3.75542969e-01
-1.91895157e-01 -5.54887772e-01 -8.60991895e-01 4.00425881e-01
7.04977632e-01 -4.20907855e-01 -2.65874982e-01 -5.14150858e-01
3.86928678e-01 -5.02598286e-01 8.02344143e-01 -6.55572057e-01
3.18816453e-01 1.09453149e-01 8.13405067e-02 -1.84164241e-01
4.60864127e-01 -7.43316650e-01 -1.85797989e-01 1.49780095e-01
-3.35164070e-01 -1.14814565e-02 2.92467892e-01 5.20140052e-01
-3.75239909e-01 -2.35641867e-01 3.19083393e-01 1.58683702e-01
-9.75851774e-01 5.33610642e-01 -3.39768976e-02 -2.27077920e-02
7.91725993e-01 -3.59934598e-01 -4.53639895e-01 -2.89104283e-01
-5.77718735e-01 5.09873390e-01 4.41064447e-01 8.05686593e-01
7.21167207e-01 -1.31663537e+00 -5.39075613e-01 1.40728593e-01
6.33725822e-01 -1.68420181e-01 2.56332725e-01 8.76919925e-01
-2.86130935e-01 4.54193860e-01 1.98166296e-02 -6.79962635e-01
-1.31823063e+00 5.63579679e-01 1.90851271e-01 -2.15859115e-01
-7.29839027e-01 9.62269247e-01 2.23044127e-01 -1.05486281e-01
2.43678138e-01 -4.16022390e-01 -1.17151976e-01 2.09418133e-01
5.34903884e-01 1.38232008e-01 -6.11415505e-02 -4.06082571e-01
-4.51349884e-01 8.15989375e-01 -1.42085046e-01 1.17713340e-01
1.34192336e+00 -2.43307590e-01 -1.15603335e-01 3.18330348e-01
1.63868332e+00 -2.03841120e-01 -1.31674528e+00 -5.41827738e-01
-2.49581605e-01 -6.71714365e-01 1.97012737e-01 -5.35041451e-01
-1.00190568e+00 1.04647136e+00 6.36606395e-01 2.00157553e-01
1.17168832e+00 2.94435054e-01 7.32499540e-01 5.04743576e-01
2.59318680e-01 -1.03251088e+00 3.17148030e-01 2.67680019e-01
1.13501012e+00 -1.38505328e+00 7.37820640e-02 -5.71253859e-02
-5.52188516e-01 1.11959815e+00 3.12202901e-01 -2.23628223e-01
8.39800239e-01 -1.23065487e-01 -5.45369424e-02 -4.23443735e-01
-7.49440670e-01 -3.29890043e-01 5.70184827e-01 2.02954277e-01
2.31585562e-01 -1.45341130e-03 -4.72345054e-01 5.11392713e-01
-1.37842402e-01 -2.89127938e-02 9.82110649e-02 9.60660160e-01
-3.20178866e-01 -1.02593696e+00 -7.27932975e-02 3.72184604e-01
-3.16737294e-01 -3.07797432e-01 -2.84665525e-01 9.45471644e-01
4.84108143e-02 6.30174756e-01 5.20250909e-02 -3.33185017e-01
3.25399697e-01 2.62742251e-01 3.56387258e-01 -4.82578158e-01
-5.74762106e-01 -1.41176030e-01 -4.31844741e-01 -8.54235590e-01
-5.11086345e-01 -3.22101057e-01 -1.06848288e+00 -1.31632492e-01
-2.72673935e-01 -1.67750388e-01 5.93634903e-01 6.76396132e-01
7.29166985e-01 3.69797945e-01 5.57527006e-01 -8.98524880e-01
-5.12804925e-01 -7.88137317e-01 -4.50339139e-01 5.22447228e-01
5.21246433e-01 -9.14416254e-01 -5.39540768e-01 -9.51766670e-02] | [10.627674102783203, 0.9711435437202454] |
80b9a5de-43f6-4570-b2b6-4998c23e9ddc | balancing-utility-and-fairness-in-submodular | 2211.00980 | null | https://arxiv.org/abs/2211.00980v4 | https://arxiv.org/pdf/2211.00980v4.pdf | Balancing Utility and Fairness in Submodular Maximization (Technical Report) | Submodular function maximization is a fundamental combinatorial optimization problem with plenty of applications -- including data summarization, influence maximization, and recommendation. In many of these problems, the goal is to find a solution that maximizes the average utility over all users, for each of whom the utility is defined by a monotone submodular function. However, when the population of users is composed of several demographic groups, another critical problem is whether the utility is fairly distributed across different groups. Although the \emph{utility} and \emph{fairness} objectives are both desirable, they might contradict each other, and, to the best of our knowledge, little attention has been paid to optimizing them jointly. To fill this gap, we propose a new problem called \emph{Bicriteria Submodular Maximization} (BSM) to balance utility and fairness. Specifically, it requires finding a fixed-size solution to maximize the utility function, subject to the value of the fairness function not being below a threshold. Since BSM is inapproximable within any constant factor, we focus on designing efficient instance-dependent approximation schemes. Our algorithmic proposal comprises two methods, with different approximation factors, obtained by converting a BSM instance into other submodular optimization problem instances. Using real-world and synthetic datasets, we showcase applications of our proposed methods in three submodular maximization problems: maximum coverage, influence maximization, and facility location. | ['Ying Wang', 'Francesco Bonchi', 'Yuchen Li', 'Yanhao Wang'] | 2022-11-02 | null | null | null | null | ['data-summarization'] | ['miscellaneous'] | [ 1.08670309e-01 5.12322068e-01 -6.77378416e-01 -3.19012552e-01
-5.91147602e-01 -6.43030882e-01 -1.19582511e-01 3.41389239e-01
-6.90131858e-02 1.19901109e+00 2.19623789e-01 -1.94844440e-01
-6.61330938e-01 -9.68574762e-01 -5.87258279e-01 -7.59027481e-01
-1.14692137e-01 6.75066054e-01 -2.45885998e-01 -2.02234969e-01
1.96060672e-01 1.30051821e-01 -1.02265620e+00 -1.97985888e-01
1.30905032e+00 1.05142474e+00 2.67820626e-01 6.44240901e-02
-1.83435231e-02 3.79348278e-01 -5.66787362e-01 -2.74460644e-01
4.65621263e-01 -3.87209237e-01 -7.63514638e-01 4.17414725e-01
3.58044580e-02 -3.27941835e-01 -1.97784066e-01 1.12243354e+00
3.84887964e-01 1.32000908e-01 5.69991052e-01 -1.72604990e+00
-7.64667094e-01 9.21345413e-01 -1.00911582e+00 5.32590672e-02
2.62271523e-01 -1.85584441e-01 1.37527704e+00 -2.41118550e-01
4.15042758e-01 1.08840179e+00 1.39160259e-02 3.26495409e-01
-1.09882450e+00 -3.48394066e-01 4.54404116e-01 -1.38736263e-01
-1.24557853e+00 -1.28263459e-01 5.64022005e-01 -9.23171267e-02
4.86903757e-01 8.13687325e-01 5.28436482e-01 6.67366460e-02
-1.52570948e-01 1.07434165e+00 7.40895867e-01 -5.89375310e-02
3.88569713e-01 4.07127589e-01 1.23471841e-01 1.75257251e-01
9.09489274e-01 -5.26628852e-01 -2.55177766e-01 -5.31140268e-01
2.86794335e-01 1.36905789e-01 -8.22853804e-01 -4.56560701e-01
-8.09475183e-01 7.69449890e-01 1.07938550e-01 -1.31502198e-02
-5.40127337e-01 1.28773674e-01 -1.22722443e-02 4.37326759e-01
3.73247743e-01 4.31431502e-01 -4.48515892e-01 2.40162253e-01
-1.11870599e+00 4.63366121e-01 1.04951739e+00 1.22280669e+00
6.98747873e-01 -1.15011655e-01 -3.42345595e-01 6.12441838e-01
3.12550545e-01 4.96600807e-01 -1.71947286e-01 -8.97336781e-01
8.65354359e-01 8.11013877e-01 4.38257366e-01 -1.28216803e+00
-2.37725645e-01 -4.94654238e-01 -7.72622049e-01 -2.85241187e-01
4.35306668e-01 -5.23803055e-01 -3.45874578e-01 1.93702877e+00
4.45089281e-01 -3.40675265e-01 -2.18426466e-01 1.19644320e+00
5.76867461e-01 9.84664679e-01 -2.81719714e-01 -1.13808143e+00
8.48654449e-01 -6.18475020e-01 -7.94634521e-01 -3.11259311e-02
4.56400245e-01 -4.68084902e-01 4.84817028e-01 3.48962069e-01
-1.41454840e+00 4.57130969e-01 -9.65673923e-01 2.92362005e-01
2.03606844e-01 -3.99903536e-01 6.47993922e-01 9.65465665e-01
-8.78892541e-01 -2.68120039e-03 -4.33887402e-03 -2.76326418e-01
5.04942179e-01 5.78142643e-01 -5.51147722e-02 -3.12184900e-01
-8.48423600e-01 4.96897638e-01 2.63917685e-01 -1.21942408e-01
-7.61070251e-01 -7.69809484e-01 -4.96902406e-01 4.49980706e-01
9.15304780e-01 -8.35053444e-01 9.47345555e-01 -7.01408327e-01
-8.80936980e-01 6.85439944e-01 -2.81022817e-01 -2.19193608e-01
6.05463982e-01 3.68231535e-01 5.35165630e-02 -1.26150042e-01
1.09273233e-01 2.76200790e-02 3.63134176e-01 -1.50645006e+00
-8.22563887e-01 -7.38790214e-01 5.84582210e-01 5.19423306e-01
-7.39692450e-01 -2.72042714e-02 -4.91913170e-01 -3.06350827e-01
6.87216073e-02 -4.89160478e-01 -6.37050211e-01 -2.68692762e-01
-5.94109118e-01 -2.66078323e-01 4.87188131e-01 -5.28717041e-01
1.71790135e+00 -1.81085062e+00 3.15368533e-01 5.79359770e-01
2.61431783e-01 -2.65429795e-01 -6.08759895e-02 4.11792547e-01
5.08932829e-01 2.56035835e-01 -4.69720900e-01 -8.15916955e-02
1.66909501e-01 3.59572798e-01 5.60149364e-02 7.73517549e-01
-5.06467402e-01 7.29044795e-01 -7.35034645e-01 -3.19150269e-01
-2.63291866e-01 -3.71470958e-01 -7.21591830e-01 -3.76557671e-02
-4.58604634e-01 -5.83448000e-02 -7.10036039e-01 9.78084385e-01
1.17861307e+00 -2.40275562e-01 5.32004833e-01 3.43444079e-01
-3.67709473e-02 -2.80640274e-01 -1.60320425e+00 1.06872451e+00
-2.67146617e-01 6.07594475e-02 6.16308153e-01 -1.38233614e+00
7.71867454e-01 8.91083032e-02 1.09135175e+00 -3.84972006e-01
3.11486512e-01 2.94414967e-01 -2.72010416e-01 -2.55988955e-01
6.59612894e-01 -1.95696518e-01 -1.58254728e-01 9.22693551e-01
-4.07905817e-01 1.76489413e-01 4.68156844e-01 4.50710237e-01
7.85460889e-01 -8.28883767e-01 6.20313466e-01 -5.29423952e-01
4.64444816e-01 -2.34829932e-02 7.45519459e-01 6.01685107e-01
-3.88380187e-03 6.43733621e-01 1.05704880e+00 1.36555936e-02
-8.21452320e-01 -8.90385449e-01 4.46028747e-02 9.59120274e-01
6.88997149e-01 -1.06014647e-01 -6.57332063e-01 -6.18222535e-01
4.75285053e-01 6.98611379e-01 -4.62516755e-01 1.65570050e-01
-1.48160592e-01 -1.17371082e+00 -1.56752944e-01 -1.34938821e-01
2.13375181e-01 -5.23715317e-01 -1.97182730e-01 2.35915750e-01
-2.97379225e-01 -8.37148666e-01 -1.00453615e+00 -2.16749951e-01
-8.37704599e-01 -1.08603525e+00 -8.74226928e-01 -6.39267743e-01
1.08387232e+00 7.45034873e-01 9.85543251e-01 1.53013095e-01
3.69454473e-02 3.90550017e-01 -2.15324014e-01 -4.19492126e-01
1.76429644e-01 1.22775942e-01 4.50773351e-02 1.73407540e-01
2.91553978e-03 -5.56143820e-01 -5.94483256e-01 5.61187029e-01
-1.14728975e+00 -3.75098407e-01 3.37614149e-01 3.37841600e-01
7.88041711e-01 3.63320827e-01 8.98382962e-01 -9.25680697e-01
9.69569862e-01 -1.13140786e+00 -5.92938006e-01 4.84239310e-01
-6.71901643e-01 -1.89492747e-01 6.36348546e-01 -1.31715804e-01
-8.53126049e-01 -2.82956004e-01 3.53054911e-01 1.48594931e-01
5.80912292e-01 7.50793636e-01 -8.48485172e-01 1.82361826e-01
1.75684616e-01 3.03388655e-01 4.35561500e-02 -1.96795732e-01
2.92885214e-01 8.29615176e-01 9.39957276e-02 -6.47703826e-01
5.38513601e-01 5.70456564e-01 1.65625185e-01 -7.68297374e-01
-8.14007580e-01 -5.56692004e-01 1.70275420e-01 -2.63693720e-01
1.65886223e-01 -5.86827219e-01 -1.04420137e+00 4.23264466e-02
-1.00675380e+00 2.21420810e-01 -3.52993131e-01 1.22270882e-01
-5.57630718e-01 5.06694674e-01 2.29951128e-01 -1.17708135e+00
-4.87456739e-01 -8.60946715e-01 3.74988109e-01 3.62165481e-01
5.90381622e-02 -7.37559736e-01 -1.19792953e-01 6.61735296e-01
2.63443977e-01 5.08957505e-01 7.72214830e-01 -3.40438843e-01
-9.13548827e-01 -1.96186274e-01 -2.68748373e-01 7.90428147e-02
2.50166416e-01 -4.29071873e-01 -2.44144529e-01 -6.00044608e-01
-6.65630624e-02 -2.53498536e-02 6.79155409e-01 7.61795402e-01
1.37864614e+00 -9.91160154e-01 -3.03299725e-01 5.94316244e-01
1.46258998e+00 2.00164840e-01 3.88076216e-01 1.80319116e-01
2.38738641e-01 6.62324667e-01 7.43545592e-01 1.21497762e+00
7.03827500e-01 5.23136795e-01 8.65637839e-01 1.77895337e-01
6.47407353e-01 -1.14431217e-01 8.24609846e-02 2.74581641e-01
-1.64993674e-01 -9.44407165e-01 -3.59271824e-01 8.37430060e-01
-2.24324083e+00 -1.02850282e+00 -2.31530204e-01 2.39429927e+00
5.90305984e-01 -2.72937566e-01 4.90040988e-01 1.86976269e-02
7.96282828e-01 2.48735353e-01 -7.82623172e-01 -5.70887923e-01
-4.40512687e-01 -4.11864042e-01 7.56977916e-01 5.46218395e-01
-5.98730862e-01 2.69454956e-01 5.73581839e+00 8.35002780e-01
-5.70947528e-01 4.78265099e-02 1.01269066e+00 -7.17451274e-01
-1.12308216e+00 1.59197804e-02 -6.56522632e-01 5.70516169e-01
1.74844906e-01 -1.05807674e+00 7.24656582e-01 7.66061842e-01
6.56816781e-01 -3.63178134e-01 -8.50113988e-01 7.79516160e-01
8.01123381e-02 -1.23747957e+00 6.07873909e-02 4.08684760e-01
1.37113309e+00 -5.60923994e-01 7.38536045e-02 -7.94949755e-02
1.15622662e-01 -1.04060292e+00 4.43384230e-01 1.49626657e-01
6.39344931e-01 -1.19483054e+00 5.43607652e-01 6.24544382e-01
-9.29484129e-01 -2.94921339e-01 -5.46822131e-01 -3.70438285e-02
6.79600120e-01 1.13098049e+00 -3.51316571e-01 7.31398582e-01
4.64711398e-01 1.48589522e-01 3.38900387e-01 1.30956924e+00
9.72868726e-02 2.67850697e-01 -5.83214998e-01 -2.87121326e-01
1.93744689e-01 -3.63807559e-01 7.71103382e-01 8.86646271e-01
4.19102937e-01 5.05867422e-01 2.99878925e-01 8.84536266e-01
-4.63259786e-01 5.83784401e-01 -4.58248168e-01 2.86216177e-02
5.68412781e-01 1.26571178e+00 -4.21829790e-01 1.30386040e-01
-3.26652735e-01 5.15188992e-01 6.48623630e-02 3.88124675e-01
-8.94279778e-01 -1.52583435e-01 9.74181056e-01 4.81068134e-01
2.33334824e-02 -2.86381375e-02 -9.41417694e-01 -8.73461545e-01
3.21581334e-01 -5.64622402e-01 7.61912405e-01 -4.41238247e-02
-9.97422457e-01 -6.66515157e-02 5.87511472e-02 -7.70546079e-01
3.07318926e-01 -8.06807578e-02 -8.10192645e-01 8.27340066e-01
-1.55026329e+00 -6.32150471e-01 -2.22183332e-01 4.84720021e-01
1.24394260e-01 -8.26945305e-02 -2.00440921e-02 4.20370132e-01
-5.20433784e-01 6.78221583e-01 3.63029957e-01 -7.18147814e-01
1.23491384e-01 -1.11920059e+00 -5.41724265e-01 7.52080202e-01
-4.82782364e-01 3.96522760e-01 8.85760009e-01 -4.30417687e-01
-1.55813754e+00 -1.01730967e+00 1.02722025e+00 6.09513633e-02
2.48380601e-01 1.31772412e-02 -4.87132281e-01 3.18476319e-01
1.14737339e-01 -2.12521002e-01 7.19306052e-01 4.14426439e-03
4.04470980e-01 -3.16893160e-01 -1.66891408e+00 5.95240116e-01
1.29422438e+00 4.11669850e-01 -1.31403180e-02 4.90011007e-01
7.18734086e-01 -1.23836830e-01 -7.23459899e-01 4.95519489e-01
2.74794430e-01 -7.21718609e-01 6.86785698e-01 -6.65280163e-01
4.89162683e-01 -3.15143794e-01 -4.62131470e-01 -1.26796889e+00
-2.63857454e-01 -1.04205239e+00 -4.50848967e-01 1.17145383e+00
5.13121128e-01 -6.57838702e-01 1.13658547e+00 8.61443996e-01
2.01260045e-01 -1.25062585e+00 -9.68098640e-01 -7.99108207e-01
-3.38026546e-02 -6.02952167e-02 9.32235479e-01 9.92751718e-01
3.11946124e-01 1.61946759e-01 -7.11112380e-01 3.19295108e-01
7.68855810e-01 5.54588556e-01 7.44508028e-01 -1.00128317e+00
-4.08821642e-01 -4.10294890e-01 8.44823793e-02 -1.33041286e+00
-1.80166528e-01 -9.75077510e-01 -2.82525182e-01 -2.09419131e+00
7.11288631e-01 -6.02288902e-01 -1.54560998e-01 3.63517880e-01
-6.85078874e-02 -2.72941977e-01 3.95115644e-01 2.14073956e-02
-8.66579652e-01 6.32008016e-01 1.55325007e+00 -2.81047553e-01
-5.11672378e-01 5.34790158e-01 -1.62035966e+00 2.63799399e-01
7.44240582e-01 -4.44933027e-01 -5.91453910e-01 -3.98433298e-01
3.75010282e-01 5.27661622e-01 -2.77508587e-01 -2.01257601e-01
2.37463042e-01 -1.00588334e+00 -3.96153361e-01 -6.66671932e-01
4.42694454e-03 -9.62999463e-01 3.31823856e-01 3.30465943e-01
-5.56461979e-04 -1.87156096e-01 -2.59573758e-01 5.80994189e-01
-6.93225339e-02 -4.44003701e-01 7.40003824e-01 -5.29026054e-02
-2.66214967e-01 7.55313635e-01 -1.67606980e-01 3.52232456e-01
1.56901896e+00 -4.02129441e-01 -4.78342861e-01 -8.47335994e-01
-2.86726594e-01 1.17150831e+00 3.64609867e-01 1.18661232e-01
6.01353407e-01 -1.19170535e+00 -9.89590049e-01 -4.03342187e-01
-3.28400582e-02 2.64462531e-01 4.30004269e-01 1.00874066e+00
-2.45260209e-01 6.42023861e-01 1.47596583e-01 -2.12080687e-01
-1.12157154e+00 5.26307523e-01 1.96022198e-01 -4.38325793e-01
2.26632953e-01 6.33349836e-01 2.59723157e-01 -4.12948698e-01
2.58316308e-01 1.81974962e-01 -6.76400736e-02 3.29282671e-01
2.19115153e-01 8.70637059e-01 -2.98948139e-01 -2.75934070e-01
-2.40318850e-01 1.24265142e-01 1.07073158e-01 1.10340603e-01
1.50821209e+00 -3.77328813e-01 -5.71835101e-01 -3.38865995e-01
1.01527596e+00 9.54818279e-02 -7.15888619e-01 -2.90965050e-01
-1.90940559e-01 -8.16729307e-01 -1.81613714e-02 -6.33859694e-01
-1.34352887e+00 -1.65278278e-02 -2.27005273e-01 6.09239221e-01
1.43035436e+00 1.00041822e-01 9.36239004e-01 2.01440230e-01
6.95502043e-01 -1.27233195e+00 -2.62111783e-01 2.08771467e-01
8.46867502e-01 -1.13195050e+00 3.49776804e-01 -6.32811368e-01
-6.26643836e-01 7.05527067e-01 6.84648037e-01 7.20203668e-02
6.95641816e-01 4.36486378e-02 -6.39363945e-01 4.23966795e-02
-6.67370677e-01 -2.19127148e-01 1.96019143e-01 1.77305833e-01
8.00540820e-02 4.58421946e-01 -1.22751939e+00 9.93233144e-01
-1.42322577e-04 -2.39020661e-02 6.77152038e-01 7.37134695e-01
-9.56582427e-01 -9.77365196e-01 -6.07997298e-01 9.01646793e-01
-4.71239477e-01 2.16584101e-01 -4.09891635e-01 3.87838930e-01
1.23990417e-01 1.23886395e+00 -1.45556256e-01 7.28466287e-02
3.61071825e-01 -8.11688006e-01 3.76937032e-01 -4.93673474e-01
-1.62168548e-01 -2.13230848e-01 9.15442482e-02 -4.26368192e-02
-1.99830443e-01 -6.69094026e-01 -1.12365055e+00 -7.37954140e-01
-4.57754642e-01 2.78943688e-01 3.54541779e-01 7.97818840e-01
-5.26606366e-02 1.74131930e-01 1.15205622e+00 -1.18477225e-01
-8.99157941e-01 -5.26673496e-01 -1.10457253e+00 9.14352760e-02
3.25566381e-02 -2.52142310e-01 -2.60264039e-01 -6.97469175e-01] | [6.565415382385254, 4.938504695892334] |
d2cd4ec2-26e1-4984-9320-b11256ee6854 | yolactedge-real-time-instance-segmentation-on | 2012.12259 | null | https://arxiv.org/abs/2012.12259v2 | https://arxiv.org/pdf/2012.12259v2.pdf | YolactEdge: Real-time Instance Segmentation on the Edge | We propose YolactEdge, the first competitive instance segmentation approach that runs on small edge devices at real-time speeds. Specifically, YolactEdge runs at up to 30.8 FPS on a Jetson AGX Xavier (and 172.7 FPS on an RTX 2080 Ti) with a ResNet-101 backbone on 550x550 resolution images. To achieve this, we make two improvements to the state-of-the-art image-based real-time method YOLACT: (1) applying TensorRT optimization while carefully trading off speed and accuracy, and (2) a novel feature warping module to exploit temporal redundancy in videos. Experiments on the YouTube VIS and MS COCO datasets demonstrate that YolactEdge produces a 3-5x speed up over existing real-time methods while producing competitive mask and box detection accuracy. We also conduct ablation studies to dissect our design choices and modules. Code and models are available at https://github.com/haotian-liu/yolact_edge. | ['Yong Jae Lee', 'Fanyi Xiao', 'Rafael A. Rivera Soto', 'Haotian Liu'] | 2020-12-22 | null | null | null | null | ['real-time-instance-segmentation'] | ['computer-vision'] | [-2.35732824e-01 -2.79024243e-01 -1.77676797e-01 -2.38467723e-01
-5.73431671e-01 -5.75356543e-01 2.70475537e-01 -3.49678993e-01
-5.54860294e-01 2.23930627e-01 -2.24820971e-01 -6.13189578e-01
3.08409154e-01 -5.45662940e-01 -7.19143033e-01 -1.60068318e-01
-3.83347631e-01 1.82567671e-01 6.68678463e-01 1.16551742e-01
1.45297438e-01 4.48667765e-01 -1.38609934e+00 2.81893700e-01
5.68309367e-01 1.37546897e+00 -1.03618287e-01 1.12137973e+00
1.88789934e-01 8.95525694e-01 -2.92126596e-01 -2.33271495e-01
9.03349280e-01 -9.44578126e-02 -6.76670849e-01 9.62081030e-02
7.10989237e-01 -6.35282874e-01 -5.04088938e-01 7.17525065e-01
4.23591375e-01 2.30711680e-02 -1.10683860e-02 -1.27232385e+00
-3.52193356e-01 4.83255088e-01 -8.35288405e-01 9.14392233e-01
-1.40833482e-02 5.18306196e-01 8.24288011e-01 -8.42020094e-01
6.34112835e-01 8.89053702e-01 5.04568994e-01 4.19732571e-01
-9.96811926e-01 -7.96864152e-01 1.42505011e-02 1.64824009e-01
-1.31275654e+00 -6.80351794e-01 3.04947674e-01 -2.51657695e-01
1.15819550e+00 2.73283839e-01 8.16951811e-01 7.95360982e-01
3.77735585e-01 7.99486876e-01 1.18866432e+00 1.42082805e-02
1.87200218e-01 -2.73824960e-01 -3.26870531e-02 1.07175255e+00
1.27717212e-01 2.24125490e-01 -7.28086591e-01 8.46250653e-02
1.06890357e+00 -2.40681916e-01 -2.75618166e-01 -1.00691482e-01
-1.14503503e+00 5.21384001e-01 3.98681939e-01 4.50668111e-02
-1.57261282e-01 6.54342532e-01 6.90866351e-01 2.14238048e-01
7.31457829e-01 3.95464391e-01 -5.26517570e-01 -5.51107526e-01
-1.18782151e+00 1.11951893e-02 6.65872097e-01 1.20824552e+00
4.05276477e-01 4.26765651e-01 4.84256931e-02 3.08696151e-01
2.42219791e-01 2.35988051e-01 3.97488236e-01 -1.30564904e+00
3.87034744e-01 1.59513995e-01 -1.04762234e-01 -7.25739360e-01
-3.35141420e-01 -4.22954619e-01 -2.78263956e-01 3.60853463e-01
3.65986943e-01 -3.85247082e-01 -1.13695741e+00 9.76825356e-01
5.07192433e-01 6.34349525e-01 -4.29436535e-01 1.31773293e+00
9.90246296e-01 7.08736122e-01 -1.31518409e-01 2.04336196e-01
1.52670538e+00 -1.62212038e+00 -2.17183337e-01 -1.55955687e-01
6.26568437e-01 -8.18266571e-01 1.07595336e+00 5.64049780e-01
-1.24344397e+00 -5.52025378e-01 -1.18029273e+00 -3.21358830e-01
-9.47515592e-02 2.27952048e-01 9.16917086e-01 8.61137450e-01
-1.06797993e+00 7.68512189e-01 -1.46643543e+00 -7.30110854e-02
5.95766306e-01 4.83712703e-01 -5.56413569e-02 1.72123328e-01
-5.16420126e-01 3.24099302e-01 9.77764558e-03 9.87608209e-02
-8.28698039e-01 -1.10942733e+00 -7.09999979e-01 4.64221723e-02
8.23650479e-01 -5.16479015e-01 1.42732584e+00 -8.53026330e-01
-1.73951781e+00 6.37840152e-01 2.80850325e-02 -7.47029722e-01
7.67421126e-01 -6.14281178e-01 -2.81158388e-01 5.63139498e-01
6.98972866e-02 8.79494131e-01 7.85806417e-01 -7.58292198e-01
-6.86907232e-01 -4.17569354e-02 -1.66250728e-02 1.59057304e-01
2.57000066e-02 3.04859072e-01 -8.70408714e-01 -6.14407539e-01
-2.28406012e-01 -9.87587094e-01 -2.90798903e-01 2.70226955e-01
-4.11362380e-01 1.76664397e-01 9.37218726e-01 -6.10114455e-01
8.73425364e-01 -2.11469030e+00 -3.89697552e-01 4.15678285e-02
5.66651464e-01 5.32229960e-01 -9.38197672e-02 -1.04089845e-02
1.05296820e-01 1.63029909e-01 6.36199117e-02 -5.65876245e-01
-3.52897137e-01 -5.13451546e-02 1.40677374e-02 6.75275207e-01
-1.23252176e-01 9.87707019e-01 -6.95577204e-01 -4.19077992e-01
5.23205698e-01 5.62872469e-01 -5.61167538e-01 -1.46987975e-01
7.45007545e-02 3.92330587e-01 -3.38188142e-01 1.04949057e+00
6.39062524e-01 -3.41685355e-01 -9.92124453e-02 -3.21378827e-01
-5.19687176e-01 2.08094090e-01 -1.07142913e+00 1.85059583e+00
-2.80269682e-01 1.00667715e+00 4.34631199e-01 -5.68024993e-01
4.70945746e-01 1.48617506e-01 7.38707483e-01 -7.35200942e-01
6.49090886e-01 1.68183714e-01 -7.58771673e-02 -3.13089132e-01
6.88963711e-01 3.06903601e-01 3.48353893e-01 2.82372922e-01
-5.13144284e-02 7.92690087e-03 3.37265640e-01 2.28897378e-01
1.34988022e+00 2.89120048e-01 -1.41846135e-01 -4.45110857e-01
-8.91116858e-02 1.36510730e-01 7.28813589e-01 6.65107906e-01
-5.34713387e-01 7.83800840e-01 3.63123477e-01 -4.70549613e-01
-1.09364629e+00 -9.59477544e-01 -1.27170995e-01 8.84080350e-01
4.06676918e-01 -8.24427962e-01 -7.26330042e-01 -6.97764099e-01
-7.11529031e-02 3.87488425e-01 -4.65788513e-01 3.38228881e-01
-5.51220655e-01 -5.75237334e-01 5.04007638e-01 8.52121353e-01
8.98476422e-01 -6.87411785e-01 -1.28992748e+00 2.60089576e-01
3.35441828e-01 -1.59000182e+00 -8.47148478e-01 -1.00306548e-01
-8.40148926e-01 -1.09515905e+00 -3.74735206e-01 -3.22003007e-01
4.15535748e-01 5.05930007e-01 1.20241928e+00 1.19792230e-01
-7.67162263e-01 2.90995300e-01 -5.21975398e-01 -1.13510817e-01
2.09788650e-01 7.24147856e-02 -6.64435029e-02 -1.94852620e-01
9.26729813e-02 -4.13338035e-01 -1.09929931e+00 4.73343521e-01
-6.59369171e-01 3.56688023e-01 2.37751812e-01 4.58671689e-01
6.35820985e-01 -1.38265297e-01 2.04003938e-02 -7.40334570e-01
1.81419343e-01 -2.05290541e-01 -1.00899243e+00 -2.15095207e-01
-6.78845048e-01 -3.15359116e-01 5.95209897e-01 -2.49505013e-01
-8.12898695e-01 1.66878834e-01 -1.53001159e-01 -7.87807703e-01
1.87811494e-01 9.43648517e-02 4.48952705e-01 -5.15693069e-01
4.57160443e-01 -1.33657768e-01 5.38484976e-02 -3.07740599e-01
4.28157389e-01 4.95223016e-01 6.04798794e-01 -5.45718014e-01
4.11841929e-01 8.67335260e-01 -1.90683663e-01 -8.04056168e-01
-5.44502556e-01 -6.86844647e-01 -3.99553508e-01 -3.45704764e-01
9.98302400e-01 -1.01802480e+00 -8.79257977e-01 4.48136866e-01
-6.71216667e-01 -8.30430686e-01 -3.36955070e-01 4.95426476e-01
-4.63531375e-01 1.23221204e-01 -9.60812569e-01 -3.95327121e-01
-5.95180511e-01 -1.31764555e+00 1.13359261e+00 6.06480956e-01
8.67688581e-02 -6.01474941e-01 -2.89817572e-01 4.87210780e-01
5.74318349e-01 2.01506972e-01 -1.02893069e-01 -1.44100919e-01
-8.72883260e-01 1.11961648e-01 -7.04810619e-01 3.13341588e-01
-1.47663280e-01 4.41278905e-01 -7.52056122e-01 -4.53053534e-01
-1.30056858e-01 -1.29666507e-01 8.41471612e-01 6.57758713e-01
1.32759750e+00 8.40915591e-02 -3.47017437e-01 1.12467027e+00
1.53537202e+00 3.88224185e-01 5.07848144e-01 3.84190947e-01
8.27950418e-01 -5.06355986e-02 6.11902773e-01 3.91567081e-01
3.03182125e-01 6.91601396e-01 4.00419444e-01 -3.40928674e-01
-3.49600136e-01 7.99846277e-03 2.14156374e-01 6.08784258e-01
-1.88776776e-01 -1.96554452e-01 -8.67934346e-01 5.12342095e-01
-1.80221403e+00 -5.97649634e-01 -1.78246856e-01 2.03333282e+00
3.31245482e-01 3.57357413e-01 3.77917379e-01 -3.50335091e-01
5.70191562e-01 2.93521643e-01 -7.73213208e-01 -4.03830111e-01
3.41048986e-01 4.37713951e-01 1.09205568e+00 3.72055352e-01
-1.15511096e+00 1.21335363e+00 5.86351156e+00 8.09975266e-01
-1.47405100e+00 3.47884208e-01 1.00675356e+00 -7.20445216e-01
3.13220084e-01 1.44512892e-01 -7.79197216e-01 5.43840587e-01
9.54265535e-01 -1.92987487e-01 5.66492200e-01 1.00762188e+00
2.40272447e-01 -3.02791804e-01 -6.96700692e-01 1.08324504e+00
-1.14108464e-02 -1.60813940e+00 -5.40900707e-01 2.16389090e-01
5.10630429e-01 6.38476193e-01 -1.70968622e-01 1.20875411e-01
2.85998046e-01 -9.67597723e-01 8.30700338e-01 5.20420484e-02
9.42577720e-01 -6.64793909e-01 3.38963777e-01 -1.25210121e-01
-1.53148472e+00 2.39549324e-01 -3.61328334e-01 -2.80848090e-02
4.98726010e-01 5.34114182e-01 -6.31390333e-01 3.98545682e-01
1.13705981e+00 7.58669376e-01 -5.38318396e-01 1.18733644e+00
-1.15037262e-01 9.48872864e-01 -7.89865017e-01 2.40947947e-01
4.10757452e-01 -1.12238221e-01 7.15762854e-01 1.25188935e+00
2.34676063e-01 2.41815984e-01 1.61044851e-01 8.28729272e-01
-1.81423306e-01 -2.17765078e-01 -2.82472223e-01 4.75042313e-02
2.81253308e-01 1.60224831e+00 -1.37030232e+00 -4.01833892e-01
-5.10376036e-01 1.10352921e+00 -8.16841125e-02 1.95210591e-01
-1.31769013e+00 -3.42710286e-01 8.36008608e-01 2.92451411e-01
6.35324657e-01 -7.73958027e-01 -3.93740624e-01 -1.32745278e+00
2.30768602e-02 -5.39667666e-01 1.36990935e-01 -8.30718279e-01
-8.72062087e-01 8.14310908e-01 -1.11049540e-01 -1.13971364e+00
3.02639157e-01 -7.79847026e-01 -5.33735991e-01 3.12496960e-01
-1.46323729e+00 -8.81631374e-01 -5.21727741e-01 2.80019969e-01
9.27628458e-01 8.19217507e-03 2.18465656e-01 4.72005635e-01
-8.69157970e-01 5.58700979e-01 -5.10035753e-02 1.88337177e-01
3.72517914e-01 -1.21630406e+00 1.03643441e+00 1.07938814e+00
1.75304875e-01 3.82501423e-01 5.60586095e-01 -7.04723001e-01
-1.65467334e+00 -1.10431957e+00 8.07561800e-02 -2.18369335e-01
6.91056311e-01 -4.74199265e-01 -6.44426942e-01 8.73356581e-01
3.76460344e-01 6.84419155e-01 3.51866007e-01 -2.66783595e-01
-1.33060753e-01 2.96708625e-02 -1.08396053e+00 7.44907975e-01
1.30910718e+00 -5.47168255e-02 9.66912657e-02 3.47449303e-01
6.51433051e-01 -1.08460486e+00 -8.68723333e-01 3.88997316e-01
6.82667315e-01 -1.29982018e+00 9.10367727e-01 -3.34536374e-01
4.43313628e-01 -3.84448588e-01 1.07596599e-01 -1.04722095e+00
-1.37066871e-01 -1.03514421e+00 -1.40580267e-01 8.45077574e-01
2.37883389e-01 -7.20887125e-01 1.12354243e+00 6.69105589e-01
-3.42076302e-01 -1.20531380e+00 -1.12673593e+00 -1.06926918e+00
-4.17194903e-01 -7.59348750e-01 3.83093119e-01 4.99401689e-01
-2.90879369e-01 4.16989438e-02 -1.88914791e-01 1.53323397e-01
4.57969993e-01 9.14299190e-02 7.69951522e-01 -4.35492098e-01
-3.90691310e-01 -3.30716223e-01 -4.79809016e-01 -1.33575356e+00
-2.64597356e-01 -5.85898519e-01 -1.50868848e-01 -1.20121157e+00
-1.64126962e-01 -5.04091322e-01 5.41973636e-02 5.87377310e-01
1.54078349e-01 6.77814722e-01 5.12113512e-01 1.76354751e-01
-7.81074643e-01 4.85309690e-01 1.19007695e+00 2.69504577e-01
-2.76402533e-01 -2.62229681e-01 -5.67973197e-01 7.42746770e-01
7.96450853e-01 -4.50615674e-01 -3.71297210e-01 -6.79629445e-01
-9.85682830e-02 3.65103292e-03 4.31881607e-01 -1.30426836e+00
2.60141432e-01 7.39836693e-02 2.52561748e-01 -3.64474595e-01
5.06787062e-01 -5.80021262e-01 6.11087345e-02 4.48240191e-01
1.43107608e-01 4.43128943e-01 5.27393401e-01 2.03595042e-01
1.07718788e-01 2.88573969e-02 7.30307519e-01 -1.26563028e-01
-8.60993683e-01 5.44203281e-01 -2.99057066e-01 1.53264210e-01
1.25497162e+00 -3.58570695e-01 -7.89717853e-01 -1.80865586e-01
-4.81230259e-01 4.29770976e-01 6.05911732e-01 4.81025249e-01
5.38541675e-01 -1.01864028e+00 -5.28688669e-01 3.15112382e-01
-4.02215928e-01 6.32222882e-03 2.50421435e-01 9.93861258e-01
-1.22278714e+00 3.20003420e-01 -1.87339887e-01 -7.89049864e-01
-1.41902661e+00 1.19764566e-01 4.55523759e-01 -2.02646367e-02
-1.14661610e+00 1.20648050e+00 4.67750654e-02 7.55871981e-02
-1.40204996e-01 -4.43788260e-01 3.81648213e-01 -3.76120597e-01
4.97486919e-01 6.95274234e-01 2.42815703e-01 -3.09225440e-01
-4.81582075e-01 4.27710891e-01 -2.28136182e-01 -3.74304026e-01
1.15467155e+00 2.52950587e-03 3.64865988e-01 -2.52045561e-02
9.99796689e-01 1.15507729e-01 -1.62275851e+00 2.87874371e-01
-4.73871410e-01 -9.01051223e-01 4.83997732e-01 -6.37488484e-01
-1.66248035e+00 3.75380754e-01 6.09571218e-01 1.56191364e-01
1.03595126e+00 -7.85033479e-02 1.26775384e+00 -1.90340385e-01
3.49658012e-01 -1.04061162e+00 -8.30009654e-02 2.32825205e-01
5.31227112e-01 -1.06585622e+00 2.26188973e-01 -8.05062175e-01
-5.70310295e-01 1.05584991e+00 8.38777184e-01 -4.54343289e-01
6.43214166e-01 7.06674993e-01 1.61264300e-01 -4.46714640e-01
-7.96430528e-01 -1.21995308e-01 1.41487256e-01 9.64906663e-02
3.75866145e-01 7.37239271e-02 -2.88137019e-01 5.07310703e-02
-3.41930777e-01 1.30473062e-01 7.25294650e-01 1.19565475e+00
-1.43382251e-01 -7.61997581e-01 -2.10353926e-01 7.10167348e-01
-6.81764424e-01 -1.60034567e-01 8.29928443e-02 8.35827351e-01
-1.62337627e-02 9.95960534e-01 3.21337402e-01 -4.69582915e-01
2.64525920e-01 -4.70423818e-01 4.28221464e-01 -6.05125964e-01
-7.18034208e-01 2.37860486e-01 2.46914670e-01 -1.36882901e+00
-2.11626351e-01 -6.62734091e-01 -1.32062578e+00 -4.44324911e-01
-3.55426669e-01 -1.28143564e-01 9.00493264e-01 6.88181579e-01
9.30366635e-01 6.82682216e-01 3.61890674e-01 -1.19555759e+00
-5.48404418e-02 -5.43888330e-01 -4.36786950e-01 -9.49555188e-02
1.27327546e-01 -6.42048895e-01 -3.49388808e-01 -5.10852318e-03] | [9.227652549743652, 0.0010232661152258515] |
0785f0b8-d981-4531-beeb-0daf93e9b8b0 | deep-metric-learning-with-soft-orthogonal | 2306.13055 | null | https://arxiv.org/abs/2306.13055v1 | https://arxiv.org/pdf/2306.13055v1.pdf | Deep Metric Learning with Soft Orthogonal Proxies | Deep Metric Learning (DML) models rely on strong representations and similarity-based measures with specific loss functions. Proxy-based losses have shown great performance compared to pair-based losses in terms of convergence speed. However, proxies that are assigned to different classes may end up being closely located in the embedding space and hence having a hard time to distinguish between positive and negative items. Alternatively, they may become highly correlated and hence provide redundant information with the model. To address these issues, we propose a novel approach that introduces Soft Orthogonality (SO) constraint on proxies. The constraint ensures the proxies to be as orthogonal as possible and hence control their positions in the embedding space. Our approach leverages Data-Efficient Image Transformer (DeiT) as an encoder to extract contextual features from images along with a DML objective. The objective is made of the Proxy Anchor loss along with the SO regularization. We evaluate our method on four public benchmarks for category-level image retrieval and demonstrate its effectiveness with comprehensive experimental results and ablation studies. Our evaluations demonstrate the superiority of our proposed approach over state-of-the-art methods by a significant margin. | ['Mahdi Eftekhari', 'Mahdi Shariatzadeh', 'Mahvash Mohazzebi', 'Dorsa Rahmatian', 'Monireh Moshavash', 'Farid Saberi-Movahed', 'Mohammad K. Ebrahimpour', 'Farshad Saberi-Movahed'] | 2023-06-22 | null | null | null | null | ['metric-learning', 'metric-learning', 'retrieval'] | ['computer-vision', 'methodology', 'methodology'] | [ 1.78224087e-01 -3.59586686e-01 -3.22134167e-01 -5.34610271e-01
-1.13615537e+00 -4.94967729e-01 6.56602681e-01 1.27715468e-01
-5.27646840e-01 4.78090018e-01 2.13322490e-01 7.39744231e-02
-2.95524985e-01 -6.02890491e-01 -6.72155082e-01 -7.41241217e-01
-1.08193874e-01 1.95378304e-01 -1.79229632e-01 1.01631105e-01
1.77555710e-01 3.20881993e-01 -1.27519393e+00 1.61984921e-01
7.85706639e-01 1.44070208e+00 1.07771203e-01 -1.85579136e-02
2.63610810e-01 7.28020132e-01 -4.05844688e-01 -5.97695827e-01
7.00177968e-01 -2.02968016e-01 -5.54227889e-01 -7.31673539e-02
6.38006151e-01 -4.97275203e-01 -5.78656197e-01 1.17231154e+00
5.68457961e-01 1.20084502e-01 8.79326344e-01 -1.47744477e+00
-9.77294147e-01 2.68481880e-01 -8.58902216e-01 -5.41710854e-03
-3.23940590e-02 -1.40045574e-02 1.72377503e+00 -1.26641893e+00
4.24215794e-01 1.10968626e+00 5.18706679e-01 3.18973660e-01
-1.40990436e+00 -6.83921039e-01 8.13549310e-02 4.70081121e-01
-1.59767187e+00 -3.44586313e-01 8.92127275e-01 -2.61377305e-01
4.72238153e-01 2.56244570e-01 2.25289911e-01 9.46877122e-01
-8.69158581e-02 9.26578462e-01 9.91473854e-01 -2.72574246e-01
1.61642253e-01 3.61477852e-01 7.93411732e-02 6.97844684e-01
9.29423347e-02 9.82616320e-02 -6.03595674e-01 -1.11664720e-01
4.93315518e-01 1.79995477e-01 -3.98838967e-01 -8.15138638e-01
-1.27986646e+00 1.11767209e+00 8.75099301e-01 -7.37021565e-02
-4.85720932e-01 3.04310620e-01 4.29942042e-01 2.96336800e-01
2.68541664e-01 5.24251282e-01 -1.53976724e-01 2.61411250e-01
-8.58329415e-01 3.30384374e-01 3.23981553e-01 8.77795637e-01
6.80732608e-01 -5.56369901e-01 -5.06743193e-01 1.19294488e+00
3.33418220e-01 2.08810359e-01 4.20479745e-01 -8.94139051e-01
6.95776880e-01 6.26821220e-01 -9.37899947e-02 -1.45377636e+00
-1.19841797e-02 -6.56251907e-01 -7.00971305e-01 1.91404566e-01
1.39757246e-01 3.40140104e-01 -6.92642033e-01 2.13342142e+00
1.89141840e-01 1.63061112e-01 -1.54439330e-01 1.12043643e+00
5.02203584e-01 5.59476078e-01 -1.19799217e-02 2.53827095e-01
1.28058064e+00 -9.90297794e-01 -3.58521163e-01 -2.17724442e-01
5.52990615e-01 -7.06064165e-01 1.31344593e+00 1.31904930e-01
-1.06591868e+00 -2.72131413e-01 -1.29608631e+00 -2.44219154e-01
-1.76482856e-01 2.49956101e-01 4.88604814e-01 2.45810807e-01
-7.14839697e-01 5.97420514e-01 -6.16588831e-01 6.75524399e-02
6.04456067e-01 2.66216397e-01 -3.75195742e-01 -4.02356833e-01
-1.23314178e+00 6.78936958e-01 -4.63913344e-02 -2.75342446e-02
-8.34633231e-01 -8.93327296e-01 -8.33525658e-01 1.89337224e-01
1.56859025e-01 -5.39199471e-01 8.18119645e-01 -7.30947673e-01
-9.26858068e-01 9.58777308e-01 1.38132110e-01 -4.67106789e-01
7.41542518e-01 -2.23851100e-01 -1.85931891e-01 2.83912152e-01
3.80513489e-01 7.60300696e-01 7.64402926e-01 -1.21708918e+00
-6.33893073e-01 -1.88134313e-01 3.33530456e-01 3.26196373e-01
-9.14071858e-01 -1.73770115e-01 -6.51230037e-01 -1.04146051e+00
2.94487894e-01 -9.28639650e-01 3.19263190e-02 6.61300480e-01
-4.87784326e-01 -9.94473994e-02 6.29623234e-01 -5.36569476e-01
1.37595594e+00 -2.22219253e+00 2.33411014e-01 2.84261972e-01
2.78952628e-01 1.89429373e-01 -4.34048027e-01 3.28717858e-01
-6.58444613e-02 -2.05160826e-02 -3.68558466e-01 -6.02439046e-01
2.95960695e-01 1.19546115e-01 -3.15151423e-01 5.90035379e-01
3.14264297e-01 7.17349529e-01 -9.05429482e-01 -4.47097570e-01
1.47603229e-01 5.50194442e-01 -8.34411204e-01 2.62057155e-01
1.32482365e-01 -4.98644868e-03 -3.01613748e-01 7.51008630e-01
7.90776789e-01 -4.01405036e-01 -1.62026688e-01 -4.99668241e-01
3.09732020e-01 4.21772778e-01 -9.78618026e-01 1.61745393e+00
-5.75574160e-01 5.54300249e-01 -2.05676645e-01 -1.10755336e+00
6.82615161e-01 -3.62441875e-02 4.62838024e-01 -8.50191653e-01
-1.16496705e-01 3.06297660e-01 -1.47709414e-01 -1.83130607e-01
2.85768181e-01 4.52026501e-02 -3.88852991e-02 2.91806370e-01
-6.90551400e-02 1.80274203e-01 5.19052846e-03 2.84317404e-01
1.09504902e+00 -1.38241261e-01 -1.74839869e-02 -1.71105966e-01
6.49429798e-01 -5.27660966e-01 7.70212173e-01 6.49243772e-01
-3.40526521e-01 7.45114923e-01 6.09558225e-01 -1.41030952e-01
-9.75698829e-01 -1.26520073e+00 -2.86107481e-01 8.07562172e-01
3.02590311e-01 -3.17385703e-01 -3.90268326e-01 -9.20843005e-01
2.20809206e-01 4.86163706e-01 -7.69734204e-01 -5.20293057e-01
-3.72364730e-01 -6.83584273e-01 3.49405408e-01 4.35410351e-01
6.57435238e-01 -5.45415819e-01 -2.57846117e-01 2.68696453e-02
-2.12458238e-01 -1.05867815e+00 -7.52445817e-01 1.88101783e-01
-5.47708452e-01 -1.05228412e+00 -8.62825215e-01 -8.26000869e-01
9.29090023e-01 6.22810304e-01 1.04396319e+00 2.11468972e-02
