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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0ae8ad1e-79d8-42d7-a2d3-1fd0899cf530 | temporal-knowledge-graph-reasoning-with-low | 2204.04783 | null | https://arxiv.org/abs/2204.04783v1 | https://arxiv.org/pdf/2204.04783v1.pdf | Temporal Knowledge Graph Reasoning with Low-rank and Model-agnostic Representations | Temporal knowledge graph completion (TKGC) has become a popular approach for reasoning over the event and temporal knowledge graphs, targeting the completion of knowledge with accurate but missing information. In this context, tensor decomposition has successfully modeled interactions between entities and relations. Their effectiveness in static knowledge graph completion motivates us to introduce Time-LowFER, a family of parameter-efficient and time-aware extensions of the low-rank tensor factorization model LowFER. Noting several limitations in current approaches to represent time, we propose a cycle-aware time-encoding scheme for time features, which is model-agnostic and offers a more generalized representation of time. We implement our methods in a unified temporal knowledge graph embedding framework, focusing on time-sensitive data processing. The experiments show that our proposed methods perform on par or better than the state-of-the-art semantic matching models on two benchmarks. | ['Günter Neumann', 'Saadullah Amin', 'Ioannis Dikeoulias'] | 2022-04-10 | null | https://aclanthology.org/2022.repl4nlp-1.12 | https://aclanthology.org/2022.repl4nlp-1.12.pdf | repl4nlp-acl-2022-5 | ['temporal-knowledge-graph-completion'] | ['knowledge-base'] | [-3.25503498e-01 -2.69572347e-01 -6.69834554e-01 -3.09481502e-01
-2.77346849e-01 -6.80679321e-01 6.78012371e-01 4.66567636e-01
-3.86074752e-01 1.48840994e-01 4.06769753e-01 -3.25138360e-01
-7.24357665e-01 -1.04381633e+00 -4.16403145e-01 -2.88812608e-01
-5.07432103e-01 6.87157512e-01 4.65922892e-01 -3.14686686e-01
8.22377726e-02 6.58542812e-01 -1.62463021e+00 4.68818098e-01
4.82495844e-01 8.99984479e-01 -4.57345158e-01 5.18089056e-01
-2.84043521e-01 1.23958755e+00 -1.22514009e-01 -7.90387690e-01
1.11292303e-01 3.16314489e-01 -1.31415355e+00 -2.03097478e-01
2.16649473e-01 -2.53336757e-01 -1.41974306e+00 6.40591800e-01
-8.61273259e-02 6.34573936e-01 2.21377149e-01 -1.71136332e+00
-7.56899893e-01 9.15745795e-01 -3.26243341e-01 5.83938956e-01
6.94979608e-01 -3.45137239e-01 1.21577466e+00 -5.76504469e-01
7.94604123e-01 1.46994841e+00 5.82739174e-01 1.97496474e-01
-1.19035351e+00 -2.89932996e-01 4.05019194e-01 1.08617890e+00
-1.62500477e+00 -2.28715956e-01 8.48918021e-01 -3.87157738e-01
1.45767999e+00 4.73835170e-01 6.24238789e-01 8.64162982e-01
1.58667922e-01 9.58669603e-01 8.12573135e-01 -3.28558236e-01
5.12862168e-02 -4.44857776e-01 6.27967358e-01 7.41381943e-01
1.80897772e-01 2.06441522e-01 -1.05781543e+00 -5.99069297e-01
4.51389432e-01 4.07022297e-01 -2.05348670e-01 -6.30231798e-01
-1.67574334e+00 7.38796532e-01 3.21065754e-01 4.49783981e-01
-3.18183810e-01 6.60194039e-01 8.40139508e-01 3.80209088e-01
5.42519987e-01 4.00911011e-02 -5.57719529e-01 -4.63051230e-01
-6.63759887e-01 3.84401888e-01 9.46515143e-01 1.03148723e+00
7.39873111e-01 -8.80399421e-02 -5.37565649e-01 4.03947204e-01
5.54541238e-02 1.80143073e-01 1.82265207e-01 -1.09912658e+00
3.54713291e-01 1.00414574e+00 1.40184194e-01 -1.37161481e+00
-4.58996028e-01 -1.77885041e-01 -3.10507685e-01 -6.54733360e-01
1.71824574e-01 7.45637774e-01 -8.01414549e-01 1.64595592e+00
4.89689171e-01 7.70338774e-01 1.64732486e-01 9.29352820e-01
5.82933664e-01 4.58307803e-01 1.98056340e-01 -2.29571939e-01
1.74033654e+00 -9.73683715e-01 -1.04993486e+00 1.32267401e-01
7.29260206e-01 -5.14045656e-01 5.89542747e-01 1.32039979e-01
-8.09193790e-01 -9.57688242e-02 -8.99427712e-01 -4.05256897e-01
-8.73232961e-01 -1.78290531e-01 1.83106220e+00 5.62319994e-01
-8.87953758e-01 8.27977538e-01 -1.36298537e+00 -3.82775128e-01
5.82488403e-02 1.83616206e-01 -5.36791146e-01 -3.86709243e-01
-1.46608365e+00 7.90008783e-01 5.19996881e-01 5.75636700e-02
-7.46324062e-01 -9.86780405e-01 -1.08114100e+00 9.70678329e-02
6.99197233e-01 -9.03936386e-01 1.25227416e+00 -2.66734585e-02
-1.12248909e+00 6.47329211e-01 -3.33327174e-01 -6.09791279e-01
2.19277799e-01 -1.04579493e-01 -9.50166583e-01 3.27917963e-01
1.26749754e-01 8.57891236e-03 6.62268460e-01 -6.98818743e-01
-5.39660335e-01 -6.36957645e-01 5.66644847e-01 -9.55557004e-02
-6.84858859e-01 1.65116712e-01 -9.72133994e-01 -7.27141619e-01
3.53331149e-01 -8.45925689e-01 -1.37054041e-01 -1.59534454e-01
-4.25209366e-02 -5.39163113e-01 9.38122869e-01 -6.90021276e-01
1.76129985e+00 -1.83462250e+00 4.78271127e-01 1.88058496e-01
4.57150161e-01 -1.94544807e-01 5.32884859e-02 9.56470609e-01
-1.75141796e-01 -2.37940431e-01 7.90468231e-02 -4.11827505e-01
3.04480463e-01 6.75540149e-01 -5.74721456e-01 5.99259138e-01
-2.21761361e-01 1.13300538e+00 -1.33366287e+00 -5.24616420e-01
2.23735958e-01 6.66649997e-01 -2.85484672e-01 -1.19026281e-01
-3.22513938e-01 -1.66508719e-01 -6.61824167e-01 8.37448299e-01
3.90377074e-01 -3.14727366e-01 5.21690130e-01 -7.80906439e-01
1.03711635e-02 1.34869367e-01 -1.23104429e+00 2.29150724e+00
-2.86972553e-01 1.32574260e-01 -4.87209022e-01 -8.88468027e-01
4.46686655e-01 4.72949952e-01 1.03073049e+00 -6.80272877e-01
-4.63467166e-02 -4.86668348e-02 -5.25136113e-01 -4.40813184e-01
9.75167871e-01 3.29693228e-01 -2.85342693e-01 5.49539268e-01
1.89099863e-01 3.83503944e-01 5.95151663e-01 7.17936158e-01
1.53471148e+00 3.09272587e-01 -6.73924387e-03 3.18953730e-02
4.18508559e-01 9.73341241e-02 7.69335449e-01 2.66323656e-01
-1.92664772e-01 -1.22992508e-01 3.85586172e-01 -7.59050667e-01
-6.43879533e-01 -9.83615816e-01 1.74603179e-01 1.27062500e+00
1.19143263e-01 -1.20413470e+00 -1.48091808e-01 -7.79019296e-01
5.61473429e-01 5.96525013e-01 -8.39949012e-01 -3.93921703e-01
-5.28218567e-01 -4.40108210e-01 6.62760079e-01 8.50739539e-01
8.15814883e-02 -2.09831342e-01 -3.28702152e-01 3.85857671e-01
-4.79382843e-01 -1.44853914e+00 -6.37650728e-01 -1.28629103e-01
-1.03988755e+00 -1.41218758e+00 -1.97992176e-01 -2.55386621e-01
4.19940859e-01 6.92913592e-01 1.10289633e+00 -1.30719841e-01
-4.62818623e-01 1.18397999e+00 -5.85389555e-01 -1.71759760e-03
1.66725367e-01 -2.05106735e-01 2.19705984e-01 3.58901411e-01
5.56904495e-01 -8.42176259e-01 -6.65850341e-01 3.72051507e-01
-1.08529246e+00 -1.58523157e-01 -1.64525732e-02 6.10545754e-01
6.03820920e-01 3.66776794e-01 -4.66951588e-03 -6.59678578e-01
5.15083909e-01 -4.50914532e-01 -6.38775229e-01 5.83947480e-01
-9.54721510e-01 4.63819474e-01 2.51057267e-01 -5.20834386e-01
-1.08949256e+00 -2.00576082e-01 6.89159930e-01 -1.20942461e+00
6.01483703e-01 8.15875471e-01 1.96688250e-01 -3.55497956e-01
4.89202946e-01 1.58154577e-01 -3.65579158e-01 -5.33516347e-01
9.03848529e-01 4.03048322e-02 6.35804534e-01 -1.15288222e+00
8.46760333e-01 1.05442524e+00 3.35139424e-01 -1.16066888e-01
-7.64383793e-01 -1.00018597e+00 -5.29642761e-01 -1.18411928e-01
5.07206023e-01 -9.14707839e-01 -1.22484183e+00 9.19482633e-02
-1.14068770e+00 1.20613880e-01 -3.69642884e-01 5.89650035e-01
-6.40115321e-01 8.53059649e-01 -7.70886719e-01 -7.10538805e-01
-2.84706384e-01 -6.77195251e-01 1.08749592e+00 -2.24767625e-01
1.59888025e-02 -1.22281301e+00 2.71724880e-01 6.38536274e-01
3.25371772e-01 3.83103371e-01 8.83748770e-01 -4.97447789e-01
-9.34912860e-01 -2.87887067e-01 -2.93468863e-01 -1.86738685e-01
-9.89274681e-02 -1.13385275e-01 -8.56522918e-01 -3.55139375e-01
-4.18479919e-01 6.81493282e-02 1.01090443e+00 -1.63272992e-01
1.10313869e+00 -2.38562837e-01 -7.51364350e-01 7.08125472e-01
1.48031735e+00 -3.32586989e-02 5.42767107e-01 2.81246662e-01
8.44707787e-01 3.08586866e-01 8.68544996e-01 6.38503790e-01
1.13836253e+00 8.72400701e-01 2.84072220e-01 5.16954720e-01
-1.59502551e-01 -3.50816160e-01 1.68627322e-01 9.67182636e-01
-6.19485140e-01 7.99698606e-02 -9.25623357e-01 1.08726335e+00
-2.52212453e+00 -1.06716168e+00 -2.74432778e-01 2.00407434e+00
6.85786724e-01 -2.01824144e-01 1.80536881e-02 3.87812853e-01
4.35929388e-01 3.85556936e-01 -3.11864883e-01 -3.94282997e-01
-3.19509842e-02 2.39363685e-01 8.39712858e-01 3.04320574e-01
-1.06605852e+00 1.09841049e+00 6.36644936e+00 7.60199189e-01
-6.05693877e-01 5.38840473e-01 -4.42912638e-01 4.60057370e-02
-6.37574911e-01 4.50723410e-01 -4.35452312e-01 6.96592173e-03
1.17766428e+00 -6.88345730e-01 8.44927013e-01 7.18745947e-01
-1.66198149e-01 2.91418940e-01 -1.34899354e+00 1.27587330e+00
-9.59647540e-03 -1.60456419e+00 6.53259009e-02 -1.93805009e-01
5.43399274e-01 -1.92598447e-01 -3.31319749e-01 5.39815247e-01
3.46540451e-01 -5.97313166e-01 6.79007471e-01 8.52640569e-01
5.66331863e-01 -7.04576135e-01 3.59288156e-01 -2.53398865e-01
-1.90694642e+00 -1.67894900e-01 -2.10822120e-01 -4.15269360e-02
2.34890267e-01 7.37847388e-01 -5.84235132e-01 1.48293591e+00
8.01402807e-01 9.20008302e-01 -4.23779190e-01 7.48299479e-01
-1.81339085e-02 3.12913239e-01 -2.01210454e-01 4.39604759e-01
-7.42385611e-02 4.88660745e-02 5.84861815e-01 1.14408755e+00
5.22099994e-02 5.33657931e-02 2.83554673e-01 6.81790233e-01
5.61346188e-02 -1.65573284e-01 -4.41993117e-01 -6.24829054e-01
5.13214886e-01 1.14970386e+00 -5.82333922e-01 -3.22666466e-01
-6.60509527e-01 1.13086867e+00 4.45044309e-01 5.82707942e-01
-9.67412770e-01 -3.00681502e-01 9.38318253e-01 -2.65653282e-01
3.40922654e-01 -6.65001512e-01 9.11533833e-02 -1.56436157e+00
2.84186959e-01 -5.05005240e-01 1.32947767e+00 -6.21447027e-01
-1.21458054e+00 2.38653585e-01 3.92775446e-01 -1.00788832e+00
-1.13562554e-01 -5.17415106e-01 5.60030900e-02 4.90458578e-01
-1.68766594e+00 -1.64204836e+00 -2.90699661e-01 1.25532448e+00
-7.89833441e-03 1.68075357e-02 1.04933369e+00 7.12418616e-01
-2.76517242e-01 5.47520280e-01 -2.44522646e-01 2.34863572e-02
4.51707125e-01 -1.12909257e+00 3.28141063e-01 8.65137517e-01
2.78447956e-01 8.69685292e-01 5.98596632e-01 -6.73059404e-01
-2.58719802e+00 -1.17194510e+00 1.08771265e+00 -6.18131518e-01
1.19379866e+00 -9.47699174e-02 -8.06253493e-01 1.23074627e+00
-2.23110050e-01 2.42245466e-01 6.03297174e-01 4.73002106e-01
-1.30120778e+00 -4.44990009e-01 -9.47442889e-01 6.28503263e-01
1.45357716e+00 -1.04776895e+00 -7.98132062e-01 5.50183833e-01
1.20627105e+00 -4.57301795e-01 -1.72950912e+00 5.40520847e-01
4.53741491e-01 -5.30056179e-01 1.29566765e+00 -9.77853060e-01
-2.38869295e-01 -5.82914650e-01 -3.94680411e-01 -7.17335641e-01
-5.16955316e-01 -9.64091599e-01 -1.18560827e+00 1.07336390e+00
-1.35936216e-01 -6.29071951e-01 7.92923510e-01 6.79131150e-01
-8.02504793e-02 -4.32790875e-01 -1.04780340e+00 -1.08317173e+00
-6.10966444e-01 -9.06510353e-01 8.75188410e-01 1.49794388e+00
4.35187787e-01 -8.40846971e-02 -3.46258432e-01 4.22481060e-01
8.20652068e-01 5.74219048e-01 3.90699953e-01 -1.24837899e+00
-1.44270167e-01 -1.74772829e-01 -9.19250131e-01 -5.61793745e-01
4.38433021e-01 -1.15755630e+00 -8.27416778e-01 -1.66823125e+00
2.37916708e-01 -2.09396958e-01 -5.69212556e-01 8.77900362e-01
1.93287835e-01 -2.49997005e-01 -1.24073155e-01 2.04495177e-01
-8.83278131e-01 6.29527211e-01 1.09448862e+00 -4.84672844e-01
1.22784391e-01 -3.58920544e-01 -3.99989307e-01 1.03282586e-01
4.04540122e-01 -2.59393454e-01 -7.71083534e-01 -5.10030627e-01
6.12918735e-01 2.99235582e-01 4.22315717e-01 -5.64577878e-01
6.89791203e-01 -4.29116070e-01 -3.60389978e-01 -6.63475752e-01
6.96758568e-01 -9.96173322e-01 5.98174512e-01 2.05387115e-01
5.55511490e-02 2.05766588e-01 2.99724638e-01 1.12602496e+00
-3.82701665e-01 3.75799358e-01 -2.02680565e-02 8.04974958e-02
-1.34144151e+00 7.97089815e-01 1.10425092e-01 -2.56260633e-01
1.00983083e+00 -1.53144049e-02 -6.60856247e-01 -1.10294022e-01
-8.12604427e-01 4.69563782e-01 1.57411098e-01 8.56068373e-01
7.06170082e-01 -1.72362614e+00 -2.90809363e-01 -3.81266892e-01
7.31018782e-01 -4.73004788e-01 5.34227192e-01 9.90789771e-01
-2.05894083e-01 8.73338163e-01 1.09664164e-01 -3.29248816e-01
-9.56509113e-01 1.37290907e+00 2.83726808e-02 -5.26474953e-01
-6.80648565e-01 4.37788635e-01 -2.69143909e-01 -3.33515048e-01
3.29343319e-01 -4.83792633e-01 7.15828687e-03 8.16089734e-02
5.04701257e-01 8.98710966e-01 4.26122844e-01 -4.56681192e-01
-7.25227296e-01 4.14440274e-01 -4.85715121e-02 3.98016572e-02
1.26389802e+00 -4.00660969e-02 -6.98516250e-01 5.23134828e-01
1.04405928e+00 -2.79157996e-01 -5.22261441e-01 -7.07913339e-01
2.63882101e-01 -7.11340547e-01 3.98953736e-01 -5.17998874e-01
-1.10588467e+00 3.12535048e-01 3.08995306e-01 1.11018792e-01
1.05555892e+00 -3.30256931e-02 9.81724501e-01 5.65678477e-01
9.02424872e-01 -1.09593523e+00 -4.38665599e-02 5.21903157e-01
7.12898374e-01 -7.74261117e-01 1.34322181e-01 -7.83936739e-01
-2.61530042e-01 1.14018011e+00 4.25725877e-01 3.07826668e-01
7.72950947e-01 -1.07584558e-01 -4.10301030e-01 -5.83549738e-01
-1.12413490e+00 -5.13766110e-01 5.02119958e-01 5.56934357e-01
7.81201646e-02 4.26067740e-01 -5.87071717e-01 6.22703731e-01
1.69209644e-01 2.40061849e-01 1.77523687e-01 1.26057398e+00
1.71661079e-01 -1.33543229e+00 -4.23188895e-01 9.61118862e-02
-2.86355406e-01 1.57342210e-01 -4.82560601e-03 6.26568735e-01
-2.75325119e-01 1.05756295e+00 -1.60974964e-01 -7.31314301e-01
5.45642316e-01 2.19356745e-01 7.42715240e-01 -2.45564625e-01
-3.26817632e-01 -4.16620195e-01 4.41287428e-01 -1.35452545e+00
-6.23413324e-01 -5.38551927e-01 -1.40974724e+00 -6.25310957e-01
-1.53604439e-02 2.19222039e-01 5.86906433e-01 6.24387562e-01
6.89952791e-01 7.22534895e-01 4.04267251e-01 -4.41252500e-01
-3.12959015e-01 -3.54164928e-01 -6.90084755e-01 7.00898767e-01
-6.99808449e-02 -1.18523264e+00 -1.99336365e-01 1.40495077e-02] | [8.552275657653809, 7.89233922958374] |
e30166e3-1e50-4877-b64d-44c28008d5f0 | dermatologist-like-explainable-ai-enhances | 2303.12806 | null | https://arxiv.org/abs/2303.12806v1 | https://arxiv.org/pdf/2303.12806v1.pdf | Dermatologist-like explainable AI enhances trust and confidence in diagnosing melanoma | Although artificial intelligence (AI) systems have been shown to improve the accuracy of initial melanoma diagnosis, the lack of transparency in how these systems identify melanoma poses severe obstacles to user acceptance. Explainable artificial intelligence (XAI) methods can help to increase transparency, but most XAI methods are unable to produce precisely located domain-specific explanations, making the explanations difficult to interpret. Moreover, the impact of XAI methods on dermatologists has not yet been evaluated. Extending on two existing classifiers, we developed an XAI system that produces text and region based explanations that are easily interpretable by dermatologists alongside its differential diagnoses of melanomas and nevi. To evaluate this system, we conducted a three-part reader study to assess its impact on clinicians' diagnostic accuracy, confidence, and trust in the XAI-support. We showed that our XAI's explanations were highly aligned with clinicians' explanations and that both the clinicians' trust in the support system and their confidence in their diagnoses were significantly increased when using our XAI compared to using a conventional AI system. The clinicians' diagnostic accuracy was numerically, albeit not significantly, increased. This work demonstrates that clinicians are willing to adopt such an XAI system, motivating their future use in the clinic. | ['Titus J. Brinker', 'Eva Krieghoff-Henning', 'Stefan Fröhling', 'Kamran Ghoreschi', 'Jochen S. Utikal', 'Bastian Schilling', 'Matthias Goebeler', 'Konstantin Drexler', 'Sebastian Haferkamp', 'Michael Erdmann', 'Markus V. Heppt', 'Frank Friedrich Gellrich', 'Wiebke Sondermann', 'Sören Korsing', 'Gabriela Poch', 'Mar Llamas Velasco', 'Raymond L. Barnhill', 'İrem Özdemir', 'Sergei Sarap', 'Sandra Peternel', 'Tanya Foreman', 'Linda Alhajwan', 'Jovana Majstorovic', 'Iva Crnaric', 'Emmanouil Chousakos', 'Sebastian Podlipnik', 'Cristian Navarrete-Dechent', 'Philipp Tschandl', 'Harald Kittler', 'Christoph Wies', 'Carina Nogueira Garcia', 'Tabea-Clara Bucher', 'Sarah Hobelsberger', 'Katja Hauser', 'Tirtha Chanda'] | 2023-03-17 | null | null | null | null | ['melanoma-diagnosis'] | ['computer-vision'] | [ 2.62591928e-01 1.12567246e+00 -4.39959794e-01 -6.54226780e-01
-4.87847954e-01 -5.56184292e-01 3.51926476e-01 5.18860102e-01
-1.31773710e-01 8.69913518e-01 5.39159715e-01 -9.26379800e-01
-6.49459779e-01 -3.07896376e-01 -2.31679231e-01 -3.14954966e-01
5.22285938e-01 6.51966631e-01 -2.68519998e-01 2.72413224e-01
3.84114325e-01 4.86159682e-01 -1.18218338e+00 7.69237399e-01
1.32945573e+00 6.21856213e-01 -5.12738347e-01 7.53053069e-01
-1.56205162e-01 9.89457190e-01 -8.91745865e-01 -6.00156009e-01
-4.43780795e-02 -2.94413775e-01 -8.25625002e-01 2.01950744e-01
3.82385612e-01 -6.33967578e-01 2.25096807e-01 3.39774519e-01
5.91215901e-02 -7.65908480e-01 8.12261641e-01 -1.54039037e+00
-9.53101337e-01 6.97032809e-01 -1.57250583e-01 -9.44103375e-02
6.07338428e-01 4.62985605e-01 5.43402731e-01 -3.65978122e-01
6.06202841e-01 9.20856535e-01 8.85016024e-01 7.94205844e-01
-9.29953337e-01 -6.11317515e-01 1.77652955e-01 -8.26598182e-02
-1.04193676e+00 -2.76393652e-01 5.49886711e-02 -4.07598913e-01
9.24541652e-01 9.70391273e-01 1.25784385e+00 7.74445653e-01
6.62799060e-01 3.74953568e-01 1.30677247e+00 -5.66365957e-01
3.89731497e-01 9.79400039e-01 5.24857163e-01 7.54274130e-01
8.22136879e-01 2.75897700e-03 -6.42841697e-01 -5.36712468e-01
6.02936625e-01 -6.54751509e-02 -3.25496078e-01 2.22564921e-01
-1.16065419e+00 8.08335662e-01 6.11222565e-01 9.31615680e-02
-2.97270715e-01 -6.82536736e-02 5.85448071e-02 1.70524403e-01
4.16373044e-01 6.55444264e-01 -1.26340777e-01 -1.07817225e-01
-5.90180397e-01 -1.82357728e-01 1.04696488e+00 5.73341727e-01
-2.13622570e-01 -3.59192729e-01 -3.57583016e-02 3.72179627e-01
4.67754304e-01 4.33807194e-01 2.78451115e-01 -9.81542706e-01
-3.17027926e-01 1.15922940e+00 1.77517995e-01 -9.34613466e-01
-5.03594398e-01 -5.95722318e-01 -6.64803982e-01 6.91295743e-01
6.25202060e-01 -1.12250589e-01 -9.16878760e-01 1.04360342e+00
4.52193767e-02 -5.79528093e-01 2.78861165e-01 9.92073655e-01
8.69315207e-01 -4.51243706e-02 2.10060954e-01 1.71697661e-01
1.39942086e+00 -7.83092737e-01 -7.18018830e-01 -2.98117548e-01
9.54461396e-01 -6.95591450e-01 9.82556581e-01 6.78056896e-01
-1.04723787e+00 -4.54225950e-02 -1.05126107e+00 1.81307375e-01
-2.45836452e-01 2.50208050e-01 1.05236053e+00 1.10420918e+00
-1.20381892e+00 2.04674795e-01 -6.13781571e-01 -9.22286570e-01
5.98214269e-01 5.22709489e-01 -6.82827473e-01 3.80879454e-02
-6.48382485e-01 1.37393653e+00 -1.64757520e-01 -1.31697310e-02
-1.69340909e-01 -8.82027447e-01 -6.31708741e-01 -1.26009792e-01
-6.45042956e-02 -1.21984327e+00 1.43697166e+00 -1.26497877e+00
-9.96315062e-01 8.72719526e-01 -3.06585431e-01 -4.18498814e-01
5.45469642e-01 1.18848160e-01 -5.26785374e-01 2.71650136e-01
9.80194584e-02 1.03782737e+00 3.13583583e-01 -1.24881232e+00
-8.25272977e-01 -3.49566549e-01 1.68319374e-01 2.51143962e-01
-3.42435360e-01 -2.70290166e-01 1.83373824e-01 -1.51832953e-01
-2.10566632e-02 -9.04498637e-01 -2.74405003e-01 9.88526225e-01
-3.55524391e-01 -3.82975824e-02 5.82246423e-01 -6.95829868e-01
9.84169722e-01 -2.07970142e+00 -7.08848894e-01 4.49316323e-01
7.73550034e-01 3.29959333e-01 2.79352099e-01 2.73720086e-01
-9.29386169e-02 6.94840968e-01 1.80477098e-01 1.79454789e-01
-1.62107989e-01 4.93830860e-01 1.23976126e-01 1.92097962e-01
4.10638809e-01 6.81568980e-01 -8.16810966e-01 -6.19943142e-01
3.52872431e-01 7.33151019e-01 -4.60310817e-01 -6.33645952e-02
2.03369677e-01 2.82205522e-01 -6.15365021e-02 7.18333304e-01
3.53177279e-01 -7.58536756e-01 1.92807958e-01 2.05973536e-01
-4.61057723e-02 1.47384763e-01 -6.57850325e-01 8.85047019e-01
-2.94825017e-01 6.74555480e-01 -1.40491491e-02 2.67855246e-02
8.06950152e-01 5.67924440e-01 2.16933265e-01 -2.52769947e-01
-1.51076451e-01 4.38882530e-01 5.39947510e-01 -6.22975707e-01
3.93599644e-02 -1.39411092e-01 4.86378908e-01 7.11955547e-01
-8.29306901e-01 -4.01210129e-01 -3.60731214e-01 4.87776130e-01
1.10540581e+00 -5.11459708e-01 7.29417741e-01 -2.78349429e-01
1.20861918e-01 7.70528853e-01 1.46962881e-01 9.41865802e-01
-2.55785078e-01 6.84647620e-01 4.60834414e-01 -7.39670217e-01
-7.16158628e-01 -7.38758683e-01 -3.92339259e-01 1.30930096e-01
-7.52003491e-02 -3.67057353e-01 -8.07347476e-01 -8.13227773e-01
2.20929742e-01 1.36300838e+00 -8.91045749e-01 -1.55238211e-01
5.15106559e-01 -3.41669828e-01 2.66465634e-01 7.59126246e-01
1.90423742e-01 -6.09478474e-01 -1.26279581e+00 3.62927765e-02
5.81354536e-02 -4.78149176e-01 -2.70249993e-01 -7.63172507e-02
-7.62099504e-01 -1.20927656e+00 -5.61720252e-01 -1.51929736e-01
1.25841808e+00 2.83519089e-01 9.79991376e-01 7.21853971e-01
-6.59121931e-01 7.30713010e-01 -3.17533493e-01 -1.15249777e+00
-8.34661663e-01 -3.57038736e-01 -2.26466879e-01 -6.10624433e-01
5.95032334e-01 2.35415384e-01 -8.37194204e-01 4.04971570e-01
-9.14129198e-01 5.91826499e-01 8.84367466e-01 7.99678326e-01
1.61286011e-01 -3.25445980e-01 3.06304574e-01 -1.42888486e+00
8.98275733e-01 -2.48809054e-01 1.92139402e-01 2.82201827e-01
-1.17243195e+00 -1.95726529e-01 1.39128000e-01 -2.79701173e-01
-1.02099824e+00 -2.91219484e-02 1.49245322e-01 4.88541722e-01
-2.10266322e-01 6.35770857e-01 4.89269108e-01 -4.52504039e-01
1.04746437e+00 -5.27693331e-01 7.57006168e-01 2.83893734e-01
-2.33209413e-02 1.12717593e+00 3.69365722e-01 1.29384518e-01
4.36693817e-01 4.61000949e-01 -3.41283321e-01 -5.03865063e-01
-7.78842509e-01 -4.93543893e-02 -3.19430590e-01 -3.39147627e-01
5.60205996e-01 -6.54797137e-01 -8.64533901e-01 -1.05207108e-01
-1.06262946e+00 -2.15906769e-01 -4.30093408e-01 6.40878320e-01
-7.06370473e-02 1.52065724e-01 -1.83959052e-01 -9.86203611e-01
-5.94948232e-01 -1.19280159e+00 8.01213801e-01 5.13200343e-01
-1.53057885e+00 -1.12356389e+00 -3.20283055e-01 7.17035055e-01
7.11140692e-01 2.37488180e-01 1.20204806e+00 -7.67572582e-01
-2.07384139e-01 -5.37156880e-01 -5.84395826e-01 -1.75371408e-01
4.98634338e-01 5.90566754e-01 -9.01501834e-01 7.75824860e-02
-2.27412030e-01 -5.00771739e-02 1.15516812e-01 5.64649463e-01
8.56211782e-01 -6.66200101e-01 -7.77673304e-01 1.83480904e-01
9.61196959e-01 5.35645962e-01 4.01314795e-01 3.51690531e-01
1.00270301e-01 8.69339406e-01 8.16514611e-01 3.17633599e-01
3.60585690e-01 3.59844416e-01 3.88838202e-01 -7.04182804e-01
-2.04609960e-01 -2.32460648e-02 -1.85484543e-01 8.62488523e-02
-8.52950513e-02 -1.18155278e-01 -1.22758496e+00 4.35280323e-01
-1.73769438e+00 -2.83364058e-01 -4.46252882e-01 1.99646068e+00
5.37372947e-01 2.88844258e-01 -1.80573210e-01 2.06190854e-01
3.53546977e-01 -8.76538396e-01 -7.70039380e-01 -1.07849407e+00
2.48190239e-01 -1.59294993e-01 3.26458544e-01 7.69523144e-01
-2.85796404e-01 1.76280543e-01 7.74178314e+00 -6.75750747e-02
-9.43018198e-01 -1.89218625e-01 9.73333180e-01 -1.10961266e-01
-8.44444811e-01 5.17954156e-02 -3.85201126e-01 2.74021626e-01
9.31070209e-01 -4.31736350e-01 -1.26499727e-01 7.61931539e-01
3.65425974e-01 -5.64526439e-01 -1.24606097e+00 6.90290689e-01
3.21020484e-02 -1.64575005e+00 4.61740308e-02 3.24282020e-01
2.59945840e-01 -8.66761625e-01 3.41461152e-01 -3.07344258e-01
4.82409388e-01 -1.50517011e+00 5.45525730e-01 7.58223891e-01
1.03383565e+00 -5.83834648e-01 1.18115318e+00 -1.50761262e-01
-2.10121393e-01 -3.72077301e-02 -2.56883968e-02 -3.91924948e-01
-3.48248571e-01 5.51310658e-01 -1.96824086e+00 1.72561496e-01
5.42568803e-01 1.92048773e-01 -8.73664320e-01 1.04156804e+00
-2.21169665e-01 6.65191174e-01 -2.37127051e-01 -4.01752740e-01
1.65865302e-01 1.79012284e-01 1.51810214e-01 1.01290441e+00
9.80170891e-02 4.82546359e-01 -5.35622358e-01 6.39148831e-01
5.88075697e-01 2.10303124e-02 -7.23189116e-01 -1.71729103e-01
4.00418639e-01 1.09412479e+00 -6.62203074e-01 -5.00029683e-01
-3.30409974e-01 8.35436046e-01 -1.89403832e-01 1.00115374e-01
-5.30434012e-01 -1.32478207e-01 5.81901312e-01 5.60205460e-01
-6.57953203e-01 4.66427296e-01 -1.08909690e+00 -5.46594381e-01
-1.19431414e-01 -1.22662735e+00 6.35832191e-01 -1.37973547e+00
-1.04905856e+00 7.60839880e-01 -2.55290359e-01 -1.14679027e+00
-5.15860438e-01 -7.64246881e-01 -4.58546758e-01 9.57871377e-01
-1.10263455e+00 -1.33020627e+00 -7.98864782e-01 1.61133539e-02
1.29208714e-01 -4.35234904e-02 1.19813240e+00 -6.35469556e-01
-2.64824718e-01 8.22239220e-01 -2.26197466e-01 -2.10515127e-01
1.16977000e+00 -1.25019646e+00 5.50909154e-02 1.95134014e-01
-2.59876370e-01 9.73840296e-01 6.75634325e-01 -7.89834976e-01
-1.07425618e+00 -5.61360419e-01 9.96114612e-01 -8.83127093e-01
3.46917421e-01 3.37785065e-01 -7.82861531e-01 7.00862646e-01
3.71950626e-01 -4.60730076e-01 1.33033895e+00 2.87496775e-01
-1.89596012e-01 -3.17296945e-02 -1.64558792e+00 8.64820063e-01
6.78465545e-01 -2.27670938e-01 -3.82637173e-01 6.67554736e-01
3.65465432e-01 -2.67396063e-01 -8.87610197e-01 2.80540496e-01
8.43016088e-01 -9.93350387e-01 6.08244777e-01 -6.43560886e-01
7.37595499e-01 -8.36010724e-02 4.24388826e-01 -1.26409161e+00
-4.91703808e-01 -1.99252963e-01 5.44061005e-01 8.01280022e-01
9.40069675e-01 -1.05931759e+00 9.49030578e-01 1.87955332e+00
-7.76453270e-03 -9.00298238e-01 -5.24493098e-01 -2.06894185e-02
-3.27652425e-01 -1.56029135e-01 3.95855337e-01 8.53514850e-01
6.93211257e-01 -1.13432728e-01 2.18354791e-01 4.03340161e-01
3.47039491e-01 -2.08825245e-01 5.08689046e-01 -1.22552264e+00
-2.48574048e-01 -3.31772059e-01 -7.03905642e-01 1.06765412e-01
-5.22339940e-01 -7.23055661e-01 -4.96083319e-01 -2.14205027e+00
5.29965341e-01 -5.74244797e-01 1.69857308e-01 1.03048718e+00
-2.75803149e-01 3.77194397e-03 3.36125344e-02 3.70214134e-01
7.88279474e-02 -5.01575649e-01 9.46385682e-01 -2.33644515e-01
-8.01038519e-02 -6.43457696e-02 -1.72306120e+00 6.84196651e-01
6.80024803e-01 -3.47635746e-01 -6.27344310e-01 -5.33727944e-01
9.21785906e-02 1.33632198e-01 4.43105459e-01 -9.70400631e-01
5.69996655e-01 -2.21209407e-01 9.48949277e-01 -3.31675440e-01
2.55671114e-01 -8.11412692e-01 7.17394829e-01 1.07854986e+00
-4.56475049e-01 -4.81199138e-02 6.06999576e-01 1.90966740e-01
-2.36064028e-02 -3.99394453e-01 4.35497493e-01 -5.45526445e-02
-1.76110581e-01 -3.81700575e-01 -7.48379707e-01 -6.51886344e-01
1.22021914e+00 -6.99946225e-01 -6.23090923e-01 -7.97012448e-01
-6.49454594e-01 8.55567530e-02 1.00883794e+00 -1.22999638e-01
8.11725140e-01 -7.06355512e-01 -7.05499709e-01 9.33644027e-02
3.54921758e-01 -5.49505390e-02 1.11502789e-01 9.46642160e-01
-9.53956842e-01 6.61080062e-01 -3.83228213e-01 -5.49063027e-01
-1.63897610e+00 3.36535901e-01 3.41961384e-01 -3.33710713e-03
-6.47129178e-01 8.97064090e-01 1.28444955e-01 -2.50832200e-01
4.20068443e-01 -4.01434124e-01 6.39172941e-02 -4.28628623e-01
7.95576453e-01 5.81623148e-03 -6.95860162e-02 1.85889080e-01
-5.52746296e-01 2.45485783e-01 -7.13297904e-01 -3.82307380e-01
9.20003414e-01 7.01190680e-02 6.21939078e-03 2.15509787e-01
1.69289604e-01 1.34581402e-01 -7.56378651e-01 4.08015847e-01
-2.66653627e-01 -7.96473324e-01 -2.02668577e-01 -1.83929586e+00
-6.85453415e-01 6.80497110e-01 4.98310357e-01 2.10956082e-01
1.03518462e+00 -1.97100103e-01 1.02533035e-01 1.07818320e-01
8.48321766e-02 -5.75069606e-01 -1.63062260e-01 -5.65482855e-01
9.11396027e-01 -1.32709444e+00 3.17529738e-01 -8.80441785e-01
-1.24817681e+00 1.23528063e+00 6.57854199e-01 6.70688570e-01
2.62621909e-01 4.39292759e-01 8.82418931e-01 -3.31864327e-01
-1.06293285e+00 4.66014743e-01 1.13777705e-01 1.04689574e+00
7.41668940e-01 3.42397928e-01 -5.04056990e-01 4.76881713e-01
-2.47783497e-01 4.71867144e-01 9.71318185e-01 7.65521646e-01
-1.41577050e-01 -9.18056548e-01 -7.08612680e-01 8.58508050e-01
-4.19920906e-02 1.52292013e-01 -1.27727640e+00 9.85240221e-01
-7.16045797e-02 1.28575051e+00 1.01649072e-02 -2.55695969e-01
1.15385465e-01 -4.55436297e-02 1.71726391e-01 -5.89834929e-01
-8.16533029e-01 -4.12619978e-01 5.68532228e-01 -3.48537803e-01
-6.68649795e-03 -5.70308089e-01 -1.29318082e+00 -4.74960268e-01
-2.63133466e-01 2.99307168e-01 8.98220122e-01 7.07801759e-01
5.86442351e-01 5.63197672e-01 2.27024388e-02 3.47457826e-01
-3.10555667e-01 -4.63263273e-01 -2.67075628e-01 3.07113022e-01
1.75666869e-01 -2.49415249e-01 -3.18332106e-01 -1.04119867e-01] | [8.653736114501953, 5.8241868019104] |
02d7bc30-25f9-43d7-b354-e91ec6815a94 | dude-deep-unsigned-distance-embeddings-for-hi | 2011.0257 | null | https://arxiv.org/abs/2011.02570v2 | https://arxiv.org/pdf/2011.02570v2.pdf | DUDE: Deep Unsigned Distance Embeddings for Hi-Fidelity Representation of Complex 3D Surfaces | High fidelity representation of shapes with arbitrary topology is an important problem for a variety of vision and graphics applications. Owing to their limited resolution, classical discrete shape representations using point clouds, voxels and meshes produce low quality results when used in these applications. Several implicit 3D shape representation approaches using deep neural networks have been proposed leading to significant improvements in both quality of representations as well as the impact on downstream applications. However, these methods can only be used to represent topologically closed shapes which greatly limits the class of shapes that they can represent. As a consequence, they also often require clean, watertight meshes for training. In this work, we propose DUDE - a Deep Unsigned Distance Embedding method which alleviates both of these shortcomings. DUDE is a disentangled shape representation that utilizes an unsigned distance field (uDF) to represent proximity to a surface, and a normal vector field (nVF) to represent surface orientation. We show that a combination of these two (uDF+nVF) can be used to learn high fidelity representations for arbitrary open/closed shapes. As opposed to prior work such as DeepSDF, our shape representations can be directly learnt from noisy triangle soups, and do not need watertight meshes. Additionally, we propose novel algorithms for extracting and rendering iso-surfaces from the learnt representations. We validate DUDE on benchmark 3D datasets and demonstrate that it produces significant improvements over the state of the art. | ['Maneesh Singh', 'Laszlo Jeni', 'Aurobrata Ghosh', 'Sarthak Sharma', 'Rahul Venkatesh'] | 2020-11-04 | null | null | null | null | ['3d-shape-representation'] | ['computer-vision'] | [ 1.16355792e-01 -4.56623593e-03 1.86547026e-01 -1.50972798e-01
-5.24674058e-01 -5.95172346e-01 8.14950049e-01 2.49575928e-01
1.16111383e-01 5.15988648e-01 5.90991601e-02 -2.14697391e-01
-2.54136082e-02 -1.41163838e+00 -8.04715633e-01 -6.02068245e-01
1.04172416e-01 6.42081618e-01 8.17628726e-02 -3.55781257e-01
2.41273060e-01 9.96084690e-01 -1.65796781e+00 1.43653676e-01
8.38845015e-01 1.05765915e+00 -2.24098697e-01 3.09789807e-01
-4.15623605e-01 -1.91223145e-01 -3.83412331e-01 -1.71396434e-01
4.01531518e-01 1.88433945e-01 -4.45116192e-01 -1.88988075e-01
8.44650328e-01 -4.48539734e-01 -5.12011051e-01 7.91731179e-01
4.82168257e-01 1.72508016e-01 1.07016897e+00 -1.01068211e+00
-9.34867859e-01 -2.92128235e-01 -7.12635219e-01 -2.80169100e-01
2.83624917e-01 -8.16870481e-02 8.93297672e-01 -1.23068202e+00
6.36282742e-01 1.40166438e+00 8.99025798e-01 4.01771814e-01
-1.62101305e+00 -7.08425581e-01 4.95469756e-03 -3.82961214e-01
-1.34148574e+00 -1.75685748e-01 1.04930830e+00 -5.26287854e-01
9.99745846e-01 3.99818689e-01 7.91208267e-01 9.31541741e-01
8.91009644e-02 4.73339081e-01 9.74587321e-01 -1.37299433e-01
2.05018997e-01 -2.63232797e-01 -1.38429150e-01 7.15183854e-01
5.20646989e-01 1.60481274e-01 -8.89207572e-02 -5.49770176e-01
1.49922478e+00 6.31655455e-02 -3.43685240e-01 -7.81283379e-01
-1.09548807e+00 9.20090556e-01 7.84896970e-01 1.59541383e-01
-2.51870632e-01 4.52053159e-01 1.53453514e-01 7.64085799e-02
7.74586022e-01 3.87384146e-01 -1.56763330e-01 -8.93972889e-02
-6.33239627e-01 4.70722675e-01 7.22428858e-01 7.89786577e-01
7.99901009e-01 2.01759383e-01 2.97549386e-02 8.88562381e-01
3.57547462e-01 5.94800770e-01 -1.04447052e-01 -8.62573385e-01
2.58343607e-01 7.74714351e-01 1.67461634e-01 -1.55659735e+00
-2.35049501e-01 -2.43774161e-01 -1.11810267e+00 7.37519264e-01
1.73615769e-01 3.09553593e-01 -1.09802580e+00 1.53700650e+00
4.75475729e-01 5.00369310e-01 -2.57200897e-01 8.75218987e-01
1.12319684e+00 7.79335201e-01 -3.80573303e-01 3.89915109e-01
1.10439563e+00 -3.53930712e-01 -4.71422762e-01 1.62665933e-01
5.36931932e-01 -6.83602631e-01 9.45238590e-01 2.47742802e-01
-1.10274005e+00 -3.70359570e-01 -1.16549408e+00 -5.86814165e-01
-4.22809720e-01 -6.81377873e-02 6.92355812e-01 3.66905063e-01
-1.09206271e+00 8.56312394e-01 -9.76147532e-01 1.06606975e-01
7.53079414e-01 3.93524915e-01 -4.03434902e-01 -2.04039678e-01
-8.11258197e-01 6.44811869e-01 -2.46726811e-01 3.64514105e-02
-4.05658633e-01 -1.00226927e+00 -1.06424522e+00 1.02508724e-01
-1.16803147e-01 -8.53960276e-01 7.29804516e-01 -3.05861562e-01
-1.40258670e+00 7.86235988e-01 -1.49663419e-01 7.32901460e-03
4.62419361e-01 5.33878393e-02 -2.18079798e-02 1.58224385e-02
-1.35002732e-01 5.65030754e-01 7.57501602e-01 -1.55295789e+00
-3.22799236e-02 -4.50422078e-01 2.01086015e-01 6.39090538e-02
-1.05612889e-01 -7.30628729e-01 -9.18645337e-02 -9.36126709e-01
5.27051747e-01 -7.89063811e-01 -1.93603292e-01 9.18905497e-01
-1.29395440e-01 -3.05223197e-01 1.24451482e+00 -5.02953827e-01
8.41243923e-01 -2.24176812e+00 3.36388886e-01 3.50472599e-01
6.12214208e-01 3.40366215e-01 -1.86590835e-01 3.83662403e-01
1.19906113e-01 4.55753326e-01 -5.04041612e-01 -4.70536053e-01
1.52925387e-01 5.46580851e-01 -4.12953049e-01 5.91156721e-01
6.23854995e-01 9.21884894e-01 -1.00443208e+00 -4.05642912e-02
3.45262021e-01 1.29499364e+00 -6.76233232e-01 3.20284329e-02
-2.62706190e-01 3.25945765e-01 -5.05802214e-01 3.91571254e-01
1.05673754e+00 -1.02361023e-01 -1.51476756e-01 -4.47527200e-01
-3.75014432e-02 4.37473297e-01 -1.15358007e+00 1.69888580e+00
-7.21931159e-01 5.51661909e-01 1.44063219e-01 -8.97679210e-01
1.32351077e+00 3.27904314e-01 4.34394538e-01 -6.01084292e-01
2.62410007e-02 3.73484582e-01 -3.01576465e-01 2.38262629e-03
3.68582278e-01 -2.03772828e-01 1.85316265e-01 4.26452696e-01
-3.36313426e-01 -6.63270772e-01 -2.82748401e-01 7.64157847e-02
9.09722626e-01 2.57427812e-01 -4.53173853e-02 -2.57292837e-01
2.15030149e-01 -3.14201176e-01 5.48589528e-01 1.35904059e-01
2.95593351e-01 1.02655542e+00 3.91279995e-01 -6.00056410e-01
-1.13318884e+00 -1.35427594e+00 -4.56916541e-01 3.81195217e-01
3.18871528e-01 -3.67093623e-01 -2.68166184e-01 -2.67117113e-01
5.25086582e-01 5.52859008e-01 -6.06554627e-01 -1.19388618e-01
-8.55050802e-01 -2.81812310e-01 3.29095572e-01 7.99300194e-01
1.05276652e-01 -5.71120977e-01 -4.58762854e-01 1.37950495e-01
3.17582488e-01 -8.44301999e-01 -2.96971440e-01 -2.48502582e-01
-1.16257131e+00 -9.95013356e-01 -6.96927905e-01 -7.45419681e-01
8.63562226e-01 4.78031069e-01 1.14479494e+00 3.59580159e-01
-2.26579234e-01 2.84019053e-01 -9.88621637e-02 -2.86615252e-01
-2.51750439e-01 -3.20669323e-01 -4.29791957e-02 -5.07420450e-02
9.20643955e-02 -1.10341978e+00 -6.92179561e-01 1.67201295e-01
-1.01242435e+00 2.21038118e-01 2.77098417e-01 9.02085185e-01
9.86046970e-01 -3.14898193e-01 5.20271719e-01 -7.23413408e-01
5.63831449e-01 -4.11933124e-01 -6.50209844e-01 -9.79208425e-02
-3.17324340e-01 2.62387276e-01 5.38827419e-01 -4.03271735e-01
-6.34352565e-01 -1.40483707e-01 -3.27317417e-01 -8.18484962e-01
-3.25789079e-02 3.91135395e-01 6.89923167e-02 -3.64434928e-01
5.57905972e-01 -1.66731700e-02 3.36097986e-01 -8.23823273e-01
4.47926044e-01 4.35299069e-01 2.68728554e-01 -8.41911674e-01
9.24331486e-01 8.02440047e-01 3.81294042e-01 -1.04567671e+00
-2.67501682e-01 -7.56718516e-02 -6.49395704e-01 2.13527247e-01
4.93448406e-01 -7.06255257e-01 -5.45242131e-01 1.99697986e-01
-1.32916260e+00 -6.94881901e-02 -3.51815522e-01 1.42947227e-01
-5.79245269e-01 4.77912426e-01 -4.92478251e-01 -7.43257165e-01
-3.71614873e-01 -1.14092338e+00 1.42119288e+00 -1.29985780e-01
-1.85171887e-01 -1.11319613e+00 -8.89306888e-02 -1.09572709e-01
5.98179579e-01 1.09237707e+00 1.31771958e+00 3.23798247e-02
-6.82753444e-01 -2.52004713e-01 -4.00336385e-01 1.41896665e-01
3.19998890e-01 1.23727635e-01 -8.15358996e-01 -4.83534336e-01
-2.81334847e-01 -1.98635191e-01 7.33740449e-01 1.64878488e-01
1.30042577e+00 -2.65346497e-01 -3.96336019e-01 9.16703224e-01
1.36819041e+00 -1.63561299e-01 6.06332541e-01 -5.30239232e-02
9.69081759e-01 3.14422309e-01 1.26838535e-01 3.73870254e-01
2.63627350e-01 7.75146902e-01 6.38467371e-01 -2.19663575e-01
-3.43721062e-01 -4.01323318e-01 -1.98834818e-02 8.74341607e-01
-3.26290190e-01 3.19271535e-02 -8.80941451e-01 4.47048187e-01
-1.59238744e+00 -5.22858381e-01 -4.33138460e-01 2.34707975e+00
6.14949107e-01 -3.58949304e-02 -2.02300295e-01 3.48895580e-01
2.98299700e-01 3.12348455e-01 -5.83397269e-01 -5.80849707e-01
-1.57997057e-01 6.67977452e-01 1.89361367e-02 5.63773930e-01
-8.25251102e-01 4.81668115e-01 5.59602308e+00 6.15803778e-01
-1.35878384e+00 -2.64756940e-03 3.21979046e-01 -3.55357490e-02
-9.99269426e-01 -2.12477013e-01 -2.58649558e-01 1.51233792e-01
3.95005763e-01 -1.83862403e-01 2.20938444e-01 5.49825847e-01
-1.13108673e-03 3.31713647e-01 -1.27592981e+00 1.27680910e+00
-1.43280998e-01 -1.59933782e+00 4.37498957e-01 3.02614808e-01
7.11745262e-01 -5.21975867e-02 2.06430405e-01 4.82619368e-02
2.24234872e-02 -1.44872713e+00 6.07800484e-01 4.05946225e-01
1.31065452e+00 -7.38986135e-01 3.17084551e-01 3.14434141e-01
-1.41602075e+00 4.16576564e-01 -6.43002570e-01 -6.98077753e-02
8.73062909e-02 7.37288177e-01 -5.14768422e-01 6.59226954e-01
3.64034563e-01 8.61451924e-01 -6.46163076e-02 8.80083680e-01
-4.65809517e-02 3.05359662e-01 -7.41761625e-01 1.29506126e-01
2.78081536e-01 -5.08055627e-01 7.20944166e-01 8.26117337e-01
5.21156132e-01 2.89318651e-01 1.60643458e-01 1.34652650e+00
-1.76943749e-01 -1.15886956e-01 -1.03576708e+00 -2.70877667e-02
5.18156230e-01 9.57303643e-01 -5.07563889e-01 -1.19827129e-01
-3.67626816e-01 7.95599878e-01 4.48212653e-01 4.14633155e-01
-4.37923729e-01 -3.72067362e-01 1.13642514e+00 5.69227457e-01
3.56342673e-01 -6.56470180e-01 -5.28168499e-01 -1.10160255e+00
2.33914018e-01 -4.53837097e-01 -1.53285444e-01 -5.71575761e-01
-1.40943682e+00 4.83013064e-01 -1.22064240e-01 -1.35663211e+00
1.25665724e-01 -7.63395667e-01 -5.87923288e-01 1.14604557e+00
-1.67762339e+00 -1.16126025e+00 -3.78671676e-01 1.35531738e-01
1.68379351e-01 2.86655337e-01 1.12158263e+00 3.66118968e-01
-1.51744753e-01 5.01320839e-01 2.41633311e-01 1.60990387e-01
1.40576214e-01 -1.23458219e+00 7.85353899e-01 2.55618960e-01
2.45332465e-01 5.88919997e-01 3.30313087e-01 -5.79383075e-01
-1.82471323e+00 -1.13878286e+00 4.15583163e-01 -4.15229708e-01
1.53915048e-01 -5.34706235e-01 -1.41006684e+00 4.65900153e-01
-3.31779003e-01 6.35310590e-01 3.92785460e-01 -7.22157359e-02
-6.42575979e-01 1.32159993e-01 -1.27245557e+00 5.61592519e-01
1.19955540e+00 -4.41313833e-01 -4.95455652e-01 1.42653927e-01
6.97880328e-01 -6.25310838e-01 -1.17868769e+00 5.70549190e-01
5.47166109e-01 -8.85673106e-01 1.28560901e+00 -4.84619856e-01
4.42133099e-01 -3.01795810e-01 -2.19020426e-01 -1.42825663e+00
-4.91801858e-01 -4.23051715e-01 -4.66867745e-01 9.38710570e-01
1.33423628e-02 -7.99891412e-01 7.21318960e-01 5.52282929e-01
-1.97059870e-01 -1.43280387e+00 -1.16565835e+00 -7.69454956e-01
5.77134013e-01 -1.63311854e-01 9.74910140e-01 1.08904314e+00
-5.14101207e-01 8.58768001e-02 9.00736824e-02 1.48454562e-01
6.60734773e-01 4.56460804e-01 6.66347623e-01 -1.70386231e+00
1.10830903e-01 -5.86679876e-01 -7.42690504e-01 -1.28074133e+00
9.42759961e-02 -1.19998193e+00 -2.36188367e-01 -1.87291980e+00
-4.22907680e-01 -1.10295844e+00 2.77806483e-02 3.13439220e-01
1.89608902e-01 4.35822874e-01 -3.13459113e-02 -7.81308766e-03
3.56860816e-01 1.01005185e+00 1.50849009e+00 -2.46347219e-01
-7.36100376e-02 -2.81069785e-01 -4.36963975e-01 7.92958438e-01
5.64428508e-01 -2.94030368e-01 -3.33641469e-01 -9.03735399e-01
3.51165116e-01 -7.27172345e-02 6.39716148e-01 -6.70486331e-01
-2.12211281e-01 -6.48157159e-03 4.74371821e-01 -6.49715960e-01
8.09303582e-01 -7.90961206e-01 2.57610470e-01 2.03730196e-01
1.68348104e-01 1.11228399e-01 3.83158356e-01 7.12765157e-01
-1.55948222e-01 7.27770925e-02 7.20252812e-01 2.55862512e-02
-1.36303365e-01 6.50034308e-01 1.60848647e-01 -7.34988898e-02
8.55584562e-01 -4.86985803e-01 -2.01707318e-01 -3.02011669e-01
-3.73240918e-01 1.30950492e-02 8.60912263e-01 4.19597805e-01
1.09585166e+00 -1.87936997e+00 -7.35630214e-01 5.92395961e-01
-1.52335390e-02 7.29838789e-01 1.44065134e-02 5.41903317e-01
-7.87649095e-01 9.85230356e-02 -1.38320893e-01 -7.41693020e-01
-8.98667216e-01 3.08750242e-01 2.72573292e-01 9.87380370e-02
-1.13226128e+00 7.26719201e-01 4.21112388e-01 -5.64291120e-01
1.68317519e-02 -6.64058864e-01 6.10745028e-02 -1.73680961e-01
3.06057990e-01 3.50932479e-01 2.66120195e-01 -8.48353207e-01
-1.47540510e-01 1.04008877e+00 1.94805965e-01 2.37999320e-01
1.61318374e+00 3.78547490e-01 -1.85767978e-01 4.06248599e-01
1.52020037e+00 3.94444764e-02 -1.25619948e+00 -2.20616773e-01
-4.43428665e-01 -9.22309577e-01 2.08807424e-01 -2.58838564e-01
-1.18918097e+00 1.37138379e+00 3.89483005e-01 1.74339116e-01
6.84789479e-01 -2.00077146e-01 1.13198686e+00 1.36474773e-01
5.82920194e-01 -4.81768906e-01 -9.96000413e-03 4.82625186e-01
1.36939180e+00 -9.89473522e-01 1.02439351e-01 -8.04310858e-01
-3.04684378e-02 1.25873208e+00 3.01241875e-01 -5.67533553e-01
8.56739938e-01 3.10505450e-01 -2.39164174e-01 -3.62160206e-01
-5.23349464e-01 1.26208976e-01 6.23525560e-01 7.51031399e-01
4.92754847e-01 1.51345849e-01 -1.54078946e-01 2.31939107e-01
-1.85532957e-01 -2.92553008e-01 3.98971647e-01 9.96762872e-01
-1.96283698e-01 -1.17190099e+00 -3.61396939e-01 7.68441558e-01
-2.27244794e-02 1.73609555e-01 -3.18839252e-01 6.02497578e-01
6.39999518e-03 3.87179404e-01 3.78553689e-01 -2.50763714e-01
5.49797714e-01 -2.50260055e-01 6.65854275e-01 -7.31064379e-01
-2.13425741e-01 -1.31970197e-01 -9.37301889e-02 -5.47353089e-01
-1.29168764e-01 -3.80637139e-01 -1.45666361e+00 -4.78013992e-01
-1.76659405e-01 -1.70807093e-01 6.01717174e-01 4.09130126e-01
7.99578488e-01 2.62550414e-01 4.90440637e-01 -1.44546366e+00
-5.30650079e-01 -5.30628204e-01 -4.49940234e-01 7.47019768e-01
4.34981674e-01 -1.29059851e+00 -2.45169774e-01 -4.70838904e-01] | [8.616474151611328, -3.6587820053100586] |
3e6ec533-5035-4dd6-9c45-6508656d7bdf | learning-free-iris-segmentation-revisited-a | 1901.01575 | null | http://arxiv.org/abs/1901.01575v1 | http://arxiv.org/pdf/1901.01575v1.pdf | Learning-Free Iris Segmentation Revisited: A First Step Toward Fast Volumetric Operation Over Video Samples | Subject matching performance in iris biometrics is contingent upon fast,
high-quality iris segmentation. In many cases, iris biometrics acquisition
equipment takes a number of images in sequence and combines the segmentation
and matching results for each image to strengthen the result. To date,
segmentation has occurred in 2D, operating on each image individually. But such
methodologies, while powerful, do not take advantage of potential gains in
performance afforded by treating sequential images as volumetric data. As a
first step in this direction, we apply the Flexible Learning-Free
Reconstructoin of Neural Volumes (FLoRIN) framework, an open source
segmentation and reconstruction framework originally designed for neural
microscopy volumes, to volumetric segmentation of iris videos. Further, we
introduce a novel dataset of near-infrared iris videos, in which each subject's
pupil rapidly changes size due to visible-light stimuli, as a test bed for
FLoRIN. We compare the matching performance for iris masks generated by FLoRIN,
deep-learning-based (SegNet), and Daugman's (OSIRIS) iris segmentation
approaches. We show that by incorporating volumetric information, FLoRIN
achieves a factor of 3.6 to an order of magnitude increase in throughput with
only a minor drop in subject matching performance. We also demonstrate that
FLoRIN-based iris segmentation maintains this speedup on low-resource hardware,
making it suitable for embedded biometrics systems. | ['Walter Scheirer', 'Camila Carballo', 'Jeffery Kinnison', 'Mateusz Trokielewicz', 'Adam Czajka'] | 2019-01-06 | null | null | null | null | ['iris-segmentation'] | ['medical'] | [ 4.86120075e-01 -2.47248858e-01 -2.69449428e-02 -3.99172068e-01
-7.40773857e-01 -5.86304605e-01 1.20012179e-01 1.07173391e-01
-4.38371420e-01 3.87786180e-01 -9.98493284e-02 -4.24418181e-01
-3.90933268e-02 -5.56888700e-01 -4.78265256e-01 -7.04173982e-01
9.33254287e-02 5.76112211e-01 -7.32530933e-03 2.31695011e-01
2.28569999e-01 8.86492968e-01 -1.61457586e+00 1.65311232e-01
7.81285405e-01 7.71224499e-01 -4.24909890e-01 1.03478527e+00
-9.26817395e-03 6.15924038e-02 -4.33430582e-01 -4.12327588e-01
8.54590714e-01 -5.01097322e-01 -1.04188132e+00 4.14308488e-01
1.29053807e+00 -6.33650005e-01 -1.92715555e-01 8.25535178e-01
6.94238365e-01 -9.39438716e-02 2.61911571e-01 -6.16784275e-01
-2.17239246e-01 1.45233423e-01 -9.70289230e-01 3.81294340e-01
2.55220473e-01 8.01355422e-01 5.66890121e-01 -4.00010020e-01
7.51872480e-01 8.80220175e-01 7.74943233e-01 7.33708858e-01
-1.63092828e+00 -3.95444393e-01 -5.90539992e-01 -5.81437767e-01
-1.34014523e+00 -6.41670942e-01 1.49669841e-01 -6.18685842e-01
9.81552482e-01 5.57018518e-01 9.63921666e-01 1.99538603e-01
1.12909637e-01 5.90855300e-01 1.49086511e+00 -3.76005381e-01
-1.16039403e-01 -1.01471446e-01 3.71623486e-02 7.27020502e-01
1.91025034e-01 3.87388408e-01 -2.44042531e-01 -1.62630856e-01
1.02350402e+00 6.21108431e-03 -2.81337887e-01 7.20941881e-03
-1.08164167e+00 2.33904049e-01 2.33194992e-01 2.42415130e-01
-1.91571325e-01 -1.60513952e-01 1.49808392e-01 3.10014307e-01
5.23636878e-01 5.14923990e-01 -2.68779248e-02 -1.94931000e-01
-1.63610601e+00 7.78556466e-02 6.10047758e-01 5.51993668e-01
6.28693640e-01 -2.97530383e-01 -2.06726968e-01 4.66229141e-01
2.35097304e-01 4.78803724e-01 1.94505200e-01 -7.51174688e-01
-1.65216833e-01 8.51931989e-01 -1.33072091e-02 -4.77144808e-01
-6.01322651e-01 -3.43511343e-01 -6.50700271e-01 4.57718521e-01
9.61646736e-01 -6.19979575e-03 -1.33434021e+00 1.03924513e+00
6.78892374e-01 3.11204374e-01 -3.23401183e-01 1.12445140e+00
7.99393713e-01 6.19633049e-02 -1.75900877e-01 -2.22051039e-01
1.30555642e+00 -4.98476088e-01 -9.05471072e-02 3.98252070e-01
5.01564920e-01 -1.13037062e+00 9.98494923e-01 5.05415618e-01
-1.35149515e+00 -4.16798860e-01 -7.65791774e-01 -1.91577077e-01
-9.82115567e-02 4.27708700e-02 6.51149988e-01 1.07126844e+00
-1.38983989e+00 6.41490638e-01 -9.86913621e-01 -5.28774202e-01
7.47348487e-01 1.13400209e+00 -2.71579802e-01 3.97940129e-02
-3.22603852e-01 4.50114071e-01 -2.72722133e-02 3.34471166e-02
-2.14643091e-01 -1.00735962e+00 -5.42647719e-01 -2.79440284e-01
-8.30759481e-02 -6.68005407e-01 1.02141726e+00 -7.73720026e-01
-1.54194438e+00 1.62881196e+00 -4.64441121e-01 -4.71353173e-01
4.22461838e-01 2.96337783e-01 -5.30113578e-02 3.91217053e-01
-3.26438934e-01 9.09236789e-01 6.87737703e-01 -8.48225355e-01
-5.36494076e-01 -8.09689641e-01 -3.70977111e-02 -3.86318453e-02
5.21481074e-02 3.22158277e-01 -5.37099957e-01 -1.04808956e-01
6.84147421e-03 -1.01931095e+00 -2.30169564e-01 2.16638669e-01
-3.62184376e-01 4.93420735e-02 4.76661533e-01 -7.44069695e-01
8.64269912e-01 -2.04339623e+00 -1.76467597e-01 4.79340583e-01
6.49855137e-01 6.86169028e-01 -2.37286478e-01 -1.49319202e-01
1.40067294e-01 1.33531645e-01 -2.71178275e-01 -4.85642999e-01
-4.49833781e-01 -2.19209958e-02 1.66744933e-01 8.87359262e-01
-5.07583842e-02 9.62129772e-01 -6.50088847e-01 -6.88617766e-01
4.90933806e-01 6.23931468e-01 -5.06815016e-01 1.56031046e-02
-6.02177391e-03 9.82847154e-01 2.51344722e-02 1.12404537e+00
9.19918478e-01 -5.28336942e-01 7.68913776e-02 -2.09279463e-01
-2.80137271e-01 -6.27122968e-02 -9.36039329e-01 1.71926129e+00
-1.06660932e-01 6.47701025e-01 2.76129216e-01 -4.95159626e-01
6.62747920e-01 3.04722577e-01 9.50215340e-01 -6.33195400e-01
3.17237377e-01 4.55780923e-01 1.73313126e-01 -4.29344535e-01
3.12137097e-01 -3.01927835e-01 6.80780649e-01 6.26153469e-01
-1.06485374e-01 -2.27891192e-01 3.18233252e-01 -1.52203798e-01
8.08753610e-01 3.70023213e-02 2.70319302e-02 -1.21975437e-01
4.13672537e-01 1.44477680e-01 2.73496360e-01 5.33025742e-01
-5.40201843e-01 6.65574372e-01 3.59460711e-01 -7.06777871e-01
-1.18454754e+00 -8.48433316e-01 -8.39658141e-01 3.49261910e-01
8.68645310e-02 -1.76128760e-01 -1.11954665e+00 -5.00068486e-01
2.24163949e-01 -3.06413084e-01 -4.38948572e-01 6.89925075e-01
-5.32131076e-01 -1.06193006e+00 6.39744341e-01 7.39550814e-02
2.78113484e-01 -7.35740066e-01 -7.92821407e-01 4.30554599e-02
3.11358422e-01 -9.22393024e-01 -5.06956041e-01 -6.12598419e-01
-1.06776154e+00 -1.45040524e+00 -7.98679471e-01 -5.12166798e-01
9.77471471e-01 7.78474808e-02 9.79495049e-01 5.66269100e-01
-1.00953925e+00 5.35783231e-01 1.81859136e-01 -1.87653080e-01
-2.75753766e-01 -1.04112998e-01 9.45887417e-02 2.24288493e-01
7.44292557e-01 -3.80766004e-01 -9.68458295e-01 1.10543422e-01
-8.06907415e-01 2.61472706e-02 4.37659234e-01 9.11830008e-01
1.00900197e+00 -3.59596819e-01 -1.93272993e-01 -8.76721978e-01
2.36702651e-01 1.38387531e-01 -1.15607238e+00 1.33913502e-01
-9.40678000e-01 -2.64562517e-01 2.95563281e-01 -1.86741978e-01
-7.03955233e-01 2.89438725e-01 -1.07439213e-01 -4.83899087e-01
-2.93905109e-01 8.75112116e-02 5.35954535e-01 -9.45213854e-01
7.77644157e-01 1.41116753e-01 6.34706616e-01 -3.73678863e-01
1.16091833e-01 9.24135506e-01 5.72871506e-01 -4.68637347e-01
5.94883621e-01 8.80998313e-01 3.44691068e-01 -8.08220983e-01
-4.48052764e-01 -8.39815378e-01 -8.43272567e-01 -1.60156503e-01
7.24652767e-01 -4.83578742e-01 -1.28996360e+00 7.52963901e-01
-7.45658875e-01 -3.91260445e-01 -4.85125154e-01 5.47228932e-01
-4.16954458e-01 4.01898086e-01 -8.70596290e-01 -6.31987631e-01
-5.94271302e-01 -1.62190557e+00 1.21948838e+00 7.43222892e-01
-5.60166948e-02 -8.56464148e-01 1.12841673e-01 8.51737678e-01
4.27988976e-01 4.76000071e-01 5.08730829e-01 -3.26117188e-01
-9.52148676e-01 -1.34176582e-01 -5.59118211e-01 1.96865827e-01
2.61503965e-01 2.62739658e-01 -1.15988779e+00 -6.40609741e-01
-1.97211310e-01 -1.58160493e-01 7.40523577e-01 8.47679794e-01
1.12101352e+00 -5.84188476e-02 -3.90086681e-01 1.21030414e+00
1.67868710e+00 1.15335114e-01 8.50113928e-01 -5.75595833e-02
6.90147042e-01 6.07229054e-01 1.95753381e-01 1.87072426e-01
5.15539460e-02 6.56157553e-01 2.68494599e-02 -8.97812724e-01
-4.63170975e-01 2.28970811e-01 -3.35226387e-01 4.20472950e-01
-5.27308941e-01 2.33918637e-01 -9.89968479e-01 6.13399267e-01
-1.33341563e+00 -8.77523065e-01 -2.78437942e-01 2.59200692e+00
1.16989410e+00 -4.12380368e-01 4.67434585e-01 -1.27332613e-01
5.31814933e-01 -2.34539106e-01 -7.92989016e-01 -2.99856573e-01
-4.95821722e-02 7.85507560e-01 7.05941975e-01 6.73669219e-01
-9.88251269e-01 9.02915299e-01 6.75077629e+00 5.28146267e-01
-1.47695303e+00 -9.23539028e-02 9.87564564e-01 -6.29699409e-01
-6.70258477e-02 6.02730662e-02 -8.19336295e-01 3.01856339e-01
1.07778382e+00 3.16798657e-01 7.11512625e-01 1.58674628e-01
2.01499462e-01 -3.86632353e-01 -9.26999569e-01 1.22915852e+00
-1.72829852e-01 -1.64264381e+00 -2.16046393e-01 6.68235838e-01
7.48746514e-01 1.51227534e-01 3.14210176e-01 -4.37028408e-01
-1.42883673e-01 -1.35019338e+00 -4.09323752e-01 6.47849262e-01
1.47210312e+00 -5.67233145e-01 6.31733775e-01 -2.55387336e-01
-9.09620166e-01 4.04365718e-01 -2.69687861e-01 2.84779876e-01
-1.86985299e-01 6.08499050e-01 -1.10924697e+00 2.07512766e-01
4.66371149e-01 5.00699461e-01 -6.77367270e-01 1.31781971e+00
5.32008648e-01 6.66586280e-01 -3.27462107e-01 4.29008305e-01
4.34153117e-02 -5.76310992e-01 6.37643576e-01 1.13497126e+00
-7.53771067e-02 2.75705814e-01 -7.48905465e-02 1.00059772e+00
-1.80258781e-01 2.42102340e-01 -3.47416639e-01 -9.16017368e-02
1.04879506e-01 1.38638663e+00 -8.27117026e-01 -2.53813058e-01
-6.51739657e-01 7.69454479e-01 -1.71896785e-01 2.29560003e-01
-3.20476234e-01 -2.62561142e-01 7.65286922e-01 5.00418305e-01
-8.83467868e-02 6.68231100e-02 -4.61318314e-01 -1.02541971e+00
-8.68870169e-02 -1.13139927e+00 9.44872946e-02 -3.28234881e-01
-1.07290137e+00 2.96751499e-01 -4.11901444e-01 -1.02344346e+00
-5.65749779e-02 -5.83073854e-01 -2.36589223e-01 1.41432512e+00
-1.64017272e+00 -1.32089090e+00 -3.64506602e-01 6.31220758e-01
8.87104943e-02 -1.63075551e-01 7.20477998e-01 3.69136065e-01
-5.56948304e-01 8.38971734e-01 7.05797225e-02 2.82948017e-01
7.36783803e-01 -1.31245685e+00 4.83893126e-01 8.86209667e-01
2.13474721e-01 9.00231779e-01 2.86682069e-01 -5.52011907e-01
-1.58906031e+00 -7.70737648e-01 6.85950577e-01 -5.20566106e-01
1.66483760e-01 7.53236264e-02 -7.81447411e-01 6.32901847e-01
1.30142868e-01 3.64446729e-01 1.03727078e+00 7.03724623e-02
-1.20498361e-02 -1.49685219e-01 -1.68700755e+00 4.52359229e-01
8.64794910e-01 -5.33430338e-01 -9.35157984e-02 6.76606417e-01
1.73478574e-01 -1.22581220e+00 -1.34870327e+00 3.32913637e-01
7.72405267e-01 -1.25092590e+00 7.84544110e-01 -4.74607527e-01
1.73141539e-01 -4.01640028e-01 3.23022068e-01 -5.54714918e-01
1.55894324e-01 -1.11252916e+00 1.42693385e-01 9.06956851e-01
2.39765942e-01 -8.00061166e-01 1.21625364e+00 9.37264383e-01
1.11850522e-01 -9.67942297e-01 -9.68198180e-01 -3.81457806e-01
-5.50617725e-02 1.37271849e-03 8.51186216e-01 6.97407603e-01
-9.51467007e-02 -3.83215249e-01 3.41199376e-02 3.09255254e-02
8.89336169e-01 4.09225553e-01 1.03195572e+00 -1.28240311e+00
-3.53990316e-01 -4.55216348e-01 -8.09284389e-01 -9.26189482e-01
-2.65398264e-01 -8.56051683e-01 -4.62781668e-01 -1.15692556e+00
3.58407557e-01 -5.49147427e-01 -1.38718739e-01 6.32597625e-01
-1.93856191e-02 9.30355370e-01 -4.97658700e-02 4.70717669e-01
-7.06396252e-02 -5.14871657e-01 1.63089001e+00 -1.77262038e-01
-6.17445290e-01 1.07542157e-01 -5.43904126e-01 2.60515749e-01
5.92593431e-01 4.25696671e-02 -1.70279250e-01 -1.41093299e-01
-1.65073767e-01 1.05666518e-01 3.79129380e-01 -9.74709034e-01
3.68350744e-01 5.90336584e-02 5.16657412e-01 -3.99182528e-01
1.94176063e-01 -4.63903069e-01 1.08320028e-01 4.31043237e-01
-5.51882498e-02 -3.79824102e-01 5.85950732e-01 -1.38892662e-02
-1.93276823e-01 1.52091175e-01 1.15931511e+00 -1.63840368e-01
-1.66594431e-01 7.36728489e-01 1.26840428e-01 -6.99806632e-03
8.09507012e-01 -8.25399816e-01 -4.27871525e-01 1.99792534e-01
-7.02274144e-01 5.91918044e-02 9.88036215e-01 -1.34592071e-01
6.08719230e-01 -5.88968217e-01 -7.96916068e-01 7.13330150e-01
-5.43455146e-02 8.88324976e-02 4.18933630e-01 1.48567879e+00
-8.95736992e-01 5.66618204e-01 -2.03378737e-01 -1.11203074e+00
-1.82278585e+00 2.56223887e-01 8.91599476e-01 -1.07150584e-01
-7.82750666e-01 1.01927793e+00 -1.39342397e-01 -2.87070423e-01
-8.32203850e-02 -3.72879863e-01 1.50753275e-01 -2.52456009e-01
7.46939182e-01 1.64633229e-01 2.81294793e-01 -6.47597730e-01
-1.46332249e-01 9.44889426e-01 -2.25724027e-01 1.02858715e-01
9.23772871e-01 -1.70547403e-02 -4.30151403e-01 -8.01032931e-02
9.11885977e-01 2.36964837e-01 -1.16111207e+00 -2.40020752e-01
-4.48205829e-01 -8.12221527e-01 3.28951031e-01 -8.69926691e-01
-1.27962422e+00 7.55387366e-01 1.03797626e+00 -5.67574315e-02
1.41721296e+00 -2.23009884e-01 1.10330927e+00 -3.88485044e-01
3.12090158e-01 -7.28938401e-01 -7.99916565e-01 -1.49229750e-01
2.00245649e-01 -1.33703530e+00 1.73330784e-01 -3.13274652e-01
-8.16197172e-02 1.18052316e+00 3.82241040e-01 2.35828057e-01
3.50102454e-01 3.43414783e-01 4.11680967e-01 -4.64641929e-01
-9.45860296e-02 -3.93235117e-01 6.64769650e-01 6.75846994e-01
7.10786223e-01 2.16516718e-01 -2.17364430e-01 -4.92381871e-01
-1.63022548e-01 2.65299201e-01 4.68767941e-01 6.07036948e-01
-2.42035147e-02 -1.38492990e+00 -4.38576549e-01 9.23529267e-01
-6.08532190e-01 -1.73129126e-01 -2.43469432e-01 6.93575442e-01
2.80374497e-01 7.84778714e-01 3.67718756e-01 -2.25674748e-01
9.24048480e-03 -1.09301075e-01 9.46602106e-01 -6.21256053e-01
-1.04966736e+00 3.06257963e-01 -5.19050360e-01 -8.07935953e-01
-6.58616245e-01 -7.86408842e-01 -1.13921595e+00 -7.75234640e-01
-1.79424301e-01 -2.84900308e-01 6.78984046e-01 8.82681251e-01
5.19796968e-01 6.23743683e-02 4.27785993e-01 -6.41763508e-01
-1.32779241e-01 -5.50544739e-01 -7.88964629e-01 2.94532031e-01
6.39143527e-01 -1.70758188e-01 -2.96440363e-01 1.84322029e-01] | [3.7429420948028564, -3.6317481994628906] |
03c54bc3-b6a9-4c47-8848-c957390e2982 | efficient-detection-of-botnet-traffic-by | 2107.02896 | null | https://arxiv.org/abs/2107.02896v1 | https://arxiv.org/pdf/2107.02896v1.pdf | Efficient Detection of Botnet Traffic by features selection and Decision Trees | Botnets are one of the online threats with the biggest presence, causing billionaire losses to global economies. Nowadays, the increasing number of devices connected to the Internet makes it necessary to analyze large amounts of network traffic data. In this work, we focus on increasing the performance on botnet traffic classification by selecting those features that further increase the detection rate. For this purpose we use two feature selection techniques, Information Gain and Gini Importance, which led to three pre-selected subsets of five, six and seven features. Then, we evaluate the three feature subsets along with three models, Decision Tree, Random Forest and k-Nearest Neighbors. To test the performance of the three feature vectors and the three models we generate two datasets based on the CTU-13 dataset, namely QB-CTU13 and EQB-CTU13. We measure the performance as the macro averaged F1 score over the computational time required to classify a sample. The results show that the highest performance is achieved by Decision Trees using a five feature set which obtained a mean F1 score of 85% classifying each sample in an average time of 0.78 microseconds. | ['Enrique Alegre', 'Eduardo Fidalgo', 'Víctor González-Castro', 'Javier Velasco-Mata'] | 2021-06-30 | null | null | null | null | ['traffic-classification'] | ['miscellaneous'] | [-1.04197234e-01 -6.29730582e-01 -1.15030371e-01 2.22932864e-02
-4.84011583e-02 -4.63875681e-01 6.59660876e-01 1.76768824e-01
-7.67821491e-01 8.23792815e-01 -4.32840049e-01 -6.51448369e-01
-4.53939974e-01 -1.10686195e+00 2.47843087e-01 -5.95107794e-01
-3.41278404e-01 6.58781052e-01 8.84911239e-01 -4.88271266e-02
7.67105997e-01 7.83625722e-01 -1.51718485e+00 7.37322345e-02
8.76669645e-01 1.11571109e+00 -1.00896016e-01 6.35517359e-01
-2.14604154e-01 5.06360948e-01 -8.75870824e-01 -5.03533125e-01
4.81052428e-01 -1.75345555e-01 -9.09822941e-01 -1.83640838e-01
-2.96587110e-01 -2.76032120e-01 -8.33951086e-02 9.44653690e-01
3.42890978e-01 -4.95791994e-02 7.23663270e-01 -1.52660954e+00
2.33004048e-01 3.42875510e-01 -5.80831051e-01 8.11695516e-01
3.00477371e-02 4.37679499e-01 9.33743417e-01 -3.83298278e-01
7.06914842e-01 1.07597899e+00 3.85092556e-01 1.81489632e-01
-1.27310920e+00 -7.77282178e-01 -3.24020654e-01 7.42950559e-01
-1.24174261e+00 -1.59488380e-01 5.57811856e-01 -6.92859173e-01
9.04589534e-01 2.76503265e-01 4.82569605e-01 8.33325505e-01
1.99922591e-01 1.55859469e-02 1.47817624e+00 -1.49710298e-01
1.69891715e-01 5.45267582e-01 3.97819489e-01 5.17821550e-01
5.19628882e-01 1.17660291e-01 1.37811124e-01 -4.59893525e-01
4.17487711e-01 3.84210236e-02 1.51911303e-01 2.68995792e-01
-8.71151209e-01 1.13166845e+00 3.32492948e-01 6.65283024e-01
-5.07334471e-01 -2.05668539e-01 6.28629744e-01 4.28240567e-01
2.70693570e-01 3.91816050e-01 -5.46537459e-01 -4.40311223e-01
-5.98622441e-01 -8.66118371e-02 8.54323804e-01 3.32045257e-01
9.11653876e-01 -1.17563099e-01 1.25097821e-03 5.91434240e-01
-1.93466693e-02 4.91920084e-01 5.50485432e-01 -5.09812117e-01
3.63132805e-01 7.81748414e-01 -1.05732754e-01 -1.28097034e+00
-4.70064580e-01 -3.28014135e-01 -6.79080963e-01 2.80489624e-01
6.76683784e-01 -2.06593186e-01 -4.72858816e-01 1.31555760e+00
2.55166411e-01 -1.60456151e-02 -2.86821634e-01 4.28400427e-01
4.69952673e-01 5.63085675e-01 2.35139672e-02 -3.16708803e-01
1.22848201e+00 -5.15191793e-01 -1.53632298e-01 5.36640823e-01
5.57130754e-01 -8.21413398e-01 8.12866867e-01 4.21568245e-01
-2.64950216e-01 -3.19014043e-01 -5.54131687e-01 7.73039579e-01
-8.22538316e-01 -1.24465391e-01 3.61838877e-01 8.92642915e-01
-7.38945246e-01 7.93561935e-01 -4.00479406e-01 -5.72959065e-01
4.15733039e-01 4.70436692e-01 -2.54885465e-01 3.59629929e-01
-1.00278342e+00 8.46370041e-01 4.20306504e-01 -6.88944578e-01
-4.88458425e-01 -3.18886012e-01 9.97936651e-02 2.04687774e-01
2.76129037e-01 -4.07488598e-03 5.97880661e-01 -4.82331723e-01
-1.25671434e+00 5.17185390e-01 4.63073328e-02 -5.19329786e-01
5.09715974e-01 6.40769839e-01 -5.39158285e-01 1.29167452e-01
1.82091221e-01 2.72970021e-01 8.15685749e-01 -9.78545487e-01
-9.77381229e-01 -4.88985121e-01 -1.47875808e-02 -7.05451488e-01
-4.85683113e-01 3.88563663e-01 2.22791135e-01 -2.69120991e-01
-1.21317834e-01 -7.74225771e-01 -1.63665578e-01 -7.30232477e-01
-4.80113059e-01 -4.62026149e-01 1.38982654e+00 -7.23866105e-01
1.37479556e+00 -1.77212071e+00 -3.25530142e-01 7.50036240e-01
3.46120507e-01 7.24348247e-01 2.48518698e-02 3.84450942e-01
9.65562165e-02 4.80982363e-01 1.91736012e-03 2.07757577e-01
-1.64396301e-01 -5.00437580e-02 -1.99151099e-01 1.44687191e-01
1.22276202e-01 3.95130783e-01 -7.01814294e-01 -4.78873610e-01
3.85319144e-01 2.86015689e-01 -6.27966940e-01 3.51449624e-02
6.61538243e-02 5.82326829e-01 -5.47151804e-01 4.66769457e-01
6.11657619e-01 -1.37907550e-01 1.06933773e-01 8.48261546e-03
-1.78090662e-01 3.55821669e-01 -9.79426503e-01 3.82202834e-01
-4.93560582e-01 5.76861799e-01 -4.96422023e-01 -8.99444818e-01
1.21464360e+00 1.55656844e-01 8.26474786e-01 -6.74672604e-01
5.60450375e-01 3.59723121e-01 3.34022015e-01 -3.57250154e-01
5.93483523e-02 2.51292437e-01 1.66101620e-01 7.31917143e-01
-1.37591422e-01 2.56552994e-01 7.64366567e-01 8.75431523e-02
1.30841374e+00 -8.03877950e-01 4.53152090e-01 -1.47765681e-01
8.62368524e-01 -1.14671566e-01 2.97955394e-01 5.45911133e-01
-6.30354702e-01 -1.44437164e-01 9.43980277e-01 -7.07387745e-01
-8.42056394e-01 -8.09066296e-01 -3.01843863e-02 7.02249050e-01
-3.59828733e-02 -3.91809255e-01 -7.57360578e-01 -1.27235889e+00
-1.22443616e-01 5.30454874e-01 -2.30595067e-01 9.00686607e-02
-5.60453355e-01 -9.02031302e-01 6.00804687e-01 -1.72342300e-01
8.07906747e-01 -1.24449635e+00 -8.55193317e-01 7.26950727e-03
-1.68491200e-01 -1.05513644e+00 3.11474241e-02 -4.16505076e-02
-7.43333638e-01 -1.43576825e+00 -9.69437882e-03 -3.02970886e-01
3.83133858e-01 1.78410694e-01 6.88350558e-01 2.85051465e-01
-4.98222798e-01 -2.48398215e-01 -6.98594689e-01 -5.68088777e-02
-4.01843786e-01 4.21332359e-01 1.50968537e-01 1.51832938e-01
5.86461663e-01 -9.80947852e-01 -2.89156258e-01 5.16231954e-01
-4.34548557e-01 -4.56212670e-01 6.33210599e-01 5.42568862e-01
4.96761166e-02 5.81743181e-01 6.86726928e-01 -8.11014116e-01
5.67026556e-01 -8.46589446e-01 -7.57137239e-01 -9.91921127e-02
-6.82776093e-01 -8.65035579e-02 9.98182356e-01 -4.75695044e-01
-4.91834968e-01 -2.82749385e-01 -1.40001982e-01 -6.07710704e-02
-2.19350591e-01 9.66401771e-02 9.93724838e-02 -2.09393203e-01
7.53670156e-01 7.37171918e-02 -1.17486775e-01 -4.36155915e-01
-6.71721995e-02 1.07895827e+00 -2.04100013e-01 -2.21424326e-01
1.02228081e+00 2.52156854e-01 1.59399658e-01 -1.13359225e+00
2.97539146e-03 -5.19012034e-01 -5.99878669e-01 -3.80705833e-01
7.21158624e-01 1.60985798e-01 -1.09246719e+00 3.67762476e-01
-1.23729181e+00 8.79165083e-02 4.46476527e-02 3.97360682e-01
-1.57706887e-01 4.96626765e-01 -2.94442952e-01 -8.65839183e-01
-4.55966890e-01 -1.38201356e+00 4.10280585e-01 8.50952119e-02
-1.40024871e-01 -6.26315534e-01 -8.76518190e-02 1.34631321e-01
7.05080509e-01 2.15001702e-01 1.14676249e+00 -1.08029437e+00
-4.54127818e-01 -2.93816954e-01 -7.34576762e-01 3.31131756e-01
3.20397735e-01 2.04695746e-01 -7.05399036e-01 -2.82130152e-01
-1.55319303e-01 3.15262228e-02 7.61191905e-01 6.85281828e-02
1.03579736e+00 -3.99874866e-01 -3.71788025e-01 1.55649185e-01
1.45670795e+00 8.47063422e-01 5.44603944e-01 3.46136749e-01
2.68036664e-01 5.99573910e-01 4.25588161e-01 6.06932163e-01
-7.29342997e-02 8.23259473e-01 6.19700491e-01 3.85633737e-01
8.77531096e-02 -1.34702906e-01 6.41199499e-02 6.65779769e-01
-3.15989316e-01 -1.80398807e-01 -1.00287044e+00 2.52785772e-01
-1.37863564e+00 -1.24650490e+00 -1.44524515e-01 2.40073204e+00
2.63882607e-01 4.36727464e-01 7.76972353e-01 7.52259016e-01
9.08531964e-01 -7.86798596e-02 -1.51609913e-01 -4.71488029e-01
4.20674980e-01 4.97430086e-01 6.42886877e-01 5.49561441e-01
-1.11350250e+00 9.04642463e-01 5.67793798e+00 1.07284033e+00
-1.47554600e+00 7.87583739e-02 7.15779185e-01 3.86392921e-01
3.09210569e-01 7.21519291e-02 -7.45576024e-01 8.90060365e-01
1.20099187e+00 -2.46681646e-01 6.34480298e-01 7.66519308e-01
2.93018609e-01 -7.69417882e-02 -2.72839874e-01 8.13749075e-01
-3.87168109e-01 -9.45425510e-01 1.87846750e-01 4.87335324e-01
4.83770669e-01 2.42084637e-02 -6.92063719e-02 2.70433396e-01
3.31639022e-01 -8.78326178e-01 1.46223068e-01 1.61846682e-01
7.17813373e-01 -8.22789431e-01 9.19222713e-01 4.90545839e-01
-1.13956535e+00 -5.53131700e-01 -3.95031840e-01 -1.63955227e-01
-2.66218521e-02 7.64335275e-01 -1.21091247e+00 5.42130113e-01
6.38729036e-01 3.79818410e-01 -6.99173331e-01 1.22799134e+00
8.49404186e-02 8.30355525e-01 -4.89251673e-01 -5.39085865e-01
1.74478859e-01 -2.50763685e-01 7.89005280e-01 1.05335903e+00
3.15314561e-01 -2.72846669e-01 9.95486602e-02 6.84533477e-01
2.56970674e-01 3.85743350e-01 -6.53876841e-01 1.59476548e-01
8.31660509e-01 1.43227339e+00 -1.08520114e+00 -2.82156318e-01
-1.15929969e-01 5.58550954e-01 1.22732751e-01 -5.82587570e-02
-1.02047575e+00 -8.46858799e-01 5.49791396e-01 3.57130975e-01
2.48649552e-01 -2.40349442e-01 -2.47925311e-01 -8.27196181e-01
-2.52015799e-01 -7.25776792e-01 3.11061174e-01 2.88092066e-02
-1.19677079e+00 9.91693676e-01 4.21453565e-02 -1.17358005e+00
-2.75467813e-01 -7.41541028e-01 -8.52614045e-01 6.80759966e-01
-1.10091233e+00 -7.05416560e-01 -3.35429817e-01 4.26691175e-01
3.26514602e-01 -4.72915649e-01 6.81299567e-01 6.29501402e-01
-8.06774259e-01 4.57249343e-01 5.02675325e-02 1.85845569e-01
2.00367883e-01 -7.74894178e-01 2.68295825e-01 6.80824399e-01
2.17883307e-02 4.42957252e-01 4.25454050e-01 -5.83964169e-01
-5.88505030e-01 -9.49819744e-01 1.04474652e+00 -2.48642161e-01
7.74356008e-01 -3.37543078e-02 -4.09185737e-01 3.61150861e-01
-3.91277283e-01 -1.25539750e-01 3.18113595e-01 -2.58560240e-01
-4.05883580e-01 -2.96089470e-01 -1.77620792e+00 2.73905337e-01
6.87142491e-01 -9.98379588e-02 -2.48735789e-02 1.21902071e-01
3.72700959e-01 6.24221027e-01 -8.95642459e-01 2.64406055e-01
5.62484503e-01 -1.30913258e+00 7.26228893e-01 -6.47529066e-01
3.90990190e-02 -3.12383771e-01 -1.65711969e-01 -1.04148495e+00
-3.06103259e-01 -3.80318135e-01 2.79755950e-01 1.29385483e+00
5.28836668e-01 -1.17482150e+00 8.53784740e-01 -1.05664469e-01
7.48984814e-01 -8.88094127e-01 -1.17849302e+00 -8.97673905e-01
-9.84153897e-02 -1.27617642e-01 7.27891564e-01 7.80004561e-01
9.62198749e-02 3.80332083e-01 -1.46306828e-01 -3.66825700e-01
5.90342283e-01 -8.69405568e-02 8.77656817e-01 -1.67495477e+00
-7.39888474e-02 -6.86438262e-01 -1.04501605e+00 -5.12048900e-01
-9.20891911e-02 -7.27292299e-01 -5.95852673e-01 -9.68002319e-01
2.27287322e-01 -8.72018397e-01 -1.23546056e-01 3.38268876e-01
1.32631883e-01 4.72321540e-01 1.89856663e-01 2.17787862e-01
-2.99061120e-01 1.01347707e-01 8.07826400e-01 -6.11855043e-03
-2.36464173e-01 4.70154524e-01 -2.73168832e-01 6.13327563e-01
1.34793639e+00 -5.28658271e-01 -3.01951736e-01 1.44362360e-01
-4.70039457e-01 -2.87749499e-01 1.52364865e-01 -1.32321596e+00
-1.09643281e-01 -2.94792086e-01 1.35183379e-01 -4.64214057e-01
1.46063074e-01 -9.48660612e-01 6.54303133e-02 9.78654087e-01
2.05357552e-01 2.12925691e-02 -2.40233287e-01 2.01743692e-01
-2.00400930e-02 -3.63554180e-01 1.21431184e+00 -3.35974107e-03
-5.62815487e-01 3.59170467e-01 -6.28626585e-01 -2.13729590e-01
1.24848485e+00 -2.53263265e-01 -5.33071995e-01 -2.46589154e-01
-4.40673381e-01 -2.65910894e-01 1.62377357e-01 1.26063153e-01
3.42190981e-01 -9.94172335e-01 -5.34596324e-01 3.58717680e-01
-2.24335849e-01 -8.70806158e-01 -1.69156075e-01 1.00930166e+00
-6.67400956e-01 6.35102093e-01 -6.78471863e-01 -4.79979873e-01
-1.61808372e+00 4.94108588e-01 1.34715363e-01 -8.42332661e-01
-1.20442100e-01 3.14263374e-01 -5.93965650e-01 -3.95088196e-01
-8.90350863e-02 6.12573363e-02 -7.49904275e-01 5.29191755e-02
3.53748649e-01 1.25449133e+00 8.87661520e-03 -8.14380109e-01
-4.16690499e-01 6.70158327e-01 1.09559540e-02 -2.25980692e-02
1.03426790e+00 3.01626533e-01 -4.66751218e-01 -9.84954163e-02
1.31442142e+00 1.86570391e-01 -3.03437918e-01 1.08286329e-01
5.26211500e-01 -8.52124572e-01 -4.09795254e-01 -6.34257853e-01
-1.11941707e+00 7.86197305e-01 6.76430643e-01 9.17509794e-01
1.04727530e+00 -3.48959565e-01 9.19280112e-01 3.72715980e-01
8.15829575e-01 -4.93418366e-01 -2.40581250e-03 7.83613682e-01
2.08153456e-01 -9.98086929e-01 -1.86122507e-01 -6.57482207e-01
-2.99520373e-01 1.15386033e+00 6.95350826e-01 -2.02493206e-01
8.86717618e-01 -1.04947582e-01 -1.56685099e-01 -9.40640550e-03
-5.76740503e-01 -4.19927716e-01 -2.53080398e-01 6.83673859e-01
6.70867711e-02 2.71791846e-01 -8.92030656e-01 2.71061778e-01
-4.46062088e-01 -1.52733833e-01 3.04338336e-01 5.01165390e-01
-8.37276340e-01 -1.38097882e+00 -1.32254809e-01 9.98421371e-01
-5.18605828e-01 1.79398209e-01 -5.63559711e-01 7.38611937e-01
2.31213838e-01 1.39568937e+00 -2.50539333e-02 -1.17671788e+00
2.16200933e-01 -4.44063582e-02 9.16964263e-02 -1.97428539e-01
-6.50639236e-01 -6.89258575e-01 1.34815723e-01 -3.91637295e-01
-5.43968007e-02 -3.18548411e-01 -1.00579727e+00 -1.04150784e+00
-5.34629941e-01 5.77114105e-01 8.10822606e-01 8.20099771e-01
2.32538253e-01 1.50586277e-01 1.26780653e+00 -6.61873519e-01
-6.32853925e-01 -1.25370884e+00 -4.56704557e-01 4.80848640e-01
-1.38118938e-01 -8.80034864e-01 -6.88129544e-01 -5.17049313e-01] | [5.2309794425964355, 7.239346981048584] |
dbd9b945-b543-4f4c-bb2f-3b6c2945f2bd | unconstrained-monocular-3d-human-pose | null | null | http://openaccess.thecvf.com/content_cvpr_2013/html/Yu_Unconstrained_Monocular_3D_2013_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2013/papers/Yu_Unconstrained_Monocular_3D_2013_CVPR_paper.pdf | Unconstrained Monocular 3D Human Pose Estimation by Action Detection and Cross-Modality Regression Forest | This work addresses the challenging problem of unconstrained 3D human pose estimation (HPE) from a novel perspective. Existing approaches struggle to operate in realistic applications, mainly due to their scene-dependent priors, such as background segmentation and multi-camera network, which restrict their use in unconstrained environments. We therfore present a framework which applies action detection and 2D pose estimation techniques to infer 3D poses in an unconstrained video. Action detection offers spatiotemporal priors to 3D human pose estimation by both recognising and localising actions in space-time. Instead of holistic features, e.g. silhouettes, we leverage the flexibility of deformable part model to detect 2D body parts as a feature to estimate 3D poses. A new unconstrained pose dataset has been collected to justify the feasibility of our method, which demonstrated promising results, significantly outperforming the relevant state-of-the-arts. | ['Roberto Cipolla', 'Tae-Kyun Kim', 'Tsz-Ho Yu'] | 2013-06-01 | null | null | null | cvpr-2013-6 | ['monocular-3d-human-pose-estimation'] | ['computer-vision'] | [ 3.27324748e-01 1.03464007e-01 -1.41464069e-01 -1.17426496e-02
-5.18075585e-01 -4.60754395e-01 5.55917323e-01 -4.06974673e-01
-5.84490955e-01 5.48040152e-01 2.86251843e-01 5.43530464e-01
1.55941918e-01 -2.30943039e-01 -4.74182755e-01 -4.26748455e-01
-1.99139401e-01 5.75103104e-01 4.63901550e-01 -2.98763454e-01
-5.66109456e-03 7.22180426e-01 -1.59306586e+00 1.20988246e-02
3.27623427e-01 7.58459926e-01 -1.17815576e-01 8.93157303e-01
4.61370528e-01 3.14562976e-01 -5.39261580e-01 -4.05145139e-01
6.20045304e-01 -4.89128709e-01 -4.87349778e-01 7.14198172e-01
5.89847624e-01 -6.02608204e-01 -3.91636044e-01 6.17744565e-01
6.15189970e-01 3.58749717e-01 4.59414184e-01 -1.17929280e+00
3.43813330e-01 -3.41307968e-01 -5.98490953e-01 1.27950478e-02
1.15990829e+00 1.95514187e-01 7.28108346e-01 -7.55175114e-01
8.76720190e-01 1.48914826e+00 7.17264175e-01 6.93172395e-01
-1.03129029e+00 -2.28561342e-01 3.69763047e-01 -1.85188334e-02
-1.43937051e+00 -4.61673915e-01 9.43070531e-01 -5.11406720e-01
9.82069969e-01 4.61276233e-01 1.16703749e+00 1.41100740e+00
2.29592681e-01 1.16860545e+00 9.75841880e-01 -5.42788744e-01
6.63105473e-02 -3.29716176e-01 -4.26030040e-01 6.31552398e-01
1.64594293e-01 4.82656062e-02 -7.17655838e-01 -2.05495465e-03
1.19596910e+00 5.79178259e-02 -1.49804950e-01 -7.67270565e-01
-1.35385346e+00 3.60869259e-01 1.40780017e-01 -5.80498166e-02
-5.04232287e-01 1.47115335e-01 2.93747008e-01 -1.73250780e-01
4.80216175e-01 8.47768188e-02 -5.45964181e-01 -2.94332176e-01
-9.65247035e-01 6.20710909e-01 6.20412052e-01 9.15028453e-01
3.77644151e-01 -1.65673837e-01 -1.56081155e-01 4.47895199e-01
5.05761921e-01 4.51461196e-01 1.13174498e-01 -1.08937919e+00
4.85223502e-01 5.75637221e-01 3.77137333e-01 -1.00878024e+00
-4.91819352e-01 4.85628471e-02 -3.02066058e-01 1.04347043e-01
6.83176041e-01 -1.97039396e-01 -1.01866162e+00 1.39919603e+00
9.13716495e-01 3.43011990e-02 -2.87148744e-01 1.28625703e+00
4.68324363e-01 9.87624675e-02 -7.04847649e-02 -3.01243961e-02
1.35354340e+00 -8.76199126e-01 -6.40065312e-01 -6.24849617e-01
3.07842940e-01 -5.39400637e-01 5.72436750e-01 5.83560288e-01
-1.12113070e+00 -5.99437118e-01 -8.19962144e-01 6.95346147e-02
-9.90542546e-02 7.76532218e-02 6.48476720e-01 6.95609510e-01
-6.48620546e-01 4.47358191e-01 -1.17100918e+00 -7.51840830e-01
2.04205111e-01 4.36007261e-01 -6.15021169e-01 3.02837808e-02
-9.60599542e-01 9.86541808e-01 5.62360883e-01 3.46450150e-01
-8.99559736e-01 -7.58738890e-02 -1.10648072e+00 -4.51544970e-01
9.14547384e-01 -8.74420226e-01 1.13112175e+00 -6.88135326e-01
-1.75084186e+00 1.06444502e+00 5.43054938e-02 -2.60494500e-01
1.18034410e+00 -8.41675282e-01 -7.38097131e-02 5.82915127e-01
-1.43018752e-01 6.43922925e-01 9.04107213e-01 -1.08361006e+00
-3.66084993e-01 -5.87705970e-01 2.43995935e-01 7.21867681e-01
-9.80746746e-02 1.74770370e-01 -1.04333258e+00 -6.87969863e-01
3.10819179e-01 -1.14882731e+00 -5.54901719e-01 3.29216033e-01
-5.37923038e-01 1.50822237e-01 7.77345419e-01 -9.09527838e-01
1.03083301e+00 -1.80454969e+00 6.06385469e-01 6.94097131e-02
1.32733464e-01 2.02861637e-01 2.42534652e-01 2.81866819e-01
2.52729088e-01 -3.82444769e-01 -1.33725420e-01 -5.32831848e-01
1.46165788e-01 3.92688394e-01 2.99350262e-01 8.84005904e-01
2.91390747e-01 8.91524196e-01 -8.86293292e-01 -8.46713185e-01
6.96171999e-01 6.43193662e-01 -5.90938270e-01 3.46489280e-01
-2.51349092e-01 8.67561638e-01 -6.87675893e-01 8.31566155e-01
3.45840126e-01 7.05936179e-02 2.73897588e-01 -6.06505722e-02
6.11413568e-02 -2.73251414e-01 -1.58708572e+00 2.17214417e+00
4.62476052e-02 1.53450862e-01 2.38794610e-01 -7.18349040e-01
5.00874043e-01 5.19888043e-01 8.45196307e-01 -8.90328214e-02
2.80614793e-01 -7.10000237e-03 -2.55726188e-01 -6.02665126e-01
3.41465265e-01 8.41193425e-04 -3.17066312e-01 -1.11347660e-01
9.25000012e-02 -1.79100305e-01 1.05551690e-01 -3.59458886e-02
9.99451041e-01 1.07875681e+00 6.92381322e-01 3.34854983e-02
5.02894342e-01 -6.45110011e-02 5.05274117e-01 4.34154361e-01
-6.25841320e-01 8.27377260e-01 1.84160352e-01 -3.98515016e-01
-7.83397675e-01 -1.16005969e+00 8.77453685e-02 9.26915944e-01
2.45504215e-01 -4.69009489e-01 -8.33595216e-01 -8.64900053e-01
-1.12103496e-03 -1.70845017e-02 -4.98743385e-01 2.55154639e-01
-9.05178905e-01 -4.71281439e-01 3.87716293e-01 6.41278148e-01
4.30522144e-01 -7.69063652e-01 -1.35117698e+00 1.90152884e-01
-2.39668280e-01 -1.51379192e+00 -4.71385688e-01 -4.91918474e-02
-8.48232269e-01 -1.21514392e+00 -1.08588982e+00 -2.52026141e-01
5.51869988e-01 -9.79953483e-02 9.78156924e-01 -2.79719204e-01
-6.14064395e-01 9.68544781e-01 -4.34357494e-01 -2.54782528e-01
3.10135838e-02 -3.40322375e-01 3.62065703e-01 1.04188152e-01
1.43615201e-01 -3.56841624e-01 -8.21633041e-01 5.62010586e-01
-3.73206705e-01 -8.48637968e-02 6.02476716e-01 5.00409424e-01
6.21387124e-01 -1.74023882e-01 -8.77789333e-02 -4.73242581e-01
-7.25883665e-03 -1.89753454e-02 -3.54805976e-01 1.76231727e-01
1.07173063e-01 -4.72214937e-01 3.39188017e-02 -5.51269174e-01
-1.09149015e+00 6.44096315e-01 -1.23497033e-02 -5.58014452e-01
-6.10846519e-01 5.47868051e-02 -5.10827065e-01 -1.44317314e-01
4.76848215e-01 -5.16302772e-02 4.01589870e-02 -4.26981539e-01
3.01063776e-01 2.42616132e-01 7.34872997e-01 -5.79912484e-01
7.40273595e-01 6.39425814e-01 2.50321329e-01 -8.77754092e-01
-6.93199039e-01 -8.31159532e-01 -1.48928130e+00 -7.27904618e-01
1.08811843e+00 -1.13660228e+00 -7.20776737e-01 4.35141414e-01
-1.12091851e+00 -9.12744328e-02 -1.20224915e-01 7.35746682e-01
-9.90550876e-01 6.46155834e-01 -5.84600866e-01 -1.29129910e+00
1.07990004e-01 -8.66008461e-01 1.64025760e+00 -1.41871274e-01
-6.49117827e-01 -9.54405069e-01 -9.70657915e-02 6.28535211e-01
-1.98620752e-01 1.17111957e+00 1.70253236e-02 -3.86822373e-01
-5.04315019e-01 -5.54481804e-01 3.53943616e-01 5.60462065e-02
-1.79599270e-01 -2.84207880e-01 -8.99281561e-01 -3.47589433e-01
-7.58733079e-02 -2.34203860e-01 3.42728049e-01 4.89684433e-01
6.97831213e-01 -7.61040002e-02 -5.28689444e-01 4.22221392e-01
1.02341974e+00 -1.52157620e-01 4.71579880e-01 3.14333946e-01
8.02460909e-01 7.89045036e-01 9.68298197e-01 8.64406466e-01
2.65424192e-01 1.06964612e+00 5.15759408e-01 9.72520094e-03
-1.78332612e-01 -3.26904118e-01 4.66563016e-01 4.82872486e-01
-7.07020521e-01 -1.46304876e-01 -9.85016704e-01 3.04247320e-01
-1.95272171e+00 -8.00343156e-01 -3.40283737e-02 2.15465355e+00
4.04758185e-01 2.44211465e-01 5.18085897e-01 1.51168406e-01
7.05337107e-01 1.04270987e-01 -4.24574286e-01 3.07228208e-01
2.12619290e-01 -4.49247509e-02 3.43670040e-01 2.28433982e-01
-1.40677655e+00 8.83487761e-01 6.57007122e+00 4.72627282e-01
-5.27416229e-01 -3.42275836e-02 4.44758497e-02 -3.28436077e-01
2.57582575e-01 -1.29461661e-01 -7.54891098e-01 1.03566356e-01
3.79295886e-01 4.46802557e-01 1.10822320e-02 7.70812571e-01
2.44522691e-01 -3.83275181e-01 -1.19466317e+00 1.20980072e+00
3.71178061e-01 -4.99510616e-01 -2.42218062e-01 2.81811863e-01
5.07689178e-01 -4.54073817e-01 -1.77897587e-01 1.70320526e-01
-5.82295507e-02 -7.00170040e-01 1.00183892e+00 7.50641644e-01
5.94203353e-01 -6.26327455e-01 4.13319111e-01 6.52438104e-01
-1.32610297e+00 4.76648584e-02 1.44205958e-01 -4.03838426e-01
5.49891174e-01 2.18126237e-01 -6.79981709e-01 6.97416961e-01
6.73076034e-01 8.22549224e-01 -4.61884171e-01 9.88327801e-01
-3.35226983e-01 7.48862028e-02 -5.70242107e-01 2.97043681e-01
-5.93482330e-02 -1.07574314e-01 8.10639262e-01 1.10230923e+00
1.42858088e-01 1.92183599e-01 8.26627672e-01 3.91939610e-01
4.12700415e-01 -3.49293686e-02 -5.03368974e-01 4.76155058e-02
-7.29311407e-02 1.11132801e+00 -1.05624509e+00 -1.09000929e-01
-3.91545683e-01 1.42494678e+00 -1.22571573e-01 2.77243733e-01
-9.33586657e-01 1.57865062e-02 5.00312090e-01 4.06996161e-01
3.54014397e-01 -7.13109314e-01 5.38385175e-02 -1.35791874e+00
1.85941160e-01 -8.31336617e-01 5.29982746e-01 -6.22228563e-01
-8.87116253e-01 2.72965878e-01 6.28082156e-01 -1.44456530e+00
-5.52876532e-01 -7.33119965e-01 -1.67915329e-01 4.98477072e-01
-9.79304671e-01 -1.47835004e+00 -2.82819539e-01 7.24633753e-01
8.07647645e-01 2.08162040e-01 6.72935009e-01 1.02404013e-01
-4.21676219e-01 3.60449553e-01 -5.59660792e-01 1.89563558e-01
7.61581659e-01 -1.18181348e+00 3.06689501e-01 1.04295456e+00
1.68085158e-01 4.98091727e-01 8.63704681e-01 -9.33783412e-01
-1.70015216e+00 -5.94730854e-01 5.56825280e-01 -1.18191683e+00
3.80474448e-01 -6.91770196e-01 -2.32030064e-01 7.67823637e-01
-2.75670707e-01 2.84932584e-01 4.99277443e-01 1.39781768e-02
7.90387541e-02 3.83007735e-01 -1.20567405e+00 6.60719931e-01
1.54344893e+00 -2.47239009e-01 -6.70904934e-01 2.65521199e-01
2.65849024e-01 -9.50722218e-01 -9.77890968e-01 6.64782763e-01
8.40040505e-01 -9.58012104e-01 1.28638542e+00 -3.52574646e-01
3.45518440e-02 -3.76983464e-01 -3.00550252e-01 -6.93993270e-01
-3.65880951e-02 -7.13046610e-01 -7.33250260e-01 7.48088777e-01
-2.63994366e-01 -1.47439763e-01 1.05672026e+00 6.82134211e-01
1.65479425e-02 -5.67595720e-01 -1.07322681e+00 -8.33451569e-01
-4.70280081e-01 -4.49231684e-01 1.80851027e-01 5.27999222e-01
-1.58032730e-01 -3.61959711e-02 -6.88544989e-01 2.54209608e-01
7.11832345e-01 -1.16204128e-01 1.21019340e+00 -1.32109070e+00
-4.70161140e-01 -6.11265702e-03 -9.60462034e-01 -1.24167645e+00
1.22398555e-01 -3.32919985e-01 2.63146967e-01 -1.24743748e+00
1.08128227e-01 2.76462227e-01 1.64100036e-01 3.06909710e-01
-1.78445891e-01 4.92017388e-01 3.62037867e-01 9.82255265e-02
-9.17402208e-01 4.30632949e-01 1.41399455e+00 2.12418258e-01
-7.84592982e-03 2.64996383e-02 2.21136473e-02 1.07448483e+00
3.07348400e-01 -2.14629069e-01 -3.58258843e-01 -9.34251919e-02
-1.97598994e-01 3.06572288e-01 7.50648618e-01 -1.23771644e+00
9.68129188e-03 -1.89534053e-01 9.23669636e-01 -6.73866212e-01
8.33207250e-01 -9.75009739e-01 2.90486723e-01 5.00667989e-01
6.54328540e-02 6.95717633e-02 1.30398557e-01 8.70382309e-01
-5.43323010e-02 1.23611607e-01 4.21435922e-01 -7.21109986e-01
-1.11210847e+00 4.39034730e-01 -1.79829612e-01 4.27705012e-02
1.36094475e+00 -8.52752328e-01 5.10860384e-01 -3.02109599e-01
-1.07515109e+00 1.07468367e-01 6.37711942e-01 4.33217168e-01
6.91839337e-01 -1.17396355e+00 -5.70633531e-01 2.72489727e-01
1.47437140e-01 1.14134140e-01 2.82143891e-01 1.02773857e+00
-5.10339022e-01 3.08116466e-01 -3.16893905e-01 -9.76242125e-01
-1.52924287e+00 4.36520070e-01 2.90033549e-01 -1.86968729e-01
-8.52373004e-01 6.67524099e-01 6.97362721e-02 -3.61581922e-01
2.94550985e-01 -9.47657824e-02 -3.58662792e-02 1.79689366e-03
2.31443927e-01 5.10256410e-01 -1.70790493e-01 -1.05062342e+00
-5.61783195e-01 8.24228883e-01 1.93403006e-01 -2.63326824e-01
1.06572664e+00 -3.61894459e-01 2.72864312e-01 3.79862249e-01
9.43016827e-01 -1.04948178e-01 -1.86695433e+00 4.10347395e-02
6.04357105e-03 -8.59166443e-01 -3.13360363e-01 -6.00001872e-01
-5.81018388e-01 6.75862670e-01 5.15709996e-01 -3.03801268e-01
9.51893926e-01 1.90958362e-02 6.82573259e-01 2.65993387e-01
8.38419735e-01 -1.42638886e+00 3.96071970e-01 4.20082718e-01
9.24194813e-01 -1.27445662e+00 3.46610516e-01 -6.24630988e-01
-6.76294744e-01 1.08407402e+00 6.94647074e-01 -1.47210464e-01
3.84357095e-01 1.21445335e-01 4.97836135e-02 -1.79822221e-01
-1.82339996e-01 -5.15081108e-01 7.20044553e-01 6.91266119e-01
4.11776394e-01 -3.72597091e-02 -1.82848305e-01 2.00676978e-01
7.45425373e-02 -3.57836578e-03 1.39547318e-01 1.28809571e+00
-1.55412763e-01 -9.90262210e-01 -7.40815461e-01 1.07682599e-02
-4.89126563e-01 5.36060035e-01 -4.97789949e-01 1.17808700e+00
2.97161579e-01 8.37020099e-01 -2.96967745e-01 -2.79571623e-01
6.11105144e-01 1.76007479e-01 1.00664294e+00 -7.41084218e-01
-3.83813679e-01 5.62282085e-01 1.50251940e-01 -9.46668744e-01
-8.75238597e-01 -1.06226516e+00 -9.83560801e-01 1.69282317e-01
-2.92247891e-01 -4.06208694e-01 3.42400581e-01 1.06669128e+00
4.37794961e-02 3.19003224e-01 4.12038807e-03 -1.62107992e+00
-5.04508734e-01 -7.46508539e-01 -6.77400768e-01 6.34925485e-01
2.68979460e-01 -1.09071362e+00 -1.07319362e-01 1.86677516e-01] | [7.098392963409424, -0.9059575200080872] |
2d081236-9c51-4b4d-9065-e42fc6529315 | the-dku-dukeece-diarization-system-for-the | 2210.01677 | null | https://arxiv.org/abs/2210.01677v1 | https://arxiv.org/pdf/2210.01677v1.pdf | The DKU-DukeECE Diarization System for the VoxCeleb Speaker Recognition Challenge 2022 | This paper discribes the DKU-DukeECE submission to the 4th track of the VoxCeleb Speaker Recognition Challenge 2022 (VoxSRC-22). Our system contains a fused voice activity detection model, a clustering-based diarization model, and a target-speaker voice activity detection-based overlap detection model. Overall, the submitted system is similar to our previous year's system in VoxSRC-21. The difference is that we use a much better speaker embedding and a fused voice activity detection, which significantly improves the performance. Finally, we fuse 4 different systems using DOVER-lap and achieve 4.75 of the diarization error rate, which ranks the 1st place in track 4. | ['Ming Li', 'Kangyue Wang', 'Yucong Zhang', 'Ming Cheng', 'Xiaoyi Qin', 'Weiqing Wang'] | 2022-10-04 | null | null | null | null | ['activity-detection'] | ['computer-vision'] | [-8.26353282e-02 2.48894513e-01 -1.19340330e-01 -4.97754902e-01
-1.32412410e+00 -4.83479172e-01 6.20099425e-01 -1.97208717e-01
-4.78725016e-01 3.01411539e-01 7.26377666e-01 -2.35867679e-01
3.17028522e-01 2.75783271e-01 -3.13892394e-01 -5.99323869e-01
-8.51628035e-02 3.16199630e-01 8.99305567e-02 -1.39153242e-01
-2.98711389e-01 2.81734973e-01 -1.21852612e+00 1.79696172e-01
4.43865478e-01 8.15589070e-01 -4.14648920e-01 1.15117979e+00
2.09680140e-01 4.41344500e-01 -7.99997270e-01 -7.22277910e-02
6.93534166e-02 -6.49542987e-01 -6.80723548e-01 -2.86152303e-01
6.51468694e-01 -9.43221152e-02 -5.78228354e-01 6.87901795e-01
9.61242139e-01 2.58202970e-01 5.23148835e-01 -1.25510299e+00
-3.32734197e-01 1.02862298e+00 -3.01187217e-01 6.75559402e-01
5.51316142e-01 -7.85964280e-02 1.12186718e+00 -1.13616276e+00
3.43742877e-01 1.38350725e+00 6.14388049e-01 8.91502202e-01
-1.16334331e+00 -7.56276608e-01 1.11899354e-01 3.74233544e-01
-1.71043634e+00 -1.26736557e+00 6.63484752e-01 -2.65833437e-01
1.13424921e+00 8.43420804e-01 5.74233174e-01 1.47076988e+00
-5.52663565e-01 1.29267883e+00 6.85748696e-01 -4.00907815e-01
4.00981218e-01 1.71942294e-01 3.02573740e-01 1.60218894e-01
-3.90362650e-01 2.96384811e-01 -6.39649451e-01 -3.41396928e-01
2.32869789e-01 -6.00865364e-01 -3.36356133e-01 2.92883694e-01
-1.22469103e+00 7.43542731e-01 -5.98321222e-02 5.03770173e-01
-3.54095809e-02 3.18591855e-02 4.06649679e-01 4.18193549e-01
4.78546232e-01 1.06178567e-01 -1.93314910e-01 -4.11914557e-01
-1.44845390e+00 1.86168134e-01 1.12159383e+00 8.97422910e-01
-1.75846796e-02 6.07868671e-01 -4.11890417e-01 1.19066000e+00
7.97734380e-01 5.40061295e-01 7.21461296e-01 -8.60866845e-01
2.69211382e-01 -2.61836141e-01 -2.86253393e-01 -1.79621890e-01
-1.53819546e-01 -4.66664284e-01 -4.56255615e-01 -1.69583544e-01
1.28395796e-01 -3.01167965e-01 -8.05521488e-01 1.65792310e+00
2.31573239e-01 5.43855369e-01 1.10160530e-01 9.02185619e-01
1.00290275e+00 7.96101034e-01 -1.57743156e-01 -3.27788740e-01
1.36912489e+00 -1.13131070e+00 -1.01170266e+00 -5.58130071e-03
2.88354933e-01 -9.35483992e-01 4.55719292e-01 4.98257935e-01
-1.08455944e+00 -4.65896398e-01 -1.25975621e+00 1.95787385e-01
-7.12557584e-02 -5.51893264e-02 1.07122988e-01 1.03568745e+00
-1.37399399e+00 1.22084599e-02 -6.50095642e-01 -3.16590548e-01
-8.47608298e-02 2.02539846e-01 -1.01297997e-01 3.55755389e-01
-1.44515455e+00 6.84936464e-01 1.03036091e-01 -7.90891796e-02
-8.74107361e-01 -8.99385393e-01 -7.77554691e-01 -1.89533755e-01
6.39378577e-02 3.43555585e-02 1.67075288e+00 -4.17439759e-01
-2.03796363e+00 8.80006015e-01 -4.38549131e-01 -7.40893304e-01
6.99611485e-01 -1.09793894e-01 -1.09854400e+00 -7.27012530e-02
-2.53960371e-01 6.98297441e-01 7.79095888e-01 -7.14513958e-01
-6.22683048e-01 -9.68985632e-02 -6.47730708e-01 2.83837378e-01
-1.06426880e-01 4.69576359e-01 -7.47067809e-01 -8.36125195e-01
-8.50052834e-02 -8.53865623e-01 2.42478698e-01 -5.28793156e-01
-6.77322209e-01 -7.65462756e-01 8.62165630e-01 -9.11256552e-01
1.44789469e+00 -2.75152564e+00 9.45180133e-02 2.30282113e-01
2.85031080e-01 2.94300616e-01 -1.79956034e-01 3.54220361e-01
-3.29289347e-01 1.20656028e-01 5.47299273e-02 -8.86982918e-01
4.24661636e-01 -5.41184396e-02 -3.93074185e-01 5.75706899e-01
-4.80500683e-02 6.59974277e-01 -6.77679300e-01 -2.98268974e-01
1.42347485e-01 5.29647350e-01 -4.88661468e-01 2.47231156e-01
1.90057293e-01 9.99656469e-02 8.91334265e-02 6.35589719e-01
6.16360366e-01 4.62326884e-01 2.80999690e-02 -2.09177863e-02
-3.28267574e-01 6.81648374e-01 -1.21603513e+00 1.51320004e+00
-1.16962396e-01 1.04540479e+00 3.89738023e-01 -5.59037983e-01
7.78358281e-01 9.27570462e-01 5.79957366e-01 -3.01039159e-01
6.22062795e-02 1.96057260e-01 9.47791785e-02 -1.59588218e-01
4.18106765e-01 7.41859153e-02 -9.57022235e-02 2.74668604e-01
2.14469463e-01 -1.87587216e-01 -1.22490317e-01 3.26327831e-01
1.08942699e+00 -5.25382400e-01 7.66232982e-02 -3.58968139e-01
4.72932667e-01 -6.04964793e-01 6.09544516e-01 6.32196546e-01
-8.87375295e-01 8.18495393e-01 1.67513222e-01 3.18831563e-01
-7.15611815e-01 -1.30393600e+00 -3.96827370e-01 1.02845478e+00
-3.61400187e-01 -5.51440835e-01 -8.94635797e-01 -6.14493072e-01
2.76810247e-02 8.88307512e-01 -5.47827959e-01 -9.35577694e-03
-6.74317002e-01 -2.94774711e-01 1.51160955e+00 4.91848260e-01
9.17939097e-02 -7.05998361e-01 1.64342478e-01 1.38812110e-01
-3.23651940e-01 -1.13365817e+00 -1.33504593e+00 2.74246544e-01
-4.55349386e-01 -4.76271749e-01 -7.94104636e-01 -8.98616374e-01
-2.51842558e-01 1.63854271e-01 5.97574472e-01 -5.34456313e-01
-2.00655133e-01 3.72607440e-01 -2.07956031e-01 -4.49489504e-01
-5.63604712e-01 9.93543863e-02 5.35020530e-01 2.71586217e-02
7.44658411e-01 -1.87897995e-01 -3.17308635e-01 3.13923717e-01
-3.53710532e-01 -5.01858354e-01 2.67321259e-01 8.76756430e-01
1.88941777e-01 -4.13752884e-01 8.25443089e-01 -3.43058974e-01
5.99135399e-01 -3.56439412e-01 -4.49157715e-01 9.92581472e-02
-6.00554705e-01 -1.90371022e-01 1.77450478e-02 -4.81146991e-01
-7.94939935e-01 2.61748452e-02 -7.00709283e-01 -5.90215862e-01
-2.70608991e-01 -1.05228778e-02 -4.51409012e-01 3.00856352e-01
5.34534752e-01 2.91170657e-01 -9.76293683e-02 -8.34480166e-01
4.92156625e-01 1.20775402e+00 7.21621394e-01 -4.47936393e-02
4.38102007e-01 -3.64909843e-02 -1.00558913e+00 -1.31719184e+00
-2.85076916e-01 -9.31587219e-01 -2.55715489e-01 2.53363475e-02
8.50975752e-01 -1.30744624e+00 -7.81019568e-01 5.25525570e-01
-1.21273518e+00 -1.01570822e-01 -5.32990992e-01 8.09886754e-01
-1.50176197e-01 3.97507578e-01 -7.36016929e-01 -1.13772118e+00
-4.32265937e-01 -9.80866790e-01 1.00701451e+00 7.59707689e-02
-5.21054745e-01 -7.53217161e-01 6.71564758e-01 6.18501246e-01
5.19115567e-01 -3.41078430e-01 1.72769830e-01 -1.17894709e+00
5.18891327e-02 -1.27495199e-01 2.59647727e-01 6.44830585e-01
7.49839097e-02 2.44473629e-02 -1.58340037e+00 -3.23016852e-01
-1.30461335e-01 2.03499496e-02 9.62478101e-01 3.60266387e-01
6.94057643e-01 -3.40952784e-01 -3.06096643e-01 2.82634109e-01
4.88656521e-01 3.77893865e-01 5.32853127e-01 -3.85317326e-01
4.50383157e-01 4.19591248e-01 1.35757238e-01 3.18277866e-01
3.92144918e-01 1.16684079e+00 -2.13074446e-01 8.44608545e-02
-6.29994571e-01 -7.64682740e-02 9.46084917e-01 1.65362597e+00
2.76367545e-01 -4.19481158e-01 -7.29584217e-01 8.35355818e-01
-1.56454968e+00 -1.23193610e+00 -2.14078277e-01 1.96024120e+00
1.07688975e+00 -1.86157942e-01 6.94175124e-01 3.88351351e-01
7.14701593e-01 3.04761112e-01 -4.15644348e-01 -5.77763438e-01
-2.86050528e-01 2.16797233e-01 2.13884026e-01 9.33360636e-01
-1.07286322e+00 9.15733516e-01 8.22336960e+00 8.88328552e-01
-1.08596659e+00 4.39084142e-01 6.94858357e-02 -5.82103550e-01
-2.38965839e-01 -5.22956431e-01 -1.08670640e+00 5.44813514e-01
1.63382745e+00 -2.58831948e-01 5.01187682e-01 7.03111708e-01
2.15399697e-01 2.46180862e-01 -1.23510683e+00 1.26062799e+00
5.96739292e-01 -8.94811273e-01 -3.70574743e-01 2.14597344e-01
3.32995385e-01 4.71957445e-01 1.55467004e-01 5.54563701e-01
3.92031014e-01 -9.98275340e-01 9.24289882e-01 3.37142587e-01
7.67901778e-01 -6.59079194e-01 5.60622990e-01 -1.22309349e-01
-1.26604640e+00 2.16183513e-01 1.35745332e-01 5.54081142e-01
1.59163639e-01 4.99733895e-01 -1.31592441e+00 4.22185689e-01
5.89252353e-01 4.16659504e-01 -2.23625541e-01 1.11658514e+00
-1.82979062e-01 1.31765759e+00 -5.38094699e-01 -1.56146094e-01
-1.68620914e-01 3.80164057e-01 1.17187440e+00 1.68691599e+00
9.65668261e-03 -2.39746690e-01 -5.25449961e-02 4.77878839e-01
-3.57631624e-01 -2.38822447e-03 -2.15701640e-01 -1.56726465e-01
9.02233124e-01 1.02330387e+00 9.21116397e-02 -5.13568223e-01
-1.01462409e-01 1.06244898e+00 -3.21976133e-02 3.98811609e-01
-8.78255665e-01 -6.20171905e-01 1.05575418e+00 -3.46522272e-01
3.14557642e-01 -1.17536880e-01 -1.15895011e-01 -1.15254855e+00
-2.37423077e-01 -8.85713100e-01 2.55270571e-01 -2.99989969e-01
-1.19237626e+00 8.47014606e-01 -8.79969597e-02 -8.29834998e-01
-5.58938205e-01 -2.34122977e-01 -8.20685387e-01 8.94725263e-01
-1.07361567e+00 -8.37638617e-01 2.14446217e-01 4.82944995e-01
7.18443334e-01 -4.28426445e-01 9.04835165e-01 6.60907865e-01
-8.69302392e-01 1.44677162e+00 2.18280613e-01 2.41123304e-01
1.06231189e+00 -1.40375483e+00 6.44585311e-01 8.11890185e-01
4.08737421e-01 4.90165919e-01 8.24599862e-01 -1.17126837e-01
-1.27200878e+00 -1.06588244e+00 1.25963783e+00 -7.49624074e-01
6.90730393e-01 -8.04229140e-01 -7.27555871e-01 6.30245805e-01
6.73617303e-01 -1.59565300e-01 1.03474402e+00 3.86769205e-01
-5.38053155e-01 -2.30751559e-01 -8.93894494e-01 3.88668925e-01
8.11771810e-01 -8.98669481e-01 -1.12250149e+00 9.48470011e-02
9.62210596e-01 -1.70602039e-01 -8.89313757e-01 1.13471635e-01
6.93224967e-01 -3.47203195e-01 7.74951875e-01 -3.67395997e-01
-5.73876023e-01 -2.92736322e-01 -5.07730365e-01 -1.41243863e+00
-2.94339180e-01 -8.98529589e-01 -4.49050993e-01 1.61943316e+00
5.86950958e-01 -5.08972466e-01 4.07903612e-01 1.13551602e-01
-3.66109043e-01 -2.99833030e-01 -1.61310124e+00 -1.18409872e+00
6.16206303e-02 -6.35275126e-01 3.72586697e-01 9.90838528e-01
3.13939989e-01 5.05049348e-01 -4.05198991e-01 6.78702164e-03
5.69974184e-01 -5.03781617e-01 5.54556131e-01 -1.16297400e+00
-4.48300689e-01 -6.85963631e-01 -3.83038849e-01 -1.22976398e+00
-4.21853550e-03 -9.14400697e-01 3.19873869e-01 -1.01419699e+00
-1.59488827e-01 3.29401693e-03 -6.73134565e-01 5.05074084e-01
2.32071374e-02 3.78729045e-01 2.50453711e-01 1.52348563e-01
-6.00796700e-01 6.08916283e-01 3.83801073e-01 -3.13888013e-01
-5.21847904e-01 -2.71280156e-03 -6.42791033e-01 1.59931213e-01
6.97789252e-01 -2.73105621e-01 -4.64231409e-02 -1.36387749e-02
-6.05138838e-01 -9.02140364e-02 1.14806250e-01 -8.51534843e-01
2.84827501e-01 3.75940531e-01 -5.80183230e-03 -6.91418290e-01
6.82318628e-01 -1.13022417e-01 -2.99513023e-02 4.13608551e-01
-3.55434358e-01 -3.57600719e-01 4.60648924e-01 4.66156632e-01
-3.13533157e-01 2.72106797e-01 7.55224049e-01 5.46124518e-01
-3.19307119e-01 -9.78993066e-03 -8.92745972e-01 9.77610350e-02
7.41043746e-01 3.30674760e-02 -1.85485855e-01 -3.12465966e-01
-9.21748817e-01 2.49105632e-01 -2.68091857e-01 9.47793186e-01
4.59077686e-01 -1.37383723e+00 -1.17489934e+00 3.44189256e-01
2.28768915e-01 -7.65497208e-01 2.34581143e-01 1.05300808e+00
8.00879374e-02 6.47517681e-01 4.56617713e-01 -7.34880507e-01
-1.53688538e+00 2.72439811e-02 3.22758287e-01 1.71868950e-01
-3.96069854e-01 1.17633033e+00 -1.27217248e-01 -4.70207423e-01
6.61048591e-01 -3.75618190e-01 -1.36393651e-01 1.77376971e-01
9.03384209e-01 5.66451430e-01 2.46680692e-01 -8.25037718e-01
-9.86700535e-01 1.74810849e-02 -2.99450904e-01 -8.08273375e-01
8.13221872e-01 -1.34593651e-01 4.26748067e-01 7.72537887e-01
1.28292310e+00 4.91405994e-01 -8.39829266e-01 -2.53073841e-01
1.16169542e-01 8.60585421e-02 4.41022217e-01 -9.78855133e-01
-6.65457845e-01 8.92621398e-01 9.34063911e-01 1.87476143e-01
7.33934164e-01 1.51707649e-01 9.28285360e-01 4.93692011e-02
-2.51558721e-01 -1.28788137e+00 -2.44024172e-01 7.74784505e-01
1.12001836e+00 -9.19569910e-01 -4.21565741e-01 -1.75138548e-01
-6.23760819e-01 7.07882762e-01 3.92183006e-01 2.33535916e-01
8.63592684e-01 4.11428362e-01 3.06049973e-01 3.08706671e-01
-8.76084805e-01 -6.89149052e-02 5.98248124e-01 5.44885814e-01
6.38356924e-01 3.94107550e-01 -1.19662412e-01 8.35047245e-01
-4.28558826e-01 -5.36297441e-01 8.46359655e-02 5.03301680e-01
-3.41415912e-01 -1.04140127e+00 -4.06691760e-01 -8.53951052e-02
-3.26135784e-01 -2.21753046e-01 -7.92630136e-01 3.26318949e-01
-1.92963645e-01 1.45428038e+00 8.29294175e-02 -7.72148132e-01
4.72032815e-01 5.80716133e-01 2.57676989e-01 -5.52925706e-01
-8.05282414e-01 6.76418245e-01 3.00183833e-01 -5.55267513e-01
-5.70289008e-02 -9.40399885e-01 -1.00938416e+00 -4.52388048e-01
-4.13670450e-01 5.36622763e-01 9.06531274e-01 7.85386562e-01
5.61458051e-01 7.76591659e-01 7.85039723e-01 -4.72950399e-01
-7.19870567e-01 -1.47103000e+00 -7.31035292e-01 -1.62751302e-01
7.24116683e-01 -3.19984525e-01 -6.08484626e-01 -1.63698897e-01] | [14.470669746398926, 5.986634731292725] |
be19a46d-b8b9-4e7a-a3e8-220b146417b4 | improving-rnn-transducer-with-target-speaker | 2011.13393 | null | https://arxiv.org/abs/2011.13393v2 | https://arxiv.org/pdf/2011.13393v2.pdf | Improving RNN Transducer With Target Speaker Extraction and Neural Uncertainty Estimation | Target-speaker speech recognition aims to recognize target-speaker speech from noisy environments with background noise and interfering speakers. This work presents a joint framework that combines time-domain target-speaker speech extraction and Recurrent Neural Network Transducer (RNN-T). To stabilize the joint-training, we propose a multi-stage training strategy that pre-trains and fine-tunes each module in the system before joint-training. Meanwhile, speaker identity and speech enhancement uncertainty measures are proposed to compensate for residual noise and artifacts from the target speech extraction module. Compared to a recognizer fine-tuned with a target speech extraction model, our experiments show that adding the neural uncertainty module significantly reduces 17% relative Character Error Rate (CER) on multi-speaker signals with background noise. The multi-condition experiments indicate that our method can achieve 9% relative performance gain in the noisy condition while maintaining the performance in the clean condition. | ['Dong Yu', 'Meng Yu', 'Shinji Watanabe', 'Chao Weng', 'Chunlei Zhang', 'Jiatong Shi'] | 2020-11-26 | null | null | null | null | ['speech-extraction'] | ['speech'] | [ 4.60826367e-01 2.28921577e-01 2.44984940e-01 -5.03330410e-01
-1.48038280e+00 -2.65848607e-01 1.93429410e-01 -5.53169429e-01
-2.41199777e-01 2.56971657e-01 4.28071558e-01 -4.45925057e-01
2.19936639e-01 5.80766499e-02 -4.29144889e-01 -8.52134466e-01
1.17240779e-01 -3.99329737e-02 -4.63209860e-03 -8.69232640e-02
-1.27349764e-01 3.51549774e-01 -1.50840092e+00 2.46034160e-01
8.27376008e-01 8.08043361e-01 4.08913851e-01 1.04261398e+00
-6.26791120e-02 6.92973495e-01 -1.16884243e+00 -2.84801833e-02
9.20307711e-02 -5.59105515e-01 -3.15921754e-01 2.00922921e-01
2.77758062e-01 -9.82140452e-02 -4.89258975e-01 1.23960495e+00
1.00526273e+00 4.41089869e-01 4.14565086e-01 -7.76402295e-01
-3.23054433e-01 1.22204411e+00 -2.76553154e-01 4.10773307e-01
2.42136255e-01 -5.80017567e-02 5.33138633e-01 -1.04502213e+00
-2.27676034e-01 1.46607602e+00 5.57892084e-01 8.67079377e-01
-1.08352995e+00 -9.81063008e-01 1.56237930e-01 1.69627778e-02
-1.57630754e+00 -1.41304016e+00 8.68167818e-01 4.62917937e-03
1.32085598e+00 5.86264551e-01 -1.14216968e-01 1.33132184e+00
-1.40518457e-01 6.76268041e-01 6.58925235e-01 -6.27841294e-01
2.34611407e-01 2.05739930e-01 2.42770731e-01 1.85276031e-01
-2.70653754e-01 5.39346516e-01 -7.69796729e-01 -8.90879855e-02
3.99151504e-01 -6.42176151e-01 -4.19780403e-01 4.16167706e-01
-8.56009126e-01 4.34077889e-01 -1.63797975e-01 6.46333873e-01
-4.49783236e-01 -1.42154843e-01 3.36650133e-01 4.71684247e-01
5.45996487e-01 6.96339086e-02 -4.98928159e-01 -3.16714495e-01
-1.26128602e+00 -5.44996202e-01 8.88482809e-01 1.11785245e+00
1.75590053e-01 1.16142154e+00 -1.81458384e-01 1.27298331e+00
5.72199881e-01 9.27955985e-01 9.20459270e-01 -5.12195647e-01
4.71779704e-01 -5.56388125e-02 -2.11650059e-01 -4.07558978e-01
-4.57021259e-02 -7.60029674e-01 -8.05547059e-01 -4.64443117e-02
2.46653352e-02 -4.19253647e-01 -1.05266666e+00 1.67554426e+00
2.40241751e-01 4.97735143e-01 6.99968278e-01 6.32392824e-01
7.15386927e-01 1.11172163e+00 -2.92511523e-01 -9.06885743e-01
1.11505580e+00 -1.10335350e+00 -1.21323085e+00 -5.32257736e-01
1.24226034e-01 -9.62372303e-01 8.21791172e-01 4.08595473e-01
-1.20635545e+00 -7.27743983e-01 -1.16120315e+00 3.99699003e-01
1.65934227e-02 3.59163284e-01 -4.92228270e-01 1.31499195e+00
-9.63084936e-01 -6.61229119e-02 -6.92936242e-01 1.00847781e-01
-2.86317110e-01 4.23863977e-01 -5.19441217e-02 3.93163443e-01
-1.32162488e+00 1.02306461e+00 1.74447104e-01 3.15483004e-01
-1.09200466e+00 -4.11077619e-01 -8.55869949e-01 2.18767583e-01
2.87894636e-01 -8.62073675e-02 1.63164222e+00 -8.86889100e-01
-2.40511203e+00 2.85722286e-01 -6.95756733e-01 -5.59436202e-01
1.35176599e-01 -2.08960623e-01 -1.16435528e+00 -1.53774634e-01
-3.46785754e-01 -2.62545031e-02 1.49822390e+00 -1.11200309e+00
-6.44926012e-01 -2.13940293e-01 -9.04922009e-01 2.86908239e-01
-3.25271279e-01 5.64446270e-01 -3.14422786e-01 -7.60355890e-01
3.48683655e-01 -6.11938417e-01 -8.93041119e-02 -9.66107905e-01
-4.94540393e-01 -5.06039299e-02 9.99236405e-01 -9.29059803e-01
1.56111836e+00 -2.41989160e+00 -1.27309235e-02 3.83674592e-01
-4.29107487e-01 5.65816462e-01 -1.45171955e-01 -2.99677271e-02
-2.42588460e-01 -3.25971097e-02 -1.47938237e-01 -7.44355917e-01
-6.46779090e-02 4.40825485e-02 -4.13013369e-01 2.76015103e-01
1.75049320e-01 4.54661876e-01 -3.73931140e-01 -2.91422457e-01
3.01328152e-01 8.72252643e-01 -4.19356525e-02 3.62818420e-01
2.55224615e-01 3.74641538e-01 2.42560748e-02 5.26778579e-01
5.93401432e-01 4.42587078e-01 1.90548778e-01 -1.21422699e-02
-1.29903004e-01 8.01361263e-01 -1.45894992e+00 1.24585927e+00
-5.87072730e-01 6.12205029e-01 8.53176773e-01 -6.72343552e-01
1.24026620e+00 9.35068727e-01 -1.69359908e-01 -5.73408067e-01
3.77446383e-01 3.74105781e-01 2.57385343e-01 -3.22903365e-01
4.91616040e-01 -3.11858922e-01 2.74924487e-02 1.28505036e-01
1.63930491e-01 -1.26300320e-01 -5.30020773e-01 -1.91933587e-01
8.76818836e-01 -5.18079638e-01 1.13705873e-01 -1.32172123e-01
9.43511605e-01 -8.68587077e-01 6.52796030e-01 6.41965985e-01
-3.88172328e-01 4.12317783e-01 -2.35817865e-01 4.04051244e-01
-7.28071630e-01 -1.01531911e+00 2.95723532e-03 1.36166155e+00
-2.19605252e-01 -5.89511953e-02 -9.27066982e-01 -1.75706029e-01
-4.05493200e-01 1.21972001e+00 -7.24651143e-02 -3.02307397e-01
-7.50456452e-01 -3.72672260e-01 1.23747051e+00 4.13178831e-01
1.65984407e-01 -8.00902069e-01 7.02574998e-02 4.04933959e-01
-3.56004059e-01 -1.06740534e+00 -1.05645931e+00 6.18049085e-01
-4.80639219e-01 -1.63718283e-01 -7.12285757e-01 -1.05054080e+00
4.14274722e-01 1.07272834e-01 4.20416504e-01 -5.17883420e-01
2.80735344e-01 1.56982392e-01 -1.53114215e-01 -3.02880585e-01
-1.16421902e+00 -1.61114316e-02 6.96533203e-01 2.33546674e-01
3.10231715e-01 -5.46144426e-01 1.99178234e-01 5.68283916e-01
-5.98723233e-01 -4.50769931e-01 5.15945375e-01 9.00127769e-01
2.10739464e-01 3.53389025e-01 9.69029963e-01 -8.67818445e-02
7.96591580e-01 1.27973221e-02 -6.40319467e-01 3.80528659e-01
-5.22444010e-01 1.22542650e-01 2.82219797e-01 -9.41827297e-01
-1.56155896e+00 1.40250012e-01 -4.49298561e-01 -5.12104094e-01
-1.64061278e-01 3.58956128e-01 -7.86404669e-01 1.80244967e-01
8.48083079e-01 5.59544623e-01 4.10629287e-02 -5.45076311e-01
2.43922189e-01 1.45463336e+00 8.24943900e-01 -2.86016941e-01
8.01399708e-01 -4.25084203e-01 -8.29874277e-01 -1.14575505e+00
-4.19666588e-01 -5.19873798e-01 -2.69732147e-01 -1.03955448e-01
3.11102062e-01 -1.09412992e+00 -6.96393728e-01 7.01456845e-01
-1.23590732e+00 -3.97361554e-02 -8.64259973e-02 8.67670238e-01
-1.37047172e-01 3.66780370e-01 -6.51938915e-01 -1.67124772e+00
-6.90656960e-01 -1.19244897e+00 8.34262013e-01 1.19243391e-01
6.05930714e-03 -4.77274984e-01 -2.10762277e-01 4.57523316e-01
7.94044673e-01 -7.80704975e-01 2.76395500e-01 -1.01925874e+00
-1.20473072e-01 -6.93518147e-02 3.72852504e-01 1.03177679e+00
4.79096383e-01 -3.59266214e-02 -1.63835359e+00 -2.19138354e-01
6.75160468e-01 1.38934553e-01 5.99163115e-01 6.12929761e-01
4.54022944e-01 -5.85558534e-01 -1.24605171e-01 2.02502236e-01
7.34674692e-01 8.48138034e-01 5.77469647e-01 -2.81475365e-01
4.30763483e-01 5.97285450e-01 2.31575027e-01 1.71856508e-01
-7.05494806e-02 5.18527925e-01 -1.26334786e-01 2.20008433e-01
-1.91070542e-01 7.09603727e-02 1.14195728e+00 1.59027326e+00
5.35089493e-01 -4.42193568e-01 -6.95070267e-01 3.86248380e-01
-1.36483860e+00 -1.10500133e+00 2.09822282e-01 2.28436041e+00
1.02132213e+00 3.39466959e-01 3.31536382e-02 4.47739810e-01
1.30576921e+00 5.08374581e-03 -5.87370396e-01 -4.74424601e-01
-3.81827533e-01 4.06392291e-02 3.53061020e-01 1.17220223e+00
-8.50836217e-01 8.78935337e-01 7.09269905e+00 1.04152262e+00
-1.45127952e+00 1.03439100e-01 4.14654344e-01 -2.52830416e-01
-1.02025993e-01 -4.77606356e-01 -1.04253721e+00 2.14878976e-01
1.87639439e+00 -4.34225589e-01 6.05895579e-01 8.31735611e-01
3.88277292e-01 3.31338018e-01 -1.11364055e+00 1.05322611e+00
3.81432146e-01 -6.35839641e-01 -3.15777838e-01 -7.23872110e-02
2.41851479e-01 1.90056920e-01 2.40449369e-01 5.53265333e-01
1.55070692e-01 -9.47033405e-01 8.93438399e-01 2.99384683e-01
6.81998372e-01 -8.22843313e-01 4.95839894e-01 6.04705453e-01
-1.13910532e+00 -1.59954324e-01 -4.65981327e-02 3.07810605e-01
3.14825445e-01 6.65560007e-01 -1.31268489e+00 4.51845258e-01
4.22524661e-01 -1.28105626e-01 -5.23932762e-02 6.65603042e-01
-1.53250307e-01 1.14779186e+00 -5.97247720e-01 -2.93284655e-02
-3.22142005e-01 9.63822752e-02 9.83411431e-01 1.50173485e+00
5.92890501e-01 -1.09871635e-02 -1.54817015e-01 5.87430835e-01
4.90146084e-03 -2.48991866e-02 -2.09035605e-01 -1.20525897e-01
8.71144772e-01 8.94103646e-01 -6.88831806e-02 -4.61973578e-01
2.02139504e-02 1.00050557e+00 -1.26465574e-01 5.37642300e-01
-8.47834885e-01 -5.58079004e-01 6.71252728e-01 -4.23200727e-01
4.53728467e-01 -1.30023733e-01 -3.18225682e-01 -1.04427457e+00
1.47210985e-01 -1.13639796e+00 -1.62878886e-01 -7.19945133e-01
-1.07718253e+00 1.16519630e+00 -3.71303797e-01 -1.07657802e+00
-7.69605279e-01 -2.75967419e-01 -5.07181644e-01 1.26682818e+00
-1.28470409e+00 -8.23872507e-01 5.00280857e-01 5.20457268e-01
8.18951964e-01 -4.34780747e-01 1.00612414e+00 3.99001241e-01
-9.47186530e-01 1.18765843e+00 2.32543245e-01 1.31129786e-01
5.33504725e-01 -9.22503293e-01 5.14920831e-01 1.29161000e+00
4.37399931e-02 8.12191606e-01 8.89533103e-01 -4.99720782e-01
-1.25060558e+00 -9.97625291e-01 1.00049365e+00 -1.55669138e-01
4.92071718e-01 -6.13108277e-01 -1.05962241e+00 4.98069376e-01
3.62034380e-01 -2.76124150e-01 7.20344186e-01 1.02346957e-01
-4.96878386e-01 -2.59179890e-01 -1.14750957e+00 3.98956954e-01
5.72218835e-01 -1.01314282e+00 -1.00568879e+00 -2.96225786e-01
1.07859910e+00 -5.47337651e-01 -6.20010734e-01 3.23937207e-01
3.91810119e-01 -5.41887701e-01 7.42631972e-01 2.66120285e-02
-5.44225931e-01 -5.00646532e-01 -4.83068854e-01 -1.53476882e+00
-1.68412954e-01 -1.15738201e+00 -9.01712626e-02 1.70941556e+00
8.42089117e-01 -6.71058655e-01 4.00855064e-01 6.41050935e-01
-6.17837727e-01 8.49999711e-02 -1.44543028e+00 -1.08523226e+00
-1.92503124e-01 -7.81098545e-01 3.86618406e-01 5.38996160e-01
1.57628730e-01 6.76961958e-01 -5.47577679e-01 8.18534195e-01
4.58177537e-01 -6.37220442e-01 3.12535077e-01 -7.74720490e-01
-3.41332942e-01 -3.33264649e-01 -5.39901145e-02 -1.27909040e+00
3.51092875e-01 -4.29236621e-01 7.76128352e-01 -1.03840399e+00
-6.10461652e-01 1.65852606e-01 -4.87846494e-01 3.34086269e-01
-2.06333071e-01 -3.82334381e-01 -1.19755663e-01 -7.29748458e-02
-1.71509996e-01 7.17212498e-01 5.83720446e-01 -3.09734762e-01
-6.21342540e-01 4.62592751e-01 -6.31025255e-01 4.53058749e-01
6.89756036e-01 -5.60788631e-01 -4.68041509e-01 -1.61291555e-01
-6.70976996e-01 3.36858541e-01 -1.62178487e-01 -1.12073147e+00
5.96890390e-01 1.48961753e-01 8.98066908e-02 -7.94097424e-01
7.46929169e-01 -9.01117146e-01 1.87307313e-01 3.58015835e-01
-5.34910977e-01 -4.35714304e-01 5.37752748e-01 4.30625260e-01
-4.10478592e-01 -1.82410941e-01 8.81707907e-01 3.98485363e-01
-1.39788806e-01 -3.52751613e-01 -9.99255657e-01 -3.83276552e-01
5.15095651e-01 -2.90806115e-01 -6.95520043e-02 -5.82726359e-01
-8.86569321e-01 -7.48397857e-02 -3.16376269e-01 5.68421960e-01
6.99602723e-01 -1.13802063e+00 -8.78769279e-01 5.83641112e-01
-1.67788669e-01 -2.95612872e-01 4.27266955e-01 5.03014684e-01
4.47454631e-01 4.69075531e-01 4.52446550e-01 -5.59248388e-01
-1.65094364e+00 4.34185266e-01 7.07639873e-01 1.39020026e-01
-1.35586321e-01 1.27579045e+00 -8.99341404e-02 -5.68208516e-01
8.18439662e-01 -6.20383620e-01 -6.28785184e-03 -1.89017758e-01
7.86675572e-01 3.46478492e-01 3.83725911e-01 -1.02991009e+00
-5.47698200e-01 8.42824727e-02 -1.73416689e-01 -7.11289287e-01
9.28630233e-01 -5.77690899e-01 2.03266874e-01 7.45098293e-01
1.10308921e+00 4.03686345e-01 -8.36130381e-01 -6.12390578e-01
7.13537931e-02 1.85038634e-02 3.72915059e-01 -9.70580220e-01
-6.77980006e-01 7.18036652e-01 8.61817837e-01 3.32151115e-01
1.24521792e+00 -2.03625053e-01 6.35219991e-01 5.34519017e-01
-3.13595757e-02 -1.21309423e+00 -1.22423694e-01 8.04333746e-01
1.04275393e+00 -8.96600783e-01 -6.96112514e-01 -3.26014668e-01
-7.27031708e-01 8.93949926e-01 4.61043030e-01 3.72176796e-01
8.62589657e-01 7.47898757e-01 5.95609307e-01 3.65692228e-01
-8.38096440e-01 -9.21943784e-02 3.86820614e-01 5.71345687e-01
3.83525133e-01 1.34527117e-01 4.64943349e-01 9.02955413e-01
-5.37834167e-01 -4.48478073e-01 2.70918816e-01 5.92722356e-01
-7.59780467e-01 -9.58013594e-01 -9.37122166e-01 7.50071853e-02
-5.38659453e-01 -4.22259301e-01 -3.10769081e-01 2.42183264e-03
-4.48595554e-01 1.74429071e+00 -1.20310038e-01 -7.52459824e-01
5.57203293e-01 6.26345098e-01 4.51283306e-02 -5.11171103e-01
-9.29758310e-01 1.01757812e+00 2.16182038e-01 -7.06213489e-02
-2.83197075e-01 -6.43692434e-01 -1.34384954e+00 1.52698517e-01
-1.12229085e+00 5.71807921e-01 1.11939549e+00 9.48633373e-01
3.99034470e-01 9.59094107e-01 1.00069773e+00 -6.43518329e-01
-9.97524917e-01 -1.53469586e+00 -6.36048853e-01 -3.43040437e-01
8.33849788e-01 -1.06050327e-01 -8.42789531e-01 1.57902956e-01] | [14.588767051696777, 6.1720428466796875] |
33f1e755-f285-46d5-b1a3-42cfd0c4fbb8 | rethinking-ai-explainability-and-plausibility | 2303.17707 | null | https://arxiv.org/abs/2303.17707v2 | https://arxiv.org/pdf/2303.17707v2.pdf | The XAI Alignment Problem: Rethinking How Should We Evaluate Human-Centered AI Explainability Techniques | Setting proper evaluation objectives for explainable artificial intelligence (XAI) is vital for making XAI algorithms follow human communication norms, support human reasoning processes, and fulfill human needs for AI explanations. In this position paper, we examine the most pervasive human-grounded concept in XAI evaluation, explanation plausibility. Plausibility measures how reasonable the machine explanation is compared to the human explanation. Plausibility has been conventionally formulated as an important evaluation objective for AI explainability tasks. We argue against this idea, and show how optimizing and evaluating XAI for plausibility is sometimes harmful, and always ineffective in achieving model understandability, transparency, and trustworthiness. Specifically, evaluating XAI algorithms for plausibility regularizes the machine explanation to express exactly the same content as human explanation, which deviates from the fundamental motivation for humans to explain: expressing similar or alternative reasoning trajectories while conforming to understandable forms or language. Optimizing XAI for plausibility regardless of the model decision correctness also jeopardizes model trustworthiness, because doing so breaks an important assumption in human-human explanation that plausible explanations typically imply correct decisions, and vice versa; and violating this assumption eventually leads to either undertrust or overtrust of AI models. Instead of being the end goal in XAI evaluation, plausibility can serve as an intermediate computational proxy for the human process of interpreting explanations to optimize the utility of XAI. We further highlight the importance of explainability-specific evaluation objectives by differentiating the AI explanation task from the object localization task. | ['Ghassan Hamarneh', 'Xiaoxiao Li', 'Weina Jin'] | 2023-03-30 | null | null | null | null | ['object-localization'] | ['computer-vision'] | [ 3.51301670e-01 1.12565100e+00 -2.93444782e-01 -6.61985219e-01
-8.13436508e-02 -4.99095052e-01 9.78276134e-01 6.76394925e-02
1.37312198e-02 6.09939098e-01 4.99869287e-01 -8.55212271e-01
-5.66770434e-01 -6.33350492e-01 -6.37010634e-01 -4.87779155e-02
4.77362901e-01 7.77828753e-01 -6.32242322e-01 1.70526773e-01
3.12325150e-01 4.41464394e-01 -1.39418352e+00 2.46210247e-01
1.10531974e+00 8.39709640e-01 -3.31068367e-01 6.00221097e-01
-2.55149417e-02 1.32882130e+00 -4.85837162e-01 -7.69311965e-01
4.44692485e-02 -8.24944913e-01 -1.39900672e+00 -3.95947807e-02
3.15898925e-01 -4.06004667e-01 5.46058863e-02 1.02495229e+00
-4.65053946e-01 -1.73488900e-01 8.35241735e-01 -1.90451121e+00
-1.14131105e+00 1.15863550e+00 1.99320361e-01 -1.94253176e-01
4.22133565e-01 6.01250827e-01 1.39395952e+00 -3.62019718e-01
2.78639942e-01 1.59725559e+00 3.67918491e-01 8.14236701e-01
-1.28234506e+00 -3.73642653e-01 1.87863261e-01 2.47145724e-02
-9.86968696e-01 -4.76499289e-01 3.87451828e-01 -5.12984216e-01
8.63183260e-01 9.49450195e-01 8.77163470e-01 1.10700798e+00
3.78178447e-01 7.11590171e-01 1.14791429e+00 -4.98647273e-01
3.74065876e-01 4.30000991e-01 7.30185270e-01 5.38869381e-01
1.11696219e+00 6.21896446e-01 -6.19766176e-01 -2.17429191e-01
5.34286439e-01 -1.72005832e-01 -5.12584686e-01 4.48800586e-02
-1.53400826e+00 7.61266530e-01 3.58190984e-01 1.49195820e-01
-5.98707080e-01 4.82369751e-01 -5.26797399e-02 4.47143406e-01
-1.42364189e-01 1.30496490e+00 -3.65010798e-01 -1.48117959e-01
-4.71234143e-01 5.38028777e-01 7.87155330e-01 7.03261256e-01
5.80911219e-01 1.42387435e-01 -1.07153706e-01 -1.97852999e-01
8.11253965e-01 5.03415823e-01 3.85524422e-01 -1.44710290e+00
-4.68818396e-02 9.35498357e-01 2.46699899e-01 -8.96604478e-01
-3.91219497e-01 -4.78572637e-01 -5.09555280e-01 6.14910126e-01
5.75571537e-01 1.36691257e-01 -5.02651691e-01 1.92954516e+00
8.60881135e-02 -4.87738699e-01 3.27024460e-01 1.20972002e+00
5.85582137e-01 5.13373733e-01 4.79711682e-01 -7.07229450e-02
1.35083008e+00 -7.80502558e-01 -6.66515410e-01 -5.51200926e-01
5.36879539e-01 -3.38064700e-01 1.65912974e+00 3.67549926e-01
-1.24522853e+00 -3.29363555e-01 -1.19661963e+00 -2.94844985e-01
9.72075611e-02 -3.67952913e-01 1.23495615e+00 6.20914102e-01
-7.74929404e-01 5.32354414e-01 -5.73714077e-01 -1.37835398e-01
3.64235789e-01 2.83644497e-01 -2.76016891e-01 1.73091367e-01
-8.69446695e-01 1.34373987e+00 2.52964616e-01 1.47001334e-02
-7.78580904e-01 -5.49691856e-01 -9.15802300e-01 6.20962203e-01
4.02701259e-01 -1.16981256e+00 1.34638202e+00 -1.69356132e+00
-1.13692248e+00 7.09127665e-01 -2.14601353e-01 -6.62004828e-01
5.97271502e-01 -2.14333147e-01 -3.25141519e-01 -2.33591437e-01
1.74889028e-01 8.63309979e-01 6.83966994e-01 -1.38539541e+00
-3.46438646e-01 -3.93266499e-01 4.11709040e-01 1.61577135e-01
1.45355791e-01 -2.13043988e-01 4.03031200e-01 -2.50680029e-01
3.32118183e-01 -8.99270535e-01 6.98486418e-02 3.00854772e-01
-6.28755450e-01 -1.68768376e-01 2.22751379e-01 -2.90139914e-01
1.06077433e+00 -1.88839591e+00 -2.03290105e-01 4.58955199e-01
7.25564837e-01 -2.62219548e-01 -4.47052009e-02 3.98632102e-02
-1.76500022e-01 7.86061525e-01 -2.80603729e-02 3.87306102e-02
6.94477618e-01 1.71629161e-01 -5.80951631e-01 3.14864755e-01
2.33950481e-01 1.37682617e+00 -7.00998843e-01 -2.83279479e-01
9.58371460e-02 2.87029952e-01 -5.63070059e-01 8.56119990e-02
-3.29607397e-01 3.15149218e-01 -3.91008705e-01 4.67050701e-01
2.15677187e-01 -6.49964571e-01 3.27212326e-02 4.08346727e-02
1.22353509e-02 6.91926718e-01 -7.82748342e-01 9.10289228e-01
-1.23439208e-01 7.97353148e-01 -4.42108452e-01 -4.23011512e-01
5.47580540e-01 3.28259051e-01 -1.72262222e-01 -4.48643297e-01
2.81120628e-01 2.67875433e-01 5.13778448e-01 -3.79764855e-01
2.17353672e-01 -4.98228103e-01 1.12919651e-01 9.98994708e-01
-4.71609116e-01 -5.02738595e-01 -3.44260752e-01 2.67217517e-01
7.48186886e-01 1.07385200e-02 6.69365525e-01 -5.37875116e-01
3.83728176e-01 4.72184330e-01 5.73986650e-01 9.92598414e-01
-1.43960729e-01 5.34355819e-01 6.32233977e-01 -9.77448702e-01
-9.73817706e-01 -9.74671304e-01 -1.75578725e-02 4.33617204e-01
3.92017812e-01 -2.56328791e-01 -8.94524693e-01 -7.78315961e-01
5.97552955e-02 1.83384228e+00 -7.78475404e-01 -4.70991462e-01
-1.43802628e-01 -5.26704006e-02 5.46856523e-01 4.18605089e-01
2.86764532e-01 -1.17101657e+00 -1.33428240e+00 -2.06474483e-01
-2.36030653e-01 -4.80681986e-01 -1.74113125e-01 6.47217482e-02
-5.73125541e-01 -1.08443308e+00 1.09724298e-01 1.47382155e-01
8.04092228e-01 2.91178674e-01 1.33434582e+00 1.02835202e+00
3.35510015e-01 5.13459563e-01 -2.22334623e-01 -8.95730495e-01
-1.02088475e+00 -2.80673951e-01 6.37156591e-02 -4.76667315e-01
5.69932401e-01 -2.79616177e-01 -2.90792912e-01 3.60782027e-01
-9.48427796e-01 5.74556291e-01 2.68097967e-01 5.71387589e-01
2.21536294e-01 -7.75337741e-02 3.05208921e-01 -9.75555539e-01
7.35976636e-01 -2.08961010e-01 -5.27423561e-01 6.25645578e-01
-1.47756445e+00 4.23877716e-01 3.70831013e-01 -2.77095944e-01
-1.07675779e+00 -5.46033025e-01 3.85977119e-01 1.38211429e-01
-2.21744195e-01 4.62922335e-01 -4.52714056e-01 1.48994923e-01
8.79017413e-01 -4.13400799e-01 1.27712712e-01 1.93593130e-01
5.73116720e-01 2.89157420e-01 3.80821615e-01 -8.87742043e-01
8.97725284e-01 3.54625612e-01 -1.18442029e-01 -2.51405060e-01
-9.19220090e-01 4.82481807e-01 -1.25992522e-01 -1.49517655e-01
9.03147638e-01 -3.81281912e-01 -1.25598907e+00 -4.59972471e-01
-1.33253169e+00 -3.25021148e-01 -7.05155492e-01 6.80564880e-01
-7.26278841e-01 1.04717202e-01 -1.06954582e-01 -9.61086631e-01
-3.42849314e-01 -1.33657384e+00 7.05552161e-01 5.61170373e-03
-1.26238036e+00 -7.77705014e-01 -3.36425215e-01 4.44142491e-01
4.48558211e-01 1.78201437e-01 1.36647499e+00 -8.93050671e-01
-6.86385810e-01 -2.28874996e-01 -1.35743558e-01 3.65417600e-02
1.78198814e-02 1.59045711e-01 -1.07239962e+00 3.72022659e-01
3.86328518e-01 -2.59468049e-01 3.70869011e-01 3.82033765e-01
6.84775710e-01 -1.00385678e+00 -2.60705538e-02 2.39198670e-01
1.06441927e+00 1.68149650e-01 6.33966267e-01 5.16008019e-01
4.47868437e-01 1.00076544e+00 4.85434771e-01 2.34515905e-01
4.44167942e-01 3.69284600e-01 4.05919164e-01 -3.20658498e-02
-3.98093648e-02 -4.27752525e-01 1.86116174e-01 2.54802755e-03
-9.48726386e-03 -3.49897861e-01 -1.18820000e+00 1.41760886e-01
-1.92646480e+00 -1.25294781e+00 -4.63816106e-01 2.17635274e+00
5.61255813e-01 2.39758566e-01 -1.77814171e-01 1.26672179e-01
2.80859888e-01 -4.13094491e-01 -5.44056177e-01 -6.99708939e-01
-1.07260711e-01 -4.50664997e-01 -3.15711312e-02 1.02587605e+00
-2.50792027e-01 8.36861789e-01 6.76671982e+00 1.00916445e-01
-7.74502754e-01 -9.69414562e-02 8.88218105e-01 3.69033553e-02
-1.34814644e+00 5.16758800e-01 -1.10920042e-01 6.18959218e-02
8.24422121e-01 -5.46930790e-01 7.05485582e-01 9.54691231e-01
3.17990392e-01 -1.53532149e-02 -1.96008658e+00 5.97467244e-01
-2.41890371e-01 -1.39545512e+00 4.63811576e-01 3.49687010e-01
3.24017465e-01 -7.68010437e-01 2.09606662e-01 -3.79256159e-02
3.75764221e-01 -1.65915799e+00 1.67934716e+00 4.75767165e-01
3.27887326e-01 -4.74280328e-01 6.61405921e-01 4.37331110e-01
-5.10038793e-01 -8.14491510e-02 -3.10194433e-01 -6.37491941e-01
9.19318572e-02 3.80406469e-01 -8.45226586e-01 -7.55433068e-02
9.72036794e-02 1.03488810e-01 -4.97169226e-01 3.43725681e-01
-7.39492714e-01 5.71878552e-01 -9.02887434e-02 3.66047919e-02
2.94877648e-01 -1.13648564e-01 7.08051801e-01 6.90922201e-01
7.61712790e-02 5.27940989e-01 -4.80320632e-01 1.74426985e+00
3.00238878e-01 -3.76902789e-01 -6.78737402e-01 -3.54400426e-01
6.03461087e-01 6.92204475e-01 -5.13327837e-01 -5.42271912e-01
-1.89916715e-01 6.82103336e-01 5.27349971e-02 4.85688388e-01
-1.02406251e+00 4.20680135e-01 7.83964157e-01 1.22030877e-01
-6.73894227e-01 1.74604699e-01 -1.30380893e+00 -1.05138540e+00
7.82506168e-02 -1.37799525e+00 2.15703458e-01 -1.12211800e+00
-9.29900110e-01 7.71378934e-01 1.02509998e-01 -9.06948447e-01
-4.12449986e-01 -5.81261277e-01 -7.58850217e-01 8.64675343e-01
-1.27840078e+00 -1.35118198e+00 -3.70058835e-01 1.02767244e-01
3.87360096e-01 1.45125449e-01 7.50713766e-01 -7.35666275e-01
-3.17187637e-01 3.99963498e-01 -8.11753392e-01 -5.41327357e-01
1.85846552e-01 -1.26163113e+00 7.33714104e-01 9.39811409e-01
3.97213250e-01 1.32664251e+00 1.08003294e+00 -7.91608274e-01
-1.38624918e+00 -6.39418483e-01 1.23708117e+00 -9.35165882e-01
3.73807251e-01 3.10334802e-01 -1.00447500e+00 1.08524776e+00
1.95647374e-01 -6.92084670e-01 5.88318408e-01 2.04981729e-01
-5.42828679e-01 3.25192124e-01 -1.23172915e+00 9.33640897e-01
1.05142164e+00 -4.69999939e-01 -8.64002347e-01 1.62848860e-01
1.03612959e+00 5.80128953e-02 -4.41270858e-01 1.98183700e-01
7.62896955e-01 -1.12069857e+00 6.56778216e-01 -9.10053849e-01
6.50194705e-01 -5.06207466e-01 -2.29162604e-01 -7.97019064e-01
-5.46571136e-01 -7.00872064e-01 1.26045778e-01 1.02804756e+00
8.56811047e-01 -7.78517902e-01 5.36749125e-01 2.14327884e+00
-8.42848569e-02 -2.61837482e-01 -6.11153364e-01 -7.00857699e-01
-5.00565059e-02 -9.12867427e-01 1.38612413e+00 1.01372910e+00
4.78910774e-01 2.39257216e-01 -1.99991614e-02 3.78864110e-01
6.89964235e-01 1.60928041e-01 8.69075298e-01 -1.31953907e+00
-3.62151295e-01 -8.56237769e-01 -2.61247367e-01 -7.14022636e-01
4.58287746e-01 -7.85074770e-01 -9.36229061e-03 -1.51142776e+00
2.67270625e-01 -4.22807813e-01 3.17985296e-01 7.44523764e-01
-2.36104235e-01 -3.70397627e-01 5.72297931e-01 6.23866498e-01
-2.39240944e-01 2.55146980e-01 1.06175292e+00 -2.52122134e-01
2.93902662e-02 -1.72656685e-01 -1.44288146e+00 1.10130823e+00
5.38971126e-01 -3.80837381e-01 -8.04831505e-01 -7.47106314e-01
6.32814467e-01 -3.57705392e-02 8.32404196e-01 -7.66296208e-01
-3.93462786e-03 -8.30790520e-01 2.53860235e-01 2.51988590e-01
9.34283212e-02 -1.18408871e+00 8.69968295e-01 6.10449672e-01
-8.75500500e-01 1.67734340e-01 -5.99234141e-02 1.64037347e-01
1.31552130e-01 -4.22860324e-01 4.61788952e-01 -8.96091536e-02
-4.17511493e-01 -1.73807427e-01 -3.55087101e-01 -2.38449201e-01
6.60529137e-01 -5.71685851e-01 -4.85179335e-01 -5.53240180e-01
-3.67660254e-01 8.34719539e-02 9.90455866e-01 1.10605560e-01
6.95648491e-01 -1.13808596e+00 -7.48751223e-01 1.05473049e-01
2.79293239e-01 -3.45811635e-01 -2.06852183e-01 5.31440377e-01
-4.57932204e-01 5.03415227e-01 -1.41411975e-01 -3.11614037e-01
-8.10469210e-01 6.29151344e-01 4.53132868e-01 2.10223988e-01
-6.17176175e-01 6.75938189e-01 9.09335256e-01 -4.28943485e-02
1.31000042e-01 -6.85650527e-01 1.33198142e-01 -8.59301627e-01
7.06769884e-01 2.87992686e-01 -4.09040481e-01 -3.76329094e-01
-3.65312785e-01 4.87321913e-02 1.85038477e-01 -4.74389106e-01
8.46195579e-01 -3.11546534e-01 -1.77796796e-01 3.03247005e-01
3.28741461e-01 -1.28751650e-01 -9.66133654e-01 2.80685782e-01
1.11445054e-01 -5.47370613e-01 -1.73132364e-02 -1.30623722e+00
-3.43770921e-01 8.14786851e-01 -5.96879795e-02 3.99440408e-01
5.39409816e-01 1.43652633e-01 3.58410776e-01 5.30261397e-01
3.83077353e-01 -4.49571937e-01 -3.09011370e-01 -7.11912215e-02
1.34048116e+00 -1.38332677e+00 2.12975070e-01 -2.93629438e-01
-1.12613487e+00 1.01946306e+00 8.74014497e-01 5.57202041e-01
-6.81044236e-02 -9.50023681e-02 1.63472459e-01 -6.01432681e-01
-8.22427034e-01 2.17299029e-01 5.43707728e-01 5.00148296e-01
5.54572225e-01 4.80156630e-01 -2.77363867e-01 9.71893489e-01
-6.62676215e-01 -2.06728160e-01 5.02233565e-01 3.35965484e-01
-5.36163211e-01 -6.48510695e-01 -5.27377427e-01 2.36996636e-01
-7.69039020e-02 -1.42020538e-01 -1.02436399e+00 1.05168569e+00
-1.09477587e-01 1.28621852e+00 -1.61854252e-01 -1.66339830e-01
-6.85596764e-02 -3.08395061e-03 1.98519498e-01 -3.23293477e-01
-7.07191765e-01 -4.89447117e-01 2.83284873e-01 -7.38474965e-01
-8.07246119e-02 -2.86638290e-01 -1.69590819e+00 -8.31679702e-01
-4.20729905e-01 4.49507236e-01 5.18957555e-01 1.34966981e+00
2.60926872e-01 9.63171944e-02 6.93460554e-02 -1.31897017e-01
-7.23386645e-01 -4.57558364e-01 -6.30724505e-02 5.85330307e-01
4.42911059e-01 -3.60721380e-01 -5.81868827e-01 -2.75872126e-02] | [8.969355583190918, 5.95249605178833] |
75134729-78f0-4b5b-a9a5-3fee56cbbd1a | a-systematic-framework-to-discover-pattern | 1711.06955 | null | http://arxiv.org/abs/1711.06955v1 | http://arxiv.org/pdf/1711.06955v1.pdf | A systematic framework to discover pattern for web spam classification | Web spam is a big problem for search engine users in World Wide Web. They use
deceptive techniques to achieve high rankings. Although many researchers have
presented the different approach for classification and web spam detection
still it is an open issue in computer science. Analyzing and evaluating these
websites can be an effective step for discovering and categorizing the features
of these websites. There are several methods and algorithms for detecting those
websites, such as decision tree algorithm. In this paper, we present a
systematic framework based on CHAID algorithm and a modified string matching
algorithm (KMP) for extract features and analysis of these websites. We
evaluated our model and other methods with a dataset of Alexa Top 500 Global
Sites and Bing search engine results in 500 queries. | ['Jiang Xiaohui', 'Yuan Chi', 'Wang Yongli', 'Jelodar Hamed'] | 2017-11-19 | null | null | null | null | ['spam-detection'] | ['natural-language-processing'] | [-3.66457850e-01 -5.13481677e-01 -1.99543178e-01 -1.99579671e-02
-7.11354196e-01 -1.06372404e+00 7.82269835e-01 1.77303076e-01
-2.96499044e-01 5.92363417e-01 -3.47753875e-02 -5.36992431e-01
-2.94602931e-01 -1.17470098e+00 -1.94827065e-01 -3.88626516e-01
-5.08902445e-02 4.57831085e-01 1.17632520e+00 -4.25577253e-01
1.02863026e+00 4.64663953e-01 -1.47012758e+00 4.75481659e-01
9.87554193e-01 9.34325218e-01 -2.50353009e-01 6.02481484e-01
-6.54807031e-01 5.56829691e-01 -6.83867812e-01 -1.00127351e+00
5.80000162e-01 -4.03711736e-01 -1.02732313e+00 -3.25782061e-01
5.45374870e-01 -1.43410757e-01 -3.93689811e-01 1.63084149e+00
3.42832893e-01 -1.64957345e-02 8.57237041e-01 -1.28625822e+00
-2.98859835e-01 1.14556633e-01 -6.22247636e-01 6.47534132e-01
3.73989820e-01 -2.63221055e-01 9.36390340e-01 -3.61948371e-01
5.78462660e-01 1.49053645e+00 4.98049974e-01 2.46323630e-01
-5.99269867e-01 -7.42968738e-01 -3.28244448e-01 8.50984037e-01
-1.05300343e+00 1.57861680e-01 6.58269167e-01 -2.90739179e-01
5.37467837e-01 6.01862729e-01 3.76025021e-01 8.80748630e-01
1.92927107e-01 6.10320330e-01 1.43571758e+00 -4.92242664e-01
1.97510242e-01 6.08952880e-01 1.07753813e+00 6.90773189e-01
8.59436870e-01 -1.92533955e-02 -2.62273222e-01 -9.82640266e-01
1.56176329e-01 -5.90238310e-02 1.87988475e-01 -1.59698203e-01
-1.93430573e-01 1.45031011e+00 1.67412430e-01 2.55534798e-01
-1.57436267e-01 -2.94300854e-01 5.93245149e-01 5.97623169e-01
1.11247964e-01 3.93625200e-02 -3.36153895e-01 9.30220857e-02
-6.06851935e-01 3.72977406e-01 1.37901556e+00 5.79352200e-01
4.03619409e-01 -3.57029676e-01 2.77271301e-01 7.91754365e-01
4.85528469e-01 5.33886790e-01 1.11927533e+00 -7.45039761e-01
2.82894224e-01 1.01081479e+00 2.89214790e-01 -1.42021775e+00
1.75766274e-01 5.33039309e-02 -2.61790395e-01 3.94693434e-01
4.87629145e-01 3.34665090e-01 -7.67872572e-01 7.35712409e-01
2.81964600e-01 -1.50897413e-01 -4.13084388e-01 8.12556088e-01
7.27550685e-01 4.24713492e-01 6.80971742e-02 -9.54250630e-04
1.64063990e+00 -1.06237483e+00 -5.26946008e-01 1.02322094e-01
2.60846734e-01 -1.06988478e+00 8.61716509e-01 7.78526247e-01
-6.04902864e-01 9.42100510e-02 -8.85692716e-01 2.51826257e-01
-1.11356008e+00 -6.21910393e-01 6.35384321e-01 1.19902825e+00
-6.77865028e-01 5.36563694e-01 -3.38074774e-01 -8.91732097e-01
3.10022891e-01 3.91368568e-01 -1.87446877e-01 1.11497171e-01
-1.06208432e+00 9.61116672e-01 3.99588406e-01 -7.15844810e-01
-4.90481853e-01 4.08336937e-01 5.87035641e-02 2.70592682e-02
3.74544442e-01 -5.99374399e-02 1.10458803e+00 -1.04977655e+00
-9.82534170e-01 1.06110501e+00 1.20382167e-01 -7.00119197e-01
5.71679711e-01 -2.29565389e-02 -8.00184608e-01 2.62167037e-01
9.38540474e-02 -4.57808018e-01 1.00934434e+00 -1.12122428e+00
-9.07002628e-01 -8.30596089e-01 -4.79170561e-01 -2.30966493e-01
-4.74448532e-01 5.28161705e-01 -1.50397509e-01 -4.46869165e-01
4.47705239e-01 -6.56395137e-01 1.02575779e-01 -4.64450002e-01
-3.26056600e-01 -5.59145153e-01 1.09064138e+00 -1.10334396e+00
1.55298102e+00 -1.52463317e+00 -6.99419320e-01 8.56473863e-01
2.69966513e-01 7.78472364e-01 2.30427876e-01 7.47761130e-01
2.47753188e-01 7.12732971e-01 3.33873898e-01 8.61284852e-01
-6.47130609e-02 -8.08417797e-02 -3.88007641e-01 4.61087823e-01
-6.16552055e-01 3.38736087e-01 -8.08300316e-01 -8.26871932e-01
8.13301876e-02 -2.10507974e-01 -1.27986297e-01 2.62973532e-02
-8.17755982e-02 -4.43989426e-01 -9.70401227e-01 9.84366775e-01
9.95540321e-01 -1.60650000e-01 2.46440634e-01 -1.42059803e-01
6.98199645e-02 3.77038479e-01 -1.14353633e+00 4.54555929e-01
9.24265310e-02 4.70197022e-01 -3.94216040e-03 -1.06464159e+00
8.80428731e-01 7.13838786e-02 -1.27754286e-02 -6.59695148e-01
3.46980214e-01 7.28970587e-01 -4.88098599e-02 -9.57554400e-01
1.37769133e-02 2.46383533e-01 3.04023743e-01 5.79861522e-01
-1.40256107e-01 4.27428901e-01 5.41727066e-01 3.28626424e-01
1.68607450e+00 -2.71304429e-01 5.43199360e-01 -3.10631603e-01
1.14279675e+00 5.40004909e-01 1.30012274e-01 1.07401311e+00
-7.25178361e-01 7.12987632e-02 4.33290482e-01 -7.52702177e-01
-1.09174693e+00 -1.15509546e+00 -2.25307625e-02 1.03027904e+00
2.91171998e-01 -4.16525751e-01 -6.85037374e-01 -1.42001462e+00
3.17648977e-01 6.00240052e-01 -1.68727353e-01 -5.21657281e-02
-6.14129543e-01 -8.63939941e-01 4.37385708e-01 -4.15906549e-01
1.08257306e+00 -8.75254452e-01 -6.93031549e-02 1.84099242e-01
-2.05970973e-01 -5.08167088e-01 -1.19611792e-01 -1.14529252e-01
-9.79304552e-01 -1.69076037e+00 -4.91418280e-02 -9.63959455e-01
4.55568731e-01 7.70019889e-01 1.11614370e+00 4.03326333e-01
-5.73037863e-01 6.56460896e-02 -7.00680614e-01 -2.61242151e-01
-7.26708472e-01 6.88665658e-02 -2.35448167e-01 -4.11160171e-01
1.26416445e+00 -4.17674780e-01 -5.93189478e-01 6.35074794e-01
-8.15535486e-01 -9.19815898e-01 6.59782112e-01 6.55002892e-01
-2.53554314e-01 5.74024379e-01 3.92107069e-01 -1.16251528e+00
1.09571445e+00 -8.91462147e-01 -8.36995125e-01 3.35968405e-01
-1.06337035e+00 1.20318569e-01 6.14720941e-01 -2.32830927e-01
-9.81767058e-01 -1.80950284e-01 -4.27031517e-02 3.70316744e-01
-6.66892231e-02 8.65993500e-02 2.70444644e-03 -5.88639617e-01
9.79514241e-01 3.42073917e-01 2.72874117e-01 -1.06794095e+00
-6.83752745e-02 1.41499674e+00 -1.38895437e-01 -1.95857093e-01
1.05334866e+00 4.24999088e-01 -2.14209601e-01 -6.60816252e-01
-6.76437557e-01 -1.03507650e+00 -4.73647192e-02 -6.24141879e-02
4.01092678e-01 -2.07220286e-01 -9.56115007e-01 4.23422575e-01
-9.46910322e-01 7.91368842e-01 7.55437255e-01 5.76316603e-02
2.77508106e-02 1.01056302e+00 -5.89322031e-01 -8.82027388e-01
-6.49837017e-01 -5.74051857e-01 2.84510702e-01 2.26971865e-01
-4.95637804e-02 -7.56926775e-01 2.70672709e-01 7.85555601e-01
4.89554375e-01 -2.12565005e-01 1.08670819e+00 -1.58748591e+00
-6.00076318e-01 -8.25899184e-01 -4.56666112e-01 4.59532768e-01
2.05597207e-02 2.87603438e-02 -4.25407469e-01 -1.01828553e-01
1.59990773e-01 7.35226870e-02 7.23917603e-01 -2.52175361e-01
9.67646122e-01 -8.21017146e-01 -6.53846622e-01 6.29870743e-02
1.92913163e+00 3.83165330e-01 6.85638309e-01 1.26515949e+00
-1.53535381e-02 5.20030677e-01 5.06417155e-01 1.94743857e-01
-2.87354380e-01 4.97173309e-01 5.08861184e-01 8.51811051e-01
6.00565821e-02 -3.22982579e-01 3.51016045e-01 6.45127058e-01
6.28874525e-02 -2.70751029e-01 -9.49689388e-01 1.76306173e-01
-1.86150753e+00 -1.34587419e+00 -7.09398329e-01 2.38219881e+00
5.13667345e-01 3.15496683e-01 4.35646683e-01 1.20854042e-01
1.25096130e+00 -2.42273539e-01 -2.16660932e-01 -5.20406306e-01
-9.84386057e-02 2.42317557e-01 1.02643836e+00 3.93355638e-01
-1.28345656e+00 1.05406272e+00 6.65806532e+00 1.21433294e+00
-5.89242756e-01 2.04358816e-01 3.32152337e-01 5.04277825e-01
3.64828221e-02 2.02334926e-01 -9.21141267e-01 9.34234321e-01
9.81181562e-01 -5.37413955e-01 5.25998414e-01 1.31139755e+00
1.29215956e-01 -3.35001200e-01 -6.89151138e-02 8.03501010e-01
3.14661682e-01 -1.00298798e+00 4.55035090e-01 1.04336701e-01
4.74546432e-01 4.82257232e-02 -4.13912714e-01 1.49761453e-01
7.74654031e-01 -7.24521041e-01 1.14516467e-02 8.89971852e-02
-3.75757843e-01 -5.55334687e-01 1.01583898e+00 3.80041063e-01
-3.75596434e-01 -3.78403246e-01 -5.46353817e-01 1.32514060e-01
-2.74716258e-01 3.85670573e-01 -6.21346354e-01 1.58538863e-01
1.16640341e+00 3.58042447e-03 -1.19251573e+00 1.73539305e+00
4.31253091e-02 1.01502323e+00 -4.86120015e-01 -9.48679566e-01
2.26929069e-01 -6.80804193e-01 7.54290223e-01 1.00039589e+00
1.79294631e-01 -2.87899166e-01 -2.98326224e-01 3.72324347e-01
2.22312883e-01 5.44219434e-01 -6.99175298e-01 -1.05202220e-01
5.67855000e-01 1.43228996e+00 -9.59041059e-01 -4.67089266e-01
-4.59811032e-01 9.65369761e-01 -2.61041755e-03 2.14004163e-02
-6.05185509e-01 -7.29300439e-01 2.30110273e-01 5.70075810e-01
5.70202470e-02 -9.94663388e-02 -1.16195641e-01 -1.06036663e+00
-2.03925055e-02 -1.31180966e+00 9.66618299e-01 -4.00045753e-01
-1.72050822e+00 4.80272382e-01 -3.08011383e-01 -1.24258351e+00
1.23927347e-01 -9.76883769e-01 -6.18063331e-01 5.94452083e-01
-1.08854306e+00 -9.60491300e-01 -6.12692475e-01 6.06043756e-01
4.93853658e-01 -7.54153192e-01 5.20417571e-01 2.05701709e-01
-2.17394963e-01 7.14061037e-02 7.98807919e-01 2.24559605e-01
7.30830967e-01 -1.14356601e+00 -1.89388152e-02 6.28244877e-01
-5.31355999e-02 8.02588820e-01 9.45275664e-01 -9.09726202e-01
-1.35603142e+00 -4.13856775e-01 8.37968767e-01 -6.18586779e-01
1.02591252e+00 -6.07095025e-02 -8.57087433e-01 3.68812412e-01
2.90479243e-01 -6.97746336e-01 6.35560811e-01 -2.29515284e-01
-5.71039796e-01 -1.16515137e-01 -1.51806509e+00 5.58401704e-01
7.80576706e-01 1.18792444e-01 -8.18139255e-01 7.53571451e-01
-2.20531714e-03 5.56860328e-01 -2.18871012e-01 -1.83427528e-01
5.59789896e-01 -1.37673903e+00 1.05089986e+00 -9.16805387e-01
4.15036492e-02 -2.03238130e-01 -1.47284707e-02 -8.41513336e-01
-4.32640910e-01 -4.25374091e-01 -1.06906090e-02 1.23365283e+00
3.44361216e-02 -1.01857638e+00 1.34243667e+00 5.23984551e-01
7.78830767e-01 1.12121515e-01 -8.57265711e-01 -9.53655362e-01
1.23694368e-01 2.29525894e-01 3.52584710e-03 7.11376965e-01
-1.49723375e-02 3.72368991e-01 -9.94095281e-02 -2.16812596e-01
1.15322626e+00 -3.95271834e-03 5.94358921e-01 -1.57997012e+00
1.41737342e-01 -7.18926489e-01 -6.52623296e-01 -3.80215257e-01
-1.21895030e-01 -8.38905394e-01 -7.07450807e-01 -1.28195739e+00
5.60808837e-01 -7.36217648e-02 -5.01449816e-02 -2.13676561e-02
-4.04108167e-02 1.07819028e-01 -2.61097759e-01 7.21604526e-01
-6.58473551e-01 -1.62574619e-01 5.39000750e-01 -1.53122591e-02
2.42010921e-01 3.68963033e-01 -4.78193998e-01 8.06243539e-01
1.14860368e+00 -8.70771885e-01 1.36590630e-01 4.21551615e-01
3.76314521e-01 -4.62750673e-01 3.98030758e-01 -8.43019724e-01
2.28658885e-01 -3.17994893e-01 1.71347186e-01 -6.85171902e-01
-2.47280121e-01 -1.01850462e+00 -8.13098252e-02 9.90212440e-01
-2.53254510e-02 1.18754439e-01 -4.52775747e-01 8.99152577e-01
-3.93421113e-01 -1.12064564e+00 9.11907732e-01 -7.75991619e-01
-8.82128119e-01 -6.83586374e-02 -6.78245962e-01 3.60545814e-02
1.05991721e+00 -1.61993548e-01 -8.65278304e-01 -4.79269534e-01
-3.16597372e-01 6.80058822e-02 4.64966476e-01 5.99962831e-01
2.88714737e-01 -1.16296601e+00 -5.59825480e-01 -3.28882635e-02
-4.68552895e-02 -1.20510888e+00 -4.12194163e-01 3.45223844e-01
-1.33378565e+00 5.04259884e-01 -4.42593485e-01 1.73628777e-01
-1.77037644e+00 7.25854695e-01 1.57240540e-01 -2.02781573e-01
-9.09378752e-02 3.39398295e-01 -5.42261779e-01 -3.19814801e-01
1.51002303e-01 9.69467759e-01 -6.32848382e-01 -1.10301197e-01
5.64401150e-01 1.00738800e+00 1.44883007e-01 -4.72612321e-01
-4.84219253e-01 2.73111254e-01 -4.62153763e-01 -4.73006964e-02
9.86939490e-01 -1.42813250e-01 -8.24714601e-01 -1.18084803e-01
1.38955557e+00 2.26012334e-01 4.62235659e-02 -3.50582367e-03
8.18548858e-01 -9.75157440e-01 -1.10323213e-01 -1.12619650e+00
-6.22655332e-01 5.29679418e-01 9.82208073e-01 1.00659549e+00
5.52292287e-01 -2.24358454e-01 9.01042640e-01 5.61108232e-01
4.84631509e-01 -1.56284916e+00 1.46668747e-01 3.15830052e-01
5.05914450e-01 -1.61318207e+00 1.78886112e-02 -8.70881140e-01
-1.57436132e-01 1.35247695e+00 7.09661841e-01 -5.35692692e-01
8.78446221e-01 -2.42676765e-01 6.03486262e-02 -3.09943140e-01
-5.26033998e-01 -2.72770435e-01 -3.93288359e-02 5.63641608e-01
4.50695544e-01 -1.72124609e-01 -1.61682117e+00 5.62899411e-01
5.98117383e-03 -1.85529172e-01 5.25655568e-01 1.11088443e+00
-1.27186584e+00 -1.51083636e+00 -6.36616588e-01 1.03844726e+00
-9.49613094e-01 -2.58989841e-01 -8.77791405e-01 7.79022932e-01
-4.40060169e-01 1.12558579e+00 -3.96479517e-01 -4.67997998e-01
5.48107252e-02 4.00675476e-01 5.08452114e-03 -9.50564072e-02
-7.18176007e-01 -2.88063407e-01 4.48718935e-01 -4.38103437e-01
1.02036402e-01 -5.75611293e-01 -7.07591295e-01 -6.57804847e-01
-5.96068740e-01 8.18833709e-01 9.18440104e-01 3.33611310e-01
2.69700915e-01 -3.88645053e-01 7.37985969e-01 1.92360252e-01
-1.23988760e+00 -9.28193450e-01 -7.73273587e-01 1.07891846e+00
-3.93010050e-01 -5.04779994e-01 -1.03823733e+00 -1.99860081e-01] | [7.822587013244629, 10.010924339294434] |
1a3c89ae-dff7-4076-a6a3-0ca44bab7f54 | distance-sensitive-offline-reinforcement | 2205.11027 | null | https://arxiv.org/abs/2205.11027v3 | https://arxiv.org/pdf/2205.11027v3.pdf | When Data Geometry Meets Deep Function: Generalizing Offline Reinforcement Learning | In offline reinforcement learning (RL), one detrimental issue to policy learning is the error accumulation of deep Q function in out-of-distribution (OOD) areas. Unfortunately, existing offline RL methods are often over-conservative, inevitably hurting generalization performance outside data distribution. In our study, one interesting observation is that deep Q functions approximate well inside the convex hull of training data. Inspired by this, we propose a new method, DOGE (Distance-sensitive Offline RL with better GEneralization). DOGE marries dataset geometry with deep function approximators in offline RL, and enables exploitation in generalizable OOD areas rather than strictly constraining policy within data distribution. Specifically, DOGE trains a state-conditioned distance function that can be readily plugged into standard actor-critic methods as a policy constraint. Simple yet elegant, our algorithm enjoys better generalization compared to state-of-the-art methods on D4RL benchmarks. Theoretical analysis demonstrates the superiority of our approach to existing methods that are solely based on data distribution or support constraints. | ['Ya-Qin Zhang', 'Jingjing Liu', 'Xiangyu Zhu', 'Haoran Xu', 'Xianyuan Zhan', 'Jianxiong Li'] | 2022-05-23 | null | null | null | null | ['d4rl'] | ['robots'] | [-4.22966301e-01 3.50608766e-01 -5.50858140e-01 -1.05600610e-01
-8.00352335e-01 -8.44005883e-01 2.24551216e-01 1.82817429e-01
-6.64799392e-01 1.04551721e+00 -4.18503731e-02 -4.16729838e-01
-2.59820282e-01 -7.38708913e-01 -1.05204821e+00 -7.83141434e-01
-2.48950928e-01 5.61393142e-01 9.88774747e-02 -4.07685399e-01
2.05745369e-01 7.22033679e-01 -1.15493762e+00 -2.78634042e-01
1.09818220e+00 1.30584025e+00 2.55525149e-02 3.16325277e-01
3.34344385e-03 7.42744386e-01 -8.48761559e-01 -2.60866553e-01
7.96348870e-01 -3.08581144e-01 -4.39102024e-01 -1.48120344e-01
2.82333761e-01 -6.48458958e-01 -6.00302160e-01 9.39312994e-01
7.12754011e-01 5.09696543e-01 5.93284726e-01 -1.12094462e+00
-9.67990220e-01 5.08228719e-01 -5.55915117e-01 8.94475281e-02
1.06951334e-02 4.35911566e-01 9.75394487e-01 -6.95221663e-01
4.55363661e-01 1.23620439e+00 5.79058111e-01 5.65293550e-01
-1.37900829e+00 -4.87523556e-01 5.76619446e-01 -1.32767215e-01
-1.21200287e+00 -1.65537089e-01 6.75818086e-01 -1.14570267e-01
8.77782404e-01 -1.51112508e-02 7.30120540e-01 1.01841128e+00
2.22453609e-01 1.27381277e+00 1.12001669e+00 -1.31799996e-01
7.35168457e-01 9.98774990e-02 -6.21753216e-01 7.76056409e-01
6.51758760e-02 6.53602123e-01 -3.67295474e-01 -1.08682171e-01
1.12609088e+00 -2.12651879e-01 -1.09481022e-01 -9.54754651e-01
-7.09496021e-01 1.12149858e+00 7.18402267e-01 -2.58222550e-01
-2.98085570e-01 5.18414497e-01 4.38706726e-01 5.42680979e-01
4.87383783e-01 9.02186513e-01 -6.05653584e-01 -5.07605731e-01
-6.00119472e-01 6.58712447e-01 6.46283805e-01 1.02587974e+00
6.79833591e-01 1.51446089e-01 -3.45261186e-01 7.41975486e-01
7.51105621e-02 4.47915792e-01 4.22848046e-01 -1.15307271e+00
6.34285450e-01 3.85386407e-01 4.34589088e-01 -6.46713376e-01
-4.03855622e-01 -8.39885831e-01 -4.57139999e-01 6.14349067e-01
7.77458608e-01 -4.63219792e-01 -7.25354612e-01 1.82036817e+00
4.44861293e-01 -1.34121731e-01 1.35084346e-01 1.10083663e+00
2.56109592e-02 3.36811692e-01 -1.18146084e-01 -1.85259700e-01
5.68109035e-01 -1.04204094e+00 -4.83596951e-01 -1.06249936e-02
8.07172537e-01 -7.99630210e-02 1.32398129e+00 6.89843237e-01
-1.12273335e+00 -4.32438552e-01 -9.28953350e-01 -2.83883559e-03
-3.01785707e-01 5.56169227e-02 7.83728480e-01 5.41917741e-01
-1.06258404e+00 9.35444951e-01 -7.47001588e-01 8.80127698e-02
8.66129994e-01 5.48094571e-01 5.26256189e-02 2.48890966e-01
-1.10057056e+00 1.05012155e+00 3.64695758e-01 -6.79121539e-02
-1.24260318e+00 -9.44747269e-01 -6.47864878e-01 -1.40374154e-01
9.15563047e-01 -3.53344470e-01 1.47843957e+00 -1.11133742e+00
-2.27966022e+00 3.58123928e-01 4.80439335e-01 -8.71601760e-01
1.03103828e+00 -6.75542057e-01 -8.12572613e-02 9.00276378e-02
-2.21226200e-01 8.60484719e-01 1.11773527e+00 -1.15895438e+00
-6.54373050e-01 -3.32102567e-01 3.21259350e-01 4.85831857e-01
-3.53963703e-01 -7.02257693e-01 -2.58761823e-01 -6.83721423e-01
-4.72317487e-01 -6.85308278e-01 -4.65553910e-01 4.20041353e-01
-2.37007469e-01 -5.23180902e-01 7.77684629e-01 -2.10100174e-01
1.22432554e+00 -2.03546786e+00 2.33067065e-01 3.24281186e-01
2.24769965e-01 1.75290421e-01 -2.53172845e-01 3.71648818e-01
3.17530096e-01 -1.67361144e-02 -1.57346472e-01 -5.75826988e-02
4.73362207e-01 5.97590387e-01 -7.11960852e-01 8.49322200e-01
1.93407312e-01 1.01761961e+00 -1.22763062e+00 -1.69829965e-01
1.05636902e-01 2.37455405e-02 -7.60284364e-01 1.58812344e-01
-6.57944322e-01 5.53664982e-01 -7.37595737e-01 5.59791923e-01
5.57120264e-01 2.45689508e-02 -1.76196173e-02 3.15850019e-01
-2.85540849e-01 9.80861019e-03 -8.91821742e-01 2.05991459e+00
-6.10179484e-01 3.99124354e-01 -8.32057372e-02 -1.27672708e+00
1.23466778e+00 -1.85379237e-01 6.71360075e-01 -9.82016861e-01
1.52640700e-01 3.03713351e-01 -1.17960945e-01 -3.24535012e-01
4.41352129e-01 1.35783523e-01 3.86752561e-02 1.75590366e-01
-5.51810488e-03 -4.35062081e-01 -5.36716171e-02 -2.44776607e-01
9.81623113e-01 7.50837088e-01 1.47489980e-01 -4.70131129e-01
1.90813392e-01 -6.52150810e-02 4.78257895e-01 1.00934374e+00
-5.00218391e-01 1.74693778e-01 8.90122056e-01 -5.02843201e-01
-1.17173219e+00 -1.15883422e+00 -2.13544220e-01 1.16365731e+00
3.37263107e-01 -8.32012296e-02 -4.70389396e-01 -1.32191181e+00
8.35432947e-01 8.05040479e-01 -8.21685672e-01 -3.53941441e-01
-7.37956643e-01 -1.92123234e-01 7.02292562e-01 7.98920095e-01
3.63298595e-01 -7.77313828e-01 -7.95581162e-01 4.01999205e-01
7.17633784e-01 -7.52787054e-01 -5.05762219e-01 5.36469936e-01
-1.05114186e+00 -7.89155662e-01 -1.03435326e+00 -5.31690121e-01
5.88128865e-01 -2.45698899e-01 1.10327053e+00 -2.23360032e-01
-3.20463553e-02 5.04638255e-01 -2.45401129e-01 -5.82645655e-01
-1.52522951e-01 2.40184903e-01 2.09008098e-01 -3.82377326e-01
1.30075350e-01 -4.62527007e-01 -8.73293817e-01 4.61534381e-01
-5.64709902e-01 -5.45153081e-01 5.48070669e-01 1.05625176e+00
6.92954421e-01 -7.01655373e-02 9.46268916e-01 -6.59188569e-01
7.08681762e-01 -5.47314465e-01 -1.09467256e+00 9.21808258e-02
-8.09563696e-01 4.72624213e-01 1.29522920e+00 -6.31871700e-01
-6.82692349e-01 4.16281633e-03 -9.15528089e-02 -8.98649693e-01
2.09529430e-01 8.77961069e-02 1.06344946e-01 -2.37853676e-01
8.59283864e-01 1.82611212e-01 3.88848066e-01 -3.91553700e-01
6.15127265e-01 2.35518098e-01 3.71515572e-01 -1.19655859e+00
6.82721436e-01 5.61020494e-01 7.67536163e-02 -4.83393610e-01
-9.83489215e-01 -1.25018805e-01 -2.56919026e-01 -1.48636073e-01
3.49173218e-01 -7.44295418e-01 -1.02241814e+00 7.39017725e-02
-4.72911119e-01 -1.06100476e+00 -1.03646243e+00 3.58702630e-01
-1.07497752e+00 7.84644708e-02 -1.79338664e-01 -1.04758966e+00
4.98964004e-02 -9.27171230e-01 1.09691083e+00 2.63770312e-01
2.46222556e-01 -1.13749325e+00 1.90361351e-01 -3.36676985e-01
4.23253298e-01 3.67903590e-01 7.03688860e-01 -4.71032858e-01
-2.23644972e-01 1.74583644e-01 2.75987945e-03 4.31444645e-01
-3.03252190e-02 -3.13226581e-01 -7.61908770e-01 -6.84656560e-01
-1.23307064e-01 -9.34898615e-01 7.98545539e-01 4.97198492e-01
1.94252133e+00 -2.99781829e-01 -1.10780463e-01 8.85278761e-01
1.46387291e+00 2.03475609e-01 5.27884781e-01 6.31191969e-01
3.43989402e-01 3.26646924e-01 1.04751825e+00 9.19751585e-01
1.96287766e-01 6.02138281e-01 7.77702212e-01 -7.35956132e-02
2.87025481e-01 -6.52869940e-01 4.71675128e-01 1.06672809e-01
1.52261555e-01 -2.29596913e-01 -4.54926163e-01 4.13843036e-01
-2.07105374e+00 -5.22841275e-01 5.47765791e-01 2.10609031e+00
9.90197182e-01 2.85567611e-01 5.56264281e-01 -2.98207998e-01
2.70629048e-01 6.92887530e-02 -1.39094484e+00 -6.46957636e-01
-1.69532835e-01 4.44847524e-01 1.01717699e+00 3.67327213e-01
-9.94595349e-01 1.01120162e+00 6.28677750e+00 1.23605919e+00
-1.16340196e+00 -1.20912515e-01 6.81161106e-01 -3.14458311e-01
-3.95465434e-01 -2.13359341e-01 -8.28271627e-01 3.06849718e-01
5.06309807e-01 1.06340885e-01 7.96309114e-01 1.25737512e+00
2.71057516e-01 -1.18795540e-02 -1.18656373e+00 8.43266785e-01
-3.68213505e-01 -1.34242988e+00 -3.56502473e-01 2.40660250e-01
8.82106006e-01 -5.38689196e-02 4.62809294e-01 8.98966372e-01
7.11003721e-01 -1.20046556e+00 9.75519419e-01 4.71263707e-01
9.93135989e-01 -1.18796766e+00 2.41244823e-01 4.72931325e-01
-6.92255139e-01 -6.14294648e-01 -7.53335476e-01 2.61425432e-02
-1.84049144e-01 2.15194434e-01 -6.62092984e-01 3.34411144e-01
4.79837030e-01 7.59060264e-01 -2.18643591e-01 9.05200899e-01
-3.78108323e-01 3.96046102e-01 -5.28550446e-01 -3.79753053e-01
8.08954775e-01 -2.43609771e-01 5.56856334e-01 7.30920196e-01
1.17349222e-01 -1.08815372e-01 4.79056656e-01 9.69345808e-01
-2.01523185e-01 1.17893100e-01 -6.87001050e-01 -1.12541668e-01
3.58104497e-01 7.98358083e-01 -2.99246728e-01 -5.55212647e-02
-1.93922281e-01 7.55581617e-01 9.30003107e-01 4.69945788e-01
-9.96128023e-01 -3.35814446e-01 6.75136626e-01 -6.47780076e-02
7.23076642e-01 -3.91201735e-01 -2.45009944e-01 -9.84544814e-01
5.29189175e-03 -8.94259572e-01 3.66459906e-01 -1.43962488e-01
-1.50304091e+00 1.33442521e-01 -1.39100164e-01 -1.19959354e+00
-1.98313907e-01 -9.01610851e-01 -3.09056997e-01 4.49935347e-01
-1.71380782e+00 -6.41838729e-01 1.71589911e-01 6.03804231e-01
5.29079199e-01 -4.23096329e-01 6.04615569e-01 8.08751732e-02
-4.22162443e-01 1.11068475e+00 7.14690030e-01 -1.87367514e-01
6.45483077e-01 -1.66824317e+00 1.43233985e-01 3.03291976e-01
4.50528041e-02 2.94914752e-01 5.47720790e-01 -4.47665185e-01
-1.63803220e+00 -1.03482771e+00 -2.49463424e-01 -3.14899951e-01
7.96524644e-01 -4.55863923e-01 -5.76737642e-01 4.37428057e-01
-1.39086947e-01 4.73707318e-01 1.93300501e-01 -4.78448346e-02
-1.34611264e-01 -3.83354932e-01 -1.17807508e+00 6.99523270e-01
1.10685134e+00 -1.31503135e-01 -2.37597808e-01 3.34596694e-01
7.63233423e-01 -8.94880831e-01 -9.87291038e-01 1.69558331e-01
4.57010061e-01 -7.90565372e-01 9.52354074e-01 -1.00464272e+00
2.07099736e-01 -6.99406862e-02 -7.03708529e-02 -1.59661114e+00
3.98286730e-02 -1.22365439e+00 -7.50443101e-01 5.77811360e-01
2.66299546e-01 -7.46775806e-01 9.95080352e-01 2.08716452e-01
-3.49266618e-01 -1.51919973e+00 -9.92175221e-01 -1.23217237e+00
7.96055079e-01 -2.48954237e-01 6.94275320e-01 6.70133293e-01
-5.00109978e-02 -2.67011017e-01 -2.68469721e-01 -2.86054378e-03
5.25885880e-01 1.18913755e-01 9.49897647e-01 -6.95784390e-01
-7.45935321e-01 -6.76593065e-01 -2.66359281e-02 -1.91783655e+00
2.77392328e-01 -7.53390312e-01 5.78123480e-02 -1.12925220e+00
-6.06714070e-01 -1.02534783e+00 -3.64379883e-01 4.91532654e-01
1.32566795e-01 -1.89580232e-01 1.44222364e-01 5.35422005e-02
-6.23814583e-01 1.25781846e+00 1.94799376e+00 5.67601435e-02
-5.47559917e-01 6.60996959e-02 -6.33908272e-01 5.53924978e-01
8.63982737e-01 -3.41411769e-01 -6.79165542e-01 -4.19544071e-01
3.13818663e-01 9.99164432e-02 2.34942839e-01 -7.06046999e-01
-1.20424270e-03 -4.17032301e-01 5.88211894e-01 -4.32881981e-01
1.23927802e-01 -7.21817434e-01 -7.98098803e-01 5.06490290e-01
-5.38848221e-01 1.13237975e-03 3.62651259e-01 7.26872146e-01
1.36150956e-01 -2.93029305e-02 7.95178533e-01 5.14173470e-02
-5.80471694e-01 5.82604229e-01 -8.13943371e-02 6.42681181e-01
1.08840251e+00 -1.25694647e-01 -2.70029545e-01 -3.07979524e-01
-5.39683223e-01 5.70458591e-01 4.01401728e-01 2.69302905e-01
5.13444424e-01 -1.26888525e+00 -4.87656295e-01 1.36591583e-01
-1.62740305e-01 3.67849827e-01 -1.91828460e-01 7.99006164e-01
-5.66284418e-01 1.65549010e-01 -1.49215549e-01 -5.30110896e-01
-5.66855557e-02 6.91640437e-01 5.84439754e-01 -3.63772810e-01
-9.20622647e-01 9.18377340e-01 1.11833252e-01 -4.64552611e-01
6.13136709e-01 -6.51366591e-01 9.85452309e-02 -2.25065514e-01
1.34031281e-01 3.69897783e-01 -2.34489948e-01 2.08415374e-01
-9.04001668e-02 2.49416888e-01 -9.71042812e-02 -8.70947018e-02
1.39481163e+00 2.17984873e-03 5.95144689e-01 3.30553532e-01
1.09554636e+00 7.53113776e-02 -2.28225732e+00 -3.33897062e-02
-7.19150379e-02 -5.99161565e-01 1.41991330e-02 -9.16676939e-01
-1.05191934e+00 6.66351318e-01 5.67861676e-01 2.33610153e-01
9.45312202e-01 -3.21203172e-01 7.69612312e-01 6.38662755e-01
4.32139844e-01 -1.74431634e+00 4.01573747e-01 5.82647979e-01
1.07237315e+00 -1.24703574e+00 1.10471368e-01 2.88287818e-01
-8.10827136e-01 1.28427184e+00 8.54463875e-01 -9.59005296e-01
5.55270374e-01 2.23063529e-01 -3.38714719e-02 -9.18892119e-03
-7.65670180e-01 -1.66755527e-01 9.29803252e-02 6.52337492e-01
-8.63225311e-02 -1.35816261e-01 -1.90031037e-01 3.21059078e-01
-2.84820944e-01 -1.96218237e-01 7.96065927e-02 9.54356194e-01
-4.36793536e-01 -1.08050942e+00 -7.70646930e-02 2.39222154e-01
-3.21315408e-01 3.00701261e-01 -1.20597854e-01 1.32661295e+00
-1.56686418e-02 4.23212260e-01 1.76632613e-01 -5.08443778e-03
3.09321344e-01 -5.10437965e-01 8.17858219e-01 -3.57990623e-01
-5.96459389e-01 4.62255068e-02 -1.32401824e-01 -9.61896956e-01
1.75131977e-01 -5.17979622e-01 -1.46943951e+00 -6.16717599e-02
-3.08565140e-01 1.15340628e-01 5.99556863e-01 9.29463506e-01
3.31918538e-01 4.24627841e-01 8.83113027e-01 -5.53082347e-01
-1.56898177e+00 -5.48145354e-01 -7.55904555e-01 1.46092623e-01
5.90192735e-01 -1.02891600e+00 -2.52224356e-01 -7.35299170e-01] | [4.125679016113281, 2.301802396774292] |
