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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
d272e659-fd91-41c8-bc1a-1a45e16c80b0
|
femda-une-methode-de-classification-robuste
|
2307.01954
| null |
https://arxiv.org/abs/2307.01954v1
|
https://arxiv.org/pdf/2307.01954v1.pdf
|
FEMDA: Une méthode de classification robuste et flexible
|
Linear and Quadratic Discriminant Analysis (LDA and QDA) are well-known classical methods but can heavily suffer from non-Gaussian distributions and/or contaminated datasets, mainly because of the underlying Gaussian assumption that is not robust. This paper studies the robustness to scale changes in the data of a new discriminant analysis technique where each data point is drawn by its own arbitrary Elliptically Symmetrical (ES) distribution and its own arbitrary scale parameter. Such a model allows for possibly very heterogeneous, independent but non-identically distributed samples. The new decision rule derived is simple, fast, and robust to scale changes in the data compared to other state-of-the-art method
|
['Frederic Pascal', 'Matthieu Jonckheere', 'Pierre Houdouin']
|
2023-07-04
| null | null | null | null |
['classification-1']
|
['methodology']
|
[-5.18764138e-01 -6.30276918e-01 3.83623205e-02 -3.92189622e-01
-7.88485646e-01 -8.46730173e-01 5.91193676e-01 5.93427867e-02
-1.79631680e-01 1.04506993e+00 -2.54901201e-01 -4.20674272e-02
-3.96452755e-01 -4.76019323e-01 -1.26972228e-01 -1.21287048e+00
-3.01353186e-01 8.69178712e-01 4.91434932e-01 1.68261126e-01
2.98958510e-01 9.35520053e-01 -1.55682564e+00 -1.89957231e-01
9.01188254e-01 8.29635024e-01 -2.52401948e-01 5.79542696e-01
3.58006597e-01 -1.17874881e-02 -5.66052496e-01 -3.54750901e-01
3.73221457e-01 -1.89642653e-01 -1.76781118e-01 8.09975415e-02
2.86043465e-01 -4.96736877e-02 2.97883190e-02 1.17856634e+00
7.71099031e-01 2.39431590e-01 1.54737544e+00 -1.59968269e+00
-8.14298451e-01 -1.52565897e-01 -6.92646801e-01 -1.36578456e-02
-4.21718732e-02 -4.19014730e-02 4.92512643e-01 -1.00625157e+00
3.44556898e-01 1.32941318e+00 8.77173245e-01 -2.47870609e-02
-1.72052455e+00 -5.71573138e-01 -4.90415365e-01 1.48434117e-01
-1.62383640e+00 -4.09229428e-01 9.12931383e-01 -7.88856447e-01
7.66960233e-02 1.37647301e-01 1.27318934e-01 9.91826355e-01
5.29855490e-01 4.37264293e-01 1.62545824e+00 -4.14021432e-01
4.65393424e-01 3.29908729e-01 9.31079611e-02 1.94041952e-01
4.44509268e-01 3.12039126e-02 7.44670108e-02 -5.65281451e-01
4.09643352e-01 -2.83112705e-01 3.45017552e-01 -1.04608274e+00
-7.88727820e-01 1.16693330e+00 -3.32742073e-02 4.54632193e-01
-3.99348706e-01 -4.18824226e-01 5.09931266e-01 2.06361383e-01
5.25040090e-01 1.00379311e-01 -4.88252729e-01 -8.31076577e-02
-9.57263947e-01 4.32963073e-01 5.05053699e-01 6.79738998e-01
6.40437722e-01 1.22432634e-01 -4.95667160e-02 1.02974892e+00
3.80175650e-01 7.90097356e-01 6.55929565e-01 -6.27429545e-01
-9.52834636e-03 3.33824724e-01 2.19493613e-01 -1.06060338e+00
-6.65818930e-01 -4.18642193e-01 -9.44199383e-01 6.93086922e-01
8.55717540e-01 -1.99550897e-01 -5.59714258e-01 1.45877528e+00
5.41186035e-01 -4.90613490e-01 -4.23145071e-02 6.28690243e-01
3.88709158e-01 2.83992171e-01 3.47460136e-02 -3.90837699e-01
1.07941818e+00 -4.16927844e-01 -7.61425555e-01 8.69218037e-02
3.45306069e-01 -1.01560068e+00 7.64068723e-01 7.47493982e-01
-5.95217884e-01 -4.93616879e-01 -8.76210809e-01 -1.19109452e-01
-5.54292321e-01 5.14255702e-01 2.55780041e-01 9.72100854e-01
-7.63154507e-01 4.64294314e-01 -5.35280347e-01 -2.84771502e-01
1.96434885e-01 4.70321357e-01 -8.31825793e-01 3.65829170e-02
-8.07993710e-01 8.98389935e-01 1.39160141e-01 -1.87855551e-03
-2.46073470e-01 -3.29866767e-01 -6.86799586e-01 -1.68029621e-01
-5.36915623e-02 -1.39598727e-01 7.08504796e-01 -6.94211185e-01
-1.64196980e+00 8.42941642e-01 -2.19048962e-01 1.67455468e-02
6.21911228e-01 4.97778989e-02 -5.92502117e-01 -1.91576242e-01
1.89019144e-01 9.29419473e-02 1.13954556e+00 -1.09655488e+00
-1.19108558e-01 -8.86744201e-01 -8.99934888e-01 -1.55265778e-01
-3.15705985e-02 2.08441228e-01 1.41141973e-02 -7.08812118e-01
2.83055246e-01 -1.03678524e+00 1.83366835e-01 8.75205025e-02
-3.43051940e-01 -4.19704497e-01 7.89944887e-01 -8.18262875e-01
1.21897006e+00 -2.67317653e+00 1.62409499e-01 3.51388425e-01
-2.09997386e-01 1.64234936e-01 1.03171825e-01 6.22095525e-01
-1.81695312e-01 -1.18674412e-01 -2.02517256e-01 -2.98747838e-01
4.07154679e-01 1.41863406e-01 9.56751779e-02 1.25536859e+00
-1.14241533e-01 1.72294807e-02 -7.07690954e-01 -3.63186657e-01
1.79969907e-01 5.16437769e-01 -1.58659052e-02 -1.08771987e-01
3.94257277e-01 2.67133594e-01 -2.79994160e-01 7.67303824e-01
1.33699417e+00 3.60295355e-01 -8.07483774e-03 -2.98173040e-01
-2.10512236e-01 -4.56180364e-01 -1.74239171e+00 9.46679294e-01
1.00870416e-01 8.62125039e-01 1.42879978e-01 -1.10343897e+00
1.42749703e+00 2.38691255e-01 3.36120278e-01 -2.16842875e-01
2.04033121e-01 7.35438108e-01 1.92056045e-01 -3.84399652e-01
2.73577005e-01 -5.28617322e-01 2.12509725e-02 3.50444391e-02
1.43285677e-01 -7.47207627e-02 5.10619618e-02 -4.05272335e-01
5.17799616e-01 -1.38625756e-01 6.95115030e-01 -8.55529428e-01
7.14854062e-01 -4.57566470e-01 4.86570030e-01 2.83300638e-01
-5.13876021e-01 5.73251069e-01 7.50973105e-01 -2.49197111e-01
-1.17656779e+00 -1.17771721e+00 -8.19208741e-01 8.13675046e-01
-5.82722835e-02 1.06979743e-01 -4.21849966e-01 -6.00588977e-01
4.99751210e-01 6.42520249e-01 -7.62566447e-01 2.02566348e-02
-8.59855413e-02 -9.57955599e-01 5.62261045e-01 3.92907381e-01
1.47636533e-01 -4.61470276e-01 1.42771989e-01 1.52440116e-01
4.54502463e-01 -7.53367841e-01 -2.53891677e-01 2.03826308e-01
-9.87759590e-01 -1.14051664e+00 -8.47797751e-01 -4.82223898e-01
5.75740814e-01 -4.24818620e-02 7.98205376e-01 -7.08340883e-01
-1.38375759e-01 3.51455212e-02 -2.16846302e-01 -3.20535302e-01
-5.42043746e-01 -1.43819451e-01 4.40781683e-01 3.59857708e-01
7.79446006e-01 -4.43687081e-01 -4.69797552e-01 8.45492840e-01
-8.20282817e-01 -8.79332483e-01 4.16450262e-01 8.71766090e-01
5.70697188e-01 4.88690943e-01 9.51057732e-01 -4.39853519e-01
1.01170588e+00 -6.67646468e-01 -7.14738071e-01 1.41436696e-01
-7.31739044e-01 1.45945907e-01 8.25347900e-01 -5.33504486e-01
-6.68412507e-01 5.93473129e-02 2.01854527e-01 -2.58738637e-01
-5.54645956e-01 3.02162170e-01 -3.81640464e-01 -2.98436075e-01
1.01377475e+00 1.53883651e-01 4.68298316e-01 -7.91145205e-01
2.90189952e-01 1.03278148e+00 4.27692503e-01 -6.16372228e-01
8.03382814e-01 3.62821579e-01 6.44146681e-01 -1.14708138e+00
-9.57113504e-03 -5.48844159e-01 -1.24050760e+00 -6.15989324e-03
5.96405804e-01 -5.45040607e-01 -4.01112378e-01 1.03927588e+00
-5.99131703e-01 -5.33437263e-03 -3.52944881e-02 7.36991346e-01
-4.98654574e-01 6.73595607e-01 -2.14695722e-01 -1.09482753e+00
-5.12838289e-02 -1.15721226e+00 8.29868078e-01 7.11641908e-02
-8.24652687e-02 -1.36089230e+00 2.66114920e-01 -5.17011061e-02
4.82525349e-01 5.36206603e-01 7.71361172e-01 -1.00532663e+00
3.82648081e-01 -7.88932502e-01 -1.01011701e-01 8.52761865e-01
4.22467947e-01 5.81183732e-01 -8.28065634e-01 -5.46150029e-01
-4.02157158e-02 -1.95684612e-01 2.74504364e-01 4.72103506e-01
6.63641095e-01 -1.35534644e-01 -2.05098346e-01 2.70809084e-01
1.32669544e+00 2.21876666e-01 5.80532551e-01 4.09076840e-01
5.12043051e-02 6.77305758e-01 5.36840796e-01 5.12029350e-01
8.13679099e-02 9.21231627e-01 9.86702442e-02 5.29733561e-02
2.38788605e-01 3.75814885e-01 3.55656177e-01 3.57763350e-01
-4.90103289e-02 -2.91255433e-02 -6.72957659e-01 4.13550138e-01
-1.61069608e+00 -1.08209646e+00 -6.44908249e-01 2.58536267e+00
6.45450890e-01 -2.23259866e-01 6.10044479e-01 4.66775894e-01
6.97837949e-01 -2.12147221e-01 -4.47913408e-01 -4.56492722e-01
-4.87326831e-01 9.08967033e-02 6.08517528e-01 2.80655324e-01
-1.32653844e+00 3.56318116e-01 7.04693937e+00 1.16102183e+00
-1.30450106e+00 -9.97899249e-02 2.24077970e-01 4.36437100e-01
1.66245699e-01 -2.09819555e-01 -6.44652963e-01 8.93994212e-01
6.99982584e-01 -3.16572994e-01 6.06443286e-02 1.02112830e+00
1.82076216e-01 -4.23111647e-01 -8.47693443e-01 9.80443180e-01
6.14519417e-02 -3.86236966e-01 -3.75545710e-01 3.29133928e-01
6.74050570e-01 -2.43577451e-01 3.51456612e-01 1.36887282e-01
1.77901879e-01 -8.65560830e-01 4.94186640e-01 5.56521893e-01
7.35597908e-01 -8.17037165e-01 1.10303938e+00 3.13257247e-01
-7.65579224e-01 2.64981780e-02 -7.06931114e-01 1.33718684e-01
-1.81263596e-01 9.11034048e-01 -5.75912714e-01 5.18718064e-01
3.56030911e-01 4.09727246e-01 -8.17208707e-01 1.34585488e+00
8.06779191e-02 4.94676173e-01 -6.49899483e-01 -7.78642148e-02
4.59711514e-02 -6.21107161e-01 6.49072349e-01 1.14191091e+00
7.82982290e-01 -2.96329886e-01 -4.36627259e-03 5.25541484e-01
5.08564174e-01 4.46948826e-01 -5.99234283e-01 2.56175846e-01
5.61440706e-01 1.18859744e+00 -7.84531653e-01 -1.71880662e-01
-5.24908543e-01 7.41537511e-01 -9.68413129e-02 3.93326938e-01
-3.07224989e-01 -3.11498374e-01 5.94201684e-01 1.98074862e-01
4.08769161e-01 -4.57341850e-01 -2.40853623e-01 -1.01369822e+00
-6.67589484e-03 -7.37151206e-01 6.00987136e-01 -5.96187711e-01
-2.10918641e+00 4.72903073e-01 2.61753649e-01 -1.55369830e+00
-3.46749932e-01 -1.07344091e+00 -3.48723382e-01 1.18091965e+00
-1.12685239e+00 -1.16912174e+00 9.08794925e-02 7.84389496e-01
7.84625039e-02 -5.84830821e-01 9.66743052e-01 3.72548848e-01
-3.50327462e-01 6.53012455e-01 1.06976891e+00 -2.05820695e-01
1.29092062e+00 -1.43274736e+00 -3.02225113e-01 5.53887188e-01
-4.81400162e-01 5.21501005e-01 1.05994844e+00 -4.48078930e-01
-1.11598217e+00 -5.95739484e-01 5.11080265e-01 -3.92574102e-01
8.31382453e-01 -2.03271538e-01 -1.06112087e+00 3.37531149e-01
-2.20662966e-01 4.05596972e-01 1.21674442e+00 -3.35260527e-03
-3.37768197e-01 -1.58883855e-01 -1.58975673e+00 1.11971848e-01
-2.52320096e-02 -2.37745345e-01 -5.79175234e-01 3.06133687e-01
-6.36202455e-01 3.55654559e-03 -1.08998990e+00 3.82547468e-01
8.48387063e-01 -1.15831912e+00 9.68311846e-01 -6.10035062e-01
-3.23711991e-01 -5.01801372e-01 -3.72522026e-01 -1.46774638e+00
-6.99399471e-01 -4.43446726e-01 1.05843492e-01 1.14081728e+00
1.60155356e-01 -8.12776029e-01 3.45641255e-01 4.72704411e-01
2.06526741e-01 -3.85963261e-01 -1.36839223e+00 -1.38126063e+00
5.95031500e-01 -1.15843713e-01 3.38386297e-01 9.58881676e-01
-1.15384579e-01 2.85875976e-01 -4.79503691e-01 1.32020235e-01
8.99178743e-01 -1.30228251e-01 1.06413662e+00 -1.90325260e+00
-2.51409020e-02 -5.42300463e-01 -1.08751225e+00 -4.31359142e-01
1.14541560e-01 -5.47788262e-01 -1.47806421e-01 -1.03568125e+00
-2.20880404e-01 -5.75004220e-01 9.14212614e-02 -4.66656014e-02
-4.29613590e-02 4.19366658e-01 -2.07425967e-01 3.07771385e-01
-1.71059985e-02 5.48521638e-01 9.58737552e-01 2.14890331e-01
-3.34106147e-01 3.34741473e-01 -2.86110729e-01 6.92750275e-01
5.77238381e-01 -4.23056066e-01 -2.68394083e-01 3.39814425e-01
2.61876360e-02 -2.05495447e-01 2.21566617e-01 -8.79958749e-01
-4.60498873e-03 -1.95691586e-02 5.57584345e-01 -5.51247954e-01
8.14448595e-02 -9.46352303e-01 4.96904016e-01 1.00150347e-01
2.21331418e-01 4.10652198e-02 2.12248042e-01 4.25135791e-01
-3.30365092e-01 -4.51713622e-01 1.23044348e+00 3.16426635e-01
-2.57212371e-01 -1.62918717e-02 -4.17510688e-01 -2.78878510e-01
1.36570454e+00 -3.47484440e-01 -2.98498333e-01 -3.14245760e-01
-9.15166914e-01 -2.83406913e-01 6.94458127e-01 4.38173339e-02
7.07957894e-02 -1.65455782e+00 -8.49401474e-01 3.91174108e-01
8.03667307e-03 -3.67817074e-01 3.00601721e-01 1.20119429e+00
-4.61834610e-01 1.87325984e-01 -4.51991171e-01 -5.56831539e-01
-1.43200803e+00 1.00814736e+00 2.35990912e-01 1.34524271e-01
-2.33111069e-01 3.18075359e-01 -6.40396774e-02 -5.27202547e-01
-2.14334130e-01 4.13503870e-03 -2.66735226e-01 5.11805356e-01
3.50099206e-01 9.87616956e-01 8.41248035e-02 -1.12557518e+00
-4.59531575e-01 9.01725709e-01 2.69221395e-01 4.00594436e-02
1.02252173e+00 -4.33513895e-02 -3.09164017e-01 6.85116827e-01
1.32879710e+00 4.55574155e-01 -9.83801126e-01 -1.12177171e-01
-1.41842782e-01 -6.79673374e-01 -1.07718825e-01 -5.16647100e-01
-5.88128924e-01 7.66665459e-01 8.85328770e-01 7.19441295e-01
8.84279370e-01 -1.57787278e-01 4.02512029e-02 6.22690190e-03
9.76514220e-02 -1.10648799e+00 -3.11092943e-01 2.11934060e-01
1.08060217e+00 -1.24174213e+00 3.94540876e-01 -4.15904671e-01
-5.77226400e-01 1.67415476e+00 9.96044427e-02 -2.76541114e-01
1.00662673e+00 2.07410827e-01 6.20149449e-02 1.01307802e-01
-4.98760715e-02 9.47421193e-02 7.47507930e-01 1.07073498e+00
4.59448397e-01 2.52541214e-01 -7.21084535e-01 7.19256580e-01
1.46564236e-02 -5.55589534e-02 2.87909389e-01 6.49172425e-01
-2.40536794e-01 -1.23990142e+00 -9.21492040e-01 3.20909679e-01
-5.08041203e-01 4.66477603e-01 -2.18829960e-01 1.16888678e+00
3.67955081e-02 5.88396668e-01 -1.29773961e-02 -1.16050132e-01
2.22616389e-01 4.06489551e-01 2.41031244e-01 -5.99968284e-02
9.55183133e-02 3.99957061e-01 -3.31201881e-01 -7.64816403e-02
-4.49190229e-01 -1.24492967e+00 -7.70994067e-01 -4.49483126e-01
-5.43686450e-01 1.80929631e-01 8.11560392e-01 6.71229005e-01
3.31274629e-01 -1.61510274e-01 7.77034342e-01 -7.65768409e-01
-9.25470889e-01 -1.36251068e+00 -1.22405219e+00 3.56524825e-01
2.65629649e-01 -1.23986101e+00 -9.56979871e-01 8.77835378e-02]
|
[7.79159688949585, 4.186161041259766]
|
a4b17dbc-0ffe-4015-9ead-661633c78187
|
evolutionary-multitasking-with-solution-space
|
2212.05679
| null |
https://arxiv.org/abs/2212.05679v2
|
https://arxiv.org/pdf/2212.05679v2.pdf
|
Evolutionary Multitasking with Solution Space Cutting for Point Cloud Registration
|
Point cloud registration (PCR) is a popular research topic in computer vision. Recently, the registration method in an evolutionary way has received continuous attention because of its robustness to the initial pose and flexibility in objective function design. However, most evolving registration methods cannot tackle the local optimum well and they have rarely investigated the success ratio, which implies the probability of not falling into local optima and is closely related to the practicality of the algorithm. Evolutionary multi-task optimization (EMTO) is a widely used paradigm, which can boost exploration capability through knowledge transfer among related tasks. Inspired by this concept, this study proposes a novel evolving registration algorithm via EMTO, where the multi-task configuration is based on the idea of solution space cutting. Concretely, one task searching in cut space assists another task with complex function landscape in escaping from local optima and enhancing successful registration ratio. To reduce unnecessary computational cost, a sparse-to-dense strategy is proposed. In addition, a novel fitness function robust to various overlap rates as well as a problem-specific metric of computational cost is introduced. Compared with 8 evolving approaches, 4 traditional approaches and 3 deep learning approaches on the object-scale and scene-scale registration datasets, experimental results demonstrate that the proposed method has superior performances in terms of precision and tackling local optima.
|
['Qiguang Miao', 'Wenping Ma', 'Yibo Liu', 'Zedong Tang', 'Hangqi Ding', 'Maoguo Gong', 'Peiran Gong', 'Wu Yue']
|
2022-12-12
| null | null | null | null |
['point-cloud-registration']
|
['computer-vision']
|
[ 1.92764059e-01 -6.22278214e-01 2.28128314e-01 -5.05557517e-03
-5.26739061e-01 8.36359262e-02 2.43486837e-01 5.93072362e-02
-6.28589272e-01 4.61685926e-01 -2.37755954e-01 3.73344183e-01
-8.29712510e-01 -7.17309117e-01 -4.18624699e-01 -1.12169480e+00
3.68319191e-02 4.63674605e-01 3.26724499e-01 -5.23180723e-01
5.24801612e-01 3.69565785e-01 -1.61378288e+00 -4.27511901e-01
1.58506548e+00 7.80429184e-01 5.51008582e-01 -2.85562098e-01
-2.05259278e-01 -2.50146717e-01 -6.74261153e-01 -3.39668751e-01
3.25672835e-01 -2.75743902e-01 -3.56178552e-01 -1.23001896e-01
-8.87240022e-02 5.27085125e-01 2.05330819e-01 1.26280165e+00
9.29058492e-01 3.00334811e-01 4.37397450e-01 -1.30094707e+00
-5.39703190e-01 9.38694999e-02 -8.90623391e-01 2.47562125e-01
4.68387417e-02 6.75489902e-02 6.91838145e-01 -8.97126794e-01
5.00314891e-01 1.00768828e+00 7.79970229e-01 3.04447740e-01
-9.26092505e-01 -7.12209761e-01 1.12717159e-01 4.76815164e-01
-1.74706852e+00 2.09282376e-02 1.02793503e+00 -3.36828381e-01
5.58613479e-01 3.42259973e-01 8.28951478e-01 6.92095101e-01
5.77599943e-01 3.98681194e-01 1.20800781e+00 -2.38219619e-01
2.04892427e-01 9.56037045e-02 -1.45460546e-01 4.68490064e-01
4.15779620e-01 1.82828799e-01 -4.88146871e-01 -7.83046056e-03
5.29974520e-01 -2.17246152e-02 -4.83240843e-01 -5.70207417e-01
-1.18343198e+00 7.05250263e-01 7.08388269e-01 5.96248090e-01
-4.55659449e-01 -1.93199322e-01 2.46926203e-01 1.09647542e-01
3.93617988e-01 5.90102494e-01 -4.09839660e-01 -1.41852990e-01
-7.41403162e-01 1.96852222e-01 3.65682364e-01 4.81272459e-01
6.98097706e-01 3.84851135e-02 -7.01525286e-02 8.89286458e-01
3.37938577e-01 2.54321605e-01 7.43203461e-01 -4.90994364e-01
3.16842079e-01 1.08995068e+00 -2.57994235e-01 -1.42336226e+00
-4.22261000e-01 -8.09952736e-01 -8.43414307e-01 2.37919256e-01
-9.15733576e-02 3.50671709e-02 -5.63047290e-01 1.53846204e+00
7.14594841e-01 2.57237673e-01 -3.42921689e-02 1.10067451e+00
6.97194278e-01 5.51796019e-01 -8.68791193e-02 -5.59531987e-01
1.41955435e+00 -8.42160046e-01 -7.48794675e-01 -1.80492088e-01
1.52304217e-01 -8.05381536e-01 8.15923810e-01 2.13625073e-01
-7.47231305e-01 -7.75478005e-01 -1.01232541e+00 4.20398742e-01
-3.40928763e-01 -1.38033912e-01 6.11289620e-01 6.50579453e-01
-8.82992327e-01 5.44439256e-01 -5.71282029e-01 -3.25026363e-01
4.83870775e-01 4.47398186e-01 -9.62693840e-02 1.52186230e-02
-1.11740863e+00 1.07972121e+00 6.51107967e-01 4.78123218e-01
-1.73575446e-01 -6.10042870e-01 -5.21867573e-01 -1.97883677e-02
4.72864956e-01 -7.55960822e-01 4.75001097e-01 -9.84750569e-01
-1.51431286e+00 4.77290630e-01 -2.17328537e-02 1.08872270e-02
5.83691716e-01 -9.86819044e-02 -3.44498664e-01 -2.13463977e-01
-1.61105264e-02 5.10755002e-01 8.66667807e-01 -1.10468030e+00
-4.42683548e-01 -4.94926780e-01 -4.33607370e-01 6.60212517e-01
-6.54709935e-01 1.81163833e-01 -4.09471393e-01 -7.18906343e-01
4.10893738e-01 -8.02024603e-01 -4.58583266e-01 1.40103959e-02
1.93949804e-01 -3.52758378e-01 8.96442831e-01 -3.61163765e-01
1.29861772e+00 -2.02850366e+00 6.30946338e-01 3.57082427e-01
-6.33451110e-03 3.15147698e-01 2.38312539e-02 1.98345751e-01
3.06874812e-02 -5.67713007e-02 -5.02417147e-01 -1.56795070e-01
-1.05042584e-01 1.37352243e-01 3.15626085e-01 5.48819184e-01
2.48165533e-01 9.47723746e-01 -8.39179218e-01 -8.23108017e-01
2.21143544e-01 5.62966049e-01 -4.73896772e-01 -8.59330222e-02
1.75463483e-02 6.68990850e-01 -7.39279568e-01 7.29294896e-01
9.25279081e-01 -7.24119470e-02 -1.93577841e-01 -2.91398048e-01
-4.26952869e-01 -6.25801861e-01 -1.48180890e+00 1.77722967e+00
-4.05154318e-01 1.87091947e-01 7.31889382e-02 -1.07661450e+00
1.43996894e+00 9.60750356e-02 7.80099809e-01 -8.37085426e-01
3.09798717e-01 4.79000032e-01 9.52349752e-02 -4.32342350e-01
5.11331141e-01 -1.00480877e-01 9.42658931e-02 1.24874912e-01
-2.68875897e-01 -1.81338474e-01 -5.38561493e-02 -4.52921748e-01
5.95286369e-01 4.72150683e-01 2.85826266e-01 -4.57246184e-01
7.67601132e-01 8.59339237e-02 9.83455002e-01 1.24417067e-01
-1.79395631e-01 4.77270067e-01 -2.70348072e-01 -5.84421992e-01
-7.07447410e-01 -6.86720788e-01 -4.84776288e-01 6.27018273e-01
8.18295062e-01 -1.78304955e-01 -5.87824702e-01 -2.71427721e-01
-1.45619214e-01 2.95839399e-01 -4.96947289e-01 -5.08068621e-01
-8.88974726e-01 -1.16680551e+00 4.82917428e-02 1.15789838e-01
8.30673873e-01 -1.15805686e+00 -9.44267750e-01 3.06508929e-01
-1.99634150e-01 -6.30950272e-01 -4.55609053e-01 -1.55290559e-01
-1.16783392e+00 -9.15100753e-01 -9.21087444e-01 -9.18658733e-01
7.60373592e-01 4.04350311e-01 8.31645489e-01 3.81155282e-01
-4.71109927e-01 1.11457177e-01 -4.20616597e-01 -4.34255064e-01
-1.88016117e-01 2.06448719e-01 -7.77473152e-02 2.81988621e-01
2.22942948e-01 -5.66106319e-01 -5.47475219e-01 6.39866590e-01
-7.98598766e-01 -1.44768685e-01 7.43706524e-01 8.31691861e-01
6.81052566e-01 3.02228540e-01 5.75794518e-01 -9.04996097e-02
7.34735966e-01 -3.60132486e-01 -7.35066175e-01 4.44863319e-01
-8.68542671e-01 8.82033110e-02 3.58394802e-01 -5.65466583e-01
-1.14507425e+00 1.05339222e-01 7.13869631e-02 -6.47780597e-01
2.52125978e-01 5.39293528e-01 -1.42683312e-01 -6.10227466e-01
4.93380189e-01 4.55173790e-01 2.09852904e-01 -2.55123943e-01
-1.21178534e-02 3.64549458e-01 2.18665510e-01 -7.57856369e-01
1.02623510e+00 2.76276410e-01 1.82828754e-01 -5.21636367e-01
-2.98699260e-01 -4.34326082e-01 -5.53946674e-01 -6.33680820e-01
9.26272750e-01 -5.33055007e-01 -9.05914962e-01 6.85648918e-01
-1.20709431e+00 4.37170923e-01 -1.55104682e-01 5.17857432e-01
-2.41817981e-01 4.72530365e-01 4.20338213e-02 -8.06203485e-01
-5.95669568e-01 -1.41685271e+00 9.56113696e-01 7.04169452e-01
1.44202709e-01 -8.71063888e-01 9.47045758e-02 2.12430358e-01
6.84138119e-01 5.49712956e-01 6.05571210e-01 -2.91300982e-01
-6.78952932e-01 5.42396344e-02 -9.41662956e-03 -2.93875691e-02
1.52186621e-02 -5.79117276e-02 -5.28123617e-01 -4.50037330e-01
4.46621865e-01 1.28321256e-02 5.00017107e-01 4.63536441e-01
9.18051422e-01 7.02647790e-02 -5.34379423e-01 8.03861916e-01
1.62057471e+00 5.50389171e-01 7.21324861e-01 9.36646163e-01
4.28654283e-01 6.30902052e-01 1.00866508e+00 3.92966986e-01
2.83639073e-01 9.42392051e-01 5.67523122e-01 -1.67086765e-01
1.56864569e-01 1.76808313e-01 1.56741813e-01 9.41416025e-01
-3.71476948e-01 8.65483284e-02 -9.08849716e-01 3.26308578e-01
-1.90992224e+00 -8.01085055e-01 -1.80581391e-01 2.17878699e+00
6.52167678e-01 1.05450816e-01 -1.78963676e-01 2.02631410e-02
1.10311818e+00 4.64890823e-02 -4.56421465e-01 -1.88080698e-01
-2.49487579e-01 2.63137609e-01 2.11710945e-01 1.39572144e-01
-9.40342844e-01 6.15806520e-01 4.73565578e+00 1.09659708e+00
-1.28019595e+00 3.08845699e-01 3.15745413e-01 6.77750409e-02
-1.95570797e-01 -4.50674221e-02 -7.23068833e-01 7.20249593e-01
2.50750661e-01 -3.65997761e-01 3.35121006e-01 7.03511894e-01
7.43450895e-02 -6.76103681e-02 -4.58692700e-01 1.19145322e+00
2.50639796e-01 -1.09418380e+00 -1.13597400e-01 1.62864715e-01
7.88846254e-01 -2.43503794e-01 1.06366068e-01 1.93676651e-01
-3.59612316e-01 -8.55453491e-01 6.93509340e-01 6.30573094e-01
5.05731106e-01 -8.54508638e-01 9.80916858e-01 4.37820554e-01
-1.57411802e+00 -2.36528099e-01 -3.01961750e-01 3.43781501e-01
4.09244269e-01 3.06407869e-01 -2.70291865e-01 1.20296991e+00
9.77197587e-01 6.15406752e-01 -5.66063225e-01 1.50307274e+00
1.69776693e-01 -1.32501006e-01 -3.67870361e-01 -2.60628909e-01
1.29805654e-01 -8.07519376e-01 9.93430078e-01 7.48324096e-01
6.15350068e-01 1.17739469e-01 2.57944196e-01 8.69482338e-01
4.53972965e-01 4.07064646e-01 -3.61421347e-01 4.19387877e-01
5.53868413e-01 1.29105937e+00 -9.63109612e-01 3.04659069e-01
-8.35116133e-02 8.58039737e-01 1.19020849e-01 6.88710883e-02
-1.05644357e+00 -3.72661948e-01 3.21188986e-01 -2.09057286e-01
2.94793427e-01 -2.01154739e-01 -3.20252299e-01 -8.99969816e-01
2.92605221e-01 -7.73594558e-01 2.73724824e-01 -4.07876134e-01
-1.09844828e+00 9.04887736e-01 4.87288572e-02 -1.54036105e+00
4.42570508e-01 -2.26327732e-01 -7.97288775e-01 8.72543514e-01
-1.61247420e+00 -1.03972661e+00 -7.02690184e-01 5.81621945e-01
7.27624536e-01 -2.71745622e-01 4.05219853e-01 5.41417837e-01
-7.62377620e-01 4.62890536e-01 1.88812092e-02 -5.23809850e-01
5.62237382e-01 -7.85778224e-01 -3.39236297e-02 8.36035132e-01
-8.38683546e-02 6.08850718e-01 5.40826499e-01 -7.07344651e-01
-1.45154858e+00 -6.35862529e-01 6.64165676e-01 -5.95601276e-02
3.05392921e-01 3.36159244e-02 -1.12422860e+00 -5.79161718e-02
6.47830516e-02 -7.64084831e-02 2.49257579e-01 -5.51424585e-02
3.18259060e-01 -3.60851973e-01 -1.16080129e+00 5.94643414e-01
1.22330463e+00 2.72016853e-01 -6.52722299e-01 1.12957262e-01
5.72258770e-01 -6.19288862e-01 -1.00251293e+00 7.81955481e-01
3.43110621e-01 -8.15713823e-01 1.03521693e+00 -8.37915689e-02
-7.79553503e-03 -5.70816696e-01 1.90457270e-01 -1.22187483e+00
-3.89891207e-01 -6.45537138e-01 1.09660387e-01 1.20248401e+00
3.59921753e-02 -8.65410268e-01 5.33639073e-01 2.32736185e-01
-4.24128622e-01 -1.17286766e+00 -1.40568316e+00 -9.16102111e-01
-1.37124926e-01 1.11808419e-01 6.94120407e-01 1.00335348e+00
-6.05125189e-01 7.13205785e-02 -8.42761993e-02 1.75703123e-01
5.83723545e-01 2.98166484e-01 5.99209845e-01 -1.60982072e+00
-5.91873787e-02 -8.01168978e-01 -4.35055822e-01 -6.44928217e-01
-1.68416828e-01 -8.89275670e-01 4.33393382e-02 -1.49702680e+00
1.57951757e-01 -9.44283187e-01 -2.97207952e-01 9.32749733e-03
-4.74769711e-01 5.79233840e-03 3.61652315e-01 5.10577381e-01
-5.75713754e-01 1.06650114e+00 1.48339617e+00 1.32697588e-02
-3.19930464e-01 3.32067162e-02 -5.20839393e-01 4.75376219e-01
6.11849010e-01 -6.80825412e-01 -3.03779751e-01 -4.14570034e-01
2.37704754e-01 -1.79916635e-01 1.36299014e-01 -1.12122571e+00
5.17482042e-01 -1.44586429e-01 9.53347757e-02 -5.15125990e-01
2.94395983e-01 -1.09834516e+00 6.50500298e-01 8.35975707e-01
3.76567185e-01 4.99531358e-01 9.66615528e-02 6.67338014e-01
-3.30064058e-01 -3.60164315e-01 8.35441828e-01 -4.98612225e-02
-9.21417415e-01 3.70427936e-01 1.39051661e-01 -1.37553094e-02
1.52410495e+00 -8.97554338e-01 8.24652892e-03 3.72581154e-01
-2.81776518e-01 4.80998069e-01 4.63973761e-01 7.20989585e-01
7.31190383e-01 -1.32479894e+00 -8.17784667e-01 1.68442637e-01
-5.09399585e-02 1.88764811e-01 4.23619896e-01 1.02833533e+00
-5.24163663e-01 -6.85578287e-02 -3.30837905e-01 -9.67324793e-01
-1.29009402e+00 3.75991762e-01 5.19253731e-01 -2.28568882e-01
-4.56829756e-01 8.07459056e-01 -1.88902184e-01 -3.56204450e-01
1.40152320e-01 1.48911744e-01 -4.69153225e-01 2.69868881e-01
3.20346770e-03 5.85864067e-01 1.47679403e-01 -6.81744754e-01
-6.51108921e-01 1.28712285e+00 2.06383899e-01 2.33589411e-01
1.34160197e+00 -3.45218228e-03 -3.13265175e-01 5.79008972e-03
1.05013537e+00 -1.86845049e-01 -1.04454792e+00 -2.15795889e-01
1.18448883e-01 -7.38343537e-01 8.42454582e-02 -4.95218754e-01
-1.26812756e+00 5.25344372e-01 9.41485524e-01 -1.51903272e-01
1.42141461e+00 -2.31483176e-01 8.85050416e-01 2.85953544e-02
6.94558740e-01 -1.19438243e+00 4.77587402e-01 2.74611980e-01
1.06567955e+00 -1.27863288e+00 2.12923288e-01 -3.86180311e-01
-5.94585538e-01 1.15037572e+00 9.35452580e-01 4.21498567e-02
5.06853938e-01 1.42438844e-01 -5.53605109e-02 -4.27183121e-01
-2.29724824e-01 -3.98882180e-02 4.25099790e-01 4.00315225e-01
-1.39161842e-02 -3.71532053e-01 -8.95662427e-01 3.69853795e-01
-9.54550281e-02 -1.24592267e-01 -1.58231020e-01 9.82881010e-01
-4.31870788e-01 -1.13201106e+00 -6.59794629e-01 5.41084707e-02
-1.21783383e-01 1.64105460e-01 1.33142620e-01 7.63738036e-01
5.53632736e-01 6.35611594e-01 1.86718535e-02 -3.02125663e-01
4.85366136e-01 -3.45760554e-01 3.72776210e-01 -2.62081325e-01
-9.66586769e-01 3.24394293e-02 -4.58221793e-01 -3.71025652e-01
-7.50948370e-01 -7.86973238e-01 -1.10344183e+00 -3.85449361e-03
-8.54534447e-01 3.79776925e-01 6.65776968e-01 8.82235229e-01
4.43642467e-01 6.28559351e-01 8.47891867e-01 -8.86255145e-01
-4.98077512e-01 -6.05042577e-01 -8.36862028e-02 1.57080993e-01
-1.83947697e-01 -1.10782266e+00 -1.38107523e-01 -4.29600716e-01]
|
[5.7641377449035645, 3.4495413303375244]
|
0ba621ce-17e1-4633-9a06-4ed0411494ad
|
decomposed-inductive-procedure-learning
|
2110.13233
| null |
https://arxiv.org/abs/2110.13233v1
|
https://arxiv.org/pdf/2110.13233v1.pdf
|
Decomposed Inductive Procedure Learning
|
Recent advances in machine learning have made it possible to train artificially intelligent agents that perform with super-human accuracy on a great diversity of complex tasks. However, the process of training these capabilities often necessitates millions of annotated examples -- far more than humans typically need in order to achieve a passing level of mastery on similar tasks. Thus, while contemporary methods in machine learning can produce agents that exhibit super-human performance, their rate of learning per opportunity in many domains is decidedly lower than human-learning. In this work we formalize a theory of Decomposed Inductive Procedure Learning (DIPL) that outlines how different forms of inductive symbolic learning can be used in combination to build agents that learn educationally relevant tasks such as mathematical, and scientific procedures, at a rate similar to human learners. We motivate the construction of this theory along Marr's concepts of the computational, algorithmic, and implementation levels of cognitive modeling, and outline at the computational-level six learning capacities that must be achieved to accurately model human learning. We demonstrate that agents built along the DIPL theory are amenable to satisfying these capacities, and demonstrate, both empirically and theoretically, that DIPL enables the creation of agents that exhibit human-like learning performance.
|
['Kenneth Koedinger', 'Erik Harpstead', 'Christopher MacLellan', 'Daniel Weitekamp']
|
2021-10-25
| null | null | null | null |
['procedure-learning']
|
['computer-vision']
|
[ 2.50159442e-01 6.26443446e-01 2.17186585e-01 -3.09133738e-01
-3.36437792e-01 -5.87255001e-01 9.89940464e-01 3.25107574e-01
-5.01697958e-01 7.36216426e-01 -3.65962595e-01 -6.84129298e-01
-3.08368772e-01 -1.03317428e+00 -7.69490361e-01 -2.35096931e-01
-1.69925094e-01 7.63871968e-01 3.15858930e-01 -3.66211116e-01
3.35804433e-01 5.39622903e-01 -2.04270601e+00 2.02714294e-01
1.33810008e+00 3.09068233e-01 1.94820091e-01 6.93745136e-01
-1.56519905e-01 1.57683372e+00 -5.21730483e-01 -4.65380996e-01
1.18903641e-03 -4.85439807e-01 -1.05936420e+00 1.53115481e-01
1.16318472e-01 -1.35828480e-01 1.18576020e-01 7.84057796e-01
1.62174135e-01 3.07753593e-01 1.01445055e+00 -1.22800708e+00
-7.26770103e-01 7.30162680e-01 -5.76831736e-02 -1.97580427e-01
7.54805088e-01 1.75080508e-01 6.10957563e-01 -7.12150156e-01
4.33691204e-01 1.12860501e+00 6.88609004e-01 8.44664156e-01
-1.17869508e+00 -4.91009116e-01 -2.01024726e-01 -1.04539365e-01
-1.05840337e+00 -3.04941148e-01 4.42975938e-01 -8.31522226e-01
8.08779478e-01 4.00674865e-02 1.17237568e+00 5.06939530e-01
-2.38910820e-02 8.48923266e-01 1.21727717e+00 -8.44320893e-01
2.84762949e-01 4.50728059e-01 3.81773189e-02 1.10707819e+00
3.30890298e-01 5.57493746e-01 -7.04472899e-01 2.28781909e-01
1.03762805e+00 -3.76323640e-01 1.82813451e-01 -5.46549261e-01
-1.11145782e+00 8.16700518e-01 3.02874178e-01 5.32209277e-01
-2.61562526e-01 2.12874830e-01 1.86313257e-01 5.68874538e-01
-9.32244062e-02 1.17366230e+00 -4.31056857e-01 3.69799472e-02
-6.58603311e-01 4.03634161e-01 9.74050820e-01 1.06838191e+00
6.63648665e-01 4.32747573e-01 4.80977118e-01 5.06393194e-01
1.15404241e-01 2.43601456e-01 6.10203981e-01 -1.32188118e+00
-2.36456513e-01 8.41052830e-01 2.16199592e-01 -7.55585194e-01
-4.29103851e-01 -6.30027533e-01 -4.11076844e-01 8.45706940e-01
5.03847539e-01 -2.14847662e-02 -6.40974104e-01 1.77879417e+00
2.60631531e-01 9.73622948e-02 3.37435365e-01 4.52349097e-01
6.79464519e-01 7.12906778e-01 4.65645075e-01 -3.62232208e-01
1.08264410e+00 -8.87769163e-01 -1.93684369e-01 -2.40993381e-01
1.10010302e+00 -3.63352329e-01 1.39204729e+00 5.75117171e-01
-1.77062011e+00 -5.87895751e-01 -1.05690873e+00 1.11726977e-01
-5.06350517e-01 -4.42385137e-01 1.28671610e+00 9.33807492e-01
-1.09585452e+00 5.45835972e-01 -5.63384771e-01 -7.46772811e-02
2.37811387e-01 4.55296934e-01 -1.90842241e-01 1.80091664e-01
-9.21155274e-01 1.46785450e+00 7.74220347e-01 -4.04289663e-01
-1.19810116e+00 -9.10562992e-01 -9.34780478e-01 1.86385080e-01
2.88628072e-01 -8.03756893e-01 1.61936724e+00 -1.40517640e+00
-1.42181611e+00 1.15191555e+00 1.75149977e-01 -4.01005179e-01
3.47611278e-01 1.21618085e-01 -1.92524493e-01 1.09567232e-01
-6.21182062e-02 6.68694377e-01 3.23062718e-01 -1.44116199e+00
-8.49436224e-01 -7.56767616e-02 6.25736654e-01 3.77349466e-01
-2.51764178e-01 1.16226740e-01 1.67711139e-01 -4.82741028e-01
2.45938823e-03 -7.72233486e-01 -2.43546933e-01 -2.23800510e-01
4.37091470e-01 -6.13122642e-01 1.06405117e-01 -1.39408931e-01
8.28552783e-01 -1.88810241e+00 2.90899187e-01 3.30895483e-01
4.54113275e-01 3.86216760e-01 1.82022508e-02 2.32375160e-01
4.25555371e-02 4.86570559e-02 -1.73087329e-01 1.84148848e-01
3.37761670e-01 1.43155754e-01 -9.92439967e-03 -5.08781746e-02
-3.85625549e-02 8.88530910e-01 -1.20112443e+00 -5.74001133e-01
1.65113583e-01 1.35396123e-01 -8.48351777e-01 4.35451448e-01
-4.95250076e-01 4.62069929e-01 -5.15441716e-01 2.55607069e-01
-2.62771100e-01 -2.46000037e-01 2.32428744e-01 7.53385603e-01
-4.41530198e-02 2.42127240e-01 -9.87883031e-01 1.32691133e+00
-5.96277475e-01 6.43225312e-01 -1.18365347e-01 -1.07454264e+00
6.81218743e-01 4.69346911e-01 1.45594388e-01 -6.33770585e-01
-1.24162752e-02 3.50359231e-01 4.37617242e-01 -5.06742001e-01
2.43954524e-01 -5.28625011e-01 -4.03732173e-02 7.10847080e-01
1.05172656e-01 -8.59418929e-01 3.31406265e-01 9.19110700e-03
8.09711158e-01 3.46515179e-01 3.26225907e-01 -4.37286258e-01
8.18383276e-01 3.43865156e-01 3.94016713e-01 8.48453939e-01
-1.08305346e-02 -4.01658475e-01 2.26856947e-01 -7.42643297e-01
-1.19092166e+00 -9.85982060e-01 4.07684222e-03 1.80496132e+00
-1.44106060e-01 3.20259817e-02 -1.17536712e+00 -6.69391006e-02
-3.02301168e-01 1.16926599e+00 -4.77247208e-01 -1.22797117e-01
-6.15823269e-01 -4.68550891e-01 6.76673532e-01 3.74124050e-01
4.77632582e-01 -1.60458207e+00 -9.50928092e-01 3.00691366e-01
4.47837204e-01 -8.15091372e-01 4.68604982e-01 6.49740621e-02
-6.80956125e-01 -1.08782232e+00 -2.70371646e-01 -1.43730509e+00
8.61479223e-01 -5.41797057e-02 1.38366771e+00 8.03127110e-01
-2.13785872e-01 7.44790971e-01 -2.37165362e-01 -8.90744448e-01
-9.77570891e-01 -1.49843246e-01 1.70688003e-01 -7.64866829e-01
3.78616691e-01 -7.74715543e-01 -7.50820264e-02 -6.79382449e-03
-9.30594802e-01 5.94305217e-01 4.91180301e-01 7.43310332e-01
-9.02886689e-02 4.33640838e-01 8.20176303e-01 -9.42048907e-01
6.86105430e-01 -2.24472493e-01 -6.37118757e-01 4.76073414e-01
-6.85612559e-01 1.36518300e-01 6.52931213e-01 -4.00757492e-01
-1.22128856e+00 -8.91068727e-02 1.02598719e-01 4.14100707e-01
-3.13393533e-01 6.92522883e-01 7.47275576e-02 -4.31464761e-01
1.06409013e+00 3.52612853e-01 -2.45583914e-02 9.60160568e-02
3.71146739e-01 3.81167084e-01 5.02380133e-01 -1.30222571e+00
1.07985830e+00 -2.31412143e-01 2.33640298e-02 -8.04325461e-01
-9.99973774e-01 1.75673276e-01 -5.83482444e-01 -4.88622069e-01
6.48290813e-01 -7.01892734e-01 -1.17883456e+00 4.53994721e-01
-6.85750902e-01 -9.66488183e-01 -3.91867578e-01 3.95358860e-01
-1.02195692e+00 -2.20985487e-01 -5.72491229e-01 -9.38013554e-01
9.71228108e-02 -1.17469478e+00 2.36761317e-01 3.60235453e-01
-3.89221758e-01 -1.23358440e+00 -1.88271962e-02 4.33637440e-01
2.51902878e-01 2.30752751e-01 1.56035054e+00 -8.17349851e-01
-4.97974247e-01 8.11531395e-02 2.85316855e-01 5.72004244e-02
-3.64427924e-01 -1.85513645e-01 -7.29472101e-01 -4.31645103e-02
4.56358716e-02 -8.57809603e-01 5.96455112e-02 -7.21858144e-02
9.23612356e-01 -1.30022317e-01 -4.61147428e-02 1.48904413e-01
1.21459126e+00 6.19020581e-01 5.47678947e-01 5.22385597e-01
3.48771840e-01 9.43193853e-01 3.74157727e-01 -1.59026697e-01
4.59644735e-01 3.78946185e-01 -1.65650636e-01 1.01493351e-01
4.53140363e-02 -2.54443318e-01 1.87323868e-01 1.00499260e+00
-5.42003214e-01 2.38971069e-01 -1.38623559e+00 5.67693651e-01
-1.68413937e+00 -1.27054036e+00 1.43516526e-01 2.03313112e+00
1.39247179e+00 5.00964701e-01 2.80560970e-01 3.24693173e-01
4.28359807e-01 -4.32061881e-01 -2.29741976e-01 -8.34826469e-01
2.54774332e-01 5.25663137e-01 -1.68700293e-01 7.53004611e-01
-6.32785499e-01 1.19922113e+00 7.53114033e+00 5.02959430e-01
-5.10523260e-01 -1.51844889e-01 4.85426366e-01 3.19974124e-01
-3.72335792e-01 -1.90152451e-01 -3.67847383e-01 8.22589472e-02
1.33523035e+00 -3.49496782e-01 8.00990105e-01 9.01374698e-01
-4.53395061e-02 -5.16059995e-02 -1.47619390e+00 5.70233285e-01
-1.25654684e-02 -1.45365512e+00 1.94088370e-01 -1.25212789e-01
1.03945589e+00 -7.96256781e-01 -5.40603064e-02 8.74975145e-01
9.83110547e-01 -1.52282858e+00 7.57060945e-01 4.13268477e-01
3.46401662e-01 -1.00078642e+00 2.96637028e-01 9.55592334e-01
-7.63098419e-01 -2.98926920e-01 -1.02238856e-01 -5.91603279e-01
-3.96453232e-01 -7.47129247e-02 -9.64040279e-01 -5.92607036e-02
1.88199267e-01 1.65842742e-01 -5.90386927e-01 8.87480617e-01
-4.72502202e-01 3.94380987e-01 3.18968035e-02 -5.29017985e-01
2.47461572e-01 -1.57189637e-01 1.02718316e-01 1.04038012e+00
1.41072005e-01 8.06519091e-01 2.16851935e-01 8.02354753e-01
1.61120310e-01 -1.41580077e-02 -7.16344595e-01 -1.38383016e-01
7.06605136e-01 7.35400915e-01 -5.84898055e-01 -5.72595298e-01
-3.05827647e-01 4.41229761e-01 5.47082126e-01 1.66219667e-01
-6.81809366e-01 -1.68421075e-01 2.82797843e-01 1.78039983e-01
-2.55130440e-01 -5.13007760e-01 -4.83193874e-01 -7.52051055e-01
-4.49545056e-01 -1.33290303e+00 -5.98527156e-02 -9.82370853e-01
-8.34108055e-01 2.07108051e-01 2.42111608e-02 -6.58002138e-01
-6.51333451e-01 -7.40945280e-01 -5.03550589e-01 5.89543581e-01
-9.61562335e-01 -1.23262310e+00 -2.76629478e-01 6.23537004e-01
4.10280079e-01 -6.85020924e-01 1.17632401e+00 -1.56134725e-01
-1.02752358e-01 4.40617591e-01 -3.48202378e-01 2.26849839e-01
1.42770140e-02 -1.46198845e+00 1.91691980e-01 5.46671093e-01
9.60284024e-02 7.34465063e-01 7.53465295e-01 -3.98970813e-01
-1.27002478e+00 -6.38187945e-01 8.62569690e-01 -6.03950560e-01
5.27276933e-01 -4.20333371e-02 -7.55836129e-01 9.07215655e-01
1.04306944e-01 -7.06766725e-01 8.01004112e-01 1.91448212e-01
-3.42104882e-01 1.52538642e-01 -1.15611792e+00 8.91575158e-01
1.09902942e+00 -6.86439931e-01 -1.36298704e+00 5.53954065e-01
6.44357502e-01 -2.62047619e-01 -1.01168072e+00 2.14534074e-01
5.29505253e-01 -1.03488266e+00 1.17618930e+00 -9.43630278e-01
5.68575978e-01 -3.24029535e-01 2.45510638e-01 -1.21942878e+00
-4.23115075e-01 -6.85293138e-01 -1.21957511e-02 8.49628568e-01
4.42733556e-01 -5.42756498e-01 6.96545720e-01 9.51731563e-01
-3.51052225e-01 -5.71790516e-01 -2.69844621e-01 -7.55054891e-01
5.95771790e-01 -4.22725081e-01 5.40894806e-01 1.29653275e+00
4.37609702e-01 3.91883820e-01 2.00896367e-01 -2.18261071e-02
7.52482653e-01 -1.27696395e-01 7.68553197e-01 -1.68788624e+00
-3.25940698e-01 -7.80630171e-01 -4.30184513e-01 -4.71718997e-01
5.92075229e-01 -1.01020646e+00 1.20635726e-01 -1.40581119e+00
8.89647678e-02 -8.33642423e-01 1.01175644e-02 3.90415341e-01
9.80963334e-02 -2.27320828e-02 1.47521570e-01 6.16434328e-02
-4.61040676e-01 -7.64201060e-02 1.25365937e+00 2.30725259e-01
-2.22413048e-01 -2.84543216e-01 -7.86286891e-01 1.42965722e+00
8.07800651e-01 -2.10414976e-01 -7.17760026e-01 -3.36252481e-01
7.17397094e-01 1.12990905e-02 4.35824037e-01 -1.39789689e+00
5.48608005e-01 -6.96255565e-01 6.08360767e-01 1.99341044e-01
9.98388007e-02 -7.62231171e-01 8.89559761e-02 7.40418434e-01
-7.59731352e-01 -1.20780177e-01 1.21694997e-01 4.69757207e-02
8.85559767e-02 -7.15942860e-01 9.15123880e-01 -4.91599828e-01
-9.21203732e-01 -2.86325932e-01 -7.45008886e-01 2.31958479e-01
1.34200621e+00 -3.02677184e-01 -2.14151040e-01 -2.74599463e-01
-8.48580241e-01 1.66315317e-01 6.73493564e-01 -1.58063263e-01
4.56161261e-01 -1.06469727e+00 -6.44521534e-01 1.32053286e-01
-2.01411873e-01 -4.40291204e-02 -2.68138528e-01 3.25046808e-01
-9.83729362e-01 3.64444435e-01 -6.47973478e-01 -1.47948533e-01
-9.05203521e-01 6.59376264e-01 5.65527380e-01 -9.76320803e-02
-4.13459271e-01 9.48077738e-01 3.47426444e-01 -7.75947750e-01
3.47357839e-01 8.03364962e-02 -3.03466797e-01 -2.79966712e-01
7.28644848e-01 1.61130965e-01 -3.38712811e-01 -2.22488597e-01
5.45366518e-02 4.31156099e-01 2.32970089e-01 -2.75013149e-01
1.36824131e+00 3.79584283e-01 -1.61836088e-01 3.06889325e-01
4.22264218e-01 -8.85075778e-02 -9.86308098e-01 -7.22706169e-02
2.88216382e-01 -2.76789684e-02 -1.58217981e-01 -9.60199118e-01
-2.75058001e-01 7.85203755e-01 -2.93035730e-04 5.39776683e-01
9.55275595e-01 -1.76147401e-01 2.41965935e-01 9.26264763e-01
7.89011240e-01 -1.18738341e+00 2.65583843e-01 5.74555457e-01
6.98022187e-01 -1.04525137e+00 1.78529248e-01 -3.97515655e-01
-4.20254290e-01 1.09474099e+00 8.12971890e-01 -2.32500985e-01
2.47298285e-01 3.18281591e-01 -3.44533235e-01 -2.29807124e-01
-7.23733008e-01 -1.60895944e-01 2.26339996e-01 8.64957571e-01
6.84870958e-01 1.78838849e-01 -8.70569572e-02 2.04089493e-01
-7.06348836e-01 1.90322757e-01 5.75319886e-01 1.29719889e+00
-1.16830075e+00 -9.90537822e-01 -4.86189961e-01 1.08141631e-01
-1.92805976e-01 -3.70736048e-02 -3.80457491e-01 1.03796434e+00
4.04456526e-01 9.02877033e-01 4.85306270e-02 -8.85993987e-02
6.97660893e-02 4.10726905e-01 1.06143117e+00 -8.53809237e-01
-8.66327524e-01 -5.87154686e-01 3.54777098e-01 -1.98283657e-01
-7.30783165e-01 -5.19995630e-01 -1.65868723e+00 -7.53791213e-01
2.12554619e-01 4.59367156e-01 4.07545388e-01 1.18944407e+00
-4.23950583e-01 5.71491241e-01 2.19657719e-01 -9.43991125e-01
-4.19487923e-01 -6.47716463e-01 -3.56939971e-01 2.90460318e-01
-3.60566936e-02 -7.22650170e-01 -3.97284746e-01 4.88181502e-01]
|
[4.135249614715576, 1.4664220809936523]
|
88b4e9a8-a8c2-4601-b02e-a291deeaf9a3
|
deeptract-a-probabilistic-deep-learning
|
1812.05129
| null |
https://arxiv.org/abs/1812.05129v3
|
https://arxiv.org/pdf/1812.05129v3.pdf
|
DeepTract: A Probabilistic Deep Learning Framework for White Matter Fiber Tractography
|
We present DeepTract, a deep-learning framework for estimating white matter fibers orientation and streamline tractography. We adopt a data-driven approach for fiber reconstruction from diffusion weighted images (DWI), which does not assume a specific diffusion model. We use a recurrent neural network for mapping sequences of DWI values into probabilistic fiber orientation distributions. Based on these estimations, our model facilitates both deterministic and probabilistic streamline tractography. We quantitatively evaluate our method using the Tractometer tool, demonstrating competitive performance with state-of-the art classical and machine learning based tractography algorithms. We further present qualitative results of bundle-specific probabilistic tractography obtained using our method. The code is publicly available at: https://github.com/itaybenou/DeepTract.git.
|
['Tammy Riklin-Raviv', 'Itay Benou']
|
2018-12-12
| null | null | null | null |
['probabilistic-deep-learning', 'white-matter-fiber-tractography']
|
['computer-vision', 'medical']
|
[-5.06721079e-01 -4.70543265e-01 -6.20319955e-02 -2.25457072e-01
-6.31316304e-01 -7.41553664e-01 4.00284082e-01 -4.60943848e-01
-5.54117441e-01 9.81037557e-01 8.09613943e-01 -3.69506747e-01
-3.60298961e-01 -6.71604335e-01 -6.01308584e-01 -7.02367544e-01
-5.72302997e-01 7.94054091e-01 1.37758106e-01 2.78389573e-01
2.97664583e-01 7.49298871e-01 -5.26521266e-01 1.86580092e-01
6.15226030e-01 5.38224936e-01 5.00211895e-01 9.37761128e-01
1.17644831e-03 6.68252766e-01 -1.54415786e-01 -1.28073856e-01
1.16438180e-01 -3.71301025e-01 -9.72334802e-01 -3.15703630e-01
2.55212396e-01 -8.41818631e-01 -8.60288858e-01 1.08739161e+00
6.98197663e-01 -2.23180994e-01 9.45780456e-01 -7.43207812e-01
-7.08763480e-01 9.31868851e-01 -2.04888389e-01 8.78108144e-01
-1.89941674e-02 5.01476943e-01 9.33381617e-01 -7.57753313e-01
1.16579521e+00 8.45957279e-01 7.55851626e-01 5.30847549e-01
-1.56711423e+00 -3.44883025e-01 -4.23731096e-02 2.99536377e-01
-1.19344985e+00 -1.39990747e-01 4.03868228e-01 -8.91198933e-01
1.05154228e+00 -3.04781228e-01 1.26739049e+00 1.30745590e+00
5.62406301e-01 9.21037078e-01 1.65516245e+00 1.20701902e-01
8.76676571e-03 -5.50089061e-01 2.41695106e-01 7.56968200e-01
-4.16993760e-02 5.03874242e-01 -1.99108094e-01 -4.95800264e-02
1.35021329e+00 -2.19131336e-02 -2.44608700e-01 -1.55788928e-01
-1.90725732e+00 6.58475220e-01 5.87203264e-01 3.54202390e-01
-6.06810510e-01 7.92756796e-01 4.74121749e-01 1.59535572e-01
4.34997976e-01 -1.11159667e-01 -2.01852903e-01 -3.08016032e-01
-1.31593716e+00 6.70979559e-01 5.97668111e-01 6.76749051e-01
3.00861925e-01 1.37325495e-01 -2.05334887e-01 6.78700089e-01
6.04756296e-01 6.19474351e-01 4.21671510e-01 -1.42533183e+00
1.49010971e-01 -1.99083149e-01 -1.22222923e-01 -3.87226343e-01
-8.22498322e-01 -6.91545904e-01 -7.72246659e-01 4.96090382e-01
8.11474979e-01 -3.52101982e-01 -8.20962071e-01 1.36670768e+00
-2.86128651e-02 1.11232460e-01 -6.58915460e-01 1.24776316e+00
3.63897890e-01 3.43675822e-01 -1.07953548e-01 2.02364698e-01
1.11628246e+00 -6.67376518e-01 -5.75158060e-01 5.43400705e-01
7.71321356e-01 -7.09162951e-01 7.44910479e-01 4.08700019e-01
-1.24121571e+00 1.86409637e-01 -5.69044828e-01 -1.50483012e-01
4.76839542e-02 2.59910643e-01 4.13388699e-01 4.81156141e-01
-1.49652219e+00 9.68061328e-01 -1.40404141e+00 -2.12843180e-01
8.94755244e-01 4.36871685e-02 -2.19900936e-01 5.91370724e-02
-8.03859711e-01 8.04372132e-01 2.19726060e-02 7.89755285e-02
-1.40851593e+00 -9.58421171e-01 -1.81688830e-01 -3.22751999e-01
-2.97084004e-01 -1.02127004e+00 1.17298281e+00 -1.29069567e-01
-1.38089752e+00 6.47333205e-01 -3.50217819e-01 -7.99211144e-01
7.33323097e-01 -1.94728822e-01 -1.58091225e-02 4.97656792e-01
8.18066224e-02 7.42270291e-01 5.67299902e-01 -9.42370594e-01
-8.05188566e-02 -2.91238278e-01 -2.51407802e-01 -7.47542158e-02
2.71665305e-01 2.02635199e-01 2.80195832e-01 -1.00357294e+00
2.05093101e-01 -9.82718349e-01 -1.87542528e-01 6.24491453e-01
-8.62988949e-01 -1.47303447e-01 1.07956223e-01 -9.80914474e-01
8.17035675e-01 -1.30743873e+00 3.66357982e-01 3.01283896e-01
7.89236665e-01 -2.41314143e-01 -1.14565782e-01 4.63783681e-01
-2.55369812e-01 6.44915253e-02 -5.69499075e-01 -3.17856759e-01
-4.15366814e-02 -6.12401823e-03 -2.86183298e-01 9.48441505e-01
-3.00959468e-01 9.30014133e-01 -1.06945956e+00 -2.51808494e-01
3.12690318e-01 6.21716321e-01 -5.29840529e-01 7.30445534e-02
-6.97719082e-02 7.87370443e-01 -1.41796827e-01 4.09164190e-01
5.84174812e-01 -1.15042880e-01 1.52128428e-01 -4.28203791e-01
-3.11716020e-01 5.40720344e-01 -7.83943295e-01 2.07345819e+00
-3.08472127e-01 6.57641888e-01 -2.48066485e-01 -8.11237335e-01
3.87173295e-01 3.69826198e-01 8.19040537e-01 -1.23765752e-01
1.04487330e-01 3.63068193e-01 4.15737778e-01 -3.44475150e-01
-1.05203815e-01 -2.37724125e-01 7.71352649e-01 1.28415298e+00
5.97253919e-01 -1.01362579e-02 3.45740527e-01 3.90623659e-01
1.20963430e+00 6.86773300e-01 -2.05864340e-01 -6.23808861e-01
-2.86981929e-02 -6.55905250e-03 2.17419162e-01 6.36366129e-01
-4.33220059e-01 8.57565880e-01 5.32900453e-01 -5.00283599e-01
-1.43857849e+00 -1.86376476e+00 -4.45816875e-01 3.74380529e-01
-3.83821785e-01 -2.84564018e-01 -1.00362146e+00 -5.84525824e-01
-8.97935256e-02 4.61303622e-01 -5.75767398e-01 5.02126753e-01
-5.94075322e-01 -1.05375159e+00 6.60173535e-01 5.47851920e-01
4.20032740e-02 -1.07559776e+00 -9.00217891e-02 4.34946060e-01
-2.69919038e-01 -8.63969207e-01 -5.53013742e-01 -2.62732446e-01
-1.30287194e+00 -8.31330478e-01 -1.39230120e+00 -4.37746733e-01
5.84945023e-01 1.96511243e-02 9.74193990e-01 5.42216152e-02
-5.12306750e-01 1.73745260e-01 -6.07404523e-02 2.82302976e-01
-3.56333137e-01 2.89876431e-01 2.79889286e-01 -5.33646345e-01
9.03093070e-02 -1.19167948e+00 -1.31199622e+00 1.45560011e-01
-5.57505250e-01 -6.42359676e-03 1.47393778e-01 5.92922390e-01
6.51277483e-01 -3.81216586e-01 4.91705567e-01 -4.77822334e-01
8.40996742e-01 -7.30472088e-01 -4.66211319e-01 -2.31262073e-01
-3.95061046e-01 1.15328774e-01 2.49144167e-01 -1.54681951e-01
-6.55939996e-01 -2.82861978e-01 -6.96401954e-01 -2.45121434e-01
-2.68580168e-01 3.46463889e-01 5.88150144e-01 -1.26877993e-01
6.95341051e-01 3.38808268e-01 3.93238872e-01 -5.98707795e-01
6.84541583e-01 2.95750588e-01 2.97246337e-01 -5.91847420e-01
5.75750291e-01 9.77026761e-01 -2.54833251e-01 -3.46694022e-01
-4.04644161e-01 -2.34521598e-01 -1.13384366e+00 -6.17374957e-01
7.32416987e-01 -5.95590234e-01 -7.23878682e-01 7.49727011e-01
-1.26315522e+00 -8.17461669e-01 -1.73651740e-01 1.02368844e+00
-1.10640550e+00 4.44215149e-01 -1.14879179e+00 -4.90859330e-01
-6.44088089e-01 -1.45100248e+00 6.89686179e-01 -3.86035919e-01
-4.32716042e-01 -1.36815798e+00 4.74252611e-01 -2.26599872e-02
5.55065751e-01 3.61657725e-03 7.76733875e-01 -1.30694047e-01
-7.66719997e-01 2.53902048e-01 -4.54778314e-01 3.03196996e-01
7.13675795e-03 -8.55461359e-02 -5.82838953e-01 -1.32543087e-01
-2.02525467e-01 -7.96917975e-02 9.03526723e-01 9.76294875e-01
1.19019973e+00 -1.33550927e-01 -1.82801634e-01 9.03320312e-01
1.39813197e+00 -5.12096643e-01 5.79209805e-01 4.13309187e-01
8.17233026e-01 5.73470235e-01 -2.39577696e-01 3.24772537e-01
5.63000560e-01 5.41853189e-01 3.28878999e-01 9.64396670e-02
-8.34985733e-01 -7.01624006e-02 1.60918906e-01 1.19193435e+00
-6.11536145e-01 1.15896441e-01 -1.04827583e+00 7.97184229e-01
-1.59915102e+00 -1.17465174e+00 -3.52028251e-01 1.82476366e+00
1.04939497e+00 5.22518866e-02 7.04509914e-01 -1.45410195e-01
5.60016811e-01 9.92970914e-03 -7.26127923e-01 -3.26314941e-02
-2.04026978e-02 2.21695192e-02 7.27446377e-01 6.35241449e-01
-6.50828481e-01 8.27725291e-01 6.93796110e+00 7.15130329e-01
-9.78668749e-01 6.68447971e-01 3.46452832e-01 -4.97631639e-01
-4.43058670e-01 -2.46238619e-01 -3.80746603e-01 5.94524205e-01
8.85768473e-01 7.32749254e-02 1.12776375e+00 2.38084137e-01
7.41830707e-01 8.22227001e-02 -5.36355197e-01 7.21994877e-01
-6.00832641e-01 -1.85611355e+00 -1.28631338e-01 3.93882185e-01
5.69848657e-01 1.32192194e+00 1.30399419e-02 -4.70095187e-01
6.84630811e-01 -8.53856862e-01 1.03773355e+00 1.08040476e+00
7.94654787e-01 -3.84604394e-01 4.11349297e-01 -3.17173451e-02
-8.13966215e-01 3.06556225e-01 -3.65617126e-01 2.74970353e-01
8.54270041e-01 1.15809989e+00 -6.18393064e-01 2.89155334e-01
8.24293077e-01 1.21195030e+00 9.34233814e-02 1.50508344e+00
-5.20884037e-01 1.05912316e+00 -4.41416502e-01 2.44612694e-01
1.94554687e-01 -4.73045021e-01 9.76961613e-01 1.34231043e+00
3.54358703e-01 -2.64387518e-01 -3.47709984e-01 1.53972828e+00
-8.34611356e-02 -1.89302772e-01 -4.10514891e-01 -6.33479506e-02
3.79605651e-01 1.32081044e+00 -1.00179780e+00 -2.90644586e-01
-3.83479223e-02 9.04099643e-01 5.79898238e-01 6.18734181e-01
-4.99070495e-01 -1.08694874e-01 8.85499358e-01 3.29720676e-01
1.43296331e-01 -9.40182507e-01 -6.52573466e-01 -1.31882584e+00
-9.53094736e-02 -2.69512057e-01 -7.84961060e-02 -8.29674363e-01
-1.66721284e+00 6.96065843e-01 -6.99307024e-02 -1.04441583e+00
-2.08238170e-01 -7.52000630e-01 -7.14825511e-01 1.18176675e+00
-1.52116883e+00 -9.75220084e-01 3.24336022e-01 5.68739057e-01
2.81624645e-01 -1.05947711e-01 5.94414651e-01 2.18514949e-01
-3.28610927e-01 -6.16568811e-02 4.33065653e-01 2.28373170e-01
3.72491628e-01 -1.64299369e+00 9.91929471e-01 7.65839040e-01
1.01985581e-01 1.00440502e+00 6.26331866e-01 -8.96246314e-01
-1.07919788e+00 -1.04108024e+00 5.69005132e-01 -4.72995371e-01
1.39811802e+00 -3.99879783e-01 -7.44820297e-01 8.47435594e-01
1.45858228e-01 2.76075095e-01 5.30460596e-01 6.36851694e-03
-2.06707209e-01 2.64992833e-01 -1.09031689e+00 5.86811662e-01
1.41525781e+00 -6.91383719e-01 -5.24633229e-01 8.27846110e-01
3.19125324e-01 -2.00252667e-01 -1.38969684e+00 -2.00644240e-01
1.00161099e+00 -8.97686601e-01 1.27021408e+00 -3.08036923e-01
3.70278955e-01 -2.12559581e-01 3.09335202e-01 -1.72663033e+00
-4.75741118e-01 -4.82247561e-01 -2.99044013e-01 5.18385530e-01
3.97934526e-01 -9.17140901e-01 6.78127170e-01 -9.72852781e-02
-1.09312870e-01 -9.42319751e-01 -1.10047495e+00 -9.80265141e-01
8.05278182e-01 -6.00458622e-01 4.61324602e-01 4.82625782e-01
-8.14125091e-02 -5.13588488e-01 -6.87470287e-02 -1.20556042e-01
1.10508001e+00 -1.20164640e-01 -9.72287506e-02 -7.08338857e-01
-1.27886534e-01 -7.26853728e-01 -3.71873111e-01 -9.97789800e-01
2.26849973e-01 -1.60591435e+00 -9.86429825e-02 -2.03541660e+00
2.01532859e-02 -5.17802417e-01 -3.46180022e-01 2.97509760e-01
3.16541076e-01 5.63742816e-01 1.53139606e-01 7.55338609e-01
3.26132067e-02 5.02072215e-01 1.56590927e+00 1.38630837e-01
1.93894431e-01 -1.54225066e-01 -1.09592222e-01 7.34860718e-01
1.29652226e+00 -8.34058166e-01 -1.96926996e-01 -7.90980875e-01
1.05163760e-01 -2.31441073e-02 7.26349056e-01 -9.68768477e-01
7.82697275e-02 3.38210583e-01 3.83181244e-01 -6.10273540e-01
7.13033080e-02 -2.63891190e-01 -2.03903705e-01 6.61564946e-01
-4.01513487e-01 2.52624243e-01 -1.53328955e-01 2.78701812e-01
5.72572529e-01 -2.92951465e-01 6.43843353e-01 -3.46895427e-01
-1.47151649e-01 8.21108162e-01 -1.07701814e+00 2.23392714e-02
2.80498028e-01 8.52736756e-02 -5.70964456e-01 -1.60529912e-01
-1.27027953e+00 -1.97694510e-01 3.00198644e-01 -3.42207141e-02
8.00875366e-01 -1.28849876e+00 -1.05067050e+00 -1.80780768e-01
-3.38637471e-01 -8.03608656e-01 4.37342048e-01 1.35734689e+00
-1.00339746e+00 3.74760568e-01 -5.70847452e-01 -6.82607770e-01
-4.21107531e-01 1.28862560e-01 7.39923358e-01 -1.34228393e-01
-1.33505177e+00 6.86854124e-01 -1.56467825e-01 -7.87663102e-01
-3.57880294e-01 -6.26623750e-01 -1.38188421e-03 -2.68727750e-01
6.78754508e-01 6.31901264e-01 -2.96734776e-02 -4.28723931e-01
-1.86295047e-01 5.30610800e-01 -1.12662688e-01 -7.46688545e-01
1.47435057e+00 -4.50765610e-01 -4.34849858e-01 4.14930969e-01
1.16831219e+00 -3.40733975e-01 -1.46179509e+00 -2.33998731e-01
-8.14712048e-02 -1.49247319e-01 5.91952980e-01 -1.15323448e+00
-1.27897894e+00 1.10947585e+00 8.49928260e-01 -5.31027839e-02
4.30331796e-01 -7.44777396e-02 1.31126714e+00 -9.17588547e-03
5.84427297e-01 -6.68857574e-01 -4.13381815e-01 4.40328807e-01
9.98074234e-01 -7.40050972e-01 -5.55397421e-02 -1.08188026e-01
-2.19862789e-01 1.40923393e+00 -1.01735517e-01 -8.07049096e-01
1.05289841e+00 4.69820201e-01 1.71840027e-01 -4.76173967e-01
-4.92572248e-01 -2.51087546e-01 1.75113276e-01 8.84161770e-01
4.75102752e-01 3.03323120e-01 -3.52043092e-01 1.75932162e-02
-3.63705963e-01 5.12542665e-01 6.40585124e-01 6.08714342e-01
-3.31676871e-01 -1.07166171e+00 -9.98975709e-02 8.18260431e-01
-4.08877015e-01 -5.96524298e-01 3.53227615e-01 2.87366569e-01
-3.25785220e-01 2.75447547e-01 5.35012335e-02 -4.09416147e-02
-2.72603482e-01 -3.51211518e-01 9.23493624e-01 -4.81301218e-01
-4.08271730e-01 1.37082413e-01 8.92722830e-02 -7.40661919e-01
-4.82284069e-01 -9.73440766e-01 -1.50303233e+00 -3.77679348e-01
2.96157569e-01 -2.28171706e-01 9.11896884e-01 1.09141016e+00
2.63347507e-01 6.17149889e-01 1.99382752e-01 -1.12182415e+00
-2.65066832e-01 -1.06594884e+00 -8.10441613e-01 -1.37163192e-01
4.44475293e-01 -8.34217429e-01 -1.89605996e-01 4.33559641e-02]
|
[13.849977493286133, -2.3789420127868652]
|
d3927fe6-fc88-44d9-ae58-ec3c26c1c876
|
w2n-switching-from-weak-supervision-to-noisy
|
2207.12104
| null |
https://arxiv.org/abs/2207.12104v1
|
https://arxiv.org/pdf/2207.12104v1.pdf
|
W2N:Switching From Weak Supervision to Noisy Supervision for Object Detection
|
Weakly-supervised object detection (WSOD) aims to train an object detector only requiring the image-level annotations. Recently, some works have managed to select the accurate boxes generated from a well-trained WSOD network to supervise a semi-supervised detection framework for better performance. However, these approaches simply divide the training set into labeled and unlabeled sets according to the image-level criteria, such that sufficient mislabeled or wrongly localized box predictions are chosen as pseudo ground-truths, resulting in a sub-optimal solution of detection performance. To overcome this issue, we propose a novel WSOD framework with a new paradigm that switches from weak supervision to noisy supervision (W2N). Generally, with given pseudo ground-truths generated from the well-trained WSOD network, we propose a two-module iterative training algorithm to refine pseudo labels and supervise better object detector progressively. In the localization adaptation module, we propose a regularization loss to reduce the proportion of discriminative parts in original pseudo ground-truths, obtaining better pseudo ground-truths for further training. In the semi-supervised module, we propose a two tasks instance-level split method to select high-quality labels for training a semi-supervised detector. Experimental results on different benchmarks verify the effectiveness of W2N, and our W2N outperforms all existing pure WSOD methods and transfer learning methods. Our code is publicly available at https://github.com/1170300714/w2n_wsod.