-2.29711741e-01 1.70160890e-01 -3.68898958e-01 1.24498876e-02
5.08110337e-02 -7.72630423e-02 -1.06271937e-01 3.12830418e-01
4.58807677e-01 -5.64244986e-01 -1.22913527e+00 5.29364944e-01
-9.67190206e-01 -3.02870393e-01 6.70459747e-01 1.23156655e+00
6.43020570e-01 -1.39190733e-01 5.52984238e-01 -5.73783755e-01
4.69875932e-01 -6.61819518e-01 -6.18512392e-01 2.91407436e-01
-8.45980525e-01 1.61556631e-01 5.80743074e-01 -2.94875920e-01
-5.11474669e-01 -2.06994563e-01 3.47674817e-01 -7.03600287e-01
6.15164518e-01 4.68838096e-01 -3.64848852e-01 -7.27182403e-02
4.06510592e-01 1.17855608e-01 -3.66606303e-02 -3.65892172e-01
3.29159766e-01 6.05074227e-01 3.65233988e-01 -5.25476098e-01
9.63803232e-01 5.68490922e-01 -1.95652596e-03 -4.47829247e-01
-8.99082541e-01 -4.68761265e-01 -1.27515942e-01 9.30709019e-02
4.30793524e-01 -1.08105111e+00 -3.16635519e-01 3.37113082e-01
-8.59108269e-01 9.06166285e-02 -8.43691379e-02 5.95843971e-01
-3.37449074e-01 4.00754511e-01 -5.07433474e-01 -4.41199422e-01
-1.84080601e-01 -1.21300948e+00 9.80945051e-01 -4.99413349e-02
-2.78502822e-01 -7.75347531e-01 -1.44142047e-01 3.83174419e-01
3.37801844e-01 1.71750665e-01 1.08802164e+00 -4.82710779e-01
-6.86651945e-01 -4.92736787e-01 -6.59486353e-01 7.94312656e-01
1.12816274e-01 -2.65000999e-01 -9.14887011e-01 -5.32920361e-01
-1.46121174e-01 -4.76238191e-01 1.16745830e+00 6.69610575e-02
1.46722949e+00 -3.41146708e-01 -1.72708437e-01 8.91437471e-01
1.61346173e+00 -2.23955497e-01 4.96928841e-01 5.13316214e-01
6.44854188e-01 6.57553911e-01 7.35879958e-01 4.67331201e-01
3.98671389e-01 1.01134360e+00 5.89546800e-01 -1.33846616e-02
-1.94499463e-01 -3.15155715e-01 3.23221087e-01 6.93192005e-01
5.20985007e-01 -2.73886412e-01 -6.42286420e-01 7.41063952e-01
-1.77835393e+00 -7.24462628e-01 2.68053740e-01 2.42529917e+00
8.47623587e-01 1.23370491e-01 -1.40902638e-01 2.49228746e-01
6.50107861e-01 4.03685421e-01 -6.75241649e-01 -1.36613073e-02
-1.62915289e-01 3.61302793e-02 5.56164205e-01 2.68803477e-01
-1.30385339e+00 5.17029941e-01 5.20251751e+00 9.13508117e-01
-1.16272235e+00 8.40029493e-02 7.37739325e-01 -5.89231193e-01
-4.28647012e-01 -8.58765617e-02 -5.80800891e-01 7.48888552e-01
3.81716967e-01 -8.05585235e-02 3.55593354e-01 6.78726792e-01
-4.80903462e-02 2.22036511e-01 -1.33014846e+00 1.11907184e+00
-1.07009960e-02 -1.19083393e+00 1.80388093e-01 1.49207130e-01
7.87550509e-01 2.69984990e-01 5.54004431e-01 2.36253113e-01
3.56219113e-02 -8.98424447e-01 8.64097357e-01 3.35558981e-01
7.84257352e-01 -8.75441134e-01 5.47429681e-01 -1.38165966e-01
-9.82174575e-01 -3.03581804e-01 -7.11333573e-01 2.50904828e-01
-1.60848990e-01 7.28774846e-01 -4.12027955e-01 3.06628227e-01
6.38579547e-01 9.59954143e-01 -4.50683683e-01 1.05641639e+00
-2.28434637e-01 4.71663266e-01 -2.43410721e-01 1.37236774e-01
4.45581466e-01 -2.81252265e-01 4.68059421e-01 8.28565836e-01
3.45625311e-01 -2.96531498e-01 9.40289535e-03 9.44639862e-01
-5.38732111e-01 1.48865253e-01 -4.61387753e-01 -1.62151270e-02
7.13677287e-01 1.11204314e+00 -1.67863861e-01 -7.14137480e-02
-6.38626695e-01 1.09667504e+00 5.47911823e-01 3.21848959e-01
-1.00826502e+00 -5.23135483e-01 1.11004651e+00 7.57052526e-02
4.15397137e-01 1.97109550e-01 -2.56275088e-01 -1.14500058e+00
5.97141087e-01 -8.25983882e-01 5.11352479e-01 -3.96397293e-01
-1.69091237e+00 4.25921053e-01 -3.14230829e-01 -1.72426343e+00
1.31112471e-01 -4.92130429e-01 -3.13170314e-01 8.58329654e-01
-2.01237321e+00 -1.12556636e+00 -2.69522846e-01 5.22031009e-01
1.08431168e-01 -2.53949881e-01 6.86525166e-01 8.18382502e-01
-5.60118973e-01 1.15545201e+00 4.59542423e-01 2.36563519e-01
9.09265518e-01 -1.17233479e+00 -5.39795831e-02 7.20101178e-01
1.27433628e-01 6.72974527e-01 3.00136656e-01 -7.86659345e-02
-1.23680544e+00 -1.21996772e+00 7.58466423e-01 -2.92401314e-01
7.39996314e-01 -4.68234479e-01 -8.24369729e-01 3.24989110e-01
-1.69812799e-01 5.23540020e-01 8.71882796e-01 -1.71464443e-01
-9.12578046e-01 -5.03680110e-01 -1.06251395e+00 6.49078250e-01
9.67006207e-01 -7.53728926e-01 -2.80604124e-01 3.66481960e-01
5.21853745e-01 9.02590379e-02 -6.95861280e-01 4.50482517e-01
7.18345702e-01 -1.04924953e+00 1.24317527e+00 -4.49439228e-01
8.62849355e-01 -2.88310677e-01 -6.28844142e-01 -1.13329577e+00
-3.93289298e-01 -3.33707958e-01 -6.42741546e-02 1.29394042e+00
4.23819661e-01 -5.65557301e-01 7.12817729e-01 6.26253605e-01
1.18002452e-01 -1.17922425e+00 -1.00848460e+00 -9.71163571e-01
1.55052885e-01 -2.81773537e-01 5.95247090e-01 8.84340107e-01
-2.35503361e-01 1.28714904e-01 -4.87657785e-01 1.81604758e-01
8.38253200e-01 1.13206364e-01 4.51294750e-01 -9.68846619e-01
-2.69135207e-01 -6.80947661e-01 -6.69406116e-01 -1.08058429e+00
2.52610773e-01 -1.04391170e+00 -2.14620563e-03 -1.28123677e+00
3.64971340e-01 -8.22290003e-01 -9.81513798e-01 3.39269966e-01
-2.71777153e-01 4.70123887e-01 2.05492780e-01 4.86702144e-01
-5.57365894e-01 8.34127307e-01 8.43872607e-01 -4.12022620e-01
1.86308131e-01 -2.26421073e-01 -7.60769129e-01 5.45992315e-01
5.81677973e-01 -6.05884254e-01 -5.24694979e-01 -7.03142107e-01
3.45664650e-01 -4.09535170e-01 6.15800202e-01 -9.12088275e-01
2.34654397e-02 2.00207829e-01 3.43930572e-01 -3.14173669e-01
4.68108654e-01 -9.76320088e-01 -3.32257211e-01 4.10889536e-01
-7.18749821e-01 -1.38208836e-01 -2.80105203e-01 7.34475970e-01
-4.99999464e-01 -1.41535431e-01 9.49929237e-01 1.72895133e-01
-4.24081862e-01 5.48576713e-01 4.51925755e-01 1.54510304e-01
1.01909041e+00 -3.25676277e-02 -7.25260526e-02 -4.01467741e-01
-1.13642886e-01 4.41846579e-01 6.59749508e-01 5.65006733e-01
7.36754477e-01 -1.77534306e+00 -7.26879656e-01 1.41160294e-01
6.20040119e-01 -2.54432887e-01 3.92486453e-02 8.20356131e-01
-3.86342019e-01 3.63004655e-01 -4.44834270e-02 -5.14466703e-01
-1.08882511e+00 5.27778387e-01 2.76238650e-01 -4.26471323e-01
-4.16051745e-01 1.06947696e+00 5.15798867e-01 -3.64656270e-01
5.15366137e-01 -1.87983755e-02 1.09769545e-01 2.34441370e-01
5.61625659e-01 2.21926823e-01 7.72300959e-02 -5.60515940e-01
-5.50137162e-01 5.04495740e-01 -3.24329764e-01 -7.81092048e-02
1.35002565e+00 -1.51256427e-01 -1.19764626e-01 1.75819427e-01
2.02048063e+00 -1.32730380e-01 -1.42641556e+00 -7.51977384e-01
2.23873034e-02 -8.95224035e-01 2.88742274e-01 -7.10084200e-01
-1.28021932e+00 9.44894135e-01 6.30665660e-01 -1.53979465e-01
1.20101547e+00 -7.50593320e-02 1.06959677e+00 2.06998631e-01
1.61641285e-01 -1.19823527e+00 3.46928388e-01 1.08632036e-01
1.00671375e+00 -1.56925154e+00 4.46202755e-02 -2.18263119e-01
-4.86984044e-01 7.34384775e-01 3.45603704e-01 -3.27669173e-01
5.99874020e-01 -1.86038062e-01 -4.39029410e-02 -5.04908822e-02
-6.55908823e-01 8.15517455e-02 5.37453771e-01 3.28933179e-01
2.80486375e-01 3.70504931e-02 -3.77631664e-01 3.56825620e-01
1.08196162e-01 -3.28028381e-01 3.51730920e-02 7.65684903e-01
1.00732492e-02 -1.10350358e+00 -9.92687345e-02 5.55849433e-01
-5.57097316e-01 -2.27591306e-01 -1.71546638e-01 4.98387367e-01
-1.06871173e-01 8.33280146e-01 -5.81885017e-02 -3.62090200e-01
2.54354268e-01 -2.96596915e-01 3.02359432e-01 -3.06833118e-01
-1.94291174e-01 -1.97608992e-01 -2.38927603e-01 -7.22294986e-01
-8.46787691e-02 -6.43857419e-01 -9.71143007e-01 -1.73625603e-01
-2.74065673e-01 -3.75348963e-02 4.97864366e-01 4.60108846e-01
4.66960222e-01 2.88972020e-01 1.25248075e+00 -4.15996641e-01
-1.11449146e+00 -6.27495229e-01 -5.86126745e-01 7.32147217e-01
4.76199031e-01 -8.57021391e-01 -5.04840553e-01 -4.80704933e-01] | [9.432486534118652, 3.088984727859497] |
dacf78fa-9c26-4d39-8c5c-77ab4c567930 | female-mosquito-detection-by-means-of-ai | 2306.10843 | null | https://arxiv.org/abs/2306.10843v1 | https://arxiv.org/pdf/2306.10843v1.pdf | Female mosquito detection by means of AI techniques inside release containers in the context of a Sterile Insect Technique program | The Sterile Insect Technique (SIT) is a biological pest control technique based on the release into the environment of sterile males of the insect species whose population is to be controlled. The entire SIT process involves mass-rearing within a biofactory, sorting of the specimens by sex, sterilization, and subsequent release of the sterile males into the environment. The reason for avoiding the release of female specimens is because, unlike males, females bite, with the subsequent risk of disease transmission. In the case of Aedes mosquito biofactories for SIT, the key point of the whole process is sex separation. This process is nowadays performed by a combination of mechanical devices and AI-based vision systems. However, there is still a possibility of false negatives, so a last stage of verification is necessary before releasing them into the environment. It is known that the sound produced by the flapping of adult male mosquitoes is different from that produced by females, so this feature can be used to detect the presence of females in containers prior to environmental release. This paper presents a study for the detection of females in Aedes mosquito release vessels for SIT programs. The containers used consist of PVC a tubular design of 8.8cm diameter and 12.5cm height. The containers were placed in an experimental setup that allowed the recording of the sound of mosquito flight inside of them. Each container was filled with 250 specimens considering the cases of (i) only male mosquitoes, (ii) only female mosquitoes, and (iii) 75% males and 25% females. Case (i) was used for training and testing, whereas cases (ii) and (iii) were used only for testing. Two algorithms were implemented for the detection of female mosquitoes: an unsupervised outlier detection algorithm (iForest) and a one-class SVM trained with male-only recordings. | ['Pedro Zuccarello', 'David Almenar', 'Jordi Grau-Haro', 'Javier Naranjo-Alcazar'] | 2023-06-19 | null | null | null | null | ['outlier-detection'] | ['methodology'] | [ 1.56915814e-01 -2.71665066e-01 3.19884270e-01 2.44846195e-02
5.40126979e-01 -8.11380208e-01 3.82022202e-01 5.49189210e-01
-6.86418056e-01 7.71685064e-01 -2.17106834e-01 -4.21776026e-01
6.28091842e-02 -8.09245050e-01 -5.24065912e-01 -1.13545096e+00
-4.44078267e-01 4.34929281e-01 8.56969953e-02 5.04539795e-02
2.34933749e-01 9.19298947e-01 -1.91859996e+00 3.04477364e-02
7.53744900e-01 7.24308372e-01 2.40446821e-01 6.68587625e-01
-1.23074301e-01 -1.26604989e-01 -9.21346426e-01 -2.10382178e-01
3.47968817e-01 -3.08982998e-01 1.34379998e-01 -8.70487373e-03
4.03186632e-03 -6.20067477e-01 6.60011232e-01 8.79350603e-01
2.91910470e-01 -1.53505310e-01 9.29909229e-01 -1.16299021e+00
1.25668421e-01 2.11473137e-01 -6.95081651e-01 1.74840093e-01
5.63748777e-01 2.06281170e-01 1.89015642e-01 -6.23167157e-01
4.19256955e-01 1.12801874e+00 4.05817866e-01 5.33749759e-01
-1.31679749e+00 -7.19004154e-01 -3.57695132e-01 -2.63303727e-01
-1.37452281e+00 -3.15439433e-01 3.71052295e-01 -6.79327011e-01
7.12220728e-01 3.41674089e-01 1.05363858e+00 9.98718083e-01
7.77167439e-01 7.60712055e-03 1.11327195e+00 -2.65107542e-01
6.49297655e-01 5.66579759e-01 2.78820321e-02 7.99177215e-02
9.50790644e-01 2.83448756e-01 -3.91931795e-02 -3.82336378e-01
2.87887961e-01 3.06413263e-01 -1.41177744e-01 -3.46373431e-02
-4.79317367e-01 9.01880860e-01 -2.60890462e-02 6.42302334e-01
-4.43413854e-01 -5.66936433e-01 3.93904656e-01 1.72483344e-02
2.68586785e-01 2.86504030e-01 -3.82964581e-01 7.36736432e-02
-1.10112047e+00 1.99651092e-01 9.72649097e-01 3.40483367e-01
2.90734679e-01 1.48335680e-01 1.50938213e-01 4.38220441e-01
4.96403873e-01 1.09270978e+00 5.00042975e-01 -5.33684436e-03
1.78694606e-01 7.48356998e-01 2.13292867e-01 -1.14590061e+00
-2.08175659e-01 -1.27113387e-01 -5.95697701e-01 3.29311192e-01
5.68080962e-01 -4.77074742e-01 -8.45492840e-01 1.50825596e+00
4.93624300e-01 -2.64108390e-01 1.17080547e-01 5.29400706e-01
5.39629400e-01 1.22411096e+00 6.71458468e-02 -8.09100330e-01
1.37130499e+00 -1.93805248e-02 -7.07550883e-01 1.95445314e-01
5.62509447e-02 -7.04861104e-01 5.42258084e-01 6.72186375e-01
-6.83254063e-01 -3.50289434e-01 -1.17791474e+00 6.60245836e-01
-9.54023600e-01 9.26438197e-02 3.82592529e-01 9.16463435e-01
-8.36783588e-01 4.84782457e-01 -9.34041798e-01 -6.60326600e-01
-2.22487062e-01 7.38105536e-01 -4.54764783e-01 3.40546727e-01
-4.92940664e-01 5.83331048e-01 -6.95202574e-02 5.58112442e-01
-8.13505113e-01 -1.55006275e-01 -7.06213593e-01 -8.57179910e-02
-2.50138402e-01 -9.77519676e-02 3.60004395e-01 -7.17383981e-01
-1.45184481e+00 8.14087152e-01 -3.42091233e-01 -3.77209544e-01
6.43132031e-01 -2.96601743e-01 -2.53439665e-01 2.20322251e-01
2.32162148e-01 4.24330294e-01 8.75646651e-01 -1.22736621e+00
-5.09188294e-01 -9.51790035e-01 -3.61687273e-01 -2.95719445e-01
-1.41892076e-01 1.15373962e-01 2.67011851e-01 -3.69287610e-01
-3.63028273e-02 -7.68125474e-01 5.02520949e-02 -3.14715058e-01
-4.10286933e-01 -5.70290312e-02 1.01056910e+00 -7.31444240e-01
8.84862900e-01 -2.22098780e+00 -2.92983167e-02 3.39349389e-01
-1.74032703e-01 4.85750109e-01 2.19556898e-01 6.50511384e-01
4.95180786e-02 1.12812728e-01 -2.38683417e-01 -3.51057708e-01
-4.19431448e-01 6.18401030e-03 -3.36897284e-01 4.85739917e-01
1.41676113e-01 -4.13548350e-01 -6.50350392e-01 -4.14550751e-01
2.60300398e-01 7.79117942e-01 -3.21943283e-01 5.42414546e-01
6.37688860e-02 5.52188635e-01 -3.61274093e-01 9.28633511e-01
1.12089622e+00 8.96493971e-01 -1.04219139e-01 4.34207451e-03
-7.06932962e-01 -1.72767520e-01 -1.34577227e+00 4.86484200e-01
-1.35324121e-01 2.45722562e-01 4.48509187e-01 -4.83545959e-01
1.10602689e+00 4.84801590e-01 5.93669295e-01 -7.73927793e-02
3.30285847e-01 5.02677739e-01 -6.66901749e-03 -6.59719706e-01
3.70474428e-01 1.68203507e-02 3.98024887e-01 3.09042811e-01
-1.00121677e-01 -2.15247154e-01 6.62042677e-01 -1.62041694e-01
5.16446650e-01 -7.79572353e-02 1.33712351e-01 -3.84376287e-01
5.98117113e-01 -1.93340302e-01 6.96785390e-01 2.96615452e-01
-2.27998272e-01 1.08780831e-01 3.89101654e-01 -5.30501068e-01
-6.22710288e-01 -1.01877010e+00 -6.58499241e-01 5.02911210e-01
6.06646314e-02 -3.98259051e-02 -9.71021175e-01 -1.21497631e-01
9.96286124e-02 6.30152941e-01 -5.84324777e-01 2.52209418e-02
-2.75103778e-01 -1.02477419e+00 2.43823379e-01 -1.49265349e-01
4.36311960e-01 -1.13409269e+00 -1.20857978e+00 5.75145781e-02
3.10116380e-01 -5.63964784e-01 1.72470942e-01 4.94529724e-01
-1.14492953e+00 -1.16994250e+00 -4.79710132e-01 -4.02114034e-01
8.71906936e-01 1.54493749e-01 4.17120188e-01 2.03483045e-01
-4.54436749e-01 4.38683070e-02 -2.78385580e-01 -7.29188859e-01
-4.28590536e-01 -4.05188799e-01 4.55652803e-01 1.41021714e-01
9.37930465e-01 -5.05688369e-01 -5.12950480e-01 2.56984740e-01
-9.83775079e-01 -4.93780792e-01 1.34308830e-01 5.45635462e-01
3.74778777e-01 3.41555387e-01 1.33553028e-01 -4.23263133e-01
4.43486869e-01 -3.80760431e-01 -9.03050244e-01 3.14762481e-02
2.33730394e-02 -2.55921543e-01 8.10638368e-01 -3.92113388e-01
-5.96332073e-01 1.61600754e-01 9.62793082e-02 3.64038311e-02
-7.14659989e-01 -5.94517551e-02 -4.58129138e-01 1.38715595e-01
3.71379673e-01 8.93395245e-02 1.54005304e-01 -3.53400350e-01
-5.76757491e-01 8.40405345e-01 1.43330351e-01 -1.55043647e-01
8.16229641e-01 3.94296229e-01 2.66186059e-01 -1.34676540e+00
1.41402818e-02 -2.32304618e-01 -5.03056407e-01 -3.22434664e-01
1.25319159e+00 -3.59952688e-01 -8.65794659e-01 7.49114096e-01
-1.16498995e+00 2.09834769e-01 1.66546866e-01 8.59136820e-01
1.77124292e-01 1.90869048e-01 -5.28295040e-01 -1.37513065e+00
-5.45636892e-01 -1.41714680e+00 8.66950631e-01 6.65600300e-01
-2.13056542e-02 -7.47078300e-01 1.90779567e-01 2.31883265e-02
3.16247404e-01 4.66590434e-01 6.02143764e-01 -6.83361053e-01
3.09144258e-02 -5.64145863e-01 3.30197513e-01 2.46495664e-01
4.64573234e-01 8.37619781e-01 -9.60599124e-01 -5.34880280e-01
6.00060523e-01 6.58854172e-02 5.64380467e-01 6.34926319e-01
4.08175707e-01 -2.50831604e-01 -5.32451630e-01 5.06954789e-01
1.60901415e+00 9.61119831e-01 4.83601481e-01 1.34883970e-01
1.58588842e-01 8.45107853e-01 6.07122719e-01 4.76394325e-01
-2.61436969e-01 5.61500847e-01 6.10177934e-01 -6.39792681e-02
5.70575118e-01 -2.26672320e-03 7.47010350e-01 2.04856202e-01
-4.43657875e-01 -4.19277638e-01 -5.19652963e-01 2.76579827e-01
-9.04086828e-01 -1.00454879e+00 -2.84623265e-01 2.93287802e+00
5.88440955e-01 -8.92686918e-02 3.39040756e-01 5.16879559e-01
8.91214550e-01 -5.73612154e-01 -5.77184074e-02 -8.67845178e-01
1.32926628e-01 2.14949980e-01 5.66022515e-01 5.06295264e-01
-1.13023031e+00 4.70115662e-01 5.53886366e+00 2.08671153e-01
-1.51822197e+00 -5.36081374e-01 2.68485844e-01 -3.68878357e-02
2.56410360e-01 -9.55231637e-02 -1.36201262e+00 8.97185922e-01
5.96364796e-01 3.56926203e-01 2.95199484e-01 4.58501607e-01
3.68889600e-01 -8.90857458e-01 -9.35723364e-01 6.53078079e-01
1.76838294e-01 -1.84282467e-01 1.68266073e-01 4.53488141e-01
1.93777278e-01 -4.63586509e-01 -2.20362931e-01 -1.61707625e-01
-3.94447893e-01 -7.09708393e-01 3.29304308e-01 2.92980015e-01
3.03524941e-01 -7.68354177e-01 1.25063384e+00 3.41397196e-01
-1.07505071e+00 -9.74046886e-02 -6.34477139e-01 -8.66337419e-02
1.03085764e-01 6.36826813e-01 -7.90928721e-01 2.97924578e-01
8.25082541e-01 2.38300785e-01 -5.73609054e-01 1.21411920e+00
1.27370954e-02 4.79699433e-01 -9.48601186e-01 -2.84934282e-01
-4.26916033e-02 -8.41512620e-01 9.48450267e-01 1.03974438e+00
5.80904186e-01 -1.58056960e-01 -1.07928880e-01 8.46148372e-01
7.22946346e-01 1.58431694e-01 -1.11206353e+00 -4.21923809e-02
3.56386214e-01 9.10298645e-01 -1.13101339e+00 -7.04286620e-02
-4.75436933e-02 6.90515459e-01 -5.13190806e-01 2.60612577e-01
-5.10642111e-01 -6.90934062e-01 3.03696454e-01 3.05261612e-01
3.73102009e-01 -1.66555762e-01 -2.80770790e-02 -8.45368564e-01
1.26280800e-01 -5.36404788e-01 2.77959794e-01 -2.06867784e-01
-9.79747772e-01 6.19646013e-01 3.54236186e-01 -1.16587818e+00
-2.59905934e-01 -7.87615061e-01 -7.09536016e-01 6.76877975e-01
-6.75025761e-01 -7.44695902e-01 -2.61982977e-01 2.35187516e-01
3.00117910e-01 -3.01313698e-01 1.06405091e+00 8.90079066e-02
-7.15880692e-01 2.86684245e-01 2.06999630e-01 -1.45183399e-01
3.32615227e-01 -9.87889469e-01 -5.83496988e-01 8.51742148e-01
-3.04605633e-01 8.54136169e-01 1.07347965e+00 -1.03306365e+00
-1.36811268e+00 -6.31357193e-01 1.17813838e+00 -6.07108548e-02
1.27940714e-01 -7.29479074e-01 -5.30003011e-01 2.46426210e-01
1.42575994e-01 -6.41434193e-01 8.88972819e-01 -3.01945716e-01
1.21659093e-01 -2.78932899e-01 -1.77737308e+00 4.26784694e-01
1.64207369e-01 9.15555507e-02 -6.14730835e-01 3.28660995e-01
2.44802590e-02 5.47009856e-02 -5.56653500e-01 3.53583038e-01
7.19218314e-01 -1.01478016e+00 7.20538616e-01 -2.15680644e-01
1.36461645e-01 -6.64418101e-01 1.13744065e-02 -9.88367498e-01
2.99536943e-01 -2.74412811e-01 5.90004742e-01 1.43282712e+00
2.53965467e-01 -8.09749961e-01 3.01946074e-01 4.59874183e-01
2.87644476e-01 -2.77277499e-01 -8.76557648e-01 -8.12957406e-01
-1.86813936e-01 3.73562038e-01 2.51321763e-01 3.54521424e-01
-1.69786550e-02 1.55404896e-01 -1.70998111e-01 3.80413294e-01
6.66427195e-01 -9.64385718e-02 6.51908278e-01 -1.42121768e+00
-2.50085533e-01 -1.39331624e-01 -7.09444106e-01 -1.66540951e-01
-3.41477692e-01 -5.33829778e-02 -1.72057405e-01 -1.25422466e+00
-6.14862069e-02 7.03893825e-02 3.61925542e-01 2.38399506e-01
1.66130349e-01 1.11519873e-01 1.01625454e-02 -2.77526360e-02
3.68159413e-01 1.76233560e-01 5.11068106e-01 -1.63665600e-02
-8.86220157e-01 4.67622250e-01 2.12967128e-01 5.03874540e-01
8.09097946e-01 -6.21433735e-01 -3.17430586e-01 -1.42498851e-01
-7.18074590e-02 -1.41281277e-01 1.93579763e-01 -1.12001395e+00
-1.66078396e-02 3.30177024e-02 6.33071840e-01 -8.68094146e-01
3.61825734e-01 -1.41757631e+00 4.77959424e-01 1.04197598e+00
5.02205431e-01 2.55488575e-01 4.21856672e-01 2.46150345e-01
-1.22243166e-02 -6.30541325e-01 7.41003454e-01 2.97811609e-02
2.52115011e-01 -2.48910740e-01 -9.84748840e-01 -7.70974517e-01
1.34572101e+00 -7.86302745e-01 -9.09146219e-02 -4.03360315e-02
-6.36793971e-01 -5.82902655e-02 5.59345543e-01 -1.36329874e-01
3.94161463e-01 -5.29601872e-01 -5.20369768e-01 6.11013889e-01
-2.22058758e-01 -8.10452104e-02 4.66396585e-02 7.73875058e-01
-9.19898987e-01 2.86257952e-01 -5.86103916e-01 -6.83182359e-01
-1.61511767e+00 7.06130981e-01 1.88839450e-01 1.14316173e-01
-1.33564532e-01 5.73081434e-01 9.18020606e-02 1.47608206e-01
4.05766070e-01 -4.67194438e-01 -7.81480551e-01 1.63724959e-01
8.25392544e-01 3.91668737e-01 4.33467887e-02 -6.63491666e-01
-4.18930441e-01 7.23759294e-01 7.62999430e-02 -5.60156256e-02
1.37532413e+00 4.35476720e-01 -5.74477851e-01 5.30802965e-01
7.24991083e-01 5.78125775e-01 -7.67133653e-01 9.18003142e-01
-3.46750259e-01 -4.80473816e-01 -3.15611869e-01 -5.27140498e-01
-8.20053279e-01 8.79229724e-01 8.68109107e-01 4.72895652e-01
1.08200586e+00 -5.15128195e-01 1.75716668e-01 1.55399412e-01
6.61156178e-01 -9.15436924e-01 -4.91332620e-01 1.15673535e-01
7.52887070e-01 -9.93626297e-01 -1.57249551e-02 -3.64840746e-01
-5.40614910e-02 1.31462944e+00 3.97892088e-01 -2.05119506e-01
6.58771157e-01 3.91070873e-01 -1.58331662e-01 -1.54764295e-01
-3.66599858e-01 2.74306148e-01 -2.55730122e-01 7.81414688e-01
4.94638681e-01 2.24487931e-01 -9.78926361e-01 4.89633024e-01
-2.78560221e-01 -3.35790329e-02 4.43213493e-01 1.07772374e+00
-4.10886973e-01 -9.61836278e-01 -8.88738453e-01 4.45620328e-01
-5.78009963e-01 2.23009601e-01 -6.62848294e-01 7.28012681e-01
5.54271638e-01 1.43961370e+00 3.16474795e-01 -2.35378265e-01
2.44348198e-01 -6.15316257e-02 2.64494568e-01 -3.34971428e-01
-9.65713978e-01 1.22148618e-01 -3.76987427e-01 -9.60968509e-02
-2.52448380e-01 -7.50515580e-01 -9.47757065e-01 -3.35644096e-01
-3.85424614e-01 7.44649947e-01 1.24053085e+00 6.75167024e-01
-3.39366868e-02 -2.64781922e-01 8.79405379e-01 -7.82079458e-01
-1.43085331e-01 -1.11079037e+00 -9.39104140e-01 6.29526377e-02
1.57743037e-01 -8.81147623e-01 -1.18356895e+00 6.26929551e-02] | [12.849227905273438, 0.31324324011802673] |
c59b1911-18af-49f9-b6f4-134c141bbfad | visibility-aware-pixelwise-view-selection-for | 2302.07182 | null | https://arxiv.org/abs/2302.07182v1 | https://arxiv.org/pdf/2302.07182v1.pdf | Visibility-Aware Pixelwise View Selection for Multi-View Stereo Matching | The performance of PatchMatch-based multi-view stereo algorithms depends heavily on the source views selected for computing matching costs. Instead of modeling the visibility of different views, most existing approaches handle occlusions in an ad-hoc manner. To address this issue, we propose a novel visibility-guided pixelwise view selection scheme in this paper. It progressively refines the set of source views to be used for each pixel in the reference view based on visibility information provided by already validated solutions. In addition, the Artificial Multi-Bee Colony (AMBC) algorithm is employed to search for optimal solutions for different pixels in parallel. Inter-colony communication is performed both within the same image and among different images. Fitness rewards are added to validated and propagated solutions, effectively enforcing the smoothness of neighboring pixels and allowing better handling of textureless areas. Experimental results on the DTU dataset show our method achieves state-of-the-art performance among non-learning-based methods and retrieves more details in occluded and low-textured regions. | ['Minglun Gong', 'Yukun Shi', 'Zhentao Huang'] | 2023-02-14 | null | null | null | null | ['stereo-matching-1'] | ['computer-vision'] | [ 4.01522368e-01 -3.24490815e-01 4.00886796e-02 -2.19839424e-01
-4.97905284e-01 -4.58678097e-01 2.28921890e-01 1.44325584e-01
-1.33674026e-01 7.65079618e-01 -3.20481732e-02 9.04752091e-02
-2.74713755e-01 -1.10451555e+00 -4.63696837e-01 -1.05945349e+00
1.97500736e-01 5.35583615e-01 8.78741503e-01 -8.88284892e-02
5.50028384e-01 5.83320320e-01 -2.17324972e+00 4.93058711e-01
1.10264421e+00 7.80594945e-01 6.71581388e-01 4.89711761e-01
5.33451065e-02 4.78252530e-01 -2.76780784e-01 -2.86978573e-01
4.61740702e-01 -2.41220951e-01 -3.38092744e-01 5.17262816e-01
8.45428407e-01 -1.36892218e-02 2.57292569e-01 1.07679057e+00
5.42530298e-01 1.83033571e-01 2.25778520e-01 -1.02055705e+00
-1.20095782e-01 -4.04700115e-02 -1.11738038e+00 2.13831604e-01
4.69960570e-01 2.21701078e-02 8.69753778e-01 -8.46507668e-01
1.00945938e+00 1.12174118e+00 2.96320558e-01 1.92017406e-01
-1.26664269e+00 -4.51943755e-01 5.21678865e-01 3.27496946e-01
-1.35276794e+00 -2.37532824e-01 1.09443605e+00 -1.56553343e-01
7.07015991e-01 6.44998670e-01 9.56972957e-01 3.14973772e-01
3.17450643e-01 5.22946596e-01 1.49022293e+00 -4.77645516e-01
2.72498578e-01 1.09168246e-01 -1.88871145e-01 8.52761626e-01
2.50525683e-01 4.14788395e-01 -8.72307122e-01 -4.18653607e-01
7.08345532e-01 -4.83235940e-02 -5.95062494e-01 -1.08952332e+00
-1.06749809e+00 8.95761907e-01 5.19801497e-01 2.34008163e-01
-6.94891989e-01 -2.42471606e-01 -1.44611493e-01 6.90520927e-02
5.50368726e-01 3.81386466e-02 -4.71596122e-01 3.02366614e-01
-9.10506189e-01 1.37839034e-01 3.17964256e-01 5.50709367e-01
1.11748290e+00 -2.79004663e-01 -2.21536662e-02 1.02292323e+00
2.36879215e-01 3.91598672e-01 -1.49767429e-01 -1.20873106e+00
4.80197966e-01 8.69163632e-01 3.94591466e-02 -1.13837171e+00
-2.38837987e-01 -5.55351675e-01 -6.23875201e-01 8.16904426e-01
1.89205721e-01 1.69731632e-01 -8.80217850e-01 1.21281803e+00
1.03829408e+00 8.89646560e-02 3.00207231e-02 9.29345489e-01
6.41644120e-01 4.64961588e-01 -5.21972239e-01 -4.12421107e-01
9.74731326e-01 -1.20704973e+00 -3.39934051e-01 -3.47337425e-01
-8.57417956e-02 -1.01636088e+00 5.40047944e-01 6.57148421e-01
-1.10794973e+00 -5.19377470e-01 -9.46899772e-01 2.43716612e-01
-2.69614011e-01 -9.66491327e-02 5.92170000e-01 6.26304686e-01
-1.18983436e+00 4.38108563e-01 -4.35030609e-01 -2.55053878e-01
2.91223556e-01 3.74200344e-01 -3.83885562e-01 -6.53900266e-01
-4.70108539e-01 6.02037132e-01 1.94968119e-01 7.50338733e-02
-5.79915166e-01 -5.23350596e-01 -8.01089823e-01 -1.15749918e-01
5.44088066e-01 -8.80747080e-01 3.04619670e-01 -9.70383108e-01
-1.09057510e+00 9.23298478e-01 -4.15199637e-01 -6.88052177e-02
5.47476590e-01 1.94177642e-01 -2.23721355e-01 2.30772495e-01
1.15600146e-01 8.71228158e-01 9.42514420e-01 -1.91272652e+00
-1.19528854e+00 -5.74090242e-01 2.76877373e-01 6.52032971e-01
3.88526567e-03 -2.74836510e-01 -1.05279243e+00 -4.80889320e-01
6.47459269e-01 -8.27510953e-01 -3.16711128e-01 2.27484673e-01
-4.71238866e-02 1.01489067e-01 7.85737574e-01 -5.36063612e-01
1.17539430e+00 -1.83002794e+00 5.42993009e-01 6.00866675e-01
7.69068673e-02 -9.10422653e-02 -1.72144786e-01 2.67176151e-01
1.99288338e-01 -4.83929932e-01 -2.20183074e-01 -2.75262564e-01
-5.83602369e-01 4.16087329e-01 3.37962002e-01 5.30017495e-01
-1.74223334e-01 1.80177540e-01 -7.99121439e-01 -7.98077524e-01
6.77276969e-01 5.27467430e-01 -7.58834124e-01 8.04954618e-02
3.91659047e-03 7.38357127e-01 -2.77527601e-01 9.64022756e-01
1.17836547e+00 -1.34639174e-01 2.17444971e-01 -2.22069517e-01
-3.16789925e-01 -4.11439091e-01 -1.76375759e+00 1.83500874e+00
-6.47390246e-01 2.45232448e-01 3.21568012e-01 -4.83348489e-01
7.92396128e-01 -1.89004019e-01 6.14225626e-01 -6.41428411e-01
-1.14178747e-01 1.07584849e-01 -1.27628073e-01 -2.62908071e-01
4.11853373e-01 4.13565457e-01 3.86338592e-01 2.64094740e-01
-4.45936799e-01 -1.69720814e-01 3.52059722e-01 -1.20332390e-01
6.28073335e-01 3.12011421e-01 2.94470280e-01 -1.40183315e-01
9.73158479e-01 1.51770160e-01 8.48576188e-01 6.11682892e-01
-4.03672531e-02 8.71273458e-01 -2.62834519e-01 -6.93980575e-01
-7.21599817e-01 -9.84007478e-01 -2.26967081e-01 8.34509313e-01
7.33305097e-01 -2.41205856e-01 -5.27802646e-01 -4.82198089e-01
6.07595779e-03 5.54981589e-01 -6.59130335e-01 4.31967556e-01
-6.74709022e-01 -6.81176841e-01 -7.54887044e-01 1.41721547e-01
5.57439566e-01 -8.27525437e-01 -9.99749720e-01 3.02339673e-01
-3.97634804e-01 -8.39143693e-01 -3.25238347e-01 5.39971031e-02
-1.01571631e+00 -1.32179999e+00 -6.62589431e-01 -6.83490098e-01
9.72674727e-01 7.72309184e-01 1.21664643e+00 4.68791008e-01
-3.29315126e-01 3.20955217e-01 -2.09996045e-01 -4.49449476e-03
-1.97276503e-01 -3.00974101e-01 -5.72992980e-01 2.75187194e-01
4.15921882e-02 -6.04549646e-01 -7.39462614e-01 6.56476557e-01
-6.74852073e-01 3.22310418e-01 4.05076921e-01 9.52956557e-01
1.19182193e+00 3.05100083e-01 -2.45978177e-01 -6.14834726e-01
4.60692458e-02 -2.52370052e-02 -1.10746193e+00 7.47588158e-01
-5.93909740e-01 -1.02715783e-01 1.45821944e-01 -2.07359537e-01
-1.20597637e+00 4.27531213e-01 3.61899108e-01 -5.87913096e-01
-1.31732598e-01 1.38449252e-01 -9.43851843e-02 -6.05207801e-01
6.07447401e-02 2.55094022e-01 -2.89539218e-01 -2.86620617e-01
1.68737650e-01 4.03098553e-01 3.29153568e-01 -2.81320989e-01
7.50112474e-01 8.99524450e-01 1.36780500e-01 -6.48912787e-01
-5.25692940e-01 -6.38025284e-01 -7.28142381e-01 -8.40763986e-01
9.83927667e-01 -8.96364450e-01 -4.39562201e-01 3.27382684e-01
-1.04275346e+00 1.50720626e-01 7.19190687e-02 2.82890975e-01
-3.13520283e-01 4.20832038e-01 1.28297344e-01 -9.84596133e-01
-2.43931875e-01 -1.39620411e+00 1.26659596e+00 3.23307782e-01
1.68731213e-01 -9.26078141e-01 9.30664390e-02 7.94468522e-01
2.29989544e-01 3.44612300e-01 7.62201428e-01 2.96803117e-01
-1.15213692e+00 6.63339272e-02 -1.54753716e-03 -1.25697604e-03
9.41331238e-02 1.51922889e-02 -9.99846876e-01 -3.69018257e-01
-8.83799270e-02 1.06488653e-01 8.07609975e-01 6.81511760e-01
9.02125478e-01 1.60582855e-01 -5.88197291e-01 7.20052958e-01
1.95814836e+00 2.66912907e-01 5.84341228e-01 7.09929705e-01
5.14785349e-01 9.36293721e-01 9.87083137e-01 5.20957887e-01
4.45555091e-01 9.19537127e-01 8.31784189e-01 -3.16296875e-01
-1.59478217e-01 5.88088892e-02 -1.98541060e-01 2.93034732e-01
-1.91062450e-01 -4.86218750e-01 -6.31799161e-01 5.16036570e-01
-1.85910940e+00 -9.12488699e-01 -3.82904708e-01 2.48570418e+00
3.32624793e-01 1.95382051e-02 -1.76602289e-01 1.45108148e-01
8.99204552e-01 2.75738984e-01 -4.12664771e-01 -2.10636348e-01
-4.68365967e-01 -1.35065615e-03 5.34417450e-01 7.26384223e-01
-1.04461861e+00 7.21551239e-01 5.20085049e+00 7.10401118e-01
-9.13102686e-01 1.09890372e-01 6.84570611e-01 -2.77674377e-01
-4.57824647e-01 1.48083046e-01 -6.35573149e-01 3.17322880e-01
-1.95771694e-01 3.48570377e-01 7.26385653e-01 5.33368289e-01
1.12508699e-01 -9.88266289e-01 -7.01117456e-01 1.28259575e+00
3.91794145e-01 -1.40902328e+00 -8.86699036e-02 2.09110036e-01
1.42464089e+00 4.53198031e-02 -1.72884136e-01 -6.39361739e-01
2.17010334e-01 -4.46728826e-01 6.52611196e-01 3.20021272e-01
2.41506353e-01 -9.13732588e-01 5.91289997e-01 2.67722160e-01
-1.41417015e+00 -2.93945879e-01 -3.89257133e-01 3.94153953e-01
2.65277356e-01 6.33111298e-01 -3.02089661e-01 9.68370080e-01
9.92146611e-01 4.84491289e-01 -6.63311720e-01 1.31321478e+00
-1.33819014e-01 -1.74498662e-01 -3.89508098e-01 2.75158018e-01
1.10994384e-01 -5.44285715e-01 7.65447915e-01 4.39560443e-01
2.29461297e-01 -1.29719973e-01 3.60993534e-01 5.18384933e-01
4.84497875e-01 1.00586265e-01 -5.54072618e-01 9.05930877e-01
4.85886425e-01 1.28418088e+00 -9.79505837e-01 -1.90182686e-01
-5.81004083e-01 1.13085091e+00 3.81380141e-01 2.69659013e-01
-6.86311066e-01 1.91777706e-01 3.91102731e-01 1.46076620e-01
8.21583807e-01 -7.20202029e-02 -3.35240573e-01 -1.03591168e+00
2.45987430e-01 -8.06021333e-01 7.36724496e-01 -9.27214622e-01
-1.01348639e+00 7.22339928e-01 -1.82054546e-02 -1.58433914e+00
-5.91127761e-03 -6.42366484e-02 -5.16377091e-01 9.66224968e-01
-1.97411811e+00 -1.11072648e+00 -6.58347368e-01 7.16595411e-01
7.84171462e-01 -8.81359130e-02 5.87715983e-01 1.77799717e-01
-1.85910210e-01 -5.43927513e-02 1.15492105e-01 -5.90298235e-01
4.14987028e-01 -9.65331912e-01 -1.58760846e-01 9.94387686e-01
2.38110036e-01 9.84589607e-02 5.71946740e-01 -6.75351381e-01
-1.26296675e+00 -7.57675052e-01 7.69753397e-01 -4.03018929e-02
5.16168252e-02 -5.29792644e-02 -6.01850569e-01 -1.54175900e-03
4.45377648e-01 -1.76786892e-02 4.36146230e-01 -2.01510638e-01
8.79522264e-02 -4.12110120e-01 -1.26672721e+00 4.33459610e-01
1.07185948e+00 -2.04706956e-02 -1.28022164e-01 3.11371744e-01
2.83544473e-02 -5.33260763e-01 -6.44854903e-01 4.88679379e-01
5.11353552e-01 -1.59679413e+00 1.27253342e+00 6.56291097e-02
9.86368805e-02 -6.73602402e-01 -3.32508683e-01 -1.32543206e+00
-4.11732972e-01 -3.12264383e-01 2.20987096e-01 1.05938661e+00
1.96062192e-01 -6.68671966e-01 8.94424856e-01 3.56033593e-01
1.03818223e-01 -7.31866539e-01 -1.25050294e+00 -5.52202523e-01
-6.07072949e-01 9.74975992e-03 5.50390661e-01 7.38478124e-01
-7.00598657e-01 -4.36610915e-02 -2.84963965e-01 7.47546315e-01
1.22332990e+00 9.52313185e-01 8.57312500e-01 -1.42812479e+00
-5.17474890e-01 -2.01419756e-01 -3.17151606e-01 -6.26997054e-01
-1.92929804e-01 -5.36492765e-01 -7.11789876e-02 -1.68700719e+00
3.09809983e-01 -6.21813834e-01 -4.88757454e-02 5.27114943e-02
-3.51354390e-01 4.59969670e-01 2.00295597e-01 1.44510806e-01
-5.22355497e-01 2.94509053e-01 1.38077545e+00 -1.46885365e-01
-2.97306329e-01 -6.70522079e-02 -1.79015636e-01 7.65869021e-01
5.68832099e-01 -4.81222689e-01 -4.35306698e-01 -6.55447960e-01
-2.04005488e-03 1.29997477e-01 3.04476708e-01 -1.18266499e+00
3.35068256e-01 -2.45890692e-01 5.53879738e-01 -1.17892480e+00
6.60224319e-01 -1.28225708e+00 5.48280299e-01 5.39687395e-01
1.70699149e-01 2.69900531e-01 -6.29625991e-02 7.13301301e-01
-2.21925288e-01 -2.06806481e-01 8.75494897e-01 -3.29210609e-01
-8.94608915e-01 1.65685713e-01 -2.92238463e-02 -2.49322981e-01
1.37413335e+00 -8.00906897e-01 -1.19150072e-01 -1.86390013e-01
-6.30873978e-01 4.45721477e-01 1.02732885e+00 2.90443152e-01
9.29651558e-01 -9.65188086e-01 -6.98897898e-01 4.63879377e-01
3.35302472e-01 -2.64681410e-02 5.71142197e-01 6.24826014e-01
-5.65721154e-01 1.08140662e-01 -3.37659389e-01 -9.75648761e-01
-1.99282265e+00 5.31269491e-01 3.48546356e-01 -2.48888552e-01
-4.66651171e-01 8.85470629e-01 2.64547795e-01 -2.20103532e-01
3.07350129e-01 4.62531298e-02 -4.46789593e-01 1.96322072e-02
3.34822536e-01 5.33178866e-01 1.67278275e-01 -6.92035377e-01
-4.09287721e-01 1.19074583e+00 4.88422439e-02 -1.03718266e-01
1.48256004e+00 -4.99206066e-01 -2.42429078e-01 2.54918952e-02
8.90513420e-01 3.36857021e-01 -1.41907179e+00 -1.82442591e-01
-4.51070786e-01 -1.19043458e+00 5.13770521e-01 -8.79209518e-01
-1.60219550e+00 4.84991252e-01 1.06422484e+00 -1.21711731e-01
1.53579032e+00 -1.03230871e-01 5.12688100e-01 -1.82048544e-01
9.25817907e-01 -1.11838794e+00 -8.30607116e-02 -1.35383978e-01
9.09248233e-01 -1.23898685e+00 4.15589273e-01 -9.67216372e-01
-5.45053542e-01 1.21462107e+00 7.63960481e-01 2.13965565e-01
5.09651184e-01 2.06064552e-01 8.70250389e-02 -3.48832071e-01
-6.78324759e-01 -2.62242377e-01 2.34009758e-01 9.05704081e-01
-5.39643355e-02 -2.62737632e-01 -3.89247596e-01 -5.11183619e-01
3.31044853e-01 -3.36366445e-01 2.83539951e-01 1.19988525e+00
-4.77889866e-01 -1.26549911e+00 -8.60089242e-01 2.51913190e-01
-5.08623868e-02 -1.13252096e-01 -2.60360330e-01 4.72296655e-01
4.57805276e-01 1.29341650e+00 1.48478448e-01 -1.11443944e-01
2.06961960e-01 -4.89031434e-01 7.04272628e-01 -4.31421727e-01
-7.21688867e-01 5.07954895e-01 1.57975331e-01 -7.54165232e-01
-9.75046873e-01 -9.36983824e-01 -8.01877439e-01 -1.42235279e-01
-6.02704942e-01 -1.42796692e-02 5.89483500e-01 4.08229470e-01
3.93540591e-01 3.73456061e-01 9.91591811e-01 -1.02437139e+00
4.91460003e-02 -3.03730309e-01 -3.59144479e-01 1.55248225e-01
6.87662093e-03 -9.10162926e-01 -2.64379978e-01 -1.68088108e-01] | [9.309090614318848, -2.4804604053497314] |
39ea6b0b-1d83-4d2e-8029-79b439488ece | convolutional-networks-are-inherently | null | null | https://openreview.net/forum?id=5oF-Z7Uk0tH | https://openreview.net/pdf?id=5oF-Z7Uk0tH | Convolutional Networks are Inherently Foveated | When convolutional layers apply no padding, central pixels have more ways to contribute to the convolution than peripheral pixels. Such discrepancy grows exponentially with the number of layers, leading to implicit foveation of the input pixels. We show that this discrepancy can persist even when padding is applied. In particular, with the commonly-used zero-padding, foveation effects are significantly reduced but not eliminated. We explore how different aspects of convolution arithmetic impact the spread and magnitude of foveation, and elaborate on which alternative padding techniques can mitigate it. Finally, we compare our findings with foveation in human vision, concluding that both effects are likely of the same nature and have similar implications.