839a9a97-36c6-4912-afc4-69b720fd0978 | optimizing-slam-evaluation-footprint-through | 2209.06316 | null | https://arxiv.org/abs/2209.06316v2 | https://arxiv.org/pdf/2209.06316v2.pdf | Optimizing SLAM Evaluation Footprint Through Dynamic Range Coverage Analysis of Datasets | Simultaneous Localization and Mapping (SLAM) is considered an ever-evolving problem due to its usage in many applications. Evaluation of SLAM is done typically using publicly available datasets which are increasing in number and the level of difficulty. Each dataset provides a certain level of dynamic range coverage that is a key aspect of measuring the robustness and resilience of SLAM. In this paper, we provide a systematic analysis of the dynamic range coverage of datasets based on a number of characterization metrics, and our analysis shows a huge level of redundancy within and between datasets. Subsequently, we propose a dynamic programming (DP) algorithm for eliminating the redundancy in the evaluation process of SLAM by selecting a subset of sequences that matches a single or multiple dynamic range coverage objectives. It is shown that, with the help of dataset characterization and DP selection algorithm, a reduction in the evaluation effort can be achieved while maintaining the same level of coverage. We also study how the evaluation process of a real-world SLAM system can be optimized utilizing the method proposed. | ['Hong Zhang', 'Islam Ali'] | 2022-09-13 | null | null | null | null | ['simultaneous-localization-and-mapping'] | ['computer-vision'] | [ 3.87585193e-01 -4.43837970e-01 4.10298593e-02 -6.37705982e-01
-6.87561393e-01 -7.03356028e-01 4.22081411e-01 6.87316597e-01
-5.84405601e-01 8.61505210e-01 -7.77557939e-02 -2.28158571e-02
-7.92193711e-01 -9.15598452e-01 -6.10145152e-01 -4.68752831e-01
-4.03168082e-01 7.08229184e-01 3.98358256e-01 -3.15180302e-01
6.34889483e-01 8.90429437e-01 -1.92130589e+00 -4.98213589e-01
9.07069862e-01 8.90189469e-01 4.26843762e-01 4.55845863e-01
-5.40655330e-02 1.67480469e-01 -8.25047612e-01 2.00199738e-01
4.82620239e-01 -2.04928041e-01 -5.16972065e-01 -1.27962008e-01
2.38145843e-01 1.79316625e-01 2.02353582e-01 1.12705112e+00
6.13282442e-01 1.45756200e-01 1.60703614e-01 -1.62787044e+00
5.59118986e-01 1.44446805e-01 -2.51664191e-01 3.95136289e-02
6.59351707e-01 -1.00186005e-01 6.91336930e-01 -4.25484926e-01
7.38714516e-01 9.00455117e-01 7.19546974e-01 -2.87485391e-01
-1.32072794e+00 -5.38518906e-01 -1.98327661e-01 4.01710197e-02
-1.76652932e+00 -3.36565882e-01 5.32989979e-01 -4.84117687e-01
8.71890306e-01 4.18569803e-01 7.31849313e-01 2.23889887e-01
6.15813911e-01 -2.39970516e-02 9.91760850e-01 -3.87264609e-01
4.10084218e-01 2.08754279e-02 9.26494896e-02 4.56551015e-01
9.17879105e-01 1.60360768e-01 -7.23638296e-01 -4.22237754e-01
1.90234497e-01 -2.56153524e-01 -1.10404082e-01 -1.01619482e+00
-1.29504263e+00 7.32746959e-01 1.94818795e-01 2.63345420e-01
-3.07040274e-01 1.65739164e-01 4.63854313e-01 3.64279896e-01
6.73142588e-03 6.81935608e-01 -3.18494946e-01 -3.79373878e-01
-9.73866642e-01 2.87547171e-01 7.93261826e-01 9.01764631e-01
1.28335464e+00 -2.93708235e-01 4.71248865e-01 4.09133017e-01
4.05971818e-02 6.37142956e-01 1.61081761e-01 -7.90034115e-01
3.00639451e-01 9.70037341e-01 3.20841223e-01 -1.42473054e+00
-6.82787418e-01 -3.56972933e-01 -4.74655747e-01 2.96221942e-01
2.47458816e-02 4.61342037e-02 -4.16003585e-01 1.71746624e+00
5.09098589e-01 -1.83635220e-01 1.19285695e-01 5.87785959e-01
1.32876590e-01 3.49826992e-01 -3.19994062e-01 -3.05746764e-01
8.98787320e-01 -2.96954930e-01 -6.20386004e-01 -3.03135663e-01
6.60509646e-01 -8.19774270e-01 7.16742814e-01 2.87690878e-01
-4.20853853e-01 -3.55368346e-01 -1.50784421e+00 2.55610466e-01
-2.96606004e-01 -2.66381413e-01 6.17009223e-01 6.55142426e-01
-9.89510059e-01 4.47315931e-01 -7.87401974e-01 -7.91147292e-01
-3.80742759e-01 6.04529858e-01 -5.78428984e-01 -1.34278715e-01
-9.31887448e-01 1.33414745e+00 5.84117234e-01 -2.27781072e-01
-5.29459476e-01 -4.02203918e-01 -9.71811652e-01 -3.54187340e-01
3.94207239e-01 -3.94406438e-01 5.48942685e-01 -4.65854436e-01
-1.03252780e+00 7.17018664e-01 -1.11671910e-01 -6.07086182e-01
6.56093419e-01 -4.34196815e-02 -3.58758152e-01 -1.81701139e-01
3.14263225e-01 3.53417277e-01 3.33794147e-01 -1.20617294e+00
-7.89049506e-01 -4.15602237e-01 -4.83403765e-02 3.34148049e-01
2.58405507e-01 -4.78247292e-02 -2.28041455e-01 -1.41537450e-02
6.58850193e-01 -1.21938634e+00 -4.30555046e-01 -3.14200103e-01
-8.38847682e-02 3.03841472e-01 6.12923205e-01 -2.90491194e-01
1.21893811e+00 -2.01503754e+00 2.59192526e-01 7.17498779e-01
-2.95655817e-01 -2.46933594e-01 -2.64038052e-02 9.24241602e-01
4.42174286e-01 -6.16003908e-02 -4.56407368e-01 -2.80824095e-01
-1.55356452e-01 6.29168332e-01 -1.42680287e-01 8.75678003e-01
-3.71549904e-01 2.45471314e-01 -8.80492270e-01 -4.63836849e-01
4.46103156e-01 1.85646296e-01 -3.99355054e-01 7.66935423e-02
3.35535072e-02 5.75694799e-01 -1.79160118e-01 6.93592310e-01
7.47356832e-01 1.24898754e-01 4.86371130e-01 -2.66850274e-02
-5.22560000e-01 2.38149077e-01 -1.55470657e+00 1.90579093e+00
-5.94443083e-01 6.64719760e-01 -1.21190161e-01 -5.92245996e-01
1.32102120e+00 -1.62960142e-01 7.54138589e-01 -5.90443909e-01
1.46722689e-01 5.10806918e-01 1.46488249e-02 -2.11727053e-01
1.13259196e+00 2.23480448e-01 -6.32089019e-01 4.85866606e-01
-2.12166920e-01 -3.81439865e-01 3.50215524e-01 -1.18159436e-01
1.20403421e+00 -6.16049208e-02 7.99132943e-01 -5.86606443e-01
6.12628877e-01 4.78796631e-01 6.11181378e-01 8.26838791e-01
-2.68053234e-01 1.48530245e-01 2.96526521e-01 -4.18205529e-01
-1.06480145e+00 -6.45367026e-01 -3.50760341e-01 5.94026744e-01
7.97776103e-01 -3.25164795e-01 -2.58169800e-01 -1.25541046e-01
4.99944568e-01 6.14418149e-01 -4.37893718e-01 -1.44663602e-01
-4.88157719e-01 -7.37837553e-01 4.81024891e-01 -2.33612612e-01
3.52524996e-01 -6.68866158e-01 -1.38179934e+00 1.16150141e-01
-1.58280596e-01 -1.05242753e+00 2.04764217e-01 2.48107538e-01
-8.57620597e-01 -1.07802057e+00 2.10618794e-01 -2.34760270e-01
6.71934545e-01 4.92294997e-01 9.41558838e-01 1.45271719e-01
-3.21115285e-01 2.92329311e-01 -4.90762979e-01 -3.20351630e-01
-4.54255760e-01 1.56556711e-01 4.30408508e-01 -2.71407664e-01
1.38763741e-01 -6.82439923e-01 -2.36066878e-01 5.44055283e-01
-7.72506356e-01 -1.90107986e-01 5.51842690e-01 4.34957892e-01
9.34614122e-01 3.78161788e-01 2.62262762e-01 -6.03610814e-01
5.27808666e-01 -5.07376313e-01 -1.16599834e+00 1.89013124e-01
-8.78701389e-01 2.14842454e-01 1.14684984e-01 -1.12206221e-01
-4.39765245e-01 3.01731050e-01 9.27406698e-02 1.49359480e-01
1.57957137e-01 8.30310643e-01 -2.62337536e-01 -9.29922223e-01
5.61747491e-01 2.25599021e-01 1.53155804e-01 -2.35111713e-01
1.69546932e-01 3.97671163e-01 4.47420955e-01 -5.11093974e-01
8.25121582e-01 6.31084919e-01 5.19848466e-01 -7.59410203e-01
-3.76126349e-01 -6.48893952e-01 -6.19800687e-01 -3.58000338e-01
1.74243018e-01 -4.68501061e-01 -7.19013691e-01 1.62523553e-01
-8.28264534e-01 3.79386656e-02 -2.14930996e-01 4.59875345e-01
-6.44867361e-01 4.87907737e-01 3.34843189e-01 -8.96110058e-01
-5.97441122e-02 -1.37291241e+00 9.49622691e-01 -1.13465391e-01
-2.72582233e-01 -5.60037792e-01 5.09026587e-01 -1.58026330e-02
5.11009932e-01 8.38355243e-01 4.91162509e-01 -5.22983134e-01
-7.40201354e-01 -5.58228850e-01 -3.78827378e-02 -1.51417509e-01
1.02541678e-01 -2.31153473e-01 -3.76186043e-01 -6.90467119e-01
4.99998853e-02 2.00533733e-01 2.61756688e-01 7.62899816e-02
5.43090701e-01 2.34473608e-02 -3.66868407e-01 7.91271508e-01
2.08212590e+00 1.50731042e-01 5.22085667e-01 8.26032996e-01
1.86284304e-01 7.37830222e-01 1.36487293e+00 6.08709276e-01
1.35722637e-01 8.64345074e-01 8.47050488e-01 3.93715024e-01
3.05203199e-01 -7.48106092e-02 8.28636810e-02 3.70474100e-01
3.07151675e-01 -1.10889576e-01 -1.15429819e+00 6.14560008e-01
-1.89636087e+00 -6.60272837e-01 -7.39507228e-02 2.73285937e+00
2.94808418e-01 5.60719110e-02 -1.28860489e-01 4.51376736e-01
7.40396500e-01 3.04745853e-01 -4.25281554e-01 -2.72131234e-01
-1.36854798e-01 -2.81330585e-01 1.21307421e+00 7.99992502e-01
-7.26459801e-01 6.38415992e-01 6.49628496e+00 3.58110189e-01
-1.22831726e+00 -1.26845837e-01 -4.23832476e-01 4.64951992e-02
-2.17619300e-01 4.20040578e-01 -8.01448941e-01 4.85074848e-01
9.45979416e-01 -5.56529343e-01 2.13756934e-01 8.55182528e-01
1.38409093e-01 -8.59974146e-01 -7.93452919e-01 9.52264905e-01
1.95615470e-01 -1.17085528e+00 -2.68201679e-01 4.34394509e-01
6.29673183e-01 2.92543828e-01 -4.19828802e-01 -2.31359035e-01
1.35633916e-01 -5.83413541e-01 7.93537676e-01 4.03814554e-01
6.42237842e-01 -9.64950502e-01 9.91884589e-01 4.04878318e-01
-1.11440933e+00 -2.31875982e-02 -2.67461866e-01 -1.41692936e-01
3.70355725e-01 7.31071711e-01 -1.10057366e+00 8.98398221e-01
4.82042015e-01 1.58861980e-01 -4.95615542e-01 1.46954358e+00
-1.01183020e-01 -1.05123080e-01 -6.84696794e-01 -1.93252236e-01
-1.07904747e-01 -2.16765240e-01 9.98828292e-01 1.04828668e+00
5.61149836e-01 -4.38051462e-01 2.89338797e-01 3.60970229e-01
3.60467196e-01 2.31944188e-01 -8.72705579e-01 2.71135926e-01
9.27692235e-01 9.14223790e-01 -6.26410425e-01 5.51303811e-02
5.68902493e-02 5.65152228e-01 4.15920950e-02 -1.58870086e-01
-6.66206121e-01 -4.68710572e-01 8.99758875e-01 2.66794935e-02
-2.45426729e-01 -8.79976869e-01 -4.42458212e-01 -6.96898282e-01
1.01912364e-01 -6.33757710e-01 3.36741596e-01 -4.02538866e-01
-4.86436635e-01 6.65178239e-01 1.52403101e-01 -1.49843526e+00
-3.39018077e-01 -7.30806589e-02 1.18329093e-01 8.01415920e-01
-1.47967827e+00 -7.63606548e-01 -7.38618851e-01 2.38349825e-01
1.53807893e-01 9.06129628e-02 7.73953199e-01 3.55307400e-01
-1.94150761e-01 2.20829681e-01 1.04362018e-01 -5.77023506e-01
6.01733029e-01 -7.98801064e-01 1.38129383e-01 1.18395221e+00
-1.95231065e-01 7.31239557e-01 1.30267143e+00 -8.98968518e-01
-1.63417399e+00 -8.98900986e-01 1.07939637e+00 -2.11930543e-01
6.04149282e-01 -2.22697377e-01 -5.38085103e-01 5.43881476e-01
-3.68373841e-01 -3.40882331e-01 5.22702873e-01 6.09907582e-02
4.66245506e-03 -3.54588002e-01 -1.51826441e+00 1.70417875e-01
9.24585462e-01 -2.77829498e-01 -4.38932031e-01 2.69144654e-01
5.72499812e-01 -5.06079435e-01 -8.35113406e-01 7.81225502e-01
6.23904586e-01 -1.18482590e+00 7.16735601e-01 1.07557371e-01
-4.36920315e-01 -8.54725420e-01 -5.61843455e-01 -1.02155495e+00
1.14320084e-01 -4.66360807e-01 3.00238520e-01 9.76753592e-01
5.47437705e-02 -9.28021252e-01 7.03303695e-01 3.39213938e-01
1.21521495e-01 -2.89813548e-01 -1.19542384e+00 -1.18300128e+00
-7.74393141e-01 -4.11378801e-01 7.81153560e-01 7.06580579e-01
-3.31018448e-01 -1.06593847e-01 -4.01210070e-01 5.83775163e-01
8.96495998e-01 3.52415144e-01 1.37665975e+00 -1.47269857e+00
5.26584722e-02 -2.38678623e-02 -1.07824898e+00 -4.59042192e-01
-2.41347879e-01 -5.70973217e-01 2.40788877e-01 -1.55937827e+00
-1.39189944e-01 -7.93034196e-01 -1.61149472e-01 2.61957385e-02
3.79795104e-01 7.70835280e-02 1.19816564e-01 5.40290952e-01
-4.63431120e-01 1.67091966e-01 4.40169334e-01 3.04809362e-01
-3.21140617e-01 -4.18446437e-02 1.03076333e-02 3.92164528e-01
6.97904944e-01 -8.17456722e-01 -2.83769131e-01 -4.08198267e-01
6.79926574e-01 1.53916821e-01 1.24543151e-02 -1.41692507e+00
3.80098134e-01 -3.83870691e-01 -3.28639656e-01 -9.61902261e-01
4.16432619e-01 -1.06226957e+00 8.55333209e-01 8.92971754e-01
4.42302711e-02 3.23614508e-01 1.66536704e-01 6.04806483e-01
-2.74053782e-01 -3.65567088e-01 8.37147236e-01 1.53933987e-01
-1.18057179e+00 2.88127027e-02 -1.81563552e-02 -4.72627491e-01
1.13808835e+00 -4.71013248e-01 -1.05542354e-01 -2.42834628e-01
-1.64614260e-01 3.98870170e-01 1.16212130e+00 3.57483029e-01
3.69991660e-01 -1.04653454e+00 -4.61890846e-01 1.64230511e-01
5.48898935e-01 -1.29022256e-01 1.58704415e-01 8.95628393e-01
-1.04816997e+00 3.12112987e-01 -4.95590270e-01 -8.02700043e-01
-1.35967076e+00 3.13636959e-01 2.14679047e-01 -3.25043470e-01
-3.27481449e-01 3.53481203e-01 -5.94175875e-01 -5.04212260e-01
1.18764997e-01 4.63112108e-02 7.83504322e-02 1.84645429e-01
2.25159168e-01 4.69771147e-01 2.34668255e-01 -9.08854365e-01
-7.63968945e-01 8.06389511e-01 6.60061359e-01 -3.06651860e-01
1.27705359e+00 -5.73720694e-01 -5.19020736e-01 4.76274759e-01
9.74914551e-01 4.91649568e-01 -7.61992514e-01 5.26759550e-02
4.16456223e-01 -8.31782997e-01 -3.20735991e-01 -4.66398060e-01
-4.85233366e-01 2.81082064e-01 5.88242888e-01 1.12332128e-01
1.03716838e+00 -3.74883056e-01 4.10888851e-01 4.24125880e-01
1.32094669e+00 -1.01060808e+00 -4.10250485e-01 5.99876642e-01
8.07904243e-01 -1.13865149e+00 4.75051522e-01 -3.79267961e-01
-3.39901865e-01 7.45284677e-01 1.08459398e-01 -2.64101386e-01
2.51229614e-01 3.12939018e-01 -9.82781500e-02 -1.49294332e-01
-2.10586295e-01 -3.30594182e-01 -2.04046786e-01 5.37571073e-01
-1.71706319e-01 1.47618651e-01 -9.20928359e-01 -1.04737103e-01
-5.69425583e-01 -2.23370776e-01 6.04230225e-01 1.30776334e+00
-1.01545620e+00 -1.24331498e+00 -4.68653888e-01 5.59461378e-02
7.03545287e-02 2.49340653e-01 -2.88772017e-01 9.89140451e-01
4.38814312e-02 9.11946774e-01 -7.95323104e-02 -6.64099813e-01
3.56839091e-01 -2.20040083e-01 4.01462615e-01 -3.46101522e-01
-2.86627293e-01 -5.37511110e-01 3.54105622e-01 -6.82745516e-01
-3.45511317e-01 -7.85180926e-01 -1.22970605e+00 -3.43814880e-01
-3.16595882e-01 4.14191574e-01 1.45744836e+00 7.22046316e-01
5.63606799e-01 2.16603562e-01 6.64290190e-01 -5.49607754e-01
-3.93188685e-01 -5.36366940e-01 -5.77136815e-01 1.50533333e-01
3.16072017e-01 -9.15539801e-01 -4.31236744e-01 -5.51823258e-01] | [7.316073894500732, -2.045354127883911] |
ae4a3f3f-db7b-4d64-bc70-d61666413fd2 | 190910390 | 1909.1039 | null | https://arxiv.org/abs/1909.10390v1 | https://arxiv.org/pdf/1909.10390v1.pdf | GNTeam at 2018 n2c2: Feature-augmented BiLSTM-CRF for drug-related entity recognition in hospital discharge summaries | Monitoring the administration of drugs and adverse drug reactions are key parts of pharmacovigilance. In this paper, we explore the extraction of drug mentions and drug-related information (reason for taking a drug, route, frequency, dosage, strength, form, duration, and adverse events) from hospital discharge summaries through deep learning that relies on various representations for clinical named entity recognition. This work was officially part of the 2018 n2c2 shared task, and we use the data supplied as part of the task. We developed two deep learning architecture based on recurrent neural networks and pre-trained language models. We also explore the effect of augmenting word representations with semantic features for clinical named entity recognition. Our feature-augmented BiLSTM-CRF model performed with F1-score of 92.67% and ranked 4th for entity extraction sub-task among submitted systems to n2c2 challenge. The recurrent neural networks that use the pre-trained domain-specific word embeddings and a CRF layer for label optimization perform drug, adverse event and related entities extraction with micro-averaged F1-score of over 91%. The augmentation of word vectors with semantic features extracted using available clinical NLP toolkits can further improve the performance. Word embeddings that are pre-trained on a large unannotated corpus of relevant documents and further fine-tuned to the task perform rather well. However, the augmentation of word embeddings with semantic features can help improve the performance (primarily by boosting precision) of drug-related named entity recognition from electronic health records. | ['Haifa Alrdahi', 'Goran Nenadic', 'Maksim Belousov', 'Ghada Alfattni', 'Nikola Milosevic'] | 2019-09-23 | null | null | null | null | ['entity-extraction'] | ['natural-language-processing'] | [ 8.10637251e-02 1.96266055e-01 -5.00027180e-01 -4.13086325e-01
-9.52057838e-01 -4.46457356e-01 3.15133214e-01 8.61954629e-01
-7.79925823e-01 7.33240545e-01 6.74510300e-01 -5.94607890e-01
-6.23806566e-03 -7.28397787e-01 -5.43651521e-01 -3.92500728e-01
-3.20183814e-01 5.97861052e-01 -5.69650054e-01 4.54246700e-02
-4.01717350e-02 5.45796335e-01 -8.10859501e-01 5.10553420e-01
7.81174779e-01 8.71082783e-01 -5.71113229e-02 5.84218264e-01
-2.28614956e-01 5.17477155e-01 -7.56554902e-01 -1.73177645e-01
-1.37665510e-01 -7.66212866e-02 -8.59174371e-01 -2.87945807e-01
-2.52496928e-01 -2.24623844e-01 -1.08580969e-01 6.95409536e-01
8.98178220e-01 8.20706785e-02 8.00594747e-01 -5.18464565e-01
-1.10303044e+00 5.65398753e-01 -1.51995763e-01 2.07377046e-01
4.38833445e-01 9.24865007e-02 8.41928124e-01 -8.52407813e-01
7.35873222e-01 7.52617300e-01 7.52452135e-01 6.93378568e-01
-9.31124449e-01 -6.88099682e-01 -1.45733416e-01 -1.68500077e-02
-1.47326910e+00 -4.10438627e-01 3.12222298e-02 -6.59895301e-01
1.88755059e+00 -1.46284014e-01 1.97842762e-01 1.28935552e+00
6.24709249e-01 4.88133997e-01 6.20645642e-01 -1.40792713e-01
2.16824755e-01 2.26494208e-01 3.37193459e-01 6.08225167e-01
3.85851622e-01 1.56806499e-01 8.67914874e-03 -6.97760403e-01
4.83361065e-01 3.51357877e-01 -2.81052411e-01 2.02757463e-01
-9.44563270e-01 1.19326711e+00 3.08145732e-01 4.18923408e-01
-8.02736223e-01 -1.41195059e-01 6.93941355e-01 -6.71847165e-02
5.18023193e-01 1.02404690e+00 -1.21467721e+00 -4.02043946e-02
-8.44163060e-01 6.55286014e-02 9.28995013e-01 8.08414042e-01
1.40118629e-01 -8.72594789e-02 -6.27685249e-01 9.57368910e-01
3.33703667e-01 2.52080947e-01 9.49910700e-01 -1.20344296e-01
3.54380816e-01 7.60702014e-01 8.76498818e-02 -5.68726838e-01
-9.53098416e-01 -4.96416032e-01 -7.14324951e-01 -4.34660405e-01
-1.16032548e-01 -5.26244164e-01 -1.43107200e+00 1.71750581e+00
1.88777313e-01 1.11461587e-01 4.88025367e-01 5.49668312e-01
1.24417078e+00 5.88694692e-01 1.00971198e+00 -1.33113474e-01
1.99947762e+00 -8.29806089e-01 -1.12012017e+00 1.00241475e-01
1.15471947e+00 -8.96148682e-01 3.70635599e-01 -4.96135019e-02
-7.07242608e-01 -2.16843635e-01 -1.00166833e+00 -1.39402062e-01
-9.72166896e-01 3.22294861e-01 5.54644048e-01 5.30446053e-01
-6.49820864e-01 6.75599635e-01 -1.03346288e+00 -2.43120000e-01
8.20147157e-01 6.76319003e-01 -4.23485577e-01 -2.21327320e-03
-1.59018970e+00 1.38479865e+00 3.01801503e-01 -1.55397654e-01
-6.35812998e-01 -1.20506036e+00 -1.33433318e+00 1.76555261e-01
-5.69383428e-02 -9.30973351e-01 1.18146849e+00 -1.46831244e-01
-1.29251707e+00 6.73405290e-01 -1.25843212e-01 -5.21764696e-01
-2.01732665e-01 -2.21818805e-01 -8.78311932e-01 -4.68000285e-02
2.58622229e-01 4.36936885e-01 4.06081192e-02 -1.40550539e-01
-3.09228569e-01 -5.00185013e-01 -4.43524688e-01 -7.52697513e-02
-2.10187986e-01 2.48722300e-01 -7.19827339e-02 -7.70348549e-01
-7.42518604e-01 -7.88952410e-01 -7.04910934e-01 -6.54949844e-01
-4.16169524e-01 -6.11888111e-01 1.17169261e-01 -8.33038330e-01
1.41961670e+00 -1.86121273e+00 -3.76671165e-01 -3.23253460e-02
2.57135928e-01 6.90804303e-01 -4.71358031e-01 5.06437957e-01
-5.68871498e-01 5.04009366e-01 -6.84114918e-02 -9.86900926e-02
-1.82384104e-01 -1.41678870e-01 1.79082360e-02 4.37775105e-01
8.87556911e-01 1.18712211e+00 -8.37590933e-01 -1.45858034e-01
1.58736929e-01 9.82521057e-01 -4.86656815e-01 3.94347161e-01
-9.66808647e-02 2.00872809e-01 -8.15879226e-01 5.20348966e-01
5.99688888e-01 -5.57633162e-01 1.71402082e-01 -2.46696621e-01
1.24435902e-01 9.38636780e-01 -7.29469299e-01 1.76995552e+00
-7.37890184e-01 -1.08191008e-02 -5.36774457e-01 -7.56867111e-01
6.63924396e-01 1.01322579e+00 8.98391902e-01 -6.42807245e-01
2.89769650e-01 1.77005500e-01 -8.98417011e-02 -8.25746119e-01
3.14881146e-01 -3.24637771e-01 -1.47946388e-01 3.38100672e-01
3.53136122e-01 4.15195078e-01 -2.30171934e-01 -1.46131530e-01
1.42509651e+00 -3.19004729e-02 8.04205358e-01 -9.32098404e-02
3.97003800e-01 1.92043632e-02 4.90491629e-01 3.61654490e-01
-1.09817959e-01 4.77899402e-01 1.66319400e-01 -5.22868156e-01
-8.96349788e-01 -6.15651608e-01 -8.03952515e-01 6.97368562e-01
-6.65706515e-01 -6.12007022e-01 -4.43841398e-01 -7.79760003e-01
1.72620006e-02 6.49294138e-01 -6.80955529e-01 -1.52782932e-01
-3.67400736e-01 -1.23346007e+00 7.50620842e-01 6.86621785e-01
-2.48502120e-01 -1.27340615e+00 -2.83844978e-01 8.81858170e-01
3.12602013e-01 -1.26097071e+00 -7.15502620e-01 6.08296394e-01
-6.86265290e-01 -1.34332275e+00 -9.73260462e-01 -8.95814538e-01
4.06660259e-01 -6.37886107e-01 1.02099550e+00 -2.54550487e-01
-4.44243997e-01 7.62272701e-02 -2.28002161e-01 -6.27208829e-01
-2.49525025e-01 2.79191405e-01 -6.56794384e-02 -3.70657861e-01
1.04382372e+00 -1.54779017e-01 -8.69289696e-01 -1.35266498e-01
-9.26509380e-01 -3.23706985e-01 5.08742571e-01 7.41796732e-01
6.35013819e-01 -5.96151412e-01 9.57357764e-01 -1.26393521e+00
9.38018680e-01 -8.35195124e-01 -2.58017063e-01 2.59997845e-01
-8.34827542e-01 4.74645853e-01 6.88612223e-01 -3.84179085e-01
-7.84823298e-01 3.06042194e-01 -6.41942263e-01 -4.20649862e-03
-3.21730524e-01 7.94901848e-01 3.06998491e-02 5.87088287e-01
6.95906222e-01 -1.99184269e-01 -2.36556247e-01 -5.89725018e-01
3.28546286e-01 1.07549322e+00 2.96191312e-02 -1.28383398e-01
1.27822869e-02 -7.40697682e-02 -2.61565596e-01 -4.85390186e-01
-9.80287194e-01 -7.44433820e-01 -1.33460477e-01 9.31540489e-01
1.40541422e+00 -1.24848437e+00 -8.55284870e-01 8.95351022e-02
-1.47612786e+00 8.33116919e-02 -1.90271139e-01 9.69149470e-01
-1.24292932e-01 -3.38328183e-02 -8.98275614e-01 -4.97049123e-01
-1.00924289e+00 -1.34830439e+00 9.89907861e-01 2.41079032e-01
-5.80345988e-01 -1.12948930e+00 6.18396044e-01 4.32977453e-02
5.90837598e-01 3.28410119e-01 1.10456538e+00 -1.72452760e+00
9.23945680e-02 -3.72613519e-01 -2.39055857e-01 1.36903465e-01
6.18495643e-01 -3.97275329e-01 -8.77043486e-01 5.78884482e-02
-6.88144341e-02 -8.78619328e-02 9.44991648e-01 7.96445727e-01
1.17908502e+00 -4.41063613e-01 -4.59413856e-01 5.81816852e-01
1.43822145e+00 6.07475698e-01 5.96076846e-01 1.47448167e-01
6.31755829e-01 4.15175617e-01 3.76633182e-02 4.76805657e-01
5.17420828e-01 5.76749146e-01 -1.26851037e-01 -3.58617455e-01
-5.48561290e-03 -2.96294969e-02 1.09160937e-01 5.89708447e-01
2.47880697e-01 -1.88141435e-01 -9.03075755e-01 7.32564449e-01
-1.48325229e+00 -4.44834590e-01 -9.31726620e-02 1.94117427e+00
1.36248147e+00 -3.59677374e-01 -1.09789982e-01 -6.24181986e-01
6.57391489e-01 -1.26614213e-01 -5.94256043e-01 -9.21497464e-01
9.42364261e-02 1.10128486e+00 8.17309916e-01 4.15009439e-01
-1.22123671e+00 8.11904728e-01 5.78093719e+00 6.79294348e-01
-1.02121270e+00 2.35399053e-01 7.25507915e-01 2.19756812e-01
-2.43278742e-01 -4.53852922e-01 -1.14808583e+00 3.94900799e-01
1.75587070e+00 -2.65228264e-02 -1.49947004e-02 5.74970663e-01
4.17941153e-01 4.91429746e-01 -1.25865364e+00 8.17171454e-01
2.81790774e-02 -1.71224463e+00 5.28229922e-02 1.83474585e-01
6.26903653e-01 4.26343173e-01 3.08443159e-02 4.44151938e-01
5.29765189e-01 -1.65461218e+00 -2.77305335e-01 3.95187706e-01
1.07976472e+00 -6.34335041e-01 1.21118772e+00 -2.44733021e-01
-9.91754770e-01 2.97952503e-01 -2.52916396e-01 3.90021831e-01
2.52304882e-01 6.90647542e-01 -1.38369751e+00 6.44649565e-01
2.92722523e-01 1.04505301e+00 -3.36213082e-01 1.02658546e+00
-1.80784732e-01 4.39721763e-01 -1.83913797e-01 -8.00979659e-02
2.75603622e-01 1.70735255e-01 -6.85784128e-03 1.50034082e+00
1.53050452e-01 2.78934866e-01 8.21141005e-02 8.17281008e-01
-4.76736009e-01 4.68071043e-01 -6.04255140e-01 -6.08006716e-01
2.66488820e-01 1.17413688e+00 -3.04404289e-01 -5.34525871e-01
-4.08511519e-01 7.12465465e-01 9.86517295e-02 2.57883132e-01
-8.83554578e-01 -5.42965293e-01 8.72503042e-01 -8.27150121e-02
4.47051167e-01 3.64460528e-01 -2.87177265e-01 -1.05674636e+00
-4.29520816e-01 -7.81711519e-01 6.98850155e-01 -2.89211005e-01
-1.54233587e+00 1.05099845e+00 -4.65653777e-01 -1.15479302e+00
-3.91254425e-01 -7.43607521e-01 -3.64104033e-01 1.28171086e+00
-1.76394653e+00 -8.94997835e-01 4.76545036e-01 3.49045515e-01
3.93738180e-01 -3.40641856e-01 1.66973090e+00 7.12933779e-01
-7.34806240e-01 7.14200020e-01 1.14605807e-01 4.22010273e-01
8.48793447e-01 -1.10147285e+00 5.41728616e-01 -3.16115879e-02
-1.22278836e-02 1.05397165e+00 3.01777035e-01 -7.15678692e-01
-8.42409194e-01 -1.62117040e+00 1.53124678e+00 -7.79506564e-01
6.03746355e-01 -1.60965040e-01 -9.72340703e-01 5.78709722e-01
1.90669030e-01 -1.65544301e-02 1.54817927e+00 2.94087917e-01
-3.05184573e-01 4.31307226e-01 -1.16829932e+00 2.52884924e-01
4.15262431e-01 -7.54198015e-01 -8.66354406e-01 8.50644946e-01
1.05090463e+00 -3.84444863e-01 -1.53728974e+00 3.53619456e-01
3.80082250e-01 2.96723247e-01 1.07303321e+00 -1.60362458e+00
6.00880206e-01 -4.71276008e-02 7.07724616e-02 -1.27472126e+00
-2.18808293e-01 -3.53356779e-01 1.05451353e-01 8.59288812e-01
1.19078362e+00 -6.38312399e-01 5.00758469e-01 8.03438306e-01
-1.27919227e-01 -1.07990348e+00 -5.98652303e-01 -4.34056789e-01
2.61479855e-01 -2.77253568e-01 7.06370890e-01 1.10705876e+00
2.41700634e-01 7.36175418e-01 -9.22915339e-02 2.15394333e-01
-5.45780472e-02 -1.15839727e-01 -2.89853872e-03 -1.02450597e+00
-4.80054878e-02 -1.77532360e-01 -2.73911744e-01 -5.17662704e-01
2.44504705e-01 -1.13302279e+00 -2.72797883e-01 -1.82251740e+00
3.20820868e-01 -2.62975633e-01 -8.19325089e-01 9.77162898e-01
-3.98953915e-01 -1.50864273e-01 -2.60447741e-01 -9.76750106e-02
-4.53873217e-01 4.02598351e-01 8.73532414e-01 -2.97406822e-01
-6.43803179e-01 1.47569515e-02 -9.17249620e-01 3.40841085e-01
6.39010906e-01 -9.67599750e-01 1.70631602e-01 -3.24776292e-01
1.27623126e-01 1.88641176e-01 -2.37619862e-01 -7.72349536e-02
1.14386708e-01 5.40287681e-02 5.55221260e-01 -2.36192241e-01
-7.89465606e-02 -6.62300587e-01 2.41311174e-02 6.16345108e-01
-5.93645632e-01 2.08839193e-01 5.02705991e-01 5.40178657e-01
-1.70151636e-01 -1.72298208e-01 4.61574167e-01 -3.21031287e-02
-1.14127867e-01 4.35281307e-01 -5.67213416e-01 -1.04899509e-02
6.72552466e-01 1.24059223e-01 -2.45389402e-01 1.92118645e-01
-1.07001269e+00 3.01622808e-01 -2.25610197e-01 5.10549784e-01
4.82480258e-01 -1.20333517e+00 -7.93801725e-01 5.60285486e-02
3.99841547e-01 -2.22665519e-01 3.29189688e-01 6.80682659e-01
-3.96274239e-01 1.02195823e+00 5.57416230e-02 -1.37673482e-01
-8.75431061e-01 7.39929736e-01 3.42174053e-01 -1.01035714e+00
-4.46766645e-01 6.64447308e-01 2.46603653e-01 -6.80562556e-01
6.62892237e-02 -5.93562007e-01 -7.61173189e-01 7.03526884e-02
6.74188793e-01 -2.23443449e-01 6.10651672e-01 -3.60505044e-01
-8.96036685e-01 3.57391745e-01 -4.81304944e-01 3.84076208e-01
1.96659243e+00 5.22427201e-01 1.62868336e-01 -1.34247780e-01
1.85194516e+00 -8.06696564e-02 -4.34491873e-01 -1.09482609e-01
2.39505991e-01 3.40600908e-01 1.46365508e-01 -1.30247474e+00
-1.09305155e+00 6.07625008e-01 8.82536471e-01 -4.33639914e-01
7.69560397e-01 -4.16650586e-02 1.02779412e+00 1.28575787e-01
-2.76813447e-01 -8.15181494e-01 -4.58698362e-01 5.01211226e-01
5.18472612e-01 -1.21174169e+00 4.71921340e-02 5.25455661e-02
-6.66028023e-01 9.93681669e-01 1.37656912e-01 -9.91825238e-02
9.28677261e-01 2.69852161e-01 -4.73466190e-03 -6.85147643e-01
-7.36370325e-01 -1.00147106e-01 6.50878251e-01 4.18884665e-01
1.00938630e+00 3.06985736e-01 -6.01251662e-01 1.28971982e+00
1.61245719e-01 2.49321371e-01 3.83157909e-01 6.93870664e-01
1.63489416e-01 -1.39883804e+00 1.35199711e-01 6.25719666e-01
-1.14150965e+00 -6.70642912e-01 -9.29183662e-02 4.08851862e-01
9.08009037e-02 9.24774051e-01 -1.07580088e-01 -8.47779959e-02
7.04183042e-01 3.18951130e-01 6.99369088e-02 -1.26859581e+00
-1.18292856e+00 -1.47235133e-02 2.36535072e-01 -5.79857171e-01
-3.98947060e-01 -1.55036360e-01 -1.53300142e+00 2.05146745e-01
-4.45492864e-01 3.99899453e-01 8.82904828e-01 1.03390634e+00
1.00117183e+00 8.69438589e-01 3.19868296e-01 -8.44290555e-02
-5.57524443e-01 -1.08296835e+00 -3.28701317e-01 3.87150705e-01
4.78055418e-01 -3.37539315e-01 -4.84248139e-02 2.51356568e-02] | [8.414437294006348, 8.706585884094238] |
b3232ac6-6ef0-4d78-8d84-b48fb77c5e0d | graph-similarity-using-pagerank-and | 2002.05158 | null | https://arxiv.org/abs/2002.05158v2 | https://arxiv.org/pdf/2002.05158v2.pdf | Fast and Scalable Complex Network Descriptor Using PageRank and Persistent Homology | The PageRank of a graph is a scalar function defined on the node set of the graph which encodes nodes centrality information of the graph. In this article, we use the PageRank function along with persistent homology to obtain a scalable graph descriptor and utilize it to compare the similarities between graphs. For a given graph $G(V,E)$, our descriptor can be computed in $O(|E|\alpha(|V|))$, where $\alpha$ is the inverse Ackermann function which makes it scalable and computable on massive graphs. We show the effectiveness of our method by utilizing it on multiple shape mesh datasets. | ['Mustafa Hajij', 'Elizabeth Munch', 'Paul Rosen'] | 2020-02-12 | null | null | null | null | ['graph-similarity'] | ['graphs'] | [-3.36868197e-01 4.06624705e-01 -1.71250463e-01 2.78459787e-02
-2.54822314e-01 -7.87499905e-01 1.97670057e-01 8.23996365e-01
-2.50335395e-01 2.03840807e-01 5.02938963e-02 -2.02214420e-01
-2.78514653e-01 -1.42757738e+00 -6.56005144e-01 -1.73286319e-01
-1.02876174e+00 4.03885126e-01 5.39542854e-01 -2.97313124e-01
3.74666572e-01 5.17899156e-01 -1.06014037e+00 -3.19412857e-01
4.84928250e-01 1.07104897e+00 -1.19001709e-01 8.34028482e-01
2.89191548e-02 4.50240821e-01 -3.12138766e-01 -5.16931176e-01
3.69194984e-01 -1.78077519e-01 -1.12226188e+00 -8.03844854e-02
4.69496161e-01 3.07487249e-02 -6.92973435e-01 1.06061482e+00
1.66348562e-01 1.11968882e-01 6.53753698e-01 -1.25305593e+00
-1.47358879e-01 3.49556774e-01 -6.78932726e-01 1.98160425e-01
7.70719290e-01 -1.54590204e-01 1.50885081e+00 -7.74444699e-01
1.08224213e+00 1.17067647e+00 8.51481855e-01 -3.53512734e-01
-1.11302936e+00 -3.30261648e-01 -2.88157254e-01 -1.87133700e-02
-1.52824736e+00 1.78702980e-01 7.81171381e-01 -4.11409289e-01
6.39948606e-01 3.56224984e-01 1.16874897e+00 -2.64413077e-02
2.20233187e-01 -2.00488120e-02 9.30611014e-01 -7.85548538e-02
3.11541319e-01 -7.41193473e-01 3.30378652e-01 1.42232919e+00
5.74442565e-01 -1.93385646e-01 -4.05503184e-01 -4.95761901e-01
8.42024148e-01 -2.06334412e-01 -1.25513121e-01 -5.78166962e-01
-1.08971834e+00 8.29030156e-01 8.68772686e-01 -3.28012779e-02
-1.85565278e-01 5.74330449e-01 3.78646255e-01 4.44813907e-01
2.75015205e-01 2.99232006e-01 -1.18418947e-01 -5.74734546e-02
-4.42996860e-01 7.73584917e-02 9.70328867e-01 8.91357183e-01
1.35850251e+00 -4.19229239e-01 3.94906372e-01 5.23888886e-01
2.25097299e-01 5.14982045e-01 -4.83859688e-01 -1.23204660e+00
2.40280986e-01 1.18719172e+00 -3.32659841e-01 -1.82713079e+00
-6.22742355e-01 -7.21171796e-02 -1.06128585e+00 -2.47275014e-03
2.84702510e-01 4.53293264e-01 -4.91809994e-01 1.67600751e+00
6.54805064e-01 8.19679573e-02 -4.28994209e-01 5.86952746e-01
7.86296606e-01 2.92127997e-01 -1.30658567e-01 -1.14526473e-01
1.28080738e+00 -6.03374124e-01 -9.08413753e-02 3.00170153e-01
6.85582399e-01 -6.22244656e-01 1.06230187e+00 -1.08722992e-01
-1.24580276e+00 6.54686168e-02 -1.11563873e+00 -1.27530033e-02
-4.15188342e-01 -4.67852354e-01 5.72793961e-01 2.92219609e-01
-1.36567819e+00 8.40662420e-01 -8.23656857e-01 -3.04780781e-01
8.49408433e-02 5.11767507e-01 -6.29124105e-01 -5.65082766e-02
-1.03108466e+00 6.09046400e-01 5.23022413e-02 -3.43639553e-01
-5.63778996e-01 -5.41071057e-01 -9.02460814e-01 1.43117830e-02
3.43179792e-01 -5.74711621e-01 4.81165826e-01 -2.21152246e-01
-1.01134419e+00 7.93866515e-01 1.24567509e-01 -8.55122209e-02
1.60059392e-01 6.51305020e-01 -3.60613950e-02 6.54147625e-01
9.07233134e-02 2.19007537e-01 3.65835190e-01 -9.51171637e-01
-2.11107209e-02 -4.51988578e-01 3.97638887e-01 1.89010799e-02
-7.38002434e-02 -2.10052490e-01 -5.01362324e-01 -4.77296084e-01
5.61845064e-01 -1.07017612e+00 -2.64159471e-01 3.82825106e-01
-2.80768812e-01 -3.58425647e-01 5.07071257e-01 -4.55032676e-01
1.32943273e+00 -2.32125187e+00 1.31993994e-01 1.09218192e+00
1.14947045e+00 -2.57786334e-01 -1.03334181e-01 7.96750605e-01
1.52211875e-01 4.05045718e-01 -2.57040739e-01 6.17068291e-01
-1.31744072e-01 1.86562881e-01 3.38763654e-01 9.59125102e-01
-1.60540149e-01 5.44115484e-01 -9.72704947e-01 -7.45218217e-01
-2.50618041e-01 3.77069086e-01 -7.38971293e-01 2.19585504e-02
6.00129254e-02 -1.28844008e-01 -5.61877906e-01 4.82186526e-01
7.47593045e-01 -7.73052573e-01 6.70209169e-01 -3.76622736e-01
3.10477559e-02 9.47861001e-02 -1.47360957e+00 1.39393032e+00
-7.77932405e-02 4.52845901e-01 3.35052550e-01 -7.97882318e-01
7.58205593e-01 -1.65227786e-01 8.42015028e-01 -4.08788353e-01
1.90664455e-01 2.03761116e-01 -2.25183353e-01 7.40922391e-02
2.22904548e-01 4.15430725e-01 -1.75493345e-01 6.38735473e-01
-1.71750233e-01 -5.39412610e-02 4.79663700e-01 8.47026587e-01
1.96558785e+00 -5.68802655e-01 4.46112335e-01 -7.29958951e-01
3.37448895e-01 -8.59293863e-02 1.74575448e-01 2.40812719e-01
-1.92114159e-01 2.06494659e-01 9.57087755e-01 -3.77971917e-01
-1.10376084e+00 -1.35367870e+00 1.37401730e-01 9.29337561e-01
6.76573873e-01 -1.20382416e+00 -6.60255730e-01 -5.67682862e-01
2.44678766e-01 -3.73873442e-01 -6.71061277e-01 -2.52378248e-02
-6.36725426e-01 -1.69166043e-01 2.61947274e-01 3.56771618e-01
3.42771590e-01 -5.30201793e-01 -4.23197389e-01 -7.85046071e-02
-5.15231378e-02 -7.98684895e-01 -9.64580536e-01 -5.00294924e-01
-8.42913508e-01 -1.79429495e+00 -1.60134003e-01 -8.74427497e-01
1.08725274e+00 3.22184324e-01 1.17729759e+00 7.46826947e-01
-4.00490791e-01 6.71543777e-01 -2.39509359e-01 3.53148997e-01
-2.06455410e-01 -8.52909219e-03 4.28044889e-03 -3.25558811e-01
-6.09076262e-01 -7.55520165e-01 -9.91103411e-01 4.02408093e-01
-8.41385484e-01 -1.48785248e-01 1.46592095e-01 5.90971947e-01
8.94053698e-01 -4.25870940e-02 -6.93970323e-02 -7.53232300e-01
6.62044287e-01 -3.92221302e-01 -7.93148279e-01 3.55123967e-01
-7.78250635e-01 2.84791440e-01 4.22427267e-01 -1.41818404e-01
1.41374096e-01 2.37045661e-02 1.34728819e-01 -9.44411159e-02
9.17415559e-01 8.44888389e-01 1.92916080e-01 -8.35045278e-01
3.29771042e-01 -2.51117200e-01 5.06619886e-02 -3.84238154e-01
5.34086585e-01 4.23670076e-02 4.76059645e-01 -5.42877197e-01
7.70105004e-01 5.80855846e-01 9.04825985e-01 -9.03772593e-01
-1.54896602e-01 -4.57813889e-01 -4.69507337e-01 -3.48270118e-01
4.48367953e-01 -5.45442700e-01 -1.46971869e+00 -5.42591885e-02
-9.88989234e-01 6.22792058e-02 4.08167159e-03 3.52259487e-01
-3.89793396e-01 6.80704951e-01 -6.25829935e-01 -4.67593759e-01
-3.94461006e-01 -7.20671535e-01 8.31015944e-01 -2.59570479e-01
-2.29833767e-01 -9.53309834e-01 5.06263793e-01 -6.00412153e-02
4.12238657e-01 6.58267379e-01 1.34800589e+00 -2.99271345e-01
-7.16156244e-01 -2.40059495e-01 -5.85531831e-01 -1.85839742e-01
-1.54530391e-01 2.45751828e-01 2.16250449e-01 -4.21360254e-01
-6.12111211e-01 2.98027229e-03 6.30656123e-01 -1.33475810e-01
9.55099881e-01 -8.28623295e-01 -3.24067563e-01 7.25010276e-01
1.59470403e+00 -5.37667036e-01 4.63607222e-01 4.69301455e-02
7.90245235e-01 1.52308017e-01 2.43122622e-01 5.20522058e-01
6.11381948e-01 4.10685599e-01 5.51662207e-01 -1.72374863e-02
-2.81200022e-01 -2.85214752e-01 -6.06903853e-03 1.37420511e+00
-3.80753547e-01 1.40286475e-01 -1.14261711e+00 2.90403306e-01
-1.59160209e+00 -5.34852028e-01 -3.32511306e-01 2.08583689e+00
5.74386418e-01 2.64683575e-03 1.74766615e-01 -1.89262420e-01
8.28007996e-01 3.26434195e-01 -3.26642931e-01 -3.60446095e-01
-1.74438683e-04 7.18244016e-01 8.39603245e-01 7.46205211e-01
-5.57295680e-01 6.17084742e-01 7.26493597e+00 5.93213320e-01
-9.30593014e-01 5.29273190e-02 1.54178530e-01 1.83385342e-01
-4.61571604e-01 3.57404619e-01 5.84610701e-02 4.01734650e-01
6.74697995e-01 -7.46326029e-01 6.51706398e-01 8.73317063e-01
-3.06500375e-01 -1.96842343e-01 -7.21972644e-01 9.79886830e-01
-2.27042511e-01 -1.38136005e+00 -8.16548616e-02 3.86773974e-01
4.96422231e-01 1.19694799e-01 -2.06243902e-01 -2.24674314e-01
4.38599676e-01 -9.01901066e-01 1.36820942e-01 1.39036328e-01
1.04363215e+00 -8.37166607e-01 5.87224849e-02 -8.47736225e-02
-1.98633230e+00 3.93065155e-01 -2.15907350e-01 1.46201879e-01
1.83488987e-03 5.46396792e-01 -7.14299202e-01 5.50839901e-01
5.60216725e-01 4.02328789e-01 -5.25964618e-01 9.08105135e-01
7.81693012e-02 1.37649924e-01 -5.78347147e-01 -7.03414008e-02
-2.00899988e-02 -5.51561117e-01 7.03472197e-01 8.19441259e-01
3.41583997e-01 4.03062254e-01 5.67108512e-01 2.93343514e-01
-6.15220964e-01 5.46630621e-01 -7.56687641e-01 -2.37181216e-01
5.38081169e-01 1.26628733e+00 -1.11629844e+00 -2.58045137e-01
-2.21345216e-01 6.60114825e-01 7.76277721e-01 -7.08442703e-02
-5.60151339e-01 -6.14844620e-01 7.85244286e-01 5.33618927e-01
2.78660446e-01 -8.65209997e-01 6.76567256e-02 -9.49756622e-01
1.92065388e-02 -5.35432339e-01 5.55430055e-01 -5.79925478e-01
-1.18267632e+00 5.40680945e-01 -4.92904037e-02 -8.37202907e-01
2.40381539e-01 -6.36313140e-01 -4.21706170e-01 3.30498725e-01
-7.53642440e-01 -6.91117704e-01 -3.39984328e-01 7.92477787e-01
-4.80789006e-01 3.34650755e-01 8.78723860e-01 1.21901929e-01
-2.24829212e-01 3.05487186e-01 -9.22399014e-02 2.03338251e-01
2.11352095e-01 -1.13860536e+00 4.17977214e-01 4.41993058e-01
-9.37109962e-02 6.39729619e-01 5.00278831e-01 -8.59053314e-01
-2.14358306e+00 -8.74455750e-01 6.31105602e-01 -6.53107017e-02
1.08680105e+00 -3.27361226e-01 -6.82865977e-01 4.67671484e-01
-1.03806853e-01 4.64974165e-01 5.68668067e-01 4.63641807e-02
-7.94610381e-01 -2.71948665e-01 -1.22629201e+00 6.40674651e-01
1.47371066e+00 -6.36467278e-01 1.01455048e-01 3.06464940e-01
6.91820323e-01 -4.00293946e-01 -1.66818011e+00 6.06878519e-01
6.75133169e-01 -7.17606187e-01 9.99340236e-01 -2.93126523e-01
-2.12850925e-02 -1.83399200e-01 -1.71609625e-01 -1.11988008e+00
-4.60178941e-01 -7.07327962e-01 -2.07298055e-01 5.10816038e-01
3.21848512e-01 -7.98162997e-01 6.67720795e-01 2.73384869e-01
3.92017305e-01 -1.01049554e+00 -1.14479864e+00 -5.52848458e-01
-2.01212242e-01 -5.80957755e-02 7.85930693e-01 9.44666803e-01
4.03235257e-01 1.83838263e-01 2.19684750e-01 1.30935118e-01
8.69540691e-01 2.06454441e-01 6.48244500e-01 -1.47197783e+00
-6.46815002e-02 -3.97227347e-01 -9.60576177e-01 -6.38815820e-01
1.38364406e-02 -1.35421693e+00 -4.28933352e-01 -1.47834551e+00
3.55201036e-01 -5.69482803e-01 -8.08245242e-02 3.82101476e-01
1.95684150e-01 4.23194557e-01 1.60896391e-01 2.21020117e-01
-8.57690692e-01 3.34747732e-01 1.35921049e+00 -4.62262183e-02
1.84461966e-01 -5.82448959e-01 -4.08630073e-01 8.13069165e-01
6.25189185e-01 -4.20782089e-01 -5.92223294e-02 -2.34994993e-01
7.57550836e-01 3.81125659e-01 2.86224693e-01 -7.29776978e-01
2.33327121e-01 -9.52021703e-02 -9.82445627e-02 -2.91982353e-01
1.89819887e-01 -5.67821383e-01 4.31999147e-01 7.29462981e-01
-9.51858386e-02 7.27202296e-01 -3.92177433e-01 8.43894064e-01
-1.36081457e-01 3.39061528e-01 7.81611204e-01 -1.24880463e-01
-3.85663897e-01 8.17885876e-01 3.21391046e-01 2.78890342e-01
9.77352023e-01 1.09020978e-01 -6.48898780e-01 -4.76998389e-01
-5.52764595e-01 1.50717601e-01 1.01515675e+00 -1.48692518e-01
7.38527000e-01 -1.58127177e+00 -3.90000731e-01 -4.29090410e-02
1.38144284e-01 -2.54518300e-01 2.61654127e-02 1.04171383e+00
-1.12479711e+00 -6.54944479e-02 -1.92783877e-01 -5.21707118e-01
-1.50615036e+00 3.54997575e-01 4.15652916e-02 -4.36705083e-01
-5.21050096e-01 3.85146171e-01 -2.31925994e-02 -1.49392560e-01
4.96627344e-03 -3.60946506e-02 1.83958501e-01 -1.59475282e-01
1.46171167e-01 8.29412818e-01 -6.04326129e-02 -9.05584693e-01
-7.11954117e-01 8.95350575e-01 3.55621189e-01 -2.11984992e-01
1.21490622e+00 5.17723113e-02 -8.66383672e-01 2.63689756e-02
1.70596278e+00 3.31179559e-01 -7.06312835e-01 3.15652601e-02
-4.29299176e-02 -5.18431306e-01 -2.66002804e-01 -1.01662755e-01
-1.14025223e+00 2.79063225e-01 2.98353702e-01 6.07975125e-01
9.35956180e-01 4.72057074e-01 8.55788469e-01 3.52885932e-01
6.92562521e-01 -9.63502586e-01 2.30880111e-01 5.70878506e-01
7.92583644e-01 -7.65401125e-01 3.61729324e-01 -1.02933502e+00
-1.67538494e-01 1.00204754e+00 2.20875472e-01 -6.76156700e-01
1.34529293e+00 2.66704559e-02 -4.15692180e-01 -7.25973785e-01
-4.23634410e-01 -2.64794737e-01 3.45159650e-01 1.82212308e-01
3.83022338e-01 2.54459351e-01 -8.46201777e-01 6.29783645e-02
-4.68082935e-01 -3.56709480e-01 4.47027087e-01 1.02536750e+00
-7.36001074e-01 -1.33560288e+00 2.98417341e-02 6.78329408e-01
-2.60451645e-01 3.91241834e-02 -6.65988266e-01 7.75176346e-01
-2.58358300e-01 6.76657796e-01 1.63544472e-02 -6.22924149e-01
3.20826024e-01 -4.72378433e-01 6.30816400e-01 -4.14331436e-01
-4.39471066e-01 -4.23501581e-01 5.58244102e-02 -9.59714532e-01
-1.63312614e-01 -2.39939824e-01 -1.60770488e+00 -1.02055049e+00
2.66294982e-02 2.26327851e-01 8.17050099e-01 4.13676649e-01
6.72410131e-01 -3.89429778e-02 9.03820932e-01 -3.27469051e-01
-1.48602426e-01 -4.47194993e-01 -8.79235029e-01 6.46349788e-01
2.43653338e-02 -7.39809275e-01 -3.99823844e-01 -4.14608657e-01] | [7.086126804351807, 5.672863483428955] |
02af9966-0e0a-4c93-8d1c-25b50104c920 | no-reference-image-quality-assessment-with-1 | null | null | https://www.mdpi.com/2076-3417/12/1/101 | https://www.mdpi.com/2076-3417/12/1/101 | No-Reference Image Quality Assessment with Convolutional Neural Networks and Decision Fusion | No-reference image quality assessment (NR-IQA) has always been a difficult research problem because digital images may suffer very diverse types of distortions and their contents are extremely various. Moreover, IQA is also a very hot topic in the research community since the number and role of digital images in everyday life is continuously growing. Recently, a huge amount of effort has been devoted to exploiting convolutional neural networks and other deep learning techniques for no-reference image quality assessment. Since deep learning relies on a massive amount of labeled data, utilizing pretrained networks has become very popular in the literature. In this study, we introduce a novel, deep learning-based NR-IQA architecture that relies on the decision fusion of multiple image quality scores coming from different types of convolutional neural networks. The main idea behind this scheme is that a diverse set of different types of networks is able to better characterize authentic image distortions than a single network. The experimental results show that our method can effectively estimate perceptual image quality on four large IQA benchmark databases containing either authentic or artificial distortions. These results are also confirmed in significance and cross database tests. | ['Domonkos Varga'] | 2021-12-23 | null | null | null | applied-sciences-2021-12 | ['no-reference-image-quality-assessment'] | ['computer-vision'] | [ 1.68223590e-01 -6.39524698e-01 1.35499820e-01 -4.39756393e-01
-9.18568373e-01 -2.65821606e-01 4.38984513e-01 -4.87488881e-02
-3.48331630e-01 5.45149684e-01 1.49359889e-02 -5.89640290e-02
-2.56862372e-01 -9.44191217e-01 -5.77528596e-01 -8.87492836e-01
6.43606950e-03 -1.03395417e-01 1.61993176e-01 -4.12640721e-01
3.29376429e-01 4.77450877e-01 -1.71684647e+00 2.17993736e-01
1.10054767e+00 1.56235242e+00 1.15246251e-01 6.35034263e-01
1.42458886e-01 8.33708525e-01 -9.29116070e-01 -7.60308206e-01
4.38517451e-01 -4.94711190e-01 -4.30098683e-01 3.83450687e-02
5.46014190e-01 -4.55530405e-01 -5.00384629e-01 1.45481157e+00
9.82087553e-01 1.30535781e-01 4.88814026e-01 -1.21988368e+00
-9.04086649e-01 3.63028318e-01 -5.10760546e-01 4.30132061e-01
1.89122722e-01 2.88227707e-01 9.25541461e-01 -7.34753311e-01
8.37424845e-02 1.28744984e+00 5.03005564e-01 5.07918335e-02
-9.27747667e-01 -7.58211136e-01 -3.30374479e-01 9.98556852e-01
-1.34434485e+00 -3.91028851e-01 9.44596767e-01 -2.47245222e-01
4.51543033e-01 7.72629008e-02 3.08524519e-01 9.96741414e-01
4.34108078e-01 6.25864387e-01 1.33838880e+00 -2.99796611e-01
2.01483011e-01 1.08757697e-01 -1.45901069e-01 2.93061316e-01
2.43346050e-01 2.39448041e-01 -3.47319871e-01 3.28902662e-01
6.24152303e-01 -1.18040673e-01 -5.61083078e-01 -3.35102528e-01
-1.11505723e+00 6.17321670e-01 5.28856516e-01 5.60325384e-01
-3.55796248e-01 7.62819275e-02 5.74612439e-01 5.25212526e-01
1.85983375e-01 3.87909770e-01 -1.23751529e-01 -1.15274541e-01
-8.93826962e-01 1.13488600e-01 3.41089010e-01 3.84843081e-01
5.17007589e-01 2.83628434e-01 -9.65045542e-02 1.17683566e+00
1.35773988e-02 6.58130407e-01 5.55959821e-01 -9.96878803e-01
3.30899328e-01 5.39376438e-01 1.87003076e-01 -1.65342939e+00
-2.53843576e-01 -6.99278474e-01 -1.41518259e+00 7.50369191e-01
5.38982868e-01 2.28069827e-01 -5.61256289e-01 1.60150337e+00
1.07466383e-02 -3.35398391e-02 1.38833165e-01 1.07634294e+00
7.09613025e-01 7.33531892e-01 -3.23292427e-02 -1.49338022e-01
1.18317962e+00 -6.58797443e-01 -9.24339354e-01 2.54815102e-01
1.92780755e-02 -8.84942353e-01 1.01237404e+00 8.76831949e-01
-1.16339433e+00 -1.27304626e+00 -1.45109725e+00 -1.02620147e-01
-4.62851763e-01 1.49716154e-01 3.41499507e-01 7.59461641e-01
-1.31772566e+00 5.53190768e-01 -2.73087382e-01 -2.16007661e-02
5.62520802e-01 2.95190573e-01 -5.27944088e-01 -4.85326618e-01
-1.52515280e+00 9.04742837e-01 4.61290628e-01 2.55799472e-01
-6.90639198e-01 -3.48442525e-01 -6.75684273e-01 2.48666227e-01
3.25228870e-01 -4.40743059e-01 8.16601992e-01 -1.35002124e+00
-1.35143387e+00 6.97257221e-01 4.06012833e-01 -3.54370713e-01
5.54470003e-01 -1.36223719e-01 -1.02612913e+00 3.05528879e-01
-1.26627445e-01 3.61916631e-01 9.93267715e-01 -1.33205402e+00
-6.44245863e-01 -4.71388549e-01 2.79055715e-01 5.20370826e-02
-3.90602261e-01 2.24386286e-02 -4.55122679e-01 -8.23298633e-01
-3.10671963e-02 -4.97628719e-01 1.88216433e-01 2.32152238e-01
-2.02237397e-01 -2.09567875e-01 6.75496578e-01 -7.44550943e-01
1.15903211e+00 -2.05425000e+00 -5.50921224e-02 8.49536806e-02
2.98589885e-01 7.22302973e-01 -2.94596165e-01 1.73534811e-01
-2.03404039e-01 7.31624961e-02 -2.72405773e-01 2.26422325e-01
1.36438370e-01 -2.50880748e-01 -5.17043173e-02 4.27720934e-01
2.64685094e-01 7.38704085e-01 -7.95745790e-01 -4.71562237e-01
5.79121649e-01 7.17600226e-01 -8.87038931e-02 3.70662719e-01
2.35348642e-01 3.41123670e-01 -1.12810314e-01 6.75069988e-01
1.12457633e+00 -2.57929474e-01 -1.94177210e-01 -6.89454496e-01
-3.15824561e-02 -3.03599000e-01 -1.30313313e+00 1.68725884e+00
-6.05596185e-01 7.99698949e-01 -9.12872031e-02 -1.19851708e+00
9.53175128e-01 2.88596064e-01 3.10961038e-01 -1.45239103e+00
4.54801649e-01 3.71629983e-01 3.19377512e-01 -6.80846989e-01
4.38731700e-01 1.58511326e-02 9.48113762e-03 2.45768189e-01
1.31830782e-01 8.80338624e-02 3.90714198e-01 -1.13966011e-01
9.33110595e-01 -1.78621650e-01 2.82171577e-01 5.27787432e-02
1.07809448e+00 -5.41431665e-01 6.57449245e-01 6.63197279e-01
-6.21269524e-01 6.75680935e-01 3.01086426e-01 -5.20161629e-01
-1.22146153e+00 -1.03647792e+00 -3.86140466e-01 7.51298547e-01
4.96167004e-01 2.31568098e-01 -6.41910791e-01 -3.07591617e-01
-3.12418580e-01 3.50235075e-01 -5.92104077e-01 -3.02493960e-01
-5.18733501e-01 -7.14619935e-01 4.40496385e-01 4.64684933e-01
1.11887610e+00 -1.29757142e+00 -4.00892794e-01 1.44713983e-01
-2.50210047e-01 -1.17771137e+00 -1.36500373e-01 -2.28477851e-01
-5.26430547e-01 -1.19984090e+00 -1.07450783e+00 -5.13478577e-01
1.72104284e-01 4.34382856e-01 1.29354584e+00 2.71401592e-02
-2.41238102e-01 1.90268047e-02 -5.92015088e-01 -1.84499472e-01
-5.98218024e-01 -2.25328490e-01 -8.06716084e-02 4.61091369e-01
3.23400468e-01 -6.23397768e-01 -8.80476773e-01 4.91190523e-01
-1.25399625e+00 -1.99238345e-01 9.99442160e-01 8.19176555e-01
4.77350682e-01 4.83668685e-01 8.78518939e-01 -3.86843532e-01
7.14881659e-01 -4.02921766e-01 -7.82042801e-01 3.23802918e-01
-4.14354920e-01 -5.92705011e-02 8.76920581e-01 -1.50403917e-01
-1.03235269e+00 -6.15639329e-01 -5.03940523e-01 -4.12806422e-01
-2.89836586e-01 3.64823669e-01 -7.42748618e-01 -1.62475511e-01
5.74731231e-01 3.11964035e-01 -1.46645322e-01 -2.57056415e-01
3.20314884e-01 7.36583948e-01 8.87016892e-01 -1.96792215e-01
9.35152888e-01 2.89849281e-01 5.51052466e-02 -5.69515526e-01
-4.38245505e-01 -2.91956097e-01 -4.14262295e-01 -5.23279428e-01
8.17005277e-01 -8.06656837e-01 -8.48765552e-01 1.10213780e+00
-1.04999435e+00 -1.96017735e-02 1.14965923e-02 4.71435726e-01
-2.94910133e-01 6.26671553e-01 -3.47270787e-01 -5.70330501e-01
-4.91835564e-01 -1.55296016e+00 7.06871331e-01 4.15395498e-01
4.31937486e-01 -8.47358525e-01 -7.85421729e-02 3.71366441e-01
7.28605866e-01 3.83267224e-01 7.94954419e-01 -3.35421354e-01
-6.46101594e-01 -3.11181039e-01 -8.17119837e-01 7.74947524e-01
3.34159195e-01 -1.30824581e-01 -1.04099166e+00 -3.56623143e-01
1.69720799e-02 -5.48871636e-01 6.61455333e-01 3.00648451e-01
1.50788569e+00 -2.87022535e-02 3.97209108e-01 6.17312074e-01
1.70092964e+00 3.74501795e-01 1.01688051e+00 4.22068179e-01
5.15342951e-01 3.48116815e-01 4.91978824e-01 3.68096054e-01
1.64555863e-01 9.37379956e-01 7.16608882e-01 -3.33188474e-01
-2.25664452e-01 2.69039810e-01 1.75055966e-01 8.46315503e-01
-1.67406380e-01 -4.66876894e-01 -7.09504545e-01 4.61512059e-01
-1.35809433e+00 -1.15835667e+00 -8.51224817e-04 2.15281367e+00
7.02902138e-01 9.85958353e-02 -1.02703003e-02 7.76984453e-01
8.66120219e-01 1.80946186e-01 -8.18851888e-01 -3.41936588e-01
-4.64761466e-01 2.97638267e-01 2.15252116e-01 -1.61976457e-01
-1.15741694e+00 3.80519420e-01 5.50242233e+00 1.09909987e+00
-1.26082218e+00 -2.93657966e-02 9.59301412e-01 3.49686056e-01
9.52445567e-02 -7.86225975e-01 -1.80249274e-01 7.04836607e-01
1.09193194e+00 -1.47004306e-01 2.72968858e-01 6.78762615e-01
3.07239830e-01 -2.35432133e-01 -7.08365619e-01 1.39205456e+00
2.12624878e-01 -1.10490000e+00 1.90816090e-01 -6.74324483e-02
7.83377349e-01 -3.31196517e-01 5.56079507e-01 9.97025222e-02
9.91502553e-02 -1.09022367e+00 6.13334000e-01 5.71775675e-01
1.03144872e+00 -1.02580225e+00 1.23502278e+00 -1.28386170e-02
-1.06071758e+00 -2.89526463e-01 -6.77327871e-01 1.92298666e-01
3.59668094e-03 8.43001544e-01 1.09847030e-02 9.04724002e-01
9.81423199e-01 6.66643679e-01 -8.00499141e-01 1.35529900e+00
-1.30454376e-01 3.25860947e-01 2.52994746e-01 3.94519538e-01
4.28860541e-03 -2.13576630e-01 1.83385521e-01 9.53546703e-01
4.93958086e-01 8.44874233e-02 -2.65165776e-01 7.91643620e-01
-3.22572619e-01 2.32860923e-01 -3.85689378e-01 1.70455158e-01
1.21521287e-01 1.34850943e+00 -5.72426081e-01 -3.34608763e-01
-5.45457363e-01 9.09610212e-01 -8.75759348e-02 1.81878537e-01
-8.80870640e-01 -7.04398870e-01 6.57209992e-01 -1.87917352e-01
2.42051393e-01 -5.66241853e-02 -2.91150436e-02 -1.09873784e+00
4.65779789e-02 -1.20863473e+00 1.28601670e-01 -9.46511030e-01
-1.38435304e+00 8.46673906e-01 -3.68942261e-01 -1.60349846e+00
-5.92855290e-02 -7.00680077e-01 -4.42052633e-01 9.48940456e-01
-1.92086554e+00 -7.12550879e-01 -8.17165554e-01 8.25794399e-01
6.45411789e-01 -1.60225287e-01 5.47537684e-01 7.40696907e-01
-5.81315637e-01 9.04676914e-01 2.71699488e-01 4.76195604e-01
7.56715775e-01 -1.20375025e+00 1.68416008e-01 1.03479683e+00
-5.53532504e-02 3.77805591e-01 5.28113484e-01 -3.44558507e-02
-1.39566731e+00 -1.03305578e+00 3.01846594e-01 2.97898818e-02
3.58060718e-01 9.90180150e-02 -1.18317068e+00 -2.04615578e-01
5.15625894e-01 2.65327960e-01 5.93424320e-01 -3.85558605e-01
-4.26186591e-01 -7.13705838e-01 -1.19433320e+00 2.67216593e-01
6.02673948e-01 -4.86206084e-01 -2.42917374e-01 -1.42525569e-01
4.62332934e-01 -7.82963187e-02 -9.48511720e-01 6.05156124e-01
4.56161916e-01 -1.58305466e+00 1.11749768e+00 1.49615556e-01
6.42782569e-01 -4.61898983e-01 -2.94231951e-01 -1.31397247e+00
-7.85199329e-02 -2.03330815e-02 5.40836528e-02 1.18260908e+00
-1.61762834e-02 -4.39904511e-01 4.03270632e-01 3.17159057e-01
-1.10096447e-01 -5.97411573e-01 -7.36901462e-01 -6.81467831e-01
4.12208997e-02 -4.55511272e-01 6.67091906e-01 8.28826249e-01
-5.64037085e-01 3.10987346e-02 -6.22783959e-01 1.42723218e-01
8.52395535e-01 2.76800152e-02 7.57355809e-01 -1.20826006e+00
-2.44470388e-01 -6.63804770e-01 -8.75135362e-01 -7.09696651e-01
-1.04425676e-01 -6.01112425e-01 -5.28334230e-02 -1.53020644e+00
8.58303681e-02 -3.38278562e-01 -8.14141035e-01 -9.25580226e-03
-3.39420348e-01 6.81714892e-01 9.60919559e-02 7.13817999e-02
-7.45799065e-01 6.96740627e-01 1.40543902e+00 -4.92867678e-01
1.31483480e-01 -6.58158511e-02 -5.47381639e-01 4.09956187e-01
7.81453729e-01 -9.41239018e-03 -4.83726352e-01 -4.26109880e-01
2.98282474e-01 2.16377303e-01 3.27703983e-01 -1.65442419e+00
1.68297827e-01 8.30165073e-02 7.30311692e-01 -5.46533763e-01
1.94613621e-01 -9.02885795e-01 1.79971591e-01 2.07272828e-01
-3.57022494e-01 1.26716211e-01 8.61981884e-02 5.51379263e-01
-7.03693688e-01 8.05661380e-02 1.24683559e+00 -1.87706828e-01
-1.06661332e+00 3.61482352e-01 4.76750955e-02 -6.11631759e-02
7.88814306e-01 -3.50371510e-01 -1.61217198e-01 -4.87218857e-01
-3.65445703e-01 -1.92315772e-01 3.95709127e-01 4.93943900e-01
8.73751104e-01 -1.45595515e+00 -8.03664863e-01 -6.45007147e-03
2.65638679e-01 -3.89056742e-01 6.56611979e-01 6.00842714e-01
-5.15794694e-01 2.98991024e-01 -7.87630200e-01 -6.36774957e-01
-1.04596758e+00 7.38047063e-01 4.19528484e-01 -2.63144523e-01
-3.90401989e-01 6.37886405e-01 1.69895202e-01 -5.97141683e-03
1.76673040e-01 -1.55279587e-03 -5.92290401e-01 -7.73455109e-03
9.59076941e-01 5.89296699e-01 3.32448155e-01 -1.05268216e+00
7.31149502e-03 6.25555158e-01 2.26438307e-04 1.73562378e-01
1.16977906e+00 -3.00351769e-01 4.21301946e-02 3.10484111e-01
1.45755434e+00 -3.26884091e-01 -1.13704610e+00 -3.86985868e-01
-1.70828551e-01 -8.41778159e-01 2.73463607e-01 -1.01353908e+00
-1.56781065e+00 1.30112112e+00 1.21108294e+00 2.35187948e-01
1.73997688e+00 -4.94224101e-01 8.86471868e-01 9.53517407e-02
5.63340247e-01 -1.01130581e+00 4.35235858e-01 -7.76795223e-02
1.00986052e+00 -1.68238640e+00 -1.31427363e-01 6.25294745e-02
-3.23246926e-01 1.19539225e+00 5.22324502e-01 -6.32710904e-02
2.91885853e-01 -1.40550435e-01 3.52835298e-01 1.17792979e-01
-1.73579320e-01 -1.70936942e-01 3.75081301e-01 6.88026249e-01
2.87524223e-01 -1.95844457e-01 -2.89355844e-01 4.82996106e-01
-4.52434197e-02 1.32899925e-01 5.79768896e-01 4.09433186e-01
-3.48268837e-01 -1.15837431e+00 -5.93842208e-01 4.08451647e-01
-7.69627810e-01 6.40956089e-02 1.81947529e-01 7.33172357e-01
3.03449839e-01 1.30496490e+00 -1.61302537e-01 -5.57118535e-01
3.23757291e-01 -3.91029119e-01 2.56005406e-01 1.35320693e-01
-3.46623808e-01 -2.93537796e-01 -5.01309276e-01 -7.73687363e-01
-7.00367153e-01 -2.56097257e-01 -7.25049436e-01 -4.67323303e-01
-2.22630918e-01 -1.42416671e-01 7.45948613e-01 7.33741701e-01
1.38597995e-01 7.14016378e-01 9.90497530e-01 -7.82546937e-01
-3.37771416e-01 -8.83285046e-01 -6.34003699e-01 7.86506712e-01
3.77986521e-01 -7.88516581e-01 -2.98112750e-01 4.99939993e-02] | [11.809365272521973, -1.8432087898254395] |
fa8888a2-a33e-4685-ae42-2100f4715951 | mask-reference-image-quality-assessment | 2302.1377 | null | https://arxiv.org/abs/2302.13770v2 | https://arxiv.org/pdf/2302.13770v2.pdf | Mask Reference Image Quality Assessment | Understanding semantic information is an essential step in knowing what is being learned in both full-reference (FR) and no-reference (NR) image quality assessment (IQA) methods. However, especially for many severely distorted images, even if there is an undistorted image as a reference (FR-IQA), it is difficult to perceive the lost semantic and texture information of distorted images directly. In this paper, we propose a Mask Reference IQA (MR-IQA) method that masks specific patches of a distorted image and supplements missing patches with the reference image patches. In this way, our model only needs to input the reconstructed image for quality assessment. First, we design a mask generator to select the best candidate patches from reference images and supplement the lost semantic information in distorted images, thus providing more reference for quality assessment; in addition, the different masked patches imply different data augmentations, which favors model training and reduces overfitting. Second, we provide a Mask Reference Network (MRNet): the dedicated modules can prevent disturbances due to masked patches and help eliminate the patch discontinuity in the reconstructed image. Our method achieves state-of-the-art performances on the benchmark KADID-10k, LIVE and CSIQ datasets and has better generalization performance across datasets. The code and results are available in the supplementary material. | ['Anlong Ming', 'Limin Liu', 'Shuai He', 'Pengxiang Xiao'] | 2023-02-27 | null | null | null | null | ['image-quality-assessment'] | ['computer-vision'] | [ 5.48692524e-01 -1.18114553e-01 3.10389232e-02 -3.43648374e-01
-7.63304889e-01 -2.92611241e-01 3.09187174e-01 -2.26826161e-01
-9.09936279e-02 6.28495038e-01 2.51499146e-01 1.25186637e-01
-2.12030858e-01 -9.68428195e-01 -7.44690835e-01 -9.62395012e-01
3.02530348e-01 -1.15860589e-01 3.20329517e-01 -3.92636567e-01
2.56677032e-01 3.54134321e-01 -1.54332829e+00 6.16113067e-01
1.23263562e+00 1.20811522e+00 5.35085499e-01 2.18604639e-01
-6.76228479e-02 6.41001940e-01 -8.55075240e-01 -2.74403781e-01
3.35158408e-01 -4.97402161e-01 -6.56649470e-01 2.35899910e-01
3.94777775e-01 -4.35633630e-01 -5.78369856e-01 1.34590161e+00
6.31432891e-01 9.50631350e-02 4.79651034e-01 -1.07256544e+00
-1.03033829e+00 2.70908386e-01 -7.39850938e-01 3.15922946e-01
2.00998127e-01 2.79827595e-01 4.74102110e-01 -9.35416639e-01
5.23015916e-01 1.34337139e+00 4.37409222e-01 4.82766598e-01
-1.19416189e+00 -6.10977411e-01 1.83253989e-01 5.90016007e-01
-1.51769841e+00 -5.75587273e-01 7.88049102e-01 -1.81424886e-01
5.62404454e-01 3.82622570e-01 4.55457151e-01 8.53143215e-01
2.57856071e-01 6.81319833e-01 1.19245887e+00 -1.55305997e-01
8.88318866e-02 8.25242326e-02 -1.40962213e-01 3.41298759e-01
2.72435974e-02 2.71432281e-01 -3.99141431e-01 2.24394575e-01
6.90935612e-01 1.84683383e-01 -8.19838941e-01 -1.05571866e-01
-1.32475889e+00 3.85326862e-01 7.01306581e-01 2.44336501e-01
-5.40600896e-01 -3.55611563e-01 1.50703460e-01 5.43801963e-01
3.31456244e-01 1.84271783e-01 -2.51300335e-01 3.43307763e-01
-8.62983942e-01 7.08012667e-04 1.45331353e-01 7.01977491e-01
8.25746536e-01 2.17477128e-01 -5.39296985e-01 1.16797650e+00
-7.10079446e-02 7.24765778e-01 5.58182895e-01 -9.21635330e-01
5.70013821e-01 5.98961115e-01 1.84102550e-01 -1.32428813e+00
-1.21709384e-01 -7.42645621e-01 -1.43647730e+00 3.56820524e-01
1.98530316e-01 3.98328930e-01 -1.01811862e+00 1.57680082e+00
1.24960296e-01 2.10788727e-01 2.83588111e-01 1.20987976e+00
1.12547636e+00 8.27235103e-01 -3.44922721e-01 -2.75837451e-01
1.13382888e+00 -9.57789421e-01 -7.98503578e-01 -1.47230431e-01
1.37543783e-01 -8.15824151e-01 1.28971624e+00 6.41519010e-01
-9.63418484e-01 -9.94904041e-01 -1.20758533e+00 4.21061851e-02
-2.59968609e-01 2.03158289e-01 2.78131030e-02 3.63731772e-01
-1.04150331e+00 5.72867393e-01 -3.17608088e-01 2.12434083e-02
6.15781665e-01 5.57148457e-02 -5.23389459e-01 -7.09583759e-01
-1.32037461e+00 7.80350566e-01 2.52058685e-01 1.59029976e-01
-1.12030435e+00 -7.21805930e-01 -7.49708116e-01 2.64913333e-03
4.19147760e-01 -2.90475905e-01 7.23118603e-01 -1.32944739e+00
-1.12559068e+00 6.87447011e-01 -3.55660580e-02 -1.88293099e-01
4.10540730e-01 1.53009817e-01 -7.60223150e-01 2.27528155e-01
-5.49295312e-03 8.04534197e-01 1.01001978e+00 -1.53993440e+00
-4.19766247e-01 -3.93287688e-01 1.10696048e-01 2.91293174e-01
-3.01202200e-02 -2.66886294e-01 -6.51823103e-01 -1.13270116e+00
3.75790238e-01 -3.38584244e-01 5.82259558e-02 1.51223361e-01
-3.44929039e-01 2.37429947e-01 8.30311894e-01 -1.05802178e+00
1.14331329e+00 -2.49588490e+00 -7.43880449e-03 1.93130970e-01
8.42100829e-02 3.48983705e-01 -6.45300210e-01 -2.10886523e-02
-2.17885971e-01 4.24656756e-02 -5.32274246e-01 5.25055006e-02
-2.80006170e-01 3.29068512e-01 -1.89429790e-01 3.28881800e-01
1.96319714e-01 8.07277441e-01 -7.26956129e-01 -2.84298301e-01
2.26169497e-01 6.83818877e-01 -2.74185449e-01 2.18613282e-01
7.92214721e-02 7.41258681e-01 -1.23525858e-01 6.85482800e-01
1.24692154e+00 -1.42049193e-01 -2.74563611e-01 -6.98155820e-01
2.03567684e-01 5.52306511e-02 -1.49139249e+00 1.86627805e+00
-4.67234939e-01 2.30321333e-01 1.97406635e-01 -1.07839394e+00
1.05707085e+00 8.62051174e-02 2.59798467e-01 -1.53522217e+00
-1.94818631e-01 8.92097056e-02 -1.08996652e-01 -6.13472462e-01
2.89102197e-01 -3.55282761e-02 2.77267873e-01 1.90661177e-01
-1.16786174e-01 1.26434445e-01 -8.04155413e-03 1.24111742e-01
7.75761366e-01 -8.98247361e-02 4.27817414e-03 -1.55840278e-01
8.08480918e-01 -2.29517534e-01 8.92163455e-01 5.54432511e-01
-1.56281456e-01 1.13912380e+00 2.07377940e-01 -3.40303987e-01
-9.67114687e-01 -1.22077179e+00 -2.98080146e-01 6.15616143e-01
7.30020046e-01 -2.75111031e-02 -6.61658645e-01 -6.22991979e-01
-2.49461189e-01 5.50075412e-01 -6.03475332e-01 -4.64744121e-01
-4.07956362e-01 -9.37806785e-01 2.70958096e-01 2.81585693e-01
9.38831329e-01 -1.16267526e+00 -1.65822990e-02 8.31634402e-02
-5.45671344e-01 -6.78852916e-01 -7.25422859e-01 -2.04376265e-01
-7.01347947e-01 -1.20716190e+00 -9.18560028e-01 -6.64322019e-01
9.12910402e-01 7.56780982e-01 1.18503129e+00 3.55421215e-01
-1.20232716e-01 -1.55374691e-01 -4.70116317e-01 2.48786211e-02
-3.51472259e-01 -4.85566825e-01 -2.39235491e-01 2.93779224e-01
-8.34169462e-02 -4.44142401e-01 -9.66979325e-01 6.07161105e-01
-1.29538941e+00 1.88229889e-01 8.18306565e-01 1.08595920e+00
1.07718873e+00 6.13975048e-01 6.63192511e-01 -6.26143456e-01
4.13099736e-01 -4.32801872e-01 -3.90012085e-01 3.39053750e-01
-7.15470970e-01 -1.13468781e-01 8.06680381e-01 -4.05895889e-01
-1.08221304e+00 -2.34042987e-01 -2.72422642e-01 -5.17356873e-01
-2.09667340e-01 1.69978842e-01 -8.19500268e-01 -6.43024594e-02
5.90359271e-01 6.62172854e-01 4.76103574e-02 -7.65909016e-01
2.17414737e-01 6.31830096e-01 8.22344542e-01 -2.53552198e-01
7.17712641e-01 5.52456558e-01 -2.12879673e-01 -4.67797697e-01
-7.52532721e-01 -2.84390926e-01 -3.35486740e-01 -8.69547427e-02
5.53822517e-01 -1.01313591e+00 -2.16861099e-01 7.37517178e-01
-8.93071055e-01 -2.44981900e-01 -3.50301355e-01 1.58637986e-01
-2.57273018e-01 4.08021092e-01 -3.56160939e-01 -3.66646826e-01
-4.16191936e-01 -1.37193990e+00 9.60229397e-01 4.28257078e-01
4.28625494e-01 -5.69458604e-01 -3.88505250e-01 3.23727399e-01
6.20350361e-01 7.58989230e-02 8.83101285e-01 -1.48079708e-01
-6.06681585e-01 8.55557621e-02 -5.21865726e-01 8.35749745e-01
4.49397892e-01 -4.70580608e-01 -9.48567569e-01 -4.51614678e-01
1.62423432e-01 -2.72092689e-02 1.01636016e+00 3.72778386e-01
1.51041210e+00 -6.04759514e-01 2.69944835e-02 8.26901674e-01
1.58621442e+00 3.63062024e-01 1.25054479e+00 3.41009170e-01
6.48167968e-01 5.83795667e-01 7.71707058e-01 1.02649033e-01
3.36502075e-01 8.30604315e-01 4.60979819e-01 -4.78007227e-01
-6.02300882e-01 -2.24878088e-01 3.32980335e-01 6.95546687e-01
2.77650088e-01 -3.20488721e-01 -4.74873304e-01 5.70913613e-01
-1.44034433e+00 -9.48365510e-01 -1.27814069e-01 2.37050462e+00
9.01095331e-01 -1.57393981e-02 -1.92689255e-01 5.10268271e-01
9.05554235e-01 1.87664434e-01 -8.12923074e-01 5.76820597e-02
-6.46706820e-01 1.66914657e-01 3.08481842e-01 3.96405280e-01
-9.25953150e-01 4.46148634e-01 5.66035604e+00 1.32951784e+00
-1.08504939e+00 2.75801420e-01 1.10940087e+00 -1.16437614e-01
-4.45792198e-01 -2.69763470e-01 -2.73420542e-01 8.35111916e-01
5.08146584e-01 4.53582071e-02 7.61953533e-01 5.00477314e-01
4.35410738e-01 -2.67741442e-01 -7.01535940e-01 1.28879118e+00
1.93682298e-01 -1.18826830e+00 3.47124130e-01 -2.52901167e-01
8.35509121e-01 -2.15731680e-01 1.49084494e-01 7.91366026e-02
-1.42528966e-01 -1.15873981e+00 7.18021750e-01 8.37735891e-01
1.12346423e+00 -8.45442891e-01 1.01391923e+00 3.35643440e-01
-1.00804412e+00 -6.92429468e-02 -7.07095981e-01 2.83346802e-01
-8.83160904e-02 1.04978573e+00 -1.15288176e-01 9.64330375e-01
9.09666181e-01 8.07857513e-01 -9.06668186e-01 1.07826328e+00
-1.16676092e-01 3.65155131e-01 8.55481699e-02 8.39763045e-01
-2.26158351e-01 -2.42215872e-01 3.73160273e-01 7.95092642e-01
4.24384892e-01 2.56381869e-01 4.24876809e-02 8.61573577e-01
-1.41649872e-01 3.11610717e-02 -3.42794567e-01 4.66867238e-01
4.26578552e-01 1.01490140e+00 -5.04916966e-01 -3.91877800e-01
-4.21362966e-01 1.13050818e+00 -2.29005814e-01 4.28189367e-01
-6.46973193e-01 -4.40788776e-01 5.36898613e-01 2.03186437e-01
2.11046323e-01 4.32090580e-01 -3.37175220e-01 -1.32081497e+00
3.93490970e-01 -1.24508035e+00 3.94014090e-01 -1.05236113e+00
-1.45826602e+00 6.62121713e-01 -2.57773548e-01 -1.50957322e+00
3.53384048e-01 -8.50611776e-02 -5.20390034e-01 1.18150151e+00
-1.79928470e+00 -7.75303841e-01 -8.24855983e-01 6.98131323e-01
5.36080837e-01 -2.80071758e-02 4.14546102e-01 7.45074332e-01
-5.23017168e-01 6.69550598e-01 2.49626100e-01 6.03967607e-02
8.25458229e-01 -9.39924836e-01 2.10097030e-01 9.93597329e-01
-4.28167768e-02 2.11973637e-01 4.29423183e-01 -5.78674853e-01
-1.20066237e+00 -1.37317514e+00 2.36106217e-01 -1.13042509e-02
-9.13484395e-02 3.43373790e-02 -1.45508373e+00 -1.53847353e-03
3.49220186e-02 2.71512568e-01 1.50314867e-01 -6.08304083e-01
-4.29082334e-01 -7.68127084e-01 -1.42797863e+00 2.13433757e-01
1.01603413e+00 -4.45824593e-01 -3.72161150e-01 1.21098876e-01
8.01688969e-01 -3.61244887e-01 -7.46641755e-01 6.63312733e-01
2.69067675e-01 -1.23371136e+00 1.17350006e+00 7.78851658e-02
3.19465309e-01 -9.12533462e-01 -2.99174368e-01 -1.48192525e+00
-3.93684119e-01 -5.11545949e-02 8.84263366e-02 1.35080457e+00
1.39581800e-01 -5.53003192e-01 4.05166894e-01 -1.72316022e-02
-2.27674097e-01 -8.13970029e-01 -7.37197101e-01 -8.52920651e-01
-4.93628979e-02 -9.82696414e-02 1.02783060e+00 1.01247263e+00
-7.11906731e-01 -1.93730310e-01 -4.51604247e-01 2.94425696e-01
7.04603851e-01 1.33558407e-01 4.56717461e-01 -9.36674833e-01
-2.77381301e-01 -3.84957612e-01 -3.41289073e-01 -1.05119920e+00
-2.40152851e-01 -8.28196645e-01 2.96638510e-03 -1.66904747e+00
3.06686282e-01 -6.10346675e-01 -5.74736297e-01 4.68747795e-01
-3.84271741e-01 6.56043828e-01 6.63078651e-02 3.87619019e-01
-5.85946739e-01 7.51255691e-01 1.74288034e+00 -5.07749677e-01
-9.94662056e-04 -1.82300568e-01 -9.76081252e-01 2.88033068e-01
7.07697451e-01 -4.69566226e-01 -5.19129932e-01 -5.00421703e-01
-6.82470649e-02 1.03413448e-01 5.86864293e-01 -1.05648327e+00
-1.08698353e-01 -2.04185426e-01 8.22110832e-01 -6.95239961e-01
4.28157002e-02 -6.94369614e-01 4.87084627e-01 3.42398673e-01
-2.25801855e-01 -8.14874992e-02 7.08660111e-02 5.20962059e-01
-6.37030065e-01 -8.66047069e-02 1.23870599e+00 -2.24417314e-01
-7.75100827e-01 3.51868451e-01 2.26967007e-01 -6.20921794e-03
7.18194008e-01 -3.75426441e-01 -5.69467068e-01 -3.18020731e-01
-6.41921997e-01 2.05442727e-01 7.97392786e-01 5.43738365e-01
9.89251375e-01 -1.55134904e+00 -9.20873940e-01 4.91624683e-01
1.44753724e-01 2.42420390e-01 9.28538203e-01 7.72155821e-01
-3.34653229e-01 9.87380967e-02 -4.18931067e-01 -5.79732835e-01
-1.04758799e+00 8.44393671e-01 4.78576481e-01 -7.56800026e-02
-7.99353302e-01 5.36933780e-01 6.39243126e-01 -2.43304521e-01
2.48064846e-01 6.01913687e-03 -3.91201198e-01 -1.67604417e-01
1.09144986e+00 3.69969130e-01 4.06841248e-01 -8.58088970e-01
-2.18027085e-01 5.53412497e-01 4.69545983e-02 1.79764703e-01
1.21457911e+00 -5.12064874e-01 -1.63482353e-01 1.41458467e-01
1.14606678e+00 -1.48985982e-02 -1.39097548e+00 -4.12986130e-01
-4.53290403e-01 -8.84240985e-01 2.22458795e-01 -1.19321597e+00
-1.62171793e+00 8.19051445e-01 1.09597230e+00 3.05971950e-02
1.85595095e+00 -1.67396635e-01 7.60664046e-01 -3.44003807e-03
3.34828764e-01 -8.15502822e-01 1.83432922e-01 6.94329515e-02
1.28784680e+00 -1.18495536e+00 -6.22892939e-02 -3.00781518e-01
-6.67593956e-01 7.41606355e-01 7.15279341e-01 3.21974605e-02
4.63888556e-01 3.35170585e-03 2.66061395e-01 2.81810854e-03
-5.77728808e-01 -1.78397633e-02 3.99975747e-01 9.16078329e-01
-1.18196420e-01 -1.32362127e-01 -1.95310771e-01 7.94269621e-01
-5.06480969e-03 -1.64495826e-01 5.54571092e-01 4.08390462e-01
-3.38915437e-01 -9.75751162e-01 -7.65594780e-01 5.43093562e-01
-3.12306046e-01 -1.89976558e-01 -5.85567951e-02 2.76171237e-01
5.44455826e-01 1.18432724e+00 4.13455442e-02 -5.03300548e-01
6.08731329e-01 -3.38155508e-01 3.13831598e-01 -4.04009223e-01
-2.78989822e-01 1.05494939e-01 -3.49543363e-01 -7.44477570e-01
-4.33025897e-01 -2.02354297e-01 -1.15670860e+00 -3.09732884e-01
-2.03966409e-01 2.61286274e-03 3.91987830e-01 7.27360368e-01
4.98456955e-01 8.32462430e-01 8.80365789e-01 -6.79174900e-01
-1.92419291e-01 -9.70509827e-01 -7.16200471e-01 7.07218170e-01
3.90218258e-01 -7.31338739e-01 -3.42799932e-01 1.89638004e-01] | [11.743398666381836, -1.8976538181304932] |
0a038e35-efc7-4399-89aa-7661f18e152f | balanced-mixture-of-supernets-for-learning | 2306.11982 | null | https://arxiv.org/abs/2306.11982v1 | https://arxiv.org/pdf/2306.11982v1.pdf | Balanced Mixture of SuperNets for Learning the CNN Pooling Architecture | Downsampling layers, including pooling and strided convolutions, are crucial components of the convolutional neural network architecture that determine both the granularity/scale of image feature analysis as well as the receptive field size of a given layer. To fully understand this problem, we analyse the performance of models independently trained with each pooling configurations on CIFAR10, using a ResNet20 network, and show that the position of the downsampling layers can highly influence the performance of a network and predefined downsampling configurations are not optimal. Network Architecture Search (NAS) might be used to optimize downsampling configurations as an hyperparameter. However, we find that common one-shot NAS based on a single SuperNet does not work for this problem. We argue that this is because a SuperNet trained for finding the optimal pooling configuration fully shares its parameters among all pooling configurations. This makes its training hard, because learning some configurations can harm the performance of others. Therefore, we propose a balanced mixture of SuperNets that automatically associates pooling configurations to different weight models and helps to reduce the weight-sharing and inter-influence of pooling configurations on the SuperNet parameters. We evaluate our proposed approach on CIFAR10, CIFAR100, as well as Food101 and show that in all cases, our model outperforms other approaches and improves over the default pooling configurations. | ['Marco Pedersoli', 'Matthew Toews', 'Mehraveh Javan'] | 2023-06-21 | null | null | null | null | ['architecture-search'] | ['methodology'] | [-1.14772163e-01 -2.96176314e-01 1.43299818e-01 -2.61195719e-01
1.63954988e-01 -5.77057123e-01 3.37423116e-01 -3.53592681e-04
-8.56957376e-01 4.97634441e-01 6.45440221e-02 -1.79979518e-01
-3.47305655e-01 -8.50070715e-01 -9.03236926e-01 -8.42472613e-01
2.49222964e-02 2.23532710e-02 7.91149735e-01 -2.61596859e-01
3.61946315e-01 6.52471602e-01 -1.51472163e+00 4.71633852e-01
3.60670507e-01 1.11104035e+00 5.33111334e-01 5.31574428e-01
-9.17661116e-02 5.55683076e-01 -9.32096660e-01 -2.26334706e-01
5.80121160e-01 -3.38101313e-02 -7.63553619e-01 -3.08489025e-01
2.77598441e-01 -1.01633146e-01 8.73931777e-03 1.13350523e+00
5.66239834e-01 1.76719353e-01 3.29921067e-01 -8.22344542e-01
-2.61674106e-01 1.27611315e+00 -2.94873238e-01 6.91815615e-01
-7.57331252e-01 1.13663636e-01 8.43657851e-01 -6.82700336e-01
4.60856289e-01 1.01570380e+00 5.83258629e-01 3.39912027e-01
-1.21983731e+00 -7.73083031e-01 3.45193595e-01 5.96442595e-02
-1.44685781e+00 -4.09793973e-01 4.57492560e-01 -2.89843500e-01
1.24433732e+00 3.20294835e-02 7.34471321e-01 7.87465036e-01
1.85614631e-01 2.49723166e-01 7.11203814e-01 -3.95839334e-01
5.46726286e-01 2.28996456e-01 2.70272672e-01 4.55717087e-01
5.38512826e-01 -3.30135196e-01 -3.83775800e-01 3.88405770e-02
8.78956378e-01 1.36076137e-01 -2.92286634e-01 -3.03909481e-01
-1.00046575e+00 6.47494912e-01 8.99330735e-01 5.78631282e-01
-3.64355564e-01 1.90508366e-01 2.90199459e-01 4.31575626e-01
-7.26322234e-02 1.09636343e+00 -7.99964070e-01 1.80451334e-01
-7.76429832e-01 2.27725103e-01 5.28259814e-01 5.22710562e-01
8.65872681e-01 -2.05525249e-01 -4.23291534e-01 8.68341982e-01
-7.59272650e-02 -1.90803006e-01 8.18990946e-01 -5.06518364e-01
6.80013120e-01 9.68261003e-01 -1.13820761e-01 -6.85296059e-01
-7.59904802e-01 -7.31335640e-01 -6.90416276e-01 2.88884133e-01
5.63779652e-01 -5.03097653e-01 -8.50542843e-01 1.82208991e+00
-6.78614825e-02 2.42085740e-01 8.47882777e-02 8.19954336e-01
6.29846573e-01 3.21957469e-01 3.59803677e-01 2.95437127e-01
1.50512147e+00 -9.92139101e-01 -1.22160345e-01 -6.10092819e-01
6.62180603e-01 -6.92722082e-01 9.62004364e-01 1.27688646e-01
-1.07919621e+00 -7.01897204e-01 -1.44625700e+00 2.89436966e-01
-6.59867823e-01 4.69144583e-01 4.79582876e-01 5.01684606e-01
-1.01253545e+00 1.10920036e+00 -6.83097780e-01 -2.55686492e-01
5.87079763e-01 6.80478871e-01 -2.41189480e-01 1.59783646e-01
-1.25372231e+00 1.00041306e+00 9.08799708e-01 9.40542147e-02
-4.67598557e-01 -8.71479690e-01 -3.77017438e-01 8.28121364e-01
1.73770875e-01 -5.64828277e-01 9.40202296e-01 -9.10465360e-01
-1.31793332e+00 3.48340392e-01 2.56869763e-01 -7.97624826e-01
2.08535552e-01 6.51379600e-02 -1.04728147e-01 -2.48802215e-01
-5.64772844e-01 8.96816313e-01 9.18540776e-01 -6.64655745e-01
-6.03414357e-01 -2.78113335e-01 3.05645585e-01 1.95315689e-01
-5.76431036e-01 8.24671909e-02 -2.61004537e-01 -4.76751536e-01
2.04810396e-01 -9.38042104e-01 -3.80662262e-01 -5.71434200e-01
-4.24396008e-01 -1.49315625e-01 3.68700027e-01 -9.60633457e-02
1.40514421e+00 -2.30796862e+00 4.20969613e-02 3.46637607e-01
-7.75495172e-02 5.43983817e-01 -1.49920851e-01 -9.02281329e-02
-3.17554414e-01 3.98414105e-01 2.79659390e-01 -2.88390189e-01
-2.26035178e-01 -5.96155338e-02 -5.28391749e-02 1.31593987e-01
3.26839894e-01 7.08779156e-01 -3.41127276e-01 -6.54609874e-02
2.07867980e-01 5.82414210e-01 -8.05227757e-01 -1.00471400e-01
-5.61052337e-02 -2.38011219e-02 -3.05676639e-01 6.42258450e-02
8.29799473e-01 -4.52019006e-01 1.80166587e-01 -3.59850019e-01
-2.93582380e-01 2.81907499e-01 -1.47910690e+00 1.29241788e+00
-3.32001388e-01 4.38128084e-01 -4.62798566e-01 -7.95045316e-01
9.64232683e-01 7.72950053e-02 -1.60968721e-01 -5.56847453e-01
4.39586431e-01 1.23266280e-01 4.94773686e-01 -2.14684740e-01
3.60363901e-01 1.36033446e-01 2.01403156e-01 3.47812325e-01
3.23331922e-01 4.39535141e-01 3.03892612e-01 -2.69384682e-01
1.14535880e+00 -2.85972625e-01 2.86859423e-01 -4.15734112e-01
4.14045334e-01 -4.63772207e-01 5.29442489e-01 1.02942991e+00
1.13460511e-01 7.92093873e-01 9.07325566e-01 -6.72216535e-01
-1.14527726e+00 -5.74624002e-01 -9.13699418e-02 1.34793079e+00
-1.09234847e-01 -2.73679614e-01 -1.01283693e+00 -4.67975229e-01
-3.44222069e-01 5.13774812e-01 -9.64858413e-01 -3.04570824e-01
-6.85794234e-01 -1.19778693e+00 4.16497260e-01 5.56111932e-01
7.73976862e-01 -1.36805332e+00 -1.20062709e+00 1.89351991e-01
2.40567401e-01 -1.09631717e+00 -2.75220782e-01 7.16039777e-01
-8.51757050e-01 -1.06223655e+00 -5.92307627e-01 -7.29839563e-01
8.97240639e-01 1.95598856e-01 9.90847468e-01 3.92147452e-01
-7.97157884e-02 -6.21406198e-01 -2.29156420e-01 -5.11834800e-01
-1.84611499e-01 8.15648735e-01 -1.85278520e-01 -4.33753664e-03
3.74676079e-01 -6.01365030e-01 -8.79800677e-01 5.68768978e-01
-8.90547812e-01 5.94977476e-02 7.88914382e-01 4.22346115e-01
4.79811341e-01 3.02020848e-01 4.78471041e-01 -8.62108409e-01
6.60508335e-01 -3.50813955e-01 -8.04397702e-01 4.75521743e-01
-3.05326670e-01 5.83568573e-01 8.37745428e-01 -6.50667369e-01
-7.09531724e-01 6.97638988e-02 6.74541518e-02 -5.01001000e-01
-9.92184803e-02 2.50786930e-01 -8.86506811e-02 -2.35497832e-01
8.58103454e-01 -1.58294871e-01 -3.43068212e-01 -7.08655536e-01
-8.92498642e-02 1.83065206e-01 -7.45565742e-02 -1.27187476e-01
4.73397523e-01 2.67930001e-01 -1.55677423e-01 -5.18754840e-01
-6.83997512e-01 -1.59124210e-01 -5.17794192e-01 3.06344688e-01
7.95428574e-01 -7.35189497e-01 -6.03837192e-01 4.66914654e-01
-1.19217265e+00 -5.01423538e-01 -2.61246771e-01 4.22876030e-01
7.28288665e-02 -4.27941352e-01 -4.07738686e-01 -2.59395033e-01
-2.55175650e-01 -1.49152291e+00 3.74080360e-01 5.31290710e-01
-4.77979369e-02 -6.45570338e-01 -1.63571730e-01 -3.08670610e-01
8.74094546e-01 -2.78239250e-01 1.05416107e+00 -8.44681621e-01
-4.40239340e-01 1.94721706e-02 -3.61224294e-01 4.94920135e-01
-1.64988518e-01 -1.61554024e-01 -1.11691678e+00 -9.70059037e-02
-8.47484320e-02 2.14824509e-02 1.27270091e+00 6.11245275e-01
1.51203203e+00 -3.16952735e-01 -2.07004800e-01 7.89047837e-01
1.57126594e+00 -1.98835358e-02 9.11621392e-01 8.38917494e-01
3.46333385e-01 4.17614192e-01 -1.22120880e-01 3.80378366e-01
-7.58963674e-02 4.52408552e-01 5.52519977e-01 6.57312497e-02
-9.70694572e-02 -2.77366899e-02 1.14917740e-01 2.43149698e-02
-2.68689871e-01 -1.69247106e-01 -6.22271359e-01 2.86595464e-01
-1.60487282e+00 -7.98306823e-01 5.22836328e-01 2.22513294e+00
5.99953651e-01 5.69535196e-01 -1.25065327e-01 1.90236285e-01
7.57704675e-01 1.21851444e-01 -4.90602583e-01 -3.83236259e-01
-2.61620373e-01 4.54073697e-01 8.94335568e-01 1.85714945e-01
-1.04816282e+00 1.05243886e+00 5.98447990e+00 6.77742183e-01
-1.38337469e+00 -3.50154564e-02 7.24613190e-01 -2.95235544e-01
6.17391057e-02 -1.35532528e-01 -1.16591644e+00 5.28347850e-01
8.21277440e-01 2.19958633e-01 6.55301094e-01 8.97302866e-01
1.54736131e-01 -2.41754577e-02 -9.98270035e-01 7.26789296e-01
-2.05413952e-01 -1.68284309e+00 1.23363584e-01 -2.75109950e-02
7.75351822e-01 4.28780884e-01 2.40899939e-02 2.54935920e-01
1.45100266e-01 -1.08922744e+00 7.75635719e-01 2.85781235e-01
8.14743415e-02 -8.07113886e-01 9.69393492e-01 1.67630613e-01
-9.90758896e-01 -5.22714972e-01 -8.51188481e-01 -9.10396278e-02
-3.84085983e-01 4.44432497e-01 -1.02043533e+00 -1.02123290e-01
1.03883719e+00 -3.04029193e-02 -1.02678943e+00 1.29361403e+00
-3.96692932e-01 4.18010592e-01 -3.23768765e-01 -2.11908773e-01
2.45277584e-01 2.91657358e-01 1.40913263e-01 1.08438826e+00
4.65367973e-01 -2.16395587e-01 -3.02033573e-01 8.39680493e-01
-3.56082499e-01 7.78655484e-02 4.61708149e-03 2.27315575e-01
7.06083477e-01 1.28162742e+00 -1.12402964e+00 -1.45712644e-01
-1.48692906e-01 4.25427884e-01 5.65189898e-01 5.01643836e-01
-7.59611785e-01 -4.24160659e-01 8.69334698e-01 2.24212766e-01
9.28837478e-01 2.04427186e-02 -4.35194194e-01 -9.21325445e-01
-9.71815735e-02 -5.59179127e-01 4.43980217e-01 -5.34754455e-01
-7.10526049e-01 9.86973464e-01 -1.09526955e-01 -1.02674901e+00
8.04602951e-02 -7.63893902e-01 -7.96653688e-01 8.19056213e-01
-1.74389172e+00 -5.99438071e-01 -2.54321754e-01 4.43053693e-01
3.75021875e-01 -1.43242404e-01 6.80552185e-01 8.17035809e-02
-8.75232339e-01 7.27252901e-01 -2.73383915e-01 2.00964183e-01
4.90453511e-01 -8.78251970e-01 5.72582006e-01 9.19138968e-01
6.86745793e-02 7.85970271e-01 7.68845439e-01 -1.28488734e-01
-8.50072980e-01 -1.17630458e+00 7.84418106e-01 5.53379692e-02
4.20356750e-01 -3.74757916e-01 -1.07530749e+00 4.77354705e-01
2.13674139e-02 2.05701068e-01 4.73344147e-01 1.27589613e-01
-2.71763593e-01 -3.20095748e-01 -1.10496676e+00 7.13961124e-01
9.01775956e-01 2.42856652e-01 -4.78530884e-01 9.51724313e-03
7.31253624e-01 -2.30501607e-01 -4.99171883e-01 3.06704998e-01
3.53831112e-01 -1.22684741e+00 9.82599378e-01 -5.92701137e-01
2.25650385e-01 -2.32175663e-01 2.75180154e-02 -1.53645146e+00
-5.62752366e-01 -2.45580226e-01 2.24308938e-01 8.58806789e-01
8.00268948e-01 -9.18284655e-01 6.83699429e-01 5.97761035e-01
1.09661082e-02 -8.13924432e-01 -7.16619372e-01 -5.83090544e-01
6.44799136e-03 -3.07806492e-01 1.25838494e+00 4.49645936e-01
-3.78734589e-01 8.85466784e-02 7.83692002e-02 4.19427544e-01
2.72494555e-01 -2.49168739e-01 3.78880620e-01 -1.26911938e+00
-3.76161933e-01 -8.78803194e-01 -4.60762382e-01 -5.98045170e-01
-3.02663241e-02 -5.87906003e-01 -1.64988235e-01 -9.95345592e-01
6.17238246e-02 -8.85669827e-01 -8.78221452e-01 7.68035769e-01
-1.30122587e-01 3.17770183e-01 2.85319895e-01 7.46327341e-02
-3.90920877e-01 -2.93062925e-02 1.00852406e+00 1.27337843e-01
-4.37909961e-01 6.02256432e-02 -9.26141977e-01 8.17290545e-01
1.07480645e+00 -5.13904095e-01 -4.78074819e-01 -7.18974054e-01
7.04308510e-01 -6.57117367e-01 3.96452904e-01 -1.43433797e+00
4.41293687e-01 3.67925763e-02 7.12279201e-01 -7.97904842e-03
-1.22894911e-04 -6.38782024e-01 1.55273378e-02 4.83835906e-01
-5.01311541e-01 9.89067703e-02 4.43080992e-01 4.85336185e-02
1.04232647e-01 -7.82947779e-01 9.66258109e-01 -4.21013951e-01
-8.04806232e-01 2.75659561e-01 1.32015878e-02 -9.12369490e-02
7.09764957e-01 -1.67483851e-01 -3.63784313e-01 3.99400175e-01
-6.88290358e-01 1.39249563e-01 2.81806558e-01 5.29376566e-01
2.62693465e-01 -8.53256583e-01 -4.96770471e-01 5.83726346e-01
-1.11098573e-01 4.52588722e-02 4.09205079e-01 6.11221790e-01
-4.48580384e-01 1.96861401e-01 -4.41129178e-01 -2.27277815e-01
-1.15307367e+00 4.24717396e-01 7.43736982e-01 -3.99129987e-01
-5.02408385e-01 1.12463832e+00 8.22465792e-02 -2.36560218e-02
3.72973680e-01 -6.82997584e-01 -6.55365884e-01 6.39981255e-02
9.13952291e-01 9.89692509e-02 5.94811857e-01 -1.25108510e-01
-3.32480192e-01 4.98607785e-01 -3.08256000e-01 1.75909191e-01
1.56263423e+00 1.37948975e-01 -3.58407423e-02 1.15610808e-01
1.03620708e+00 -5.20960510e-01 -1.42279148e+00 -2.36897007e-01
-1.22485168e-01 -1.78776339e-01 1.75055131e-01 -6.70013845e-01
-1.27302873e+00 7.96161532e-01 7.06754982e-01 2.21792519e-01
1.07521307e+00 -1.74129065e-02 3.41330767e-01 6.15710974e-01
1.80055946e-01 -1.16593802e+00 -6.53936267e-02 6.08946145e-01
7.51420319e-01 -1.01099408e+00 -1.71318114e-01 -1.91166829e-02
-5.08239686e-01 1.24997699e+00 8.55087519e-01 -4.69223410e-01
7.08274961e-01 3.42449009e-01 -1.85974300e-01 -9.91434529e-02
-8.54262710e-01 5.92111461e-02 6.83932155e-02 -2.18952410e-02
2.18443841e-01 -8.67476240e-02 7.14313686e-02 6.44652367e-01
-5.15601337e-01 6.38210624e-02 3.03719103e-01 7.27356315e-01
-6.24758601e-01 -9.94162977e-01 -2.05029279e-01 4.19489115e-01
-5.04758239e-01 -2.52434403e-01 8.29134360e-02 5.72469950e-01
6.82695866e-01 4.31524396e-01 3.70742649e-01 -4.02191639e-01
4.45890009e-01 2.59636253e-01 4.82222646e-01 -5.77822924e-01
-1.00884509e+00 -2.75470018e-01 -3.01564395e-01 -3.19489449e-01
-1.51108086e-01 -2.99763113e-01 -1.12331951e+00 -1.90400377e-01
-3.13369483e-01 1.97470821e-02 8.16151738e-01 1.02563393e+00
4.93455648e-01 7.85012960e-01 4.59257126e-01 -9.39277351e-01
-7.52935588e-01 -1.11517453e+00 -4.39012170e-01 1.64667174e-01
2.18746066e-01 -6.74019277e-01 -5.53827405e-01 -3.97756249e-01] | [8.729609489440918, 2.8569869995117188] |
4146a025-c8e8-4b1c-aeb2-d50ccd154557 | object-detection-with-deep-reinforcement | 2208.04511 | null | https://arxiv.org/abs/2208.04511v1 | https://arxiv.org/pdf/2208.04511v1.pdf | Object Detection with Deep Reinforcement Learning | Object localization has been a crucial task in computer vision field. Methods of localizing objects in an image have been proposed based on the features of the attended pixels. Recently researchers have proposed methods to formulate object localization as a dynamic decision process, which can be solved by a reinforcement learning approach. In this project, we implement a novel active object localization algorithm based on deep reinforcement learning. We compare two different action settings for this MDP: a hierarchical method and a dynamic method. We further perform some ablation studies on the performance of the models by investigating different hyperparameters and various architecture changes. | ['Ruofeng Li', 'Manoosh Samiei'] | 2022-08-09 | null | null | null | null | ['active-object-localization'] | ['computer-vision'] | [-6.09303731e-03 -5.38332909e-02 -1.17437966e-01 -4.92521673e-01
-6.54543579e-01 -5.00245929e-01 7.24748135e-01 1.53164968e-01
-9.41182792e-01 7.92318463e-01 -1.05844615e-02 1.93885297e-01
-2.71892399e-01 -6.09542847e-01 -5.12721956e-01 -9.69674349e-01
-5.65417707e-02 3.59119058e-01 6.53163612e-01 2.17119902e-01
7.12774217e-01 6.54714048e-01 -1.31384492e+00 7.98291415e-02
5.74200153e-01 8.14243793e-01 5.01435220e-01 7.25785613e-01
-2.02778235e-01 8.36135626e-01 -6.84121609e-01 2.71171838e-01
3.05850059e-01 -4.40279663e-01 -8.80400836e-01 2.61276066e-01
3.52286369e-01 -1.77394748e-01 7.33734742e-02 1.06164312e+00
6.59901142e-01 3.55530441e-01 5.61576009e-01 -1.35960937e+00
-5.01410127e-01 4.93719906e-01 -6.30242825e-01 6.11043274e-01
3.10742915e-01 1.05748095e-01 7.70144701e-01 -5.87366760e-01
4.19570655e-01 1.25460613e+00 2.89273769e-01 5.07614970e-01
-1.34682202e+00 -3.66936237e-01 4.87368882e-01 7.08593309e-01
-1.34019744e+00 -3.97389442e-01 7.11540103e-01 -4.00158197e-01
8.48140597e-01 -1.96686476e-01 6.70396924e-01 9.58776295e-01
2.80757099e-01 9.03646052e-01 1.33341229e+00 -7.72246838e-01
7.35705316e-01 1.17967576e-01 9.80547518e-02 7.09191799e-01
1.88311592e-01 2.27910161e-01 -4.40838903e-01 -1.97562009e-01
8.22126508e-01 -1.15164548e-01 1.63159277e-02 -6.48287177e-01
-9.32610214e-01 1.15809298e+00 5.26397407e-01 3.37575167e-01
-5.18598020e-01 6.07887387e-01 2.06904653e-02 4.88667749e-02
1.79274157e-01 5.44154584e-01 -3.36958170e-01 1.76633075e-01
-5.85553169e-01 2.19789252e-01 6.60216630e-01 4.52877939e-01
8.47938418e-01 -2.00355560e-01 -4.35912728e-01 5.60428679e-01
8.43494356e-01 2.00582087e-01 6.39152288e-01 -9.31988418e-01
-2.30188593e-02 6.06114686e-01 4.87704873e-01 -8.35951090e-01
-3.10849607e-01 -7.64698088e-02 -1.56008229e-01 6.95553601e-01
2.11247459e-01 -2.84398794e-01 -9.27189410e-01 1.63204312e+00
4.74330813e-01 2.28917554e-01 -9.71664190e-02 9.36392546e-01
6.43774152e-01 6.51516676e-01 5.02720714e-01 -1.80300444e-01
7.65745103e-01 -1.22889602e+00 -7.75189817e-01 -1.40461236e-01
3.32381755e-01 -5.39249003e-01 6.40783429e-01 3.48807782e-01
-9.69970942e-01 -6.41171932e-01 -1.05436051e+00 2.58877635e-01
-4.82677907e-01 1.30889878e-01 6.29400969e-01 5.65556467e-01
-1.31156743e+00 4.96399045e-01 -1.11759353e+00 -4.42851573e-01
4.07820284e-01 7.38484859e-01 -9.09356773e-02 3.81554931e-01
-7.96382308e-01 1.05402482e+00 6.80180669e-01 2.42512390e-01
-1.19875455e+00 -6.04829527e-02 -5.32603562e-01 -2.21889019e-02
2.20069826e-01 -4.84535575e-01 1.23472619e+00 -1.33742619e+00
-1.79242873e+00 7.45782375e-01 2.05069464e-02 -7.78053105e-01
3.51317078e-01 -3.14179748e-01 -1.08750619e-01 2.53159076e-01
-2.98638828e-02 1.01845181e+00 1.05617404e+00 -1.31253028e+00
-8.83883953e-01 -3.17077041e-01 4.09661829e-01 4.00526792e-01
-1.20115625e-02 9.83501524e-02 -1.17377900e-02 -3.18433076e-01
1.33144140e-01 -9.66390193e-01 -4.95158702e-01 1.85178503e-01
1.94213726e-02 -5.88915110e-01 8.31381083e-01 -2.20309138e-01
7.85604417e-01 -2.12315392e+00 1.82272315e-01 5.07855974e-03
3.34511581e-03 1.67009100e-01 -2.08358273e-01 1.97853506e-01
1.17836989e-01 -7.46187568e-02 9.37629864e-02 -2.22489864e-01
-4.94549759e-02 1.08008593e-01 -4.03751075e-01 5.44750333e-01
8.27091932e-02 8.69281769e-01 -8.12125921e-01 -7.27690935e-01
1.72797024e-01 3.19478184e-01 -3.69352728e-01 3.16825390e-01
-2.49943912e-01 4.25305396e-01 -6.41646266e-01 5.06350398e-01
4.64124262e-01 -1.61593601e-01 5.27812988e-02 -8.85700285e-02
-4.16130573e-01 -1.19083412e-01 -1.28384340e+00 1.67403018e+00
-2.20915467e-01 6.03614867e-01 -2.00093351e-02 -1.23847532e+00
7.82108486e-01 1.68735445e-01 5.60447156e-01 -7.21739709e-01
1.81835428e-01 -2.00733677e-01 1.87653661e-01 -6.16292059e-01
3.55554581e-01 1.37044266e-01 3.27551991e-01 4.47734833e-01
2.14776844e-01 1.12589337e-01 3.43730986e-01 -1.70776486e-01
1.17136657e+00 4.35178429e-01 3.07155937e-01 -3.46935809e-01
4.79352623e-01 3.06098089e-02 3.68244529e-01 1.24759793e+00
-6.46475494e-01 2.34190479e-01 1.58782437e-01 -3.76248509e-01
-5.58610797e-01 -1.03762352e+00 -1.14102945e-01 1.19605911e+00
5.04320502e-01 2.01686360e-02 -6.78035259e-01 -7.83594072e-01
-1.16324522e-01 3.64234567e-01 -7.28144228e-01 -3.28177959e-01
-6.40980303e-01 -9.84827220e-01 2.89255947e-01 5.30385077e-01
8.78019929e-01 -1.42322862e+00 -1.22166026e+00 2.57129163e-01
2.23315462e-01 -8.03745687e-01 3.91485803e-02 5.45350432e-01
-8.24807942e-01 -9.00179267e-01 -5.84624469e-01 -9.79167461e-01
6.83500409e-01 1.67644888e-01 9.07276094e-01 -9.95879546e-02
-4.81271654e-01 7.76082814e-01 -6.45608008e-01 -5.72197855e-01
-1.16668671e-01 1.28345549e-01 -7.96000287e-02 1.81765124e-01
1.90242901e-01 -2.11152151e-01 -8.86132479e-01 2.60722458e-01
-7.71653891e-01 -2.61636585e-01 8.77629757e-01 6.22554302e-01
5.94446361e-01 -2.70371474e-02 4.93131399e-01 -6.16554797e-01
5.74305773e-01 -2.39948705e-01 -1.00503421e+00 4.98867244e-01
-5.17210484e-01 3.74975979e-01 -6.40557036e-02 -5.47167420e-01
-1.10348845e+00 6.86281264e-01 1.07624218e-01 -1.59098238e-01
-2.76338875e-01 2.28383616e-01 -2.86463834e-02 -4.95152175e-01
5.95387220e-01 9.08633545e-02 -2.41533399e-01 -1.34402573e-01
3.13571662e-01 5.07594109e-01 -1.10598207e-01 -4.68501657e-01
4.65498269e-01 4.29319322e-01 -1.75162390e-01 -5.64192533e-01
-8.54536951e-01 -5.56355417e-01 -6.37374759e-01 -3.56158078e-01
8.56658280e-01 -6.66336298e-01 -9.11339045e-01 3.91835332e-01
-1.12119567e+00 -6.32490695e-01 -3.80387247e-01 6.91639721e-01
-7.02970147e-01 1.61024123e-01 -1.92638889e-01 -6.77928805e-01
-1.33305266e-01 -1.23972726e+00 9.12473023e-01 5.70274115e-01
4.96222936e-02 -1.12684810e+00 5.87843418e-01 -9.80458595e-03
5.36087930e-01 -2.49008127e-02 4.92550105e-01 -5.16784430e-01
-9.34379578e-01 8.23712647e-02 -1.20958894e-01 1.08762383e-01
1.51524380e-01 1.46432877e-01 -8.76571536e-01 -4.28826094e-01
6.00998662e-02 -4.28854764e-01 9.91109490e-01 6.88488543e-01
1.07816148e+00 -1.26987189e-01 -4.75918114e-01 3.26615274e-01
1.59912777e+00 6.22255862e-01 6.58954442e-01 5.50231636e-01
1.29151836e-01 3.70851249e-01 7.24659801e-01 2.95542508e-01
-1.70968354e-01 6.14859998e-01 6.04911387e-01 -1.38433194e-02
-4.31014895e-02 1.42559364e-01 2.21089900e-01 1.22931659e-01
1.91676710e-02 -3.01733106e-01 -1.05645549e+00 4.21151847e-01
-2.16451883e+00 -1.00994873e+00 4.50189441e-01 1.89560556e+00
6.87733114e-01 2.75958359e-01 2.44418364e-02 -6.09081835e-02
8.61102521e-01 -2.24203381e-04 -5.89919150e-01 -1.42748728e-01
1.01259962e-01 2.55221099e-01 4.28261399e-01 5.33846974e-01
-1.36430514e+00 1.01418591e+00 7.63353348e+00 5.09325922e-01
-1.47051823e+00 9.88698602e-02 3.32362533e-01 1.98422089e-01
4.25345540e-01 2.62555093e-01 -1.04446316e+00 2.94093728e-01
5.73924422e-01 -2.48047728e-02 1.16607681e-01 1.06281078e+00
2.37920731e-01 -7.69708037e-01 -1.08986068e+00 1.00456119e+00
1.37315720e-01 -1.10331202e+00 2.21608840e-02 -4.17837799e-02
7.20980406e-01 -8.04180503e-02 1.68815628e-01 2.60709316e-01
4.33932096e-01 -8.26435685e-01 6.34442210e-01 6.78355277e-01
-1.41010970e-01 -4.37646866e-01 4.42022681e-01 2.78818160e-01
-8.65187347e-01 -4.99684215e-01 -3.93239439e-01 -5.88457137e-02
-5.94112687e-02 1.17662854e-01 -9.07487571e-01 -1.70134440e-01
6.41199529e-01 4.87666458e-01 -9.01401281e-01 1.69717872e+00
-3.20279807e-01 6.29033208e-01 -2.28952706e-01 -3.85572284e-01
2.48970732e-01 -9.97677371e-02 3.59032720e-01 9.00464594e-01
-5.22412173e-02 1.64929293e-02 3.66653860e-01 1.00141644e+00
3.17758262e-01 -9.96483024e-03 -3.56559306e-01 4.73679006e-02
2.46720359e-01 1.28013623e+00 -1.25606012e+00 -1.33127928e-01
-1.82222232e-01 9.37314034e-01 5.11025310e-01 4.05472279e-01
-9.15225744e-01 -3.12423706e-01 7.37966448e-02 1.90732479e-02
4.71161216e-01 -3.83304238e-01 2.36004233e-01 -9.14223611e-01
-3.93010437e-01 -3.85647923e-01 2.47861356e-01 -8.36711347e-01
-9.40405428e-01 4.48541403e-01 2.42932141e-01 -9.91196096e-01
-7.64131173e-02 -6.62726343e-01 -5.67946494e-01 4.82702225e-01
-1.28606606e+00 -7.62460411e-01 -3.69770646e-01 5.73611498e-01
6.28544450e-01 -2.41178125e-01 7.25416362e-01 7.34183192e-02
-6.48402572e-01 1.93027496e-01 1.48596078e-01 1.49401411e-01
5.82160890e-01 -1.27630115e+00 -1.98141247e-01 7.38804162e-01
5.11800349e-01 4.66906399e-01 7.41454303e-01 -3.48524630e-01
-1.03703952e+00 -7.95681059e-01 2.80941337e-01 -2.59798974e-01
4.19289470e-01 -1.65319860e-01 -5.58116376e-01 7.15293050e-01
6.03840709e-01 1.88064232e-01 3.21519613e-01 -2.32829377e-02
1.94381848e-01 -3.01473320e-01 -1.12314081e+00 3.68290961e-01
6.64799929e-01 -1.48144871e-01 -6.88676953e-01 2.68333107e-01
3.42913449e-01 -1.54775456e-01 -2.36899704e-01 4.29270267e-01
2.40932092e-01 -8.99554610e-01 8.10871124e-01 -4.45950061e-01
-1.05671868e-01 -4.64527458e-01 8.77340324e-03 -1.19894147e+00
-4.76199716e-01 -1.99366584e-01 -1.68975055e-01 9.15544033e-01
2.34454468e-01 -4.51986432e-01 8.42883468e-01 3.05626750e-01
2.96595305e-01 -6.28369927e-01 -8.18203270e-01 -5.05419016e-01
-3.02047431e-01 2.62072653e-01 -4.45150100e-02 5.12670934e-01
-3.45082104e-01 4.63933408e-01 6.65343329e-02 1.52270198e-01
4.43546534e-01 2.76648819e-01 4.05851811e-01 -1.04625034e+00
-2.86015451e-01 -3.08101952e-01 -6.88107908e-01 -8.61188591e-01
2.10125387e-01 -5.35679400e-01 3.69089097e-01 -1.62814903e+00
3.89378905e-01 -4.66598630e-01 -6.06865704e-01 7.04689145e-01
2.90617067e-02 1.38026416e-01 -1.33512281e-02 1.21614717e-01
-1.11581993e+00 5.75261891e-01 8.41147840e-01 -2.79265136e-01
-3.70648086e-01 1.26407832e-01 -1.72786057e-01 7.10678518e-01
1.06444716e+00 -6.46968663e-01 -4.97282565e-01 -4.18257803e-01
2.86087304e-01 -1.28572360e-01 5.42366445e-01 -1.30662596e+00
4.89936233e-01 -3.20422292e-01 5.60946167e-01 -5.48050523e-01
2.96272188e-01 -7.87565708e-01 -1.95092589e-01 6.55588031e-01
-6.14522278e-01 6.46594390e-02 7.92323723e-02 6.35517895e-01
-2.91196138e-01 -6.47926509e-01 1.03508556e+00 -4.01511312e-01
-1.29951501e+00 1.43873245e-01 -5.51874578e-01 -1.84405908e-01
1.48077440e+00 -3.21011096e-01 7.84551667e-04 -3.87581848e-02
-1.07200372e+00 4.16762345e-02 7.55557120e-02 4.38270301e-01
6.80674672e-01 -1.05559695e+00 -3.99240166e-01 -3.58613618e-02
-1.50333107e-01 -2.37087369e-01 -1.50830835e-01 7.40089118e-01
-7.65609562e-01 2.07686275e-01 -5.99384367e-01 -7.28134096e-01
-1.22579980e+00 6.33162737e-01 7.10777700e-01 -1.15088530e-01
-2.06604078e-01 8.42763543e-01 8.89905021e-02 -4.34870571e-02
5.13632715e-01 -9.44817439e-02 -5.06744623e-01 1.47342324e-01
3.36262912e-01 3.27645719e-01 -3.46096978e-02 -3.43299747e-01
-3.52172107e-01 6.18534386e-01 -2.14926407e-01 -2.70848066e-01
1.22685492e+00 -2.48827636e-01 6.13164715e-03 4.61003929e-01
8.45953047e-01 -3.86889607e-01 -1.30933475e+00 -1.08151548e-01
3.06663632e-01 -3.38409483e-01 3.04601401e-01 -8.71304035e-01
-9.54212070e-01 6.72182322e-01 1.42027950e+00 2.16700286e-01
1.01503575e+00 -1.01535385e-02 -5.71102574e-02 6.87374413e-01
5.17948091e-01 -1.34003699e+00 6.04050696e-01 2.93275416e-01
9.01505172e-01 -1.54027951e+00 1.75451607e-01 4.36457843e-02
-4.66198355e-01 1.10388899e+00 8.57549548e-01 -5.25409341e-01
7.78347909e-01 1.80478334e-01 1.96163028e-01 -1.12037472e-02
-6.24231756e-01 -4.89186138e-01 7.54827783e-02 4.35562491e-01
2.15506777e-01 -3.90822470e-01 -3.43201727e-01 9.18911919e-02
4.44692135e-01 1.34000452e-02 2.73557693e-01 1.21959448e+00
-8.41075063e-01 -1.09144509e+00 -3.61542374e-01 1.91036984e-02
-2.70645916e-01 2.89989591e-01 -4.97348994e-01 7.23185837e-01
3.47952574e-01 8.36818516e-01 8.93260092e-02 1.51756510e-01
4.68308665e-03 -1.76495656e-01 9.17962551e-01 -6.81099594e-01
-3.64133894e-01 -7.25752190e-02 -5.67690432e-01 -4.80398208e-01
-1.02031136e+00 -5.68502069e-01 -1.33624673e+00 3.54871720e-01
-5.75021505e-01 1.72002450e-01 8.05458188e-01 9.01703596e-01
2.70923495e-01 3.91052753e-01 5.17204523e-01 -8.79120469e-01
-6.19979203e-01 -9.47851777e-01 -5.05007505e-01 1.50267258e-01
2.63969123e-01 -8.98195148e-01 -2.84563929e-01 8.06706306e-03] | [9.329421043395996, 0.6355604529380798] |
d68afc92-a84c-4f8f-8e79-cf931adcfb8b | pre-training-contextualized-world-models-with | 2305.18499 | null | https://arxiv.org/abs/2305.18499v1 | https://arxiv.org/pdf/2305.18499v1.pdf | Pre-training Contextualized World Models with In-the-wild Videos for Reinforcement Learning | Unsupervised pre-training methods utilizing large and diverse datasets have achieved tremendous success across a range of domains. Recent work has investigated such unsupervised pre-training methods for model-based reinforcement learning (MBRL) but is limited to domain-specific or simulated data. In this paper, we study the problem of pre-training world models with abundant in-the-wild videos for efficient learning of downstream visual control tasks. However, in-the-wild videos are complicated with various contextual factors, such as intricate backgrounds and textured appearance, which precludes a world model from extracting shared world knowledge to generalize better. To tackle this issue, we introduce Contextualized World Models (ContextWM) that explicitly model both the context and dynamics to overcome the complexity and diversity of in-the-wild videos and facilitate knowledge transfer between distinct scenes. Specifically, a contextualized extension of the latent dynamics model is elaborately realized by incorporating a context encoder to retain contextual information and empower the image decoder, which allows the latent dynamics model to concentrate on essential temporal variations. Our experiments show that in-the-wild video pre-training equipped with ContextWM can significantly improve the sample-efficiency of MBRL in various domains, including robotic manipulation, locomotion, and autonomous driving. | ['Mingsheng Long', 'Chaoyi Deng', 'Haoyu Ma', 'Jialong Wu'] | 2023-05-29 | null | null | null | null | ['unsupervised-pre-training', 'model-based-reinforcement-learning'] | ['methodology', 'reasoning'] | [ 1.10025570e-01 -2.15377986e-01 -4.23839271e-01 -2.00054199e-01
-2.43228734e-01 -2.91030467e-01 5.49150348e-01 -3.56723905e-01
-3.87852162e-01 7.10614681e-01 4.05573472e-02 8.37168843e-02
-7.37909600e-02 -6.26028717e-01 -1.07626879e+00 -8.61326933e-01
-1.68904319e-01 2.33347118e-02 3.61376494e-01 -4.63544667e-01
1.95435286e-02 2.18846649e-01 -1.67608309e+00 2.99497634e-01
8.62257361e-01 5.66181421e-01 8.13356161e-01 6.41145110e-01
2.00281441e-02 1.12710643e+00 -4.33320314e-01 1.92688420e-01
1.76404923e-01 -6.63249433e-01 -3.05274367e-01 3.64381582e-01
3.38679045e-01 -4.77540463e-01 -7.59719193e-01 8.46657693e-01
3.33716661e-01 3.87892962e-01 3.47501308e-01 -1.30200660e+00
-4.13192540e-01 1.60912916e-01 -4.23147291e-01 9.72304195e-02
1.44355237e-01 8.19923580e-01 5.93374789e-01 -5.36197364e-01
1.07885575e+00 1.17838395e+00 3.34606558e-01 8.03570688e-01
-1.22437859e+00 -7.57019281e-01 6.12912238e-01 5.36796689e-01
-1.03032386e+00 -2.85839468e-01 9.38625932e-01 -4.98337269e-01
8.16215098e-01 -3.11414212e-01 9.62308466e-01 1.63799655e+00
3.79022986e-01 8.57393801e-01 9.66296136e-01 -3.07761729e-01
3.12957197e-01 -2.65692323e-01 -4.09255207e-01 8.11467230e-01
1.36889920e-01 4.76880878e-01 -9.68327880e-01 3.81252050e-01
1.16110301e+00 4.79171835e-02 -3.23839098e-01 -9.41053987e-01
-1.33390355e+00 6.50061369e-01 4.41255689e-01 3.03206891e-02
-1.99366421e-01 3.65937620e-01 3.93793881e-01 3.51568729e-01
1.22262817e-02 5.15482366e-01 -5.00919700e-01 -1.95195720e-01
-6.74295604e-01 1.04905635e-01 5.46964347e-01 1.24099195e+00
9.79335189e-01 3.75857323e-01 6.35620803e-02 7.08923161e-01
2.07670763e-01 5.32128394e-01 7.60147095e-01 -1.24591351e+00
7.06712961e-01 3.96425754e-01 3.00035626e-02 -9.16391432e-01
-1.56363875e-01 -3.42060596e-01 -7.55436122e-01 2.54044682e-01
2.44035751e-01 -2.45528385e-01 -1.05497313e+00 2.08668017e+00
3.56000036e-01 6.33918881e-01 1.53697446e-01 9.83553529e-01
3.95290256e-01 7.10949838e-01 9.99054536e-02 -2.23219529e-01
9.09629107e-01 -1.28687155e+00 -6.78624988e-01 -4.54101503e-01
6.33904755e-01 -4.22785372e-01 1.26737058e+00 3.72595996e-01
-6.75707996e-01 -8.90761375e-01 -1.19392943e+00 2.62394771e-02
-3.02946061e-01 9.18065384e-03 5.09989381e-01 5.41376881e-02
-7.24395394e-01 6.76988125e-01 -1.18927944e+00 -3.80191088e-01
3.20934713e-01 2.82314479e-01 -7.08499670e-01 -3.69172901e-01
-1.06876421e+00 8.33431661e-01 6.21199846e-01 4.34978493e-02
-1.39575791e+00 -5.74055195e-01 -1.02167690e+00 -2.73240179e-01
7.69880354e-01 -8.40433598e-01 1.09373295e+00 -1.05876696e+00
-1.97685659e+00 2.81881779e-01 3.62350754e-02 -3.01870018e-01
4.92168516e-01 -3.79364312e-01 -2.40392983e-01 2.54015744e-01
-5.37425317e-02 6.95127308e-01 1.27180362e+00 -1.32373667e+00
-4.89449799e-01 -2.17055544e-01 3.91205072e-01 3.49508524e-01
-2.55935609e-01 -5.20004869e-01 -6.20003700e-01 -7.82437265e-01
-1.47304624e-01 -1.04769039e+00 -3.99236590e-01 4.44501303e-02
1.31050691e-01 3.44744235e-01 1.20136547e+00 -4.68059152e-01
1.00030196e+00 -2.21253157e+00 5.10026455e-01 -2.42351547e-01
1.53972387e-01 2.72624791e-01 -3.62039477e-01 5.78710854e-01
1.21703930e-01 -4.31035966e-01 -3.30559425e-02 -2.84088671e-01
-1.66144133e-01 9.60587978e-01 -3.26002449e-01 2.52875656e-01
4.74987745e-01 8.39600027e-01 -1.19347298e+00 -5.46017826e-01
3.02303284e-01 3.59746099e-01 -8.54943752e-01 4.40559983e-01
-6.26028240e-01 1.13905001e+00 -6.27139747e-01 3.26434314e-01
3.05389524e-01 -2.17366412e-01 3.74805301e-01 -1.77275702e-01
-8.30432400e-02 -6.07632697e-02 -1.08597696e+00 2.11809444e+00
-6.18958473e-01 6.35387540e-01 1.89365581e-01 -1.20675874e+00
6.35208726e-01 4.38810997e-02 4.39508349e-01 -8.43191683e-01
-6.00921400e-02 9.32822078e-02 1.58287838e-01 -8.71702135e-01
3.96817356e-01 -7.21308887e-02 6.15756437e-02 1.90212894e-02
4.73791450e-01 -3.84230345e-01 3.05103600e-01 9.70559940e-02
9.20381606e-01 9.08415377e-01 -2.67359372e-02 2.82206889e-02
3.01095039e-01 1.11073814e-01 8.42585325e-01 6.33569300e-01
-3.03035140e-01 3.90601218e-01 3.02670121e-01 -3.80030692e-01
-1.02523041e+00 -9.20920312e-01 3.06972086e-01 1.10402870e+00
4.60145980e-01 -5.36574006e-01 -5.64269900e-01 -5.54521680e-01
-1.32215768e-01 3.98020297e-01 -5.05510628e-01 -3.78307223e-01
-8.92989755e-01 -4.21626538e-01 1.10149689e-01 5.72768688e-01
6.79299295e-01 -1.14133024e+00 -1.01855052e+00 4.29280162e-01
-1.44535840e-01 -1.49346149e+00 -3.66579324e-01 3.70939881e-01
-1.00400352e+00 -1.15471637e+00 -4.20434266e-01 -6.78874195e-01
5.37443995e-01 3.18364352e-01 8.72252464e-01 -1.68735012e-01
-3.65608960e-01 5.33305109e-01 -5.81649661e-01 -7.48290643e-02
-3.47154766e-01 -2.69775987e-01 1.19004980e-01 5.48468158e-02
-2.10004047e-01 -8.52578163e-01 -4.77582067e-01 4.23091173e-01
-9.80447173e-01 4.69902843e-01 6.42240763e-01 1.34164762e+00
6.11533403e-01 -1.52743295e-01 4.16857094e-01 -5.05779982e-01
1.44993728e-02 -3.33748013e-01 -5.84012628e-01 1.25870436e-01
-1.19479470e-01 2.24853486e-01 7.96308875e-01 -8.90304327e-01
-1.11391151e+00 1.77881390e-01 9.19573680e-02 -8.33006740e-01
-7.10922256e-02 5.24549425e-01 -2.37567350e-01 -5.69553524e-02
3.72100770e-01 3.34408164e-01 5.01601472e-02 -2.39647880e-01
4.07169938e-01 1.63145959e-01 5.81260324e-01 -8.68573248e-01
7.88344741e-01 4.44612533e-01 1.51574060e-01 -1.00136554e+00
-6.88343942e-01 -2.37013444e-01 -7.44677961e-01 -3.97499174e-01
9.23212230e-01 -1.20841467e+00 -2.54894614e-01 6.03700936e-01
-7.72438049e-01 -1.01727009e+00 -4.00509298e-01 8.41948032e-01
-1.04468083e+00 2.83380777e-01 -5.68838596e-01 -4.43883032e-01
3.99002850e-01 -1.32603514e+00 8.63958418e-01 2.95660049e-01
1.70344070e-01 -8.56348872e-01 1.39057398e-01 1.66984826e-01
3.27055663e-01 3.83105516e-01 8.79376650e-01 5.19875661e-02
-7.55851984e-01 1.91983759e-01 1.18391037e-01 3.70898485e-01
8.59193802e-02 -8.22640806e-02 -7.41709530e-01 -4.13170576e-01
-4.73929532e-02 -6.21811807e-01 9.25422490e-01 1.63479120e-01
1.12470543e+00 -1.81026515e-02 -2.72143960e-01 7.82614172e-01
1.36569118e+00 2.89216429e-01 7.22979426e-01 2.83750921e-01
8.61812115e-01 5.32226086e-01 8.17374945e-01 4.10953462e-01
3.33034962e-01 7.32397199e-01 5.42570949e-01 8.25648904e-02
-2.22136423e-01 -5.30836582e-01 7.47469842e-01 9.87587154e-01
-1.71204612e-01 -4.28132676e-02 -6.19876862e-01 4.94744629e-01
-2.07383919e+00 -9.75021303e-01 3.84917051e-01 1.83359563e+00
8.51133883e-01 8.00670758e-02 -7.34252706e-02 -2.34192088e-01
5.12318552e-01 3.44629824e-01 -8.86119664e-01 1.78839453e-02
-8.93756375e-02 2.38857269e-02 2.42186099e-01 2.77274281e-01
-1.06285727e+00 1.10034323e+00 5.46324587e+00 9.33388472e-01
-1.31504416e+00 5.51651865e-02 2.06004515e-01 -2.66276821e-02
5.52335083e-02 1.50442973e-01 -6.90346122e-01 4.81175631e-01
6.75866842e-01 2.02654600e-01 4.38148379e-01 8.94614756e-01
2.87554115e-01 -1.49177164e-01 -1.15730274e+00 9.43045735e-01
-2.19001412e-01 -1.12231302e+00 3.99532765e-02 6.28131703e-02
9.95788634e-01 1.29529312e-01 6.92221746e-02 7.07749605e-01
2.50306517e-01 -6.46199822e-01 7.46937096e-01 6.37381494e-01
7.14618444e-01 -4.73796755e-01 2.92295665e-01 6.82102144e-01
-1.12074184e+00 -4.16399628e-01 -4.04944062e-01 -1.71323359e-01
5.98120987e-02 2.20791385e-01 -4.54658747e-01 6.42601073e-01
6.10108614e-01 1.14206362e+00 -3.69584233e-01 7.48360634e-01
-1.88321978e-01 4.90177780e-01 -3.17994475e-01 1.74086943e-01
4.23166245e-01 -2.38718584e-01 4.72770631e-01 9.47188437e-01
3.88297230e-01 1.78139322e-02 4.97609168e-01 5.93992054e-01
1.50584668e-01 -2.49963641e-01 -7.80902684e-01 -1.52218029e-01
1.95456035e-02 7.52245247e-01 -4.41528589e-01 -3.50685418e-01
-5.61662078e-01 8.36167514e-01 4.13744152e-01 6.31706357e-01
-9.84514773e-01 -1.27595693e-01 7.35608160e-01 3.54799367e-02
6.70405447e-01 -8.19369256e-01 3.20519030e-01 -1.61212420e+00
1.33728283e-02 -8.65792036e-01 1.08108506e-01 -8.17436814e-01
-1.00352335e+00 2.03896597e-01 2.91360468e-01 -1.55913138e+00
-3.17328304e-01 -7.88900197e-01 -4.45907205e-01 3.91759962e-01
-1.75205624e+00 -1.24065518e+00 -5.37136495e-01 8.87727380e-01
9.25877154e-01 -2.20532984e-01 5.63774824e-01 1.85194686e-01
-7.29209900e-01 2.25065961e-01 2.85676628e-01 -1.89945269e-02
8.75737846e-01 -9.45437968e-01 1.16436714e-02 8.67358506e-01
9.17368978e-02 5.55162609e-01 6.51138186e-01 -6.10958397e-01
-1.89958024e+00 -1.24758470e+00 5.93365319e-02 -1.83680743e-01
8.38327944e-01 -4.03790414e-01 -7.57013083e-01 6.58737540e-01
9.89594832e-02 2.90374517e-01 2.49947086e-01 -1.70679703e-01
-3.52403581e-01 -2.00004607e-01 -6.43467724e-01 9.40862417e-01
1.32295036e+00 -5.70447266e-01 -4.46087062e-01 9.48764570e-03
7.59823561e-01 -4.61098701e-01 -6.35393023e-01 4.93707776e-01
6.45953953e-01 -9.29172099e-01 8.98747504e-01 -7.33320951e-01
6.11043036e-01 -4.58211631e-01 -1.97220877e-01 -1.46898127e+00
-1.09469339e-01 -7.01644659e-01 -5.09325683e-01 8.79725635e-01
-7.44710490e-02 -4.46509272e-01 6.51043475e-01 1.59994647e-01
-3.19966435e-01 -6.56422794e-01 -7.59888947e-01 -9.49612439e-01
-3.20753157e-02 -4.66184825e-01 2.66399264e-01 8.56139600e-01
-3.76877822e-02 2.36852631e-01 -6.60415292e-01 1.26171023e-01
3.12221646e-01 1.70975402e-01 1.08710015e+00 -6.92029178e-01
-6.42630935e-01 -6.14913218e-02 -5.54695964e-01 -1.45478308e+00
2.91929752e-01 -4.09828991e-01 2.81675696e-01 -1.19206274e+00
5.78782745e-02 -3.53429884e-01 -3.14467996e-01 4.15172935e-01
-1.27545014e-01 1.03770336e-02 3.69647861e-01 2.61000544e-01
-7.97676921e-01 9.93035197e-01 1.83978558e+00 -2.03739956e-01
-3.32394838e-01 -3.87805909e-01 -6.44864440e-02 6.89944446e-01
7.23220170e-01 -3.14901501e-01 -9.28751886e-01 -6.67061329e-01
-6.62486115e-03 2.44829580e-01 5.62464952e-01 -1.33195293e+00
2.27745444e-01 -5.22660375e-01 2.84658343e-01 -3.14189732e-01
5.60414016e-01 -8.83379042e-01 2.88318079e-02 4.89989519e-01
-1.71703845e-01 -3.30433547e-02 3.83454412e-01 1.10459101e+00
-4.68346357e-01 5.26781790e-02 6.92838073e-01 -2.64859974e-01
-1.28022695e+00 3.07527333e-01 -3.70406866e-01 1.54584453e-01
1.11523390e+00 -2.49831200e-01 -2.73558915e-01 -3.40278327e-01
-8.86316717e-01 3.46487582e-01 4.55370635e-01 5.20324409e-01
5.60840905e-01 -1.20891869e+00 -3.33735049e-01 5.20589769e-01
2.68514037e-01 1.82245672e-01 5.85444093e-01 8.21911395e-01
-5.37726521e-01 1.47155181e-01 -6.80642247e-01 -8.46108615e-01
-9.17956054e-01 7.79478788e-01 2.13019237e-01 -5.25754809e-01
-8.25489521e-01 5.62654138e-01 6.22929215e-01 -3.07073146e-01
1.15332633e-01 -5.30696511e-01 1.18949106e-02 -2.02327669e-01
2.49530718e-01 6.72650412e-02 -3.56012046e-01 -5.08022964e-01
-7.24914344e-03 7.90176809e-01 7.90591910e-02 -8.62288103e-02
1.38332939e+00 -2.51137972e-01 2.99911678e-01 5.46163023e-01
1.11599374e+00 -3.30697984e-01 -2.12063718e+00 -1.85421571e-01
-2.22664684e-01 -4.25382584e-01 -1.23785049e-01 -5.36074698e-01
-9.53692257e-01 1.20658636e+00 4.30807650e-01 -3.79795611e-01
1.18571889e+00 -4.15518939e-01 6.96063340e-01 7.33572483e-01
8.01211774e-01 -1.40208149e+00 6.95350409e-01 8.89905393e-01
9.16115403e-01 -1.27334273e+00 -1.16267793e-01 -2.03274414e-01
-8.34415197e-01 1.14067972e+00 9.83150303e-01 -2.71632671e-01
6.40200615e-01 1.68223977e-01 -1.08958788e-01 1.22215092e-01
-9.93865252e-01 -3.38411450e-01 -2.37523988e-02 9.35066879e-01
-2.51014531e-01 -2.84349233e-01 2.09278286e-01 3.90612066e-01
2.33003303e-01 1.82402611e-01 4.00640458e-01 1.29919791e+00
-3.04054677e-01 -1.19468606e+00 -7.98908249e-03 -5.78249916e-02
4.85485326e-03 2.31133640e-01 1.44028917e-01 1.16579497e+00
4.32796389e-01 7.23854125e-01 -1.77896887e-01 -4.99484837e-01
2.79616475e-01 2.49433480e-02 7.93480217e-01 -6.13134980e-01
-9.42403525e-02 2.32940927e-01 2.05343291e-02 -7.95679510e-01
-6.58578992e-01 -3.53229851e-01 -1.07784951e+00 4.74391952e-02
-2.58976430e-01 -1.15368567e-01 4.93288040e-01 1.01100481e+00
5.16471386e-01 7.73857772e-01 4.59172875e-01 -1.31594396e+00
-4.52735454e-01 -8.18493068e-01 -5.18066585e-01 4.07023400e-01
7.02569485e-01 -1.17118084e+00 -1.15527205e-01 5.69068313e-01] | [7.468553066253662, -0.04944642633199692] |
b61fed4b-3bbc-4eee-b6e8-0ded94388731 | semi-supervised-counterfactual-explanations-1 | 2303.12634 | null | https://arxiv.org/abs/2303.12634v1 | https://arxiv.org/pdf/2303.12634v1.pdf | Semi-supervised counterfactual explanations | Counterfactual explanations for machine learning models are used to find minimal interventions to the feature values such that the model changes the prediction to a different output or a target output. A valid counterfactual explanation should have likely feature values. Here, we address the challenge of generating counterfactual explanations that lie in the same data distribution as that of the training data and more importantly, they belong to the target class distribution. This requirement has been addressed through the incorporation of auto-encoder reconstruction loss in the counterfactual search process. Connecting the output behavior of the classifier to the latent space of the auto-encoder has further improved the speed of the counterfactual search process and the interpretability of the resulting counterfactual explanations. Continuing this line of research, we show further improvement in the interpretability of counterfactual explanations when the auto-encoder is trained in a semi-supervised fashion with class tagged input data. We empirically evaluate our approach on several datasets and show considerable improvement in-terms of several metrics. | ['Satyam Dwivedi', 'Sumanta Mukherjee', 'Shravan Kumar Sajja'] | 2023-03-22 | semi-supervised-counterfactual-explanations | https://openreview.net/forum?id=o6ndFLB1DST | https://openreview.net/pdf?id=o6ndFLB1DST | null | ['counterfactual-explanation'] | ['miscellaneous'] | [ 5.31472802e-01 8.94002140e-01 -5.47478497e-01 -5.24153411e-01
-5.35307407e-01 -4.44148332e-01 1.05752158e+00 -5.62564805e-02
-4.24527638e-02 1.36626434e+00 6.71602428e-01 -6.76318169e-01
-3.21225405e-01 -7.98078954e-01 -1.09199476e+00 -5.25814354e-01
7.07257092e-02 5.79170167e-01 -3.75375718e-01 3.21921766e-01
3.42840672e-01 3.55414301e-01 -1.68285203e+00 6.10831857e-01
6.96552038e-01 6.74664199e-01 -1.37041613e-01 5.38689494e-01
4.47305888e-02 8.24266076e-01 -4.63706225e-01 -7.30000198e-01
3.45719546e-01 -7.25091875e-01 -9.90231514e-01 1.25020221e-01
2.99879372e-01 -7.32156560e-02 -6.97738081e-02 9.49948549e-01
-3.28227319e-02 -1.92062128e-02 1.09011042e+00 -1.65518427e+00
-6.46088421e-01 9.32022274e-01 -1.46268204e-01 -3.69298123e-02
1.41453862e-01 2.32962102e-01 1.13990140e+00 -5.95875025e-01
6.11822605e-01 1.52200282e+00 3.90801609e-01 7.04111576e-01
-1.56131852e+00 -7.98619211e-01 1.03575863e-01 1.65431529e-01
-6.63542569e-01 -4.37997937e-01 8.25014174e-01 -3.96608204e-01
9.19164538e-01 3.22294354e-01 4.44007486e-01 1.24928594e+00
4.40923810e-01 5.65040052e-01 1.19100666e+00 -6.91531837e-01
5.38960874e-01 4.59605992e-01 -2.41812259e-01 4.62541670e-01
5.62420845e-01 7.30629265e-01 -4.06947106e-01 -4.18424755e-01
4.62277442e-01 -2.64617559e-02 -2.09036589e-01 -9.58474755e-01
-1.02674234e+00 1.28409386e+00 5.40401995e-01 1.56993076e-01
-6.10099494e-01 4.21407223e-01 3.58535796e-01 3.89988422e-01
6.42620146e-01 9.31153297e-01 -8.53358984e-01 2.40654722e-01
-8.41859996e-01 4.61017638e-01 6.95983410e-01 4.78141040e-01
5.48801661e-01 2.43173633e-02 -1.51189610e-01 2.87528485e-01
2.90994614e-01 4.45813090e-01 8.93828452e-01 -1.07866120e+00
6.11170053e-01 6.54366076e-01 3.58878255e-01 -5.88366568e-01
-5.57223409e-02 -4.48946536e-01 -4.18009520e-01 4.72689599e-01
5.18281877e-01 -2.60425240e-01 -7.31114686e-01 1.83327293e+00
3.14341038e-01 3.37996818e-02 4.30077523e-01 7.84716904e-01
-2.62380928e-01 4.16053653e-01 1.25670418e-01 -5.64077318e-01
8.12311351e-01 -5.81736982e-01 -4.91005242e-01 -3.00336212e-01
8.67278278e-01 -5.62326133e-01 9.34807181e-01 3.55651379e-02
-8.10374916e-01 -4.48562384e-01 -1.10584331e+00 3.88706744e-01
-1.17925897e-01 -2.19289362e-02 9.19119000e-01 6.14182889e-01
-4.99102414e-01 9.80068445e-01 -5.69697142e-01 6.10727035e-02
5.41924357e-01 3.87758374e-01 -3.25700790e-01 1.25569612e-01
-1.18959534e+00 9.40743148e-01 7.40577400e-01 -5.51632226e-01
-8.08762729e-01 -8.69953096e-01 -8.27996433e-01 4.23828036e-01
3.40804219e-01 -9.19076979e-01 1.33965373e+00 -1.57335854e+00
-1.15273857e+00 5.56266844e-01 -1.49561763e-01 -1.06198752e+00
9.80771244e-01 -6.65516406e-02 -3.32347572e-01 -3.30548316e-01
4.70329195e-01 6.19730473e-01 8.77423167e-01 -1.11073589e+00
-8.48343134e-01 -3.80564213e-01 4.41281199e-02 1.57378882e-01
1.81825042e-01 -3.95096362e-01 6.72548652e-01 -5.43360889e-01
-1.58860952e-01 -1.01164186e+00 -2.36684650e-01 -1.79806322e-01
-5.59549272e-01 -1.83032632e-01 7.11124480e-01 -3.01158261e-02
8.41613293e-01 -1.98011124e+00 -2.78191179e-01 8.54439810e-02
-2.04557195e-01 -1.56638436e-02 3.11606497e-01 5.95788956e-02
-9.52779412e-01 3.77465516e-01 -2.98304677e-01 -1.79429520e-02
1.33764967e-01 1.04711674e-01 -1.09098709e+00 5.26864111e-01
1.89560682e-01 7.20022023e-01 -8.12252998e-01 -2.67859221e-01
2.18269199e-01 -2.26247117e-01 -6.86409533e-01 2.82216787e-01
-4.33413684e-01 -8.10154900e-02 -2.97268480e-01 -2.79375404e-01
4.71825510e-01 -1.43381819e-01 3.19958955e-01 3.16240549e-01
-6.02866709e-02 7.94305980e-01 -1.10124671e+00 1.05931377e+00
-6.79233849e-01 7.34569550e-01 -9.08228457e-01 -1.03786528e+00
6.39233768e-01 6.71503901e-01 3.71117182e-02 -2.66448528e-01
-1.03889540e-01 4.31348085e-01 3.32430363e-01 -2.18859300e-01
3.23089838e-01 -8.81289780e-01 -3.01607773e-02 9.22949195e-01
7.87921697e-02 2.79707555e-02 -2.47806400e-01 -1.56487525e-01
4.87550735e-01 1.72478184e-01 9.16248024e-01 -2.94932008e-01
2.27321416e-01 2.25157723e-01 3.45920473e-01 9.73238170e-01
1.36225060e-01 4.86792535e-01 7.16204762e-01 -6.20178282e-01
-1.20054483e+00 -1.19767988e+00 -8.95467252e-02 4.29444104e-01
-4.43577677e-01 1.42909586e-01 -5.99682033e-01 -1.44461393e+00
2.04446003e-01 1.69453144e+00 -1.02472401e+00 -6.20832920e-01
-1.53238341e-01 -6.08975112e-01 2.74430305e-01 4.56230551e-01
3.03586721e-01 -1.13334787e+00 -9.22236621e-01 4.58538122e-02
-1.25077128e-01 -3.01509827e-01 -1.88664898e-01 2.87724674e-01
-1.11987889e+00 -1.35479891e+00 -3.17106932e-01 -7.31673539e-02
7.13759005e-01 -1.50811836e-01 8.68709922e-01 -1.93885729e-01
1.65915683e-01 -3.13701749e-01 7.19854757e-02 -7.93444932e-01
-7.34254122e-01 -2.31338233e-01 2.22659692e-01 -3.99317145e-02
3.97816837e-01 -4.96909052e-01 -3.55958939e-01 -8.08512270e-02
-5.88141739e-01 3.02148074e-01 4.72241372e-01 1.06610107e+00
2.09926262e-01 2.12542415e-01 8.31823170e-01 -1.24605691e+00
5.00502288e-01 -6.83616459e-01 -5.64859152e-01 1.36986136e-01
-1.12128556e+00 7.77617216e-01 1.02480018e+00 -5.75467587e-01
-1.49193478e+00 1.00467868e-01 2.56144911e-01 -3.44773769e-01
-3.96677792e-01 1.09218456e-01 -2.72699833e-01 8.73783171e-01
8.30912769e-01 2.94466391e-02 -9.27643180e-02 -3.19632947e-01
5.82897782e-01 6.93348765e-01 3.80124271e-01 -1.52924955e-01
6.52011395e-01 5.80298483e-01 1.56418562e-01 -6.02439158e-02
-1.08748353e+00 8.47614706e-02 -3.99117112e-01 2.60163337e-01
4.57179874e-01 -5.23926914e-01 -4.20722961e-01 -3.07528496e-01
-1.16036916e+00 -1.12204492e-01 -9.23794150e-01 7.32008994e-01
-1.12883306e+00 -2.38204226e-01 3.35799128e-01 -1.10929418e+00
8.88335630e-02 -9.13128555e-01 7.19848096e-01 -9.86005440e-02
-7.99450696e-01 -1.08312309e+00 8.40999633e-02 1.62527621e-01
-2.12273933e-02 2.83568650e-01 1.27999914e+00 -1.07168269e+00
-3.41336071e-01 -2.41878077e-01 -1.08182140e-01 1.46478176e-01
1.29097790e-01 -3.45447540e-01 -1.15711474e+00 -2.47867387e-02
1.92812130e-01 -8.99393260e-02 1.00474596e+00 6.02695227e-01
9.67670083e-01 -9.45713401e-01 -5.72005987e-01 1.00583084e-01
1.27791286e+00 3.09933454e-01 4.34466422e-01 3.21096808e-01
3.50316912e-02 8.22829425e-01 7.16448247e-01 2.35465214e-01
1.03161991e-01 6.48362279e-01 4.14499551e-01 4.32968475e-02
8.99208039e-02 -7.99007356e-01 3.96571279e-01 -5.18300772e-01
3.20905983e-01 -1.40489936e-01 -4.37213451e-01 9.47118580e-01
-1.97059536e+00 -1.46033573e+00 -1.17900819e-01 2.26577783e+00
7.49413669e-01 2.56245255e-01 -1.12063088e-01 3.56086165e-01
6.09345913e-01 -2.43154708e-02 -6.96120083e-01 -8.80470634e-01
2.40595251e-01 -1.48944572e-01 4.05283153e-01 7.37269700e-01
-9.95590508e-01 5.34468055e-01 6.35054445e+00 5.45986712e-01
-1.07457769e+00 2.98927948e-02 8.52705538e-01 -2.16921315e-01
-7.34054089e-01 4.08170670e-01 -3.82072896e-01 6.58378541e-01
1.15855074e+00 -6.57733083e-01 1.59871489e-01 1.06198919e+00
4.06273901e-01 -1.66303422e-02 -1.67327523e+00 4.14181352e-01
-2.13226199e-01 -1.64642549e+00 3.70371193e-01 4.15619284e-01
9.14087474e-01 -3.75314802e-01 1.62543505e-01 2.14509070e-01
6.04643941e-01 -1.09388316e+00 1.04154396e+00 4.91838425e-01
7.11498618e-01 -9.49518621e-01 9.79054749e-01 6.64125562e-01
-2.81024963e-01 -4.40499723e-01 -3.50172281e-01 -6.07829690e-01
-1.47898078e-01 4.99843419e-01 -1.57933915e+00 3.63348633e-01
8.82061422e-02 2.58746117e-01 -2.37846598e-01 5.42542577e-01
-5.61024427e-01 7.11933076e-01 1.37624830e-01 -3.29115912e-02
5.18506654e-02 2.89294928e-01 4.18778658e-01 9.30137813e-01
3.06565315e-01 -1.43607512e-01 -2.50115931e-01 1.29339588e+00
-3.70836407e-02 -2.20182642e-01 -1.10191762e+00 1.82931334e-01
2.65048742e-01 3.77122402e-01 -3.59822482e-01 -6.13960624e-01
-4.03261557e-03 8.23417127e-01 2.33839333e-01 2.17700958e-01
-8.11915457e-01 -1.91076875e-01 7.50076652e-01 1.09614581e-01
2.10364580e-01 7.60340512e-01 -6.32835627e-01 -1.10153747e+00
-8.12326595e-02 -6.96341634e-01 5.62107086e-01 -9.09480751e-01
-1.07461274e+00 1.61855206e-01 2.31072038e-01 -1.04135263e+00
-1.16667354e+00 -3.52910876e-01 -7.19531476e-01 1.13412082e+00
-1.26054072e+00 -7.88431525e-01 6.14329398e-01 7.47846588e-02
6.60984337e-01 -2.45798975e-01 9.18340325e-01 -5.95177650e-01
4.29185554e-02 3.37112427e-01 3.12575395e-03 -2.46515587e-01
2.29365587e-01 -1.47165513e+00 4.00087982e-01 9.40156996e-01
3.59732866e-01 6.19973063e-01 1.17338276e+00 -6.43306673e-01
-4.88998115e-01 -1.20743227e+00 1.54024398e+00 -5.59291124e-01
4.71974313e-01 -8.68279710e-02 -5.76487005e-01 9.37097490e-01
1.35917187e-01 -3.76481026e-01 5.97727120e-01 2.28214785e-01
-2.54554302e-01 2.55903721e-01 -1.34689522e+00 7.04404354e-01
8.18949103e-01 -5.42450607e-01 -1.25358808e+00 2.13507861e-01
7.89170742e-01 5.34816971e-03 -1.00594215e-01 -5.65531664e-02
6.04315281e-01 -1.02503347e+00 8.22955906e-01 -1.27779281e+00
9.20173645e-01 -1.08696349e-01 -1.60000280e-01 -1.75750375e+00
-1.84410781e-01 -3.73681009e-01 -1.19061172e-01 1.03093469e+00
7.41475642e-01 -7.55005419e-01 7.35372126e-01 7.11642742e-01
2.97354221e-01 -7.01686919e-01 -1.11608589e+00 -6.39585018e-01
1.06922217e-01 -5.15784264e-01 1.10161006e+00 8.55862439e-01
3.44612122e-01 3.81243527e-01 -2.66275495e-01 4.94050272e-02
6.23672247e-01 7.80655980e-01 6.65722966e-01 -1.10806549e+00
-4.51279253e-01 -2.07186565e-01 -2.66446590e-01 -6.14516139e-01
5.85220814e-01 -1.00970304e+00 -9.30633917e-02 -1.12053275e+00
4.22161639e-01 -9.00637507e-02 -1.01980530e-01 4.73060757e-01
-2.49739662e-01 -1.01922326e-01 3.77674341e-01 1.55004382e-01
-2.28467621e-02 5.20198822e-01 8.87928843e-01 1.52472660e-01
-6.23181388e-02 3.91260147e-01 -9.82477427e-01 9.53707099e-01
7.73210883e-01 -8.99938107e-01 -6.09070718e-01 -3.96361545e-04
6.51566163e-02 2.71067441e-01 6.22160137e-01 -3.91860574e-01
-3.05116594e-01 -3.55611175e-01 5.90247512e-01 -5.61460592e-02
1.17752656e-01 -9.93124187e-01 3.10752422e-01 9.37988281e-01
-1.06349349e+00 -1.78995188e-02 -4.72655445e-02 7.69940495e-01
9.49829072e-03 -4.75471467e-01 6.50111675e-01 -1.59355357e-01
-2.52454847e-01 -2.99358755e-01 -4.06735390e-01 -5.69602028e-02
1.04561603e+00 -2.14486614e-01 -1.08349457e-01 -5.20261765e-01
-5.62693596e-01 -1.10284075e-01 5.11862159e-01 5.23902535e-01
4.49400395e-01 -1.27132702e+00 -6.25339031e-01 2.82309413e-01
1.19064555e-01 -5.14304280e-01 -2.19419688e-01 3.44697237e-01
1.56881452e-01 9.61621881e-01 -1.72592729e-01 -8.16012546e-02
-9.59951341e-01 7.16176212e-01 6.09457254e-01 -5.93864024e-01
-3.72717828e-01 2.61044770e-01 4.05506700e-01 -4.74127591e-01
-3.89104068e-01 -3.81692410e-01 -1.00312702e-01 -4.98139858e-02
3.15547168e-01 3.77156198e-01 -1.34264335e-01 -3.81833941e-01
-1.90739945e-01 -4.23128605e-01 1.16384432e-01 -4.17747408e-01
1.45937884e+00 2.80859955e-02 3.50348949e-01 5.89487255e-01
1.18502557e+00 -2.89605279e-02 -1.45204353e+00 8.65261778e-02
3.66985023e-01 -8.56876135e-01 -6.60818219e-02 -1.14950073e+00
-5.04485726e-01 7.28151679e-01 4.75265801e-01 3.28414768e-01
8.51329446e-01 1.47420883e-01 2.88056955e-02 1.21204473e-01
2.77723134e-01 -8.16291809e-01 -3.06846619e-01 -8.19529742e-02
1.05697453e+00 -1.26965046e+00 5.46470843e-02 -1.56685226e-02
-7.99582064e-01 8.97862613e-01 1.51350573e-01 -2.70347178e-01
2.36557946e-01 -2.23514900e-01 -8.91345069e-02 -1.83835521e-01
-1.24563527e+00 1.94106460e-01 2.91365236e-01 4.92908478e-01
6.86851561e-01 4.85643655e-01 -3.31969678e-01 5.26053607e-01
-7.56038904e-01 8.70926902e-02 6.47628963e-01 1.69209689e-01
-8.03046301e-02 -7.31954396e-01 -2.76431024e-01 7.06040680e-01
-5.67000091e-01 -4.00508977e-02 -5.48188210e-01 9.47664201e-01
1.92519382e-01 9.33952630e-01 2.29733929e-01 7.18648732e-02
2.53789186e-01 5.72578490e-01 1.64936021e-01 -7.73365796e-01
-3.00312012e-01 -3.42785209e-01 2.25354731e-01 -4.90751654e-01
-2.30328277e-01 -9.76169050e-01 -1.23632646e+00 1.22048985e-03
-5.38914919e-01 6.14659250e-01 5.84010363e-01 1.34168088e+00
3.35877597e-01 3.58979791e-01 7.95574188e-01 -3.16931337e-01
-1.14790654e+00 -9.27622318e-01 -3.16155046e-01 6.14886165e-01
6.63269639e-01 -6.09610438e-01 -5.62613845e-01 1.70578972e-01] | [8.652080535888672, 5.625154495239258] |
98f93a91-9a3a-48db-89ee-80123b6d936b | multi-hypothesis-3d-human-pose-estimation | 2210.11179 | null | https://arxiv.org/abs/2210.11179v1 | https://arxiv.org/pdf/2210.11179v1.pdf | Multi-hypothesis 3D human pose estimation metrics favor miscalibrated distributions | Due to depth ambiguities and occlusions, lifting 2D poses to 3D is a highly ill-posed problem. Well-calibrated distributions of possible poses can make these ambiguities explicit and preserve the resulting uncertainty for downstream tasks. This study shows that previous attempts, which account for these ambiguities via multiple hypotheses generation, produce miscalibrated distributions. We identify that miscalibration can be attributed to the use of sample-based metrics such as minMPJPE. In a series of simulations, we show that minimizing minMPJPE, as commonly done, should converge to the correct mean prediction. However, it fails to correctly capture the uncertainty, thus resulting in a miscalibrated distribution. To mitigate this problem, we propose an accurate and well-calibrated model called Conditional Graph Normalizing Flow (cGNFs). Our model is structured such that a single cGNF can estimate both conditional and marginal densities within the same model - effectively solving a zero-shot density estimation problem. We evaluate cGNF on the Human~3.6M dataset and show that cGNF provides a well-calibrated distribution estimate while being close to state-of-the-art in terms of overall minMPJPE. Furthermore, cGNF outperforms previous methods on occluded joints while it remains well-calibrated. | ['Fabian H. Sinz', 'Mohammad Bashiri', 'R. James Cotton', 'Paweł A. Pierzchlewicz'] | 2022-10-20 | null | null | null | null | ['multi-hypotheses-3d-human-pose-estimation', '3d-human-pose-estimation'] | ['computer-vision', 'computer-vision'] | [-5.53936735e-02 3.18714380e-01 -2.78421789e-01 -3.12832177e-01
-1.12779403e+00 -1.41238764e-01 4.63884860e-01 -2.76922524e-01
-2.68761545e-01 1.06925225e+00 4.75080580e-01 5.61654717e-02
-1.01758003e-01 -6.27962887e-01 -9.44033325e-01 -4.93478894e-01
-7.47737139e-02 9.97187078e-01 4.35531288e-01 1.72535017e-01
1.14664182e-01 2.77758151e-01 -1.58630359e+00 -1.76722050e-01
1.03615224e+00 4.69863325e-01 1.81390479e-01 6.58009410e-01
-3.70275564e-02 4.13067132e-01 -8.23573530e-01 -5.51952243e-01
1.98512286e-01 -2.11859152e-01 -7.11620450e-01 1.32465005e-01
9.01306748e-01 -6.66927993e-01 -2.73737460e-01 1.11329508e+00
4.75250363e-01 2.24937037e-01 1.03064060e+00 -1.38064146e+00
-1.70011207e-01 2.81855434e-01 -9.27281201e-01 -9.79780555e-02
3.09878260e-01 -1.02744535e-01 9.09698844e-01 -8.94413769e-01
8.14746022e-01 1.89694238e+00 7.72322655e-01 4.22699302e-01
-1.68572903e+00 -5.71083307e-01 1.11644454e-01 -2.13159189e-01
-1.57118452e+00 -3.05909365e-01 3.82997543e-01 -5.62446237e-01
7.29988098e-01 -1.60554886e-01 6.36763811e-01 1.32576656e+00
7.13559687e-01 6.20483279e-01 7.75376201e-01 -2.68455803e-01
4.04815137e-01 -2.53858894e-01 -1.62429780e-01 5.66655695e-01
5.84048390e-01 1.95157066e-01 -8.97360504e-01 -2.67908514e-01
1.18479633e+00 -4.82337207e-01 -3.13581467e-01 -6.44910872e-01
-1.07726097e+00 6.97684109e-01 3.73746812e-01 -2.80133665e-01
-1.11890234e-01 4.57359463e-01 -3.50343511e-02 -2.92441756e-01
5.92625737e-01 2.41674513e-01 -2.46935785e-01 -5.30557871e-01
-1.07913959e+00 6.80339694e-01 8.83599579e-01 9.85181034e-01
8.96368206e-01 -1.21613503e-01 -3.78027596e-02 8.35266113e-01
7.04021871e-01 5.96624434e-01 -4.82065603e-02 -1.43698704e+00
4.04649228e-01 2.19963118e-01 1.94175407e-01 -1.05191827e+00
-2.31486574e-01 -5.97485483e-01 -5.62405407e-01 5.28933525e-01
7.82542467e-01 -9.70719829e-02 -1.31561947e+00 2.05832291e+00
4.88741398e-01 1.28459021e-01 -2.72714376e-01 9.57097769e-01
1.08545668e-01 5.29560983e-01 1.72599092e-01 1.16020486e-01
1.04362309e+00 -7.23802328e-01 -7.93137908e-01 -6.76762402e-01
3.54656667e-01 -8.19968820e-01 9.44339633e-01 4.22868311e-01
-7.86959648e-01 -2.78152168e-01 -9.71458912e-01 -1.95290931e-02
3.16922665e-01 -2.43656561e-01 4.54382002e-01 5.32730520e-01
-8.22705805e-01 7.58458555e-01 -1.24916768e+00 -5.04616380e-01
3.07689607e-01 1.23619467e-01 -4.70772415e-01 -3.06119055e-01
-9.12855983e-01 1.24723387e+00 3.34696740e-01 -4.87637855e-02
-8.83728743e-01 -8.75044048e-01 -1.02945566e+00 -2.84712702e-01
4.09366429e-01 -8.64687800e-01 1.21810269e+00 -3.22797894e-01
-1.30306733e+00 3.13562036e-01 -3.35918635e-01 -2.10045829e-01
6.82987094e-01 -6.70534670e-01 1.19819120e-01 1.25859380e-01
3.31468076e-01 1.13940561e+00 7.59058475e-01 -1.54187751e+00
-4.17536587e-01 -3.07598114e-01 -5.27687855e-02 3.32416117e-01
1.80972293e-01 -6.48142695e-01 -8.06583166e-01 -6.03637993e-01
3.68239909e-01 -1.03947866e+00 -3.43021423e-01 2.96867818e-01
-5.65621555e-01 1.75005063e-01 5.78966141e-01 -7.37050891e-01
1.11030543e+00 -1.88987291e+00 3.87344927e-01 1.83654502e-01
1.79543495e-01 -2.29065686e-01 1.28433943e-01 3.55729133e-01
4.75990206e-01 1.25057533e-01 -4.20166314e-01 -8.51909935e-01
2.00441182e-01 8.65586758e-01 -9.52426866e-02 7.25475311e-01
3.07424664e-01 5.10220826e-01 -9.44249928e-01 -8.02102268e-01
5.95793307e-01 7.32627749e-01 -8.67165565e-01 1.15447395e-01
-3.06893528e-01 5.58750987e-01 -9.90603566e-02 6.45243704e-01
8.63736331e-01 -1.54349089e-01 9.70117822e-02 -3.60231757e-01
2.76648790e-01 5.03644720e-02 -1.36348999e+00 1.90425920e+00
-1.93965539e-01 4.68622178e-01 3.71612795e-02 -4.33663160e-01
8.68696988e-01 -2.05642879e-01 4.33782130e-01 -1.22080080e-01
1.21654570e-01 2.91818827e-01 -1.81922108e-01 -8.36449936e-02
7.17850208e-01 -3.55433434e-01 -9.48272496e-02 4.15752009e-02
4.04482991e-01 -4.09389883e-01 3.74135636e-02 3.69694322e-01
9.80099082e-01 7.15147614e-01 1.60074577e-01 -4.62714046e-01
-2.56135672e-01 7.69272447e-02 7.95471191e-01 7.72625804e-01
-2.58766413e-01 1.12937009e+00 5.70428908e-01 7.16625378e-02
-1.00882471e+00 -1.58335364e+00 -2.16576949e-01 4.10073429e-01
3.47763151e-01 -6.00993454e-01 -8.62221718e-01 -3.53301913e-01
2.43364155e-01 8.71151924e-01 -4.39772934e-01 -2.24621043e-01
-2.83398092e-01 -8.15571427e-01 3.82091463e-01 5.43439865e-01
3.74280423e-01 -2.85617918e-01 -5.30433059e-01 2.19617739e-01
-4.08022076e-01 -1.12846446e+00 -4.36841041e-01 1.04890969e-02
-8.80161703e-01 -1.07576132e+00 -8.64311218e-01 -2.08721295e-01
5.77804446e-01 -1.15161449e-01 1.32798684e+00 -2.49659792e-01
-2.37063393e-01 3.43528241e-01 -5.69938980e-02 -3.14357668e-01
-3.99240971e-01 1.05389521e-01 6.99899569e-02 -4.17055547e-01
2.27300406e-01 -6.66460991e-01 -6.01801991e-01 4.05684412e-01
-6.14312470e-01 8.39725733e-02 6.00940704e-01 8.07862341e-01
7.61585355e-01 -1.61571965e-01 3.69560838e-01 -6.45394623e-01
4.22314882e-01 -4.06700999e-01 -4.47094917e-01 -6.94169151e-03
-7.12354004e-01 5.17010212e-01 -3.77279259e-02 -5.61628878e-01
-1.24934757e+00 1.45194575e-01 -1.09904818e-01 -7.15144992e-01
-6.52326345e-02 3.65123540e-01 -2.23328352e-01 1.09765738e-01
7.67860234e-01 -3.53423327e-01 3.78703386e-01 -5.61514795e-01
4.52945113e-01 3.60390812e-01 8.08951139e-01 -9.55958843e-01
6.87820733e-01 4.09394503e-01 1.36498094e-01 -7.38682985e-01
-8.88601363e-01 -3.10743541e-01 -6.06952548e-01 -5.01665175e-01
6.85334384e-01 -1.03093708e+00 -3.88037205e-01 5.30424416e-01
-1.04821789e+00 -2.06285298e-01 -2.09974006e-01 7.37576902e-01
-6.48925841e-01 5.87901235e-01 -5.76343715e-01 -9.87891197e-01
1.50715619e-01 -1.22836626e+00 1.51348948e+00 1.57145381e-01
-6.10172331e-01 -1.13116515e+00 3.71633351e-01 2.28292242e-01
2.03735933e-01 4.97665524e-01 7.32077539e-01 -6.56503364e-02
-5.06820202e-01 2.35632877e-03 -1.06183574e-01 2.16109455e-01
9.14498940e-02 1.67833537e-01 -9.27158594e-01 -2.49819696e-01
-3.33307743e-01 -2.37162739e-01 8.20674658e-01 6.37728393e-01
5.71280420e-01 1.34515733e-01 -4.73739237e-01 3.66699725e-01
1.21529365e+00 -3.72028708e-01 8.59199405e-01 2.39069968e-01
7.08205342e-01 5.79346716e-01 7.75329113e-01 4.50606883e-01
6.22866988e-01 7.47472584e-01 6.56221688e-01 2.28043646e-01
-4.24301326e-01 -6.97392225e-01 2.14672893e-01 4.47018981e-01
7.02157095e-02 -4.20134604e-01 -9.13112462e-01 5.49873412e-01
-2.04130220e+00 -6.70276761e-01 -6.14809282e-02 2.30659676e+00
8.91337335e-01 3.46946359e-01 -8.96586627e-02 -1.48025885e-01
7.08665788e-01 1.75737411e-01 -5.53380966e-01 7.99480677e-02
1.12015367e-01 1.86060682e-01 6.76835299e-01 9.94935095e-01
-8.64059806e-01 8.24325800e-01 6.82133102e+00 9.92560863e-01
-4.84553665e-01 -1.34200782e-01 4.47279155e-01 -6.34350181e-02
-4.19191927e-01 1.99737981e-01 -9.91232336e-01 4.31531549e-01
8.21637094e-01 -5.78487338e-03 1.38738811e-01 8.84495258e-01
5.43314703e-02 -7.39489257e-01 -1.03088653e+00 8.71576846e-01
1.15457639e-01 -1.03425205e+00 1.90450609e-01 3.67349267e-01
5.24626791e-01 -8.52181315e-02 -1.98672563e-01 1.96810171e-01
5.02133548e-01 -9.40087318e-01 1.01853430e+00 6.03432655e-01
6.21978045e-01 -7.81636536e-01 6.01094127e-01 4.05558646e-01
-1.08489931e+00 5.31665087e-01 -4.59787250e-01 -5.05207814e-02
7.35303402e-01 9.47968781e-01 -8.85983884e-01 5.52569807e-01
5.61589777e-01 6.38331890e-01 -2.54947454e-01 1.23403013e+00
-3.65592331e-01 3.67516547e-01 -7.32501626e-01 2.80190617e-01
-1.34336710e-01 -1.56817529e-02 6.36296272e-01 8.50786150e-01
5.25245488e-01 -3.31470132e-01 2.84924477e-01 9.92094815e-01
1.79624036e-01 -4.97447848e-01 -2.70997673e-01 1.76973462e-01
6.47030056e-01 7.16590762e-01 -6.34142816e-01 -4.33505662e-02
-4.87558134e-02 9.95988548e-01 4.48420733e-01 3.32857013e-01
-8.95755947e-01 6.95691770e-03 9.88484442e-01 2.01471671e-01
8.86033252e-02 -4.35077041e-01 -2.16315642e-01 -1.22349894e+00
-1.63640194e-02 -4.89446938e-01 1.61229685e-01 -7.09095657e-01
-1.48635840e+00 3.57172221e-01 6.40584767e-01 -1.13724589e+00
-5.11564791e-01 -6.43836975e-01 -1.55650064e-01 7.73110926e-01
-1.20068169e+00 -9.05565500e-01 -2.15868786e-01 1.59468189e-01
4.89643216e-01 4.35969621e-01 7.08621383e-01 4.62109625e-01
-4.66519266e-01 4.46760088e-01 -2.80122221e-01 -2.87412554e-01
9.55067158e-01 -1.35487139e+00 4.21504229e-01 9.66990471e-01
7.66196698e-02 5.57912767e-01 1.14072621e+00 -1.12237036e+00
-1.17517233e+00 -1.01654351e+00 5.01485646e-01 -5.54316521e-01
4.70940977e-01 -2.67375022e-01 -9.29616868e-01 6.49687350e-01
-2.13940397e-01 1.04920961e-01 2.78763384e-01 1.10828660e-01
-2.02971369e-01 3.25979471e-01 -1.17373908e+00 4.68565851e-01
1.27929938e+00 -2.50691682e-01 -6.63386285e-01 1.37407154e-01
7.13043928e-01 -9.00904238e-01 -9.96287346e-01 5.17487526e-01
7.25163698e-01 -9.93667960e-01 1.00898623e+00 -8.01979527e-02
3.83162558e-01 -2.94283152e-01 -3.52934390e-01 -1.35415053e+00
-7.52164498e-02 -3.83856773e-01 -3.18902999e-01 1.17127371e+00
3.03585529e-01 -5.27452052e-01 1.01473117e+00 9.21592772e-01
-1.71641603e-01 -6.56213343e-01 -1.22853875e+00 -9.62800264e-01
1.36819452e-01 -6.56161070e-01 1.80006742e-01 4.38020438e-01
-3.30242515e-01 2.81236172e-01 -6.30364478e-01 3.47533405e-01
1.12481868e+00 -3.42621654e-01 1.05072320e+00 -1.28794038e+00
-3.35933208e-01 -5.32803833e-02 -5.45591056e-01 -1.28287649e+00
1.64155900e-01 -3.14422131e-01 4.91992444e-01 -1.85708058e+00
9.84134302e-02 -4.30718720e-01 3.54129642e-01 1.92979798e-01
-1.53336644e-01 1.08888298e-01 6.90231249e-02 1.51656821e-01
-2.73090541e-01 6.95455253e-01 1.28146291e+00 1.01262033e-01
-8.31222236e-02 -2.39685252e-01 -4.22971159e-01 8.52138937e-01
5.18033326e-01 -5.75893342e-01 -6.46853805e-01 -5.56149900e-01
5.12511060e-02 -1.61183514e-02 4.59390730e-01 -1.22253501e+00
8.05015340e-02 -2.72515208e-01 4.22525197e-01 -8.47196043e-01
7.68142462e-01 -6.61787689e-01 4.51582432e-01 3.25729996e-01
1.65317774e-01 -2.11390212e-01 2.74825364e-01 9.65990007e-01
-8.52543861e-02 -2.42242962e-01 6.25743806e-01 -1.87550858e-02
-7.74292231e-01 1.29275575e-01 -3.47981244e-01 2.87313938e-01
8.13918591e-01 -2.81226486e-01 -3.69918913e-01 -4.91760254e-01
-5.99183977e-01 2.21268043e-01 8.16661358e-01 2.33813241e-01
5.35514355e-01 -1.30136311e+00 -4.27881449e-01 -4.01372612e-02
1.14802802e-02 5.24937451e-01 1.99052557e-01 6.15935564e-01
-6.52364433e-01 -1.21364161e-01 7.11237416e-02 -1.03547788e+00
-8.18934679e-01 -1.69183969e-01 4.05038953e-01 -1.70268312e-01
-6.83531344e-01 8.07816565e-01 7.99763873e-02 -5.02027929e-01
3.09769541e-01 -4.55786586e-01 3.23310852e-01 -1.20507628e-01
1.85423389e-01 3.93104881e-01 -2.01493531e-01 -6.10749662e-01
-4.69940186e-01 7.80635595e-01 1.56912819e-01 -4.73787785e-01
9.36620533e-01 -2.58239120e-01 1.64342418e-01 3.72196406e-01
1.01185942e+00 3.66429538e-02 -1.78447258e+00 4.04923297e-02
-1.77264005e-01 -8.93096685e-01 9.70764458e-02 -5.73020101e-01
-8.82823586e-01 7.51329720e-01 3.52707624e-01 -5.01612186e-01
5.84179401e-01 1.24035023e-01 5.78503191e-01 1.38643250e-01
8.14008653e-01 -9.49189663e-01 1.49377361e-01 4.17875648e-01
7.56858468e-01 -1.06628811e+00 3.86781901e-01 -8.09599996e-01
-6.54290438e-01 9.16962028e-01 9.90224898e-01 -1.53365746e-01
6.84103966e-01 5.63466489e-01 -6.14142492e-02 -6.62281644e-03
-5.65814972e-01 2.90024951e-02 3.71615291e-01 8.22227716e-01
1.74060151e-01 -4.13989499e-02 6.15458265e-02 2.61645079e-01
-5.28769970e-01 -7.42560402e-02 5.20835400e-01 1.11919534e+00
-5.36276519e-01 -1.09212792e+00 -5.13545036e-01 4.18850780e-01
-8.58202428e-02 2.74305344e-01 3.76563519e-02 1.03560197e+00
-2.45846435e-01 8.26097131e-01 2.49710605e-01 -5.34507871e-01
2.82516837e-01 -1.69455811e-01 7.11407363e-01 -6.52212083e-01
3.24087948e-01 3.25542003e-01 3.47619236e-01 -7.32103527e-01
-5.32260358e-01 -4.89230752e-01 -1.30606210e+00 -4.42079753e-01
-5.72620869e-01 -1.40054703e-01 4.05769557e-01 9.56571519e-01
4.24772918e-01 5.46224475e-01 -1.52684271e-01 -1.20274997e+00
-6.75048590e-01 -1.05824614e+00 -5.93482137e-01 3.17299396e-01
1.76757887e-01 -1.37705147e+00 -6.78058565e-01 -1.10228278e-01] | [7.115455150604248, -1.0833992958068848] |
dfae93c0-4a22-4b2d-8c3d-07642fa5d461 | nlp-cuet-lt-edi-eacl2021-multilingual-code | 2103.00464 | null | https://arxiv.org/abs/2103.00464v1 | https://arxiv.org/pdf/2103.00464v1.pdf | NLP-CUET@LT-EDI-EACL2021: Multilingual Code-Mixed Hope Speech Detection using Cross-lingual Representation Learner | In recent years, several systems have been developed to regulate the spread of negativity and eliminate aggressive, offensive or abusive contents from the online platforms. Nevertheless, a limited number of researches carried out to identify positive, encouraging and supportive contents. In this work, our goal is to identify whether a social media post/comment contains hope speech or not. We propose three distinct models to identify hope speech in English, Tamil and Malayalam language to serve this purpose. To attain this goal, we employed various machine learning (support vector machine, logistic regression, ensemble), deep learning (convolutional neural network + long short term memory) and transformer (m-BERT, Indic-BERT, XLNet, XLM-Roberta) based methods. Results indicate that XLM-Roberta outdoes all other techniques by gaining a weighted $f_1$-score of $0.93$, $0.60$ and $0.85$ respectively for English, Tamil and Malayalam language. Our team has achieved $1^{st}$, $2^{nd}$ and $1^{st}$ rank in these three tasks respectively. | ['Mohammed Moshiul Hoque', 'Omar Sharif', 'Eftekhar Hossain'] | 2021-02-28 | null | https://aclanthology.org/2021.ltedi-1.25 | https://aclanthology.org/2021.ltedi-1.25.pdf | eacl-ltedi-2021-4 | ['multilingual-text-classification', 'hope-speech-detection'] | ['miscellaneous', 'natural-language-processing'] | [-3.85861546e-01 9.17819068e-02 -2.89014071e-01 -2.02967495e-01
-3.19212228e-01 -3.44612658e-01 9.26179886e-01 3.56650054e-01
-4.58522141e-01 1.01315892e+00 5.18709064e-01 -5.19456446e-01
-3.38149279e-01 -7.02917516e-01 -1.75018087e-01 -3.04836661e-01
-2.04161808e-01 -2.96535529e-02 -2.73550600e-01 -9.05355275e-01
9.16796803e-01 3.03788483e-01 -1.14593077e+00 3.43500972e-01
7.65616894e-01 8.48488867e-01 -4.89340156e-01 5.98268747e-01
-2.86708146e-01 1.44134593e+00 -5.10859191e-01 -7.90363908e-01
-1.15190238e-01 -2.89746940e-01 -8.90207171e-01 -5.63524604e-01
-1.20101929e-01 -1.31238624e-01 -1.64712891e-01 9.68463838e-01
5.55111229e-01 2.21329078e-01 7.55974770e-01 -1.18538761e+00
-7.29201555e-01 9.71484303e-01 -8.89870167e-01 5.65773070e-01
3.62549573e-01 -1.48842365e-01 9.97682750e-01 -7.81740904e-01
4.26402569e-01 1.10544491e+00 6.67111039e-01 2.93743551e-01
-6.11605048e-01 -1.08696330e+00 -1.10677615e-01 -1.74184904e-01
-1.10143912e+00 -4.30411845e-01 1.15453959e+00 -5.01610518e-01
1.08552635e+00 2.07865596e-01 5.69198787e-01 1.21858644e+00
3.87210160e-01 6.34557307e-01 1.45622444e+00 -3.69185507e-01
-1.50355086e-01 6.47714853e-01 5.04407704e-01 7.33179033e-01
-2.45619625e-01 -3.26719992e-02 -8.60745966e-01 -2.76512474e-01
2.39376694e-01 -2.95847714e-01 1.16328768e-01 7.24408090e-01
-5.65552115e-01 1.40646601e+00 3.46766233e-01 6.70138657e-01
-5.39008379e-01 -2.54537314e-01 6.49842441e-01 6.41982675e-01
7.60995805e-01 4.34964687e-01 -3.95857185e-01 -5.99886835e-01
-7.97471225e-01 7.54370391e-02 8.08284402e-01 3.23772401e-01
4.15877551e-01 3.12372774e-01 -1.57219323e-03 1.11725509e+00
3.20563018e-01 2.92103320e-01 7.08823442e-01 -4.30872947e-01
2.49745265e-01 6.24302149e-01 5.60018271e-02 -1.47815418e+00
-5.45241654e-01 -6.07337296e-01 -9.26540256e-01 1.16733760e-01
-6.50713295e-02 -4.85155851e-01 -5.63822865e-01 1.60508370e+00
1.31959952e-02 -2.97094226e-01 9.46657136e-02 3.62147152e-01
1.05506444e+00 8.80701244e-01 3.20464998e-01 -4.62702572e-01
9.68992770e-01 -8.73856544e-01 -6.47275209e-01 -2.67093867e-01
6.52174532e-01 -1.19702363e+00 8.42834949e-01 6.64366901e-01
-1.17459631e+00 -2.85031974e-01 -9.95613098e-01 1.48880914e-01
-6.13049984e-01 -1.73280701e-01 7.86134362e-01 9.16438103e-01
-9.37403619e-01 5.79286695e-01 -2.09472463e-01 -2.78129905e-01
5.46140313e-01 5.69341123e-01 -3.88415873e-01 5.45494914e-01
-1.40691888e+00 1.00427616e+00 9.44226421e-03 -1.18885055e-01
-6.17320061e-01 -2.54716009e-01 -4.70050097e-01 -1.43602967e-01
-2.47216225e-02 6.40262142e-02 8.16003621e-01 -1.25545824e+00
-1.37364566e+00 1.18937826e+00 6.69100210e-02 -6.17520154e-01
1.60360545e-01 -2.13019878e-01 -7.16192007e-01 -2.12567002e-01
-8.35177302e-03 2.95831919e-01 6.60179734e-01 -7.76542366e-01
-6.36073709e-01 -5.58131874e-01 1.72547281e-01 2.93801595e-02
-7.78655648e-01 7.17308342e-01 5.96511841e-01 -6.55539811e-01
-5.01762740e-02 -7.37354815e-01 1.88092515e-01 -7.38974094e-01
-5.88176250e-01 -6.29430652e-01 7.10414827e-01 -9.32946205e-01
1.48432636e+00 -1.94678020e+00 -1.45632759e-01 4.15771365e-01
3.66872311e-01 5.55875361e-01 3.29100728e-01 7.37195790e-01
-4.85164084e-04 5.98180175e-01 3.01594108e-01 -5.57641052e-02
1.10914335e-02 2.13603452e-02 -4.82021905e-02 4.96667713e-01
-1.08072251e-01 5.50204933e-01 -4.99368429e-01 -5.75703979e-01
7.93578103e-02 4.48456198e-01 -4.26927447e-01 1.61443688e-02
2.82988757e-01 3.05496249e-02 -5.59227765e-01 7.21957147e-01
4.72802490e-01 -5.38434945e-02 7.85889998e-02 2.72730052e-01
-4.72403884e-01 2.90431768e-01 -8.07910562e-01 8.91171515e-01
-3.39049906e-01 7.16963589e-01 3.01294237e-01 -1.19379151e+00
1.49463046e+00 3.05576652e-01 3.77297461e-01 -7.76781082e-01
7.96755433e-01 3.28102797e-01 2.35248148e-01 -6.72269702e-01
7.68394411e-01 -5.56128144e-01 -1.81727633e-01 4.65464652e-01
-8.74785036e-02 3.27891111e-01 -7.19924569e-02 3.84010702e-01
9.37736273e-01 -3.93303484e-01 4.54381287e-01 -3.09201151e-01
8.41320336e-01 -3.18528354e-01 2.76169956e-01 6.41181350e-01
-8.19538713e-01 -2.47782003e-02 7.84373939e-01 -2.75847703e-01
-7.07904577e-01 -6.83509409e-01 9.81840789e-02 1.73108542e+00
-2.38539398e-01 -1.47200808e-01 -5.10932803e-01 -5.29524922e-01
-3.60571504e-01 9.82427716e-01 -5.81891000e-01 -2.60386735e-01
-5.99119425e-01 -5.99200428e-01 7.39155173e-01 5.94493225e-02
7.16075540e-01 -1.23846149e+00 -2.86020637e-01 1.32331073e-01
-1.06770806e-01 -4.06738728e-01 5.59109785e-02 2.48533770e-01
-5.11068821e-01 -6.14103138e-01 -4.26080555e-01 -6.80937409e-01
1.38962641e-01 -1.83310866e-01 8.75099540e-01 1.15287267e-01
8.64101350e-02 -1.73894539e-01 -5.79805315e-01 -6.54558539e-01
-3.03335398e-01 2.58869320e-01 8.93116370e-03 -5.62333353e-02
7.26159573e-01 -6.70314968e-01 -3.58000487e-01 -2.41133288e-01
-5.90146005e-01 -2.23511800e-01 4.45672154e-01 7.64649153e-01
-3.29457402e-01 -5.67110665e-02 9.75411892e-01 -9.71117020e-01
1.21180975e+00 -9.51491535e-01 -1.05650462e-02 -3.21657300e-01
-7.11096346e-01 -3.05166543e-01 5.68407536e-01 -3.70577902e-01
-9.98337567e-01 -3.55496496e-01 -7.91398108e-01 1.39913663e-01
7.11689815e-02 8.98617446e-01 4.18682128e-01 1.10374562e-01
9.38411772e-01 2.65327960e-01 -6.68292865e-02 -3.89243066e-01
-4.65269797e-02 1.18666637e+00 1.44758159e-02 5.66272140e-02
6.67301416e-01 2.60570347e-01 -4.83153850e-01 -9.18779373e-01
-8.90957177e-01 -4.43255067e-01 -3.18882108e-01 -6.13290727e-01
7.66108990e-01 -6.92726254e-01 -8.10633957e-01 4.69619125e-01
-9.20881093e-01 1.76775753e-01 3.57477158e-01 4.54367012e-01
5.74507304e-02 8.37262198e-02 -6.56356096e-01 -1.37817121e+00
-9.96657312e-01 -6.72552884e-01 2.00500160e-01 5.21962345e-01
-7.28313088e-01 -8.76487315e-01 -1.04708128e-01 7.50100017e-01
8.52276504e-01 3.66519630e-01 8.40832531e-01 -1.20838702e+00
3.06173354e-01 -4.21234369e-01 -1.31287932e-01 5.30447543e-01
-8.77090320e-02 2.06926107e-01 -8.28224480e-01 1.11835375e-01
2.99647242e-01 -6.71978533e-01 7.22508132e-01 5.23663878e-01
5.42228639e-01 -6.15486324e-01 -2.25082375e-02 -8.24913569e-03
1.02357471e+00 6.17985547e-01 8.11862409e-01 5.42219579e-01
1.17636912e-01 5.63775659e-01 5.09954333e-01 8.04652929e-01
1.13017090e-01 -1.70131400e-02 4.36259568e-01 8.73062834e-02
4.18269962e-01 -3.46866459e-01 7.12750196e-01 7.01542020e-01
4.26678620e-02 -1.98679417e-01 -8.50777090e-01 6.37813985e-01
-1.37553537e+00 -1.22658122e+00 -3.73468816e-01 1.72080910e+00
8.98236334e-01 7.10569501e-01 3.23645651e-01 5.92027545e-01
6.16444468e-01 5.38351893e-01 -1.16006672e-01 -1.20747674e+00
-1.81640044e-01 4.60574061e-01 1.74028695e-01 5.65685213e-01
-1.21952879e+00 9.66642141e-01 4.93429184e+00 9.67739522e-01
-1.48388648e+00 4.09787297e-02 1.07870734e+00 -2.06571877e-01
-1.95697114e-01 -1.39308080e-01 -6.18353605e-01 3.61184508e-01
1.27130079e+00 -4.21111286e-01 2.92847008e-01 9.95805502e-01
4.88296121e-01 -1.25881582e-01 -3.80304813e-01 8.84983659e-01
3.11108828e-01 -1.46031356e+00 -4.51816618e-01 -3.67639437e-02
6.16526604e-01 3.04519266e-01 4.48148787e-01 7.87735701e-01
3.50553453e-01 -1.30451071e+00 4.62387979e-01 3.50184530e-01
5.45270704e-02 -1.01863551e+00 8.47214997e-01 7.29216337e-01
-5.38935959e-01 -2.64087588e-01 6.89719245e-03 -7.81786561e-01
3.94570455e-02 5.40972173e-01 -7.11771011e-01 1.35902092e-01
7.96083450e-01 5.46027422e-01 -4.06844646e-01 2.84857422e-01
-1.54018655e-01 9.16153073e-01 -2.29898527e-01 -8.24191213e-01
6.94983840e-01 -1.70620997e-02 5.06487072e-01 1.20425594e+00
7.54934028e-02 2.84271985e-01 -5.09642586e-02 5.12005746e-01
-2.01279536e-01 7.03898907e-01 -6.44648790e-01 -3.79942089e-01
8.87313336e-02 1.34393096e+00 -5.46614349e-01 -2.59042889e-01
7.00250035e-03 4.06276613e-01 2.25262791e-01 -1.65294021e-01
-9.69994783e-01 -6.36109054e-01 1.31574973e-01 2.87452996e-01
-2.36018226e-01 -5.04845427e-03 -5.59132099e-01 -8.14710557e-01
-4.54397976e-01 -9.26537156e-01 4.84520167e-01 -6.59078956e-01
-1.22979808e+00 6.90437913e-01 -2.21724257e-01 -6.00078881e-01
-2.21088082e-01 -3.96938473e-01 -6.42881513e-01 8.87113810e-01
-1.17442036e+00 -1.09738111e+00 6.95356205e-02 5.06603956e-01
4.61131334e-01 -6.13568366e-01 5.16113639e-01 3.62719476e-01
-6.12316191e-01 4.75742817e-01 9.73758637e-04 3.64220977e-01
4.68368322e-01 -7.24331796e-01 -5.19409001e-01 4.22980756e-01
-2.17581511e-01 8.59739304e-01 9.14806545e-01 -4.88202244e-01
-1.12895894e+00 -6.75548851e-01 1.51233518e+00 -1.66973636e-01
8.64354551e-01 -2.88208853e-03 -4.79791760e-01 4.55328584e-01
6.39689386e-01 -5.97080052e-01 8.85046840e-01 3.33430588e-01
-2.34363362e-01 -1.75752103e-01 -1.39894640e+00 5.97883582e-01
5.26336432e-01 -3.52032810e-01 -5.32839417e-01 5.03790677e-01
2.48845309e-01 2.14224607e-01 -8.89972627e-01 4.34841253e-02
7.45744646e-01 -1.41196990e+00 6.74202621e-01 -7.73238242e-01
1.11870992e+00 4.10883665e-01 -1.81185946e-01 -8.76977861e-01
-2.76794285e-01 -6.71326101e-01 3.14895630e-01 1.48023748e+00
7.88153768e-01 -6.70870781e-01 8.28583241e-01 5.15482843e-01
-1.27146095e-01 -9.09830153e-01 -1.03951955e+00 -4.92158011e-02
5.45311153e-01 -3.14459920e-01 -9.55987126e-02 1.22058094e+00
2.50390112e-01 8.25576484e-01 -9.60134983e-01 -4.50527906e-01
4.40287553e-02 -3.13858747e-01 6.09368920e-01 -1.22693098e+00
8.65234360e-02 -6.85171545e-01 -1.17550604e-01 -5.58469534e-01
1.87314808e-01 -7.12751925e-01 -6.44501567e-01 -1.34027994e+00
1.51309028e-01 -2.79469281e-01 -5.64376354e-01 5.20491600e-01
3.97563457e-01 2.12703362e-01 -5.01681603e-02 -6.48118183e-02
-3.09090734e-01 3.78693581e-01 7.74183393e-01 -1.02474108e-01
-2.67327040e-01 -9.20805242e-03 -1.29525805e+00 9.24163818e-01
1.28297365e+00 -4.97645676e-01 -2.57807463e-01 6.58315122e-02
5.73105216e-01 1.31935775e-02 6.13267273e-02 -4.67828810e-01
3.18664238e-02 -4.88637239e-01 2.25720987e-01 -6.51910245e-01
5.40274262e-01 -1.81991994e-01 -2.74562865e-01 6.91626906e-01
-6.33893907e-01 1.81572318e-01 -8.60524550e-02 2.37486884e-01
-3.64110053e-01 -5.28305948e-01 1.05633485e+00 -3.52852970e-01
-3.32603127e-01 -1.43475518e-01 -6.04036570e-01 1.57918911e-02
9.88942325e-01 -2.53717303e-01 -4.76731449e-01 -7.49513149e-01
-5.29320776e-01 -9.41302255e-02 -2.02597275e-01 4.62538630e-01
6.65331781e-01 -8.58293056e-01 -8.39775026e-01 7.45195802e-03
-2.17033923e-01 -9.13273513e-01 3.16490024e-01 1.20072567e+00
-5.00158548e-01 3.63067806e-01 -3.13963443e-01 6.80191293e-02
-1.49569631e+00 1.41298622e-01 1.72800392e-01 -4.04400915e-01
8.97226930e-02 1.32089388e+00 -5.17943382e-01 -4.09350872e-01
7.66444486e-03 4.11286712e-01 -8.81804943e-01 3.28676790e-01
3.52923125e-01 5.59111297e-01 -1.41320914e-01 -1.09858084e+00
-3.48776132e-01 -5.28642386e-02 -3.47516149e-01 -2.30044186e-01
1.45176721e+00 1.01074189e-01 -3.96153092e-01 4.87219185e-01
1.38598371e+00 3.54686230e-01 -6.85655028e-02 1.56215861e-01
7.57531896e-02 -3.10826391e-01 2.88801521e-01 -9.69576180e-01
-8.71230662e-01 7.64980972e-01 5.11950970e-01 8.18589211e-01
9.25086498e-01 -2.22044513e-01 7.41367936e-01 8.88686255e-02
-1.56430483e-01 -1.56842971e+00 1.94728181e-01 8.34723592e-01
9.04377699e-01 -1.31609404e+00 -5.43906316e-02 1.31167844e-01
-7.71075189e-01 1.02284014e+00 5.43816984e-01 -2.61995792e-01
1.03448856e+00 6.30428717e-02 7.80756818e-04 -2.57010639e-01
-8.63012493e-01 1.82617143e-01 -1.76493019e-01 4.19241160e-01
1.09800816e+00 -2.49534938e-02 -1.09502172e+00 7.07437634e-01
-6.59041345e-01 7.75532797e-02 6.71641052e-01 9.51190829e-01
-8.06528330e-01 -7.46277571e-01 -2.10577846e-01 9.45071399e-01
-1.20928752e+00 -2.45500177e-01 -8.51500988e-01 8.74159753e-01
1.10494100e-01 1.41445971e+00 -3.13549697e-01 -8.02128434e-01
-3.91475623e-03 1.39000535e-01 -3.43642205e-01 -1.82025567e-01
-1.26382399e+00 2.71045923e-01 7.39057720e-01 1.23419076e-01
-5.08659005e-01 -4.14685965e-01 -1.11204338e+00 -9.37697172e-01
-3.78220081e-01 3.96108329e-01 7.85286665e-01 9.69209552e-01
3.65027040e-02 1.32545263e-01 6.90639675e-01 -2.14579344e-01
-6.07711673e-01 -1.36498344e+00 -6.45814002e-01 3.22355896e-01
4.66401828e-03 -5.60172439e-01 -5.50758541e-01 -3.61593753e-01] | [8.926568031311035, 10.669601440429688] |
d27b3655-9403-477a-b274-b7f38e621feb | time-matters-multi-scale-temporalization-of | 1801.05853 | null | http://arxiv.org/abs/1801.05853v1 | http://arxiv.org/pdf/1801.05853v1.pdf | Time Matters: Multi-scale Temporalization of Social Media Popularity | The evolution of social media popularity exhibits rich temporality, i.e.,
popularities change over time at various levels of temporal granularity. This
is influenced by temporal variations of public attentions or user activities.
For example, popularity patterns of street snap on Flickr are observed to
depict distinctive fashion styles at specific time scales, such as season-based
periodic fluctuations for Trench Coat or one-off peak in days for Evening
Dress. However, this fact is often overlooked by existing research of
popularity modeling. We present the first study to incorporate multiple
time-scale dynamics into predicting online popularity. We propose a novel
computational framework in the paper, named Multi-scale Temporalization, for
estimating popularity based on multi-scale decomposition and structural
reconstruction in a tensor space of user, post, and time by joint low-rank
constraints. By considering the noise caused by context inconsistency, we
design a data rearrangement step based on context aggregation as preprocessing
to enhance contextual relevance of neighboring data in the tensor space. As a
result, our approach can leverage multiple levels of temporal characteristics
and reduce the noise of data decomposition to improve modeling effectiveness.