|
['WangMeng Zuo', 'Erjin Zhou', 'Bowen Dong', 'Yiping Bao', 'Zitong Huang']
|
2022-07-25
| null | null | null | null |
['weakly-supervised-object-detection']
|
['computer-vision']
|
[ 2.07621396e-01 2.68637985e-01 -3.47546965e-01 -4.60710257e-01
-9.73570466e-01 -3.18327218e-01 4.02089477e-01 -2.47833300e-02
-6.42999828e-01 5.85431993e-01 -2.03995347e-01 1.71817064e-01
2.02974215e-01 -6.25217378e-01 -8.93769622e-01 -8.00110698e-01
3.77426982e-01 5.60233533e-01 1.01585865e+00 1.13851644e-01
5.98420314e-02 2.23003253e-01 -1.51536322e+00 2.73737431e-01
8.81733179e-01 1.14479113e+00 7.50563979e-01 2.02353820e-01
-9.98333022e-02 6.12400532e-01 -4.97554928e-01 -1.94896057e-01
5.18415093e-01 -4.94660318e-01 -6.29921079e-01 3.43930095e-01
4.10073370e-01 -1.66982785e-01 3.11294701e-02 1.39977825e+00
4.38249528e-01 -1.35274038e-01 6.61545277e-01 -1.20658052e+00
-4.12161469e-01 5.10790884e-01 -7.45373070e-01 8.07478949e-02
-7.23321959e-02 3.07195097e-01 9.60841298e-01 -1.27679348e+00
6.04162872e-01 1.04552472e+00 6.21417642e-01 7.42019236e-01
-1.48545754e+00 -8.08720469e-01 3.31449598e-01 1.45889774e-01
-1.73610282e+00 -2.84871489e-01 7.75757015e-01 -3.88045281e-01
3.34907353e-01 1.39068156e-01 4.34786975e-01 8.97318184e-01
-2.92005658e-01 1.06351769e+00 1.14567566e+00 -5.62070787e-01
2.42009968e-01 6.89491153e-01 1.43384889e-01 8.93833399e-01
4.27191705e-01 1.13820404e-01 -3.56819540e-01 -7.96364918e-02
6.33278906e-01 1.91660672e-01 -2.38901913e-01 -7.68848658e-01
-1.11608362e+00 6.26557350e-01 8.78814101e-01 4.52480853e-01
-1.36486009e-01 -2.42408931e-01 3.04780483e-01 -4.97103892e-02
6.75921202e-01 3.29475820e-01 -4.70126957e-01 6.04605138e-01
-1.01886070e+00 6.83346763e-02 3.81365538e-01 1.02209294e+00
1.05189526e+00 -3.98090512e-01 -6.07616425e-01 8.77343774e-01
4.79149401e-01 5.02216935e-01 5.09463251e-01 -5.54873466e-01
4.28088009e-01 1.08728468e+00 4.06561166e-01 -6.96425319e-01
-2.47751340e-01 -6.51336551e-01 -6.45405233e-01 2.11345717e-01
5.49520493e-01 4.74586114e-02 -1.15462351e+00 1.52453530e+00
5.64408302e-01 2.63880789e-01 -2.53074557e-01 1.26473355e+00
8.01030099e-01 6.16237581e-01 1.96190417e-01 -1.31382436e-01
1.28099096e+00 -1.26351261e+00 -3.85006934e-01 -4.70948994e-01
7.49958515e-01 -6.04253232e-01 1.12058914e+00 1.76728308e-01
-6.72202051e-01 -9.40503538e-01 -1.07344794e+00 1.92615762e-01
-2.93324411e-01 9.17700529e-01 1.98354404e-02 3.88775200e-01
-7.88995683e-01 4.16579276e-01 -7.94719040e-01 -3.00982982e-01
7.04409897e-01 2.29441836e-01 -1.54253662e-01 -7.66751766e-02
-9.01911736e-01 8.59281361e-01 9.09800947e-01 1.82496309e-01
-1.21319413e+00 -4.82472271e-01 -6.02206171e-01 -2.36288160e-01
7.46988714e-01 -3.74305904e-01 1.23452091e+00 -1.11810982e+00
-1.09759426e+00 1.22498488e+00 -2.57474538e-02 -3.44972879e-01
6.63251460e-01 -1.72138065e-02 -1.83560058e-01 3.04497802e-03
4.71893966e-01 8.51918578e-01 9.38979924e-01 -1.65512359e+00
-1.01419878e+00 -3.64765167e-01 -1.84936207e-02 1.31203100e-01
-3.43478084e-01 1.85201913e-02 -6.54014230e-01 -4.67365563e-01
3.60856801e-01 -8.56131315e-01 -3.66340965e-01 2.91421652e-01
-6.33965433e-01 -6.30485415e-01 8.26872289e-01 -1.76129416e-01
9.32436943e-01 -2.19487858e+00 -3.10950004e-03 -1.35091413e-02
3.39987844e-01 6.27587497e-01 -1.60176113e-01 -1.10080682e-01
5.23295812e-02 -1.75840169e-01 -4.46185589e-01 -7.62208104e-01
-1.25781894e-01 1.67060360e-01 -1.27218887e-01 5.63875556e-01
5.37102997e-01 7.38572955e-01 -1.08713853e+00 -8.95975709e-01
1.77757263e-01 1.76916718e-01 -2.04812482e-01 4.85584170e-01
-3.94213587e-01 5.45547962e-01 -5.90786994e-01 6.52503908e-01
8.16375911e-01 -3.42076987e-01 -2.11635724e-01 -2.24705651e-01
-1.73775926e-01 -8.65057558e-02 -1.40101218e+00 1.64202166e+00
-2.12834492e-01 1.52254134e-01 1.15338728e-01 -1.14768958e+00
1.06923103e+00 -1.00576431e-02 2.15452015e-01 -4.51769054e-01
2.92590380e-01 4.53221321e-01 -2.17897251e-01 -4.15165603e-01
1.13674648e-01 -1.92284271e-01 9.23311785e-02 1.36136204e-01
4.18496996e-01 -1.12977326e-01 2.98816234e-01 1.37595460e-01
8.67495179e-01 4.35126096e-01 2.30097145e-01 -1.97494552e-01
8.33840728e-01 2.51001358e-01 9.00044799e-01 9.04477179e-01
-5.23294032e-01 8.23736012e-01 1.98570445e-01 -3.35755527e-01
-9.98475730e-01 -8.77606928e-01 -5.68498433e-01 1.30138147e+00
5.38504064e-01 -1.74998149e-01 -9.32938278e-01 -1.26401520e+00
-1.30968794e-01 5.05871773e-01 -6.57996893e-01 -2.01026335e-01
-3.77458841e-01 -7.29016602e-01 3.83130640e-01 5.17637074e-01
5.98064899e-01 -1.17432833e+00 -3.53546798e-01 7.65643939e-02
2.36402638e-02 -9.48265374e-01 -5.43743551e-01 4.96441036e-01
-6.78787649e-01 -1.01851010e+00 -9.61670041e-01 -1.03656149e+00
1.19030070e+00 5.23125172e-01 8.41779053e-01 2.58027554e-01
-3.41916502e-01 -1.31747276e-01 -4.94003356e-01 -4.30179149e-01
-3.86958629e-01 5.55876940e-02 1.28275767e-01 3.00967544e-01
4.90374446e-01 8.91446311e-04 -6.09502316e-01 7.29454994e-01
-9.11597252e-01 1.86977815e-02 9.03967261e-01 8.76880348e-01
8.63205731e-01 -2.06437349e-01 5.38774252e-01 -1.08431613e+00
5.54588996e-02 -3.33617538e-01 -9.15256858e-01 2.67979860e-01
-7.70875096e-01 4.15553674e-02 7.69670844e-01 -4.93931681e-01
-1.22099519e+00 6.25210524e-01 6.04319759e-02 -5.44517934e-01
-4.14110243e-01 2.13931524e-03 -2.08577260e-01 -1.18434347e-01
9.61410463e-01 4.41168904e-01 -1.77584097e-01 -4.92689908e-01
3.14953297e-01 6.46135092e-01 4.29937601e-01 -3.20242226e-01
1.15649927e+00 5.89311481e-01 -4.95251149e-01 -2.58789629e-01
-1.46375525e+00 -9.57597375e-01 -9.90856647e-01 -2.04137519e-01
8.51198256e-01 -8.91813517e-01 -1.08407952e-01 2.99407572e-01
-1.09121943e+00 -2.87795633e-01 -5.28977096e-01 4.39585894e-01
-2.91661948e-01 2.12288231e-01 -2.94936240e-01 -8.44357789e-01
-2.32504353e-01 -1.08385313e+00 1.41951823e+00 3.65356803e-01
2.80609697e-01 -5.58697760e-01 1.02817841e-01 3.47201556e-01
-4.42334125e-03 -3.99879292e-02 1.98148578e-01 -8.65466177e-01
-6.75811768e-01 -5.05113840e-01 -6.02042735e-01 6.54406369e-01
-1.18360400e-01 -3.99043739e-01 -1.03321135e+00 -3.45012248e-01
-6.69098049e-02 -7.11541474e-01 1.12109566e+00 2.18949452e-01
1.08934438e+00 -8.37322846e-02 -6.93016052e-01 4.87048715e-01
1.45231104e+00 -1.36038050e-01 1.21071257e-01 2.84677505e-01
6.02489948e-01 4.75864232e-01 1.06378126e+00 2.64581800e-01
1.02534555e-01 5.87255538e-01 4.25098628e-01 -2.70893395e-01
-2.45421693e-01 -5.69007099e-01 4.12364751e-01 3.63431603e-01
-2.78259888e-02 -4.83023897e-02 -7.35833526e-01 5.10301352e-01
-1.99343181e+00 -6.77090824e-01 -1.88038334e-01 2.19054818e+00
9.25064743e-01 4.80666459e-01 2.94587016e-01 7.39539973e-03
1.05269158e+00 1.54745892e-01 -6.88503683e-01 4.38551575e-01
3.58346589e-02 -2.51268327e-01 6.08504176e-01 1.72952309e-01
-1.31616306e+00 1.06719017e+00 4.60448837e+00 1.26379204e+00
-8.63617241e-01 5.76368690e-01 6.52647257e-01 4.48896848e-02
1.83618397e-01 3.54981422e-02 -1.38615561e+00 4.71390456e-01
3.67617816e-01 2.02130124e-01 -8.78797844e-02 1.23965418e+00
2.56417900e-01 -2.07238659e-01 -9.69265103e-01 8.47813010e-01
1.59465447e-01 -1.08302712e+00 -2.49537095e-01 -3.06433201e-01
9.17832375e-01 2.13802770e-01 -2.09383458e-01 5.47295868e-01
1.87492177e-01 -4.92599696e-01 9.81743932e-01 2.76572615e-01
6.32165253e-01 -2.70004064e-01 9.37557459e-01 9.24175620e-01
-1.35876048e+00 -2.32357711e-01 -7.03429520e-01 2.70071000e-01
-6.50457442e-02 6.33700192e-01 -1.00565815e+00 3.84950161e-01
8.58509660e-01 6.47408664e-01 -8.84548545e-01 1.23454404e+00
-6.25056088e-01 6.87246799e-01 -3.26111823e-01 -2.01410428e-01
2.29929030e-01 -1.12028100e-01 5.55199206e-01 1.20854306e+00
-6.17660359e-02 7.44834915e-02 5.27561665e-01 1.25189626e+00
-5.20514436e-02 1.07186295e-01 -2.78797746e-01 5.09677827e-01
3.09421837e-01 1.60304916e+00 -1.08408809e+00 -4.58351851e-01
-3.45540494e-01 8.69179606e-01 5.95079184e-01 2.46948048e-01
-8.59853923e-01 -5.29666245e-02 -1.24036357e-01 3.35393786e-01
2.32863158e-01 6.99550211e-02 -2.15935156e-01 -1.16461396e+00
2.26685837e-01 -4.56897259e-01 4.82430577e-01 -6.36737645e-01
-1.34173083e+00 7.08818614e-01 4.16248403e-02 -1.52033722e+00
2.56507605e-01 -6.64669096e-01 -5.54246664e-01 6.47180438e-01
-1.68097818e+00 -1.31658661e+00 -5.46488285e-01 3.31265986e-01
7.37104833e-01 2.14981772e-02 4.41973358e-01 2.95331270e-01
-7.95315802e-01 4.63531137e-01 1.37257483e-02 3.63487124e-01
8.20739567e-01 -1.34613407e+00 -1.62625462e-02 9.06119585e-01
2.15266570e-01 3.00805509e-01 4.27301019e-01 -5.74803829e-01
-8.01164746e-01 -1.55336440e+00 5.89114130e-01 -4.76086855e-01
3.65236551e-01 -6.43827975e-01 -9.93615270e-01 5.29780209e-01
-2.15043932e-01 6.19282305e-01 2.38222703e-01 -1.69230029e-01
-1.70212418e-01 -3.76125485e-01 -1.17313993e+00 2.84554601e-01
1.23385227e+00 -3.01854342e-01 -7.44585156e-01 6.72399938e-01
7.05888093e-01 -2.89617777e-01 -2.91365504e-01 5.79922259e-01
3.01981196e-02 -8.65960658e-01 8.41128647e-01 -1.83410630e-01
1.56539991e-01 -8.84086609e-01 3.42629910e-01 -1.03567672e+00
-3.23011160e-01 -9.11780000e-02 8.25189874e-02 1.20143616e+00
4.96857584e-01 -4.75919604e-01 9.50365126e-01 2.35154286e-01
-1.97951272e-01 -7.74997234e-01 -8.94739389e-01 -9.63998616e-01
-4.15629297e-01 -2.60603428e-01 1.73602745e-01 7.25838423e-01
-3.28066617e-01 2.47233391e-01 -1.77424461e-01 4.03315753e-01
8.24323356e-01 2.82299668e-01 8.39534521e-01 -1.14561546e+00
-1.42079413e-01 -2.85307854e-01 -4.13129777e-01 -1.38112354e+00
1.78144779e-02 -1.07080412e+00 6.99596524e-01 -1.36306334e+00
6.48200035e-01 -6.65236175e-01 -4.44328696e-01 7.39364803e-01
-4.42339659e-01 6.90540552e-01 8.93291757e-02 5.52253246e-01
-1.15114427e+00 6.78785801e-01 1.29547632e+00 -2.01750785e-01
-2.35352308e-01 3.14732641e-01 -3.61202270e-01 8.73420656e-01
5.14418185e-01 -8.84396315e-01 -2.47055694e-01 -1.64995626e-01
-2.79047310e-01 -3.38418156e-01 5.98448277e-01 -9.99503195e-01
3.67778838e-01 3.17103975e-02 4.82562244e-01 -7.55616426e-01
7.33175203e-02 -8.79177034e-01 -4.58192170e-01 5.67716122e-01
-3.04398715e-01 -7.70146489e-01 -1.52091205e-01 7.29722023e-01
-2.72852659e-01 -6.18064582e-01 1.16177273e+00 -1.36691317e-01
-9.04482543e-01 4.57993090e-01 9.06513929e-02 -2.10316875e-03
1.32855368e+00 -2.71465600e-01 -1.95215195e-02 2.01280028e-01
-7.62388051e-01 4.73068327e-01 2.37769738e-01 3.42055321e-01
5.11582911e-01 -1.24941158e+00 -7.44937062e-01 1.60564139e-01
5.87022245e-01 5.38057804e-01 -3.29738371e-02 7.88849533e-01
-1.96705580e-01 2.69694865e-01 1.72242239e-01 -1.00865471e+00
-9.83206511e-01 6.67627752e-01 4.23397690e-01 -2.85180569e-01
-4.38674778e-01 1.21089041e+00 3.47912937e-01 -6.95608020e-01
4.08488184e-01 -3.91365618e-01 -2.18799204e-01 -1.55478204e-02
5.13463199e-01 9.94660333e-02 1.04452148e-02 -5.95159650e-01
-3.93969953e-01 4.72271442e-01 -8.93463343e-02 1.74822375e-01
1.18450022e+00 -4.01877239e-02 9.93995443e-02 4.47220355e-01
9.76351440e-01 -2.98412204e-01 -1.59829593e+00 -7.02591181e-01
2.63445973e-01 -3.88749719e-01 8.13114643e-02 -6.12724841e-01
-9.54336345e-01 5.79927504e-01 8.19603205e-01 8.49049091e-02
1.05440271e+00 5.37967324e-01 4.88083184e-01 2.57140696e-01
4.70879078e-01 -1.13263845e+00 2.70016044e-01 1.96089655e-01
7.66240895e-01 -1.64821839e+00 -6.33949488e-02 -5.47259808e-01
-5.95906615e-01 8.23545814e-01 1.24791825e+00 -3.71262223e-01
5.32253504e-01 1.20644547e-01 3.81157696e-02 -5.83348833e-02
-2.81781822e-01 -5.46938956e-01 5.01568019e-01 3.60442668e-01
4.88523394e-02 -5.88760264e-02 -4.24307048e-01 8.28733444e-01
3.83672953e-01 3.99188437e-02 -9.45491120e-02 7.69874215e-01
-7.84779668e-01 -1.01956570e+00 -5.87676585e-01 4.10283685e-01
6.98198229e-02 1.35423794e-01 -4.17698175e-01 5.60152531e-01
7.18013227e-01 7.54255056e-01 -1.24511592e-01 -2.21796736e-01
4.89961505e-01 3.23012061e-02 1.76863536e-01 -1.07966232e+00
-3.77110243e-01 1.81159943e-01 -3.35819513e-01 -4.49871689e-01
-6.40434861e-01 -4.87036139e-01 -1.37974882e+00 4.82378483e-01
-1.03669131e+00 1.88212037e-01 3.19562167e-01 9.60695863e-01
1.48804083e-01 4.25851732e-01 7.70720720e-01 -1.04458559e+00
-7.70589471e-01 -1.24081326e+00 -5.65889001e-01 5.19338489e-01
2.12758213e-01 -7.88965344e-01 -4.26508397e-01 1.06085040e-01]
|
[9.215463638305664, 1.297062635421753]
|
634706f6-fec9-4d80-a261-6c0c57f244b1
|
ontology-aware-network-for-zero-shot-sketch
|
2302.10040
| null |
https://arxiv.org/abs/2302.10040v1
|
https://arxiv.org/pdf/2302.10040v1.pdf
|
Ontology-aware Network for Zero-shot Sketch-based Image Retrieval
|
Zero-Shot Sketch-Based Image Retrieval (ZSSBIR) is an emerging task. The pioneering work focused on the modal gap but ignored inter-class information. Although recent work has begun to consider the triplet-based or contrast-based loss to mine inter-class information, positive and negative samples need to be carefully selected, or the model is prone to lose modality-specific information. To respond to these issues, an Ontology-Aware Network (OAN) is proposed. Specifically, the smooth inter-class independence learning mechanism is put forward to maintain inter-class peculiarity. Meanwhile, distillation-based consistency preservation is utilized to keep modality-specific information. Extensive experiments have demonstrated the superior performance of our algorithm on two challenging Sketchy and Tu-Berlin datasets.
|
['Deqiang Cheng', 'Ziqiang Wang', 'He Jiang', 'Haoxiang Zhang']
|
2023-02-20
| null | null | null | null |
['sketch-based-image-retrieval']
|
['computer-vision']
|
[ 2.71738052e-01 -4.26681459e-01 -7.03918815e-01 -4.97786313e-01
-8.85359108e-01 -1.05003670e-01 6.17331386e-01 -8.19132011e-03
-2.77078271e-01 6.42904758e-01 -4.52184305e-02 2.27362871e-01
-6.18295431e-01 -6.97219968e-01 -2.71597713e-01 -7.13032365e-01
9.45916101e-02 3.25493038e-01 3.07188720e-01 -1.76279023e-01
3.97524744e-01 3.12553704e-01 -1.53125262e+00 3.13647866e-01
9.08915102e-01 1.17928588e+00 6.39583766e-02 9.63762179e-02
-4.63452309e-01 6.41809762e-01 -1.96697593e-01 -4.23718184e-01
3.44475806e-01 -3.02732050e-01 -5.63308537e-01 1.87658280e-01
5.03306806e-01 -2.59024739e-01 -3.66449207e-01 1.37803829e+00
5.65034509e-01 2.19963223e-01 6.78897262e-01 -1.63392079e+00
-7.30360627e-01 1.49970233e-01 -7.88750589e-01 -2.14337148e-02
1.41034350e-01 1.68861833e-03 1.23594487e+00 -1.17454946e+00
9.51806486e-01 1.40595150e+00 2.47364879e-01 5.01867533e-01
-1.16810334e+00 -8.27133358e-01 2.45804027e-01 7.37890482e-01
-1.61181068e+00 -1.63731053e-01 1.41163218e+00 -8.90520290e-02
1.28781006e-01 2.87148625e-01 6.60484076e-01 1.17317080e+00
-1.79577813e-01 1.04988611e+00 9.77833688e-01 -5.36055744e-01
1.89151645e-01 3.86277199e-01 1.68082938e-02 5.71351051e-01
2.08102107e-01 -1.63459349e-02 -7.64020741e-01 -7.49290660e-02
8.84875238e-01 3.65667015e-01 -1.37304321e-01 -8.82319033e-01
-1.08388877e+00 7.28288114e-01 3.66422772e-01 4.20995653e-01
-8.70156959e-02 -2.37950206e-01 6.22071445e-01 4.00317758e-01
4.69022840e-01 7.24779889e-02 -1.69365719e-01 1.73248246e-01
-1.04749405e+00 1.29771128e-01 4.77984250e-01 1.02859056e+00
7.05089509e-01 -2.38642082e-01 -2.82595426e-01 1.27940381e+00
2.95796961e-01 1.16486199e-01 3.00986081e-01 -8.41506124e-01
5.10177851e-01 7.65553236e-01 -2.12397665e-01 -1.34706342e+00
-4.58056144e-02 -2.74531573e-01 -1.13706112e+00 9.09379348e-02
2.69573808e-01 4.70230669e-01 -9.82636273e-01 1.69754660e+00
3.51098925e-01 1.64975703e-01 -1.99915931e-01 1.21278381e+00
8.91356826e-01 4.14765328e-01 1.46765694e-01 -1.71857744e-01
1.25887001e+00 -7.66929507e-01 -8.84067476e-01 1.43200278e-01
5.76227605e-02 -7.51053095e-01 1.19724810e+00 2.29338765e-01
-9.98635352e-01 -4.19215947e-01 -1.28711152e+00 -4.97087352e-02
-4.88059044e-01 -1.70647442e-01 5.83110690e-01 4.03355956e-01
-5.52699566e-01 3.14782679e-01 -3.35200638e-01 -3.71528119e-01
6.96412504e-01 4.27918397e-02 -5.77555835e-01 -6.60571337e-01
-1.29743540e+00 6.13638520e-01 5.88757038e-01 7.25891218e-02
-5.53021312e-01 -7.61586487e-01 -6.51332200e-01 1.98572427e-02
7.55561709e-01 -5.57912052e-01 5.35860002e-01 -8.97818685e-01
-1.33914995e+00 7.08224118e-01 4.82463241e-02 7.39327818e-02
6.71638131e-01 2.84159064e-01 -6.95206881e-01 4.80194420e-01
-7.38700926e-02 6.39076829e-01 9.49334502e-01 -1.42605817e+00
-5.22730708e-01 -4.50272292e-01 1.02382086e-01 3.26180667e-01
-8.41925085e-01 -1.96600243e-01 -9.02894676e-01 -1.00250113e+00
6.06945217e-01 -7.80436277e-01 2.13883668e-02 6.27250075e-01
-2.26046026e-01 -2.54548669e-01 9.68207717e-01 -4.63833779e-01
1.21406078e+00 -2.02600384e+00 1.94695622e-01 4.46672648e-01
8.43270048e-02 2.17127249e-01 -4.97131646e-01 5.14751613e-01
7.24133328e-02 -6.26577213e-02 -4.46728975e-01 -1.79081634e-01
-3.80355604e-02 1.49181828e-01 -1.76917732e-01 3.08882177e-01
2.55243272e-01 7.19932318e-01 -9.19735491e-01 -1.02490520e+00
4.47681308e-01 6.25037134e-01 -4.58653569e-01 -8.24019983e-02
-8.75812620e-02 2.18377456e-01 -4.47982401e-01 1.19776464e+00
9.32339072e-01 -3.15014184e-01 1.45185649e-01 -4.08913374e-01
3.41532737e-01 -3.65433425e-01 -1.25100088e+00 2.09886456e+00
-2.99420685e-01 4.17962134e-01 -3.74557525e-02 -9.65208232e-01
9.07833338e-01 1.51246727e-01 6.57093048e-01 -1.00449049e+00
5.89555688e-02 1.87556878e-01 -2.75457412e-01 -4.97916847e-01
6.02558851e-01 -3.73281449e-01 1.01586729e-01 3.99390124e-02
2.46776063e-02 7.13735148e-02 3.60265113e-02 3.61851543e-01
5.52225590e-01 1.90154716e-01 1.32459596e-01 -2.70034224e-01
7.13716805e-01 -1.76860988e-01 7.63757408e-01 5.02036452e-01
-3.93971533e-01 8.62720490e-01 2.79504776e-01 -2.72509515e-01
-8.65086019e-01 -1.00311649e+00 -3.07561576e-01 9.56843615e-01
8.60354066e-01 -4.84654129e-01 -1.40808225e-01 -7.08236575e-01
-6.04166836e-02 4.21240330e-01 -6.26466751e-01 -3.60892832e-01
-2.74348050e-01 -5.93618214e-01 1.79151028e-01 2.17644170e-01
7.15382695e-01 -7.88530350e-01 -1.13871373e-01 8.75580907e-02
-3.49452525e-01 -7.57056832e-01 -4.22952801e-01 -3.72491896e-01
-7.97109425e-01 -1.24879313e+00 -1.10371983e+00 -7.39526927e-01
7.36550272e-01 5.52320659e-01 7.36606598e-01 1.99949324e-01
-4.72569436e-01 3.91528428e-01 -3.88415992e-01 -1.02235042e-01
2.13680729e-01 -6.09621368e-02 -2.51595646e-01 2.44655147e-01
3.13325852e-01 -6.53718114e-01 -7.44754732e-01 4.04991776e-01
-1.11366796e+00 -5.99179789e-02 5.35176039e-01 1.25074315e+00
7.32082188e-01 4.78246510e-02 6.82480514e-01 -6.98415041e-01
5.53216875e-01 -2.35114381e-01 -4.38295096e-01 7.88807631e-01
-8.93322647e-01 -5.36301807e-02 4.05641764e-01 -6.62071943e-01
-1.20166397e+00 -1.56392813e-01 2.94681340e-01 -8.93232822e-01
2.52914369e-01 5.25731266e-01 -4.37788963e-01 -1.51127219e-01
1.20672241e-01 2.93797582e-01 -2.12866012e-02 -3.62810224e-01
3.63235176e-01 4.58628625e-01 4.61099297e-01 -7.81358659e-01
7.85776913e-01 6.99267983e-01 1.98417544e-01 -7.71756172e-01
-5.71916878e-01 -5.90240598e-01 -4.07165676e-01 -3.81003112e-01
4.95387822e-01 -8.47439170e-01 -4.98520195e-01 3.79504174e-01
-8.76586914e-01 2.56853133e-01 -1.47371545e-01 3.34810853e-01
-2.82360643e-01 7.27565765e-01 -5.72070241e-01 -1.00888419e+00
-4.09253061e-01 -8.82067323e-01 8.52924824e-01 3.50541204e-01
2.13485837e-01 -7.70889163e-01 1.30934507e-01 3.40955436e-01
3.59036297e-01 6.63508028e-02 9.38882113e-01 -4.86199051e-01
-6.94025576e-01 -4.36775118e-01 -6.93073511e-01 2.05755770e-01
7.54416138e-02 5.63976243e-02 -7.88812339e-01 -2.45538622e-01
-3.24044138e-01 -4.52955753e-01 1.07675421e+00 3.94243635e-02
1.43862343e+00 -1.87442303e-02 -1.97559953e-01 3.69341403e-01
1.49820423e+00 -9.14543215e-03 8.00857782e-01 2.46963307e-01
6.15018964e-01 6.41326070e-01 1.00018418e+00 4.31269318e-01
4.63127136e-01 7.81866372e-01 3.66355777e-01 5.10790832e-02
-1.59234479e-01 -4.28804994e-01 -1.69399574e-01 7.23207831e-01
6.42693117e-02 -1.44124880e-01 -3.89218301e-01 5.41516066e-01
-2.05678105e+00 -1.12561798e+00 1.64717913e-01 2.34454155e+00
7.37224579e-01 3.04195154e-02 -4.65351418e-02 1.91126525e-01
9.01383698e-01 4.61141855e-01 -4.94285524e-01 2.47019932e-01
-4.48489338e-01 -1.51705518e-01 1.10128939e-01 1.97937995e-01
-1.04988396e+00 7.60902524e-01 5.38092375e+00 1.39710522e+00
-1.09287560e+00 7.09209591e-02 4.74383563e-01 -7.01596960e-02
-5.58924377e-01 1.61077291e-01 -3.10399890e-01 5.65993965e-01
-4.54801507e-02 -1.55955106e-01 3.64295840e-01 7.21373498e-01
-2.19505936e-01 -1.60804391e-01 -7.09535718e-01 1.38771105e+00
2.92075425e-01 -1.20579886e+00 2.55832106e-01 -7.39681125e-02
6.47120476e-01 -4.39796388e-01 1.03836842e-01 2.92288870e-01
-2.21682876e-01 -4.87348169e-01 5.94416082e-01 8.22173119e-01
9.77597237e-01 -9.43022907e-01 4.29975688e-01 1.57672241e-01
-1.38880336e+00 1.70545597e-02 -5.48315585e-01 2.34491572e-01
6.32357299e-02 6.11694992e-01 -2.23177478e-01 9.45398688e-01
6.07811630e-01 9.01053369e-01 -5.30170739e-01 1.19652545e+00
-1.45470705e-02 1.97998300e-01 -2.76807517e-01 1.38814405e-01
8.13395754e-02 -4.39938575e-01 6.67217612e-01 7.97473192e-01
8.58086050e-02 8.39902982e-02 1.18519299e-01 6.64122462e-01
-1.09181024e-01 1.25850976e-01 -4.20057565e-01 2.66491938e-02
5.22999763e-01 1.47911501e+00 -7.63450027e-01 -3.80932838e-01
-2.69465476e-01 1.05202448e+00 1.58716306e-01 3.20380628e-01
-7.28786349e-01 -4.69411314e-01 5.30670643e-01 -1.01497509e-01
2.71864206e-01 4.09777723e-02 -4.29055691e-01 -1.50555336e+00
2.74550349e-01 -4.99320805e-01 8.09505820e-01 -6.67891622e-01
-1.76954234e+00 2.14354888e-01 -4.81602037e-03 -1.73051822e+00
2.29631081e-01 -1.95182905e-01 -5.84018171e-01 4.71541673e-01
-1.93754160e+00 -1.20777559e+00 -2.90634394e-01 7.87235975e-01
5.90497017e-01 -2.01244161e-01 5.58029294e-01 7.95811772e-01
-7.51441419e-01 9.00374115e-01 2.51071266e-04 4.25367132e-02
9.87031400e-01 -7.80156493e-01 -5.31364024e-01 6.28769040e-01
9.23780799e-02 6.23075008e-01 4.77842689e-01 -6.53854311e-01
-1.43604529e+00 -9.16519642e-01 7.83012748e-01 -1.05486400e-02
5.28998673e-01 -2.00269043e-01 -1.16246319e+00 1.34532869e-01
-5.54209650e-02 4.43067968e-01 5.60151577e-01 1.50350869e-01
-7.75148332e-01 -4.69568491e-01 -1.21925819e+00 7.22498178e-01
1.21427321e+00 -6.71258092e-01 -4.26977009e-01 1.55753180e-01
3.41177344e-01 -1.24362692e-01 -8.47635508e-01 6.17873907e-01
9.21223879e-01 -8.46282005e-01 1.18980002e+00 -6.00697339e-01
5.80377102e-01 -3.74433935e-01 -2.50949204e-01 -7.91593730e-01
-1.82888061e-01 -1.25557691e-01 -7.69891441e-02 1.42862606e+00
9.94503424e-02 -3.53170246e-01 6.72987282e-01 7.56367683e-01
2.42231429e-01 -6.44767225e-01 -1.04444933e+00 -9.55951512e-01
-1.77877024e-01 -2.99984306e-01 4.77062285e-01 1.11680555e+00
-9.15926043e-03 1.85895041e-01 -8.43600273e-01 -1.45256400e-01
8.09425235e-01 3.60547096e-01 5.66651762e-01 -1.37699080e+00
5.96599691e-02 -4.58278239e-01 -4.47356611e-01 -5.72635412e-01
2.76958123e-02 -8.68903399e-01 -7.84355476e-02 -1.45248353e+00
5.36431193e-01 -6.82819784e-01 -9.58627403e-01 3.53646457e-01
-2.15617582e-01 4.08453703e-01 3.84695679e-01 4.02926981e-01
-9.95847702e-01 1.00721312e+00 1.32357538e+00 -4.10990924e-01
-3.94442454e-02 -3.16641241e-01 -4.40361172e-01 3.90372753e-01
5.54131687e-01 -4.02246773e-01 -6.07637823e-01 -5.89758866e-02
1.40481219e-01 1.55615970e-01 6.05602562e-01 -7.67364562e-01
4.47698444e-01 -4.17029291e-01 3.36823136e-01 -7.93755174e-01
6.24609709e-01 -1.12251925e+00 3.96768786e-02 1.73259169e-01
-4.83831406e-01 -3.69960755e-01 -2.03593418e-01 9.85510886e-01
-4.70264882e-01 -1.61744989e-02 7.53849447e-01 2.16253754e-02
-9.98516321e-01 7.28588760e-01 2.68220007e-01 -4.47476357e-02
8.51553857e-01 -3.12962264e-01 -1.99400470e-01 -2.05014974e-01
-2.88811356e-01 3.08085531e-01 5.40190399e-01 6.75613761e-01
8.69240820e-01 -1.78236282e+00 -4.98965293e-01 1.69495136e-01
7.67192125e-01 -3.24631661e-01 8.16108108e-01 9.31112587e-01
-1.53001389e-02 1.34627238e-01 -3.06245744e-01 -5.59885740e-01
-1.24612451e+00 7.86549926e-01 -5.63008636e-02 -2.66545892e-01
-6.98687494e-01 6.70337558e-01 2.50068884e-02 -3.05296689e-01
3.03806573e-01 3.82870376e-01 -9.17396098e-02 2.85527080e-01
5.49017191e-01 3.95302445e-01 -1.34805590e-01 -3.62865746e-01
-5.79784453e-01 6.65394425e-01 -5.46437092e-02 -2.04084143e-01
1.23006690e+00 -1.37570933e-01 -1.18941970e-01 5.08962452e-01
1.22264409e+00 -3.87340575e-01 -1.15079606e+00 -5.47679663e-01
6.56718239e-02 -8.75029147e-01 1.20322458e-01 -7.22158730e-01
-1.22312319e+00 8.19459558e-01 6.22387946e-01 -1.38811812e-01
1.25705922e+00 -3.46913934e-01 8.50383818e-01 3.88421118e-01
4.19221878e-01 -1.39983308e+00 2.70215869e-01 1.33850306e-01
8.32606792e-01 -1.55313408e+00 1.73179552e-01 -5.61645329e-01
-5.42461812e-01 1.07937169e+00 5.74425459e-01 -2.13464033e-02
7.74577618e-01 -1.64195284e-01 -3.12989622e-01 -7.20569566e-02
-5.85075319e-01 -1.66772440e-01 5.76085985e-01 4.31865573e-01
-6.24463186e-02 -1.24113768e-01 -5.78378260e-01 5.11512876e-01
6.33617163e-01 2.44608775e-01 -7.29283467e-02 1.09999931e+00
-1.36095107e-01 -1.23360384e+00 -2.19975770e-01 4.15376484e-01
-9.29970015e-03 1.36302076e-02 -2.98488975e-01 8.49455237e-01
-5.14804907e-02 7.35577822e-01 -2.08010301e-01 -3.78904879e-01
1.19569011e-01 3.48656103e-02 6.54913366e-01 -1.13930300e-01
-1.82135940e-01 1.10068463e-01 -2.46234149e-01 -4.96383280e-01
-6.97915435e-01 -3.80971909e-01 -9.28236663e-01 -1.68083698e-01
-4.99458998e-01 -1.67051852e-01 4.93946880e-01 7.50448525e-01
3.68219018e-01 2.11107120e-01 7.22057402e-01 -6.14278376e-01
-3.13293993e-01 -5.35300076e-01 -8.01113486e-01 4.43765998e-01
4.66884896e-02 -9.47612524e-01 -3.20539117e-01 -3.07432353e-01]
|
[11.576814651489258, 0.7023018002510071]
|
86610d76-055a-46d7-a454-72adf8fa2e8d
|
image-clustering-without-ground-truth
|
1610.07758
| null |
http://arxiv.org/abs/1610.07758v1
|
http://arxiv.org/pdf/1610.07758v1.pdf
|
Image Clustering without Ground Truth
|
Cluster analysis has become one of the most exercised research areas over the
past few decades in computer science. As a consequence, numerous clustering
algorithms have already been developed to find appropriate partitions of a set
of objects. Given multiple such clustering solutions, it is a challenging task
to obtain an ensemble of these solutions. This becomes more challenging when
the ground truth about the number of clusters is unavailable. In this paper, we
introduce a crowd-powered model to collect solutions of image clustering from
the general crowd and pose it as a clustering ensemble problem with variable
number of clusters. The varying number of clusters basically reflects the crowd
workers' perspective toward a particular set of objects. We allow a set of
crowd workers to independently cluster the images as per their perceptions. We
address the problem by finding out centroid of the clusters using an
appropriate distance measure and prioritize the likelihood of similarity of the
individual cluster sets. The effectiveness of the proposed method is
demonstrated by applying it on multiple artificial datasets obtained from
crowd.
|
['Tripti Prasad', 'Abhisek Dash', 'Sujoy Chatterjee', 'Malay Bhattacharyya']
|
2016-10-25
| null | null | null | null |
['clustering-ensemble']
|
['graphs']
|
[-1.56944469e-02 -1.94571003e-01 4.88209426e-01 -1.99452311e-01
-3.20775330e-01 -6.94504380e-01 6.44974470e-01 4.03871983e-01
-5.22013068e-01 3.61325771e-01 5.82474619e-02 6.94862455e-02
-3.99357021e-01 -5.89671195e-01 -2.38885865e-01 -1.01851034e+00
3.85549143e-02 9.23413217e-01 4.07701850e-01 1.90114379e-02
6.82402611e-01 3.78974527e-01 -2.03720427e+00 1.41445830e-01
9.87149179e-01 5.60711384e-01 5.31282544e-01 7.14116633e-01
-1.52539432e-01 3.57678413e-01 -9.65663552e-01 -8.84308442e-02
5.36224008e-01 -3.68956834e-01 -7.60187447e-01 8.88455808e-01
-1.34484768e-01 4.41356033e-01 5.37526369e-01 1.21736658e+00
4.48685706e-01 3.99954736e-01 8.62276316e-01 -1.36663961e+00
-2.49130130e-01 4.22516227e-01 -8.69560540e-01 3.16790909e-01
2.39457607e-01 1.18115336e-01 6.98188245e-01 -5.96733689e-01
4.27121669e-01 1.27531290e+00 1.61049604e-01 3.08209062e-01
-1.06206238e+00 -2.96970218e-01 1.50788099e-01 2.42323324e-01
-1.77613604e+00 -2.62639850e-01 8.00199091e-01 -8.34611058e-01
3.28344285e-01 3.11539888e-01 4.59123462e-01 2.39938557e-01
-7.14522153e-02 3.28110963e-01 1.26863217e+00 -4.94380772e-01
7.00722694e-01 5.87400734e-01 7.85251036e-02 3.83350141e-02
4.77461725e-01 -6.61410391e-01 -9.39824581e-02 -2.96963304e-01
2.07372338e-01 1.62503406e-01 -1.43016785e-01 -5.19367814e-01
-1.02322495e+00 6.47861540e-01 3.15371305e-01 5.08966625e-01
-5.29235780e-01 -2.68102795e-01 -1.01846894e-02 -1.99773863e-01
3.63583356e-01 3.37137043e-01 4.73927259e-02 3.11693609e-01
-9.58120167e-01 3.70828032e-01 6.69827700e-01 7.84646034e-01
1.02838230e+00 -5.07518709e-01 1.17900677e-01 4.97113317e-01
3.29569668e-01 1.76233351e-01 4.13501173e-01 -1.04180419e+00
3.58321756e-01 9.66289043e-01 4.60894495e-01 -1.63593543e+00
-1.01050049e-01 -2.11201221e-01 -8.75641942e-01 2.25693673e-01
4.05001700e-01 -3.30561846e-01 -5.85749567e-01 1.32415199e+00
7.76335001e-01 1.42170548e-01 2.09657568e-02 9.19478297e-01
2.74908394e-01 5.98277211e-01 -1.55333042e-01 -5.09490132e-01
1.15467501e+00 -4.34685022e-01 -7.03855097e-01 1.88089013e-02
1.62835285e-01 -7.74352074e-01 4.33651149e-01 4.27689761e-01
-8.64625156e-01 -6.70870364e-01 -7.73165226e-01 6.51297808e-01
-3.53316873e-01 -1.96397547e-02 1.74400583e-01 6.21628642e-01
-1.28878021e+00 1.45748943e-01 -5.15863359e-01 -6.60093725e-01
6.71292022e-02 6.86206341e-01 -1.42178740e-02 -3.22796255e-02
-5.51589549e-01 5.93480647e-01 6.20028794e-01 2.42386922e-01
-5.40555716e-01 1.23991244e-01 -2.32243821e-01 -2.81765938e-01
4.90609258e-01 -4.94770199e-01 7.96456397e-01 -1.19676828e+00
-8.41886580e-01 8.97887886e-01 -4.04405951e-01 -5.70652112e-02
5.41835666e-01 2.85817951e-01 -2.91414738e-01 9.15692076e-02
4.76198524e-01 4.68843699e-01 7.10088551e-01 -1.93402004e+00
-1.04957747e+00 -6.44291759e-01 -3.09361607e-01 3.93617570e-01
-1.44707099e-01 1.80158749e-01 -5.24469495e-01 -8.28613117e-02
2.30437264e-01 -1.06532323e+00 -6.42329156e-01 -4.97553945e-01
-5.97618401e-01 -5.02873302e-01 7.37777174e-01 -1.43716797e-01
1.22015083e+00 -2.08682227e+00 3.02700013e-01 6.19470835e-01
4.96192306e-01 2.56564260e-01 2.76054949e-01 4.87233818e-01
1.92110807e-01 1.26688048e-01 -3.76114368e-01 -3.63194555e-01
-2.68558800e-01 2.05951959e-01 1.27486736e-01 5.80298305e-01
5.01573943e-02 6.77644014e-02 -9.11652923e-01 -9.80018079e-01
2.23916829e-01 2.00725511e-01 -4.89869237e-01 4.31712598e-01
-4.08174135e-02 7.32658923e-01 -6.65465355e-01 2.36589447e-01
8.75360489e-01 -2.48911142e-01 4.40734625e-01 1.33652121e-01
-2.52022624e-01 -6.97117209e-01 -1.87707329e+00 1.02776742e+00
2.93061793e-01 2.40495041e-01 2.49431223e-01 -9.76648211e-01
9.10057247e-01 3.98241132e-01 6.69919431e-01 7.72781968e-02
4.05446589e-01 1.52003184e-01 2.55232990e-01 -7.60664225e-01
6.76603377e-01 -2.56351471e-01 -2.47474462e-02 7.32543409e-01
-3.61409307e-01 1.79652289e-01 4.76147622e-01 4.28693853e-02
8.79779220e-01 -4.73617405e-01 5.54986000e-01 -3.55076253e-01
6.65900230e-01 3.06946278e-01 4.23307091e-01 7.35831439e-01
-4.31453735e-01 7.39330590e-01 -1.82723941e-03 -3.86918187e-01
-8.64957690e-01 -8.71918023e-01 1.38101846e-01 7.27520108e-01
5.01753330e-01 7.44329393e-02 -1.04248047e+00 -2.69291729e-01
-3.01043749e-01 2.35085532e-01 -6.14983857e-01 3.19369465e-01
-5.52133024e-01 -9.30305719e-01 6.31758943e-02 -1.07296780e-01
5.43169022e-01 -1.08440781e+00 -8.50771666e-01 2.85943002e-01
-4.62904125e-01 -8.31154227e-01 -1.32089481e-01 1.54360905e-02
-4.94592816e-01 -1.23434484e+00 -5.89109659e-01 -8.63706827e-01
1.07217205e+00 8.69834304e-01 9.04383183e-01 5.36354125e-01
-1.53295174e-01 3.85256290e-01 -6.58167124e-01 -6.58662140e-01
-3.77158314e-01 3.10043320e-02 2.96322614e-01 7.28031814e-01
6.03117108e-01 -4.15684015e-01 -5.26825547e-01 5.10667562e-01
-1.12750161e+00 -3.60628724e-01 2.50727415e-01 9.46622416e-02
4.00503963e-01 9.61481512e-01 3.71218890e-01 -6.73775554e-01
1.02806115e+00 -8.85351419e-01 -2.86670804e-01 3.54877830e-01
-1.36765003e-01 -1.24321379e-01 5.52349329e-01 -4.06258523e-01
-1.09319627e+00 2.43807450e-01 6.56256080e-01 -2.64319271e-01
-6.72302544e-01 2.40497351e-01 -3.43404442e-01 2.80219287e-01
6.32822514e-01 3.81547064e-02 4.93045188e-02 -2.53997445e-01
3.09715927e-01 9.73615110e-01 5.65415561e-01 -5.75246871e-01
8.80614161e-01 7.81645596e-01 -1.96949333e-01 -1.00024295e+00
-5.91224611e-01 -1.17893982e+00 -1.09839129e+00 -7.27963984e-01
1.11627042e+00 -6.30988896e-01 -8.24871004e-01 1.72738746e-01
-1.11125195e+00 2.85831898e-01 1.19817384e-01 2.09333822e-01
-2.18955785e-01 5.20404100e-01 2.14928523e-01 -1.30228496e+00
1.57199372e-02 -1.25704885e+00 8.35132241e-01 3.60218823e-01
-4.57101613e-01 -8.92140865e-01 1.09720506e-01 4.46474016e-01
5.40203601e-02 5.79403877e-01 5.28901637e-01 -7.01422513e-01
-6.44806266e-01 -1.95167482e-01 7.42768943e-02 -9.29586068e-02
4.60979968e-01 9.99397188e-02 -8.44060719e-01 -2.99083769e-01
3.53650481e-01 2.20081910e-01 5.02918959e-01 5.11154354e-01
6.58481538e-01 -2.78743893e-01 -6.57895148e-01 8.95472988e-02
1.29615295e+00 4.99450952e-01 2.75784224e-01 3.66128415e-01
6.69390976e-01 1.14162302e+00 6.97198212e-01 6.46140873e-01
4.04806346e-01 3.59829426e-01 4.10122961e-01 -3.54181393e-03
4.12292629e-01 3.08792919e-01 -1.07485309e-01 7.10768819e-01
-3.59615833e-01 -6.15646183e-01 -1.10390449e+00 7.05701590e-01
-1.99328411e+00 -1.11351395e+00 -2.57693678e-01 2.12355065e+00
2.74789393e-01 -1.96680859e-01 5.44610023e-01 6.47022069e-01
1.55310643e+00 -2.53563106e-01 -3.03346723e-01 -8.34712572e-03
-2.36639660e-02 -3.49536985e-01 9.18038115e-02 4.92723107e-01
-1.02529025e+00 6.87799454e-01 5.58076859e+00 5.73853314e-01
-7.44026959e-01 -1.83457926e-01 7.74115682e-01 1.45438641e-01
3.05078626e-02 2.19899133e-01 -7.64477372e-01 7.06887126e-01
3.49137962e-01 -1.80645436e-01 4.23795730e-01 3.35323900e-01
3.99279058e-01 -6.47311687e-01 -9.26579654e-01 8.51503134e-01
1.61689550e-01 -7.45195270e-01 -1.56089766e-02 4.78771985e-01
9.85498846e-01 -3.98924768e-01 8.75137672e-02 -3.48868489e-01
6.29629195e-01 -1.06009400e+00 6.18528306e-01 5.65883458e-01
8.22669938e-02 -8.66946578e-01 8.78767371e-01 9.84712899e-01
-1.18124020e+00 -3.59455138e-01 -3.37573498e-01 -4.29023504e-01
2.21778750e-01 5.92615843e-01 -1.20332098e+00 4.27509457e-01
7.94677496e-01 8.78234282e-02 -7.01686502e-01 1.26247764e+00
3.43000710e-01 3.44098240e-01 -4.23072428e-01 -2.38697782e-01
2.17294425e-01 -4.99824554e-01 5.48029244e-01 8.97568822e-01
4.36022490e-01 3.83839399e-01 5.78442633e-01 7.12470055e-01
4.36476499e-01 1.55178189e-01 -6.45150006e-01 4.80490476e-01
8.46585035e-01 1.34208608e+00 -1.50307274e+00 -2.46034637e-01
1.65545449e-01 6.60783529e-01 3.70985925e-01 2.74538606e-01
-4.20761079e-01 5.54396249e-02 2.62782902e-01 2.27509201e-01
1.27538398e-01 -2.05129072e-01 -1.03565097e-01 -6.37555063e-01
1.62010770e-02 -8.46214235e-01 4.38585609e-01 -5.37408173e-01
-1.25536752e+00 7.09944904e-01 2.39083648e-01 -1.37751663e+00
-4.83729243e-02 -1.51524663e-01 -7.16897964e-01 6.68876946e-01
-8.51798058e-01 -6.95785820e-01 -5.06250560e-01 5.49575984e-01
3.93806934e-01 -2.25541353e-01 3.36457729e-01 -1.24449894e-01
-5.44264197e-01 -1.97320998e-01 1.11738667e-01 1.20168805e-01
4.76894796e-01 -1.16280532e+00 -2.88091511e-01 1.05077326e+00
-5.72725274e-02 7.84962356e-01 1.05451107e+00 -7.35089719e-01
-7.73157120e-01 -1.00451922e+00 8.92471015e-01 -6.06580317e-01
1.93768322e-01 -1.28724456e-01 -7.12925971e-01 2.69109637e-01
4.51914996e-01 -2.33920470e-01 7.82551229e-01 -2.87833631e-01
4.33323830e-01 5.79731502e-02 -1.28166163e+00 5.07419169e-01
6.21797204e-01 1.14830367e-01 -5.46128452e-01 1.80780470e-01
1.48999974e-01 1.38286337e-01 -4.94437605e-01 -5.51298186e-02
-1.10909261e-01 -1.40959573e+00 6.60897136e-01 -1.77271172e-01
4.39037271e-02 -1.07886195e+00 -2.46907592e-01 -1.37869334e+00
-3.66409808e-01 -5.14177263e-01 5.74621081e-01 1.43985188e+00
2.87593286e-02 -3.88770908e-01 6.91869736e-01 5.70968688e-01
3.31800222e-01 -1.92563504e-01 -7.71240711e-01 -4.09630567e-01
-2.76199818e-01 -3.70391496e-02 6.07813418e-01 1.04436588e+00
1.50991930e-02 3.06029230e-01 -9.27197337e-02 5.42559028e-01
9.38951731e-01 2.12210611e-01 1.08732903e+00 -1.65328348e+00
-8.45269188e-02 -3.83556455e-01 -4.49923992e-01 -5.65508604e-01
2.51449496e-02 -6.32264018e-01 3.69403839e-01 -1.52376854e+00
4.86621618e-01 -6.42917454e-01 1.43136099e-01 -1.04377478e-01
-5.15476525e-01 1.03069432e-02 3.70768458e-01 5.91311574e-01
-1.02019489e+00 1.12239480e-01 1.06189632e+00 1.03647210e-01
-3.53051543e-01 1.71029717e-01 -7.94013262e-01 8.34993124e-01
9.41981912e-01 -5.95750570e-01 -4.88438815e-01 -2.15748772e-01
1.00056166e-02 -7.28281680e-03 1.19041540e-01 -1.38348949e+00
7.39549756e-01 -3.15142453e-01 2.97701031e-01 -6.37241185e-01
1.69890478e-01 -1.03598130e+00 6.45103276e-01 2.02697873e-01
-1.92528635e-01 3.09887826e-01 -3.32668275e-01 8.45896780e-01
-2.00623915e-01 -2.21208975e-01 6.82181060e-01 -6.41059279e-01
-6.22413933e-01 2.86277756e-02 -5.39660335e-01 -7.15429150e-03
1.64367008e+00 -7.37426162e-01 1.68023318e-01 -3.09408665e-01
-7.48304605e-01 4.01904821e-01 7.12870359e-01 2.31604546e-01
4.05232280e-01 -9.78600085e-01 -6.53879642e-01 -8.71811807e-02
5.34995310e-02 3.78188848e-01 2.30313808e-01 6.05713546e-01
-4.92338598e-01 1.36300623e-01 -1.71531260e-01 -9.01548982e-01
-1.62272108e+00 7.84451902e-01 8.31335932e-02 -7.04881549e-02
-7.52096549e-02 5.04034698e-01 1.74752727e-01 -2.82029957e-01
1.11955613e-01 1.60700306e-01 -6.80920243e-01 1.90203965e-01
4.48860466e-01 7.17376530e-01 -2.94970185e-01 -1.14523315e+00
-3.03488791e-01 7.74033189e-01 2.07705289e-01 -2.77575940e-01
1.31550717e+00 -5.10876775e-01 -4.49733466e-01 5.56363821e-01
8.60900640e-01 -8.48518014e-02 -9.96428490e-01 -1.43445730e-01
2.87409276e-01 -5.05094588e-01 -5.24440169e-01 -1.56130940e-01
-8.20089579e-01 5.35433412e-01 4.34110969e-01 6.77564085e-01
1.10199249e+00 1.77354947e-01 1.23814456e-01 1.58194780e-01
5.38510501e-01 -1.17268085e+00 2.13644966e-01 4.46849689e-02
4.45438027e-01 -1.38384235e+00 1.43217640e-02 -4.42033172e-01
-8.04847360e-01 6.41173422e-01 4.62488681e-01 -2.45296136e-01
7.55579293e-01 5.71065955e-02 3.49701077e-01 -4.29509163e-01
-3.58365387e-01 -4.59776103e-01 -2.77389660e-02 8.85345280e-01
2.53798813e-01 4.25869912e-01 -2.81238735e-01 1.44219071e-01
-3.48403454e-01 -3.09482068e-01 7.32040048e-01 8.95372331e-01
-9.10121322e-01 -8.61862361e-01 -1.21827900e+00 2.40873590e-01
-4.76948351e-01 4.82095778e-01 -6.38171673e-01 5.54444432e-01
6.61394238e-01 1.75396478e+00 1.00616612e-01 -2.82495797e-01
6.96002916e-02 -2.40380839e-01 7.45375603e-02 -6.67068481e-01
-6.54390872e-01 1.61430761e-01 -3.29667538e-01 1.09679557e-01
-1.08236396e+00 -8.09362292e-01 -1.21128309e+00 -2.51814961e-01
-3.67605656e-01 6.53201997e-01 3.64597827e-01 9.89148080e-01
2.13473290e-01 2.01333761e-02 9.47270274e-01 -1.23387003e+00
-1.42619208e-01 -7.68938839e-01 -6.96235120e-01 7.77757287e-01
1.60232365e-01 -6.92005455e-01 -4.49680060e-01 3.88402462e-01]
|
[7.644064426422119, 4.599337100982666]
|
fe898684-6990-428a-b050-9fd01924c1fe
|
metaue-model-based-meta-learning-for
|
2303.06543
| null |
https://arxiv.org/abs/2303.06543v1
|
https://arxiv.org/pdf/2303.06543v1.pdf
|
MetaUE: Model-based Meta-learning for Underwater Image Enhancement
|
The challenges in recovering underwater images are the presence of diverse degradation factors and the lack of ground truth images. Although synthetic underwater image pairs can be used to overcome the problem of inadequately observing data, it may result in over-fitting and enhancement degradation. This paper proposes a model-based deep learning method for restoring clean images under various underwater scenarios, which exhibits good interpretability and generalization ability. More specifically, we build up a multi-variable convolutional neural network model to estimate the clean image, background light and transmission map, respectively. An efficient loss function is also designed to closely integrate the variables based on the underwater image model. The meta-learning strategy is used to obtain a pre-trained model on the synthetic underwater dataset, which contains different types of degradation to cover the various underwater environments. The pre-trained model is then fine-tuned on real underwater datasets to obtain a reliable underwater image enhancement model, called MetaUE. Numerical experiments demonstrate that the pre-trained model has good generalization ability, allowing it to remove the color degradation for various underwater attenuation images such as blue, green and yellow, etc. The fine-tuning makes the model able to adapt to different underwater datasets, the enhancement results of which outperform the state-of-the-art underwater image restoration methods. All our codes and data are available at \url{https://github.com/Duanlab123/MetaUE}.
|
['Yuping Duan', 'Ke Tang', 'Haorui Yan', 'Zhenwei Zhang']
|
2023-03-12
| null | null | null | null |
['underwater-image-restoration', 'image-enhancement']
|
['computer-vision', 'computer-vision']
|
[ 1.27028301e-01 -3.68317902e-01 7.82389343e-01 -4.59362060e-01
-7.46928096e-01 -1.53826252e-01 -7.67039955e-02 -3.11491549e-01
-4.23946857e-01 7.26070821e-01 1.36711285e-01 -3.57294045e-02
-1.02825642e-01 -8.93326700e-01 -8.49625230e-01 -1.23500848e+00
-1.16339810e-01 -4.11172211e-01 9.37961116e-02 -5.18926561e-01
1.45856932e-01 1.20222710e-01 -1.61070967e+00 -3.79245058e-02
1.32478261e+00 9.07984495e-01 8.02098274e-01 6.86822951e-01
3.75617519e-02 5.52338362e-01 -5.63946545e-01 -2.04369307e-01
4.51591462e-01 -4.20340747e-01 -1.20972440e-01 1.00108504e-01
2.00626135e-01 -7.58295596e-01 -7.23490655e-01 1.48400044e+00
9.48364496e-01 3.20654631e-01 3.22511524e-01 -7.79799104e-01
-8.20273876e-01 3.24779570e-01 -4.89806175e-01 1.93793237e-01
-1.48969963e-01 1.51217997e-01 4.46358711e-01 -7.73921549e-01
6.64739609e-02 1.27464890e+00 7.22101092e-01 5.37760854e-01
-6.65170729e-01 -8.21604311e-01 -2.33261790e-02 4.56053615e-01
-1.11154246e+00 -4.14012671e-01 6.34358168e-01 -1.50098398e-01
1.33155391e-01 1.10976458e-01 5.67371845e-01 7.08990037e-01
3.42540085e-01 5.15313864e-01 1.40834963e+00 -2.16635153e-01
-2.50438950e-03 -1.52734682e-01 -8.47987756e-02 6.21840715e-01
3.04951370e-01 1.30573586e-01 -2.57747144e-01 2.80033588e-01
7.86832631e-01 2.13306442e-01 -8.92733157e-01 2.23238111e-01
-6.72388375e-01 4.90580589e-01 6.24373257e-01 -1.42756343e-01
-2.16074392e-01 7.05938367e-03 2.33431026e-01 5.32534838e-01
4.00372654e-01 1.30979776e-01 -5.55329740e-01 2.06624597e-01
-4.63805616e-01 3.37300040e-02 5.35701036e-01 7.58950591e-01
1.15811968e+00 4.81956244e-01 1.47273645e-01 1.26645613e+00
6.00245774e-01 8.42017472e-01 5.90060055e-01 -1.09211731e+00
5.00025570e-01 1.69139102e-01 1.48255199e-01 -8.03410649e-01
-4.09152716e-01 -4.66308594e-01 -1.09118748e+00 5.26958287e-01
3.22797671e-02 -4.25337583e-01 -1.18546855e+00 1.62826872e+00
2.00992912e-01 2.78254479e-01 5.84951341e-01 1.05402780e+00
1.13591862e+00 1.07321775e+00 -2.58428186e-01 -3.45670938e-01
9.54397619e-01 -9.61101472e-01 -1.00575304e+00 -3.36006075e-01
1.82104737e-01 -5.74415624e-01 1.01269948e+00 4.22613323e-01
-9.48981404e-01 -6.48607790e-01 -1.39217007e+00 -2.86401417e-02
-7.82032609e-02 3.13385599e-03 2.71688879e-01 3.82014513e-01
-1.04244053e+00 6.77879632e-01 -7.58053124e-01 1.90358926e-02
2.99250841e-01 1.20675087e-01 -3.81549478e-01 -6.17110610e-01
-1.38175428e+00 7.81303763e-01 3.74893636e-01 7.24560320e-01
-1.37845051e+00 -4.59784746e-01 -9.93968666e-01 -6.61430955e-02
3.02269924e-02 -4.09834206e-01 9.57186937e-01 -9.54430223e-01
-1.39829814e+00 2.22655505e-01 1.10106111e-01 1.16350129e-01
5.30349016e-01 -2.05873922e-01 -5.67115426e-01 2.07550600e-01
1.42959664e-02 1.07655466e-01 8.77629638e-01 -1.81648517e+00
-5.99588931e-01 -2.89014816e-01 1.44673496e-01 4.32194471e-01
-4.14905548e-01 -2.03410178e-01 -6.96736395e-01 -5.37786901e-01
3.81801933e-01 -5.75591922e-01 -2.43505076e-01 2.24172458e-01
-2.05088094e-01 4.91903156e-01 7.88352191e-01 -1.12867773e+00
8.25371802e-01 -2.18656516e+00 1.50999606e-01 -1.45441353e-01
-1.16630234e-01 2.42354125e-01 -4.26777393e-01 4.20830429e-01
-1.11833038e-02 2.56653372e-02 -6.03983045e-01 -4.59871829e-01
-2.63408333e-01 6.70898020e-01 1.29380189e-02 7.47010589e-01
-1.09081134e-01 2.56724238e-01 -9.27929401e-01 -3.93311799e-01
2.72657663e-01 6.82999015e-01 -1.72182173e-01 5.73933005e-01
1.46293953e-01 6.54905319e-01 -3.34067613e-01 7.40775168e-01
1.26414001e+00 2.20845208e-01 -1.09683372e-01 -5.09258866e-01
-2.23152608e-01 -3.81786793e-01 -1.30733919e+00 1.44327426e+00
-6.18207753e-01 6.30138874e-01 5.33651233e-01 -8.70933831e-01
1.18034613e+00 1.49425149e-01 1.33221269e-01 -1.01632130e+00
2.34077826e-01 3.61518234e-01 -1.80133998e-01 -1.21077859e+00
3.19234550e-01 -4.56047356e-01 3.92153591e-01 9.11964849e-03
-2.20369428e-01 -1.25368135e-02 7.25624785e-02 -1.04007706e-01
6.26428604e-01 2.08309025e-01 -2.84398526e-01 -3.12391758e-01
5.48387229e-01 -4.76566762e-01 8.92778456e-01 6.55685067e-01
-9.72985700e-02 9.11222339e-01 -2.90382169e-02 -3.44252020e-01
-1.04007435e+00 -8.75982285e-01 -3.44330341e-01 7.88846970e-01
9.58539367e-01 3.60697597e-01 -5.54944217e-01 2.85202619e-02
-3.45965862e-01 3.56335700e-01 -5.80081820e-01 -3.57511610e-01
-4.56465304e-01 -1.25396025e+00 3.11050206e-01 3.56643796e-01
1.01277411e+00 -1.00239122e+00 -1.32763371e-01 6.14029691e-02
-4.76794302e-01 -8.51487756e-01 -9.51633900e-02 9.78071615e-02
-8.04742038e-01 -1.04225719e+00 -8.16206634e-01 -9.19631362e-01
7.84889758e-01 5.56543469e-01 6.90397620e-01 5.09074450e-01
1.47573621e-04 4.50354628e-02 -6.31760657e-01 -3.93036544e-01
-5.21905541e-01 -7.44682789e-01 1.25129640e-01 1.37605563e-01
-2.45761812e-01 -6.21043861e-01 -9.15423691e-01 3.29868108e-01
-1.40239525e+00 -8.04402456e-02 6.18286669e-01 1.12192619e+00
4.70354021e-01 3.92993212e-01 4.03173089e-01 -3.71968538e-01
4.56433773e-01 -6.22892559e-01 -5.07555187e-01 1.81861296e-01
-4.40671951e-01 -1.25611395e-01 6.45378590e-01 -4.31278795e-01
-1.37240100e+00 -2.38793641e-01 -4.61489797e-01 -3.03539038e-01
1.36064783e-01 7.58282661e-01 -4.46667343e-01 -3.84819150e-01
3.67473096e-01 5.67444384e-01 1.98634058e-01 -7.96588242e-01
-2.02452570e-01 7.87387908e-01 5.76799452e-01 -3.26918572e-01
1.06511760e+00 5.18508255e-01 -1.74732655e-01 -9.82992053e-01
-8.00087512e-01 -2.21664369e-01 -1.74459547e-01 -4.14086729e-01
7.19912350e-01 -1.38863254e+00 -3.93407762e-01 1.20251894e+00
-8.64827096e-01 -6.97290003e-01 2.80929923e-01 4.79892582e-01
-1.24178313e-01 8.68752718e-01 -9.61445332e-01 -7.87911177e-01
-4.83300239e-01 -1.31104255e+00 7.23580897e-01 8.13087463e-01
1.02885199e+00 -1.04277050e+00 -4.33524745e-03 2.66526520e-01
4.43654209e-01 3.00599575e-01 5.48933327e-01 1.36856303e-01
-4.67281967e-01 -3.25245261e-02 -3.46977621e-01 9.91865754e-01
2.20030010e-01 9.23471227e-02 -9.94424760e-01 -6.88408613e-01
1.18873306e-01 -3.70604068e-01 1.11458933e+00 3.55828941e-01
1.20226145e+00 -2.06982434e-01 2.30813906e-01 1.27254999e+00
1.90074658e+00 1.82693407e-01 1.13807452e+00 8.24536562e-01
6.27913594e-01 5.52883923e-01 6.02226496e-01 4.33857471e-01
5.77446997e-01 2.40219668e-01 1.05489206e+00 -4.29873914e-01
-9.28258747e-02 -6.87097525e-03 3.49287599e-01 9.36270893e-01
-2.98200667e-01 -3.21296066e-01 -5.78947186e-01 5.69753408e-01
-1.47460759e+00 -7.81627595e-01 -2.78617114e-01 2.02923083e+00
8.05837154e-01 -3.08711767e-01 -6.12049520e-01 8.25056806e-02
6.74721241e-01 1.65301651e-01 -6.40656650e-01 -3.03318948e-02
-4.85926926e-01 -1.78918958e-01 8.40787768e-01 6.00066483e-01
-1.03652501e+00 4.49233115e-01 5.46960497e+00 5.48729420e-01
-1.15880430e+00 1.09044649e-01 4.29685980e-01 4.67777848e-01
-4.88431811e-01 -1.50485873e-01 -4.91492122e-01 6.71945274e-01
5.20904124e-01 1.45672038e-01 4.64225113e-01 4.17311519e-01
6.17797017e-01 1.20832235e-01 -2.60814339e-01 8.92570138e-01
1.94439933e-01 -9.18040395e-01 -7.43583497e-03 -1.44270122e-01
8.88558745e-01 2.50789374e-01 -2.20444843e-01 3.07061374e-01
2.88083613e-01 -7.87399769e-01 7.35885859e-01 8.81463587e-01
7.91037858e-01 -3.26849371e-01 1.11398757e+00 3.19618255e-01
-9.39375401e-01 -4.62137043e-01 -8.53107929e-01 -1.42334417e-01
2.25623846e-01 3.64979059e-01 -2.82456726e-02 7.79463410e-01
1.25904083e+00 7.44070649e-01 -4.65147853e-01 1.45436394e+00
-4.88657176e-01 5.64284444e-01 -1.67161092e-01 2.73003340e-01
-4.11987789e-02 -6.09581172e-01 4.65041369e-01 1.13843036e+00
7.11160719e-01 4.85244066e-01 1.08134188e-02 4.05348033e-01
-5.91455512e-02 -8.92729610e-02 -9.07099918e-02 5.28406560e-01
3.62293214e-01 1.22887599e+00 -6.82772100e-02 -2.06527963e-01
-4.37011093e-01 9.19436574e-01 -5.14055379e-02 7.51243591e-01
-7.19454288e-01 -6.04809403e-01 7.29024112e-01 -3.61615926e-01
2.52947807e-01 -2.72506982e-01 -1.45487294e-01 -1.23887467e+00
3.18677910e-02 -9.01005864e-01 6.51364252e-02 -1.22270775e+00
-1.31458735e+00 7.20733166e-01 -1.89841509e-01 -1.52461946e+00
5.46621919e-01 -6.70309484e-01 -9.03172076e-01 1.07636929e+00
-2.28139567e+00 -1.04799378e+00 -1.19700456e+00 4.97279733e-01
5.63905060e-01 3.82930152e-02 4.91777688e-01 5.76089561e-01
-9.20845926e-01 3.83458912e-01 7.17427552e-01 9.15691704e-02
7.22120941e-01 -1.18284428e+00 -1.88053802e-01 1.30090761e+00
-7.49917507e-01 3.28992993e-01 1.10825849e+00 -5.24330616e-01
-1.53973937e+00 -1.17816973e+00 -6.19293489e-02 3.12745482e-01
5.72043300e-01 2.31850758e-01 -1.40440083e+00 4.08465087e-01
1.41737670e-01 1.15085416e-01 4.86888379e-01 -6.08336270e-01
-8.48265961e-02 -6.27791226e-01 -1.12606680e+00 4.36705738e-01
6.49126947e-01 -1.20553695e-01 -3.18300247e-01 2.60898650e-01
7.91233182e-01 -7.02420413e-01 -1.03389966e+00 6.45960569e-01
4.92595702e-01 -9.55512166e-01 9.59709585e-01 1.87435057e-02
8.31258237e-01 -6.84871078e-01 -3.47772539e-01 -1.53361225e+00
-2.43087292e-01 -3.15237910e-01 2.12492913e-01 1.24131274e+00
3.91167194e-01 -7.91032970e-01 2.02764884e-01 3.51634860e-01
-7.26286829e-01 -5.48032105e-01 -7.68202007e-01 -4.91350025e-01
5.24404421e-02 -8.91343728e-02 4.85724151e-01 7.12640345e-01
-5.79433322e-01 -1.03422917e-01 -8.61062109e-01 9.18763041e-01
1.05069399e+00 -7.12596253e-03 8.33553910e-01 -9.14193690e-01
-1.88734546e-01 5.97706437e-02 -1.74471334e-01 -1.17751396e+00
-2.11901173e-01 -2.83424407e-01 5.77475071e-01 -2.12493753e+00
1.04800344e-01 -5.29407203e-01 -4.13283139e-01 5.75673759e-01
-5.86181760e-01 5.88188052e-01 -9.97875109e-02 3.97041470e-01
-1.41813993e-01 1.03159666e+00 1.74440944e+00 -2.93797344e-01
-6.62258044e-02 1.39783286e-02 -7.88234890e-01 4.73897249e-01
8.93875182e-01 -3.23121607e-01 -1.30180359e-01 -1.11812949e+00
7.59864971e-02 1.52775124e-01 3.06032836e-01 -1.01462913e+00
2.06148654e-01 -1.92732617e-01 4.51146930e-01 -2.37387657e-01
4.78897721e-01 -7.22653508e-01 2.06903219e-01 5.30094385e-01
9.06974524e-02 -2.75375634e-01 2.10756063e-01 7.14719236e-01
-4.93806750e-01 -5.23324788e-01 1.11179698e+00 -3.10444862e-01
-1.03858232e+00 4.93657976e-01 -2.06726696e-02 -2.45195746e-01
5.83777070e-01 -3.78924012e-01 -6.62984908e-01 -5.82234621e-01
-3.27846080e-01 5.94120085e-01 6.34779871e-01 2.19000921e-01
9.62964654e-01 -8.55880916e-01 -1.12175536e+00 1.62055090e-01
7.89557304e-03 1.85005412e-01 8.53829920e-01 6.55812979e-01
-1.09426510e+00 -8.39645863e-01 -3.39480877e-01 -3.79117608e-01
-1.06133521e+00 2.38407075e-01 8.36328804e-01 2.57491916e-01
-7.46384382e-01 8.10256898e-01 1.67271703e-01 -4.63860184e-01
7.94050321e-02 -3.19192782e-02 -4.19666678e-01 -3.09016764e-01
8.17776084e-01 4.71026808e-01 -2.15473637e-01 -6.13866806e-01
8.66263881e-02 8.00896108e-01 4.20631498e-01 3.16401720e-01
1.73383987e+00 -7.57656515e-01 -3.25857341e-01 1.57918289e-01
1.24411428e+00 -4.03552763e-02 -1.82353902e+00 -2.27112457e-01
-7.65436172e-01 -6.39861286e-01 1.98819578e-01 -6.95933819e-01
-1.50183153e+00 9.46649432e-01 1.00392246e+00 1.73755199e-01
1.58817542e+00 -4.19495374e-01 9.32568252e-01 3.47836286e-01
1.60428703e-01 -8.97209227e-01 4.64821570e-02 4.59679276e-01
9.59922552e-01 -1.55500031e+00 1.83448210e-01 -1.72289073e-01
-4.94917929e-01 1.38748968e+00 6.65552020e-01 -8.53457153e-02
4.60414767e-01 2.98789889e-01 6.93650067e-01 -1.68976914e-02
-7.24210441e-02 -1.01245731e-01 -2.56450802e-01 6.57049119e-01
-1.31888911e-01 -1.03916064e-01 -2.59314299e-01 5.36570251e-01
9.09621175e-03 -2.12585181e-01 9.86017823e-01 6.63334191e-01
-7.76141167e-01 -7.77860761e-01 -5.77867985e-01 2.26308689e-01
-4.63636875e-01 -1.02944553e-01 5.46126306e-01 5.01892805e-01
2.48751804e-01 1.16249859e+00 -3.59843761e-01 -4.79147136e-01
4.92523730e-01 -7.69341528e-01 3.01281959e-02 -2.51816720e-01
7.80034170e-04 3.63218784e-02 -6.83772415e-02 -1.74426764e-01
-7.08174407e-01 -3.59844565e-01 -1.12503183e+00 3.79361100e-02
-4.62167442e-01 2.20681265e-01 7.73881316e-01 8.56737673e-01
-2.39218861e-01 6.94515109e-01 9.97153461e-01 -1.29253805e+00
-5.29252470e-01 -1.29548967e+00 -8.45390499e-01 4.52982515e-01
7.60293305e-01 -5.73940396e-01 -9.13803816e-01 1.94989651e-01]
|
[10.702558517456055, -3.531609296798706]
|
0f289bf9-e89c-4395-87c5-2533f41c9fd1
|
network-free-unsupervised-semantic
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Feng_Network-Free_Unsupervised_Semantic_Segmentation_With_Synthetic_Images_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Feng_Network-Free_Unsupervised_Semantic_Segmentation_With_Synthetic_Images_CVPR_2023_paper.pdf
|
Network-Free, Unsupervised Semantic Segmentation With Synthetic Images
|
We derive a method that yields highly accurate semantic segmentation maps without the use of any additional neural network, layers, manually annotated training data, or supervised training. Our method is based on the observation that the correlation of a set of pixels belonging to the same semantic segment do not change when generating synthetic variants of an image using the style mixing approach in GANs. We show how we can use GAN inversion to accurately semantically segment synthetic and real photos as well as generate large training image-semantic segmentation mask pairs for downstream tasks.