| ['Orion Reblitz-Richardson', 'David Adkins', 'Narine Kokhlikyan', 'Vivek Miglani', 'Bilal Alsallakh'] | 2021-10-12 | null | null | null | neurips-workshop-svrhm-2021-12 | ['foveation'] | ['computer-vision'] | [ 2.86149681e-01 1.32749185e-01 1.99704021e-01 -4.73685078e-02
5.61117567e-02 -8.12986016e-01 4.88143981e-01 6.15541749e-02
-8.90041709e-01 4.39997673e-01 3.68802637e-01 -6.61986887e-01
1.53857514e-01 -8.15222919e-01 -8.36595833e-01 -5.17499924e-01
-5.97235933e-02 -8.27415168e-01 3.99249345e-01 -1.57783106e-01
4.79904741e-01 3.06147486e-01 -1.61270082e+00 6.20660126e-01
6.09971225e-01 6.67405367e-01 2.07142800e-01 7.62138903e-01
4.04430255e-02 6.78528190e-01 -7.89868116e-01 -3.68896961e-01
6.85380101e-01 -1.48969516e-01 -5.21654665e-01 -1.14942029e-01
9.32159007e-01 -5.17598510e-01 -4.04494405e-01 1.17283440e+00
3.44025999e-01 7.80083910e-02 4.85331208e-01 -8.83219957e-01
-7.69826472e-01 5.51899910e-01 -8.69538307e-01 5.14923036e-01
-1.54413134e-01 4.48055983e-01 7.00243413e-01 -8.76995921e-01
4.28525358e-01 1.35882032e+00 8.65834653e-01 3.47302049e-01
-1.39811659e+00 -3.29191953e-01 1.94929793e-01 -2.28747219e-01
-1.35044444e+00 -5.52090287e-01 1.90253988e-01 -4.18500572e-01
1.17976356e+00 1.52966514e-01 6.90746009e-01 3.87139976e-01
3.59667033e-01 3.20503175e-01 1.31832659e+00 -6.41995907e-01
-7.84632340e-02 3.56896669e-02 2.65419960e-01 4.44903672e-01
6.75529838e-01 1.49687812e-01 -3.15964729e-01 -6.30470284e-04
1.33792090e+00 -2.25066721e-01 -5.22749186e-01 2.01416269e-01
-1.10413718e+00 4.60014224e-01 6.06527925e-01 1.29319921e-01
-5.43059886e-01 4.99254972e-01 1.52359724e-01 1.89853758e-01
1.01011142e-01 8.14256251e-01 -3.47278982e-01 1.10924453e-01
-8.00569832e-01 3.91949028e-01 5.11898875e-01 6.55479491e-01
7.67573893e-01 -1.74313858e-01 -3.30627322e-01 5.58606088e-01
1.10394478e-01 1.87355429e-01 2.17006519e-01 -1.18820930e+00
4.76125866e-01 3.43143672e-01 3.57844204e-01 -8.44963610e-01
-3.48316312e-01 -4.46102768e-01 -2.72156715e-01 1.08796251e+00
1.24745464e+00 -5.78739643e-01 -8.30325127e-01 1.86210501e+00
-4.12959546e-01 -1.75886348e-01 -1.11531518e-01 8.77153218e-01
2.86967784e-01 3.06714445e-01 3.85289580e-01 1.28433555e-01
1.70333433e+00 -7.45730221e-01 -6.55079842e-01 -6.81184411e-01
3.65504116e-01 -1.06585145e+00 1.35257101e+00 3.21769983e-01
-1.35383701e+00 -5.36755979e-01 -1.35664606e+00 -3.71786416e-01
-3.10588926e-01 4.82096337e-02 8.33144009e-01 7.92480111e-01
-1.30648136e+00 8.31944227e-01 -6.50786459e-01 -1.07341863e-01
4.81975049e-01 2.87167758e-01 -1.75708008e-03 2.07674697e-01
-9.42088366e-01 1.21454561e+00 2.65817076e-01 6.02667592e-03
2.16314178e-02 -8.53482842e-01 -4.99688089e-01 3.09743285e-01
3.82502526e-02 -6.25885546e-01 1.46362162e+00 -1.24855733e+00
-7.92101026e-01 8.92278075e-01 -3.80550116e-01 -6.67068601e-01
3.75612199e-01 -3.07946265e-01 -2.47964144e-01 2.80572325e-02
-2.68092990e-01 1.03674483e+00 8.15083921e-01 -9.71827209e-01
-4.54662383e-01 -1.95110679e-01 4.25424993e-01 3.00705731e-01
-2.19394550e-01 -1.15218041e-02 -1.77838832e-01 -4.84378308e-01
1.27184778e-01 -5.94183385e-01 -2.16619596e-01 5.64298034e-01
4.92392555e-02 -4.36070282e-03 4.94367689e-01 -4.37404901e-01
1.41369140e+00 -2.50623345e+00 -6.92201436e-01 -4.36407077e-04
5.22732496e-01 3.43381673e-01 -2.06067488e-01 1.12467989e-01
-3.32699299e-01 2.68101454e-01 3.80869173e-02 -7.60573475e-03
-4.91588898e-02 -3.66703495e-02 -3.66370052e-01 4.41589504e-01
5.13104439e-01 9.67438161e-01 -6.36363506e-01 -1.17630817e-01
1.61178485e-01 5.37638366e-01 -6.28722847e-01 -5.55670500e-01
7.28584677e-02 -3.06954592e-01 2.23174602e-01 2.55122960e-01
1.07658601e+00 -3.09432417e-01 -3.81067693e-02 -3.52358162e-01
-6.89828038e-01 6.65731788e-01 -1.06135643e+00 1.11015606e+00
-6.39601648e-02 1.31698179e+00 9.24247131e-02 -2.59181529e-01
4.55545843e-01 6.18101358e-02 -2.59825796e-01 -7.83428252e-01
2.17101783e-01 2.40990072e-01 9.70080078e-01 -1.78153276e-01
8.79997671e-01 -1.66198149e-01 5.72998703e-01 5.21668792e-01
-3.27622682e-01 3.40575911e-02 1.91758454e-01 5.88579737e-02
9.93984044e-01 -9.35218632e-02 3.98008853e-01 -5.98595679e-01
-1.77403510e-01 -2.78566480e-02 1.10807642e-01 9.07215953e-01
-4.64986056e-01 6.16635501e-01 9.08000886e-01 -2.54991353e-01
-1.24132216e+00 -9.15735066e-01 -4.08445776e-01 8.97816837e-01
4.70856391e-02 -3.35128039e-01 -1.14277077e+00 -1.88511342e-01
6.58140611e-03 6.60100579e-01 -7.21994400e-01 -3.97465751e-02
-4.14139718e-01 -7.63427377e-01 7.16777086e-01 7.39108562e-01
4.93041873e-01 -9.97467697e-01 -1.34101212e+00 1.83731280e-02
8.33996385e-02 -8.96763980e-01 -5.53667784e-01 3.22877705e-01
-1.07220268e+00 -9.85757351e-01 -8.28724980e-01 -4.79971409e-01
9.11282718e-01 7.81168044e-01 9.96144593e-01 3.46271962e-01
-2.79283226e-01 6.85455650e-02 1.03446282e-01 -6.49682581e-01
-6.61637262e-02 -2.64970750e-01 -3.28757972e-01 -5.17235994e-01
6.41764522e-01 -4.56157625e-01 -1.03215563e+00 1.00359797e-01
-8.07440877e-01 1.62505031e-01 7.79804230e-01 4.16428745e-01
4.81367260e-02 4.62869704e-02 2.15382174e-01 -8.09913635e-01
9.27587986e-01 -1.09978527e-01 -5.52142203e-01 -2.76667684e-01
-5.98414004e-01 9.59271714e-02 4.12994504e-01 -6.92937195e-01
-9.22356963e-01 -3.19290489e-01 2.59557962e-01 -2.37963602e-01
-8.41727629e-02 5.79308420e-02 3.69391292e-01 -2.82119989e-01
1.25069499e+00 -2.71745771e-01 1.21682763e-01 -1.38584018e-01
3.89761329e-01 1.83948562e-01 5.85555077e-01 -2.29443058e-01
4.64595020e-01 6.19296134e-01 -2.61041820e-01 -7.76484251e-01
-3.47234488e-01 5.80543652e-02 -4.16134715e-01 6.69931620e-02
8.96323264e-01 -8.26933026e-01 -7.34575212e-01 5.12960911e-01
-1.51727223e+00 -6.59590662e-01 -3.24127257e-01 6.67911828e-01
-1.80125043e-01 4.77105677e-02 -6.71907067e-01 -7.29546130e-01
7.60253742e-02 -1.12200665e+00 2.33346581e-01 5.22742629e-01
-5.90004325e-01 -7.95958817e-01 -4.29136902e-01 -3.67652386e-01
5.63730478e-01 -1.09671786e-01 9.32835698e-01 -8.81040692e-02
-5.78931451e-01 1.50464147e-01 -8.05175781e-01 3.33235085e-01
2.08021924e-01 1.45748436e-01 -1.33325815e+00 9.93618593e-02
-4.74579707e-02 1.80722833e-01 1.01154435e+00 7.38553822e-01
8.27915549e-01 -3.42128187e-01 -1.99882671e-01 4.58328754e-01
1.44871271e+00 5.18001132e-02 1.16578567e+00 3.87585282e-01
1.90091908e-01 7.75100350e-01 2.63528645e-01 2.02724457e-01
1.48852495e-02 2.09745884e-01 2.69093215e-01 -4.50135231e-01
-5.13329685e-01 -1.15096465e-01 2.54650377e-02 -2.58524064e-02
-3.33023906e-01 -1.21079691e-01 -7.61151969e-01 4.93088335e-01
-1.45425284e+00 -1.04975784e+00 -5.25777042e-01 2.39802885e+00
8.83362710e-01 4.37056541e-01 -1.24501199e-01 1.18601687e-01
7.67661631e-01 6.60418496e-02 -4.26737964e-01 -6.31032109e-01
-2.45725110e-01 4.12580788e-01 1.14861619e+00 5.56853473e-01
-8.00476372e-01 7.74530768e-01 8.75229549e+00 3.05499017e-01
-1.26417470e+00 -1.96378946e-01 6.42742753e-01 -2.02068105e-01
-2.52916634e-01 4.21272777e-02 -5.44853330e-01 4.13946360e-01
6.15146697e-01 -5.69103612e-03 4.41292524e-01 4.32700604e-01
3.05887669e-01 -6.36068523e-01 -8.91805708e-01 6.58265293e-01
-4.12230492e-01 -1.13232827e+00 -3.13368440e-01 2.40403906e-01
7.56119609e-01 -7.06603453e-02 4.37678128e-01 -1.13900803e-01
1.44198060e-01 -1.24261069e+00 7.64069557e-01 4.56351668e-01
6.87841117e-01 -5.16320288e-01 3.34464192e-01 -5.20568453e-02
-8.43543112e-01 3.54045816e-02 -5.51661193e-01 -9.16451633e-01
-7.10106045e-02 5.37407339e-01 -7.71196127e-01 -4.39443231e-01
4.48945403e-01 -1.17165163e-01 -6.72423065e-01 1.26306427e+00
-3.97635043e-01 4.59842175e-01 -3.53110731e-01 1.51681721e-01
2.29849964e-01 4.73516844e-02 1.79642558e-01 1.09733582e+00
2.73272485e-01 2.30483159e-01 -8.28644931e-01 1.27275515e+00
2.23695084e-01 -2.29683369e-01 -5.70442021e-01 -1.49676427e-01
8.06156874e-01 1.07118988e+00 -8.09435070e-01 -3.66844922e-01
-1.07117534e+00 1.06440508e+00 2.94040948e-01 8.23663235e-01
-6.86882257e-01 -6.99922204e-01 9.88047838e-01 4.65836346e-01
4.27310705e-01 -4.39549685e-01 -1.00820792e+00 -6.53611541e-01
8.86818767e-03 -6.66053832e-01 -2.10756287e-01 -9.44316685e-01
-7.34765410e-01 2.03591660e-01 -2.01798990e-01 -7.57839441e-01
3.89956683e-02 -1.04022896e+00 -8.72876823e-01 1.32158387e+00
-1.49833608e+00 -4.53135103e-01 -1.62907034e-01 5.38679302e-01
1.21414706e-01 4.46924955e-01 5.91917157e-01 2.04924494e-01
-7.08756596e-02 7.86980569e-01 -4.43922192e-01 2.31282473e-01
9.15249765e-01 -1.12999415e+00 6.49621010e-01 1.05907416e+00
-1.07589044e-01 1.31439006e+00 7.62484133e-01 -2.68905938e-01
-1.02365720e+00 -6.27878189e-01 8.50802541e-01 -3.56064409e-01
4.02240813e-01 -1.52779415e-01 -1.09048021e+00 8.08091402e-01
5.26489854e-01 -3.89019102e-02 5.01385689e-01 1.84929371e-01
-6.08870447e-01 3.85148972e-01 -9.57073033e-01 1.13965404e+00
9.06925678e-01 -4.68999237e-01 -5.67929149e-01 -2.47221097e-01
7.07064271e-01 -3.71029854e-01 -3.87650251e-01 -3.68842706e-02
7.73712039e-01 -1.39509690e+00 8.27612936e-01 -2.42170200e-01
4.79319811e-01 -3.79573256e-01 -3.14680114e-02 -1.12192833e+00
-5.43489754e-01 -5.17903805e-01 2.00264469e-01 7.67300189e-01
6.34966075e-01 -1.01630759e+00 4.53846365e-01 9.81851339e-01
-2.22836971e-01 -3.84944290e-01 -4.47129548e-01 -4.21645671e-01
4.09689248e-01 -3.82708132e-01 4.26615894e-01 6.25958622e-01
2.21137062e-01 -6.21935166e-02 2.36001834e-01 1.00616693e-01
3.30721617e-01 -4.76158261e-01 4.72318500e-01 -9.17143643e-01
-3.71533513e-01 -8.96971822e-01 -3.27401698e-01 -1.36035204e+00
-5.25727153e-01 -3.42717171e-01 2.92668790e-02 -1.26773953e+00
-2.34160535e-02 -3.46848994e-01 -1.83096692e-01 6.13527536e-01
-3.96975100e-01 5.11968076e-01 4.39572304e-01 6.12357184e-02
9.10049155e-02 -3.88778120e-01 1.62396514e+00 3.62278104e-01
-1.93652183e-01 -1.14570029e-01 -1.27541590e+00 9.78180170e-01
8.80530894e-01 4.00783606e-02 -4.08213139e-01 -8.23402226e-01
3.77475947e-01 -3.81796688e-01 8.18111777e-01 -1.03455079e+00
2.21429318e-01 -5.12320548e-02 6.77203119e-01 -3.54132086e-01
2.20682532e-01 -4.65056032e-01 -3.29804927e-01 5.42024732e-01
-2.67602384e-01 2.57032126e-01 9.09142971e-01 6.08528331e-02
1.44049108e-01 -5.59575260e-01 8.47381890e-01 -1.36303812e-01
-7.92121112e-01 -3.02537054e-01 -8.83611679e-01 -7.27699399e-02
7.71922350e-01 -4.73997802e-01 -7.75964916e-01 -2.11245611e-01
-5.04007936e-01 -2.28957579e-01 6.91239476e-01 4.06521298e-02
3.12818795e-01 -1.12304211e+00 -5.08202493e-01 2.82789946e-01
-3.16485196e-01 -2.67861605e-01 1.38073057e-01 8.86379480e-01
-4.85359371e-01 3.43073428e-01 -3.27318788e-01 -1.02487154e-01
-1.09601188e+00 4.20571208e-01 4.19903785e-01 3.64399046e-01
-4.91767824e-01 9.80111063e-01 6.32683992e-01 5.43064713e-01
1.59937352e-01 -7.21213341e-01 1.03460796e-01 3.42417732e-02
9.47189748e-01 4.31926399e-01 -1.87390283e-01 -1.15066078e-02
-2.54687816e-01 4.33329672e-01 -1.90909848e-01 -4.43498135e-01
5.24933100e-01 -3.65294553e-02 -7.46510550e-02 3.37796032e-01
8.28921735e-01 3.12999159e-01 -1.73650110e+00 -1.63917243e-02
-5.90522885e-01 -5.47752261e-01 -1.94187686e-02 -7.38759458e-01
-7.86796987e-01 1.16684842e+00 5.47069490e-01 2.58984178e-01
1.06008112e+00 -1.09069817e-01 3.43833745e-01 7.34832734e-02
-2.59877127e-02 -8.02011013e-01 -1.23930030e-01 5.56192100e-01
6.33684576e-01 -7.68858135e-01 -3.62499431e-02 -5.89741886e-01
-4.29060698e-01 1.08891058e+00 8.45847726e-01 -4.62979525e-01
3.96297127e-01 7.05605984e-01 9.21858102e-02 -5.37560433e-02
-7.40417182e-01 -3.47400010e-01 1.02349348e-01 8.72521698e-01
9.56495106e-01 -6.94641024e-02 -7.79240012e-01 4.49178606e-01
-5.20609200e-01 -1.11517116e-01 6.22305572e-01 8.37639213e-01
-7.00046182e-01 -8.09436023e-01 -6.59363925e-01 4.27320868e-01
-4.47576851e-01 -7.72666693e-01 -1.53740287e-01 9.89643455e-01
4.10345167e-01 8.50512147e-01 6.81018472e-01 -2.48455666e-02
1.35134220e-01 1.85091794e-01 9.28024292e-01 -5.42502940e-01
-5.49207151e-01 1.45851851e-01 -1.80327788e-01 -3.31127018e-01
-1.91561449e-02 -4.90570724e-01 -1.51082540e+00 -5.52068293e-01
-3.72812569e-01 -3.92401099e-01 3.94024134e-01 5.47583282e-01
4.70550716e-01 8.06979954e-01 -8.60290378e-02 -5.99960327e-01
-5.37590444e-01 -1.05842996e+00 -5.13525188e-01 1.79734752e-01
4.49696273e-01 -4.29832965e-01 -4.46287572e-01 1.46247566e-01] | [9.88267993927002, 2.1617813110351562] |
934d2ede-a255-4b66-adff-9b49c20a73da | vakyansh-asr-toolkit-for-low-resource-indic | 2203.16512 | null | https://arxiv.org/abs/2203.16512v2 | https://arxiv.org/pdf/2203.16512v2.pdf | Vakyansh: ASR Toolkit for Low Resource Indic languages | We present Vakyansh, an end to end toolkit for Speech Recognition in Indic languages. India is home to almost 121 languages and around 125 crore speakers. Yet most of the languages are low resource in terms of data and pretrained models. Through Vakyansh, we introduce automatic data pipelines for data creation, model training, model evaluation and deployment. We create 14,000 hours of speech data in 23 Indic languages and train wav2vec 2.0 based pretrained models. These pretrained models are then finetuned to create state of the art speech recognition models for 18 Indic languages which are followed by language models and punctuation restoration models. We open source all these resources with a mission that this will inspire the speech community to develop speech first applications using our ASR models in Indic languages. | ['Vivek Raghavan', 'Rishabh Gaur', 'Ankur Dhuriya', 'Neeraj Chhimwal', 'Priyanshi Shah', 'Anirudh Gupta', 'Harveen Singh Chadha'] | 2022-03-30 | null | null | null | null | ['punctuation-restoration'] | ['natural-language-processing'] | [-1.30650714e-01 -1.50371520e-02 6.07555360e-02 -7.20431089e-01
-1.05768728e+00 -6.84328377e-01 5.15162349e-01 -4.54874843e-01
-3.57044667e-01 1.85506701e-01 7.81286359e-01 -1.02161050e+00
4.54351246e-01 -3.39951783e-01 -2.45852441e-01 -1.51154563e-01
-6.50709271e-02 8.28425467e-01 -2.54724920e-01 -6.09793007e-01
-2.11475804e-01 3.31012189e-01 -9.17338133e-01 3.53370041e-01
5.88711321e-01 4.23222035e-01 2.92714983e-01 1.29488420e+00
-2.69459546e-01 8.46699297e-01 -8.18704963e-01 -3.36690336e-01
1.13730997e-01 -1.23980634e-01 -1.10066557e+00 4.39925194e-02
5.01046062e-01 -3.30472797e-01 -7.37324476e-01 5.58679640e-01
8.19932282e-01 2.58720547e-01 4.86384630e-02 -6.48357391e-01
-1.08888626e+00 1.37879968e+00 2.00223446e-01 5.08436501e-01
3.49862315e-02 6.15758337e-02 9.32938337e-01 -1.35118771e+00
5.66524982e-01 1.42794979e+00 6.33204460e-01 8.75491560e-01
-8.39744091e-01 -5.98340631e-01 -1.03075594e-01 2.14822680e-01
-1.31022608e+00 -1.42345762e+00 6.10147834e-01 -1.71598941e-01
1.68539155e+00 4.83180732e-01 2.69841433e-01 1.24982166e+00
-6.85998380e-01 1.09582925e+00 8.74558568e-01 -6.62171841e-01
-1.13602437e-01 1.64530233e-01 5.39739370e-01 4.84591633e-01
-6.05861187e-01 -1.74067795e-01 -6.87937319e-01 8.88454169e-02
4.55517948e-01 -6.31208599e-01 1.13251463e-01 7.61489153e-01
-1.06033349e+00 7.26856947e-01 2.08784059e-01 3.21991950e-01
-2.46463597e-01 -1.63065314e-01 5.05903542e-01 3.99638891e-01
4.69104260e-01 2.95636624e-01 -8.26395333e-01 -5.94504535e-01
-9.69421208e-01 -1.55978873e-01 6.38227344e-01 1.00256503e+00
4.11179960e-01 1.04146862e+00 -1.03449732e-01 1.83917022e+00
5.17531574e-01 8.38639379e-01 7.09870994e-01 -5.16105533e-01
6.99079454e-01 2.01111361e-01 -4.74414110e-01 -2.89566100e-01
-3.99727337e-02 -3.08620572e-01 -4.74964589e-01 -4.56828207e-01
-2.57114228e-02 -5.11859894e-01 -1.53453648e+00 1.35626662e+00
-1.54693071e-02 7.82038718e-02 4.72290665e-01 5.74732304e-01
1.20180726e+00 1.21843660e+00 1.07214339e-01 1.93983346e-01
1.05041432e+00 -1.25207114e+00 -7.91841388e-01 -6.41540647e-01
6.11399114e-01 -1.24371743e+00 1.23270357e+00 2.63663709e-01
-1.21231675e+00 -5.95464945e-01 -6.68904066e-01 -3.92494470e-01
-6.34978354e-01 3.23599041e-01 5.00951469e-01 9.79316533e-01
-1.32464147e+00 -7.02561289e-02 -8.92519832e-01 -5.58120608e-01
3.78350973e-01 1.62819088e-01 -3.37997347e-01 -5.01550958e-02
-1.14008176e+00 1.06794715e+00 1.81411162e-01 1.46902770e-01
-1.07202733e+00 -6.82889819e-01 -1.10583234e+00 -1.44921750e-01
-2.86954165e-01 1.41526863e-01 1.52223444e+00 -6.57297134e-01
-1.66688538e+00 9.20706987e-01 -3.40590715e-01 -6.17290318e-01
-5.98480813e-02 6.30063042e-02 -1.08616543e+00 -4.55590934e-01
-1.44975036e-01 5.78398049e-01 4.58920628e-01 -7.85788059e-01
-5.01318634e-01 -1.95563018e-01 -5.23251832e-01 2.85086691e-01
-4.00341570e-01 7.61204183e-01 -8.06838274e-01 -6.77367330e-01
-9.29144546e-02 -7.57592440e-01 -2.03522936e-01 -7.20101476e-01
-6.65615022e-01 -2.36560076e-01 1.14138865e+00 -1.51881099e+00
1.39394164e+00 -2.05624366e+00 -2.03081816e-01 1.95518270e-01
-4.24466908e-01 1.01316810e+00 -6.84472859e-01 5.86104810e-01
2.22301800e-02 3.35693270e-01 -1.13861628e-01 -7.13222027e-01
1.15240611e-01 5.33263803e-01 -5.72286069e-01 9.54342335e-02
5.04380703e-01 8.14718068e-01 -5.33430099e-01 1.49884254e-01
5.66746414e-01 1.04602647e+00 -2.95371562e-01 5.36519289e-01
1.31730780e-01 4.10119370e-02 1.01142883e-01 8.58078897e-01
8.06238353e-01 5.92178285e-01 3.17882933e-02 3.55590761e-01
-3.38062346e-01 1.14721966e+00 -7.34221756e-01 1.35704446e+00
-7.54541218e-01 9.13976789e-01 3.84300321e-01 -6.96682215e-01
1.24973357e+00 4.63224053e-01 -9.58405584e-02 -5.38494408e-01
-6.96718320e-02 3.76503050e-01 1.16717279e-01 -2.81756818e-01
6.88968837e-01 1.73398107e-01 -2.15132441e-02 1.05867609e-01
4.16930199e-01 -5.34497678e-01 -3.78614734e-03 1.32439002e-01
1.09574318e+00 -4.43367630e-01 -1.03691712e-01 -3.66474874e-02
4.44275081e-01 1.28336683e-01 4.42658246e-01 6.17298484e-01
-2.91806549e-01 7.33550191e-01 -1.63842022e-01 -4.81833369e-01
-1.30363333e+00 -1.11488903e+00 -2.21326873e-01 1.51382923e+00
-9.21178818e-01 -5.51495969e-01 -6.99316204e-01 -4.65906233e-01
-2.12384954e-01 7.85281062e-01 -2.63907343e-01 2.76024014e-01
-7.35338748e-01 -5.17817080e-01 1.28192794e+00 3.95913392e-01
3.99454027e-01 -1.40943205e+00 4.83464390e-01 3.81259024e-01
7.17279688e-02 -1.16269171e+00 -6.72455907e-01 2.45387807e-01
-2.09857985e-01 -4.38222885e-01 -5.74339986e-01 -1.19354117e+00
6.20737858e-02 1.88674331e-01 1.21353173e+00 1.25123426e-01
-2.22357973e-01 1.90452352e-01 -4.92410660e-01 -5.90725601e-01
-9.38088715e-01 3.30752730e-01 1.56180680e-01 -1.44433141e-01
5.45367002e-01 -1.51216820e-01 9.83428806e-02 -1.34442180e-01
-7.03168035e-01 -3.37161347e-02 1.96018904e-01 6.62987471e-01
4.10024703e-01 -3.80547702e-01 5.75250506e-01 -7.94616997e-01
5.70660949e-01 -4.86006141e-01 -5.26080728e-01 3.68336320e-01
-2.14271069e-01 -3.95253688e-01 6.29093885e-01 -1.53326869e-01
-1.16356099e+00 1.62176996e-01 -9.50120628e-01 -1.68334082e-01
-2.94966161e-01 7.12060869e-01 -2.06538022e-01 1.98620752e-01
4.88432646e-01 3.93559694e-01 -1.90609157e-01 -8.20111215e-01
6.55410290e-01 1.61031663e+00 5.83756387e-01 -1.42960697e-01
6.84226990e-01 -1.48846895e-01 -1.11918736e+00 -1.85294890e+00
-4.41584229e-01 -5.15781581e-01 -5.67129493e-01 6.03410415e-02
7.46242642e-01 -1.02341735e+00 -1.29355684e-01 8.17417979e-01
-1.14509451e+00 -6.95051670e-01 -1.06695153e-01 2.82064378e-01
-6.78323880e-02 -1.46739259e-02 -1.03128147e+00 -1.00880885e+00
-7.40104258e-01 -1.17079067e+00 8.30715835e-01 1.85637310e-01
-9.76276025e-02 -1.08741891e+00 3.08247566e-01 6.79842472e-01
9.72059727e-01 -8.70175242e-01 7.28017092e-01 -9.66987491e-01
-1.18515052e-01 -6.56529739e-02 2.69474983e-02 8.00375819e-01
2.33233541e-01 4.24212217e-01 -1.39761984e+00 -1.99665591e-01
-6.79833710e-01 -4.88487750e-01 7.07868874e-01 4.42667723e-01
1.04397404e+00 -3.84711683e-01 1.84804067e-01 7.85656154e-01
6.81999385e-01 4.49311107e-01 6.91472471e-01 3.36078703e-02
8.66766989e-01 5.00659406e-01 -2.52139457e-02 1.66921586e-01
5.75687766e-01 4.48836505e-01 -1.45228818e-01 -3.10306281e-01
-5.95978200e-01 -3.62363547e-01 7.49639809e-01 1.87520671e+00
5.36416352e-01 -2.55286366e-01 -1.47760594e+00 1.01774490e+00
-1.17564106e+00 -9.03196454e-01 -2.06864804e-01 2.03988886e+00
1.04294276e+00 -3.61676276e-01 8.52755383e-02 -1.94035828e-01
5.68302274e-01 3.42174351e-01 -1.70125216e-01 -9.62634206e-01
-5.85767686e-01 7.40390897e-01 3.61742824e-01 1.11908722e+00
-1.28407454e+00 1.93163848e+00 7.49068499e+00 7.76795447e-01
-1.41578293e+00 1.84880927e-01 7.56999850e-01 1.10345729e-01
-4.10562843e-01 2.73022931e-02 -1.16835332e+00 1.67121768e-01
1.61358011e+00 -2.04526171e-01 8.37907910e-01 9.64043140e-01
4.68946785e-01 7.18232393e-01 -3.89553696e-01 8.26627374e-01
1.28208905e-01 -1.52664816e+00 -1.67821851e-02 -1.24375664e-01
5.63659787e-01 1.09611917e+00 7.62359947e-02 8.01564693e-01
9.94978249e-01 -1.29809809e+00 5.33180416e-01 -9.87840071e-02
7.90740132e-01 -8.33236217e-01 4.36019719e-01 -5.31628029e-04
-9.22711074e-01 1.79791734e-01 -6.07327878e-01 1.55485854e-01
3.68503422e-01 1.81488782e-01 -1.54076672e+00 2.22463459e-02
5.53046525e-01 6.06270492e-01 -8.13043296e-01 7.77167022e-01
-9.49982852e-02 1.58550668e+00 -4.64854300e-01 1.10100903e-01
4.01445836e-01 1.55138941e-02 4.75804418e-01 1.81158972e+00
2.87732780e-01 -2.14828789e-01 1.45795047e-01 2.75084078e-01
-1.95229843e-01 4.15021449e-01 -6.34806037e-01 -5.52517593e-01
9.34288025e-01 1.12493193e+00 -1.97236881e-01 -5.59047997e-01
-4.87507582e-01 9.30321276e-01 5.14367223e-01 3.82968754e-01
-2.91311115e-01 -4.58658904e-01 1.14365375e+00 -2.60666996e-01
1.43922016e-01 -6.10461891e-01 -1.31782010e-01 -1.26669967e+00
-2.95178205e-01 -1.35972214e+00 2.83643305e-01 -7.46895194e-01
-1.38040984e+00 1.08522367e+00 -6.29450798e-01 -3.60400975e-01
-3.19012582e-01 -8.51735771e-01 -7.78975189e-01 1.17216766e+00
-1.36264849e+00 -1.66904593e+00 1.50453985e-01 5.05024970e-01
1.31908488e+00 -8.03026915e-01 1.28927553e+00 3.73022646e-01
-9.38286185e-01 6.91441536e-01 1.16193488e-01 7.24891067e-01
4.69015121e-01 -1.27524495e+00 1.28904891e+00 1.16034293e+00
6.24207377e-01 7.27908373e-01 3.10693830e-01 -6.41585112e-01
-1.59857392e+00 -1.29710257e+00 1.42424381e+00 -6.05862498e-01
9.68293846e-01 -7.67083108e-01 -9.96279776e-01 1.07941842e+00
6.09179974e-01 -1.10000201e-01 9.87676144e-01 4.42923963e-01
-5.09085894e-01 -7.95965418e-02 -7.16296315e-01 6.41422391e-01
6.15637004e-01 -1.00830960e+00 -5.08795857e-01 4.09697771e-01
8.79567087e-01 -2.85372257e-01 -6.57442391e-01 -9.56812799e-02
3.49742830e-01 -2.72520959e-01 7.50487983e-01 -9.31648791e-01
-1.35834396e-01 -5.02357073e-02 -6.04375184e-01 -1.63315606e+00
-1.45870954e-01 -8.50263238e-01 2.97539175e-01 1.75070024e+00
1.13148189e+00 -4.54936177e-01 4.41504151e-01 4.96892482e-01
-7.51332760e-01 -1.31837666e-01 -1.01510918e+00 -6.81311965e-01
3.73243809e-01 -1.16546106e+00 7.50822127e-01 1.34548366e+00
-1.38011634e-01 5.26637375e-01 -4.48833406e-01 2.81983525e-01
1.42920107e-01 -6.56813562e-01 7.87057400e-01 -5.01075327e-01
-2.68896639e-01 -2.78734654e-01 -3.49604756e-01 -7.98922956e-01
2.89936513e-01 -1.18088758e+00 1.87413052e-01 -1.52941680e+00
-2.46860430e-01 -7.92655230e-01 -1.11838661e-01 9.96203482e-01
1.57897230e-02 1.89765692e-01 1.82748660e-01 -1.00738481e-01
-2.65444797e-02 4.66358632e-01 5.49424410e-01 -2.35111624e-01
-5.14209032e-01 1.70716438e-02 -5.11274815e-01 2.60282844e-01
8.88413906e-01 -3.93373333e-02 -2.32359931e-01 -1.22081721e+00
-4.05068338e-01 -3.05507034e-01 -1.67471498e-01 -8.30105186e-01
4.68977503e-02 -7.30389431e-02 1.80194601e-01 -7.78760016e-01
5.46245456e-01 -2.58868754e-01 -2.18581453e-01 2.32070517e-02
-3.20760518e-01 1.66177139e-01 5.87096810e-01 -4.71267968e-01
-2.74287403e-01 -1.14634857e-02 7.69603610e-01 6.69043697e-03
-9.67424214e-01 3.08378935e-01 -8.70619535e-01 1.43286496e-01
2.50432342e-01 2.56578028e-01 -5.80836952e-01 -6.58088803e-01
-8.01758349e-01 1.34901240e-01 -8.11136514e-03 1.01162279e+00
6.36355162e-01 -1.23414230e+00 -1.23459148e+00 7.82314658e-01
6.83369562e-02 -3.17067862e-01 2.27021128e-01 2.98047125e-01
-7.60122955e-01 5.13471127e-01 1.76675767e-01 -1.70246258e-01
-1.50564778e+00 -7.77957812e-02 3.52028519e-01 1.77613571e-01
-1.92390800e-01 1.28208184e+00 -3.92043918e-01 -1.36693597e+00
3.11094910e-01 -1.90993622e-01 -1.97679207e-01 -1.60741881e-01
8.69762063e-01 1.56541422e-01 3.85299623e-01 -1.22572422e+00
-6.19630933e-01 -1.12564482e-01 -3.09147656e-01 -6.37176633e-01
1.63635612e+00 9.29381885e-03 9.54153389e-02 2.95692354e-01
1.11438310e+00 4.38961029e-01 -7.83445001e-01 -2.06115365e-01
8.43677372e-02 -1.33231953e-01 4.16903555e-01 -1.08165336e+00
-8.84052813e-01 1.14416826e+00 5.57076991e-01 9.32740141e-03
7.98626125e-01 7.00255707e-02 8.80298257e-01 4.10943270e-01
-1.62621573e-01 -1.37683809e+00 -5.28679013e-01 1.51554930e+00
8.94237876e-01 -1.13995969e+00 -7.20030189e-01 -2.99712196e-02
-9.24400866e-01 8.44884336e-01 4.29595500e-01 2.33373865e-02
7.83623815e-01 5.48950553e-01 1.08066463e+00 8.02146792e-02
-9.17477489e-01 -4.76104885e-01 2.81839818e-01 9.96556044e-01
9.76109922e-01 4.87974167e-01 2.33298257e-01 5.53066611e-01
-9.07631278e-01 -5.23863673e-01 4.77906823e-01 6.59619272e-01
-3.39515775e-01 -1.46746862e+00 -3.15938145e-01 2.42278129e-01
-5.45984089e-01 -7.62817562e-01 -7.38570154e-01 5.21446168e-01
-3.52444798e-01 1.38277686e+00 2.39291459e-01 -5.16277850e-01
4.55764681e-01 3.91828984e-01 -1.89806670e-01 -1.03367627e+00
-7.58536637e-01 3.72165740e-01 6.78590357e-01 -1.85490787e-01
4.12351876e-01 -5.26446402e-01 -1.08811224e+00 -3.68359834e-01
-2.25085482e-01 1.29476473e-01 1.17631185e+00 6.56693339e-01
4.61505651e-01 2.07064077e-01 8.03163409e-01 -6.34902656e-01
-3.85901779e-01 -1.45775008e+00 -3.45288098e-01 -4.95319907e-03
3.10637325e-01 4.11569357e-01 -1.74744800e-01 1.50773391e-01] | [14.311922073364258, 6.805300712585449] |
62b9e64e-3633-4219-8fe0-55e0ee9c6310 | improved-wavelets-for-image-compression-from | 2203.02556 | null | https://arxiv.org/abs/2203.02556v1 | https://arxiv.org/pdf/2203.02556v1.pdf | Improved Wavelets for Image Compression from Unitary Circuits | We benchmark the efficacy of several novel orthogonal, symmetric, dilation-3 wavelets, derived from a unitary circuit based construction, towards image compression. The performance of these wavelets is compared across several photo databases against the CDF-9/7 wavelets in terms of the minimum number of non-zero wavelet coefficients needed to obtain a specified image quality, as measured by the multi-scale structural similarity index (MS-SSIM). The new wavelets are found to consistently offer better compression efficiency than the CDF-9/7 wavelets across a broad range of image resolutions and quality requirements, averaging 7-8% improved compression efficiency on high-resolution photo images when high-quality (MS-SSIM = 0.99) is required. | ['Glen Evenbly', 'James C. McCord'] | 2022-03-04 | null | null | null | null | ['ms-ssim'] | ['computer-vision'] | [ 6.71371460e-01 -4.39594716e-01 -2.52387494e-01 -3.54215801e-02
-8.51501048e-01 -4.40369248e-02 3.49045873e-01 1.91600174e-01
-4.08581913e-01 3.46074611e-01 5.68276346e-01 -9.70541220e-03
-4.14779752e-01 -8.32166731e-01 -1.75545290e-01 -8.60553801e-01
-5.59890747e-01 -4.55673844e-01 2.46369034e-01 -3.05458993e-01
6.78774357e-01 6.90212727e-01 -1.76567030e+00 3.86686265e-01
2.61136413e-01 1.50838828e+00 1.99489847e-01 1.05085933e+00
5.43200970e-01 3.75629753e-01 -4.50486749e-01 -4.23407763e-01
3.66200119e-01 -4.92663741e-01 -4.36976314e-01 1.37116939e-01
6.23617649e-01 -4.34273273e-01 -4.04536337e-01 8.87324929e-01
7.75687814e-01 -9.36257392e-02 5.41901171e-01 -8.33309352e-01
-8.44831765e-01 3.45336422e-02 -6.26816332e-01 6.05305612e-01
6.37384713e-01 -1.05590895e-02 1.02042305e+00 -9.18536007e-01
6.58457875e-01 1.14663386e+00 1.02861059e+00 -5.75425476e-02
-1.76806450e+00 -2.80945927e-01 -1.21686125e+00 2.27462769e-01
-1.43026066e+00 -8.18915844e-01 5.86368144e-01 -1.40408143e-01
1.25049925e+00 4.19578314e-01 3.70606512e-01 5.62213421e-01
8.23346198e-01 -3.85886252e-01 1.07236373e+00 -6.01636171e-01
1.80683717e-01 -2.76523679e-01 -2.26812139e-01 3.76332462e-01
5.19387364e-01 2.35691115e-01 -6.10279560e-01 -3.40405464e-01
1.07856238e+00 -4.58759487e-01 -2.98194736e-01 8.31730515e-02
-9.82859790e-01 8.91838670e-01 1.67718172e-01 7.16786385e-01
-5.99059522e-01 2.39184529e-01 4.59396273e-01 6.56401634e-01
3.28935415e-01 1.70102894e-01 -9.14803520e-02 -3.78036797e-02
-9.87458646e-01 1.12495527e-01 3.97807986e-01 8.51440907e-01
4.44186985e-01 4.34685260e-01 -1.16560981e-01 1.02579653e+00
-7.28356093e-02 5.25952518e-01 6.45493925e-01 -1.56263053e+00
2.26729751e-01 -7.06115067e-02 -9.54046100e-03 -1.39242113e+00
-1.69131994e-01 -2.54171550e-01 -1.06691110e+00 3.78100276e-01
-1.18267216e-01 4.76622343e-01 -6.10905588e-01 1.23332226e+00
-1.05314732e-01 -3.54385048e-01 4.89267886e-01 4.95559007e-01
8.32194090e-01 7.17936933e-01 -1.71174303e-01 -8.02944601e-01
1.39186692e+00 -2.07060039e-01 -6.99955225e-01 2.87500650e-01
9.49114840e-03 -1.43180811e+00 9.01381731e-01 3.17018658e-01
-1.66841161e+00 -1.05131102e+00 -1.25693190e+00 -3.10223699e-02
3.35408866e-01 -6.46294430e-02 5.69591634e-02 6.37247503e-01
-1.09616661e+00 7.61573851e-01 -3.91594887e-01 -3.53064567e-01
4.13776368e-01 1.07259732e-02 -5.29918134e-01 -3.10457021e-01
-5.64201951e-01 7.96984375e-01 1.17846355e-01 -8.28735471e-01
-3.80507618e-01 -6.81457579e-01 -7.20151365e-01 4.11501348e-01
-4.31650281e-01 -4.47399050e-01 7.25028813e-01 -3.88473898e-01
-9.58599806e-01 7.51681685e-01 -2.07732216e-01 -8.54848385e-01
1.86818615e-01 2.92814076e-01 -7.33932197e-01 1.07651436e+00
-1.04045253e-02 6.08926892e-01 1.05507970e+00 -1.26706696e+00
-5.91187418e-01 1.91169027e-02 -3.33591640e-01 -3.84741947e-02
-4.47398067e-01 3.27235699e-01 1.24550216e-01 -9.30709243e-01
4.50164437e-01 -4.66914117e-01 -5.15602902e-02 2.59721633e-02
2.10470319e-01 1.22905158e-01 9.55908120e-01 -8.48816752e-01
1.16962481e+00 -2.37498403e+00 -2.41220668e-01 2.08969846e-01
-8.53345767e-02 1.97839648e-01 -3.71379882e-01 7.52098978e-01
-1.73599318e-01 1.20093472e-01 -2.85504341e-01 1.10976189e-01
-5.03237545e-01 2.39665225e-01 -2.29966357e-01 6.66643441e-01
1.44711316e-01 3.32649171e-01 -4.18021858e-01 -6.26852214e-01
2.30133086e-01 8.40332806e-01 -4.61472034e-01 9.07445475e-02
6.98700130e-01 -1.02750063e-01 1.26442552e-01 8.88245046e-01
7.60373116e-01 1.06078275e-01 -1.75486237e-01 -4.86323774e-01
-9.55054909e-03 -2.19147354e-01 -9.48797405e-01 1.07847106e+00
-1.71764120e-01 7.74153650e-01 3.84306796e-02 -1.16851997e+00
1.22818947e+00 6.85605109e-01 1.04275954e+00 -9.63929892e-01
-1.17273189e-01 2.37773716e-01 2.77861059e-02 -7.12186515e-01
3.84082705e-01 -2.63404638e-01 3.88749719e-01 8.83803442e-02
1.55862391e-01 -3.75653148e-01 6.45225704e-01 -5.61166927e-02
1.17219806e+00 -4.61509764e-01 5.23026586e-01 -6.14108145e-01
5.02899766e-01 -2.28719905e-01 3.21406037e-01 3.14183533e-01
-1.70092314e-01 8.88995945e-01 1.53192773e-01 -4.64366466e-01
-1.85713768e+00 -1.09125757e+00 -6.76496148e-01 5.10244250e-01
-1.26781315e-02 -5.30751348e-01 -4.74181503e-01 4.13006455e-01
3.59935127e-02 5.27060032e-01 -4.96058017e-02 -9.20633748e-02
-4.57885116e-01 -6.83635473e-01 5.25809050e-01 2.36189336e-01
6.13424480e-01 -9.87245500e-01 -1.31125641e+00 2.81911105e-01
-2.07247913e-01 -1.52501690e+00 -3.25477064e-01 -7.56431147e-02
-1.21203172e+00 -1.03428769e+00 -6.52838588e-01 -6.86118066e-01
4.88565356e-01 6.77848577e-01 1.05776763e+00 1.67694896e-01
-8.49988282e-01 6.73339486e-01 -6.87624335e-01 5.93200475e-02
-5.69096804e-01 -8.28511953e-01 6.25449494e-02 -1.77313641e-01
-8.02902952e-02 -1.10675156e+00 -8.12013268e-01 2.66831666e-01
-1.23923266e+00 -2.58365184e-01 6.10136688e-01 9.45826292e-01
7.91225076e-01 6.75880849e-01 3.99803877e-01 1.10286593e-01
8.07955563e-01 4.96072508e-02 -1.72353581e-01 -1.27587512e-01
-9.05843318e-01 -3.86063099e-01 5.37167549e-01 -5.49305856e-01
-4.81568515e-01 -2.39762470e-01 -2.63602883e-02 -2.96331793e-01
9.92881507e-02 3.48822623e-01 5.61414063e-01 -5.83055258e-01
1.01098108e+00 5.70751250e-01 5.08986562e-02 -3.10752720e-01
2.84609627e-02 4.93613005e-01 1.05900300e+00 -2.92387456e-01
7.93410242e-01 6.37113392e-01 4.81461763e-01 -1.37412941e+00
3.44599038e-02 -6.34708285e-01 -3.97119790e-01 4.38071266e-02
8.17697108e-01 -7.96896160e-01 -4.67101157e-01 1.57638833e-01
-9.75410938e-01 3.18452120e-01 -5.43146789e-01 6.04468107e-01
-1.00785542e+00 6.88465536e-01 -6.13771617e-01 -8.39113653e-01
-6.08767450e-01 -1.04967892e+00 8.44457090e-01 7.48180202e-04
-1.80005267e-01 -4.54361886e-01 -3.10928404e-01 1.67128488e-01
7.92571843e-01 6.65009260e-01 1.24063182e+00 1.46127179e-01
-8.92887339e-02 -3.74628067e-01 -3.75346005e-01 6.04477942e-01
2.11985156e-01 1.87660262e-01 -5.83023190e-01 -6.27779424e-01
3.73593748e-01 -3.93321097e-01 7.39053369e-01 7.33642042e-01
8.93507659e-01 -5.15641868e-01 2.42260084e-01 6.23476505e-01
1.86248720e+00 4.00100976e-01 1.14959013e+00 1.07353464e-01
-3.25240105e-01 3.31132025e-01 2.66470402e-01 7.51110852e-01
-2.66599268e-01 6.56869769e-01 3.60197276e-01 -9.95099619e-02
-5.41879177e-01 1.52235433e-01 2.00328276e-01 9.60177481e-01
-4.97478724e-01 1.09236449e-01 -3.31550330e-01 5.46642721e-01
-1.01227486e+00 -1.51259518e+00 -1.29255772e-01 2.16759849e+00
6.79938614e-01 2.54105777e-01 9.05435830e-02 8.64141643e-01
6.15819752e-01 2.05531940e-01 -5.97016066e-02 -6.00800395e-01
-4.63146299e-01 7.12774754e-01 6.68621838e-01 3.08710009e-01
-7.89526463e-01 1.24462776e-01 7.84419107e+00 7.83071399e-01
-7.82303870e-01 -7.59365931e-02 4.94434685e-01 8.27208161e-02
1.24608027e-02 -8.69108178e-03 -2.98266679e-01 3.83139431e-01
1.11881399e+00 -5.23181915e-01 3.59010130e-01 6.98360860e-01
3.41711700e-01 -3.56711715e-01 -6.26100481e-01 1.22295034e+00
1.60023525e-01 -1.54966533e+00 9.15749297e-02 2.73293436e-01
6.91401243e-01 -3.56199652e-01 9.48931798e-02 -6.18581593e-01
-4.57048595e-01 -1.09052277e+00 4.89300072e-01 3.35013419e-01
1.27977765e+00 -7.31358349e-01 5.49040318e-01 -4.04609181e-02
-1.40266216e+00 -2.13047370e-01 -8.87960851e-01 2.00969830e-01
1.83087051e-01 6.23561502e-01 -1.80965707e-01 5.25453269e-01
1.06050730e+00 2.30501622e-01 -1.64012909e-01 1.17092621e+00
5.48337221e-01 3.60032439e-01 -2.87341267e-01 5.69798470e-01
3.17394570e-03 -1.40057594e-01 5.55061162e-01 1.27755165e+00
1.03316641e+00 5.87911844e-01 -1.64013401e-01 1.19570777e-01
1.76373333e-01 1.16749421e-01 -5.69698870e-01 3.18448059e-02
4.83688414e-01 8.85148644e-01 -6.53112769e-01 -3.28229427e-01
-3.34066421e-01 8.20351839e-01 -6.32716179e-01 9.97596905e-02
-2.88201362e-01 -6.30663276e-01 5.61416388e-01 2.08488658e-01
7.16253281e-01 -4.78405029e-01 -2.52344966e-01 -5.89997470e-01
1.91995148e-02 -1.01611531e+00 2.94624537e-01 -1.04309547e+00
-1.10624158e+00 5.90391219e-01 2.03209206e-01 -1.35382724e+00
-8.88951197e-02 -1.70879364e-01 -5.09625673e-01 7.49974251e-01
-1.25063801e+00 -7.97174335e-01 -2.52664924e-01 5.32585144e-01
5.67773223e-01 -4.14657891e-01 1.18444324e+00 3.27117473e-01
1.86106727e-01 5.39326072e-01 3.61758992e-02 -3.73604238e-01
2.85606593e-01 -5.72942317e-01 2.47434244e-01 8.06372583e-01
1.86730087e-01 5.53311594e-02 1.06552541e+00 -4.88140404e-01
-1.46080661e+00 -5.74767530e-01 8.74348342e-01 3.39235544e-01
1.56950682e-01 3.88230324e-01 -8.00232530e-01 9.01795626e-02
2.86163539e-01 2.63886631e-01 6.39884770e-01 -7.53658056e-01
-5.42176723e-01 -2.31657684e-01 -1.79880607e+00 1.57055661e-01
9.58089590e-01 -4.19102579e-01 -4.26819801e-01 1.49771169e-01
6.61579728e-01 1.19175389e-03 -1.80060005e+00 7.77384520e-01
8.54271293e-01 -1.42866242e+00 1.46744144e+00 -1.39921144e-01
8.06394398e-01 2.96941157e-02 -9.86655712e-01 -8.17899227e-01
-7.90537179e-01 -5.56404054e-01 1.32651597e-01 8.20642114e-01
-3.83637543e-03 -4.44209605e-01 4.48780179e-01 -2.44929045e-01
6.70029074e-02 -8.46940875e-01 -1.48282027e+00 -9.41168308e-01
-5.07085800e-01 -1.87461451e-01 2.14544848e-01 6.78425312e-01
-2.42366381e-02 7.92194381e-02 -3.72267663e-01 -1.18261144e-01
8.42610061e-01 -1.30467266e-01 5.19321382e-01 -9.95563090e-01
-1.75270230e-01 -4.95266557e-01 -1.08209395e+00 -3.70744348e-01
-6.76718473e-01 -3.70192796e-01 -5.37412703e-01 -1.42768073e+00
4.73282859e-02 -9.18314457e-02 -9.54543725e-02 1.01483695e-01
5.20021915e-01 8.24237943e-01 2.30606988e-01 5.95532477e-01
4.16284084e-01 3.92485946e-01 9.20916736e-01 7.84137249e-02
9.29727629e-02 -2.16389492e-01 -5.25672793e-01 4.80789810e-01
5.84180474e-01 -3.00906897e-01 -4.09336060e-01 -8.14893004e-03
-2.68348604e-01 5.53856254e-01 1.80170462e-01 -1.61993885e+00
-1.32987678e-01 -1.48501009e-01 7.06171989e-01 -5.10718822e-01
6.65956676e-01 -9.39088464e-01 5.33391714e-01 9.14997339e-01
-3.50074828e-01 4.12388086e-01 -7.92272389e-02 3.39783907e-01
-3.89281809e-01 -2.94672430e-01 1.19604349e+00 3.70578095e-02
-5.23682356e-01 -5.82533069e-02 -1.66151494e-01 -4.30777788e-01
1.12649107e+00 -7.33828187e-01 -1.56785399e-01 -5.05176127e-01
-4.08095211e-01 -8.01454484e-01 5.20704091e-01 -1.72098771e-01
1.10006917e+00 -1.60932016e+00 -9.99123812e-01 4.52142358e-01
1.52960673e-01 -7.34564960e-01 7.52964169e-02 7.24437892e-01
-7.38697171e-01 2.05087274e-01 -6.52018726e-01 -5.57344794e-01
-1.74825215e+00 3.54863703e-01 -9.40289423e-02 -7.60500804e-02
-9.95627344e-01 4.70244735e-01 -4.81295645e-01 8.23971748e-01
-1.04304500e-01 -1.27207473e-01 -2.29246691e-02 -1.72535539e-01
6.09093308e-01 7.64336228e-01 1.58508658e-01 -9.24668789e-01
8.06984007e-02 9.17863727e-01 5.42569757e-01 -2.08760232e-01
1.56270039e+00 -7.25087821e-02 -2.08863094e-01 -8.64812061e-02
1.50761235e+00 1.82957798e-02 -9.60864604e-01 -8.24340507e-02
-1.61748484e-01 -9.07036841e-01 3.95297147e-02 -3.04012418e-01
-8.00914347e-01 3.46963286e-01 9.90957439e-01 7.63364017e-01
1.75006390e+00 -2.36333162e-01 1.01880825e+00 1.01586960e-01
3.23858052e-01 -1.01083624e+00 3.52639228e-01 -7.87172616e-02
1.20138025e+00 -9.16882336e-01 7.38020957e-01 -4.30118322e-01
-3.20490539e-01 1.50107038e+00 -1.94495603e-01 -5.54294169e-01
6.49177372e-01 3.03163767e-01 -2.78859437e-01 -5.57545908e-02
-4.90000367e-01 -6.99101388e-02 1.36424169e-01 8.36566567e-01
5.43493450e-01 -2.02231221e-02 -7.89365470e-01 -3.79702508e-01
-4.29856241e-01 -7.99966156e-02 4.63014990e-01 9.99681950e-01
-9.32261586e-01 -9.55859244e-01 -6.19869113e-01 4.99238938e-01
-7.43040144e-01 -3.55632487e-03 3.82342130e-01 7.18270779e-01
-2.25108173e-02 1.14548171e+00 1.40710384e-01 -3.91803056e-01
3.35503608e-01 -1.71248451e-01 6.57637239e-01 4.88523617e-02
-1.96983263e-01 4.64535862e-01 2.08767965e-01 -6.25206172e-01
-9.14143026e-01 -3.45284343e-01 -8.73984814e-01 -7.66063809e-01
-6.41584769e-02 -1.54806793e-01 8.57243598e-01 4.13079709e-01
1.02849975e-01 2.29683071e-01 9.70029771e-01 -8.06497812e-01
-8.63432825e-01 -9.87008214e-01 -5.12059629e-01 6.06839120e-01
3.30787450e-01 -2.82661140e-01 -2.84011245e-01 6.83911741e-01] | [11.59280014038086, -2.0822393894195557] |
0f4ac962-6e2b-4f03-ab6e-134a36338d04 | spiq-data-free-per-channel-static-input | 2203.14642 | null | https://arxiv.org/abs/2203.14642v1 | https://arxiv.org/pdf/2203.14642v1.pdf | SPIQ: Data-Free Per-Channel Static Input Quantization | Computationally expensive neural networks are ubiquitous in computer vision and solutions for efficient inference have drawn a growing attention in the machine learning community. Examples of such solutions comprise quantization, i.e. converting the processing values (weights and inputs) from floating point into integers e.g. int8 or int4. Concurrently, the rise of privacy concerns motivated the study of less invasive acceleration methods, such as data-free quantization of pre-trained models weights and activations. Previous approaches either exploit statistical information to deduce scalar ranges and scaling factors for the activations in a static manner, or dynamically adapt this range on-the-fly for each input of each layers (also referred to as activations): the latter generally being more accurate at the expanse of significantly slower inference. In this work, we argue that static input quantization can reach the accuracy levels of dynamic methods by means of a per-channel input quantization scheme that allows one to more finely preserve cross-channel dynamics. We show through a thorough empirical evaluation on multiple computer vision problems (e.g. ImageNet classification, Pascal VOC object detection as well as CityScapes semantic segmentation) that the proposed method, dubbed SPIQ, achieves accuracies rivalling dynamic approaches with static-level inference speed, significantly outperforming state-of-the-art quantization methods on every benchmark. | ['Kevin Bailly', 'Matthieu Cord', 'Arnaud Dapogny', 'Edouard Yvinec'] | 2022-03-28 | null | null | null | null | ['data-free-quantization', 'data-free-quantization'] | ['computer-vision', 'methodology'] | [ 4.33324933e-01 -5.23703471e-02 -8.60620141e-02 -5.11675358e-01
-7.86211073e-01 -6.83760464e-01 6.35552347e-01 4.29853052e-01
-1.17852485e+00 7.71740913e-01 -4.16065127e-01 -4.24307257e-01
-5.68539370e-04 -7.88620114e-01 -1.04150295e+00 -8.28848660e-01
9.30076465e-02 2.86990106e-01 3.39142919e-01 -8.03951174e-03
3.00030380e-01 4.01278257e-01 -1.69914746e+00 1.94649145e-01