We evaluate our approach on two large-scale Flickr image datasets with over 1.8
million photos in total, for the task of popularity prediction. The results
show that our approach significantly outperforms state-of-the-art popularity
prediction techniques, with a relative improvement of 10.9%-47.5% in terms of
prediction accuracy. | ['Wen-Huang Cheng', 'Yongdong Zhang', 'Bo Wu', 'Tao Mei'] | 2017-12-12 | null | null | null | null | ['social-media-popularity-prediction', 'social-media-popularity-prediction'] | ['miscellaneous', 'time-series'] | [-3.55938524e-01 -9.43804801e-01 -6.14532292e-01 -1.10099159e-01
-3.17941785e-01 -4.47898597e-01 4.24639106e-01 2.82510400e-01
-9.59972143e-02 4.06903803e-01 6.13182127e-01 9.32141766e-02
-2.31503308e-01 -6.84554458e-01 -6.68538392e-01 -5.90392768e-01
-2.73839623e-01 1.74504146e-01 2.73349077e-01 -1.60064548e-01
3.25963795e-01 1.20067686e-01 -1.58370888e+00 4.89147007e-01
7.42259979e-01 1.03034616e+00 1.54027387e-01 2.04454049e-01
-1.04614124e-01 8.61031532e-01 2.18094531e-02 -3.39335382e-01
1.93812698e-01 1.35980099e-01 -4.85235006e-01 3.91749233e-01
4.69391793e-01 -3.04715008e-01 -8.14862549e-01 8.75165224e-01
7.53549561e-02 8.76841396e-02 2.42133468e-01 -1.44071329e+00
-5.82277596e-01 8.43276441e-01 -1.11137378e+00 3.53636980e-01
3.80143642e-01 -6.23257421e-02 1.42084324e+00 -7.85599291e-01
6.48331523e-01 8.93073261e-01 6.82104111e-01 -3.00058305e-01
-1.37566757e+00 -7.40449965e-01 5.43992519e-01 7.04453409e-01
-1.72019529e+00 -1.16599530e-01 1.14721596e+00 -6.42065227e-01
4.88308340e-01 3.76542836e-01 1.11809886e+00 1.15144086e+00
3.47553432e-01 8.71922731e-01 1.21212208e+00 2.20297053e-01
1.13165043e-02 -3.10823739e-01 4.19041924e-02 3.71803284e-01
2.87490635e-04 -5.65159917e-01 -8.28402758e-01 -6.74151957e-01
7.32066631e-01 3.77540439e-01 -4.93085869e-02 -3.31204981e-01
-1.39955974e+00 5.31368852e-01 2.73733377e-01 3.02956522e-01
-6.37929201e-01 6.23202957e-02 4.66155916e-01 3.73758644e-01
8.55723083e-01 1.23577245e-01 -6.05471373e-01 -5.69208443e-01
-1.29927909e+00 3.21176678e-01 5.41919291e-01 9.26199377e-01
6.32986963e-01 -9.94400233e-02 3.35057303e-02 1.06570995e+00
2.11268961e-01 6.02522790e-01 4.78260368e-01 -7.21724510e-01
3.21456492e-01 3.84519458e-01 6.65641055e-02 -1.56610417e+00
-3.28387529e-01 -6.69889152e-01 -9.39866662e-01 -9.40113068e-01
4.09159839e-01 5.23646832e-01 -5.46252191e-01 1.60055852e+00
4.92672920e-01 8.45810890e-01 -7.48926461e-01 7.16469169e-01
4.98203397e-01 8.53104591e-01 1.51777431e-01 -6.02708459e-01
1.38111353e+00 -7.87865758e-01 -4.71285462e-01 4.92116183e-01
3.44708353e-01 -8.82234633e-01 9.89340007e-01 5.59762061e-01
-8.33788395e-01 -4.21982944e-01 -3.16373259e-01 1.04940705e-01
5.11906780e-02 -1.47718281e-01 5.93554199e-01 4.90973704e-02
-6.10500693e-01 6.49758637e-01 -7.34577298e-01 -4.35210675e-01
7.97557831e-02 7.24406242e-02 -1.23732038e-01 -1.71846449e-01
-1.06173718e+00 1.78986177e-01 -1.45167457e-02 -2.51105219e-01
-4.83898252e-01 -1.01401067e+00 -2.81391948e-01 -2.56446656e-02
3.94447833e-01 -4.87377673e-01 9.81936216e-01 -9.82440770e-01
-1.15170145e+00 3.97903353e-01 -2.25417420e-01 -3.32144588e-01
4.77919906e-01 -3.22363734e-01 -9.39359903e-01 4.20691371e-02
1.71169594e-01 1.98553905e-01 1.05095828e+00 -8.63055289e-01
-7.09324002e-01 -1.41050011e-01 2.29960028e-02 3.05978698e-03
-6.71815872e-01 -2.27891430e-02 -9.50550318e-01 -1.26885343e+00
4.26075697e-01 -1.10263586e+00 -2.17688859e-01 -2.13015467e-01
-1.85187146e-01 -1.05000593e-01 7.22554505e-01 -6.99190974e-01
1.98417318e+00 -2.25385904e+00 2.90583968e-01 4.41194803e-01
2.72217244e-01 -4.62468088e-01 -2.09968925e-01 7.52545178e-01
1.95985526e-01 -6.20045140e-03 2.20160738e-01 -1.28959462e-01
-3.51270102e-02 2.73092657e-01 -7.52312958e-01 6.40496969e-01
-4.13264066e-01 6.93642437e-01 -1.12775099e+00 -4.71721351e-01
9.72956419e-02 3.59118342e-01 -7.87491560e-01 -9.90697742e-02
-1.79545805e-01 5.33612728e-01 -5.62870800e-01 8.08989346e-01
6.67913914e-01 -7.87737727e-01 3.23808491e-01 -5.38727760e-01
-4.43607241e-01 3.00488174e-01 -1.23485327e+00 1.54551458e+00
-2.66705990e-01 3.17295313e-01 -1.16814993e-01 -6.25187874e-01
3.05440217e-01 2.44365335e-01 1.16662765e+00 -5.34181893e-01
1.49416840e-02 2.26779178e-01 -2.18439892e-01 -3.11653823e-01
1.04448092e+00 3.02824348e-01 -1.44049942e-01 4.47378188e-01
-3.05743665e-01 4.30608541e-01 2.02980950e-01 3.40622663e-01
1.08080256e+00 -2.02660546e-01 2.03602195e-01 -2.80503929e-01
2.49832049e-01 -1.26616061e-01 6.41679287e-01 5.58796465e-01
-4.78837788e-02 4.70795095e-01 4.59605217e-01 -5.46960354e-01
-1.34175694e+00 -8.15817893e-01 -2.82739550e-01 1.38605368e+00
2.88469434e-01 -9.88684595e-01 -8.03369731e-02 -3.03509325e-01
3.43767285e-01 2.07353219e-01 -5.79980254e-01 3.92109632e-01
-6.02019310e-01 -7.84482896e-01 -2.65050423e-03 1.30829677e-01
3.38576585e-01 -5.20912945e-01 2.62643576e-01 4.52544868e-01
-4.77718115e-01 -1.43767107e+00 -6.84948802e-01 -6.39178097e-01
-1.00040817e+00 -7.05502689e-01 -8.03720891e-01 -3.62218022e-01
5.39639115e-01 8.86354387e-01 1.11494911e+00 5.58056459e-02
7.32648447e-02 7.26316810e-01 -6.36266887e-01 4.54106122e-01
2.32239962e-01 1.64244249e-01 3.49564314e-01 5.19356906e-01
8.79779831e-02 -1.25001466e+00 -8.14002514e-01 7.71654427e-01
-9.19452667e-01 2.26185903e-01 4.54916000e-01 6.37438118e-01
8.09750080e-01 3.81627142e-01 8.07261541e-02 -5.21081030e-01
4.45427924e-01 -1.25421858e+00 -4.61662024e-01 -8.53694603e-02
-3.79572809e-01 -3.32390010e-01 7.52648711e-01 -9.21481609e-01
-4.25527811e-01 -1.84585124e-01 3.51080745e-01 -7.84790218e-01
3.16173673e-01 7.82734334e-01 4.04939711e-01 -3.40616405e-02
1.30317032e-01 5.12347400e-01 -4.85494912e-01 -7.51072347e-01
4.01501775e-01 4.82638776e-01 7.08931535e-02 -8.66204321e-01
9.88062620e-01 6.81987405e-01 -1.70205623e-01 -1.23365200e+00
-6.85733914e-01 -8.68239939e-01 -3.39571893e-01 -3.06113094e-01
2.81124055e-01 -1.43688822e+00 -4.04038340e-01 6.87192023e-01
-5.70566058e-01 -1.03221834e-02 -1.48546860e-01 4.51966256e-01
-1.55737400e-01 7.57312477e-01 -7.95242012e-01 -4.56814557e-01
3.84410173e-02 -8.39853346e-01 8.75322402e-01 -2.12412938e-01
-1.99002415e-01 -7.07313597e-01 1.28616512e-01 3.93270642e-01
5.20296156e-01 -1.46061540e-01 7.64328182e-01 -2.53623039e-01
-8.59038413e-01 -2.90188398e-02 -2.45989636e-01 -8.50226507e-02
1.09128326e-01 3.27408195e-01 -3.26921642e-01 -3.08019042e-01
-4.62227851e-01 -3.46834548e-02 6.08543754e-01 3.58719945e-01
1.62051892e+00 -6.13733828e-01 -2.15368077e-01 3.91878724e-01
1.20987499e+00 -4.83399898e-01 4.95014518e-01 2.80771464e-01
1.04035938e+00 2.60126740e-01 6.47866905e-01 1.13077152e+00
8.00367475e-01 1.05789948e+00 1.57093704e-01 5.14038801e-01
2.92908698e-01 -3.92906219e-01 4.00380731e-01 1.52535439e+00
-4.32412922e-01 -1.86971929e-02 -7.41126657e-01 7.99517930e-01
-2.12044477e+00 -1.24967134e+00 -2.67514795e-01 2.13427448e+00
6.39287412e-01 4.15424518e-02 6.26892626e-01 -5.67549951e-02
7.40282238e-01 5.90586185e-01 -4.11667466e-01 7.51621574e-02
3.36310007e-02 -2.46801049e-01 7.04530239e-01 1.73592344e-01
-1.10744870e+00 7.92247355e-01 5.60474443e+00 1.16649771e+00
-1.30369568e+00 1.79739952e-01 4.60854471e-01 -3.21072251e-01
-5.37150979e-01 -5.84341809e-02 -4.60237771e-01 9.22332764e-01
7.07242250e-01 -2.70478129e-01 7.59375036e-01 8.49058688e-01
5.29300272e-01 8.89083594e-02 -5.76544642e-01 1.23755419e+00
-1.33112773e-01 -1.07217741e+00 1.46869406e-01 3.49903613e-01
9.29979920e-01 3.02431345e-01 3.35786164e-01 1.83284611e-01
3.50133181e-02 -3.43628913e-01 8.04690301e-01 5.33407390e-01
3.84227157e-01 -4.42666709e-01 1.97512761e-01 2.46167243e-01
-1.74979115e+00 -2.19346210e-01 -2.04752132e-01 -1.72679946e-01
4.66630161e-01 1.20346093e+00 -2.52770662e-01 4.89981264e-01
7.53450871e-01 1.29429042e+00 -3.73962551e-01 1.18531466e+00
1.72820687e-01 9.45687652e-01 -5.17299771e-01 2.05119729e-01
1.48284346e-01 -3.12066644e-01 7.69366920e-01 9.44558263e-01
5.51770389e-01 1.88756183e-01 4.10396278e-01 2.08456188e-01
-1.17996901e-01 3.75900358e-01 -2.09055454e-01 -3.01169246e-01
5.27846932e-01 1.29485357e+00 -5.50715208e-01 -2.48688653e-01
-4.90451783e-01 8.84464085e-01 9.48555842e-02 2.48015001e-01
-1.34314835e+00 4.18194771e-01 7.92638659e-01 7.18642414e-01
6.46649003e-01 -7.56900787e-01 2.53392577e-01 -1.86365116e+00
2.09335506e-01 -8.32477093e-01 2.37911433e-01 -4.49848413e-01
-1.85384011e+00 1.87869847e-01 9.99158397e-02 -1.62528503e+00
3.70867878e-01 -1.52123556e-01 -3.48122656e-01 1.51210427e-01
-1.28578556e+00 -1.26572835e+00 -8.58148485e-02 7.61849463e-01
4.95520830e-01 1.36276886e-01 2.77269453e-01 8.18397880e-01
-5.43000400e-01 4.31635022e-01 4.58550632e-01 -2.98055381e-01
5.12274265e-01 -7.99939096e-01 2.92061180e-01 5.56429267e-01
2.93025017e-01 6.96503162e-01 8.00192058e-01 -8.16083789e-01
-1.78709364e+00 -1.12204170e+00 1.04675627e+00 -2.29660392e-01
1.53259730e+00 -1.55356616e-01 -6.66835427e-01 5.00342131e-01
-1.50784597e-01 2.11439177e-01 8.68697464e-01 5.29523909e-01
-7.31852412e-01 -2.97705054e-01 -7.43528068e-01 7.64027119e-01
1.27062309e+00 -6.42728209e-01 -2.49238253e-01 3.37807178e-01
4.58207875e-01 -1.72318622e-01 -1.25978720e+00 4.35647964e-01
1.19968545e+00 -5.48855543e-01 9.09599066e-01 -5.07231653e-01
5.17248034e-01 -4.72954571e-01 -5.31648099e-01 -9.18044090e-01
-8.86343181e-01 -7.55069911e-01 -5.00612795e-01 1.08411384e+00
1.15006596e-01 -3.35599154e-01 6.24063790e-01 4.16195661e-01
2.76582390e-01 -8.89828026e-01 -1.03921044e+00 -5.43222249e-01
-5.67263842e-01 -5.69262505e-01 5.12774765e-01 1.20240629e+00
-1.36316298e-02 -2.47537927e-03 -1.08722782e+00 7.97765553e-02
6.12646818e-01 5.02455950e-01 1.03115392e+00 -9.87949371e-01
-4.54188973e-01 -3.65223020e-01 -5.32804549e-01 -1.43893063e+00
-2.21771240e-01 -5.42675316e-01 -6.01467907e-01 -1.02356029e+00
5.64317822e-01 -5.72163820e-01 -7.06445336e-01 1.25802979e-01
7.61461928e-02 3.48924398e-01 2.22442210e-01 7.69795001e-01
-1.15435588e+00 6.22253895e-01 1.28284705e+00 -2.46991187e-01
-9.49104950e-02 -1.51387796e-01 -4.15188283e-01 7.00213075e-01
5.96762419e-01 -3.30286235e-01 -2.62253523e-01 -2.64069378e-01
6.87690020e-01 -8.18322748e-02 1.55771941e-01 -9.60091114e-01
6.88128173e-02 -4.81134951e-01 3.25634889e-02 -7.57584453e-01
4.17503983e-01 -9.63138700e-01 7.83381522e-01 1.14091054e-01
-7.45422617e-02 3.24233174e-01 -4.88448031e-02 1.00602198e+00
-2.34668389e-01 6.18577898e-01 3.13998014e-01 6.64317980e-02
-7.54423499e-01 7.82173753e-01 -5.01308143e-01 -8.70278776e-02
8.10558975e-01 -2.30263565e-02 -2.08394498e-01 -5.09781420e-01
-7.21735656e-01 1.72217399e-01 8.04625213e-01 6.76000297e-01
2.09731266e-01 -1.65471947e+00 -5.51054835e-01 -1.86119184e-01
1.81731939e-01 -7.31631458e-01 8.02167773e-01 1.33035862e+00
-2.26037726e-01 1.32765830e-01 -1.40005693e-01 -6.94991112e-01
-1.21675909e+00 5.49916923e-01 -4.07822758e-01 -5.43149590e-01
-5.91539145e-01 5.42197824e-01 7.29737803e-02 8.85274559e-02
-1.94249041e-02 -4.58632171e-01 -2.37193078e-01 3.91527712e-01
3.42178106e-01 5.29538333e-01 -2.84208596e-01 -1.15880597e+00
-2.20877931e-01 8.61007869e-01 -2.95151114e-01 2.92808026e-01
1.48541617e+00 -3.78064811e-01 -3.56178433e-01 7.65806973e-01
1.12968087e+00 4.86446500e-01 -8.76740634e-01 -5.19535720e-01
-1.73405319e-01 -8.62245202e-01 -7.21496046e-02 -1.88504323e-01
-1.22252774e+00 1.75128520e-01 2.81295180e-01 6.47509396e-01
1.17108738e+00 3.17922197e-02 1.26904130e+00 1.30521074e-01
7.84852326e-01 -1.13091671e+00 4.02398258e-02 5.95764875e-01
7.53996789e-01 -9.78939950e-01 4.62150991e-01 -6.02453947e-01
-6.24335408e-01 7.48794258e-01 2.60599762e-01 -3.06358248e-01
1.06147075e+00 -4.95018482e-01 -4.60140944e-01 -1.28918812e-01
-8.94437611e-01 6.55221492e-02 5.83454967e-01 -4.50593948e-01
2.63359845e-01 3.68345767e-01 -6.67370737e-01 4.80489612e-01
-1.47578239e-01 -3.66773717e-02 2.72991031e-01 8.31637323e-01
-2.99741149e-01 -1.33007836e+00 -1.81672990e-01 6.70883119e-01
-6.07307792e-01 -1.99366152e-01 9.62182879e-02 4.29718971e-01
1.62813514e-01 5.22901177e-01 3.46095301e-02 -7.56472111e-01
9.77330841e-03 -2.95696855e-01 2.49615416e-01 -2.88715482e-01
-4.12500501e-01 3.95415127e-01 1.29565045e-01 -8.20458531e-01
-6.23993218e-01 -1.03713870e+00 -4.20998394e-01 -7.73975194e-01
-2.70810694e-01 2.77643174e-01 3.56447816e-01 6.18224680e-01
7.72922993e-01 9.20850039e-02 9.65141058e-01 -8.97168636e-01
-1.98585659e-01 -6.92906320e-01 -9.44628060e-01 6.68640077e-01
1.36179969e-01 -7.94903696e-01 -4.43952173e-01 3.39370668e-02] | [9.635824203491211, 5.467423439025879] |
7d3b58f6-3b95-4c97-abe9-b7a8108c73e2 | xalign-cross-lingual-fact-to-text-alignment | 2202.00291 | null | https://arxiv.org/abs/2202.00291v2 | https://arxiv.org/pdf/2202.00291v2.pdf | XAlign: Cross-lingual Fact-to-Text Alignment and Generation for Low-Resource Languages | Multiple critical scenarios (like Wikipedia text generation given English Infoboxes) need automated generation of descriptive text in low resource (LR) languages from English fact triples. Previous work has focused on English fact-to-text (F2T) generation. To the best of our knowledge, there has been no previous attempt on cross-lingual alignment or generation for LR languages. Building an effective cross-lingual F2T (XF2T) system requires alignment between English structured facts and LR sentences. We propose two unsupervised methods for cross-lingual alignment. We contribute XALIGN, an XF2T dataset with 0.45M pairs across 8 languages, of which 5402 pairs have been manually annotated. We also train strong baseline XF2T generation models on the XAlign dataset. | ['Vasudeva Varma', 'Manish Gupta', 'Anubhav Sharma', 'Bhavyajeet Singh', 'Shivprasad Sagare', 'Tushar Abhishek'] | 2022-02-01 | null | null | null | null | ['data-to-text-generation'] | ['natural-language-processing'] | [ 9.61815342e-02 7.10246623e-01 -3.32331359e-01 -5.00388205e-01
-1.53546453e+00 -8.10084224e-01 1.06905568e+00 -2.07314994e-02
-2.62201309e-01 1.58125544e+00 8.31692576e-01 -5.65699637e-01
2.51333505e-01 -1.01939499e+00 -1.02932966e+00 1.02302119e-01
3.38830173e-01 1.01404238e+00 -1.66447699e-01 -5.81294715e-01
1.19129740e-01 -4.24239546e-01 -1.08881640e+00 8.22810829e-01
1.56243825e+00 2.07591474e-01 2.51294553e-01 7.95106113e-01
-6.97288215e-01 8.75979722e-01 -9.24111605e-01 -1.11310267e+00
2.27503046e-01 -7.65507638e-01 -1.12755930e+00 -5.57232976e-01
1.03446293e+00 -3.20867002e-02 1.24757998e-01 7.76190758e-01
7.03384757e-01 -2.56913513e-01 7.99890876e-01 -9.81275022e-01
-8.85784388e-01 1.76912344e+00 -3.54409933e-01 -1.70641944e-01
7.40651786e-01 -2.10744422e-02 1.13026023e+00 -9.11293983e-01
1.33026600e+00 1.33527064e+00 6.65017605e-01 8.09441328e-01
-9.54129100e-01 -4.48667467e-01 -4.46599424e-01 -3.42485130e-01
-1.17583966e+00 -6.10170126e-01 1.94848001e-01 -2.33662099e-01
1.59009862e+00 1.23103112e-01 3.64798844e-01 1.30174541e+00
2.73132205e-01 7.76332200e-01 1.36755681e+00 -1.02080929e+00
-3.61882389e-01 3.53973687e-01 -2.31107518e-01 5.53728700e-01
5.86659372e-01 -2.02679217e-01 -8.83784950e-01 1.19881660e-01
2.71953970e-01 -1.15644276e+00 2.40870025e-02 5.60974240e-01
-1.63587737e+00 8.01347792e-01 -1.59408465e-01 3.01644415e-01
2.81149764e-02 4.40831110e-02 7.18784988e-01 3.91630352e-01
6.48819506e-01 7.08848953e-01 -7.68891811e-01 -2.81670749e-01
-7.73888588e-01 5.81922054e-01 9.56694067e-01 1.80576932e+00
5.01587868e-01 8.19991380e-02 -2.71557301e-01 9.42580521e-01
9.05624032e-03 9.71502841e-01 8.60798478e-01 -4.53982860e-01
1.39296818e+00 3.81273866e-01 1.00191176e-01 -4.46440071e-01
-1.19767472e-01 -6.40610307e-02 -5.67126274e-01 -4.22525942e-01
3.08037788e-01 -6.25780284e-01 -6.32988572e-01 1.62443233e+00
1.37992889e-01 -9.19026375e-01 7.95320272e-01 1.80071563e-01
1.15547776e+00 8.24854016e-01 -1.08042479e-01 -1.62901476e-01
1.31066477e+00 -8.58436644e-01 -8.32575142e-01 -3.05204064e-01
1.13839376e+00 -1.32820368e+00 1.18484652e+00 -3.85423601e-02
-1.21116316e+00 -6.51960015e-01 -8.21448326e-01 -5.01080573e-01
-9.06143665e-01 6.78992331e-01 5.87785006e-01 8.87107730e-01
-9.60747838e-01 -5.36275283e-02 -4.15190607e-01 -6.98228896e-01
-1.81957811e-01 -3.23581636e-01 -5.25657475e-01 -8.14653635e-02
-1.55815411e+00 1.11026692e+00 9.50034499e-01 -2.66146362e-01
-5.23132861e-01 -6.41451120e-01 -8.76615345e-01 -6.41601741e-01
3.53499621e-01 -8.86750579e-01 1.35080242e+00 -4.59765017e-01
-1.27946126e+00 1.07848084e+00 -3.06358010e-01 -6.43042445e-01
7.23629117e-01 -7.31344461e-01 -6.26457334e-01 -3.51932079e-01
7.82220781e-01 8.14513028e-01 2.27350920e-01 -1.01320052e+00
-7.33808935e-01 -4.58539464e-02 -2.66526550e-01 2.99429685e-01
2.20743101e-02 1.68692067e-01 -9.37584490e-02 -9.76531625e-01
-3.56006473e-01 -8.40118587e-01 4.34857309e-02 -8.44435036e-01
-9.07152593e-01 -4.81875032e-01 3.95857215e-01 -1.18691075e+00
1.29673386e+00 -1.22273111e+00 -2.97865689e-01 -1.59297496e-01
-5.43982208e-01 2.40988091e-01 -1.48894683e-01 1.09310675e+00
7.01515302e-02 6.55923367e-01 -6.23839051e-02 -2.23022431e-01
3.49952102e-01 1.72147244e-01 -8.01622450e-01 -3.90891105e-01
2.49227285e-01 1.20148921e+00 -9.76274729e-01 -9.52227473e-01
-1.21834062e-01 1.08265676e-01 -4.75679725e-01 3.28413516e-01
-6.53862715e-01 8.89618173e-02 -2.63864487e-01 4.37237173e-01
3.70531708e-01 3.94992888e-01 4.87663984e-01 -4.23237197e-02
-4.43366587e-01 1.13208520e+00 -5.82607627e-01 1.88987982e+00
-8.67877305e-01 5.78844607e-01 -7.96520591e-01 -1.36301935e-01
9.43428755e-01 4.33926165e-01 -2.06330102e-02 -5.81722796e-01
-2.97519088e-01 7.74439037e-01 -1.53005213e-01 -5.67032874e-01
1.20212746e+00 -8.04411545e-02 -6.54182732e-01 5.46160996e-01
3.48339081e-01 -7.51993120e-01 8.61555576e-01 4.98280615e-01
7.05790579e-01 7.63648212e-01 4.89004314e-01 -2.53511637e-01
3.02614033e-01 5.01934767e-01 3.22869569e-01 9.08681393e-01
5.96372485e-01 4.76692349e-01 2.34446377e-01 -2.80497044e-01
-1.46378458e+00 -9.12732005e-01 -2.56795019e-01 6.31773472e-01
-5.50877690e-01 -1.15113890e+00 -9.94455338e-01 -9.64177132e-01
-2.47765660e-01 1.53748798e+00 -2.38318503e-01 3.12953085e-01
-1.07539046e+00 -8.53378594e-01 1.12934506e+00 2.77637273e-01
6.41299903e-01 -1.19340622e+00 -1.37131318e-01 4.10519451e-01
-9.12939906e-01 -1.35254169e+00 -4.84602958e-01 -2.37837464e-01
-3.74843717e-01 -7.65981734e-01 -2.35226214e-01 -8.03304195e-01
4.90354300e-01 -4.24464971e-01 1.79285562e+00 -3.74592304e-01
-9.87974852e-02 2.41850279e-02 -6.02707982e-01 -6.39590204e-01
-1.18628502e+00 5.81570685e-01 1.36076987e-01 -8.10489416e-01
5.34541368e-01 9.54469945e-03 2.10942224e-01 -9.88548845e-02
-6.63810670e-01 5.60242534e-01 5.47369599e-01 7.28769481e-01
5.85068285e-01 -8.17322135e-02 7.22984374e-01 -1.53069210e+00
7.54431307e-01 -2.78592110e-01 -5.36646724e-01 7.23555326e-01
-4.56054419e-01 1.97514534e-01 8.55056465e-01 3.02076280e-01
-1.73670638e+00 -4.18641686e-01 -3.31318051e-01 6.47585809e-01
-6.11462630e-03 8.25647175e-01 -3.26064467e-01 7.38129795e-01
7.87549615e-01 1.54746220e-01 -6.01024389e-01 -3.32800090e-01
1.07774627e+00 9.43858445e-01 7.76805580e-01 -1.03026450e+00
7.48709023e-01 -1.31725192e-01 -4.06691849e-01 -7.60411382e-01
-1.01902068e+00 7.14116767e-02 -9.13132012e-01 -6.52510375e-02
9.73453820e-01 -1.08613324e+00 1.06048353e-01 3.86554599e-01
-1.39688873e+00 -4.57912147e-01 -3.80186737e-01 4.08289194e-01
-7.56741464e-01 7.54757077e-02 -5.51400661e-01 -5.60573637e-01
-7.58185387e-01 -5.39124072e-01 1.30447924e+00 -5.21405414e-02
-6.44678950e-01 -9.67917621e-01 5.62582016e-01 6.34471476e-01
1.41891748e-01 2.27520347e-01 9.80773270e-01 -4.25354987e-01
-5.07838786e-01 -1.63631678e-01 2.32343152e-02 1.09304100e-01
3.62093925e-01 3.28556001e-01 -3.88311058e-01 -7.15062544e-02
-4.37574238e-01 -7.82383144e-01 6.40556455e-01 1.23357669e-01
4.09680039e-01 -7.87779570e-01 -3.69644076e-01 4.39469218e-01
1.36341906e+00 -2.88448986e-02 7.94495165e-01 3.86499852e-01
8.61621678e-01 8.76325369e-01 1.04140472e+00 8.44456255e-02
9.17898178e-01 6.16771936e-01 -3.90123904e-01 6.60754070e-02
-6.80241585e-01 -9.67104554e-01 8.16743135e-01 1.29228222e+00
9.21696126e-02 -6.92423999e-01 -1.06712127e+00 8.17353070e-01
-1.54084349e+00 -1.27328861e+00 -4.38150883e-01 1.96839356e+00
1.60301852e+00 6.45082295e-02 -3.18664134e-01 -4.60403264e-01
5.40219843e-01 -9.58735240e-04 5.42021282e-02 -5.00218511e-01
-7.23341405e-01 1.83873281e-01 3.61512810e-01 7.63263524e-01
-9.55938995e-01 1.49218094e+00 6.24240065e+00 1.05669045e+00
-6.25717282e-01 3.79385054e-02 5.08278728e-01 -7.96617493e-02
-8.48837912e-01 3.71711224e-01 -1.59022284e+00 5.50325930e-01
9.77846384e-01 -5.82681298e-01 -1.49156461e-02 7.18197584e-01
-1.54899305e-03 3.82898375e-02 -8.89517367e-01 5.69404364e-01
3.82320553e-01 -1.47307014e+00 4.64793354e-01 -1.28920376e-01
1.25899982e+00 2.12150872e-01 -2.54236937e-01 5.46162248e-01
1.10649264e+00 -9.93790507e-01 1.05838990e+00 6.00240171e-01
1.36811543e+00 -8.18101287e-01 7.59056211e-01 2.71239907e-01
-1.01148856e+00 6.05060160e-01 -4.74715471e-01 4.81972456e-01
6.63488030e-01 9.45325315e-01 -1.33736777e+00 1.00294495e+00
5.80989361e-01 6.05815291e-01 -7.72895098e-01 2.18871504e-01
-6.00841045e-01 5.78943849e-01 -2.49785230e-01 -1.70329362e-01
7.87139237e-02 -2.12403908e-01 6.16909742e-01 1.58829165e+00
7.20436692e-01 -3.61610085e-01 2.38653734e-01 5.75559020e-01
-3.58771473e-01 4.89722729e-01 -1.07887447e+00 -4.05686736e-01
5.32220542e-01 1.16257441e+00 -3.95666063e-01 -7.87086546e-01
-3.94489884e-01 8.59565437e-01 5.32796264e-01 -5.64653240e-02
-4.63128686e-01 -6.51056230e-01 2.82035679e-01 2.76670122e-04
3.38167362e-02 -2.21484601e-01 -2.40073860e-01 -1.55861712e+00
2.76591212e-01 -1.26658082e+00 4.49919343e-01 -1.01684725e+00
-1.59382570e+00 1.02180851e+00 1.20761216e-01 -9.87027407e-01
-1.05407107e+00 -3.79276901e-01 -2.65413612e-01 1.01782048e+00
-1.26818347e+00 -1.77016258e+00 1.39597014e-01 1.80899963e-01
6.60100758e-01 -6.07205689e-01 9.58936095e-01 1.51564158e-03
-2.22968891e-01 6.69897974e-01 1.60537794e-01 3.73701692e-01
1.18168962e+00 -1.69183457e+00 1.15613210e+00 1.25965643e+00
2.69463181e-01 1.02552080e+00 7.08735108e-01 -1.43867862e+00
-1.03912067e+00 -1.59179854e+00 1.97013342e+00 -1.13878572e+00
6.92803681e-01 -7.48590827e-01 -3.69951516e-01 9.94880557e-01
9.87947822e-01 -7.92828143e-01 7.35815763e-01 1.56898245e-01
-3.96710247e-01 1.45424962e-01 -8.25781882e-01 8.97689521e-01
1.10306203e+00 -7.23943233e-01 -6.97321296e-01 7.64266908e-01
9.55778599e-01 -6.44489169e-01 -1.07432938e+00 1.73540473e-01
4.20105904e-01 -4.92095023e-01 5.29141426e-01 -5.91483414e-01
7.85693049e-01 -5.12346148e-01 -2.65844673e-01 -1.56514001e+00
3.20525557e-01 -9.19414282e-01 2.97794998e-01 1.93032086e+00
8.85551393e-01 -5.40197968e-01 3.29226166e-01 1.98434964e-01
-4.84731019e-01 -1.80127472e-01 -7.39742577e-01 -9.55349803e-01
5.03226817e-01 -3.53564024e-01 8.80586803e-01 1.05212533e+00
2.60226756e-01 8.32207501e-01 -5.43426216e-01 -2.56333917e-01
3.49753916e-01 1.86345696e-01 1.18446755e+00 -5.68956077e-01
-1.62105471e-01 -8.16471800e-02 2.16981202e-01 -7.08294570e-01
2.07320705e-01 -1.32713115e+00 1.58849403e-01 -1.75625634e+00
1.89066291e-01 -6.77463770e-01 9.28462327e-01 6.86958015e-01
-4.25956964e-01 1.18294790e-01 -9.90266874e-02 -3.43105048e-02
-1.12640426e-01 6.71727538e-01 8.55829179e-01 5.72433397e-02
8.44114572e-02 -4.78811264e-01 -6.86964214e-01 3.34035993e-01
9.79095757e-01 -5.85374117e-01 -4.24617469e-01 -8.22227895e-01
7.48639882e-01 -1.24439113e-01 -3.37744355e-01 -8.34559500e-01
-5.76455593e-02 -1.02116100e-01 3.03704202e-01 -9.28574920e-01
-1.89506859e-01 -4.01655845e-02 2.39719883e-01 3.16833258e-02
-4.65952307e-01 6.65822983e-01 1.75680831e-01 -2.25462973e-01
-3.09521019e-01 -5.54114103e-01 1.17982596e-01 -5.39878666e-01
-3.47292811e-01 -7.19995350e-02 -3.76728207e-01 7.18483210e-01
5.00730157e-01 1.26156583e-01 -1.07586610e+00 -2.20013902e-01
8.52749795e-02 1.71550989e-01 5.17284095e-01 5.83425164e-01
2.35579982e-01 -1.43769455e+00 -1.45468259e+00 7.11605474e-02
3.23509812e-01 -1.24791630e-01 -1.83509111e-01 2.18286619e-01
-7.29451120e-01 9.48431075e-01 -2.41491660e-01 -5.50764017e-02
-9.31946337e-01 3.16025436e-01 5.47968261e-02 -6.78567350e-01
-2.60764569e-01 6.84846103e-01 -1.93679318e-01 -1.24241710e+00
-5.70721865e-01 8.58984143e-02 -1.11832462e-01 7.69503340e-02
4.64063972e-01 3.43288034e-01 2.41401300e-01 -8.74036372e-01
5.81253022e-02 3.08172107e-01 -4.76599097e-01 -6.34383380e-01
1.06168652e+00 -1.22232012e-01 -5.78925192e-01 4.11026001e-01
8.58702838e-01 8.95023465e-01 -2.88043737e-01 1.25976190e-01
2.42375657e-01 -2.60610938e-01 -5.34880638e-01 -1.07007682e+00
-3.94473106e-01 5.42780578e-01 -1.73696980e-01 4.79929261e-02
6.43356442e-01 9.40072164e-03 9.66952920e-01 5.23534775e-01
5.78097582e-01 -1.42201221e+00 -1.51715174e-01 9.33984458e-01
1.07773399e+00 -1.14552677e+00 -7.73018003e-02 -4.90021586e-01
-7.01637506e-01 9.50277627e-01 1.03300369e+00 4.25931305e-01
7.53913745e-02 2.71727741e-01 3.52281988e-01 -1.36742577e-01
-1.04195547e+00 -1.30106300e-01 1.39422283e-01 8.36229920e-01
1.17897987e+00 2.43767530e-01 -7.50017881e-01 2.49112532e-01
-1.49389839e+00 -2.68411040e-01 7.70093560e-01 6.10562682e-01
7.42971897e-02 -1.64337003e+00 -1.14799039e-02 6.02127016e-01
-6.10874951e-01 -8.15576673e-01 -5.13730109e-01 9.66522992e-01
2.28976846e-01 8.50736439e-01 -9.25620869e-02 -1.97355021e-02
1.61653310e-01 2.12576538e-01 5.25443971e-01 -9.40404058e-01
-6.89904690e-01 -1.17749326e-01 1.07623982e+00 -3.15143652e-02
-3.39425176e-01 -9.11677837e-01 -1.02446997e+00 -3.85942072e-01
-2.27344021e-01 3.23638082e-01 5.99837780e-01 6.63082004e-01
1.58860385e-01 1.94627136e-01 1.59744069e-01 -9.41906944e-02
-1.01080395e-01 -1.27095544e+00 -1.04296528e-01 3.06135386e-01
-2.93083876e-01 1.54239625e-01 -1.23070227e-02 3.64421934e-01] | [11.478921890258789, 9.461713790893555] |
803f1815-49b7-4f09-8dc6-b7c2bc4634a8 | lsta-long-short-term-attention-for-egocentric | 1811.10698 | null | http://arxiv.org/abs/1811.10698v3 | http://arxiv.org/pdf/1811.10698v3.pdf | LSTA: Long Short-Term Attention for Egocentric Action Recognition | Egocentric activity recognition is one of the most challenging tasks in video
analysis. It requires a fine-grained discrimination of small objects and their
manipulation. While some methods base on strong supervision and attention
mechanisms, they are either annotation consuming or do not take spatio-temporal
patterns into account. In this paper we propose LSTA as a mechanism to focus on
features from spatial relevant parts while attention is being tracked smoothly
across the video sequence. We demonstrate the effectiveness of LSTA on
egocentric activity recognition with an end-to-end trainable two-stream
architecture, achieving state of the art performance on four standard
benchmarks. | ['Swathikiran Sudhakaran', 'Oswald Lanz', 'Sergio Escalera'] | 2018-11-26 | lsta-long-short-term-attention-for-egocentric-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Sudhakaran_LSTA_Long_Short-Term_Attention_for_Egocentric_Action_Recognition_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Sudhakaran_LSTA_Long_Short-Term_Attention_for_Egocentric_Action_Recognition_CVPR_2019_paper.pdf | cvpr-2019-6 | ['egocentric-activity-recognition'] | ['computer-vision'] | [ 1.20619088e-02 -6.13310337e-01 -4.43523049e-01 -2.50366986e-01
-4.57154721e-01 -4.45398539e-01 6.73875511e-01 -9.45518091e-02
-6.09941304e-01 3.15109611e-01 5.78077614e-01 2.83847481e-01
-1.33336663e-01 -4.12134320e-01 -7.41694808e-01 -6.33084774e-01
-4.77465242e-01 1.17680550e-01 4.26831096e-01 5.60660027e-02
3.61653954e-01 5.78686118e-01 -1.31231761e+00 3.30013722e-01
2.58635670e-01 1.09207499e+00 8.32119435e-02 1.04349291e+00
4.06946063e-01 1.29521871e+00 -3.06579530e-01 1.71011478e-01
1.27868474e-01 -3.97233814e-01 -7.55107701e-01 2.11785063e-01
7.09559679e-01 -4.54966784e-01 -8.91146541e-01 7.39894331e-01
4.36141610e-01 4.64327037e-01 6.53909206e-01 -1.11828327e+00
-4.42860812e-01 2.27408320e-01 -6.53308094e-01 1.19455767e+00
4.47705507e-01 1.51989385e-01 6.82428598e-01 -1.17370760e+00
4.48725939e-01 8.21929693e-01 4.00647014e-01 4.31685656e-01
-9.90923584e-01 -4.54010278e-01 5.45804262e-01 7.16127455e-01
-1.33289039e+00 -8.04827869e-01 6.94134295e-01 -6.88341200e-01
1.29251647e+00 1.72553258e-03 7.83465207e-01 1.19121790e+00
2.79182255e-01 1.13766885e+00 6.18134618e-01 -7.17785135e-02
3.37932557e-01 -5.39782107e-01 8.84301122e-03 7.37005353e-01
-1.42828211e-01 -3.31977040e-01 -9.63728547e-01 9.69594121e-02
1.06034636e+00 4.44418192e-01 -4.05697793e-01 -8.68425667e-01
-1.47482073e+00 3.17135274e-01 4.49213386e-01 2.11262628e-01
-6.52552843e-01 5.72660029e-01 7.28050828e-01 -1.84112899e-02
6.59214079e-01 2.25199610e-01 -4.29743618e-01 -7.28120267e-01
-8.90220582e-01 1.88559946e-02 1.25462756e-01 1.22096431e+00
4.57660615e-01 1.31360993e-01 -6.15569711e-01 4.76551563e-01
-9.59379151e-02 3.14381659e-01 6.49969697e-01 -8.92160714e-01
6.09657824e-01 5.93010962e-01 8.77947882e-02 -7.70657361e-01
-2.75248230e-01 -4.44893241e-01 -5.32519937e-01 9.04211774e-02
2.96515375e-01 -1.95101798e-02 -9.57950473e-01 1.59494054e+00
1.65091455e-01 7.18102157e-01 -1.55317292e-01 1.13911748e+00
4.57408816e-01 4.67703134e-01 3.10171217e-01 -1.24898061e-01
1.18101597e+00 -1.57073200e+00 -6.37057960e-01 -2.60735899e-01
6.89588964e-01 -1.69156313e-01 1.09209180e+00 2.94831187e-01
-1.42089033e+00 -4.90009248e-01 -8.25843334e-01 -1.61314785e-01
-2.10019708e-01 1.02919377e-01 5.71076930e-01 2.89473563e-01
-6.70707047e-01 5.30466914e-01 -1.29135036e+00 -4.52355474e-01
9.26889181e-01 1.92626089e-01 -7.89247930e-01 1.62461579e-01
-5.52686512e-01 6.28989697e-01 1.45103559e-01 9.19933319e-02
-1.41635907e+00 -6.41608417e-01 -1.03382421e+00 4.11997855e-01
5.86125851e-01 -6.49908841e-01 1.27039623e+00 -1.05712485e+00
-1.33742607e+00 8.17674458e-01 -2.96028137e-01 -5.11978686e-01
5.23612559e-01 -7.25159883e-01 -2.17142642e-01 3.98597240e-01
1.71583310e-01 2.33901724e-01 8.89989018e-01 -3.62710685e-01
-7.44448841e-01 -5.35616159e-01 3.48860621e-02 4.09269273e-01
-4.43853617e-01 3.41403693e-01 -6.76237404e-01 -9.84015405e-01
-8.82254690e-02 -8.69609535e-01 2.34498829e-02 1.31634086e-01
1.01385087e-01 -3.94933164e-01 1.09486055e+00 -3.17939579e-01
1.19933820e+00 -2.24799371e+00 4.69234169e-01 -2.55669922e-01
2.72486150e-01 2.96121269e-01 5.11643216e-02 2.14936778e-01
-1.67034775e-01 -2.30591893e-01 -4.06208336e-02 -2.95225680e-01
-1.46358728e-01 -1.00358136e-01 -3.00819516e-01 9.60678458e-01
2.72978514e-01 1.10925078e+00 -1.13626134e+00 -2.94424325e-01
3.67782742e-01 3.29236388e-01 -7.04626977e-01 4.30029422e-01
1.07894972e-01 4.17343050e-01 -5.38879454e-01 7.14984477e-01
1.56427681e-01 -2.47859046e-01 -1.64758697e-01 -3.01618464e-02
7.98175018e-03 3.54739636e-01 -8.21544051e-01 2.28206277e+00
-2.98219234e-01 9.09319043e-01 -1.87055260e-01 -9.63798940e-01
2.30307832e-01 3.77878904e-01 7.60518551e-01 -7.39603519e-01
1.50617942e-01 -2.50286013e-01 -6.10710010e-02 -7.25255013e-01
3.36057544e-01 2.40011349e-01 -1.41166344e-01 5.98918676e-01
3.06662709e-01 4.74823922e-01 3.33132565e-01 1.48599669e-01
1.37643194e+00 5.96118748e-01 4.49191302e-01 -2.38105893e-01
5.74322462e-01 -5.08156829e-02 6.31812692e-01 6.22842550e-01
-6.64363146e-01 7.86944687e-01 3.42560560e-01 -6.08173907e-01
-9.69805300e-01 -9.60705698e-01 3.36698413e-01 1.73342264e+00
2.18889028e-01 -5.86899757e-01 -7.40962625e-01 -1.08768630e+00
-4.16972101e-01 2.80692041e-01 -9.33264613e-01 -3.53696227e-01
-8.96774471e-01 -2.98241526e-01 4.67674941e-01 1.25559187e+00
4.32576686e-01 -1.13381052e+00 -1.16752732e+00 1.58961669e-01
-1.00636289e-01 -1.28022873e+00 -1.01709366e+00 1.10941678e-01
-7.28796065e-01 -1.25557435e+00 -9.95447755e-01 -5.66389978e-01
6.08136415e-01 7.70969748e-01 1.03476441e+00 -4.65702772e-01
-2.29120120e-01 7.95059443e-01 -2.01696590e-01 -3.81776839e-01
5.49593031e-01 1.17114440e-01 2.12249413e-01 2.79585689e-01
5.43908596e-01 -4.59218234e-01 -1.05046248e+00 5.06963491e-01
-4.61694956e-01 -1.27079576e-01 3.93003106e-01 6.31395757e-01
6.00479841e-01 -3.69498670e-01 2.62098700e-01 -6.83891237e-01
1.84382588e-01 -5.31106830e-01 -2.56263375e-01 1.51264071e-01
-8.59935880e-02 -1.81554019e-01 5.75440466e-01 -4.52846378e-01
-9.03342307e-01 3.12516391e-01 2.63185471e-01 -1.09131229e+00
-2.39895776e-01 2.60197818e-01 -1.48253843e-01 3.11519895e-02
6.03055179e-01 4.06032681e-01 -3.14767033e-01 -3.28909487e-01
1.27519593e-01 2.66282558e-01 7.31208503e-01 -3.76264453e-01
3.05072784e-01 8.48492682e-01 -2.59333372e-01 -8.01449955e-01
-1.00387478e+00 -9.14665520e-01 -9.09284711e-01 -4.39163715e-01
9.37524796e-01 -1.15297163e+00 -5.54281592e-01 4.40415829e-01
-8.31694901e-01 -6.39375508e-01 -4.13825691e-01 7.26562381e-01
-1.06819534e+00 4.74147141e-01 -5.04147112e-01 -5.85520387e-01
-3.11172068e-01 -8.36545646e-01 1.24103570e+00 1.58471882e-01
-3.36267561e-01 -8.32740843e-01 3.16514939e-01 3.20763141e-01
2.69553095e-01 9.74973142e-02 2.79280752e-01 -2.96480268e-01
-6.72481060e-01 -3.62637639e-01 -1.24613680e-01 -3.90505232e-02
2.08965510e-01 -1.37480557e-01 -1.06154180e+00 -3.64211917e-01
-1.50644546e-02 -3.58130604e-01 9.00695503e-01 4.32951927e-01
1.51529992e+00 -1.67147830e-01 -3.88182998e-01 8.38571548e-01
1.00118434e+00 -6.80109710e-02 5.56220829e-01 2.33409643e-01
9.20611203e-01 4.74740624e-01 9.47701931e-01 4.32381839e-01
2.89959401e-01 8.58396232e-01 5.68731189e-01 1.19402215e-01
7.06577748e-02 -1.35587946e-01 5.65460503e-01 3.44048202e-01
-3.95839185e-01 -2.54336238e-01 -8.25072169e-01 8.06346178e-01
-2.33704257e+00 -1.41552246e+00 4.60713506e-02 2.09852624e+00
1.83742598e-01 1.53072476e-01 4.69402015e-01 -3.72792827e-03
5.27646661e-01 3.78232032e-01 -6.73911035e-01 -1.90582812e-01
1.01414889e-01 -1.09775774e-01 5.79116046e-01 2.32990701e-02
-1.64913416e+00 9.87366736e-01 6.65267754e+00 4.13911134e-01
-1.22369480e+00 2.65183240e-01 2.96774626e-01 -6.40642583e-01
4.94667023e-01 -3.74918044e-01 -3.95408213e-01 4.78214532e-01
8.18474710e-01 -1.71183631e-01 1.68733552e-01 1.12020087e+00
4.31308240e-01 -2.15060055e-01 -1.47895193e+00 1.23213351e+00
3.45099866e-01 -1.22077286e+00 -1.76434606e-01 -5.53293340e-03
7.19154298e-01 3.32877576e-01 -1.13389350e-01 2.85464406e-01
-8.37098062e-02 -1.04753149e+00 1.06636965e+00 6.18580878e-01
5.96596301e-01 -5.44942021e-01 3.49094540e-01 4.53687221e-01
-1.45094073e+00 -3.30162793e-01 -2.31472179e-01 -2.79470682e-01
2.78172672e-01 -9.64162275e-02 -4.16579098e-01 2.21004104e-03
9.48979974e-01 1.15265179e+00 -4.29556817e-01 1.33286273e+00
-2.32036486e-01 6.78873837e-01 -1.05740376e-01 1.48949116e-01
4.62267965e-01 7.52244741e-02 6.78130269e-01 1.59463310e+00
1.59858033e-01 2.15972349e-01 1.63148716e-01 3.69230837e-01
-8.52428228e-02 -9.19776931e-02 -8.67948830e-01 -3.95161286e-02
-6.47343770e-02 1.11651421e+00 -7.48512089e-01 -4.91440475e-01
-4.91566986e-01 1.15850294e+00 5.79429626e-01 2.88084537e-01
-1.07144332e+00 -2.83384979e-01 9.09264088e-01 2.62131691e-01
6.12037659e-01 -6.24248981e-01 1.56403750e-01 -1.37400901e+00
2.04443976e-01 -4.74597186e-01 7.87736297e-01 -8.64631653e-01
-9.15578008e-01 2.72469342e-01 -8.40953961e-02 -1.44072366e+00
-3.28355491e-01 -5.90438187e-01 -6.93722486e-01 5.25209665e-01
-1.37667894e+00 -1.08531010e+00 -7.67940342e-01 9.06051159e-01
1.13430095e+00 -1.15664840e-01 6.91118181e-01 4.92240220e-01
-7.89577007e-01 4.32273686e-01 -1.55083403e-01 3.85238767e-01
7.22579598e-01 -1.31524777e+00 4.77505565e-01 1.00922906e+00
3.37713033e-01 5.78417838e-01 4.65938479e-01 -2.59730220e-01
-1.56799138e+00 -1.13814044e+00 4.81065899e-01 -8.06951344e-01
6.12963200e-01 -6.01699531e-01 -7.72538722e-01 9.71370220e-01
1.60925880e-01 7.02593029e-01 6.36321604e-01 4.57446799e-02
-4.22708482e-01 4.45239851e-03 -6.64642453e-01 5.16261697e-01
1.49002028e+00 -7.83243179e-01 -6.00734055e-01 3.09033871e-01
1.39729157e-01 -4.55471992e-01 -6.18112326e-01 2.80273616e-01
6.73030913e-01 -9.06180322e-01 9.47957635e-01 -1.23096430e+00
4.08862114e-01 -5.35703242e-01 4.59794141e-02 -1.07731175e+00
-7.07127094e-01 -5.84810913e-01 -7.89573073e-01 7.88159311e-01
-8.18693638e-02 -1.27061412e-01 1.11094964e+00 2.59716511e-01
-2.10407838e-01 -6.61283195e-01 -8.12559426e-01 -7.94795036e-01
-4.64466780e-01 -5.18643260e-01 1.53938562e-01 8.73962998e-01
3.57629091e-01 3.66567135e-01 -4.89127636e-01 1.53369186e-02
4.45305705e-01 -6.28736988e-02 9.85736787e-01 -6.95519030e-01
-1.40246451e-01 -6.35790884e-01 -9.01417255e-01 -1.28034234e+00
1.39287725e-01 -5.43631494e-01 1.10942736e-01 -1.32695901e+00
4.01154369e-01 3.52016389e-02 -6.89369261e-01 2.82649189e-01
-6.62051067e-02 6.08305871e-01 -2.10415453e-01 1.69554815e-01
-1.30070698e+00 7.02973008e-01 9.29542482e-01 -1.97725743e-01
-1.12997785e-01 5.26822433e-02 -3.58713686e-01 8.09672892e-01
6.22404635e-01 -3.86660427e-01 -6.07917726e-01 -6.52060151e-01
-1.10559285e-01 -1.36024043e-01 4.37307954e-01 -1.23407984e+00
4.83651966e-01 -3.52719426e-01 4.50442940e-01 -5.81948221e-01
6.14024520e-01 -8.10751200e-01 -2.87277371e-01 2.69244194e-01
-3.23077500e-01 1.78500980e-01 1.53860837e-01 9.32897925e-01
-2.27161080e-01 2.01280400e-01 6.04386330e-01 -1.15155548e-01
-1.02917969e+00 7.04502761e-01 -6.38251543e-01 1.97753429e-01
1.46680629e+00 -4.04621631e-01 -1.23686261e-01 -2.32666537e-01
-7.61558235e-01 3.00280631e-01 4.90751207e-01 5.94775856e-01
5.95270813e-01 -1.21944892e+00 -5.64366221e-01 1.14337847e-01
4.58296686e-01 4.40124981e-02 4.57817584e-01 1.15183055e+00
-5.81323981e-01 6.75719142e-01 -4.27040279e-01 -8.78469646e-01
-1.37914753e+00 8.77177775e-01 5.39231479e-01 3.46152335e-02
-8.15728784e-01 8.44586194e-01 7.52688944e-01 2.89793879e-01
5.30021310e-01 -2.59048939e-01 -3.65970463e-01 2.48196023e-03
7.48748481e-01 5.94427466e-01 -1.01326220e-01 -7.23372042e-01
-4.89753842e-01 6.53410256e-01 -9.77542624e-02 1.98291644e-01
1.40547192e+00 -1.95983276e-01 2.81050533e-01 5.36277533e-01
9.13756728e-01 -1.42987534e-01 -1.85111332e+00 -2.68949181e-01
1.15252294e-01 -8.88216972e-01 1.30920783e-01 -2.77446240e-01
-9.85208273e-01 1.09247553e+00 5.35706341e-01 1.63459759e-02
1.05083203e+00 1.20091988e-02 6.57944202e-01 4.58302051e-01
3.00527185e-01 -1.17515337e+00 4.90319014e-01 5.69162488e-01
9.02614117e-01 -1.24855196e+00 6.50394112e-02 -2.89622933e-01
-8.32620382e-01 8.49045038e-01 8.11839104e-01 -4.51718777e-01
5.24606943e-01 8.02124664e-02 -2.24693254e-01 -3.83580118e-01
-9.47594702e-01 -2.92853147e-01 3.47070336e-01 4.06666696e-01
4.17060137e-01 -3.14581901e-01 1.61006361e-01 3.74450713e-01
5.27831674e-01 1.02728732e-01 3.68616074e-01 1.30958223e+00
-2.99664736e-01 -3.58186573e-01 4.05731499e-02 2.85117209e-01
-6.50632441e-01 2.54309863e-01 -2.90856838e-01 4.66478199e-01
-1.71200871e-01 4.75149363e-01 3.61761481e-01 -6.80055320e-02
5.55548072e-01 -6.69552982e-02 7.08157063e-01 -8.66428196e-01
-5.52046120e-01 5.28586805e-02 -6.56136647e-02 -9.77715671e-01
-6.41747177e-01 -9.85386670e-01 -1.02631176e+00 -9.13253427e-02
-4.23260890e-02 -9.22081918e-02 3.05033267e-01 9.68567729e-01
6.11400127e-01 6.43798888e-01 5.13426125e-01 -1.27415359e+00
-1.15507849e-01 -9.67499793e-01 -4.33519542e-01 3.88327390e-01
5.13873994e-01 -8.46809328e-01 9.49874148e-02 2.29635581e-01] | [8.415735244750977, 0.6257859468460083] |
b4304a3e-ce17-4dc6-b74f-3589f1c3bc50 | clinical-language-understanding-evaluation | 2209.14377 | null | https://arxiv.org/abs/2209.14377v1 | https://arxiv.org/pdf/2209.14377v1.pdf | Clinical Language Understanding Evaluation (CLUE) | Clinical language processing has received a lot of attention in recent years, resulting in new models or methods for disease phenotyping, mortality prediction, and other tasks. Unfortunately, many of these approaches are tested under different experimental settings (e.g., data sources, training and testing splits, metrics, evaluation criteria, etc.) making it difficult to compare approaches and determine state-of-the-art. To address these issues and facilitate reproducibility and comparison, we present the Clinical Language Understanding Evaluation (CLUE) benchmark with a set of four clinical language understanding tasks, standard training, development, validation and testing sets derived from MIMIC data, as well as a software toolkit. It is our hope that these data will enable direct comparison between approaches, improve reproducibility, and reduce the barrier-to-entry for developing novel models or methods for these clinical language understanding tasks. | ['Dina Demner-Fushman', 'Travis R. Goodwin'] | 2022-09-28 | null | null | null | null | ['mortality-prediction'] | ['medical'] | [ 1.98064655e-01 -1.29089087e-01 -3.59104127e-01 -5.68251848e-01
-8.39763880e-01 -4.26937222e-01 4.61169451e-01 9.81681764e-01
-4.58302408e-01 5.91343701e-01 4.89118248e-01 -6.41633213e-01
-2.22701222e-01 -5.24610102e-01 1.64918199e-01 -2.90177435e-01
-3.63049835e-01 8.14602554e-01 -9.43699386e-03 1.80559546e-01
1.22443035e-01 3.59573126e-01 -1.04053879e+00 7.95009673e-01
6.51593983e-01 6.06979609e-01 -1.14588991e-01 6.68318629e-01
-1.57625854e-01 7.43923604e-01 -4.31690007e-01 -3.11308205e-01
-7.65137002e-02 -5.22566915e-01 -9.53871250e-01 -3.58731776e-01
-9.11943689e-02 1.66156754e-01 1.57164320e-01 6.12253308e-01
8.23825538e-01 -2.84061134e-01 5.19249737e-01 -9.43575263e-01
-3.96294355e-01 4.88723218e-01 -1.65268838e-01 2.13343829e-01
3.99114311e-01 3.65294933e-01 5.54314554e-01 -4.12252039e-01
9.32405055e-01 1.28720546e+00 8.22973132e-01 8.20375621e-01
-1.32239807e+00 -5.00611067e-01 -2.05428332e-01 5.83319515e-02
-1.28802264e+00 -4.26301271e-01 1.03113897e-01 -6.37385547e-01
9.72752094e-01 2.34707072e-01 3.26989919e-01 1.08437288e+00
4.27646279e-01 7.28973329e-01 1.12148285e+00 -3.82641107e-01
1.90306440e-01 4.67783034e-01 4.55498785e-01 6.58902466e-01
1.86753988e-01 1.80892631e-01 -3.52458507e-01 -5.68545520e-01
2.13451296e-01 -1.52168795e-01 -4.03614253e-01 -1.14898965e-01
-1.24575496e+00 8.17488372e-01 -1.53931737e-01 4.04197872e-01
-8.78837705e-02 -4.34675455e-01 8.87823761e-01 4.27669913e-01
5.14780402e-01 5.10782421e-01 -8.37390780e-01 -4.36732233e-01
-8.87142837e-01 1.47862673e-01 1.15754879e+00 5.00459969e-01
1.69038832e-01 -4.27162141e-01 -1.63128793e-01 8.24528813e-01
3.71249706e-01 8.33674148e-02 8.37322176e-01 -5.99362433e-01
3.84881020e-01 6.16423666e-01 -2.32814088e-01 -6.17915511e-01
-9.08891678e-01 -3.65738183e-01 -9.14021134e-01 -9.85931978e-02
5.14999390e-01 -2.27307111e-01 -8.32926929e-01 1.75102901e+00
-9.26853914e-04 -6.49331231e-03 3.18547338e-01 5.66943765e-01
7.84666300e-01 2.06668839e-01 5.97372949e-01 -2.50061899e-01
1.54100716e+00 -5.27728438e-01 -4.95555043e-01 -2.69899994e-01
1.47651267e+00 -8.58045220e-01 1.02587640e+00 5.90166569e-01
-9.76060987e-01 -3.68113935e-01 -6.92925572e-01 -1.00952476e-01
-4.81168151e-01 3.37016061e-02 5.63622296e-01 7.41186202e-01
-9.49531972e-01 4.47331488e-01 -1.23992443e+00 -7.09111333e-01
2.36267611e-01 3.33714008e-01 -3.97358477e-01 -1.67408943e-01
-1.24607253e+00 1.11249864e+00 4.60322589e-01 -2.14252561e-01
-6.43662810e-01 -1.14572787e+00 -6.97744727e-01 -1.95767194e-01
-1.25459179e-01 -1.08296287e+00 1.08872843e+00 -7.00418651e-01
-1.06379139e+00 1.35539925e+00 -8.56474712e-02 -4.43276227e-01
5.44291198e-01 -1.93259090e-01 -5.95513642e-01 -3.31123322e-01
1.21087497e-02 4.97629762e-01 3.95067818e-02 -6.96787179e-01
-3.58571440e-01 -6.32239699e-01 -5.20782888e-01 -1.22089893e-01
1.29931480e-01 4.53543872e-01 -2.73064345e-01 -5.11170387e-01
-2.96803981e-01 -7.76839793e-01 -3.80263805e-01 1.52464872e-02
-3.23086619e-01 -2.06100829e-02 3.49627525e-01 -8.16599727e-01
1.38768554e+00 -2.32232165e+00 5.78634441e-02 -2.86208671e-02
4.27853733e-01 4.61724788e-01 -2.20896691e-01 6.12937093e-01
-3.84943634e-01 6.75787151e-01 -1.28422841e-01 -3.68964285e-01
-4.42744374e-01 8.55782479e-02 5.31637520e-02 3.02833200e-01
3.76399964e-01 6.62533700e-01 -8.64446461e-01 -6.57704055e-01
2.31310904e-01 4.96898770e-01 -5.38640141e-01 2.53987014e-01
-5.06018624e-02 5.91663778e-01 -7.44493783e-01 5.39035857e-01
3.03571403e-01 -4.17796701e-01 3.06111664e-01 1.28392950e-01
4.58957367e-02 5.44170916e-01 -8.50945115e-01 1.51601124e+00
-3.85659397e-01 4.20617163e-01 -1.06367663e-01 -8.98356795e-01
7.19951153e-01 4.11843568e-01 5.14057755e-01 -4.42934275e-01
2.66614884e-01 7.70604983e-02 3.86456162e-01 -6.32304609e-01
-1.59919158e-01 -4.80822951e-01 3.83334011e-02 4.43554223e-01
-2.59221256e-01 1.29944533e-01 2.22070754e-01 4.40292098e-02
1.33409750e+00 -3.89832318e-01 6.97963655e-01 -1.64068788e-01
6.67216003e-01 1.64480954e-01 7.27006495e-01 5.74872196e-01
-4.98854220e-01 7.66492307e-01 7.20013738e-01 -6.27211750e-01
-6.69271111e-01 -9.85726893e-01 -5.95751166e-01 7.14771032e-01
-5.23961008e-01 -7.64496386e-01 -5.55840790e-01 -5.69612801e-01
5.66043742e-02 7.39638746e-01 -6.66421056e-01 -2.95829982e-01
-3.36888134e-01 -1.24410617e+00 9.49175358e-01 4.63743836e-01
2.83886045e-02 -1.07380927e+00 -5.25194526e-01 4.16615725e-01
-1.54263273e-01 -1.06235993e+00 -1.51656732e-01 1.81060940e-01
-1.17746282e+00 -1.41368854e+00 -2.83239245e-01 -6.10786676e-01
3.93339276e-01 -2.69898772e-01 1.47328138e+00 2.15085462e-01
-5.00003576e-01 2.57545918e-01 -3.68400604e-01 -7.48912394e-01
-1.04073811e+00 2.22569853e-01 3.40380077e-03 -3.73027325e-01
6.47506058e-01 -1.28856644e-01 -5.31095207e-01 2.01861694e-01
-1.04772723e+00 2.38004223e-01 5.80392241e-01 8.24636519e-01
5.71597219e-01 -3.61945897e-01 6.72205627e-01 -1.15841389e+00
1.02587044e+00 -6.83651388e-01 -2.77476728e-01 6.11491919e-01
-9.81008589e-01 2.80545413e-01 6.80410743e-01 -1.45553946e-01
-7.00261474e-01 -4.08491760e-01 -1.65770575e-01 6.28456324e-02
-5.07414818e-01 9.56622422e-01 1.82193180e-03 3.29859346e-01
8.01373899e-01 1.28344312e-01 3.33156615e-01 -7.60200560e-01
1.17056593e-01 9.03101146e-01 1.08357795e-01 -4.48647082e-01
1.28849447e-01 1.79830655e-01 -2.51022965e-01 -5.97917676e-01
-6.66640162e-01 -5.08203566e-01 -5.70234299e-01 4.31323975e-01
9.27530944e-01 -8.67064178e-01 -2.75034010e-01 2.15741426e-01
-1.00318027e+00 -3.36970925e-01 -8.69210511e-02 6.21521175e-01
-2.93904096e-01 3.04285347e-01 -7.26152539e-01 -3.37815911e-01
-6.48216188e-01 -1.44584620e+00 9.72153306e-01 -7.86840618e-02
-9.39460993e-01 -1.38437808e+00 4.14756685e-01 5.25927424e-01
5.01556396e-01 2.77670056e-01 1.53753710e+00 -1.06301904e+00
-6.52734190e-02 -2.88142949e-01 -2.10969731e-01 3.45123321e-01
3.30132633e-01 9.59810466e-02 -6.31935358e-01 -2.91137159e-01
-7.25709349e-02 -4.64675725e-01 5.67309320e-01 4.03916448e-01
1.32737195e+00 9.93267894e-02 -6.12530649e-01 5.45914769e-01
1.11001277e+00 2.26653963e-01 4.59598988e-01 9.87546444e-02
2.28821233e-01 7.12924778e-01 4.00899559e-01 3.03265274e-01
4.38316554e-01 5.29024303e-01 -2.46924236e-01 -2.45705634e-01
-7.85177276e-02 1.24494374e-01 2.15645984e-01 8.89525831e-01
2.19921067e-01 -1.93659201e-01 -1.54059124e+00 4.51175958e-01
-1.62746668e+00 -4.69218105e-01 -7.24630207e-02 2.28455400e+00
1.12748194e+00 6.99641258e-02 3.77985649e-02 -1.42840549e-01
1.99764952e-01 -1.79919556e-01 -4.50979024e-01 -7.45251596e-01
1.76565439e-01 2.35297635e-01 5.14589325e-02 3.79033089e-01
-7.46511161e-01 7.19710708e-01 7.14962530e+00 5.35214722e-01
-1.41990292e+00 1.22461375e-02 1.12316632e+00 1.55756797e-03
-4.42230329e-02 -1.59379810e-01 -6.40629828e-01 3.36025357e-01
1.37007344e+00 -2.95246601e-01 1.76569268e-01 5.29125631e-01
7.70373821e-01 -6.58782497e-02 -1.41729259e+00 7.77067542e-01
3.11258286e-02 -1.08309495e+00 -2.03461915e-01 -1.53901458e-01
1.99781790e-01 4.73624527e-01 -2.82803446e-01 2.78313965e-01
2.96396255e-01 -1.17119122e+00 7.80714527e-02 3.91983002e-01
9.89289224e-01 -1.67216927e-01 9.91535246e-01 3.02643389e-01
-7.97914147e-01 1.89537182e-01 3.13020572e-02 1.00724541e-01
2.11797729e-01 6.54081523e-01 -1.21142280e+00 8.00596058e-01
5.36317825e-01 6.81435287e-01 -7.34847963e-01 1.00523841e+00
2.38179311e-01 8.86080921e-01 -9.98604894e-02 2.73471922e-02
-1.17513398e-02 6.55868575e-02 2.24607378e-01 1.53495741e+00
-1.23503149e-01 6.36584759e-02 4.61506993e-01 7.16043890e-01
1.39664441e-01 5.45779765e-01 -4.27978784e-01 -4.49315399e-01
1.53913647e-01 1.02491641e+00 -7.40498424e-01 -4.32424605e-01
-5.57502925e-01 3.07262152e-01 1.19047999e-01 1.37304038e-01
-5.79257190e-01 -1.00289114e-01 9.63768959e-01 4.27417010e-01
-6.29308939e-01 -6.81712329e-02 -5.62251747e-01 -1.25827169e+00
-4.54212725e-01 -1.41564822e+00 8.75367939e-01 -4.25048649e-01
-1.46086085e+00 6.60929561e-01 1.22285336e-01 -9.41207349e-01
-3.49158317e-01 -7.37152100e-01 -4.00623977e-01 9.13093567e-01
-1.30696881e+00 -7.33755350e-01 -2.64523208e-01 9.77199078e-02
3.60218018e-01 -2.62724757e-01 1.20901918e+00 4.08672035e-01
-6.55873358e-01 4.89969015e-01 7.73334578e-02 2.35003665e-01
1.12057638e+00 -9.44721997e-01 1.38783082e-01 2.16255829e-01
-1.12566046e-01 7.90659070e-01 5.87862372e-01 -5.26090145e-01
-9.89782393e-01 -1.13151538e+00 9.75638211e-01 -8.62278223e-01
7.37592876e-01 -1.24425866e-01 -1.10376406e+00 5.70077896e-01
-1.38388291e-01 -3.25812101e-01 1.49227607e+00 5.19163311e-01
-2.46730626e-01 -4.20413492e-03 -1.14615846e+00 4.40487653e-01
7.46801257e-01 -3.59094769e-01 -6.46638572e-01 4.10731703e-01
4.09826219e-01 -2.08877295e-01 -1.22234583e+00 6.06873751e-01
5.76627612e-01 -7.24704504e-01 8.13219726e-01 -1.27381384e+00
4.52766418e-01 -2.02412754e-01 1.70151561e-01 -1.20563197e+00
-1.69496492e-01 -2.96269536e-01 4.01476949e-01 9.49113667e-01
7.89886117e-01 -8.22067440e-01 5.65669417e-01 7.51804650e-01
2.43197083e-02 -1.16373467e+00 -6.82285368e-01 -3.74084383e-01
3.51220220e-01 -7.35266864e-01 4.05039161e-01 1.01959527e+00
2.51716077e-01 3.58785212e-01 8.59501362e-02 -5.12604676e-02
2.03082860e-01 1.95194092e-02 5.87272584e-01 -1.20749724e+00
-2.78549314e-01 -5.51128089e-01 -6.61179662e-01 -3.20505857e-01
5.55189152e-04 -1.17025375e+00 -4.57865447e-01 -1.52980292e+00
4.79281723e-01 -7.33303130e-01 -5.01626432e-01 9.40994620e-01
-3.97981495e-01 -1.45334974e-01 -3.89049836e-02 2.34290347e-01
-4.28145051e-01 1.15855977e-01 8.40483904e-01 -6.52673021e-02
-4.50779885e-01 7.04081077e-03 -1.02799845e+00 6.16033196e-01
9.57491994e-01 -6.67736113e-01 -4.68190759e-01 -4.11366612e-01
4.56504151e-02 3.07285875e-01 1.47790417e-01 -8.33632708e-01
7.57087115e-03 -1.52337939e-01 1.85485616e-01 -1.75710633e-01
-2.17114925e-01 -3.45145136e-01 2.55113900e-01 9.05743659e-01
-6.19362772e-01 3.38235140e-01 4.78782266e-01 2.59442389e-01
-3.53823811e-01 -2.78592855e-03 6.66516960e-01 -1.25699073e-01
-4.40598667e-01 1.60475254e-01 -4.35731918e-01 5.72562158e-01
7.86471128e-01 -1.35565192e-01 -2.34825745e-01 -9.76489484e-02
-1.04505634e+00 5.04760683e-01 3.84898782e-01 8.19133818e-01
3.12132835e-01 -6.79545879e-01 -1.27037144e+00 2.78050691e-01
4.47892725e-01 -1.13093346e-01 1.50377557e-01 1.13826072e+00
-7.03776300e-01 6.85895622e-01 4.50179540e-02 -5.73259890e-01
-1.66008842e+00 4.86947328e-01 4.73555118e-01 -9.32124197e-01
-3.67606521e-01 2.67979681e-01 3.78425211e-01 -5.59803247e-01
1.05014421e-01 -5.11042953e-01 3.43256444e-02 -1.97654679e-01
6.74747229e-01 9.76633281e-02 2.51667231e-01 -1.76169649e-01
-6.04994416e-01 3.07098269e-01 -4.03582215e-01 1.88150555e-01
1.20866585e+00 2.09013879e-01 -1.28729865e-01 6.34914517e-01
1.07788432e+00 -2.42689446e-01 -2.26770550e-01 -7.14312717e-02
4.47745651e-01 -2.05687545e-02 -7.13698864e-02 -1.26129699e+00
-7.87486434e-01 1.05061853e+00 9.14936483e-01 1.00633010e-01
1.24343979e+00 7.76542202e-02 4.17109966e-01 9.68404859e-02
2.88027585e-01 -7.25560009e-01 -4.97544259e-01 3.48093182e-01
6.74730301e-01 -1.39534187e+00 9.45352390e-02 -4.35827047e-01
-6.14501715e-01 1.08362067e+00 2.64425248e-01 5.35506487e-01
8.93959999e-01 5.02505839e-01 5.85255265e-01 -2.19477341e-01
-1.17371213e+00 1.56560525e-01 1.91230386e-01 5.42456746e-01
1.22912192e+00 1.82439432e-01 -7.36959159e-01 7.19501615e-01
-1.06062651e-01 5.82284272e-01 2.91040182e-01 7.53866792e-01
-7.80344335e-03 -1.80799043e+00 -1.11412637e-01 7.39514947e-01
-8.25204253e-01 -2.84190267e-01 -4.37582076e-01 1.04892421e+00
1.50223330e-01 8.29830825e-01 -2.81285912e-01 -2.06507891e-01
3.56285751e-01 4.34015870e-01 1.81813255e-01 -1.03351748e+00
-8.04810107e-01 -1.59279719e-01 4.60703701e-01 -4.81129706e-01
-1.15748078e-01 -7.69143105e-01 -1.33988655e+00 -2.07444504e-01
-1.82919577e-01 2.25173429e-01 3.30945253e-01 9.91127968e-01
9.77251232e-01 5.48412263e-01 3.22984569e-02 9.67108533e-02
-6.44885898e-01 -1.14363062e+00 -2.90544778e-01 5.25158167e-01
-7.24602863e-02 -3.13537717e-01 -2.03635290e-01 2.01271102e-01] | [8.456910133361816, 8.393970489501953] |
631d0ce6-d577-4a0c-91c6-ddd7634560c6 | perada-parameter-efficient-and-generalizable | 2302.06637 | null | https://arxiv.org/abs/2302.06637v1 | https://arxiv.org/pdf/2302.06637v1.pdf | PerAda: Parameter-Efficient and Generalizable Federated Learning Personalization with Guarantees | Personalized Federated Learning (pFL) has emerged as a promising solution to tackle data heterogeneity across clients in FL. However, existing pFL methods either (1) introduce high communication and computation costs or (2) overfit to local data, which can be limited in scope, and are vulnerable to evolved test samples with natural shifts. In this paper, we propose PerAda, a parameter-efficient pFL framework that reduces communication and computational costs and exhibits superior generalization performance, especially under test-time distribution shifts. PerAda reduces the costs by leveraging the power of pretrained models and only updates and communicates a small number of additional parameters from adapters. PerAda has good generalization since it regularizes each client's personalized adapter with a global adapter, while the global adapter uses knowledge distillation to aggregate generalized information from all clients. Theoretically, we provide generalization bounds to explain why PerAda improves generalization, and we prove its convergence to stationary points under non-convex settings. Empirically, PerAda demonstrates competitive personalized performance (+4.85% on CheXpert) and enables better out-of-distribution generalization (+5.23% on CIFAR-10-C) on different datasets across natural and medical domains compared with baselines, while only updating 12.6% of parameters per model based on the adapter. | ['Anima Anandkumar', 'Bo Li', 'Chaowei Xiao', 'Daguang Xu', 'Wenda Chu', 'De-An Huang', 'Chulin Xie'] | 2023-02-13 | null | null | null | null | ['personalized-federated-learning'] | ['methodology'] | [-2.53448009e-01 -3.10149759e-01 -4.52113926e-01 -5.00727654e-01
-1.21894324e+00 -6.58836365e-01 -3.10175139e-02 -9.31089297e-02
-3.55714411e-01 8.17432582e-01 -1.87655911e-02 -2.64232635e-01
-4.96594220e-01 -7.44715750e-01 -9.91244555e-01 -8.96633923e-01
-3.17851335e-01 7.37861514e-01 1.29758298e-01 1.81029648e-01
-3.80496025e-01 4.05323774e-01 -1.21380377e+00 4.29635793e-01
1.17417717e+00 1.29396009e+00 -8.55641216e-02 4.77344006e-01
-2.25626361e-02 6.15785122e-01 -6.76953077e-01 -7.16420472e-01
2.89347738e-01 -1.14292942e-01 -7.54342496e-01 -1.58214346e-01
3.69522542e-01 -5.55730879e-01 -3.43397707e-01 8.71469438e-01
8.47612977e-01 1.30583465e-01 2.21641749e-01 -1.41078186e+00
-7.44248807e-01 1.00728416e+00 -4.01115179e-01 1.89776555e-01
-3.21973301e-02 3.69120717e-01 8.88926804e-01 -5.10451496e-01
3.47662181e-01 1.06859386e+00 9.28158641e-01 6.04921460e-01
-1.37263441e+00 -6.62052572e-01 3.39556247e-01 2.10126385e-01
-1.34262180e+00 -1.98686659e-01 2.52704680e-01 2.25533415e-02
8.88988137e-01 4.88808811e-01 1.28818095e-01 1.37888479e+00
-1.03422649e-01 1.07160258e+00 8.12144697e-01 1.36663288e-01
5.77260911e-01 2.83320338e-01 3.61919641e-01 3.79511744e-01
1.01278178e-01 -1.36277094e-01 -6.84226274e-01 -8.38820100e-01
3.04991364e-01 3.85285854e-01 -6.22139931e-01 -3.40880901e-01
-7.30804920e-01 7.79204249e-01 4.76088077e-01 -4.07917276e-02
-5.08205354e-01 9.05870497e-02 5.22432029e-01 4.08237457e-01
3.49904746e-01 1.49151593e-01 -9.75522399e-01 -3.61530334e-01
-5.58485568e-01 1.63450390e-01 9.52612936e-01 1.14059377e+00
8.18354130e-01 -1.86869457e-01 -5.74055254e-01 9.86068487e-01
-3.23754132e-01 7.19773233e-01 7.28541136e-01 -9.47419763e-01
7.33056128e-01 7.32599974e-01 -7.23747686e-02 -6.17651224e-01
-2.96998888e-01 -9.03046012e-01 -7.98824728e-01 -3.97101492e-01
2.18375787e-01 -4.00403053e-01 -4.85873044e-01 2.02888560e+00
5.80738187e-01 3.16112310e-01 -5.43837575e-03 8.36817384e-01
3.44200790e-01 3.70949984e-01 -1.28114015e-01 -4.38427515e-02
9.93532300e-01 -1.08632517e+00 -2.83115387e-01 3.74143384e-03
9.32914555e-01 -2.20824435e-01 1.50442970e+00 5.64780951e-01
-1.05734813e+00 4.42573950e-02 -4.92156982e-01 2.31665656e-01
-1.13168158e-01 -3.38616341e-01 5.96381426e-01 7.44319201e-01
-9.07822907e-01 6.15085661e-01 -8.98083866e-01 -5.48776314e-02
9.30249453e-01 4.40393448e-01 -2.27267951e-01 -4.76325035e-01
-8.05660129e-01 2.64418066e-01 5.47473101e-05 -3.82408828e-01
-8.74668896e-01 -1.33327520e+00 -3.01057696e-01 5.07593811e-01
6.26817048e-01 -1.03409743e+00 1.37021506e+00 -7.78719306e-01
-1.46792102e+00 2.13022485e-01 2.73357723e-02 -6.44110978e-01
8.54030371e-01 -3.11315656e-01 -2.81593531e-01 -1.41466465e-02
-2.68221796e-01 1.93289697e-01 4.00670558e-01 -8.79237533e-01
-8.19237590e-01 -6.27603531e-01 5.12909405e-02 -8.31392482e-02
-9.08768058e-01 -4.71052885e-01 -7.52107799e-01 -5.43926299e-01
-2.67219216e-01 -8.77064228e-01 -7.86413848e-02 -8.64923075e-02
-2.71817267e-01 -3.96682143e-01 7.97957242e-01 -2.97552139e-01
1.21418536e+00 -2.36908889e+00 -1.39154270e-01 3.13245386e-01
2.67563432e-01 2.96292633e-01 -3.60217959e-01 3.13396126e-01
4.17916507e-01 -2.05782950e-02 -2.28482112e-01 -2.84834504e-01
3.95592064e-01 4.48286057e-01 -2.41640523e-01 4.67325032e-01
-4.18062240e-01 8.72073174e-01 -7.17978776e-01 -1.56519219e-01
-4.76989299e-01 6.18695796e-01 -1.25583613e+00 1.28237516e-01
-4.58782762e-01 1.83195263e-01 -5.56877673e-01 6.71399951e-01
9.17993844e-01 -7.08019972e-01 2.82275379e-01 -1.71628371e-01
5.49875915e-01 2.35658139e-01 -1.10073054e+00 1.48835087e+00
-7.49566793e-01 1.00664824e-01 1.66889250e-01 -8.97740066e-01
5.05858779e-01 1.16104104e-01 5.65341294e-01 -6.39226496e-01
4.07952331e-02 2.26054326e-01 -4.44796443e-01 -4.83498573e-01
-6.85064793e-02 2.07764775e-01 1.39938325e-01 7.30867505e-01
-5.87109029e-02 6.54804528e-01 -9.40611511e-02 1.41994476e-01
1.53672194e+00 -3.74131173e-01 -2.66015530e-01 -8.05549473e-02
2.34449238e-01 -3.48606765e-01 8.09059024e-01 9.37965930e-01
-4.21021730e-02 1.66753113e-01 4.43918407e-01 -4.15439099e-01
-5.29582739e-01 -1.17012322e+00 -1.46280721e-01 1.60214937e+00
4.36307713e-02 -2.95273960e-01 -7.35226750e-01 -1.07069576e+00
6.98657274e-01 8.76950026e-01 -5.29717863e-01 -2.57442981e-01
-3.93327057e-01 -1.02350628e+00 5.83427131e-01 6.28807664e-01
4.74991530e-01 -5.25541306e-01 -3.40224266e-01 2.78084695e-01
-1.67800754e-01 -9.70815837e-01 -9.69745219e-01 7.38358870e-02
-8.52054536e-01 -1.08122599e+00 -6.59172058e-01 -2.47207135e-01
5.68202376e-01 3.75266641e-01 1.05683589e+00 -1.09125994e-01
-1.73367471e-01 6.52738452e-01 -3.42224509e-01 -2.47575283e-01
-7.66454488e-02 3.70736301e-01 2.71790326e-02 2.00056225e-01
3.07247639e-01 -6.88571274e-01 -9.07256842e-01 6.18887007e-01
-9.05551851e-01 -5.34937263e-01 4.65422362e-01 1.09235644e+00
6.05300605e-01 -1.19168781e-01 7.87215233e-01 -1.25653720e+00
6.20782316e-01 -8.63584280e-01 -5.52521825e-01 4.99233574e-01
-9.56211805e-01 1.65238138e-02 8.14898610e-01 -7.51147687e-01
-1.07243621e+00 -3.70818347e-01 9.62839872e-02 -6.79153085e-01
2.04747126e-01 3.00039023e-01 -2.10307255e-01 1.92591567e-02
8.59250963e-01 2.82223761e-01 7.61295557e-02 -8.48724604e-01
3.66580367e-01 9.42126155e-01 5.33079267e-01 -1.02127361e+00
2.13382885e-01 4.77485538e-01 -3.73408079e-01 -1.91481382e-01
-8.70068789e-01 -2.88373351e-01 9.30348411e-02 2.87976116e-01
-4.91035581e-02 -8.96860301e-01 -1.32637239e+00 3.55372578e-01
-5.84274590e-01 -7.88160980e-01 -4.51072365e-01 4.59297597e-01
-4.39792812e-01 2.68149018e-01 -7.36804307e-01 -4.43349689e-01
-9.22155678e-01 -1.04496706e+00 7.56060243e-01 8.73158872e-02
1.38257101e-01 -8.78741086e-01 -1.56630993e-01 4.61150795e-01
8.37103903e-01 5.68764322e-02 8.72098088e-01 -8.26347411e-01
-5.05659878e-01 -2.04021364e-01 -1.34084731e-01 5.48588216e-01
9.05817822e-02 -3.86097819e-01 -1.03126097e+00 -7.66943395e-01
-9.78091806e-02 -3.87373239e-01 5.27302861e-01 1.21679105e-01
1.81221128e+00 -9.18625474e-01 -4.16924357e-01 1.14387608e+00
1.32149827e+00 -1.35440797e-01 2.54159570e-02 1.95624664e-01
3.15665275e-01 2.01308012e-01 1.49091780e-01 1.04901648e+00
4.60127413e-01 5.46820939e-01 5.19834936e-01 3.06209713e-01
-4.29393686e-02 -7.19071329e-02 4.15204734e-01 5.41048765e-01
4.70410168e-01 -4.29343611e-01 -7.10169971e-01 5.10248601e-01
-1.86486888e+00 -6.50127232e-01 1.04430050e-01 2.36363006e+00
1.11303306e+00 -2.46605203e-01 4.27112162e-01 -3.19991887e-01
5.76070368e-01 -5.24062753e-01 -1.35572040e+00 -5.37378192e-02
-2.99231589e-01 1.67426601e-01 7.66507447e-01 8.80445763e-02
-4.07919556e-01 4.72397923e-01 5.75922155e+00 9.35886264e-01
-1.15991843e+00 6.15182817e-01 7.23456919e-01 -7.44011939e-01
-3.27109843e-01 -4.67334956e-01 -7.62624502e-01 8.33206534e-01
1.06241322e+00 -4.36565399e-01 7.86238372e-01 1.37083983e+00
-2.80464858e-01 3.23594391e-01 -1.37886775e+00 1.11801445e+00
-1.75297186e-01 -1.41186500e+00 7.09149009e-03 1.51475400e-01
9.09923732e-01 6.41770244e-01 2.98152804e-01 4.92017150e-01
5.64491034e-01 -6.80172563e-01 3.18809837e-01 3.71836364e-01
7.31079221e-01 -9.63795245e-01 8.79032612e-01 5.47120333e-01
-6.24206185e-01 -5.17868042e-01 -4.14398402e-01 4.80282217e-01
-2.06250831e-01 7.48212874e-01 -7.93514132e-01 5.25280356e-01
1.15430892e+00 1.09372780e-01 -3.53787005e-01 1.08978748e+00
3.72376323e-01 8.19129825e-01 -6.74145103e-01 4.01219726e-02
9.76604596e-02 1.09278016e-01 1.90862134e-01 1.18848228e+00
4.47072208e-01 -6.97753131e-02 1.42679408e-01 5.91074049e-01
-5.41743398e-01 2.29732707e-01 -1.15581630e-02 3.32607090e-01
9.00045276e-01 1.02995729e+00 2.28306785e-01 -4.09296274e-01
-3.46262813e-01 9.82487917e-01 5.98987758e-01 3.98202747e-01
-1.06686783e+00 -2.59511918e-01 8.60536575e-01 7.80183598e-02
4.74668533e-01 3.24027717e-01 1.23734418e-02 -1.05947685e+00
2.50650734e-01 -1.04805160e+00 1.06774175e+00 -2.17413396e-01
-1.89956295e+00 6.23565078e-01 -2.87098736e-01 -8.64199758e-01
-5.50165549e-02 -2.52868652e-01 -3.29806328e-01 7.09207952e-01
-1.42799401e+00 -9.85896230e-01 -4.42134172e-01 1.03856099e+00
1.00795716e-01 -1.50128916e-01 7.87172437e-01 7.33467221e-01
-8.17057133e-01 1.68158436e+00 6.72407210e-01 -2.37277910e-01
9.59263146e-01 -1.07879031e+00 -7.31842518e-02 2.68287688e-01
-3.20148379e-01 6.36368454e-01 3.28989744e-01 -2.37553269e-01
-1.87616968e+00 -1.38185573e+00 3.69315684e-01 -3.27635944e-01
6.12948954e-01 -1.00669682e-01 -1.09579527e+00 7.68634617e-01
-1.77123621e-01 5.29778481e-01 1.02415490e+00 3.75289142e-01
-7.21943200e-01 -8.74031723e-01 -1.55657148e+00 4.30189639e-01
1.25874984e+00 -2.79182076e-01 1.23056300e-01 6.91449165e-01
7.31371045e-01 -5.19416809e-01 -1.23299718e+00 3.04168880e-01
4.35769647e-01 -1.04070890e+00 7.00291753e-01 -7.49447048e-01
-4.07504588e-01 1.79114625e-01 -3.16985786e-01 -1.22450554e+00
-1.91261113e-01 -1.12993741e+00 -4.75872993e-01 1.33168924e+00
3.56730759e-01 -1.29770565e+00 9.69705284e-01 8.13650191e-01
-5.19268997e-02 -1.11594331e+00 -1.01888263e+00 -1.19137990e+00
4.75940946e-03 -4.71000075e-01 1.21726286e+00 8.34174514e-01
2.20064130e-02 -5.77900885e-03 -9.87290516e-02 3.02834243e-01
6.54963672e-01 1.52791232e-01 1.01840043e+00 -8.44962060e-01
-7.60556817e-01 -4.90762740e-01 6.90823048e-02 -1.21236205e+00
8.11045319e-02 -1.01836276e+00 -4.05527443e-01 -8.99544954e-01
3.15930992e-01 -8.73299956e-01 -5.47434509e-01 9.04920936e-01
-1.96376860e-01 -2.64644027e-01 9.84151587e-02 2.80501246e-01
-7.86891520e-01 6.88927054e-01 9.24013734e-01 9.08210594e-03
-3.33454639e-01 3.10039520e-01 -8.81349266e-01 2.60537297e-01
7.92739213e-01 -4.08783168e-01 -5.41240752e-01 -6.29242599e-01
1.32954225e-01 -8.10506940e-02 1.96722060e-01 -7.28723884e-01
3.54063421e-01 -8.96751583e-02 7.92462826e-02 -2.49398649e-01
-3.45354788e-02 -8.13516855e-01 2.85110086e-01 4.09430891e-01
-2.07994878e-01 1.14786021e-01 1.40932560e-01 7.58894503e-01
1.21482961e-01 2.62959898e-01 7.55994618e-01 2.27094606e-01
-3.01721059e-02 7.31735826e-01 2.37418532e-01 2.65454948e-01
1.05179679e+00 1.96946904e-01 -8.29677582e-01 -2.28398666e-01
-4.70977783e-01 6.98003411e-01 3.68865579e-01 5.51673472e-02
2.11053222e-01 -1.19459307e+00 -5.57200670e-01 1.00000396e-01
-4.65439595e-02 1.81572393e-01 8.24301422e-01 1.02399349e+00
-2.86860555e-01 2.20229477e-01 2.65273482e-01 -6.84897006e-01
-1.00269854e+00 7.60744035e-01 4.56159830e-01 -2.12112159e-01
-7.35635042e-01 1.04845941e+00 1.71764895e-01 -6.08715296e-01
7.47208536e-01 -1.77337021e-01 6.19328558e-01 -2.40171000e-01
6.93287313e-01 9.02006507e-01 3.43591988e-01 9.93678719e-02
-5.13541162e-01 1.21892363e-01 -1.80587754e-01 4.12554055e-01
1.58861554e+00 -1.14363143e-02 9.52322185e-02 -7.26437056e-03
1.39130485e+00 2.28084028e-02 -1.56065392e+00 -7.79602528e-01
-2.50759214e-01 -5.20928442e-01 5.96827790e-02 -1.14647460e+00
-1.46162760e+00 4.18184966e-01 5.52394032e-01 4.94306982e-02
1.38675404e+00 3.12840901e-02 1.16980195e+00 5.02968013e-01
5.21131992e-01 -9.35071409e-01 -6.18801042e-02 6.36212230e-02
6.64833724e-01 -9.99828160e-01 -1.69091061e-01 -1.35255024e-01
-6.09236598e-01 8.22653353e-01 6.42862022e-01 2.57088453e-01
6.68400347e-01 4.06477243e-01 -2.12703511e-01 9.41060036e-02
-1.37638283e+00 2.98496485e-01 -6.35158494e-02 4.99918580e-01
7.32567580e-03 2.16491982e-01 5.15584014e-02 1.13558841e+00
-7.92429447e-02 1.44703090e-01 3.22122052e-02 8.04756522e-01
-1.83855128e-02 -9.73466992e-01 -3.06968629e-01 5.88886440e-01
-5.93726516e-01 1.27117693e-01 1.32957026e-01 5.29048741e-01
5.84184937e-02 7.87178934e-01 1.01531610e-01 -2.74885952e-01
4.95397896e-01 2.23171171e-02 2.84931958e-01 -2.06016287e-01
-6.92376852e-01 2.72500012e-02 -1.90641865e-01 -8.74704301e-01
1.92581207e-01 -5.94670415e-01 -1.24404430e+00 -8.70614171e-01
-3.12568665e-01 2.99914956e-01 4.85208601e-01 6.14758849e-01
1.25644600e+00 2.97459722e-01 9.54300940e-01 -2.06684619e-01
-1.43004858e+00 -8.23242843e-01 -6.26910627e-01 3.76423687e-01
1.66807026e-01 -3.45867485e-01 -5.78309596e-01 -3.09573799e-01] | [5.848239421844482, 6.272885322570801] |
0af30134-17e5-4c64-8ac1-527aaad2bf35 | comparing-computational-architectures-for | 2210.04107 | null | https://arxiv.org/abs/2210.04107v1 | https://arxiv.org/pdf/2210.04107v1.pdf | Comparing Computational Architectures for Automated Journalism | The majority of NLG systems have been designed following either a template-based or a pipeline-based architecture. Recent neural models for data-to-text generation have been proposed with an end-to-end deep learning flavor, which handles non-linguistic input in natural language without explicit intermediary representations. This study compares the most often employed methods for generating Brazilian Portuguese texts from structured data. Results suggest that explicit intermediate steps in the generation process produce better texts than the ones generated by neural end-to-end architectures, avoiding data hallucination while better generalizing to unseen inputs. Code and corpus are publicly available. | ['Fabio G. Cozman', 'Marcos M. José', 'João Gabriel M. Campos', 'Yan V. Sym'] | 2022-10-08 | null | null | null | null | ['data-to-text-generation'] | ['natural-language-processing'] | [ 8.20744131e-03 9.10920799e-01 3.44978720e-01 -4.01884466e-01
-8.01660836e-01 -5.26964843e-01 1.14373314e+00 1.92536220e-01
-3.68520230e-01 1.06099296e+00 7.66816139e-01 -3.67855161e-01
1.84490144e-01 -9.80280757e-01 -4.70806867e-01 -2.53183931e-01
4.47905928e-01 1.06462359e+00 -2.82987475e-01 -4.43753898e-01
1.55787230e-01 2.33879149e-01 -1.31194711e+00 1.00105846e+00
8.02833736e-01 4.99679327e-01 2.72883952e-01 8.27527761e-01
-5.25796533e-01 8.50382686e-01 -1.06240141e+00 -6.97984517e-01
1.14914440e-02 -7.69933820e-01 -1.00425231e+00 -2.02741519e-01
3.04691885e-02 -1.25443727e-01 -2.18964815e-01 5.36207020e-01
1.01830947e+00 9.65429172e-02 7.89722025e-01 -8.05368900e-01
-1.28728640e+00 1.14277482e+00 3.15771371e-01 -1.02078706e-01
6.81093633e-01 5.20006359e-01 6.06505990e-01 -1.16867650e+00
7.98190951e-01 1.41488814e+00 4.48571354e-01 9.32014763e-01
-1.27482378e+00 -1.88232422e-01 -3.66338372e-01 -1.79206580e-01
-1.06183493e+00 -6.41977787e-01 6.08327150e-01 -3.87196779e-01
1.50640380e+00 3.45568545e-02 4.71404910e-01 1.59146392e+00
1.45523772e-01 8.42918694e-01 7.61310518e-01 -6.37672246e-01
2.01974794e-01 2.50600219e-01 -3.26604664e-01 2.86337167e-01
3.74988943e-01 1.18843004e-01 -6.03181362e-01 -6.64558709e-02
3.32744092e-01 -6.78258896e-01 -8.98695290e-02 2.93801099e-01
-1.26387513e+00 1.04508913e+00 4.48679209e-01 4.83249307e-01
-8.04780245e-01 -2.00248331e-01 5.70478976e-01 2.39729092e-01
5.44816434e-01 9.40051138e-01 -3.00698340e-01 -1.23909347e-01
-1.15589499e+00 5.22228181e-01 1.08974612e+00 1.16421318e+00
4.07821447e-01 5.50710082e-01 -7.41576433e-01 5.08958519e-01
4.53023732e-01 3.22773784e-01 1.22863257e+00 -2.55057782e-01
8.41019034e-01 5.88723540e-01 1.74325168e-01 -7.09010243e-01
-4.60867763e-01 -3.65518689e-01 -6.61923051e-01 8.22829604e-02
4.49388862e-01 -6.64908469e-01 -9.92015898e-01 1.51973629e+00
-4.97430377e-02 -6.99829340e-01 7.46757686e-01 4.60610718e-01
1.25294113e+00 9.03404057e-01 2.64851630e-01 7.86455348e-02
1.09253883e+00 -9.77062106e-01 -1.01158655e+00 -2.75694460e-01
5.25950313e-01 -8.07739377e-01 1.25976610e+00 4.03212011e-01
-1.52024269e+00 -7.89759278e-01 -7.49269307e-01 -6.24369621e-01
-8.32576394e-01 2.62165993e-01 4.28431388e-03 7.29082406e-01
-1.35120749e+00 5.29370308e-01 -5.39866447e-01 -4.26980555e-01
3.93668979e-01 6.66192695e-02 -2.77562618e-01 3.74766946e-01
-1.38687730e+00 1.23717654e+00 1.09704375e+00 2.77996838e-01
-9.46379423e-01 -3.71846825e-01 -8.76757443e-01 -1.55030759e-02
1.52673079e-02 -9.88700271e-01 1.60475302e+00 -1.04435420e+00
-1.86872959e+00 7.52586544e-01 -1.10709116e-01 -8.60331655e-01
8.51781428e-01 -3.75269473e-01 -4.00930107e-01 1.47631383e-02
5.80721349e-02 9.13904905e-01 5.80080450e-01 -1.22517550e+00
-7.05358535e-02 -7.93423802e-02 -2.04499036e-01 1.85934797e-01
-1.69978663e-01 -5.54457568e-02 3.29253465e-01 -7.60607541e-01
-3.91992539e-01 -4.22451228e-01 -3.65884811e-01 -4.25307631e-01
-6.94047689e-01 -5.89576542e-01 5.43835402e-01 -8.12483788e-01
1.12177372e+00 -1.46174729e+00 2.06236467e-02 -4.13936645e-01
-1.83422521e-01 6.75131977e-01 -3.52163017e-01 1.20419836e+00
-9.22574773e-02 3.47196788e-01 -1.35476310e-02 -5.40587664e-01
2.69209057e-01 -4.05322798e-02 -5.67694604e-01 -2.41442636e-01
6.66834354e-01 1.22048604e+00 -1.17498851e+00 -3.42863351e-01
2.20647767e-01 5.22624612e-01 -4.89148378e-01 6.49268508e-01
-6.99217677e-01 3.21974248e-01 -2.43170992e-01 9.35485438e-02
2.95632601e-01 -1.27429441e-02 -6.79397881e-02 6.67416081e-02
-1.65071204e-01 7.26887584e-01 -6.47770286e-01 1.91789305e+00
-7.38248646e-01 7.70117581e-01 -3.11036587e-01 -5.14147937e-01
1.33642995e+00 9.49664831e-01 -2.77710497e-01 -4.85224694e-01
2.29750738e-01 3.73533487e-01 1.53311431e-01 -8.00927937e-01
9.49080408e-01 -4.59600717e-01 -5.61796427e-02 5.02702832e-01
3.39252859e-01 -3.96376014e-01 4.62308288e-01 6.96525127e-02
8.37858081e-01 5.88731587e-01 5.01940370e-01 -1.27574369e-01
4.61981565e-01 3.22767884e-01 -1.06972985e-01 8.42700660e-01
2.31820554e-01 9.50315177e-01 3.96883637e-01 -4.37004566e-01
-1.11655128e+00 -9.00072515e-01 2.73162365e-01 8.71438980e-01
-7.29929328e-01 -3.34766239e-01 -1.13340652e+00 -5.78701377e-01
-4.81465131e-01 1.62443209e+00 -6.30332410e-01 -2.11052284e-01
-4.47575271e-01 -3.80333602e-01 1.03465140e+00 3.80789548e-01
2.39791572e-01 -1.98263490e+00 -8.33259523e-01 7.76881874e-01
-3.86366814e-01 -8.45335305e-01 -6.18780777e-02 1.69152007e-01
-8.86264145e-01 -4.88437712e-01 -8.78627002e-01 -5.58258116e-01
6.29330695e-01 -5.14831781e-01 1.42216539e+00 -2.27101564e-01
3.78599092e-02 -1.67792380e-01 -5.94662845e-01 -7.87433803e-01
-1.12284553e+00 5.08438051e-01 -4.43444401e-01 -1.57022461e-01
4.39914256e-01 -3.19819480e-01 -2.75325447e-01 -3.75492275e-01
-1.21590447e+00 3.57604414e-01 8.04355562e-01 7.66238809e-01
8.94288942e-02 -6.03956521e-01 1.02890909e+00 -9.76566672e-01
1.48460972e+00 -7.40740657e-01 -9.01070237e-02 5.78896254e-02
-3.78109366e-01 4.56405461e-01 1.27420759e+00 -3.57298106e-01
-1.32594287e+00 1.61850095e-01 -5.65206647e-01 -1.69357941e-01
-7.15226948e-01 6.91778541e-01 -2.92138815e-01 9.10507560e-01
1.16078115e+00 4.83969957e-01 -9.66078043e-02 -5.16782820e-01
7.47910738e-01 8.35628033e-01 5.50746083e-01 -5.61273813e-01
5.61751366e-01 1.24696381e-01 -4.72053587e-01 -6.97445095e-01
-7.47564375e-01 2.45656222e-01 -7.95651257e-01 -3.90696898e-02
1.05183923e+00 -7.46033788e-01 -9.98058766e-02 4.12622362e-01
-1.78053832e+00 -5.10409355e-01 -7.29019761e-01 1.26563564e-01
-7.36290574e-01 -6.67970181e-02 -3.29198718e-01 -8.10770631e-01
-9.56809938e-01 -7.83592999e-01 1.13358462e+00 3.21199328e-01
-6.88020289e-01 -1.12944853e+00 3.72964919e-01 1.48237213e-01
7.02048123e-01 3.69303167e-01 8.45495820e-01 -1.31255031e+00
6.84997663e-02 -2.82661080e-01 3.55603546e-02 1.69619203e-01
3.45817185e-03 -7.79513195e-02 -1.22429907e+00 -5.85186742e-02
-6.53444827e-02 -4.56537843e-01 4.92205232e-01 -1.17121615e-01
6.74732685e-01 -8.46140742e-01 4.26848605e-02 2.22349167e-01
1.16720963e+00 1.37990892e-01 7.63810396e-01 1.72659203e-01
4.38036978e-01 1.04450381e+00 1.87453061e-01 3.50468993e-01
4.85132396e-01 2.81500727e-01 3.85464787e-01 -1.43925324e-01
-3.66836935e-01 -6.74288690e-01 6.27040505e-01 4.95794386e-01
2.90544480e-01 -9.64988530e-01 -1.05633736e+00 9.40792620e-01
-1.71549165e+00 -1.16265070e+00 -4.30229783e-01 1.81659496e+00
1.12198889e+00 1.21660315e-01 9.28578675e-02 -7.09611773e-02
6.08444214e-01 2.84556132e-02 -2.55507976e-01 -9.87698615e-01
-1.55220583e-01 5.31760633e-01 -1.65379509e-01 4.88463879e-01
-5.92195213e-01 1.02978468e+00 6.70292377e+00 4.46762621e-01
-1.12053001e+00 1.36093080e-01 4.76788402e-01 -1.63670287e-01
-4.91864890e-01 -1.56708792e-01 -6.43341064e-01 4.29672152e-01
1.47419155e+00 -4.61206615e-01 1.18935198e-01 7.67304957e-01
6.85093343e-01 2.64767110e-01 -1.33150458e+00 5.31571388e-01
1.41945630e-01 -1.35712481e+00 7.30628908e-01 -1.22337200e-01
6.67508066e-01 7.48509839e-02 -2.81796873e-01 4.67990041e-01
5.14600813e-01 -1.26508963e+00 1.21382701e+00 7.01363266e-01
5.10488272e-01 -5.27854085e-01 7.33188748e-01 5.12051523e-01
-5.81007123e-01 1.30002052e-01 -3.27299058e-01 -3.25377464e-01
4.19945568e-01 4.44698453e-01 -1.44103611e+00 7.67673731e-01
5.92891537e-02 1.66119978e-01 -8.64438236e-01 7.19321012e-01
-5.14270127e-01 4.64962333e-01 -6.63903281e-02 -4.06964689e-01
5.21624088e-01 6.48118705e-02 5.10929465e-01 1.68302763e+00
5.58273375e-01 -3.36090267e-01 -5.02162464e-02 1.43257940e+00
3.76098529e-02 3.75649154e-01 -1.02650392e+00 -3.35696250e-01
4.06343579e-01 1.23182213e+00 -5.05100012e-01 -4.84404981e-01
-4.16813456e-02 1.08656406e+00 3.49302441e-01 3.56162161e-01
-3.89782429e-01 -8.08197975e-01 -3.84475328e-02 2.55643368e-01
1.08328812e-01 1.36891077e-03 -4.59472299e-01 -8.74692380e-01
-8.84894356e-02 -1.03422391e+00 1.66876510e-01 -1.11561227e+00
-1.34486163e+00 1.23420107e+00 -3.61118093e-02 -8.58516335e-01
-1.18907320e+00 -5.67055643e-01 -8.90691757e-01 1.28729844e+00
-1.10746992e+00 -1.30991435e+00 -2.02471778e-01 4.07712758e-01
8.71466219e-01 -3.40739042e-01 1.19688749e+00 -1.00792632e-01
-2.96655118e-01 3.93537790e-01 -1.38845056e-01 2.61594981e-01
4.21917677e-01 -1.46651328e+00 7.77236342e-01 1.01418841e+00
1.25045106e-01 6.37861729e-01 8.56346250e-01 -8.14930916e-01
-1.00204396e+00 -1.26619923e+00 1.53150952e+00 -5.06164014e-01
3.83271188e-01 -5.14377534e-01 -9.52235103e-01 5.13773561e-01
1.29513621e+00 -3.91007572e-01 5.94681561e-01 -2.95096755e-01
-1.56697892e-02 3.50711226e-01 -1.26636076e+00 7.83953369e-01
7.68815041e-01 -3.74704480e-01 -1.01872981e+00 3.95709515e-01
6.47434711e-01 -4.16154981e-01 -5.63002229e-01 -9.18823108e-02
1.96105689e-01 -7.94676125e-01 3.85591626e-01 -8.90149772e-01
8.35805535e-01 -2.72670865e-01 3.11535358e-01 -1.74266958e+00
-3.95110510e-02 -1.16714311e+00 -7.43383765e-02 1.34583485e+00
8.93118799e-01 -3.97296578e-01 4.51484531e-01 4.14751709e-01
-4.36095357e-01 -4.12930906e-01 -8.37044299e-01 -5.38139343e-01
5.80741704e-01 -1.90802976e-01 7.36891508e-01 6.15418851e-01
-1.19122295e-02 8.71738255e-01 -2.02841580e-01 -3.44749272e-01
4.72028293e-02 -2.73132771e-01 8.37535381e-01 -1.02309632e+00
-1.59301907e-01 -6.82791352e-01 1.45184070e-01 -6.09184921e-01
1.22669913e-01 -1.31312382e+00 3.47035438e-01 -2.06476617e+00
-2.90158272e-01 1.85794830e-01 3.90637279e-01 6.31182551e-01
-2.17410564e-01 5.22640608e-02 2.42135644e-01 -7.90717304e-02
-1.72903314e-01 7.30048239e-01 1.20544422e+00 1.34125754e-01
-2.74262816e-01 -1.99436888e-01 -8.57312620e-01 4.06686366e-01
1.03627980e+00 -5.49644947e-01 -5.16439021e-01 -8.65278661e-01
3.74653757e-01 -1.06230915e-01 1.45204619e-01 -1.01244688e+00
4.96522635e-02 1.58860266e-01 5.13050437e-01 -6.36025131e-01
9.79797691e-02 -4.49810177e-01 1.63581759e-01 5.15939534e-01
-8.47456276e-01 3.43799382e-01 3.43763202e-01 5.37725091e-02
-2.67328441e-01 -7.72288561e-01 6.04894638e-01 -4.50760990e-01
3.13068517e-02 -2.03059331e-01 -8.45222652e-01 1.08999804e-01
7.82037139e-01 -1.42271325e-01 -4.65841502e-01 -7.76633203e-01
-8.59484732e-01 -1.00863554e-01 5.44882193e-02 6.07111812e-01
5.21320164e-01 -1.26826298e+00 -1.35381782e+00 1.09488390e-01
-1.38268424e-02 1.68383673e-01 -1.46402925e-01 2.67283022e-01
-6.75955296e-01 6.51158690e-01 -3.10071200e-01 -7.05753714e-02
-5.05982637e-01 5.74891806e-01 4.40999627e-01 -6.19434416e-01
-3.14699829e-01 4.68596339e-01 -2.90407270e-01 -7.43241191e-01
4.57935920e-03 -2.67728180e-01 -3.02753478e-01 3.67244929e-01
6.46341443e-01 3.32359672e-01 2.69735903e-01 -5.12241721e-01
1.24863207e-01 -3.39594722e-01 -3.69284377e-02 -6.51661038e-01
1.23257518e+00 2.04057917e-01 2.04983190e-01 5.01865149e-01
8.18010628e-01 -9.86015648e-02 -8.11644673e-01 -6.19621278e-05
3.47511411e-01 1.19592652e-01 -2.58465707e-01 -1.33673286e+00
-5.96667469e-01 1.29332912e+00 1.82970494e-01 5.60037971e-01
8.45440626e-01 -1.91013470e-01 5.71157396e-01 6.67175770e-01
-3.12131383e-02 -1.01368451e+00 1.09355807e-01 8.46421957e-01
1.55155635e+00 -9.65129375e-01 -5.06293058e-01 2.11316168e-01
-7.59556532e-01 1.46330833e+00 6.67101026e-01 -3.40414196e-02
1.07622348e-01 2.49076754e-01 2.90513337e-01 -5.92150651e-02
-1.07805455e+00 -2.06526026e-01 2.06342071e-01 1.00255203e+00
1.03624928e+00 -1.28441676e-01 -4.96346623e-01 8.27813268e-01
-7.58791983e-01 1.16638519e-01 6.83845818e-01 6.29222035e-01
-1.03525616e-01 -1.25549948e+00 -3.63122404e-01 2.18752384e-01
-4.32230085e-01 -5.01412928e-01 -9.97540295e-01 8.22297215e-01
4.37579341e-02 1.31342781e+00 6.24781623e-02 3.18476632e-02
5.65843880e-01 6.32463276e-01 2.82377720e-01 -1.12927413e+00
-1.39981067e+00 -1.87387004e-01 4.95134085e-01 -1.46304742e-01
-5.13258055e-02 -6.20114148e-01 -1.38276052e+00 1.35649359e-02
-9.70068350e-02 2.60037243e-01 6.64617300e-01 8.72859120e-01
6.66358292e-01 6.61032617e-01 4.02686089e-01 -1.04061663e+00
-6.08962655e-01 -1.78424668e+00 1.83955282e-01 4.21904534e-01
1.62417993e-01 2.25943178e-01 7.40605667e-02 2.49427035e-01] | [11.594347953796387, 9.059464454650879] |
e4e875e1-c058-45e7-8910-f002a738651f | do-we-train-on-test-data-the-impact-of-near | 2304.04653 | null | https://arxiv.org/abs/2304.04653v1 | https://arxiv.org/pdf/2304.04653v1.pdf | Do We Train on Test Data? The Impact of Near-Duplicates on License Plate Recognition | This work draws attention to the large fraction of near-duplicates in the training and test sets of datasets widely adopted in License Plate Recognition (LPR) research. These duplicates refer to images that, although different, show the same license plate. Our experiments, conducted on the two most popular datasets in the field, show a substantial decrease in recognition rate when six well-known models are trained and tested under fair splits, that is, in the absence of duplicates in the training and test sets. Moreover, in one of the datasets, the ranking of models changed considerably when they were trained and tested under duplicate-free splits. These findings suggest that such duplicates have significantly biased the evaluation and development of deep learning-based models for LPR. The list of near-duplicates we have found and proposals for fair splits are publicly available for further research at https://raysonlaroca.github.io/supp/lpr-train-on-test/ | ['David Menotti', 'Rodrigo Minetto', 'Alceu S. Britto Jr.', 'Valter Estevam', 'Rayson Laroca'] | 2023-04-10 | null | null | null | null | ['license-plate-recognition'] | ['computer-vision'] | [-1.31712094e-01 -4.77904797e-01 -8.97659361e-03 -4.72120792e-01
-1.07963550e+00 -9.01203930e-01 8.89413476e-01 -6.32364690e-01
-4.17039752e-01 7.00106740e-01 -5.38097806e-02 -2.99308300e-01
5.07822409e-02 -5.38345993e-01 -1.05185652e+00 -5.03887475e-01
2.85678267e-01 5.54751515e-01 2.56561995e-01 2.15727016e-02
5.19725919e-01 4.18493569e-01 -1.55861413e+00 4.70470339e-01
5.93515337e-01 7.96202123e-01 -3.53183299e-02 3.65504920e-01
2.78753489e-01 6.96432292e-01 -6.96791351e-01 -1.14401078e+00
9.01172161e-01 -1.58398390e-01 -5.23239553e-01 2.92769492e-01
1.05899262e+00 -2.96654344e-01 -9.78257000e-01 9.65808809e-01
3.51992279e-01 -1.79728959e-02 6.25479460e-01 -1.25904965e+00
-8.56680512e-01 3.49916816e-01 -4.94624466e-01 3.78781140e-01
4.45632413e-02 4.50311869e-01 8.66731524e-01 -1.26892662e+00
7.22784877e-01 7.94727027e-01 9.53804731e-01 4.62701589e-01
-9.60300863e-01 -9.62499976e-01 -3.43583256e-01 3.68232340e-01
-1.79099786e+00 -9.04679954e-01 5.90468943e-01 -6.25745535e-01
7.02471435e-01 2.94628561e-01 2.31300473e-01 1.23217392e+00
1.91128105e-01 5.91751814e-01 1.22767854e+00 -3.89191687e-01
7.82730356e-02 4.05221432e-01 1.61267057e-01 2.90668488e-01
5.44346035e-01 7.65346885e-02 -4.68539417e-01 4.08305041e-02
5.94189167e-01 1.06835514e-02 -2.21787557e-01 -5.05562983e-02
-8.42660487e-01 5.46569765e-01 -2.93165445e-02 2.73302674e-01
-7.85254836e-02 -1.63709581e-01 3.01484019e-01 2.71980166e-01
3.95246148e-01 6.05572701e-01 -3.98364633e-01 -4.30647656e-03
-1.14337659e+00 1.65133655e-01 6.73039317e-01 1.15230203e+00
7.80664623e-01 -1.60012946e-01 3.91787477e-02 1.04431951e+00
7.52127767e-02 2.96651065e-01 6.72232509e-01 -8.03520262e-01
7.48410106e-01 4.36055720e-01 -2.50320532e-03 -9.14123118e-01
8.29322711e-02 -4.12789881e-01 -3.27147692e-01 2.46862933e-01
5.96722960e-01 2.36408580e-02 -9.11523640e-01 1.16316974e+00
-5.45213282e-01 4.32776809e-01 6.65083453e-02 7.82046854e-01
9.82493162e-01 2.94976979e-01 -4.89185490e-02 4.52610880e-01
1.30073047e+00 -9.13072705e-01 -2.28582144e-01 -3.83501709e-01
5.76117694e-01 -9.77449834e-01 9.82114851e-01 2.57050872e-01
-7.61378944e-01 -5.90481639e-01 -1.26709425e+00 1.68894634e-01
-5.59082866e-01 5.22736013e-01 1.39465451e-01 6.72540069e-01
-1.01796377e+00 5.51644146e-01 -2.20345706e-01 -3.40084940e-01
8.42054069e-01 2.67006308e-01 -8.41085613e-01 -5.46427548e-01
-1.06316841e+00 9.79204595e-01 8.45048353e-02 1.14393912e-01
-9.26531017e-01 -5.36629438e-01 -4.33204025e-01 -1.80468962e-01
3.50280672e-01 6.12227805e-02 1.08908987e+00 -1.00359356e+00
-1.05846524e+00 1.45950532e+00 8.13998654e-02 -3.23226631e-01
9.32594121e-01 -1.60181180e-01 -7.37902582e-01 -1.95395261e-01
4.98102531e-02 4.90335166e-01 6.74998760e-01 -1.35686994e+00
-5.98540545e-01 -4.10003364e-01 -3.20737697e-02 -1.38724633e-02
1.59869090e-01 2.11446375e-01 -8.36816430e-01 -4.67208385e-01
-6.32868558e-02 -1.20557547e+00 1.58158556e-01 -3.02501768e-01
-5.12086689e-01 -1.43470570e-01 6.01954937e-01 -7.70906389e-01
8.07856023e-01 -2.42341185e+00 -6.21141493e-01 -1.28844408e-02
1.56571597e-01 5.43983698e-01 -2.22323596e-01 3.59080255e-01
-2.51590222e-01 6.40740395e-01 -9.64289904e-02 -4.39099520e-01
1.59787819e-01 9.34816722e-04 -4.53949600e-01 7.41952538e-01
3.25245082e-01 7.66351104e-01 -2.74830401e-01 -2.47244835e-01
1.46110609e-01 4.96086292e-02 -4.47371639e-02 -1.50971591e-01
3.06107759e-01 5.74274808e-02 1.47931948e-02 6.60177588e-01
1.10606515e+00 -7.60413930e-02 -1.11353248e-01 -5.90141565e-02
-2.42444023e-01 2.71354824e-01 -9.08290446e-01 1.11740792e+00
-8.54259580e-02 1.34927475e+00 -5.18182576e-01 -5.59550226e-01
1.03409386e+00 1.67750061e-01 6.71862066e-02 -1.02986979e+00
8.94885734e-02 5.77216506e-01 6.44183904e-02 -3.65732163e-01
6.81989312e-01 -1.65403664e-01 5.65155922e-03 1.83066070e-01
5.48327118e-02 1.69336841e-01 3.99931371e-01 -1.24031805e-01
9.84874666e-01 -1.34466171e-01 -2.66261071e-01 -1.34830743e-01
5.14886498e-01 5.63820340e-02 6.81269765e-01 9.91151989e-01
-5.77033699e-01 1.15045714e+00 4.50803876e-01 -3.25819373e-01
-1.33095694e+00 -1.05905867e+00 -4.62433964e-01 7.94053316e-01
1.40011206e-01 -2.19997093e-02 -3.96966070e-01 -6.58978045e-01
1.66691676e-01 7.74934590e-01 -6.81395292e-01 -5.29209450e-02
-5.47141552e-01 -6.22552156e-01 1.16909695e+00 3.18690896e-01
6.11565948e-01 -7.00921834e-01 -2.80852653e-02 -2.66036302e-01
5.06395176e-02 -1.22937572e+00 -8.62190276e-02 1.20148854e-03
-6.63258612e-01 -1.25039589e+00 -8.95430088e-01 -7.64451146e-01
4.62620407e-01 3.51028144e-01 1.19502783e+00 2.32646629e-01
-5.40782623e-02 2.03635663e-01 -3.11748981e-01 -5.33559501e-01
-6.04577541e-01 -9.61175933e-02 -1.00940198e-01 -7.60164484e-02
8.01318288e-01 -2.21617445e-01 -2.96230525e-01 8.94547224e-01
-8.86129260e-01 -2.37073153e-01 7.48228669e-01 2.76319236e-01
2.88237482e-01 -5.55889234e-02 5.57329655e-01 -6.45504951e-01
5.39255798e-01 -5.82748353e-01 -7.62273133e-01 3.91496003e-01
-3.15729886e-01 -3.25344861e-01 5.56034923e-01 -2.57779330e-01
-7.53636718e-01 -2.74238199e-01 -9.31843147e-02 -7.57504940e-01
-5.68613231e-01 2.28866011e-01 -4.88871746e-02 -2.10746557e-01
6.39063418e-01 4.14589047e-01 -1.75436229e-01 -4.70063508e-01
-1.97343677e-01 9.10429418e-01 4.00314063e-01 -2.42764518e-01
9.21121895e-01 3.34181875e-01 -3.35864395e-01 -8.11566472e-01
-2.37993315e-01 -5.93348861e-01 -5.10310233e-01 -4.04596835e-01
4.34651673e-01 -1.01608336e+00 -3.50357264e-01 9.55391109e-01
-8.73841107e-01 -1.58100903e-01 -1.34977013e-01 7.30448663e-01
-4.00275052e-01 2.91164279e-01 -4.49382931e-01 -4.78998691e-01
2.96327978e-01 -1.41506839e+00 7.77254283e-01 3.75424296e-01
-2.32887603e-02 -6.93442345e-01 1.84509471e-01 7.64170110e-01
3.73835206e-01 -1.95698678e-01 5.86354852e-01 -1.34494662e+00
-5.68192959e-01 -6.28268719e-01 -1.93725705e-01 5.99376738e-01
-4.32649069e-03 3.82263958e-01 -1.34109938e+00 -5.85046113e-02
-3.56584191e-01 -4.93851066e-01 1.03538191e+00 2.18851402e-01
6.85715735e-01 -8.44979957e-02 -1.70739725e-01 3.88904184e-01
1.52633858e+00 2.54013062e-01 1.28281546e+00 5.82762241e-01
3.23262870e-01 4.74448532e-01 2.06445694e-01 2.11542681e-01
-4.39164601e-03 7.23813653e-01 7.52192587e-02 4.75759953e-02
-2.28601411e-01 -9.92063954e-02 4.68112081e-01 5.33007324e-01
-1.64698809e-01 -4.86273468e-01 -1.23609567e+00 5.39055169e-01
-1.38655937e+00 -1.04208517e+00 -3.43362033e-01 2.33125114e+00
4.78303879e-01 2.35230073e-01 2.62177307e-02 -1.72787786e-01
1.12849307e+00 2.07543179e-01 -4.25867945e-01 -1.70443788e-01
-5.07621348e-01 -1.42654851e-01 9.26614583e-01 1.92741081e-01
-1.28319907e+00 8.66316438e-01 6.74842358e+00 9.22162652e-01
-1.36678863e+00 -5.69961704e-02 8.48672032e-01 -7.98044130e-02
3.42263980e-03 -1.61490694e-01 -8.11374664e-01 6.14608705e-01
1.05474961e+00 -2.10003272e-01 1.60258457e-01 7.88948596e-01
3.15162651e-02 -8.71336013e-02 -1.19190562e+00 1.10583973e+00
2.63945818e-01 -1.53115320e+00 -8.05478767e-02 2.43306994e-01
7.10093617e-01 6.43059015e-01 3.94454807e-01 5.47874689e-01
1.23526827e-01 -1.05265129e+00 9.28295851e-01 6.17636144e-01
8.48391414e-01 -5.73200166e-01 1.03844917e+00 4.33737896e-02
-6.36973441e-01 -1.47538353e-02 -7.24758327e-01 1.70223191e-01
-3.50696862e-01 4.09811914e-01 -9.63431060e-01 5.38794219e-01
6.73027813e-01 7.38086700e-01 -1.21277821e+00 1.41302657e+00
6.28325865e-02 8.97298634e-01 -4.38713767e-02 1.89460188e-01
2.32748911e-01 -2.53647089e-01 5.64620554e-01 1.28951061e+00
2.93345869e-01 -4.01542544e-01 -3.86019170e-01 9.37365472e-01
-5.04857123e-01 -1.53058052e-01 -7.27350831e-01 -3.44863050e-02
4.72698748e-01 8.95315528e-01 -6.16719246e-01 -1.96392521e-01
-7.96626627e-01 6.65397346e-01 4.75675575e-02 4.34383094e-01
-1.19754088e+00 -2.19845325e-01 6.22641146e-01 4.28408802e-01
3.92495006e-01 -7.81344548e-02 -2.11652204e-01 -1.28103483e+00
5.04992604e-01 -9.01721895e-01 2.26195529e-01 -7.97306240e-01
-1.64897883e+00 4.93862778e-01 -6.40915260e-02 -1.67387092e+00
1.00880206e-01 -9.90396321e-01 -7.88287878e-01 9.22009230e-01
-1.36407387e+00 -9.54243779e-01 -7.88148344e-02 4.53511447e-01
6.04114711e-01 -5.95408618e-01 3.86993676e-01 6.51511729e-01
-7.79285014e-01 9.17278469e-01 6.10679567e-01 8.12290132e-01
1.00282717e+00 -6.37961507e-01 5.58792472e-01 9.68402028e-01
1.26744390e-01 5.77251732e-01 5.89632273e-01 -5.29018104e-01
-1.00476015e+00 -1.02641535e+00 6.16978884e-01 -5.73030889e-01
5.70436537e-01 -3.40147823e-01 -1.03591299e+00 8.62254381e-01
8.78784209e-02 -1.92385986e-02 1.00494397e+00 -8.56227577e-02
-6.45989239e-01 -6.04553744e-02 -1.07542861e+00 5.49444854e-01
7.01622009e-01 -5.62630355e-01 -7.52540290e-01 2.38011569e-01
1.27050102e-01 -5.74806370e-02 -6.84685349e-01 1.03347525e-01
7.77570128e-01 -1.16664338e+00 7.32301533e-01 -7.13062882e-01
7.31221497e-01 -3.86349231e-01 -4.77387577e-01 -8.65827978e-01
-3.50181997e-01 3.07805799e-02 6.49470448e-01 1.29415739e+00
7.36565232e-01 -8.95349145e-01 7.72676289e-01 9.45510387e-01
-4.01072443e-01 -6.52658463e-01 -1.20991814e+00 -1.24816394e+00
4.01793808e-01 -4.58972007e-01 4.31847781e-01 9.48702335e-01
-4.85454470e-01 -1.24906220e-01 -2.02465132e-01 9.51157957e-02
3.03807169e-01 -1.98381618e-01 8.72007012e-01 -1.21226060e+00
-6.54102266e-02 -5.78351498e-01 -9.32010174e-01 -6.55851662e-01
2.56146878e-01 -9.18118715e-01 4.94532287e-02 -1.32140803e+00
4.07640934e-01 -3.35883081e-01 -2.41192043e-01 4.12307858e-01
1.71425581e-01 7.57044137e-01 5.58959544e-01 8.59572411e-01
-7.70444274e-01 2.58116484e-01 9.32621241e-01 -2.15955585e-01
2.14930803e-01 1.04648292e-01 -6.57988727e-01 7.49881208e-01
8.90495121e-01 -6.96069539e-01 4.80595939e-02 -4.70037103e-01
1.39050931e-01 -4.48924869e-01 5.46851099e-01 -1.29382849e+00
2.61492580e-01 1.07434720e-01 6.43145978e-01 -4.57754135e-01
2.80977100e-01 -7.77948737e-01 4.10844892e-01 1.91149190e-01
-2.62900382e-01 1.38688847e-01 5.35042286e-01 2.46913671e-01
-3.14094037e-01 -5.42171896e-01 7.90498376e-01 -1.73328832e-01
-1.09528601e+00 5.20038605e-03 -6.33180559e-01 1.99692279e-01
1.09611499e+00 -7.72593617e-01 -6.83130741e-01 -1.95291311e-01
-4.55630332e-01 -1.81723654e-01 7.87239254e-01 6.46799922e-01
5.09557366e-01 -1.21154165e+00 -1.14340305e+00 2.93572489e-02
4.64436591e-01 -4.86920297e-01 3.37901860e-01 6.97818518e-01
-7.13498175e-01 6.30243182e-01 -4.80375946e-01 -5.00420272e-01
-1.17946863e+00 1.68803230e-01 6.21666610e-01 -1.89870149e-01
-3.69174808e-01 6.29712820e-01 2.57789940e-02 -5.69503725e-01
-7.54471496e-02 4.45654206e-02 1.17743939e-01 5.48582673e-02
4.64586467e-01 5.98610342e-01 4.32570010e-01 -1.10485339e+00
-5.75997710e-01 3.48628223e-01 -3.05836767e-01 -1.94613989e-02
1.34620011e+00 2.31348187e-01 5.34070209e-02 4.58760738e-01
1.29585958e+00 1.70732826e-01 -1.18557477e+00 -2.24787407e-02
-6.79609552e-02 -9.01704669e-01 -3.53074759e-01 -7.91301429e-01
-1.05486381e+00 5.50942600e-01 5.67400932e-01 -1.04292713e-01
6.36967778e-01 7.68639445e-02 2.28832796e-01 5.85433006e-01
4.02984262e-01 -1.02961087e+00 -3.27206969e-01 7.16830671e-01
1.07794893e+00 -1.20943999e+00 1.43049091e-01 6.13930775e-03
-6.50173366e-01 1.19156218e+00 6.49448276e-01 -3.38427395e-01
4.15139616e-01 1.74226075e-01 2.01411545e-01 3.89670469e-02
-5.98814189e-01 1.83268800e-01 1.67293057e-01 5.39876103e-01
4.24243301e-01 -1.40461437e-02 -5.04463650e-02 6.14928365e-01
-1.76440760e-01 9.05979872e-02 8.62581551e-01 7.21256971e-01
-1.32605150e-01 -7.21002877e-01 -5.73740363e-01 7.35621154e-01
-3.85325134e-01 -8.14705640e-02 -7.49192536e-01 1.25329494e+00
2.62936145e-01 9.09827650e-01 3.06139588e-01 -5.98380744e-01
4.68990564e-01 4.09432620e-01 2.18045101e-01 -3.13473880e-01
-5.03636956e-01 -4.69474286e-01 7.80389458e-02 -2.23897681e-01
-2.02435985e-01 -7.70813286e-01 -7.32488692e-01 -5.79429686e-01
-4.26945895e-01 -1.01181187e-01 4.81397986e-01 9.91069555e-01
4.36871499e-01 2.99672842e-01 4.39357996e-01 -8.31388950e-01
-5.58926225e-01 -1.06368959e+00 -7.62157977e-01 6.59593284e-01
5.13353460e-02 -6.31704688e-01 -6.38243616e-01 1.97500184e-01] | [9.845483779907227, -4.907479763031006] |
c4a8d58a-cfbc-4408-a055-4d4e6c41d7f0 | online-training-through-time-for-spiking | 2210.04195 | null | https://arxiv.org/abs/2210.04195v2 | https://arxiv.org/pdf/2210.04195v2.pdf | Online Training Through Time for Spiking Neural Networks | Spiking neural networks (SNNs) are promising brain-inspired energy-efficient models. Recent progress in training methods has enabled successful deep SNNs on large-scale tasks with low latency. Particularly, backpropagation through time (BPTT) with surrogate gradients (SG) is popularly used to achieve high performance in a very small number of time steps. However, it is at the cost of large memory consumption for training, lack of theoretical clarity for optimization, and inconsistency with the online property of biological learning and rules on neuromorphic hardware. Other works connect spike representations of SNNs with equivalent artificial neural network formulation and train SNNs by gradients from equivalent mappings to ensure descent directions. But they fail to achieve low latency and are also not online. In this work, we propose online training through time (OTTT) for SNNs, which is derived from BPTT to enable forward-in-time learning by tracking presynaptic activities and leveraging instantaneous loss and gradients. Meanwhile, we theoretically analyze and prove that gradients of OTTT can provide a similar descent direction for optimization as gradients based on spike representations under both feedforward and recurrent conditions. OTTT only requires constant training memory costs agnostic to time steps, avoiding the significant memory costs of BPTT for GPU training. Furthermore, the update rule of OTTT is in the form of three-factor Hebbian learning, which could pave a path for online on-chip learning. With OTTT, it is the first time that two mainstream supervised SNN training methods, BPTT with SG and spike representation-based training, are connected, and meanwhile in a biologically plausible form. Experiments on CIFAR-10, CIFAR-100, ImageNet, and CIFAR10-DVS demonstrate the superior performance of our method on large-scale static and neuromorphic datasets in small time steps. | ['Zhouchen Lin', 'Di He', 'Zongpeng Zhang', 'Qingyan Meng', 'Mingqing Xiao'] | 2022-10-09 | null | null | null | null | ['event-data-classification', 'gesture-recognition'] | ['computer-vision', 'computer-vision'] | [ 5.93943819e-02 -4.65616673e-01 2.66509652e-01 -2.29593948e-01
-6.87970370e-02 -2.24614158e-01 1.88182741e-01 -1.80720061e-01
-7.70430923e-01 9.92686093e-01 -3.29348534e-01 -2.55389154e-01
-1.22026257e-01 -9.02082145e-01 -1.25472331e+00 -9.05635715e-01
-4.96846586e-02 -5.15830927e-02 5.49444079e-01 -3.54582965e-01
2.90294826e-01 5.62148571e-01 -1.29745042e+00 1.11256048e-01
6.91334426e-01 1.20815957e+00 4.97381330e-01 3.27792823e-01
-2.21235737e-01 6.24262691e-01 -3.74180555e-01 -8.13795775e-02
3.80799919e-01 -5.55536807e-01 -2.13495210e-01 -5.79373896e-01
1.33052804e-02 5.63285947e-02 -7.76460707e-01 8.40863049e-01
8.62638175e-01 -8.62334371e-02 3.19541216e-01 -1.03375578e+00
-6.90583169e-01 6.45728290e-01 -1.99993566e-01 3.62538308e-01
-4.92122084e-01 3.98476005e-01 3.88701528e-01 -7.45694220e-01
4.08703357e-01 7.81613052e-01 8.93798172e-01 9.64045346e-01
-1.29974163e+00 -1.07555580e+00 2.57299155e-01 1.78476900e-01
-1.21896458e+00 -3.30615997e-01 6.03163064e-01 -3.45851816e-02
1.43393517e+00 -2.23238189e-02 1.30094230e+00 1.14173591e+00
4.57842797e-01 6.22695863e-01 1.26418900e+00 -7.79408291e-02
7.02040315e-01 -3.27759951e-01 2.26864457e-01 7.06405580e-01
2.94578761e-01 2.31782332e-01 -8.49540234e-01 1.33491501e-01
1.20831466e+00 2.03384936e-01 -4.43340153e-01 1.17545441e-01
-1.14628255e+00 3.94332916e-01 9.85800445e-01 4.00234580e-01
-1.58222005e-01 7.80097842e-01 4.86672819e-01 3.26006591e-01
-5.11269122e-02 2.47132480e-01 -3.82649690e-01 -1.86512306e-01
-8.77765894e-01 -2.74710376e-02 6.63586497e-01 6.58720136e-01
7.11868286e-01 6.76217914e-01 -1.34523943e-01 6.93062842e-01
9.41556171e-02 6.34136796e-01 8.70802104e-01 -7.66744435e-01
1.08099490e-01 5.57382047e-01 -3.61682087e-01 -6.07212842e-01
-5.78324378e-01 -6.90865815e-01 -1.20182467e+00 4.29307878e-01
3.37629825e-01 -3.78490686e-02 -1.09500396e+00 1.98831952e+00
-1.50268838e-01 4.91471380e-01 -1.38220519e-01 9.93886888e-01
4.72419292e-01 8.14389467e-01 -1.00044541e-01 -2.45031536e-01
1.11900687e+00 -8.27735007e-01 -5.04640758e-01 -4.15010035e-01
4.96030003e-01 5.77477515e-02 1.07191229e+00 2.65791774e-01
-1.10412133e+00 -4.56040472e-01 -1.26221645e+00 -8.51729289e-02
-4.98125404e-01 -1.73026398e-01 9.19395983e-01 5.31989992e-01
-1.32535255e+00 9.23981726e-01 -1.20995069e+00 -2.15488970e-01
6.63038731e-01 7.69190609e-01 8.40671584e-02 2.44609341e-01
-9.50177014e-01 6.99136794e-01 2.26121485e-01 2.91351885e-01
-1.01604152e+00 -8.61393094e-01 -3.23366821e-01 1.19642369e-01
-3.99754494e-01 -7.92805433e-01 7.53119886e-01 -8.67288947e-01
-1.93687809e+00 6.13533497e-01 -3.42702746e-01 -9.72131371e-01
2.05618158e-01 1.99519292e-01 -2.69427210e-01 -2.15084717e-01
-4.24343884e-01 8.36124420e-01 6.23991907e-01 -7.34814942e-01
-1.34988815e-01 -4.08026695e-01 -1.33645564e-01 -2.69501626e-01
-7.98140347e-01 -1.85692325e-01 -1.48241073e-01 -6.11842394e-01
3.19069207e-01 -8.25486302e-01 -2.09753856e-01 4.46791500e-01
8.90517756e-02 -6.25885651e-02 5.95662713e-01 -2.75971025e-01
1.03996110e+00 -2.02102733e+00 1.32512162e-02 1.11590028e-01
1.39671996e-01 2.73191035e-01 -2.53637195e-01 5.64975105e-02
2.24814504e-01 -3.59658487e-02 -5.04882753e-01 -1.10679744e-02
-1.00808181e-01 2.75537223e-01 -5.82218170e-01 3.08553308e-01
2.24123314e-01 1.36651146e+00 -7.29761899e-01 -1.25967816e-01
-1.73380390e-01 5.82493603e-01 -5.97735286e-01 -1.18648939e-01
-5.32968305e-02 4.05651599e-01 -1.18096396e-01 5.69640219e-01
4.79378462e-01 -3.05301309e-01 -1.55325100e-01 -2.97388226e-01
-4.24239904e-01 3.76609087e-01 -6.69672787e-01 1.93502557e+00
-4.96677458e-01 6.66127205e-01 -1.66006647e-02 -1.25562847e+00
1.14071512e+00 1.75790526e-02 2.95168400e-01 -1.31361616e+00
1.83023363e-01 7.62093186e-01 1.93843082e-01 5.96729778e-02
-2.92323232e-01 2.48813629e-03 2.61094987e-01 2.93776959e-01
3.06675971e-01 1.42617136e-01 1.14225879e-01 -8.32295567e-02
1.36815906e+00 1.13489605e-01 -3.64578426e-01 -4.65519667e-01
1.24707781e-01 -3.78370136e-01 7.57129908e-01 6.93620980e-01
-1.03804491e-01 3.25999826e-01 1.52387423e-02 -4.49782431e-01
-1.05738652e+00 -1.13869643e+00 -4.36026335e-01 7.94080079e-01
1.98646173e-01 -7.33606592e-02 -6.99508905e-01 7.70065337e-02
-1.67472497e-01 1.86174139e-01 -4.19820994e-01 -4.36085314e-01
-9.51883376e-01 -1.02929592e+00 1.01157558e+00 7.21563578e-01
8.47269595e-01 -1.20476556e+00 -9.61905897e-01 6.77707553e-01
3.42717737e-01 -8.87739837e-01 -3.48496795e-01 9.43451285e-01
-1.37264514e+00 -5.25308788e-01 -7.90870965e-01 -1.07508647e+00
7.49579489e-01 -3.67989764e-02 6.69259906e-01 3.17015760e-02
-4.72118080e-01 -1.79388866e-01 1.84894130e-01 -3.52197111e-01
3.48737121e-01 -1.19002543e-01 1.93581417e-01 -2.33952910e-01
2.23945335e-01 -1.34016383e+00 -9.65643048e-01 2.28179842e-01
-6.93231761e-01 3.00560534e-01 5.33645034e-01 1.11085224e+00
1.02207661e+00 -4.29256797e-01 8.82113874e-01 -4.39230353e-01
3.97981137e-01 -3.77054095e-01 -7.52242327e-01 1.91726774e-01
-7.92252004e-01 3.54959100e-01 1.08484995e+00 -7.32438862e-01
-5.95739424e-01 -5.09079508e-02 -3.52232248e-01 -4.48861599e-01
3.70350242e-01 2.58549005e-01 2.33590245e-01 -6.77721322e-01
7.98022211e-01 8.41754973e-01 -3.14249456e-01 -2.14285150e-01
-6.50340915e-02 1.05566688e-01 4.87038910e-01 -6.88519299e-01
4.65722859e-01 6.86628222e-01 9.44472104e-02 -3.63520801e-01
-2.95591503e-01 1.24268338e-01 8.45786482e-02 -1.43826157e-01
5.20735860e-01 -7.82292545e-01 -1.00115669e+00 9.37016428e-01
-1.17024219e+00 -7.37912893e-01 -2.81428844e-01 5.07183313e-01
-5.11845410e-01 -1.32134140e-01 -1.19455183e+00 -7.67776906e-01
-7.92529285e-01 -7.94122815e-01 4.65753883e-01 5.93244195e-01
3.03257972e-01 -7.90547907e-01 -5.45557588e-03 -3.49158734e-01
9.89654601e-01 7.52646178e-02 9.86446917e-01 -1.93816647e-01
-7.84564018e-01 8.92809182e-02 -3.08585763e-01 1.02187008e-01
-1.93164498e-01 -2.52204597e-01 -1.09412444e+00 -4.69914496e-01
3.23427647e-01 -5.70660233e-01 1.17063868e+00 6.09892309e-01
1.54315579e+00 -2.64176577e-01 -4.39203620e-01 1.25164759e+00
1.81815147e+00 3.73739243e-01 7.87178993e-01 3.73494834e-01
5.08513927e-01 6.23808280e-02 -3.50672871e-01 3.11237603e-01
1.08654067e-01 3.89148116e-01 4.39853966e-01 -3.78506929e-02
-1.98662236e-01 -3.26497078e-01 6.45041823e-01 1.49652529e+00
-1.50063515e-01 1.06163189e-01 -6.76981330e-01 4.38878983e-01
-1.87308776e+00 -8.05768013e-01 9.89283398e-02 2.23846865e+00
1.08128822e+00 3.77333432e-01 -1.96760163e-01 1.47641435e-01
5.70690095e-01 -2.40949810e-01 -1.35060275e+00 -3.72061908e-01
-5.95908940e-01 8.29415798e-01 7.13793635e-01 -5.25122099e-02
-3.98702174e-01 6.32068992e-01 5.54628849e+00 9.17825878e-01
-1.62234116e+00 2.66252637e-01 5.96466362e-01 -3.80905330e-01
-3.43865097e-01 -1.77398380e-02 -1.03176022e+00 7.58483171e-01
1.32707524e+00 -1.93374798e-01 9.22940791e-01 5.46773732e-01
1.12297438e-01 3.08593869e-01 -1.02095878e+00 1.32397604e+00
-5.35657465e-01 -1.71406412e+00 -3.88172418e-02 -1.28247753e-01
6.97585464e-01 6.07848704e-01 -2.78290361e-02 4.90758687e-01
5.74508160e-02 -1.04133928e+00 8.67577136e-01 4.81360674e-01
6.05233729e-01 -5.53055406e-01 3.20101857e-01 4.36454654e-01
-1.31077838e+00 -2.87776768e-01 -8.51516247e-01 -1.97093815e-01
1.80630311e-02 7.77588248e-01 5.01991212e-02 -4.15737294e-02
9.27646279e-01 8.47907484e-01 -2.54291803e-01 1.35237837e+00
2.08140700e-03 7.78080463e-01 -6.91609621e-01 -6.13349020e-01
1.34789363e-01 -2.11195573e-01 1.06397890e-01 1.28584623e+00
5.69318354e-01 1.15954593e-01 -3.47196043e-01 1.38784719e+00
-4.25639391e-01 -1.47573054e-01 -3.84210587e-01 -1.15595661e-01
7.88551152e-01 1.17142117e+00 -8.98655593e-01 -3.46189812e-02
-1.26784995e-01 9.66627538e-01 6.36844635e-01 4.58348870e-01
-9.65537250e-01 -7.01079488e-01 5.75137317e-01 4.22731675e-02
2.89539069e-01 -4.15937811e-01 -6.31133616e-01 -1.16955984e+00
2.98224509e-01 -3.25582266e-01 -2.68326670e-01 -6.32786751e-01
-1.16760278e+00 7.56438494e-01 -7.49598980e-01 -1.10361588e+00
2.18834013e-01 -8.88509452e-01 -8.61758351e-01 6.69338584e-01
-1.76133192e+00 -6.44608200e-01 -2.53921717e-01 9.21523809e-01
2.89530158e-01 1.15424059e-01 8.15112829e-01 4.55084682e-01
-6.50593877e-01 6.86747611e-01 3.85929257e-01 -4.27467786e-02
2.32557431e-01 -8.23813796e-01 6.02057517e-01 7.22636461e-01
3.90275232e-02 7.68928587e-01 2.28452608e-01 -2.46674269e-01
-1.77996683e+00 -1.04218376e+00 4.72360432e-01 2.88633972e-01
7.38203704e-01 -6.82422101e-01 -1.18378353e+00 1.88615918e-01
-5.56186866e-03 3.31453919e-01 2.87266374e-01 -5.24413586e-01
-3.92017931e-01 -5.02023637e-01 -9.65231419e-01 7.17848420e-01
1.73961675e+00 -6.09823525e-01 -1.75058827e-01 3.25923890e-01
7.19399035e-01 -1.83946311e-01 -6.61279261e-01 3.50673318e-01
5.62251925e-01 -7.21265554e-01 7.36991525e-01 -2.67236650e-01
2.20747381e-01 -3.72004420e-01 -7.12516755e-02 -1.16527212e+00
-2.72372752e-01 -6.38724804e-01 -3.03657472e-01 1.00534570e+00
3.65047604e-01 -1.11530387e+00 8.17411840e-01 3.43292654e-01
-4.87771183e-01 -1.24420130e+00 -1.24326003e+00 -1.17586339e+00
2.53404826e-01 -2.24620223e-01 2.94053972e-01 4.43100929e-01
6.86441362e-02 5.89755401e-02 -5.49545139e-03 -1.79670468e-01
6.49632990e-01 1.04034938e-01 1.25278145e-01 -1.02326262e+00
-3.97339612e-01 -7.60912418e-01 -5.19058943e-01 -1.25677454e+00
-1.60897113e-02 -1.10344613e+00 2.50226676e-01 -1.51560187e+00
3.79028567e-03 -6.44978285e-01 -8.97081375e-01 6.31509840e-01
3.53669852e-01 1.82997629e-01 -9.70992222e-02 4.53963280e-01
-2.56651580e-01 7.44760036e-01 1.00518465e+00 -2.13081807e-01
-2.18177568e-02 -3.62162560e-01 -2.23456115e-01 4.45509881e-01
9.09511924e-01 -6.87680721e-01 -4.70894307e-01 -8.26872051e-01
2.90251940e-01 -2.70026922e-01 4.55908895e-01 -1.58833373e+00
8.50400090e-01 5.01702912e-02 4.12829846e-01 5.21052368e-02
4.14166421e-01 -5.38333714e-01 2.43202046e-01 9.80463743e-01
-2.07070321e-01 3.46846022e-02 4.53135878e-01 5.59158802e-01
-8.49489197e-02 -4.13514562e-02 8.94651532e-01 -1.79987270e-02
-6.39624298e-01 5.65546036e-01 -3.93397242e-01 3.04171313e-02
6.16088152e-01 -5.76589286e-01 -5.64419448e-01 2.13161260e-01
-3.32598537e-01 6.62417933e-02 1.88841984e-01 -3.58540309e-03
9.25933421e-01 -1.37088048e+00 -4.02457446e-01 4.16181982e-01
-2.49145791e-01 -1.66239545e-01 1.23678505e-01 9.57347155e-01
-5.13346553e-01 4.29693401e-01 -7.43578255e-01 -6.11389399e-01
-3.24203283e-01 3.14721942e-01 6.87027216e-01 -9.99847278e-02
-5.67444682e-01 1.17570603e+00 2.26073682e-01 -2.32901871e-01
3.10593992e-01 -4.71903205e-01 3.39096099e-01 -5.68043053e-01
4.48523939e-01 5.97870238e-02 2.09209070e-01 2.23810449e-01
-4.39806044e-01 7.98001170e-01 1.43235385e-01 -6.94196820e-02
1.60959470e+00 2.30623856e-01 -2.15857372e-01 5.64285696e-01
1.14418519e+00 -4.63160872e-01 -1.56874049e+00 -8.69886056e-02
-4.69789356e-02 8.67366269e-02 1.12358317e-01 -6.95687950e-01
-1.45290673e+00 1.21590161e+00 1.00378025e+00 -3.45247477e-01
1.30705488e+00 -5.05609632e-01 1.32159019e+00 7.61054218e-01
9.36546862e-01 -1.01111901e+00 3.06963176e-02 6.10785306e-01
6.58069253e-01 -7.92314589e-01 -6.22358143e-01 1.78348586e-01
6.00497425e-02 1.33080482e+00 9.50902820e-01 -5.72656929e-01
7.81856894e-01 7.98142195e-01 -3.75262529e-01 4.69422638e-02
-1.00205100e+00 7.17960745e-02 -2.79037774e-01 3.74521136e-01
3.25580001e-01 -2.35244274e-01 -4.74301934e-01 8.46566379e-01
8.51944610e-02 4.70715940e-01 1.27132133e-01 9.98335660e-01
-6.19406462e-01 -8.07092190e-01 2.69508123e-01 5.72239935e-01
-3.09054524e-01 -4.55375552e-01 1.40326127e-01 2.61212081e-01
2.32962910e-02 4.06803370e-01 2.56579630e-02 -5.85599482e-01
1.99700922e-01 -3.05438638e-02 7.46639907e-01 -1.53702840e-01
-9.63215172e-01 -1.91856444e-01 -5.30066073e-01 -5.16445339e-01
-8.43748525e-02 4.92838472e-02 -1.97117257e+00 -4.45274472e-01
-3.48388314e-01 -6.49746880e-02 9.78086293e-01 8.65378559e-01
6.22511625e-01 6.24511003e-01 4.81798321e-01 -9.58829999e-01
-4.88599777e-01 -4.98828292e-01 -4.88821000e-01 5.68427630e-02
-7.34594325e-03 -3.48956972e-01 -3.12404275e-01 -1.01311907e-01] | [8.2272310256958, 2.4872119426727295] |
5eb47235-de91-45e1-a7c3-185875c1fd5a | uncertainty-aware-blind-image-quality | 2005.13983 | null | https://arxiv.org/abs/2005.13983v6 | https://arxiv.org/pdf/2005.13983v6.pdf | Uncertainty-Aware Blind Image Quality Assessment in the Laboratory and Wild | Performance of blind image quality assessment (BIQA) models has been significantly boosted by end-to-end optimization of feature engineering and quality regression. Nevertheless, due to the distributional shift between images simulated in the laboratory and captured in the wild, models trained on databases with synthetic distortions remain particularly weak at handling realistic distortions (and vice versa). To confront the cross-distortion-scenario challenge, we develop a \textit{unified} BIQA model and an approach of training it for both synthetic and realistic distortions. We first sample pairs of images from individual IQA databases, and compute a probability that the first image of each pair is of higher quality. We then employ the fidelity loss to optimize a deep neural network for BIQA over a large number of such image pairs. We also explicitly enforce a hinge constraint to regularize uncertainty estimation during optimization. Extensive experiments on six IQA databases show the promise of the learned method in blindly assessing image quality in the laboratory and wild. In addition, we demonstrate the universality of the proposed training strategy by using it to improve existing BIQA models. | ['Kede Ma', 'Xiaokang Yang', 'Weixia Zhang', 'Guangtao Zhai'] | 2020-05-28 | null | null | null | null | ['blind-image-quality-assessment'] | ['computer-vision'] | [ 3.00955139e-02 -2.39726439e-01 4.69918758e-01 -5.89399874e-01
-1.39724720e+00 -6.82283342e-01 6.16615713e-01 -2.02089086e-01
-4.58222806e-01 5.67169368e-01 3.40038925e-01 -8.74800608e-02
-2.34247565e-01 -5.00510037e-01 -9.10695910e-01 -6.18889213e-01
-3.82565111e-02 2.80541092e-01 -3.40917140e-01 6.71620443e-02
1.21891461e-01 4.09439772e-01 -1.55222571e+00 1.60496607e-01
1.04307771e+00 1.27673066e+00 -9.52196792e-02 8.56869698e-01
6.36676311e-01 6.02666497e-01 -8.09376001e-01 -9.29152548e-01
9.14358258e-01 -5.27771533e-01 -6.82312667e-01 4.08690184e-01
9.68155861e-01 -8.05480301e-01 -6.33205593e-01 1.36828244e+00
9.13910806e-01 5.57185486e-02 6.52686179e-01 -1.13564825e+00
-9.50649083e-01 1.67421848e-02 -3.22125584e-01 1.54537037e-01
6.02929354e-01 7.70497680e-01 1.09468007e+00 -9.93544936e-01
3.08219612e-01 1.14857566e+00 4.60125148e-01 5.66104472e-01
-1.23405802e+00 -3.81505817e-01 -3.67636159e-02 2.43121877e-01
-1.24090207e+00 -8.66095662e-01 3.33066791e-01 -5.27065456e-01
7.88072646e-01 8.45741555e-02 4.90433663e-01 1.00666642e+00
-1.05848592e-02 7.22737908e-01 1.39376497e+00 -2.69359738e-01
2.90542424e-01 2.00040609e-01 -3.66073608e-01 5.84309936e-01
6.60534278e-02 5.27379036e-01 -6.15758181e-01 -6.41882122e-02
5.11777580e-01 -3.73834759e-01 -6.90333188e-01 -6.08456314e-01
-1.28045440e+00 5.20272613e-01 6.57755196e-01 -2.43231580e-01
-3.07669610e-01 9.18102190e-02 1.10656479e-02 5.47295153e-01
1.51709899e-01 7.23741949e-01 -2.95536276e-02 4.88232896e-02
-1.16893685e+00 5.43330669e-01 5.78516006e-01 9.05146658e-01
5.46299100e-01 2.71926913e-02 -5.91918528e-01 8.43061388e-01
3.01708877e-01 7.24892020e-01 3.65409076e-01 -1.16657174e+00
5.93906164e-01 4.97992374e-02 5.86826980e-01 -8.39253247e-01
-6.24607019e-02 -4.51599866e-01 -5.28732777e-01 6.46376252e-01
7.02815711e-01 1.12073861e-01 -1.19351292e+00 1.66677487e+00
1.83522645e-02 -2.38283336e-01 1.31323576e-01 1.31154811e+00
2.94660360e-01 5.11064112e-01 -1.56760484e-01 -1.25336424e-01
1.08080375e+00 -7.96455264e-01 -5.41491330e-01 -1.63103282e-01
5.94956009e-03 -7.42813170e-01 1.21872783e+00 5.67574918e-01
-1.51437032e+00 -5.32638252e-01 -1.31362820e+00 -9.62768346e-02
-1.72486246e-01 1.47933349e-01 6.24695644e-02 7.84047067e-01
-1.26761472e+00 6.29847765e-01 -4.76523012e-01 -6.76281527e-02
7.54074275e-01 4.00477052e-01 -3.70176524e-01 -4.21916813e-01
-1.04690433e+00 9.25227702e-01 -1.11352876e-01 1.77186057e-01
-1.44190991e+00 -8.01912725e-01 -8.92892599e-01 -7.64129534e-02
1.60472523e-02 -9.91100371e-01 1.27451742e+00 -1.01416302e+00
-1.52294731e+00 9.97520566e-01 -3.83554734e-02 -5.36883712e-01
8.61968100e-01 -3.94425333e-01 -5.07503808e-01 2.83264786e-01
9.63346511e-02 6.86133444e-01 1.21391046e+00 -1.45291877e+00
-3.87491107e-01 -5.09872854e-01 3.64331365e-01 3.38127166e-01
-8.82860795e-02 1.59770280e-01 -5.08712590e-01 -5.59549332e-01
-1.52263150e-01 -5.51071107e-01 1.53197767e-02 3.44272494e-01
-4.03368622e-01 4.32971984e-01 1.42648220e-01 -8.99955332e-01
8.61750424e-01 -2.03856087e+00 2.55834192e-01 2.39380255e-01
2.30715424e-01 2.16000542e-01 -3.45624328e-01 1.98161993e-02
5.47425523e-02 -1.58888876e-01 -5.33128798e-01 -6.62335038e-01
1.80199116e-01 -5.31380158e-03 -2.40943089e-01 7.33763635e-01
5.51945686e-01 8.22454751e-01 -8.41683269e-01 -2.52260298e-01
1.60022989e-01 4.66570467e-01 -7.57029414e-01 7.85121381e-01
-1.06118649e-01 5.74464917e-01 6.69363812e-02 7.26935804e-01
8.97170842e-01 -8.58387351e-02 -2.60237277e-01 -4.42942321e-01
2.31999829e-01 2.54222095e-01 -1.07897687e+00 1.73221195e+00
-5.25176942e-01 4.51909900e-01 1.07674338e-01 -6.61571145e-01
4.38364685e-01 2.96250939e-01 7.14704096e-02 -1.07219195e+00
1.12743229e-01 2.01066807e-01 8.21932480e-02 -4.16449040e-01
2.86784232e-01 -2.40599871e-01 1.51291743e-01 3.42239827e-01
4.13364917e-01 -5.28056324e-01 4.04468551e-02 8.16516131e-02
7.62475550e-01 6.38075322e-02 -1.10566124e-01 -9.06208158e-02
5.11575878e-01 -4.26701576e-01 3.83682102e-01 7.97426224e-01
-6.60158277e-01 1.24568987e+00 3.90408218e-01 -2.49586061e-01
-1.46392989e+00 -1.55000699e+00 -3.04384559e-01 6.99496031e-01
1.64766982e-01 -1.86237413e-02 -8.39193940e-01 -7.75897861e-01
9.29193050e-02 5.96170783e-01 -4.75753307e-01 -2.28633866e-01
-1.48819134e-01 -8.23673606e-01 4.26488549e-01 2.54281759e-01
7.06247330e-01 -5.58700502e-01 -4.17590082e-01 -1.10824861e-01
-1.94361389e-01 -1.25971913e+00 -6.41830564e-01 -2.04165533e-01
-3.37514430e-01 -9.37606215e-01 -9.99525070e-01 -4.69094425e-01
5.60846567e-01 3.06235570e-02 1.40675318e+00 -1.82693645e-01
-3.05226624e-01 6.07783556e-01 -2.66178120e-02 -1.63825020e-01
-4.85300452e-01 -6.56543911e-01 3.36218446e-01 2.76925862e-01
-1.80743653e-02 -4.48937178e-01 -9.45348740e-01 4.59499329e-01
-9.86907423e-01 -3.50087881e-01 3.62049550e-01 9.28067446e-01
5.37659347e-01 -2.60075118e-04 3.66126537e-01 -2.77144194e-01
6.61845803e-01 -2.37962797e-01 -8.94999564e-01 3.75771731e-01
-6.57620132e-01 1.77789524e-01 4.14687395e-01 -2.02408284e-01
-9.65040386e-01 -1.27867952e-01 5.84195601e-03 -5.64938188e-01
4.79687750e-02 5.71819767e-02 -5.44643044e-01 -3.02961946e-01
7.19765127e-01 1.08638242e-01 5.29518463e-02 -1.99286342e-01
4.93292004e-01 6.63357735e-01 9.53329682e-01 -6.36008382e-01
1.04986405e+00 4.03718829e-01 -1.60370305e-01 -3.85065943e-01
-6.98869765e-01 -1.59249380e-01 -2.95250326e-01 -2.89223641e-01
8.63407314e-01 -1.22009575e+00 -6.08645320e-01 7.44039655e-01
-8.85855615e-01 -2.49807656e-01 -3.52341890e-01 4.95043278e-01
-9.18346941e-01 5.31535029e-01 -2.69153863e-01 -7.79239237e-01
-2.14854151e-01 -1.62589800e+00 1.13865638e+00 1.79623321e-01
2.45087221e-01 -6.51808798e-01 -2.33651455e-02 5.94826460e-01
4.42535460e-01 -3.07188630e-02 6.43903852e-01 -1.76746473e-01
-7.78672993e-01 -3.04604411e-01 -4.90619957e-01 8.53682578e-01
-1.14685250e-02 -3.38871777e-01 -1.32863855e+00 -5.37638664e-01
1.60079375e-01 -7.53441870e-01 7.37572074e-01 3.67119342e-01
1.09106314e+00 -2.67456651e-01 4.36920583e-01 1.02989054e+00
1.42948806e+00 -9.59754512e-02 7.14902282e-01 2.92056829e-01
3.22666109e-01 5.15646338e-01 3.51587206e-01 2.67563373e-01
3.50579441e-01 8.18492234e-01 6.52066052e-01 -7.75886234e-03
-4.02537107e-01 -1.62150070e-01 5.38905144e-01 3.98381501e-01
1.62628889e-01 -3.16618085e-01 -7.15179443e-01 7.85350263e-01
-1.30693102e+00 -7.33303607e-01 4.41632956e-01 2.55076218e+00
9.87093091e-01 -2.47405656e-02 1.46946266e-01 -5.44667768e-04
4.57972854e-01 6.92018941e-02 -7.02833295e-01 -2.16559887e-01
-4.59814578e-01 8.73823240e-02 4.37386483e-01 6.37530982e-01
-1.22081447e+00 5.02598166e-01 6.89065599e+00 4.05161768e-01
-8.45153868e-01 -1.34024933e-01 8.83963525e-01 -4.58343983e-01
-2.04144984e-01 -2.19497636e-01 -3.11142117e-01 4.53976810e-01
1.00569236e+00 -8.02755505e-02 9.33560610e-01 5.01436830e-01
1.95322242e-02 -1.68046728e-01 -1.42760682e+00 1.15641248e+00
2.24165291e-01 -9.02701199e-01 2.60562450e-01 4.05787751e-02
1.00489390e+00 -2.13642232e-02 8.01476061e-01 -5.78386849e-03
4.02026385e-01 -1.38128686e+00 9.42663372e-01 6.35350406e-01
1.01931775e+00 -5.15059352e-01 7.29451060e-01 -8.40508938e-02
-5.80572605e-01 -2.24709719e-01 -5.23348331e-01 1.92192733e-01
3.08358781e-02 5.91448307e-01 -5.28339267e-01 6.75820172e-01
8.49385083e-01 4.90418851e-01 -8.09076130e-01 1.30693448e+00
-1.69439971e-01 2.46574372e-01 -1.38648838e-01 5.62462986e-01
6.98097870e-02 -2.43335158e-01 6.25281215e-01 9.88582194e-01
4.09880191e-01 -1.25045896e-01 -3.42687488e-01 1.18943179e+00
-3.30162019e-01 -2.89685309e-01 -4.85995650e-01 8.42386186e-02
2.00507224e-01 8.42647851e-01 1.78131118e-01 -1.80163741e-01
-4.68235224e-01 1.36841500e+00 2.59605050e-01 5.56916475e-01
-6.75210834e-01 -2.36754820e-01 1.01356292e+00 -9.28681195e-02
2.91781783e-01 1.42566198e-02 -2.66788542e-01 -1.44197488e+00
4.53822702e-01 -1.28678477e+00 2.36028016e-01 -1.18366742e+00
-1.57607031e+00 7.30349958e-01 -2.03821778e-01 -1.46582520e+00
-2.46329129e-01 -8.15355718e-01 -3.70436043e-01 1.40832245e+00
-1.81354249e+00 -9.94425178e-01 -2.32477829e-01 8.01603556e-01
2.57124484e-01 -2.45391130e-01 5.74259818e-01 3.85626554e-01
-3.11829865e-01 1.02581036e+00 1.98378280e-01 1.40190229e-01
1.00350869e+00 -1.52086902e+00 3.89927328e-01 1.19880128e+00
1.10581830e-01 4.14235145e-01 8.40255857e-01 -1.01657093e-01
-1.34608245e+00 -1.04714048e+00 4.66686428e-01 -8.01453114e-01
4.67325926e-01 -2.74316221e-01 -7.57151008e-01 3.93522084e-01
2.99552500e-01 3.33360672e-01 3.49045455e-01 -3.47121418e-01
-8.79960835e-01 -1.38091683e-01 -1.52008712e+00 4.06533659e-01
9.25328851e-01 -9.75028396e-01 -5.44739425e-01 3.71261775e-01
6.92969143e-01 -4.49319631e-01 -9.49542880e-01 2.91895986e-01
4.82791692e-01 -1.41942823e+00 1.23429656e+00 -5.39774537e-01
6.25462294e-01 -3.07619512e-01 -5.29513419e-01 -1.64374149e+00
-2.81458478e-02 -5.85764647e-01 -5.62295020e-02 1.07905829e+00
4.21910286e-01 -3.48243564e-01 3.71219069e-01 7.54793704e-01
2.39519924e-02 -4.50923026e-01 -1.06392407e+00 -9.35053825e-01
2.95972615e-01 -4.31645513e-01 8.08635890e-01 4.51963484e-01
-2.31642991e-01 -2.05214679e-01 -5.58595300e-01 5.44548392e-01
1.06694329e+00 -1.75761849e-01 6.15487218e-01 -6.78188741e-01
-6.20051384e-01 -5.19878626e-01 -6.99664772e-01 -9.56508100e-01
-1.92632362e-01 -5.75201154e-01 2.75788516e-01 -1.24375474e+00
1.43492833e-01 -1.67738229e-01 -2.38669053e-01 -1.94317251e-01
-4.88364428e-01 3.90808940e-01 2.77108252e-01 2.16356650e-01
-5.25290310e-01 6.97610915e-01 1.33695817e+00 -3.36158365e-01
1.02635577e-01 -1.56092197e-02 -5.79051018e-01 3.44503909e-01
3.74634266e-01 -1.08260937e-01 -3.59808207e-01 -8.44002426e-01
2.57061869e-01 6.38553053e-02 6.26624405e-01 -1.32742274e+00
1.03135131e-01 2.27007404e-01 6.05627775e-01 4.59651137e-03
4.81052428e-01 -8.65299106e-01 -2.45637521e-01 2.48089973e-02
-3.91416371e-01 -6.90893680e-02 1.25300229e-01 6.00479066e-01
-2.78999031e-01 6.02808520e-02 1.21794415e+00 3.10103763e-02
-2.17326388e-01 4.75597471e-01 1.71836242e-01 3.69783759e-01
6.87464297e-01 -3.70093472e-02 -3.27628344e-01 -6.58620834e-01
-6.73641503e-01 3.24863940e-01 7.06075191e-01 2.42813677e-01
7.47553229e-01 -1.43271053e+00 -8.81649554e-01 4.89617586e-01
4.85302180e-01 -1.27056882e-01 3.04197699e-01 6.21255875e-01
-3.91659737e-01 1.25771463e-01 -3.42778206e-01 -5.36207736e-01
-8.35213602e-01 6.74366236e-01 9.37365592e-01 -1.41307816e-01
-2.37688512e-01 1.04008281e+00 2.56061852e-01 -3.88976842e-01
5.40251493e-01 -1.83413044e-01 1.62663788e-01 -2.96264797e-01
7.57080197e-01 1.65871829e-01 3.23973835e-01 -7.02141583e-01
-1.79282695e-01 4.92579788e-01 -6.99642720e-03 -5.28553665e-01
1.08155286e+00 -3.45203608e-01 1.16253108e-01 1.19313151e-01
1.35937607e+00 -7.62808621e-02 -1.65790844e+00 -3.52640986e-01
-3.04992259e-01 -9.84508097e-01 2.08223298e-01 -1.33717084e+00
-1.10175800e+00 9.57518637e-01 1.09875977e+00 -7.94121027e-02
1.35323095e+00 -1.28785431e-01 5.24563313e-01 1.72417536e-01
2.40296841e-01 -8.62277985e-01 1.98230252e-01 9.89495683e-03
1.18949497e+00 -1.63400388e+00 -1.12314373e-01 1.63424850e-01
-7.74539232e-01 7.91404009e-01 3.73043865e-01 -1.33415446e-01
5.21014214e-01 -1.71486497e-01 2.83232033e-01 -3.38264778e-02
-3.54141444e-01 -9.05893594e-02 6.34447932e-01 9.99334633e-01
1.28437176e-01 1.69011019e-02 2.89794028e-01 2.61095583e-01
-3.14566851e-01 -7.44010434e-02 5.76563001e-01 4.71854180e-01
-1.07053921e-01 -8.33900094e-01 -5.79871297e-01 3.47394973e-01
-4.53700393e-01 -1.92371503e-01 -8.75024647e-02 3.50372016e-01
-2.94549968e-02 1.15063310e+00 -1.46597307e-02 -3.50760072e-01
5.69328725e-01 3.66524197e-02 6.90924883e-01 -1.57896683e-01
-5.40930033e-01 -2.20819145e-01 -1.08712666e-01 -9.34013307e-01
-3.04033697e-01 -7.34224081e-01 -5.05010366e-01 -1.95318148e-01
3.38172801e-02 -1.43553287e-01 6.02559745e-01 7.34525561e-01
2.51442552e-01 3.25873941e-01 8.59466016e-01 -7.48232365e-01
-1.05333495e+00 -7.83327699e-01 -7.59059489e-01 6.95758522e-01
9.71055388e-01 -5.41019797e-01 -6.69368327e-01 8.75379369e-02] | [11.888123512268066, -1.801232099533081] |
7a8f5e96-eff3-469d-be64-bec8347809b5 | language-id-prediction-from-speech-using-self | 2104.11985 | null | https://arxiv.org/abs/2104.11985v1 | https://arxiv.org/pdf/2104.11985v1.pdf | Language ID Prediction from Speech Using Self-Attentive Pooling and 1D-Convolutions | This memo describes NTR-TSU submission for SIGTYP 2021 Shared Task on predicting language IDs from speech. Spoken Language Identification (LID) is an important step in a multilingual Automated Speech Recognition (ASR) system pipeline. For many low-resource and endangered languages, only single-speaker recordings may be available, demanding a need for domain and speaker-invariant language ID systems. In this memo, we show that a convolutional neural network with a Self-Attentive Pooling layer shows promising results for the language identification task. | ['Nikolay Mikhaylovskiy', 'Roman Bedyakin'] | 2021-04-24 | null | null | null | null | ['spoken-language-identification'] | ['speech'] | [-6.16265126e-02 -7.30723068e-02 -2.00155482e-01 -6.15178764e-01
-1.04229653e+00 -7.51910567e-01 6.33392632e-01 -1.31449535e-01
-8.52969468e-01 6.62335038e-01 4.55734402e-01 -6.60721123e-01
4.60502297e-01 2.28721015e-02 -3.67392719e-01 -2.29593039e-01
-7.81411454e-02 7.55301118e-01 -2.61223316e-03 -3.69259745e-01
-1.77370906e-01 8.31057906e-01 -1.09034264e+00 5.56210577e-01
4.73856360e-01 5.04162908e-01 2.90741056e-01 1.07403934e+00
-1.99265584e-01 1.04221082e+00 -6.65239692e-01 -8.01870823e-02
-4.42713164e-02 -2.42921680e-01 -1.21378493e+00 -3.60756159e-01
3.26201648e-01 -4.59998220e-01 -6.66264176e-01 8.64597023e-01
8.91328156e-01 1.10475115e-01 4.14893895e-01 -7.13788450e-01
-5.62679887e-01 1.01539862e+00 8.43560919e-02 6.73981011e-01
5.02846062e-01 1.61560893e-01 7.59488165e-01 -1.08151472e+00
6.06356561e-01 1.68691957e+00 5.23976624e-01 9.88863587e-01
-1.01641643e+00 -6.78490937e-01 -6.27404973e-02 -1.89450718e-02
-1.57517374e+00 -1.25024450e+00 3.33359301e-01 -3.68623465e-01
1.72340763e+00 2.44505018e-01 -3.52433398e-02 1.24252725e+00
-3.28801006e-01 1.14890862e+00 8.36322486e-01 -2.62605965e-01
1.72074568e-02 3.70740384e-01 3.68216842e-01 3.62549841e-01
-2.66874105e-01 -1.71562299e-01 -8.99089217e-01 -2.51191229e-01
2.85875618e-01 -3.86802733e-01 -7.77288228e-02 6.13469958e-01
-1.26253116e+00 7.24839866e-01 6.81913868e-02 5.75496078e-01
-3.17847461e-01 -3.40889186e-01 8.24966967e-01 4.03148443e-01
8.20358336e-01 3.64875853e-01 -5.07513046e-01 -2.09060267e-01
-9.88280594e-01 -3.42947632e-01 8.02410781e-01 6.20831788e-01
5.03017128e-01 4.69792902e-01 -1.16965830e-01 1.48501682e+00
1.53846502e-01 6.77984416e-01 7.89939523e-01 -2.74565428e-01
2.35492468e-01 2.80168623e-01 -2.38875970e-01 -1.44307822e-01
-5.70860982e-01 -1.16432652e-01 -7.20697999e-01 -2.42028400e-01
2.28333130e-01 -3.57102543e-01 -9.02009487e-01 1.42139697e+00
-1.48623466e-01 1.11087322e-01 4.94036704e-01 7.73193181e-01
1.17459667e+00 7.56925166e-01 1.71287432e-01 -3.24030034e-02
1.32844055e+00 -7.74559319e-01 -6.55888677e-01 -4.53968853e-01
6.63844764e-01 -7.87713706e-01 6.85391903e-01 1.34001762e-01
-8.79158854e-01 -2.35571310e-01 -8.26290131e-01 -2.47697815e-01
-5.96194148e-01 3.01543891e-01 2.47903228e-01 7.54614294e-01
-1.41918504e+00 -5.73598035e-02 -7.22811580e-01 -9.75263000e-01
2.29270160e-02 4.56802785e-01 -7.52985477e-01 7.06851250e-03
-1.39137828e+00 1.13321507e+00 4.01912659e-01 6.35449663e-02
-1.05668652e+00 -5.53423703e-01 -9.12203968e-01 -1.99340105e-01
-2.73861308e-02 1.20585918e-01 1.40918469e+00 -7.08597302e-01
-1.79486525e+00 1.40770781e+00 -4.39426959e-01 -9.14958894e-01
2.71833271e-01 -7.99680278e-02 -9.09567475e-01 1.95737910e-02
-8.03184509e-02 7.31770396e-01 5.16831815e-01 -5.06712675e-01
-5.14426291e-01 -3.57827842e-01 -5.36106706e-01 1.60945624e-01
-3.76012385e-01 1.17930198e+00 -1.11794889e-01 -3.80239636e-01
-1.88236162e-01 -6.61561430e-01 3.46011743e-02 -7.51624227e-01
-3.87284815e-01 -7.17262924e-01 8.65763187e-01 -1.34324920e+00
1.00823295e+00 -2.30594206e+00 -2.21574783e-01 -1.84980795e-01
-1.75765734e-02 7.91272044e-01 -1.89390257e-01 3.03837299e-01
-1.38851637e-02 1.44764662e-01 8.86756033e-02 -6.13399386e-01
-4.56722900e-02 -1.05211567e-02 -4.91167784e-01 4.89567608e-01
3.36236566e-01 5.87145388e-01 -6.85427010e-01 -1.09195516e-01
2.84183294e-01 6.42662406e-01 -6.59695789e-02 4.52340186e-01
-4.16761450e-02 5.69239676e-01 8.46160501e-02 5.76478958e-01
4.48858738e-01 8.77116099e-02 -3.85939851e-02 2.73114383e-01
-5.90318680e-01 1.07501292e+00 -7.69947827e-01 1.46175122e+00
-5.58292031e-01 8.09883893e-01 6.04517519e-01 -7.10965157e-01
1.30672848e+00 6.78718686e-01 -7.42677897e-02 -6.03689611e-01
1.50657758e-01 5.80221713e-01 6.30630972e-03 -1.20568924e-01
4.99763578e-01 3.24170701e-02 1.59419619e-03 4.56610471e-01
4.48592335e-01 1.59845442e-01 -2.99236298e-01 1.71711206e-01
1.15988517e+00 -6.87264740e-01 2.77441263e-01 -4.54878658e-01
8.10102582e-01 -1.78014383e-01 5.80655098e-01 8.48148763e-01
-6.47483826e-01 7.02883601e-01 -8.61568376e-02 -5.76793790e-01
-9.51081932e-01 -8.44439864e-01 -2.80265480e-01 1.42821336e+00
-8.30509484e-01 1.02282334e-02 -7.23233342e-01 -2.40877867e-01
-2.03368902e-01 7.18323052e-01 -8.23105574e-02 6.31455779e-02
-6.77682042e-01 -4.24928933e-01 1.45788276e+00 2.46460706e-01
2.93392986e-01 -1.25649631e+00 3.54467630e-01 3.98905486e-01
-1.46784037e-01 -1.48433328e+00 -6.78181767e-01 4.20719355e-01
5.64400740e-02 -4.27374214e-01 -1.11802673e+00 -1.20318532e+00
6.53967261e-02 5.32477461e-02 7.86817551e-01 -4.34256226e-01
-3.26621741e-01 4.11957264e-01 -5.31776026e-02 -4.47532892e-01
-1.01295972e+00 5.24976194e-01 8.60464215e-01 2.23990574e-01
1.00326180e+00 3.54521833e-02 -1.43900156e-01 1.71743721e-01
-3.68095845e-01 -2.06717163e-01 1.36848792e-01 5.86965382e-01
1.80856943e-01 -6.82129085e-01 9.93149340e-01 -2.27387860e-01
5.82107544e-01 -4.00554359e-01 -6.97784841e-01 4.90456372e-01
1.10508567e-02 -1.31908745e-01 5.81166148e-01 -3.82484674e-01
-9.29655612e-01 4.05248016e-01 -6.01049066e-01 -2.25324839e-01
-6.49437249e-01 1.84846014e-01 -2.06510887e-01 2.27722287e-01
6.08081222e-01 6.65876746e-01 1.23216048e-01 -6.76461637e-01
1.37563571e-01 1.65650618e+00 7.66456008e-01 -1.95634142e-01
1.12740539e-01 8.57654586e-02 -8.91093373e-01 -1.76203930e+00
-3.69574219e-01 -8.03613305e-01 -6.30714536e-01 4.30956343e-03
1.20955002e+00 -1.29627597e+00 -7.84335554e-01 8.55674326e-01
-1.42638564e+00 -3.99118096e-01 2.45480984e-01 7.31544912e-01
-1.50990725e-01 -1.79658122e-02 -7.25271583e-01 -1.20211697e+00
-7.08843231e-01 -1.23495400e+00 8.64382923e-01 4.17136028e-02
-4.62902367e-01 -8.17919314e-01 3.54336232e-01 3.55595052e-01
7.72394478e-01 -7.94116020e-01 2.20562235e-01 -1.66197109e+00
-1.23404920e-01 -2.79947370e-01 -8.73112157e-02 6.48062110e-01
1.05308719e-01 -1.84640035e-01 -1.40731061e+00 -3.22949708e-01
-4.39934939e-01 -5.79652429e-01 9.08182442e-01 3.34800482e-01
6.21486664e-01 -2.01030463e-01 -3.46402302e-02 4.70736444e-01
4.84776288e-01 3.26685190e-01 1.36622027e-01 -3.81799042e-03
7.97258377e-01 7.28678346e-01 -3.66076380e-01 2.34218210e-01
4.27764624e-01 6.11218452e-01 -5.68028152e-01 8.47197622e-02
-3.12861979e-01 -1.83472946e-01 7.55082667e-01 9.84332025e-01
6.96244657e-01 -2.22719312e-01 -1.56518006e+00 9.20118392e-01
-1.41136420e+00 -8.11332405e-01 1.50879577e-01 2.24679303e+00
8.39388072e-01 -3.49592417e-01 4.46100235e-01 -3.54030848e-01
1.19788826e+00 1.53825417e-01 -5.77318490e-01 -8.30339253e-01
-6.92434847e-01 7.60295019e-02 5.08221030e-01 8.62448394e-01
-1.09720969e+00 1.39074135e+00 7.28314877e+00 7.09208071e-01
-1.42684400e+00 2.97438622e-01 5.41390121e-01 2.72303075e-02
3.73539107e-04 -5.24341941e-01 -1.38314986e+00 2.40883335e-01
1.71674430e+00 -4.49186653e-01 7.43147969e-01 8.03215683e-01
2.47662678e-01 2.47967795e-01 -1.00651908e+00 1.17096651e+00
3.01625401e-01 -1.17600548e+00 -1.98564738e-01 -1.97673216e-01
8.09429958e-02 1.15480530e+00 5.13643585e-02 2.14056805e-01
7.23959506e-01 -1.25049043e+00 2.68870354e-01 4.47948650e-02
1.01168227e+00 -8.50004554e-01 6.85748637e-01 3.31022471e-01
-8.95524561e-01 1.39169231e-01 -2.09190533e-01 1.02815293e-01
1.13114409e-01 2.17187330e-01 -1.46178353e+00 -1.30424365e-01
6.54336452e-01 4.97567505e-01 -3.85681003e-01 6.95590973e-01
1.11847743e-02 1.09781897e+00 -5.14104784e-01 -2.46422142e-01
1.56586498e-01 4.51987743e-01 7.48313963e-01 1.72046745e+00
6.27605319e-02 -1.40571490e-01 1.06569819e-01 5.49337327e-01
-4.09458667e-01 3.41708958e-01 -8.80775452e-01 -5.26238143e-01
5.47643125e-01 1.03194678e+00 -4.50440019e-01 -3.40131849e-01
-3.89745086e-01 1.05876553e+00 4.76092875e-01 2.04194680e-01
3.57514024e-02 -2.67342985e-01 1.07137990e+00 -2.71765709e-01
-4.46586087e-02 -4.09075260e-01 -8.19323137e-02 -1.18831253e+00
-5.34641147e-01 -9.90224898e-01 4.20813560e-01 -3.70195806e-01
-1.43118095e+00 9.56675529e-01 -4.94435072e-01 -7.47895360e-01
-6.78704917e-01 -7.74699330e-01 -4.56476003e-01 1.33289623e+00
-1.16953051e+00 -1.27810776e+00 4.93001014e-01 5.79281509e-01
6.88202202e-01 -1.01668870e+00 1.24606323e+00 4.58667696e-01
-9.49263453e-01 1.05366814e+00 1.95269421e-01 5.72701573e-01
9.14584279e-01 -9.51838195e-01 8.55455756e-01 8.67518425e-01
1.92071106e-02 6.80205286e-01 3.74528855e-01 -5.11868000e-01
-1.35307705e+00 -1.14781320e+00 1.44760501e+00 -4.24657136e-01
9.89241481e-01 -8.86509895e-01 -1.10309756e+00 8.55263770e-01
3.59499484e-01 -9.02929083e-02 7.93321490e-01 3.18174362e-01
-6.25271499e-01 2.49826722e-02 -1.06437469e+00 3.36478740e-01
5.71006000e-01 -1.38275433e+00 -5.37058055e-01 5.05385339e-01
1.03040266e+00 -7.04946145e-02 -5.78446925e-01 -3.51610184e-02
3.84731799e-01 -2.85124481e-01 8.00020874e-01 -7.06456959e-01
-2.01236457e-01 -1.09074898e-01 -2.85704345e-01 -1.25009859e+00
-1.29123509e-01 -7.72763073e-01 3.62858832e-01 1.51690364e+00
5.21743417e-01 -9.52703178e-01 2.71430969e-01 5.78322887e-01
-2.63611734e-01 3.13644350e-01 -1.30197227e+00 -1.01946592e+00
2.81573355e-01 -4.85859156e-01 4.67165887e-01 1.05469584e+00
1.33047372e-01 6.11758709e-01 -4.44375992e-01 4.23134118e-01
2.27797478e-01 -8.90017033e-01 3.77925187e-01 -9.33860660e-01
2.14634053e-02 -4.48016644e-01 -5.11403918e-01 -6.25932693e-01
8.25133502e-01 -1.16652131e+00 2.21860200e-01 -1.17596841e+00
5.48576713e-02 -4.21624593e-02 -4.61746216e-01 6.51417077e-01
1.26653135e-01 5.30867875e-02 -4.94598374e-02 1.16159372e-01
-7.05536842e-01 2.00333118e-01 3.86843085e-02 -3.49031121e-01
-3.18339556e-01 1.21823996e-02 -3.26937526e-01 5.82360089e-01
9.02378678e-01 -2.38633052e-01 1.11647010e-01 -6.52659357e-01
-4.06664908e-01 -5.11162274e-04 6.00914732e-02 -8.15123856e-01
4.18283671e-01 2.33177587e-01 -6.04338311e-02 -9.03215468e-01
2.17500806e-01 -1.17727704e-01 -2.37465292e-01 4.33673620e-01
-5.90582013e-01 -9.55844298e-02 4.71294820e-01 -1.83254987e-01
-3.69843066e-01 5.89315668e-02 9.34863031e-01 -6.77876547e-02
-8.46881509e-01 2.50130177e-01 -1.06931782e+00 -4.00319882e-02
2.91999012e-01 2.12941006e-01 -4.34801012e-01 -3.37641537e-01
-6.52306199e-01 3.21977228e-01 1.28800735e-01 8.84982765e-01
4.05288219e-01 -1.00150788e+00 -1.24861825e+00 4.85261500e-01
4.69354004e-01 -4.90210801e-01 3.69903743e-01 5.22913754e-01
-2.93383807e-01 9.59200442e-01 4.38215733e-02 -3.75636369e-01
-1.44091332e+00 2.16219261e-01 5.23185611e-01 1.87557429e-01
-3.56519312e-01 1.18293512e+00 9.26809385e-02 -1.18135941e+00
6.51438713e-01 2.00705841e-01 -2.42569759e-01 1.32694647e-01
1.40665507e+00 3.03107053e-01 4.13102776e-01 -1.16622365e+00
-9.79645729e-01 -3.13169539e-01 -5.93514144e-01 -4.91452843e-01
1.19331205e+00 -2.02214852e-01 -2.32741624e-01 8.02450716e-01
1.45490670e+00 -2.47745380e-01 -5.17078280e-01 -5.24404228e-01
1.48501173e-01 2.87995756e-01 3.90043676e-01 -8.50148499e-01
-2.74579316e-01 1.11262143e+00 7.84959137e-01 1.20065756e-01
5.24727046e-01 3.47143382e-01 8.27213109e-01 7.19482303e-01
1.82405829e-01 -1.18878937e+00 -5.76128960e-01 1.23936176e+00
1.02853048e+00 -1.59540772e+00 -7.42253840e-01 1.72776565e-01
-5.78392386e-01 9.17027116e-01 4.94570583e-01 3.09606999e-01
5.97443283e-01 3.70621145e-01 5.25158465e-01 1.72693893e-01
-6.92498326e-01 -3.11904103e-01 2.70292908e-01 7.39718139e-01
7.62578011e-01 3.91834676e-01 1.65142678e-02 5.94792426e-01
-2.63379842e-01 -4.74644393e-01 2.70716846e-01 5.18325031e-01
-4.52989787e-01 -8.86848152e-01 -4.61440861e-01 1.49734870e-01
-7.24078596e-01 -4.45321232e-01 -8.76953065e-01 6.77040145e-02
-5.40692747e-01 1.23487091e+00 3.77159804e-01 -3.40803862e-01
4.38183732e-02 5.47469079e-01 -1.35658696e-01 -8.77449036e-01
-7.87569404e-01 1.58702694e-02 4.70269620e-01 -1.99128628e-01
1.63083673e-01 -7.09008098e-01 -1.12241864e+00 -3.90945345e-01
1.31308049e-01 -1.05502211e-01 1.03250372e+00 8.80129397e-01
4.30588245e-01 2.06066117e-01 6.79665565e-01 -6.19317055e-01
-3.61429542e-01 -1.18830740e+00 -6.50051415e-01 -1.91562399e-01
7.40495265e-01 2.06504967e-02 -4.73442584e-01 -3.14100474e-01] | [14.143684387207031, 6.653419494628906] |
3f2b1c08-87cb-4166-aad5-4139b1eacc1c | what-leads-to-generalization-of-object | 2008.057 | null | https://arxiv.org/abs/2008.05700v1 | https://arxiv.org/pdf/2008.05700v1.pdf | What leads to generalization of object proposals? | Object proposal generation is often the first step in many detection models. It is lucrative to train a good proposal model, that generalizes to unseen classes. This could help scaling detection models to larger number of classes with fewer annotations. Motivated by this, we study how a detection model trained on a small set of source classes can provide proposals that generalize to unseen classes. We systematically study the properties of the dataset - visual diversity and label space granularity - required for good generalization. We show the trade-off between using fine-grained labels and coarse labels. We introduce the idea of prototypical classes: a set of sufficient and necessary classes required to train a detection model to obtain generalized proposals in a more data-efficient way. On the Open Images V4 dataset, we show that only 25% of the classes can be selected to form such a prototypical set. The resulting proposals from a model trained with these classes is only 4.3% worse than using all the classes, in terms of average recall (AR). We also demonstrate that Faster R-CNN model leads to better generalization of proposals compared to a single-stage network like RetinaNet. | ['Rui Wang', 'Dhruv Mahajan', 'Vignesh Ramanathan'] | 2020-08-13 | null | null | null | null | ['object-proposal-generation'] | ['computer-vision'] | [ 1.07080355e-01 4.92789894e-01 -6.96612075e-02 -4.74946052e-01
-5.30050218e-01 -6.41495407e-01 6.79632723e-01 1.34963512e-01
-6.18531704e-01 5.06595910e-01 -2.36321673e-01 -1.22924127e-01
2.74406940e-01 -8.57225716e-01 -7.35195935e-01 -5.02975523e-01
4.44310494e-02 5.81335127e-01 9.39334095e-01 -9.32508707e-02
2.86633939e-01 7.70724118e-01 -1.88569379e+00 5.04213095e-01
5.14213800e-01 6.94194496e-01 4.13813710e-01 5.38449943e-01
-1.75849404e-02 1.06453426e-01 -6.56039000e-01 -2.53462195e-01
6.99232340e-01 -3.27791512e-01 -8.16355705e-01 2.45167553e-01
1.04014361e+00 -3.16183746e-01 1.02032751e-01 1.05314112e+00
3.68501157e-01 1.00216217e-01 9.73199606e-01 -8.86750281e-01
-5.08435905e-01 5.41195810e-01 -5.52857101e-01 3.30542773e-01
-1.47503734e-01 3.71794492e-01 9.68794703e-01 -1.07014275e+00
8.40883136e-01 1.26042390e+00 4.77325678e-01 1.00053596e+00
-1.42871964e+00 -6.48718715e-01 2.94290692e-01 -1.74387664e-01
-1.56810486e+00 -2.67421633e-01 3.69940490e-01 -4.24105585e-01
8.08813035e-01 3.75007451e-01 5.99202931e-01 1.01798737e+00
-2.80159488e-02 5.51669061e-01 1.09591174e+00 -5.91416299e-01
4.32982564e-01 6.05708063e-01 4.56634223e-01 7.65542388e-01
8.58332634e-01 2.95371234e-01 -1.24862686e-01 -1.16195045e-02
8.60162735e-01 3.94248404e-02 -8.41968134e-02 -5.06104231e-01
-9.63760912e-01 9.29358482e-01 1.06770349e+00 2.37541005e-01
-8.00070539e-02 6.64492100e-02 1.57585487e-01 2.79427350e-01
1.84127614e-01 8.69243562e-01 -3.70910972e-01 6.27865970e-01
-1.05891967e+00 1.28203139e-01 6.21359408e-01 1.09968841e+00
1.04388237e+00 -1.00634463e-01 -3.30246866e-01 7.51852036e-01
1.57448113e-01 2.58627206e-01 4.45188850e-01 -7.66380012e-01
-3.98396477e-02 8.30082715e-01 1.10421240e-01 -4.51178849e-01
-6.05501115e-01 -9.94594395e-01 -5.39261758e-01 4.89812136e-01
6.90532863e-01 3.12803276e-02 -1.39621329e+00 1.75321126e+00
2.46262908e-01 -1.95840254e-01 -5.42976782e-02 8.18251133e-01
7.44797409e-01 5.22147059e-01 4.53421660e-02 -5.07271066e-02
1.23938942e+00 -9.32460904e-01 9.81450304e-02 -4.81622458e-01
8.39796603e-01 -7.71046460e-01 1.00779390e+00 3.06495160e-01
-1.04943705e+00 -8.78449976e-01 -1.11051774e+00 -9.61871445e-03
-4.80380476e-01 5.57177901e-01 4.83108580e-01 7.79045939e-01
-1.18938935e+00 4.11506832e-01 -5.37467360e-01 -8.85201454e-01
6.42718613e-01 4.41079766e-01 -2.17611164e-01 -8.49511772e-02
-4.33646888e-01 1.04709041e+00 7.97742307e-01 -2.10777581e-01
-1.22496998e+00 -4.13783997e-01 -3.40313137e-01 5.11842892e-02
3.43508989e-01 -6.19813919e-01 1.09181499e+00 -1.02705395e+00
-8.18450928e-01 1.15257323e+00 1.20865487e-01 -4.89604682e-01
4.63957399e-01 9.83108729e-02 -2.95600649e-02 -6.27378076e-02
9.30163115e-02 1.55900514e+00 9.60974455e-01 -1.40492666e+00
-9.83454883e-01 -1.89760864e-01 2.40419224e-01 -1.08325683e-01
-1.80913433e-01 -1.42408326e-01 -2.65960932e-01 -3.87252837e-01
1.95973381e-01 -1.20989120e+00 -4.51469541e-01 2.63759732e-01
-4.41762954e-01 -3.93773377e-01 5.19035459e-01 1.14967227e-01
7.09956884e-01 -2.01294303e+00 -1.83524907e-01 2.14437023e-01
3.10066223e-01 4.36017066e-01 -3.24026495e-01 4.08396497e-02
2.84976009e-02 1.83083206e-01 1.31903827e-01 -2.44472206e-01
-9.16426033e-02 2.89129108e-01 -4.20424849e-01 3.67763042e-01
5.41600049e-01 7.12962329e-01 -8.38540077e-01 -3.41474563e-01
-2.53468025e-02 2.23205969e-01 -7.93645501e-01 6.76363185e-02
-3.01901639e-01 1.85598597e-01 -4.24579620e-01 5.43160677e-01
6.82985723e-01 -4.01995063e-01 -1.57614633e-01 -1.93506122e-01
-5.71094975e-02 1.15693338e-01 -1.17328751e+00 1.20321715e+00
-3.40114325e-01 7.95933485e-01 -4.49302346e-01 -7.57088602e-01
1.17872834e+00 -2.21143126e-01 -3.19703162e-01 -3.12991560e-01
2.11737767e-01 5.57120681e-01 4.05826867e-01 -7.10939914e-02
5.11756599e-01 -1.31419986e-01 5.04331551e-02 4.49799985e-01
3.84251624e-01 -7.78585672e-02 2.70323724e-01 1.99142814e-01
9.75321651e-01 -2.20315933e-01 4.12842304e-01 -4.05707300e-01
1.97947219e-01 1.34660974e-01 5.06072521e-01 1.38761914e+00
-2.69014180e-01 7.83681393e-01 2.44105682e-01 -7.73298740e-01
-1.29085016e+00 -9.86230552e-01 -5.00924766e-01 1.28235352e+00
1.12098344e-01 -1.31806329e-01 -6.69599056e-01 -8.85277987e-01
-2.04743907e-01 6.67721987e-01 -8.23907018e-01 -2.84941077e-01
-4.38412130e-01 -9.70830142e-01 4.60367918e-01 6.22728348e-01
3.18921864e-01 -1.16783810e+00 -9.48945880e-01 -4.74048890e-02
3.03619236e-01 -7.85694778e-01 -1.75203174e-01 4.74940896e-01
-9.29813504e-01 -9.62406039e-01 -6.75684035e-01 -7.87377298e-01
1.09925508e+00 4.76686537e-01 1.17744505e+00 3.49016339e-01
-6.11284018e-01 2.38611490e-01 -3.88776660e-01 -5.79393387e-01
-6.97855413e-01 2.11228639e-01 -3.80500732e-03 -2.23105907e-01
4.86832887e-01 -1.21931538e-01 -6.27306402e-01 5.11218786e-01
-7.98514724e-01 -7.32816234e-02 9.58226562e-01 8.18882167e-01
4.17631805e-01 -3.74786317e-01 5.41801691e-01 -8.80868673e-01
1.31683409e-01 -2.29890883e-01 -7.16685593e-01 4.55464512e-01
-4.93098974e-01 3.02304089e-01 2.97675371e-01 -7.96628535e-01
-7.92694688e-01 4.06385571e-01 1.13944553e-01 -1.17959589e-01
-3.34391981e-01 -2.88076103e-01 3.09389442e-01 -3.79802465e-01
1.51028824e+00 1.26961991e-01 -4.05467242e-01 -3.89987975e-01
6.45568192e-01 4.90150511e-01 2.96427816e-01 -5.57771206e-01
7.58988380e-01 6.82106733e-01 9.94101614e-02 -8.29954684e-01
-1.02161002e+00 -7.07570374e-01 -8.67675662e-01 -2.03678869e-02
6.43669903e-01 -7.39766359e-01 -1.54753447e-01 -1.54204398e-01
-1.33420300e+00 -2.77515054e-01 -6.02952719e-01 2.81465411e-01
-4.71875817e-01 9.49504897e-02 -2.48442680e-01 -7.16836154e-01
-1.12265825e-01 -1.15632093e+00 1.21937346e+00 3.52215171e-01
-8.80759731e-02 -5.88933051e-01 2.57272186e-04 -1.23156421e-01
2.49378353e-01 -5.67984022e-02 5.85678995e-01 -1.03363419e+00
-8.92056465e-01 -3.45620304e-01 -5.60180664e-01 2.79936582e-01
-2.00278237e-01 1.37667105e-01 -1.24469912e+00 -5.49806118e-01
-2.21811697e-01 -5.28778255e-01 1.56856501e+00 2.97910511e-01
9.56709027e-01 -1.41526163e-01 -7.37512231e-01 4.23043966e-01
1.53078198e+00 -3.38613689e-02 4.83076453e-01 2.15297639e-01
3.74239981e-01 5.77376544e-01 5.80370963e-01 1.14071995e-01
-1.64430469e-01 8.27672839e-01 4.32296634e-01 -1.78752333e-01
-4.80706424e-01 -1.29382715e-01 4.68996502e-02 1.62007049e-01
-1.70675665e-01 -1.71181053e-01 -8.60197842e-01 7.30943561e-01
-1.64920580e+00 -8.42059493e-01 -1.40537238e-02 2.42475986e+00
7.34269142e-01 5.06598115e-01 4.82347310e-01 -1.19853191e-01
8.17545295e-01 -3.74111444e-01 -2.22713009e-01 -3.51829469e-01
-9.96752363e-03 1.60889789e-01 5.91985703e-01 4.27634388e-01
-1.01324761e+00 9.73388493e-01 6.99885845e+00 9.04348195e-01
-9.93057132e-01 2.37863079e-01 6.12138510e-01 -2.69481707e-02
1.62413232e-02 1.68424964e-01 -1.57609665e+00 2.06903666e-01
7.70166218e-01 2.77628720e-01 -1.22826800e-01 1.04142630e+00
-1.85528249e-01 -3.41996434e-03 -1.37806332e+00 7.54910767e-01
-4.22534114e-03 -1.49486589e+00 3.42462450e-01 1.15505077e-01
9.08860743e-01 2.76141763e-01 -3.42945987e-03 5.08185625e-01
4.39494550e-01 -8.54279578e-01 7.73601353e-01 2.79261529e-01
4.70018536e-01 -3.52576405e-01 6.31046772e-01 5.27247250e-01
-8.79548132e-01 -3.81594032e-01 -1.12101507e+00 1.22178048e-01
-1.54821351e-01 3.22086900e-01 -1.33892083e+00 -3.21420543e-02
5.89350343e-01 2.02290058e-01 -1.14654469e+00 1.65069246e+00
-7.37777874e-02 4.05330986e-01 -5.72929978e-01 -2.50409067e-01
3.45095307e-01 2.29082555e-01 3.77802938e-01 1.43324578e+00
3.10609311e-01 -5.03086299e-02 3.24272603e-01 1.02648854e+00
-4.00781296e-02 -1.38665721e-01 -5.77256918e-01 4.84773725e-01
5.27490914e-01 1.43776071e+00 -1.10964561e+00 -5.73934555e-01
-3.08411479e-01 6.54309392e-01 7.52111077e-01 2.36863390e-01
-6.43876255e-01 5.16775027e-02 -1.83504063e-03 3.27490419e-01
4.62297618e-01 1.19551040e-01 -6.41501173e-02 -8.88857901e-01
-2.03165352e-01 -4.45272923e-01 4.54880297e-01 -6.95102096e-01
-1.33893299e+00 8.23252678e-01 4.72039506e-02 -1.33941042e+00
-1.18461167e-02 -9.21061277e-01 -5.72332740e-01 6.47752166e-01
-1.43693125e+00 -1.20888615e+00 -4.88778830e-01 1.59911439e-01
5.80969810e-01 -1.09891199e-01 6.58013880e-01 1.94774084e-02
-4.07646745e-01 7.08485842e-01 -2.61640519e-01 1.10323224e-02
6.41605794e-01 -1.29688752e+00 3.64320189e-01 8.73746276e-01
5.39318025e-01 6.63626790e-01 6.60883784e-01 -4.28683579e-01
-4.89438355e-01 -1.51795566e+00 7.28181481e-01 -6.56888604e-01
2.50038058e-01 -2.81765312e-01 -9.26006079e-01 7.06934929e-01
-1.33540019e-01 3.98642838e-01 3.12077254e-01 1.68041855e-01
-6.21849537e-01 -1.22528009e-01 -1.21649456e+00 5.94082415e-01
1.07252371e+00 -1.64983153e-01 -7.48374999e-01 5.25189161e-01
6.24598742e-01 -1.33123621e-01 -3.44299883e-01 3.39979023e-01
3.94551635e-01 -1.12128758e+00 1.04706633e+00 -6.82758808e-01
1.72818708e-03 -4.08594847e-01 -1.23547405e-01 -1.06918597e+00
-5.43279588e-01 -2.68471569e-01 3.58486652e-01 8.83173943e-01
6.97100580e-01 -5.77015519e-01 8.17779660e-01 2.95790672e-01
-1.90910086e-01 -5.05762458e-01 -8.43246996e-01 -1.21300578e+00
1.77105233e-01 -1.14220999e-01 1.69524550e-01 6.53571546e-01
-3.31561118e-01 3.81047130e-01 -5.17514572e-02 2.28606060e-01
6.96848691e-01 1.20171987e-01 9.33641255e-01 -1.47564435e+00
-3.57953548e-01 -5.63252926e-01 -7.55959630e-01 -1.19173193e+00
-3.69010121e-01 -9.81474817e-01 1.81642339e-01 -1.24951816e+00
5.47619820e-01 -9.83760655e-01 -2.92535335e-01 7.17048705e-01
-1.41918987e-01 6.23363078e-01 2.73793191e-01 4.64517802e-01
-6.92293167e-01 -6.10819878e-03 1.17450583e+00 -1.13196984e-01
-3.34131330e-01 1.54122487e-01 -6.31194174e-01 7.01859057e-01
6.93492234e-01 -7.39631951e-01 -2.86601365e-01 -1.42946303e-01
1.88958108e-01 -4.91738498e-01 6.44960523e-01 -1.15895629e+00
1.34761363e-01 -4.61459272e-02 5.51928580e-01 -7.10662842e-01
3.62986147e-01 -6.20860100e-01 -2.03282446e-01 7.24147975e-01
-5.11247814e-01 -3.89617592e-01 2.68983811e-01 6.84206009e-01
2.47627944e-01 -6.58993781e-01 1.20682657e+00 -3.14313024e-01
-8.19752216e-01 1.66109353e-01 -7.63389543e-02 2.91532986e-02
1.13083696e+00 -5.78547776e-01 -5.78177571e-01 1.87925458e-01
-9.33874071e-01 -1.92372099e-01 6.17718279e-01 3.23880494e-01
4.62932557e-01 -1.11381054e+00 -7.83027351e-01 1.90829992e-01
5.05414188e-01 7.13328924e-03 -2.93416500e-01 5.50307989e-01
-4.18462604e-01 4.40651238e-01 -3.62248123e-01 -9.72386181e-01
-1.09332001e+00 7.62293994e-01 5.24222255e-01 1.04780808e-01
-5.98976314e-01 1.19866228e+00 6.93990886e-01 -2.45918572e-01
1.19509622e-01 -6.39421463e-01 -6.30566627e-02 -2.69034896e-02
5.32634437e-01 1.72991082e-01 1.54219475e-02 -4.20620650e-01
-2.93047279e-01 5.97650111e-01 -4.26032454e-01 7.41920397e-02
1.03125501e+00 1.57018229e-01 3.01239669e-01 9.54039320e-02
9.00555670e-01 -1.82469696e-01 -1.39249241e+00 -2.47407317e-01
4.94566523e-02 -5.26498854e-01 2.95665278e-03 -6.96382344e-01
-8.28158736e-01 9.25485194e-01 9.86222863e-01 4.56445813e-01
8.18327427e-01 4.21329409e-01 7.89074376e-02 5.89397252e-01
5.08015275e-01 -7.87688553e-01 3.49919647e-01 1.91454515e-01
8.22557330e-01 -1.43367946e+00 7.02372519e-03 -5.42896688e-01
-3.46979737e-01 1.19712222e+00 7.87033796e-01 -4.94745016e-01
3.97208452e-01 -1.71031386e-01 -9.38112438e-02 -3.15507442e-01
-7.25559175e-01 -6.04534745e-01 5.61781049e-01 7.35919058e-01
2.05543712e-01 1.41075253e-01 -1.57929137e-01 2.24767745e-01
-3.97607349e-02 -2.77490973e-01 8.01514030e-01 6.07217431e-01
-1.22862971e+00 -9.69652295e-01 -3.34696740e-01 5.37877202e-01
-8.84142239e-04 3.41003016e-02 -4.96448874e-01 9.57827806e-01
5.10141492e-01 6.94886029e-01 2.06410244e-01 -8.95382240e-02
1.40391484e-01 -9.10437778e-02 7.02064335e-01 -1.30879188e+00
-4.38432455e-01 -7.92209059e-02 -2.48154271e-02 -2.94143736e-01
-4.10525620e-01 -3.68734837e-01 -9.85179305e-01 8.19868445e-02
-9.24805582e-01 -2.75844187e-01 4.17602420e-01 8.49143863e-01
1.57747030e-01 2.87428916e-01 3.74000460e-01 -9.18918908e-01
-6.32761955e-01 -9.51117992e-01 -4.26758230e-01 2.60060608e-01
2.95213640e-01 -7.36227095e-01 -5.16012132e-01 -1.06346227e-01] | [9.360478401184082, 1.597768783569336] |
1af98938-77b1-42f3-8543-ae80580e424f | graph-contrastive-learning-with-multi | 2307.04322 | null | https://arxiv.org/abs/2307.04322v1 | https://arxiv.org/pdf/2307.04322v1.pdf | Graph Contrastive Learning with Multi-Objective for Personalized Product Retrieval in Taobao Search | In e-commerce search, personalized retrieval is a crucial technique for improving user shopping experience. Recent works in this domain have achieved significant improvements by the representation learning paradigm, e.g., embedding-based retrieval (EBR) and collaborative filtering (CF). EBR methods do not sufficiently exploit the useful collaborative signal and are difficult to learn the representations of long-tail item well. Graph-based CF methods improve personalization by modeling collaborative signal within the user click graph. However, existing Graph-based methods ignore user's multiple behaviours, such as click/purchase and the relevance constraint between user behaviours and items.In this paper, we propose a Graph Contrastive Learning with Multi-Objective (GCL-MO) collaborative filtering model, which solves the problems of weak relevance and incomplete personalization in e-commerce search. Specifically, GCL-MO builds a homogeneous graph of items and then optimizes a multi-objective function of personalization and relevance. Moreover, we propose a modified contrastive loss for multi-objectives graph learning, which avoids the mutual suppression among positive samples and thus improves the generalization and robustness of long-tail item representations. These learned item embeddings are then used for personalized retrieval by constructing an efficient offline-to-online inverted table. GCL-MO outperforms the online collaborative filtering baseline in both offline/online experimental metrics and shows a significant improvement in the online A/B testing of Taobao search. | ['Xiaoyi Zeng', 'Qingwen Liu', 'Yun Zhong', 'Sen Li', 'Chao Zhang', 'Longbin Li'] | 2023-07-10 | null | null | null | null | ['contrastive-learning', 'graph-learning', 'contrastive-learning', 'representation-learning', 'retrieval', 'collaborative-filtering'] | ['computer-vision', 'graphs', 'methodology', 'methodology', 'methodology', 'miscellaneous'] | [-1.36367440e-01 -4.82336462e-01 -6.30878150e-01 -3.65137428e-01
-7.49457777e-01 -4.93869156e-01 2.14212343e-01 4.72968519e-01
-4.56169516e-01 2.07217395e-01 2.67469317e-01 -2.33352631e-01
-9.60658967e-01 -1.06666398e+00 -5.48154354e-01 -4.49559689e-01
-3.38966399e-01 4.18560207e-01 2.64042497e-01 -6.27399564e-01
1.93203241e-01 1.72181875e-01 -1.30898106e+00 1.84324488e-01
1.07252860e+00 1.33025038e+00 2.26433069e-01 4.78568226e-01
-1.32801175e-01 1.16446227e-01 -2.13653982e-01 -7.83432662e-01
4.55720991e-01 -3.34968239e-01 -2.71140069e-01 -1.61633700e-01
3.32696438e-01 -2.85279248e-02 -8.10584247e-01 1.07928324e+00
7.94421554e-01 6.61074936e-01 6.29839897e-01 -1.23542917e+00
-1.54008055e+00 6.91121638e-01 -6.71062827e-01 3.23678493e-01
3.58629614e-01 -2.41170034e-01 1.73100889e+00 -9.44170773e-01
3.63182306e-01 1.21067548e+00 6.73812270e-01 2.46378943e-01
-1.30343056e+00 -6.56312168e-01 6.21660292e-01 5.23856223e-01
-1.45916831e+00 3.10096711e-01 1.00582755e+00 -1.01826251e-01
7.22381175e-01 5.02226651e-01 6.35421515e-01 9.95650053e-01
1.58128530e-01 1.19138908e+00 4.94420946e-01 -1.98904052e-01
4.67060171e-02 2.74272114e-01 5.47518909e-01 5.86216450e-01
4.43024427e-01 1.65620103e-01 -5.91673732e-01 -3.28760147e-01
6.95997536e-01 6.14304364e-01 -1.85204163e-01 -5.31855047e-01
-7.73246169e-01 1.24893844e+00 8.28967035e-01 2.71172673e-01
-4.29460526e-01 -3.69880199e-02 3.92270684e-01 6.34460151e-01
4.17909473e-01 3.88612509e-01 -4.73437220e-01 5.03234744e-01
-5.10549009e-01 1.02523386e-01 6.62182450e-01 1.10167074e+00
7.55234122e-01 -1.73542067e-01 -4.07763183e-01 1.10335386e+00
7.72671700e-01 4.07523781e-01 8.79695714e-01 -3.65083814e-01
4.87876505e-01 6.36561334e-01 -1.47932068e-01 -1.51321089e+00
-1.74745172e-01 -8.35871458e-01 -7.42744386e-01 -4.52616751e-01
-9.64306146e-02 1.15521103e-01 -5.67325950e-01 1.45626843e+00
3.10335755e-01 -7.42711201e-02 -2.42865279e-01 9.94610846e-01
7.36246645e-01 5.48603296e-01 9.81830433e-02 -1.63861737e-01
1.08411920e+00 -1.19774568e+00 -8.29300761e-01 -1.89171106e-01
4.97875839e-01 -7.13111222e-01 1.17595959e+00 2.89766282e-01
-8.74076068e-01 -6.29770517e-01 -9.88267422e-01 -7.36928657e-02
-5.66314161e-01 -1.77712694e-01 1.16013753e+00 8.81107509e-01
-7.22837090e-01 6.85149550e-01 -2.60953814e-01 5.92001565e-02
3.52147102e-01 6.34737432e-01 -9.99019593e-02 -4.02474284e-01
-1.68504870e+00 4.16725814e-01 3.38111222e-01 -5.31835146e-02
-3.55802566e-01 -7.88783908e-01 -7.70775437e-01 3.32164526e-01
5.86427212e-01 -7.25104868e-01 7.03661919e-01 -7.16934204e-01
-1.34890795e+00 3.30927998e-01 2.56016523e-01 -3.55620265e-01
3.69930983e-01 -3.44908684e-01 -8.55634153e-01 -2.14605227e-01
-1.14656925e-01 8.53097513e-02 1.02784753e+00 -1.17308748e+00
-6.59761965e-01 -7.13123500e-01 9.05869752e-02 2.63460815e-01
-8.01814497e-01 -4.83983517e-01 -9.74859834e-01 -9.42492068e-01
6.89733699e-02 -6.96255326e-01 -2.88949460e-01 -2.29925439e-01
1.09438851e-01 -5.10891855e-01 5.91131389e-01 -7.26159036e-01
1.75936627e+00 -2.26223373e+00 3.29208523e-02 5.90528429e-01
1.16766557e-01 3.01386714e-01 -6.48467720e-01 6.06264055e-01
1.08753897e-01 1.49178892e-01 4.58746940e-01 -2.32856855e-01
3.85001808e-01 8.80716518e-02 -1.07135259e-01 2.92800367e-01
-3.47463489e-01 1.33817184e+00 -1.06802237e+00 -4.27987963e-01
2.71375049e-02 5.29749155e-01 -9.12198067e-01 6.93685561e-02
-1.96150951e-02 -1.03425823e-01 -6.32756770e-01 5.79396844e-01
5.89359164e-01 -4.09537941e-01 3.31119090e-01 -3.42829108e-01
3.91628087e-01 4.98743877e-02 -1.19067276e+00 1.67336726e+00
-7.93501616e-01 -3.58745479e-03 -5.10108583e-02 -8.84241283e-01
9.54995275e-01 -1.31957069e-01 5.21669805e-01 -1.24625921e+00
-6.76539689e-02 1.35341868e-01 -2.89078444e-01 -2.38980398e-01
8.22410464e-01 3.41807425e-01 -4.77143079e-02 3.57005954e-01
1.71480790e-01 3.75986844e-01 1.35696828e-01 4.16365534e-01
8.74805093e-01 -1.69458345e-01 1.49580672e-01 -1.29940152e-01
4.38726246e-01 -5.40152729e-01 3.97157431e-01 9.32650328e-01
-4.50478829e-02 3.43985766e-01 -2.87602078e-02 6.06479831e-02
-5.94577432e-01 -1.08788645e+00 1.79660082e-01 1.71264207e+00
5.47788918e-01 -6.79660380e-01 8.03299919e-02 -9.99621809e-01
6.54525459e-01 7.38589168e-01 -6.34846032e-01 -6.44789040e-01
-4.27186519e-01 -8.08368027e-01 -2.25058272e-01 4.85448241e-01
2.12935165e-01 -6.30254447e-01 5.11350274e-01 3.22367817e-01
6.23455569e-02 -4.36622024e-01 -1.43954170e+00 2.22290922e-02
-9.96862471e-01 -8.45144451e-01 -7.17207372e-01 -8.79555941e-01
4.28140342e-01 7.52785504e-01 9.22739148e-01 -5.73423412e-03
-3.22073013e-01 8.08685184e-01 -6.26587212e-01 -7.78977014e-03
3.20037007e-01 6.59299195e-02 -6.51310757e-02 3.73448372e-01
5.34510195e-01 -3.84497315e-01 -1.13246143e+00 7.87486255e-01
-1.06499672e+00 -8.22352648e-01 7.55430877e-01 1.21298528e+00
7.84316242e-01 2.90354580e-01 9.55396950e-01 -9.84957457e-01
1.24661756e+00 -8.38669479e-01 -5.09899259e-01 5.12255728e-01
-1.33199537e+00 2.71330904e-02 5.48806190e-01 -8.30787003e-01
-8.11611950e-01 -4.10512984e-01 -4.18529734e-02 -5.40481746e-01
7.18546629e-01 7.61099219e-01 -1.25373110e-01 -2.11680472e-01
5.46886444e-01 2.38649085e-01 -2.45599121e-01 -6.03840590e-01
6.86143517e-01 6.95232928e-01 4.31814184e-03 -1.78743526e-01
6.68327570e-01 -6.10471191e-03 -1.19707413e-01 -6.83986485e-01
-6.83789074e-01 -1.19253659e+00 -7.26201832e-02 4.68354188e-02
3.79871279e-01 -6.98492944e-01 -9.46312666e-01 -1.13368712e-01
-4.47395861e-01 9.22292843e-02 -4.17784959e-01 7.90006161e-01
-2.30123982e-01 8.39794397e-01 -5.41869462e-01 -7.74188280e-01
-5.74223459e-01 -6.40147269e-01 7.92822301e-01 2.43842110e-01
2.40576982e-01 -1.26497960e+00 4.02457044e-02 4.46109474e-01
5.53204894e-01 -4.87807065e-01 1.02239168e+00 -9.52936888e-01
-5.36338806e-01 -6.95764184e-01 -2.79999942e-01 4.64256585e-01
1.93942383e-01 -6.49451077e-01 -4.70694155e-01 -5.82738101e-01
-2.33923003e-01 -4.56948802e-02 1.07424307e+00 3.09738368e-01
1.06505907e+00 -3.01977932e-01 -3.54925156e-01 4.58511293e-01
1.49989927e+00 3.11467975e-01 5.46030343e-01 1.53638095e-01
6.38445854e-01 3.83619368e-01 7.96350598e-01 4.20860082e-01
2.92353570e-01 7.09058225e-01 2.65866458e-01 2.48982698e-01
-1.14581980e-01 -7.03757823e-01 2.77153194e-01 1.02965117e+00
7.41030872e-02 -4.88622546e-01 -1.25142053e-01 3.80563527e-01
-2.13788486e+00 -6.76539302e-01 5.78984208e-02 2.57030439e+00
6.11181796e-01 -4.47119237e-04 1.38112962e-01 8.10480211e-03
8.68600726e-01 1.38873413e-01 -6.65231586e-01 -9.44257081e-02
1.11787980e-02 2.04075381e-01 7.07790017e-01 4.50655371e-01
-1.06870770e+00 6.76504016e-01 5.51751709e+00 1.38577831e+00
-5.74952245e-01 2.45540351e-01 1.36131361e-01 8.49863440e-02
-7.80124962e-01 -3.89420122e-01 -1.00895619e+00 4.11038727e-01
5.63616037e-01 -3.79205853e-01 7.30348349e-01 1.01379621e+00
-1.69590056e-01 3.30069780e-01 -9.67457891e-01 1.29087222e+00
3.58017057e-01 -1.06442606e+00 2.65390813e-01 6.95608333e-02
7.91913271e-01 -2.12268665e-01 4.69473094e-01 8.39472532e-01
2.63414145e-01 -6.12901688e-01 1.68523446e-01 6.54420018e-01
3.40509951e-01 -7.69426882e-01 6.79124415e-01 7.33214021e-02
-1.26161253e+00 -4.35517520e-01 -5.38694739e-01 6.00375473e-01
8.44384581e-02 5.41186810e-01 -2.49341980e-01 9.23296630e-01
6.02732658e-01 7.74998486e-01 -6.97160482e-01 1.39130211e+00
7.48066306e-02 4.34479058e-01 -2.02333927e-01 -4.78016973e-01
1.55507863e-01 -6.43589914e-01 4.91838902e-01 9.95151818e-01
1.95029497e-01 -1.89890772e-01 2.77442157e-01 5.58150351e-01
-2.97930688e-01 6.40162647e-01 -3.67746204e-01 -4.58486229e-01
1.24054849e-01 1.16859734e+00 -3.88692260e-01 -5.12889288e-02
-6.38949037e-01 1.14341712e+00 2.54241645e-01 4.91371810e-01
-6.43719673e-01 -6.73459947e-01 4.85829264e-01 5.10242283e-02
7.33982146e-01 -1.89123407e-01 1.52837351e-01 -1.17144382e+00
5.44088371e-02 -7.36064374e-01 8.11349273e-01 -1.91079438e-01
-1.90700567e+00 2.24362060e-01 -2.92624921e-01 -1.15487862e+00
1.49565309e-01 -4.75473613e-01 -2.51593500e-01 7.72509158e-01
-1.63310325e+00 -1.06727254e+00 -5.75521169e-03 8.60143721e-01
4.88852531e-01 -2.44192377e-01 6.36213481e-01 8.69662285e-01
-3.69038552e-01 1.18581331e+00 4.59857017e-01 -2.02356920e-01
7.49414146e-01 -1.12939394e+00 1.04641967e-01 2.43293867e-01
3.95005733e-01 1.01800776e+00 2.24398002e-01 -8.45873356e-01
-1.91355073e+00 -1.14288557e+00 8.20941269e-01 -9.04586241e-02
9.52039778e-01 -2.89617985e-01 -8.12386096e-01 5.96529007e-01
-1.20719098e-01 -4.19276990e-02 9.98846292e-01 6.70232654e-01
-6.20760143e-01 -4.22924995e-01 -9.81267214e-01 6.18286967e-01
1.25526679e+00 -5.94910204e-01 -5.17350793e-01 6.66814983e-01
8.43681037e-01 2.12560177e-01 -1.20070350e+00 2.02052474e-01
6.31649017e-01 -6.63626730e-01 1.30659556e+00 -7.56734848e-01
-2.42382452e-01 5.00522628e-02 -3.53370994e-01 -1.37693942e+00
-8.17687452e-01 -7.62229204e-01 -3.28823388e-01 1.00379968e+00
5.12950599e-01 -9.83927608e-01 7.13678658e-01 3.32098037e-01
5.78912869e-02 -8.41580391e-01 -4.40315962e-01 -1.02262831e+00
-2.54443496e-01 -2.65002787e-01 5.57123899e-01 7.25941539e-01
-1.30580459e-02 6.29954517e-01 -5.25381923e-01 3.85222919e-02
6.81436360e-01 4.39987034e-01 4.44724917e-01 -1.25356054e+00
-7.89690554e-01 -5.00142932e-01 -3.84352744e-01 -1.45740974e+00
-1.81464374e-01 -1.32010365e+00 -5.27457595e-01 -1.46931064e+00
5.44764102e-02 -4.75647599e-01 -8.55643213e-01 -2.86728144e-02
-4.06638294e-01 2.21004888e-01 1.67530224e-01 1.86083168e-01
-1.02596378e+00 7.80687928e-01 1.39246798e+00 -3.66893470e-01
-5.18187821e-01 3.34590912e-01 -8.36062074e-01 9.42200348e-02
5.36226034e-01 -2.47068763e-01 -5.84921479e-01 -1.77791864e-01
7.17576504e-01 -3.52299362e-02 -1.57387361e-01 -3.76911640e-01
3.23244482e-01 1.26184717e-01 3.81907910e-01 -5.05152047e-01
3.29076499e-01 -9.41543519e-01 2.53408179e-02 2.76713878e-01
-5.75487137e-01 -8.86622816e-02 -2.49602586e-01 1.34367549e+00
-2.57261753e-01 -2.45838076e-01 3.37919682e-01 -6.34453213e-03
-7.11294353e-01 7.63942838e-01 1.63105920e-01 -1.32817209e-01
7.95869708e-01 6.90076277e-02 -4.41245586e-02 -4.72829401e-01
-9.12853539e-01 6.00771427e-01 -1.63372859e-01 8.14010024e-01
7.38384664e-01 -1.66652262e+00 -4.04055625e-01 2.20493913e-01
3.47238123e-01 -7.06241608e-01 5.20606101e-01 7.15707242e-01
5.44153340e-02 5.00444531e-01 1.58578664e-01 -1.21230371e-01
-1.11856902e+00 1.08164096e+00 -1.19717799e-01 -5.40846050e-01
-2.84853786e-01 9.22513485e-01 1.98893979e-01 -3.79824042e-01
4.10091817e-01 1.19561791e-01 -5.01490533e-01 4.71915603e-01
3.91362220e-01 6.37535274e-01 1.09547429e-01 -3.32333446e-01
-8.58188793e-02 4.88145769e-01 -5.32570779e-01 2.71984935e-01
1.20542085e+00 -4.19635952e-01 2.69348264e-01 2.47279763e-01
1.58738554e+00 2.59550571e-01 -7.27626503e-01 -5.31693518e-01
-3.78009342e-02 -8.70002270e-01 3.43004256e-01 -6.86934352e-01
-1.21303654e+00 4.63974327e-01 8.47827792e-01 5.02018273e-01
9.69237626e-01 -1.85317814e-01 1.11004639e+00 4.73451793e-01
4.63125646e-01 -1.31358898e+00 3.32403898e-01 2.93135166e-01
9.46086586e-01 -1.19130993e+00 5.40289022e-02 -4.57860798e-01
-6.20799720e-01 8.16803634e-01 3.00699085e-01 -2.81474203e-01
1.03376865e+00 -4.96634156e-01 -2.06826985e-01 -1.58013791e-01
-5.04965842e-01 -4.07026142e-01 7.69994259e-01 4.77082878e-01
1.84431642e-01 1.66951343e-01 -8.00068498e-01 9.46646452e-01
3.63504618e-01 -1.58448160e-01 -3.80059838e-01 6.63862109e-01
-2.08414167e-01 -1.24979520e+00 5.34232408e-02 7.90903926e-01
-3.17590773e-01 -2.93250740e-01 -2.15915129e-01 5.48769236e-01
-2.62874812e-01 1.00018370e+00 -3.61275494e-01 -8.30889821e-01
5.06114662e-01 -3.10109891e-02 4.09061015e-01 -4.47542280e-01
-7.83719957e-01 3.06141138e-01 -2.17189685e-01 -6.85361266e-01
-1.47338975e-02 -3.98384839e-01 -7.34845877e-01 -6.87207654e-02
-1.02543771e+00 5.54531336e-01 6.56762600e-01 2.49442920e-01
6.49009347e-01 4.55604583e-01 1.02741599e+00 -5.76431334e-01
-9.54966307e-01 -7.93563902e-01 -1.20099258e+00 6.67107940e-01
-2.04047840e-02 -6.80579066e-01 -5.99870801e-01 -6.21000051e-01] | [10.182642936706543, 5.6233229637146] |
2ae4aa78-b3d4-4955-88e1-85d874d9af4c | multi-head-attention-mechanism-learning-for | 2307.04075 | null | https://arxiv.org/abs/2307.04075v1 | https://arxiv.org/pdf/2307.04075v1.pdf | Multi-Head Attention Mechanism Learning for Cancer New Subtypes and Treatment Based on Cancer Multi-Omics Data | Due to the high heterogeneity and clinical characteristics of cancer, there are significant differences in multi-omics data and clinical features among subtypes of different cancers. Therefore, the identification and discovery of cancer subtypes are crucial for the diagnosis, treatment, and prognosis of cancer. In this study, we proposed a generalization framework based on attention mechanisms for unsupervised contrastive learning (AMUCL) to analyze cancer multi-omics data for the identification and characterization of cancer subtypes. AMUCL framework includes a unsupervised multi-head attention mechanism, which deeply extracts multi-omics data features. Importantly, a decoupled contrastive learning model (DMACL) based on a multi-head attention mechanism is proposed to learn multi-omics data features and clusters and identify new cancer subtypes. This unsupervised contrastive learning method clusters subtypes by calculating the similarity between samples in the feature space and sample space of multi-omics data. Compared to 11 other deep learning models, the DMACL model achieved a C-index of 0.002, a Silhouette score of 0.801, and a Davies Bouldin Score of 0.38 on a single-cell multi-omics dataset. On a cancer multi-omics dataset, the DMACL model obtained a C-index of 0.016, a Silhouette score of 0.688, and a Davies Bouldin Score of 0.46, and obtained the most reliable cancer subtype clustering results for each type of cancer. Finally, we used the DMACL model in the AMUCL framework to reveal six cancer subtypes of AML. By analyzing the GO functional enrichment, subtype-specific biological functions, and GSEA of AML, we further enhanced the interpretability of cancer subtype analysis based on the generalizable AMUCL framework. | ['Shaoliang Peng', 'Liwen Xu', 'Pengfei Rong', 'Zhichao Feng', 'Lian Wang', 'Yutao Dou', 'Dazhen Liu', 'Liangrui Pan'] | 2023-07-09 | null | null | null | null | ['contrastive-learning', 'contrastive-learning'] | ['computer-vision', 'methodology'] | [-3.71493638e-01 -2.75511622e-01 -4.07103807e-01 -1.13500886e-01
-7.77310073e-01 -2.35799029e-01 3.63208681e-01 6.28483951e-01
-1.90365002e-01 4.92768019e-01 2.99914092e-01 -2.88573265e-01
-5.84804952e-01 -9.41724181e-01 -3.45237970e-01 -1.31559718e+00
2.08351970e-01 4.89258677e-01 -4.60203528e-01 1.16050206e-01
-3.69495363e-04 5.59720337e-01 -1.32340550e+00 6.54850721e-01
9.78055954e-01 1.02877080e+00 2.23735526e-01 4.80778039e-01
-3.10591847e-01 2.45123446e-01 -2.82179654e-01 9.86185372e-02
-3.35475475e-01 -3.88734788e-01 -4.91554081e-01 -2.19078690e-01
7.71925598e-02 2.20328584e-01 -4.80774194e-01 9.51762497e-01
5.86160183e-01 -3.22811037e-01 9.79557395e-01 -1.28501546e+00
-7.69716978e-01 1.79192349e-01 -3.18562567e-01 6.24375045e-02
2.70213354e-02 6.69378191e-02 1.18084335e+00 -8.77176344e-01
4.96589094e-01 9.83769596e-01 5.20913303e-01 5.49831867e-01
-1.06349194e+00 -7.24888504e-01 1.04113474e-01 1.94392547e-01
-1.56732285e+00 -1.73520580e-01 2.60580063e-01 -5.15372217e-01
8.31084609e-01 7.55578995e-01 8.92307401e-01 7.47716725e-01
6.02689147e-01 8.08278561e-01 7.56169021e-01 -8.15279782e-02
2.02809229e-01 -1.03826761e-01 4.06844199e-01 8.18727493e-01
3.29573989e-01 -3.33260782e-02 -1.43905461e-01 -3.48825663e-01
2.07385063e-01 8.92082810e-01 -1.26450241e-01 1.38614014e-01
-1.28979218e+00 7.92347908e-01 5.96737802e-01 5.12806952e-01
-1.85973287e-01 -5.78637235e-02 3.38797927e-01 -3.79719981e-03
4.39377636e-01 4.96808141e-01 -5.23589849e-01 4.14445519e-01
-5.31434417e-01 -1.59536928e-01 2.89390147e-01 7.77956128e-01
6.33985639e-01 -3.50630552e-01 -3.82303357e-01 9.11514640e-01
4.47184712e-01 4.29889083e-01 9.41951990e-01 -3.11289370e-01
-3.46902966e-01 1.22674453e+00 -5.89221239e-01 -8.33616734e-01
-9.79129076e-01 -6.99633539e-01 -1.23124874e+00 -5.11869192e-01
6.35255873e-02 1.98923409e-01 -6.67591810e-01 1.64080691e+00
3.99249434e-01 2.57499725e-01 8.55171159e-02 7.69357026e-01
1.29953277e+00 3.61811459e-01 2.43023053e-01 -2.01741710e-01
1.61769700e+00 -6.70016229e-01 -6.28362477e-01 4.06835824e-01
1.11335814e+00 -1.84400707e-01 1.13294411e+00 1.22161292e-01
-3.93052906e-01 -2.84683824e-01 -7.38063276e-01 -8.84324908e-02
-6.11692846e-01 2.69342601e-01 8.44986916e-01 3.73411477e-01
-6.34014070e-01 3.84417504e-01 -6.62651837e-01 -4.53286469e-01
6.72464013e-01 4.11487818e-01 -4.84701276e-01 -1.84958339e-01
-9.08959150e-01 2.00705141e-01 4.11110878e-01 -7.38347769e-02
-8.97433162e-01 -1.08702981e+00 -6.18650079e-01 2.72619694e-01
-1.57224014e-01 -1.06816018e+00 5.02550840e-01 -6.29572511e-01
-1.34424829e+00 8.52140367e-01 -3.36745948e-01 2.40768090e-01
-9.01710019e-02 1.18307099e-01 -5.73743284e-01 9.89705324e-02
2.79062182e-01 4.29720968e-01 3.32948491e-02 -9.09716249e-01
-8.10339093e-01 -7.63081968e-01 -4.14547026e-01 1.52601689e-01
-5.81220388e-01 -3.95343989e-01 -2.84259439e-01 -3.26219976e-01
2.60185957e-01 -9.74906147e-01 -2.15090975e-01 -2.26860836e-01
-4.24046606e-01 -2.65906572e-01 6.53412223e-01 -2.63670951e-01
1.20219982e+00 -2.44669008e+00 3.98555607e-01 9.22938138e-02
7.03349411e-01 -5.28136343e-02 -1.61542028e-01 1.60143748e-01
-1.41901523e-01 4.46141094e-01 5.31904399e-02 -2.18914612e-03
-1.55431554e-01 -9.64572057e-02 7.55778924e-02 8.34057570e-01
3.16673636e-01 1.21671808e+00 -1.07762647e+00 -3.02515268e-01
6.92063151e-03 3.41317564e-01 -6.20237470e-01 3.70723605e-01
-8.99162069e-02 2.95736670e-01 -5.08702159e-01 1.07784867e+00
4.60523903e-01 -5.08971930e-01 1.17631942e-01 -3.05592179e-01
-5.88380285e-02 -1.82835683e-01 -3.92691642e-01 1.35532272e+00
-3.34293067e-01 4.92208660e-01 -2.19790459e-01 -8.34155679e-01
7.90112436e-01 1.86297506e-01 9.53863561e-01 -5.62540650e-01
2.79609203e-01 3.18779767e-01 3.38037372e-01 -6.49452388e-01
-4.42259684e-02 -3.33294064e-01 -2.18032837e-01 1.58318162e-01
2.73874979e-02 3.55982333e-01 -7.69074038e-02 9.91616324e-02
1.24421906e+00 -6.62352324e-01 2.47290522e-01 -6.47147596e-01
6.66470051e-01 4.75310311e-02 7.65815020e-01 3.99239719e-01
-9.29256231e-02 5.39094925e-01 6.52890503e-01 -4.71197993e-01
-7.64933586e-01 -8.50647330e-01 -6.70897245e-01 9.84975934e-01
1.77934900e-01 -4.53271002e-01 -2.14469850e-01 -6.53771341e-01
2.30439574e-01 3.52935553e-01 -8.20334017e-01 -6.73317194e-01
1.32144123e-01 -1.56532741e+00 8.30700636e-01 3.09446841e-01
4.19837803e-01 -6.56446993e-01 1.60288677e-01 -5.03380448e-02
-1.11024655e-01 -7.67325878e-01 -4.25045848e-01 3.38201255e-01
-6.79809511e-01 -1.31900454e+00 -6.10729337e-01 -6.67050719e-01
8.85303974e-01 2.35056385e-01 5.52548587e-01 2.14569494e-01
-4.92601842e-01 1.26580233e-02 -1.88464895e-01 -6.09197021e-01
-2.74949163e-01 1.82378799e-01 1.17898189e-01 2.12633878e-01
6.95879817e-01 -1.80101946e-01 -5.57656884e-01 3.35041285e-01
-7.96346724e-01 1.27575696e-01 8.68319809e-01 1.26089346e+00
9.20848787e-01 -4.53373976e-02 9.56424415e-01 -8.29320669e-01
3.56428742e-01 -1.09942007e+00 -2.00686410e-01 1.95647985e-01
-7.72815704e-01 -1.20444037e-01 8.07996929e-01 -3.24736834e-01
-5.92605531e-01 -7.04902485e-02 -2.44146526e-01 -3.21925223e-01
-4.28152889e-01 6.92364097e-01 -6.93605840e-01 1.82685658e-01
3.14338475e-01 4.07466441e-01 1.22425050e-01 -2.47867107e-01
-1.19411692e-01 1.08776736e+00 1.06538236e-01 -3.12848866e-01
-1.13534771e-01 4.85462695e-01 2.91207612e-01 -9.78295267e-01
-9.10542786e-01 -8.21213841e-01 -5.36430657e-01 1.83172509e-01
1.10402524e+00 -1.05691242e+00 -1.23212922e+00 6.65468812e-01
-6.23528719e-01 -2.92418718e-01 1.66152567e-01 6.71145380e-01
-2.80448347e-01 3.46459687e-01 -9.06347096e-01 -2.69862503e-01
-5.36850810e-01 -1.10649383e+00 1.23026884e+00 6.65942580e-02
-2.31690690e-01 -1.00320768e+00 2.28647649e-01 3.40426534e-01
3.66425402e-02 1.56895339e-01 1.56208622e+00 -9.56131637e-01
-1.42214537e-01 -2.85302818e-01 -5.67425787e-01 -2.53359228e-01
4.99529153e-01 -1.12721883e-02 -1.07086349e+00 -1.87689260e-01
-3.24734926e-01 1.25251561e-01 9.60244715e-01 4.96335655e-01
1.69537032e+00 -2.05892056e-01 -9.12481725e-01 1.15945208e+00
1.32864785e+00 3.48632127e-01 3.90371710e-01 1.72350228e-01
1.01755154e+00 2.13102579e-01 3.30217302e-01 3.04232776e-01
4.11441147e-01 4.17526364e-01 5.40983379e-01 -2.04619512e-01
1.64286509e-01 8.71638060e-02 7.55829886e-02 8.53237569e-01
1.49454936e-01 -3.30233365e-01 -1.02305543e+00 4.70762461e-01
-1.80230761e+00 -8.12495708e-01 -5.42877734e-01 1.90650332e+00
6.36009812e-01 -3.71425509e-01 3.64040583e-02 -3.23814511e-01
6.86277688e-01 -3.93293232e-01 -8.51324081e-01 -8.06524456e-02
-3.68959576e-01 -2.39351094e-01 2.32313693e-01 1.26786619e-01
-1.00295520e+00 6.13358438e-01 5.96977854e+00 8.06917787e-01
-1.30211782e+00 -1.16799220e-01 8.60082567e-01 -1.96078360e-01
-3.71080488e-01 -3.01761478e-01 -1.00835240e+00 8.03177536e-01
9.39946353e-01 -2.25161925e-01 -5.62696904e-02 5.60544670e-01
1.51260018e-01 2.07465887e-01 -1.12193632e+00 1.00501657e+00
-1.03238359e-01 -1.51894224e+00 1.44820169e-01 6.58073008e-01
6.60161912e-01 2.40297586e-01 -3.87270339e-02 3.18022192e-01
1.86711356e-01 -1.24025369e+00 4.56953086e-02 6.56103194e-01
9.83842969e-01 -7.58011401e-01 1.00618601e+00 3.29544872e-01
-1.01494813e+00 -4.23707902e-01 -4.72104698e-01 1.72447413e-01
-5.33793032e-01 9.90524173e-01 -7.26178169e-01 7.44230032e-01
5.97585022e-01 1.04571855e+00 -4.42461640e-01 9.67051983e-01
2.63772726e-01 3.28303248e-01 -4.12110128e-02 -2.77480215e-01
5.87126054e-02 -2.22781122e-01 1.87155187e-01 1.47303534e+00
4.95698512e-01 2.69540608e-01 1.65760338e-01 8.55005324e-01
-2.85668913e-02 3.34609419e-01 -1.54105887e-01 -3.17292035e-01
4.00130659e-01 1.71245551e+00 -5.62678337e-01 -3.87181073e-01
-2.15187430e-01 6.25316381e-01 3.20532829e-01 1.83397785e-01
-8.48188579e-01 -4.65431929e-01 1.00579000e+00 -3.89146395e-02
-1.80015206e-01 1.78859472e-01 -3.03452015e-01 -1.38045514e+00
-6.17243111e-01 -7.95579135e-01 7.99288630e-01 -2.60848880e-01
-1.68332040e+00 4.65747476e-01 -3.66003394e-01 -1.04245222e+00
3.43331456e-01 -8.32776487e-01 -7.04865694e-01 7.84713626e-01
-1.29720688e+00 -1.15296388e+00 -7.59080708e-01 3.66206110e-01
2.07015663e-01 -4.79233861e-01 1.19980407e+00 3.66347551e-01
-1.24922156e+00 6.24359906e-01 4.96117800e-01 5.10825776e-02
6.76682949e-01 -1.17957830e+00 -3.74437004e-01 -3.55616882e-02
-3.59385014e-01 5.40900171e-01 5.57323471e-02 -5.91783822e-01
-1.59193707e+00 -1.60026968e+00 6.12700880e-01 -1.20290995e-01
6.30983770e-01 -2.12659091e-01 -9.48951960e-01 5.44790685e-01
-2.73455292e-01 1.81164011e-01 1.68092465e+00 3.20586294e-01
-1.47956923e-01 -1.12658784e-01 -1.09086156e+00 5.87556243e-01
8.37347150e-01 -5.19537628e-01 -5.27166799e-02 5.42046309e-01
7.20388353e-01 -5.52954376e-02 -1.47028756e+00 4.77673858e-01
7.56447852e-01 -5.16188264e-01 8.48176062e-01 -8.85648608e-01
4.47874367e-01 -3.47073317e-01 -3.74755681e-01 -1.31755900e+00
-1.03419709e+00 1.68763682e-01 -4.99469638e-02 9.07725275e-01
3.07484567e-01 -7.97925234e-01 6.25102937e-01 2.61978328e-01
-6.56873167e-01 -1.10825348e+00 -7.46656597e-01 -4.08095241e-01
3.93194199e-01 -4.59468141e-02 8.16664934e-01 1.12500525e+00
3.53521019e-01 3.78957629e-01 2.42652759e-01 1.39263108e-01
3.43164861e-01 3.14863652e-01 6.06359243e-01 -1.40566599e+00
-2.01081514e-01 -9.31123197e-01 -3.86689603e-01 -4.77707386e-01
4.74913150e-01 -1.59206462e+00 -4.24677223e-01 -1.43619859e+00
7.76086092e-01 -4.89628047e-01 -8.18299472e-01 5.02267182e-01
-4.00900543e-01 -7.42453486e-02 -1.72010854e-01 3.11461568e-01
-5.01539886e-01 5.88551819e-01 9.13535893e-01 -5.44642150e-01
-2.10314319e-01 -1.66176245e-01 -1.13945782e+00 4.58818972e-01
7.97677338e-01 -3.39492828e-01 5.04653566e-02 -2.58362629e-02
1.34299532e-01 -2.09187329e-01 3.99002105e-01 -6.60008132e-01
2.23678634e-01 -4.34094429e-01 9.28524911e-01 -6.09167039e-01
1.23538107e-01 -5.04408896e-01 4.64116156e-01 1.00520241e+00
-2.33673885e-01 -3.32779050e-01 3.63822490e-01 4.93853301e-01
-1.29702404e-01 6.43460974e-02 5.75775087e-01 -9.00856256e-02
-4.98130739e-01 6.93016231e-01 -5.61204851e-01 -3.60279679e-01
1.06222534e+00 -8.02982897e-02 -4.80658859e-01 2.65244991e-01
-6.74386084e-01 2.70002902e-01 3.44134510e-01 1.84092551e-01
5.66311538e-01 -1.58605039e+00 -7.96191990e-01 5.88211596e-01
5.38685620e-01 1.68061659e-01 6.03737175e-01 1.15641809e+00
-3.14439625e-01 5.13087332e-01 -1.78689629e-01 -9.38327312e-01
-1.23441172e+00 6.27724648e-01 5.35849988e-01 -1.65162697e-01
-3.70945781e-01 5.33423662e-01 8.68527532e-01 -6.58356011e-01
-2.67474234e-01 -1.16193347e-01 -2.03409687e-01 2.26239696e-01
4.89977717e-01 5.25865853e-01 1.34847090e-01 -6.11705899e-01
-5.25980890e-01 6.49367332e-01 -1.08734518e-01 5.64088404e-01
1.41729844e+00 1.38990045e-01 -5.42156458e-01 5.48473954e-01
1.49227679e+00 1.28586832e-02 -5.32135189e-01 2.46061206e-01
-1.02929547e-01 -2.32305378e-01 1.32918581e-01 -6.53176725e-01
-1.07517445e+00 6.67898059e-01 4.69444513e-01 -2.44028047e-02
1.14810693e+00 2.26164147e-01 5.54015219e-01 2.07341298e-01
-1.66194797e-01 -6.61423624e-01 -8.49018153e-03 5.97968578e-01
5.31979620e-01 -1.16610205e+00 5.12837153e-03 -2.68132001e-01
-2.30871081e-01 1.24417341e+00 7.32550442e-01 1.80705354e-01
8.05580795e-01 2.33231127e-01 -8.31554234e-02 -5.25611103e-01
-9.36286032e-01 -4.39213775e-02 4.63435799e-01 3.14579457e-01
6.92148566e-01 3.15181166e-01 -1.99989513e-01 1.21228611e+00
5.41121960e-02 -9.38030109e-02 1.61932588e-01 2.24115223e-01
-4.68225986e-01 -7.72345543e-01 -1.64207965e-01 7.34162807e-01
-4.13060039e-01 -5.99381514e-02 -5.43002248e-01 4.98349756e-01
1.44480273e-01 8.05488646e-01 3.31989050e-01 -6.62146151e-01
-4.21869233e-02 1.39667536e-03 1.01646937e-01 -6.44740403e-01
-6.17071211e-01 3.35910738e-01 -2.79751867e-01 -2.76832938e-01
1.58752482e-02 -4.79092121e-01 -1.57260323e+00 -2.70807266e-01
-5.95768273e-01 7.30240792e-02 4.01868790e-01 8.62679422e-01
9.80163991e-01 8.51575315e-01 7.89517939e-01 -2.96431869e-01
-1.07455645e-02 -8.85818958e-01 -7.49723911e-01 4.19006139e-01
1.46334589e-01 -4.21633601e-01 -3.61254960e-01 -4.66519296e-01] | [6.013308525085449, 5.692407131195068] |