|
['Aleix Martinez', 'Eduard Ramon', 'Wentong Liao', 'Raghudeep Gadde', 'Qianli Feng']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['unsupervised-semantic-segmentation']
|
['computer-vision']
|
[ 1.01613808e+00 7.33234286e-01 2.49307364e-01 -6.10929966e-01
-8.36563528e-01 -9.50461328e-01 5.64258814e-01 -4.38579232e-01
-1.25053421e-01 8.17606330e-01 -1.66450903e-01 -1.77251920e-02
4.47458863e-01 -9.70962942e-01 -1.06172752e+00 -5.42527378e-01
6.47091210e-01 7.11822510e-01 3.08467686e-01 -1.48290768e-01
-4.73356061e-02 4.16117251e-01 -1.05093229e+00 2.02087015e-01
8.98189664e-01 8.97182703e-01 2.30184704e-01 7.20151782e-01
-1.81469291e-01 4.32270378e-01 -7.33238876e-01 -4.34737027e-01
7.78101921e-01 -1.13503361e+00 -7.95082986e-01 4.94053334e-01
4.07548517e-01 -3.60857733e-02 -4.63508926e-02 1.09960830e+00
7.48433322e-02 6.78059831e-02 7.39453435e-01 -1.10988009e+00
-5.02005756e-01 5.40113270e-01 -5.17081678e-01 -4.56519425e-01
-9.61671844e-02 1.00284450e-01 5.65938890e-01 -3.74472260e-01
8.05194676e-01 9.46973920e-01 7.80600727e-01 6.65662766e-01
-1.45568514e+00 -4.38080043e-01 -1.44828379e-01 -3.27203542e-01
-1.19013786e+00 -3.55640262e-01 9.46845949e-01 -3.28383863e-01
2.82092541e-01 1.30994812e-01 8.10221136e-01 9.98027980e-01
-3.14164817e-01 6.06514215e-01 1.17073774e+00 -5.94903827e-01
3.35941851e-01 1.17839158e-01 -3.49194407e-01 6.82014823e-01
9.59869772e-02 1.21090114e-02 -1.70581594e-01 3.00397038e-01
1.04887378e+00 -3.68973285e-01 -2.17635259e-01 -3.40050459e-01
-1.05010903e+00 7.09036589e-01 4.51894999e-01 1.06577992e-01
-2.48114228e-01 4.92159814e-01 8.04204419e-02 1.44834906e-01
4.43491042e-01 7.63602674e-01 -3.56872052e-01 2.16808796e-01
-1.23545730e+00 -5.47312498e-02 4.23012793e-01 1.07533789e+00
1.12743068e+00 4.31601018e-01 8.96942616e-02 5.98461211e-01
-1.53147414e-01 6.19004190e-01 2.78636068e-01 -1.59676659e+00
1.91925153e-01 3.38510484e-01 4.40642573e-02 -4.82502967e-01
1.88522432e-02 -3.01705807e-01 -5.63063204e-01 1.91706270e-01
3.81207347e-01 -4.28030819e-01 -1.40994132e+00 1.90934575e+00
1.19802274e-01 2.19958439e-01 3.08906287e-01 3.94785851e-01
3.41653526e-01 3.95242244e-01 -9.53094065e-02 1.53967410e-01
7.84726679e-01 -1.09250522e+00 -4.32902515e-01 -4.55151975e-01
5.13371229e-01 -6.58699095e-01 1.05601561e+00 2.34172903e-02
-1.28552973e+00 -5.42664170e-01 -1.14135420e+00 -8.07579756e-02
-2.26396829e-01 5.73891811e-02 5.84803224e-01 7.24505782e-01
-1.29515326e+00 5.49747646e-01 -7.30410814e-01 -2.19709456e-01
7.08899498e-01 2.06925631e-01 -2.79692918e-01 7.90268853e-02
-6.77476287e-01 6.54488981e-01 5.45909941e-01 -1.65671147e-02
-9.86483634e-01 -5.20838439e-01 -9.20476854e-01 -1.27716810e-01
1.28464684e-01 -7.80876458e-01 1.20820844e+00 -2.03762102e+00
-1.55983496e+00 1.00380051e+00 -1.16660453e-01 -3.64380240e-01
7.45274186e-01 7.86560774e-02 -7.13243783e-02 2.66358942e-01
1.89151734e-01 1.30614018e+00 1.04209185e+00 -1.50210166e+00
-4.11722809e-01 -2.07601771e-01 -1.25310123e-01 1.94371760e-01
2.82794893e-01 -4.07034010e-01 -4.24884290e-01 -7.77712047e-01
1.37451589e-01 -1.14470553e+00 -4.06546026e-01 -1.62774950e-01
-7.31961131e-01 4.92973983e-01 8.91672194e-01 -8.01829517e-01
2.12453857e-01 -2.05319166e+00 2.02648476e-01 4.18461382e-01
-2.09224150e-01 2.22568914e-01 -4.19198424e-01 -6.00677952e-02
6.32400718e-03 2.69970328e-01 -9.09006834e-01 -3.32369119e-01
-2.47949407e-01 3.67243379e-01 -4.44223046e-01 1.59026995e-01
2.87609667e-01 1.20021296e+00 -8.43210220e-01 -3.82023394e-01
2.76929975e-01 3.44135404e-01 -4.90620106e-01 3.35019588e-01
-5.95270872e-01 1.10517263e+00 -1.54309645e-01 2.09169134e-01
5.42492807e-01 -1.45617589e-01 1.57684401e-01 -5.53407483e-02
3.63996297e-01 1.73031211e-01 -7.58074164e-01 2.06777263e+00
-4.51486140e-01 8.05817366e-01 -1.13007516e-01 -1.14211905e+00
6.95068240e-01 1.21588118e-01 3.52164924e-01 -6.15151167e-01
1.27120867e-01 2.93663710e-01 -2.97924757e-01 2.20007554e-04
3.06123585e-01 -2.20461920e-01 -1.34804383e-01 5.88009119e-01
2.05907479e-01 -9.43206728e-01 1.89667687e-01 5.64649664e-02
7.11426198e-01 4.03222769e-01 -3.42089176e-01 -1.15309142e-01
1.79058492e-01 1.94661707e-01 5.00772357e-01 6.76217496e-01
3.45406085e-01 1.13898945e+00 5.33960640e-01 -7.85174891e-02
-1.56683433e+00 -1.33268464e+00 4.15065251e-02 4.14075226e-01
2.27170199e-01 1.23034306e-01 -1.49826193e+00 -7.84368753e-01
-4.49795663e-01 1.04890585e+00 -6.35605872e-01 -1.64617777e-01
-6.01561844e-01 -7.05653012e-01 7.41123438e-01 6.29504025e-01
8.77094448e-01 -1.08425176e+00 -3.89086962e-01 -4.95962538e-02
-2.62936622e-01 -1.30866194e+00 -4.54408050e-01 1.24842882e-01
-8.58361244e-01 -1.13513923e+00 -8.03695083e-01 -9.96053815e-01
1.08156085e+00 1.49496784e-02 1.20706177e+00 -9.75312218e-02
-1.01517521e-01 2.89183378e-01 -1.54900402e-01 -2.11013988e-01
-8.92750561e-01 2.08623499e-01 -6.84653223e-01 -9.94020700e-02
-2.48375773e-01 -6.46251082e-01 -3.85260999e-01 3.74963462e-01
-1.08385241e+00 6.89551771e-01 3.35864991e-01 6.82100654e-01
7.76466012e-01 -5.99277802e-02 4.04917687e-01 -1.21053529e+00
1.35539904e-01 -8.65205303e-02 -7.02231824e-01 3.40358049e-01
-3.89697373e-01 2.64178932e-01 4.90483224e-01 -8.55680704e-02
-1.06774914e+00 4.33368206e-01 -1.77741945e-01 -1.13068044e-01
-3.90799761e-01 -5.70306145e-02 -2.98539013e-01 -1.94871515e-01
7.87843525e-01 1.64056003e-01 -6.46382496e-02 -1.88505560e-01
8.36611867e-01 3.58670324e-01 8.32673669e-01 -4.47018206e-01
8.71653259e-01 8.31509531e-01 7.26750717e-02 -6.78532064e-01
-8.85423958e-01 1.23145975e-01 -9.05130267e-01 5.95867187e-02
1.22543359e+00 -7.81297565e-01 6.29752353e-02 5.55081129e-01
-1.18583763e+00 -1.01386774e+00 -8.16393316e-01 1.08119078e-01
-9.76225853e-01 1.20247334e-01 -4.45023447e-01 -2.50234693e-01
-2.13951409e-01 -1.08618474e+00 1.19699407e+00 1.77420929e-01
-1.19763620e-01 -8.56843293e-01 -2.13282377e-01 4.59482342e-01
1.76653266e-01 4.70534265e-01 8.40588570e-01 -2.90819645e-01
-7.74142504e-01 -2.56928168e-02 -3.83458704e-01 6.82084024e-01
3.89625043e-01 -6.51101321e-02 -9.38212812e-01 9.92339104e-02
-1.19858772e-01 -3.83049071e-01 8.85664463e-01 5.16498923e-01
1.16928923e+00 -2.32929572e-01 -2.06629947e-01 9.60562885e-01
1.46083081e+00 2.87416130e-01 8.63857746e-01 -2.29851112e-01
1.09852576e+00 7.46454775e-01 -6.50750399e-02 -3.80719900e-01
1.17128879e-01 5.04762888e-01 1.00031264e-01 -5.02174497e-01
-4.66791630e-01 -5.01212180e-01 9.95872356e-03 3.70081216e-01
1.17976606e-01 -4.73422617e-01 -6.91686869e-01 6.24906838e-01
-1.56147420e+00 -7.45295644e-01 -8.39850772e-03 2.06098938e+00
8.34003150e-01 -1.08870715e-01 -1.35896459e-01 -4.34088297e-02
8.20500910e-01 -3.20025906e-02 -6.88493133e-01 -3.35148543e-01
-3.84458274e-01 6.20779514e-01 9.43113565e-01 5.93529165e-01
-9.14782345e-01 1.40674031e+00 7.46894550e+00 6.77859187e-01
-1.03330076e+00 2.39841387e-01 1.07954800e+00 1.39751166e-01
-6.28397644e-01 1.22532569e-01 -1.77252874e-01 3.52513969e-01
7.20663548e-01 2.24965557e-01 6.68576181e-01 5.76845407e-01
1.21328486e-02 -3.55916053e-01 -9.04688716e-01 7.08835185e-01
1.09508805e-01 -1.52466631e+00 1.15583725e-01 -1.86525181e-01
1.35523546e+00 -5.52582406e-02 7.86891356e-02 -3.39627624e-01
8.56596649e-01 -1.20398486e+00 1.00331604e+00 3.66644025e-01
1.04350090e+00 -4.82563049e-01 4.43575531e-01 1.43795852e-02
-6.90895736e-01 4.73287493e-01 -1.41948715e-01 3.87343317e-01
3.90560120e-01 5.87723434e-01 -8.36361051e-01 3.29768300e-01
2.00234056e-01 5.31081080e-01 -5.72359085e-01 6.13491893e-01
-7.22445607e-01 6.27719164e-01 -3.08800876e-01 6.66077554e-01
1.85686097e-01 -5.18955469e-01 2.28464797e-01 5.96218884e-01
3.46669525e-01 -1.43955871e-01 -1.20833190e-02 1.35198462e+00
-1.60735101e-01 -2.42277950e-01 -6.71579659e-01 -1.91393301e-01
1.65099949e-01 9.00200844e-01 -1.19620717e+00 -4.84999895e-01
-9.80416536e-02 1.66825509e+00 9.50534120e-02 6.77040339e-01
-7.41481245e-01 -1.65490255e-01 3.87996942e-01 9.42479298e-02
4.70162958e-01 -3.19974542e-01 -8.97580087e-01 -1.04852116e+00
2.49337088e-02 -5.36598802e-01 -1.13426775e-01 -1.17845356e+00
-9.48204279e-01 6.53152227e-01 -2.03806102e-01 -7.24850833e-01
-5.24331748e-01 -1.74673513e-01 -7.53104866e-01 8.43486249e-01
-1.19750285e+00 -1.39770508e+00 -3.67504627e-01 4.27845895e-01
3.22026938e-01 -2.00499571e-03 8.50897789e-01 -7.43304491e-02
-3.10195327e-01 4.15072441e-01 5.96721796e-03 4.07305658e-01
3.82321686e-01 -1.34856725e+00 8.04063857e-01 1.06548250e+00
2.43978724e-01 6.01207651e-02 5.91843724e-01 -6.62955165e-01
-4.39524770e-01 -1.32978511e+00 3.23439449e-01 -3.04952919e-01
7.11629614e-02 -4.40556467e-01 -4.92041528e-01 1.02094626e+00
3.88534248e-01 -1.33771285e-01 4.65562046e-01 -4.30501521e-01
-2.26521671e-01 3.90587114e-02 -1.26386583e+00 5.74204683e-01
1.23328626e+00 -6.07302785e-01 -3.05341840e-01 3.60533148e-01
8.89879584e-01 -4.75738049e-01 -3.85399401e-01 4.65591222e-01
2.63872445e-01 -1.09650481e+00 9.03980196e-01 -4.75569159e-01
5.71832597e-01 -3.43832254e-01 -1.34539172e-01 -1.49246335e+00
2.82390296e-01 -4.65675235e-01 5.85999966e-01 1.04269946e+00
6.81028306e-01 -5.17021775e-01 1.03536785e+00 7.17875481e-01
-3.29252630e-01 -6.65727034e-02 -5.85333228e-01 -7.48973310e-01
1.13036938e-01 -1.89141899e-01 8.12209547e-01 8.62848818e-01
-6.93409264e-01 2.52388328e-01 -2.59573534e-02 -4.23891880e-02
5.36826015e-01 2.82827258e-01 8.67659569e-01 -9.76983905e-01
-4.09403801e-01 -2.74897218e-01 -3.90272021e-01 -7.62445152e-01
4.13008928e-01 -8.44828010e-01 2.58843660e-01 -1.59610713e+00
-1.85877979e-01 -8.19756806e-01 2.09079891e-01 5.33450186e-01
8.08059052e-02 8.60811353e-01 -1.92975309e-02 7.93484002e-02
-2.56393909e-01 5.31837702e-01 1.37792826e+00 -1.65616319e-01
-1.38807371e-01 -1.38695255e-01 -7.08143830e-01 8.08083534e-01
9.31490421e-01 -4.77241576e-01 -7.69896865e-01 -6.41584277e-01
2.23486036e-01 -1.48805186e-01 5.23718417e-01 -1.07998800e+00
-4.49717581e-01 -1.28412023e-02 5.80503643e-01 -8.66595954e-02
3.08673471e-01 -7.38697588e-01 6.53469801e-01 2.35537767e-01
-4.75635797e-01 -1.65348887e-01 1.51036650e-01 3.90974402e-01
-1.37518778e-01 -3.52943093e-01 1.06598139e+00 -4.95537192e-01
-5.27596474e-01 9.24538821e-02 -1.15161203e-01 3.05031836e-01
1.09518778e+00 -2.97802240e-01 -5.01253754e-02 -5.01125455e-01
-6.97337687e-01 -1.74724683e-01 1.21298575e+00 9.72969234e-02
1.63364429e-02 -1.23374724e+00 -3.40222448e-01 4.55096573e-01
-2.27960974e-01 4.90901589e-01 4.35288325e-02 3.63373518e-01
-8.89815390e-01 3.63415778e-02 -3.77153039e-01 -5.56063652e-01
-6.26847267e-01 2.14388385e-01 6.50669694e-01 1.00423634e-01
-4.71712172e-01 8.24691057e-01 4.73644078e-01 -6.15188956e-01
-3.56444985e-01 -1.34273633e-01 5.49886286e-01 -3.46804261e-01
1.30987406e-01 -9.53781828e-02 2.58652922e-02 -6.17780209e-01
1.29524991e-01 5.99459052e-01 5.10789335e-01 -6.95022941e-01
1.13865244e+00 -9.01792645e-02 -3.27987745e-02 2.99842745e-01
1.18979406e+00 -7.11635426e-02 -1.61673260e+00 7.51158148e-02
-3.04344654e-01 -4.29527551e-01 -1.08788736e-01 -8.21120203e-01
-1.45591938e+00 6.92181528e-01 4.76110160e-01 -4.66109551e-02
1.19285119e+00 9.61523429e-02 1.11367011e+00 -5.31502329e-02
3.75229686e-01 -1.30239534e+00 2.57147588e-02 2.17052326e-01
5.77758908e-01 -1.00156033e+00 -2.99850404e-01 -7.51208663e-01
-6.68485641e-01 7.79316783e-01 3.90914708e-01 -3.58792782e-01
3.76990318e-01 2.97131658e-01 3.02221835e-01 -3.71874087e-02
-8.70046616e-02 -2.78165728e-01 1.34975269e-01 9.61467624e-01
1.34181544e-01 2.18979180e-01 -2.51168832e-02 9.00936499e-02
-5.12801826e-01 -1.33901358e-01 5.27483463e-01 6.35417104e-01
-1.10840105e-01 -1.31508088e+00 -7.32069984e-02 3.31674933e-01
-1.50189251e-01 -2.32339263e-01 -6.27971709e-01 4.48631734e-01
2.15431929e-01 6.17323399e-01 4.73781139e-01 -2.45349124e-01
-1.91428632e-01 2.70142198e-01 8.75897706e-01 -7.24691033e-01
-2.41302252e-01 -4.68451418e-02 2.73980815e-02 -4.78416741e-01
-5.92314422e-01 -5.41331708e-01 -1.44999802e+00 -1.00894555e-01
-1.66345537e-01 -3.77983861e-02 7.51232028e-01 1.14264345e+00
3.07920814e-01 5.40889323e-01 5.26033938e-01 -8.11345875e-01
2.87048131e-01 -7.80828297e-01 -4.96280789e-01 6.01355433e-01
3.72289196e-02 -2.39030123e-01 -1.97291344e-01 5.37702739e-01]
|
[11.469635963439941, -0.37535035610198975]
|
7ee0199c-7b2a-4929-ba72-8ca4d828d699
|
drone-data-aware-low-rank-compression-for
| null | null |
http://proceedings.neurips.cc/paper/2021/hash/f56de5ef149cf0aedcc8f4797031e229-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/f56de5ef149cf0aedcc8f4797031e229-Paper.pdf
|
DRONE: Data-aware Low-rank Compression for Large NLP Models
|
The representations learned by large-scale NLP models such as BERT have been widely used in various tasks. However, the increasing model size of the pre-trained models also brings efficiency challenges, including inference speed and model size when deploying models on mobile devices. Specifically, most operations in BERT consist of matrix multiplications. These matrices are not low-rank and thus canonical matrix decomposition could not find an efficient approximation. In this paper, we observe that the learned representation of each layer lies in a low-dimensional space. Based on this observation, we propose DRONE (data-aware low-rank compression), a provably optimal low-rank decomposition of weight matrices, which has a simple closed form solution that can be efficiently computed. DRONE can be applied to both fully connected and self-attention layers appearing in the BERT model. In addition to compressing standard models, out method can also be used on distilled BERT models to further improve compression rate. Experimental results show that DRONE is able to improve both model size and inference speed with limited loss in accuracy. Specifically, DRONE alone achieves 1.92x speedup on the MRPC task with only 1.5% loss in accuracy, and when DRONE is combined with distillation, it further achieves over 12.3x speedup on various natural language inference tasks.
|
['Cho-Jui Hsieh', 'Inderjit Dhillon', 'Hsiang-Fu Yu', 'Pei-Hung Chen']
|
2021-12-01
| null |
https://openreview.net/forum?id=sthiz9zeXGG
|
https://openreview.net/pdf?id=sthiz9zeXGG
|
neurips-2021-12
|
['low-rank-compression']
|
['computer-code']
|
[-3.30822542e-02 8.88232961e-02 -2.69210398e-01 -2.24366426e-01
-6.85699761e-01 -4.31222916e-01 2.76244223e-01 -6.29965812e-02
-6.10678434e-01 4.68971521e-01 3.90282162e-02 -5.44158041e-01
-3.09765637e-01 -8.62474144e-01 -1.11806738e+00 -3.35069627e-01
-6.99835569e-02 8.53950083e-01 -7.99595043e-02 -1.11627474e-01
-1.34205103e-01 3.86177480e-01 -1.47476792e+00 5.65533102e-01
9.81783152e-01 9.33692455e-01 7.39915609e-01 9.01258945e-01
1.38925239e-02 9.29411650e-01 -3.63537818e-01 -6.46813869e-01
3.26127499e-01 3.09412748e-01 -7.54804790e-01 -3.42076659e-01
7.32586920e-01 -7.79710352e-01 -8.84943128e-01 8.35850656e-01
2.72063851e-01 2.22032622e-01 4.70358253e-01 -1.24728584e+00
-2.90482998e-01 9.68751013e-01 -3.90691072e-01 3.04671198e-01
-5.88404946e-02 -2.82802761e-01 1.32955039e+00 -1.08478534e+00
2.59292603e-01 1.43454492e+00 5.12583017e-01 4.49071646e-01
-1.11746573e+00 -8.15446138e-01 2.62171775e-01 4.48762745e-01
-1.45915854e+00 -5.15244842e-01 2.94018328e-01 1.01542450e-01
1.39280570e+00 4.66782063e-01 6.48714960e-01 9.22172129e-01
-9.48403999e-02 1.22416341e+00 5.18632293e-01 -1.79391161e-01
4.57143113e-02 -7.62653872e-02 2.50022918e-01 1.05680692e+00
6.51902974e-01 -3.73495281e-01 -5.91405690e-01 -1.41199768e-01
4.28711444e-01 2.95109123e-01 -1.89999521e-01 2.41563004e-02
-1.08687556e+00 7.95097172e-01 6.30048692e-01 1.02235585e-01
-3.15732479e-01 6.93955660e-01 3.33692014e-01 1.80738911e-01
3.38367850e-01 3.50655586e-01 -7.67658472e-01 -4.08843189e-01
-1.02411556e+00 3.56040359e-01 9.20116484e-01 1.08057392e+00
7.91793227e-01 -1.24898277e-01 2.98323371e-02 9.29794133e-01
2.05500662e-01 7.42459893e-01 6.86780572e-01 -1.01686513e+00
1.14280367e+00 5.60523331e-01 -2.25718915e-01 -1.32374108e+00
-3.93777728e-01 -6.41827047e-01 -1.18640041e+00 -6.81178153e-01
8.28891769e-02 -1.51261568e-01 -5.53043306e-01 1.62150335e+00
1.02629073e-01 5.10891378e-01 4.88477573e-02 5.59644401e-01
5.18936932e-01 9.87513542e-01 -2.71913707e-01 7.09732249e-02
1.41534555e+00 -1.37626791e+00 -7.17556059e-01 -6.59425259e-01
1.09857297e+00 -4.39016789e-01 1.19712818e+00 6.21071517e-01
-1.26674664e+00 -4.25473660e-01 -1.23696458e+00 -6.53058350e-01
-3.43450993e-01 5.23308337e-01 1.00130391e+00 3.70644212e-01
-8.83725107e-01 7.01072335e-01 -1.15141165e+00 9.68805179e-02
5.73052347e-01 4.91420239e-01 -2.50168830e-01 -7.30412841e-01
-1.13349569e+00 7.97896981e-01 5.76151848e-01 2.86087453e-01
-7.67112076e-01 -9.00626481e-01 -7.67603219e-01 5.28518438e-01
5.17679989e-01 -7.88592458e-01 1.13499141e+00 -1.44860759e-01
-1.12374032e+00 2.72029847e-01 -5.72228372e-01 -1.01579332e+00
2.56382376e-01 -7.38172174e-01 -1.72304690e-01 2.16703057e-01
-3.17771763e-01 6.91225410e-01 7.05826521e-01 -6.78940535e-01
-7.76425183e-01 -3.88060689e-01 4.38097924e-01 2.87719280e-01
-8.79333735e-01 -4.21533942e-01 -9.91534472e-01 -3.45665127e-01
8.11460242e-02 -1.15520012e+00 -2.08752975e-01 -2.34935079e-02
-1.33113369e-01 -3.07794094e-01 6.89569652e-01 -7.76515782e-01
1.72785580e+00 -1.96642613e+00 1.47379458e-01 2.20224291e-01
5.16911864e-01 5.34970641e-01 -2.72473484e-01 2.78464526e-01
4.52023715e-01 1.60136029e-01 -1.92857191e-01 -6.65930808e-01
5.40875904e-02 6.73115551e-01 -4.69036222e-01 1.90599352e-01
-1.39505425e-02 1.14174092e+00 -8.08380604e-01 -3.39433283e-01
7.69766560e-03 5.55031121e-01 -1.09203517e+00 3.09920534e-02
-2.16935039e-01 -5.03049254e-01 -1.11971587e-01 2.96305656e-01
8.31336141e-01 -5.76444507e-01 3.83086979e-01 -3.35086018e-01
4.82394934e-01 6.57208085e-01 -1.12449312e+00 1.74976301e+00
-1.10342658e+00 7.73464859e-01 3.88877876e-02 -1.12072682e+00
4.33098018e-01 3.83807831e-02 1.12222053e-01 -3.88129979e-01
-8.84172618e-02 1.99533761e-01 8.16878676e-02 -2.63918012e-01
5.92768192e-01 3.80333215e-01 2.60464698e-01 4.79639441e-01
1.61470205e-01 8.85899514e-02 5.90517521e-01 5.69569647e-01
1.18271792e+00 -4.63846654e-01 1.09434165e-01 -4.02351841e-03
3.63268197e-01 -3.34541857e-01 3.11485440e-01 7.14385748e-01
4.56785649e-01 2.79092222e-01 3.79666120e-01 -4.25249845e-01
-8.04339349e-01 -6.89340830e-01 -7.47547904e-03 1.25679171e+00
-2.55859077e-01 -1.22067523e+00 -6.90207958e-01 -6.36221230e-01
7.41892681e-03 6.39952123e-01 -2.40539581e-01 -3.57359827e-01
-7.08515167e-01 -8.38293731e-01 6.23954773e-01 7.19234765e-01
8.01035464e-01 -5.16284823e-01 -3.78815055e-01 2.27388933e-01
-4.65734154e-01 -1.49849927e+00 -3.28231901e-01 1.48166075e-01
-1.31997120e+00 -1.08890963e+00 -2.83861399e-01 -5.86344421e-01
8.65663052e-01 7.08069861e-01 1.04955101e+00 4.81459141e-01
-5.24300560e-02 9.26144570e-02 -1.97486863e-01 -4.86233950e-01
-1.05486745e-02 5.98921061e-01 6.06203340e-02 -1.46426946e-01
3.76715630e-01 -6.94990575e-01 -4.34087843e-01 4.90764454e-02
-9.41922963e-01 2.31770188e-01 7.72275984e-01 8.92153978e-01
5.17228365e-01 3.77287179e-01 1.71971112e-01 -1.04260516e+00
6.20770216e-01 -3.72630388e-01 -5.76860309e-01 3.42800766e-01
-6.15974188e-01 3.87880176e-01 8.61668289e-01 -3.73841405e-01
-7.53583610e-01 3.68417725e-02 -2.48607427e-01 -6.72815740e-01
4.22783107e-01 9.08886611e-01 -6.83365315e-02 2.67443359e-01
5.15901387e-01 1.96255893e-01 -2.35202350e-02 -7.47087836e-01
5.55759788e-01 7.20113099e-01 3.37347567e-01 -6.03367329e-01
8.48066926e-01 4.55711901e-01 -6.40900880e-02 -9.34222281e-01
-1.22421932e+00 -4.59381491e-01 -4.15735573e-01 4.14313555e-01
3.22333515e-01 -1.40308404e+00 -6.64911926e-01 -6.93275686e-03
-1.17560971e+00 -4.11629438e-01 1.30819017e-02 6.62300408e-01
-3.17116052e-01 3.49949598e-01 -7.97252774e-01 -6.28195524e-01
-7.13282108e-01 -9.26591516e-01 1.02742541e+00 -3.18379909e-01
6.66961353e-03 -8.75029027e-01 -4.09512192e-01 7.06650615e-01
3.22081506e-01 -5.26950598e-01 1.08318341e+00 -5.47080159e-01
-6.83866918e-01 -3.89430076e-01 -3.48051548e-01 4.78974819e-01
-3.79160196e-01 -2.63972819e-01 -9.18202519e-01 -6.34548306e-01
-6.11751601e-02 -4.31242138e-01 1.12354493e+00 1.23983376e-01
1.73625875e+00 -7.56544709e-01 -4.25051451e-01 8.36160183e-01
1.27277195e+00 -1.99280292e-01 5.69848299e-01 -6.67260364e-02
9.79833007e-01 1.69231862e-01 5.75433016e-01 4.23202217e-01
4.26980883e-01 6.01230204e-01 4.04951930e-01 1.10441804e-01
-2.60980520e-03 -6.28772199e-01 5.11338115e-01 1.32971525e+00
-2.31540963e-01 -2.22869828e-01 -8.56231451e-01 2.91329861e-01
-2.00804305e+00 -7.67812192e-01 4.02859524e-02 2.14906383e+00
7.67525077e-01 7.02066049e-02 -2.15627477e-01 3.86331499e-01
3.39508057e-01 1.36315465e-01 -5.40704370e-01 -2.82429338e-01
1.04735538e-01 2.65838206e-01 7.02874601e-01 5.71093261e-01
-8.66179526e-01 1.05168641e+00 5.99485922e+00 1.18340492e+00
-9.08100843e-01 2.84630686e-01 4.81748402e-01 -5.79093874e-01
-1.13846935e-01 -1.61006778e-01 -1.23039305e+00 4.12199736e-01
1.43110573e+00 -1.57710537e-01 9.27980423e-01 1.20361781e+00
-6.41433969e-02 3.85286629e-01 -1.27962613e+00 1.39533567e+00
6.72798455e-02 -1.58598971e+00 3.61069024e-01 3.49082351e-01
6.70465529e-01 4.02234614e-01 -4.34994958e-02 8.31478834e-01
1.80680200e-01 -1.00398743e+00 3.06720883e-01 1.13441244e-01
7.94352114e-01 -9.75675106e-01 6.91156805e-01 9.53570604e-01
-1.22188473e+00 -2.96936363e-01 -1.13031578e+00 -1.98151663e-01
6.50153235e-02 7.65609801e-01 -9.01786506e-01 2.60579377e-01
6.57849491e-01 7.05948114e-01 -5.72611272e-01 6.10878289e-01
-1.96664035e-01 7.33120322e-01 -7.84761012e-01 -3.74792740e-02
2.20421314e-01 -1.27801016e-01 1.36191800e-01 1.16147470e+00
4.43822116e-01 1.21635973e-01 -5.53184636e-02 5.12364745e-01
-5.52152932e-01 4.23865691e-02 -6.05749965e-01 -2.06856146e-01
6.51250541e-01 1.16318691e+00 -1.52526289e-01 -6.81013942e-01
-4.42570984e-01 9.67571020e-01 9.05231476e-01 2.26613417e-01
-1.07404625e+00 -1.86429754e-01 7.23411739e-01 2.32144207e-01
4.88508761e-01 -4.68850523e-01 3.05193290e-02 -1.52310205e+00
1.98777214e-01 -1.00640261e+00 3.53107214e-01 -6.35160625e-01
-8.44603658e-01 4.10286337e-01 1.00953825e-01 -8.27695191e-01
-1.68440267e-01 -7.72638977e-01 -1.34010136e-01 3.91279429e-01
-1.54468071e+00 -8.49443734e-01 -2.25275233e-01 6.46608233e-01
6.41576707e-01 -1.86969370e-01 9.07163084e-01 6.74824178e-01
-8.11544716e-01 9.93259609e-01 3.91767830e-01 3.28209922e-02
2.42584988e-01 -1.11335266e+00 5.05979657e-01 8.24638307e-01
7.27695227e-01 1.05247164e+00 1.44407958e-01 -3.32539439e-01
-1.93917704e+00 -1.39754105e+00 1.19659400e+00 -1.53346986e-01
5.44629395e-01 -6.89313352e-01 -7.35698819e-01 9.32625830e-01
-1.99742794e-01 1.31053448e-01 5.96346617e-01 4.02479023e-01
-3.90059233e-01 -4.36007351e-01 -8.32055986e-01 6.94684565e-01
1.40083373e+00 -6.83500051e-01 -3.51387531e-01 8.24836254e-01
9.76592779e-01 -5.92928827e-01 -7.55524278e-01 2.15623021e-01
3.83424789e-01 -4.27317262e-01 1.19378757e+00 -8.01587403e-01
5.27702034e-01 5.15005402e-02 -3.86484802e-01 -1.04697526e+00
-3.21084559e-01 -6.09730065e-01 -1.09991169e+00 7.83550382e-01
6.31121576e-01 -5.18474102e-01 9.42814946e-01 7.99171805e-01
8.97192657e-02 -1.12480664e+00 -7.98661232e-01 -7.90723920e-01
-1.78258285e-01 -8.82719576e-01 7.24117875e-01 5.57187796e-01
-1.86038643e-01 7.72271276e-01 -5.66055119e-01 2.66701430e-01
3.74083579e-01 3.88926603e-02 9.78495896e-01 -1.25112319e+00
-5.47657132e-01 1.89018138e-02 -2.64595654e-02 -1.96232700e+00
2.75568008e-01 -1.31181645e+00 -3.24106723e-01 -1.59966123e+00
8.88090357e-02 -5.64214826e-01 -1.54062063e-01 5.61505497e-01
-1.56888321e-01 2.29775477e-02 4.62017596e-01 2.35031858e-01
-6.40746295e-01 5.51846981e-01 1.00146353e+00 -3.04021478e-01
1.43264327e-02 6.11542119e-03 -6.66348398e-01 7.59797513e-01
9.05497968e-01 -5.13617516e-01 -7.30349302e-01 -1.00672507e+00
6.55350804e-01 -2.15673983e-01 -6.44180039e-03 -1.10303020e+00
4.25567389e-01 3.38008344e-01 1.56027228e-01 -7.18743980e-01
6.82814121e-01 -1.02947426e+00 -1.67706817e-01 6.06358171e-01
-3.14939409e-01 2.63871670e-01 1.80720538e-01 7.27535665e-01
-1.58738434e-01 -3.93624514e-01 4.39756989e-01 -8.87829438e-02
-3.43444675e-01 5.70300877e-01 -1.55855194e-01 -3.55929546e-02
4.71026123e-01 2.44083360e-01 -2.82181740e-01 -2.84265488e-01
-4.18632686e-01 2.70894140e-01 -1.75172865e-01 2.84255266e-01
8.02858591e-01 -1.19487083e+00 -5.88019490e-01 1.67643473e-01
-2.51191258e-01 5.59816420e-01 2.79904664e-01 8.25096369e-01
-7.39315271e-01 8.46582115e-01 3.18489611e-01 -4.03446048e-01
-1.29956484e+00 5.88036180e-01 -5.23087308e-02 -8.17457259e-01
-6.10361218e-01 9.25246894e-01 4.42906842e-02 -4.07708555e-01
4.35989410e-01 -7.92035103e-01 -3.70626291e-03 7.86609575e-02
8.37345958e-01 6.17777288e-01 3.14332426e-01 -1.50744170e-01
-5.87001070e-02 2.79005915e-01 -3.93003285e-01 1.51053414e-01
1.52091897e+00 6.69281110e-02 -1.96444571e-01 9.14168656e-02
1.51811516e+00 -5.35098389e-02 -9.97549534e-01 -4.45096493e-01
-2.75144637e-01 -4.80057389e-01 3.95099550e-01 -4.01086181e-01
-1.43125927e+00 1.16404986e+00 2.43655890e-01 -1.73928753e-01
9.73415315e-01 -3.37940603e-01 1.20646095e+00 1.33319974e+00
4.80221659e-01 -1.04855835e+00 -6.28218800e-02 7.05127835e-01
9.74383175e-01 -1.05912018e+00 2.22981900e-01 -4.34608042e-01
-4.89219934e-01 8.92404497e-01 4.88754332e-01 -1.09748825e-01
6.81856692e-01 2.11101323e-01 -6.62635505e-01 -7.02489242e-02
-1.36587691e+00 1.04769759e-01 2.56701499e-01 2.69524604e-01
2.37993032e-01 1.55181244e-01 -2.14214846e-01 7.24200487e-01
-5.60445726e-01 -7.81942904e-02 2.20154688e-01 5.35703778e-01
-4.46763545e-01 -8.74493718e-01 6.16939832e-03 8.46183658e-01
-2.11596206e-01 -5.05087078e-01 -1.47268446e-02 6.28429234e-01
-3.89847308e-02 9.51996446e-01 1.75588220e-01 -5.58365405e-01
1.94207087e-01 1.48361260e-02 3.19477022e-01 -5.62071264e-01
-3.12380373e-01 -3.22526842e-01 2.12733105e-01 -7.78324783e-01
3.79597954e-02 -2.09479898e-01 -1.27480412e+00 -7.24591732e-01
-3.90110105e-01 1.32137388e-01 7.85292625e-01 7.25497782e-01
7.25531876e-01 3.15618366e-01 2.06111774e-01 -6.74447834e-01
-8.44376624e-01 -1.16310179e+00 -3.61441314e-01 1.28783017e-01
5.63241960e-03 -4.27381247e-01 -4.11878109e-01 -1.43596321e-01]
|
[8.708028793334961, 3.601494073867798]
|
08e7b326-6f4a-49d3-96f0-53a1c65ae0f6
|
first-explore-then-exploit-meta-learning
|
2307.02276
| null |
https://arxiv.org/abs/2307.02276v1
|
https://arxiv.org/pdf/2307.02276v1.pdf
|
First-Explore, then Exploit: Meta-Learning Intelligent Exploration
|
Standard reinforcement learning (RL) agents never intelligently explore like a human (i.e. by taking into account complex domain priors and previous explorations). Even the most basic intelligent exploration strategies such as exhaustive search are only inefficiently or poorly approximated by approaches such as novelty search or intrinsic motivation, let alone more complicated strategies like learning new skills, climbing stairs, opening doors, or conducting experiments. This lack of intelligent exploration limits sample efficiency and prevents solving hard exploration domains. We argue a core barrier prohibiting many RL approaches from learning intelligent exploration is that the methods attempt to explore and exploit simultaneously, which harms both exploration and exploitation as the goals often conflict. We propose a novel meta-RL framework (First-Explore) with two policies: one policy learns to only explore and one policy learns to only exploit. Once trained, we can then explore with the explore policy, for as long as desired, and then exploit based on all the information gained during exploration. This approach avoids the conflict of trying to do both exploration and exploitation at once. We demonstrate that First-Explore can learn intelligent exploration strategies such as exhaustive search and more, and that it outperforms dominant standard RL and meta-RL approaches on domains where exploration requires sacrificing reward. First-Explore is a significant step towards creating meta-RL algorithms capable of learning human-level exploration which is essential to solve challenging unseen hard-exploration domains.
|
['Jeff Clune', 'Ben Norman']
|
2023-07-05
| null | null | null | null |
['meta-learning', 'reinforcement-learning-1']
|
['methodology', 'methodology']
|
[ 9.38775390e-02 5.30947924e-01 -6.22570455e-01 2.22339749e-01
-7.75151849e-01 -8.38616192e-01 6.14344954e-01 3.26164998e-02
-1.01844966e+00 1.31877553e+00 -5.92992157e-02 -3.95632297e-01
-2.73569763e-01 -8.72060359e-01 -8.45139802e-01 -8.74432504e-01
-4.15659636e-01 8.11981082e-01 6.65841624e-02 -2.77959973e-01
4.74563658e-01 3.69514734e-01 -1.59540069e+00 -4.81255919e-01
1.02381861e+00 5.80911815e-01 4.18474704e-01 6.68493927e-01
-1.68345407e-01 7.51211643e-01 -4.54411119e-01 8.43145177e-02
3.25691551e-01 -5.59461713e-01 -9.46368873e-01 -2.71234870e-01
-3.97933573e-01 -4.09347773e-01 -4.45536561e-02 8.47263694e-01
4.73491967e-01 6.84920430e-01 3.30740988e-01 -9.84951437e-01
-4.85899478e-01 7.57116497e-01 -7.02658474e-01 2.46845171e-01
4.42677945e-01 6.47560477e-01 8.82850885e-01 -2.82862425e-01
5.52193403e-01 1.19281459e+00 1.98114619e-01 1.01880646e+00
-1.24588394e+00 -5.56695163e-01 6.06589794e-01 -7.10003823e-02
-6.51691437e-01 -2.19178721e-01 6.35872066e-01 1.15962900e-01
1.08833766e+00 2.05301106e-01 1.10325062e+00 1.48760009e+00
-2.49936637e-02 1.29855216e+00 1.39057422e+00 -4.00559276e-01
8.84992898e-01 -4.83059557e-03 -3.36586028e-01 6.51947439e-01
3.12899977e-01 9.40305471e-01 -6.17059827e-01 -1.87466577e-01
8.21555972e-01 -1.73387334e-01 -9.78992358e-02 -7.04686105e-01
-1.15690160e+00 9.74830449e-01 3.26270580e-01 2.28200957e-01
-5.51779032e-01 3.97340149e-01 3.43653917e-01 6.69337392e-01
-1.06131159e-01 1.32667494e+00 -6.22974932e-01 -6.09675705e-01
-8.14369678e-01 5.58042645e-01 7.28386402e-01 6.55508280e-01
1.13139796e+00 1.96246833e-01 1.25554025e-01 4.68740404e-01
1.48969710e-01 3.19917947e-01 9.26090062e-01 -1.23021400e+00
3.74139428e-01 4.45404083e-01 5.39686263e-01 -2.84970909e-01
-5.31510472e-01 -7.00268209e-01 -3.24825794e-01 8.53381097e-01
3.05941612e-01 -5.29668212e-01 -7.85736263e-01 2.05477405e+00
4.40134913e-01 -1.64742962e-01 3.39925706e-01 7.48833060e-01
1.42035022e-01 3.90713841e-01 1.99834466e-01 -4.64407265e-01
9.17804718e-01 -1.17432797e+00 -3.73652548e-01 -5.96136928e-01
7.22669184e-01 3.60825062e-02 1.32031488e+00 7.78152227e-01
-1.53234220e+00 -3.59475702e-01 -1.04257762e+00 2.42313191e-01
-3.68332624e-01 -5.03861308e-01 9.15100753e-01 6.38373017e-01
-9.44844902e-01 8.87436330e-01 -9.65233445e-01 -1.06289029e-01
6.15662754e-01 4.07098562e-01 -1.21711515e-01 1.14928268e-01
-1.02190113e+00 1.01546323e+00 7.23008454e-01 -2.38481849e-01
-1.38931119e+00 -3.98484409e-01 -7.96019971e-01 3.88686806e-02
1.00766611e+00 -5.35893023e-01 1.42979228e+00 -1.04735160e+00
-1.86750638e+00 5.27910709e-01 3.51835117e-02 -7.98212171e-01
7.79558718e-01 -6.55234814e-01 -1.21475768e-03 -7.79639184e-02
-7.13766692e-03 1.01244926e+00 6.88488007e-01 -1.28535795e+00
-6.44299328e-01 -2.25574046e-01 3.80093396e-01 7.73652613e-01
-1.12374976e-01 -5.00361383e-01 -1.89328581e-01 -3.49426895e-01
-1.12403549e-01 -9.82210159e-01 -6.42712057e-01 -2.12335020e-01
-2.12789118e-01 -2.36484692e-01 5.29154718e-01 1.83671769e-02
9.45343435e-01 -1.87395561e+00 3.03994447e-01 2.39278674e-01
8.81309733e-02 2.76598662e-01 -5.25361478e-01 3.57670456e-01
2.69233257e-01 1.62999853e-01 -1.10254228e-01 -1.26086026e-01
1.64981022e-01 4.96689796e-01 -3.69775891e-01 1.52872667e-01
-7.71885887e-02 1.34294486e+00 -1.54201710e+00 -1.69698864e-01
7.82870129e-02 2.96116676e-02 -7.88874686e-01 3.88194531e-01
-8.81686211e-01 8.50311875e-01 -7.27573574e-01 7.63102949e-01
2.93783873e-01 -2.56479114e-01 2.88173318e-01 7.16028035e-01
-1.96590900e-01 4.47245777e-01 -1.26629531e+00 1.71353471e+00
-5.87419212e-01 2.88719594e-01 -8.06171596e-02 -9.98783469e-01
8.39721918e-01 1.26259960e-02 3.11045289e-01 -1.17635751e+00
-1.45916611e-01 4.40770328e-01 -8.24108571e-02 -2.88465530e-01
2.98683316e-01 -1.77735046e-01 -6.89387135e-03 8.42959166e-01
-1.12994522e-01 -2.38302916e-01 2.79150784e-01 -1.20456792e-01
1.25828600e+00 7.82893836e-01 5.94745100e-01 -1.84777990e-01
3.04140598e-01 1.61348149e-01 6.73546731e-01 1.30241084e+00
-2.03329533e-01 -1.34217963e-01 5.84444463e-01 -6.58862472e-01
-7.87542522e-01 -1.22909629e+00 3.50874394e-01 1.53628898e+00
2.17999324e-01 -2.18894240e-02 -4.44836944e-01 -9.97905374e-01
-2.69389469e-02 1.03929842e+00 -7.45029569e-01 -2.65816778e-01
-7.99410224e-01 -5.77064335e-01 3.48751098e-01 4.77314889e-01
5.89222074e-01 -1.75387084e+00 -1.55257201e+00 4.26019877e-01
2.16315344e-01 -2.30237648e-01 -7.18816146e-02 9.91899073e-01
-1.13491881e+00 -9.63511407e-01 -7.43568659e-01 -4.25050616e-01
4.86088425e-01 -4.98233475e-02 1.31053483e+00 3.54883909e-01
-2.09239990e-01 5.56095898e-01 -2.62430966e-01 -2.26442397e-01
-1.55810609e-01 2.17805669e-01 1.46227300e-01 -8.26582611e-01
3.43902797e-01 -8.88249874e-01 -7.06268191e-01 1.72862992e-01
-6.46661818e-01 -2.37661213e-01 9.53588128e-01 1.07759213e+00
5.82301199e-01 2.62745529e-01 7.63069928e-01 -8.75670552e-01
8.63090396e-01 -6.44809604e-01 -8.00596952e-01 2.62741089e-01
-8.38176847e-01 5.83733320e-01 4.81696814e-01 -8.05707574e-01
-1.03148949e+00 -1.39088646e-01 -6.77535608e-02 -2.34154597e-01
-2.14029700e-01 2.55415022e-01 5.29480167e-02 -1.53835922e-01
8.75763476e-01 5.21557391e-01 -8.65582451e-02 -4.38916147e-01
3.68317485e-01 -8.24438483e-02 3.18913430e-01 -1.02726007e+00
8.19631696e-01 1.96632981e-01 -4.61920872e-02 -5.83422601e-01
-7.87702858e-01 -1.00507811e-01 -2.91790515e-01 5.51525876e-02
4.85217959e-01 -6.82260573e-01 -1.11381829e+00 -1.47921622e-01
-4.83978599e-01 -9.63979840e-01 -9.63030636e-01 4.45667922e-01
-1.10302973e+00 2.15187579e-01 -1.96429580e-01 -1.22574592e+00
-6.24183044e-02 -1.09367371e+00 6.10072613e-01 6.84172332e-01
-3.75004947e-01 -1.01049685e+00 4.30359513e-01 -2.46334262e-02
5.56376874e-01 2.75363207e-01 7.90384054e-01 -6.32727206e-01
-8.60947490e-01 4.55311716e-01 3.76312286e-01 -3.01535696e-01
-3.76605168e-02 -6.24382675e-01 -5.64868271e-01 -6.42095029e-01
-5.99831417e-02 -1.15683103e+00 9.69734430e-01 1.56324551e-01
1.03749013e+00 -4.95783567e-01 -4.28561866e-01 6.64341092e-01
1.29164052e+00 4.38579828e-01 4.87883925e-01 1.03496563e+00
6.52706996e-02 4.56833780e-01 8.26343536e-01 7.13475108e-01
2.14840546e-01 2.03787938e-01 6.64268196e-01 2.15121463e-01
4.79644984e-01 -6.19421601e-01 3.75656575e-01 -5.78026474e-02
-1.85710460e-01 -1.24165274e-01 -6.97870731e-01 5.84697247e-01
-1.80987549e+00 -9.67354178e-01 8.55434835e-01 2.22002316e+00
1.11663008e+00 5.22615314e-01 4.00788456e-01 -8.97262916e-02
-6.99675828e-02 1.41544655e-01 -1.41019583e+00 -7.53405035e-01
5.39668696e-03 4.30330247e-01 3.13036680e-01 6.49838030e-01
-8.28402519e-01 1.23216271e+00 6.73763371e+00 5.39932132e-01
-7.87634134e-01 -1.73138186e-01 4.52501833e-01 -5.55351198e-01
-6.77725613e-01 1.71724752e-01 -7.60899484e-01 1.77365571e-01
7.06090569e-01 9.04880092e-02 1.09571707e+00 1.14766932e+00
-2.19376177e-01 -5.58254600e-01 -1.15097344e+00 5.20329535e-01
-4.73244399e-01 -1.08244479e+00 -1.46995321e-01 2.25026965e-01
7.73513615e-01 -2.59421160e-03 2.65586704e-01 8.94530475e-01
1.13646770e+00 -1.27370965e+00 5.18904865e-01 4.06299502e-01
2.95208007e-01 -8.53231251e-01 2.27459833e-01 1.02983677e+00
-7.83254027e-01 -6.20454609e-01 -4.01472092e-01 -1.02840222e-01
-8.04407224e-02 1.75050318e-01 -5.63601911e-01 7.59021118e-02
6.23139739e-01 3.09158444e-01 -2.15646684e-01 9.33580101e-01
-6.08815551e-01 3.09674978e-01 -3.69497985e-01 -4.54145193e-01
7.79278636e-01 -1.57731533e-01 6.35755658e-01 7.69617260e-01
2.92028010e-01 7.18019083e-02 3.53287101e-01 1.09208584e+00
2.25502551e-01 -1.51781619e-01 -6.24003589e-01 -3.36200565e-01
4.98540640e-01 8.32639873e-01 -4.97483701e-01 -1.47727907e-01
-7.54387975e-02 8.16549480e-01 8.39454114e-01 5.16747773e-01
-4.71798599e-01 4.37904894e-03 4.39152271e-01 -2.00100064e-01
4.44864661e-01 -3.15351516e-01 -7.50856474e-02 -9.91029978e-01
-2.35779986e-01 -1.16102219e+00 5.14861822e-01 -3.82799566e-01
-9.03509617e-01 4.56941396e-01 -3.49503756e-02 -8.39084089e-01
-6.40926838e-01 -3.83887917e-01 -6.06114566e-01 7.74892807e-01
-2.03296161e+00 -6.48041427e-01 8.64254981e-02 4.63190675e-01
7.55105257e-01 -3.38016629e-01 7.52550840e-01 -6.61477029e-01
-3.04452747e-01 5.42725444e-01 2.86680106e-02 -4.87222791e-01
1.90204084e-01 -1.53080821e+00 2.89501071e-01 3.05822879e-01
1.49902806e-01 7.82692730e-01 6.82901680e-01 -7.42025495e-01
-1.49275386e+00 -2.57955253e-01 2.91993648e-01 -1.99615151e-01
4.47451949e-01 -2.43475184e-01 -8.63122404e-01 5.98287106e-01
1.16133973e-01 -1.85681388e-01 4.43082273e-01 4.87128407e-01
-1.57479644e-01 3.93894017e-01 -1.11685538e+00 9.18420970e-01
1.08202517e+00 -1.42067730e-01 -8.45343530e-01 6.94942921e-02
5.88384986e-01 -4.04424399e-01 -2.83373863e-01 3.06722194e-01
5.56380868e-01 -1.20459139e+00 1.03319860e+00 -7.21951067e-01
8.67966563e-02 -6.47066336e-04 8.96457583e-02 -1.47154248e+00
-2.38243714e-01 -1.19953775e+00 -6.72422767e-01 5.36868870e-01
3.97962183e-01 -7.66969025e-01 1.18931830e+00 3.82692039e-01
2.97175288e-01 -1.16111600e+00 -9.17126238e-01 -1.09057701e+00
4.20098037e-01 -5.45418225e-02 7.06331611e-01 6.13489866e-01
2.25510389e-01 -8.23802799e-02 -4.01782960e-01 -1.67199895e-01
7.39412189e-01 2.78527766e-01 7.94381440e-01 -1.00156963e+00
-7.30346560e-01 -6.38815939e-01 4.40433919e-01 -1.31503642e+00
1.85017809e-01 -3.38758975e-01 1.64277583e-01 -1.19826841e+00
3.31413746e-02 -7.84386933e-01 -4.75877404e-01 5.29266119e-01
-1.60566211e-01 -3.47917080e-01 -3.83284353e-02 2.54375011e-01
-9.26607668e-01 7.93574810e-01 1.55861831e+00 5.48957810e-02
-7.84484565e-01 2.36649774e-02 -9.41854417e-01 7.89146960e-01
1.01221383e+00 -4.27645743e-01 -7.58668542e-01 -6.74153939e-02
6.32127643e-01 2.31256291e-01 1.89422414e-01 -1.02655995e+00
1.64003000e-01 -6.79701149e-01 5.96596777e-01 -3.34878147e-01
2.69645810e-01 -5.78800678e-01 -2.74688721e-01 8.22659075e-01
-6.85279965e-01 2.94913957e-03 2.49357745e-01 7.65679955e-01
2.39961982e-01 -4.75075632e-01 7.73525834e-01 -8.39175582e-01
-7.62877882e-01 1.45575136e-01 -4.23685074e-01 4.36652809e-01
1.14874530e+00 -4.44824278e-01 -1.44175768e-01 -3.79951119e-01
-9.60524738e-01 7.91264176e-01 5.47159195e-01 1.80732265e-01
4.69549447e-01 -1.00094521e+00 -3.22777689e-01 2.13351086e-01
-5.06375134e-02 1.64934009e-01 4.35000584e-02 4.41317290e-01
-9.38657597e-02 3.46066236e-01 -4.47407424e-01 -1.42320514e-01
-7.28677630e-01 9.90489483e-01 4.12376136e-01 -9.37296212e-01
-8.07813108e-01 9.67007518e-01 1.99826464e-01 -6.32276773e-01
5.94220579e-01 -2.46596597e-02 -2.55066067e-01 -1.72529414e-01
5.58097124e-01 4.05499220e-01 -5.96885502e-01 2.06406593e-01
-1.11407727e-01 4.84692276e-01 -3.01480472e-01 -4.55176294e-01
1.36704433e+00 -1.35057271e-01 3.40013504e-01 4.05997217e-01
6.14159882e-01 -1.48160562e-01 -1.76130664e+00 -2.33796015e-01
2.17643395e-01 -5.63818455e-01 -8.76110420e-02 -1.19305122e+00
-5.12071490e-01 8.79920363e-01 3.57600540e-01 1.01259783e-01
1.08944011e+00 -6.51236027e-02 6.63622797e-01 1.08130538e+00
8.17198813e-01 -1.51858521e+00 7.17054069e-01 4.56984341e-01
7.39940107e-01 -1.24173784e+00 1.20688088e-01 5.20453334e-01
-8.63758326e-01 8.98932815e-01 9.65091288e-01 -2.90046364e-01
1.52791277e-01 1.62799090e-01 -2.54301846e-01 -1.37065768e-01
-1.05192780e+00 -3.77915323e-01 1.87275968e-02 7.84761012e-01
-3.48903984e-02 -9.06767175e-02 -8.89919698e-03 1.66076437e-01
-2.16919348e-01 6.10926710e-02 1.84224963e-01 1.33817184e+00
-9.47448313e-01 -1.32520187e+00 -2.64018446e-01 2.42942736e-01
-2.50949979e-01 1.80948824e-01 -2.27879345e-01 9.14481223e-01
5.78928478e-02 5.80420196e-01 -1.31680623e-01 4.85420264e-02
-3.84653993e-02 -8.05267245e-02 6.51280105e-01 -5.03303289e-01
-5.92296958e-01 5.14093190e-02 -1.04286432e-01 -9.41702008e-01
-1.49001092e-01 -7.18370438e-01 -1.32846653e+00 -4.32036668e-02
-8.99489075e-02 4.80391979e-01 3.88982773e-01 9.66535151e-01
4.10321765e-02 2.63288051e-01 6.51843071e-01 -8.03998411e-01
-1.12171614e+00 -4.85575974e-01 -6.18219852e-01 -9.87723917e-02
7.34919965e-01 -8.97116065e-01 -5.12144864e-01 -6.90675557e-01]
|
[3.9212708473205566, 1.7372854948043823]
|
2b7abf0b-33a5-4c79-aa04-168359ce33a3
|
part-based-pseudo-label-refinement-for
|
2203.14675
| null |
https://arxiv.org/abs/2203.14675v1
|
https://arxiv.org/pdf/2203.14675v1.pdf
|
Part-based Pseudo Label Refinement for Unsupervised Person Re-identification
|
Unsupervised person re-identification (re-ID) aims at learning discriminative representations for person retrieval from unlabeled data. Recent techniques accomplish this task by using pseudo-labels, but these labels are inherently noisy and deteriorate the accuracy. To overcome this problem, several pseudo-label refinement methods have been proposed, but they neglect the fine-grained local context essential for person re-ID. In this paper, we propose a novel Part-based Pseudo Label Refinement (PPLR) framework that reduces the label noise by employing the complementary relationship between global and part features. Specifically, we design a cross agreement score as the similarity of k-nearest neighbors between feature spaces to exploit the reliable complementary relationship. Based on the cross agreement, we refine pseudo-labels of global features by ensembling the predictions of part features, which collectively alleviate the noise in global feature clustering. We further refine pseudo-labels of part features by applying label smoothing according to the suitability of given labels for each part. Thanks to the reliable complementary information provided by the cross agreement score, our PPLR effectively reduces the influence of noisy labels and learns discriminative representations with rich local contexts. Extensive experimental results on Market-1501 and MSMT17 demonstrate the effectiveness of the proposed method over the state-of-the-art performance. The code is available at https://github.com/yoonkicho/PPLR.