7.11424649e-01 1.46279430e+00 -1.83167979e-01 7.33824909e-01
-1.55166566e-01 2.32630238e-01 -7.01391518e-01 -8.68233025e-01
5.43254912e-01 -6.84568956e-02 -6.10881627e-01 -1.21861644e-01
6.89826965e-01 -2.78940409e-01 -1.68988109e-01 1.46599603e+00
3.10744822e-01 -2.04885621e-02 3.37603182e-01 -1.11853325e+00
-4.60556149e-01 7.26410210e-01 -5.60972273e-01 3.04378271e-01
-2.80496120e-01 2.06754595e-01 1.11363804e+00 -6.22539222e-01
4.35274780e-01 1.22755063e+00 6.25243247e-01 4.09097314e-01
-1.57388544e+00 -6.90857708e-01 1.28339678e-01 2.91068643e-01
-1.58291090e+00 -4.71995234e-01 3.72965276e-01 -2.57058382e-01
1.05223179e+00 2.48440415e-01 5.36501050e-01 8.05353165e-01
1.58332199e-01 6.66051626e-01 1.21569502e+00 -3.56398284e-01
8.23138475e-01 1.15923718e-01 1.73191190e-01 6.76817358e-01
4.75457132e-01 -4.71354239e-02 -5.91997623e-01 -1.68862581e-01
7.60226250e-01 -1.34620368e-01 -2.96723843e-01 -1.97338164e-01
-1.12768614e+00 1.00020313e+00 5.98466992e-01 6.06246218e-02
-4.46928620e-01 6.64464831e-01 8.45602930e-01 4.15720016e-01
2.84711003e-01 2.10226223e-01 -6.59450352e-01 -1.85202867e-01
-9.93886352e-01 3.83420318e-01 8.00038040e-01 7.10919380e-01
9.22123253e-01 8.97440985e-02 -2.18199179e-01 5.73414922e-01
6.16504736e-02 2.88155973e-01 5.27250051e-01 -1.03204846e+00
5.92612565e-01 2.43066445e-01 -5.65454066e-02 -9.11101878e-01
-1.34803489e-01 -3.67557704e-01 -1.06580842e+00 2.90590763e-01
7.19767094e-01 -8.76273289e-02 -1.13777936e+00 1.67036235e+00
2.67986923e-01 1.30161211e-01 -4.18498851e-02 8.74299467e-01
1.26762375e-01 6.65809810e-01 1.93116263e-01 -1.28944874e-01
1.51453590e+00 -6.12263143e-01 -5.99934280e-01 -7.28739426e-02
2.81702340e-01 -3.78072917e-01 9.46581960e-01 8.66670489e-01
-1.05022943e+00 -5.88990271e-01 -1.27038884e+00 -1.79383695e-01
-7.25011706e-01 -6.74893558e-02 8.37607861e-01 9.84847426e-01
-1.23198247e+00 8.14617932e-01 -1.15945876e+00 6.25185203e-03
8.22756648e-01 7.26496220e-01 -7.11770952e-02 1.91258684e-01
-1.23650825e+00 6.84862077e-01 7.96393573e-01 2.45136827e-01
-5.25915265e-01 -7.16460526e-01 -7.05594838e-01 2.06345156e-01
3.91171247e-01 -6.31501138e-01 1.25006056e+00 -1.35346735e+00
-1.60244453e+00 7.47667193e-01 7.54563585e-02 -1.20933533e+00
7.29464829e-01 -1.83272555e-01 5.30669242e-02 2.35050097e-01
-2.63407707e-01 1.20402074e+00 1.31569135e+00 -8.85427535e-01
-7.70548105e-01 -5.22072673e-01 1.26913264e-01 -2.51724452e-01
-5.97096682e-01 -2.58686930e-01 -6.14203393e-01 -4.60712790e-01
-3.50851417e-02 -8.65793586e-01 -5.07895470e-01 3.98668140e-01
-3.65255743e-01 -2.30885223e-01 6.60928190e-01 -4.90033209e-01
1.11168456e+00 -2.04917789e+00 2.10969220e-03 3.61671746e-01
2.71164626e-01 5.62108755e-01 2.21626118e-01 -1.76530182e-01
7.40150288e-02 9.06850472e-02 -3.24116766e-01 -5.17145634e-01
2.14424580e-01 5.57955503e-01 -3.05687249e-01 6.38110876e-01
3.36484641e-01 9.73802269e-01 -7.92517066e-01 -5.12714207e-01
3.27895552e-01 8.02291274e-01 -8.03942144e-01 -3.43373895e-01
-2.43962646e-01 -4.59732637e-02 -3.25295001e-01 4.55920726e-01
5.71391642e-01 -2.55686879e-01 3.69958393e-02 -3.65895897e-01
-1.67535767e-01 2.10604772e-01 -1.29527330e+00 1.66112196e+00
-4.52013761e-01 6.66348040e-01 1.38860762e-01 -1.08706093e+00
6.47175491e-01 2.12042674e-01 7.76698142e-02 -6.34778380e-01
3.72738928e-01 2.95100927e-01 -1.36820868e-01 7.62994960e-02
5.88233471e-01 1.19763799e-01 -4.04110104e-02 -1.29761219e-01
7.66610354e-02 6.41018674e-02 1.67712539e-01 -1.32250667e-01
9.09666598e-01 -2.87199825e-01 4.31241065e-01 -2.45327786e-01
4.33408886e-01 -1.06711313e-01 3.34466130e-01 9.33349371e-01
-3.48030001e-01 4.51332152e-01 4.70227957e-01 -4.08495098e-01
-9.90763426e-01 -1.00101161e+00 -3.20890993e-01 1.07767928e+00
-1.85555100e-01 -3.52538168e-01 -1.12659216e+00 -3.81168813e-01
8.84527639e-02 3.03595155e-01 -7.00646460e-01 -1.19424956e-02
-4.84874040e-01 -7.49629259e-01 8.19669664e-01 5.70502877e-01
8.45259964e-01 -7.56030023e-01 -1.29863000e+00 3.82898539e-01
3.90018791e-01 -1.17767978e+00 -1.16361663e-01 7.88644910e-01
-9.29404616e-01 -5.11282086e-01 -6.38985634e-01 -2.83903867e-01
4.68553603e-01 -3.19089592e-01 8.87238801e-01 -1.03233464e-01
-4.47860330e-01 1.43424958e-01 -9.65342522e-02 -5.38228571e-01
-9.46441218e-02 2.74329931e-01 -6.79469630e-02 2.90906191e-01
3.29950541e-01 -5.60153842e-01 -7.28257537e-01 -3.09601307e-01
-1.11964571e+00 -1.25904948e-01 6.54253364e-01 9.73694742e-01
9.68997121e-01 2.79253554e-02 3.00348163e-01 -1.00587118e+00
5.54974139e-01 -1.73852816e-01 -1.09438086e+00 -2.66867317e-02
-8.09070289e-01 2.97087371e-01 9.23197031e-01 -4.53686655e-01
-5.81928909e-01 3.36757541e-01 -3.05701435e-01 -7.01390862e-01
-1.48648784e-01 2.69710600e-01 5.55887707e-02 -2.57147074e-01
6.51163280e-01 1.90146506e-01 -2.14636073e-01 -1.53521448e-01
6.54712737e-01 6.17064238e-01 8.84844184e-01 -5.04724205e-01
5.87953389e-01 5.54964006e-01 5.62937409e-02 -7.16246188e-01
-5.87995172e-01 -2.68429041e-01 -5.41413128e-01 7.43738860e-02
1.02314258e+00 -8.45333934e-01 -9.45927620e-01 4.43753153e-01
-1.22772741e+00 -1.64220646e-01 -3.34908396e-01 1.91596821e-01
-3.70592386e-01 4.31099273e-02 -7.36541152e-01 -6.69847190e-01
-3.98017704e-01 -1.38773882e+00 1.14905679e+00 2.37214908e-01
-6.83913529e-02 -8.55622947e-01 -3.82396996e-01 5.07298484e-02
4.90520149e-01 3.35767537e-01 7.72644222e-01 -4.94087905e-01
-7.73501277e-01 -2.25084633e-01 -3.92611414e-01 5.80313742e-01
-2.61063635e-01 -1.22728489e-01 -1.31528330e+00 -2.45253727e-01
-1.16990685e-01 -3.43724102e-01 1.06418777e+00 4.48547870e-01
1.56894147e+00 -5.00915289e-01 -7.34808445e-02 9.41657782e-01
1.74532938e+00 1.89110450e-02 6.49463117e-01 2.80841619e-01
6.87060118e-01 8.41689929e-02 2.62318164e-01 6.74269497e-01
1.45865560e-01 5.87800324e-01 5.37876546e-01 -2.04955246e-02
2.13579118e-01 5.15643656e-02 -1.23896881e-03 4.84668165e-01
-4.54878099e-02 -6.37426600e-02 -7.49236107e-01 4.02354687e-01
-1.64365029e+00 -6.58689260e-01 1.42481968e-01 2.19995761e+00
1.01041615e+00 4.96977180e-01 -1.85681865e-01 4.88112330e-01
4.39418346e-01 1.76884517e-01 -7.47642577e-01 -8.97073150e-01
2.08320692e-01 7.03169048e-01 1.23366213e+00 3.70985985e-01
-1.18329680e+00 8.33389938e-01 5.81818581e+00 9.70389664e-01
-1.39454222e+00 1.99413463e-01 9.69754577e-01 -3.01164091e-02
-2.43248828e-02 -3.47366124e-01 -9.55325007e-01 4.56438571e-01
1.23454022e+00 2.59378523e-01 8.03421140e-01 9.66780961e-01
-1.41954571e-01 -6.96028844e-02 -1.19850326e+00 1.30222857e+00
-2.60414809e-01 -1.60202444e+00 2.16859519e-01 5.29007390e-02
7.30775476e-01 2.23787650e-01 2.61722893e-01 1.81479260e-01
8.45853016e-02 -1.01202989e+00 8.74445617e-01 1.60025433e-01
8.77092719e-01 -9.28676069e-01 7.08605289e-01 1.43624857e-01
-1.07525599e+00 -1.02176778e-01 -3.90712529e-01 -5.08247726e-02
-1.47726655e-01 6.16454303e-01 -5.82481503e-01 2.95485377e-01
8.89572501e-01 3.15208852e-01 -5.92497051e-01 7.56963968e-01
2.01358367e-02 7.62953937e-01 -5.15375853e-01 -1.90028295e-01
5.85662663e-01 -1.88283980e-01 1.07517548e-01 1.39723182e+00
-8.46568421e-02 4.30631265e-03 -1.03482433e-01 9.42495644e-01
-2.97523826e-01 -1.62415057e-01 -2.11384147e-01 -5.42054400e-02
5.70809364e-01 1.03984892e+00 -9.20519948e-01 -4.53113884e-01
-3.04536819e-01 1.11213434e+00 3.31917495e-01 2.83788830e-01
-9.03376758e-01 -4.74748939e-01 8.13744962e-01 -2.59707272e-01
7.50734568e-01 -2.80587137e-01 -6.91706181e-01 -7.38557458e-01
3.31062488e-02 -7.83649266e-01 3.00084472e-01 -2.62055755e-01
-9.19174731e-01 4.59276587e-01 1.35584734e-03 -7.96244800e-01
-2.37230211e-01 -8.33129287e-01 -1.97127551e-01 6.94610000e-01
-1.70227826e+00 -6.32274270e-01 -7.88975358e-02 7.00527012e-01
4.39666122e-01 1.61679462e-01 6.92172348e-01 4.24322456e-01
-4.26343352e-01 9.73313630e-01 2.17256576e-01 1.48156583e-01
2.56320268e-01 -1.29614949e+00 4.08783346e-01 7.87397325e-01
3.30528557e-01 6.73223555e-01 7.55026996e-01 -3.32460180e-02
-1.65519226e+00 -1.14565599e+00 6.43231094e-01 1.59439836e-02
6.39121354e-01 -5.23906052e-01 -9.40465927e-01 4.47374880e-01
3.97615209e-02 3.54618192e-01 1.40371233e-01 -3.10631126e-01
-5.04208803e-01 -3.85125250e-01 -1.20218194e+00 6.85518265e-01
6.68564439e-01 -5.73407769e-01 -2.34657183e-01 4.54057828e-02
6.31863534e-01 -4.74150747e-01 -7.14984357e-01 2.40209419e-02
5.15755892e-01 -1.06692588e+00 1.08898079e+00 -8.33846554e-02
2.68934786e-01 -2.62558997e-01 -3.02726269e-01 -9.48452532e-01
1.55560791e-01 -7.62089849e-01 -1.13856889e-01 9.52212751e-01
2.32428789e-01 -9.27527845e-01 8.88919711e-01 6.53861105e-01
5.70373200e-02 -9.22267973e-01 -1.32194042e+00 -5.45401514e-01
-3.92538682e-02 -6.11788154e-01 4.61443633e-01 6.38965786e-01
-5.60133159e-01 8.77130926e-02 1.27365217e-01 2.95464575e-01
6.26911998e-01 -2.03669190e-01 4.45544958e-01 -1.00049281e+00
-5.16000569e-01 -7.99989641e-01 -8.07958782e-01 -1.02460945e+00
7.22112656e-02 -7.67592788e-01 9.67867449e-02 -8.99425447e-01
-3.03396434e-01 -3.74029815e-01 -4.14610863e-01 5.95305860e-01
-3.87773365e-02 5.48248172e-01 1.62689641e-01 -1.06831647e-01
-5.06272256e-01 3.72547179e-01 9.42632258e-01 -2.92282850e-01
-5.82682565e-02 -2.29055569e-01 -4.98938084e-01 6.63814366e-01
7.16774344e-01 -5.31566441e-01 -3.59107077e-01 -5.10934532e-01
1.59807816e-01 -1.82062969e-01 6.58687890e-01 -1.22027338e+00
4.38928187e-01 1.08060010e-01 4.10232157e-01 -3.18968356e-01
4.17450488e-01 -8.89181912e-01 -9.36937183e-02 6.12853527e-01
-4.50375795e-01 8.78715217e-02 3.48988265e-01 7.09798396e-01
-2.80389935e-01 -7.68410638e-02 9.11519885e-01 -4.70188335e-02
-7.71643460e-01 3.41414720e-01 -2.01134324e-01 -4.34743613e-02
8.25485468e-01 -4.16128933e-01 -7.70301139e-03 1.51064864e-03
-3.65811199e-01 -6.28845915e-02 1.01317093e-01 1.57470897e-01
5.00827789e-01 -9.35507417e-01 -3.94986272e-01 2.68224478e-01
-2.30612472e-01 1.93338796e-01 -1.26047343e-01 7.66402423e-01
-5.66231668e-01 5.07336378e-01 -1.92849219e-01 -7.88957059e-01
-1.06544125e+00 4.74003077e-01 3.69553119e-01 -2.46553510e-01
-5.85615575e-01 1.12342632e+00 -8.25829953e-02 3.50905880e-02
6.02021277e-01 -9.89762485e-01 2.88058609e-01 6.76063448e-02
6.35341525e-01 1.95335865e-01 3.79355520e-01 -2.14977324e-01
-3.52045029e-01 3.64078701e-01 -1.46289617e-01 -2.68062919e-01
1.05305910e+00 9.81171504e-02 -7.48770013e-02 4.90852863e-01
1.36646688e+00 -5.19296646e-01 -1.48179948e+00 -1.15961559e-01
3.20774503e-03 -2.94173807e-01 3.81369233e-01 -6.03844345e-01
-1.31190741e+00 1.04593742e+00 8.56261730e-01 2.47734040e-01
1.29871225e+00 -3.36764365e-01 9.21383858e-01 7.38934517e-01
6.16258085e-01 -1.19632089e+00 -3.86657536e-01 3.87590736e-01
3.40175688e-01 -1.11364949e+00 -1.03164844e-01 -2.68063098e-01
-3.15609545e-01 1.10837889e+00 2.17668682e-01 -2.79646337e-01
6.42918587e-01 5.41939318e-01 -3.51743028e-02 1.52738944e-01
-7.53028631e-01 -3.97300906e-02 1.49514332e-01 3.14341396e-01
9.33149680e-02 1.76151395e-01 -2.21712440e-01 4.49261308e-01
-2.02990755e-01 1.09756328e-01 1.53060034e-01 9.19294059e-01
-3.29946697e-01 -8.97370160e-01 -4.31762248e-01 5.96306801e-01
-6.46600723e-01 -2.53780246e-01 -5.44981100e-02 7.47560263e-01
3.92095208e-01 6.33415937e-01 2.73759663e-01 -2.90210098e-01
2.00469851e-01 4.82529737e-02 5.35645783e-01 -1.75310612e-01
-9.93651450e-01 -2.15642124e-01 -2.65881658e-01 -7.47016191e-01
-2.94001549e-01 -5.76793909e-01 -1.47200739e+00 -2.71105468e-01
-2.48371527e-01 -1.18655108e-01 9.93196607e-01 9.77896988e-01
2.01079965e-01 5.66553116e-01 1.89471379e-01 -1.04735076e+00
-1.04679346e+00 -4.87305254e-01 -4.03524995e-01 2.50853032e-01
5.44764161e-01 -3.57684106e-01 -2.76110858e-01 2.89165050e-01] | [8.607053756713867, 3.1164307594299316] |
b77758c0-f739-43ed-8931-5f6561f6d471 | bootstrapped-masked-autoencoders-for-vision | 2207.07116 | null | https://arxiv.org/abs/2207.07116v1 | https://arxiv.org/pdf/2207.07116v1.pdf | Bootstrapped Masked Autoencoders for Vision BERT Pretraining | We propose bootstrapped masked autoencoders (BootMAE), a new approach for vision BERT pretraining. BootMAE improves the original masked autoencoders (MAE) with two core designs: 1) momentum encoder that provides online feature as extra BERT prediction targets; 2) target-aware decoder that tries to reduce the pressure on the encoder to memorize target-specific information in BERT pretraining. The first design is motivated by the observation that using a pretrained MAE to extract the features as the BERT prediction target for masked tokens can achieve better pretraining performance. Therefore, we add a momentum encoder in parallel with the original MAE encoder, which bootstraps the pretraining performance by using its own representation as the BERT prediction target. In the second design, we introduce target-specific information (e.g., pixel values of unmasked patches) from the encoder directly to the decoder to reduce the pressure on the encoder of memorizing the target-specific information. Thus, the encoder focuses on semantic modeling, which is the goal of BERT pretraining, and does not need to waste its capacity in memorizing the information of unmasked tokens related to the prediction target. Through extensive experiments, our BootMAE achieves $84.2\%$ Top-1 accuracy on ImageNet-1K with ViT-B backbone, outperforming MAE by $+0.8\%$ under the same pre-training epochs. BootMAE also gets $+1.0$ mIoU improvements on semantic segmentation on ADE20K and $+1.3$ box AP, $+1.4$ mask AP improvement on object detection and segmentation on COCO dataset. Code is released at https://github.com/LightDXY/BootMAE. | ['Nenghai Yu', 'Fang Wen', 'Dong Chen', 'Lu Yuan', 'Weiming Zhang', 'Dongdong Chen', 'Ting Zhang', 'Jianmin Bao', 'Xiaoyi Dong'] | 2022-07-14 | null | null | null | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 1.82604611e-01 5.01604795e-01 -1.48422867e-01 -4.24269319e-01
-7.05124795e-01 -1.37243062e-01 2.01428935e-01 -3.96922708e-01
-7.56766438e-01 5.26536465e-01 -8.90650749e-02 -2.23327756e-01
4.35161740e-01 -8.97141516e-01 -1.42008758e+00 -7.10654318e-01
2.19023693e-02 3.97523582e-01 4.50223356e-01 -8.42852518e-02
-6.58578128e-02 5.99124702e-03 -1.46278894e+00 7.51000822e-01
6.14481509e-01 1.32510233e+00 7.40613878e-01 7.34458864e-01
-5.37742488e-02 9.86377180e-01 -5.59682488e-01 -5.01205385e-01
3.85629863e-01 -3.71507764e-01 -6.25214100e-01 -2.98192333e-02
5.23491979e-01 -5.27311265e-01 -5.46692491e-01 1.10304296e+00
2.90274799e-01 -2.02037737e-01 5.22557139e-01 -1.04463232e+00
-6.02719367e-01 1.04210138e+00 -7.40262091e-01 3.96127224e-01
-6.11231506e-01 3.29700172e-01 8.82556617e-01 -9.98127580e-01
3.73770148e-01 9.51834500e-01 6.17254555e-01 8.28968763e-01
-1.04268694e+00 -7.06716835e-01 2.11642385e-01 5.66829741e-01
-1.30408597e+00 -4.03123409e-01 4.72198904e-01 -4.01495755e-01
1.14248002e+00 5.55808432e-02 6.93383694e-01 9.56300259e-01
1.82916373e-01 1.17064977e+00 1.10894084e+00 -4.34275478e-01
2.41445154e-01 4.43914056e-01 2.84874171e-01 9.28815484e-01
-2.09262241e-02 3.82749140e-01 -6.23614848e-01 3.58192503e-01
7.43950248e-01 -3.66208881e-01 -2.70179600e-01 5.72318323e-02
-7.67110527e-01 8.33664656e-01 8.03452492e-01 -5.69334738e-02
-3.66650909e-01 5.35651505e-01 2.84605026e-01 2.19034463e-01
2.62534171e-01 1.64007232e-01 -7.29973495e-01 -1.56281486e-01
-1.05138838e+00 -2.96510279e-01 4.19993073e-01 7.83428073e-01
1.17188001e+00 3.63300025e-01 -1.93513870e-01 8.39507401e-01
3.60677421e-01 5.20452678e-01 7.14984238e-01 -6.60088778e-01
4.12175119e-01 5.79794466e-01 -2.67084479e-01 -4.38035518e-01
-7.15953112e-02 -5.91663897e-01 -7.00874388e-01 3.10009032e-01
1.98257610e-01 -2.31083885e-01 -1.60748661e+00 1.87216067e+00
-8.44558030e-02 4.05854344e-01 2.83167124e-01 8.03328574e-01
1.00416291e+00 8.67390931e-01 1.30242612e-02 1.93636432e-01
1.47532094e+00 -1.36615586e+00 -1.99058220e-01 -8.13360453e-01
4.65731740e-01 -5.79213679e-01 1.05052698e+00 3.11682940e-01
-1.08340836e+00 -7.81664252e-01 -1.17356479e+00 4.98180762e-02
-2.27023691e-01 5.31772375e-01 4.98507708e-01 5.77386916e-01
-1.23456848e+00 6.11012816e-01 -1.00369692e+00 7.65993372e-02
7.29998171e-01 6.76513493e-01 -7.74104446e-02 8.46493691e-02
-9.87422526e-01 9.74565208e-01 6.31519079e-01 1.25550821e-01
-1.31357014e+00 -7.50438988e-01 -8.57590616e-01 3.05455983e-01
4.40815240e-01 -3.59072149e-01 1.39160621e+00 -1.48797512e+00
-1.56229484e+00 7.89940655e-01 -1.83593094e-01 -9.50523257e-01
3.41873080e-01 -1.83458969e-01 -1.15967818e-01 1.93145588e-01
2.39130035e-02 1.49583304e+00 1.10918939e+00 -1.09251976e+00
-6.95112348e-01 7.29627982e-02 -1.87034875e-01 1.11428536e-01
-2.87622899e-01 -2.38327235e-01 -6.76087320e-01 -5.72652161e-01
-1.21245116e-01 -8.65665913e-01 -1.57124296e-01 -2.05191061e-01
-3.56769383e-01 -8.40160809e-03 7.95896709e-01 -7.80367315e-01
8.25551569e-01 -2.35681987e+00 -1.65273324e-01 4.70096618e-02
2.86510587e-01 6.50881946e-01 -2.77752846e-01 -1.62411362e-01
-4.52805795e-02 -1.40687630e-01 -2.54379630e-01 -5.00665843e-01
-2.32476339e-01 4.62702513e-01 -4.11376953e-01 9.76143628e-02
5.29011488e-01 1.29330790e+00 -5.32191098e-01 -3.41711044e-01
1.45512223e-01 2.99192846e-01 -5.64056456e-01 9.45871025e-02
-3.64927799e-01 5.50286733e-02 -5.55379651e-02 4.56289858e-01
6.64950013e-01 -4.62582231e-01 -6.62368461e-02 -1.76479593e-01
-5.71682006e-02 3.44589055e-01 -9.73203003e-01 1.37404847e+00
-2.59545833e-01 8.45115602e-01 1.41433612e-01 -1.09564924e+00
8.63855898e-01 2.80916363e-01 1.49723843e-01 -7.30918646e-01
3.03116143e-01 1.02121145e-01 1.71543419e-01 -1.04705036e-01
4.59241927e-01 2.05920283e-02 2.16113046e-01 2.70671815e-01
3.67037296e-01 2.35968739e-01 2.23147310e-03 2.16490790e-01
9.39734221e-01 5.80816604e-02 -4.66295369e-02 -1.58062771e-01
3.09367716e-01 3.07194352e-01 8.07930291e-01 7.88931191e-01
-2.60204911e-01 4.42830533e-01 3.53522569e-01 -3.20552021e-01
-1.00104964e+00 -1.04269111e+00 -5.05183674e-02 1.08230054e+00
1.35358036e-01 -3.77766937e-01 -8.90505910e-01 -8.05116057e-01
-1.61220327e-01 6.03179097e-01 -8.21528554e-01 -3.78777981e-01
-6.83928251e-01 -8.98945987e-01 6.22200549e-01 9.52450693e-01
9.83258486e-01 -1.27488124e+00 -8.12358797e-01 2.22730160e-01
-1.24467544e-01 -9.89738345e-01 -3.67037922e-01 6.62704229e-01
-7.94306993e-01 -7.73995936e-01 -5.59339762e-01 -8.34562480e-01
7.27280378e-01 1.53149202e-01 1.02198470e+00 -1.01504020e-01
-3.15050960e-01 1.65816262e-01 -3.55517328e-01 -7.85006166e-01
-4.68311012e-01 -4.69845980e-02 -1.99292392e-01 -3.58755738e-02
4.62961346e-01 -2.73389578e-01 -7.73241818e-01 4.01450068e-01
-6.41283154e-01 5.64181149e-01 9.48086560e-01 1.06202793e+00
7.69403875e-01 -1.22956313e-01 4.73445475e-01 -5.30062199e-01
-9.48581845e-02 -2.72757679e-01 -7.20767379e-01 1.06426850e-01
-6.65604472e-01 6.75202832e-02 5.26513398e-01 -6.02871597e-01
-8.37395370e-01 2.53169864e-01 -3.08749884e-01 -6.44366443e-01
7.42119700e-02 3.51767659e-01 -3.23797725e-02 -5.20280115e-02
6.92572594e-01 4.58805025e-01 7.16079399e-03 -5.81061900e-01
3.38311821e-01 6.09140873e-01 6.73247635e-01 -3.46605718e-01
6.03162169e-01 3.95072967e-01 -4.72595960e-01 -4.86726999e-01
-7.89569974e-01 -2.38063410e-01 -2.54034311e-01 -7.94235244e-02
9.65203822e-01 -1.21249866e+00 -5.80495179e-01 5.60121357e-01
-8.65747690e-01 -8.14850450e-01 -6.30097091e-01 4.64488208e-01
-5.23236096e-01 8.71355236e-02 -8.29119265e-01 -4.12424505e-01
-6.72619283e-01 -1.39571083e+00 5.51168203e-01 4.66411173e-01
6.56682476e-02 -3.77956152e-01 -2.97400773e-01 4.27046627e-01
4.89144027e-01 -3.22600156e-01 7.89916992e-01 -5.11952281e-01
-8.10144424e-01 3.36161517e-02 -5.59432387e-01 9.15597498e-01
-2.44081870e-01 -3.00207108e-01 -1.37010920e+00 -2.00109959e-01
6.38418049e-02 -4.31747347e-01 1.35875154e+00 6.75179899e-01
1.13425517e+00 -2.85324633e-01 -2.58283824e-01 8.12553823e-01
1.40709412e+00 3.23085845e-01 7.76386201e-01 4.20383722e-01
6.35281324e-01 1.93199113e-01 3.65893394e-01 1.18792988e-01
4.14766312e-01 4.96851474e-01 5.83839893e-01 -1.48696482e-01
-4.52367514e-01 -2.83589989e-01 6.92505538e-01 7.72087097e-01
5.36981747e-02 5.19905314e-02 -8.38351130e-01 6.69603705e-01
-1.84766614e+00 -6.87028289e-01 8.01481530e-02 1.94232655e+00
1.07909334e+00 3.84604067e-01 -4.67548110e-02 -1.79963082e-01
6.67251408e-01 -9.66085717e-02 -7.90201247e-01 -4.88588661e-01
-8.99059400e-02 6.45106375e-01 8.35835457e-01 4.24335092e-01
-1.18622649e+00 1.39197969e+00 5.28426886e+00 1.15268719e+00
-1.43939722e+00 2.99242139e-01 8.76061440e-01 -1.85007945e-01
2.07649440e-01 -7.79033005e-02 -1.14644027e+00 6.59665346e-01
1.08491898e+00 1.16040923e-01 3.87555629e-01 1.25764143e+00
-5.60197532e-02 -3.31608415e-01 -9.96468484e-01 9.09133196e-01
-1.15416594e-01 -1.57717586e+00 -1.33688211e-01 -3.94615019e-03
4.99682426e-01 4.81037676e-01 2.46648893e-01 6.27305388e-01
3.11883152e-01 -1.16072822e+00 1.01418757e+00 1.95490554e-01
8.35673153e-01 -4.47005302e-01 7.84037769e-01 4.23300117e-01
-1.02150393e+00 -9.29128155e-02 -6.78934872e-01 -5.57801574e-02
-6.95088739e-03 5.29214263e-01 -1.28102970e+00 5.74232154e-02
8.69440317e-01 4.61583704e-01 -6.32413447e-01 9.20402527e-01
-5.59051633e-01 1.18813515e+00 -3.06948364e-01 1.43071055e-01
3.65093529e-01 8.37053880e-02 3.72575700e-01 1.26655674e+00
4.27497402e-02 -2.96630543e-02 -4.98759523e-02 9.16266441e-01
-2.06382483e-01 -1.68150201e-01 5.71719483e-02 9.92620066e-02
2.99663424e-01 1.08454382e+00 -7.10895658e-01 -7.49014318e-01
-2.84488261e-01 1.07731080e+00 4.30962443e-01 3.26321721e-01
-1.09205973e+00 -2.02127725e-01 5.45618176e-01 -6.73426762e-02
9.27723527e-01 6.62860647e-02 -3.26837033e-01 -1.00749063e+00
-9.18363631e-02 -8.29296887e-01 3.60146701e-01 -8.29600155e-01
-9.23867166e-01 7.86367774e-01 -2.09283099e-01 -9.21993852e-01
-4.28174473e-02 -8.31476212e-01 -6.54351950e-01 9.16744411e-01
-1.64426100e+00 -1.12730241e+00 -2.30886355e-01 5.33966541e-01
6.35334074e-01 -6.07611500e-02 6.94303513e-01 1.81534737e-01
-8.07679534e-01 8.19219828e-01 -1.00756278e-02 3.95495892e-01
5.40132999e-01 -1.31707072e+00 4.99198854e-01 1.00456011e+00
3.40240121e-01 2.94706136e-01 3.69938314e-01 -5.75240970e-01
-9.83100712e-01 -1.30244279e+00 5.44902384e-01 -2.19334573e-01
3.73510599e-01 -4.70834076e-01 -8.89805138e-01 1.04263282e+00
8.04908052e-02 -7.29119405e-02 4.17871773e-01 -1.18633077e-01
-3.89952987e-01 -2.15999946e-01 -1.02254212e+00 4.44692969e-01
5.78765213e-01 -3.05528045e-01 -6.47160888e-01 1.23395637e-01
8.38056386e-01 -6.08233333e-01 -4.98931766e-01 4.17870045e-01
3.18317682e-01 -7.59819388e-01 9.45917130e-01 -1.74484506e-01
6.42345011e-01 -2.42519498e-01 -1.81068540e-01 -1.08014572e+00
-2.95501262e-01 -2.23187953e-01 -3.33773881e-01 9.22817349e-01
6.68535590e-01 -4.80122298e-01 1.29629958e+00 4.70083684e-01
-4.50331688e-01 -1.11866570e+00 -9.71854866e-01 -7.51466990e-01
1.83790419e-02 -6.71582818e-01 4.61282253e-01 5.75228810e-01
-4.65550274e-01 2.28548557e-01 -4.47835356e-01 2.92308897e-01
3.87096077e-01 -6.30689859e-02 5.95788538e-01 -6.13380134e-01
-7.85228670e-01 -3.99666756e-01 -3.91561538e-01 -1.44851661e+00
-2.30907947e-01 -9.42749679e-01 2.41704836e-01 -1.39629054e+00
3.41023237e-01 -5.05080402e-01 -4.98178750e-01 1.04204297e+00
-3.50196183e-01 5.82387567e-01 3.84561509e-01 1.76402450e-01
-3.05070043e-01 6.04682684e-01 1.04177582e+00 -2.56426096e-01
-3.17388505e-01 -6.28813282e-02 -6.76373184e-01 8.78416836e-01
9.39902782e-01 -5.94080091e-01 -3.78929585e-01 -6.75241530e-01
-7.59519786e-02 -3.58762264e-01 6.24813020e-01 -1.17632294e+00
9.62619632e-02 1.30992129e-01 5.93350291e-01 -6.85762584e-01
6.23822570e-01 -5.24779558e-01 -1.23799704e-01 7.05822110e-01
-4.33385521e-02 -2.29097426e-01 5.84868550e-01 4.06844437e-01
-1.43749043e-01 -4.10349369e-01 9.67025638e-01 -3.08020979e-01
-1.31326258e+00 2.25950122e-01 -3.92796427e-01 -8.31898898e-02
8.45799565e-01 -5.12385726e-01 -4.46873665e-01 -1.47933275e-01
-9.15823340e-01 3.09330434e-01 1.83061138e-01 9.75085199e-02
7.71739781e-01 -8.48881721e-01 -6.06960177e-01 2.45539591e-01
-1.36094779e-01 1.87041238e-01 3.60495955e-01 8.24105203e-01
-4.28438872e-01 2.53946036e-01 -2.36609191e-01 -7.15297520e-01
-1.24214685e+00 4.72387224e-01 4.75683808e-01 -2.47873202e-01
-7.67968833e-01 1.60856545e+00 4.31403607e-01 -7.73077011e-02
2.76205331e-01 -3.88037384e-01 2.17143893e-02 -1.57139316e-01
5.57111263e-01 1.02731444e-01 1.02811098e-01 -3.27746719e-01
-3.81991655e-01 2.56917953e-01 -5.15710652e-01 -1.84694618e-01
1.40097654e+00 1.02832846e-01 7.44678406e-03 9.24244076e-02
1.01252055e+00 -2.31378660e-01 -1.70078444e+00 -3.81602287e-01
-1.68432981e-01 -1.49679795e-01 2.06299245e-01 -1.15865910e+00
-1.28492677e+00 9.62488413e-01 7.66126692e-01 -5.23916245e-01
1.27749753e+00 1.89352721e-01 8.84165049e-01 2.17138395e-01
1.53207883e-01 -1.08486617e+00 2.04267710e-01 5.41727483e-01
6.15282118e-01 -1.17985201e+00 -2.84394830e-01 -3.35941702e-01
-8.64643037e-01 8.26297998e-01 1.04396188e+00 -1.96918890e-01
6.20207608e-01 2.86510110e-01 2.37955764e-01 -1.17025837e-01
-7.42996335e-01 -2.07819834e-01 3.19874555e-01 4.28685397e-01
-2.20952537e-02 2.56726801e-01 1.06189266e-01 9.17949259e-01
-4.88385916e-01 -9.56276283e-02 3.59804899e-01 7.62777686e-01
-6.49684012e-01 -9.59110737e-01 -2.98314840e-01 6.90013170e-01
-2.99651742e-01 -5.37272453e-01 -3.34503576e-02 6.28712952e-01
5.49847364e-01 7.56163061e-01 3.05434942e-01 -5.44768274e-01
-4.11774516e-02 2.14631572e-01 3.26225758e-01 -7.25015104e-01
-5.76313317e-01 1.22400001e-01 -1.20255351e-01 -5.14173746e-01
-4.15231176e-02 -2.60661840e-01 -1.36792374e+00 -1.46681398e-01
-3.64734262e-01 1.48137659e-01 6.88664198e-01 8.11405063e-01
5.70893407e-01 6.32740855e-01 3.08464706e-01 -6.32736266e-01
-5.70521176e-01 -1.02511048e+00 -3.83876979e-01 -4.73470092e-02
2.27796450e-01 -4.40220535e-01 -5.50922342e-02 1.73106715e-01] | [9.565206527709961, 0.8753634095191956] |
644fe61e-85de-47f1-8378-d1ab0ffda335 | entire-space-learning-framework-unbias | 2303.00276 | null | https://arxiv.org/abs/2303.00276v1 | https://arxiv.org/pdf/2303.00276v1.pdf | Entire Space Learning Framework: Unbias Conversion Rate Prediction in Full Stages of Recommender System | Recommender system is an essential part of online services, especially for e-commerce platform. Conversion Rate (CVR) prediction in RS plays a significant role in optimizing Gross Merchandise Volume (GMV) goal of e-commerce. However, CVR suffers from well-known Sample Selection Bias (SSB) and Data Sparsity (DS) problems. Although existing methods ESMM and ESM2 train with all impression samples over the entire space by modeling user behavior paths, SSB and DS problems still exist. In real practice, the online inference space are samples from previous stage of RS process, rather than the impression space modeled by existing methods. Moreover, existing methods solve the DS problem mainly by building behavior paths of their own specific scene, ignoring the behaviors in various scenes of e-commerce platform. In this paper, we propose Entire Space Learning Framework: Unbias Conversion Rate Prediction in Full Stages of Recommender System, solving SSB and DS problems by reformulating GMV goal in a novel manner. Specifically, we rebuild the CVR on the entire data space with samples from previous stage of RS process, unifying training and online inference space. Moreover, we explicitly introduce purchase samples from other scenes of e-commerce platform in model learning process. Online A/B test and offline experiments show the superiority of our framework. Our framework has been deployed in rank stage of Taobao recommendation, providing recommendation service for hundreds of millions of consumers everyday. | ['Junfeng Ge', 'Tao Zhuang', 'Qiwei Chen', 'Shanshan Lyu'] | 2023-03-01 | null | null | null | null | ['selection-bias'] | ['natural-language-processing'] | [-3.15725356e-01 -5.82896352e-01 -5.68337023e-01 -6.48329318e-01
-5.50390065e-01 -5.90541422e-01 2.33917281e-01 -2.93129802e-01
-5.20291887e-02 -3.16528883e-03 1.17794402e-01 -4.64656919e-01
-4.34474885e-01 -9.06759977e-01 -8.17417979e-01 -5.11752844e-01
-2.67866869e-02 5.13770878e-01 -4.96541932e-02 -5.94160318e-01
4.36988264e-01 2.42610589e-01 -1.26636648e+00 5.14836431e-01
8.60156775e-01 1.48666501e+00 4.02849793e-01 3.32004249e-01
-2.88668841e-01 6.44218266e-01 -1.93698868e-01 -6.82051480e-01
7.19568253e-01 -1.67474523e-01 -2.72265911e-01 2.50311524e-01
2.38444790e-01 -4.97154802e-01 -5.30942619e-01 1.04187739e+00
2.03571171e-01 4.18968409e-01 5.23688793e-01 -1.21520329e+00
-1.03655183e+00 7.85341740e-01 -5.58615625e-01 6.10921495e-02
8.64542127e-02 -1.22716673e-01 1.26822639e+00 -1.06265533e+00
4.43680018e-01 1.10972714e+00 5.47954977e-01 1.08799398e-01
-9.22529459e-01 -6.47436976e-01 8.25887978e-01 3.68998587e-01
-1.18617880e+00 2.38932073e-02 7.41415679e-01 -2.26156399e-01
4.00225103e-01 4.72552508e-01 8.91888678e-01 1.05711973e+00
-7.07173022e-03 1.51685119e+00 9.03152585e-01 2.43480965e-01
2.02945322e-01 6.24397397e-01 3.61992568e-01 2.32922927e-01
-3.22818682e-02 6.35720342e-02 -2.72763401e-01 7.91535527e-02
1.07213640e+00 6.55654490e-01 1.35123923e-01 -3.42931837e-01
-8.22747588e-01 9.54945266e-01 4.47106183e-01 -1.40183926e-01
-3.15135211e-01 -2.86504537e-01 1.07164428e-01 6.92443132e-01
2.89537638e-01 2.71975398e-01 -9.25144553e-01 4.49937396e-02
-7.14964032e-01 1.61454991e-01 8.83943319e-01 1.59637165e+00
4.33265179e-01 -6.00351654e-02 -8.60460177e-02 1.11368501e+00
6.62483394e-01 4.83030260e-01 6.73042357e-01 -5.58524728e-01
6.21115208e-01 5.59131861e-01 1.69211656e-01 -1.09455109e+00
-1.31042063e-01 -8.15317273e-01 -7.82962143e-01 -4.54377741e-01
3.92516464e-01 2.16001365e-02 -5.08769333e-01 1.08752882e+00
4.08436209e-01 2.48944625e-01 -2.62812495e-01 1.16003525e+00
6.93852603e-01 9.78507102e-01 -1.49991781e-01 -4.53393549e-01
1.18014240e+00 -1.33953202e+00 -5.65866053e-01 2.40350366e-02
5.88494003e-01 -7.37619519e-01 1.29962885e+00 1.08436465e+00
-8.97552669e-01 -8.35271955e-01 -8.39939475e-01 -1.36398762e-01
-1.41675860e-01 4.96676177e-01 1.20324481e+00 5.26196301e-01
-3.05606365e-01 6.92308545e-01 -3.49859387e-01 -1.86928958e-02
3.42134446e-01 3.58842611e-01 2.35898808e-01 -3.32883507e-01
-1.21890354e+00 3.26492608e-01 -2.42165029e-01 5.35704136e-01
-9.21583652e-01 -8.30806017e-01 -5.67562461e-01 1.45448446e-01
7.24149346e-01 -1.32455856e-01 1.27337074e+00 -1.09230256e+00
-1.71107376e+00 2.67569602e-01 2.39060804e-01 -2.69849390e-01
7.68993139e-01 -5.23544490e-01 -1.04318738e+00 -6.49206758e-01
-3.42398554e-01 -1.66592419e-01 5.66075385e-01 -1.34117162e+00
-9.61833000e-01 -5.98380029e-01 2.13781163e-01 2.80400842e-01
-1.96849301e-01 -8.74298140e-02 -8.89594138e-01 -7.05914676e-01
4.96459663e-01 -8.88772666e-01 -4.13100928e-01 -1.82560921e-01
-9.75678861e-02 -2.47092634e-01 4.40447122e-01 -8.72130752e-01
1.54520631e+00 -2.28773999e+00 -2.68994093e-01 5.56998909e-01
-8.89922231e-02 1.16647296e-01 -3.67268443e-01 1.33901775e-01
3.00606757e-01 -2.03495353e-01 4.91918951e-01 -7.88433775e-02
2.14113787e-01 1.99725285e-01 -5.31375706e-01 3.56846392e-01
-5.47964275e-01 7.51734793e-01 -8.67187858e-01 -2.42563501e-01
2.18013629e-01 2.41125926e-01 -7.55966246e-01 1.65412307e-01
-2.21516564e-01 2.60467321e-01 -6.84880078e-01 9.16647375e-01
9.80552614e-01 -3.55328500e-01 3.57739359e-01 -6.65625513e-01
-1.23634517e-01 3.60100687e-01 -1.57207525e+00 1.74340236e+00
-9.03386116e-01 -2.06600562e-01 1.26123160e-01 -1.02682209e+00
1.06103694e+00 -2.66578168e-01 7.09840059e-01 -1.06407034e+00
1.95322409e-01 1.10578395e-01 -5.98106114e-03 -5.68717599e-01
6.24172986e-01 -2.61287447e-02 5.55221748e-04 2.81791598e-01
8.01312327e-02 4.21378255e-01 -3.66775356e-02 1.10535748e-01
3.60642672e-01 3.74062479e-01 3.69071476e-02 -1.45577952e-01
3.11583787e-01 -2.18708172e-01 6.58259571e-01 8.37950826e-01
-3.51447985e-02 3.85166407e-01 1.52924761e-01 -4.56657350e-01
-9.54197586e-01 -1.07989609e+00 -2.40800813e-01 1.41187763e+00
6.75442040e-01 -4.70562875e-01 -1.90047354e-01 -9.37103093e-01
1.84739947e-01 8.64606798e-01 -4.59673613e-01 -5.25712781e-02
-5.67491472e-01 -8.12555313e-01 -4.11339283e-01 5.25680959e-01
4.50941861e-01 -6.50467753e-01 5.03649056e-01 3.74980271e-01
1.44334650e-02 -8.57087612e-01 -9.24156129e-01 -3.78324330e-01
-1.10326648e+00 -9.08494174e-01 -4.32768792e-01 -6.10151172e-01
4.73775536e-01 8.19524050e-01 1.05591488e+00 -3.36377993e-02
-1.27144620e-01 1.16774827e-01 -5.67400873e-01 -1.60632238e-01
-2.91403271e-02 -1.65638641e-01 3.60435769e-02 4.21369016e-01
6.39918208e-01 -2.53500104e-01 -1.01744032e+00 1.02428246e+00
-5.56130409e-01 -1.96437493e-01 6.76028609e-01 5.48996150e-01
8.22655678e-01 3.23444486e-01 4.79245961e-01 -1.30872190e+00
2.84930527e-01 -1.07026255e+00 -8.02660584e-01 3.52434665e-01
-1.06139851e+00 -4.55697805e-01 8.24207127e-01 -7.51832485e-01
-1.05092657e+00 -2.53374308e-01 -3.13719422e-01 -4.29933161e-01
1.76177397e-01 3.78155202e-01 -4.73901719e-01 4.68170106e-01
2.04767779e-01 3.61747444e-01 -1.16622508e-01 -9.48179305e-01
3.92605931e-01 9.11809921e-01 3.69476550e-03 -3.45680207e-01
5.25852680e-01 2.28930265e-01 -3.26098114e-01 -3.43717873e-01
-1.02744901e+00 -9.53049421e-01 -2.78234869e-01 -7.81099349e-02
1.56868234e-01 -1.05814719e+00 -1.03137159e+00 9.93167609e-02
-2.56636262e-01 -1.00024983e-01 -2.63561398e-01 7.37096906e-01
-2.47083053e-01 1.42061189e-01 -7.83563614e-01 -8.89732301e-01
-2.07204312e-01 -1.10634744e+00 9.27509427e-01 4.47525866e-02
3.26252937e-01 -9.36822593e-01 -3.33670259e-01 7.02117920e-01
3.39057833e-01 -4.99173939e-01 4.94568169e-01 -8.88504207e-01
-7.43779302e-01 -3.34826201e-01 -2.51842022e-01 5.80531538e-01
3.23421657e-02 -2.00890929e-01 -5.48935771e-01 -3.67707968e-01
2.32757777e-01 1.95626602e-01 6.16342783e-01 5.38428545e-01
1.57489765e+00 -5.65691888e-01 -8.63520205e-02 7.80749321e-01
1.56818414e+00 4.50859845e-01 6.35313988e-01 3.28042567e-01
7.97943532e-01 5.93521118e-01 1.31285715e+00 6.28442109e-01
6.24475956e-01 7.20342517e-01 2.64889568e-01 8.08730442e-03
1.12459026e-01 -6.85230792e-01 5.76206565e-01 1.07826698e+00
-1.46674767e-01 -2.05873236e-01 6.97268248e-02 5.23907356e-02
-2.00898528e+00 -8.64792407e-01 -3.70436341e-01 2.36837530e+00
4.18976545e-01 -5.96006997e-02 4.40482825e-01 -1.82540163e-01
5.01202226e-01 -1.97425708e-01 -8.05919349e-01 -6.92068338e-02
-5.71673736e-03 -3.64044130e-01 6.44693851e-01 3.04560572e-01
-8.91254961e-01 5.41147947e-01 5.39097691e+00 1.04540694e+00
-8.93832028e-01 1.82064176e-01 7.59679437e-01 -4.45367575e-01
-6.05713725e-01 4.18464653e-02 -1.18696630e+00 8.16691160e-01
6.08694494e-01 2.87980586e-01 8.49109232e-01 1.31294990e+00
2.69239217e-01 3.27469975e-01 -1.13327181e+00 1.34132290e+00
2.17731759e-01 -1.13241911e+00 7.27556050e-02 2.30810001e-01
7.71694303e-01 -1.15012929e-01 3.36290419e-01 1.03768563e+00
4.07104075e-01 -7.03837991e-01 8.15330207e-01 4.61851329e-01
5.48067570e-01 -5.51816463e-01 7.17723727e-01 5.00049174e-01
-1.13334417e+00 -6.83867455e-01 -6.58984303e-01 2.45622382e-01
2.68782228e-01 5.44829607e-01 -4.34034139e-01 5.06393850e-01
6.43398702e-01 9.88111556e-01 -2.02812433e-01 9.70467925e-01
1.96613461e-01 5.61640859e-01 -1.51284784e-01 -1.12663582e-01
1.37550354e-01 -1.00479293e+00 2.06428424e-01 8.84148002e-01
7.26116240e-01 2.26290047e-01 1.95319235e-01 6.76045120e-01
-1.89071342e-01 5.67406178e-01 8.77680853e-02 2.64978290e-01
3.30311328e-01 1.43307579e+00 -2.83208191e-01 -2.59474576e-01
-7.48047650e-01 8.52530122e-01 -1.39208347e-01 3.12301219e-01
-1.10109913e+00 1.68589711e-01 5.67785025e-01 4.94448394e-01
4.83493716e-01 1.36367500e-01 -1.15368463e-01 -1.52570105e+00
-7.64093101e-02 -1.13012934e+00 4.34682786e-01 -3.83373767e-01
-1.57519090e+00 3.73667777e-02 -3.78306329e-01 -1.74698853e+00
5.34896672e-01 -7.98123181e-01 -4.43887770e-01 5.75539887e-01
-1.52601755e+00 -1.25465715e+00 -1.34095997e-01 6.37340248e-01
1.19289649e+00 -4.23636466e-01 3.00569862e-01 8.25858474e-01
-7.21122801e-01 7.71085143e-01 7.45396912e-01 -2.34345153e-01
5.75628996e-01 -1.13931859e+00 2.64093965e-01 4.63619739e-01
3.10548782e-01 1.00511289e+00 5.01961112e-01 -6.49235964e-01
-1.95867717e+00 -1.09980249e+00 3.43282282e-01 -2.48438776e-01
8.16963255e-01 -5.05473673e-01 -6.32566929e-01 8.12363446e-01
-5.42545855e-01 -5.14642633e-02 9.46820498e-01 8.32500815e-01
-3.28784764e-01 -6.04283035e-01 -1.13943458e+00 4.52856213e-01
1.17668951e+00 -9.01000649e-02 -1.03112392e-01 6.24615848e-01
5.02357423e-01 -3.06249142e-01 -1.21352839e+00 2.04197556e-01
9.31274295e-01 -6.13727868e-01 1.25942266e+00 -8.35267901e-01
3.74893963e-01 -1.74432397e-01 -5.46413183e-01 -1.16787982e+00
-6.54127240e-01 -5.04026949e-01 -5.82771242e-01 1.21374094e+00
4.76622283e-01 -4.31021333e-01 8.33318472e-01 7.12233543e-01
-1.24119118e-01 -8.63144934e-01 -1.26413569e-01 -7.38968492e-01
-1.71888888e-01 -4.44168389e-01 9.07003999e-01 8.99334371e-01
-1.58334121e-01 4.44826812e-01 -7.85628676e-01 1.37569040e-01
6.94602907e-01 6.20567560e-01 9.85472620e-01 -1.08228600e+00
-1.00998509e+00 -3.45268190e-01 2.03127027e-01 -2.01455760e+00
-5.58862329e-01 -8.50801766e-01 -2.77128398e-01 -1.31843829e+00
1.83690816e-01 -1.08546269e+00 -6.07078612e-01 -3.44192564e-01
6.49922937e-02 -1.11547736e-02 2.72710115e-01 5.78104973e-01
-8.64734054e-01 3.44527662e-01 1.78932536e+00 2.20277384e-02
-2.80571520e-01 6.35373354e-01 -1.04383004e+00 5.51935494e-01
4.65022802e-01 -5.74903414e-02 -8.05990040e-01 -3.18269789e-01
5.99718332e-01 1.82718411e-01 -4.88439612e-02 -2.07439288e-01
8.80302489e-02 -4.00857806e-01 5.34474909e-01 -7.85947859e-01
2.67140925e-01 -9.82688665e-01 3.35223764e-01 4.96952794e-02
-4.23162848e-01 -2.11046264e-01 -4.32939589e-01 7.74399221e-01
-1.01026893e-01 -1.49201483e-01 3.72704655e-01 -2.05216110e-01
-6.97191954e-01 8.44240963e-01 2.22982258e-01 -4.00275767e-01
8.08559358e-01 -1.63249463e-01 -4.15579230e-03 -2.82291859e-01
-1.02096450e+00 3.90600681e-01 6.83878660e-02 7.21498966e-01
4.10023451e-01 -1.40399480e+00 -5.46977758e-01 3.76790673e-01
6.60995916e-02 -8.62900317e-02 7.79024720e-01 9.58443284e-01
-1.36842638e-01 9.00702029e-02 2.50090152e-01 -2.88872957e-01
-8.41989398e-01 9.08305705e-01 -7.17633462e-04 -3.00511271e-01
-5.01742899e-01 9.24408376e-01 4.04018849e-01 -6.79187357e-01
2.24006832e-01 -4.14219081e-01 -5.89627862e-01 7.64104873e-02
6.88517928e-01 5.86904824e-01 2.89448649e-02 -3.98109466e-01
2.50127673e-01 2.99299598e-01 -5.53150952e-01 4.59232032e-01
1.51205635e+00 -5.69471121e-01 3.77870768e-01 5.12353003e-01
1.35355413e+00 1.94608182e-01 -1.34449422e+00 -4.08764690e-01
-4.36082810e-01 -9.09287453e-01 1.75847575e-01 -9.09855723e-01
-1.31157637e+00 4.86637622e-01 5.90301216e-01 3.30320746e-01
1.01937222e+00 -1.73733771e-01 1.12408495e+00 1.80092737e-01
6.86317086e-01 -1.52156162e+00 -8.25074688e-02 1.03441633e-01
8.13704908e-01 -1.54339254e+00 4.90378588e-02 -5.45912743e-01
-1.05092657e+00 7.80202031e-01 6.13428831e-01 -4.78964567e-01
1.19101059e+00 -1.20566256e-01 -1.53538764e-01 -7.79688917e-03
-6.04479849e-01 3.24948639e-01 2.37057179e-01 1.67714387e-01
3.38970304e-01 3.61026645e-01 -3.15402955e-01 1.39051998e+00
-9.93366465e-02 9.68993641e-03 1.60462260e-01 5.13167500e-01
-2.08023563e-01 -1.08285725e+00 -7.86420144e-03 9.06346738e-01
-2.80124694e-01 -1.94468886e-01 4.35396194e-01 6.79385841e-01
8.43617022e-02 8.46281588e-01 9.02892426e-02 -6.76168203e-01
6.59188807e-01 -5.37695229e-01 4.95114177e-01 -4.14887995e-01
-5.75330496e-01 6.40948236e-01 4.93000448e-02 -8.15369129e-01
6.22491948e-02 -9.99798238e-01 -8.81744564e-01 -3.73283178e-01
-7.81748354e-01 2.08631814e-01 9.89678621e-01 7.29188502e-01
2.42562115e-01 3.37248683e-01 1.29470503e+00 -3.95780504e-01
-1.18111682e+00 -1.10035765e+00 -1.33444607e+00 7.19918072e-01
-3.81700997e-03 -6.28904521e-01 -3.77592444e-01 -1.32016554e-01] | [10.126322746276855, 5.577022552490234] |
c322c099-f7c1-47ae-bfd6-ae026a86c68e | automatic-gloss-level-data-augmentation-for | null | null | https://aclanthology.org/2022.lrec-1.734 | https://aclanthology.org/2022.lrec-1.734.pdf | Automatic Gloss-level Data Augmentation for Sign Language Translation | Securing sufficient data to enable automatic sign language translation modeling is challenging. The data insufficiency issue exists in both video and text modalities; however, fewer studies have been performed on text data augmentation compared to video data. In this study, we present three methods of augmenting sign language text modality data, comprising 3,052 Gloss-level Korean Sign Language (GKSL) and Word-level Korean Language (WKL) sentence pairs. Using each of the three methods, the following number of sentence pairs were created: blank replacement 10,654, sentence paraphrasing 1,494, and synonym replacement 899. Translation experiment results using the augmented data showed that when translating from GKSL to WKL and from WKL to GKSL, Bi-Lingual Evaluation Understudy (BLEU) scores improved by 0.204 and 0.170 respectively, compared to when only the original data was used. The three contributions of this study are as follows. First, we demonstrated that three different augmentation techniques used in existing Natural Language Processing (NLP) can be applied to sign language. Second, we propose an automatic data augmentation method which generates quality data by utilizing the Korean sign language gloss dictionary. Lastly, we publish the Gloss-level Korean Sign Language 13k dataset (GKSL13k), which has verified data quality through expert reviews. | ['Gahgene Gweon', 'Byungcheon Yoon', 'Suna Shin', 'Saim Shin', 'Han-Mu Park', 'Jin Yea Jang'] | null | null | null | null | lrec-2022-6 | ['sign-language-translation'] | ['computer-vision'] | [ 3.13281298e-01 -3.62990320e-01 -4.84664857e-01 -2.36954734e-01