390e8e26-d559-4c3b-856f-12542921daf8 | guiding-users-to-where-to-give-color-hints | 2210.1427 | null | https://arxiv.org/abs/2210.14270v1 | https://arxiv.org/pdf/2210.14270v1.pdf | Guiding Users to Where to Give Color Hints for Efficient Interactive Sketch Colorization via Unsupervised Region Prioritization | Existing deep interactive colorization models have focused on ways to utilize various types of interactions, such as point-wise color hints, scribbles, or natural-language texts, as methods to reflect a user's intent at runtime. However, another approach, which actively informs the user of the most effective regions to give hints for sketch image colorization, has been under-explored. This paper proposes a novel model-guided deep interactive colorization framework that reduces the required amount of user interactions, by prioritizing the regions in a colorization model. Our method, called GuidingPainter, prioritizes these regions where the model most needs a color hint, rather than just relying on the user's manual decision on where to give a color hint. In our extensive experiments, we show that our approach outperforms existing interactive colorization methods in terms of the conventional metrics, such as PSNR and FID, and reduces required amount of interactions. | ['Jaegul Choo', 'Daesik Kim', 'Mohammad Azam Khan', 'Haneol Lee', 'Yeojeong Park', 'Juntae Kim', 'Soyoung Yang', 'Junsoo Lee', 'Youngin Cho'] | 2022-10-25 | null | null | null | null | ['colorization'] | ['computer-vision'] | [ 3.87665220e-02 -2.60933399e-01 -4.61972617e-02 -3.07690740e-01
-3.22696894e-01 -6.92615271e-01 5.83895385e-01 2.33606860e-01
-3.65968704e-01 3.40636969e-01 9.17565301e-02 -4.65098143e-01
1.66011810e-01 -7.51879036e-01 -3.29937249e-01 -4.36330795e-01
2.85438418e-01 1.47858456e-01 3.32759619e-01 -1.72916681e-01
9.64138925e-01 5.69785774e-01 -1.20679951e+00 3.33872765e-01
9.66757774e-01 7.27077246e-01 8.49243551e-02 7.43536472e-01
-6.17264926e-01 5.86819828e-01 -5.57631969e-01 -3.02496493e-01
2.22804293e-01 -6.23822093e-01 -5.15137017e-01 -9.74528193e-02
3.43396515e-01 -8.31962168e-01 2.88092326e-02 1.05899954e+00
3.84820998e-01 1.93588600e-01 5.03960252e-01 -1.03257763e+00
-7.56552875e-01 5.61654866e-01 -9.63559270e-01 1.84758529e-02
3.98238420e-01 2.49882251e-01 8.79371762e-01 -9.79521394e-01
6.12242997e-01 1.24357629e+00 2.23972499e-01 6.76188707e-01
-1.14103007e+00 -6.88518405e-01 5.48299730e-01 1.88175544e-01
-1.34121859e+00 -2.89797008e-01 1.24833286e+00 -3.27796787e-01
2.76035815e-01 5.40532172e-01 7.32595026e-01 4.76261616e-01
-1.89098313e-01 1.00762343e+00 1.09970081e+00 -7.55588412e-01
5.79660833e-01 -3.60908061e-02 3.60299125e-02 9.51186717e-01
-1.36920810e-01 -1.88271314e-01 -6.10496461e-01 -4.42048967e-01
1.32725739e+00 1.92552179e-01 -3.09141368e-01 -6.11157000e-01
-1.02987754e+00 4.66094315e-01 3.17623556e-01 1.74463093e-01
-4.09660995e-01 4.39761579e-01 2.76359707e-01 -7.49170855e-02
4.93204981e-01 5.50808132e-01 -3.04967284e-01 -4.07561362e-01
-1.02825880e+00 2.04303041e-01 5.02630591e-01 8.37796330e-01
9.72799420e-01 -2.21015319e-01 -3.49268198e-01 8.64583313e-01
3.31615180e-01 5.36308698e-02 -1.23542443e-01 -1.18605781e+00
3.58193934e-01 9.98803854e-01 4.62367237e-01 -1.20356500e+00
2.15967029e-01 1.27596706e-01 -5.22857606e-01 7.05772936e-01
3.11005771e-01 -1.90325543e-01 -1.00011790e+00 1.45771396e+00
3.79606217e-01 -8.37298632e-02 -3.66333425e-01 1.21417546e+00
4.90133792e-01 6.85661495e-01 1.06163882e-03 1.08545475e-01
1.17678010e+00 -9.22987700e-01 -6.57827199e-01 4.53117266e-02
2.37337381e-01 -9.10673082e-01 1.73671257e+00 7.62452483e-01
-1.05367029e+00 -4.51548636e-01 -9.49862063e-01 -1.93643123e-01
-3.34662676e-01 3.21440399e-01 5.86506784e-01 5.56995928e-01
-1.08305192e+00 7.72037983e-01 -8.37836504e-01 -3.52743119e-01
2.75563717e-01 2.01167148e-02 1.17509015e-01 -2.86017880e-02
-8.00492704e-01 3.44276696e-01 -2.55543385e-02 2.47568563e-01
-7.73826957e-01 -6.13938630e-01 -3.59616548e-01 2.57248968e-01
4.84731674e-01 -3.93701702e-01 1.22196400e+00 -1.14141691e+00
-1.85191965e+00 4.42329526e-01 -1.13359585e-01 1.51713520e-01
8.54003072e-01 -5.78171849e-01 1.92645475e-01 3.08304101e-01
-4.49590772e-01 8.68124604e-01 8.37784648e-01 -1.74723852e+00
-6.51080906e-01 -2.16865540e-01 7.21987188e-01 3.72779727e-01
-4.61997539e-01 4.31990959e-02 -1.25614345e+00 -7.19643891e-01
3.16606313e-02 -8.91480684e-01 -4.96967852e-01 7.10985720e-01
-7.77114868e-01 -1.29962891e-01 1.13543570e+00 -5.22981882e-01
1.51980698e+00 -2.12798452e+00 2.96371896e-02 5.01722813e-01
3.06562513e-01 2.35679582e-01 7.18424888e-03 6.07133389e-01
2.41170496e-01 4.06447321e-01 -1.64854318e-01 -3.28774929e-01
1.60638794e-01 -3.47563401e-02 -1.22038506e-01 3.31082428e-03
-1.98447570e-01 3.67559880e-01 -8.62454057e-01 -5.41025400e-01
4.96584058e-01 5.03620625e-01 -6.53822362e-01 4.59228426e-01
-3.55408162e-01 1.59015983e-01 -4.75573838e-01 5.37036002e-01
9.89626288e-01 -5.07788025e-02 1.17619611e-01 -2.94631183e-01
-3.62024903e-01 2.41535291e-01 -1.17868316e+00 1.72045040e+00
-4.91807640e-01 5.89736938e-01 1.18452765e-01 -2.51536548e-01
8.66897821e-01 -7.62817413e-02 2.57772326e-01 -5.34835994e-01
-4.91340682e-02 -1.91244453e-01 -4.01451379e-01 -1.73902124e-01
5.96081257e-01 6.26631618e-01 1.27786055e-01 9.13102031e-01
-8.92617106e-01 1.95488200e-01 2.42090896e-01 6.36460304e-01
1.00779331e+00 5.58277190e-01 5.78620434e-02 -2.29943797e-01
2.72876948e-01 -1.93473265e-01 2.38812476e-01 7.50012100e-01
1.79148130e-02 6.15431190e-01 8.69262874e-01 -4.80070919e-01
-9.49692011e-01 -8.01359236e-01 5.24324477e-01 1.38809478e+00
5.40489376e-01 -7.30208874e-01 -1.04431164e+00 -8.34468722e-01
-3.26451391e-01 6.56307459e-01 -7.34892666e-01 1.54804304e-01
-7.04852283e-01 -1.06256619e-01 2.56468207e-02 3.96528542e-01
5.69479108e-01 -1.24122047e+00 -1.13594449e+00 1.24837076e-02
1.07501850e-01 -3.80100280e-01 -9.91935849e-01 -2.41705522e-01
-9.71190035e-01 -1.03199017e+00 -1.08649266e+00 -4.30115193e-01
1.26849866e+00 5.58583081e-01 8.60678911e-01 6.82104826e-01
-2.08368048e-01 2.88958192e-01 -5.84340632e-01 5.02714217e-02
-1.10384405e-01 -9.33171138e-02 -5.31111300e-01 2.70161405e-02
-1.08001426e-01 -5.14324486e-01 -1.26494527e+00 3.99572700e-01
-9.74401176e-01 6.81079865e-01 5.60217023e-01 5.37901282e-01
5.43646634e-01 -3.05980206e-01 -7.03564659e-02 -1.31617844e+00
9.66235280e-01 2.17815742e-01 -5.35454988e-01 3.54338884e-01
-5.80780506e-01 4.19581741e-01 7.53500402e-01 -4.57124650e-01
-1.09066176e+00 6.21807054e-02 9.16682854e-02 -3.44475716e-01
-6.40565157e-02 3.52467358e-01 -5.86670265e-02 -2.02916026e-01
5.06539881e-01 9.15550143e-02 -5.35325110e-01 -5.74798882e-01
7.88169444e-01 3.86938661e-01 3.51598501e-01 -7.73789287e-01
8.50336373e-01 2.16229036e-01 -4.78392154e-01 -3.99154574e-01
-1.69724643e-01 -2.16514796e-01 -5.30277491e-01 -4.72840577e-01
5.44844866e-01 -2.05610245e-01 -8.35462511e-01 1.93358824e-01
-1.32135725e+00 -4.30379242e-01 2.58127272e-01 -1.25015214e-01
-1.80869728e-01 5.91277480e-01 -5.17610669e-01 -1.11187184e+00
-5.22179008e-01 -1.13685751e+00 9.39753711e-01 4.48850274e-01
-2.25388035e-01 -6.20898664e-01 -1.36643231e-01 -1.67100802e-01
4.91440654e-01 3.72603774e-01 1.21615613e+00 1.36214844e-03
-1.03429675e+00 -2.12252945e-01 -4.83144283e-01 -1.28929600e-01
4.09217566e-01 4.55848277e-01 -7.14655161e-01 -8.45183209e-02
-7.36585498e-01 -3.03939153e-02 5.46023309e-01 -1.42615885e-02
1.69667292e+00 -5.30992925e-01 -3.71848851e-01 5.20264268e-01
1.43464220e+00 6.75620079e-01 8.23306262e-01 3.36976945e-01
7.41095304e-01 2.82574177e-01 5.66494644e-01 6.97664440e-01
1.85694262e-01 5.98681152e-01 4.95143056e-01 -6.85367584e-01
1.16378919e-03 -3.64832789e-01 -8.00847560e-02 3.56103510e-01
-2.49614820e-01 -2.83477575e-01 -5.40137053e-01 1.64667502e-01
-1.93537569e+00 -5.23483694e-01 2.71885961e-01 2.24846768e+00
7.62552321e-01 3.46148223e-01 1.52257994e-01 1.29165158e-01
7.24051654e-01 2.43639663e-01 -7.21519828e-01 -5.59669852e-01
5.74894965e-01 7.25508109e-02 1.67864233e-01 4.97775733e-01
-6.98658049e-01 1.11606419e+00 6.10344315e+00 5.37397623e-01
-1.23190641e+00 -2.49130353e-01 8.53540897e-01 1.54586554e-01
-5.84098101e-01 3.77112001e-01 -2.24209875e-01 4.79547262e-01
3.58508565e-02 1.48031160e-01 8.04115891e-01 9.14073825e-01
5.80953360e-01 -5.34281194e-01 -1.09477520e+00 9.56248105e-01
-1.53162088e-02 -1.21070457e+00 2.58883238e-01 -1.84236765e-01
3.78674954e-01 -9.03010130e-01 4.89243045e-02 -3.14330915e-04
4.05887097e-01 -3.61559033e-01 7.38141596e-01 6.29715025e-01
7.69266427e-01 -8.98004174e-01 -1.15123624e-02 4.75626513e-02
-1.18108761e+00 3.93439561e-01 -2.92519219e-02 4.38960753e-02
4.47593220e-02 2.26125926e-01 -8.43109965e-01 1.60904050e-01
6.36947691e-01 1.12313062e-01 -6.64177656e-01 1.07765198e+00
-3.56060326e-01 6.34693205e-01 -1.10237710e-01 -3.54474634e-01
1.61103144e-01 -2.54760951e-01 9.23564658e-02 1.24207687e+00
2.40199342e-01 4.06478524e-01 1.85720056e-01 9.52185869e-01
1.50896251e-01 2.96065867e-01 -1.53081790e-01 -9.68894958e-02
5.48676074e-01 1.31699252e+00 -1.11071002e+00 -4.29336756e-01
-2.07616374e-01 1.47463930e+00 2.40144357e-01 6.83794439e-01
-9.18066740e-01 -8.89903247e-01 3.93171310e-01 4.00619775e-01
1.15632899e-01 -6.03864789e-01 -4.14301723e-01 -7.00390339e-01
-1.11687273e-01 -6.39600396e-01 7.36703575e-02 -1.24777210e+00
-5.84426403e-01 4.61719811e-01 -6.78581819e-02 -1.12663925e+00
1.09977491e-01 -3.65817994e-01 -1.07112002e+00 8.71809483e-01
-1.19983220e+00 -8.61257553e-01 -5.17262042e-01 4.91939187e-01
6.31814718e-01 2.70298779e-01 5.06408513e-01 2.26431206e-01
-5.12148976e-01 6.72608256e-01 -2.43600950e-01 6.91151023e-02
8.23736131e-01 -1.42918432e+00 5.69308341e-01 8.87350023e-01
-3.76376659e-02 9.46344614e-01 8.73313069e-01 -4.49870676e-01
-1.46542597e+00 -4.04225051e-01 4.33000356e-01 2.09940836e-01
7.92525485e-02 -3.55992436e-01 -9.05602932e-01 1.15720913e-01
4.17201936e-01 -2.00120732e-01 3.98327082e-01 8.61434042e-02
-2.81834126e-01 -2.85356194e-01 -1.02720118e+00 1.23515069e+00
9.57702816e-01 -3.37324589e-01 2.32938603e-02 5.28163202e-02
4.59037811e-01 -3.12101811e-01 -5.40746972e-02 6.26695855e-03
7.54639804e-01 -1.15488029e+00 9.23901260e-01 -2.71742553e-01
4.27856326e-01 -6.61683023e-01 2.64512628e-01 -1.17009389e+00
-5.20424843e-01 -8.57612789e-01 3.07204109e-02 1.34474027e+00
3.72316629e-01 -4.40728664e-02 1.01499498e+00 1.17792189e+00
-1.60932302e-01 -8.44151139e-01 -2.92237431e-01 -5.26849320e-03
-5.79547524e-01 -1.07315809e-01 6.23385549e-01 7.91724622e-01
1.92851620e-03 -2.06737131e-01 -4.78506148e-01 -4.86314371e-02
3.51257473e-01 2.08142698e-01 1.15166318e+00 -8.63933265e-01
-1.99479014e-01 -7.39240468e-01 1.56215146e-01 -1.26687574e+00
-5.51905930e-01 -2.68885106e-01 1.68106601e-01 -1.85780442e+00
2.24039227e-01 -6.13109350e-01 -4.01018769e-01 6.97301805e-01
-3.95772994e-01 1.28136635e-01 4.47418928e-01 1.26776725e-01
-8.01250458e-01 2.88344026e-01 1.37859333e+00 -8.29014108e-02
-5.86043894e-01 -2.75291234e-01 -8.76560569e-01 8.31110418e-01
6.12225413e-01 -2.10420117e-01 -5.23019314e-01 -5.53798556e-01
2.67887533e-01 2.19670251e-01 1.79893374e-01 -9.75728035e-01
3.79639030e-01 -7.74845123e-01 5.53738773e-01 -7.93628931e-01
2.17990011e-01 -7.02250004e-01 1.73710566e-02 3.71921152e-01
-6.88552320e-01 4.96230960e-01 1.74020231e-01 3.70026857e-01
2.21294522e-01 -1.06047310e-01 6.64901972e-01 -2.98038065e-01
-7.24585295e-01 2.61845112e-01 -4.68795091e-01 -4.39286411e-01
8.50985706e-01 -3.94022226e-01 -8.63627046e-02 -5.41127503e-01
-5.25366962e-01 1.22198038e-01 6.40895426e-01 2.13392675e-01
8.45817268e-01 -1.35500741e+00 -1.38907894e-01 2.18381435e-02
4.26367158e-04 -3.03547174e-01 2.21818551e-01 1.76955268e-01
-9.72055733e-01 -3.61687839e-02 -1.02759287e-01 -3.09510201e-01
-1.37289417e+00 5.85956037e-01 9.78566930e-02 -7.04121962e-02
-6.54473662e-01 7.45536864e-01 3.55433792e-01 1.38149694e-01
5.75421035e-01 -5.42118847e-01 -2.72402894e-02 -2.43194014e-01
6.13474429e-01 3.15729767e-01 -1.78930297e-01 1.24801010e-01
-2.92168230e-01 6.92077458e-01 -3.06971312e-01 -3.95577967e-01
8.69266391e-01 -2.32043594e-01 -6.03558868e-02 2.40015134e-01
7.37726033e-01 2.15082645e-01 -1.68480027e+00 -2.87539922e-02
-1.55960038e-01 -9.84727025e-01 -4.30245735e-02 -1.24136496e+00
-1.10344183e+00 9.42662716e-01 7.26232767e-01 2.45127395e-01
1.39017129e+00 -4.45075542e-01 7.80693233e-01 3.85298312e-01
4.37162220e-01 -1.19201541e+00 4.36155617e-01 1.67893320e-01
9.95071173e-01 -8.53351414e-01 6.08494170e-02 -3.97139817e-01
-6.24874592e-01 1.22606039e+00 1.12258101e+00 -2.17690691e-01
4.66069728e-01 2.59268939e-01 2.58350909e-01 -6.61964789e-02
-5.59036434e-01 1.96690500e-01 1.87052324e-01 3.64557564e-01
7.06074715e-01 1.02004498e-01 -5.28652847e-01 1.81803420e-01
1.57053858e-01 3.08487900e-02 2.95185804e-01 1.05707395e+00
-6.14998698e-01 -1.33136284e+00 -5.09295285e-01 2.31752157e-01
-1.32629216e-01 -1.31843910e-01 -7.86151230e-01 5.15663981e-01
-6.02826886e-02 5.81609964e-01 -8.81770626e-02 -3.70282829e-01
3.38146359e-01 -1.65327862e-01 2.79239804e-01 -3.98965746e-01
-4.53752458e-01 2.37669528e-01 -2.05228373e-01 -6.01035297e-01
-8.71537402e-02 -1.27375945e-01 -1.24880505e+00 -5.06186485e-01
-2.02866286e-01 1.08418301e-01 5.91537595e-01 6.48638189e-01
5.05334139e-01 3.71659607e-01 6.60489678e-01 -1.16798615e+00
1.25561208e-01 -7.89794445e-01 -1.00902081e-01 3.61434788e-01
1.56271502e-01 -4.91331041e-01 6.03918433e-02 -3.20390761e-02] | [11.405937194824219, -1.0073308944702148] |
93405ef5-2ed6-4874-9d11-359d934523c0 | detection-of-gravitational-waves-using | 1910.08245 | null | https://arxiv.org/abs/1910.08245v1 | https://arxiv.org/pdf/1910.08245v1.pdf | Detection of gravitational waves using topological data analysis and convolutional neural network: An improved approach | The gravitational wave detection problem is challenging because the noise is typically overwhelming. Convolutional neural networks (CNNs) have been successfully applied, but require a large training set and the accuracy suffers significantly in the case of low SNR. We propose an improved method that employs a feature extraction step using persistent homology. The resulting method is more resilient to noise, more capable of detecting signals with varied signatures and requires less training. This is a powerful improvement as the detection problem can be computationally intense and is concerned with a relatively large class of wave signatures. | ['Jae-Hun Jung', 'Christopher Bresten'] | 2019-10-18 | null | null | null | null | ['gravitational-wave-detection'] | ['miscellaneous'] | [ 4.04398292e-01 -2.96440274e-01 4.64829892e-01 5.32035753e-02
-7.96505988e-01 -6.01231158e-01 6.24637306e-01 -1.43068284e-01
-6.82075202e-01 6.91512167e-01 -1.82073534e-01 -1.23190694e-01
-2.12372899e-01 -1.05273843e+00 -4.88357663e-01 -1.12025297e+00
-3.95683229e-01 3.11315298e-01 8.69344056e-01 -5.88550091e-01
2.77945250e-01 7.14457393e-01 -1.76115263e+00 -1.49713680e-02
3.47032249e-01 1.13320923e+00 2.62429789e-02 7.91538060e-01
2.50918537e-01 4.69107509e-01 -5.22529185e-01 -1.52621552e-01
4.48212862e-01 -5.65139294e-01 -4.94354606e-01 -4.78827477e-01
4.83709425e-01 2.13661760e-01 -4.49622899e-01 1.18465483e+00
9.79649723e-01 9.06857625e-02 3.62336934e-01 -8.11649024e-01
6.22087717e-02 2.09171996e-01 -2.99334913e-01 4.86579061e-01
4.01861757e-01 9.45624709e-03 7.21821487e-01 -8.26085150e-01
5.31631112e-01 7.15238988e-01 1.07782626e+00 2.79505312e-01
-1.01357126e+00 -5.69095314e-01 -7.86319256e-01 4.52661783e-01
-1.34464598e+00 -3.49798828e-01 8.76002252e-01 -3.64508122e-01
8.45582426e-01 4.51246083e-01 6.06015921e-01 9.13340569e-01
4.37906496e-02 1.66205317e-01 1.15142846e+00 -4.47821230e-01
2.46997267e-01 -2.12462097e-01 -6.30577654e-02 7.72619784e-01
4.04049128e-01 4.30587709e-01 -4.13543969e-01 -1.71617731e-01
5.94750762e-01 -1.93812817e-01 -6.36833489e-01 -9.59364027e-02
-8.86018038e-01 1.00939953e+00 3.60696167e-01 7.30416596e-01
-1.67913973e-01 1.87003553e-01 4.24144626e-01 7.36589015e-01
3.49925794e-02 7.02063680e-01 -4.66979355e-01 -2.05805853e-01
-1.13941133e+00 5.42674422e-01 9.96709466e-01 3.71423364e-01
7.66272426e-01 2.52682596e-01 2.85611004e-01 5.79970241e-01
-6.40222579e-02 2.84399718e-01 5.83518147e-01 -6.14286482e-01
5.65944202e-02 4.61689681e-01 -1.72661468e-01 -1.18006849e+00
-9.25758362e-01 -7.76803553e-01 -9.05562460e-01 7.57705390e-01
4.85813349e-01 -3.17398697e-01 -8.15674543e-01 1.50242233e+00
3.89754698e-02 1.38976038e-01 1.71398088e-01 9.66091752e-01
1.09218884e+00 3.92761648e-01 -3.49228948e-01 -7.83351213e-02
1.21200168e+00 -3.48102301e-01 -4.37971830e-01 -2.51902252e-01
4.28826511e-01 -7.65157223e-01 4.76915389e-01 3.83315027e-01
-8.66438866e-01 -3.31030846e-01 -1.26723707e+00 2.31205299e-01
-6.26757979e-01 -3.94540578e-02 8.33676279e-01 7.57064700e-01
-1.04980564e+00 8.47318649e-01 -5.78909159e-01 -3.60342175e-01
3.50661218e-01 5.87683260e-01 -4.75922376e-01 1.23713359e-01
-1.28095126e+00 9.78040636e-01 5.14831066e-01 -4.06639986e-02
-5.67151189e-01 -3.12042266e-01 -7.14342713e-01 1.51591375e-01
1.88452736e-01 -3.07227641e-01 7.75239646e-01 -7.46592939e-01
-1.62400043e+00 5.81232846e-01 4.07029808e-01 -4.40484673e-01
6.96389735e-01 1.24846071e-01 -3.83028507e-01 2.00230032e-01
-2.74773061e-01 -9.30767953e-02 9.31013405e-01 -7.20964968e-01
-2.80109614e-01 -2.15167597e-01 -4.15179841e-02 -2.48106092e-01
8.76905620e-02 3.36936295e-01 4.73518204e-03 -6.45151138e-01
5.98892212e-01 -1.04129243e+00 -1.72857255e-01 -2.82904446e-01
1.26702100e-01 -4.73987162e-02 9.14549947e-01 -3.22186261e-01
5.40439248e-01 -1.99654961e+00 1.35044739e-01 2.60028273e-01
1.36382654e-01 5.07830560e-01 1.08623549e-01 5.75530291e-01
-1.37689307e-01 -2.59199291e-02 -3.59187484e-01 3.11237574e-01
-2.42587656e-01 7.29339793e-02 -6.78468719e-02 6.81100309e-01
1.72001287e-01 4.87162262e-01 -6.13939643e-01 -2.25713819e-01
-4.10628617e-02 4.29088265e-01 -5.08449793e-01 1.88443348e-01
1.80831045e-01 4.76174027e-01 -2.23399803e-01 7.38369286e-01
6.02120697e-01 1.57115813e-02 -2.58890629e-01 -3.14540625e-01
-5.07227957e-01 -2.41050068e-02 -1.34028220e+00 1.31638551e+00
-1.89370081e-01 1.11280918e+00 1.49107575e-01 -1.45763838e+00
1.14041758e+00 3.63965511e-01 3.24235588e-01 -4.42199230e-01
5.00029147e-01 3.47030193e-01 3.49984795e-01 -9.83445048e-01
2.26790398e-01 -5.66486299e-01 1.07914910e-01 1.01894937e-01
2.61639565e-01 -1.73721835e-01 2.58490834e-02 -1.10959619e-01
1.54109430e+00 -4.04625088e-02 1.05284296e-01 -2.08855510e-01
3.75948071e-01 -8.18243548e-02 7.59993136e-01 1.00104463e+00
-1.34192020e-01 6.61365688e-01 2.72808015e-01 -5.97860754e-01
-7.40304708e-01 -5.08932054e-01 -3.08436543e-01 7.70939589e-01
1.57391667e-01 -1.69129863e-01 -3.13040763e-01 -2.09622845e-01
-2.24634781e-01 3.79564427e-02 -2.75282681e-01 -2.34006166e-01
-8.16802740e-01 -1.22628474e+00 7.55644798e-01 3.63684773e-01
5.70034623e-01 -9.36312735e-01 -8.34167838e-01 3.53401959e-01
5.68489879e-02 -1.16559935e+00 5.40550351e-01 4.07178819e-01
-7.76014924e-01 -1.21242797e+00 -6.01174176e-01 -8.72670174e-01
2.83928365e-01 -9.50035900e-02 8.88564646e-01 3.52432132e-01
-7.12369561e-01 2.40334123e-03 -6.67448223e-01 -6.44903362e-01
-1.35640278e-01 -2.69375801e-01 1.62285760e-01 -1.99664407e-03
2.75035292e-01 -8.95123959e-01 -3.30930591e-01 1.06661655e-01
-7.30178714e-01 -4.93873537e-01 8.07471931e-01 1.21409202e+00
5.24345525e-02 2.75191158e-01 6.53093874e-01 -6.45824492e-01
3.94189864e-01 -2.74640620e-01 -7.75841236e-01 -2.20740303e-01
-3.45399499e-01 -5.47248721e-02 7.26003110e-01 -3.24186802e-01
-6.78926647e-01 2.21124008e-01 -3.96275789e-01 7.98126385e-02
-3.07825416e-01 6.07911527e-01 2.22779587e-01 -1.20500696e+00
1.03699028e+00 1.68789297e-01 -1.66745529e-01 -7.40738034e-01
-1.67911142e-01 3.50730866e-01 6.87406898e-01 -1.30793944e-01
1.10702503e+00 3.91324043e-01 6.20627224e-01 -1.20413291e+00
-4.13065076e-01 -4.61887151e-01 -6.52453780e-01 -3.74681115e-01
3.40732127e-01 -6.16991222e-01 -5.80876529e-01 5.93945503e-01
-1.02238441e+00 1.18533283e-01 1.97076406e-02 9.16089118e-01
-2.99821436e-01 5.74351847e-01 -4.52699631e-01 -7.33698010e-01
-2.36533105e-01 -9.07998264e-01 4.75727618e-01 3.08736533e-01
3.49893987e-01 -6.42954111e-01 2.59573311e-01 -1.47162557e-01
8.38566363e-01 7.27175295e-01 5.70015192e-01 -7.05161572e-01
-5.49002111e-01 -8.12246501e-01 -3.04823369e-01 3.56739722e-02
-2.58895665e-01 -2.95932889e-02 -1.10969913e+00 -3.92506689e-01
9.75455046e-02 -4.42995042e-01 1.24833250e+00 2.25834876e-01
8.29540431e-01 -3.99797000e-02 -2.17772424e-01 7.92191923e-01
1.67892909e+00 -5.34944870e-02 6.72100127e-01 5.38763642e-01
3.57925624e-01 5.33626974e-01 2.40984753e-01 2.61247903e-01
-3.97284240e-01 7.89336860e-01 5.60466528e-01 -3.48227769e-01
-1.65249288e-01 6.27816975e-01 1.68698788e-01 7.12490559e-01
-6.23833537e-01 8.96228552e-02 -8.99367332e-01 4.74066406e-01
-1.78505647e+00 -1.35616362e+00 -5.14705539e-01 1.89360261e+00
6.15829706e-01 1.32203877e-01 9.75436196e-02 6.40675962e-01
3.61031055e-01 -1.18289031e-01 -1.21651791e-01 -1.07783675e-01
-3.92556012e-01 5.94820380e-01 6.81984723e-01 3.35515290e-01
-1.20116937e+00 6.12448335e-01 7.62954473e+00 5.09849489e-01
-1.37519169e+00 8.54532346e-02 -2.99978554e-01 4.79554087e-02
3.97350490e-01 -2.25413702e-02 -4.27123040e-01 3.92203897e-01
7.73057580e-01 1.24904774e-01 -7.64075816e-02 7.86563575e-01
-1.00083947e-01 -2.09350199e-01 -6.26039684e-01 1.16173005e+00
1.20981112e-02 -1.39282393e+00 -4.03688133e-01 -2.28920475e-01
2.95498043e-01 4.98922825e-01 -2.27933273e-01 4.98302206e-02
-1.63335368e-01 -9.01312172e-01 4.26773489e-01 5.63076079e-01
6.19562149e-01 -7.12580681e-01 1.23523080e+00 3.12167436e-01
-9.43943679e-01 -9.88973007e-02 -8.55160475e-01 -4.76086348e-01
4.66414839e-02 7.45953619e-01 -6.06336951e-01 4.91480172e-01
9.39107656e-01 2.36390695e-01 -4.67324436e-01 1.75736427e+00
-3.60971130e-02 7.63703644e-01 -6.12253129e-01 -2.35661536e-01
2.23312989e-01 -1.24476433e-01 8.87014627e-01 1.51941884e+00
6.18154705e-01 2.96547621e-01 2.80898929e-01 4.80828762e-01
1.03593081e-01 9.12934542e-02 -7.91126609e-01 2.72490084e-01
3.20333280e-02 1.47008181e+00 -7.56299198e-01 -1.18663952e-01
-4.09454525e-01 6.86472058e-01 5.91676049e-02 1.34046316e-01
-3.93197894e-01 -9.34281766e-01 4.04366910e-01 -6.26741871e-02
4.66497570e-01 -4.05647546e-01 4.72148098e-02 -1.07865441e+00
-2.16195896e-01 -5.27506948e-01 4.16856855e-01 -4.39140558e-01
-1.05624700e+00 8.43952477e-01 -3.74219149e-01 -1.24611497e+00
-1.36821359e-01 -1.07458806e+00 -8.54617119e-01 7.86723554e-01
-1.74322283e+00 -8.90750766e-01 -5.65339446e-01 5.63925147e-01
1.21476188e-01 -3.98142189e-01 9.19991851e-01 5.69691360e-01
-3.78483832e-01 5.04240036e-01 1.48210615e-01 2.47950241e-01
4.48202997e-01 -1.19382167e+00 -6.64155558e-02 1.00655603e+00
-8.88025854e-03 2.04763517e-01 1.10009432e+00 -3.71690869e-01
-1.59980202e+00 -8.06400537e-01 7.60126472e-01 -2.39418045e-01
7.26839244e-01 -3.43599558e-01 -8.97472680e-01 1.47631660e-01
-1.18171729e-01 3.33856463e-01 7.40046799e-01 -2.14597099e-02
-2.25140810e-01 1.48230210e-01 -1.27119422e+00 1.26957104e-01
9.62722480e-01 -4.26093072e-01 -7.24614024e-01 4.54082817e-01
2.36188427e-01 -4.05246526e-01 -6.08337939e-01 5.95354140e-01
5.81948578e-01 -1.03017008e+00 9.36135709e-01 -2.25047320e-01
-3.02708507e-01 -5.25605738e-01 6.13766871e-02 -1.04916525e+00
-5.71935713e-01 -6.73173368e-01 1.00785151e-01 6.62921071e-01
3.43675733e-01 -6.57505393e-01 7.36085892e-01 4.90870439e-02
-8.20767432e-02 -3.76322120e-01 -1.27918231e+00 -1.13985169e+00
-3.72175634e-01 -2.64184415e-01 1.69086561e-01 9.48791444e-01
2.26112399e-02 2.11186826e-01 -5.72384000e-01 3.80956024e-01
5.92749238e-01 3.23232114e-01 4.78064328e-01 -1.82396865e+00
-4.58781630e-01 -3.33673716e-01 -1.11408126e+00 -3.49216342e-01
-2.24332839e-01 -8.02961648e-01 2.97196805e-01 -1.02777815e+00
-1.85684532e-01 -2.61218369e-01 -2.09924385e-01 5.78116715e-01
1.58797249e-01 9.29363728e-01 -2.16959104e-01 2.18226582e-01
-2.48599440e-01 3.17680746e-01 7.42775440e-01 1.39377238e-02
-1.08483844e-02 9.65323597e-02 -2.09072843e-01 9.49254334e-01
8.86469841e-01 -6.58752918e-01 3.06360751e-01 3.76401208e-02
5.41349947e-01 -3.15815777e-01 5.97885668e-01 -1.48450100e+00
4.74743485e-01 2.41178572e-01 4.84143883e-01 -4.30818051e-01
3.85971874e-01 -5.52858591e-01 3.34822446e-01 7.85711527e-01
1.41901091e-01 4.01708037e-02 8.40581506e-02 5.26326299e-01
-3.23025078e-01 -5.84613681e-01 1.09943128e+00 -3.94386142e-01
-7.92100728e-01 7.21790269e-02 -4.54820961e-01 -2.83298433e-01
6.40264928e-01 -2.59628654e-01 -9.57722962e-02 -3.51109713e-01
-7.18433857e-01 -3.53879094e-01 3.20842087e-01 -8.21206644e-02
5.21785259e-01 -1.10221112e+00 -8.52050900e-01 2.53668487e-01
-2.95712613e-02 -2.71553665e-01 -1.25333443e-01 1.05140352e+00
-8.29503834e-01 4.38484782e-03 -3.06123108e-01 -4.42471504e-01
-1.47059798e+00 2.95639515e-01 4.90031600e-01 1.60021722e-01
-1.05808425e+00 1.20835221e+00 -5.07215738e-01 -1.01075515e-01
6.29526600e-02 1.70561448e-01 -6.28502786e-01 7.69298300e-02
6.63086236e-01 5.28179348e-01 3.85922551e-01 -8.39729071e-01
-3.20318311e-01 9.19063330e-01 2.61824876e-01 1.95228547e-01
1.81060719e+00 4.30746436e-01 -4.25806224e-01 -7.57546872e-02
1.16469944e+00 1.67254508e-02 -8.45557511e-01 -2.91131049e-01
1.56741083e-01 -5.55740833e-01 4.87757564e-01 -6.20416880e-01
-1.06350231e+00 7.77399540e-01 8.99987042e-01 5.67584157e-01
1.13754427e+00 -1.97434779e-02 5.71714282e-01 6.45239472e-01
4.16420817e-01 -1.07012677e+00 -2.69938167e-02 8.66954565e-01
9.30090547e-01 -1.09693480e+00 9.52951089e-02 -2.67387271e-01
2.43805632e-01 1.62580180e+00 2.50494480e-01 -3.77971023e-01
7.35265493e-01 3.95972282e-01 2.60136323e-03 -7.87917793e-01
-1.91916287e-01 -6.57916427e-01 2.73163289e-01 6.20007575e-01
2.42191553e-01 -1.72261864e-01 -7.28682220e-01 4.69867080e-01
-3.53071690e-01 -2.72450894e-01 6.85574055e-01 1.04880428e+00
-7.94931233e-01 -1.01759911e+00 -5.15290201e-01 2.99392223e-01
-7.93773532e-01 -2.16445830e-02 -4.49810773e-01 5.10642886e-01
3.72249633e-01 7.44292974e-01 -4.24775094e-01 -4.91742641e-01
1.77693650e-01 1.41199976e-01 4.72568482e-01 -3.29641432e-01
-4.37535077e-01 1.31062893e-02 1.24644510e-01 -6.14516735e-01
-7.67407119e-01 -6.11401618e-01 -1.03848279e+00 -1.29298344e-01
-7.77275026e-01 1.74432591e-01 8.12340021e-01 1.00622141e+00
1.66925952e-01 4.62868482e-01 6.58425927e-01 -1.14110756e+00
-3.04068357e-01 -1.02797985e+00 -6.95963860e-01 3.02435786e-01
5.33222437e-01 -1.00488305e+00 -7.06628978e-01 -2.18370959e-01] | [7.567573070526123, 3.0552990436553955] |
afab361d-930a-4faf-ab6e-06529918ad31 | generative-sequential-recommendation-with | 2306.11114 | null | https://arxiv.org/abs/2306.11114v1 | https://arxiv.org/pdf/2306.11114v1.pdf | Generative Sequential Recommendation with GPTRec | Sequential recommendation is an important recommendation task that aims to predict the next item in a sequence. Recently, adaptations of language models, particularly Transformer-based models such as SASRec and BERT4Rec, have achieved state-of-the-art results in sequential recommendation. In these models, item ids replace tokens in the original language models. However, this approach has limitations. First, the vocabulary of item ids may be many times larger than in language models. Second, the classical Top-K recommendation approach used by these models may not be optimal for complex recommendation objectives, including auxiliary objectives such as diversity, coverage or coherence. Recent progress in generative language models inspires us to revisit generative approaches to address these challenges. This paper presents the GPTRec sequential recommendation model, which is based on the GPT-2 architecture. GPTRec can address large vocabulary issues by splitting item ids into sub-id tokens using a novel SVD Tokenisation algorithm based on quantised item embeddings from an SVD decomposition of the user-item interaction matrix. The paper also presents a novel Next-K recommendation strategy, which generates recommendations item-by-item, considering already recommended items. The Next-K strategy can be used for producing complex interdependent recommendation lists. We experiment with GPTRec on the MovieLens-1M dataset and show that using sub-item tokenisation GPTRec can match the quality of SASRec while reducing the embedding table by 40%. We also show that the recommendations generated by GPTRec on MovieLens-1M using the Next-K recommendation strategy match the quality of SASRec in terms of NDCG@10, meaning that the model can serve as a strong starting point for future research. | ['Craig Macdonald', 'Aleksandr V. Petrov'] | 2023-06-19 | null | null | null | null | ['sequential-recommendation'] | ['miscellaneous'] | [-1.16231576e-01 -2.75021911e-01 -4.25425470e-01 -1.84715867e-01
-5.79836309e-01 -7.15767801e-01 7.37164915e-01 -3.75179350e-02
-4.22668993e-01 4.35525179e-01 8.37949395e-01 -4.60792273e-01
-5.38022399e-01 -9.32871759e-01 -6.36484623e-01 -5.13390362e-01
-1.92720041e-01 7.15280354e-01 1.01476490e-01 -6.23187184e-01
4.32158589e-01 1.69835821e-01 -1.83619320e+00 7.88603842e-01
5.38464189e-01 6.08859122e-01 4.45769638e-01 7.19711840e-01
-4.53679055e-01 4.32619989e-01 -2.28702232e-01 -6.07884586e-01
3.70539784e-01 -5.99631846e-01 -5.33509851e-01 -9.03681442e-02
2.77197510e-01 -2.55016714e-01 -1.43316969e-01 5.43376148e-01
4.72060621e-01 5.36852062e-01 9.98593569e-01 -9.23627138e-01
-1.13573122e+00 1.38989854e+00 -1.90300941e-02 6.38971254e-02
3.27858716e-01 -5.32344937e-01 1.80295181e+00 -1.28803444e+00
5.54793656e-01 1.13544846e+00 5.90858042e-01 4.77589875e-01
-1.47723079e+00 -2.60552257e-01 4.92755413e-01 2.22137809e-01
-1.39222944e+00 -9.98085514e-02 2.69214779e-01 -3.02896559e-01
1.46510231e+00 5.45675159e-01 6.80233240e-01 1.10883951e+00
2.45070651e-01 7.09140182e-01 7.16415226e-01 -5.15298784e-01
1.80066705e-01 4.65346366e-01 3.14558238e-01 1.98976934e-01
4.43798095e-01 1.21532351e-01 -5.86514831e-01 -3.53213429e-01
8.28086436e-01 2.44517997e-01 -3.28676440e-02 -2.24367335e-01
-1.08319104e+00 1.35731101e+00 -3.15120518e-02 4.72888380e-01
-5.73891461e-01 9.26262960e-02 2.72267342e-01 7.14492798e-01
7.16247320e-01 6.82509124e-01 -5.28534412e-01 -1.48043975e-01
-1.12737560e+00 5.49865305e-01 9.64209259e-01 1.05438864e+00
2.86195517e-01 5.85647598e-02 -4.03477162e-01 1.10017240e+00
6.81555152e-01 4.04790461e-01 6.68930829e-01 -6.11162722e-01
2.99229860e-01 1.66507989e-01 1.01445086e-01 -9.11250472e-01
-1.92972645e-01 -7.88631022e-01 -6.23906314e-01 -3.23563993e-01
-4.95079234e-02 4.55274470e-02 -6.66228771e-01 1.43626404e+00
-4.05592466e-04 4.89875898e-02 4.27409485e-02 8.16085815e-01
8.64173353e-01 8.62374306e-01 -1.38851121e-01 -3.47226053e-01
1.22636461e+00 -1.03187931e+00 -6.68959796e-01 3.09259482e-02
8.26166630e-01 -8.47922146e-01 9.97361541e-01 6.16190135e-01
-9.68103588e-01 -7.36981273e-01 -8.26189399e-01 2.08193064e-01
-3.98816556e-01 1.47150591e-01 6.22034907e-01 9.48250711e-01
-1.23573053e+00 6.86817110e-01 -3.87693018e-01 -5.27618647e-01
-2.02903047e-01 4.36707288e-01 -3.29888016e-02 -1.96730122e-01
-1.43636656e+00 6.80632770e-01 2.56940067e-01 -3.41811329e-01
-4.53377545e-01 -7.04740942e-01 -5.83714724e-01 2.55560696e-01
3.47680777e-01 -8.52823436e-01 9.97281492e-01 -5.98647177e-01
-1.58804393e+00 9.93384346e-02 -6.79614544e-02 -5.28484404e-01
-1.52353585e-01 -3.09073776e-01 -7.87093043e-01 -4.11964327e-01
-1.71524644e-01 3.44582647e-01 7.75994241e-01 -9.89239395e-01
-7.93713450e-01 -6.90947752e-03 1.31805673e-01 4.80676413e-01
-6.39645994e-01 5.37155233e-02 -5.26462615e-01 -1.00704801e+00
-2.10827351e-01 -1.00689709e+00 -4.13459569e-01 -9.05314505e-01
3.04538906e-02 -4.54954743e-01 3.01361904e-02 -3.16015273e-01
1.94097483e+00 -1.80949366e+00 3.99614364e-01 4.50696647e-01
-3.75650488e-02 1.60157725e-01 -6.09180689e-01 1.09608936e+00
8.73172376e-03 3.53934854e-01 5.38494051e-01 -5.33594787e-01
2.56888330e-01 3.21407348e-01 -5.91191590e-01 7.22415298e-02
-3.12598854e-01 9.89087522e-01 -7.15515375e-01 -1.60219390e-02
9.31133851e-02 4.58779156e-01 -1.14086103e+00 1.57960221e-01
-2.41382822e-01 -2.98191905e-01 -1.49792686e-01 6.04300275e-02
2.34951332e-01 -2.39813596e-01 3.81034285e-01 -8.37908089e-02
-1.15258746e-01 7.21516609e-01 -1.41529989e+00 1.45640111e+00
-6.62768722e-01 2.12821201e-03 -4.99799520e-01 -7.76286960e-01
8.74082923e-01 4.36191678e-01 4.70887929e-01 -5.25563240e-01
-1.39455006e-01 1.36806294e-01 -2.84734759e-02 -1.39421090e-01
1.16882074e+00 -2.36099422e-01 -1.29356578e-01 8.70014429e-01
2.11425394e-01 1.26166746e-01 4.64641094e-01 5.60687661e-01
8.34817290e-01 4.58948910e-02 2.28627741e-01 -3.03640068e-01
3.73654515e-01 -2.92904288e-01 1.42680898e-01 1.09560156e+00
8.12399268e-01 5.70754290e-01 3.53799909e-02 -2.54151434e-01
-1.00852072e+00 -8.30888927e-01 6.84841275e-02 1.65431058e+00
-3.24028790e-01 -1.20252240e+00 -2.04107881e-01 -8.28737438e-01
1.50941297e-01 1.32166791e+00 -6.70726299e-01 -1.46067843e-01
-2.72961795e-01 -8.95812988e-01 1.60986260e-01 4.81186777e-01
-5.30627608e-01 -9.18289542e-01 1.93381786e-01 6.98105395e-01
-8.51653144e-02 -6.93863094e-01 -9.29198146e-01 2.00759098e-01
-9.01564479e-01 -5.17544866e-01 -7.58905172e-01 -5.88645577e-01
5.05350590e-01 6.09935999e-01 1.40766537e+00 -5.41055873e-02
3.37331176e-01 6.26064062e-01 -1.15703619e+00 -9.25202295e-02
-4.35572386e-01 1.86118528e-01 5.88776886e-01 4.84040529e-02
5.47768712e-01 -5.35602927e-01 -4.09804434e-01 4.85851169e-01
-1.07772982e+00 3.49223893e-03 3.55867684e-01 8.48512471e-01
6.17541850e-01 1.23334959e-01 7.96680152e-01 -1.13503587e+00
1.10712850e+00 -7.61584818e-01 -1.56656250e-01 1.87243432e-01
-1.25729024e+00 2.18710691e-01 1.06604481e+00 -6.31036282e-01
-6.91891491e-01 -5.77959359e-01 -4.98799860e-01 -2.21692801e-01
2.45393842e-01 8.82924736e-01 1.54907927e-01 3.99721682e-01
6.85432196e-01 4.94813710e-01 -2.59602010e-01 -8.75039279e-01
8.02208364e-01 8.59218895e-01 -2.27516532e-01 -3.83566618e-01
5.36116183e-01 -7.34096300e-03 -3.83535266e-01 -5.99019289e-01
-7.84870803e-01 -7.51650393e-01 -3.86615634e-01 -3.78295057e-03
3.33354414e-01 -8.05538476e-01 -1.93853810e-01 -1.52256742e-01
-6.79604828e-01 -1.13276020e-01 -6.02551103e-01 7.30051756e-01
-4.87014055e-01 4.03960556e-01 -6.39162421e-01 -7.31815577e-01
-6.08385623e-01 -9.32608426e-01 7.90287495e-01 -1.99930131e-01
-3.57286751e-01 -1.30309737e+00 3.69048417e-01 1.45618930e-01
6.07403755e-01 -7.96543002e-01 9.64195251e-01 -1.13774121e+00
-1.69658825e-01 -2.40812778e-01 3.26806247e-01 4.92996603e-01
1.61384135e-01 -3.02399874e-01 -2.14721411e-01 -5.91607988e-01
-3.54281545e-01 9.94694233e-03 8.77819359e-01 3.36419672e-01
8.10002804e-01 -3.93643469e-01 -7.87809342e-02 3.93659383e-01
1.41334331e+00 7.47267753e-02 6.59171641e-01 1.58566579e-01
6.67118728e-01 3.32870394e-01 7.65825808e-01 6.13767862e-01
4.21765029e-01 1.11162806e+00 6.75684661e-02 4.65987712e-01
-2.18010738e-01 -5.78635573e-01 8.37894857e-01 1.41944993e+00
-2.81598508e-01 -8.22071552e-01 -2.28718594e-01 4.66580212e-01
-1.93193531e+00 -1.12347043e+00 -2.42156968e-01 2.42202854e+00
5.77423692e-01 3.71048562e-02 3.46700162e-01 4.93899584e-02
1.53557897e-01 -4.78637181e-02 -2.64334381e-02 -8.31156611e-01
-7.02310875e-02 2.76434571e-01 6.03370249e-01 5.58096170e-01
-6.28838003e-01 9.64365184e-01 6.48893642e+00 1.11655676e+00
-6.84752107e-01 8.72630328e-02 1.15803510e-01 -4.12295014e-01
-8.98545861e-01 -9.45129171e-02 -1.42149663e+00 5.79292238e-01
1.35265732e+00 -2.87243813e-01 7.82792389e-01 5.63490093e-01
3.42287660e-01 3.47236216e-01 -1.17416751e+00 8.98746073e-01
5.63423097e-01 -1.41631889e+00 5.93148708e-01 4.30355847e-01
8.11609983e-01 -1.80838406e-02 3.04362386e-01 7.93675721e-01
6.07186794e-01 -9.06671166e-01 5.35734355e-01 6.07313871e-01
6.60473049e-01 -8.82050872e-01 7.94765353e-01 4.30375725e-01
-1.14851165e+00 -3.23937945e-02 -8.37588787e-01 7.27722645e-02
4.40358698e-01 5.80909193e-01 -8.15698385e-01 8.58210325e-01
5.02617061e-01 1.02482414e+00 -3.37712109e-01 8.79382253e-01
1.54783860e-01 1.00004578e+00 -2.38141268e-01 -3.43926877e-01
3.20305079e-01 -5.79887629e-01 5.92180431e-01 1.39292943e+00
9.10417557e-01 4.85611446e-02 7.21250847e-02 3.72944713e-01
4.90986630e-02 7.54967511e-01 -3.94020438e-01 -2.25544244e-01
3.28573197e-01 1.02321911e+00 -5.05689502e-01 -3.80996227e-01
-5.71759522e-01 1.02226734e+00 2.45075583e-01 2.07017884e-01
-7.11130619e-01 3.89652997e-02 7.54640341e-01 2.90044010e-01
9.67533171e-01 -1.32770240e-01 9.47068781e-02 -1.23883927e+00
-5.21268308e-01 -1.16551197e+00 3.22073430e-01 -4.04261172e-01
-1.55506825e+00 6.30068839e-01 9.44145918e-02 -1.42163789e+00
-5.31060040e-01 -5.31698883e-01 -1.74368665e-01 8.87982607e-01
-9.88937318e-01 -9.90074217e-01 4.47161853e-01 5.01887858e-01
8.64355505e-01 -6.23938262e-01 1.21765721e+00 4.66190994e-01
-1.51498362e-01 8.67040813e-01 7.78715789e-01 -5.17913103e-01
7.11540341e-01 -1.30884933e+00 5.84114671e-01 6.95157826e-01
9.08322453e-01 1.05390930e+00 7.77968287e-01 -7.01062441e-01
-1.45045722e+00 -1.26029432e+00 1.28201485e+00 -6.68742478e-01
6.07473016e-01 -4.60216910e-01 -6.27378166e-01 5.94390810e-01
1.13066293e-01 -4.88313645e-01 1.26138556e+00 6.75306559e-01
-4.34045345e-01 -4.77930605e-02 -6.25604868e-01 7.78746784e-01
1.17889953e+00 -2.95119852e-01 -6.07854068e-01 4.81773347e-01
9.04776812e-01 7.81853646e-02 -1.11159098e+00 1.17696635e-02
8.05358827e-01 -7.93358982e-01 1.07711709e+00 -9.51861560e-01
2.25355834e-01 -9.64044556e-02 -5.34635544e-01 -1.72660112e+00
-1.20543957e+00 -3.94502610e-01 -4.66510892e-01 1.25232017e+00
6.96134031e-01 -4.41333026e-01 5.98612964e-01 1.40162081e-01
-2.36931182e-02 -7.95445681e-01 -4.58801270e-01 -1.03953087e+00
1.55669466e-01 -7.87334681e-01 7.53393948e-01 4.84873861e-01
-1.80757232e-02 6.45339191e-01 -9.21233237e-01 -2.37839580e-01
2.64738232e-01 2.46331409e-01 6.91640854e-01 -1.15480614e+00
-8.52276683e-01 -3.60389918e-01 4.97623384e-02 -1.54415405e+00
-1.57606885e-01 -1.33103859e+00 -3.48382801e-01 -1.77806771e+00
1.43777549e-01 -5.61460316e-01 -7.25922167e-01 2.41404474e-01
-2.26152502e-02 4.18134391e-01 2.70553857e-01 1.57372594e-01
-5.65272689e-01 5.11212170e-01 1.00304401e+00 1.28221989e-01
-4.30496424e-01 2.78131068e-01 -1.10334778e+00 1.15915693e-01
5.90597391e-01 -5.84993899e-01 -7.90527880e-01 -2.50072032e-01
8.80562067e-01 -1.67487204e-01 -2.78854579e-01 -4.52259272e-01
8.55672732e-02 -6.95630461e-02 -5.59992529e-02 -6.62778318e-01
3.02808374e-01 -8.13015819e-01 5.79862177e-01 5.98064512e-02
-6.05510890e-01 2.32287928e-01 -1.26855537e-01 7.16639698e-01
8.84041041e-02 -4.32138890e-01 -7.23702600e-03 -1.10368468e-02
-4.91208404e-01 3.14206779e-01 -8.29124868e-01 -2.52791822e-01
4.46251631e-01 -1.14445530e-01 1.69355541e-01 -6.59053028e-01
-7.55472481e-01 -5.98821864e-02 2.30829015e-01 9.01213348e-01
7.63097823e-01 -1.57026446e+00 -1.05613184e+00 3.40924770e-01
3.27966630e-01 -8.42215717e-01 2.54680365e-01 6.80629432e-01
2.13679411e-02 8.39560151e-01 2.68489808e-01 -5.57227358e-02
-1.55342245e+00 7.57100701e-01 -2.09282309e-01 -7.11768925e-01
-4.78453398e-01 9.58682716e-01 -4.02339250e-02 -4.73894924e-01
2.28192266e-02 -2.61847436e-01 -7.75077403e-01 3.48217875e-01
6.94674194e-01 1.56343788e-01 1.28065675e-01 -5.94747961e-01
-7.97142461e-02 4.38116014e-01 -4.64110613e-01 -2.79634714e-01
1.53735864e+00 -3.88470799e-01 -2.83083804e-02 5.10483921e-01
1.07327044e+00 4.04442281e-01 -3.87795866e-01 -4.73455936e-01
-2.15112582e-01 -4.59332883e-01 2.84093440e-01 -8.35505843e-01
-8.68688464e-01 3.20647657e-01 2.63313174e-01 4.32968915e-01
8.86459947e-01 -1.11027770e-01 6.90053165e-01 3.02694827e-01
6.78027332e-01 -8.86338532e-01 -1.77750528e-01 7.46482134e-01
8.88050735e-01 -7.44294107e-01 -1.78356550e-03 -7.30429143e-02
-8.23477209e-01 9.51929390e-01 7.96709284e-02 -1.75065741e-01
9.32245016e-01 -4.05279696e-02 -2.52750754e-01 5.16858809e-02
-1.29668713e+00 -3.03343385e-01 8.39540958e-01 3.56570363e-01
8.57980072e-01 3.08147311e-01 -8.48311305e-01 1.14474642e+00
-3.90800476e-01 -1.21011555e-01 4.80166197e-01 2.94609606e-01
-4.63134855e-01 -1.81054902e+00 -1.97148472e-01 1.02117276e+00
-4.30648923e-01 -7.17000008e-01 -9.95504409e-02 2.85734296e-01
1.44998446e-01 1.19420552e+00 -1.07908435e-01 -9.04159784e-01
2.75712788e-01 1.09142959e-01 4.15342480e-01 -1.03221393e+00
-8.20767999e-01 3.15468460e-01 3.15197885e-01 -3.61128390e-01
-1.99371949e-01 -7.93780208e-01 -6.98985934e-01 -3.93583357e-01
-4.98008549e-01 6.69162095e-01 4.77166981e-01 7.01492071e-01
5.51833689e-01 4.54911858e-01 7.54003704e-01 -6.29123211e-01
-6.82029009e-01 -1.12488401e+00 -1.00798845e+00 4.36382651e-01
-2.30276838e-01 -5.24427474e-01 -3.62567604e-01 -1.44286171e-01] | [10.127577781677246, 5.713368892669678] |
3f4f8294-0e22-4c4c-86d2-e3deb2b49bb2 | when-a-computer-cracks-a-joke-automated | 2109.08702 | null | https://arxiv.org/abs/2109.08702v1 | https://arxiv.org/pdf/2109.08702v1.pdf | When a Computer Cracks a Joke: Automated Generation of Humorous Headlines | Automated news generation has become a major interest for new agencies in the past. Oftentimes headlines for such automatically generated news articles are unimaginative as they have been generated with ready-made templates. We present a computationally creative approach for headline generation that can generate humorous versions of existing headlines. We evaluate our system with human judges and compare the results to human authored humorous titles. The headlines produced by the system are considered funny 36\% of the time by human evaluators. | ['Mika Hämäläinen', 'Khalid Alnajjar'] | 2021-09-17 | null | null | null | null | ['headline-generation', 'news-generation'] | ['natural-language-processing', 'natural-language-processing'] | [-1.05182163e-01 9.87590253e-01 7.33767301e-02 -2.58171201e-01
-8.46958995e-01 -5.85934341e-01 1.05630529e+00 1.00749843e-01
-3.56262863e-01 1.28412688e+00 1.21558642e+00 -8.87081251e-02
3.10478359e-01 -5.83285570e-01 -3.03773552e-01 -1.18834004e-01
4.90829915e-01 7.70304739e-01 -2.32230090e-02 -8.16087127e-01
7.00896502e-01 2.38645732e-01 -1.26518679e+00 5.93968809e-01
6.63348734e-01 2.75476724e-01 3.75076495e-02 9.07347739e-01
-3.24031383e-01 1.40249956e+00 -1.33926070e+00 -9.52552676e-01
-1.03691384e-01 -7.99922287e-01 -9.79519784e-01 2.91019559e-01
4.62122470e-01 -2.71860480e-01 -2.49757320e-01 1.08220088e+00
6.24904096e-01 1.39746010e-01 6.88137054e-01 -8.10437739e-01
-5.65073729e-01 1.36668015e+00 -3.51786017e-01 4.51910757e-02
9.33793724e-01 -6.05190471e-02 1.16257882e+00 -8.19077253e-01
1.31259894e+00 1.38514328e+00 4.84953582e-01 4.98739243e-01
-1.17232466e+00 -3.12919557e-01 -4.60919231e-01 2.83003859e-02
-1.00000954e+00 -5.77856541e-01 7.47987866e-01 -5.99596441e-01
9.43038046e-01 6.87854469e-01 5.16524971e-01 9.76251662e-01
4.82267976e-01 7.18503535e-01 8.46853137e-01 -3.79765391e-01
3.95219177e-02 5.41488111e-01 -3.16771604e-02 2.43787691e-01
1.56113654e-01 -4.55297112e-01 -3.00970584e-01 -4.75492835e-01
4.36774373e-01 -5.18138468e-01 -4.73154724e-01 6.44587755e-01
-1.14283574e+00 1.19284463e+00 2.03205004e-01 2.71117151e-01
-6.93903029e-01 -2.22675145e-01 6.33156836e-01 1.58332348e-01
6.88829720e-01 1.33965671e+00 2.38743931e-01 -2.72244632e-01
-1.11321390e+00 1.05824685e+00 1.38503194e+00 1.15813804e+00
1.50519073e-01 1.86732426e-01 -5.77548623e-01 8.25344503e-01
-3.27169120e-01 7.89840579e-01 5.31150579e-01 -7.91954935e-01
2.61100233e-01 3.10038775e-01 8.19547772e-01 -1.50756347e+00
-3.35004032e-01 -4.29629385e-01 -3.29557121e-01 7.54267499e-02
1.83549181e-01 -4.91135955e-01 -3.19889545e-01 8.81544411e-01
-1.80012777e-01 -1.07336748e+00 -8.55393615e-03 6.75433815e-01
1.02429152e+00 1.25619638e+00 -2.39524290e-01 -8.58855784e-01
1.06871152e+00 -1.02617085e+00 -1.38891709e+00 8.17420613e-03
3.45763981e-01 -1.43729782e+00 1.23356736e+00 6.55781686e-01
-1.48851264e+00 -2.89647996e-01 -9.58199501e-01 9.45780799e-02
-6.57810420e-02 -4.37138639e-02 -4.21212502e-02 2.52661675e-01
-6.77113175e-01 5.89914083e-01 9.40643251e-02 -3.09616446e-01
5.71665354e-02 -3.28672349e-01 -5.96062951e-02 3.93885314e-01
-1.10755348e+00 1.19105709e+00 3.32069665e-01 -6.70745730e-01
-3.81933212e-01 -3.48346204e-01 -5.05179942e-01 -1.49668843e-01
2.66822129e-01 -4.30530846e-01 2.23353434e+00 -7.38474250e-01
-1.21678531e+00 7.37647176e-01 -1.47869974e-01 -5.50823987e-01
8.22056770e-01 -5.15181243e-01 -6.29571617e-01 3.12661529e-01
4.58475888e-01 4.48477894e-01 8.65767121e-01 -1.30455863e+00
-5.89063406e-01 3.99258792e-01 -9.43869650e-02 1.65683657e-01
-2.68243402e-01 5.93969107e-01 3.06658685e-01 -1.05416596e+00
-6.83835983e-01 -7.27554262e-01 -2.78128773e-01 -6.40137732e-01
-8.74285579e-01 -1.74436018e-01 6.27819180e-01 -8.71670723e-01
1.81951845e+00 -1.60647440e+00 -1.57827228e-01 4.44843210e-02
2.72303909e-01 7.52597004e-02 2.46484429e-01 9.09984171e-01
-4.40115854e-02 2.95907080e-01 3.94567698e-01 1.87128365e-01
2.25910991e-01 -2.10866928e-01 -8.18439901e-01 -5.83177358e-02
-2.08690614e-01 6.92497611e-01 -1.13918805e+00 -6.68476760e-01
-3.40784431e-01 -1.26090169e-01 -4.06242788e-01 4.63308781e-01
-5.03150403e-01 -2.08752126e-01 -1.72471777e-01 2.46969089e-01
9.01547372e-02 -2.27961317e-01 -1.03324942e-01 -1.55108631e-01
-4.64740783e-01 5.14185369e-01 -4.90402550e-01 7.93250144e-01
-2.50140488e-01 9.55217361e-01 -4.94115472e-01 2.73297489e-01
1.20159519e+00 5.57475328e-01 -1.02778748e-01 -5.54531455e-01
2.49334350e-01 1.81656376e-01 -1.66304171e-01 -6.52553022e-01
1.32667363e+00 -7.32295573e-01 -4.02065575e-01 9.42827523e-01
-2.44351625e-01 -7.15129316e-01 5.55877745e-01 6.23924255e-01
8.82945359e-01 -3.01578671e-01 8.31733704e-01 1.42396183e-03
1.51269987e-01 7.75485396e-01 -3.10679805e-02 8.84454846e-01
2.43594974e-01 7.87259161e-01 6.53015077e-01 -5.99682570e-01
-1.67921948e+00 -5.36436379e-01 2.10789852e-02 1.02181661e+00
-4.95864391e-01 -6.96309686e-01 -1.08361173e+00 -3.91073376e-01
-5.47350347e-01 1.59849989e+00 -4.67045933e-01 3.34546149e-01
-4.96889859e-01 -4.22291785e-01 4.28441823e-01 6.10722639e-02
-4.41718735e-02 -1.43644226e+00 -7.32983947e-01 8.41735125e-01
-6.52547956e-01 -1.04964519e+00 -5.49476564e-01 -1.86805665e-01
-2.29084626e-01 -7.63790488e-01 -9.16349411e-01 -3.77254784e-01
5.60060382e-01 2.47215852e-01 1.29675543e+00 -1.10205896e-01
-1.53032169e-01 -1.96816102e-01 -8.38719249e-01 -8.05868268e-01
-1.18839979e+00 -1.16245791e-01 4.70585078e-02 -4.29336607e-01
-1.66256493e-03 -2.55745292e-01 -8.29232577e-03 1.34472489e-01
-1.13441277e+00 5.20032823e-01 1.97716802e-01 8.33092272e-01
-1.68765150e-02 -6.01340160e-02 7.48060346e-01 -1.59607458e+00
1.48545074e+00 -3.29243720e-01 -1.04726113e-01 -2.40575485e-02
-4.39329952e-01 6.98771626e-02 9.98321831e-01 -2.76033938e-01
-1.37938941e+00 -2.50222802e-01 -1.94154233e-01 2.20299825e-01
-3.04411966e-02 6.33222580e-01 4.44682121e-01 4.13626164e-01
1.58086205e+00 -1.88556299e-01 -1.65990144e-01 -4.23357636e-01
5.18516600e-01 8.88155282e-01 8.01870465e-01 -3.30343843e-01
1.00191116e+00 2.22160786e-01 -6.55127883e-01 -8.86367917e-01
-1.19682610e+00 -2.65697837e-01 3.90316695e-02 -7.50327170e-01
5.08977056e-01 -6.22366667e-01 -1.00945942e-01 2.20023304e-01
-1.84959781e+00 1.01495929e-01 -5.67411959e-01 8.97422284e-02
-5.43885469e-01 1.18969113e-01 -5.93015373e-01 -5.19720018e-01
-8.11400950e-01 -6.18554175e-01 5.24864435e-01 4.90789622e-01
-1.31464744e+00 -6.15964055e-01 4.58784044e-01 2.79605389e-01
4.09322381e-01 5.03856659e-01 9.24117088e-01 -1.09905112e+00
2.40378678e-01 -8.55747342e-01 -2.52616465e-01 -2.98278797e-02
-3.49904001e-02 3.35740447e-01 -8.15710664e-01 3.17944378e-01
-9.78422165e-02 -4.13145959e-01 1.60011470e-01 6.63816258e-02
2.26457089e-01 -1.56644416e+00 -5.07622212e-02 -3.24129403e-01
9.96971428e-01 2.94460744e-01 7.42456377e-01 5.23820817e-01
2.99471438e-01 6.20224595e-01 6.66501939e-01 1.19315600e+00
3.30755636e-02 5.85139990e-01 -2.46764794e-02 4.00524540e-03
-5.64502887e-02 -6.22225642e-01 3.75419319e-01 9.06283081e-01
-9.98097733e-02 -4.79875922e-01 -7.91202664e-01 5.92338800e-01
-1.72831917e+00 -1.65375686e+00 -5.94404519e-01 1.46687782e+00
1.04948366e+00 5.63043952e-01 4.19116288e-01 -2.58637946e-02
8.44747841e-01 2.67678797e-01 6.84643313e-02 -1.01665223e+00
-2.49720633e-01 1.84737056e-01 2.16553792e-01 6.65741861e-01
-6.57556951e-01 7.65400231e-01 7.45441914e+00 6.83408022e-01
-7.71308482e-01 -1.51963368e-01 5.00579238e-01 -2.00514823e-01
-5.70309699e-01 -5.47263399e-02 -8.35811019e-01 2.93922871e-01
8.89596045e-01 -1.09290075e+00 7.93310702e-02 1.42229629e+00
9.06913519e-01 -1.29719555e-01 -7.99519420e-01 7.37518847e-01
5.88433981e-01 -1.77084565e+00 4.46574718e-01 -2.59782881e-01
1.19281209e+00 -5.21653771e-01 -2.78955728e-01 1.12861998e-01
5.47643363e-01 -1.01308286e+00 1.05514348e+00 6.65643096e-01
3.98816079e-01 -1.01063299e+00 9.81321633e-01 3.18802685e-01
-1.74977705e-01 2.05551878e-01 -5.16731501e-01 -3.51326913e-01
8.74778748e-01 8.54429364e-01 -1.56997538e+00 -3.12815905e-02
1.49739116e-01 1.45918861e-01 -7.34688699e-01 1.28392589e+00
-4.66380417e-01 5.19080281e-01 1.79545388e-01 -6.92682147e-01
2.92469174e-01 2.55307257e-01 9.56418753e-01 1.76116586e+00
2.12947398e-01 2.11222097e-01 1.38484150e-01 6.59841120e-01
-9.32949781e-02 5.17520785e-01 -7.73044825e-01 -2.97332704e-01
2.85693139e-01 1.41295028e+00 -5.49507558e-01 -6.87303960e-01
1.35313198e-01 7.93275416e-01 -8.26985985e-02 9.27805379e-02
-7.53382921e-01 -9.25741553e-01 -9.58931968e-02 5.54556012e-01
-2.64869094e-01 3.27204764e-01 -4.10775959e-01 -8.15853953e-01
-4.32744533e-01 -1.26517344e+00 7.85439685e-02 -1.42298305e+00
-1.28373694e+00 1.29346728e+00 1.26180828e-01 -1.18613231e+00
-7.67459989e-01 -1.36588231e-01 -9.79985118e-01 4.76161748e-01
-5.66964388e-01 -5.81738830e-01 -1.66735739e-01 -1.25614107e-01
8.90225768e-01 -1.04934126e-01 7.02437282e-01 -7.08040670e-02
-1.54064476e-01 3.14396739e-01 -2.88726598e-01 -7.84800798e-02
8.43514979e-01 -1.11753201e+00 5.87226331e-01 7.30206728e-01
5.93986325e-02 3.99279147e-01 1.84452701e+00 -7.77642846e-01
-3.82840455e-01 -7.36382186e-01 1.73365843e+00 -2.03286201e-01
1.04067922e+00 1.67384550e-01 -6.79564714e-01 2.45839670e-01
9.41702902e-01 -1.12493753e+00 7.92641521e-01 -2.47769356e-01
-2.87974626e-01 4.16388720e-01 -9.21399951e-01 8.50415647e-01
4.16909993e-01 -1.63970321e-01 -1.13089371e+00 9.43956971e-01
8.65305662e-01 -4.16182369e-01 -3.02146137e-01 -2.63799518e-01
3.01406354e-01 -7.95418501e-01 1.86943233e-01 -7.91355550e-01
1.26455677e+00 -3.68801117e-01 2.46876970e-01 -1.51137900e+00
-3.51781487e-01 -1.17365086e+00 4.11662310e-01 1.15996909e+00
8.98704529e-01 8.05634912e-03 1.07609160e-01 8.37621987e-01
-4.05555755e-01 -2.23876476e-01 -1.91596404e-01 -5.34139335e-01
-4.79300767e-01 -6.82859644e-02 2.20690653e-01 8.78564119e-01
8.62609327e-01 9.59298849e-01 -8.52744162e-01 -6.44101381e-01
2.59777725e-01 9.45287570e-02 1.17741394e+00 -9.61648345e-01
-3.90260428e-01 -6.05966210e-01 -3.71940322e-02 -7.15116560e-01
-2.80645669e-01 -5.84830284e-01 2.70507365e-01 -1.50152159e+00
4.88309860e-01 4.50720996e-01 6.37240708e-01 4.48893487e-01
3.24491598e-02 2.17247888e-01 3.66077960e-01 2.95777977e-01
-5.04972875e-01 2.78971851e-01 1.42577600e+00 -3.61102521e-02
-4.81802940e-01 -1.57582536e-01 -1.03337288e+00 9.12955880e-01
8.20636153e-01 -7.09513605e-01 -9.32843518e-03 3.01671714e-01
9.13881600e-01 1.71148539e-01 4.91098724e-02 -9.98932600e-01
-2.12517641e-02 -5.28590262e-01 1.31705835e-01 -6.61185622e-01
-1.55816406e-01 -1.15435869e-01 6.07049048e-01 1.66942373e-01
-8.24140787e-01 3.56142461e-01 -1.12295337e-01 1.73759311e-01
-2.94417650e-01 -8.06367576e-01 9.04656053e-01 -4.09384191e-01
-1.52474076e-01 -4.19042140e-01 -8.78184676e-01 3.98561031e-01
7.18734682e-01 -7.34612644e-02 -7.70608604e-01 -1.18006206e+00
-4.13288265e-01 -3.52772534e-01 5.89148641e-01 2.35218778e-01
6.10782087e-01 -1.14163327e+00 -1.12343943e+00 -8.16003740e-01
1.79259524e-01 -4.86585915e-01 -3.25422972e-01 4.36590880e-01
-1.16032100e+00 3.93912911e-01 -3.87062609e-01 4.03180718e-01
-1.04936385e+00 4.79815841e-01 -2.22865134e-01 -2.57526457e-01
-8.59620750e-01 5.28900802e-01 1.19492114e-02 2.30501562e-01
-1.39676286e-02 1.69348896e-01 -5.17243803e-01 4.96799529e-01
1.24366820e+00 4.36038315e-01 -1.06989145e-01 -7.93206513e-01
3.30355853e-01 -3.50261539e-01 -5.86473823e-01 -6.47197485e-01
1.41226399e+00 7.25741088e-02 7.96920527e-03 5.32165647e-01
7.82413423e-01 4.09853578e-01 -5.23993194e-01 1.14317112e-01
1.60005540e-01 -4.11247194e-01 -2.92649299e-01 -8.86173964e-01
-2.45702296e-01 2.28398025e-01 -4.57461208e-01 6.79703772e-01
5.51274896e-01 9.99839306e-02 1.07154632e+00 6.45278215e-01
1.03541300e-01 -1.50603533e+00 3.81428212e-01 4.89665627e-01
1.59592652e+00 -6.95466042e-01 4.76440430e-01 -8.99118260e-02
-1.21945441e+00 1.37708390e+00 3.70077819e-01 -7.02269226e-02
-4.36210819e-02 6.38118908e-02 3.32809389e-01 -4.73186940e-01
-9.65645254e-01 2.63631374e-01 1.83037341e-01 4.41683173e-01
7.48950005e-01 1.81919992e-01 -1.01906312e+00 7.08932340e-01
-9.99397218e-01 -2.24606588e-01 1.63247275e+00 7.72656798e-01
-1.11457908e+00 -5.65837443e-01 -7.39884138e-01 5.55064082e-01
-6.54977798e-01 -1.68534577e-01 -9.53387320e-01 6.16464376e-01
-5.33137500e-01 1.04603744e+00 -4.35554713e-01 -3.64967763e-01
4.39388722e-01 1.23518355e-01 -7.33040720e-02 -9.21635985e-01
-9.19479251e-01 3.53654563e-01 1.06109583e+00 1.15726450e-02
-1.59127340e-01 -5.87487459e-01 -1.25634825e+00 -6.74199641e-01
-3.48884881e-01 5.70550561e-01 5.73951483e-01 7.49646604e-01
-3.18608359e-02 1.10393986e-01 8.85614395e-01 -9.27114248e-01
-6.62248731e-01 -1.12158084e+00 -3.01338166e-01 6.55641139e-01
1.81219094e-02 4.24964167e-03 -3.51274490e-01 6.15900457e-01] | [12.237801551818848, 9.379389762878418] |
6ab24947-d9f9-4833-a183-64f958102314 | deep-curiosity-search-intra-life-exploration | 1806.00553 | null | http://arxiv.org/abs/1806.00553v3 | http://arxiv.org/pdf/1806.00553v3.pdf | Deep Curiosity Search: Intra-Life Exploration Can Improve Performance on Challenging Deep Reinforcement Learning Problems | Traditional exploration methods in RL require agents to perform random
actions to find rewards. But these approaches struggle on sparse-reward domains
like Montezuma's Revenge where the probability that any random action sequence
leads to reward is extremely low. Recent algorithms have performed well on such
tasks by encouraging agents to visit new states or perform new actions in
relation to all prior training episodes (which we call across-training
novelty). But such algorithms do not consider whether an agent exhibits
intra-life novelty: doing something new within the current episode, regardless
of whether those behaviors have been performed in previous episodes. We
hypothesize that across-training novelty might discourage agents from
revisiting initially non-rewarding states that could become important stepping
stones later in training. We introduce Deep Curiosity Search (DeepCS), which
encourages intra-life exploration by rewarding agents for visiting as many
different states as possible within each episode, and show that DeepCS matches
the performance of current state-of-the-art methods on Montezuma's Revenge. We
further show that DeepCS improves exploration on Amidar, Freeway, Gravitar, and
Tutankham (many of which are hard exploration games). Surprisingly, DeepCS
doubles A2C performance on Seaquest, a game we would not have expected to
benefit from intra-life exploration because the arena is small and already
easily navigated by naive exploration techniques. In one run, DeepCS achieves a
maximum training score of 80,000 points on Seaquest, higher than any methods
other than Ape-X. The strong performance of DeepCS on these sparse- and
dense-reward tasks suggests that encouraging intra-life novelty is an
interesting, new approach for improving performance in Deep RL and motivates
further research into hybridizing across-training and intra-life exploration
methods. | ['Jeff Clune', 'Christopher Stanton'] | 2018-06-01 | null | https://openreview.net/forum?id=BkeDEoCctQ | https://openreview.net/pdf?id=BkeDEoCctQ | null | ['montezumas-revenge'] | ['playing-games'] | [-2.35860065e-01 1.26376480e-01 -2.75640100e-01 5.46870120e-02
-6.74430728e-01 -8.30809057e-01 8.42611194e-01 -9.17335972e-02
-1.09506965e+00 1.42637205e+00 1.83529213e-01 -4.57393348e-01
-2.17651129e-01 -8.40301514e-01 -8.63304436e-01 -6.87724292e-01
-7.61113703e-01 8.55677128e-01 1.06553838e-01 -5.71216047e-01
3.17015588e-01 3.52156788e-01 -1.52816653e+00 -2.01179996e-01
8.07302356e-01 3.62842023e-01 2.07833424e-01 6.14852726e-01
1.74311414e-01 7.03297853e-01 -8.06843817e-01 2.74027959e-02
5.21222532e-01 -7.60553300e-01 -9.16797221e-01 -4.32780862e-01
-2.35053033e-01 -4.47397113e-01 -7.99739733e-02 8.80127788e-01
4.28709298e-01 6.58803582e-01 1.40682653e-01 -1.05938828e+00
-2.23284364e-01 1.13309526e+00 -6.93396270e-01 4.13879722e-01
4.19809937e-01 6.41987264e-01 9.76054609e-01 -3.20528775e-01
7.71885991e-01 1.20423794e+00 3.72354925e-01 7.62619913e-01
-1.27580249e+00 -8.08828354e-01 4.97875452e-01 -9.20126513e-02
-6.90293849e-01 -1.59453943e-01 3.22885454e-01 7.18724951e-02
1.21989107e+00 3.81572336e-01 1.19255912e+00 1.37435961e+00
1.14076667e-01 1.07139158e+00 1.52351487e+00 -3.76657769e-02
9.20762360e-01 -2.48536333e-01 -5.02229691e-01 4.55698073e-01
2.93880045e-01 9.59044933e-01 -6.67436838e-01 -2.32369468e-01
9.26500261e-01 -1.43081516e-01 -5.54541834e-02 -5.92898607e-01
-1.35661197e+00 1.13336265e+00 6.59927905e-01 3.63957286e-01
-6.19417429e-01 5.71850777e-01 3.38804930e-01 5.51007092e-01
-1.87226325e-01 1.47871745e+00 -5.17631054e-01 -9.97278333e-01
-6.98550701e-01 6.28244340e-01 7.07728326e-01 6.25091493e-01
8.23750377e-01 2.57859856e-01 -6.40556123e-03 3.33908290e-01
-1.59768090e-01 2.77231961e-01 7.30414569e-01 -1.49981868e+00
3.09570283e-01 4.58376437e-01 4.78858650e-01 -3.56551260e-01
-6.28942311e-01 -8.16835701e-01 -3.75853688e-01 8.84686947e-01
4.30919379e-01 -6.67325258e-01 -7.60125935e-01 2.03404498e+00
1.83722854e-01 -9.52979699e-02 4.88870293e-01 9.57975328e-01
1.74929082e-01 6.31822824e-01 -1.42773986e-01 -2.87829667e-01
1.00469947e+00 -1.18995500e+00 -3.09507877e-01 -7.79633641e-01
9.32783306e-01 -5.09441346e-02 1.25836504e+00 6.68236494e-01
-1.26098263e+00 -9.92897600e-02 -8.17450225e-01 5.39887130e-01
-2.55720586e-01 -4.89365250e-01 1.38435292e+00 5.90296447e-01
-9.17416394e-01 8.69953156e-01 -1.03789759e+00 -2.38966301e-01
4.76574332e-01 4.66390342e-01 -2.69115537e-01 1.94703534e-01
-1.13124645e+00 1.00772405e+00 5.56321442e-01 -1.36539385e-01
-1.57918775e+00 -2.45621711e-01 -6.47104979e-01 1.32804632e-01
1.11496949e+00 -4.12864149e-01 1.54322267e+00 -8.72784615e-01
-1.77600539e+00 2.88470536e-01 2.55456239e-01 -7.94610500e-01
5.68713486e-01 -3.36755633e-01 4.68340702e-02 -3.03877413e-01
2.55855381e-01 1.04313803e+00 3.85073334e-01 -1.13900316e+00
-6.83269083e-01 -1.38109416e-01 4.87888753e-01 7.99191415e-01
1.21066347e-01 -4.11273092e-01 -4.81992820e-03 -4.15273577e-01
-1.19085282e-01 -1.18717432e+00 -8.53906989e-01 -5.34650266e-01
-1.87066525e-01 -2.80215204e-01 2.60123879e-01 3.06817114e-01
7.79046059e-01 -1.73909402e+00 3.30267280e-01 2.69892335e-01
-7.07169622e-02 1.10894516e-01 -5.00424266e-01 5.62445521e-01
1.94208622e-01 1.63725734e-01 -7.90937319e-02 -1.22170269e-01
1.35194182e-01 6.14477813e-01 -2.60085046e-01 1.33333653e-01
-3.64936292e-01 1.11705172e+00 -1.54305983e+00 -5.83804771e-02
-9.52399522e-03 -1.23623997e-01 -7.43720293e-01 1.19330198e-01
-5.78961849e-01 6.01780653e-01 -6.06325388e-01 5.03946424e-01
2.49184474e-01 -1.75876796e-01 1.96130797e-01 9.06725943e-01
-3.57294828e-01 5.21505892e-01 -1.06110322e+00 1.88446343e+00
-2.61256248e-01 4.19455707e-01 -1.23581879e-01 -6.75429702e-01
5.74868202e-01 -4.71224748e-02 3.46735626e-01 -1.06431687e+00
1.25036091e-01 3.94905061e-01 3.89370799e-01 2.86638513e-02
7.91367710e-01 -1.47419974e-01 -2.68599123e-01 8.66777360e-01
-3.44446689e-01 -3.87503266e-01 5.60626864e-01 3.74441504e-01
1.53261077e+00 3.10323417e-01 2.83834398e-01 -1.30695120e-01
-1.90919518e-01 2.06646398e-01 6.56506598e-01 1.45138323e+00
-2.21239209e-01 5.50046824e-02 8.79619896e-01 -6.15787387e-01
-7.73870945e-01 -1.07913971e+00 3.20594937e-01 1.33181894e+00
3.36004198e-01 -4.29418802e-01 -4.15274054e-01 -1.06481218e+00
-1.66383018e-07 1.05121922e+00 -1.11927104e+00 -1.79194465e-01
-5.50120175e-01 -5.40251434e-01 5.63492537e-01 4.09241259e-01
5.70898354e-01 -1.82180309e+00 -1.41219831e+00 3.28323901e-01
9.15359035e-02 -3.30071360e-01 -3.05732101e-01 1.02946818e+00
-1.00979602e+00 -8.61429930e-01 -8.53908598e-01 -2.81730026e-01
4.07636851e-01 2.65117176e-02 1.26169336e+00 9.06600617e-03
-6.79417402e-02 1.77502900e-01 -4.47404027e-01 -7.66042843e-02
-1.14414580e-01 2.69806355e-01 2.39061981e-01 -9.72368538e-01
2.50191569e-01 -6.92324460e-01 -7.05448687e-01 3.96965563e-01
-5.02921581e-01 -8.91481116e-02 8.93761039e-01 9.46037412e-01
3.04153472e-01 7.31633529e-02 6.97456717e-01 -6.08214259e-01
8.80447805e-01 -6.58676565e-01 -8.11035454e-01 -1.79566368e-01
-6.07358158e-01 4.91631746e-01 3.88101518e-01 -7.47257411e-01
-8.06434155e-01 -1.47248626e-01 1.00758724e-01 -1.74292699e-01
8.17289874e-02 5.10153413e-01 5.77270865e-01 7.28254765e-02
1.10514069e+00 2.87827879e-01 1.76312886e-02 -1.97020262e-01
2.86173731e-01 -1.80824175e-01 2.64060676e-01 -8.39719534e-01
7.40231037e-01 2.16552973e-01 -8.11515301e-02 -1.93452209e-01
-6.49498940e-01 4.70321178e-02 2.47838020e-01 7.39407241e-02
5.37055373e-01 -8.32731605e-01 -1.38607085e+00 9.49815139e-02
-5.84483743e-01 -1.05630744e+00 -1.05511594e+00 6.48043036e-01
-7.53707707e-01 5.93654637e-04 -2.90816844e-01 -9.78580415e-01
1.88811377e-01 -1.31349874e+00 5.76713204e-01 6.68072402e-01
-3.90175730e-01 -7.00271308e-01 5.40290952e-01 9.25898086e-03
5.29211581e-01 1.03930883e-01 2.11810753e-01 -6.32106960e-01
-8.00019801e-01 3.86613578e-01 2.91466206e-01 -2.39634037e-01
-3.99110094e-02 -6.39837444e-01 -2.87253708e-01 -6.84664726e-01
-2.61707574e-01 -1.10686731e+00 9.10633624e-01 3.59987468e-01
7.89448977e-01 -4.19583559e-01 -4.41860497e-01 4.80575234e-01
1.00613666e+00 5.85128009e-01 6.38116360e-01 1.12212646e+00
-5.30753173e-02 2.46837616e-01 9.89043534e-01 7.84859896e-01
1.90211162e-01 6.10414445e-01 9.05956328e-01 -6.65149582e-06
4.98493284e-01 -3.55617344e-01 5.98159611e-01 -1.45246923e-01
-1.27911821e-01 -1.75603330e-01 -6.25807583e-01 7.17869997e-01
-1.96431291e+00 -1.16332257e+00 5.03315747e-01 2.13948941e+00
1.20424974e+00 5.44178605e-01 4.25452888e-01 -3.14390838e-01
8.21304247e-02 2.51346946e-01 -1.16140211e+00 -7.68951714e-01
-8.27765986e-02 2.65206903e-01 5.68700969e-01 6.54805064e-01
-6.11282647e-01 1.34472048e+00 6.47126532e+00 7.71214068e-01
-7.17105985e-01 -1.23812363e-01 5.32236218e-01 -7.92736828e-01
-4.13288534e-01 3.14940393e-01 -7.65465081e-01 3.02510738e-01
6.66014373e-01 -2.05361862e-02 1.04709899e+00 1.15631533e+00
-1.79389529e-02 -8.10395420e-01 -1.15920806e+00 4.77283895e-01
-3.34522396e-01 -1.27002740e+00 -3.57553333e-01 4.05280322e-01
1.11794412e+00 3.41752738e-01 3.30284327e-01 9.05267835e-01
1.39429462e+00 -1.29015410e+00 5.09057939e-01 1.03778861e-01
3.39841515e-01 -1.01669610e+00 5.34437358e-01 6.83916926e-01
-6.93592250e-01 -3.37650150e-01 -2.48687208e-01 -4.62331831e-01
1.23010173e-01 2.95379339e-03 -9.81308877e-01 1.56838566e-01
8.42261314e-01 3.24968338e-01 -4.03027534e-01 9.43516850e-01
-4.65007812e-01 3.08490902e-01 -6.53180182e-01 -5.48073649e-01
1.14805889e+00 -1.12170666e-01 7.27506518e-01 5.45398235e-01
3.99008632e-01 1.23955332e-01 2.40159020e-01 1.01735771e+00
6.12471253e-02 -3.81568462e-01 -7.16540754e-01 -3.21124673e-01
5.11532247e-01 9.53180492e-01 -6.65265977e-01 -2.54300922e-01
2.21719101e-01 7.68290162e-01 4.51838374e-01 2.52073616e-01
-7.81228125e-01 -2.13239938e-01 6.39900327e-01 -2.45092526e-01
3.74456912e-01 -1.35841876e-01 -1.22505024e-01 -1.06067836e+00
-3.02235484e-01 -1.27048743e+00 3.30812573e-01 -7.23572075e-01
-7.01426625e-01 6.49289370e-01 -4.15964723e-02 -9.58699346e-01
-6.34672046e-01 -4.55394667e-03 -6.97053790e-01 5.01440585e-01
-1.51610529e+00 -4.30665284e-01 9.55452546e-02 4.92868423e-01
6.98555052e-01 -4.05778676e-01 7.26437569e-01 -5.13072193e-01
-3.49834621e-01 4.25606459e-01 1.77567840e-01 -4.14115369e-01
3.72434616e-01 -1.50817382e+00 5.90362310e-01 5.87529957e-01
2.86600232e-01 7.18347311e-01 8.47555578e-01 -9.16484892e-01
-1.18112409e+00 -3.23098183e-01 1.88192725e-01 -2.89751977e-01
5.44070780e-01 -2.23927706e-01 -5.47158659e-01 8.60424340e-01
2.88299114e-01 -2.71616250e-01 1.95602521e-01 6.16514981e-01
-9.24670789e-03 3.56874883e-01 -9.16509390e-01 9.93469894e-01
1.14931262e+00 5.76926433e-02 -7.65864253e-01 2.26274863e-01
4.70580250e-01 -5.93911231e-01 -3.23814362e-01 2.11447015e-01
4.03218478e-01 -1.35419583e+00 7.99606264e-01 -5.60965657e-01
2.17670500e-01 -1.11200020e-01 1.60085082e-01 -1.71785605e+00
-1.86592609e-01 -1.26246059e+00 -5.69962300e-02 4.24918205e-01
5.23703933e-01 -8.04648638e-01 1.37405956e+00 2.28791669e-01
-8.51151496e-02 -9.52913642e-01 -9.64072108e-01 -1.16213596e+00
2.87182480e-01 -7.51170516e-02 7.05046475e-01 8.19665074e-01
3.02218586e-01 1.40748352e-01 -5.38533747e-01 -2.43141189e-01
4.87531364e-01 1.07558131e-01 1.05442905e+00 -8.96166563e-01
-6.46248579e-01 -6.23277426e-01 3.29907149e-01 -1.31972587e+00
-1.48141664e-02 -5.14337003e-01 2.20078602e-01 -1.41668355e+00
1.61423847e-01 -8.85048270e-01 -1.66236669e-01 8.47642243e-01
-1.86220333e-02 -2.31350828e-02 3.13910902e-01 2.52208620e-01
-1.06336570e+00 9.51650202e-01 1.63743579e+00 1.76249355e-01
-7.91204453e-01 -1.06387613e-02 -8.79251719e-01 6.96475744e-01
8.96177411e-01 -5.78649759e-01 -6.04219258e-01 -1.00319788e-01
6.87025011e-01 3.43637526e-01 1.00091681e-01 -1.14018178e+00
2.47792616e-01 -6.28315389e-01 2.29950711e-01 -4.36603308e-01
6.15187228e-01 -6.15067184e-01 1.68094188e-01 9.15770829e-01
-5.65392971e-01 2.28479847e-01 3.00753295e-01 5.68190813e-01
1.83556587e-01 -3.80798995e-01 5.79897046e-01 -5.58746338e-01
-6.62255585e-01 -3.05711515e-02 -8.00626695e-01 4.27113593e-01
1.32223570e+00 -5.51521778e-01 -4.63895828e-01 -6.33816957e-01
-7.15673864e-01 7.26818621e-01 6.37786269e-01 1.42153487e-01
4.73981887e-01 -1.00123012e+00 -4.81920332e-01 9.20840800e-02
-1.66208565e-01 1.62516147e-01 3.02897319e-02 6.36122823e-01
-2.72524923e-01 1.60191923e-01 -6.00265920e-01 -2.39115223e-01
-7.97699809e-01 4.87911403e-01 4.14395094e-01 -8.90390635e-01
-6.92066133e-01 1.18249857e+00 2.01003462e-01 -5.56695521e-01
3.94632727e-01 -3.11865062e-01 -8.99013877e-02 4.55592386e-03
1.48535341e-01 3.44535053e-01 -5.68113208e-01 1.63604379e-01
-3.14111710e-01 1.86452404e-01 -4.68470573e-01 -5.87577045e-01
1.43058717e+00 1.44299001e-01 3.58213037e-01 2.98190206e-01
5.16885519e-01 -6.21819161e-02 -1.90184712e+00 4.92440797e-02
-1.41041845e-01 -6.15846992e-01 -6.56375512e-02 -1.41116238e+00
-8.99479985e-01 5.21803677e-01 2.97613174e-01 2.46395007e-01
8.43617082e-01 -2.40950305e-02 7.10056543e-01 1.02366042e+00
7.95334816e-01 -1.19223452e+00 6.90724671e-01 7.14817286e-01
7.95224428e-01 -1.25276709e+00 1.64690137e-01 5.96243083e-01
-1.03275824e+00 6.25402212e-01 9.05354559e-01 -2.86198825e-01
-1.72075778e-01 1.31092221e-01 -3.39846700e-01 -3.74921590e-01
-1.15654469e+00 -4.29686934e-01 -3.64449888e-01 5.39550602e-01
-2.02208623e-01 7.67052174e-02 -6.13329224e-02 1.01848990e-01
-5.33784509e-01 -4.32199910e-02 6.90344155e-01 1.28441596e+00
-7.07145631e-01 -1.17627394e+00 -1.82874084e-01 3.64424646e-01
-2.15849355e-01 2.33226921e-02 -4.23847467e-01 1.07435977e+00
-9.99349281e-02 7.10778773e-01 6.95995912e-02 -1.38239652e-01
1.17219370e-02 -4.24650967e-01 4.90363657e-01 -6.35387123e-01
-1.00897932e+00 6.57350942e-02 2.32204825e-01 -1.00765324e+00
-1.74876582e-02 -8.57882917e-01 -1.48557770e+00 -4.90356594e-01
-2.76539803e-01 7.26443350e-01 4.01311457e-01 7.21878529e-01
2.44927719e-01 4.94623691e-01 5.39683998e-01 -9.90445852e-01
-7.20138729e-01 -6.15289569e-01 -5.69570959e-01 -7.62478933e-02
4.22640592e-01 -9.05691445e-01 -6.08046234e-01 -9.78169024e-01] | [3.8818957805633545, 1.638156533241272] |
b2d2413f-214b-406f-8a0b-25bdea309ab2 | revisiting-data-augmentation-in-model | 2305.13232 | null | https://arxiv.org/abs/2305.13232v1 | https://arxiv.org/pdf/2305.13232v1.pdf | Revisiting Data Augmentation in Model Compression: An Empirical and Comprehensive Study | The excellent performance of deep neural networks is usually accompanied by a large number of parameters and computations, which have limited their usage on the resource-limited edge devices. To address this issue, abundant methods such as pruning, quantization and knowledge distillation have been proposed to compress neural networks and achieved significant breakthroughs. However, most of these compression methods focus on the architecture or the training method of neural networks but ignore the influence from data augmentation. In this paper, we revisit the usage of data augmentation in model compression and give a comprehensive study on the relation between model sizes and their optimal data augmentation policy. To sum up, we mainly have the following three observations: (A) Models in different sizes prefer data augmentation with different magnitudes. Hence, in iterative pruning, data augmentation with varying magnitudes leads to better performance than data augmentation with a consistent magnitude. (B) Data augmentation with a high magnitude may significantly improve the performance of large models but harm the performance of small models. Fortunately, small models can still benefit from strong data augmentations by firstly learning them with "additional parameters" and then discard these "additional parameters" during inference. (C) The prediction of a pre-trained large model can be utilized to measure the difficulty of data augmentation. Thus it can be utilized as a criterion to design better data augmentation policies. We hope this paper may promote more research on the usage of data augmentation in model compression. | ['Kaisheng Ma', 'Linfeng Zhang', 'Muzhou Yu'] | 2023-05-22 | null | null | null | null | ['model-compression'] | ['methodology'] | [ 2.88579553e-01 1.26530960e-01 -4.84937340e-01 -4.07531470e-01
2.58440226e-02 6.14963211e-02 3.25887948e-01 1.77411377e-01
-6.12118602e-01 6.73637092e-01 7.19938427e-02 -4.08451915e-01
-2.49703247e-02 -9.42101002e-01 -7.25930393e-01 -7.91490495e-01
2.23019019e-01 2.46958867e-01 2.42144749e-01 -1.94449052e-01
2.26711154e-01 5.60051978e-01 -1.61870158e+00 1.90620720e-01
8.12022746e-01 1.25061500e+00 5.56551099e-01 2.42899254e-01
-2.42676690e-01 7.44628072e-01 -6.79124832e-01 -3.27896595e-01
3.58723670e-01 -3.01855266e-01 -5.45470417e-01 -2.27976054e-01
-1.17046956e-03 -6.82956517e-01 -7.27454901e-01 1.21225667e+00
4.54663306e-01 -2.05307584e-02 4.27903563e-01 -1.03756166e+00
-4.04540271e-01 8.63534868e-01 -3.71550888e-01 3.81529421e-01
-3.47058952e-01 5.54394461e-02 5.81023097e-01 -5.74599802e-01
2.15022177e-01 8.93957973e-01 6.02327228e-01 5.82913578e-01
-8.36639404e-01 -9.02427495e-01 1.53917417e-01 3.27364802e-01
-1.52395773e+00 -6.18168950e-01 9.52230036e-01 3.78933619e-03
7.87196457e-01 3.84633631e-01 8.94793451e-01 7.64473975e-01
-2.61722505e-01 7.46962070e-01 6.78547621e-01 -5.30345082e-01
2.71017462e-01 2.49301687e-01 1.00451395e-01 5.97646177e-01
5.52761555e-01 4.15360779e-02 -3.48836541e-01 1.21417306e-01
9.19903874e-01 5.61108030e-02 -1.79224223e-01 1.16676666e-01
-7.94024348e-01 8.48506987e-01 5.62633395e-01 4.29628968e-01
-3.42990339e-01 1.77653536e-01 5.53096294e-01 3.11341405e-01
2.34900355e-01 5.67252815e-01 -6.85931444e-01 -2.01953769e-01
-9.62153256e-01 3.07210356e-01 5.13631463e-01 1.12156534e+00
6.31260991e-01 4.44293320e-01 -9.33307558e-02 1.07241750e+00
6.39839321e-02 1.20665126e-01 9.08144534e-01 -8.52857530e-01
6.78971350e-01 7.47016966e-01 -3.33935231e-01 -9.54801440e-01
-3.29258651e-01 -6.22336686e-01 -1.42393482e+00 -1.47969425e-01
1.69169605e-01 -1.74309850e-01 -8.86541128e-01 1.67461598e+00
3.58233154e-02 1.69526741e-01 -1.78512428e-02 5.67817211e-01
8.02023113e-01 6.59503043e-01 7.34578744e-02 -4.79460657e-01
1.22689569e+00 -9.86246526e-01 -9.40900862e-01 -2.71837115e-01
1.06661952e+00 -6.23131692e-01 1.15649605e+00 5.57820559e-01
-1.22608757e+00 -7.50253320e-01 -1.27931952e+00 -4.82316688e-02
-3.41103941e-01 3.05200964e-01 8.52179050e-01 7.31995702e-01
-8.33568931e-01 8.22258174e-01 -7.89294302e-01 8.43041241e-02
7.28809476e-01 5.55816472e-01 3.75021026e-02 -2.35885084e-01
-1.34172833e+00 8.68607879e-01 8.61722469e-01 9.41912904e-02
-4.70176995e-01 -6.62487388e-01 -5.48745930e-01 3.20206106e-01
3.46463025e-01 -5.66528857e-01 1.01848722e+00 -7.17955232e-01
-1.10225356e+00 2.42515862e-01 -1.52781278e-01 -8.29323828e-01
2.70338148e-01 -3.05827796e-01 -4.62257892e-01 2.64891563e-03
-4.77819830e-01 7.97225595e-01 6.35222018e-01 -1.03131759e+00
-6.32143080e-01 -2.70338506e-01 -5.87079898e-02 1.74100325e-01
-1.05737746e+00 -1.75108567e-01 -6.14367723e-01 -8.88405740e-01
3.24506849e-01 -6.37180269e-01 -4.53131318e-01 4.72051650e-02
-2.76893646e-01 -1.12730324e-01 9.73801136e-01 -5.63286006e-01
1.75472605e+00 -2.14269233e+00 -7.10054860e-02 4.45403792e-02
2.36755639e-01 7.62384415e-01 -1.51801154e-01 4.09645624e-02
6.44392744e-02 4.34638858e-01 -1.43968657e-01 -2.82587141e-01
-3.75616670e-01 6.74444020e-01 -2.57139862e-01 -1.72915042e-03
2.25971028e-01 7.10577428e-01 -4.83303070e-01 -6.20913863e-01
1.20093361e-01 5.15024841e-01 -6.81834102e-01 8.93692523e-02
-9.43477228e-02 2.82114167e-02 -5.02808154e-01 5.57453811e-01
7.43373394e-01 -1.25930712e-01 -7.27761537e-02 -4.99026656e-01
1.76551174e-02 3.38808775e-01 -1.15010810e+00 1.26323533e+00
-1.71131492e-01 5.35842657e-01 -8.01257789e-02 -1.42431450e+00
1.16067064e+00 2.56255418e-01 3.58129978e-01 -7.82882690e-01
3.71701866e-01 1.45448655e-01 2.96286285e-01 -5.01257062e-01
5.87328017e-01 -4.87313904e-02 4.04792011e-01 9.86256078e-02
-1.30854368e-01 7.27205258e-03 3.82497728e-01 5.46484105e-02
9.26354468e-01 -3.89792502e-01 1.82954893e-01 6.39938377e-03
3.09889883e-01 -2.80667126e-01 7.62025118e-01 6.41662240e-01
-4.24656533e-02 3.44208926e-01 5.03919959e-01 -5.48336923e-01
-1.38439357e+00 -3.05603355e-01 -3.57172608e-01 9.27279294e-01
-4.67813946e-02 -5.86166620e-01 -7.66093254e-01 -3.70229661e-01
-2.40243480e-01 6.78956568e-01 -4.08686906e-01 -6.11785114e-01
-6.88208938e-01 -1.12782598e+00 7.90598452e-01 8.41154158e-01
9.02311206e-01 -9.18814540e-01 -4.36587304e-01 -3.30346785e-02
-6.93153217e-02 -1.13496459e+00 -8.69219750e-02 4.71400857e-01
-1.66936803e+00 -7.33078599e-01 -5.11142790e-01 -8.39264154e-01
8.79497111e-01 1.67852923e-01 7.83930302e-01 6.62718654e-01
2.47637987e-01 -5.33931077e-01 -5.79369843e-01 -8.36514890e-01
-3.44928980e-01 3.70497644e-01 3.41856718e-01 -4.68874276e-01
3.18063885e-01 -8.06103826e-01 -4.37073797e-01 1.23794660e-01
-1.17327130e+00 3.37051123e-01 9.05896664e-01 7.30820537e-01
6.87543869e-01 5.33961594e-01 6.48048401e-01 -9.63807762e-01
8.29564333e-01 -2.21388012e-01 -4.50093895e-01 6.24684393e-02
-9.68093514e-01 3.32061768e-01 1.03044343e+00 -7.10785687e-01
-7.81673849e-01 -1.60868347e-01 -3.66931379e-01 -7.29131103e-01
3.76927368e-02 6.70617044e-01 -3.29648942e-01 -8.60906243e-02
5.27637124e-01 2.90848792e-01 -1.50951836e-02 -7.10085869e-01
9.27276909e-02 5.19331574e-01 2.87581384e-01 -3.81451994e-01
5.30614197e-01 6.43389672e-02 1.55084759e-01 -7.92550087e-01
-6.03155255e-01 4.06881347e-02 -5.41707873e-01 1.89104930e-01
3.43780816e-01 -7.74112523e-01 -3.04537624e-01 4.54422742e-01
-9.80406106e-01 -3.12384665e-01 -4.87776786e-01 5.56508720e-01
-1.37131974e-01 4.58237827e-01 -7.30977058e-01 -6.90189958e-01
-4.13243979e-01 -9.92407858e-01 3.44271272e-01 3.19443285e-01
3.74184474e-02 -7.30892777e-01 -5.28294265e-01 2.01666877e-01
4.79754329e-01 -3.31649601e-01 1.22949696e+00 -8.74383092e-01
-4.17730659e-01 -3.91335428e-01 -3.84496689e-01 7.50758052e-01
-3.19462381e-02 -1.05967380e-01 -9.18554485e-01 -1.16462357e-01
3.47696096e-01 -2.63879031e-01 8.61439943e-01 4.47819442e-01
1.95577073e+00 -6.73288167e-01 -3.07171971e-01 8.57002258e-01
1.26090252e+00 4.30661142e-01 9.37470794e-01 2.35525131e-01
6.81357622e-01 2.91281343e-01 5.01839638e-01 5.13396144e-01
-5.62064797e-02 3.92174155e-01 5.59415877e-01 -1.16745114e-01
-7.45829642e-02 -2.64660656e-01 -9.35535431e-02 1.27414787e+00
-3.71449053e-01 -1.03334062e-01 -6.98209107e-01 2.85085976e-01
-1.54729199e+00 -6.30867422e-01 -2.04327442e-02 2.26076269e+00
1.08597755e+00 4.81887549e-01 -2.45886609e-01 7.41387844e-01
4.83274221e-01 9.04309824e-02 -5.76825082e-01 -3.23324859e-01
-1.72392935e-01 1.74917921e-01 5.72420895e-01 7.09416568e-02
-8.02913189e-01 7.51262665e-01 6.57841825e+00 1.34659111e+00
-1.13567317e+00 -4.34215665e-02 9.11071002e-01 -3.04185361e-01
-2.22026572e-01 -1.13876529e-01 -1.14886272e+00 7.43082166e-01
1.04769349e+00 -1.40297525e-02 4.32582438e-01 9.66790676e-01
1.44457087e-01 -8.43734946e-03 -1.00307643e+00 9.61949825e-01
-2.61959076e-01 -1.44113874e+00 4.76084113e-01 1.81898504e-01
6.75958157e-01 -1.56460792e-01 -1.30436689e-01 4.37840164e-01
-2.25848764e-01 -1.01525986e+00 3.04253876e-01 3.79232109e-01
6.63534999e-01 -9.49870944e-01 9.99245107e-01 6.54223800e-01
-9.27244365e-01 -3.31346840e-01 -9.47937250e-01 -2.85767794e-01
5.10174260e-02 8.26987326e-01 -5.77325106e-01 2.60729998e-01
5.31682074e-01 3.77312779e-01 -6.06367886e-01 1.07459688e+00
-7.16173053e-02 7.76735842e-01 -5.87297380e-01 -5.46731465e-02
9.65420995e-03 -3.20730478e-01 1.23618409e-01 9.52488065e-01
3.80314291e-01 4.04536933e-01 -2.56132156e-01 5.61397076e-01
-3.56144428e-01 1.61828473e-01 -4.34673667e-01 -2.99297303e-01
8.09743702e-01 8.38850260e-01 -5.40301383e-01 -6.84013188e-01
-4.87129569e-01 4.47264165e-01 3.70831668e-01 2.15900809e-01
-7.10651398e-01 -2.53413945e-01 2.89047390e-01 3.04292232e-01
1.98797882e-01 -1.50038555e-01 -8.63175213e-01 -8.82814944e-01
6.68974072e-02 -6.96547449e-01 2.19457388e-01 -6.21678948e-01
-7.06411004e-01 6.13136470e-01 3.30144204e-02 -1.13045228e+00
-2.56733969e-03 -4.22483623e-01 -5.04471064e-01 6.75074935e-01
-1.44054615e+00 -6.88190043e-01 -3.21058482e-01 3.40929478e-01
5.32358050e-01 -3.89339507e-01 8.12428057e-01 6.24992549e-01
-8.24213684e-01 9.35083449e-01 1.57089785e-01 1.09635897e-01
2.04693735e-01 -5.42467952e-01 6.51514977e-02 7.45457709e-01
5.42901643e-02 5.74539959e-01 6.53569400e-01 -5.70316970e-01
-1.13521659e+00 -1.11803234e+00 8.88345659e-01 1.65425003e-01
2.14533553e-01 8.17229152e-02 -1.34976339e+00 4.53328490e-01
-2.18457893e-01 -5.62111055e-03 5.51118970e-01 1.40963554e-01
1.35010434e-02 -4.34906870e-01 -1.09648955e+00 7.23980486e-01
9.79431570e-01 -1.12389065e-01 -4.23566461e-01 2.84792811e-01
9.26894248e-01 -3.10578257e-01 -9.18164611e-01 7.98574865e-01
3.79532278e-01 -8.09046924e-01 9.43585932e-01 -7.27741599e-01
7.82973289e-01 1.93297371e-01 -1.52658537e-01 -9.24085140e-01
-2.10565135e-01 -7.50225782e-02 -7.15510488e-01 1.20195138e+00
2.96709388e-01 -4.56222802e-01 1.12165284e+00 8.22425723e-01
-1.86696231e-01 -1.36470938e+00 -7.81035542e-01 -6.78499401e-01
1.09491140e-01 -5.44704199e-01 8.15088928e-01 9.68828142e-01
-1.38433456e-01 1.59257218e-01 -5.22370458e-01 -1.16908453e-01
2.14991644e-01 -4.03235525e-01 4.33521688e-01 -1.23603427e+00
-1.17051110e-01 -4.86654401e-01 -2.52231926e-01 -1.39237392e+00
-2.37965524e-01 -5.77627957e-01 -3.17706019e-01 -1.31603086e+00
7.13564306e-02 -9.83868301e-01 -3.58939558e-01 5.91105640e-01
-1.37689501e-01 1.58517107e-01 4.32608239e-02 4.68109667e-01
-6.82323053e-02 7.34208405e-01 1.38809574e+00 3.95796746e-02
-3.17858040e-01 2.08275974e-01 -9.00930285e-01 8.49114895e-01
1.07112741e+00 -4.46584195e-01 -6.73085988e-01 -5.92533231e-01
3.16567659e-01 -5.85277230e-02 1.78987645e-02 -1.12401235e+00
3.31868172e-01 -1.16969578e-01 5.37134230e-01 -5.80554366e-01
4.04867142e-01 -9.70409095e-01 -1.01197295e-01 6.43132389e-01
-3.81145447e-01 2.59339008e-02 3.39114547e-01 3.87480140e-01
-2.89667308e-01 -8.40982616e-01 7.94165552e-01 -1.49396017e-01
-5.65171421e-01 3.67874205e-01 -1.97199613e-01 -1.14882804e-01
6.35054231e-01 -3.74006420e-01 -6.90824538e-02 -2.13832140e-01
-4.96924192e-01 5.60516641e-02 2.45079115e-01 2.03968674e-01
6.74642801e-01 -1.34913790e+00 -3.80092740e-01 5.86684465e-01
-3.29202622e-01 5.57233095e-01 1.34064168e-01 7.21621454e-01
-5.02967656e-01 5.44776559e-01 -2.42077857e-01 -1.99135423e-01
-1.30803978e+00 6.59539819e-01 7.93635026e-02 -2.63567805e-01
-4.75956440e-01 9.07011569e-01 1.03478506e-02 3.92706133e-02
6.06463432e-01 -4.92202699e-01 -3.90609086e-01 -2.23695114e-02
6.35720253e-01 4.20390099e-01 2.50054896e-01 -2.07667485e-01
2.02008933e-01 1.34705052e-01 -5.00071824e-01 4.79837418e-01
1.26894081e+00 1.09071016e-01 5.84469847e-02 1.39694229e-01
1.02219641e+00 -4.35620278e-01 -1.11178136e+00 -3.16773564e-01
-4.25178826e-01 -4.84993488e-01 4.18435812e-01 -4.26280767e-01
-1.47443557e+00 1.05188143e+00 5.01216948e-01 2.60696560e-01
1.54825163e+00 -1.91791043e-01 7.92351305e-01 6.49183512e-01
1.96902946e-01 -1.30967712e+00 1.74051095e-02 4.58362132e-01
7.11635768e-01 -9.80394423e-01 3.38919133e-01 -4.96943206e-01
-4.72998947e-01 9.65754569e-01 8.59962463e-01 5.37461340e-02
6.64859414e-01 4.63550568e-01 -3.44293058e-01 5.39158098e-03
-7.23643243e-01 5.58799021e-02 1.32099733e-01 4.79923129e-01
3.43505859e-01 -2.03289971e-01 -5.58358312e-01 9.03925717e-01
-4.01216894e-01 2.96608478e-01 3.06885839e-01 7.91543245e-01
-6.71921670e-01 -1.17916000e+00 -1.62743062e-01 1.14856362e+00
-3.82678241e-01 -3.84770274e-01 -3.73474471e-02 7.84128308e-01
4.50599164e-01 6.67211771e-01 1.58981800e-01 -6.26023531e-01
3.01089317e-01 1.66232903e-02 3.85517716e-01 -3.59244674e-01
-3.47746879e-01 -1.55781537e-01 5.87115623e-02 -2.34040424e-01
-2.03108102e-01 -2.94207513e-01 -1.18834639e+00 -6.04047298e-01
-6.84509993e-01 5.98477833e-02 6.18728995e-01 9.53755260e-01
2.36847907e-01 7.33683407e-01 2.53840774e-01 -6.20001376e-01
-7.51056433e-01 -1.25642610e+00 -6.61627412e-01 2.31679648e-01
-8.29708576e-02 -5.13914406e-01 -3.20749879e-01 1.38078872e-02] | [8.544032096862793, 3.117164134979248] |
e2ad35e4-6219-41ad-ab9d-21370aa7c226 | analyzing-the-impact-of-climate-change-on | 2302.01887 | null | https://arxiv.org/abs/2302.01887v2 | https://arxiv.org/pdf/2302.01887v2.pdf | Analyzing the impact of climate change on critical infrastructure from the scientific literature: A weakly supervised NLP approach | Natural language processing (NLP) is a promising approach for analyzing large volumes of climate-change and infrastructure-related scientific literature. However, best-in-practice NLP techniques require large collections of relevant documents (corpus). Furthermore, NLP techniques using machine learning and deep learning techniques require labels grouping the articles based on user-defined criteria for a significant subset of a corpus in order to train the supervised model. Even labeling a few hundred documents with human subject-matter experts is a time-consuming process. To expedite this process, we developed a weak supervision-based NLP approach that leverages semantic similarity between categories and documents to (i) establish a topic-specific corpus by subsetting a large-scale open-access corpus and (ii) generate category labels for the topic-specific corpus. In comparison with a months-long process of subject-matter expert labeling, we assign category labels to the whole corpus using weak supervision and supervised learning in about 13 hours. The labeled climate and NCF corpus enable targeted, efficient identification of documents discussing a topic (or combination of topics) of interest and identification of various effects of climate change on critical infrastructure, improving the usability of scientific literature and ultimately supporting enhanced policy and decision making. To demonstrate this capability, we conduct topic modeling on pairs of climate hazards and NCFs to discover trending topics at the intersection of these categories. This method is useful for analysts and decision-makers to quickly grasp the relevant topics and most important documents linked to the topic. | ['Prasanna Balaprakash', 'Leslie-Anne Levy', 'John K Hutchison', 'Duane R. Verner', 'Joshua David Bergerson', 'Tanwi Mallick'] | 2023-02-03 | null | null | null | null | ['semantic-textual-similarity'] | ['natural-language-processing'] | [ 1.85668021e-01 4.44450676e-02 -3.09406906e-01 -3.88563901e-01
-1.03272021e+00 -1.03258717e+00 8.31681430e-01 9.97571826e-01
-3.38172495e-01 6.22713506e-01 4.46761936e-01 -1.01821125e+00
-1.24823473e-01 -1.12368369e+00 -7.57445395e-01 -5.56398809e-01
-3.72314937e-02 7.13233232e-01 -8.86863917e-02 1.46873727e-01
5.50112307e-01 4.87221539e-01 -1.42310607e+00 2.06957817e-01
1.12052667e+00 6.71937466e-01 6.13134444e-01 1.18680798e-01
-9.50487077e-01 3.41103613e-01 -6.95727766e-01 1.65206909e-01
4.62948754e-02 -6.24989383e-02 -8.77969205e-01 -1.92816317e-01
3.02421689e-01 1.64469123e-01 3.37391645e-01 9.18585122e-01
2.00045392e-01 2.28160799e-01 7.51527250e-01 -9.35280800e-01
-2.61704832e-01 5.44961214e-01 -6.61791980e-01 4.93913531e-01
1.34203598e-01 -1.52314052e-01 1.09772980e+00 -7.41236031e-01
5.30093312e-01 1.07567310e+00 3.83041710e-01 -1.18831784e-01
-9.02298033e-01 -9.31927085e-01 4.28897947e-01 1.38512790e-01
-1.18796730e+00 -2.90288702e-02 8.53844523e-01 -9.54054534e-01
8.46440673e-01 9.30929780e-02 5.04803360e-01 6.89043880e-01
1.73655763e-01 2.57225275e-01 1.01380992e+00 -4.52006668e-01
6.07169151e-01 2.94899255e-01 6.03575230e-01 1.34389624e-01
1.86450884e-01 -4.36916262e-01 -3.63401800e-01 -3.72203827e-01
1.00257523e-01 2.08062023e-01 -5.88139594e-02 2.34823436e-01
-1.15136373e+00 9.79853809e-01 2.68761396e-01 4.48445916e-01
-7.01701462e-01 -9.40418392e-02 4.45930898e-01 -1.71303824e-02
9.09554124e-01 6.11580193e-01 -6.10263944e-01 1.49460793e-01
-1.28765941e+00 3.25054795e-01 8.72368276e-01 6.58952236e-01
1.16492617e+00 -3.83855641e-01 -8.97262171e-02 6.99701726e-01
3.04889470e-01 7.22032666e-01 3.58101279e-02 -6.72713459e-01
6.02763593e-01 7.07641602e-01 2.33416423e-01 -1.21571589e+00
-6.07081950e-01 -3.67562264e-01 -6.13189697e-01 -1.86391979e-01
3.56771886e-01 -2.73513824e-01 -9.63873208e-01 1.49008751e+00
4.96226639e-01 2.68812906e-02 -4.90378290e-02 7.73457825e-01
6.42084479e-01 1.30630589e+00 6.24234080e-01 -2.58500338e-01
1.72361445e+00 -4.51292336e-01 -6.96316123e-01 -2.10094184e-01
3.86794388e-01 -6.85225010e-01 1.21985149e+00 -7.71810785e-02
-4.23130214e-01 -2.02751428e-01 -6.39590561e-01 7.10741282e-02
-1.04056454e+00 -1.47780687e-01 4.48442012e-01 2.01596767e-01
-7.50911295e-01 2.38565087e-01 -4.04272407e-01 -5.47996163e-01
3.78001541e-01 4.10433486e-02 3.16182971e-02 -4.68058586e-02
-1.52229822e+00 8.97573411e-01 3.55085194e-01 -1.16985582e-01
-9.69983220e-01 -1.23020434e+00 -6.94544971e-01 4.85559821e-01
4.78085071e-01 -2.63677180e-01 1.04995441e+00 -2.95727044e-01
-8.55537593e-01 7.92395592e-01 -4.98523593e-01 -3.55181545e-01
-7.62564763e-02 -1.68881297e-01 -3.92948151e-01 4.53021944e-01
7.42189944e-01 4.89277363e-01 4.49377090e-01 -1.20726395e+00
-9.53789353e-01 -3.79575968e-01 2.41584927e-02 1.59768000e-01
-6.43724561e-01 5.27766466e-01 -2.15723455e-01 -5.53715467e-01
-1.34455068e-02 -6.49558127e-01 -1.33421421e-01 -4.77152050e-01
-4.09004092e-01 -7.20665753e-01 1.09636664e+00 -9.91294801e-01
1.07430577e+00 -1.94008911e+00 -4.21052814e-01 4.27301615e-01
2.52375424e-01 -1.23678267e-01 -1.80074219e-02 5.63718498e-01
6.63908646e-02 5.63182116e-01 -2.30608135e-01 -1.12636050e-03
-1.30329812e-02 -4.95131500e-02 -9.55183387e-01 3.98614854e-01
1.46324098e-01 4.45540935e-01 -1.10078990e+00 -5.24382114e-01
1.76721424e-01 2.10018143e-01 -1.86769590e-01 4.68006991e-02
-6.10653698e-01 5.37690580e-01 -6.85760438e-01 5.25803447e-01
4.23891366e-01 -3.34004641e-01 1.33233696e-01 -4.96297292e-02
-4.29327250e-01 7.53574431e-01 -7.38321364e-01 1.05961645e+00
-7.43828535e-01 9.71011281e-01 -9.68440175e-02 -1.27146542e+00
1.02413940e+00 3.70851874e-01 7.50347197e-01 -5.93375921e-01
-2.68184483e-01 6.89970925e-02 -4.58483487e-01 -4.74700540e-01
4.15753722e-01 -3.12023818e-01 -3.05730999e-01 1.01128054e+00
-2.19973430e-01 -2.47845590e-01 2.46196315e-01 2.10637331e-01
8.66393447e-01 -3.85764420e-01 5.45518100e-02 -6.87599242e-01
3.96364361e-01 4.78960931e-01 5.35523772e-01 7.03043044e-01
-1.22899441e-02 3.32353115e-01 5.24087667e-01 -6.33406520e-01
-1.01971030e+00 -7.37939119e-01 -2.92112947e-01 1.23595965e+00
-1.38631970e-01 -2.88594842e-01 -3.99244040e-01 -6.42136216e-01
-1.10382391e-02 1.21616256e+00 -5.10487616e-01 3.81583482e-01
-5.02741873e-01 -8.28439653e-01 2.01410681e-01 2.38564685e-01
3.17630947e-01 -1.05336618e+00 -4.04664665e-01 1.51718140e-01
-5.41539311e-01 -9.25293267e-01 -2.76934117e-01 3.25116396e-01
-3.77868563e-01 -1.11976004e+00 -6.36951268e-01 -6.92374885e-01
8.01475525e-01 4.23687041e-01 1.06737721e+00 -3.93534213e-01
1.12206683e-01 1.66178599e-01 -1.52616009e-01 -8.97261143e-01
-4.49279934e-01 3.34363788e-01 6.98152930e-03 -4.17849153e-01
7.71912396e-01 -4.04348969e-01 -4.56319928e-01 1.44178465e-01
-8.11662734e-01 -1.54563099e-01 2.54388839e-01 3.23946536e-01
4.55517471e-01 5.29955566e-01 9.43102121e-01 -7.22628236e-01
7.19149053e-01 -1.00083303e+00 -8.99017632e-01 3.60613555e-01
-7.19417512e-01 -1.41618982e-01 6.30689323e-01 -2.25864172e-01
-1.22736931e+00 -2.56979555e-01 2.02306360e-01 1.10614197e-02
-4.31624919e-01 1.25399029e+00 -3.29180248e-02 5.37335515e-01
6.47897005e-01 1.90749168e-01 -2.72507399e-01 -5.82804620e-01
2.78292537e-01 9.79046226e-01 5.31727791e-01 -8.06219339e-01
8.85727286e-01 4.93025392e-01 -2.71805316e-01 -7.75162399e-01
-1.25842559e+00 -1.03500271e+00 -4.71304864e-01 -3.59370738e-01
8.94451797e-01 -1.14313340e+00 -3.46494883e-01 -2.50254702e-02
-1.36305332e+00 -2.31005207e-01 -4.58942726e-02 4.78921860e-01
1.00408055e-01 5.48988394e-02 -1.55858293e-01 -6.20521843e-01
-4.92045820e-01 -7.64672160e-01 1.01455629e+00 2.74060041e-01
-3.73146683e-01 -1.20428729e+00 2.35173166e-01 5.11860549e-01
4.32565153e-01 1.48967579e-01 1.33607423e+00 -8.74204218e-01
-2.00332567e-01 -2.29154766e-01 -4.04864997e-01 -8.14392567e-02
3.89632642e-01 -6.43379688e-02 -9.23106194e-01 -8.33622292e-02
-7.48369889e-03 5.76927364e-02 6.93221927e-01 4.10542309e-01
1.09298897e+00 -4.62786078e-01 -4.85032082e-01 4.37654043e-03
1.19585836e+00 3.56376737e-01 2.10512131e-01 6.68057323e-01
6.20794356e-01 1.23814082e+00 5.89810967e-01 3.74246716e-01
6.97818220e-01 3.80761117e-01 -1.11555181e-01 -2.85954654e-01
2.91727871e-01 -1.00481227e-01 1.81280613e-01 4.43470389e-01
3.12822402e-01 -4.14811634e-02 -1.69119895e+00 1.08168864e+00
-1.46800518e+00 -1.04430950e+00 -2.93186288e-02 1.96009481e+00
1.17234612e+00 5.73171899e-02 -3.71985212e-02 -1.99653327e-01
9.67104137e-01 1.39894873e-01 -4.55795735e-01 -1.12114601e-01
5.18455915e-03 -6.78680018e-02 4.32899445e-01 3.43566477e-01
-1.13596547e+00 9.93366957e-01 5.99088287e+00 7.52188683e-01
-1.29452443e+00 2.36511678e-02 8.22371960e-01 2.13904437e-02
-6.38159156e-01 2.61266738e-01 -9.92296040e-01 5.38333178e-01
1.32416439e+00 -5.49801409e-01 1.61198869e-01 8.28300893e-01
9.83882725e-01 -1.83651209e-01 -7.50341713e-01 3.28067064e-01
-2.89407909e-01 -1.81055653e+00 -3.33078988e-02 1.33563176e-01
8.41729820e-01 3.16133082e-01 -2.96443462e-01 2.73957193e-01
4.76760805e-01 -7.89780796e-01 6.27138019e-01 2.25645974e-01
6.46752179e-01 -6.19781375e-01 7.49065101e-01 2.89314359e-01
-1.13979483e+00 -5.61756827e-02 -4.80979592e-01 -3.46144401e-02
1.41182736e-01 1.31774116e+00 -8.78095686e-01 3.81263345e-01
1.04699314e+00 5.54233134e-01 -2.69784182e-01 8.13825965e-01
-2.97879338e-01 1.29009974e+00 -6.15208626e-01 7.87704289e-02
5.23574889e-01 -1.98134094e-01 4.94347751e-01 1.35006142e+00
1.60730615e-01 1.36334524e-01 3.24802935e-01 1.10012269e+00
-5.95070273e-02 2.59440511e-01 -5.45412242e-01 -3.11615795e-01
9.17307794e-01 1.29778409e+00 -1.11201811e+00 -5.11480570e-01
-4.12494540e-01 -7.79280290e-02 9.21099260e-03 5.44224977e-01
-5.50570607e-01 -4.67190206e-01 3.69626433e-01 2.90237904e-01
1.40156448e-01 -2.40098253e-01 -7.63353109e-01 -7.56154716e-01
-9.49038565e-03 -6.54679775e-01 3.67000490e-01 -5.46248317e-01
-1.25530767e+00 2.29104921e-01 2.71027297e-01 -1.06136429e+00
-1.51157647e-01 -6.22656271e-02 -1.06588960e+00 1.17701268e+00
-1.79378843e+00 -1.03748775e+00 -2.58515120e-01 1.46585181e-01
4.45359141e-01 -8.21010768e-02 6.27496362e-01 1.14303373e-01
-5.34430444e-01 -2.19390720e-01 2.77194142e-01 8.42834543e-03
7.34030008e-01 -1.25173759e+00 4.13188964e-01 7.03740835e-01
-2.13787720e-01 9.12682116e-01 8.37543011e-01 -9.72242832e-01
-8.87486815e-01 -1.46331692e+00 1.33273602e+00 -2.77263641e-01
9.46727633e-01 -4.95951831e-01 -1.09749389e+00 3.40036750e-01
2.96620756e-01 -5.78615785e-01 1.10640526e+00 5.25903225e-01
-2.54965723e-01 -1.44853339e-01 -8.89849126e-01 4.77965295e-01
2.91207999e-01 -7.75358140e-01 -9.20556068e-01 9.82972383e-01
9.82623219e-01 1.60445392e-01 -7.19912291e-01 6.68528676e-02
1.86153218e-01 -3.18089239e-02 7.26112843e-01 -4.33546036e-01
5.09209454e-01 -4.54158157e-01 6.23235367e-02 -1.15921402e+00
-1.73084140e-01 -5.05851507e-01 3.00472617e-01 1.42761612e+00
6.24707699e-01 -4.34994400e-01 5.50072074e-01 8.20296824e-01
-1.77371338e-01 -3.91736329e-01 -5.88517904e-01 -6.17015898e-01
4.93404895e-01 -5.67306995e-01 6.88733757e-01 1.28010178e+00
3.26938987e-01 2.93279052e-01 2.05947682e-01 4.94796902e-01
5.83387911e-01 5.97263575e-01 5.62444150e-01 -1.52025509e+00
4.84948874e-01 -3.83029252e-01 3.92347217e-01 -4.54459369e-01
5.34847498e-01 -8.10267389e-01 2.48701006e-01 -1.89810574e+00
2.09657699e-01 -7.19952881e-01 -2.51274198e-01 9.34338570e-01
-2.88927794e-01 -2.17790797e-01 -7.11098239e-02 5.20452499e-01
-4.01286930e-01 2.62980938e-01 5.88068247e-01 -4.98763740e-01
-4.09230411e-01 2.68442277e-02 -1.02294564e+00 6.40607178e-01
8.77085924e-01 -6.56664491e-01 -2.70211905e-01 -2.27008179e-01
3.53224814e-01 -3.94899607e-01 2.39562556e-01 -7.22121119e-01
4.63558704e-01 -7.03327000e-01 1.03685737e-01 -1.10962129e+00
-5.38190007e-01 -6.59225881e-01 -5.32550365e-02 1.34494394e-01
-5.20746946e-01 -2.24855214e-01 4.94895756e-01 3.92608583e-01
-4.28517818e-01 -1.97802737e-01 3.94193918e-01 -1.38305619e-01
-7.05168962e-01 2.92607918e-02 -9.12460268e-01 4.73450217e-03
8.91075671e-01 2.14866564e-01 -5.40858448e-01 -4.06656116e-01
-4.05396372e-01 6.52935505e-01 2.48565376e-01 3.96939337e-01
1.99173883e-01 -7.39620745e-01 -7.97690988e-01 -2.67148852e-01
1.39925390e-01 3.85377258e-01 2.22223952e-01 4.05570984e-01
-4.18523252e-01 8.55418146e-01 2.04602033e-01 -3.97808671e-01
-7.49046922e-01 4.80884492e-01 -1.18494630e-01 -4.11361247e-01
-5.97842753e-01 3.63792807e-01 4.81052846e-01 -6.15650713e-01
1.72297835e-01 -1.99559867e-01 -5.79087973e-01 5.53033710e-01
6.78736031e-01 1.74208701e-01 4.20529582e-02 -3.98160100e-01
-4.54897225e-01 3.40876341e-01 -9.71685350e-02 -2.21910164e-01
1.62731290e+00 -1.93516538e-01 -4.23855841e-01 3.89887124e-01
1.07306147e+00 3.73547967e-03 -1.02670002e+00 -3.64130229e-01
4.99653786e-01 7.24261105e-02 4.01648402e-01 -9.91809607e-01
-8.67914736e-01 9.17872369e-01 2.03874305e-01 2.87763476e-01
9.33854818e-01 2.58974820e-01 5.95844090e-01 5.62756956e-01
-1.69227123e-02 -1.26716948e+00 -2.89366633e-01 4.30145770e-01
9.23108339e-01 -1.23519635e+00 4.96590436e-02 -3.58240932e-01
-2.29965568e-01 1.06090534e+00 4.02776510e-01 4.58935052e-01
8.94575477e-01 2.33117953e-01 2.55329192e-01 -6.43018961e-01
-5.55952311e-01 -1.10816069e-01 3.85934025e-01 4.32324618e-01
3.00561309e-01 6.21282309e-02 -3.31491679e-01 5.13407171e-01
-2.34287649e-01 -2.16384530e-01 4.49539185e-01 8.43149245e-01
-6.77878261e-01 -7.20844030e-01 -7.48179674e-01 5.94178200e-01
-5.04463017e-01 -4.62359369e-01 -1.95494711e-01 3.13299507e-01
-6.82360530e-02 1.23762190e+00 3.57402503e-01 6.55184239e-02
-5.84393293e-02 1.69252038e-01 -6.00011289e-01 -9.49331820e-01
-5.97063124e-01 1.82794943e-01 8.77741575e-02 -1.29357591e-01
-3.89698416e-01 -8.40395749e-01 -1.44974101e+00 -1.82081699e-01
-1.76770732e-01 6.44197702e-01 1.10006249e+00 1.41161788e+00
5.68334758e-01 3.12358528e-01 7.53637075e-01 -8.08383346e-01
-6.95917010e-02 -1.09513175e+00 -4.63113457e-01 3.90374452e-01
1.55593440e-01 -5.40198863e-01 -4.38206822e-01 2.89748698e-01] | [10.328680038452148, 7.23297643661499] |
b1e16304-dd0a-4076-b744-510200129955 | fredom-fairness-domain-adaptation-approach-to | 2304.02135 | null | https://arxiv.org/abs/2304.02135v1 | https://arxiv.org/pdf/2304.02135v1.pdf | FREDOM: Fairness Domain Adaptation Approach to Semantic Scene Understanding | Although Domain Adaptation in Semantic Scene Segmentation has shown impressive improvement in recent years, the fairness concerns in the domain adaptation have yet to be well defined and addressed. In addition, fairness is one of the most critical aspects when deploying the segmentation models into human-related real-world applications, e.g., autonomous driving, as any unfair predictions could influence human safety. In this paper, we propose a novel Fairness Domain Adaptation (FREDOM) approach to semantic scene segmentation. In particular, from the proposed formulated fairness objective, a new adaptation framework will be introduced based on the fair treatment of class distributions. Moreover, to generally model the context of structural dependency, a new conditional structural constraint is introduced to impose the consistency of predicted segmentation. Thanks to the proposed Conditional Structure Network, the self-attention mechanism has sufficiently modeled the structural information of segmentation. Through the ablation studies, the proposed method has shown the performance improvement of the segmentation models and promoted fairness in the model predictions. The experimental results on the two standard benchmarks, i.e., SYNTHIA $\to$ Cityscapes and GTA5 $\to$ Cityscapes, have shown that our method achieved State-of-the-Art (SOTA) performance. | ['Khoa Luu', 'Jackson Cothren', 'Bhiksha Raj', 'Ngan Le', 'Thanh-Dat Truong'] | 2023-04-04 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Truong_FREDOM_Fairness_Domain_Adaptation_Approach_to_Semantic_Scene_Understanding_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Truong_FREDOM_Fairness_Domain_Adaptation_Approach_to_Semantic_Scene_Understanding_CVPR_2023_paper.pdf | cvpr-2023-1 | ['scene-segmentation'] | ['computer-vision'] | [ 1.19137824e-01 3.87343645e-01 -9.09288228e-02 -8.47293496e-01
-3.53680074e-01 1.31226435e-01 4.48521703e-01 -4.19568531e-02
-6.76276565e-01 8.13233614e-01 -1.21295363e-01 -1.13068543e-01
-1.58579439e-01 -6.85262620e-01 -4.73478615e-01 -7.37133861e-01
4.51540470e-01 3.46460223e-01 3.43631834e-01 -3.21329862e-01
3.16348195e-01 6.68795928e-02 -1.54666245e+00 -1.36305332e-01
1.57841992e+00 1.28994620e+00 1.48141205e-01 9.85272601e-02
-1.09335937e-01 8.09199214e-01 -4.64202166e-01 -8.19919527e-01
2.65771329e-01 -5.30377984e-01 -8.64170134e-01 2.67904717e-02
1.96099848e-01 -2.72544533e-01 -1.42675787e-01 1.39539576e+00
5.26470542e-01 5.13239026e-01 5.05363584e-01 -1.63220501e+00
-4.60606158e-01 5.87859035e-01 -7.41794288e-01 1.19746111e-01
-3.28675628e-01 2.91059285e-01 8.84285152e-01 -4.05417800e-01
2.46643066e-01 1.48670304e+00 2.60235637e-01 7.30850875e-01
-7.51609623e-01 -1.07008779e+00 5.52676320e-01 5.09257078e-01
-1.30442703e+00 -2.86267102e-01 7.78013408e-01 -4.54595059e-01
3.57005477e-01 2.03348190e-01 3.56845319e-01 9.57348526e-01
1.03848904e-01 7.97650635e-01 1.26511502e+00 -4.90305834e-02
4.13594604e-01 3.91696632e-01 3.85735542e-01 2.49972001e-01
2.43662521e-01 2.73118883e-01 -1.85042128e-01 2.92639166e-01
2.08484232e-01 -3.08604211e-01 1.24354668e-01 -4.18869525e-01
-6.62767828e-01 1.01359284e+00 6.52831733e-01 -1.38491347e-01
-1.40360028e-01 5.12818582e-02 7.52949357e-01 -2.38497898e-01
7.43385196e-01 9.45967808e-02 -1.46673799e-01 2.54452769e-02
-7.72708893e-01 3.23626578e-01 3.19689631e-01 1.14163709e+00
6.00019574e-01 1.20298944e-01 -4.90301609e-01 8.08508992e-01
3.80809814e-01 4.74370450e-01 3.57876003e-01 -1.19864285e+00
5.42789638e-01 4.59590822e-01 1.43663451e-01 -1.06898403e+00
-3.63833845e-01 -5.11377454e-01 -1.11911976e+00 3.57348830e-01
4.20215398e-01 -1.30037040e-01 -8.35713267e-01 1.96056914e+00
4.37135130e-01 2.17678443e-01 2.43327975e-01 1.27556908e+00
7.49206305e-01 3.34444523e-01 7.71970570e-01 -1.49212433e-02
1.24995792e+00 -1.04011703e+00 -7.63349771e-01 -2.86906332e-01
2.89668471e-01 -5.27996898e-01 1.09374130e+00 1.40299484e-01
-8.26104045e-01 -8.50253820e-01 -9.28322136e-01 -6.65225536e-02
-3.11784536e-01 -2.19363421e-01 5.22078633e-01 9.69529092e-01
-4.35510278e-01 3.73155683e-01 -5.97885847e-01 -3.96724015e-01
9.71576452e-01 1.00513175e-01 7.38854930e-02 -9.09372866e-02
-1.63557124e+00 8.83778512e-01 7.75302112e-01 5.16007468e-02
-9.44894075e-01 -5.73532760e-01 -8.19658399e-01 1.08161807e-01
4.45805520e-01 -5.70590794e-01 1.20190883e+00 -1.19319844e+00
-1.47271240e+00 9.47508633e-01 -7.86530524e-02 -7.79614866e-01
1.06249249e+00 -2.57566899e-01 -5.31145096e-01 -1.28603399e-01
3.98810446e-01 8.54073226e-01 5.67295074e-01 -1.39357662e+00
-1.11908782e+00 -3.10455948e-01 2.28411555e-01 4.42938447e-01
-2.13105366e-01 -9.26491693e-02 -2.53061563e-01 -4.52517986e-01
-3.22958410e-01 -8.55566502e-01 -4.33053762e-01 -8.99467394e-02
-4.50766385e-01 -2.16561973e-01 6.85284615e-01 -6.14696622e-01
1.22570610e+00 -2.43328404e+00 -2.66680479e-01 2.74980903e-01
1.23262100e-01 4.18436289e-01 1.58543035e-01 -3.50344628e-01
1.41359434e-01 2.93886781e-01 -6.94260776e-01 -3.06198120e-01
3.43518555e-01 2.43759587e-01 -1.00775383e-01 4.48558867e-01
1.53289773e-02 5.40216744e-01 -8.48048508e-01 -7.13825941e-01
4.86729294e-01 1.64111868e-01 -6.48252428e-01 3.32218438e-01
-2.31577486e-01 7.34788835e-01 -7.08059371e-01 1.81483567e-01
1.23693025e+00 2.50462711e-01 -3.41794431e-01 8.94747488e-03
4.31135185e-02 -2.01824993e-01 -1.02974653e+00 1.60068798e+00
-2.24869713e-01 5.19910455e-02 -7.66156912e-02 -1.18857288e+00
9.56717014e-01 -2.70743128e-02 3.40664804e-01 -1.28359735e+00
4.25834119e-01 1.68319210e-01 8.07899311e-02 -3.79346818e-01
7.31621861e-01 -2.69097388e-01 -4.03426558e-01 -1.13885738e-01
-3.17726821e-01 -2.92420059e-01 4.98627424e-02 7.24459365e-02
2.02201575e-01 8.75849053e-02 3.42751056e-01 -6.22834265e-01
8.04922044e-01 2.26843268e-01 1.09020674e+00 5.03598154e-01
-1.02663350e+00 4.95845765e-01 4.49208289e-01 -1.88434824e-01
-8.36351871e-01 -8.49607706e-01 -1.20409235e-01 1.07132256e+00
8.83067966e-01 1.39752716e-01 -1.35634732e+00 -8.10772598e-01
-1.62110344e-01 1.37919784e+00 -6.67122543e-01 -5.98642051e-01
-2.26300552e-01 -7.69223750e-01 7.39361644e-01 4.18624789e-01
1.31499147e+00 -1.00537169e+00 -8.20468605e-01 -1.48921618e-02
-3.27611566e-01 -1.36781526e+00 -4.87343788e-01 -1.10802136e-01
-5.63024879e-01 -9.86277163e-01 -6.67380929e-01 -5.27014375e-01
3.97210479e-01 1.16163142e-01 9.15289521e-01 -1.35669842e-01
-2.87270639e-02 -1.51720811e-02 -3.40385556e-01 -6.21667802e-01
-2.70652801e-01 -1.44071924e-02 -1.79087415e-01 3.38395208e-01
4.77775246e-01 -1.90754190e-01 -6.98461294e-01 6.38712347e-01
-8.35639596e-01 1.63527250e-01 2.34732941e-01 8.22732866e-01
6.24481499e-01 4.44889128e-01 6.88071489e-01 -1.29762053e+00
5.34377635e-01 -4.56409246e-01 -5.84584773e-01 1.33698627e-01
-8.68060231e-01 -1.72122434e-01 6.53731406e-01 8.15899577e-03
-1.84279060e+00 -1.17800325e-01 -1.51409194e-01 -2.01613188e-01
-4.11560416e-01 2.05608562e-01 -7.31719971e-01 2.64460146e-01
4.83319581e-01 -8.00273493e-02 -1.13520533e-01 -7.30703399e-02
5.55840850e-01 7.30899870e-01 4.87102807e-01 -6.90739155e-01
5.17022967e-01 4.34969693e-01 -1.00369707e-01 -5.50877452e-01
-7.26895213e-01 -3.00950915e-01 -3.63077015e-01 -1.92541599e-01
1.26051402e+00 -1.00563836e+00 -6.75285578e-01 7.52212048e-01
-1.03754866e+00 -3.04267526e-01 -3.25462192e-01 4.10012543e-01
-7.31489182e-01 4.82947558e-01 -2.12899119e-01 -1.02887893e+00
-2.24908143e-01 -1.31841552e+00 6.23857379e-01 6.78653955e-01
-5.02988882e-02 -7.41752446e-01 -3.78878474e-01 6.21738970e-01
5.16856492e-01 2.66476810e-01 8.42934370e-01 -6.91950321e-01
-4.19343799e-01 1.14986286e-01 -5.90863407e-01 5.82201242e-01
-5.93845136e-02 -2.03462869e-01 -1.21244729e+00 7.63042644e-02
9.27906558e-02 -2.14972079e-01 8.47009063e-01 4.27988887e-01
1.48948860e+00 -5.48959859e-02 -7.60859326e-02 4.99783278e-01
1.17815924e+00 4.24397200e-01 9.77444470e-01 3.87914717e-01
6.12362206e-01 8.65919232e-01 1.14767969e+00 6.12764537e-01
5.97675085e-01 4.86290812e-01 7.77339995e-01 -2.84079194e-01
-1.30089045e-01 -2.74397582e-01 1.69890355e-02 2.91395694e-01
-6.77731410e-02 -3.16839337e-01 -6.96496487e-01 4.23884779e-01
-2.08036065e+00 -7.33069539e-01 -3.57002705e-01 2.20538068e+00
4.62193072e-01 3.98536652e-01 1.69969611e-02 -1.44281060e-01
1.06740701e+00 2.16537908e-01 -9.34253693e-01 -5.79480052e-01
-8.13367292e-02 2.28219051e-02 6.36768937e-01 4.50292826e-01
-1.24391079e+00 1.28378701e+00 5.28255272e+00 1.29112172e+00
-8.41725588e-01 2.71948427e-01 1.15282941e+00 2.15668634e-01
-2.27995649e-01 -1.27628297e-01 -5.58581531e-01 7.45881557e-01
6.45236909e-01 -5.52695572e-01 2.73260534e-01 9.92206573e-01
5.88243604e-01 -3.26060832e-01 -6.04604304e-01 8.76226604e-01
-1.86296046e-01 -4.51517254e-01 -4.95586265e-03 -1.45776644e-01
8.06069493e-01 -3.02855730e-01 2.91369081e-01 4.48505104e-01
5.63256621e-01 -1.01484299e+00 9.30899262e-01 3.48104656e-01
7.35522449e-01 -1.12493730e+00 9.73758996e-01 4.25576806e-01
-9.57712948e-01 -8.34506676e-02 -6.03886664e-01 3.15281679e-03
2.45935589e-01 7.66913891e-01 -3.33940655e-01 8.65160406e-01
8.54208767e-01 6.23967528e-01 -4.91139710e-01 9.17219877e-01
-2.87166774e-01 6.16933286e-01 7.93690905e-02 8.70548263e-02
2.70468116e-01 -1.88007295e-01 3.46145838e-01 1.12063396e+00
9.85078886e-02 2.06402272e-01 1.23859040e-01 1.10435331e+00
2.46670134e-02 2.43556544e-01 -2.72677928e-01 3.70335966e-01
1.68739080e-01 1.11540318e+00 -6.56184673e-01 -4.39477891e-01
-1.81743637e-01 8.26906502e-01 1.48567289e-01 4.09735620e-01
-1.38474858e+00 -2.42934674e-01 9.50212240e-01 -5.24359159e-02
2.71517430e-02 1.67369619e-01 -8.78039956e-01 -8.81332934e-01
-2.08531931e-01 -7.22410142e-01 5.30351102e-01 -5.72556555e-01
-1.40686715e+00 6.41372979e-01 1.18121222e-01 -1.17346168e+00
2.77709216e-01 -2.30928972e-01 -6.08301401e-01 1.05558896e+00
-1.83311188e+00 -9.57168400e-01 -5.25290310e-01 7.66197145e-01
4.50109422e-01 -1.74083039e-01 4.77227002e-01 4.44259733e-01
-9.03137207e-01 8.45417976e-01 -1.15063354e-01 3.91696617e-02
7.83300042e-01 -9.78207171e-01 2.37596229e-01 1.08192015e+00
-5.76811612e-01 1.52697816e-01 7.57992625e-01 -5.18042922e-01
-2.03419417e-01 -1.37614763e+00 6.19716346e-01 -3.62087637e-02
1.80907741e-01 -1.38624817e-01 -8.87973487e-01 3.15961808e-01
2.74383426e-01 2.05937289e-02 4.44432557e-01 -1.65107265e-01
-1.83210641e-01 -3.96428376e-01 -1.55973494e+00 5.41349113e-01
1.15459645e+00 -2.16334879e-01 -4.07391340e-01 -9.76639017e-02
9.44673300e-01 -5.44973791e-01 -4.28172529e-01 5.22370577e-01
2.75426567e-01 -1.27496660e+00 7.11663365e-01 -5.68702102e-01
5.67194402e-01 -2.85112321e-01 -2.66245306e-01 -1.23361969e+00
-4.24586356e-01 -1.02623925e-01 4.77538735e-01 1.56300569e+00
3.36137563e-01 -8.51443648e-01 7.54132986e-01 1.29950249e+00
-3.16464067e-01 -2.57614404e-01 -1.04631114e+00 -7.39279985e-01
4.05187190e-01 -5.86822629e-01 7.89606929e-01 7.83046544e-01
-3.46532702e-01 3.55913229e-02 -4.77385670e-01 2.02865899e-01
6.90179825e-01 -1.48868978e-01 7.29622483e-01 -1.03305602e+00
8.13535228e-02 -6.88606322e-01 -3.63915980e-01 -8.94707203e-01
4.40689892e-01 -7.81703234e-01 2.96125144e-01 -1.36818087e+00
2.70364612e-01 -6.74979925e-01 -5.62363386e-01 1.52071133e-01
-5.49255073e-01 -1.55086070e-01 3.19688946e-01 -7.34955966e-02
-8.28143239e-01 9.85662341e-01 1.37507141e+00 -3.24901164e-01
9.89576951e-02 2.46183321e-01 -1.12855542e+00 6.44154012e-01
1.07394898e+00 -5.93287945e-01 -6.44134879e-01 -3.66613150e-01
-3.91251713e-01 -2.02177897e-01 4.36052650e-01 -1.08489287e+00
6.73225373e-02 -5.04930496e-01 -6.19123206e-02 -2.09812701e-01
3.31476480e-02 -1.06901515e+00 -1.08451314e-01 4.76705313e-01
-5.37041783e-01 -5.11473656e-01 3.03963274e-01 7.07113564e-01
-4.10682350e-01 -2.10232418e-02 1.26418269e+00 -3.59324031e-02
-1.21962762e+00 4.57816154e-01 3.17281894e-02 5.28583586e-01
1.29887295e+00 -2.50363976e-01 -1.38176501e-01 -1.74767420e-01
-4.73180979e-01 6.93222463e-01 2.85496682e-01 4.75145608e-01
3.44492197e-01 -1.29024482e+00 -8.01334441e-01 -2.55732778e-02
4.00657922e-01 2.27505475e-01 7.96414495e-01 6.71539903e-01
-2.34469950e-01 2.14444920e-01 -4.54162389e-01 -6.01095736e-01
-9.68016207e-01 6.37331307e-01 4.43469912e-01 -2.71101892e-01
-3.28672439e-01 8.39968920e-01 7.71561444e-01 -4.74711716e-01
2.16998696e-01 -2.35974804e-01 -3.78104210e-01 -1.35655880e-01
2.44328558e-01 5.37813008e-01 -2.13190526e-01 -8.84032965e-01
-4.54610974e-01 4.25428212e-01 2.15934157e-01 1.94410998e-02
7.29161739e-01 -5.94387949e-01 1.65004224e-01 1.27396077e-01
8.03415060e-01 -3.74111116e-01 -1.56128883e+00 -1.66858956e-01
-9.00802240e-02 -6.59474611e-01 -6.47825748e-02 -9.84445393e-01
-1.42349422e+00 1.16014755e+00 8.37873399e-01 -6.20815977e-02
1.14485300e+00 -4.26290542e-01 9.26383734e-01 -1.78041428e-01
5.36082864e-01 -1.42882550e+00 -3.88499707e-01 4.44691151e-01
5.56515157e-01 -1.59346974e+00 -3.32136095e-01 -6.38000190e-01
-1.29991937e+00 5.42612612e-01 1.06585455e+00 -5.34580387e-02
6.57511294e-01 -2.46052742e-01 1.61522180e-01 1.60051063e-01
-7.55041018e-02 -3.53416055e-01 2.26777762e-01 7.68820703e-01
1.97944209e-01 5.03533304e-01 -5.57308316e-01 1.13559282e+00
-1.35674372e-01 1.37913957e-01 3.07837605e-01 3.21257561e-01
-4.40064996e-01 -7.49772787e-01 -2.45759636e-01 2.43956670e-01
-3.55709314e-01 9.30599943e-02 -7.70040080e-02 6.88229680e-01
5.13250291e-01 1.29136264e+00 -1.39524594e-01 -2.92433232e-01
5.95942497e-01 -1.64942294e-01 -1.86671928e-01 -3.58360201e-01
-3.80434752e-01 -2.81909764e-01 3.48605923e-02 -6.53763711e-01
-6.07478142e-01 -4.19050723e-01 -1.54317033e+00 -5.18811107e-01
-1.77629575e-01 2.37629429e-01 3.93594265e-01 9.86700535e-01
2.76675999e-01 7.97598898e-01 5.88374496e-01 -3.96208525e-01
-3.66194576e-01 -7.39088356e-01 -7.87655890e-01 7.64153957e-01
-2.01527283e-01 -8.25181186e-01 -2.25984141e-01 -1.56308517e-01] | [9.599556922912598, 1.466704249382019] |
a337aaf6-f8a6-4806-98db-959bdf8edcf5 | resper-computationally-modelling-resisting | 2101.10545 | null | https://arxiv.org/abs/2101.10545v1 | https://arxiv.org/pdf/2101.10545v1.pdf | RESPER: Computationally Modelling Resisting Strategies in Persuasive Conversations | Modelling persuasion strategies as predictors of task outcome has several real-world applications and has received considerable attention from the computational linguistics community. However, previous research has failed to account for the resisting strategies employed by an individual to foil such persuasion attempts. Grounded in prior literature in cognitive and social psychology, we propose a generalised framework for identifying resisting strategies in persuasive conversations. We instantiate our framework on two distinct datasets comprising persuasion and negotiation conversations. We also leverage a hierarchical sequence-labelling neural architecture to infer the aforementioned resisting strategies automatically. Our experiments reveal the asymmetry of power roles in non-collaborative goal-directed conversations and the benefits accrued from incorporating resisting strategies on the final conversation outcome. We also investigate the role of different resisting strategies on the conversation outcome and glean insights that corroborate with past findings. We also make the code and the dataset of this work publicly available at https://github.com/americast/resper. | ['Carolyn Penstein Rosé', 'Haogang Bao', 'Xinru Yan', 'Meredith Riggs', 'Surya Shekhar Chakraborty', 'Rishabh Joshi', 'Sayan Sinha', 'Ritam Dutt'] | 2021-01-26 | null | https://aclanthology.org/2021.eacl-main.7 | https://aclanthology.org/2021.eacl-main.7.pdf | eacl-2021-2 | ['persuasion-strategies'] | ['computer-vision'] | [ 7.59190857e-01 7.24470198e-01 -1.93884388e-01 -6.98917270e-01
-3.96772772e-01 -5.53930223e-01 1.28864312e+00 2.85881966e-01
-5.97575903e-01 7.49271393e-01 1.07536352e+00 -8.73803675e-01
-4.58179533e-01 -3.96032363e-01 -7.14156181e-02 -3.23927045e-01
2.90666103e-01 3.91328663e-01 -3.53994101e-01 -4.57735986e-01
8.68594050e-01 -4.27829800e-03 -1.24225986e+00 4.75438803e-01
6.67480767e-01 2.90149122e-01 1.88402623e-01 6.70309484e-01
7.27383122e-02 1.16869259e+00 -4.84112501e-01 -7.15712070e-01
-5.37107065e-02 -7.12430239e-01 -1.23142827e+00 7.11831897e-02
7.03479070e-03 -1.71412826e-01 -1.44292012e-01 6.21540129e-01
6.22731566e-01 1.49704635e-01 6.94472015e-01 -1.23319268e+00
-7.32460916e-01 1.33294523e+00 -1.60104573e-01 4.59672332e-01
6.12575173e-01 2.66351700e-01 1.15099180e+00 -1.94898710e-01
6.13166809e-01 1.58617187e+00 6.51309550e-01 8.57670009e-01
-1.28327692e+00 -5.74597716e-01 2.34247848e-01 3.36380780e-01
-6.15323782e-01 -1.05195677e+00 1.07669997e+00 -5.05360186e-01
6.95455253e-01 3.33250344e-01 7.32962310e-01 1.93199551e+00
-4.23162356e-02 8.08392763e-01 1.64976156e+00 -3.61257553e-01
6.91895336e-02 -4.10975777e-02 4.34185535e-01 4.21125561e-01
5.63401692e-02 1.29976571e-01 -7.44406998e-01 -4.29767311e-01
2.59404629e-01 -4.45406973e-01 -1.95937529e-01 2.47229263e-01
-1.07948053e+00 1.13023090e+00 1.98601708e-01 5.09645462e-01
-5.28067172e-01 -9.89027247e-02 4.88615900e-01 4.53323275e-01
5.80140531e-01 5.85271537e-01 -3.10506701e-01 -6.55231595e-01
-3.83477211e-01 4.09348309e-01 1.06266451e+00 2.83268750e-01
3.57448936e-01 -1.75612941e-01 -1.84676394e-01 1.09525287e+00
3.58973891e-01 -1.85401589e-02 4.02568877e-01 -1.19004619e+00
2.94386178e-01 5.22420645e-01 2.80554324e-01 -1.10946918e+00
-8.36710036e-01 1.73429847e-02 -4.81581181e-01 -1.67941853e-01
5.67721605e-01 -2.79216915e-01 -5.48242740e-02 2.02514505e+00
2.68736839e-01 -1.96841165e-01 2.23814300e-03 7.65268862e-01
4.76838648e-01 4.23878385e-03 3.79371464e-01 -5.15296221e-01
1.37930584e+00 -6.63722813e-01 -6.76688790e-01 -5.98814487e-01
8.01291049e-01 -8.04161191e-01 1.17244482e+00 8.50345567e-02
-1.03080618e+00 -2.61401713e-01 -7.63967693e-01 -1.61584079e-01
1.30662009e-01 -3.75013947e-01 9.69596505e-01 7.36600816e-01
-9.20595706e-01 5.73313892e-01 -5.57194829e-01 -5.67579567e-01
5.92814624e-01 -2.83485889e-01 1.40998643e-02 2.70289093e-01
-1.27705884e+00 1.17064238e+00 4.99319062e-02 3.02336663e-01
-5.87320328e-01 -4.00326222e-01 -6.37889922e-01 -1.75031036e-01
7.15723932e-01 -7.40331590e-01 1.63695002e+00 -1.34044373e+00
-1.67371964e+00 1.00658262e+00 -1.51232228e-01 -2.81647235e-01
6.44170165e-01 -1.38441533e-01 -8.49004388e-02 -1.49354413e-01
2.39038855e-01 2.68145025e-01 6.50268793e-01 -7.68618345e-01
-4.28034455e-01 -2.88146049e-01 3.55997562e-01 5.28455675e-01
-1.79581180e-01 2.10033938e-01 7.37719715e-01 -4.48471576e-01
-2.88921595e-01 -9.12487864e-01 -1.28835246e-01 -6.50792003e-01
-6.01961553e-01 -8.85624707e-01 1.86270148e-01 -4.54641283e-01
1.07787704e+00 -1.85850537e+00 2.97473311e-01 -1.31612211e-01
4.83262956e-01 -2.94721145e-02 -1.57602757e-01 7.40452886e-01
5.77754118e-02 5.31341195e-01 -5.32068759e-02 -3.24817717e-01
5.31444430e-01 2.46105045e-02 -1.36076942e-01 5.41017413e-01
-2.28490517e-01 1.13329530e+00 -6.85953319e-01 -1.77146018e-01
-1.92713458e-02 4.17859614e-01 -4.52383488e-01 2.35543177e-01
-3.04906331e-02 4.72673208e-01 -6.16529882e-01 1.17091589e-01
2.42375225e-01 -3.81772310e-01 8.54502618e-01 4.38701034e-01
-1.81024536e-01 1.17875886e+00 -4.22384053e-01 1.24558175e+00
-1.60131097e-01 6.55836403e-01 4.62683737e-01 -1.09388614e+00
6.77616894e-01 3.12524617e-01 -1.72756940e-01 -6.92854404e-01
4.37026560e-01 -4.53271903e-02 8.85878623e-01 -3.79555702e-01
4.22005504e-01 -5.46765864e-01 -2.95124769e-01 1.33241928e+00
-5.45741618e-01 -9.22396779e-03 1.12053836e-02 4.62921441e-01
9.59488273e-01 -1.45072257e-02 6.36322260e-01 -4.19957072e-01
6.06723785e-01 -6.91078156e-02 6.00749195e-01 1.09767592e+00
-4.32911962e-01 -2.47216836e-01 7.71662116e-01 -4.55105871e-01
-1.01223588e+00 -3.60540181e-01 6.80290759e-02 1.46132147e+00
-2.04688147e-01 -3.50818723e-01 -7.45334327e-01 -5.95151067e-01
-2.09988922e-01 1.14194381e+00 -7.41149604e-01 -1.16306208e-01
-7.56217062e-01 -5.09924054e-01 6.74518824e-01 2.13831961e-01
3.88202369e-01 -1.42189491e+00 -8.64391983e-01 2.71313548e-01
-6.48319721e-01 -9.90852058e-01 -2.22710535e-01 -2.30740651e-01
-5.13450742e-01 -1.28489304e+00 1.10666521e-01 -1.80587232e-01
1.12535125e-02 4.23721075e-01 1.06744421e+00 8.32725465e-01
-2.51772460e-02 6.86992705e-01 -4.09356475e-01 -6.22498393e-01
-1.01912880e+00 2.37149477e-01 1.09444723e-01 -2.77862370e-01
5.61685622e-01 -6.78810954e-01 -5.52642524e-01 3.10489625e-01
-4.09193188e-01 5.40074229e-01 1.68992206e-01 5.64347863e-01
-4.64018881e-01 -4.19077605e-01 1.02166867e+00 -9.31058049e-01
1.55033839e+00 -7.69921720e-01 -4.84482609e-02 -2.54007429e-01
-8.31162214e-01 -4.13678765e-01 3.02367210e-01 -2.39557579e-01
-1.29979753e+00 -8.99646640e-01 -3.06959569e-01 6.33621991e-01
-5.17314672e-01 3.92455667e-01 2.74654299e-01 4.48196769e-01
6.54873490e-01 -6.87843412e-02 4.03494447e-01 -2.28469923e-01
3.38899493e-01 7.94538796e-01 -8.12316015e-02 -9.20278430e-01
4.39194083e-01 2.76195109e-01 -4.44805652e-01 -8.87547195e-01
-1.27226269e+00 5.67245670e-02 -2.99024165e-01 -4.01613384e-01
7.70345390e-01 -4.88886386e-01 -1.40815771e+00 2.75622755e-01
-1.06800830e+00 -8.36357772e-01 2.08395571e-01 1.99284732e-01
-9.08129692e-01 5.11105299e-01 -6.39530599e-01 -1.24210000e+00
-4.56858009e-01 -8.01021576e-01 5.25289357e-01 1.52954683e-01
-1.11371398e+00 -1.25712550e+00 -9.13153216e-02 1.02650785e+00
6.09275341e-01 -3.63990963e-02 9.64589655e-01 -1.09961736e+00
-3.24381813e-02 3.18593740e-01 1.69408366e-01 -9.27219093e-02
-9.22345668e-02 -4.74325299e-01 -7.14580119e-01 1.29963309e-01
4.30114299e-01 -7.31292009e-01 6.35563195e-01 3.62282306e-01
5.24221778e-01 -5.87238193e-01 -2.49973014e-01 -8.04952979e-02
4.64786351e-01 -2.77098175e-02 4.19271588e-01 5.64113975e-01
2.81568408e-01 1.15981781e+00 6.31205797e-01 3.16151530e-01
8.68812263e-01 5.41911006e-01 -4.32367548e-02 3.44240665e-01
2.68413246e-01 -4.03265029e-01 3.52014661e-01 3.70608985e-01
-1.35358900e-01 -4.38798904e-01 -1.03087425e+00 5.02274096e-01
-1.95890200e+00 -1.30793035e+00 -2.60339975e-01 1.69363332e+00
9.87728953e-01 3.13743562e-01 2.10386172e-01 2.83979952e-01
5.82238913e-01 6.30102694e-01 -1.49509013e-01 -8.77693832e-01
-6.98087215e-02 -9.58355442e-02 -1.67544872e-01 1.04341865e+00
-6.67155504e-01 1.10842192e+00 6.40029383e+00 3.08449686e-01
-5.83223283e-01 2.52151787e-01 8.47214997e-01 -7.13403746e-02
-5.40175438e-01 1.02535509e-01 -4.74937469e-01 2.48885617e-01
1.08135808e+00 -4.22373325e-01 6.70791388e-01 2.40161538e-01
6.20023549e-01 -2.22130150e-01 -1.37305176e+00 2.22257137e-01
-4.68761250e-02 -1.19826257e+00 -5.65701485e-01 2.04274952e-01
1.59313604e-01 -1.47853583e-01 -1.91983864e-01 3.14055622e-01
6.85216367e-01 -1.07027841e+00 7.37761855e-01 3.52672487e-01
2.32519373e-01 -3.45637321e-01 5.30394971e-01 7.80964375e-01
-1.26610398e-01 -2.45997131e-01 -2.77902223e-02 -1.05991423e+00
5.73753655e-01 7.35127926e-02 -1.27391267e+00 4.66615483e-02
3.50949824e-01 5.25819302e-01 -2.04092249e-01 -2.65543312e-01
-7.83500075e-01 1.05450976e+00 -3.21587436e-02 -3.65852088e-01
1.90453872e-01 -2.72889078e-01 7.82709002e-01 9.40233529e-01
-3.96581173e-01 4.60138291e-01 5.48789604e-03 8.44345331e-01
8.39736238e-02 8.53428319e-02 -6.51020288e-01 -5.81860721e-01
7.79765546e-01 1.35811830e+00 -6.72751725e-01 -3.15356344e-01
-1.65565312e-01 6.17689669e-01 7.15950489e-01 3.17557305e-01
-5.04838586e-01 4.10195470e-01 8.54521990e-01 5.27743772e-02
-9.72868949e-02 -2.58568287e-01 -5.06327331e-01 -9.54145432e-01
-2.06925794e-01 -1.03195643e+00 2.66838104e-01 -7.09911883e-01
-1.51622713e+00 1.50648057e-01 2.10048556e-02 -1.03153652e-02
-5.37652075e-01 -2.53184795e-01 -6.82233930e-01 1.01869690e+00
-1.19468796e+00 -9.94908392e-01 -1.17059939e-01 1.63153917e-01
7.31838346e-01 9.30650532e-03 9.31356668e-01 -4.05598015e-01
-5.75964868e-01 3.57622683e-01 -8.32580745e-01 -1.45052776e-01
6.01545513e-01 -9.03089523e-01 5.54517567e-01 4.98415768e-01
-1.70671701e-01 9.91598010e-01 9.55337882e-01 -6.16209686e-01
-1.18236458e+00 -1.95498571e-01 1.28975940e+00 -9.17904198e-01
1.04434741e+00 -5.00086248e-01 -8.76188517e-01 9.93977666e-01
7.15211928e-01 -9.99406815e-01 1.15188885e+00 6.83882058e-01
-3.39694142e-01 6.98775768e-01 -9.32807028e-01 9.17064428e-01
1.76937795e+00 -4.59465533e-01 -1.06079924e+00 3.59189987e-01
4.37717468e-01 -4.93457802e-02 -6.18928373e-01 -7.04793632e-02
6.49468899e-01 -1.29290736e+00 7.77474463e-01 -8.96908045e-01
8.69145632e-01 3.57644111e-01 2.92383004e-02 -1.36169481e+00
-6.82316303e-01 -8.20359409e-01 1.97494507e-01 1.11933935e+00
2.71088988e-01 -1.00062227e+00 3.72969538e-01 7.13903666e-01
4.32439595e-02 -5.53636849e-01 -9.60884273e-01 5.48106879e-02
2.91402608e-01 -5.41734993e-01 4.33137596e-01 1.27334213e+00
2.73062229e-01 1.03702199e+00 -4.10565853e-01 -3.05518031e-01
5.32295108e-01 5.98784275e-02 9.17568624e-01 -1.48268628e+00
-2.06465676e-01 -7.63183415e-01 1.72539204e-01 -8.33937347e-01
8.58851314e-01 -1.10585678e+00 -7.49689862e-02 -1.70430303e+00
3.33761930e-01 -5.00751615e-01 2.34092057e-01 7.54512012e-01
-3.46877277e-01 -2.98622716e-02 3.14366937e-01 1.89948976e-01
-4.82169390e-01 6.03353441e-01 9.18993294e-01 2.84924924e-01
-1.74602553e-01 -2.27624863e-01 -1.80518401e+00 9.53171611e-01
1.18123746e+00 -3.32694203e-01 -5.12723505e-01 -2.08450779e-01
4.28393006e-01 7.82712847e-02 4.88363802e-01 1.17063887e-01
-1.19566858e-01 -5.43471396e-01 -5.27548715e-02 1.07303053e-01
1.52124003e-01 -3.00140947e-01 -7.56453797e-02 3.08021188e-01
-8.48193705e-01 -6.78998753e-02 8.95715281e-02 5.69493115e-01
5.67708910e-01 -2.31481299e-01 5.35691142e-01 -8.10074732e-02
-2.52745956e-01 -5.11314034e-01 -9.61892188e-01 1.99306265e-01
8.02033365e-01 -2.08851382e-01 -6.11436188e-01 -8.20731282e-01
-5.00648737e-01 2.65024722e-01 4.50584978e-01 8.30948114e-01
2.93196976e-01 -9.35011685e-01 -1.10886645e+00 -2.15881228e-01
-2.41831258e-01 -6.45814359e-01 2.72379667e-01 1.20318341e+00
1.50213957e-01 3.71421695e-01 -1.03634395e-01 -2.17271879e-01
-1.39311361e+00 2.90904760e-01 2.79858083e-01 9.41362754e-02
-7.75491536e-01 7.42407560e-01 1.30632058e-01 -4.71230328e-01
-3.38748455e-01 5.23767054e-01 -4.05371010e-01 3.21563601e-01
6.06155515e-01 5.62257230e-01 -4.38180089e-01 -6.34010017e-01
-3.60893518e-01 -1.87261164e-01 -3.99655759e-01 -4.27172780e-01
1.34815621e+00 -4.81484801e-01 -3.32002312e-01 6.16768837e-01
5.27586162e-01 1.92139402e-01 -9.20011282e-01 -2.74681121e-01
2.59222806e-01 -3.32487404e-01 -2.53767163e-01 -9.69183385e-01
-6.10202812e-02 4.35074240e-01 -2.88500279e-01 6.73484147e-01
5.85781217e-01 1.62280947e-01 1.44232854e-01 2.21823990e-01
2.33728841e-01 -1.13096356e+00 -4.66597676e-02 5.58135152e-01
1.23526525e+00 -1.13336515e+00 1.54221565e-01 -3.69360238e-01
-9.55570400e-01 6.58598602e-01 3.97041619e-01 6.76409528e-02
1.86744004e-01 1.52670592e-01 2.96265990e-01 -4.23953950e-01
-1.41100276e+00 8.96277875e-02 -1.38666272e-01 5.25486827e-01
1.09279835e+00 2.71833301e-01 -1.23834860e+00 8.12909782e-01
-7.24310994e-01 -1.42566234e-01 8.09793115e-01 9.19422686e-01
-2.56733596e-01 -1.06679964e+00 -8.44209567e-02 3.89579594e-01
-5.48106372e-01 -2.42633775e-01 -1.18057799e+00 5.82146943e-01
-4.37337995e-01 1.50291002e+00 2.02047895e-03 -2.48611301e-01
1.31983712e-01 5.22927701e-01 3.29041719e-01 -5.41903138e-01
-8.52112591e-01 -3.61392759e-02 9.72275674e-01 -5.55869341e-01
-8.37314308e-01 -1.09883249e+00 -9.64782119e-01 -8.27163100e-01
-1.83874741e-02 2.44426951e-01 3.63664985e-01 1.07882571e+00
4.94560748e-01 3.19130927e-01 3.16702425e-01 -6.30951762e-01
-7.60487795e-01 -1.49926054e+00 -2.35738754e-02 5.42260766e-01
8.44567269e-02 -6.34607673e-01 -6.20365858e-01 -2.79244304e-01] | [12.723514556884766, 7.960143089294434] |
7faaa1c6-0f0b-4094-ab51-024dfd09e755 | multi-view-active-fine-grained-recognition | 2206.01153 | null | https://arxiv.org/abs/2206.01153v1 | https://arxiv.org/pdf/2206.01153v1.pdf | Multi-View Active Fine-Grained Recognition | As fine-grained visual classification (FGVC) being developed for decades, great works related have exposed a key direction -- finding discriminative local regions and revealing subtle differences. However, unlike identifying visual contents within static images, for recognizing objects in the real physical world, discriminative information is not only present within seen local regions but also hides in other unseen perspectives. In other words, in addition to focusing on the distinguishable part from the whole, for efficient and accurate recognition, it is required to infer the key perspective with a few glances, e.g., people may recognize a "Benz AMG GT" with a glance of its front and then know that taking a look at its exhaust pipe can help to tell which year's model it is. In this paper, back to reality, we put forward the problem of active fine-grained recognition (AFGR) and complete this study in three steps: (i) a hierarchical, multi-view, fine-grained vehicle dataset is collected as the testbed, (ii) a simple experiment is designed to verify that different perspectives contribute differently for FGVC and different categories own different discriminative perspective, (iii) a policy-gradient-based framework is adopted to achieve efficient recognition with active view selection. Comprehensive experiments demonstrate that the proposed method delivers a better performance-efficient trade-off than previous FGVC methods and advanced neural networks. | ['Zhanyu Ma', 'Yongbin Li', 'Ting-En Lin', 'Dongliang Chang', 'Heqing Wang', 'Wenqing Yu', 'Ruoyi Du'] | 2022-06-02 | null | null | null | null | ['fine-grained-image-classification'] | ['computer-vision'] | [ 3.56481262e-02 -3.41547787e-01 -3.28142077e-01 -3.90598506e-01
-4.14475381e-01 -7.21443653e-01 6.43291771e-01 -1.87687963e-01
3.39982286e-03 5.36497891e-01 2.33129129e-01 -8.34567547e-02
-7.61341676e-02 -7.18185723e-01 -6.78950131e-01 -1.00971985e+00
8.31783488e-02 2.07954571e-01 4.02327687e-01 -1.71522930e-01
4.58676785e-01 9.10944104e-01 -1.77360260e+00 4.68599886e-01
5.08838177e-01 1.29470229e+00 2.93901205e-01 4.01613027e-01
9.62062925e-02 8.63220990e-01 -5.55426717e-01 -2.70638257e-01
4.17342871e-01 -2.89373938e-02 -5.71537316e-01 4.60940659e-01
7.76551366e-01 -5.79650283e-01 -2.72684544e-01 1.06347179e+00
1.21644117e-01 2.18518615e-01 7.44440436e-01 -1.17676389e+00
-5.08894265e-01 -1.00133710e-01 -4.58134085e-01 4.62620050e-01
2.19028994e-01 3.50140482e-01 7.97352135e-01 -1.06073344e+00
4.78275031e-01 1.35728931e+00 2.42801726e-01 3.86604369e-01
-7.89830625e-01 -6.76542819e-01 8.19229543e-01 6.00234509e-01
-1.30394650e+00 -4.86943841e-01 1.00984490e+00 -5.81457376e-01
6.04312837e-01 4.62394506e-01 7.04133093e-01 1.32536733e+00
4.20114815e-01 8.85919273e-01 1.34266925e+00 -1.16229393e-01
9.94127318e-02 3.01220655e-01 2.32214347e-01 6.51415646e-01
2.50516623e-01 4.71703112e-01 -4.11217958e-01 1.49509832e-01
3.27171475e-01 3.40810597e-01 -2.39220634e-01 -5.50450325e-01
-1.21042275e+00 6.19053066e-01 3.58483076e-01 2.30409488e-01
-2.87958592e-01 -2.28340417e-01 2.18136087e-01 2.71315813e-01
3.99626940e-01 2.77816206e-02 -1.75664470e-01 1.54000461e-01
-8.46944034e-01 6.48832694e-02 6.55796707e-01 7.10282028e-01
8.76345217e-01 3.24895501e-01 -2.31255710e-01 5.13890386e-01
3.30079943e-01 5.15301526e-01 3.54504853e-01 -7.85092294e-01
4.55636680e-01 6.78068101e-01 1.27123296e-01 -1.42826581e+00
-7.80589059e-02 -8.08029294e-01 -9.97313380e-01 6.83156610e-01
3.79393667e-01 2.86277413e-01 -9.47004318e-01 1.23332381e+00
4.12755787e-01 2.71480344e-02 -2.40545779e-01 1.17214179e+00
8.40638399e-01 6.81690216e-01 -7.74284527e-02 -2.95941412e-01
1.41178095e+00 -1.07899201e+00 -4.51373577e-01 -3.73496473e-01
7.03853443e-02 -6.37922108e-01 7.30056047e-01 5.53371906e-01
-4.45331037e-01 -9.95289326e-01 -1.26514173e+00 1.88371256e-01
-8.03263068e-01 3.17522176e-02 5.40799618e-01 5.65018713e-01
-7.82974005e-01 1.68866843e-01 -4.17095780e-01 -2.84742057e-01
3.33625585e-01 -7.40308687e-02 -5.91980517e-01 -3.66858482e-01
-8.47794592e-01 7.57032573e-01 3.51305068e-01 4.28637505e-01
-1.39376652e+00 -4.40483779e-01 -6.08750343e-01 5.06939832e-03
6.73793197e-01 -4.49519515e-01 6.23084962e-01 -1.14648771e+00
-1.05117929e+00 7.06054628e-01 -3.94743949e-01 -3.50392580e-01
7.81846702e-01 -6.98796660e-02 -6.32342756e-01 9.76089165e-02
9.03161019e-02 2.51401454e-01 1.14211595e+00 -1.67477143e+00
-8.48327875e-01 -7.16979027e-01 3.26349080e-01 2.46850505e-01
4.63541597e-02 -7.65459687e-02 -4.02481198e-01 -6.02207780e-01
2.10421368e-01 -7.34902263e-01 -8.39650445e-03 6.32939935e-02
-3.25634003e-01 -1.05530091e-01 1.33055198e+00 -7.16053963e-01
1.04600608e+00 -2.28998923e+00 -7.75137395e-02 1.76415101e-01
4.60072488e-01 3.23119432e-01 2.08322585e-01 3.71776700e-01
1.96472302e-01 -1.00556545e-01 1.29525304e-01 -4.93121818e-02
-6.60684854e-02 2.37301871e-01 -5.48404574e-01 6.45511150e-01
-3.59320641e-03 7.87435532e-01 -6.72671556e-01 -4.86285031e-01
5.44213653e-01 3.14325571e-01 -1.97243422e-01 1.69264928e-01
9.89654660e-02 4.91036862e-01 -7.37557650e-01 1.15308869e+00
8.81016135e-01 -1.94198161e-01 -1.66987740e-02 -6.27778172e-01
-3.26196581e-01 -6.14764333e-01 -1.31486917e+00 9.02020216e-01
-1.73732862e-01 6.18859708e-01 2.90311351e-02 -1.25661063e+00
9.19029355e-01 -4.09019515e-02 2.19866902e-01 -1.04440391e+00
4.30694297e-02 -8.03126674e-03 -2.94292748e-01 -5.86290240e-01
5.56162953e-01 1.86573908e-01 -9.28219482e-02 -1.41813150e-02
-2.11001843e-01 4.85306710e-01 1.14156993e-03 5.61673827e-02
7.13444114e-01 2.71767471e-02 5.12652993e-01 -1.20052151e-01
7.07636774e-01 2.62321346e-02 6.44012988e-01 7.59456277e-01
-6.06774211e-01 5.38903475e-01 1.04891255e-01 -7.18411267e-01
-6.35881603e-01 -1.11012352e+00 3.38180549e-02 1.03548527e+00
7.73639143e-01 -2.84931231e-02 -4.32349473e-01 -9.02354717e-01
1.38862520e-01 6.35993242e-01 -8.59295607e-01 -8.69064033e-02
-5.24017572e-01 -3.20271671e-01 1.50446117e-01 4.10233498e-01
9.12704229e-01 -7.03437567e-01 -7.19877541e-01 -1.48166269e-01
-1.65958077e-01 -1.06510127e+00 -3.73190075e-01 1.36755839e-01
-6.01123512e-01 -1.14730334e+00 -5.73760569e-01 -5.77827454e-01
5.13422787e-01 9.42648649e-01 1.04845369e+00 -1.07275471e-01
-1.06046870e-01 4.02609408e-01 -3.18071455e-01 -3.02786916e-01
3.23928371e-02 -3.12057197e-01 -4.22705598e-02 7.05315351e-01
3.53667289e-01 -4.01948631e-01 -8.04479778e-01 5.92124164e-01
-4.68519568e-01 -7.32903704e-02 8.69337678e-01 7.15599179e-01
6.84944451e-01 2.91626275e-01 2.55723804e-01 -6.83402359e-01
1.35494888e-01 -4.26484287e-01 -6.08987153e-01 4.93389934e-01
-6.07094646e-01 -2.07380936e-01 7.41314948e-01 -2.45567158e-01
-1.21010065e+00 -3.19328234e-02 1.24125250e-01 -9.02433574e-01
-5.56122482e-01 4.41256864e-03 -4.81629819e-01 -2.34600767e-01
3.81946027e-01 6.57082617e-01 -1.63736343e-01 -4.23071325e-01
2.09256381e-01 6.00690722e-01 3.90451580e-01 -2.74650365e-01
1.06736517e+00 8.31426799e-01 -1.81534901e-01 -8.59878242e-01
-8.91518176e-01 -7.10465193e-01 -7.77942359e-01 -7.08864689e-01
1.00599265e+00 -9.84324932e-01 -7.77734578e-01 2.66851187e-01
-7.87874103e-01 8.65201578e-02 5.07716350e-02 3.27561080e-01
-2.07910299e-01 4.52778071e-01 2.56346948e-02 -9.31759238e-01
6.67273030e-02 -1.17300236e+00 1.19103777e+00 3.57645184e-01
2.37714440e-01 -8.91827822e-01 -2.14394718e-01 7.58672655e-01
4.95966792e-01 2.65423506e-01 6.94459081e-01 -4.71909314e-01
-1.07734740e+00 -1.15311041e-01 -4.00853068e-01 3.57922554e-01
1.09419517e-01 -5.26093785e-03 -1.21424150e+00 -4.34925377e-01
5.09087853e-02 -7.65105337e-02 8.87914121e-01 1.71380401e-01
1.11309254e+00 -3.87330502e-01 -5.88253617e-01 7.56325841e-01
1.56973243e+00 5.98100364e-01 4.75020170e-01 3.68916273e-01
8.50283325e-01 5.87159812e-01 9.18500423e-01 2.03344002e-01
3.45219612e-01 8.62442672e-01 7.72499084e-01 -1.83489025e-01
-2.99062490e-01 -2.35905021e-01 4.54245776e-01 4.11318153e-01
-2.46685028e-01 -2.50224590e-01 -4.76004809e-01 4.43316966e-01
-1.62954211e+00 -1.27664065e+00 1.75551921e-01 2.16037679e+00
-4.81771603e-02 1.78953633e-01 7.69046620e-02 1.45578086e-01
6.85047626e-01 6.26382947e-01 -7.30971932e-01 -1.33603603e-01
-2.71852940e-01 -4.26185250e-01 4.14691180e-01 2.63105541e-01
-1.24317324e+00 7.07024395e-01 5.41855240e+00 1.13195646e+00
-1.53115070e+00 1.18059643e-01 8.67166817e-01 7.46567175e-02
-1.85898229e-01 -1.51404496e-02 -9.05645967e-01 5.48479617e-01
2.37883776e-01 1.52431563e-01 2.65094668e-01 1.07817340e+00
2.17276528e-01 -1.66828156e-01 -8.76918316e-01 1.18424070e+00
3.98270458e-01 -1.11876833e+00 2.21649244e-01 2.72942364e-01
5.72838664e-01 -1.30996227e-01 1.10144563e-01 2.58049279e-01
5.98207377e-02 -1.04421914e+00 9.89673913e-01 6.91795409e-01
5.16381741e-01 -6.03892386e-01 5.30928075e-01 5.61565340e-01
-1.47423089e+00 -3.97102475e-01 -4.17872250e-01 1.59022406e-01
-1.21081801e-04 5.45391262e-01 -6.04888558e-01 8.87515128e-01
7.47679114e-01 6.05221629e-01 -7.36152470e-01 9.34321642e-01
1.52585551e-01 4.94325340e-01 2.16561824e-01 8.94755349e-02
4.21774894e-01 -2.56917000e-01 6.57070160e-01 1.16225767e+00
1.79144934e-01 1.21580986e-02 3.48955005e-01 6.53716505e-01
3.31636310e-01 -1.78214461e-01 -8.50554585e-01 1.12228066e-01
2.16927648e-01 1.35226941e+00 -8.25125158e-01 -2.65183628e-01
-4.60520774e-01 1.03542471e+00 6.97145611e-02 7.06565320e-01
-8.36008847e-01 -6.52834699e-02 5.21168590e-01 2.45048061e-01
5.99908769e-01 -2.73379058e-01 -1.78528596e-02 -1.28458345e+00
8.79490003e-02 -8.94999743e-01 3.86223972e-01 -7.48997569e-01
-1.14200068e+00 6.32500112e-01 -8.33051130e-02 -1.66475701e+00
-1.46303698e-01 -6.77163124e-01 -4.59238082e-01 7.85782456e-01
-1.61437607e+00 -1.41023564e+00 -7.22141981e-01 6.99566185e-01
1.01726592e+00 -3.18090290e-01 3.44508886e-01 2.69946784e-01
-3.57587308e-01 5.76697886e-01 1.80904791e-01 1.56573609e-01
4.08054233e-01 -9.08906102e-01 -9.92496684e-02 1.08356464e+00
2.03718081e-01 4.12298501e-01 6.84269667e-01 -5.32388926e-01
-1.68467689e+00 -1.16157913e+00 5.99396050e-01 -5.52046895e-01
2.89863795e-01 -3.64805460e-01 -7.67354608e-01 4.14165646e-01
7.99747556e-02 1.49307787e-01 1.85425714e-01 -8.16355795e-02
-4.26134229e-01 -6.15971088e-01 -1.07450175e+00 2.98591435e-01
1.04161048e+00 -5.68192244e-01 -5.12803435e-01 1.19905777e-01
1.99248195e-01 -1.75494030e-01 -4.99047995e-01 5.64198196e-01
8.71097982e-01 -1.35474122e+00 1.08280241e+00 -5.05458355e-01
9.54591557e-02 -5.98596334e-01 -5.92167377e-01 -1.21848238e+00
-5.52669048e-01 6.05001599e-02 -2.16866329e-01 1.13852966e+00
-3.93797383e-02 -7.34758914e-01 7.59292305e-01 -2.07358375e-02
-2.13083208e-01 -8.06485355e-01 -9.53710854e-01 -7.43364513e-01
-3.99580687e-01 -3.06369334e-01 4.40791249e-01 8.11972737e-01
-8.09827685e-01 2.16915295e-01 -5.75987041e-01 5.54082394e-01
8.57733190e-01 5.80077350e-01 8.52040470e-01 -1.23549342e+00
-2.88899273e-01 -2.26913080e-01 -5.71153879e-01 -1.12114096e+00
-8.33775401e-02 -3.93513381e-01 -6.90578371e-02 -1.45113969e+00
3.13733906e-01 -2.38268301e-01 -4.61885691e-01 1.05964042e-01
-1.47745293e-02 3.51007342e-01 3.67250592e-01 3.53373975e-01
-7.78123677e-01 4.05702412e-01 1.31675458e+00 -5.98065674e-01
4.34649467e-01 2.01215342e-01 -7.94519663e-01 5.53459048e-01
3.74492913e-01 7.13326875e-03 -5.84354997e-01 -4.39049564e-02
-1.02953665e-01 1.98728547e-01 7.64413655e-01 -1.06229615e+00
2.77867526e-01 -4.30118769e-01 7.24475801e-01 -9.72201288e-01
4.83298242e-01 -1.04363668e+00 3.14826578e-01 2.67665267e-01
1.80031523e-01 2.29342887e-03 -1.58475026e-01 8.54786515e-01
-5.63382745e-01 1.25797704e-01 7.38412738e-01 -4.82621700e-01
-1.52773452e+00 3.44478786e-01 -3.72617871e-01 -1.88278139e-01
1.18895435e+00 -8.84318292e-01 -5.50381303e-01 -3.03521365e-01
-5.32782614e-01 -1.95315387e-02 3.96422476e-01 5.97656965e-01
6.06534064e-01 -1.22917449e+00 -5.21621406e-01 3.31280708e-01
3.86045307e-01 -3.80689859e-01 8.38326037e-01 8.15718591e-01
-1.92311451e-01 6.59684420e-01 -3.19668263e-01 -8.10105979e-01
-1.44607997e+00 7.99954057e-01 4.33111817e-01 -6.63044751e-02
-5.24748802e-01 5.53872228e-01 9.42599416e-01 -6.49167895e-02
2.08217546e-01 -1.47527352e-01 -5.71285188e-01 4.23271596e-01
5.83746016e-01 3.32579106e-01 1.59908131e-01 -1.06953120e+00
-4.30705935e-01 7.82058060e-01 -1.92955017e-01 4.71916944e-01
9.63439286e-01 -5.40281057e-01 3.28574032e-01 5.50694823e-01
1.20770347e+00 2.36306444e-01 -1.39365661e+00 -2.40135461e-01
-4.09096956e-01 -8.97027075e-01 1.16738170e-01 -8.95995319e-01
-1.12509751e+00 8.77433836e-01 1.10736465e+00 4.13241684e-01
1.16237032e+00 8.62053111e-02 2.96311945e-01 2.40586460e-01
7.00035095e-01 -1.01371682e+00 2.15716243e-01 3.42461109e-01
8.32225919e-01 -1.47495067e+00 5.87228406e-03 -4.27080244e-01
-7.41311550e-01 1.16299307e+00 6.52331829e-01 -1.78775862e-02
4.20574248e-01 -5.74354306e-02 1.81681305e-01 -3.76893312e-01
-6.99600697e-01 -2.28555009e-01 5.02917111e-01 6.09228671e-01
-1.90692708e-01 1.67112485e-01 3.03960055e-01 1.95503637e-01
2.09280271e-02 -2.90778637e-01 1.41661867e-01 5.46960771e-01
-6.59841359e-01 -5.39104521e-01 -5.65666497e-01 4.23305959e-01
-2.19303548e-01 3.03871840e-01 -3.96909863e-01 9.39472497e-01
4.93709713e-01 1.01978803e+00 4.55661938e-02 -5.64304531e-01
2.63396680e-01 -2.43607879e-01 3.17471892e-01 -2.05990851e-01
-3.08416426e-01 -7.36087710e-02 1.84718058e-01 -8.92551720e-01
-4.32209432e-01 -6.34780228e-01 -5.83015323e-01 -2.61676937e-01
-8.96248221e-02 5.15138991e-02 4.09024715e-01 1.22178710e+00
2.41278097e-01 5.30631661e-01 9.72579718e-01 -9.32724237e-01
-3.60587299e-01 -5.85763931e-01 -6.01881564e-01 2.98418611e-01
6.13260686e-01 -1.04774737e+00 -5.99789679e-01 -1.13409415e-01] | [9.690919876098633, 1.940043568611145] |
71b9582d-fd7b-4198-a075-06191caf9e8f | counterfactual-explanation-and-causal | 2009.08856 | null | https://arxiv.org/abs/2009.08856v2 | https://arxiv.org/pdf/2009.08856v2.pdf | Counterfactual Explanation and Causal Inference in Service of Robustness in Robot Control | We propose an architecture for training generative models of counterfactual conditionals of the form, 'can we modify event A to cause B instead of C?', motivated by applications in robot control. Using an 'adversarial training' paradigm, an image-based deep neural network model is trained to produce small and realistic modifications to an original image in order to cause user-defined effects. These modifications can be used in the design process of image-based robust control - to determine the ability of the controller to return to a working regime by modifications in the input space, rather than by adaptation. In contrast to conventional control design approaches, where robustness is quantified in terms of the ability to reject noise, we explore the space of counterfactuals that might cause a certain requirement to be violated, thus proposing an alternative model that might be more expressive in certain robotics applications. So, we propose the generation of counterfactuals as an approach to explanation of black-box models and the envisioning of potential movement paths in autonomous robotic control. Firstly, we demonstrate this approach in a set of classification tasks, using the well known MNIST and CelebFaces Attributes datasets. Then, addressing multi-dimensional regression, we demonstrate our approach in a reaching task with a physical robot, and in a navigation task with a robot in a digital twin simulation. | ['Simón C. Smith', 'Subramanian Ramamoorthy'] | 2020-09-18 | null | null | null | null | ['counterfactual-explanation'] | ['miscellaneous'] | [ 7.58288443e-01 9.20193732e-01 7.98093528e-02 -1.70311421e-01
-2.55862087e-01 -6.62061930e-01 1.22139633e+00 -2.40978688e-01
-5.75664222e-01 9.60331440e-01 6.12618960e-02 -6.10397637e-01
-4.29852635e-01 -8.95849526e-01 -1.30685592e+00 -7.97684610e-01
-4.59085070e-02 3.84897768e-01 -1.48318827e-01 -3.89608055e-01
3.95827949e-01 6.51300669e-01 -1.52279377e+00 1.67080626e-01
4.59186077e-01 6.61399662e-01 1.93164334e-01 4.63710785e-01
4.53859508e-01 6.19774640e-01 -6.25673711e-01 -1.68230787e-01
4.39856827e-01 -4.95253325e-01 -7.02563345e-01 2.14203447e-01
-3.12513043e-03 -1.41396359e-01 -2.52680272e-01 1.12357879e+00
2.74583220e-01 1.36536747e-01 1.15076339e+00 -1.46742141e+00
-6.76310956e-01 6.54471815e-01 -6.50552139e-02 -2.38707364e-01
3.94428670e-01 6.31344020e-01 5.68803728e-01 -2.45153144e-01
8.43485534e-01 1.62242842e+00 3.68732005e-01 8.29237282e-01
-1.68649697e+00 -4.42635626e-01 7.20556304e-02 -1.06343202e-01
-8.62301946e-01 -3.10827494e-01 6.28120244e-01 -5.63479066e-01
5.83036661e-01 1.95907533e-01 5.31445146e-01 1.67307365e+00
6.34711385e-01 5.04474878e-01 1.28373182e+00 -4.65854317e-01
6.81640506e-01 -2.79576816e-02 -7.58577466e-01 3.59932125e-01
3.71628433e-01 8.43159735e-01 3.37259658e-02 -8.58287215e-02
8.34016085e-01 -2.14587450e-01 -4.45283562e-01 -9.26007032e-01
-1.38718009e+00 1.06424260e+00 7.18707561e-01 1.91709787e-01
-5.91289461e-01 5.77813804e-01 1.98619038e-01 4.62046236e-01
-4.94441688e-02 1.21129966e+00 -5.61279535e-01 3.58665317e-01
-2.83719867e-01 7.10003376e-01 7.90436268e-01 6.66113257e-01
4.22922283e-01 3.14166844e-01 -2.98457384e-01 5.84832439e-03
3.31042171e-01 2.38534689e-01 6.88569009e-01 -1.18009210e+00
2.83644229e-01 3.73459607e-01 4.59552228e-01 -5.55899382e-01
-4.15454000e-01 -2.05113426e-01 -6.51924133e-01 1.14359164e+00
5.60792804e-01 -4.59176898e-01 -1.02613640e+00 2.03661704e+00
2.07235530e-01 2.04778723e-02 4.03963566e-01 1.08594334e+00
-1.20062806e-01 3.94136161e-01 5.99412173e-02 -2.38500208e-01
1.01413357e+00 -2.48169675e-01 -4.45153326e-01 -3.10116976e-01
5.85921288e-01 -2.39349991e-01 1.21862209e+00 4.60629255e-01
-8.87086213e-01 -4.35388058e-01 -1.23016310e+00 5.24627686e-01
-5.74483275e-01 -2.98637390e-01 5.05382001e-01 4.63093638e-01
-9.01932895e-01 1.03083777e+00 -5.05965590e-01 -4.10776973e-01
4.25599039e-01 2.93476492e-01 -4.17164087e-01 1.80291951e-01
-1.19895673e+00 1.22390831e+00 7.85902381e-01 7.67568424e-02
-1.36469328e+00 -4.87997979e-01 -8.96034002e-01 3.79494466e-02
5.44713497e-01 -9.55214858e-01 1.21650290e+00 -1.29744661e+00
-1.71730721e+00 7.09452212e-01 6.72599733e-01 -9.35120106e-01
9.75952625e-01 7.90894181e-02 -1.84090555e-01 -2.05716819e-01
6.14652522e-02 1.02396488e+00 1.16402042e+00 -1.43359983e+00
-3.26887965e-01 -3.57851326e-01 5.91004193e-01 1.31968364e-01
4.54633892e-01 -3.94825011e-01 3.37147802e-01 -6.87249899e-01
-1.10815890e-01 -1.28503394e+00 -8.29478681e-01 1.38453633e-01
-7.59579659e-01 2.92434245e-01 6.90300822e-01 1.04955649e-02
3.82301033e-01 -1.93045712e+00 3.81058306e-01 2.32277319e-01
-1.96949780e-01 3.24574411e-02 -2.04659715e-01 3.19849014e-01
-4.69299197e-01 3.04907799e-01 -5.53923249e-01 6.91010281e-02
3.48396957e-01 2.51987487e-01 -6.22368336e-01 6.49156511e-01
4.58282083e-01 8.69047523e-01 -7.70839393e-01 1.42742228e-02
3.45798671e-01 1.72327623e-01 -5.24306953e-01 1.07520007e-01
-6.99199557e-01 7.63741136e-01 -5.59146345e-01 -2.77947206e-02
1.75818160e-01 4.07331109e-01 2.03878641e-01 3.55875075e-01
-3.59500945e-02 1.41407056e-02 -1.14264393e+00 1.56832838e+00
-4.58505332e-01 5.15442967e-01 -4.91718017e-02 -1.26035833e+00
8.42594683e-01 3.09460044e-01 5.21625429e-02 -5.92502952e-01
4.29143280e-01 2.95783337e-02 3.56802434e-01 -3.85062695e-01
2.20624447e-01 -5.47150791e-01 -4.24484074e-01 3.72650743e-01
-1.95134044e-01 -7.08450854e-01 -1.48384701e-02 -2.32772246e-01
1.18128300e+00 5.86236537e-01 3.42460930e-01 -3.79159868e-01
3.57227564e-01 2.50134766e-01 2.56586969e-01 8.86826515e-01
-1.37281761e-01 5.53475320e-01 7.79648781e-01 -3.07788014e-01
-1.18861628e+00 -9.99962330e-01 6.40169755e-02 5.66268027e-01
-1.66657213e-02 4.11234468e-01 -8.28729212e-01 -7.21190810e-01
1.25853404e-01 1.43600702e+00 -9.05420363e-01 -7.08127201e-01
-5.55282950e-01 -6.24941230e-01 4.30807173e-01 1.04829289e-01
3.12866896e-01 -1.50721526e+00 -1.33983123e+00 2.38669947e-01
2.67470390e-01 -6.64989352e-01 -6.43748865e-02 4.68642324e-01
-7.29518533e-01 -1.27483201e+00 -5.06954968e-01 -4.62360114e-01
5.81485689e-01 -4.02499527e-01 8.82435083e-01 -3.14503938e-01
-1.61446065e-01 4.53702956e-01 -1.01822622e-01 -7.38207102e-01
-9.61321414e-01 -4.72495049e-01 3.46258909e-01 -9.00718942e-02
-2.29807347e-01 -5.82609355e-01 -5.95009089e-01 1.47503793e-01
-1.22820747e+00 -4.30999100e-02 8.10697436e-01 7.59413838e-01
3.57319146e-01 6.71543777e-02 7.13081896e-01 -7.33697116e-01
8.67132843e-01 -4.75828648e-01 -8.12697053e-01 -3.74013260e-02
-5.91653109e-01 4.84989375e-01 8.47279131e-01 -9.45517302e-01
-1.06264925e+00 4.21994537e-01 1.32524773e-01 -6.01593733e-01
-6.32627010e-01 3.70580047e-01 -4.24870610e-01 1.93011209e-01
1.04462647e+00 1.18898898e-01 2.19628140e-01 -6.22836761e-02
8.32247198e-01 1.35989308e-01 6.47893131e-01 -5.95954359e-01
7.75521338e-01 5.06758451e-01 4.07847792e-01 -3.29294175e-01
-2.21842363e-01 4.40910250e-01 -5.62914312e-01 -1.06629677e-01
8.56464565e-01 -5.59969783e-01 -8.34343970e-01 5.92597835e-02
-1.13285828e+00 -6.76572502e-01 -8.00683558e-01 4.97702003e-01
-1.38901865e+00 -1.73448592e-01 2.09761169e-02 -8.14958334e-01
3.42997342e-01 -1.28734589e+00 7.92857885e-01 7.23138899e-02
-2.61904031e-01 -7.87852466e-01 6.92145824e-02 -3.00747067e-01
1.38770580e-01 6.68444097e-01 1.01098454e+00 -7.28210688e-01
-4.81656611e-01 -1.93203866e-01 2.01905727e-01 1.80245176e-01
-1.06119759e-01 -3.36351156e-01 -8.04755509e-01 -3.20234656e-01
2.91201383e-01 -2.59832799e-01 5.64019322e-01 5.62852681e-01
9.42118108e-01 -7.29971588e-01 -5.53106606e-01 2.79983282e-01
1.37720227e+00 5.70847094e-01 9.82982337e-01 6.37609839e-01
8.89533013e-02 9.16622221e-01 8.63786459e-01 2.36041993e-01
-1.61338270e-01 7.44026959e-01 1.09032476e+00 1.49813190e-01
2.71285325e-01 -4.18431342e-01 4.92742985e-01 -6.22309387e-01
1.48787737e-01 -3.53025228e-01 -5.98240852e-01 5.07479489e-01
-1.92784560e+00 -1.05825388e+00 -4.66325358e-02 2.21845865e+00
3.53117615e-01 4.45413411e-01 -3.84425325e-03 2.74351016e-02
7.58023083e-01 -2.73929238e-02 -6.69512093e-01 -6.49639249e-01
1.14666417e-01 -5.57536408e-02 5.85822761e-01 3.71505737e-01
-1.09188390e+00 6.70933366e-01 6.04381752e+00 4.91786808e-01
-1.18247330e+00 -1.33140564e-01 6.61195755e-01 -1.62547529e-02
-2.35489339e-01 1.49772704e-01 -1.52056962e-01 4.61088657e-01
1.13857639e+00 -7.86535814e-02 5.07593036e-01 8.36780667e-01
5.96135020e-01 -4.29150797e-02 -1.48175752e+00 4.85965818e-01
-1.93736479e-01 -1.35938299e+00 1.88776836e-01 2.35417739e-01
4.65466022e-01 -2.77182758e-01 2.09675640e-01 3.84952158e-01
6.51136577e-01 -1.15760159e+00 1.20961714e+00 6.86508298e-01
6.22538865e-01 -6.42989576e-01 3.87016594e-01 7.27907002e-01
-3.85504127e-01 -4.16313350e-01 -1.82626918e-01 -3.20875317e-01
5.56843504e-02 -6.52254745e-02 -1.04968703e+00 3.04808259e-01
2.94471234e-01 -7.95404464e-02 -9.66916978e-02 6.45328164e-01
-4.76757407e-01 1.39201730e-01 -2.23350450e-01 -2.12875053e-01
3.00944269e-01 -1.88202426e-01 9.15279806e-01 8.24206471e-01
3.54546577e-01 -1.16533615e-01 -2.38179207e-01 1.39707696e+00
2.77695209e-01 -3.65701407e-01 -1.31051755e+00 2.79960990e-01
1.39933273e-01 8.23552668e-01 -6.38189614e-01 -1.16008045e-02
2.12247625e-01 1.06882954e+00 1.65110275e-01 5.53273141e-01
-9.80881035e-01 -3.04477781e-01 5.43828249e-01 1.73723362e-02
2.78863043e-01 -1.92926023e-02 -1.39081091e-01 -8.41594398e-01
-1.39643863e-01 -9.21154082e-01 -3.12281754e-02 -1.05355704e+00
-9.85048175e-01 3.49161804e-01 2.40169257e-01 -1.40154815e+00
-7.73522139e-01 -8.16275656e-01 -7.63161838e-01 6.68245912e-01
-1.12992656e+00 -9.38662529e-01 3.14910829e-01 5.08122444e-01
3.74044359e-01 -1.63538709e-01 6.83750868e-01 -3.90895993e-01
-2.40789786e-01 -1.19616993e-01 -1.09408908e-01 -2.69034028e-01
4.51770186e-01 -1.39893985e+00 3.86703789e-01 8.45831454e-01
1.43191904e-01 4.26571965e-01 1.39246762e+00 -4.99826998e-01
-1.27327538e+00 -1.22551656e+00 2.64484167e-01 -5.64734459e-01
6.44361556e-01 -2.47821391e-01 -6.16393447e-01 8.28139961e-01
-1.55934934e-02 -1.62657127e-01 -9.98645090e-03 -3.93703431e-01
2.91752648e-02 2.34913677e-01 -1.46763182e+00 1.09069800e+00
9.31797385e-01 -1.46584973e-01 -1.06563485e+00 2.51522928e-01
7.21448362e-01 -1.57015204e-01 -4.99960124e-01 3.53599250e-01
4.65719461e-01 -8.38028371e-01 1.02378190e+00 -1.03910124e+00
5.78034759e-01 -2.94150591e-01 -1.04425289e-01 -1.86659360e+00
-1.28451034e-01 -7.46270478e-01 2.61214375e-01 8.18402946e-01
4.90302116e-01 -7.10457265e-01 6.60554171e-01 5.07606149e-01
-2.76236653e-01 -6.20628536e-01 -1.24470949e+00 -7.34203339e-01
3.85996163e-01 -5.62457621e-01 6.22915864e-01 8.04694593e-01
-1.06172226e-01 2.92209625e-01 -2.22735122e-01 3.49459231e-01
2.58400619e-01 -1.08216777e-01 9.67834711e-01 -1.03025591e+00
-3.20790678e-01 -5.35872102e-01 -4.07307178e-01 -6.21554792e-01
4.44766432e-01 -5.74142873e-01 4.86300439e-01 -1.20942760e+00
-3.27493936e-01 -3.08560252e-01 6.13823766e-03 1.68994799e-01
2.90646017e-01 -9.16396752e-02 4.06697839e-01 1.87082440e-02
-4.97099496e-02 6.79016173e-01 1.22684562e+00 -2.25828603e-01
-2.32127249e-01 1.75927624e-01 -6.06605291e-01 9.59997714e-01
7.66519606e-01 -5.04432678e-01 -4.59296316e-01 2.38985643e-02
2.46370986e-01 3.44007194e-01 8.17827880e-01 -9.00843143e-01
-2.58275032e-01 -4.53632861e-01 3.74345243e-01 1.21087395e-01
3.21809739e-01 -1.14543629e+00 3.81780356e-01 9.62996364e-01
-4.52740341e-01 -4.30434719e-02 2.43148997e-01 9.04643595e-01
1.18208401e-01 -4.60597277e-01 7.52951384e-01 -3.23484987e-01
-6.51954710e-01 -1.16601467e-01 -7.22231865e-01 -4.13104415e-01
1.30206203e+00 -1.61072344e-01 -2.26236343e-01 -5.52640438e-01
-8.63081038e-01 4.86276485e-02 5.64454377e-01 5.89606643e-01
5.18658280e-01 -1.14551890e+00 -5.32481372e-01 1.10243164e-01
5.66939563e-02 -1.87610939e-01 -7.21164867e-02 4.23885256e-01
-3.37120712e-01 3.73315454e-01 -4.53011334e-01 -3.47307771e-01
-4.92902219e-01 1.11985481e+00 7.14177489e-01 -2.11590961e-01
-4.24086273e-01 2.64966398e-01 6.57418489e-01 -6.59639239e-01
-2.09416702e-01 -5.39686918e-01 -2.68222064e-01 -3.45516831e-01
2.99887322e-02 -5.06190732e-02 -2.14799583e-01 -3.77244949e-01
-1.58055406e-02 -4.95603643e-02 4.85879213e-01 -4.62207794e-01
1.15490115e+00 -1.25199363e-01 3.22145641e-01 3.79959106e-01
7.14154899e-01 -3.74008060e-01 -1.69028854e+00 4.10127699e-01
1.35137856e-01 -1.46613419e-01 -2.13982537e-01 -9.65401828e-01
-6.37251079e-01 6.33658171e-01 8.12973142e-01 3.40390503e-01
8.91666591e-01 -1.13661326e-01 -2.20536098e-01 6.74163342e-01
4.06702965e-01 -9.42226112e-01 8.54215473e-02 2.58391351e-01
1.54445672e+00 -1.26663125e+00 -2.62326777e-01 1.23322271e-01
-6.78569376e-01 9.45339024e-01 4.99737322e-01 -3.93705785e-01
3.22711647e-01 4.98749800e-02 -1.54002696e-01 -2.42404073e-01
-6.64711058e-01 -1.47703692e-01 1.56544745e-01 8.71238708e-01
-1.14677206e-01 1.19198941e-01 -2.35137478e-01 3.54519516e-01
-3.31014007e-01 -2.08031456e-03 8.46699536e-01 8.15856874e-01
-2.25501865e-01 -8.43403876e-01 -5.49466074e-01 3.37836623e-01
-3.24491978e-01 2.45889887e-01 -3.57927322e-01 1.39567304e+00
3.04522157e-01 7.34250724e-01 2.18498692e-01 -4.72490303e-02
6.19290352e-01 9.80572477e-02 5.11295259e-01 -6.47594154e-01
-1.75446376e-01 -2.38813713e-01 1.53421909e-01 -5.67669749e-01
-4.15660620e-01 -7.78980792e-01 -1.05517030e+00 1.48938343e-01
-1.18783906e-01 -1.56949654e-01 7.90996850e-01 1.16492283e+00
1.15168318e-01 7.79677510e-01 3.60112607e-01 -1.28688562e+00
-8.67413640e-01 -1.00974095e+00 -4.41766948e-01 5.06209075e-01
4.40165311e-01 -8.72785747e-01 -3.73239517e-01 1.22268707e-01] | [4.502810478210449, 1.8757786750793457] |
ff6e57aa-5df2-4998-8554-2f07e1c41438 | rethinking-class-relations-absolute-relative | 2001.03919 | null | https://arxiv.org/abs/2001.03919v4 | https://arxiv.org/pdf/2001.03919v4.pdf | Rethinking Class Relations: Absolute-relative Supervised and Unsupervised Few-shot Learning | The majority of existing few-shot learning methods describe image relations with binary labels. However, such binary relations are insufficient to teach the network complicated real-world relations, due to the lack of decision smoothness. Furthermore, current few-shot learning models capture only the similarity via relation labels, but they are not exposed to class concepts associated with objects, which is likely detrimental to the classification performance due to underutilization of the available class labels. To paraphrase, children learn the concept of tiger from a few of actual examples as well as from comparisons of tiger to other animals. Thus, we hypothesize that in fact both similarity and class concept learning must be occurring simultaneously. With these observations at hand, we study the fundamental problem of simplistic class modeling in current few-shot learning methods. We rethink the relations between class concepts, and propose a novel Absolute-relative Learning paradigm to fully take advantage of label information to refine the image representations and correct the relation understanding in both supervised and unsupervised scenarios. Our proposed paradigm improves the performance of several the state-of-the-art models on publicly available datasets. | ['Songlei Jian', 'Piotr Koniusz', 'Hongdong Li', 'Hongguang Zhang', 'Philip H. S. Torr'] | 2020-01-12 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Zhang_Rethinking_Class_Relations_Absolute-Relative_Supervised_and_Unsupervised_Few-Shot_Learning_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Zhang_Rethinking_Class_Relations_Absolute-Relative_Supervised_and_Unsupervised_Few-Shot_Learning_CVPR_2021_paper.pdf | cvpr-2021-1 | ['unsupervised-few-shot-learning', 'unsupervised-few-shot-image-classification'] | ['computer-vision', 'computer-vision'] | [ 5.61325550e-01 3.12566608e-01 -4.57807690e-01 -6.54723406e-01
2.09794529e-02 -3.13401222e-01 7.19135463e-01 6.94436252e-01
-2.43655160e-01 6.27530515e-01 -1.73383877e-01 2.32580483e-01
-4.21538264e-01 -1.09462476e+00 -7.59669781e-01 -4.43190902e-01
5.15531786e-02 6.38421535e-01 5.14055669e-01 -1.71896845e-01
1.17113173e-03 8.49769711e-02 -2.25642800e+00 2.85637945e-01
1.09078896e+00 9.26731288e-01 1.59347236e-01 2.17979327e-01
-3.50256145e-01 1.19023204e+00 -3.09777796e-01 -6.63362145e-01
-8.40907171e-02 -5.00740707e-01 -8.53428960e-01 8.39941204e-02
6.69042289e-01 -3.68952215e-01 -3.34150553e-01 1.38367832e+00
-7.52259493e-02 5.73888958e-01 9.76685643e-01 -1.47770667e+00
-7.15663552e-01 8.31398070e-01 -4.22505379e-01 1.99484617e-01
2.49191031e-01 -1.59069449e-01 1.20370650e+00 -6.36085570e-01
8.01542163e-01 1.10003316e+00 4.35827255e-01 6.43586397e-01
-1.30475497e+00 -7.03846157e-01 5.19353867e-01 8.59674931e-01
-1.40344834e+00 -2.91503161e-01 7.80326247e-01 -5.96025229e-01
8.62512290e-01 7.72891473e-03 1.07418501e+00 9.48160827e-01
-3.20367545e-01 7.27543712e-01 8.98664057e-01 -5.39852679e-01
4.04823393e-01 1.51270360e-01 5.82247198e-01 7.82833755e-01
3.07360470e-01 8.46874416e-02 -7.16722012e-01 1.60627797e-01
3.93125504e-01 3.19873005e-01 -1.09318808e-01 -8.21192920e-01
-7.86441505e-01 7.44033992e-01 6.22923911e-01 4.36586499e-01
-1.66670419e-02 9.77425426e-02 3.14001441e-02 3.05283695e-01
5.93849540e-01 4.29162711e-01 -1.68131381e-01 1.01435918e-03
-7.99366295e-01 -4.32983302e-02 7.22105324e-01 1.31076658e+00
1.25618339e+00 -3.86846811e-01 9.61713940e-02 9.63434577e-01
2.17728332e-01 -8.33669968e-04 4.29189801e-01 -8.00970256e-01
1.79613218e-01 7.69074023e-01 -4.24449712e-01 -9.64732051e-01
-9.52349454e-02 -3.68725747e-01 -5.83805323e-01 7.58791342e-02
2.99895614e-01 3.75563771e-01 -1.14352667e+00 1.74077249e+00
3.18902045e-01 8.56268644e-01 4.74604480e-02 5.11201203e-01
1.10628152e+00 3.76821846e-01 1.77154139e-01 -3.32310617e-01
1.22310412e+00 -9.60588217e-01 -7.63857126e-01 -4.42891359e-01
7.62187123e-01 -2.58602351e-01 8.26238573e-01 1.21318139e-01
-6.00246251e-01 -5.61405480e-01 -1.33346522e+00 -1.36333415e-02
-7.64294744e-01 -5.68419755e-01 1.01600933e+00 5.06673753e-01
-5.04551411e-01 8.99648786e-01 -8.63558173e-01 -7.19769299e-01
8.94578218e-01 2.39383385e-01 -3.67311239e-01 -5.63793361e-01
-1.25231791e+00 1.09859979e+00 6.51168942e-01 -2.52910793e-01
-9.86879408e-01 -8.35176229e-01 -1.12889361e+00 3.46588552e-01
6.31532371e-01 -3.55113268e-01 1.13243139e+00 -7.91308165e-01
-1.03769958e+00 1.03480089e+00 1.06898159e-01 -4.09193397e-01
1.11646548e-01 -1.41864553e-01 -2.30136037e-01 2.55361438e-01
-8.01459327e-02 8.20204318e-01 6.77243829e-01 -1.27567697e+00
-7.30734169e-01 -2.27776870e-01 3.92076790e-01 2.73437917e-01
-4.79764640e-01 -3.63705814e-01 -1.88715130e-01 -3.91731977e-01
3.12160164e-01 -6.75525427e-01 -4.24298532e-02 3.37466538e-01
-7.18331859e-02 -4.25198197e-01 5.97790301e-01 1.52554780e-01
8.53444517e-01 -2.05407858e+00 -1.76263060e-02 -3.33078690e-02
3.96735996e-01 3.37986469e-01 -1.67094961e-01 3.36641073e-01
-2.89446980e-01 -1.13317616e-01 -1.31450504e-01 -1.56318054e-01
-2.91017443e-01 6.65519834e-01 -2.20920116e-01 3.93816978e-01
1.11701228e-01 8.32091630e-01 -1.41044343e+00 -7.83805549e-01
3.44373733e-01 2.68827707e-01 -5.30011892e-01 3.27441186e-01
-3.45676988e-01 1.44770756e-01 -2.50508606e-01 6.03421986e-01
4.16191041e-01 -2.51382858e-01 2.37385839e-01 -1.62923902e-01
2.23095462e-01 2.53108412e-01 -1.06175160e+00 1.89068902e+00
-3.23741555e-01 6.51784778e-01 -7.20524192e-01 -1.47456539e+00
7.75667191e-01 1.98527977e-01 4.73699123e-01 -5.09136736e-01
1.80453062e-01 -1.03489503e-01 2.19741300e-01 -6.79437339e-01
-3.80803570e-02 -4.37673748e-01 4.60139126e-01 4.36209083e-01
6.09793961e-01 -3.20388794e-01 3.57706606e-01 3.60401630e-01
8.86500001e-01 2.96651334e-01 6.93059087e-01 -7.60165676e-02
2.19965950e-02 5.65499887e-02 7.92759657e-01 8.46407712e-01
-2.45106936e-01 5.24107814e-01 3.72376055e-01 -4.02815163e-01
-5.30903041e-01 -1.00146556e+00 -8.43245164e-02 1.35619366e+00
5.43927729e-01 -6.46115482e-01 -4.58064020e-01 -6.96940780e-01
-2.13537857e-01 9.53090489e-01 -9.22991216e-01 -5.35887480e-01
-1.77949980e-01 -5.83752930e-01 2.03554049e-01 4.32344347e-01
2.76214778e-01 -8.94118547e-01 -9.19935167e-01 1.83548287e-01
9.89828035e-02 -1.01054156e+00 1.14234254e-01 5.23250937e-01
-8.37174356e-01 -1.30900109e+00 -3.12507302e-01 -9.44303393e-01
7.61677742e-01 3.87313575e-01 1.11003578e+00 4.56689835e-01
-4.45446581e-01 3.19200993e-01 -6.88522160e-01 -4.04232442e-01
-2.02328086e-01 -2.29564503e-01 2.17842050e-02 -6.37223572e-02
6.64201915e-01 -9.69880879e-01 -2.52241552e-01 1.51891097e-01
-7.74969101e-01 2.92714655e-01 3.50868195e-01 9.63863671e-01
4.31002825e-01 3.75311464e-01 3.56953025e-01 -1.27874589e+00
-5.33605553e-03 -5.55355728e-01 -4.24723893e-01 6.94948852e-01
-7.26501286e-01 1.52054131e-01 2.76974410e-01 -1.05395210e+00
-1.19326985e+00 5.04135676e-02 2.94800311e-01 -6.24158859e-01
-3.80156517e-01 5.09325325e-01 -8.89526606e-02 -1.26409441e-01
6.81352258e-01 1.05762996e-01 -2.65594631e-01 -4.11009133e-01
6.68300807e-01 2.76353568e-01 4.67302948e-01 -6.71288788e-01
7.09195793e-01 5.50961018e-01 1.26576543e-01 -8.27935874e-01
-1.47018433e+00 -5.71707547e-01 -1.02977073e+00 -2.20369264e-01
8.27924132e-01 -7.63971686e-01 -4.45946038e-01 3.28083098e-01
-1.03415990e+00 -1.77255198e-01 -4.08384085e-01 5.13263345e-01
-6.23624146e-01 2.11261928e-01 -5.12776017e-01 -5.28289735e-01
2.67412782e-01 -8.80862236e-01 4.13684338e-01 5.12485981e-01
-1.99779257e-01 -9.58601892e-01 1.26218081e-01 1.71985716e-01
7.79404640e-02 -4.45368774e-02 1.42521906e+00 -1.03669035e+00
-4.86270547e-01 -1.88403279e-02 -1.95579186e-01 4.10672463e-02
3.01468521e-01 5.96089326e-02 -1.16766977e+00 -8.25671479e-02
-1.08073018e-01 -6.91646516e-01 9.23262656e-01 7.47126266e-02
1.03102803e+00 -3.72647718e-02 -5.11456788e-01 5.22484303e-01
1.28888214e+00 2.25076988e-01 4.41231877e-01 1.78858303e-02
8.60116005e-01 1.12257802e+00 8.78205895e-01 2.77731955e-01
3.80582362e-01 6.15999758e-01 4.31285292e-01 3.83177847e-01
-3.23967189e-01 -5.48177481e-01 -2.32673377e-01 8.57233346e-01
1.97224766e-02 -1.05217835e-02 -1.10790718e+00 7.06699729e-01
-1.99807072e+00 -1.02279603e+00 1.72985867e-01 2.07862711e+00
1.10878348e+00 2.10294232e-01 -3.34387481e-01 1.31424040e-01
8.59071970e-01 1.33770734e-01 -5.15807509e-01 8.24596062e-02
1.56706020e-01 2.22576067e-01 1.05066083e-01 2.49589548e-01
-1.11551738e+00 1.18771183e+00 6.17210579e+00 9.17323172e-01
-7.67725885e-01 6.90607056e-02 3.60363305e-01 2.67703235e-02
-1.21989533e-01 3.93124580e-01 -7.10429609e-01 7.66354874e-02
5.85333169e-01 -2.41267383e-01 3.19313794e-01 1.01787615e+00
-6.21842504e-01 -1.85125709e-01 -1.83659458e+00 1.08173728e+00
4.32931215e-01 -1.22889900e+00 1.23445131e-01 -2.13128969e-01
7.47531116e-01 -3.20223838e-01 -1.94666088e-01 5.58153749e-01
3.67636085e-01 -1.08919632e+00 6.55314028e-01 4.79923487e-01
5.76996982e-01 -5.04529357e-01 5.47280073e-01 6.12746835e-01
-1.16379142e+00 -2.68419906e-02 -5.61112344e-01 -5.34797847e-01
-1.88789979e-01 3.62673044e-01 -9.32158351e-01 2.95489073e-01
5.94873667e-01 1.16112959e+00 -8.16797495e-01 1.14774323e+00
-6.04734302e-01 4.59810972e-01 -3.17222118e-01 -8.81644860e-02
-2.99745630e-02 3.89074013e-02 9.05534998e-02 6.16252840e-01
2.21807342e-02 6.15368426e-01 2.32803956e-01 9.39287245e-01
3.08968462e-02 -1.01913184e-01 -6.44266069e-01 -1.95126474e-01
7.76517212e-01 1.04193401e+00 -1.03683150e+00 -6.24484241e-01
-7.21072853e-01 5.68073511e-01 7.54592299e-01 1.95765957e-01
-5.03950417e-01 -2.50655711e-01 6.31036818e-01 -1.57791108e-01
2.67482787e-01 7.83843175e-02 -1.90745518e-01 -1.32872617e+00
-3.59499693e-01 -5.36463976e-01 5.52707672e-01 -6.87142432e-01
-1.36180079e+00 1.96939945e-01 4.80016619e-01 -1.34892559e+00
-1.94109246e-01 -3.73778492e-01 -7.68387139e-01 2.35759094e-01
-1.53210580e+00 -1.26454806e+00 -3.69646907e-01 4.24973339e-01
6.92052364e-01 -6.39059693e-02 9.30127919e-01 2.52124041e-01
-3.65550607e-01 4.97501522e-01 -3.13890159e-01 -2.64413059e-02
6.91628754e-01 -1.00284290e+00 1.50877193e-01 7.47667313e-01
6.34022295e-01 7.70652056e-01 8.95063519e-01 -7.25222349e-01
-8.70915234e-01 -7.07343996e-01 8.30167651e-01 -2.25678831e-01
7.43908465e-01 -4.27288651e-01 -1.28196943e+00 6.23195589e-01
-7.71680400e-02 3.85552764e-01 9.52038050e-01 4.10904258e-01
-7.17589259e-01 7.79933354e-04 -1.00820363e+00 4.82937872e-01
1.32254040e+00 -5.75224876e-01 -1.21888757e+00 2.51287609e-01
7.84156919e-01 3.37919034e-02 -6.92546546e-01 4.55208421e-01
5.34451485e-01 -7.96517909e-01 9.22367573e-01 -8.05412650e-01
6.51874065e-01 -1.55469850e-01 -2.23957285e-01 -1.36502302e+00
-1.73180461e-01 8.08954239e-03 -3.61562252e-01 1.43381488e+00
1.17929749e-01 -2.40775496e-01 8.55403900e-01 7.41371393e-01
7.30084851e-02 -7.59518385e-01 -7.86220312e-01 -8.78103793e-01
-2.39928495e-02 -4.43544537e-01 5.76960742e-01 1.42642486e+00
3.72195512e-01 4.07671332e-01 -3.60203117e-01 1.20550536e-01
8.41859281e-01 3.16686094e-01 2.62830466e-01 -1.70347476e+00
-3.36402386e-01 -1.73692524e-01 -8.59907448e-01 -7.56755769e-01
3.86638254e-01 -9.07826722e-01 2.20128730e-01 -1.49865687e+00
5.17659903e-01 -6.96073174e-01 -4.06110913e-01 7.18151748e-01
-3.61544758e-01 1.85350075e-01 2.29245260e-01 1.87392145e-01
-7.67713666e-01 6.88878894e-01 1.03367293e+00 -4.69441712e-01
8.57815519e-02 -9.72495377e-02 -4.09770668e-01 1.09305573e+00
3.34588259e-01 -8.15989912e-01 -9.33125973e-01 -4.08038974e-01
2.97640681e-01 -1.77144602e-01 2.28706330e-01 -1.11278689e+00
5.50279617e-01 -5.83920240e-01 2.77109027e-01 -4.03594285e-01
4.98861223e-01 -1.01149786e+00 -3.01445611e-02 4.02326137e-01
-6.82821929e-01 -7.53215492e-01 -3.18746746e-01 7.58402944e-01
-1.64714649e-01 -5.69397807e-01 9.68444645e-01 -1.50281399e-01
-1.42222714e+00 4.09894168e-01 -4.61810939e-02 2.04748616e-01
1.11269927e+00 -3.19078028e-01 -5.41231692e-01 -2.78287143e-01
-6.92533314e-01 2.88772374e-01 5.03783524e-01 4.37515050e-01
7.07572818e-01 -1.21327806e+00 -1.69818729e-01 2.45705068e-01
6.74747288e-01 1.45038649e-01 2.69818515e-01 5.09691060e-01
-2.31113210e-01 1.00470111e-02 -4.95847046e-01 -4.69404906e-01
-1.33111298e+00 9.67618108e-01 1.48053572e-01 2.21324533e-01
-8.31479967e-01 1.22782540e+00 3.93979490e-01 -2.17673346e-01
5.41955411e-01 -2.18715221e-01 -6.95390999e-01 4.59086895e-01
5.76060772e-01 1.75310060e-01 -1.85297430e-01 -5.15418291e-01
-3.28327268e-01 5.98544776e-01 -3.98333997e-01 2.07539752e-01
1.31159067e+00 -8.45418721e-02 5.63170724e-02 8.16905141e-01
1.03296530e+00 -6.64703548e-01 -1.12319040e+00 -5.89691520e-01
1.58243403e-01 -6.42248213e-01 -1.92684576e-01 -5.42413771e-01
-8.09541523e-01 1.14231002e+00 5.24077356e-01 -1.02989271e-01
9.38836336e-01 3.12848270e-01 4.02427375e-01 6.83607221e-01
4.91972268e-01 -1.00064540e+00 2.15728119e-01 4.56112057e-01
1.30059779e-01 -1.44766271e+00 3.20812285e-01 -1.01374125e+00
-4.25930530e-01 9.47377086e-01 8.67043912e-01 2.41788779e-03
7.53938735e-01 -5.23265973e-02 -1.66482836e-01 -3.57445002e-01
-1.15835297e+00 -4.52513844e-01 3.24364930e-01 8.33857179e-01
4.01022822e-01 -5.52053116e-02 -1.62982285e-01 4.86347497e-01
-1.17465574e-03 -1.54935762e-01 5.86416245e-01 1.13527119e+00
-7.24044204e-01 -1.02229679e+00 2.13671848e-01 6.02221191e-01
-6.12187237e-02 -1.51396006e-01 -2.63453066e-01 4.96664047e-01
5.22457361e-01 9.24843788e-01 1.75814390e-01 -3.22444826e-01
2.01634690e-02 1.58456728e-01 6.61071658e-01 -1.32652164e+00
-7.38871098e-02 -4.62158442e-01 -2.79449746e-02 -3.41016412e-01
-6.98690057e-01 -3.02596569e-01 -1.16952980e+00 -1.22263245e-02
-4.93726254e-01 4.58613448e-02 3.74881804e-01 1.51826668e+00
-2.00283974e-01 4.88725960e-01 1.83717310e-01 -6.31190658e-01
-3.77039522e-01 -8.49150121e-01 -7.09882677e-01 6.34868205e-01
1.41999975e-01 -1.27643895e+00 -3.26161057e-01 1.55891418e-01] | [10.168471336364746, 2.5303750038146973] |
3ae90784-eba3-4e18-b6ce-03bb8c9e3947 | how-old-is-gpt-the-humbel-framework-for | 2305.14195 | null | https://arxiv.org/abs/2305.14195v2 | https://arxiv.org/pdf/2305.14195v2.pdf | How Old is GPT?: The HumBEL Framework for Evaluating Language Models using Human Demographic Data | While large pre-trained language models (LMs) find greater use across NLP, existing evaluation protocols do not consider how LM language use aligns with particular human demographic groups, which can be an important consideration in conversational AI applications. To remedy this gap, we consider how LM language skills can be measured and compared to human sub-populations. We suggest clinical techniques from Speech Language Pathology, which has well-established norms for acquisition of language skills, organized by (human) age. We conduct evaluation with a domain expert (i.e., a clinically licensed speech language pathologist), and also propose automated techniques to substitute clinical evaluation at scale. We find LM capability varies widely depending on task with GPT-3.5 mimicking the ability of a typical 6-9 year old at tasks requiring inference about word meanings and simultaneously outperforming a typical 21 year old at memorization. GPT-3.5 (InstructGPT) also has trouble with social language use, exhibiting less than 50\% of the tested pragmatic skills. It shows errors in understanding particular word parts-of-speech and associative word relations, among other lexical features. Ultimately, findings reiterate the importance of considering demographic alignment and conversational goals when using these models as public-facing tools. Our framework will be publicly available via code, data, and a python package. | ['Malihe Alikhani', 'Jennifer C. Gates', 'Anthony Sicilia'] | 2023-05-23 | null | null | null | null | ['memorization'] | ['natural-language-processing'] | [-1.61453649e-01 6.89945579e-01 -2.13428900e-01 -1.91119641e-01
-8.91304910e-01 -3.77566040e-01 4.69550133e-01 2.71284282e-01
-7.40896821e-01 4.50550795e-01 7.89738774e-01 -6.69919252e-01
-3.12465996e-01 -3.02674115e-01 -9.12359655e-02 -1.73116148e-01
1.02056101e-01 8.71100247e-01 -9.43025500e-02 -2.18855411e-01
2.57204086e-01 1.13218889e-01 -1.02088916e+00 4.56822425e-01
1.22139812e+00 1.43922716e-01 5.66306710e-01 5.55864215e-01
-1.96656540e-01 8.83664906e-01 -6.54504776e-01 -5.22030115e-01
-4.17517275e-01 -3.99912059e-01 -1.26191401e+00 -1.03620850e-01
6.25321507e-01 -6.13660991e-01 -1.07116409e-01 7.79818237e-01
6.33480430e-01 -2.77523566e-02 6.53062582e-01 -8.55640531e-01
-8.28723550e-01 1.27119124e+00 1.76071942e-01 4.85541314e-01
7.81692028e-01 5.12497306e-01 9.86377776e-01 -5.63140631e-01
6.40237808e-01 1.62014377e+00 7.05227733e-01 1.00695884e+00
-1.11620271e+00 -7.27436364e-01 3.89791340e-01 7.50811920e-02
-8.53536248e-01 -6.97792172e-01 3.20554972e-01 -6.85287893e-01
1.51059067e+00 -6.30372092e-02 9.76503432e-01 1.33604169e+00
-3.26847136e-01 6.34242237e-01 1.37062621e+00 -4.29440111e-01
2.59339623e-02 2.81429410e-01 2.82185167e-01 4.07268554e-01
2.68399835e-01 -2.58646876e-01 -1.01962137e+00 -2.62054324e-01
1.78711027e-01 -8.09558153e-01 -2.41772920e-01 4.35231090e-01
-1.51474285e+00 5.87654889e-01 -1.19050346e-01 9.56873357e-01
-1.12317897e-01 -2.05370426e-01 4.78894442e-01 6.72889173e-01
2.88112134e-01 7.63710380e-01 -5.06678224e-01 -9.00341392e-01
-4.67506617e-01 6.84430376e-02 1.00752544e+00 7.71684349e-01
-8.48258436e-02 2.78694853e-02 4.61151777e-03 1.62840188e+00
3.71913821e-01 5.11637330e-01 9.75487173e-01 -9.99047101e-01
6.11359537e-01 6.29058838e-01 -4.78282362e-01 -4.92306352e-01
-6.74819589e-01 -9.98524874e-02 -4.38594557e-02 -1.65975690e-01
7.98485041e-01 -7.05539435e-02 -4.42346692e-01 2.34097433e+00
1.13618903e-01 -4.00654286e-01 6.33118674e-02 1.37664706e-01
9.43196177e-01 1.70451730e-01 8.24118972e-01 -5.33852637e-01
1.40355742e+00 -4.87175882e-01 -4.16493356e-01 -9.35946703e-01
1.21836329e+00 -7.22678840e-01 1.37756968e+00 4.05691326e-01
-1.49117315e+00 -3.55899602e-01 -9.46591139e-01 -1.38747692e-01
-2.40557402e-01 -2.47868806e-01 7.54380524e-01 1.02013028e+00
-1.55148089e+00 4.08446968e-01 -8.04084718e-01 -9.57410753e-01
1.15367338e-01 6.39987886e-01 -5.92719972e-01 2.62760788e-01
-1.16179490e+00 1.45850945e+00 3.55563939e-01 -3.50659370e-01
-3.17424506e-01 -5.47922015e-01 -8.30451608e-01 -3.53007674e-01
-1.62939653e-01 -6.21836126e-01 1.55563068e+00 -6.75915897e-01
-1.47975993e+00 1.43470359e+00 -2.17590779e-02 -1.28373846e-01
1.82990789e-01 -1.71431750e-01 -7.81941772e-01 2.10418224e-01
3.20694298e-01 6.11608088e-01 4.53076422e-01 -5.61442435e-01
-6.98276222e-01 -3.97132546e-01 3.55835527e-01 6.35550439e-01
-6.49739146e-01 4.16570544e-01 1.62122250e-01 -4.44412619e-01
2.14697361e-01 -7.87197173e-01 1.84191525e-01 -2.31387541e-02
-3.76310915e-01 -6.29025042e-01 1.00043811e-01 -1.08179200e+00
1.51297414e+00 -1.94872320e+00 5.94099015e-02 1.63641542e-01
3.31874669e-01 1.36281863e-01 -2.70799603e-02 6.78899765e-01
8.85272548e-02 5.62671065e-01 -2.72701820e-03 -4.48886544e-01
1.45094618e-01 -8.85088742e-02 2.05522403e-01 3.71935755e-01
-1.10065900e-01 7.32822776e-01 -9.91962373e-01 -5.08697808e-01
4.80926633e-02 8.92953798e-02 -7.81291187e-01 2.21406184e-02
8.98188129e-02 6.82413459e-01 -7.12820739e-02 4.68335092e-01
-9.91532952e-03 -1.07781552e-01 4.92501408e-01 6.46178961e-01
-6.19956292e-02 1.01467144e+00 -5.16511023e-01 1.32257915e+00
-6.34440362e-01 6.66211128e-01 3.22672427e-01 -6.05033338e-01
5.50008714e-01 4.25906688e-01 -1.55090271e-02 -4.74045455e-01
8.86877552e-02 5.54981112e-01 1.21806788e+00 -6.45661950e-01
3.58898146e-03 -1.89574540e-01 1.41386017e-01 5.91647267e-01
-1.85493529e-01 -5.22194624e-01 1.45141184e-01 1.15978077e-01
1.26176786e+00 -5.05928397e-01 5.06184280e-01 -6.15756691e-01
7.18244314e-01 -1.80173635e-01 3.86261165e-01 7.21443832e-01
-2.51000315e-01 -2.08670329e-02 6.12980664e-01 2.93658137e-01
-9.78352845e-01 -8.97422373e-01 -2.20124438e-01 1.45812058e+00
-6.24153733e-01 -3.77697557e-01 -1.09925008e+00 -1.34973362e-01
-9.48895738e-02 1.24561954e+00 -1.63838297e-01 -2.52732426e-01
-7.73898721e-01 -1.72937587e-01 7.00461924e-01 4.62522984e-01
1.49307251e-01 -1.58723116e+00 4.72786129e-02 3.72251689e-01
-1.86198324e-01 -1.27118516e+00 -3.91145140e-01 -4.03045267e-01
-4.99230325e-01 -7.77419806e-01 -6.29341722e-01 -1.13032949e+00
4.11516368e-01 -2.67191470e-01 9.74786758e-01 5.23721576e-01
1.97154522e-01 9.16643798e-01 -2.73485363e-01 -3.61657828e-01
-1.13218021e+00 3.73457700e-01 5.11225700e-01 -8.80665123e-01
7.38373160e-01 -7.76402771e-01 -4.74864364e-01 6.09502085e-02
-2.10080236e-01 2.17137963e-01 5.70366740e-01 6.92880750e-01
-4.07301366e-01 -8.18868458e-01 7.12164700e-01 -7.15153813e-01
1.09076786e+00 -5.23757279e-01 2.45439723e-01 1.63413256e-01
-5.46382546e-01 -4.25382078e-01 3.01169574e-01 -9.60186720e-01
-7.52536416e-01 -5.86752832e-01 -5.70444167e-01 5.27622640e-01
-2.40466073e-01 3.57884675e-01 -4.45401706e-02 1.60819590e-01
6.37667179e-01 9.14764106e-02 4.35186952e-01 -3.49786669e-01
6.60837144e-02 1.21302414e+00 3.51528347e-01 -9.43811715e-01
4.84827012e-01 -1.43690631e-01 -8.39487731e-01 -1.41933417e+00
-6.03257537e-01 -1.51051059e-01 -4.63493854e-01 -1.42320752e-01
8.81024778e-01 -8.19299757e-01 -9.38228667e-01 7.33442783e-01
-9.09894168e-01 -8.75211537e-01 8.49962905e-02 8.95289004e-01
-7.89474249e-01 3.64561796e-01 -7.56328166e-01 -7.49802649e-01
-5.44202924e-01 -1.19782794e+00 7.44013608e-01 -1.60620406e-01
-1.35988057e+00 -1.14883590e+00 -2.95066629e-02 7.25069523e-01
3.21930319e-01 -2.89168149e-01 1.50277901e+00 -1.36328638e+00
1.59583434e-01 1.63937479e-01 5.46892881e-02 3.93640131e-01
-1.19593263e-01 -4.16147500e-01 -6.11180305e-01 -2.45833606e-01
-2.20714957e-02 -5.74520588e-01 3.31424624e-02 1.76567942e-01
6.81864560e-01 -2.00706825e-01 -2.18564257e-01 1.63527578e-01
4.66520607e-01 3.91675442e-01 1.08396322e-01 2.18058348e-01
4.71526921e-01 9.75156903e-01 -8.09381437e-03 6.29462376e-02
9.11998272e-01 5.38974404e-01 -5.45677185e-01 5.39641559e-01
-3.25401872e-01 -3.70880634e-01 7.51721859e-01 1.51910293e+00
2.64576167e-01 -8.73023421e-02 -1.42977202e+00 6.65493608e-01
-1.15216947e+00 -8.32134843e-01 2.42644083e-02 2.05293202e+00
1.17328703e+00 3.27524006e-01 3.21426570e-01 6.23033345e-02
5.99347591e-01 6.52014883e-03 -8.27975512e-01 -7.19133735e-01
2.38393098e-02 2.54213691e-01 6.73207082e-03 9.07758296e-01
-2.17752278e-01 1.18062091e+00 6.60567999e+00 6.98771477e-01
-9.64879334e-01 3.37919593e-01 8.04039061e-01 4.39172201e-02
-4.67270076e-01 -2.01058269e-01 -7.72995114e-01 4.50679988e-01
1.29620230e+00 -4.99043941e-01 6.76747680e-01 5.46832085e-01
1.98138192e-01 -1.67397797e-01 -1.73775613e+00 8.74879241e-01
1.12017937e-01 -5.19353747e-01 -2.00889364e-01 1.69686690e-01
1.20831534e-01 -4.00987528e-02 1.53928593e-01 6.46416068e-01
2.99355149e-01 -1.15580201e+00 7.27672458e-01 3.40206474e-01
1.05967164e+00 -2.00316936e-01 1.62793890e-01 5.38743377e-01
-7.52499163e-01 -2.02375159e-01 1.46840245e-01 -6.33848965e-01
1.78345233e-01 -6.95125535e-02 -1.17714059e+00 -5.76064289e-01
3.62830013e-01 2.43745387e-01 -3.41070801e-01 5.63423932e-01
-2.90638000e-01 7.12790668e-01 -6.54883981e-01 -6.62044942e-01
-1.25305718e-02 2.24017650e-01 5.76538503e-01 9.70975816e-01
5.03383398e-01 2.89992124e-01 1.61268972e-02 3.72982591e-01
2.10812196e-01 7.03889728e-01 -4.95956987e-01 -5.69861948e-01
1.02142620e+00 8.06410432e-01 -6.60419941e-01 -2.96381176e-01
-7.25354850e-01 6.07126057e-01 6.47642076e-01 -1.27102286e-01
4.04584557e-02 1.58538207e-01 9.61126924e-01 2.93578178e-01
-5.81944227e-01 -4.32539135e-01 -3.87568921e-01 -7.03650832e-01
-4.54873554e-02 -1.28424823e+00 1.46440670e-01 -6.26177549e-01
-1.50685585e+00 2.91847050e-01 1.55176774e-01 -9.22227740e-01
-6.96413100e-01 -9.70202684e-01 -5.88624239e-01 8.01000357e-01
-4.28973198e-01 -1.08912635e+00 -7.10806577e-03 3.02126884e-01
7.98257470e-01 -4.01814163e-01 7.65233696e-01 -4.27326597e-02
-5.40147781e-01 7.34581649e-01 -3.92450094e-01 8.23482573e-02
6.05120063e-01 -1.20672858e+00 6.73197806e-01 3.77973646e-01