|
['Sung-Eui Yoon', 'Seunghoon Hong', 'Woo Jae Kim', 'Yoonki Cho']
|
2022-03-28
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Cho_Part-Based_Pseudo_Label_Refinement_for_Unsupervised_Person_Re-Identification_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Cho_Part-Based_Pseudo_Label_Refinement_for_Unsupervised_Person_Re-Identification_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['person-retrieval', 'unsupervised-person-re-identification']
|
['computer-vision', 'computer-vision']
|
[-1.20225407e-01 -2.41724104e-01 -2.99666356e-02 -6.86209917e-01
-8.39361191e-01 -4.31590468e-01 6.25939310e-01 1.71208426e-01
-4.69415277e-01 6.21400177e-01 5.24384916e-01 5.92575610e-01
-3.28771144e-01 -5.99582314e-01 -2.63457537e-01 -9.15230155e-01
3.11114103e-01 4.31058675e-01 -1.33613840e-01 6.80765286e-02
-1.95264425e-02 1.58571839e-01 -1.76998127e+00 4.99220341e-02
9.98664796e-01 7.59523511e-01 -1.88156217e-02 1.32698510e-02
-1.88922714e-02 2.08989248e-01 -3.63629401e-01 -3.13285172e-01
2.77736217e-01 -3.78049403e-01 -7.73468256e-01 2.53099740e-01
3.29529017e-01 -2.04333067e-01 -3.84660602e-01 1.15575910e+00
7.07742274e-01 4.67885345e-01 8.34657669e-01 -9.74740982e-01
-6.78601682e-01 2.42611304e-01 -5.87337971e-01 -6.15815585e-03
5.33693433e-01 -1.26543030e-01 1.01195526e+00 -1.13636136e+00
3.58231574e-01 1.32438385e+00 7.60071874e-01 7.61740029e-01
-1.33594680e+00 -8.18021894e-01 3.95361722e-01 3.35614145e-01
-1.99455619e+00 -4.23943877e-01 8.20768774e-01 -4.31725293e-01
4.54540819e-01 2.62593031e-01 2.94661939e-01 8.52819502e-01
-4.97535318e-01 7.34949946e-01 1.24534440e+00 -4.86510158e-01
8.56577605e-03 2.66961694e-01 6.72255933e-01 7.65672684e-01
5.92039935e-02 -4.56484221e-03 -4.59535152e-01 -3.46920580e-01
5.45570970e-01 5.31905472e-01 -2.72557765e-01 -3.34702551e-01
-1.04153466e+00 6.13528073e-01 5.27706742e-01 4.13095027e-01
-1.40381768e-01 -2.21634448e-01 3.40777695e-01 2.82222703e-02
4.86178249e-01 -4.75940964e-04 -1.99984506e-01 2.77387589e-01
-7.79384255e-01 1.55297294e-01 4.15138662e-01 9.03719366e-01
1.10856330e+00 -5.70292890e-01 -4.77247477e-01 1.27399349e+00
4.46445614e-01 4.44906414e-01 6.50398195e-01 -7.35633433e-01
2.21596569e-01 9.10660028e-01 2.00248748e-01 -8.79251063e-01
-4.99035984e-01 -5.68985105e-01 -9.86810029e-01 -3.13569427e-01
4.92397487e-01 1.05583027e-01 -9.79345560e-01 1.80107021e+00
5.40480673e-01 3.33616376e-01 -1.19116858e-01 1.05313802e+00
9.31089342e-01 3.24194074e-01 2.97503591e-01 -1.50435492e-01
1.51967227e+00 -9.16883945e-01 -6.39493287e-01 1.08401895e-01
7.66776264e-01 -4.37722266e-01 7.21228004e-01 3.37187457e-03
-4.33448970e-01 -9.55122828e-01 -7.54419446e-01 8.42173994e-02
-3.76676202e-01 4.63544577e-01 3.77654165e-01 8.07103038e-01
-8.95099938e-01 3.97892058e-01 -5.30894399e-01 -3.45694363e-01
3.13453406e-01 5.95163167e-01 -5.82875252e-01 -3.59987408e-01
-1.27341652e+00 4.15167451e-01 3.50520164e-01 3.32820982e-01
-3.70148033e-01 -4.04052824e-01 -8.10923219e-01 5.04577253e-03
3.39209646e-01 -4.84041899e-01 8.41693759e-01 -6.99170351e-01
-1.26663530e+00 8.62698078e-01 -5.31982183e-01 -5.11468463e-02
2.06530854e-01 -2.38438651e-01 -5.40880024e-01 1.91304430e-01
3.90501320e-01 6.17123008e-01 7.82374144e-01 -1.39699388e+00
-6.10173225e-01 -5.82590938e-01 -2.77772129e-01 3.86517406e-01
-4.96467799e-01 -2.74920035e-02 -6.26407623e-01 -6.56168044e-01
3.00717205e-01 -1.00463498e+00 -3.13124895e-01 -3.61315787e-01
-2.78250515e-01 -8.74667346e-01 4.49134678e-01 -7.36884475e-01
1.20064640e+00 -2.23475933e+00 -1.41690835e-01 6.05931640e-01
3.11208606e-01 2.57833451e-01 -1.34775609e-01 1.52507961e-01
-1.73410997e-02 6.40756916e-04 -1.19935043e-01 -6.52932048e-01
-2.04590196e-03 4.94082868e-02 6.39819959e-03 5.85127056e-01
-1.35397594e-02 7.72384346e-01 -8.52040529e-01 -6.94987774e-01
3.14024419e-01 5.38496077e-01 -2.62652248e-01 2.57968664e-01
3.83673698e-01 7.87158728e-01 -7.87418962e-01 7.57882655e-01
7.62007952e-01 -2.19587713e-01 2.18242288e-01 -2.51871735e-01
9.01209041e-02 -4.37540896e-02 -1.40766978e+00 1.62585151e+00
7.05337226e-02 -1.99295416e-01 -1.66802987e-01 -1.11464965e+00
1.16243446e+00 2.62690783e-01 5.99062562e-01 -6.22729003e-01
6.64877668e-02 2.44006157e-01 -5.39769471e-01 -2.45450318e-01
2.93075532e-01 -9.60977341e-04 -2.74219871e-01 3.65867078e-01
2.10316535e-02 6.09920740e-01 1.54402032e-01 1.75729960e-01
6.78667843e-01 3.19528520e-01 3.16783339e-01 -2.50287473e-01
1.03288949e+00 -4.56862956e-01 8.58649731e-01 7.62970328e-01
-4.69947755e-01 8.43331277e-01 -2.58490086e-01 -4.58283931e-01
-6.28043294e-01 -9.23672616e-01 -2.52440929e-01 1.18870056e+00
4.78498697e-01 -7.01833069e-01 -7.81663239e-01 -1.07981288e+00
9.10886601e-02 2.84031749e-01 -6.69605434e-01 -1.66296050e-01
-5.85215032e-01 -7.57313848e-01 4.55356300e-01 5.05291939e-01
6.89721048e-01 -8.25575352e-01 1.79339558e-01 1.21749915e-01
-4.48768407e-01 -8.38470221e-01 -8.29137564e-01 -8.52383748e-02
-6.18178308e-01 -9.00434196e-01 -9.70720887e-01 -1.05970562e+00
1.01487112e+00 6.27517939e-01 7.15334415e-01 3.82413149e-01
-8.75775740e-02 5.82622170e-01 -5.63815653e-01 3.22399378e-01
9.36634913e-02 4.16351482e-02 5.13406336e-01 3.73538524e-01
8.47725272e-01 -3.74342471e-01 -6.40380979e-01 6.37741804e-01
-5.70817292e-01 -2.65153825e-01 4.08952057e-01 1.07112134e+00
8.80829513e-01 2.72186726e-01 7.07404852e-01 -7.65068471e-01
3.28835994e-01 -4.36750174e-01 -1.90856129e-01 4.74524856e-01
-7.03828096e-01 2.35763177e-01 4.31090057e-01 -4.33900684e-01
-1.28948164e+00 2.29242742e-01 8.96131713e-03 -2.21922815e-01
-4.77596700e-01 2.41989791e-01 -4.15572077e-01 -6.41698763e-02
3.47394586e-01 3.14149350e-01 -1.57085642e-01 -8.73601377e-01
3.69282365e-01 8.72934818e-01 5.33635199e-01 -8.09161484e-01
8.89787316e-01 4.83373374e-01 -1.72423065e-01 -4.37772661e-01
-1.01163924e+00 -1.05548155e+00 -8.25085342e-01 -1.59365952e-01
6.98343575e-01 -1.20063043e+00 -6.77977622e-01 4.31494802e-01
-6.36173725e-01 2.31233314e-01 -2.33788535e-01 5.04405379e-01
-1.66625604e-01 9.10717368e-01 -3.94538671e-01 -8.17006946e-01
-4.81188446e-01 -9.25733626e-01 1.13573790e+00 7.25049198e-01
-2.36683890e-01 -6.62950933e-01 9.17541534e-02 5.34130573e-01
3.34588662e-02 -1.21641345e-01 4.99289542e-01 -8.38072538e-01
-2.42508829e-01 -3.75146270e-01 -5.35457015e-01 2.12270230e-01
3.51300031e-01 -6.45427346e-01 -9.68840778e-01 -5.38247883e-01
-2.32155338e-01 -2.65874922e-01 1.09657109e+00 1.17128521e-01
1.03914773e+00 -8.28794762e-02 -6.02814436e-01 4.00726259e-01
1.14069188e+00 -1.74219772e-01 2.99875736e-01 2.98000485e-01
9.57294345e-01 8.30389678e-01 7.67528594e-01 6.36708260e-01
5.74213088e-01 8.69676292e-01 -1.66934744e-01 9.30140764e-02
-1.99484885e-01 -3.57589513e-01 1.97739318e-01 6.11999691e-01
-3.36652845e-01 7.66304061e-02 -6.25950396e-01 4.91595775e-01
-2.03885341e+00 -8.03010404e-01 -9.67576355e-02 2.33545232e+00
8.77533674e-01 -2.80886412e-01 2.26268589e-01 2.34246194e-01
1.09914851e+00 -1.03112921e-01 -2.74756253e-01 2.84366488e-01
-9.94211137e-02 -6.47422597e-02 3.74929428e-01 2.96617806e-01
-1.36890340e+00 8.56559575e-01 4.81983805e+00 1.14822257e+00
-3.53979349e-01 2.61120677e-01 4.99355495e-01 1.54413745e-01
-9.81275961e-02 -1.02351993e-01 -1.24987364e+00 5.71032882e-01
5.78898489e-01 2.21652374e-01 2.79923171e-01 8.41123462e-01
4.78735566e-02 -1.15119286e-01 -1.00462067e+00 1.30562556e+00
2.20184162e-01 -7.14136124e-01 -6.74642846e-02 8.90535340e-02
7.47939110e-01 -4.90838945e-01 -6.00457154e-02 4.31151330e-01
2.19759822e-01 -7.23485589e-01 5.00571847e-01 8.95979524e-01
7.29984999e-01 -8.66519749e-01 9.76740599e-01 2.37671927e-01
-1.58309758e+00 -1.63337931e-01 -4.04070705e-01 1.31544292e-01
3.21964659e-02 6.00288451e-01 -5.85942686e-01 7.70168126e-01
8.97351325e-01 9.74148571e-01 -8.84784043e-01 9.96254146e-01
-2.69909739e-01 4.11541075e-01 -3.27500403e-01 2.89998025e-01
-2.39596456e-01 -3.02480608e-01 3.76092911e-01 1.19747591e+00
1.37473330e-01 2.44777650e-01 5.46732664e-01 6.49232209e-01
5.13795763e-03 2.24030763e-01 -1.05139360e-01 2.69941688e-01
5.66590130e-01 1.21479285e+00 -6.17637336e-01 -4.34950471e-01
-3.81106764e-01 1.14159906e+00 4.01795745e-01 5.33625185e-01
-4.84541327e-01 -1.46446899e-01 5.85678697e-01 -8.38475600e-02
1.66123629e-01 -9.80183110e-02 -1.80630922e-01 -1.18250549e+00
1.53942704e-01 -5.68009913e-01 7.73798525e-01 -2.22081020e-01
-1.69675088e+00 3.80665481e-01 1.01506405e-01 -1.41075647e+00
-1.63734809e-01 -1.86377972e-01 -1.59422070e-01 9.72250640e-01
-1.43340874e+00 -1.40264881e+00 -4.92862880e-01 7.26781428e-01
3.60526055e-01 -2.41106287e-01 1.01803088e+00 3.80497068e-01
-7.93833077e-01 9.93464231e-01 2.38444716e-01 2.64325798e-01
1.09718871e+00 -1.08120668e+00 -3.74617390e-02 6.74270332e-01
-1.66905653e-02 1.01037788e+00 3.26745778e-01 -8.49281788e-01
-8.41909826e-01 -1.08976126e+00 1.00794554e+00 -3.35865855e-01
1.00054152e-01 -2.73805022e-01 -9.48974729e-01 3.94259661e-01
-3.93874437e-01 2.14624062e-01 9.09267902e-01 3.60092044e-01
-6.74934685e-01 -2.78538764e-01 -1.27827764e+00 2.80342102e-01
1.26553643e+00 -7.16774166e-01 -5.43296099e-01 2.66319662e-01
5.36977589e-01 1.47029221e-01 -9.43929911e-01 4.83192503e-01
5.97868383e-01 -7.42843091e-01 1.16030657e+00 -1.06814869e-01
-4.18477297e-01 -7.94645369e-01 -3.55785117e-02 -9.14573491e-01
-8.38358283e-01 -1.48246735e-01 8.93604308e-02 1.82209766e+00
-1.52136326e-01 -7.22290933e-01 7.26806402e-01 9.13735271e-01
1.61236897e-01 -4.10278946e-01 -9.51018631e-01 -7.98373580e-01
-3.92747939e-01 -8.61970428e-03 6.64683044e-01 8.42564046e-01
-1.32757109e-02 1.46892399e-01 -5.00661910e-01 2.98552811e-01
8.81770194e-01 1.56640634e-01 5.87004006e-01 -1.48656237e+00
-2.01540679e-01 -1.87147528e-01 -4.58186209e-01 -1.09823203e+00
3.27430218e-01 -1.01859450e+00 1.29550427e-01 -1.31518364e+00
6.94383621e-01 -7.95777857e-01 -7.20648646e-01 7.17802405e-01
-6.19806588e-01 4.97428179e-01 3.51505950e-02 8.08470666e-01
-1.10888731e+00 6.82905257e-01 8.09521914e-01 -1.81704596e-01
-3.41968268e-01 1.00863120e-02 -7.26217210e-01 5.47453642e-01
8.03792000e-01 -4.90935534e-01 -9.28510055e-02 -3.96605134e-02
-2.89158314e-01 -5.47965169e-01 4.38307166e-01 -1.08031142e+00
2.69057274e-01 2.11884663e-01 7.85815716e-01 -5.77724934e-01
3.16030294e-01 -7.64894605e-01 1.65535107e-01 1.86457649e-01
-3.84454906e-01 -4.34304893e-01 -3.33419621e-01 8.55843186e-01
-1.29739255e-01 -2.60232627e-01 6.34140909e-01 -4.65340652e-02
-8.98665965e-01 2.77402699e-01 -4.39770743e-02 -2.92615473e-01
7.68313825e-01 -2.27893427e-01 -1.02769844e-01 -1.26459807e-01
-8.79538119e-01 4.24671948e-01 4.68206912e-01 3.30286771e-01
6.05848372e-01 -1.52211356e+00 -6.71911240e-01 4.04513478e-01
4.21762675e-01 -2.23750472e-01 5.47649980e-01 7.31795251e-01
2.80489236e-01 3.67158979e-01 5.75927719e-02 -5.61596274e-01
-1.39930642e+00 6.65751219e-01 1.91848725e-01 -3.48483801e-01
-6.21607482e-01 8.16925287e-01 4.49106961e-01 -5.46815157e-01
2.75837034e-01 2.35556290e-01 -6.91714287e-01 7.09758922e-02
7.79956043e-01 4.28217232e-01 -1.23986103e-01 -1.07635128e+00
-5.46894014e-01 9.16584432e-01 -4.15625781e-01 6.62309006e-02
1.11312222e+00 -5.58132112e-01 -8.95768106e-02 1.36304438e-01
1.12360740e+00 -1.57101697e-03 -1.21729612e+00 -8.11521053e-01
8.04433376e-02 -5.01914382e-01 -2.83972502e-01 -5.27612567e-01
-8.41695249e-01 3.74795407e-01 9.71827507e-01 -1.80705220e-01
1.17412889e+00 2.09226027e-01 8.11347663e-01 1.85552135e-01
4.87058282e-01 -1.23073363e+00 2.08404548e-02 2.26841778e-01
4.35918480e-01 -1.31130052e+00 8.38981476e-03 -5.82712531e-01
-5.02869785e-01 7.72768319e-01 5.37400484e-01 -4.18913141e-02
5.96427441e-01 -3.82287681e-01 -6.19037710e-02 9.56385583e-02
5.01601696e-02 -6.31147742e-01 5.64939916e-01 7.29107678e-01
2.68148065e-01 2.89261729e-01 -5.68329751e-01 1.09232390e+00
1.43998325e-01 -7.89205134e-02 -1.85617059e-01 7.33224452e-01
-4.54099715e-01 -1.38200784e+00 -6.52770996e-01 3.92173111e-01
-2.22980902e-01 8.11809897e-02 -2.68079877e-01 2.80635864e-01
5.14432132e-01 1.27432823e+00 -2.43700176e-01 -5.86617649e-01
1.47924319e-01 2.34942168e-01 2.35890865e-01 -5.57173193e-01
-4.27203387e-01 2.60856360e-01 4.54045497e-02 -3.55959207e-01
-6.95814550e-01 -7.91542470e-01 -1.38200140e+00 -1.20221414e-02
-3.54567766e-01 4.14394259e-01 2.25637376e-01 1.07412708e+00
3.88724834e-01 4.38473262e-02 7.62798309e-01 -8.32496703e-01
-5.52841604e-01 -1.05365396e+00 -7.85654902e-01 8.64096105e-01
1.07404917e-01 -9.64991570e-01 -3.49964082e-01 8.27647671e-02]
|
[14.814799308776855, 1.0656388998031616]
|
505081bf-4558-4903-82bd-5924ce453ef0
|
scanet-self-paced-semi-curricular-attention
|
2304.08444
| null |
https://arxiv.org/abs/2304.08444v1
|
https://arxiv.org/pdf/2304.08444v1.pdf
|
SCANet: Self-Paced Semi-Curricular Attention Network for Non-Homogeneous Image Dehazing
|
The presence of non-homogeneous haze can cause scene blurring, color distortion, low contrast, and other degradations that obscure texture details. Existing homogeneous dehazing methods struggle to handle the non-uniform distribution of haze in a robust manner. The crucial challenge of non-homogeneous dehazing is to effectively extract the non-uniform distribution features and reconstruct the details of hazy areas with high quality. In this paper, we propose a novel self-paced semi-curricular attention network, called SCANet, for non-homogeneous image dehazing that focuses on enhancing haze-occluded regions. Our approach consists of an attention generator network and a scene reconstruction network. We use the luminance differences of images to restrict the attention map and introduce a self-paced semi-curricular learning strategy to reduce learning ambiguity in the early stages of training. Extensive quantitative and qualitative experiments demonstrate that our SCANet outperforms many state-of-the-art methods. The code is publicly available at https://github.com/gy65896/SCANet.
|
['Wenqi Ren', 'Shengfeng He', 'Jingxiang Qu', 'Yuxu Lu', 'Ryan Wen Liu', 'Yuan Gao', 'Yu Guo']
|
2023-04-17
| null | null | null | null |
['image-dehazing']
|
['computer-vision']
|
[ 1.21027268e-01 -2.09211946e-01 2.26982147e-01 -1.53073877e-01
-5.39492071e-01 -1.31448194e-01 2.49338925e-01 -3.59060854e-01
-1.05015874e-01 7.06934988e-01 3.85712802e-01 -1.75270498e-01
-6.94611892e-02 -8.37696731e-01 -1.03112710e+00 -9.49886560e-01
4.16694611e-01 -9.07991678e-02 5.17552912e-01 -3.47367406e-01
3.21031451e-01 1.85280815e-01 -1.89437115e+00 4.46349621e-01
1.68029952e+00 6.91781461e-01 5.99100113e-01 7.37832248e-01
-3.18191461e-02 1.04499781e+00 -7.93412447e-01 -1.33621544e-01
2.45732963e-01 -5.55001557e-01 -5.16336977e-01 1.23285659e-01
1.07395875e+00 -6.30253553e-01 -6.21204078e-01 1.33160293e+00
6.83273375e-01 3.25048983e-01 5.31514049e-01 -9.88608658e-01
-1.33549201e+00 1.16974816e-01 -8.24071288e-01 5.99912405e-01
-7.69220069e-02 3.77518088e-01 2.82662839e-01 -9.59573686e-01
2.70728856e-01 1.13072634e+00 4.18318868e-01 6.34114802e-01
-1.01321375e+00 -8.93760800e-01 2.10030258e-01 5.08957684e-01
-1.49632037e+00 -3.44437599e-01 9.95043576e-01 -3.42961788e-01
3.18200380e-01 3.39962631e-01 7.68486619e-01 9.26449716e-01
5.29524148e-01 8.16192746e-01 1.36974716e+00 -4.11541313e-01
1.63889840e-01 1.29727945e-01 -1.93762600e-01 7.28015542e-01
5.55010080e-01 2.24826485e-01 -5.72740555e-01 5.01307130e-01
9.31807160e-01 1.76071778e-01 -5.59848130e-01 -3.44081819e-01
-7.67684102e-01 4.62918967e-01 6.19815111e-01 7.24945217e-02
-3.83166075e-01 1.55409276e-01 -3.42654407e-01 1.83221191e-01
9.35567439e-01 6.24483943e-01 -4.95876260e-02 2.24164516e-01
-8.34018469e-01 3.15262318e-01 1.48012623e-01 9.71055806e-01
9.62643027e-01 5.07553935e-01 -3.29559058e-01 7.42219627e-01
6.35529608e-02 5.37780821e-01 4.23816264e-01 -9.86731231e-01
2.12794885e-01 2.85499156e-01 2.02931091e-01 -7.43471801e-01
2.49541149e-01 -4.59393919e-01 -9.04364944e-01 6.54648006e-01
1.14622965e-01 -1.98251233e-01 -1.37673330e+00 1.38322902e+00
5.37587464e-01 5.18681288e-01 -7.71769211e-02 1.23166990e+00
9.95761633e-01 9.29507732e-01 -4.21723947e-02 1.77918762e-01
1.12521315e+00 -1.45391738e+00 -1.14560914e+00 -2.61529118e-01
-1.40470862e-01 -9.27446485e-01 1.29368019e+00 2.61477143e-01
-1.30490375e+00 -6.52689636e-01 -1.18809545e+00 -4.12731946e-01
-5.56693494e-01 -2.52834529e-01 2.59849250e-01 4.18095618e-01
-1.26314366e+00 4.25734192e-01 -5.56379855e-01 1.63189396e-01
7.15405107e-01 4.83721606e-02 -1.93637554e-02 -7.64387846e-01
-1.25722408e+00 8.87264311e-01 1.62626758e-01 6.21528178e-02
-1.17963111e+00 -1.24741828e+00 -1.16753519e+00 1.39551535e-01
3.58837157e-01 -8.20513427e-01 1.00984848e+00 -1.24859774e+00
-1.53854597e+00 5.63735008e-01 -7.85498843e-02 5.78850433e-02
3.90941352e-01 -5.26553273e-01 -3.17302853e-01 2.09558859e-01
8.44798386e-02 6.53341830e-01 1.31853116e+00 -1.69263923e+00
-4.89977717e-01 -3.28262635e-02 -5.50264120e-02 7.11763263e-01
-1.85138583e-01 -2.20111176e-01 -4.32413012e-01 -9.66487169e-01
-2.68377095e-01 -4.52027649e-01 -1.40288651e-01 3.72560062e-02
-2.78929681e-01 2.13261738e-01 9.45836842e-01 -7.77941465e-01
1.12900376e+00 -2.23890448e+00 6.52536601e-02 -3.92495960e-01
5.33720970e-01 4.56411481e-01 -2.08868295e-01 2.00492740e-02
-7.83349052e-02 -5.10486634e-03 -3.57345313e-01 -2.81173080e-01
-2.33295694e-01 1.54428920e-02 -2.18884051e-01 5.57854176e-01
4.83494878e-01 8.29813600e-01 -1.15567875e+00 -4.43645179e-01
5.95113575e-01 9.22386885e-01 -4.72220212e-01 6.21500909e-01
-2.02859715e-01 5.82472324e-01 -1.41411185e-01 6.49453163e-01
1.07438707e+00 -7.84078911e-02 -5.88065624e-01 2.95866877e-02
-4.76762652e-01 1.53466344e-01 -8.10328126e-01 1.45349753e+00
-4.51002985e-01 9.76001382e-01 1.91740900e-01 -5.78499675e-01
5.13276637e-01 1.26495555e-01 1.84341669e-01 -9.60586250e-01
7.26350257e-03 -4.26882617e-02 -2.41337776e-01 -7.49770224e-01
6.41148746e-01 -1.26270905e-01 5.47083676e-01 1.18138656e-01
-7.11639896e-02 -4.63732630e-01 -1.11026242e-01 -4.55486104e-02
6.93722963e-01 2.91611161e-02 -6.30936548e-02 -5.20589828e-01
3.74833018e-01 -1.55609384e-01 4.91879702e-01 6.82981133e-01
-2.14655325e-01 1.11600745e+00 5.16774282e-02 -5.44857681e-01
-1.13357401e+00 -1.16129029e+00 -2.91473921e-02 1.01665246e+00
5.77548623e-01 -6.75276220e-02 -8.81725192e-01 -3.43706816e-01
-1.68989956e-01 7.38442540e-01 -8.02458346e-01 -5.17036140e-01
-4.43830252e-01 -5.44631481e-01 -1.03534468e-01 2.30260622e-02
1.05853748e+00 -9.48068202e-01 -3.42300475e-01 -2.00189054e-02
-2.25679889e-01 -9.37738895e-01 -8.22480619e-01 -7.65898600e-02
-5.52953959e-01 -8.92081261e-01 -1.18679738e+00 -1.06317055e+00
9.71813440e-01 1.06738126e+00 1.14745259e+00 2.83015579e-01
-3.17833573e-01 1.33621320e-01 -1.93696722e-01 -7.74276376e-01
-1.27121985e-01 -1.70767173e-01 -3.00702035e-01 1.30420640e-01
1.64239705e-01 -5.84142447e-01 -1.15226352e+00 2.73253798e-01
-1.25505078e+00 4.65256661e-01 6.55574977e-01 8.15392673e-01
5.12439907e-01 6.33033991e-01 1.29410759e-01 -9.92481053e-01
3.10600668e-01 -5.79032421e-01 -6.78992629e-01 3.84300314e-02
-6.07655942e-01 -7.20299333e-02 4.80429232e-01 -6.48914814e-01
-1.52414811e+00 -2.46976450e-01 9.07299817e-02 -7.79116511e-01
-3.29610705e-01 9.45754573e-02 -3.00268352e-01 -4.66224641e-01
5.84698498e-01 5.20432234e-01 -2.71341532e-01 -3.32186282e-01
2.42169365e-01 4.69589710e-01 6.69391036e-01 -4.66755033e-01
1.31065297e+00 5.43705881e-01 -3.42340499e-01 -8.75263751e-01
-1.19429028e+00 -4.34753448e-01 -3.82099539e-01 -3.57115746e-01
1.07895410e+00 -1.35683060e+00 -2.16796145e-01 9.20665026e-01
-9.25311148e-01 -9.83285785e-01 -2.40056038e-01 2.09924579e-01
-1.75251141e-01 2.16210067e-01 -5.80010593e-01 -7.71793962e-01
-1.95863023e-01 -1.00866759e+00 9.67079401e-01 6.46642625e-01
5.35895288e-01 -8.75268042e-01 -3.44932601e-02 6.04579031e-01
8.93336117e-01 2.11272866e-01 7.54379451e-01 3.18043947e-01
-1.10733032e+00 3.18934113e-01 -4.19877559e-01 4.49039578e-01
4.01280522e-01 -1.59301028e-01 -1.10125542e+00 -2.04461962e-01
1.25406340e-01 -9.22347605e-02 1.14493310e+00 6.01852059e-01
1.40461075e+00 -4.83464092e-01 2.68818945e-01 1.10453808e+00
1.46723580e+00 8.27845037e-02 9.94308531e-01 5.23249269e-01
1.13083792e+00 6.02034211e-01 4.54672962e-01 5.08098267e-02
4.88684088e-01 4.32564497e-01 5.37690163e-01 -6.96570754e-01
-7.47394145e-01 -1.70735925e-01 1.46236315e-01 5.59707999e-01
-7.24112848e-03 -2.85913110e-01 -7.51366794e-01 8.79395068e-01
-1.47317123e+00 -7.53447652e-01 -9.64059960e-03 1.97978139e+00
1.16039968e+00 -1.04646936e-01 -3.49149287e-01 -2.63272613e-01
8.14825296e-01 3.00349891e-01 -3.76666456e-01 -1.61144868e-01
-1.93768620e-01 2.27379888e-01 5.28389215e-01 7.83584535e-01
-1.25760794e+00 1.19480741e+00 5.66759109e+00 7.95185506e-01
-1.12594438e+00 1.45450488e-01 8.11964750e-01 -2.15015784e-01
-5.45041502e-01 -4.57968116e-01 -4.68972594e-01 9.30048048e-01
6.80580080e-01 -4.14817594e-02 6.13909602e-01 5.24031818e-01
2.13531598e-01 -1.83498457e-01 -4.68976647e-01 9.57390964e-01
3.15731019e-01 -1.30224335e+00 7.75461569e-02 -1.78298846e-01
1.48593056e+00 -8.34972039e-02 6.68991089e-01 2.95604080e-01
5.22304416e-01 -1.05948126e+00 7.65885413e-01 4.72521156e-01
9.24720049e-01 -6.92480147e-01 6.70039594e-01 -4.00735587e-02
-9.29651856e-01 3.77642289e-02 -4.74258214e-01 5.83130233e-02
-2.49735385e-01 5.97202718e-01 -2.45984897e-01 5.43110192e-01
1.06511760e+00 6.63876772e-01 -8.10559392e-01 1.40307963e+00
-4.78966117e-01 5.18718541e-01 2.41122782e-01 4.24395800e-01
2.38219693e-01 -2.74998337e-01 4.71401542e-01 1.09246051e+00
3.33542109e-01 4.91607219e-01 -1.97047576e-01 9.64121103e-01
8.50134715e-02 -3.54738891e-01 -6.30257845e-01 3.87708306e-01
2.37820476e-01 8.49843442e-01 -2.46391058e-01 -4.40424085e-01
-4.96248275e-01 1.13034749e+00 2.25021794e-01 8.03141117e-01
-8.17456603e-01 -4.75518674e-01 8.64865482e-01 2.87630945e-01
4.31155533e-01 -6.04025237e-02 -3.91431630e-01 -1.25104666e+00
-2.20008269e-01 -1.05211425e+00 1.29898101e-01 -1.26870728e+00
-1.29064262e+00 5.52879155e-01 9.03389305e-02 -1.08641100e+00
4.13011432e-01 -4.28483605e-01 -9.26430106e-01 1.01370811e+00
-2.24176908e+00 -1.02952933e+00 -1.04797292e+00 7.68694460e-01
9.46861982e-01 1.16148852e-01 2.44263560e-01 5.39266586e-01
-7.35601366e-01 4.10331964e-01 2.18853205e-01 -1.09182507e-01
9.52881634e-01 -1.53593647e+00 3.15992117e-01 1.31527746e+00
-4.24364626e-01 3.44351441e-01 9.17388201e-01 -5.89691639e-01
-1.18708217e+00 -1.48099458e+00 4.84224945e-01 -4.09503549e-01
3.26761931e-01 -4.74932998e-01 -1.31773257e+00 4.55920488e-01
7.13593006e-01 2.62534052e-01 2.89506763e-01 -4.43173856e-01
-2.46896148e-01 -2.71674007e-01 -8.86137605e-01 9.10801470e-01
7.41435468e-01 -3.02920520e-01 -3.81389022e-01 1.89103335e-01
1.04957354e+00 -8.05521071e-01 -3.88825387e-01 2.93172091e-01
1.58521891e-01 -1.03514159e+00 1.00891364e+00 -1.72240764e-01
7.61729956e-01 -4.52310264e-01 1.40703604e-01 -1.60789967e+00
-6.81397021e-01 -7.99591601e-01 -3.00704777e-01 8.79138350e-01
-6.07634746e-02 -5.54201603e-01 8.24047804e-01 3.86146754e-01
-5.30337214e-01 -6.00344121e-01 -5.92718959e-01 -5.04680812e-01
1.36294171e-01 2.83713609e-01 7.34628618e-01 1.12191427e+00
-7.17518806e-01 9.87513959e-02 -6.13884151e-01 5.93732595e-01
9.78534639e-01 4.81863320e-02 5.99285126e-01 -8.31293046e-01
8.54416862e-02 -4.77710426e-01 -1.38691962e-01 -9.49087858e-01
2.75295544e-02 -1.37092680e-01 3.78369451e-01 -1.54904521e+00
2.16881961e-01 -1.44190609e-01 -4.68666673e-01 3.18301648e-01
-8.14222097e-01 4.35669631e-01 -2.01543063e-01 4.09070440e-02
-7.24253595e-01 9.95770216e-01 1.95195079e+00 -3.93623739e-01
-1.58601582e-01 -2.18556225e-01 -9.30682480e-01 4.13755745e-01
1.00143278e+00 -4.01762456e-01 -7.05759704e-01 -8.92862916e-01
-2.17048779e-01 -3.68439853e-01 3.99598062e-01 -1.06056678e+00
2.37173885e-01 -4.28488314e-01 7.48685420e-01 -4.55362618e-01
2.68407255e-01 -7.00351655e-01 -1.30915418e-01 3.32594573e-01
-1.94494158e-01 -8.30958188e-02 4.23160881e-01 4.74574238e-01
-3.31354678e-01 6.20559938e-02 9.68199074e-01 -2.31274888e-01
-7.36899376e-01 4.35600996e-01 -5.69056153e-01 6.17935993e-02
9.10564721e-01 -1.70998633e-01 -8.66426468e-01 -5.56820810e-01
-9.74724516e-02 2.47039795e-01 7.35526681e-01 4.76466596e-01
9.77450550e-01 -1.21511436e+00 -7.62528062e-01 4.55229133e-01
-2.47116871e-02 5.74640155e-01 7.28875756e-01 6.04277611e-01
-9.37076330e-01 9.28847566e-02 -4.73833054e-01 -2.50353694e-01
-1.14437330e+00 7.70019412e-01 5.09787619e-01 1.97268665e-01
-7.99966097e-01 9.79840159e-01 9.56294179e-01 -4.31176834e-02
2.15139091e-01 -1.54368520e-01 -1.16497956e-01 -5.96850753e-01
7.59508312e-01 2.37528592e-01 -1.16515674e-01 -3.38254511e-01
6.64432719e-02 4.48262900e-01 -2.24400446e-01 1.64918616e-01
1.26078987e+00 -4.57116246e-01 -1.57743081e-01 2.89230257e-01
8.25729668e-01 2.88946889e-02 -1.93305075e+00 -2.95482606e-01
-5.79589307e-01 -1.11146498e+00 5.36756516e-01 -8.09906483e-01
-1.28425038e+00 8.35067451e-01 7.52991855e-01 -3.65141258e-02
1.47726595e+00 -2.59719282e-01 8.07689428e-01 -1.37679845e-01
-1.32775113e-01 -6.22865438e-01 4.80871141e-01 3.54394108e-01
8.34136248e-01 -1.36135435e+00 1.04409449e-01 -5.89001775e-01
-6.28161967e-01 5.82625747e-01 1.29159927e+00 -1.29102349e-01
4.83838469e-01 1.07290305e-01 3.87957484e-01 -3.51838380e-01
-7.07097530e-01 -3.21783632e-01 6.20986462e-01 8.13662827e-01
2.26896703e-01 -2.74223417e-01 3.22115242e-01 3.53804082e-02
-2.52667889e-02 -4.31924701e-01 7.76730657e-01 9.23648536e-01
-5.97876787e-01 -5.12244642e-01 -6.11322343e-01 4.78295982e-02
-4.50527370e-01 -4.45573360e-01 -1.65434569e-01 7.57710040e-01
3.57024103e-01 9.12933230e-01 1.50766462e-01 -2.48342827e-01
1.91133380e-01 -3.29157501e-01 5.99950314e-01 -6.60579205e-01
-2.45960116e-01 2.48679280e-01 -3.64365190e-01 -4.33407754e-01
-1.39233246e-01 -2.66788185e-01 -5.74308455e-01 -5.18235624e-01
-2.98153460e-02 9.10718832e-03 4.54218775e-01 4.81591880e-01
4.16022331e-01 1.11904156e+00 7.15307653e-01 -9.23361480e-01
7.19959065e-02 -8.94246876e-01 -5.61040044e-01 4.55093116e-01
9.69647288e-01 -6.99520886e-01 -4.31401283e-01 1.31686732e-01]
|
[10.942460060119629, -3.109165906906128]
|
afdf8f1c-6eb5-4513-b932-38cbd4589e85
|
realistic-face-reenactment-via-self
|
2003.12957
| null |
https://arxiv.org/abs/2003.12957v1
|
https://arxiv.org/pdf/2003.12957v1.pdf
|
Realistic Face Reenactment via Self-Supervised Disentangling of Identity and Pose
|
Recent works have shown how realistic talking face images can be obtained under the supervision of geometry guidance, e.g., facial landmark or boundary. To alleviate the demand for manual annotations, in this paper, we propose a novel self-supervised hybrid model (DAE-GAN) that learns how to reenact face naturally given large amounts of unlabeled videos. Our approach combines two deforming autoencoders with the latest advances in the conditional generation. On the one hand, we adopt the deforming autoencoder to disentangle identity and pose representations. A strong prior in talking face videos is that each frame can be encoded as two parts: one for video-specific identity and the other for various poses. Inspired by that, we utilize a multi-frame deforming autoencoder to learn a pose-invariant embedded face for each video. Meanwhile, a multi-scale deforming autoencoder is proposed to extract pose-related information for each frame. On the other hand, the conditional generator allows for enhancing fine details and overall reality. It leverages the disentangled features to generate photo-realistic and pose-alike face images. We evaluate our model on VoxCeleb1 and RaFD dataset. Experiment results demonstrate the superior quality of reenacted images and the flexibility of transferring facial movements between identities.
|
['Yong liu', 'Jiangning Zhang', 'Yusu Pan', 'Xianfang Zeng', 'Mengmeng Wang']
|
2020-03-29
| null | null | null | null |
['face-reenactment']
|
['computer-vision']
|
[-1.80796739e-02 3.69696736e-01 1.45557463e-01 -7.05898583e-01
-6.90884233e-01 -5.24720430e-01 5.07099271e-01 -1.13392913e+00
1.77057251e-01 6.95061445e-01 4.16983217e-01 3.96683872e-01
1.77530780e-01 -7.01535225e-01 -1.01140141e+00 -9.05813634e-01
4.74143118e-01 3.70718360e-01 -4.62404221e-01 -2.86862820e-01
-3.42194855e-01 6.20739639e-01 -1.64799726e+00 2.04259798e-01
5.35808921e-01 9.96058524e-01 -2.41077486e-02 4.41656619e-01
3.12507361e-01 5.91454625e-01 -4.22165483e-01 -7.97864914e-01
6.16657019e-01 -4.30809140e-01 -4.06469464e-01 5.50507069e-01
9.20489609e-01 -9.48444903e-01 -8.72885048e-01 8.68921340e-01
6.09878957e-01 -2.93847620e-02 4.63077366e-01 -1.42031968e+00
-8.85401130e-01 3.07203203e-01 -7.16414332e-01 -2.47799814e-01
6.26710892e-01 2.99914628e-01 6.76775992e-01 -1.02387702e+00
8.20297539e-01 1.49176228e+00 5.54816782e-01 8.62644255e-01
-1.12318861e+00 -9.51094031e-01 3.57018858e-02 -1.17119282e-01
-1.57781017e+00 -8.39894414e-01 1.16455078e+00 -4.87440407e-01
1.86175093e-01 4.19086404e-02 7.00640202e-01 1.61446571e+00
-6.14055730e-02 5.91122627e-01 9.03261960e-01 -1.57534942e-01
-1.25420034e-01 -8.08909237e-02 -8.94706786e-01 9.36507821e-01
-1.37325954e-02 4.21398759e-01 -5.76222837e-01 1.22048393e-01
1.26760256e+00 2.10063234e-01 -4.91826773e-01 -4.16297555e-01
-1.06215906e+00 7.83520401e-01 2.38221705e-01 3.12658772e-02
-3.35442305e-01 -7.42103010e-02 -1.20639130e-02 6.04729243e-02
4.47825462e-01 3.83742303e-02 -3.14350247e-01 2.27375001e-01
-8.58803153e-01 2.22811878e-01 5.59857905e-01 1.08077359e+00
9.55712795e-01 4.01501805e-01 -1.48722917e-01 6.05838597e-01
3.45462203e-01 6.00212395e-01 4.08552676e-01 -1.18112981e+00
1.79791689e-01 2.30095848e-01 -6.94552287e-02 -1.11202490e+00
6.01476021e-02 -2.31890425e-01 -8.80428135e-01 2.17397884e-01
1.82472646e-01 -3.57243091e-01 -9.55721915e-01 2.13671923e+00
5.82826674e-01 6.75182462e-01 1.24086902e-01 9.28558826e-01
9.31705117e-01 4.56893057e-01 -1.66008607e-01 -9.99264717e-02
1.16682780e+00 -8.40049982e-01 -9.23144877e-01 8.75346586e-02
-8.75367001e-02 -7.48534560e-01 5.50645351e-01 2.04917844e-02
-1.25439894e+00 -7.27869749e-01 -9.20531631e-01 -2.47363433e-01
-2.06887610e-02 2.05931798e-01 5.18841445e-01 4.93537188e-01
-1.11838174e+00 4.32344019e-01 -7.74090052e-01 3.91725674e-02
4.87666428e-01 3.97791594e-01 -8.68808925e-01 -8.76645669e-02
-1.08862996e+00 4.17620659e-01 1.35372519e-01 2.15136185e-01
-1.19106472e+00 -7.80238688e-01 -1.20866048e+00 2.91679036e-02
2.83120364e-01 -9.68780637e-01 9.25232410e-01 -1.34716976e+00
-1.95762289e+00 9.72674787e-01 2.38675084e-02 -7.82916695e-03
7.33039081e-01 -6.25086799e-02 -4.61724639e-01 4.03190821e-01
9.27990526e-02 1.00365078e+00 1.31704700e+00 -1.54960668e+00
-2.33285278e-01 -6.48077905e-01 1.40532479e-01 3.34205091e-01
-3.82289708e-01 -3.60167682e-01 -6.15494490e-01 -1.00401270e+00
-9.41075683e-02 -1.08552539e+00 2.71893233e-01 2.96382606e-01
-3.61976773e-01 1.07362211e-01 1.12574387e+00 -9.56940055e-01
5.76531827e-01 -2.20932484e+00 4.96457666e-01 -1.29254803e-01
3.08598667e-01 -8.93833581e-03 -3.23671699e-01 9.32929367e-02
-4.21615362e-01 -7.63594806e-02 -3.59074445e-04 -6.27294898e-01
-1.14386946e-01 3.72074038e-01 -3.65162015e-01 5.91971278e-01
4.58258986e-01 1.05295193e+00 -7.49433339e-01 -4.36095744e-01
2.58452147e-01 1.11850190e+00 -8.95051122e-01 5.77447474e-01
-9.92156129e-05 1.05464041e+00 -4.08171594e-01 5.59748709e-01
1.01933026e+00 -5.73143102e-02 1.98356688e-01 -7.25172520e-01
2.21430570e-01 -3.19771647e-01 -1.11078978e+00 2.05495048e+00
-5.05043149e-01 3.42425853e-01 2.45567665e-01 -8.87515306e-01
8.03491771e-01 6.47127450e-01 6.51081562e-01 -3.60650718e-01
2.84823656e-01 -4.44310084e-02 -3.48649293e-01 -5.73736191e-01
6.57989979e-02 -4.11290884e-01 1.87850267e-01 2.36878261e-01
6.99921489e-01 -1.44310743e-01 -3.11517537e-01 3.35156657e-02
5.16949236e-01 6.15841389e-01 -6.62009120e-02 2.58757025e-02
5.80503881e-01 -8.88059914e-01 8.77667725e-01 -9.06176344e-02
3.61739509e-02 1.06303966e+00 1.85986444e-01 -5.99472225e-01
-9.35475647e-01 -1.19478607e+00 2.32328717e-02 8.13968718e-01
3.83861512e-02 -1.44914031e-01 -1.05668724e+00 -6.26121819e-01
2.81007006e-03 2.22847492e-01 -9.63987172e-01 -2.38364846e-01
-6.48392916e-01 -1.59476191e-01 5.13831139e-01 5.83022118e-01
6.50228322e-01 -7.73523986e-01 -7.43642896e-02 -1.43920869e-01
-3.33077997e-01 -1.36731982e+00 -9.04086590e-01 -5.71759403e-01
-3.96950126e-01 -9.63805020e-01 -8.78399789e-01 -8.37403655e-01
8.22142184e-01 1.42038599e-01 9.20225978e-01 -1.20038137e-01
-2.57420033e-01 5.16627312e-01 -1.20296605e-01 -7.03928545e-02
-1.98980838e-01 -4.45858091e-01 4.12938505e-01 8.19533885e-01
-1.52060255e-01 -1.08610487e+00 -8.02353501e-01 3.78434271e-01
-9.24993753e-01 1.71894029e-01 5.46183825e-01 9.17151332e-01
5.61988473e-01 -1.90564200e-01 5.80049098e-01 -5.89719594e-01
9.66384858e-02 -4.69969660e-01 -3.83176655e-01 2.43171170e-01
5.57484403e-02 3.03015998e-03 6.28660619e-01 -5.38522542e-01
-1.46730423e+00 3.51958185e-01 -3.04053664e-01 -1.13960946e+00
-1.01277038e-01 -1.63656712e-01 -8.25098813e-01 -2.38890395e-01
3.01098496e-01 3.08578372e-01 2.05131248e-01 -2.73479611e-01
6.62363529e-01 4.19061661e-01 9.04043913e-01 -8.10601950e-01
1.13059711e+00 8.29606891e-01 -2.57475257e-01 -7.31099069e-01
-7.04899728e-01 1.88633159e-01 -7.28411198e-01 -3.35894614e-01
1.17270255e+00 -1.27306426e+00 -9.74272668e-01 5.01909375e-01
-1.02707946e+00 -1.38733819e-01 -3.87913167e-01 3.39574426e-01
-8.00140738e-01 3.44610602e-01 -5.35634637e-01 -4.39405918e-01
-9.92281884e-02 -1.25155950e+00 1.54912460e+00 4.60634023e-01
2.04891726e-01 -8.71615529e-01 -4.36409786e-02 6.44496858e-01
6.02106452e-02 6.69537842e-01 2.55070597e-01 -2.36904129e-01
-6.47928655e-01 -1.20686665e-01 8.27102065e-02 4.41293091e-01
3.60316783e-01 9.99257937e-02 -1.22931755e+00 -4.74219561e-01
2.50596613e-01 -5.02611518e-01 4.69416171e-01 1.36051759e-01
1.37983525e+00 -6.24620736e-01 -1.36311069e-01 1.13238323e+00
1.01187742e+00 4.44929637e-02 6.71626508e-01 -4.49900448e-01
1.02078056e+00 7.59264529e-01 1.52389780e-01 4.26968575e-01
5.19296944e-01 8.17900479e-01 3.64102662e-01 -6.72770292e-02
-3.45590830e-01 -7.38062620e-01 3.79329443e-01 8.05674136e-01
-3.93803984e-01 -6.48319721e-02 -3.63147974e-01 1.86825842e-01
-1.37224400e+00 -1.04613638e+00 7.62287199e-01 1.90498328e+00
7.17960238e-01 -4.31190968e-01 -9.86351669e-02 -9.06179026e-02
7.96704173e-01 7.49478936e-02 -7.66786039e-01 3.86062592e-01
-2.66463190e-01 2.38123193e-01 -1.57125667e-01 4.99247164e-01
-8.38394880e-01 9.30536747e-01 4.85437250e+00 6.46075845e-01
-1.39419091e+00 7.47332796e-02 7.26370335e-01 -9.38476548e-02
-3.30746710e-01 -1.65681452e-01 -5.29004753e-01 4.69301075e-01
5.16605914e-01 -1.11960463e-01 4.96858299e-01 8.31125438e-01
-9.21671614e-02 5.22143006e-01 -1.14533174e+00 1.29387009e+00
5.79266965e-01 -1.28542733e+00 4.17789519e-01 2.00874835e-01
9.00804639e-01 -5.39059877e-01 4.17551398e-01 1.49846375e-01
2.56735921e-01 -1.08523023e+00 8.95847380e-01 6.69968486e-01
1.33237457e+00 -7.76150823e-01 3.94883901e-01 1.49698742e-02
-1.11040986e+00 1.25693548e-02 -8.25999677e-02 2.08289087e-01
2.14093789e-01 1.31391063e-01 -5.03984272e-01 8.06466341e-01
7.09381759e-01 8.15757513e-01 -2.50087380e-01 2.81609863e-01
-1.89408973e-01 1.04787737e-01 -1.67988256e-01 8.94369245e-01
-1.64376974e-01 -4.06632215e-01 4.31254297e-01 6.58998668e-01
4.84048903e-01 4.57336456e-01 1.07757814e-01 8.64000499e-01
-4.85177249e-01 -2.05330938e-01 -7.54695833e-01 1.31692812e-01
4.41949695e-01 1.46950650e+00 -2.52196997e-01 -1.62059575e-01
-4.83914435e-01 1.34670794e+00 2.64543742e-01 4.87225592e-01
-8.60137403e-01 1.03556842e-01 8.51710081e-01 1.41532868e-01
4.74855751e-01 -3.34331505e-02 3.55715841e-01 -1.59511733e+00
1.08600996e-01 -8.95011187e-01 7.96595141e-02 -9.78182316e-01
-1.37102866e+00 8.41868997e-01 -1.11059956e-01 -1.24008369e+00
-3.70532334e-01 -4.72045749e-01 -5.68791449e-01 6.93611503e-01
-1.25036716e+00 -1.68129313e+00 -5.98234355e-01 1.04184151e+00
5.59039474e-01 -2.91459858e-01 8.32346857e-01 4.00934726e-01
-6.42123699e-01 9.81369436e-01 -2.72594333e-01 5.75475812e-01
9.07536447e-01 -9.18831825e-01 1.69736370e-01 4.82226998e-01
3.18898141e-01 6.75966680e-01 3.71239066e-01 -4.31338727e-01
-1.52788448e+00 -1.22673106e+00 2.36231327e-01 -4.48901653e-01
1.73877478e-01 -5.12114465e-01 -6.90134645e-01 1.05729139e+00
2.51846373e-01 5.42965114e-01 8.03164482e-01 -4.37964976e-01
-5.77300787e-01 -3.65970641e-01 -1.28706706e+00 6.19248152e-01
1.31022012e+00 -7.60952711e-01 -4.07894760e-01 2.38649547e-01
7.73618639e-01 -6.72623336e-01 -1.17082000e+00 4.32345808e-01
7.71202564e-01 -1.14760113e+00 1.11283696e+00 -5.30469000e-01
6.42094195e-01 -2.56124318e-01 -2.45718971e-01 -1.22767448e+00
-8.95800292e-02 -9.62800801e-01 -1.17584459e-01 1.46262538e+00
-1.64028436e-01 -4.97035921e-01 7.32685566e-01 4.57253218e-01
-7.33731315e-03 -8.94395351e-01 -9.35000777e-01 -4.70301598e-01
9.29881483e-02 1.60378546e-01 9.76509809e-01 1.15278304e+00
-5.14684439e-01 4.76126552e-01 -8.20436418e-01 2.60984272e-01
6.77141547e-01 7.98072144e-02 9.81545091e-01 -1.04911530e+00
-3.57628226e-01 2.91298889e-02 -5.73102415e-01 -1.15655971e+00
6.69644117e-01 -7.16523170e-01 -2.90070027e-01 -7.29234338e-01
1.40158907e-01 -2.05391590e-02 2.60371625e-01 3.98286045e-01
-6.29667565e-02 5.79205096e-01 1.45695180e-01 2.54565049e-02
-1.86965629e-01 1.12436116e+00 1.62529171e+00 1.88793149e-02
1.11836538e-01 -3.08591396e-01 -6.86872482e-01 9.24925923e-01
4.59547162e-01 -7.79403970e-02 -5.83965302e-01 -6.43068969e-01
-1.03774369e-01 4.61040378e-01 5.25665760e-01 -8.47072423e-01
1.40352547e-01 -6.39232770e-02 7.79552817e-01 -2.94911832e-01
6.68715715e-01 -9.33093131e-01 4.74857956e-01 -4.92463335e-02
-1.84068717e-02 5.54413535e-02 1.17996670e-01 8.09916735e-01
-3.18737656e-01 4.50798035e-01 8.06671083e-01 -3.34250554e-02
-3.31087619e-01 9.26537454e-01 3.51509422e-01 2.24843365e-03
1.14426625e+00 -1.80198103e-01 1.43952131e-01 -7.62417793e-01
-9.03895319e-01 5.76971611e-03 6.69144869e-01 7.52396703e-01
7.06177711e-01 -1.65326178e+00 -8.06740403e-01 7.08694637e-01
-6.90629482e-02 2.46217161e-01 7.88060248e-01 3.74642193e-01
-3.62862289e-01 2.76015271e-02 -5.99845886e-01 -6.07051790e-01
-1.04990959e+00 6.58399045e-01 5.21557629e-01 2.32588544e-01
-6.10329568e-01 8.86879742e-01 7.71270275e-01 -5.32545924e-01
-5.70096895e-02 3.30070853e-01 -4.50347550e-02 -7.21298307e-02
5.90388298e-01 -2.36770213e-02 -1.60677999e-01 -1.26559412e+00
-8.78590941e-02 9.55785751e-01 -1.21570945e-01 -1.74352556e-01
1.45942855e+00 -1.83138132e-01 9.18155983e-02 2.57412065e-02
1.51476920e+00 2.34639943e-01 -1.97937858e+00 -1.76797926e-01
-8.19875419e-01 -7.13213623e-01 -1.21955514e-01 -4.38569814e-01
-1.80933249e+00 7.91623890e-01 5.08717239e-01 -3.80538493e-01
1.21844995e+00 -5.03787361e-02 7.73977578e-01 -9.18892864e-03
3.76598865e-01 -6.34692848e-01 5.07656991e-01 6.46252511e-03
1.33055019e+00 -1.24528706e+00 -2.73546427e-01 -6.02688134e-01
-6.61383331e-01 1.11455989e+00 8.12560320e-01 -2.32255355e-01
7.27152526e-01 3.21866959e-01 8.65298137e-02 -1.22693621e-01
-3.59976530e-01 2.00815469e-01 2.23988473e-01 7.78264284e-01
1.46727219e-01 -8.02222732e-03 3.31261784e-01 7.03491449e-01
-5.67977607e-01 -1.48905441e-03 2.64943153e-01 4.63863462e-01
3.18226546e-01 -9.10847366e-01 -3.12113523e-01 -5.31749427e-02
-2.47779444e-01 1.55173719e-01 -2.30144441e-01 7.84209847e-01
4.17881012e-01 7.47846305e-01 2.35993117e-01 -5.53528905e-01
2.11386904e-01 -5.47721162e-02 9.01803911e-01 -5.68134367e-01
-9.46132764e-02 6.39688969e-02 -4.34997261e-01 -6.96588337e-01
-3.73725235e-01 -5.37126958e-01 -8.94356966e-01 -3.39868665e-01
-1.32095411e-01 -2.06634961e-03 3.11397731e-01 7.49458373e-01
5.49017608e-01 4.85471189e-01 8.77926946e-01 -1.33888626e+00
-4.74137813e-01 -8.48771453e-01 -5.53064883e-01 6.58337235e-01
5.47733605e-01 -1.04912829e+00 -3.15148354e-01 4.89970237e-01]
|
[12.845610618591309, -0.21419517695903778]
|
9325de24-25e7-4dad-8342-3e655d81b1c6
|
detecting-stylistic-deception
| null | null |
https://aclanthology.org/W12-0414
|
https://aclanthology.org/W12-0414.pdf
|
Detecting Stylistic Deception
| null |
['Patrick Juola']
|
2012-04-01
| null | null | null |
ws-2012-4
|
['deception-detection']
|
['miscellaneous']
|
[-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]
|
[-7.246927738189697, 3.7058560848236084]
|
364426f8-fa56-4065-87ed-3b35819ee3ce
|
uncertainty-based-network-for-few-shot-image
|
2205.08157
| null |
https://arxiv.org/abs/2205.08157v1
|
https://arxiv.org/pdf/2205.08157v1.pdf
|
Uncertainty-based Network for Few-shot Image Classification
|
The transductive inference is an effective technique in the few-shot learning task, where query sets update prototypes to improve themselves. However, these methods optimize the model by considering only the classification scores of the query instances as confidence while ignoring the uncertainty of these classification scores. In this paper, we propose a novel method called Uncertainty-Based Network, which models the uncertainty of classification results with the help of mutual information. Specifically, we first data augment and classify the query instance and calculate the mutual information of these classification scores. Then, mutual information is used as uncertainty to assign weights to classification scores, and the iterative update strategy based on classification scores and uncertainties assigns the optimal weights to query instances in prototype optimization. Extensive results on four benchmarks show that Uncertainty-Based Network achieves comparable performance in classification accuracy compared to state-of-the-art method.
|
['Tong Lu', 'Tao Wang', 'Yin-Dong Zheng', 'Chunhao Cai', 'Qian Xu', 'Minglei Yuan']
|
2022-05-17
| null | null | null | null |
['few-shot-image-classification']
|
['computer-vision']
|
[ 3.55853513e-02 1.53431334e-02 -6.04712248e-01 -7.02208102e-01
-7.97654927e-01 -1.09551013e-01 4.03787911e-01 2.13043258e-01
-3.85969102e-01 8.15743148e-01 9.06938091e-02 2.54695028e-01
-5.28040886e-01 -1.12461829e+00 -4.61525381e-01 -4.37595695e-01
1.50933906e-01 7.55988777e-01 6.00228012e-01 4.76856343e-02
3.41030478e-01 -8.42198953e-02 -1.98986471e+00 3.52590352e-01
1.07504904e+00 1.40575540e+00 -5.50450496e-02 2.75025457e-01
-5.66384792e-01 7.58621395e-01 -5.84533036e-01 -4.83806282e-01
6.19667508e-02 -2.92260885e-01 -5.15537024e-01 -3.92514884e-01
-2.41253570e-01 -2.91861683e-01 -1.73917845e-01 1.27049541e+00
2.83603996e-01 7.61432171e-01 7.86438346e-01 -1.22967422e+00
-6.75735176e-01 9.00504768e-01 -8.61015469e-02 -5.90981543e-03
3.60905856e-01 -3.35940450e-01 1.16823637e+00 -1.06450713e+00
3.30086499e-01 1.21992195e+00 4.29367065e-01 2.89889127e-01
-1.06802022e+00 -5.95741391e-01 1.27685726e-01 7.86777973e-01
-1.61541641e+00 -3.38574409e-01 7.31684566e-01 -2.95682758e-01
8.37354958e-01 3.50238800e-01 5.84093392e-01 2.72301763e-01
-6.07871115e-02 8.73416841e-01 6.39635384e-01 -4.73806143e-01
6.84149325e-01 5.10773003e-01 4.70435381e-01 7.88999677e-01
8.01480338e-02 3.54722947e-01 -5.42953432e-01 -3.54252219e-01
8.41323063e-02 3.53459895e-01 -1.52235106e-01 -3.78941476e-01
-7.89300025e-01 9.25446570e-01 5.94772220e-01 1.15792356e-01
-1.89806387e-01 1.35723539e-02 1.12937033e-01 2.55878538e-01
7.02715874e-01 2.72234976e-01 -5.37318051e-01 -1.54915582e-02
-7.22101569e-01 -6.52649179e-02 9.16065335e-01 9.21478152e-01
1.07008195e+00 -3.27919394e-01 -6.94538236e-01 1.06483018e+00
7.04115152e-01 2.72530138e-01 5.41311085e-01 -7.39708364e-01
2.37716407e-01 7.53370047e-01 6.63739294e-02 -9.97702539e-01
8.31328630e-02 -2.79822409e-01 -5.37274778e-01 -8.37229006e-03
-2.11302832e-01 -2.31257498e-01 -1.13736987e+00 1.69833732e+00
4.11200225e-01 6.63961411e-01 2.84504831e-01 7.89345801e-01
9.28391516e-01 7.10880339e-01 -5.35801649e-02 -6.72732890e-01
7.47074008e-01 -8.14095199e-01 -8.16015542e-01 2.33223364e-02
4.22992200e-01 -6.89429402e-01 7.18548357e-01 1.33396834e-01
-6.29687786e-01 -4.43625420e-01 -1.23874581e+00 4.39096153e-01
-4.56318229e-01 -9.57405344e-02 7.22607791e-01 6.05229318e-01
-4.61300850e-01 9.67212260e-01 -6.64744914e-01 -8.59646276e-02
2.50436187e-01 3.24672401e-01 1.76908731e-01 -5.27637713e-02
-1.59295404e+00 9.65757370e-01 7.83795714e-01 -1.06500670e-01
-4.95565891e-01 -7.70337701e-01 -9.60950851e-01 1.59571216e-01
6.43474996e-01 -5.70533216e-01 1.25462127e+00 -4.89068776e-01
-1.73236728e+00 2.39820749e-01 -2.70637929e-01 -3.74123693e-01
1.38422921e-01 -1.00121861e-02 -4.06661540e-01 -4.25541103e-01
1.58239100e-02 5.70674717e-01 5.25551915e-01 -1.00496399e+00
-9.56554830e-01 -2.55921036e-01 -4.39982824e-02 3.88717115e-01
-2.07851887e-01 -3.19078743e-01 -7.73026764e-01 -1.98827833e-01
5.45825839e-01 -5.12393296e-01 -1.71650141e-01 -1.29264653e-01
-2.64179140e-01 -5.80901265e-01 6.61157429e-01 -5.76990619e-02
1.51886737e+00 -1.91073895e+00 2.06699550e-01 4.04001564e-01
-1.09439068e-01 7.03492239e-02 1.24991722e-01 -9.90176126e-02
4.27338243e-01 -3.41614406e-03 -4.27202106e-01 -2.17678964e-01
1.46520689e-01 4.62317795e-01 -1.97054386e-01 -7.17636421e-02
1.11745253e-01 8.25331032e-01 -1.06871355e+00 -6.87921703e-01
3.13053727e-01 1.22156605e-01 -5.54531157e-01 4.43121076e-01
-4.63060051e-01 -1.67183489e-01 -6.02396190e-01 7.17414975e-01
4.21505660e-01 -7.73693696e-02 -5.01515120e-02 -4.43933696e-01
2.31507316e-01 -5.13773412e-02 -1.31546271e+00 1.66272473e+00
-2.82365739e-01 2.00083613e-01 -6.71569943e-01 -8.79377306e-01
1.03438699e+00 -2.08552480e-02 3.68145436e-01 -3.22686374e-01
1.85893834e-01 -1.17040947e-01 -8.94796327e-02 -5.17229617e-01
3.00373852e-01 -8.66953060e-02 -2.24872276e-01 4.38273132e-01
2.49925792e-01 -5.88369668e-01 2.12023348e-01 1.84586063e-01
6.68619394e-01 -8.04098099e-02 2.94348657e-01 2.47433111e-01
4.50802356e-01 -1.90590620e-01 9.23738658e-01 8.13097596e-01
-2.13516936e-01 5.08779824e-01 2.57274032e-01 -1.94171742e-01
-6.11179650e-01 -1.31660509e+00 -1.34237677e-01 1.08057249e+00
3.63030821e-01 -3.34899962e-01 -4.23845887e-01 -8.15491021e-01
2.19282672e-01 1.09596431e+00 -7.03114390e-01 -6.56794190e-01
3.70880574e-01 -8.15598905e-01 1.51729137e-01 3.40167105e-01
4.73378003e-01 -7.75310397e-01 -2.75426637e-02 2.48160690e-01
-1.58369333e-01 -3.71048748e-01 -4.55532581e-01 -4.82300706e-02
-7.91306078e-01 -1.08211970e+00 -1.58222049e-01 -4.92208421e-01
4.97539282e-01 -1.02203652e-01 8.13467920e-01 -9.40508321e-02
-2.28628963e-01 3.28848004e-01 -3.43231261e-01 -5.83262682e-01
3.71470265e-02 -2.51504898e-01 1.71413630e-01 1.02323934e-01
6.18607044e-01 -3.69823456e-01 -3.78038883e-01 3.02401096e-01
-6.51828945e-01 -2.76279092e-01 4.85691011e-01 9.83579576e-01
9.69538987e-01 1.33358091e-01 7.33214319e-01 -9.57184196e-01
8.21013212e-01 -6.67328477e-01 -6.19260013e-01 8.72129738e-01
-1.21608937e+00 4.33373213e-01 3.18621881e-02 -4.87467676e-01
-1.36869633e+00 -3.85899320e-02 1.45410404e-01 -5.50886691e-01
3.15847456e-01 8.93725038e-01 1.04239294e-02 3.36455777e-02
5.36053717e-01 3.80347064e-03 -9.62942094e-02 -2.50892907e-01
6.34990215e-01 7.61524141e-01 3.66113335e-01 -4.95639652e-01
3.38360131e-01 1.33619532e-01 -2.36342415e-01 -2.98763722e-01
-1.29628837e+00 -6.40158057e-01 -3.52275997e-01 -2.37294078e-01
6.04997456e-01 -5.69867790e-01 -4.68657732e-01 2.26726219e-01
-9.78999972e-01 2.48747751e-01 -5.13708115e-01 8.82119060e-01
-4.49796528e-01 2.08971277e-01 -3.75462353e-01 -1.09374988e+00
-5.25466442e-01 -1.04044402e+00 6.63162708e-01 7.16004908e-01
-1.39201770e-03 -8.40016067e-01 1.88655064e-01 1.34754226e-01
4.33837891e-01 -1.49794713e-01 9.30522501e-01 -9.77043450e-01
-5.77225566e-01 -5.19735217e-01 -3.09901744e-01 3.23165894e-01
1.52195245e-01 1.27708867e-01 -9.35418248e-01 1.99920744e-01
6.77646846e-02 -2.73772717e-01 1.00054872e+00 3.71890932e-01
1.05646801e+00 -1.21859655e-01 -4.06262785e-01 2.74649560e-01
1.37136340e+00 3.00568044e-01 4.70907718e-01 -3.06088090e-01
3.53210866e-01 3.89310330e-01 1.10187411e+00 7.26238370e-01
3.56404305e-01 4.75784302e-01 2.40567401e-01 6.61203623e-01
3.85654807e-01 -1.89120889e-01 -3.37891169e-02 1.08695865e+00
2.73031354e-01 2.50608884e-02 -5.75215578e-01 2.40864635e-01
-2.40181637e+00 -9.84772980e-01 3.41812819e-01 2.51823115e+00
1.08614171e+00 2.79856861e-01 -5.07990062e-01 -1.41788080e-01
9.17149067e-01 -6.90415874e-02 -8.55755985e-01 -7.47214183e-02
3.03235352e-01 4.36572097e-02 4.03062627e-02 7.28595316e-01
-9.25962090e-01 9.86361504e-01 6.74509287e+00 1.01144505e+00
-6.52581692e-01 1.75965831e-01 4.60863709e-01 -3.35613400e-01
-3.19869131e-01 2.12418720e-01 -8.07756484e-01 5.67362368e-01
6.32401526e-01 -7.26805806e-01 5.40307462e-01 9.51822460e-01
-2.24004269e-01 -3.40505838e-01 -1.24032164e+00 8.85666609e-01
2.05710560e-01 -1.33658516e+00 2.45033987e-02 -5.94415784e-01
9.78423357e-01 6.65345415e-02 -3.80510613e-02 8.87928307e-01
4.82030451e-01 -4.72706974e-01 3.01741511e-01 1.44491971e+00
4.18460250e-01 -9.77328658e-01 1.05971611e+00 3.93610239e-01
-1.12026560e+00 -1.23306036e-01 -6.85605824e-01 -5.63569069e-02
2.96950843e-02 8.46106470e-01 -9.14788127e-01 4.38005567e-01
5.84166944e-01 6.02314472e-01 -3.54535490e-01 1.36038077e+00
-3.21604669e-01 3.67512077e-01 -4.83540982e-01 -5.11345029e-01
-5.96318021e-02 -2.28348389e-01 3.54036301e-01 7.81867862e-01
3.82631272e-01 4.01697665e-01 4.07008350e-01 1.10396123e+00
-1.59274027e-01 9.11917835e-02 -3.50574672e-01 3.18558291e-02
9.49793458e-01 9.69587624e-01 -2.81454504e-01 -7.17483044e-01
-1.66498825e-01 5.81810057e-01 4.76413399e-01 2.07487404e-01
-7.28213727e-01 -7.10976183e-01 5.63492358e-01 -7.67212987e-01
1.29661426e-01 3.96804720e-01 -2.45588884e-01 -9.85006273e-01
-1.44281862e-02 -3.10228746e-02 6.71979725e-01 -7.65607774e-01
-1.44587815e+00 2.13255048e-01 1.55697942e-01 -1.20747757e+00
-4.55806643e-01 -4.82034981e-02 -8.75604630e-01 6.46330416e-01
-1.23094463e+00 -4.64743316e-01 -2.12239787e-01 2.25476995e-01
5.16933143e-01 -2.11785257e-01 8.98288429e-01 8.81104916e-03
-6.14723682e-01 5.76354265e-01 4.07008260e-01 -9.40541625e-02
7.23795176e-01 -1.16335392e+00 -2.51511395e-01 3.88517618e-01
7.62883425e-02 5.88747025e-01 5.93479097e-01 -8.25331450e-01
-9.56463635e-01 -1.06919146e+00 7.64977038e-01 -3.35303545e-01
5.85950255e-01 2.96938986e-01 -9.53609943e-01 3.09214592e-01
-2.16090262e-01 1.69442922e-01 8.93724680e-01 2.50414014e-01
-2.65204221e-01 -3.76660377e-01 -1.38804662e+00 4.31615889e-01
7.07825184e-01 -5.33751428e-01 -1.06358945e+00 4.94061828e-01
1.30601680e+00 -3.02805066e-01 -9.40150201e-01 7.40826845e-01
7.12750673e-01 -7.13933051e-01 7.59383082e-01 -5.70640504e-01
6.80865673e-03 -4.91569579e-01 -4.26099807e-01 -1.46061039e+00
-2.11261213e-01 4.29728217e-02 -4.41128284e-01 1.18913031e+00
7.72169948e-01 -5.68143785e-01 8.02584827e-01 1.10292017e+00
4.84424643e-02 -9.35552716e-01 -9.57744837e-01 -5.23649216e-01
-5.17802775e-01 -6.23627663e-01 8.00251126e-01 7.99477637e-01
3.14607292e-01 2.88933218e-01 -2.22161993e-01 2.15678811e-02
8.23076606e-01 2.03571096e-01 2.52085835e-01 -1.42831147e+00
-3.75565290e-01 -3.27400029e-01 -5.72014928e-01 -4.43283170e-01
1.30391195e-01 -7.42036998e-01 3.73501897e-01 -1.47101068e+00
4.41311151e-01 -5.39284408e-01 -8.34205508e-01 2.89631397e-01
-4.92372483e-01 -1.98152021e-01 5.17183244e-02 2.92658806e-01
-1.06075907e+00 8.71915042e-01 7.85399556e-01 -4.46815759e-01
-5.28462470e-01 3.37345183e-01 -4.13459867e-01 6.47133529e-01
6.49366319e-01 -6.56361997e-01 -6.82596803e-01 -2.14398772e-01
1.20433159e-01 4.57344726e-02 -2.57233828e-01 -1.11771119e+00
8.75465274e-01 -3.39092940e-01 3.76823962e-01 -8.64077687e-01
4.44557637e-01 -7.12649584e-01 1.20900325e-01 4.25226808e-01
-6.46028221e-01 -6.08292162e-01 -2.20278323e-01 1.07568085e+00
-3.25393766e-01 -6.41486824e-01 5.57214260e-01 -3.15948278e-02
-8.62938583e-01 5.14704168e-01 -2.45770793e-02 -1.20841578e-01
1.05310941e+00 3.77887897e-02 -1.17742315e-01 -2.59430200e-01
-9.87393558e-01 6.63359702e-01 1.57754426e-03 5.32813668e-01
9.93610322e-01 -1.67078996e+00 -5.01324475e-01 1.57848001e-01
4.56342906e-01 3.18065360e-02 1.16804257e-01 5.36214769e-01
1.66136220e-01 3.67869794e-01 3.06364655e-01 -6.39700830e-01
-1.04920352e+00 5.64088464e-01 2.71385908e-01 -1.48368359e-01
3.05067241e-01 1.05105078e+00 -2.40562439e-01 -6.83316231e-01
5.51295221e-01 -1.99320629e-01 -5.27728319e-01 3.68597478e-01
5.52054822e-01 5.49866378e-01 -2.99155060e-02 -1.46507874e-01
-3.34213495e-01 4.58712935e-01 -2.30075583e-01 -2.90354669e-01
1.03346455e+00 1.84572447e-04 -1.93801075e-01 8.11917126e-01
1.22693503e+00 -5.89650393e-01 -8.97539020e-01 -9.14538741e-01
1.51343554e-01 -5.73504925e-01 2.62852937e-01 -9.04008150e-01
-8.81943107e-01 5.14869392e-01 8.57699394e-01 -5.14524132e-02
8.61089528e-01 1.44784171e-02 3.47181141e-01 8.70384753e-01
5.36074877e-01 -1.35248661e+00 3.94987222e-03 6.24934554e-01
6.06604874e-01 -1.50984645e+00 1.36743203e-01 -3.43054235e-01
-5.66853166e-01 8.71060252e-01 8.91307175e-01 7.27716982e-02
1.03686392e+00 -2.02933773e-01 -3.08650196e-01 9.75692272e-02
-1.15164304e+00 -4.80811179e-01 6.05868638e-01 3.81391227e-01
1.53212920e-01 2.88787901e-01 -4.79931474e-01 8.97699237e-01
2.23456666e-01 1.81219026e-01 -7.88032040e-02 9.87440407e-01
-9.39258695e-01 -8.16401422e-01 -2.20110700e-01 9.86725569e-01
7.32513517e-02 -1.68934479e-01 -3.12359124e-01 -3.65803670e-03
2.02558041e-01 1.23493516e+00 9.76271033e-02 -9.16081965e-01
3.41892928e-01 2.98948854e-01 3.03435504e-01 -1.01894760e+00
-7.43087009e-02 -3.69962662e-01 -2.57437080e-02 -4.01387393e-01
-2.28716239e-01 -4.08193529e-01 -1.40781105e+00 2.84898337e-02
-1.04936552e+00 6.09321415e-01 7.71915913e-01 1.26369250e+00
4.75245655e-01 3.59855413e-01 8.52817774e-01 -5.79235256e-01
-9.18364048e-01 -1.08528376e+00 -4.77131248e-01 1.28207400e-01
-1.95046306e-01 -9.68480647e-01 -6.16093159e-01 -4.48304266e-01]
|
[10.136016845703125, 3.3736631870269775]
|
bda6d78e-7f84-47d1-805a-89b93c4d35e6
|
graph-based-3d-multi-person-pose-estimation
|
2109.05885
| null |
https://arxiv.org/abs/2109.05885v1
|
https://arxiv.org/pdf/2109.05885v1.pdf
|
Graph-Based 3D Multi-Person Pose Estimation Using Multi-View Images
|
This paper studies the task of estimating the 3D human poses of multiple persons from multiple calibrated camera views. Following the top-down paradigm, we decompose the task into two stages, i.e. person localization and pose estimation. Both stages are processed in coarse-to-fine manners. And we propose three task-specific graph neural networks for effective message passing. For 3D person localization, we first use Multi-view Matching Graph Module (MMG) to learn the cross-view association and recover coarse human proposals. The Center Refinement Graph Module (CRG) further refines the results via flexible point-based prediction. For 3D pose estimation, the Pose Regression Graph Module (PRG) learns both the multi-view geometry and structural relations between human joints. Our approach achieves state-of-the-art performance on CMU Panoptic and Shelf datasets with significantly lower computation complexity.