-1.05058324e+00 -4.44075108e-01 4.64099348e-01 -3.74554068e-01
-8.51325154e-01 1.00261784e+00 1.03071988e+00 -2.15943947e-01
2.37471819e-01 -2.83568621e-01 -4.90952581e-01 -3.32667500e-01
4.34911042e-01 1.48893848e-01 -1.15254916e-01 -2.28037626e-01
1.58465385e-01 6.41450705e-03 -1.38309681e+00 5.45439363e-01
1.15210938e+00 5.15505910e-01 -7.01541919e-03 6.89178348e-01
-3.70984107e-01 6.23179138e-01 -5.54487884e-01 -5.24261713e-01
4.14140671e-01 -8.07676077e-01 -5.80614984e-01 3.80547089e-03
9.46239114e-01 -5.91845274e-01 -3.90060991e-01 9.71864343e-01
7.97132790e-01 -8.95260870e-02 4.15888339e-01 -1.47070658e+00
-1.09750772e+00 5.45584142e-01 -2.24027142e-01 -2.35435843e-01
5.57793438e-01 3.71279031e-01 7.68991649e-01 -1.11713839e+00
1.11571538e+00 9.47500229e-01 5.23527026e-01 9.19113517e-01
-7.80011117e-01 -8.54020417e-01 7.16182068e-02 2.36072391e-01
-1.04207194e+00 -2.48345926e-01 5.89900136e-01 -2.92536557e-01
9.80877995e-01 1.02237850e-01 1.19523799e+00 1.25051796e+00
8.37116987e-02 1.06234813e+00 1.60037315e+00 -7.34566867e-01
-1.42553553e-01 -4.10456583e-02 3.43005598e-01 7.47616172e-01
5.22925735e-01 1.12432994e-01 -1.15065670e+00 1.46952597e-02
5.13627470e-01 -4.67626005e-01 -2.15627730e-01 2.94125360e-02
-1.65727079e+00 4.79830265e-01 -1.64496139e-01 2.83937335e-01
-3.40315104e-01 -4.18174118e-02 6.28943324e-01 7.18469024e-01
6.73882850e-03 2.24146485e-01 -5.79226851e-01 -6.30254984e-01
-6.58492208e-01 2.45167792e-01 4.15168673e-01 1.39951146e+00
2.27782540e-02 1.54931337e-01 -2.46019289e-01 7.70392299e-01
4.90760505e-01 1.23268127e+00 9.39987242e-01 -6.87867165e-01
9.16345358e-01 7.75955796e-01 -1.06331110e-01 -5.47000706e-01
-3.11978966e-01 2.98036486e-01 -8.13069105e-01 6.78043514e-02
3.58090401e-01 -1.18570976e-01 -1.50808156e+00 1.83260465e+00
-1.90932691e-01 -3.18107754e-01 5.15696645e-01 9.17944491e-01
1.28128421e+00 2.61111945e-01 3.44238073e-01 -4.39492345e-01
1.52771807e+00 -1.02483952e+00 -1.25817096e+00 7.86740333e-02
1.10890639e+00 -1.16439295e+00 1.80785048e+00 4.19869125e-01
-1.01841509e+00 -4.19389993e-01 -7.49992549e-01 -2.70517617e-01
-5.54623663e-01 4.88732576e-01 4.72424746e-01 6.30874395e-01
-1.02942157e+00 -2.07887843e-01 -6.21451557e-01 -6.46789253e-01
1.20535515e-01 1.22705773e-01 -7.55428910e-01 -2.32595608e-01
-1.28022635e+00 1.12494087e+00 3.59547675e-01 -9.35005620e-02
-8.31781700e-02 -4.65642035e-01 -8.04981828e-01 -7.37406731e-01
3.28899473e-02 -8.20769489e-01 1.13576138e+00 -8.99700582e-01
-1.63168466e+00 8.35979462e-01 -4.76910412e-01 -1.48304224e-01
5.01956701e-01 -4.53124076e-01 -8.50063562e-01 -1.08208612e-03
-5.01704291e-02 7.08876729e-01 5.44171035e-01 -8.69376600e-01
-6.99689388e-01 -2.04365268e-01 -1.68308794e-01 5.68367183e-01
-3.59196693e-01 3.59513938e-01 -6.13074660e-01 -1.12118530e+00
2.62058109e-01 -9.95786905e-01 2.88184956e-02 2.07083732e-01
-1.21142678e-01 -1.02259934e-01 5.13380468e-01 -1.35434663e+00
1.67985570e+00 -1.93710363e+00 1.82281986e-01 1.55104399e-01
-5.87167628e-02 6.36231244e-01 -5.14668405e-01 4.85375255e-01
6.62434250e-02 2.56473631e-01 -3.65976274e-01 -3.11686605e-01
-8.35882500e-02 5.09640396e-01 -8.02539065e-02 -3.40007581e-02
-1.69371925e-02 1.35224724e+00 -8.51468086e-01 -7.39246249e-01
3.53044331e-01 2.80914277e-01 -6.87213182e-01 -1.93152860e-01
1.54933304e-01 3.79618019e-01 -2.85918452e-02 1.11714911e+00
4.53194261e-01 2.66788870e-01 -1.67151049e-01 -4.75069791e-01
-1.40803471e-01 1.90603733e-01 -9.98167634e-01 1.92995334e+00
-3.34662616e-01 7.78439820e-01 -4.45363075e-01 -8.57061222e-02
6.22382939e-01 5.02600312e-01 6.57957315e-01 -9.32634294e-01
2.11469188e-01 5.24983346e-01 -1.12929188e-01 -1.17263031e+00
6.68562710e-01 -1.66296631e-01 -1.77142531e-01 4.34369564e-01
-6.45680651e-02 -2.88468897e-01 6.12394214e-01 2.26820439e-01
8.39106441e-01 1.27490267e-01 3.49704862e-01 3.97943825e-01
5.31244218e-01 3.39200497e-01 4.83086228e-01 4.87822473e-01
-3.75917345e-01 5.48872709e-01 -2.09796578e-02 -1.36081278e-01
-1.26174915e+00 -1.02193058e+00 1.54972017e-01 5.20662487e-01
-2.67526656e-01 -7.34189808e-01 -5.17630279e-01 -7.72303522e-01
-1.95192993e-01 7.13211119e-01 -1.85953543e-01 -1.13089629e-01
-8.13941896e-01 -5.75890005e-01 1.08764899e+00 7.66569555e-01
8.50335240e-01 -1.18099129e+00 -1.91105008e-01 -2.57461928e-02
-7.46988058e-01 -1.30865276e+00 -8.89308274e-01 -5.80956936e-01
-9.78989363e-01 -1.14638841e+00 -8.35221708e-01 -1.20424819e+00
9.40040410e-01 5.77417761e-02 7.32160568e-01 -1.25608683e-01
-6.19725436e-02 7.42947459e-01 -8.96998107e-01 -4.14653331e-01
-6.31984055e-01 -2.68324435e-01 4.57480848e-01 -4.34720457e-01
9.29749846e-01 -4.30262759e-02 -5.57572721e-03 1.71123341e-01
-1.05590796e+00 4.71026659e-01 9.74607289e-01 1.08633244e+00
5.48719704e-01 -6.56136513e-01 3.38272601e-01 -1.64404824e-01
1.05054379e+00 3.37919623e-01 -3.31175327e-01 5.91195285e-01
-8.51707876e-01 1.45232514e-01 3.18927556e-01 -7.27819860e-01
-7.88986623e-01 1.05325557e-01 -2.66040117e-01 -6.80511892e-02
-9.20045450e-02 8.04207265e-01 -9.81261283e-02 -9.80200991e-02
3.50692868e-01 6.71391487e-01 3.22487980e-01 -5.41982353e-01
6.04548991e-01 1.11973464e+00 5.46900272e-01 -3.29201996e-01
7.93434143e-01 1.37961879e-01 -2.72756070e-01 -1.00552797e+00
-1.89671129e-01 -3.40246141e-01 -8.48301470e-01 -4.96937543e-01
8.04685533e-01 -7.49113441e-01 -4.79094565e-01 1.03138781e+00
-1.00629234e+00 -8.78687277e-02 -6.67098284e-01 1.16537201e+00
-5.85569084e-01 7.53710687e-01 -2.80398548e-01 -5.21928608e-01
-4.71783727e-01 -9.92624640e-01 9.04957891e-01 -7.93678388e-02
-3.18934023e-01 -3.95387173e-01 8.03679749e-02 8.44034731e-01
2.55664498e-01 3.36142629e-02 8.68075073e-01 -2.47617185e-01
-2.18151912e-01 -5.12860775e-01 -2.43318751e-01 7.70295978e-01
5.92131078e-01 -2.71946546e-02 -1.96341664e-01 -2.11935848e-01
-3.72426540e-01 -4.80823755e-01 4.74656880e-01 2.27539137e-01
5.36999047e-01 -4.20882910e-01 4.08740640e-01 4.09633785e-01
1.17638946e+00 1.98049948e-01 8.56996655e-01 2.90112406e-01
7.42513716e-01 2.65208632e-01 9.18124199e-01 5.11183590e-02
5.62025309e-01 7.68110931e-01 -3.45831960e-01 -2.55105421e-02
-8.04984570e-01 -6.51131034e-01 7.98709512e-01 1.69084394e+00
-5.36698639e-01 -8.78675431e-02 -6.90497100e-01 5.26258349e-01
-1.65315318e+00 -7.91277647e-01 -5.16418874e-01 2.14400840e+00
1.02374697e+00 -1.91826105e-01 6.01571873e-02 2.47658834e-01
1.58689126e-01 -2.03433141e-01 -3.61755967e-01 -2.41051912e-01
-6.83854103e-01 4.86398861e-02 6.20681703e-01 7.09420919e-01
-7.66506374e-01 1.25533807e+00 6.43335772e+00 3.95680666e-01
-9.15416300e-01 -4.53161970e-02 -3.79590809e-01 -3.85996372e-01
-3.39285284e-01 -1.30627736e-01 -7.86484182e-01 6.01129055e-01
7.05981135e-01 -3.40526164e-01 4.39718127e-01 3.24889421e-01
7.21514940e-01 -1.49089277e-01 -6.81951225e-01 1.40556264e+00
6.13907218e-01 -1.07788229e+00 8.00140381e-01 1.08917383e-02
8.18685532e-01 3.17589566e-02 -1.62366778e-01 3.80649984e-01
-8.62783566e-03 -6.32219136e-01 5.46895862e-01 8.18651080e-01
1.08840764e+00 -1.43423200e-01 9.58189130e-01 1.03660047e-01
-1.23221791e+00 1.83288142e-01 1.71337530e-01 4.93259914e-02
6.99414492e-01 -8.33924860e-02 -5.37138104e-01 5.72990656e-01
4.18958962e-01 6.08445406e-01 -7.03298330e-01 9.53669429e-01
-4.71687347e-01 6.32585764e-01 -5.52038908e-01 -2.09130198e-01
7.13225231e-02 -1.53176129e-01 6.67285800e-01 1.00437427e+00
3.60036910e-01 2.06567958e-01 -1.59209892e-01 7.60304481e-02
-2.26122625e-02 6.75459683e-01 -7.12995052e-01 -4.49090362e-01
1.29667968e-01 4.04087573e-01 -1.75313354e-01 -6.21594667e-01
-7.82057345e-01 1.20632231e+00 -3.10516477e-01 5.72063923e-01
-6.74945831e-01 -9.75822583e-02 6.07581854e-01 -1.12829991e-01
-4.23198938e-01 -5.73609531e-01 -5.49793601e-01 -1.56340754e+00
6.47136509e-01 -1.22071683e+00 5.04593670e-01 -9.93870378e-01
-1.32402813e+00 2.98146725e-01 1.48525044e-01 -1.86987662e+00
-1.30619556e-01 -7.57469058e-01 2.54319519e-01 8.89760852e-01
-1.43598461e+00 -1.62887895e+00 -4.80895340e-01 7.21405149e-01
7.55796850e-01 -4.78817582e-01 8.79120171e-01 6.89910293e-01
-9.39527303e-02 9.60503697e-01 -5.09274565e-02 3.61534148e-01
9.93503749e-01 -7.53444850e-01 2.46753797e-01 1.05751264e+00
1.86341420e-01 6.80814862e-01 2.79200494e-01 -1.12403393e+00
-1.40933514e+00 -8.74447048e-01 1.80631661e+00 -5.44119716e-01
5.09525239e-01 9.15154889e-02 -5.53969562e-01 4.53400254e-01
1.00052483e-01 -4.15298581e-01 8.34145188e-01 -5.09974778e-01
-1.32350832e-01 2.25995883e-01 -1.15289450e+00 1.14999413e+00
1.42237055e+00 -5.34134626e-01 -8.89439106e-01 3.04548353e-01
7.16272891e-01 -5.23549139e-01 -9.04618204e-01 5.25210381e-01
1.01883745e+00 -6.62774816e-02 6.64683819e-01 -1.23556089e+00
4.76323009e-01 -5.63669920e-01 -4.51105028e-01 -1.00103295e+00
1.04738623e-01 -2.66326129e-01 -2.13213980e-01 9.88468766e-01
5.31873405e-01 -5.50613344e-01 7.13321149e-01 9.30179060e-01
-3.28427136e-01 -4.95550543e-01 -1.05983639e+00 -1.00180459e+00
-6.96393698e-02 -6.98002160e-01 3.27585727e-01 9.11063373e-01
4.92997617e-02 -2.35274900e-02 -8.41057479e-01 -1.85901612e-01
3.97978574e-01 -3.33514273e-01 9.86966014e-01 -6.58736706e-01
1.48767352e-01 -4.47276771e-01 -5.87808132e-01 -1.04210222e+00
-2.35367462e-01 -1.04010725e+00 -1.73074290e-01 -2.03628230e+00
5.24337701e-02 1.06654473e-01 -7.25073218e-02 9.09044623e-01
-7.31700435e-02 5.23320198e-01 4.45534199e-01 2.45301351e-01
5.90827242e-02 6.68562889e-01 1.59773493e+00 -1.79592848e-01
-5.21058023e-01 -2.13107362e-01 -4.24623072e-01 7.51897037e-01
7.78483808e-01 -1.57723516e-01 -1.90263093e-01 -6.53428316e-01
1.00105353e-01 -4.65913981e-01 2.88250089e-01 -7.37836480e-01
2.28816658e-01 -2.92963475e-01 -7.82305077e-02 -9.68100607e-01
2.24661916e-01 -9.94926453e-01 -8.34645405e-02 5.81258655e-01
-4.75765616e-01 3.01562488e-01 3.06282192e-01 1.97901756e-01
-3.63674551e-01 -6.95862398e-02 4.49060380e-01 7.13443309e-02
-1.03823221e+00 -7.91844446e-03 -7.76403189e-01 2.25469679e-01
8.39259028e-01 -5.47027588e-01 -2.78067559e-01 -4.20399398e-01
-7.07372367e-01 2.34090686e-01 3.29711348e-01 9.63858426e-01
1.05340755e+00 -1.87180245e+00 -8.61677349e-01 4.71134752e-01
5.81338584e-01 -5.11489451e-01 1.30060151e-01 1.10640514e+00
-5.85477471e-01 5.82954109e-01 -4.40451145e-01 -1.14381582e-01
-1.95018780e+00 6.77570421e-03 -2.16175586e-01 1.81962620e-03
-7.37397730e-01 4.56180304e-01 -8.07788670e-01 -6.78831339e-01
4.93186980e-01 -9.36171114e-01 -1.08707570e-01 5.51434830e-02
3.98470849e-01 3.30177158e-01 -1.10526279e-01 -8.73465240e-01
-2.64484048e-01 9.23615992e-01 -5.97225875e-03 -5.72043002e-01
9.16862190e-01 -2.08319917e-01 -1.83448866e-02 4.28033441e-01
8.99313271e-01 1.57296155e-02 -2.12606728e-01 -3.60746413e-01
-1.10103168e-01 -5.70318818e-01 -2.82832265e-01 -1.31409669e+00
-6.30842268e-01 4.04671341e-01 8.90916526e-01 -7.50626266e-01
1.44760549e+00 -2.68780231e-01 1.07418227e+00 6.76888406e-01
4.80527341e-01 -1.58873653e+00 -1.13336667e-01 7.54113615e-01
1.23813987e+00 -1.41386843e+00 -9.92185101e-02 -1.94608510e-01
-9.96987104e-01 9.66915607e-01 5.69474518e-01 3.30784678e-01
4.60242420e-01 1.62677333e-01 6.09413385e-01 7.49830380e-02
-3.23371917e-01 -2.93097854e-01 4.96242553e-01 6.13500893e-01
4.54420716e-01 -3.39673199e-02 -1.33359218e+00 5.37067533e-01
-4.96183902e-01 6.55443966e-01 6.13128901e-01 1.25476110e+00
-4.78172296e-05 -1.51529205e+00 -3.85857165e-01 7.62156129e-01
1.24227375e-01 -5.09355426e-01 -7.14563549e-01 9.29725885e-01
1.50772676e-01 7.96589315e-01 -4.16158110e-01 -5.63238204e-01
7.43135750e-01 5.10352135e-01 5.81349075e-01 -3.71534437e-01
-3.98852915e-01 -1.11595653e-01 3.34152281e-01 -4.11555290e-01
-8.98266733e-01 -8.17658782e-01 -1.26437569e+00 -1.06217392e-01
-2.42677674e-01 -2.23236784e-01 7.77760684e-01 1.02117705e+00
1.14709102e-01 3.28985393e-01 -9.29123089e-02 -8.70696455e-02
-3.23416084e-01 -1.20396054e+00 -3.23635161e-01 5.74620187e-01
7.24645481e-02 -3.92173678e-01 -2.46423557e-01 6.54280782e-01] | [9.17392349243164, -6.496230602264404] |
1a1e4505-9e09-4aa5-b4af-a533dba3a48f | building-low-resource-ner-models-using-non | 2006.09627 | null | https://arxiv.org/abs/2006.09627v2 | https://arxiv.org/pdf/2006.09627v2.pdf | Building Low-Resource NER Models Using Non-Speaker Annotation | In low-resource natural language processing (NLP), the key problems are a lack of target language training data, and a lack of native speakers to create it. Cross-lingual methods have had notable success in addressing these concerns, but in certain common circumstances, such as insufficient pre-training corpora or languages far from the source language, their performance suffers. In this work we propose a complementary approach to building low-resource Named Entity Recognition (NER) models using ``non-speaker'' (NS) annotations, provided by annotators with no prior experience in the target language. We recruit 30 participants in a carefully controlled annotation experiment with Indonesian, Russian, and Hindi. We show that use of NS annotators produces results that are consistently on par or better than cross-lingual methods built on modern contextual representations, and have the potential to outperform with additional effort. We conclude with observations of common annotation patterns and recommended implementation practices, and motivate how NS annotations can be used in addition to prior methods for improved performance. For more details, http://cogcomp.org/page/publication_view/941 | ['Dan Roth', 'Tatiana Tsygankova', 'Stephen Mayhew', 'Francesca Marini'] | 2020-06-17 | null | null | null | null | ['low-resource-named-entity-recognition'] | ['natural-language-processing'] | [ 1.46383733e-01 2.10283458e-01 -2.08902031e-01 -7.97753930e-01
-1.47955120e+00 -8.19836915e-01 4.95538235e-01 3.82570863e-01
-9.17688131e-01 9.03721690e-01 7.27065384e-01 -4.64406997e-01
2.83759445e-01 -1.34580553e-01 -3.47426414e-01 -2.56326169e-01
3.96354586e-01 5.50241292e-01 -1.18128052e-02 -2.35366210e-01
5.85412532e-02 2.62773663e-01 -9.50272977e-01 2.31139690e-01
7.58056164e-01 2.29364768e-01 2.71338999e-01 3.08498770e-01
-4.45650578e-01 8.10830653e-01 -6.90519333e-01 -5.87836862e-01
2.52914261e-02 -2.98691988e-01 -1.12633622e+00 -2.99838006e-01
4.57617700e-01 2.49072149e-01 2.15921670e-01 9.02871668e-01
7.67001867e-01 3.48887593e-01 3.16926181e-01 -6.63420260e-01
-7.44323850e-01 1.00218403e+00 -2.23535031e-01 3.25719535e-01
4.79944348e-01 -1.07044064e-01 1.03275013e+00 -9.25313771e-01
8.58737707e-01 1.12104440e+00 9.97762382e-01 8.74719262e-01
-1.43190789e+00 -5.59996605e-01 1.41314030e-01 -2.60524303e-01
-1.49671614e+00 -1.09597588e+00 4.42782313e-01 -2.83778936e-01
1.15550995e+00 8.15175399e-02 -1.05262429e-01 1.17699611e+00
-3.18331629e-01 5.44516742e-01 1.13175023e+00 -1.00222790e+00
3.79466303e-02 5.68985403e-01 1.81451872e-01 3.35784644e-01
1.82552531e-01 -2.61129826e-01 -6.12639308e-01 -2.47832894e-01
4.33654398e-01 -6.39055014e-01 -2.65890151e-01 2.67259348e-02
-1.20665717e+00 7.57558823e-01 -1.52856139e-02 9.69650865e-01
-2.72618890e-01 -2.44398639e-01 6.77037299e-01 1.40114933e-01
5.69136143e-01 8.14011276e-01 -1.04676831e+00 -3.58958840e-01
-8.60559285e-01 -9.56658348e-02 1.12605870e+00 1.03445220e+00
5.92570245e-01 1.45026043e-01 1.28433242e-01 1.29601812e+00
1.84788361e-01 3.14970195e-01 8.18760991e-01 -8.26048195e-01
6.30280495e-01 2.94145256e-01 3.60053688e-01 -6.98435128e-01
-4.58665907e-01 -9.81461182e-02 -3.14841300e-01 -1.81080237e-01
6.89605474e-01 -4.52911764e-01 -6.26866043e-01 1.96622610e+00
2.05276847e-01 -3.52685273e-01 4.33074445e-01 4.74300623e-01
5.49592614e-01 5.26601434e-01 8.66702676e-01 -3.31381589e-01
1.66219950e+00 -6.78392351e-01 -7.41832137e-01 -6.21817291e-01
1.06505108e+00 -1.13236916e+00 1.23823142e+00 -2.23705489e-02
-1.03821266e+00 -5.04790664e-01 -6.38630211e-01 -3.51464242e-01
-6.18646920e-01 3.61802667e-01 5.99211693e-01 1.01922941e+00
-1.02366972e+00 4.59201097e-01 -7.88925350e-01 -8.69473279e-01
-8.27735439e-02 2.06831172e-02 -6.35107040e-01 -9.48193222e-02
-1.23314250e+00 1.29750144e+00 5.32212019e-01 1.77178070e-01
-3.01358908e-01 -6.16232574e-01 -1.07201958e+00 -2.78980404e-01
3.72637421e-01 6.72744662e-02 1.54652441e+00 -9.65794802e-01
-1.23535049e+00 1.20147097e+00 -3.32724273e-01 -1.17408328e-01
2.25314379e-01 -6.05339527e-01 -7.25208640e-01 -3.07226658e-01
5.36505580e-01 4.91654634e-01 7.12513402e-02 -1.02845907e+00
-6.64350212e-01 -4.21810985e-01 -2.05307409e-01 3.39274108e-01
-2.71419019e-01 9.85390961e-01 -1.65334597e-01 -5.18687069e-01
-1.70941204e-01 -9.41910326e-01 -3.63645494e-01 -5.99421620e-01
-1.36667535e-01 -5.44188797e-01 3.45293581e-01 -1.03155112e+00
1.12327361e+00 -2.12101531e+00 -3.66154701e-01 -4.01452780e-02
-3.23826522e-01 5.30186117e-01 -7.11646006e-02 6.23786151e-01
-1.04238354e-01 5.54079056e-01 -1.54953167e-01 -4.16813642e-01
8.92496631e-02 3.59797567e-01 -1.20200172e-01 3.91868055e-01
2.86596924e-01 6.08226120e-01 -9.74235117e-01 -6.14477754e-01
-7.68622011e-02 5.98484337e-01 -1.42211793e-02 -1.74885709e-02
6.84230477e-02 5.67594349e-01 -2.22002432e-01 3.99368167e-01
1.86707735e-01 1.31650075e-01 6.09673619e-01 9.33001116e-02
-6.35588229e-01 1.03883648e+00 -1.31380081e+00 1.77391577e+00
-7.16202557e-01 5.28723598e-01 3.83967936e-01 -7.67421126e-01
8.06033075e-01 6.93871856e-01 -1.13960192e-01 -4.85818535e-01
3.93857993e-02 5.61676025e-01 2.53323186e-02 -4.18205082e-01
6.47773802e-01 -3.46721709e-01 -4.21760648e-01 5.35886168e-01
1.57357454e-01 4.08831500e-02 3.27815920e-01 -1.02880634e-01
1.00647771e+00 3.99390012e-01 7.87017703e-01 -5.16570330e-01
4.96782362e-01 3.37953776e-01 1.00466132e+00 7.22569227e-01
-4.94354367e-01 1.86913475e-01 2.29289100e-01 -2.01685384e-01
-1.19938791e+00 -6.07871234e-01 -4.10073519e-01 1.68363249e+00
-5.10850370e-01 -4.89869356e-01 -7.30494320e-01 -5.82295895e-01
-5.92738450e-01 1.12187874e+00 -3.38175595e-01 3.77859861e-01
-1.11069953e+00 -5.25866508e-01 1.10547006e+00 7.55424082e-01
2.42360637e-01 -1.37711883e+00 -3.77099097e-01 4.98309076e-01
-4.65944558e-01 -1.23764610e+00 -3.54746968e-01 3.85330230e-01
-5.36909699e-01 -6.63174570e-01 -5.70065558e-01 -1.15148401e+00
4.74499375e-01 -1.06697172e-01 1.14911914e+00 -7.37593994e-02
6.71545556e-03 3.29842269e-01 -4.94694084e-01 -4.89711970e-01
-6.64769351e-01 3.56796682e-01 2.11911768e-01 -5.05721331e-01
9.11975801e-01 -2.48463124e-01 4.90355156e-02 3.19605321e-01
-7.16226578e-01 -2.59769082e-01 5.03082395e-01 7.39109457e-01
3.91921997e-01 -1.67032778e-01 6.90978169e-01 -1.33675694e+00
5.28110981e-01 -3.28508824e-01 -3.34615976e-01 3.32397580e-01
-3.98285270e-01 1.56867914e-02 5.86788237e-01 -2.75038391e-01
-1.58753669e+00 2.55602807e-01 -5.18557608e-01 2.77162224e-01
-7.56458104e-01 6.60393775e-01 -3.97729129e-01 2.71072894e-01
1.08548725e+00 -2.06114694e-01 -3.30496311e-01 -9.07859623e-01
3.89523596e-01 7.47792244e-01 4.89461541e-01 -8.63886774e-01
5.72828293e-01 -1.18569262e-01 -7.03777015e-01 -1.04507565e+00
-1.06237292e+00 -6.41218603e-01 -9.13778484e-01 2.51598209e-01
8.78432155e-01 -1.01915848e+00 -1.48861736e-01 -7.75559014e-03
-1.22831082e+00 -5.64159811e-01 -3.31598520e-01 6.76809430e-01
-2.40726158e-01 1.50620401e-01 -7.31847465e-01 -8.13717127e-01
-2.40917772e-01 -6.42674804e-01 7.78780758e-01 2.67384052e-01
-7.62716591e-01 -1.27385259e+00 1.96018934e-01 4.73017246e-01
4.73614395e-01 -3.50186899e-02 8.31468999e-01 -1.44268656e+00
2.47017413e-01 -3.12611610e-01 4.21774797e-02 4.74051535e-01
1.61000013e-01 -2.45269269e-01 -1.03308558e+00 -1.53149022e-02
-1.13717755e-02 -5.32853901e-01 1.12165757e-01 -2.55077779e-02
3.14732134e-01 -2.66876429e-01 -2.90948659e-01 -6.31216988e-02
1.36783826e+00 1.53881371e-01 3.74223381e-01 3.07715088e-01
6.47420168e-01 9.70370650e-01 4.23151791e-01 2.09715106e-02
4.70715106e-01 5.68621099e-01 -6.09366477e-01 9.55510661e-02
-2.14505848e-02 -3.69738638e-01 4.86600727e-01 9.88613784e-01
-1.59785487e-02 -1.35696590e-01 -1.32627046e+00 8.51984978e-01
-1.59313881e+00 -8.78967047e-01 -3.43880691e-02 2.18150091e+00
1.14763105e+00 -2.54091751e-02 1.59665886e-02 -2.42363274e-01
9.75490630e-01 3.11398059e-02 -6.69913739e-02 -6.20074391e-01
-2.14247242e-01 2.98753977e-01 4.63077962e-01 5.96147478e-01
-1.13173687e+00 1.35728955e+00 6.10072803e+00 6.57140970e-01
-8.51853251e-01 4.91615176e-01 3.58759969e-01 3.28014225e-01
-1.09689310e-01 2.22076789e-01 -1.26086748e+00 1.43783599e-01
1.60536087e+00 -1.94359988e-01 2.18920067e-01 9.67892468e-01
9.33150426e-02 1.03414111e-01 -9.94242132e-01 6.02508783e-01
2.71634132e-01 -8.14035237e-01 -5.62609494e-01 -1.66609854e-01
4.77011919e-01 4.12907511e-01 -5.86525440e-01 5.81531763e-01
7.75711179e-01 -7.85945892e-01 7.37600982e-01 -6.25267858e-03
7.74231493e-01 -4.98900265e-01 9.06043649e-01 4.21559006e-01
-1.16193795e+00 2.46335596e-01 -4.48677272e-01 6.60969242e-02
5.14137506e-01 1.55124709e-01 -8.49605322e-01 4.62799102e-01
6.27365589e-01 2.06761181e-01 -5.36140501e-01 7.44275928e-01
-6.20135665e-01 9.51080441e-01 -4.59670752e-01 -1.04133159e-01
1.64925516e-01 1.49301276e-01 3.53403598e-01 1.69655836e+00
1.23248622e-01 2.90130287e-01 3.70306313e-01 3.08498859e-01
-1.81621641e-01 8.25624824e-01 -6.94406748e-01 -2.25248620e-01
6.11885548e-01 1.34567487e+00 -7.94294894e-01 -3.51767898e-01
-7.13056445e-01 8.56585205e-01 7.09316969e-01 2.78295845e-01
-3.42579633e-01 -4.69013691e-01 4.84163016e-01 2.53104847e-02
6.96333274e-02 -3.97410393e-01 -2.96404630e-01 -1.16881549e+00
3.98728391e-03 -9.90565896e-01 6.06195152e-01 -6.01942301e-01
-1.46016467e+00 8.66992533e-01 -9.25816447e-02 -7.08092809e-01
-3.67033482e-01 -8.10534060e-01 -2.93014526e-01 1.10580266e+00
-1.15914190e+00 -1.22235692e+00 2.29793057e-01 2.97185868e-01
6.12326741e-01 3.17732356e-02 1.29377413e+00 4.51547831e-01
-5.50379515e-01 6.10012412e-01 -1.04363680e-01 5.73870361e-01
1.10885918e+00 -1.16461027e+00 4.16887313e-01 9.76566851e-01
4.54311490e-01 9.77082253e-01 6.73573017e-01 -7.10809231e-01
-8.59476328e-01 -8.88146341e-01 1.75688982e+00 -7.86354899e-01
7.64663517e-01 -4.23395127e-01 -1.04277635e+00 1.07127988e+00
5.08080363e-01 -1.58766538e-01 1.29518294e+00 8.04627001e-01
-3.59200209e-01 2.87182301e-01 -1.10117650e+00 5.62685788e-01
9.68905449e-01 -7.78430998e-01 -1.04479682e+00 3.68102819e-01
6.38654530e-01 -3.24620783e-01 -1.00975919e+00 2.43137572e-02
3.96786094e-01 -3.11002076e-01 4.62322176e-01 -7.95686245e-01
-1.15346044e-01 -2.72318840e-01 -2.28140429e-01 -1.29681873e+00
-3.23603183e-01 -6.28361940e-01 8.86467636e-01 1.97068012e+00
8.40183377e-01 -6.08410180e-01 3.66778314e-01 1.15860379e+00
-4.08791870e-01 -3.64701310e-03 -9.50210214e-01 -8.15053701e-01
2.53084391e-01 -6.69001043e-01 3.15750539e-01 1.35000706e+00
2.27080137e-01 7.45140612e-01 -1.83721259e-01 1.72712460e-01
2.34721273e-01 -5.60304344e-01 5.06314456e-01 -1.31062329e+00
-2.82557607e-02 -1.44540310e-01 -2.18130991e-01 -5.45262218e-01
5.65702140e-01 -9.04487193e-01 5.66425145e-01 -1.42404735e+00
-1.82220817e-01 -8.62995028e-01 -3.24010223e-01 1.02238238e+00
-8.41228142e-02 2.37793595e-01 1.17589764e-01 2.68952221e-01
-4.72369164e-01 6.59841001e-02 4.71883059e-01 4.73167390e-01
-4.11653399e-01 -2.23063529e-01 -1.05300128e+00 9.58982766e-01
9.03070927e-01 -7.75879860e-01 -2.60573886e-02 -7.65236020e-01
2.60424376e-01 -2.08740756e-01 -1.32309094e-01 -1.03729868e+00
2.40796730e-01 -1.26495615e-01 4.48906690e-01 2.90077515e-02
1.18887000e-01 -5.98652780e-01 8.92798826e-02 -4.75310674e-03
-5.73969781e-01 3.51556599e-01 4.67472225e-01 1.29632413e-01
-1.43670380e-01 -7.24824488e-01 6.66192412e-01 -3.90894115e-01
-8.40228319e-01 -2.44977355e-01 -5.56743562e-01 5.28399169e-01
6.55696034e-01 2.55204085e-02 -2.98583031e-01 -3.70779522e-02
-7.54869223e-01 -6.89124390e-02 5.12978137e-01 4.16091114e-01
-7.96468332e-02 -1.16047204e+00 -8.59613657e-01 -1.63377319e-02
1.81766212e-01 -3.22746813e-01 -8.23793747e-03 6.87087953e-01
-4.15247589e-01 5.88798881e-01 -9.52701867e-02 2.37630289e-02
-1.03329825e+00 3.19363207e-01 3.54726054e-02 -2.48784661e-01
-5.26360571e-01 7.87844718e-01 -8.48527104e-02 -7.96297252e-01
1.11112885e-01 2.05236167e-01 -3.35558802e-01 2.13575289e-01
6.66414976e-01 1.86318606e-01 1.64358854e-01 -1.21633697e+00
-6.40069485e-01 4.69984040e-02 -3.15814644e-01 -6.24561131e-01
1.32034111e+00 -3.25245321e-01 3.36623713e-02 7.15794146e-01
8.67882848e-01 7.04045355e-01 -5.85490227e-01 -5.16167819e-01
6.74357891e-01 -1.93422839e-01 -2.78795868e-01 -8.54581654e-01
-4.36430782e-01 6.07256413e-01 2.82515615e-01 1.92000419e-01
6.64482355e-01 2.35835284e-01 5.78312218e-01 5.72575152e-01
4.46591318e-01 -1.51695311e+00 -5.93506098e-01 7.14205384e-01
6.12487674e-01 -1.24764204e+00 -2.16368467e-01 -3.99553597e-01
-9.33888376e-01 8.88843298e-01 6.08212054e-01 1.72291383e-01
5.63496649e-01 1.76317021e-01 5.86731076e-01 -1.40184671e-01
-5.03145933e-01 -2.69687176e-01 -1.04598850e-01 7.40523160e-01
1.19400156e+00 1.88609868e-01 -5.60791552e-01 7.03416944e-01
-3.96077096e-01 -2.65036583e-01 3.87937278e-01 1.17561185e+00
-3.23417872e-01 -1.51563060e+00 -3.27483177e-01 1.38874933e-01
-1.05506599e+00 -5.34947276e-01 -3.04166466e-01 9.23559189e-01
1.28383622e-01 1.15229511e+00 1.34688159e-02 2.74903148e-01
5.09813786e-01 8.57292950e-01 1.09566882e-01 -1.14476943e+00
-7.74252713e-01 1.75760090e-01 9.02192473e-01 -2.26780787e-01
-5.73539793e-01 -1.04097664e+00 -1.12358356e+00 -2.61471290e-02
-3.59519213e-01 6.34419084e-01 8.09505105e-01 1.03775978e+00
2.32894570e-01 -2.54051797e-02 4.16548476e-02 -6.57616436e-01
-4.24942613e-01 -1.24898660e+00 -4.78401273e-01 3.86217117e-01
-2.47634619e-01 -3.18877131e-01 -2.93317646e-01 3.99374157e-01] | [10.105134963989258, 9.786418914794922] |
db45b429-f12c-4075-8ec6-e9700744a6a2 | train-on-small-play-the-large-scaling-up | 2107.08387 | null | https://arxiv.org/abs/2107.08387v1 | https://arxiv.org/pdf/2107.08387v1.pdf | Train on Small, Play the Large: Scaling Up Board Games with AlphaZero and GNN | Playing board games is considered a major challenge for both humans and AI researchers. Because some complicated board games are quite hard to learn, humans usually begin with playing on smaller boards and incrementally advance to master larger board strategies. Most neural network frameworks that are currently tasked with playing board games neither perform such incremental learning nor possess capabilities to automatically scale up. In this work, we look at the board as a graph and combine a graph neural network architecture inside the AlphaZero framework, along with some other innovative improvements. Our ScalableAlphaZero is capable of learning to play incrementally on small boards, and advancing to play on large ones. Our model can be trained quickly to play different challenging board games on multiple board sizes, without using any domain knowledge. We demonstrate the effectiveness of ScalableAlphaZero and show, for example, that by training it for only three days on small Othello boards, it can defeat the AlphaZero model on a large board, which was trained to play the large board for $30$ days. | ['Ran El-Yaniv', 'Shai Ben-Assayag'] | 2021-07-18 | null | null | null | null | ['board-games'] | ['playing-games'] | [-3.57968882e-02 4.30720598e-01 6.94153178e-03 1.07787266e-01
-2.40770906e-01 -7.07036316e-01 -1.95335858e-02 2.01744527e-01
-5.08771896e-01 6.28454089e-01 -5.48206508e-01 -6.56275332e-01
-1.90561399e-01 -1.27030408e+00 -9.13865924e-01 2.06681546e-02
-4.85880017e-01 9.69549656e-01 8.46459270e-01 -8.70856762e-01
-1.24093555e-01 1.23476312e-01 -1.23185372e+00 2.78681755e-01
6.73520148e-01 8.26893806e-01 3.07533909e-02 1.15682721e+00
2.89036781e-01 1.52040648e+00 -8.82185340e-01 -6.82099044e-01
7.38953233e-01 -5.94786853e-02 -7.87217736e-01 -3.07983994e-01
6.41363561e-01 -8.50002885e-01 -6.33276761e-01 1.00249434e+00
1.64801061e-01 9.49924290e-02 -1.47555843e-02 -1.48687375e+00
-1.53364345e-01 1.15249407e+00 -3.40544224e-01 2.59190053e-01
1.15697421e-01 2.92254895e-01 1.23769391e+00 -2.35859975e-02
4.78279680e-01 7.18826950e-01 1.00029624e+00 4.99239087e-01
-8.30612540e-01 -9.77824032e-01 2.42983907e-01 1.99286684e-01
-1.03442621e+00 -1.31875813e-01 6.08587980e-01 -1.89633369e-01
1.28849685e+00 -2.23645839e-04 1.30171108e+00 6.36721909e-01
2.54719019e-01 6.71840489e-01 6.21128321e-01 -1.90846875e-01
3.16264063e-01 -7.48188138e-01 1.83908597e-01 9.65504348e-01
6.73482239e-01 1.25688180e-01 -2.71410406e-01 7.61172771e-02
1.06273699e+00 -8.92614275e-02 1.49575889e-01 -4.38612133e-01
-6.16932929e-01 7.56217897e-01 7.09637702e-01 1.01223417e-01
-1.82944179e-01 7.59656429e-01 4.97154444e-01 7.65062988e-01
1.65366456e-01 1.18382573e+00 -6.04776859e-01 -5.76601267e-01
-7.79159009e-01 5.27840853e-01 1.19570601e+00 8.70011330e-01
9.07608569e-01 1.69737816e-01 4.71869648e-01 4.98770207e-01
-1.91650987e-01 7.65369236e-02 3.24686319e-01 -1.18044460e+00
4.82448697e-01 7.39693880e-01 -2.02639356e-01 -1.18748295e+00
-6.48008645e-01 -3.62473935e-01 -6.10251606e-01 4.84236121e-01
6.58945262e-01 -6.35072708e-01 -7.06664145e-01 1.48973811e+00
1.70491755e-01 4.63318229e-01 -2.95281053e-01 4.74604338e-01
8.24245751e-01 5.63605845e-01 -4.60605562e-01 4.39700812e-01
1.19861150e+00 -1.39002943e+00 7.57155567e-02 -6.56587481e-01
1.01559651e+00 3.73412631e-02 1.02229130e+00 1.11961544e+00
-1.47968209e+00 -5.25550425e-01 -1.67083132e+00 2.59737849e-01
-2.05547720e-01 -3.03386480e-01 1.15555286e+00 8.54840636e-01
-1.48930311e+00 8.95047128e-01 -1.02345932e+00 -1.57441795e-01
5.19185305e-01 9.52779114e-01 -4.88290429e-01 -2.41556689e-01
-1.15353906e+00 8.88911128e-01 5.64790785e-01 -2.26596072e-02
-1.30085659e+00 -7.03165233e-01 -7.24788845e-01 2.90878803e-01
9.84092236e-01 -7.02364743e-01 1.68587017e+00 -1.02313042e+00
-1.89055133e+00 5.93511283e-01 8.58568251e-01 -8.54641318e-01
4.05474901e-01 -3.04845512e-01 -3.55301099e-03 1.03829525e-01
-2.50818301e-02 6.88050330e-01 6.07092679e-01 -5.53746819e-01
-5.60725570e-01 -1.12741040e-02 1.30229950e+00 3.53890151e-01
-3.98254812e-01 -2.21022457e-01 -4.54133034e-01 -1.61034405e-01
-1.73108369e-01 -1.01809919e+00 -5.22508681e-01 -2.53097564e-01
-6.99228123e-02 -2.84070671e-01 9.46404710e-02 -3.25557679e-01
1.09536576e+00 -1.91023529e+00 -4.71734963e-02 2.85864949e-01
9.81514513e-01 2.55312115e-01 -3.40735406e-01 3.47073913e-01
7.44703189e-02 7.51429498e-02 4.43918347e-01 3.77113461e-01
-7.96474516e-03 1.84513181e-01 3.48556012e-01 1.23476349e-01
-1.14226723e-02 9.60566878e-01 -1.17144775e+00 1.86286177e-02
-2.47120690e-02 -2.84564734e-01 -1.28476334e+00 9.15384442e-02
-3.79736423e-01 -3.19761217e-01 -1.03979215e-01 3.72700274e-01
4.42447454e-01 -5.66985846e-01 5.59225261e-01 4.81742740e-01
3.03947508e-01 3.07628393e-01 -1.11888814e+00 1.80300045e+00
-4.57738876e-01 8.24633300e-01 2.27536544e-01 -9.78333652e-01
7.34817624e-01 -5.93542047e-02 1.98345572e-01 -5.99323988e-01
3.57705146e-01 1.54590055e-01 9.38571513e-01 -1.60913356e-02
6.04679644e-01 -5.75587861e-02 -1.33070663e-01 4.66577142e-01
1.40980482e-01 -6.15470350e-01 7.15417743e-01 3.62914741e-01
2.12966394e+00 -3.05590540e-01 1.72395036e-01 2.78346334e-02
-1.09750750e-02 2.59250224e-01 2.69306421e-01 1.19410002e+00
-3.04415196e-01 1.88419804e-01 9.67813671e-01 -9.22677100e-01
-9.96411443e-01 -8.56638312e-01 7.62644291e-01 1.56618571e+00
1.88848272e-01 -8.89196694e-01 -8.66414785e-01 -6.85946643e-01
-1.75708935e-01 1.50251046e-01 -6.26582801e-01 -4.13403928e-01
-7.28669465e-01 -1.63208678e-01 7.20396996e-01 7.25954831e-01
7.83664942e-01 -1.15132630e+00 -7.90804088e-01 4.08612728e-01
3.54817867e-01 -7.76769936e-01 -1.60358444e-01 4.13584977e-01
-7.32044518e-01 -1.31477714e+00 -3.06847602e-01 -8.11502934e-01
3.40679735e-01 3.52828413e-01 1.42969191e+00 7.11251974e-01
-2.39182308e-01 1.25558212e-01 -4.48762804e-01 -4.86766934e-01
-2.29483560e-01 7.67636955e-01 -1.86144456e-01 -8.21894526e-01
1.39996428e-02 -1.01226997e+00 -4.70314443e-01 1.28140271e-01
-6.88838780e-01 1.62523061e-01 6.36136234e-01 7.60073185e-01
-3.16903591e-01 4.59166139e-01 1.83757082e-01 -1.38940251e+00
9.65535700e-01 -2.67014772e-01 -9.02691245e-01 -1.79150086e-02
-2.26947948e-01 -3.24091345e-01 8.06018472e-01 -5.83130419e-01
-1.83183923e-01 -1.89942002e-01 -2.60665804e-01 -2.00763240e-01
1.80301532e-01 7.46888697e-01 -6.12061918e-02 -2.96208382e-01
1.10403836e+00 -2.62400508e-01 -7.75365755e-02 7.86064789e-02
4.61458832e-01 2.55161077e-02 3.03408384e-01 -8.65635753e-01
8.25730979e-01 1.25923023e-01 -1.31984025e-01 -4.99557853e-01
-6.79775357e-01 -2.70097911e-01 -3.76515925e-01 -3.79174024e-01
4.60886419e-01 -8.98512661e-01 -1.24507594e+00 7.19654858e-01
-7.75302768e-01 -1.22640157e+00 -3.62136394e-01 1.21697839e-02
-4.17037994e-01 -1.11312427e-01 -8.96230519e-01 -1.17631420e-01
-6.37265071e-02 -8.43561649e-01 4.11088586e-01 3.13716173e-01
-4.24629539e-01 -8.30077827e-01 3.07912290e-01 4.42796052e-01
5.37808120e-01 -4.77965809e-02 8.65335882e-01 -8.07251394e-01
-5.22522211e-01 -5.23145080e-01 -1.19293056e-01 2.54504025e-01
-2.23128960e-01 -5.06514125e-02 -4.04859066e-01 -5.74149966e-01
-3.18505734e-01 -8.75440538e-01 5.88338733e-01 3.81714910e-01
9.33344960e-01 -1.15351103e-01 -4.04218733e-02 7.24154413e-01
1.14015126e+00 3.33194375e-01 6.56677008e-01 8.26249301e-01
6.91879153e-01 2.17763111e-01 1.18280463e-01 2.96792865e-01
5.26355624e-01 2.39395052e-01 8.27168345e-01 7.25052953e-02
1.22997627e-01 -3.21495086e-01 3.43390137e-01 7.77555764e-01
-3.98823172e-01 -2.13320568e-01 -1.30643189e+00 2.23261058e-01
-1.74861169e+00 -7.52738595e-01 1.93184495e-01 1.75578415e+00
6.78697646e-01 9.77275014e-01 4.61435735e-01 4.62751836e-02
3.83217514e-01 2.02643573e-01 -7.07878828e-01 -4.48669642e-01
2.34676972e-01 9.52006638e-01 6.53023183e-01 5.25354624e-01
-1.17929351e+00 1.45711768e+00 7.20291138e+00 7.59810686e-01
-1.20996881e+00 -7.25791454e-02 5.12769759e-01 -3.86519283e-01
2.05255106e-01 2.62411952e-01 -3.54532152e-01 -8.86020958e-02
9.37166393e-01 -1.89917982e-01 8.05724204e-01 1.30012107e+00
-5.45211732e-01 -1.45086572e-02 -1.07912767e+00 8.72707367e-01
-2.58190688e-02 -1.55082738e+00 -1.76245391e-01 9.09031928e-02
7.65177250e-01 2.43865684e-01 -1.42547682e-01 1.04953396e+00
1.39660215e+00 -1.45507061e+00 5.98676383e-01 -1.82178006e-01
4.86159474e-01 -7.18756735e-01 6.68997943e-01 5.40310204e-01
-9.90836740e-01 -2.19929516e-01 -6.96521103e-01 -1.18016994e+00
-5.44513345e-01 -1.34798409e-02 -8.80439103e-01 1.44840211e-01
7.10390031e-01 7.04652250e-01 -7.68605947e-01 1.15036094e+00
-6.23734653e-01 5.99987090e-01 -5.54514349e-01 -4.44712102e-01
3.18258673e-01 5.02304323e-02 9.44037512e-02 2.36488014e-01
1.67018205e-01 1.46358624e-01 5.19452333e-01 1.91511109e-01
-4.48968828e-01 -9.51077640e-02 -6.47463739e-01 -2.94415027e-01
3.93853605e-01 1.21074617e+00 -9.54209387e-01 -3.54437798e-01
-3.73140275e-01 7.38660574e-01 1.00422752e+00 1.22101940e-02
-9.48428452e-01 -5.81073999e-01 4.77199614e-01 4.18383598e-01
5.12168229e-01 -4.06248897e-01 -1.18066832e-01 -1.12682295e+00
-7.34406337e-02 -1.28112149e+00 3.33808541e-01 -9.34565425e-01
-9.99952972e-01 7.37712026e-01 -1.72258362e-01 -8.34562123e-01
-8.28806758e-02 -1.05773091e+00 -8.87231052e-01 1.57504082e-01
-8.44349861e-01 -9.70079958e-01 -4.06260550e-01 5.23328781e-01
2.29707539e-01 -2.46672854e-01 5.27392268e-01 5.63310683e-02
-5.60531676e-01 7.86320210e-01 -1.38644546e-01 4.63856727e-01
2.90044993e-01 -1.26735282e+00 7.73426294e-01 5.92518508e-01
3.77682537e-01 4.99671608e-01 4.73321885e-01 -5.33519447e-01
-1.32164729e+00 -6.19358420e-01 1.28224511e-02 -3.14101517e-01
1.12387061e+00 -6.25872016e-01 -4.54350084e-01 8.55693698e-01
2.65181303e-01 -1.01975583e-01 4.42926645e-01 8.49254370e-01
-5.31316340e-01 -4.48563039e-01 -5.63695967e-01 7.74065554e-01
1.27126849e+00 -3.21409136e-01 -4.83364016e-01 2.72203982e-01
6.43436909e-01 -7.55182445e-01 -6.22249126e-01 3.03268526e-02
6.03571296e-01 -1.08179832e+00 7.36373842e-01 -7.89588749e-01
6.61528647e-01 -1.32106617e-02 3.03680599e-01 -1.54855335e+00
-4.57470298e-01 -8.35371077e-01 4.71099541e-02 3.62974375e-01
5.01190186e-01 -6.42743707e-01 1.72213125e+00 6.01900339e-01
-1.82617873e-01 -6.39698148e-01 -7.95692205e-01 -8.54323685e-01
3.52564424e-01 -6.17573559e-01 7.18056619e-01 7.38277137e-01
4.51711595e-01 5.22084415e-01 -3.75345349e-01 -2.47316454e-02
-4.94036190e-02 -1.58170536e-01 1.43212092e+00 -1.47864711e+00
-9.98615503e-01 -6.75797164e-01 -7.42663860e-01 -1.41305768e+00
-1.43821627e-01 -8.29787791e-01 1.31736934e-01 -1.40098286e+00
7.10917683e-03 -4.64168221e-01 -3.25603902e-01 6.71615064e-01
2.00529266e-02 3.20114881e-01 4.13425177e-01 -1.62530448e-02
-1.10194004e+00 -3.76822352e-02 1.27366483e+00 -4.79575455e-01
-1.01224869e-01 -9.62932259e-02 -1.05772579e+00 1.00278378e+00
7.84988761e-01 -5.83628833e-01 -7.83077955e-01 -6.34565711e-01
1.25336671e+00 7.29716867e-02 -5.51589578e-03 -1.70470297e+00
5.91155648e-01 -4.96701971e-02 1.02407739e-01 1.53151795e-01
2.82440066e-01 -5.69793642e-01 -3.81455645e-02 7.30122089e-01
-4.75437529e-02 3.27136278e-01 4.07232761e-01 2.93246359e-01
-1.49325937e-01 -4.34034199e-01 3.29615682e-01 -4.31463927e-01
-6.45873368e-01 4.51613963e-01 -6.64844036e-01 3.66796196e-01
1.30883873e+00 -4.77364272e-01 -4.85683709e-01 -8.01853716e-01
-8.57409060e-01 4.16386634e-01 6.56653523e-01 -5.86135611e-02
3.67852151e-01 -8.91223133e-01 -4.30387050e-01 1.01382144e-01
-1.10601470e-01 3.23236912e-01 3.53868634e-01 3.99819613e-01
-1.59980512e+00 -1.87242702e-02 -6.12263083e-01 -2.32748628e-01
-1.22417331e+00 3.92428547e-01 6.12435162e-01 -1.08341599e+00
-4.90717828e-01 1.43422103e+00 2.44683996e-01 -6.46086037e-01
9.88826081e-02 -6.31820917e-01 -4.32686880e-02 -2.43849829e-01
3.54401529e-01 1.54502824e-01 2.46842831e-01 3.42818290e-01
-1.06057413e-01 1.47477597e-01 -5.07181227e-01 4.95985933e-02
1.65590560e+00 8.50439310e-01 -6.87255561e-02 3.76777388e-02
4.63828117e-01 -1.74792990e-01 -1.32594836e+00 1.87993601e-01
-2.19176099e-01 -2.12116227e-01 -2.16022804e-01 -6.48425579e-01
-1.16220200e+00 8.44022810e-01 -3.84229571e-02 4.80264485e-01
9.86299217e-01 -3.11323494e-01 8.15545321e-01 1.04030418e+00
7.30812788e-01 -9.99104798e-01 5.92510283e-01 1.11076140e+00
3.93767059e-01 -7.72490442e-01 9.91932154e-02 -2.92979687e-01
-1.65803477e-01 1.15662980e+00 1.25596964e+00 -9.68001187e-01
6.83001459e-01 4.82932359e-01 -6.76111430e-02 -4.69246864e-01
-1.22237551e+00 1.53485034e-02 -2.50842869e-01 7.79933333e-01
1.05709553e-01 1.15433350e-01 2.66487181e-01 6.72647893e-01
-8.05300415e-01 5.87944724e-02 8.80540907e-01 1.09204185e+00
-5.04389584e-01 -1.14548802e+00 1.07079772e-02 7.73761928e-01
-1.93095475e-01 -4.53722954e-01 -6.32560074e-01 1.14124823e+00
2.44244576e-01 7.22671866e-01 1.88581377e-01 -8.16972971e-01
3.92307520e-01 -2.95697302e-01 8.64422500e-01 -9.44315314e-01
-1.07383490e+00 -5.67353427e-01 3.87308359e-01 -5.96241891e-01
1.64427862e-01 -2.17017889e-01 -7.84059227e-01 -1.03145003e+00
-4.13795054e-01 1.56229883e-01 5.20247258e-02 7.47166574e-01
9.62262154e-02 7.54942954e-01 7.83351958e-02 -7.59385765e-01
-4.02964205e-01 -8.80985200e-01 -6.88465238e-01 7.04414994e-02
-1.37167454e-01 -4.63566273e-01 -1.66384375e-03 -5.35182059e-01] | [3.4751126766204834, 1.4457818269729614] |
869f3fbd-11c6-41c3-95df-fb77ed29c91a | automatic-piano-transcription-with | 2307.04305 | null | https://arxiv.org/abs/2307.04305v1 | https://arxiv.org/pdf/2307.04305v1.pdf | Automatic Piano Transcription with Hierarchical Frequency-Time Transformer | Taking long-term spectral and temporal dependencies into account is essential for automatic piano transcription. This is especially helpful when determining the precise onset and offset for each note in the polyphonic piano content. In this case, we may rely on the capability of self-attention mechanism in Transformers to capture these long-term dependencies in the frequency and time axes. In this work, we propose hFT-Transformer, which is an automatic music transcription method that uses a two-level hierarchical frequency-time Transformer architecture. The first hierarchy includes a convolutional block in the time axis, a Transformer encoder in the frequency axis, and a Transformer decoder that converts the dimension in the frequency axis. The output is then fed into the second hierarchy which consists of another Transformer encoder in the time axis. We evaluated our method with the widely used MAPS and MAESTRO v3.0.0 datasets, and it demonstrated state-of-the-art performance on all the F1-scores of the metrics among Frame, Note, Note with Offset, and Note with Offset and Velocity estimations. | ['Yuki Mitsufuji', 'Wei-Hsiang Liao', 'Yuhta Takida', 'Yukara Ikemiya', 'Taketo Akama', 'Keisuke Toyama'] | 2023-07-10 | null | null | null | null | ['music-transcription'] | ['music'] | [-3.75351869e-02 -5.30978322e-01 -6.24809088e-03 1.23525783e-01