-3.21863860e-01 1.07073140e+00 5.72980762e-01 -8.61789763e-01
-7.75401115e-01 -1.93236321e-01 1.28955042e+00 -6.25012219e-01
1.14435804e+00 -4.22328234e-01 -7.45463550e-01 7.49421060e-01
4.62176323e-01 -1.02069247e+00 8.95652115e-01 5.12914598e-01
1.83328375e-01 3.02498788e-01 -1.25391424e+00 1.12551308e+00
1.95218801e+00 -9.01669562e-01 -9.64477122e-01 7.90298104e-01
8.23433936e-01 -3.60176295e-01 -1.10984349e+00 1.74143508e-01
5.81438720e-01 -8.04596901e-01 8.16073895e-01 -3.35035443e-01
1.76324755e-01 6.06927097e-01 2.70003706e-01 -1.47148156e+00
-1.51881929e-02 -7.61052132e-01 2.05326378e-01 1.16486895e+00
7.63266861e-01 -9.83677745e-01 3.99896830e-01 1.13644874e+00
-2.20529377e-01 -8.58046293e-01 -1.13815546e+00 -4.38550442e-01
8.39720905e-01 -8.31234336e-01 3.06337535e-01 9.44239318e-01
9.12787616e-01 2.10526705e-01 1.36993393e-01 -3.06667443e-02
-6.92682639e-02 -9.24376309e-01 5.41749954e-01 -1.30302536e+00
-3.48603666e-01 -1.02551782e+00 -2.49579579e-01 -7.07884967e-01
6.12995148e-01 -1.01381481e+00 -3.34800631e-01 -1.74282360e+00
-8.03018734e-02 -4.45498139e-01 3.86268288e-01 5.03247440e-01
-1.56707406e-01 -4.43869174e-01 1.94632217e-01 -7.83989504e-02
1.20843006e-02 1.26941442e-01 1.09157181e+00 1.81435809e-01
-4.09105331e-01 -1.23205088e-01 -1.24426174e+00 1.17709839e+00
1.05695629e+00 -2.09015191e-01 -8.29942882e-01 -4.10159230e-01
3.35154444e-01 -2.89763808e-02 -1.97950035e-01 -1.15203607e+00
2.67824650e-01 -1.99429601e-01 -6.63680583e-03 2.95771509e-02
5.32891989e-01 -2.15365127e-01 -1.74628928e-01 5.72066128e-01
-3.25719565e-01 3.33362907e-01 9.76482928e-02 -4.21650201e-01
2.84516901e-01 -5.48632979e-01 5.75309753e-01 -3.79387081e-01
-3.57326329e-01 -1.23187285e-02 -7.99336076e-01 6.13651454e-01
8.38297665e-01 -3.35746050e-01 -5.24950266e-01 -8.28609109e-01
-1.00289249e+00 3.39989096e-01 4.26156908e-01 4.88368154e-01
3.48933488e-01 -8.43096972e-01 -6.45048261e-01 3.45796980e-02
2.00255021e-01 -4.44439054e-01 1.41930312e-01 1.08778393e+00
-7.29490995e-01 3.59319031e-01 2.01063558e-01 1.95778278e-03
-1.22759044e+00 5.88569380e-02 2.77425945e-01 -5.87696303e-03
-5.07423162e-01 1.11323905e+00 1.67332917e-01 -6.13755763e-01
2.21271634e-01 -4.30833012e-01 -3.37382436e-01 2.99881995e-01
6.67867780e-01 5.39464235e-01 -3.67667556e-01 -8.56285930e-01
-2.42148608e-01 4.59522188e-01 -4.22935039e-01 -5.61471164e-01
1.01131272e+00 -1.76720008e-01 -3.02163422e-01 7.73856997e-01
8.74622107e-01 5.64350545e-01 -3.95110637e-01 -1.93749189e-01
2.83445895e-01 1.79808840e-01 -4.31030720e-01 -7.67797053e-01
-2.89508730e-01 5.28002739e-01 2.33649790e-01 1.62886769e-01
4.58584338e-01 4.55340147e-01 7.65319943e-01 6.64900422e-01
3.97028923e-01 -1.57856762e+00 -7.72512704e-02 7.21037209e-01
1.30260587e+00 -9.10140395e-01 -1.48958504e-01 -3.76091510e-01
-6.85926855e-01 6.91809475e-01 1.01821899e+00 2.21297964e-01
6.69503033e-01 8.16027820e-02 1.77156344e-01 1.19341491e-02
-7.24489868e-01 -3.15247476e-01 1.02534540e-01 7.62443542e-01
9.36589122e-01 2.90400028e-01 -1.15259111e+00 6.48593605e-01
-1.31848824e+00 -4.83897895e-01 3.83299738e-01 5.97731411e-01
-4.34698343e-01 -1.30117142e+00 -1.45476550e-01 8.02972078e-01
-5.15513837e-01 -5.18076718e-01 -5.52356422e-01 9.65508938e-01
8.06023031e-02 1.19825900e+00 2.30150059e-01 -2.66055226e-01
3.58872026e-01 6.19042337e-01 1.41874194e-01 -1.09280634e+00
-7.87261605e-01 -3.24472666e-01 8.35462153e-01 -2.58976102e-01
-1.41926676e-01 -1.23370957e+00 -1.26151121e+00 -4.57256377e-01
1.41297966e-01 -3.09087723e-01 3.36344838e-01 1.27377558e+00
-5.81285683e-04 6.86097071e-02 -2.51802862e-01 -4.16301906e-01
-4.72432524e-01 -1.74057376e+00 -3.17809224e-01 1.83615550e-01
1.57074377e-01 -5.05089760e-01 -7.12111950e-01 -5.67531943e-01] | [10.810999870300293, 9.657529830932617] |
a5e31e92-90cc-4294-b1ec-3150dc1ea842 | audio-cover-song-identification-using | 1712.00166 | null | https://arxiv.org/abs/1712.00166v2 | https://arxiv.org/pdf/1712.00166v2.pdf | Audio Cover Song Identification using Convolutional Neural Network | In this paper, we propose a new approach to cover song identification using a CNN (convolutional neural network). Most previous studies extract the feature vectors that characterize the cover song relation from a pair of songs and used it to compute the (dis)similarity between the two songs. Based on the observation that there is a meaningful pattern between cover songs and that this can be learned, we have reformulated the cover song identification problem in a machine learning framework. To do this, we first build the CNN using as an input a cross-similarity matrix generated from a pair of songs. We then construct the data set composed of cover song pairs and non-cover song pairs, which are used as positive and negative training samples, respectively. The trained CNN outputs the probability of being in the cover song relation given a cross-similarity matrix generated from any two pieces of music and identifies the cover song by ranking on the probability. Experimental results show that the proposed algorithm achieves performance better than or comparable to the state-of-the-art. | ['Sang Keun Choe', 'Sungkyun Chang', 'Kyogu Lee', 'Juheon Lee'] | 2017-12-01 | null | null | null | null | ['cover-song-identification'] | ['music'] | [ 5.60422003e-01 -5.28347909e-01 -3.08014117e-02 -2.63343066e-01
-6.16706133e-01 -8.46554160e-01 3.21913093e-01 1.43256122e-02
-1.53965473e-01 4.00228560e-01 9.52548981e-02 2.70775408e-01
-4.66841340e-01 -1.25534809e+00 -7.99433768e-01 -5.52965701e-01
-3.99260014e-01 4.26419765e-01 -1.14020087e-01 -1.03251934e-02
3.18405122e-01 1.52111322e-01 -2.11391544e+00 4.85890031e-01
4.07303363e-01 1.38233697e+00 2.34950464e-02 6.73457503e-01
7.68500343e-02 2.82516092e-01 -9.10675824e-01 -2.38374248e-02
6.17817879e-01 -8.10074329e-01 -8.57283950e-01 2.88754776e-02
5.96669137e-01 -5.40114455e-02 -3.11914414e-01 1.13481224e+00
3.36447179e-01 -2.78187240e-03 5.78067899e-01 -1.20806134e+00
-4.64246064e-01 9.77675915e-01 -3.71524900e-01 2.57525325e-01
4.87571359e-01 -2.71492392e-01 1.58361554e+00 -6.88113391e-01
2.95458972e-01 8.36567640e-01 7.92446017e-01 3.04075062e-01
-9.82374012e-01 -1.09007692e+00 -4.06176537e-01 2.26527721e-01
-1.86074412e+00 7.31456801e-02 8.50552678e-01 -4.22339052e-01
6.09690309e-01 7.72268891e-01 1.28813219e+00 4.88493800e-01
-1.71127319e-01 8.62852216e-01 7.95410514e-01 -5.66618204e-01
6.05482459e-02 8.47914442e-02 2.15315670e-02 3.23636562e-01
-2.59291660e-02 1.22861020e-01 -7.08039224e-01 -2.32347026e-01
4.76774812e-01 8.17092136e-02 -3.50928366e-01 -3.54359448e-02
-1.26941383e+00 8.21600020e-01 6.46589220e-01 7.29581416e-01
-1.85051203e-01 1.63322270e-01 1.53225690e-01 4.96931314e-01
2.98988938e-01 6.87425196e-01 -3.22820581e-02 2.21686482e-01
-1.32340586e+00 4.36286896e-01 1.00184143e+00 5.28765500e-01
9.27586079e-01 -1.65750518e-01 -8.83456916e-02 6.76868916e-01
2.17697561e-01 2.65818149e-01 4.72364485e-01 -2.40879476e-01
2.63058156e-01 6.81814194e-01 -3.40992868e-01 -1.31201041e+00
1.94059163e-02 -1.08694994e+00 -7.94736385e-01 -1.14724323e-01
1.14291452e-01 3.00948530e-01 -5.03153384e-01 1.68503606e+00
-8.46285671e-02 6.34050846e-01 1.72715575e-01 9.23305690e-01
1.01869273e+00 6.33372664e-01 -7.10629046e-01 -5.33715077e-02
1.09182322e+00 -7.44426012e-01 -4.90040272e-01 1.11648798e-01
1.12119772e-01 -9.92185473e-01 8.74972165e-01 5.08473814e-01
-4.82760608e-01 -7.90970504e-01 -1.60642970e+00 6.76319480e-01
-4.41910297e-01 3.48479927e-01 4.43138510e-01 5.16802132e-01
-8.29340756e-01 1.07225454e+00 -1.54526263e-01 -2.50604302e-01
3.65885019e-01 6.61550939e-01 -2.10807696e-01 2.06429988e-01
-1.42320621e+00 1.67434514e-01 7.83496141e-01 -4.45227660e-02
-1.26067090e+00 -4.65515614e-01 -5.38981199e-01 1.60252154e-01
1.40654773e-01 -3.37360442e-01 9.46068585e-01 -1.43430471e+00
-8.08696151e-01 9.53639269e-01 4.65113699e-01 -5.32083213e-01
1.27157941e-01 -1.24239765e-01 -6.46275163e-01 -1.75336614e-01
2.72724748e-01 4.02748853e-01 1.01559591e+00 -1.35016298e+00
-9.66264069e-01 -1.08274929e-01 1.62937075e-01 2.97301948e-01
-6.12091303e-01 -4.58947010e-02 -4.29392040e-01 -7.28668332e-01
2.33056158e-01 -1.02825427e+00 2.88976848e-01 -4.03318226e-01
-9.06898320e-01 -2.03026813e-02 9.10993755e-01 -2.62111604e-01
1.44596398e+00 -2.22927165e+00 7.56748486e-03 6.32591784e-01
8.22370052e-02 2.19765693e-01 -3.78994793e-01 6.25812113e-01
-4.09287363e-01 1.85335532e-01 -5.43874443e-01 1.73111111e-01
-2.54510134e-01 -5.40947020e-02 -5.41205823e-01 4.02327448e-01
6.17242008e-02 5.07389247e-01 -9.15314734e-01 -1.70508862e-01
-1.38085186e-01 4.83391970e-01 -2.14708984e-01 5.22774637e-01
-1.94201052e-01 2.46988341e-01 -1.22349009e-01 5.49922287e-01
6.32527888e-01 1.18165471e-01 2.65833467e-01 -1.38285264e-01
-8.46009236e-03 3.85577530e-01 -1.54741728e+00 1.45646286e+00
-7.78892040e-02 7.61771202e-01 -4.47955310e-01 -8.56319904e-01
1.20262527e+00 1.87538236e-01 4.37659472e-01 -1.34576470e-01
1.81063831e-01 4.18180555e-01 2.44692743e-01 -3.68384093e-01
3.47269505e-01 -2.52428621e-01 -7.19863325e-02 8.33155334e-01
3.19659650e-01 -8.42509493e-02 2.60299891e-01 -1.67550057e-01
8.25422406e-01 -1.81603640e-01 2.39081651e-01 -1.42052516e-01
7.69717455e-01 -1.19568989e-01 3.97891551e-01 7.62572348e-01
4.11193997e-01 8.00832689e-01 2.85538524e-01 -6.18731260e-01
-7.59715199e-01 -9.37057734e-01 -8.68811756e-02 9.06765401e-01
1.06706053e-01 -6.55920506e-01 -8.04481387e-01 -6.25786901e-01
6.82258084e-02 1.00535028e-01 -8.57748747e-01 -1.74784541e-01
-3.47731799e-01 -6.35486305e-01 1.00952399e+00 4.14076388e-01
6.63477600e-01 -1.23799431e+00 -3.58272165e-01 6.26411289e-02
-1.70450091e-01 -4.56333101e-01 -5.77577412e-01 3.45913708e-01
-5.63104272e-01 -1.19784284e+00 -4.94393468e-01 -1.25282073e+00
2.47835994e-01 3.69571239e-01 1.21500719e+00 3.49401563e-01
-4.23139006e-01 -6.81821723e-03 -4.73994493e-01 -5.70993245e-01
-2.37914711e-01 3.09999377e-01 3.30275983e-01 4.99629378e-01
3.89226407e-01 -8.56486976e-01 -4.59660381e-01 1.62449032e-01
-1.12566268e+00 -1.34850413e-01 4.37660664e-01 8.63872409e-01
8.48867238e-01 6.30759299e-01 5.10539055e-01 -6.39107525e-01
7.29765415e-01 -4.15214866e-01 -2.56649047e-01 1.21274173e-01
-3.17549139e-01 -1.96270019e-01 5.35914242e-01 -5.20413637e-01
-5.70910843e-03 3.38670313e-01 6.11299649e-02 -5.25168240e-01
-1.31320953e-01 7.37306058e-01 -3.04112554e-01 1.29064284e-02
4.62549269e-01 5.77069283e-01 -2.94465184e-01 -6.10892832e-01
1.45217419e-01 1.08064890e+00 6.09715044e-01 -1.91083923e-01
1.20064068e+00 3.78656507e-01 -2.27362573e-01 -6.02141201e-01
-1.12449706e+00 -6.07738495e-01 -7.65248239e-01 -3.55606288e-01
6.22117758e-01 -8.33592594e-01 -5.59040427e-01 3.20704103e-01
-8.05391550e-01 3.26482922e-01 -5.34233689e-01 5.60449660e-01
-4.51686978e-01 9.18236747e-02 -1.30995587e-01 -8.68633151e-01
-5.44010699e-01 -6.75659180e-01 8.91680539e-01 2.62569427e-01
-3.37081701e-01 -6.92290843e-01 5.56806445e-01 -9.87505615e-02
6.45487309e-02 3.90598655e-01 8.68436575e-01 -1.18241644e+00
-6.27831638e-01 -5.16101182e-01 2.41955332e-02 5.18215597e-01
3.41149271e-01 -1.35140747e-01 -1.40338314e+00 -3.56232941e-01
9.74147096e-02 -2.28319362e-01 1.11260533e+00 1.63593411e-01
1.23922169e+00 -4.52809781e-01 -1.11616388e-01 7.83521831e-01
1.57311809e+00 2.44503900e-01 4.81898397e-01 3.57822716e-01
6.29298806e-01 5.15657842e-01 6.93097711e-01 3.19241315e-01
-2.01179355e-01 6.71712995e-01 6.09152436e-01 1.01287305e-01
7.29935914e-02 -5.21827579e-01 2.08087131e-01 1.09827495e+00
5.41768447e-02 -2.16069952e-01 -7.08756149e-01 7.50814199e-01
-1.72723198e+00 -1.10982418e+00 5.02168834e-02 2.47543788e+00
9.10736024e-01 1.16765767e-01 2.32952058e-01 9.53208625e-01
8.89262259e-01 2.44646341e-01 -2.57626384e-01 -3.62467840e-02
-2.54226148e-01 4.55424011e-01 1.81383759e-01 1.68096989e-01
-1.51869500e+00 6.62942052e-01 6.46218157e+00 1.15652382e+00
-1.34338856e+00 -2.52542436e-01 2.35746875e-01 -2.07490206e-01
-2.82867253e-01 -1.12854257e-01 -4.48080063e-01 1.70830294e-01
5.34454644e-01 -1.93649799e-01 7.44985700e-01 5.44285417e-01
-5.14458120e-01 6.61144555e-02 -1.26471436e+00 1.14443231e+00
4.65318859e-01 -1.29154587e+00 2.40275666e-01 7.54641965e-02
7.04076409e-01 -1.55726433e-01 2.90096074e-01 2.87722144e-02
-3.12477916e-01 -1.33272278e+00 8.04883718e-01 4.37273800e-01
9.46957469e-01 -1.00012219e+00 1.00324106e+00 3.17977220e-01
-1.66773272e+00 7.12288469e-02 -3.85251820e-01 -3.14230114e-01
-5.36629140e-01 4.96429503e-01 -1.07293081e+00 7.98927009e-01
8.34779441e-01 1.14863491e+00 -5.05859375e-01 1.22152662e+00
-3.35540138e-02 8.16002131e-01 -1.98057100e-01 -1.03019387e-01
5.88482097e-02 -5.93206473e-02 7.60443151e-01 1.09652662e+00
4.26099688e-01 -5.46832383e-01 1.84457242e-01 1.05251825e+00
-3.51105988e-01 2.63401031e-01 -8.30110788e-01 -4.41130370e-01
3.43479455e-01 1.35087490e+00 -8.57464135e-01 -3.07745159e-01
1.02014495e-02 8.73938620e-01 -1.17713049e-01 -1.17386170e-01
-5.67881823e-01 -7.08484590e-01 7.99783885e-01 -4.75060530e-02
4.22272027e-01 3.50193828e-01 -1.23957224e-01 -9.58917677e-01
-8.26953258e-03 -9.42518175e-01 3.36793482e-01 -5.00622809e-01
-1.41378593e+00 9.69607532e-01 -2.71666944e-01 -2.03654790e+00
-1.92013204e-01 -3.25283617e-01 -7.19927251e-01 8.61897409e-01
-1.18921661e+00 -1.10620522e+00 -2.30604783e-01 5.95365942e-01
3.20238769e-01 -6.71884418e-01 1.19298422e+00 3.16276073e-01
1.50980828e-02 5.75336039e-01 1.80982485e-01 6.60121620e-01
3.63196135e-01 -1.19040155e+00 3.06911379e-01 4.63153720e-01
1.07781601e+00 5.25236428e-01 5.31463265e-01 -5.63570142e-01
-1.15014935e+00 -1.17817736e+00 7.68691719e-01 -2.22168952e-01
3.86690974e-01 -4.30099338e-01 -6.47868514e-01 2.70982742e-01
2.20189601e-01 -2.12801054e-01 1.35894024e+00 1.05338238e-01
-6.31324589e-01 -2.94647247e-01 -9.40316737e-01 8.27403292e-02
9.49174225e-01 -8.95022511e-01 -6.74107969e-01 2.80624747e-01
5.64445913e-01 -1.81026295e-01 -7.60821164e-01 3.71912241e-01
9.27294374e-01 -8.23171735e-01 9.00561750e-01 -6.18746638e-01
4.91630018e-01 -7.34176397e-01 -5.29661417e-01 -1.18982518e+00
-2.57968128e-01 -2.08110631e-01 1.33118883e-01 1.20959187e+00
3.17144036e-01 -4.94946614e-02 6.21586680e-01 -6.52309299e-01
1.99638784e-01 -7.87343442e-01 -8.52571726e-01 -1.04189336e+00
-3.67173910e-01 -6.53112173e-01 9.80248332e-01 1.02713692e+00
-7.67528638e-02 4.37365562e-01 -6.12908840e-01 2.22508222e-01
4.16768789e-01 8.99221539e-01 6.79180205e-01 -1.58695424e+00
-5.44490099e-01 -2.85852164e-01 -7.87071347e-01 -5.48262596e-01
2.97582820e-02 -1.31321979e+00 1.92736626e-01 -1.28600383e+00
5.06808341e-01 -2.94590771e-01 -8.16510677e-01 4.78666246e-01
1.43804237e-01 1.11584210e+00 3.25209916e-01 4.55313653e-01
-6.10465467e-01 4.12658989e-01 9.20511782e-01 -5.95566869e-01
-2.54382372e-01 2.91577637e-01 -7.30817199e-01 6.66261911e-01
8.51531625e-01 -8.33197415e-01 -1.02536857e-01 -5.88576421e-02
3.09711874e-01 -2.36697018e-01 2.42151335e-01 -1.58687401e+00
-1.40780047e-01 2.58855164e-01 4.39757109e-01 -8.13647032e-01
2.22425789e-01 -7.64539659e-01 3.91554117e-01 7.02556908e-01
-6.20229185e-01 -3.28380466e-01 2.82567050e-02 5.75028598e-01
-6.59561515e-01 -3.85940999e-01 5.52121043e-01 3.82297598e-02
-1.74070969e-01 2.69654959e-01 -2.12775275e-01 -2.01671198e-01
8.06715548e-01 -3.18839073e-01 3.89393240e-01 -6.39175713e-01
-7.68930137e-01 -3.21720243e-01 1.05618298e-01 6.69593513e-01
7.92844415e-01 -1.76274872e+00 -7.74751544e-01 5.37080526e-01
4.06219989e-01 -3.98988068e-01 3.07386015e-02 3.77762914e-01
-4.35237229e-01 3.49284232e-01 -3.94452006e-01 -6.65809155e-01
-1.70920765e+00 3.76466811e-01 3.72344226e-01 -1.29823416e-01
-2.29880780e-01 9.73541141e-01 5.04325368e-02 -4.39305961e-01
4.17213440e-01 -1.10438019e-01 -6.22144997e-01 2.69018650e-01
6.89903080e-01 -1.05734080e-01 -1.82747170e-02 -9.30750549e-01
-4.61537659e-01 8.81556511e-01 1.70433268e-01 -2.13641852e-01
1.44382405e+00 4.13969874e-01 -4.35306162e-01 7.95453131e-01
1.37085688e+00 2.20711187e-01 -5.34181476e-01 -3.19592059e-01
-1.62073269e-01 -7.98562527e-01 -3.28098357e-01 -6.82344973e-01
-1.31627035e+00 8.04896832e-01 1.02441347e+00 7.27307260e-01
1.28943300e+00 8.96826982e-02 7.42793918e-01 2.93195456e-01
2.09116653e-01 -8.12216580e-01 2.15755850e-02 5.24053633e-01
9.61994886e-01 -9.30069923e-01 -1.04451582e-01 -2.47643560e-01
-2.02617049e-01 1.17670131e+00 3.82365435e-01 -6.08069122e-01
8.56955230e-01 1.17116138e-01 1.16364338e-01 -1.72580555e-01
-4.15853083e-01 -6.25857115e-01 8.85130703e-01 6.00603104e-01
5.05502641e-01 2.26724342e-01 -2.77279288e-01 7.81872809e-01
-6.94544375e-01 -2.03673109e-01 1.96144357e-01 5.39483190e-01
-4.64352071e-01 -1.17810655e+00 -5.06250620e-01 7.13946462e-01
-4.55823451e-01 -1.91848829e-01 -1.03181338e+00 4.64897037e-01
7.35501885e-01 8.97798359e-01 2.40182638e-01 -1.43046594e+00
1.16952270e-01 -1.00891456e-01 2.22932711e-01 -8.82534325e-01
-1.13998997e+00 1.77011967e-01 -1.73649833e-01 -8.23787525e-02
-5.83311915e-01 -2.96690255e-01 -8.83757651e-01 -6.44627213e-03
-6.97303891e-01 3.95326376e-01 3.81180763e-01 8.51510406e-01
1.82479043e-02 5.77410758e-01 1.02715516e+00 -6.36987627e-01
-2.08502144e-01 -1.06559658e+00 -9.93006408e-01 4.49950218e-01
3.21306914e-01 -3.53053808e-01 -2.50729233e-01 -7.43143559e-02] | [15.777467727661133, 5.180514812469482] |
546b30d0-935e-4d11-8066-fd0d59c1e2e0 | bwbaugh-hierarchical-sentiment-analysis-with | null | null | https://aclanthology.org/S13-2090 | https://aclanthology.org/S13-2090.pdf | bwbaugh : Hierarchical sentiment analysis with partial self-training | null | ['Wesley Baugh'] | 2013-06-01 | null | null | null | semeval-2013-6 | ['subjectivity-analysis'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.3230366706848145, 3.8184146881103516] |
9084bb86-b7a8-4d5b-b176-aaae36fd843d | homography-from-two-orientation-and-scale | 1906.11927 | null | https://arxiv.org/abs/1906.11927v1 | https://arxiv.org/pdf/1906.11927v1.pdf | Homography from two orientation- and scale-covariant features | This paper proposes a geometric interpretation of the angles and scales which the orientation- and scale-covariant feature detectors, e.g. SIFT, provide. Two new general constraints are derived on the scales and rotations which can be used in any geometric model estimation tasks. Using these formulas, two new constraints on homography estimation are introduced. Exploiting the derived equations, a solver for estimating the homography from the minimal number of two correspondences is proposed. Also, it is shown how the normalization of the point correspondences affects the rotation and scale parameters, thus achieving numerically stable results. Due to requiring merely two feature pairs, robust estimators, e.g. RANSAC, do significantly fewer iterations than by using the four-point algorithm. When using covariant features, e.g. SIFT, the information about the scale and orientation is given at no cost. The proposed homography estimation method is tested in a synthetic environment and on publicly available real-world datasets. | ['Zuzana Kukelova', 'Daniel Barath'] | 2019-06-27 | homography-from-two-orientation-and-scale-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Barath_Homography_From_Two_Orientation-_and_Scale-Covariant_Features_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Barath_Homography_From_Two_Orientation-_and_Scale-Covariant_Features_ICCV_2019_paper.pdf | iccv-2019-10 | ['homography-estimation'] | ['computer-vision'] | [-5.66508994e-02 -2.06110895e-01 -1.94342464e-01 -2.21558452e-01
-2.13307127e-01 -8.41412783e-01 6.34109497e-01 9.37300995e-02
-4.87450153e-01 5.65782011e-01 -3.33765835e-01 1.29775077e-01
-1.92272291e-01 -7.36153483e-01 -4.67667490e-01 -6.23458326e-01
3.99908982e-02 4.25591767e-01 2.63892829e-01 -2.20799953e-01
8.11416924e-01 1.01827776e+00 -1.44109082e+00 -6.78723156e-01
6.12773180e-01 9.97335792e-01 -5.21375379e-03 7.92059839e-01
3.27315211e-01 -8.34847465e-02 -3.50953817e-01 -4.03255045e-01
8.19552779e-01 -2.52623737e-01 -3.76915783e-01 4.04095620e-01
7.49718606e-01 -7.44147718e-01 -2.73810387e-01 1.25648534e+00
5.42182140e-02 1.46008000e-01 5.73651552e-01 -1.11819959e+00
-1.43077210e-01 -8.33180696e-02 -5.61370254e-01 -3.39409143e-01
5.94525218e-01 -1.73572198e-01 1.12201953e+00 -9.31050539e-01
8.67532134e-01 1.10221565e+00 6.60320759e-01 -1.95846781e-01
-1.06449282e+00 -1.59099966e-01 -4.43774253e-01 3.14938426e-02
-1.54849708e+00 -2.45041177e-01 8.87200296e-01 -3.69828135e-01
6.99048102e-01 4.53489453e-01 5.81169486e-01 3.40427667e-01
4.74272639e-01 6.56925738e-02 8.02469969e-01 -7.11704612e-01
6.55585155e-02 2.17528239e-01 -2.07890533e-02 7.84966290e-01
9.62240875e-01 1.32228613e-01 -1.86245725e-01 -3.00516754e-01
1.41242611e+00 5.14665060e-02 -2.12807786e-02 -1.16636491e+00
-1.37363005e+00 8.80955935e-01 3.89684349e-01 2.90203035e-01
-1.63270414e-01 1.78100020e-01 1.92114249e-01 1.19129963e-01
1.40463382e-01 6.08206928e-01 -1.61831230e-01 -7.79676065e-02
-6.10050440e-01 -2.27575749e-02 7.43179083e-01 1.46899819e+00
1.12964475e+00 -8.22657272e-02 7.79488504e-01 3.36161256e-01
3.73928696e-01 8.32241058e-01 1.30725697e-01 -9.64462519e-01
3.84658605e-01 6.07944489e-01 5.63849807e-01 -1.60604429e+00
-5.91962814e-01 -2.53643751e-01 -7.11994290e-01 1.23155043e-01
5.68132222e-01 4.29864004e-02 -6.33297324e-01 1.27872097e+00
4.29458290e-01 -2.73650050e-01 -1.31383792e-01 9.68757689e-01
5.26096784e-02 1.13056824e-01 -6.90616608e-01 -1.15182564e-01
1.38735497e+00 -4.70050424e-01 -6.91033661e-01 -6.26998097e-02
6.27677262e-01 -1.48749256e+00 4.32311296e-01 2.97976226e-01
-9.91847336e-01 -5.94335198e-01 -1.42402375e+00 -2.62408316e-01
-4.25428659e-01 5.95014215e-01 6.53816044e-01 6.33991838e-01
-9.18026745e-01 9.03298438e-01 -9.52588081e-01 -7.38502681e-01
-5.52549422e-01 6.23116672e-01 -5.24235785e-01 5.26836216e-01
-8.03690970e-01 1.22904754e+00 2.89344698e-01 3.19026738e-01
1.13673218e-01 -1.97933540e-01 -8.67908895e-01 9.77582112e-02
6.45614117e-02 -5.68419576e-01 9.45244431e-01 -5.94126821e-01
-1.72016430e+00 8.23930681e-01 -1.66631326e-01 -2.29523510e-01
8.36684942e-01 -1.60311788e-01 -3.01429611e-02 5.56862295e-01
1.47050202e-01 2.65326262e-01 9.18168724e-01 -1.07505786e+00
-5.31078935e-01 -4.34644759e-01 1.24214225e-01 1.81410670e-01
-5.09574190e-02 -3.21087241e-01 -4.06851202e-01 -3.78495336e-01
9.44139600e-01 -1.20228517e+00 -3.61834943e-01 1.96313024e-01
-3.88720036e-01 3.02665740e-01 6.37245059e-01 -6.78281367e-01
5.82237184e-01 -2.12130237e+00 3.19290876e-01 7.82332897e-01
7.28328452e-02 -1.84464693e-01 1.95730239e-01 6.08713329e-01
1.67688087e-01 -2.01830178e-01 2.28331417e-01 -7.83857480e-02
-1.14482120e-01 1.24223717e-01 -1.82181150e-01 1.13660622e+00
-1.42122835e-01 2.25501299e-01 -7.10093141e-01 -2.58736283e-01
7.08709717e-01 5.50343394e-01 -3.92289639e-01 -8.32124799e-02
4.94045407e-01 3.50629210e-01 -5.54695785e-01 3.41999978e-01
1.01306164e+00 4.13594209e-02 4.05091375e-01 -6.47853673e-01
-5.36286414e-01 1.70984194e-01 -1.68133771e+00 1.42240131e+00
-5.96901238e-01 5.72866917e-01 6.81749880e-02 -5.73392630e-01
1.25112736e+00 1.29096508e-01 5.45568347e-01 -4.22297925e-01
5.59501112e-01 5.74155688e-01 -8.79213214e-02 9.74273309e-02
7.73254156e-01 7.07113668e-02 -3.11699118e-02 -2.07299981e-02
5.68469316e-02 -5.54352403e-01 4.68407422e-01 3.96970622e-02
5.68252146e-01 1.80379063e-01 9.17653024e-01 -5.09692073e-01
8.02826345e-01 -4.53521572e-02 5.22299170e-01 3.20898324e-01
7.54854679e-02 3.19640577e-01 3.09792906e-01 -5.28872848e-01
-1.49442315e+00 -9.08346295e-01 -3.13178331e-01 1.31948218e-01
6.49986923e-01 -5.06820023e-01 -4.95313078e-01 -5.22933938e-02
3.21560651e-01 8.95406082e-02 -4.00875896e-01 -3.06970142e-02
-6.76222563e-01 -1.50294214e-01 1.17304102e-02 4.30941671e-01
5.83802521e-01 -1.35622248e-01 -1.05420732e+00 1.05822518e-01
1.21641867e-01 -1.29988182e+00 -3.87431711e-01 -1.40447468e-01
-1.22955167e+00 -1.31115580e+00 -3.53001773e-01 -5.48709154e-01
1.15397441e+00 6.56810939e-01 5.85698843e-01 1.26458332e-01
-3.17930132e-01 5.34870803e-01 -2.12665856e-01 6.45614490e-02
-1.10338442e-01 -1.67835101e-01 3.70113999e-01 3.18679847e-02
2.14137256e-01 -5.50315261e-01 -4.89801466e-01 8.01160991e-01
-4.47779804e-01 -6.35142177e-02 6.12718701e-01 7.50753939e-01
6.54909253e-01 -1.62120342e-01 -8.38029850e-03 -5.04960060e-01
3.80531140e-02 3.65568161e-01 -1.34607303e+00 1.17260389e-01
-5.42950213e-01 2.63600260e-01 7.52236247e-01 -3.38413507e-01
-9.37710583e-01 5.48747540e-01 3.98247689e-01 -3.37097645e-01
-1.84414082e-03 7.17134401e-02 -7.08794445e-02 -8.56828153e-01
4.36017811e-01 -4.17787917e-02 4.20067273e-02 -4.74344611e-01
5.55921197e-01 4.16727006e-01 5.09811103e-01 -4.37209725e-01
1.37157261e+00 8.17415059e-01 7.74226427e-01 -1.25422966e+00
-7.75501207e-02 -1.01250648e+00 -1.35682046e+00 4.37643901e-02
6.26419246e-01 -7.04385996e-01 -8.75600755e-01 2.95545965e-01
-1.27164817e+00 6.58433676e-01 -1.29911020e-01 1.00742328e+00
-9.52852070e-01 7.32963681e-01 -3.65567565e-01 -7.38845170e-01
-8.12070072e-02 -1.26048827e+00 1.02780485e+00 1.35356620e-01
-2.56560534e-01 -1.02594936e+00 4.42449600e-02 -1.85515478e-01
2.03184977e-01 3.32417548e-01 5.49416363e-01 -1.95503071e-01
-9.27045226e-01 -5.15241921e-01 -2.03403234e-01 -3.55545878e-02
3.93161505e-01 3.06101322e-01 -5.28236449e-01 -5.43637812e-01
2.13939607e-01 3.39213938e-01 2.33307078e-01 1.55232698e-01
3.38814348e-01 -2.50287682e-01 -3.20739597e-01 1.00792277e+00
1.65687859e+00 3.43383670e-01 3.66207629e-01 5.18487155e-01
5.88412464e-01 4.44174767e-01 9.50850546e-01 5.03866732e-01
-1.42612187e-02 9.38008308e-01 4.50813502e-01 4.89778258e-02
1.98649138e-01 -3.47067356e-01 -5.90458326e-02 1.03573143e+00
-5.02426744e-01 5.30525208e-01 -4.64396685e-01 3.40810657e-01
-1.71205223e+00 -5.95973492e-01 -3.98543924e-01 2.63322163e+00
3.27180952e-01 -9.19797421e-02 -9.10325795e-02 1.70546740e-01
6.42728567e-01 7.36571616e-03 -4.32731897e-01 -3.75535965e-01
4.50854786e-02 2.75867004e-02 1.25453913e+00 8.81847382e-01
-1.10604274e+00 7.63248801e-01 6.17293406e+00 1.79828271e-01
-1.09131455e+00 -4.66972560e-01 -3.38227004e-01 4.70830858e-01
2.89962124e-02 6.45361066e-01 -9.42024589e-01 -9.77826491e-03
3.49849075e-01 -5.16718149e-01 3.87404323e-01 1.12059271e+00
1.28727674e-01 -2.51730591e-01 -1.01137245e+00 1.28195548e+00
1.53179795e-01 -9.43873227e-01 4.60826419e-02 2.60644823e-01
5.64984381e-01 -4.23369646e-01 -9.71824080e-02 -3.86963338e-01
-2.16916353e-01 -1.16668522e-01 5.71401894e-01 2.48923063e-01
5.77739954e-01 -7.05229104e-01 7.44360268e-01 8.07836279e-02
-1.43231142e+00 2.55198359e-01 -7.24281430e-01 -7.20498338e-02
1.52500078e-01 4.99904811e-01 -9.64791954e-01 8.87820244e-01
6.90762773e-02 7.44878948e-01 -6.00827992e-01 1.24311292e+00
-3.45143259e-01 -2.73740947e-01 -7.00015724e-01 -2.90126726e-02
-7.76684284e-03 -8.20016086e-01 6.82721496e-01 9.45871890e-01
7.10217357e-01 -8.27203095e-02 -5.11425175e-02 5.41990757e-01
2.50837356e-01 3.04441035e-01 -5.87794185e-01 1.76583469e-01
3.11700761e-01 1.45830274e+00 -9.59418178e-01 -1.03930332e-01
-3.52169901e-01 9.99117374e-01 -2.59059042e-01 2.95149088e-01
-5.89959264e-01 -7.42656589e-01 6.67369843e-01 1.51061267e-01
2.89214283e-01 -8.95932198e-01 -3.21429104e-01 -1.45126450e+00
1.64776832e-01 -5.53892136e-01 -5.70929199e-02 -4.21086371e-01
-7.31980979e-01 2.79962659e-01 2.72158086e-01 -1.66965759e+00
-5.64511120e-01 -9.68518317e-01 -3.66811693e-01 7.60782719e-01
-1.24297094e+00 -8.89882326e-01 -2.85638511e-01 4.16025609e-01
1.87271640e-01 1.46175340e-01 7.12543964e-01 9.21553969e-02
1.23885628e-02 4.11126822e-01 2.69942909e-01 3.08350120e-02
7.69603848e-01 -1.10370469e+00 3.83298457e-01 9.12291706e-01
9.92634222e-02 1.11611140e+00 1.04614758e+00 -6.13298297e-01
-1.77526772e+00 -3.52417201e-01 8.25020432e-01 -1.46577209e-01
8.82196546e-01 -4.45124120e-01 -4.93312269e-01 9.31455433e-01
-1.72088444e-01 -9.53081623e-02 1.06348708e-01 -8.87342542e-02
-3.02459270e-01 -1.42289087e-01 -1.26304734e+00 3.33519876e-01
8.42537463e-01 -5.18614352e-01 -3.35016221e-01 3.78380179e-01
2.45368853e-01 -6.27131462e-01 -9.40437675e-01 1.44183129e-01
7.73826063e-01 -1.10115206e+00 1.04306567e+00 -1.44822329e-01
-3.88665169e-01 -5.14902949e-01 -1.96397096e-01 -1.21293390e+00
-3.24212283e-01 -7.61688948e-01 2.92020261e-01 8.88472497e-01
-9.67004076e-02 -1.12783003e+00 5.85380912e-01 3.38877261e-01
2.95083493e-01 -1.29400104e-01 -1.03368759e+00 -1.08798718e+00
-3.99332404e-01 3.27539504e-01 3.69377732e-01 8.53638172e-01
7.65886307e-02 1.34070992e-01 -3.80095869e-01 5.94253361e-01
1.00724316e+00 2.95461804e-01 1.08732104e+00 -1.44880354e+00
-1.91900238e-01 -1.24344610e-01 -1.21509743e+00 -1.24147892e+00
-1.74673274e-01 -3.45572174e-01 -3.33987057e-01 -8.73633325e-01
-2.65651852e-01 -2.54969627e-01 2.72324800e-01 -2.38786079e-03
4.42602426e-01 6.21587932e-02 2.91962862e-01 3.21522057e-01
8.91763493e-02 3.17995906e-01 1.07585275e+00 2.40759745e-01
-2.07421303e-01 7.43683428e-02 4.79502641e-02 1.14827883e+00
8.24369490e-01 -1.14214465e-01 -2.02092275e-01 -2.46991083e-01
2.22238526e-01 1.35733873e-01 2.06878006e-01 -1.03957009e+00
3.28549623e-01 -1.71152860e-01 4.35946226e-01 -6.79871559e-01
5.27024984e-01 -1.07621312e+00 2.97407746e-01 5.13247490e-01
1.62448570e-01 4.48492795e-01 -5.47717661e-02 2.60380715e-01
-1.70164824e-01 -4.55459744e-01 7.80533791e-01 2.91884621e-03
-4.89556491e-01 -4.11784910e-02 9.07038748e-02 -6.06379390e-01
1.06348729e+00 -4.34571326e-01 -1.73339590e-01 -5.11308670e-01
-6.85267687e-01 -3.51525903e-01 9.15283203e-01 2.24873766e-01
5.08024395e-01 -1.48334193e+00 -2.06760585e-01 5.76186895e-01
-3.34276259e-02 -2.86006719e-01 -4.42006588e-02 1.04858768e+00
-1.00266683e+00 7.83364654e-01 -5.23623824e-01 -6.73932910e-01
-1.44287336e+00 4.45779294e-01 1.94438070e-01 -5.64868040e-02
-3.50720078e-01 1.30524514e-02 -6.32479936e-02 -3.40967178e-01
-8.48115832e-02 -6.66414738e-01 1.79445669e-01 -1.13831744e-01
3.13849241e-01 6.75514638e-01 5.72182424e-02 -1.16045165e+00
-3.85893762e-01 1.40052259e+00 5.70080057e-02 -1.71582043e-01
9.96080816e-01 -3.54355007e-01 -1.45750985e-01 1.83426082e-01
1.34161901e+00 3.87042969e-01 -9.46102738e-01 4.72163968e-02
-8.59130174e-02 -9.93093550e-01 -2.16958135e-01 4.35784608e-02
-7.48304904e-01 8.28629911e-01 5.17361224e-01 2.62888312e-01
8.44621837e-01 -5.66240609e-01 3.58128101e-01 5.62614441e-01
9.16018903e-01 -8.95825624e-01 -4.00005251e-01 4.57376927e-01
8.59089017e-01 -9.48227465e-01 5.57490051e-01 -9.15892780e-01
-3.85843962e-02 1.61952841e+00 2.60162562e-01 -5.96862316e-01
3.47314566e-01 -3.91607396e-02 -2.61269137e-02 2.33305439e-01
4.65855971e-02 -5.85850216e-02 2.25101471e-01 4.41557288e-01
3.85187596e-01 5.21592386e-02 -8.50697160e-01 -4.50674206e-01
-4.64546800e-01 -3.02851081e-01 7.72891521e-01 7.86921799e-01
-5.32586157e-01 -1.37288201e+00 -7.19211757e-01 -6.19557314e-02
-7.94729441e-02 3.59466463e-01 -4.44101423e-01 1.29088056e+00
-2.68647313e-01 6.88396037e-01 9.33937132e-02 -2.47128397e-01
5.24649501e-01 -3.67254972e-01 8.28195274e-01 -1.29468992e-01
-1.62212148e-01 3.15740705e-01 -8.80338028e-02 -7.94450462e-01
-5.47617435e-01 -5.70740521e-01 -1.29388833e+00 -3.41227233e-01
-5.74186802e-01 2.31743902e-01 1.08188045e+00 5.59175372e-01
2.12880403e-01 -1.90692082e-01 7.07515001e-01 -9.78193521e-01
-9.34326828e-01 -6.35300159e-01 -8.38916004e-01 3.98005575e-01
3.46991062e-01 -9.57219005e-01 -8.09582710e-01 -1.41636193e-01] | [7.912501811981201, -2.31968092918396] |
698b1093-bc4e-4e86-8f5e-3d83e6707e47 | sutav-a-turkish-audio-visual-database | null | null | https://aclanthology.org/L12-1262 | https://aclanthology.org/L12-1262.pdf | SUTAV: A Turkish Audio-Visual Database | This paper contains information about the ''''''``Sabanci University Turkish Audio-Visual (SUTAV)'''''''' database. The main aim of collecting SUTAV database was to obtain a large audio-visual collection of spoken words, numbers and sentences in Turkish language. The database was collected between 2006 and 2010 during ''''''``Novel approaches in audio-visual speech recognition'''''''' project which is funded by The Scientific and Technological Research Council of Turkey (TUBITAK). First part of the database contains a large corpus of Turkish language and contains standart quality videos. The second part is relatively small compared to the first one and contains recordings of spoken digits in high quality videos. Although the main aim to collect SUTAV database was to obtain a database for audio-visual speech recognition applications, it also contains useful data that can be used in other kinds of multimodal research like biometric security and person verification. The paper presents information about the data collection process and the the spoken content. It also contains a sample evaluation protocol and recognition results that are obtained with a small portion of the database. | ['Ibrahim Saygin Topkaya', 'Hakan Erdogan'] | 2012-05-01 | null | null | null | lrec-2012-5 | ['person-identification', 'audio-visual-speech-recognition'] | ['computer-vision', 'speech'] | [-1.78655609e-01 -4.01310384e-01 2.14197308e-01 -3.66511256e-01
-1.04973328e+00 -3.32359016e-01 4.66170818e-01 2.19878197e-01
-5.30792117e-01 6.10472381e-01 2.88416594e-01 -1.69316277e-01
1.95207670e-01 -4.48922068e-01 -5.15636504e-01 -8.76369178e-01
9.17932391e-02 1.98241100e-01 4.78434283e-03 -1.69821069e-01
3.59592468e-01 5.69916964e-01 -1.94093335e+00 5.88578880e-01
2.83586740e-01 1.04248619e+00 3.16474378e-01 1.02617967e+00
1.76114701e-02 6.83051586e-01 -7.34047055e-01 -3.29714894e-01
-7.76871368e-02 -3.84127736e-01 -7.98590899e-01 4.37600017e-01
4.48634148e-01 -5.20707250e-01 -3.68617356e-01 8.60632479e-01
1.00934100e+00 2.21814618e-01 6.52870357e-01 -1.16850019e+00
-6.29848123e-01 4.02318418e-01 -2.29663596e-01 3.09537858e-01
8.00573349e-01 1.08275056e-01 7.00844765e-01 -1.07508087e+00
6.19010627e-01 1.30913401e+00 3.17480803e-01 5.33923209e-01
-7.24510908e-01 -5.06112516e-01 -1.75509334e-01 4.63771790e-01
-1.77127922e+00 -9.26351964e-01 7.41801262e-01 -5.99437654e-01
1.08006334e+00 2.53896713e-01 8.06665480e-01 1.06888199e+00
-4.06444401e-01 1.04007494e+00 9.28503990e-01 -9.64314163e-01
1.06305845e-01 6.26222730e-01 2.45857492e-01 4.15780365e-01
-2.45395109e-01 1.08427010e-01 -4.24237043e-01 -5.64247509e-03
4.71914560e-01 -4.53206778e-01 -3.23923647e-01 9.03623737e-03
-9.34119940e-01 7.34790623e-01 -1.85289532e-01 7.71829069e-01
-2.82937825e-01 -4.54345644e-01 6.72656596e-01 6.26303852e-01
1.56624749e-01 -5.36432087e-01 -2.13133544e-01 -4.99904424e-01
-6.64885044e-01 7.16920868e-02 6.80877268e-01 9.68659937e-01
2.87145287e-01 6.84436858e-01 2.89906919e-01 1.51806819e+00
6.55692935e-01 1.08965683e+00 4.70658869e-01 -7.62222111e-01
3.78083587e-01 2.81322569e-01 2.35139113e-02 -8.96440804e-01
-4.64334525e-02 2.83165783e-01 -3.99780482e-01 1.32742241e-01
6.02345467e-01 -1.77353904e-01 -8.90218675e-01 1.15169191e+00
1.30644009e-01 -6.38995886e-01 4.46686566e-01 9.43530083e-01
1.61193419e+00 1.09221268e+00 -3.37229490e-01 -5.34850717e-01
1.82963526e+00 -4.04777676e-01 -9.87279236e-01 4.26380783e-01
4.81283128e-01 -1.23829353e+00 1.13814557e+00 5.28331399e-01
-1.17297494e+00 -4.93836075e-01 -6.97396755e-01 9.91958901e-02
-3.67776275e-01 8.03073168e-01 2.32162997e-01 1.14527154e+00
-1.29814041e+00 -4.36705917e-01 -2.30714276e-01 -7.70659149e-01
2.81084061e-01 1.75326079e-01 -7.98951387e-01 -2.95238137e-01
-1.14977562e+00 8.90557349e-01 1.19620785e-01 8.61958414e-02
-8.07461977e-01 -3.06039508e-02 -1.14122856e+00 -1.65963769e-01
1.73121810e-01 1.12040982e-01 1.23617232e+00 -9.93089974e-01
-1.44152701e+00 9.80041742e-01 -5.72529614e-01 2.17869785e-02
2.06766695e-01 1.40619442e-01 -7.61119246e-01 6.20875835e-01
-1.25660986e-01 5.64994633e-01 8.02058399e-01 -1.01214647e+00
-4.94302839e-01 -4.55488265e-01 -4.39582676e-01 1.43341318e-01
-1.52908787e-01 6.51900053e-01 -4.99033719e-01 -9.20077026e-01
-2.05546737e-01 -6.24543488e-01 5.69735229e-01 -3.99451196e-01
-1.40738100e-01 -4.13575023e-01 1.03895974e+00 -9.99086142e-01
1.01256716e+00 -2.21345019e+00 -1.53042495e-01 2.69927144e-01
-2.45116547e-01 5.47178626e-01 -6.70534521e-02 8.93004417e-01
-1.29511401e-01 -8.09297264e-02 1.37356997e-01 -1.99417815e-01
-2.04155013e-01 9.00447741e-03 -1.14811838e-01 7.28311598e-01
-1.06509559e-01 2.61755794e-01 -5.42615712e-01 -5.70167780e-01
5.05564392e-01 6.39258742e-01 -2.73595572e-01 -2.02404246e-01
3.65549624e-01 2.04374403e-01 -2.79547453e-01 1.12340772e+00
5.47086954e-01 5.83968937e-01 -2.96091497e-01 -2.16900229e-01
-3.29263270e-01 1.40697151e-01 -1.24967360e+00 1.13310230e+00
1.57444850e-01 1.26522386e+00 2.15656042e-01 -1.35491943e+00
1.00336540e+00 9.94490683e-01 3.69104236e-01 -7.16112733e-01
5.41658521e-01 2.38212556e-01 -1.98689029e-01 -9.22063351e-01
6.59885705e-01 -2.43924677e-01 1.33538663e-01 2.91924715e-01
1.06552921e-01 -1.21338457e-01 4.18518960e-01 1.03920460e-01
3.51286143e-01 -3.23101878e-01 1.02785893e-01 -5.67757040e-02
1.01618028e+00 -1.03893969e-02 3.18613231e-01 1.81746453e-01
-4.83908713e-01 4.37941551e-01 3.83943558e-01 -1.57365441e-01
-1.12902272e+00 -7.78983295e-01 -4.69641089e-01 7.70726085e-01
-5.69777191e-01 -2.52054125e-01 -7.33300388e-01 1.42237749e-02
-2.77931660e-01 4.03010070e-01 -3.46014053e-01 3.65910113e-01
-1.29152790e-01 -2.51636565e-01 9.07772064e-01 4.23848838e-01
5.91353416e-01 -1.10904884e+00 -1.53978139e-01 -2.61368513e-01
-6.43486857e-01 -1.12239850e+00 -4.88492757e-01 -4.24979120e-01
-4.54971761e-01 -1.13155425e+00 -1.26133156e+00 -1.07615805e+00
5.42798162e-01 3.11329722e-01 4.98561382e-01 -3.73082012e-01
-6.97980702e-01 8.33837271e-01 -6.52773440e-01 -7.57242560e-01
-5.71751058e-01 -5.59138358e-01 3.41355383e-01 1.14889741e-01
6.96089029e-01 1.18073791e-01 -6.98026717e-02 3.65398586e-01
-6.97709203e-01 -6.61162436e-01 3.35906923e-01 8.03081334e-01
3.01221937e-01 2.24220976e-01 5.38159192e-01 -1.82551339e-01
7.31393039e-01 1.48688182e-01 -6.58337831e-01 1.84604257e-01
5.13828620e-02 -6.65645182e-01 2.25456432e-01 -3.13384801e-01
-9.30271864e-01 -2.36521587e-01 -4.44008440e-01 -4.33901638e-01
-6.65696681e-01 3.12450022e-01 -3.88492912e-01 2.37727940e-01
4.70826894e-01 6.37626588e-01 2.75114715e-01 -4.27590072e-01
-1.89960897e-01 1.53755963e+00 6.38103068e-01 -3.47830057e-01
2.86742330e-01 2.36999348e-01 -3.97484303e-01 -1.82414007e+00
2.08114654e-01 -8.00366163e-01 -4.39832360e-01 -6.53197706e-01
7.58434653e-01 -1.11415339e+00 -1.13299453e+00 1.01537609e+00
-8.85748327e-01 2.58542508e-01 -5.90347052e-02 9.87885356e-01
-3.30010951e-01 5.88791192e-01 -7.32762575e-01 -1.53021717e+00
-1.95932910e-01 -1.45803130e+00 1.02168703e+00 2.00277299e-01
-9.82492343e-02 -8.45422089e-01 -2.45509967e-02 6.74606264e-01
4.44836318e-02 -1.72141969e-01 4.67098653e-01 -5.56830049e-01
-1.44800562e-02 -4.72041696e-01 -1.81054443e-01 7.18741894e-01
1.36382326e-01 4.86180454e-01 -1.27542961e+00 -5.23299396e-01
-1.02546759e-01 -4.68973756e-01 5.68162084e-01 5.47415137e-01
1.43870115e-01 -2.95250088e-01 1.57864571e-01 -3.07496905e-01
1.08080709e+00 1.02487171e+00 1.03390336e+00 1.44564003e-01
1.02438420e-01 1.08195162e+00 6.38794303e-01 4.29538608e-01
4.49239641e-01 9.95810330e-01 -1.50985882e-01 2.80413389e-01
-6.68389499e-02 6.32075146e-02 8.96356761e-01 1.13069427e+00
-2.58573800e-01 8.91291648e-02 -7.63567805e-01 8.46302867e-01
-1.29077661e+00 -1.33557355e+00 -2.69890189e-01 2.20695829e+00
4.23700750e-01 -5.93732178e-01 8.61726046e-01 7.82547593e-01
8.53882968e-01 -1.88804641e-01 2.28830472e-01 -8.72915208e-01
-2.79526114e-01 -9.61959437e-02 -2.26503164e-01 6.44100428e-01
-9.69948649e-01 6.64318621e-01 5.89057207e+00 8.90819967e-01
-1.46454263e+00 -3.07214320e-01 2.24325940e-01 -6.21141382e-02
1.59142211e-01 -4.09074217e-01 -7.60099828e-01 5.31439364e-01
1.22657144e+00 5.69164976e-02 -4.77852765e-03 5.11820495e-01
6.94694102e-01 -5.90078592e-01 -7.14171767e-01 1.48186362e+00
6.06684089e-01 -1.07166409e+00 2.00534165e-01 1.92162693e-01
7.06064478e-02 -3.80620658e-01 -4.12237737e-03 9.10973027e-02
-2.14559197e-01 -9.03434038e-01 8.26080024e-01 3.06857765e-01
1.05917180e+00 -1.07858348e+00 9.61555004e-01 1.44685984e-01
-1.12503493e+00 -6.50564060e-02 -4.40975308e-01 2.92209834e-01
-1.87235978e-02 3.84153873e-01 -7.90709198e-01 6.15189433e-01
1.05780637e+00 5.55165291e-01 -2.94474572e-01 1.08218610e+00
7.76414648e-02 6.00561321e-01 -5.45159392e-02 -3.47184032e-01
1.27106989e-02 -2.92952478e-01 7.39511728e-01 1.38758016e+00
3.85056436e-01 2.47076318e-01 -1.59648851e-01 2.28806525e-01
2.08053946e-01 6.15092754e-01 -9.72258806e-01 -5.16593575e-01
5.71739674e-01 7.59365439e-01 -5.88920295e-01 -3.02930504e-01
-5.01419902e-01 5.55499434e-01 -4.37702298e-01 3.20398360e-01
-4.42002863e-01 -7.73324430e-01 3.65086257e-01 -2.25086287e-02
2.90622622e-01 4.11242880e-02 1.03077017e-01 -1.03758347e+00
4.78556789e-02 -9.90219653e-01 3.57262850e-01 -1.10636973e+00
-8.41360688e-01 7.44437754e-01 2.01993898e-01 -1.34519541e+00
-4.62798119e-01 -7.02700377e-01 -6.36932179e-02 1.02213979e+00
-8.10710728e-01 -1.15601075e+00 -1.21011995e-01 8.63751650e-01
7.48909175e-01 -8.85206461e-01 9.58885312e-01 5.35872340e-01
-5.82861423e-01 7.58512020e-01 1.47683471e-01 5.70080698e-01
6.75156355e-01 -7.82410979e-01 -4.07148033e-01 5.53116143e-01
-1.99130177e-02 5.11387467e-01 7.68158019e-01 -2.49301985e-01
-1.30331540e+00 -5.28430104e-01 1.36566472e+00 -1.50607631e-01
3.30718279e-01 -3.70424837e-02 -4.75446999e-01 4.64679450e-01
4.77209926e-01 -4.97486860e-01 1.07490349e+00 -4.68198031e-01
-1.29387021e-01 -2.74151415e-01 -1.47526109e+00 3.26821297e-01
5.54670282e-02 -9.19079125e-01 -6.66459978e-01 9.80070904e-02
-1.50901213e-01 -2.31290102e-01 -8.14490020e-01 -1.32260859e-01
8.27530324e-01 -7.07863629e-01 9.40663576e-01 -2.87990302e-01
2.87034422e-01 -2.01565459e-01 -4.94965136e-01 -1.01681077e+00
2.11371928e-01 -3.17601621e-01 6.11769497e-01 1.49955475e+00
9.78244543e-02 -6.51178420e-01 4.30576086e-01 1.57195374e-01
-2.73547500e-01 -2.62751013e-01 -1.24359322e+00 -7.29762375e-01
-1.39842212e-01 -7.88437784e-01 2.03604668e-01 7.82481372e-01
2.72613525e-01 1.34249598e-01 -4.71074253e-01 -9.30849463e-02
5.51236212e-01 -4.77553159e-01 8.01751316e-01 -8.78883481e-01
8.05133358e-02 -2.78579295e-01 -8.85151625e-01 -7.26231515e-01
-5.96977174e-02 -4.01956022e-01 -1.51696399e-01 -1.26507628e+00
1.17472932e-01 6.68418556e-02 2.72184879e-01 3.57093334e-01
4.04991955e-01 4.51011449e-01 1.55554473e-01 2.50467211e-01
1.69136405e-01 2.80172318e-01 1.15560079e+00 -3.84828568e-01
-2.68607527e-01 1.01303592e-01 -3.22043747e-01 3.30074281e-01
4.88450766e-01 7.63438940e-02 -4.66049969e-01 -1.99847713e-01
-4.37463284e-01 4.49030250e-01 4.44791824e-01 -4.08178449e-01
1.89938590e-01 1.49114862e-01 2.97137529e-01 -1.00883853e+00
9.03146386e-01 -6.42233670e-01 7.08208680e-02 3.62119496e-01
-1.10011008e-02 2.60376066e-01 4.05769140e-01 6.90533519e-02
-8.11561346e-01 -1.49095908e-01 1.15650880e+00 -1.56580776e-01
-7.77134597e-01 -3.51907939e-01 -1.06702936e+00 -3.55875432e-01
1.26249123e+00 -6.24038041e-01 -3.08856159e-01 -9.91494417e-01
-1.00298035e+00 8.71385038e-02 3.65794867e-01 2.49733657e-01
1.07836378e+00 -1.57903922e+00 -7.65693963e-01 2.61249006e-01
1.51080400e-01 -4.01015908e-01 6.20302200e-01 1.06773961e+00
-8.36729467e-01 1.04757786e+00 -3.69921178e-01 -8.24361742e-01
-2.16811967e+00 2.74119645e-01 4.29256521e-02 5.94273388e-01
-6.35205805e-01 6.92609072e-01 -2.06726938e-01 -6.10932633e-02
5.41343153e-01 1.87490284e-02 -7.37238824e-01 4.94550258e-01
8.50409329e-01 4.45918858e-01 1.53273135e-01 -1.56819773e+00
-5.81372201e-01 4.27525669e-01 2.21609008e-02 -6.13083065e-01
1.15159059e+00 -4.59591210e-01 -1.34690553e-01 9.66642141e-01
1.36970282e+00 -7.52329640e-03 -2.23594755e-01 -7.47815296e-02
-4.05417830e-01 -7.51511812e-01 -1.21669300e-01 -6.69378221e-01
-1.22253978e+00 1.17134893e+00 7.32819259e-01 -4.56275186e-03
1.11679614e+00 7.23080039e-02 4.56816196e-01 3.80663067e-01
3.64724696e-01 -1.32837880e+00 -1.25992149e-01 5.08421659e-01
1.10444331e+00 -1.09455287e+00 -3.51373196e-01 -2.06021711e-01
-8.64618540e-01 1.33178222e+00 -1.80159155e-02 5.58275163e-01
5.20438433e-01 8.48421082e-03 5.23781240e-01 8.45941156e-02
-4.08835530e-01 -2.79357791e-01 3.65477234e-01 1.09401941e+00
5.97389638e-01 -1.98331267e-01 -3.30168188e-01 4.31590319e-01
-4.15050894e-01 9.61821601e-02 8.47723901e-01 1.02669311e+00
-4.86727685e-01 -9.98161197e-01 -1.17813051e+00 1.40666336e-01
-6.86612546e-01 2.75860988e-02 -5.51653385e-01 8.36757958e-01
-2.15561047e-01 1.40098274e+00 -6.77561238e-02 -2.51433879e-01
4.46641803e-01 1.29865780e-01 5.89471042e-01 -1.46216840e-01
-3.06089193e-01 6.78460777e-01 4.55209076e-01 -7.46977981e-03
-6.95351958e-01 -1.11719418e+00 -9.71532345e-01 -7.49002755e-01
-2.31449887e-01 6.07456207e-01 8.35938692e-01 7.99031079e-01
-2.25781843e-01 2.90801048e-01 5.32052040e-01 -7.80705988e-01
-4.92856652e-02 -1.21009219e+00 -1.08022213e+00 2.84617901e-01
2.64923841e-01 -5.76392531e-01 -1.86464369e-01 4.06544536e-01] | [14.318987846374512, 5.1222076416015625] |
0029b3f2-514e-4e6b-8ae9-6770144cf9e2 | epic-ensemble-of-partial-point-clouds-for | 2303.11419 | null | https://arxiv.org/abs/2303.11419v2 | https://arxiv.org/pdf/2303.11419v2.pdf | EPiC: Ensemble of Partial Point Clouds for Robust Classification | Robust point cloud classification is crucial for real-world applications, as consumer-type 3D sensors often yield partial and noisy data, degraded by various artifacts. In this work we propose a general ensemble framework, based on partial point cloud sampling. Each ensemble member is exposed to only partial input data. Three sampling strategies are used jointly, two local ones, based on patches and curves, and a global one of random sampling. We demonstrate the robustness of our method to various local and global degradations. We show that our framework significantly improves the robustness of top classification netowrks by a large margin. Our experimental setting uses the recently introduced ModelNet-C database by Ren et al.[24], where we reach SOTA both on unaugmented and on augmented data. Our unaugmented mean Corruption Error (mCE) is 0.64 (current SOTA is 0.86) and 0.50 for augmented data (current SOTA is 0.57). We analyze and explain these remarkable results through diversity analysis. Our code is available at: https://github.com/yossilevii100/EPiC | ['Guy Gilboa', 'Meir Yossef Levi'] | 2023-03-20 | null | null | null | null | ['point-cloud-classification'] | ['computer-vision'] | [-3.04395556e-02 -3.86550635e-01 -1.94902578e-03 -3.38203758e-02
-1.06247294e+00 -7.79907167e-01 5.89714348e-01 3.82494256e-02
-8.10699984e-02 8.03449035e-01 -2.01471031e-01 -7.79886618e-02
-5.27469106e-02 -8.37957025e-01 -1.07836676e+00 -1.10157037e+00
-2.43529961e-01 2.94214755e-01 3.07356060e-01 -1.69888631e-01
3.42459768e-01 6.93300724e-01 -1.96574700e+00 1.69273153e-01
1.13850880e+00 1.22929466e+00 7.23959804e-02 7.41198361e-01
2.85563976e-01 2.88717598e-01 -8.49366307e-01 -1.18230335e-01
5.15238881e-01 2.14064404e-01 -1.71091348e-01 5.36372550e-02
5.85160851e-01 -9.31506008e-02 5.00721782e-02 1.13114727e+00
6.81326568e-01 -1.64768040e-01 7.07523346e-01 -1.63292873e+00
-3.21142167e-01 2.47034520e-01 -6.34874880e-01 -6.11506738e-02
9.51642618e-02 3.29208314e-01 6.90359354e-01 -9.75241005e-01
4.27742988e-01 9.53305900e-01 7.76131272e-01 2.41687730e-01
-1.29009080e+00 -1.00295901e+00 1.34076431e-01 4.36629280e-02
-1.69412088e+00 -4.48241532e-01 8.61834347e-01 -5.83011985e-01
5.82129717e-01 5.27495146e-01 5.38142622e-01 1.39354682e+00
2.91066736e-01 5.52010715e-01 1.12298930e+00 -1.37287229e-02
6.84410155e-01 5.57036838e-03 2.09926233e-01 1.18457444e-01
6.61016285e-01 3.34957957e-01 -5.13984621e-01 -6.94494605e-01
4.90669787e-01 7.42687285e-03 -3.86724144e-01 -4.76503909e-01
-1.05053818e+00 4.29541647e-01 4.10992354e-01 -5.01746079e-03
-4.42669809e-01 1.31793454e-01 1.29510105e-01 3.49326670e-01
6.00624323e-01 6.12498671e-02 -6.46670640e-01 -1.44571230e-01
-7.40710974e-01 4.09096479e-01 6.17789090e-01 1.29398274e+00
7.69200802e-01 -1.57499820e-01 2.17670873e-01 8.79333615e-01
2.18679979e-01 1.10644424e+00 -1.18111782e-02 -8.99360359e-01
5.33358276e-01 5.15086770e-01 3.99227023e-01 -1.06948400e+00
-2.69000620e-01 -6.51836991e-01 -1.15957999e+00 4.95222300e-01
1.88862368e-01 -1.04313225e-01 -8.13195646e-01 1.42973602e+00
4.02367085e-01 4.68043566e-01 -1.05303511e-01 7.79599845e-01
5.93549728e-01 3.87413353e-01 -1.97714224e-01 -2.27642238e-01
1.00311983e+00 -4.68437761e-01 -3.67849171e-01 1.08169325e-01
3.42070818e-01 -5.60852289e-01 9.93499994e-01 9.69374835e-01
-6.91927254e-01 -3.91722918e-01 -1.21299160e+00 5.80517530e-01
-1.84923574e-01 2.49855574e-02 1.02741256e-01 7.60512471e-01
-9.06519294e-01 7.56555319e-01 -9.90280986e-01 -2.69411117e-01
6.71582937e-01 1.47803783e-01 -2.77602226e-01 -9.52987969e-02
-7.20458031e-01 5.29669046e-01 -3.23353671e-02 1.09425403e-01
-7.72150159e-01 -7.00897872e-01 -3.73248905e-01 -4.20588821e-01
2.88864404e-01 -6.81181490e-01 8.48127902e-01 -4.70672339e-01
-1.26064527e+00 4.84148681e-01 -1.74984545e-01 -2.19338089e-01
7.94772446e-01 -3.90072465e-01 -4.53061998e-01 -1.17301829e-01
3.09799835e-02 1.44913539e-01 8.03335667e-01 -1.86026287e+00
-5.48893094e-01 -7.97521472e-01 -4.16704118e-01 -1.35632649e-01
-2.04043575e-02 -2.03378126e-01 -2.01815203e-01 -6.66790962e-01
5.16667843e-01 -1.08207154e+00 -2.62794346e-01 9.36408415e-02
-4.97499496e-01 2.36069709e-02 7.49643445e-01 -4.04626042e-01
1.01010954e+00 -2.10111523e+00 -4.67385575e-02 5.52803993e-01
2.11660996e-01 -4.17842343e-02 8.48424900e-03 4.04558957e-01
2.73304079e-02 4.08222497e-01 -7.04408109e-01 -5.05153239e-01
-1.65784746e-01 2.39681378e-01 -4.92753714e-01 7.79843807e-01
8.30982998e-02 3.38154197e-01 -8.66130888e-01 -1.23815080e-02
3.00859779e-01 3.96882176e-01 -3.27058077e-01 4.79379371e-02
-3.89804095e-02 5.12133956e-01 -3.35075289e-01 1.30142272e+00
1.35750365e+00 1.09088413e-01 -2.28887409e-01 -1.02981620e-01
-9.67565626e-02 -1.44002184e-01 -1.58899212e+00 1.58264840e+00
-2.15796143e-01 2.40310699e-01 3.05155486e-01 -4.78508562e-01
1.05074871e+00 1.65913731e-01 5.93899071e-01 -2.66943097e-01
1.58784956e-01 4.75356132e-01 -3.13264191e-01 -1.22021362e-01
6.64891243e-01 1.62719741e-01 -1.95883438e-02 3.74289006e-01
-3.38073939e-01 -3.43383580e-01 -2.25309312e-01 3.52865532e-02
1.32780755e+00 7.85449613e-03 3.67242545e-02 -4.94628638e-01
2.49006584e-01 -7.14614391e-02 5.27128279e-01 9.55123603e-01
-4.24849927e-01 1.17672348e+00 3.11265677e-01 -3.63165401e-02
-1.10536373e+00 -1.21102583e+00 -4.89794225e-01 4.29998308e-01
4.24433887e-01 -4.66916025e-01 -5.28935790e-01 -5.79649329e-01
5.69739878e-01 7.83836186e-01 -4.68241543e-01 -1.38279721e-01
-4.21986312e-01 -7.88777173e-01 5.66089392e-01 3.54385644e-01
3.57800603e-01 -6.41411006e-01 -3.45240146e-01 -1.25906467e-01
-5.11608645e-02 -9.79833245e-01 9.03421342e-02 -6.58354834e-02
-1.20212436e+00 -1.08309174e+00 -3.33457977e-01 4.64750119e-02
5.26080251e-01 3.65741223e-01 1.24557555e+00 2.12254792e-01
1.54780343e-01 2.63116986e-01 -6.06301129e-01 -5.84678173e-01
-2.83822000e-01 1.52514204e-01 5.86772382e-01 -5.91507088e-03
1.52937889e-01 -9.76028025e-01 -6.00550056e-01 7.02358246e-01
-8.28849852e-01 -3.66108626e-01 2.99723953e-01 6.84528649e-01
7.57793665e-01 -5.42838536e-02 3.77643675e-01 -5.75391829e-01
4.38404709e-01 -8.76006901e-01 -7.35502422e-01 -1.24432035e-01
-5.77376425e-01 -2.29441240e-01 3.31362814e-01 -2.35790581e-01
-4.27074045e-01 5.10198250e-02 -1.24247067e-01 -8.85449588e-01
-4.51669574e-01 1.73958540e-02 -3.05493832e-01 -1.98623672e-01
8.98946881e-01 7.52456859e-02 -1.68844655e-01 -6.94934487e-01
1.63808137e-01 1.07122397e+00 4.57688034e-01 -8.55493367e-01
1.02124035e+00 6.75504148e-01 -8.77347067e-02 -1.00027537e+00
-1.00154616e-01 -3.77349138e-01 -6.03726864e-01 -3.40671718e-01
1.04632661e-01 -1.02272904e+00 -5.94661593e-01 7.91362405e-01
-1.18126440e+00 -1.76505804e-01 -2.72822648e-01 1.55180663e-01
-3.87129962e-01 3.19613189e-01 -4.46576364e-02 -1.08381939e+00
-2.98275828e-01 -1.19739187e+00 1.51670110e+00 -3.52933437e-01
5.33446930e-02 -4.08818871e-01 1.43559679e-01 5.71742170e-02
2.84226209e-01 7.63366759e-01 2.19215333e-01 -4.23818737e-01
-7.33961403e-01 -2.58542866e-01 2.98763178e-02 4.65696275e-01
1.03160463e-01 3.62697303e-01 -1.17989314e+00 -5.52943468e-01
1.29354209e-01 2.94624548e-02 8.50843549e-01 1.45434842e-01
1.37437475e+00 2.50326302e-02 -4.08699393e-01 6.91269338e-01
1.53078532e+00 -1.14147037e-01 8.10358405e-01 2.66775101e-01
6.31040990e-01 1.30986333e-01 6.77506804e-01 6.53504014e-01
9.22794938e-02 6.64919138e-01 9.26777303e-01 3.65414590e-01
9.36522782e-02 6.68672249e-02 2.90174752e-01 7.06831276e-01
-4.56302941e-01 -3.58018160e-01 -1.18532693e+00 6.19448841e-01
-1.85042024e+00 -7.58778870e-01 -4.99281973e-01 2.42281342e+00
4.41868871e-01 9.43546519e-02 -4.47425321e-02 2.50434399e-01
8.34063828e-01 1.97415128e-01 -5.65720558e-01 1.62356779e-01
-3.75039101e-01 1.39499217e-01 8.09993446e-01 3.23035240e-01
-1.06948459e+00 4.32534218e-01 5.82849169e+00 8.12953055e-01
-8.65842700e-01 1.91993937e-01 3.66731197e-01 -3.36166501e-01
-1.83039367e-01 -2.85445124e-01 -6.30630493e-01 9.08065617e-01
8.84400845e-01 1.14697367e-02 6.38107657e-02 8.88350785e-01
1.45824939e-01 -3.02002400e-01 -9.08849478e-01 1.19962502e+00
4.94916458e-03 -1.12208891e+00 -2.19627768e-01 2.88403571e-01
8.48810673e-01 6.64381087e-01 -1.07198462e-01 2.26634555e-03
3.96572024e-01 -6.83302045e-01 9.44331884e-01 5.42309403e-01
8.46390009e-01 -6.53261781e-01 8.42731118e-01 4.90046084e-01
-1.04188073e+00 4.46815696e-03 -3.21218133e-01 -1.42019019e-01
1.70254111e-02 1.23003411e+00 -4.18206304e-01 8.54998827e-01
1.09544885e+00 6.99043751e-01 -5.30678630e-01 1.17969143e+00
-1.38391122e-01 6.49816871e-01 -9.12188172e-01 -2.07189675e-02
-4.47750807e-01 -1.77572429e-01 1.26957810e+00 8.44386041e-01
6.95321798e-01 7.80748054e-02 -2.52260659e-02 7.63546348e-01
-2.22015679e-02 -2.21313938e-01 -8.35362673e-01 8.37388635e-01
8.96831274e-01 1.01152444e+00 -2.79586047e-01 -2.50662953e-01
9.54075456e-02 7.45847583e-01 7.49009326e-02 3.94483805e-01
-7.91345358e-01 -2.04618663e-01 1.21581316e+00 6.81959614e-02
2.05494791e-01 -3.65740955e-01 -7.76185870e-01 -1.41463518e+00
4.50552583e-01 -6.94584668e-01 5.90272211e-02 -7.98052669e-01
-1.53783607e+00 5.88383496e-01 7.68778250e-02 -1.88521540e+00
1.26441970e-01 -6.07402742e-01 -4.36585784e-01 6.93732619e-01
-1.09699750e+00 -8.24213564e-01 -7.75149167e-01 4.30550754e-01
2.09986180e-01 -1.29449457e-01 6.09545588e-01 4.25076336e-01
-6.21635556e-01 5.32910824e-01 3.27036858e-01 -2.90971577e-01
6.18494928e-01 -1.01250911e+00 5.45898616e-01 8.92366588e-01
-1.47314504e-01 3.83101881e-01 9.32172775e-01 -7.37603366e-01
-1.45389748e+00 -1.32429934e+00 3.78009945e-01 -1.03087437e+00
5.20990491e-01 -5.96714199e-01 -9.87805188e-01 4.53213185e-01
-1.10606343e-01 2.74838179e-01 3.05440634e-01 -7.32114073e-03
-3.44889462e-01 -3.36491913e-01 -1.59251726e+00 4.48050082e-01
1.28573203e+00 -1.83750555e-01 -1.96888581e-01 3.11268359e-01
7.27666438e-01 -6.01106942e-01 -9.93316889e-01 8.22079659e-01
5.31889737e-01 -1.20019901e+00 8.25862288e-01 -9.69457328e-02
2.19782650e-01 -7.19703615e-01 -7.77781069e-01 -1.43187666e+00
-8.31961557e-02 -4.04323697e-01 -2.73554295e-01 1.06860960e+00
3.11611444e-01 -1.08397651e+00 7.76777148e-01 4.18903530e-01
-3.84345770e-01 -5.96314013e-01 -1.16624236e+00 -1.08369446e+00
1.85902417e-01 -9.74038720e-01 1.03893411e+00 9.60333943e-01
-3.14780444e-01 -2.14831606e-01 -1.71471596e-01 7.30028927e-01
1.03653526e+00 -3.29313353e-02 1.19381392e+00 -1.43181229e+00
6.59032539e-03 -2.30786994e-01 -5.04390657e-01 -7.88872600e-01
-4.12323087e-01 -6.99647784e-01 -7.27677643e-02 -1.03615093e+00
-9.63615905e-03 -7.54680037e-01 -4.26120818e-01 3.18972260e-01
4.29512002e-02 3.98289800e-01 1.93976581e-01 5.60695469e-01
-3.51852089e-01 5.81235051e-01 7.50421345e-01 -7.33696446e-02
-1.98505878e-01 2.34053165e-01 -4.15584475e-01 5.55762410e-01
1.24253654e+00 -6.03034914e-01 1.19787328e-01 -3.91558141e-01
1.32785276e-01 -3.07245672e-01 6.42365694e-01 -1.45468438e+00
7.06105232e-02 -9.84167680e-02 2.15841040e-01 -1.07587314e+00
3.10955614e-01 -1.17570877e+00 7.41450846e-01 2.83659756e-01
3.76845211e-01 -3.81111763e-02 3.14725935e-01 6.33040130e-01
-6.62705526e-02 1.56060308e-01 6.73665643e-01 1.62233144e-01
-3.28523815e-01 3.64916921e-01 1.23144045e-01 -4.03536916e-01
1.04663563e+00 -2.66073316e-01 -7.04169154e-01 -1.43598333e-01
-6.17229819e-01 2.85063028e-01 9.70680356e-01 3.29879493e-01
5.58496058e-01 -1.57275915e+00 -9.37748969e-01 4.54937190e-01
4.65072602e-01 3.40145797e-01 2.26363823e-01 7.99022079e-01
-4.79320407e-01 -1.63249597e-01 6.60859048e-02 -1.34267771e+00
-1.17321754e+00 2.25983426e-01 3.08227420e-01 2.55301863e-01
-4.14047360e-01 6.23296499e-01 -3.66205573e-01 -7.42191494e-01
2.67141372e-01 -6.12386048e-01 3.45512420e-01 -6.17618598e-02
2.92243212e-01 7.42085755e-01 6.02319539e-01 -6.17712975e-01
-5.01580119e-01 8.31435025e-01 3.53220165e-01 7.80319721e-02
1.34676111e+00 1.47591541e-02 -1.57692879e-01 5.70896566e-01
9.66841340e-01 2.45563269e-01 -1.10433030e+00 -7.22214058e-02
-2.89669454e-01 -7.58473039e-01 -1.14134960e-01 -7.23276556e-01
-1.06473386e+00 5.20252585e-01 8.74830008e-01 4.48026955e-01
1.17766082e+00 -1.26903155e-03 5.06730855e-01 2.54046261e-01
9.60355163e-01 -7.63571918e-01 -4.61690098e-01 4.28641558e-01
1.22697973e+00 -1.23707151e+00 2.33809605e-01 -4.96907085e-01
-2.97305256e-01 6.18177176e-01 4.23197657e-01 -3.77061456e-01
7.99038887e-01 4.35288191e-01 -3.29642594e-02 -1.79890186e-01
-8.99803340e-01 3.34008746e-02 -1.38169512e-01 7.39666343e-01
-1.88823998e-01 2.71024406e-01 7.90075213e-02 6.16644740e-01
-3.81811768e-01 -5.75949736e-02 4.90819842e-01 9.11420822e-01
-4.50653702e-01 -8.41989279e-01 -9.80805099e-01 5.34096718e-01
-5.20818010e-02 2.09437326e-01 -3.87265205e-01 7.02710748e-01
2.47703001e-01 9.86694455e-01 3.04951429e-01 -8.80606532e-01
7.41447091e-01 -1.66316986e-01 1.98736131e-01 -2.55698919e-01
-3.94469768e-01 -6.68171048e-02 1.00323465e-03 -7.91914165e-01
-4.07023013e-01 -1.00980818e+00 -7.81779528e-01 -6.19031787e-01
-5.04059732e-01 -9.02617276e-02 9.09922719e-01 2.89768010e-01
7.28558719e-01 3.76128018e-01 9.99736488e-01 -1.22808945e+00
-4.93343592e-01 -1.01232350e+00 -7.43390501e-01 2.23727852e-01
5.73040307e-01 -9.34047163e-01 -8.83804500e-01 -2.66229212e-01] | [7.686522483825684, -2.8788421154022217] |
f36c6d10-6e7f-446f-8aa9-cf31dd7cd6cd | gres-generalized-referring-expression-1 | 2306.00968 | null | https://arxiv.org/abs/2306.00968v1 | https://arxiv.org/pdf/2306.00968v1.pdf | GRES: Generalized Referring Expression Segmentation | Referring Expression Segmentation (RES) aims to generate a segmentation mask for the object described by a given language expression. Existing classic RES datasets and methods commonly support single-target expressions only, i.e., one expression refers to one target object. Multi-target and no-target expressions are not considered. This limits the usage of RES in practice. In this paper, we introduce a new benchmark called Generalized Referring Expression Segmentation (GRES), which extends the classic RES to allow expressions to refer to an arbitrary number of target objects. Towards this, we construct the first large-scale GRES dataset called gRefCOCO that contains multi-target, no-target, and single-target expressions. GRES and gRefCOCO are designed to be well-compatible with RES, facilitating extensive experiments to study the performance gap of the existing RES methods on the GRES task. In the experimental study, we find that one of the big challenges of GRES is complex relationship modeling. Based on this, we propose a region-based GRES baseline ReLA that adaptively divides the image into regions with sub-instance clues, and explicitly models the region-region and region-language dependencies. The proposed approach ReLA achieves new state-of-the-art performance on the both newly proposed GRES and classic RES tasks. The proposed gRefCOCO dataset and method are available at https://henghuiding.github.io/GRES. | ['Xudong Jiang', 'Henghui Ding', 'Chang Liu'] | 2023-06-01 | gres-generalized-referring-expression | http://openaccess.thecvf.com//content/CVPR2023/html/Liu_GRES_Generalized_Referring_Expression_Segmentation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Liu_GRES_Generalized_Referring_Expression_Segmentation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['referring-expression', 'generalized-referring-expression-segmentation', 'referring-expression-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.29843056e-01 -2.74327397e-02 -3.14791918e-01 -5.53907692e-01
-9.29176867e-01 -5.49763501e-01 4.39491093e-01 -1.75512701e-01
-3.05887520e-01 5.11346102e-01 -7.66022727e-02 -2.02653304e-01
2.31347248e-01 -6.11072421e-01 -6.55998707e-01 -5.05141675e-01
4.77723658e-01 3.38930100e-01 5.69008887e-01 -4.02050823e-01
8.56336430e-02 4.60089833e-01 -1.23332930e+00 4.76176232e-01
8.35296810e-01 1.10505867e+00 2.60553807e-01 2.66425610e-01
-1.48640603e-01 6.88566685e-01 -7.88335383e-01 -4.43658859e-01
1.69892728e-01 -5.02198696e-01 -9.37727809e-01 5.61919110e-03
1.00932986e-01 -1.52112432e-02 4.20716256e-02 9.86301303e-01
3.70553941e-01 2.97438264e-01 6.75740361e-01 -1.35990429e+00
-8.44135284e-01 7.22150505e-01 -1.02718985e+00 3.52647424e-01
3.65814358e-01 4.39513139e-02 1.07503951e+00 -9.58816886e-01
6.15541220e-01 1.38216424e+00 4.45764989e-01 4.71421480e-01
-8.77881765e-01 -8.57544720e-01 6.55318856e-01 3.13314833e-02
-1.72098231e+00 -1.44105315e-01 6.17451191e-01 -3.88339221e-01
8.39720666e-01 3.96636724e-01 5.02515733e-01 1.01531863e+00
-7.43814334e-02 1.14884734e+00 1.29895139e+00 -3.33482653e-01
3.05914599e-02 -1.38691906e-02 2.63031363e-01 5.36482871e-01
-2.25123912e-01 -3.54134232e-01 -2.93432057e-01 5.25032543e-02
8.03602934e-01 -2.79272527e-01 -3.64822656e-01 1.06804490e-01
-1.15377820e+00 6.17541611e-01 4.98346776e-01 5.38471699e-01
-2.08605960e-01 1.43850431e-01 4.48140711e-01 -6.48796856e-02
6.55967832e-01 2.40342528e-01 -5.06721914e-01 -7.43511394e-02
-8.58432770e-01 3.24832141e-01 4.20267552e-01 1.40030122e+00
7.09216595e-01 -1.75672755e-01 -5.35821557e-01 1.05983496e+00
1.69220462e-01 9.94223654e-02 3.92349720e-01 -8.12708437e-01
4.79335874e-01 9.27664638e-01 -4.87140529e-02 -9.23379183e-01
-4.37282771e-01 -5.13431370e-01 -7.05665290e-01 -2.24410385e-01
3.02533597e-01 -3.62968296e-02 -8.95544648e-01 1.81609654e+00
5.28584063e-01 4.10131544e-01 1.52290231e-02 1.00078917e+00
1.14214325e+00 6.96491063e-01 2.56275922e-01 -1.84659377e-01
1.50945687e+00 -1.29093993e+00 -6.81348801e-01 -2.48040199e-01
8.19114268e-01 -7.29002059e-01 1.55905986e+00 2.50994474e-01
-8.40547323e-01 -4.94981200e-01 -6.85840368e-01 -1.79094896e-01
-5.26544690e-01 3.90050024e-01 6.37462258e-01 2.47772112e-01
-8.27194571e-01 -5.56780919e-02 -6.11764252e-01 -2.77391672e-01
4.28750932e-01 3.32116455e-01 -2.86442995e-01 -1.11535758e-01
-1.34305465e+00 6.54482901e-01 6.35280788e-01 3.86021405e-01
-4.97622281e-01 -7.43948400e-01 -8.69493365e-01 -1.97647586e-01
8.86388719e-01 -5.10295510e-01 1.30625927e+00 -1.20741260e+00
-1.20020747e+00 1.30240941e+00 -3.64842236e-01 -1.61147356e-01
5.55348873e-01 -3.16203862e-01 -4.42015141e-01 1.05004221e-01
3.79780889e-01 7.75454104e-01 5.16320050e-01 -1.36026669e+00
-4.72844541e-01 -1.91554978e-01 4.01689947e-01 2.13891566e-01
2.77690053e-01 6.35225713e-01 -1.12640929e+00 -9.05235350e-01
2.14726236e-02 -7.67047942e-01 -1.01271585e-01 -3.12704027e-01
-7.31459439e-01 -6.79794312e-01 7.45356441e-01 -3.85815382e-01
1.34540403e+00 -2.26711559e+00 6.05505053e-03 -1.95119306e-02
1.29623964e-01 2.88426399e-01 -1.32345244e-01 1.00979008e-01
-3.22605699e-01 4.10114497e-01 -4.78594154e-01 -3.09747696e-01
-7.84998909e-02 2.84419805e-01 -1.43232808e-01 2.31927127e-01
3.68118525e-01 1.17453182e+00 -8.78428221e-01 -7.84469783e-01
-9.09315720e-02 2.13607088e-01 -2.41303220e-01 3.19580108e-01
-5.22334337e-01 7.38359571e-01 -7.38793731e-01 8.14538896e-01
8.26119125e-01 -3.13694954e-01 -2.09253311e-01 -3.27929020e-01
-2.85708215e-02 -3.73930670e-02 -1.10466826e+00 1.54472566e+00
-4.70301747e-01 2.63440818e-01 -7.35931695e-02 -8.43854904e-01
1.07621741e+00 -5.24679385e-02 4.94771093e-01 -6.16476893e-01
2.83442914e-01 3.09180319e-01 -6.05319776e-02 -4.67275113e-01
5.34204185e-01 -1.32699460e-01 -3.92709225e-01 1.64132640e-01
-1.32108241e-01 -1.36920996e-02 3.25026989e-01 4.20885503e-01
8.84814262e-01 5.29814363e-01 6.38349533e-01 -3.39885503e-01
7.70483732e-01 -1.73050035e-02 9.73111868e-01 6.09325051e-01
-2.35758096e-01 7.95056581e-01 8.12858462e-01 3.46570276e-02
-5.37106931e-01 -8.11290383e-01 -2.01316312e-01 1.29297936e+00
5.61038792e-01 -5.27211726e-01 -8.45800042e-01 -9.65520501e-01
-3.71260703e-01 8.73334646e-01 -7.59064674e-01 2.02468321e-01
-6.28214777e-01 -7.96435535e-01 7.27704763e-01 6.86910212e-01
8.40543270e-01 -1.23539376e+00 -4.18382436e-01 -2.67856196e-02
-3.15647304e-01 -1.47673380e+00 -7.47981906e-01 -1.91368654e-01
-2.52688348e-01 -1.21306407e+00 -8.90570402e-01 -9.74682748e-01
7.05865204e-01 8.61147884e-03 1.26215327e+00 8.57779533e-02
-1.03812348e-02 2.82412678e-01 -6.89378023e-01 -1.79277286e-01
-2.43073851e-01 2.61161834e-01 -4.97654706e-01 2.15351284e-01
3.57481569e-01 -2.82963455e-01 -5.84277749e-01 7.19871700e-01
-9.77285326e-01 4.53476518e-01 5.39519370e-01 6.25064671e-01
1.15072656e+00 -2.56355435e-01 7.50549197e-01 -1.16648555e+00
5.90755284e-01 -7.06130207e-01 -5.52084744e-01 5.51997960e-01
-2.07634285e-01 -2.29638368e-01 6.52481735e-01 -5.35623968e-01
-1.19950879e+00 -3.77091989e-02 -2.71326333e-01 -4.64916825e-01
-3.22508812e-01 5.97830892e-01 -4.40748513e-01 2.35527635e-01
4.35656726e-01 8.71859640e-02 -7.55311489e-01 -3.71577829e-01
4.70152557e-01 6.60736382e-01 7.48110235e-01 -1.01752377e+00
3.30992162e-01 1.81366622e-01 -1.07626870e-01 -5.31612575e-01
-1.09786522e+00 -6.96399033e-01 -6.96937501e-01 -1.31069660e-01
8.78053546e-01 -9.13381159e-01 -4.08704847e-01 6.14443302e-01
-1.35323715e+00 -5.73143423e-01 -2.06092611e-01 8.09326954e-03
-4.63109255e-01 2.50558019e-01 -5.34489512e-01 -7.49058902e-01
-2.20732972e-01 -1.32000411e+00 1.39259851e+00 3.67713451e-01
-8.39591697e-02 -7.98734307e-01 -3.67309511e-01 3.14441085e-01
7.00927079e-02 5.49335659e-01 5.59054732e-01 -7.82132864e-01
-4.72661644e-01 2.22339436e-01 -6.57554090e-01 1.84864059e-01
9.16926563e-02 -9.58435610e-03 -6.99806988e-01 9.41421092e-02
-1.47244483e-01 -3.83567333e-01 7.71147847e-01 1.22470923e-01
1.45905352e+00 -8.42332467e-02 -2.74966657e-01 6.47836745e-01
1.29258931e+00 3.57857198e-01 7.25264728e-01 2.47150332e-01
8.73836398e-01 6.60597563e-01 1.32411051e+00 1.84980571e-01
6.04948401e-01 8.81136537e-01 1.94245994e-01 -3.37269306e-01
-9.32717696e-02 -7.30831176e-02 2.03926444e-01 6.40184939e-01
-2.98338383e-02 -5.29850364e-01 -1.10596299e+00 6.48120403e-01
-2.07064939e+00 -4.11315024e-01 -4.76631492e-01 1.72066021e+00
9.34918225e-01 -6.84178770e-02 1.96286254e-02 -4.28120971e-01
9.30507004e-01 1.81330964e-01 -7.08986342e-01 -2.39939749e-01
-3.78082842e-01 1.98874190e-01 1.57266304e-01 3.23636949e-01
-1.12539768e+00 1.51390827e+00 4.88670158e+00 1.33085072e+00
-1.07043672e+00 1.93460077e-01 9.66232777e-01 1.48770392e-01
-2.02011809e-01 -4.10509519e-02 -1.07098722e+00 3.22700262e-01
5.94147325e-01 -1.71605796e-01 6.66151047e-02 7.12606132e-01
2.76226014e-01 -3.46874177e-01 -9.20528114e-01 1.02959943e+00
1.55083075e-01 -8.47251296e-01 -6.89460859e-02 -4.05448854e-01
6.76698446e-01 -9.11315158e-02 -1.32505313e-01 3.97765815e-01
7.55868927e-02 -1.06124246e+00 8.76056314e-01 2.65636176e-01
1.03616095e+00 -5.93384922e-01 8.59512150e-01 1.91812307e-01
-1.41697681e+00 2.59613037e-01 3.93485092e-03 3.50270420e-01
2.17292160e-01 3.39716226e-01 -4.35909480e-01 8.32715750e-01
7.54719079e-01 7.14135230e-01 -7.43819237e-01 7.66417086e-01
-5.54571211e-01 7.13855386e-01 -4.40495491e-01 1.45076692e-01
4.23165888e-01 -2.25108489e-01 2.94467926e-01 1.44191086e+00
1.54918626e-01 3.48829210e-01 3.41040730e-01 1.31460702e+00
-2.26337090e-01 5.59502304e-01 -2.26698905e-01 2.41180986e-01
4.51479703e-01 1.29054427e+00 -1.02503705e+00 -4.08582687e-01
-4.90079731e-01 9.73326623e-01 2.80041605e-01 5.48035800e-01
-1.22808552e+00 -2.76429117e-01 2.09997863e-01 1.88564971e-01
2.99007654e-01 2.13508774e-03 -1.72075331e-01 -1.12399888e+00
1.19301490e-01 -7.83044457e-01 6.12637818e-01 -1.13893354e+00
-1.21885443e+00 7.19697952e-01 5.27298212e-01 -1.04098392e+00
-9.71143395e-02 -4.93567616e-01 -5.44828892e-01 9.96013284e-01
-1.60991180e+00 -1.69950116e+00 -5.08747041e-01 6.75631762e-01
7.24023759e-01 1.57131314e-01 4.46280569e-01 2.93231159e-01
-9.49758768e-01 7.26048172e-01 -5.13588905e-01 2.51353055e-01
7.39393532e-01 -1.08736920e+00 3.92794818e-01 9.26994085e-01
1.21581607e-01 7.75765896e-01 4.25524116e-01 -6.38872147e-01
-6.94014966e-01 -1.27564657e+00 5.41298687e-01 -1.02591306e-01
6.66337192e-01 -4.80346203e-01 -1.08276474e+00 9.81258869e-01
2.57447250e-02 4.51360375e-01 4.23339754e-01 -9.21989977e-02
-3.59019071e-01 1.73261896e-01 -1.14032900e+00 7.46915579e-01
1.25801456e+00 -3.38519871e-01 -4.35566843e-01 3.80939275e-01
9.35570180e-01 -7.99784720e-01 -8.59920681e-01 6.35094762e-01
1.54577062e-01 -9.45853531e-01 6.92710817e-01 -4.74849790e-01
4.94842887e-01 -5.55753767e-01 -2.27206215e-01 -8.97747815e-01
2.24980898e-02 -5.80740809e-01 1.20278582e-01 1.73451495e+00
4.04421479e-01 -7.00624228e-01 4.95198339e-01 5.77331007e-01
-4.03417915e-01 -1.23786378e+00 -8.95790756e-01 -7.88194776e-01
2.27037132e-01 -6.43078923e-01 7.29561627e-01 9.85812545e-01
-4.17644292e-01 3.76238048e-01 -3.68181728e-02 2.08538279e-01
1.04253478e-01 3.58398020e-01 6.23664439e-01 -5.41261315e-01
-4.16192263e-01 -3.81035954e-01 -2.85127670e-01 -1.40579164e+00
4.38772321e-01 -8.96081984e-01 5.77408299e-02 -1.47186196e+00
2.06738830e-01 -7.01036632e-01 -2.57702261e-01 7.39702880e-01
-4.73598927e-01 3.37495297e-01 2.49273136e-01 1.87132373e-01
-8.59375298e-01 4.92836863e-01 1.40826452e+00 2.96144057e-02
-2.14374959e-01 9.65426862e-03 -9.45210516e-01 9.20393109e-01
9.76594985e-01 -3.46026301e-01 -5.01140177e-01 -2.47065425e-01
1.15967304e-01 -1.91365927e-01 3.68865371e-01 -4.06306446e-01
1.50274523e-02 -3.80129308e-01 -9.84791517e-02 -7.77182698e-01
1.76623538e-01 -4.67469186e-01 -2.57901195e-02 -2.83941358e-01
-2.07271695e-01 7.57935122e-02 3.49576205e-01 2.41346225e-01
-3.99450868e-01 -2.38194615e-01 7.98259258e-01 -2.45391145e-01
-1.20657539e+00 1.58375889e-01 4.28329483e-02 5.76337397e-01
1.30316412e+00 -1.87517464e-01 -3.12828809e-01 -2.20124632e-01
-7.33219683e-01 4.67850924e-01 2.52666622e-01 5.07204890e-01
5.89872241e-01 -1.20123637e+00 -6.88379645e-01 -2.20041573e-01
3.99894238e-01 7.37361491e-01 3.37546170e-01 1.04862475e+00
-4.22222614e-01 1.76441520e-01 3.07605863e-01 -6.50177479e-01
-1.37156081e+00 4.42247599e-01 5.65269887e-01 -4.47200507e-01
-5.67664504e-01 1.13453043e+00 8.50231528e-01 -5.66821396e-01
-1.33912846e-01 -5.07676780e-01 -2.78022170e-01 -1.86041579e-01
3.11142325e-01 1.19583309e-02 -2.45134175e-01 -1.11841869e+00
-3.44470710e-01 7.73363113e-01 -1.20474711e-01 7.69162923e-02
9.17658389e-01 -9.92879197e-02 -3.15051585e-01 6.55576468e-01
1.13600886e+00 1.53020963e-01 -1.03860772e+00 -2.05903128e-01
7.66220316e-03 -3.89962971e-01 -1.96361199e-01 -1.00040054e+00
-1.29249656e+00 7.02493429e-01 2.21747443e-01 -2.34073341e-01
1.50951779e+00 4.78378564e-01 7.07394898e-01 -1.15713306e-01
3.96777630e-01 -8.07649672e-01 2.30146060e-03 4.88009185e-01
1.20571232e+00 -1.05913436e+00 -7.16226771e-02 -1.08462858e+00
-8.66443276e-01 8.24996769e-01 1.11973238e+00 4.64061350e-02
3.70909750e-01 3.27085704e-01 8.03267658e-02 -3.01787674e-01
-3.95724118e-01 -3.73568058e-01 3.63735706e-01 3.67616415e-01
5.01697242e-01 -2.87934002e-02 -5.29179692e-01 1.07991159e+00
-1.37179032e-01 9.08206180e-02 2.26496860e-01 7.29974031e-01
1.03032701e-01 -1.10856867e+00 -3.97757232e-01 3.43975455e-01
-6.01745069e-01 -1.69994310e-01 -4.79338616e-01 9.94830310e-01
4.57718581e-01 8.13221276e-01 -2.37949993e-02 -1.79484993e-01
5.89754403e-01 -1.98977605e-01 2.61660665e-01 -6.79368854e-01
-7.20192194e-01 3.10787588e-01 1.21474572e-01 -5.81177771e-01
-6.71718657e-01 -6.59816027e-01 -1.79521859e+00 1.65663943e-01
-4.50485021e-01 8.62250570e-03 2.03791857e-01 1.02224195e+00
3.10154349e-01 7.73466706e-01 2.64429718e-01 -4.13362145e-01
-5.46996817e-02 -8.73837233e-01 -5.06522715e-01 5.69414079e-01
-1.47817910e-01 -7.98159480e-01 -5.82368486e-02 -1.00072779e-01] | [10.240772247314453, 1.170952320098877] |
1466f9c9-cc5b-431c-8604-be94ad4afee2 | action-unit-detection-with-region-adaptation | 1704.03067 | null | http://arxiv.org/abs/1704.03067v1 | http://arxiv.org/pdf/1704.03067v1.pdf | Action Unit Detection with Region Adaptation, Multi-labeling Learning and Optimal Temporal Fusing | Action Unit (AU) detection becomes essential for facial analysis. Many
proposed approaches face challenging problems in dealing with the alignments of
different face regions, in the effective fusion of temporal information, and in
training a model for multiple AU labels. To better address these problems, we
propose a deep learning framework for AU detection with region of interest
(ROI) adaptation, integrated multi-label learning, and optimal LSTM-based
temporal fusing. First, ROI cropping nets (ROI Nets) are designed to make sure
specifically interested regions of faces are learned independently; each
sub-region has a local convolutional neural network (CNN) - an ROI Net, whose
convolutional filters will only be trained for the corresponding region.
Second, multi-label learning is employed to integrate the outputs of those
individual ROI cropping nets, which learns the inter-relationships of various
AUs and acquires global features across sub-regions for AU detection. Finally,
the optimal selection of multiple LSTM layers to form the best LSTM Net is
carried out to best fuse temporal features, in order to make the AU prediction
the most accurate. The proposed approach is evaluated on two popular AU
detection datasets, BP4D and DISFA, outperforming the state of the art
significantly, with an average improvement of around 13% on BP4D and 25% on
DISFA, respectively. | ['Zhigang Zhu', 'Wei Li', 'Farnaz Abitahi'] | 2017-04-10 | action-unit-detection-with-region-adaptation-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Li_Action_Unit_Detection_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Li_Action_Unit_Detection_CVPR_2017_paper.pdf | cvpr-2017-7 | ['action-unit-detection'] | ['computer-vision'] | [ 2.86765486e-01 -1.98926240e-01 -2.44443938e-01 -1.46439925e-01
-7.67867148e-01 3.91860912e-03 2.46600911e-01 -2.62717128e-01
-1.50133207e-01 2.53200531e-01 -7.66078159e-02 3.85572255e-01
2.24957854e-01 -7.67979741e-01 -6.03967071e-01 -9.50794160e-01
-3.07836324e-01 -1.58742871e-02 3.79008204e-01 1.04238456e-02
-1.56764667e-02 7.23524988e-01 -1.53380895e+00 6.17102087e-01
5.59654772e-01 1.80164802e+00 5.23526734e-03 2.57233739e-01
1.65598918e-04 1.13730872e+00 -4.24863279e-01 -1.69049188e-01
2.47872099e-01 -4.04643506e-01 -5.07055342e-01 1.35012940e-01
6.57039642e-01 -7.12688327e-01 -2.04562485e-01 9.92693782e-01
4.91671413e-01 8.75144675e-02 8.32804382e-01 -1.02603662e+00
-2.65900612e-01 2.99108177e-01 -1.01890194e+00 2.21141383e-01
-7.69292936e-02 -1.39188766e-03 8.98521543e-01 -1.03882539e+00
4.00735319e-01 1.49464381e+00 4.09124702e-01 6.88349545e-01
-8.74920905e-01 -1.06607842e+00 2.29186580e-01 3.37054491e-01
-1.53095424e+00 -6.09204471e-01 7.31164753e-01 -4.42756206e-01
7.70670414e-01 -1.76496074e-01 4.48653549e-01 1.10839045e+00
4.41383868e-01 9.30443585e-01 7.45180249e-01 -2.40055770e-01
-2.83487320e-01 -1.66999131e-01 -1.52860150e-01 1.03238368e+00
-5.22763729e-01 6.92850426e-02 -4.29071546e-01 8.42654482e-02
9.98116255e-01 1.43859997e-01 7.70852193e-02 1.91440403e-01
-7.10645437e-01 6.99526489e-01 7.44943082e-01 5.47249377e-01
-5.32677650e-01 2.35650063e-01 5.52868724e-01 1.68910608e-01
6.67351604e-01 -1.36428773e-01 -2.88890183e-01 4.33121592e-01
-8.24042976e-01 -3.02850176e-03 -2.71692444e-02 5.54500401e-01
1.02304769e+00 6.67800456e-02 -7.25940824e-01 1.13697731e+00
4.04036403e-01 2.88599670e-01 3.21879536e-01 -7.46801913e-01
3.57628077e-01 8.50390315e-01 2.28025243e-02 -1.10085261e+00
-5.04791379e-01 -3.48064691e-01 -7.38587558e-01 3.64042699e-01
4.26349849e-01 -3.76555324e-01 -9.71090674e-01 1.84830141e+00
3.50095958e-01 6.91797256e-01 -2.16637049e-02 7.22539425e-01
8.81017268e-01 7.88758337e-01 2.80453444e-01 -4.70420361e-01
1.42777610e+00 -1.05082595e+00 -5.79305112e-01 -4.12412137e-01
7.77143598e-01 -7.02695191e-01 2.60454029e-01 1.31361097e-01
-9.26951528e-01 -9.12897885e-01 -9.21328008e-01 9.00209025e-02
-2.17854545e-01 8.05022299e-01 4.62115742e-02 3.49647067e-02
-9.70211029e-01 2.90702462e-01 -4.53897923e-01 -2.07369074e-01
8.52586627e-01 6.68558896e-01 -4.56342548e-01 -8.16155151e-02
-1.17547190e+00 9.14584041e-01 3.46963018e-01 5.27603805e-01
-1.47455657e+00 -3.50282788e-01 -9.03790772e-01 -2.28537861e-02
4.08603042e-01 -6.64798617e-02 1.08286512e+00 -1.32239628e+00
-1.16583860e+00 7.63517797e-01 -1.48207292e-01 -3.95648479e-01
2.63683259e-01 5.14880568e-03 -4.00138229e-01 2.81121761e-01
-3.02317087e-02 1.01279330e+00 1.25855041e+00 -1.08399343e+00
-9.91055191e-01 -4.82044250e-01 -5.54003567e-02 1.59367815e-01
-5.52080154e-01 5.49885631e-01 -3.48463863e-01 -5.20382226e-01
-2.59847995e-02 -6.67959511e-01 1.65241197e-01 8.99669677e-02
-1.29373129e-02 -7.95030475e-01 1.36755335e+00 -9.25594747e-01
1.24140358e+00 -2.25679541e+00 3.60712260e-01 -5.77262379e-02
2.67526746e-01 5.42144239e-01 -4.45904642e-01 -7.21879750e-02
-8.19294900e-02 -3.28249544e-01 1.11248329e-01 -5.85039735e-01
-4.92813110e-01 7.94565827e-02 1.21734915e-02 4.70072627e-01
7.72017777e-01 8.22030306e-01 -6.17427289e-01 -8.70821118e-01
3.43103111e-01 3.51666391e-01 -1.04387537e-01 4.35258657e-01
-3.19767803e-01 4.82844383e-01 -5.02492726e-01 9.98230219e-01
7.27179706e-01 1.22026615e-01 -1.69946522e-01 -5.08930564e-01
-1.15166403e-01 -3.49106431e-01 -1.01677024e+00 1.25979662e+00
-4.90853667e-01 4.30663288e-01 1.32053927e-01 -1.04804850e+00
1.23388124e+00 5.36838353e-01 7.87576079e-01 -7.62674034e-01
5.57395101e-01 2.31318057e-01 -3.02311536e-02 -6.07183099e-01
-2.26521417e-01 1.20108552e-01 1.85418367e-01 4.48724031e-01
3.26877058e-01 6.66675150e-01 3.30017656e-02 -3.08611721e-01
7.87637115e-01 1.89385042e-01 1.79138139e-01 1.36137784e-01
9.42052543e-01 -5.50565720e-01 8.29506457e-01 2.20955104e-01
-5.57260215e-01 3.44535202e-01 5.29131293e-01 -7.03420997e-01
-5.59267640e-01 -5.42059243e-01 -1.45563960e-01 1.35348833e+00
1.40176326e-01 4.99611199e-02 -8.30430329e-01 -1.15981758e+00
-7.53565580e-02 2.04487741e-01 -1.15058100e+00 -2.85901636e-01
-7.07518697e-01 -6.11277580e-01 4.92163986e-01 6.62731647e-01
7.98604429e-01 -1.56005681e+00 -5.85379958e-01 2.80268192e-01
-1.19016409e-01 -1.16957307e+00 -5.30367434e-01 1.14579745e-01
-5.99111140e-01 -1.10196626e+00 -8.53566051e-01 -1.05365658e+00
6.05739534e-01 2.51633823e-01 4.83732939e-01 5.29323854e-02
-3.40816587e-01 -1.78779125e-01 -4.17078882e-01 -2.38085195e-01
-2.04873502e-01 -2.26620749e-01 -2.56353077e-02 8.46698701e-01
4.14188862e-01 -2.77332008e-01 -5.70746779e-01 6.14815652e-01
-5.79891026e-01 -2.17090830e-01 8.50226164e-01 9.93591428e-01
3.92748356e-01 -1.75984368e-01 7.39201427e-01 -4.73788768e-01
3.91843654e-02 -3.65065902e-01 -5.03241658e-01 5.54694593e-01
3.98302963e-03 -2.88856506e-01 4.42886114e-01 -5.44562578e-01
-1.17982876e+00 1.63007423e-01 -7.77238756e-02 -9.24274147e-01
-1.36638731e-01 8.80517215e-02 -2.86062956e-02 -5.06029367e-01
4.97986168e-01 1.23650387e-01 1.81707993e-01 -1.73007265e-01
1.84018277e-02 7.23788917e-01 2.38796204e-01 -2.86825389e-01
2.62271971e-01 3.16213429e-01 -9.55212712e-02 -5.97906232e-01
-1.04751205e+00 -3.80984753e-01 -8.84918630e-01 -7.93592811e-01
1.29546416e+00 -1.00165570e+00 -6.12170100e-01 8.67320955e-01
-1.23152816e+00 -3.19556743e-01 3.06397945e-01 2.73171216e-01
-2.57319689e-01 6.25081360e-02 -7.90969908e-01 -9.27399278e-01
-5.22792995e-01 -1.37123191e+00 1.33621538e+00 6.45080924e-01
7.58402944e-02 -7.26543546e-01 -2.77917802e-01 2.52834737e-01
4.41627353e-02 3.37432683e-01 7.13968217e-01 -4.18031484e-01
-3.77920896e-01 -2.15542793e-01 -7.66146600e-01 5.64098120e-01
3.24522346e-01 2.84040719e-01 -1.24466419e+00 -3.56220335e-01
-1.12098023e-01 -6.02321804e-01 1.00569212e+00 5.64586818e-01
9.86441135e-01 -2.52347887e-01 -6.30233109e-01 2.56444246e-01
1.14346313e+00 5.20250678e-01 4.43344831e-01 -1.85129493e-01
8.28925371e-01 7.64469147e-01 1.11254823e+00 4.14753705e-01
-1.02911837e-01 9.48510945e-01 8.87651920e-01 -1.17229432e-01
-2.61359543e-01 2.46315166e-01 6.76919818e-01 1.62886351e-01
-1.17086105e-01 8.50643590e-03 -4.73197520e-01 5.37071407e-01
-1.95874655e+00 -8.49865615e-01 4.19337898e-01 2.01111984e+00
3.77527297e-01 3.18716168e-02 4.39168960e-01 -2.53314435e-01
8.09837043e-01 4.04784262e-01 -5.17996848e-01 -3.13174486e-01
-6.52645379e-02 1.02726027e-01 1.73060551e-01 2.01157272e-01
-1.51206779e+00 1.25115609e+00 5.08324909e+00 1.15080047e+00
-1.39245570e+00 3.50735009e-01 9.92671490e-01 6.16126433e-02
5.07176518e-01 -4.37549889e-01 -1.06957519e+00 3.15056115e-01
5.88918746e-01 4.15033460e-01 1.39029264e-01 7.09561050e-01
3.12090218e-01 -1.22710094e-01 -8.68317187e-01 8.34084630e-01
2.39939034e-01 -9.66211379e-01 -8.91856756e-03 -1.33419866e-02
8.57892215e-01 -4.33918983e-02 1.31543398e-01 1.75330743e-01
7.55396411e-02 -1.11339402e+00 5.93412220e-01 5.71884692e-01
6.50304735e-01 -9.89819229e-01 9.16419804e-01 2.06547827e-01
-1.67299008e+00 -6.32489204e-01 -5.63362420e-01 2.66010135e-01
-1.26301542e-01 8.95072892e-02 -7.04921722e-01 4.70704973e-01
8.34972560e-01 1.13500452e+00 -5.12854874e-01 7.28846431e-01
-1.10065982e-01 1.72664106e-01 -1.73312545e-01 1.57024264e-01
5.63846529e-01 -2.94731081e-01 2.48344898e-01 1.04888177e+00
4.54587966e-01 2.54604574e-02 2.82807350e-01 6.34481072e-01
-1.30388945e-01 2.85395533e-01 -5.48396051e-01 1.44606039e-01
2.07401618e-01 1.64447355e+00 -5.45463026e-01 -1.15909524e-01
-6.07798040e-01 8.85811865e-01 8.03159952e-01 1.96445212e-01
-9.42410767e-01 3.22267972e-02 7.20482051e-01 -1.38970643e-01
4.10985142e-01 6.27788305e-02 2.07613915e-01 -6.11187339e-01
-1.83030427e-01 -6.81239009e-01 7.89769173e-01 -5.18505812e-01
-1.12831640e+00 8.00500274e-01 -1.10001102e-01 -1.38260448e+00
-2.32579082e-01 -7.10592747e-01 -9.07792211e-01 7.35521495e-01
-1.31858683e+00 -1.71900022e+00 -1.82222515e-01 6.46804333e-01
7.71450400e-01 -1.99113443e-01 5.36658049e-01 5.46324193e-01
-1.10782325e+00 8.54885340e-01 -3.77203077e-01 4.29555774e-01
7.88174450e-01 -6.62402809e-01 -2.28422210e-01 9.49918628e-01
7.68281445e-02 1.05991326e-02 -6.21013902e-03 -6.22110903e-01
-8.40397120e-01 -1.55449677e+00 7.32263207e-01 -4.87426817e-02
4.30130690e-01 -2.04301372e-01 -7.45132685e-01 7.79374659e-01
7.07740523e-03 4.03688997e-01 2.21943855e-01 -2.43724287e-01
-1.84274375e-01 -5.19605517e-01 -1.08472979e+00 4.97607768e-01
8.66843820e-01 -3.47890288e-01 -3.83841656e-02 3.85659158e-01
4.41003770e-01 -3.39918584e-01 -9.32915747e-01 7.86416650e-01
6.03704393e-01 -9.69698846e-01 8.66556108e-01 -3.28891337e-01
3.45092267e-01 -3.28552842e-01 1.95090380e-02 -9.25306141e-01
-2.34576151e-01 -1.95903152e-01 -1.86289266e-01 1.34309781e+00
1.73779547e-01 -3.96756798e-01 6.16897106e-01 3.54520790e-02
-7.82316774e-02 -1.16008723e+00 -1.05153763e+00 -3.10168475e-01
-3.85222942e-01 -2.40977019e-01 2.37039924e-01 6.80086374e-01
-3.85001004e-01 1.44598052e-01 -7.84424543e-01 3.03080440e-01
5.02785325e-01 5.50020412e-02 4.22290295e-01 -1.06717873e+00
2.29173258e-01 -5.75446546e-01 -4.71008480e-01 -8.66597593e-01
3.64899069e-01 -5.54236412e-01 9.78157148e-02 -1.27167332e+00
9.00296122e-02 -3.30363572e-01 -4.70137239e-01 1.03447485e+00
-2.16776907e-01 6.53037429e-01 5.67275099e-02 6.38156161e-02
-5.93994737e-01 6.18104696e-01 1.42013586e+00 -3.45810562e-01
-3.60285550e-01 2.49446958e-01 -2.13775441e-01 5.96485138e-01
4.16959226e-01 -2.26954713e-01 -1.07454769e-01 -2.51524568e-01
-5.36507428e-01 1.86063543e-01 3.14053386e-01 -1.25381041e+00
2.33763948e-01 1.07261194e-02 7.48622239e-01 -7.26496518e-01
5.13059616e-01 -8.29881489e-01 -2.40232512e-01 5.69657385e-01
-8.02702084e-02 -2.93132812e-01 4.10893828e-01 3.67351055e-01
-3.89266610e-01 -1.53173044e-01 1.21001494e+00 -1.10942021e-01
-1.11657536e+00 8.75387907e-01 -2.61034548e-01 -4.85555351e-01
1.56086922e+00 -2.01987997e-01 -1.77445225e-02 -1.30094990e-01
-7.49529123e-01 4.82059509e-01 -1.34986147e-01 6.28524125e-01
7.23242223e-01 -1.44517350e+00 -8.09076369e-01 3.59654665e-01
1.73137516e-01 -1.89206246e-02 6.74906611e-01 9.98363376e-01
-2.11956054e-01 2.40692586e-01 -5.58575988e-01 -7.81385362e-01
-1.61380398e+00 4.06119496e-01 7.15771317e-01 -2.69716650e-01
-1.77243829e-01 1.28443563e+00 5.83688915e-01 -5.75664528e-02
2.24913552e-01 3.95334326e-02 -9.08696592e-01 6.35375082e-01
7.06020176e-01 2.19822153e-01 2.62651723e-02 -1.14268327e+00
-3.00842732e-01 9.85937178e-01 -3.57307583e-01 4.33038890e-01
1.05204391e+00 -5.75194731e-02 -3.90216172e-01 1.04843900e-01
1.41251624e+00 -3.93053412e-01 -1.61845207e+00 -5.55780828e-01
-2.89029449e-01 -1.36787385e-01 9.32655931e-02 -3.68112564e-01
-1.48418486e+00 9.85170841e-01 8.62024665e-01 -2.66139716e-01
1.45931327e+00 2.76648570e-02 8.22279871e-01 -1.73072889e-01
1.81500122e-01 -7.75936723e-01 5.23631573e-01 3.03832293e-01
8.06332946e-01 -1.22431350e+00 -3.36415887e-01 -2.81637520e-01
-6.27362728e-01 1.31355417e+00 1.29174256e+00 -1.49351954e-01
7.19353735e-01 6.57738233e-03 1.67859763e-01 -2.99109459e-01
-5.87614954e-01 -4.15847450e-01 5.34105420e-01 2.55885810e-01
6.41117916e-02 -2.73254931e-01 6.76702261e-02 3.42281789e-01
6.51778638e-01 -4.81703281e-02 -3.59146357e-01 5.99948227e-01
-6.14924133e-01 -1.01473439e+00 -3.17548007e-01 6.55403852e-01
-3.71639758e-01 2.88286120e-01 -3.88922334e-01 6.55996799e-01
7.33376086e-01 8.43632877e-01 3.20122212e-01 -7.07642496e-01
-3.48365046e-02 1.93947535e-02 4.62563068e-01 -6.25383496e-01
-5.55616498e-01 4.56198841e-01 -1.33835658e-01 -6.75846040e-01
-5.14827609e-01 -7.42960751e-01 -1.07018960e+00 1.39800891e-01
-4.55058396e-01 -2.26360068e-01 1.61929727e-02 1.30397522e+00
6.70982376e-02 5.57648182e-01 1.03016245e+00 -9.97683704e-01
-1.08946264e-01 -1.30374420e+00 -5.27706921e-01 -9.50480904e-03
3.20880711e-01 -1.09219170e+00 -1.42940775e-01 -2.67264426e-01] | [13.591064453125, 1.6044058799743652] |
715abdc4-a379-40cf-9c3d-d3cd66c95d23 | autoshape-an-autoencoder-shapelet-approach | 2208.04313 | null | https://arxiv.org/abs/2208.04313v2 | https://arxiv.org/pdf/2208.04313v2.pdf | AUTOSHAPE: An Autoencoder-Shapelet Approach for Time Series Clustering | Time series shapelets are discriminative subsequences that have been recently found effective for time series clustering (TSC). The shapelets are convenient for interpreting the clusters. Thus, the main challenge for TSC is to discover high-quality variable-length shapelets to discriminate different clusters. In this paper, we propose a novel autoencoder-shapelet approach (AUTOSHAPE), which is the first study to take the advantage of both autoencoder and shapelet for determining shapelets in an unsupervised manner. An autoencoder is specially designed to learn high-quality shapelets. More specifically, for guiding the latent representation learning, we employ the latest self-supervised loss to learn the unified embeddings for variable-length shapelet candidates (time series subsequences) of different variables, and propose the diversity loss to select the discriminating embeddings in the unified space. We introduce the reconstruction loss to recover shapelets in the original time series space for clustering. Finally, we adopt Davies Bouldin index (DBI) to inform AUTOSHAPE of the clustering performance during learning. We present extensive experiments on AUTOSHAPE. To evaluate the clustering performance on univariate time series (UTS), we compare AUTOSHAPE with 15 representative methods using UCR archive datasets. To study the performance of multivariate time series (MTS), we evaluate AUTOSHAPE on 30 UEA archive datasets with 5 competitive methods. The results validate that AUTOSHAPE is the best among all the methods compared. We interpret clusters with shapelets, and can obtain interesting intuitions about clusters in two UTS case studies and one MTS case study, respectively. | ['Grace Lai-Hung Wong', 'Daphne Ngar-yin Mah', 'Sourav S Bhowmick', 'Jianliang Xu', 'Byron Choi', 'Guozhong Li'] | 2022-08-06 | null | null | null | null | ['time-series-clustering'] | ['time-series'] | [-5.68078995e-01 -6.70002222e-01 9.55896974e-02 -1.65290296e-01
-6.56207860e-01 -7.39181399e-01 3.78776997e-01 1.28628597e-01
-2.45578066e-01 2.22119287e-01 9.67186540e-02 -1.08527921e-01
-6.63696408e-01 -5.46312392e-01 -4.52488601e-01 -1.16420949e+00
-6.27322793e-01 5.03106058e-01 -3.53042722e-01 9.73335002e-03
3.09270825e-02 7.06238568e-01 -1.62929320e+00 2.13027120e-01
7.81877577e-01 1.18793726e+00 8.27282295e-02 2.44723499e-01
5.38278967e-02 2.75593966e-01 -6.78788722e-01 6.62067309e-02
4.25042450e-01 -3.01247239e-01 -3.17263842e-01 2.29275241e-01
-3.20987821e-01 1.63477644e-01 -2.55424142e-01 7.11659014e-01
4.86977607e-01 4.05096948e-01 1.01997340e+00 -1.61164486e+00
-6.44506454e-01 5.45277357e-01 -5.18396795e-01 4.03540403e-01
-1.95811957e-01 -1.62372380e-01 1.16306937e+00 -8.63904715e-01
3.69143754e-01 1.10995054e+00 8.26636791e-01 -1.34571204e-02
-1.01586914e+00 -5.19350111e-01 -1.26736581e-01 6.26602530e-01
-1.56491601e+00 -6.53370395e-02 1.23748803e+00 -5.95988989e-01
7.27653325e-01 4.25033152e-01 5.84672630e-01 9.64925051e-01
5.10516018e-02 8.28063011e-01 9.04585719e-01 -2.89643854e-01
5.59599757e-01 -1.56871289e-01 2.10220888e-01 2.91475534e-01
-1.98080204e-02 2.98088025e-02 2.32513752e-02 -2.44009763e-01
5.79530060e-01 3.88152182e-01 -3.38647664e-01 -3.26770574e-01
-1.19155443e+00 9.97703195e-01 2.50553221e-01 7.71034062e-01
-5.82242131e-01 -2.82627255e-01 5.76261163e-01 6.34735703e-01
6.55349672e-01 3.71297747e-01 -5.79142332e-01 -2.20887467e-01
-7.26239145e-01 -2.06876814e-01 5.42526484e-01 5.18801749e-01
5.77018976e-01 2.59481400e-01 -1.29607186e-01 9.38351095e-01
6.00425974e-02 3.61477137e-01 9.79754388e-01 -7.98735023e-01
3.57586265e-01 6.68420255e-01 -3.20157975e-01 -1.22665048e+00
-4.39940184e-01 -6.91139400e-01 -1.08169866e+00 -1.38496175e-01
1.31981134e-01 -1.13707103e-01 -4.29311693e-01 1.32584798e+00
2.03544691e-01 5.22116661e-01 7.30971321e-02 8.27727914e-01
5.33235192e-01 9.74709392e-01 -4.15618807e-01 -5.56207776e-01
1.12130904e+00 -6.68145657e-01 -5.75296819e-01 6.16990268e-01
6.35842323e-01 -5.12283444e-01 8.00289094e-01 4.98889565e-01
-4.75192070e-01 -7.71206081e-01 -8.73974383e-01 3.74270350e-01
-4.71474230e-01 5.45608103e-01 3.96224231e-01 3.58985364e-01
-6.96742654e-01 9.45014179e-01 -1.14279437e+00 -2.51579732e-01
1.80638328e-01 1.34961709e-01 -4.02522773e-01 3.60740870e-01
-1.05145621e+00 2.51778066e-01 6.52780533e-01 -8.26504529e-02
-6.19989872e-01 -6.89643919e-01 -6.59639657e-01 3.02956514e-02
-5.23770526e-02 -2.06112131e-01 5.14921427e-01 -9.56565797e-01
-1.14832127e+00 5.49288094e-01 -7.64295384e-02 -5.90195894e-01
1.12473838e-01 1.49726272e-01 -8.29042912e-01 4.25116718e-01
1.47294641e-01 2.12450877e-01 1.06335163e+00 -1.12051117e+00
-4.09360498e-01 -4.31163192e-01 -4.85813141e-01 -1.05979115e-01
-9.05834079e-01 -1.03792980e-01 -2.89827108e-01 -1.03732347e+00
2.29039744e-01 -8.70372355e-01 1.35223893e-03 -4.35204536e-01
-1.66390017e-01 -7.31686056e-01 1.04682624e+00 -6.60505831e-01
1.60505164e+00 -2.63010025e+00 1.93896398e-01 4.09534693e-01
1.37728497e-01 -3.74005623e-02 -1.94145590e-01 6.06365561e-01
-5.90148091e-01 -1.04974005e-02 -3.10060829e-01 -4.29595470e-01
8.52811038e-02 5.48600554e-01 -5.32942951e-01 6.50854349e-01
2.23898530e-01 5.52355111e-01 -7.33339369e-01 -3.45715970e-01
2.71282226e-01 4.56465721e-01 -4.29214060e-01 3.18053424e-01
1.68238685e-01 4.27067935e-01 -4.04472649e-01 4.98790741e-01
6.20934010e-01 -2.61595339e-01 -1.04298055e-01 -4.77787286e-01
-3.15013468e-01 -2.22171441e-01 -1.07566929e+00 1.29999375e+00
-2.65327156e-01 5.79990745e-01 -4.39711988e-01 -1.32909000e+00
1.21352017e+00 4.31251228e-01 1.16216278e+00 -5.37100554e-01
2.18745694e-01 2.16957167e-01 -4.36034650e-02 -6.06610179e-01
1.37341470e-01 -9.16820839e-02 6.40984774e-02 4.13960814e-01
9.70132798e-02 3.90724778e-01 3.29634160e-01 -2.48892620e-01
9.91820335e-01 -2.35528633e-01 6.51482642e-02 -2.87585735e-01
6.23303592e-01 -1.74798608e-01 6.41341925e-01 2.34481856e-01
-1.80232093e-01 8.37358057e-01 4.07109201e-01 -7.05858648e-01
-1.26339662e+00 -1.08174670e+00 -2.43373960e-01 8.46583307e-01
-2.24164948e-01 -5.36638081e-01 -3.55595201e-01 -5.34808218e-01
-1.54643785e-02 6.55522227e-01 -7.09672213e-01 -1.60188690e-01
-7.02651858e-01 -8.30737174e-01 5.46608150e-01 6.79517448e-01
4.42223549e-02 -1.05886137e+00 -4.72339749e-01 1.97355852e-01
-3.26341540e-01 -7.82327950e-01 -3.78718287e-01 4.15385723e-01
-9.79827404e-01 -9.97430921e-01 -6.76180422e-01 -8.08305502e-01
5.98411322e-01 3.02245438e-01 8.30753684e-01 -1.87180281e-01
-9.26504284e-02 4.30533677e-01 -9.54193890e-01 -1.74821958e-01
-2.31320754e-01 4.48172987e-02 3.37354898e-01 4.35039490e-01
5.38554132e-01 -9.73920286e-01 -4.52961683e-01 3.73394728e-01
-1.01819932e+00 -4.75961983e-01 3.15252602e-01 9.04898882e-01
9.30550098e-01 8.50035071e-01 4.44695592e-01 -1.54544622e-01
7.74028957e-01 -8.12169552e-01 -3.72375488e-01 1.47285789e-01
-4.43453819e-01 5.43984734e-02 1.38403130e+00 -5.01507759e-01
-3.59317392e-01 -2.73777515e-01 -1.02270516e-02 -1.30988133e+00
-1.48449972e-01 7.40654469e-01 7.04780743e-02 3.48326564e-01
4.51428175e-01 6.43095732e-01 1.02034084e-01 -7.99877524e-01
7.07729757e-02 6.63881838e-01 2.88570315e-01 -6.90921485e-01
8.37670147e-01 6.84670806e-01 -2.73266226e-01 -9.06915128e-01
-3.92359853e-01 -7.20439672e-01 -8.00311625e-01 -1.41976118e-01
8.58806014e-01 -7.99415171e-01 -6.51502013e-01 5.13332933e-02
-8.95677447e-01 -1.43861575e-02 -5.01422405e-01 6.17645621e-01
-5.68388462e-01 5.47101200e-01 -4.52187926e-01 -7.96394050e-01
-2.64866590e-01 -8.91918540e-01 1.15119600e+00 -9.31872576e-02
-1.25365958e-01 -1.15667212e+00 3.57565284e-01 -1.00378662e-01
9.65146497e-02 5.28478682e-01 9.86791790e-01 -1.04386091e+00
2.53105052e-02 -9.70918462e-02 1.48504496e-01 5.48090875e-01
3.21952283e-01 1.13665961e-01 -8.19324315e-01 -6.73929095e-01
4.44555372e-01 4.22636755e-02 7.48125017e-01 4.68917787e-01
1.66068292e+00 -3.36848170e-01 -1.65854886e-01 8.36511910e-01
1.32094026e+00 6.77589595e-01 4.55414951e-01 4.39563066e-01
6.86666131e-01 4.94660407e-01 5.17501771e-01 7.93534875e-01
2.69950092e-01 4.17734027e-01 5.11515625e-02 9.19960588e-02
3.84337842e-01 -8.39507394e-03 6.48582041e-01 1.89579487e+00
-2.55670100e-01 3.96505976e-03 -9.06239152e-01 6.78522646e-01
-1.80944669e+00 -1.11867750e+00 -2.53826100e-02 2.03107214e+00
5.52368104e-01 -1.94963887e-01 3.89365792e-01 7.44237244e-01
7.83060968e-01 2.71247357e-01 -4.45435673e-01 -9.41936448e-02
-2.39829183e-01 2.56967187e-01 4.34197448e-02 -8.44843239e-02
-1.44478369e+00 3.43310446e-01 5.03926516e+00 1.15336263e+00
-1.28191352e+00 1.01358429e-01 4.96181011e-01 1.51570067e-01
-1.38947502e-01 -1.66890457e-01 -2.65969485e-01 7.66226172e-01
1.04235327e+00 -3.13485682e-01 4.96093750e-01 8.16670239e-01
3.73302132e-01 7.27748573e-01 -1.04802561e+00 1.34523463e+00
6.03633560e-03 -9.81231809e-01 1.34360380e-04 -1.32427350e-01
6.22905612e-01 -2.97607798e-02 2.19173729e-01 4.14243609e-01
-7.26336315e-02 -8.09688687e-01 4.08690065e-01 4.59570199e-01
4.19855237e-01 -9.36089814e-01 7.14337528e-01 1.60953224e-01
-1.56658626e+00 -3.19060177e-01 -5.44285119e-01 2.16066211e-01
-4.33803648e-02 7.57399917e-01 -6.72741473e-01 1.07641959e+00
7.72707164e-01 1.29407418e+00 -5.29213846e-01 1.20791268e+00
2.66289115e-01 8.62355649e-01 -3.84493440e-01 1.59273446e-01
3.15282762e-01 -6.28231645e-01 6.66341126e-01 1.09741330e+00
7.55911767e-01 1.06471978e-01 3.16430032e-01 8.01062465e-01
3.24226677e-01 1.99449390e-01 -3.84916723e-01 -2.70896077e-01
7.14394927e-01 1.18922234e+00 -8.19172621e-01 -1.88985094e-01
-1.98036745e-01 8.06378424e-01 -6.92662299e-02 3.87401849e-01
-7.93892264e-01 -3.76689702e-01 6.11092329e-01 -2.54178047e-01
6.69463277e-01 -4.71057534e-01 -5.92200272e-02 -1.28262770e+00
1.90132469e-01 -9.79289711e-01 7.40534782e-01 -6.24126434e-01
-1.89487803e+00 9.07097340e-01 -5.14918901e-02 -2.02632618e+00
-1.96114972e-01 -6.40924633e-01 -6.53459311e-01 3.51183742e-01
-1.15796459e+00 -7.56408334e-01 -2.29523867e-01 8.80893111e-01
5.25188625e-01 -4.55848575e-01 8.29725385e-01 5.48811197e-01
-7.88119376e-01 5.25381923e-01 7.98021913e-01 4.48013783e-01
5.20472050e-01 -1.20518112e+00 6.73092604e-02 6.12574100e-01
4.02058244e-01 6.63555861e-01 5.16295254e-01 -3.82289290e-01
-1.51921403e+00 -1.38282263e+00 3.21290195e-01 -2.92160809e-01
8.55762303e-01 -9.72533748e-02 -1.07116914e+00 4.40443277e-01
-1.58674025e-03 -1.86533928e-02 1.05668890e+00 5.32541648e-02
-3.21767062e-01 -3.56273293e-01 -7.26581991e-01 3.58148009e-01
7.13257968e-01 -5.82233369e-01 -7.50539243e-01 1.40089080e-01
6.74859762e-01 3.25063199e-01 -1.36082935e+00 4.71602261e-01
3.43794674e-01 -8.77954960e-01 9.77558076e-01 -5.55096865e-01
3.58999878e-01 -5.53801656e-01 -3.45900953e-01 -1.51294744e+00
-5.67629397e-01 -4.00973797e-01 -1.52745202e-01 1.27753842e+00
-2.27195770e-02 -6.97035015e-01 4.85134333e-01 -1.56658828e-01
-2.72926807e-01 -7.57656515e-01 -1.10651159e+00 -1.26591635e+00
9.34494957e-02 -5.56574047e-01 7.62594581e-01 1.54704571e+00
-9.14601907e-02 -1.64969694e-02 -2.20255017e-01 1.80380896e-01
6.05928123e-01 5.52770793e-01 5.27952313e-01 -1.41470444e+00
-3.38241786e-01 -5.21509767e-01 -5.78458548e-01 -7.36095786e-01
2.84796745e-01 -1.15006661e+00 -3.34294885e-01 -9.81465459e-01
-1.36173695e-01 -3.50349456e-01 -7.66906440e-01 3.35600853e-01
6.47142902e-02 -1.89998806e-01 1.95483863e-01 5.20307004e-01
-4.89175826e-01 1.08822453e+00 8.52005720e-01 -1.78299680e-01
-3.47266167e-01 -3.63021307e-02 -2.86777973e-01 4.80474085e-01
8.03346992e-01 -4.22928989e-01 -3.74273568e-01 -2.68937975e-01
-1.62645191e-01 2.31914967e-02 2.56717592e-01 -1.15383589e+00
2.58756936e-01 6.99486658e-02 3.78303766e-01 -1.08891344e+00
1.39504388e-01 -1.09120369e+00 2.94019431e-01 1.75703496e-01
-8.98597296e-03 4.50119257e-01 7.03201890e-02 5.12323320e-01
-5.98593712e-01 3.22100483e-02 4.20560360e-01 2.36003548e-01
-5.99385798e-01 5.07842362e-01 -2.11683795e-01 -1.76717341e-02
8.42785120e-01 -2.98341304e-01 1.56451911e-01 -2.32199237e-01
-8.19314837e-01 1.96387619e-01 2.42016852e-01 5.72205722e-01
6.50465906e-01 -1.85552633e+00 -7.49768019e-01 4.44116086e-01
1.44165039e-01 -2.84433007e-01 2.74603069e-01 9.91775751e-01
-3.72654945e-01 2.80631959e-01 -3.35692823e-01 -8.36226523e-01
-8.63236010e-01 8.79143715e-01 1.05078258e-01 -1.57193258e-01
-8.12261462e-01 4.02132064e-01 4.80147116e-02 -4.13410217e-01
1.87811762e-01 -3.43411803e-01 -5.03504574e-01 4.39722270e-01
4.40677166e-01 4.73401934e-01 9.03586596e-02 -5.93965054e-01
-3.59668821e-01 7.94173598e-01 2.60437340e-01 1.24345586e-01
1.82908559e+00 -2.74465680e-02 -3.66381437e-01 8.59928906e-01
1.77878952e+00 -1.41726330e-01 -1.09380472e+00 -6.20891191e-02
2.00004503e-01 -3.53256881e-01 -1.51797116e-01 -1.09437510e-01
-1.24727249e+00 9.53750193e-01 7.41294742e-01 7.48820364e-01
1.46180344e+00 -1.79333806e-01 8.62339854e-01 1.85478777e-01
3.05616707e-01 -1.05280864e+00 2.00117946e-01 5.11471152e-01
9.34391975e-01 -9.31813061e-01 -3.29222143e-01 6.85778484e-02
-5.79152346e-01 1.44748712e+00 1.39263824e-01 -4.23947722e-01
9.04200256e-01 -8.54431302e-04 7.17457430e-03 -2.54278898e-01
-7.50845373e-01 -1.58064961e-01 4.57231551e-01 3.37703764e-01
3.03227782e-01 3.02952170e-01 -2.94116825e-01 8.59173477e-01
-4.06297445e-01 -4.59755689e-01 5.71205989e-02 3.92729729e-01
-9.69454795e-02 -1.07415581e+00 -6.14077985e-01 4.26872671e-01
-2.46693984e-01 2.76931494e-01 -2.30301946e-01 6.32437050e-01
2.16811851e-01 9.36379135e-01 3.15678716e-01 -6.86936021e-01
2.45406061e-01 6.21503964e-02 4.92505059e-02 -1.65235281e-01
-5.36013424e-01 3.54181141e-01 -4.22061533e-01 -2.34797537e-01
-4.01368231e-01 -7.92790890e-01 -1.17957735e+00 -1.39275268e-01
-8.57960880e-02 5.18688440e-01 3.18479836e-01 7.62074411e-01
5.09769320e-01 5.11831880e-01 1.39252603e+00 -7.16173291e-01
-5.00058472e-01 -6.99556589e-01 -8.64642084e-01 8.48961651e-01
2.63032675e-01 -6.93551719e-01 -6.61115825e-01 6.28717840e-02] | [7.323685646057129, 3.2488014698028564] |
11750aca-58c2-4bb2-8278-b36181292b22 | learning-robust-video-synchronization-without | 1610.05985 | null | http://arxiv.org/abs/1610.05985v3 | http://arxiv.org/pdf/1610.05985v3.pdf | Learning Robust Video Synchronization without Annotations | Aligning video sequences is a fundamental yet still unsolved component for a
broad range of applications in computer graphics and vision. Most classical
image processing methods cannot be directly applied to related video problems
due to the high amount of underlying data and their limit to small changes in
appearance. We present a scalable and robust method for computing a non-linear
temporal video alignment. The approach autonomously manages its training data
for learning a meaningful representation in an iterative procedure each time
increasing its own knowledge. It leverages on the nature of the videos
themselves to remove the need for manually created labels. While previous
alignment methods similarly consider weather conditions, season and
illumination, our approach is able to align videos from data recorded months
apart. | ['Ido Freeman', 'Patrick Wieschollek', 'Hendrik P. A. Lensch'] | 2016-10-19 | null | null | null | null | ['video-synchronization', 'video-alignment'] | ['computer-vision', 'computer-vision'] | [ 4.95596558e-01 -4.86973882e-01 -2.41396993e-01 -3.13377321e-01
-3.88598591e-01 -8.01767886e-01 6.45753801e-01 -1.52831096e-02
-5.01859128e-01 5.14617682e-01 -8.26347023e-02 -9.14606005e-02
-1.51044717e-02 -3.00997376e-01 -6.70666873e-01 -6.32744908e-01
-3.04783404e-01 3.69447708e-01 5.38734674e-01 -2.49606580e-01
3.32031250e-01 5.07652879e-01 -1.72553849e+00 2.39592567e-02
3.38921815e-01 6.74546242e-01 2.67243385e-01 1.04682624e+00
1.26444682e-01 8.30511153e-01 -1.96646869e-01 -2.07786426e-01
6.01829290e-01 -6.14413559e-01 -5.88664114e-01 5.64001858e-01
1.03649020e+00 -2.78389812e-01 -2.39884555e-01 1.06353283e+00
1.10616229e-01 4.08953935e-01 5.40302038e-01 -1.46214914e+00
-2.18208712e-02 -7.41435140e-02 -6.84960902e-01 4.53962177e-01
6.51432216e-01 -8.40667561e-02 8.02009463e-01 -6.15227103e-01
8.90513182e-01 7.82324374e-01 8.47091138e-01 3.42545956e-01
-1.21235394e+00 -3.23178738e-01 3.25405955e-01 5.23717940e-01
-1.18100917e+00 -6.01079643e-01 7.11219490e-01 -7.19090700e-01
8.91107798e-01 5.09583354e-01 9.44829822e-01 8.38856161e-01
-3.92358974e-02 4.03625071e-01 1.01163733e+00 -6.09136283e-01
2.20345289e-01 -2.05090508e-01 -1.54780075e-01 6.64039552e-01
-4.35016267e-02 3.07863224e-02 -5.26512146e-01 -2.40266602e-02
1.01515710e+00 -1.01724036e-01 -2.33154088e-01 -6.96177304e-01
-1.47423744e+00 5.12524486e-01 -2.36962978e-02 2.91996658e-01
-2.97289461e-01 1.31201625e-01 3.00435871e-01 5.74790299e-01
2.57009268e-01 3.67074609e-01 -4.97517318e-01 -2.26876244e-01
-1.20315599e+00 3.48730385e-01 8.36631477e-01 8.85287225e-01
9.59134400e-01 4.92436141e-02 3.03176969e-01 4.40268248e-01
7.07176104e-02 3.12520564e-01 5.01668990e-01 -1.31714141e+00
1.10260129e-01 2.25924373e-01 2.04809293e-01 -1.27879810e+00
-2.28245810e-01 8.08420554e-02 -5.60764372e-01 5.64862967e-01
6.38335824e-01 1.05690822e-01 -7.18694866e-01 1.56883645e+00
6.20821476e-01 6.08527839e-01 -1.81247503e-01 7.75208950e-01
3.81956190e-01 4.28199321e-01 -1.20973237e-01 -4.79243428e-01
1.10047090e+00 -1.14311695e+00 -5.38168669e-01 -3.52281988e-01
2.99423218e-01 -1.18092799e+00 6.28348768e-01 4.68269944e-01
-8.72311056e-01 -5.56811929e-01 -9.90803659e-01 -6.71081021e-02
-2.10108444e-01 -9.77945700e-02 5.30165553e-01 3.59439671e-01
-1.31548584e+00 6.30417824e-01 -7.60219872e-01 -7.53427088e-01
7.54065141e-02 5.36317885e-01 -8.08550298e-01 1.37065183e-02
-5.83696723e-01 9.77082491e-01 4.19427365e-01 3.17274705e-02
-3.95802408e-01 -2.18604684e-01 -8.66184533e-01 -3.53251040e-01
6.27320707e-01 -6.59559250e-01 1.31161964e+00 -1.55539489e+00
-1.54943120e+00 1.04963362e+00 -3.81751955e-01 -3.66075605e-01
6.36126220e-01 -2.42945686e-01 -1.82180166e-01 2.51675159e-01
-1.39282197e-01 7.00419009e-01 1.34672725e+00 -1.04795206e+00
-7.96025157e-01 -1.50876030e-01 1.03262313e-01 4.50826615e-01
-2.73826092e-01 1.46812513e-01 -8.05476308e-01 -9.10292745e-01
1.05109043e-01 -1.23102665e+00 -4.84352261e-01 8.86483714e-02
2.31959596e-01 2.64860302e-01 9.37640131e-01 -7.22800970e-01
9.41939354e-01 -1.89143002e+00 6.47687376e-01 7.56974444e-02
1.05602130e-01 1.53576672e-01 -2.86949217e-01 3.58360708e-01
-1.38304263e-01 -2.32751817e-01 -1.40874803e-01 -2.04521716e-01
-2.80104578e-01 3.69707167e-01 -1.35133356e-01 7.60431528e-01
1.24412134e-01 6.29325211e-01 -1.02528119e+00 -6.81844175e-01
5.66507876e-01 3.78271788e-01 -5.64542651e-01 3.05868983e-01
-2.01340929e-01 5.84904730e-01 -1.39765963e-01 5.83772242e-01
4.31689233e-01 7.33347610e-02 3.26530248e-01 -2.77010262e-01
-2.84584403e-01 -2.36428052e-01 -1.41571641e+00 1.74187446e+00
-1.92821741e-01 9.23718750e-01 1.18911713e-01 -1.25529230e+00
6.08465731e-01 3.33099753e-01 9.59857106e-01 -5.05442202e-01
2.86362041e-03 8.60470086e-02 -3.84770148e-03 -7.39798903e-01
5.75735450e-01 2.08938755e-02 1.61155194e-01 4.19318259e-01
4.63181734e-02 -1.74640059e-01 5.30426800e-01 -2.54719183e-02
1.00535822e+00 5.61340094e-01 7.33640730e-01 -1.60875142e-01
7.51292765e-01 1.91290408e-01 6.26283228e-01 4.92802262e-01
-2.88340449e-01 7.82275438e-01 1.11988477e-01 -9.01889205e-01
-1.32799566e+00 -7.85148203e-01 1.18099093e-01 1.25507414e+00
2.60334939e-01 -3.63303334e-01 -7.18008578e-01 -5.41884363e-01
-2.78472155e-01 -1.99296087e-01 -6.97333276e-01 2.18293965e-01
-7.90314853e-01 -4.57282484e-01 8.06179345e-02 3.23377728e-01
1.15013860e-01 -1.01427948e+00 -7.86358893e-01 1.58034205e-01
-2.29845077e-01 -1.34578991e+00 -5.99139571e-01 1.81735406e-04
-7.66830981e-01 -1.24581540e+00 -4.49826837e-01 -7.89945066e-01
7.20956743e-01 5.17146587e-01 1.17429233e+00 1.37288362e-01
-4.56157148e-01 8.19178700e-01 -2.83982724e-01 -2.77394474e-01
-4.28530037e-01 -3.77669334e-02 2.25887403e-01 1.94162145e-01
5.18578768e-01 -6.19555831e-01 -2.72372931e-01 4.91460383e-01
-9.19213533e-01 9.30580422e-02 4.59250122e-01 6.64669633e-01
6.29481792e-01 -9.92828682e-02 8.89903456e-02 -7.05059111e-01
1.24586120e-01 -2.52136260e-01 -7.68652976e-01 3.32803875e-01
-1.40386388e-01 -1.04717210e-01 4.30084527e-01 -6.04232550e-01
-6.31950319e-01 7.35609889e-01 2.05771565e-01 -5.33065736e-01
-4.59404409e-01 1.48482502e-01 9.79478881e-02 -4.81815070e-01
3.53268176e-01 4.97489460e-02 1.85056597e-01 -4.77585495e-02
3.31429660e-01 1.28522530e-01 8.72470915e-01 -5.22396266e-01
1.09046113e+00 6.10820055e-01 1.79652706e-01 -1.04469776e+00
-6.63096786e-01 -7.97779143e-01 -1.27265131e+00 -6.18343651e-01
9.33234632e-01 -7.82010615e-01 -3.70874882e-01 4.74278778e-01
-1.03140318e+00 -2.62765974e-01 -1.82762474e-01 5.79850078e-01
-9.18249667e-01 7.68046021e-01 -2.65736520e-01 -4.59424287e-01
1.14767261e-01 -1.04014444e+00 9.68837023e-01 1.12789050e-01
-4.70511526e-01 -1.19385338e+00 4.66396481e-01 2.38559783e-01
2.47952119e-01 4.87725109e-01 4.24611300e-01 -2.83000678e-01
-6.07830107e-01 -2.73817629e-01 6.90631494e-02 2.68768430e-01
3.88443291e-01 4.18122143e-01 -8.03496659e-01 -2.96811670e-01
1.63675740e-01 -1.58789262e-01 4.48964924e-01 3.04446429e-01
7.15089500e-01 -2.88612396e-01 -2.95791086e-02 5.91975272e-01
1.52808607e+00 1.00799426e-01 4.55058366e-01 7.28423655e-01
8.91592145e-01 8.82683337e-01 6.54862165e-01 2.86429763e-01
1.82510585e-01 1.05733073e+00 4.43910331e-01 -1.78685129e-01
-1.90485623e-02 2.90701240e-01 4.57681835e-01 8.49474847e-01
-5.45571864e-01 6.11342974e-02 -8.19824755e-01 4.18178737e-01
-2.22539520e+00 -1.50896788e+00 -1.51026800e-01 2.40930653e+00
6.48487270e-01 -2.56507814e-01 3.54727209e-01 1.44184649e-01
7.20668852e-01 1.55454695e-01 -3.79230112e-01 -2.31269673e-01
-1.96143910e-01 2.64109541e-02 5.63329756e-01 5.02833009e-01
-1.40475416e+00 7.21782863e-01 7.17376614e+00 2.60572404e-01
-1.06160212e+00 -2.27446839e-01 3.22899908e-01 -2.20815927e-01
2.07150504e-01 1.27979487e-01 -4.01040494e-01 2.70052671e-01
6.55260503e-01 -2.10191920e-01 7.74570704e-01 6.11108780e-01
2.41974413e-01 -3.21341574e-01 -1.37970138e+00 1.19558382e+00
4.94159430e-01 -1.20970750e+00 -8.69135559e-02 1.38824536e-02
8.91089737e-01 -3.56536098e-02 -1.60549030e-01 -2.91038960e-01
6.76504001e-02 -8.94474506e-01 8.22967350e-01 5.81803977e-01
3.07975978e-01 -4.39392567e-01 4.37173218e-01 1.63027450e-01
-1.26080048e+00 -4.54476429e-03 -1.73126966e-01 -4.06131566e-01
2.78663158e-01 1.46057695e-01 -5.39743662e-01 4.73733991e-01
7.77223647e-01 9.65776563e-01 -6.98696733e-01 1.12608922e+00
1.04684338e-01 2.39054397e-01 -4.71418291e-01 4.50136483e-01
1.95419848e-01 -5.72622359e-01 5.29042184e-01 1.21560669e+00
3.98691028e-01 -1.21646598e-01 6.42853320e-01 -1.85733497e-01
2.76873380e-01 2.84729004e-01 -8.38665426e-01 1.60907775e-01
2.76434422e-01 1.27304447e+00 -1.01530135e+00 -3.54132682e-01
-7.86728203e-01 1.14829695e+00 1.80705950e-01 2.25337923e-01
-7.65637755e-01 1.11004308e-01 6.65958583e-01 1.16991445e-01
4.98396486e-01 -6.88977838e-01 5.84179573e-02 -1.32993197e+00
2.10133463e-01 -1.23308456e+00 4.88630921e-01 -4.96384412e-01
-1.06855881e+00 5.96527755e-01 1.55952752e-01 -1.66686952e+00
-6.03897214e-01 -5.92434227e-01 -4.82519120e-01 4.07665491e-01
-1.54387379e+00 -1.12153947e+00 -5.68183780e-01 8.59922767e-01
7.46570587e-01 -1.53594404e-01 8.19000244e-01 3.04282069e-01
-5.65080881e-01 9.57004502e-02 2.78268546e-01 -8.58900603e-03
1.05885780e+00 -1.19784057e+00 2.81272948e-01 1.24544632e+00
6.32768869e-01 4.14671183e-01 1.06253433e+00 -2.60112047e-01
-1.43165803e+00 -7.33280957e-01 7.57655382e-01 -4.74277645e-01
8.77059340e-01 -2.09184825e-01 -9.20087576e-01 8.00566971e-01
4.50816303e-01 1.91533983e-01 6.57595158e-01 -2.40509003e-01
-4.37346667e-01 -1.02767847e-01 -6.41049147e-01 6.16788566e-01
8.76708150e-01 -4.58111942e-01 -4.73679662e-01 3.86207104e-01
1.21518455e-01 -5.54073930e-01 -6.92845643e-01 2.36809522e-01
6.95854485e-01 -1.10052407e+00 1.06708848e+00 -5.66474855e-01
3.38106304e-01 -6.43397689e-01 -2.05288976e-01 -7.72492528e-01
-2.21199170e-01 -8.80946517e-01 -4.60755788e-02 1.07556450e+00
-1.56621665e-01 -2.58402050e-01 8.06172073e-01 6.06348991e-01
2.35843763e-01 -3.28712463e-01 -8.60974073e-01 -8.07871282e-01
-4.66000408e-01 -3.17441911e-01 1.43518820e-01 1.19812572e+00
-2.58526385e-01 1.21972337e-01 -8.20199072e-01 2.07871258e-01
8.17074418e-01 2.39687428e-01 1.22566497e+00 -1.28456891e+00
-3.94968271e-01 -5.15309513e-01 -9.61234808e-01 -7.89976060e-01
1.52064517e-01 -4.16914493e-01 2.27150694e-01 -1.09052110e+00
8.49812776e-02 -3.14362079e-01 -1.59068346e-01 3.49489748e-01
-9.38536003e-02 8.34549367e-01 1.86411425e-01 3.75770509e-01
-7.22542107e-01 -3.27863283e-02 6.87882304e-01 9.14810002e-02
-4.83770184e-02 -1.95963338e-01 -1.45015612e-01 9.28025782e-01
7.33363390e-01 -3.56208563e-01 -5.54988265e-01 -5.53062737e-01
1.04961723e-01 -2.04156801e-01 3.49339128e-01 -1.07619941e+00
3.50221187e-01 -5.95235705e-01 2.59935468e-01 -1.24253996e-01
2.62048006e-01 -1.02850318e+00 5.16807854e-01 2.91922510e-01
-5.26695251e-02 5.78743756e-01 1.80556834e-01 5.47488928e-01
-3.93888891e-01 -4.22489196e-01 6.86463773e-01 -2.76737899e-01
-1.30577493e+00 3.81040603e-01 -4.76339430e-01 -1.75740466e-01
1.49470401e+00 -6.55163825e-01 1.24417342e-01 -4.30474818e-01
-6.55478179e-01 1.74584631e-02 1.07092893e+00 6.33939326e-01
3.68895113e-01 -1.26993358e+00 -6.00298047e-01 1.34550154e-01
6.82302192e-03 -2.00398326e-01 2.10794792e-01 8.52300704e-01
-8.45370531e-01 6.24376945e-02 -6.89201474e-01 -8.40057135e-01
-1.89494860e+00 5.67960262e-01 1.07920408e-01 -3.43573876e-02
-6.73545539e-01 4.70606536e-01 -7.28037674e-03 3.44087556e-02
1.46677807e-01 5.77454492e-02 -2.81497240e-01 2.06946149e-01
5.51754713e-01 2.87028849e-01 2.05716230e-02 -1.18010938e+00
-2.86674410e-01 1.03357220e+00 -8.79279822e-02 -1.15391292e-01
1.44277573e+00 -3.81138265e-01 -2.21203789e-01 4.81843859e-01
1.08017206e+00 -5.96633274e-03 -1.48547721e+00 -1.63946524e-01
2.25707233e-01 -7.74061739e-01 -2.20516548e-01 -5.79177439e-02
-9.37339485e-01 4.40669149e-01 4.75746214e-01 1.89166054e-01
1.32407343e+00 -1.58469155e-01 4.75405723e-01 5.63735485e-01
2.27851659e-01 -1.09223294e+00 1.24105550e-01 4.28263783e-01
5.65304756e-01 -1.36862159e+00 4.54602242e-01 -3.79455060e-01
-5.60640156e-01 1.41193151e+00 4.75827664e-01 -1.41893372e-01
4.02350247e-01 4.54620361e-01 4.86427635e-01 5.82248643e-02
-6.18265450e-01 -2.23434880e-01 3.08229119e-01 8.80660355e-01
4.48659748e-01 -4.45645332e-01 -1.41918421e-01 -4.51500416e-01
-8.71718600e-02 -1.59233361e-01 7.14677691e-01 1.16213179e+00
-3.29311222e-01 -1.50669003e+00 -4.44413155e-01 1.49559096e-01
-3.76907945e-01 1.58135727e-01 -3.08675140e-01 7.78358996e-01
2.10401878e-01 5.46839833e-01 1.37463361e-01 -2.40335330e-01
4.66691069e-02 -1.03944056e-01 8.00808966e-01 -4.07602817e-01
-3.31635058e-01 2.40764543e-01 -1.61357149e-01 -7.84482777e-01
-1.44892275e+00 -1.26864135e+00 -8.34752142e-01 -1.62017927e-01
-1.46087274e-01 -1.51267067e-01 6.03925884e-01 1.03682375e+00
2.58004107e-02 1.56323850e-01 7.06350446e-01 -1.40940917e+00
2.31423918e-02 -3.69784504e-01 -1.97223887e-01 6.88922048e-01
4.50896621e-01 -5.85797608e-01 -1.54434398e-01 8.72460186e-01] | [8.61434555053711, -0.5172134041786194] |
6031aa14-859d-47b3-aea5-7057a6f110d3 | tadil-task-agnostic-domain-incremental | 2306.11955 | null | https://arxiv.org/abs/2306.11955v1 | https://arxiv.org/pdf/2306.11955v1.pdf | TADIL: Task-Agnostic Domain-Incremental Learning through Task-ID Inference using Transformer Nearest-Centroid Embeddings | Machine Learning (ML) models struggle with data that changes over time or across domains due to factors such as noise, occlusion, illumination, or frequency, unlike humans who can learn from such non independent and identically distributed data. Consequently, a Continual Learning (CL) approach is indispensable, particularly, Domain-Incremental Learning. In this paper, we propose a novel pipeline for identifying tasks in domain-incremental learning scenarios without supervision. The pipeline comprises four steps. First, we obtain base embeddings from the raw data using an existing transformer-based model. Second, we group the embedding densities based on their similarity to obtain the nearest points to each cluster centroid. Third, we train an incremental task classifier using only these few points. Finally, we leverage the lightweight computational requirements of the pipeline to devise an algorithm that decides in an online fashion when to learn a new task using the task classifier and a drift detector. We conduct experiments using the SODA10M real-world driving dataset and several CL strategies. We demonstrate that the performance of these CL strategies with our pipeline can match the ground-truth approach, both in classical experiments assuming task boundaries, and also in more realistic task-agnostic scenarios that require detecting new tasks on-the-fly | ['David Ellison', 'Ajay Dholakia', 'Jordi Guitart', 'Peini Liu', 'Gusseppe Bravo-Rocca'] | 2023-06-21 | null | null | null | null | ['incremental-learning'] | ['methodology'] | [ 2.22574651e-01 -3.24342281e-01 -1.68932959e-01 -5.05706191e-01
-8.64680886e-01 -7.20071256e-01 6.86189175e-01 4.82699037e-01
-6.97745979e-01 6.53549910e-01 -1.58974975e-01 -1.72480464e-01
-2.18986139e-01 -4.29375112e-01 -8.22983027e-01 -5.20982206e-01
-6.26278445e-02 8.50951493e-01 5.87036967e-01 1.38992086e-01
1.98653802e-01 3.07016283e-01 -1.84828734e+00 1.25846803e-01
1.00326228e+00 9.35401559e-01 3.54804307e-01 5.38350105e-01
-2.97277011e-02 6.00827157e-01 -5.24533331e-01 -1.56034440e-01
3.28241259e-01 -2.33051293e-02 -5.61244369e-01 1.27382800e-01
4.16864842e-01 -2.65477985e-01 -1.83218986e-01 7.69369900e-01
4.09941703e-01 1.61668181e-01 7.49369740e-01 -1.50683367e+00
-3.13556343e-01 2.14895919e-01 -4.88028616e-01 3.11465174e-01
1.13343641e-01 3.58361959e-01 5.72735310e-01 -1.09140766e+00
5.26771128e-01 1.00014615e+00 8.08064759e-01 5.43276548e-01
-1.40481222e+00 -6.16982639e-01 3.95375371e-01 4.34965342e-01
-1.11811721e+00 -5.34497440e-01 7.66189933e-01 -7.94111669e-01
7.82439172e-01 -3.87715638e-01 3.52337331e-01 1.34678221e+00
7.61482567e-02 8.13486934e-01 9.22626555e-01 -4.55070525e-01
7.34620631e-01 3.97821903e-01 1.97240308e-01 3.51350188e-01
3.38999808e-01 -5.08217625e-02 -9.22150016e-01 -2.24675626e-01
1.58292890e-01 1.41865268e-01 9.16384161e-03 -8.40611875e-01
-1.23943090e+00 4.81123149e-01 2.89110512e-01 8.78856257e-02
-4.53446716e-01 3.93734500e-02 4.68516111e-01 2.77379215e-01
6.27133191e-01 2.21392810e-01 -5.98610520e-01 -2.24234864e-01
-9.85506594e-01 3.76015514e-01 5.54657519e-01 9.42153037e-01
1.02535820e+00 -2.57480741e-01 -6.23078234e-02 6.46713793e-01
-5.82272559e-02 2.69492000e-01 6.80959225e-01 -7.39330232e-01
4.72468108e-01 5.69518447e-01 3.65842700e-01 -5.51797867e-01
-4.30476576e-01 -4.60857481e-01 -4.39410329e-01 1.32934093e-01
4.86447603e-01 -1.86916530e-01 -8.31456900e-01 1.70308530e+00
5.37742376e-01 4.13007677e-01 1.69086605e-02 6.32418334e-01
2.39206925e-02 2.57057011e-01 7.12477937e-02 2.49790847e-02
1.19096220e+00 -6.19882107e-01 -6.33276343e-01 -6.88288689e-01
8.10476184e-01 -3.58089834e-01 1.19282663e+00 4.00687367e-01
-5.91399908e-01 -7.72857368e-01 -1.14569318e+00 -5.23381531e-02
-4.46473390e-01 2.00046316e-01 2.69770145e-01 3.85755122e-01
-9.19499636e-01 4.52073127e-01 -1.02745855e+00 -4.80966240e-01
6.28389180e-01 1.48150697e-01 -3.21873873e-01 -3.04040700e-01
-9.74366069e-01 8.60367894e-01 5.44331193e-01 -1.25665456e-01
-1.03311741e+00 -9.78673398e-01 -9.37514305e-01 -2.57674038e-01
2.93552309e-01 -6.01542234e-01 1.33274150e+00 -7.90196061e-01
-1.09737766e+00 7.95446455e-01 -6.31862700e-01 -7.60969520e-01
7.08592296e-01 -3.35364252e-01 -2.23055914e-01 -1.93058044e-01
2.64172524e-01 6.43499732e-01 1.03156543e+00 -1.25666583e+00
-9.78859007e-01 -6.96216166e-01 -1.18143417e-01 1.59760669e-01
-5.53760469e-01 -3.86924446e-01 -7.37416744e-02 -1.48122087e-01
-4.32684878e-03 -9.65698481e-01 2.61468347e-02 2.17779130e-02
-7.46001899e-02 -5.17226219e-01 1.13336742e+00 -5.30708015e-01
9.97621596e-01 -2.52581644e+00 -5.84749691e-02 -8.05305019e-02
2.60575920e-01 8.56349617e-02 1.16443545e-01 2.32455775e-01
8.30024555e-02 -4.24683958e-01 -3.84939879e-01 -7.47565448e-01
6.66534156e-02 2.06216455e-01 -2.93812007e-01 4.31930602e-01
5.24733007e-01 6.81799531e-01 -1.19052875e+00 -2.98440486e-01
9.64624360e-02 1.32833779e-01 -4.42842305e-01 1.38741910e-01
-3.08321774e-01 4.29411680e-01 -1.13435090e-01 5.41520178e-01
6.27310157e-01 -5.03163300e-02 1.56126142e-01 2.14841533e-02
-1.11866504e-01 2.96956897e-01 -1.18183053e+00 1.92738843e+00
-5.53026915e-01 9.73347127e-01 -1.71140596e-01 -1.27905369e+00
9.67766047e-01 9.95891020e-02 3.63889456e-01 -5.26896000e-01
-2.55192339e-01 3.23686272e-01 -4.46318351e-02 -4.23388481e-01
4.24954474e-01 9.04456973e-02 -1.67613223e-01 4.53786492e-01
2.85697877e-01 9.29761901e-02 1.30234405e-01 7.45783821e-02
1.52249169e+00 1.56207114e-01 7.87429810e-02 -1.78281114e-01
2.77129710e-01 3.14931422e-01 6.79864764e-01 7.49146283e-01
-6.11007690e-01 4.72900450e-01 4.29978192e-01 -6.48486316e-01
-1.01725066e+00 -1.17077827e+00 -2.67432809e-01 1.22959745e+00
4.33764532e-02 -1.28024757e-01 -5.54721415e-01 -8.11316431e-01
3.75623912e-01 8.81777763e-01 -7.01508701e-01 -5.35901606e-01
-4.24710274e-01 -5.95224202e-01 1.86376661e-01 6.17232502e-01
4.33065921e-01 -7.59079456e-01 -8.81385624e-01 4.18528497e-01
-7.85247311e-02 -1.27572453e+00 -2.47217402e-01 6.42239749e-01
-8.83945942e-01 -1.08114505e+00 -4.54504132e-01 -8.31937909e-01
6.22126818e-01 3.45736742e-01 9.98571992e-01 -5.06363690e-01
-2.74184018e-01 5.03158629e-01 -6.54986799e-02 -7.31307387e-01
-1.51671305e-01 2.44133547e-01 3.47501159e-01 2.27160782e-01
9.17148709e-01 -4.23093289e-01 -5.83892703e-01 2.13613346e-01
-6.91987038e-01 -2.96141416e-01 5.28341413e-01 6.95442259e-01
6.04929447e-01 2.47320831e-01 8.86750996e-01 -8.00208569e-01
7.35787749e-01 -6.40091121e-01 -6.70559168e-01 1.65934145e-01
-6.05308950e-01 1.15852557e-01 6.04394734e-01 -7.18288064e-01
-1.02455652e+00 4.86815631e-01 3.69932652e-01 -7.17562318e-01
-3.55247676e-01 2.11897671e-01 -1.09197669e-01 4.00287628e-01
1.08768213e+00 3.44281584e-01 -7.24413618e-02 -4.94492054e-01
2.88700938e-01 9.92228270e-01 7.52781153e-01 -5.95388055e-01
9.38956559e-01 6.40878379e-01 -3.43573838e-01 -5.81294596e-01
-9.28078234e-01 -7.04999983e-01 -1.17201173e+00 -1.63864344e-01
6.85140848e-01 -1.21039939e+00 -3.81356657e-01 5.38191557e-01
-1.21461642e+00 -5.24747670e-01 -4.08672065e-01 5.74003696e-01
-5.48062384e-01 -4.89355847e-02 -1.23938888e-01 -8.32552016e-01
9.96551961e-02 -9.14328814e-01 1.16171563e+00 1.90551087e-01
-3.05054903e-01 -1.06143427e+00 1.45286158e-01 2.10243136e-01
3.39465708e-01 1.50074258e-01 7.38835335e-01 -9.00485694e-01
-3.81689012e-01 -3.08896929e-01 -6.31589890e-02 4.61526841e-01
2.39743382e-01 -3.70597959e-01 -1.27006567e+00 -4.55226630e-01
-3.85120302e-03 -3.80930841e-01 9.39845443e-01 1.92908332e-01
1.20044971e+00 1.59276485e-01 -5.37831783e-01 4.64318722e-01
1.25669289e+00 7.73469731e-02 1.97859392e-01 5.53128242e-01
7.04594731e-01 7.21357405e-01 7.04292238e-01 4.70420033e-01
6.75360441e-01 4.35035348e-01 3.11057806e-01 1.68953925e-01
-8.49451348e-02 -4.14231628e-01 4.97607052e-01 4.05582041e-01
4.41006780e-01 1.67450204e-01 -1.19429398e+00 1.00947773e+00
-1.96768725e+00 -7.48311222e-01 9.77136865e-02 2.45653057e+00
8.88364434e-01 4.15168732e-01 1.24635488e-01 2.54491389e-01
6.18103743e-01 -1.74536616e-01 -1.15125024e+00 -1.32174045e-01
2.39619538e-01 6.64622188e-02 5.31693578e-01 2.31012166e-01
-1.26318014e+00 8.81677151e-01 5.74139023e+00 3.28177363e-01
-1.09043062e+00 2.50209689e-01 3.24093819e-01 -2.35917658e-01
-8.99186283e-02 -1.04162119e-01 -9.40735519e-01 5.39173722e-01
1.06202281e+00 -4.33228970e-01 2.91054189e-01 1.01742589e+00
1.77804027e-02 -2.51277536e-01 -1.57005417e+00 1.02866483e+00
1.26968399e-01 -8.69626701e-01 -3.61639827e-01 -6.00455180e-02
6.62169576e-01 2.13871703e-01 1.48079365e-01 5.57723165e-01
4.99476522e-01 -5.01865923e-01 7.22268939e-01 4.09856021e-01
6.19962931e-01 -6.07471645e-01 5.79135716e-01 8.12351048e-01
-9.99595642e-01 -3.26144904e-01 -3.53564739e-01 -8.19872469e-02
-1.02040179e-01 9.04854000e-01 -1.18049514e+00 4.03844982e-01
7.77269781e-01 8.04010332e-01 -7.16121554e-01 1.06279743e+00
-8.19120929e-02 5.33105254e-01 -2.69726485e-01 3.50906730e-01
-1.58232972e-02 2.17243046e-01 3.64786237e-01 1.01627660e+00
3.80634516e-01 -5.04646122e-01 1.73163697e-01 7.57659912e-01
-6.28893450e-02 -3.15897167e-01 -9.10174012e-01 2.94198900e-01
8.78046751e-01 1.10063696e+00 -4.19471473e-01 -3.90390754e-01
-4.03817236e-01 1.13538241e+00 5.11062086e-01 4.62225884e-01
-7.07273543e-01 -3.50583255e-01 8.33348870e-01 3.72005105e-01
2.14438766e-01 -3.81920129e-01 -2.36475632e-01 -1.08573556e+00
4.08257127e-01 -6.30674660e-01 3.56397390e-01 -5.30195296e-01
-1.45703566e+00 3.57259929e-01 1.39308706e-01 -1.30871439e+00
-3.70087087e-01 -4.87242281e-01 -4.72255200e-01 8.79509747e-01
-1.89043105e+00 -7.99117088e-01 -6.00451291e-01 6.28399432e-01
6.97393298e-01 -1.41929701e-01 6.30631983e-01 2.94249922e-01
-6.33437514e-01 4.63215768e-01 2.68173933e-01 6.67389203e-03
1.12088692e+00 -1.34370971e+00 4.93625283e-01 7.52674222e-01
8.01356807e-02 4.52066779e-01 5.47276318e-01 -4.84121829e-01
-1.16829205e+00 -1.38263798e+00 8.97296250e-01 -8.00765991e-01
6.34848595e-01 -9.15052474e-01 -1.13508785e+00 8.60842466e-01
-1.26161993e-01 3.10937822e-01 5.85719585e-01 2.86078602e-01
-4.02775913e-01 -4.37510282e-01 -1.00433195e+00 1.12156868e-01
1.06168509e+00 -6.14069343e-01 -7.59331942e-01 4.72967714e-01
5.99126339e-01 -3.35984319e-01 -5.15243471e-01 2.87105769e-01
3.73355001e-01 -7.82160163e-01 7.29560196e-01 -5.55915833e-01
1.17623195e-01 -4.31683987e-01 -5.63934036e-02 -1.57889962e+00
-3.01225007e-01 -3.52160841e-01 -1.24346025e-01 1.12066090e+00
3.01734328e-01 -7.82626987e-01 6.45707011e-01 5.23190558e-01
-1.48664087e-01 -4.01289403e-01 -9.96639669e-01 -1.07410622e+00
1.71318557e-02 -6.15349889e-01 5.93384206e-01 9.97192740e-01
-2.41369754e-01 2.32053012e-01 1.09165180e-02 2.85168558e-01
8.34177434e-01 -1.59675404e-01 1.03948355e+00 -1.67021298e+00
-7.87639618e-02 6.40718713e-02 -5.29436767e-01 -9.63981330e-01
3.01081419e-01 -7.53940880e-01 3.35580409e-01 -1.44770980e+00
5.97781055e-02 -7.22392440e-01 -5.04093587e-01 4.95921701e-01
-2.65025318e-01 -3.02788615e-01 -6.97326213e-02 5.13584971e-01
-8.88716459e-01 5.12368381e-01 5.66156149e-01 -1.09688826e-01
-3.10238779e-01 -7.34055489e-02 -5.13651729e-01 6.68879569e-01
7.88978040e-01 -7.46684790e-01 -5.90968430e-01 -6.45373404e-01
5.36116250e-02 -5.28282523e-01 4.85907614e-01 -1.38938916e+00
5.28790653e-01 1.53902277e-01 5.93994439e-01 -6.28033936e-01
1.35591060e-01 -8.80581319e-01 -1.63403347e-01 3.56224835e-01
-2.18391478e-01 5.54861920e-03 3.45712274e-01 8.04161847e-01
-1.43301457e-01 -1.28781766e-01 6.77475095e-01 1.35392055e-01
-9.61985171e-01 1.42915308e-01 -1.22426294e-01 2.49309063e-01
1.37567973e+00 -1.71306908e-01 -3.17403316e-01 1.40546891e-03
-5.74032843e-01 4.38577384e-01 4.93844926e-01 6.71951592e-01
4.29173470e-01 -1.24215579e+00 -6.32577658e-01 4.32174474e-01
6.21864378e-01 4.08502162e-01 7.24953189e-02 6.52059376e-01
-1.19971680e-02 2.44180962e-01 -2.64919233e-02 -1.06424832e+00
-8.55845571e-01 5.51859260e-01 2.21416369e-01 -1.72597840e-01
-4.52771127e-01 6.36608779e-01 1.39859959e-01 -5.58421075e-01
3.66730601e-01 -2.83958554e-01 8.14468600e-03 3.57540131e-01
5.21139920e-01 1.69703156e-01 4.28529412e-01 -1.97616324e-01
-5.69987059e-01 1.57492191e-01 -2.44904339e-01 -7.73281232e-02
1.22986376e+00 -7.08124638e-02 4.07459468e-01 1.00559616e+00
1.11235905e+00 -4.38460439e-01 -1.75604200e+00 -6.45470679e-01
4.65408146e-01 -5.05309820e-01 -6.51735300e-03 -5.15219331e-01
-5.29938102e-01 9.58374083e-01 8.98075759e-01 -3.00837145e-03
9.73549128e-01 7.63967261e-02 6.23657823e-01 2.73093015e-01
4.31617796e-01 -1.43592501e+00 2.35092774e-01 4.11397785e-01
4.53166515e-01 -1.45831621e+00 -2.74470031e-01 2.29728445e-01
-6.00390434e-01 8.51885140e-01 6.75547242e-01 6.66906312e-02
6.71202600e-01 3.11708122e-01 6.83178753e-02 4.42311689e-02
-1.10904515e+00 -2.95106739e-01 4.02276702e-02 9.72241700e-01
8.96340907e-02 -2.36501023e-02 1.85701922e-01 4.39321816e-01
-9.69200656e-02 3.08608264e-01 3.49997759e-01 1.10873568e+00
-4.97398585e-01 -9.82329428e-01 -2.82696664e-01 4.77419734e-01
6.30163401e-03 2.62266308e-01 -2.38677233e-01 5.72680831e-01
3.74855459e-01 8.74121964e-01 4.97740775e-01 -4.03092355e-01
6.35054052e-01 5.43561518e-01 2.93063223e-01 -7.35768855e-01
-8.74427855e-02 -4.25888866e-01 -2.93899775e-01 -4.17410046e-01
-1.19446941e-01 -1.09305274e+00 -1.10144913e+00 6.86270511e-03
-4.33735363e-02 -2.48191506e-01 9.02074695e-01 8.65230143e-01
6.94483042e-01 5.77415645e-01 7.31658459e-01 -7.78478563e-01
-7.54918277e-01 -9.80477691e-01 -3.34959388e-01 6.55223668e-01
4.53992367e-01 -9.59301651e-01 -4.97411042e-01 3.00694585e-01] | [9.877602577209473, 2.773505210876465] |
85871c5f-f59a-401f-ba06-dcfa565cdeff | sea-sentence-encoder-assembly-for-video | 2011.12091 | null | https://arxiv.org/abs/2011.12091v1 | https://arxiv.org/pdf/2011.12091v1.pdf | SEA: Sentence Encoder Assembly for Video Retrieval by Textual Queries | Retrieving unlabeled videos by textual queries, known as Ad-hoc Video Search (AVS), is a core theme in multimedia data management and retrieval. The success of AVS counts on cross-modal representation learning that encodes both query sentences and videos into common spaces for semantic similarity computation. Inspired by the initial success of previously few works in combining multiple sentence encoders, this paper takes a step forward by developing a new and general method for effectively exploiting diverse sentence encoders. The novelty of the proposed method, which we term Sentence Encoder Assembly (SEA), is two-fold. First, different from prior art that use only a single common space, SEA supports text-video matching in multiple encoder-specific common spaces. Such a property prevents the matching from being dominated by a specific encoder that produces an encoding vector much longer than other encoders. Second, in order to explore complementarities among the individual common spaces, we propose multi-space multi-loss learning. As extensive experiments on four benchmarks (MSR-VTT, TRECVID AVS 2016-2019, TGIF and MSVD) show, SEA surpasses the state-of-the-art. In addition, SEA is extremely ease to implement. All this makes SEA an appealing solution for AVS and promising for continuously advancing the task by harvesting new sentence encoders. | ['Gang Yang', 'Jiaqi Ji', 'Chaoxi Xu', 'Fangming Zhou', 'Xirong Li'] | 2020-11-24 | null | null | null | null | ['ad-hoc-video-search'] | ['computer-vision'] | [ 2.78673768e-01 -6.11010969e-01 -2.92769551e-01 -3.02501649e-01
-1.15949214e+00 -4.09241229e-01 5.86172581e-01 1.91545472e-01
-4.66889322e-01 5.84680319e-01 5.61307192e-01 9.19419341e-03
-7.01969340e-02 -3.74920964e-01 -9.47060287e-01 -4.56386417e-01
7.80640021e-02 4.93745133e-02 3.95803988e-01 -2.19459057e-01
2.61527300e-01 -2.32540499e-02 -1.71491754e+00 6.76217258e-01
5.44341028e-01 1.30828404e+00 5.56780696e-01 4.38117057e-01
-3.21915299e-01 9.91638660e-01 -2.36585259e-01 -7.06660330e-01
1.08448379e-01 -5.53410590e-01 -1.00102997e+00 1.58375744e-02
5.50254583e-01 -2.84930438e-01 -5.18097818e-01 8.80831540e-01
6.64126456e-01 1.36927590e-01 6.42504215e-01 -1.30976236e+00
-8.57947588e-01 4.16659981e-01 -4.94219542e-01 1.08443432e-01
7.24715531e-01 -9.88156274e-02 1.36907852e+00 -9.02275443e-01
7.54084229e-01 1.27810585e+00 6.93286717e-01 4.18471277e-01
-9.04476523e-01 -4.43588495e-01 6.39714822e-02 6.47346377e-01
-1.65811658e+00 -4.91470158e-01 6.23102546e-01 -3.83441478e-01
1.02914250e+00 4.51285988e-01 3.93459737e-01 1.30220532e+00
6.49196506e-02 1.37685382e+00 6.17528141e-01 -2.86247551e-01
8.76767784e-02 8.79095122e-02 -1.53356567e-01 5.53579926e-01
-9.28149968e-02 -3.22155118e-01 -7.61319399e-01 -4.51942394e-03
3.50113153e-01 1.64147168e-01 -4.29276526e-01 -5.47598958e-01
-1.30626059e+00 8.37133527e-01 2.41198301e-01 4.31857318e-01
-1.75844178e-01 1.54925108e-01 1.01378226e+00 5.76867044e-01
3.01603884e-01 3.09214771e-01 -2.97933966e-01 -3.29860181e-01
-1.15029156e+00 2.27074504e-01 5.86951196e-01 1.15562463e+00
6.62857234e-01 -1.69671759e-01 -4.65615004e-01 8.62690985e-01
2.43602768e-01 3.37425202e-01 8.87072802e-01 -7.28336871e-01
6.03414237e-01 4.93747354e-01 -8.41027275e-02 -1.01019061e+00
4.04322855e-02 -3.97864372e-01 -7.39565253e-01 -4.41618234e-01
6.47463743e-03 3.52971911e-01 -6.39533162e-01 1.61626649e+00
-8.21748301e-02 4.72006261e-01 2.51912832e-01 9.21589136e-01
1.02478611e+00 7.23466516e-01 -3.58251780e-02 -2.25230739e-01
1.31493390e+00 -1.21803713e+00 -6.44455850e-01 -7.70680904e-02
7.39582479e-01 -8.47470522e-01 1.21089089e+00 1.56319737e-01
-1.15583277e+00 -7.70595551e-01 -1.10318363e+00 -1.95600063e-01
-3.90039414e-01 -9.51852798e-02 4.50745821e-01 2.88562536e-01
-1.13222635e+00 5.21969318e-01 -5.53874850e-01 -5.68570733e-01
3.80717546e-01 6.72871619e-02 -3.78742933e-01 -2.20092729e-01
-1.50998378e+00 7.13239849e-01 3.95122886e-01 -3.15360099e-01
-7.29541361e-01 -5.90364456e-01 -9.67234969e-01 1.41688913e-01
6.67934120e-01 -8.43607128e-01 1.16164184e+00 -1.09759212e+00
-1.20222652e+00 8.15750003e-01 -3.38335067e-01 -6.89090192e-01
3.87305111e-01 -3.46328408e-01 -5.13233423e-01 4.09121096e-01
3.15314263e-01 6.87206864e-01 1.11526930e+00 -9.18042302e-01
-5.51096499e-01 -2.12884560e-01 1.07232137e-02 3.12843859e-01
-7.82264590e-01 3.00282896e-01 -8.60112011e-01 -7.27558851e-01
-1.37462318e-01 -6.58487260e-01 -4.84039262e-03 4.34786193e-02
-2.23626077e-01 -4.50808972e-01 7.05978572e-01 -4.61162984e-01
1.50239444e+00 -2.28210330e+00 4.25358415e-01 -2.15178400e-01
5.89507595e-02 3.67878556e-01 -3.26041013e-01 8.25565815e-01
4.15597446e-02 6.37249947e-02 -1.47739515e-01 -5.23245692e-01
9.47167352e-02 1.38341978e-01 -4.63474572e-01 2.55947411e-01
2.50430167e-01 1.12555873e+00 -1.07075787e+00 -8.05889845e-01
1.09448515e-01 4.75334853e-01 -4.91947502e-01 1.52984113e-01
-5.30182011e-02 9.50722247e-02 -4.70658928e-01 8.01661551e-01
4.02649999e-01 -4.29887176e-01 -1.91632673e-01 -4.27079439e-01
-1.98260192e-02 2.27780432e-01 -1.12449229e+00 2.21835780e+00
-3.32937211e-01 6.30720317e-01 -2.67829210e-01 -1.28637183e+00
7.14273334e-01 4.19421136e-01 7.79717863e-01 -9.45857346e-01
-2.34268252e-02 4.32806551e-01 -7.50466168e-01 -8.64478409e-01
7.63574064e-01 1.33658528e-01 -2.07525581e-01 2.34047234e-01
4.39418793e-01 1.86440945e-01 3.79826248e-01 3.94799262e-01
1.03636050e+00 2.08278671e-01 2.55967468e-01 7.46616051e-02
8.44295502e-01 -2.18933672e-01 4.49752629e-01 7.51691163e-01
-2.06547379e-01 7.42312908e-01 3.11755866e-01 -3.49308908e-01
-9.79902208e-01 -1.05849540e+00 -3.60359363e-02 1.10021567e+00
3.83707970e-01 -9.43811238e-01 -4.28345054e-01 -7.87131369e-01
7.51119554e-02 3.22905481e-01 -3.16911817e-01 -2.97016323e-01
-3.82148176e-01 -1.75970063e-01 5.65933824e-01 5.82257926e-01
4.82999802e-01 -9.14798141e-01 -5.62325597e-01 1.24240525e-01
-6.05493903e-01 -1.43118143e+00 -7.98887372e-01 -7.43038207e-02
-6.44682646e-01 -9.15373147e-01 -1.10361528e+00 -8.46614778e-01
1.15975112e-01 6.10712528e-01 1.23517358e+00 -1.42946810e-01
-3.54558975e-01 7.32708514e-01 -8.72181833e-01 -1.21606901e-01
-2.31479988e-01 9.13926885e-02 8.73474311e-03 2.18491688e-01
4.77348536e-01 -3.35106254e-01 -5.55521965e-01 2.13388339e-01
-1.04826605e+00 -3.57757695e-02 6.19714558e-01 8.67301285e-01
7.37090588e-01 -1.75197780e-01 6.30554974e-01 -4.63911504e-01
5.27522445e-01 -8.04644346e-01 -1.91983551e-01 6.45030975e-01
-4.68668908e-01 8.92327577e-02 5.31836569e-01 -2.38094315e-01
-6.23607576e-01 3.75823565e-02 -1.61412701e-01 -7.93717742e-01
7.79281482e-02 7.08800912e-01 1.19399048e-01 1.98846057e-01
2.66114950e-01 6.41176999e-01 1.03824899e-01 -5.71511686e-01
1.72295421e-01 8.26827586e-01 4.65779543e-01 -2.37638086e-01
6.40196741e-01 3.10978949e-01 -1.46654710e-01 -8.67215395e-01
-5.75985432e-01 -9.20739353e-01 -3.77150416e-01 -2.33767375e-01
1.01688504e+00 -1.16075420e+00 -5.46813726e-01 2.97931433e-01
-1.11661482e+00 2.67128408e-01 -1.84162259e-01 5.13005972e-01
-7.05984592e-01 7.52268255e-01 -3.55854154e-01 -6.72871888e-01
-3.22606295e-01 -1.39349997e+00 1.34503162e+00 -7.76434243e-02
-1.57045856e-01 -8.42608511e-01 5.21174958e-03 4.66773838e-01
4.60117668e-01 -1.17332317e-01 4.66095030e-01 -6.98955476e-01
-6.51518226e-01 -1.53214648e-01 -2.35693738e-01 4.93410558e-01
-1.52116697e-02 -1.72751606e-01 -7.18428493e-01 -5.45019567e-01
4.56358567e-02 -7.68882453e-01 1.20425129e+00 9.90083441e-02
1.37773538e+00 -1.65332973e-01 -1.71782374e-01 7.01239884e-01
1.61949599e+00 1.22353500e-02 6.29681051e-01 5.29556692e-01
5.36685169e-01 2.84996599e-01 6.93336487e-01 4.80363101e-01
5.20326734e-01 1.02903378e+00 4.46167380e-01 6.11225441e-02
-1.34342358e-01 -3.16945612e-01 6.76212966e-01 1.11028039e+00
1.85854793e-01 -3.42667639e-01 -5.23018062e-01 6.00774944e-01
-2.01244211e+00 -1.12429166e+00 1.43671229e-01 2.11767387e+00
7.59628236e-01 -1.26304433e-01 1.72748134e-01 1.19926870e-01
4.64721233e-01 4.14829850e-01 -3.79669935e-01 -2.74031073e-01
-3.28750700e-01 6.17001280e-02 4.85797465e-01 1.77670464e-01
-1.26620221e+00 7.54339874e-01 5.32601261e+00 1.12510252e+00
-1.05189693e+00 1.30466223e-01 3.31299335e-01 -2.08270296e-01
-3.92175347e-01 -1.30628988e-01 -7.84465194e-01 7.44988620e-01
8.49426150e-01 -2.53729254e-01 2.53470600e-01 7.43874907e-01
-7.52557963e-02 1.30424619e-01 -1.30716085e+00 1.37694061e+00
4.85337973e-01 -1.63689804e+00 3.43959540e-01 -4.75151867e-01
4.63997811e-01 2.17789084e-01 9.88578871e-02 4.63750631e-01
-3.15830439e-01 -8.01840723e-01 8.89448762e-01 5.00769496e-01
8.95722687e-01 -4.97953862e-01 8.65773022e-01 2.46913135e-02
-1.45047045e+00 -7.14132041e-02 -3.80599588e-01 2.93811709e-01
3.27701807e-01 8.87467414e-02 -4.55871522e-01 9.95105088e-01
8.63068581e-01 1.18831944e+00 -7.38342762e-01 1.04237390e+00
3.50385427e-01 2.76371717e-01 4.83375154e-02 -6.96141794e-02
5.07303417e-01 9.96633619e-02 6.27527237e-01 1.47927880e+00
3.60807151e-01 -4.11007077e-01 2.66368330e-01 5.61468244e-01
-1.71953052e-01 3.53367388e-01 -6.80194795e-01 -1.24203384e-01
4.21287924e-01 7.55474150e-01 -3.04417193e-01 -3.16207081e-01
-8.77751112e-01 1.37307799e+00 1.56027496e-01 3.72457713e-01
-9.99131322e-01 -3.61205369e-01 5.49219191e-01 -7.16047287e-02
6.55617297e-01 5.53044397e-03 1.68191448e-01 -1.37393200e+00
5.12944400e-01 -9.16714609e-01 4.23994839e-01 -7.33891964e-01
-1.36754608e+00 6.94817185e-01 4.54735234e-02 -1.78303266e+00
-2.92193234e-01 -2.53027052e-01 -2.38703519e-01 5.76011777e-01
-1.93285418e+00 -1.16438913e+00 -1.64877221e-01 8.62940967e-01
1.00235331e+00 -5.24324954e-01 7.29120255e-01 7.28007138e-01
-3.37837189e-01 9.66970801e-01 2.59120852e-01 -5.35281301e-02
9.76996899e-01 -8.95930409e-01 1.96461022e-01 6.95439160e-01
4.53017741e-01 3.16502512e-01 4.86064136e-01 -3.58189672e-01
-1.73856759e+00 -1.01495969e+00 1.08266222e+00 -2.18573257e-01
8.19955230e-01 -2.40269840e-01 -8.34775805e-01 3.24962348e-01
3.88871193e-01 -2.31095124e-02 7.39479840e-01 -2.53327489e-01
-4.97317761e-01 -3.00491840e-01 -6.97626889e-01 4.76151079e-01
1.08616447e+00 -8.27421486e-01 -6.07050836e-01 3.55883032e-01
1.00449693e+00 -1.38204470e-01 -7.77544320e-01 5.00123322e-01
5.06147087e-01 -1.05388927e+00 1.23718834e+00 -7.35664785e-01
7.23735631e-01 -1.13971472e-01 -6.72843397e-01 -8.96819949e-01
-9.31789726e-02 -6.84534132e-01 -3.62679362e-01 1.28570962e+00
7.65886605e-02 -2.75085419e-01 5.51546156e-01 1.03571683e-01
-3.42927724e-01 -1.04780614e+00 -1.08807397e+00 -1.08667767e+00
-1.84973404e-01 -4.58004355e-01 5.45140088e-01 9.03053880e-01
-1.11339360e-01 3.94225210e-01 -6.04494512e-01 -2.27922544e-01
5.28398931e-01 7.85841644e-02 4.84311491e-01 -9.81940985e-01
-3.35027874e-01 -5.28019607e-01 -7.64653802e-01 -1.44847202e+00
2.31737405e-01 -1.09829032e+00 -1.50800481e-01 -1.44571579e+00
4.47237432e-01 -9.70176831e-02 -5.02486706e-01 2.32464403e-01
-1.38262495e-01 3.17987770e-01 3.61137152e-01 3.88592511e-01
-1.24577940e+00 7.97428668e-01 1.09836805e+00 -3.25207382e-01
9.86331180e-02 -1.73293382e-01 -6.38849616e-01 4.85861987e-01
4.21822578e-01 -2.06430376e-01 -6.24590814e-01 -8.20356488e-01
2.68205523e-01 1.16588995e-01 4.21051025e-01 -1.03245258e+00
4.42490846e-01 9.28134844e-02 -5.29586673e-02 -7.59048462e-01
4.31659460e-01 -8.90486002e-01 -8.69623274e-02 3.30362886e-01
-5.20525336e-01 1.37364596e-01 -6.24090284e-02 7.67362535e-01
-8.11913788e-01 -3.41011792e-01 4.57751095e-01 -1.78487107e-01
-1.16837096e+00 3.85878056e-01 -1.88428193e-01 3.63815486e-01
9.99136984e-01 -4.61095512e-01 -4.11694316e-04 -4.81753170e-01
-3.69080305e-01 4.77325171e-01 3.21373314e-01 8.49216104e-01
9.51574504e-01 -1.49641144e+00 -9.01421189e-01 9.68745723e-02
4.44893479e-01 -2.97874957e-01 3.89471591e-01 8.34201992e-01
-1.85739651e-01 6.82542622e-01 -6.10227883e-02 -8.36169779e-01
-1.42647207e+00 7.46340513e-01 -7.94243142e-02 -1.70634225e-01
-5.34624279e-01 1.08995605e+00 6.11291826e-02 1.64047852e-01
5.14178157e-01 3.34008713e-03 -2.08439216e-01 2.25349560e-01
6.08093202e-01 1.73941791e-01 1.22718796e-01 -8.15725923e-01
-3.92839462e-01 5.80058813e-01 -2.94349670e-01 1.78809166e-01
1.43799400e+00 -4.25227970e-01 5.95651683e-04 5.71467817e-01
1.86780548e+00 -3.22490901e-01 -9.36939001e-01 -5.26802301e-01
1.87756896e-01 -6.39305472e-01 -5.96451759e-02 -3.72583568e-01
-1.03605258e+00 6.86021388e-01 4.29706901e-01 1.47918165e-01
1.29075801e+00 2.78994590e-01 1.21830082e+00 4.37146634e-01
2.85069555e-01 -1.13012266e+00 4.83506203e-01 4.96318430e-01
9.51891780e-01 -1.43024671e+00 -1.01760611e-01 -1.44366473e-01
-8.69852245e-01 9.62014019e-01 4.32226956e-01 5.86359762e-02
4.38991725e-01 -1.28501251e-01 -3.59246135e-01 -8.73872787e-02
-8.27171564e-01 -2.68306047e-01 4.75171894e-01 2.92954832e-01
4.68739599e-01 -3.57005745e-01 -3.53514016e-01 6.27219439e-01
1.90591022e-01 2.05506861e-01 2.42812991e-01 1.01757765e+00
-3.38552684e-01 -1.24509120e+00 -4.78814216e-03 4.98199791e-01
-5.16483545e-01 -2.27401167e-01 -2.30132133e-01 4.60276276e-01
-2.88370699e-02 7.40104675e-01 3.00589614e-02 -6.35829628e-01
2.63445675e-01 1.71542287e-01 3.66196662e-01 -4.41040695e-01
-4.58618969e-01 -4.20812592e-02 -1.39989793e-01 -8.48841965e-01
-6.40245914e-01 -7.53585875e-01 -8.16888630e-01 2.05913745e-02
-2.47444376e-01 1.38823986e-01 5.60119569e-01 8.77539217e-01
4.92717087e-01 4.62919861e-01 8.13040972e-01 -5.08347511e-01
-7.98427701e-01 -6.65010452e-01 -3.59980643e-01 7.24886596e-01
5.02025783e-01 -7.34846711e-01 -2.84196109e-01 6.17068075e-02] | [10.331851959228516, 0.9297490119934082] |
ac093da3-fa09-4771-8ac0-28bf575ce71a | synergetic-reconstruction-from-2d-pose-and-3d | 2001.05613 | null | https://arxiv.org/abs/2001.05613v2 | https://arxiv.org/pdf/2001.05613v2.pdf | Synergetic Reconstruction from 2D Pose and 3D Motion for Wide-Space Multi-Person Video Motion Capture in the Wild | Although many studies have investigated markerless motion capture, the technology has not been applied to real sports or concerts. In this paper, we propose a markerless motion capture method with spatiotemporal accuracy and smoothness from multiple cameras in wide-space and multi-person environments. The proposed method predicts each person's 3D pose and determines the bounding box of multi-camera images small enough. This prediction and spatiotemporal filtering based on human skeletal model enables 3D reconstruction of the person and demonstrates high-accuracy. The accurate 3D reconstruction is then used to predict the bounding box of each camera image in the next frame. This is feedback from the 3D motion to 2D pose, and provides a synergetic effect on the overall performance of video motion capture. We evaluated the proposed method using various datasets and a real sports field. The experimental results demonstrate that the mean per joint position error (MPJPE) is 31.5 mm and the percentage of correct parts (PCP) is 99.5% for five people dynamically moving while satisfying the range of motion (RoM). Video demonstration, datasets, and additional materials are posted on our project page. | ['Yoshihiko Nakamura', 'Takuya Ohashi', 'Yosuke Ikegami'] | 2020-01-16 | null | null | null | null | ['markerless-motion-capture'] | ['computer-vision'] | [-9.65972468e-02 -4.25564468e-01 -2.82824665e-01 1.09726235e-01
-6.93037987e-01 -3.29265356e-01 2.64841497e-01 -5.00354528e-01
-5.26887059e-01 5.90924084e-01 2.60295331e-01 4.24287021e-01
1.98773459e-01 -4.16670233e-01 -6.39757812e-01 -4.33106035e-01
-3.04178391e-02 3.48093331e-01 9.43125963e-01 3.95873524e-02
3.18825930e-01 5.25175869e-01 -1.24002695e+00 -4.16666362e-03
4.71124798e-01 7.68744946e-01 3.22491705e-01 1.02571428e+00
5.63373387e-01 4.02710676e-01 -4.63505477e-01 -2.64643848e-01
5.90031862e-01 -1.97043777e-01 -1.05899081e-01 3.22496355e-01
8.74792933e-01 -6.45393610e-01 -5.67258775e-01 7.28967726e-01
8.14833522e-01 3.48862022e-01 1.63406774e-01 -1.07362616e+00
1.25131570e-02 -1.19770512e-01 -9.03140783e-01 3.85864228e-01
8.93236279e-01 2.36701339e-01 2.03802943e-01 -9.17765439e-01
9.07238543e-01 1.00990403e+00 9.31464255e-01 6.77184403e-01
-5.66133678e-01 -6.50917888e-01 -9.14445743e-02 1.10324770e-01
-1.59201753e+00 -4.24120605e-01 7.75151968e-01 -5.53009272e-01
5.75573623e-01 4.28416729e-01 1.11189830e+00 6.65578425e-01
4.95785415e-01 6.91415489e-01 5.76811373e-01 -4.66992795e-01
-6.31826743e-02 -2.76766837e-01 -7.79123008e-02 7.63388693e-01
5.27407706e-01 2.18154624e-01 -7.96248555e-01 3.21351103e-02
1.42793310e+00 2.20182434e-01 -3.82020503e-01 -4.78063285e-01
-1.42861176e+00 1.83034658e-01 -2.66323108e-02 -2.25456040e-02
-3.46232265e-01 4.65707898e-01 5.28269522e-02 -4.38701987e-01
1.69678256e-01 -2.99701184e-01 1.76320225e-01 -6.85877681e-01
-1.10796309e+00 5.52668393e-01 3.69101614e-01 1.22333241e+00
1.92478448e-01 6.00359514e-02 4.06235904e-02 5.79038322e-01
3.06385189e-01 8.83834004e-01 3.87442708e-01 -1.34283864e+00
6.72919989e-01 6.16852164e-01 5.40655971e-01 -1.22861207e+00
-3.61965567e-01 -7.89319575e-02 -5.57103813e-01 3.39312404e-01
6.28620207e-01 -3.70465159e-01 -6.45549774e-01 1.20432353e+00
7.58360088e-01 4.90582138e-01 -4.62758571e-01 1.62805271e+00
5.38228273e-01 5.47788143e-01 -1.69190481e-01 -1.33443594e-01
1.30098677e+00 -8.32573056e-01 -7.05506444e-01 -5.70944965e-01
1.61691949e-01 -9.10869241e-01 7.70307243e-01 4.48639661e-01
-1.37627912e+00 -8.11183095e-01 -9.83920157e-01 1.07643373e-01
5.64534545e-01 3.84038329e-01 1.93850264e-01 6.38843000e-01
-6.57175183e-01 4.39972967e-01 -1.12954640e+00 -4.13004726e-01
-1.09063126e-01 3.99787068e-01 -4.59217906e-01 1.71141190e-04
-1.00300741e+00 7.78660655e-01 -8.23351294e-02 2.49811485e-01
-5.93399167e-01 -5.92296243e-01 -6.44298315e-01 -3.28799367e-01
3.47486556e-01 -9.88311231e-01 1.05413556e+00 -4.85136896e-01
-1.51302135e+00 8.19821835e-01 -7.84752890e-02 -2.60086566e-01
9.54781592e-01 -8.41211498e-01 -2.44942367e-01 5.31575203e-01
2.00155690e-01 5.63754797e-01 5.77202082e-01 -9.88723040e-01
-8.58930290e-01 -5.57500422e-01 -2.87955880e-01 6.28613830e-01
8.86742175e-02 -5.52046932e-02 -1.14362359e+00 -6.60672009e-01
4.20444250e-01 -1.32144773e+00 -2.49424830e-01 4.35486794e-01
-1.83721587e-01 1.97476566e-01 6.73102975e-01 -7.40664899e-01
1.31634045e+00 -2.00857449e+00 -1.03442170e-01 7.24912435e-03
-1.65558189e-01 1.97672129e-01 5.04564285e-01 2.88738273e-02
4.87787813e-01 -3.59998167e-01 2.73272604e-01 -2.44444534e-01
-3.47358465e-01 -7.29045272e-02 1.32500589e-01 8.04772556e-01
-5.58858871e-01 6.55935287e-01 -6.31304920e-01 -8.29978764e-01
5.63800931e-01 5.78973591e-01 -4.70913380e-01 2.37027124e-01
4.48006868e-01 4.35193807e-01 -6.04849339e-01 7.65766382e-01
6.49991512e-01 8.58957022e-02 -6.09882548e-03 -2.11731881e-01
-2.68383682e-01 -4.01146442e-01 -1.81317425e+00 1.95058990e+00
8.44474062e-02 5.60432076e-01 7.32531920e-02 -1.97169662e-01
9.57032740e-01 3.36861640e-01 8.55178595e-01 -3.64208758e-01
2.60886159e-02 6.86336011e-02 -3.79882872e-01 -7.80884266e-01
8.53030145e-01 -6.76383749e-02 -7.68314376e-02 1.89083517e-01
-5.21750391e-01 1.92906752e-01 8.13996419e-02 -2.43637078e-02
6.71944201e-01 6.57929122e-01 3.15144032e-01 9.32070687e-02
2.96990603e-01 3.32913190e-01 8.88092697e-01 5.38700521e-01
-5.90600789e-01 1.05345738e+00 -2.85499871e-01 -4.47713941e-01
-1.20050550e+00 -9.29261684e-01 2.58938104e-01 4.38697726e-01
8.04656386e-01 -2.97206044e-01 -6.34682000e-01 9.19954255e-02
7.08475709e-02 3.04321665e-02 -2.25041956e-01 4.71457690e-02
-1.18066418e+00 -2.63577491e-01 4.37621802e-01 7.84800470e-01
6.19133890e-01 -4.62597609e-01 -1.24656272e+00 1.56407133e-01
-4.19802845e-01 -1.23932600e+00 -9.47418630e-01 -8.60623717e-01
-1.07242990e+00 -1.16656709e+00 -1.10902834e+00 -6.72623515e-01
3.63288373e-01 5.55143237e-01 6.10840380e-01 -2.58306805e-02
-1.89895868e-01 5.50154328e-01 -1.08141162e-01 -4.86551262e-02
8.92712697e-02 -4.03140187e-01 4.06874478e-01 -5.39089069e-02
1.15617514e-01 -1.87286571e-01 -1.14570022e+00 6.54443920e-01
-2.57848948e-01 3.02977204e-01 2.95291841e-01 4.02558684e-01
7.49267578e-01 -3.62481654e-01 -1.56974941e-01 -2.13733360e-01
1.68573126e-01 -2.59781815e-02 -5.81262469e-01 1.84815992e-02
-2.10492030e-01 -6.71011806e-01 5.71649559e-02 -7.20663905e-01
-1.05270100e+00 5.13430953e-01 9.69589874e-02 -7.01563597e-01
-2.03436747e-01 -1.47525638e-01 -3.18050454e-03 -3.06737833e-02
5.16185880e-01 2.39408746e-01 1.04628354e-01 -3.34382296e-01
4.29189689e-02 4.98082995e-01 1.12997031e+00 -3.31250131e-01
4.72019345e-01 8.04413438e-01 8.45332071e-02 -7.73701131e-01
-3.62393767e-01 -7.52753139e-01 -9.33741808e-01 -9.85210121e-01
9.81652021e-01 -1.31845605e+00 -1.00976956e+00 4.18766886e-01
-9.44990277e-01 -7.11515322e-02 4.97434987e-03 1.23908448e+00
-6.13316596e-01 6.92279637e-01 -6.09486222e-01 -1.01389050e+00
-4.44738060e-01 -9.28776205e-01 1.22359061e+00 5.13106704e-01
-4.06464010e-01 -6.98177576e-01 1.95359036e-01 5.94454587e-01
-1.42724887e-01 5.43760419e-01 -1.56601071e-01 2.49339938e-01
-8.17475975e-01 -6.48612440e-01 4.15058196e-01 -3.92466515e-01
-2.76279032e-01 -5.55832125e-02 -4.85867828e-01 -1.92483261e-01
-1.46201000e-01 2.82667935e-01 3.33186269e-01 1.02091622e+00
3.76082659e-01 5.88944256e-02 -6.20838642e-01 6.09315813e-01
1.35991919e+00 1.96373850e-01 7.47764707e-01 6.96243465e-01
8.04421306e-01 2.74075896e-01 1.21896553e+00 6.70533299e-01
3.08750570e-01 1.33193731e+00 3.34617794e-01 2.91641712e-01
-4.02595252e-01 -4.52250361e-01 6.00319624e-01 7.15902209e-01
-6.03941441e-01 -2.97833867e-02 -9.21755314e-01 4.67423886e-01
-2.00013614e+00 -1.29080808e+00 -6.43036127e-01 2.39127183e+00
3.76922309e-01 1.45326778e-01 4.48418468e-01 -2.33569555e-02
1.14686680e+00 -1.49508551e-01 -4.52498347e-01 -3.44186462e-02
6.44803792e-02 -5.18409908e-01 6.68276787e-01 5.56100845e-01
-9.25179780e-01 6.98079646e-01 6.30486393e+00 4.36428964e-01
-8.38758469e-01 -4.38584089e-02 -5.19201159e-02 -4.76136565e-01
2.73362398e-01 -1.80774499e-02 -1.07211423e+00 6.27206624e-01
6.23808384e-01 -8.73813704e-02 -2.37608120e-01 7.77090728e-01
6.46899045e-01 -5.51922381e-01 -6.96278691e-01 1.35511518e+00
1.69659480e-01 -1.42178786e+00 -2.24254772e-01 7.93018639e-02
6.36603832e-01 -4.48449016e-01 -2.16722369e-01 -1.61312088e-01
-1.76313013e-01 -3.89439970e-01 1.10480738e+00 9.94819522e-01
7.16924608e-01 -6.72278404e-01 4.08156037e-01 7.61141717e-01
-1.40414178e+00 1.59253050e-02 -3.68171871e-01 -2.54655629e-01
7.56311595e-01 1.96312115e-01 -6.16870940e-01 3.06064039e-01
6.13841355e-01 5.99654794e-01 -3.43277037e-01 1.35467160e+00
6.22858815e-02 3.34763736e-01 -3.50702643e-01 7.85938576e-02
-2.05966771e-01 -2.63476491e-01 9.42692220e-01 1.10266638e+00
6.06813371e-01 5.12955189e-01 3.52125466e-01 3.18478495e-01
4.43894058e-01 -6.86654225e-02 -3.31643850e-01 6.43279254e-01
5.58407128e-01 1.02221107e+00 -5.48451543e-01 -3.53503615e-01
-2.50476182e-01 1.14429736e+00 -2.20264897e-01 -1.64388064e-02
-1.06249964e+00 -9.37891975e-02 5.11624813e-01 6.13237321e-01
1.35264903e-01 -5.50503075e-01 -2.11193874e-01 -1.22882128e+00
3.39301199e-01 -3.03917438e-01 4.21724260e-01 -1.05187309e+00
-6.90295339e-01 1.34618372e-01 9.89147648e-02 -1.92775393e+00
-3.62366736e-01 -2.15530530e-01 -4.16380972e-01 6.57551169e-01
-6.97738469e-01 -1.05246818e+00 -5.73006392e-01 6.57060742e-01
7.14585364e-01 9.67397727e-03 4.11324173e-01 5.52014410e-01
-5.63268542e-01 3.98900807e-01 -1.57713503e-01 2.40798295e-01
8.50835741e-01 -7.02868283e-01 2.28241667e-01 1.15432525e+00
-7.14253858e-02 4.48041320e-01 7.62383819e-01 -1.05290020e+00
-1.73416209e+00 -8.03917468e-01 7.13629127e-01 -6.75712347e-01
1.27229452e-01 -2.80033089e-02 -4.44959164e-01 5.96914709e-01
-2.56775945e-01 2.64538229e-01 3.19349438e-01 -5.63778937e-01
4.44635689e-01 -7.68217146e-02 -9.92983222e-01 5.94558716e-01
1.15937531e+00 1.00107700e-01 -5.74702144e-01 -6.20895103e-02
2.47771755e-01 -1.16588783e+00 -9.60436523e-01 1.36618987e-01
1.16365087e+00 -9.58210647e-01 1.23800552e+00 -1.89630017e-01
1.39366165e-01 -6.56144440e-01 -7.56963119e-02 -6.42495334e-01
-3.56338501e-01 -6.00720763e-01 -3.18156540e-01 8.78082037e-01
-2.10172996e-01 1.19375169e-01 1.32393646e+00 1.06902742e+00
-1.56838715e-01 -6.26280725e-01 -1.09507644e+00 -6.24021113e-01
-5.46355724e-01 -4.21093702e-01 9.16916430e-02 5.44125855e-01
3.97728421e-02 -8.95444304e-02 -1.01425648e+00 3.82616252e-01
8.70104492e-01 2.93785483e-01 1.33125770e+00 -8.50216925e-01
-1.52997464e-01 1.17686965e-01 -8.41253877e-01 -1.51503205e+00
-5.24622321e-01 -2.31457055e-01 -7.54412711e-02 -1.51956797e+00
3.03832024e-01 -1.57480195e-01 2.63605326e-01 -9.64446589e-02
-2.91772962e-01 4.89240587e-01 4.41154897e-01 6.05612755e-01
-6.45634234e-01 1.21852256e-01 1.48053908e+00 2.00818151e-01
-2.56277800e-01 3.00475210e-01 -6.17290400e-02 8.91062796e-01
4.48239177e-01 -4.14925456e-01 -1.58617392e-01 -5.29528797e-01
-1.71270594e-01 6.34715676e-01 7.11658835e-01 -1.40188861e+00
5.94805181e-01 -3.67304653e-01 9.17978287e-01 -9.17478740e-01
8.34077179e-01 -8.36423397e-01 7.91656673e-01 8.28498483e-01
7.22128600e-02 2.53550678e-01 1.13876355e-04 7.16434360e-01
1.04320832e-01 4.78947610e-02 6.25765324e-01 -2.57040918e-01
-1.06462049e+00 3.98561418e-01 -1.31336898e-01 -3.13375518e-02
1.25020981e+00 -1.22373927e+00 6.91511631e-02 -5.14704287e-01
-9.13047075e-01 1.44603536e-01 7.69610226e-01 3.54453474e-01
1.06893337e+00 -1.64792013e+00 -6.93318665e-01 1.04615338e-01
-1.71616912e-01 -1.42214224e-01 5.34915566e-01 1.03326929e+00
-9.17938769e-01 3.61229807e-01 -4.57253993e-01 -1.08081806e+00
-1.67130029e+00 9.93611589e-02 3.62119347e-01 1.35871693e-01
-1.07160819e+00 4.40743238e-01 -2.50056535e-01 2.89013237e-01
-5.66722564e-02 -3.64804864e-02 -9.20108259e-02 -4.18736309e-01
6.65244222e-01 8.76806378e-01 -3.45595568e-01 -1.09801292e+00
-5.50682306e-01 1.25343883e+00 4.27521795e-01 -4.58731979e-01
1.07150173e+00 -7.02818751e-01 6.19467199e-01 2.84803599e-01
7.09915817e-01 4.21802014e-01 -1.90989709e+00 1.29568979e-01
-3.34110588e-01 -1.21725190e+00 -4.09021407e-01 -2.15286300e-01
-9.51527059e-01 6.29092634e-01 8.86765778e-01 -4.31156844e-01
8.11905265e-01 -2.22191751e-01 1.00370061e+00 -2.73970574e-01
5.83288074e-01 -1.22935879e+00 9.58908051e-02 9.00978148e-02
6.18537009e-01 -8.64635229e-01 4.53484148e-01 -5.92911839e-01
-9.36489224e-01 1.12302327e+00 8.47273648e-01 -4.62794125e-01
2.99451560e-01 3.30035806e-01 8.85639116e-02 -3.92352268e-02
-4.31240052e-01 1.75650597e-01 3.45875353e-01 7.19611228e-01
3.12188327e-01 -1.03917560e-02 -3.78186971e-01 5.08256793e-01
-1.38289198e-01 3.07033449e-01 6.49358928e-01 1.10078156e+00
-6.59081519e-01 -6.69903994e-01 -1.00545681e+00 -9.57873091e-02
-5.05803704e-01 6.17590487e-01 -5.04340082e-02 1.02650833e+00
1.90803558e-01 7.31979549e-01 1.10837713e-01 -3.44001681e-01
6.21715665e-01 -3.67159307e-01 7.16705441e-01 -3.67113650e-01
-3.60308856e-01 4.94260669e-01 1.52768105e-01 -7.69979954e-01
-5.24210632e-01 -7.47008741e-01 -1.42620289e+00 -4.51121330e-01
-3.81873608e-01 -7.20956326e-02 4.80275869e-01 3.67421359e-01
3.24847758e-01 3.91163165e-03 1.99814200e-01 -1.14171648e+00
-1.47476852e-01 -6.31979287e-01 -5.63372314e-01 6.50030375e-01
1.44074187e-01 -7.96188712e-01 1.71733081e-01 4.05232280e-01] | [7.229074478149414, -0.8978018760681152] |
4b3bd3a3-e6cc-4e7e-a285-46a879a7656f | multi-granularity-generator-for-temporal | 1811.11524 | null | http://arxiv.org/abs/1811.11524v2 | http://arxiv.org/pdf/1811.11524v2.pdf | Multi-granularity Generator for Temporal Action Proposal | Temporal action proposal generation is an important task, aiming to localize
the video segments containing human actions in an untrimmed video. In this
paper, we propose a multi-granularity generator (MGG) to perform the temporal
action proposal from different granularity perspectives, relying on the video
visual features equipped with the position embedding information. First, we
propose to use a bilinear matching model to exploit the rich local information
within the video sequence. Afterwards, two components, namely segment proposal
producer (SPP) and frame actionness producer (FAP), are combined to perform the
task of temporal action proposal at two distinct granularities. SPP considers
the whole video in the form of feature pyramid and generates segment proposals
from one coarse perspective, while FAP carries out a finer actionness
evaluation for each video frame. Our proposed MGG can be trained in an
end-to-end fashion. By temporally adjusting the segment proposals with
fine-grained frame actionness information, MGG achieves the superior
performance over state-of-the-art methods on the public THUMOS-14 and
ActivityNet-1.3 datasets. Moreover, we employ existing action classifiers to
perform the classification of the proposals generated by MGG, leading to
significant improvements compared against the competing methods for the video
detection task. | ['Shih-Fu Chang', 'Yuan Liu', 'Yifeng Zhang', 'Lin Ma', 'Wei Liu'] | 2018-11-28 | multi-granularity-generator-for-temporal-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Liu_Multi-Granularity_Generator_for_Temporal_Action_Proposal_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Liu_Multi-Granularity_Generator_for_Temporal_Action_Proposal_CVPR_2019_paper.pdf | cvpr-2019-6 | ['temporal-action-proposal-generation'] | ['computer-vision'] | [ 4.61370170e-01 -1.30127892e-01 -3.32628578e-01 -8.41024145e-03
-7.60177016e-01 -1.40832037e-01 7.67972946e-01 -6.53485879e-02
-4.64806795e-01 5.08775592e-01 5.93211710e-01 2.34413236e-01
1.57773450e-01 -6.52045667e-01 -6.10165060e-01 -8.91755819e-01
-3.31543274e-02 -1.33695379e-02 7.72694886e-01 2.27056053e-02
3.15021545e-01 2.55451232e-01 -1.60639870e+00 6.97255254e-01
6.19089544e-01 1.33644617e+00 2.89637834e-01 4.77227747e-01
2.03817889e-01 1.17415202e+00 -3.67095530e-01 -2.10455477e-01
3.37482512e-01 -4.01590586e-01 -6.17291212e-01 4.42515701e-01
4.70596313e-01 -5.79846859e-01 -5.12892067e-01 8.77200961e-01
4.20290977e-01 2.85847962e-01 4.26174551e-01 -1.29988372e+00
-7.59051517e-02 5.24254143e-01 -6.82874799e-01 4.43887502e-01
5.80013335e-01 5.26773870e-01 1.03575301e+00 -1.07035148e+00
7.79660463e-01 1.33897579e+00 3.03087234e-01 3.93273056e-01
-9.33158278e-01 -5.76899946e-01 4.33338255e-01 7.67771363e-01
-1.38033724e+00 -4.38005030e-01 9.16206837e-01 -6.46732330e-01
7.44128823e-01 5.22515178e-02 8.12591255e-01 1.21843600e+00
2.01804221e-01 1.13610339e+00 7.52030253e-01 -1.69280276e-01
1.55615836e-01 -5.44440687e-01 -4.47177440e-01 6.40158772e-01
-2.28000641e-01 4.05226499e-02 -7.31489480e-01 -4.76513058e-02
9.31870818e-01 2.70963758e-01 -4.75794882e-01 -4.13189948e-01
-1.75822043e+00 6.35780275e-01 3.71553034e-01 3.17296654e-01
-7.77881265e-01 1.51961654e-01 5.93649328e-01 -9.99039784e-02
4.86618131e-01 1.05765007e-01 -2.54182845e-01 -2.30818033e-01
-1.10300744e+00 3.08007866e-01 2.43414715e-01 7.98963547e-01
6.62320018e-01 -8.32648575e-02 -1.12535727e+00 5.06493926e-01
3.08200032e-01 3.73570211e-02 5.25646627e-01 -1.03882194e+00
8.64313900e-01 6.77129209e-01 3.12555879e-01 -1.05859327e+00
-1.36429965e-01 -2.25726619e-01 -7.97625959e-01 -2.28799544e-02
4.21130151e-01 5.17502129e-02 -7.82797158e-01 1.53714454e+00
5.42002380e-01 5.91952205e-01 -1.90244451e-01 1.02131140e+00
5.85167468e-01 7.56128252e-01 4.19672519e-01 -2.87670135e-01
1.55584431e+00 -1.34666073e+00 -5.25862396e-01 -7.53399879e-02
6.19163811e-01 -5.96642852e-01 7.84154475e-01 3.39863360e-01
-1.06471121e+00 -8.52038205e-01 -8.46523583e-01 5.72746582e-02
1.18346907e-01 7.44091630e-01 1.11982137e-01 7.42009804e-02
-8.74850214e-01 4.95133460e-01 -1.00134516e+00 -1.84802279e-01
5.91555357e-01 2.95359036e-03 -4.14698213e-01 -1.74509827e-02
-1.11876428e+00 4.98849154e-01 8.39834273e-01 2.15352625e-01
-1.13498962e+00 -4.36221868e-01 -9.47261512e-01 1.15853108e-01
6.81251109e-01 -6.22706711e-01 1.01285100e+00 -1.08284128e+00
-1.42261434e+00 4.98655379e-01 -1.21465407e-01 -7.14938879e-01
8.16727579e-01 -9.57957879e-02 -1.74760997e-01 6.81812465e-01
2.78534055e-01 8.64890754e-01 1.16459787e+00 -5.80139399e-01
-1.15333021e+00 -1.66460872e-01 1.57069504e-01 3.05263013e-01
-2.02713266e-01 1.34036928e-01 -7.72023022e-01 -1.19877517e+00
6.31522834e-02 -7.67593682e-01 -1.78516194e-01 1.06097437e-01
-2.89291859e-01 -5.87536633e-01 8.04963946e-01 -8.40300441e-01
1.50629592e+00 -2.15866280e+00 3.79505426e-01 -1.68913200e-01
2.87103027e-01 3.07607383e-01 -1.56940445e-01 2.46999219e-01
1.28712356e-01 -3.38893205e-01 -1.52033553e-01 -2.76664972e-01
-2.05526836e-02 -2.25135371e-01 -9.71229374e-02 4.47687835e-01
3.43179971e-01 9.46045280e-01 -1.14961374e+00 -7.83756495e-01
5.24002790e-01 2.18141362e-01 -7.02684283e-01 3.88112098e-01
-1.77152827e-01 6.00582838e-01 -8.04768980e-01 6.61248028e-01
3.84791285e-01 -2.32466206e-01 8.13413486e-02 -5.21280706e-01
-2.05434799e-01 4.70103286e-02 -1.19270825e+00 1.87256777e+00
-2.44656354e-01 3.68962854e-01 -1.83847681e-01 -9.87507701e-01
6.60174310e-01 5.12789726e-01 8.43641043e-01 -5.44038415e-01
2.56708842e-02 6.03798330e-02 -1.24717951e-01 -5.80660582e-01
3.75187308e-01 1.55448347e-01 -1.16134122e-01 3.16065073e-01
2.02181324e-01 6.12637222e-01 5.43925941e-01 1.03333332e-01
1.35386467e+00 5.95600426e-01 6.93955898e-01 5.43227866e-02
1.03478456e+00 -3.72655898e-01 7.80929565e-01 4.69765753e-01
-4.60977525e-01 6.79581463e-01 4.99404967e-01 -5.85399091e-01
-8.21959853e-01 -8.77453208e-01 3.15882057e-01 1.22989345e+00
3.09033096e-01 -4.72934514e-01 -7.73928344e-01 -1.03352797e+00
-1.97616249e-01 3.17706972e-01 -7.73948967e-01 -1.04494832e-01
-8.76740694e-01 -3.42320204e-01 2.32493997e-01 7.30005801e-01
1.04811108e+00 -1.45902121e+00 -9.97791350e-01 4.73145425e-01
-5.82196712e-01 -1.40571487e+00 -9.50964928e-01 -3.05376261e-01
-6.69845581e-01 -1.05473506e+00 -9.90047097e-01 -4.95264828e-01
4.46835130e-01 1.58805579e-01 7.91579783e-01 -4.12855223e-02
-2.09793970e-01 2.11534068e-01 -6.68872654e-01 3.95806968e-01
-2.70683408e-01 -1.38924137e-01 -1.34297341e-01 8.75045419e-01
2.93925911e-01 -4.10199583e-01 -1.14554417e+00 6.22095048e-01
-9.08196449e-01 3.80570769e-01 8.52059543e-01 6.44808292e-01
5.76523066e-01 -1.64315030e-01 3.35288048e-01 -3.17741096e-01
2.75347847e-02 -3.53946388e-01 -3.57986748e-01 3.37136924e-01
2.48004310e-02 -5.83775938e-02 5.28197587e-01 -4.03198421e-01
-1.08528674e+00 3.20657074e-01 -9.67462137e-02 -7.45934188e-01
-2.57370591e-01 6.20882437e-02 -3.77742290e-01 7.26362020e-02
4.58850652e-01 5.16971648e-01 -3.59872311e-01 -4.57917869e-01
3.72163415e-01 4.31208014e-01 5.09216309e-01 -3.74086022e-01
5.58110654e-01 5.04288554e-01 -2.21661359e-01 -5.93608737e-01
-6.45276129e-01 -7.09768176e-01 -7.55509138e-01 -4.67094749e-01
1.28456867e+00 -1.05022776e+00 -4.53434765e-01 6.73515737e-01
-1.27549279e+00 -1.69677407e-01 -2.20526680e-01 5.94329119e-01
-8.06605160e-01 6.11122847e-01 -5.02133846e-01 -4.79931772e-01
-3.04788768e-01 -1.33519518e+00 1.46125829e+00 7.05377097e-05
3.17250714e-02 -5.76750219e-01 -1.59285665e-01 4.13608432e-01
5.36816241e-03 4.88902003e-01 3.20202738e-01 -3.43519688e-01
-9.69642520e-01 -7.81228989e-02 -2.38729969e-01 4.16794419e-01
1.58375844e-01 -9.86665264e-02 -6.57041907e-01 -3.58694047e-01
-8.88958275e-02 -2.16691807e-01 1.07459319e+00 4.34080869e-01
1.19064021e+00 -5.36356747e-01 -3.68268102e-01 5.11848330e-01
1.11793888e+00 1.44264296e-01 7.60648131e-01 2.49117091e-01
8.57226789e-01 3.19671184e-01 1.24052632e+00 7.93894410e-01
2.19340861e-01 1.15256202e+00 3.83714944e-01 1.22426778e-01
-2.19853222e-01 -3.31718743e-01 6.85112536e-01 1.93713635e-01
-4.45894241e-01 -2.12607592e-01 -4.72656310e-01 4.21863496e-01
-2.21494555e+00 -1.34019172e+00 2.69619972e-01 2.16624689e+00
5.66078305e-01 1.46546766e-01 3.66095364e-01 5.23018502e-02
9.48165894e-01 6.43834054e-01 -3.25650185e-01 3.24860930e-01
1.69255272e-01 -6.64679930e-02 3.87214720e-01 -3.63146588e-02
-1.59553754e+00 8.85211349e-01 4.48658514e+00 1.23528135e+00
-1.00416946e+00 2.85257757e-01 5.62723637e-01 -1.12606995e-01
2.30535671e-01 -8.93657282e-02 -8.49209309e-01 8.47180605e-01
5.22532523e-01 4.88687456e-02 7.05186650e-02 6.87690079e-01
6.60913050e-01 -1.79339245e-01 -1.03280139e+00 1.02783656e+00
1.32187337e-01 -1.43463218e+00 2.63774425e-01 -1.40925467e-01
6.79489434e-01 -2.23998338e-01 -1.93478480e-01 3.04372460e-01
-1.98004559e-01 -4.52247828e-01 1.15965819e+00 6.52233362e-01
6.64469600e-01 -5.43611109e-01 5.92956960e-01 2.60643333e-01
-1.84951544e+00 -5.04043341e-01 -1.12213813e-01 1.80430442e-01
4.48209882e-01 2.94105709e-01 -5.79213321e-01 6.04739547e-01
6.23443604e-01 1.14551842e+00 -5.82753181e-01 1.22807360e+00
-3.37345511e-01 4.56626445e-01 8.34526569e-02 3.02103609e-01
4.59522158e-01 -1.26951948e-01 6.07738435e-01 1.07641768e+00
4.89979893e-01 1.35498852e-01 5.11520565e-01 6.24428213e-01
-3.85394692e-02 1.04892656e-01 -2.03520924e-01 5.52554093e-02
2.86135882e-01 1.32267869e+00 -8.10680509e-01 -6.20775342e-01
-4.98234123e-01 1.25584614e+00 1.23550273e-01 2.78128445e-01
-1.15298676e+00 -1.24571204e-01 5.27932823e-01 3.17397565e-01
7.62459636e-01 -2.83144955e-02 5.30739307e-01 -1.28581023e+00
3.36019248e-01 -1.01249218e+00 6.04342997e-01 -6.65831923e-01
-8.62679183e-01 5.15459359e-01 1.02894090e-01 -1.98717320e+00
-3.86861652e-01 -3.35805207e-01 -6.57567322e-01 4.39009070e-01
-1.13807344e+00 -1.31509936e+00 -4.54752862e-01 6.39368713e-01
1.10734129e+00 -2.65518315e-02 1.94376618e-01 4.16064501e-01
-5.65764904e-01 3.95597130e-01 -3.80845398e-01 2.84667313e-01
5.57408512e-01 -8.67730498e-01 3.42647463e-01 1.11006999e+00
2.41271675e-01 7.86794573e-02 3.13496768e-01 -6.34380937e-01
-9.27815378e-01 -1.46117747e+00 8.02546978e-01 -3.63141209e-01
6.40579104e-01 -1.83578938e-01 -6.87821746e-01 4.62831080e-01
2.84138205e-03 3.27290386e-01 1.71699762e-01 -6.60283744e-01
-6.50689527e-02 -1.20064579e-01 -8.60616922e-01 5.40013850e-01
1.20593464e+00 -3.48936975e-01 -6.24712706e-01 2.54040539e-01
7.07071602e-01 -3.71211827e-01 -8.34252775e-01 4.96958464e-01
5.32549322e-01 -1.12291372e+00 1.02169013e+00 -3.84092152e-01
6.06744707e-01 -7.04464495e-01 -5.48659936e-02 -7.38868058e-01
-4.92216229e-01 -6.75768316e-01 -3.71686488e-01 1.01169062e+00
-1.39886990e-01 -1.58079997e-01 6.43961132e-01 5.10070920e-02
-2.01582491e-01 -8.86660814e-01 -1.12896454e+00 -6.47044837e-01
-6.49487734e-01 -3.38185132e-01 4.82984543e-01 3.86985958e-01
-1.70439273e-01 9.10640210e-02 -6.47199512e-01 -9.30584297e-02
5.55976927e-01 1.29445702e-01 6.97895467e-01 -7.84073412e-01
-4.74802643e-01 -4.78529036e-01 -8.25283468e-01 -1.38039029e+00
3.15604843e-02 -6.00743353e-01 1.02905065e-01 -1.31876624e+00
1.70432433e-01 1.68653145e-01 -5.76026797e-01 4.43325073e-01
-3.79934609e-01 5.84230006e-01 4.53711808e-01 2.44401231e-01
-1.02448821e+00 8.31590652e-01 1.30192268e+00 -2.40561023e-01
-1.32025287e-01 1.67424485e-01 -1.98234171e-01 8.49712610e-01
3.62220019e-01 -2.63286978e-01 -2.99837977e-01 -5.62484860e-02
-4.37929899e-01 3.01628768e-01 7.56403804e-01 -1.37426770e+00
7.22859949e-02 -1.46708488e-01 3.77376854e-01 -7.57095933e-01
2.84056306e-01 -5.68881035e-01 -1.52956313e-02 5.82352877e-01
-3.33744138e-01 -1.07356355e-01 -2.17577308e-01 9.06362832e-01
-3.61555338e-01 8.05477798e-02 8.15965235e-01 -5.98387085e-02
-1.14283562e+00 7.16128945e-01 -3.09638560e-01 2.41493154e-02
1.36749792e+00 -2.39858329e-01 2.51205377e-02 2.88600773e-02
-8.04926217e-01 8.48080367e-02 3.59037757e-01 5.76007485e-01
7.38541603e-01 -1.58461010e+00 -7.44122386e-01 1.88036025e-01
4.14698839e-01 -3.40224564e-01 6.23436689e-01 1.35388768e+00
-4.30450350e-01 2.58469015e-01 -3.38176697e-01 -7.93964744e-01
-1.26961291e+00 7.19380617e-01 1.63022086e-01 -5.16995430e-01
-8.47753108e-01 7.67839909e-01 5.55309474e-01 3.13756943e-01
1.89794257e-01 -5.23208916e-01 -5.10780036e-01 2.33711183e-01
7.35158145e-01 4.31914985e-01 -1.88424379e-01 -9.79452193e-01
-3.34077358e-01 6.63334906e-01 1.10904500e-01 6.97163641e-02
9.50129569e-01 -4.76517938e-02 1.79543018e-01 1.56855695e-02
1.15202093e+00 -3.86961609e-01 -1.86028135e+00 -3.55069011e-01
-1.95737407e-02 -7.79539585e-01 -1.11669488e-01 -3.02842140e-01
-1.24189997e+00 7.94785023e-01 5.34945130e-01 -1.10639520e-01
1.36742759e+00 4.22210852e-03 8.54357958e-01 -4.31913957e-02
5.55285513e-01 -1.08746195e+00 4.63351399e-01 1.56381220e-01
9.39519823e-01 -1.09976888e+00 -4.04886752e-02 -3.63423347e-01
-7.62160659e-01 1.11324906e+00 6.41287863e-01 -1.12143205e-02
2.75332630e-01 -3.02997738e-01 -3.37627649e-01 5.57788042e-03
-6.66973948e-01 -3.69296610e-01 5.82746983e-01 3.05295438e-01
1.66739479e-01 -2.85484251e-02 -5.73353827e-01 4.21151221e-01
4.89800751e-01 2.88523853e-01 1.30211562e-01 6.65393889e-01
-5.20130932e-01 -9.55235183e-01 -3.39863718e-01 4.41100419e-01
-2.77005106e-01 4.97464649e-02 -2.49734502e-02 5.22025585e-01
5.30444026e-01 7.41975784e-01 2.75390092e-02 -4.58226591e-01
1.89940482e-01 -1.33350998e-01 4.29178119e-01 -5.48619926e-01
-3.80977124e-01 3.09514463e-01 -3.06118149e-02 -1.20356655e+00
-6.74333394e-01 -9.13547158e-01 -1.09415483e+00 2.16495663e-01
6.58684224e-02 -3.52214761e-02 -1.41000703e-01 1.05994678e+00
3.77550662e-01 5.21096528e-01 7.32746065e-01 -1.35949314e+00
-4.73025233e-01 -1.05408657e+00 -4.39494461e-01 6.10608637e-01
1.94892183e-01 -9.17977512e-01 -2.32933909e-01 3.06786627e-01] | [8.537554740905762, 0.48761439323425293] |
c17420d2-3a94-420e-9e47-9d2495679f0d | universal-domain-adaptation-from-foundation | 2305.11092 | null | https://arxiv.org/abs/2305.11092v1 | https://arxiv.org/pdf/2305.11092v1.pdf | Universal Domain Adaptation from Foundation Models | Foundation models (e.g., CLIP or DINOv2) have shown their impressive learning and transferring capabilities on a wide range of visual tasks, by training on a large corpus of data and adapting to specific downstream tasks. It is, however, interesting that foundation models have not been fully explored for universal domain adaptation (UniDA), which is to learn models using labeled data in a source domain and unlabeled data in a target one, such that the learned models can successfully adapt to the target data. In this paper, we make comprehensive empirical studies of state-of-the-art UniDA methods using foundation models. We first demonstrate that, while foundation models greatly improve the performance of the baseline methods that train the models on the source data alone, existing UniDA methods generally fail to improve over the baseline. This suggests that new research efforts are very necessary for UniDA using foundation models. To this end, we propose a very simple method of target data distillation on the CLIP model, and achieves consistent improvement over the baseline across all the UniDA benchmarks. Our studies are under a newly proposed evaluation metric of universal classification rate (UCR), which is threshold- and ratio-free and addresses the threshold-sensitive issue encountered when using the existing H-score metric. | ['Kui Jia', 'Bin Deng'] | 2023-05-18 | null | null | null | null | ['universal-domain-adaptation'] | ['computer-vision'] | [ 9.81564298e-02 -2.81290203e-01 -4.61726457e-01 -4.10580575e-01
-9.28808570e-01 -6.73639953e-01 7.98195779e-01 -3.07070732e-01
-4.26400542e-01 8.41182351e-01 1.03315756e-01 -1.56964928e-01
2.74153024e-01 -2.89091945e-01 -6.45030379e-01 -7.20581234e-01
7.02369064e-02 5.97553015e-01 6.01939559e-01 -1.04694225e-01
-8.08457360e-02 5.30461550e-01 -1.49965000e+00 4.27475095e-01
6.24834716e-01 9.62744772e-01 2.97265887e-01 6.22423589e-01
6.53419569e-02 4.86595899e-01 -7.67245471e-01 -4.15586948e-01
2.37952247e-01 -5.45792699e-01 -8.26445639e-01 6.87540397e-02
8.37213814e-01 -3.81970525e-01 -4.44558531e-01 8.04588497e-01
7.57133484e-01 1.73616588e-01 1.01183009e+00 -1.15771890e+00
-1.04176557e+00 2.52860159e-01 -5.20278096e-01 4.74637061e-01
9.65974946e-03 1.41670723e-02 1.05925226e+00 -1.12751842e+00
8.43274236e-01 1.19327605e+00 4.91125286e-01 9.86076772e-01
-1.40176129e+00 -7.46195972e-01 3.45577806e-01 2.66679257e-01
-1.18211603e+00 -4.59668338e-01 5.28920233e-01 -6.76276028e-01
9.64106441e-01 8.48368462e-03 -6.53770613e-03 1.60464680e+00
6.67987689e-02 1.00750053e+00 1.22786582e+00 -6.45090699e-01
1.58182040e-01 3.79659951e-01 1.31274432e-01 1.83940396e-01
8.20447877e-02 3.40165854e-01 -4.16764885e-01 2.10301429e-01
7.86127865e-01 -3.39395761e-01 -4.17650938e-02 -8.00429344e-01
-1.19088113e+00 7.95696735e-01 4.85388786e-01 1.32106856e-01
8.05208534e-02 -2.13781714e-01 5.30241966e-01 4.31392670e-01
5.88891208e-01 6.02290630e-01 -6.84522212e-01 6.92468062e-02
-8.68528843e-01 2.74189115e-01 5.21096349e-01 1.29240525e+00
6.20770872e-01 1.67012542e-01 -4.63671595e-01 1.02186847e+00
-2.75824051e-02 4.96597648e-01 6.53507531e-01 -8.00215781e-01
3.90232146e-01 4.02805805e-01 5.48771024e-03 -3.53460014e-01
-2.48854086e-01 -5.75029790e-01 -7.14655101e-01 5.03665984e-01
6.43544197e-01 -1.25915632e-01 -1.33584261e+00 1.84643412e+00
1.79024130e-01 2.86557823e-01 2.87933171e-01 7.79860675e-01
8.21617007e-01 6.42614245e-01 2.62434065e-01 -1.04711037e-02
1.14148962e+00 -1.29359913e+00 -3.80375594e-01 -4.25795525e-01
5.10123312e-01 -7.52799690e-01 1.38638043e+00 4.07652527e-01
-8.60128403e-01 -8.46269429e-01 -1.30329621e+00 -8.37087780e-02
-6.07896864e-01 2.88646340e-01 4.45044041e-01 4.46979195e-01
-9.39562023e-01 1.78235412e-01 -5.38266242e-01 -7.04939425e-01
4.95872229e-01 1.74659967e-01 -5.19845128e-01 -3.69263589e-01
-1.06220353e+00 1.13284147e+00 3.17971170e-01 -4.93964344e-01
-1.21216691e+00 -7.30217636e-01 -5.89352310e-01 -1.35048732e-01
1.79725081e-01 -4.73394871e-01 1.54873490e+00 -1.19207978e+00
-1.34785676e+00 1.08050084e+00 1.06858537e-01 -5.86732864e-01
3.56540024e-01 -2.78980851e-01 -6.68612003e-01 -3.36692668e-02
2.71888729e-02 1.07926130e+00 7.56533265e-01 -1.45284331e+00
-9.54019666e-01 -1.70748174e-01 1.21332757e-01 2.41760463e-01
-4.82023954e-01 3.69898006e-02 -5.99637747e-01 -7.97026336e-01
-5.27350247e-01 -8.72883618e-01 -1.95513647e-02 -6.36541620e-02
-1.36532351e-01 -3.59993130e-01 9.32122648e-01 -4.75561261e-01
1.40605772e+00 -2.26455164e+00 2.43690312e-01 -1.09752670e-01
-5.92587814e-02 6.11880004e-01 -4.61849958e-01 1.64443657e-01
-5.23431562e-02 -3.31338346e-02 -2.07051560e-01 -2.07314655e-01
-9.59560499e-02 2.12874278e-01 -5.16258478e-01 1.04925364e-01
1.69641003e-01 6.61965251e-01 -7.73351967e-01 -2.37013012e-01
-6.81426236e-03 2.69384831e-01 -5.14637709e-01 3.93689305e-01
-2.70706236e-01 3.03255379e-01 -1.32537872e-01 5.56144714e-01
5.28277934e-01 -3.27648073e-01 -9.11778025e-03 -1.79918572e-01
9.52689908e-03 7.56387636e-02 -9.69433069e-01 1.82389164e+00
-4.06414241e-01 8.93213034e-01 -3.06668133e-01 -1.03655863e+00
9.87268507e-01 2.64901400e-01 2.37577662e-01 -7.45037556e-01
-1.81861907e-01 3.61385882e-01 6.49829283e-02 -1.87015042e-01
2.82443434e-01 1.29687428e-01 -2.08899871e-01 1.34041652e-01
5.29182374e-01 1.55052483e-01 3.20470154e-01 1.90354243e-01
1.00196803e+00 4.57583576e-01 4.46492255e-01 -2.74402916e-01
4.08537686e-01 3.35850954e-01 3.94866168e-01 7.53365636e-01
-4.38382417e-01 9.07143712e-01 2.04904348e-01 -4.67163980e-01
-1.27490163e+00 -1.38353217e+00 -3.83004457e-01 1.48244989e+00
1.29822582e-01 -7.76810199e-02 -5.77275693e-01 -1.18210864e+00
8.90240148e-02 7.15080440e-01 -8.00094485e-01 -3.03100377e-01
-2.58267850e-01 -6.25563979e-01 6.08958781e-01 9.31406498e-01
5.55189729e-01 -8.72523248e-01 -3.29160184e-01 2.32386500e-01
2.73487210e-01 -1.23354805e+00 -5.33724427e-01 2.88070202e-01
-7.30029047e-01 -7.75014162e-01 -1.07738853e+00 -1.03295374e+00
4.84250605e-01 4.62142885e-01 1.24701476e+00 -5.00736415e-01
-6.30536154e-02 5.05338550e-01 -4.22974437e-01 -4.32071596e-01
-3.80930185e-01 2.86791176e-01 1.39163449e-01 -1.42539740e-01
7.64445782e-01 -4.32341516e-01 -4.15156424e-01 6.16397917e-01
-7.31998801e-01 -6.50122687e-02 8.69491279e-01 9.64147925e-01
7.11225569e-01 -6.05144262e-01 1.04845345e+00 -1.03376746e+00
5.70249081e-01 -5.47460973e-01 -4.81773347e-01 3.83812875e-01
-7.68060029e-01 -1.13029450e-01 5.03898382e-01 -7.39787579e-01
-1.30818045e+00 2.27411598e-01 6.13011494e-02 -6.79418504e-01
-3.88271928e-01 3.72141480e-01 -1.12550132e-01 1.07994840e-01