|
['Wanli Ouyang', 'Dong Liu', 'Chen Qian', 'Lei Bai', 'Wentao Liu', 'Sheng Jin', 'Size Wu']
|
2021-09-13
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Wu_Graph-Based_3D_Multi-Person_Pose_Estimation_Using_Multi-View_Images_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Wu_Graph-Based_3D_Multi-Person_Pose_Estimation_Using_Multi-View_Images_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['3d-pose-estimation', '3d-multi-person-pose-estimation']
|
['computer-vision', 'computer-vision']
|
[-2.13003278e-01 -4.94015180e-02 8.67401361e-02 -3.56903195e-01
-7.75846481e-01 -3.91338736e-01 4.54505324e-01 -4.42644246e-02
-4.92121220e-01 3.38880986e-01 3.11256379e-01 3.98662627e-01
1.18133172e-01 -6.35286510e-01 -8.15390170e-01 -3.47197771e-01
-4.37485203e-02 1.09891903e+00 1.25870690e-01 -1.40792698e-01
-1.03669852e-01 6.23070955e-01 -1.21556234e+00 -1.39644267e-02
1.10208958e-01 6.02399349e-01 -2.98935950e-01 1.06565034e+00
6.30450368e-01 3.38369280e-01 -3.76384467e-01 -5.45841336e-01
4.80880439e-01 1.03925005e-01 -6.61757708e-01 5.39720058e-01
9.14778411e-01 -5.10667145e-01 -5.53728640e-01 8.32506359e-01
8.09244812e-01 2.13420331e-01 4.91286635e-01 -1.22730494e+00
-2.29746506e-01 9.01072249e-02 -8.94280434e-01 -1.27661094e-01
1.01724207e+00 -6.64477423e-02 8.83747697e-01 -1.14538252e+00
7.79907048e-01 1.64450490e+00 9.94298160e-01 5.67261219e-01
-1.05649233e+00 -4.10921454e-01 4.37861204e-01 4.15777788e-02
-1.72447062e+00 -2.62974381e-01 7.09351242e-01 -5.39845049e-01
1.03716826e+00 -1.80237480e-02 1.16939330e+00 1.17401707e+00
3.84109408e-01 5.81033170e-01 7.20153809e-01 -2.28882000e-01
-1.13632329e-01 -3.15572798e-01 1.44356444e-01 1.16142666e+00
5.52097738e-01 2.17350945e-02 -6.65366352e-01 -1.94218531e-01
1.21596646e+00 2.93231666e-01 -1.54066071e-01 -8.67991030e-01
-1.25180495e+00 7.97363102e-01 5.15898108e-01 -3.29449326e-01
-4.98984247e-01 4.22941446e-01 2.47766465e-01 5.70920371e-02
5.30324697e-01 4.83610220e-02 -4.40126002e-01 3.20632309e-01
-6.15819693e-01 7.04433262e-01 9.13858712e-01 1.09513342e+00
5.54199994e-01 -3.42390120e-01 -2.88085695e-02 6.85128629e-01
6.94294155e-01 5.54326117e-01 -2.46058106e-01 -1.01755142e+00
7.30755925e-01 7.44291127e-01 2.52974957e-01 -1.16280389e+00
-8.09071779e-01 -5.55132270e-01 -9.02977884e-01 4.30526435e-02
6.09926045e-01 -2.54618526e-01 -1.03833091e+00 1.57699633e+00
8.03152263e-01 -1.03718154e-02 -3.86396140e-01 1.28272367e+00
8.74655366e-01 3.47394317e-01 -3.48360427e-02 2.51169235e-01
1.66873014e+00 -1.16979599e+00 -2.27172941e-01 -6.42600596e-01
2.74874985e-01 -4.60142553e-01 3.74413818e-01 1.89703554e-01
-1.38143051e+00 -7.87486255e-01 -9.96185780e-01 -4.02002752e-01
-1.84512347e-01 3.13105673e-01 3.88685286e-01 5.11661947e-01
-1.09457994e+00 3.75780612e-01 -1.00841105e+00 -6.50356174e-01
4.19282876e-02 6.50823891e-01 -7.95825064e-01 -2.64199048e-01
-9.68280494e-01 8.23874533e-01 1.08897619e-01 3.48659843e-01
-8.17724347e-01 -2.90294468e-01 -1.27512407e+00 -2.20674559e-01
5.41565537e-01 -1.71670687e+00 9.73535419e-01 -2.18831658e-01
-1.30136228e+00 1.15068460e+00 -6.03720136e-02 -1.66734889e-01
7.55948901e-01 -6.37796044e-01 1.17839854e-02 2.13286921e-01
1.84741080e-01 7.95857251e-01 9.47653592e-01 -1.30219913e+00
-5.50908089e-01 -9.99739110e-01 9.32882726e-02 7.04340875e-01
3.78177375e-01 -1.32240623e-01 -1.38080442e+00 -4.14892256e-01
5.69059789e-01 -1.31428158e+00 -5.12201488e-01 1.47292033e-01
-6.62670016e-01 -2.37092301e-01 2.76036263e-01 -1.01987934e+00
7.22743273e-01 -1.55063617e+00 7.37116754e-01 4.43252802e-01
5.65781653e-01 -4.19922292e-01 2.63534896e-02 2.76969910e-01
1.14011355e-01 -5.82909942e-01 3.00416380e-01 -9.77129221e-01
1.02897041e-01 -4.60913964e-02 4.07779723e-01 9.71978962e-01
-1.79528430e-01 1.09132397e+00 -5.81780374e-01 -4.55466926e-01
2.64688700e-01 8.05172920e-01 -6.74985707e-01 3.08406234e-01
1.79320365e-01 4.12383378e-01 -3.92877877e-01 7.27916956e-01
5.54299951e-01 -5.98597527e-01 3.99924695e-01 -4.31034982e-01
2.55225122e-01 -2.38043070e-02 -1.60246396e+00 2.11615825e+00
-8.96017998e-02 -1.50201255e-02 2.99902409e-01 -5.47196746e-01
6.19763255e-01 2.40508527e-01 5.86474419e-01 1.00251243e-01
1.77913934e-01 -3.47432047e-01 -5.24151683e-01 -1.84951574e-01
6.13804996e-01 2.13205621e-01 -4.44926023e-01 3.13802510e-01
2.68406004e-01 3.13338518e-01 -7.71944597e-02 2.02946067e-01
7.77607977e-01 7.04865515e-01 5.74873567e-01 -1.97975487e-02
5.22934735e-01 -2.38354251e-01 4.78715003e-01 5.70550621e-01
-2.23286659e-01 6.86752379e-01 1.58559531e-01 -8.86863828e-01
-1.07017720e+00 -1.30223346e+00 6.09507143e-01 1.16774881e+00
3.70397002e-01 -5.69573879e-01 -9.24135327e-01 -5.96808732e-01
2.82202691e-01 -1.80699587e-01 -7.16152668e-01 1.38726026e-01
-9.88632262e-01 -4.22909558e-01 3.01290661e-01 7.30609179e-01
5.27760923e-01 -5.14560223e-01 -6.70206845e-01 1.82425343e-02
-3.55937511e-01 -1.36459637e+00 -1.07204485e+00 -4.01045740e-01
-6.93500519e-01 -1.27104115e+00 -9.40665901e-01 -7.89475083e-01
8.80953789e-01 4.84103143e-01 1.28142405e+00 1.57004166e-02
-1.45996496e-01 8.07570636e-01 1.56546444e-01 3.33299637e-02
1.40012354e-01 1.31336063e-01 5.13841212e-01 -8.19972008e-02
3.46488506e-01 -4.39469874e-01 -7.82293081e-01 3.19851935e-01
6.77891672e-02 1.28337577e-01 5.33485591e-01 4.88143206e-01
8.29120636e-01 -2.86325574e-01 -1.28425866e-01 -7.45593786e-01
3.43783945e-01 1.49922207e-01 -5.97407877e-01 3.08508188e-01
-1.38076827e-01 -1.70521542e-01 1.13380350e-01 -2.45923296e-01
-7.35368550e-01 6.09318495e-01 -5.39534055e-02 -5.72508156e-01
-3.99226785e-01 7.07854554e-02 -3.33967000e-01 -1.99804291e-01
4.13303137e-01 6.35830387e-02 -1.64884284e-01 -4.42178369e-01
4.48469520e-01 1.35168508e-01 7.13423073e-01 -3.78500730e-01
9.87980068e-01 4.96285290e-01 9.60536599e-02 -7.20510840e-01
-9.14156377e-01 -7.24680841e-01 -1.26869857e+00 -5.60201526e-01
1.41025031e+00 -1.54477966e+00 -1.20961022e+00 5.11624217e-01
-1.36176169e+00 -9.46645662e-02 1.39958367e-01 5.22672236e-01
-5.66077113e-01 5.29598653e-01 -9.39961076e-01 -7.31871486e-01
-5.09312510e-01 -1.01830423e+00 1.84247863e+00 -4.26864959e-02
-3.27588081e-01 -8.76065612e-01 1.50865018e-01 6.71537519e-01
-4.19117033e-01 4.47579890e-01 1.03403658e-01 -2.75471598e-01
-8.51383924e-01 -5.01406670e-01 -8.17117915e-02 -3.31649154e-01
-4.09683317e-01 -6.60262644e-01 -7.53040254e-01 -8.79866958e-01
-2.62907773e-01 -1.52466759e-01 7.21881151e-01 6.81861818e-01
5.03071129e-01 -9.90118831e-02 -6.86077595e-01 9.94405091e-01
1.08155930e+00 -4.96355146e-01 2.08441660e-01 2.22527891e-01
1.23768711e+00 5.93928754e-01 3.66899639e-01 4.14958030e-01
9.43189085e-01 1.00744927e+00 2.13650733e-01 7.56298453e-02
-1.81241065e-01 -4.94497359e-01 3.05142164e-01 6.27434790e-01
-6.17871046e-01 -1.59658998e-01 -8.25515389e-01 6.66224509e-02
-2.00179553e+00 -8.01065266e-01 1.74949095e-02 2.09100676e+00
6.95781112e-02 2.53596127e-01 5.65222144e-01 -3.14127058e-01
9.60625470e-01 2.09359273e-01 -3.60697180e-01 3.17718595e-01
3.60385805e-01 -3.86066884e-01 4.95452017e-01 7.57705331e-01
-1.41691685e+00 9.29778695e-01 6.06958389e+00 1.34184763e-01
-3.00307423e-01 7.55139217e-02 2.02311486e-01 -1.30399853e-01
2.70179152e-01 -3.60960484e-01 -1.32605553e+00 -1.99447781e-01
4.11978722e-01 6.61705613e-01 4.83804494e-01 7.56626844e-01
-1.36900961e-01 6.00596219e-02 -1.25138402e+00 1.41275620e+00
4.30208415e-01 -1.05133128e+00 1.44202057e-02 2.23446667e-01
5.29580951e-01 -8.83204862e-03 -3.53508621e-01 1.69852957e-01
3.59487802e-01 -6.26012027e-01 8.01074266e-01 7.84061372e-01
5.91916800e-01 -9.00532544e-01 7.12616980e-01 4.26282316e-01
-1.73787713e+00 1.53041512e-01 -2.35640734e-01 3.10047306e-02
5.01277745e-01 2.72671223e-01 -5.70419669e-01 7.19280481e-01
8.44576478e-01 7.10798264e-01 -5.41044831e-01 7.50155985e-01
-3.90388101e-01 -2.59650737e-01 -3.00757080e-01 2.80346572e-01
-2.28820577e-01 -2.52673596e-01 7.40139902e-01 1.07096624e+00
1.79904684e-01 7.62591735e-02 7.98474073e-01 6.00226045e-01
-2.63308771e-02 -3.70297134e-01 -4.34977621e-01 6.82873487e-01
3.03299010e-01 1.35792482e+00 -7.00651348e-01 -3.56164515e-01
-3.46639246e-01 1.47214699e+00 6.79606557e-01 3.64908814e-01
-7.52552748e-01 9.09779221e-02 6.08769000e-01 1.78160265e-01
1.65018573e-01 -7.18191743e-01 -2.60282736e-02 -1.46104395e+00
1.90632753e-02 -7.38032103e-01 6.75274014e-01 -7.53537297e-01
-1.30486560e+00 4.46204633e-01 1.03788592e-01 -9.65478718e-01
-4.79380250e-01 -6.04989469e-01 -1.44195244e-01 8.66297007e-01
-7.84606516e-01 -1.80359960e+00 -5.51150441e-01 8.32803667e-01
5.49385548e-01 -1.25406533e-02 7.34987497e-01 1.12480015e-01
-5.82784474e-01 5.41665971e-01 -7.88983762e-01 5.38982451e-01
6.93612874e-01 -1.30732632e+00 1.06900084e+00 8.82674932e-01
1.04437090e-01 7.07064450e-01 4.48469013e-01 -1.14874971e+00
-1.66717231e+00 -1.19159639e+00 1.07111800e+00 -9.26982999e-01
1.26994446e-01 -7.69146860e-01 -2.33155295e-01 1.07954311e+00
-1.91638216e-01 4.33323998e-03 3.37731183e-01 4.61036891e-01
-4.46924001e-01 1.21075191e-01 -9.19894934e-01 6.12043262e-01
1.38386393e+00 -5.80178857e-01 -5.28190076e-01 3.47059160e-01
5.67670286e-01 -9.62203622e-01 -9.21584189e-01 5.75936995e-02
7.88890719e-01 -8.43039870e-01 1.65481913e+00 -5.96525490e-01
-1.25612766e-01 -4.13243026e-01 -1.72905982e-01 -9.88643289e-01
-6.93053007e-01 -4.93178725e-01 -5.17880499e-01 5.92814028e-01
8.57707337e-02 -2.22378284e-01 1.19397140e+00 4.93393987e-01
1.62360117e-01 -6.15506172e-01 -8.31588984e-01 -3.71821254e-01
-5.82941711e-01 -2.94987112e-01 2.36959904e-01 5.25691152e-01
-3.26945662e-01 8.77499759e-01 -7.60640800e-01 6.98229015e-01
1.23624825e+00 1.59289852e-01 1.24986422e+00 -1.39647150e+00
-5.94740391e-01 1.27534986e-01 -4.74795073e-01 -1.42480671e+00
-3.02806846e-03 -6.67062044e-01 5.85968234e-02 -1.65961623e+00
4.96464968e-01 2.89906293e-01 1.98505864e-01 5.80474548e-02
-3.07234496e-01 4.54711825e-01 3.00303102e-01 2.07238019e-01
-9.25327599e-01 1.83719322e-01 1.20312548e+00 -2.72887964e-02
-2.58245945e-01 3.77565801e-01 -3.98452222e-01 1.06073821e+00
3.68358433e-01 -2.41001561e-01 -5.92467710e-02 -5.45661211e-01
1.70003548e-01 4.19501811e-01 1.03297865e+00 -1.20594633e+00
4.91362780e-01 2.08027914e-01 1.06218743e+00 -1.00739741e+00
9.52137291e-01 -6.83672667e-01 2.30202302e-01 7.18674481e-01
-1.59961894e-01 5.60448170e-01 -2.47865692e-01 8.85296583e-01
3.55220705e-01 4.98293072e-01 6.87911987e-01 -5.89958787e-01
-5.22992313e-01 8.14428270e-01 -8.91618654e-02 -1.47393465e-01
6.79461420e-01 -2.89283246e-01 1.19383179e-01 -5.37659943e-01
-1.23494458e+00 2.73104429e-01 4.61884201e-01 4.71974939e-01
8.22161078e-01 -1.38944399e+00 -6.69405937e-01 3.50061387e-01
-7.95794427e-02 1.06229700e-01 3.14181566e-01 7.68850446e-01
-5.66188931e-01 4.34381366e-01 6.27878588e-03 -8.64648879e-01
-1.47276199e+00 4.64668483e-01 5.03777802e-01 -5.54573119e-01
-8.23689282e-01 1.09821188e+00 1.75116077e-01 -8.37842405e-01
3.58024418e-01 -5.24608009e-02 -1.98513702e-01 -1.04019523e-01
4.11241412e-01 6.64783418e-01 -2.62380093e-01 -1.05640805e+00
-5.82669854e-01 1.06314337e+00 8.31577852e-02 -2.83878624e-01
1.37230432e+00 -4.48386937e-01 -2.76674330e-03 1.31774291e-01
1.04126799e+00 -1.71007916e-01 -1.45034742e+00 -1.41162753e-01
-2.32262760e-01 -2.81618088e-01 -2.94969201e-01 -4.15976197e-01
-8.05988908e-01 7.42163658e-01 5.70849478e-01 -2.95333713e-01
6.66934133e-01 1.75651908e-01 8.01245451e-01 5.99731922e-01
6.41686976e-01 -1.17954147e+00 1.98395818e-01 6.11106694e-01
9.06960130e-01 -1.01557052e+00 3.95358026e-01 -7.31859028e-01
-3.61752540e-01 9.16651666e-01 8.55112195e-01 -4.15536880e-01
4.85173821e-01 -6.01339117e-02 -9.71133858e-02 -4.23609167e-01
-4.69278634e-01 -1.36640146e-01 8.39578748e-01 6.39672101e-01
2.37934798e-01 1.05471916e-01 3.20950270e-01 5.18127918e-01
-1.85082480e-01 -2.53969580e-01 8.39107297e-03 6.62553191e-01
-5.27568400e-01 -8.33719194e-01 -7.41901994e-01 5.80406599e-02
-1.45712733e-01 3.23381513e-01 -5.76355338e-01 1.03604126e+00
1.63272873e-01 7.12260842e-01 -6.45786524e-02 -6.03721619e-01
7.53257692e-01 5.13453037e-02 8.88167739e-01 -5.38910389e-01
-5.44444501e-01 4.82411623e-01 2.93734580e-01 -9.76603210e-01
-5.71601510e-01 -8.20745409e-01 -9.07748222e-01 -4.14219767e-01
-1.30026087e-01 -2.11796910e-01 2.68691510e-01 1.01469696e+00
2.76601583e-01 4.10452008e-01 6.68866336e-02 -1.53717220e+00
-4.71709132e-01 -7.77299702e-01 -5.32084227e-01 4.24462646e-01
2.92692184e-01 -7.71985173e-01 2.83295006e-01 -4.77281865e-03]
|
[7.02625846862793, -0.9148258566856384]
|
72bbbbc5-577b-4b71-b44a-ed9f69e0a8ce
|
a-frustratingly-easy-approach-for-joint
|
2010.12812
| null |
https://arxiv.org/abs/2010.12812v2
|
https://arxiv.org/pdf/2010.12812v2.pdf
|
A Frustratingly Easy Approach for Entity and Relation Extraction
|
End-to-end relation extraction aims to identify named entities and extract relations between them. Most recent work models these two subtasks jointly, either by casting them in one structured prediction framework, or performing multi-task learning through shared representations. In this work, we present a simple pipelined approach for entity and relation extraction, and establish the new state-of-the-art on standard benchmarks (ACE04, ACE05 and SciERC), obtaining a 1.7%-2.8% absolute improvement in relation F1 over previous joint models with the same pre-trained encoders. Our approach essentially builds on two independent encoders and merely uses the entity model to construct the input for the relation model. Through a series of careful examinations, we validate the importance of learning distinct contextual representations for entities and relations, fusing entity information early in the relation model, and incorporating global context. Finally, we also present an efficient approximation to our approach which requires only one pass of both entity and relation encoders at inference time, achieving an 8-16$\times$ speedup with a slight reduction in accuracy.
|
['Danqi Chen', 'Zexuan Zhong']
|
2020-10-24
| null |
https://aclanthology.org/2021.naacl-main.5
|
https://aclanthology.org/2021.naacl-main.5.pdf
|
naacl-2021-4
|
['joint-entity-and-relation-extraction']
|
['natural-language-processing']
|
[ 1.67401940e-01 9.73568439e-01 -3.50344568e-01 -5.99791825e-01
-1.11454201e+00 -3.77152205e-01 7.71713674e-01 5.86259127e-01
-3.96246374e-01 9.75397408e-01 4.36207741e-01 -4.76411343e-01
-1.03423834e-01 -9.79164600e-01 -8.82171214e-01 -1.69036627e-01
-3.96256834e-01 7.79175758e-01 1.98892936e-01 -1.58619940e-01
-3.06313753e-01 3.04780066e-01 -1.15931714e+00 4.02300686e-01
5.53544819e-01 9.71530557e-01 -3.43389869e-01 8.80462766e-01
-1.89543054e-01 1.26080406e+00 -5.96941650e-01 -1.09642291e+00
5.13770431e-02 -3.82204317e-02 -1.59809387e+00 -3.16185534e-01
1.42052159e-01 -2.62745649e-01 -5.96143425e-01 4.53670949e-01
3.80963534e-01 -4.97767255e-02 4.33386803e-01 -1.02685118e+00
-7.15788603e-01 1.13355219e+00 -4.35029894e-01 1.69114515e-01
4.19038028e-01 -5.53893924e-01 1.67453027e+00 -7.95448303e-01
8.58104587e-01 8.93334866e-01 8.49515378e-01 1.96597651e-01
-1.32829785e+00 -5.29331386e-01 1.55302063e-01 1.79891735e-01
-1.54783916e+00 -6.81886494e-01 2.17483401e-01 -2.38919199e-01
1.93794692e+00 1.39296442e-01 1.40298560e-01 7.01572239e-01
4.70546745e-02 7.88328648e-01 5.99345446e-01 -5.19912422e-01
-2.09155530e-01 1.04508795e-01 3.25149298e-01 5.76586246e-01
5.44174671e-01 -1.33904470e-02 -3.70040178e-01 -2.43602172e-01
4.91944045e-01 -4.82814252e-01 -9.35691223e-02 -1.04078442e-01
-9.47945297e-01 5.40937960e-01 5.44568896e-01 3.68478626e-01
-3.49917799e-01 2.23510742e-01 4.02110845e-01 2.71516055e-01
7.47743368e-01 3.70033056e-01 -1.05988121e+00 -7.17891753e-03
-8.12313914e-01 3.20426404e-01 1.33864248e+00 1.27378237e+00
8.75748098e-01 -6.03555977e-01 -1.93583935e-01 6.55600309e-01
2.83586532e-01 -8.75858516e-02 5.96260801e-02 -4.80247647e-01
7.63824940e-01 7.00462461e-01 1.31322118e-02 -7.30815649e-01
-4.51132178e-01 -4.40160513e-01 -5.28591335e-01 -3.01013172e-01
2.83389062e-01 -4.78441626e-01 -8.26631665e-01 1.68873906e+00
4.58694398e-01 3.26620936e-01 4.42818224e-01 2.21258998e-01
1.13144875e+00 4.59948540e-01 4.44106042e-01 -8.09640959e-02
1.59727061e+00 -1.15675426e+00 -8.20361674e-01 -4.37579364e-01
1.16772890e+00 -7.21624434e-01 1.19106054e-01 -1.86293378e-01
-1.24592733e+00 -4.64349985e-01 -1.08332539e+00 -6.83653474e-01
-7.00551808e-01 4.07912672e-01 1.02365792e+00 3.55935186e-01
-8.71092141e-01 8.37735713e-01 -9.13447857e-01 -2.68866122e-01
4.23253626e-01 7.74428368e-01 -6.98731899e-01 1.76118314e-01
-1.44876683e+00 1.35809350e+00 6.56780839e-01 1.95784066e-02
-2.40840256e-01 -7.79296219e-01 -1.18604350e+00 2.53263354e-01
5.35908937e-01 -6.83714747e-01 1.31015229e+00 -2.96368808e-01
-1.24316132e+00 1.02581823e+00 -5.45408666e-01 -7.95856953e-01
-6.64974516e-03 -8.13379526e-01 -5.70880294e-01 -1.92298412e-01
1.85938552e-01 3.62025917e-01 3.84458825e-02 -1.14439940e+00
-7.54430056e-01 -1.71700835e-01 3.82002681e-01 2.08539274e-02
1.41293585e-01 5.51682949e-01 -5.24485171e-01 -3.84399414e-01
-1.35243997e-01 -7.35870779e-01 -3.77326965e-01 -4.77401257e-01
-7.09530234e-01 -6.25361621e-01 3.85512471e-01 -9.43961740e-01
1.40732861e+00 -1.80493987e+00 1.66817933e-01 9.94517561e-03
4.08107847e-01 4.00713593e-01 1.10088781e-01 6.46092355e-01
-5.06094992e-01 2.21679017e-01 -2.66503066e-01 -6.96984649e-01
-9.13357269e-03 2.63154268e-01 -1.26135781e-01 1.26484975e-01
9.12340403e-01 1.33645082e+00 -9.23992097e-01 -5.80450475e-01
-2.72769153e-01 4.83398765e-01 -3.12655866e-01 4.48123664e-01
-8.51759985e-02 1.06068682e-02 -2.61630684e-01 4.50355738e-01
4.28929418e-01 -5.77446580e-01 7.26267457e-01 -2.60433525e-01
1.45088166e-01 1.17450976e+00 -1.23693120e+00 1.53348577e+00
-6.66021168e-01 6.33964658e-01 -4.27308381e-02 -1.07519853e+00
9.32800293e-01 7.61495531e-01 5.22153378e-01 -2.88060099e-01
5.39098457e-02 4.46086854e-01 2.38910541e-02 -3.78184050e-01
5.35708487e-01 8.86914656e-02 -1.77067101e-01 3.29420716e-01
7.39119053e-01 2.05209956e-01 2.72350967e-01 2.38975897e-01
1.35748160e+00 5.45465887e-01 8.50257397e-01 -1.72097627e-02
4.64666724e-01 -2.62311786e-01 6.06811404e-01 4.36696827e-01
2.14951873e-01 3.25127900e-01 7.47579455e-01 -5.85109472e-01
-7.36460865e-01 -7.84591675e-01 -1.50334463e-01 1.13201523e+00
-1.85861737e-01 -9.92175937e-01 -3.53407532e-01 -1.06833625e+00
-3.10878549e-02 7.41035819e-01 -7.42894292e-01 -7.40615055e-02
-1.05890667e+00 -9.03462529e-01 8.10015202e-01 9.36037183e-01
3.29263270e-01 -8.92544210e-01 -3.15203846e-01 4.08543468e-01
-1.13924488e-01 -1.53748059e+00 2.78580219e-01 8.39002430e-01
-5.66206694e-01 -9.90662217e-01 -3.46952736e-01 -8.97167504e-01
3.11357915e-01 -3.72842640e-01 1.70428562e+00 -1.74721554e-02
-1.02450877e-01 -2.67282695e-01 -3.20232153e-01 -3.29682231e-01
-1.21884920e-01 6.68956816e-01 -4.33141589e-01 -3.09995532e-01
7.47013032e-01 -6.56336129e-01 -1.44433826e-01 -1.97168976e-01
-3.90682876e-01 1.21095456e-01 8.73326778e-01 7.45927632e-01
3.83653343e-01 -1.48727551e-01 5.88629723e-01 -1.53182065e+00
2.77997434e-01 -7.37789989e-01 -1.08513452e-01 5.16309440e-01
-7.22142875e-01 3.43756914e-01 3.53168517e-01 1.78689629e-01
-1.25773537e+00 3.01938951e-01 -4.24398571e-01 1.34062797e-01
-2.68087953e-01 6.23982251e-01 -3.32268953e-01 3.71129394e-01
4.95763302e-01 -3.43727112e-01 -4.83613789e-01 -6.05050862e-01
6.92017138e-01 6.08936906e-01 6.09009087e-01 -6.13289297e-01
8.08571577e-01 1.39672130e-01 -1.77474335e-01 -2.85935283e-01
-1.29449439e+00 -4.79191393e-01 -1.04039812e+00 5.80662131e-01
8.95771921e-01 -1.26003838e+00 -4.92857933e-01 2.33938848e-03
-1.50053418e+00 -2.29883149e-01 -4.00379092e-01 3.47282350e-01
-2.46648133e-01 9.68552828e-02 -1.05461204e+00 -5.07266164e-01
-5.56247652e-01 -7.69917130e-01 1.32828736e+00 1.46473780e-01
-4.89495248e-01 -1.01496112e+00 2.44093597e-01 2.98907936e-01
1.72650293e-01 1.92496002e-01 9.05019820e-01 -1.28614712e+00
-5.63581049e-01 -3.18320215e-01 -6.08815849e-01 -4.97232787e-02
1.00233391e-01 -1.35207251e-01 -1.17259681e+00 1.38699055e-01
-5.35564542e-01 -3.61703843e-01 9.77195859e-01 -1.21470638e-01
8.60429287e-01 -2.04818681e-01 -7.73760319e-01 5.94519556e-01
1.37183154e+00 2.31182575e-03 6.96930051e-01 1.81054801e-01
8.73843968e-01 6.79930806e-01 4.59952772e-01 9.87412408e-02
9.47846055e-01 7.24785507e-01 3.84700000e-02 -1.93148822e-01
-2.70081043e-01 -2.09673390e-01 -6.00847267e-02 5.96428394e-01
-4.46768433e-01 -2.34304205e-01 -8.78730357e-01 8.29837501e-01
-1.99714625e+00 -7.13127971e-01 -1.53707623e-01 1.85963154e+00
1.25325239e+00 2.20130622e-01 -1.06614254e-01 2.72991974e-02
4.18866217e-01 1.57229841e-01 -3.78397964e-02 -5.69045663e-01
5.25762774e-02 1.02460432e+00 6.85954809e-01 6.30414188e-01
-1.51055050e+00 1.25552201e+00 6.66550016e+00 5.09680808e-01
-6.19196117e-01 3.07208556e-03 6.36769652e-01 1.40062362e-01
-2.12121472e-01 4.52533692e-01 -1.13164794e+00 -1.30024344e-01
1.45767319e+00 -9.95249003e-02 -4.38048467e-02 8.03185046e-01
-4.62063283e-01 2.44815182e-02 -1.52996099e+00 4.36573476e-01
-1.98936105e-01 -1.39285576e+00 -3.77898842e-01 7.51820728e-02
4.72423881e-01 1.57079950e-01 -6.63088083e-01 5.66891074e-01
6.93246901e-01 -1.28821182e+00 2.79549748e-01 3.20473790e-01
7.77985275e-01 -6.01493716e-01 1.19995034e+00 -6.05503693e-02
-1.56698430e+00 2.79437184e-01 -1.19327836e-01 -1.96444109e-01
3.43296558e-01 7.07962453e-01 -8.60701203e-01 1.23797369e+00
4.14864689e-01 7.33767867e-01 -5.37193418e-01 5.31622112e-01
-4.64106143e-01 5.19383967e-01 -3.57741088e-01 1.91639215e-01
-5.70070781e-02 2.66951531e-01 1.39620349e-01 1.69053936e+00
-1.21196441e-01 1.76658094e-01 8.93533975e-02 6.36625290e-01
-3.84043604e-01 9.30485949e-02 -4.34502244e-01 -3.92869823e-02
5.22791147e-01 1.31184161e+00 -3.67636591e-01 -5.88515162e-01
-8.25664878e-01 8.37679684e-01 9.81576025e-01 1.65965110e-01
-8.50723386e-01 -7.99416304e-01 5.21707118e-01 -2.32473075e-01
6.83031499e-01 -1.60511017e-01 -4.98381615e-01 -1.19807899e+00
8.45601782e-02 -5.12311339e-01 3.80445838e-01 -1.59384742e-01
-1.05260909e+00 8.43151510e-01 1.02528550e-01 -5.32972932e-01
-6.09275103e-01 -4.65028942e-01 -5.08802176e-01 1.14127731e+00
-1.62221670e+00 -1.43859041e+00 5.26896119e-02 1.16071835e-01
7.14271516e-02 2.16431811e-01 1.35910678e+00 5.87778270e-01
-8.53630722e-01 7.83558547e-01 -3.36011022e-01 7.50255644e-01
6.71508610e-01 -1.60262632e+00 9.77256179e-01 7.51949668e-01
4.83784407e-01 8.29881370e-01 3.94512087e-01 -5.99335730e-01
-1.01202643e+00 -1.07140398e+00 1.98888326e+00 -6.20990455e-01
7.88107753e-01 -4.35934097e-01 -9.07397866e-01 1.36148095e+00
5.47487080e-01 3.94738972e-01 1.01582229e+00 1.04222655e+00
-5.35930812e-01 9.56455693e-02 -8.55314136e-01 2.43496835e-01
1.18783939e+00 -4.82901961e-01 -7.55460620e-01 2.05429882e-01
9.69874680e-01 -5.69915950e-01 -1.50428271e+00 6.97106898e-01
4.95347679e-01 -6.36485279e-01 9.76266921e-01 -8.62987697e-01
6.66093528e-01 -6.34051114e-02 -1.61165923e-01 -9.90009487e-01
-4.28305417e-01 -4.80754077e-01 -8.87355685e-01 1.51786220e+00
1.00606060e+00 -4.82759804e-01 7.48716772e-01 7.60843694e-01
6.54810220e-02 -1.34259832e+00 -5.49450219e-01 -3.87800187e-01
3.15235034e-02 -2.89209545e-01 6.39575005e-01 8.38611722e-01
1.37748957e-01 9.86530960e-01 -1.88895777e-01 3.51373911e-01
2.19615936e-01 3.62062097e-01 7.45356321e-01 -1.15076065e+00
-6.40400231e-01 -1.24425925e-01 -3.11319798e-01 -1.08086932e+00
4.01053786e-01 -8.12526464e-01 6.92686124e-04 -1.80491936e+00
5.43870740e-02 -5.67051113e-01 -3.40565383e-01 7.81707644e-01
-5.46489418e-01 2.44014561e-02 -1.22000292e-01 2.84683760e-02
-6.35128796e-01 2.19324127e-01 5.08188784e-01 9.29205939e-02
-1.31268248e-01 3.16953138e-02 -9.84856606e-01 5.34390330e-01
5.82275450e-01 -4.76201475e-01 -2.36932471e-01 -5.00861228e-01
3.26203942e-01 -9.16051120e-02 8.39728639e-02 -7.13362455e-01
2.36413360e-01 3.03174675e-01 4.54835415e-01 -3.41748625e-01
3.96924317e-01 -5.02006233e-01 -9.28212404e-02 3.82401869e-02
-5.39746165e-01 -1.97329804e-01 2.16694355e-01 4.54191178e-01
-4.33828920e-01 -2.09844872e-01 2.76402056e-01 -7.72971958e-02
-5.76808035e-01 1.63511127e-01 8.20291713e-02 1.34971187e-01
1.01857901e+00 3.75855446e-01 -1.97730407e-01 -1.07514657e-01
-1.07267523e+00 2.60809094e-01 -1.88623294e-01 2.67277718e-01
6.58802688e-02 -1.03972816e+00 -9.20121431e-01 -1.99512206e-02
4.20065448e-02 4.33365136e-01 -1.56059027e-01 7.24352658e-01
-3.23348582e-01 5.47186971e-01 3.32290500e-01 -8.12890753e-02
-1.24798143e+00 4.39189672e-01 2.59441197e-01 -1.13647676e+00
-5.54837167e-01 1.25605893e+00 -1.76216424e-01 -5.12414634e-01
7.20549980e-03 -4.00268227e-01 -2.64151692e-01 1.20036758e-01
3.55792820e-01 6.08085096e-02 4.18808848e-01 -5.60445368e-01
-6.78765416e-01 1.63214371e-01 -4.49033558e-01 7.55634680e-02
1.46116376e+00 7.82009140e-02 -2.19363734e-01 3.03011239e-01
1.41395390e+00 2.49905929e-01 -7.95645118e-01 -5.84346533e-01
6.93938673e-01 7.71965925e-03 -2.04263508e-01 -8.40243459e-01
-8.33437264e-01 5.87233424e-01 -2.43990198e-01 3.61264050e-01
8.93220723e-01 5.08481026e-01 9.89638865e-01 3.78068328e-01
3.23176622e-01 -7.07314014e-01 -6.66805565e-01 5.65563142e-01
5.60759187e-01 -1.29625487e+00 2.98814833e-01 -9.51291263e-01
-4.94735032e-01 8.57213676e-01 5.58284223e-01 -3.71566474e-01
8.69122982e-01 8.19865704e-01 -2.40353703e-01 -2.09272563e-01
-1.13739753e+00 -6.40191436e-01 4.08008933e-01 3.44511867e-01
1.20059204e+00 2.23926648e-01 -3.67918700e-01 8.70002210e-01
-2.10862279e-01 -1.31823361e-01 -5.06667644e-02 1.07398593e+00
-2.97279228e-02 -1.58343208e+00 3.94024104e-01 4.92120385e-01
-8.90997231e-01 -4.56311047e-01 -6.10198557e-01 8.37280393e-01
2.74089843e-01 9.28220987e-01 1.13252081e-01 -4.62911814e-01
3.60202223e-01 4.42494124e-01 4.46918756e-01 -9.46490109e-01
-7.63833284e-01 -2.05163166e-01 1.00924516e+00 -6.50194287e-01
-5.74726045e-01 -6.53317332e-01 -1.34856570e+00 -2.91413255e-02
-5.58138669e-01 1.99696273e-01 3.13248158e-01 1.28441453e+00
7.57645309e-01 8.08756709e-01 2.42425159e-01 -4.45824951e-01
-1.75147668e-01 -1.19022202e+00 -9.94694233e-02 1.64560527e-01
6.45376667e-02 -5.74127913e-01 1.26046807e-01 2.81274356e-02]
|
[9.376158714294434, 8.782285690307617]
|
61c8102a-e32d-46f2-9ec0-dc30a9cc8bb2
|
augmenting-multi-turn-text-to-sql-datasets
|
2210.12096
| null |
https://arxiv.org/abs/2210.12096v1
|
https://arxiv.org/pdf/2210.12096v1.pdf
|
Augmenting Multi-Turn Text-to-SQL Datasets with Self-Play
|
The task of context-dependent text-to-SQL aims to convert multi-turn user utterances to formal SQL queries. This is a challenging task due to both the scarcity of training data from which to learn complex contextual dependencies and to generalize to unseen databases. In this paper we explore augmenting the training datasets using self-play, which leverages contextual information to synthesize new interactions to adapt the model to new databases. We first design a SQL-to-text model conditioned on a sampled goal query, which represents a user's intent, that then converses with a text-to-SQL semantic parser to generate new interactions. We then filter the synthesized interactions and retrain the models with the augmented data. We find that self-play improves the accuracy of a strong baseline on SParC and CoSQL, two widely used cross-domain text-to-SQL datasets. Our analysis shows that self-play simulates various conversational thematic relations, enhances cross-domain generalization and improves beam-search.
|
['Linfeng Song', 'Phil Blunsom', 'Tao Yu', 'Zihuiwen Ye', 'Qi Liu']
|
2022-10-21
| null | null | null | null |
['sql-to-text', 'text-to-sql']
|
['computer-code', 'computer-code']
|
[ 2.70824075e-01 5.20256996e-01 -1.12320401e-01 -1.14129972e+00
-1.30129778e+00 -8.87994945e-01 7.22264111e-01 1.24373786e-01
-2.78368592e-01 8.02780926e-01 6.04619145e-01 -4.24567312e-01
1.85923815e-01 -9.78402793e-01 -1.08286715e+00 1.74086675e-01
1.66667670e-01 1.25339627e+00 3.74308228e-01 -7.09137321e-01
-2.84945875e-01 1.12881005e-01 -1.66708088e+00 1.21838188e+00
9.54023540e-01 6.77839756e-01 2.35458791e-01 7.90617824e-01
-5.63497901e-01 9.98648643e-01 -6.50782347e-01 -5.80170214e-01
-3.58019359e-02 -4.36015576e-01 -1.24794841e+00 -1.84552118e-01
1.55751199e-01 -1.53437436e-01 7.82941654e-02 3.20496798e-01
4.18353558e-01 2.48271957e-01 3.13142031e-01 -1.10527635e+00
-5.07435203e-01 1.07201815e+00 3.60229671e-01 -1.97971296e-02
9.31708634e-01 1.11154489e-01 1.16525221e+00 -7.96939552e-01
1.22019684e+00 1.54303157e+00 5.90518951e-01 7.89557457e-01
-1.35672927e+00 -4.60068226e-01 3.88963073e-02 -1.66572303e-01
-1.02730525e+00 -6.50710404e-01 5.39173245e-01 -8.83945376e-02
1.37256205e+00 6.06457889e-01 2.81849027e-01 1.51962614e+00
-4.55497921e-01 7.34512687e-01 7.55075097e-01 -6.40334547e-01
1.83308497e-01 5.85304797e-01 7.04810694e-02 4.54635352e-01
-3.94443393e-01 -6.38184696e-02 -8.38067710e-01 -1.81712300e-01
7.50326738e-02 -3.70563984e-01 3.43521833e-02 -1.02695331e-01
-8.34376395e-01 7.42806435e-01 2.36247659e-01 3.97393778e-02
-2.08327755e-01 -2.90467948e-01 5.54248989e-01 5.52088559e-01
4.58447903e-01 9.83952999e-01 -8.42551649e-01 -3.92407119e-01
-3.37723762e-01 9.18916464e-01 1.48763382e+00 1.45514882e+00
8.41636777e-01 -4.98050183e-01 -8.09573680e-02 1.10355270e+00
-1.49111822e-01 4.47340041e-01 4.53017086e-01 -1.07977915e+00
8.83030176e-01 8.64738524e-01 1.20758740e-02 -4.02011633e-01
-3.64645213e-01 -2.45287269e-01 -3.01296394e-02 -4.86246347e-01
3.46989214e-01 -3.26978028e-01 -6.81755304e-01 1.79781365e+00
4.19763207e-01 -4.01648432e-01 5.78125179e-01 3.84551704e-01
1.07414103e+00 6.45344317e-01 3.10888678e-01 -1.13478079e-02
1.25205493e+00 -7.67391384e-01 -4.25649643e-01 -6.48433745e-01
1.05979300e+00 -5.21135628e-01 1.68135393e+00 2.18717456e-01
-1.05766881e+00 -6.24581099e-01 -7.03405142e-01 -5.01848996e-01
-6.86175108e-01 -3.26873064e-01 5.62206686e-01 3.69814306e-01
-6.78700566e-01 3.10432255e-01 -7.55422831e-01 -7.92231321e-01
-1.35467043e-02 1.18552642e-02 -2.70705849e-01 -3.07869196e-01
-1.40149438e+00 8.25390279e-01 7.16607213e-01 -5.54215968e-01
-4.09117967e-01 -9.38707709e-01 -8.26879740e-01 -1.15557514e-01
8.74086261e-01 -7.05800414e-01 1.72897565e+00 -6.25425279e-01
-1.57915211e+00 8.38314116e-01 -5.08792698e-01 -4.56268072e-01
4.18184966e-01 -2.38434181e-01 -3.78927797e-01 -2.31054723e-01
3.65488172e-01 7.13519752e-01 8.89463723e-02 -1.22811496e+00
-7.89982378e-01 -5.42056918e-01 2.03381807e-01 4.74814236e-01
-2.39991900e-02 -7.01084137e-02 -8.41936588e-01 -3.09462458e-01
2.92514525e-02 -9.30292964e-01 4.57526883e-03 -8.02439928e-01
-6.28184974e-01 -4.54464734e-01 6.78721488e-01 -5.56506455e-01
1.21329832e+00 -1.91003478e+00 1.68761581e-01 1.79983959e-01
-1.70466945e-01 -3.14847112e-01 -1.01783320e-01 8.56728792e-01
-7.23656043e-02 1.38856411e-01 -5.14700562e-02 -4.32839066e-01
8.61420110e-03 6.54223680e-01 -5.59746206e-01 -5.50009966e-01
2.93464899e-01 1.07734644e+00 -6.22874439e-01 -5.47421098e-01
-1.73790619e-01 -6.43057674e-02 -1.11628926e+00 7.46392310e-01
-1.21630991e+00 3.89698207e-01 -5.87514639e-01 4.62030500e-01
3.07830215e-01 -2.83720434e-01 6.48969650e-01 -1.75127715e-01
1.73048779e-01 8.36794913e-01 -8.22156012e-01 1.89019263e+00
-8.38153124e-01 2.22079217e-01 -1.90009087e-01 -9.58148658e-01
9.98032451e-01 5.93008697e-02 1.46562353e-01 -9.11419213e-01
-3.36262524e-01 2.56372005e-01 -4.93757039e-01 -1.02338791e+00
5.71741581e-01 -2.56238014e-01 -6.50578618e-01 3.33514333e-01
2.33885035e-01 -5.54794312e-01 1.44004568e-01 4.65851873e-01
1.15387011e+00 2.83127993e-01 5.54610677e-02 6.91132396e-02
2.24773958e-01 4.93690461e-01 3.48163545e-01 8.84797335e-01
5.72101593e-01 3.07137489e-01 6.53240919e-01 -3.61213952e-01
-7.47844517e-01 -1.01848137e+00 1.29088789e-01 1.86625528e+00
-2.85139531e-01 -6.47504628e-01 -6.63275599e-01 -9.76934433e-01
2.94928789e-01 1.43717694e+00 -4.41559553e-01 -1.28193805e-02
-7.06588507e-01 -4.20985729e-01 5.48719168e-01 4.56420273e-01
3.08755368e-01 -1.16010165e+00 -3.82953137e-01 4.70271081e-01
-6.64076209e-01 -1.52587199e+00 -1.30989224e-01 6.38694704e-01
-7.14497507e-01 -9.50628042e-01 2.62625396e-01 -6.73522353e-01
1.40085146e-01 -4.34201986e-01 1.70866203e+00 -1.30213007e-01
-1.76809147e-01 3.21595281e-01 -5.15935481e-01 -5.63519359e-01
-1.13901925e+00 6.55300140e-01 -4.76168752e-01 -3.76822948e-01
8.71041417e-01 -5.13029635e-01 2.35864613e-02 3.23838234e-01
-1.00573814e+00 2.28262872e-01 2.62742102e-01 7.73129642e-01
2.83324748e-01 -2.02590287e-01 6.54508829e-01 -1.82319319e+00
7.96340287e-01 -5.15719473e-01 -3.87381524e-01 4.85875100e-01
-4.25653696e-01 5.09363890e-01 7.23968744e-01 -1.80683911e-01
-1.58295536e+00 1.28728554e-01 -2.30865598e-01 6.77295849e-02
-2.40576476e-01 7.83629000e-01 -3.84805411e-01 7.69309223e-01
1.35093319e+00 -4.75515658e-03 -1.10723093e-01 -5.35439432e-01
8.21418226e-01 8.56760323e-01 8.68585765e-01 -1.26567852e+00
5.43583155e-01 4.15003188e-02 -5.22413671e-01 -7.02817440e-01
-1.05896235e+00 -3.90897840e-01 -4.82106060e-01 2.51452923e-01
9.09491599e-01 -8.32618952e-01 -7.49820411e-01 6.55469373e-02
-1.22874451e+00 -7.69112706e-01 -4.34480727e-01 7.00728148e-02
-5.47215283e-01 -2.20554560e-01 -5.18884003e-01 -6.10905230e-01
-4.74371687e-02 -8.90108585e-01 1.18917203e+00 2.39997469e-02
-5.00437438e-01 -9.87136900e-01 5.44621833e-02 7.14749098e-01
3.59100461e-01 1.40232861e-01 1.37193894e+00 -1.29372764e+00
-6.15358829e-01 -1.09157920e-01 1.16940722e-01 2.09142745e-01
-5.39230183e-02 -2.53813297e-01 -9.89168465e-01 8.98484066e-02
-6.95804283e-02 -9.72642183e-01 3.44168901e-01 -3.25476587e-01
1.16633904e+00 -4.61707354e-01 -5.58776796e-01 6.27839267e-01
1.05123866e+00 4.03794527e-01 4.43975300e-01 1.20362490e-01
2.82068193e-01 9.37549710e-01 6.08210325e-01 3.58818889e-01
8.92387211e-01 7.71121621e-01 3.90570387e-02 1.39630377e-01
1.60960674e-01 -7.41704464e-01 -9.76187810e-02 5.47721684e-01
4.91029710e-01 -3.34838420e-01 -1.17319894e+00 4.27324206e-01
-1.66782296e+00 -7.55952299e-01 3.89879137e-01 1.88894188e+00
1.56809902e+00 3.25931370e-01 7.15475529e-02 -6.14701271e-01
2.82685965e-01 -1.27036199e-01 -7.45002031e-01 -2.27045432e-01
-5.30341826e-02 6.39211297e-01 1.21280774e-01 8.84349823e-01
-6.86003804e-01 1.43025482e+00 5.79415464e+00 3.79537165e-01
-9.91396904e-01 3.09262015e-02 6.26122117e-01 -7.82207996e-02
-5.85320890e-01 7.49133825e-02 -1.11606765e+00 7.76625574e-02
1.21145463e+00 -2.51890630e-01 6.79296494e-01 9.88109410e-01
-2.20173955e-01 7.51187280e-02 -1.78546774e+00 5.00543475e-01
6.91088960e-02 -1.33429074e+00 1.41832829e-01 -4.17425811e-01
3.09309632e-01 3.81631441e-02 -2.67970562e-01 1.12641883e+00
8.61531854e-01 -1.07650006e+00 3.70543182e-01 3.64582449e-01
7.93197155e-01 -3.91115695e-01 3.39190334e-01 5.06168187e-01
-7.15334296e-01 -1.03730960e-02 -1.81964934e-02 1.73667610e-01
4.68646735e-02 6.93671852e-02 -1.60396063e+00 4.26254660e-01
7.46487558e-01 3.82361203e-01 -6.90688550e-01 2.30462495e-02
-5.86057007e-02 3.47140163e-01 -5.10468662e-01 -1.95757940e-01
-9.56187099e-02 1.23273775e-01 2.87856847e-01 1.32903016e+00
8.64778459e-02 2.16051936e-01 3.85284185e-01 9.85062182e-01
-2.84131289e-01 1.41402408e-01 -7.18522370e-01 -1.76640779e-01
8.79876375e-01 8.39255214e-01 -1.96730401e-02 -6.59374595e-01
-4.95423973e-01 1.00471509e+00 5.01471937e-01 6.80549502e-01
-4.42790300e-01 -4.61259365e-01 5.20740211e-01 4.05414909e-01
1.60010621e-01 1.83831170e-01 -1.82552204e-01 -1.22562516e+00
1.95463717e-01 -1.51065469e+00 7.66345084e-01 -9.36473072e-01
-1.15014470e+00 7.10307956e-01 3.80232483e-01 -3.84053797e-01
-1.05468440e+00 -2.96850860e-01 -2.74139524e-01 8.89600635e-01
-1.04436922e+00 -1.19429111e+00 -4.26617563e-01 5.93162358e-01
7.86410868e-01 -2.61383593e-01 1.06386113e+00 1.29302561e-01
-1.55537769e-01 6.13865912e-01 -3.64055127e-01 2.05221087e-01
9.87368226e-01 -1.32108974e+00 8.11507642e-01 5.04167318e-01
-5.22667132e-02 9.72514808e-01 8.88172269e-01 -8.60086918e-01
-1.40886068e+00 -1.19830632e+00 1.07229602e+00 -9.99702513e-01
4.16044325e-01 -9.92803991e-01 -1.17386830e+00 1.31658971e+00
2.04384491e-01 -6.81902245e-02 6.85651183e-01 5.65229654e-01
-6.86493516e-01 -3.39562505e-01 -1.00090325e+00 6.33033633e-01
1.27841258e+00 -8.42279315e-01 -9.73694921e-01 4.32048589e-01
1.30357230e+00 -8.40463698e-01 -8.58690560e-01 3.73323232e-01
3.92404407e-01 -9.90635991e-01 9.67419386e-01 -1.27233636e+00
3.62205207e-01 2.79191643e-01 -6.01618052e-01 -1.34100246e+00
3.13157022e-01 -7.09026396e-01 3.72218311e-01 1.28013575e+00
8.76138330e-01 -6.60695672e-01 9.84825909e-01 1.02899587e+00
-2.71258205e-01 -5.00705421e-01 -6.92492843e-01 -5.02811253e-01
1.50890410e-01 -6.63854003e-01 1.00796914e+00 9.08289254e-01
3.22716922e-01 9.21867311e-01 1.50257528e-01 -5.79677615e-03
2.18808085e-01 1.75629228e-01 1.36335194e+00 -1.09371448e+00
-7.12616563e-01 1.29109561e-01 3.30981612e-01 -1.38453186e+00
2.31422380e-01 -9.50401485e-01 1.07982099e-01 -1.24852657e+00
-7.10752234e-02 -7.35484600e-01 4.93053049e-01 5.96897125e-01
-2.47666270e-01 -5.17126501e-01 -6.81003109e-02 -5.41669019e-02
-6.22211576e-01 3.93502772e-01 7.41812110e-01 7.74679333e-02
-7.82100856e-01 6.69748560e-02 -9.75299835e-01 2.55348444e-01
4.85875934e-01 -4.54818755e-01 -8.07071388e-01 -5.08421957e-01
4.12700444e-01 5.65966964e-01 -4.36992869e-02 -7.59456396e-01
2.45668352e-01 -3.02399009e-01 1.37359247e-01 -4.58115458e-01
5.19312084e-01 -4.55754280e-01 1.39023855e-01 3.26237120e-02
-8.74998212e-01 1.90501645e-01 3.20585102e-01 3.42551917e-01
-3.23613465e-01 -1.58945173e-01 3.74859840e-01 -4.62581605e-01
-5.01313150e-01 -2.57245243e-01 -5.45730814e-02 9.02817726e-01
6.62738502e-01 1.47197500e-01 -4.95569319e-01 -6.93170428e-01
-9.48019624e-01 4.33494449e-01 2.22717300e-01 7.31575191e-01
2.31326193e-01 -9.72703278e-01 -6.82286024e-01 4.08839196e-01
5.36598623e-01 4.42703426e-01 -9.60257798e-02 5.45315482e-02
-4.41808194e-01 6.05913758e-01 2.28484482e-01 -8.49116743e-01
-1.19707167e+00 4.40227866e-01 3.54595512e-01 -3.26602012e-01
-2.96905905e-01 1.05576754e+00 7.99262598e-02 -1.23507881e+00
4.10014868e-01 -5.76539338e-01 1.42435774e-01 -2.73445025e-02
9.27076489e-02 -5.18879555e-02 2.77212411e-01 -5.41493036e-02
-2.74736673e-01 3.85993421e-02 -3.35724622e-01 -5.10799229e-01
1.38457382e+00 -1.63923316e-02 -1.20923240e-02 5.00282466e-01
1.40339398e+00 1.03242323e-01 -8.64182174e-01 -6.92340314e-01
4.82484460e-01 -3.42186242e-01 -6.11638308e-01 -1.22659993e+00
-3.93251002e-01 5.21997094e-01 -5.33661395e-02 2.86693126e-01
8.62495601e-01 4.32169527e-01 6.69659257e-01 1.07786703e+00
2.86276579e-01 -1.09951341e+00 3.58024567e-01 7.62089550e-01
1.02917540e+00 -1.25178051e+00 -3.65025520e-01 -5.00498295e-01
-9.24921513e-01 1.01056671e+00 1.06954801e+00 5.50917923e-01
3.56693238e-01 4.58603770e-01 4.37532008e-01 -3.02736491e-01
-1.29211867e+00 -1.58780180e-02 -2.58337576e-02 6.27360284e-01
5.71657777e-01 -9.26889479e-02 3.39812487e-01 6.80445254e-01
-8.45869064e-01 1.27667278e-01 2.24434540e-01 1.03859365e+00
-1.22742899e-01 -1.52241206e+00 -6.08070903e-02 5.55564582e-01
-1.75318733e-01 -2.27320910e-01 -5.46443582e-01 1.01277411e+00
3.92857566e-03 9.38270688e-01 1.11182921e-01 -3.77797544e-01
8.42004061e-01 8.51881504e-01 1.82404935e-01 -1.24065435e+00
-7.62006044e-01 -1.77627951e-01 8.60573471e-01 -6.99104130e-01
8.93561468e-02 -7.22860992e-01 -1.40748727e+00 -8.53908584e-02
2.14484315e-02 5.10084271e-01 6.34724557e-01 7.80733466e-01
7.24915028e-01 4.21775550e-01 3.58690828e-01 1.64700449e-02
-7.53925622e-01 -1.00083816e+00 8.96364376e-02 9.47346926e-01
1.62981927e-01 -1.78991303e-01 -6.50352612e-02 2.05068320e-01]
|
[10.029142379760742, 7.9431023597717285]
|
73add660-3db0-4810-9538-6d0e7fafcbeb
|
pc-rgnn-point-cloud-completion-and-graph
|
2012.10412
| null |
https://arxiv.org/abs/2012.10412v3
|
https://arxiv.org/pdf/2012.10412v3.pdf
|
PC-RGNN: Point Cloud Completion and Graph Neural Network for 3D Object Detection
|
LiDAR-based 3D object detection is an important task for autonomous driving and current approaches suffer from sparse and partial point clouds of distant and occluded objects. In this paper, we propose a novel two-stage approach, namely PC-RGNN, dealing with such challenges by two specific solutions. On the one hand, we introduce a point cloud completion module to recover high-quality proposals of dense points and entire views with original structures preserved. On the other hand, a graph neural network module is designed, which comprehensively captures relations among points through a local-global attention mechanism as well as multi-scale graph based context aggregation, substantially strengthening encoded features. Extensive experiments on the KITTI benchmark show that the proposed approach outperforms the previous state-of-the-art baselines by remarkable margins, highlighting its effectiveness.
|
['Yunhong Wang', 'Di Huang', 'Yanan Zhang']
|
2020-12-18
| null | null | null | null |
['point-cloud-completion']
|
['computer-vision']
|
[-7.09639788e-02 3.55177745e-02 -5.80696203e-02 -3.40585291e-01
-8.17266166e-01 -3.09138864e-01 6.28765166e-01 1.46272093e-01
-3.28623086e-01 3.25968891e-01 -1.00064874e-01 -2.69642770e-01
-5.06530236e-03 -8.12997401e-01 -9.16809440e-01 -5.10682642e-01
1.57580495e-01 6.32370353e-01 7.35943317e-01 -3.69774967e-01
4.58935887e-01 9.24907804e-01 -1.85113633e+00 -1.18468098e-01
1.08184791e+00 1.12745643e+00 3.90360683e-01 2.68012673e-01
-2.30043232e-01 5.50900042e-01 -1.34551868e-01 -4.76474613e-01
3.69255215e-01 4.75732982e-01 -2.30764508e-01 2.69677103e-01
9.41399157e-01 -4.30657655e-01 -5.09384155e-01 1.18576467e+00
2.14569330e-01 2.39883289e-01 3.45720083e-01 -1.25782597e+00
-5.98356664e-01 3.97015251e-02 -1.00030625e+00 -2.11265106e-02
4.63635921e-02 3.14201951e-01 1.16906142e+00 -1.45572221e+00
4.55426365e-01 1.48637080e+00 6.87992334e-01 1.06795035e-01
-1.09293985e+00 -7.56199658e-01 6.67624652e-01 4.81766582e-01
-1.55378783e+00 -1.58307493e-01 1.20530009e+00 -3.17866027e-01
1.07354879e+00 1.47355897e-02 7.31912911e-01 6.35130644e-01
1.57524496e-01 6.25525832e-01 8.42900515e-01 1.20475842e-02
1.39587119e-01 -2.05903828e-01 1.50838822e-01 8.07736814e-01
5.40047348e-01 2.34374449e-01 -5.27261734e-01 -9.08384919e-02
6.19613528e-01 4.29198474e-01 -4.81911562e-02 -9.18576777e-01
-9.99785841e-01 8.32073689e-01 1.01535630e+00 -9.78188664e-02
-5.11741340e-01 3.08054298e-01 6.52656481e-02 -1.66370228e-01
6.92341924e-01 -2.78222561e-03 -1.37748554e-01 4.34273183e-01
-7.69001484e-01 5.72610080e-01 2.89434582e-01 1.35970831e+00
1.16482627e+00 -5.65389991e-02 -7.15696141e-02 5.42069793e-01
5.39778531e-01 5.69450200e-01 -3.00247163e-01 -8.73329699e-01
8.86287391e-01 1.09109414e+00 3.19505744e-02 -1.17659557e+00
-3.96513551e-01 -6.25842750e-01 -6.78498805e-01 5.54886997e-01
-1.65123060e-01 4.20392185e-01 -1.13419938e+00 1.39241683e+00
7.05132723e-01 6.63194716e-01 -1.17105156e-01 1.01474822e+00
1.05811143e+00 4.82447833e-01 1.22205848e-02 2.52331197e-01
1.29252195e+00 -1.09159958e+00 -4.73498970e-01 -6.87159956e-01
5.08476458e-02 -4.22290951e-01 7.43765235e-01 3.30125764e-02
-1.00694489e+00 -7.14441836e-01 -1.16615498e+00 -6.29248261e-01
-3.90446365e-01 2.31028330e-02 6.56159937e-01 2.59180725e-01
-1.12530982e+00 3.91494423e-01 -8.19002688e-01 -9.46322456e-02
8.30651641e-01 3.05813909e-01 -4.23484623e-01 -6.09203577e-01
-5.26152372e-01 7.88059354e-01 2.27337390e-01 3.52127910e-01
-7.75504529e-01 -7.34921098e-01 -1.04673707e+00 1.25590727e-01
5.26204109e-01 -1.00433969e+00 8.63461912e-01 -1.25626266e-01
-1.10919821e+00 8.06765974e-01 -3.82325113e-01 -4.45451796e-01
4.86983240e-01 -4.87892330e-01 -1.14205470e-02 1.90967113e-01
2.88182586e-01 9.49791670e-01 8.26792359e-01 -1.67263091e+00
-9.52908397e-01 -8.56249452e-01 9.44975764e-02 4.17375416e-01
1.53184801e-01 -3.72034252e-01 -9.36023414e-01 -2.98856556e-01
5.63509703e-01 -8.50888133e-01 -5.58448076e-01 2.94776112e-01
-3.26669037e-01 -5.84056854e-01 1.09129655e+00 -3.35085481e-01
5.83892167e-01 -2.05484557e+00 2.06360191e-01 1.01558886e-01
6.31864846e-01 1.74602300e-01 -1.51890516e-01 2.67626286e-01
2.20530346e-01 -2.22012773e-01 -4.68061954e-01 -9.40953910e-01
7.27325529e-02 4.66832966e-01 -4.28225070e-01 4.62002069e-01
8.70297313e-01 1.21300268e+00 -8.91209483e-01 -4.12495315e-01
5.82841337e-01 7.18165815e-01 -6.12636864e-01 4.11895066e-02
-2.71538675e-01 4.33799088e-01 -6.17931008e-01 9.03119922e-01
1.22857869e+00 -2.13307932e-01 -2.87021309e-01 -1.55793354e-01
-2.70054698e-01 2.46900320e-01 -1.05689204e+00 2.04399323e+00
-8.54013935e-02 3.55935335e-01 1.56076297e-01 -9.78529751e-01
1.01503420e+00 -2.01026410e-01 4.15763944e-01 -5.49377918e-01
4.33324687e-02 1.64880708e-01 -4.29961890e-01 -1.72117949e-01
8.33740711e-01 1.19782984e-01 1.33304745e-01 -2.03022331e-01
-1.99651480e-01 -4.18924481e-01 -3.11524093e-01 3.08445841e-01
1.07163167e+00 3.90538812e-01 4.44610827e-02 -1.64436060e-03
5.40474236e-01 2.19731912e-01 6.87518060e-01 5.27850807e-01
-3.52658957e-01 8.88690650e-01 1.46838769e-01 -4.25559103e-01
-8.07487488e-01 -9.15325999e-01 8.88988189e-03 5.40226877e-01
7.41944313e-01 -2.89469808e-01 -3.08837891e-01 -8.31000984e-01
4.53498274e-01 6.87409759e-01 -4.55356687e-01 -9.03105959e-02
-7.91846097e-01 -2.18299359e-01 -9.62787718e-02 8.01336825e-01
4.07352537e-01 -8.50802362e-01 -4.66361672e-01 2.71832079e-01
-1.98687896e-01 -1.59451294e+00 -2.06519574e-01 1.76132157e-01
-1.14494467e+00 -9.86827552e-01 -3.92976910e-01 -6.65930331e-01
6.33651078e-01 1.19741762e+00 1.31468058e+00 1.50861084e-01
-6.14617616e-02 8.79708454e-02 -2.72270143e-01 -6.72504306e-01
2.85597473e-01 -1.43156976e-01 -1.89888388e-01 4.21836600e-02
5.37764549e-01 -8.00556004e-01 -7.29904294e-01 2.26295471e-01
-5.73455691e-01 -6.49039671e-02 1.02735877e+00 6.16364837e-01
1.14692795e+00 -2.79242635e-01 3.52101445e-01 -6.72580004e-01
-3.04266587e-02 -4.06623036e-01 -9.48526144e-01 -1.12234928e-01
-5.00692904e-01 -1.82398930e-01 1.48024961e-01 8.95677656e-02
-7.68847883e-01 4.77602512e-01 -2.28614628e-01 -1.16379678e+00
-2.05563575e-01 2.55483061e-01 -4.17867064e-01 -4.75906014e-01
2.24126399e-01 2.00682923e-01 -1.48186892e-01 -4.99796718e-01
6.25402272e-01 2.22164184e-01 7.30196774e-01 -3.01520139e-01
1.35402060e+00 9.92489517e-01 2.64334768e-01 -5.87887168e-01
-8.29493463e-01 -8.25236917e-01 -8.47590923e-01 -1.34005755e-01
8.22929740e-01 -1.43023038e+00 -4.55814302e-01 1.99649557e-01
-1.35712671e+00 -8.19932893e-02 -3.56308848e-01 4.10745680e-01
-6.23669624e-01 4.25433546e-01 -6.28898069e-02 -8.08418453e-01
-2.52608329e-01 -1.13930535e+00 1.79264700e+00 4.39325012e-02
5.83447516e-01 -3.96175236e-01 -4.95513827e-02 5.78100085e-01
-7.63400458e-03 3.56713921e-01 5.36489010e-01 -2.56604463e-01
-1.49483085e+00 -2.91080028e-01 -8.04866314e-01 1.46057263e-01
-2.47788727e-01 -1.96790874e-01 -1.11506248e+00 -2.79595435e-01
-1.89217813e-02 -5.08664399e-02 1.28100526e+00 2.71707267e-01
1.19093680e+00 1.80394664e-01 -5.98013461e-01 8.29985857e-01
1.45439768e+00 -2.55239487e-01 6.47222757e-01 8.29297304e-02
1.13811207e+00 6.03668809e-01 9.04391587e-01 1.25322014e-01
9.29130912e-01 6.20042622e-01 1.15307117e+00 -1.32282585e-01
-3.82054597e-01 -4.13331836e-01 -2.03131847e-02 5.58282852e-01
-1.64979488e-01 2.52292659e-02 -7.55131721e-01 6.23033702e-01
-2.05010772e+00 -7.43981898e-01 -3.89320135e-01 1.92793977e+00
3.35299447e-02 3.65206778e-01 -4.07903194e-02 -2.14553401e-02
6.18655384e-01 4.06886578e-01 -7.93542266e-01 2.56161004e-01
-5.22385649e-02 6.79970384e-02 5.88789225e-01 4.06537682e-01
-1.07006299e+00 1.09640682e+00 5.30662489e+00 7.18824446e-01
-7.87378013e-01 1.35347247e-01 2.03065246e-01 -3.92206647e-02
-2.68827379e-01 1.22717686e-01 -1.08434868e+00 5.92835359e-02
2.54982054e-01 1.66917056e-01 -6.34452607e-03 1.03029287e+00
1.62103809e-02 1.71007246e-01 -8.14850807e-01 1.01134634e+00
1.65431902e-01 -1.40907800e+00 1.31544396e-01 3.68052065e-01
7.34351695e-01 6.99729502e-01 -1.13736615e-01 3.21177125e-01
2.68621802e-01 -5.51670015e-01 9.29000556e-01 5.13502181e-01
5.91744781e-01 -9.27743852e-01 5.48399746e-01 3.61325711e-01
-1.64442003e+00 -8.76002908e-02 -7.74405420e-01 -7.19458563e-04
1.89947918e-01 7.64278769e-01 -5.03052711e-01 1.07180631e+00
8.26602817e-01 1.13123465e+00 -7.55231738e-01 1.31009531e+00
-2.77227134e-01 2.49951601e-01 -3.76878440e-01 2.43212447e-01
4.74143147e-01 -2.85454869e-01 9.18724239e-01 8.10982704e-01
3.75754744e-01 2.33368203e-01 4.03809339e-01 1.16140795e+00
-1.26340732e-01 -1.55201450e-01 -9.66340899e-01 5.30257225e-01
4.12757039e-01 1.60156476e+00 -5.61874211e-01 -2.75712252e-01
-7.41710722e-01 4.81514722e-01 7.71336317e-01 3.41211885e-01
-6.89586282e-01 -1.83420435e-01 8.75973046e-01 1.66135862e-01
7.73195148e-01 -6.65198743e-01 -4.33239341e-01 -1.17584825e+00
4.32412207e-01 -3.64708275e-01 1.49876118e-01 -8.32800984e-01
-1.29856825e+00 5.20692945e-01 -1.77242681e-01 -1.46625078e+00
2.96185285e-01 -5.39743662e-01 -6.16546094e-01 8.99146497e-01
-2.48325968e+00 -1.56425011e+00 -8.15496504e-01 6.45996392e-01
5.99738359e-01 1.36021614e-01 2.19743773e-01 2.80663222e-01
-4.04994369e-01 1.17497377e-01 -2.54012078e-01 -1.47399664e-01
3.77571106e-01 -1.09940577e+00 9.37289536e-01 9.80891526e-01
2.67967552e-01 4.95271921e-01 2.47062236e-01 -8.45839620e-01
-1.66310060e+00 -1.62177980e+00 7.26203620e-01 -4.97459769e-01
4.42003399e-01 -5.02530277e-01 -1.16296530e+00 6.85637474e-01
-3.77587066e-03 4.72185224e-01 -2.46590879e-02 -6.99067339e-02
-4.82059211e-01 -2.05687612e-01 -8.59381139e-01 4.67078686e-01
1.49380934e+00 -4.34531599e-01 -5.43754876e-01 4.47173268e-01
1.11403131e+00 -6.84577942e-01 -4.63890642e-01 7.19686568e-01
1.46911144e-01 -1.09705055e+00 1.37766767e+00 -2.32067525e-01
1.61069751e-01 -6.49314582e-01 -2.71393627e-01 -9.23293114e-01
-5.39083600e-01 -4.23653483e-01 -3.81460369e-01 1.01360214e+00
9.54825804e-03 -6.94726765e-01 9.81468022e-01 2.80638665e-01
-7.96778321e-01 -7.75186539e-01 -1.14205873e+00 -7.69913793e-01
-1.81894809e-01 -6.74255073e-01 7.61914492e-01 6.07637465e-01
-7.15685189e-01 5.35847068e-01 -7.60832652e-02 7.15009809e-01
9.61724520e-01 4.80812490e-01 1.10952365e+00 -1.54841876e+00
1.43265188e-01 -5.36245763e-01 -7.70503283e-01 -1.42544055e+00
2.02937424e-01 -8.20921898e-01 2.02634305e-01 -1.71220803e+00
-2.55021118e-02 -5.79051435e-01 -2.21354723e-01 2.62186736e-01
-5.71471989e-01 3.69742155e-01 1.91664368e-01 2.74031579e-01
-8.07977557e-01 9.62133527e-01 1.18512940e+00 -2.10514024e-01
-8.47625136e-02 5.29299229e-02 -6.54675484e-01 7.69635260e-01
3.71446460e-01 -5.39002657e-01 -2.48481348e-01 -6.83126509e-01
5.22813387e-02 -2.80676067e-01 9.69279408e-01 -1.26274586e+00
3.61466765e-01 -1.12492129e-01 2.07930908e-01 -1.56757784e+00
7.93422401e-01 -1.03501403e+00 -2.01650202e-01 1.94104165e-01
2.13329867e-01 2.09848315e-01 3.52708250e-01 1.27328444e+00
-2.15200931e-01 1.54270425e-01 6.11492157e-01 -7.87334219e-02
-9.80231225e-01 8.54604661e-01 3.65134835e-01 -1.85656309e-01
1.07868195e+00 -2.25369155e-01 -3.50368381e-01 -4.14019711e-02
-2.39093453e-01 6.17003977e-01 5.07493675e-01 5.99435687e-01
1.01144302e+00 -1.37134111e+00 -8.15423191e-01 3.62089187e-01
5.23601830e-01 8.51064622e-01 3.04805994e-01 8.81260574e-01
-3.67307514e-01 4.03551698e-01 7.10126385e-02 -1.06827211e+00
-1.07044792e+00 7.27637708e-01 -1.33217836e-04 -1.63331389e-01
-1.21933043e+00 8.48523557e-01 4.55711842e-01 -5.23547471e-01
1.22165062e-01 -5.40176451e-01 -2.81367779e-01 -1.47000298e-01
2.98665524e-01 9.79783833e-02 3.56937408e-01 -8.88879001e-01
-4.73322749e-01 1.15296912e+00 -2.33871303e-02 2.84569532e-01
1.53531337e+00 -3.82880479e-01 5.40480949e-02 2.07980782e-01
9.49483573e-01 -6.67352304e-02 -1.62423897e+00 -5.78500152e-01
-1.62475839e-01 -7.35457182e-01 3.02210748e-01 -3.40128630e-01
-1.22767746e+00 1.07220900e+00 4.50038552e-01 5.85049018e-02
8.90841246e-01 5.82733564e-02 7.62998521e-01 5.01920283e-01
4.87330735e-01 -6.10884309e-01 3.14448066e-02 4.70818818e-01
1.12624526e+00 -1.51574910e+00 1.99876338e-01 -8.00406098e-01
-4.39522684e-01 7.73364007e-01 6.57209337e-01 -5.39080203e-01
5.15321314e-01 -1.81400582e-01 -1.24765933e-01 -5.69205642e-01
-6.32953465e-01 -6.75487280e-01 6.22619212e-01 6.99209094e-01
-4.39682752e-01 -8.02719668e-02 1.00379638e-01 3.29601228e-01
9.29554999e-02 -1.85303345e-01 4.20140363e-02 9.92927849e-01
-6.14251256e-01 -7.46144652e-01 -3.07819694e-01 2.78570205e-01
1.43283218e-01 -4.49241549e-02 -4.59401578e-01 8.94851387e-01
5.92990369e-02 9.06640589e-01 1.68607965e-01 -4.79724824e-01
6.66854501e-01 -8.77841413e-02 1.27015159e-01 -6.43304408e-01
-3.76076549e-01 4.60569821e-02 -2.13214830e-01 -1.04146481e+00
-4.55450475e-01 -8.70799184e-01 -1.19844460e+00 -2.51190923e-02
-5.08653283e-01 -3.18999082e-01 8.12887907e-01 7.51091719e-01
9.31469500e-01 6.58362150e-01 5.93113601e-01 -1.48086083e+00
-3.19999427e-01 -6.47886038e-01 -3.57076705e-01 2.29348481e-01
5.54139972e-01 -1.05639017e+00 -2.17631459e-01 -4.74586576e-01]
|
[7.9801483154296875, -2.8016834259033203]
|
8cf063c6-d81f-4075-bcb5-aac0624b58e3
|
understanding-spatial-relations-through-1
|
2007.09551
| null |
https://arxiv.org/abs/2007.09551v1
|
https://arxiv.org/pdf/2007.09551v1.pdf
|
Understanding Spatial Relations through Multiple Modalities
|
Recognizing spatial relations and reasoning about them is essential in multiple applications including navigation, direction giving and human-computer interaction in general. Spatial relations between objects can either be explicit -- expressed as spatial prepositions, or implicit -- expressed by spatial verbs such as moving, walking, shifting, etc. Both these, but implicit relations in particular, require significant common sense understanding. In this paper, we introduce the task of inferring implicit and explicit spatial relations between two entities in an image. We design a model that uses both textual and visual information to predict the spatial relations, making use of both positional and size information of objects and image embeddings. We contrast our spatial model with powerful language models and show how our modeling complements the power of these, improving prediction accuracy and coverage and facilitates dealing with unseen subjects, objects and relations.