-9.71584916e-01 -7.33424306e-01 1.53168380e-01 -1.94881037e-01
6.37875721e-02 2.88136452e-01 3.93590540e-01 -8.59231204e-02
-7.02907145e-02 -4.72389847e-01 -4.31775600e-01 -7.53353119e-01
1.08010080e-02 2.31505826e-01 2.50666738e-01 -1.36036739e-01
3.09301764e-01 2.72647124e-02 -1.46577251e+00 5.40819526e-01
6.90675914e-01 1.36513114e+00 1.58956885e-01 1.10498571e+00
1.24836536e-02 9.27096486e-01 -6.67993784e-01 -2.69693226e-01
-1.86228901e-02 -8.15357566e-01 -4.97360051e-01 -2.54440755e-01
2.93393731e-01 -2.12734938e-01 -4.80142266e-01 8.22118402e-01
8.47041667e-01 8.87278020e-02 5.85394442e-01 -9.65081334e-01
-3.87514949e-01 1.13729787e+00 -4.46721673e-01 4.68719929e-01
3.26507002e-01 1.39416948e-01 1.36368763e+00 -8.57043982e-01
2.70291030e-01 9.04182553e-01 1.02055478e+00 -3.16761946e-03
-9.24239755e-01 -8.73077929e-01 -2.10117176e-01 5.93263268e-01
-1.47714567e+00 -5.19197583e-01 1.06642449e+00 -7.12919295e-01
1.02907121e+00 1.68952629e-01 7.81058729e-01 8.80749941e-01
4.83729035e-01 6.24692917e-01 7.05238461e-01 -3.82234395e-01
-8.24008733e-02 -4.29224193e-01 -2.95393374e-02 2.00477630e-01
-6.02656960e-01 1.21096358e-01 -1.04383099e+00 1.74536675e-01
9.57600653e-01 -3.72023404e-01 -3.87156159e-01 2.81195849e-01
-1.30514276e+00 3.96299243e-01 3.64165276e-01 4.75395381e-01
-5.11313438e-01 3.57943624e-01 7.53669202e-01 1.39166877e-01
3.38928431e-01 3.01371992e-01 -3.86941940e-01 -9.39285815e-01
-1.31675506e+00 1.16832010e-01 6.48321152e-01 7.61014044e-01
-4.88746874e-02 3.20562959e-01 -7.41724968e-01 8.31322014e-01
1.77369639e-01 3.31585586e-01 7.55789757e-01 -9.30351079e-01
7.26888776e-01 1.03122853e-01 8.95216987e-02 -7.35967338e-01
-3.54989827e-01 -8.61334622e-01 -7.35584080e-01 -1.82910457e-01
3.80043149e-01 -5.66692166e-02 -6.69115067e-01 1.77404642e+00
2.12552294e-01 4.40061301e-01 -1.40903145e-01 1.16330147e+00
6.97164178e-01 8.44493508e-01 -1.46427706e-01 -3.53659034e-01
1.69832575e+00 -1.12627184e+00 -1.12587714e+00 1.47084832e-01
1.26986504e-01 -1.26531816e+00 1.32159650e+00 5.07269800e-01
-1.28986406e+00 -1.08814442e+00 -1.17357206e+00 -4.91666168e-01
1.52716532e-01 8.16583395e-01 1.87421516e-01 1.23713881e-01
-7.35214889e-01 9.60622907e-01 -8.07634830e-01 -1.59676336e-02
-9.78899300e-02 1.49455398e-01 1.63949713e-01 6.48468256e-01
-1.43678212e+00 3.28816295e-01 8.90209079e-02 3.16886723e-01
-7.65060306e-01 -8.44125509e-01 -6.61834300e-01 4.65485245e-01
-1.44996509e-01 -4.24822420e-01 1.71105647e+00 -7.38545477e-01
-1.80151081e+00 4.51123714e-01 -2.39924699e-01 -3.97572607e-01
4.64471430e-01 -4.55321878e-01 -6.49189174e-01 4.64237900e-03
1.09326340e-01 4.13520753e-01 1.04484189e+00 -4.74376947e-01
-8.94295394e-01 -2.83634603e-01 -9.99418721e-02 4.11144674e-01
-9.08647180e-02 -2.47440472e-01 -6.38969600e-01 -1.08906639e+00
1.30502224e-01 -1.07487011e+00 5.49195766e-01 -5.54666281e-01
-5.50834954e-01 -1.63044572e-01 7.70088315e-01 -1.18186069e+00
1.98958695e+00 -2.54945040e+00 4.19529080e-01 -8.47771987e-02
-8.32490250e-03 -1.92337967e-02 1.40661255e-01 4.58135366e-01
-3.64024416e-02 -2.54312009e-01 1.63543075e-01 -2.90525645e-01
2.16063648e-01 -3.25961709e-01 -5.36870182e-01 2.11128935e-01
-7.58692846e-02 6.38352633e-01 -7.06214428e-01 -3.09638321e-01
2.97565460e-02 8.03484261e-01 -6.29226923e-01 2.72830099e-01
-1.10236732e-02 4.71102744e-01 6.66306764e-02 4.79635507e-01
2.33805746e-01 -8.04626197e-02 9.41191837e-02 -6.54016197e-01
-4.68174398e-01 1.07428014e+00 -9.83938336e-01 1.82483232e+00
-5.24574041e-01 1.08196056e+00 -2.26251096e-01 -3.38773727e-01
8.12386811e-01 8.41955185e-01 5.95822692e-01 -6.21747196e-01
2.45940313e-01 3.23426634e-01 3.58835608e-01 -4.92345840e-01
5.83413005e-01 -1.35708109e-01 -6.32459521e-02 3.54604423e-01
2.24509865e-01 -9.12483111e-02 2.94303387e-01 -3.02392274e-01
7.99571991e-01 2.81725913e-01 1.55576885e-01 1.09471353e-02
4.97256726e-01 -4.38837111e-01 5.64727485e-01 1.30671471e-01
-8.69572535e-03 9.69764233e-01 6.83043182e-01 -3.09607476e-01
-1.01373899e+00 -9.46214080e-01 1.57300293e-01 1.27229881e+00
-2.48374283e-01 -1.08218753e+00 -7.81814694e-01 -2.59111375e-01
-1.73107907e-01 5.93299150e-01 -6.63990378e-01 -2.52491832e-01
-5.46425641e-01 -1.72220781e-01 7.59675562e-01 7.35506415e-01
6.19340301e-01 -1.17590082e+00 -4.58325893e-01 5.75819492e-01
-8.02981913e-01 -9.68743324e-01 -1.30692220e+00 3.49611193e-01
-6.37634277e-01 -8.64043176e-01 -7.06500709e-01 -6.52246356e-01
-1.85761273e-01 -5.43557182e-02 9.06521678e-01 -5.50553858e-01
1.47269875e-01 -4.86505926e-02 -4.15738583e-01 -3.88001502e-01
3.52963582e-02 4.79485929e-01 -1.68013021e-01 2.06178620e-01
2.79686630e-01 -9.11546528e-01 -7.61112928e-01 3.55404019e-01
-5.33296764e-01 2.26396874e-01 2.69053340e-01 5.56364536e-01
7.34293461e-01 1.22732729e-01 4.92806315e-01 -3.70055616e-01
6.99144065e-01 -1.01913258e-01 -5.44287205e-01 -9.76317301e-02
-1.21871382e-01 -1.43079057e-01 7.03760087e-01 -6.97939217e-01
-6.32829964e-01 1.43742770e-01 -4.70825136e-01 -6.65867805e-01
5.84719658e-01 4.94096875e-01 -2.52590925e-01 4.64017779e-01
5.70329964e-01 2.25869283e-01 -4.53237653e-01 -6.09337211e-01
1.13048740e-01 1.06860411e+00 8.00133705e-01 -2.97378272e-01
5.02174854e-01 1.93381514e-02 -2.72148341e-01 -6.66175365e-01
-1.06915665e+00 -3.68951112e-01 -6.53987348e-01 -3.56740922e-01
9.78170097e-01 -1.05432093e+00 -6.37961328e-01 5.97250819e-01
-1.26608098e+00 -4.39673722e-01 -3.79659414e-01 7.43423223e-01
-8.24849784e-01 -1.52980968e-01 -8.47677827e-01 -5.14512539e-01
-6.71472013e-01 -1.29245949e+00 1.11938691e+00 2.79381752e-01
-5.92635393e-01 -5.40418148e-01 2.87742615e-01 1.89345419e-01
4.78482157e-01 -1.34557173e-01 6.79809928e-01 -1.19877808e-01
-3.70060772e-01 -1.54201640e-02 1.39491767e-01 3.31639767e-01
1.68179363e-01 1.12564869e-01 -1.28175914e+00 -6.95892051e-02
9.71175507e-02 6.05860054e-02 7.24836230e-01 6.79968655e-01
1.14213824e+00 -2.97790766e-01 1.64081559e-01 9.22438085e-01
9.46759880e-01 5.55906653e-01 8.41446579e-01 -5.39467558e-02
8.61925423e-01 8.63938853e-02 4.81794864e-01 4.45024282e-01
5.58571875e-01 1.05976975e+00 -1.49626330e-01 7.04871044e-02
-4.73350883e-01 -4.67648774e-01 6.08910203e-01 1.45050573e+00
-2.53392935e-01 -1.60396054e-01 -6.25397265e-01 5.34015238e-01
-1.71936059e+00 -1.09654689e+00 1.72242299e-02 2.21493959e+00
1.15190077e+00 1.53172478e-01 2.43069872e-01 6.87693775e-01
6.69722199e-01 1.87589690e-01 -6.43903971e-01 -5.56426108e-01
7.15472475e-02 2.22964540e-01 -5.16019622e-03 4.30153698e-01
-1.13349962e+00 7.09771633e-01 5.58030319e+00 9.30731118e-01
-1.59497297e+00 9.31615680e-02 3.42646241e-01 -2.05920979e-01
2.08252873e-02 -1.54667869e-01 -6.07807636e-01 6.58425331e-01
1.12462425e+00 -7.86737949e-02 7.17672050e-01 5.32480001e-01
4.49491322e-01 2.02213347e-01 -1.16668582e+00 1.24478304e+00
-1.47455618e-01 -9.59033668e-01 -2.31720060e-01 -7.81646967e-02
5.27756393e-01 -1.63166374e-01 -6.89061433e-02 4.74729359e-01
-4.80750710e-01 -8.52785587e-01 1.08313274e+00 5.04445374e-01
1.25698256e+00 -9.59041774e-01 3.51258248e-01 -3.03059872e-02
-2.04868221e+00 2.81565962e-03 6.84971139e-02 -7.83145055e-02
1.35146499e-01 5.28683484e-01 -6.58604205e-01 2.82683522e-01
6.15913928e-01 8.84876490e-01 3.78001258e-02 1.12025678e+00
-4.84086096e-01 9.79555190e-01 -1.42585233e-01 5.16622782e-01
-6.79608285e-02 -1.34254480e-02 5.55052876e-01 1.35326827e+00
7.51558363e-01 7.01893121e-02 -1.09415390e-01 4.84841049e-01
-8.24402422e-02 -6.71647787e-02 -1.05007902e-01 -2.56575316e-01
6.89227700e-01 1.11735332e+00 -5.48822522e-01 -2.25427359e-01
-1.18691370e-01 1.02157903e+00 7.34998286e-02 2.52436697e-01
-1.39739919e+00 -8.65864754e-01 6.21044219e-01 1.25466660e-01
6.30332112e-01 -2.55367309e-01 -1.45738542e-01 -1.00558400e+00
1.09589167e-01 -8.75614703e-01 4.09545749e-01 -1.10618222e+00
-9.05713677e-01 9.48782444e-01 -5.67127466e-01 -1.74446714e+00
-6.64103210e-01 -2.16144010e-01 -4.98986244e-01 9.55736101e-01
-1.15585804e+00 -9.22537744e-01 -2.23782867e-01 6.14331424e-01
6.74876332e-01 1.03152446e-01 8.66010249e-01 7.97357261e-01
-4.15326089e-01 7.33631313e-01 -2.94605106e-01 2.16959476e-01
7.87700713e-01 -1.30123901e+00 4.39838976e-01 6.86438084e-01
4.68129128e-01 2.90850759e-01 5.84272802e-01 -4.13357139e-01
-1.07057238e+00 -1.00059056e+00 1.25285852e+00 -1.69116948e-02
5.87215006e-01 -3.69937569e-01 -6.30382955e-01 6.46186173e-01
1.76904723e-01 -2.21425250e-01 6.49560928e-01 1.82988364e-02
-3.29987317e-01 -4.68982607e-01 -4.08660382e-01 4.64189649e-01
7.57455766e-01 -1.04796374e+00 -4.32204187e-01 -4.39897850e-02
8.07809830e-01 -9.30969715e-01 -9.65890706e-01 1.98274568e-01
1.04309928e+00 -9.42274034e-01 7.63438702e-01 -3.78246047e-02
5.93143106e-01 -5.72508037e-01 -8.49648863e-02 -1.49579060e+00
-5.61933219e-01 -9.96317804e-01 -3.82735848e-01 1.38553739e+00
3.83215219e-01 4.81503047e-02 1.98132262e-01 -2.73256570e-01
-3.52344185e-01 -5.83160222e-01 -9.22455072e-01 -5.61942697e-01
-4.92519051e-01 -6.47160769e-01 5.42372823e-01 6.91732645e-01
1.54391110e-01 7.31963754e-01 -6.10951304e-01 3.03214118e-02
2.55222023e-01 2.99618483e-01 4.81441021e-01 -9.93068457e-01
-5.49713850e-01 -4.38658267e-01 -2.24648476e-01 -1.09097207e+00
-3.30331087e-01 -5.41267335e-01 1.21981189e-01 -1.30734491e+00
-5.87873757e-02 2.97593679e-02 -5.47787130e-01 2.83883244e-01
-1.12116359e-01 4.27642196e-01 4.01641697e-01 4.36662525e-01
-2.25221798e-01 5.86882651e-01 1.28943574e+00 -1.49134591e-01
-6.09232843e-01 3.18142563e-01 -5.03109694e-01 6.99141741e-01
8.54174614e-01 -2.90132016e-01 -5.56262136e-01 -5.18450856e-01
3.98486778e-02 5.44892251e-01 2.67837327e-02 -1.57834125e+00
3.54722887e-01 1.81732044e-01 3.41111779e-01 -9.46015060e-01
6.59913778e-01 -5.81754506e-01 3.63943011e-01 2.07367748e-01
-4.67543244e-01 4.09015894e-01 3.52468610e-01 -1.29929138e-02
-5.45253396e-01 3.55867296e-01 6.11637354e-01 2.17953354e-01
-2.57727683e-01 1.65680841e-01 -1.43752649e-01 1.42175004e-01
5.18020332e-01 5.33753745e-02 -1.29475310e-01 -4.97447819e-01
-8.73898387e-01 -1.73703387e-01 -1.02981411e-01 5.55729866e-01
3.42341810e-01 -1.74093580e+00 -7.22937822e-01 2.41588399e-01
-6.25744909e-02 -2.41508305e-01 5.61322749e-01 1.12893260e+00
-4.43475515e-01 5.02354562e-01 -3.13038498e-01 -6.53584480e-01
-1.26130450e+00 1.41447201e-01 5.37944973e-01 -4.06090319e-01
-5.16620398e-01 8.10320735e-01 1.51685372e-01 1.67071983e-01
5.52605569e-01 -7.58365989e-01 -3.44932348e-01 3.78568262e-01
7.37318873e-01 2.42956564e-01 2.00661585e-01 -7.77448833e-01
-3.70313227e-01 8.57943892e-01 2.96317190e-01 -5.49226761e-01
1.00681341e+00 -9.99670476e-02 2.78800242e-02 9.88741994e-01
1.06457186e+00 3.93565685e-01 -1.36087251e+00 1.29390240e-01
-2.31092632e-01 -1.85949966e-01 2.23097458e-01 -9.14691567e-01
-1.24656880e+00 1.17796659e+00 4.98636663e-01 2.18918547e-01
1.49343860e+00 -4.32146162e-01 1.02735329e+00 -2.32364222e-01
5.36058210e-02 -1.06704915e+00 7.38703925e-03 9.08935845e-01
1.07521641e+00 -4.92496580e-01 -3.73280972e-01 -1.02529675e-01
-6.52074277e-01 1.25936449e+00 3.81746799e-01 -3.41062504e-03
6.19593203e-01 4.17540938e-01 2.74279982e-01 1.55355809e-02
-8.87520909e-01 -1.01705104e-01 7.08899558e-01 1.85121804e-01
1.05471265e+00 1.91353306e-01 -2.29055941e-01 1.02714801e+00
-9.16775703e-01 1.27852634e-01 2.38565505e-01 3.89772266e-01
-1.26141474e-01 -8.84442568e-01 -3.29575896e-01 -9.26008727e-03
-7.30941534e-01 -2.43053496e-01 -3.26650381e-01 4.57374543e-01
2.56329268e-01 8.55297029e-01 2.80485034e-01 -8.10800731e-01
5.58396161e-01 2.78331816e-01 5.30990899e-01 -3.05966645e-01
-1.08209443e+00 8.43176186e-01 3.16671729e-02 -5.93653440e-01
-2.18528192e-02 -5.98118544e-01 -1.33171010e+00 -2.25030817e-03
-1.32671267e-01 2.67782390e-01 5.88218808e-01 6.28649175e-01
4.51066762e-01 1.27280354e+00 6.41225934e-01 -1.09527183e+00
-2.97234714e-01 -1.33675945e+00 -6.89071298e-01 1.95962727e-01
5.64460695e-01 -3.95457000e-01 -2.38007188e-01 3.80632967e-01] | [15.705209732055664, 5.483864784240723] |
0da2c11a-39ce-4258-8254-1fd0dda3484f | learning-to-classify-images-without-labels | 2005.12320 | null | https://arxiv.org/abs/2005.12320v2 | https://arxiv.org/pdf/2005.12320v2.pdf | SCAN: Learning to Classify Images without Labels | Can we automatically group images into semantically meaningful clusters when ground-truth annotations are absent? The task of unsupervised image classification remains an important, and open challenge in computer vision. Several recent approaches have tried to tackle this problem in an end-to-end fashion. In this paper, we deviate from recent works, and advocate a two-step approach where feature learning and clustering are decoupled. First, a self-supervised task from representation learning is employed to obtain semantically meaningful features. Second, we use the obtained features as a prior in a learnable clustering approach. In doing so, we remove the ability for cluster learning to depend on low-level features, which is present in current end-to-end learning approaches. Experimental evaluation shows that we outperform state-of-the-art methods by large margins, in particular +26.6% on CIFAR10, +25.0% on CIFAR100-20 and +21.3% on STL10 in terms of classification accuracy. Furthermore, our method is the first to perform well on a large-scale dataset for image classification. In particular, we obtain promising results on ImageNet, and outperform several semi-supervised learning methods in the low-data regime without the use of any ground-truth annotations. The code is made publicly available at https://github.com/wvangansbeke/Unsupervised-Classification. | ['Luc van Gool', 'Wouter Van Gansbeke', 'Marc Proesmans', 'Stamatios Georgoulis', 'Simon Vandenhende'] | 2020-05-25 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/1057_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123550273.pdf | eccv-2020-8 | ['unsupervised-image-classification'] | ['computer-vision'] | [ 9.07967016e-02 -1.38734177e-01 -1.99741960e-01 -4.35305417e-01
-1.10908520e+00 -5.97535729e-01 5.36949754e-01 2.69098729e-01
-7.51794338e-01 4.57407504e-01 -3.45646148e-03 3.02199647e-02
5.02670072e-02 -4.57112074e-01 -6.96794391e-01 -8.38560641e-01
-4.46506999e-02 4.25506115e-01 1.45921901e-01 8.71305019e-02
1.11240931e-01 9.10197422e-02 -1.59789014e+00 3.61165404e-01
7.76503026e-01 1.13007510e+00 2.25883856e-01 3.82805020e-01
1.40611663e-01 9.04665947e-01 -2.32464507e-01 -1.09630466e-01
2.41957337e-01 -4.06321943e-01 -8.92001450e-01 4.96369153e-01
4.42133933e-01 1.60773784e-01 4.24472243e-02 1.06440413e+00
3.51370811e-01 5.56033626e-02 8.23311031e-01 -1.07503867e+00
-4.60402161e-01 4.37509447e-01 -6.86191261e-01 -1.37357295e-01
-1.05067536e-01 4.13654633e-02 1.04141879e+00 -1.08035862e+00
4.55187500e-01 8.51663828e-01 5.15971780e-01 4.24065590e-01
-1.36710167e+00 -6.16058111e-01 2.37723053e-01 1.62209854e-01
-1.66239965e+00 -5.59468329e-01 7.61044979e-01 -6.99985325e-01
5.96986830e-01 1.79359838e-01 2.08494380e-01 7.48900294e-01
-3.50896776e-01 8.29652607e-01 1.29035592e+00 -7.17057228e-01
3.56210113e-01 1.41206324e-01 3.19964737e-01 7.70686150e-01
1.11878030e-01 -8.06513950e-02 -1.73722401e-01 1.29478564e-02
5.01708508e-01 6.51364326e-02 -2.13947892e-01 -5.38446784e-01
-1.20410454e+00 9.59831178e-01 7.40060568e-01 4.41523403e-01
-2.88569033e-01 1.54986754e-01 3.21951807e-01 1.80778980e-01
6.94826782e-01 3.94157797e-01 -4.45597708e-01 2.15477020e-01
-1.02454543e+00 -6.38429895e-02 6.11347675e-01 7.58581519e-01
8.57256889e-01 -1.90628946e-01 1.69919748e-02 9.06832933e-01
2.60357082e-01 1.64900392e-01 5.71277976e-01 -8.75845015e-01
2.97968626e-01 5.66894412e-01 -8.26453790e-02 -8.52025151e-01
-4.01449203e-01 -7.12564111e-01 -9.54790354e-01 2.31002808e-01
5.94819725e-01 -3.06460802e-02 -1.09144366e+00 1.52823448e+00
1.73759028e-01 3.14005554e-01 9.10881162e-02 9.35902834e-01
8.38156343e-01 3.83264661e-01 1.09974302e-01 1.69994272e-02
1.29120815e+00 -1.20595765e+00 -3.39004844e-01 -1.77024260e-01
7.68282950e-01 -8.06365311e-01 1.22628129e+00 5.87324619e-01
-6.63748145e-01 -5.88026881e-01 -8.72467637e-01 1.38811439e-01
-2.70925432e-01 5.96671045e-01 5.35938859e-01 5.86624384e-01
-9.38392878e-01 3.99885923e-01 -8.77155125e-01 -3.85135114e-01
7.48586357e-01 3.63881439e-01 -4.83571947e-01 -3.13364029e-01
-5.52979350e-01 4.95476514e-01 6.43637836e-01 -7.57433772e-02
-7.92615235e-01 -4.23406959e-01 -6.99873686e-01 -1.18653357e-01
5.35261869e-01 -3.86750668e-01 1.08748984e+00 -1.33271587e+00
-1.30223501e+00 1.01888204e+00 -9.97556224e-02 -6.13769650e-01
5.02059877e-01 -1.25425220e-01 -8.75116065e-02 4.31586057e-01
8.85595605e-02 9.13074434e-01 7.08657086e-01 -1.57497144e+00
-6.07663214e-01 -2.52646774e-01 -7.36157745e-02 2.84920111e-02
-5.40944695e-01 -6.00781292e-02 -7.74854720e-01 -6.71199977e-01
8.67412537e-02 -1.23329520e+00 -5.05204856e-01 -7.20315278e-02
-4.67338443e-01 -2.80234873e-01 5.09096980e-01 -2.72276551e-01
7.84482062e-01 -2.17272091e+00 -5.59359789e-02 2.33045131e-01
2.28032887e-01 3.71283054e-01 -1.47975281e-01 1.49991810e-01
-9.30344090e-02 1.21792264e-01 -5.37633300e-01 -6.60186231e-01
-1.06423870e-01 8.60434491e-03 -3.50350589e-02 5.69000661e-01
3.53658676e-01 8.96969557e-01 -8.10320616e-01 -4.96721774e-01
3.64055365e-01 4.77080882e-01 -6.52946413e-01 3.79688758e-03
-1.56221032e-01 6.60515010e-01 -2.99222916e-01 5.35980463e-01
4.08356041e-01 -5.48162818e-01 1.18275248e-01 -1.40714228e-01
-6.73764646e-02 1.58280227e-02 -1.02537668e+00 1.85193634e+00
-4.43789274e-01 4.10140812e-01 -2.01289818e-01 -1.51558149e+00
7.22943306e-01 2.88695186e-01 5.45882761e-01 -5.76835036e-01
2.03283370e-01 2.21616998e-01 3.53219956e-02 -3.03683609e-01
6.37916848e-02 -8.09714049e-02 -1.54126152e-01 1.68514460e-01
2.62969077e-01 1.71642840e-01 3.78607780e-01 1.48037225e-01
8.64771664e-01 -4.96404693e-02 2.63931096e-01 -5.02517343e-01
5.85672140e-01 2.00769603e-01 5.07703185e-01 6.46324217e-01
-3.63206454e-02 1.10745132e+00 2.25402698e-01 -2.07654670e-01
-7.12097824e-01 -8.64106178e-01 -1.16402552e-01 1.09612393e+00
5.64698596e-04 -4.79217529e-01 -8.96594703e-01 -9.03659761e-01
-2.76361048e-01 4.90416616e-01 -5.21657646e-01 1.33087989e-02
-2.55534291e-01 -8.91947567e-01 3.32644135e-01 5.54487526e-01
5.03586769e-01 -9.40092742e-01 -2.82570839e-01 9.37572271e-02
-3.29107761e-01 -1.33359921e+00 -2.83462524e-01 3.27122420e-01
-8.28730345e-01 -1.12995732e+00 -6.37924433e-01 -1.09819603e+00
1.03192008e+00 3.19822460e-01 1.06677353e+00 2.11277142e-01
-3.97513002e-01 3.74521434e-01 -6.17837787e-01 -3.42584103e-01
2.87067275e-02 2.58997560e-01 -2.49741241e-01 3.22404444e-01
3.98574442e-01 -4.05766666e-01 -6.98004305e-01 4.65722680e-01
-7.90666938e-01 8.93909670e-03 7.40155280e-01 7.43705928e-01
9.56155002e-01 1.97257727e-01 6.17933095e-01 -1.06439137e+00
9.02512763e-03 -4.84408408e-01 -5.46776831e-01 7.04831481e-02
-4.51187998e-01 -4.96596284e-02 7.74778366e-01 -3.90682191e-01
-6.83930337e-01 7.89458632e-01 -1.44888490e-01 -4.96113539e-01
-6.01302564e-01 6.41498804e-01 -2.54861563e-01 -3.22509781e-02
7.72658467e-01 2.29951233e-01 -8.59939307e-02 -6.98146999e-01
4.45251226e-01 6.93117559e-01 6.14309490e-01 -5.40409446e-01
9.30045307e-01 7.47720540e-01 -3.35143924e-01 -8.39869499e-01
-1.04088187e+00 -8.97519827e-01 -9.26584005e-01 5.34932464e-02
1.07592905e+00 -1.27327275e+00 -2.12289050e-01 3.28637064e-01
-6.43075407e-01 -5.16961575e-01 -2.60757565e-01 6.26906395e-01
-6.27999604e-01 3.82872105e-01 -3.52450818e-01 -4.38592702e-01
-2.60150462e-01 -1.09525383e+00 9.30981398e-01 5.25459126e-02
-3.17349136e-02 -8.75872076e-01 -2.35138476e-01 6.36282325e-01
1.35088235e-01 3.68814468e-01 5.36815703e-01 -7.31634021e-01
-5.16813993e-01 -2.41700128e-01 -2.60671109e-01 6.58332229e-01
1.40523285e-01 -2.20519543e-01 -1.09765863e+00 -4.95536417e-01
-1.20622823e-02 -6.83241069e-01 1.30683374e+00 2.51257956e-01
1.47778070e+00 -2.69576639e-01 -2.28915900e-01 5.35443366e-01
1.57254195e+00 -1.96405426e-01 4.87280995e-01 1.53141871e-01
8.60332787e-01 6.09057426e-01 6.36771619e-01 1.84878647e-01
5.35738885e-01 6.36032283e-01 4.03872073e-01 -2.38087207e-01
-3.21958631e-01 -2.40702685e-02 2.03171372e-01 7.76324034e-01
-8.86215940e-02 -1.01743616e-01 -1.10276711e+00 7.10910261e-01
-2.04611230e+00 -7.13409126e-01 -2.56388932e-01 2.28950548e+00
8.10031593e-01 2.57966101e-01 2.79510558e-01 2.51902133e-01
6.77329123e-01 -2.34785497e-01 -2.46194154e-01 3.48616004e-01
1.15352228e-01 2.23203018e-01 4.99239832e-01 3.52090329e-01
-1.65336573e+00 1.18436992e+00 4.61456156e+00 9.06402707e-01
-1.19848764e+00 2.56657302e-01 8.32575738e-01 1.77832786e-02
2.37375453e-01 5.64563610e-02 -5.36701739e-01 4.72981393e-01
7.16695487e-01 2.54503459e-01 1.12431750e-01 7.87005126e-01
9.48516876e-02 -1.41373843e-01 -1.09532952e+00 1.10800087e+00
1.89901009e-01 -1.06823456e+00 -1.98828593e-01 5.70580810e-02
8.87547076e-01 3.37113053e-01 9.30095911e-02 1.86501294e-01
2.37978771e-01 -1.14733016e+00 7.10225224e-01 3.45355272e-01
6.14242256e-01 -7.97219217e-01 6.79818451e-01 5.09900928e-01
-1.18311298e+00 2.30002776e-02 -3.66632819e-01 -3.46872285e-02
-2.67686874e-01 8.08666110e-01 -5.44734836e-01 5.38425028e-01
7.41471827e-01 8.17918062e-01 -7.84030378e-01 1.33858263e+00
-3.18104237e-01 1.08414352e+00 -4.40072924e-01 2.97851801e-01
4.17927831e-01 -1.61266066e-02 1.98811311e-02 1.29834223e+00
2.23678928e-02 -6.26416877e-02 7.33527720e-01 5.34385443e-01
-3.03075939e-01 2.41292775e-01 -3.02559495e-01 1.28934398e-01
1.06756374e-01 1.42697847e+00 -1.19871044e+00 -3.52662623e-01
-4.44499135e-01 8.68007123e-01 4.22540158e-01 3.69002432e-01
-6.68269932e-01 -3.14254373e-01 1.73238233e-01 1.41702607e-01
6.41981363e-01 -2.52871722e-01 -2.09627226e-01 -1.28316915e+00
1.50371835e-01 -7.64722586e-01 4.79491383e-01 -3.07417333e-01
-1.44715238e+00 6.88451052e-01 -1.63651422e-01 -1.41587198e+00
-2.67807275e-01 -4.53208804e-01 -4.29837674e-01 3.25601369e-01
-1.79028344e+00 -1.27226210e+00 -3.72272998e-01 7.47230887e-01
6.24752581e-01 -1.69141725e-01 8.43174875e-01 4.09180373e-01
-4.62615550e-01 6.25586152e-01 3.52710962e-01 5.84645092e-01
7.58892119e-01 -1.28967953e+00 5.29586039e-02 8.27462435e-01
6.12628460e-01 3.06976259e-01 3.98009330e-01 -2.46074408e-01
-1.02846634e+00 -1.53473258e+00 7.68140435e-01 -4.98997629e-01
4.70357239e-01 -6.75344110e-01 -9.21514571e-01 5.01765847e-01
8.69909748e-02 6.73259079e-01 8.24927390e-01 9.79600623e-02
-6.48013532e-01 -5.19841947e-02 -9.37480569e-01 2.64237672e-01
9.57261324e-01 -3.54178786e-01 -3.68594527e-01 4.89717752e-01
4.31187510e-01 -1.15222177e-02 -8.13097477e-01 3.81466985e-01
2.40517199e-01 -7.66429961e-01 9.30331469e-01 -3.56791615e-01
3.88080627e-01 -5.57786405e-01 -2.52906084e-01 -1.14742470e+00
-2.84785718e-01 -2.80996531e-01 1.61089033e-01 1.27903605e+00
4.26906466e-01 -4.93939698e-01 8.58628809e-01 1.58440724e-01
-8.87052193e-02 -6.90996468e-01 -6.10216498e-01 -9.92328584e-01
2.39624932e-01 -4.36741233e-01 9.69789177e-03 1.12917578e+00
-2.40976155e-01 4.81409848e-01 -1.89253718e-01 1.49283648e-01
8.98897171e-01 3.56231451e-01 8.00955176e-01 -1.33496249e+00
-3.12017977e-01 -4.06716198e-01 -5.49174488e-01 -8.56665194e-01
3.32758397e-01 -1.18644190e+00 2.26353973e-01 -1.43846154e+00
3.07722837e-01 -6.40377998e-01 -6.29515350e-01 7.15606451e-01
-1.64901525e-01 8.14310074e-01 4.32719618e-01 5.59331596e-01
-1.05256295e+00 5.22353351e-01 6.86264873e-01 -1.88848987e-01
-5.74427992e-02 -3.96692306e-02 -8.11457396e-01 9.64529097e-01
1.15778041e+00 -7.18929231e-01 -5.24135530e-01 -3.78019750e-01
-4.35072452e-01 -5.58923602e-01 5.63540578e-01 -1.25341868e+00
3.25395584e-01 1.38371512e-01 3.62497300e-01 -3.41255665e-01
8.57438296e-02 -8.80759835e-01 -2.67410666e-01 3.62509519e-01
-3.55536878e-01 -4.31485742e-01 -9.79812164e-03 5.59396625e-01
-4.27222043e-01 -1.43555224e-01 9.13848877e-01 -1.42435521e-01
-8.59007061e-01 2.85684437e-01 -2.53296822e-01 2.24410132e-01
1.05375218e+00 3.13872099e-02 -1.09738432e-01 -2.06519127e-01
-7.95742035e-01 2.47787908e-01 5.01651227e-01 4.19082850e-01
4.58186865e-01 -1.10612035e+00 -8.66021872e-01 -8.59857649e-02
4.18828756e-01 2.35029340e-01 5.90315536e-02 9.52162325e-01
-3.13180417e-01 3.02739084e-01 1.43937413e-02 -9.04662549e-01
-1.22755992e+00 5.19952118e-01 1.85148180e-01 -2.13880777e-01
-5.70204735e-01 7.27581739e-01 3.63745987e-01 -4.46024746e-01
4.25233692e-01 -2.02423111e-01 -1.88605040e-01 1.74848884e-01
4.55308288e-01 -1.16376594e-01 2.62686551e-01 -7.49991596e-01
-3.45596582e-01 7.16294646e-01 -2.78363407e-01 5.82239628e-02
1.51804435e+00 -5.68971038e-02 1.18481427e-01 4.21260059e-01
1.59668219e+00 -2.60961413e-01 -1.28692305e+00 -4.72373694e-01
3.10464203e-01 -3.57826710e-01 1.76823676e-01 -8.48289669e-01
-1.35345912e+00 8.26148391e-01 7.59395659e-01 -1.19348923e-02
1.16100419e+00 2.33752698e-01 5.44989288e-01 5.32153130e-01
2.85499513e-01 -1.00652313e+00 2.40938753e-01 3.66313875e-01
6.70526862e-01 -1.72739506e+00 -1.67851355e-02 -5.43699980e-01
-7.03152061e-01 7.41829395e-01 5.11399508e-01 -4.09448326e-01
8.19114387e-01 7.95480460e-02 1.86834231e-01 -1.18579321e-01
-6.14627421e-01 -7.36894250e-01 4.83470201e-01 5.19156158e-01
5.44534087e-01 3.77154537e-02 -1.60670400e-01 5.96784115e-01
-1.12945504e-01 4.10709493e-02 2.57259011e-01 9.11264718e-01
-4.96946365e-01 -1.12411177e+00 -2.44753972e-01 3.92155319e-01
-5.76631069e-01 -1.32283270e-01 -4.52971786e-01 7.43923187e-01
1.62920505e-01 1.14178085e+00 -6.04017638e-03 -2.47905016e-01
1.09552622e-01 -4.89727687e-03 3.47611487e-01 -8.20753098e-01
-4.15728390e-01 3.81649792e-01 -1.11762390e-01 -4.53773558e-01
-7.26232350e-01 -5.69653988e-01 -1.41122127e+00 2.83848047e-01
-3.86435509e-01 2.22026393e-01 6.05067551e-01 8.63357008e-01
4.18984860e-01 3.08453232e-01 6.88266337e-01 -8.74844849e-01
-2.89256752e-01 -7.28713274e-01 -3.35999846e-01 6.85127914e-01
2.31888026e-01 -5.30514479e-01 -2.75869578e-01 4.07543957e-01] | [9.456124305725098, 2.5536465644836426] |
95c96ede-eac9-4dbd-a17d-87203babbe47 | hemlets-posh-learning-part-centric-heatmap | 2003.04894 | null | https://arxiv.org/abs/2003.04894v3 | https://arxiv.org/pdf/2003.04894v3.pdf | HEMlets PoSh: Learning Part-Centric Heatmap Triplets for 3D Human Pose and Shape Estimation | Estimating 3D human pose from a single image is a challenging task. This work attempts to address the uncertainty of lifting the detected 2D joints to the 3D space by introducing an intermediate state-Part-Centric Heatmap Triplets (HEMlets), which shortens the gap between the 2D observation and the 3D interpretation. The HEMlets utilize three joint-heatmaps to represent the relative depth information of the end-joints for each skeletal body part. In our approach, a Convolutional Network (ConvNet) is first trained to predict HEMlets from the input image, followed by a volumetric joint-heatmap regression. We leverage on the integral operation to extract the joint locations from the volumetric heatmaps, guaranteeing end-to-end learning. Despite the simplicity of the network design, the quantitative comparisons show a significant performance improvement over the best-of-grade methods (e.g. $20\%$ on Human3.6M). The proposed method naturally supports training with "in-the-wild" images, where only weakly-annotated relative depth information of skeletal joints is available. This further improves the generalization ability of our model, as validated by qualitative comparisons on outdoor images. Leveraging the strength of the HEMlets pose estimation, we further design and append a shallow yet effective network module to regress the SMPL parameters of the body pose and shape. We term the entire HEMlets-based human pose and shape recovery pipeline HEMlets PoSh. Extensive quantitative and qualitative experiments on the existing human body recovery benchmarks justify the state-of-the-art results obtained with our HEMlets PoSh approach. | ['Xiaoguang Han', 'Nianjuan Jiang', 'Kun Zhou', 'Kui Jia', 'Jiangbo Lu'] | 2020-03-10 | null | null | null | null | ['3d-human-pose-and-shape-estimation'] | ['computer-vision'] | [-1.42368048e-01 5.36757171e-01 -2.08915085e-01 -1.95867792e-01
-6.59433544e-01 -1.03268228e-01 3.36838543e-01 -2.73375630e-01
-5.60853899e-01 3.46055329e-01 2.06516832e-01 3.62363338e-01
1.93673879e-01 -4.65823710e-01 -9.26555812e-01 -4.33416605e-01
-2.06504822e-01 8.13899159e-01 2.87241906e-01 -3.79400313e-01
-1.76297247e-01 3.88969272e-01 -1.43679535e+00 -5.88197894e-02
4.54974115e-01 1.28137243e+00 -2.79269010e-01 4.44851846e-01
3.73091638e-01 3.37415367e-01 -3.80846411e-01 -5.92981160e-01
6.62693143e-01 -1.92734480e-01 -6.53461099e-01 3.09253693e-01
6.51122749e-01 -6.55530572e-01 -4.83376086e-01 5.14864564e-01
6.96283996e-01 -9.88562591e-03 5.82593799e-01 -1.22869968e+00
-1.42450169e-01 9.74072739e-02 -9.06518817e-01 -4.03842241e-01
6.25362813e-01 4.80231673e-01 8.14783156e-01 -1.09380603e+00
6.95063412e-01 1.31980658e+00 1.10169864e+00 5.34251750e-01
-1.04033506e+00 -6.05118990e-01 1.00469686e-01 -2.14837521e-01
-1.46095002e+00 -1.93319857e-01 9.25388277e-01 -5.70979059e-01
8.31202447e-01 -4.90445644e-04 1.07759035e+00 1.03127134e+00
2.72244275e-01 9.16257739e-01 8.47434461e-01 -3.89993489e-01
1.93983093e-02 -3.99491787e-01 -2.10861474e-01 1.11379921e+00
2.58115828e-01 1.22460328e-01 -6.27313673e-01 -3.71779539e-02
1.02970707e+00 1.05499528e-01 -1.52440473e-01 -9.92560327e-01
-1.20774484e+00 5.00270069e-01 8.48767519e-01 -2.96212077e-01
-4.40402746e-01 4.44700956e-01 3.84783506e-01 -1.69352770e-01
5.29209971e-01 7.79656544e-02 -4.28114504e-01 1.30671756e-02
-1.04241467e+00 5.59868455e-01 5.91722310e-01 8.80996764e-01
6.72008216e-01 -1.56906381e-01 -1.80874199e-01 6.53379083e-01
5.42137086e-01 7.05251515e-01 2.91904751e-02 -9.73200977e-01
6.69589281e-01 8.13179910e-01 8.17114040e-02 -8.69138360e-01
-7.21169710e-01 -4.58047450e-01 -7.87035704e-01 3.56926441e-01
6.09565735e-01 -1.24929890e-01 -1.25785577e+00 1.68138826e+00
7.57827878e-01 -2.51224607e-01 -4.20988709e-01 1.38378775e+00
4.40141827e-01 2.38096610e-01 8.30584541e-02 3.80532742e-01
1.43641150e+00 -1.05252361e+00 -1.94320053e-01 -5.06133318e-01
4.20810223e-01 -3.84081006e-01 9.85105395e-01 2.57552266e-01
-1.29208767e+00 -6.97898984e-01 -1.00170815e+00 -3.26921582e-01
-2.22737193e-02 4.11538571e-01 5.06458461e-01 4.00011480e-01
-6.73697948e-01 6.41809046e-01 -1.30904865e+00 -3.00974578e-01
2.97974586e-01 5.02663374e-01 -7.05016375e-01 1.16146272e-02
-1.01385403e+00 9.41787958e-01 1.68507680e-01 6.49683654e-01
-7.91246414e-01 -6.29764199e-01 -1.21076500e+00 -2.89202124e-01
4.85022098e-01 -1.19096899e+00 9.79056954e-01 -3.18716466e-01
-1.60297966e+00 1.04087400e+00 2.10797623e-01 -3.50524902e-01
1.01704216e+00 -7.95396805e-01 2.86443710e-01 3.91346097e-01
1.00247808e-01 1.03709638e+00 9.65564251e-01 -1.18780828e+00
-3.01476926e-01 -6.66169107e-01 -8.42164904e-02 2.47240216e-01
-6.42223880e-02 -4.13913667e-01 -9.58309531e-01 -7.31669366e-01
3.29587519e-01 -1.35358453e+00 -3.44820410e-01 5.41393101e-01
-6.10246241e-01 4.34318781e-02 4.30813253e-01 -9.73942220e-01
9.96384442e-01 -1.96012878e+00 6.57219291e-01 3.33349824e-01
2.19897255e-01 -4.01919410e-02 1.68546721e-01 2.04868034e-01
1.70334116e-01 -3.04784566e-01 -3.91004771e-01 -9.18155909e-01
1.87138647e-01 2.56355554e-01 1.80832207e-01 8.23097587e-01
3.53792399e-01 1.11444020e+00 -6.51131034e-01 -6.60771310e-01
3.36400986e-01 7.29415834e-01 -7.22022712e-01 4.47921842e-01
-3.38796936e-02 4.97971952e-01 -2.35823020e-01 8.53900373e-01
5.27605653e-01 -2.01988056e-01 7.93663338e-02 -5.15420020e-01
3.13973725e-01 4.64134524e-03 -1.08872807e+00 2.35742760e+00
-3.28996599e-01 -3.49706854e-03 6.58909380e-02 -6.66684091e-01
7.32996464e-01 2.51505882e-01 7.55502284e-01 -4.63180780e-01
1.11739881e-01 1.65232912e-01 -1.81478113e-01 -2.51513451e-01
2.48106450e-01 -1.31450295e-01 -3.79431337e-01 6.24029823e-02
3.03144664e-01 -8.68054256e-02 -1.26631796e-01 -9.06203017e-02
8.32325041e-01 9.10673499e-01 1.77780375e-01 -1.27669513e-01
2.82621682e-01 -1.26613900e-01 5.44344962e-01 2.47665361e-01
-2.46070489e-01 9.79106903e-01 4.37176853e-01 -4.64021027e-01
-1.19912744e+00 -1.39233851e+00 2.06452385e-01 8.71756911e-01
1.23064034e-01 -4.56649363e-01 -9.06018257e-01 -8.59338880e-01
3.45483869e-01 -1.13432072e-01 -8.81966650e-01 -1.45550817e-01
-9.14146602e-01 -3.91443521e-01 6.50520861e-01 9.40942705e-01
5.44465959e-01 -7.05595613e-01 -1.07215130e+00 -1.93799008e-03
-2.93543369e-01 -1.17110264e+00 -4.15435553e-01 8.51302296e-02
-9.47196245e-01 -1.02264190e+00 -1.10889292e+00 -4.75768775e-01
7.32728481e-01 -2.41323337e-01 1.00508201e+00 3.43005247e-02
-5.00324130e-01 3.32171023e-01 -1.56767145e-01 2.77793929e-02
1.05989300e-01 1.04425289e-01 1.21646695e-01 -2.05361038e-01
-1.45129412e-01 -5.73494673e-01 -1.09935999e+00 4.76202875e-01
-4.82578397e-01 2.08584741e-01 7.69305229e-01 8.62480342e-01
6.92747295e-01 -4.46379066e-01 2.90204529e-02 -3.08052182e-01
-1.32287771e-01 -1.20909497e-01 -2.94973075e-01 3.27413008e-02
-3.57821912e-01 2.00972542e-01 2.62184560e-01 -3.74908805e-01
-8.50981176e-01 5.81062138e-01 -4.05592918e-01 -5.82237005e-01
-5.65339737e-02 2.94914782e-01 -2.98132539e-01 1.12784110e-01
5.78656793e-01 -1.55037940e-01 4.10542816e-01 -6.99842930e-01
4.22878861e-01 1.07257061e-01 8.88442576e-01 -8.39689612e-01
9.07064915e-01 6.89267933e-01 2.76474863e-01 -5.22725642e-01
-9.30218875e-01 -5.94936967e-01 -1.04067802e+00 -3.98138523e-01
1.03246307e+00 -1.14989436e+00 -7.89914250e-01 5.22805214e-01
-1.02978611e+00 -3.81566197e-01 -3.64917219e-01 4.66893524e-01
-7.90429294e-01 5.27271688e-01 -8.60781729e-01 -7.81192422e-01
-6.19061470e-01 -1.13431287e+00 1.79565024e+00 -2.25389734e-01
-4.02637005e-01 -5.40640593e-01 3.34557630e-02 5.40214837e-01
-5.57655357e-02 7.48449445e-01 5.33169031e-01 -9.01101381e-02
-2.39477441e-01 -4.69896942e-01 2.97396723e-02 2.92889237e-01
-3.61120284e-01 -3.80341828e-01 -9.40132022e-01 -4.85623747e-01
-2.76205003e-01 -5.04481375e-01 8.03742111e-01 3.61694276e-01
8.57794881e-01 -1.51061282e-01 -2.14198798e-01 6.99025989e-01
1.08554614e+00 -7.04587281e-01 7.21091211e-01 3.36198151e-01
1.02280354e+00 8.05592239e-01 8.19929421e-01 5.80987513e-01
4.98590499e-01 1.00489128e+00 6.74413562e-01 -3.41335833e-01
-4.08775181e-01 -6.30180955e-01 5.22038639e-01 5.68730533e-01
-5.25272131e-01 1.95601135e-01 -8.40342581e-01 3.51349235e-01
-1.93237090e+00 -4.86857563e-01 7.64102563e-02 2.21777248e+00
7.89999843e-01 3.54876041e-01 4.45956111e-01 8.87510255e-02
3.87737602e-01 7.06527308e-02 -5.57347596e-01 2.44883582e-01
2.83634633e-01 3.21452707e-01 4.64609593e-01 3.61465335e-01
-1.13084328e+00 7.55983353e-01 5.35100651e+00 5.38078964e-01
-9.03702021e-01 5.53672947e-02 2.83850908e-01 -2.49253646e-01
2.40656078e-01 -5.70809133e-02 -8.42724323e-01 2.08423197e-01
5.08078992e-01 6.66533649e-01 -9.94001850e-02 8.47906530e-01
3.36924158e-02 -6.36828914e-02 -1.24025965e+00 1.03487182e+00
1.03119537e-01 -6.66255474e-01 -1.13015734e-01 2.02446103e-01
4.68969584e-01 -5.62039167e-02 2.49003991e-02 3.31453294e-01
-1.26680925e-01 -9.16439116e-01 1.06431019e+00 6.23241723e-01
8.30699563e-01 -6.34630978e-01 6.86218143e-01 3.92187029e-01
-1.37790036e+00 1.72696740e-03 -1.19737014e-01 -6.14242554e-02
3.92295778e-01 5.79181731e-01 -6.86034024e-01 7.38783836e-01
8.60342920e-01 5.01730382e-01 -7.47275233e-01 8.25437248e-01
-4.51565951e-01 2.29801491e-01 -6.18857563e-01 6.00574315e-01
-6.87200427e-02 8.67688730e-02 3.22875470e-01 1.14739144e+00
2.02882096e-01 -1.61932781e-01 4.47901189e-01 8.88565958e-01
-3.05558164e-02 -8.74005482e-02 -2.16277599e-01 3.84898454e-01
-3.18909883e-02 1.22052550e+00 -5.51206589e-01 -1.66966274e-01
-7.77217448e-02 1.27397835e+00 3.98738742e-01 1.04424551e-01
-1.01985133e+00 -5.22734337e-02 5.17485619e-01 6.78253055e-01
2.59855539e-01 -4.19759929e-01 -2.83700585e-01 -1.16003346e+00
4.79846537e-01 -6.95518315e-01 3.27664882e-01 -5.92005074e-01
-1.20005357e+00 2.96756059e-01 3.78064603e-01 -1.26330829e+00
-6.80714369e-01 -6.84880316e-01 -1.05422609e-01 7.37736344e-01
-1.07553840e+00 -1.67399895e+00 -4.67817932e-01 4.75980192e-01
3.39156181e-01 4.19590384e-01 6.46363318e-01 1.66820213e-01
-4.16591376e-01 7.52661705e-01 -7.22466290e-01 4.44642544e-01
7.74717867e-01 -1.17183638e+00 7.18774498e-01 7.03851819e-01
-2.13371500e-01 6.04797840e-01 6.15361989e-01 -8.10439825e-01
-1.58898211e+00 -8.63030314e-01 4.87371713e-01 -6.99076414e-01
2.70847023e-01 -6.73396647e-01 -5.42378306e-01 5.82835078e-01
-2.95617938e-01 4.22428995e-01 3.71884465e-01 6.23763129e-02
-4.56840724e-01 -2.84504821e-03 -1.13107514e+00 5.01776040e-01
1.33634162e+00 -3.09357017e-01 -6.64822042e-01 7.08402097e-02
5.43969154e-01 -7.14891732e-01 -1.24710059e+00 8.80905449e-01
1.07738805e+00 -8.62044811e-01 1.31749892e+00 -3.77205759e-01
6.55420899e-01 -4.88382071e-01 -1.09804608e-01 -9.39725101e-01
-5.89023009e-02 -2.97817320e-01 -5.94130218e-01 7.62434423e-01
1.12041987e-01 -8.06058869e-02 1.23995578e+00 7.15108871e-01
-1.43417984e-01 -1.05923128e+00 -1.01702869e+00 -7.39632905e-01
9.10482630e-02 -3.23463619e-01 1.65048063e-01 4.58628029e-01
-2.07368713e-02 1.62504643e-01 -6.22009754e-01 1.61837146e-01
9.68714356e-01 -5.59854656e-02 1.18018031e+00 -1.20119071e+00
-6.13903046e-01 -6.90947324e-02 -6.71100974e-01 -1.37822449e+00
-1.92727055e-03 -4.88030910e-01 3.07729304e-01 -1.34734738e+00
1.99159294e-01 -1.73399389e-01 -5.85903376e-02 7.70989656e-01
-1.26565054e-01 6.35789037e-01 4.09960359e-01 1.97791204e-01
-5.21479130e-01 6.97579265e-01 1.30074275e+00 -3.45176719e-02
-1.04986079e-01 -1.50297403e-01 -1.16358556e-01 9.33307171e-01
4.05000061e-01 -2.68300474e-01 -1.62874997e-01 -3.60373884e-01
1.03967361e-01 5.38683124e-02 9.85048890e-01 -1.29513490e+00
7.77202547e-02 2.23020762e-01 7.82743990e-01 -8.28990698e-01
7.25045025e-01 -8.78764451e-01 2.66819566e-01 7.54108548e-01
-1.54785797e-01 -5.43954642e-03 -3.39749753e-02 4.55846936e-01
7.80511126e-02 2.86901563e-01 7.00057685e-01 -2.01366797e-01
-4.57615286e-01 4.72485155e-01 3.41693193e-01 1.36559546e-01
8.17713082e-01 -3.65340948e-01 9.64454785e-02 -1.69494748e-01
-8.15661728e-01 1.34778976e-01 7.59181201e-01 3.28910470e-01
7.33950257e-01 -1.31687033e+00 -6.20119572e-01 2.31814176e-01
2.46530339e-01 5.06983876e-01 2.47565284e-01 1.18494499e+00
-5.67509234e-01 1.71592981e-01 -3.11527699e-01 -9.63408053e-01
-9.71001804e-01 3.79146904e-01 4.18004870e-01 -4.99081939e-01
-9.41022456e-01 5.83159864e-01 1.95738390e-01 -4.83413279e-01
3.31423461e-01 -4.29621339e-01 3.51296812e-01 -2.19102576e-01
7.02800453e-02 4.65567112e-01 1.35325372e-01 -8.20371509e-01
-4.31727320e-01 8.57750952e-01 7.80833811e-02 -2.93573260e-01
1.16208673e+00 4.87790164e-03 2.25071698e-01 1.52981296e-01
1.38935912e+00 -4.37153727e-02 -1.79001474e+00 3.39058042e-02
-3.63236517e-01 -3.76525432e-01 -3.61075282e-01 -7.36506522e-01
-1.14980114e+00 1.03896093e+00 6.98099673e-01 -6.12434983e-01
8.36593211e-01 8.83899480e-02 1.06412995e+00 5.72737902e-02
5.66609919e-01 -1.18958449e+00 2.57923782e-01 2.13985503e-01
9.88909006e-01 -1.24249923e+00 3.30639184e-01 -5.75146675e-01
-4.58534211e-01 9.69392359e-01 7.37758398e-01 -2.64598131e-01
5.21705627e-01 1.48968160e-01 -4.15004641e-02 -4.27223593e-01
-1.66819066e-01 -2.12887004e-01 6.53573871e-01 5.67003667e-01
4.61760283e-01 2.16237344e-02 -2.59297401e-01 5.12330770e-01
-2.73034811e-01 -9.39173102e-02 -1.46120518e-01 1.14984310e+00
-1.91542074e-01 -9.06713188e-01 -5.94685256e-01 2.55563091e-02
-3.08376193e-01 2.45867774e-01 -4.00643021e-01 1.08284056e+00
2.16998816e-01 4.72881615e-01 -1.28796101e-01 -5.30423522e-01
6.71850443e-01 3.21748219e-02 7.86531448e-01 -4.11805391e-01
-5.08596480e-01 2.51475573e-01 9.81507823e-03 -9.79716539e-01
-4.15652335e-01 -3.93810749e-01 -1.56447709e+00 -8.10362548e-02
-1.39895812e-01 -3.20652097e-01 5.69715023e-01 8.89396787e-01
2.67164558e-01 3.01601499e-01 1.44085661e-01 -1.57162464e+00
-8.61241043e-01 -9.00372744e-01 -3.74471068e-01 7.56070673e-01
1.18623897e-01 -1.16980731e+00 -1.09050028e-01 2.93750651e-02] | [7.004674911499023, -0.96905118227005] |