1.17119622e+00 4.86042947e-02 -2.36290395e-01 -6.50753975e-01
4.70837086e-01 8.97740781e-01 9.85060632e-01 -5.73297322e-01
7.10991681e-01 2.66558647e-01 -3.22802961e-01 -5.42315960e-01
-1.07182515e+00 -6.77826583e-01 -9.97360408e-01 1.51967689e-01
8.50285113e-01 -1.26281786e+00 2.95254141e-01 4.55440581e-01
-9.98631537e-01 -6.80212855e-01 -2.34027117e-01 4.20266390e-01
-5.15546262e-01 1.24597013e-01 -2.32937261e-01 -2.26484299e-01
-3.60594429e-02 -1.07354164e+00 7.65173316e-01 3.28207612e-01
-1.65120453e-01 -1.24287963e+00 4.38122779e-01 9.25022811e-02
4.62689310e-01 6.29800111e-02 7.95968175e-01 -1.03985560e+00
-2.21664563e-01 -1.54870465e-01 -4.02442247e-01 7.03878045e-01
2.20205262e-01 -1.01010486e-01 -1.26446342e+00 -5.10329247e-01
-5.53750753e-01 -6.21290565e-01 9.77761090e-01 4.07520652e-01
1.19750011e+00 6.37406632e-02 -4.84745651e-01 6.61196053e-01
1.29844308e+00 3.04127723e-01 7.12533593e-01 5.45400381e-01
4.88024503e-01 3.03432256e-01 6.99507117e-01 1.92400247e-01
3.45229506e-01 9.66888845e-01 1.41494751e-01 -3.64313364e-01
-7.87364423e-01 -2.50606328e-01 5.94230950e-01 4.22892541e-01
-3.51239443e-02 -4.12382275e-01 -9.15670037e-01 8.99498999e-01
-1.79366386e+00 -7.38908768e-01 9.61692557e-02 2.29092002e+00
9.08211291e-01 1.57641232e-01 4.04883295e-01 -4.48096901e-01
7.47646272e-01 -1.22953519e-01 -8.03340971e-01 -4.29480731e-01
-3.89815748e-01 1.94411620e-01 3.80822450e-01 2.56649017e-01
-1.51170611e+00 1.10660625e+00 7.09797573e+00 8.86144698e-01
-1.09712362e+00 2.69738972e-01 6.24572515e-01 -1.05698794e-01
9.34542492e-02 -1.83082893e-01 -1.08489227e+00 3.29297453e-01
1.02025759e+00 -2.30215967e-01 1.96890190e-01 1.19019866e+00
-1.77228853e-01 1.41814485e-01 -1.38832569e+00 9.09215629e-01
1.63249657e-01 -1.25027132e+00 2.20144406e-01 -3.00220996e-02
1.13323939e+00 2.40759209e-01 2.48388052e-01 7.64393389e-01
6.53749347e-01 -9.43115175e-01 3.41825247e-01 2.93906890e-02
1.23760700e+00 -5.05796671e-01 6.47830486e-01 -3.31766047e-02
-8.34297776e-01 1.53346825e-02 -6.33957684e-01 2.29777172e-01
-1.06762357e-01 2.01401994e-01 -7.71005630e-01 3.90026689e-01
7.89128363e-01 8.45805109e-01 -7.91478813e-01 1.28755116e+00
-2.61302859e-01 8.31640422e-01 -2.13875659e-02 4.65849757e-01
3.77598047e-01 2.94273466e-01 4.90617782e-01 1.53824854e+00
3.78221691e-01 -2.82763869e-01 2.30099067e-01 3.51558715e-01
-1.83260724e-01 1.58945933e-01 -7.95127392e-01 1.71150863e-01
4.46874917e-01 1.09753013e+00 -2.81775326e-01 -3.79786372e-01
-8.63165796e-01 1.07880259e+00 5.26881278e-01 6.18421257e-01
-7.32728601e-01 -2.67186701e-01 6.71785474e-01 1.65453553e-02
5.30453563e-01 -1.25913903e-01 -2.88594484e-01 -1.25537694e+00
-1.27542004e-01 -1.00967133e+00 8.71354997e-01 -7.07838595e-01
-1.61705554e+00 5.94091296e-01 1.97839588e-01 -1.49371505e+00
-2.80591369e-01 -8.23834896e-01 -4.75451976e-01 7.19941914e-01
-1.76227319e+00 -1.34122896e+00 -1.42345265e-01 8.76119196e-01
9.19705629e-01 -6.57879055e-01 8.04853737e-01 3.95521879e-01
-5.12088180e-01 1.07144272e+00 4.04562801e-01 1.38378814e-01
1.43551016e+00 -1.32583618e+00 4.20886636e-01 1.03288877e+00
3.52987349e-01 3.35997999e-01 5.40028930e-01 -4.03580040e-01
-9.05613303e-01 -1.15760016e+00 5.69012046e-01 -6.95551217e-01
5.36402285e-01 -4.43985164e-01 -1.10931337e+00 1.02692461e+00
3.63515079e-01 2.38271385e-01 6.92745686e-01 2.32350528e-01
-7.00651407e-01 -2.81198502e-01 -9.85377848e-01 5.07363677e-01
1.18758035e+00 -2.59074450e-01 -6.62603498e-01 2.45173723e-01
5.94523072e-01 -4.54471469e-01 -8.71976078e-01 4.25043553e-01
4.58474487e-01 -7.17676103e-01 9.94668007e-01 -9.88635182e-01
4.90683138e-01 -2.28634626e-01 -3.69437724e-01 -1.63542485e+00
-6.75160229e-01 -2.74424255e-01 -2.50569135e-01 1.42187464e+00
7.11907446e-01 -4.66396511e-01 4.22027767e-01 3.88487637e-01
-4.18225497e-01 -4.55463260e-01 -1.01678514e+00 -1.19720304e+00
4.57577795e-01 -1.83483288e-01 3.00780147e-01 1.01838505e+00
-9.05600041e-02 6.14136159e-01 -6.47235632e-01 1.07144400e-01
3.75502080e-01 -2.73863561e-02 1.03339362e+00 -1.26912177e+00
-2.36620903e-01 -3.77990007e-01 -3.64022672e-01 -1.37080157e+00
6.59602657e-02 -9.26136613e-01 -2.08558626e-02 -1.37695992e+00
2.65385032e-01 -6.09078169e-01 -7.67384171e-01 5.95025361e-01
-2.20866367e-01 3.65501434e-01 1.56391248e-01 5.82995355e-01
-6.71891987e-01 4.12392855e-01 1.22579539e+00 -2.43853688e-01
-3.47011209e-01 2.21541338e-02 -8.67768288e-01 4.97380137e-01
7.60641336e-01 -2.74319321e-01 -6.50550187e-01 -7.03098357e-01
-3.23894024e-01 -5.07484853e-01 1.19147807e-01 -1.20314562e+00
1.18764266e-01 -2.79553622e-01 4.81053442e-01 -2.61791229e-01
1.35639995e-01 -7.46108770e-01 -3.06781888e-01 1.86436940e-02
-4.55600709e-01 -2.70108953e-02 4.88485694e-01 6.82871938e-01
-3.02816689e-01 -5.02061881e-02 1.12427390e+00 1.23848327e-01
-1.52526450e+00 3.25114131e-01 -1.21235766e-01 2.44869098e-01
1.18851709e+00 -9.44127887e-02 -5.87140024e-01 -2.32713148e-01
-7.45456159e-01 2.04816639e-01 4.86875802e-01 7.43807852e-01
4.52664673e-01 -1.62290621e+00 -9.10838664e-01 6.59144223e-02
5.50983369e-01 -1.52870223e-01 1.01279527e-01 4.92804617e-01
-9.28816944e-02 4.38846320e-01 -5.87905347e-01 -7.67020941e-01
-1.00945675e+00 7.20005453e-01 2.47666672e-01 -3.99244815e-01
-4.02767003e-01 7.87728190e-01 6.96526110e-01 -4.18760628e-01
2.94783771e-01 1.37446551e-02 -2.74394035e-01 -6.13213331e-03
6.88632309e-01 1.73389539e-01 2.47062445e-02 -4.87291604e-01
-2.83762693e-01 4.32406574e-01 -4.12324756e-01 -6.00613803e-02
1.19003022e+00 -1.11093946e-01 5.70839882e-01 3.82260591e-01
1.00287783e+00 -1.03618093e-01 -1.76538324e+00 -4.82828766e-01
-7.67489523e-02 -4.12827492e-01 1.83622371e-02 -1.23297572e+00
-8.92178416e-01 1.01449382e+00 8.91106546e-01 -1.59948424e-01
1.22555161e+00 1.77067265e-01 4.65756655e-01 3.51189405e-01
8.73491168e-02 -1.23547578e+00 2.72086114e-01 5.02329588e-01
9.46416557e-01 -1.41713655e+00 -1.01973284e-02 -1.10058440e-03
-1.06788671e+00 1.04767466e+00 1.20798635e+00 9.51047055e-03
4.23711568e-01 5.81741706e-02 3.10341567e-01 2.15136915e-01
-8.93617988e-01 -4.49917644e-01 5.23733616e-01 1.18593669e+00
3.11582267e-01 1.38827935e-02 -2.52124425e-02 5.81086636e-01
2.80141413e-01 -3.13743018e-02 4.06247884e-01 5.91804206e-01
-5.19604921e-01 -1.36165929e+00 -1.70551077e-01 3.48127365e-01
-3.69175732e-01 4.59869578e-02 -4.30039555e-01 1.11259663e+00
5.87842874e-02 6.19138598e-01 1.10429421e-01 -4.02462691e-01
5.18884718e-01 3.03314388e-01 3.49047422e-01 -8.63015652e-01
-2.18277857e-01 -9.03760940e-02 1.44640386e-01 -1.62995860e-01
-5.58303893e-01 -7.28042305e-01 -7.83710182e-01 1.90891922e-01
-3.09021417e-02 -2.32469201e-01 3.90365362e-01 7.25825608e-01
5.15054584e-01 2.65306920e-01 4.73141581e-01 -8.31965744e-01
-6.14313185e-01 -1.05051768e+00 -3.49096835e-01 6.17322147e-01
2.82821059e-01 -1.01944053e+00 -2.02478707e-01 3.39450955e-01] | [10.122467041015625, 2.6717028617858887] |
99e96940-5679-4875-be3f-48110ef91cdc | fedet-a-communication-efficient-federated | 2306.15347 | null | https://arxiv.org/abs/2306.15347v1 | https://arxiv.org/pdf/2306.15347v1.pdf | FedET: A Communication-Efficient Federated Class-Incremental Learning Framework Based on Enhanced Transformer | Federated Learning (FL) has been widely concerned for it enables decentralized learning while ensuring data privacy. However, most existing methods unrealistically assume that the classes encountered by local clients are fixed over time. After learning new classes, this assumption will make the model's catastrophic forgetting of old classes significantly severe. Moreover, due to the limitation of communication cost, it is challenging to use large-scale models in FL, which will affect the prediction accuracy. To address these challenges, we propose a novel framework, Federated Enhanced Transformer (FedET), which simultaneously achieves high accuracy and low communication cost. Specifically, FedET uses Enhancer, a tiny module, to absorb and communicate new knowledge, and applies pre-trained Transformers combined with different Enhancers to ensure high precision on various tasks. To address local forgetting caused by new classes of new tasks and global forgetting brought by non-i.i.d (non-independent and identically distributed) class imbalance across different local clients, we proposed an Enhancer distillation method to modify the imbalance between old and new knowledge and repair the non-i.i.d. problem. Experimental results demonstrate that FedET's average accuracy on representative benchmark datasets is 14.1% higher than the state-of-the-art method, while FedET saves 90% of the communication cost compared to the previous method. | ['Jing Xiao', 'Jianzong Wang', 'Xiaoyang Qu', 'Chenghao Liu'] | 2023-06-27 | null | null | null | null | ['class-incremental-learning', 'incremental-learning'] | ['computer-vision', 'methodology'] | [-1.91247120e-01 -7.01774359e-02 -1.76970869e-01 -3.50769371e-01
-7.42295742e-01 -4.48281884e-01 2.44946927e-01 -2.82489937e-02
-4.17549312e-01 1.13607299e+00 -2.84514967e-02 -6.03577234e-02
-4.10812311e-02 -8.26552749e-01 -9.40396011e-01 -9.90915835e-01
6.61181435e-02 3.85660619e-01 3.02755386e-01 2.71894813e-01
-1.63564131e-01 1.43951118e-01 -1.62345135e+00 5.10874450e-01
1.05107582e+00 1.13448465e+00 -7.20459595e-02 1.46884201e-02
-3.28538418e-01 9.93650138e-01 -8.09773862e-01 -7.83608854e-01
2.91145533e-01 -6.62883595e-02 -7.91029871e-01 -4.10447717e-01
2.59779721e-01 -5.62059700e-01 -4.83306110e-01 1.03338408e+00
6.38319492e-01 -1.01958796e-01 1.43298849e-01 -1.63556767e+00
-5.73440850e-01 8.07013273e-01 -6.20458722e-01 1.00229988e-02
-2.14638636e-01 1.75551012e-01 4.51172054e-01 -7.46738017e-01
4.74703878e-01 1.01603115e+00 9.07212615e-01 6.67804003e-01
-1.09980607e+00 -1.32066119e+00 4.57918912e-01 4.21925992e-01
-1.39493036e+00 -4.49120700e-01 5.19786060e-01 -7.25958049e-02
6.04761243e-01 3.82122904e-01 2.85869092e-01 1.00116932e+00
3.00588161e-01 9.44560111e-01 1.19446981e+00 5.51595129e-02
3.66628677e-01 5.15565634e-01 -1.56377777e-02 3.49563986e-01
4.91675168e-01 -3.90489288e-02 -7.24478185e-01 -5.47859490e-01
1.88311413e-01 7.27324128e-01 -3.81374061e-01 -5.97065747e-01
-9.02778506e-01 5.17741561e-01 4.91619080e-01 1.52027786e-01
-2.78689176e-01 -7.93534443e-02 5.83351433e-01 5.13945878e-01
6.52244270e-01 -2.32427716e-01 -1.03705776e+00 7.42264464e-02
-6.60025358e-01 9.89828259e-02 8.23375285e-01 9.17767942e-01
1.00373065e+00 -2.88433164e-01 -3.47529203e-01 5.50071955e-01
-1.44933939e-01 4.45678920e-01 6.90130889e-01 -7.02103555e-01
5.63302815e-01 8.59308243e-01 -1.45293981e-01 -8.43100309e-01
-9.02447179e-02 -8.26129258e-01 -1.18216825e+00 -8.85406435e-02
2.57687509e-01 -1.30168691e-01 -5.62957883e-01 1.93667972e+00
7.13492751e-01 1.82824850e-01 1.52458191e-01 7.34412551e-01
4.39735919e-01 4.83255744e-01 7.47313052e-02 -5.47387123e-01
1.12946987e+00 -9.56481516e-01 -7.97100782e-01 1.44909590e-01
5.41958034e-01 -5.63069105e-01 9.04173613e-01 4.62077349e-01
-7.57851243e-01 -1.80070445e-01 -7.31233656e-01 1.75862268e-01
-3.50157589e-01 -3.03769678e-01 7.25091040e-01 6.64895594e-01
-6.88677371e-01 4.48810726e-01 -6.04556859e-01 -4.55869399e-02
1.06495106e+00 4.80217040e-01 -4.65713054e-01 -3.58996660e-01
-1.15859771e+00 5.17881930e-01 9.24816802e-02 -2.92990714e-01
-1.12035847e+00 -1.05791605e+00 -2.45355591e-01 2.43984386e-01
5.74232697e-01 -8.74000907e-01 1.41807258e+00 -8.27976704e-01
-1.17468536e+00 4.47234273e-01 -5.67083880e-02 -4.62492347e-01
8.60147834e-01 -2.05384076e-01 -5.73877215e-01 -3.24279577e-01
-9.40572247e-02 -7.91616924e-03 8.87032330e-01 -1.24370861e+00
-9.83560264e-01 -8.08365762e-01 -1.05775096e-01 -4.04576585e-03
-1.03902018e+00 -3.66496235e-01 -4.31434363e-01 -5.19510806e-01
-2.36202508e-01 -5.44887304e-01 1.47089045e-02 2.59340316e-01
-2.34797820e-01 -2.16144815e-01 1.24569416e+00 -4.79326606e-01
1.34707701e+00 -2.22018123e+00 -3.73291343e-01 3.26124318e-02
5.32962143e-01 5.10803461e-01 -3.72062102e-02 2.78817803e-01
2.44941548e-01 -1.35968298e-01 -1.84852302e-01 -5.29969454e-01
-3.70410532e-02 4.84500587e-01 -4.73739237e-01 5.00381649e-01
-4.15868163e-01 6.07258618e-01 -7.70833313e-01 -3.50380301e-01
-2.48359412e-01 7.30946660e-01 -6.37978077e-01 1.68511674e-01
-1.75036445e-01 4.72153157e-01 -4.81265992e-01 6.46620035e-01
1.15865803e+00 -2.68364787e-01 2.60669798e-01 -1.02509998e-01
1.08064808e-01 1.35006443e-01 -1.23354089e+00 1.53892863e+00
-5.12176991e-01 -1.39116794e-01 2.69089967e-01 -8.02030802e-01
7.47669339e-01 4.12099242e-01 3.50351691e-01 -7.34136999e-01
2.21538514e-01 2.98406929e-01 -6.12517118e-01 -3.59759927e-01
3.51032754e-03 -4.50656675e-02 8.70027766e-02 6.54098094e-01
-9.15032253e-02 5.36085486e-01 -4.85191405e-01 2.19747826e-01
1.40659177e+00 -2.76931047e-01 -1.06614284e-01 -3.71399261e-02
5.17419636e-01 -2.63494819e-01 1.16281533e+00 6.04636192e-01
-3.52368981e-01 1.99800476e-01 2.16851234e-01 -8.33755493e-01
-5.69169104e-01 -9.70681608e-01 1.74642373e-02 1.19859159e+00
1.73501939e-01 -3.57676595e-01 -7.29277074e-01 -1.21810269e+00
5.37869453e-01 5.63571215e-01 -5.15557051e-01 -4.99298215e-01
-3.24515432e-01 -7.85131991e-01 4.67954844e-01 1.85569689e-01
8.33186984e-01 -6.75822198e-01 -3.24891746e-01 2.04840526e-01
-3.87085915e-01 -6.48882449e-01 -6.83102071e-01 1.49679616e-01
-9.03315842e-01 -1.23839283e+00 -5.29881656e-01 -5.39974332e-01
8.04719210e-01 6.01871490e-01 8.60130370e-01 -1.69346165e-02
-2.80709594e-01 1.82403222e-01 -1.03555009e-01 -4.73428845e-01
-1.58343688e-02 2.73076504e-01 2.14287028e-01 4.05559480e-01
4.14916158e-01 -7.88872123e-01 -7.92997301e-01 4.54084039e-01
-1.09314585e+00 -2.24041507e-01 5.51732361e-01 9.03246045e-01
5.60345948e-01 4.55719560e-01 7.72977293e-01 -1.23021066e+00
3.56713951e-01 -7.57846653e-01 -3.99813056e-01 4.01295274e-01
-1.22987378e+00 3.52341086e-02 9.48050320e-01 -6.88116610e-01
-1.29092872e+00 6.10435344e-02 3.28489035e-01 -6.00126624e-01
2.74529040e-01 1.06539810e-02 -4.30495948e-01 -1.75471038e-01
4.56122488e-01 4.17424798e-01 2.41038531e-01 -8.81181359e-01
2.76383966e-01 1.05683041e+00 5.26010334e-01 -5.55958807e-01
7.16821551e-01 5.81585467e-01 -3.96420270e-01 5.68780825e-02
-8.57252955e-01 -1.32978752e-01 -1.37796164e-01 2.38333195e-02
1.11183606e-01 -1.24616444e+00 -1.06465709e+00 8.81225049e-01
-1.01519883e+00 -1.66798353e-01 -4.57755685e-01 2.23747194e-01
-9.42177549e-02 1.24278583e-01 -4.82723862e-01 -6.77890897e-01
-8.88268471e-01 -7.62711883e-01 4.66199905e-01 3.07391554e-01
2.53121912e-01 -5.80590963e-01 -7.48240277e-02 6.20207071e-01
9.37047839e-01 3.39743942e-02 8.22932065e-01 -6.30558133e-01
-7.41680145e-01 -2.36911640e-01 -1.61181167e-01 5.04444420e-01
1.71535522e-01 -5.47328055e-01 -1.19555426e+00 -8.17976654e-01
1.82584971e-01 -3.81411761e-01 7.87481844e-01 -2.77846694e-01
1.60565019e+00 -9.78010356e-01 -5.28418779e-01 7.17775166e-01
1.35002720e+00 -4.15647700e-02 3.94445419e-01 2.01646373e-01
4.75688517e-01 2.29733452e-01 4.41192538e-01 9.58543181e-01
7.49136746e-01 3.30657959e-01 5.52570760e-01 2.09795073e-01
-1.18586123e-01 -4.87891614e-01 2.99378306e-01 9.05219197e-01
5.07130444e-01 -3.53128165e-02 -5.24289429e-01 7.32443750e-01
-1.98296726e+00 -7.83249557e-01 9.96547937e-02 2.49125934e+00
1.27566028e+00 -4.94485460e-02 -1.45326272e-01 1.59377977e-01
7.41202295e-01 -1.56836525e-01 -1.10516977e+00 -4.89986800e-02
-1.35684192e-01 6.61771446e-02 7.18226671e-01 -2.68673263e-02
-7.48304963e-01 5.13502955e-01 5.25380182e+00 1.04077005e+00
-1.13802958e+00 7.95590162e-01 6.88652396e-01 -4.32710886e-01
-3.73280138e-01 -8.62439498e-02 -9.08668995e-01 8.01241994e-01
7.05363095e-01 -4.81698394e-01 6.56597018e-01 9.69469249e-01
-4.38478380e-01 1.20013997e-01 -8.94213557e-01 1.02343023e+00
-2.46242955e-02 -1.15169120e+00 1.00427754e-01 -1.39470145e-01
8.74732256e-01 8.65492150e-02 2.10385963e-01 5.77789247e-01
5.05867541e-01 -5.67309022e-01 4.81743515e-01 6.29277289e-01
9.05744016e-01 -1.09759057e+00 7.72817135e-01 7.69131541e-01
-9.64145601e-01 -5.70886850e-01 -4.41640139e-01 -7.94518813e-02
-4.20259744e-01 1.12915468e+00 -4.77111101e-01 5.94095409e-01
1.17192352e+00 3.91240686e-01 -4.87625569e-01 1.03350973e+00
1.29390612e-01 5.65840006e-01 -4.21480864e-01 3.23436618e-01
-3.96540314e-01 3.92008841e-01 2.39242360e-01 6.36099279e-01
3.07035983e-01 1.97282564e-02 1.08070202e-01 3.71761233e-01
-5.92978656e-01 6.28595352e-02 -3.85406047e-01 4.93441463e-01
9.53036726e-01 1.20812631e+00 5.78209572e-02 -3.85217279e-01
-2.87550122e-01 1.15135646e+00 6.24078572e-01 1.79795519e-01
-6.65655673e-01 -5.17414153e-01 9.39068496e-01 1.21866360e-01
2.95701236e-01 4.34033990e-01 -1.66761994e-01 -1.36460972e+00
3.70500594e-01 -1.02267277e+00 9.03344452e-01 -2.85419285e-01
-1.68387079e+00 6.00598097e-01 -5.81341088e-01 -1.00067401e+00
4.03632939e-01 1.85756490e-01 -4.81915146e-01 6.25910997e-01
-1.69203532e+00 -1.14125228e+00 -3.38494182e-01 1.11249137e+00
1.98870301e-02 -2.15026915e-01 9.11525071e-01 8.58981907e-01
-7.38017201e-01 1.28021848e+00 5.58576524e-01 -2.96242654e-01
1.19941938e+00 -7.04610169e-01 -4.61182818e-02 5.44757426e-01
-3.56654346e-01 5.71120441e-01 3.23426872e-01 -5.77115297e-01
-1.68129289e+00 -1.64067566e+00 1.11211669e+00 -4.63443667e-01
1.52035192e-01 -4.16637659e-01 -1.10854900e+00 7.40606070e-01
-1.05649017e-01 5.16348422e-01 7.25012898e-01 -1.34412810e-01
-8.52492929e-01 -1.02727914e+00 -1.76852083e+00 1.46743685e-01
1.10540199e+00 -4.13421839e-01 -1.88067570e-01 4.88605857e-01
1.00400066e+00 -2.13421851e-01 -8.64718199e-01 4.14888173e-01
3.68787259e-01 -1.00404572e+00 7.72669375e-01 -6.08701169e-01
-1.10861942e-01 -3.28694642e-01 -5.85639738e-02 -1.16803610e+00
-3.52814257e-01 -8.97535384e-01 -5.68663239e-01 1.68127465e+00
9.11417603e-02 -1.16319835e+00 8.37559640e-01 8.09256494e-01
2.16112852e-01 -6.69542253e-01 -1.29332817e+00 -8.74175906e-01
-1.17113344e-01 -6.00806028e-02 1.34115803e+00 1.20557714e+00
-1.30991682e-01 3.61751318e-02 -5.62049627e-01 1.98170125e-01
9.48292911e-01 2.65971541e-01 1.01888728e+00 -1.32537746e+00
-1.42270729e-01 -1.42131727e-02 -1.13319971e-01 -6.99442983e-01
1.77056380e-02 -1.00020719e+00 -2.76042610e-01 -1.09055221e+00
5.57933211e-01 -8.32141817e-01 -8.13476384e-01 1.11922741e+00
-2.08772212e-01 5.72382323e-02 8.90560076e-02 4.75028217e-01
-9.61102188e-01 9.34078097e-01 1.10164738e+00 -1.49386197e-01
-6.66065663e-02 1.77048743e-01 -9.63073194e-01 2.68141329e-01
7.79636741e-01 -8.59582543e-01 -2.86787391e-01 -6.18914604e-01
7.91623432e-04 -1.73551992e-01 1.53798446e-01 -1.07216072e+00
6.07024610e-01 -4.72246744e-02 4.13525611e-01 -4.33157772e-01
-4.73343171e-02 -1.31514597e+00 5.42830646e-01 7.67169356e-01
-7.65102059e-02 -1.67672515e-01 5.97811230e-02 8.02138090e-01
-1.93930492e-02 4.52840745e-01 6.91952169e-01 7.60785937e-02
-1.74157664e-01 7.34401464e-01 5.85851744e-02 4.71645333e-02
1.36804223e+00 2.98565716e-01 -6.67015433e-01 -1.40729785e-01
-3.04863483e-01 5.76184392e-01 4.75279450e-01 4.41500485e-01
3.09587538e-01 -1.38060427e+00 -4.90163416e-01 5.75951636e-01
8.72397348e-02 1.67488009e-01 6.33957028e-01 7.68085420e-01
8.88497606e-02 6.49182424e-02 -5.12364320e-02 -2.23196685e-01
-1.26214814e+00 9.67160881e-01 1.32732138e-01 -3.12031716e-01
-5.67476392e-01 8.95787179e-01 7.15666339e-02 -8.54708672e-01
8.24957073e-01 6.64342120e-02 3.50801110e-01 -9.47024487e-03
8.17862153e-01 5.66569567e-01 3.43843520e-01 -1.13010488e-01
-4.48992908e-01 6.81181699e-02 -4.57778364e-01 7.01780796e-01
1.46947873e+00 -3.79336327e-01 -3.20089221e-01 2.70845503e-01
1.17262363e+00 -3.62110734e-02 -1.43937695e+00 -9.39053118e-01
-4.68404412e-01 -7.02141643e-01 1.21144083e-04 -1.10117018e+00
-1.61070871e+00 7.41887927e-01 8.85955930e-01 -2.52752304e-01
1.37672961e+00 -3.75276923e-01 1.43758917e+00 1.45863846e-01
9.45432723e-01 -9.37133253e-01 -2.85448432e-01 2.46452034e-01
3.87506068e-01 -1.04139650e+00 -1.54961506e-02 -2.06163749e-01
-4.53866839e-01 7.16022491e-01 8.47801149e-01 4.58744198e-01
8.51039529e-01 2.60734320e-01 5.56530356e-02 1.48188859e-01
-1.18215680e+00 4.66445863e-01 -3.57066482e-01 4.75164175e-01
-2.59603649e-01 1.32763535e-01 -2.61850208e-01 1.22633529e+00
1.12273484e-01 5.53774953e-01 9.91120040e-02 1.08240891e+00
-1.97201893e-01 -1.13933432e+00 -4.40921426e-01 6.08386099e-01
-7.15117276e-01 2.36490533e-01 -4.88394313e-02 9.75724310e-02
5.53642392e-01 8.35468233e-01 -8.01279321e-02 -6.37853086e-01
3.51006866e-01 1.48409128e-01 3.00925542e-02 -1.65668577e-01
-9.28562164e-01 -3.87500584e-01 -4.40115333e-01 -6.41611755e-01
2.98783015e-02 -3.38764399e-01 -1.14732635e+00 -7.93321848e-01
-3.24831128e-01 4.08077925e-01 5.27200162e-01 6.01768613e-01
1.07971108e+00 3.79256546e-01 1.07334971e+00 -9.70607921e-02
-1.19332767e+00 -7.15394139e-01 -6.62769139e-01 4.04709637e-01
3.81210148e-01 -5.53186357e-01 -5.40203571e-01 -2.22570673e-01] | [5.8534932136535645, 6.328486442565918] |
b8a24e15-2acb-447d-85ca-871698ae10fc | multipar-supervised-irregular-tensor | 2208.00993 | null | https://arxiv.org/abs/2208.00993v2 | https://arxiv.org/pdf/2208.00993v2.pdf | MULTIPAR: Supervised Irregular Tensor Factorization with Multi-task Learning | Tensor factorization has received increasing interest due to its intrinsic ability to capture latent factors in multi-dimensional data with many applications such as recommender systems and Electronic Health Records (EHR) mining. PARAFAC2 and its variants have been proposed to address irregular tensors where one of the tensor modes is not aligned, e.g., different users in recommender systems or patients in EHRs may have different length of records. PARAFAC2 has been successfully applied on EHRs for extracting meaningful medical concepts (phenotypes). Despite recent advancements, current models' predictability and interpretability are not satisfactory, which limits its utility for downstream analysis. In this paper, we propose MULTIPAR: a supervised irregular tensor factorization with multi-task learning. MULTIPAR is flexible to incorporate both static (e.g. in-hospital mortality prediction) and continuous or dynamic (e.g. the need for ventilation) tasks. By supervising the tensor factorization with downstream prediction tasks and leveraging information from multiple related predictive tasks, MULTIPAR can yield not only more meaningful phenotypes but also better predictive performance for downstream tasks. We conduct extensive experiments on two real-world temporal EHR datasets to demonstrate that MULTIPAR is scalable and achieves better tensor fit with more meaningful subgroups and stronger predictive performance compared to existing state-of-the-art methods. | ['Sivasubramanium Bhavani', 'Xiaoqian Jiang', 'Joyce C Ho', 'Li Xiong', 'Jian Lou', 'Yifei Ren'] | 2022-08-01 | null | null | null | null | ['mortality-prediction'] | ['medical'] | [-3.82924139e-01 -4.56270278e-01 -2.35175744e-01 -2.84298211e-01
-5.08124411e-01 -4.09740150e-01 -1.08232811e-01 3.32007319e-01
1.91489860e-01 3.85890096e-01 8.21514547e-01 -4.54271972e-01
-8.25176060e-01 -4.62963551e-01 -1.95481023e-03 -7.08735466e-01
-6.26426995e-01 4.64214474e-01 -2.05895498e-01 -1.89640790e-01
-2.39722997e-01 1.87365204e-01 -9.49469626e-01 6.78786457e-01
9.42469120e-01 8.60709310e-01 -1.27608001e-01 4.56515819e-01
2.89994866e-01 6.86071992e-01 -7.83761963e-02 -3.78548503e-01
1.71143129e-01 2.41681755e-01 -5.83412886e-01 -2.61132531e-02
-1.33355260e-01 -3.14186126e-01 -4.91880119e-01 4.20680702e-01
4.03392196e-01 -1.92078128e-01 5.33296824e-01 -1.21184742e+00
-7.33243644e-01 5.89230597e-01 -5.94938874e-01 3.24168622e-01
2.19182551e-01 3.31675448e-02 1.20497465e+00 -8.69680643e-01
1.14788227e-01 1.12729847e+00 8.51047277e-01 9.80100781e-02
-1.43146241e+00 -5.01233757e-01 1.77755371e-01 1.44394696e-01
-1.20435238e+00 -7.11795241e-02 4.90261286e-01 -8.29587400e-01
6.99906468e-01 6.18833959e-01 6.00805283e-01 1.01192117e+00
4.14710134e-01 7.33030379e-01 8.84375513e-01 4.36796576e-01
-1.28730282e-01 -3.13366830e-01 3.59240592e-01 5.63427925e-01
2.21520022e-01 -4.34463248e-02 -4.04860824e-01 -1.05024111e+00
6.90242946e-01 1.01867926e+00 -3.25200617e-01 4.22869846e-02
-2.04573274e+00 5.75805664e-01 3.05920303e-01 2.52949357e-01
-9.12065744e-01 -1.23308480e-01 5.41391909e-01 2.91495442e-01
5.55610359e-01 4.11616981e-01 -8.32591832e-01 -1.33427322e-01
-7.25676000e-01 1.14177465e-01 4.87698853e-01 7.05926836e-01
3.08639765e-01 -3.15034725e-02 -4.17497933e-01 9.00196493e-01
2.82758713e-01 4.42109525e-01 6.74345851e-01 -6.42791152e-01
4.96419102e-01 8.73317003e-01 1.54273938e-02 -1.36290741e+00
-8.11294258e-01 -6.40128374e-01 -1.47329628e+00 -1.08529198e+00
1.49428502e-01 -2.40396813e-01 -6.24305427e-01 1.37632465e+00
3.34469795e-01 6.73799634e-01 1.66168548e-02 8.69868279e-01
5.98053575e-01 4.47872430e-01 1.79831326e-01 -4.42988634e-01
1.81612539e+00 -6.58850789e-01 -7.20195711e-01 3.89647037e-01
1.03549278e+00 -7.55535364e-01 6.44365430e-01 6.79578185e-01
-5.12887239e-01 -2.91240036e-01 -3.68043035e-01 8.76671225e-02
1.03488527e-01 4.69080627e-01 1.20494664e+00 3.05497319e-01
-7.38960803e-01 5.62608361e-01 -1.24429381e+00 -1.87114090e-01
3.08803201e-01 3.33988100e-01 -4.11185414e-01 -3.94141883e-01
-1.07348537e+00 1.20981731e-01 -3.21213691e-03 2.13667959e-01
-6.22369945e-01 -1.05733573e+00 -4.15213287e-01 1.22210413e-01
2.52692133e-01 -1.31416023e+00 7.66519427e-01 -1.59690529e-01
-9.14008915e-01 2.35334747e-02 -2.06417292e-01 -1.10994481e-01
3.53412665e-02 -3.38695049e-01 -7.44240463e-01 9.04279649e-02
2.05630109e-01 -3.24689709e-02 6.34873927e-01 -6.73870206e-01
-4.23794150e-01 -7.17450678e-01 -5.62336249e-03 -8.70287791e-02
-8.04535151e-01 -5.23057468e-02 -2.78106004e-01 -1.00658607e+00
5.34752250e-01 -1.37538445e+00 -5.31403780e-01 -3.19064349e-01
-5.21055162e-01 -3.31465364e-01 7.22900748e-01 -8.11864138e-01
1.60295725e+00 -2.08065414e+00 3.86114389e-01 1.35314316e-01
9.39136565e-01 -6.18263669e-02 -3.19240838e-02 6.04993582e-01
-3.62867057e-01 2.74983287e-01 1.69275388e-01 -2.31666401e-01
-5.04699230e-01 3.77682030e-01 -2.50805110e-01 4.41970497e-01
3.26650888e-02 8.63093615e-01 -1.02132213e+00 -3.51035476e-01
-1.66726962e-01 5.46447814e-01 -7.47874141e-01 1.85195968e-01
1.80538595e-01 7.77304113e-01 -9.78537679e-01 8.21980894e-01
5.10392904e-01 -9.23271954e-01 3.43081415e-01 -5.37641406e-01
3.32507156e-02 8.99287909e-02 -8.38265717e-01 1.49319077e+00
-2.01898411e-01 -5.64017445e-02 -3.13611507e-01 -1.07987595e+00
5.93453288e-01 7.48351693e-01 1.49350595e+00 -3.30140591e-02
-1.79293618e-01 5.87717332e-02 3.85620624e-01 -8.53745043e-01
2.86369503e-01 -1.70546472e-01 4.77318726e-02 5.07146537e-01
-3.02171618e-01 7.33203351e-01 -3.32013257e-02 3.58153015e-01
1.43795991e+00 -2.75423855e-01 1.36494368e-01 -3.21788460e-01
3.22591513e-01 -3.20711099e-02 9.45605040e-01 1.48797423e-01
1.07817106e-01 4.10626322e-01 5.03055036e-01 -6.65100217e-01
-6.29872143e-01 -7.68503487e-01 -3.59287709e-01 1.04056776e+00
-2.06873447e-01 -9.70580220e-01 1.05067581e-01 -5.86393893e-01
4.02249813e-01 7.86942467e-02 -7.72341847e-01 -8.81477073e-02
-3.40303361e-01 -1.36741555e+00 3.94803196e-01 6.25695646e-01
-1.67937562e-01 -1.89600781e-01 -5.47648408e-02 3.82084787e-01
-6.57781303e-01 -1.05890286e+00 -5.31394839e-01 -2.81085700e-01
-1.53437114e+00 -1.12532187e+00 -7.02287436e-01 -1.77227125e-01
6.91280007e-01 7.71540701e-01 9.69862103e-01 1.17909536e-01
4.42869551e-02 5.27213693e-01 -5.65365016e-01 -9.38967988e-02
5.74289113e-02 7.21348897e-02 5.40558040e-01 4.92916256e-01
2.37007514e-01 -6.11775279e-01 -9.82879996e-01 6.94409788e-01
-9.85803306e-01 1.67303637e-01 6.50341094e-01 1.07925665e+00
6.58596277e-01 1.71442300e-01 6.33615613e-01 -8.51367652e-01
7.78995514e-01 -1.20349467e+00 9.76384953e-02 2.69710541e-01
-9.19767499e-01 4.75710221e-02 6.79881334e-01 -5.52498460e-01
-6.35231078e-01 -3.40918720e-01 2.21028939e-01 -7.60948062e-01
1.34149477e-01 1.21148288e+00 3.97437215e-01 4.08023387e-01
4.25338179e-01 7.07078055e-02 -1.02898315e-01 -8.09907258e-01
-1.59410477e-01 5.71128488e-01 -2.55706012e-01 -7.56800354e-01
7.19385743e-01 4.63523477e-01 2.09540367e-01 -5.67998171e-01
-8.83615792e-01 -9.55468953e-01 -5.57277262e-01 1.48753658e-01
8.03438544e-01 -1.23441637e+00 -9.77427065e-01 1.37367249e-01
-8.00384760e-01 2.59551525e-01 1.55791342e-01 8.98114204e-01
-7.38085667e-03 4.76124674e-01 -8.40462208e-01 -5.38738847e-01
-5.02633870e-01 -1.14874148e+00 1.03506792e+00 -3.04730594e-01
-9.34545919e-02 -1.01191461e+00 6.93482533e-02 6.47641420e-01
3.11866432e-01 2.47610196e-01 1.11829519e+00 -6.23479068e-01
-4.26839143e-01 -1.19937479e-01 -2.35172227e-01 5.14382962e-03
5.05958438e-01 1.92800444e-03 -4.73497868e-01 -4.07854378e-01
-1.78577542e-01 1.73273236e-02 6.91142857e-01 3.43879431e-01
1.37862313e+00 -6.22742534e-01 -6.39690995e-01 6.59623384e-01
9.95966792e-01 1.82797182e-02 2.04429701e-01 -6.72322586e-02
1.11648595e+00 5.50709069e-01 7.26933002e-01 7.78032601e-01
1.00771475e+00 6.67794108e-01 5.30396178e-02 -1.35097519e-01
4.50297028e-01 2.82430556e-02 1.72207698e-01 1.53841269e+00
-5.81721544e-01 2.33710676e-01 -1.24172544e+00 4.13981229e-01
-2.29716516e+00 -6.84556484e-01 -6.23305380e-01 2.00813842e+00
6.27559900e-01 -3.96784574e-01 2.97174931e-01 -4.30425033e-02
2.79336661e-01 -1.42520174e-01 -6.24198914e-01 7.60406256e-02
4.51298691e-02 -3.20937127e-01 2.81314552e-01 -2.54168063e-01
-9.31863010e-01 2.70112306e-01 5.58280611e+00 3.82244378e-01
-1.18819726e+00 3.78924966e-01 8.44385386e-01 -1.15832038e-01
-4.76816416e-01 -8.23465511e-02 -3.18366319e-01 4.53327417e-01
8.86852145e-01 -6.27434859e-03 3.79085600e-01 6.56754732e-01
6.53193235e-01 4.56913173e-01 -1.00166750e+00 1.08684862e+00
-3.64749432e-01 -1.19730496e+00 8.08497220e-02 4.85275209e-01
7.64326692e-01 3.38284045e-01 9.97403935e-02 2.63672471e-01
7.61903450e-02 -7.51673579e-01 5.49777523e-02 8.40227544e-01
8.48561704e-01 -3.70822459e-01 8.29951763e-01 2.65024453e-01
-1.38154817e+00 -3.96562427e-01 -2.85252124e-01 2.62308046e-02
7.73008093e-02 1.08167303e+00 -9.27166760e-01 1.01739335e+00
8.28283131e-01 1.30758119e+00 -4.51411098e-01 8.01543236e-01
3.45885873e-01 9.24345434e-01 -1.58179954e-01 3.49361807e-01
5.47225326e-02 -4.84938323e-01 4.97964382e-01 9.17718828e-01
6.22559428e-01 4.88559455e-01 7.40459383e-01 3.54532987e-01
2.53428578e-01 4.54551160e-01 -4.08345878e-01 -3.72571081e-01
1.67419717e-01 1.42542481e+00 -3.03476244e-01 -4.10435170e-01
-6.40803337e-01 6.05360150e-01 -5.69127575e-02 5.22680879e-01
-7.49889314e-01 3.09569478e-01 1.17347884e+00 4.33596432e-01
-1.48846935e-02 -5.69112897e-01 -4.01593119e-01 -1.91531134e+00
-2.28931040e-01 -8.47192109e-01 8.67223918e-01 -4.13910449e-01
-1.68222642e+00 6.71100259e-01 -1.95159048e-01 -1.64717269e+00
-6.42167851e-02 -4.35794443e-01 -1.75211489e-01 7.51106501e-01
-1.39716482e+00 -1.32109833e+00 -2.71736979e-01 1.00068665e+00
1.88557237e-01 -1.96150526e-01 1.05850244e+00 7.02313423e-01
-9.54033792e-01 4.16321278e-01 2.68284440e-01 1.44952357e-01
8.68889391e-01 -1.17399311e+00 -9.83123779e-02 5.09128153e-01
-2.29532570e-01 1.22572875e+00 3.15534264e-01 -7.61949182e-01
-1.95687020e+00 -1.36197960e+00 6.48426950e-01 -4.87619847e-01
1.01125109e+00 5.17751426e-02 -8.72356951e-01 6.69428587e-01
-4.70572591e-01 2.20782489e-01 1.58586764e+00 8.06959271e-01
-5.27392149e-01 -4.00811434e-01 -7.59106100e-01 4.87715393e-01
9.59334910e-01 -3.09323698e-01 -1.59454077e-01 6.13740265e-01
6.83387816e-01 -1.70792803e-01 -1.85971260e+00 6.97697997e-01
6.29692435e-01 -6.24220133e-01 1.10347009e+00 -1.18817031e+00
4.65897352e-01 -3.71484786e-01 -1.09282091e-01 -1.29246938e+00
-1.02391553e+00 -6.47960961e-01 -4.87715989e-01 8.00893128e-01
2.91128457e-01 -7.91423678e-01 3.81134719e-01 7.67120242e-01
-1.50843814e-01 -1.16446674e+00 -7.93579638e-01 -3.94412488e-01
-4.23681051e-01 -4.98302966e-01 7.24980831e-01 1.47654724e+00
1.75602227e-01 5.96306860e-01 -8.36066782e-01 4.63974059e-01
4.37976450e-01 4.69220459e-01 6.33392215e-01 -1.57334161e+00
-6.05479777e-01 -6.55062795e-02 -3.58778000e-01 -9.10709977e-01
-3.70902896e-01 -1.09238625e+00 -7.59068727e-01 -1.47893643e+00
4.98016477e-01 -1.01578605e+00 -7.22811699e-01 9.06487882e-01
-5.15781820e-01 -1.91516113e-02 -5.09234145e-02 8.13730419e-01
-4.69749868e-01 5.26407480e-01 1.36789119e+00 -1.07000664e-01
-3.10034007e-01 1.90029502e-01 -1.04408085e+00 4.05676037e-01
6.67656481e-01 -4.77121204e-01 -5.62680364e-01 -4.37419027e-01
4.47137922e-01 5.73959947e-01 9.56435800e-02 -6.21825635e-01
-6.99704438e-02 -3.67037237e-01 2.79420584e-01 -3.05374056e-01
3.21687013e-01 -8.71726513e-01 4.98157054e-01 5.78570485e-01
-4.32231873e-02 5.40197551e-01 -7.47702643e-02 9.17070985e-01
-1.96018606e-01 6.27691686e-01 6.56618550e-03 7.36343935e-02
-9.19588506e-02 1.05362570e+00 -8.77635106e-02 -3.84534299e-01
8.08732152e-01 1.50651768e-01 -3.72510195e-01 -1.73593134e-01
-9.73432302e-01 5.18317044e-01 2.19194535e-02 5.81960797e-01
7.93889999e-01 -1.41580367e+00 -1.05903959e+00 9.70927626e-03
5.25789320e-01 -7.66242370e-02 6.67667508e-01 1.73278785e+00
3.72073166e-02 5.30578434e-01 2.22679198e-01 -9.05864894e-01
-1.28981709e+00 7.98772871e-01 -6.66288659e-02 -5.86904347e-01
-7.50044107e-01 2.87290096e-01 4.41795290e-01 -2.84056276e-01
-2.54687577e-01 -5.80573797e-01 -3.72721523e-01 1.88974679e-01
5.07904887e-01 4.38996166e-01 5.18285818e-02 -6.41075730e-01
-3.75661939e-01 3.73699874e-01 -1.29944503e-01 5.72533846e-01
1.75502264e+00 -1.00499853e-01 -4.78399336e-01 5.66776693e-01
1.05843616e+00 6.73685074e-02 -5.68159878e-01 -4.15116519e-01
-5.67046134e-03 -5.51276982e-01 2.04052940e-01 -6.08394861e-01
-1.26555610e+00 6.92461669e-01 3.78725857e-01 3.52132320e-01
1.23526216e+00 -1.28259510e-01 1.14032817e+00 2.86531419e-01
4.17124331e-01 -4.04129267e-01 8.54506120e-02 3.62701654e-01
7.60798514e-01 -1.17655361e+00 9.50061381e-02 -4.25741732e-01
-9.45092261e-01 1.24876010e+00 2.11926699e-01 2.34009981e-01
1.18719041e+00 -4.83804196e-02 2.01971196e-02 -4.71665382e-01
-1.17865872e+00 1.96891993e-01 6.95982218e-01 1.66270956e-01
7.70767510e-01 6.81391120e-01 -3.46174419e-01 9.41879690e-01
2.35387702e-02 1.10383227e-01 2.93915212e-01 5.01445234e-01
2.43477106e-01 -1.13928676e+00 -4.38877344e-01 1.25124216e+00
-7.07361341e-01 -3.05435628e-01 2.03885421e-01 3.44527476e-02
1.85620219e-01 9.83970404e-01 -3.95933956e-01 -8.25586677e-01
2.30919406e-01 -8.93069953e-02 -1.13828570e-01 -7.46792376e-01
-5.73058307e-01 5.35845697e-01 -1.49165676e-03 -7.87884951e-01
-4.70155686e-01 -9.89208817e-01 -1.04201853e+00 -3.48435163e-01
-2.14198872e-01 2.82076925e-01 3.52477789e-01 8.16920757e-01
1.03147161e+00 6.94304109e-01 8.71703207e-01 -3.02460551e-01
-5.20426452e-01 -1.10913360e+00 -6.46939695e-01 4.65221196e-01
3.47803295e-01 -7.97033489e-01 1.46034928e-02 1.58420369e-01] | [6.465212821960449, 5.963247776031494] |
43a6fab4-d34f-497c-86ab-7a4796296646 | grammatical-error-detection-and-correction | null | null | https://aclanthology.org/W14-1710 | https://aclanthology.org/W14-1710.pdf | Grammatical Error Detection and Correction using a Single Maximum Entropy Model | null | ['Zhongye Jia', 'Peilu Wang', 'Hai Zhao'] | 2014-06-01 | null | null | null | ws-2014-6 | ['grammatical-error-detection'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.423455715179443, 3.640964984893799] |
dcbc8545-083f-4623-a4dc-0a728f3b2406 | saliency-guided-local-full-reference-image | null | null | https://www.mdpi.com/2624-6120/3/3/28 | https://www.mdpi.com/2624-6120/3/3/28 | Saliency-Guided Local Full-Reference Image Quality Assessment | Research and development of image quality assessment (IQA) algorithms have been in the focus of the computer vision and image processing community for decades. The intent of IQA methods is to estimate the perceptual quality of digital images correlating as high as possible with human judgements. Full-reference image quality assessment algorithms, which have full access to the distortion-free images, usually contain two phases: local image quality estimation and pooling. Previous works have utilized visual saliency in the final pooling stage. In addition to this, visual saliency was utilized as weights in the weighted averaging of local image quality scores, emphasizing image regions that are salient to human observers. In contrast to this common practice, visual saliency is applied in the computation of local image quality in this study, based on the observation that local image quality is determined both by local image degradation and visual saliency simultaneously. Experimental results on KADID-10k, TID2013, TID2008, and CSIQ have shown that the proposed method was able to improve the state-of-the-art’s performance at low computational costs. | ['Domonkos Varga'] | 2022-06-29 | null | null | null | signals-2022-6 | ['image-quality-estimation'] | ['computer-vision'] | [ 2.55908400e-01 -3.20640117e-01 -3.06673581e-03 -1.61067173e-01
-5.27124166e-01 -1.88218970e-02 3.55647177e-01 5.11842251e-01
-3.53971452e-01 4.64366674e-01 8.29669684e-02 -2.31724270e-02
-1.67497024e-01 -7.20018804e-01 -1.88932821e-01 -6.58804059e-01
1.50654437e-02 -4.36732918e-01 5.84478021e-01 -1.16943397e-01
7.69417048e-01 4.63734359e-01 -1.87134778e+00 8.56695622e-02
1.19966924e+00 1.16824424e+00 5.41034281e-01 6.49079263e-01
1.47138536e-02 7.00317383e-01 -6.66481614e-01 -1.89276949e-01
9.58166942e-02 -5.25479794e-01 -7.18055964e-01 3.88376534e-01
2.37741798e-01 -1.77075744e-01 2.76121777e-02 1.47121954e+00
6.23293400e-01 3.07244267e-02 5.19913435e-01 -1.35847890e+00
-7.96530783e-01 -9.38588232e-02 -7.23398685e-01 7.93326318e-01
2.96222836e-01 1.26692489e-01 7.91057944e-01 -1.09915006e+00
2.78606176e-01 1.17393816e+00 1.62680507e-01 -1.17428705e-01
-9.56130981e-01 -4.45140928e-01 -5.81752844e-02 8.93578053e-01
-1.41078329e+00 -3.06973547e-01 9.88891602e-01 -3.74188125e-01
6.60108387e-01 1.38339862e-01 7.11527050e-01 1.38087854e-01
6.00400209e-01 6.29058719e-01 1.37526357e+00 -6.73147738e-01
4.42710429e-01 2.02960774e-01 2.35760808e-02 5.40075123e-01
2.98637778e-01 1.26940995e-01 -6.49502099e-01 3.47748213e-02
6.71615958e-01 -4.83143963e-02 -2.55146056e-01 -4.39517081e-01
-1.05792284e+00 6.26546264e-01 7.66985774e-01 4.05114859e-01
-7.70111740e-01 -2.87918627e-01 2.02849269e-01 1.94967419e-01
4.05101746e-01 3.08762461e-01 -1.05964420e-02 1.17949590e-01
-1.11347234e+00 1.07008610e-02 1.22678883e-01 5.72559536e-01
7.98816144e-01 3.43445241e-02 -2.93408513e-01 7.31452346e-01
2.11813509e-01 6.94135189e-01 5.62071681e-01 -9.31431055e-01
1.16553716e-01 7.45550334e-01 3.44091564e-01 -1.59861934e+00
-1.30562782e-01 -4.23800558e-01 -7.34188497e-01 8.72235537e-01
2.23942667e-01 4.31726277e-01 -9.48139668e-01 1.30455554e+00
2.30306879e-01 -5.03307767e-02 5.57423942e-02 1.23221993e+00
5.38981855e-01 6.50920510e-01 2.78092414e-01 -3.75580609e-01
1.36180091e+00 -8.90520096e-01 -9.09151673e-01 -1.47656068e-01
-7.12331608e-02 -1.06504178e+00 1.20380676e+00 3.55841219e-01
-1.34985673e+00 -1.04822469e+00 -1.36632299e+00 9.47937891e-02
-4.00564373e-01 3.59597355e-02 2.01856658e-01 5.94640911e-01
-1.26869059e+00 4.25013483e-01 -4.96022940e-01 -3.79370660e-01
3.39445263e-01 -3.77469398e-02 -1.37700453e-01 -1.72882959e-01
-1.25143898e+00 1.35356569e+00 1.43418044e-01 4.43042703e-02
-9.19390857e-01 -4.95902002e-01 -6.52995527e-01 1.28117114e-01
2.11136773e-01 -3.77106547e-01 9.32113588e-01 -1.27960980e+00
-1.20486653e+00 7.78289318e-01 -2.67057925e-01 -4.44825262e-01
2.16250673e-01 -1.29348263e-01 -6.70863926e-01 6.27240181e-01
2.35896766e-01 6.50760055e-01 9.26476419e-01 -1.26630688e+00
-8.50157559e-01 -2.94954330e-01 5.87582067e-02 5.49957573e-01
-6.47110045e-02 2.36000776e-01 -4.75836337e-01 -5.63733459e-01
2.72786051e-01 -5.09141922e-01 -3.60554717e-02 2.11672693e-01
1.27749443e-01 -6.21100552e-02 8.53478789e-01 -8.41467202e-01
1.20874894e+00 -2.02433181e+00 -1.05645061e-01 1.93224788e-01
1.44474104e-01 5.01979470e-01 -4.75433059e-02 7.83483759e-02
8.18183720e-02 4.31001671e-02 -2.08041161e-01 8.01459774e-02
-2.45437801e-01 -2.38786057e-01 3.34013067e-02 6.89355791e-01
4.14362013e-01 6.91015124e-01 -8.37696850e-01 -9.32404339e-01
6.47785366e-01 3.48135740e-01 -1.18141204e-01 2.63164997e-01
3.05635542e-01 3.46997045e-02 -2.54880548e-01 8.51466596e-01
8.82407844e-01 -2.65561461e-01 -1.65429637e-01 -3.90695006e-01
-4.35048580e-01 -1.45029038e-01 -1.13086891e+00 1.44440460e+00
-3.03070664e-01 7.17832923e-01 -3.07825729e-02 -7.31296480e-01
9.46119726e-01 2.48902932e-01 4.38306779e-01 -1.36480844e+00
8.42963159e-02 2.06691936e-01 -3.71711999e-02 -5.11616647e-01
7.23232925e-01 8.58702734e-02 2.57716835e-01 1.14897400e-01
3.37151289e-02 -2.56280214e-01 3.18969727e-01 1.17616102e-01
5.66731870e-01 -1.94923319e-02 6.93442285e-01 -4.28015918e-01
8.11471343e-01 4.25650785e-03 5.34220815e-01 4.65679824e-01
-8.41187596e-01 6.75030351e-01 6.45752996e-02 -4.18680869e-02
-1.19757450e+00 -1.27402508e+00 -1.62830144e-01 6.61937892e-01
9.37511384e-01 -2.18473822e-02 -7.59598076e-01 -9.38361660e-02
-4.20710564e-01 4.57910776e-01 -4.58429128e-01 -2.53520131e-01
-2.05491021e-01 -5.04859924e-01 -9.75984037e-02 1.59151316e-01
9.83461082e-01 -1.47173917e+00 -1.09136343e+00 1.88145787e-01
-2.30622485e-01 -8.41174066e-01 -4.23422754e-01 -2.24265441e-01
-8.14453423e-01 -1.13817823e+00 -1.18362868e+00 -8.36353481e-01
7.00765014e-01 7.47514307e-01 1.07533073e+00 2.62059152e-01
-3.69395018e-01 3.50564755e-02 -5.10963082e-01 -3.69929045e-01
-1.16103262e-01 -3.48317027e-01 -1.76832974e-01 8.04551318e-02
1.77675888e-01 -2.44082525e-01 -9.00354564e-01 4.73070771e-01
-1.06347251e+00 1.15605637e-01 9.00663733e-01 5.70271492e-01
5.94198465e-01 4.16024268e-01 4.88856167e-01 -2.88871974e-01
6.61572099e-01 -2.02582523e-01 -6.22443438e-01 3.60987306e-01
-9.21254873e-01 -2.70628154e-01 2.72831529e-01 9.72993672e-03
-1.25425863e+00 -4.09269333e-01 2.11322874e-01 -2.85553992e-01
-8.03980529e-02 3.87067348e-01 -3.44971359e-01 -2.93553203e-01
5.04372895e-01 3.88868988e-01 -3.75540070e-02 -1.32481128e-01
1.09038271e-01 6.40579700e-01 6.63161755e-01 -3.85760819e-03
6.77093923e-01 2.95180023e-01 1.44034639e-01 -7.27667093e-01
-5.03532827e-01 -7.19853282e-01 -4.55438018e-01 -7.38215029e-01
9.32708025e-01 -9.38388884e-01 -5.24910867e-01 8.54330540e-01
-7.95703709e-01 1.79997534e-01 2.42891144e-02 4.61849511e-01
-4.00041014e-01 5.78016579e-01 -1.43462047e-01 -7.91872799e-01
-4.56451237e-01 -1.34822512e+00 7.18948126e-01 7.86749303e-01
1.35597680e-02 -7.82141626e-01 -3.76090556e-01 1.40172333e-01
6.73393130e-01 1.55344039e-01 7.28602886e-01 2.75874227e-01
-4.85517383e-01 2.76853945e-02 -6.46219611e-01 5.14274299e-01
3.87950420e-01 -1.26115039e-01 -8.15100610e-01 -1.78736210e-01
-1.41592289e-03 1.39112294e-01 3.89253497e-01 6.78221941e-01
8.21472883e-01 -4.56237830e-02 8.21629018e-02 1.30993292e-01
1.85428631e+00 4.59688097e-01 1.00885105e+00 4.85560745e-01
2.74551421e-01 4.43179518e-01 1.19959748e+00 2.66130626e-01
2.73245335e-01 8.47522557e-01 5.60559809e-01 -5.51570773e-01
-5.46546876e-01 1.68593638e-02 3.39896202e-01 7.09033370e-01
1.63055897e-01 -1.02319732e-01 -7.66503990e-01 9.33167577e-01
-1.60345984e+00 -9.08132672e-01 -1.32286981e-01 2.17043066e+00
6.58391297e-01 1.50945097e-01 -1.05644614e-01 6.65660501e-01
9.11044300e-01 1.23881258e-01 -4.43191886e-01 -4.83302444e-01
-2.18698218e-01 3.12254969e-02 5.26326776e-01 3.97541791e-01
-1.01443756e+00 6.55772150e-01 6.14572954e+00 8.52205753e-01
-1.18173790e+00 1.45907044e-01 7.43754029e-01 8.68896544e-02
-7.33510554e-02 -7.84485936e-02 -2.33151183e-01 8.03815663e-01
7.56640017e-01 -3.46114486e-01 3.09971333e-01 6.03206635e-01
5.36930263e-01 -8.00249040e-01 -3.50873560e-01 1.03886449e+00
2.25809470e-01 -9.10540998e-01 3.58153321e-02 -1.72964245e-01
8.52761805e-01 -4.38627332e-01 3.97662282e-01 -2.42557719e-01
-8.06894675e-02 -8.56873274e-01 9.07351434e-01 7.30781198e-01
4.51711386e-01 -9.54255700e-01 1.01506591e+00 -1.94398351e-02
-1.12236679e+00 -9.26245600e-02 -4.53862339e-01 -6.54723644e-02
2.17229411e-01 8.52720618e-01 -3.56201619e-01 5.54656625e-01
1.18878818e+00 5.55632949e-01 -9.67719316e-01 1.41131330e+00
-3.10982704e-01 3.89662355e-01 1.40374586e-01 1.41273186e-01
1.20532811e-01 -1.22065991e-01 4.43844080e-01 8.16480935e-01
3.32852125e-01 1.18031383e-01 -3.06113530e-02 7.66755819e-01
2.49940053e-01 2.90986985e-01 -1.76579133e-01 2.14212686e-01
3.00830960e-01 1.05748606e+00 -7.85385787e-01 -4.49774444e-01
-3.37235093e-01 9.77652848e-01 -2.08269596e-01 2.62219161e-01
-7.00386703e-01 -6.25811040e-01 5.23230135e-01 4.66461703e-02
2.32936785e-01 -1.86346963e-01 -5.42973518e-01 -7.02843785e-01
2.24023238e-02 -7.24125922e-01 7.10872188e-02 -1.33629584e+00
-8.46357226e-01 5.03750801e-01 -4.99906801e-02 -1.49470091e+00
8.31531063e-02 -1.92220137e-01 -5.13493359e-01 1.30233133e+00
-1.86251032e+00 -8.15993607e-01 -6.22774720e-01 5.58398008e-01
6.32303357e-01 5.41434996e-02 4.14591640e-01 2.17975393e-01
-1.96042761e-01 2.44277284e-01 -7.29352683e-02 -2.63653398e-01
5.87494969e-01 -1.16034055e+00 1.50132030e-01 1.40952528e+00
-2.51493931e-01 2.40465716e-01 9.09533381e-01 -5.18299997e-01
-8.61587465e-01 -8.04982901e-01 9.07925487e-01 1.89867392e-01
2.78977931e-01 3.48757833e-01 -9.05495286e-01 -3.50310504e-01
5.56940258e-01 3.22713554e-02 2.31307670e-01 -5.12668967e-01
1.84651881e-01 -4.08826709e-01 -1.39463532e+00 5.43969333e-01
4.90501553e-01 -5.24932384e-01 -5.27300537e-01 -3.57217997e-01
4.77501154e-01 1.58741083e-02 -8.37395549e-01 2.89027274e-01
3.43328267e-01 -1.09066129e+00 1.05987966e+00 3.49649042e-01
3.39696676e-01 -9.65231180e-01 -6.78524971e-02 -1.24397016e+00
-5.18644750e-01 5.02733514e-02 2.03399971e-01 1.21225941e+00
-3.99249047e-02 -1.91081777e-01 3.34988683e-01 2.71692365e-01
-8.91643241e-02 -3.98398817e-01 -7.41733789e-01 -4.68483388e-01
-5.14282942e-01 -5.83940074e-02 5.13741910e-01 4.69752580e-01
-1.53272703e-01 -7.17129782e-02 -3.39211136e-01 4.28806990e-01
9.23338413e-01 1.27417043e-01 3.64677042e-01 -1.06622243e+00
2.67691374e-01 -6.10633969e-01 -8.66450131e-01 -4.32445943e-01
-4.29565191e-01 -3.77699614e-01 1.98909104e-01 -1.73744738e+00
3.39236289e-01 -5.30450344e-02 -8.63380790e-01 1.75352722e-01
-5.81155717e-01 6.99060321e-01 3.32752705e-01 2.58793384e-01
-7.23247588e-01 5.30029178e-01 1.47498512e+00 -4.17764693e-01
-1.51617751e-02 -2.06176341e-01 -6.78841174e-01 4.28117096e-01
8.43480587e-01 -3.05164039e-01 -4.88946110e-01 -1.83435783e-01
-4.78482321e-02 -1.10816106e-01 4.69965219e-01 -1.38882673e+00
2.58490920e-01 -3.12358618e-01 5.22904336e-01 -6.57489419e-01
1.19519018e-01 -9.24069047e-01 -2.67304038e-03 5.03318906e-01
-2.37310633e-01 1.95474014e-01 1.48254663e-01 2.71624953e-01
-7.33789742e-01 -9.61230025e-02 1.28832674e+00 6.55276980e-03
-1.40095901e+00 -3.58999595e-02 -4.02925491e-01 -3.20245802e-01
1.17819655e+00 -5.32077372e-01 -2.39000082e-01 -3.56693119e-01
-3.13231021e-01 4.12018523e-02 6.26331747e-01 5.26776195e-01
9.69871342e-01 -1.30310929e+00 -6.41321003e-01 9.62610021e-02
3.53233874e-01 -5.72226882e-01 6.08240306e-01 8.44035089e-01
-6.87141657e-01 3.10115188e-01 -8.43832135e-01 -6.31496072e-01
-1.38765407e+00 7.48936594e-01 1.33095697e-01 -6.26550615e-02
-2.89895803e-01 3.26591104e-01 3.60836857e-03 4.80056316e-01
7.59554729e-02 -2.23025903e-01 -6.19914532e-01 -8.72723907e-02
6.82869554e-01 5.54250121e-01 1.93003237e-01 -1.01310658e+00
-5.02712309e-01 7.57834733e-01 1.50450423e-01 -7.69867972e-02
8.74838412e-01 -7.26089597e-01 2.16973238e-02 3.04550678e-01
1.01193953e+00 -2.06608996e-01 -1.15108812e+00 -3.19049746e-01
8.91094841e-03 -8.82465005e-01 5.05382001e-01 -1.00025690e+00
-1.16002381e+00 9.37875152e-01 1.40822005e+00 1.65474206e-01
1.78166294e+00 -2.43627831e-01 4.87148345e-01 -3.01152885e-01
4.44369555e-01 -1.23387289e+00 2.65732557e-01 -4.25151400e-02
8.50890100e-01 -1.35303617e+00 2.53724426e-01 -1.79162413e-01
-6.76121354e-01 7.15666652e-01 4.76151735e-01 -5.61750196e-02
5.10213375e-01 -1.89074501e-01 2.52623886e-01 6.54144734e-02
-3.95096600e-01 -3.88629377e-01 6.37749016e-01 6.85980439e-01
1.42688528e-01 4.81970161e-02 -6.00433052e-01 1.88152507e-01
2.60440558e-01 1.46761656e-01 6.36758685e-01 9.15708780e-01
-8.37986887e-01 -6.68606639e-01 -6.05597138e-01 1.78376511e-01
-6.08195901e-01 -6.71535879e-02 1.26712918e-01 4.10035491e-01
2.17782348e-01 1.58032167e+00 -1.70798507e-02 -2.48420164e-01
3.12729627e-01 -4.09684926e-01 1.88560516e-01 -1.86708048e-01
-3.13393295e-01 -1.01151422e-01 -4.54046756e-01 -7.12088883e-01
-7.94308484e-01 -4.93593633e-01 -1.07919252e+00 -1.56687513e-01
-2.82046944e-01 2.73496807e-01 1.03589809e+00 7.32071698e-01
2.19582230e-01 7.49618590e-01 6.61348879e-01 -8.64303112e-01
-1.44861732e-02 -9.84540164e-01 -8.52885425e-01 4.75010097e-01
3.78599197e-01 -8.11105728e-01 -2.82824129e-01 2.83878088e-01] | [11.72187614440918, -1.9510142803192139] |
a7c85c94-44a1-4110-baa8-a708852366e1 | semantic-visual-guided-transformer-for-few | 2303.15494 | null | https://arxiv.org/abs/2303.15494v1 | https://arxiv.org/pdf/2303.15494v1.pdf | Semantic-visual Guided Transformer for Few-shot Class-incremental Learning | Few-shot class-incremental learning (FSCIL) has recently attracted extensive attention in various areas. Existing FSCIL methods highly depend on the robustness of the feature backbone pre-trained on base classes. In recent years, different Transformer variants have obtained significant processes in the feature representation learning of massive fields. Nevertheless, the progress of the Transformer in FSCIL scenarios has not achieved the potential promised in other fields so far. In this paper, we develop a semantic-visual guided Transformer (SV-T) to enhance the feature extracting capacity of the pre-trained feature backbone on incremental classes. Specifically, we first utilize the visual (image) labels provided by the base classes to supervise the optimization of the Transformer. And then, a text encoder is introduced to automatically generate the corresponding semantic (text) labels for each image from the base classes. Finally, the constructed semantic labels are further applied to the Transformer for guiding its hyperparameters updating. Our SV-T can take full advantage of more supervision information from base classes and further enhance the training robustness of the feature backbone. More importantly, our SV-T is an independent method, which can directly apply to the existing FSCIL architectures for acquiring embeddings of various incremental classes. Extensive experiments on three benchmarks, two FSCIL architectures, and two Transformer variants show that our proposed SV-T obtains a significant improvement in comparison to the existing state-of-the-art FSCIL methods. | ['Qinmu Peng', 'Chengxiang Lei', 'Jingyi Zhang', 'Sichao Fu', 'Wenhao Qiu'] | 2023-03-27 | null | null | null | null | ['class-incremental-learning', 'few-shot-class-incremental-learning'] | ['computer-vision', 'methodology'] | [ 3.64576161e-01 -1.11533493e-01 -2.63573796e-01 -4.47909236e-01
-6.02009535e-01 -4.52182651e-01 7.24505305e-01 -6.29255697e-02
-9.17034149e-02 4.23624605e-01 1.12715378e-01 8.70569202e-04
-1.64799407e-01 -8.58225405e-01 -6.63280666e-01 -8.50722551e-01
2.20369741e-01 3.04322124e-01 5.91955602e-01 -2.03193143e-01
3.89488876e-01 2.10134789e-01 -1.80543172e+00 2.18300387e-01
7.71512985e-01 1.42028058e+00 3.47111702e-01 6.84363097e-02
-3.66303772e-01 8.04947495e-01 -3.14102441e-01 -2.31906846e-01
2.19807059e-01 -4.11981195e-01 -6.36532307e-01 3.32101673e-01
2.91896522e-01 -1.88076526e-01 -4.59190518e-01 9.67608273e-01
3.82247686e-01 2.75644839e-01 7.38652229e-01 -1.45809293e+00
-1.01465821e+00 5.05311430e-01 -4.33528572e-01 8.51025432e-03
8.84655118e-02 1.20051846e-01 1.18261337e+00 -1.35676527e+00
7.49398708e-01 1.28803849e+00 4.76386726e-01 5.41621447e-01
-8.26438069e-01 -7.19042063e-01 5.26616752e-01 6.16459310e-01
-1.47879171e+00 -3.93044561e-01 9.24582958e-01 -3.42612535e-01
8.36847425e-01 -1.03349410e-01 7.02783108e-01 8.65327120e-01
-1.45519614e-01 1.16655183e+00 8.88699353e-01 -5.23190320e-01
2.44757563e-01 2.86929011e-01 2.55422950e-01 9.36989069e-01
-1.25273749e-01 -7.42066950e-02 -6.45643711e-01 2.19827205e-01
5.25230348e-01 2.87790596e-01 -1.49400070e-01 -6.90690100e-01
-9.83898640e-01 8.40334773e-01 5.76739371e-01 3.87653440e-01
4.00495203e-03 1.72784552e-02 6.00991011e-01 3.38399082e-01
4.39423233e-01 1.67384535e-01 -3.84485304e-01 6.89432546e-02
-6.84377372e-01 -1.14136726e-01 3.23465228e-01 1.32639992e+00
9.00827646e-01 -1.91301983e-02 -6.43948615e-01 1.02059698e+00
2.21536443e-01 4.04476404e-01 6.62781358e-01 -6.38328791e-01
4.99808788e-01 1.06096923e+00 -2.45286644e-01 -8.50133061e-01
-1.01648644e-01 -6.05781734e-01 -7.28888988e-01 -7.58884847e-02
-4.05303091e-02 4.32859242e-01 -9.93285418e-01 1.57003033e+00
3.72361273e-01 2.98322976e-01 1.90776393e-01 5.20870328e-01
7.81335413e-01 8.65151763e-01 -1.06597774e-01 -3.54628079e-02
1.12001860e+00 -1.24948394e+00 -4.52519536e-01 -2.00971097e-01
6.43169940e-01 -4.43161100e-01 1.29635882e+00 2.35772654e-01
-6.15050375e-01 -8.49750042e-01 -1.17981911e+00 -1.13735506e-02
-5.20770848e-01 3.33733320e-01 4.63882029e-01 2.74305254e-01
-8.00998449e-01 4.32829499e-01 -7.92082787e-01 -1.80233985e-01
9.76541340e-01 2.49727160e-01 -3.57255697e-01 -4.48192716e-01
-1.17851746e+00 5.17852724e-01 4.89435732e-01 5.09336330e-02
-1.08804774e+00 -6.90584660e-01 -1.06671083e+00 2.38028288e-01
4.00129706e-01 -3.16104800e-01 1.07892990e+00 -1.07312846e+00
-1.51005328e+00 7.30379999e-01 -8.02449882e-02 -1.42181873e-01
2.45784968e-01 -1.30119408e-02 -1.56681001e-01 1.26033232e-01
3.34983021e-01 7.82617390e-01 1.07282782e+00 -1.07757282e+00
-8.41336906e-01 -2.97759444e-01 1.69445872e-01 3.07855308e-01
-8.87489796e-01 -2.45448992e-01 -7.99160421e-01 -5.99487245e-01
-4.36926298e-02 -8.83279681e-01 -1.79443303e-02 3.02765876e-01
-3.11907858e-01 -7.35626459e-01 9.73419189e-01 -1.74180031e-01
1.34779561e+00 -2.45734048e+00 2.19453454e-01 -6.21489584e-02
-3.19779143e-02 5.55138648e-01 -3.20221066e-01 3.85738492e-01
2.63361931e-02 -2.48182237e-01 -2.73564965e-01 -2.94952750e-01
1.07254922e-01 2.13267326e-01 -3.38957429e-01 3.15492839e-01
3.40596586e-01 1.05267894e+00 -1.03672695e+00 -5.90590000e-01
4.41310853e-01 3.84584635e-01 -5.05594552e-01 3.70013267e-01
-2.34786466e-01 1.84541211e-01 -6.13192022e-01 6.05988801e-01
5.86942792e-01 -4.62347388e-01 -1.00655980e-01 -3.51037979e-01
-8.45343992e-02 -6.69062436e-02 -8.77251267e-01 1.81033802e+00
-4.24795657e-01 3.85754883e-01 -5.72010398e-01 -1.49171650e+00
1.00749278e+00 1.34190872e-01 3.78052294e-01 -9.20993924e-01
8.04303959e-02 2.55883247e-01 -2.85452008e-01 -3.73793095e-01
6.21822774e-02 -1.78573936e-01 -1.62565231e-01 1.51589096e-01
4.79107201e-01 7.60004595e-02 3.69404584e-01 3.00982445e-01
8.59659195e-01 3.10786366e-01 2.67113298e-01 -6.48460910e-02
9.63167131e-01 -1.39457673e-01 7.50397682e-01 5.56072950e-01
-2.86987782e-01 4.50877070e-01 1.64588973e-01 -4.37232494e-01
-9.03678536e-01 -1.15346169e+00 -3.08593720e-01 1.21808362e+00
3.63910854e-01 -5.06364584e-01 -5.69959402e-01 -1.18806028e+00
-4.07253057e-02 7.27145493e-01 -7.13837624e-01 -7.10783958e-01
-3.22327286e-01 -4.39300060e-01 1.72850341e-01 8.28885376e-01
7.02019930e-01 -1.13022149e+00 -4.18405950e-01 3.39789420e-01
3.08500566e-02 -1.03269196e+00 -4.95375663e-01 1.93206087e-01
-7.40904331e-01 -9.93473589e-01 -6.45802379e-01 -1.06624031e+00
7.91583836e-01 5.01922607e-01 7.52880514e-01 2.25311499e-02
-4.29227471e-01 4.52541322e-01 -7.14770734e-01 -3.33133996e-01
-6.33845627e-02 1.57194301e-01 -2.79269099e-01 3.36539626e-01
4.82750148e-01 -5.11155546e-01 -5.70571303e-01 3.96478653e-01
-9.91001964e-01 1.98755711e-01 5.33945799e-01 9.43723321e-01
5.67107260e-01 8.82794410e-02 8.54345083e-01 -1.06657064e+00
-2.33801361e-02 -4.05601710e-01 -5.26177645e-01 4.68817770e-01
-6.99357092e-01 2.47160211e-01 9.59198952e-01 -3.65304291e-01
-1.25318146e+00 9.42362472e-02 -1.42142043e-01 -6.62924528e-01
6.02204725e-02 4.11980838e-01 -2.68955261e-01 -5.36109060e-02
3.48596007e-01 5.02440929e-01 -1.96968302e-01 -3.75266045e-01
5.62686980e-01 7.80748427e-01 3.61588210e-01 -5.03943801e-01
1.02624547e+00 4.73860919e-01 -4.96086963e-02 -6.23890579e-01
-1.41513622e+00 -5.61084569e-01 -6.56686723e-01 -1.00940831e-01
6.10093117e-01 -9.43893433e-01 -1.05819531e-01 7.16858864e-01
-8.58598888e-01 -2.41014257e-01 -4.95674700e-01 2.10198224e-01
-5.47046185e-01 2.58490205e-01 -3.86235505e-01 -4.05938536e-01
-4.56891000e-01 -1.25944436e+00 1.36270678e+00 3.94353598e-01
3.91228348e-01 -1.06087589e+00 2.76903547e-02 1.24756090e-01
5.64751446e-01 -1.42435923e-01 1.09684956e+00 -5.90708852e-01
-6.22842669e-01 -2.21009180e-01 -3.92587095e-01 5.79129219e-01
3.89297545e-01 -2.71618932e-01 -1.12165880e+00 -5.72540224e-01
-2.10036531e-01 -7.57501900e-01 1.02091777e+00 9.05138031e-02
1.49588096e+00 1.48794167e-02 -5.19527912e-01 7.75774479e-01
1.56331587e+00 2.56187946e-01 6.11607015e-01 3.85534495e-01
7.78481603e-01 3.27517211e-01 9.71805274e-01 5.29936910e-01
4.16286170e-01 6.75875485e-01 3.88795644e-01 4.89566550e-02
-3.82659376e-01 -4.45011467e-01 4.78061914e-01 1.01349282e+00
1.67732239e-01 -1.16029866e-01 -4.89062965e-01 6.19500220e-01
-1.91811037e+00 -6.92719162e-01 4.43581194e-01 2.04966474e+00
8.34904909e-01 1.02429762e-01 -3.12742621e-01 1.92919284e-01
7.03173280e-01 2.64018506e-01 -8.17310989e-01 5.48378378e-02
7.10171759e-02 2.67418325e-01 1.38468221e-01 1.43174678e-01
-1.32128131e+00 1.25126493e+00 4.98133469e+00 1.25980234e+00
-1.09490108e+00 1.39173001e-01 4.51531559e-01 7.72537291e-02
-2.74210870e-01 1.14600509e-01 -1.19349122e+00 3.34910035e-01
5.99179924e-01 -3.74510139e-01 2.25119352e-01 1.20331037e+00
-2.39015773e-01 3.22799057e-01 -1.14307570e+00 1.05708456e+00
3.88691962e-01 -1.24881661e+00 4.00242001e-01 -2.08648324e-01
7.92593777e-01 -1.20700501e-01 1.17361821e-01 8.65802884e-01
8.39832574e-02 -6.54517531e-01 7.44346499e-01 4.23025310e-01
1.20150840e+00 -7.93738246e-01 6.28255606e-01 2.67040581e-01
-1.61727202e+00 -3.76053452e-01 -8.24277163e-01 2.15383977e-01
-1.67512577e-02 5.69310009e-01 -6.59135759e-01 6.56657398e-01
5.60099125e-01 1.36937106e+00 -9.53661740e-01 9.38105583e-01
-3.91446471e-01 6.59023702e-01 -6.28931597e-02 5.25303930e-02
4.26813245e-01 -1.32315606e-01 2.78779060e-01 9.81031537e-01
3.32304806e-01 -1.81289315e-01 4.47503597e-01 6.40973568e-01
-2.19337389e-01 2.50586152e-01 -7.14371860e-01 -1.32492363e-01
6.02736056e-01 1.29042649e+00 -6.47566557e-01 -5.00877023e-01
-7.46901631e-01 1.10748839e+00 6.73133910e-01 1.46174714e-01
-7.75767922e-01 -6.01268768e-01 3.92835230e-01 -1.22264072e-01
6.49476051e-01 1.24825507e-01 3.35719526e-01 -1.31015146e+00
1.24309562e-01 -5.90517342e-01 3.94278586e-01 -6.63147390e-01
-1.38402617e+00 6.56968951e-01 -2.93994080e-02 -1.40223956e+00
-6.05766997e-02 -7.56800830e-01 -5.25967598e-01 3.34662378e-01
-1.83762717e+00 -1.50804210e+00 -4.47147489e-01 8.81714880e-01
9.42882001e-01 -4.15175676e-01 8.51640046e-01 3.31433266e-01
-7.25585520e-01 8.41361582e-01 1.68217614e-01 9.84854549e-02
5.92396975e-01 -1.13125753e+00 3.44192356e-01 7.19925940e-01
1.88236952e-01 4.13626075e-01 3.62138897e-02 -3.76054555e-01
-1.35210538e+00 -1.54437351e+00 5.87270439e-01 -1.77266926e-01
6.70976102e-01 -5.18547952e-01 -8.77321839e-01 8.07193339e-01
5.03112003e-02 4.86041129e-01 4.87684906e-01 -1.75532594e-01
-5.21606982e-01 -4.38617408e-01 -7.31026471e-01 3.27051759e-01
1.34001005e+00 -6.25825644e-01 -6.07964337e-01 3.25684428e-01
7.90491700e-01 -2.07350980e-02 -5.21802187e-01 5.03692687e-01
4.31634665e-01 -7.23443151e-01 9.48550403e-01 -5.18523097e-01
3.19763660e-01 -2.54534662e-01 -2.10771978e-01 -1.31769156e+00
-5.25339723e-01 -1.13025017e-01 -5.97791979e-03 1.56734705e+00
1.52685314e-01 -5.62305093e-01 6.14185750e-01 -1.57172997e-02
-4.86895025e-01 -9.56400990e-01 -8.42227817e-01 -8.56347144e-01
2.35941727e-02 -3.66897464e-01 3.96534622e-01 7.95287490e-01
-2.06721306e-01 5.88074505e-01 -2.34186813e-01 -1.89116806e-01
6.07412696e-01 3.87739778e-01 6.37485743e-01 -1.32595408e+00
-1.85218543e-01 -2.45716482e-01 -6.36352479e-01 -1.17655826e+00
2.83593476e-01 -1.40503860e+00 8.16152543e-02 -1.51736593e+00
4.97509599e-01 -7.20446348e-01 -6.74104869e-01 8.81788909e-01
-4.74003494e-01 1.71677619e-01 2.46364772e-01 2.08062187e-01
-9.66566563e-01 1.08874154e+00 1.25469828e+00 -3.18895727e-01
1.24615636e-02 -1.24630518e-01 -7.06280649e-01 6.43603444e-01
5.77587128e-01 -4.88796830e-01 -9.41370964e-01 -2.24809870e-01
7.18393624e-02 -4.49536890e-01 2.07379416e-01 -1.12111747e+00
2.48097122e-01 -7.82083869e-02 2.56214321e-01 -4.82876718e-01
1.52847081e-01 -7.87324727e-01 -3.46950829e-01 1.85805261e-01
-2.38569975e-01 -4.04212683e-01 -4.04514670e-02 6.61742151e-01
-3.14044297e-01 -4.53617424e-01 8.74253750e-01 -2.09125802e-01
-1.04937851e+00 6.88178837e-01 -2.38639116e-02 2.34702572e-01
1.07418537e+00 -1.10342711e-01 -2.78944939e-01 2.91795693e-02
-3.91589582e-01 3.22908610e-01 3.81739676e-01 6.00170493e-01
8.67054224e-01 -1.59661698e+00 -4.30411816e-01 5.43434978e-01
7.13996530e-01 5.73517345e-02 2.17996225e-01 6.24865770e-01
-1.05614319e-01 4.17839259e-01 -2.44664535e-01 -6.97846174e-01
-9.31082785e-01 9.58912551e-01 6.24204054e-02 -3.25798869e-01
-8.64133000e-01 9.37215686e-01 6.95113599e-01 -4.13177848e-01
3.65893990e-01 3.84194255e-02 -3.54415685e-01 1.12012535e-01
6.47963941e-01 1.12447195e-01 2.17042267e-02 -5.73683739e-01
-3.86069804e-01 9.49374855e-01 -3.82084996e-01 1.92628682e-01
1.51521099e+00 -1.31872296e-01 1.27222121e-01 6.20638013e-01
1.52148998e+00 -3.17150146e-01 -1.39433694e+00 -6.46664858e-01
-4.31830250e-02 -4.05378193e-01 9.41127613e-02 -4.58730489e-01
-1.38578033e+00 1.11645949e+00 6.75613105e-01 -1.74862191e-01
1.23871648e+00 5.88402338e-02 8.79842460e-01 4.53320444e-01
6.47793412e-01 -1.12596989e+00 4.03741032e-01 4.69478101e-01
7.89827108e-01 -1.11644840e+00 -2.91103274e-01 -5.47226310e-01
-6.61949694e-01 1.15379572e+00 6.51568830e-01 -1.56270176e-01
7.27979779e-01 -6.33983612e-02 -1.54655442e-01 -3.06937754e-01
-7.53072917e-01 -3.49821389e-01 3.83926600e-01 5.10285854e-01
2.26943821e-01 -2.33311683e-01 -8.50453898e-02 5.93281567e-01
2.32968137e-01 1.55649796e-01 4.97373268e-02 9.62187707e-01
-6.77621722e-01 -1.27329636e+00 1.20502397e-01 4.38414276e-01
4.74934876e-02 -1.02555931e-01 -1.04290687e-01 5.68864763e-01
3.39822769e-01 6.71541691e-01 -8.36483985e-02 -4.30906922e-01
4.80426818e-01 2.43474215e-01 5.71833730e-01 -9.45138335e-01
-2.47819856e-01 -1.50695056e-01 -2.27796540e-01 -5.62623084e-01
-4.07222301e-01 -5.34214318e-01 -1.30568373e+00 3.58163238e-01
-4.45858955e-01 2.15740308e-01 3.12130451e-01 1.05332899e+00
3.62744868e-01 5.32376051e-01 1.00677931e+00 -8.00777435e-01
-6.11197829e-01 -8.77956808e-01 -5.32390535e-01 6.29960775e-01
1.18683070e-01 -1.16686809e+00 -5.16806543e-01 -1.59666978e-03] | [9.89794921875, 2.5240213871002197] |
afded3b7-1c67-498a-8c04-0c2c2ccb6975 | cped-a-large-scale-chinese-personalized-and-1 | 2205.14727 | null | https://arxiv.org/abs/2205.14727v1 | https://arxiv.org/pdf/2205.14727v1.pdf | CPED: A Large-Scale Chinese Personalized and Emotional Dialogue Dataset for Conversational AI | Human language expression is based on the subjective construal of the situation instead of the objective truth conditions, which means that speakers' personalities and emotions after cognitive processing have an important influence on conversation. However, most existing datasets for conversational AI ignore human personalities and emotions, or only consider part of them. It's difficult for dialogue systems to understand speakers' personalities and emotions although large-scale pre-training language models have been widely used. In order to consider both personalities and emotions in the process of conversation generation, we propose CPED, a large-scale Chinese personalized and emotional dialogue dataset, which consists of multi-source knowledge related to empathy and personal characteristic. These knowledge covers gender, Big Five personality traits, 13 emotions, 19 dialogue acts and 10 scenes. CPED contains more than 12K dialogues of 392 speakers from 40 TV shows. We release the textual dataset with audio features and video features according to the copyright claims, privacy issues, terms of service of video platforms. We provide detailed description of the CPED construction process and introduce three tasks for conversational AI, including personality recognition, emotion recognition in conversations as well as personalized and emotional conversation generation. Finally, we provide baseline systems for these tasks and consider the function of speakers' personalities and emotions on conversation. Our motivation is to propose a dataset to be widely adopted by the NLP community as a new open benchmark for conversational AI research. The full dataset is available at https://github.com/scutcyr/CPED. | ['Xiangmin Xu', 'Qianfeng Tie', 'Wenjing Han', 'Minlie Huang', 'Jianxin Pang', 'Xiaofen Xing', 'Weiquan Fan', 'YiRong Chen'] | 2022-05-29 | null | null | null | null | ['personality-trait-recognition', 'personality-recognition-in-conversation', 'emotion-recognition-in-conversation', 'dialog-act-classification', 'emotional-dialogue-acts', 'personalized-and-emotional-conversation', 'open-domain-dialog', 'conversational-response-generation'] | ['computer-vision', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [-2.88165003e-01 3.62323731e-01 8.22779164e-02 -7.54911840e-01
-2.95490623e-01 -3.93183529e-01 7.31990397e-01 -4.12258267e-01
-1.31684497e-01 8.31945717e-01 8.34617972e-01 4.48797613e-01
2.37007767e-01 -5.55919647e-01 1.20891230e-02 -6.40514314e-01
1.19234391e-01 6.20629013e-01 -4.94746268e-01 -6.77080274e-01
8.98080841e-02 1.46711860e-02 -1.49724627e+00 8.05605710e-01
7.37414896e-01 1.03655362e+00 -1.94870472e-01 8.77541125e-01
-4.15272623e-01 8.99478197e-01 -1.03930163e+00 -1.05114090e+00
-2.88234740e-01 -6.77156448e-01 -1.28967834e+00 4.21247065e-01
-2.25920856e-01 -3.92605484e-01 -2.08849907e-01 7.97316432e-01
9.04890120e-01 2.40176991e-01 7.82972217e-01 -1.62910855e+00
-8.04341853e-01 9.50498223e-01 -7.30093792e-02 -1.59183249e-01
9.44547057e-01 1.37548387e-01 9.62245464e-01 -5.08546472e-01
5.99822998e-01 1.44212234e+00 4.67721254e-01 1.04161072e+00
-6.20911062e-01 -5.88967860e-01 -9.05075669e-02 3.16555321e-01
-9.24817026e-01 -6.99101269e-01 1.01158714e+00 -3.09476972e-01
7.48849869e-01 5.80144525e-01 9.42876518e-01 1.95282292e+00
-2.23736048e-01 1.17381740e+00 1.14469373e+00 -3.50220054e-01
-1.29489437e-01 8.09617281e-01 1.07389838e-01 3.15573961e-01
-7.93385983e-01 -3.94095093e-01 -7.79168308e-01 -3.88580829e-01
3.67371768e-01 -4.74217981e-01 -4.02630538e-01 3.48112494e-01
-1.20612967e+00 8.24253261e-01 -5.94068021e-02 3.70275795e-01
-4.07096863e-01 -4.92690831e-01 8.79218340e-01 6.33111775e-01
5.20832717e-01 5.47288358e-01 -4.61604089e-01 -8.57356787e-01
8.80193547e-04 2.78150290e-01 1.49343646e+00 9.87806082e-01
4.85595793e-01 -1.88055202e-01 -4.51893836e-01 1.73768568e+00
-6.12337999e-02 2.98352003e-01 7.73289680e-01 -1.09520710e+00
1.41434193e-01 6.29887402e-01 -1.20188147e-02 -1.27283847e+00
-3.58778119e-01 2.31077164e-01 -9.10393178e-01 -6.85693920e-01
1.55421555e-01 -7.48131335e-01 1.09463148e-01 1.77452409e+00
3.04189414e-01 -2.39189923e-01 6.31165206e-01 7.95262694e-01
1.54545581e+00 9.46106434e-01 -1.92275897e-01 -5.78139007e-01
1.59490180e+00 -9.43611681e-01 -1.06420135e+00 4.51817848e-02
4.42689389e-01 -8.58083129e-01 1.31156898e+00 4.92743134e-01
-1.09667790e+00 -1.78318307e-01 -4.70992744e-01 -1.07884802e-01
-4.50546563e-01 2.14356661e-01 8.90612781e-01 7.67727256e-01
-7.86176145e-01 1.13504231e-01 6.61712140e-02 -5.22686005e-01
8.13382566e-02 1.74053609e-01 -2.70357907e-01 4.43248391e-01
-1.65988719e+00 7.66868174e-01 9.38057303e-02 2.44726196e-01
-4.70968157e-01 -7.65357688e-02 -6.15544915e-01 -1.24319382e-01
2.91150510e-01 -3.97441924e-01 1.50840926e+00 -1.46822131e+00
-2.32989812e+00 1.13631344e+00 -1.44440264e-01 -8.59911963e-02
5.24190545e-01 -1.17642760e-01 -6.71501040e-01 1.52521133e-01
-1.87303290e-01 7.31246948e-01 5.25844038e-01 -1.06797552e+00
-5.63445807e-01 -1.21501923e-01 1.30711004e-01 5.06758094e-01
-7.08916008e-01 3.91335070e-01 -6.98625386e-01 -3.45006377e-01
-5.41066468e-01 -1.05287933e+00 1.48926541e-01 -6.23933554e-01
-5.57243228e-01 -6.92234457e-01 5.81997514e-01 -7.25826800e-01
1.04808104e+00 -2.13537598e+00 1.11115769e-01 -5.82702644e-02
5.01956157e-02 2.80233789e-02 -3.26359123e-02 4.55191046e-01
1.98281333e-01 1.63886324e-02 1.39101118e-01 -3.76366049e-01
4.39561307e-01 1.37411594e-01 -1.74946591e-01 -3.70620973e-02
-5.23890220e-02 7.25411057e-01 -7.98460424e-01 -6.55636907e-01
-1.61517531e-01 3.98369163e-01 -4.27458465e-01 6.56739712e-01
-1.42980605e-01 5.96866667e-01 -6.80735826e-01 3.77018660e-01
3.45000565e-01 1.97747365e-01 1.58081993e-01 -1.41082779e-01
3.51774246e-02 4.64803994e-01 -7.80340612e-01 1.37550902e+00
-3.92366439e-01 7.48744607e-01 3.35647613e-01 -5.94274700e-01
1.23411644e+00 8.68647099e-01 6.35861993e-01 -4.20709699e-01
4.66367602e-01 -1.33962557e-01 4.33989987e-02 -9.09987628e-01
7.44288266e-01 -2.37684801e-01 -6.42319620e-01 6.16501749e-01
1.34224864e-02 -3.44155818e-01 8.67221281e-02 3.26403886e-01
6.32737458e-01 -4.20084953e-01 2.51756400e-01 -3.61256860e-02
7.29984522e-01 -2.17032999e-01 7.13982284e-01 2.48974159e-01
-5.31115711e-01 2.04968110e-01 1.07728434e+00 -2.43750289e-01
-6.31977439e-01 -4.32269722e-01 -9.45392773e-02 1.44056165e+00
-1.71194255e-01 -4.91855830e-01 -8.68752658e-01 -3.83938223e-01
-4.77862984e-01 6.76160991e-01 -6.22880220e-01 -1.08048126e-01
-1.02495804e-01 -6.83019698e-01 9.26950157e-01 2.65308451e-02
8.69728982e-01 -1.69264209e+00 -9.89470780e-02 1.61290064e-01
-9.68803644e-01 -1.27049088e+00 -4.87001270e-01 -4.87870425e-01
-6.29168004e-02 -7.54330873e-01 -5.72015345e-01 -7.60167599e-01
1.60849005e-01 -2.14326799e-01 1.24292600e+00 -5.50203025e-02
-4.12683822e-02 6.84102654e-01 -6.39167905e-01 -6.74600244e-01
-7.01994956e-01 2.98738480e-03 8.63653794e-02 3.06880891e-01
7.23260522e-01 -4.98800874e-01 -2.75410980e-01 4.69917834e-01
-3.12043250e-01 2.39597157e-01 2.68649161e-01 8.59797180e-01
-1.54681116e-01 7.64522776e-02 8.76208186e-01 -8.77014458e-01
1.47539520e+00 -4.83513892e-01 3.50730419e-01 2.00911835e-01
9.87800434e-02 -5.36644816e-01 4.27430958e-01 -5.36478281e-01
-1.68440700e+00 -3.18418205e-01 -4.26651776e-01 4.63726856e-02
-5.03367662e-01 2.92342693e-01 -4.84111726e-01 4.35953230e-01
3.05445045e-01 1.04839385e-01 1.67762101e-01 -1.13099821e-01
2.40428105e-01 1.41441095e+00 5.37790418e-01 -1.08430672e+00
-1.52955815e-01 -4.78515364e-02 -7.74952054e-01 -1.28896570e+00
-7.13232338e-01 -3.14580977e-01 -3.55913877e-01 -7.30328798e-01
8.38726044e-01 -9.01253879e-01 -1.44781530e+00 7.29608595e-01
-1.26106775e+00 -2.24024683e-01 1.13339432e-01 4.44152623e-01
-7.56984770e-01 3.14658254e-01 -1.10111570e+00 -1.03304553e+00
-6.08829200e-01 -9.87732112e-01 6.13551617e-01 2.85154790e-01
-7.22867131e-01 -8.80603790e-01 -6.59167543e-02 8.42488170e-01
2.31433362e-01 1.38356537e-01 6.28741980e-01 -7.80981243e-01
3.36249709e-01 4.65344638e-02 1.13491617e-01 5.14343143e-01
-1.48850931e-02 2.26052552e-01 -1.13615954e+00 3.44526380e-01
2.29359090e-01 -1.02735174e+00 1.85402349e-01 4.71311249e-02
1.02606750e+00 -7.27071166e-01 1.16027005e-01 7.63827413e-02
3.86050552e-01 2.95190424e-01 7.79275954e-01 -9.54961255e-02
4.42548364e-01 1.44864964e+00 6.51999772e-01 9.47773755e-01
8.69248331e-01 5.98282635e-01 -2.28805184e-01 2.06885830e-01
5.04697978e-01 3.14887933e-04 8.14739764e-01 1.32819974e+00
-3.52685481e-01 -3.62384856e-01 -7.22939432e-01 4.02350515e-01
-1.76735222e+00 -1.24523675e+00 -1.73765078e-01 1.62693441e+00
1.38477433e+00 -2.86669105e-01 4.27981734e-01 -2.27891713e-01
7.52712548e-01 1.33345142e-01 -4.16942924e-01 -1.12493312e+00
-3.50978464e-01 -4.51754808e-01 -4.07001615e-01 4.45465326e-01
-7.85949349e-01 1.04434943e+00 5.52254581e+00 7.70647347e-01
-9.86957252e-01 1.28751686e-02 9.90550637e-01 -1.69929042e-01
-1.78464070e-01 -4.71041501e-01 -7.01109111e-01 3.56793582e-01
9.91430223e-01 -4.46279794e-01 6.25933170e-01 8.21915448e-01
2.07108051e-01 1.53005168e-01 -1.09827530e+00 1.22512007e+00
2.91523397e-01 -6.93645477e-01 -2.42734149e-01 -2.16627326e-02
4.62539732e-01 -3.62222224e-01 -6.43697679e-02 7.82607436e-01
3.08525890e-01 -9.57974494e-01 3.33685577e-01 5.49381018e-01
4.87964034e-01 -8.15476894e-01 7.49587297e-01 4.79678512e-01
-6.74276531e-01 9.03249159e-02 -3.47592890e-01 -2.49937087e-01
1.65814027e-01 2.39143476e-01 -8.95104647e-01 6.76586106e-02
8.45124841e-01 7.01320648e-01 -1.39771178e-01 1.05131477e-01
-1.17811583e-01 5.22304773e-01 -1.68709293e-01 -5.58406293e-01
1.11868136e-01 -3.85041267e-01 4.67265576e-01 1.43393099e+00
2.22626075e-01 6.69330716e-01 2.71509495e-02 6.37434185e-01
-2.55801290e-01 6.72137141e-01 -4.88790125e-01 -4.05373305e-01
5.49417973e-01 1.50813913e+00 -7.71116391e-02 -4.42626625e-01
-4.56422359e-01 1.05125809e+00 2.34818727e-01 1.25621110e-01
-7.91769385e-01 -2.29301497e-01 7.72057891e-01 -6.23338163e-01
-3.19734573e-01 3.13100934e-01 9.82618257e-02 -1.34873331e+00
-2.20541254e-01 -1.48173285e+00 3.50861311e-01 -8.02109241e-01
-1.64380658e+00 8.51133883e-01 -2.00318068e-01 -9.25210476e-01
-5.17802060e-01 -4.72423911e-01 -7.40502894e-01 6.63169205e-01
-7.71168828e-01 -1.13290644e+00 -4.51592505e-01 8.66557300e-01
7.24746704e-01 -4.73030150e-01 1.21651196e+00 1.85216427e-01
-8.28617930e-01 6.02078557e-01 -2.97750473e-01 3.69805753e-01
1.17687154e+00 -1.00128257e+00 -2.10603356e-01 -1.91734105e-01
-3.55431765e-01 4.70516920e-01 8.42962742e-01 -3.33132118e-01
-1.14140928e+00 -4.34115380e-01 1.04660225e+00 -4.22526568e-01
6.81787372e-01 -4.91631955e-01 -8.09166193e-01 7.16294825e-01
9.09409821e-01 -9.52701807e-01 1.27681422e+00 5.80154181e-01
1.11684419e-01 9.51355770e-02 -1.18090272e+00 8.74446213e-01
8.93156111e-01 -4.07895625e-01 -6.57248378e-01 4.86501813e-01
4.53806013e-01 -3.22369128e-01 -1.26640081e+00 7.28051588e-02
7.09001303e-01 -1.21795321e+00 6.48693085e-01 -5.12590826e-01
7.57420897e-01 4.95989263e-01 -1.70705076e-02 -1.45389032e+00
-1.66806374e-02 -9.88088787e-01 2.62922972e-01 1.92395937e+00
3.17674428e-01 -6.62180722e-01 6.02902591e-01 1.14474368e+00
-1.68155953e-01 -7.04048276e-01 -5.69271326e-01 1.46332756e-02
7.04895332e-02 -4.14373398e-01 8.72405350e-01 1.37889814e+00
8.32101405e-01 9.78236318e-01 -8.78671467e-01 -3.43405128e-01
2.03917567e-02 1.67115599e-01 1.22317004e+00 -1.02167726e+00
-2.77361572e-01 -6.77973688e-01 8.76003876e-03 -1.00691366e+00
6.69842005e-01 -3.80732298e-01 6.06917478e-02 -1.14602697e+00
2.35241428e-01 -2.80116737e-01 3.69983345e-01 3.26052904e-01
-1.24174990e-01 -1.35113955e-01 -4.12792452e-02 2.40326952e-02
-5.18794477e-01 1.03310919e+00 1.47663367e+00 1.25888353e-02
-3.71863335e-01 1.01149052e-01 -8.30426157e-01 8.69731903e-01
9.92905557e-01 1.35624126e-01 -4.63874966e-01 -7.49244317e-02
9.32758227e-02 4.64693606e-01 -3.76508608e-02 -5.10919333e-01
1.84357837e-01 -3.87153476e-01 1.89881563e-01 -3.19956869e-01
1.02912509e+00 -5.49062848e-01 -1.11252740e-01 -1.04573071e-01
-7.14242339e-01 -1.22883655e-01 -1.02130035e-02 1.05085224e-01
-5.71447551e-01 -1.77688390e-01 5.22580743e-01 -4.00364846e-01
-7.10088313e-01 1.47374019e-01 -6.28457010e-01 1.90024033e-01
9.82720852e-01 -1.82788000e-02 -4.04776424e-01 -1.24600983e+00
-8.78156900e-01 6.08433247e-01 1.50851771e-01 6.09516799e-01
5.10852516e-01 -1.25205243e+00 -9.76571023e-01 -1.34983286e-01
1.66943476e-01 -4.08226639e-01 7.67578900e-01 7.53211677e-01
-5.64709865e-02 2.05824345e-01 -4.71115649e-01 -1.48028359e-01
-1.70542836e+00 1.51734918e-01 2.87533462e-01 5.07776067e-02
-2.80885786e-01 9.94340837e-01 2.05953225e-01 -8.52661133e-01
3.80444556e-01 2.03286514e-01 -7.23298192e-01 4.40935254e-01
5.54973960e-01 3.09694737e-01 -5.57528913e-01 -1.09666526e+00
2.20583603e-02 -5.37765697e-02 -2.40752622e-02 -2.85185903e-01
1.15663099e+00 -5.04289210e-01 -4.91752088e-01 7.43937910e-01
1.22395194e+00 2.24032536e-01 -7.17093647e-01 -2.77070314e-01
-3.24296594e-01 -3.64015162e-01 -4.50466484e-01 -8.91628563e-01
-8.32045436e-01 8.10020387e-01 -3.15367877e-02 5.22815585e-01
1.01982903e+00 1.10448629e-01 1.09664333e+00 5.33125103e-01
1.10847302e-01 -1.58762097e+00 3.44614983e-01 8.76143515e-01
1.37550271e+00 -1.24867880e+00 -2.77353287e-01 -5.07483304e-01
-1.78074408e+00 1.02011561e+00 1.00441062e+00 4.73006517e-01
4.74674940e-01 -1.30925238e-01 5.35547018e-01 -7.22196698e-02
-1.23084211e+00 1.39682695e-01 4.93614115e-02 5.60453057e-01
8.12294900e-01 4.76910681e-01 -3.41012508e-01 1.35298312e+00
-1.04624331e+00 -2.71971464e-01 6.00624979e-01 4.04470772e-01
-1.18457116e-01 -1.14477587e+00 -2.53730476e-01 2.64492512e-01
-4.54781502e-01 1.38276756e-01 -1.13436019e+00 5.06485403e-01
3.81411165e-02 1.38227439e+00 -3.13978642e-02 -6.20585620e-01
3.83369654e-01 3.18349868e-01 8.74532983e-02 -4.13188487e-01
-8.34695697e-01 -2.69530388e-03 1.04088247e+00 -2.54931599e-01
-6.31652713e-01 -7.26972520e-01 -1.07689774e+00 -8.62538517e-01
-2.43455935e-02 4.97953862e-01 4.28353369e-01 8.03703845e-01
2.88595200e-01 1.18615992e-01 8.81113410e-01 -6.56301498e-01
-1.80338129e-01 -1.37061048e+00 -7.36892998e-01 7.22411811e-01
-2.27542907e-01 -3.60030204e-01 -3.99745613e-01 7.01055229e-02] | [13.025507926940918, 6.30573034286499] |
89514220-9923-4470-ae23-425574a5e9ff | reviewing-evolution-of-learning-functions-and | 2305.14397 | null | https://arxiv.org/abs/2305.14397v1 | https://arxiv.org/pdf/2305.14397v1.pdf | Reviewing Evolution of Learning Functions and Semantic Information Measures for Understanding Deep Learning | A new trend in deep learning, represented by Mutual Information Neural Estimation (MINE) and Information Noise Contrast Estimation (InfoNCE), is emerging. In this trend, similarity functions and Estimated Mutual Information (EMI) are used as learning and objective functions. Coincidentally, EMI is essentially the same as Semantic Mutual Information (SeMI) proposed by the author 30 years ago. This paper first reviews the evolutionary histories of semantic information measures and learning functions. Then, it briefly introduces the author's semantic information G theory with the rate-fidelity function R(G) (G denotes SeMI, and R(G) extends R(D)) and its applications to multi-label learning, the maximum Mutual Information (MI) classification, and mixture models. Then it discusses how we should understand the relationship between SeMI and Shan-non's MI, two generalized entropies (fuzzy entropy and coverage entropy), Autoencoders, Gibbs distributions, and partition functions from the perspective of the R(G) function or the G theory. An important conclusion is that mixture models and Restricted Boltzmann Machines converge because SeMI is maximized, and Shannon's MI is minimized, making information efficiency G/R close to 1. A potential opportunity is to simplify deep learning by using Gaussian channel mixture models for pre-training deep neural networks' latent layers without considering gradients. It also discusses how the SeMI measure is used as the reward function (reflecting purposiveness) for reinforcement learning. The G theory helps interpret deep learning but is far from enough. Combining semantic information theory and deep learning will accelerate their development. | ['Chenguang Lu'] | 2023-05-23 | null | null | null | null | ['multi-label-learning'] | ['methodology'] | [-1.06970116e-01 2.60038167e-01 -2.04938099e-01 -3.32067102e-01
-2.52737254e-01 -2.89844066e-01 7.01862812e-01 -1.12867951e-01
-8.22240233e-01 8.55751455e-01 1.08514763e-01 9.30156372e-03
-7.02685237e-01 -9.90871906e-01 -3.41376513e-01 -7.73008943e-01
-1.73085183e-01 2.96646178e-01 -9.30420831e-02 8.39836746e-02
3.71451557e-01 3.70820999e-01 -1.87320912e+00 -1.92601010e-02
1.04036653e+00 1.03665733e+00 5.03054678e-01 5.13760090e-01
-5.01213193e-01 9.73949552e-01 -5.12526631e-01 -4.57975686e-01
-1.30570874e-01 -7.51822174e-01 -1.11590087e+00 -5.27355194e-01
-4.67215598e-01 -9.57609341e-02 -3.37081611e-01 1.30866635e+00
3.93304616e-01 4.49611723e-01 1.35135329e+00 -1.17946422e+00
-8.35165858e-01 8.51153195e-01 -1.80727914e-01 1.29081801e-01
-9.51120630e-02 -2.80402303e-01 8.74136806e-01 -5.32033384e-01
2.52139598e-01 1.20284450e+00 7.60428846e-01 8.22117209e-01
-9.39676225e-01 -5.68517268e-01 -3.61966521e-01 3.67501527e-01
-1.40299881e+00 3.83780524e-02 5.78921854e-01 -3.95722210e-01
1.07317126e+00 1.75898552e-01 7.19801724e-01 6.48570359e-01
4.34219509e-01 1.07979167e+00 1.12409639e+00 -6.08356476e-01
5.52232683e-01 6.18532240e-01 6.72128126e-02 5.97983718e-01
2.43862718e-01 3.33113939e-01 -2.96720028e-01 1.76844582e-01
8.63156736e-01 2.05137935e-02 -8.79086182e-02 -9.45586935e-02
-6.86448753e-01 1.04789424e+00 4.51102793e-01 9.62124288e-01
-2.00248003e-01 1.60910606e-01 1.99490950e-01 3.60016793e-01
4.60133702e-01 4.46897566e-01 -3.19446117e-01 -2.19747648e-01
-9.53699887e-01 -3.96006070e-02 1.13478673e+00 5.75885057e-01
9.28595126e-01 1.76860854e-01 2.73787022e-01 1.14122856e+00
7.34164774e-01 2.81816274e-01 9.41508234e-01 -1.23054063e+00
-2.38433510e-01 3.47341895e-01 -3.45987201e-01 -6.78437531e-01
-5.50904930e-01 -7.08724976e-01 -8.95875156e-01 8.91375095e-02
3.16885382e-01 -9.85254571e-02 -6.24815583e-01 2.05214524e+00
-1.66600123e-01 -2.71109253e-01 3.28224748e-01 7.09834158e-01
8.54002297e-01 4.23500299e-01 1.67191476e-01 -1.94354340e-01
1.09360182e+00 -5.51106453e-01 -8.98217797e-01 -3.21022779e-01
6.72587812e-01 -3.38332236e-01 5.90093791e-01 2.14457735e-01
-9.78712201e-01 -5.18884659e-01 -1.26379883e+00 2.03333393e-01
-7.85107136e-01 -3.14219654e-01 8.85871291e-01 9.85810459e-01
-1.28233993e+00 1.08205664e+00 -8.58220458e-01 -2.92353034e-01
4.97974098e-01 4.42607820e-01 -2.24112853e-01 1.73735797e-01
-1.57956624e+00 1.26434648e+00 8.58840764e-01 -3.10542017e-01
-7.00585425e-01 -2.55391836e-01 -8.43016088e-01 6.11524507e-02
-1.35083809e-01 -8.09179544e-01 7.86813974e-01 -1.13964403e+00
-1.85135508e+00 9.02232885e-01 3.22654814e-01 -3.94693792e-01
1.63778603e-01 -2.18778998e-01 -4.01374489e-01 9.72339064e-02
-2.61699617e-01 1.03718174e+00 3.40939850e-01 -1.15511549e+00
-3.50224495e-01 -5.72530329e-01 -1.78563848e-01 4.52045977e-01
-3.33312690e-01 -2.10467845e-01 1.13918290e-01 -2.73272693e-01
4.60069180e-01 -5.38469851e-01 -3.66682634e-02 -4.19815868e-01
-1.35488048e-01 -3.68057668e-01 1.94006398e-01 -6.56094968e-01
1.04568994e+00 -2.14539647e+00 4.58617285e-02 2.87114263e-01
3.25884849e-01 5.27825095e-02 9.23411548e-03 2.28694960e-01
-1.91344351e-01 2.05573559e-01 -5.97107291e-01 1.89906985e-01
7.59179816e-02 3.33923638e-01 2.97440499e-01 4.58019286e-01
-1.65463805e-01 8.68493497e-01 -8.59148324e-01 -2.77261972e-01
2.83583850e-01 6.36791527e-01 -6.42246068e-01 -6.25881040e-03
2.36261025e-01 1.28898501e-01 -8.86568520e-03 9.50500444e-02
6.98126614e-01 -3.57082546e-01 1.75878569e-01 -6.53233007e-02
-4.01329324e-02 1.79753870e-01 -9.60367501e-01 1.62490916e+00
-4.53691095e-01 8.17944527e-01 -4.45446849e-01 -1.19469047e+00
1.05023038e+00 2.04609364e-01 5.39214075e-01 -7.11551070e-01
4.72822994e-01 2.30072543e-01 1.09947361e-01 -3.64458650e-01
1.43766820e-01 -4.20418113e-01 2.62645423e-01 4.93342429e-01
8.50945234e-01 -1.67191308e-02 -2.74218805e-02 1.28738424e-02
7.85833597e-01 8.02007914e-02 5.19582212e-01 -5.68937957e-01
2.42205173e-01 -5.74447751e-01 1.43725678e-01 8.16020310e-01
-1.86460912e-01 3.11242104e-01 2.70801753e-01 5.06377742e-02
-9.52953815e-01 -1.35719287e+00 -4.27070558e-01 9.42549527e-01
2.29139715e-01 -1.00590959e-01 -1.10271668e+00 -2.70817220e-01
-2.28881076e-01 1.06045473e+00 -6.56214654e-01 -6.28929853e-01
1.20179743e-01 -1.15975416e+00 6.18081212e-01 3.99257034e-01
8.43321264e-01 -9.47501779e-01 -4.51605380e-01 1.21899620e-01
-1.79474592e-01 -3.54391783e-01 1.58383414e-01 7.47208476e-01
-1.02158260e+00 -7.84942329e-01 -7.98250258e-01 -4.86798018e-01
2.17258200e-01 -2.04178527e-01 1.03891182e+00 -4.80226785e-01
-2.38242954e-01 5.72107255e-01 -1.79180324e-01 -2.63325095e-01
-6.03212893e-01 -2.14525476e-01 1.92259908e-01 -4.26322132e-01
8.15003932e-01 -7.85994887e-01 -6.44952118e-01 2.82936603e-01
-1.14401376e+00 -6.42777458e-02 6.52221560e-01 7.58834541e-01
2.93636858e-01 1.97236195e-01 6.56749010e-01 -5.56550026e-01
6.53454125e-01 -6.11284435e-01 -1.14949517e-01 1.19115613e-01
-1.05584788e+00 4.82099831e-01 -2.85824426e-02 -2.72475690e-01
-1.22620177e+00 -5.98411858e-01 -3.93413961e-01 -2.33179405e-01
-1.74134418e-01 4.05206800e-01 -2.55091935e-01 1.06864303e-01
8.07503343e-01 3.90258819e-01 1.29187942e-01 -3.67797792e-01
4.97934341e-01 8.68097901e-01 2.89583415e-01 -1.71002790e-01
-1.42503589e-01 2.89524436e-01 -2.71316856e-01 -7.44005501e-01
-7.68976867e-01 -3.64167571e-01 -6.01397514e-01 -3.77461761e-01
1.08654118e+00 -5.53875506e-01 -1.04368639e+00 5.92668474e-01
-9.43070471e-01 -5.70456833e-02 -4.42489535e-01 1.05404198e+00
-9.61722493e-01 3.45436990e-01 -8.53118896e-01 -1.27268803e+00
-2.48348787e-01 -9.22276080e-01 2.51748234e-01 4.66379166e-01
-1.66292861e-01 -1.43423057e+00 1.60703167e-01 2.86876857e-01
5.69808125e-01 -1.41133249e-01 8.47604871e-01 -1.09859884e+00
1.72925293e-02 1.56590510e-02 -2.15613961e-01 9.35264468e-01
1.15811624e-01 -3.58135760e-01 -1.23727226e+00 1.88928634e-01
7.14603543e-01 -8.54166001e-02 1.30716932e+00 7.18981087e-01
9.83333588e-01 -4.54164505e-01 -1.14191316e-01 5.24955869e-01
1.65244043e+00 7.57742107e-01 8.67696226e-01 1.60738304e-01
4.36446190e-01 6.83600545e-01 -1.00500867e-01 4.86128718e-01
2.82320201e-01 8.34629219e-03 2.56694436e-01 3.25277567e-01
1.27653894e-03 -1.28578499e-01 4.61567789e-01 1.47412646e+00
-3.13763529e-01 -1.71157196e-01 -7.18744338e-01 1.61655337e-01
-1.57064199e+00 -1.22406280e+00 1.17123909e-01 2.26797509e+00
9.05011117e-01 1.18805155e-01 -8.52578580e-02 2.35491931e-01
8.63445282e-01 -1.91455811e-01 -6.19629264e-01 -4.25661206e-01
-3.49221498e-01 1.51814476e-01 5.86670160e-01 5.68804324e-01
-1.02380049e+00 7.06408143e-01 7.17956305e+00 1.35962903e+00
-7.79542029e-01 4.03825551e-01 8.21938455e-01 1.69970542e-01
-3.95976245e-01 -1.04531944e-01 -4.48805422e-01 4.95797962e-01
1.35142946e+00 1.39624417e-01 6.27910137e-01 8.96715343e-01
-4.83100414e-01 -5.23667037e-01 -8.67867708e-01 1.22257984e+00
7.87436217e-02 -1.11178780e+00 -1.98052034e-01 2.38573089e-01
6.76358044e-01 2.55696505e-01 2.03752816e-01 3.54022443e-01
5.96370161e-01 -9.77595747e-01 4.41892922e-01 9.77922499e-01
5.26325226e-01 -1.01057613e+00 8.65104735e-01 4.10290450e-01
-7.25544572e-01 -3.48434523e-02 -6.41878009e-01 -3.91356498e-02
-2.85309821e-01 9.62343276e-01 -4.93399471e-01 3.64144713e-01
4.63597119e-01 5.46813846e-01 -1.31613612e-01 8.39858830e-01
4.40554097e-02 4.41660017e-01 -4.88122642e-01 -4.07785743e-01
2.24012241e-01 -4.72748280e-01 4.98738319e-01 1.19158280e+00
2.52837956e-01 -1.74973890e-01 -5.22090673e-01 1.30254102e+00
1.95339635e-01 1.23889357e-01 -5.97853661e-01 -2.39986345e-01
4.03106362e-01 1.01581335e+00 -9.76539969e-01 -3.20586890e-01
-2.82597035e-01 1.09970593e+00 5.30098975e-02 2.05905557e-01
-6.62358582e-01 -4.99160469e-01 4.47839290e-01 -2.77513862e-01
-1.48453832e-01 1.59492329e-01 -3.09148103e-01 -8.86810601e-01
-6.79620802e-01 -2.99815238e-01 3.21202725e-01 -5.59443712e-01
-1.36196804e+00 4.89744097e-01 2.40411624e-01 -9.30464566e-01
-3.62536311e-01 -7.32910156e-01 -1.03936642e-01 7.83158243e-01
-1.00769973e+00 -3.18376809e-01 2.02188149e-01 3.76564771e-01
6.59908056e-02 -3.16445589e-01 9.09857869e-01 2.68523425e-01
-3.00203323e-01 4.94468749e-01 8.62909555e-01 1.03961445e-01
6.78216070e-02 -1.48485184e+00 -1.15586191e-01 3.39715518e-02
2.59870917e-01 4.24285412e-01 5.68449438e-01 -3.94164801e-01
-7.48123288e-01 -4.58770752e-01 4.85829324e-01 -2.20596075e-01
4.88659412e-01 6.80238977e-02 -8.14260542e-01 1.75696731e-01
9.25766826e-02 -6.70879364e-01 1.17650056e+00 -6.77828118e-02
-1.59136876e-01 1.67085499e-01 -1.39809525e+00 3.68779778e-01
9.09631670e-01 -7.52288222e-01 -4.62309092e-01 2.44745806e-01
5.22774637e-01 1.99221358e-01 -1.22202015e+00 4.57512230e-01
7.84286797e-01 -1.43509626e+00 7.43890345e-01 -3.21441323e-01
1.01413704e-01 2.10139945e-01 -4.24513966e-01 -1.26712573e+00
-5.28182387e-01 -6.48654476e-02 -6.55139843e-03 9.52814996e-01
4.31350231e-01 -7.74276912e-01 4.98715132e-01 6.04637563e-01
4.96809632e-02 -8.06750357e-01 -1.04392517e+00 -8.45389962e-01
2.78424650e-01 -7.12537110e-01 3.45338374e-01 1.09758341e+00
4.26338494e-01 1.38860270e-01 7.25216605e-03 -5.63596845e-01
6.05262458e-01 -5.06654143e-01 -1.52529106e-01 -1.46304679e+00
-4.03095782e-01 -9.38985109e-01 -5.93884885e-01 -7.92916179e-01
8.05531591e-02 -1.32235038e+00 -7.92664438e-02 -1.59721684e+00
5.41738331e-01 -2.50408277e-02 -8.93542290e-01 2.33576521e-01
1.35265693e-01 -9.85973999e-02 -8.92398786e-03 2.78843850e-01
-4.47902501e-01 6.28986120e-01 1.04871798e+00 1.59797929e-02
-2.83752620e-01 5.63429184e-02 -6.77491605e-01 1.14026058e+00
1.07674766e+00 -4.52491254e-01 -4.38851595e-01 8.09953734e-02
5.31294763e-01 -9.57330130e-03 1.20596580e-01 -1.35741079e+00
1.56255841e-01 1.23750627e-01 5.67158759e-01 -2.62814581e-01
3.27495575e-01 -5.67689121e-01 3.26384276e-01 5.48781991e-01
-5.50417602e-01 -3.87782186e-01 -4.81751896e-02 3.40822637e-01
-1.70078024e-01 -8.72798860e-01 1.03025663e+00 -5.40839553e-01
-6.16947055e-01 4.68255468e-02 -6.27772152e-01 5.26328338e-03
7.07965136e-01 -4.43585068e-01 -3.82791087e-02 -4.18356091e-01
-1.10875475e+00 -4.38470691e-01 1.25406340e-01 2.16510043e-01
7.30386376e-01 -1.38779318e+00 -4.79882389e-01 7.12756887e-02
-2.86255866e-01 -4.86603588e-01 5.05180001e-01 9.12326992e-01
-4.19782549e-01 4.08028692e-01 -2.87961662e-01 -4.97484952e-01
-4.54181850e-01 3.70182186e-01 6.31753504e-01 -1.54816762e-01
-1.72224596e-01 1.05197954e+00 3.83188754e-01 -4.62355226e-01
1.03855796e-01 8.57730508e-02 -4.83613670e-01 2.24484488e-01
6.00059748e-01 6.62067592e-01 -1.84239760e-01 -4.10083503e-01
-3.52355301e-01 1.26713514e-01 1.77513864e-02 -5.20968616e-01
1.14193213e+00 -2.62335271e-01 -3.57134432e-01 9.52289104e-01
1.58972549e+00 -6.67624712e-01 -9.79818106e-01 -1.31762624e-01
1.92953274e-01 1.29512995e-01 3.33244264e-01 -8.11563253e-01
-1.02146518e+00 9.08064365e-01 1.15247047e+00 3.27720165e-01
9.36725318e-01 2.27354020e-01 3.31419855e-01 3.57663453e-01
1.30587265e-01 -1.58092916e+00 7.28642792e-02 5.34976840e-01
4.73613918e-01 -9.23572779e-01 -3.03190738e-01 1.65843531e-01
-6.65737867e-01 9.89185214e-01 3.76179665e-01 -3.34993284e-03
1.31466973e+00 4.36111927e-01 -2.50900328e-01 -9.24953967e-02
-5.17194092e-01 -5.05315542e-01 3.84738594e-01 6.02202833e-01
7.84167588e-01 3.03393930e-01 -4.37050104e-01 6.85230792e-01
-4.36362982e-01 -1.03179887e-01 -1.09733408e-02 6.01825476e-01
-1.12392902e+00 -6.63970351e-01 -1.66898116e-01 7.00856447e-01
-5.42573035e-01 -2.46573836e-01 -9.01879221e-02 5.18708706e-01
2.87378818e-01 8.92907739e-01 2.84749985e-01 -7.30427921e-01
-5.45415938e-01 3.68050426e-01 7.99366951e-01 -1.50699809e-01
-7.33050033e-02 -1.15739457e-01 -3.65753800e-01 -3.64141554e-01
-6.98224664e-01 -1.97834224e-01 -1.31639397e+00 -3.42481613e-01
-7.23863482e-01 4.56475824e-01 1.27638865e+00 1.27583182e+00
-7.43642598e-02 5.26770353e-01 3.63149464e-01 -6.48317158e-01
-2.91847855e-01 -1.20209551e+00 -1.11008680e+00 4.41863202e-02
-1.89286709e-01 -7.07245290e-01 -5.94542861e-01 -2.75922716e-01] | [7.9361467361450195, 3.588212251663208] |
357dc3d4-5563-47d8-b01a-0ab3f0157c46 | testing-of-detection-tools-for-ai-generated | 2306.15666 | null | https://arxiv.org/abs/2306.15666v2 | https://arxiv.org/pdf/2306.15666v2.pdf | Testing of Detection Tools for AI-Generated Text | Recent advances in generative pre-trained transformer large language models have emphasised the potential risks of unfair use of artificial intelligence (AI) generated content in an academic environment and intensified efforts in searching for solutions to detect such content. The paper examines the general functionality of detection tools for artificial intelligence generated text and evaluates them based on accuracy and error type analysis. Specifically, the study seeks to answer research questions about whether existing detection tools can reliably differentiate between human-written text and ChatGPT-generated text, and whether machine translation and content obfuscation techniques affect the detection of AI-generated text. The research covers 12 publicly available tools and two commercial systems (Turnitin and PlagiarismCheck) that are widely used in the academic setting. The researchers conclude that the available detection tools are neither accurate nor reliable and have a main bias towards classifying the output as human-written rather than detecting AI-generated text. Furthermore, content obfuscation techniques significantly worsen the performance of tools. The study makes several significant contributions. First, it summarises up-to-date similar scientific and non-scientific efforts in the field. Second, it presents the result of one of the most comprehensive tests conducted so far, based on a rigorous research methodology, an original document set, and a broad coverage of tools. Third, it discusses the implications and drawbacks of using detection tools for AI-generated text in academic settings. | ['Lorna Waddington', 'Petr Šigut', 'Olumide Popoola', 'Jean Guerrero-Dib', 'Tomáš Foltýnek', 'Sonja Bjelobaba', 'Alla Anohina-Naumeca', 'Debora Weber-Wulff'] | 2023-06-21 | null | null | null | null | ['machine-translation'] | ['natural-language-processing'] | [ 3.07933360e-01 5.03211617e-01 -1.71158478e-01 2.28047565e-01
-9.43058431e-01 -1.02799320e+00 1.11290169e+00 2.76348025e-01
-1.70683071e-01 5.69571614e-01 3.65157098e-01 -8.53290319e-01
1.03115007e-01 -5.88761628e-01 -3.08270454e-01 -3.45792264e-01
6.15688026e-01 6.38852298e-01 -2.62955993e-01 -5.22511899e-02
9.98174250e-01 3.33652973e-01 -1.32691216e+00 4.44338053e-01
1.01924777e+00 2.59073853e-01 -2.27682784e-01 9.68616962e-01
-4.25261617e-01 1.41786015e+00 -1.64329588e+00 -1.01842535e+00
-8.79464671e-03 -7.61201620e-01 -1.13127613e+00 -6.31741732e-02
7.84942508e-01 -4.17677552e-01 -1.25943720e-01 1.16350198e+00
6.17904603e-01 -4.77608025e-01 6.30223513e-01 -1.34144628e+00
-1.11482084e+00 9.65676069e-01 -6.57169297e-02 5.59218168e-01
7.13693559e-01 4.22377795e-01 6.86340451e-01 -6.42976642e-01
6.98880315e-01 1.44122338e+00 5.43895841e-01 4.10075665e-01
-1.08873594e+00 -9.69882131e-01 -5.74338913e-01 -1.42695069e-01
-1.16408443e+00 -4.95120049e-01 5.91178298e-01 -9.71556783e-01
9.37441945e-01 6.08880937e-01 5.48287988e-01 1.57065248e+00
4.33054984e-01 6.03692830e-01 1.24912548e+00 -8.23519826e-01
8.62455517e-02 7.78186858e-01 1.39410108e-01 6.10998213e-01
6.73599005e-01 -1.83046833e-01 -3.12867701e-01 -6.17951393e-01
3.90349925e-01 -5.63889325e-01 6.02391250e-02 4.47442979e-01
-1.02527928e+00 1.28645408e+00 -3.13793093e-01 8.18524778e-01
-2.15585325e-02 -3.90661865e-01 5.41954160e-01 3.77382278e-01
7.80019045e-01 9.38803077e-01 2.49735117e-01 -7.80223548e-01
-1.14853048e+00 2.70686775e-01 1.30478036e+00 9.52914655e-01
2.08678126e-01 3.99588734e-01 -2.38212869e-01 5.85432887e-01
3.26370448e-01 4.75010961e-01 9.68583286e-01 -6.99908793e-01
7.08531439e-01 6.80017471e-01 -1.47085986e-03 -1.56985378e+00
1.41075611e-01 -3.01514894e-01 -3.01205069e-01 8.86822492e-02
5.74550748e-01 -4.77363229e-01 -2.72364497e-01 8.97749662e-01
-7.40689486e-02 -3.80289584e-01 -6.26235828e-02 4.66413558e-01
1.10583830e+00 5.79192996e-01 9.10450891e-02 1.72404270e-03
1.30006814e+00 -7.79123187e-01 -9.51334476e-01 -4.75274116e-01
1.15650678e+00 -1.40223730e+00 1.16515672e+00 3.05238426e-01
-1.16452289e+00 -2.24284694e-01 -9.62412894e-01 -7.76271746e-02
-6.08589232e-01 2.67110821e-02 2.73341388e-01 1.54527664e+00
-8.61790538e-01 3.62741590e-01 -2.06463560e-01 -4.53350335e-01
4.19644535e-01 -2.64737099e-01 1.01461336e-01 2.44875535e-01
-9.53239262e-01 1.00738347e+00 1.85747311e-01 -2.92352974e-01
-6.18119597e-01 -5.96087158e-01 -5.48900247e-01 -1.32760599e-01
2.46221781e-01 -6.62123412e-02 1.55624866e+00 -1.34725308e+00
-1.42020714e+00 1.27472472e+00 1.25905752e-01 -3.86260778e-01
8.30767453e-01 -2.62111157e-01 -4.78713572e-01 2.42062792e-01
3.57534647e-01 -1.21289089e-01 9.14806247e-01 -9.49532151e-01
-2.64706969e-01 -2.39753753e-01 -2.90438652e-01 -1.30963117e-01
-6.93226218e-01 7.31511712e-01 5.60799122e-01 -1.04404175e+00
-7.07908690e-01 -6.56140387e-01 2.94094622e-01 -6.07827783e-01
-6.79453731e-01 -4.62266117e-01 9.45774674e-01 -8.45213056e-01
1.67423582e+00 -1.84108853e+00 -5.48139036e-01 -3.35547328e-02
5.29701412e-01 7.39871740e-01 -2.38208845e-02 9.12751615e-01
3.75087373e-02 1.09189320e+00 3.37044477e-01 -3.48199606e-02
8.25594664e-02 -2.80565709e-01 -5.95440388e-01 5.47673464e-01
-1.23430537e-02 9.62824941e-01 -8.64292622e-01 -6.53835714e-01
1.37913376e-01 2.56399632e-01 -1.17821582e-01 1.49980128e-01
-4.87669632e-02 -6.73988312e-02 -5.60356081e-01 4.82885778e-01
2.91345507e-01 -1.97482169e-01 -9.55023989e-02 6.56956017e-01
-5.70579350e-01 7.30650485e-01 -5.82029760e-01 6.38606012e-01
-1.73046976e-01 1.27809787e+00 1.82996970e-02 -7.37215698e-01
1.01746404e+00 6.27343416e-01 -1.20244287e-01 -3.73652905e-01
5.19460738e-01 3.60960901e-01 8.64148885e-02 -5.89216292e-01
5.71053982e-01 2.06325009e-01 -7.15403184e-02 9.82097149e-01
-1.35827726e-02 -4.40220445e-01 3.43548618e-02 4.84070361e-01
1.08786249e+00 -2.78240949e-01 3.64333600e-01 -1.19801424e-01
5.55250585e-01 2.96949267e-01 -2.33642265e-01 1.17437816e+00
-3.44624251e-01 1.61379769e-01 5.73768079e-01 -2.37304747e-01
-1.14674103e+00 -3.97555560e-01 -2.11527646e-02 1.18877029e+00
-5.17444313e-01 -8.16894531e-01 -1.31538725e+00 -7.94162750e-01
-2.72943527e-01 1.37975287e+00 -4.50986564e-01 -2.41361201e-01
-3.09908956e-01 -4.43671137e-01 1.12499404e+00 -3.18429023e-02
3.43226224e-01 -1.33244705e+00 -8.51750433e-01 1.81685686e-01
-3.96196991e-01 -9.59201157e-01 -2.01646820e-01 -5.24604201e-01
-7.05762148e-01 -9.39690292e-01 -4.82302219e-01 -6.42653525e-01
3.97255749e-01 2.88273364e-01 1.16004848e+00 2.97537148e-01
-4.92884815e-01 4.96469826e-01 -4.33348596e-01 -9.48054016e-01
-1.40801919e+00 1.53668433e-01 -5.00373840e-01 -4.46850210e-01
1.07684970e+00 -9.27584693e-02 1.30441576e-01 1.31353796e-01
-8.83979321e-01 -2.19598100e-01 5.40436208e-01 5.33675671e-01
-6.12447858e-01 1.02790870e-01 6.14847958e-01 -9.88058865e-01
1.38016140e+00 -5.36652684e-01 -3.82209063e-01 1.24415047e-01
-8.85052323e-01 -3.31563056e-01 7.72815764e-01 -6.10034585e-01
-9.92420852e-01 -6.70223773e-01 8.67634546e-03 -4.49314155e-03
-5.08597374e-01 3.32867682e-01 2.92418689e-01 -2.83738822e-01
1.01054454e+00 1.29166976e-01 8.12335685e-02 -3.10787439e-01
1.23002619e-01 1.15187693e+00 2.33023167e-01 -3.06764007e-01
9.92923081e-01 -3.26262154e-02 -7.06022859e-01 -1.19485712e+00
-6.23196006e-01 -3.31353188e-01 -3.06028545e-01 -4.90752280e-01
4.89398569e-01 -6.02621138e-01 -4.42925274e-01 4.51336503e-01
-1.26323128e+00 -1.78632557e-01 -8.18390474e-02 3.23624313e-01
-3.25223595e-01 5.95865309e-01 -7.54518390e-01 -1.10465765e+00
-7.40232110e-01 -9.19449866e-01 6.15196943e-01 4.11628224e-02
-9.70623314e-01 -9.90505338e-01 -1.17575722e-02 9.26275253e-01
4.30413634e-01 1.47385120e-01 7.81619251e-01 -1.31527698e+00
-1.24138221e-01 -7.63201952e-01 1.93070471e-02 1.24064937e-01
-1.21585637e-01 5.30644238e-01 -9.21783864e-01 -3.95085886e-02
2.87371933e-01 -5.00485599e-01 -6.59786239e-02 -1.25570938e-01
5.24029255e-01 -1.27174675e+00 -8.80721733e-02 -2.05395177e-01
9.61483002e-01 3.15973103e-01 6.58153355e-01 7.54072547e-01
4.88624215e-01 7.91505158e-01 2.65018761e-01 4.45816398e-01
-9.46109071e-02 1.73189402e-01 -4.82916692e-03 4.75467414e-01
4.75267209e-02 -5.32461882e-01 5.79916358e-01 7.98013389e-01
1.67050138e-01 -5.51870227e-01 -9.68473434e-01 4.02299702e-01
-1.41883123e+00 -1.34277713e+00 -4.48446155e-01 2.00966120e+00
7.56527722e-01 3.38633627e-01 3.81769329e-01 1.55811593e-01
7.08749533e-01 -2.07307655e-02 6.61772192e-02 -1.15819764e+00
9.72229466e-02 1.49114653e-01 4.10038412e-01 3.48657757e-01
-7.55830884e-01 9.04132664e-01 7.47844982e+00 9.05065179e-01
-9.88521039e-01 6.63696826e-02 6.71502531e-01 6.57092556e-02
-1.69378713e-01 -1.66457057e-01 -8.27152133e-01 8.11838865e-01
1.65361404e+00 -8.33871484e-01 1.48787528e-01 1.11261642e+00
3.00553173e-01 1.04695782e-01 -8.03013742e-01 6.16171479e-01
6.36184633e-01 -1.29904270e+00 1.69568285e-01 2.54284322e-01
3.86812568e-01 -3.41778785e-01 1.26972914e-01 3.56517762e-01
6.47396386e-01 -1.13570857e+00 1.03466737e+00 -7.90773928e-02
3.67865860e-01 -5.55624843e-01 9.20706332e-01 6.67006731e-01
-2.47644663e-01 3.56025924e-03 -1.17741488e-01 -3.60459864e-01
-2.41478145e-01 2.88034022e-01 -1.28791261e+00 -2.99677197e-02
4.59870964e-01 3.30566823e-01 -1.02039897e+00 7.23012090e-01
-4.23092753e-01 1.24439895e+00 -5.68963680e-03 -6.55572414e-01
3.95871818e-01 -2.69600630e-01 9.39934015e-01 1.58503187e+00
1.07010730e-01 -2.03410476e-01 -1.39679611e-01 1.14349997e+00
2.76592318e-02 2.55060017e-01 -1.11083996e+00 -9.94640291e-01
6.80734158e-01 1.25185192e+00 -7.23156273e-01 -5.03795683e-01
-4.74246740e-01 7.32303262e-01 9.68061164e-02 6.50212914e-02
-5.08476377e-01 -5.12001514e-01 1.33495718e-01 3.95821631e-01
-1.53745830e-01 8.73863511e-03 -6.30629241e-01 -9.69594121e-01
-2.17232004e-01 -1.53435397e+00 3.81261021e-01 -7.15622425e-01
-1.11975408e+00 4.36356634e-01 -5.33649512e-02 -7.60103405e-01
-6.95226252e-01 -3.82949382e-01 -7.10673451e-01 8.04501355e-01
-5.16117632e-01 -8.45957220e-01 -1.24573968e-01 6.27310500e-02
6.85720265e-01 -3.43466014e-01 9.45139229e-01 5.21280542e-02
-7.22068727e-01 7.75673032e-01 1.17330082e-01 5.96403599e-01
7.83343077e-01 -9.04638588e-01 4.50869232e-01 8.98408592e-01
2.88722888e-02 9.37975526e-01 8.77520740e-01 -8.92780006e-01
-1.23060775e+00 -8.23498547e-01 1.22808528e+00 -9.69819486e-01
1.04095948e+00 -4.21895921e-01 -8.62782121e-01 9.03778195e-01
5.26475012e-01 -9.70480859e-01 9.46468592e-01 -3.64610344e-01
-3.48575681e-01 8.04750681e-01 -1.29084504e+00 6.07067287e-01
4.26862836e-01 -6.51309669e-01 -1.05902791e+00 6.48709893e-01
4.45873022e-01 -1.29061714e-01 -7.01057196e-01 -4.46144968e-01
4.74847078e-01 -8.24091613e-01 6.48130119e-01 -7.17567980e-01
9.60626841e-01 5.04534125e-01 4.91166204e-01 -9.47757483e-01
-3.32744926e-01 -1.16189098e+00 -9.17625874e-02 1.64521706e+00
3.33713472e-01 -8.22882235e-01 6.54516280e-01 1.09250140e+00
5.99517748e-02 -3.00152991e-02 -5.79397202e-01 -6.41433477e-01
4.90210503e-01 -9.74836797e-02 1.44679904e-01 1.30358231e+00
3.32553476e-01 6.07954979e-01 -8.68479759e-02 -4.02409971e-01
5.31216621e-01 -1.93809226e-01 1.01668310e+00 -1.24667048e+00
1.69963986e-01 -8.40668201e-01 -4.69075233e-01 -3.67595941e-01
9.06493068e-02 -7.30210483e-01 -3.43586266e-01 -1.17326546e+00
1.09343559e-01 2.02730075e-01 6.29216909e-01 3.17337990e-01
-2.29087956e-02 2.19613254e-01 3.13335329e-01 4.69931304e-01
-2.00690553e-01 -6.41675889e-02 8.59319091e-01 2.17553438e-03
-8.94575715e-02 -7.40401223e-02 -1.06502104e+00 7.75152206e-01
9.73543286e-01 -6.70071781e-01 -2.06334144e-01 -6.56486303e-02
3.54622483e-01 -5.01934528e-01 5.09865522e-01 -8.50802600e-01
-2.13712212e-02 -1.40250579e-01 2.54320562e-01 -3.58794928e-01
-3.17186117e-01 -5.98879516e-01 -1.23623118e-01 4.30730969e-01
-6.20722890e-01 3.74003112e-01 3.10387701e-01 2.76852936e-01
1.08311407e-01 -9.18496132e-01 6.43214881e-01 -3.59277099e-01
-1.26523867e-01 -4.32590216e-01 -1.36726856e+00 4.85427499e-01
1.14647090e+00 -5.22284091e-01 -6.82671607e-01 -7.65197635e-01
1.11450844e-01 -3.97106171e-01 3.67304921e-01 5.98557711e-01
2.68631339e-01 -8.59282613e-01 -8.32873046e-01 7.75270090e-02
2.80639287e-02 -7.26571262e-01 -3.55148673e-01 4.73027587e-01
-7.43253946e-01 7.55306125e-01 -1.29570961e-01 -1.72035452e-02
-1.54202688e+00 6.26504481e-01 8.22074898e-03 -9.84494668e-03
-6.99134111e-01 5.58297098e-01 -2.33819738e-01 -1.40302807e-01
2.35562548e-01 1.41961768e-01 -2.78860956e-01 9.93863344e-02
8.60208511e-01 8.46808612e-01 3.63467112e-02 -9.02384281e-01
1.71011444e-02 -2.32568026e-01 -2.66873807e-01 -2.71295339e-01
8.73833358e-01 -9.28686485e-02 -2.31984258e-01 6.29646003e-01
1.17136955e+00 3.00053686e-01 6.76403800e-03 -2.85876915e-02
6.68805540e-02 -6.37073100e-01 1.28191620e-01 -9.04954612e-01
-4.43416595e-01 8.27216506e-01 1.50239289e-01 9.84359801e-01
3.42494726e-01 -1.65580437e-01 5.59471369e-01 1.85586691e-01
-5.26482686e-02 -1.24235523e+00 3.00129533e-01 4.65079218e-01
9.66490865e-01 -8.36074114e-01 1.42346546e-01 -3.77157092e-01
-5.38809359e-01 1.36358023e+00 6.13613844e-01 2.18406916e-01
2.52621770e-01 3.05072546e-01 1.44136429e-01 -4.00683403e-01
-4.94568706e-01 5.34249187e-01 2.01620758e-01 4.79934126e-01
9.97875392e-01 5.15133105e-02 -7.05448925e-01 3.73472840e-01
-9.89411533e-01 1.25618810e-02 1.00816369e+00 1.10551953e+00
-3.83292377e-01 -7.90812433e-01 -1.04007554e+00 6.62876546e-01
-1.13941693e+00 -1.71044499e-01 -1.47918248e+00 8.33306670e-01
-4.05059010e-01 1.58455896e+00 -1.56489685e-02 -3.83920312e-01
-1.84426665e-01 4.46919173e-01 -1.13968581e-01 -6.77136242e-01
-1.23878777e+00 -1.52013704e-01 4.38986659e-01 -1.00831874e-01
-1.54125184e-01 -6.47109509e-01 -5.07404029e-01 -9.13407564e-01
-5.56665838e-01 4.19120878e-01 5.65878868e-01 8.04315150e-01
5.01366138e-01 9.66706574e-02 1.42631471e-01 -3.97777617e-01
-7.03046143e-01 -1.30311120e+00 -5.94304539e-02 3.94837111e-01
1.57113820e-01 -1.06769070e-01 -7.67496884e-01 1.14405863e-01] | [8.5275239944458, 10.01546859741211] |
77deb21c-3359-41e7-97e2-fb5fb99c41c0 | unsupervised-visible-infrared-person-re | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wu_Unsupervised_Visible-Infrared_Person_Re-Identification_via_Progressive_Graph_Matching_and_Alternate_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wu_Unsupervised_Visible-Infrared_Person_Re-Identification_via_Progressive_Graph_Matching_and_Alternate_CVPR_2023_paper.pdf | Unsupervised Visible-Infrared Person Re-Identification via Progressive Graph Matching and Alternate Learning | Unsupervised visible-infrared person re-identification is a challenging task due to the large modality gap and the unavailability of cross-modality correspondences. Cross-modality correspondences are very crucial to bridge the modality gap. Some existing works try to mine cross-modality correspondences, but they focus only on local information. They do not fully exploit the global relationship across identities, thus limiting the quality of the mined correspondences. Worse still, the number of clusters of the two modalities is often inconsistent, exacerbating the unreliability of the generated correspondences. In response, we devise a Progressive Graph Matching method to globally mine cross-modality correspondences under cluster imbalance scenarios. PGM formulates correspondences mining as a graph matching process and considers the global information by minimizing the global matching cost, where the matching cost measures the dissimilarity of clusters. Besides, PGM adopts a progressive strategy to address the imbalance issue with multiple dynamic matching processes. Based on PGM, we design an Alternate Cross Contrastive Learning (ACCL) module to reduce the modality gap with the mined cross-modality correspondences, while mitigating the effect of noise in correspondences through an alternate scheme. Extensive experiments demonstrate the reliability of the generated correspondences and the effectiveness of our method. | ['Mang Ye', 'Zesen Wu'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['person-re-identification', 'graph-matching'] | ['computer-vision', 'graphs'] | [ 1.99105665e-01 -2.35427186e-01 -2.04532132e-01 -1.86730638e-01
-7.13386297e-01 -2.94571042e-01 4.79062051e-01 1.54273286e-01
-1.51429832e-01 3.36392730e-01 3.69245827e-01 2.21931145e-01
-5.11406302e-01 -8.43808413e-01 -3.56599212e-01 -6.44632459e-01
5.00047028e-01 1.69644475e-01 6.99893087e-02 -2.16257408e-01
2.90031046e-01 -1.12213353e-02 -1.50241411e+00 2.38807499e-01
1.17225683e+00 7.46149957e-01 8.99421126e-02 -7.90524110e-02
-3.73711407e-01 2.86015123e-01 -3.01597059e-01 -5.50471902e-01
4.15232420e-01 -7.75151670e-01 -6.37595654e-01 3.14105660e-01
4.77939159e-01 1.83629945e-01 -3.86142939e-01 1.39087129e+00
6.15185916e-01 1.77765433e-02 3.78945529e-01 -1.66680264e+00
-3.87932301e-01 3.27768862e-01 -1.13736761e+00 8.92227590e-02
6.36141062e-01 7.61911320e-03 1.06951809e+00 -8.22474003e-01
4.06049907e-01 1.20827985e+00 8.50589991e-01 3.72851104e-01
-1.22227466e+00 -7.86190987e-01 3.54034543e-01 4.59931195e-01
-1.58846152e+00 -4.82605070e-01 1.07734835e+00 -4.13290292e-01
3.48593444e-01 3.55966419e-01 5.44793367e-01 8.41068149e-01
-3.64839286e-01 3.65996569e-01 1.15683913e+00 -6.91132486e-01
-1.21905446e-01 1.35796577e-01 3.24547514e-02 6.39862657e-01
2.37820789e-01 -1.14447344e-02 -8.36188674e-01 -1.88388526e-01
4.85288948e-01 1.74851060e-01 -2.40318984e-01 -2.68888354e-01
-1.44639564e+00 4.64737684e-01 3.68710488e-01 4.08487469e-01
-1.46160141e-01 -5.50687075e-01 3.11889708e-01 4.08262134e-01
3.11232179e-01 1.40171349e-01 1.02976821e-01 1.12802804e-01
-8.35198402e-01 -1.55169452e-02 3.79779965e-01 9.27309334e-01
9.63246226e-01 -4.69876051e-01 -6.29479736e-02 1.06699646e+00
3.48321825e-01 2.28771999e-01 4.44544792e-01 -5.39473355e-01
8.81752312e-01 1.15802324e+00 -2.07892746e-01 -1.60515225e+00
-3.87818187e-01 -3.47622156e-01 -1.16728377e+00 -1.38904914e-01
5.28988719e-01 -2.22249720e-02 -4.25793767e-01 1.86668861e+00
5.47585189e-01 2.83334285e-01 -2.90311754e-01 1.02015197e+00
8.46873105e-01 2.71728605e-01 1.29011661e-01 -4.04534280e-01
1.20875287e+00 -7.96000719e-01 -7.38851070e-01 -1.53493717e-01
4.59073246e-01 -8.02331209e-01 8.57287943e-01 -5.11948727e-02
-8.67937446e-01 -6.79932177e-01 -9.11467433e-01 1.83772177e-01
-1.26357675e-01 -4.17243019e-02 3.46567571e-01 4.75480229e-01
-7.35724568e-01 3.45580727e-01 -2.88133562e-01 -3.97547185e-01
1.47340283e-01 3.88507247e-01 -5.78245163e-01 -2.24354669e-01
-1.21575260e+00 5.90771139e-01 5.68648756e-01 4.00347352e-01
1.90169886e-01 -6.04707837e-01 -6.20295823e-01 -2.10625440e-01
5.19832730e-01 -7.16188908e-01 3.97607297e-01 -1.06866384e+00
-1.02076066e+00 8.57085288e-01 -2.06901491e-01 1.79752529e-01
6.85098231e-01 3.18597943e-01 -7.57799864e-01 7.32013062e-02
2.01508820e-01 4.14564103e-01 6.65373445e-01 -1.34569502e+00
-8.84373009e-01 -6.48241818e-01 -1.49891093e-01 6.03460014e-01
-6.44088209e-01 -2.90875554e-01 -7.33656287e-01 -6.65748656e-01
6.17974579e-01 -9.32550609e-01 -3.96712497e-02 -5.44406846e-02
-5.62017560e-01 -3.27088565e-01 5.78045905e-01 -7.62607753e-01
1.30147445e+00 -2.23982215e+00 2.15628743e-01 6.65078461e-01
3.90997499e-01 -8.53651389e-02 -3.03145051e-01 4.38347757e-01
-1.24618605e-01 -1.64459407e-01 -2.49068007e-01 -3.68621647e-01
-1.28128588e-01 -1.56264640e-02 1.19225442e-01 5.39836049e-01
-1.73265308e-01 7.01436400e-01 -8.10329556e-01 -7.67915308e-01
1.62454441e-01 2.30910048e-01 -2.07474753e-01 1.40615538e-01
1.86167911e-01 6.89495742e-01 -3.16004723e-01 6.37583017e-01
9.59118783e-01 -2.71079421e-01 3.88009459e-01 -6.49323225e-01
1.81827154e-02 -1.99910834e-01 -1.67092586e+00 1.82489610e+00
-1.92576885e-01 1.92094252e-01 -7.50575811e-02 -1.17782652e+00
1.03694904e+00 6.64372370e-02 9.80109870e-01 -1.04575217e+00
-8.32834616e-02 2.44891539e-01 -1.42146289e-01 -5.32933831e-01
2.99060524e-01 -1.88670576e-01 -1.63880941e-02 4.03236717e-01
-2.29222357e-01 7.07126796e-01 4.33855206e-02 2.21006155e-01
6.48929715e-01 -3.02618109e-02 1.02973133e-01 -7.24310428e-03
7.39251018e-01 -1.60549283e-01 8.33562016e-01 5.66161394e-01
-2.06654429e-01 7.81594813e-01 1.78308964e-01 -1.41775176e-01
-8.87600899e-01 -9.51528311e-01 8.51531699e-02 7.26402819e-01
8.30835998e-01 -6.02197945e-01 -4.84362751e-01 -6.73854232e-01
-4.02707718e-02 1.80008020e-02 -4.58127111e-01 -3.11923802e-01
-5.67779839e-01 -8.75059664e-01 3.76991034e-01 2.36102521e-01
8.23253155e-01 -7.27711558e-01 1.12675130e-01 -2.24628784e-02
-8.91321003e-01 -1.11733198e+00 -7.03805327e-01 -6.71062648e-01
-7.48092353e-01 -1.25747132e+00 -5.06808758e-01 -9.03166056e-01
9.10647273e-01 7.27700293e-01 9.17893589e-01 4.70796853e-01
-8.72600451e-02 3.59287709e-01 -3.47784251e-01 7.82327279e-02
-2.52550602e-01 9.47063789e-02 6.09495975e-02 4.44142759e-01
7.00921893e-01 -7.09496856e-01 -6.34001136e-01 4.17560041e-01
-5.70994854e-01 2.25882336e-01 4.32641834e-01 9.93441105e-01
6.48929298e-01 4.15088177e-01 6.98146820e-01 -7.31167793e-01
6.09147549e-01 -4.50048774e-01 -3.90929550e-01 6.06933355e-01
-8.67195487e-01 -7.89494291e-02 4.48405594e-01 -4.72435027e-01
-1.09521449e+00 9.15902779e-02 8.36579353e-02 -2.52424359e-01
-3.49211581e-02 4.53421354e-01 -4.95663792e-01 -2.59802788e-01
3.97772342e-01 2.49146849e-01 1.50779530e-01 -5.13065875e-01
2.09873885e-01 7.32295215e-01 6.35593772e-01 -4.72515523e-01
1.10655475e+00 5.60540974e-01 4.80843857e-02 -4.43416655e-01
-6.80843949e-01 -7.96134770e-01 -7.89734483e-01 -3.66829693e-01
5.42313933e-01 -1.02935064e+00 -8.15017879e-01 5.36556304e-01
-7.96420932e-01 3.46056134e-01 -5.95655479e-02 4.28039193e-01
-1.98716074e-01 9.52400267e-01 -3.39066535e-01 -6.41974866e-01
-2.82146513e-01 -8.70135903e-01 8.66807103e-01 4.08489525e-01
-1.13709293e-01 -8.67187738e-01 4.82800230e-02 6.91275299e-01
8.20207521e-02 2.83826411e-01 7.55994022e-01 -2.72526979e-01
-4.67720389e-01 -2.11715877e-01 -5.08356512e-01 -2.11418182e-01
5.48275530e-01 -4.36681896e-01 -7.62266040e-01 -4.24184293e-01
-2.43396491e-01 1.10674404e-01 6.31611884e-01 -2.24894211e-02
9.27398980e-01 -1.56649008e-01 -4.07443017e-01 4.25076306e-01
1.41020072e+00 -1.04485102e-01 5.87141931e-01 5.38478255e-01
1.17628407e+00 1.03231919e+00 6.93080783e-01 4.12132919e-01
7.33618736e-01 9.19009089e-01 9.37704816e-02 -1.75546438e-01
-1.70800984e-01 -4.96592671e-01 7.43538439e-02 1.04344726e+00
-1.72059953e-01 8.97888094e-02 -6.81213856e-01 5.14435768e-01
-2.16220069e+00 -1.08638918e+00 -2.65635699e-01 2.47408295e+00
8.47690821e-01 -3.90329733e-02 4.17794049e-01 3.42040509e-01
1.19550633e+00 -9.06446055e-02 -3.04361641e-01 3.37000757e-01
-3.07144284e-01 -3.58417928e-01 1.51951402e-01 2.80512214e-01
-8.98261189e-01 5.21974862e-01 5.25460434e+00 8.61762404e-01
-8.38593543e-01 4.90438007e-02 2.39381179e-01 1.74622700e-01
-4.50589865e-01 1.20500319e-01 -6.25903070e-01 8.75075281e-01
3.07227492e-01 -6.08991878e-03 4.18889523e-01 2.74185985e-01
5.73439747e-02 -1.79336190e-01 -9.39492881e-01 1.46574295e+00
1.43157020e-01 -1.06295860e+00 3.29065346e-03 2.13862836e-01
7.05748737e-01 -4.53795820e-01 -1.96955666e-01 -9.06883031e-02
-1.69131875e-01 -6.80237293e-01 5.12753367e-01 6.78728461e-01
7.55548418e-01 -8.99745226e-01 6.31666422e-01 4.24673647e-01
-1.48007488e+00 -1.28769636e-01 -2.45683700e-01 1.89900964e-01
4.67473008e-02 7.74569988e-01 -2.85147667e-01 1.12731242e+00
7.15711296e-01 8.03458393e-01 -8.32243919e-01 1.06569064e+00
1.95663080e-01 1.16861440e-01 -2.68608302e-01 4.72102821e-01
-3.35268259e-01 -6.30164981e-01 6.02650404e-01 9.47677076e-01
2.13470519e-01 3.40911262e-02 4.87113416e-01 6.66244686e-01
3.33409943e-03 3.11617076e-01 -4.43450123e-01 2.61583358e-01
7.43342400e-01 1.15058243e+00 -5.06963551e-01 -4.78911139e-02
-5.83389997e-01 9.73625302e-01 2.45073289e-01 1.62301525e-01
-5.89941621e-01 -1.76686287e-01 3.63099635e-01 1.80354998e-01
-2.90726781e-01 -6.80043027e-02 -4.90865260e-01 -1.34305036e+00
4.76960748e-01 -9.69672561e-01 9.32639778e-01 -3.10158759e-01
-1.85915124e+00 2.88055480e-01 -6.68980032e-02 -1.65094745e+00
4.22642119e-02 1.95659965e-01 -4.92967993e-01 8.98572683e-01
-1.54719305e+00 -1.37918675e+00 -6.87505066e-01 8.63862395e-01
2.72010773e-01 -1.96584657e-01 5.05324364e-01 6.79980218e-01
-7.57205129e-01 1.12393403e+00 -1.59699008e-01 1.14919566e-01
9.28698063e-01 -8.78451109e-01 3.36486921e-02 9.53061521e-01
7.77999759e-02 6.53601289e-01 4.93884414e-01 -7.80844390e-01
-1.33204377e+00 -9.35359716e-01 9.14666951e-01 -1.47790924e-01
3.36347848e-01 -2.73591578e-01 -1.12499332e+00 2.33400673e-01
2.15093722e-03 -1.77172020e-01 7.24638462e-01 4.05436844e-01
-5.24865806e-01 -3.82251769e-01 -1.03658140e+00 5.43809474e-01
1.44047117e+00 -6.44643009e-01 -3.96777630e-01 1.87250331e-01
3.29120845e-01 -2.37245291e-01 -9.69901383e-01 5.39758801e-01
4.72868651e-01 -9.60318506e-01 9.34590101e-01 -2.06842288e-01
1.03031747e-01 -7.15427876e-01 1.11709237e-01 -1.05566704e+00
-4.97839034e-01 -3.28691334e-01 7.70435557e-02 1.87745237e+00
1.55658573e-01 -6.73638105e-01 6.78503811e-01 7.25033879e-01
3.18754941e-01 -2.51259744e-01 -9.05547738e-01 -7.29291797e-01
-3.63026887e-01 1.99737083e-02 6.98581874e-01 1.39131391e+00
2.04540953e-01 3.20836633e-01 -5.70529997e-01 2.53565431e-01
9.23323870e-01 5.08218467e-01 7.61364520e-01 -1.35695088e+00
-2.64916301e-01 -3.63503426e-01 -4.31627899e-01 -6.97730899e-01
9.27488953e-02 -8.99299502e-01 4.04726304e-02 -1.36867869e+00
7.04977930e-01 -8.58016551e-01 -3.71438950e-01 3.67869884e-01
-4.85870361e-01 2.93739021e-01 3.19436699e-01 7.17752397e-01
-6.36922836e-01 4.10326481e-01 1.13929784e+00 -2.12437645e-01
-3.26838166e-01 -8.94284174e-02 -9.04228866e-01 4.71050382e-01
6.91697538e-01 -2.73118585e-01 -5.39739788e-01 -2.48798683e-01
2.21325859e-01 -8.88944939e-02 4.04642254e-01 -9.86383140e-01
7.32572675e-01 -3.19130898e-01 4.11554813e-01 -5.84954739e-01
1.19983114e-01 -9.45659339e-01 4.60677892e-01 5.75074330e-02
-1.40288368e-01 1.67935535e-01 -2.57506192e-01 8.71277332e-01
-3.74899566e-01 8.43016133e-02 6.80463910e-01 -1.38602063e-01
-8.16254795e-01 4.19104695e-01 2.78151840e-01 9.42151528e-04
8.38299870e-01 -5.91155410e-01 -2.54609257e-01 -2.11024970e-01
-5.79508483e-01 4.17176783e-01 6.43475294e-01 7.05665112e-01
5.30502498e-01 -1.65892446e+00 -6.80192053e-01 1.89867094e-01
4.16470766e-01 -9.86032113e-02 5.92829823e-01 1.06179595e+00
1.19857028e-01 -1.61840364e-01 -2.06766173e-01 -6.30947232e-01
-1.45855010e+00 5.64176023e-01 3.81715268e-01 -1.75877079e-01
-7.08830774e-01 5.03119171e-01 1.06121816e-01 -4.19173747e-01
1.72852531e-01 6.74048841e-01 -4.10990179e-01 3.21809500e-01
4.90928054e-01 4.40225035e-01 9.76071507e-02 -8.83849204e-01
-4.77425694e-01 9.36163545e-01 7.89193362e-02 4.93089817e-02
1.12151527e+00 -6.96674287e-01 -4.19285834e-01 3.22088629e-01
1.07732022e+00 1.06007963e-01 -9.62365270e-01 -6.93668365e-01
3.64179548e-04 -8.38222504e-01 -3.27700347e-01 -4.45464730e-01
-1.20577395e+00 4.56755161e-01 6.90667033e-01 7.78400898e-02
1.44766366e+00 -1.94012389e-01 9.67528224e-01 -1.44909501e-01
2.98795372e-01 -1.33165157e+00 2.18487740e-01 -4.67426069e-02
4.05670345e-01 -1.46118963e+00 1.93080828e-01 -9.10980463e-01
-5.47930777e-01 1.00561428e+00 8.43661487e-01 3.95267636e-01
4.64698642e-01 -1.34382278e-01 -1.06864505e-01 -1.74079910e-01
-1.89974025e-01 -2.41496503e-01 4.56237346e-01 6.72480762e-01
1.37444511e-01 5.93599584e-03 -5.20375729e-01 3.05475891e-01
1.99796464e-02 -2.00420514e-01 1.64102256e-01 7.46289313e-01
-8.48596245e-02 -1.38609111e+00 -5.90773642e-01 3.44757468e-01
-2.03134604e-02 1.34675518e-01 -4.42713141e-01 4.42717999e-01
4.29158658e-01 1.27225697e+00 -1.31417662e-01 -7.11736619e-01
4.45458055e-01 -2.18952104e-01 4.23021585e-01 -1.67566136e-01
-3.99067014e-01 1.36043534e-01 -5.28542511e-03 -2.66889215e-01
-9.18443024e-01 -7.37561703e-01 -9.84209180e-01 -6.03907883e-01
-5.30336797e-01 7.08033815e-02 1.95693865e-01 1.05250752e+00
4.93942618e-01 1.71251357e-01 9.87913370e-01 -2.31865942e-01
-3.09412420e-01 -6.69051826e-01 -4.67335075e-01 9.77370441e-01
1.39228940e-01 -6.36115015e-01 -2.05803096e-01 5.94852380e-02] | [14.72988224029541, 0.9946154952049255] |
f7e766d1-762a-421a-92b9-aee94bd4db00 | decomposed-knowledge-distillation-for-class | 2210.05941 | null | https://arxiv.org/abs/2210.05941v1 | https://arxiv.org/pdf/2210.05941v1.pdf | Decomposed Knowledge Distillation for Class-Incremental Semantic Segmentation | Class-incremental semantic segmentation (CISS) labels each pixel of an image with a corresponding object/stuff class continually. To this end, it is crucial to learn novel classes incrementally without forgetting previously learned knowledge. Current CISS methods typically use a knowledge distillation (KD) technique for preserving classifier logits, or freeze a feature extractor, to avoid the forgetting problem. The strong constraints, however, prevent learning discriminative features for novel classes. We introduce a CISS framework that alleviates the forgetting problem and facilitates learning novel classes effectively. We have found that a logit can be decomposed into two terms. They quantify how likely an input belongs to a particular class or not, providing a clue for a reasoning process of a model. The KD technique, in this context, preserves the sum of two terms (i.e., a class logit), suggesting that each could be changed and thus the KD does not imitate the reasoning process. To impose constraints on each term explicitly, we propose a new decomposed knowledge distillation (DKD) technique, improving the rigidity of a model and addressing the forgetting problem more effectively. We also introduce a novel initialization method to train new classifiers for novel classes. In CISS, the number of negative training samples for novel classes is not sufficient to discriminate old classes. To mitigate this, we propose to transfer knowledge of negatives to the classifiers successively using an auxiliary classifier, boosting the performance significantly. Experimental results on standard CISS benchmarks demonstrate the effectiveness of our framework. | ['Bumsub Ham', 'Junghyup Lee', 'SangHoon Lee', 'Youngmin Oh', 'Donghyeon Baek'] | 2022-10-12 | null | null | null | null | ['class-incremental-semantic-segmentation'] | ['computer-vision'] | [ 5.92276931e-01 2.43533224e-01 -2.52017498e-01 -4.48223531e-01
-2.71011710e-01 -8.27932596e-01 3.48136276e-01 3.09368551e-01
-4.86129314e-01 9.02819276e-01 -4.74256426e-01 -2.47093230e-01
6.99013355e-04 -9.89735961e-01 -9.54710066e-01 -9.84012425e-01
2.76807934e-01 2.59279698e-01 7.60360837e-01 1.69486806e-01
2.48006746e-01 5.52446544e-01 -1.64235520e+00 3.75131160e-01
1.03464210e+00 9.83886003e-01 2.75423735e-01 3.54808033e-01
-2.42159441e-01 6.20680749e-01 -6.64457142e-01 -4.55068499e-01
3.33184153e-01 -3.17005545e-01 -9.25130367e-01 1.22752145e-03
4.54652756e-01 -1.79238647e-01 1.40526295e-01 1.09292865e+00
-1.38918146e-01 2.00477362e-01 5.21213531e-01 -1.30633187e+00
-5.47416866e-01 6.07150137e-01 -6.56077802e-01 6.12408444e-02
-8.99952129e-02 1.28745511e-01 8.53233337e-01 -8.25446129e-01
6.64148152e-01 9.22797024e-01 7.43957341e-01 5.58856368e-01
-1.20650971e+00 -6.62784457e-01 6.36333525e-01 2.90989578e-01
-1.40954161e+00 -1.46970972e-01 8.94720316e-01 -4.24825400e-01
4.67381924e-01 4.53395247e-01 9.37993169e-01 7.87212968e-01
-3.64038385e-02 1.08915484e+00 1.20948160e+00 -4.72168446e-01
6.21728361e-01 4.37355369e-01 4.45230722e-01 7.45467484e-01
3.88844043e-01 -2.35083327e-01 -4.82236415e-01 6.18377998e-02
5.23925245e-01 2.01342270e-01 -2.46845767e-01 -7.82061338e-01
-8.13296020e-01 6.06194019e-01 5.36123693e-01 3.07175279e-01
3.76610719e-02 -2.05876336e-01 3.40625435e-01 3.19455862e-01
3.03717375e-01 4.35915232e-01 -9.22824860e-01 7.45999590e-02
-8.10452700e-01 5.62983984e-03 6.10561728e-01 6.77089930e-01
1.38096833e+00 -3.56619269e-01 -1.55801207e-01 7.03574717e-01
-1.53373063e-01 2.10739136e-01 4.69249129e-01 -8.69936705e-01
7.53275752e-02 9.87784505e-01 -1.14846796e-01 -8.63646686e-01
-1.71150804e-01 -5.56994677e-01 -7.28992164e-01 2.21907243e-01
4.81473565e-01 2.54409343e-01 -1.19391990e+00 1.90011585e+00
5.36028385e-01 3.67268413e-01 5.85753992e-02 4.73642856e-01
1.93078622e-01 4.07264918e-01 -2.34248042e-02 -3.36015493e-01
9.37984765e-01 -9.63570178e-01 -3.54740739e-01 -3.45371038e-01
5.07225513e-01 -2.91241437e-01 1.29302740e+00 4.30029213e-01
-7.14389443e-01 -5.31951010e-01 -1.16938353e+00 5.75770214e-02
-8.01011443e-01 -1.11048244e-01 8.15041482e-01 6.33241296e-01
-6.66383624e-01 7.24678934e-01 -1.05252981e+00 -7.96754733e-02
6.13428771e-01 3.76441091e-01 -1.89955652e-01 -9.09520239e-02
-1.04097652e+00 6.87607467e-01 7.42314875e-01 -1.13078423e-01
-4.82472599e-01 -8.43262494e-01 -8.31004441e-01 2.37489656e-01
7.52075374e-01 -7.10430741e-01 1.13554239e+00 -1.42937732e+00
-1.32300222e+00 6.74401939e-01 -1.84688181e-01 -3.99353117e-01
5.22612453e-01 -1.65780373e-02 -2.32290879e-01 1.52446985e-01
1.79130375e-01 7.17152357e-01 1.25768268e+00 -1.45783603e+00
-8.23090374e-01 -4.18045938e-01 3.63896847e-01 4.92844060e-02
-5.52337527e-01 -7.02540576e-01 -4.78922814e-01 -6.34329617e-01
3.99121642e-01 -9.86560225e-01 -2.10740790e-03 2.59727687e-01
-2.49143079e-01 -1.37799114e-01 8.56153846e-01 -2.20625550e-01
1.22229111e+00 -2.30991626e+00 4.98643741e-02 3.60500723e-01
1.67948321e-01 3.88594151e-01 5.49593940e-02 -2.58910984e-01
-8.18278193e-02 1.26807526e-01 -7.01036036e-01 -2.19044775e-01
-3.27042937e-01 7.82695889e-01 -3.41657668e-01 4.04977938e-03
4.97874618e-01 8.13836753e-01 -1.05261421e+00 -3.22653711e-01
5.51844276e-02 3.03425491e-01 -6.71920776e-01 -1.66948736e-01
-3.48262191e-01 3.28116745e-01 -2.40301132e-01 5.97660124e-01
8.90718937e-01 -3.20213944e-01 1.82172865e-01 -2.01342791e-01
1.63311772e-02 -5.09199910e-02 -1.32159281e+00 1.62632251e+00
-3.10868382e-01 1.55674636e-01 -3.22038054e-01 -1.10348439e+00
6.14465296e-01 -2.63670355e-01 1.64542198e-01 -5.67063630e-01
-1.20939128e-01 2.14245707e-01 -2.30555758e-01 -2.52262533e-01
3.28199565e-01 -2.90692896e-01 8.40909854e-02 6.77569509e-02
1.00789629e-01 1.22025982e-01 1.80019289e-01 1.72282472e-01
1.00449467e+00 1.91419736e-01 2.10566431e-01 -2.50099540e-01
4.37789768e-01 -7.52841530e-04 1.05521476e+00 1.02458620e+00
-1.52736232e-01 4.40716326e-01 5.03192842e-01 -4.56680238e-01
-5.48531055e-01 -1.32719088e+00 7.14822765e-03 1.00967789e+00
2.98528880e-01 -2.68596649e-01 -7.44723916e-01 -1.24545777e+00
1.02071397e-01 8.21284413e-01 -8.58945370e-01 -5.40757895e-01
-5.12016475e-01 -7.05355525e-01 1.90178409e-01 6.85541987e-01
4.35446888e-01 -6.96491122e-01 -8.10656905e-01 1.47490695e-01
-9.02610570e-02 -8.61304700e-01 -3.40939343e-01 6.05501354e-01
-1.03862345e+00 -1.34349763e+00 -6.58683479e-01 -7.65244186e-01
1.06739986e+00 3.32196027e-01 7.22370803e-01 9.29783136e-02
-2.52762198e-01 3.41155112e-01 -3.68886083e-01 -1.73998401e-01
-2.94359028e-01 1.76961556e-01 -2.08728190e-04 1.54187754e-01
2.00682729e-01 -6.36521637e-01 -4.52815562e-01 1.22643642e-01
-1.03971207e+00 3.05606723e-01 5.89726806e-01 1.04115021e+00
6.46467686e-01 4.54073161e-01 5.39905250e-01 -1.15107346e+00
-4.00629193e-02 -3.00622374e-01 -5.03626049e-01 4.76188540e-01
-8.39664757e-01 3.37382823e-01 7.62347460e-01 -8.13694239e-01
-1.19098377e+00 3.09459329e-01 9.89987403e-02 -3.84542912e-01
-4.58325557e-02 4.32300538e-01 -2.55308598e-01 -7.98450336e-02
4.29272622e-01 2.50934869e-01 -1.97785869e-01 -4.68261421e-01
4.27375525e-01 2.24855378e-01 6.21655345e-01 -8.13091874e-01
8.06236386e-01 6.48126185e-01 -1.59664035e-01 -5.27518690e-01
-1.09713221e+00 -3.99574637e-01 -8.83260548e-01 -3.49147581e-02
4.57667619e-01 -5.75881004e-01 -6.06476784e-01 6.25033438e-01
-9.53691304e-01 -3.33688885e-01 -7.14200735e-01 1.74915493e-01
-2.03523353e-01 3.81187797e-01 -3.35194737e-01 -6.63555205e-01
7.19008818e-02 -9.05037344e-01 6.74360931e-01 4.11203533e-01
-7.91057423e-02 -8.44312966e-01 -1.83386207e-01 1.86589822e-01
1.95443332e-01 1.86631203e-01 1.24276531e+00 -4.63476688e-01
-5.23908198e-01 2.33085416e-02 -2.63841093e-01 5.85766196e-01
2.87828773e-01 -6.54586032e-02 -1.07749569e+00 -3.65062892e-01
9.32872221e-02 -1.07112996e-01 1.33664072e+00 -1.82284161e-01
1.42698228e+00 -4.74776685e-01 -3.81497383e-01 3.84015292e-01
1.37044024e+00 4.58442807e-01 5.32516837e-01 4.03443158e-01
7.05514073e-01 3.15715224e-01 4.50868756e-01 3.05722237e-01
3.40597093e-01 3.54437679e-01 1.96191594e-01 4.98755313e-02
-1.88170031e-01 -3.16326708e-01 3.00107121e-01 7.08104670e-01
2.11500093e-01 2.68815085e-02 -8.64014387e-01 5.47448158e-01
-1.85245979e+00 -5.54031789e-01 2.08448797e-01 2.28477001e+00
1.29668069e+00 4.98898655e-01 -3.43692303e-01 4.91391987e-01
6.06000006e-01 -1.98928937e-01 -1.04733884e+00 -1.27258942e-01
-1.09456249e-01 4.19111848e-01 3.57665390e-01 5.45299470e-01
-8.84853184e-01 1.03181672e+00 5.44104767e+00 8.87967527e-01
-1.21185207e+00 -3.64163257e-02 5.75308681e-01 1.48348227e-01
-3.68099123e-01 3.51319075e-01 -7.72846699e-01 3.25906485e-01
2.88639545e-01 -6.77863359e-02 4.28978473e-01 9.22600329e-01
-3.57258111e-01 -5.39002717e-01 -1.24595165e+00 7.64235020e-01
6.13086261e-02 -1.09191108e+00 3.27857912e-01 -4.23454911e-01
7.58535385e-01 -4.99742210e-01 1.22044481e-01 4.42870021e-01
8.19205865e-02 -4.20726120e-01 8.45046759e-01 5.13910234e-01
4.55612183e-01 -6.84846401e-01 4.38476562e-01 3.54228169e-01
-1.12362075e+00 -2.36785248e-01 -1.11301787e-01 -1.30664796e-01
-3.77639562e-01 8.06270838e-01 -8.50338936e-01 5.13124168e-01
8.68617833e-01 6.71053946e-01 -1.04923820e+00 7.83162475e-01
-2.84577906e-01 4.44051772e-01 -2.84417838e-01 4.14559215e-01
8.68366584e-02 -2.28819605e-02 3.74896854e-01 8.81521344e-01
4.70347218e-02 5.17627820e-02 1.32713437e-01 8.62978816e-01
-6.97066858e-02 -2.15899184e-01 -2.23608211e-01 1.55021787e-01
3.76675516e-01 9.77673590e-01 -1.14142644e+00 -4.77208853e-01
-2.70851493e-01 1.30686724e+00 4.52209383e-01 3.62108201e-01
-6.77486122e-01 -3.41207683e-01 4.53635663e-01 4.34620529e-02
4.91557032e-01 -7.98633620e-02 -5.59457362e-01 -1.28047657e+00
4.17999357e-01 -5.33618450e-01 5.60404122e-01 -6.48755908e-01
-1.23125517e+00 1.66016772e-01 -4.57452163e-02 -8.54623675e-01
9.06446278e-02 -5.15608191e-01 -2.58899242e-01 3.74989808e-01
-1.87387097e+00 -1.07640171e+00 -3.28691185e-01 6.18023753e-01
4.22281444e-01 2.41844490e-01 7.34863102e-01 2.48212069e-01
-6.02685869e-01 7.07598329e-01 3.70053807e-03 -1.01744652e-01
8.12770545e-01 -1.29008555e+00 -4.74519422e-03 8.47275794e-01
6.90252008e-03 8.34056377e-01 4.98876631e-01 -6.53697789e-01
-1.09707999e+00 -1.20189452e+00 6.39348924e-01 -1.88871816e-01
5.26183367e-01 -3.48773867e-01 -1.34796727e+00 7.18198121e-01
-4.14609104e-01 4.67080716e-03 6.18112147e-01 1.96039349e-01
-7.04425395e-01 -3.81415129e-01 -1.22374785e+00 5.32960534e-01
1.00183010e+00 -5.11357725e-01 -6.85544908e-01 5.80183938e-02
8.32348704e-01 -1.38512969e-01 -4.54784781e-01 7.05966592e-01
6.11538529e-01 -9.01585817e-01 7.89975166e-01 -3.71071756e-01
4.82937768e-02 -6.13647163e-01 1.09677771e-02 -1.07502639e+00
-1.91557541e-01 -2.66005754e-01 -4.49087650e-01 1.31292260e+00
4.08302695e-01 -8.22422504e-01 8.43384504e-01 7.18547940e-01
-5.68147153e-02 -7.05105007e-01 -9.85171378e-01 -1.12783563e+00
1.62870467e-01 -3.32449287e-01 5.92627704e-01 1.31650519e+00
-1.53238356e-01 2.59624481e-01 -6.13874346e-02 2.22178221e-01
3.81425202e-01 1.57282919e-01 5.07045746e-01 -1.37758720e+00
-3.81511331e-01 -3.39625806e-01 -3.63088280e-01 -9.85392749e-01
1.14325240e-01 -8.86294663e-01 -2.00341605e-02 -1.10952103e+00
4.99244958e-01 -8.26979041e-01 -5.46980858e-01 1.17649114e+00
-4.05055434e-01 7.99726769e-02 2.05095857e-01 1.76861182e-01
-5.44309556e-01 4.72414017e-01 1.10413384e+00 -3.90964150e-01
-4.70403939e-01 -7.22710788e-02 -8.59682560e-01 8.93064260e-01
8.44164908e-01 -6.60078466e-01 -5.43864608e-01 -3.60622257e-01
3.11287433e-01 -4.99053061e-01 5.79299808e-01 -9.72946227e-01
4.13572341e-01 -1.64373979e-01 5.01484931e-01 -4.33097333e-01
1.67796999e-01 -9.51261461e-01 1.13109492e-01 6.49836361e-01
-7.69876465e-02 -5.88867605e-01 3.75579029e-01 6.74110651e-01
-6.48829341e-02 -2.97356069e-01 1.07765162e+00 -8.54699388e-02
-9.92302299e-01 8.93692151e-02 -7.84206092e-02 -4.51455936e-02
1.10335314e+00 -4.12399292e-01 -1.40995726e-01 3.57266158e-01
-8.91591668e-01 2.67081499e-01 6.36217415e-01 5.35465181e-01
5.19807935e-01 -1.01269150e+00 -1.00288391e-01 4.94109899e-01
2.91050643e-01 3.71668369e-01 2.11240426e-01 5.79492927e-01
-1.13127507e-01 -1.22543126e-01 -8.41843188e-02 -6.00884438e-01
-1.11294639e+00 8.17357719e-01 1.75570399e-01 -2.74575680e-01
-5.28703094e-01 8.20218146e-01 3.66710693e-01 -3.39238197e-01
2.57400304e-01 -5.63368320e-01 -4.31409031e-02 3.07765126e-01
4.43684906e-01 2.66727209e-01 1.32672623e-01 -5.77333458e-02
-3.68431121e-01 5.55349112e-01 -5.39546013e-01 1.75885603e-01
1.20376086e+00 -2.04854622e-01 -1.50123641e-01 8.61303270e-01
1.07144856e+00 -1.61095187e-01 -1.51034474e+00 -4.11692530e-01
1.72637030e-01 -4.12865341e-01 -1.52122095e-01 -1.17251623e+00
-1.05513871e+00 7.36914754e-01 6.49784863e-01 -9.13604423e-02
1.46243715e+00 2.35507488e-02 8.34075749e-01 5.39969742e-01
3.51894796e-01 -1.24379849e+00 1.42854780e-01 6.34421289e-01
4.99132842e-01 -1.19831491e+00 -1.43219873e-01 -6.01786375e-01
-4.81609672e-01 1.22720993e+00 6.13031149e-01 3.18030536e-01
5.77182770e-01 2.19852656e-01 -2.21010208e-01 1.09121732e-01
-3.45230639e-01 -1.37175858e-01 5.44547737e-02 3.71073037e-01
-2.90229589e-01 3.53779010e-02 -1.11845881e-01 8.69477034e-01
6.67415783e-02 1.80232018e-01 3.35101545e-01 1.23381984e+00
-5.49029410e-01 -1.10578859e+00 -1.09857097e-01 2.41530493e-01
-1.39063438e-02 -1.62736669e-01 -4.13069785e-01 6.37503445e-01
6.96277559e-01 6.07838690e-01 3.04519888e-02 -3.07528764e-01
3.04273367e-01 3.29662055e-01 5.06537855e-01 -7.01752663e-01
-3.03596377e-01 -3.06952387e-01 -5.06324410e-01 -3.48102152e-01
-4.39645320e-01 -5.67920566e-01 -1.45304871e+00 -9.20546949e-02
-5.30420780e-01 9.54042673e-02 4.47466880e-01 1.01614463e+00
2.55019158e-01 5.87706149e-01 7.48613238e-01 -3.42820257e-01
-3.59355778e-01 -2.88597077e-01 -4.89645600e-01 3.29278767e-01
3.34510475e-01 -8.71456027e-01 -4.50408876e-01 3.16366047e-01] | [9.40876293182373, 2.2631661891937256] |
a0dc900a-0ecf-443b-bdce-1305c3b921cd | automatic-objects-removal-for-scene | 1501.0597 | null | http://arxiv.org/abs/1501.05970v1 | http://arxiv.org/pdf/1501.05970v1.pdf | Automatic Objects Removal for Scene Completion | With the explosive growth of web-based cameras and mobile devices, billions
of photographs are uploaded to the internet. We can trivially collect a huge
number of photo streams for various goals, such as 3D scene reconstruction and
other big data applications. However, this is not an easy task due to the fact
the retrieved photos are neither aligned nor calibrated. Furthermore, with the
occlusion of unexpected foreground objects like people, vehicles, it is even
more challenging to find feature correspondences and reconstruct realistic
scenes. In this paper, we propose a structure based image completion algorithm
for object removal that produces visually plausible content with consistent
structure and scene texture. We use an edge matching technique to infer the
potential structure of the unknown region. Driven by the estimated structure,
texture synthesis is performed automatically along the estimated curves. We
evaluate the proposed method on different types of images: from highly
structured indoor environment to the natural scenes. Our experimental results
demonstrate satisfactory performance that can be potentially used for
subsequent big data processing: 3D scene reconstruction and location
recognition. | ['Kun Hua', 'Jianjun Yang', 'Yin Wang', 'Ju Shen', 'Wei Wang', 'Honggang Wang'] | 2015-01-23 | null | null | null | null | ['3d-scene-reconstruction'] | ['computer-vision'] | [ 3.98887187e-01 -3.04046482e-01 3.40722680e-01 -2.54356295e-01
-4.97716844e-01 -5.15555918e-01 4.52989101e-01 -9.51891094e-02
-1.11606017e-01 5.48730314e-01 1.40062049e-01 -1.67285446e-02
4.30997349e-02 -7.95280397e-01 -6.81073785e-01 -6.06802106e-01
3.88312697e-01 5.09474039e-01 5.29010773e-01 9.05234143e-02
3.19090337e-01 7.73762703e-01 -1.86227679e+00 1.28844842e-01
7.45289683e-01 1.11507082e+00 6.75923944e-01 4.71992522e-01
-1.97040528e-01 5.65381646e-01 -2.16990858e-01 -3.74362975e-01
6.09312654e-01 -5.76974340e-02 -2.37394542e-01 7.60123014e-01
6.28375828e-01 -5.98346770e-01 -2.32101575e-01 1.20065820e+00
2.06522346e-01 3.78078520e-02 3.09145153e-01 -9.87128079e-01
-9.18795094e-02 -2.40848154e-01 -8.72345269e-01 -1.16941959e-01
6.92221940e-01 3.62027250e-02 3.37101012e-01 -1.08834052e+00
8.89059067e-01 1.17252636e+00 5.19728720e-01 -5.38204471e-03
-8.07926059e-01 -3.16741288e-01 4.77393344e-02 3.62131596e-01
-1.51561689e+00 -6.27998531e-01 9.32495832e-01 -2.11065769e-01
4.88115460e-01 3.15475404e-01 5.31381369e-01 7.81109333e-01
-4.88516241e-02 3.82575363e-01 1.05030394e+00 -4.33019012e-01
3.12821895e-01 3.44826162e-01 -3.34659696e-01 5.54211497e-01
4.00752306e-01 -7.58551583e-02 -3.82173091e-01 1.27218533e-02
7.72899270e-01 4.38482165e-01 -3.81493032e-01 -4.85692501e-01
-1.35784054e+00 1.55204711e-02 1.65944859e-01 -1.30824409e-02
-5.09411275e-01 -1.41033590e-01 -1.11800246e-01 2.75681484e-02
4.37832147e-01 -1.42762557e-01 -1.09633274e-01 8.17840844e-02
-6.30205929e-01 5.57952598e-02 5.72157025e-01 1.31155789e+00
9.07181025e-01 2.99393237e-02 4.50410277e-01 8.72763813e-01
1.87425137e-01 7.14154959e-01 1.16512410e-01 -1.20145190e+00
6.01659894e-01 7.12767422e-01 4.69182402e-01 -1.41864216e+00
-2.26507589e-01 -1.38879612e-01 -1.15461004e+00 9.18211490e-02
5.70348561e-01 1.81103393e-01 -5.79201043e-01 9.94657516e-01
9.03802693e-01 4.42463994e-01 -2.18905866e-01 1.10468292e+00
5.86217344e-01 6.95529044e-01 -4.18293387e-01 -1.95732534e-01
1.33343327e+00 -6.94770873e-01 -5.45573890e-01 -4.16004568e-01
3.17119509e-02 -1.11589098e+00 7.81179547e-01 4.54357952e-01
-9.49568808e-01 -5.19516826e-01 -7.33177066e-01 -9.32931751e-02
2.29771808e-02 1.19039334e-01 5.38331568e-01 4.02195364e-01
-7.97074735e-01 3.02409828e-01 -5.16148269e-01 -5.63269436e-01
3.06350827e-01 6.84709549e-02 -6.68637455e-01 -7.19937682e-01
-4.30425674e-01 5.13514578e-01 2.64644533e-01 5.04936278e-01
-6.31470621e-01 -3.71858090e-01 -7.37695396e-01 4.18698601e-02
4.95171219e-01 -8.26898336e-01 6.96787655e-01 -9.34024096e-01
-1.09754062e+00 8.87872934e-01 -2.74600208e-01 -3.96511368e-02
6.71705604e-01 -1.06831945e-01 -1.15752012e-01 1.45601779e-01
4.55597043e-03 2.52844244e-01 9.60838556e-01 -1.43123937e+00
-7.48297811e-01 -6.43148959e-01 -1.42191440e-01 4.90232140e-01
-2.86617607e-01 -5.88095710e-02 -7.84151673e-01 -2.27827355e-01
4.44669545e-01 -7.98670650e-01 -3.50325495e-01 4.21904773e-01
-4.10969794e-01 3.52034330e-01 9.01447654e-01 -5.68133175e-01
6.18093491e-01 -2.29155159e+00 -8.78429338e-02 1.67225271e-01
3.04929018e-02 -4.42019738e-02 -1.40850544e-01 2.99888104e-01
3.82939726e-01 -2.78043658e-01 -1.65679678e-01 -5.58922052e-01
-4.10225689e-01 4.87941392e-02 -3.14698219e-01 5.57283103e-01
-1.69879898e-01 4.25854772e-01 -7.44891346e-01 -5.80221951e-01
6.83372676e-01 5.57912588e-01 -4.81992811e-01 3.40769500e-01
-5.02356254e-02 6.57422543e-01 -5.38652599e-01 8.11416864e-01
1.07846212e+00 -2.02488661e-01 1.07036315e-01 -3.43267888e-01
-3.11448246e-01 -2.71944791e-01 -1.65666187e+00 1.65253675e+00
-4.18835074e-01 6.67389810e-01 3.14145595e-01 -7.43621945e-01
9.60484326e-01 -1.14016375e-02 6.15391910e-01 -5.66652179e-01
1.39654174e-01 1.88169777e-01 -5.66702902e-01 -5.91483295e-01
7.70162523e-01 1.25480235e-01 1.72340766e-01 1.97922453e-01
-4.67096090e-01 -5.03186822e-01 -3.35710905e-02 7.60802403e-02
8.86229813e-01 1.51142344e-01 1.84937045e-01 6.71519488e-02
5.86042106e-01 1.68781117e-01 7.40884066e-01 3.22463185e-01
1.26764789e-01 9.93640780e-01 -7.07466602e-02 -6.79992557e-01
-1.56495428e+00 -1.01554573e+00 -8.76799077e-02 1.82674423e-01
5.88444889e-01 -1.59450784e-01 -5.85542738e-01 -1.12743140e-03
-1.49336785e-01 2.15208337e-01 1.15642343e-02 3.38395387e-01
-4.79642421e-01 -5.40357411e-01 -1.58751398e-01 4.78538796e-02
9.32373047e-01 -7.64236093e-01 -5.17622530e-01 9.44177434e-02
-4.32813048e-01 -1.61345255e+00 -3.85685295e-01 -5.29886365e-01
-8.61559689e-01 -1.07433772e+00 -5.51563740e-01 -8.74296188e-01
1.09814560e+00 1.10183287e+00 8.15279186e-01 3.54924977e-01
-5.82463682e-01 4.17762786e-01 -1.91238850e-01 -1.04735911e-01
-1.30605474e-01 -5.03337741e-01 2.98975110e-02 6.05833948e-01
2.58218236e-02 -7.00341046e-01 -9.11806285e-01 6.04795456e-01
-9.61290002e-01 5.88134766e-01 4.92414057e-01 5.69412172e-01
7.30472744e-01 5.00502050e-01 7.34852031e-02 -6.11253738e-01
9.85857993e-02 -3.34621519e-01 -9.74808335e-01 1.93736076e-01
-1.12538368e-01 -2.91396856e-01 6.53879225e-01 -3.49530637e-01
-1.37530696e+00 4.95539188e-01 5.13680950e-02 -5.17723799e-01
-3.81948978e-01 -2.29395665e-02 -4.00146425e-01 -1.57723650e-01
2.17544317e-01 3.67551893e-01 -1.74813703e-01 -5.29107571e-01
1.34781793e-01 8.53797793e-01 5.72111547e-01 -2.69950420e-01
1.01816893e+00 9.26663578e-01 9.10920203e-02 -1.46818519e+00
-5.44582665e-01 -6.08860910e-01 -7.00560927e-01 -5.08265793e-01
7.61800528e-01 -9.58671153e-01 -6.79932058e-01 5.93061030e-01
-1.31014264e+00 -1.21411964e-01 -7.00140595e-02 4.61276740e-01
-3.29725087e-01 7.45193243e-01 -2.86865175e-01 -7.62028992e-01
-1.81920990e-01 -1.01558852e+00 1.35393107e+00 2.95548916e-01
3.09345305e-01 -6.35850847e-01 -1.82899728e-01 6.97256565e-01
2.33335406e-01 3.04732889e-01 7.10738003e-01 3.69734377e-01
-1.57119250e+00 -3.64262998e-01 -5.64295650e-01 1.16675355e-01
6.49630576e-02 8.15719664e-02 -1.04517543e+00 5.19835996e-03
1.15266845e-01 3.97362448e-02 3.26877534e-01 3.27119261e-01
1.14379954e+00 -1.89426363e-01 -2.97583699e-01 7.70821035e-01
1.59697902e+00 -8.04180726e-02 9.43452835e-01 2.53278553e-01
8.51948440e-01 8.17233622e-01 7.43169785e-01 6.25765145e-01
4.89484310e-01 8.12684596e-01 5.17505825e-01 1.10678263e-01
-2.18200341e-01 -3.25748384e-01 7.88213685e-02 7.26133764e-01
-2.77278543e-01 -2.88538069e-01 -7.34045565e-01 4.08931464e-01
-1.73008060e+00 -1.02766061e+00 -4.48790163e-01 2.60479760e+00
2.30347708e-01 -5.83170988e-02 -4.35397774e-01 9.54130478e-03
8.10127139e-01 -1.38614913e-02 -4.38941061e-01 2.04932675e-01
-3.10925037e-01 -4.18713272e-01 4.70167994e-01 3.97170812e-01
-5.81116140e-01 7.93532073e-01 5.07938719e+00 5.93898237e-01
-9.26029861e-01 -8.06304663e-02 6.41666651e-01 1.91340089e-01
-3.70728195e-01 1.58340991e-01 -8.13172281e-01 4.05104399e-01
3.41821283e-01 -2.22053621e-02 4.34062719e-01 7.05634058e-01
3.79583985e-01 -5.96651316e-01 -8.44557464e-01 1.42771780e+00
1.82404935e-01 -1.31946182e+00 9.31702927e-03 2.07107797e-01
1.01301432e+00 -7.29542226e-02 -3.12391728e-01 -4.32578921e-01
-1.08198710e-01 -5.49576342e-01 7.24633455e-01 5.86652100e-01
6.75438344e-01 -4.95100319e-01 5.04156232e-01 7.38723218e-01
-1.16793203e+00 -9.18357670e-02 -7.96490073e-01 -1.34784833e-01
5.79350173e-01 8.84296715e-01 -8.81481946e-01 4.26970392e-01
6.42340004e-01 6.93245828e-01 -5.94643474e-01 1.51386416e+00
-7.84423016e-03 -3.11455596e-02 -5.46809793e-01 9.68950614e-02
-2.50807852e-01 -6.09416008e-01 6.17319047e-01 5.24316907e-01
6.97102726e-01 3.21682036e-01 2.63640016e-01 5.63444674e-01
-9.57171470e-02 1.97956458e-01 -9.75688815e-01 4.61477935e-01
4.60487843e-01 1.50611949e+00 -9.90478635e-01 -1.82207689e-01
-5.86775839e-01 1.24608195e+00 1.62036598e-01 2.87934035e-01
-5.42916834e-01 3.56341302e-02 4.95058358e-01 4.26503807e-01
7.17619359e-02 -4.91781771e-01 -4.20524999e-02 -1.56033051e+00
2.95238554e-01 -6.88703477e-01 -6.37813881e-02 -1.30072331e+00
-1.02925301e+00 4.76762056e-01 -1.24733992e-01 -1.61286700e+00
-6.96029216e-02 -3.32201898e-01 -4.21832442e-01 3.78857315e-01
-1.48782241e+00 -9.68157709e-01 -8.34551990e-01 8.15784931e-01
8.46844494e-01 2.43798405e-01 6.03668332e-01 5.74145615e-01
-3.97800386e-01 -1.59545049e-01 4.89352226e-01 -1.74830124e-01
6.05298579e-01 -6.66355133e-01 1.86686561e-01 9.63909626e-01
1.34607717e-01 2.19334841e-01 5.90400815e-01 -7.12385535e-01
-1.81564879e+00 -1.06699085e+00 7.67384648e-01 -3.58653486e-01
2.89724439e-01 -5.99888563e-01 -6.15536869e-01 4.02609169e-01
-1.07623041e-01 2.00438499e-01 1.17062561e-01 -4.54129636e-01
7.28846854e-03 -5.09786308e-01 -1.12640333e+00 6.22042894e-01
1.18830407e+00 -1.69403836e-01 -8.28664675e-02 4.91236061e-01
3.11010301e-01 -4.66192335e-01 -4.79924321e-01 2.18113780e-01
5.96748412e-01 -1.24698973e+00 1.20019519e+00 3.32344137e-02
2.22064987e-01 -4.76506859e-01 -3.18458080e-01 -9.95164812e-01
6.55920133e-02 -4.09849972e-01 3.40903908e-01 1.22440267e+00
-6.55965880e-02 -4.38154250e-01 9.50049222e-01 1.00354290e+00
-5.05491942e-02 -1.07429869e-01 -8.26411903e-01 -5.10602176e-01
-9.71764505e-01 -3.57268304e-01 5.10927498e-01 7.19480813e-01
-5.11964560e-01 2.24416867e-01 -6.82286978e-01 4.94165093e-01
1.19575489e+00 5.83828270e-01 1.29378664e+00 -1.23016870e+00
-1.28750056e-01 8.16848949e-02 -7.35860467e-01 -1.21633184e+00
-1.44268975e-01 -3.21589291e-01 9.35366750e-02 -1.57618129e+00
2.54042536e-01 -7.88664401e-01 4.15546477e-01 -1.71427310e-01
1.88749656e-01 6.19195759e-01 1.25108853e-01 3.00811470e-01
-5.71989357e-01 4.59119111e-01 1.21045280e+00 4.98595228e-03
-1.52111918e-01 1.47907352e-02 -2.05863193e-01 9.67292607e-01
5.13861299e-01 -2.40045637e-01 -4.04966712e-01 -7.42082059e-01
3.56903017e-01 3.34713221e-01 5.05332947e-01 -1.08732188e+00
2.77484536e-01 -3.53562176e-01 5.82832992e-01 -7.68538296e-01
6.16040170e-01 -1.28119504e+00 6.56210482e-01 1.93003528e-02
2.32199207e-01 -1.40358970e-01 -6.13539517e-02 7.19894528e-01
-1.37504295e-01 -2.18340367e-01 5.89262724e-01 -2.16615975e-01
-9.38200116e-01 3.98851544e-01 -1.41001478e-01 -2.31884778e-01
1.23735273e+00 -5.15443861e-01 -2.39164755e-01 -5.56042433e-01
-1.53935149e-01 6.75438717e-02 8.98805857e-01 4.64281082e-01
1.18083894e+00 -1.29532611e+00 -6.03784442e-01 6.01691127e-01
8.35746974e-02 4.74371493e-01 6.39974713e-01 6.91062152e-01
-9.15949583e-01 1.54643565e-01 -2.02301949e-01 -8.38208258e-01
-1.52486730e+00 5.23296773e-01 -3.15522440e-02 2.59464532e-01
-9.54766631e-01 2.35416472e-01 4.61070597e-01 -2.13363901e-01
1.73021510e-01 -2.25572571e-01 2.53036708e-01 -3.13309938e-01
8.64445090e-01 3.77363682e-01 6.22166358e-02 -7.52182841e-01
-1.20221004e-01 9.35840964e-01 1.78884059e-01 1.12044573e-01
1.62676322e+00 -7.16463029e-01 -1.18506536e-01 7.38704577e-02
9.92855787e-01 2.26613551e-01 -1.38334882e+00 -1.74190238e-01
-2.30057597e-01 -1.23451769e+00 3.23001929e-02 -1.33520484e-01
-1.01622105e+00 8.95833790e-01 4.40485179e-01 -7.55912289e-02
1.13176990e+00 -7.80158490e-02 8.19435179e-01 4.10345256e-01
9.75075841e-01 -9.58200932e-01 6.88261120e-03 1.85653597e-01
7.00417876e-01 -1.30142128e+00 2.07107142e-01 -6.15107596e-01
-4.23165470e-01 1.05765629e+00 4.05296832e-01 -3.38531956e-02
4.76860970e-01 1.11878388e-01 -1.05394356e-01 -2.11068261e-02
-4.68090564e-01 -6.11051433e-02 9.39356908e-03 6.34227395e-01
-2.24107012e-01 -3.44043165e-01 1.68247610e-01 -2.17924550e-01
1.11955173e-01 -1.70263082e-01 8.84579003e-01 5.42648554e-01
-6.56336784e-01 -9.25012112e-01 -8.95975769e-01 2.63704985e-01
-1.10614352e-01 9.62792337e-02 -9.93296281e-02 4.08622563e-01
-1.09955572e-01 9.41772342e-01 -7.32052140e-03 1.04024401e-02
2.60709167e-01 -2.60520816e-01 5.31410456e-01 -5.04782498e-01
1.25869319e-01 2.85947591e-01 1.18605174e-01 -6.20944798e-01
-5.45763612e-01 -6.96814060e-01 -9.80225623e-01 -3.72444332e-01
-3.77316386e-01 -2.15318397e-01 1.06196129e+00 5.50299168e-01
4.39608067e-01 -2.08822042e-01 8.61040175e-01 -1.08349729e+00
-2.39626449e-02 -5.34536302e-01 -6.32672310e-01 6.82246089e-01
1.88572869e-01 -5.42370081e-01 -3.30887675e-01 3.29042971e-01] | [9.042449951171875, -2.531083345413208] |
d2344266-aac7-4de6-9c1c-d89c76194e35 | revisit-the-fundamental-theorem-of-linear | 2108.04432 | null | https://arxiv.org/abs/2108.04432v1 | https://arxiv.org/pdf/2108.04432v1.pdf | Revisit the Fundamental Theorem of Linear Algebra | This survey is meant to provide an introduction to the fundamental theorem of linear algebra and the theories behind them. Our goal is to give a rigorous introduction to the readers with prior exposure to linear algebra. Specifically, we provide some details and proofs of some results from (Strang, 1993). We then describe the fundamental theorem of linear algebra from different views and find the properties and relationships behind the views. The fundamental theorem of linear algebra is essential in many fields, such as electrical engineering, computer science, machine learning, and deep learning. This survey is primarily a summary of purpose, significance of important theories behind it. The sole aim of this survey is to give a self-contained introduction to concepts and mathematical tools in theory behind the fundamental theorem of linear algebra and rigorous analysis in order to seamlessly introduce its properties in four subspaces in subsequent sections. However, we clearly realize our inability to cover all the useful and interesting results and given the paucity of scope to present this discussion, e.g., the separated analysis of the (orthogonal) projection matrices. We refer the reader to literature in the field of linear algebra for a more detailed introduction to the related fields. Some excellent examples include (Rose, 1982; Strang, 2009; Trefethen and Bau III, 1997; Strang, 2019, 2021). | ['Jun Lu'] | 2021-08-10 | null | null | null | null | ['electrical-engineering'] | ['miscellaneous'] | [ 1.09373681e-01 1.22926710e-02 -2.63779789e-01 -4.01361063e-02
-1.59388304e-01 -7.48914599e-01 3.69118422e-01 -3.22472364e-01
8.62584561e-02 5.19715846e-01 7.30862692e-02 -7.16075480e-01
-8.57508957e-01 -5.45665383e-01 -5.10172665e-01 -9.05739248e-01
-7.51244545e-01 -1.30158424e-01 -5.29428244e-01 -3.71661752e-01
3.68156314e-01 9.41663146e-01 -1.21780503e+00 2.48530153e-02
6.13647521e-01 9.60897624e-01 -7.25739747e-02 5.91934919e-01
6.70212209e-02 7.24933326e-01 2.97228917e-02 -6.08709395e-01
6.71513915e-01 -4.35372531e-01 -9.15179133e-01 3.39018703e-01
5.39321899e-01 -2.53426939e-01 -8.06556106e-01 1.25869691e+00
1.84796393e-01 1.94796622e-01 7.09507108e-01 -1.67441320e+00
-6.74778104e-01 5.48519313e-01 -5.08399546e-01 1.73221767e-01
4.89090234e-02 -2.94462353e-01 1.27052820e+00 -1.30888593e+00
1.51220053e-01 1.02879012e+00 8.01959097e-01 2.38121763e-01
-1.10854876e+00 -1.97628021e-01 1.13017363e-02 6.94780350e-01
-1.43767202e+00 -5.05001247e-01 8.73614192e-01 -6.86819494e-01
5.51517308e-01 6.89139307e-01 6.91557229e-01 5.20017385e-01
3.70572090e-01 9.33456421e-01 1.16296124e+00 -8.52064431e-01
-9.16978791e-02 2.63093829e-01 7.17049181e-01 7.13433027e-01
3.48544121e-01 2.46505797e-01 -2.43378907e-01 -1.91706195e-02
8.95942807e-01 -2.64447276e-02 -2.50586838e-01 -1.17240024e+00
-1.23532653e+00 1.15981030e+00 4.20708984e-01 5.23270249e-01
-1.94290459e-01 -1.96460932e-01 2.77459413e-01 4.80220079e-01
-8.31725821e-02 3.28967959e-01 -1.18618213e-01 2.15274081e-01
-7.48412549e-01 3.89916301e-01 1.02197039e+00 1.18397379e+00
5.90075910e-01 3.19123507e-01 3.74923944e-01 6.26310110e-01
5.30171134e-02 5.04367828e-01 8.80079642e-02 -1.12737513e+00
1.89106837e-01 -6.67310581e-02 -4.08993036e-01 -1.38896501e+00
-4.64397579e-01 -6.81460559e-01 -1.26386333e+00 2.25069970e-01
2.55290926e-01 1.46365166e-03 -2.84068525e-01 1.56201291e+00
-2.71788418e-01 -2.98408240e-01 1.27898663e-01 7.98705578e-01
8.47040832e-01 6.29672170e-01 -5.70919394e-01 -6.80521309e-01
1.09289384e+00 -8.51286471e-01 -7.81484187e-01 -1.11554533e-01
5.26334465e-01 -9.15670812e-01 5.39598346e-01 2.17469081e-01
-1.26891005e+00 -2.96061516e-01 -1.08108795e+00 -9.25952345e-02
-2.35616013e-01 2.73959398e-01 1.16757834e+00 5.97501814e-01
-1.07695580e+00 5.10909379e-01 -9.28035855e-01 -9.19570565e-01
1.27205163e-01 4.08953995e-01 -5.47769308e-01 -1.46394581e-01
-1.03327620e+00 1.19663990e+00 3.45979154e-01 2.77306557e-01
-2.18113959e-01 -7.49892056e-01 -9.62405622e-01 -8.09435397e-02
3.14057171e-01 -6.21268094e-01 1.14711034e+00 -7.08256483e-01
-1.24927342e+00 1.04464972e+00 -2.75694549e-01 -5.30022740e-01
3.08258444e-01 -2.30861843e-01 -2.30683461e-01 1.46465991e-02
-6.31157309e-02 1.22904234e-01 3.31120253e-01 -1.16136670e+00
-3.45644683e-01 -5.72541535e-01 3.37716848e-01 1.72175765e-01
-4.04339105e-01 1.74136460e-01 -1.74876064e-01 -5.13284326e-01
5.38539171e-01 -1.07544804e+00 -2.54034072e-01 -1.82333484e-01
-3.94645005e-01 -2.03851238e-01 5.14004707e-01 -3.66665602e-01
1.41321015e+00 -2.06206131e+00 5.56405127e-01 3.60590577e-01
3.87313664e-01 1.52867455e-02 2.22253092e-02 1.15053964e+00
-9.45707619e-01 -2.25277007e-01 -3.13785613e-01 2.79019743e-01
2.76071578e-02 8.59019831e-02 -7.48629868e-01 1.09425366e+00
-3.21768969e-01 1.00250912e+00 -6.67810261e-01 -1.37210786e-01
5.50644815e-01 3.96373957e-01 -2.56804645e-01 -3.18195999e-01
6.81289434e-01 6.31706268e-02 -1.43695801e-01 4.49809045e-01
8.85644197e-01 -1.00404680e-01 1.64377347e-01 -6.67522967e-01
-4.66666788e-01 2.81408250e-01 -1.49572492e+00 1.22803092e+00
-2.38704309e-01 1.07120728e+00 4.49823141e-01 -1.78070581e+00
4.42071497e-01 3.83851051e-01 7.22848058e-01 -3.20439249e-01
9.64615718e-02 2.69370228e-01 3.21172208e-01 -3.89598370e-01
3.51870567e-01 -2.68305063e-01 -5.54168858e-02 4.86256301e-01
-1.61393970e-01 -2.92143002e-02 5.41465580e-01 5.41821957e-01
6.33957028e-01 -4.64597344e-01 9.73999679e-01 -7.27995992e-01
9.31792200e-01 -6.23419881e-02 1.54949933e-01 7.36518323e-01
-2.92513877e-01 2.07421541e-01 4.42520529e-01 -6.65819228e-01
-1.15521789e+00 -1.21615279e+00 -5.83399415e-01 1.02121472e+00
-1.42236248e-01 -6.41865313e-01 -4.06915665e-01 -1.83794662e-01
-3.49358618e-02 5.00790954e-01 -7.11021006e-01 5.44249937e-02
-5.49478352e-01 -8.76784801e-01 1.47503912e-01 6.35475218e-01
5.64946949e-01 -5.78910232e-01 -4.20731260e-03 -4.03331906e-01
-4.84434739e-02 -8.55254233e-01 3.43966000e-02 1.78638205e-01
-1.16460502e+00 -1.13115776e+00 -7.70796478e-01 -9.71145749e-01
6.99900210e-01 6.82728827e-01 8.44391882e-01 -3.86897214e-02
-2.00386360e-01 7.42332518e-01 1.47107452e-01 -1.62277132e-01
-2.76026249e-01 -1.29421040e-01 5.36421835e-01 -3.14878047e-01
5.34414649e-01 -6.87674165e-01 -8.86494815e-02 2.38596931e-01
-7.40273118e-01 2.39509009e-02 7.29754925e-01 7.65275121e-01
6.71056956e-02 3.12648475e-01 -2.83882115e-02 -6.42540216e-01
5.54579437e-01 -2.85886645e-01 -7.17587948e-01 4.56483476e-02
-4.93938535e-01 -1.66977942e-01 5.83715737e-01 5.80011122e-02
-4.60583508e-01 -1.54270362e-02 4.25889976e-02 -3.64442877e-02
1.66492969e-01 8.36237550e-01 -1.48385376e-01 -3.60935777e-01
7.25449085e-01 5.71068108e-01 1.57844022e-01 -4.03679520e-01
6.51455939e-01 5.91035604e-01 5.26634753e-01 -5.22267759e-01
1.18358827e+00 4.99412477e-01 6.96375191e-01 -1.18598473e+00
-1.06485081e+00 -6.51042938e-01 -1.22222304e+00 -2.32767373e-01
2.27900237e-01 -6.13912463e-01 -7.53026962e-01 3.37242365e-01
-9.03843999e-01 1.22233652e-01 -8.96558445e-03 7.73968101e-01
-1.03003311e+00 5.71671903e-01 -8.19172084e-01 -9.20844972e-01
-1.18257385e-03 -1.01681054e+00 4.38523352e-01 2.05452833e-02
-7.51045421e-02 -1.41242182e+00 1.91624705e-02 2.50000805e-01
1.97935864e-01 2.26521362e-02 9.23209846e-01 -4.99054521e-01
-4.95116949e-01 -3.17860514e-01 -1.79364547e-01 5.50037324e-01
-1.39894057e-02 -1.04002520e-01 -7.94082284e-01 -4.44865972e-01
2.67447472e-01 -6.58673719e-02 7.44601309e-01 4.76991445e-01
1.02635515e+00 -5.75725794e-01 -3.30239326e-01 6.86130643e-01
1.54723299e+00 1.01985179e-01 6.82948709e-01 4.77762967e-01
6.45974100e-01 9.38288927e-01 2.48075753e-01 5.82380816e-02
4.58081625e-02 5.29561579e-01 2.92446613e-01 -7.06124231e-02
1.95867941e-01 2.98489183e-01 4.03886080e-01 1.11868680e+00
-5.40439427e-01 2.66528606e-01 -5.78885019e-01 3.95603210e-01
-1.83669841e+00 -1.38378954e+00 -5.68531513e-01 2.45805430e+00
4.18879181e-01 -3.17418367e-01 3.57439578e-01 5.49493194e-01
8.96353602e-01 3.02363299e-02 -5.11680655e-02 -5.11525452e-01
-2.55500466e-01 -1.11456014e-01 8.77610624e-01 8.51006508e-01
-1.32344174e+00 6.82315052e-01 7.49595547e+00 5.71879327e-01
-8.33024085e-01 -4.62271690e-01 8.67294893e-02 -1.28431935e-02
-3.84897217e-02 1.01424158e-01 -6.34964287e-01 -2.35299561e-02
5.87823749e-01 -8.25827479e-01 7.28074729e-01 8.42809200e-01
2.21643195e-01 3.00535351e-01 -1.41122687e+00 1.22373879e+00
3.83396000e-01 -1.23224711e+00 -1.55824408e-01 5.85544646e-01
8.05705011e-01 -1.77379161e-01 4.21736538e-01 2.39104539e-01
5.89567311e-02 -1.11763513e+00 3.19117159e-01 1.25942469e-01
3.25414658e-01 -7.49118745e-01 6.03777826e-01 4.04845297e-01
-1.04303789e+00 -2.45861083e-01 -6.38686657e-01 -6.64569557e-01
-4.36864514e-03 8.29517305e-01 -2.72826165e-01 9.25462604e-01
4.50298011e-01 9.06023741e-01 -1.47707343e-01 1.09959519e+00
3.02476846e-02 2.92169809e-01 -3.08612734e-01 1.02860399e-01
3.12609285e-01 -5.92533112e-01 5.93690276e-01 1.07206976e+00
2.34665960e-01 3.54417592e-01 9.52387080e-02 8.59618962e-01
3.06658745e-01 2.50868529e-01 -9.27065313e-01 -2.58439958e-01
1.49662241e-01 1.41725767e+00 -4.90100831e-01 -1.55518204e-01
-1.03218675e+00 5.97795725e-01 1.85996350e-02 4.28933829e-01
-3.97425681e-01 -6.40328705e-01 7.40393400e-01 1.39124513e-01
1.14401728e-01 -6.60860896e-01 -7.35311747e-01 -1.09467363e+00
-1.10435702e-01 -9.36176419e-01 4.05267358e-01 -3.61447841e-01
-1.38870275e+00 1.20031610e-01 5.22458315e-01 -1.12939835e+00
-3.90492052e-01 -1.27413177e+00 -4.37001318e-01 1.09311640e+00
-8.63818765e-01 -9.62323010e-01 -1.77827664e-02 4.02815968e-01
2.31600821e-01 -2.97179759e-01 7.08989024e-01 2.19391555e-01
-5.98085761e-01 3.12469751e-01 6.04003787e-01 5.07122695e-01
3.92768979e-01 -1.23542035e+00 2.40477473e-01 9.87010598e-01
1.44588903e-01 1.12016833e+00 8.65959764e-01 -6.71239346e-02
-1.89102781e+00 -6.74082994e-01 1.01282644e+00 -4.94359106e-01
9.80463564e-01 -2.13910669e-01 -5.86210251e-01 1.34848225e+00
2.85051644e-01 -3.12986970e-01 8.76206636e-01 2.82817870e-01
-1.00799233e-01 -1.79565161e-01 -5.86466074e-01 8.11497211e-01
8.84258151e-01 -3.79579604e-01 -7.06759751e-01 6.31388485e-01
-4.14891616e-02 -3.10582638e-01 -7.62080312e-01 4.55334991e-01
9.05720294e-01 -1.29853201e+00 1.33733094e+00 -7.19094932e-01
1.71205252e-01 -9.54013541e-02 -4.63881642e-01 -9.50740695e-01
-1.04363143e+00 -6.16853416e-01 -1.39136275e-03 5.75506926e-01
1.79221462e-02 -6.78327978e-01 6.26465201e-01 4.49585229e-01
-3.34702432e-01 -8.51288736e-01 -7.59696245e-01 -1.10275853e+00
6.11601830e-01 -5.43511391e-01 -7.53331110e-02 1.08424032e+00
6.93275154e-01 4.00174469e-01 -3.78728062e-01 1.93364143e-01
8.08136642e-01 3.56982231e-01 8.96621943e-01 -1.16552198e+00
3.85504253e-02 -7.73119926e-01 -9.73981142e-01 -1.38066864e+00
1.18479520e-01 -1.26003218e+00 -3.48971605e-01 -1.68143249e+00
4.68119651e-01 -8.93556401e-02 -2.26493731e-01 9.22625661e-02
3.50542694e-01 4.46335763e-01 4.55274761e-01 3.16544324e-01
-2.04921037e-01 1.96394041e-01 1.08901453e+00 3.14778313e-02
-1.32462559e-02 4.49208915e-01 -1.09255552e+00 9.69099164e-01
6.11601651e-01 1.62659228e-01 -4.94637042e-01 -2.83828735e-01
3.14146876e-01 -1.41832486e-01 4.23751563e-01 -9.58730876e-01
2.45804712e-01 -2.95980930e-01 3.43885690e-01 -7.74995625e-01
2.46628970e-01 -9.46822643e-01 -1.63446113e-01 6.65687442e-01
-2.92826831e-01 2.94745713e-01 6.72274306e-02 1.46562591e-01
-1.57103509e-01 -5.35983503e-01 8.45691144e-01 -1.71398699e-01
-8.85519505e-01 2.99203098e-01 -4.27662194e-01 -2.68449694e-01
1.24833226e+00 -9.52972323e-02 -8.20444301e-02 -8.00244927e-01
-8.02907407e-01 3.05583291e-02 3.04867476e-01 7.45922849e-02
2.54225820e-01 -1.68394637e+00 -6.13877952e-01 2.21541151e-01
-7.58706331e-02 -4.23235118e-01 3.46687108e-01 1.31988192e+00
-6.16064906e-01 1.28529310e+00 -2.32259676e-01 -5.09758770e-01
-1.36934066e+00 9.05662656e-01 3.30905825e-01 3.52594145e-02
-4.65222001e-01 6.11182868e-01 5.98542154e-01 -9.36568975e-02
3.30868661e-01 -5.72970361e-02 -1.93660948e-02 -1.81685746e-01
7.36048043e-01 8.64155948e-01 9.18409303e-02 -8.59472454e-01
-6.05293274e-01 6.44065320e-01 9.27554071e-02 -1.09813787e-01
1.22139573e+00 -3.26832235e-01 -7.15556145e-01 7.42661536e-01
1.56543541e+00 1.15976326e-01 -3.90194774e-01 -3.04725677e-01
-3.42005074e-01 -3.24493170e-01 -1.61588099e-03 -2.36140311e-01
-1.00634217e+00 1.26713479e+00 2.69022971e-01 6.62409782e-01
1.20496583e+00 5.85211366e-02 3.54727864e-01 6.68343723e-01
1.11093320e-01 -8.30363393e-01 -3.62640053e-01 8.33319008e-01
1.16221905e+00 -9.32986438e-01 4.11279500e-01 -8.80844235e-01
-2.16258034e-01 1.62385476e+00 2.08150670e-01 -1.78691342e-01
9.42257464e-01 1.80530921e-01 -1.14432305e-01 2.69643101e-03
-4.36738968e-01 -2.77755588e-01 3.83719802e-01 8.05930138e-01
8.20804596e-01 -1.39085814e-01 -5.54647326e-01 1.21220246e-01
-7.56299973e-01 -1.64450839e-01 7.03425944e-01 7.36862183e-01
-5.07221580e-01 -1.07184446e+00 -7.73324132e-01 3.13086361e-01
-2.54112929e-01 -3.19598347e-01 -3.04912269e-01 1.01178014e+00
-2.79875904e-01 6.26643181e-01 1.62976950e-01 -2.05620989e-01
7.54608307e-03 3.27564143e-02 9.92554843e-01 -4.14330840e-01
2.24140301e-01 1.36662321e-02 -1.49274647e-01 -2.16892183e-01
-5.00812769e-01 -9.14538920e-01 -7.09608495e-01 -1.06648314e+00
-8.63550231e-02 3.04880619e-01 6.72130644e-01 9.85175371e-01
-1.01442277e-01 2.54075825e-01 6.83156431e-01 -5.98155320e-01
-8.07597458e-01 -8.78968418e-01 -1.05570924e+00 -1.21530153e-01
4.99770999e-01 -5.53256631e-01 -5.48273623e-01 1.97828069e-01] | [7.4805707931518555, 4.224154949188232] |
722de6bb-a3e9-489e-b7d4-93cb1068e1cc | nuig-at-tiad-combining-unsupervised-nlp-and | null | null | https://aclanthology.org/2020.globalex-1.15 | https://aclanthology.org/2020.globalex-1.15.pdf | NUIG at TIAD: Combining Unsupervised NLP and Graph Metrics for Translation Inference | In this paper, we present the NUIG system at the TIAD shard task. This system includes graph-based metrics calculated using novel algorithms, with an unsupervised document embedding tool called ONETA and an unsupervised multi-way neural machine translation method. The results are an improvement over our previous system and produce the highest precision among all systems in the task as well as very competitive F-Measure results. Incorporating features from other systems should be easy in the framework we describe in this paper, suggesting this could very easily be extended to an even stronger result. | ['John Philip McCrae', 'Mihael Arcan'] | 2020-05-01 | null | null | null | lrec-2020-5 | ['document-embedding'] | ['methodology'] | [ 1.31731898e-01 9.56446007e-02 -2.87793010e-01 -2.99949646e-01
-6.19496226e-01 -3.96011025e-01 1.17832196e+00 2.14160994e-01
-7.46228516e-01 8.34763348e-01 4.85875815e-01 -4.09435123e-01
-3.03533614e-01 -8.21356118e-01 -8.81294087e-02 -4.85530347e-01
-3.12049925e-01 1.01576734e+00 5.57277322e-01 -6.56127512e-01
3.74079168e-01 3.12142849e-01 -1.21531868e+00 2.99337715e-01
3.62720281e-01 7.47178495e-01 -1.31815046e-01 9.60850716e-01
-1.93698451e-01 6.97341800e-01 -6.63916111e-01 -6.43436074e-01
2.40828171e-01 -3.67650568e-01 -1.06082499e+00 -3.29885930e-01
6.12090290e-01 1.15139512e-02 -4.85116690e-01 6.21554971e-01
6.59388721e-01 9.10791159e-02 8.76889229e-01 -1.13544309e+00
-8.17824841e-01 8.10931206e-01 -2.60008246e-01 5.63185513e-01
4.96658623e-01 -3.50773633e-01 1.22645736e+00 -8.60180676e-01
7.81180799e-01 1.34956753e+00 7.29368329e-01 2.57162303e-01
-1.30517030e+00 -1.54582530e-01 -5.78645527e-01 1.37518629e-01
-1.24021590e+00 -3.45325619e-01 5.53697467e-01 -2.46201307e-01
1.67828345e+00 3.88633102e-01 3.21853638e-01 9.20061946e-01
5.83924890e-01 5.89481056e-01 1.40897584e+00 -8.21074784e-01
-5.73234633e-02 2.51141638e-01 3.58987391e-01 8.90807688e-01
4.88047868e-01 8.46499950e-02 -5.90967596e-01 -2.22358868e-01
2.65902579e-01 -1.83589593e-01 1.37174070e-01 -3.88856351e-01
-1.72750258e+00 9.94190753e-01 4.92002070e-01 1.04301047e+00
-3.80368158e-02 2.29027912e-01 6.86278045e-01 9.39853132e-01
9.79114234e-01 8.08307767e-01 -4.25927430e-01 -3.16598356e-01
-1.05149043e+00 2.47627839e-01 1.22417831e+00 6.15197778e-01
5.22057533e-01 1.82877898e-01 -4.51922156e-02 6.29526019e-01
7.98646435e-02 1.95192814e-01 7.44940102e-01 -7.33056068e-01
5.98914802e-01 3.52728426e-01 -2.77441949e-01 -1.16664994e+00
-5.56753397e-01 -3.31930012e-01 -5.16179144e-01 1.60806656e-01
1.30302653e-01 -5.95507361e-02 -7.59320617e-01 1.13368368e+00
-2.69789007e-02 -2.76139468e-01 2.04364747e-01 5.87697327e-01
7.80005574e-01 5.24875045e-01 -2.07878605e-01 -9.73820314e-02
1.13923538e+00 -1.19687307e+00 -1.05054927e+00 1.36313424e-01
1.08171010e+00 -1.10003436e+00 7.22808897e-01 4.00191873e-01
-8.74411106e-01 -4.53456938e-01 -1.23262012e+00 -5.79735860e-02
-1.09415436e+00 -2.25848749e-01 9.74866569e-01 7.74594247e-01
-1.85647190e+00 9.23119068e-01 -5.24200261e-01 -9.27204967e-01
-1.23407461e-01 5.05641460e-01 -5.50494969e-01 1.55969605e-01
-1.34799027e+00 1.57895005e+00 6.15764260e-01 -1.92670032e-01
-2.80546010e-01 1.63880102e-02 -1.00526845e+00 -3.46292078e-01
1.65943936e-01 -7.13080704e-01 1.19245172e+00 -4.29665238e-01
-1.36448157e+00 8.47428441e-01 4.74105403e-02 -8.49926531e-01
2.91721821e-01 -1.06769726e-01 -5.01486123e-01 2.82784421e-02
1.52009130e-01 4.96096075e-01 4.04473484e-01 -7.94562638e-01
-1.16005957e-01 -2.92936474e-01 -3.15720364e-02 2.40744010e-01
-6.00897312e-01 2.16175660e-01 -1.69283390e-01 -6.55278742e-01
-1.64575979e-01 -8.83506238e-01 -4.25410606e-02 -2.05119580e-01
-2.94284761e-01 -5.16885281e-01 8.56353223e-01 -7.11551785e-01
1.35520148e+00 -1.85464633e+00 4.23934251e-01 1.37748033e-01
4.54924375e-01 1.82570294e-01 -1.77124366e-01 1.03973961e+00
-7.34611601e-02 2.41237402e-01 -3.15625131e-01 -4.90936428e-01
2.99599707e-01 5.14885783e-01 1.23316664e-02 3.60393196e-01
1.23397097e-01 9.28744197e-01 -8.98380816e-01 -5.61169922e-01
1.05201945e-01 2.45181113e-01 4.17625671e-03 -2.58354750e-02
-3.57545540e-02 -4.81350124e-01 -2.59275347e-01 3.76637220e-01
1.45617053e-01 -3.46966945e-02 1.89624146e-01 -3.79671417e-02
-2.05448627e-01 4.34945434e-01 -8.66369903e-01 1.83708692e+00
-3.90249342e-01 8.36148679e-01 -4.08594340e-01 -1.06878459e+00
1.11404383e+00 2.36556098e-01 4.05447811e-01 -6.27851307e-01
1.68656811e-01 2.63058662e-01 2.27543727e-01 -3.10568035e-01
9.86044288e-01 -5.06411977e-02 -1.43518791e-01 7.29651451e-01
3.70036602e-01 -2.57790834e-01 3.92579824e-01 5.94581723e-01
1.32376230e+00 -4.57522385e-02 3.51298541e-01 -6.98820889e-01
1.24697812e-01 4.97702397e-02 -2.98623741e-01 6.09564126e-01
-2.82694511e-02 4.71244842e-01 4.41741258e-01 -4.56395894e-01
-1.30242980e+00 -9.13519561e-01 -1.09982118e-01 8.73343587e-01
-3.57823789e-01 -1.01468110e+00 -4.73249912e-01 -8.73173833e-01
1.46112189e-01 6.65328443e-01 -9.51736987e-01 2.57623959e-02
-2.05292389e-01 -6.73734248e-01 8.44244301e-01 4.82385695e-01
3.44667286e-01 -8.81910741e-01 -4.62470427e-02 3.61023754e-01
1.71273023e-01 -9.19472694e-01 -5.33639729e-01 5.22326112e-01
-9.13807333e-01 -5.53535104e-01 -7.01921463e-01 -9.41521645e-01
1.87413260e-01 2.16398209e-01 1.14111876e+00 1.10345006e-01
-2.31593207e-01 2.94216573e-01 -6.40160143e-01 -4.04491872e-01
-4.74242508e-01 4.21972841e-01 2.82143742e-01 -4.00720060e-01
6.35167778e-01 -3.01124156e-01 -1.58854872e-01 1.64808512e-01
-1.07206762e+00 -2.57692337e-01 4.65033531e-01 8.42680275e-01
2.48797640e-01 -1.08639166e-01 5.31055033e-01 -1.08265305e+00
1.17629755e+00 -2.89364487e-01 -2.61511672e-02 2.15107482e-02
-1.26078093e+00 3.94764006e-01 5.01597822e-01 1.14806723e-02
-5.10377109e-01 -3.26668948e-01 -2.99476087e-01 -1.43359050e-01
1.29788071e-01 6.48580253e-01 1.21839598e-01 -3.11376780e-01
8.65957558e-01 2.00820312e-01 1.05134189e-01 -5.84634900e-01
5.82551837e-01 9.04097140e-01 4.77755293e-02 -2.74820000e-01
8.42483222e-01 7.66278580e-02 1.49966693e-02 -8.68496478e-01
-5.25047183e-01 -6.82841301e-01 -7.22244859e-01 5.47704510e-02
8.41387451e-01 -6.26412451e-01 4.92573977e-02 2.00081274e-01
-1.19886792e+00 -6.57320470e-02 -3.77712309e-01 5.82213759e-01
-4.94586408e-01 4.72078264e-01 -7.80692756e-01 -4.22402412e-01
-5.13876140e-01 -8.48430455e-01 1.14123726e+00 -3.00567210e-01
-4.22078371e-01 -1.57911134e+00 8.09687614e-01 6.29784018e-02
7.33007550e-01 2.12832794e-01 6.29573166e-01 -1.18099427e+00
8.86372253e-02 -3.13115448e-01 -3.23684484e-01 5.88777006e-01
2.96848029e-01 2.77664252e-02 -6.59839690e-01 -4.45694596e-01
-2.97879905e-01 -4.18135941e-01 1.15468729e+00 -3.21717076e-02
6.19965136e-01 -1.73396587e-01 -3.14555585e-01 1.67470604e-01
1.53581023e+00 -3.65622388e-03 6.66109085e-01 5.77319980e-01
6.24366283e-01 3.81668061e-01 4.07817274e-01 -1.87314767e-02
5.58076441e-01 8.97663653e-01 -9.46002454e-03 -1.45921826e-01
-4.95724946e-01 -7.97545686e-02 4.61073786e-01 1.46948647e+00
-2.48497993e-01 -3.77205819e-01 -1.04528785e+00 5.57013631e-01
-1.85011017e+00 -1.00140285e+00 -3.64573061e-01 1.79875124e+00
4.65943754e-01 3.15067500e-01 3.07918847e-01 2.95275271e-01
3.89043063e-01 5.58718741e-01 1.19346999e-01 -1.25312674e+00
-2.44265869e-01 5.73896945e-01 4.58949745e-01 6.14744961e-01
-9.60601270e-01 1.01162422e+00 8.09287739e+00 6.64538205e-01
-7.47799873e-01 3.32713574e-01 1.32633165e-01 1.94427013e-01
-3.58157396e-01 -1.01378858e-01 -5.48962772e-01 3.11449111e-01
1.44320476e+00 -3.00243795e-01 2.92192459e-01 4.31023449e-01
-2.03411892e-01 9.90601331e-02 -9.85607207e-01 7.48684347e-01
7.82388747e-01 -1.21364331e+00 8.23322758e-02 2.36481935e-01
4.94332254e-01 4.23693925e-01 -1.44431964e-01 2.31724232e-01
3.01079601e-01 -7.76664138e-01 1.43978328e-01 4.52408582e-01
7.58019388e-01 -6.85944021e-01 9.24201071e-01 -8.79669264e-02
-1.00523674e+00 5.06158948e-01 -5.41601717e-01 -3.21115404e-01
-2.31797859e-01 4.98244286e-01 -1.03874612e+00 9.92031574e-01
4.55208093e-01 9.77898598e-01 -9.65769231e-01 1.05772996e+00
4.37740386e-02 5.77015758e-01 -3.34027201e-01 -6.16894066e-01
4.74716395e-01 -1.55528858e-01 7.49391079e-01 1.58464122e+00
2.28198782e-01 -5.35079300e-01 -3.34453047e-03 2.10354291e-02
4.99503873e-02 3.59123915e-01 -1.34320807e+00 -1.28138736e-01
-8.92706960e-02 1.28891885e+00 -5.93282044e-01 -5.12960374e-01
-4.44569409e-01 1.13990462e+00 3.20851594e-01 -2.76049525e-02
-5.90102851e-01 -8.19798648e-01 1.24822684e-01 -1.12206601e-01
2.10231692e-01 -5.01051128e-01 -6.41613528e-02 -1.37423635e+00
-1.03639975e-01 -6.53564990e-01 5.96768260e-01 -7.90125430e-01
-1.66506600e+00 1.15483642e+00 8.49802345e-02 -1.23344827e+00
-5.12593269e-01 -8.15893054e-01 -5.83396792e-01 6.52665734e-01
-1.34849274e+00 -1.00583911e+00 3.01351398e-01 4.59753305e-01
3.24041635e-01 -5.28286159e-01 1.31725919e+00 3.95577163e-01
-2.55903542e-01 7.01664031e-01 4.57631916e-01 5.82670383e-02
9.91223335e-01 -1.69017529e+00 5.82904935e-01 7.42874265e-01
8.04374039e-01 3.72456133e-01 8.19233239e-01 -3.49862754e-01
-1.32148242e+00 -8.73785317e-01 1.39221871e+00 -7.11531520e-01
1.31627917e+00 -4.42006260e-01 -6.60285413e-01 5.45758545e-01
8.81962776e-01 -4.32903051e-01 8.29117000e-01 3.66335630e-01
-5.14543116e-01 -2.44123279e-03 -8.95003021e-01 2.37167612e-01
1.08968747e+00 -5.25912464e-01 -1.25633037e+00 7.08544195e-01
8.15714896e-01 -1.23799689e-01 -1.16550505e+00 1.71410203e-01
4.11831349e-01 -5.79631150e-01 6.86821103e-01 -7.83424914e-01
3.92396450e-01 -1.42790437e-01 -2.67773330e-01 -1.64043486e+00
-3.95299435e-01 -5.03457963e-01 -1.45987406e-01 9.92620945e-01
8.65935028e-01 -1.00927222e+00 5.41246116e-01 7.60118440e-02
-1.35121033e-01 -6.44161165e-01 -1.15218675e+00 -1.18208158e+00
1.54605895e-01 -2.31111184e-01 1.82383493e-01 1.14393294e+00
3.78448784e-01 7.92807043e-01 -4.87425238e-01 -4.15344179e-01
5.80531418e-01 5.28225526e-02 6.31842315e-01 -1.15098298e+00
-3.75390857e-01 -5.24274588e-01 -9.63815570e-01 -4.64836389e-01
1.36183247e-01 -1.36295521e+00 -2.70613462e-01 -1.87685394e+00
2.01304466e-01 -9.46942568e-02 -4.93967235e-01 7.22992778e-01
1.78728908e-01 6.93834722e-01 1.63231239e-01 2.05799654e-01
-7.29153156e-01 3.96451771e-01 1.13399434e+00 -2.60467470e-01
2.33605713e-01 -3.64925921e-01 -6.68958783e-01 2.31801167e-01
1.02023482e+00 -5.66797435e-01 -2.10669681e-01 -6.00106895e-01
-4.46672924e-02 -2.44289428e-01 4.18604724e-02 -8.84333432e-01
1.30018219e-01 5.62073350e-01 2.46747360e-01 -2.71419138e-01
3.90236378e-01 -6.94631040e-01 8.63828813e-04 7.16195166e-01
-2.98802793e-01 7.96965063e-01 1.49088338e-01 4.06910241e-01
-4.12845641e-01 -1.23549804e-01 3.07458103e-01 -1.06402105e-02
-6.38078392e-01 1.55494779e-01 -4.27196592e-01 -3.24291497e-01
9.04285908e-01 -2.36218601e-01 -6.01501167e-01 -3.50411624e-01
-6.64407611e-01 4.34832834e-02 4.08585697e-01 6.59986198e-01
6.43395305e-01 -1.47797227e+00 -9.27181542e-01 -6.36483356e-02
3.57499182e-01 -8.91863167e-01 -4.96228546e-01 8.41526926e-01
-6.23253345e-01 6.34233534e-01 -2.80464709e-01 -2.79726565e-01
-1.18588340e+00 5.12171090e-01 1.02325328e-01 -5.50420165e-01
-3.92421544e-01 2.82221854e-01 -6.55261695e-01 -6.61068976e-01
-2.33778447e-01 1.21368365e-02 -3.25229555e-01 2.51205772e-01
4.32118207e-01 2.47121707e-01 4.03196871e-01 -6.78840578e-01
-4.57273990e-01 4.32857513e-01 -3.25179309e-01 -1.27773717e-01
1.49888778e+00 4.78106737e-02 -4.18976158e-01 8.28142226e-01
1.58918214e+00 -2.98908185e-02 -3.50000635e-02 -2.60763198e-01
1.87886074e-01 -3.21254700e-01 3.38552415e-01 -9.22883570e-01
-6.22227132e-01 7.56568909e-01 7.29488432e-01 8.06668878e-01
6.89691901e-01 6.15905523e-02 7.32016861e-01 6.71586335e-01
5.63971221e-01 -8.57782781e-01 -9.18237492e-02 5.41755736e-01
6.51721537e-01 -1.03786361e+00 2.04158098e-01 -3.25963162e-02
-4.94848788e-01 1.36759925e+00 2.03324825e-01 -4.15810794e-01
6.77524865e-01 3.66073608e-01 4.83104177e-02 -4.11378652e-01
-9.02930081e-01 -4.96685296e-01 3.71778011e-01 4.90674466e-01
7.31938779e-01 8.51524100e-02 -9.93758976e-01 -2.03388527e-01
-2.17686504e-01 -3.34477365e-01 5.78313053e-01 9.69154060e-01
-3.27167749e-01 -1.59640586e+00 5.63098118e-02 7.72313237e-01
-3.26944500e-01 -3.09303015e-01 -9.28410113e-01 1.14138067e+00
-2.99102515e-01 7.93998361e-01 -4.48018909e-02 -9.29088712e-01
3.86602670e-01 2.52709389e-01 6.27942622e-01 -7.51029193e-01
-6.93937004e-01 -2.91335613e-01 7.26994038e-01 -4.17391241e-01
-7.15325952e-01 -5.25478780e-01 -7.13128388e-01 -6.00681722e-01
-5.41344583e-01 3.97954702e-01 7.55439878e-01 7.62059093e-01
2.45259479e-01 3.49150866e-01 6.45720065e-01 -7.58827984e-01
-3.38327467e-01 -1.25661278e+00 -6.41934276e-01 1.45195723e-01
-1.03881240e-01 -5.00765145e-01 -5.10842264e-01 -2.34910890e-01] | [10.600967407226562, 8.748727798461914] |
854e0a1a-3648-41a5-87df-5fdf3a6ac029 | interpretable-machine-learning-based-on | 2305.1567 | null | https://arxiv.org/abs/2305.15670v1 | https://arxiv.org/pdf/2305.15670v1.pdf | Interpretable Machine Learning based on Functional ANOVA Framework: Algorithms and Comparisons | In the early days of machine learning (ML), the emphasis was on developing complex algorithms to achieve best predictive performance. To understand and explain the model results, one had to rely on post hoc explainability techniques, which are known to have limitations. Recently, with the recognition that interpretability is just as important, researchers are compromising on small increases in predictive performance to develop algorithms that are inherently interpretable. While doing so, the ML community has rediscovered the use of low-order functional ANOVA (fANOVA) models that have been known in the statistical literature for some time. This paper starts with a description of challenges with post hoc explainability and reviews the fANOVA framework with a focus on main effects and second-order interactions. This is followed by an overview of two recently developed techniques: Explainable Boosting Machines or EBM (Lou et al., 2013) and GAMI-Net (Yang et al., 2021b). The paper proposes a new algorithm, called GAMI-Lin-T, that also uses trees like EBM, but it does linear fits instead of piecewise constants within the partitions. There are many other differences, including the development of a new interaction filtering algorithm. Finally, the paper uses simulated and real datasets to compare selected ML algorithms. The results show that GAMI-Lin-T and GAMI-Net have comparable performances, and both are generally better than EBM. | ['Jie Chen', 'Aijun Zhang', 'Agus Sudjianto', 'Vijayan N. Nair', 'Linwei Hu'] | 2023-05-25 | null | null | null | null | ['interpretable-machine-learning'] | ['methodology'] | [-6.25796616e-03 1.34432644e-01 -2.71782398e-01 -6.26636803e-01