|
['Dan Roth', 'Soham Dan', 'Hangfeng He']
|
2020-07-19
|
understanding-spatial-relations-through
|
https://aclanthology.org/2020.lrec-1.288
|
https://aclanthology.org/2020.lrec-1.288.pdf
|
lrec-2020-5
|
['implicit-relations']
|
['natural-language-processing']
|
[-4.43350784e-02 1.45634755e-01 -2.73542166e-01 -6.27955794e-01
9.13963318e-02 -8.27476978e-01 9.90408957e-01 5.34981310e-01
-5.89641571e-01 6.50180817e-01 5.58353841e-01 -5.91399252e-01
-4.50349271e-01 -1.00820720e+00 -6.88694656e-01 -3.09305459e-01
-2.91201115e-01 3.71769696e-01 5.28069139e-01 -1.29501849e-01
4.66242254e-01 7.70175040e-01 -1.48051715e+00 3.93695772e-01
5.05145431e-01 8.28631759e-01 3.42050284e-01 5.00374734e-01
-3.86897117e-01 1.27441514e+00 -1.61116600e-01 -3.18983287e-01
-1.48473814e-01 1.81449294e-01 -1.14740372e+00 -2.29438052e-01
2.32768193e-01 9.49060638e-03 -4.52193081e-01 7.22759306e-01
9.31827873e-02 4.21945393e-01 8.59487176e-01 -1.35644758e+00
-1.16106498e+00 3.53823125e-01 -5.31309366e-01 2.72722930e-01
5.50874412e-01 -9.65961292e-02 1.17995644e+00 -1.00660408e+00
6.91888213e-01 1.20706904e+00 6.62907958e-01 1.44493505e-01
-1.30949080e+00 -2.02805713e-01 3.19725662e-01 6.61936224e-01
-1.63098454e+00 -4.64164376e-01 5.86466670e-01 -6.22301817e-01
1.31134558e+00 4.76367235e-01 6.06800497e-01 5.41490257e-01
-1.56278774e-01 8.55202019e-01 1.03073418e+00 -7.14667857e-01
1.15649380e-01 2.60183781e-01 1.96208909e-01 5.15927970e-01
-2.68891584e-02 -2.02892404e-02 -5.64121068e-01 8.69044811e-02
1.01728177e+00 -4.26757708e-02 -1.73603445e-01 -6.65102363e-01
-1.42410302e+00 6.30098760e-01 9.42649782e-01 5.70204794e-01
-2.87369519e-01 3.91210496e-01 1.74916182e-02 -9.51225907e-02
3.19645673e-01 6.76565111e-01 -3.89978141e-01 -1.28717780e-01
-5.96814632e-01 2.79158413e-01 6.46012723e-01 1.24018228e+00
8.75893235e-01 -6.42259240e-01 8.66218135e-02 6.75761223e-01
3.30008239e-01 2.99525112e-01 2.96774596e-01 -7.36740470e-01
3.73897105e-01 7.55108058e-01 3.78490508e-01 -1.62695479e+00
-5.98190606e-01 7.73823410e-02 -5.16229570e-01 -1.71846151e-02
3.03252637e-01 4.70640033e-01 -8.43771219e-01 1.83675468e+00
2.51364768e-01 1.92749277e-01 -3.07989448e-01 7.16090441e-01
9.28566754e-01 6.64736450e-01 5.72778761e-01 1.34640306e-01
1.55675793e+00 -8.87253642e-01 -8.42953801e-01 -5.46918631e-01
1.00604475e+00 -5.26016533e-01 1.09194195e+00 -2.08754838e-01
-9.63901341e-01 -3.29549134e-01 -6.74310088e-01 -7.79907048e-01
-1.23536325e+00 9.25536901e-02 8.02797258e-01 1.73377737e-01
-1.10678554e+00 2.70518005e-01 -6.66361690e-01 -6.87569737e-01
4.47131038e-01 3.97325248e-01 -7.46039629e-01 9.10379440e-02
-1.09433532e+00 1.36789548e+00 2.80184776e-01 6.90289363e-02
2.25353375e-01 -6.18956029e-01 -1.30742455e+00 -5.31392135e-02
2.45029658e-01 -8.11046481e-01 8.66821468e-01 -5.41389704e-01
-7.25236416e-01 1.26243901e+00 -5.27920842e-01 -3.71432632e-01
1.27921313e-01 -4.73767489e-01 -3.40018302e-01 -3.07428777e-01
3.31833154e-01 1.04698396e+00 2.35073939e-01 -1.17075574e+00
-6.02519572e-01 -5.97432375e-01 4.48755682e-01 5.31315386e-01
-1.52215743e-02 7.67108351e-02 -3.38259429e-01 -4.30840850e-01
4.50296879e-01 -8.43449056e-01 -1.60552308e-01 5.91807306e-01
-6.89490020e-01 -3.55826110e-01 8.82195175e-01 -6.14170432e-01
1.22805786e+00 -2.35261011e+00 1.84887543e-01 2.82607764e-01
1.56922877e-01 -2.79119853e-02 -2.60250475e-02 4.66077328e-01
-3.01122785e-01 1.69562861e-01 1.77151500e-03 -4.27933149e-02
1.83540642e-01 6.36375070e-01 -4.69129324e-01 4.20363367e-01
2.03565404e-01 1.37811244e+00 -1.07488298e+00 -5.57823777e-01
5.30781209e-01 4.20486629e-01 -3.11140060e-01 8.33832547e-02
-1.37578905e-01 2.80219167e-01 -3.76857847e-01 2.34290227e-01
4.85399663e-01 -3.74772191e-01 3.19050908e-01 -3.12990844e-01
-4.17622179e-01 7.42968500e-01 -1.15234864e+00 1.61644268e+00
-5.88219762e-01 8.94286275e-01 -4.37669218e-01 -8.68918240e-01
6.31540418e-01 9.30548832e-02 1.08338684e-01 -9.94160473e-01
-2.06228331e-01 -6.36865497e-02 -1.47950709e-01 -8.27224433e-01
6.65836990e-01 -1.48906872e-01 2.00712979e-02 3.59533012e-01
-3.93920481e-01 -1.65026367e-01 1.28034145e-01 1.30087420e-01
8.34057629e-01 4.64427739e-01 9.35736358e-01 -4.29270864e-01
4.45643157e-01 1.57958865e-02 6.83870167e-02 5.01386702e-01
6.27387986e-02 3.04076165e-01 5.16868651e-01 -7.49613523e-01
-8.63805711e-01 -1.38416219e+00 -7.76025727e-02 1.16673958e+00
5.39434791e-01 -6.22226357e-01 -1.09128624e-01 -4.19320285e-01
1.35005563e-01 1.00218725e+00 -1.02406347e+00 1.31267875e-01
-6.85708046e-01 -1.53571323e-01 2.00360328e-01 1.04167604e+00
2.19742626e-01 -1.23697972e+00 -1.01720428e+00 -2.25865126e-01
-1.77662894e-01 -1.29447997e+00 -2.94483334e-01 3.33153009e-01
-5.49682140e-01 -9.73135233e-01 -1.80616945e-01 -9.81003284e-01
7.07168043e-01 2.12717816e-01 1.34032655e+00 9.05554518e-02
-4.27998215e-01 4.15595859e-01 -3.20646375e-01 -3.34719449e-01
4.48834479e-01 -7.16472790e-02 -6.85851052e-02 -2.48438716e-01
5.98581791e-01 -8.11429977e-01 -4.92654204e-01 3.03037345e-01
-7.06496894e-01 3.25452745e-01 3.19932014e-01 5.86353242e-01
4.83909786e-01 -2.01293528e-01 -5.95807619e-02 -8.66556108e-01
2.76143730e-01 -5.08242309e-01 -2.15205118e-01 4.33019817e-01
-8.12912807e-02 2.40844220e-01 3.26894410e-02 -4.66560960e-01
-8.41359675e-01 7.07401410e-02 1.39058620e-01 4.15274035e-03
-4.33820128e-01 6.00439548e-01 -2.64571667e-01 1.28486067e-01
7.52375126e-01 2.09295541e-01 -3.61275643e-01 -2.46771604e-01
9.80157256e-01 3.02558959e-01 4.43755180e-01 -6.09907627e-01
7.23399401e-01 5.42895734e-01 2.77118891e-01 -1.04580927e+00
-9.23557878e-01 -5.90201676e-01 -1.31351161e+00 2.11006001e-01
1.22201395e+00 -7.20556438e-01 -6.73966527e-01 -1.26500160e-01
-1.38043272e+00 -2.75493383e-01 -5.57665706e-01 3.69751811e-01
-5.37949800e-01 4.08153981e-02 -4.01712894e-01 -7.48954296e-01
5.88962138e-01 -8.07833433e-01 1.12586033e+00 3.69356722e-02
-8.59404981e-01 -1.41921282e+00 -3.47502008e-02 -4.78881560e-02
3.38663667e-01 4.25680019e-02 1.26095247e+00 -5.26667953e-01
-8.95740271e-01 9.97457877e-02 -7.28523552e-01 -3.76224279e-01
3.74991357e-01 -2.93279856e-01 -6.76304162e-01 5.27237475e-01
-5.34674644e-01 -1.44209653e-01 4.56113070e-01 3.51411968e-01
1.20747614e+00 -4.82409358e-01 -7.47778952e-01 5.32492936e-01
1.34663093e+00 2.37893820e-01 9.10528421e-01 4.83196557e-01
7.78562725e-01 9.92178082e-01 5.57291448e-01 2.17905328e-01
7.19863057e-01 9.00652111e-01 2.58611858e-01 9.36263520e-03
2.26905551e-02 -4.11571831e-01 -2.82876998e-01 3.16338241e-01
-1.91366151e-01 -9.90666971e-02 -1.11436760e+00 7.66174495e-01
-1.96172798e+00 -1.12362885e+00 -4.64017779e-01 1.80729890e+00
8.46955240e-01 -3.15443248e-01 -1.35809258e-01 9.20028090e-02
3.40256691e-01 2.82409132e-01 -1.52420700e-01 -3.64955276e-01
-2.64303386e-02 1.20332733e-01 2.89430827e-01 7.10490763e-01
-1.26420724e+00 1.18990409e+00 6.68719864e+00 4.43934411e-01
-1.03033352e+00 1.58086587e-02 3.68989378e-01 1.35255709e-01
-5.10943413e-01 -3.34329088e-03 -3.08876574e-01 -1.02724358e-01
3.54613811e-01 6.21392839e-02 2.84484476e-01 6.95769370e-01
-4.25652526e-02 -3.61802965e-01 -1.53260815e+00 1.22240090e+00
5.95970377e-02 -1.36475956e+00 8.48907232e-03 1.13307303e-02
2.83007562e-01 -5.13041139e-01 2.58736759e-01 2.66494360e-02
3.27190727e-01 -1.48247719e+00 9.83037710e-01 6.39012754e-01
6.14431262e-01 -2.50616014e-01 4.01667416e-01 3.59429866e-01
-1.29350305e+00 8.28248039e-02 -6.44899383e-02 -5.67546427e-01
2.71265119e-01 2.27110222e-01 -5.92662632e-01 3.36879313e-01
7.54506648e-01 7.83586144e-01 -6.28965557e-01 8.37573409e-01
-5.98280966e-01 -1.08782174e-02 -4.71310794e-01 -1.69628754e-01
1.11508869e-01 -2.48780146e-01 1.58258051e-01 1.14451241e+00
2.17985854e-01 5.56740880e-01 -1.23746082e-01 1.12461245e+00
4.34680641e-01 1.23243287e-01 -1.02787375e+00 1.11731857e-01
5.14489472e-01 9.05519605e-01 -7.51012444e-01 -1.79099813e-01
-5.31751275e-01 8.86580408e-01 6.97379172e-01 5.32264113e-01
-8.80344689e-01 -2.19825223e-01 7.41677761e-01 5.68960667e-01
2.09168807e-01 -6.14610672e-01 -6.18465841e-01 -8.08067441e-01
1.42450020e-01 -7.94959068e-02 1.28947794e-01 -1.40585113e+00
-1.09410465e+00 4.19429392e-01 2.72689968e-01 -1.01346743e+00
-1.29644170e-01 -1.01028061e+00 -3.38716924e-01 9.68724787e-01
-1.21564841e+00 -1.48774111e+00 -3.29386234e-01 4.90356714e-01
2.23278388e-01 3.59959304e-01 8.32850099e-01 8.39312747e-02
-1.22692659e-02 1.72517806e-01 -5.41862190e-01 2.57840276e-01
5.14266253e-01 -1.35783756e+00 3.32214683e-01 3.55300337e-01
7.07369804e-01 1.12078214e+00 7.67001748e-01 -3.92746717e-01
-9.72667217e-01 -8.45305741e-01 1.55650973e+00 -9.13206100e-01
8.15987945e-01 -4.36930180e-01 -8.02894831e-01 1.21939480e+00
1.81643695e-01 3.63506258e-01 7.89588451e-01 6.84303939e-01
-7.83919036e-01 2.99950719e-01 -7.95883060e-01 1.11639225e+00
1.48978031e+00 -9.32748199e-01 -9.70325410e-01 2.59350270e-01
5.54276943e-01 -4.99772757e-01 -4.79529023e-01 2.80904293e-01
4.87452060e-01 -9.48743999e-01 1.44698918e+00 -9.02589917e-01
4.50642288e-01 -3.72626007e-01 -3.91938716e-01 -1.18861675e+00
-7.07619190e-01 2.00024128e-01 -1.25016253e-02 8.56480420e-01
7.26486087e-01 -4.85830694e-01 7.08002627e-01 8.75758708e-01
2.22111464e-01 -7.07070470e-01 -7.49611020e-01 -6.89523339e-01
-5.87460175e-02 -5.78832328e-01 6.25780463e-01 1.27115631e+00
4.53098327e-01 4.88375604e-01 -1.44769043e-01 2.14842334e-01
7.04688728e-02 2.11821109e-01 5.97667336e-01 -1.13724613e+00
-8.26424081e-03 -4.44237173e-01 -9.53939319e-01 -1.48043799e+00
3.42524648e-01 -7.19707012e-01 -5.52186891e-02 -1.91228843e+00
-7.02882037e-02 -9.01150107e-01 1.28791511e-01 7.15903938e-01
8.60666558e-02 1.88015044e-01 1.29077509e-01 7.84515403e-03
-7.36407518e-01 3.58496606e-01 1.13343263e+00 -1.37508512e-01
-1.23482279e-01 -4.04360652e-01 -5.31042516e-01 9.77711380e-01
4.08579648e-01 -1.74926639e-01 -5.80588341e-01 -7.44213462e-01
7.89743364e-01 -1.05236806e-01 7.50295997e-01 -7.83199668e-01
4.64980304e-01 -5.49837947e-01 5.02792239e-01 -3.80432814e-01
6.34283304e-01 -1.15983355e+00 2.60122772e-02 8.98226723e-02
-6.54681563e-01 3.58664155e-01 2.44226918e-01 5.06704748e-01
-2.89415568e-01 -7.16061071e-02 2.13031888e-01 -1.15865126e-01
-1.24966049e+00 9.37460288e-02 -1.58297479e-01 1.79430973e-02
1.22849905e+00 -3.72005582e-01 -3.23689103e-01 -4.92261499e-01
-1.22336423e+00 6.29698262e-02 4.61235464e-01 5.36270559e-01
7.23300815e-01 -1.49248862e+00 -1.60411060e-01 1.08723067e-01
4.03839469e-01 5.73883802e-02 1.59972116e-01 9.85353649e-01
-7.29966581e-01 5.46899259e-01 -1.04926400e-01 -5.74133337e-01
-1.24872673e+00 1.03053701e+00 7.69467130e-02 6.26406670e-02
-5.18248677e-01 1.00424707e+00 7.37470686e-01 -5.23402572e-01
5.65379262e-02 -5.47932982e-01 -4.35179293e-01 -7.34561533e-02
4.35439289e-01 2.56619975e-02 -4.02486026e-01 -1.04947007e+00
-6.91328466e-01 8.50977480e-01 3.11690867e-01 -2.49066636e-01
1.17530775e+00 -2.35453025e-01 -4.93497849e-01 8.56042922e-01
1.09669828e+00 1.86896026e-01 -7.14197099e-01 -3.92669499e-01
3.77355427e-01 -6.95011139e-01 -3.88182670e-01 -7.67648458e-01
-4.18022841e-01 1.12095428e+00 2.00386778e-01 4.14536953e-01
8.57155919e-01 5.31999946e-01 2.05802321e-01 3.21186393e-01
4.70334113e-01 -6.80380285e-01 2.63125487e-02 5.29299915e-01
9.86804545e-01 -1.05639076e+00 1.29303992e-01 -9.55812752e-01
-5.79576075e-01 9.55757022e-01 6.12828076e-01 1.30927071e-01
8.63938272e-01 3.91305447e-01 -1.60756409e-01 -3.20999980e-01
-6.91092134e-01 -3.64427835e-01 7.24969327e-01 8.55321348e-01
7.27071643e-01 2.67665535e-01 -1.02395602e-01 2.08966017e-01
-3.86538833e-01 -3.94302085e-02 -1.26190439e-01 1.07584488e+00
-2.88124710e-01 -9.53119993e-01 -2.24823818e-01 2.24421009e-01
1.41713381e-01 -3.96432668e-01 -3.18406880e-01 1.03370619e+00
6.03515923e-01 6.56615436e-01 7.15174258e-01 -2.17522740e-01
3.59611392e-01 4.27881395e-03 7.01236844e-01 -6.16313994e-01
1.25413686e-02 -6.70040905e-01 2.03126967e-01 -6.61691010e-01
-6.35055065e-01 -6.42854929e-01 -1.33303607e+00 -2.42152661e-01
-7.41140768e-02 -5.98545745e-02 4.60330814e-01 1.22296429e+00
2.51725435e-01 3.52176160e-01 -1.51378572e-01 -6.27282441e-01
-4.40174788e-02 -6.70196414e-01 -5.03129840e-01 5.29506505e-01
2.98670590e-01 -8.16141903e-01 -2.36888304e-02 2.62498885e-01]
|
[10.448945999145508, 1.782551884651184]
|
b3a0c5b5-c000-412c-b954-b2bc7e33949e
|
arrowgan-learning-to-generate-videos-by
|
2101.03710
| null |
https://arxiv.org/abs/2101.03710v1
|
https://arxiv.org/pdf/2101.03710v1.pdf
|
ArrowGAN : Learning to Generate Videos by Learning Arrow of Time
|
Training GANs on videos is even more sophisticated than on images because videos have a distinguished dimension: time. While recent methods designed a dedicated architecture considering time, generated videos are still far from indistinguishable from real videos. In this paper, we introduce ArrowGAN framework, where the discriminators learns to classify arrow of time as an auxiliary task and the generators tries to synthesize forward-running videos. We argue that the auxiliary task should be carefully chosen regarding the target domain. In addition, we explore categorical ArrowGAN with recent techniques in conditional image generation upon ArrowGAN framework, achieving the state-of-the-art performance on categorical video generation. Our extensive experiments validate the effectiveness of arrow of time as a self-supervisory task, and demonstrate that all our components of categorical ArrowGAN lead to the improvement regarding video inception score and Frechet video distance on three datasets: Weizmann, UCFsports, and UCF-101.
|
['Hyeran Byun', 'Youngjung Uh', 'Kibeom Hong']
|
2021-01-11
| null | null | null | null |
['conditional-image-generation']
|
['computer-vision']
|
[ 1.17198028e-01 9.91107449e-02 -3.77478957e-01 -2.11219475e-01
-6.72784865e-01 -6.50913596e-01 1.07177019e+00 -1.05783379e+00
-3.38550024e-02 8.39217901e-01 4.32281524e-01 -3.06760669e-01
2.59891152e-01 -8.17860544e-01 -1.14027154e+00 -8.72681916e-01
-1.58312038e-01 9.72241312e-02 -2.74186373e-01 -3.95778604e-02
7.63088018e-02 1.12222675e-02 -1.44246840e+00 4.06829566e-01
8.15679729e-01 9.99816895e-01 -6.92847818e-02 1.01102352e+00
4.99039739e-01 1.25523353e+00 -7.10306227e-01 -5.40911376e-01
5.92503905e-01 -1.13759208e+00 -4.18255538e-01 1.09234199e-01
5.70578873e-01 -6.85261488e-01 -7.54505992e-01 8.42667997e-01
3.93469840e-01 8.52950513e-02 8.80289376e-01 -1.76182246e+00
-1.18624318e+00 8.55999231e-01 -2.36459941e-01 8.55975002e-02
4.49659884e-01 6.98899984e-01 8.30034614e-01 -6.22966468e-01
8.66716564e-01 1.09664583e+00 4.75380450e-01 9.90851104e-01
-8.13091755e-01 -8.73192728e-01 1.48119703e-01 2.44234771e-01
-1.04952145e+00 -3.91580105e-01 6.95849538e-01 -6.20855212e-01
6.34956419e-01 2.19596788e-01 6.83838546e-01 1.86707437e+00
-4.72099595e-02 7.92602241e-01 9.74538743e-01 -2.55416244e-01
1.90455928e-01 -3.12763564e-02 -6.14420056e-01 6.07213914e-01
-3.62288058e-02 4.66859579e-01 -3.39465946e-01 4.97109413e-01
1.06095397e+00 -7.77509585e-02 -5.50446093e-01 -7.16756731e-02
-1.57511413e+00 8.99381876e-01 2.60825247e-01 1.18500583e-01
-1.87752053e-01 6.54675245e-01 4.24929589e-01 4.44758117e-01
2.81196415e-01 4.45852548e-01 2.08746325e-02 -4.35481608e-01
-8.40814292e-01 2.99802929e-01 6.15590632e-01 1.53089893e+00
3.62351060e-01 4.89370376e-01 -7.17848718e-01 8.13360661e-02
-1.64319634e-01 5.36013067e-01 5.56241751e-01 -1.32077503e+00
6.34860694e-01 4.71126512e-02 -5.62663637e-02 -8.04388225e-01
3.14000487e-01 -3.59831810e-01 -9.63901877e-01 1.12265110e-01
3.50797653e-01 -4.03435081e-01 -1.07040441e+00 2.07785487e+00
-1.67839453e-01 5.35068333e-01 1.26457408e-01 9.60215330e-01
7.78182805e-01 9.39647913e-01 2.88114697e-02 -9.07409284e-03
9.90043998e-01 -1.37424302e+00 -7.14759886e-01 1.95003211e-01
4.44430739e-01 -2.76376009e-01 1.03740406e+00 4.95487183e-01
-1.19801736e+00 -9.04688954e-01 -9.50975895e-01 -3.14463000e-03
-2.80777335e-01 3.37285787e-01 6.61492825e-01 6.51078761e-01
-1.20371640e+00 6.25103176e-01 -4.50878590e-01 -1.76339492e-01
4.60187852e-01 7.25032017e-02 -4.98254359e-01 9.29877833e-02
-1.39370811e+00 5.37315190e-01 4.00146276e-01 -1.41854212e-01
-1.65283453e+00 -6.41539574e-01 -9.00006890e-01 5.77291772e-02
2.38560066e-01 -1.05749655e+00 1.11745608e+00 -1.31606245e+00
-1.61488163e+00 7.48786628e-01 2.86868542e-01 -8.70895207e-01
1.12526441e+00 -1.91518977e-01 -5.28663099e-01 3.59073013e-01
1.55586615e-01 1.11261642e+00 1.36020458e+00 -1.04462504e+00
-6.18076563e-01 1.59042597e-01 4.44841623e-01 6.37359247e-02
-3.08261663e-01 -5.05952001e-01 -3.80810410e-01 -1.13090765e+00
-8.41053605e-01 -1.07103741e+00 2.79492348e-01 -3.06638777e-01
-4.29454029e-01 -1.87557444e-01 9.61249530e-01 -7.56934047e-01
1.23237503e+00 -2.26148558e+00 4.63223368e-01 -3.94972563e-01
2.05411077e-01 4.23592664e-02 -3.45785648e-01 4.37238991e-01
-3.80845159e-01 3.80775481e-01 4.44972999e-02 -1.99994951e-01
1.31820142e-01 6.97489921e-03 -6.27004147e-01 5.16428769e-01
3.10436994e-01 1.04852629e+00 -9.78492618e-01 -3.79258007e-01
1.93854764e-01 4.64093626e-01 -7.85004318e-01 5.35143077e-01
-4.44130689e-01 7.18989849e-01 -9.38761514e-03 6.34058475e-01
3.43295485e-01 -1.47573858e-01 -1.47522926e-01 -1.06239609e-01
7.50026107e-02 -9.92167443e-02 -6.03243232e-01 1.86132669e+00
-2.37191811e-01 7.72075176e-01 -5.22974968e-01 -9.16117013e-01
3.89099360e-01 5.35188019e-01 3.29738528e-01 -5.97275496e-01
1.57717884e-01 -5.14399447e-02 -2.12315209e-02 -6.27513468e-01
4.05924618e-01 -6.71185479e-02 -9.93281752e-02 2.39687532e-01
4.11769718e-01 -9.92201865e-02 4.45136100e-01 3.16781104e-01
9.76699531e-01 7.60263979e-01 5.02422601e-02 -2.72767525e-03
3.43371570e-01 -3.14431041e-01 3.62256348e-01 6.95358992e-01
-8.39044526e-03 8.96669567e-01 8.12134266e-01 -2.34590083e-01
-1.31010723e+00 -1.07801807e+00 4.51694310e-01 9.97370839e-01
-2.56178621e-02 -4.24296528e-01 -1.18417156e+00 -1.18969679e+00
-2.67917573e-01 9.29798007e-01 -1.05444682e+00 -2.86033839e-01
-6.06440187e-01 -2.25307420e-01 8.55088174e-01 7.41919219e-01
6.26678944e-01 -1.06774426e+00 -5.10878265e-01 -2.46027082e-01
-2.16447443e-01 -1.23213637e+00 -8.59129727e-01 -2.08721876e-01
-5.60929835e-01 -9.41678345e-01 -1.03426886e+00 -6.83729172e-01
6.69291496e-01 2.08997279e-01 1.12323666e+00 -1.83161631e-01
-1.84281729e-02 3.80152166e-01 -4.95836109e-01 -9.56175625e-02
-4.67321575e-01 -1.00283474e-02 -1.35515556e-01 1.63641348e-02
-6.93283975e-02 -4.49456006e-01 -8.56345594e-01 2.91459739e-01
-9.81561005e-01 2.89388686e-01 4.96729612e-01 1.03979552e+00
2.23163962e-01 3.87931243e-02 6.12199306e-01 -9.06399071e-01
3.57981265e-01 -7.54404426e-01 -5.23753941e-01 6.48870543e-02
-5.08173823e-01 -4.43323329e-02 1.01133978e+00 -5.88869750e-01
-1.04316449e+00 -7.68955722e-02 9.16647092e-02 -8.28765213e-01
-1.29638299e-01 1.10248588e-01 -1.63432375e-01 2.27144212e-01
5.55612922e-01 3.67908418e-01 -1.19443104e-01 1.51485696e-01
7.65534222e-01 4.50539470e-01 7.17887223e-01 -6.32911026e-01
9.28573608e-01 5.96629918e-01 -5.48337400e-02 -4.63588327e-01
-4.88284081e-01 1.43139720e-01 -2.09124267e-01 -5.05371571e-01
1.19916558e+00 -1.25507641e+00 -6.31130219e-01 4.52567011e-01
-9.59348977e-01 -6.33972108e-01 -5.00954568e-01 3.78377736e-01
-1.05716145e+00 2.23574236e-01 -6.27996981e-01 -4.92928803e-01
-1.49804443e-01 -1.20437539e+00 1.12176001e+00 2.13986456e-01
-1.32525384e-01 -9.67510581e-01 -1.18913181e-01 2.56643891e-01
4.11102533e-01 7.37914741e-01 6.07353926e-01 -3.10915709e-01
-8.32488775e-01 -1.14162542e-01 -2.97395363e-02 6.17441058e-01
5.29909274e-03 2.21019581e-01 -8.86679769e-01 -3.89873564e-01
1.11428127e-01 -4.26115662e-01 9.49720621e-01 3.16423982e-01
1.41191268e+00 -5.18060386e-01 3.93269882e-02 1.06268406e+00
1.45441198e+00 3.55232686e-01 1.01461005e+00 1.82060286e-01
7.42121577e-01 3.54676902e-01 6.43044889e-01 3.92682552e-01
3.23880345e-01 4.26835239e-01 5.46900809e-01 1.57101173e-02
-4.68328178e-01 -6.70230389e-01 8.13237369e-01 7.07138062e-01
-6.19816482e-01 -7.04601169e-01 -3.00419778e-01 4.63780642e-01
-1.74621677e+00 -1.63708305e+00 9.14377570e-02 1.88509798e+00
6.49204373e-01 7.54439039e-03 3.65121067e-01 -2.44169086e-02
8.31969082e-01 1.68611303e-01 -4.85934198e-01 1.65450182e-02
-3.69361758e-01 -3.65889370e-02 4.95453626e-01 6.58635274e-02
-1.12623703e+00 8.29334378e-01 6.38859892e+00 7.11929083e-01
-1.33092701e+00 1.95827350e-01 1.01376140e+00 -1.17511772e-01
-2.65822917e-01 -7.19706416e-02 -4.54431593e-01 1.06499636e+00
9.48219001e-01 -2.22851872e-01 6.98063552e-01 9.77820456e-01
4.73469831e-02 3.73307884e-01 -1.58270550e+00 1.16035426e+00
3.61274570e-01 -1.39846468e+00 2.17273191e-01 1.46796614e-01
1.13118482e+00 -3.77591997e-01 4.97237802e-01 6.93983316e-01
2.26962209e-01 -1.39496028e+00 1.24517524e+00 4.41914737e-01
1.33562446e+00 -4.97242868e-01 4.05179322e-01 6.21152557e-02
-9.35439110e-01 -1.28751218e-01 6.37904508e-04 -8.68324563e-02
1.90153956e-01 9.72376689e-02 -6.16271317e-01 5.64056456e-01
4.97844666e-01 1.02774680e+00 -5.03899157e-01 5.61959922e-01
-5.10569394e-01 5.82594693e-01 2.94679046e-01 2.71000355e-01
3.65985453e-01 -2.22961992e-01 2.55937517e-01 1.35474467e+00
7.60307908e-01 -3.74364741e-02 -2.20341265e-01 9.16099131e-01
-4.17283624e-01 -4.67423916e-01 -1.00331593e+00 -4.27123547e-01
3.11858684e-01 1.00930023e+00 -4.88938808e-01 -6.60405636e-01
-4.16722655e-01 1.15270066e+00 -2.87703071e-02 6.22703135e-01
-1.79303169e+00 -3.80975008e-01 4.39800888e-01 1.80717826e-01
6.28072381e-01 -9.85992625e-02 8.53658542e-02 -1.47189081e+00
-7.99155012e-02 -1.09509599e+00 2.95984566e-01 -1.04484117e+00
-1.10457003e+00 6.51004255e-01 5.13215996e-02 -1.76438463e+00
-6.76768661e-01 -5.11019528e-01 -5.76432407e-01 5.15766323e-01
-1.28883445e+00 -1.23361361e+00 -5.54164886e-01 7.36986995e-01
6.94680810e-01 -4.42795485e-01 4.23799962e-01 3.99557620e-01
-4.17651623e-01 7.67554343e-01 -1.22952543e-01 3.79997820e-01
8.49481225e-01 -1.32947111e+00 4.19476688e-01 1.12303317e+00
1.35984449e-02 4.40691173e-01 5.78443885e-01 -4.73776847e-01
-1.85846031e+00 -1.38151050e+00 1.58824861e-01 -4.36762393e-01
4.89334911e-01 -3.93911391e-01 -2.43238240e-01 1.11868858e+00
8.55070591e-01 -7.99717009e-02 4.34344500e-01 -6.75299704e-01
-5.57041049e-01 -6.33682162e-02 -1.06276798e+00 7.01098084e-01
1.47087860e+00 -5.80495894e-01 -3.28801423e-01 3.77350241e-01
9.27185953e-01 -4.87237871e-01 -7.61796415e-01 2.57101774e-01
5.43226302e-01 -1.20876825e+00 8.86959195e-01 -6.61669254e-01
1.10737455e+00 -3.37863892e-01 -1.77096665e-01 -1.48353004e+00
-3.62965763e-01 -9.74987388e-01 -3.84554803e-01 1.36406267e+00
1.29558876e-01 -4.38252360e-01 7.74683774e-01 1.14058077e-01
-2.39412799e-01 -3.79939228e-01 -5.66536546e-01 -1.14338243e+00
-3.75585281e-03 -1.63082570e-01 6.99484169e-01 9.69172239e-01
-2.66262203e-01 4.22431827e-01 -9.63209808e-01 -2.07664415e-01
5.87871909e-01 -6.63449615e-02 1.06346333e+00 -4.88478482e-01
-5.81114590e-01 -4.34140980e-01 -5.09378135e-01 -1.00629389e+00
3.64051551e-01 -6.77858531e-01 2.36560870e-02 -1.17721176e+00
5.78008108e-02 -1.38664886e-01 -1.63319744e-02 2.09671795e-01
-4.41173501e-02 3.10801774e-01 3.49709570e-01 -2.23083105e-02
-6.20610654e-01 6.01507485e-01 1.43086970e+00 -3.23991656e-01
3.90475452e-01 -3.80780935e-01 -8.00197244e-01 3.63990426e-01
6.63420022e-01 -1.90825000e-01 -8.21442664e-01 -8.42493176e-01
-4.56590317e-02 3.66615027e-01 3.62607509e-01 -1.25755227e+00
-2.39113122e-02 -2.03964233e-01 4.22926068e-01 -2.00210258e-01
3.68147433e-01 -6.49173439e-01 5.98717928e-01 5.23845911e-01
-4.09985036e-01 3.01964551e-01 -2.97351956e-01 6.65585995e-01
-3.75414938e-01 1.77924514e-01 6.91544771e-01 -3.50583971e-01
-6.87329888e-01 4.87787426e-01 -9.22254026e-02 2.06473023e-01
1.47305667e+00 -2.14121759e-01 -6.74765408e-01 -8.19041550e-01
-4.08274293e-01 -5.59923612e-03 6.99458957e-01 6.00061238e-01
5.95996857e-01 -1.60508287e+00 -8.02128017e-01 1.47504598e-01
-3.59821282e-02 -2.04097927e-01 3.28365386e-01 6.30124867e-01
-6.23660684e-01 5.75221717e-01 -5.42206585e-01 -5.09886920e-01
-7.95866191e-01 1.04344058e+00 -4.53439094e-02 -7.33666867e-02
-4.69191551e-01 8.78254294e-01 6.19804502e-01 1.33502722e-01
3.88342738e-01 -4.70515668e-01 1.15935743e-01 -6.46507889e-02
4.31193084e-01 3.56533021e-01 -3.12991768e-01 -5.22269905e-01
4.75934148e-03 4.13607627e-01 2.53343105e-01 -1.92134112e-01
1.27488959e+00 1.10314496e-01 1.61818385e-01 1.35577589e-01
1.25474346e+00 -3.25827375e-02 -1.66782773e+00 5.01403093e-01
-5.33307016e-01 -5.63412905e-01 -4.75067884e-01 -7.48927236e-01
-1.29598260e+00 8.23190570e-01 3.84742230e-01 3.81136984e-01
1.40817916e+00 -1.95592210e-01 7.89789259e-01 -1.21853217e-01
3.60901564e-01 -7.76536405e-01 4.01951015e-01 1.76984280e-01
1.03466594e+00 -1.14597261e+00 -3.42294097e-01 -1.58985302e-01
-9.67696965e-01 9.96572673e-01 9.02048886e-01 -4.33528870e-01
1.66795030e-01 3.55003588e-02 -2.70322978e-01 2.18616188e-01
-9.72418010e-01 1.31241590e-01 4.32795547e-02 7.39243925e-01
4.10432428e-01 1.16712041e-01 -1.82271451e-01 4.69946176e-01
-3.00907940e-01 3.35701823e-01 6.07708991e-01 6.47225201e-01
2.86007941e-01 -8.99214983e-01 -1.64903358e-01 2.21483245e-01
-5.66223323e-01 -6.31035492e-02 -1.91135462e-02 9.64984655e-01
3.12232047e-01 6.78252876e-01 6.56908378e-02 -6.76282942e-01
2.39659958e-02 -1.71842083e-01 6.70942545e-01 -3.15447867e-01
-4.38921541e-01 -1.63584247e-01 7.53313005e-02 -6.06593013e-01
-5.65175891e-01 -4.31345046e-01 -8.55493605e-01 -3.54917556e-01
-6.63487911e-02 2.95753717e-01 5.27452409e-01 5.10255396e-01
5.60191989e-01 7.17474639e-01 7.40634620e-01 -9.42025602e-01
-5.54448962e-01 -8.94914389e-01 -2.18779474e-01 8.33311856e-01
3.57091933e-01 -5.07982016e-01 -5.05539536e-01 7.18841851e-01]
|
[10.8882474899292, -0.5894728899002075]
|
4d2419c4-4bc5-432d-81aa-d2987210da30
|
the-cultivated-practices-of-text-to-image
|
2306.11393
| null |
https://arxiv.org/abs/2306.11393v1
|
https://arxiv.org/pdf/2306.11393v1.pdf
|
The Cultivated Practices of Text-to-Image Generation
|
Humankind is entering a novel creative era in which anybody can synthesize digital information using generative artificial intelligence (AI). Text-to-image generation, in particular, has become vastly popular and millions of practitioners produce AI-generated images and AI art online. This chapter first gives an overview of the key developments that enabled a healthy co-creative online ecosystem around text-to-image generation to rapidly emerge, followed by a high-level description of key elements in this ecosystem. A particular focus is placed on prompt engineering, a creative practice that has been embraced by the AI art community. It is then argued that the emerging co-creative ecosystem constitutes an intelligent system on its own - a system that both supports human creativity, but also potentially entraps future generations and limits future development efforts in AI. The chapter discusses the potential risks and dangers of cultivating this co-creative ecosystem, such as the bias inherent in today's training data, potential quality degradation in future image generation systems due to synthetic data becoming common place, and the potential long-term effects of text-to-image generation on people's imagination, ambitions, and development.
|
['Jonas Oppenlaender']
|
2023-06-20
| null | null | null | null |
['prompt-engineering']
|
['natural-language-processing']
|
[ 6.11333907e-01 5.18915415e-01 2.56341219e-01 2.16087982e-01
-1.29603222e-01 -6.40972733e-01 1.04342079e+00 -4.30472344e-01
9.90303233e-03 6.65559530e-01 6.56960905e-01 -8.02925527e-02
4.39455472e-02 -1.01180613e+00 -6.36525095e-01 -3.20287436e-01
4.00967062e-01 3.37448508e-01 -4.91267860e-01 -4.12233710e-01
5.42820811e-01 3.30014914e-01 -1.93739879e+00 4.08941865e-01
1.17746031e+00 6.19589746e-01 1.60078481e-01 7.35564411e-01
-4.24597323e-01 1.03714216e+00 -9.66484547e-01 -8.63040686e-01
2.44430810e-01 -9.48223889e-01 -5.52792192e-01 2.76333630e-01
3.93310696e-01 -2.26931974e-01 -5.44086620e-02 8.79228413e-01
4.58002269e-01 -3.00839722e-01 4.95350838e-01 -1.28370368e+00
-1.56234992e+00 3.60815853e-01 -9.38478187e-02 -1.20464213e-01
4.30937111e-01 8.34130466e-01 4.20342147e-01 -8.11766148e-01
1.11829901e+00 1.30826080e+00 7.49333978e-01 4.28615838e-01
-1.30544364e+00 -7.97971070e-01 -2.77976364e-01 -1.94740668e-01
-1.32420325e+00 -5.10780275e-01 8.33884656e-01 -8.56228948e-01
8.42934489e-01 3.63765776e-01 1.72411597e+00 1.19130242e+00
4.87434298e-01 5.88418067e-01 1.13829064e+00 -7.20616400e-01
9.35686976e-02 4.40131009e-01 -8.00344646e-01 4.57688063e-01
4.27349031e-01 1.75243229e-01 -7.41636097e-01 1.80087641e-01
1.05059552e+00 -3.04399341e-01 1.17514014e-01 1.34972811e-01
-1.38726807e+00 7.62354076e-01 2.81171709e-01 6.62379920e-01
-6.31077349e-01 3.92733037e-01 6.83816373e-02 3.06751698e-01
6.74080610e-01 1.05550170e+00 3.70330155e-01 -5.38312256e-01
-1.03055465e+00 4.74932224e-01 6.11924827e-01 5.99447131e-01
4.95956600e-01 4.16403353e-01 -1.58143602e-02 7.85357714e-01
2.77782083e-01 7.26801217e-01 5.19249439e-01 -1.18779719e+00
-1.04554750e-01 8.06856215e-01 2.24049166e-02 -1.35714710e+00
2.40455985e-01 -5.34569919e-01 -5.29041111e-01 5.82541406e-01
1.06929295e-01 -3.34274709e-01 -8.92754793e-01 1.30454099e+00
-2.77057160e-02 -3.27563941e-01 -8.43467116e-02 6.86107635e-01
6.56448424e-01 6.84715867e-01 2.31731087e-02 -7.19245896e-02
9.65501666e-01 -7.61051893e-01 -8.61603737e-01 -5.27393579e-01
2.83226296e-02 -1.01588023e+00 1.20997536e+00 3.41725081e-01
-1.56298625e+00 -4.79194909e-01 -1.21451998e+00 -3.92190218e-02
-6.18037641e-01 -2.86669582e-01 8.23311985e-01 1.00113702e+00
-1.12131739e+00 5.65839410e-01 -3.82600099e-01 -6.66532636e-01
8.11745167e-01 -4.67020608e-02 -5.22063114e-02 -9.21800062e-02
-9.19482470e-01 9.17318583e-01 1.60316899e-01 -5.95844015e-02
-2.45421171e-01 -8.23104143e-01 -3.62448812e-01 -4.98135835e-01
6.22260012e-02 -1.17525256e+00 9.96936142e-01 -1.83661675e+00
-1.41783977e+00 9.10849214e-01 4.97319363e-02 -1.70425221e-01
7.84252167e-01 -2.90156364e-01 -5.85588455e-01 5.65679334e-02
3.59665006e-01 9.61132169e-01 7.67058790e-01 -1.51054168e+00
-3.66593897e-01 -2.04665050e-01 -4.07238662e-01 2.34083697e-01
-4.78163064e-01 -7.40861446e-02 8.65623057e-02 -1.03878796e+00
-3.93027700e-02 -8.95006061e-01 1.05279889e-02 1.44814357e-01
-1.01534419e-01 -2.04180498e-02 7.67168403e-01 -4.99812841e-01
9.96249020e-01 -2.05846643e+00 -3.58365536e-01 -3.95429991e-02
2.92089731e-01 1.97142303e-01 -4.33469452e-02 1.04767585e+00
3.96632046e-01 6.93887591e-01 1.94482580e-01 9.22477338e-03
4.76559177e-02 3.10588647e-02 -1.90370008e-01 -1.28223583e-01
3.34476024e-01 1.23376620e+00 -9.66640115e-01 -3.55827689e-01
-1.57751888e-02 8.68481994e-01 -2.90501714e-01 -2.07056716e-01
-2.70159513e-01 4.38774794e-01 -2.34222651e-01 7.90332496e-01
4.37112272e-01 -2.94167876e-01 2.51052808e-02 2.93730527e-01
-7.33730733e-01 -1.38068587e-01 -5.39760947e-01 1.41497540e+00
-3.57271880e-01 1.18887055e+00 -3.20388973e-01 -1.24104835e-01
1.11076701e+00 2.75729865e-01 2.76919782e-01 -1.04448032e+00
-4.41604815e-02 4.25833136e-01 1.66049302e-01 -6.35523796e-01
7.48198330e-01 -5.48869729e-01 2.91789144e-01 7.76547611e-01
-4.23269570e-01 -7.12213576e-01 1.62469313e-01 5.37521131e-02
8.09391499e-01 3.62903118e-01 -5.17787002e-02 -2.29424238e-01
-1.06212376e-02 5.40835738e-01 1.17543690e-01 6.21762514e-01
5.37884682e-02 5.13037086e-01 -1.60132516e-02 -6.54408634e-01
-1.49214458e+00 -1.16657794e+00 1.24707088e-01 6.85975432e-01
-3.57533395e-02 -2.21111640e-01 -7.09522486e-01 5.91807067e-02
9.27843601e-02 1.06482708e+00 -5.53151488e-01 -1.75050661e-01
-1.23360254e-01 -4.74650204e-01 5.32989800e-01 7.27786273e-02
7.56415009e-01 -1.43730068e+00 -8.97094429e-01 3.19525689e-01
-2.32195348e-01 -6.71695650e-01 -9.14243311e-02 -6.49828017e-01
-8.12556684e-01 -4.42571938e-01 -9.45021510e-01 -7.22020984e-01
8.37302983e-01 2.60642618e-01 1.24068666e+00 4.83021066e-02
-5.75908184e-01 5.89480758e-01 -2.66897827e-01 -9.66005504e-01
-9.78759646e-01 -3.56252730e-01 -2.94394642e-01 -2.39969090e-01
9.85071212e-02 -8.69314313e-01 -8.00213099e-01 -3.29766236e-02
-9.71142352e-01 8.12553942e-01 8.43689978e-01 3.85466754e-01
2.06972793e-01 1.90563127e-01 9.73379672e-01 -7.22504735e-01
8.29430163e-01 -4.13060457e-01 1.68999255e-01 2.34888256e-01
-9.80234921e-01 -3.46249551e-01 1.86800390e-01 -5.89814007e-01
-1.38287091e+00 -2.65955508e-01 3.92306954e-01 2.12332606e-02
2.64773816e-02 5.47378480e-01 2.46457849e-02 4.01216969e-02
9.53774393e-01 2.34281421e-01 4.68730092e-01 -1.40671059e-01
5.86914301e-01 7.36023843e-01 6.37073696e-01 -2.39691243e-01
1.02056289e+00 4.60358709e-01 -1.98212445e-01 -1.11658561e+00
-1.58320948e-01 3.33727688e-01 -2.64757216e-01 -9.16780770e-01
8.13907385e-01 -8.76616955e-01 -1.92141905e-01 7.19980896e-01
-1.11446917e+00 -4.02489036e-01 -8.82370234e-01 1.45364851e-01
-4.69984233e-01 -2.01356977e-01 -2.10727856e-01 -9.58306253e-01
-4.46734220e-01 -7.30582297e-01 5.00335097e-01 5.50097644e-01
-6.20224893e-01 -8.35448563e-01 9.57734287e-02 9.50092018e-01
7.29424715e-01 8.70382071e-01 7.96742618e-01 1.15495836e-02
-7.88663685e-01 -2.08645359e-01 -2.32322738e-01 2.74640530e-01
1.79491878e-01 1.56066522e-01 -8.42976391e-01 1.75584763e-01
-8.80209208e-02 -3.56981486e-01 2.47239098e-01 -3.32998075e-02
1.65155157e-01 -7.73973525e-01 -1.95590824e-01 1.27360135e-01
1.51837146e+00 6.28768623e-01 9.39882815e-01 4.81278360e-01
5.29396653e-01 5.46033919e-01 2.89581209e-01 3.60452712e-01
3.37776035e-01 -2.51950994e-02 -4.69478443e-02 -1.43060327e-01
-4.69266117e-01 -5.62431455e-01 3.65514070e-01 9.64137673e-01
-4.73879814e-01 -2.95366704e-01 -1.00956547e+00 6.51349247e-01
-1.61613166e+00 -1.19786489e+00 -1.79257557e-01 1.94110036e+00
7.90048718e-01 8.06214064e-02 2.00844392e-01 -3.04184854e-02
6.11065447e-01 -9.29946080e-02 -5.68363249e-01 -6.94673538e-01
-3.50095719e-01 3.45495820e-01 1.18447475e-01 2.29795761e-02
-3.73699129e-01 8.97604644e-01 7.64187956e+00 3.68530780e-01
-1.12119174e+00 -1.58333220e-02 7.05950916e-01 -1.10457942e-01
-8.39871407e-01 -9.44910347e-02 -3.91978584e-02 4.69833136e-01
7.93318391e-01 -7.93673337e-01 6.42471850e-01 6.12844408e-01
3.42573762e-01 -2.47716874e-01 -4.94229794e-01 7.18831241e-01
4.63610679e-01 -1.82419479e+00 2.62664080e-01 3.68385404e-01
1.18443072e+00 -2.79146612e-01 3.83469164e-01 -2.54314303e-01
3.02897483e-01 -1.20103383e+00 1.07729280e+00 7.58464098e-01
9.45664823e-01 -8.20837080e-01 1.25284523e-01 7.53037035e-02
-6.53098226e-01 -1.30770251e-01 1.44202366e-01 -6.07270956e-01
1.58117458e-01 6.47191048e-01 -9.00251865e-01 5.35121150e-02
4.93472636e-01 2.69445449e-01 -7.72495389e-01 7.95234442e-01
-7.16113150e-02 4.20922399e-01 6.26196861e-02 -1.76575601e-01
-3.98991667e-02 -2.60607362e-01 6.74425602e-01 1.06158662e+00
6.68302834e-01 9.08348933e-02 -5.11695445e-01 1.43845499e+00
1.84787780e-01 -4.70420755e-02 -1.00799811e+00 -1.15551186e+00
4.97607261e-01 9.08212304e-01 -8.98995042e-01 -2.56221920e-01
-1.90863118e-01 1.19466805e+00 -3.11427891e-01 2.35221699e-01
-5.13121247e-01 -3.45412254e-01 4.28425401e-01 6.69310570e-01
-2.17803001e-01 -3.09276074e-01 -7.58599460e-01 -6.24450147e-01
-2.65550882e-01 -1.18789423e+00 -3.68937373e-01 -1.22209048e+00
-1.15121794e+00 4.32916373e-01 -2.47734517e-01 -7.70701230e-01
-3.47264618e-01 -2.68530697e-02 -6.58011734e-01 7.13482499e-01
-6.38628840e-01 -1.61558151e+00 -4.97353077e-01 -4.49473709e-02
5.88606954e-01 -2.11823896e-01 6.59244299e-01 -7.04172328e-02
-1.31420210e-01 4.40009385e-01 1.24566033e-01 -3.01208436e-01
3.83008748e-01 -7.62203217e-01 8.43077004e-01 6.93666756e-01
1.02989458e-01 5.17812729e-01 7.29805708e-01 -1.06401062e+00
-1.61979043e+00 -6.60317361e-01 1.05734074e+00 -5.00038266e-01
5.64516842e-01 -1.32121220e-01 -2.77428359e-01 5.60901225e-01
6.44246519e-01 -8.68921340e-01 8.39160502e-01 -2.96607435e-01
2.49070562e-02 2.68485874e-01 -1.17311788e+00 1.05506647e+00
1.38175738e+00 -4.27754074e-01 -3.93681914e-01 3.54061425e-01
7.58441687e-01 1.05417691e-01 -7.91837156e-01 -1.11856289e-01
1.05608058e+00 -1.08740878e+00 1.01410544e+00 -4.44593392e-02
1.02377737e+00 -1.11368023e-01 2.56675363e-01 -1.10997689e+00
-6.21065140e-01 -1.25198483e+00 3.81510407e-01 1.43035889e+00
3.15609574e-01 -7.14287221e-01 8.03720057e-01 9.18323994e-01
-1.45144954e-01 -3.10141832e-01 -5.42890370e-01 -7.01731622e-01
1.57176405e-01 -3.53219718e-01 7.06067622e-01 1.02371383e+00
2.29382273e-02 2.98406810e-01 -2.28416964e-01 -5.66632509e-01
4.34780449e-01 -1.97156295e-01 1.05132031e+00 -1.27186024e+00
7.11521208e-02 -7.80059636e-01 -5.58586478e-01 -2.06340015e-01
-6.22853220e-01 -8.17605436e-01 -9.15196240e-02 -1.98415196e+00
1.75629705e-01 -3.92302647e-02 3.88529092e-01 7.51177818e-02
2.15526402e-01 7.70593882e-01 7.26092815e-01 4.84775335e-01
-2.16382067e-03 1.12384215e-01 2.06321764e+00 1.72714852e-02
-4.85103488e-01 -6.20873868e-01 -1.25678158e+00 6.07336640e-01
7.37041116e-01 -3.37703936e-02 -6.29924595e-01 -2.96791106e-01
8.88392568e-01 -4.89662707e-01 5.67944109e-01 -1.41734111e+00
6.10124832e-03 -5.32682419e-01 8.09515059e-01 -2.31092080e-01
2.96882808e-01 -5.82962751e-01 1.12361920e+00 5.27534485e-01
-4.36332710e-02 9.14792791e-02 2.13096783e-01 1.24157719e-01
1.65265083e-01 1.71098427e-03 5.27894139e-01 -3.50109458e-01
-4.16202426e-01 -4.61450994e-01 -7.19165683e-01 2.08371654e-02
1.10057747e+00 -9.83752429e-01 -5.40535450e-01 -5.83685875e-01
-5.19368589e-01 -2.40254432e-01 8.79856884e-01 5.97160757e-01
5.12233496e-01 -1.50100338e+00 -7.74741411e-01 1.42401680e-01
-4.03229743e-02 -2.03528494e-01 2.08291233e-01 3.56076658e-01
-8.62163067e-01 -2.03468706e-02 -6.24174178e-01 1.42581016e-02
-9.09908354e-01 2.87179768e-01 -7.78905824e-02 2.41553545e-01
-8.66106510e-01 8.92624736e-01 3.79674323e-02 2.69897908e-01
-3.42992663e-01 2.43879065e-01 2.69515038e-01 4.85473201e-02
4.23218668e-01 6.85386837e-01 -6.23287022e-01 -7.32509434e-01
-1.06515726e-02 2.93548644e-01 2.78122187e-01 -6.37376249e-01
1.24226320e+00 -1.18577629e-01 -2.01634467e-01 7.45063066e-01
5.77194512e-01 -1.25032365e-01 -1.05284727e+00 4.23628032e-01
-4.93685186e-01 -7.79703319e-01 -3.31632569e-02 -1.57717907e+00
-8.00267756e-01 6.30947471e-01 5.77720582e-01 4.00154263e-01
9.76952672e-01 -6.05070367e-02 1.11065972e+00 -1.51432201e-01
-2.71557514e-02 -1.42002332e+00 7.14375615e-01 -9.32954103e-02
1.56047177e+00 -6.74400628e-01 2.76488483e-01 -2.25653276e-01
-7.79680431e-01 9.92013335e-01 5.32104015e-01 1.13106996e-01
2.27471188e-01 1.89918995e-01 1.89520597e-01 -2.87252396e-01
-6.28633916e-01 -5.08560501e-02 1.68273762e-01 9.86122251e-01
5.90155125e-01 1.33550465e-01 -5.78561604e-01 2.52869189e-01
-6.67189658e-01 7.04666495e-01 5.26960552e-01 9.88723576e-01
-4.88372535e-01 -1.18021905e+00 -7.59700239e-01 3.98427933e-01
-2.05684185e-01 6.71573952e-02 -1.09240174e+00 9.63087380e-01
8.94984782e-01 1.00535440e+00 3.53971541e-01 -4.92898524e-01
-5.87803721e-02 2.86507785e-01 5.71575940e-01 -2.06857726e-01
-8.56149554e-01 3.90851963e-03 2.23621726e-01 1.11072036e-02
-3.05715561e-01 -8.19688559e-01 -9.85881448e-01 -7.28849232e-01
-1.03420064e-01 -3.06947410e-01 1.04315078e+00 5.89539111e-01
9.00282204e-01 3.54198635e-01 2.79963583e-01 -7.64203429e-01
3.92046601e-01 -8.10508609e-01 -2.24108025e-01 1.68687701e-01
-2.43666947e-01 -2.33319134e-01 -2.72702694e-01 4.57920939e-01]
|
[9.375419616699219, 6.3292717933654785]
|
5a642ad9-36f3-42c8-984f-ce8bc9f9b2e0
|
patchbatch-a-batch-augmented-loss-for-optical
|
1512.01815
| null |
http://arxiv.org/abs/1512.01815v2
|
http://arxiv.org/pdf/1512.01815v2.pdf
|
PatchBatch: a Batch Augmented Loss for Optical Flow
|
We propose a new pipeline for optical flow computation, based on Deep
Learning techniques. We suggest using a Siamese CNN to independently, and in
parallel, compute the descriptors of both images. The learned descriptors are
then compared efficiently using the L2 norm and do not require network
processing of patch pairs. The success of the method is based on an innovative
loss function that computes higher moments of the loss distributions for each
training batch. Combined with an Approximate Nearest Neighbor patch matching
method and a flow interpolation technique, state of the art performance is
obtained on the most challenging and competitive optical flow benchmarks.
|
['Lior Wolf', 'David Gadot']
|
2015-12-06
|
patchbatch-a-batch-augmented-loss-for-optical-1
|
http://openaccess.thecvf.com/content_cvpr_2016/html/Gadot_PatchBatch_A_Batch_CVPR_2016_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2016/papers/Gadot_PatchBatch_A_Batch_CVPR_2016_paper.pdf
|
cvpr-2016-6
|
['patch-matching']
|
['computer-vision']
|
[-3.75309885e-01 -5.37244916e-01 -8.05746838e-02 -2.38434821e-01
-5.61282814e-01 -6.46229088e-01 6.93909883e-01 3.65119964e-01
-7.63374090e-01 7.37877011e-01 1.15908735e-01 1.67837381e-01
-8.48238692e-02 -7.21950412e-01 -5.89280605e-01 -4.25414443e-01
-2.01197386e-01 3.18208933e-01 4.13079083e-01 -3.08246128e-02
7.76191592e-01 1.00779974e+00 -1.48811007e+00 1.68200999e-01
6.17368817e-01 1.29410052e+00 -2.18871713e-01 8.78215194e-01
-7.31481165e-02 1.08301997e+00 -3.15152347e-01 -5.05817533e-01
8.36602807e-01 -4.06311095e-01 -1.13257384e+00 -2.86612630e-01
1.46643281e+00 -6.99974179e-01 -6.50792480e-01 6.86260462e-01
4.51801509e-01 7.35247076e-01 4.41441864e-01 -1.07682562e+00
-3.41001123e-01 -2.50772983e-01 -1.94937140e-01 5.35757303e-01
4.53844100e-01 2.67148942e-01 1.17482960e+00 -9.66855586e-01
1.03334630e+00 1.17000186e+00 1.03845620e+00 2.27684632e-01
-1.53410256e+00 -2.03199774e-01 -3.33785951e-01 4.89419132e-01
-1.26175380e+00 -4.70438838e-01 6.39257610e-01 -5.68879604e-01
1.29585552e+00 -2.27539778e-01 1.00847948e+00 3.29074770e-01
2.54879355e-01 5.46380341e-01 7.43763566e-01 -2.23790377e-01
2.11802825e-01 -8.10297430e-02 -2.56710082e-01 9.71375704e-01
-7.01486170e-02 3.20122421e-01 -5.68035066e-01 6.57184049e-02
8.66419673e-01 1.27319574e-01 -3.51297677e-01 -6.10733569e-01
-1.25275075e+00 8.04659188e-01 9.02155578e-01 1.44479394e-01
-3.94339263e-01 4.72421259e-01 5.92312157e-01 2.63122022e-01
4.68200535e-01 4.35089767e-01 -2.14721426e-01 -2.99411833e-01
-1.40563011e+00 4.80963826e-01 9.12810087e-01 6.34331286e-01
1.38855660e+00 -3.79203647e-01 -3.96435618e-01 5.53801060e-01
2.48990014e-01 2.71844774e-01 1.02427773e-01 -1.60190070e+00
2.30704427e-01 2.30446160e-01 1.41678482e-01 -1.44907331e+00
-1.62647441e-01 -9.85675603e-02 -5.82650542e-01 5.36947608e-01
8.20514441e-01 2.48462379e-01 -5.99861324e-01 1.34900832e+00
3.46397281e-01 8.32097352e-01 -3.21549535e-01 8.49432290e-01
3.83799791e-01 5.19034028e-01 -1.36069730e-01 1.22256987e-01
6.90707743e-01 -1.34182906e+00 -2.35126078e-01 2.52089411e-01
6.79710865e-01 -9.86789227e-01 8.54121447e-01 2.66626060e-01
-1.40433979e+00 -7.58235455e-01 -9.83491957e-01 -8.01547587e-01
-2.50202775e-01 -1.82541415e-01 6.00708902e-01 2.67869532e-01
-1.48223639e+00 1.33861113e+00 -9.37116504e-01 -1.44112706e-01
8.39588702e-01 3.71237934e-01 -7.00874746e-01 -2.34407216e-01
-6.62362218e-01 7.90301979e-01 1.91911124e-02 1.56648889e-01
-6.41817153e-01 -1.36593938e+00 -9.79113519e-01 1.26698077e-01
-3.49329054e-01 -9.36555743e-01 8.06050599e-01 -9.15702701e-01
-1.82490337e+00 8.97372484e-01 -2.93890268e-01 -5.37775576e-01
6.77229047e-01 -1.93462014e-01 2.56071568e-01 7.43658304e-01
7.42253140e-02 9.64265704e-01 8.41344297e-01 -6.16092622e-01
-7.19308555e-01 -1.25845194e-01 -5.29846251e-02 -1.89267203e-01
-4.69649918e-02 -2.54676759e-01 -1.56428486e-01 -5.08585811e-01
-1.72497630e-01 -8.02159429e-01 -2.34462291e-01 7.89820075e-01
1.83218986e-01 -2.26340175e-01 6.90164983e-01 -5.89831531e-01
8.00380647e-01 -2.06422091e+00 1.09581560e-01 3.62241000e-01
4.63708729e-01 4.57582265e-01 -2.02016130e-01 2.40694538e-01
1.51953623e-01 -2.84472346e-01 -1.93237856e-01 -6.60295010e-01
-1.05766244e-01 1.38407545e-02 -3.23530853e-01 8.01250517e-01
5.35725832e-01 8.83036137e-01 -1.18019950e+00 -5.09474218e-01
5.41855454e-01 8.02341640e-01 -1.23497665e+00 4.63092595e-01
8.86559188e-02 5.15756488e-01 1.74270660e-01 9.17237997e-02
8.24130535e-01 -1.08145021e-01 -3.51218253e-01 -5.77633083e-01
-1.61546439e-01 6.24881685e-01 -1.15915537e+00 2.49958253e+00
-5.56786776e-01 1.09011197e+00 -1.73477218e-01 -1.02750874e+00
8.77107441e-01 1.58772413e-02 7.13579059e-01 -5.47075868e-01
8.62123910e-03 5.52265465e-01 -1.74485028e-01 -2.62041152e-01
2.92490691e-01 -2.29656175e-02 7.34205961e-01 4.27880108e-01
5.18101394e-01 -4.25217509e-01 6.60451949e-01 1.00643143e-01
1.17212939e+00 4.61794555e-01 -1.49787040e-02 -5.03662944e-01
9.33033109e-01 -1.03738569e-02 4.85998511e-01 6.04500473e-01
-6.61971271e-01 7.43773818e-01 5.72179437e-01 -1.15510380e+00
-1.25425506e+00 -1.12332869e+00 -1.39754474e-01 5.95676899e-01
1.42443359e-01 -4.46611941e-01 -5.73771060e-01 -8.43263149e-01
3.01834226e-01 -6.17133360e-03 -5.76322079e-01 -7.44798928e-02
-9.80802357e-01 2.85345986e-02 4.85222548e-01 5.40652931e-01
4.97590482e-01 -9.33887899e-01 -6.58357859e-01 3.52321029e-01
9.42307040e-02 -1.21554828e+00 -7.90845692e-01 -2.84896851e-01
-8.78106177e-01 -1.16021872e+00 -7.49543428e-01 -8.53616953e-01
4.71463054e-01 1.35312065e-01 1.35800838e+00 3.63309920e-01
-5.70988178e-01 3.19953233e-01 3.18791308e-02 3.49711776e-01
-6.43666685e-02 5.86260222e-02 -3.43748719e-01 1.99856222e-01
2.37258613e-01 -7.23543882e-01 -1.17706549e+00 4.77784052e-02
-5.80004573e-01 -5.23088872e-01 1.21028259e-01 8.73201251e-01
6.17978990e-01 -6.35668516e-01 -1.17687099e-01 -3.80683690e-01
2.66325831e-01 -1.08084075e-01 -8.46938014e-01 -1.43972356e-02
-5.32025456e-01 2.25339279e-01 7.43967354e-01 -1.72428340e-01
-6.59567475e-01 2.68349648e-01 -1.22712232e-01 -8.23802352e-01
-4.91961874e-02 -1.45821832e-02 6.26131296e-01 -7.62747705e-01
6.18561447e-01 -9.72801596e-02 2.81506479e-01 -2.41983935e-01
4.47856247e-01 3.24268304e-02 6.50818467e-01 -4.87720251e-01
5.90727091e-01 9.09056842e-01 6.54530406e-01 -6.33590043e-01
-6.37659311e-01 -7.14121401e-01 -9.40962553e-01 -1.72841638e-01
8.12903583e-01 -6.48874998e-01 -1.09866214e+00 5.05870044e-01
-1.54148376e+00 -3.56269985e-01 -5.47712743e-01 9.77950513e-01
-8.53501678e-01 4.87511128e-01 -8.65001976e-01 -2.15130568e-01
-4.09138441e-01 -1.31616068e+00 1.06722164e+00 1.58021927e-01
-4.47211936e-02 -1.36427069e+00 6.74609840e-01 1.00330450e-01
7.00625122e-01 1.56320348e-01 3.92239898e-01 -3.27435017e-01
-9.03700888e-01 -1.98836252e-01 -6.72488570e-01 7.08439648e-01
-3.76264937e-02 3.42771769e-01 -8.85226250e-01 -3.36600155e-01
-3.48434806e-01 -3.86083305e-01 1.09314394e+00 4.43359077e-01
9.56739664e-01 3.03643327e-02 8.17493275e-02 1.21550298e+00
1.61514699e+00 -3.84738654e-01 8.65198016e-01 1.13070197e-01
8.04793119e-01 6.33386672e-01 9.48680788e-02 3.71230692e-01
3.46105874e-01 6.99396193e-01 2.10557967e-01 -1.14595648e-02
-5.47346175e-01 -1.76942974e-01 1.55973881e-01 6.04811192e-01
-1.74454451e-01 2.68727958e-01 -8.20324183e-01 5.71395159e-01
-1.72066808e+00 -1.19072556e+00 -1.19758964e-01 2.34522367e+00
6.33698046e-01 -1.27232417e-01 1.25268660e-03 -3.46832387e-02
3.35044146e-01 2.35915259e-01 -2.02988774e-01 -6.19356334e-01
1.21014668e-02 9.27997530e-01 4.92305875e-01 1.00379908e+00
-1.19249117e+00 9.35817420e-01 6.87661552e+00 5.42389631e-01
-1.18135953e+00 -2.27661543e-02 5.13210356e-01 -1.74650803e-01
7.58989081e-02 2.17650741e-01 -5.81225276e-01 4.15422916e-01
7.39939153e-01 1.69201307e-02 5.72438240e-01 5.38739145e-01
9.21341032e-02 -2.01176777e-01 -1.27311516e+00 1.21179533e+00
1.46184042e-01 -1.98223901e+00 4.50417213e-02 -3.48082669e-02
8.77781510e-01 2.93736756e-01 -1.77056089e-01 -1.99888334e-01
-1.76456571e-01 -8.53546143e-01 5.42696118e-01 7.78338671e-01
4.17160898e-01 -7.07737267e-01 6.11019552e-01 -3.57240409e-01
-1.30695868e+00 2.22730134e-02 -4.69393194e-01 -1.69618905e-01
1.44695580e-01 6.00984514e-01 -3.50592792e-01 2.09410205e-01
7.95215726e-01 1.49668777e+00 -3.63580853e-01 1.52225065e+00
-1.17277488e-01 1.01296604e-01 -3.39550763e-01 2.69364387e-01
4.38780695e-01 -4.26357150e-01 3.93328905e-01 1.35162723e+00
1.19891733e-01 -4.12365168e-01 1.67107180e-01 1.08241820e+00
-1.41414344e-01 2.43201435e-01 -5.00318229e-01 3.75930518e-01
2.76218265e-01 1.40785015e+00 -4.99271989e-01 -3.50002229e-01
-4.89691406e-01 1.09892702e+00 7.53157794e-01 2.00534478e-01
-3.59219760e-01 -8.47065508e-01 1.02731013e+00 -1.94504354e-02
4.57495362e-01 -5.05979419e-01 -8.06627423e-02 -1.21337366e+00
2.56035805e-01 -2.51289427e-01 8.15033317e-02 -4.80803788e-01
-1.42378283e+00 3.67707849e-01 -3.84089738e-01 -1.18356633e+00
-3.75846088e-01 -7.75384605e-01 -7.89933562e-01 1.11761534e+00
-1.99302411e+00 -7.96727002e-01 -4.46949244e-01 8.94613504e-01
3.07351202e-02 4.81657451e-03 5.54030120e-01 7.26175964e-01
-3.11455458e-01 5.86708844e-01 -8.03323835e-02 3.53688776e-01
9.57055151e-01 -1.29309416e+00 4.75956231e-01 8.03198218e-01
1.59608856e-01 4.24250454e-01 3.24612498e-01 -4.43385430e-02
-1.16394246e+00 -8.76046240e-01 1.15670168e+00 -3.33692014e-01
7.31666028e-01 -1.52876094e-01 -9.61327612e-01 4.52201456e-01
1.88344568e-01 1.06871867e+00 3.79249305e-01 -3.59834820e-01
-6.24108016e-01 -4.38632220e-01 -1.27076638e+00 6.91862777e-02
9.66081798e-01 -7.85322547e-01 -2.56433040e-01 3.73533547e-01
3.26577455e-01 -3.52557600e-01 -1.04774022e+00 1.52223438e-01
7.48346150e-01 -1.41118622e+00 1.18970013e+00 -7.11385667e-01
6.46320581e-01 -3.03488880e-01 2.06158850e-02 -1.09918809e+00
-3.32909048e-01 -9.41208005e-01 -1.16952024e-01 7.16030002e-01
1.21590033e-01 -5.79982162e-01 9.42954898e-01 5.07744610e-01
-1.18264645e-01 -6.44617021e-01 -1.08499277e+00 -7.55470335e-01
1.64317444e-01 -1.73872694e-01 2.64171153e-01 6.99676752e-01
-1.38141692e-01 -1.97844267e-01 -1.18013300e-01 -2.36395210e-01
7.47927785e-01 5.20737991e-02 8.35564852e-01 -1.19640076e+00
-2.57540673e-01 -7.63424218e-01 -1.22873878e+00 -1.22659636e+00
6.19197428e-01 -1.04441237e+00 5.64301498e-02 -1.21910298e+00
-3.04405153e-01 -3.43715161e-01 -3.51016104e-01 2.69260108e-01
-7.43448883e-02 6.22321367e-01 3.23486000e-01 2.54653484e-01
-5.37789762e-01 5.87110221e-01 1.31157362e+00 -5.95022291e-02
-1.82299927e-01 -3.11790675e-01 2.49546185e-01 3.39277148e-01
5.26988864e-01 -3.52320105e-01 -1.63077220e-01 -6.28010094e-01
7.95854479e-02 -3.19880843e-01 6.62770391e-01 -1.28720486e+00
4.12618369e-01 9.39773768e-02 3.04272681e-01 -2.13000476e-01
2.24837318e-01 -6.79441988e-01 -3.65053594e-01 6.07927084e-01
-3.74936908e-01 2.15992525e-01 1.42604068e-01 2.71063685e-01
-4.46139514e-01 -3.89152057e-02 1.15071976e+00 1.27453566e-01
-5.78850985e-01 7.66531467e-01 2.12796614e-01 4.25807565e-01
8.63163292e-01 -1.72815263e-01 -1.34881034e-01 -5.32905124e-02
-4.51538295e-01 -1.96199313e-01 4.30101097e-01 1.30867511e-01
6.02626741e-01 -1.48599088e+00 -8.88081193e-01 5.65573335e-01
-1.94243386e-01 -1.75533459e-01 4.41948548e-02 1.17766666e+00
-1.35572481e+00 2.51722127e-01 -7.30077088e-01 -7.04200506e-01
-6.34472609e-01 4.48475569e-01 8.12123418e-01 -3.13165814e-01
-5.25515974e-01 9.46769297e-01 -3.28284234e-01 -1.56466097e-01
1.23117670e-01 -2.28228644e-01 2.29307503e-01 5.23688318e-03
7.86276877e-01 9.00735915e-01 3.12724352e-01 -6.43833995e-01
-4.94237065e-01 8.61871243e-01 9.67513025e-02 -3.03964056e-02
1.18439364e+00 4.43057753e-02 -4.71071601e-01 1.30741596e-02
2.03683805e+00 -1.38182431e-01 -1.65455818e+00 -2.08425447e-01
-1.28866971e-01 -1.03750062e+00 3.02953303e-01 -2.16315478e-01
-1.56890666e+00 1.25732636e+00 6.93120122e-01 -1.63570046e-01
9.33323085e-01 -4.55217510e-01 1.11377501e+00 3.51852894e-01
-2.32305303e-02 -1.00581920e+00 2.13642325e-02 6.82209671e-01
6.47650003e-01 -1.25197661e+00 -2.28304211e-02 -1.61191091e-01
1.45062909e-01 1.54563415e+00 3.21449548e-01 -1.05047965e+00
1.04443181e+00 5.77749200e-02 8.20545554e-02 1.04770206e-01
-5.76686919e-01 -2.17318788e-01 4.83815700e-01 5.79350173e-01
6.22273982e-01 -5.84224224e-01 -3.03351581e-01 -6.09896362e-01
3.92057411e-02 1.67577937e-01 2.04588249e-01 6.89201355e-01
-1.07123397e-01 -1.23228002e+00 1.59843609e-01 2.83815861e-01
-2.74574816e-01 -2.18511820e-01 1.24116167e-01 4.60551113e-01
8.60568732e-02 6.27336860e-01 5.42412043e-01 -5.48300818e-02
3.79657269e-01 -2.83838719e-01 9.01000321e-01 -2.15423912e-01
-8.26284647e-01 -3.54901493e-01 -5.27180016e-01 -1.39649451e+00
-7.41089404e-01 -5.01833856e-01 -1.10081327e+00 -7.68632948e-01
3.24322134e-01 -4.06472236e-02 5.20683765e-01 8.40456486e-01
6.23652160e-01 -9.22495872e-02 7.96549678e-01 -1.42005587e+00
-3.88708204e-01 -4.62971807e-01 -3.32989573e-01 8.75752568e-01
6.40957892e-01 -5.66973805e-01 -5.16376078e-01 3.44506316e-02]
|
[8.77652359008789, -1.8855361938476562]
|
5ffa8c44-1bfc-48e9-a50f-63e4a9fbcdcb
|
image-decomposition-using-a-robust-regression
|
1609.03874
| null |
http://arxiv.org/abs/1609.03874v2
|
http://arxiv.org/pdf/1609.03874v2.pdf
|
Image Decomposition Using a Robust Regression Approach
|
This paper considers how to separate text and/or graphics from smooth
background in screen content and mixed content images and proposes an algorithm
to perform this segmentation task. The proposed methods make use of the fact
that the background in each block is usually smoothly varying and can be
modeled well by a linear combination of a few smoothly varying basis functions,
while the foreground text and graphics create sharp discontinuity. This
algorithm separates the background and foreground pixels by trying to fit pixel
values in the block into a smooth function using a robust regression method.