b53b34d5-e523-4d41-a006-bbf87539ba4f | clrernet-improving-confidence-of-lane | 2305.08366 | null | https://arxiv.org/abs/2305.08366v1 | https://arxiv.org/pdf/2305.08366v1.pdf | CLRerNet: Improving Confidence of Lane Detection with LaneIoU | Lane marker detection is a crucial component of the autonomous driving and driver assistance systems. Modern deep lane detection methods with row-based lane representation exhibit excellent performance on lane detection benchmarks. Through preliminary oracle experiments, we firstly disentangle the lane representation components to determine the direction of our approach. We show that correct lane positions are already among the predictions of an existing row-based detector, and the confidence scores that accurately represent intersection-over-union (IoU) with ground truths are the most beneficial. Based on the finding, we propose LaneIoU that better correlates with the metric, by taking the local lane angles into consideration. We develop a novel detector coined CLRerNet featuring LaneIoU for the target assignment cost and loss functions aiming at the improved quality of confidence scores. Through careful and fair benchmark including cross validation, we demonstrate that CLRerNet outperforms the state-of-the-art by a large margin - enjoying F1 score of 81.43% compared with 80.47% of the existing method on CULane, and 86.47% compared with 86.10% on CurveLanes. | ['Yusuke Uchida', 'Hiroto Honda'] | 2023-05-15 | null | null | null | null | ['lane-detection'] | ['computer-vision'] | [-4.08983618e-01 1.64641395e-01 -4.39795852e-01 -5.41867614e-01
-1.08965349e+00 -3.91834259e-01 4.89322424e-01 -2.10308373e-01
-5.04434705e-01 7.07704902e-01 -1.41564133e-02 -6.91952825e-01
1.21821046e-01 -4.88702565e-01 -7.75146246e-01 -6.12154245e-01
1.87364314e-02 2.37114504e-01 6.27694666e-01 -4.18500125e-01
3.79597038e-01 6.38369024e-01 -1.61612546e+00 -1.52400061e-01
1.22856677e+00 1.03847349e+00 -4.48185578e-02 6.63526118e-01
8.31187963e-02 7.75177240e-01 -2.72095859e-01 -5.35806715e-01
4.77782607e-01 5.36170974e-02 1.28471118e-03 -3.40002894e-01
8.34026933e-01 -4.42326099e-01 -5.26673794e-01 1.00066948e+00
3.34312499e-01 -1.54201880e-01 5.73538065e-01 -1.55452955e+00
1.35311365e-01 1.77211121e-01 -7.73162603e-01 1.15973718e-01
1.61162801e-02 3.81976038e-01 1.12146473e+00 -1.13887155e+00
5.01416385e-01 1.08090329e+00 8.42321694e-01 2.14782059e-02
-8.73027205e-01 -8.64739954e-01 1.16951860e-01 5.68452716e-01
-1.64310229e+00 -4.44597423e-01 6.86711371e-01 -4.55030352e-01
4.76991087e-01 1.46557212e-01 2.17782184e-01 6.66963696e-01
5.71633518e-01 1.03685713e+00 1.20365775e+00 -1.56930760e-01
-2.20741287e-01 2.44908109e-01 4.82205182e-01 1.05984616e+00
5.57967961e-01 5.01642704e-01 -2.97646284e-01 3.18011492e-01
3.08790177e-01 -4.01868522e-01 -1.32947683e-01 -8.78262043e-01
-1.20601439e+00 7.86080599e-01 4.92504478e-01 -4.80105251e-01
-8.78936276e-02 3.98415774e-02 3.56197238e-01 -2.67363898e-02
-1.68501437e-01 3.47048044e-01 6.55738935e-02 -1.16187766e-01
-6.85941041e-01 4.05116200e-01 5.09700418e-01 1.31791711e+00
8.63848567e-01 1.31466672e-01 -3.29276770e-01 5.98796427e-01
2.50454038e-01 7.98664212e-01 -7.81736225e-02 -8.24611902e-01
6.72116578e-01 4.66764718e-01 2.52188236e-01 -9.85180318e-01
-6.00271046e-01 -8.69665563e-01 -6.34585083e-01 5.58752656e-01
6.86317444e-01 -1.37197211e-01 -8.21302295e-01 1.44386649e+00
3.17914598e-02 3.49640727e-01 2.12984204e-01 9.78846252e-01
4.39170033e-01 4.65199172e-01 -6.30240366e-02 1.04741111e-01
1.17880225e+00 -1.03914893e+00 -6.16703093e-01 -6.80991888e-01
8.75993133e-01 -7.36275792e-01 8.81650567e-01 3.36076498e-01
-4.51703250e-01 -6.90402091e-01 -1.53299510e+00 4.74973992e-02
-1.85700804e-01 7.17952669e-01 4.13114548e-01 6.64519489e-01
-7.47225165e-01 -1.52645074e-02 -2.80165941e-01 -1.44228756e-01
1.95917845e-01 -1.07652113e-01 -4.43026215e-01 -1.23424754e-01
-1.27153277e+00 1.27083230e+00 3.02148044e-01 2.64670938e-01
-7.29583144e-01 -6.00417316e-01 -8.30372214e-01 -2.16144785e-01
6.54638767e-01 -1.54358447e-01 1.12506294e+00 -1.25427797e-01
-1.21424472e+00 8.09671521e-01 -2.91862726e-01 -1.07257855e+00
1.13289547e+00 -4.94660020e-01 -6.03311121e-01 -1.58362240e-01
2.62252033e-01 7.94838190e-01 5.47925353e-01 -1.42217886e+00
-1.42209172e+00 -1.30730905e-02 -7.85857365e-02 1.28775045e-01
5.26350737e-01 -6.05369210e-01 -5.18465400e-01 -8.16126913e-03
2.75576651e-01 -1.07379186e+00 -1.91818208e-01 -1.63833708e-01
-7.91485190e-01 -1.32326543e-01 7.66414285e-01 -4.29973543e-01
1.39247203e+00 -2.13444662e+00 -6.81202114e-01 3.57833922e-01
3.67807716e-01 3.14954251e-01 1.61665633e-01 2.53873110e-01
1.83031246e-01 -3.14047217e-01 4.09319298e-03 -4.86024544e-02
1.26211137e-01 5.72582195e-03 -7.29084074e-01 7.16172278e-01
3.83047014e-01 7.76057243e-01 -8.20457637e-01 -4.94146615e-01
4.71250027e-01 1.63065381e-02 -2.53661454e-01 1.97684973e-01
3.58677775e-01 9.21516418e-02 -3.29239488e-01 3.93834352e-01
9.26752567e-01 3.23027164e-01 -8.29858929e-02 -3.26189935e-01
-5.27375758e-01 1.57891855e-01 -1.16906154e+00 1.02440214e+00
-3.27727199e-01 1.21388566e+00 -2.73601443e-01 -4.14823204e-01
1.38492680e+00 -3.34016621e-01 -7.95330480e-02 -8.86752725e-01
6.53930306e-02 5.18297076e-01 4.65152621e-01 -2.26056144e-01
7.52421319e-01 2.52444625e-01 -3.67319345e-01 -4.28768098e-01
-4.99724150e-01 1.29996970e-01 -5.60020916e-02 1.56905726e-01
8.72729838e-01 -9.65586528e-02 3.99866164e-01 -3.34761113e-01
8.08527827e-01 1.17545746e-01 5.68199217e-01 9.94394481e-01
-7.87940264e-01 4.76008266e-01 8.39107454e-01 -3.10285896e-01
-1.12942433e+00 -1.15304315e+00 -3.10158104e-01 6.24108970e-01
5.79033315e-01 4.40858975e-02 -6.36375308e-01 -6.07981980e-01
4.23163265e-01 1.16151524e+00 -4.92345512e-01 -2.08568737e-01
-5.85934162e-01 -3.49237800e-01 7.61369407e-01 7.77360082e-01
6.74294770e-01 -3.37917894e-01 -6.81891978e-01 -1.25636486e-02
2.74790794e-01 -1.40498030e+00 -3.11381429e-01 8.68541449e-02
-1.75978601e-01 -1.23976696e+00 -3.72284144e-01 -6.06714487e-01
3.47231627e-01 5.68831384e-01 8.27309012e-01 -2.72060037e-01
3.15594673e-02 -1.73623532e-01 -1.67553555e-02 -5.02634943e-01
-3.33139062e-01 1.37368530e-01 1.11042978e-02 1.65487289e-01
3.53221416e-01 1.07284211e-01 -7.33130693e-01 6.37661755e-01
6.66731922e-03 4.03380483e-01 1.11834109e+00 8.72087657e-01
5.31888843e-01 -4.36885446e-01 7.60106862e-01 -9.47049201e-01
3.72703671e-01 -2.85452336e-01 -1.04497612e+00 8.81557614e-02
-1.06608081e+00 1.76868826e-01 5.46345472e-01 1.78340867e-01
-8.02070498e-01 2.49407902e-01 -3.29094648e-01 -5.57685435e-01
-1.77912593e-01 6.84455708e-02 -2.43726552e-01 -1.21389806e-01
5.41814566e-01 1.05279975e-01 1.29152894e-01 6.02939017e-02
6.16946578e-01 5.68190038e-01 1.06081033e+00 -8.32288638e-02
8.32992733e-01 5.05751908e-01 1.92240387e-01 -6.49067402e-01
-7.98973799e-01 -8.08177412e-01 -5.96148491e-01 -5.00666261e-01
7.96875894e-01 -1.03491390e+00 -9.14375722e-01 4.03030217e-01
-9.98728812e-01 4.22819965e-02 8.45217705e-02 5.45184374e-01
-5.11451900e-01 2.00257123e-01 6.55689687e-02 -1.07310677e+00
1.12777509e-01 -1.38003862e+00 1.05663049e+00 6.64172322e-02
2.69203484e-01 -5.66758275e-01 -1.76754504e-01 2.48426348e-01
1.44799590e-01 4.19277251e-01 5.40617406e-01 -7.07220912e-01
-9.30158138e-01 -7.86913157e-01 -6.53898001e-01 2.31220514e-01
-4.87746984e-01 -4.93672676e-02 -1.20329714e+00 -1.12870693e-01
-5.85164785e-01 1.61211267e-01 1.19970703e+00 2.34752506e-01
9.10170734e-01 1.23023503e-01 -5.28642774e-01 5.43924809e-01
1.30412114e+00 3.24064553e-01 8.24476898e-01 5.89489996e-01
8.61694455e-01 6.26397669e-01 1.16443622e+00 2.44697146e-02
6.95565760e-01 7.45940506e-01 5.46898305e-01 -1.93367034e-01
-1.21805169e-01 -5.65696299e-01 3.21799308e-01 5.06443501e-01
4.62801635e-01 -2.49401450e-01 -1.13101554e+00 5.84586620e-01
-2.05109429e+00 -7.54768491e-01 -4.75635439e-01 2.35947680e+00
7.56565556e-02 1.00752759e+00 2.48830706e-01 3.04555092e-02
6.88374281e-01 2.89612085e-01 -7.12593436e-01 -3.90939683e-01
-1.48435637e-01 -5.48840582e-01 1.46136570e+00 7.03995943e-01
-1.32336581e+00 1.07704937e+00 5.70124531e+00 1.00317776e+00
-9.65253413e-01 -2.18907163e-01 6.27821922e-01 3.67449701e-01
3.20083089e-02 4.33815131e-03 -1.46165431e+00 3.17332655e-01
8.61537576e-01 -1.43651262e-01 -1.14810258e-01 1.16497922e+00
3.59735608e-01 -1.99164689e-01 -9.53400552e-01 9.65322435e-01
9.90019664e-02 -1.23848271e+00 -2.46680275e-01 9.41805094e-02
5.77025294e-01 1.65787041e-01 1.97107404e-01 5.61507225e-01
1.64572895e-01 -9.31471884e-01 1.11307847e+00 6.21144772e-01
8.24547708e-01 -1.02871442e+00 1.18728685e+00 5.60298264e-01
-1.19958174e+00 -1.23484582e-01 -2.62979150e-01 1.96672276e-01
1.98731720e-01 3.58601391e-01 -1.44737506e+00 5.33566415e-01
1.26418352e-01 5.15162826e-01 -6.27897680e-01 1.27460659e+00
-4.30951834e-01 7.06212401e-01 -1.23497389e-01 -8.44284892e-02
5.83588839e-01 -1.55946836e-01 8.60238850e-01 1.40861964e+00
1.00000843e-01 -2.55520999e-01 3.10512006e-01 6.73249066e-01
1.36822954e-01 -5.68379872e-02 -9.18103278e-01 7.40859807e-01
6.82070255e-01 1.15872717e+00 -2.40311638e-01 -5.46594039e-02
-4.05678660e-01 1.94564283e-01 2.67832786e-01 1.56453684e-01
-1.28096437e+00 -7.26329744e-01 1.01535273e+00 3.41706157e-01
2.34857142e-01 -3.16365838e-01 -4.66675580e-01 -5.38780093e-01
7.56327808e-02 -3.95825088e-01 -3.62714455e-02 -6.49510562e-01
-8.54894936e-01 7.11824596e-01 3.03566153e-03 -1.85366488e+00
-2.45177090e-01 -9.38219905e-01 -5.47520578e-01 7.81370997e-01
-2.07567096e+00 -1.04117227e+00 -5.18366098e-01 -1.26538515e-01
5.17777383e-01 -3.55823338e-01 -6.22449210e-03 2.81534225e-01
-7.30605960e-01 1.12735736e+00 1.49038553e-01 1.46352082e-01
7.70747483e-01 -1.18230987e+00 6.24651432e-01 1.26467681e+00
-2.63968080e-01 2.42439017e-01 9.05461550e-01 -4.04826909e-01
-1.32166791e+00 -1.34683335e+00 6.53245986e-01 -5.10796249e-01
7.88856864e-01 -3.25644702e-01 -7.90817082e-01 4.15214747e-01
-1.18566103e-01 -8.41721147e-02 -7.00419620e-02 -1.41077951e-01
-3.84147197e-01 -4.55359608e-01 -7.95609474e-01 8.05962682e-01
1.09754276e+00 -4.00556564e-01 -4.05105501e-01 -2.13416629e-02
5.11333644e-01 -6.53983533e-01 -2.54257441e-01 7.11696446e-01
7.45141864e-01 -1.17630756e+00 9.11880553e-01 -3.22222471e-01
1.11733891e-01 -7.78440893e-01 -1.03505857e-01 -9.28198457e-01
-2.46828377e-01 -3.37859899e-01 2.07151368e-01 8.39194655e-01
6.69003487e-01 -1.05254853e+00 7.52612710e-01 1.82975769e-01
-6.88003123e-01 -9.42193568e-01 -9.51793075e-01 -9.19177115e-01
-1.04027562e-01 -7.37376451e-01 4.00395095e-01 3.32496524e-01
-4.00951266e-01 3.37687910e-01 -4.54696655e-01 6.21764004e-01
7.40792096e-01 2.33965423e-02 1.29914212e+00 -1.14217198e+00
3.38898182e-01 -6.41906083e-01 -8.74117792e-01 -1.49979007e+00
2.91672021e-01 -8.39735925e-01 5.46136260e-01 -1.37757373e+00
-2.13187605e-01 -5.36843896e-01 -3.08182478e-01 1.87059462e-01
-2.65122741e-01 7.73929292e-03 9.97334495e-02 2.85686851e-01
-7.11647213e-01 3.79496783e-01 1.18034399e+00 -5.37037253e-02
-1.39159374e-02 1.31788224e-01 -5.04961908e-01 7.85650015e-01
7.85032570e-01 -6.96360879e-03 -3.47792238e-01 4.08229977e-02
-6.13734350e-02 -9.61177144e-03 4.39363927e-01 -1.43164062e+00
4.23645079e-01 -1.77949034e-02 1.13518406e-02 -1.18945575e+00
2.39898652e-01 -6.28471911e-01 -2.64343411e-01 4.54459846e-01
-4.05278534e-01 1.11797787e-01 3.29962045e-01 8.53565574e-01
-1.99520975e-01 1.39552757e-01 9.38689411e-01 6.53656662e-01
-1.64524710e+00 -3.01371608e-02 -2.42560968e-01 4.87532765e-02
1.31216407e+00 -5.78516483e-01 -7.94288099e-01 -4.46721494e-01
-1.54255807e-01 6.40884280e-01 2.46296495e-01 5.84827900e-01
6.63417339e-01 -1.24330449e+00 -7.37345517e-01 4.03927177e-01
6.78280771e-01 -3.63262504e-01 -1.05043128e-01 1.07826495e+00
-7.44710267e-01 9.41784501e-01 -1.29066110e-01 -8.36647272e-01
-9.39983726e-01 3.14836323e-01 4.36682105e-01 -2.81586591e-02
-6.77959979e-01 6.04782462e-01 4.44692165e-01 -1.81873724e-01
2.86253512e-01 -6.05098903e-01 -2.74348497e-01 -2.13316679e-01
1.93425462e-01 5.51692963e-01 2.00145721e-01 -9.44569528e-01
-5.01021147e-01 5.35476804e-01 -8.54923353e-02 1.29902855e-01
5.40290833e-01 -1.56839371e-01 6.55216157e-01 4.22927320e-01
1.01874137e+00 2.93637902e-01 -1.64018667e+00 1.82488397e-01
1.48958102e-01 -5.53009629e-01 1.03943296e-01 -7.67465413e-01
-6.77103221e-01 9.65601087e-01 9.57660735e-01 -5.64044304e-02
5.39560378e-01 -3.54562461e-01 7.79544771e-01 4.32470024e-01
5.25040209e-01 -9.02076602e-01 -5.52456915e-01 7.13747263e-01
7.90308774e-01 -1.59597671e+00 -1.93899870e-01 -7.13952303e-01
-1.01204801e+00 9.54987347e-01 7.65874922e-01 -3.30568075e-01
4.67420876e-01 1.79806083e-01 3.33260834e-01 1.29056321e-02
-7.16568053e-01 -5.09484589e-01 6.47375524e-01 3.40971380e-01
7.17701688e-02 4.30581212e-01 -4.50688183e-01 3.07419747e-01
-3.53074193e-01 -4.90701407e-01 4.19117272e-01 4.01538223e-01
-9.52105224e-01 -5.47507763e-01 -2.92098403e-01 5.38211942e-01
2.39915654e-01 1.61991447e-01 -6.91980645e-02 1.28745317e+00
-8.67835060e-03 7.96111107e-01 1.51161522e-01 -8.30900252e-01
7.20534027e-01 -2.66647693e-02 -2.36000359e-01 -9.63632539e-02
1.75982311e-01 -3.03267509e-01 5.14938414e-01 -6.91443264e-01
4.47952330e-01 -6.11586213e-01 -1.29178011e+00 -3.68285924e-01
-3.52492362e-01 9.64161288e-03 5.38733542e-01 7.07912624e-01
4.71999586e-01 4.04811949e-01 8.78265500e-01 -5.69763303e-01
-6.44197822e-01 -5.73877037e-01 -3.19946289e-01 3.96608599e-02
5.71052670e-01 -1.05208242e+00 -4.46206301e-01 -5.74924409e-01] | [7.914958953857422, -1.5462111234664917] |
22c9fb68-29c6-4ce2-b42f-dd8b7221491d | document-embedding-for-scientific-articles | 2107.05151 | null | https://arxiv.org/abs/2107.05151v1 | https://arxiv.org/pdf/2107.05151v1.pdf | Document Embedding for Scientific Articles: Efficacy of Word Embeddings vs TFIDF | Over the last few years, neural network derived word embeddings became popular in the natural language processing literature. Studies conducted have mostly focused on the quality and application of word embeddings trained on public available corpuses such as Wikipedia or other news and social media sources. However, these studies are limited to generic text and thus lack technical and scientific nuances such as domain specific vocabulary, abbreviations, or scientific formulas which are commonly used in academic context. This research focuses on the performance of word embeddings applied to a large scale academic corpus. More specifically, we compare quality and efficiency of trained word embeddings to TFIDF representations in modeling content of scientific articles. We use a word2vec skip-gram model trained on titles and abstracts of about 70 million scientific articles. Furthermore, we have developed a benchmark to evaluate content models in a scientific context. The benchmark is based on a categorization task that matches articles to journals for about 1.3 million articles published in 2017. Our results show that content models based on word embeddings are better for titles (short text) while TFIDF works better for abstracts (longer text). However, the slight improvement of TFIDF for larger text comes at the expense of 3.7 times more memory requirement as well as up to 184 times higher computation times which may make it inefficient for online applications. In addition, we have created a 2-dimensional visualization of the journals modeled via embeddings to qualitatively inspect embedding model. This graph shows useful insights and can be used to find competitive journals or gaps to propose new journals. | ['R. Karimi', 'J. Truong', 'H. J. Meijer'] | 2021-07-11 | null | null | null | null | ['document-embedding'] | ['methodology'] | [-6.14968896e-01 -6.38725236e-02 -5.23775041e-01 6.35739416e-02
-5.25412858e-01 -5.19387424e-01 8.23318243e-01 8.40986550e-01
-9.03042018e-01 4.35513645e-01 6.32491350e-01 -6.13354027e-01
-1.94327682e-01 -9.61232305e-01 -4.00388658e-01 -2.42802471e-01
-2.95291822e-02 1.52225986e-01 -2.47188479e-01 -5.42240404e-02
6.36698782e-01 4.97880250e-01 -1.36342955e+00 -3.66069913e-01
8.40701282e-01 6.81497574e-01 3.91547233e-01 2.77425051e-01
-8.07679236e-01 1.67176023e-03 -8.92719328e-01 -3.68101150e-01
-2.33003050e-02 1.11737147e-01 -4.52952206e-01 -4.89544332e-01
4.40207660e-01 1.29331544e-01 -7.42457211e-01 1.12570846e+00
4.43350881e-01 2.54781395e-01 7.95335829e-01 -7.54473269e-01
-1.61916625e+00 6.58494115e-01 -4.70006168e-01 4.93721783e-01
1.72529653e-01 -2.19066143e-01 1.30295479e+00 -1.11320460e+00
8.05126667e-01 1.28440845e+00 3.19382280e-01 2.44715810e-01
-8.53191614e-01 -5.47367632e-01 9.16673839e-02 2.99575597e-01
-1.30599976e+00 1.00491770e-01 4.40374374e-01 -5.72489619e-01
9.57536340e-01 1.13716468e-01 5.34641266e-01 1.28989232e+00
6.65414691e-01 2.46208489e-01 7.45116651e-01 -5.09409845e-01
8.66918117e-02 4.57247734e-01 7.10313797e-01 3.48204643e-01
9.51400816e-01 -4.07748550e-01 -2.93425858e-01 -1.66723326e-01
6.06885791e-01 3.00035626e-01 -2.53815770e-01 1.47920966e-01
-1.46635735e+00 1.28977740e+00 2.74388582e-01 6.78424597e-01
-2.20417649e-01 -6.26737997e-02 5.73927701e-01 2.10516348e-01
7.19379306e-01 1.14096379e+00 -2.79294938e-01 -2.15390205e-01
-7.70541608e-01 3.80159885e-01 7.23481536e-01 7.21958280e-01
3.77125263e-01 1.60662502e-01 -2.76277274e-01 1.09698427e+00
3.39360297e-01 5.41786253e-01 8.44510376e-01 -6.44186497e-01
1.47968337e-01 6.45414114e-01 -5.56490570e-02 -1.44002295e+00
-2.62679696e-01 -6.89022481e-01 -6.06963277e-01 -3.81295323e-01
3.15178096e-01 3.53915319e-02 -7.49842167e-01 1.28802454e+00
-1.86244324e-01 -7.44771287e-02 5.87010151e-03 7.92067468e-01
1.23132408e+00 1.13838780e+00 2.18434721e-01 7.22726583e-02
1.88844991e+00 -9.23182011e-01 -9.53154445e-01 -1.64482042e-01
8.26070428e-01 -9.95923877e-01 1.10788965e+00 9.87784639e-02
-8.06022942e-01 -6.38798773e-01 -8.93079221e-01 -2.76763678e-01
-1.14481807e+00 9.21426564e-02 5.61352849e-01 4.71920431e-01
-8.99050534e-01 5.19827843e-01 -4.54172283e-01 -6.40198648e-01
2.61881173e-01 -9.87352878e-02 -3.72522682e-01 -2.21740544e-01
-1.41378343e+00 1.26404619e+00 7.70748854e-02 -5.63670158e-01
-3.46098930e-01 -1.21663761e+00 -9.65462208e-01 3.47845048e-01
-1.02913894e-01 -2.69152850e-01 7.80692458e-01 -2.86461055e-01
-1.00588799e+00 7.84048736e-01 -1.63674906e-01 -3.12376350e-01
-1.27693892e-01 -3.71092230e-01 -7.89772868e-01 1.55615574e-02
2.98848212e-01 3.95501018e-01 1.02296742e-02 -7.46762872e-01
-3.54412645e-01 -2.68093079e-01 1.63715146e-02 -1.81297749e-01
-1.29565156e+00 1.11093909e-01 -7.49976099e-01 -1.02169442e+00
-3.46514225e-01 -6.79753482e-01 -1.99439779e-01 3.44477631e-02
1.01509616e-01 -7.21539855e-01 6.38374329e-01 -6.86878979e-01
1.63992381e+00 -2.12886429e+00 1.07139602e-01 3.31555530e-02
3.24409306e-01 2.19058588e-01 -3.54355663e-01 6.02031648e-01
-4.25511375e-02 6.98342800e-01 3.54560047e-01 -6.73451200e-02
1.26185000e-01 2.11747885e-02 -3.76735061e-01 4.99655902e-01
-3.37414034e-02 7.17130184e-01 -9.01478291e-01 -4.52234030e-01
3.77704054e-02 5.52623093e-01 -4.17067587e-01 -4.17213850e-02
1.34349078e-01 -3.56035531e-01 -6.00876927e-01 6.40314400e-01
5.16836882e-01 -2.96063632e-01 -1.88119663e-03 -2.44351730e-01
-5.07759929e-01 3.73962581e-01 -6.08660460e-01 1.56122911e+00
-7.92568326e-01 1.11999023e+00 -4.16946918e-01 -1.11847591e+00
1.07343185e+00 1.73546746e-01 3.91844749e-01 -6.05924666e-01
3.37300688e-01 1.12045102e-01 2.46937379e-01 -4.61203754e-01
8.35785925e-01 2.13158965e-01 -1.57665357e-01 4.74704742e-01
2.63182372e-01 9.63707492e-02 4.91809040e-01 4.02491093e-01
8.62811625e-01 -2.38594130e-01 6.85033873e-02 -7.56508887e-01
2.34599486e-01 4.60238988e-03 3.32324564e-01 6.31158054e-01
-2.04977244e-02 4.71188337e-01 5.22179425e-01 -4.22465295e-01
-1.27448857e+00 -8.22063088e-01 -6.54670119e-01 9.61427271e-01
-9.37891193e-03 -8.49782228e-01 -3.73454720e-01 -1.72652066e-01
3.96690875e-01 9.68109429e-01 -8.91353548e-01 -1.62511036e-01
-2.65958756e-01 -8.01093459e-01 3.54231745e-01 4.71450597e-01
-1.90987796e-01 -8.20103347e-01 -2.63884068e-01 2.31957123e-01
3.16957891e-01 -9.61717904e-01 -5.19130051e-01 3.35060097e-02
-8.25881183e-01 -8.24481308e-01 -1.18329453e+00 -9.59779501e-01
4.50493366e-01 2.77014434e-01 1.03855991e+00 -2.06408262e-01
-4.07604575e-01 4.47602600e-01 -5.22721410e-01 -7.41470993e-01
1.67044252e-02 2.10577220e-01 4.17931169e-01 -5.01105547e-01
9.83817995e-01 -8.11193064e-02 -4.82642502e-01 -1.30925149e-01
-9.58798528e-01 -5.02838433e-01 5.38953185e-01 8.34578633e-01
1.92519575e-01 -4.23058033e-01 7.67941892e-01 -6.04635000e-01
1.12842333e+00 -9.01141047e-01 -4.90665078e-01 1.89415544e-01
-1.16112804e+00 1.16780438e-01 6.73922718e-01 -6.31925583e-01
-5.88918149e-01 -9.49171424e-01 -4.62108552e-02 -4.06899333e-01
1.92386165e-01 1.04361105e+00 4.22573954e-01 1.60708115e-01
6.09618485e-01 1.39655069e-01 9.05465931e-02 -7.47548461e-01
2.59211570e-01 1.03700566e+00 4.33608145e-03 -7.22572029e-01
6.45851791e-01 -4.15754728e-02 -3.13497096e-01 -1.12168229e+00
-6.65683389e-01 -6.23132229e-01 -8.34256858e-02 5.77391945e-02
8.41857135e-01 -9.21282530e-01 -5.35870373e-01 -2.49352470e-01
-1.29964089e+00 4.27443534e-01 -5.35594225e-02 9.81904447e-01
2.57847279e-01 4.45729971e-01 -7.35288858e-01 -2.95397818e-01
-3.29467416e-01 -1.07466316e+00 6.09748244e-01 3.86621237e-01
-5.36124051e-01 -1.37576783e+00 1.59368426e-01 1.12522967e-01
6.84745431e-01 -1.38958082e-01 1.27101469e+00 -8.65299046e-01
1.90955531e-02 -3.96699488e-01 -4.45407718e-01 2.55134612e-01
2.21802130e-01 3.45335126e-01 -7.70651400e-01 -1.99335247e-01
-4.87745553e-01 1.72121730e-02 9.53839064e-01 4.98061985e-01
1.53759444e+00 -1.98540106e-01 -5.25972903e-01 4.12898451e-01
1.45191813e+00 3.07251662e-01 5.79663694e-01 9.21860158e-01
6.96232557e-01 6.28674209e-01 4.40624058e-01 4.01411086e-01
1.88680619e-01 4.22688246e-01 -4.48943004e-02 1.17482059e-01
5.60567304e-02 -9.02871788e-02 1.71917289e-01 1.35816085e+00
-9.81759354e-02 -3.71846914e-01 -1.21096277e+00 9.64406550e-01
-1.21670556e+00 -6.53733730e-01 -1.38142526e-01 2.14783573e+00
9.35500145e-01 1.71434298e-01 -2.70209700e-01 -2.36862287e-01
4.71657276e-01 2.88773507e-01 -1.16846435e-01 -9.27042425e-01
-8.35273415e-02 4.82600451e-01 7.13025570e-01 1.86157003e-01
-7.72374988e-01 7.53004730e-01 6.23229170e+00 9.75129545e-01
-1.17094171e+00 5.37266918e-02 4.28924859e-01 -1.68426596e-02
-7.42980063e-01 -1.49764389e-01 -8.74169528e-01 6.13496006e-01
1.36827874e+00 -8.85567188e-01 -1.79022983e-01 1.00453019e+00
2.01679692e-01 2.40051270e-01 -9.13430095e-01 9.84718680e-01
7.68565834e-02 -1.81717002e+00 4.23776090e-01 2.97866970e-01
6.03569269e-01 -1.77717134e-02 2.68288225e-01 6.38553977e-01
-4.83730063e-03 -1.28518534e+00 2.22183332e-01 4.51982111e-01
8.91398132e-01 -6.66148901e-01 9.62497771e-01 -1.73610479e-01
-7.83260822e-01 3.46092843e-02 -9.01213884e-01 -1.13506600e-01
-1.80891767e-01 8.25140297e-01 -3.64859760e-01 4.76091892e-01
8.51213872e-01 9.89707589e-01 -5.30722857e-01 8.97961855e-01
2.18064100e-01 7.07460999e-01 6.39228374e-02 -5.86014450e-01
5.04515827e-01 -3.67184728e-01 3.09210986e-01 1.54003692e+00
6.62063062e-01 -2.65683293e-01 -2.44014189e-01 9.20260251e-01
-5.23221791e-01 5.42164683e-01 -8.70739400e-01 -8.46178949e-01
6.13097072e-01 1.37525690e+00 -5.44951975e-01 -4.43854362e-01
-9.23180819e-01 4.25380617e-01 1.89787120e-01 4.11733240e-01
-6.82460666e-01 -8.32352996e-01 1.03323650e+00 2.09568858e-01
7.03151301e-02 -4.08491969e-01 -3.58441114e-01 -1.19565737e+00
-3.38527590e-01 -6.15853250e-01 6.64771572e-02 -4.79898095e-01
-1.61609936e+00 5.31512797e-01 -4.75859828e-02 -1.18282592e+00
2.61132121e-01 -1.06787074e+00 -7.16444373e-01 1.10499966e+00
-1.32646060e+00 -4.80749339e-01 3.03535559e-03 -1.26385629e-01
7.00989246e-01 -5.83228052e-01 1.16545713e+00 4.85092908e-01
-7.15181530e-01 6.98865056e-01 7.90274143e-01 2.69766688e-01
1.04562235e+00 -1.09806883e+00 2.21835434e-01 3.37131739e-01
5.13280369e-02 1.29449821e+00 7.41384685e-01 -5.26381552e-01
-1.57031155e+00 -9.19370592e-01 1.36254847e+00 -3.52314979e-01
1.12613654e+00 -1.96798891e-01 -1.09540689e+00 3.75205696e-01
6.14277661e-01 -1.92026049e-01 1.08392322e+00 5.11920154e-01
-3.91238868e-01 -3.82925905e-02 -7.85072327e-01 8.50211740e-01
4.82354522e-01 -3.20302159e-01 -8.70053172e-01 4.80122387e-01
8.90310049e-01 1.07661054e-01 -1.42635131e+00 1.11470334e-02
5.52860439e-01 -2.07246430e-02 1.02191353e+00 -7.58532763e-01
9.33936477e-01 1.14875644e-01 -3.97081226e-02 -1.50648618e+00
-6.42995596e-01 8.17817375e-02 1.06454916e-01 1.18757367e+00
4.28809911e-01 -6.97151303e-01 3.86579990e-01 5.22271216e-01
-9.63940844e-02 -1.03788304e+00 -5.91551542e-01 -8.74941051e-01
7.50701904e-01 -1.58281907e-01 3.09603900e-01 1.28288937e+00
2.77415603e-01 2.65001267e-01 1.44880682e-01 -3.59858900e-01
3.36098284e-01 1.41556755e-01 4.07036901e-01 -1.52760863e+00
3.08559984e-01 -8.05936575e-01 -4.91957784e-01 -7.20354974e-01
3.54420990e-01 -1.06401169e+00 -6.48479581e-01 -1.88785374e+00
1.61479145e-01 -2.80798763e-01 -5.46465635e-01 1.50963888e-01
-2.61856973e-01 1.12177255e-02 2.03854796e-02 2.46576250e-01
-4.50565182e-02 5.98891318e-01 1.17684913e+00 -4.44581568e-01
2.50566334e-01 -7.06848979e-01 -1.00335968e+00 5.05838513e-01
7.84055054e-01 -4.21372294e-01 -1.35390371e-01 -4.97410834e-01
1.65743187e-01 -3.73587608e-01 -8.83548260e-02 -6.67138755e-01
1.02494210e-01 -2.15522885e-01 4.77556705e-01 -7.44552240e-02
5.17466329e-02 -4.82513338e-01 -4.35507357e-01 2.08738849e-01
-4.78094399e-01 4.91094828e-01 5.64764619e-01 5.28627455e-01
-4.52807784e-01 -5.21787941e-01 3.69811982e-01 -1.50844734e-02
-6.09942198e-01 1.98776647e-01 -6.62925422e-01 2.07917057e-02
8.73217463e-01 -7.79595375e-02 -2.69548923e-01 -2.91711599e-01
-2.27248505e-01 1.27426907e-01 3.84701699e-01 8.84805143e-01
5.91743469e-01 -1.34386277e+00 -7.83073127e-01 -2.35955104e-01
4.06151384e-01 -5.31431854e-01 -1.67434160e-02 4.53712344e-01
-6.69551492e-01 8.75260413e-01 -2.62329072e-01 1.54823512e-02
-8.70838523e-01 6.27539158e-01 -2.76461244e-01 -9.31679904e-02
-5.08799970e-01 5.78695178e-01 1.11255139e-01 -1.96159974e-01
3.90685201e-01 -4.37394917e-01 -7.08940983e-01 4.68463004e-01
4.75174308e-01 3.46590251e-01 -1.64689690e-01 -4.26822990e-01
-3.99797857e-01 6.47977889e-01 -2.21756771e-01 5.79742752e-02
1.57275987e+00 2.81895489e-01 -1.49646819e-01 6.54698670e-01
1.55058038e+00 3.44279021e-01 -1.42619193e-01 -2.16037050e-01
7.67183900e-02 -4.81285512e-01 4.64347869e-01 -3.86421472e-01
-1.02300942e+00 9.37651575e-01 5.10091960e-01 1.34362713e-01
3.46543163e-01 -1.72806188e-01 7.96323717e-01 3.36023152e-01
-8.64892080e-02 -1.25364125e+00 -2.31426150e-01 5.67459404e-01
7.96923816e-01 -1.06049395e+00 2.57052571e-01 1.24600112e-01
-3.78692448e-01 1.60859561e+00 4.62133259e-01 -1.71430662e-01
9.28443313e-01 5.11035398e-02 6.22205138e-02 -2.41856501e-01
-5.30294716e-01 1.76246449e-01 5.98167658e-01 2.32750341e-01
1.08766389e+00 6.42743753e-03 -1.09052682e+00 7.73030221e-01
-1.51146561e-01 -1.91513613e-01 7.54584670e-01 6.60859883e-01
-3.90111625e-01 -1.22852170e+00 -3.68547589e-01 8.76067400e-01
-9.04890418e-01 -3.48212838e-01 -3.42329554e-02 7.69921839e-01
-1.34654045e-01 7.14101195e-01 3.54290962e-01 -2.13393867e-01
1.60788596e-01 1.13497764e-01 6.81656003e-02 -7.59892583e-01
-4.06387687e-01 -1.42812520e-01 -1.40797481e-01 -3.04134772e-03
-1.92518085e-01 -4.09442395e-01 -9.97561753e-01 -5.91401160e-01
-1.48930028e-01 5.98647177e-01 9.39817727e-01 4.90244359e-01
5.49529612e-01 8.01876605e-01 1.93584368e-01 -5.06136954e-01
-3.94262820e-01 -1.36387181e+00 -6.30223930e-01 3.16613257e-01
-8.14779550e-02 -8.55883062e-01 -5.52133203e-01 -2.17745036e-01] | [10.185735702514648, 8.524608612060547] |
869e8a76-98b9-4ef5-8899-f3dff4c22576 | a-goal-driven-tree-structured-neural-model | null | null | https://www.ijcai.org/Proceedings/2019/736 | https://www.ijcai.org/Proceedings/2019/0736.pdf | A Goal-Driven Tree-Structured Neural Model for Math Word Problems | Most existing neural models for math word problems exploit Seq2Seq model to generate solution
expressions sequentially from left to right, whose
results are far from satisfactory due to the lack
of goal-driven mechanism commonly seen in human problem solving. This paper proposes a treestructured neural model to generate expression tree
in a goal-driven manner. Given a math word problem, the model first identifies and encodes its goal
to achieve, and then the goal gets decomposed into
sub-goals combined by an operator in a top-down
recursive way. The whole process is repeated until the goal is simple enough to be realized by a
known quantity as leaf node. During the process,
two-layer gated-feedforward networks are designed
to implement each step of goal decomposition, and
a recursive neural network is used to encode fulfilled subtrees into subtree embeddings, which provides a better representation of subtrees than the
simple goals of subtrees. Experimental results on
the dataset Math23K have shown that our treestructured model outperforms significantly several
state-of-the-art models. | ['Zhipeng Xie and Shichao Sun'] | 2019-08-10 | null | null | null | null | ['math-word-problem-solving', 'math-word-problem-solving', 'math-word-problem-solving'] | ['knowledge-base', 'reasoning', 'time-series'] | [ 3.38315725e-01 4.83717471e-01 1.36560783e-01 -5.86928368e-01
-5.12407303e-01 -5.48596978e-01 1.45178184e-01 1.73603699e-01
-1.86110720e-01 6.46379113e-01 4.60628539e-01 -4.79369462e-01
-7.36706480e-02 -1.35848749e+00 -5.42957783e-01 -5.46363771e-01
-2.95594316e-02 4.75439548e-01 -4.07427326e-02 -4.15666401e-01
3.05278659e-01 1.39993414e-01 -1.41134691e+00 5.75128973e-01
8.36644888e-01 6.88188314e-01 2.96451926e-01 6.79755151e-01
-8.36217225e-01 1.09955037e+00 -4.51277673e-01 -3.65593433e-01
1.21928044e-01 -7.70667255e-01 -1.16084146e+00 -5.83113730e-02
1.46384597e-01 -2.04703227e-01 -3.43910977e-02 1.07203066e+00
2.99098432e-01 4.00669813e-01 5.65006077e-01 -8.45791757e-01
-8.02172482e-01 1.13863075e+00 -1.16889335e-01 -2.50311792e-01
3.77587318e-01 6.63544163e-02 1.42823672e+00 -5.21920979e-01
4.37090784e-01 1.34009814e+00 3.42682481e-01 8.33873868e-01
-1.20084095e+00 -5.61433792e-01 3.90454859e-01 2.30427012e-02
-1.17501688e+00 -7.85773695e-02 8.17751825e-01 -1.39400780e-01
1.33295727e+00 1.95853561e-01 7.80366182e-01 6.80057526e-01
2.46176705e-01 6.84045672e-01 7.67391562e-01 -4.32758629e-01
3.89749259e-01 -2.57091463e-01 3.18949461e-01 9.99980867e-01
-1.46085277e-01 -1.17158972e-01 -2.97271401e-01 7.44975731e-02
8.81292880e-01 -1.51802480e-01 -1.02846079e-01 -1.09210879e-01
-7.13519394e-01 9.92054939e-01 6.63585544e-01 5.01821280e-01
-5.51303089e-01 4.27746356e-01 2.38329113e-01 5.08931100e-01
3.34262669e-01 7.98410237e-01 -4.66812998e-01 -1.34988263e-01
-9.34476733e-01 5.92675507e-01 7.75763571e-01 7.41546035e-01
7.47084618e-01 2.49056175e-01 -4.93397623e-01 7.44183838e-01
1.03712365e-01 -4.83104140e-02 4.99125212e-01 -8.90580833e-01
5.23017108e-01 1.14570570e+00 -1.27788678e-01 -8.75522971e-01
-4.28749204e-01 -4.30001318e-01 -7.48712838e-01 1.63302839e-01
5.24059415e-01 -2.24368006e-01 -1.13991451e+00 1.98587990e+00
1.88703805e-01 1.76070437e-01 2.53948033e-01 8.74280572e-01
1.02348733e+00 1.16950512e+00 2.58395433e-01 -3.40866111e-02
1.20771956e+00 -9.63308454e-01 -5.31924725e-01 -2.20283359e-01
1.05844247e+00 -1.38432696e-01 1.14742398e+00 3.99368644e-01
-1.36717463e+00 -5.76442182e-01 -6.74160838e-01 -2.96390623e-01
-3.31694275e-01 -2.35633045e-01 8.25499594e-01 6.73920870e-01
-1.21322393e+00 6.34708881e-01 -3.70391518e-01 -9.69586745e-02
3.68311465e-01 2.27024242e-01 -1.54134795e-01 -2.06005663e-01
-1.27842462e+00 6.38303459e-01 6.67881012e-01 1.00148521e-01
-1.03294396e+00 -8.50403130e-01 -1.31071031e+00 5.30856848e-01
3.97064805e-01 -7.58466721e-01 1.23221016e+00 -7.85762489e-01
-1.61016583e+00 8.33669782e-01 -2.40376517e-01 -4.94663775e-01
3.11701605e-03 5.81205636e-02 4.43422049e-02 -1.91156819e-01
-8.82629380e-02 7.89654613e-01 7.63275266e-01 -9.00005937e-01
-5.52092493e-01 -2.86383092e-01 5.20380318e-01 2.86057711e-01
-3.15060228e-01 7.69087533e-03 -3.47853184e-01 -4.46100324e-01
1.47203192e-01 -4.32131618e-01 -6.92813277e-01 -5.07444024e-01
-3.11029315e-01 -8.21209908e-01 1.14517182e-01 -4.51122463e-01
1.75152004e+00 -1.88826597e+00 6.40527368e-01 8.56150240e-02
3.32645178e-01 1.87642679e-01 -5.40354848e-01 4.60976481e-01
-3.78904790e-01 2.60743022e-01 -2.99175382e-01 -1.71730429e-01
1.18355565e-01 3.02222759e-01 -2.91409343e-01 -1.08273491e-01
3.88385206e-01 1.15521133e+00 -1.09256744e+00 -2.86434710e-01
1.06249556e-01 -7.84180760e-02 -9.46763039e-01 6.23697758e-01
-8.15798402e-01 -1.19902536e-01 -6.25241637e-01 3.91148537e-01
5.54597557e-01 -1.77997425e-01 3.72023731e-01 2.08243012e-01
-5.42114675e-02 7.28927612e-01 -1.18551207e+00 1.69602799e+00
-6.88740134e-01 1.15375437e-01 -2.35315353e-01 -1.23745799e+00
1.14339173e+00 1.10730492e-01 8.98347516e-03 -5.70417941e-01
1.60195529e-01 -3.20118815e-02 1.26133174e-01 -5.30182898e-01
4.43473995e-01 -4.68558431e-01 -3.88124347e-01 4.17788029e-01
4.83376272e-02 -4.68314111e-01 4.32491720e-01 1.79752588e-01
1.19067132e+00 3.68537933e-01 2.56931275e-01 -1.31039500e-01
7.65146077e-01 -1.37800155e-02 6.18923962e-01 7.67815828e-01
2.61932552e-01 4.94261682e-01 8.59180510e-01 -7.04795599e-01
-7.14450836e-01 -8.86232972e-01 5.78395426e-01 1.20194125e+00
-2.41082549e-01 -6.98733330e-01 -9.37768519e-01 -6.79298341e-01
-2.21110508e-01 1.16525006e+00 -6.17566168e-01 -2.37985507e-01
-8.63479555e-01 -3.60802770e-01 5.51665783e-01 4.93934840e-01
5.28542817e-01 -1.65433562e+00 -7.91495502e-01 6.04163945e-01
-3.63458470e-02 -7.59846091e-01 -1.33730710e-01 5.08039653e-01
-9.62699175e-01 -8.41578722e-01 -5.50492525e-01 -1.08493936e+00
7.59774625e-01 -1.58741713e-01 1.11454749e+00 4.70172495e-01
-1.98804244e-01 -2.16365546e-01 -2.82441765e-01 -2.11117148e-01
-2.55020767e-01 9.45064873e-02 -3.86450887e-01 -1.28409922e-01
4.20827121e-01 -6.33180618e-01 -9.20994803e-02 -3.88282329e-01
-7.25794733e-01 2.07852304e-01 4.02087480e-01 8.18632305e-01
2.92969733e-01 4.05303031e-01 5.75552821e-01 -8.80713522e-01
1.01509190e+00 -4.37572122e-01 -6.98189199e-01 2.59153843e-01
-1.53761283e-01 4.19175208e-01 1.17166305e+00 -1.80470064e-01
-1.18951511e+00 6.38408528e-04 -3.07603359e-01 -1.12963788e-01
-4.09322113e-01 8.20186317e-01 -2.00334370e-01 3.55504155e-01
5.60990393e-01 4.82818812e-01 -2.15585008e-01 -4.00347620e-01
6.08525991e-01 2.41613969e-01 3.52107137e-01 -9.63717818e-01
6.51324272e-01 -2.19947845e-01 4.64719273e-02 -7.14506328e-01
-1.00465333e+00 -3.16920802e-02 -2.58882940e-01 -6.88553527e-02
1.02910936e+00 -4.66362506e-01 -7.30738521e-01 6.17844701e-01
-1.36824548e+00 -6.66870356e-01 -4.78793919e-01 5.04875034e-02
-5.94590962e-01 1.23587035e-01 -7.40123570e-01 -1.06844902e+00
-2.88913995e-01 -9.17360425e-01 7.13386476e-01 5.49149275e-01
-3.59479755e-01 -9.12980556e-01 -2.08353624e-01 1.90018341e-01
3.81603062e-01 1.11198969e-01 1.64490616e+00 -6.87126279e-01
-6.55661106e-01 5.96357509e-02 -1.67615101e-01 3.13008934e-01
1.70080047e-02 -3.86993766e-01 -4.35302764e-01 2.64539242e-01
2.15261262e-02 -7.07643867e-01 7.12922335e-01 3.85614753e-01
1.51176524e+00 -3.35132509e-01 -1.54397395e-02 6.06563032e-01
1.40084410e+00 2.48165846e-01 8.08977127e-01 1.23181783e-01
3.33262146e-01 5.72501063e-01 4.14148867e-01 2.16098309e-01
5.63422263e-01 1.40989676e-01 5.80009043e-01 1.15492381e-01
5.10603860e-02 -4.63404238e-01 2.74553329e-01 5.36989331e-01
-6.01620898e-02 -4.29752946e-01 -8.82933557e-01 6.78177655e-01
-1.71155751e+00 -8.02575529e-01 -2.19689861e-01 1.89619672e+00
1.02064538e+00 2.70328134e-01 -1.49034157e-01 2.25883842e-01
3.23004872e-01 4.11920279e-01 -2.04659149e-01 -1.04325497e+00
3.55252057e-01 7.27680981e-01 -2.69796532e-02 6.25861764e-01
-6.57660007e-01 1.67684305e+00 6.40362978e+00 8.40041518e-01
-9.34040546e-01 -4.02877331e-01 5.39355397e-01 -3.94326076e-03
-5.08245707e-01 -8.47657472e-02 -7.77015567e-01 2.61307904e-03
8.71282279e-01 -2.80067295e-01 6.75813437e-01 6.70644224e-01
1.24735720e-01 7.34316930e-02 -1.25970256e+00 6.91709518e-01
-1.73920631e-01 -1.30554771e+00 3.67539942e-01 -3.48596990e-01
4.22069639e-01 -5.79408348e-01 -1.37809485e-01 8.84618759e-01
7.45840549e-01 -1.39578259e+00 4.65887696e-01 3.73714864e-01
3.65147978e-01 -9.79334414e-01 5.20154178e-01 7.77443051e-01
-1.15494275e+00 -2.92552769e-01 -5.37514031e-01 -7.69717157e-01
2.16771960e-01 3.75135809e-01 -6.75644934e-01 6.21628463e-01
4.25543725e-01 4.35129881e-01 -1.87837258e-01 5.47281504e-01
-8.13172698e-01 6.57157779e-01 -2.04691470e-01 -6.09214365e-01
8.16516936e-01 -4.62216198e-01 2.91386604e-01 1.07949829e+00
3.85197550e-01 6.29814744e-01 1.19890295e-01 1.33102691e+00
-8.25748965e-02 1.79598898e-01 -5.57923496e-01 -3.64977539e-01
1.88113987e-01 1.22176325e+00 -5.67718267e-01 -3.66398066e-01
-1.93874940e-01 9.37951565e-01 9.57245588e-01 3.96907091e-01
-7.98574150e-01 -5.32398880e-01 6.21747613e-01 -4.95041236e-02
3.55482042e-01 -3.22410047e-01 -3.98239821e-01 -9.59372759e-01
-1.71597600e-01 -9.49460864e-01 5.33488333e-01 -9.54500318e-01
-9.86122072e-01 4.58493531e-01 5.45498244e-02 -3.99662435e-01
-3.40639830e-01 -7.19110966e-01 -9.28126156e-01 1.35377347e+00
-1.30170202e+00 -9.94951546e-01 -8.05969313e-02 3.57401699e-01
7.40808010e-01 -3.39129046e-02 1.20376241e+00 -2.01691493e-01
-4.31466639e-01 4.15855467e-01 -6.63501084e-01 1.89778566e-01
-1.04524978e-01 -1.29523993e+00 5.36064684e-01 8.47218513e-01
8.66077095e-03 8.96100998e-01 5.38444638e-01 -6.28320456e-01
-1.22392547e+00 -1.05821371e+00 1.46916270e+00 -5.38724251e-02
5.21682322e-01 -4.58964020e-01 -9.56644177e-01 5.96819520e-01
2.54011393e-01 -3.69106680e-01 6.40594602e-01 1.94541425e-01
-2.86173135e-01 1.40331432e-01 -1.01467979e+00 9.44238305e-01
1.25105071e+00 -1.57665163e-02 -1.22049654e+00 1.06978439e-01
9.88973796e-01 -5.32778561e-01 -4.90664542e-01 2.17440799e-01
1.84624270e-01 -8.39716077e-01 7.36523509e-01 -1.45989203e+00
1.09367740e+00 -2.39362136e-01 -1.31730109e-01 -1.55967617e+00
-7.17910051e-01 -5.85191369e-01 2.81496830e-02 1.14776611e+00
5.05144954e-01 -4.57682639e-01 9.84136701e-01 5.76679885e-01
-2.51049489e-01 -1.11701131e+00 -7.58727610e-01 -6.24987125e-01
3.87813658e-01 -5.31439364e-01 8.58683288e-01 7.80573070e-01
1.83078095e-01 5.46230257e-01 -2.57054597e-01 -4.92337160e-02
3.32277238e-01 4.65976626e-01 5.72204471e-01 -8.84566367e-01
-2.46500775e-01 -5.85704088e-01 -9.86780599e-02 -1.40440190e+00
3.60639066e-01 -1.21499228e+00 4.76027317e-02 -1.94754481e+00
-9.64287966e-02 -4.30819094e-01 -2.79588312e-01 7.14635670e-01
-3.90736252e-01 -2.37004146e-01 2.31157750e-01 -7.35191941e-01
-3.51325005e-01 5.44927716e-01 1.27355349e+00 -9.11219195e-02
-2.80980527e-01 -3.29476416e-01 -1.07989073e+00 7.43358433e-01
8.55986953e-01 -5.17436981e-01 -5.51564395e-01 -5.28407991e-01
3.42393309e-01 3.30922216e-01 -3.43480594e-02 -7.95279384e-01
1.95131361e-01 -5.31388581e-01 2.66943127e-01 -3.74568164e-01
2.07035601e-01 -6.83422089e-01 -2.23380864e-01 5.79657793e-01
-6.43296063e-01 -4.29550149e-02 1.84294343e-01 7.97823146e-02
-3.49095538e-02 -6.90154493e-01 6.04297161e-01 -4.53177482e-01
-5.82359195e-01 2.23767251e-01 -5.67811906e-01 1.87474087e-01
8.96745205e-01 -1.76308274e-01 1.88124776e-01 -4.13353086e-01
-7.36942828e-01 5.79980195e-01 -1.20409191e-01 4.96369809e-01
6.84929252e-01 -1.32874882e+00 -7.63167799e-01 3.13039541e-01
-1.33243918e-01 4.17114735e-01 1.93106741e-01 4.54714559e-02
-4.04621631e-01 3.32959175e-01 -4.22960408e-02 -1.90121368e-01
-1.24628854e+00 6.34477377e-01 3.96010250e-01 -8.17365229e-01
-5.34501255e-01 1.31633210e+00 3.11003536e-01 -6.76637173e-01
3.36123645e-01 -6.94416225e-01 -4.07864898e-01 -9.48584005e-02
6.34713352e-01 -1.44372582e-02 -1.72477052e-01 -1.65744543e-01
-1.28652230e-01 5.50931513e-01 -3.88017222e-02 1.07230239e-01
1.56139565e+00 3.59392047e-01 -3.79450738e-01 1.53921425e-01
9.09096956e-01 -3.66627783e-01 -6.20061398e-01 -1.27961352e-01
4.92360815e-03 -2.67103285e-01 -1.47459045e-01 -7.51272678e-01
-9.74368513e-01 1.08319032e+00 -2.09558442e-01 2.28234962e-01
1.25447583e+00 -1.58124954e-01 8.17947924e-01 4.96198982e-01
2.60988653e-01 -9.86049235e-01 2.29929820e-01 1.10945928e+00
9.93176043e-01 -5.09118795e-01 -5.01723886e-01 -5.00038743e-01
-5.10922313e-01 1.30237341e+00 8.70767355e-01 -3.27095687e-01
1.84434071e-01 3.68788898e-01 -1.89626992e-01 -2.35255152e-01
-9.82984722e-01 -2.09443420e-01 3.02145425e-02 4.69673276e-01
6.00488961e-01 3.65000516e-02 -5.19875109e-01 1.10163593e+00
-5.88875473e-01 2.91154206e-01 3.96624476e-01 9.23934639e-01
-6.74952447e-01 -1.24070334e+00 -2.21168593e-01 3.53144199e-01
-4.57651228e-01 -4.11655813e-01 -3.80699158e-01 5.41278362e-01
1.45096704e-01 8.27168524e-01 1.16252728e-01 -1.91109300e-01
4.30793762e-01 5.24998844e-01 7.31220722e-01 -1.15613878e+00
-9.79143500e-01 -3.41523081e-01 3.45829964e-01 -6.25331879e-01
2.26995170e-01 -2.79301882e-01 -1.81391346e+00 -1.78878546e-01
4.75403629e-02 2.22669289e-01 2.43291557e-01 8.81619155e-01
1.04426797e-02 9.71531332e-01 4.05643404e-01 -4.99076962e-01
-7.33431160e-01 -7.08616436e-01 -3.70498836e-01 4.42079037e-01
-8.53767395e-02 -2.27447748e-01 -1.78412691e-01 -1.11963682e-01] | [9.807136535644531, 7.473447799682617] |