-3.48270595e-01 -3.87363315e-01 4.96887326e-01 3.04348916e-01
1.86347261e-01 8.67527723e-01 -2.77949665e-02 -7.84327984e-01
-6.69916153e-01 -4.86350656e-01 -8.99106801e-01 -4.49964225e-01
-3.06032717e-01 4.26385045e-01 -2.06286535e-01 -5.21470867e-02
4.26231146e-01 3.42577025e-02 -2.01648974e+00 5.15686452e-01
1.15519619e+00 5.26148081e-01 2.68402368e-01 6.92099690e-01
-2.17343606e-02 6.51187956e-01 -4.60881531e-01 -4.56370205e-01
-1.15936860e-01 -7.87700295e-01 -6.37832046e-01 -4.16502804e-01
2.15017885e-01 1.01592094e-01 3.81960630e-01 3.22292715e-01
2.09838882e-01 3.93702164e-02 7.17084706e-01 -1.80558848e+00
-5.26753247e-01 7.70013094e-01 -5.55152357e-01 -1.98342055e-01
2.51065165e-01 -4.69715632e-02 1.01085734e+00 -9.29913878e-01
1.46727473e-01 1.47590208e+00 1.20089126e+00 1.92558110e-01
-1.74483097e+00 -8.06792676e-01 2.19791502e-01 4.69657332e-01
-1.14480150e+00 6.07059821e-02 3.81022274e-01 -6.36260033e-01
1.09973145e+00 9.68492329e-01 8.33084702e-01 7.43503869e-01
6.42908931e-01 5.92445433e-01 1.64050746e+00 -8.35934937e-01
3.15273494e-01 1.90133020e-01 5.89351416e-01 3.58041078e-01
4.56365943e-01 4.82886612e-01 -5.82114637e-01 -3.24020535e-01
3.97525102e-01 -1.21547803e-01 3.62790115e-02 -4.91683602e-01
-8.28494668e-01 1.27354431e+00 4.87972587e-01 3.33616585e-01
-3.19867194e-01 6.11588098e-02 3.08271736e-01 1.24294162e-01
7.57552683e-01 4.55747873e-01 -7.06096649e-01 -1.42486751e-01
-1.00855708e+00 6.12984955e-01 6.76038086e-01 4.09178287e-01
6.23626709e-01 -1.65010914e-01 -7.06247613e-02 7.38420367e-01
5.36784947e-01 7.19912797e-02 1.55900419e-01 -5.66248238e-01
-6.12090491e-02 6.58642948e-01 -8.03887546e-02 -1.07365763e+00
-6.65463746e-01 -4.24528271e-01 -9.59731162e-01 3.43917012e-01
3.15618396e-01 2.20873907e-01 -9.01660740e-01 1.70013285e+00
3.82958382e-01 -1.57292277e-01 -5.61587036e-01 6.92294955e-01
6.49612784e-01 5.58535755e-01 4.80245411e-01 -3.07834268e-01
1.02809691e+00 -9.55990374e-01 -9.32823718e-01 -2.63604343e-01
6.61886394e-01 -6.76953554e-01 1.17499208e+00 7.49888361e-01
-8.13351035e-01 -8.24621022e-01 -1.04338670e+00 2.79109180e-02
-4.70976830e-01 -1.90827504e-01 1.41513002e+00 1.00564206e+00
-9.65097189e-01 8.12985659e-01 -8.76712799e-01 -3.49377304e-01
2.13375509e-01 6.19284451e-01 -3.30457866e-01 6.38370635e-03
-9.81208324e-01 1.04147446e+00 1.33094311e-01 1.03153892e-01
-4.46004540e-01 -1.04082048e+00 -6.51966929e-01 -1.19189687e-01
1.81595460e-02 -8.59986305e-01 9.26236153e-01 -1.10392261e+00
-1.14371490e+00 5.70069730e-01 -4.15896982e-01 -4.51560169e-01
3.14562351e-01 -2.93847859e-01 -3.27791087e-02 -6.98493779e-01
4.80703311e-03 6.53651893e-01 3.33493888e-01 -1.40694189e+00
-5.50884426e-01 -5.71756184e-01 2.98379678e-02 -1.01436563e-01
1.41331553e-01 3.52157392e-02 1.83485031e-01 -6.38804018e-01
1.71184599e-01 -8.71365964e-01 -2.30012774e-01 -4.69292760e-01
-2.99881846e-01 -2.43331671e-01 5.13990343e-01 -7.84847915e-01
1.71817052e+00 -1.74910653e+00 -4.58789468e-02 1.36489525e-01
3.62959743e-01 -9.72952470e-02 1.86939299e-01 5.94166815e-01
-6.90747082e-01 3.74691159e-01 -4.62300330e-01 -2.84697592e-01
-3.55738662e-02 3.97846639e-01 -1.30712941e-01 2.85622776e-01
-8.53961185e-02 8.79345059e-01 -5.59648752e-01 -1.11051001e-01
4.89917070e-01 4.35102761e-01 -7.87969232e-01 9.60058868e-02
2.18348160e-01 5.37216365e-01 2.41194680e-01 4.06769425e-01
8.98720264e-01 -2.05140337e-02 1.91319317e-01 9.62580442e-02
-4.61526215e-01 3.36192608e-01 -1.11244726e+00 9.24813151e-01
-4.43307936e-01 7.60090590e-01 -6.39864206e-02 -1.34683347e+00
8.62609148e-01 1.11371830e-01 4.43504095e-01 -2.18826696e-01
5.52074648e-02 3.56997728e-01 4.36985433e-01 -1.53033853e-01
7.84859806e-02 -2.95829892e-01 2.67908871e-01 1.42002463e-01
-1.73322082e-01 -2.28973389e-01 1.20409932e-02 -1.80644870e-01
9.60226119e-01 3.00391614e-01 8.73601496e-01 -7.04569042e-01
3.69435757e-01 1.46137774e-01 4.55346882e-01 9.93017912e-01
1.59099057e-01 3.59113157e-01 3.72757316e-01 -6.88530207e-01
-7.80357003e-01 -8.57044458e-01 -6.86367333e-01 1.12485397e+00
-2.82066584e-01 -6.58526361e-01 -9.06000495e-01 -5.51011741e-01
3.58720869e-01 1.25002348e+00 -1.01113009e+00 -8.22900310e-02
-8.62384886e-02 -1.10241044e+00 1.04756895e-02 3.58509064e-01
2.05216408e-01 -7.49079287e-01 -5.20389020e-01 9.86041427e-02
-1.49794951e-01 -2.07272291e-01 2.58828700e-01 7.81144142e-01
-1.19377291e+00 -1.02806497e+00 -1.18829295e-01 -1.24967530e-01
5.92783332e-01 3.61952513e-01 1.33049369e+00 2.46406153e-01
-3.91767234e-01 2.25201249e-01 -4.67620313e-01 -9.03245330e-01
-3.96847218e-01 -6.92189410e-02 -6.42044917e-02 -4.39257890e-01
6.60735846e-01 -6.38608336e-01 -2.95232743e-01 5.76246202e-01
-8.24395835e-01 5.41030407e-01 4.74114537e-01 1.08730960e+00
5.49196422e-01 1.43671140e-01 4.40348476e-01 -9.97897744e-01
3.81317139e-01 -5.76919794e-01 -3.47477764e-01 1.72996998e-01
-1.20776129e+00 -1.22868337e-01 3.20304900e-01 -2.99500674e-01
-7.07716525e-01 -1.25809774e-01 -8.40824470e-02 3.44396532e-02
-7.68318865e-03 8.34328473e-01 -2.42668077e-01 -1.32780090e-01
7.61206985e-01 -3.49719137e-01 2.44327318e-02 -5.85998416e-01
2.52668560e-01 6.97051287e-01 -1.47169441e-01 -3.61563772e-01
5.53363562e-01 5.02828360e-02 1.35188192e-01 -6.37238801e-01
-6.97953939e-01 -1.90861523e-02 -6.88580811e-01 -3.84538144e-01
5.72038472e-01 -5.30146420e-01 -8.20079982e-01 2.31133640e-01
-8.57459486e-01 -5.28902709e-01 -5.10855988e-02 5.73395014e-01
-6.12068653e-01 -1.16190445e-02 -1.63059101e-01 -1.12912905e+00
-5.45924343e-02 -1.07332122e+00 7.12014258e-01 3.97605114e-02
-9.08635437e-01 -1.01362431e+00 7.47579262e-02 5.44492006e-01
4.11371350e-01 4.13233191e-01 1.19803858e+00 -6.89138353e-01
-2.26123035e-01 -1.93879098e-01 4.99152541e-02 1.00131318e-01
1.91255525e-01 2.31837347e-01 -1.08149159e+00 -2.25251578e-02
7.11495206e-02 1.46973923e-01 6.27819359e-01 1.15265274e+00
1.28479934e+00 -2.64120072e-01 -4.88879621e-01 4.60052609e-01
1.09067178e+00 3.63619924e-01 5.82525194e-01 4.94229972e-01
4.74125206e-01 9.63755250e-01 8.53857934e-01 2.04476848e-01
5.31206310e-01 8.25767756e-01 3.89547110e-01 -2.41576940e-01
1.62803590e-01 -3.87413472e-01 2.65182465e-01 7.66071379e-01
-3.29326957e-01 -7.08919615e-02 -9.30918753e-01 1.66661203e-01
-2.13115263e+00 -8.15762520e-01 -1.03591764e+00 2.50950289e+00
4.97808695e-01 1.83842257e-02 2.09525123e-01 5.25578260e-01
5.97880125e-01 -4.36092585e-01 -3.21415991e-01 -1.00443661e+00
-1.21762045e-01 1.65821850e-01 3.55597168e-01 7.38927543e-01
-1.00567603e+00 5.22481680e-01 7.33933783e+00 8.38448167e-01
-7.19020784e-01 9.15490538e-02 1.08104134e+00 2.29276821e-01
-3.76618713e-01 3.89340609e-01 -5.15019417e-01 4.18117285e-01
1.15906966e+00 -1.14875846e-01 5.52493215e-01 9.49174166e-01
6.27569616e-01 -6.45542324e-01 -1.12668908e+00 7.25906014e-01
-2.32826367e-01 -8.60734224e-01 -2.49321848e-01 1.64992645e-01
6.02262497e-01 -4.95878756e-01 -7.32833743e-02 5.45053840e-01
1.98094979e-01 -1.50009656e+00 6.42732620e-01 4.57003415e-01
2.64472932e-01 -7.98314631e-01 9.15553927e-01 3.47876042e-01
-9.30592895e-01 -8.04362968e-02 -4.73002613e-01 -8.90551090e-01
-2.34484375e-01 1.05096209e+00 -7.28485346e-01 7.09386468e-01
1.04788828e+00 6.45283043e-01 -5.68739533e-01 1.05221188e+00
-3.13773304e-01 1.09699905e+00 -2.20763713e-01 -3.76055352e-02
-1.94313973e-01 -2.07425147e-01 2.24335715e-01 1.00098467e+00
3.70615661e-01 1.03456199e-01 -1.37440741e-01 8.43539953e-01
7.61624753e-01 2.68886000e-01 -5.56053460e-01 2.06383988e-01
2.93405563e-01 1.23050511e+00 -7.56575227e-01 -3.22271176e-02
-3.89779747e-01 4.77696508e-01 -1.09389918e-02 -1.08571248e-02
-1.04057109e+00 9.56514999e-02 6.60470426e-01 3.80819559e-01
-1.26744747e-01 3.33714038e-02 -9.83762324e-01 -5.37512600e-01
-2.54165709e-01 -1.24362564e+00 5.63684583e-01 -9.27450299e-01
-1.26920748e+00 1.28632829e-01 4.42582399e-01 -8.92338395e-01
-2.33114719e-01 -6.56856298e-01 -3.28910172e-01 1.01952195e+00
-6.87546849e-01 -1.25153136e+00 -3.59223455e-01 6.36208579e-02
5.17114997e-01 3.22689295e-01 9.78324592e-01 1.69415861e-01
-3.68783236e-01 3.52631271e-01 2.81110734e-01 -7.82715082e-01
7.05033004e-01 -1.35891640e+00 3.30672681e-01 3.05589914e-01
7.23086968e-02 9.91617918e-01 1.19729519e+00 -8.05730402e-01
-1.11712503e+00 -6.10693038e-01 9.85295594e-01 -7.62212396e-01
3.19504619e-01 -5.67002773e-01 -1.02039707e+00 7.92192698e-01
2.35677168e-01 -5.67549765e-01 1.04056704e+00 7.49312520e-01
7.30126053e-02 -1.44203931e-01 -1.03818345e+00 5.01100540e-01
9.01747048e-01 1.70989856e-01 -3.51425529e-01 1.57253519e-01
3.82836610e-01 -3.61929893e-01 -7.16696620e-01 7.91176736e-01
8.62730026e-01 -1.38497233e+00 8.59662831e-01 -6.49949253e-01
4.43258941e-01 -2.40595728e-01 -9.82758924e-02 -1.35872293e+00
-5.42955756e-01 -4.49209243e-01 2.02009901e-01 1.26187968e+00
4.12452966e-01 -6.65648162e-01 5.66566646e-01 1.05524015e+00
-9.76421237e-02 -9.27087724e-01 -9.06218767e-01 -8.30373287e-01
1.87698230e-01 -8.54765713e-01 5.77942610e-01 9.09882128e-01
1.06338203e-01 3.59084785e-01 -5.21276593e-01 -2.22048491e-01
5.20717621e-01 6.32227361e-02 1.16642153e+00 -1.60226905e+00
-4.47597206e-01 -4.56004739e-01 -5.33595085e-01 -4.88258123e-01
-1.42466381e-01 -8.91657531e-01 -1.69595294e-02 -1.48300564e+00
6.02198303e-01 -1.73244879e-01 -3.00733596e-01 6.65030658e-01
-5.94155967e-01 -3.47798690e-02 -2.75883300e-04 2.87424903e-02
1.27541184e-01 4.08919185e-01 8.59764874e-01 6.93625137e-02
-4.22986686e-01 2.67917991e-01 -8.79939854e-01 7.08923578e-01
7.61539578e-01 -7.05959737e-01 -4.86435473e-01 2.67116092e-02
2.99692452e-01 -1.65350169e-01 6.02902174e-01 -8.77518833e-01
-3.17216337e-01 -2.08127424e-01 6.33897901e-01 -7.34433174e-01
-6.17728308e-02 -8.20953548e-01 9.90274906e-01 6.98812306e-01
-3.26489478e-01 4.01577473e-01 6.76066279e-01 1.94260687e-01
-1.46605253e-01 -2.16918156e-01 5.29162765e-01 2.49790475e-01
-2.28034422e-01 -3.41164351e-01 -4.93378431e-01 -5.85433245e-01
1.11578691e+00 -4.25920367e-01 -3.57631664e-03 -6.30436838e-01
-6.60353541e-01 2.00166374e-01 1.45565465e-01 4.10285503e-01
1.46140769e-01 -1.13366473e+00 -6.30875707e-01 6.21428192e-02
-1.20255284e-01 -4.82383162e-01 5.38685381e-01 1.23951137e+00
-3.60089988e-01 7.53587663e-01 -1.23650655e-01 -7.50607252e-01
-1.59924912e+00 7.62996554e-01 -1.43598148e-03 -3.58755529e-01
-2.21600756e-01 8.41141939e-01 6.42471850e-01 -7.71553993e-01
-2.18497366e-01 -3.05234551e-01 -6.45201281e-02 1.82138998e-02
3.51777911e-01 6.02606833e-01 9.69565734e-02 -2.45938703e-01
-5.03038466e-01 3.33960921e-01 2.78274089e-01 3.95181403e-02
1.54609156e+00 -1.10890493e-01 -5.06088078e-01 8.75298619e-01
7.34876573e-01 -1.85621679e-02 -7.57216513e-01 6.31950915e-01
1.74258649e-01 -5.97339451e-01 8.40075240e-02 -1.21820366e+00
-4.88773763e-01 9.16931570e-01 6.31487906e-01 4.80048895e-01
1.23045921e+00 -2.61509776e-01 -1.59552827e-01 -1.78839713e-01
3.41199189e-01 -7.12477684e-01 -5.39283752e-01 2.59980768e-01
1.07318890e+00 -1.14293039e+00 3.34847212e-01 -6.10810935e-01
-4.50312853e-01 1.04932606e+00 5.06146252e-01 3.18784207e-01
7.27256656e-01 2.04722553e-01 -1.31866947e-01 2.15000380e-02
-7.21296608e-01 4.09284718e-02 5.65991282e-01 6.41330361e-01
9.83960211e-01 3.16749096e-01 -1.06295931e+00 1.14079142e+00
-5.76498985e-01 7.04619288e-03 7.25909024e-02 5.08169174e-01
-3.72150093e-01 -1.34154308e+00 -8.68825436e-01 7.15332687e-01
-2.76356816e-01 -7.09085390e-02 -6.71398044e-01 1.40171242e+00
2.90324539e-01 1.35789025e+00 -1.63584307e-01 -7.51491427e-01
1.44257858e-01 3.57201129e-01 2.63601393e-01 -2.73099393e-01
-8.20563018e-01 -3.06626558e-02 -2.01966818e-02 -5.20963907e-01
-3.76920104e-01 -8.15732002e-01 -8.05180192e-01 -6.86439455e-01
-5.87830603e-01 4.20130432e-01 9.11117911e-01 8.71693254e-01
2.85706371e-01 7.69204438e-01 5.32232285e-01 -8.53266358e-01
-9.36856419e-02 -1.10557091e+00 -4.56266522e-01 -1.10835889e-02
-9.27241594e-02 -1.08986688e+00 -6.13449454e-01 -1.30674735e-01] | [8.434959411621094, 5.338274955749512] |
98c74a2f-b3ba-48d5-9c66-53fb95ca7a7b | knowledge-cross-distillation-for-membership | 2111.01363 | null | https://arxiv.org/abs/2111.01363v3 | https://arxiv.org/pdf/2111.01363v3.pdf | Knowledge Cross-Distillation for Membership Privacy | A membership inference attack (MIA) poses privacy risks for the training data of a machine learning model. With an MIA, an attacker guesses if the target data are a member of the training dataset. The state-of-the-art defense against MIAs, distillation for membership privacy (DMP), requires not only private data for protection but a large amount of unlabeled public data. However, in certain privacy-sensitive domains, such as medicine and finance, the availability of public data is not guaranteed. Moreover, a trivial method for generating public data by using generative adversarial networks significantly decreases the model accuracy, as reported by the authors of DMP. To overcome this problem, we propose a novel defense against MIAs that uses knowledge distillation without requiring public data. Our experiments show that the privacy protection and accuracy of our defense are comparable to those of DMP for the benchmark tabular datasets used in MIA research, Purchase100 and Texas100, and our defense has a much better privacy-utility trade-off than those of the existing defenses that also do not use public data for the image dataset CIFAR10. | ['Hikaru Tsuchida', 'Isamu Teranishi', 'Junki Mori', 'Kunihiro Ito', 'Batnyam Enkhtaivan', 'Rishav Chourasia'] | 2021-11-02 | null | null | null | null | ['membership-inference-attack'] | ['computer-vision'] | [ 1.23256646e-01 3.43685985e-01 -1.84938997e-01 -4.67932224e-01
-1.09318304e+00 -1.10827875e+00 4.10911769e-01 1.34056151e-01
-3.58173311e-01 1.13932550e+00 -4.05006409e-01 -7.54668653e-01
7.52467737e-02 -1.48105705e+00 -1.01988733e+00 -9.25274134e-01
6.11748286e-02 5.12537718e-01 9.07491427e-04 -8.80246311e-02
-6.91900328e-02 5.21867514e-01 -8.93238664e-01 5.01504779e-01
7.21393406e-01 1.09795403e+00 -8.46503258e-01 2.75305092e-01
3.24390411e-01 4.30709124e-01 -8.23172688e-01 -1.12750888e+00
8.97572577e-01 -2.92415380e-01 -8.41993928e-01 -3.92740995e-01
4.85153407e-01 -4.51715201e-01 -6.27832472e-01 1.30137086e+00
1.80257693e-01 -2.30251446e-01 4.80797738e-01 -1.78080475e+00
-6.71560943e-01 8.19905221e-01 -4.52248544e-01 -2.65573442e-01
1.85628831e-02 5.79553246e-02 8.68503928e-01 -1.60197690e-01
5.43161690e-01 1.11572254e+00 4.77193981e-01 8.39033425e-01
-1.26137841e+00 -1.31698716e+00 -2.33425081e-01 4.46088910e-02
-1.68453944e+00 -2.33738676e-01 4.48065996e-01 -2.32206628e-01
2.54851997e-01 7.86155403e-01 8.87633860e-02 1.20395815e+00
2.04000369e-01 5.21455765e-01 1.25751722e+00 -5.67896701e-02
4.86511350e-01 6.36698902e-01 1.05314985e-01 5.54168165e-01
8.12567055e-01 3.47561479e-01 -1.75136894e-01 -1.15160823e+00
5.17503381e-01 5.13993278e-02 -4.93571043e-01 -6.07528567e-01
-6.94354773e-01 1.16962540e+00 1.67597547e-01 -1.49059355e-01
-2.59712785e-02 7.39854425e-02 3.13426763e-01 6.14376545e-01
3.21012288e-01 4.85081613e-01 -4.21964437e-01 4.56071407e-01
-7.44942188e-01 4.63591427e-01 1.19575489e+00 8.61000240e-01
6.94571316e-01 -1.61317259e-01 -3.42332497e-02 1.20727839e-02
2.25800350e-01 6.50199056e-01 2.35403374e-01 -8.39614809e-01
6.09937012e-01 3.87448102e-01 2.35819399e-01 -1.17281604e+00
3.30328315e-01 -1.86650559e-01 -8.85809720e-01 3.94786090e-01
8.38555455e-01 -4.29316789e-01 -7.72756398e-01 2.12354064e+00
4.76528108e-01 1.79122090e-02 6.48789763e-01 5.22243977e-01
3.87640893e-01 6.38794899e-01 -6.39834180e-02 4.49512750e-02
1.33536029e+00 -3.54216188e-01 -6.49682403e-01 -4.02998732e-04
6.19902194e-01 -2.96804994e-01 5.37592709e-01 3.90835345e-01
-9.43615556e-01 7.34687522e-02 -1.17216718e+00 1.01234354e-01
-6.81401432e-01 -3.23850304e-01 9.07322288e-01 1.43486845e+00
-5.48008859e-01 4.77084100e-01 -6.37129426e-01 -8.21029861e-03
9.87875342e-01 5.88698149e-01 -8.75933647e-01 -8.72512162e-02
-1.61897707e+00 5.03623903e-01 5.03337741e-01 -3.92844856e-01
-1.21563435e+00 -6.63445890e-01 -8.91380250e-01 3.00477654e-01
6.04488909e-01 -6.05944395e-01 7.98441648e-01 -6.58430099e-01
-1.12477982e+00 1.12083697e+00 4.16080058e-01 -1.11362159e+00
1.00494230e+00 8.63801911e-02 -4.08975691e-01 5.47667034e-02
-2.68936008e-01 3.80165488e-01 8.31831515e-01 -1.40578020e+00
-5.17044842e-01 -6.25064671e-01 5.57893276e-01 -3.78019482e-01
-3.69087487e-01 -1.70529783e-01 6.77358434e-02 -6.11837029e-01
-2.33059511e-01 -1.05208135e+00 -5.21925688e-01 1.95142016e-01
-7.85622716e-01 1.49132922e-01 1.23456872e+00 -5.77711046e-01
1.06832230e+00 -2.37516689e+00 -3.96468788e-01 8.94046009e-01
2.38591686e-01 5.50915182e-01 1.32141471e-01 2.59378523e-01
-7.46458471e-02 6.44508958e-01 -5.13686180e-01 -2.32955962e-01
1.08724408e-01 3.74334157e-01 -8.29926729e-01 7.01684833e-01
-2.29499221e-01 6.53841317e-01 -5.83482683e-01 -1.79354355e-01
-2.28658825e-01 2.66327649e-01 -7.47979820e-01 7.29412958e-02
-2.83815116e-01 3.19351733e-01 -4.25678939e-01 4.81836289e-01
1.24344838e+00 -4.14010584e-01 3.60696346e-01 1.89213872e-01
4.71522689e-01 2.43788119e-02 -1.01294899e+00 1.17262781e+00
-3.50451954e-02 2.11609200e-01 1.38125747e-01 -6.82087481e-01
8.18151116e-01 4.94694382e-01 2.60082155e-01 4.29561846e-02
2.74552017e-01 8.92513990e-02 -3.86027321e-02 -2.70522647e-02
5.19298986e-02 -7.66220763e-02 -4.31172818e-01 7.24877000e-01
-3.49031746e-01 9.92813185e-02 -3.66830260e-01 2.82507151e-01
9.45763528e-01 -4.85453606e-01 1.32458225e-01 -2.92189181e-01
6.00397289e-01 -1.92413166e-01 8.46810222e-01 9.76271689e-01
-1.08984016e-01 5.53107440e-01 7.47649908e-01 -4.11947191e-01
-8.45656157e-01 -1.09951246e+00 -1.21372372e-01 4.51141566e-01
2.97225155e-02 -4.52361077e-01 -1.13920844e+00 -1.29526281e+00
3.87993217e-01 7.20954180e-01 -7.06975937e-01 -2.58471429e-01
-2.75729209e-01 -8.48540723e-01 1.05450165e+00 2.26413235e-01
8.74385238e-01 -4.51918185e-01 -1.56732053e-01 -2.99733162e-01
-2.91752249e-01 -1.21517360e+00 -5.48349082e-01 -1.87490910e-01
-4.24865663e-01 -1.41637087e+00 -2.85699368e-01 -4.03459877e-01
9.11699235e-01 4.16960195e-02 6.14835978e-01 -5.05155399e-02
-1.00558460e-01 8.37888718e-02 1.32298365e-01 -6.57272100e-01
-8.13696384e-01 -7.03334659e-02 3.72961648e-02 3.21721196e-01
3.60402495e-01 -5.69329619e-01 -4.76573914e-01 3.43942553e-01
-1.06925273e+00 -2.64020801e-01 2.15135813e-01 6.91068232e-01
5.13312399e-01 3.60452563e-01 5.66697299e-01 -1.76172924e+00
5.16323984e-01 -6.86690569e-01 -1.00134337e+00 4.05823022e-01
-5.29172957e-01 -8.58384520e-02 7.77158976e-01 -5.10444939e-01
-7.82923400e-01 1.11629620e-01 -1.31592095e-01 -6.16692841e-01
-1.94557011e-01 1.62694231e-01 -6.67037308e-01 -3.47500503e-01
6.99828565e-01 1.05044849e-01 1.93488911e-01 -3.53906184e-01
2.49878630e-01 6.40836716e-01 5.72614908e-01 -6.37874246e-01
1.14925301e+00 6.94933355e-01 3.64975423e-01 -2.96826392e-01
-7.04619646e-01 1.70864165e-01 -5.37848920e-02 6.74655974e-01
7.51696706e-01 -7.76431501e-01 -1.12289035e+00 6.76450431e-01
-9.15127337e-01 4.18335125e-02 -4.03214514e-01 3.21475029e-01
-5.18164873e-01 5.94547868e-01 -4.87463474e-01 -7.22532153e-01
-5.50704241e-01 -1.06561255e+00 2.24909738e-01 4.72096987e-02
-1.04762120e-02 -8.43676448e-01 -3.71481538e-01 7.49447763e-01
3.15812081e-01 8.74041438e-01 1.10860944e+00 -1.24416387e+00
-8.17241192e-01 -8.05069804e-01 1.84058528e-02 5.56853712e-01
2.10727870e-01 -2.32097894e-01 -1.14995766e+00 -6.21614754e-01
3.72860700e-01 -3.66516829e-01 6.21177852e-01 -5.58881611e-02
1.83933330e+00 -1.16008389e+00 -2.02149838e-01 8.49611998e-01
1.32603645e+00 3.03985626e-01 8.24841321e-01 1.45172328e-01
4.35916573e-01 5.52167475e-01 4.78382915e-01 6.15957022e-01
2.24288642e-01 3.02556038e-01 6.56941354e-01 -1.81144774e-01
7.53033221e-01 -4.64499086e-01 1.35197580e-01 -3.21588963e-01
2.19944283e-01 -4.94213879e-01 -5.42840481e-01 3.40825021e-01
-1.80148673e+00 -1.03524172e+00 6.24704435e-02 2.64504504e+00
1.14539254e+00 -5.16527519e-03 2.70849224e-02 1.35867730e-01
5.01136124e-01 6.42698258e-02 -4.68637049e-01 -4.86213326e-01
-2.61368394e-01 3.03259760e-01 1.08750784e+00 4.91050571e-01
-1.38507950e+00 7.04204619e-01 6.49835253e+00 8.88557196e-01
-8.04937780e-01 2.03223333e-01 1.12472105e+00 -1.24183096e-01
-4.44598764e-01 1.95621073e-01 -6.67153537e-01 5.49671113e-01
1.09217799e+00 -6.27605081e-01 3.98954719e-01 1.05873346e+00
-3.80797178e-01 3.15218955e-01 -1.25348306e+00 7.84690678e-01
-1.26174390e-01 -1.54647660e+00 5.90975769e-02 5.98463774e-01
6.78237498e-01 -4.70596701e-01 5.59478164e-01 1.51658803e-01
7.33587682e-01 -1.27907050e+00 2.78013200e-01 1.40699044e-01
7.74158835e-01 -1.35361242e+00 7.86037624e-01 5.33016562e-01
-3.72337073e-01 -5.24762161e-02 -5.09537995e-01 4.22291130e-01
-2.12746769e-01 4.20694321e-01 -7.18430340e-01 5.81839740e-01
5.21003723e-01 -2.15533420e-01 -2.17539668e-01 5.44614851e-01
-4.39725101e-01 7.11273193e-01 -5.16968071e-01 3.41449708e-01
2.57112712e-01 -6.85718656e-02 4.93053645e-01 8.32882762e-01
5.21805845e-02 3.19677442e-01 1.50949404e-01 1.02365887e+00
-6.35748208e-01 -3.76456976e-02 -8.03903043e-01 -1.60730809e-01
5.74535131e-01 8.06543469e-01 3.75738554e-02 -2.58112133e-01
-6.11839779e-02 8.98341119e-01 -1.65337950e-01 1.86327979e-01
-9.07120585e-01 -2.71995932e-01 1.08122087e+00 3.21663022e-01
1.24226108e-01 8.66057500e-02 -2.53064841e-01 -1.16357529e+00
-1.07195348e-01 -1.18900216e+00 8.97364080e-01 4.55270372e-02
-1.58664203e+00 4.19128418e-01 3.26912887e-02 -1.09803200e+00
-1.99326128e-01 -3.39032233e-01 -5.53642571e-01 1.10618126e+00
-1.39927971e+00 -1.39195716e+00 2.47250721e-01 1.03809977e+00
-4.36172366e-01 -3.14098448e-01 1.26663756e+00 1.48783728e-01
-4.14920717e-01 1.43208182e+00 1.87354505e-01 5.06794810e-01
6.26771986e-01 -9.96170759e-01 6.46049321e-01 9.50029552e-01
5.24846762e-02 9.49793756e-01 5.07091641e-01 -6.43569291e-01
-1.36041892e+00 -1.20520043e+00 8.45540524e-01 -4.48098660e-01
3.35095525e-01 -6.59968555e-01 -1.05712116e+00 1.07833481e+00
-4.52717254e-03 2.86115110e-01 1.26169157e+00 -1.91736072e-01
-7.26883709e-01 -1.33785486e-01 -2.15975666e+00 3.44182521e-01
4.33980972e-01 -6.27920926e-01 -1.41143054e-01 5.46400189e-01
7.45967329e-01 -4.93724883e-01 -8.41114640e-01 2.83455789e-01
4.62809056e-01 -8.01169097e-01 9.92737353e-01 -1.03780437e+00
1.20524708e-02 -2.51185060e-01 -3.63389701e-01 -7.73510277e-01
4.85435277e-02 -9.57363605e-01 -3.71197045e-01 1.19501674e+00
5.20460725e-01 -1.24975216e+00 1.23826087e+00 1.09974134e+00
7.87353337e-01 -4.63555217e-01 -1.19438529e+00 -9.68129456e-01
4.40936148e-01 1.35142980e-02 1.10813582e+00 1.33245540e+00
-2.46258631e-01 -4.19720501e-01 -6.56048119e-01 6.74259543e-01
9.98441637e-01 1.96827114e-01 1.05973232e+00 -1.02727449e+00
-4.29048359e-01 2.41068736e-01 -4.46141690e-01 -4.45998251e-01
2.22569510e-01 -9.02262866e-01 -5.82868934e-01 -8.32458317e-01
2.74921190e-02 -6.58643484e-01 -4.42908078e-01 9.00580406e-01
4.96293344e-02 2.69957483e-01 2.27643520e-01 -2.34995615e-02
7.15625063e-02 2.35319704e-01 9.41498458e-01 -3.24318767e-01
1.99445352e-01 4.75198239e-01 -1.06587446e+00 6.87554598e-01
9.33801293e-01 -8.40067804e-01 -6.09639764e-01 1.32550672e-01
-8.08142591e-03 2.09632695e-01 4.07839686e-01 -8.86734426e-01
1.34381831e-01 -5.20368993e-01 1.90834910e-01 -4.28152561e-01
3.64841044e-01 -1.12818325e+00 6.24648273e-01 7.47782111e-01
-1.80493444e-01 -2.12692901e-01 9.73806605e-02 6.61415040e-01
-1.55399904e-01 -1.52087256e-01 1.11624563e+00 -1.76771089e-01
-6.80618882e-02 7.42597938e-01 -1.24718413e-01 8.05667192e-02
1.49094903e+00 6.56832755e-02 -7.49163389e-01 -4.96236771e-01
-6.84381485e-01 2.14839667e-01 7.93417633e-01 1.16511531e-01
5.21812856e-01 -1.14547229e+00 -7.49307990e-01 4.93392110e-01
-3.86580117e-02 -1.21558867e-01 2.10333601e-01 2.29988307e-01
-4.20857489e-01 2.23488852e-01 -2.49145642e-01 7.08910674e-02
-1.58513129e+00 9.00446951e-01 3.17827523e-01 -4.18405741e-01
-4.13502306e-01 8.06110620e-01 4.56113577e-01 -2.42025822e-01
2.63338894e-01 1.82904471e-02 3.09610397e-01 -3.36440653e-01
6.01575136e-01 2.02254370e-01 -1.14402913e-01 -3.81749988e-01
-2.90271670e-01 -2.60900289e-01 -3.61011803e-01 -6.07216172e-02
1.05544043e+00 2.47941509e-01 -1.60855249e-01 -3.47999990e-01
1.27227628e+00 4.10507441e-01 -8.79125714e-01 -1.77458778e-01
-3.80130708e-01 -1.02511859e+00 -2.78422743e-01 -7.94362545e-01
-1.30419481e+00 7.85448313e-01 4.94960636e-01 2.36954659e-01
1.03536642e+00 -3.61357003e-01 1.07412660e+00 3.69281679e-01
6.08852446e-01 -6.59750104e-01 -4.84322309e-01 1.06505647e-01
6.23836935e-01 -1.23412871e+00 7.69207925e-02 -7.35952675e-01
-7.70900249e-01 5.99678218e-01 5.71188092e-01 -3.38142030e-02
8.28996301e-01 3.92676562e-01 4.35729325e-02 2.27366120e-01
-6.59518361e-01 5.44413626e-01 -2.75186226e-02 7.04031765e-01
-4.29254591e-01 3.03739578e-01 -3.42742711e-01 1.07610273e+00
-3.42536807e-01 -2.66327977e-01 6.22958720e-01 1.05210149e+00
9.43303406e-02 -1.61950767e+00 -3.54505688e-01 1.60884023e-01
-1.22210455e+00 2.55228654e-02 -5.32520235e-01 6.37379885e-01
1.24595210e-01 7.83911169e-01 -3.16671580e-01 -3.99784356e-01
-1.94248296e-02 -3.26644592e-02 1.48153082e-01 -4.86586034e-01
-8.39811206e-01 -5.40806532e-01 1.09666832e-01 -4.78112042e-01
8.80216286e-02 -4.06741112e-01 -1.12909329e+00 -9.81964946e-01
-4.12576020e-01 3.70845318e-01 6.18961155e-01 4.55767334e-01
3.50373596e-01 -1.94315076e-01 1.05643153e+00 3.07132691e-01
-9.01208818e-01 -3.56111228e-01 -7.40731001e-01 5.19330204e-01
2.27339149e-01 -1.51954427e-01 -4.82015818e-01 -1.35509819e-01] | [5.9221014976501465, 7.092617511749268] |
cd20a9b5-e94c-4a8f-b4b4-5b84dab123ae | back-attention-knowledge-transfer-for-low | 1906.01183 | null | https://arxiv.org/abs/1906.01183v4 | https://arxiv.org/pdf/1906.01183v4.pdf | Converse Attention Knowledge Transfer for Low-Resource Named Entity Recognition | In recent years, great success has been achieved in many tasks of natural language processing (NLP), e.g., named entity recognition (NER), especially in the high-resource language, i.e., English, thanks in part to the considerable amount of labeled resources. However, most low-resource languages do not have such an abundance of labeled data as high-resource English, leading to poor performance of NER in these low-resource languages. Inspired by knowledge transfer, we propose Converse Attention Network, or CAN in short, to improve the performance of NER in low-resource languages by leveraging the knowledge learned in pretrained high-resource English models. CAN first translates low-resource languages into high-resource English using an attention based translation module. In the process of translation, CAN obtain the attention matrices that align the two languages. Furthermore, CAN use the attention matrices to align the high-resource semantic features from a pretrained high-resource English model with the low-resource semantic features. As a result, CAN obtains aligned high-resource semantic features to enrich the representations of low-resource languages. Experiments on four low-resource NER datasets show that CAN achieves consistent and significant performance improvements, which indicates the effectiveness of CAN. | ['Chunyan Miao', 'Huanhuan Chen', 'Huixiong Yi', 'Shengfei Lyu', 'Yong liu', 'Linghao Sun'] | 2019-06-04 | null | null | null | null | ['low-resource-named-entity-recognition'] | ['natural-language-processing'] | [-2.81733930e-01 -1.93071529e-01 -3.45414996e-01 -3.72715592e-01
-8.32718432e-01 -5.10278285e-01 3.63142699e-01 -2.00735524e-01
-9.97118413e-01 8.25753927e-01 6.43156528e-01 -1.03561960e-01
3.24328214e-01 -9.60397243e-01 -6.75588012e-01 -1.18909262e-01
5.64883828e-01 5.24275243e-01 -3.06006670e-01 -3.12019914e-01
-2.39481449e-01 2.54990280e-01 -1.01716554e+00 2.66702235e-01
1.18145359e+00 5.74066639e-01 5.59427559e-01 -7.77552575e-02
-7.14882433e-01 5.64757109e-01 -2.56272614e-01 -5.84149003e-01
6.44975454e-02 -4.27502245e-01 -1.00111139e+00 -3.98702621e-01
-3.10592800e-01 -2.75808270e-03 -4.09364790e-01 1.18790507e+00
4.74782407e-01 3.81654948e-01 2.71007031e-01 -8.15479040e-01
-1.42075276e+00 9.95712101e-01 -3.45435709e-01 2.33067557e-01
2.72060245e-01 -1.97302982e-01 1.04324913e+00 -1.51682353e+00
7.08206296e-01 1.24175584e+00 5.77338159e-01 5.57392657e-01
-7.96451509e-01 -8.63683820e-01 2.49856245e-02 7.16517493e-02
-1.64311326e+00 -4.57264543e-01 4.07407194e-01 3.56687675e-03
1.31165373e+00 -1.86324105e-01 1.19294018e-01 9.67248559e-01
-2.15408519e-01 6.46588206e-01 8.47390413e-01 -5.55430293e-01
-2.92570263e-01 2.45640501e-01 -8.96846280e-02 5.99900246e-01
2.48596936e-01 -2.08176136e-01 -4.38896060e-01 2.12250561e-01
8.02265584e-01 2.92745322e-01 -1.80902019e-01 4.73333567e-01
-1.50470459e+00 7.65135229e-01 7.47443676e-01 9.26192224e-01
-7.03747272e-01 -2.36278877e-01 4.57983762e-01 2.26119876e-01
4.92032081e-01 7.55956471e-01 -8.12874734e-01 6.65644854e-02
-3.25488955e-01 -5.08326113e-01 5.38867354e-01 1.37708986e+00
1.15348113e+00 1.45189881e-01 -8.01071525e-02 1.14020824e+00
1.23375468e-01 8.08258951e-01 8.22999239e-01 -5.12954712e-01
9.48055804e-01 8.46277773e-01 1.18770085e-01 -7.39208460e-01
-3.53982568e-01 -3.11860085e-01 -1.06457484e+00 -8.66902709e-01
-7.56497532e-02 -3.96623284e-01 -7.72097588e-01 2.03326106e+00
5.11079058e-02 -2.37851012e-02 6.08167827e-01 9.10617352e-01
9.12343442e-01 1.07828963e+00 5.46011090e-01 6.04816414e-02
1.55112112e+00 -1.21973217e+00 -7.10814595e-01 -5.46383858e-01
7.74689019e-01 -6.91773772e-01 1.37357950e+00 -5.15605271e-01
-7.06322610e-01 -6.24203265e-01 -5.40898144e-01 -4.09740597e-01
-7.99120426e-01 5.45262337e-01 6.63539112e-01 1.95211425e-01
-7.71267116e-01 3.89963031e-01 -5.31327307e-01 -7.44847178e-01
1.85658187e-01 1.79963395e-01 -6.80693507e-01 -5.27994812e-01
-1.83516037e+00 1.02757514e+00 7.32855678e-01 3.16098511e-01
-4.65059727e-01 -6.39445066e-01 -1.09744895e+00 2.33780012e-01
3.04983079e-01 -6.04539871e-01 8.49295139e-01 -1.04980791e+00
-1.31190312e+00 8.82709622e-01 -2.12782994e-01 -6.84482828e-02
-7.33443769e-03 -5.46719372e-01 -7.76517987e-01 -1.54383481e-01
6.37981236e-01 7.57473886e-01 7.99440444e-02 -8.43400657e-01
-6.66674197e-01 -1.95928127e-01 7.86835924e-02 4.19197500e-01
-7.36009717e-01 5.57181776e-01 -6.83471203e-01 -5.93242764e-01
-1.47056401e-01 -7.16195941e-01 -3.41856122e-01 -7.21971452e-01
-2.36746565e-01 -4.13283825e-01 1.29935279e-01 -8.24228287e-01
1.12008417e+00 -2.10423136e+00 9.90942866e-02 -1.89286292e-01
-2.49171108e-01 3.87550145e-01 -5.66402137e-01 4.72658366e-01
-5.05214110e-02 6.13232553e-01 -1.07859001e-01 -2.61636358e-02
-8.38323310e-02 3.59393269e-01 -4.33886051e-01 -8.63803551e-03
6.65837944e-01 1.12313211e+00 -1.21526849e+00 -3.91364098e-01
-3.57927233e-02 5.57906628e-01 -3.30464661e-01 4.15110141e-01
1.84059054e-01 5.29070914e-01 -8.08180213e-01 5.11694849e-01
2.96829998e-01 -2.75117159e-01 1.60423666e-01 -2.31948182e-01
-2.69033343e-01 5.59417486e-01 -5.44398129e-01 1.85399354e+00
-1.18499660e+00 1.96254909e-01 -3.43349963e-01 -7.38477647e-01
9.54750240e-01 4.67249602e-01 2.68515229e-01 -9.12529528e-01
8.80485624e-02 6.30648732e-01 -6.66967556e-02 -2.91999966e-01
7.12299824e-01 -3.63189459e-01 -4.10349458e-01 4.12625521e-01
3.81579846e-01 2.23433658e-01 1.54511943e-01 5.08737750e-02
8.34964275e-01 1.28838003e-01 5.26109815e-01 -1.87569723e-01
6.43267751e-01 4.56115976e-02 9.47896242e-01 2.71453083e-01
-4.61066253e-02 2.82730818e-01 -9.62286368e-02 -2.38167241e-01
-1.12176394e+00 -6.83487475e-01 -1.25900850e-01 1.41588295e+00
1.53790966e-01 -4.76108819e-01 -5.97368956e-01 -7.74362147e-01
-4.10172731e-01 6.82415843e-01 -4.62199867e-01 -1.36250257e-01
-7.98972845e-01 -7.85042286e-01 8.11847925e-01 8.60845804e-01
7.57706344e-01 -1.62542641e+00 2.02777926e-02 4.54521805e-01
-4.67742503e-01 -1.49855828e+00 -6.86158180e-01 1.53435066e-01
-5.83241820e-01 -5.34991980e-01 -8.16820025e-01 -1.11412585e+00
7.96769381e-01 3.12528968e-01 1.22207296e+00 -1.13923857e-02
1.52833834e-01 3.93608026e-02 -5.83128810e-01 -2.80469775e-01
-2.62930006e-01 4.31765795e-01 2.78661370e-01 -7.94351920e-02
8.68387997e-01 -1.86967865e-01 7.46267091e-04 4.22972918e-01
-8.92849028e-01 8.92592445e-02 8.97561014e-01 1.03735137e+00
8.11241210e-01 -8.42246562e-02 9.17366147e-01 -8.66267562e-01
2.20695585e-01 -8.00503135e-01 -3.62378418e-01 3.78293067e-01
-2.57642835e-01 1.71056569e-01 1.33116853e+00 -3.68182868e-01
-1.19457984e+00 -2.87222356e-01 -2.35940799e-01 -3.59981745e-01
-1.27123117e-01 1.02595389e+00 -7.02941000e-01 4.22155857e-01
3.62722695e-01 2.37463981e-01 -8.31645250e-01 -7.02305019e-01
4.28499520e-01 9.09065902e-01 4.49380875e-01 -7.68483698e-01
7.11080968e-01 1.66687313e-02 -5.16529799e-01 -7.55332351e-01
-1.33755660e+00 -4.70828146e-01 -7.20350862e-01 4.16362256e-01
1.05851066e+00 -1.25465798e+00 1.69648919e-02 2.56727815e-01
-1.31334984e+00 -1.64749235e-01 -3.08998764e-01 7.71606922e-01
-2.05816060e-01 -1.27357021e-01 -8.90645981e-01 -3.97985637e-01
-4.63939548e-01 -8.72805953e-01 9.28035915e-01 4.11951721e-01
9.60330293e-02 -1.08563495e+00 -1.34973183e-01 1.68965518e-01
6.08487844e-01 -4.16727722e-01 1.07139444e+00 -1.03606355e+00
-3.58163029e-01 -1.64482936e-01 -5.95539391e-01 4.21330690e-01
2.43978292e-01 -5.28027594e-01 -6.93898022e-01 -1.18633948e-01
-2.33695805e-01 -3.92662019e-01 5.99538326e-01 -2.00732410e-01
9.60824132e-01 -3.16250771e-01 -1.85168251e-01 5.77196062e-01
1.54367924e+00 -7.06423679e-03 4.45751816e-01 4.12712425e-01
1.07085168e+00 5.88775158e-01 7.51734316e-01 2.15374120e-02
6.66585267e-01 5.19609630e-01 -3.17720115e-01 -3.21612448e-01
-1.72370821e-01 -6.10265911e-01 6.29797220e-01 1.61100972e+00
-7.89864138e-02 -3.24515477e-02 -1.18791914e+00 7.47079432e-01
-1.61396527e+00 -6.52732015e-01 2.83732355e-01 1.89964736e+00
1.10555065e+00 -3.27234656e-01 -4.61290628e-01 -7.98832238e-01
1.06810212e+00 -3.74287218e-02 -5.68947315e-01 -2.02443957e-01
-3.36081803e-01 1.72363773e-01 4.97528225e-01 1.20474912e-01
-1.05305004e+00 1.63748443e+00 4.84518576e+00 7.64654398e-01
-1.04130387e+00 4.59850639e-01 3.71728033e-01 3.76760721e-01
-4.07838106e-01 1.63616892e-02 -1.14388359e+00 4.74710345e-01
1.19433880e+00 -7.35693634e-01 5.66024005e-01 9.85459387e-01
-1.09255627e-01 7.04841495e-01 -9.06470478e-01 7.60359645e-01
1.10800363e-01 -1.05133295e+00 4.14889544e-01 -1.00567028e-01
9.22130108e-01 5.57187200e-01 -2.83778787e-01 8.60531032e-01
5.92949092e-01 -9.73399758e-01 3.82802576e-01 2.75136590e-01
1.31247687e+00 -9.83848870e-01 1.24517512e+00 3.91020805e-01
-1.44694543e+00 9.38891247e-02 -9.98483956e-01 2.53211141e-01
1.44289196e-01 3.90372127e-01 -4.25351828e-01 8.35321367e-01
7.01808214e-01 8.95298183e-01 -2.12913632e-01 6.22908652e-01
-6.25177920e-01 5.11643589e-01 -2.62783259e-01 7.47908056e-02
5.33212543e-01 -3.31358552e-01 1.94199950e-01 1.50216246e+00
5.71029484e-01 2.27296337e-01 4.43337530e-01 8.38423014e-01
-9.81278360e-01 7.98489928e-01 -7.06946015e-01 -4.46517527e-01
6.54670417e-01 1.40870821e+00 -3.15230608e-01 -5.24407387e-01
-7.66641498e-01 1.11163485e+00 7.55382001e-01 3.96939158e-01
-6.40467227e-01 -6.97872937e-01 5.91989994e-01 -3.33614498e-01
1.43268481e-01 -9.14389417e-02 1.56176565e-02 -1.69406402e+00
-7.48191923e-02 -7.96014607e-01 4.41549361e-01 -7.64536083e-01
-1.87399113e+00 1.13917804e+00 -5.26399255e-01 -1.07106781e+00
-6.63308650e-02 -4.97672647e-01 -2.12494925e-01 1.38558412e+00
-1.78110707e+00 -1.50984132e+00 5.22248931e-02 6.62975967e-01
6.53716683e-01 -2.90977299e-01 1.24814999e+00 5.87823868e-01
-7.90792346e-01 5.99145949e-01 1.24309987e-01 8.52023661e-01
9.98963952e-01 -9.50259745e-01 6.01265550e-01 1.00499105e+00
3.38549405e-01 1.06033349e+00 1.20228995e-03 -6.01480365e-01
-1.38140583e+00 -1.57655442e+00 1.58737421e+00 -3.16095531e-01
9.06025887e-01 -2.66667873e-01 -1.08642650e+00 9.70518470e-01
2.32405171e-01 1.88124523e-01 1.01458740e+00 2.74653703e-01
-5.68900883e-01 1.24775127e-01 -8.08976471e-01 7.06363082e-01
1.13397026e+00 -9.68614995e-01 -9.95124519e-01 2.97359765e-01
1.07879496e+00 -1.22332826e-01 -9.89014030e-01 2.79884607e-01
-7.36722052e-02 7.04753166e-03 7.04413176e-01 -1.11780274e+00
5.70689201e-01 -2.94338495e-01 -3.79348338e-01 -1.52026510e+00
-5.03343523e-01 -1.54289842e-01 4.26727027e-01 1.69411814e+00
6.46076739e-01 -7.43253350e-01 1.40583307e-01 6.05962098e-01
-3.33008885e-01 -3.81303251e-01 -8.01725864e-01 -8.81968558e-01
2.90018260e-01 -2.77629286e-01 9.07117665e-01 1.45588708e+00
-1.12060487e-01 8.75401080e-01 -3.23018700e-01 1.78490728e-01
7.15562254e-02 3.17672670e-01 4.17842329e-01 -9.59841669e-01
4.47278395e-02 -9.08959433e-02 -6.65634871e-02 -9.53258395e-01
9.32431102e-01 -1.49648559e+00 1.80920258e-01 -1.51285434e+00
5.01588106e-01 -8.16681385e-01 -8.23945880e-01 1.01280999e+00
-6.09172463e-01 2.79082477e-01 2.14991361e-01 3.91242832e-01
-7.40182877e-01 7.22888470e-01 9.69601154e-01 1.22006811e-01
-1.48882627e-01 -6.24477088e-01 -1.05855441e+00 7.67296195e-01
6.26718402e-01 -7.35691369e-01 2.27297232e-01 -1.01832485e+00
4.01942044e-01 -1.64360762e-01 -2.24555850e-01 -6.47526801e-01
1.48100272e-01 -2.50407785e-01 4.54250723e-01 -1.06566571e-01
-1.04100816e-01 -8.08877468e-01 -2.36488789e-01 -1.17627591e-01
-3.63054365e-01 3.72087687e-01 2.00822473e-01 2.67301112e-01
-6.24494612e-01 -2.74577200e-01 6.54633462e-01 -3.54758739e-01
-1.15044010e+00 5.77156126e-01 -3.89519483e-02 4.64285374e-01
4.84169006e-01 3.40185434e-01 -3.28165233e-01 2.25605373e-03
-4.35268998e-01 3.31857085e-01 5.48291922e-01 7.83028483e-01
2.72250623e-01 -1.62747121e+00 -9.15642679e-01 2.25272939e-01
3.47563803e-01 -6.65925667e-02 2.47059286e-01 5.65265059e-01
-1.78461879e-01 5.21201909e-01 -4.30080682e-01 -6.97894068e-03
-5.79124749e-01 7.03735054e-01 1.24384567e-01 -5.48857749e-01
-4.43909883e-01 5.56726336e-01 2.89257795e-01 -9.19468105e-01
-4.60540384e-01 1.64782435e-01 -3.68068725e-01 -1.83249246e-02
6.22051120e-01 1.97962616e-02 1.20975152e-02 -1.14445937e+00
-6.18615746e-01 6.18261993e-01 -9.23092216e-02 -3.71634290e-02
1.57208443e+00 -3.10897142e-01 -4.26567376e-01 3.85260850e-01
1.28447378e+00 2.17012972e-01 -5.60911179e-01 -6.23363137e-01
3.15255374e-01 -2.95109957e-01 -7.89340213e-02 -7.16274023e-01
-1.02237022e+00 1.06802499e+00 4.12841626e-02 -4.93624568e-01
1.07416892e+00 1.02470234e-01 1.02740252e+00 8.33566070e-01
7.07823873e-01 -1.19019401e+00 -2.90204942e-01 1.22317183e+00
7.00558603e-01 -1.16577303e+00 -5.29834628e-01 -2.84521371e-01
-8.97591114e-01 7.70869553e-01 7.26579487e-01 -5.25976345e-02
4.64234114e-01 6.92666695e-02 2.57312715e-01 2.08766103e-01
-5.40295482e-01 -4.54246730e-01 2.71055788e-01 4.63248283e-01
7.77842999e-01 1.58569425e-01 -2.28258654e-01 1.23783886e+00
-1.37940034e-01 -2.44150423e-02 3.15087736e-01 5.18234193e-01
-2.33681872e-01 -1.21177697e+00 -4.42510657e-02 2.87706226e-01
-7.41331756e-01 -7.17805266e-01 -1.58843175e-01 6.39449835e-01
1.50523409e-01 8.57949853e-01 2.81042233e-02 -1.46879837e-01
5.25979042e-01 2.54483312e-01 -1.45030692e-01 -1.07936597e+00
-7.55237758e-01 -1.82075977e-01 1.06801227e-01 -6.07671797e-01
-3.19140464e-01 -2.70716429e-01 -1.49430585e+00 -1.40060365e-01
-3.42171162e-01 5.11771679e-01 5.47970116e-01 9.36775506e-01
6.25863731e-01 3.40222269e-01 6.47136927e-01 -4.10605192e-01
-3.87073129e-01 -1.03423512e+00 -5.43244362e-01 6.87191606e-01
-3.54241729e-01 -4.07630295e-01 -1.26420930e-01 -2.21449748e-01] | [9.909741401672363, 9.686399459838867] |
3962fb86-8e9d-428d-bd2f-2274c7f690ff | thompson-sampling-for-combinatorial-pure-1 | 2206.0915 | null | https://arxiv.org/abs/2206.09150v1 | https://arxiv.org/pdf/2206.09150v1.pdf | Thompson Sampling for (Combinatorial) Pure Exploration | Existing methods of combinatorial pure exploration mainly focus on the UCB approach. To make the algorithm efficient, they usually use the sum of upper confidence bounds within arm set $S$ to represent the upper confidence bound of $S$, which can be much larger than the tight upper confidence bound of $S$ and leads to a much higher complexity than necessary, since the empirical means of different arms in $S$ are independent. To deal with this challenge, we explore the idea of Thompson Sampling (TS) that uses independent random samples instead of the upper confidence bounds, and design the first TS-based algorithm TS-Explore for (combinatorial) pure exploration. In TS-Explore, the sum of independent random samples within arm set $S$ will not exceed the tight upper confidence bound of $S$ with high probability. Hence it solves the above challenge, and achieves a lower complexity upper bound than existing efficient UCB-based algorithms in general combinatorial pure exploration. As for pure exploration of classic multi-armed bandit, we show that TS-Explore achieves an asymptotically optimal complexity upper bound. | ['Jun Zhu', 'Siwei Wang'] | 2022-06-18 | thompson-sampling-for-combinatorial-pure | https://openreview.net/forum?id=7N-6ZLyFUXz | https://openreview.net/pdf?id=7N-6ZLyFUXz | null | ['thompson-sampling'] | ['methodology'] | [-1.25140920e-01 4.08828259e-01 -8.57264996e-01 -7.18555599e-02
-1.59998763e+00 -9.49747562e-01 5.58950100e-03 1.23092411e-02
-3.68094295e-01 1.20822656e+00 -2.26009905e-01 -1.00487888e+00
-6.93916857e-01 -1.01142764e+00 -1.02833033e+00 -8.00299346e-01
-2.54430294e-01 1.00956535e+00 1.84820026e-01 9.93846431e-02
2.62397557e-01 1.88963488e-01 -1.08754909e+00 -2.90994365e-02
7.02689826e-01 1.48702836e+00 -3.45893353e-02 4.84036893e-01
-4.57989126e-01 1.83846729e-04 -6.17278159e-01 -1.94769189e-01
5.66405356e-01 -6.43760502e-01 -9.06724870e-01 -1.76391914e-01
-2.57536799e-01 -3.76572222e-01 8.72499868e-02 1.31280160e+00
-4.73513454e-03 1.56695247e-01 4.63912755e-01 -1.06115389e+00
-1.14282608e-01 1.64004314e+00 -1.31656194e+00 1.09485611e-01
3.15705538e-01 -3.75469655e-01 1.14367533e+00 -1.29512265e-01
3.38133484e-01 1.28506720e+00 4.38962251e-01 1.31237790e-01
-1.32675147e+00 -8.39489102e-01 6.81212425e-01 -8.17752853e-02
-1.27027440e+00 -1.49832904e-01 4.46621448e-01 2.23352969e-01
5.13934851e-01 8.04267049e-01 8.54530811e-01 6.43941164e-01
-4.26256269e-01 1.28063822e+00 1.34578860e+00 -7.84522295e-01
8.48434329e-01 6.67541698e-02 3.00743788e-01 2.06908554e-01
8.72511029e-01 3.83140624e-01 -2.11356461e-01 -4.79947746e-01
6.20571077e-01 7.06374720e-02 -2.87884384e-01 -5.84634483e-01
-8.91072631e-01 1.27246284e+00 3.05940121e-01 -4.29918757e-03
-1.00345753e-01 7.24152148e-01 1.81431860e-01 1.76779553e-01
2.62637157e-02 3.14322591e-01 -5.93052208e-01 -1.96641952e-01
-1.20945513e+00 3.17438513e-01 9.39347684e-01 1.34824800e+00
6.12735808e-01 -3.69266868e-01 -4.55810964e-01 3.71454269e-01
3.55809718e-01 7.25354135e-01 -8.47009569e-02 -9.39810455e-01
7.46601701e-01 1.76891848e-01 6.77482188e-01 -4.24206972e-01
-3.37623898e-03 -5.73713005e-01 -3.69888812e-01 1.15001008e-01
7.93249428e-01 -2.31792897e-01 -9.50879037e-01 1.78278828e+00
4.31774497e-01 -4.78676200e-01 -3.44871610e-01 6.75431907e-01
-1.74712956e-01 5.67850232e-01 -4.54222798e-01 -8.12805712e-01
1.25756502e+00 -1.00285828e+00 -6.90052927e-01 -4.27459270e-01
6.61466122e-01 -4.44448203e-01 8.39064538e-01 6.63067222e-01
-1.30233681e+00 3.37251276e-01 -1.26165676e+00 7.82338202e-01
-8.06589127e-02 -3.73074383e-01 1.04200220e+00 1.23112714e+00
-5.13336718e-01 2.81340331e-01 -6.47299945e-01 2.75280058e-01
4.89935517e-01 2.96795994e-01 2.07381055e-01 -4.28270429e-01
-7.61984944e-01 4.95937049e-01 6.61928236e-01 8.32642242e-02
-8.01613986e-01 -3.03230643e-01 -6.44327700e-01 2.46028394e-01
1.30028772e+00 -2.35077977e-01 1.50031984e+00 -2.92978257e-01
-1.20743322e+00 2.55006284e-01 -2.35430121e-01 -5.13171375e-01
6.64731920e-01 -6.44521490e-02 3.65629792e-01 -1.74001772e-02
1.73663571e-01 -8.70027691e-02 4.45794910e-01 -1.28802764e+00
-8.85361016e-01 -5.44586599e-01 2.11259514e-01 -1.33208513e-01
1.14692435e-01 -2.49423906e-01 -6.11458242e-01 -3.29375356e-01
6.71317935e-01 -1.03845680e+00 -7.39778399e-01 -6.53875291e-01
-6.71649992e-01 -1.05054185e-01 -1.24822557e-01 1.59784108e-01
1.67001259e+00 -1.85856152e+00 -3.26830000e-01 9.01571929e-01
-1.80935804e-02 -1.03051111e-01 -8.54080170e-03 3.72820079e-01
1.12811774e-01 2.78396815e-01 -7.46624172e-02 -1.79969557e-02
2.19027087e-01 2.30328351e-01 -3.94760728e-01 3.62365901e-01
-5.93335629e-01 9.20286715e-01 -1.06471443e+00 -2.63012737e-01
-1.87762961e-01 -6.90938652e-01 -5.12170076e-01 -1.29999653e-01
-6.56481147e-01 -2.06102550e-01 -8.82195532e-01 6.54631674e-01
1.04135644e+00 -3.10004205e-01 5.14056742e-01 4.61330682e-01
-2.77557205e-02 2.12039918e-01 -1.67949522e+00 1.31873226e+00
-2.03441232e-01 1.29364148e-01 2.76480466e-01 -1.32053781e+00
5.57797372e-01 -6.89815432e-02 2.98282683e-01 -4.72042292e-01
3.80742460e-01 5.85931122e-01 -1.44643694e-01 1.19168557e-01
2.12549448e-01 -2.92482167e-01 -3.61836940e-01 1.02621782e+00
-5.39014041e-01 -1.24224506e-01 1.79078653e-01 2.73637325e-01
1.23346460e+00 -1.98215798e-01 3.75046045e-01 -3.93099606e-01
1.83107704e-02 6.09763302e-02 6.22990549e-01 1.61847401e+00
-1.53979108e-01 2.96485752e-01 9.22408879e-01 -1.36277229e-01
-5.67277670e-01 -1.08036864e+00 -1.48737922e-01 9.35991943e-01
3.51323515e-01 -3.76032591e-01 -6.94773793e-01 -8.13858628e-01
2.69453645e-01 8.38056207e-01 -1.17993510e+00 1.94693893e-01
-1.23117641e-01 -7.37965882e-01 -4.41399924e-02 5.08807600e-01
3.27510208e-01 -5.34451604e-01 -5.83492100e-01 5.04616916e-01
-1.01491250e-01 -6.75344527e-01 -5.52646160e-01 7.41135180e-01
-7.46127605e-01 -1.26358402e+00 -8.19015086e-01 -2.09357932e-01
4.35513735e-01 5.24921536e-01 1.00817478e+00 -2.79476374e-01
-9.56809074e-02 1.22491360e-01 -6.16996348e-01 -7.01813042e-01
4.45795879e-02 4.88123260e-02 -3.79542470e-01 -4.25044596e-01
3.14850271e-01 -3.32118303e-01 -7.67993033e-01 4.81294274e-01
-6.17584646e-01 -4.98494387e-01 6.44218564e-01 9.38156188e-01
7.94945717e-01 2.65074998e-01 5.80989778e-01 -7.62951851e-01
5.61844051e-01 -4.88841146e-01 -1.09812188e+00 5.54719567e-01
-7.05830395e-01 3.45136404e-01 9.90614817e-02 -6.54968619e-01
-5.47425628e-01 -2.74299353e-01 2.40375698e-01 -3.39121997e-01
4.78082955e-01 8.63231242e-01 1.88320369e-01 2.23378286e-01
4.82171655e-01 5.49421757e-02 -1.11876957e-01 -4.68060821e-01
4.39260036e-01 4.93164748e-01 1.56391695e-01 -9.02144253e-01
3.85814250e-01 4.54292029e-01 1.00025244e-01 -5.38524948e-02
-1.28905034e+00 -3.71214598e-01 3.74417365e-01 2.02067599e-01
2.39573549e-02 -4.94583607e-01 -1.18856621e+00 -3.41440499e-01
-6.77097917e-01 -5.03822744e-01 -6.55488610e-01 6.74154103e-01
-8.86933506e-01 2.87177145e-01 -3.54861058e-02 -1.74049771e+00
-2.19631433e-01 -1.26011503e+00 6.57162547e-01 1.81011960e-01
-2.03513011e-01 -3.08259547e-01 -1.97084602e-02 7.55159184e-02
1.79490894e-01 3.13662887e-01 8.75271201e-01 -7.50287533e-01
-8.16973031e-01 -5.81417859e-01 -3.37620676e-01 -1.95719466e-01
-3.71053326e-03 -2.81520963e-01 -1.96472347e-01 -4.62395698e-01
-7.75608271e-02 -4.19109493e-01 1.10293901e+00 1.00406504e+00
1.28562570e+00 -4.46876049e-01 -8.98713052e-01 2.99379081e-01
1.29497075e+00 5.87969601e-01 6.45766020e-01 7.78242350e-01
-3.84368718e-01 2.70312399e-01 1.09854281e+00 1.12059307e+00
4.44384739e-02 5.18428743e-01 5.39683104e-01 4.53009367e-01
6.29210055e-01 -5.52270710e-02 -2.74528302e-02 -2.09563389e-01
1.62170723e-01 -3.92356515e-01 -5.78748882e-01 8.06635618e-01
-2.06243920e+00 -9.00451124e-01 1.61450747e-02 2.72853065e+00
1.00360501e+00 6.31942630e-01 4.26645786e-01 3.46636325e-01
6.85776651e-01 -5.27482256e-02 -7.64017224e-01 -7.63432741e-01
1.95159242e-01 3.14043224e-01 9.77917314e-01 5.89785278e-01
-7.22793579e-01 4.44235355e-01 7.41978979e+00 1.52063739e+00
-3.22697490e-01 3.25640403e-02 8.98117959e-01 -7.90757477e-01
-6.89089298e-01 2.56243616e-01 -1.12898636e+00 5.30453324e-01
7.02636778e-01 -2.97904372e-01 6.15370452e-01 1.19872463e+00
-1.58678561e-01 -8.21108341e-01 -1.11649382e+00 8.12391102e-01
-3.84080827e-01 -1.48454928e+00 -3.76121581e-01 3.88659865e-01
9.40727949e-01 -2.79393762e-01 2.37211362e-01 3.04622203e-01
1.06162751e+00 -9.87743914e-01 9.13763523e-01 -2.14532629e-01
9.28645074e-01 -1.11291242e+00 6.58556879e-01 6.10722780e-01
-1.08547604e+00 -4.56504256e-01 -3.51404339e-01 1.05729282e-01
4.19277400e-01 8.55868101e-01 -4.64845002e-01 5.29109120e-01
9.42252576e-01 -3.53200316e-01 2.88692713e-01 1.17096686e+00
-3.48690599e-02 3.99201989e-01 -1.04527473e+00 -5.12709796e-01
8.39690268e-01 -2.54394710e-01 4.71689224e-01 8.09107423e-01
5.55774450e-01 2.36518025e-01 1.73322469e-01 7.81116545e-01
2.34746337e-01 -1.16850577e-01 -2.33411655e-01 -7.75029734e-02
9.48035061e-01 6.79830313e-01 -8.75085711e-01 -4.16512638e-01
-1.59699321e-01 2.12594688e-01 2.12962434e-01 2.05297709e-01
-1.01281273e+00 -5.96502006e-01 3.09424043e-01 1.98398717e-02
8.68743896e-01 -1.34795234e-02 -5.68662047e-01 -5.01160681e-01
1.38891965e-01 -8.29964101e-01 7.20030785e-01 -4.02623378e-02
-8.55323613e-01 5.37951767e-01 4.29533184e-01 -9.52563763e-01
-2.24992827e-01 -5.38209736e-01 -5.14215827e-01 8.97460938e-01
-1.09709239e+00 -5.84449470e-01 2.31589720e-01 3.18685174e-01
2.04514161e-01 1.08742714e-01 6.11058056e-01 -4.40478921e-01
-6.35526597e-01 9.27504897e-01 6.91446662e-01 -3.52799654e-01
2.09376454e-01 -1.12618446e+00 -7.13376794e-03 5.95678926e-01
-1.84965894e-01 7.93903351e-01 8.01199079e-01 -5.90014517e-01
-1.32570386e+00 -5.46171427e-01 3.29324938e-02 -1.65525839e-01
6.50424540e-01 -1.10374406e-01 -2.26550534e-01 7.82892823e-01
-1.18588045e-01 -1.14673987e-01 8.90041292e-01 6.57501638e-01
-5.04403949e-01 -9.71971080e-02 -1.22607052e+00 6.75448239e-01
9.23070312e-01 1.66611791e-01 -3.49232584e-01 2.69669592e-01
6.35392904e-01 -3.58809084e-01 -5.88823438e-01 5.17832816e-01
8.28575552e-01 -1.07198322e+00 8.95393491e-01 -5.76064706e-01
-1.02193677e-03 5.97279705e-02 -5.11343598e-01 -1.03281343e+00
-1.03794843e-01 -1.13219333e+00 -4.95053858e-01 8.45947742e-01
8.70346069e-01 -7.30660796e-01 1.17612171e+00 8.01232278e-01
2.19629943e-01 -1.21532559e+00 -1.31327736e+00 -1.21644104e+00
2.49199614e-01 -5.86985946e-01 1.01620090e+00 3.94206285e-01
4.89663035e-01 -2.49305606e-01 -1.62509501e-01 -1.47871271e-01
9.89030898e-01 7.98577487e-01 6.06134593e-01 -8.04065764e-01
-8.55137169e-01 -7.00966835e-01 4.84939367e-01 -1.48339009e+00
-4.30811018e-01 -3.12955350e-01 2.85331547e-01 -1.34768033e+00
5.17016828e-01 -1.16288960e+00 -4.34557021e-01 2.34972864e-01
-1.17170990e-01 4.38923854e-03 8.57600477e-03 -1.72654688e-01
-8.34079206e-01 3.46949399e-01 1.37678432e+00 -2.36033559e-01
-4.48752880e-01 2.79376686e-01 -1.19406927e+00 6.47157848e-01
5.87088704e-01 -5.89901805e-01 -3.56107235e-01 -1.12007642e-02
6.01704419e-01 6.27449751e-01 -4.05961186e-01 -3.68362844e-01
6.98595308e-03 -6.20117605e-01 1.96095437e-01 -1.26265681e+00
1.17346659e-01 -7.24248588e-01 1.87645793e-01 7.09972441e-01
-3.72900218e-01 -4.38036054e-01 1.91362068e-01 9.54094648e-01
1.31873369e-01 -7.10280120e-01 5.16582191e-01 -3.31203818e-01
2.16746554e-01 2.09999010e-01 -2.92037368e-01 4.13277112e-02
1.16071582e+00 -4.24136639e-01 -3.44582379e-01 -5.74707925e-01
-7.63173163e-01 6.68686450e-01 9.60946828e-02 -3.53406638e-01
2.25108385e-01 -1.38755810e+00 -2.76374608e-01 -7.98586383e-02
1.44808695e-01 3.54660183e-01 1.17144071e-01 9.36900675e-01
-9.17216018e-02 5.90418637e-01 4.11174834e-01 -3.45433950e-01
-6.45199001e-01 1.06508398e+00 -2.70788446e-02 -7.06908703e-01
-2.02554494e-01 1.22407210e+00 1.30604789e-01 9.69328284e-02
3.60492438e-01 -4.16469574e-01 3.86196345e-01 3.96546945e-02
7.03326821e-01 4.77182448e-01 -2.87590832e-01 4.33935672e-01
-4.06732649e-01 3.30056340e-01 -3.62140089e-01 -5.98200560e-01
1.22897959e+00 -1.90683439e-01 3.37361265e-03 1.02272071e-01
8.77164543e-01 2.52282739e-01 -9.71259236e-01 -3.82768661e-01
8.10064599e-02 -8.84524107e-01 -1.38547877e-02 -9.18187499e-01
-9.09845889e-01 4.10084158e-01 1.30770952e-01 7.76520669e-01
8.98791492e-01 4.24956053e-01 6.22466028e-01 4.38156277e-01
9.90127981e-01 -1.26247978e+00 -2.57460356e-01 2.59042978e-01
8.81353438e-01 -1.10845101e+00 3.05743247e-01 -5.27676404e-01
-3.11276376e-01 7.68900216e-01 1.55300692e-01 -9.87635851e-02
5.51932514e-01 4.55053985e-01 -5.60759068e-01 5.36296852e-02
-5.63197196e-01 -3.96296829e-01 -4.95244376e-02 2.12145403e-01
-5.02643920e-03 3.51429969e-01 -5.67337513e-01 1.11088705e+00
-2.33255163e-01 -1.35296300e-01 2.05664784e-01 1.10396254e+00
-7.41196096e-01 -1.27890003e+00 -8.32412004e-01 7.77764916e-01
-5.87990761e-01 6.51060045e-02 7.69562572e-02 7.13131726e-01
-5.45321226e-01 1.26201725e+00 -5.31718414e-03 -1.23880982e-01
7.35464990e-02 -3.59746337e-01 8.76813114e-01 -2.91981637e-01
1.42302319e-01 6.42809987e-01 2.97886729e-01 -7.01343536e-01
-3.60343754e-02 -6.52257442e-01 -8.54336381e-01 -6.13513052e-01
-1.00752676e+00 7.85456419e-01 5.19788086e-01 8.58419895e-01
-9.54972729e-02 1.63628146e-01 1.01309407e+00 -6.56208634e-01
-1.27884185e+00 -8.84282589e-01 -1.13612509e+00 -3.07876229e-01
2.04091430e-01 -8.48229170e-01 -5.85495949e-01 -9.13963139e-01] | [4.543247699737549, 3.3196988105773926] |
9f17bd19-a88b-49d9-ae31-b89f1e56f013 | artificial-pupil-dilation-for-data-1 | 2212.12733 | null | https://arxiv.org/abs/2212.12733v1 | https://arxiv.org/pdf/2212.12733v1.pdf | Artificial Pupil Dilation for Data Augmentation in Iris Semantic Segmentation | Biometrics is the science of identifying an individual based on their intrinsic anatomical or behavioural characteristics, such as fingerprints, face, iris, gait, and voice. Iris recognition is one of the most successful methods because it exploits the rich texture of the human iris, which is unique even for twins and does not degrade with age. Modern approaches to iris recognition utilize deep learning to segment the valid portion of the iris from the rest of the eye, so it can then be encoded, stored and compared. This paper aims to improve the accuracy of iris semantic segmentation systems by introducing a novel data augmentation technique. Our method can transform an iris image with a certain dilation level into any desired dilation level, thus augmenting the variability and number of training examples from a small dataset. The proposed method is fast and does not require training. The results indicate that our data augmentation method can improve segmentation accuracy up to 15% for images with high pupil dilation, which creates a more reliable iris recognition pipeline, even under extreme dilation. | ['Andres Valenzuela', 'David A. Benalcazar', 'Daniel P. Benalcazar'] | 2022-12-24 | artificial-pupil-dilation-for-data | https://ieeexplore.ieee.org/document/9935749 | https://github.com/dpbenalcazar/ArtificialDilation/blob/main/benalcazar2022dilation.pdf | 2022-ieee-sixth-ecuador-technical-chapters | ['iris-recognition', 'pupil-dilation', 'iris-segmentation'] | ['computer-vision', 'computer-vision', 'medical'] | [ 3.23806763e-01 1.95722971e-02 -5.21526039e-01 -4.72583741e-01
-8.92693922e-02 -4.70523387e-01 2.50022173e-01 3.97598632e-02
-2.45477483e-01 2.40139440e-01 2.33745333e-02 -2.45676219e-01
-1.86425466e-02 -5.20020545e-01 -2.97995448e-01 -6.02251351e-01
2.03649297e-01 4.27728087e-01 -1.89108163e-01 2.60422707e-01
3.86879385e-01 8.21627200e-01 -1.97363067e+00 -3.31408270e-02
1.12227356e+00 8.97441447e-01 -5.00048876e-01 7.23761797e-01
-1.55079782e-01 1.56080887e-01 -6.74279213e-01 -3.69673699e-01
4.59921300e-01 -5.58358669e-01 -8.64145458e-01 2.43053481e-01
1.03012466e+00 -5.56765139e-01 -6.94671785e-03 1.01803648e+00
6.59609497e-01 -1.46958336e-01 2.60225445e-01 -5.10324001e-01
-5.06433487e-01 2.12872952e-01 -8.46010029e-01 -5.32910787e-02
1.97037354e-01 5.12159824e-01 3.55652690e-01 -1.31501988e-01
4.27518696e-01 9.88948226e-01 6.71952426e-01 8.07330072e-01
-1.17387021e+00 -5.64593971e-01 -5.70223629e-01 -2.82582551e-01
-1.29304099e+00 -5.06338418e-01 3.74336749e-01 -5.81551850e-01
6.11887395e-01 5.15512884e-01 8.86134505e-01 5.37517130e-01
-2.84636050e-01 5.70533693e-01 1.52041459e+00 -5.32165289e-01
-2.77798802e-01 -3.80999334e-02 3.77037749e-02 7.14669704e-01
2.38448337e-01 3.75075221e-01 -1.99877664e-01 1.79923877e-01
9.72519696e-01 -8.33403915e-02 7.77157471e-02 -4.75775823e-02
-1.14184463e+00 2.69366682e-01 3.33407491e-01 2.61971861e-01
-1.42471507e-01 -2.69830048e-01 1.73705101e-01 7.71364719e-02
3.12909722e-01 8.48177314e-01 -3.00096601e-01 -4.17204291e-01
-1.20483637e+00 -2.21344277e-01 4.74690706e-01 3.94699365e-01
6.42247796e-01 -2.20354617e-01 -3.38974863e-01 9.10326838e-01
2.10822105e-01 6.17425203e-01 4.76324439e-01 -7.42249906e-01
-1.55571103e-01 1.19911575e+00 -1.42169580e-01 -5.73419094e-01
-5.73690057e-01 -5.44692874e-01 -6.46148503e-01 3.95213306e-01
8.62822413e-01 -3.05173039e-01 -1.50806975e+00 1.25142837e+00
5.75220525e-01 2.93902040e-01 -3.06022763e-01 9.33744848e-01
9.17698622e-01 -8.56480747e-02 -1.68648258e-01 1.27063066e-01
1.25129128e+00 -6.84380829e-01 -4.26733971e-01 -1.17107928e-02
4.28556293e-01 -1.06419349e+00 9.07418072e-01 3.58307064e-01
-8.18200231e-01 -6.96317017e-01 -8.49390566e-01 -1.60830840e-01
-2.93054193e-01 5.00348389e-01 8.66796017e-01 1.34791923e+00
-8.69785309e-01 5.20324051e-01 -7.77214825e-01 -4.38285947e-01
6.48791134e-01 9.63412106e-01 -3.98660719e-01 2.09041268e-01
-4.83248889e-01 5.88447511e-01 2.77954161e-01 3.80657837e-02
-1.97916385e-02 -6.72488868e-01 -8.57430995e-01 -2.51486182e-01
-1.34515524e-01 -5.49170136e-01 8.69914949e-01 -1.06050754e+00
-1.76942372e+00 1.33516395e+00 -4.88868415e-01 -2.80536413e-01
1.95122033e-01 -5.94562329e-02 -3.82853657e-01 1.64515674e-01
-3.11859548e-01 7.33687222e-01 1.03859568e+00 -7.43400156e-01
-5.93226910e-01 -6.97180927e-01 -2.82460898e-01 -1.00795127e-01
-2.24463776e-01 3.19060594e-01 -6.13694727e-01 -3.96271884e-01
1.38813943e-01 -1.01930618e+00 8.49683881e-02 -1.85252309e-01
-3.93658489e-01 -2.55400464e-02 5.08318424e-01 -9.64543939e-01
1.01192868e+00 -2.19258666e+00 -2.21861109e-01 5.84784269e-01
4.11773533e-01 9.64833677e-01 -8.87536854e-02 -2.74725080e-01
-1.90143377e-01 3.15367490e-01 -8.43117163e-02 -2.77537525e-01
-4.94946331e-01 1.87533554e-02 1.93862990e-01 6.15141630e-01
1.76059723e-01 9.67208922e-01 -6.58846557e-01 -3.82494569e-01
5.22370934e-01 5.30773103e-01 -2.90495962e-01 4.54078913e-02
3.34281996e-02 9.90990162e-01 1.94179243e-03 9.95546877e-01
7.87651658e-01 -1.46621704e-01 -1.14278160e-01 -2.14423798e-02
-9.19956565e-02 9.72109735e-02 -1.08795321e+00 1.61535072e+00
-1.49672076e-01 7.45105624e-01 -9.96100456e-02 -5.30573428e-01
1.15221286e+00 1.78622186e-01 6.04986668e-01 -7.67448545e-01
2.50863165e-01 2.81417221e-01 4.85164523e-01 -6.05544865e-01
2.60271966e-01 1.05272502e-01 6.23342395e-01 5.52607358e-01
-2.34496355e-01 7.99975619e-02 2.50630289e-01 -5.09979665e-01
3.68053764e-01 -3.78182158e-03 -6.94315210e-02 1.82996038e-02
4.81275409e-01 -2.68153071e-01 4.80200708e-01 5.15516281e-01
-2.16402084e-01 7.23253071e-01 4.13671464e-01 -7.91107655e-01
-1.10360670e+00 -5.62200606e-01 -6.90458834e-01 3.85386050e-01
1.41263440e-01 8.25566798e-02 -1.08312917e+00 -5.43223083e-01
1.87348589e-01 -5.20059094e-02 -6.26783371e-01 1.60250977e-01
-2.37503171e-01 -8.87332261e-01 6.28331423e-01 1.77124634e-01
5.60606539e-01 -7.62012541e-01 -3.80658001e-01 -3.01759839e-01
1.92575544e-01 -8.94916117e-01 -4.32306916e-01 -6.54889166e-01
-1.05622005e+00 -1.38584661e+00 -7.86509037e-01 -6.81307197e-01
1.16685343e+00 -2.22235501e-01 6.69644654e-01 4.07137960e-01
-8.02484095e-01 5.12180803e-03 3.53071392e-02 -5.10075927e-01
-3.22785109e-01 4.05939817e-02 2.20878616e-01 4.94560748e-01
8.37577879e-01 -1.23543106e-01 -8.24261487e-01 3.39534014e-01
-5.53895473e-01 3.23230922e-02 5.94543397e-01 8.08221936e-01
6.82906866e-01 3.92976329e-02 -2.55425628e-02 -8.29341590e-01
3.02258253e-01 2.29176655e-01 -7.21680760e-01 2.20184132e-01
-8.37444723e-01 -2.71594394e-02 1.34406999e-01 -4.96901542e-01
-8.17232788e-01 2.77176648e-01 4.69277799e-02 -1.58109307e-01
-5.29778779e-01 2.03038529e-01 2.49902189e-01 -7.00094879e-01
7.73857892e-01 6.63292184e-02 6.91344738e-01 -7.14168966e-01
2.37250715e-01 1.05877900e+00 7.70460129e-01 -4.69821990e-01
4.46701199e-01 5.29832661e-01 2.07009912e-01 -9.30731297e-01
-4.69713628e-01 -7.23709166e-01 -9.53763962e-01 -1.44428283e-01
4.99942720e-01 -3.78554970e-01 -1.11712849e+00 9.69541788e-01
-6.45051062e-01 -1.24293037e-01 -4.45982218e-01 8.02539051e-01
5.58154378e-03 4.75632280e-01 -5.29112935e-01 -7.35038340e-01
-5.59224546e-01 -1.34375513e+00 8.95831823e-01 1.01518452e+00
-2.71155208e-01 -7.96550989e-01 -2.68059224e-03 8.79194260e-01
3.75827461e-01 3.73190045e-01 6.45672321e-01 -5.41254878e-01
-3.92929614e-01 -5.51868141e-01 -4.29616064e-01 4.36698198e-01
5.34747779e-01 5.17623842e-01 -1.10896504e+00 -4.07164574e-01
-2.83660263e-01 1.12397848e-02 7.11633205e-01 5.82051277e-01
1.29677725e+00 -1.06665999e-01 -1.72694385e-01 1.14773679e+00
1.14699411e+00 2.39670068e-01 9.20819640e-01 2.24623963e-01
6.30497336e-01 7.18030035e-01 2.03371868e-01 2.82611866e-02
7.22794607e-02 5.07775128e-01 2.41308331e-01 -7.63812006e-01
-6.20369375e-01 9.26016178e-03 -3.41058314e-01 1.91151246e-01
-5.28283060e-01 4.37279880e-01 -1.21292198e+00 7.42946029e-01
-1.18236947e+00 -5.76251030e-01 -1.85856447e-01 2.74747348e+00
1.07431376e+00 -2.56329149e-01 4.53233480e-01 9.01288837e-02
8.34448814e-01 -4.45956469e-01 -7.52760708e-01 -5.87320805e-01
-1.31071419e-01 9.47704971e-01 6.60674274e-01 5.19905031e-01
-1.13175285e+00 9.20095265e-01 7.08926678e+00 3.83145928e-01
-1.61759233e+00 -4.61660355e-01 8.30345213e-01 -1.85441941e-01
1.45512447e-01 -1.88431159e-01 -7.37151623e-01 5.20927548e-01
9.64585185e-01 3.31363887e-01 5.37680566e-01 2.97674030e-01
1.46198511e-01 -2.42207140e-01 -8.84524226e-01 1.13419521e+00
1.82502586e-02 -1.31301999e+00 -1.95342854e-01 3.27375114e-01
7.70453930e-01 -1.76991090e-01 5.36662996e-01 -4.04222965e-01
-2.22836673e-01 -1.51982319e+00 -2.78961986e-01 7.55950630e-01
1.36961997e+00 -6.59529865e-01 8.38612676e-01 -1.39876977e-01
-4.92375314e-01 7.51389489e-02 1.07723093e-02 3.42795849e-02
-3.97613674e-01 4.15468276e-01 -1.14169431e+00 2.34217137e-01
5.22328556e-01 7.01153636e-01 -1.02645135e+00 1.60459816e+00
-3.12858000e-02 6.91984236e-01 -4.92712438e-01 2.53992677e-01
-2.90247500e-01 -3.59919995e-01 4.11153913e-01 8.61183822e-01
1.33523762e-01 -1.47771195e-01 -2.41645664e-01 8.22871506e-01
-5.47873601e-02 2.97473341e-01 -2.28925556e-01 -3.60585272e-01
1.67455688e-01 1.05248749e+00 -6.45382226e-01 -1.31673977e-01
-3.90676737e-01 6.46376789e-01 -2.91907042e-01 1.94478676e-01
-1.43896222e-01 -5.70262015e-01 8.19130838e-01 2.14527041e-01
-1.50596499e-01 3.40600051e-02 -9.15508211e-01 -8.42055619e-01
1.18315585e-01 -1.12364519e+00 6.22268878e-02 -2.40391806e-01
-8.60788465e-01 2.98492700e-01 -7.14202106e-01 -1.09534812e+00
-1.01510033e-01 -7.18425870e-01 -3.66401494e-01 1.46587431e+00
-1.43376172e+00 -1.56164348e+00 -3.44586998e-01 3.51303816e-01
-6.40712604e-02 -5.11900842e-01 9.07865226e-01 4.00790364e-01
-9.35111880e-01 1.02469492e+00 -8.64199102e-02 6.17020845e-01
8.51026773e-01 -1.31690347e+00 5.66030502e-01 1.01259720e+00
1.21730782e-01 9.48569596e-01 1.60642654e-01 -5.77304304e-01
-1.12823033e+00 -5.64089179e-01 7.99824595e-01 -4.83514041e-01
1.05434060e-01 1.54041499e-01 -7.75194764e-01 2.99013823e-01
-1.45540714e-01 1.03362963e-01 9.87223446e-01 4.94049877e-01
-3.77262861e-01 -3.24850410e-01 -1.39133561e+00 4.36491847e-01
4.76548761e-01 -6.19105041e-01 -3.61864537e-01 2.46088922e-01
1.14785686e-01 -1.09601700e+00 -1.19944882e+00 5.99309504e-01
9.43961322e-01 -1.04031384e+00 8.93764496e-01 -6.90462053e-01
2.22365797e-01 -3.96453172e-01 6.60251498e-01 -7.36183286e-01
6.98372200e-02 -8.78750324e-01 -1.16597474e-01 1.19349098e+00
4.34841096e-01 -7.61391580e-01 1.14247644e+00 1.06620252e+00
2.69636005e-01 -6.59647942e-01 -7.29379773e-01 -5.34882367e-01
-1.25044048e-01 -2.32414845e-02 1.10476589e+00 9.96974647e-01
-1.37468770e-01 -3.86348814e-01 -2.30472356e-01 2.41241157e-01
6.72232926e-01 3.41284484e-01 9.70968962e-01 -1.43324327e+00
-1.03826359e-01 -7.03749776e-01 -8.93800676e-01 -7.85839319e-01
-2.23417819e-01 -7.79073060e-01 -5.02105653e-01 -1.13103211e+00
3.61044593e-02 -6.05190516e-01 -8.93830806e-02 7.85343766e-01
-2.26718500e-01 6.53266132e-01 -2.89798856e-01 1.98598146e-01
2.08361730e-01 -3.59078407e-01 1.61313272e+00 -2.18636185e-01
-5.97169816e-01 4.04212922e-01 -6.91551447e-01 5.13436794e-01
7.58694530e-01 1.07664630e-01 -7.98265636e-02 -3.75355899e-01
-1.27615288e-01 -3.25822264e-01 1.80194870e-01 -8.45616102e-01
2.90784568e-01 6.31269589e-02 7.48354614e-01 -1.63255185e-01
-7.81297460e-02 -5.54796159e-01 1.40328454e-02 4.89464611e-01
-1.30183473e-01 -5.69951057e-01 5.48552573e-01 -3.63502763e-02
-2.14651883e-01 3.80541733e-03 1.05885494e+00 1.47302628e-01
-5.49069464e-01 4.53726500e-01 2.50548512e-01 -1.45628199e-01
9.01314020e-01 -7.46555746e-01 -4.02379632e-01 3.90796959e-02
-8.43159676e-01 1.00036822e-01 8.18771839e-01 4.15956229e-01
4.10520077e-01 -7.58775532e-01 -7.40088463e-01 9.93282020e-01
1.53991222e-01 7.87680820e-02 1.22557230e-01 1.06729233e+00
-9.28949714e-01 3.94616663e-01 -3.35952073e-01 -9.75405335e-01
-1.66477334e+00 3.99290025e-01 8.14815521e-01 3.42321515e-01
-5.90107322e-01 9.21405613e-01 -4.38334376e-01 -3.19847763e-01
2.55289227e-01 -3.21527749e-01 -4.72215682e-01 -1.23321958e-01
8.50097120e-01 3.47668856e-01 1.53376579e-01 -7.15432525e-01
-2.96445340e-02 1.03789222e+00 -2.30389193e-01 2.67789394e-01
9.11786258e-01 8.34308863e-02 -6.43265545e-01 -1.53804928e-01
7.83249140e-01 1.59795776e-01 -9.45442140e-01 -2.22870156e-01
-1.54828757e-01 -1.05043852e+00 2.19167456e-01 -1.08135152e+00
-1.23505175e+00 9.79397833e-01 1.28687465e+00 -2.51706392e-02
1.21451259e+00 -3.06583613e-01 8.52991641e-01 -2.76142389e-01
3.31567563e-02 -9.66486692e-01 -5.91313839e-01 1.99891794e-02
2.84232348e-01 -1.50834286e+00 -3.68238203e-02 -2.22181439e-01
-2.94209838e-01 1.24403512e+00 4.45089966e-01 4.25862491e-01
4.13713664e-01 -6.94688456e-03 4.40243214e-01 -1.71225309e-01
2.23044172e-01 -4.35912549e-01 9.51433957e-01 8.58423650e-01
6.57169759e-01 2.24487603e-01 -3.62192869e-01 -8.08571279e-02
-3.77809793e-01 1.57219470e-02 2.40812913e-01 3.45094532e-01
-7.74378181e-02 -1.56816447e+00 -3.64746004e-01 8.29724908e-01
-7.09498167e-01 -6.64545000e-02 -5.85442662e-01 3.74389917e-01
5.48653543e-01 8.50601852e-01 2.45633721e-01 -3.06461364e-01
9.05259550e-02 1.08918205e-01 6.42241359e-01 -7.12641776e-01
-7.28749156e-01 1.29578725e-01 -2.97810495e-01 -5.21262646e-01
-6.00243628e-01 -7.30596185e-01 -9.76115346e-01 -4.63488489e-01
-2.61450559e-01 -2.14565411e-01 9.57348108e-01 9.93775010e-01
5.08569539e-01 1.55181214e-01 5.56845009e-01 -2.93420911e-01
-6.63664863e-02 -7.98215210e-01 -6.34172142e-01 4.97036755e-01
7.67679751e-01 -3.04126501e-01 -1.18055373e-01 2.07700551e-01] | [3.7452142238616943, -3.6305134296417236] |
55fc8946-49e5-4ad9-a7a4-c2857d808dda | small-coupling-expansion-for-multiple | 2210.03463 | null | https://arxiv.org/abs/2210.03463v2 | https://arxiv.org/pdf/2210.03463v2.pdf | Small Coupling Expansion for Multiple Sequence Alignment | The alignment of biological sequences such as DNA, RNA, and proteins, is one of the basic tools that allow to detect evolutionary patterns, as well as functional/structural characterizations between homologous sequences in different organisms. Typically, state-of-the-art bioinformatics tools are based on profile models that assume the statistical independence of the different sites of the sequences. Over the last years, it has become increasingly clear that homologous sequences show complex patterns of long-range correlations over the primary sequence as a consequence of the natural evolution process that selects genetic variants under the constraint of preserving the functional/structural determinants of the sequence. Here, we present a new alignment algorithm based on message passing techniques that overcomes the limitations of profile models. Our method is based on a new perturbative small-coupling expansion of the free energy of the model that assumes a linear chain approximation as the $0^\mathrm{th}$-order of the expansion. We test the potentiality of the algorithm against standard competing strategies on several biological sequences. | ['Andrea Pagnani', 'Louise Budzynski'] | 2022-10-07 | null | null | null | null | ['multiple-sequence-alignment'] | ['medical'] | [ 5.07569671e-01 -3.77766877e-01 1.36184484e-01 -3.15593541e-01
-1.73682913e-01 -4.71946687e-01 4.32455271e-01 6.59805417e-01
-5.34478068e-01 9.58052397e-01 1.02719657e-01 -4.98738199e-01
-2.11717427e-01 -6.02671623e-01 -6.41010344e-01 -1.14237571e+00
-1.23051912e-01 5.32844424e-01 5.83632708e-01 -4.83178467e-01
5.68255723e-01 4.74301726e-01 -1.41110790e+00 2.13541165e-01
9.39107597e-01 4.81683165e-01 3.20048571e-01 6.47170305e-01
-2.95881748e-01 2.28433330e-02 -3.50100845e-01 -4.86533821e-01
1.63630441e-01 -8.08916986e-01 -7.33882844e-01 -3.85125071e-01
8.59067068e-02 4.90557551e-01 -1.67008527e-02 1.05722117e+00
4.36593801e-01 4.89438605e-03 6.63785338e-01 -6.33526802e-01
-2.30636492e-01 7.58848190e-02 -4.18982804e-01 2.18666434e-01
2.97241688e-01 1.82501346e-01 1.14131474e+00 -4.95335728e-01
8.30301702e-01 9.86583889e-01 5.70518434e-01 2.80875504e-01
-1.66770566e+00 -6.75007403e-02 -3.15734923e-01 3.20761681e-01
-1.23363566e+00 2.94320937e-02 4.06919539e-01 -5.90379953e-01
1.09691453e+00 4.11081642e-01 7.19875216e-01 6.90133572e-01
7.95444667e-01 3.02342866e-02 1.00675607e+00 -6.31009638e-01
3.95389438e-01 -3.03431392e-01 1.74451530e-01 5.95325768e-01
1.83086544e-01 -7.60883689e-02 -5.06847441e-01 -8.91730428e-01
1.03443332e-01 -1.15029648e-01 -4.49270904e-01 -7.15938389e-01
-9.91231263e-01 7.71894574e-01 -1.35211349e-01 5.74456513e-01
-3.53016526e-01 -2.19187140e-01 3.21631938e-01 4.21429006e-03
2.84187526e-01 3.03906173e-01 -3.61069173e-01 -2.91042000e-01
-7.91236758e-01 1.88159004e-01 8.76200855e-01 3.15063924e-01
7.80839801e-01 -5.73942363e-01 2.86842406e-01 7.11147547e-01
2.21962288e-01 1.51982248e-01 4.82519656e-01 -4.63434011e-01
-2.01744750e-01 5.63119233e-01 2.70060934e-02 -7.67196417e-01
-2.84184366e-01 -3.60700428e-01 -7.31694996e-01 7.08756670e-02
5.48185647e-01 1.83749288e-01 -4.18664128e-01 1.77297294e+00
5.81998289e-01 -7.40673542e-02 -6.31700456e-02 3.92149329e-01
4.39426042e-02 6.61394894e-01 6.31518289e-02 -6.62242830e-01
9.35464442e-01 -3.43773186e-01 -1.50530815e-01 2.03326821e-01
4.83870119e-01 -8.61388445e-01 6.36769593e-01 1.97371617e-01
-8.54869187e-01 -3.11527312e-01 -9.83493090e-01 1.87984169e-01
2.75217183e-02 -6.44187391e-01 2.21573204e-01 6.36617422e-01
-9.16346967e-01 1.10893691e+00 -7.89707363e-01 -6.20680273e-01
-4.52606827e-01 2.52794445e-01 -3.70833755e-01 1.64912701e-01
-9.42648828e-01 9.13882673e-01 4.50917304e-01 -5.02376296e-02
-2.33608529e-01 -4.94368166e-01 -2.19764695e-01 1.13046013e-01
2.67460831e-02 -4.00933653e-01 8.38855624e-01 -1.07807887e+00
-1.25932193e+00 9.63247120e-01 -3.91540736e-01 -1.44025013e-01
3.71534258e-01 1.56158954e-02 -8.15908536e-02 1.83562152e-02
-3.37937206e-01 9.37582403e-02 3.48990619e-01 -6.10745966e-01
-9.85455737e-02 -4.47789103e-01 -4.26168799e-01 -9.63926315e-02
3.41188222e-01 7.87253082e-02 5.71504831e-02 -3.92349452e-01
1.07570447e-01 -1.17729223e+00 -4.78374749e-01 -1.65325701e-01
-6.04101382e-02 1.13965005e-01 4.35918644e-02 -5.68736851e-01
9.53352988e-01 -2.19548941e+00 8.41075659e-01 6.60828650e-01
-1.15852438e-01 3.69914055e-01 1.13481686e-01 1.13728964e+00
-4.28207606e-01 -5.77956624e-02 -5.65226793e-01 4.80667084e-01
-2.92907923e-01 2.49345541e-01 -6.93628043e-02 6.17883861e-01
-1.71306431e-01 3.51787776e-01 -7.40337610e-01 -1.80062085e-01
-1.16284221e-01 5.89162409e-01 -4.23789918e-01 3.13537829e-02
-4.31115746e-01 4.65037137e-01 -3.11196655e-01 -1.28186628e-01
4.80007529e-01 -3.77105445e-01 9.89945352e-01 -1.19707212e-01
-5.33989549e-01 4.88525540e-01 -7.48307109e-01 1.39811516e+00
3.07907283e-01 2.20907539e-01 -1.59206867e-01 -1.12437761e+00
7.86845684e-01 1.53231382e-01 6.24060929e-01 -2.54630566e-01
1.16793960e-02 4.96503413e-01 7.39584744e-01 -1.19307749e-01
1.83219403e-01 -2.48929262e-01 2.72712857e-01 2.85288751e-01
-6.86155781e-02 2.37133473e-01 3.15395564e-01 4.93891910e-02
9.85268235e-01 2.86253721e-01 7.20470369e-01 -6.72577202e-01
7.25976765e-01 -1.88469850e-02 5.19224763e-01 5.58719218e-01
8.87289867e-02 4.38116491e-01 8.32602024e-01 -4.76210594e-01
-1.49112678e+00 -7.08729565e-01 -2.71169126e-01 8.16034257e-01
-8.05467218e-02 -4.88484979e-01 -9.60287690e-01 -7.15420842e-02
-1.27115205e-01 5.43132126e-01 -4.18094963e-01 -4.57653821e-01
-5.54378331e-01 -1.28995728e+00 4.42471445e-01 -1.79872066e-01
9.47593749e-02 -8.98866475e-01 -6.31489515e-01 4.30671096e-01
-8.03464204e-02 -5.84247351e-01 -2.87607312e-01 3.76619905e-01
-9.34491336e-01 -1.02589846e+00 -5.40769577e-01 -3.09348345e-01
4.22521114e-01 6.11200742e-02 7.47985542e-01 2.66711771e-01
-6.03210747e-01 -1.32008314e-01 -2.45926201e-01 1.19554764e-02
-8.49503100e-01 2.40322910e-02 1.39819473e-01 8.72253701e-02
3.42559159e-01 -8.53074133e-01 -5.59914291e-01 3.10352832e-01
-9.66066480e-01 8.58173892e-02 5.15254796e-01 8.35832477e-01
6.09010637e-01 -3.66480500e-01 5.59544526e-02 -5.69202185e-01
3.99155468e-01 -3.45645487e-01 -7.69703150e-01 6.50875330e-01
-3.95701796e-01 6.44995034e-01 5.85973561e-01 -1.90552011e-01
-7.42581844e-01 1.53526574e-01 -3.52907091e-01 4.21375692e-01
-2.62727253e-02 5.74464321e-01 -2.74397016e-01 -2.08931729e-01
5.85084081e-01 8.58090401e-01 1.04916967e-01 -6.03320837e-01
-3.15075442e-02 4.83212888e-01 2.44445473e-01 -7.81191766e-01
3.10999900e-01 3.83695036e-01 5.01261055e-01 -1.25087178e+00
-3.09951186e-01 -4.05474275e-01 -6.47585332e-01 -5.57970665e-02
7.15465665e-01 -1.54363289e-01 -9.29046392e-01 4.92324233e-01
-1.04758584e+00 5.10223545e-02 1.01767220e-01 4.47615176e-01
-6.63972139e-01 9.71096456e-01 -4.75550801e-01 -6.68772519e-01
-1.47737414e-01 -1.23215163e+00 7.22648084e-01 3.21570039e-02
-2.38559738e-01 -7.35867560e-01 8.12879562e-01 5.39236031e-02
2.58872122e-01 2.11809099e-01 1.64728642e+00 -7.04759777e-01
-4.01124865e-01 4.18596454e-02 2.86478996e-01 1.67737633e-01
1.06611699e-01 4.29146618e-01 -4.44870591e-01 -1.31784543e-01
1.65706594e-02 2.41370484e-01 7.81014323e-01 1.86543345e-01
4.93984848e-01 -3.20831165e-02 -3.57701778e-01 2.44701907e-01
1.43427575e+00 3.43458235e-01 6.48893774e-01 2.92287588e-01
1.52599797e-01 6.78536713e-01 3.29894394e-01 2.26912752e-01
-4.43870574e-01 9.79104519e-01 1.83256790e-01 3.74904513e-01
5.85483015e-01 -1.47740647e-01 3.21605533e-01 9.07311201e-01
-3.00419509e-01 -1.09042384e-01 -1.05244911e+00 1.78229377e-01
-1.92969668e+00 -1.20411444e+00 -2.52209127e-01 2.64636350e+00
8.49916160e-01 1.93240181e-01 2.12704450e-01 -2.89850794e-02
8.76298904e-01 -4.41723093e-02 -4.66478288e-01 -5.90151548e-01
-1.98621839e-01 3.70911539e-01 4.40880477e-01 6.10003650e-01
-4.63377804e-01 3.62039924e-01 6.61697054e+00 6.44745171e-01
-1.18991983e+00 -2.37093195e-01 2.71029353e-01 8.17771852e-02
-2.31446758e-01 3.67025167e-01 -5.74915826e-01 6.28578126e-01
1.12070143e+00 -2.04668030e-01 1.23060174e-01 6.14702165e-01
9.87162143e-02 -2.05795258e-01 -7.69413054e-01 4.44834709e-01
-1.66050583e-01 -1.31519330e+00 -1.89712375e-01 3.70789826e-01
3.17692906e-01 2.27937594e-01 -1.82723597e-01 -4.65905488e-01
1.57297879e-01 -7.15429783e-01 3.53713721e-01 4.18564498e-01
3.46028954e-01 -6.80764377e-01 6.68775737e-01 5.65012157e-01
-7.97036350e-01 3.42405677e-01 -4.74467307e-01 -3.86661552e-02
2.03673467e-01 6.63466692e-01 -6.80355132e-01 5.10586858e-01
2.85042286e-01 3.21227670e-01 -1.95613384e-01 1.11314929e+00
7.53641576e-02 5.06812215e-01 -4.21047539e-01 -3.55922878e-01
5.27350791e-02 -8.54777038e-01 5.39150119e-01 1.01020932e+00
3.55775446e-01 1.74666390e-01 -1.87935252e-02 2.92492628e-01
3.44565123e-01 4.89777416e-01 -2.33461067e-01 -3.09368849e-01
1.70118157e-02 8.14401448e-01 -8.26175153e-01 -1.45752374e-02
-4.94967222e-01 8.10932696e-01 3.82170439e-01 8.50773975e-02
-5.25744617e-01 -1.39830261e-01 9.21741545e-01 2.01212853e-01
4.61913884e-01 -5.36956072e-01 4.26259905e-01 -1.04799092e+00
9.10946727e-02 -1.07878184e+00 1.69903949e-01 -3.46214801e-01
-1.01267803e+00 4.86864179e-01 -1.15193725e-01 -6.65409923e-01
-3.28875631e-01 -6.26994610e-01 -1.77206278e-01 1.03365386e+00
-8.97541463e-01 -3.92708033e-01 3.20439100e-01 1.31923795e-01
3.46032344e-02 1.24417230e-01 9.51135159e-01 1.98275596e-02
-4.00370032e-01 5.38899973e-02 9.90461826e-01 -4.12065148e-01
5.85114777e-01 -8.15744340e-01 4.77441072e-01 5.28352022e-01
6.57049194e-02 8.96769106e-01 1.27010369e+00 -8.27641845e-01
-1.14155126e+00 -4.21837032e-01 1.11468220e+00 -1.24303000e-02
6.88842535e-01 -4.72645879e-01 -1.21476090e+00 3.63900959e-01
9.67413411e-02 -3.58326077e-01 1.12344778e+00 -3.23662907e-02
-3.35937798e-01 1.26320451e-01 -8.66944671e-01 5.02829254e-01
8.56017113e-01 -3.29500675e-01 -4.60825354e-01 2.37499416e-01
3.31691712e-01 1.13052711e-01 -7.30012774e-01 2.74569511e-01
8.41286361e-01 -1.32188606e+00 6.95153594e-01 -9.59512293e-01
1.19855285e-01 -3.25121492e-01 -3.45659047e-01 -1.04226971e+00
-4.77236480e-01 -7.31947958e-01 4.74678397e-01 6.68282568e-01
3.98222357e-01 -6.31279647e-01 4.43509489e-01 2.86208153e-01
1.80959761e-01 -4.60939914e-01 -1.26977003e+00 -6.22867227e-01
-3.89683247e-03 2.07517400e-01 2.32327968e-01 5.49114108e-01
3.45217139e-01 4.19904083e-01 -2.63648689e-01 -9.03142020e-02
3.85386676e-01 1.54492408e-01 4.74561602e-01 -1.28978992e+00
-7.82962084e-01 -3.17132831e-01 -7.45812893e-01 -6.91053987e-01
-2.17405017e-02 -5.06312609e-01 1.74176358e-02 -9.00672972e-01
4.88358557e-01 -2.53155213e-02 -2.50109673e-01 -4.02490161e-02
-1.02516443e-01 -1.80835426e-01 -1.84154361e-01 3.15772742e-01
-2.41619259e-01 3.99513841e-01 5.26355207e-01 2.95170844e-01
7.25151822e-02 -5.20636179e-02 -1.13009520e-01 6.28197134e-01
6.82242692e-01 -6.18215084e-01 -8.90589282e-02 8.09414908e-02
8.89347196e-01 -3.48124653e-02 6.96426630e-02 -6.89376235e-01
-1.07406914e-01 -3.57752681e-01 -1.14317827e-01 -2.32566133e-01
2.24005759e-01 -6.58556581e-01 9.02678728e-01 9.81813729e-01
-3.86802316e-01 3.71664107e-01 3.51597331e-02 5.73216140e-01
-2.23380211e-03 -6.03766799e-01 1.06582522e+00 -2.07716048e-01
-1.42906576e-01 -1.38239071e-01 -6.66100800e-01 -2.60400534e-01
8.67827415e-01 1.38527562e-03 -1.13068514e-01 -1.22058265e-01
-4.60642904e-01 -4.43916380e-01 9.58632112e-01 -4.52907896e-03
9.92572159e-02 -5.88277817e-01 -5.79908550e-01 7.70661011e-02
-9.63959023e-02 -6.77160621e-01 2.70614952e-01 9.01565671e-01
-9.37718928e-01 5.63410163e-01 -4.54985529e-01 -5.10453105e-01
-1.80931580e+00 6.88055992e-01 4.19732720e-01 -3.38669270e-01
-2.33168736e-01 3.58614057e-01 1.59112066e-01 -2.54717767e-01
-4.31348473e-01 1.89052939e-01 1.10523045e-01 -2.91382134e-01
3.98812771e-01 3.54365975e-01 2.21259743e-01 -8.99053752e-01
-5.13879657e-01 6.46129072e-01 -3.31079364e-01 -1.08700983e-01
1.26060677e+00 5.95503114e-02 -6.49817169e-01 5.58484912e-01
9.14933264e-01 1.75802737e-01 -8.71742785e-01 -8.75302479e-02
2.82829195e-01 -3.06543410e-01 -6.00405157e-01 -3.86106700e-01
-1.95619971e-01 8.02084446e-01 4.36870933e-01 1.62470847e-01
7.56166160e-01 -7.15402812e-02 4.24304247e-01 3.08647394e-01
5.37839949e-01 -7.45641291e-01 -4.51237530e-01 4.03066248e-01
3.88733804e-01 -5.91096580e-01 8.86358693e-02 -4.29741949e-01
-6.06299192e-02 1.09334898e+00 -3.59851979e-02 1.49761615e-02
2.10952476e-01 -4.34735417e-02 -2.56985605e-01 1.41727373e-01
-8.48921835e-01 5.61648561e-03 1.21075541e-01 2.01938525e-01
8.90759289e-01 -9.74318236e-02 -1.16845751e+00 -1.38873741e-01
-1.15312515e-02 -2.64796317e-01 3.33211511e-01 1.08854020e+00
-5.48933029e-01 -1.86571229e+00 -1.60226613e-01 2.00246405e-02
-7.35501826e-01 -3.25888842e-01 -8.69935870e-01 5.79763114e-01
-9.00541916e-02 5.71674407e-01 -1.94159940e-01 3.09480750e-03
2.22882144e-02 6.24686778e-01 7.77723014e-01 -2.98767239e-01
-6.27159059e-01 8.05646107e-02 -7.56619275e-02 -2.33230412e-01
-4.14293796e-01 -9.37774777e-01 -1.03569686e+00 -3.08321416e-01
-3.43130678e-01 8.44167590e-01 7.82112658e-01 1.00890648e+00
5.70339680e-01 7.33478577e-04 3.84963095e-01 -4.93362874e-01
-7.36171901e-01 -7.60871828e-01 -4.44707841e-01 4.32863414e-01
9.52374116e-02 -5.14637113e-01 -2.28814095e-01 -2.24027917e-01] | [4.839676856994629, 5.192108631134033] |
4f3b62f0-2796-420d-991a-99bdc29cf291 | sparseness-meets-deepness-3d-human-pose | 1511.09439 | null | http://arxiv.org/abs/1511.09439v2 | http://arxiv.org/pdf/1511.09439v2.pdf | Sparseness Meets Deepness: 3D Human Pose Estimation from Monocular Video | This paper addresses the challenge of 3D full-body human pose estimation from
a monocular image sequence. Here, two cases are considered: (i) the image
locations of the human joints are provided and (ii) the image locations of
joints are unknown. In the former case, a novel approach is introduced that
integrates a sparsity-driven 3D geometric prior and temporal smoothness. In the
latter case, the former case is extended by treating the image locations of the
joints as latent variables. A deep fully convolutional network is trained to
predict the uncertainty maps of the 2D joint locations. The 3D pose estimates
are realized via an Expectation-Maximization algorithm over the entire
sequence, where it is shown that the 2D joint location uncertainties can be
conveniently marginalized out during inference. Empirical evaluation on the
Human3.6M dataset shows that the proposed approaches achieve greater 3D pose
estimation accuracy over state-of-the-art baselines. Further, the proposed
approach outperforms a publicly available 2D pose estimation baseline on the
challenging PennAction dataset. | ['Xiaowei Zhou', 'Spyridon Leonardos', 'Kosta Derpanis', 'Kostas Daniilidis', 'Menglong Zhu'] | 2015-11-30 | sparseness-meets-deepness-3d-human-pose-1 | http://openaccess.thecvf.com/content_cvpr_2016/html/Zhou_Sparseness_Meets_Deepness_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Zhou_Sparseness_Meets_Deepness_CVPR_2016_paper.pdf | cvpr-2016-6 | ['monocular-3d-human-pose-estimation'] | ['computer-vision'] | [-3.11712548e-02 3.61128300e-01 -1.45368144e-01 -3.08440149e-01
-7.69348443e-01 -2.03208223e-01 6.18681431e-01 -4.70887303e-01
-6.14411891e-01 6.87417328e-01 3.21692914e-01 3.24957103e-01
1.25704437e-01 -8.15957859e-02 -1.02516270e+00 -6.82894766e-01
-4.87347357e-02 7.86929190e-01 1.17027201e-01 2.06783898e-02
5.87826669e-02 4.41496670e-01 -1.19246972e+00 -4.71212238e-01
4.90202516e-01 1.16860759e+00 1.37363106e-01 5.19196332e-01
3.51932257e-01 4.15279180e-01 -4.87785846e-01 -1.87118620e-01
5.60156465e-01 -4.51005474e-02 -5.23683906e-01 6.21360719e-01
5.77554524e-01 -7.04606771e-01 -5.16722202e-01 9.25924122e-01
6.02409124e-01 1.28221199e-01 6.37644708e-01 -1.17013609e+00
-2.92601180e-03 -1.39938369e-01 -9.17631745e-01 -3.19735348e-01
6.36668265e-01 1.42575338e-01 7.29798973e-01 -9.84248877e-01
7.87270069e-01 1.36837721e+00 7.12185025e-01 2.22658381e-01
-1.06446207e+00 -3.33200544e-01 2.24736035e-01 5.81095219e-02
-1.58606040e+00 -3.44928473e-01 8.31509352e-01 -7.22760081e-01
7.73884296e-01 -2.88422108e-01 6.76739573e-01 1.01096117e+00
4.76908088e-01 1.04746199e+00 8.57155621e-01 -4.80182022e-01
1.17827848e-01 -3.96175355e-01 -2.86961079e-01 8.55004668e-01
1.47774369e-01 1.51640579e-01 -7.00266600e-01 -8.25693607e-02
9.79124963e-01 1.32539690e-01 -2.20645651e-01 -8.29164982e-01
-1.36262405e+00 4.91061449e-01 3.40424150e-01 -2.03020751e-01
-6.27450645e-01 4.20605421e-01 2.65936047e-01 -2.73628265e-01
6.02853119e-01 1.28986120e-01 -3.86159331e-01 -1.52913585e-01
-9.82315004e-01 7.87834287e-01 4.94803846e-01 1.21881425e+00
4.39292431e-01 -9.67131853e-02 -1.80133149e-01 5.75659096e-01
5.47403812e-01 6.06229722e-01 1.25268668e-01 -1.11891770e+00
7.54541636e-01 3.38195652e-01 7.00271845e-01 -9.29612517e-01
-5.79493165e-01 -4.42745119e-01 -5.24423182e-01 1.80107951e-01
6.74138606e-01 -2.97322571e-01 -1.18239701e+00 1.74192023e+00
6.63000405e-01 -4.53372784e-02 -3.44986171e-01 1.30429900e+00
3.94426614e-01 4.25196558e-01 -2.37281695e-01 1.90272793e-01
1.33927286e+00 -9.50292230e-01 -8.29146981e-01 -5.44906020e-01
1.72177766e-04 -6.70747876e-01 4.38068539e-01 3.59172076e-01
-1.27160215e+00 -6.96875691e-01 -1.03955162e+00 -7.45959133e-02
8.00534487e-02 5.85944653e-01 3.79937112e-01 2.84430653e-01
-7.07544267e-01 4.54863548e-01 -1.23470163e+00 -2.73088872e-01
-5.90508152e-03 3.34306777e-01 -4.77416784e-01 -1.08660482e-01
-1.10807431e+00 9.81285214e-01 2.48903304e-01 5.83912432e-01
-9.24517214e-01 -3.05919111e-01 -1.12114871e+00 -2.37488016e-01
7.08131671e-01 -9.76302147e-01 1.34673262e+00 -3.88144970e-01
-1.66938770e+00 9.38197255e-01 -1.93659052e-01 -3.87383699e-01
9.69312787e-01 -9.30807948e-01 3.87978226e-01 3.81856173e-01
2.07809821e-01 8.48097086e-01 1.04075813e+00 -1.14143395e+00
-3.56053829e-01 -5.58238447e-01 5.33565879e-03 5.69420934e-01
2.96201944e-01 -4.24295217e-01 -1.00848985e+00 -7.35261738e-01
5.34621716e-01 -1.40128064e+00 -3.23675513e-01 4.50303465e-01
-7.08360910e-01 -7.55763277e-02 4.29905564e-01 -9.90885973e-01
7.64441669e-01 -1.84636962e+00 6.84592307e-01 1.71235830e-01
3.22307497e-02 -1.20136857e-01 3.21343958e-01 2.73403764e-01
1.89769074e-01 -4.50881481e-01 -1.44563273e-01 -8.61647308e-01
2.89606839e-01 2.70275861e-01 2.27982737e-02 8.50819170e-01
2.08862036e-01 9.66304541e-01 -6.23560727e-01 -5.38936615e-01
4.45560217e-01 6.11124218e-01 -5.44249475e-01 4.43085611e-01
-1.94434494e-01 7.72837460e-01 -4.27935779e-01 5.53872049e-01
5.93094826e-01 -3.20059329e-01 1.88383773e-01 -4.27432299e-01
6.09496869e-02 1.33104682e-01 -1.32499075e+00 2.31122804e+00
-8.48156437e-02 1.67287946e-01 1.11927204e-01 -6.88366354e-01
6.80105507e-01 5.03730834e-01 6.51273668e-01 -1.73922807e-01
1.49943724e-01 -1.18632363e-02 -3.02833408e-01 -3.63856673e-01
4.79365319e-01 -7.57448301e-02 -1.88685045e-01 1.12444356e-01
1.46239281e-01 -2.27750987e-01 -9.22470167e-02 -4.27541286e-02
6.99713945e-01 9.45972145e-01 4.29657102e-01 -9.45421308e-02
3.16901326e-01 -1.18997835e-01 7.44241893e-01 5.13585865e-01
-2.49890760e-01 9.58715856e-01 4.07075942e-01 -2.63755947e-01
-1.05145121e+00 -1.40372765e+00 1.53957903e-01 3.70755792e-01
1.74190998e-01 -1.68831825e-01 -6.18592024e-01 -4.61526841e-01
2.22349197e-01 1.68064743e-01 -6.21915579e-01 4.08088267e-02
-7.79882371e-01 -3.28808039e-01 2.23966807e-01 7.06214070e-01
6.42280698e-01 -4.15807307e-01 -8.93377125e-01 1.15206830e-01
-6.07248187e-01 -1.37849987e+00 -6.94558144e-01 -8.28240365e-02
-9.19638693e-01 -9.42266941e-01 -1.11502099e+00 -5.77569008e-01
8.09162319e-01 -8.23910087e-02 7.35089481e-01 -3.42734158e-01
-3.56525153e-01 4.82739091e-01 -9.41995159e-02 -1.65963590e-01
2.22758517e-01 -1.69906467e-01 3.42311203e-01 -9.02356580e-02
9.73977819e-02 -3.78320873e-01 -7.86628008e-01 4.60118800e-01
-3.77383441e-01 1.41108721e-01 6.65168285e-01 9.42655146e-01
8.54629278e-01 -1.93994537e-01 2.40794361e-01 -3.65352541e-01
-3.07057574e-02 -1.61881432e-01 -6.23485684e-01 1.84924491e-02
-2.34631762e-01 1.98740616e-01 1.99411400e-02 -5.30412674e-01
-1.40670419e+00 6.91857576e-01 -6.04884923e-02 -6.67559147e-01
-2.34042928e-01 4.10329580e-01 -3.63044858e-01 1.52535096e-01
2.15286940e-01 5.98331876e-02 2.61626005e-01 -6.88595474e-01
2.67271727e-01 2.81443417e-01 8.39873254e-01 -8.00325871e-01
8.50286424e-01 6.67217076e-01 1.89655647e-01 -6.72128558e-01
-7.94674993e-01 -5.36580682e-01 -1.11816728e+00 -2.51695067e-01
1.11333966e+00 -1.33284664e+00 -5.70468485e-01 6.54421866e-01
-1.27099383e+00 -1.13046534e-01 2.82798856e-02 8.06990206e-01
-9.86509264e-01 7.00826287e-01 -6.74271584e-01 -9.05562878e-01
-1.18632793e-01 -1.37583399e+00 1.74618340e+00 -9.25341696e-02
-5.06540358e-01 -7.18601346e-01 -1.01937957e-01 4.33628738e-01
-2.71677583e-01 5.16651094e-01 5.67352057e-01 -1.80305049e-01
-5.73325157e-01 -6.05932951e-01 2.07930841e-02 1.09293737e-01
-4.49338518e-02 -4.95011628e-01 -6.24483287e-01 -3.56128097e-01
-4.27558646e-02 -3.19484681e-01 4.74122167e-01 8.25867295e-01
7.38800943e-01 -1.75355777e-01 -3.65677088e-01 4.72900897e-01
9.82373416e-01 -8.25700834e-02 5.13244689e-01 1.93589702e-01
6.84225380e-01 6.29399896e-01 1.01063931e+00 7.37415910e-01
4.46606249e-01 1.12429452e+00 4.82931077e-01 8.65335837e-02
7.92911053e-02 -6.00114703e-01 2.70648301e-01 4.97117251e-01
-2.69731969e-01 9.03701223e-03 -6.53963566e-01 4.93568450e-01
-2.01690412e+00 -4.84081924e-01 4.01287109e-01 2.18980646e+00
6.79711521e-01 1.87580273e-01 1.69502765e-01 -1.34375304e-01
7.16713369e-01 2.45303839e-01 -7.48861432e-01 4.33257669e-01
2.10188732e-01 -1.01627268e-01 4.89454418e-01 5.72906792e-01
-1.24445009e+00 7.45505393e-01 6.00042105e+00 4.16839451e-01
-6.71688735e-01 -2.57896744e-02 1.93139836e-01 -3.28508288e-01
3.35156113e-01 -6.82000518e-02 -1.01021683e+00 2.72854000e-01
1.69337153e-01 2.91264415e-01 -1.23966709e-01 7.43414998e-01
2.64658123e-01 -3.32335621e-01 -1.15047312e+00 1.08889890e+00
1.51737303e-01 -7.74288952e-01 -3.01241875e-01 2.86408395e-01
6.69459939e-01 -2.48781174e-01 1.86023470e-02 -6.39831871e-02
1.03305705e-01 -6.62788272e-01 1.02608109e+00 9.08619523e-01
6.62565351e-01 -7.48578846e-01 6.55964136e-01 5.24343133e-01
-1.13537729e+00 1.08140364e-01 -1.89468384e-01 -2.67535746e-01
7.08664894e-01 5.87221026e-01 -7.24733055e-01 5.97069383e-01
7.07494497e-01 7.09272504e-01 -2.93029487e-01 9.69778657e-01
-6.79554701e-01 2.48297483e-01 -5.02685010e-01 3.96730393e-01
1.52782192e-02 -1.18858010e-01 7.97467589e-01 7.42999196e-01
2.31640175e-01 -1.72503795e-02 3.69780332e-01 9.28568959e-01
1.57322034e-01 -3.01297754e-01 -3.52897346e-01 2.61771560e-01
2.19209626e-01 9.56029952e-01 -4.89660323e-01 -9.00649726e-02
-2.15661824e-01 1.27244031e+00 1.53319940e-01 4.87471968e-01
-8.62668633e-01 -1.60610810e-01 6.53522789e-01 5.63155860e-02
4.65625793e-01 -6.76178396e-01 -1.20724283e-01 -1.32004380e+00
3.68508458e-01 -6.08475864e-01 1.96698949e-01 -9.44600284e-01
-1.22278237e+00 7.89630413e-02 5.31235754e-01 -1.21038091e+00
-8.89219582e-01 -6.91322088e-01 -1.95692629e-01 8.56293738e-01
-1.07587576e+00 -1.08184302e+00 -1.10872149e-01 3.73459816e-01
4.98790085e-01 1.95746601e-01 4.03593749e-01 1.18667655e-01
-2.04355463e-01 4.12638336e-01 -2.62814909e-01 2.55402356e-01
7.57738173e-01 -1.10386002e+00 5.85057855e-01 9.25637126e-01
-1.05886757e-01 7.04325914e-01 8.39451194e-01 -9.60467458e-01
-1.46381986e+00 -7.66757190e-01 8.94593179e-01 -4.69114125e-01
2.53067344e-01 -5.25981605e-01 -4.32159901e-01 8.92374396e-01
-2.58704633e-01 -3.99114341e-02 1.17413543e-01 -1.36366948e-01
-7.45319501e-02 1.97589099e-01 -1.01898921e+00 6.05567694e-01
1.13357627e+00 -2.75855303e-01 -8.27202916e-01 3.03168267e-01
5.46036422e-01 -8.63654971e-01 -9.60679591e-01 4.88498718e-01
9.38095152e-01 -6.20330215e-01 1.26868641e+00 -3.76509607e-01
3.75843078e-01 -5.30684650e-01 -1.96929157e-01 -1.06871819e+00
5.02849519e-02 -5.55020452e-01 -3.75216573e-01 6.47527754e-01
-6.38270285e-04 -2.83524632e-01 9.85329151e-01 7.49824047e-01
6.43026903e-02 -7.25907147e-01 -1.26453590e+00 -6.76975250e-01
-1.08743548e-01 -1.71548516e-01 1.58496439e-01 2.47089952e-01
-1.91066250e-01 1.26844630e-01 -9.50244427e-01 2.90566117e-01
9.12194550e-01 9.82787833e-02 1.04101717e+00 -1.02946901e+00
-5.64535797e-01 1.40777066e-01 -6.22024596e-01 -1.97022152e+00
2.83150077e-01 -2.48310715e-01 4.69342858e-01 -1.35064554e+00
1.23965867e-01 2.36573368e-01 2.37639144e-01 2.29193777e-01
-1.71085179e-01 2.91281305e-02 2.73584932e-01 1.36232451e-01
-4.77030367e-01 7.68178523e-01 1.24384439e+00 4.05732468e-02
7.85910860e-02 3.30670327e-01 1.93306245e-02 9.86260414e-01
3.37817699e-01 -2.58176446e-01 -3.17086399e-01 -3.24244291e-01
-6.04703929e-03 4.21019256e-01 7.62013197e-01 -1.02891970e+00
2.54475415e-01 -7.58047029e-02 7.42505908e-01 -1.20198441e+00
9.18603420e-01 -8.97461951e-01 2.60561407e-01 5.03501713e-01
-2.44973212e-01 -8.15210417e-02 -7.66138211e-02 8.61850917e-01
-9.78691317e-03 -1.61078181e-02 6.26282752e-01 -3.16067129e-01
-5.07370055e-01 4.18937534e-01 -2.30767459e-01 6.12783581e-02
7.31229067e-01 -3.21537435e-01 2.69149572e-01 -6.65738940e-01
-1.00205672e+00 2.46012971e-01 4.35126573e-01 4.49339449e-01
7.40532637e-01 -1.40310478e+00 -4.69853550e-01 8.58106092e-02
9.92691070e-02 4.02893782e-01 2.61660606e-01 1.03190076e+00
-2.53180206e-01 5.11531830e-01 -8.25404823e-02 -9.80353057e-01
-1.07939231e+00 3.17363799e-01 4.07382756e-01 -3.75546038e-01
-7.73840308e-01 7.27834463e-01 3.35091770e-01 -5.67069709e-01
4.24163014e-01 -1.84615836e-01 2.56910026e-01 -2.75323302e-01
2.19474390e-01 4.52262729e-01 -2.57822126e-01 -1.08326137e+00
-2.91774780e-01 8.69362772e-01 9.43571851e-02 -3.54590029e-01
1.28872728e+00 -4.69546646e-01 2.25704566e-01 4.61908638e-01
1.21475816e+00 -2.61848807e-01 -1.94776630e+00 -4.51974809e-01
-1.02743812e-01 -5.97836375e-01 -2.65806794e-01 -6.86069131e-01
-9.28270519e-01 9.11906600e-01 4.94176060e-01 -8.58745635e-01
7.59282887e-01 -2.18707360e-02 7.07556546e-01 2.88985223e-01
5.73292553e-01 -1.15192938e+00 1.35193497e-01 4.47394490e-01
1.02037096e+00 -1.10378850e+00 3.62422138e-01 -5.66333652e-01
-5.47399998e-01 1.02571321e+00 6.45437598e-01 -2.62447327e-01
6.83535457e-01 3.39877531e-02 -7.09220767e-02 -8.14586654e-02
-3.75146925e-01 -1.13205820e-01 7.59410203e-01 3.94310832e-01
3.20698917e-01 -6.36705905e-02 -3.42251092e-01 4.72145140e-01
-7.67354295e-02 4.25088815e-02 1.16928279e-01 1.15309656e+00
-2.06479892e-01 -7.97105610e-01 -6.79846466e-01 -6.93257600e-02
-4.54506904e-01 2.50851840e-01 -2.30017349e-01 9.26584423e-01
-3.28547023e-02 6.82842731e-01 -5.49770333e-02 -1.39932349e-01
4.79482412e-01 2.85028756e-01 7.64439642e-01 -4.64505166e-01
5.35954796e-02 5.87950289e-01 6.92052096e-02 -8.17323744e-01
-5.18574774e-01 -8.12053084e-01 -1.24062550e+00 2.09627599e-01
-3.07201445e-01 -1.87973857e-01 7.56956279e-01 1.19566762e+00
2.75373638e-01 2.42537379e-01 3.32755558e-02 -1.56143713e+00
-9.96930063e-01 -9.14036453e-01 -6.80584967e-01 4.35766667e-01
3.37785900e-01 -1.31910610e+00 -2.76265085e-01 1.26417086e-01] | [7.016590118408203, -0.9877634048461914] |
5f1a0621-0430-476d-bc23-26a667b67bdf | optimal-gradient-sliding-and-its-application | 2205.15136 | null | https://arxiv.org/abs/2205.15136v1 | https://arxiv.org/pdf/2205.15136v1.pdf | Optimal Gradient Sliding and its Application to Distributed Optimization Under Similarity | We study structured convex optimization problems, with additive objective $r:=p + q$, where $r$ is ($\mu$-strongly) convex, $q$ is $L_q$-smooth and convex, and $p$ is $L_p$-smooth, possibly nonconvex. For such a class of problems, we proposed an inexact accelerated gradient sliding method that can skip the gradient computation for one of these components while still achieving optimal complexity of gradient calls of $p$ and $q$, that is, $\mathcal{O}(\sqrt{L_p/\mu})$ and $\mathcal{O}(\sqrt{L_q/\mu})$, respectively. This result is much sharper than the classic black-box complexity $\mathcal{O}(\sqrt{(L_p+L_q)/\mu})$, especially when the difference between $L_q$ and $L_q$ is large. We then apply the proposed method to solve distributed optimization problems over master-worker architectures, under agents' function similarity, due to statistical data similarity or otherwise. The distributed algorithm achieves for the first time lower complexity bounds on {\it both} communication and local gradient calls, with the former having being a long-standing open problem. Finally the method is extended to distributed saddle-problems (under function similarity) by means of solving a class of variational inequalities, achieving lower communication and computation complexity bounds. | ['Gesualdo Scutari', 'Alexander Gasnikov', 'Ekaterina Borodich', 'Aleksandr Beznosikov', 'Dmitry Kovalev'] | 2022-05-30 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-3.28809023e-01 2.06478015e-01 5.77987023e-02 -1.40348420e-01
-1.21896267e+00 -4.97759968e-01 -2.00310707e-01 4.93536025e-01
-9.06795323e-01 9.74067926e-01 -4.46407795e-01 -3.06741804e-01
-8.27284694e-01 -1.03601682e+00 -9.95973587e-01 -1.03131902e+00
-9.22464788e-01 6.91095412e-01 -2.40311008e-02 -3.61331612e-01
2.81110227e-01 2.79085785e-01 -1.46415067e+00 -4.17889774e-01
1.16342914e+00 1.48565578e+00 5.59641086e-02 7.68164277e-01
-8.59138817e-02 6.94697559e-01 -4.17500466e-01 -3.52400333e-01
6.67409837e-01 -3.38394642e-01 -9.12838757e-01 -1.24709979e-01
2.52921820e-01 6.48783967e-02 3.71229835e-02 1.47745454e+00
4.84769344e-01 6.45985067e-01 1.86288595e-01 -1.25027037e+00
-2.53360838e-01 4.81815547e-01 -1.04193425e+00 1.95012808e-01
9.50025171e-02 1.73482776e-01 1.06102908e+00 -6.22096360e-01
4.82753456e-01 1.16880214e+00 5.93616962e-01 7.65555575e-02
-1.12182593e+00 -7.02448308e-01 4.97975707e-01 -2.25676060e-01
-1.53952730e+00 -2.43251938e-02 3.37669194e-01 -3.36952865e-01
8.98892879e-01 5.17254412e-01 3.58329237e-01 -3.30442399e-01
1.84520036e-01 4.56021011e-01 1.20472550e+00 -2.22892374e-01
4.86140609e-01 -7.18502775e-02 2.59760171e-01 1.09241521e+00
1.46367729e-01 -7.94058740e-02 -2.78764516e-01 -3.90322119e-01
5.11696815e-01 6.10592514e-02 -3.46049815e-01 1.06140137e-01
-7.83101439e-01 1.06938910e+00 3.52518111e-01 2.40756661e-01
-2.27814198e-01 5.09252071e-01 2.33505338e-01 4.59442437e-01
6.96192622e-01 1.92234188e-01 -7.04174519e-01 -2.46448427e-01
-1.02109623e+00 4.24174815e-01 9.84702587e-01 1.06048942e+00
1.35498941e+00 -5.56464912e-03 -7.64565766e-02 7.64947057e-01
1.10121772e-01 8.94625962e-01 -7.82633796e-02 -1.40732074e+00
1.03648782e+00 5.82283139e-01 4.02721822e-01 -1.01043034e+00
-6.25102460e-01 -4.41203535e-01 -1.07315886e+00 3.85573655e-01
7.28455245e-01 -5.97313941e-01 -2.13497296e-01 1.97403705e+00
5.04419744e-01 -5.86037710e-02 -3.43573004e-01 7.76991427e-01
-2.56806500e-02 8.71617973e-01 -3.42483461e-01 -7.73997724e-01
1.27446854e+00 -1.02413666e+00 -1.81879833e-01 -9.62864831e-02
7.32000649e-01 -7.25006461e-01 1.14856720e+00 3.66767049e-01
-1.73905730e+00 -1.95093185e-01 -5.13180137e-01 6.12247959e-02
-2.37844199e-01 -2.51052797e-01 5.28612733e-01 7.94035792e-01
-1.26175344e+00 7.71357417e-01 -8.54063630e-01 2.94370025e-01
2.98022717e-01 8.59506249e-01 -5.89188486e-02 -7.18487501e-02
-6.50977433e-01 3.35419863e-01 -2.03586608e-01 1.85476169e-01
-7.05311716e-01 -7.59703398e-01 -5.23932934e-01 1.13352343e-01
7.78774917e-01 -4.60610092e-01 6.65490568e-01 -6.00115240e-01
-1.33999658e+00 6.85292125e-01 -2.29008108e-01 -1.86209813e-01
6.16897345e-01 1.62238255e-01 2.57785797e-01 -3.34176496e-02
3.25176597e-01 -2.35051010e-02 7.03947783e-01 -9.37932670e-01
-7.25000441e-01 -1.02390850e+00 3.21318388e-01 3.38272780e-01
-2.65830904e-01 1.16067655e-01 -2.38929480e-01 -3.50937545e-01
-1.30216599e-01 -9.29290414e-01 -7.84068108e-01 -1.95484206e-01
-2.09804788e-01 -2.97874242e-01 2.50863165e-01 -6.53704166e-01
1.36996877e+00 -1.89422631e+00 2.92303771e-01 8.56179059e-01
5.46228409e-01 -3.38528268e-02 3.33987810e-02 4.23550546e-01
3.95741969e-01 1.61487907e-01 -5.30101061e-01 -4.94828761e-01
1.78330705e-01 1.68824822e-01 1.36945456e-01 7.59452641e-01
-5.56445837e-01 4.32964206e-01 -7.42529809e-01 -2.76671350e-01
-3.00709605e-01 1.22359701e-01 -8.91802073e-01 6.97881961e-03
-2.38506496e-01 2.20462471e-01 -7.12966084e-01 3.49905282e-01
9.93215024e-01 -4.19811428e-01 1.97788194e-01 2.64821321e-01
-4.02521878e-01 -1.47568449e-01 -1.88420117e+00 1.43834591e+00
-7.00309753e-01 -2.53466032e-02 1.29361796e+00 -1.61318862e+00
7.68025875e-01 -1.46883011e-01 1.03809261e+00 -5.82228303e-01
1.57615080e-01 3.79628539e-01 -3.34642977e-01 -2.79062480e-01
4.70387727e-01 -3.80406916e-01 -1.32777601e-01 7.58964598e-01
-4.82522666e-01 -2.66443133e-01 5.91038823e-01 1.48395762e-01
1.32018006e+00 -3.24101746e-01 -1.35949150e-01 -8.18902552e-01
7.88170695e-01 -6.75965920e-02 6.36945188e-01 8.34896266e-01
-2.32334867e-01 -9.39424410e-02 9.16549087e-01 -2.20194221e-01
-6.39802098e-01 -9.67803001e-01 1.03281550e-02 1.65656149e+00
4.68813896e-01 -3.62008065e-01 -9.46150362e-01 -5.07166922e-01
2.01169387e-01 2.72810876e-01 -5.38436949e-01 3.62820983e-01
-8.27304959e-01 -1.19085479e+00 1.69658184e-01 2.76863396e-01
4.50670391e-01 -8.58483553e-01 -4.98804927e-01 3.60588998e-01
1.15801297e-01 -4.87417519e-01 -9.02942061e-01 3.24482083e-01
-9.57759082e-01 -1.13368130e+00 -7.57665217e-01 -5.82500875e-01
9.76538956e-01 1.58589259e-02 1.10768950e+00 3.65416199e-01
-3.16197187e-01 5.12371004e-01 2.79653687e-02 -3.05467993e-01
2.27925941e-01 -6.49015531e-02 1.48144430e-02 5.41541167e-02
-3.22063476e-01 -5.45733392e-01 -9.24399376e-01 3.82766813e-01
-9.17331636e-01 -5.43363035e-01 1.24620877e-01 8.74203682e-01
9.54413533e-01 2.19013855e-01 3.93450379e-01 -8.72612000e-01
7.81172454e-01 -4.36913371e-01 -1.11634421e+00 9.93647054e-02
-6.91258550e-01 2.06120744e-01 1.06624722e+00 -2.51552314e-01
-7.60611653e-01 -2.37997726e-01 -1.09116413e-01 -2.30809376e-01
5.44553220e-01 5.20197392e-01 3.21959794e-01 -2.45741218e-01
5.71528256e-01 1.44859344e-01 2.38150135e-01 -4.78510410e-01
3.57249051e-01 2.63769925e-01 1.83992773e-01 -1.03305483e+00
5.08137584e-01 8.47420156e-01 1.46535799e-01 -6.06104672e-01
-6.00928485e-01 -1.87621832e-01 1.51340395e-01 1.43318057e-01
5.42489767e-01 -5.31970918e-01 -1.69810498e+00 -8.30185413e-02
-7.34215975e-01 -5.72598994e-01 -6.58421040e-01 4.14676100e-01
-7.55269945e-01 3.32516879e-01 -6.12888515e-01 -1.12223315e+00
-5.80086350e-01 -1.40772521e+00 7.98534334e-01 1.24927819e-01
3.60792398e-01 -1.02261627e+00 1.42216608e-01 5.45194387e-01
6.32895768e-01 7.15578347e-02 6.31034613e-01 -2.58301169e-01
-7.10184634e-01 -2.08916932e-01 -5.08684516e-01 3.74474436e-01
-2.76318312e-01 -3.53297085e-01 -1.54920220e-01 -7.95104861e-01
2.96292365e-01 -1.62209183e-01 5.82078457e-01 5.39256811e-01
1.38895690e+00 -9.02331531e-01 -2.99188912e-01 6.39226496e-01
1.59373760e+00 1.30875021e-01 2.51288831e-01 -1.09877892e-01
3.04867148e-01 3.93453747e-01 3.66634935e-01 9.98257995e-01
5.42169988e-01 4.68261153e-01 6.00420892e-01 -9.18787047e-02
4.68042403e-01 5.14448404e-01 5.06623924e-01 7.16468096e-01
-4.66109842e-01 1.23825386e-01 -7.03121662e-01 5.35650849e-01
-1.88856459e+00 -7.03260064e-01 -8.95239115e-02 2.52630830e+00
7.84931004e-01 -1.29573241e-01 3.50272834e-01 -1.66364804e-01
7.20809698e-01 8.92258212e-02 -5.34615993e-01 -7.83834994e-01
1.96250603e-01 1.00255358e+00 9.18184757e-01 7.24613190e-01
-5.89858413e-01 4.81441498e-01 4.74289322e+00 1.34964836e+00
-8.89304876e-01 4.41963971e-01 7.86272705e-01 -6.16864979e-01
-1.48139209e-01 5.65797947e-02 -6.46974981e-01 8.51201475e-01
7.41616249e-01 -1.32844001e-01 1.09743655e+00 1.06068134e+00
2.77052432e-01 -4.93051380e-01 -8.78653288e-01 9.43401873e-01
-2.63822019e-01 -1.35092580e+00 -9.15225804e-01 5.18043876e-01
1.03450882e+00 1.58014759e-01 6.95653036e-02 2.75691748e-01
7.93890238e-01 -1.02421021e+00 4.11667019e-01 -3.32094766e-02
6.45608664e-01 -9.79144037e-01 5.45245469e-01 5.75871587e-01
-1.67336726e+00 -3.11236322e-01 -3.72149825e-01 -3.35759908e-01
1.43562764e-01 7.67421246e-01 1.38446108e-01 5.57314098e-01
1.00903010e+00 3.67051624e-02 1.70274734e-01 6.25509202e-01
3.81728292e-01 -1.86347831e-02 -9.04356778e-01 -1.25045374e-01
4.93215233e-01 -7.98026919e-01 4.57623184e-01 9.92321074e-01
4.66828048e-01 3.84730071e-01 7.73145318e-01 6.45332813e-01
-3.88211966e-01 6.47920251e-01 -2.32598767e-01 4.23091501e-01
2.50023484e-01 1.11985779e+00 -8.17351997e-01 -3.70323598e-01
-2.04170182e-01 6.37859821e-01 6.07712805e-01 2.84433246e-01
-8.44416022e-01 -7.01651335e-01 8.78685832e-01 8.70736167e-02
2.94904321e-01 -3.97544980e-01 -3.42270643e-01 -9.89675462e-01
4.52513993e-01 -4.68059659e-01 8.18222165e-01 2.05673650e-01
-1.25766587e+00 3.77526581e-01 -1.33139268e-01 -6.96491897e-01
-3.25696468e-02 -5.05965531e-01 -4.72546518e-01 8.52435350e-01
-1.30732608e+00 -5.43823361e-01 1.15151731e-02 9.55344260e-01
-3.38303968e-02 -4.38397191e-02 5.53734958e-01 4.09123242e-01
-6.23375356e-01 6.33179605e-01 6.51790261e-01 -2.12108746e-01
1.31939039e-01 -1.21525669e+00 -6.06019855e-01 6.52657449e-01
-2.78052121e-01 5.16334713e-01 5.46920180e-01 -2.89881766e-01
-1.72534299e+00 -8.34428668e-01 6.50283873e-01 -9.60335061e-02
7.39078224e-01 -9.47799757e-02 -5.25106490e-01 5.19420862e-01
1.11832261e-01 3.57992232e-01 5.35987079e-01 3.17299329e-02
1.43818958e-02 -6.45973146e-01 -1.60824609e+00 2.90014803e-01
1.01850152e+00 -1.76965833e-01 2.55850941e-01 6.23976171e-01
3.61429304e-01 -4.87604022e-01 -1.30409110e+00 1.07134320e-02
1.16441876e-01 -1.11267686e+00 9.63260472e-01 -1.91633940e-01
3.04739904e-02 -2.07105443e-01 -2.56483018e-01 -8.86041999e-01
1.74690321e-01 -1.09573054e+00 1.13863535e-02 6.75698817e-01
4.32861954e-01 -1.01568091e+00 8.77362788e-01 7.82961309e-01
-1.90148070e-01 -1.06923914e+00 -1.51821387e+00 -8.06623399e-01
5.46350241e-01 -4.97448862e-01 3.54372382e-01 7.05600917e-01
1.42003983e-01 -1.53982058e-01 -2.61221796e-01 -1.99966300e-02
7.25993156e-01 3.09640735e-01 6.69598401e-01 -7.79555082e-01
-8.55623364e-01 -7.40089297e-01 1.39425308e-01 -1.08368897e+00
-1.04794828e-02 -1.03214049e+00 -6.05890341e-02 -1.15998423e+00
1.98796377e-01 -1.12561440e+00 -1.99821919e-01 3.79580647e-01
-7.97933638e-02 5.26968986e-02 4.31382418e-01 8.76827762e-02
-8.11376691e-01 3.97619635e-01 1.26270032e+00 -4.75462750e-02
-5.40863514e-01 1.52275011e-01 -6.74278796e-01 6.69367433e-01
3.83046359e-01 -4.68596786e-01 -2.11681306e-01 -6.09284818e-01
6.77184105e-01 7.43593395e-01 1.09797046e-01 -4.80331361e-01
4.29385662e-01 -5.53688467e-01 -4.20791626e-01 -1.50940225e-01
4.35434669e-01 -6.46060169e-01 -1.09485820e-01 6.33864522e-01
-1.36179984e-01 3.78467381e-01 -1.78909779e-01 4.05963808e-01
-3.93739343e-03 -1.97798014e-01 9.78219271e-01 -3.66931498e-01
7.29412064e-02 6.27430379e-01 -1.27711132e-01 4.95215148e-01
1.21974707e+00 4.11245413e-02 -2.62133092e-01 -3.91401678e-01
-7.00727880e-01 7.00975776e-01 1.85423777e-01 -4.31453109e-01
1.87359363e-01 -8.30965161e-01 -6.78211451e-01 -4.53661270e-02
-5.40980518e-01 5.83177209e-01 6.14453018e-01 1.39870131e+00
-5.71323216e-01 8.74120966e-02 3.09328645e-01 -4.49705601e-01
-7.90628731e-01 6.36053622e-01 2.97222614e-01 -9.52736974e-01
-2.43181050e-01 1.27300417e+00 -2.56665722e-02 -4.20780569e-01
2.36450717e-01 -3.68842840e-01 6.13225281e-01 -5.98603599e-02
3.58326644e-01 1.31512225e+00 -4.77503380e-03 -2.04275578e-01
-4.64394361e-01 6.64079010e-01 2.65047938e-01 -2.30135739e-01
1.42988443e+00 -3.56962949e-01 -7.23474324e-01 -1.43903986e-01
1.47167742e+00 2.46460348e-01 -1.19943941e+00 -1.64376736e-01
-2.16467515e-01 -3.19837630e-01 -2.92337865e-01 -4.48701918e-01
-1.56092155e+00 7.27659643e-01 6.47610247e-01 5.29987752e-01
1.20818830e+00 4.65691201e-02 8.19767773e-01 3.03165108e-01
7.51799941e-01 -1.73086131e+00 4.26406637e-02 6.08475924e-01
6.27787232e-01 -1.15783095e+00 1.35273233e-01 -3.05272490e-01
-2.76101559e-01 8.56125772e-01 3.63788724e-01 -5.46726048e-01
9.12802756e-01 4.01702106e-01 -4.42901075e-01 -2.55362958e-01
-4.28736240e-01 -1.72829196e-01 -8.77499953e-02 -5.26233949e-02
3.41396242e-01 3.21710914e-01 -8.41322005e-01 6.64810419e-01
-2.67198890e-01 -2.00094163e-01 2.08358720e-01 1.09096372e+00
-4.91435379e-01 -1.29159033e+00 -5.49295723e-01 6.81300461e-01
-6.70753956e-01 1.31161273e-01 3.27278405e-01 6.64287508e-01
3.65822047e-01 1.00330484e+00 1.49202928e-01 2.95473635e-01
7.35405460e-02 -1.22182220e-01 3.79943669e-01 -2.92128861e-01
-9.57499921e-01 2.32646793e-01 -2.19661608e-01 -8.51983607e-01
-2.36299336e-01 -4.32627052e-01 -1.64517760e+00 -6.75287306e-01
-1.64269760e-01 5.57236731e-01 6.52120948e-01 9.98847306e-01
3.64194244e-01 7.28892758e-02 9.56904411e-01 -7.16738999e-01
-9.83862996e-01 -5.79312682e-01 -9.15398955e-01 3.22559178e-01
-7.83014745e-02 -4.09633905e-01 -3.10623288e-01 -3.27003300e-01] | [6.363692283630371, 4.606513023376465] |
a051109b-6bb3-43cd-bc6d-d69fe59be29a | text-classification-and-clustering-with | 2107.14597 | null | https://arxiv.org/abs/2107.14597v1 | https://arxiv.org/pdf/2107.14597v1.pdf | Text Classification and Clustering with Annealing Soft Nearest Neighbor Loss | We define disentanglement as how far class-different data points from each other are, relative to the distances among class-similar data points. When maximizing disentanglement during representation learning, we obtain a transformed feature representation where the class memberships of the data points are preserved. If the class memberships of the data points are preserved, we would have a feature representation space in which a nearest neighbour classifier or a clustering algorithm would perform well. We take advantage of this method to learn better natural language representation, and employ it on text classification and text clustering tasks. Through disentanglement, we obtain text representations with better-defined clusters and improve text classification performance. Our approach had a test classification accuracy of as high as 90.11% and test clustering accuracy of 88% on the AG News dataset, outperforming our baseline models -- without any other training tricks or regularization. | ['Abien Fred Agarap'] | 2021-07-23 | null | null | null | null | ['text-clustering'] | ['natural-language-processing'] | [-1.68782398e-02 8.56308639e-02 -4.69486862e-01 -5.69501042e-01
-8.58377457e-01 -8.49588573e-01 8.60987902e-01 5.59846878e-01
-3.31062853e-01 5.59797704e-01 5.48543096e-01 -2.49762863e-01
-3.20183754e-01 -7.64944077e-01 -3.11755121e-01 -7.59122670e-01
4.50550057e-02 1.03163135e+00 -4.20188576e-01 -4.63049337e-02
4.02713567e-01 2.92989969e-01 -1.60033143e+00 7.07454026e-01
6.49633646e-01 7.12599695e-01 -3.36251318e-01 6.71372175e-01
-1.68261714e-02 6.60436749e-01 -7.28488982e-01 -1.26948565e-01
3.18818957e-01 -1.66011661e-01 -9.45256650e-01 7.46845827e-02
4.70157087e-01 5.33727221e-02 -4.78426784e-01 8.16185057e-01
5.66969961e-02 3.61834347e-01 1.26621425e+00 -1.11221182e+00
-9.29029703e-01 7.88061023e-01 -8.18168283e-01 1.20988339e-01
4.37140375e-01 -4.98195112e-01 1.53746748e+00 -8.57460916e-01
3.51878375e-01 1.24323928e+00 4.96392190e-01 3.37794155e-01
-1.80256104e+00 -8.76583755e-01 1.30762443e-01 5.57655357e-02
-1.55693269e+00 -5.24413705e-01 5.34961760e-01 -5.36767840e-01
9.66933548e-01 3.70915473e-01 3.61551553e-01 9.50889528e-01
5.49367033e-02 6.80006087e-01 8.53607833e-01 -5.77658236e-01
1.28216654e-01 2.85488516e-01 7.61835098e-01 5.78712642e-01
4.78322029e-01 -2.44692117e-01 -4.02147889e-01 -3.56447160e-01
3.56669992e-01 3.91134232e-01 -3.42940360e-01 -6.89509094e-01
-1.32184052e+00 1.28899479e+00 5.82895577e-01 4.62271631e-01
-1.81903113e-02 -7.66814128e-02 3.49430412e-01 5.16558111e-01
6.17915511e-01 7.52221048e-01 -4.94342387e-01 -3.01392912e-03
-1.00956154e+00 2.41690241e-02 7.88720250e-01 8.83202672e-01
8.60034883e-01 -1.15980618e-01 3.28336656e-02 8.34605277e-01
2.51222223e-01 1.76521719e-01 9.95651722e-01 -6.86068296e-01
6.17627263e-01 9.89998460e-01 -2.24533454e-01 -1.31953979e+00
-5.29268980e-01 -2.44901210e-01 -1.11774290e+00 7.73816928e-02
5.37867129e-01 2.52775624e-02 -7.98231840e-01 1.58079934e+00
-9.99619216e-02 -2.07220428e-02 3.08901995e-01 5.42040586e-01
5.18953741e-01 8.14128518e-01 -3.32365900e-01 -2.94954032e-01
1.33442330e+00 -5.45052469e-01 -4.57674563e-01 -9.28657353e-02
1.14167953e+00 -6.01477504e-01 1.01684844e+00 4.15678561e-01
-8.16596746e-01 -4.17863101e-01 -1.28691995e+00 -5.69405183e-02
-5.15321672e-01 -1.34879705e-02 6.97938979e-01 7.12628305e-01
-8.00290942e-01 5.64676225e-01 -7.54544199e-01 -1.90414727e-01
4.80410844e-01 4.68010694e-01 -7.87764072e-01 -1.16568565e-01
-8.24459851e-01 8.07129383e-01 5.72685480e-01 -4.65464175e-01
-3.42131853e-01 -6.63819551e-01 -9.17802989e-01 1.74963102e-01
5.35541363e-02 -6.27627432e-01 8.71380985e-01 -9.02388036e-01
-9.02380109e-01 1.01560080e+00 -1.97028860e-01 -2.57345945e-01
1.88074052e-01 -2.29324892e-01 -5.21217167e-01 -2.39146911e-02
2.10174814e-01 3.52846950e-01 8.23836088e-01 -1.21676254e+00
-3.76142740e-01 -6.25351608e-01 -3.73257905e-01 3.98354411e-01
-5.11039853e-01 -2.67519355e-01 1.83684714e-02 -6.12617731e-01
5.27740836e-01 -1.01972818e+00 -1.17753614e-02 -5.31943515e-02
-5.30948877e-01 -3.70487839e-01 9.71283972e-01 -2.66468227e-01
1.16325951e+00 -2.35795164e+00 1.63090944e-01 4.49714482e-01
6.77260578e-01 8.40566531e-02 -6.94136694e-02 3.10202450e-01
-5.02139270e-01 3.27816904e-01 9.83300582e-02 -2.19213754e-01
2.70221476e-02 3.63232553e-01 -4.74871367e-01 7.16131091e-01
4.63429801e-02 6.67965829e-01 -8.00494850e-01 -1.76235050e-01
1.56820700e-01 4.11078900e-01 -8.33920181e-01 -2.13976093e-02
-2.03596577e-02 -1.11510468e-04 -4.09090579e-01 4.21447940e-02
5.07677913e-01 -4.14638132e-01 3.07800978e-01 -9.25051272e-02
4.56971586e-01 6.04111433e-01 -1.25670314e+00 1.48533881e+00
-4.33880657e-01 1.04421914e+00 -3.69107842e-01 -1.15749371e+00
1.10657644e+00 1.83030263e-01 4.17742789e-01 -3.46432447e-01
-8.38782638e-02 -2.05371097e-01 1.55510619e-01 -1.41188815e-01
5.73949695e-01 -1.08336480e-02 -1.59412593e-01 8.41558874e-01
-9.91432518e-02 -3.31225502e-03 -7.76774958e-02 5.79633415e-01
1.02864182e+00 -4.40360636e-01 6.62004232e-01 -4.92516667e-01
5.93873858e-02 -4.00002338e-02 1.37573466e-01 6.64723158e-01
3.92222740e-02 7.17437923e-01 6.58172548e-01 -4.90459144e-01
-8.52249801e-01 -1.14157200e+00 -3.35307032e-01 1.24731493e+00
-2.50527114e-01 -8.59860957e-01 -1.68618679e-01 -9.13798094e-01
1.91229224e-01 9.46071744e-01 -1.04725075e+00 -4.65596616e-01
-2.42777228e-01 -7.07535684e-01 4.21548098e-01 6.06702685e-01
3.56891230e-02 -4.38937217e-01 -1.34029835e-02 -2.31606320e-01
-2.47121826e-01 -4.25913215e-01 -4.75771308e-01 5.84261239e-01
-8.65003645e-01 -1.21696389e+00 -2.60379672e-01 -8.66394877e-01
8.78475666e-01 4.36681151e-01 1.10770142e+00 -2.40747277e-02
-1.42375216e-01 2.29526795e-02 -4.37189460e-01 2.77110413e-02
-3.32174718e-01 1.29933223e-01 8.04086700e-02 -1.03591532e-01
6.58540070e-01 -3.97195667e-01 -3.96192193e-01 3.22697014e-01
-7.28267968e-01 -1.83337331e-01 3.22977871e-01 1.03849697e+00
3.01731497e-01 2.53469527e-01 5.30390799e-01 -1.17209947e+00
7.00366735e-01 -6.77289665e-01 -3.38984095e-02 2.25682184e-01
-8.01847994e-01 3.53573054e-01 6.96738303e-01 -5.81258476e-01
-4.99627322e-01 -5.68649126e-03 4.36452508e-01 -2.68571675e-01
-2.12943316e-01 4.88138467e-01 -1.47599086e-01 5.98042667e-01
1.22013938e+00 8.52008685e-02 3.41505073e-02 -2.76748985e-01
6.88964605e-01 9.98057783e-01 2.15019390e-01 -6.12029970e-01
6.03697956e-01 3.32203150e-01 -3.02716881e-01 -6.88751698e-01
-1.04720795e+00 -7.57305682e-01 -8.60191107e-01 4.86927837e-01
6.61691189e-01 -9.96362150e-01 -5.71870685e-01 -1.97564140e-01
-9.00344491e-01 2.49354735e-01 -3.23192984e-01 6.56283557e-01
-3.09421241e-01 3.62749875e-01 -2.79018402e-01 -4.34354007e-01
-7.27772564e-02 -9.58109260e-01 9.94006157e-01 -1.54312223e-01
-8.13281119e-01 -1.20315003e+00 2.67471503e-02 3.95789921e-01
2.01277193e-02 7.51887187e-02 1.37525725e+00 -1.51948369e+00
1.36733335e-02 -5.35917461e-01 -2.97050267e-01 5.58922105e-02
3.85505915e-01 -6.34863228e-02 -8.56901407e-01 -6.04455113e-01
-1.68196201e-01 -4.07366067e-01 1.04829085e+00 2.45380878e-01
1.14760125e+00 -6.62784338e-01 -6.17723286e-01 4.97698396e-01
9.53964055e-01 -9.44256410e-02 3.91128004e-01 1.55366883e-01
7.35548437e-01 6.59038484e-01 1.44816652e-01 5.23401022e-01
2.31785700e-01 7.18705416e-01 3.14967963e-03 2.49021173e-01
8.96078497e-02 -2.78853655e-01 2.07935885e-01 8.21348727e-01
3.25550526e-01 -2.98681825e-01 -9.80665624e-01 3.46940815e-01
-1.83324420e+00 -9.73232031e-01 -2.92386740e-01 2.09212112e+00
8.78604770e-01 3.17459702e-02 1.23904675e-01 4.28852886e-01
7.05231071e-01 1.04916245e-01 -5.35504460e-01 -3.37316275e-01
-4.03562635e-02 4.73437160e-02 1.00476041e-01 5.26178241e-01
-1.09771931e+00 7.23505437e-01 6.53290796e+00 6.39840603e-01
-6.23568952e-01 -4.34681386e-01 5.69526732e-01 -3.49326760e-01
-3.80977929e-01 -8.60206932e-02 -6.17244959e-01 1.58481404e-01
8.91270757e-01 -3.77878666e-01 4.43957984e-01 8.16683769e-01
-1.47659466e-01 2.19769776e-01 -1.75316691e+00 1.21049595e+00
3.83935690e-01 -1.36367095e+00 4.37773496e-01 1.52169928e-01
7.25519717e-01 -1.16616197e-01 2.83875763e-01 4.04211700e-01
8.94460499e-01 -1.49418771e+00 1.88637570e-01 3.48981589e-01
5.99476159e-01 -9.01575267e-01 5.01440942e-01 4.57545280e-01
-9.25840616e-01 4.53584129e-03 -6.80616677e-01 -1.93249732e-01
-5.83068848e-01 6.31498992e-01 -1.11696017e+00 4.18216258e-01
4.41331297e-01 1.10317588e+00 -8.16410959e-01 6.34085238e-01
-1.03875272e-01 5.89399397e-01 -4.32034805e-02 -1.48147240e-01
-5.41710630e-02 -1.02833331e-01 1.59073949e-01 1.22285616e+00
1.34203479e-01 4.95253094e-02 2.11000666e-01 8.21422875e-01
-2.71128923e-01 5.62605709e-02 -7.79731512e-01 -1.10473529e-01
6.97246611e-01 1.10174060e+00 -6.20519400e-01 -5.21545827e-01
-2.73708552e-01 8.52140307e-01 6.17979109e-01 3.49033892e-01
-3.02861363e-01 -6.29960358e-01 9.10257518e-01 -9.38523710e-02
1.36216739e-02 -2.47956708e-01 -5.93404174e-01 -1.41326904e+00
-2.00997442e-01 -9.60784853e-01 7.95659482e-01 -6.97176278e-01
-1.66839695e+00 5.51430821e-01 -1.31110728e-01 -1.14378905e+00
-5.17184913e-01 -7.13972628e-01 -4.11927283e-01 6.75351262e-01
-8.88697505e-01 -8.52391183e-01 4.86546457e-02 7.20103264e-01
3.07624698e-01 -5.06229639e-01 1.24277794e+00 -2.95108408e-01
-3.88084501e-01 9.29039240e-01 7.29486167e-01 6.12391055e-01
7.41810083e-01 -1.27376163e+00 1.29114777e-01 3.87952745e-01
7.84108758e-01 1.00135219e+00 3.71665031e-01 -3.79716814e-01
-1.05873942e+00 -1.05427504e+00 9.06650066e-01 -9.17686641e-01
8.40572417e-01 -6.39815390e-01 -1.10619295e+00 1.11137915e+00
1.06410570e-01 -1.43834939e-02 1.28723693e+00 7.21386075e-01
-1.17586386e+00 1.10846728e-01 -9.78208601e-01 6.70529962e-01
7.63168991e-01 -7.20107973e-01 -1.08246934e+00 4.67600465e-01
5.68803668e-01 1.27450824e-01 -8.81832421e-01 -6.26682788e-02
5.60780108e-01 -7.77093172e-01 1.03243637e+00 -1.07575452e+00
3.93067151e-01 -1.14354901e-01 -5.31664491e-01 -1.61336243e+00
-6.59545541e-01 -2.43114844e-01 -2.77334992e-02 1.18148422e+00
7.01745629e-01 -7.20162332e-01 8.67182314e-01 7.25154996e-01
1.23326115e-01 -4.34461713e-01 -8.41636419e-01 -5.89558542e-01
5.88978589e-01 -2.47170374e-01 4.09884691e-01 1.57827508e+00
7.39131272e-01 8.02993834e-01 2.25684363e-02 8.86285976e-02
4.24327523e-01 4.14567053e-01 5.70918500e-01 -1.61874425e+00
-1.93179354e-01 -5.84706426e-01 -6.97779596e-01 -1.03316879e+00
6.24985516e-01 -1.56228459e+00 -3.27048689e-01 -1.37972152e+00
4.69343454e-01 -4.36012030e-01 -2.64836103e-01 7.29608893e-01
-2.04165369e-01 8.49780161e-04 -8.95408075e-03 5.41096509e-01
-5.87280750e-01 4.25418407e-01 9.71782446e-01 -3.08852255e-01
-1.81318447e-01 -1.16083756e-01 -1.23020661e+00 6.49476945e-01
8.28083932e-01 -6.43045008e-01 -5.05897760e-01 -4.77203369e-01
1.20788530e-01 -5.58364056e-02 6.80313110e-02 -6.12844706e-01
2.32219473e-01 2.15955023e-02 9.96190429e-01 -4.21304315e-01
3.35828900e-01 -8.47502172e-01 -1.06131606e-01 2.74999082e-01
-1.03984439e+00 1.06719382e-01 -1.36355646e-02 7.50919402e-01
-1.25701025e-01 -1.03790008e-01 7.66921341e-01 6.32236600e-02
-1.98198780e-01 -1.02121383e-01 -5.70234537e-01 3.04442614e-01
1.00971770e+00 -2.33327314e-01 -5.85228384e-01 -3.09742630e-01
-7.42296338e-01 4.49702144e-02 5.58346510e-01 6.04783177e-01
5.87121785e-01 -1.40658617e+00 -8.98262918e-01 7.15942800e-01
4.57091182e-01 -8.35363865e-02 -2.82139271e-01 2.68537581e-01
-5.37139773e-02 4.15896863e-01 -3.90526243e-02 -6.21203899e-01
-1.17507422e+00 6.23160124e-01 1.73904762e-01 -1.01742297e-01
-5.38540184e-01 6.28372192e-01 2.96596527e-01 -7.32749879e-01
8.29087868e-02 -3.77144963e-01 -2.95520067e-01 3.53079408e-01
5.78676105e-01 2.47951806e-01 1.40570551e-01 -6.80799901e-01
-4.07220989e-01 4.79734242e-01 -6.20842516e-01 -1.60539709e-02
1.28567779e+00 2.01043189e-02 3.01608518e-02 7.11713016e-01
1.63383305e+00 3.67488898e-02 -8.82221937e-01 -3.63546908e-01
1.65180355e-01 -7.05432653e-01 1.76622331e-01 -8.22445571e-01
-7.89573252e-01 8.59761178e-01 2.08827138e-01 5.16678095e-01
6.23475492e-01 3.38947535e-01 -5.41231874e-03 7.07716048e-01
-2.04445928e-01 -8.21297467e-01 3.79397511e-01 6.31668329e-01
7.76580632e-01 -1.27429366e+00 1.26459777e-01 -1.83372334e-01
-8.00936282e-01 1.19741464e+00 3.49929124e-01 -4.04191792e-01
8.66004169e-01 7.44129494e-02 5.01356013e-02 -2.65674412e-01
-1.03411162e+00 2.43617836e-02 6.63346708e-01 8.49366426e-01
7.72330165e-01 1.91587254e-01 2.02911735e-01 6.19286776e-01
-6.46191835e-01 -6.85463309e-01 4.47322875e-01 5.90938091e-01
-5.62515855e-01 -7.76567876e-01 -2.88833231e-01 9.14799392e-01
2.47239675e-02 -1.79194927e-01 -7.22667336e-01 8.39329541e-01
-3.83027971e-01 1.29191971e+00 5.90842545e-01 -5.53831398e-01
2.06435695e-01 4.30526227e-01 2.82882363e-01 -1.00528646e+00
-3.75676990e-01 -1.62470937e-01 2.64079496e-02 -2.43988261e-01
-5.36936335e-02 -7.65630722e-01 -1.56695735e+00 -6.65413797e-01
-5.67042530e-01 4.98473704e-01 4.05511379e-01 8.52208674e-01
4.59129840e-01 2.38498643e-01 8.47963095e-01 -6.11672223e-01
-6.56162918e-01 -8.44946384e-01 -8.69405031e-01 6.36807263e-01
4.92557079e-01 -4.63844836e-01 -4.63028282e-01 -1.61698200e-02] | [10.294556617736816, 6.711891174316406] |
4f906550-d214-45a7-8dc2-f188d0ce938c | data-and-physics-driven-learning-models-for | 2204.01706 | null | https://arxiv.org/abs/2204.01706v1 | https://arxiv.org/pdf/2204.01706v1.pdf | Data and Physics Driven Learning Models for Fast MRI -- Fundamentals and Methodologies from CNN, GAN to Attention and Transformers | Research studies have shown no qualms about using data driven deep learning models for downstream tasks in medical image analysis, e.g., anatomy segmentation and lesion detection, disease diagnosis and prognosis, and treatment planning. However, deep learning models are not the sovereign remedy for medical image analysis when the upstream imaging is not being conducted properly (with artefacts). This has been manifested in MRI studies, where the scanning is typically slow, prone to motion artefacts, with a relatively low signal to noise ratio, and poor spatial and/or temporal resolution. Recent studies have witnessed substantial growth in the development of deep learning techniques for propelling fast MRI. This article aims to (1) introduce the deep learning based data driven techniques for fast MRI including convolutional neural network and generative adversarial network based methods, (2) survey the attention and transformer based models for speeding up MRI reconstruction, and (3) detail the research in coupling physics and data driven models for MRI acceleration. Finally, we will demonstrate through a few clinical applications, explain the importance of data harmonisation and explainable models for such fast MRI techniques in multicentre and multi-scanner studies, and discuss common pitfalls in current research and recommendations for future research directions. | ['Guang Yang', 'Yonina C. Eldar', 'Daniel Rueckert', 'Pietro Lio', 'Zidong Wang', 'Yang Li', 'Zhifan Gao', 'Yinzhe Wu', 'Huanjun Wu', 'Yang Nan', 'Yingying Fang', 'Jiahao Huang'] | 2022-04-01 | null | null | null | null | ['explainable-models'] | ['computer-vision'] | [ 4.06612933e-01 2.23192304e-01 -1.33867353e-01 -5.98220885e-01
-1.07835698e+00 -3.30026895e-01 3.80670667e-01 1.29970402e-01
-3.60759825e-01 5.15886009e-01 4.70671117e-01 -5.58048546e-01
-6.12920642e-01 -4.53526646e-01 -4.40116554e-01 -1.05294335e+00
-5.34625888e-01 8.15072715e-01 8.98710266e-02 -4.18464616e-02
-5.79197109e-02 5.90452313e-01 -5.05193949e-01 4.58427519e-01
4.86631006e-01 6.61593080e-01 2.13551119e-01 9.20099616e-01
6.79231584e-02 1.32966864e+00 -3.20969075e-01 -1.55672818e-01
-1.27375484e-01 -6.33584917e-01 -1.29137981e+00 -1.64185137e-01
-1.11149706e-01 -6.15163505e-01 -5.40711224e-01 8.42028797e-01
1.13315415e+00 -4.45275707e-03 6.54681206e-01 -8.34905863e-01
-6.66379333e-01 9.20465112e-01 -5.41002691e-01 9.42870140e-01
-2.53597766e-01 4.45943892e-01 -1.13017730e-01 -6.53963745e-01
6.32368088e-01 6.81023300e-01 9.51242149e-01 8.83993447e-01
-1.18228829e+00 -3.95459414e-01 -1.31462425e-01 5.80842867e-02
-8.07482660e-01 -3.71450156e-01 5.51113665e-01 -6.28625631e-01
9.65706587e-01 2.76917577e-01 5.77435553e-01 1.20163369e+00
8.89647067e-01 4.50653881e-01 1.21919620e+00 -1.15372702e-01
1.12430058e-01 -2.81698555e-01 9.22782943e-02 3.94545406e-01
-1.24995999e-01 3.90925258e-01 -1.43564209e-01 -2.76024193e-02
8.75280619e-01 -1.77611426e-01 -2.98545718e-01 -2.66376734e-01
-1.29389358e+00 9.70161974e-01 8.44498754e-01 7.33731210e-01
-6.85454965e-01 2.37839535e-01 7.45521367e-01 1.26036391e-01
4.25681531e-01 2.92723626e-01 -2.72543162e-01 -5.30378930e-02
-1.24496949e+00 1.46809533e-01 1.45478621e-01 5.00473261e-01
-3.46768379e-01 4.14925903e-01 -1.27472863e-01 7.08116829e-01
1.89379796e-01 2.78709739e-01 9.32630897e-01 -8.52795660e-01
-1.03641033e-01 -1.53316855e-01 -3.85836244e-01 -6.13054574e-01
-1.01719046e+00 -5.51319122e-01 -1.20189309e+00 1.60223514e-01
2.41822109e-01 -3.28756034e-01 -1.40133655e+00 1.41246939e+00
2.37876669e-01 4.74294648e-02 -1.52020887e-01 1.30143583e+00
1.20868671e+00 3.20559829e-01 5.38515449e-01 -1.30312234e-01
1.40593958e+00 -7.14659750e-01 -7.02137947e-01 -2.31733516e-01
8.49789441e-01 -6.93965197e-01 8.08678091e-01 2.93244272e-01
-1.44543076e+00 -3.57287169e-01 -8.21207106e-01 -2.19547618e-02
-4.61975411e-02 -6.35052800e-01 8.66315186e-01 7.81667888e-01
-1.14282203e+00 7.02604353e-01 -1.53266490e+00 -6.46739006e-02
8.66486728e-01 7.81512976e-01 -2.88579255e-01 -1.60338700e-01
-1.20763016e+00 1.13422894e+00 1.82363018e-01 3.08542252e-01
-1.34811473e+00 -1.23528600e+00 -6.32853031e-01 -3.17463994e-01
-1.03458501e-01 -8.42916369e-01 1.52308547e+00 -8.00285876e-01
-1.08312619e+00 1.05523312e+00 3.87673736e-01 -8.19151580e-01
4.48951811e-01 2.18022987e-01 -6.69783115e-01 1.93856075e-01
1.18553579e-01 7.59965658e-01 6.05337441e-01 -9.40166473e-01
-1.46935642e-01 -3.58462453e-01 -2.03258932e-01 1.39830187e-01
3.32286358e-01 3.29946727e-01 1.27703488e-01 -7.28674233e-01
1.37433097e-01 -8.75715911e-01 -7.82126546e-01 -8.47308412e-02
-2.04879165e-01 4.97312397e-01 7.17134356e-01 -8.42072308e-01
6.97480440e-01 -1.85721540e+00 -1.74957156e-01 -5.13255149e-02
6.07840896e-01 2.09782541e-01 1.45111039e-01 -6.68276772e-02
-7.24725306e-01 1.40791163e-01 -5.60351431e-01 4.12853509e-01
-6.06362343e-01 1.73413321e-01 7.20709488e-02 7.84768641e-01
1.85442362e-02 1.38213992e+00 -1.00941801e+00 -4.24363673e-01
6.48403347e-01 7.39931703e-01 -3.86829913e-01 -2.76983269e-02
3.72702330e-01 1.33868206e+00 -4.95147467e-01 5.12752116e-01
5.31842589e-01 -2.79272765e-01 4.49867174e-02 -3.68277133e-01
5.12263849e-02 3.24479997e-01 -5.00311852e-01 1.94492424e+00
-6.22555077e-01 6.27551138e-01 1.93851203e-01 -1.14117217e+00
2.77668804e-01 6.49706960e-01 1.10063303e+00 -1.03103602e+00
4.21156943e-01 3.08666348e-01 7.71676540e-01 -7.93397665e-01
-4.39910963e-02 -8.78662229e-01 1.62689865e-01 5.12590945e-01
-6.39093667e-02 -2.65680730e-01 -5.41421175e-01 1.64841756e-01
1.07715273e+00 -2.25173011e-02 -1.76094368e-01 -3.06973130e-01
2.41596609e-01 3.75869930e-01 1.78198770e-01 6.95720375e-01
-6.23179018e-01 8.41435790e-01 2.33446553e-01 -6.87817872e-01
-1.22607851e+00 -9.18024421e-01 -5.24699807e-01 7.23287404e-01
-2.15079606e-01 2.21160784e-01 -9.17761743e-01 -5.49669802e-01
-4.96128887e-01 6.35595620e-01 -8.13907444e-01 -3.81990820e-01
-9.03923571e-01 -1.41813922e+00 5.73179960e-01 8.44377220e-01
2.13682409e-02 -1.23158026e+00 -7.82566011e-01 5.81133306e-01
-2.21313193e-01 -8.33328784e-01 -2.15885371e-01 5.01909971e-01
-1.27622843e+00 -9.40490127e-01 -1.11492574e+00 -9.07028019e-01
5.21356881e-01 -1.30293025e-02 1.19403875e+00 -3.58211733e-02
-5.77380598e-01 2.32137382e-01 -1.11192718e-01 -4.18433607e-01
-6.39337063e-01 -7.83322155e-02 -3.15208025e-02 -6.00784600e-01
-1.35817118e-02 -5.12584865e-01 -1.08380163e+00 1.05718933e-01
-1.09095991e+00 2.52051324e-01 5.19855022e-01 1.15175581e+00
7.29371846e-01 7.45677799e-02 5.14407218e-01 -1.05858111e+00
6.24274552e-01 -7.21794844e-01 7.29262382e-02 -5.57558723e-02
-5.50721228e-01 -1.00459643e-01 1.88626915e-01 -2.44528294e-01
-1.06178427e+00 -1.71548232e-01 -7.03813493e-01 -2.05136120e-01
-1.17318928e-01 5.60091257e-01 5.40988028e-01 -4.20930654e-01
8.89409542e-01 9.30132568e-02 3.05614680e-01 -2.80411452e-01
2.15634435e-01 4.20137495e-01 7.14253187e-01 -1.65271029e-01
3.77883017e-01 7.25315154e-01 6.99029639e-02 -5.84253848e-01
-4.89055514e-01 7.61470497e-02 -7.23367512e-01 -3.11232716e-01
1.03043818e+00 -5.49036086e-01 -2.38271534e-01 4.09677118e-01
-8.42954278e-01 -4.67798114e-01 -7.25799084e-01 6.93389297e-01
-7.31796384e-01 6.58856034e-02 -1.13697338e+00 -3.23229134e-01
-8.43289614e-01 -1.86969435e+00 7.56051421e-01 1.68801889e-01
-2.45662302e-01 -1.33482826e+00 3.73630319e-03 2.44348079e-01
9.08812702e-01 7.35001504e-01 1.15696919e+00 -5.73519945e-01
-1.84977770e-01 -1.20795451e-01 -1.08051844e-01 -8.15668795e-03
1.41329393e-01 -4.49582368e-01 -8.92468333e-01 -2.67639488e-01
5.88509798e-01 -2.57334411e-01 3.58343214e-01 1.40816426e+00
1.50257325e+00 1.59599453e-01 -6.86995536e-02 9.58442807e-01
1.16171825e+00 4.58324492e-01 5.77138603e-01 3.60008180e-01
6.20685697e-01 5.41239977e-01 2.65531242e-01 -1.43656448e-01
1.90883592e-01 4.16484505e-01 4.71722543e-01 -6.81027770e-01
-6.44441187e-01 1.08659662e-01 -2.70997703e-01 9.96758163e-01
1.80738084e-02 1.56448141e-01 -1.19555116e+00 6.64238572e-01
-1.30735457e+00 -6.18729174e-01 -7.07453609e-01 1.66308224e+00
5.21808565e-01 4.22627255e-02 1.20753936e-01 6.90022018e-03
6.12016797e-01 -4.73084264e-02 -6.76470578e-01 -4.83037531e-01
4.77304347e-02 5.36035061e-01 8.38734567e-01 3.59757841e-01
-9.56705332e-01 5.67198813e-01 7.28725433e+00 5.40801227e-01
-1.41995645e+00 9.02324319e-01 1.12672925e+00 -1.95219710e-01
-2.21316978e-01 -1.69578761e-01 1.12933882e-01 4.43333775e-01
1.29046297e+00 9.85981710e-03 1.36396468e-01 5.45238972e-01
5.72866380e-01 -1.72335468e-03 -9.37138796e-01 7.31963754e-01
-4.23041135e-01 -1.52702510e+00 -2.72559673e-01 1.31710976e-01
5.95510840e-01 5.58773220e-01 3.05350929e-01 2.14920305e-02
3.77781433e-03 -1.40512002e+00 4.18321997e-01 2.98469961e-01
1.18914640e+00 -6.46545470e-01 8.32799911e-01 4.80225012e-02
-4.57588345e-01 3.46276999e-01 1.33985784e-02 4.15157825e-01
4.32359278e-01 3.09057981e-01 -7.62267590e-01 4.61925149e-01
6.82083130e-01 4.05101717e-01 -8.61740112e-02 1.10650313e+00
1.81864277e-01 7.56210744e-01 8.68076682e-02 6.05295777e-01
6.94098413e-01 3.13861758e-01 5.83108068e-01 1.23595250e+00
-6.00610748e-02 5.11831164e-01 -2.37061590e-01 6.04831994e-01
3.37613404e-01 -1.21115103e-01 -4.33069646e-01 2.04292357e-01
-2.46528268e-01 1.31966996e+00 -1.20148325e+00 -1.82318687e-01
-3.08752686e-01 7.86109447e-01 -3.17671239e-01 9.06533226e-02
-9.75831747e-01 1.46581471e-01 9.26041380e-02 4.51384068e-01
-3.19932967e-01 -9.65910330e-02 -5.81927657e-01 -7.77724802e-01
-5.19899249e-01 -9.58441675e-01 4.84448105e-01 -7.24606156e-01
-1.18332791e+00 8.48029315e-01 4.86848168e-02 -9.66120780e-01
-3.45926076e-01 -4.79392678e-01 -5.16410470e-01 1.04783833e+00
-1.32679439e+00 -9.88092721e-01 -9.49438959e-02 6.94567859e-01
5.87964594e-01 4.16412912e-02 9.00133789e-01 6.44050479e-01
-2.03326210e-01 1.76541865e-01 5.10597937e-02 2.04691529e-01
4.40040112e-01 -1.21707535e+00 5.04800498e-01 6.00686848e-01
-2.70276934e-01 4.05450374e-01 6.29845917e-01 -6.75401628e-01
-1.38653779e+00 -9.14914131e-01 4.33483332e-01 -3.85475069e-01
5.98399580e-01 -1.40975295e-02 -1.01729465e+00 6.17903471e-01
9.02123600e-02 4.37363118e-01 6.64145470e-01 -3.51672798e-01
5.71446061e-01 9.07104909e-02 -1.65599811e+00 2.65393436e-01
5.65363526e-01 -2.84560353e-01 -2.98911780e-01 7.79095948e-01
4.58073258e-01 -1.08517694e+00 -1.13112760e+00 4.63569850e-01
5.14392316e-01 -8.79277349e-01 1.00415254e+00 -8.02286208e-01
5.76221943e-01 2.92108089e-01 3.57477069e-01 -1.25411105e+00
-6.68556035e-01 -3.23119730e-01 6.27771839e-02 5.45309246e-01
1.48943946e-01 -3.99754733e-01 1.06002796e+00 8.64268601e-01
-4.37614262e-01 -9.16245282e-01 -1.01345372e+00 -2.57095158e-01
5.64598262e-01 -6.93532884e-01 3.96343768e-01 1.06157267e+00
-5.20998597e-01 -9.02249739e-02 -2.06856340e-01 -7.57522834e-03
6.94244981e-01 -3.15783948e-01 -8.71849731e-02 -6.99094057e-01
-1.32825986e-01 -5.95847368e-01 -4.25469398e-01 -3.48376602e-01
-4.27181572e-01 -1.02844441e+00 -7.04511479e-02 -1.74117780e+00
3.93905938e-01 -6.65106297e-01 -6.08153820e-01 2.44411767e-01
1.01576485e-01 4.75140870e-01 -1.42513439e-01 3.12081724e-01
1.48711309e-01 5.65001220e-02 1.67502153e+00 -2.11077660e-01
-1.79876294e-02 2.65376922e-02 -8.94050181e-01 6.78256273e-01
6.35147393e-01 -7.48682201e-01 -4.81188148e-01 -7.85778582e-01
-2.86956102e-01 5.71871281e-01 6.41438127e-01 -9.48319018e-01
1.03468701e-01 1.43497854e-01 7.52512693e-01 -3.18462402e-01
-5.71876504e-02 -9.58074570e-01 3.78659338e-01 1.01410890e+00
-4.44111168e-01 3.05682242e-01 3.97420973e-01 5.46641275e-02
-9.10052210e-02 -1.96578786e-01 1.21059573e+00 -4.38600123e-01
-5.49662828e-01 5.72584152e-01 -5.18267214e-01 9.19330642e-02
9.09339726e-01 -2.66168624e-01 2.66988903e-01 -3.63838673e-01
-1.33514118e+00 9.62770432e-02 -1.15657851e-01 3.57684344e-01
6.93132102e-01 -1.27266693e+00 -7.90802896e-01 -1.64037431e-03
-4.22062397e-01 1.42827690e-01 9.52345788e-01 1.54736626e+00
-8.84809673e-01 4.95402217e-01 -3.97839904e-01 -8.61760318e-01
-8.77092123e-01 4.37978268e-01 1.09191895e+00 -4.61153269e-01
-1.16805816e+00 1.02858496e+00 3.39381009e-01 -3.15103203e-01
-4.92806025e-02 -2.47997075e-01 1.34208962e-01 -3.33599389e-01
4.46236849e-01 8.45325291e-02 6.64455831e-01 -7.57495880e-01
-7.09304869e-01 3.41011822e-01 -4.09904391e-01 -1.02840967e-01
1.57037425e+00 -1.15557753e-01 2.27408633e-01 1.32838383e-01
9.82776403e-01 -5.91434896e-01 -1.07567441e+00 -1.99351599e-03
-2.04404593e-01 2.03927793e-02 8.77631843e-01 -1.15797210e+00
-1.67806470e+00 1.09895027e+00 1.20223367e+00 6.51122555e-02
1.23427856e+00 -1.35555193e-01 1.01747727e+00 -7.33705223e-01
2.21335307e-01 -6.67778671e-01 -3.35099399e-01 -1.42116463e-02
8.06432486e-01 -1.26824129e+00 -2.27121897e-02 5.34474617e-04
-9.11765695e-01 1.10977137e+00 2.53911048e-01 -5.86474985e-02
7.72217155e-01 8.09376359e-01 1.72316283e-01 -7.26318896e-01
-2.75896519e-01 1.95289478e-01 2.19578490e-01 8.36663544e-01
8.07226479e-01 5.25186434e-02 -2.24928707e-01 3.80710214e-01
-4.10538428e-02 2.52394468e-01 2.63364494e-01 1.01009357e+00
1.31118542e-03 -1.02004838e+00 -4.12726343e-01 6.63293004e-01
-1.05748665e+00 -3.10229182e-01 1.06092520e-01 7.90779948e-01
1.36069134e-01 7.04559565e-01 -1.91782892e-01 3.84594849e-03
2.86217302e-01 -1.21308826e-01 6.90132976e-01 -4.78306353e-01
-9.18700635e-01 2.36492053e-01 -2.46019006e-01 -4.92385477e-01
-4.38818872e-01 -5.38785458e-01 -1.41830754e+00 -3.50806087e-01
3.96664068e-02 -1.23064168e-01 1.00946343e+00 8.67897511e-01
8.11952874e-02 1.15506887e+00 3.73219192e-01 -8.26394856e-01
-2.31090039e-01 -7.23727882e-01 -4.56905693e-01 2.49176249e-01
4.35058355e-01 -4.41548109e-01 1.97096691e-01 -7.86170959e-02] | [14.18740463256836, -2.4251739978790283] |
62501cb4-c73c-4fa0-b00d-d87c45c9ea9c | sequence-level-knowledge-distillation-for-1 | 2305.13899 | null | https://arxiv.org/abs/2305.13899v1 | https://arxiv.org/pdf/2305.13899v1.pdf | Sequence-Level Knowledge Distillation for Class-Incremental End-to-End Spoken Language Understanding | The ability to learn new concepts sequentially is a major weakness for modern neural networks, which hinders their use in non-stationary environments. Their propensity to fit the current data distribution to the detriment of the past acquired knowledge leads to the catastrophic forgetting issue. In this work we tackle the problem of Spoken Language Understanding applied to a continual learning setting. We first define a class-incremental scenario for the SLURP dataset. Then, we propose three knowledge distillation (KD) approaches to mitigate forgetting for a sequence-to-sequence transformer model: the first KD method is applied to the encoder output (audio-KD), and the other two work on the decoder output, either directly on the token-level (tok-KD) or on the sequence-level (seq-KD) distributions. We show that the seq-KD substantially improves all the performance metrics, and its combination with the audio-KD further decreases the average WER and enhances the entity prediction metric. | ['Alessio Brutti', 'Daniele Falavigna', 'Muqiao Yang', 'Umberto Cappellazzo'] | 2023-05-23 | null | null | null | null | ['spoken-language-understanding', 'spoken-language-understanding'] | ['natural-language-processing', 'speech'] | [ 2.44512394e-01 2.61436909e-01 1.11045815e-01 -2.50963479e-01
-7.28532493e-01 -4.11856532e-01 4.69959348e-01 1.95259765e-01
-7.92299986e-01 9.37554121e-01 3.43385905e-01 -4.05275881e-01
-1.92303911e-01 -6.31884754e-01 -7.98683226e-01 -6.16010010e-01
1.32056430e-01 5.15470326e-01 5.76458931e-01 -2.61194080e-01
-1.26717687e-01 1.87815726e-01 -1.66820145e+00 2.80273616e-01
9.80297744e-01 5.76515138e-01 6.36287868e-01 8.44436586e-01
-3.20464194e-01 1.08121204e+00 -4.27965522e-01 -3.84500027e-01
-2.21398510e-02 -4.93805468e-01 -8.58764470e-01 -3.98664802e-01
1.94117635e-01 -2.30067343e-01 -6.04897559e-01 8.86221707e-01
6.21449828e-01 3.30547005e-01 3.91962677e-01 -9.92109239e-01
-4.51659918e-01 1.05014718e+00 -1.75271437e-01 2.67995387e-01
1.07573293e-01 -2.16836303e-01 8.30037653e-01 -1.29295313e+00
5.46204090e-01 1.16007900e+00 7.07043648e-01 6.18165076e-01
-8.85487616e-01 -4.22878325e-01 2.65690088e-01 6.37554526e-01
-1.56293488e+00 -7.15156138e-01 4.36395466e-01 -4.00491692e-02
1.35568178e+00 -2.07785338e-01 5.25754631e-01 9.73837554e-01
-6.97116414e-03 1.16096842e+00 7.15806067e-01 -8.10730219e-01
4.51651990e-01 1.35591596e-01 7.69492909e-02 4.74008143e-01
7.65520483e-02 4.34556045e-02 -1.31234741e+00 1.07877083e-01
2.02603132e-01 -3.25631678e-01 -3.73372257e-01 -2.98137814e-01
-8.45070362e-01 5.32303631e-01 -2.45248809e-01 4.58939821e-01
-3.52566540e-01 4.85495813e-02 4.64647293e-01 6.64125204e-01
4.54922885e-01 2.28800133e-01 -1.01382399e+00 -8.37970674e-01
-1.28478837e+00 1.61950171e-01 9.77660537e-01 9.55863535e-01
7.41415739e-01 2.12815136e-01 -2.06120133e-01 9.81084704e-01
1.85505431e-02 4.03383106e-01 8.49514902e-01 -6.51843607e-01
3.19315463e-01 -2.45744996e-02 4.73775864e-02 -3.40374857e-01
-2.22799063e-01 -6.63784325e-01 -5.75972736e-01 -2.18906924e-01
3.23717803e-01 -1.98735803e-01 -1.05789137e+00 2.16961455e+00
1.33970678e-01 4.27868485e-01 3.76639873e-01 2.24151328e-01
3.96555305e-01 7.30834067e-01 1.48894876e-01 -6.17367268e-01
8.36145163e-01 -1.02420318e+00 -8.59956682e-01 -2.80645072e-01
7.10504055e-01 -4.91289467e-01 1.11367166e+00 6.88460410e-01
-1.07118607e+00 -4.98629779e-01 -1.14111161e+00 -4.25755009e-02
-4.25189942e-01 1.01812817e-02 -9.48458165e-02 4.84262347e-01
-1.35817981e+00 6.77160144e-01 -7.62214005e-01 -4.32884932e-01
1.27972841e-01 1.31252676e-01 -5.37495539e-02 -3.09255421e-01
-1.56502962e+00 1.21627867e+00 8.52846682e-01 -1.94276318e-01
-8.90476108e-01 -9.71834421e-01 -5.65760195e-01 4.40188587e-01
4.84706461e-01 -6.64675236e-01 1.53742635e+00 -9.05676544e-01
-1.91493797e+00 3.83239448e-01 -3.51097256e-01 -6.89467490e-01
6.48147225e-01 -6.86354578e-01 -4.14632171e-01 -1.07056789e-01
-2.44536221e-01 5.96177518e-01 9.61552441e-01 -1.09688222e+00
-9.58639622e-01 -9.76663977e-02 -2.88456142e-01 4.57759291e-01
-5.47397554e-01 -5.48922002e-01 -3.43602508e-01 -7.08167613e-01
-1.79701298e-01 -7.11029112e-01 2.44920552e-01 -4.66016412e-01
-1.51499987e-01 -2.65324146e-01 8.05251598e-01 -8.35176468e-01
1.64647424e+00 -2.41388822e+00 4.31580395e-01 -4.79790615e-03
-1.53167471e-01 5.93865871e-01 -2.56594837e-01 6.55722678e-01
-2.26608887e-02 -2.90949017e-01 -2.57293463e-01 -6.31591678e-01
-5.00379205e-02 5.37525833e-01 -5.22177458e-01 -1.14993915e-01
1.87558949e-01 7.12226391e-01 -1.10424531e+00 -8.48837495e-02
-1.26207396e-01 4.10201401e-01 -6.08679473e-01 1.49119228e-01
-3.19830328e-01 8.65896791e-02 2.03637660e-01 6.83840290e-02
4.63366628e-01 1.08910114e-01 3.24801505e-01 1.11003675e-01
-8.99614990e-02 4.70486164e-01 -1.10747969e+00 2.00686765e+00
-6.57096028e-01 5.54789186e-01 -1.65428385e-01 -8.11628044e-01
7.88081944e-01 4.65909064e-01 1.22968242e-01 -7.09056377e-01
-3.63824815e-01 4.02637422e-01 4.46681939e-02 -4.07734960e-01
9.64542270e-01 -4.93722588e-01 5.08159027e-02 4.83276367e-01
7.34672368e-01 1.06300466e-01 7.94272423e-02 4.69737142e-01
1.20391035e+00 9.02046114e-02 3.85601193e-01 -2.14395002e-01
2.28755966e-01 -3.11283499e-01 4.80041355e-01 9.89297092e-01
-4.14167754e-02 4.04937476e-01 2.59597391e-01 -1.84925213e-01
-1.03531408e+00 -1.34877181e+00 3.85919392e-01 1.43480706e+00
-9.12623182e-02 -3.76843780e-01 -7.17923582e-01 -6.09660387e-01
1.68864623e-01 1.44414055e+00 -3.33149374e-01 -8.20048869e-01
-4.73761171e-01 -3.22402269e-01 7.78969526e-01 4.77472723e-01
3.82879734e-01 -7.36591876e-01 -5.36029696e-01 5.54379761e-01
-2.97449052e-01 -8.58776152e-01 -4.38120902e-01 6.92597508e-01
-6.27643704e-01 -5.49124479e-01 -7.53574491e-01 -5.46848178e-01
6.48213997e-02 -2.77837347e-02 1.02185237e+00 -1.67129233e-01
-1.83948372e-02 6.40098572e-01 -5.82352638e-01 -6.81924343e-01
-6.30662441e-01 5.83449066e-01 6.06384635e-01 -2.32139230e-02
5.46129644e-01 -7.92691469e-01 -6.88018426e-02 -2.41728798e-01
-9.60577548e-01 7.21990094e-02 7.09486306e-01 9.88855362e-01
2.83670515e-01 1.90796956e-01 1.09583449e+00 -7.43168354e-01
7.72854090e-01 -4.34506685e-01 3.14942487e-02 5.65889418e-01
-7.39086986e-01 4.31242913e-01 5.74118972e-01 -5.95733583e-01
-1.27316284e+00 1.46415666e-01 -3.58334094e-01 -4.66557026e-01
6.48490489e-02 6.92550838e-01 -1.14327185e-01 3.75254601e-01
5.50985992e-01 8.19116473e-01 -1.69522658e-01 -6.55739307e-01
6.65027380e-01 6.23966753e-01 7.59535313e-01 -5.56070507e-01
5.22380352e-01 1.83569118e-02 -6.44598722e-01 -1.15016294e+00
-1.00069559e+00 -3.99639577e-01 -8.37022007e-01 -1.54624462e-01
2.87356943e-01 -9.30159390e-01 -1.28819451e-01 6.51816070e-01
-1.19603086e+00 -6.01355374e-01 -9.06176329e-01 4.95811909e-01
-6.91767156e-01 3.53166103e-01 -4.56005603e-01 -9.35187697e-01
-2.10574701e-01 -6.14464045e-01 5.97097337e-01 2.19534755e-01
-2.59013414e-01 -8.94409418e-01 2.43642285e-01 -2.56975919e-01
7.38679826e-01 -6.36909783e-01 1.23049963e+00 -8.74928594e-01
-2.93278307e-01 2.72651136e-01 9.71386582e-02 6.15008056e-01
3.50214466e-02 -3.64765674e-01 -1.25294280e+00 -5.04561722e-01
1.62515696e-02 -4.02648389e-01 1.17152715e+00 1.13238372e-01
8.17398787e-01 -2.61745602e-01 -1.71213984e-01 2.59098746e-02
1.23423469e+00 3.21416974e-01 6.50874555e-01 3.89810167e-02
3.95293891e-01 2.48328805e-01 4.00630176e-01 6.13420963e-01
5.68505228e-01 5.38924515e-01 -9.78946611e-02 3.14234555e-01
-4.74153221e-01 -7.00399280e-01 6.50147021e-01 1.63903773e+00
1.69741526e-01 -5.64991176e-01 -1.04878330e+00 9.61605191e-01
-1.90503025e+00 -7.82836854e-01 5.16979873e-01 2.23515654e+00
1.26923716e+00 1.34035096e-01 -1.39669195e-01 2.57764608e-01
4.30023700e-01 -5.72300069e-02 -6.77230895e-01 -3.60894561e-01
-1.18022189e-01 4.94778514e-01 2.14284286e-01 7.11625457e-01
-6.85340822e-01 9.88121927e-01 6.46922255e+00 1.27816868e+00
-9.47646558e-01 4.51461345e-01 1.97355017e-01 -1.67811409e-01
-2.54716575e-01 -9.53214914e-02 -9.13315535e-01 4.48783427e-01
1.38110662e+00 -4.96764332e-01 4.37227398e-01 6.23254836e-01
-5.11924326e-02 -3.30559343e-01 -1.28430748e+00 9.80520606e-01
1.72139212e-01 -9.56893563e-01 3.04132462e-01 -4.84126091e-01
4.81407821e-01 1.12413563e-01 7.05005527e-02 6.32972598e-01
3.89215738e-01 -8.78333330e-01 1.01125658e+00 9.27143276e-01
8.38545978e-01 -8.40160728e-01 6.12689853e-01 7.20572472e-01
-9.49996889e-01 -9.48593318e-02 -2.81798899e-01 4.87634763e-02
4.59228128e-01 7.40022123e-01 -1.13551235e+00 6.41821444e-01
6.19204700e-01 4.17735189e-01 -2.92687297e-01 9.14693475e-01
-2.02741802e-01 9.67438757e-01 -3.03029031e-01 1.58753172e-02
8.59397873e-02 3.31357569e-01 7.11814284e-01 1.46196747e+00
6.06809735e-01 -1.52275721e-02 -1.97255120e-01 4.30870086e-01
-2.14224383e-01 -7.97320902e-02 -3.36725503e-01 -6.11268096e-02
8.08834314e-01 4.77833688e-01 -2.27445230e-01 -5.03522575e-01
-3.67669880e-01 1.30379748e+00 5.85912883e-01 4.37039673e-01
-4.08854872e-01 -6.36296749e-01 5.84815800e-01 -8.24106336e-02
6.85352862e-01 -4.49105889e-01 9.22940671e-02 -1.22584653e+00
1.67597272e-02 -7.44966567e-01 2.83866733e-01 -6.23068571e-01
-1.20685947e+00 4.06519979e-01 7.10963085e-02 -7.14532316e-01
-4.96107548e-01 -2.91766286e-01 -1.60300374e-01 7.24969983e-01
-1.92227459e+00 -6.55352354e-01 -6.58890791e-03 4.58525687e-01
8.80960047e-01 -3.04338008e-01 9.46590781e-01 6.37544215e-01
-3.33805904e-02 7.23170578e-01 5.70949256e-01 -2.67684042e-01
7.85442352e-01 -1.10926032e+00 3.18360209e-01 7.75412858e-01
3.83662552e-01 4.38322127e-01 9.01259482e-01 -5.17551839e-01
-1.15791357e+00 -1.06990325e+00 1.40111482e+00 -3.14639091e-01
3.75579745e-01 -3.19083869e-01 -1.28996885e+00 5.82493901e-01
1.90546215e-01 -5.60503602e-01 5.62159717e-01 6.71280771e-02
-4.20567870e-01 -1.50285155e-01 -8.08963180e-01 3.66274625e-01
9.16911364e-01 -8.13917935e-01 -9.97576714e-01 3.83238588e-03
1.06175482e+00 -3.67910802e-01 -6.92754686e-01 1.14777558e-01
5.27308226e-01 -7.61649728e-01 6.42551184e-01 -7.23503768e-01
9.03677791e-02 -2.71486610e-01 -2.30589002e-01 -1.86890304e+00
-1.49339288e-01 -5.57622612e-01 -5.04096925e-01 1.31281412e+00
6.43119574e-01 -4.04093713e-01 5.10337234e-01 1.18802764e-01
-1.87334642e-01 -5.10254204e-01 -1.35977399e+00 -1.15589058e+00
3.17658901e-01 -5.34556746e-01 4.14205134e-01 8.31176162e-01
1.64981663e-01 4.79059368e-01 -6.40340984e-01 8.55658669e-03
2.05306083e-01 -4.44799900e-01 2.84327984e-01 -9.95363653e-01
-4.50895220e-01 -6.54908642e-02 -1.87133104e-01 -1.38267612e+00
-1.46013424e-01 -7.13911891e-01 4.61804658e-01 -1.22541320e+00
7.66934603e-02 -2.84024775e-01 -5.29199481e-01 5.46184242e-01
-1.81323513e-01 -4.70065087e-01 3.08718085e-01 -2.38345638e-02
-7.02641487e-01 9.72792268e-01 5.65559268e-01 -2.46039536e-02
-3.53786618e-01 -1.67014658e-01 -4.16467637e-01 4.29746300e-01
6.45376325e-01 -6.83864892e-01 -7.15504527e-01 -6.42994225e-01
3.98100495e-01 -2.03399196e-01 -9.11404565e-02 -1.21209300e+00
6.87951684e-01 2.89120048e-01 -1.13697223e-01 -5.30615389e-01
4.35988545e-01 -5.33651054e-01 3.75993252e-02 4.75988507e-01
-4.92604941e-01 2.20402982e-02 4.21737134e-01 8.63585353e-01
-4.21063989e-01 -3.33358079e-01 8.48555207e-01 -9.90428850e-02
-9.28954363e-01 -2.39072125e-02 -5.98510802e-01 3.13464016e-01
7.63499439e-01 1.11646488e-01 -1.79375187e-01 -6.66713834e-01
-1.01085901e+00 1.20043673e-01 1.72050864e-01 6.03544474e-01
6.41480863e-01 -1.17721498e+00 -7.24157155e-01 1.90516144e-01
1.43676788e-01 -1.20880872e-01 3.00450861e-01 7.97719538e-01
-6.31453246e-02 4.73044455e-01 -1.43931545e-02 -3.94197464e-01
-1.05613995e+00 2.89002717e-01 2.90064037e-01 -3.44945401e-01
-5.11580110e-01 1.15405345e+00 -5.23963347e-02 -4.61440504e-01
6.20508194e-01 -3.68202239e-01 1.48972142e-02 2.16956601e-01
5.90503275e-01 4.54907298e-01 4.36559349e-01 -2.88127720e-01
-2.89021283e-01 6.93293735e-02 -4.85147983e-01 -4.78516459e-01
1.38907337e+00 -5.18213093e-01 2.21573636e-01 9.98334885e-01
7.20655918e-01 -2.03998089e-01 -1.14691210e+00 -7.18433678e-01
3.97896618e-01 -6.23164475e-02 -2.49234606e-02 -1.07187498e+00
-4.55500007e-01 1.09492743e+00 4.50206995e-01 7.81127512e-02
1.08409047e+00 -1.06473893e-01 8.82067502e-01 6.52491450e-01
4.95033234e-01 -1.52993762e+00 1.33953482e-01 1.29724157e+00
5.32537401e-01 -5.03710270e-01 -3.10406029e-01 -3.50908712e-02
-8.28706801e-01 1.19570112e+00 3.03630412e-01 4.67402637e-01
6.07008338e-01 3.87985796e-01 -1.44770950e-01 2.86753386e-01
-1.20808804e+00 -2.03840330e-01 -1.67039171e-01 5.51319838e-01
2.72830784e-01 -5.03209494e-02 -7.27698430e-02 7.96209037e-01
-1.70701012e-01 4.29721683e-01 5.97970724e-01 9.80864406e-01
-7.98030853e-01 -1.15362298e+00 1.90932408e-01 4.25224870e-01
-9.25804973e-02 -4.57906306e-01 -2.28780061e-01 5.38019061e-01
3.60699534e-01 7.07232118e-01 5.23249693e-02 -4.00032461e-01
5.02117455e-01 9.40451026e-01 4.36154395e-01 -7.15367317e-01
-2.52962708e-01 -3.08715999e-01 1.43153831e-01 -2.20941439e-01
-2.76173621e-01 -7.39524007e-01 -1.15666449e+00 -1.54380411e-01
-3.58267784e-01 4.13083404e-01 3.88348669e-01 1.11853957e+00
6.38884962e-01 6.31747961e-01 2.51524508e-01 -2.69650936e-01
-8.25959146e-01 -1.15876567e+00 -8.17863762e-01 -3.29450965e-02
4.28714603e-01 -6.40801489e-01 -2.73676425e-01 2.47035578e-01] | [9.967093467712402, 3.6413612365722656] |
87e6c796-4f8c-497d-8a9c-80c093b8bace | semantic-annotation-aggregation-with | null | null | https://aclanthology.org/C16-1168 | https://aclanthology.org/C16-1168.pdf | Semantic Annotation Aggregation with Conditional Crowdsourcing Models and Word Embeddings | In modern text annotation projects, crowdsourced annotations are often aggregated using item response models or by majority vote. Recently, item response models enhanced with generative data models have been shown to yield substantial benefits over those with conditional or no data models. However, suitable generative data models do not exist for many tasks, such as semantic labeling tasks. When no generative data model exists, we demonstrate that similar benefits may be derived by conditionally modeling documents that have been previously embedded in a semantic space using recent work in vector space models. We use this approach to show state-of-the-art results on a variety of semantic annotation aggregation tasks. | ['Paul Felt', 'Kevin Seppi', 'Eric Ringger'] | 2016-12-01 | semantic-annotation-aggregation-with-1 | https://aclanthology.org/C16-1168 | https://aclanthology.org/C16-1168.pdf | coling-2016-12 | ['text-annotation'] | ['natural-language-processing'] | [ 7.73535594e-02 4.39911097e-01 -2.27965221e-01 -8.84628057e-01
-1.11092699e+00 -5.29872715e-01 9.05294776e-01 2.80011475e-01
-5.15044212e-01 9.39776301e-01 7.27992475e-01 9.91776064e-02
8.52706805e-02 -6.30865872e-01 -5.48309207e-01 -3.23346227e-01
4.88509506e-01 9.61312175e-01 2.13461056e-01 -3.35940003e-01
4.00795899e-02 -3.85567725e-01 -1.48696136e+00 5.63327372e-01
6.06179357e-01 6.74027503e-01 4.02348042e-02 5.28752685e-01
-7.47982383e-01 9.62904632e-01 -7.38915622e-01 -7.29034781e-01
4.35567871e-02 -4.97635543e-01 -1.01746690e+00 4.49994504e-02
4.56821352e-01 -1.61107227e-01 1.42934784e-01 9.68432546e-01
5.11511266e-01 4.64368343e-01 8.60535800e-01 -1.36058044e+00
-1.28390968e+00 9.36561763e-01 -2.42689058e-01 -2.15074480e-01
5.22397459e-01 -2.75433868e-01 1.42767453e+00 -1.30348361e+00
7.93978930e-01 1.47698808e+00 7.01390266e-01 6.97273731e-01
-1.43419003e+00 -4.31412518e-01 2.27099270e-01 -2.16011047e-01
-1.00551093e+00 -1.46036729e-01 6.70745015e-01 -5.16044199e-01
1.05208373e+00 4.27316315e-02 2.86190242e-01 1.07212865e+00
-3.82246554e-01 1.06102145e+00 1.05833876e+00 -6.06178820e-01
3.00075233e-01 3.32173198e-01 5.25073051e-01 4.41183746e-01
3.61467928e-01 -5.79058528e-01 -7.69453883e-01 -4.98608619e-01
3.06938142e-01 1.17169388e-01 2.22369149e-01 -2.74484515e-01
-7.18210280e-01 1.38212168e+00 3.72266382e-01 4.51350838e-01
-3.95475686e-01 4.08024907e-01 2.07585633e-01 1.16739631e-01
1.35666680e+00 5.20461321e-01 -2.75453091e-01 1.17112681e-01
-9.32011425e-01 5.81281424e-01 9.47961390e-01 1.24156702e+00
1.19713855e+00 2.68912874e-02 -3.22757095e-01 9.92803872e-01
5.77522516e-01 1.94979578e-01 5.62393665e-01 -8.95737648e-01
4.84574854e-01 8.69022191e-01 5.88431776e-01 -6.69547260e-01
-3.53817374e-01 1.25323787e-01 -2.52437443e-01 -2.20886490e-04
2.97795653e-01 -4.64029044e-01 -8.69595826e-01 1.66186488e+00
3.66475791e-01 -1.33604631e-01 2.09127307e-01 9.67812717e-01
1.07728052e+00 3.28033924e-01 9.79538679e-01 1.91189006e-01
1.36244690e+00 -8.69683325e-01 -9.79327798e-01 -5.22429407e-01
1.21728992e+00 -6.98643327e-01 1.28616297e+00 -7.84459338e-02
-8.87285948e-01 -4.73365605e-01 -6.00881040e-01 -4.13303852e-01
-7.29572892e-01 1.39140263e-01 9.21504676e-01 8.26490283e-01
-1.08234215e+00 8.21250677e-02 -7.16109514e-01 -4.70192999e-01
5.41166067e-01 1.06464764e-02 -3.80126774e-01 -2.87501346e-02
-1.35182953e+00 7.94928610e-01 4.16194290e-01 -5.17716289e-01
-6.72842562e-01 -6.19298339e-01 -8.71943414e-01 6.46160566e-04
3.36663455e-01 -9.47589755e-01 1.51815677e+00 -9.81757164e-01
-1.00575209e+00 9.46301579e-01 -2.48966992e-01 -5.36804318e-01
3.33661646e-01 -5.28091073e-01 -2.38617688e-01 -2.10583970e-01
4.13285404e-01 8.64879429e-01 3.02486062e-01 -1.26472497e+00
-5.60592473e-01 -5.37426889e-01 2.31004313e-01 4.22587305e-01
-7.64838696e-01 3.61167789e-01 1.61046028e-01 -6.38968170e-01
-2.36398026e-01 -1.13073885e+00 -4.30366576e-01 -3.65189403e-01
-2.97449738e-01 -7.64118731e-01 7.43134856e-01 -4.77302015e-01
1.07139337e+00 -1.83022833e+00 -2.23866075e-01 -2.14845017e-01
2.08640441e-01 2.33939048e-02 5.01547903e-02 4.40586507e-01
2.10960507e-01 5.42912602e-01 -6.76473603e-02 -9.29242671e-01
4.77776229e-01 2.56858528e-01 -4.73352820e-01 -7.22018182e-02
1.34158790e-01 1.10127795e+00 -9.09946382e-01 -4.76986021e-01
-1.00668103e-01 3.13200474e-01 -6.42059505e-01 1.78294390e-01
-7.14308739e-01 5.41106239e-02 -6.18156016e-01 3.83126646e-01
3.28689478e-02 -7.38855839e-01 2.46719301e-01 2.31420472e-01
4.08200055e-01 2.87521094e-01 -9.91254568e-01 1.96513772e+00
-3.33623022e-01 5.01455247e-01 -3.46735030e-01 -7.44983375e-01
8.79358470e-01 5.60643733e-01 5.03008366e-01 -2.96212345e-01
-6.89376965e-02 6.17224053e-02 -4.78482395e-01 -3.36350858e-01
1.18605363e+00 -4.14291054e-01 -5.30249178e-01 8.79372537e-01
4.56631869e-01 -2.87959635e-01 1.65508360e-01 5.92712998e-01
8.49528432e-01 1.55852214e-01 3.10392588e-01 -2.71215409e-01
-2.43437186e-01 4.73401845e-01 2.16043353e-01 9.03372109e-01
1.52652755e-01 6.02603793e-01 8.81024376e-02 -2.95813888e-01
-1.11844051e+00 -6.43449783e-01 1.47853643e-01 1.96090329e+00
1.58154313e-02 -5.37021637e-01 -7.79649496e-01 -8.75815094e-01
-5.22547588e-02 9.91289973e-01 -8.09083343e-01 6.09557629e-02
-9.53385085e-02 -1.07030380e+00 5.61601043e-01 1.24840713e+00
6.97360784e-02 -8.70360613e-01 -2.31610149e-01 3.89444590e-01
-2.34949157e-01 -1.08536708e+00 -2.09318489e-01 2.13463873e-01
-5.73753774e-01 -7.90541649e-01 -6.15591943e-01 -5.09098947e-01
5.97644866e-01 3.54444504e-01 1.51387787e+00 -1.41426265e-01
3.34241497e-03 7.97500968e-01 -6.48413181e-01 -8.34068596e-01
-4.19429809e-01 1.37568280e-01 -4.03476991e-02 -1.72004104e-01
1.09794986e+00 1.19709104e-01 -4.06113714e-01 3.76745015e-01
-1.06565452e+00 -1.77450120e-01 1.42436381e-02 8.34214807e-01
5.56973100e-01 -4.34310794e-01 8.54780316e-01 -1.51228440e+00
9.19470429e-01 -8.93319190e-01 -1.43704832e-01 2.84471899e-01
-8.45840573e-01 4.50542346e-02 1.43946990e-01 -4.45318222e-01
-1.39226961e+00 4.80036698e-02 -4.67283167e-02 -2.01137260e-01
-3.23582798e-01 6.42513990e-01 2.53929757e-02 3.21505517e-01
1.16111577e+00 -4.01199400e-01 -2.97919869e-01 -4.26820129e-01
8.80514801e-01 7.29609191e-01 4.76719104e-02 -6.21979773e-01
2.97421664e-01 5.08580387e-01 -4.10991371e-01 -3.37391943e-01
-1.38314104e+00 -7.88130760e-01 -3.87162447e-01 1.21060751e-01
1.24641633e+00 -1.20518625e+00 -6.73528835e-02 -6.24346100e-02
-1.12017250e+00 -3.11936319e-01 -6.86154008e-01 4.65173304e-01
-6.24491870e-01 9.26882103e-02 -5.47656536e-01 -1.02457917e+00
-2.33405769e-01 -7.91475475e-01 1.35640132e+00 2.68171966e-01
-7.02410161e-01 -1.42274475e+00 3.09967965e-01 4.50730413e-01
5.30250192e-01 -1.05780087e-01 6.86524153e-01 -1.47713470e+00
-2.73144275e-01 -4.49690491e-01 9.35684219e-02 2.46039882e-01
-6.02345057e-02 -2.91682422e-01 -1.14559293e+00 9.42994952e-02
-3.11761022e-01 -6.58992350e-01 8.63002717e-01 2.21465409e-01
8.83050084e-01 -3.71781260e-01 -3.59143049e-01 -2.07768381e-01
1.22691453e+00 -1.87567487e-01 4.84835118e-01 4.77508921e-03
8.30434024e-01 8.00096512e-01 6.90395474e-01 4.27422583e-01
7.00366974e-01 7.23083496e-01 1.30413920e-01 -2.41209447e-01
-1.18837193e-01 -5.54462850e-01 1.25248469e-02 2.66474962e-01
-1.08303852e-01 -8.05963695e-01 -9.52234447e-01 7.25261629e-01
-2.29764628e+00 -1.02736700e+00 -4.22985822e-01 1.62858438e+00
7.55644321e-01 -3.04203302e-01 3.53553265e-01 -2.93882340e-01
6.67896390e-01 -2.84562563e-03 -1.61923468e-01 -3.32491100e-02
-3.83141875e-01 9.81097817e-02 3.73817772e-01 4.64242160e-01
-1.14246762e+00 1.26114011e+00 7.72896814e+00 6.96126759e-01
-2.88496614e-01 6.78986669e-01 5.97933114e-01 -7.44884461e-02
-7.98973024e-01 3.38222057e-01 -1.19491136e+00 5.09037852e-01
7.97363162e-01 -1.59344316e-01 -2.12748155e-01 1.13461494e+00
-3.05675626e-01 -1.20994255e-01 -1.08449030e+00 6.65959120e-01
3.18176180e-01 -1.17466950e+00 1.33817911e-01 1.45616949e-01
1.10999560e+00 -8.95457715e-02 -1.38821185e-01 4.72876757e-01
1.40280092e+00 -8.58771026e-01 5.80773890e-01 5.70248842e-01
5.82097709e-01 -2.98817992e-01 8.93966079e-01 3.44284058e-01
-7.74576008e-01 -3.07770749e-03 -5.54187953e-01 -7.99416676e-02
3.68450791e-01 7.01353073e-01 -1.20950532e+00 3.82758856e-01
5.09740055e-01 3.38347971e-01 -5.89597821e-01 6.02561414e-01
-3.46611589e-01 8.64440382e-01 -1.48677438e-01 -4.17559355e-01
3.09774131e-01 4.79534566e-02 3.16304117e-01 1.45160401e+00
4.28443670e-01 7.78393447e-02 4.01210755e-01 8.71530950e-01
-2.86419839e-01 4.26470429e-01 -8.58399868e-01 -1.09429322e-01
4.13647771e-01 1.18988931e+00 -6.41541719e-01 -7.53715456e-01
-7.07409620e-01 6.90085053e-01 3.63473088e-01 5.26913464e-01
-5.89439690e-01 6.60038739e-02 4.00512159e-01 2.54518867e-01
-2.13281941e-02 -6.28823712e-02 -4.56002474e-01 -1.04302263e+00
-2.98741847e-01 -3.26324254e-01 7.61316180e-01 -1.24059010e+00
-1.67401230e+00 3.16454500e-01 1.61904678e-01 -8.80898654e-01
-9.29944456e-01 -4.41067368e-01 -1.75115213e-01 9.25324082e-01
-7.11856246e-01 -1.40940654e+00 -4.16419178e-01 5.45336246e-01
6.53146327e-01 -2.97424674e-01 1.11676896e+00 -1.33274134e-05
1.14865221e-01 3.35697323e-01 -2.28527069e-01 1.27902672e-01
9.22155321e-01 -1.70796096e+00 3.87942761e-01 7.19040275e-01
6.60849750e-01 6.22256994e-01 7.13588178e-01 -8.95676494e-01
-8.10635448e-01 -1.13589215e+00 1.17775333e+00 -1.18570542e+00
6.88163102e-01 -5.05568564e-01 -9.62394118e-01 1.10512197e+00
2.89459646e-01 1.04613431e-01 1.33981013e+00 6.04430616e-01
-4.43388641e-01 4.96182948e-01 -9.63530958e-01 2.82827884e-01
1.16453969e+00 -7.90317118e-01 -6.89482570e-01 6.58577979e-01
8.61765087e-01 -2.36089528e-01 -7.66190648e-01 1.78442672e-02
2.51791060e-01 -3.67883325e-01 8.57534528e-01 -1.30451155e+00
4.21039850e-01 -2.12867081e-01 -5.66653848e-01 -1.62649608e+00
-4.15461689e-01 -2.70634323e-01 -5.19164428e-02 1.42153466e+00
7.60093212e-01 -4.09328252e-01 8.77483726e-01 1.39105368e+00
-2.86443889e-01 -1.26089610e-03 -5.28643847e-01 -6.90548778e-01
2.12425500e-01 -3.59753937e-01 6.69646204e-01 1.18960595e+00
8.72402713e-02 5.85058868e-01 -3.09427649e-01 -3.12287122e-01
2.63394296e-01 -2.12713499e-02 9.07467425e-01 -1.50737345e+00
-1.34173512e-01 -2.76704431e-01 -4.09275442e-01 -7.91136801e-01
5.59995890e-01 -1.02972150e+00 2.12687522e-01 -1.95146406e+00
3.97902489e-01 -5.99205196e-01 -1.48842007e-01 8.85288656e-01
-7.07764089e-01 4.78948385e-01 3.27026378e-03 3.05091947e-01
-1.07806349e+00 3.13769311e-01 5.60034394e-01 -5.29985502e-02
5.83671685e-03 -3.12572449e-01 -1.18019986e+00 7.08343863e-01
6.06894732e-01 -6.93514228e-01 -6.12845123e-01 -6.82511985e-01
6.39532208e-01 -2.11665988e-01 4.26300108e-01 -4.94478494e-01
3.18123370e-01 -3.65308337e-02 3.26639861e-01 -2.13078186e-01
3.44908297e-01 -6.87252879e-01 3.98526698e-01 -2.33006313e-01
-9.43930686e-01 -5.86746559e-02 -1.13510072e-01 9.13681030e-01
-3.40834737e-01 -4.63736594e-01 2.31390968e-01 -2.11632341e-01
-9.61274624e-01 1.28397033e-01 -1.49648592e-01 2.78681099e-01
8.51605356e-01 -1.81969777e-01 -5.86325467e-01 -4.67087388e-01
-1.09736383e+00 -5.79425730e-02 4.57074612e-01 5.74956179e-01
2.09216297e-01 -1.62949669e+00 -8.98931682e-01 -4.00044292e-01
5.58171153e-01 -9.12093744e-03 2.46031657e-01 3.35865319e-01
2.18658581e-01 3.28143388e-01 1.72754124e-01 -5.65165997e-01
-1.08319104e+00 4.93393093e-01 -1.93315938e-01 -3.44884187e-01
-2.55452782e-01 9.14615929e-01 3.53507698e-01 -4.22011435e-01
-2.21674904e-01 1.56582311e-01 -2.31316522e-01 3.86729717e-01
5.25396049e-01 3.31169128e-01 1.94883555e-01 -5.32057822e-01
-2.28368774e-01 2.11627167e-02 8.76532719e-02 -5.04775703e-01
1.23328984e+00 -2.67484367e-01 3.46912861e-01 6.23910844e-01
9.07273173e-01 -8.27434883e-02 -1.08211100e+00 -6.04883373e-01
1.34875864e-01 -3.99530292e-01 -1.00753121e-01 -8.73598814e-01
-5.82028925e-01 7.34311402e-01 3.48967761e-01 6.74370289e-01
5.95940113e-01 4.12892878e-01 3.50228280e-01 3.01095635e-01
6.35139644e-01 -1.40517914e+00 2.12021977e-01 2.90432304e-01
6.57384098e-01 -1.55259597e+00 -3.34302783e-01 -3.24189007e-01
-1.11746383e+00 6.33942723e-01 4.81232047e-01 -1.58783376e-01
5.46085238e-01 4.62600827e-01 1.40099972e-02 -4.12849367e-01
-8.11557531e-01 -4.69424069e-01 3.63918126e-01 6.85146928e-01
9.55123782e-01 4.45891649e-01 -3.17944795e-01 9.53648388e-01
-1.95824593e-01 -1.21677615e-01 3.36153358e-01 9.05727029e-01
-5.42128563e-01 -1.03602993e+00 -6.54712170e-02 6.09831095e-01
-5.20412922e-01 -1.96878895e-01 -7.54661143e-01 5.54318130e-01
-2.25985125e-01 1.18819344e+00 1.08436674e-01 -9.29870307e-02
8.53006393e-02 6.57839775e-01 2.18944371e-01 -1.14968562e+00
-6.01709366e-01 -4.27158028e-02 5.24459004e-01 -3.80876005e-01
-8.35708857e-01 -6.66035235e-01 -1.20980752e+00 -2.43225764e-03
-4.73771691e-01 3.19780380e-01 9.52701092e-01 1.06162822e+00
4.67101663e-01 2.74112433e-01 -2.71450263e-02 -4.70361412e-01
-3.58804017e-01 -1.28320742e+00 -4.95304644e-01 8.54086816e-01
-3.57977509e-01 -7.21514404e-01 -1.30699456e-01 4.88223165e-01] | [10.797410011291504, 7.973968505859375] |
6a426b82-8522-4697-a200-36a22587175a | on-the-design-of-deep-priors-for-unsupervised | 2104.07161 | null | https://arxiv.org/abs/2104.07161v1 | https://arxiv.org/pdf/2104.07161v1.pdf | On the Design of Deep Priors for Unsupervised Audio Restoration | Unsupervised deep learning methods for solving audio restoration problems extensively rely on carefully tailored neural architectures that carry strong inductive biases for defining priors in the time or spectral domain. In this context, lot of recent success has been achieved with sophisticated convolutional network constructions that recover audio signals in the spectral domain. However, in practice, audio priors require careful engineering of the convolutional kernels to be effective at solving ill-posed restoration tasks, while also being easy to train. To this end, in this paper, we propose a new U-Net based prior that does not impact either the network complexity or convergence behavior of existing convolutional architectures, yet leads to significantly improved restoration. In particular, we advocate the use of carefully designed dilation schedules and dense connections in the U-Net architecture to obtain powerful audio priors. Using empirical studies on standard benchmarks and a variety of ill-posed restoration tasks, such as audio denoising, in-painting and source separation, we demonstrate that our proposed approach consistently outperforms widely adopted audio prior architectures. | ['Andreas Spanias', 'Jayaraman J. Thiagarajan', 'Vivek Sivaraman Narayanaswamy'] | 2021-04-14 | null | null | null | null | ['audio-denoising'] | ['audio'] | [ 5.38725078e-01 -2.08459675e-01 2.44476929e-01 -3.45091611e-01
-1.00059879e+00 -4.37148124e-01 3.51687640e-01 -1.30058378e-01
-2.14767337e-01 5.68945050e-01 4.53282267e-01 -1.10366359e-01
-4.37525004e-01 -5.16214073e-01 -6.76856577e-01 -7.09932268e-01
-1.06320441e-01 -1.33724824e-01 -1.41074523e-01 -2.48654574e-01
2.89543569e-02 3.82307321e-01 -1.46539211e+00 -1.40873238e-01
7.94328570e-01 1.06575465e+00 -4.79151420e-02 7.43810236e-01
3.99285614e-01 8.26823771e-01 -5.33716023e-01 -3.09574187e-01
3.40528339e-01 -3.55551332e-01 -6.41696930e-01 1.44208595e-01
4.23972547e-01 -4.28772420e-01 -6.05300844e-01 1.08085799e+00
7.62341917e-01 5.11217296e-01 5.44040084e-01 -8.52933168e-01
-6.49089038e-01 7.97664881e-01 -3.49887490e-01 3.01032633e-01
1.33507699e-01 2.72170901e-01 1.31534040e+00 -7.77495563e-01
1.51609957e-01 1.23156798e+00 1.04156613e+00 3.17844540e-01
-1.52959704e+00 -6.54466331e-01 1.44519985e-01 1.63806647e-01
-1.28167951e+00 -9.23125625e-01 1.23292279e+00 -2.92288154e-01
5.73918343e-01 1.81025520e-01 2.34042719e-01 1.14321482e+00
-1.66542903e-01 6.52220428e-01 8.47560585e-01 -3.56940925e-01
3.10604274e-01 -2.38076180e-01 -3.86344083e-02 1.20419525e-01
-1.21623263e-01 -3.00779827e-02 -5.34857810e-01 -4.89976332e-02
8.89455140e-01 -1.84780672e-01 -7.61647463e-01 6.01833127e-03
-9.04538870e-01 6.08911932e-01 5.37899792e-01 3.06161225e-01
-3.11876327e-01 5.38047194e-01 4.77063894e-01 3.99265915e-01
6.98777497e-01 4.53581959e-01 -3.12304944e-01 -1.48908868e-02
-1.07712257e+00 1.14225090e-01 3.60648602e-01 5.77897549e-01
4.82888907e-01 4.61163700e-01 -9.80473012e-02 1.07352257e+00
2.81881928e-01 2.59074965e-03 4.07560110e-01 -1.02905273e+00
3.30987751e-01 6.74241930e-02 8.50895420e-02 -1.03084934e+00
-2.48784393e-01 -8.67118180e-01 -1.19404101e+00 9.66118872e-02
4.36852843e-01 -1.37007907e-01 -9.75885153e-01 1.93611014e+00
1.16458364e-01 7.69548476e-01 7.32680261e-02 1.14149809e+00
5.81258237e-01 6.05543911e-01 -5.56898825e-02 6.57981113e-02
1.07669640e+00 -5.82543850e-01 -7.84640849e-01 -2.64336437e-01
5.80605753e-02 -1.05684757e+00 1.29585457e+00 6.12545848e-01
-1.18659878e+00 -7.94692457e-01 -1.30025768e+00 -3.28270346e-01
1.07302107e-01 2.78324127e-01 5.09122074e-01 6.03716731e-01
-1.13101256e+00 9.08003330e-01 -7.48482049e-01 -2.78953579e-03
4.88921672e-01 3.49428177e-01 -6.18309118e-02 -7.96489939e-02
-1.12414062e+00 4.08788830e-01 5.66753931e-02 5.05627990e-01
-1.34294629e+00 -8.36457968e-01 -6.86957896e-01 3.93359005e-01
2.41303623e-01 -6.47370279e-01 1.24404442e+00 -9.96169209e-01
-1.75856054e+00 4.69208151e-01 1.86200410e-01 -8.22814405e-01
4.75462168e-01 -6.21877909e-01 -2.86302626e-01 2.89540946e-01
-3.68062317e-01 5.43485999e-01 1.41965318e+00 -1.04862595e+00
-2.44333699e-01 -5.56182563e-02 4.20128763e-01 -9.41535756e-02
-6.46054626e-01 -4.82964404e-02 -2.37050220e-01 -1.22084677e+00
1.67667046e-01 -6.93137646e-01 -3.55127543e-01 -5.45556210e-02
-3.41712773e-01 -1.01753958e-02 5.34425199e-01 -7.79135108e-01
9.48656559e-01 -2.43197298e+00 2.97089159e-01 1.48813024e-01
1.28507867e-01 1.59120634e-01 -2.52902627e-01 2.77299494e-01
-2.68869013e-01 -2.65211426e-03 -5.76204896e-01 -5.66897213e-01
1.57123342e-01 2.02297270e-01 -7.49378920e-01 6.24527335e-01
4.63639885e-01 3.01707983e-01 -6.78975463e-01 2.50130109e-02
2.15583146e-01 1.01833510e+00 -1.03455377e+00 2.73730546e-01
-2.28719786e-01 7.37131774e-01 -9.24236625e-02 3.13393593e-01
6.34116828e-01 -5.56132384e-02 6.58442974e-02 -2.24523112e-01
1.62498608e-01 6.53117836e-01 -1.19578266e+00 1.93702066e+00
-7.47675896e-01 8.67661536e-01 5.87623775e-01 -1.23405707e+00
8.74583185e-01 6.75370097e-01 4.90306616e-01 -4.18004602e-01
2.29164600e-01 2.18402326e-01 -1.09320499e-01 -2.95420915e-01
4.03327703e-01 -2.83916652e-01 4.32567179e-01 3.43495816e-01
2.75892109e-01 -4.31014895e-02 7.66402185e-02 -9.30691138e-02
9.26446259e-01 1.06410109e-01 -2.57555455e-01 -2.59780526e-01
6.51479900e-01 -6.32447124e-01 6.31492615e-01 6.08722687e-01
4.35017832e-02 9.29012120e-01 2.89558291e-01 -9.28127542e-02
-7.44166255e-01 -9.38495278e-01 -1.82375461e-01 1.30760920e+00
-2.21374333e-01 -3.08685124e-01 -8.50978196e-01 -1.79753214e-01
-3.59116673e-01 3.03349912e-01 -2.91774273e-01 -1.72492027e-01
-7.26467907e-01 -6.82435870e-01 9.01247859e-01 5.77041805e-01
3.69008482e-01 -8.92182112e-01 -2.92683363e-01 4.46042091e-01
-2.17559099e-01 -1.12896729e+00 -5.95767319e-01 4.20391679e-01
-1.01880574e+00 -9.99308348e-01 -9.55125690e-01 -8.15194428e-01
5.97264707e-01 2.53035009e-01 1.03614140e+00 7.60622025e-02
-1.94149807e-01 4.23189521e-01 -2.95125306e-01 -4.57839780e-02
2.85425922e-03 7.37430006e-02 -1.41137913e-02 3.07826221e-01
-1.89117745e-01 -1.34234369e+00 -6.91572726e-01 2.09762856e-01
-1.34585714e+00 -3.04201752e-01 2.97346652e-01 9.12592053e-01
3.80822867e-01 3.70493054e-01 7.35504329e-01 -4.77812231e-01
8.54582489e-01 -2.49414727e-01 -5.30047715e-01 -1.50629029e-01
-2.30888695e-01 4.58156914e-02 8.18400621e-01 -5.39067328e-01
-1.08937705e+00 -1.36479676e-01 -4.96606708e-01 -5.48330545e-01
-1.49370104e-01 6.33310080e-01 -2.06711888e-01 3.11008282e-02
7.47364521e-01 -7.58629758e-03 -3.09643626e-01 -9.76578772e-01
3.87088031e-01 4.68395680e-01 8.63508463e-01 -8.33446741e-01
8.62156928e-01 6.16774738e-01 -1.79509953e-01 -8.94343674e-01
-8.94375920e-01 -2.93167621e-01 -4.31095988e-01 1.32748550e-02
5.96301675e-01 -1.01145971e+00 -5.29500425e-01 4.10928607e-01
-9.89818454e-01 -4.89568204e-01 -2.47729108e-01 5.65716565e-01
-4.81162369e-01 3.65624666e-01 -9.09356534e-01 -7.49218822e-01
-3.45824957e-01 -1.34419906e+00 8.88050199e-01 8.27514231e-02
-1.15101919e-01 -9.55747664e-01 -8.33296254e-02 2.20114633e-01
6.47965908e-01 7.60622546e-02 8.03584874e-01 -2.14759022e-01
-6.21691704e-01 2.19398104e-02 -2.63119727e-01 8.15142989e-01
5.45360819e-02 -1.74318522e-01 -1.53333366e+00 -4.36008275e-01
2.51035869e-01 -2.52467692e-01 1.13233769e+00 5.60370386e-01
1.37820268e+00 -2.71780103e-01 4.31463361e-01 1.00061929e+00
1.16465712e+00 -7.79881626e-02 6.90333247e-01 7.34007880e-02
5.04183650e-01 5.52712977e-01 8.28743353e-02 5.57419360e-01
-2.67579295e-02 4.78026569e-01 4.82063383e-01 -2.03907907e-01
-3.70174736e-01 -8.43327940e-02 3.93627107e-01 8.93594086e-01
-9.04726833e-02 -2.25185007e-02 -6.46243513e-01 7.24063635e-01
-1.65742075e+00 -6.70711815e-01 2.49327078e-01 2.06714272e+00
1.13019145e+00 2.14382961e-01 -8.57522115e-02 6.98055565e-01
5.88991582e-01 2.79214174e-01 -3.31886947e-01 7.18926191e-02
-7.56573305e-02 7.30815768e-01 6.34049922e-02 5.79269111e-01
-1.19654930e+00 6.23767614e-01 6.24896526e+00 9.07972813e-01
-1.24897933e+00 1.58153832e-01 6.29351139e-01 -8.90957639e-02
-3.66218835e-01 3.89954410e-02 -2.96132982e-01 9.21736509e-02
7.67813504e-01 1.55053556e-01 6.83982253e-01 4.96283829e-01
2.73119092e-01 3.10129762e-01 -1.21340501e+00 1.04414845e+00
-2.59749740e-01 -1.16457975e+00 -2.10527658e-01 -1.51272237e-01
6.18678629e-01 -2.11725473e-01 3.84579450e-01 1.66269645e-01
1.81166857e-01 -1.25655580e+00 7.94757187e-01 3.22041333e-01
6.08402848e-01 -9.10327256e-01 4.58148897e-01 -6.84510032e-03
-9.79570448e-01 -1.41012426e-02 -4.60865974e-01 -2.87609756e-01
5.18487617e-02 9.19513166e-01 -5.97648144e-01 4.12449300e-01
7.83867836e-01 8.90781939e-01 -2.07412854e-01 1.19715822e+00
-4.11985427e-01 1.14768684e+00 -3.22189569e-01 7.04995096e-01
3.58908802e-01 -1.76528275e-01 6.40531778e-01 1.28073549e+00
2.47799709e-01 -7.68894851e-02 5.72276972e-02 7.11286247e-01
-4.63049233e-01 -4.79540229e-02 -2.59043157e-01 1.11027313e-02
1.50365651e-01 1.11824143e+00 -5.92132628e-01 -1.50500797e-02
-1.71530336e-01 5.28019130e-01 5.65027073e-02 7.53343582e-01
-8.40184510e-01 -3.78398478e-01 8.94007087e-01 1.01639368e-01
4.28786457e-01 -3.81890267e-01 -2.90788084e-01 -1.06682241e+00
3.75083461e-02 -9.28571999e-01 2.51989514e-01 -7.28611648e-01
-1.39676476e+00 5.97490549e-01 -3.08218420e-01 -1.15701413e+00
-1.51251033e-01 -5.36569178e-01 -8.02328289e-01 7.69606233e-01
-1.86046672e+00 -8.77511024e-01 -1.58277154e-01 8.48058522e-01
4.50782090e-01 4.02850471e-02 6.20129704e-01 8.69955301e-01
-6.25881672e-01 4.98143882e-01 8.49943087e-02 1.26393706e-01
9.07521605e-01 -1.10272801e+00 3.10240984e-01 1.09411418e+00
3.69984627e-01 8.22731614e-01 9.75676954e-01 -1.80811554e-01
-1.33567929e+00 -1.07478619e+00 1.84911385e-01 1.78143173e-01
7.79857874e-01 -2.45655388e-01 -1.01014650e+00 5.86901665e-01
3.32291067e-01 1.61721349e-01 6.31739080e-01 2.69934654e-01
-5.26254296e-01 -3.91842842e-01 -8.95673454e-01 5.62196612e-01
8.92179847e-01 -9.33615983e-01 -3.36462945e-01 2.75455385e-01
5.59003413e-01 -2.93894559e-01 -8.09744060e-01 4.45070803e-01
3.29531878e-01 -8.72822821e-01 1.37963474e+00 -5.70949137e-01
6.26824915e-01 -3.09499592e-01 -2.86458164e-01 -1.26448715e+00
-2.11047038e-01 -1.05403125e+00 -9.09144059e-02 1.35533166e+00
1.26126766e-01 -4.42597806e-01 4.69346404e-01 2.29093894e-01
-3.84507835e-01 -5.64764500e-01 -9.53442156e-01 -5.27684748e-01
-2.49168444e-02 -8.41267109e-01 5.48571229e-01 9.03597713e-01
-4.01426166e-01 1.54897466e-01 -7.41078138e-01 3.64109546e-01
6.00195050e-01 -1.07630506e-01 6.48688257e-01 -1.23467207e+00
-6.24046624e-01 -4.66299087e-01 -1.25984579e-01 -1.29764426e+00
1.78595752e-01 -6.85038805e-01 2.78433949e-01 -1.22746265e+00
-3.72085959e-01 -4.79731888e-01 -6.78468347e-01 4.26311165e-01
-1.28311425e-01 6.31304801e-01 1.80530325e-02 1.71321422e-01
-2.80957580e-01 7.80015945e-01 1.08131123e+00 -2.75315940e-01
-2.40232959e-01 -1.21393511e-02 -8.26218426e-01 9.86077607e-01
8.77384186e-01 -3.52335244e-01 -5.49502671e-01 -8.27824593e-01
2.82751113e-01 4.54184879e-03 5.87020516e-01 -1.18196797e+00
9.00393128e-02 2.38143876e-01 2.14829683e-01 -2.56610572e-01
6.00317478e-01 -7.54840970e-01 2.20793914e-02 4.02685851e-02
-5.54167211e-01 -3.90172064e-01 2.46095613e-01 6.90536320e-01
-5.43827593e-01 -2.69170642e-01 8.74605060e-01 9.33805332e-02
-3.52848202e-01 1.91036493e-01 -1.97386160e-01 7.49883577e-02
2.83653498e-01 1.31969020e-01 1.36836931e-01 -6.73761010e-01
-8.41708481e-01 -7.67326578e-02 9.97480527e-02 2.50372589e-01
5.70815444e-01 -1.24275267e+00 -6.95230782e-01 1.26687393e-01
-3.76860619e-01 2.63351724e-02 3.49784195e-01 1.01601684e+00
-4.28703338e-01 1.30792499e-01 -6.56617731e-02 -5.81543386e-01
-8.85910809e-01 2.56819963e-01 2.84855783e-01 -1.59019127e-01
-8.95002186e-01 9.98139203e-01 3.15582663e-01 -1.83185309e-01
8.03596199e-01 -5.66139936e-01 -8.59927461e-02 9.66081396e-02
5.35709739e-01 3.46961319e-01 1.68474048e-01 -3.88029546e-01
-9.59521458e-02 2.88992345e-01 1.47785798e-01 -1.48312420e-01
1.66786110e+00 -1.17531553e-01 4.09967043e-02 1.43412486e-01
1.08768988e+00 1.70912147e-01 -1.69130325e+00 -3.45232934e-01
-6.85838461e-02 -3.44706923e-01 5.29914439e-01 -3.99193853e-01
-1.46446860e+00 1.12669468e+00 5.15998125e-01 3.21350604e-01
1.53666222e+00 -5.49774528e-01 9.38125908e-01 3.37407708e-01
3.71767618e-02 -9.64770436e-01 3.15571398e-01 4.11025584e-01
9.78543401e-01 -9.12659168e-01 -6.97215796e-02 -2.73273438e-01
-1.16848819e-01 1.10040128e+00 1.65725902e-01 -5.16541719e-01
7.08035886e-01 2.53635257e-01 1.45129740e-01 2.16933694e-02
-3.46425384e-01 -2.60246098e-01 4.49822426e-01 4.40515399e-01
7.13579237e-01 -2.89666325e-01 3.56739387e-02 5.96375942e-01
-2.18722239e-01 -1.36204928e-01 3.59826177e-01 6.71266615e-01
-2.89615810e-01 -1.08026695e+00 -6.47801161e-01 3.19379456e-02
-9.18018818e-01 -2.45348796e-01 -1.12472668e-01 3.74484718e-01
-2.96721384e-02 1.08413088e+00 -1.34295210e-01 -1.56064764e-01
7.90352151e-02 -1.75040087e-03 5.16180396e-01 -3.03001165e-01
-7.28898346e-01 6.61803722e-01 -1.48900850e-02 -2.73983032e-01
-6.30985379e-01 -1.74923599e-01 -9.39239264e-01 -1.18659995e-01
-2.86033541e-01 1.63661391e-02 3.52747232e-01 8.91467988e-01
1.91600084e-01 9.37268734e-01 4.84638542e-01 -1.16757393e+00
-6.75642312e-01 -1.08411825e+00 -5.06648898e-01 4.09778833e-01
6.79504991e-01 -5.81180811e-01 -4.48033869e-01 2.89590418e-01] | [15.359269142150879, 5.541706085205078] |
504a450c-e55f-4988-9de3-c5158eb7525d | canonicalizing-open-knowledge-bases | null | null | https://suchanek.name/work/publications/cikm2014.pdf | https://suchanek.name/work/publications/cikm2014.pdf | Canonicalizing Open Knowledge Bases | Open information extraction approaches have led to the creation of large knowledge bases from the Web. The problem with such methods is that their entities and relations are not canonicalized, leading to redundant and ambiguous facts. For example, they may store hBarack Obama, was born in, Honolului and hObama, place of birth, Honolului. In this paper, we present an approach based on machine learning methods that can canonicalize such Open IE triples, by clustering synonymous names and phrases.