The inlier pixels that can be well represented with the smooth model will be
considered as background, while remaining outlier pixels will be considered
foreground. We have also created a dataset of screen content images extracted
from HEVC standard test sequences for screen content coding with their ground
truth segmentation result which can be used for this task. The proposed
algorithm has been tested on the dataset mentioned above and is shown to have
superior performance over other methods, such as the hierarchical k-means
clustering algorithm, shape primitive extraction and coding, and the least
absolute deviation fitting scheme for foreground segmentation.
|
['Yao Wang', 'Shervin Minaee']
|
2016-09-13
| null | null | null | null |
['foreground-segmentation']
|
['computer-vision']
|
[ 5.12794673e-01 -7.50216618e-02 1.26961410e-01 -3.16312701e-01
-5.05376220e-01 -3.75927210e-01 4.03131783e-01 -7.05453828e-02
-2.15213094e-02 5.95447183e-01 -3.18730712e-01 -2.07120582e-01
2.27754027e-01 -4.93768424e-01 -4.59766239e-01 -9.60194409e-01
3.36743206e-01 6.77034020e-01 9.25902784e-01 6.26791492e-02
5.41141272e-01 7.26511300e-01 -1.46005726e+00 2.75206298e-01
9.38358724e-01 8.12752485e-01 9.56612080e-02 9.32523787e-01
-5.49437463e-01 8.78570676e-01 -6.06320143e-01 -2.29799405e-01
3.92396718e-01 -6.37898386e-01 -3.86793673e-01 7.22758651e-01
6.08808994e-01 -3.86794627e-01 1.42010739e-02 1.29009175e+00
2.11030364e-01 1.20734796e-01 8.53389025e-01 -9.83173966e-01
-2.90953945e-02 4.78071757e-02 -1.08487582e+00 1.18052125e-01
2.55113214e-01 -2.67819673e-01 3.06336284e-01 -7.96959639e-01
6.77489519e-01 1.19122207e+00 5.88091791e-01 1.84725642e-01
-1.33797359e+00 -4.34361368e-01 -2.43710484e-02 -4.46016826e-02
-1.59803152e+00 -5.23688197e-01 8.88815463e-01 -6.01213217e-01
4.47956890e-01 6.86551273e-01 5.96222758e-01 1.02699384e-01
1.61098093e-01 7.89789736e-01 9.11011457e-01 -6.17302001e-01
2.87998497e-01 3.97776693e-01 4.35617268e-01 6.61204934e-01
4.19552863e-01 -4.57645535e-01 3.43492627e-01 -3.98604810e-01
1.12802994e+00 -1.02309331e-01 -5.19676149e-01 -4.42864567e-01
-9.52289939e-01 5.03544867e-01 -4.11216496e-03 5.23477376e-01
-3.28381479e-01 8.44423845e-02 4.78119478e-02 -8.98868516e-02
3.30242485e-01 -1.76866874e-01 -1.86509848e-01 9.56933498e-02
-1.70340014e+00 -2.70658974e-02 8.07079375e-01 1.19973433e+00
7.04618454e-01 2.86188692e-01 6.46691248e-02 8.10929000e-01
4.21402693e-01 5.53766429e-01 3.59730899e-01 -9.93494511e-01
1.07589193e-01 6.79350138e-01 2.21544817e-01 -1.36806798e+00
1.11973479e-01 2.92047858e-01 -5.86847305e-01 6.35240614e-01
5.76673210e-01 -2.85351053e-02 -1.20482552e+00 9.39735353e-01
5.14882863e-01 4.56210285e-01 -2.36861855e-01 6.36287570e-01
9.29794967e-01 9.81988788e-01 -5.21708488e-01 -5.66161633e-01
8.99694562e-01 -7.94123709e-01 -1.02149367e+00 3.65508705e-01
7.40625784e-02 -1.21217251e+00 5.67733705e-01 8.69427264e-01
-1.48585165e+00 -7.27346480e-01 -9.16548371e-01 1.07214332e-01
4.40490730e-02 3.55081856e-01 1.56821057e-01 8.66446197e-01
-1.00472605e+00 5.04627645e-01 -9.76034760e-01 -2.22747728e-01
9.82296243e-02 5.11744082e-01 -2.54156627e-02 2.41168648e-01
-2.38094106e-01 6.56833470e-01 4.76598024e-01 1.67954668e-01
-5.48503518e-01 -1.55294180e-01 -3.74531150e-01 -1.94428086e-01
3.24773282e-01 -1.47346199e-01 6.97275162e-01 -1.62314713e+00
-1.45243037e+00 9.61725116e-01 -4.89162385e-01 -1.26560107e-01
8.00679564e-01 1.35132119e-01 -4.52446103e-01 4.37734246e-01
-2.96004742e-01 1.92503810e-01 1.31241620e+00 -1.71086133e+00
-8.24953914e-01 -1.13537557e-01 -5.92539608e-01 -8.98671243e-03
2.45744914e-01 1.79049850e-01 -1.04430389e+00 -5.97647727e-01
7.23625064e-01 -8.26072812e-01 -1.79029748e-01 -1.54742643e-01
-4.18227166e-01 3.25526506e-01 1.27824020e+00 -1.23274505e+00
1.32929981e+00 -2.30409551e+00 6.85732486e-03 8.23785484e-01
8.85405391e-02 2.26274833e-01 4.70060736e-01 -1.63442969e-01
-1.38997033e-01 -3.88367437e-02 -3.89335304e-01 1.09503679e-01
-3.55999291e-01 1.16433397e-01 2.33942531e-02 7.81860650e-01
-1.76758066e-01 1.39358342e-01 -3.85705978e-01 -1.01816869e+00
4.58225429e-01 4.27860171e-01 -2.48499557e-01 7.34376609e-02
-1.48195639e-01 2.60685802e-01 -2.67784148e-01 7.66748905e-01
1.24263442e+00 1.43462464e-01 -1.69546738e-01 -2.52645016e-01
-1.31120816e-01 -5.13844252e-01 -1.94713569e+00 8.09913576e-01
3.73457670e-01 8.74872565e-01 5.50022066e-01 -7.25337148e-01
1.18086827e+00 3.33131284e-01 6.93296134e-01 -3.85821201e-02
4.59966689e-01 2.78582722e-01 -4.85146046e-03 -4.06284750e-01
6.16317689e-01 1.70155928e-01 5.75570941e-01 -9.75857154e-02
-2.89995611e-01 -6.38987720e-01 1.80530354e-01 -2.53551789e-02
5.51699936e-01 2.28104725e-01 2.76184201e-01 -3.37176651e-01
1.04387510e+00 1.52282834e-01 6.82764709e-01 4.56028461e-01
-1.74646303e-01 9.21126664e-01 4.12293822e-01 -8.52001086e-02
-1.16539240e+00 -7.90647864e-01 -2.90097505e-01 5.43474734e-01
4.32756782e-01 3.17001268e-02 -1.19651604e+00 -8.45361054e-02
-2.52924085e-01 6.90730512e-01 -2.23959610e-01 4.63535130e-01
-7.43127763e-01 -5.96407712e-01 1.46408185e-01 6.01001196e-02
6.93446636e-01 -7.34572709e-01 -3.34009022e-01 2.55455643e-01
-4.99972813e-02 -1.00500178e+00 -1.94498688e-01 7.81537369e-02
-1.17480409e+00 -1.21118164e+00 -9.72623825e-01 -9.20467675e-01
8.75413895e-01 5.28632581e-01 8.47122669e-01 2.94488430e-01
-3.46339941e-01 4.46483940e-01 -3.10514748e-01 -2.32140183e-01
-6.74659371e-01 -6.66867614e-01 -4.07304466e-01 3.85244697e-01
2.39959329e-01 -1.85072228e-01 -2.97586054e-01 5.08948982e-01
-8.08931947e-01 1.38352001e-02 1.31723389e-01 4.30612952e-01
1.03338897e+00 6.11420691e-01 -3.35506886e-01 -9.72001731e-01
1.68858066e-01 -1.80750206e-01 -9.92365956e-01 2.83316165e-01
-5.17931134e-02 -5.09298027e-01 4.21915412e-01 -3.79404426e-01
-1.38221681e+00 3.25996578e-01 1.44906282e-01 -4.80862677e-01
-4.05155092e-01 -1.47634417e-01 -5.04168689e-01 -4.23156381e-01
4.55454320e-01 2.74812222e-01 -1.97599322e-01 -4.05458808e-01
2.50169635e-01 9.02795911e-01 6.99887455e-01 -3.25633466e-01
7.66637146e-01 7.56853700e-01 2.50251181e-02 -1.53878462e+00
6.85329735e-02 -8.48987341e-01 -7.77694643e-01 -4.73147273e-01
1.13483536e+00 -6.34778023e-01 -9.94648039e-02 5.78983128e-01
-9.99992013e-01 -1.47638842e-01 -6.62368536e-02 3.64395708e-01
-6.12402558e-01 8.78735185e-01 -6.12104833e-01 -1.26167834e+00
1.50786592e-02 -1.23305929e+00 8.51542830e-01 3.09867501e-01
-1.15648754e-01 -9.61913943e-01 -2.31971711e-01 3.06639075e-01
-7.04987273e-02 5.29095709e-01 8.48196924e-01 -5.40823638e-01
-7.65655994e-01 -3.38182360e-01 -3.81029360e-02 5.30220389e-01
2.03460485e-01 9.93731678e-01 -9.05757129e-01 -1.17001913e-01
5.40812254e-01 3.59403610e-01 6.18983448e-01 9.14455295e-01
9.02444720e-01 1.36464894e-01 -4.47750121e-01 8.10412347e-01
1.72930110e+00 8.09936404e-01 1.06911504e+00 3.13071966e-01
9.60382164e-01 4.99990284e-01 7.61308968e-01 2.44113550e-01
-3.12335730e-01 5.30075014e-01 2.20858172e-01 -3.88703376e-01
3.52356508e-02 4.61049348e-01 2.07584083e-01 7.77721167e-01
-3.18881303e-01 -8.43114927e-02 -8.12883019e-01 3.11941504e-01
-1.83691847e+00 -1.10313165e+00 -1.24837649e+00 2.21402216e+00
4.86232877e-01 3.13269049e-01 1.55189440e-01 5.99559188e-01
1.16168404e+00 -3.24018180e-01 -1.43351614e-01 -5.79821527e-01
-2.69992977e-01 3.49421389e-02 6.77858174e-01 6.02888107e-01
-1.22992504e+00 8.93261492e-01 6.33537149e+00 1.10913026e+00
-1.08210421e+00 -2.75635868e-01 8.50144088e-01 2.55272418e-01
-1.37886694e-02 8.50793347e-02 -7.46782124e-01 5.98952055e-01
6.12652779e-01 2.32216390e-03 2.28538767e-01 8.93099368e-01
4.21351939e-01 -7.11073160e-01 -7.71104217e-01 1.20139742e+00
1.90651894e-01 -8.05034041e-01 -5.11936024e-02 -1.56808138e-01
1.15945065e+00 -5.42378724e-01 -1.34991109e-01 -3.45871359e-01
-8.68570432e-03 -8.76599848e-01 7.36118853e-01 6.46536946e-01
3.14345092e-01 -6.16636872e-01 6.80186629e-01 4.53465164e-01
-1.13072586e+00 2.39600718e-01 -6.60128713e-01 3.01265419e-01
6.14492828e-03 5.25052845e-01 -7.28936911e-01 3.49026620e-01
4.28222388e-01 3.42393547e-01 -6.48484051e-01 1.51041794e+00
3.24669629e-01 7.01591551e-01 -5.33191681e-01 1.63445517e-01
1.43905073e-01 -1.03899550e+00 5.32238066e-01 1.30708563e+00
4.98922467e-01 2.02081770e-01 1.34686425e-01 8.84043753e-01
6.01685107e-01 5.51212549e-01 -3.68189663e-01 1.23967290e-01
5.50443642e-02 1.12826335e+00 -1.57800579e+00 -7.58304238e-01
-5.35158038e-01 1.11648393e+00 -6.52096391e-01 4.75015819e-01
-9.17952299e-01 -2.81496286e-01 -1.00471480e-02 3.38148654e-01
4.74878997e-01 -4.30947430e-02 -5.87404191e-01 -9.61900711e-01
-1.72924593e-01 -1.03026235e+00 1.46011710e-01 -9.37948346e-01
-6.94877803e-01 6.11559629e-01 1.02976084e-01 -1.36041641e+00
7.71370530e-02 -6.42473519e-01 -6.81856453e-01 9.72659349e-01
-7.82036841e-01 -8.97769213e-01 -3.96486133e-01 8.81885171e-01
6.88656628e-01 -3.49846989e-01 2.60571063e-01 2.33376414e-01
-4.62416887e-01 -4.66591306e-03 6.71621203e-01 1.11886428e-03
4.39957291e-01 -1.23623359e+00 -9.46414694e-02 1.13227999e+00
-7.28595331e-02 4.80642796e-01 9.56850946e-01 -1.03839028e+00
-8.17047596e-01 -7.86426008e-01 3.68386120e-01 -2.62965739e-01
2.30290353e-01 -8.11639279e-02 -1.18414235e+00 5.88525355e-01
5.67142554e-02 -1.61515296e-01 4.49373603e-01 -7.45330393e-01
5.11612117e-01 1.06219739e-01 -1.41004169e+00 5.80573380e-01
2.46178031e-01 3.39697115e-02 -7.18580544e-01 1.99261710e-01
6.71073142e-03 -4.88631070e-01 -4.04222757e-01 2.12940305e-01
1.75157040e-01 -1.34333432e+00 9.49968100e-01 7.35491291e-02
8.90109390e-02 -7.46684432e-01 -1.43896699e-01 -6.36988282e-01
-2.83959210e-01 -7.58168399e-01 7.54223540e-02 1.44168806e+00
9.60817114e-02 -1.10967733e-01 1.01371384e+00 7.78455734e-01
1.77023143e-01 -1.60982475e-01 -6.45653069e-01 -5.10573089e-01
-2.48659194e-01 -9.45677608e-02 1.70470521e-01 9.41339731e-01
-4.94447768e-01 -1.55326411e-01 -2.62936354e-01 -3.60943982e-03
8.16090465e-01 4.56674695e-02 1.03384113e+00 -1.39279473e+00
-2.22733244e-01 -5.61003447e-01 -5.37762880e-01 -9.12926555e-01
-1.87496960e-01 -4.58278894e-01 1.68346867e-01 -1.55667746e+00
4.47265878e-02 -4.53953892e-01 3.28733951e-01 -4.24781799e-01
-2.63927400e-01 1.47866786e-01 2.33444080e-01 3.28663051e-01
-2.70564169e-01 -1.64220184e-01 1.20140839e+00 6.20517954e-02
-5.28598368e-01 2.97360659e-01 -1.01154953e-01 1.36283112e+00
4.46707368e-01 -2.37891883e-01 -3.42841327e-01 9.79175866e-02
-3.52871031e-01 3.55046570e-01 -6.18100390e-02 -1.04720449e+00
8.39046910e-02 -1.88270316e-01 7.73039877e-01 -9.97613430e-01
3.13200653e-01 -1.54458547e+00 6.66091979e-01 1.95261136e-01
2.84457318e-02 -1.36787280e-01 1.10973045e-01 4.90688384e-01
-2.19173416e-01 -8.01800311e-01 1.18524456e+00 -2.89573789e-01
-7.51613975e-01 -2.05333695e-01 -6.94691181e-01 -2.77114451e-01
1.39901412e+00 -1.16780472e+00 3.28257173e-01 -5.19978881e-01
-9.16174591e-01 -2.12197646e-01 8.38757515e-01 -3.85257602e-01
6.65105104e-01 -1.20168316e+00 -4.66675669e-01 2.27555200e-01
-5.47803521e-01 -8.90927389e-04 -7.95041211e-03 9.48657334e-01
-1.50330126e+00 5.98006360e-02 -1.51737764e-01 -8.68252218e-01
-1.82397008e+00 6.13422453e-01 2.76105374e-01 2.38645345e-01
-6.37611687e-01 7.63718307e-01 1.36189878e-01 2.53272563e-01
4.65875357e-01 -7.42014945e-01 -3.90698344e-01 -3.37499194e-02
3.93961638e-01 8.85912001e-01 -1.18802153e-01 -1.10445821e+00
-1.50476515e-01 9.99734223e-01 3.11829060e-01 -1.76172107e-01
8.99344265e-01 -1.88214481e-01 -4.18645561e-01 6.69518948e-01
1.03211617e+00 5.94716489e-01 -1.13644707e+00 1.14237860e-01
1.56282768e-01 -8.01293910e-01 1.54528469e-01 -2.77668089e-01
-1.06927216e+00 7.46693611e-01 7.45977998e-01 5.77032626e-01
1.28413820e+00 -6.22304440e-01 5.62316358e-01 1.23857372e-01
1.80008560e-01 -1.44846427e+00 -2.25917652e-01 2.37506434e-01
6.13220632e-01 -9.95370626e-01 2.10642427e-01 -8.68073761e-01
-7.43647933e-01 1.49605155e+00 3.98129791e-01 -6.23887718e-01
5.78541994e-01 5.27744472e-01 2.60352325e-02 1.42426610e-01
-1.34966984e-01 -1.99885637e-01 2.84402221e-01 5.63256919e-01
4.74646419e-01 -1.93102643e-01 -4.38473076e-01 9.77414027e-02
1.18747234e-01 -9.62948427e-02 7.31404662e-01 8.17952454e-01
-9.21504974e-01 -7.48984337e-01 -1.25337708e+00 4.50123370e-01
-8.00819874e-01 1.26290366e-01 -6.12056553e-01 1.02405167e+00
3.48894030e-01 9.18209791e-01 1.47973448e-01 8.49226490e-02
2.36831725e-01 7.30568469e-02 5.39312541e-01 -2.34547913e-01
-5.83890498e-01 1.13630903e+00 -1.18360527e-01 -3.36545765e-01
-4.47309822e-01 -7.12903559e-01 -1.40792048e+00 -2.97037005e-01
-5.78565717e-01 1.53495088e-01 6.13119364e-01 5.01611531e-01
-5.93782067e-01 4.37484682e-01 4.36313540e-01 -9.61351335e-01
4.88293804e-02 -6.35020316e-01 -8.81424010e-01 6.02690876e-01
2.38331348e-01 -3.31897080e-01 -4.99733180e-01 7.80341148e-01]
|
[8.991046905517578, -0.856682538986206]
|
7180709d-004b-4e6a-93c4-9a6d8afd1bef
|
multi-granularity-argument-mining-in-legal
|
2210.09472
| null |
https://arxiv.org/abs/2210.09472v2
|
https://arxiv.org/pdf/2210.09472v2.pdf
|
Multi-granularity Argument Mining in Legal Texts
|
In this paper, we explore legal argument mining using multiple levels of granularity. Argument mining has usually been conceptualized as a sentence classification problem. In this work, we conceptualize argument mining as a token-level (i.e., word-level) classification problem. We use a Longformer model to classify the tokens. Results show that token-level text classification identifies certain legal argument elements more accurately than sentence-level text classification. Token-level classification also provides greater flexibility to analyze legal texts and to gain more insight into what the model focuses on when processing a large amount of input data.
|
['Kevin Ashley', 'Huihui Xu']
|
2022-10-17
| null | null | null | null |
['sentence-classification', 'argument-mining']
|
['natural-language-processing', 'natural-language-processing']
|
[ 3.01108569e-01 4.39230800e-01 -1.06178057e+00 -4.32272315e-01
-9.76521611e-01 -5.13914883e-01 7.69226313e-01 1.17576122e+00
-3.67918670e-01 7.59991705e-01 7.21685708e-01 -1.24036849e+00
-2.17561051e-01 -1.14027774e+00 -1.48113370e-01 -6.68845847e-02
8.90290439e-02 2.51500070e-01 9.54785123e-02 -3.35374922e-01
8.73633325e-01 1.01439916e-01 -1.71021879e+00 1.05689442e+00
1.29676270e+00 6.21885717e-01 -4.22608256e-01 3.76282305e-01
-9.52036142e-01 1.45076585e+00 -1.19797373e+00 -6.78720176e-01
-1.13660023e-01 -3.33641708e-01 -1.47277045e+00 -3.94037366e-01
3.62863809e-01 1.47267118e-01 6.36739671e-01 9.91233468e-01
-2.78253332e-02 -4.19844449e-01 6.00254118e-01 -9.67829525e-01
-3.52210969e-01 1.29120779e+00 -5.44429600e-01 6.72968984e-01
7.71327734e-01 -5.15418410e-01 1.51318562e+00 -3.98938417e-01
1.00658464e+00 1.56998980e+00 4.74063724e-01 6.75584152e-02
-1.08762646e+00 -2.60559559e-01 4.47363138e-01 1.91271886e-01
-4.48486418e-01 2.34411974e-02 8.77429128e-01 -6.50890589e-01
1.05078781e+00 3.80758941e-01 7.32475042e-01 8.13434303e-01
4.27771181e-01 1.13395393e+00 1.19626331e+00 -7.06682444e-01
2.00372353e-01 -1.21784821e-01 1.08302760e+00 2.81399071e-01
7.81964600e-01 -4.94916826e-01 -2.31816888e-01 -9.06814754e-01
-3.47920507e-02 6.11190647e-02 4.41771865e-01 4.50376481e-01
-7.18983889e-01 1.51390934e+00 -2.47817427e-01 5.89503050e-01
-4.51334357e-01 2.11312738e-03 1.03871286e+00 6.47756755e-01
7.86073267e-01 4.36524123e-01 -5.62723339e-01 -6.25440896e-01
-7.47399569e-01 6.38493896e-01 1.01925242e+00 4.33988363e-01
2.25882426e-01 -1.94405660e-01 -3.33291888e-01 8.93334687e-01
2.77148336e-01 2.96930492e-01 3.89846265e-01 -5.21500349e-01
1.08599687e+00 1.01423097e+00 -1.93122298e-01 -6.52407646e-01
-3.28164577e-01 -1.65257186e-01 -1.13670960e-01 3.00658673e-01
4.43590134e-01 -1.71272591e-01 -6.61503792e-01 1.24569690e+00
2.07756102e-01 -7.70368516e-01 4.55185056e-01 1.82602599e-01
7.10809886e-01 3.59077841e-01 5.17124712e-01 -2.11847201e-01
1.92537975e+00 -4.56744283e-01 -8.51132989e-01 -1.05824113e-01
1.28140867e+00 -7.73761451e-01 1.06683755e+00 3.62552583e-01
-1.05066133e+00 -1.55604392e-01 -9.54126716e-01 -1.46662757e-01
-6.07949436e-01 -1.36535645e-01 8.65951478e-01 8.44494164e-01
-2.55471140e-01 5.08185506e-01 -2.74189800e-01 -5.66180088e-02
5.45768976e-01 -2.06093401e-01 1.46787047e-01 3.16606432e-01
-1.61601257e+00 8.45926583e-01 2.41473466e-01 -8.00144553e-01
3.02791834e-01 -4.77856845e-01 -1.02027988e+00 -6.17699232e-03
5.82348287e-01 -1.86672136e-01 1.32395387e+00 -5.40248573e-01
-7.03663409e-01 9.87979829e-01 -4.97703373e-01 -7.53088355e-01
3.49488735e-01 -1.75398126e-01 -5.12337983e-01 -1.04503399e-02
7.63669074e-01 -1.86371565e-01 5.69946766e-01 -9.15719688e-01
-9.76756215e-01 -4.69613433e-01 4.96431410e-01 -1.71945274e-01
-2.27686554e-01 6.38366103e-01 5.69505811e-01 -8.62851322e-01
-2.76657585e-02 -4.83934313e-01 -1.94990173e-01 -7.73906767e-01
-4.16770518e-01 -1.15629661e+00 7.93556213e-01 -3.37901443e-01
1.65007186e+00 -1.65264094e+00 -6.09413028e-01 3.94103438e-01
4.63813603e-01 5.26224412e-02 3.96623760e-01 9.64847267e-01
-2.85712957e-01 9.18974638e-01 9.06870142e-02 -8.80854391e-03
-8.15590248e-02 3.37571114e-01 -6.03253365e-01 1.07650176e-01
7.84558058e-02 7.19181180e-01 -8.91910613e-01 -9.30648327e-01
-2.26309761e-01 -1.58105135e-01 -3.75563085e-01 -4.46320891e-01
-3.11440974e-01 -3.86455595e-01 -1.03755331e+00 8.25314820e-01
3.78031790e-01 -1.67713076e-01 3.06145638e-01 1.48996413e-01
-4.96282399e-01 1.07096338e+00 -7.73419559e-01 1.10202825e+00
-2.74213850e-01 6.39522851e-01 -1.83114916e-01 -1.17720628e+00
6.97375953e-01 2.97427088e-01 4.81264323e-01 -9.02290463e-01
1.03900068e-01 3.14289153e-01 3.01536620e-01 -5.69284379e-01
5.43365717e-01 -1.74771041e-01 -7.13917196e-01 9.42106903e-01
-7.93020308e-01 3.15598071e-01 9.51406121e-01 3.68781835e-01
1.16516042e+00 -4.58438128e-01 6.11861408e-01 -6.69961154e-01
4.35210556e-01 6.96889520e-01 5.64274788e-01 8.49807978e-01
2.15404138e-01 -1.63608402e-01 1.10313261e+00 -6.90462768e-01
-8.41114283e-01 -7.30456233e-01 -6.96589172e-01 8.87601733e-01
-2.08106041e-01 -1.06230414e+00 -7.79522002e-01 -1.04426217e+00
4.46504205e-01 5.54763854e-01 -8.92715216e-01 4.75794435e-01
-7.02712536e-01 -7.73224890e-01 6.05722368e-01 5.89782000e-01
2.08883226e-01 -1.12125242e+00 -1.25761783e+00 3.45561773e-01
-2.83100843e-01 -7.78288007e-01 1.19520344e-01 4.17094141e-01
-7.82573879e-01 -1.52893865e+00 1.85738266e-01 -5.60065448e-01
4.61423308e-01 -6.50997609e-02 1.16006279e+00 6.56468511e-01
-4.41199273e-01 -2.13195279e-01 -1.04188585e+00 -9.49110806e-01
-5.69889009e-01 9.32459310e-02 -3.02061737e-01 -5.19880831e-01
5.97424865e-01 2.60457210e-02 -7.92905912e-02 -1.84452906e-01
-1.03305233e+00 -2.35043764e-01 4.45404172e-01 7.95141995e-01
1.85698882e-01 1.42581001e-01 8.71285260e-01 -1.58370364e+00
1.91066551e+00 -5.19261420e-01 -1.96635842e-01 4.13551062e-01
-9.30216551e-01 3.34642828e-01 7.85657048e-01 -2.77219802e-01
-9.13079917e-01 -9.40521240e-01 -4.65796858e-01 6.70924783e-01
-1.64793074e-01 1.18512917e+00 1.87936977e-01 6.68113947e-01
5.26115775e-01 -3.32389444e-01 -1.06720224e-01 -2.81002253e-01
2.84444183e-01 9.50523138e-01 -2.60053009e-01 -1.13063872e+00
5.53335726e-01 4.49282259e-01 -1.48688108e-01 -7.39685774e-01
-1.01385295e+00 -3.14015985e-01 -5.72215974e-01 -2.58521050e-01
8.82026672e-01 -2.95420408e-01 -9.28643167e-01 -1.98560715e-01
-1.12958419e+00 -1.73527718e-01 -5.11960328e-01 2.18986765e-01
-1.43201083e-01 5.15774608e-01 -7.69468188e-01 -1.13298702e+00
-3.70518416e-01 -8.65951300e-01 9.96130943e-01 -9.76432860e-02
-9.63652730e-01 -1.21113431e+00 3.90738249e-03 7.46098340e-01
-1.34058505e-01 5.42168081e-01 1.41947854e+00 -9.46425378e-01
2.86768645e-01 -4.30557728e-01 -6.53710514e-02 -1.69070005e-01
1.66764036e-01 8.10405388e-02 -5.31649411e-01 1.90563545e-01
2.57160366e-01 -3.79703313e-01 7.86000550e-01 1.77590266e-01
1.15111244e+00 -5.25142372e-01 -3.46232355e-01 -2.15464085e-01
1.10133708e+00 4.35034096e-01 5.65706611e-01 1.00943720e+00
4.53018621e-02 9.53807414e-01 1.15391445e+00 4.05925691e-01
3.21205139e-01 2.25265816e-01 6.25481233e-02 -2.02897191e-01
3.55741888e-01 -1.65197492e-01 7.18989819e-02 1.57038674e-01
5.83332889e-02 -2.74119407e-01 -1.14897811e+00 5.03267348e-01
-2.10328150e+00 -1.60051835e+00 -7.22748995e-01 1.41013181e+00
7.94181585e-01 8.39193761e-01 4.16477799e-01 8.02687228e-01
4.03840035e-01 3.35309088e-01 -2.51401305e-01 -1.20552802e+00
-7.78209046e-02 1.61491886e-01 3.72038007e-01 5.03135145e-01
-9.79781568e-01 8.23159993e-01 6.55004978e+00 1.07097709e+00
-6.53382361e-01 -1.70253396e-01 7.70336330e-01 1.47125050e-01
-7.36955881e-01 2.60872215e-01 -7.89500773e-01 6.11659825e-01
6.39475226e-01 -4.31766510e-01 -4.14803416e-01 9.43233848e-01
3.76773953e-01 -4.25622791e-01 -9.41391826e-01 4.46352124e-01
-4.40359041e-02 -1.80039144e+00 4.16842669e-01 6.67584181e-01
3.19498718e-01 -3.58082682e-01 -1.98812246e-01 3.06634754e-01
4.69890833e-01 -9.87826884e-01 9.56322074e-01 -5.62670790e-02
3.01875174e-01 -6.72737718e-01 9.29657221e-01 4.37582642e-01
-1.23906672e+00 -4.69457597e-01 -9.31322798e-02 -7.38518119e-01
3.55452389e-01 8.08684230e-01 -5.09079218e-01 7.28572130e-01
5.83785892e-01 5.29758692e-01 -3.00224751e-01 4.12433237e-01
-3.01168799e-01 7.22848535e-01 -2.08510477e-02 -4.65621084e-01
4.96883333e-01 -1.75842404e-01 1.92158774e-01 1.29155922e+00
-1.54481456e-01 3.01532745e-01 5.07748604e-01 3.58879536e-01
3.94599587e-02 4.72976029e-01 -4.75046068e-01 -4.67300117e-02
5.84683001e-01 1.00594366e+00 -9.99638319e-01 -7.44594038e-01
-5.12491822e-01 1.47121087e-01 3.50050718e-01 -6.65609911e-02
-6.20923877e-01 -4.65439528e-01 8.11464727e-01 3.74668241e-01
-7.58669972e-02 2.28338372e-02 -7.34999239e-01 -1.09284067e+00
2.75698423e-01 -7.93380439e-01 9.51010346e-01 -2.83629358e-01
-1.31782734e+00 5.41742802e-01 2.40620404e-01 -1.03373706e+00
-4.41103160e-01 -8.83812189e-01 -1.04098427e+00 6.46272540e-01
-1.42627132e+00 -7.83416629e-01 3.62385303e-01 2.29301885e-01
7.09330440e-01 1.13386894e-02 7.36732781e-01 6.32011425e-03
-1.78967521e-01 4.46354747e-01 -1.61567688e-01 6.91399097e-01
2.86980927e-01 -1.39541125e+00 4.13721830e-01 5.10558367e-01
1.28457844e-01 1.00389981e+00 6.83892131e-01 -1.15319943e+00
-9.67480242e-01 -4.95337963e-01 1.58160770e+00 -5.48957586e-01
1.03344321e+00 -2.01103285e-01 -6.57133996e-01 2.83268601e-01
3.92661303e-01 -6.87800407e-01 1.28964770e+00 8.26113164e-01
-5.35983324e-01 1.77558705e-01 -1.34250855e+00 4.92960215e-01
9.28851306e-01 -7.47812867e-01 -1.36987579e+00 3.08626980e-01
4.39293921e-01 9.27244201e-02 -1.05022693e+00 2.11780518e-01
6.43239260e-01 -6.08368456e-01 8.77431214e-01 -1.10243237e+00
6.64658368e-01 6.55917451e-02 1.42885387e-01 -8.15261483e-01
6.90371692e-02 -1.11289524e-01 9.38297436e-02 1.18255174e+00
1.00022948e+00 -8.96800816e-01 8.34916651e-01 4.90418613e-01
1.49175562e-02 -1.17380226e+00 -8.47586215e-01 -9.29680765e-01
5.46983600e-01 -7.01734662e-01 7.29095101e-01 1.24014580e+00
1.14827418e+00 5.29835641e-01 1.60138771e-01 -4.79196906e-01
7.63072133e-01 9.42307532e-01 3.87099624e-01 -1.83633721e+00
2.82401204e-01 -8.65326345e-01 -1.93767563e-01 -6.38688624e-01
4.15143192e-01 -1.00050080e+00 -5.54412067e-01 -2.03603816e+00
1.64870158e-01 -3.23698461e-01 2.88570356e-02 4.43295807e-01
-2.35335708e-01 -3.26201826e-01 -4.56836037e-02 3.19373935e-01
-3.82350802e-01 -1.40574962e-01 7.77427673e-01 -2.60176629e-01
2.35791085e-03 6.86672255e-02 -1.01043797e+00 7.40591943e-01
1.20778811e+00 -7.44466305e-01 -4.36928988e-01 -3.15528065e-01
6.55671179e-01 -4.63585146e-02 -8.72108117e-02 -3.84280473e-01
1.77656889e-01 -5.15122533e-01 5.85741363e-02 -9.29215610e-01
-9.58289504e-02 -2.74969965e-01 -6.19319201e-01 6.70218885e-01
-7.98183918e-01 3.55857939e-01 -8.98007601e-02 3.44807357e-01
-5.99471211e-01 -6.78304195e-01 2.36768931e-01 -2.10248977e-01
-1.97153658e-01 -2.02747330e-01 -9.77511525e-01 3.81689280e-01
8.76965404e-01 -5.69158137e-01 -8.67877662e-01 3.92621057e-03
-6.91057444e-01 4.25452173e-01 3.03924054e-01 4.51897115e-01
5.03693521e-01 -1.00107563e+00 -8.24315667e-01 -6.87250569e-02
3.68762612e-01 -4.35234785e-01 -3.65856081e-01 5.09041786e-01
-3.49448085e-01 5.90293527e-01 2.96035498e-01 -1.56226233e-01
-1.78945708e+00 3.17961007e-01 -3.47419292e-01 -8.49943995e-01
-7.38936007e-01 3.86747092e-01 -6.11388206e-01 -8.77746642e-02
-1.14682741e-01 -7.97500014e-01 -8.74750316e-01 7.90990114e-01
5.46512306e-01 1.81965843e-01 -1.16612643e-01 -3.36308330e-01
-4.22693998e-01 4.73743945e-01 -2.02007115e-01 -1.58424735e-01
1.35892475e+00 4.07495677e-01 -5.32843053e-01 5.35893917e-01
9.14275408e-01 5.62074959e-01 -1.72878698e-01 7.38154724e-02
9.85564351e-01 -3.77867341e-01 -2.19788417e-01 -6.18610740e-01
-2.71476299e-01 5.71649671e-01 -2.11210579e-01 1.24174976e+00
3.60967994e-01 4.49031949e-01 6.33666754e-01 5.00982404e-01
1.74275681e-01 -1.73846507e+00 -1.65623590e-01 6.99639916e-01
5.09412646e-01 -8.59088182e-01 2.29134902e-01 -7.11296320e-01
-4.78483379e-01 1.30030274e+00 3.22598815e-01 -1.03851743e-02
7.97435760e-01 6.22435033e-01 2.69671232e-01 -7.52264440e-01
-7.76434541e-01 -3.17556351e-01 -4.96114045e-02 7.47506320e-02
9.84982610e-01 4.79729460e-05 -1.25485635e+00 6.23665094e-01
-5.55012345e-01 -4.04320627e-01 6.54143035e-01 1.46304572e+00
-8.09461892e-01 -1.85701072e+00 -5.33648610e-01 1.11385643e+00
-1.04829371e+00 -4.09258932e-01 -8.47464383e-01 8.62507939e-01
6.43995404e-02 1.52532911e+00 6.75042346e-02 -3.56609337e-02
3.62691849e-01 2.17978448e-01 4.54314500e-02 -9.20074344e-01
-9.78984535e-01 -2.65959501e-01 1.01397932e+00 -2.37300366e-01
-5.78969657e-01 -1.13943553e+00 -1.68225300e+00 -3.05513233e-01
-2.20659167e-01 1.22228050e+00 5.85651457e-01 1.31105638e+00
-5.05384766e-02 5.13083220e-01 2.41815686e-01 2.41586074e-01
-5.27364433e-01 -8.00098360e-01 -3.51369858e-01 5.58563232e-01
2.02177599e-01 -5.71837127e-01 -2.26337776e-01 -2.48466805e-01]
|
[9.521957397460938, 9.590855598449707]
|
5ac37bf0-fba6-4516-ab77-487d9ec774d1
|
automerge-a-framework-for-map-assembling-and
|
2207.06965
| null |
https://arxiv.org/abs/2207.06965v4
|
https://arxiv.org/pdf/2207.06965v4.pdf
|
AutoMerge: A Framework for Map Assembling and Smoothing in City-scale Environments
|
We present AutoMerge, a LiDAR data processing framework for assembling a large number of map segments into a complete map. Traditional large-scale map merging methods are fragile to incorrect data associations, and are primarily limited to working only offline. AutoMerge utilizes multi-perspective fusion and adaptive loop closure detection for accurate data associations, and it uses incremental merging to assemble large maps from individual trajectory segments given in random order and with no initial estimations. Furthermore, after assembling the segments, AutoMerge performs fine matching and pose-graph optimization to globally smooth the merged map. We demonstrate AutoMerge on both city-scale merging (120km) and campus-scale repeated merging (4.5km x 8). The experiments show that AutoMerge (i) surpasses the second- and third- best methods by 14% and 24% recall in segment retrieval, (ii) achieves comparable 3D mapping accuracy for 120 km large-scale map assembly, (iii) and it is robust to temporally-spaced revisits. To the best of our knowledge, AutoMerge is the first mapping approach that can merge hundreds of kilometers of individual segments without the aid of GPS.
|
['Sebastian Scherer', 'Howie Choset', 'Ji Zhang', 'Ruohai Ge', 'Shiqi Zhao', 'Haowen Lai', 'Peng Yin']
|
2022-07-14
| null | null | null | null |
['loop-closure-detection']
|
['computer-vision']
|
[-3.13285947e-01 -2.06829086e-02 -1.34292468e-01 -3.93736690e-01
-1.47342205e+00 -9.44484234e-01 6.30681038e-01 7.92195857e-01
-3.57353657e-01 8.78431201e-01 -9.87216830e-02 -6.28747344e-01
-3.64275992e-01 -1.29177225e+00 -8.87476444e-01 1.42415032e-01
-4.26137686e-01 1.15951633e+00 9.84990001e-01 -3.35754991e-01
3.55092108e-01 1.03447759e+00 -1.38989723e+00 -4.70922321e-01
1.37769437e+00 6.57062769e-01 3.58584911e-01 6.58712506e-01
-2.25967646e-01 2.31884811e-02 -2.26752430e-01 -2.85653353e-01
4.40719664e-01 3.43108386e-01 -8.72047424e-01 -2.12511390e-01
1.06520474e+00 -5.17052174e-01 -3.19669187e-01 7.33111620e-01
4.37364697e-01 2.00791031e-01 3.22987974e-01 -1.35876381e+00
-3.04141920e-02 2.93296933e-01 -6.96627855e-01 1.85222358e-01
3.23464483e-01 -2.32206702e-01 6.27167463e-01 -1.14625549e+00
6.17216349e-01 9.42999542e-01 1.19185829e+00 -6.27317548e-01
-1.54352844e+00 -9.16778147e-01 1.60782665e-01 -9.61649120e-02
-2.36663723e+00 -4.45158273e-01 -5.73292002e-03 -3.85016531e-01
1.13311172e+00 6.00309491e-01 6.94037080e-01 -2.22269595e-01
3.07620484e-02 1.93933457e-01 6.98235571e-01 -1.55555243e-02
-3.99893112e-02 -1.65211037e-01 2.46921733e-01 6.05542481e-01
4.09884810e-01 -3.34710553e-02 -2.00550184e-01 -3.03362250e-01
7.64873564e-01 -1.75534621e-01 7.90412948e-02 -3.22624892e-01
-1.45792031e+00 5.42236805e-01 8.18945289e-01 -2.60368735e-01
-3.49126905e-01 2.44608954e-01 -6.66714134e-03 -5.29157929e-02
3.72153461e-01 2.56666422e-01 -2.38785706e-02 -1.44110978e-01
-1.45316350e+00 5.90761602e-01 4.30091470e-01 1.76535988e+00
1.49285495e+00 -1.55232221e-01 4.36771005e-01 4.54645663e-01
-4.07985635e-02 9.31718886e-01 -3.25156063e-01 -1.05083001e+00
9.88834858e-01 6.31228745e-01 3.51845086e-01 -1.15225971e+00
-7.53362060e-01 -1.30119681e-01 -5.86159289e-01 1.27802745e-01
9.23032463e-02 9.47131813e-02 -8.88854206e-01 1.12750769e+00
4.57376570e-01 4.42554384e-01 -3.04318517e-01 5.23475230e-01
7.16108561e-01 5.59730828e-01 3.74946110e-02 1.40499115e-01
9.43967581e-01 -6.51281059e-01 -3.83076906e-01 -4.86405224e-01
5.65710425e-01 -9.17241096e-01 4.86662209e-01 3.65140586e-04
-1.16543293e+00 -6.84632063e-01 -1.16835845e+00 -2.54023671e-01
-6.63089335e-01 -1.14484884e-01 4.11144853e-01 4.34654772e-01
-1.42038476e+00 4.49529409e-01 -7.83856690e-01 -5.13148308e-01
2.97039956e-01 5.58276296e-01 -5.25609791e-01 -9.12705585e-02
-8.25400889e-01 1.05480242e+00 5.14746249e-01 6.09153584e-02
-2.52739221e-01 -8.94690156e-01 -8.75682116e-01 -2.04180598e-01
2.14467630e-01 -6.33256435e-01 9.66503918e-01 2.71702260e-01
-5.45129657e-01 6.46717787e-01 -5.83771288e-01 -4.02272463e-01
5.02036631e-01 -3.63458663e-01 -6.28205121e-01 -2.43557408e-03
8.19220960e-01 1.10877347e+00 1.26277044e-01 -1.30275166e+00
-1.18465829e+00 -4.21088338e-01 -3.75416994e-01 3.24537426e-01
4.05388772e-01 -3.04265678e-01 -7.56440103e-01 -3.02227646e-01
7.66848266e-01 -1.02938879e+00 -5.82719922e-01 -3.27216268e-01
-2.83159226e-01 2.61958867e-01 9.07225132e-01 -7.25317895e-01
1.59763992e+00 -1.80951786e+00 -3.54171872e-01 7.06449628e-01
1.58542186e-01 4.92729172e-02 7.29749128e-02 8.99112105e-01
2.84934759e-01 3.28649670e-01 -4.46423560e-01 -3.32544297e-01
-1.66577071e-01 2.66058505e-01 -5.60248673e-01 4.37633127e-01
-1.80514067e-01 1.14763093e+00 -9.35941577e-01 -6.61613047e-01
5.74289382e-01 1.64866354e-02 -1.09376170e-01 -2.86861837e-01
3.27602774e-01 7.94050619e-02 -3.68923128e-01 8.24120760e-01
1.31223392e+00 3.00806195e-01 -1.38029456e-01 -1.30340382e-01
-8.76029313e-01 1.74948260e-01 -1.53506422e+00 1.78502262e+00
-1.29039899e-01 7.46608913e-01 8.95983651e-02 -3.21446121e-01
1.18376398e+00 -1.41386837e-01 4.98769730e-01 -6.06291234e-01
-6.99345052e-01 6.39082074e-01 -7.65303910e-01 3.26128639e-02
1.33642066e+00 2.10151792e-01 -4.83637989e-01 2.90466964e-01
-4.49241251e-01 -9.21289086e-01 2.23718092e-01 5.11237562e-01
9.61047411e-01 -3.49605046e-02 3.94362211e-01 -3.37328851e-01
1.46162271e-01 7.30389118e-01 2.64873296e-01 8.01752746e-01
2.97936294e-02 7.97813594e-01 -1.99612886e-01 -5.87148368e-01
-1.06325257e+00 -1.61462271e+00 -4.46015447e-01 5.23946166e-01
6.79156303e-01 -6.09163463e-01 -2.45414957e-01 -2.76836008e-01
6.75791085e-01 7.12354481e-01 -5.02857491e-02 4.75900739e-01
-7.93732345e-01 -5.67077518e-01 7.98584759e-01 6.86013818e-01
6.55603349e-01 -2.12916717e-01 -1.70774400e-01 5.50051868e-01
-4.13398147e-01 -1.12695408e+00 -4.33534384e-01 4.39748503e-02
-1.16233933e+00 -9.69049096e-01 -3.97991873e-02 -7.55591869e-01
7.11918950e-01 8.76129866e-01 1.07634902e+00 -4.12410721e-02
4.53671552e-02 1.12620912e-01 1.82329286e-02 -2.92008549e-01
8.82573202e-02 4.28546131e-01 2.18848456e-02 -8.10415089e-01
2.24386826e-01 -8.97426546e-01 -3.54248792e-01 8.26447845e-01
-4.07719940e-01 3.42874229e-02 5.37995517e-01 4.11707759e-02
1.01645839e+00 1.09507881e-01 6.37924433e-01 -4.82003778e-01
4.13812876e-01 -5.54684401e-01 -9.47765291e-01 1.75060898e-01
-7.63041496e-01 -3.93151999e-01 3.17070037e-02 2.17400953e-01
-6.24359965e-01 4.80116963e-01 -2.21889704e-01 6.05221745e-03
-2.47616798e-01 5.69505870e-01 3.57694104e-02 -3.84841949e-01
6.28504455e-01 1.13081321e-01 -2.78344452e-01 -3.36645871e-01
7.49768138e-01 6.63987100e-01 1.15068316e+00 -4.65450317e-01
1.16241300e+00 6.99960470e-01 5.60875423e-02 -8.28624725e-01
-1.92143589e-01 -9.44844306e-01 -1.39424324e+00 -2.50546366e-01
6.24858379e-01 -1.21650231e+00 -3.73331428e-01 2.64248341e-01
-1.00680673e+00 -1.97086006e-01 -1.94115326e-01 4.38582957e-01
-4.83210474e-01 2.75797576e-01 -1.41827404e-01 -5.53309143e-01
-1.66473597e-01 -7.84577668e-01 1.43133557e+00 6.99257106e-02
-3.02708685e-01 -5.03833055e-01 3.02006841e-01 8.46701637e-02
3.19930375e-01 4.10163999e-01 3.68352026e-01 -8.32280815e-02
-1.19052041e+00 -6.13147020e-01 -5.21399379e-01 -6.57809377e-01
1.26081243e-01 3.25059555e-02 -7.29463041e-01 -1.26415119e-01
-8.34761679e-01 3.04372877e-01 7.04154372e-01 4.46905375e-01
5.62086284e-01 1.37532309e-01 -9.50618982e-01 6.09268069e-01
1.52772701e+00 1.30789682e-01 9.10711348e-01 6.30148590e-01
6.35775447e-01 4.58536476e-01 1.08713126e+00 1.72318101e-01
1.12667203e+00 8.83929133e-01 6.05661809e-01 -2.94239759e-01
1.47946388e-01 -4.59907413e-01 -1.80484712e-01 7.84939110e-01
-2.02946588e-01 9.37776640e-02 -1.31495881e+00 7.68935978e-01
-2.16050100e+00 -1.02069104e+00 -8.39785695e-01 2.51289344e+00
2.55224496e-01 1.73470184e-01 2.75123060e-01 -3.84312510e-01
7.82174647e-01 1.35276109e-01 -3.14775199e-01 -5.22225350e-02
-2.37299427e-02 1.20809957e-01 1.40954638e+00 9.38978910e-01
-1.30538201e+00 1.28597236e+00 6.95229387e+00 7.19650805e-01
-6.77984476e-01 8.12924430e-02 -2.46314076e-03 6.43971562e-02
-3.60798746e-01 3.59076560e-01 -1.18675280e+00 2.08808258e-01
8.49958301e-01 -1.82987049e-01 1.25272483e-01 7.10135639e-01
2.74796516e-01 -5.93289614e-01 -5.46260774e-01 6.87412977e-01
-3.77049267e-01 -1.65999734e+00 -1.71758190e-01 3.43630135e-01
8.45238686e-01 6.13266766e-01 -5.15586853e-01 3.00461322e-01
5.22713959e-01 -9.02696669e-01 1.02617335e+00 5.67458391e-01
8.10810208e-01 -1.12755585e+00 5.63026965e-01 5.90788782e-01
-2.17826509e+00 2.55174011e-01 -3.61969531e-01 4.21399921e-02
9.15583849e-01 5.14346480e-01 -9.94694591e-01 1.03194368e+00
6.77319050e-01 5.69091201e-01 -8.56078565e-01 1.40017283e+00
2.08874373e-03 7.70790428e-02 -8.17625284e-01 5.13033569e-01
3.99834722e-01 -5.10133862e-01 5.58670163e-01 1.26508784e+00
7.16663837e-01 1.03425868e-01 6.50262892e-01 4.47902888e-01
3.22038800e-01 -1.20177820e-01 -8.99396539e-01 5.83018959e-01
1.18154800e+00 1.04001558e+00 -8.01586092e-01 -4.20628637e-01
-1.24956600e-01 5.66074908e-01 3.86715651e-01 2.19019696e-01
-8.60058904e-01 -5.65197706e-01 6.27206326e-01 5.63488007e-01
3.28927159e-01 -8.88759494e-01 -4.59631413e-01 -5.04957080e-01
-6.69766078e-03 -1.66298281e-02 2.12865487e-01 -8.30876470e-01
-5.07078230e-01 4.59217787e-01 3.97694707e-01 -1.50635767e+00
-1.85726389e-01 2.27254257e-01 -6.06616914e-01 9.28659976e-01
-1.54073787e+00 -1.47689033e+00 -6.58867717e-01 2.70386815e-01
3.41529638e-01 3.86239916e-01 8.19156528e-01 5.74772537e-01
-2.03645527e-01 5.94072044e-02 2.36092612e-01 -2.22507358e-01
6.64364576e-01 -1.30185485e+00 1.21821642e+00 1.00370955e+00
-4.46981490e-02 5.45597017e-01 3.93890768e-01 -1.27608943e+00
-1.01010597e+00 -1.50856256e+00 1.15380657e+00 -5.07678509e-01
6.33302093e-01 -3.14409494e-01 -8.62532675e-01 1.10772061e+00
-1.58698723e-01 -1.33208409e-01 2.07420647e-01 1.13079093e-01
1.09256938e-01 -3.61949414e-01 -1.02114773e+00 5.44670105e-01
1.31038117e+00 -4.14637744e-01 -3.77228588e-01 1.49142697e-01
5.71817875e-01 -6.92112029e-01 -1.23751795e+00 6.17328942e-01
6.95471704e-01 -8.23480189e-01 1.48946953e+00 1.14550605e-01
-3.71918052e-01 -8.22779298e-01 -3.23328167e-01 -1.16100359e+00
-5.15890062e-01 -4.22156423e-01 3.78605694e-01 1.22361004e+00
5.03930449e-01 -7.56131768e-01 7.32938349e-01 5.55658400e-01
-6.96379662e-01 -3.67633820e-01 -1.11008155e+00 -9.33074534e-01
-1.57448724e-01 -7.46993184e-01 1.17945194e+00 8.31126392e-01
-1.66013315e-01 -3.79197970e-02 -1.41591921e-01 8.68955851e-01
7.13420391e-01 3.70675772e-01 1.36754596e+00 -1.47721112e+00
7.09321499e-01 -4.87591922e-01 -6.26158059e-01 -1.15657365e+00
-2.45782435e-01 -8.67318869e-01 9.43622738e-02 -2.03964758e+00
-3.95566851e-01 -1.10263062e+00 3.96569461e-01 3.59730870e-01
-6.80341497e-02 3.43399853e-01 -1.20364010e-01 5.46027422e-01
-4.87725884e-01 1.68295443e-01 6.27847075e-01 -7.02950433e-02
-5.21459639e-01 1.29152894e-01 -4.61109519e-01 4.54607934e-01
8.72128010e-01 -4.33748215e-01 -3.22684705e-01 -4.64721978e-01
3.02477121e-01 -3.94100808e-02 2.76100785e-01 -1.28451681e+00
5.88940382e-01 -2.63991147e-01 2.07402661e-01 -1.81198144e+00
4.23747241e-01 -8.22809279e-01 9.07887876e-01 1.98927611e-01
4.54149485e-01 5.91876924e-01 6.18668318e-01 6.20812893e-01
-2.23598257e-01 3.08155757e-03 3.34964544e-01 3.90822142e-02
-1.20674574e+00 3.33695561e-01 -3.53769541e-01 -2.36348137e-01
1.22792661e+00 -6.73584521e-01 -4.85238194e-01 -2.65237361e-01
-6.82075918e-01 8.75959337e-01 7.20764458e-01 1.89053923e-01
6.49523675e-01 -1.63006222e+00 -7.05201387e-01 1.01675712e-01
2.90832639e-01 5.58844984e-01 5.44919483e-02 9.24273014e-01
-1.09454024e+00 5.23422241e-01 -5.32289483e-02 -8.99307787e-01
-1.28518021e+00 2.27315605e-01 -4.47903648e-02 -9.63223353e-02
-7.86124885e-01 5.07672548e-01 -2.68732280e-01 -9.09085691e-01
-1.18167609e-01 -4.01701719e-01 2.58086562e-01 3.49244982e-01
3.23852777e-01 8.87909591e-01 3.61410081e-01 -1.00978804e+00
-6.40136480e-01 9.77691591e-01 3.22356403e-01 -3.16782147e-01
1.23393953e+00 -6.19030535e-01 -7.58116134e-03 1.49241179e-01
8.01768959e-01 4.70805794e-01 -8.78070772e-01 1.52754858e-02
2.05022618e-01 -7.10637867e-01 -4.91135493e-02 -3.04400921e-01
-4.94954526e-01 4.76076335e-01 3.42155010e-01 8.78435522e-02
6.72580600e-01 5.85806966e-02 1.08343077e+00 3.93036604e-01
8.94114494e-01 -1.04919696e+00 -6.56376600e-01 5.36311984e-01
7.86688864e-01 -1.08513856e+00 5.36091328e-01 -6.48471534e-01
-3.32904279e-01 9.83762205e-01 4.26544249e-01 1.59691658e-03
6.58554494e-01 3.68700564e-01 -1.38406366e-01 -3.68159205e-01
-2.52132993e-02 -5.36672711e-01 1.73437372e-01 6.10256791e-01
-2.80685455e-01 4.24366891e-01 9.34488997e-02 -1.19681574e-01
-5.49610913e-01 -1.69168994e-01 2.85717666e-01 1.05319750e+00
-1.11167860e+00 -9.82135534e-01 -7.96248496e-01 5.28987885e-01
5.33852220e-01 2.89293677e-02 -2.49937251e-01 1.28061163e+00
1.02560125e-01 7.22972870e-01 3.65875840e-01 -4.76256430e-01
7.02943385e-01 -1.45007864e-01 1.00711502e-01 -4.70306665e-01
-3.44737947e-01 3.93805914e-02 3.87727827e-01 -6.34067059e-01
-2.05451641e-02 -8.45552504e-01 -1.48962200e+00 -7.67043173e-01
-4.76910412e-01 1.67599440e-01 7.85828233e-01 5.41864455e-01
7.02175975e-01 -7.67347515e-02 4.67208326e-01 -1.20466375e+00
1.41954795e-01 -6.20241344e-01 -5.76900303e-01 -2.58710206e-01
2.38045096e-01 -6.73192322e-01 8.46931562e-02 -3.18639159e-01]
|
[7.805464744567871, -2.603060722351074]
|
be5b3c62-9fe5-43a2-b230-8e3eefecf8ca
|
intelligent-systems-for-information-security
|
1401.3592
| null |
http://arxiv.org/abs/1401.3592v1
|
http://arxiv.org/pdf/1401.3592v1.pdf
|
Intelligent Systems for Information Security
|
This thesis aims to use intelligent systems to extend and improve performance
and security of cryptographic techniques. Genetic algorithms framework for
cryptanalysis problem is addressed. A novel extension to the differential
cryptanalysis using genetic algorithm is proposed and a fitness measure based
on the differential characteristics of the cipher being attacked is also
proposed. The complexity of the proposed attack is shown to be less than
quarter of normal differential cryptanalysis of the same cipher by applying the
proposed attack to both the basic Substitution Permutation Network and the
Feistel Network. The basic models of modern block ciphers are attacked instead
of actual cipher to prove that the attack is applicable to other ciphers
vulnerable to differential cryptanalysis. A new attack for block cipher based
on the ability of neural networks to perform an approximation of mapping is
proposed. A complete problem formulation is explained and implementation of the
attack on some hypothetical Feistel cipher not vulnerable to differential or
linear attacks is presented. A new block cipher based on the neural networks is
proposed. A complete cipher structure is given and a key scheduling is also
shown. The main properties of neural network being able to perform mapping
between large dimension domains in a very fast and a very small memory compared
to S-Boxes is used as a base for the cipher.
|
['Ayman M. Bahaa-Eldin']
|
2014-01-15
| null | null | null | null |
['cryptanalysis']
|
['miscellaneous']
|
[ 6.50106728e-01 1.37011945e-01 3.96300107e-01 -2.90184468e-01
2.44544640e-01 -8.10290456e-01 5.09064019e-01 3.97017956e-01
-6.53081834e-01 9.36043262e-01 -5.19756258e-01 -8.51403296e-01
-4.11777049e-01 -1.17408276e+00 -6.21748149e-01 -9.02662754e-01
-3.79195362e-01 4.04295325e-01 1.22082224e-02 -8.72861564e-01
6.30460203e-01 7.90865898e-01 -1.68769801e+00 3.90902251e-01
5.60653508e-01 4.81650054e-01 8.24817643e-02 1.17942774e+00
1.71136573e-01 3.27042192e-02 -8.72872055e-01 -1.71617776e-01
8.57735395e-01 -5.52529454e-01 -7.21576989e-01 -5.47898471e-01
-2.23079190e-01 1.62729993e-01 -1.79746971e-01 1.25037479e+00
3.93474519e-01 -1.82210132e-01 6.83545530e-01 -1.40428019e+00
3.27083796e-01 9.58844900e-01 -1.30562440e-01 1.38156563e-01
2.71806031e-01 -1.60989121e-01 4.34499472e-01 -7.37928599e-02
4.06694740e-01 1.07239795e+00 6.20870233e-01 3.92009884e-01
-7.61924624e-01 -1.10816288e+00 -4.23829585e-01 3.52417022e-01
-1.35334563e+00 3.28950435e-01 4.06107515e-01 3.36443841e-01
1.21599841e+00 5.44758260e-01 6.46727145e-01 2.37531230e-01
9.23756838e-01 -3.55252298e-03 1.38826168e+00 -7.66172588e-01
2.32694462e-01 1.68890148e-01 1.27860874e-01 5.00031710e-01
9.73432541e-01 4.57954377e-01 1.52417749e-01 -1.40936464e-01
-2.24315271e-01 -2.40745470e-01 2.96942461e-02 7.91059285e-02
-8.02321136e-01 8.73159289e-01 4.42022644e-02 6.68612421e-01
-1.77091852e-01 1.15161046e-01 6.93405271e-01 7.09635794e-01
-4.27735716e-01 6.73448443e-01 -6.27917945e-01 -2.30181336e-01
-5.75786591e-01 4.27227825e-01 1.31760430e+00 6.03636563e-01
5.81078351e-01 3.11372519e-01 7.65475214e-01 -2.43024856e-01
1.91585645e-01 5.30651748e-01 5.52770793e-01 4.52406481e-02
2.93650597e-01 7.70950377e-01 -5.27969420e-01 -7.85495400e-01
-6.95171356e-01 -5.02861738e-01 -5.83176672e-01 9.84113872e-01
3.49131167e-01 -5.74651480e-01 -7.91984081e-01 1.40321565e+00
4.70823854e-01 1.53879225e-01 8.97545636e-01 2.58293264e-02
2.96738952e-01 8.18766415e-01 -3.89730126e-01 4.91002575e-03
1.44493473e+00 -3.40985358e-01 -6.27818823e-01 2.80832857e-01
1.10890424e+00 -9.56281960e-01 2.38905530e-02 8.05113792e-01
-1.14471018e+00 -4.83430922e-01 -2.19865847e+00 6.44421577e-01
-8.91048551e-01 -4.11442161e-01 4.84267205e-01 1.80453920e+00
-8.16441059e-01 7.68258572e-01 -5.35918355e-01 -1.42499253e-01
-1.69021953e-02 1.40572989e+00 -3.95950556e-01 1.53900623e-01
-1.48429513e+00 9.95057642e-01 1.26321864e+00 1.11331604e-01
-4.21997994e-01 -5.66900730e-01 -6.86202109e-01 -1.39295282e-02
-1.71139836e-01 -2.65474111e-01 4.45498347e-01 -1.16755891e+00
-1.32263529e+00 6.42632425e-01 4.40519512e-01 -1.19664156e+00
3.89631718e-01 6.32617712e-01 -6.56376004e-01 7.67982826e-02
-5.77106416e-01 3.71352285e-01 5.02000093e-01 -6.55602753e-01
-1.03871739e+00 -2.75071800e-01 -2.71222237e-02 2.20752746e-01
1.62989926e-02 -2.68645763e-01 5.80585063e-01 -4.82824177e-01
1.33398488e-01 -1.29739308e+00 -6.93183690e-02 -8.69823277e-01
-8.06253478e-02 2.96065390e-01 1.30821204e+00 -3.28909695e-01
1.13950217e+00 -1.76815259e+00 -2.78636336e-01 1.10375047e+00
-4.35164034e-01 8.03462505e-01 8.38937387e-02 9.69230115e-01
-5.54615796e-01 -4.70911860e-02 -4.10842299e-01 6.43221200e-01
5.40040135e-02 1.76727757e-01 -2.76691671e-02 9.25026417e-01
-2.06007943e-01 2.30741471e-01 -3.36656362e-01 1.38686821e-01
-1.06430985e-01 5.50609648e-01 -2.67438024e-01 -3.28787506e-01
-8.24156310e-03 -1.84409857e-01 -1.45299673e-01 3.84249091e-01
1.37092614e+00 8.69313300e-01 5.19264162e-01 -1.09732494e-01
-8.71955752e-02 1.20987780e-01 -1.62207508e+00 1.00370193e+00
-3.55642676e-01 6.78190291e-01 -1.78024516e-01 -1.44690144e+00
1.41094232e+00 4.02864933e-01 -3.05677727e-02 -4.24031526e-01
6.08444989e-01 5.33211946e-01 1.01061571e+00 1.37791768e-01
3.10106397e-01 -2.05538154e-01 -6.51828712e-04 9.19471085e-01
-4.22545791e-01 -2.42952034e-01 4.93180901e-01 -1.80313542e-01
9.83426571e-01 1.69993117e-01 4.51950550e-01 -8.48144531e-01
1.37256646e+00 2.85515804e-02 2.63971001e-01 4.84177917e-01
1.70952156e-01 -3.92345965e-01 4.93779570e-01 -4.99829382e-01
-1.04871678e+00 -5.90861738e-01 -2.64572203e-01 2.82758474e-01
3.14356893e-01 -2.07941264e-01 -9.22323048e-01 -6.50477886e-01
-6.18918333e-04 8.24004650e-01 -6.27263069e-01 -5.34956694e-01
-7.14759886e-01 -1.13142538e+00 1.29156065e+00 -2.25816935e-01
1.04831505e+00 -7.31093407e-01 -8.29830647e-01 1.62670180e-01
8.68660927e-01 -8.46170843e-01 6.27535343e-01 6.60861075e-01
-1.06546342e+00 -1.27484357e+00 1.64647758e-01 -1.09923732e+00
7.67226279e-01 -1.87106937e-01 6.18340313e-01 6.12468958e-01
-4.79113162e-01 -1.39591843e-01 -2.76942313e-01 -1.14686060e+00
-1.43280888e+00 -7.03189373e-02 7.61555284e-02 -4.65645790e-01
7.66317427e-01 -5.53785264e-01 -4.05290216e-01 2.80161232e-01
-1.15589261e+00 -3.65391195e-01 6.63770974e-01 7.51559556e-01
-5.76440878e-02 1.15210450e+00 3.02161366e-01 -8.74906659e-01
7.18260586e-01 -1.47864074e-01 -1.05163455e+00 1.72769666e-01
-9.61416364e-01 5.54737389e-01 8.92974973e-01 -2.02358350e-01
-8.59000802e-01 3.47897075e-02 -2.08262593e-01 7.15091467e-01
-7.17573687e-02 1.73385993e-01 -3.75405312e-01 -9.38911915e-01
6.65095091e-01 3.33632857e-01 4.30978566e-01 1.70939475e-01
-2.95940667e-01 4.90546972e-01 1.92758679e-01 -3.30737531e-01
1.25327110e+00 3.92526627e-01 1.24532139e+00 -7.94989109e-01
5.37150919e-01 3.33661214e-02 -2.41239652e-01 1.57040074e-01
2.87603587e-01 -3.50445330e-01 -1.21297991e+00 5.75335741e-01
-8.40305984e-01 1.82477772e-01 2.19138056e-01 6.74854100e-01
-3.32005501e-01 5.17108858e-01 -1.16593331e-01 -8.38828623e-01
-6.72479272e-01 -1.42727578e+00 -1.44293625e-02 1.31447449e-01
7.47905523e-02 -1.20049751e+00 -1.64193168e-01 -1.85841098e-01
8.22520554e-02 5.14385819e-01 1.09648514e+00 -1.36889482e+00
-2.43317634e-01 -8.72930110e-01 3.43667209e-01 6.00214720e-01
1.79332063e-01 7.76956081e-02 -5.41821301e-01 -5.49584448e-01
5.43136895e-01 7.91785493e-02 3.81056279e-01 -5.71875907e-02
2.85991549e-01 3.58354561e-02 -3.40495199e-01 8.44437540e-01
2.01136518e+00 1.18325472e+00 9.50457573e-01 1.09865189e+00
1.68514363e-02 5.22367656e-01 8.31497312e-01 2.05743089e-01
-3.42681557e-01 1.80425316e-01 6.15798414e-01 2.57778704e-01
4.52393770e-01 4.04126763e-01 3.53703439e-01 3.00933093e-01
1.40476480e-01 -3.33216101e-01 -9.39618945e-01 -1.09704874e-01
-1.00392485e+00 -7.85145342e-01 -5.12041330e-01 2.24499822e+00
4.75340784e-01 5.72507441e-01 -1.98323503e-01 1.19187987e+00
9.53154683e-01 -5.07004499e-01 1.03091568e-01 -1.57614756e+00
-4.21157211e-01 1.06832778e+00 1.26617754e+00 8.45445275e-01
-9.43864107e-01 5.59069037e-01 5.11651993e+00 8.25088322e-01
-1.14318705e+00 -3.88517827e-01 1.89260185e-01 2.54264176e-01
1.71403900e-01 3.69480312e-01 -1.07272458e+00 4.73912627e-01
1.21671569e+00 -3.08200955e-01 2.19346687e-01 3.79016846e-01
-2.14060649e-01 -4.20706272e-01 -8.44918847e-01 6.18656635e-01
7.57773519e-02 -1.06663156e+00 2.66165793e-01 3.70838881e-01
6.65532887e-01 -5.64686000e-01 9.07305554e-02 -4.44870479e-02
-1.27362490e-01 -9.85143304e-01 1.03869036e-01 -1.44209594e-01
1.62834018e-01 -1.61451805e+00 1.13853776e+00 1.49157241e-01
-9.94593680e-01 -1.83909804e-01 -5.06264329e-01 -3.86979371e-01
-2.98891336e-01 -2.27641508e-01 -1.38311458e+00 8.32420766e-01
-3.50151490e-03 -7.40023255e-02 -3.94164979e-01 9.99686241e-01
1.25893936e-01 3.32887590e-01 -7.12146938e-01 -4.72789288e-01
7.90652752e-01 -3.02779227e-01 6.08637333e-01 1.30472052e+00
6.33945405e-01 -1.94838107e-01 -7.00970769e-01 1.69019207e-01
3.68163347e-01 2.86996067e-01 -7.35694706e-01 -9.08760801e-02
4.93835330e-01 8.96727979e-01 -1.05314338e+00 -2.62161642e-01
-1.08424537e-01 6.86098635e-01 -7.69142270e-01 -2.64332592e-01
-5.05884230e-01 -1.32640624e+00 6.17275357e-01 -7.59639964e-02
3.62798154e-01 -4.72833402e-02 -1.77363262e-01 -2.96312124e-01
-4.72166300e-01 -1.35666251e+00 3.90565932e-01 6.80145472e-02
-1.86309174e-01 6.45594358e-01 2.08012924e-01 -1.23690236e+00
-3.73811722e-01 -9.37798381e-01 -6.93647146e-01 1.17114532e+00
-1.11200392e+00 -7.98206627e-01 -4.59703244e-02 6.69533491e-01
6.26966506e-02 -1.10354722e+00 9.64406073e-01 -1.43705875e-01
-9.41506997e-02 7.57645547e-01 3.90540868e-01 -3.05381566e-01
3.28497618e-01 -9.80595946e-01 5.06772578e-01 1.18661320e+00
-2.05696017e-01 7.87975490e-01 1.17998207e+00 -8.44932258e-01
-1.75807965e+00 -4.45533752e-01 6.66782260e-01 3.67130786e-01
3.72540951e-01 -3.31302255e-01 -4.45505708e-01 2.70927161e-01
5.22319973e-01 -5.92803895e-01 6.02449119e-01 -6.54873729e-01
7.86411762e-02 -3.59752685e-01 -1.67555201e+00 3.14270347e-01
2.26660788e-01 -8.39754120e-02 -4.99319047e-01 4.25733268e-01
-1.88603461e-01 -4.99881864e-01 -4.90924299e-01 2.05492437e-01
6.42772794e-01 -1.32149744e+00 8.96986485e-01 -3.15988690e-01
-3.81803155e-01 -4.63782459e-01 4.78416570e-02 -6.86488867e-01
3.40987861e-01 -1.16003466e+00 4.15216029e-01 5.94740391e-01
4.04911697e-01 -1.07730234e+00 1.13686728e+00 -7.73570016e-02
1.06323712e-01 -5.26501536e-01 -9.46785271e-01 -9.76852059e-01
1.99158281e-01 -2.29432210e-01 9.03194070e-01 6.06080055e-01
-2.51483414e-02 -1.27960816e-01 1.40327409e-01 3.95513088e-01
6.34394407e-01 -3.46752286e-01 7.63859153e-01 -1.18266106e+00
-4.54553246e-01 -2.54606903e-01 -1.28589058e+00 4.87079173e-02
2.16706336e-01 -1.01576591e+00 -5.94018459e-01 -7.42476106e-01
-6.74783707e-01 -4.34426367e-01 -2.85152823e-01 -2.46656001e-01
3.61411482e-01 4.47961420e-01 -1.21829286e-01 -5.00620246e-01
6.71040297e-01 -4.29070115e-01 6.95640624e-01 -3.01584210e-02
-1.56668246e-01 3.64646763e-01 -3.39728534e-01 5.93933940e-01
1.05320489e+00 -7.20356286e-01 -8.50037992e-01 4.20248032e-01
6.82977378e-01 -2.02939451e-01 -3.76807153e-01 -1.48537767e+00
1.74871936e-01 2.98755467e-01 2.29856566e-01 -6.56733572e-01
8.77564996e-02 -1.22072589e+00 4.47673976e-01 1.58599651e+00
1.60324126e-01 8.20972741e-01 4.89183486e-01 3.66734535e-01
-1.24855295e-01 -9.39826846e-01 8.52341950e-01 -3.19578722e-02
-5.57306528e-01 -1.93567440e-01 -5.42636395e-01 -5.31626761e-01
1.65450549e+00 -1.14294505e+00 -2.90390067e-02 -2.10637167e-01
-4.44277197e-01 -9.35755521e-02 4.91392106e-01 -1.76484182e-01
4.91925269e-01 -7.30153620e-01 -7.09162712e-01 5.89708686e-01
-3.07246983e-01 -6.27434909e-01 3.31902876e-02 4.85714823e-01
-1.71427000e+00 6.70614362e-01 -9.89277005e-01 5.53050358e-03
-2.14801335e+00 6.26618743e-01 4.89817500e-01 -2.30849311e-01
-2.35390022e-01 7.07353055e-01 -2.79752493e-01 1.31719068e-01
5.61657399e-02 -1.24010161e-01 -2.96064168e-01 -2.05541834e-01
3.49815339e-01 5.60817182e-01 4.47724909e-01 -5.32835305e-01
-4.10937995e-01 5.98723233e-01 -9.56673399e-02 -1.37438238e-01
1.27804673e+00 1.76483050e-01 -6.02389157e-01 -4.14009273e-01
1.36814690e+00 3.41583788e-02 -3.31644088e-01 4.59857672e-01
-4.94064502e-02 -1.11020096e-01 -1.47262186e-01 -7.32540131e-01
-6.88310564e-01 6.68488026e-01 1.01783395e+00 1.34863839e-01
1.34465945e+00 -1.40126526e+00 7.35493839e-01 1.09575140e+00
7.75162503e-02 -1.17968702e+00 -7.70686269e-01 5.28201282e-01
1.79252535e-01 -6.11829162e-01 4.93101865e-01 -1.99726507e-01
-3.38167310e-01 2.00734258e+00 1.46580905e-01 -6.97697222e-01
8.27593982e-01 9.23645854e-01 -1.28367543e-01 3.72086056e-02
-4.25952435e-01 8.13620016e-02 -3.66189107e-02 8.63281250e-01
1.29075989e-01 -3.08363736e-01 -1.13840222e+00 3.95963416e-02
-5.47353625e-01 -2.76774794e-01 1.08132029e+00 1.48305762e+00
-7.07041174e-02 -2.00592232e+00 -8.71270716e-01 1.57800894e-02
-9.15047050e-01 -1.33163959e-01 -3.95071983e-01 1.52012742e+00
5.15779674e-01 7.23360240e-01 8.34424421e-03 -6.34883404e-01
-7.43681639e-02 6.38291761e-02 6.12634599e-01 -4.26197834e-02
-1.34005821e+00 -4.55492139e-01 2.08415210e-01 1.57268923e-02
-2.11767718e-01 -5.75102746e-01 -1.45736468e+00 -5.84015012e-01
-2.37603456e-01 7.72188187e-01 1.52768564e+00 8.55907142e-01
-1.88476652e-01 3.02221864e-01 6.59858406e-01 -2.56395429e-01
-3.13903362e-01 -5.22337914e-01 -7.67245293e-01 -7.39457831e-02
1.97239686e-02 -4.37770128e-01 -3.71967465e-01 -5.05395174e-01]
|
[5.728446960449219, 4.673893928527832]
|
1e00473d-a36e-4cf7-b291-e4a0612352e1
|
deep-generative-views-to-mitigate-gender
|
2208.08382
| null |
https://arxiv.org/abs/2208.08382v1
|
https://arxiv.org/pdf/2208.08382v1.pdf
|
Deep Generative Views to Mitigate Gender Classification Bias Across Gender-Race Groups
|
Published studies have suggested the bias of automated face-based gender classification algorithms across gender-race groups. Specifically, unequal accuracy rates were obtained for women and dark-skinned people. To mitigate the bias of gender classifiers, the vision community has developed several strategies. However, the efficacy of these mitigation strategies is demonstrated for a limited number of races mostly, Caucasian and African-American. Further, these strategies often offer a trade-off between bias and classification accuracy. To further advance the state-of-the-art, we leverage the power of generative views, structured learning, and evidential learning towards mitigating gender classification bias. We demonstrate the superiority of our bias mitigation strategy in improving classification accuracy and reducing bias across gender-racial groups through extensive experimental validation, resulting in state-of-the-art performance in intra- and cross dataset evaluations.