21e1fc3a-5b18-4eb4-a95c-fa3dca438932 | vasr-visual-analogies-of-situation | 2212.04542 | null | https://arxiv.org/abs/2212.04542v1 | https://arxiv.org/pdf/2212.04542v1.pdf | VASR: Visual Analogies of Situation Recognition | A core process in human cognition is analogical mapping: the ability to identify a similar relational structure between different situations. We introduce a novel task, Visual Analogies of Situation Recognition, adapting the classical word-analogy task into the visual domain. Given a triplet of images, the task is to select an image candidate B' that completes the analogy (A to A' is like B to what?). Unlike previous work on visual analogy that focused on simple image transformations, we tackle complex analogies requiring understanding of scenes. We leverage situation recognition annotations and the CLIP model to generate a large set of 500k candidate analogies. Crowdsourced annotations for a sample of the data indicate that humans agree with the dataset label ~80% of the time (chance level 25%). Furthermore, we use human annotations to create a gold-standard dataset of 3,820 validated analogies. Our experiments demonstrate that state-of-the-art models do well when distractors are chosen randomly (~86%), but struggle with carefully chosen distractors (~53%, compared to 90% human accuracy). We hope our dataset will encourage the development of new analogy-making models. Website: https://vasr-dataset.github.io/ | ['Gabriel Stanovsky', 'Roy Schwartz', 'Dafna Shahaf', 'Eli Strugo', 'Ron Yosef', 'Yonatan Bitton'] | 2022-12-08 | null | null | null | null | ['visual-reasoning', 'visual-analogies', 'common-sense-reasoning', 'visual-commonsense-reasoning', 'visual-reasoning'] | ['computer-vision', 'computer-vision', 'reasoning', 'reasoning', 'reasoning'] | [ 2.41695672e-01 -5.91035001e-02 4.68162932e-02 -4.87777531e-01
-6.00016594e-01 -9.80572343e-01 9.02563393e-01 3.79455000e-01
-6.89931273e-01 4.83084410e-01 5.09233117e-01 -3.34321320e-01
2.12707669e-01 -2.56142616e-01 -8.67018402e-01 2.20364183e-01
2.96478212e-01 5.31817317e-01 1.52028859e-01 -4.42437649e-01
5.73173583e-01 4.65564430e-01 -1.44852126e+00 9.98638868e-01
6.91330910e-01 9.09949005e-01 1.95500255e-01 5.29827356e-01
-3.99336740e-02 1.10894537e+00 -6.44930661e-01 -7.96761572e-01
5.56012034e-01 -6.39922321e-01 -1.14964294e+00 -2.89363354e-01
1.10052252e+00 -2.18250260e-01 -4.71343160e-01 1.09251463e+00
3.93027186e-01 3.99718046e-01 8.11367273e-01 -1.49703145e+00
-1.49606943e+00 3.14358145e-01 -4.40867275e-01 4.94710863e-01
1.15027082e+00 4.52057421e-01 1.01665080e+00 -1.14723670e+00
7.67580867e-01 1.40089691e+00 6.21661305e-01 5.65851390e-01
-1.64677668e+00 -7.65971363e-01 -1.21154234e-01 7.93721616e-01
-1.46153104e+00 -4.01631564e-01 4.41292852e-01 -5.00368953e-01
1.13628674e+00 5.13843954e-01 8.80676627e-01 1.41396248e+00
5.13049476e-02 4.68324929e-01 1.78573585e+00 -5.00162542e-01
2.09545344e-01 2.07885742e-01 4.97111157e-02 2.59497702e-01
-1.46350395e-02 2.56897092e-01 -5.94005287e-01 -1.03222191e-01
7.14994431e-01 -1.09496385e-01 -3.55147213e-01 -1.23850860e-01
-1.53059304e+00 5.30327499e-01 1.07258582e+00 2.32744411e-01
-3.39606464e-01 3.32270525e-02 1.10877402e-01 4.51719075e-01
-2.53427386e-01 1.23053980e+00 1.16560526e-01 2.69008368e-01
-5.03167510e-01 6.52564704e-01 4.78886873e-01 1.15809429e+00
7.36052752e-01 -3.81762564e-01 -4.26481515e-01 6.91895068e-01
-1.76895767e-01 4.34691012e-01 5.18163264e-01 -1.43901205e+00
3.58771086e-01 4.91785318e-01 3.54253024e-01 -1.07831633e+00
-3.43450457e-01 4.55756299e-02 -4.76061583e-01 3.23792100e-01
7.11900055e-01 4.95602340e-01 -8.13431740e-01 1.79187214e+00
-1.51613848e-02 1.67007089e-01 4.65855785e-02 1.14337039e+00
9.20934141e-01 5.71253955e-01 4.58747894e-01 1.66234434e-01
1.65388060e+00 -8.16819310e-01 -3.67911875e-01 -6.63378000e-01
2.82142878e-01 -1.04621542e+00 1.94265342e+00 1.36039987e-01
-1.02973366e+00 -9.46201205e-01 -9.84773636e-01 -3.80217493e-01
-5.68052769e-01 -2.44671777e-01 5.45428634e-01 1.49132043e-01
-1.26922441e+00 4.94014293e-01 8.44856054e-02 -6.75638258e-01
5.05683601e-01 -1.38288632e-01 -7.30274320e-01 -3.83853137e-01
-1.25143623e+00 1.61158717e+00 3.34812909e-01 -3.16293597e-01
-7.67698765e-01 -8.56454909e-01 -9.06509638e-01 2.71658110e-03
9.07129198e-02 -8.91522646e-01 1.46753323e+00 -1.25308895e+00
-8.00444424e-01 1.46072268e+00 -1.80961013e-01 -4.18246418e-01
4.47296172e-01 -2.61525691e-01 -5.88396490e-01 1.71017230e-01
5.81542850e-01 1.18133116e+00 4.63803381e-01 -1.44273674e+00
-3.24711323e-01 -7.55281225e-02 4.05842394e-01 3.58202785e-01
-4.14219722e-02 1.63282901e-01 -1.40828475e-01 -7.46599317e-01
-1.20030738e-01 -1.12051249e+00 -1.51036516e-01 5.45183718e-01
-9.48296636e-02 -1.64711148e-01 1.32414833e-01 -6.77184343e-01
6.23786688e-01 -2.17734408e+00 -8.76156464e-02 3.92618887e-02
2.22439542e-01 -1.13340586e-01 -4.93168741e-01 5.42692006e-01
-6.01686239e-01 -3.01433727e-03 -5.29081635e-02 -1.69332162e-01
1.89428717e-01 2.78870645e-03 -7.35235631e-01 2.16912493e-01
7.93827996e-02 1.27608693e+00 -1.25248849e+00 -6.61883116e-01
4.74694610e-01 -4.04570773e-02 -4.25609440e-01 3.71079057e-01
-4.09462564e-02 4.58014935e-01 2.13256806e-01 3.18363100e-01
5.35353422e-01 -2.76484668e-01 -6.47599101e-02 -4.89896566e-01
2.72615343e-01 1.15390688e-01 -7.99228489e-01 1.90739918e+00
-3.93659979e-01 9.39484835e-01 -8.23045552e-01 -4.48697180e-01
7.57249713e-01 1.25038996e-01 -1.96391433e-01 -1.21199083e+00
6.41850308e-02 2.72848736e-02 2.87748724e-01 -4.81758118e-01
5.76967359e-01 -3.04153144e-01 -2.68873870e-01 4.10860509e-01
6.88510621e-03 -5.51579833e-01 5.66953495e-02 4.81244832e-01
1.00412607e+00 7.60457367e-02 8.29726934e-01 -2.63844490e-01
2.47136191e-01 4.34989363e-01 1.28100425e-01 9.99819994e-01
-4.31095392e-01 6.00380540e-01 2.82223910e-01 -7.26332068e-01
-1.31586945e+00 -1.60799241e+00 1.19404763e-01 1.05264497e+00
4.43591923e-01 -4.11914557e-01 -5.88209271e-01 -3.81920069e-01
-3.28047909e-02 1.34668612e+00 -1.05271184e+00 -2.61654258e-01
-3.61449122e-01 -1.19974285e-01 5.10979235e-01 7.89207280e-01
4.57229048e-01 -1.46847463e+00 -6.64369583e-01 -3.18144649e-01
-3.75609457e-01 -1.35752201e+00 -4.86162633e-01 -3.89242887e-01
-2.36013904e-01 -1.06435847e+00 -5.63796699e-01 -7.17561841e-01
6.79006159e-01 3.91737223e-01 1.56241810e+00 -8.86548981e-02
-4.41659123e-01 5.84854066e-01 -2.86786407e-01 -4.85627651e-01
-6.61642015e-01 -6.98224247e-01 1.23937026e-01 -3.05257112e-01
5.04272342e-01 -5.24363160e-01 -8.46894979e-01 5.03163874e-01
-5.56671619e-01 3.53189886e-01 5.67048132e-01 5.76797605e-01
5.81535995e-01 -8.57711732e-01 4.36379313e-01 -4.54640597e-01
8.50085795e-01 -3.52148473e-01 -3.82316291e-01 5.53894639e-01
-3.44072759e-01 -1.37491658e-01 6.37028039e-01 -8.81681323e-01
-9.71611142e-01 5.65990843e-02 4.35986549e-01 -8.41497004e-01
-3.49011689e-01 2.39487570e-02 2.83820242e-01 7.29913497e-03
1.36980939e+00 -2.54749432e-02 -3.60629439e-01 -3.10132490e-03
1.05695987e+00 3.57938945e-01 1.22701287e+00 -6.94221675e-01
6.56467974e-01 4.96932864e-01 -4.60377894e-02 -3.69670242e-01
-9.41689909e-01 -2.77052820e-01 -5.57173729e-01 -3.76555651e-01
1.12801349e+00 -9.26697731e-01 -9.17552888e-01 -1.29711986e-01
-1.44200969e+00 -5.21298409e-01 -5.90130806e-01 4.91535872e-01
-7.19914079e-01 3.07211846e-01 -2.73207456e-01 -4.28294867e-01
-8.71708766e-02 -8.23541462e-01 8.68878245e-01 1.75483197e-01
-1.12223160e+00 -6.30984187e-01 1.31876230e-01 4.58080679e-01
2.71873295e-01 3.12690437e-01 8.83802533e-01 -6.62221611e-01
-3.15761507e-01 -1.50827974e-01 -6.33159339e-01 -5.49429171e-02
9.57579389e-02 -4.37108785e-01 -1.11483073e+00 3.47503163e-02
-7.79184327e-02 -7.53057003e-01 4.72344279e-01 -4.41338401e-03
1.32378376e+00 -9.90927815e-02 -2.54423678e-01 2.77854741e-01
1.13578308e+00 1.47243142e-01 8.39926422e-01 3.25713217e-01
6.75516188e-01 7.44118273e-01 7.74137139e-01 9.44931582e-02
4.79427695e-01 8.17885816e-01 1.65448099e-01 -1.15838453e-01
-3.29337180e-01 -4.35892284e-01 1.87836178e-02 1.61571726e-02
-6.99427724e-02 -1.14530705e-01 -1.24940407e+00 5.37069380e-01
-1.74934578e+00 -1.25717139e+00 -9.68827903e-02 2.21989441e+00
8.51292193e-01 1.48080930e-01 -7.20051862e-03 -2.18568370e-01
6.99987829e-01 -7.84677193e-02 -4.34192091e-01 -5.08703709e-01
-1.59504935e-01 2.50431925e-01 1.20312966e-01 4.90870923e-01
-8.62625122e-01 1.11904931e+00 6.82869387e+00 7.19237208e-01
-7.56694317e-01 -2.47592255e-02 4.98198271e-01 -2.07116008e-01
-4.03487027e-01 2.27000713e-01 -8.93105417e-02 2.80930012e-01
5.93706846e-01 -3.28882605e-01 8.35633934e-01 6.47316694e-01
-5.56929270e-03 -1.16267815e-01 -1.67232513e+00 1.41982543e+00
5.01224697e-01 -1.33685422e+00 4.77686942e-01 -5.71742535e-01
6.52676225e-01 -2.12886393e-01 1.72030076e-01 3.05262715e-01
4.24759567e-01 -1.49468184e+00 1.01838386e+00 7.31788695e-01
1.06170666e+00 -3.35494876e-01 2.99975812e-01 4.29573134e-02
-1.04047108e+00 -7.82468095e-02 -5.21515727e-01 -4.98783916e-01
2.41084427e-01 -1.45836964e-01 -1.00028074e+00 6.29383475e-02
1.01153398e+00 5.57694793e-01 -1.12619233e+00 7.98301458e-01
-4.74660933e-01 -6.98167086e-02 -3.75558548e-02 4.85246144e-02
-1.17604202e-02 7.88482875e-02 3.30867708e-01 1.18786454e+00
1.04682833e-01 4.24669027e-01 -2.12676544e-02 1.22230399e+00
-1.70811176e-01 1.24155253e-01 -1.01685023e+00 3.30928206e-01
8.49013448e-01 1.08395648e+00 -5.70519865e-01 -5.84869206e-01
-4.79438245e-01 1.37529778e+00 7.74408937e-01 4.76433307e-01
-9.09156740e-01 -7.53305703e-02 4.64392483e-01 7.63356164e-02
-1.60472915e-01 1.02404848e-01 -4.40562129e-01 -8.65495324e-01
6.41646311e-02 -1.13773501e+00 6.04365408e-01 -1.84718215e+00
-1.84568417e+00 7.42194653e-01 2.80925900e-01 -1.43454313e+00
-1.50981948e-01 -8.12036335e-01 -8.41297448e-01 9.57837224e-01
-9.04583812e-01 -1.18855178e+00 -9.20933187e-01 7.90370584e-01
4.82468486e-01 -1.28696496e-02 9.59955096e-01 -2.11851761e-01
2.69405067e-01 6.33529305e-01 -5.72983146e-01 5.93596958e-02
1.28842139e+00 -1.31230700e+00 9.08131182e-01 5.32167852e-01
5.64221621e-01 7.97945917e-01 8.92026186e-01 -5.67421257e-01
-7.91275442e-01 -7.20515490e-01 7.37731397e-01 -1.20634747e+00
8.71829987e-01 -2.74452537e-01 -1.04257202e+00 9.15738404e-01
6.10741198e-01 2.46497840e-01 8.75600040e-01 -2.79522806e-01
-1.01526427e+00 2.20002547e-01 -1.10339987e+00 1.25660491e+00
1.43256903e+00 -9.90289509e-01 -1.28759050e+00 3.91717672e-01
7.22837329e-01 -4.52405095e-01 -7.40818083e-01 1.16018474e-01
6.99512303e-01 -9.56387162e-01 1.34651756e+00 -1.11402524e+00
7.58615971e-01 -4.01361912e-01 -4.87328678e-01 -1.44020593e+00
-5.80954254e-01 -2.93330014e-01 2.57806718e-01 8.30473363e-01
3.28232288e-01 -5.19133091e-01 1.21432118e-01 9.45924044e-01
6.77041784e-02 -3.06435078e-01 -9.49912131e-01 -9.54093874e-01
-1.79900229e-02 -3.43252212e-01 5.21488249e-01 1.26834106e+00
2.71006614e-01 5.32935321e-01 -1.11549921e-01 -3.57448161e-02
4.43541765e-01 2.28533104e-01 8.95864248e-01 -8.62096250e-01
-3.58801961e-01 -5.23513436e-01 -5.36233723e-01 -6.58078969e-01
3.37443620e-01 -1.14450240e+00 -1.89047620e-01 -1.45536458e+00
4.46084470e-01 -1.05673201e-01 -3.64380181e-01 4.95537192e-01
-4.16601449e-01 5.70946276e-01 6.39753819e-01 3.38644534e-01
-6.42848611e-01 3.78475547e-01 1.32206595e+00 -1.25556201e-01
1.16653413e-01 -6.73819363e-01 -9.31590736e-01 7.63565719e-01
8.06131661e-01 -3.10538352e-01 -5.40063381e-01 -5.00934780e-01
1.76835105e-01 -6.18214719e-02 1.12019515e+00 -1.07416558e+00
1.08464539e-01 -3.82836163e-01 9.01072323e-01 -1.08166881e-01
5.92843711e-01 -7.98546910e-01 2.73844451e-01 3.42582643e-01
-7.05918252e-01 7.50220895e-01 5.50403893e-01 5.03630042e-01
7.23613426e-02 -1.29480347e-01 7.08184242e-01 -1.75694883e-01
-1.20610034e+00 -2.15652198e-01 -1.13692544e-01 4.26626444e-01
1.18098938e+00 -2.16078311e-01 -7.48144507e-01 -6.67438686e-01
-6.91377997e-01 3.91752757e-02 8.56133759e-01 6.72336578e-01
7.42698669e-01 -1.63630867e+00 -7.78393865e-01 -4.32091236e-01
7.89324284e-01 -5.54260492e-01 2.07652420e-01 3.83096278e-01
-4.44872022e-01 2.85209026e-02 -7.81707704e-01 -4.66527164e-01
-1.04233587e+00 1.18500650e+00 3.20302337e-01 3.45525086e-01
-4.99229133e-01 7.61424005e-01 5.62953830e-01 -1.68524176e-01
-2.69463599e-01 -2.16422886e-01 -7.05783963e-02 -2.22215459e-01
6.27589583e-01 1.59387469e-01 -5.35858035e-01 -7.11554229e-01
-5.45692146e-01 6.26720846e-01 -6.26717508e-02 -4.92045164e-01
6.69326663e-01 1.45133302e-01 -3.39654423e-02 6.42705858e-01
9.93381679e-01 -2.61269510e-01 -8.74049604e-01 -2.49292508e-01
-1.93270713e-01 -8.05122852e-01 -6.61498427e-01 -1.03805327e+00
-9.94425118e-02 7.34448850e-01 6.25871837e-01 -1.54263720e-01
9.40487921e-01 3.64543974e-01 2.50362009e-01 7.17299700e-01
3.13090712e-01 -6.64857209e-01 5.80181122e-01 2.09216148e-01
1.72540390e+00 -1.42508948e+00 1.16309412e-01 -3.10828894e-01
-1.15505862e+00 7.10302413e-01 1.09053504e+00 -3.60932022e-01
3.65037136e-02 -1.92877978e-01 2.32151031e-01 -2.18873680e-01
-5.33568621e-01 -9.27556828e-02 5.82429945e-01 8.90916586e-01
3.17968667e-01 8.79186764e-02 6.16789870e-02 5.45677960e-01
-5.97443402e-01 -1.97654128e-01 3.86571616e-01 6.04239404e-01
-4.42285612e-02 -6.20766699e-01 -4.34800506e-01 3.17906708e-01
2.34777942e-01 -4.01837736e-01 -7.79020071e-01 8.36680353e-01
4.48444858e-02 8.02953541e-01 3.91053796e-01 -2.93570817e-01
7.14287162e-01 7.73812272e-03 8.20404410e-01 -4.79416311e-01
-5.29480100e-01 -7.31246233e-01 1.29759312e-01 -8.57461989e-01
-4.19813871e-01 -5.72443426e-01 -1.04028726e+00 -3.21980387e-01
3.49017113e-01 -1.81015432e-01 1.32384598e-01 8.63588691e-01
2.53672749e-01 1.79588169e-01 1.03117697e-01 -9.48390365e-01
-3.44516486e-01 -9.48361754e-01 4.60196240e-03 1.22147202e+00
-4.23516445e-02 -7.41520226e-01 -1.32812232e-01 3.01087886e-01] | [10.719136238098145, 2.0526723861694336] |
dc97eaa8-37c9-44db-8faf-9ab5622a5c5b | denoising-diffusion-models-for-plug-and-play | 2305.08995 | null | https://arxiv.org/abs/2305.08995v1 | https://arxiv.org/pdf/2305.08995v1.pdf | Denoising Diffusion Models for Plug-and-Play Image Restoration | Plug-and-play Image Restoration (IR) has been widely recognized as a flexible and interpretable method for solving various inverse problems by utilizing any off-the-shelf denoiser as the implicit image prior. However, most existing methods focus on discriminative Gaussian denoisers. Although diffusion models have shown impressive performance for high-quality image synthesis, their potential to serve as a generative denoiser prior to the plug-and-play IR methods remains to be further explored. While several other attempts have been made to adopt diffusion models for image restoration, they either fail to achieve satisfactory results or typically require an unacceptable number of Neural Function Evaluations (NFEs) during inference. This paper proposes DiffPIR, which integrates the traditional plug-and-play method into the diffusion sampling framework. Compared to plug-and-play IR methods that rely on discriminative Gaussian denoisers, DiffPIR is expected to inherit the generative ability of diffusion models. Experimental results on three representative IR tasks, including super-resolution, image deblurring, and inpainting, demonstrate that DiffPIR achieves state-of-the-art performance on both the FFHQ and ImageNet datasets in terms of reconstruction faithfulness and perceptual quality with no more than 100 NFEs. The source code is available at {\url{https://github.com/yuanzhi-zhu/DiffPIR}} | ['Luc van Gool', 'Radu Timofte', 'Bihan Wen', 'JieZhang Cao', 'Jingyun Liang', 'Kai Zhang', 'Yuanzhi Zhu'] | 2023-05-15 | null | null | null | null | ['deblurring'] | ['computer-vision'] | [ 1.39564991e-01 -1.79314196e-01 1.74611896e-01 -2.70449907e-01
-9.75266993e-01 -1.85785845e-01 6.78886354e-01 -5.76868355e-01
-1.43631876e-01 5.66388726e-01 1.65852517e-01 -9.27327946e-02
-1.63703352e-01 -6.92660332e-01 -7.05521226e-01 -1.00805700e+00
3.15824181e-01 1.56401515e-01 1.20458961e-01 -2.43570551e-01
-1.52567970e-02 2.43173614e-01 -1.32334030e+00 5.38261384e-02
1.22901583e+00 9.90912795e-01 6.17177665e-01 4.43727046e-01
1.39582381e-01 7.86605060e-01 -3.17831725e-01 -3.39461118e-01
2.95806199e-01 -5.98460257e-01 -3.72443438e-01 -6.82595000e-02
2.01937273e-01 -6.27009928e-01 -6.59876645e-01 1.30931044e+00
7.96872973e-01 2.77512759e-01 4.24986988e-01 -7.47723162e-01
-1.41745830e+00 1.97851986e-01 -7.25435615e-01 4.17118996e-01
3.09894919e-01 3.24180663e-01 6.08316660e-01 -1.02545297e+00
5.55914521e-01 1.15814161e+00 5.69949389e-01 5.09893537e-01
-1.42534614e+00 -7.32188702e-01 -2.36334577e-01 3.97044092e-01
-1.55335581e+00 -6.98651195e-01 9.12175000e-01 -2.78269142e-01
6.14388525e-01 3.77842456e-01 3.79737228e-01 9.36952412e-01
1.96520776e-01 7.53868341e-01 1.39461350e+00 -1.60516426e-01
1.74986705e-01 -2.19227403e-01 -2.69043654e-01 4.21437651e-01
-7.80244321e-02 2.50458002e-01 -4.80694056e-01 4.38534543e-02
1.26654339e+00 -1.09216936e-01 -5.68860590e-01 4.95701171e-02
-9.10603464e-01 7.15977013e-01 6.14214897e-01 2.44006500e-01
-8.03193927e-01 2.65640885e-01 9.45170894e-02 2.23385617e-01
9.24750686e-01 1.31802499e-01 1.25158221e-01 -1.00167848e-01
-1.17345476e+00 3.17928702e-01 2.28402734e-01 6.90426350e-01
5.86478055e-01 4.70857620e-01 -2.55166322e-01 1.23025298e+00
2.97894478e-01 4.32392806e-01 4.09238338e-01 -1.37929893e+00
1.48192886e-02 2.79573575e-02 1.69779778e-01 -1.06357312e+00
1.97127342e-01 -5.19375682e-01 -1.20435631e+00 4.06134963e-01
2.51702964e-03 1.46213233e-01 -1.02115560e+00 1.51631260e+00
3.79296482e-01 4.76398170e-01 -1.17350973e-01 1.19565630e+00
8.69328201e-01 8.76186192e-01 -2.13216931e-01 -2.04196736e-01
1.35176015e+00 -1.02734613e+00 -9.52281713e-01 -2.23022580e-01
-2.99230546e-01 -1.03158319e+00 9.33066845e-01 6.77766562e-01
-1.56445718e+00 -6.80557251e-01 -1.05486166e+00 -4.88963097e-01
3.15353274e-01 -1.27175907e-02 5.58434188e-01 4.66512620e-01
-1.27185476e+00 7.33104229e-01 -1.07489848e+00 8.18682984e-02
5.43428302e-01 4.91280633e-04 -3.03622365e-01 -4.26076710e-01
-1.09625924e+00 8.61098826e-01 -6.31456375e-02 6.19244099e-01
-1.09902406e+00 -6.32682025e-01 -6.13068700e-01 -5.37515506e-02
3.04354310e-01 -9.56505418e-01 1.12670112e+00 -8.34061444e-01
-1.86992431e+00 4.68109399e-01 -2.94097900e-01 -3.08785349e-01
6.58914208e-01 -4.51450229e-01 -5.05220413e-01 3.41331363e-01
1.43288478e-01 5.54765403e-01 1.10501182e+00 -1.29908502e+00
-1.22561999e-01 -1.87149420e-01 -1.07086048e-01 1.74267888e-01
-5.68205416e-02 1.96552888e-01 -7.55853951e-01 -1.11492085e+00
3.41537625e-01 -8.09195399e-01 -1.91805556e-01 1.79470137e-01
-3.32073212e-01 4.45075519e-02 7.13204443e-01 -1.22052133e+00
1.16136527e+00 -2.13401270e+00 2.09043741e-01 -1.77507073e-01
2.08407670e-01 4.67236698e-01 -8.85024071e-02 2.86944091e-01
-1.05526887e-01 -3.76804508e-02 -2.76794821e-01 -4.44577396e-01
4.41385480e-03 2.18856707e-01 -2.92662203e-01 7.09270775e-01
5.67087010e-02 9.53484595e-01 -6.12800956e-01 -1.60840124e-01
4.10727918e-01 1.14355409e+00 -5.22694170e-01 2.44781062e-01
1.03614837e-01 8.07770252e-01 -3.51287276e-01 5.06112576e-01
8.99719775e-01 -3.06100279e-01 -2.39810236e-02 -4.38031167e-01
-9.15646553e-02 2.40146711e-01 -1.22350717e+00 1.77395391e+00
-3.26061815e-01 5.52891433e-01 2.65900701e-01 -8.89977753e-01
7.50303447e-01 5.22439361e-01 3.47647876e-01 -1.01702309e+00
-7.02240914e-02 4.29797858e-01 -2.36032233e-01 -3.59883726e-01
5.36674142e-01 -3.28203380e-01 3.65302056e-01 7.86640868e-02
-1.32577166e-01 -9.40688103e-02 2.11230651e-01 9.46356133e-02
8.59900177e-01 4.82767701e-01 5.39937653e-02 -1.65411323e-01
3.80862772e-01 -2.95482874e-01 6.39780045e-01 6.50546014e-01
1.50400370e-01 1.16873419e+00 -1.14140811e-03 -1.28910154e-01
-1.03947008e+00 -1.19414663e+00 -1.53395593e-01 6.61274612e-01
2.61772543e-01 -1.66793555e-01 -8.99614334e-01 9.59814116e-02
-5.58182716e-01 9.34729517e-01 -3.72702062e-01 -3.26159298e-02
-5.12199461e-01 -7.87312150e-01 4.36594933e-01 3.14027429e-01
9.17698443e-01 -9.07068849e-01 -3.30095023e-01 3.17764074e-01
-5.96377313e-01 -9.31673586e-01 -5.71794808e-01 -2.36148253e-01
-8.80189061e-01 -6.37531400e-01 -1.12809062e+00 -4.96745169e-01
6.02735400e-01 5.76903403e-01 8.88583124e-01 -1.16177447e-01
-6.89672306e-02 1.84840515e-01 -3.36835802e-01 1.12361826e-01
-4.33400869e-01 -4.37050492e-01 -1.61484987e-01 5.77268451e-02
-1.72288299e-01 -1.12760234e+00 -1.16609538e+00 4.77707028e-01
-1.15547800e+00 1.99293569e-01 5.86344361e-01 8.88822198e-01
8.83867681e-01 1.60355508e-01 4.32736695e-01 -5.59122384e-01
7.68737793e-01 -3.91441852e-01 -4.87515807e-01 -1.04403265e-01
-8.18965197e-01 -1.18956640e-01 5.38106859e-01 -4.49682176e-01
-1.24640620e+00 -3.34047556e-01 -3.90767455e-01 -7.86602318e-01
3.31998020e-02 6.12439692e-01 -1.84314564e-01 -7.92760774e-02
5.46348631e-01 6.06591702e-01 1.99413434e-01 -8.28442872e-01
5.91812551e-01 3.74386072e-01 7.87363946e-01 -4.79911476e-01
8.88345778e-01 5.82227826e-01 -2.97753155e-01 -7.83594012e-01
-6.05415106e-01 -2.14454606e-01 1.56962909e-02 -1.65939331e-01
8.68198097e-01 -1.18006265e+00 -5.39232612e-01 7.39311457e-01
-9.51881468e-01 -3.46298218e-01 -1.97847128e-01 3.66527379e-01
-5.30850470e-01 7.28108764e-01 -8.92651796e-01 -5.63658416e-01
-5.41002393e-01 -1.35705531e+00 7.84425974e-01 3.31919909e-01
-1.30393486e-02 -8.37148368e-01 -1.66467041e-01 6.10915959e-01
8.23294997e-01 9.33697596e-02 5.30134141e-01 5.10493219e-02
-8.29682767e-01 -1.33243665e-01 -1.82424426e-01 7.30394304e-01
1.43986300e-01 -3.25883597e-01 -9.30028975e-01 -4.56066102e-01
4.13471460e-01 -7.50793740e-02 7.92723835e-01 7.47663379e-01
1.16529465e+00 -5.37823021e-01 1.92348018e-01 8.86042893e-01
1.44122195e+00 1.79487333e-01 1.42478967e+00 3.73000741e-01
6.03286982e-01 9.09461752e-02 3.77376437e-01 2.78402746e-01
2.98062623e-01 8.99405420e-01 3.92137408e-01 -2.04974234e-01
-5.89375317e-01 -2.35640094e-01 4.12102342e-01 8.33912909e-01
-4.23453480e-01 -2.68390864e-01 -4.91302371e-01 4.33928132e-01
-1.67818999e+00 -1.06984162e+00 -1.92950860e-01 2.01636529e+00
1.03132629e+00 -1.46185175e-01 -2.91044414e-01 -5.06321043e-02
6.20227814e-01 3.99629444e-01 -6.78481758e-01 -1.20637715e-02
-2.11646214e-01 2.91918069e-01 2.17091411e-01 5.09402633e-01
-8.39148939e-01 7.68284380e-01 5.13911915e+00 1.24784303e+00
-1.12711191e+00 4.78147864e-01 8.21650982e-01 -2.58914940e-02
-3.59094024e-01 8.45447257e-02 -4.25688684e-01 5.02428412e-01
7.46345043e-01 9.13065150e-02 9.91853058e-01 5.25188327e-01
5.92143476e-01 -1.74410626e-01 -5.60363293e-01 1.03622222e+00
-1.73165470e-01 -1.32439518e+00 -1.10542797e-01 1.85797433e-03
8.19127023e-01 2.90803947e-02 2.87169993e-01 -1.14416882e-01
2.50632703e-01 -1.02964854e+00 9.98315573e-01 8.17435265e-01
8.91196012e-01 -5.26503444e-01 4.59351867e-01 2.52434164e-01
-9.42929268e-01 2.79362112e-01 -4.16206479e-01 1.78374395e-01
5.10838568e-01 8.68358552e-01 -1.90361857e-01 7.32343554e-01
8.95724654e-01 6.46910131e-01 -1.96224019e-01 1.06331992e+00
-5.48538327e-01 7.53921926e-01 -1.74751207e-01 6.24871969e-01
-4.63263318e-02 -5.38212538e-01 7.83618391e-01 7.88994074e-01
5.66608906e-01 2.49995574e-01 -1.26098529e-01 1.15632141e+00
4.44798172e-02 -2.37385944e-01 -3.66785303e-02 4.82387953e-02
2.20817804e-01 1.15116704e+00 -5.72871566e-01 -2.56947368e-01
-1.52397931e-01 1.39544117e+00 -1.71590716e-01 7.22211778e-01
-9.88506198e-01 -2.56531179e-01 6.82358086e-01 3.78538251e-01
6.68554366e-01 -2.83589810e-01 -1.67308673e-01 -1.16019845e+00
2.56424814e-01 -1.06221819e+00 -1.33932784e-01 -1.21900117e+00
-1.35369313e+00 7.15067148e-01 -2.84219552e-02 -1.17950904e+00
-2.35019758e-01 -1.41110286e-01 -4.00197476e-01 1.27549410e+00
-1.61183989e+00 -1.23857605e+00 -3.89817238e-01 6.39308393e-01
6.77227318e-01 2.39999279e-01 5.90599775e-01 5.71269095e-01
-4.77148682e-01 2.46105224e-01 4.34387207e-01 -1.85419947e-01
5.42400002e-01 -8.68662477e-01 4.37421024e-01 1.16015220e+00
4.59685363e-03 7.95994818e-01 1.08948302e+00 -6.54695868e-01
-1.48585868e+00 -8.69872808e-01 3.80467176e-01 7.31605068e-02
4.34242874e-01 6.07348904e-02 -1.20523238e+00 4.22355026e-01
5.72986186e-01 2.44988933e-01 2.19238386e-01 -4.66558278e-01
-2.09868670e-01 -1.75827205e-01 -1.10875332e+00 6.20235920e-01
9.06515896e-01 -4.97123867e-01 -2.25039288e-01 5.95278293e-02
3.96783948e-01 -5.22047758e-01 -9.91833448e-01 2.85120547e-01
3.85715067e-01 -1.26835942e+00 1.33676898e+00 1.81579351e-01
7.69765317e-01 -5.99429250e-01 -1.36612520e-01 -1.20884967e+00
-6.02134466e-01 -9.47468996e-01 -1.86299697e-01 1.42460883e+00
-4.88596149e-02 -8.76522601e-01 3.24424356e-01 5.14363229e-01
-3.14762414e-01 -7.41039574e-01 -9.23267901e-01 -8.00161481e-01
-6.33140430e-02 -3.54424149e-01 2.79087842e-01 7.87825048e-01
-6.50733054e-01 1.72006592e-01 -7.37963080e-01 2.02816218e-01
8.70089114e-01 -1.26013055e-01 5.58746994e-01 -6.67653382e-01
-6.68678463e-01 -3.72458518e-01 -2.70763598e-02 -1.49880528e+00
-1.44283742e-01 -7.71670282e-01 1.87151536e-01 -1.93824327e+00
1.07107319e-01 -4.35802042e-01 -3.45601514e-02 3.58267993e-01
-2.98329800e-01 6.13143623e-01 2.14363113e-01 6.26694381e-01
-1.81683734e-01 8.85105669e-01 1.50413156e+00 -4.63587269e-02
6.77021071e-02 -1.53708801e-01 -9.03749883e-01 6.07380927e-01
7.67597675e-01 -4.35827702e-01 -5.62086523e-01 -6.34153962e-01
6.93507716e-02 3.12009096e-01 7.52547443e-01 -7.58398414e-01
1.79609627e-01 -1.00244410e-01 2.37408295e-01 -2.33494788e-01
6.21720612e-01 -4.80715334e-01 8.87999773e-01 1.15685046e-01
-7.77894408e-02 -5.22884429e-02 1.99735597e-01 5.72416842e-01
-3.70669425e-01 -1.74728081e-01 9.36233819e-01 -1.68284833e-01
-6.76749170e-01 1.83573529e-01 -2.97521710e-01 2.61434950e-02
4.77440685e-01 -3.99604529e-01 -3.29972625e-01 -5.50741196e-01
-6.12450480e-01 -2.78099358e-01 5.48347294e-01 2.96200335e-01
8.33491266e-01 -1.21806693e+00 -9.77034152e-01 2.32347706e-03
-5.18823147e-01 1.41122729e-01 7.66730249e-01 1.09272969e+00
-5.28405666e-01 -2.39984393e-02 -1.87517684e-02 -4.08087283e-01
-9.95901763e-01 4.82659727e-01 3.21508288e-01 -2.35128447e-01
-9.35957909e-01 8.92806768e-01 3.88445437e-01 -1.21145612e-02
-1.40545607e-01 7.25994632e-02 1.76429912e-01 -3.31503361e-01
6.57004058e-01 5.18169165e-01 -2.15366501e-02 -5.47980547e-01
-5.66206127e-02 3.44463468e-01 -2.63359956e-02 -2.38980785e-01
1.62624609e+00 -3.95582855e-01 -3.30947965e-01 8.10260922e-02
9.73035693e-01 -1.96616538e-02 -1.68747377e+00 -2.01120779e-01
-4.43499386e-01 -7.52073467e-01 5.05930662e-01 -8.68212640e-01
-1.42167807e+00 8.13222051e-01 7.89275289e-01 3.01201753e-02
1.49812460e+00 -1.74960658e-01 9.56720293e-01 -3.70489746e-01
3.81902784e-01 -7.24539340e-01 1.67548150e-01 8.12165290e-02
1.32949996e+00 -8.78777981e-01 2.05452174e-01 -4.93704945e-01
-5.91393769e-01 6.85811400e-01 1.32587954e-01 -3.08489412e-01
4.74875182e-01 1.32637635e-01 2.53941957e-02 5.58084110e-03
-4.40254927e-01 1.57720625e-01 3.23998034e-01 4.77360874e-01
4.21950489e-01 -1.55722588e-01 -4.47604358e-01 3.74827147e-01
2.68869847e-02 3.52262795e-01 4.19129401e-01 6.41416550e-01
-1.81827128e-01 -1.05162930e+00 -3.75272512e-01 6.76944330e-02
-5.71965873e-01 -4.46156412e-01 3.77703160e-01 4.01233941e-01
-8.75038654e-02 1.00537670e+00 -3.18908811e-01 -1.34210214e-01
1.41230628e-01 -3.97608757e-01 4.51603830e-01 -1.22032166e-01
-4.13808614e-01 5.91952682e-01 4.70105000e-02 -6.48983002e-01
-5.39515078e-01 -5.46387970e-01 -1.04376435e+00 -4.58288223e-01
-2.92511851e-01 -9.81593132e-02 5.74505806e-01 6.29212618e-01
6.39242291e-01 7.50695527e-01 3.67687970e-01 -1.06890416e+00
-5.83308697e-01 -1.02151787e+00 -5.85851848e-01 2.09388196e-01
2.00380653e-01 -6.02973104e-01 -2.44245544e-01 1.88191473e-01] | [11.472803115844727, -2.251838445663452] |
4e4264b2-5ee4-4f86-b3fe-08e14b815215 | a-lexicon-based-graph-neural-network-for | null | null | https://aclanthology.org/D19-1096 | https://aclanthology.org/D19-1096.pdf | A Lexicon-Based Graph Neural Network for Chinese NER | Recurrent neural networks (RNN) used for Chinese named entity recognition (NER) that sequentially track character and word information have achieved great success. However, the characteristic of chain structure and the lack of global semantics determine that RNN-based models are vulnerable to word ambiguities. In this work, we try to alleviate this problem by introducing a lexicon-based graph neural network with global semantics, in which lexicon knowledge is used to connect characters to capture the local composition, while a global relay node can capture global sentence semantics and long-range dependency. Based on the multiple graph-based interactions among characters, potential words, and the whole-sentence semantics, word ambiguities can be effectively tackled. Experiments on four NER datasets show that the proposed model achieves significant improvements against other baseline models. | ['Xuanjing Huang', 'Tao Gui', 'Minlong Peng', 'Yicheng Zou', 'Zhongyu Wei', 'Qi Zhang', 'Jinlan Fu'] | 2019-11-01 | null | null | null | ijcnlp-2019-11 | ['chinese-named-entity-recognition'] | ['natural-language-processing'] | [-1.32609472e-01 -1.42307356e-01 -2.93639809e-01 -2.03611434e-01
-2.16945603e-01 -6.01714969e-01 1.20070405e-01 5.02198398e-01
-6.43073559e-01 5.97847879e-01 6.65601254e-01 -5.47956944e-01
1.90150976e-01 -1.14773393e+00 -1.95402682e-01 -3.51254106e-01
4.90708910e-02 1.75523490e-01 4.63434964e-01 -6.24049723e-01
3.47680062e-01 3.12380970e-01 -5.36359310e-01 1.40491053e-02
1.02312720e+00 3.24942380e-01 4.16618764e-01 3.00366819e-01
-1.12125623e+00 1.16649592e+00 -6.72989130e-01 -5.66249609e-01
-3.91658753e-01 -6.07287526e-01 -8.22355270e-01 -4.26177889e-01
-2.23192319e-01 4.97776456e-02 -4.54640031e-01 1.29494023e+00
5.65324962e-01 2.11025327e-01 2.45563298e-01 -5.55327117e-01
-7.85863638e-01 1.31243706e+00 -2.55612820e-01 2.83566028e-01
2.76506275e-01 -3.05389345e-01 1.30772352e+00 -6.01150990e-01
7.58124769e-01 1.09775162e+00 8.57565820e-01 6.63668573e-01
-6.33740187e-01 -5.13485491e-01 5.86675704e-01 2.33145565e-01
-1.52585900e+00 1.02396365e-02 8.39917958e-01 1.69953853e-01
1.57825840e+00 -8.00165087e-02 4.95787144e-01 9.93585944e-01
3.28075975e-01 6.41713917e-01 3.89121294e-01 -4.78419453e-01
1.01515008e-02 -2.54174352e-01 7.68529415e-01 7.82232225e-01
3.67301613e-01 -5.47511578e-01 -1.84511587e-01 1.16755098e-01
7.51189709e-01 -4.47808430e-02 -7.92913437e-02 9.95346308e-02
-7.86387146e-01 8.60229492e-01 6.59177482e-01 8.24208796e-01
-3.72140229e-01 7.98434168e-02 6.37600839e-01 1.14330426e-01
2.89705902e-01 5.02421916e-01 -5.46884954e-01 -1.22207114e-02
-5.48619449e-01 -3.83342743e-01 9.51063633e-01 9.92586732e-01
6.29490674e-01 3.39550585e-01 -2.10337028e-01 8.47947896e-01
3.67806256e-01 3.32648814e-01 8.50134492e-01 4.12095413e-02
6.83616817e-01 9.39745843e-01 -4.36395705e-01 -1.39555979e+00
-6.25734508e-01 -4.85744268e-01 -9.47215140e-01 -7.27928758e-01
7.79056922e-02 -4.17737365e-01 -1.13691187e+00 1.65224648e+00
-2.52726879e-02 3.23162615e-01 3.47685158e-01 7.43759692e-01
1.23355258e+00 8.83715153e-01 4.26946580e-01 -8.29992667e-02
1.31723595e+00 -9.26074505e-01 -1.00257409e+00 -5.05488575e-01
8.67080808e-01 -5.32858372e-01 7.25298524e-01 -2.30726570e-01
-6.57652438e-01 -3.57580543e-01 -9.42019105e-01 -1.52954042e-01
-6.90280378e-01 -1.99301302e-01 6.20286345e-01 7.38018036e-01
-1.08995259e+00 3.30087841e-01 -7.65073180e-01 -5.11287749e-01
-1.44156627e-02 3.41009140e-01 -9.81103033e-02 1.21335778e-02
-1.77784014e+00 8.59847724e-01 9.71616924e-01 7.07246900e-01
-3.93095255e-01 -1.61805153e-02 -1.07484460e+00 2.12857187e-01
4.29748356e-01 -3.17891032e-01 7.22351372e-01 -5.64212918e-01
-1.33142960e+00 3.46792608e-01 -3.05799216e-01 -3.65324348e-01
-9.22622830e-02 -2.76391786e-02 -8.16995680e-01 3.04919407e-02
-9.38838348e-03 2.20916837e-01 1.27533108e-01 -7.46786237e-01
-4.41403359e-01 -2.52670974e-01 -1.45550758e-01 3.95794511e-01
-6.56830609e-01 2.71949142e-01 -7.85332263e-01 -7.42578387e-01
1.88350275e-01 -5.04440188e-01 -5.12915075e-01 -1.03398550e+00
-7.12830424e-01 -4.81454730e-01 4.31474030e-01 -6.89670324e-01
1.79318106e+00 -1.80272150e+00 1.27982274e-01 3.34527314e-01
1.06317423e-01 4.72297251e-01 -5.02099097e-01 8.19348514e-01
9.47398096e-02 5.52482486e-01 8.89902860e-02 -2.73857471e-02
-2.23015174e-01 4.24622059e-01 -1.51720330e-01 -4.30890135e-02
3.12698662e-01 1.37869334e+00 -9.03823555e-01 -5.28039813e-01
-4.50452939e-02 4.82229173e-01 3.40173617e-02 1.55650955e-02
-2.91687012e-01 -4.15541232e-03 -8.96484375e-01 4.38601375e-01
5.15761018e-01 -3.34716558e-01 8.80113542e-01 -4.34525982e-02
1.43025760e-02 6.75901473e-01 -1.14887130e+00 1.56394196e+00
-3.84564847e-01 3.19197625e-01 -3.25019032e-01 -8.54925215e-01
1.19559181e+00 3.00506502e-01 -1.43827453e-01 -7.60019779e-01
2.54488111e-01 1.71583876e-01 -9.57394019e-02 -2.39397883e-01
6.42930448e-01 3.35489437e-02 -3.93173695e-01 3.09064478e-01
9.05808806e-02 3.07636827e-01 2.05939725e-01 4.42865998e-01
1.18625104e+00 -2.27733701e-02 3.24671328e-01 -9.40015912e-02
7.02605605e-01 1.01004124e-01 8.37207317e-01 7.10912049e-01
3.10722739e-02 4.78262752e-01 6.04983389e-01 -3.99593532e-01
-7.27960467e-01 -7.64343143e-01 4.97947067e-01 1.10817444e+00
3.15122962e-01 -7.43670344e-01 -7.35089004e-01 -7.37421632e-01
-6.01468861e-01 7.20016658e-01 -3.60153705e-01 -2.41330653e-01
-1.13272417e+00 -7.41281569e-01 8.81089628e-01 8.39562178e-01
4.75040495e-01 -1.43355048e+00 9.74151790e-02 7.39899039e-01
-3.60161752e-01 -1.30357444e+00 -6.20050848e-01 2.16058642e-01
-9.11223292e-01 -7.20793545e-01 -4.20461535e-01 -1.30595958e+00
6.44308984e-01 2.95656741e-01 1.05954325e+00 5.28751016e-01
8.39114785e-02 -1.06141359e-01 -8.29478502e-01 -1.18311033e-01
-3.01967144e-01 6.59690619e-01 -3.24508101e-01 -2.08507374e-01
6.44882143e-01 -4.00754690e-01 -1.58483565e-01 1.14954054e-01
-9.93009865e-01 -1.49980411e-01 4.97832865e-01 7.90858388e-01
3.22061747e-01 1.21191971e-01 6.92811608e-01 -1.12417233e+00
1.03814518e+00 -4.82699692e-01 -4.49437469e-01 6.87166154e-01
-6.28532350e-01 3.36001009e-01 7.85023451e-01 -3.56568247e-01
-1.39975262e+00 -2.11657107e-01 -3.45768392e-01 1.88650072e-01
-7.93058500e-02 1.18041956e+00 -3.86185974e-01 1.75534517e-01
2.39256904e-01 3.73404831e-01 -4.09943968e-01 -5.94353795e-01
5.02238989e-01 4.54690814e-01 4.27624494e-01 -4.83979315e-01
5.12559414e-01 -6.54205605e-02 -2.27020308e-01 -7.99886763e-01
-8.74299467e-01 -5.45123994e-01 -7.75009334e-01 8.06519464e-02
1.11996460e+00 -8.23046625e-01 -5.38173199e-01 6.38013899e-01
-1.51258624e+00 -8.00622776e-02 1.27315626e-01 4.51582581e-01
2.16554940e-01 6.96725428e-01 -1.17723072e+00 -7.08433688e-01
-5.78676760e-01 -8.00004244e-01 4.44680035e-01 6.89987123e-01
-8.35847184e-02 -1.37005055e+00 6.12961408e-03 1.02256099e-02
4.61526752e-01 -1.42509148e-01 1.09533656e+00 -1.17715502e+00
-4.87778038e-01 -2.84750164e-01 -2.90020108e-01 4.70672315e-03
3.49008620e-01 -3.28752637e-01 -3.00701320e-01 4.06004302e-02
-2.70002872e-01 -9.56375450e-02 9.26891804e-01 8.38432088e-02
5.25178850e-01 -1.82328478e-01 -2.86663562e-01 4.28037435e-01
1.58284593e+00 1.78349435e-01 7.22576022e-01 4.44153398e-01
1.25625193e+00 4.56272244e-01 1.54762149e-01 2.07238927e-01
7.04558194e-01 1.49926618e-01 3.11404884e-01 -5.45740426e-02
-1.08583100e-01 -6.50666833e-01 2.96641827e-01 1.59802210e+00
-1.40171861e-02 -7.78319836e-01 -1.09537375e+00 4.63434905e-01
-1.97545624e+00 -6.57437623e-01 -4.12620485e-01 1.68483305e+00
7.62174785e-01 3.70876908e-01 -2.27349550e-01 -4.04567063e-01
1.04445279e+00 4.65831071e-01 -2.49511376e-01 -5.01200378e-01
-6.07459664e-01 9.83669758e-02 6.51987016e-01 4.31521237e-01
-8.44023287e-01 1.75111341e+00 5.94617414e+00 9.14409339e-01
-1.00570273e+00 -6.24866858e-02 2.86985427e-01 6.50247574e-01
-6.70741320e-01 4.36545685e-02 -1.21551168e+00 2.25350484e-01
8.40426445e-01 -5.80780171e-02 2.96374470e-01 4.52345610e-01
1.54696004e-02 2.49571174e-01 -3.15857708e-01 5.59841931e-01
2.18361095e-01 -1.34606206e+00 4.54454243e-01 -3.55592668e-01
4.62880939e-01 2.84820676e-01 -5.65481901e-01 3.29022765e-01
6.55523002e-01 -9.77147162e-01 3.34393263e-01 4.11621481e-01
4.88112450e-01 -9.61706400e-01 1.18197393e+00 3.62148404e-01
-1.74584651e+00 1.85331792e-01 -5.36824465e-01 -4.84898128e-02
4.36487257e-01 2.85163194e-01 -6.72948778e-01 9.39463079e-01
2.84095615e-01 8.06800485e-01 -5.40682256e-01 7.94317663e-01
-9.08236742e-01 8.14981222e-01 -1.72654599e-01 -8.35829854e-01
5.74338317e-01 -4.26649034e-01 4.45153952e-01 1.65067005e+00
9.03778523e-03 2.97896147e-01 1.64826021e-01 5.88796794e-01
-2.67985374e-01 5.67564011e-01 -4.47577029e-01 -4.68048006e-01
5.39885402e-01 1.25405085e+00 -1.32545996e+00 -1.66502774e-01
-4.85416710e-01 1.18057811e+00 8.30707073e-01 4.95377153e-01
-5.53613901e-01 -8.41614723e-01 4.06938374e-01 -5.81081867e-01
5.88684559e-01 -6.58346772e-01 -1.86357558e-01 -1.35377717e+00
-1.24039993e-01 -6.69255137e-01 6.68800712e-01 -5.89555383e-01
-1.41064680e+00 9.66801167e-01 -4.30323005e-01 -6.09735548e-01
-2.36273315e-02 -5.81640840e-01 -9.52399611e-01 8.48600447e-01
-1.51387787e+00 -1.24561048e+00 2.54282385e-01 5.75404406e-01
5.01518250e-01 -1.16403542e-01 9.00812328e-01 1.21067047e-01
-8.83350134e-01 6.52279079e-01 -8.56314152e-02 9.29995775e-01
2.39388779e-01 -1.04727936e+00 7.71111369e-01 1.06335425e+00
5.30998826e-01 1.00687504e+00 3.26371789e-01 -1.18724608e+00
-1.38433075e+00 -1.08275557e+00 1.49843097e+00 -1.42470300e-01
8.87160301e-01 -5.02255321e-01 -1.05357718e+00 6.93620861e-01
2.25345924e-01 -3.09720933e-01 6.63277984e-01 3.29283655e-01
-4.29211587e-01 1.50195181e-01 -4.24048454e-01 8.09153914e-01
1.14609611e+00 -6.54415369e-01 -8.24134946e-01 3.31113930e-04
1.16339970e+00 -3.28591555e-01 -4.90198493e-01 1.48085847e-01
1.89292699e-01 -5.40119112e-01 6.77043140e-01 -8.46361637e-01
5.45900688e-02 -2.12190479e-01 9.91218388e-02 -1.22663796e+00
-4.33660746e-01 -5.81907809e-01 1.65757209e-01 1.54534256e+00
8.75251889e-01 -6.43657088e-01 7.10287809e-01 3.51931691e-01
-1.47763625e-01 -4.02385861e-01 -6.83182180e-01 -7.06593990e-01
3.45359892e-02 -4.88951951e-01 6.42100275e-01 1.02833283e+00
2.29307935e-01 9.80010986e-01 -4.61629033e-01 1.99256167e-01
1.71139538e-01 -5.83367199e-02 -7.48610795e-02 -1.10250139e+00
1.22440256e-01 -4.37313914e-01 -4.33801651e-01 -1.23138773e+00
6.00058794e-01 -9.13297474e-01 1.52049273e-01 -1.93176329e+00
2.25958332e-01 -2.50399381e-01 -6.77374184e-01 6.33101642e-01
-4.33090776e-01 -2.93234810e-02 1.68396860e-01 6.31456124e-03
-8.52540970e-01 4.73888576e-01 9.70476806e-01 -9.63711068e-02
-4.08650130e-01 -3.02059114e-01 -6.58167720e-01 5.69986045e-01
1.06713736e+00 -6.44722104e-01 -2.46417165e-01 -9.29066896e-01
7.44786561e-01 -1.10362675e-02 -3.45339358e-01 -5.38584173e-01
8.67856801e-01 -1.33484766e-01 1.93909124e-01 -6.89794600e-01
-5.76978289e-02 -6.19401932e-01 -2.32328549e-01 4.56848621e-01
-3.13491046e-01 4.29538429e-01 5.16985264e-03 7.28182971e-01
-2.58536160e-01 -4.76750493e-01 1.57761022e-01 -3.49961996e-01
-1.01670980e+00 2.73377210e-01 -6.32804155e-01 2.07085177e-01
4.96119887e-01 -2.15510339e-01 -2.39437491e-01 -4.64937508e-01
-5.07745802e-01 4.18754518e-01 2.53253244e-02 7.48326182e-01
8.47442985e-01 -1.20306456e+00 -5.95920086e-01 1.56401843e-02
-7.63843432e-02 -1.36814117e-01 3.88992466e-02 3.00543189e-01
-6.50567651e-01 4.73318726e-01 -4.99499291e-02 7.61529580e-02
-1.09174240e+00 4.60882515e-01 2.40889207e-01 -5.57049930e-01
-6.23936415e-01 8.59414935e-01 -2.69492865e-01 -6.84799969e-01
2.20211104e-01 -1.75079748e-01 -8.84652793e-01 6.60347939e-02
3.96596372e-01 2.85266191e-01 -1.70693733e-02 -8.03526342e-01
-4.48825717e-01 7.14946210e-01 -2.13992953e-01 1.16956793e-01
1.27121723e+00 -5.33281267e-01 -5.31007469e-01 3.58156443e-01
9.64081168e-01 1.92755714e-01 -3.62253755e-01 -5.48408270e-01
6.60059392e-01 2.82641649e-01 7.50758266e-03 -5.80960572e-01
-1.07234240e+00 9.22379375e-01 -5.97536229e-02 2.49601081e-01
8.80449593e-01 -6.86774775e-02 1.23135400e+00 6.75364196e-01
2.34640956e-01 -1.13670433e+00 -1.53800994e-01 1.38863158e+00
1.18170202e-01 -9.12170291e-01 -3.99825722e-01 -5.47901273e-01
-7.25042164e-01 1.48299801e+00 8.08211803e-01 8.43556598e-02
6.12105012e-01 1.50998235e-01 3.33633602e-01 -2.09604949e-01
-4.12453622e-01 -4.35914576e-01 1.33061156e-01 4.42618459e-01
4.11805153e-01 -1.05078802e-01 -6.47942960e-01 8.41642916e-01
1.29755944e-01 -3.90172631e-01 4.94025379e-01 9.31314886e-01
-5.29802740e-01 -1.49219501e+00 1.24930017e-01 1.46932572e-01
-6.34264410e-01 -6.98485851e-01 -5.82974672e-01 5.12570381e-01
-1.31339684e-01 1.19046581e+00 5.61252460e-02 -3.87728721e-01
2.98539609e-01 3.47747393e-02 -7.06384629e-02 -8.24405313e-01
-8.46489310e-01 2.02685043e-01 3.61878127e-01 -3.19423079e-01
-2.93390900e-01 -9.43465680e-02 -1.86610234e+00 -5.30773439e-02
-6.69705629e-01 5.68848789e-01 5.48550308e-01 9.79792953e-01
3.47071916e-01 5.61029971e-01 4.88474190e-01 4.39769924e-02
-3.18053812e-01 -8.99623752e-01 -6.03548586e-01 2.03129828e-01
-2.86497716e-02 7.71472231e-02 -8.96116644e-02 -3.48305434e-01] | [9.799392700195312, 9.74716854095459] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.