We also provide a detailed discussion about the different signals, features and design choices that influence the quality of synonym resolution for noun phrases in Open IE KBs, thus shedding light on the middle ground between “open” and “closed” information extraction systems. | ['Geremy Heitz', 'Luis Galárraga'] | 2014-11-03 | null | null | null | null | ['open-information-extraction'] | ['natural-language-processing'] | [-2.47664034e-01 5.40605009e-01 -4.65916395e-01 -6.22217469e-02
-4.56642509e-01 -9.02013183e-01 4.19138759e-01 5.11208296e-01
-2.85346836e-01 1.45214474e+00 3.46587330e-01 -2.22002998e-01
-6.17251694e-01 -1.09129584e+00 -5.44049561e-01 -2.44313732e-01
-2.19438761e-01 9.90537107e-01 6.03121743e-02 -3.95773947e-01
-8.10812786e-02 1.35229573e-01 -1.57966721e+00 4.49928373e-01
1.06875968e+00 6.26947343e-01 -2.00655293e-02 8.38275626e-03
-6.37417316e-01 4.26632255e-01 -4.83517170e-01 -1.02457762e+00
1.60106540e-01 7.39293098e-02 -1.20345473e+00 -6.12099826e-01
3.61581177e-01 2.89921999e-01 -3.22942108e-01 1.09639001e+00
4.22872081e-02 -2.00086340e-01 8.48161459e-01 -1.10561800e+00
-6.65530086e-01 1.36064100e+00 -1.30571127e-01 -1.42809958e-03
7.09484100e-01 -6.43974721e-01 1.78025591e+00 -8.96749973e-01
1.26874781e+00 1.02484465e+00 5.06130517e-01 1.67819947e-01
-9.83073950e-01 -6.91775978e-01 -7.47675523e-02 5.22728384e-01
-1.88029099e+00 -9.26555619e-02 2.10609138e-01 -2.23920271e-01
1.14866030e+00 4.91481751e-01 6.86077416e-01 8.99662733e-01
2.76787449e-02 4.07739788e-01 8.17925930e-01 -8.41295004e-01
-5.15069403e-02 3.84121507e-01 4.29474562e-01 3.73715699e-01
1.44434106e+00 -3.71785998e-01 -5.04514337e-01 -5.09925425e-01
2.06639782e-01 -3.10501456e-01 -2.06481546e-01 -4.57052618e-01
-1.09487486e+00 8.13903511e-01 2.34270155e-01 7.97612965e-01
-1.58534974e-01 -3.62150192e-01 2.07204461e-01 1.77819371e-01
1.19409887e-02 1.14242578e+00 -8.54057729e-01 5.77024594e-02
-5.98112166e-01 3.54422897e-01 1.38698149e+00 1.35922945e+00
1.01698971e+00 -6.99372888e-01 4.39620435e-01 9.06691134e-01
2.41919115e-01 4.30109382e-01 3.24123591e-01 -5.92912853e-01
6.52960598e-01 8.64302218e-01 2.69300312e-01 -1.14288414e+00
-6.50224328e-01 -1.22696757e-01 -3.14863563e-01 -6.83138132e-01
3.29257637e-01 -3.22344542e-01 -6.80437624e-01 1.40832388e+00
6.52784556e-02 -3.85235339e-01 4.26523805e-01 5.50223947e-01
1.17043984e+00 3.71364862e-01 1.51720986e-01 -3.21380228e-01
1.74058390e+00 -3.58332723e-01 -1.06314588e+00 -2.97779679e-01
5.88857174e-01 -6.51369572e-01 3.94736528e-01 2.62771368e-01
-8.64063263e-01 7.45862275e-02 -1.11549485e+00 -5.91056831e-02
-1.12577319e+00 -1.98113650e-01 9.81758356e-01 5.38417935e-01
-3.61419559e-01 5.31294584e-01 -2.76530683e-01 -8.86479437e-01
-7.00886315e-03 4.93529201e-01 -5.92973590e-01 6.04647957e-02
-1.88936293e+00 1.24188793e+00 1.16771209e+00 -2.33716652e-01
1.50314525e-01 -6.44541502e-01 -7.98527479e-01 2.13716522e-01
7.49309599e-01 -6.85705483e-01 5.49235165e-01 -5.17879665e-01
-6.43942535e-01 8.13939035e-01 7.43153021e-02 -4.16166484e-01
-3.17419201e-01 -1.63544297e-01 -1.12543857e+00 6.58292919e-02
5.45612872e-01 4.94206548e-01 1.38942644e-01 -1.47358274e+00
-9.20081854e-01 -4.92407709e-01 2.70071149e-01 1.42433405e-01
-4.07043487e-01 2.22532481e-01 -4.00257617e-01 -3.72974366e-01
4.47216392e-01 -9.14822698e-01 4.13326025e-02 -1.01515305e+00
-5.52532971e-01 -2.40333855e-01 2.50047415e-01 -5.35983503e-01
1.81605184e+00 -1.87316501e+00 1.19643867e-01 6.93971336e-01
3.22026759e-01 -2.37175710e-02 3.60874772e-01 8.54155898e-01
-3.18017602e-01 6.47960365e-01 1.91965410e-06 6.51179254e-01
-1.47837037e-02 7.26363480e-01 -4.30853665e-01 -1.32369414e-01
-8.26452449e-02 6.93474472e-01 -9.99163628e-01 -8.27447534e-01
-3.19471091e-01 -7.55802616e-02 -2.76575923e-01 -5.02393246e-01
-2.18624234e-01 -1.24894075e-01 -5.52832305e-01 8.30227733e-01
4.32161897e-01 -8.27150419e-02 9.50516522e-01 -4.89617735e-01
-3.69036973e-01 6.41978323e-01 -1.42459834e+00 1.30339837e+00
-8.90298784e-02 3.93930495e-01 -2.17756703e-01 -6.29854500e-01
6.99865043e-01 5.16187012e-01 6.99131429e-01 -3.71430159e-01
1.89208716e-01 6.91514909e-01 5.93006471e-03 -5.31378448e-01
1.00832927e+00 -3.33906114e-02 -4.80862677e-01 -9.94730648e-03
4.25764322e-01 -2.24505097e-01 8.51845920e-01 3.80416840e-01
1.00440669e+00 2.00520270e-02 9.64925826e-01 -4.24942076e-01
4.02427614e-01 4.23961818e-01 8.93057168e-01 4.47438896e-01
3.23947430e-01 3.12787175e-01 6.07470334e-01 -4.27748859e-01
-1.15910506e+00 -1.05898893e+00 -7.75439680e-01 4.82159287e-01
3.12497526e-01 -1.06288445e+00 -4.12177116e-01 -6.92395031e-01
1.37780368e-01 7.62736082e-01 -3.88445199e-01 1.06151260e-01
-5.78530014e-01 -7.55951822e-01 6.42429292e-01 -7.82725401e-03
1.48590758e-01 -5.96372843e-01 -2.92432547e-01 1.46633923e-01
-9.01789308e-01 -1.24675190e+00 2.87817985e-01 6.87505603e-01
-5.89824677e-01 -1.08078766e+00 -2.11623475e-01 -5.37567258e-01
4.64556098e-01 -1.82506815e-01 1.54978502e+00 -1.18861526e-01
-1.85855940e-01 1.54209569e-01 -6.65279567e-01 -5.53664386e-01
-1.41684726e-01 6.14689767e-01 3.09781373e-01 -5.13919950e-01
9.34802055e-01 -5.51435530e-01 -5.08941989e-03 1.84832364e-01
-7.31047094e-01 -3.39948982e-01 5.43503106e-01 7.97380865e-01
3.73250246e-01 8.65623504e-02 6.15673065e-01 -1.25671697e+00
5.58803141e-01 -9.07086372e-01 -3.00847739e-01 8.54562879e-01
-9.28266585e-01 3.72746468e-01 1.45290047e-01 1.86385199e-01
-1.13647711e+00 -1.73371315e-01 1.51593849e-01 2.98089206e-01
-5.99420369e-02 9.90909278e-01 -4.44114000e-01 1.17217921e-01
7.81993270e-01 -4.63114202e-01 -6.88044965e-01 -4.31308478e-01
5.99264324e-01 8.03144574e-01 2.66426623e-01 -8.05861235e-01
6.97977662e-01 3.53556782e-01 -9.68753099e-02 -9.54318225e-01
-8.22777629e-01 -6.22811496e-01 -1.01652706e+00 2.08027899e-01
8.30349743e-01 -9.74334896e-01 -2.65283227e-01 -2.19126701e-01
-1.07985604e+00 6.44527018e-01 -3.15586030e-01 5.76534212e-01
-2.42547259e-01 4.05472815e-01 -3.13638091e-01 -4.18481886e-01
2.87124491e-03 -4.85378325e-01 4.09235805e-01 3.52901161e-01
-9.58290637e-01 -8.86882663e-01 2.86496818e-01 4.61564779e-01
-3.03796858e-01 -5.32968938e-02 1.40644491e+00 -1.24672949e+00
-5.49380302e-01 -1.72080174e-01 -1.92462921e-01 -1.70046791e-01
1.46435350e-01 5.71335629e-02 -5.15917361e-01 2.62710363e-01
-5.86703718e-01 -8.32319781e-02 4.44223404e-01 -8.72266889e-02
2.19017282e-01 -3.50514770e-01 -8.55395913e-01 2.42047518e-01
1.61484194e+00 2.70706117e-01 7.27151692e-01 6.72361553e-01
3.92895371e-01 6.40155315e-01 7.60379732e-01 4.09426779e-01
4.79935825e-01 5.30544877e-01 -5.22150286e-02 3.74055624e-01
2.58670181e-01 -3.11859041e-01 -1.50105610e-01 7.11746037e-01
-2.76169121e-01 -2.11581573e-01 -1.04938745e+00 7.59998262e-01
-1.74579918e+00 -8.86286974e-01 -2.41384476e-01 2.14837718e+00
1.18585062e+00 -2.88420729e-03 -2.82928258e-01 -9.06513408e-02
6.05235219e-01 -1.90784350e-01 1.17681347e-01 -2.91715324e-01
-5.91539681e-01 3.52108210e-01 8.07345629e-01 2.82874227e-01
-8.63288939e-01 9.90232587e-01 6.39882803e+00 6.76840723e-01
-3.81232709e-01 -2.17791125e-02 -1.54990405e-01 1.82405248e-01
-6.72716320e-01 6.33089483e-01 -1.20458388e+00 2.82518983e-01
8.69056761e-01 -3.12778205e-01 1.98373109e-01 6.17798984e-01
-4.23828423e-01 -4.66308445e-01 -1.02599275e+00 8.90101492e-01
1.37271315e-01 -1.34996939e+00 2.58729458e-01 1.80708766e-01
9.06910241e-01 -6.43941239e-02 -5.68638980e-01 2.94200122e-01
5.31421006e-01 -7.41267145e-01 5.22361517e-01 4.65138257e-01
5.13469458e-01 -7.11464167e-01 8.27696204e-01 2.36037038e-02
-1.06884146e+00 -2.90616602e-01 -4.53853190e-01 4.78554741e-02
1.30746916e-01 8.09550822e-01 -7.12129235e-01 1.11104751e+00
8.99122000e-01 2.51336128e-01 -5.23671031e-01 7.45472193e-01
-4.20440882e-01 2.04210877e-01 -4.73015189e-01 -2.34336287e-01
1.03333503e-01 -3.27748537e-01 6.42918050e-01 1.09400547e+00
3.02194357e-01 3.41981113e-01 -4.43866923e-02 7.93849111e-01
-1.23763300e-01 4.50371832e-01 -9.21268404e-01 -2.29280680e-01
8.32977712e-01 1.04296124e+00 -7.40139782e-01 -3.86715353e-01
-7.46922135e-01 3.67200136e-01 3.60763609e-01 3.65584224e-01
-6.46944642e-01 -7.30723977e-01 6.82248712e-01 3.97474356e-02
1.94637075e-01 -1.40011162e-01 -3.72277081e-01 -1.48220062e+00
-2.35182862e-03 -8.85093033e-01 9.39347386e-01 -4.99068826e-01
-1.17638803e+00 4.67368722e-01 4.34868217e-01 -9.42020953e-01
-2.57508367e-01 -6.72701716e-01 -1.35355471e-02 4.32173401e-01
-9.84433293e-01 -8.92939389e-01 3.35947782e-01 2.94593513e-01
-3.03845834e-02 -1.60912961e-01 1.00232482e+00 5.84449470e-01
-4.43299562e-01 3.49931419e-01 3.36004943e-01 3.27220440e-01
9.07352269e-01 -1.22625005e+00 5.09020537e-02 6.50326252e-01
5.89575469e-01 1.15517986e+00 7.30597734e-01 -1.08391619e+00
-1.03736424e+00 -4.36752319e-01 1.68547046e+00 -6.84914947e-01
9.18098032e-01 -2.77522564e-01 -6.46175385e-01 7.89558053e-01
3.72760653e-01 -4.50767756e-01 1.09616113e+00 9.63683605e-01
-6.27576292e-01 -4.80896415e-04 -1.01178741e+00 6.89546347e-01
1.03359592e+00 -3.98426414e-01 -1.17431319e+00 3.23562443e-01
5.15749812e-01 -1.77289486e-01 -1.25696230e+00 4.61785585e-01
5.72852910e-01 -5.59812665e-01 9.79080915e-01 -8.85090888e-01
1.83883905e-01 -1.65374130e-01 -1.98908314e-01 -1.22964108e+00
-3.68104666e-01 -4.27371204e-01 -6.20945916e-02 1.24757636e+00
1.33080852e+00 -6.30739808e-01 3.72280061e-01 9.79910374e-01
1.10531159e-01 -3.48524243e-01 -1.02714229e+00 -7.40022600e-01
1.51602224e-01 -3.74243557e-01 7.12347507e-01 1.20930386e+00
1.01156950e+00 7.30621457e-01 -1.74182579e-01 3.13667916e-02
2.28320330e-01 3.05661440e-01 3.70959669e-01 -1.63292944e+00
1.81580827e-01 5.21284044e-02 -4.79831040e-01 -5.09158254e-01
1.54043108e-01 -9.55000579e-01 -4.08369273e-01 -1.49762774e+00
1.87980607e-01 -7.82561004e-01 -8.69873315e-02 4.61302519e-01
3.88926715e-02 -1.00225760e-02 6.87828958e-02 2.90038466e-01
-6.10270858e-01 -1.30967215e-01 7.15678871e-01 -8.74860361e-02
-7.83564523e-02 -4.34564650e-01 -9.93070722e-01 8.84877264e-01
5.70482016e-01 -5.44488072e-01 -3.55581604e-02 -3.75881493e-01
1.15292609e+00 -1.63816586e-01 -1.54700899e-03 -8.27929914e-01
2.68412501e-01 -2.75293022e-01 9.49650183e-02 -5.70969522e-01
1.73954546e-01 -1.13382351e+00 5.40611267e-01 1.45149350e-01
-3.71165574e-02 -6.58940673e-02 -9.16690193e-03 3.08772266e-01
-4.43232328e-01 -8.68384600e-01 3.27613771e-01 -5.26661098e-01
-8.94573748e-01 -7.61616156e-02 -2.98143029e-01 4.20601934e-01
9.74118054e-01 -1.67922422e-01 -3.33508164e-01 8.92174989e-02
-1.04268062e+00 2.37069353e-01 3.20697010e-01 5.12252212e-01
2.13909373e-01 -1.29906106e+00 -5.10187745e-01 -1.77946150e-01
4.34033781e-01 -3.19612652e-01 -7.27716684e-02 5.20561874e-01
-4.44524020e-01 9.85999942e-01 -2.05314845e-01 1.00417309e-01
-9.87627685e-01 7.22714543e-01 -7.79134706e-02 -5.38230777e-01
-4.10379738e-01 4.60827082e-01 -2.32043803e-01 -4.13098097e-01
-9.03801098e-02 -2.91362643e-01 -5.65534651e-01 7.17584729e-01
7.09525049e-02 3.88383657e-01 3.02637428e-01 -5.75135708e-01
-5.08718014e-01 2.79493809e-01 -1.09235182e-01 -1.22085493e-02
9.96149659e-01 -4.79470819e-01 -7.15547502e-01 4.82998312e-01
8.26272607e-01 4.85542357e-01 3.58809501e-01 -1.39234111e-01
6.29843056e-01 -3.06428343e-01 -5.52336156e-01 -7.60592461e-01
-5.14281154e-01 5.95014393e-02 1.37009680e-01 4.77333814e-01
5.49562693e-01 3.72389257e-01 7.95227289e-01 9.34499443e-01
8.41653228e-01 -1.46902001e+00 -9.62324798e-01 8.87398601e-01
6.30354404e-01 -1.03715742e+00 2.64920503e-01 -9.06717777e-01
-4.80842352e-01 1.24373245e+00 4.02532518e-01 2.88610846e-01
7.29416907e-01 2.65062779e-01 -1.19694762e-01 -5.74642539e-01
-5.90602875e-01 -4.57434475e-01 1.69036686e-01 4.55700994e-01
5.64432621e-01 3.43469054e-01 -1.16006446e+00 1.11449039e+00
-2.97302544e-01 -2.12130591e-01 5.97470760e-01 8.04484427e-01
-3.73288482e-01 -1.34866571e+00 -5.40092587e-01 7.18417645e-01
-5.84653616e-01 -5.10418355e-01 -7.70082593e-01 1.01630890e+00
6.91642761e-01 8.01383913e-01 -9.05913711e-02 -2.85663277e-01
2.81454057e-01 6.00010633e-01 3.40410113e-01 -7.61005521e-01
-4.02016193e-01 -3.05251390e-01 9.37763214e-01 -7.36835226e-02
-3.97194803e-01 -8.58530521e-01 -1.48829341e+00 -5.40242419e-02
-8.17721605e-01 8.74790788e-01 3.35292459e-01 1.07245743e+00
3.70765060e-01 8.18072930e-02 1.37418970e-01 9.99891460e-02
-1.25779524e-01 -7.99297452e-01 -7.30156958e-01 2.33658060e-01
-3.87341350e-01 -1.02867889e+00 -1.64741933e-01 -5.37275113e-02] | [9.321702003479004, 8.427739143371582] |
153203ee-416e-45a2-9427-3ef433b6da60 | policy-representation-via-diffusion | 2305.13122 | null | https://arxiv.org/abs/2305.13122v1 | https://arxiv.org/pdf/2305.13122v1.pdf | Policy Representation via Diffusion Probability Model for Reinforcement Learning | Popular reinforcement learning (RL) algorithms tend to produce a unimodal policy distribution, which weakens the expressiveness of complicated policy and decays the ability of exploration. The diffusion probability model is powerful to learn complicated multimodal distributions, which has shown promising and potential applications to RL. In this paper, we formally build a theoretical foundation of policy representation via the diffusion probability model and provide practical implementations of diffusion policy for online model-free RL. Concretely, we character diffusion policy as a stochastic process, which is a new approach to representing a policy. Then we present a convergence guarantee for diffusion policy, which provides a theory to understand the multimodality of diffusion policy. Furthermore, we propose the DIPO which is an implementation for model-free online RL with DIffusion POlicy. To the best of our knowledge, DIPO is the first algorithm to solve model-free online RL problems with the diffusion model. Finally, extensive empirical results show the effectiveness and superiority of DIPO on the standard continuous control Mujoco benchmark. | ['Zhouchen Lin', 'Binbin Zhou', 'Shiting Wen', 'Cong Fang', 'Yiming Yang', 'Yucun Zhong', 'Fenghao Lei', 'Zhixiong Huang', 'Long Yang'] | 2023-05-22 | null | null | null | null | ['continuous-control'] | ['playing-games'] | [-4.54414040e-01 -6.04367666e-02 -8.46378803e-01 3.87840420e-01
-5.79511762e-01 -6.56515837e-01 6.33313835e-01 -7.03039020e-02
-4.81332332e-01 1.10163355e+00 1.71861112e-01 -7.50637233e-01
-3.75344634e-01 -6.53745174e-01 -6.99204862e-01 -1.08263624e+00
-4.29679304e-01 6.02901280e-01 -2.32828259e-01 -3.84012818e-01
1.75575376e-01 3.53083611e-01 -1.21239805e+00 -2.02445492e-01
8.97580087e-01 1.01455212e+00 3.90968025e-01 6.53098166e-01
-2.17262909e-01 1.21472967e+00 -6.98932946e-01 6.08957075e-02
3.24266404e-01 -5.45861006e-01 -5.49316406e-01 -4.90556546e-02
-3.23363632e-01 -7.54298329e-01 -5.24479628e-01 1.19072866e+00
7.50557899e-01 3.66103739e-01 7.79550910e-01 -1.40838504e+00
-9.28538382e-01 9.89061475e-01 -6.82667196e-01 2.12815329e-02
3.62629712e-01 3.21248055e-01 9.89116371e-01 -4.16320413e-01
5.15552044e-01 1.54815495e+00 1.84126303e-01 7.58684874e-01
-9.08897758e-01 -3.97444397e-01 5.48492968e-01 -2.58040871e-03
-9.12666380e-01 1.56215802e-01 6.81895375e-01 -1.56409040e-01
6.05973601e-01 2.92393211e-02 9.47759926e-01 1.31722057e+00
3.20704281e-01 1.66762114e+00 1.60096657e+00 -4.37659532e-01
7.67087162e-01 -3.37396324e-01 -4.35986072e-01 7.46667922e-01
1.69059843e-01 5.82712650e-01 -3.20916682e-01 -4.47965980e-01
1.27807653e+00 1.89891947e-03 -1.37295261e-01 -6.13248229e-01
-1.11506414e+00 1.04372048e+00 1.29765376e-01 -2.27734614e-02
-5.92731357e-01 5.13759315e-01 1.81519493e-01 6.80210948e-01
2.47426286e-01 2.77337015e-01 -3.37933928e-01 -5.91130733e-01
-3.90482038e-01 4.32407469e-01 1.06535375e+00 9.20879066e-01
3.59083980e-01 3.67424279e-01 -5.98984182e-01 6.64569676e-01
4.53587383e-01 1.10516500e+00 4.62028623e-01 -1.45594394e+00
3.02576870e-01 -7.36580491e-02 6.56843364e-01 -5.24470389e-01
-2.15900511e-01 -2.56257713e-01 -7.11760819e-01 4.37252492e-01
6.19249940e-01 -6.75853848e-01 -5.03090322e-01 1.75834894e+00
3.30110341e-01 -2.09847242e-02 3.18716407e-01 7.94993043e-01
2.89128739e-02 7.87706017e-01 -9.94837135e-02 -6.44198239e-01
6.84148908e-01 -1.03353500e+00 -9.87146676e-01 1.74853206e-01
4.23115313e-01 -4.49655384e-01 1.35327744e+00 5.40443480e-01
-1.13065553e+00 3.51651907e-02 -5.88779390e-01 6.15868092e-01
-8.98002684e-02 -9.74284485e-03 8.97219360e-01 6.24229193e-01
-1.01931942e+00 6.46219492e-01 -8.34227443e-01 -1.49434164e-01
4.00752842e-01 8.69870335e-02 4.11559165e-01 -9.86057594e-02
-1.10403943e+00 8.87036264e-01 3.56384516e-01 -1.26765564e-01
-1.32792866e+00 -4.02825415e-01 -5.70983708e-01 3.84184346e-03
9.41501439e-01 -5.00086546e-01 1.77551055e+00 -8.71881843e-01
-2.43119144e+00 -5.46421222e-02 4.05088216e-02 -5.64392209e-01
7.99095571e-01 -1.29119486e-01 -1.04858831e-01 2.24343091e-01
-9.20123756e-02 4.49584126e-01 1.09498799e+00 -1.38175142e+00
-7.19862461e-01 7.13968873e-02 3.30638707e-01 4.65462267e-01
-2.51444250e-01 -3.40713948e-01 -8.49090070e-02 -7.83612192e-01
-6.04991436e-01 -7.67130733e-01 -4.80858296e-01 -1.62471980e-01
-8.92278180e-02 -5.17587900e-01 5.22911310e-01 -1.59718931e-01
1.25014126e+00 -2.02985001e+00 3.64272386e-01 4.02256221e-01
8.57986286e-02 3.39371189e-02 -3.69999409e-01 9.38535511e-01
5.40866554e-01 1.50356099e-01 -8.46623480e-02 -6.57972544e-02
8.10350716e-01 7.40419388e-01 -7.47668505e-01 5.19168019e-01
-4.37042981e-01 1.16169727e+00 -1.39064419e+00 -2.83525556e-01
1.42833337e-01 1.03968494e-02 -5.03917694e-01 2.04353705e-01
-6.66515172e-01 4.44289565e-01 -7.24232495e-01 7.73175597e-01
4.07739341e-01 -1.08762644e-01 3.05282414e-01 4.70746607e-01
-5.29166833e-02 -3.29410940e-01 -1.13968921e+00 1.47702098e+00
-4.02547717e-01 1.53073967e-01 3.21551234e-01 -7.62466788e-01
7.36082911e-01 2.32780680e-01 8.78461838e-01 -7.37217844e-01
1.80380717e-01 1.92862555e-01 -2.26462290e-01 -2.86666989e-01
4.74564642e-01 -6.45280927e-02 -7.83599988e-02 8.64686847e-01
-6.10744320e-02 -2.16674387e-01 4.03527081e-01 8.24405700e-02
7.86964834e-01 2.94338554e-01 2.94857770e-01 -3.54557484e-01
1.26954630e-01 -2.59357244e-01 3.87444019e-01 1.40415585e+00
-4.51109797e-01 -2.04808727e-01 7.73107648e-01 -1.45329282e-01
-6.13623500e-01 -1.14103949e+00 4.03140068e-01 1.31448805e+00
2.56069839e-01 -3.24082196e-01 -3.18117559e-01 -8.86995077e-01
4.28855985e-01 7.21467614e-01 -7.19621181e-01 -7.49610364e-02
-2.97555178e-01 -5.77848554e-01 5.27569592e-01 4.37576264e-01
4.77846056e-01 -1.11420310e+00 -5.08246541e-01 3.35512072e-01
-4.49856520e-02 -6.20141327e-01 -6.91232383e-01 -9.07826647e-02
-7.53918648e-01 -1.09522676e+00 -1.16179121e+00 -4.68143404e-01
3.07013899e-01 4.00857255e-02 7.40586221e-01 -3.36320102e-01
2.57270813e-01 1.26436341e+00 -3.22395653e-01 -5.68566620e-01
-4.15688038e-01 -2.15107277e-01 1.71906874e-01 -9.33337212e-02
-1.53515130e-01 -2.17929706e-01 -6.97586298e-01 2.61568159e-01
-9.61714625e-01 -5.22793233e-01 6.05211735e-01 1.03489161e+00
6.78209722e-01 1.15563758e-01 6.27042711e-01 -3.64969105e-01
1.57012093e+00 -4.91275281e-01 -9.14218783e-01 3.05338711e-01
-6.78406954e-01 3.02792668e-01 7.10701048e-01 -9.69546020e-01
-1.09178066e+00 -2.30028644e-01 4.10028435e-02 -6.64499044e-01
1.92404777e-01 5.91288507e-01 3.19446892e-01 -4.88416031e-02
3.65165740e-01 4.45117831e-01 3.85952681e-01 -3.11698228e-01
8.89967203e-01 2.57119417e-01 4.41552140e-02 -1.29207158e+00
4.04462606e-01 6.67837679e-01 6.86263805e-03 -6.87874496e-01
-7.10936010e-01 -1.75178364e-01 1.22912519e-01 -3.76897126e-01
4.97347385e-01 -5.52998722e-01 -1.33862364e+00 4.33614373e-01
-7.51051307e-01 -1.13147879e+00 -8.56353760e-01 5.81960440e-01
-1.27430618e+00 4.03439224e-01 -8.24627757e-01 -1.34760141e+00
-1.01352088e-01 -9.98548627e-01 7.53042459e-01 2.84382135e-01
3.81632447e-01 -1.37648380e+00 3.31498355e-01 -4.12711948e-01
6.58851802e-01 7.18840724e-03 5.60616314e-01 -1.09953634e-01
-3.19112837e-01 4.09052134e-01 2.13628903e-01 2.52421945e-01
-2.77857471e-05 -1.41280428e-01 -3.98585349e-01 -6.72355175e-01
1.20820761e-01 -5.66353500e-01 7.32569337e-01 6.84534967e-01
1.06627035e+00 -5.79178631e-01 -6.02583066e-02 3.64877552e-01
1.44200897e+00 4.31601763e-01 3.30038667e-01 4.42877382e-01
2.42596582e-01 1.65940791e-01 7.93556988e-01 1.12104785e+00
4.86111879e-01 6.26501739e-02 5.09925306e-01 3.96682285e-02
2.52329826e-01 -7.29747176e-01 8.91298771e-01 7.68333852e-01
-1.92582250e-01 -2.58629978e-01 -5.53745866e-01 1.55680835e-01
-2.21857762e+00 -1.13163352e+00 5.76506436e-01 2.03566909e+00
9.46757853e-01 -1.90775082e-01 5.13275743e-01 -3.27644825e-01
4.66269433e-01 2.26614520e-01 -8.48943830e-01 -4.66450334e-01
-1.84795871e-01 1.51550576e-01 8.01865458e-01 6.65585816e-01
-8.68966401e-01 1.12409890e+00 7.59377050e+00 1.32203889e+00
-1.00405884e+00 4.56485637e-02 3.00329059e-01 -1.94610842e-02
-2.79655308e-01 -1.81618363e-01 -5.83932638e-01 4.55687493e-01
4.41898823e-01 -4.77242053e-01 1.04280174e+00 9.15122032e-01
4.23923731e-01 -1.93578020e-01 -7.55108774e-01 1.05882299e+00
-4.22172785e-01 -1.30145073e+00 4.15553637e-02 2.85861701e-01
1.04153967e+00 -7.96716195e-03 4.09179956e-01 6.75485373e-01
1.14525032e+00 -1.04887891e+00 7.36526966e-01 7.09648371e-01
5.17770827e-01 -9.63624418e-01 4.26672161e-01 7.10529804e-01
-9.47480619e-01 -6.72483683e-01 -6.79659605e-01 -1.58446491e-01
8.54753181e-02 -1.01534966e-02 -4.01748747e-01 5.48806012e-01
4.25916612e-01 7.64680147e-01 1.23255931e-01 1.02327204e+00
-4.71983045e-01 7.64041305e-01 -3.43626887e-01 -5.79871595e-01
6.58845246e-01 -4.16429371e-01 6.74783349e-01 8.07520986e-01
3.25574040e-01 -1.37759075e-01 6.87521815e-01 7.39636660e-01
1.01693273e-01 1.90231487e-01 -7.53264904e-01 -4.14157420e-01
4.82728362e-01 8.03892612e-01 -4.01321679e-01 -2.54088730e-01
-2.06938535e-01 7.60786057e-01 5.26792407e-01 9.04537201e-01
-8.03311348e-01 2.28467342e-02 4.93875951e-01 -4.99621093e-01
5.39412081e-01 -5.41042387e-01 3.40045571e-01 -1.17377710e+00
-2.41705790e-01 -1.11572766e+00 3.35046023e-01 -3.73390168e-01
-1.65420258e+00 -5.78964017e-02 2.01583788e-01 -1.05123603e+00
-3.00764203e-01 -5.71027040e-01 -5.10655463e-01 4.16522503e-01
-1.67574692e+00 -7.61495233e-01 4.32356596e-01 8.63003731e-01
3.93275440e-01 -3.58659118e-01 6.59391046e-01 -2.19754145e-01
-3.98718476e-01 3.54789466e-01 8.88464093e-01 -1.95507482e-01
5.51721811e-01 -1.69245577e+00 -4.24247324e-01 3.05925280e-01
-1.39124453e-01 5.19197941e-01 6.63065791e-01 -6.70487106e-01
-2.10934901e+00 -7.74888575e-01 -1.44925475e-01 5.86425699e-02
1.11762381e+00 3.78398946e-03 -4.23596799e-01 5.65593243e-01
4.34190184e-01 -2.98830181e-01 3.88339520e-01 -2.06214905e-01
1.10874206e-01 2.01240182e-02 -1.03665853e+00 9.59770739e-01
8.28760922e-01 -2.64359534e-01 -4.24935162e-01 4.69063371e-01
5.88818312e-01 -4.34120446e-01 -1.02719104e+00 -6.35650679e-02
4.24981713e-01 -6.82128906e-01 8.30036163e-01 -5.40907621e-01
-9.14685875e-02 -1.45694211e-01 -1.32247686e-01 -1.68240917e+00
-1.33900791e-01 -1.46463692e+00 -8.51326942e-01 8.73061776e-01
2.15696096e-02 -1.01471519e+00 4.45669442e-01 -1.46305012e-02
3.31792533e-01 -1.03463626e+00 -8.49054635e-01 -1.26832044e+00
7.50600934e-01 -2.93044567e-01 7.13298261e-01 7.10650206e-01
1.61265150e-01 -1.68594226e-01 -6.03688180e-01 -1.97659910e-01
8.10860634e-01 1.89132735e-01 6.12062633e-01 -5.89434981e-01
-7.33415723e-01 -9.10304904e-01 5.51164746e-01 -1.74878955e+00
3.02472115e-01 -6.52128458e-01 -1.04001276e-01 -1.68179834e+00
-2.23612249e-01 -5.13037980e-01 -4.20129567e-01 2.33441889e-01
3.07513420e-02 -5.79439342e-01 4.30743814e-01 3.41471970e-01
-9.08230543e-01 1.24356830e+00 2.01433301e+00 -1.69484481e-01
-3.86104554e-01 1.26822188e-01 -6.17438316e-01 6.33034945e-01
1.28876460e+00 -2.73991644e-01 -8.14694047e-01 -2.24509388e-01
2.55773395e-01 4.41136181e-01 9.52821877e-03 -3.43303591e-01
3.01053703e-01 -7.99144864e-01 3.09862737e-02 -4.62711215e-01
1.33438617e-01 -5.88893890e-01 -5.18640935e-01 7.28405416e-01
-4.51067328e-01 1.53848633e-01 9.03717354e-02 1.06445825e+00
3.76313063e-03 2.15587318e-02 5.12454748e-01 -1.07771724e-01
-6.72180831e-01 5.67991734e-01 -7.74063647e-01 3.23874950e-01
1.17869711e+00 3.36544007e-01 -3.48598838e-01 -8.71590734e-01
-7.99751461e-01 6.57949984e-01 3.35602850e-01 1.13309100e-01
6.13687515e-01 -1.40736914e+00 -4.37804878e-01 -1.52290925e-01
-4.38422918e-01 -4.86173809e-01 -7.78586939e-02 9.01129663e-01
-2.48286068e-01 1.83085144e-01 -6.13144077e-02 -2.65065283e-01
-4.98358995e-01 8.45462263e-01 4.22503769e-01 -5.31374633e-01
-6.35162175e-01 3.50070357e-01 -7.19990730e-02 -3.61619920e-01
5.60241044e-01 -4.52430189e-01 -1.87741682e-01 -2.28833035e-01
4.78527129e-01 6.02932274e-01 -8.53318155e-01 7.97691569e-02
5.80926687e-02 3.00751954e-01 2.03045920e-01 -6.61629796e-01
9.92768168e-01 -1.26334846e-01 -7.85172433e-02 5.26076496e-01
4.93621141e-01 -8.15192796e-03 -1.76035237e+00 -1.89054087e-01
-5.14354557e-03 -4.27623987e-01 -1.12329192e-01 -8.67789507e-01
-9.03783500e-01 6.15032613e-01 4.45785791e-01 4.42299336e-01
7.97572315e-01 -1.66531458e-01 5.85263491e-01 7.27000594e-01
6.29844427e-01 -1.60015833e+00 2.87244439e-01 9.31232154e-01
1.03616500e+00 -8.98858607e-01 -1.97790548e-01 2.25642949e-01
-1.02898467e+00 1.06612158e+00 3.35345775e-01 -3.28845710e-01
7.03361452e-01 3.39937776e-01 -1.41454563e-01 1.67620048e-01
-9.39913750e-01 -5.65111578e-01 -1.42041743e-01 8.41166019e-01
-1.39109030e-01 3.72854739e-01 -5.09218395e-01 4.10578161e-01
1.23225398e-01 1.59688249e-01 3.19573343e-01 1.32007921e+00
-5.97540140e-01 -1.32297575e+00 -4.44074363e-01 1.34547859e-01
-3.00442964e-01 1.29046291e-01 -2.47981027e-01 8.14840972e-01
-5.31163692e-01 1.06221294e+00 -2.62664825e-01 8.04911181e-02
6.08003698e-02 -2.34590441e-01 8.82822096e-01 5.03799655e-02
-2.93005556e-01 3.99173826e-01 -1.51074871e-01 -6.95753455e-01
-2.74867773e-01 -5.48127651e-01 -1.23969293e+00 -4.80554968e-01
9.52892601e-02 3.00137013e-01 4.16795522e-01 7.72360682e-01
3.31640542e-01 3.12264293e-01 7.55739331e-01 -6.70517147e-01
-1.52644551e+00 -6.98790729e-01 -9.00468647e-01 4.88745905e-02
3.51456314e-01 -1.03257108e+00 -2.59483457e-01 -6.63961947e-01] | [4.072254657745361, 2.273719549179077] |
6b28d2df-8839-45ba-99c2-1f13f5599879 | fpcd-an-open-aerial-vhr-dataset-for-farm-pond | 2302.14554 | null | https://arxiv.org/abs/2302.14554v1 | https://arxiv.org/pdf/2302.14554v1.pdf | FPCD: An Open Aerial VHR Dataset for Farm Pond Change Detection | Change detection for aerial imagery involves locating and identifying changes associated with the areas of interest between co-registered bi-temporal or multi-temporal images of a geographical location. Farm ponds are man-made structures belonging to the category of minor irrigation structures used to collect surface run-off water for future irrigation purposes. Detection of farm ponds from aerial imagery and their evolution over time helps in land surveying to analyze the agricultural shifts, policy implementation, seasonal effects and climate changes. In this paper, we introduce a publicly available object detection and instance segmentation (OD/IS) dataset for localizing farm ponds from aerial imagery. We also collected and annotated the bi-temporal data over a time-span of 14 years across 17 villages, resulting in a binary change detection dataset called \textbf{F}arm \textbf{P}ond \textbf{C}hange \textbf{D}etection Dataset (\textbf{FPCD}). We have benchmarked and analyzed the performance of various object detection and instance segmentation methods on our OD/IS dataset and the change detection methods over the FPCD dataset. The datasets are publicly accessible at this page: \textit{\url{https://huggingface.co/datasets/ctundia/FPCD}} | ['G. Sivakumar', 'Om Damani', 'Rajiv Kumar', 'Chintan Tundia'] | 2023-02-28 | null | null | null | null | ['change-detection'] | ['computer-vision'] | [ 2.25193977e-01 -2.28043124e-01 1.10710062e-01 -3.89058709e-01
-1.82820722e-01 -8.61214578e-01 6.72101080e-01 5.45810103e-01
-3.26241851e-01 5.83115101e-01 -1.28552422e-01 -5.05184889e-01
-2.72342831e-01 -1.25669444e+00 -7.33673871e-01 -7.46193945e-01
-6.21804118e-01 3.14611048e-01 4.32960242e-01 -2.89695591e-01
-4.09812480e-02 8.28564227e-01 -1.37917125e+00 -5.06581292e-02
8.32627594e-01 7.21428812e-01 5.82125485e-01 9.15919662e-01
5.77464961e-02 -1.34032983e-02 -2.21862599e-01 -3.70001793e-02
6.32272899e-01 -2.86051124e-01 -9.12886441e-01 1.26523823e-01
4.06160265e-01 -6.67981505e-01 -2.48059332e-01 1.26855242e+00
2.88891554e-01 -1.16382591e-01 7.80057430e-01 -1.03825808e+00
-1.89593330e-01 5.94289362e-01 -1.14029217e+00 7.31103063e-01
-1.22460440e-01 2.90773124e-01 6.74318552e-01 -6.90377712e-01
8.51567328e-01 1.09569120e+00 7.67679870e-01 -3.94279122e-01
-1.16478813e+00 -6.72209203e-01 3.07078034e-01 2.02706810e-02
-1.44138896e+00 -3.04705262e-01 2.10911885e-01 -7.14222550e-01
7.67138839e-01 5.37232578e-01 8.46398175e-01 1.59628838e-01
2.24800974e-01 7.78071344e-01 1.04036093e+00 -2.89341658e-01
-4.04297411e-02 -3.63232106e-01 3.26370358e-01 6.65613592e-01
3.32407027e-01 -1.22426553e-02 2.06828222e-01 -9.15262327e-02
1.05465150e+00 3.12883019e-01 -3.34238559e-01 6.64246082e-02
-1.22562480e+00 6.35546505e-01 7.86907971e-01 4.14422393e-01
-6.72722876e-01 5.01715429e-02 2.24821180e-01 3.47122967e-01
6.02321684e-01 1.17612988e-01 -8.91795397e-01 2.32757881e-01
-1.07039726e+00 2.89796501e-01 5.72530687e-01 9.91589487e-01
1.26046383e+00 -3.45502198e-01 7.94484988e-02 7.35235095e-01
2.02043712e-01 8.00851107e-01 9.92873237e-02 -7.10209310e-01
5.39757073e-01 8.07544887e-01 4.40537423e-01 -1.08231783e+00
-7.47058749e-01 2.28766218e-01 -9.36648607e-01 -1.02987839e-02
2.93482691e-01 -4.56581086e-01 -1.23509645e+00 1.17375958e+00
5.23033321e-01 -1.22516848e-01 -2.02156603e-01 5.93583941e-01
8.63161206e-01 8.28225553e-01 1.94923282e-01 -1.15496464e-01
1.42430151e+00 -1.82033330e-01 -4.89187628e-01 -1.09365098e-01
5.29587924e-01 -4.21645582e-01 7.37690568e-01 -2.02383369e-01
-4.10064757e-01 -3.03529710e-01 -7.17234552e-01 6.38488352e-01
-7.56591499e-01 4.08133477e-01 6.07084572e-01 1.68742165e-01
-1.07804930e+00 6.08192384e-01 -1.14680564e+00 -9.94691074e-01
7.71125495e-01 3.93528312e-01 -4.28764045e-01 1.43921405e-01
-8.78854692e-01 5.71643472e-01 4.89968181e-01 6.04242444e-01
-9.56518769e-01 -6.21750474e-01 -8.62763047e-01 -2.93408275e-01
2.02525392e-01 1.56232929e-02 9.48124111e-01 -9.32772875e-01
-8.11375618e-01 1.19163549e+00 -1.67081952e-02 -5.49288869e-01
5.52210748e-01 -1.32235557e-01 -4.42457050e-01 9.46477875e-02
5.47812343e-01 8.06320071e-01 7.08566904e-01 -1.18448949e+00
-1.16038167e+00 -7.39259899e-01 -2.71155685e-02 3.26705612e-02
1.01911962e-01 1.20938107e-01 -4.26025018e-02 -8.87317955e-01
5.14211059e-01 -1.01763439e+00 -2.09739879e-01 -2.37935677e-01
-3.62591684e-01 1.50922492e-01 1.34049809e+00 -1.05682230e+00
1.24261057e+00 -1.94020343e+00 -2.01290131e-01 2.06522509e-01
-1.22731209e-01 4.08677429e-01 -1.64936893e-02 5.13444006e-01
-1.11389749e-01 3.54620367e-01 -8.73887002e-01 6.24301910e-01
-3.10387611e-01 3.87048870e-01 -1.95354983e-01 6.14104033e-01
2.47506678e-01 8.47941339e-01 -8.87697518e-01 -3.66744041e-01
4.19309229e-01 -2.40418792e-01 -8.68673157e-03 2.11745277e-02
-2.81528115e-01 5.81684470e-01 -6.41712070e-01 1.10227048e+00
9.28970754e-01 2.23887250e-01 7.02817962e-02 -5.38842864e-02
-6.75040960e-01 -4.00071621e-01 -1.33693230e+00 1.17640185e+00
2.17361614e-01 6.13222599e-01 2.87008494e-01 -9.08048570e-01
9.25509334e-01 1.00382455e-01 7.61145651e-01 -3.71621281e-01
7.26693636e-03 1.22735471e-01 -2.08025664e-01 -7.18495429e-01
7.42853045e-01 3.84714246e-01 -5.41579090e-02 1.60732746e-01
-2.41950989e-01 -4.37737793e-01 4.08820570e-01 -1.89614207e-01
1.05794728e+00 3.42238486e-01 5.61769664e-01 -6.36642992e-01
2.41360322e-01 6.29810870e-01 4.31120157e-01 6.10670090e-01
-4.86178398e-01 3.02147150e-01 3.64072621e-01 -6.80876911e-01
-8.27993453e-01 -7.60858059e-01 -6.66468680e-01 9.32704985e-01
2.90641993e-01 1.52105317e-01 -5.04206955e-01 -5.53524733e-01
3.45424473e-01 3.67454678e-01 -6.88991666e-01 5.61489463e-01
-4.45365876e-01 -1.48731470e+00 6.57295227e-01 3.31507415e-01
1.18259847e+00 -1.30479956e+00 -1.08863211e+00 3.49409729e-01
-1.72496065e-01 -6.54708743e-01 -3.34368646e-02 5.49750745e-01
-1.02188969e+00 -1.19730818e+00 -7.28591800e-01 -7.20850408e-01
6.89656258e-01 2.60432720e-01 9.40708101e-01 -3.17849934e-01
-6.71750546e-01 2.71892667e-01 -4.67534155e-01 -6.17305040e-01
-1.21453315e-01 7.90235102e-02 -3.06622505e-01 -3.22486103e-01
4.42894816e-01 -6.87843204e-01 -7.22655535e-01 5.30519545e-01
-1.08326197e+00 -8.28113258e-02 3.95151526e-01 4.00340676e-01
7.26478159e-01 3.80044997e-01 2.77680188e-01 -7.36207962e-01
9.62932780e-03 -7.97538459e-01 -9.04831231e-01 2.69459486e-01
-1.67792171e-01 -4.29790825e-01 -1.45929009e-01 -1.16383824e-02
-8.88942659e-01 4.40402538e-01 1.63685963e-01 1.83913544e-01
-8.25114250e-01 7.37400174e-01 -1.74584374e-01 2.81884104e-01
5.33900976e-01 -1.53966844e-01 -4.76927042e-01 -4.62354839e-01
1.62223771e-01 8.58073711e-01 7.45761156e-01 -1.73189059e-01
7.45258331e-01 7.12193727e-01 -3.31684083e-01 -1.14937592e+00
-2.81415433e-01 -8.34403634e-01 -1.41561031e+00 -2.06807286e-01
8.82664502e-01 -1.02371097e+00 -7.95806870e-02 9.48420048e-01
-9.05903578e-01 -8.89343202e-01 -1.56131387e-01 2.01110303e-01
-1.68351486e-01 1.81640133e-01 -2.41945848e-01 -6.90510750e-01
-4.73693669e-01 -7.04932570e-01 1.04005790e+00 4.78839546e-01
-1.70024186e-01 -7.95278370e-01 5.30601405e-02 -9.91493538e-02
6.60952032e-02 8.51125479e-01 7.54579127e-01 -1.36111885e-01
-3.82988840e-01 -6.16040044e-02 -3.94841552e-01 4.39265296e-02
6.60988152e-01 4.84950602e-01 -5.64492226e-01 -2.45589778e-01
-4.81619269e-01 2.83949614e-01 9.68965352e-01 8.48893404e-01
7.69251764e-01 -2.84829676e-01 -6.36808693e-01 5.58834791e-01
1.67970908e+00 5.25037110e-01 6.90872490e-01 6.62197053e-01
4.00339186e-01 6.00757718e-01 9.34806168e-01 8.11879456e-01
4.18967664e-01 4.05562848e-01 7.54760325e-01 -3.95075023e-01
2.85214126e-01 2.63907135e-01 1.90002605e-01 -2.02800632e-01
-3.92539859e-01 -1.79251730e-02 -1.40901673e+00 1.16546738e+00
-1.85788095e+00 -1.06683886e+00 -7.67194092e-01 1.99874914e+00
7.69523919e-01 -5.19859493e-01 3.59427533e-03 -1.08950786e-01
1.04545450e+00 1.50151059e-01 -6.71174347e-01 -3.41196265e-03
-1.80387124e-01 2.96947360e-01 1.29827821e+00 2.97569185e-01
-1.85810149e+00 1.17960632e+00 5.07815027e+00 2.00523779e-01
-1.00453186e+00 -1.33674620e-02 5.50154805e-01 4.43618685e-01
3.09664637e-01 4.87639084e-02 -8.65359962e-01 2.89688885e-01
5.05032003e-01 9.62613150e-02 4.01619971e-02 4.90348399e-01
6.35054827e-01 -8.53782237e-01 -5.30443966e-01 3.98153335e-01
-5.24035513e-01 -8.80253375e-01 -2.29560480e-01 7.72221247e-03
9.92392898e-01 5.46123028e-01 -5.15209377e-01 -3.82193215e-02
8.93407464e-01 -6.33739531e-01 6.85823202e-01 3.88186514e-01
9.79465544e-01 -2.70742685e-01 8.10991287e-01 1.60762265e-01
-1.75848532e+00 -2.09022209e-01 -2.23628312e-01 -6.84176013e-02
-9.15376842e-02 5.61053216e-01 -7.42254078e-01 7.75862217e-01
1.43899524e+00 1.02196503e+00 -8.47295344e-01 9.85092103e-01
-1.92724392e-01 8.03049982e-01 -5.45982122e-01 4.92497593e-01
4.89327878e-01 -5.38281322e-01 6.07664049e-01 1.29170573e+00
3.84805977e-01 6.42074525e-01 3.02007198e-01 6.53256714e-01
1.34481311e-01 -1.10536136e-01 -6.11371458e-01 -2.69267354e-02
3.78602743e-01 1.27539217e+00 -1.29790938e+00 -1.27398580e-01
9.47648287e-03 9.88598406e-01 -3.10753107e-01 4.05384094e-01
-7.56081223e-01 -6.77542388e-01 8.19066346e-01 2.69951224e-01
4.35866147e-01 -4.00673628e-01 -2.04560570e-02 -7.96122789e-01
-5.03918901e-02 -4.61521834e-01 5.85515380e-01 -6.06099367e-01
-9.04194474e-01 1.44201592e-01 4.93233562e-01 -1.03827608e+00
2.52799094e-01 -3.24081600e-01 -6.17260873e-01 8.70983303e-01
-1.74164414e+00 -1.20458078e+00 -9.49951172e-01 5.15729785e-01
6.27452135e-01 8.62309262e-02 7.46797979e-01 1.56857129e-02
-6.68900311e-01 -2.71959126e-01 3.98632526e-01 4.21473712e-01
2.92842269e-01 -1.23037088e+00 5.70700586e-01 1.02550554e+00
-3.46283406e-01 2.27786899e-02 4.55300838e-01 -1.11282289e+00
-9.46402788e-01 -1.76377797e+00 4.29793626e-01 -1.88830435e-01
7.80568719e-01 1.07834309e-01 -8.96051764e-01 9.40879464e-01
-1.76795185e-01 4.75614555e-02 1.52447134e-01 -5.57478786e-01
3.65767300e-01 -1.32996500e-01 -1.41252232e+00 2.50714451e-01
1.03983450e+00 4.10280004e-02 -3.72465700e-01 3.88726026e-01
2.51334548e-01 -3.36900175e-01 -1.00700104e+00 5.88922858e-01
4.31225449e-01 -6.28547251e-01 5.90617537e-01 -4.66222316e-02
4.59675997e-01 -3.95376593e-01 -3.86218339e-01 -1.15823126e+00
-3.50794524e-01 -4.09638256e-01 4.97698039e-01 1.44456708e+00
2.31944770e-01 -6.79928124e-01 4.92566288e-01 1.72932863e-01
-8.86992738e-02 -1.08834878e-01 -7.33688056e-01 -6.33717954e-01
1.03642680e-01 -1.32269040e-01 8.91244590e-01 1.01594710e+00
-5.08365810e-01 -4.35275942e-01 3.18910748e-01 8.92050087e-01
1.67433813e-01 3.28374058e-01 7.41564810e-01 -1.47219157e+00
3.33405405e-01 -3.58220190e-01 -4.63976115e-01 -5.76994836e-01
-2.34694272e-01 -8.64055812e-01 1.14232011e-01 -1.94302011e+00
2.02767011e-02 -8.84002566e-01 2.71166116e-01 1.13223028e+00
-1.17042422e-01 1.76012740e-01 -6.80867583e-02 6.35812998e-01
1.12793311e-01 2.19622523e-01 8.19305182e-01 -3.39733064e-01
-7.92012513e-01 1.86403617e-01 -8.63339603e-02 6.74750268e-01
9.68947709e-01 -5.24606526e-01 1.77364424e-01 -4.86404598e-01
-1.45988658e-01 -8.76855850e-02 6.73134267e-01 -8.78384233e-01
-2.71360070e-01 -5.17330110e-01 4.02622819e-01 -1.14492929e+00
-3.47578049e-01 -1.08239615e+00 5.56680024e-01 8.23977232e-01
2.33296812e-01 7.61914328e-02 5.06655812e-01 4.93772000e-01
-9.61700156e-02 -2.45430142e-01 8.62441421e-01 -5.08912385e-01
-1.22511733e+00 4.70157713e-01 -6.79834485e-01 -3.68258566e-01
1.32248890e+00 -2.19010338e-01 -2.40977824e-01 9.22213048e-02
-7.50461876e-01 6.48488283e-01 4.74822402e-01 3.79072309e-01
1.50916457e-01 -8.93260241e-01 -9.27275121e-01 3.86895537e-01
2.24276140e-01 2.98127949e-01 6.27991408e-02 7.91668952e-01
-1.20877481e+00 3.40919763e-01 -4.14321512e-01 -9.02005315e-01
-1.39338541e+00 -1.31183043e-01 6.96651757e-01 -1.57223731e-01
-6.73784912e-01 5.95323563e-01 -4.23001051e-02 -6.72079027e-01
-1.71985120e-01 -7.44702935e-01 -4.29224432e-01 6.66047692e-01
1.78669482e-01 5.13629019e-01 1.09184317e-01 -8.11183631e-01
-3.88938934e-01 4.60488528e-01 3.86429399e-01 6.55737594e-02
1.85170376e+00 -2.86743551e-01 -4.83285904e-01 3.04913074e-01
7.23529518e-01 -6.40933514e-01 -1.31614935e+00 -1.87337533e-01
4.79275674e-01 -4.71916199e-01 -3.41933556e-02 -9.34488118e-01
-1.13026726e+00 3.33681256e-01 1.25437367e+00 2.04543203e-01
1.23968136e+00 1.87681504e-02 3.20423305e-01 4.62905258e-01
5.92641532e-01 -9.64059412e-01 -6.14036202e-01 5.24913549e-01
1.11836648e+00 -1.40905046e+00 1.54784068e-01 -3.07969719e-01
-4.71760243e-01 1.01048911e+00 4.23068732e-01 8.89841020e-02
1.04069924e+00 2.39066482e-01 -7.53028989e-02 -4.98898268e-01
1.83487460e-01 -7.37530828e-01 -2.92983234e-01 6.28627300e-01
8.91776383e-02 6.66352451e-01 -2.35587314e-01 2.44714737e-01
-1.37461632e-01 3.50390449e-02 4.43045557e-01 1.29189754e+00
-6.84147537e-01 -6.99395239e-01 -6.60906017e-01 9.41294432e-01
-1.95369348e-01 -5.58709353e-02 -3.85168761e-01 9.85568047e-01
4.95894402e-01 9.76446688e-01 2.81396031e-01 7.28218406e-02
4.65141058e-01 -8.95282552e-02 1.44588515e-01 -6.43712640e-01
-8.60054851e-01 -9.96765867e-03 -8.69975314e-02 -1.40321821e-01
-7.53210485e-01 -1.20104158e+00 -1.35089445e+00 -2.08712325e-01
-3.92406732e-01 -2.20907748e-01 6.66917562e-01 5.88949144e-01
6.87649921e-02 2.03715578e-01 7.26461709e-01 -1.09869707e+00
6.63762465e-02 -1.18354058e+00 -1.04903436e+00 3.16012017e-02
2.06207037e-01 -3.94407451e-01 -3.33624817e-02 5.25228798e-01] | [9.378275871276855, -1.5291517972946167] |
d95e3c4e-8ba2-408d-97cf-0b5c98963986 | joint-estimation-of-room-geometry-and-modes | 1802.05879 | null | http://arxiv.org/abs/1802.05879v1 | http://arxiv.org/pdf/1802.05879v1.pdf | Joint Estimation of Room Geometry and Modes with Compressed Sensing | Acoustical behavior of a room for a given position of microphone and sound
source is usually described using the room impulse response. If we rely on the
standard uniform sampling, the estimation of room impulse response for
arbitrary positions in the room requires a large number of measurements. In
order to lower the required sampling rate, some solutions have emerged that
exploit the sparse representation of the room wavefield in the terms of plane
waves in the low-frequency domain. The plane wave representation has a simple
form in rectangular rooms. In our solution, we observe the basic axial modes of
the wave vector grid for extraction of the room geometry and then we propagate
the knowledge to higher order modes out of the low-pass version of the
measurements. Estimation of the approximate structure of the $k$-space should
lead to the reduction in the terms of number of required measurements and in
the increase of the speed of the reconstruction without great losses of
quality. | ['Hervé Lissek', 'Helena Peić Tukuljac', 'Pierre Vandergheynst', 'Thach Pham Vu'] | 2018-02-16 | null | null | null | null | ['room-impulse-response'] | ['audio'] | [ 3.25693786e-01 -6.29582703e-02 1.15840912e+00 -2.29082957e-01
-6.62335634e-01 -3.11826110e-01 1.81788921e-01 1.61501542e-01
-3.22220057e-01 5.20440519e-01 2.25115180e-01 -2.21068144e-01
-4.75240439e-01 -1.06530988e+00 -1.82605758e-01 -8.65616918e-01
2.57163253e-02 1.79201618e-01 1.49842417e-02 -1.41365185e-01
1.50827216e-02 4.52820480e-01 -1.55697191e+00 -1.32264853e-01
5.34002304e-01 1.03863382e+00 1.37737334e-01 1.02851796e+00
-2.11231917e-01 5.32423794e-01 -4.69428986e-01 3.15528840e-01
2.08708271e-01 -4.15348679e-01 -4.86484259e-01 3.30676585e-02
-4.06293496e-02 -1.18775874e-01 2.18391493e-02 1.12656033e+00
7.14125693e-01 4.37737793e-01 6.07471704e-01 -2.00220555e-01
2.07677335e-01 1.99402019e-01 -6.40039667e-02 -1.25905275e-01
6.15539908e-01 -2.07436725e-01 5.41265666e-01 -1.00803113e+00
1.70975059e-01 7.86113977e-01 7.87567139e-01 1.73115209e-01
-1.18911862e+00 -1.63795993e-01 -5.33376515e-01 8.98002535e-02
-1.55374885e+00 -6.99985623e-01 1.19317746e+00 -3.25957596e-01
5.32833040e-01 6.90760434e-01 4.20346886e-01 3.74801219e-01
-2.71999538e-01 -2.61566341e-01 1.12476885e+00 -1.04532254e+00
4.79224920e-01 1.81938663e-01 3.71135324e-01 8.57250869e-01
1.66825429e-01 1.27457589e-01 -2.53006339e-01 -3.51697356e-01
7.83752799e-01 -5.74980304e-02 -7.66129196e-01 -2.02450335e-01
-9.69164133e-01 5.52031040e-01 2.25011751e-01 8.60438526e-01
-4.60697651e-01 2.35844515e-02 -3.26058604e-02 -1.18997499e-01
1.50111869e-01 5.32248199e-01 -5.08132204e-02 -1.26798332e-01
-8.69646728e-01 7.31549934e-02 1.10250711e+00 5.26781917e-01
9.18484390e-01 1.83734402e-01 3.71174753e-01 7.83654928e-01
3.94036621e-01 8.38641882e-01 1.20807640e-01 -8.69639695e-01
3.77746135e-01 1.08766146e-01 4.66418624e-01 -1.17401969e+00
-4.99168575e-01 -3.85394424e-01 -9.32644725e-01 1.58483729e-01
7.59450495e-01 -4.13394421e-01 -4.77735400e-01 1.29941487e+00
6.27846718e-01 3.28009948e-02 1.19144753e-01 5.44219375e-01
6.26747310e-01 9.89261091e-01 -6.83262348e-01 -4.29550171e-01
1.13107502e+00 -3.95271182e-01 -8.81051481e-01 -1.19445346e-01
3.93542558e-01 -1.11925960e+00 8.11198413e-01 4.40373033e-01
-9.05309558e-01 -7.55706072e-01 -8.15254986e-01 4.90127742e-01
-1.08483888e-01 6.51239008e-02 -2.84752324e-02 9.16392565e-01
-9.64720666e-01 3.09669137e-01 -5.99415541e-01 1.89166769e-01
-4.86618251e-01 1.58585057e-01 -3.15308124e-01 -1.72698393e-01
-6.22018278e-01 5.47700644e-01 -3.00942659e-01 7.35815108e-01
-3.69782686e-01 -7.63291299e-01 -7.15125680e-01 3.08823198e-01
-3.83660605e-04 -3.56718034e-01 9.24856424e-01 -3.94915760e-01
-1.49626482e+00 8.92567113e-02 -4.54466790e-01 3.59492123e-01
2.41492733e-01 -4.51584086e-02 -5.27916431e-01 2.94742852e-01
-7.00312331e-02 -6.72689259e-01 9.80569124e-01 -1.26804352e+00
-7.23675713e-02 -2.68733382e-01 -2.88056046e-01 -3.44526693e-02
-1.71575025e-01 -5.05664825e-01 1.73945576e-01 -1.70759037e-01
6.93147957e-01 -8.15347016e-01 -6.33927047e-01 -5.66848636e-01
-4.01578844e-01 3.09358954e-01 2.09895343e-01 -9.16641831e-01
1.05592489e+00 -2.43051386e+00 -5.93744740e-02 8.16890895e-01
1.44307658e-01 -1.96599141e-01 3.36591065e-01 7.99285412e-01
-1.45689964e-01 -4.00891781e-01 -3.35299671e-01 -3.21908027e-01
-3.98555517e-01 -1.94023088e-01 -1.52199149e-01 7.53178775e-01
-6.41183615e-01 -1.56593099e-01 -4.64891434e-01 -3.48904341e-01
3.29164773e-01 5.57473004e-01 -8.78270149e-01 5.10989070e-01
3.46915156e-01 6.34010136e-01 -8.00118625e-01 -9.77161080e-02
8.74838531e-01 1.27302229e-01 -7.62049928e-02 -2.25062683e-01
-5.38639903e-01 3.43388170e-01 -1.80732203e+00 1.22986746e+00
-1.18311155e+00 3.05990487e-01 6.88299656e-01 -9.41325843e-01
1.08927596e+00 7.48199642e-01 4.80209589e-01 -3.27875704e-01
-4.88465168e-02 3.62155467e-01 -1.86500549e-01 -5.97227156e-01
3.88830528e-02 -3.03546846e-01 2.18071952e-01 1.73781767e-01
-2.47727737e-01 -4.27185416e-01 -2.72870749e-01 -5.04139721e-01
1.09000313e+00 -3.73734385e-01 3.33679587e-01 -4.30996388e-01
9.78222668e-01 -2.78439432e-01 -1.64445117e-01 4.19924349e-01
4.72517878e-01 3.36229891e-01 -1.42004237e-01 -4.41082388e-01
-7.13755846e-01 -7.04589844e-01 -3.19013864e-01 5.87384045e-01
-3.56059402e-01 -2.19410911e-01 -8.26442659e-01 2.49957949e-01
-4.61309046e-01 6.17711842e-01 -3.65221083e-01 1.48914382e-01
-9.84624267e-01 -6.81299627e-01 2.73478329e-01 -1.83793664e-01
6.38658285e-01 -8.83472025e-01 -9.00410175e-01 2.74009407e-01
-5.52277982e-01 -1.19422543e+00 -2.48664641e-03 4.51833278e-01
-8.89221013e-01 -9.57677424e-01 -6.79087162e-01 -7.19330907e-01
8.68195713e-01 3.04048117e-02 7.72756696e-01 -1.43437952e-01
-2.50449836e-01 8.09763789e-01 -3.05925697e-01 -3.30590278e-01
-4.34874654e-01 -4.30153847e-01 -1.15709025e-02 2.91781336e-01
-1.75249517e-01 -9.16748703e-01 -5.59559584e-01 1.26302093e-01
-4.70908403e-01 -4.37232405e-01 7.93054849e-02 4.53069896e-01
2.01645344e-01 4.77108777e-01 9.42513421e-02 -4.09308434e-01
4.65208441e-01 3.99764515e-02 -9.35950875e-01 -3.31348628e-01
-2.80309796e-01 2.10724503e-01 9.96724069e-01 -1.07878998e-01
-1.28371215e+00 2.45989524e-02 -6.70925140e-01 3.71657968e-01
-3.50023597e-01 3.23245615e-01 6.26208410e-02 -3.84866506e-01
7.05276012e-01 5.21654963e-01 -3.30487251e-01 -9.67395425e-01
-6.20212965e-02 4.71371472e-01 3.35704654e-01 -5.85111201e-01
1.19145977e+00 4.55888331e-01 6.42430067e-01 -1.67641270e+00
-5.27281702e-01 -9.04006541e-01 -3.94484967e-01 -2.66580224e-01
8.32482278e-01 -4.46999371e-01 -1.01585960e+00 2.68261075e-01
-1.27849567e+00 -1.61711484e-01 -5.38671076e-01 1.00650847e+00
-3.80366147e-01 5.38929462e-01 -3.23437780e-01 -1.30687749e+00
-1.37800202e-01 -9.49454367e-01 5.82498312e-01 -1.56699970e-01
-1.83403134e-01 -1.02996838e+00 4.91910726e-01 1.98583990e-01
6.69882596e-01 5.22059314e-02 1.03869498e+00 -7.56301880e-02
-5.62893033e-01 -4.05230045e-01 3.66733700e-01 2.81234622e-01
1.11788280e-01 -5.34599125e-01 -1.19715512e+00 -6.46769777e-02
1.07610250e+00 2.87268966e-01 2.19666183e-01 8.09607089e-01
8.82567286e-01 -3.09366107e-01 -4.32037283e-03 7.81943560e-01
1.90457368e+00 7.79067874e-02 5.39851606e-01 -4.20543551e-03
4.56537455e-01 7.44688928e-01 1.67254508e-01 7.26867855e-01
-1.69757739e-01 5.95443010e-01 4.62639630e-01 -1.09407127e-01
6.65984675e-02 2.28236075e-02 6.48057684e-02 1.14618909e+00
-6.26937211e-01 1.39465719e-01 -8.56157064e-01 4.55958933e-01
-1.00045896e+00 -1.04727829e+00 -3.13353479e-01 2.53285861e+00
4.29937452e-01 -1.23794578e-01 -1.22461185e-01 8.36937010e-01
3.08984548e-01 2.67938644e-01 2.48688444e-01 -4.63247925e-01
5.00285387e-01 4.18680668e-01 4.77472961e-01 1.37149394e+00
-2.91244298e-01 -1.67297453e-01 6.65855789e+00 3.04327697e-01
-1.01583338e+00 -7.10469931e-02 1.50105655e-01 4.23062384e-01
-3.95364225e-01 -2.19720289e-01 -8.68461072e-01 1.44352570e-01
1.03530252e+00 1.90049544e-01 7.10240364e-01 7.31189132e-01
1.85597703e-01 -2.07267314e-01 -9.16594386e-01 1.10369992e+00
-1.20346762e-01 -9.96747017e-01 -5.01955450e-01 2.63953418e-01
2.32679665e-01 -4.00585443e-01 -1.19146667e-01 -1.23224229e-01
-1.80709824e-01 -1.06239009e+00 4.23911065e-01 5.52133024e-01
5.86563110e-01 -6.88790023e-01 5.52481472e-01 8.17124009e-01
-1.35409045e+00 1.50978994e-02 -4.09658968e-01 -9.78542045e-02
2.58486688e-01 9.36089098e-01 -1.10460567e+00 3.83296341e-01
4.87398148e-01 -2.43830040e-01 -9.59448982e-03 9.10956323e-01
2.11559147e-01 9.62942064e-01 -7.99114227e-01 -1.15020566e-01
2.41170332e-01 -6.61847472e-01 5.15505075e-01 1.04315948e+00
8.32366228e-01 4.42684799e-01 8.63791928e-02 7.57561266e-01
4.59120810e-01 2.06569850e-01 -6.18415594e-01 5.56761622e-01
1.30100310e-01 9.59956825e-01 -4.35410291e-01 1.48396596e-01
-3.27845603e-01 5.26466310e-01 -5.20803213e-01 7.36322165e-01
-3.00271958e-01 -5.25957525e-01 1.63067952e-01 6.15107536e-01
4.51234311e-01 -6.80576801e-01 -2.57537186e-01 -5.81597507e-01
1.92712128e-01 -5.61190546e-01 -2.36328229e-01 -2.66260892e-01
-5.63658834e-01 7.84757853e-01 1.12614222e-01 -1.15189552e+00
-6.59616172e-01 -4.92822886e-01 -6.63655281e-01 1.39351737e+00
-1.39035630e+00 -4.94235069e-01 -3.72624248e-01 1.15432000e+00
2.87670940e-01 2.75153369e-01 1.44118237e+00 2.84505635e-01
1.30973712e-01 5.89152090e-02 4.63361472e-01 1.92279220e-01
1.59222737e-01 -9.73579049e-01 -2.34912544e-01 7.10256040e-01
-2.18982875e-01 8.21700215e-01 1.34009898e+00 -7.48149529e-02
-1.20088851e+00 -4.86621708e-01 9.54211175e-01 -1.88834518e-01
1.78870618e-01 -4.56201583e-01 -6.45469189e-01 2.83039600e-01
-1.51697099e-01 1.41719669e-01 8.95488918e-01 2.26366892e-01
-2.17336714e-02 -4.27303702e-01 -1.19445825e+00 1.76573262e-01
3.91747296e-01 -5.71089745e-01 -4.12461311e-01 3.72459322e-01
1.95107535e-01 3.95910554e-02 -7.65838265e-01 -1.16140038e-01
2.61760831e-01 -1.23697340e+00 1.24484897e+00 2.94296175e-01
4.67826650e-02 -2.69971311e-01 -6.71128571e-01 -1.35502219e+00
-5.36003888e-01 -7.04495132e-01 3.69439125e-01 1.03274012e+00
1.56535685e-01 -6.55695021e-01 7.21327305e-01 3.01355481e-01
-4.07653153e-02 -3.94648045e-01 -1.27810192e+00 -3.65835369e-01
-2.39035487e-01 -3.55622500e-01 3.69414449e-01 4.70665187e-01
9.07047391e-02 2.94781566e-01 -1.86954930e-01 7.89201558e-01
1.13691616e+00 2.83802480e-01 5.41531265e-01 -1.42867398e+00
-7.50576198e-01 2.23085910e-01 -1.27964780e-01 -1.16688693e+00
-3.03359777e-01 -2.72131145e-01 3.96558374e-01 -1.48750246e+00
-4.59066600e-01 -6.94049716e-01 2.12680921e-01 -4.80712384e-01
2.61856973e-01 1.62275421e-04 -9.02309865e-02 -8.06134269e-02
2.43858725e-01 3.09593022e-01 1.19179916e+00 2.42645010e-01
-2.14305162e-01 5.99681020e-01 5.15010841e-02 1.13494623e+00
4.19710368e-01 -3.73478770e-01 -3.78857762e-01 -2.84540564e-01
4.34858084e-01 4.53478187e-01 1.27210647e-01 -1.50045609e+00
2.59513617e-01 1.21939242e-01 3.43616396e-01 -5.18285155e-01
8.09932888e-01 -1.37055671e+00 2.72800356e-01 4.31685895e-01
-1.15174256e-01 -4.55092043e-02 -9.07977745e-02 4.97967899e-01
-3.85466963e-01 -6.22820139e-01 1.00737095e+00 -3.77951622e-01
9.78276413e-03 -2.55134583e-01 -5.93122005e-01 -1.92541093e-01
4.01456058e-01 -2.73661315e-01 3.36310565e-01 -6.59707308e-01
-7.58202791e-01 -6.63595378e-01 2.60847658e-02 -6.45572484e-01
5.24402976e-01 -9.51391280e-01 -4.46613878e-01 5.13257265e-01
-2.95661956e-01 2.61554480e-01 5.93149543e-01 5.59581757e-01
-7.62581646e-01 5.40022075e-01 3.79553847e-02 -3.34503680e-01
-1.21001971e+00 4.04797971e-01 6.74856544e-01 -4.12444890e-01
-6.30360544e-01 6.67252004e-01 2.43500665e-01 -3.17626625e-01
1.11847520e-01 -4.85313207e-01 -4.65887725e-01 -2.15226665e-01
6.62137747e-01 5.93081594e-01 1.98492393e-01 -6.80523753e-01
-3.29297245e-01 1.12375116e+00 8.23076248e-01 -3.85720551e-01
1.45361531e+00 -2.58208126e-01 -3.76152873e-01 6.44546926e-01
1.44044888e+00 1.17518866e+00 -6.86226487e-01 -7.85588995e-02
-4.43667114e-01 -4.71098006e-01 1.34101585e-01 -3.44319135e-01
-4.58640188e-01 9.35370207e-01 5.66762090e-01 5.53119123e-01
1.21638739e+00 -2.62360483e-01 5.27218461e-01 6.25783503e-01
6.13592625e-01 -6.76999807e-01 -1.13106824e-01 2.95265019e-01
6.04053080e-01 -6.67280495e-01 -1.20606564e-01 -8.32665682e-01
2.60434330e-01 1.47181737e+00 -1.81482434e-01 -1.92944542e-01
1.26617515e+00 5.49038470e-01 1.68064430e-01 -1.40913680e-01
3.50653119e-02 1.93648964e-01 1.77004158e-01 7.56335616e-01
6.24227226e-01 6.21284265e-03 5.28169563e-03 3.47103924e-01
-4.56325173e-01 -2.49560505e-01 6.71235383e-01 4.49333817e-01
-8.75273168e-01 -9.64008570e-01 -1.22401273e+00 -1.53937206e-01
-5.52975774e-01 -1.41083285e-01 4.13582057e-01 2.53749222e-01
3.27712744e-02 1.25518501e+00 -3.47401917e-01 3.42034586e-02
6.38296843e-01 3.01015288e-01 4.86432701e-01 -4.19637144e-01
-2.30458856e-01 4.43068951e-01 1.84380993e-01 -3.92109841e-01
-3.08503777e-01 -4.87851024e-01 -1.06239164e+00 -2.97862172e-01
-1.60240531e-01 7.42200017e-01 8.96462619e-01 6.05937600e-01
-2.90319979e-01 7.26908565e-01 8.27404439e-01 -9.12064195e-01
-8.08039725e-01 -8.44816327e-01 -1.10250568e+00 7.83554912e-02
9.73075271e-01 -1.93328131e-02 -7.73768902e-01 -4.80524376e-02] | [15.143007278442383, 5.651333332061768] |
c370120e-9a95-46dc-b7df-ee6a8f25a058 | aggression-identification-in-social-media-a | null | null | https://aclanthology.org/2020.trac-1.5 | https://aclanthology.org/2020.trac-1.5.pdf | Aggression Identification in Social Media: a Transfer Learning Based Approach | The way people communicate have changed in many ways with the outbreak of social media. One of the aspects of social media is the ability for their information producers to hide, fully or partially, their identity during a discussion; leading to cyber-aggression and interpersonal aggression. Automatically monitoring user-generated content in order to help moderating it is thus a very hot topic. In this paper, we propose to use the transformer based language model BERT (Bidirectional Encoder Representation from Transformer) (Devlin et al., 2019) to identify aggressive content. Our model is also used to predict the level of aggressiveness. The evaluation part of this paper is based on the dataset provided by the TRAC shared task (Kumar et al., 2018a). When compared to the other participants of this shared task, our model achieved the third best performance according to the weighted F1 measure on both Facebook and Twitter collections. | ['Josiane Mothe', 'Faneva risoa', 'Rami'] | 2020-05-01 | null | null | null | lrec-2020-5 | ['aggression-identification'] | ['natural-language-processing'] | [-2.59529948e-01 2.64243454e-01 -2.39605531e-02 -6.02083094e-02
-3.06929439e-01 -4.13658351e-01 9.40598607e-01 3.01405132e-01
-5.92454314e-01 6.56419158e-01 7.07556665e-01 2.96295416e-02
-1.50336832e-01 -5.17497003e-01 -1.90910384e-01 -9.91667658e-02
-2.38603234e-01 3.97384644e-01 1.56724989e-01 -4.75180745e-01
4.49847043e-01 2.09147364e-01 -1.22769523e+00 5.79789281e-01
6.34795368e-01 5.98912477e-01 -2.77276188e-01 8.15451205e-01
-1.91366211e-01 1.34503305e+00 -6.79354668e-01 -9.44329143e-01
-1.58150196e-01 -2.75236696e-01 -1.00053954e+00 -2.15849295e-01
2.88647234e-01 -4.52557653e-01 -3.18715453e-01 9.44279909e-01
3.07196617e-01 1.83818899e-02 4.38266754e-01 -1.33530331e+00
-2.11519763e-01 1.08177090e+00 -4.63901818e-01 5.19223809e-01
6.60006404e-01 8.79722238e-02 8.97024453e-01 -5.61310589e-01
7.46774971e-01 1.39603984e+00 7.74002552e-01 4.77562875e-01
-1.14230978e+00 -9.10255909e-01 -5.83342873e-02 2.84553140e-01
-1.29765594e+00 -5.48658550e-01 7.87737668e-01 -6.72758102e-01
7.31248796e-01 5.13024926e-01 9.38031733e-01 1.83470941e+00
-2.95739863e-02 6.94695652e-01 1.24807620e+00 7.72809833e-02
-1.12231076e-01 5.72478116e-01 4.07860100e-01 5.29068112e-01
-1.73612267e-01 -1.93585694e-01 -7.45792627e-01 -4.51735854e-01
-1.55731618e-01 -3.93768728e-01 1.00300252e-01 4.81181145e-01
-5.55486798e-01 1.08683681e+00 3.18058014e-01 7.01170504e-01
-2.83969909e-01 -1.36111602e-01 6.72003806e-01 3.43225777e-01
9.03566778e-01 5.32050788e-01 1.06636412e-01 -1.14137244e+00
-6.38588011e-01 4.61989313e-01 9.94259536e-01 1.35596007e-01
2.42901251e-01 -4.58964050e-01 -1.61430553e-01 9.80105400e-01
3.26671809e-01 -2.02415492e-02 6.92135215e-01 -7.27341294e-01
5.12572050e-01 7.62079597e-01 -4.09467965e-01 -1.38765025e+00
-7.84899712e-01 -5.84095359e-01 -6.39301240e-01 -2.54349500e-01
4.65990245e-01 -2.15027168e-01 -5.81137165e-02 1.86331213e+00
3.23414683e-01 2.54724443e-01 -4.81791228e-01 5.51165819e-01
7.56520927e-01 6.42399907e-01 1.28496438e-01 -1.52123213e-01
1.02910626e+00 -7.33322382e-01 -8.64794970e-01 -2.85586156e-02
9.26823020e-01 -5.95352650e-01 8.18872154e-01 5.77744305e-01
-1.01093197e+00 -2.76939780e-03 -6.97655082e-01 -1.56488791e-01
-6.10281825e-01 -4.59579885e-01 4.02812958e-01 8.98230016e-01
-9.67286646e-01 5.30279636e-01 -5.09485364e-01 -6.07043862e-01
5.93640029e-01 1.17397286e-01 -4.60301489e-01 5.00825047e-01
-1.38887298e+00 1.04530919e+00 -1.42180575e-02 -8.25563967e-02
-3.59359056e-01 -6.92916632e-01 -3.12795967e-01 -3.56164545e-01
2.90808946e-01 -2.15936497e-01 1.10242569e+00 -8.17246854e-01
-1.53661406e+00 1.18039751e+00 2.25453541e-01 -6.64669156e-01
9.95147526e-01 -6.04034841e-01 -3.49633068e-01 -2.07842320e-01
8.33186507e-02 2.90937036e-01 7.14471757e-01 -7.41046250e-01
-4.43596065e-01 -6.49992287e-01 1.21142440e-01 -5.15725166e-02
-7.78068125e-01 6.14781082e-01 5.33575006e-02 -4.33469325e-01
-5.75400412e-01 -9.72941995e-01 2.73015916e-01 -3.49494249e-01
-9.17594552e-01 -4.49087381e-01 1.09634399e+00 -1.09956110e+00
1.74951804e+00 -2.19432878e+00 4.63002056e-01 1.37932390e-01
8.08498681e-01 5.01290560e-01 3.58479440e-01 8.80745828e-01
3.16645131e-02 6.41536236e-01 -9.11479443e-02 -8.57495844e-01
5.37158512e-02 -1.50999859e-01 -1.50224686e-01 6.56835020e-01
-2.94613838e-01 6.53125942e-01 -6.57308638e-01 -3.49465877e-01
-7.40599558e-02 4.87628818e-01 -5.36426663e-01 2.03097686e-01
1.02505162e-01 6.63646579e-01 -2.40174264e-01 6.11660257e-03
3.97799402e-01 5.07216230e-02 -1.50383219e-01 2.11793020e-01
-5.05630136e-01 5.74333131e-01 -6.49485111e-01 1.08533800e+00
-5.19245803e-01 1.02596378e+00 2.85042465e-01 -4.42184538e-01
5.57052851e-01 2.31215969e-01 3.91097784e-01 -6.18426383e-01
4.47405756e-01 -3.81288938e-02 3.62771958e-01 -6.76910579e-01
3.65275294e-01 2.52938539e-01 -1.78796649e-01 7.44985998e-01
-3.38467956e-01 1.71935156e-01 2.91409582e-01 6.22100472e-01
1.22742987e+00 -5.21868050e-01 3.83645207e-01 -5.79020306e-02
7.22934425e-01 -4.95253831e-01 6.06454685e-02 6.54222786e-01
-5.68418741e-01 1.30620241e-01 9.59039569e-01 -2.58467495e-01
-8.27753425e-01 -6.09070003e-01 -2.21932083e-01 1.11235344e+00
-5.54678023e-01 -9.66617107e-01 -8.53070080e-01 -4.76445585e-01
7.87830800e-02 8.24414909e-01 -9.56816256e-01 -3.83258909e-01
-4.60778117e-01 -3.50454569e-01 1.04857910e+00 -3.34167868e-01
5.90477765e-01 -8.69900823e-01 -3.94575238e-01 2.43279800e-01
-4.87569451e-01 -1.20771551e+00 -1.49735361e-01 -4.93199110e-01
-1.39850363e-01 -1.05761123e+00 -2.83417910e-01 1.00730561e-01
-1.05162486e-01 -1.48019344e-01 5.90771973e-01 2.15472206e-01
-1.79764003e-01 6.43217564e-02 -5.49605072e-01 -5.44579923e-01
-7.34917462e-01 4.39185590e-01 -7.07704667e-03 4.68140662e-01
2.58584529e-01 -7.15845644e-01 -3.50380205e-02 -1.12224100e-02
-7.10356593e-01 2.56577641e-01 -5.10860011e-02 2.84504771e-01
-5.48200130e-01 -2.00371087e-01 2.63942599e-01 -1.11945868e+00
1.24032581e+00 -1.01891553e+00 -8.54396597e-02 -4.35147017e-01
-2.69041449e-01 -4.16195363e-01 5.61088622e-01 -3.84626150e-01
-7.18631268e-01 -7.70957053e-01 -3.62835974e-01 6.91320281e-03
-6.09577969e-02 5.07279217e-01 3.28793734e-01 1.46690264e-01
3.76354188e-01 -1.40043721e-01 1.38026416e-01 -7.19828069e-01
2.36588567e-02 1.11352444e+00 -1.49070024e-01 -7.74051622e-02
7.77254879e-01 3.41673076e-01 -8.15348551e-02 -1.43475246e+00
-9.70738411e-01 -3.96728754e-01 -4.91744757e-01 -6.77896440e-01
9.69501436e-01 -6.01915061e-01 -1.17017579e+00 8.67149413e-01
-1.16943228e+00 -3.22123975e-01 2.42799312e-01 2.33462527e-01
-1.94155350e-01 2.43599057e-01 -9.16057229e-01 -1.12946641e+00
-2.87995547e-01 -8.01057100e-01 6.12596691e-01 -1.54759243e-01
-9.34090436e-01 -1.03896773e+00 5.21903813e-01 8.04428756e-01
8.30336094e-01 4.15681958e-01 6.80515945e-01 -1.10351205e+00
2.76164621e-01 -9.47627723e-02 -1.26335010e-01 1.20408192e-01
-5.17540090e-02 1.74857169e-01 -9.95631456e-01 -1.40467957e-02
-1.07725756e-02 -4.85122770e-01 6.92507923e-01 -2.10174441e-01
1.15372598e+00 -8.81067991e-01 -7.90430307e-02 1.99051484e-01
6.21135831e-01 -2.60379553e-01 6.15285456e-01 4.53244150e-01
7.89200664e-01 9.71692324e-01 8.09302777e-02 8.11106741e-01
5.63202560e-01 1.04737926e+00 4.11337137e-01 3.41832519e-01
9.95515380e-03 -5.08607864e-01 6.01439655e-01 8.45269322e-01
1.59291998e-01 -2.94316053e-01 -1.00951600e+00 1.16241306e-01
-1.73412144e+00 -1.25110221e+00 -6.98153496e-01 1.86317527e+00
6.60964310e-01 1.71530142e-01 8.01859617e-01 2.29854032e-01
6.16828561e-01 2.46957213e-01 -4.92201746e-02 -1.07460833e+00
1.01331808e-01 -1.92107692e-01 2.45943055e-01 8.82469952e-01
-9.52999055e-01 7.98700750e-01 5.41581011e+00 6.08914435e-01
-1.11152947e+00 4.12632018e-01 6.03559256e-01 -3.93252820e-01
1.51080117e-01 -3.71057689e-01 -5.53665817e-01 1.03535450e+00
1.35825360e+00 -4.06900018e-01 6.00561440e-01 2.13161170e-01
4.26262200e-01 -1.71427891e-01 -7.40643501e-01 8.86826396e-01
4.04331177e-01 -7.93909669e-01 -3.34334999e-01 5.19734323e-01
2.64646173e-01 1.03445880e-01 3.04268181e-01 3.66217941e-01
1.25365421e-01 -1.13887215e+00 6.29534900e-01 5.93790829e-01
1.91986531e-01 -6.53081357e-01 5.12675524e-01 7.96433270e-01
-4.49054807e-01 -2.94909149e-01 3.22495908e-01 -4.70566720e-01
2.94799209e-01 5.76442659e-01 -9.31724131e-01 -1.08098879e-01
7.15910196e-01 8.59473407e-01 -8.99226785e-01 7.42273211e-01
-1.37385041e-01 1.10006201e+00 -4.77381110e-01 -4.16227818e-01
2.64447242e-01 -2.91431010e-01 1.21206033e+00 1.24217284e+00
-1.03075340e-01 -2.17392012e-01 -1.54297858e-01 7.39355326e-01
-2.13720515e-01 2.92649329e-01 -7.85061657e-01 -4.70010102e-01
2.14917913e-01 1.24817050e+00 -1.45892337e-01 5.30472547e-02
-1.31788552e-01 8.81388068e-01 9.11774576e-01 -2.00792581e-01
-7.96677470e-01 9.31254551e-02 5.82804501e-01 5.90527654e-01
-3.55094314e-01 -1.59535497e-01 -2.06251573e-02 -7.95106769e-01
-1.67805746e-01 -8.73249590e-01 2.97841132e-01 -3.66446853e-01
-1.41162610e+00 4.02239203e-01 9.22202468e-02 -5.99698842e-01
-4.49092478e-01 -2.72900969e-01 -6.05898321e-01 5.39494336e-01
-8.26067507e-01 -1.21481335e+00 -1.08600929e-01 4.54746634e-01
2.94722497e-01 -1.51288226e-01 5.97928703e-01 5.88242352e-01
-8.96612287e-01 5.51970482e-01 -1.63185105e-01 2.24909917e-01
5.41430295e-01 -9.95701194e-01 1.40925884e-01 3.96468312e-01
-9.17168707e-02 3.83404940e-01 9.52493429e-01 -5.77038109e-01
-1.08856440e+00 -8.06858420e-01 1.23941982e+00 -6.45603716e-01
1.35732174e+00 -8.61052990e-01 -9.17488873e-01 7.38944411e-01
3.24511766e-01 -7.90735662e-01 1.02687383e+00 4.27155584e-01
-3.43733311e-01 1.40297383e-01 -1.15418315e+00 6.85947239e-01
1.22480321e+00 -5.48781812e-01 -3.59733611e-01 5.51939249e-01
3.58950257e-01 -1.80734381e-01 -6.47214055e-01 -1.85295448e-01
5.58885753e-01 -1.53793406e+00 4.42408144e-01 -7.45980918e-01
7.59681284e-01 5.85931301e-01 2.65916854e-01 -1.44315267e+00
-2.82500714e-01 -7.48189330e-01 -2.10456759e-01 1.54765785e+00
3.60313654e-01 -8.68990898e-01 4.46858585e-01 5.94706953e-01
2.26158813e-01 -4.68757540e-01 -1.14443970e+00 -4.51051056e-01
2.51194835e-01 -6.63941503e-01 2.01307237e-01 1.05590641e+00
4.00190681e-01 5.02821088e-01 -5.53266764e-01 -3.35475802e-01
5.03286779e-01 -7.36325800e-01 1.13837075e+00 -1.42310548e+00
-1.64030284e-01 -9.06572759e-01 -6.95519328e-01 -4.36355710e-01
4.00908411e-01 -9.36440349e-01 -6.75635338e-01 -1.04628277e+00
3.39777946e-01 -2.03176722e-01 2.23218352e-01 1.21707894e-01
2.50272691e-01 3.61276656e-01 4.03118849e-01 1.12709306e-01
-5.46085060e-01 4.84859467e-01 8.45104933e-01 1.27151906e-01
-2.14758232e-01 3.55071612e-02 -7.55024076e-01 6.62260354e-01
9.04802084e-01 -3.99551809e-01 -7.24809021e-02 -2.96098981e-02
7.33171403e-01 -8.96011889e-02 4.49295729e-01 -9.85249698e-01
2.19454631e-01 1.22455515e-01 -2.21714675e-01 -2.74489313e-01
8.03617716e-01 -6.80011809e-01 5.89011647e-02 6.45750344e-01
-7.53445148e-01 2.01321378e-01 1.25887752e-01 3.38399589e-01
3.01394667e-02 -1.03734791e-01 6.53479278e-01 2.01094925e-01
7.45084435e-02 1.60969645e-01 -8.07724774e-01 3.10564190e-01
1.17868006e+00 6.39402866e-03 -5.12346208e-01 -9.96586323e-01
-7.25819290e-01 1.55531794e-01 7.22151697e-02 7.79745221e-01
2.45224714e-01 -9.65759277e-01 -1.12654078e+00 -1.30104236e-02
3.45644541e-02 -8.01478624e-01 2.73145705e-01 1.25030756e+00
-2.55645841e-01 4.39145029e-01 -1.04062371e-01 -1.26594603e-01
-1.48991609e+00 1.84099182e-01 3.85491103e-01 -4.65741992e-01
-4.60838020e-01 8.05608749e-01 -3.48502278e-01 -3.60504836e-01
6.49599358e-02 3.51152807e-01 -5.71605146e-01 3.63542080e-01
8.97492290e-01 9.52706158e-01 -5.46300858e-02 -1.28173244e+00
-2.36497924e-01 -7.32283443e-02 -4.80247408e-01 -3.73217732e-01
1.43708801e+00 -1.42952830e-01 -5.16150951e-01 9.08031583e-01
1.84997094e+00 5.08274853e-01 -3.53188276e-01 -2.00660154e-01
-1.16420211e-02 -7.30276167e-01 1.41331702e-01 -8.33509147e-01
-7.26451159e-01 7.51392841e-01 3.53220761e-01 9.27604616e-01
3.72204781e-01 1.57955483e-01 8.56841862e-01 -9.01853144e-02
4.00829986e-02 -1.10635161e+00 4.63957302e-02 7.98816025e-01
1.16921151e+00 -1.08546531e+00 -1.52559936e-01 -5.01866579e-01
-6.24878585e-01 8.95112097e-01 6.60757184e-01 7.74649605e-02
8.58914196e-01 1.62838593e-01 -1.22433066e-01 -4.87752974e-01
-9.82968569e-01 6.64174929e-02 6.29185289e-02 3.77678305e-01
6.36823475e-01 1.28532305e-01 -7.92024732e-01 5.34260154e-01
-7.79910088e-01 -2.58087993e-01 6.44665122e-01 4.51700479e-01
-1.30964369e-01 -9.24574554e-01 -1.85501248e-01 6.74835801e-01
-7.60843158e-01 1.57268912e-01 -1.31435132e+00 6.61560953e-01
-8.66749585e-02 1.37244093e+00 1.42319560e-01 -7.61343479e-01
2.09253326e-01 2.04220321e-02 -3.24962884e-02 -4.66337442e-01
-1.19913554e+00 -7.14318335e-01 6.95651472e-01 -6.62852526e-01
-3.63172367e-02 -9.37442660e-01 -7.87436843e-01 -8.88832033e-01
4.08211276e-02 1.21945158e-01 8.21823359e-01 1.09177923e+00
3.02873075e-01 2.92280942e-01 8.17365289e-01 -4.09354985e-01
-4.67634171e-01 -1.38389218e+00 -4.40829337e-01 7.08078325e-01
3.44601065e-01 -6.51573539e-01 -7.48480260e-01 -7.14892268e-01] | [8.715832710266113, 10.508666038513184] |
45a7d06a-63e5-4d31-a680-8da44c7c2915 | usb-universal-scale-object-detection | 2103.14027 | null | https://arxiv.org/abs/2103.14027v3 | https://arxiv.org/pdf/2103.14027v3.pdf | USB: Universal-Scale Object Detection Benchmark | Benchmarks, such as COCO, play a crucial role in object detection. However, existing benchmarks are insufficient in scale variation, and their protocols are inadequate for fair comparison. In this paper, we introduce the Universal-Scale object detection Benchmark (USB). USB has variations in object scales and image domains by incorporating COCO with the recently proposed Waymo Open Dataset and Manga109-s dataset. To enable fair comparison and inclusive research, we propose training and evaluation protocols. They have multiple divisions for training epochs and evaluation image resolutions, like weight classes in sports, and compatibility across training protocols, like the backward compatibility of the Universal Serial Bus. Specifically, we request participants to report results with not only higher protocols (longer training) but also lower protocols (shorter training). Using the proposed benchmark and protocols, we conducted extensive experiments using 15 methods and found weaknesses of existing COCO-biased methods. The code is available at https://github.com/shinya7y/UniverseNet . | ['Yosuke Shinya'] | 2021-03-25 | null | null | null | null | ['real-time-object-detection'] | ['computer-vision'] | [-1.75876066e-01 -5.29748559e-01 -3.29777002e-01 -3.94345552e-01
-6.76873386e-01 -5.36550343e-01 2.93676227e-01 -2.77123898e-01
-6.58060551e-01 5.31647742e-01 -2.38394916e-01 -1.12442032e-01
5.12198545e-02 -7.17277467e-01 -7.74890244e-01 -3.79618615e-01
-2.00549990e-01 1.85748830e-01 1.07867420e+00 -2.69403279e-01
2.13156044e-01 3.08846891e-01 -1.69868958e+00 5.29742837e-01
3.99155706e-01 1.18016231e+00 2.57769227e-01 9.41851318e-01
1.55181155e-01 7.93631852e-01 -6.86987996e-01 -5.68127871e-01
5.49883723e-01 -2.59988844e-01 -9.20830905e-01 -2.67107785e-01
9.02098954e-01 -4.35106844e-01 -4.48338687e-01 1.13334453e+00
9.29575205e-01 2.61784475e-02 2.86535084e-01 -1.58332288e+00
-7.66626179e-01 7.12185800e-01 -6.46933138e-01 8.38446200e-01
4.74848598e-02 4.92194980e-01 9.54671085e-01 -6.94339037e-01
7.08273768e-01 1.25039780e+00 9.56801951e-01 7.65007138e-01
-9.61749375e-01 -9.72473741e-01 8.10305215e-03 5.51174164e-01
-1.42050028e+00 -2.96244264e-01 2.74947554e-01 -1.31767571e-01
9.48409498e-01 5.62388897e-01 5.73247373e-01 1.43053532e+00
2.36307681e-02 8.64653230e-01 1.36657035e+00 -2.95126140e-01
4.64982018e-02 2.08485737e-01 2.80444115e-01 5.67437530e-01
6.22180164e-01 8.17491040e-02 -5.23759186e-01 3.07553858e-01
7.31890976e-01 -3.36048156e-01 3.49331312e-02 -3.11810195e-01
-1.31220305e+00 6.39977694e-01 8.86334181e-01 8.64721462e-02
1.70257717e-01 2.85086304e-01 7.25411952e-01 1.48217976e-01
8.51218030e-02 2.55450934e-01 -3.65775585e-01 -7.44862556e-02
-5.85196793e-01 4.02468115e-01 6.18168950e-01 1.45241666e+00
4.23512936e-01 -5.97568676e-02 -2.71746933e-01 9.05862033e-01
1.50188699e-01 5.76085031e-01 6.25143468e-01 -1.18719161e+00
5.23830950e-01 3.52962822e-01 2.43016295e-02 -7.97403395e-01
-5.38971663e-01 -3.16490501e-01 -4.79993522e-01 3.75948697e-01
8.01271141e-01 -5.43949902e-02 -7.50520408e-01 1.54617250e+00
5.34581602e-01 1.29792422e-01 -1.68162972e-01 1.23375463e+00
1.28342867e+00 2.94431865e-01 4.90774885e-02 3.14767450e-01
1.82204247e+00 -1.33458424e+00 -3.82125348e-01 -1.94252312e-01
5.31690598e-01 -1.00657761e+00 1.37677860e+00 2.49673307e-01
-1.03011489e+00 -7.98668265e-01 -1.44347906e+00 -2.57442594e-01
-6.36955082e-01 -5.13235386e-03 7.12757230e-01 1.00972700e+00
-7.73674130e-01 4.24922228e-01 -6.62110925e-01 -8.44334543e-01
3.85378718e-01 4.85897530e-03 -4.47455272e-02 1.30415067e-01
-1.17032647e+00 9.84094679e-01 5.86379111e-01 -1.93263933e-01
-7.99712837e-01 -4.88507271e-01 -5.43070018e-01 -1.77410468e-01
5.84145606e-01 -4.59458023e-01 1.42462170e+00 -6.25388980e-01
-1.13953459e+00 9.98873234e-01 6.57147408e-01 -6.75820887e-01
6.89886928e-01 -1.92365855e-01 -4.69899207e-01 1.89700633e-01
2.82921493e-01 1.10976899e+00 4.16589648e-01 -1.01601791e+00
-1.02637041e+00 -1.86089292e-01 6.21477485e-01 1.15923658e-01
-2.36167893e-01 2.65090317e-01 -8.76639545e-01 -7.40820825e-01
-8.37679729e-02 -9.54968154e-01 8.08541942e-03 2.83139348e-01
-3.82395983e-01 -2.20941976e-01 7.29156733e-01 -2.05001742e-01
1.17379117e+00 -2.20245600e+00 -5.24307787e-01 -2.43707776e-01
1.14229508e-01 2.39383981e-01 -3.99521977e-01 -2.78747976e-02
1.66509211e-01 1.83479160e-01 1.48839146e-01 -1.34105474e-01
9.67175215e-02 2.78223366e-01 1.11339025e-01 5.27192712e-01
8.69399160e-02 7.64105856e-01 -8.14072013e-01 -8.45274985e-01
2.12028906e-01 1.12952538e-01 -5.78893840e-01 3.77094001e-02
1.05667815e-01 5.51905334e-02 -3.24504256e-01 1.18058133e+00
8.09407115e-01 -2.04087004e-01 -1.11953050e-01 -6.41713440e-01
-2.26799488e-01 1.78363860e-01 -1.69171071e+00 1.59494352e+00
-1.06120974e-01 7.81190455e-01 -6.96526840e-02 -7.06136644e-01
4.01610494e-01 -4.90397178e-02 2.33683228e-01 -9.38845158e-01
4.17354047e-01 2.90958285e-01 1.55854523e-01 -6.04080319e-01
6.42328978e-01 3.87195021e-01 4.48594764e-02 2.14698330e-01
2.81791717e-01 -2.14269105e-02 9.21559215e-01 3.41972619e-01
9.01076674e-01 2.38525212e-01 2.75239259e-01 -3.51040155e-01
1.10452071e-01 1.51026025e-01 5.67171752e-01 1.12825751e+00
-7.86874473e-01 9.40416396e-01 3.63207966e-01 -4.62907284e-01
-1.13322341e+00 -1.23133612e+00 -5.99346995e-01 1.21139455e+00
6.59369290e-01 -6.04997218e-01 -8.47862065e-01 -7.28927612e-01
1.29454127e-02 1.79443955e-01 -7.91409552e-01 9.57193822e-02
-4.32901025e-01 -9.11070108e-01 9.90495622e-01 8.13538730e-01
8.19684088e-01 -1.24868286e+00 -1.00704193e+00 -4.48716171e-02
-3.40280801e-01 -1.33646524e+00 -6.68729305e-01 -8.35285112e-02
-7.49173522e-01 -1.49144924e+00 -6.93078101e-01 -7.07972407e-01
1.59509957e-01 6.53173208e-01 1.16668999e+00 9.80034247e-02
-7.12278068e-01 3.51515681e-01 -5.39526403e-01 -8.07575524e-01
-1.71748698e-01 2.79828101e-01 1.66591078e-01 -4.10369068e-01
2.17348546e-01 -1.61613673e-01 -1.02837384e+00 8.79854858e-01
-8.49393964e-01 4.68298271e-02 5.70820630e-01 4.71847326e-01
4.39723998e-01 -5.98365247e-01 4.98592585e-01 -8.46941590e-01
1.96199998e-01 -2.98152238e-01 -6.88433290e-01 2.54291177e-01
-6.96790755e-01 -4.45021331e-01 -5.90041727e-02 -6.69503272e-01
-8.18681717e-01 -4.16531652e-01 -3.88751067e-02 -1.39985219e-01
-6.68508634e-02 -1.86925277e-01 9.92934108e-02 -3.51624370e-01
1.17451501e+00 -1.80136383e-01 -2.81301409e-01 -5.44258416e-01
2.94376343e-01 7.20174789e-01 4.78476197e-01 -6.02862477e-01
5.05591631e-01 5.16153395e-01 -4.48639095e-01 -4.50279564e-01
-8.11676919e-01 -6.02707267e-01 -5.07251561e-01 -3.45987260e-01
8.30945373e-01 -1.05196333e+00 -6.56289756e-01 5.40322244e-01
-9.29208040e-01 -3.03984791e-01 -2.41156504e-01 6.20049536e-01
-4.11530375e-01 1.32296532e-01 -8.54908347e-01 -3.37149680e-01
-3.47868353e-01 -1.17113686e+00 8.22412848e-01 5.51546872e-01
5.56215346e-02 -4.48549211e-01 -1.77134201e-01 5.58226407e-01
5.66028357e-01 2.64864206e-01 4.66707945e-02 -4.43044990e-01
-6.49345815e-01 -1.04687877e-01 -8.06713223e-01 4.74922448e-01
-1.32340893e-01 7.74184540e-02 -9.56318855e-01 -4.28226203e-01
-3.02355856e-01 -6.75921142e-01 7.04699993e-01 2.56685466e-01
1.10431802e+00 4.35933061e-02 -2.60795802e-01 8.19353402e-01
1.35166407e+00 -6.87971562e-02 5.83396614e-01 1.00850463e+00
5.67511022e-01 3.12466800e-01 6.53716207e-01 2.94919819e-01
4.03474867e-01 6.82633817e-01 3.96516502e-01 -8.13074335e-02
-8.01132858e-01 2.59322464e-01 3.44546109e-01 6.28158450e-01
-4.87338901e-01 6.39051795e-02 -7.08694398e-01 5.08320451e-01
-1.54836226e+00 -8.87558579e-01 -3.66517156e-01 1.95066702e+00
8.81242990e-01 5.83142638e-01 5.63979030e-01 -1.96020603e-01
7.30869889e-01 3.97608951e-02 -5.29910445e-01 -2.17791080e-01
-2.32504010e-01 1.19069675e-02 7.51949728e-01 -6.26899526e-02
-1.13672519e+00 6.48387611e-01 6.95235062e+00 1.07831907e+00
-1.03241837e+00 6.72272742e-01 4.86583829e-01 -5.77803254e-01
5.36613047e-01 -1.57070294e-01 -1.09444237e+00 5.00746548e-01
8.33423257e-01 -5.42403199e-02 2.39245370e-01 1.07419455e+00
-2.64237523e-01 -1.89752847e-01 -9.91190434e-01 9.66535509e-01
8.22005495e-02 -1.29980457e+00 -1.77932501e-01 -3.35922927e-01
7.16857314e-01 5.26651084e-01 -1.79609582e-01 6.71982825e-01
5.88659979e-02 -4.67387587e-01 1.23232317e+00 -4.86541502e-02
8.25701594e-01 -2.87774235e-01 7.14858353e-01 -7.07441568e-02
-1.37442255e+00 -7.18029886e-02 -5.93555808e-01 1.62873920e-02
-1.37521148e-01 -4.56838161e-02 -3.19667548e-01 4.66653287e-01
1.52047157e+00 4.23443258e-01 -1.18003595e+00 1.32184076e+00
-7.01592788e-02 6.74169779e-01 -4.17260736e-01 -2.23900378e-01
1.15822926e-01 9.25242826e-02 5.27136385e-01 1.62586689e+00
1.19656339e-01 -1.70420438e-01 2.16952994e-01 6.08087301e-01
-2.33835727e-01 1.32313013e-01 -3.13641280e-01 4.52592075e-01
4.76672739e-01 1.51453710e+00 -1.07088661e+00 -5.22866070e-01
-7.35891581e-01 6.21265352e-01 2.24594638e-01 1.72089145e-01
-1.52529144e+00 -4.72868979e-01 5.50163627e-01 1.34615183e-01
3.01063746e-01 -6.60147984e-03 -4.03237671e-01 -1.11009538e+00
-2.21309680e-02 -1.07088065e+00 8.52419734e-01 -8.22897732e-01
-1.28669786e+00 4.95457143e-01 4.01837051e-01 -1.47311127e+00
6.04210317e-01 -8.13212872e-01 -4.90831137e-01 4.30383056e-01
-1.33220899e+00 -1.30700290e+00 -6.15088701e-01 4.27059978e-01
8.90857816e-01 -3.66388261e-02 3.82921845e-01 7.80523121e-01
-8.76257241e-01 8.32693636e-01 -8.82985964e-02 3.44933838e-01
1.12045431e+00 -1.24824941e+00 4.67577428e-01 8.59684289e-01
2.11762950e-01 5.34585834e-01 6.20028436e-01 -2.96680629e-01
-1.14935434e+00 -9.67074752e-01 1.06908992e-01 -6.21433139e-01
7.43081033e-01 -4.08442289e-01 -6.38298571e-01 7.60547996e-01
2.83191234e-01 4.82770205e-01 4.13373172e-01 1.36448145e-01
-5.19890904e-01 -3.99464458e-01 -9.93210375e-01 6.16726041e-01
1.26131070e+00 -1.17869891e-01 -3.54724169e-01 3.50386024e-01
7.11466849e-01 -8.72512281e-01 -8.21543992e-01 4.90873843e-01
9.31747437e-01 -1.23572099e+00 1.20322490e+00 -6.42977595e-01
2.56847739e-01 -4.24311876e-01 -3.23736876e-01 -6.56397879e-01
-4.12530065e-01 -2.44464856e-02 -1.33077636e-01 1.33230042e+00
4.97391045e-01 -6.99700832e-01 5.90946734e-01 1.08450882e-01
-5.25412485e-02 -7.88676262e-01 -9.53453898e-01 -1.19425821e+00
-1.54770225e-01 -5.77246487e-01 4.31843549e-01 8.04176152e-01
-3.94091249e-01 2.21736744e-01 -2.82815278e-01 1.83015645e-01
7.93794513e-01 1.15488961e-01 1.02278090e+00 -6.25311673e-01
-3.05619955e-01 -5.60326993e-01 -5.15227437e-01 -7.28012323e-01
-8.28471839e-01 -7.54019618e-01 -9.04379934e-02 -1.19170952e+00
5.50854385e-01 -4.98057991e-01 -5.48887014e-01 4.74918693e-01
-2.09914222e-01 1.11089551e+00 6.17225170e-01 3.57823849e-01
-1.16869414e+00 3.00851148e-02 1.27511799e+00 -2.12717324e-01
1.84037104e-01 -1.61859408e-01 -5.16616940e-01 7.99480200e-01
9.21875596e-01 -6.24215961e-01 -5.29439338e-02 -4.93097246e-01
2.33905539e-02 -6.79834604e-01 5.51558495e-01 -1.67301989e+00
1.45694152e-01 -3.71621810e-02 4.25694555e-01 -4.51482803e-01
3.68681662e-02 -3.32732469e-01 -1.04603797e-01 7.77225196e-01
-1.48458868e-01 1.04768723e-01 3.70846510e-01 1.94472730e-01
5.42167164e-02 -3.18020463e-01 1.03618991e+00 -1.09464556e-01
-1.19326293e+00 2.63858229e-01 -2.94302180e-02 3.12529474e-01
1.10340369e+00 -3.90729606e-01 -6.09575033e-01 2.06551999e-01
-5.64673662e-01 4.95535702e-01 3.58462572e-01 8.62557352e-01
3.04580510e-01 -1.52262723e+00 -6.88552916e-01 -2.22049698e-01
4.77883965e-01 -2.10117206e-01 2.17622429e-01 1.06311381e+00
-9.16570544e-01 2.11635828e-01 -6.54360831e-01 -7.10685849e-01
-1.32494664e+00 5.90536118e-01 3.68658423e-01 -4.33010012e-02
-5.70249498e-01 7.92291105e-01 1.33408472e-01 -5.34678042e-01
4.14400995e-01 -5.65499961e-01 -7.16281310e-02 3.41842733e-02
7.81444013e-01 7.54746199e-01 6.58523664e-02 -3.40529501e-01
-5.08412540e-01 4.25735921e-01 -1.46126393e-02 2.09557682e-01
9.58967149e-01 -2.21938387e-01 3.47620279e-01 5.23141861e-01
8.50054741e-01 -1.46999344e-01 -1.21190882e+00 -1.77683443e-01
-1.31182551e-01 -6.13654375e-01 -2.46613845e-01 -7.74286807e-01
-9.69709456e-01 5.29265702e-01 1.30294752e+00 3.67738575e-01
8.90612185e-01 1.45073548e-01 8.32720935e-01 1.84736684e-01
5.53746998e-01 -1.30707288e+00 2.94745654e-01 3.86835009e-01
8.97981048e-01 -1.61381638e+00 2.32563078e-01 -5.99372447e-01
-4.96754229e-01 9.22622681e-01 1.29846370e+00 -1.19427226e-01
2.47864291e-01 3.95765603e-01 2.96831191e-01 -1.82929814e-01
-5.28273404e-01 -4.98625308e-01 3.34136367e-01 5.83854198e-01
4.49694574e-01 2.73826346e-02 -6.34754300e-01 6.50777161e-01
-1.05599694e-01 -5.10243326e-02 3.75752926e-01 1.00900900e+00
-3.69489253e-01 -8.13027501e-01 -7.40427971e-01 3.41603518e-01
-5.07791817e-01 5.68602383e-02 -1.84485205e-02 1.23568237e+00
6.09332919e-01 7.17144608e-01 -1.50428601e-02 -2.57096678e-01
7.31556475e-01 -3.59456867e-01 7.00025618e-01 -4.89991426e-01
-4.87410694e-01 -3.09009761e-01 2.10937470e-01 -9.05482054e-01
-5.45990109e-01 -6.34528637e-01 -1.11139536e+00 -4.42603290e-01
-5.25821626e-01 -2.67510891e-01 6.27652347e-01 5.53436995e-01
1.74633443e-01 7.74637640e-01 1.95389107e-01 -1.04349458e+00
-6.67899430e-01 -1.37295389e+00 -4.86055136e-01 7.07713246e-01
-2.32987478e-01 -9.01097000e-01 -5.57809919e-02 1.81825981e-01] | [8.92793083190918, 0.02569383569061756] |
fca0c229-f233-4d99-83f8-6261652c9e50 | model-based-deep-learning-1 | 2306.04469 | null | https://arxiv.org/abs/2306.04469v1 | https://arxiv.org/pdf/2306.04469v1.pdf | Model-Based Deep Learning | Signal processing traditionally relies on classical statistical modeling techniques. Such model-based methods utilize mathematical formulations that represent the underlying physics, prior information and additional domain knowledge. Simple classical models are useful but sensitive to inaccuracies and may lead to poor performance when real systems display complex or dynamic behavior. More recently, deep learning approaches that use deep neural networks are becoming increasingly popular. Deep learning systems do not rely on mathematical modeling, and learn their mapping from data, which allows them to operate in complex environments. However, they lack the interpretability and reliability of model-based methods, typically require large training sets to obtain good performance, and tend to be computationally complex. Model-based signal processing methods and data-centric deep learning each have their pros and cons. These paradigms can be characterized as edges of a continuous spectrum varying in specificity and parameterization. The methodologies that lie in the middle ground of this spectrum, thus integrating model-based signal processing with deep learning, are referred to as model-based deep learning, and are the focus here. This monograph provides a tutorial style presentation of model-based deep learning methodologies. These are families of algorithms that combine principled mathematical models with data-driven systems to benefit from the advantages of both approaches. Such model-based deep learning methods exploit both partial domain knowledge, via mathematical structures designed for specific problems, as well as learning from limited data. We accompany our presentation with running examples, in super-resolution, dynamic systems, and array processing. We show how they are expressed using the provided characterization and specialized in each of the detailed methodologies. | ['Yonina C. Eldar', 'Nir Shlezinger'] | 2023-06-05 | null | null | null | null | ['super-resolution', 'specificity'] | ['computer-vision', 'natural-language-processing'] | [ 2.59858370e-01 -4.21265751e-01 -1.29991323e-01 -2.50634134e-01
-5.35730720e-01 -3.94143283e-01 4.18825448e-01 -4.24733534e-02
-1.65133953e-01 5.95363915e-01 -2.25988343e-01 -1.31043240e-01
-5.64702153e-01 -7.73096859e-01 -6.67994380e-01 -8.60336125e-01
-5.39949477e-01 1.57371491e-01 -5.02625406e-02 -4.67422307e-01
-4.13311422e-02 6.80940747e-01 -1.16210723e+00 2.74124801e-01
5.46298027e-01 1.45004821e+00 4.72699590e-02 6.03396595e-01
-2.31664345e-01 4.86873627e-01 -6.46190643e-01 7.50056058e-02
2.08392620e-01 -3.45218122e-01 4.57823016e-02 -2.20088005e-01
5.98250814e-02 -2.37318203e-01 -5.11170745e-01 8.83993924e-01
7.69106746e-01 -1.64741680e-01 6.46780729e-01 -1.06236172e+00
-3.36689055e-01 4.84686434e-01 -5.19739449e-01 4.53377634e-01
3.86564024e-02 1.06424466e-01 7.58471847e-01 -8.61764610e-01
3.76229584e-02 1.04316902e+00 1.19932938e+00 4.31010067e-01
-1.55946422e+00 -6.20304942e-01 -3.35602947e-02 3.63649547e-01
-1.23582339e+00 -4.74074185e-01 1.29210734e+00 -5.74872077e-01
8.34460616e-01 7.48838410e-02 7.23546326e-01 1.35276413e+00
4.97506768e-01 7.15244412e-01 1.08713603e+00 -4.67341989e-01
4.48602676e-01 1.08428858e-01 2.66598254e-01 5.52472591e-01
3.51476371e-01 7.34618247e-01 -6.50368214e-01 -1.89677387e-01
8.97073567e-01 -3.91770676e-02 -3.97357225e-01 -5.94665051e-01
-8.81494999e-01 7.50265360e-01 3.74925077e-01 5.64180434e-01
-4.79832739e-01 2.97333241e-01 6.16712689e-01 5.33593059e-01
1.65948644e-01 6.59405589e-01 -5.37057698e-01 1.37151450e-01
-1.30021834e+00 3.48842710e-01 8.48636448e-01 6.49070144e-01
6.16008937e-01 6.67214692e-01 1.64352551e-01 7.72457480e-01
2.25779921e-01 5.25565147e-01 6.35497630e-01 -6.78133368e-01
-8.54553133e-02 1.92515016e-01 8.81490931e-02 -1.14055514e+00
-1.01009715e+00 -9.86591935e-01 -1.25879478e+00 3.39503646e-01
1.94525868e-01 -2.89322466e-01 -9.01047230e-01 1.70495296e+00
-1.56282201e-01 1.95354730e-01 6.06002584e-02 8.21386039e-01
7.62636185e-01 7.44035363e-01 -1.22817457e-01 -3.22330654e-01
1.09622025e+00 -3.74051988e-01 -8.83985162e-01 -1.02519140e-01
1.71159118e-01 -5.33568799e-01 7.74772167e-01 7.89314151e-01
-1.17205787e+00 -8.50199699e-01 -1.55630422e+00 2.71494448e-01
-5.47087491e-01 -1.68791991e-02 5.48398137e-01 7.97448933e-01
-1.09792960e+00 8.53861988e-01 -1.08064079e+00 -9.08751339e-02
4.23129648e-01 4.88794982e-01 1.62885025e-01 3.63444686e-01
-1.49209905e+00 8.42967808e-01 1.84169367e-01 4.41719055e-01
-9.84837055e-01 -1.11258912e+00 -6.88168526e-01 1.80579945e-01
1.69243187e-01 -6.42291248e-01 1.24810338e+00 -9.67706323e-01
-1.71330047e+00 4.60778147e-01 1.21012673e-01 -8.65601897e-01
4.18439150e-01 -2.89268196e-01 -7.39343941e-01 1.86196253e-01
-5.36966622e-01 9.68338642e-03 1.34884965e+00 -1.43506968e+00
-1.65953159e-01 -7.53040165e-02 3.47790420e-02 -3.75603259e-01
-2.83744961e-01 -2.37256244e-01 1.31355584e-01 -7.36739039e-01
1.74346909e-01 -4.59963083e-01 -2.72053778e-01 5.01636136e-03
-2.41400629e-01 2.77795702e-01 9.00205791e-01 -4.10723597e-01
1.30654740e+00 -2.09555697e+00 -3.37450989e-02 4.07780558e-01
3.44800383e-01 5.25776625e-01 -7.82710537e-02 6.71877265e-01
-2.68629551e-01 3.60178277e-02 -3.31129611e-01 7.86634982e-02
5.98007515e-02 -6.46315515e-02 -3.81778240e-01 3.67436707e-01
2.61971772e-01 8.48148704e-01 -6.75099254e-01 -1.80964664e-01
6.04069889e-01 5.83003640e-01 -2.33877569e-01 4.79737446e-02
-1.00987971e-01 3.63048404e-01 -4.05912817e-01 5.17733216e-01
7.25631356e-01 -1.55471087e-01 1.05599090e-01 -7.89003968e-01
5.66785922e-03 -4.38433923e-02 -1.25990701e+00 1.49288011e+00
-7.77317166e-01 8.62717330e-01 5.18886685e-01 -1.82392848e+00
9.71914291e-01 4.51093107e-01 7.50612199e-01 -7.51953423e-01
2.43667528e-01 2.87545532e-01 2.54810035e-01 -5.49735248e-01
-9.96385440e-02 -6.58554077e-01 1.20345786e-01 2.44402602e-01
2.50742823e-01 -2.13854998e-01 -2.52524856e-02 -2.24062413e-01
9.20966923e-01 7.43594915e-02 6.18198574e-01 -3.87297571e-01
5.36462307e-01 -2.57452846e-01 5.04242778e-01 1.00185609e+00
-2.15617463e-01 4.24109221e-01 3.68147016e-01 -6.55879259e-01
-9.25524533e-01 -1.17436910e+00 -4.64885533e-01 7.11099923e-01
-3.03415656e-02 -1.37512729e-01 -5.01165748e-01 1.96293965e-02
-1.03723168e-01 3.70701075e-01 -4.87451136e-01 -3.98677528e-01
-7.27868915e-01 -1.17449450e+00 4.53717768e-01 6.09556973e-01
4.17478859e-01 -8.43037009e-01 -7.43819058e-01 6.50979638e-01
2.44184777e-01 -1.02388394e+00 4.06905115e-01 5.17388225e-01
-9.86718059e-01 -9.33332443e-01 -6.68660104e-01 -4.67358202e-01
1.34700254e-01 -1.14255831e-01 1.17048776e+00 -1.45509169e-01
-4.41390425e-01 5.79228878e-01 -1.66791677e-01 -7.73802221e-01
-3.93523425e-01 -1.26605377e-01 4.62987810e-01 1.67367786e-01
2.82335550e-01 -1.12280142e+00 -4.61074054e-01 2.91658798e-03
-8.23457420e-01 -2.32854933e-01 8.49659443e-01 1.11468863e+00
4.91585195e-01 2.63102382e-01 8.98367882e-01 -6.51837468e-01
7.24232137e-01 -3.35706174e-01 -6.87547386e-01 6.67655170e-02
-4.09669340e-01 6.85478523e-02 9.55588937e-01 -4.21133906e-01
-8.42978120e-01 -5.10574803e-02 -1.85384095e-01 -5.94304502e-01
-2.09416106e-01 8.03800404e-01 -2.13800520e-01 -2.10230440e-01
1.01828647e+00 5.11569679e-01 1.31055593e-01 -4.06275660e-01
8.96626189e-02 4.26501274e-01 4.23085302e-01 -6.19901538e-01
8.64974201e-01 5.76528728e-01 3.17766756e-01 -1.12981832e+00
-7.80553460e-01 -1.25735566e-01 -6.65199459e-01 -1.80978939e-01
3.85341495e-01 -6.67484879e-01 -4.98196602e-01 5.76874733e-01
-1.07492852e+00 -3.62475783e-01 -3.46536011e-01 5.92063546e-01
-7.88829327e-01 1.27407014e-01 -5.43410122e-01 -9.48646903e-01
-2.51777649e-01 -8.57806563e-01 8.73331845e-01 1.32788822e-01
-1.43227622e-01 -1.40202487e+00 -8.86045322e-02 -3.13168198e-01
8.66469204e-01 5.69183528e-01 1.13711846e+00 -5.45093358e-01
-2.46102378e-01 -3.71608555e-01 -2.01442409e-02 6.26296818e-01
1.39779776e-01 -1.68268159e-01 -1.21982443e+00 -2.94339746e-01
3.82856846e-01 -1.16212800e-01 8.19524407e-01 1.07064402e+00
1.37369084e+00 -7.58715998e-03 -3.90382916e-01 6.89037144e-01
1.62803841e+00 1.94836333e-01 2.85091460e-01 1.24301568e-01
3.65088075e-01 5.68044662e-01 2.98486613e-02 4.82701778e-01
-4.13376749e-01 6.70035720e-01 3.11478764e-01 -3.25981200e-01
-4.73682676e-03 1.37066856e-01 3.19928646e-01 8.01039040e-01
-4.24544886e-02 3.96279506e-02 -8.18784595e-01 2.06586987e-01
-1.60272181e+00 -1.15888011e+00 -1.38412759e-01 1.99346387e+00
7.10874200e-01 4.18276608e-01 2.18781561e-01 5.48573494e-01
6.09892190e-01 1.03477620e-01 -8.48279476e-01 -2.63474166e-01
-1.67429969e-01 3.76274258e-01 3.11895013e-01 3.74746948e-01
-1.09490275e+00 5.12046874e-01 7.31324911e+00 7.85266757e-01
-1.55517864e+00 -4.31281049e-03 3.16405237e-01 -1.29313543e-01
3.73127200e-02 -5.47239125e-01 -4.53971148e-01 3.34432065e-01
1.26338923e+00 -2.73796171e-01 3.22464526e-01 7.48943210e-01
5.62844038e-01 1.78535476e-01 -1.37502062e+00 1.33534884e+00
-2.44454816e-01 -1.38788605e+00 -1.19352244e-01 -8.34228769e-02
3.67847115e-01 -8.99842381e-02 3.46193165e-01 4.17054355e-01
-1.96965128e-01 -1.18446732e+00 6.27469540e-01 6.98362410e-01
4.99646425e-01 -4.45311487e-01 7.40366936e-01 3.44434112e-01
-9.69774246e-01 -3.49302679e-01 -3.61843169e-01 -1.90485001e-01
2.61161387e-01 1.05650330e+00 -2.02026889e-01 6.69634759e-01
6.42594457e-01 7.97114015e-01 -1.85622778e-02 1.10905814e+00
1.21883787e-01 7.76594222e-01 -3.93426478e-01 -1.09824941e-01
5.23954332e-02 -2.53374735e-03 7.36036301e-01 1.36858416e+00
3.14756066e-01 -7.17643201e-02 8.14934522e-02 1.18795252e+00
3.13873768e-01 -3.19958031e-01 -6.40633523e-01 -1.32550567e-01
3.91002715e-01 1.16709161e+00 -4.57036495e-01 -3.47992361e-01
-6.15578651e-01 3.16955745e-01 -2.36009479e-01 5.75896323e-01
-8.36091638e-01 -3.73307467e-01 6.13615394e-01 2.57301211e-01
4.02666181e-02 -4.48032707e-01 -5.64933896e-01 -8.40972543e-01
-9.34367925e-02 -9.36770558e-01 3.88641134e-02 -4.91086394e-01
-1.57006216e+00 4.68297720e-01 2.43943885e-01 -1.43503952e+00
-3.34243685e-01 -1.20677674e+00 -5.49505115e-01 7.45364487e-01
-1.62320709e+00 -8.23942006e-01 -2.73742855e-01 4.57456201e-01
3.90514106e-01 -3.93278092e-01 8.42323661e-01 3.97443026e-01
-3.96088809e-01 4.09297854e-01 3.38263601e-01 5.45009747e-02
3.67572874e-01 -1.26709330e+00 7.37021631e-03 5.56406677e-01
5.72265461e-02 6.44978344e-01 9.85370159e-01 3.21692154e-02
-1.53165269e+00 -8.51252556e-01 1.31021693e-01 -2.14004040e-01
7.57057309e-01 -6.51297569e-01 -1.05832791e+00 2.81681865e-01
-9.96343642e-02 4.00426984e-02 8.95977199e-01 2.45938614e-01
-1.31175712e-01 -6.02300942e-01 -9.84440207e-01 3.28335464e-01
8.16382051e-01 -4.09418881e-01 -6.60535038e-01 2.58917421e-01
2.42281631e-01 -3.06316376e-01 -8.01255882e-01 4.49307114e-01
7.33834386e-01 -1.16728759e+00 1.19859886e+00 -4.03869390e-01
8.13533589e-02 -1.97940558e-01 -1.18110836e-01 -1.47606146e+00
-4.51635897e-01 -6.82653606e-01 -4.49646473e-01 7.26765275e-01
3.56094033e-01 -5.86771190e-01 6.18867278e-01 2.03057766e-01
-2.60367393e-01 -7.29884863e-01 -8.33133698e-01 -1.22212434e+00
3.14116627e-01 -7.76195228e-01 5.15695274e-01 8.32749069e-01
-1.02278091e-01 1.26072735e-01 -3.08281243e-01 3.45897734e-01
7.61480153e-01 1.05248913e-01 4.49561685e-01 -1.46616793e+00
-3.55674744e-01 -8.05251122e-01 -6.01325631e-01 -8.14859211e-01
4.23981957e-02 -5.30587733e-01 -7.34182745e-02 -1.40836871e+00
-2.39437371e-01 -3.69760662e-01 -7.75194466e-01 1.54431328e-01
3.17527860e-01 5.82423322e-02 -1.22384906e-01 1.87072441e-01
-6.29294384e-03 4.58664954e-01 8.14430833e-01 -3.18071663e-01
-4.36770618e-01 1.19751409e-01 -5.96137226e-01 8.90202999e-01
1.02189040e+00 -1.07174560e-01 -5.08634984e-01 -2.75456429e-01
1.88740581e-01 3.39123569e-02 6.73731685e-01 -1.54951286e+00
1.68570921e-01 -1.80836823e-02 7.46674120e-01 -3.10091317e-01
5.85542560e-01 -1.05665338e+00 -6.01750538e-02 7.27784753e-01
-3.27019989e-01 -3.23334038e-01 4.56538618e-01 5.71648300e-01
-3.80843461e-01 -2.00527683e-02 1.11014962e+00 -1.80563703e-01
-7.59232819e-01 2.29479119e-01 -5.17370820e-01 -2.22053200e-01
7.17538536e-01 -4.01400387e-01 2.65458882e-01 -5.96447945e-01
-1.10368371e+00 -1.15004621e-01 -1.89489603e-01 2.10927010e-01
4.98096794e-01 -1.25697482e+00 -6.27211869e-01 4.04071420e-01
-1.97718367e-01 -3.64300668e-01 4.34714079e-01 9.81747031e-01
-2.72651553e-01 5.96497893e-01 -4.51184571e-01 -8.47898483e-01
-6.37959599e-01 7.00983584e-01 9.90593612e-01 -1.74455076e-01
-4.89456236e-01 6.28363431e-01 3.58521134e-01 -3.08775842e-01
1.55641302e-01 -5.77077270e-01 -8.26121196e-02 -3.55347060e-02
6.01058364e-01 2.11296797e-01 2.26440698e-01 -1.43332288e-01
-3.65483761e-01 7.14453220e-01 1.87185138e-01 -1.66752011e-01
1.31216502e+00 8.36661682e-02 1.57022908e-01 6.60930872e-01
1.07955468e+00 -2.51528561e-01 -1.25296462e+00 -4.08359915e-01
6.98109791e-02 7.91415200e-02 4.39154536e-01 -9.64730263e-01
-9.73947823e-01 1.45939255e+00 9.46862817e-01 6.44017994e-01
1.51454520e+00 -4.11082268e-01 6.06469393e-01 3.29161942e-01
2.88010746e-01 -1.20426595e+00 8.47483426e-02 3.80281508e-01
1.09027410e+00 -1.03453124e+00 -3.19970539e-03 -2.85726845e-01
1.06170654e-01 1.56926596e+00 4.86699522e-01 -2.62922585e-01
1.21717072e+00 7.65278339e-01 7.80889019e-02 -6.49798736e-02
-5.52833080e-01 1.25719262e-02 2.62056410e-01 1.04495716e+00
4.98480976e-01 -5.69221862e-02 -1.23831466e-01 1.09912276e+00
-2.70834535e-01 1.93599850e-01 2.20879748e-01 8.76179993e-01
-5.95333397e-01 -9.63111281e-01 -4.52753395e-01 4.74849850e-01
-3.96413058e-01 -1.11995079e-01 -5.60854562e-02 9.28004444e-01
5.84765002e-02 8.75615597e-01 -9.04189572e-02 -3.95938814e-01
3.57150495e-01 7.68260211e-02 6.07457042e-01 -3.29732478e-01
-2.77672350e-01 1.89755708e-01 -1.98072463e-01 -5.20112634e-01
-4.47918028e-01 -4.37549680e-01 -1.05643308e+00 -2.34258801e-01
-8.50496665e-02 -1.92132309e-01 6.87939525e-01 1.02508259e+00
3.43400806e-01 8.05455804e-01 4.63551521e-01 -1.14594293e+00
-7.73314059e-01 -9.67989326e-01 -7.46582508e-01 -6.44601136e-02
7.75167525e-01 -8.26643586e-01 -2.96623856e-01 1.22570850e-01] | [8.197980880737305, 2.272075891494751] |
7e46056a-1919-42fd-b0af-0eac4b7e243f | how-local-is-the-local-diversity-reinforcing | 1807.04219 | null | http://arxiv.org/abs/1807.04219v4 | http://arxiv.org/pdf/1807.04219v4.pdf | How Local is the Local Diversity? Reinforcing Sequential Determinantal Point Processes with Dynamic Ground Sets for Supervised Video Summarization | The large volume of video content and high viewing frequency demand automatic
video summarization algorithms, of which a key property is the capability of
modeling diversity. If videos are lengthy like hours-long egocentric videos, it
is necessary to track the temporal structures of the videos and enforce local
diversity. The local diversity refers to that the shots selected from a short
time duration are diverse but visually similar shots are allowed to co-exist in
the summary if they appear far apart in the video. In this paper, we propose a
novel probabilistic model, built upon SeqDPP, to dynamically control the time
span of a video segment upon which the local diversity is imposed. In
particular, we enable SeqDPP to learn to automatically infer how local the
local diversity is supposed to be from the input video. The resulting model is
extremely involved to train by the hallmark maximum likelihood estimation
(MLE), which further suffers from the exposure bias and non-differentiable
evaluation metrics. To tackle these problems, we instead devise a reinforcement
learning algorithm for training the proposed model. Extensive experiments
verify the advantages of our model and the new learning algorithm over
MLE-based methods. | ['Liqiang Wang', 'Tianbao Yang', 'Boqing Gong', 'Yandong Li'] | 2018-07-11 | how-local-is-the-local-diversity-reinforcing-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Yandong_Li_How_Local_is_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Yandong_Li_How_Local_is_ECCV_2018_paper.pdf | eccv-2018-9 | ['supervised-video-summarization'] | ['computer-vision'] | [ 1.11729644e-01 -5.04571907e-02 -3.00221890e-01 -1.63798407e-01
-6.16200447e-01 -3.68460417e-01 4.57466781e-01 1.87040679e-02
-3.24095845e-01 8.27754080e-01 4.19743150e-01 1.64256915e-01
-3.40681583e-01 -5.90653718e-01 -8.72768819e-01 -9.03127968e-01
-2.45621219e-01 8.66464004e-02 3.27945560e-01 1.90898567e-01
4.11838323e-01 3.01975012e-01 -1.57566738e+00 4.85847294e-02
1.09896672e+00 7.95207024e-01 6.81189001e-01 6.57343149e-01
9.49582979e-02 1.00750685e+00 -5.75558722e-01 -1.11749597e-01
9.48196948e-02 -7.11227298e-01 -4.47509497e-01 5.00955999e-01
3.25526267e-01 -4.44055140e-01 -4.52050745e-01 9.76723433e-01
5.08944213e-01 5.80858648e-01 7.36014485e-01 -1.28099310e+00
-8.24493170e-02 4.60287362e-01 -5.87521791e-01 4.93445188e-01
5.06076276e-01 9.83769968e-02 9.49538648e-01 -6.38737202e-01
8.40223908e-01 8.98289800e-01 3.48861158e-01 4.07779843e-01
-9.49011505e-01 -1.89504489e-01 4.91104245e-01 5.57111979e-01
-1.49688542e+00 -5.58285236e-01 1.00124013e+00 -4.85675514e-01
4.71888483e-01 1.38717830e-01 6.80978775e-01 1.04061508e+00
2.17022285e-01 9.89260018e-01 6.51163697e-01 -2.65235364e-01
5.16479731e-01 9.81282350e-03 -3.01767409e-01 6.02203131e-01
3.30476314e-02 -3.60398293e-01 -6.64959848e-01 -5.43691516e-02
6.38955653e-01 1.93702150e-02 -4.36456352e-01 -6.41038954e-01
-1.15749490e+00 6.08576417e-01 4.60190326e-02 1.54866010e-01
-6.83025301e-01 3.10178902e-02 5.17243981e-01 1.70963749e-01
2.32826263e-01 2.85913408e-01 -6.53092489e-02 -2.64803380e-01
-1.19959223e+00 4.47057694e-01 6.95610225e-01 1.08768165e+00
7.58767426e-01 -1.22141756e-01 -3.25759232e-01 6.59776986e-01
-6.46226779e-02 1.20968223e-01 5.10614753e-01 -1.10997522e+00
5.06457329e-01 3.48541707e-01 2.74930030e-01 -1.23469579e+00
2.57498752e-02 -2.77017057e-01 -7.97563910e-01 -1.50273904e-01
1.02031507e-01 -2.55306512e-01 -4.67848688e-01 1.83325744e+00
3.34969878e-01 4.05350059e-01 -2.27516573e-02 8.66631389e-01
3.35099638e-01 1.15378070e+00 -1.65542532e-02 -9.12131429e-01
8.38197947e-01 -8.38073850e-01 -8.25101674e-01 1.76294640e-01
3.68274450e-01 -4.81819153e-01 8.89367104e-01 3.36207062e-01
-1.20137405e+00 -5.24157524e-01 -1.12767458e+00 4.26783383e-01
1.95688650e-01 1.60978716e-02 7.96326175e-02 2.14212283e-01
-8.66111577e-01 7.15459228e-01 -7.77936816e-01 -2.94121504e-01
3.17136586e-01 1.63457580e-02 -1.74490765e-01 -1.26386404e-01
-1.05010700e+00 5.78921080e-01 7.74869502e-01 -4.55887169e-02
-1.01479149e+00 -3.61751586e-01 -6.87101364e-01 3.66625279e-01
9.34109807e-01 -5.85356474e-01 1.09041643e+00 -1.29030073e+00
-1.61390269e+00 3.22378367e-01 -2.26613328e-01 -4.06860441e-01
7.87396491e-01 -2.11748362e-01 -2.20310673e-01 6.67067349e-01
6.62688166e-02 4.81300026e-01 1.16880786e+00 -1.06090462e+00
-8.94159019e-01 4.15923707e-02 1.04607835e-01 6.30834401e-01
-4.41898227e-01 -3.51735167e-02 -6.33999527e-01 -6.44308567e-01
-1.23020723e-01 -7.07332313e-01 -5.72226606e-02 -2.65291721e-01
-2.75126666e-01 -3.51995438e-01 7.76757538e-01 -6.36353433e-01
1.60133243e+00 -2.15452266e+00 4.85717863e-01 -4.16263677e-02
-5.94802434e-03 1.76584750e-01 9.90818813e-02 6.51674926e-01
3.07153672e-01 -1.14974894e-01 -8.60545784e-02 -2.34220326e-02
-5.85062020e-02 2.16332644e-01 -2.60236204e-01 4.42551702e-01
7.19469339e-02 3.42184097e-01 -1.02133882e+00 -8.35281491e-01
1.56416699e-01 1.76790476e-01 -7.31588483e-01 6.05527878e-01
-3.96855831e-01 5.23151875e-01 -6.00201547e-01 3.29003394e-01
6.07264996e-01 -9.36953574e-02 2.68764734e-01 -2.54168123e-01
-2.87168920e-01 -5.62729314e-02 -1.27434301e+00 1.72126782e+00
-1.61100686e-01 4.23759550e-01 -1.66513816e-01 -1.12191689e+00
6.26534641e-01 3.54075670e-01 5.34098506e-01 -4.57153738e-01
-2.57792026e-02 -1.06614254e-01 -3.05865824e-01 -9.71369267e-01
5.90666890e-01 6.87577352e-02 -5.03780805e-02 3.13588083e-01
6.05656058e-02 2.47360036e-01 5.96537292e-01 3.62776279e-01
8.75688672e-01 3.70485753e-01 4.79873270e-01 -1.21718563e-01
6.92088246e-01 -3.35402966e-01 9.39673245e-01 8.20658803e-01
-3.00372064e-01 3.96792233e-01 7.99806476e-01 -2.06341296e-01
-1.01002204e+00 -1.04129827e+00 1.68436915e-01 1.07482505e+00
4.48972404e-01 -4.96776640e-01 -8.53364110e-01 -6.56198084e-01
-4.99937534e-01 7.05882668e-01 -2.63473779e-01 -2.34984890e-01
-6.07401788e-01 -2.79786199e-01 7.34331878e-03 1.99790463e-01
4.39304709e-01 -9.62262452e-01 -8.80255163e-01 2.73540169e-01
-4.98314500e-01 -1.03338945e+00 -7.41723061e-01 -2.36427292e-01
-7.02708364e-01 -8.99212778e-01 -9.41917717e-01 -6.89923406e-01
5.34038007e-01 3.80493313e-01 8.76516998e-01 -3.36658508e-01
1.10476434e-01 4.06702280e-01 -6.30959928e-01 -6.59286678e-02
-3.76685321e-01 7.46195167e-02 1.64156884e-01 3.57903898e-01
9.69336256e-02 -7.14892924e-01 -7.49626338e-01 3.22720647e-01
-1.03405321e+00 2.65835643e-01 5.42269588e-01 7.96952784e-01
7.27417886e-01 3.55242074e-01 8.07664812e-01 -6.03769004e-01
3.82714957e-01 -7.15711296e-01 -3.69108796e-01 5.13128757e-01
-2.01498181e-01 -9.48034823e-02 7.81971633e-01 -5.14241338e-01
-1.06555569e+00 -4.06817496e-02 1.30670175e-01 -6.25594497e-01
-8.19450840e-02 5.65940380e-01 -3.89081091e-01 4.96122450e-01
2.41872832e-01 5.71377814e-01 -5.94691075e-02 -1.60237849e-01
2.12999225e-01 5.74280620e-01 4.99072134e-01 -4.45231438e-01
4.20884103e-01 3.03376228e-01 -2.54823536e-01 -1.00718808e+00
-6.12910628e-01 -5.22398174e-01 -5.43225110e-01 -6.77122951e-01
8.44079912e-01 -9.91616368e-01 -5.44970334e-01 3.70614529e-01
-9.98712301e-01 -6.61294907e-02 -1.84613094e-01 6.40165389e-01
-9.81543243e-01 7.85569429e-01 -2.07241789e-01 -8.97695482e-01
-9.15862992e-02 -1.00151145e+00 6.26190960e-01 4.60392475e-01
-1.34495467e-01 -6.83263421e-01 8.75433683e-02 -9.47065465e-03
5.34327403e-02 2.91968763e-01 7.95021653e-01 -5.11498928e-01
-7.57384062e-01 -1.17621310e-01 7.55726546e-02 3.90011162e-01
2.07415029e-01 2.89081544e-01 -5.87194085e-01 -4.20742929e-01
9.28791910e-02 -2.26665080e-01 6.69315696e-01 6.40957952e-01
1.27463293e+00 -6.27156854e-01 -1.29739657e-01 3.85196716e-01
1.35522270e+00 2.64431685e-01 5.67016602e-01 1.67038769e-01
5.66646397e-01 7.08806515e-01 9.41085398e-01 8.15704882e-01
2.54560649e-01 8.18160474e-01 4.32962447e-01 3.60014051e-01
2.51440555e-01 -4.23610598e-01 6.14333689e-01 1.03813422e+00
-3.69869545e-02 -5.83871484e-01 -3.61465842e-01 5.81506371e-01
-2.17645812e+00 -1.46886230e+00 3.74695867e-01 2.43897605e+00
7.68098235e-01 1.17782727e-01 3.22811276e-01 -1.06246896e-01
1.00887942e+00 4.86671746e-01 -6.38045132e-01 -2.23441586e-01
6.03398271e-02 -5.40573835e-01 1.40023440e-01 2.55474299e-01
-1.11746919e+00 6.89114094e-01 5.82289743e+00 1.20834816e+00
-8.42022598e-01 -1.60852969e-01 4.60661799e-01 -3.96970093e-01
-1.96703136e-01 -2.37437300e-02 -5.81318140e-01 9.00584638e-01
8.22140694e-01 -4.38799024e-01 3.01854879e-01 8.10157835e-01
6.21362567e-01 -4.97115105e-01 -1.11285174e+00 1.03823614e+00
2.51816779e-01 -1.23502719e+00 2.97592103e-01 -2.00390741e-01
8.52192044e-01 -4.34947342e-01 -1.44414723e-01 2.02969417e-01
-2.84658998e-01 -5.52859306e-01 7.58621693e-01 7.88727760e-01
4.67528582e-01 -1.10799587e+00 3.88847262e-01 7.16018081e-01
-1.04195035e+00 -1.66193485e-01 -5.89109480e-01 2.06068084e-01
4.12844926e-01 5.69509506e-01 -6.83358014e-01 6.73692226e-01
4.14608598e-01 8.59666586e-01 -2.20737278e-01 1.20115566e+00
-1.44403741e-01 5.30231297e-01 -8.50500632e-03 -2.38300432e-02
2.08938479e-01 -3.38380307e-01 9.26341772e-01 1.25068653e+00
6.95790470e-01 8.52578320e-03 4.98780221e-01 4.71236080e-01
-5.02937324e-02 1.24259934e-01 -6.02915287e-01 -1.15030952e-01
5.98097920e-01 1.07312655e+00 -6.32241249e-01 -4.44247186e-01
-3.42660159e-01 1.08578742e+00 1.76142335e-01 3.56497526e-01
-1.02696359e+00 -3.06754023e-01 3.62229794e-01 4.93444838e-02
6.23311043e-01 -8.68122280e-02 3.46870154e-01 -1.25278878e+00
1.79370746e-01 -8.10158372e-01 4.54089046e-01 -7.61182010e-01
-1.08164394e+00 5.08074582e-01 1.87563479e-01 -1.63155293e+00
-4.69001740e-01 1.31221898e-02 -7.26058185e-01 4.41486031e-01
-1.38498712e+00 -6.24083936e-01 -2.00881332e-01 6.12823606e-01
9.98530507e-01 -2.10926712e-01 2.86205083e-01 1.89406022e-01
-6.96352184e-01 3.91821712e-01 2.51027942e-01 -3.04438084e-01
7.56381869e-01 -1.01035678e+00 -3.17798972e-01 1.03250241e+00
-7.14344159e-02 2.00388849e-01 1.07061160e+00 -6.11606300e-01
-1.28955555e+00 -9.93565559e-01 6.79360688e-01 6.56566843e-02
5.09489834e-01 -2.97691599e-02 -9.11472142e-01 4.84672338e-01
3.25667650e-01 -4.15296584e-01 6.50754750e-01 -2.68892676e-01
-5.36379870e-03 -1.99414924e-01 -7.91362345e-01 6.70812666e-01
8.97204101e-01 -3.29170287e-01 -7.24707186e-01 2.94982523e-01
6.76289856e-01 -9.90163907e-02 -5.66555798e-01 3.03655356e-01
4.37242776e-01 -1.21999288e+00 6.56696796e-01 -3.69005591e-01
6.44426405e-01 -3.41172367e-01 -1.68261170e-01 -1.35338938e+00
-3.69872779e-01 -8.17200899e-01 -4.66858864e-01 1.41351140e+00
-1.04402013e-01 -1.17119029e-01 6.57414496e-01 2.65749097e-01
-1.43733755e-01 -6.72605991e-01 -9.94624853e-01 -8.32670271e-01
-4.58002716e-01 5.97108342e-02 3.04366291e-01 7.39348471e-01
9.01220739e-02 1.53779224e-01 -8.17945957e-01 2.74462312e-01
5.91456234e-01 2.24200353e-01 7.87746847e-01 -8.79498422e-01
-3.95088673e-01 -3.47470015e-01 -4.11453635e-01 -1.27486956e+00
7.96506330e-02 -4.39379454e-01 2.98299611e-01 -1.35615706e+00
5.16210437e-01 1.56806726e-02 -4.53717291e-01 -1.13893799e-01
-3.06552827e-01 -4.21815395e-01 1.37180194e-01 3.50793511e-01
-1.16573966e+00 8.66035402e-01 1.11689377e+00 -4.60014585e-03
-5.08299589e-01 1.80015355e-01 -3.34741175e-01 7.25782633e-01
6.76777065e-01 -4.47257996e-01 -7.81129777e-01 -1.93458349e-01
2.20075592e-01 4.90209401e-01 3.68590653e-02 -1.05284655e+00
3.87394577e-01 -3.74403656e-01 7.16341957e-02 -8.00341487e-01
2.09821180e-01 -7.79001713e-01 2.13722333e-01 3.74891490e-01
-4.92346078e-01 6.20802771e-03 -2.84462035e-01 8.92065048e-01
-3.57749492e-01 -4.59788918e-01 7.26453185e-01 -1.76629469e-01
-9.02842045e-01 4.21351701e-01 -7.02845573e-01 1.33235948e-02
1.31048012e+00 -3.43368143e-01 1.56179741e-01 -6.32573962e-01
-5.57735085e-01 3.63433897e-01 4.53373581e-01 2.16920927e-01
6.10878408e-01 -1.33959734e+00 -7.50044107e-01 -9.69712883e-02
1.86534643e-01 -4.08136509e-02 7.33915210e-01 7.77297318e-01
-3.96975398e-01 6.63748309e-02 -3.30896616e-01 -6.51080966e-01
-1.15959680e+00 7.18422592e-01 -1.10158339e-01 -2.38674060e-01
-7.03693509e-01 5.09692550e-01 3.55604589e-01 2.73192644e-01
3.46584648e-01 4.57348526e-02 -4.42555249e-01 3.19042295e-01
6.68594301e-01 6.00891531e-01 -3.70416582e-01 -5.79372227e-01
-7.46611655e-02 4.39201236e-01 -2.89253861e-01 4.53574099e-02
1.29071414e+00 -6.45881951e-01 1.40100598e-01 5.86022139e-01
1.12573218e+00 6.19734451e-03 -1.81669772e+00 -2.82592058e-01
-1.46452278e-01 -7.08313525e-01 -1.20830350e-01 -2.46617392e-01
-7.80158401e-01 5.94443560e-01 3.02629977e-01 2.27864176e-01
1.35090721e+00 -5.08009195e-02 7.65415430e-01 3.73871386e-01
3.51903617e-01 -1.44879127e+00 3.00795108e-01 2.79296994e-01
8.06695640e-01 -8.98271680e-01 1.55789509e-01 -2.21541107e-01
-9.87811625e-01 9.37633276e-01 4.96797472e-01 -2.40394369e-01
2.28511229e-01 -1.96005851e-01 -4.28118587e-01 2.45851353e-02
-8.52023125e-01 6.40385002e-02 1.62833527e-01 4.17054206e-01
1.25436168e-02 -1.40892237e-01 -5.66301644e-01 3.09752941e-01
1.91586852e-01 5.03200553e-02 7.65197158e-01 8.53066981e-01
-7.15218663e-01 -6.81555390e-01 -6.48467019e-02 2.68377632e-01
-3.11736822e-01 3.12918305e-01 5.35671040e-02 4.57754672e-01
-1.10786571e-03 7.11504281e-01 8.78994614e-02 -1.72982052e-01
1.02674909e-01 7.99278393e-02 4.35094655e-01 -3.21654677e-01
3.51188593e-02 3.46227616e-01 -1.12590872e-01 -5.48614264e-01
-7.48851120e-01 -8.47920477e-01 -8.96207511e-01 -1.90274179e-01
-2.27675214e-01 2.37350792e-01 3.50276560e-01 8.39939713e-01
3.71752799e-01 5.60711324e-01 1.14513063e+00 -1.00880194e+00
-6.73793674e-01 -7.53513575e-01 -7.42409050e-01 3.85145396e-01
3.06472123e-01 -5.60726821e-01 -3.70986998e-01 3.06857347e-01] | [10.426552772521973, 0.4073573350906372] |
8c5e9c54-398e-4a13-a573-1ac23f0cdb44 | polytuplet-loss-a-reverse-approach-to | 2304.01046 | null | https://arxiv.org/abs/2304.01046v2 | https://arxiv.org/pdf/2304.01046v2.pdf | Deep Manifold Learning for Reading Comprehension and Logical Reasoning Tasks with Polytuplet Loss | The current trend in developing machine learning models for reading comprehension and logical reasoning tasks is focused on improving the models' abilities to understand and utilize logical rules. This work focuses on providing a novel loss function and accompanying model architecture that has more interpretable components than some other models by representing a common strategy employed by humans when given reading comprehension and logical reasoning tasks. This strategy involves emphasizing relative accuracy over absolute accuracy and can theoretically produce the correct answer without full knowledge of the information required to solve the question. We examine the effectiveness of applying such a strategy to train transfer learning models to solve reading comprehension and logical reasoning questions. The models were evaluated on the ReClor dataset, a challenging reading comprehension and logical reasoning benchmark. We propose the polytuplet loss function, an extension of the triplet loss function, to ensure prioritization of learning the relative correctness of answer choices over learning the true accuracy of each choice. Our results indicate that models employing polytuplet loss outperform existing baseline models. Although polytuplet loss is a promising alternative to other contrastive loss functions, further research is required to quantify the benefits it may present. | ['Ivan Rodriguez', 'Jeffrey Lu'] | 2023-04-03 | null | null | null | null | ['reading-comprehension', 'logical-reasoning'] | ['natural-language-processing', 'reasoning'] | [ 4.52276796e-01 4.85020995e-01 -1.09651484e-01 -7.32553244e-01
-7.23579288e-01 -4.36635047e-01 5.66441298e-01 4.36013401e-01
-4.36272383e-01 6.72625840e-01 2.22612813e-01 -9.86079991e-01
-3.94305050e-01 -8.92535448e-01 -8.16640258e-01 -7.25058466e-02
2.60131776e-01 6.52214170e-01 1.54023796e-01 -1.87782317e-01
5.46540618e-01 1.83506146e-01 -1.56422412e+00 6.41874671e-01
1.38381207e+00 1.02724588e+00 1.72541710e-03 8.21231067e-01
-2.88144797e-01 1.77422988e+00 -5.64452767e-01 -6.61197484e-01
1.66896097e-02 -6.82583332e-01 -1.43273628e+00 -4.39316899e-01
9.41212475e-01 -4.91582394e-01 1.77428037e-01 6.78599119e-01
1.73555806e-01 2.39578471e-01 8.61226797e-01 -1.04922771e+00
-9.33075070e-01 7.18100309e-01 8.15726146e-02 3.21324289e-01
9.89921153e-01 1.65245056e-01 1.38760126e+00 -5.63389480e-01
1.21832050e-01 1.47365940e+00 7.19589174e-01 5.11793613e-01
-1.14711022e+00 -4.16906714e-01 3.17729324e-01 7.81348646e-01
-8.89401197e-01 -2.76106119e-01 3.84936899e-01 -2.30136439e-01
1.21581268e+00 2.95396268e-01 1.90600589e-01 8.50056171e-01
2.56462634e-01 9.81456697e-01 1.34264457e+00 -8.34825635e-01
4.87135574e-02 2.80906055e-02 6.45403385e-01 8.41516852e-01
6.90819100e-02 -7.32109547e-02 -6.57356024e-01 -1.37858549e-02
3.47997129e-01 -3.96272510e-01 -4.93277252e-01 -3.02383602e-01
-8.76369774e-01 7.09775567e-01 4.27521646e-01 3.77773196e-02
-2.91541308e-01 1.14077754e-01 5.96615151e-02 8.49212587e-01
8.72825980e-02 9.10045564e-01 -6.79656565e-01 -2.87630677e-01
-9.30408716e-01 5.35991848e-01 1.18280852e+00 6.91003382e-01
3.87115061e-01 -2.05821991e-01 -6.50074720e-01 6.33212268e-01
4.24932301e-01 1.95973411e-01 1.69874504e-01 -1.38368118e+00
6.69337928e-01 8.64196241e-01 9.38943028e-02 -7.33783424e-01
-3.72827560e-01 -3.17606747e-01 -2.48646855e-01 3.81642461e-01
7.33696520e-01 1.80705011e-01 -6.56058431e-01 1.91332626e+00
-5.19033708e-02 -9.31460485e-02 9.39370170e-02 6.30984724e-01
6.55378401e-01 5.16746581e-01 2.66917527e-01 1.46384373e-01
1.32877398e+00 -9.23782289e-01 -5.93255281e-01 -8.90097171e-02
7.38390505e-01 -5.83617330e-01 1.71733689e+00 6.39138877e-01
-1.51797402e+00 -5.78563929e-01 -1.06990635e+00 -6.23799443e-01
-3.23464334e-01 -9.75882187e-02 4.94607806e-01 5.63831151e-01
-1.06449497e+00 4.46625143e-01 -4.98856902e-01 -1.02015086e-01
4.54798818e-01 7.46308044e-02 4.92614731e-02 -4.14895862e-01
-1.32126522e+00 1.47190738e+00 3.40209544e-01 -4.96846884e-02
-5.57049453e-01 -9.65495825e-01 -8.50083888e-01 6.23955429e-01
4.29255724e-01 -9.01611984e-01 1.67076218e+00 -7.81983256e-01
-1.44712663e+00 8.34541082e-01 -2.38046616e-01 -6.53082550e-01
6.54246390e-01 -4.44159091e-01 7.97873456e-03 7.13163242e-02
3.09288036e-02 5.81184268e-01 4.59546417e-01 -9.49894726e-01
-4.99619901e-01 -6.66630939e-02 6.28832221e-01 2.24466428e-01
7.40962997e-02 -2.54771680e-01 2.36820817e-01 -3.57445896e-01
1.06050827e-01 -4.70358551e-01 5.57772160e-01 2.33526215e-01
-1.59126133e-01 -8.22097719e-01 3.80020112e-01 -7.94567883e-01
1.23635781e+00 -1.56401312e+00 1.75388560e-01 6.93307593e-02
1.69638559e-01 3.78772557e-01 -2.71780282e-01 4.85392749e-01
1.11288279e-02 3.11399519e-01 -2.62516677e-01 3.84136178e-02
3.30886185e-01 7.72451013e-02 -5.63488781e-01 -1.90435737e-01
3.84803325e-01 1.05238128e+00 -8.06193173e-01 -3.14782172e-01
-4.56523895e-02 -3.08436831e-03 -7.24728584e-01 4.50857520e-01
-8.00493360e-01 -6.38238993e-03 -2.29569852e-01 4.59999263e-01
5.34065008e-01 -4.81775850e-01 3.96202579e-02 1.29531696e-01
2.66059041e-01 1.00898314e+00 -8.38961840e-01 1.28728449e+00
-6.15855813e-01 6.32926643e-01 -2.92375445e-01 -9.44266260e-01
7.81655550e-01 1.39502779e-01 -2.77257711e-01 -9.49209094e-01
-3.21504921e-02 1.01829149e-01 4.26768273e-01 -7.83457518e-01
1.61284477e-01 -2.62161374e-01 1.96020946e-01 6.86758161e-01
-1.41830221e-01 -3.98059875e-01 2.74040550e-01 2.11127073e-01
1.13663852e+00 3.81247610e-01 2.93541282e-01 -2.06850886e-01
9.01934564e-01 3.58895920e-02 9.96165052e-02 1.10759914e+00
-1.05126791e-01 2.86769092e-01 8.44924033e-01 -3.53836089e-01
-7.55231500e-01 -1.27688539e+00 8.10982734e-02 1.20341945e+00
-1.33616224e-01 -3.26293916e-01 -5.54010451e-01 -8.02827239e-01
6.48426339e-02 1.67902434e+00 -5.20720363e-01 -3.20074111e-01
-5.95974743e-01 -2.99954414e-01 7.11070776e-01 8.39838982e-01
8.81003737e-01 -1.09586704e+00 -9.91169155e-01 -5.24662808e-02
-4.43994075e-01 -8.24370801e-01 -1.03576638e-01 1.27817720e-01
-7.90714800e-01 -1.52521122e+00 -7.37462416e-02 -7.55145073e-01
5.53443611e-01 -2.03904081e-02 1.58046508e+00 6.59652591e-01
2.32308328e-01 6.69084013e-01 -5.23736954e-01 -5.94790697e-01
-4.97399777e-01 1.30081952e-01 -4.57433969e-01 -5.65270782e-01
7.87959158e-01 -1.84564441e-01 -2.36020863e-01 1.15672080e-02
-8.30920875e-01 2.41283208e-01 3.52720410e-01 9.17328298e-01
7.54126012e-02 2.21611681e-05 6.90714836e-01 -1.02872181e+00
1.33855844e+00 -4.15218472e-01 -2.52064198e-01 8.14972997e-01
-1.02173495e+00 4.02840883e-01 7.70297766e-01 -6.41601682e-02
-1.34301698e+00 -7.75692225e-01 -2.52063662e-01 6.68692663e-02
-1.85593273e-02 7.27631092e-01 3.84585932e-02 1.06393345e-01
5.36029577e-01 2.36686811e-01 1.46645352e-01 -3.39306951e-01
2.69517303e-01 2.75499374e-01 4.12815124e-01 -1.14580083e+00
5.90069294e-01 -1.64122164e-01 -9.34015214e-02 -3.76246154e-01
-1.34529591e+00 -1.03884555e-01 -2.92516261e-01 3.42378281e-02
7.11754739e-01 -5.38503110e-01 -1.02806699e+00 4.08080757e-01
-1.26031792e+00 -5.36429465e-01 -3.01488310e-01 2.51954228e-01
-7.12289393e-01 3.42879742e-01 -5.49066603e-01 -7.85325408e-01
-2.72259057e-01 -8.89651716e-01 6.59555912e-01 1.70311898e-01
-6.84736729e-01 -1.17591512e+00 -2.25650489e-01 9.06961739e-01
6.38183415e-01 -6.33014962e-02 1.89365518e+00 -9.01946723e-01
-6.38315856e-01 1.60010457e-02 -3.35864812e-01 6.99468136e-01
-5.32630235e-02 -2.20654994e-01 -7.86503315e-01 -1.61436740e-02
2.79392511e-01 -1.00045943e+00 9.38612998e-01 1.19778857e-01
1.43601692e+00 -3.85384858e-01 3.50216061e-01 3.79654199e-01
1.08276415e+00 1.38373569e-01 9.04409885e-01 4.85033721e-01
3.49804521e-01 7.08681881e-01 3.46538991e-01 -1.01527624e-01
1.00928223e+00 3.80191684e-01 1.44400597e-01 2.36016721e-01
-1.58876345e-01 -5.33614814e-01 2.56788641e-01 3.45204592e-01
7.03226924e-02 -3.32968891e-01 -1.13158894e+00 2.14845121e-01
-1.52385175e+00 -1.16284454e+00 -1.80838741e-02 1.97843969e+00
1.22394812e+00 3.34882528e-01 -1.78613007e-01 3.70309174e-01
-1.55936591e-02 -7.62215629e-02 -6.02960169e-01 -1.01482344e+00
-7.76592717e-02 6.34120107e-01 -9.90232900e-02 9.15084362e-01
-5.35942197e-01 8.63203764e-01 6.84236765e+00 5.18959701e-01
-7.98772573e-01 -2.51286864e-01 5.01940727e-01 2.66648114e-01
-5.67398965e-01 1.01883337e-01 -5.67005217e-01 6.32118061e-02
8.90059531e-01 -1.12550901e-02 4.83136177e-01 3.11676025e-01
1.77607894e-01 -4.54166621e-01 -1.61937988e+00 3.10510397e-01
2.53217340e-01 -1.11937082e+00 4.13492650e-01 -5.35048246e-01
4.47013527e-01 -6.57541633e-01 2.14716159e-02 7.07287014e-01
2.29427487e-01 -1.58350682e+00 7.80343354e-01 8.39151740e-01
3.38317633e-01 -4.08008307e-01 7.66874194e-01 6.49913967e-01
-5.47061145e-01 -1.93038121e-01 -1.35120615e-01 -8.09995830e-01
-1.69939935e-01 -2.58639641e-02 -1.03661966e+00 2.60849655e-01
4.70953375e-01 4.06640321e-01 -9.37920690e-01 8.01963389e-01
-8.50983441e-01 8.55551124e-01 -1.46585470e-02 -3.50398421e-01
1.79283798e-01 -3.76508012e-02 1.21050462e-01 9.81350482e-01
-6.02953024e-02 1.35714546e-01 -1.03380509e-01 1.26765525e+00
-1.47820175e-01 -1.14925958e-01 -1.87909871e-01 7.10377842e-02
5.96808076e-01 5.32313347e-01 -1.41040549e-01 -3.03448170e-01
-4.85810161e-01 6.55695438e-01 7.30914593e-01 3.87801886e-01
-7.83875406e-01 -3.15542042e-01 2.26653948e-01 1.80697083e-01
1.95412531e-01 -1.15861073e-01 -8.44657838e-01 -1.09372795e+00
3.55772018e-01 -1.35580707e+00 4.89969343e-01 -1.08689427e+00
-1.35577369e+00 2.47116193e-01 3.42686534e-01 -5.94679236e-01
-3.28130275e-01 -1.05591369e+00 -8.06763232e-01 1.22332430e+00
-1.91291761e+00 -1.00902879e+00 -4.62649018e-01 1.97172150e-01
5.83745718e-01 1.34360313e-03 7.84145355e-01 -2.77331740e-01
-2.39515990e-01 7.59959817e-01 -4.04844135e-01 -7.59413987e-02
5.40037334e-01 -1.60781205e+00 1.21122919e-01 6.24981344e-01
-2.49453694e-01 8.78955543e-01 7.72001982e-01 -3.64996612e-01
-1.08512723e+00 -4.67800200e-01 1.29377675e+00 -8.71258259e-01
3.97241026e-01 -2.74166521e-02 -1.32724309e+00 8.89819026e-01
4.94633585e-01 -8.93697739e-01 6.97558701e-01 2.80519277e-01
-6.48531258e-01 -7.91261718e-02 -1.25908327e+00 6.28657579e-01
7.47622073e-01 -5.16337156e-01 -1.53963327e+00 1.83986917e-01
6.06654942e-01 -3.34617794e-01 -7.05526590e-01 5.28839827e-01
5.45396805e-01 -1.24188387e+00 8.54604959e-01 -9.41499352e-01
9.07301664e-01 -1.74427390e-01 -3.11692450e-02 -1.18883455e+00
-1.71658918e-01 -1.36450931e-01 -2.30361909e-01 9.66583788e-01
6.06537521e-01 -6.98214412e-01 5.47838509e-01 1.01437283e+00
-1.14797160e-03 -1.07782733e+00 -8.30030441e-01 -5.89230359e-01
7.57447362e-01 -3.69852990e-01 5.95468402e-01 7.07165062e-01
1.03531457e-01 4.61448699e-01 2.81568039e-02 -1.19968526e-01
4.34943706e-01 6.30967915e-02 5.74216545e-01 -1.09021473e+00
-2.27955833e-01 -7.26638615e-01 2.18452141e-03 -1.27706099e+00
5.49231172e-01 -1.08479226e+00 -1.35104299e-01 -1.85694623e+00
3.46098766e-02 -1.93650141e-01 -1.95111394e-01 5.81988335e-01
-4.43195015e-01 -5.84331989e-01 3.27088296e-01 -1.87680319e-01
-6.23415828e-01 4.65494126e-01 1.42170751e+00 -2.07242996e-01
2.67933160e-01 7.70782307e-02 -9.09351826e-01 6.71134353e-01
7.99084723e-01 -9.41728354e-02 -8.53083611e-01 -8.20138097e-01
4.93104935e-01 -6.34453893e-02 6.72949553e-01 -8.87844741e-01
4.74146694e-01 -2.34998450e-01 4.17884380e-01 -4.01904106e-01
9.93086249e-02 -6.06628716e-01 -4.54595536e-01 4.96692836e-01
-1.08184516e+00 4.50428724e-01 3.58771563e-01 2.04827309e-01
-2.73444206e-01 -4.29678619e-01 6.97167933e-01 -2.06322372e-01
-6.28048122e-01 -4.86086309e-01 -2.71988630e-01 4.82603043e-01
7.86276579e-01 -2.86223918e-01 -5.52661836e-01 -5.60022235e-01
-3.78193617e-01 7.78255284e-01 2.80409932e-01 4.46406543e-01
8.39851499e-01 -9.30155337e-01 -9.44156408e-01 1.34432659e-01
-3.29966880e-02 -1.05005041e-01 -8.08317810e-02 6.81265533e-01
-8.65973353e-01 7.35316038e-01 -2.33041495e-01 -2.75654286e-01
-1.00459337e+00 3.03512186e-01 7.26140261e-01 -5.40418446e-01
-3.22185904e-01 1.01271665e+00 -1.13802135e-01 -6.46021903e-01
4.30486023e-01 -6.66882157e-01 -2.87172914e-01 -2.89338291e-01
5.30092955e-01 5.98569870e-01 1.36300445e-01 1.80510968e-01
-2.48454958e-01 3.24383646e-01 -2.46990070e-01 5.40212803e-02
9.76928711e-01 -6.35894462e-02 -2.06127435e-01 4.92590606e-01
6.37536168e-01 -2.41450295e-01 -1.03066683e+00 -2.37222463e-01
3.71698380e-01 -3.16709787e-01 -1.55010447e-01 -1.61265683e+00
-2.31870323e-01 1.06707180e+00 1.42223492e-01 1.88566506e-01
1.15191042e+00 -2.31406897e-01 5.68718731e-01 8.21164966e-01
1.98350340e-01 -9.20371294e-01 2.09749430e-01 9.84260738e-01
9.56496835e-01 -1.30468595e+00 -2.59006806e-02 -3.00502717e-01
-4.68682051e-01 1.27986228e+00 1.09121311e+00 4.18915078e-02
2.88904399e-01 5.65582141e-02 1.01139687e-01 -2.63883919e-01
-1.18980491e+00 9.68241319e-02 4.32332575e-01 3.45332801e-01
9.77923453e-01 -5.61200380e-02 -5.58776975e-01 4.58499730e-01
-5.60587764e-01 1.47058517e-01 4.87955362e-01 1.01831484e+00
-3.93679380e-01 -1.20165634e+00 -3.47964674e-01 7.34041750e-01
-2.63387650e-01 -3.53091687e-01 -6.14364266e-01 7.66241789e-01
-8.50456730e-02 1.21948624e+00 -1.00390516e-01 -5.31899594e-02
5.03865659e-01 5.76136231e-01 9.05368745e-01 -6.91172600e-01
-6.70569062e-01 -9.30716455e-01 2.13168427e-01 -2.98119664e-01
-3.52878869e-01 -4.94106323e-01 -1.22380686e+00 -4.39996213e-01
-7.64268041e-02 2.25037962e-01 6.42865598e-02 1.33225513e+00
6.71123341e-02 6.62244439e-01 1.20326225e-03 2.38361815e-03
-1.37595463e+00 -8.67234588e-01 -5.67250960e-02 5.44230759e-01
3.94283175e-01 -5.14719784e-01 -3.84974867e-01 -1.19065061e-01] | [10.007835388183594, 7.618768215179443] |
4ba502aa-e9b6-4e8e-ab03-7db26f91f978 | data-efficient-and-interpretable-tabular | 2203.02034 | null | https://arxiv.org/abs/2203.02034v2 | https://arxiv.org/pdf/2203.02034v2.pdf | Data-Efficient and Interpretable Tabular Anomaly Detection | Anomaly detection (AD) plays an important role in numerous applications. We focus on two understudied aspects of AD that are critical for integration into real-world applications. First, most AD methods cannot incorporate labeled data that are often available in practice in small quantities and can be crucial to achieve high AD accuracy. Second, most AD methods are not interpretable, a bottleneck that prevents stakeholders from understanding the reason behind the anomalies. In this paper, we propose a novel AD framework that adapts a white-box model class, Generalized Additive Models, to detect anomalies using a partial identification objective which naturally handles noisy or heterogeneous features. In addition, the proposed framework, DIAD, can incorporate a small amount of labeled data to further boost anomaly detection performances in semi-supervised settings. We demonstrate the superiority of our framework compared to previous work in both unsupervised and semi-supervised settings using diverse tabular datasets. For example, under 5 labeled anomalies DIAD improves from 86.2\% to 89.4\% AUC by learning AD from unlabeled data. We also present insightful interpretations that explain why DIAD deems certain samples as anomalies. | ['Tomas Pfister', 'Madeleine Udell', 'Sercan Arik', 'Jinsung Yoon', 'Chun-Hao Chang'] | 2022-03-03 | null | null | null | null | ['additive-models'] | ['methodology'] | [ 3.26661259e-01 8.58505666e-02 -1.69433072e-01 -6.41548634e-01
-9.42048669e-01 -7.07210004e-01 4.92431343e-01 6.40991628e-01
-5.56197390e-02 6.65681005e-01 -1.27122030e-02 -3.51293087e-01
-2.38780707e-01 -8.10347736e-01 -6.99075162e-01 -5.26302159e-01
-2.38575265e-02 2.50734687e-01 6.09525032e-02 -3.68407033e-02
2.73547292e-01 3.33990008e-01 -1.58929062e+00 1.93727985e-01
1.44127285e+00 1.28933835e+00 -6.31610751e-01 3.38794470e-01
-3.02823305e-01 8.85766506e-01 -6.00650012e-01 -5.28015137e-01
4.07495111e-01 -4.19639766e-01 -4.28057134e-01 1.89258739e-01
5.87015033e-01 -3.75771165e-01 2.61403203e-01 1.05864334e+00
1.99541211e-01 -3.36122438e-02 9.86000121e-01 -1.61776662e+00
-7.00086534e-01 5.82000136e-01 -7.27840066e-01 3.13387960e-01
1.81936249e-01 -4.64739017e-02 1.43540764e+00 -8.32077503e-01
5.06155528e-02 1.13028812e+00 5.70734441e-01 3.90927464e-01
-1.36118805e+00 -6.60055697e-01 5.99302471e-01 3.15782040e-01
-1.12036288e+00 -3.28785449e-01 7.72380590e-01 -5.75299740e-01
6.11624300e-01 5.49574137e-01 1.59396693e-01 9.79727268e-01
4.88761552e-02 9.52876985e-01 9.78539884e-01 -4.16158199e-01
5.18109322e-01 1.40847534e-01 4.67652142e-01 2.72538334e-01
8.60091507e-01 -1.24298394e-01 -4.83434707e-01 -6.14780962e-01
1.15125865e-01 4.02706027e-01 1.02784395e-01 -3.90412182e-01
-6.83794379e-01 1.03020740e+00 1.68716595e-01 -7.06416070e-02
-4.14563954e-01 -2.26956531e-01 4.53301370e-01 3.98566157e-01
9.35447931e-01 6.45045936e-01 -4.51840073e-01 -9.10539776e-02
-7.19153345e-01 2.05427349e-01 5.64834654e-01 6.89438641e-01
6.10813320e-01 2.00681418e-01 -8.04864615e-02 9.46054459e-01
5.15266240e-01 6.73916221e-01 1.78366870e-01 -8.34321141e-01
3.27370822e-01 1.17090261e+00 1.75688177e-01 -1.01876676e+00
-2.47369751e-01 -3.92266184e-01 -6.98603868e-01 4.51287329e-02
5.00524640e-01 -1.30662784e-01 -8.39908361e-01 1.54569411e+00
3.38312238e-01 1.10686891e-01 -1.62091013e-02 6.04970157e-01
5.82931697e-01 3.06045324e-01 8.78942758e-02 -1.94240898e-01
1.02472818e+00 -6.37781203e-01 -9.55027103e-01 -3.41514528e-01
7.38937497e-01 -5.46182871e-01 1.04366958e+00 6.91798091e-01
-7.81089425e-01 -2.86043305e-02 -1.04089558e+00 4.71008271e-01
-4.28556174e-01 -1.78238675e-01 4.34480160e-01 7.90152490e-01
-5.24334133e-01 3.80964190e-01 -9.39805448e-01 -4.35453594e-01
5.56464493e-01 4.71667526e-03 -1.59796581e-01 -1.80699617e-01
-1.12949884e+00 6.59289420e-01 1.94644898e-01 -1.13206543e-02
-7.08992481e-01 -6.17751420e-01 -1.11017275e+00 1.49873700e-02
8.52955937e-01 -9.68055502e-02 1.32072198e+00 -7.40080476e-01
-8.53130043e-01 5.34850419e-01 -3.66906852e-01 -7.35619426e-01
6.20331287e-01 -7.01455891e-01 -8.63867581e-01 -4.18309458e-02
3.16095114e-01 -5.31332009e-02 8.15693617e-01 -1.31782579e+00
-7.73652911e-01 -6.28529668e-01 -2.68681377e-01 -8.57587159e-02
-5.64803600e-01 -5.42135909e-02 -3.32717691e-03 -7.99608469e-01
2.15854526e-01 -7.79924929e-01 -3.51247907e-01 -5.98534010e-02
-6.39668882e-01 -7.40696564e-02 1.09573364e+00 -4.20013458e-01
1.58244479e+00 -2.15758038e+00 -5.63910723e-01 5.40019095e-01
3.89584005e-01 2.71438539e-01 2.78356135e-01 4.33253884e-01
-1.39593959e-01 2.20496699e-01 -7.30364323e-01 -2.19618663e-01
1.84208378e-01 3.54878515e-01 -5.85730433e-01 4.68056977e-01
6.14791453e-01 5.42222500e-01 -8.50853205e-01 -1.88594416e-01
1.80063784e-01 -1.32620975e-01 -6.69084191e-01 2.13320270e-01
-1.72495335e-01 4.36016798e-01 -5.19988596e-01 8.99386525e-01
7.83449173e-01 -2.72935301e-01 -1.28327429e-01 3.11155349e-01
-1.23220216e-02 2.82460041e-02 -1.47183728e+00 9.49266076e-01
-1.29725397e-01 4.83192384e-01 -2.70662427e-01 -1.39812362e+00
1.07003748e+00 9.20610353e-02 6.46511555e-01 -6.00375235e-01
-2.93935984e-01 3.61499071e-01 -7.26924837e-03 -3.60063642e-01
2.86882550e-01 2.26477265e-01 -1.72439635e-01 7.98447728e-01
-2.65000761e-01 2.04113856e-01 2.20265433e-01 3.60353231e-01
1.29700923e+00 -2.63120711e-01 5.81269741e-01 -1.61701411e-01
5.47635853e-01 2.69048624e-02 8.39406431e-01 9.74116206e-01
-2.15636611e-01 7.11900651e-01 9.14653599e-01 -5.16322136e-01
-8.02462459e-01 -1.14256084e+00 -3.72510463e-01 1.12292266e+00
-1.30297303e-01 -4.94806111e-01 -4.70629573e-01 -1.03296316e+00
2.10572034e-01 1.02231526e+00 -7.30724156e-01 -4.20867711e-01
-2.51564264e-01 -1.32479715e+00 5.74006319e-01 6.84333980e-01
3.40326756e-01 -6.74603164e-01 -3.06024462e-01 8.25659856e-02
-1.89826474e-01 -1.00071454e+00 -1.49848312e-01 3.12756538e-01
-7.85792470e-01 -1.30757511e+00 -2.07143232e-01 7.39196911e-02
8.69260192e-01 1.55516297e-01 1.14145815e+00 -9.79468599e-02
-1.60224482e-01 4.85929638e-01 -6.26329660e-01 -9.80117142e-01
-3.73082310e-01 -1.67133301e-01 2.83107817e-01 3.90222758e-01
9.13184524e-01 -3.27562958e-01 -5.37107110e-01 5.49685121e-01
-1.09306371e+00 -7.10118294e-01 5.54231584e-01 8.08470070e-01
4.14494365e-01 -7.17361122e-02 9.45988417e-01 -1.38377106e+00
6.03253722e-01 -9.69221413e-01 -5.39652705e-01 1.07705690e-01
-1.25816202e+00 1.07614413e-01 5.97013116e-01 -1.76538721e-01
-9.90417600e-01 -1.68226123e-01 4.52962518e-02 -8.19643512e-02
-4.00389165e-01 5.24912655e-01 -2.32509911e-01 2.74074495e-01
9.57834840e-01 2.98393127e-02 8.15773383e-03 -4.95743662e-01
1.87091917e-01 8.35839927e-01 2.49746233e-01 -6.80743873e-01
8.20820034e-01 5.00046611e-01 -5.65227084e-02 -7.12234437e-01
-1.08992827e+00 -5.81887245e-01 -5.05875289e-01 2.15709329e-01
4.72999930e-01 -9.49097216e-01 -1.76301137e-01 3.87242287e-01
-6.75486505e-01 3.87074170e-03 -4.03577000e-01 2.30158091e-01
-6.60494789e-02 5.90825379e-01 -2.22529918e-01 -1.19642174e+00
-2.91169047e-01 -8.61931026e-01 9.55595970e-01 -7.23501146e-02
-4.16385293e-01 -9.12155330e-01 1.15614220e-01 3.02281201e-01
3.88467550e-01 5.89954317e-01 7.57719755e-01 -1.56416285e+00
-1.23684824e-01 -5.54742873e-01 -2.85658687e-01 6.30572617e-01
4.18527484e-01 1.94487840e-01 -1.23067260e+00 -8.41511935e-02
-1.64801091e-01 -1.65477917e-01 7.59113550e-01 2.95649856e-01
1.32806766e+00 -4.60496128e-01 -4.16629352e-02 5.34231365e-02
1.02885211e+00 3.70791465e-01 3.66323322e-01 3.91549587e-01
6.23413026e-01 6.33621693e-01 8.76852036e-01 8.56007993e-01
3.61365676e-01 4.27683860e-01 5.19393504e-01 -9.75857750e-02
5.24665713e-01 -1.28219768e-01 4.35239911e-01 3.87699038e-01
1.66719645e-01 -2.62744218e-01 -1.28675473e+00 7.11581051e-01
-2.12570405e+00 -7.66480148e-01 -5.79380810e-01 2.39662123e+00
4.80069160e-01 4.15728331e-01 4.10516411e-01 5.24911523e-01
7.74324358e-01 -1.22787096e-02 -8.24866474e-01 -3.60473424e-01
-1.15490958e-01 -8.01255777e-02 3.49390119e-01 9.91722867e-02
-1.20757353e+00 3.76569420e-01 6.54751968e+00 4.89902526e-01
-6.90870583e-01 -8.94794539e-02 7.11532235e-01 9.90711302e-02
-3.35343689e-01 -1.11440942e-01 -5.61123371e-01 5.92513263e-01
1.11418903e+00 -2.79387832e-01 -1.11567006e-01 1.09371269e+00
2.70533711e-01 -1.90156952e-01 -1.29568350e+00 7.99023449e-01
1.43817931e-01 -7.09814250e-01 1.40457347e-01 2.75333464e-01
8.47124457e-01 -2.48140976e-01 2.48024911e-01 2.09372565e-01
4.79455292e-01 -1.00715971e+00 3.79654706e-01 3.20272893e-01
4.85557288e-01 -8.34162116e-01 1.06289399e+00 2.30837628e-01
-8.17339718e-01 -3.21582913e-01 -2.57313043e-01 -2.38504127e-01
-9.09322128e-02 1.05743945e+00 -9.15926933e-01 6.41678393e-01
8.43026996e-01 8.29658568e-01 -9.20935035e-01 9.89242196e-01
-1.27737224e-01 1.12028253e+00 -3.00773650e-01 2.89752603e-01
2.01286748e-01 -1.80301502e-01 5.44052064e-01 9.68220472e-01
3.52359861e-01 -1.15148880e-01 9.20863748e-02 8.40052247e-01
9.12293121e-02 3.38017166e-01 -9.20859456e-01 -2.03070626e-01
4.73206043e-01 1.03530288e+00 -4.82160896e-01 -2.13399038e-01
-7.73750484e-01 5.74872315e-01 -4.90610674e-03 3.07789117e-01
-5.80974400e-01 -2.64142483e-01 5.86406946e-01 2.58933753e-01
-1.85722318e-02 2.43578136e-01 -6.30791485e-01 -1.23117149e+00
2.13036552e-01 -9.85038042e-01 8.64194453e-01 -1.73236489e-01
-1.78044724e+00 3.19180101e-01 2.80066226e-02 -1.71863270e+00
-5.36867440e-01 -6.39324546e-01 -8.20535541e-01 4.74066824e-01
-1.23999310e+00 -8.81505668e-01 -4.59058285e-01 4.46770102e-01
4.25045639e-01 -2.99111247e-01 7.80269861e-01 3.53345096e-01
-1.00345945e+00 7.40057647e-01 2.53472626e-01 2.27938443e-01
1.10981464e+00 -1.72376871e+00 5.17039835e-01 1.27659857e+00
1.31742775e-01 5.00181377e-01 7.44338930e-01 -5.79197884e-01
-8.91376793e-01 -1.31084299e+00 4.67040896e-01 -8.10220718e-01
6.72942162e-01 -3.00852627e-01 -1.21389401e+00 7.81783104e-01
-2.21816480e-01 2.99470097e-01 1.28095734e+00 3.37784886e-01
-3.87207240e-01 -1.71155438e-01 -1.39214265e+00 5.08400440e-01
8.11734021e-01 -1.97854578e-01 -5.83889782e-01 1.22055151e-01
4.75150168e-01 -6.83519244e-02 -7.23757744e-01 4.35375750e-01
2.45144561e-01 -8.78717780e-01 6.40221715e-01 -8.40705931e-01
3.81047636e-01 -2.28949502e-01 -4.05030996e-01 -1.39007998e+00
1.52184799e-01 -3.78447622e-01 -4.85409588e-01 1.40473735e+00
5.56647658e-01 -1.05217171e+00 5.13156354e-01 1.06704092e+00
-3.74876633e-02 -4.02027637e-01 -5.61142445e-01 -8.34063590e-01
-1.10100349e-03 -9.84790504e-01 6.45481586e-01 1.15530872e+00
-1.50745809e-02 -2.51640584e-02 -4.72443283e-01 3.35028082e-01
8.50884020e-01 -1.21583521e-01 8.81996214e-01 -1.76070786e+00
1.74174368e-01 -9.69380289e-02 -6.06472254e-01 -5.36370337e-01
2.27022246e-02 -6.09859824e-01 -1.92389220e-01 -1.06697762e+00
7.74528682e-02 -4.16856974e-01 -6.88779771e-01 5.61271906e-01
-4.78363633e-01 3.15519601e-01 -2.91196376e-01 1.36360064e-01
-7.07742453e-01 4.60040152e-01 5.65027535e-01 -6.49218354e-03
-1.66236043e-01 1.43701077e-01 -1.15331686e+00 9.57623243e-01
8.04801881e-01 -5.01436830e-01 -3.97360235e-01 -5.23760840e-02
2.24665582e-01 -5.61308026e-01 2.47437254e-01 -6.76977873e-01
7.36109121e-03 -3.24550003e-01 3.92441988e-01 -5.18139064e-01
-2.37532601e-01 -1.02814806e+00 -2.06832558e-01 2.91418701e-01
-4.18527663e-01 1.84616506e-01 8.45767409e-02 9.00392830e-01
-3.80856246e-01 -2.01016307e-01 5.77624202e-01 1.19975753e-01
-6.53471053e-01 1.23678222e-01 -2.36023769e-01 4.38290209e-01
1.18594444e+00 2.08917409e-02 -4.41841006e-01 -6.72973573e-01
-6.97782815e-01 5.16892076e-01 3.04982722e-01 3.93187881e-01
4.75372374e-01 -1.43182528e+00 -7.88725734e-01 3.14677656e-01
6.96494877e-01 1.80830210e-01 1.45861387e-01 8.95419419e-01
-2.91201979e-01 2.94514328e-01 3.20060588e-02 -8.20658028e-01
-9.24404562e-01 5.15622139e-01 3.30688013e-03 -5.98123558e-02
-4.74028856e-01 2.72606581e-01 2.49152437e-01 -4.70541626e-01
2.96757877e-01 -2.49116197e-01 -1.44532621e-01 2.58358896e-01
7.39100277e-01 7.53667533e-01 2.86273152e-01 -3.78877729e-01
-4.41148847e-01 2.41121709e-01 -3.95906091e-01 9.69591290e-02
1.33133543e+00 -2.16283113e-01 1.03059383e-02 7.84427285e-01
6.96736991e-01 2.14406684e-01 -9.81780589e-01 -5.69373608e-01
4.78879184e-01 -6.41425431e-01 -1.71004519e-01 -7.68082440e-01
-9.55823243e-01 7.28867531e-01 6.20946348e-01 5.30973494e-01
1.15222299e+00 -1.24015529e-02 4.55572069e-01 5.78621924e-01
-6.80664778e-02 -1.20534432e+00 1.95431441e-01 1.49462223e-01
6.63461328e-01 -1.79519033e+00 4.43369299e-02 -4.48642433e-01
-9.77916420e-01 8.65584135e-01 9.10139859e-01 -4.60559167e-02
5.43944657e-01 3.02967951e-02 2.89116889e-01 -2.48751223e-01
-7.26107836e-01 -1.90823525e-01 5.60843766e-01 4.46250379e-01
3.84747386e-01 -8.37010443e-02 -3.08956921e-01 8.86267781e-01
2.04883993e-01 -5.07444024e-01 7.97996581e-01 9.81425166e-01
-3.60533565e-01 -9.91244793e-01 -5.94204724e-01 1.01410115e+00
-8.79941165e-01 7.96161816e-02 -5.94581664e-01 6.26067102e-01
-1.11699671e-01 1.25786698e+00 1.44218624e-01 -3.14053744e-01
4.72903252e-01 4.13355082e-01 -4.37246144e-01 -7.43407965e-01
-6.59637228e-02 7.40554482e-02 -7.03380443e-04 -5.48509836e-01
-3.58755738e-01 -9.07044768e-01 -8.97864938e-01 -1.56316459e-01
-3.44256461e-01 3.16769481e-01 2.33025134e-01 1.08657801e+00
3.86919200e-01 4.54008132e-01 8.53957474e-01 -8.24822634e-02
-6.57265544e-01 -9.90928531e-01 -7.21842289e-01 5.19920111e-01
5.07849932e-01 -8.60426843e-01 -7.53513277e-01 -1.17882453e-01] | [7.628170490264893, 2.542316198348999] |
3933ffa6-e6df-45ea-9d99-105646c77f3e | interaction-level-membership-inference-attack | 2301.10964 | null | https://arxiv.org/abs/2301.10964v2 | https://arxiv.org/pdf/2301.10964v2.pdf | Interaction-level Membership Inference Attack Against Federated Recommender Systems | The marriage of federated learning and recommender system (FedRec) has been widely used to address the growing data privacy concerns in personalized recommendation services. In FedRecs, users' attribute information and behavior data (i.e., user-item interaction data) are kept locally on their personal devices, therefore, it is considered a fairly secure approach to protect user privacy. As a result, the privacy issue of FedRecs is rarely explored. Unfortunately, several recent studies reveal that FedRecs are vulnerable to user attribute inference attacks, highlighting the privacy concerns of FedRecs. In this paper, we further investigate the privacy problem of user behavior data (i.e., user-item interactions) in FedRecs. Specifically, we perform the first systematic study on interaction-level membership inference attacks on FedRecs. An interaction-level membership inference attacker is first designed, and then the classical privacy protection mechanism, Local Differential Privacy (LDP), is adopted to defend against the membership inference attack. Unfortunately, the empirical analysis shows that LDP is not effective against such new attacks unless the recommendation performance is largely compromised. To mitigate the interaction-level membership attack threats, we design a simple yet effective defense method to significantly reduce the attacker's inference accuracy without losing recommendation performance. Extensive experiments are conducted with two widely used FedRecs (Fed-NCF and Fed-LightGCN) on three real-world recommendation datasets (MovieLens-100K, Steam-200K, and Amazon Cell Phone), and the experimental results show the effectiveness of our solutions. | ['Hongzhi Yin', 'Tieke He', 'Lizhen Cui', 'Quoc Viet Hung Nguyen', 'Chaoqun Yang', 'Wei Yuan'] | 2023-01-26 | null | null | null | null | ['inference-attack', 'membership-inference-attack'] | ['adversarial', 'computer-vision'] | [-2.32196376e-01 -4.47057724e-01 -8.29351917e-02 -5.30486763e-01
-2.73055792e-01 -1.14354765e+00 2.23663136e-01 -2.21087292e-01
-1.81491703e-01 5.34530818e-01 -4.58757207e-03 -7.50165403e-01
-2.44572088e-01 -9.89034712e-01 -5.70827067e-01 -7.38048196e-01
-8.29611495e-02 -3.87405396e-01 -4.67981771e-03 -9.04042646e-02
1.98633090e-01 4.41383660e-01 -1.34761870e+00 4.29424882e-01
8.00464869e-01 1.17988098e+00 -4.93466973e-01 2.82198757e-01
7.98905417e-02 4.26007181e-01 -4.00256425e-01 -1.07942080e+00
7.26560295e-01 -1.02459259e-01 -6.61057413e-01 -4.92977172e-01
1.14060618e-01 -7.54907668e-01 -4.77577150e-01 1.36259413e+00
4.43408668e-01 1.47330686e-01 1.96387589e-01 -1.49138737e+00
-6.00807071e-01 7.49274969e-01 -2.93169469e-01 -3.03340275e-02
4.55984414e-01 -1.07497029e-01 8.73464346e-01 -4.26992953e-01
1.24274164e-01 8.12780738e-01 5.87353647e-01 6.45958662e-01
-1.00495493e+00 -1.15925944e+00 2.82830507e-01 -5.43127283e-02
-1.59369504e+00 -2.29147568e-01 2.68604249e-01 -1.21279083e-01
1.88085213e-01 9.91583586e-01 2.62124062e-01 1.19696999e+00
7.71673620e-02 6.38019800e-01 1.02661896e+00 4.67543378e-02
4.91800666e-01 6.70849860e-01 7.42273510e-01 2.12234333e-01
8.83185744e-01 3.13157260e-01 -2.10792691e-01 -1.19004095e+00
5.52279711e-01 4.67027456e-01 -6.29549861e-01 -2.58138359e-01
-3.21874261e-01 6.74606502e-01 1.59842595e-02 -7.75083676e-02
-1.09505080e-01 -4.49574262e-01 5.78101337e-01 5.07128954e-01
2.08490789e-01 6.41281391e-03 -7.43955433e-01 1.65976360e-01
-2.10929021e-01 2.96103597e-01 1.25001550e+00 1.13758111e+00
4.73063439e-01 -2.69720197e-01 8.62276331e-02 4.66881305e-01
4.69267249e-01 3.49514544e-01 4.95325923e-01 -6.58234894e-01
2.57491887e-01 3.53373617e-01 3.76787663e-01 -1.32122874e+00
2.58243859e-01 -1.51518479e-01 -7.38440871e-01 -1.90087512e-01
4.41819757e-01 -6.58408999e-01 1.70097321e-01 1.73082399e+00
5.28261304e-01 4.44578379e-01 2.37469405e-01 8.86276126e-01
5.15483916e-01 4.13324952e-01 -1.33673716e-02 -3.91451538e-01
1.56508863e+00 -4.78194386e-01 -7.49163330e-01 5.26494443e-01
6.24407828e-01 -2.74037331e-01 9.35798049e-01 6.84862375e-01
-6.87209010e-01 -1.99205577e-01 -1.00710034e+00 3.05595309e-01
-6.18888140e-01 -1.96444422e-01 8.88177216e-01 1.60972929e+00
-4.67432052e-01 5.07766426e-01 -4.92328823e-01 -3.34799349e-01
3.25025558e-01 7.23440409e-01 -4.82265443e-01 3.89383435e-02
-1.59163666e+00 -9.35496017e-02 -1.19696669e-01 -2.40060464e-01
-3.62806827e-01 -7.33660698e-01 -4.67687845e-01 2.10933611e-01
4.75903630e-01 -3.52726191e-01 1.23200035e+00 -4.62063819e-01
-1.54513621e+00 3.86773854e-01 3.97182435e-01 -6.13806486e-01
3.56814563e-01 -2.23201945e-01 -1.04246771e+00 -2.96841085e-01
-4.92941022e-01 -6.36085451e-01 7.62431741e-01 -1.10990453e+00
-8.54575992e-01 -7.53496647e-01 4.44428712e-01 -1.34136621e-02
-1.02449763e+00 3.39871466e-01 -1.34759828e-01 -5.20428002e-01
-2.40860701e-01 -8.05662453e-01 -1.20727539e-01 -2.40748242e-01
-3.68493170e-01 7.83121064e-02 1.13006115e+00 -4.36028004e-01
1.62901700e+00 -2.53877282e+00 -5.96371233e-01 7.03202605e-01
3.52567770e-02 7.10142612e-01 3.24088871e-01 4.15941596e-01
2.35842735e-01 3.73036981e-01 5.95821999e-02 -1.13853335e-01
8.23155046e-02 2.41548762e-01 -7.59204090e-01 6.66913986e-01
-1.05322552e+00 4.62198496e-01 -5.90347588e-01 5.50682843e-02
1.31892532e-01 3.63900006e-01 -8.68164241e-01 4.97988164e-01
4.97284122e-02 1.46927118e-01 -8.69246602e-01 5.83309591e-01
1.27610636e+00 1.08245770e-02 5.87327659e-01 -1.71108127e-01
-1.39250711e-01 1.05062962e-01 -1.33243084e+00 1.18449652e+00
-2.11304873e-01 -4.52376664e-01 4.88142371e-01 -3.35663229e-01
9.29673016e-01 5.01772106e-01 2.78245240e-01 -2.48979703e-01
4.45395470e-01 6.71777204e-02 -2.28286266e-01 -4.15910244e-01
3.54537219e-01 4.03457940e-01 -2.92779565e-01 9.22286689e-01
-3.44730318e-01 8.81541967e-01 -7.26018369e-01 2.81377941e-01
1.07928991e+00 -1.85185194e-01 3.10786337e-01 -4.74992186e-01
1.07039225e+00 -7.29371130e-01 8.83287489e-01 8.45850229e-01
-3.12946141e-01 2.32638434e-01 1.81568593e-01 -4.00173128e-01
-1.54028639e-01 -7.45978773e-01 -3.16085219e-01 1.08620644e+00
4.56491321e-01 -1.02147079e+00 -9.98389006e-01 -1.42197895e+00
4.59422618e-01 9.04943526e-01 -4.21224654e-01 -4.06049848e-01
-2.62013197e-01 -7.20548809e-01 9.01681840e-01 2.13518605e-01
7.33406544e-01 -6.16580486e-01 -1.10120773e-01 -1.06425479e-01
-6.03890559e-03 -9.48043942e-01 -8.84743571e-01 -2.99720049e-01
-4.41877276e-01 -1.21122503e+00 2.42476724e-03 -1.92033395e-01
6.06521308e-01 8.01883698e-01 3.74299705e-01 2.40988880e-01
8.06669518e-02 5.40529311e-01 -3.85463923e-01 5.50983706e-03
-1.37552246e-01 -1.44686088e-01 5.78107834e-01 7.85031259e-01
8.50889027e-01 -7.83938587e-01 -7.37289727e-01 8.31243575e-01
-9.58765984e-01 -6.87349081e-01 -5.50829321e-02 4.57086921e-01
1.91940516e-01 3.19884032e-01 5.86004019e-01 -1.63400662e+00
8.88753593e-01 -7.65625358e-01 -6.02825820e-01 3.49032879e-01
-8.76634002e-01 -5.72069705e-01 1.34958684e+00 -5.82054377e-01
-1.22599971e+00 -2.39319459e-01 -3.66859436e-01 -2.31983900e-01
-4.46757257e-01 1.52507141e-01 -9.66209829e-01 -4.79486346e-01
4.71190840e-01 2.16671586e-01 -1.08234666e-01 -9.78948832e-01
3.44871610e-01 1.34635723e+00 4.92991090e-01 -7.36151576e-01
5.59229851e-01 5.10297298e-01 -3.81975174e-01 -4.07591641e-01
-5.20152986e-01 -3.17250311e-01 2.29086336e-02 1.50625974e-01
2.72964984e-01 -7.30918169e-01 -1.35620797e+00 5.84175169e-01
-6.46344185e-01 3.19772631e-01 4.20724787e-02 5.19533932e-01
-3.17296982e-01 7.68530250e-01 -8.79396915e-01 -9.47985828e-01
-8.71710896e-01 -8.97349477e-01 1.48277357e-01 3.81331146e-01
-6.72059879e-02 -7.66167462e-01 -1.11529924e-01 4.33396250e-01
4.16282594e-01 3.64529230e-02 7.28742421e-01 -1.33966255e+00
-2.54404724e-01 -6.02725208e-01 -1.00467779e-01 3.59313458e-01
4.80330467e-01 -2.51202255e-01 -9.27989721e-01 -6.91559732e-01
4.93277848e-01 1.01345584e-01 -6.19407669e-02 -2.50466198e-01
1.67501962e+00 -9.85663354e-01 -2.26825595e-01 9.73154366e-01
1.29279494e+00 3.35163713e-01 6.49848044e-01 4.32294831e-02
5.22869885e-01 4.08819646e-01 6.13270044e-01 8.49930644e-01
2.79100567e-01 5.58404684e-01 4.65396285e-01 4.99628395e-01
6.67489052e-01 -5.73706865e-01 3.59011739e-01 4.49231595e-01
1.07471928e-01 -3.69073302e-01 8.48625749e-02 -1.64222002e-01
-2.01382995e+00 -8.31229091e-01 -2.17691511e-01 2.77159739e+00
5.89672327e-01 -2.80291229e-01 2.36832276e-01 6.32184520e-02
6.84577525e-01 -1.09065987e-01 -6.03750169e-01 -6.01314545e-01
3.44684720e-02 -1.52764931e-01 7.44364858e-01 1.38505846e-01
-1.01517904e+00 6.10562563e-01 5.22834969e+00 6.03842974e-01
-7.90553331e-01 1.79039150e-01 4.66808349e-01 3.55092622e-02
-2.86000401e-01 1.32378832e-01 -8.85326982e-01 9.33912575e-01
9.19675648e-01 -3.80067468e-01 6.82611287e-01 1.09763062e+00
-1.09719604e-01 5.32123089e-01 -9.46758032e-01 9.70270395e-01
-1.47759318e-01 -9.83497500e-01 -8.05894434e-02 6.13004327e-01
3.31544638e-01 -5.27254879e-01 1.07392982e-01 4.02544379e-01
5.96744776e-01 -4.93681520e-01 8.96515325e-02 2.99927324e-01
7.19096184e-01 -1.26047873e+00 6.85201347e-01 5.40513217e-01
-8.64076674e-01 -4.51137006e-01 -6.30167544e-01 1.34209460e-02
1.96845047e-02 2.82274187e-01 4.59567979e-02 8.57899189e-01
9.31595683e-01 4.47106868e-01 -1.49132982e-01 6.98974907e-01
3.70799825e-02 9.43564832e-01 -3.70627314e-01 6.42712042e-02
-2.37304330e-01 -5.58775067e-01 3.66647810e-01 8.98036718e-01
2.16218233e-01 7.31230497e-01 -5.83749153e-02 5.65886319e-01
-2.87736505e-01 4.31818008e-01 -5.85902214e-01 2.49974936e-01
6.86992586e-01 1.46953344e+00 2.83186901e-02 9.56017748e-02
-6.06683493e-01 1.04018009e+00 1.85124636e-01 1.84583291e-01
-7.10649014e-01 -3.14526498e-01 1.26609755e+00 2.06094995e-01
1.79792881e-01 2.23473117e-01 -2.54733511e-03 -1.38059139e+00
-2.27381781e-01 -1.24174774e+00 9.28453147e-01 -3.58144045e-02
-1.61054325e+00 3.29125255e-01 -2.60636002e-01 -1.07654810e+00
7.79598504e-02 -1.36798218e-01 -7.03629851e-01 7.66152143e-01
-9.02197957e-01 -8.89108956e-01 7.03497231e-02 1.24182856e+00
-3.30888867e-01 -3.14694434e-01 1.29966867e+00 5.99672198e-01
-8.30881238e-01 1.58690512e+00 3.88954222e-01 1.80409431e-01
5.04780829e-01 -6.81711793e-01 3.49646181e-01 8.36662352e-01
3.67239006e-02 1.48038423e+00 3.27102274e-01 -6.07512474e-01
-1.96537447e+00 -1.04880142e+00 3.72370809e-01 -2.83152401e-01
2.91219801e-01 -7.24675059e-01 -1.20061302e+00 8.70746791e-01
-1.69037566e-01 3.34162980e-01 1.34946442e+00 6.25385717e-02
-8.13400745e-01 -3.38137895e-01 -1.98692989e+00 7.18998134e-01
1.00093353e+00 -6.38871372e-01 -1.21985361e-01 3.95340100e-02
9.15715396e-01 -3.51008177e-02 -9.67184305e-01 4.19017583e-01
9.50305820e-01 -1.11377501e+00 7.93194532e-01 -9.35981989e-01
-6.95434391e-01 -3.45045865e-01 -5.69549680e-01 -6.64509952e-01
-4.01207775e-01 -1.17007422e+00 -6.77768767e-01 1.83696806e+00
-8.64086077e-02 -1.08134115e+00 9.30746913e-01 1.26934040e+00
6.34583712e-01 -4.68157977e-01 -5.94320059e-01 -6.42580986e-01
-1.99106395e-01 -2.26509422e-01 1.31517041e+00 1.14214063e+00
2.70561635e-01 -1.85240895e-01 -8.06695402e-01 6.80054307e-01
7.15181410e-01 1.16003782e-01 9.93549347e-01 -1.14238596e+00
-4.60278243e-01 2.80135632e-01 -5.67356832e-02 -9.26635087e-01
1.69505477e-02 -5.53327739e-01 -6.93851650e-01 -4.43785280e-01
-1.75724149e-01 -5.95751166e-01 -7.87248433e-01 3.17794472e-01
1.90382540e-01 -1.43271908e-01 7.56121799e-02 2.29900911e-01
-3.55741978e-01 2.25769296e-01 4.74564254e-01 4.13709551e-01
-1.23169854e-01 8.67633104e-01 -1.24510074e+00 6.14035547e-01
9.75082517e-01 -4.19767112e-01 -7.56759107e-01 2.33098671e-01
-3.29922102e-02 8.01039487e-02 -1.43331308e-02 -6.27989233e-01
3.26192915e-01 -2.67440438e-01 -1.69720173e-01 -3.43354881e-01
-1.28661588e-01 -1.46391368e+00 4.05238032e-01 2.18249381e-01
-2.25220859e-01 -2.96040148e-01 -1.91582367e-01 7.82504797e-01
3.21903050e-01 -6.59316182e-02 5.19390821e-01 8.05532280e-03
-1.66128427e-01 7.69916773e-01 -2.18034893e-01 -3.83639038e-01
1.11979795e+00 -4.93734889e-03 -4.07833695e-01 -3.63633305e-01
-3.44402373e-01 2.56646752e-01 6.56741142e-01 4.81432855e-01
4.39217657e-01 -1.18954432e+00 -3.20660084e-01 7.01627374e-01
7.14229941e-02 -5.74878216e-01 4.17576849e-01 2.70238996e-01
2.25176420e-02 3.21842395e-02 4.25297990e-02 2.23534435e-01
-1.60433030e+00 1.06444085e+00 2.69661427e-01 1.42240271e-01
-7.05423653e-01 7.24981546e-01 1.84679151e-01 -6.91279829e-01
5.70609212e-01 2.37360269e-01 -1.87806576e-01 -3.37276578e-01
1.07079291e+00 6.01273060e-01 2.05523130e-02 -4.91835684e-01
-4.18549925e-01 5.24859577e-02 -5.74697077e-01 2.34295651e-01
7.57196963e-01 -5.90174556e-01 -2.17037275e-01 -2.80644894e-01
1.27824378e+00 6.05375528e-01 -8.71915817e-01 -4.49326366e-01
-2.32026026e-01 -1.08428335e+00 -1.18140064e-01 -6.20160699e-01
-1.29460704e+00 3.14028800e-01 5.17677665e-01 5.59883654e-01
9.92051125e-01 -7.09836483e-01 1.23279905e+00 1.54321119e-01
9.78675485e-01 -6.66160047e-01 -7.80524850e-01 5.81941009e-02
2.03293443e-01 -7.12310076e-01 5.16962223e-02 -6.42552733e-01
-5.13560891e-01 6.40176296e-01 8.51290286e-01 -2.70436276e-02
1.26073766e+00 3.75049770e-01 -3.90978642e-02 2.38928273e-01
-5.88683784e-01 6.53469622e-01 -1.45723581e-01 5.01980364e-01
1.35120600e-02 1.65917933e-01 -6.78374112e-01 1.85386574e+00
-8.05374682e-02 8.16299543e-02 4.81075227e-01 8.96519065e-01
-2.47088019e-02 -1.43837607e+00 -3.28946650e-01 4.07237321e-01
-1.07935762e+00 1.89243153e-01 -3.62853289e-01 1.17928661e-01
7.46611580e-02 1.20026016e+00 -3.77863288e-01 -9.73846078e-01
3.16378206e-01 -9.78902578e-02 -3.00250918e-01 -2.13039801e-01
-1.14207208e+00 -3.49203467e-01 -5.29149808e-02 -7.58064091e-01
1.95587128e-01 -5.41884303e-01 -1.10458505e+00 -8.49116921e-01
-6.18221521e-01 7.26005197e-01 5.20893753e-01 4.93262947e-01
6.23932838e-01 -1.50905460e-01 1.41138315e+00 1.50584191e-01
-1.10456383e+00 -1.49227440e-01 -1.18857288e+00 4.75564986e-01
-5.22024371e-02 -2.63498455e-01 -6.51460230e-01 -5.91007769e-01] | [5.8957929611206055, 6.778964519500732] |
6faefc78-c74a-4ffa-94ee-e484769ecf64 | distributional-constrained-reinforcement | 2302.01727 | null | https://arxiv.org/abs/2302.01727v1 | https://arxiv.org/pdf/2302.01727v1.pdf | Distributional constrained reinforcement learning for supply chain optimization | This work studies reinforcement learning (RL) in the context of multi-period supply chains subject to constraints, e.g., on production and inventory. We introduce Distributional Constrained Policy Optimization (DCPO), a novel approach for reliable constraint satisfaction in RL. Our approach is based on Constrained Policy Optimization (CPO), which is subject to approximation errors that in practice lead it to converge to infeasible policies. We address this issue by incorporating aspects of distributional RL into DCPO. Specifically, we represent the return and cost value functions using neural networks that output discrete distributions, and we reshape costs based on the associated confidence. Using a supply chain case study, we show that DCPO improves the rate at which the RL policy converges and ensures reliable constraint satisfaction by the end of training. The proposed method also improves predictability, greatly reducing the variance of returns between runs, respectively; this result is significant in the context of policy gradient methods, which intrinsically introduce significant variance during training. | ['Calvin Tsay', 'Antonio del Rio Chanona', 'Jaime Sabal Bermúdez'] | 2023-02-03 | null | null | null | null | ['policy-gradient-methods', 'distributional-reinforcement-learning'] | ['methodology', 'methodology'] | [ 2.20860783e-02 1.51245654e-01 -5.02303839e-01 -1.72137134e-02
-8.01939070e-01 -6.20746195e-01 2.50301570e-01 2.91080058e-01
-5.47194719e-01 1.20163918e+00 6.23793062e-03 -5.26384532e-01
-4.85198021e-01 -7.27366030e-01 -9.29706156e-01 -8.73492718e-01
-2.24019364e-01 5.66106915e-01 -1.60253733e-01 -1.17779247e-01
3.49612385e-01 5.26811182e-01 -1.03314018e+00 -2.05853045e-01
9.47177768e-01 1.17454195e+00 3.15177828e-01 4.40164417e-01
1.91144403e-02 8.53002012e-01 -6.28787816e-01 -1.70329213e-01
3.80758911e-01 -1.89210370e-01 -3.05140704e-01 1.37824371e-01
-2.95077473e-01 -3.29671115e-01 2.81571060e-01 1.00241756e+00
3.29299152e-01 5.20868182e-01 7.13576317e-01 -1.20400798e+00
-4.46508825e-01 7.93781459e-01 -5.05057573e-01 -1.10030755e-01
-9.87974107e-02 1.72700971e-01 1.02138662e+00 -4.76642340e-01
2.65345067e-01 1.25156856e+00 4.00532812e-01 3.79820615e-01
-1.38672566e+00 -2.28130803e-01 5.63446224e-01 -3.81587595e-01
-9.19907928e-01 -7.74575248e-02 5.90893984e-01 -4.06673402e-01
8.78143072e-01 -2.05653489e-01 4.37514424e-01 7.09289551e-01
4.95333523e-01 8.49152148e-01 1.11775899e+00 -5.29573143e-01
7.50729024e-01 2.82618672e-01 -3.26486617e-01 3.09135318e-01
3.81157398e-01 5.34926474e-01 -2.59700995e-02 -4.82008159e-02
7.25377083e-01 -8.55131820e-02 -9.77077112e-02 -5.79616845e-01
-8.03720951e-01 1.13624692e+00 1.05725616e-01 -1.06321469e-01
-6.91711605e-01 3.23198646e-01 4.19821292e-01 6.14450157e-01
4.68818814e-01 7.51982510e-01 -5.91286004e-01 -7.74199143e-02
-7.57097185e-01 5.43281198e-01 1.04736006e+00 1.02168500e+00
3.24688792e-01 4.09527987e-01 -3.56766373e-01 6.82426095e-01
1.89141169e-01 7.55257666e-01 1.16085909e-01 -1.22855568e+00
6.67642891e-01 -4.99426313e-02 7.72350907e-01 -6.50276244e-01
-3.34989369e-01 -6.29381299e-01 -2.84256011e-01 3.70989263e-01
4.90663707e-01 -6.07835531e-01 -6.61753178e-01 1.81115973e+00
5.43588772e-02 -2.31336609e-01 2.60338902e-01 8.41759741e-01
-5.63349485e-01 7.19505847e-01 8.75310302e-02 -8.62719595e-01
6.71754301e-01 -8.91774952e-01 -8.79240930e-01 1.30512401e-01
3.80327165e-01 -4.29310948e-01 1.11977875e+00 6.90466225e-01
-1.13395417e+00 -1.18808441e-01 -9.43354666e-01 7.67332137e-01
-1.87018231e-01 -1.08952202e-01 3.97547871e-01 5.70640624e-01
-6.60581708e-01 9.75132823e-01 -8.36252630e-01 1.68496564e-01
1.23349853e-01 3.83546382e-01 4.19146001e-01 2.21472338e-01
-1.03950942e+00 1.12243748e+00 3.91648799e-01 1.45893961e-01
-9.41170454e-01 -6.88784719e-01 -6.48730576e-01 3.35806966e-01
9.90624011e-01 -6.15100972e-02 1.62350392e+00 -1.00899065e+00
-1.97169757e+00 -2.70777613e-01 2.82218426e-01 -6.02383018e-01
6.75623178e-01 -3.86732161e-01 -1.69679657e-01 -3.18699400e-04
-4.55642678e-02 2.03104869e-01 9.48418260e-01 -1.35551989e+00
-7.98048794e-01 3.28939892e-02 8.29289481e-03 1.26011446e-01
-7.41606653e-02 -1.81684241e-01 7.60264024e-02 -6.79524601e-01
-4.51833934e-01 -9.70271647e-01 -5.84128082e-01 -6.44085348e-01
-1.72457948e-01 -3.55132133e-01 2.38910630e-01 -6.22820973e-01
1.19586611e+00 -1.77344739e+00 2.43317887e-01 5.79428852e-01
-4.18204635e-01 5.51318144e-03 -5.77535927e-02 5.57333112e-01
2.27503657e-01 1.44795299e-01 -3.33024472e-01 -2.48752832e-01
4.45411921e-01 5.16860604e-01 -5.16569257e-01 4.40417320e-01
3.62316430e-01 6.71049654e-01 -1.02742779e+00 -7.66680315e-02
3.29851806e-02 -2.28681818e-01 -6.60442472e-01 2.79370695e-01
-9.44612443e-01 3.26403916e-01 -4.35242176e-01 7.07314909e-01
3.23116958e-01 1.16173580e-01 6.21385753e-01 4.34271067e-01
-4.11578387e-01 -1.13367312e-01 -1.11028004e+00 1.12567782e+00
-9.61695790e-01 2.64846273e-02 8.74105394e-02 -1.16219079e+00
8.49773347e-01 2.44774714e-01 6.32404804e-01 -5.80451012e-01
1.15816213e-01 3.35591137e-01 -1.20050870e-01 -3.36026281e-01
5.41085362e-01 -2.78756350e-01 -2.10718676e-01 5.96947610e-01
-1.61710933e-01 -1.48180917e-01 3.71529162e-01 -2.04696909e-01
5.05213082e-01 5.36959529e-01 9.85141620e-02 -5.46865463e-01
1.71529427e-01 -1.11048231e-02 7.18603253e-01 8.60490203e-01
7.95500949e-02 -1.07164443e-01 1.07476032e+00 2.83831321e-02
-1.05426562e+00 -9.64510083e-01 6.63243979e-02 1.00843048e+00
-5.69053404e-02 2.72381276e-01 -2.66237199e-01 -7.81645119e-01
7.40194201e-01 1.03548598e+00 -4.43435669e-01 2.72649638e-02
-6.85921550e-01 -5.33083200e-01 -1.97706312e-01 5.74673355e-01
5.05672880e-02 -1.04189992e+00 -6.56306207e-01 7.09082544e-01
2.83077359e-01 -8.11005592e-01 -5.66901326e-01 5.15722394e-01
-7.87573278e-01 -8.11869323e-01 -9.51727152e-01 -4.92810905e-01
7.14866281e-01 -3.46251339e-01 1.02985632e+00 -3.20253372e-01
3.11686069e-01 3.62268418e-01 -2.60620296e-01 -5.19373596e-01
-3.57429415e-01 1.49075881e-01 3.82087171e-01 -1.09287754e-01
-1.95726871e-01 -3.78461808e-01 -3.32069814e-01 1.98054925e-01
-6.81714296e-01 -5.33926249e-01 4.94686157e-01 1.12454379e+00
7.30592847e-01 1.82832122e-01 1.14708090e+00 -1.02910721e+00
1.10687661e+00 -5.86318791e-01 -1.45665979e+00 5.42641461e-01
-1.28597653e+00 3.99556100e-01 1.04293978e+00 -6.30515337e-01
-9.95007277e-01 -1.73549771e-01 2.90917397e-01 -5.90664804e-01
3.68576139e-01 6.94906890e-01 1.83812961e-01 1.77411556e-01
2.00790912e-01 3.56510058e-02 2.65821278e-01 -4.77212697e-01
3.43781143e-01 4.35560286e-01 2.15600133e-01 -1.08532357e+00
5.07160723e-01 -1.71953112e-01 1.88927680e-01 -3.36572051e-01
-6.88111901e-01 -7.04212040e-02 -1.84588268e-01 -2.47835010e-01
3.88028771e-01 -5.14513791e-01 -1.10523772e+00 -6.16999194e-02
-6.91431284e-01 -8.79122019e-01 -6.18302226e-01 7.15740919e-01
-1.03186786e+00 3.94305028e-02 -5.35440207e-01 -1.36337066e+00
-1.46128222e-01 -1.04760075e+00 3.91236752e-01 1.85639307e-01
1.55517161e-01 -1.08318007e+00 -9.23332199e-03 -4.46475089e-01
4.09566075e-01 5.46614766e-01 1.07243621e+00 -3.93996239e-01
-3.00600231e-01 1.89032733e-01 1.44950256e-01 6.03526592e-01
-1.93888582e-02 -1.00394048e-01 -3.52818042e-01 -7.78646231e-01
-2.46326253e-02 -5.03015816e-01 5.23281157e-01 5.67584038e-01
9.41557467e-01 -6.47039652e-01 -7.87107572e-02 3.07521641e-01
1.77712548e+00 6.10355437e-01 5.47686592e-02 5.10405302e-01
1.53201029e-01 7.88545191e-01 1.12399101e+00 8.73651385e-01
8.62063691e-02 4.86635625e-01 3.94884080e-01 2.26817429e-01
4.67213631e-01 -3.97590727e-01 4.22710031e-01 4.78764385e-01
4.01595160e-02 -3.32929671e-01 -5.63471258e-01 6.06970489e-01
-2.05424452e+00 -7.02027440e-01 5.42404413e-01 2.33540726e+00
8.44659150e-01 3.35205168e-01 5.02936661e-01 -1.87864900e-01
5.64002633e-01 -1.87128201e-01 -9.77005005e-01 -9.83488619e-01
1.83199108e-01 2.61797041e-01 1.07953727e+00 7.09470332e-01
-6.66997373e-01 7.16674805e-01 6.43928051e+00 5.26160836e-01
-1.11354947e+00 -1.44942537e-01 5.13794899e-01 -2.76372761e-01
-3.13376755e-01 -1.14296727e-01 -6.48569882e-01 7.14004517e-01
9.62463915e-01 -2.71632135e-01 9.16651189e-01 1.02397513e+00
5.62514424e-01 -2.38182798e-01 -1.01120913e+00 3.26806575e-01
-4.84768093e-01 -1.11373460e+00 -4.02303755e-01 1.69130087e-01
1.04733491e+00 -3.50500166e-01 1.85963556e-01 6.65506363e-01
7.58938551e-01 -8.99256229e-01 9.14370775e-01 5.48011839e-01
6.06795728e-01 -1.50112951e+00 8.42832386e-01 3.34015578e-01
-7.33254731e-01 -6.77880585e-01 -4.40419912e-01 1.18293718e-01
3.24635208e-01 5.20168126e-01 -7.97962546e-01 6.14941895e-01
1.91266924e-01 1.96712956e-01 3.26113224e-01 7.95139372e-01
-3.63123834e-01 6.19846940e-01 -3.18133920e-01 -4.50409353e-01
5.43828547e-01 -4.36222911e-01 2.78934687e-01 9.48190033e-01
1.90790057e-01 -2.67894566e-01 6.70776188e-01 9.04769957e-01
2.02001587e-01 1.05788723e-01 -3.96581352e-01 -9.53765661e-02
6.22853160e-01 7.47103274e-01 -5.54817617e-01 -1.22208539e-02
-2.95005232e-01 5.06902397e-01 3.48169267e-01 5.50948143e-01
-7.52763510e-01 -5.69748700e-01 3.58248770e-01 -3.06929141e-01
6.24494612e-01 -3.98408800e-01 -1.71280488e-01 -7.09168434e-01
7.87069276e-02 -8.50220263e-01 2.07945511e-01 2.23379284e-02
-1.24426007e+00 1.27949998e-01 1.57285184e-01 -1.00244844e+00
-7.09261358e-01 -5.74110806e-01 -2.91449010e-01 7.91608095e-01
-1.94400084e+00 -5.15260816e-01 5.96251190e-01 3.03730220e-01
5.72978616e-01 -1.50758237e-01 3.85019004e-01 -1.75016616e-02
-6.35898232e-01 5.76806784e-01 6.97915792e-01 -3.30827117e-01
3.30450743e-01 -1.55296218e+00 -1.45838797e-01 5.63137114e-01
-5.04662871e-01 4.21786904e-01 8.34349334e-01 -7.31860220e-01
-1.41673911e+00 -1.08285165e+00 3.24175060e-01 8.78755599e-02
7.84673035e-01 -2.10372224e-01 -4.65601772e-01 4.37156767e-01
-4.40945402e-02 -2.87506491e-01 2.04796344e-01 1.39419079e-01
2.02614039e-01 -2.13598385e-01 -1.28335786e+00 4.53329831e-01
4.72619385e-01 -1.28107160e-01 -3.19119900e-01 4.46308941e-01
8.54373515e-01 -2.60603666e-01 -1.13899374e+00 2.59852141e-01
5.17450869e-01 -3.53674561e-01 4.71933991e-01 -7.74095178e-01
1.87538102e-01 3.75608541e-02 -2.21286654e-01 -1.77164233e+00
-2.83652872e-01 -8.86885047e-01 -3.70270938e-01 1.05297470e+00
6.16957486e-01 -8.07755888e-01 5.32225788e-01 6.30016923e-01
-1.04457073e-01 -1.02004182e+00 -8.12671006e-01 -1.34271026e+00
4.46632266e-01 -7.60013014e-02 6.09510183e-01 7.13433683e-01
1.42963141e-01 -1.80573165e-01 -4.61364985e-01 1.32048056e-01
7.63013899e-01 3.49869996e-01 2.98537552e-01 -5.91053963e-01
-6.61031842e-01 -4.12185043e-01 5.21965206e-01 -8.49335492e-01
4.03658897e-01 -5.20878494e-01 4.42882389e-01 -1.08136189e+00
-4.32977527e-01 -8.57345283e-01 -4.83450234e-01 3.04644436e-01
9.76538062e-02 -5.36683440e-01 5.45431793e-01 1.02068506e-01
-4.19041783e-01 6.61918938e-01 1.33156884e+00 6.53472617e-02
-6.30844474e-01 4.20529366e-01 -4.78332877e-01 2.75496155e-01
1.21735728e+00 -4.15844917e-01 -5.50744176e-01 -2.31133685e-01
1.34621441e-01 6.11664414e-01 -1.68258086e-01 -5.89385390e-01
-8.58908892e-02 -7.09687531e-01 1.18464477e-01 -4.59730089e-01
-7.51931742e-02 -8.99284363e-01 -9.96880755e-02 6.73767805e-01
-6.73005939e-01 3.93862486e-01 1.49342418e-01 7.74819791e-01
3.76988277e-02 -5.38729250e-01 7.11072624e-01 -2.16254219e-01
-2.76034206e-01 6.05656020e-02 -5.49801648e-01 7.54843354e-02
9.24160957e-01 3.59632552e-01 2.98536662e-03 -4.73091036e-01
-7.75560617e-01 8.68596315e-01 1.43025652e-01 6.29944354e-02
3.78650635e-01 -1.14064181e+00 -4.54151124e-01 -2.02577770e-01
-3.83868814e-01 -7.35439658e-02 -3.50109667e-01 7.38866687e-01
-2.05515236e-01 6.91390514e-01 3.49739217e-03 -2.68365711e-01
-3.26089650e-01 8.70261014e-01 4.32387888e-01 -7.06979871e-01
-3.58126909e-01 4.48411107e-01 -4.55751896e-01 -2.82675803e-01
4.76624280e-01 -5.95385075e-01 8.37634727e-02 2.01123789e-01
1.50036529e-01 4.38759387e-01 -8.43475610e-02 2.51958221e-01
-3.50760147e-02 2.54321009e-01 -8.42066035e-02 -5.36664307e-01
1.37915146e+00 -1.97246969e-02 2.50390708e-01 3.45873863e-01
7.28515267e-01 -2.88873557e-02 -1.87518060e+00 -2.44469032e-01
5.09606719e-01 -5.00723004e-01 1.39474824e-01 -1.10668623e+00
-9.97751772e-01 3.91517133e-01 4.28701222e-01 2.96727955e-01
9.29791093e-01 -5.75302899e-01 5.99919260e-01 4.99612212e-01
4.74473059e-01 -1.59808731e+00 1.33954128e-02 5.83103895e-01
7.09882438e-01 -9.85815346e-01 2.48274896e-02 1.12912998e-01
-9.10620332e-01 1.10008717e+00 3.63455415e-01 -3.58344257e-01
4.08708245e-01 4.07835245e-01 -1.16267644e-01 3.59322011e-01
-8.44727159e-01 -2.57601649e-01 -3.14335614e-01 5.54867685e-01
2.55357306e-02 3.75737816e-01 -6.27415895e-01 5.23478031e-01
1.53906941e-01 4.35605273e-02 4.78804141e-01 1.17414391e+00
-4.42953438e-01 -1.21570575e+00 -3.24772388e-01 3.76970798e-01
-4.97464716e-01 9.05826241e-02 2.80974567e-01 9.74658608e-01
-2.66493410e-01 9.15143192e-01 1.84643790e-01 1.09552309e-01
3.62936169e-01 -8.08145478e-02 5.51888645e-01 -4.14327383e-01
-5.25085092e-01 3.99894863e-01 2.12607235e-01 -2.85363674e-01
-1.75351724e-01 -7.41000175e-01 -1.27792788e+00 -2.54496723e-01
-4.39116120e-01 5.04965723e-01 7.25856125e-01 8.53813946e-01
5.82503006e-02 5.43610096e-01 1.27756596e+00 -5.74045181e-01
-1.58582127e+00 -7.38852620e-01 -8.95093977e-01 -1.10293459e-03
4.71411556e-01 -8.99090111e-01 -3.04864198e-01 -4.72195357e-01] | [4.296387195587158, 2.4677553176879883] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.