|
['Ajita Rattani', 'Sreeraj Ramachandran']
|
2022-08-17
| null | null | null | null |
['facial-attribute-classification']
|
['computer-vision']
|
[ 2.30206817e-01 2.19754279e-01 -3.79702449e-01 -7.36322284e-01
-4.18783545e-01 -4.31748420e-01 8.20625842e-01 -2.69585680e-02
-1.85451299e-01 6.64959669e-01 2.44660050e-01 -1.12154409e-01
2.06847176e-01 -7.72183359e-01 -1.22038431e-01 -5.63129187e-01
1.36300579e-01 8.77849981e-02 -4.57946241e-01 4.84458357e-02
5.46484232e-01 1.86809868e-01 -1.81439459e+00 2.74884611e-01
1.11562431e+00 9.10380185e-01 -9.26487565e-01 2.82712638e-01
-3.45136710e-02 4.96723622e-01 -6.17344737e-01 -1.08461380e+00
3.56279939e-01 -4.70293224e-01 -2.39772469e-01 -7.27374181e-02
1.09695554e+00 -3.48560989e-01 2.69888639e-02 9.61041093e-01
8.22019160e-01 -5.16615868e-01 9.11870539e-01 -1.62434638e+00
-8.20894718e-01 3.03137302e-01 -1.32496357e+00 2.26001944e-02
1.49479136e-01 2.40964174e-01 7.27371573e-01 -1.01935720e+00
3.59736830e-01 1.59256160e+00 9.59884584e-01 9.12445128e-01
-1.39581251e+00 -1.42168152e+00 2.85690576e-01 -8.73627439e-02
-1.46853387e+00 -6.51967585e-01 7.28260875e-01 -6.31602943e-01
4.63736415e-01 2.26576209e-01 7.17777491e-01 1.32914913e+00
2.85657763e-01 3.53317410e-01 1.78508115e+00 -4.38238025e-01
-7.35201091e-02 3.34316492e-01 2.86868252e-02 7.04002917e-01
6.36920333e-01 3.28353673e-01 -1.08049846e+00 -2.49408230e-01
4.65172619e-01 -2.08183050e-01 2.38498792e-01 -3.14311504e-01
-6.30145073e-01 8.82274866e-01 1.32491395e-01 -3.16360325e-01
3.21487114e-02 -6.65864870e-02 4.64599252e-01 -6.67121336e-02
9.40045118e-01 2.76478559e-01 1.88314058e-02 7.45578203e-04
-1.14548659e+00 4.02636409e-01 6.74371898e-01 6.05753362e-01
6.15881562e-01 1.27893344e-01 -4.36742991e-01 9.84156787e-01
3.59306842e-01 7.12528706e-01 -1.70265958e-01 -8.03223729e-01
4.05320644e-01 8.29678178e-01 -9.25552025e-02 -1.18662775e+00
-3.80355477e-01 -4.66396242e-01 -8.21070731e-01 5.37290573e-01
6.68411613e-01 -2.40731046e-01 -1.10464525e+00 1.92427874e+00
4.49052721e-01 -3.01390857e-01 -3.58633310e-01 7.31484175e-01
8.08874905e-01 -1.46769717e-01 6.38155282e-01 -1.57298937e-01
1.34625387e+00 -4.85620707e-01 -5.62686980e-01 -4.89889503e-01
2.03535214e-01 -7.60165393e-01 9.91176367e-01 1.66331157e-01
-9.27172124e-01 -5.17737448e-01 -9.00326908e-01 5.98738790e-02
-1.83269739e-01 2.81523407e-01 6.65153503e-01 1.52859616e+00
-7.20300615e-01 2.46499568e-01 -5.43708324e-01 -3.97840381e-01
1.15609813e+00 2.78485149e-01 -1.07667476e-01 1.41141554e-02
-6.79728448e-01 8.16661060e-01 -3.91693294e-01 3.47947739e-02
-6.81455374e-01 -1.42558074e+00 -5.73371649e-01 -3.71945769e-01
7.06017390e-02 -8.60392272e-01 8.62790644e-01 -1.21432543e+00
-9.12066877e-01 1.55867815e+00 -2.94490159e-01 -2.98118711e-01
1.01249647e+00 -3.73311698e-01 -3.99496198e-01 -1.78691987e-02
2.05819026e-01 9.55851734e-01 1.14059532e+00 -1.46835876e+00
-6.06068194e-01 -9.81147051e-01 -1.96395680e-01 1.27202243e-01
-4.60282922e-01 1.58727333e-01 3.21692735e-01 -6.71805739e-01
-1.72934644e-02 -1.01658833e+00 2.11312145e-01 1.59288168e-01
-3.82019937e-01 -1.39395073e-01 8.66945148e-01 -6.10100150e-01
1.05048323e+00 -2.10649824e+00 -2.49788642e-01 9.15043876e-02
6.91080034e-01 1.25095919e-01 2.23440528e-01 7.87653998e-02
-7.01842532e-02 4.70053911e-01 1.09043799e-01 -3.65438670e-01
-4.44819368e-02 -1.99813932e-01 -3.19431573e-01 6.94860280e-01
4.22099590e-01 8.06604803e-01 -5.03622055e-01 -7.10026562e-01
-6.51162565e-02 5.90764761e-01 -5.15820682e-01 -4.05710228e-02
2.91238576e-01 5.02600491e-01 1.36623550e-02 1.16826701e+00
1.05229628e+00 2.19743013e-01 2.84119487e-01 -2.23396212e-01
-5.22219017e-02 -1.28834927e-02 -7.21793592e-01 9.53602433e-01
-2.03408688e-01 6.87716305e-01 1.75666794e-01 -4.97782230e-01
9.36438799e-01 -1.82593837e-01 2.92230815e-01 -8.10353398e-01
1.99437723e-01 1.84551910e-01 2.64225453e-01 -2.26515085e-01
4.88583595e-01 -4.69442248e-01 1.25385046e-01 3.18107396e-01
-1.05227083e-01 1.14696119e-02 -1.31657152e-02 -8.65746215e-02
2.46363461e-01 1.86040506e-01 2.53096800e-02 -4.16027546e-01
1.49889901e-01 -1.63502544e-01 6.39595449e-01 8.46345305e-01
-7.42552042e-01 8.72419119e-01 7.82402754e-01 -2.51480937e-01
-8.40584040e-01 -1.16162574e+00 -3.39448005e-01 1.36734951e+00
4.66887131e-02 -3.55075419e-01 -8.53970885e-01 -9.62703764e-01
3.68173033e-01 5.56918859e-01 -1.13046396e+00 -3.12349409e-01
-3.30813318e-01 -1.19192278e+00 9.42719400e-01 6.70615911e-01
8.37289214e-01 -4.08652902e-01 -7.24034309e-01 -5.54432452e-01
-3.01959701e-02 -9.24251974e-01 -2.43253082e-01 -4.16844785e-01
-8.16575885e-01 -1.38436818e+00 -6.85713410e-01 -2.34245613e-01
8.56519461e-01 1.68481067e-01 1.47929454e+00 1.17852144e-01
-5.39103746e-01 5.16557693e-01 4.69869114e-02 -1.04406261e+00
-2.46314228e-01 8.02846029e-02 5.49012572e-02 1.98572889e-01
5.03161967e-01 -3.75816554e-01 -1.00693512e+00 5.57913959e-01
-3.12943757e-01 4.57903631e-02 4.33365852e-01 6.66042745e-01
2.23289937e-01 -4.25255358e-01 6.97702825e-01 -1.12607872e+00
4.86222833e-01 -2.27250636e-01 -3.42344344e-01 9.15162340e-02
-1.26426113e+00 -4.69810754e-01 -5.71161695e-02 -4.30488408e-01
-1.40932202e+00 -3.19577575e-01 3.00243586e-01 -1.08360238e-01
-1.12039469e-01 9.12922472e-02 -1.62178129e-01 -2.90078849e-01
8.43569994e-01 -2.39185691e-01 2.65890419e-01 -1.94655471e-02
1.92747802e-01 7.62634039e-01 3.01249087e-01 -6.74881101e-01
7.24513650e-01 7.00279772e-01 2.06180856e-01 -6.63162589e-01
-1.03584290e+00 -5.06408438e-02 -5.47022998e-01 -5.52108824e-01
7.10601509e-01 -1.17841411e+00 -5.96105278e-01 8.70417416e-01
-5.62199116e-01 -2.16742530e-02 3.94445248e-02 6.16256706e-02
-7.13422820e-02 1.06936619e-01 -1.50146753e-01 -1.11733878e+00
-5.23977935e-01 -7.34241128e-01 1.17396665e+00 6.14095271e-01
-6.74174845e-01 -7.67179549e-01 -2.20365837e-01 8.06057334e-01
4.12629604e-01 4.83980566e-01 9.95425105e-01 -3.11222166e-01
-4.42252792e-02 -1.50421277e-01 -6.06063068e-01 1.00150086e-01
1.37375325e-01 3.22316468e-01 -1.49376798e+00 -2.38760263e-01
-5.03445923e-01 -4.68058079e-01 9.95865464e-01 4.23706144e-01
1.24108088e+00 2.52573704e-03 -4.23878521e-01 5.62323749e-01
1.13733172e+00 -1.59860998e-01 6.43429041e-01 3.68682630e-02
9.21741307e-01 1.10235822e+00 5.96008360e-01 4.70647693e-01
6.54662848e-01 2.33969837e-01 2.95441151e-01 -2.61343479e-01
-4.41813260e-01 -4.48225856e-01 -1.20406464e-01 -2.17899725e-01
-4.84760374e-01 1.88257277e-01 -1.11883843e+00 4.34359789e-01
-1.50288498e+00 -1.02847350e+00 -1.30210295e-01 2.33060145e+00
6.65377617e-01 -1.61553733e-02 5.58133721e-01 9.28046703e-02
7.70673335e-01 3.42861921e-01 -7.45377362e-01 -3.42331707e-01
-2.63597310e-01 7.76704550e-02 4.77527261e-01 -3.99389379e-02
-1.21460831e+00 7.24976301e-01 7.61300230e+00 5.43960929e-01
-1.33714521e+00 -1.39756277e-01 1.30016673e+00 -3.23313028e-01
-1.61745250e-01 -1.72959611e-01 -9.62319374e-01 3.13895971e-01
5.48482895e-01 -1.10713817e-01 1.41461179e-01 6.99457824e-01
3.29513513e-02 -3.43408018e-01 -1.06857252e+00 1.38548541e+00
3.38460594e-01 -9.93590355e-01 -8.25425908e-02 2.88533002e-01
8.34170401e-01 -4.87481505e-01 6.63579285e-01 3.66130382e-01
-7.62627050e-02 -1.45315433e+00 8.20851505e-01 6.81356639e-02
1.12561655e+00 -8.79810393e-01 5.09275079e-01 -2.02911288e-01
-7.22768962e-01 -2.79891312e-01 3.03863678e-02 -3.42676491e-01
-4.00370717e-01 6.33167505e-01 -6.71522379e-01 3.33683521e-01
9.02730703e-01 5.59057295e-01 -1.01498294e+00 6.18069589e-01
-1.20703056e-01 7.76445091e-01 4.09732871e-02 2.34615877e-01
-4.07565713e-01 -1.75004780e-01 2.41951704e-01 1.17200112e+00
3.65446471e-02 -1.12955905e-02 -8.36154446e-02 9.15461898e-01
-9.65468064e-02 -1.11918710e-02 -5.99292219e-01 -3.56776044e-02
5.12260616e-01 1.44046831e+00 -7.68188119e-01 -5.21106683e-02
-4.17893559e-01 3.04140717e-01 7.88142458e-02 3.46896321e-01
-8.11478198e-01 7.27462471e-02 1.00676346e+00 5.60287416e-01
-4.74189818e-02 -5.27962297e-03 -1.24398887e+00 -8.24468911e-01
-5.28000332e-02 -1.17739713e+00 5.44025660e-01 -3.86358231e-01
-1.43773603e+00 -6.47396371e-02 1.10753417e-01 -9.98480678e-01
-1.75506175e-02 -4.49942023e-01 -4.61304426e-01 8.89980257e-01
-1.22723210e+00 -1.86414492e+00 -6.56390727e-01 2.20129430e-01
1.00507118e-01 -5.67616284e-01 6.92686677e-01 1.75435647e-01
-6.14969730e-01 1.10986590e+00 -6.07111752e-01 1.28896490e-01
1.16262603e+00 -1.11272919e+00 -1.07267862e-02 6.32775128e-01
-2.14378759e-01 8.37366164e-01 7.33130455e-01 -7.04003155e-01
-1.29400790e+00 -9.19447243e-01 5.15739202e-01 -6.61832690e-01
1.71662793e-01 -5.57940900e-01 -4.33752596e-01 3.84207308e-01
2.54210591e-01 5.32618212e-03 1.16148961e+00 7.52007067e-01
-1.09748554e+00 -3.01562726e-01 -1.47742653e+00 6.67083621e-01
1.28205252e+00 -5.78605890e-01 -2.29291111e-01 -1.78643584e-01
-2.07280710e-01 -5.93005002e-01 -6.03342295e-01 6.62383854e-01
1.29804814e+00 -1.41498709e+00 1.00142980e+00 -4.70209926e-01
9.44414616e-01 -2.94958819e-02 -1.19176228e-02 -1.14775515e+00
-1.59644276e-01 -2.95703143e-01 -5.51990755e-02 1.65426481e+00
2.35755876e-01 -6.88645542e-01 1.00942218e+00 8.29633117e-01
3.18235010e-01 -7.57043719e-01 -7.20366240e-01 -3.91618520e-01
3.77611637e-01 -1.44451365e-01 5.44350266e-01 8.82204473e-01
-5.45134306e-01 2.04988927e-01 -2.65487790e-01 9.00911987e-02
1.02798259e+00 2.73212790e-01 1.08947980e+00 -1.34212470e+00
2.67720610e-01 -6.00909472e-01 -3.61485124e-01 -1.69144403e-02
3.72976750e-01 -6.04418397e-01 -3.17725390e-01 -1.05569458e+00
6.27825916e-01 -4.39236969e-01 -1.11798383e-01 5.46820462e-01
-4.98103648e-01 8.25333059e-01 2.57299393e-01 8.61731097e-02
-1.92009076e-01 2.26740882e-01 1.15505755e+00 -2.85272568e-01
-1.69136133e-02 -1.60838738e-01 -1.48907661e+00 7.72732913e-01
7.05278099e-01 -4.04964089e-01 -4.92630899e-01 -2.12304756e-01
2.79656768e-01 -5.34240305e-01 7.42072284e-01 -7.42730677e-01
-2.41215989e-01 -2.78525412e-01 1.11328983e+00 -5.52002847e-01
2.27726951e-01 -3.22128117e-01 -1.64499521e-01 4.65172976e-01
-1.35543674e-01 -7.82495961e-02 3.38255614e-01 4.17404711e-01
-5.92768528e-02 5.62570453e-01 1.03962445e+00 1.02682814e-01
-3.15295994e-01 1.91209793e-01 -1.79422535e-02 5.54270267e-01
8.96938741e-01 -4.29781169e-01 -7.51246870e-01 -3.07413757e-01
-1.26524508e-01 2.30698720e-01 4.69667226e-01 5.90158701e-01
3.74225795e-01 -1.11053491e+00 -9.49580014e-01 4.41135943e-01
3.97736043e-01 -5.01047492e-01 2.69032001e-01 9.44063246e-01
-1.48096278e-01 -6.94353059e-02 -5.95968366e-01 -7.75749922e-01
-1.79434812e+00 1.24583863e-01 4.94771928e-01 1.98329389e-01
2.14756448e-02 9.27496254e-01 3.88305306e-01 -4.77067649e-01
3.96958552e-02 2.51904458e-01 -1.94139123e-01 6.02453649e-01
3.94218624e-01 9.81756806e-01 5.45455702e-02 -6.37019336e-01
-6.52933598e-01 7.37603724e-01 -7.96913579e-02 -1.24748208e-01
1.02768207e+00 -1.29358023e-01 -1.86995551e-01 1.77381709e-01
6.18130147e-01 3.33283544e-01 -1.23883116e+00 4.65301961e-01
-3.25743705e-01 -9.79069531e-01 -2.09666312e-01 -1.08505166e+00
-1.08362985e+00 7.89434373e-01 1.07915986e+00 6.47247210e-02
1.07126236e+00 -3.96169960e-01 1.76692814e-01 -4.13734704e-01
3.30033600e-01 -1.29373288e+00 5.12493700e-02 1.91590432e-02
7.22275913e-01 -1.67669070e+00 4.43321139e-01 -7.61443913e-01
-6.04049087e-01 7.89782703e-01 9.44652796e-01 9.67395902e-02
6.34330392e-01 1.71955615e-01 6.18649065e-01 -2.04142720e-01
-4.38449323e-01 4.06416133e-02 4.64409500e-01 9.50990915e-01
7.95615792e-01 2.34286070e-01 -5.59913456e-01 5.97978532e-01
-5.22735953e-01 1.38209369e-02 2.28837505e-01 6.73579872e-01
1.64735183e-01 -9.35029387e-01 -5.82672656e-01 6.18345857e-01
-8.40368330e-01 1.66078031e-01 -1.00317097e+00 8.46924603e-01
4.97327179e-01 1.14788640e+00 1.05002269e-01 -3.62060547e-01
2.92407781e-01 2.64558047e-01 8.96344364e-01 -3.52425903e-01
-8.33836377e-01 -9.20333862e-02 3.05951595e-01 -4.84953225e-01
-6.20654345e-01 -8.56468201e-01 -7.33749151e-01 -5.38559139e-01
8.42185225e-04 -5.83276451e-01 5.66172481e-01 7.18011796e-01
5.89967668e-01 1.36043459e-01 5.33682466e-01 -5.00647426e-01
-6.06986165e-01 -8.79436731e-01 -5.24879217e-01 5.64109564e-01
1.82278901e-01 -9.55571115e-01 -3.02133679e-01 -6.78761154e-02]
|
[13.024470329284668, 1.2373592853546143]
|
81865afe-17b6-42be-a521-afde4150d9da
|
koopman-type-inverse-operator-for-linear-non
|
2305.04158
| null |
https://arxiv.org/abs/2305.04158v1
|
https://arxiv.org/pdf/2305.04158v1.pdf
|
Koopman-type inverse operator for linear non-minimum phase systems with disturbances
|
In this paper, a novel Koopman-type inverse operator for linear time-invariant non-minimum phase systems with stochastic disturbances is proposed. This operator employs functions of the desired output to directly calculate the input. Furthermore, it can be applied as a data-driven approach for systems with unknown parameters yet a known relative degree, which is a departure from the majority of existing data-driven methods that are only applicable to minimum phase systems. Based on this foundation, we use the Monte Carlo approach to develop an improved Koopman-type method for addressing the issue of inaccurate parameter estimation in data-driven systems with large disturbances. The simulation results justify the tracking accuracy of Koopman-type operator.
|
['Xiaoqiang Ji', 'Yuhan Li']
|
2023-05-07
| null | null | null | null |
['type']
|
['speech']
|
[ 2.45158955e-01 -1.35312006e-01 -2.02288702e-02 1.82301611e-01
-5.95527709e-01 -6.21928632e-01 4.86756653e-01 -1.00437589e-01
-1.99458510e-01 1.16042924e+00 -3.15237343e-01 -4.13368434e-01
-8.85392845e-01 -5.00763237e-01 -3.51815671e-01 -9.85795319e-01
2.80116480e-02 4.67427760e-01 1.89487651e-01 -4.14805919e-01
2.26100355e-01 4.71185058e-01 -1.31990099e+00 -4.62456822e-01
9.45932865e-01 6.83946729e-01 -2.48825684e-01 6.98648751e-01
8.05527687e-01 3.73517901e-01 -7.04253554e-01 6.05716527e-01
4.40289468e-01 -5.61999679e-01 -2.62964308e-01 -2.10702755e-02
-3.84186029e-01 -5.77474684e-02 -4.08749670e-01 1.01122534e+00
7.08678842e-01 4.64632362e-01 9.47384298e-01 -1.23684454e+00
-1.70168102e-01 3.96627896e-02 -1.94989055e-01 2.81495035e-01
5.10117888e-01 4.73684728e-01 2.01158419e-01 -1.02989531e+00
3.39908987e-01 9.70698833e-01 8.10080171e-01 1.80375382e-01
-1.34008658e+00 -5.44219971e-01 -1.29085973e-01 4.71820030e-03
-1.71933031e+00 -2.15654761e-01 6.20689452e-01 -8.16189349e-01
7.40407765e-01 3.90305966e-01 7.36769140e-01 5.32294869e-01
4.41295654e-01 1.45112157e-01 1.38161409e+00 -4.05714989e-01
4.95674819e-01 -5.57864867e-02 6.38484955e-02 2.03081459e-01
7.45122969e-01 6.92120731e-01 -1.90320477e-01 -3.59839648e-01
7.51092851e-01 -3.70276660e-01 -5.09061813e-01 -5.89983761e-01
-9.53618169e-01 7.24052012e-01 4.60630357e-02 2.80929387e-01
-4.52867419e-01 5.36139943e-02 1.83037758e-01 3.54822576e-01
3.04076433e-01 4.50473756e-01 -3.11452329e-01 -2.00890377e-01
-7.61029720e-01 6.11146212e-01 9.88892674e-01 1.00366306e+00
3.24775845e-01 8.47190797e-01 -8.47644806e-02 -7.18301013e-02
6.07953608e-01 1.13143754e+00 9.70123485e-02 -6.62439704e-01
9.87854972e-02 5.10488153e-01 6.77272260e-01 -5.72408557e-01
-3.86046827e-01 -4.66485232e-01 -8.68231714e-01 5.97696245e-01
5.63112974e-01 -6.00304663e-01 -8.75378251e-01 1.40964103e+00
2.68541306e-01 4.03032592e-03 1.97899505e-01 8.49049032e-01
-3.69696133e-02 7.93271542e-01 -7.03062654e-01 -9.04274464e-01
8.83382201e-01 -3.02633703e-01 -1.29007578e+00 2.01270327e-01
1.94084704e-01 -8.27155054e-01 7.73869753e-01 5.92555225e-01
-8.70796204e-01 -2.98919141e-01 -1.27909005e+00 8.31899047e-01
-2.73648888e-01 -8.05268884e-02 -1.33325517e-01 7.61850834e-01
-7.48577833e-01 4.76444334e-01 -5.40762901e-01 -1.33232430e-01
-4.54872400e-01 2.21549273e-01 2.43432105e-01 5.89040637e-01
-1.34386897e+00 1.30434465e+00 3.29980850e-01 4.25587744e-01
-7.75237501e-01 -7.97256887e-01 -5.71181715e-01 -4.21556503e-01
5.98407030e-01 -5.17645836e-01 1.24565089e+00 -5.01140237e-01
-1.95879376e+00 -2.31679082e-01 -2.33989775e-01 -4.41234827e-01
4.17455584e-01 2.04144884e-02 -4.92098093e-01 -1.61975659e-02
-1.54157534e-01 -6.94463432e-01 9.84583259e-01 -1.07669318e+00
-2.63885081e-01 3.17923762e-02 -4.86042917e-01 9.01472941e-02
3.06698173e-01 -3.52307498e-01 6.27074003e-01 -4.65251744e-01
2.42840394e-01 -1.06300151e+00 -4.75391299e-01 -3.19380969e-01
-4.06476408e-01 2.58936211e-02 1.05522418e+00 -2.36474067e-01
1.44359994e+00 -1.98045170e+00 -5.81892906e-03 6.35920286e-01
-3.64326566e-01 5.14242113e-01 4.43319827e-01 1.29437530e+00
-2.09274903e-01 -5.46342619e-02 -3.32944840e-01 5.43814182e-01
-5.63573763e-02 -4.34988104e-02 -5.82220793e-01 7.71148562e-01
3.18370849e-01 3.39779884e-01 -8.76519144e-01 -6.11687303e-02
6.17466211e-01 2.97431529e-01 -2.39131734e-01 -7.24462196e-02
2.87873358e-01 6.26315475e-01 -5.29783368e-01 2.98465341e-01
5.77424705e-01 9.82512236e-02 -3.87963839e-02 -2.96355397e-01
-6.07107699e-01 -8.10380280e-02 -1.88854420e+00 7.80022442e-01
-5.10885298e-01 8.12489569e-01 2.75818378e-01 -1.03982031e+00
9.56904709e-01 7.28249550e-01 5.91874778e-01 -1.71814412e-01
5.21967292e-01 4.30851549e-01 2.22035915e-01 -3.88630539e-01
1.23466952e-02 -5.09660482e-01 -3.09481286e-02 2.12185636e-01
-1.16991445e-01 -8.48772824e-01 3.71677041e-01 -5.70754148e-02
8.95562530e-01 5.44769987e-02 8.89587522e-01 -7.76360154e-01
8.92490625e-01 2.71293074e-01 6.14692688e-01 4.04133171e-01
-3.15004885e-01 2.61561960e-01 1.83065563e-01 -1.68295145e-01
-1.07346046e+00 -1.01683319e+00 -4.76829261e-01 7.64730852e-03
3.84065896e-01 -4.09721248e-02 -4.73908275e-01 8.98970887e-02
1.18771240e-01 8.56938064e-01 -2.82124579e-01 -4.45593894e-01
-3.13704789e-01 -8.57888281e-01 1.54320985e-01 2.45274797e-01
2.15606824e-01 -2.18635425e-01 -3.77461970e-01 5.21533608e-01
2.39777490e-01 -7.28573442e-01 -2.82682031e-01 3.62107128e-01
-8.52368951e-01 -1.29611719e+00 -5.60773075e-01 -4.93527830e-01
9.60198343e-01 -1.09248742e-01 4.40014273e-01 -3.93146902e-01
-2.06899598e-01 6.30711913e-01 -1.20041803e-01 -7.26370335e-01
-5.07440627e-01 -5.24646461e-01 7.38631070e-01 1.58631071e-01
-8.42733309e-02 -3.69477451e-01 -3.96901816e-01 7.28957534e-01
-6.02256835e-01 -4.80873048e-01 2.94835299e-01 9.90416288e-01
3.65376621e-01 5.46795070e-01 9.17500079e-01 -4.95917797e-01
1.00130022e+00 -1.29927069e-01 -1.35315990e+00 -2.22278014e-01
-9.66806889e-01 4.17700969e-02 1.12315845e+00 -7.85055637e-01
-9.68774736e-01 4.07187670e-01 3.22311014e-01 -2.02335790e-01
3.05138528e-01 3.30779642e-01 -3.17421891e-02 -4.66665298e-01
7.91025221e-01 2.37746984e-01 3.37719977e-01 -9.74013358e-02
1.33124180e-02 7.86319017e-01 2.38295898e-01 -3.12933385e-01
1.48377943e+00 3.73396426e-01 5.12576580e-01 -8.59099507e-01
-4.32851702e-01 -5.85105717e-01 -3.96414489e-01 -2.81255662e-01
3.26382250e-01 -7.51548052e-01 -1.10237682e+00 6.13592207e-01
-7.39294469e-01 -2.67755836e-01 -8.54789138e-01 9.94321167e-01
-9.83612299e-01 2.21285317e-02 -2.92937100e-01 -1.55675781e+00
-2.76307911e-02 -1.00145113e+00 6.29895031e-01 3.79305959e-01
-1.73372597e-01 -1.03169668e+00 3.74020278e-01 -4.09636140e-01
6.42476201e-01 4.34945315e-01 3.59503567e-01 -3.40594321e-01
-6.13433599e-01 -6.46961808e-01 4.52696055e-01 2.95956224e-01
-6.96708858e-02 2.68426061e-01 -5.63468754e-01 -6.47298455e-01
6.04475141e-01 2.15848699e-01 -1.20296463e-01 6.63176715e-01
1.24431133e-01 -3.91885459e-01 -5.05029738e-01 2.94352919e-01
1.94186974e+00 6.15461588e-01 2.32520387e-01 3.25451881e-01
8.75434726e-02 1.43155739e-01 1.08326876e+00 6.98723614e-01
-1.92810714e-01 5.20244062e-01 3.76572549e-01 -8.16426240e-03
4.72844154e-01 -1.41035259e-01 3.09675574e-01 7.70621359e-01
2.68484596e-02 -3.27536792e-01 -6.61086619e-01 6.62087739e-01
-1.61238694e+00 -9.02128935e-01 -6.51611507e-01 2.46601725e+00
6.39912426e-01 1.39271706e-01 -7.69222230e-02 7.13700950e-01
9.48126495e-01 -2.44839966e-01 -5.36639631e-01 -5.23124337e-01
1.49001613e-01 7.03820959e-02 1.05351424e+00 6.93517506e-01
-7.04983890e-01 1.55204087e-01 7.39732075e+00 7.80417681e-01
-8.33819509e-01 -1.88093424e-01 -2.74282485e-01 3.32371801e-01
4.09433618e-02 4.06247377e-01 -9.32800949e-01 5.98132074e-01
1.22280264e+00 -1.13784838e+00 4.06955987e-01 5.86624384e-01
1.01927221e+00 -5.35078049e-01 -6.37766600e-01 5.28730273e-01
-3.06155771e-01 -6.37089312e-01 -4.98871475e-01 1.07355937e-01
1.02354789e+00 -6.15407944e-01 1.00259647e-01 1.99500501e-01
1.67727649e-01 -4.94433463e-01 3.35076600e-01 7.62272418e-01
4.50994104e-01 -7.24123895e-01 8.15250874e-01 7.75338113e-01
-1.24875760e+00 -3.05988073e-01 4.57005901e-03 -5.17450571e-01
8.32064807e-01 9.31081057e-01 -9.67120409e-01 6.10085309e-01
4.89361845e-02 2.80014664e-01 1.71227992e-01 1.44576943e+00
-2.56295800e-01 7.96846211e-01 -6.91426992e-01 -5.64425290e-01
-7.60817081e-02 -3.34326982e-01 1.14960289e+00 5.21316648e-01
6.50499701e-01 1.14379324e-01 1.76541522e-01 7.04260528e-01
1.08677173e+00 -1.06918879e-01 -9.16730523e-01 1.49992824e-01
5.98536253e-01 8.70058894e-01 -2.73692727e-01 -3.38938415e-01
-9.96900201e-02 3.84410262e-01 -7.67303169e-01 6.53622687e-01
-9.00059402e-01 -9.71671999e-01 2.88248599e-01 2.73928106e-01
-2.02642325e-02 -6.41431749e-01 -1.94074467e-01 -5.08182704e-01
-1.03688911e-02 -7.10397899e-01 4.29856926e-02 -6.41954422e-01
-9.58809793e-01 3.65246743e-01 4.38612312e-01 -2.17407012e+00
-9.85748947e-01 -4.13993031e-01 -4.93325293e-01 1.13589501e+00
-1.12378895e+00 -4.28424686e-01 9.59421471e-02 4.85629648e-01
2.26853967e-01 -5.23208939e-02 7.07073271e-01 1.28147095e-01
-5.69024384e-01 -1.22971810e-01 8.34048629e-01 -3.87560695e-01
5.89435518e-01 -1.13844109e+00 -1.88554376e-02 1.09674191e+00
-6.75572455e-01 5.79334915e-01 1.40920365e+00 -6.17163241e-01
-1.64575958e+00 -8.90351474e-01 5.99595904e-01 -3.31876785e-01
1.00801039e+00 -1.10674851e-01 -6.56818211e-01 5.01377106e-01
1.05504468e-01 1.13237418e-01 -4.35257442e-02 -6.25001788e-01
6.42200291e-01 -3.94811511e-01 -1.30174458e+00 4.69125926e-01
4.76657480e-01 -2.53335327e-01 -7.44384646e-01 2.84959286e-01
4.10253525e-01 -5.31983674e-01 -1.15016532e+00 7.79550433e-01
1.05704211e-01 -3.58438551e-01 7.29577959e-01 8.48098919e-02
-7.05956280e-01 -1.03743136e+00 1.78690314e-01 -1.91484737e+00
-9.71841365e-02 -1.45772922e+00 -1.50877342e-01 9.25771832e-01
2.69238055e-01 -1.28289390e+00 3.58735561e-01 1.09392151e-01
-4.56957072e-02 -4.98491079e-01 -1.35153556e+00 -1.51481175e+00
-1.77833885e-01 -1.69984624e-01 1.55368121e-02 4.28529441e-01
3.62768292e-01 2.57722259e-01 -2.18499318e-01 6.85842156e-01
8.15028131e-01 -1.95330635e-01 7.30105519e-01 -1.08307242e+00
-5.02088778e-02 -1.94456503e-02 -3.60259295e-01 -7.09876716e-01
-3.28718811e-01 -2.77967662e-01 6.18315697e-01 -1.61160553e+00
-4.64402318e-01 -3.46740603e-01 6.24619145e-03 -1.47770867e-01
2.97893640e-02 1.45243317e-01 -2.16279402e-01 6.23819903e-02
9.66863036e-02 5.81847906e-01 1.15502799e+00 1.82379693e-01
-3.57945055e-01 7.43985236e-01 5.05719408e-02 7.72351563e-01
8.49090099e-01 -5.24733961e-01 -8.74227047e-01 3.24182957e-01
-6.75760433e-02 2.09700584e-01 2.75685400e-01 -1.63813841e+00
3.27974349e-01 -4.26936805e-01 -9.25201178e-02 -8.50487232e-01
1.89721435e-01 -1.24025118e+00 5.81389368e-01 1.00941610e+00
1.90161273e-01 7.36670718e-02 6.38668090e-02 7.62703121e-01
-4.65098351e-01 -3.71061474e-01 9.60559547e-01 3.03777844e-01
-2.80287325e-01 -6.12864532e-02 -1.16954696e+00 1.63327515e-01
1.50317836e+00 -5.48841834e-01 -3.08530629e-01 -7.00567365e-01
-7.05287099e-01 2.03966185e-01 2.84583241e-01 -1.06822111e-01
3.93626720e-01 -1.28592086e+00 -3.92667264e-01 3.66388977e-01
-1.89735278e-01 -2.96535194e-01 1.47657216e-01 1.11367393e+00
-3.52556586e-01 7.72526026e-01 9.41934139e-02 -7.63508141e-01
-7.58318305e-01 5.59102297e-01 6.47930026e-01 2.36687660e-02
-2.53892064e-01 1.78575605e-01 -2.72937983e-01 -2.90466160e-01
-1.43366054e-01 -5.05495250e-01 1.83329523e-01 -2.17618831e-02
3.91112536e-01 7.51076221e-01 9.33504999e-02 -5.24394631e-01
-2.88399667e-01 8.34704161e-01 7.31771588e-01 -5.02848387e-01
8.80947769e-01 -2.44103178e-01 3.21075320e-01 8.85025322e-01
7.29112864e-01 1.68032646e-01 -1.32191324e+00 1.03060111e-01
-1.27843574e-01 -3.78106058e-01 9.38763376e-03 -6.01246119e-01
-5.92335165e-01 4.89587218e-01 6.97129190e-01 6.10870481e-01
1.16823232e+00 -8.72396410e-01 1.89519659e-01 2.18014479e-01
6.38833463e-01 -1.28460729e+00 -3.11039597e-01 7.12461054e-01
8.12620342e-01 -5.89213133e-01 2.47987017e-01 -6.61524296e-01
-3.30071777e-01 9.86512959e-01 5.70988715e-01 -5.96223295e-01
7.97854125e-01 9.43183303e-01 -2.08384846e-03 3.25326920e-01
-8.75583887e-01 -3.60586941e-01 1.47440001e-01 8.25322688e-01
2.30319470e-01 -1.78241923e-01 -1.00328624e+00 1.08326241e-01
3.49212974e-01 2.86985129e-01 1.04245865e+00 1.24054122e+00
-6.69517338e-01 -9.34267998e-01 -8.75209451e-01 2.02583760e-01
-1.35430023e-01 2.84072191e-01 7.60674328e-02 1.22928441e+00
-3.45391512e-01 1.15089703e+00 -1.74508959e-01 -4.96967174e-02
1.14640760e+00 -5.42224497e-02 2.19093770e-01 -4.26642478e-01
-4.42504466e-01 1.71008304e-01 1.14783838e-01 -3.54026228e-01
-1.59472421e-01 -5.99541545e-01 -1.23263490e+00 -1.21627860e-02
-8.66726160e-01 5.21079481e-01 6.10557735e-01 7.05201864e-01
1.43462256e-01 6.78945661e-01 9.65613961e-01 -6.46021485e-01
-1.11551249e+00 -9.04133856e-01 -6.68610394e-01 -6.86074868e-02
5.26408970e-01 -1.03832483e+00 -1.07352006e+00 -1.11143790e-01]
|
[5.418814182281494, 2.5676114559173584]
|
c9045753-f3f1-4174-ad14-9b55816a3596
|
a-greedy-approach-to-ell_0infty-based
|
1812.10538
| null |
http://arxiv.org/abs/1812.10538v1
|
http://arxiv.org/pdf/1812.10538v1.pdf
|
A Greedy Approach to $\ell_{0,\infty}$ Based Convolutional Sparse Coding
|
Sparse coding techniques for image processing traditionally rely on a
processing of small overlapping patches separately followed by averaging. This
has the disadvantage that the reconstructed image no longer obeys the sparsity
prior used in the processing. For this purpose convolutional sparse coding has
been introduced, where a shift-invariant dictionary is used and the sparsity of
the recovered image is maintained. Most such strategies target the $\ell_0$
"norm" or the $\ell_1$ norm of the whole image, which may create an imbalanced
sparsity across various regions in the image. In order to face this challenge,
the $\ell_{0,\infty}$ "norm" has been proposed as an alternative that "operates
locally while thinking globally". The approaches taken for tackling the
non-convexity of these optimization problems have been either using a convex
relaxation or local pursuit algorithms. In this paper, we present an efficient
greedy method for sparse coding and dictionary learning, which is specifically
tailored to $\ell_{0,\infty}$, and is based on matching pursuit. We demonstrate
the usage of our approach in salt-and-pepper noise removal and image
inpainting. A code package which reproduces the experiments presented in this
work is available at https://web.eng.tau.ac.il/~raja
|
['Raja Giryes', 'Elad Plaut']
|
2018-12-26
| null | null | null | null |
['salt-and-pepper-noise-removal']
|
['computer-vision']
|
[ 3.72188449e-01 -1.24827467e-01 2.33072508e-02 -8.97139087e-02
-5.95797837e-01 -1.85934156e-01 4.46915068e-02 -3.85949835e-02
-3.66827250e-01 5.85997164e-01 6.70294091e-02 -2.83337720e-02
-1.85551584e-01 -7.23619401e-01 -5.33400118e-01 -1.04684854e+00
4.98334244e-02 -2.45024368e-01 -1.59643739e-01 -1.77260965e-01
4.96529073e-01 4.85388368e-01 -1.58975708e+00 2.37500355e-01
7.69269407e-01 1.11348331e+00 4.61095005e-01 2.52500206e-01
-2.43979067e-01 7.21628249e-01 -3.33494902e-01 -1.18724793e-01
5.23397446e-01 -6.78645909e-01 -3.47623587e-01 2.21547633e-01
3.30430567e-01 1.59398764e-02 -3.36119324e-01 1.64609432e+00
4.36585367e-01 3.21156114e-01 2.96015233e-01 -8.43928874e-01
-4.08257216e-01 2.40855470e-01 -9.70181286e-01 2.80643851e-01
3.54710579e-01 -1.52458861e-01 7.41063058e-01 -1.03363001e+00
5.93512595e-01 7.42585301e-01 6.89928174e-01 1.85729906e-01
-1.24381816e+00 -7.04575300e-01 2.38279626e-02 1.51257202e-01
-1.78122890e+00 -4.87959594e-01 1.21942282e+00 -3.65964413e-01
7.21514523e-01 4.22789812e-01 6.34041965e-01 5.28355539e-01
3.64048839e-01 5.50816059e-01 1.19068480e+00 -7.56371975e-01
2.47906685e-01 -5.79579780e-03 -9.27937329e-02 6.59041703e-01
2.13804781e-01 3.25266831e-02 -5.74705720e-01 5.76225519e-02
8.90311837e-01 1.93039924e-01 -4.38762963e-01 -1.99471340e-01
-9.46197510e-01 9.41572666e-01 2.45140672e-01 7.64380991e-01
-6.40070319e-01 7.29118064e-02 2.15534717e-01 4.17384446e-01
6.22912228e-01 1.88408494e-01 -2.64745504e-02 8.69216025e-02
-1.47029698e+00 2.06110492e-01 6.71442807e-01 8.02274704e-01
1.00126934e+00 3.11859161e-01 1.18193783e-01 9.29829061e-01
3.63747984e-01 4.68808383e-01 4.16155636e-01 -1.11292934e+00
3.34334999e-01 5.99781811e-01 -1.37800276e-02 -1.60531938e+00
-1.60080180e-01 -3.70016992e-01 -1.15814352e+00 4.92818892e-01
3.30518454e-01 -3.73494923e-02 -9.28389370e-01 1.50808787e+00
1.96791664e-01 3.02296817e-01 -1.75545681e-02 8.90969217e-01
6.46468759e-01 7.62933195e-01 -2.95718253e-01 -6.34257019e-01
1.12600696e+00 -6.59134984e-01 -9.98435974e-01 -3.19545418e-02
2.14911804e-01 -1.18084431e+00 8.04484844e-01 5.60381830e-01
-1.34331107e+00 -4.77714360e-01 -9.58768189e-01 7.53642917e-02
-2.57406443e-01 1.53099298e-01 2.91141272e-01 5.88714421e-01
-1.04265475e+00 4.71720129e-01 -7.04376638e-01 -1.31856695e-01
2.56846905e-01 2.72265494e-01 -4.97994035e-01 -3.97535056e-01
-7.33964026e-01 7.91148007e-01 1.34862229e-01 2.97740340e-01
-4.62310642e-01 -4.32079285e-01 -9.13646340e-01 4.91321906e-02
2.48757005e-01 -2.17254490e-01 7.28482068e-01 -1.18871176e+00
-1.27901602e+00 9.58511353e-01 -2.85035461e-01 -3.41692656e-01
2.76422590e-01 -7.53884912e-02 -3.10129344e-01 2.15503812e-01
1.70540899e-01 3.10590923e-01 1.39374006e+00 -1.08261085e+00
-2.50682503e-01 -3.88947368e-01 -2.29054287e-01 1.60064220e-01
-1.69724137e-01 2.34402850e-01 -4.04244542e-01 -1.28160632e+00
6.73509061e-01 -7.40568578e-01 -4.31150436e-01 -8.34928527e-02
-7.16536716e-02 3.25132698e-01 8.00342441e-01 -8.09690952e-01
1.51843393e+00 -2.44278288e+00 2.63231516e-01 5.43977141e-01
1.96011513e-01 2.54954278e-01 8.48240703e-02 5.67235589e-01
-2.56504089e-01 -2.80822217e-01 -6.58171773e-01 -3.83647799e-01
-4.48085576e-01 2.92717248e-01 -1.61971658e-01 8.06129277e-01
-1.29707158e-01 2.49216214e-01 -6.17821991e-01 -4.58952308e-01
3.28208238e-01 7.23678529e-01 -5.79989910e-01 4.33811359e-02
1.37851179e-01 5.98715603e-01 -1.68629512e-01 7.86212444e-01
1.00839508e+00 4.15543914e-02 -3.12170777e-02 -3.11403751e-01
-6.20429337e-01 -2.91266948e-01 -1.72690070e+00 1.90918243e+00
-1.08568884e-01 4.72646832e-01 7.23540783e-01 -1.49973953e+00
1.16917014e+00 3.91549826e-01 9.16876733e-01 -6.87586606e-01
2.38977715e-01 4.25084889e-01 -2.12566823e-01 -5.35067260e-01
3.65691245e-01 -4.14075822e-01 2.84475952e-01 2.36532331e-01
-1.29494131e-01 -2.86696106e-01 3.64088446e-01 -4.07076702e-02
1.00754237e+00 4.14676480e-02 4.29366499e-01 -5.93868613e-01
6.77875996e-01 1.19816586e-01 7.31275141e-01 5.86764336e-01
-1.51373550e-01 9.18677807e-01 3.64674628e-01 -4.77317631e-01
-9.28808689e-01 -6.95445538e-01 -1.86953545e-01 7.03029990e-01
1.54798001e-01 -2.30638787e-01 -7.71414816e-01 -2.51038373e-01
-2.00649589e-01 4.57480073e-01 -5.56173444e-01 1.19692259e-01
-8.41311514e-01 -7.14340568e-01 2.69706935e-01 1.18277520e-01
6.61104679e-01 -1.13679492e+00 -6.92041934e-01 3.85373205e-01
-1.38014644e-01 -6.96537971e-01 -3.82653147e-01 2.54161090e-01
-8.85657251e-01 -9.54725385e-01 -1.07771814e+00 -7.88687527e-01
1.06912374e+00 4.76150185e-01 7.20935822e-01 2.20728174e-01
-4.18883860e-01 2.80275553e-01 -6.50395334e-01 -3.21709901e-01
6.80800155e-02 -4.18151408e-01 -8.46185014e-02 4.14662600e-01
1.52730972e-01 -7.40579367e-01 -6.59528613e-01 1.43832564e-01
-1.17723012e+00 -1.67568460e-01 6.44891858e-01 9.30769205e-01
1.01898408e+00 3.19332510e-01 3.32145751e-01 -8.52228880e-01
4.69798863e-01 -4.27649051e-01 -5.98293960e-01 -5.83349243e-02
-4.60236460e-01 -1.96293011e-01 6.72147512e-01 -2.68121779e-01
-8.83694470e-01 3.56985480e-01 -3.84810388e-01 -8.88531029e-01
7.72548914e-02 7.59302437e-01 -4.99908142e-02 -4.86971080e-01
5.46218872e-01 6.21658206e-01 2.81191051e-01 -6.12939060e-01
1.63748041e-01 3.51716429e-01 4.80450869e-01 -3.46793205e-01
6.75163805e-01 8.40237021e-01 1.22357206e-02 -1.12248671e+00
-5.52654862e-01 -7.30071902e-01 -4.22999263e-01 -1.76710159e-01
6.21634543e-01 -7.97685146e-01 -2.04392552e-01 4.97427374e-01
-9.63497519e-01 -2.51114182e-02 -6.34703219e-01 7.02108741e-01
-5.64055741e-01 6.11033380e-01 -2.72002339e-01 -7.47977376e-01
-3.37651104e-01 -1.11573970e+00 6.50685370e-01 2.56084502e-01
-7.62675107e-02 -7.80239642e-01 1.50507376e-01 1.65812999e-01
7.33751118e-01 3.12017381e-01 3.94710481e-01 -8.89461637e-02
-3.11514586e-01 -3.71510476e-01 -3.16956043e-01 7.19838440e-01
1.24292180e-01 -1.76242039e-01 -6.61528885e-01 -4.84415621e-01
6.57088161e-01 5.69104664e-02 7.82108903e-01 7.47860730e-01
1.00129104e+00 -4.50010419e-01 3.62983793e-02 7.20577717e-01
1.74075949e+00 2.45427474e-01 9.85275030e-01 3.12367529e-01
4.86737758e-01 6.07156634e-01 6.12623453e-01 7.63453066e-01
-1.29450008e-01 4.94032294e-01 4.19896901e-01 -1.28109306e-01
-1.37764886e-01 1.64977938e-01 2.44767219e-01 7.43505001e-01
-1.98185250e-01 8.41906220e-02 -6.65623128e-01 6.59444034e-01
-1.66388035e+00 -1.11290359e+00 -1.24057263e-01 2.17221212e+00
7.03409672e-01 -1.76580086e-01 -2.97207385e-01 3.98937792e-01
7.16546535e-01 3.94978940e-01 -2.24464312e-01 -3.79768699e-01
-2.18953371e-01 7.27881730e-01 5.51110327e-01 6.51648343e-01
-9.36930656e-01 6.67716920e-01 5.55027485e+00 9.95219171e-01
-1.41093147e+00 1.06450282e-01 2.75866628e-01 -4.84240539e-02
-2.02698812e-01 1.77936926e-01 -4.42764819e-01 7.18634427e-01
3.82678658e-01 1.23192802e-01 3.51390004e-01 6.69944763e-01
5.08287191e-01 -4.24927473e-01 -3.93272400e-01 1.27121449e+00
2.74805337e-01 -1.23979211e+00 -3.20417464e-01 4.05910704e-03
8.67404759e-01 -1.90469816e-01 1.06744587e-01 -1.13022134e-01
-1.80468038e-01 -1.05026317e+00 7.14908242e-01 5.46580374e-01
8.49191129e-01 -5.54966211e-01 5.93558133e-01 3.93120080e-01
-1.33866918e+00 -3.17967720e-02 -4.32592541e-01 -2.02586219e-01
1.54194579e-01 1.01742923e+00 -1.60673231e-01 5.50578415e-01
8.06704283e-01 7.60912538e-01 -3.91515531e-02 1.06087601e+00
-6.63279071e-02 3.27477038e-01 -4.70987558e-01 4.02723223e-01
2.43565530e-01 -6.17917836e-01 7.80269802e-01 1.30437183e+00
8.15853536e-01 4.53215003e-01 2.85792559e-01 7.40172863e-01
1.67434722e-01 4.61452931e-01 -7.68231988e-01 1.34391516e-01
2.11153403e-01 1.03948700e+00 -8.28508914e-01 -1.51209787e-01
-5.71582437e-01 9.38222408e-01 -1.52724206e-01 3.54982883e-01
-5.88802099e-01 -5.83048820e-01 4.56035495e-01 2.89102674e-01
5.62930286e-01 -4.12989050e-01 -4.99487251e-01 -1.34064007e+00
1.98300615e-01 -1.02104723e+00 4.10154074e-01 -3.96244675e-01
-9.34501886e-01 5.12324154e-01 -8.99500400e-02 -1.46898365e+00
1.59081340e-01 -4.33733314e-01 -6.21618629e-01 1.01447392e+00
-1.58535635e+00 -8.83618832e-01 -3.24127644e-01 1.04958272e+00
6.18509889e-01 -1.50540352e-01 6.08421683e-01 5.61813474e-01
-6.15441620e-01 2.64726728e-01 3.28768671e-01 -8.53153691e-02
4.46202725e-01 -7.13094950e-01 -4.23120260e-01 1.10604548e+00
6.68592006e-02 5.83073795e-01 9.52485561e-01 -5.87586343e-01
-1.36403358e+00 -6.25876844e-01 1.04460120e+00 3.40797991e-01
4.39667374e-01 1.62134245e-01 -9.77921009e-01 5.46820164e-01
3.64247322e-01 2.50731826e-01 6.22167230e-01 -4.41451401e-01
-2.03028426e-01 -2.26517081e-01 -1.49766588e+00 2.54352957e-01
6.51190460e-01 -3.26538622e-01 -3.72226238e-01 7.14843646e-02
1.03799373e-01 -3.75301749e-01 -8.63684595e-01 3.27739179e-01
4.24220473e-01 -1.18957853e+00 9.31899846e-01 -2.73421817e-02
2.68956363e-01 -5.24713695e-01 -5.84118426e-01 -9.21214223e-01
-3.56223196e-01 -6.73751414e-01 -1.05427779e-01 7.95186758e-01
5.13271540e-02 -5.35982668e-01 9.18556094e-01 2.39653274e-01
-1.94281623e-01 -6.99077606e-01 -1.29675746e+00 -4.42634970e-01
-2.81207800e-01 -3.88785928e-01 4.07953141e-03 1.10512745e+00
-7.05074668e-02 -2.73868501e-01 -5.39359093e-01 1.34269878e-01
8.22943032e-01 1.71328023e-01 3.26248586e-01 -8.55107903e-01
-2.25425035e-01 -3.38778764e-01 -4.01441127e-01 -9.50139642e-01
-1.69041693e-01 -6.39517903e-01 4.13680039e-02 -1.20346177e+00
-1.31531581e-01 -6.24331951e-01 -2.95304000e-01 4.34347391e-01
2.42999271e-01 4.97913688e-01 1.95888683e-01 5.13656020e-01
-1.24491900e-01 3.69845837e-01 1.15207958e+00 -1.09320931e-01
-3.00370753e-01 -1.85878053e-01 -6.49882317e-01 7.73074329e-01
8.83661747e-01 -5.60089588e-01 -2.31754631e-01 -5.35820007e-01
2.16326997e-01 7.20321089e-02 1.62231833e-01 -1.07669783e+00
3.71452510e-01 -2.09519193e-01 9.97625366e-02 -4.35872287e-01
5.31614780e-01 -1.12186193e+00 4.45529491e-01 4.86910105e-01
-1.88358929e-02 6.48056418e-02 -7.46631920e-02 4.71411765e-01
-6.74553931e-01 -6.14720404e-01 1.16036069e+00 -4.42678332e-01
-6.06562674e-01 6.45418884e-03 -3.39558601e-01 -3.23139966e-01
1.12427795e+00 -5.23968279e-01 3.15167755e-01 -3.22328418e-01
-9.13017809e-01 -2.47102350e-01 4.34006602e-01 -2.14620918e-01
8.32630694e-01 -1.13982391e+00 -8.25689077e-01 5.93886852e-01
-2.29765102e-01 -6.49834946e-02 5.46458185e-01 1.25915623e+00
-8.72668862e-01 1.27548382e-01 -3.93539548e-01 -3.59504670e-01
-1.13286984e+00 5.58995783e-01 1.71496361e-01 -3.03482682e-01
-6.39134347e-01 9.86960888e-01 -1.42586693e-01 1.25177309e-01
1.78530157e-01 -1.03101157e-01 -2.64828920e-01 2.46947631e-01
5.69963694e-01 4.88225162e-01 1.20935634e-01 -9.09655869e-01
-2.09292799e-01 8.07149470e-01 8.04391503e-02 -1.05355419e-01
1.38024068e+00 -2.54618406e-01 -6.22452378e-01 1.02218099e-01
1.28616381e+00 2.73517072e-01 -1.10334373e+00 -1.73725888e-01
-3.44338924e-01 -8.77220809e-01 2.30457947e-01 -1.55209526e-01
-1.35091305e+00 6.73065066e-01 5.06176233e-01 3.52228254e-01
1.66310191e+00 -4.38833535e-01 7.68800855e-01 -2.58638263e-02
3.13327581e-01 -1.16886902e+00 -1.24231547e-01 3.75842154e-01
1.11051846e+00 -1.12400687e+00 1.89913213e-01 -4.48831707e-01
-3.19912016e-01 1.14996183e+00 5.77538982e-02 -5.94813406e-01
9.30482090e-01 3.62728804e-01 3.43546085e-02 -3.17057669e-01
-6.83711693e-02 -2.27019861e-01 -8.07299279e-03 2.26006463e-01
4.36372876e-01 -1.98111132e-01 -8.60406101e-01 2.37914681e-01
5.69924749e-02 9.64332521e-02 2.67052770e-01 1.29764628e+00
-5.43767810e-01 -1.20915246e+00 -9.09274518e-01 3.15734774e-01
-6.84611797e-01 -2.00178176e-01 1.00570850e-01 4.49537545e-01
4.93379474e-01 8.62154722e-01 -2.80302078e-01 -1.21425405e-01
2.77231455e-01 -7.18974620e-02 4.54522818e-01 -4.91065919e-01
-4.98489201e-01 4.98683691e-01 -3.25093389e-01 -5.90384960e-01
-8.08630526e-01 -8.31647336e-01 -1.12240362e+00 -3.44287097e-01
-1.18961178e-01 1.46550909e-01 6.96080267e-01 4.82507199e-01
1.35300189e-01 1.60279423e-01 6.49540901e-01 -9.66028571e-01
-3.17900538e-01 -7.58620739e-01 -9.04899359e-01 3.91165197e-01
3.52378845e-01 -5.39180636e-01 -4.93807524e-01 3.01457584e-01]
|
[11.52099895477295, -2.2462193965911865]
|
17fb9bd7-f458-49e2-a2dd-c474d87ad3cf
|
uncertainty-aware-camera-pose-estimation-from-1
|
2107.03890
| null |
https://arxiv.org/abs/2107.03890v1
|
https://arxiv.org/pdf/2107.03890v1.pdf
|
Uncertainty-Aware Camera Pose Estimation from Points and Lines
|
Perspective-n-Point-and-Line (P$n$PL) algorithms aim at fast, accurate, and robust camera localization with respect to a 3D model from 2D-3D feature correspondences, being a major part of modern robotic and AR/VR systems. Current point-based pose estimation methods use only 2D feature detection uncertainties, and the line-based methods do not take uncertainties into account. In our setup, both 3D coordinates and 2D projections of the features are considered uncertain. We propose PnP(L) solvers based on EPnP and DLS for the uncertainty-aware pose estimation. We also modify motion-only bundle adjustment to take 3D uncertainties into account. We perform exhaustive synthetic and real experiments on two different visual odometry datasets. The new PnP(L) methods outperform the state-of-the-art on real data in isolation, showing an increase in mean translation accuracy by 18% on a representative subset of KITTI, while the new uncertain refinement improves pose accuracy for most of the solvers, e.g. decreasing mean translation error for the EPnP by 16% compared to the standard refinement on the same dataset. The code is available at https://alexandervakhitov.github.io/uncertain-pnp/.
|
['Francesc Moreno-Noguer', 'Antonio Agudo', 'Luis Ferraz Colomina', 'Alexander Vakhitov']
|
2021-07-08
|
uncertainty-aware-camera-pose-estimation-from
|
http://openaccess.thecvf.com//content/CVPR2021/html/Vakhitov_Uncertainty-Aware_Camera_Pose_Estimation_From_Points_and_Lines_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Vakhitov_Uncertainty-Aware_Camera_Pose_Estimation_From_Points_and_Lines_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['camera-localization']
|
['computer-vision']
|
[-4.39179242e-01 -7.98207745e-02 2.51376741e-02 -2.21903324e-01
-7.70264447e-01 -8.20427537e-01 6.67099774e-01 -5.11699310e-03
-4.96290982e-01 6.89170361e-01 -2.36195326e-01 2.11060256e-01
-1.94664568e-01 -5.64848185e-01 -1.07212293e+00 -4.49931026e-01
5.49938828e-02 1.12554038e+00 3.50471735e-01 -2.95831442e-01
3.04116994e-01 8.34569335e-01 -1.28265536e+00 -7.93535292e-01
7.61221170e-01 1.01870477e+00 1.24076612e-01 4.25446987e-01
2.29785725e-01 7.59982504e-04 -1.01878814e-01 -1.83863670e-01
6.74973011e-01 3.00630301e-01 -2.45451674e-01 9.34524238e-02
7.15553820e-01 -1.68847218e-01 -3.61601055e-01 1.21642172e+00
5.18504083e-01 -1.65814115e-03 4.17201668e-01 -1.28911734e+00
-1.09247692e-01 -8.58888030e-02 -6.20894909e-01 -4.07942027e-01
8.06827068e-01 1.68496415e-01 5.63883960e-01 -1.42068398e+00
8.98154259e-01 1.07946825e+00 1.12299109e+00 4.22837548e-02
-1.19150710e+00 -4.85328615e-01 -5.11789657e-02 1.45993337e-01
-2.04575396e+00 -2.16897488e-01 5.88303506e-01 -6.19547188e-01
8.10759246e-01 1.15218684e-01 7.55454421e-01 7.30357766e-01
4.86197799e-01 2.90549189e-01 8.13844621e-01 -1.90204278e-01
1.14619032e-01 6.53842613e-02 -1.17798701e-01 7.54854679e-01
7.06402481e-01 1.91630661e-01 -5.21627009e-01 -1.24753334e-01
1.03847969e+00 -1.44256204e-02 -3.10723603e-01 -1.29808080e+00
-1.61383450e+00 7.17101336e-01 4.21035260e-01 -3.81674349e-01
-2.11271867e-01 3.15645367e-01 -8.52788053e-03 -1.26263602e-02
3.60727012e-01 4.86714870e-01 -5.51305413e-01 -2.59790331e-01
-5.51545978e-01 4.74019110e-01 6.47039115e-01 1.50316000e+00
9.23340201e-01 7.44598210e-02 4.92204487e-01 3.64742696e-01
5.87595105e-01 1.34241819e+00 1.57846183e-01 -1.35403895e+00
3.98514837e-01 5.42773366e-01 5.95233560e-01 -1.09831297e+00
-7.68448353e-01 -3.91747564e-01 -6.79250896e-01 3.94147575e-01
3.35551292e-01 -1.31738326e-02 -8.54082465e-01 1.35865033e+00
5.45565367e-01 3.57530862e-02 -6.33375123e-02 1.04745758e+00
4.23389584e-01 4.44485486e-01 -8.88473272e-01 -2.99823955e-02
1.04668880e+00 -6.60899580e-01 -4.64447945e-01 -4.11974519e-01
4.50124919e-01 -1.01847124e+00 5.13566673e-01 5.25173008e-01
-8.51495504e-01 -4.17178988e-01 -1.24439514e+00 -8.94791633e-02
4.35501039e-02 2.87243694e-01 2.35646695e-01 2.85394430e-01
-8.73117089e-01 5.83488524e-01 -8.74966741e-01 -4.45039392e-01
-2.05982760e-01 6.08823299e-01 -8.32148612e-01 -2.13769898e-01
-7.87030339e-01 1.22487152e+00 1.88039586e-01 3.42299968e-01
-5.97600043e-01 -6.54869378e-01 -1.16813326e+00 -4.88439173e-01
6.73411667e-01 -9.72497940e-01 1.22679472e+00 -3.02844584e-01
-1.76166511e+00 6.28047943e-01 -1.94104090e-01 -3.82849067e-01
8.70243967e-01 -6.60095096e-01 8.40018988e-02 -5.85800558e-02
6.59762546e-02 7.44331419e-01 7.13519931e-01 -1.41543925e+00
-2.72667289e-01 -3.73056203e-01 -2.90025882e-02 5.34710050e-01
5.78215539e-01 -6.04137182e-01 -6.37382030e-01 -1.21839836e-01
8.60136807e-01 -1.52462196e+00 -5.18125892e-01 3.37660700e-01
-3.27385753e-01 3.77828121e-01 5.43403447e-01 -4.01639253e-01
2.84178585e-01 -1.80675757e+00 5.74042857e-01 2.68146932e-01
5.84612601e-02 -1.03682026e-01 2.15971291e-01 3.52007568e-01
2.15289935e-01 -3.38694572e-01 -1.67064860e-01 -7.43522584e-01
5.68412952e-02 4.15932328e-01 -8.35432708e-02 1.22574878e+00
-1.78270265e-01 4.56808925e-01 -7.61467397e-01 -2.44819120e-01
7.06010997e-01 5.96352816e-01 -3.91465306e-01 -2.20154211e-01
-1.46933421e-01 7.15072036e-01 -2.01835886e-01 6.17999852e-01
9.71868634e-01 2.79154778e-01 -2.33083397e-01 -2.50220984e-01
-5.02983749e-01 4.54030633e-02 -1.76582444e+00 2.13847852e+00
-3.21066022e-01 4.67613459e-01 1.70098603e-01 -2.08599672e-01
9.76941347e-01 1.35291785e-01 4.78454024e-01 -3.53980631e-01
2.53930002e-01 5.75021863e-01 -1.38440579e-01 1.24103032e-01
8.77937615e-01 1.28225267e-01 -3.16620678e-01 -2.72871017e-01
-7.88831189e-02 -8.93760026e-01 -7.17181116e-02 3.19526121e-02
7.72328973e-01 5.06260216e-01 5.76019406e-01 -3.51937026e-01
6.36470199e-01 3.23961407e-01 9.76357579e-01 2.85651118e-01
1.03973597e-02 1.06854320e+00 1.42536342e-01 -1.77992016e-01
-1.23895824e+00 -1.04118979e+00 -3.48266989e-01 -1.20000698e-01
7.40781069e-01 -2.82918304e-01 -3.64444762e-01 -2.71921903e-01
5.17056346e-01 5.57456434e-01 -2.73268104e-01 3.28986906e-03
-4.68774498e-01 -2.03349173e-01 9.15089920e-02 3.13415349e-01
4.16234046e-01 -2.61493891e-01 -5.14611602e-01 1.14027988e-02
2.37824619e-02 -1.33383441e+00 -2.45727196e-01 -1.93729654e-01
-8.49138856e-01 -1.13737154e+00 -5.83308339e-01 -2.38314837e-01
7.15452611e-01 3.23027492e-01 6.88793123e-01 -3.82088512e-01
2.67368015e-02 6.64493322e-01 -2.34784603e-01 -3.60392481e-01
6.49353070e-03 -1.79395810e-01 7.62935281e-01 -3.37480128e-01
-1.97908267e-01 -3.43142748e-01 -2.82409549e-01 8.59743178e-01
-2.42314905e-01 6.72835633e-02 3.67604524e-01 6.43677890e-01
9.77646232e-01 -3.92084688e-01 -3.27444524e-01 -3.45573276e-01
-1.94947019e-01 -2.10787460e-01 -1.25636053e+00 -3.06210786e-01
-4.86254364e-01 5.75378016e-02 2.89528638e-01 -3.51842135e-01
-8.17899048e-01 7.04681575e-01 -6.30785674e-02 -8.67082536e-01
1.28652066e-01 4.86766398e-01 -1.45308897e-01 -6.13811374e-01
6.58212543e-01 -1.94998703e-03 8.23259577e-02 -4.31521147e-01
2.05184400e-01 1.84832573e-01 4.67967451e-01 -4.89064276e-01
1.14369118e+00 6.81230247e-01 5.90162992e-01 -7.25233972e-01
-3.71394336e-01 -6.80607617e-01 -1.09577537e+00 -1.86462834e-01
5.72284698e-01 -1.18122137e+00 -5.49095750e-01 6.25085175e-01
-1.30359709e+00 -8.47941414e-02 -4.32906061e-01 1.08300626e+00
-8.59470546e-01 6.41465724e-01 -1.57411128e-01 -6.74621761e-01
-1.89217948e-03 -1.67588818e+00 1.28015864e+00 1.62175728e-03
-7.29112327e-02 -6.10222995e-01 3.76591682e-01 7.73234665e-03
8.57562013e-03 3.61895204e-01 8.84072203e-03 -8.62421766e-02
-8.19045126e-01 -5.68329871e-01 3.90732810e-02 5.08745685e-02
-1.69901162e-01 1.55045167e-01 -5.57576120e-01 -4.59546924e-01
-4.96326312e-02 1.82365164e-01 3.90053570e-01 2.22611561e-01
1.81500137e-01 2.41550446e-01 -3.71377707e-01 7.85619378e-01
1.68995190e+00 -1.96171403e-01 4.39079702e-01 5.36549687e-01
8.59512091e-01 2.65395015e-01 1.14565325e+00 5.40052593e-01
6.35560334e-01 1.28261662e+00 1.13392591e+00 4.07916039e-01
1.59285933e-01 -1.81993201e-01 4.45527434e-01 7.33898163e-01
-2.84159899e-01 2.64967114e-01 -1.13155031e+00 4.66196716e-01
-2.15182447e+00 -1.72904745e-01 -5.94644368e-01 2.69181585e+00
3.03343892e-01 1.32581115e-01 -2.61763781e-01 -1.55367702e-01
6.04845524e-01 -6.12161122e-02 -6.35346413e-01 4.39227410e-02
1.77604388e-02 -2.86134690e-01 1.17175031e+00 9.84730422e-01
-7.74116993e-01 9.34842169e-01 4.74500799e+00 3.35968375e-01
-9.80262816e-01 1.20544732e-01 -4.68774259e-01 -1.54711381e-01
1.52497059e-02 3.36939573e-01 -1.28796172e+00 1.05020463e-01
7.46486127e-01 -2.31300980e-01 1.45750880e-01 1.10955441e+00
5.07150292e-02 -3.67196441e-01 -1.01781797e+00 1.23598325e+00
1.73074648e-01 -1.26258564e+00 -3.88617426e-01 3.14322323e-01
7.17626810e-01 5.48405290e-01 -2.58836836e-01 4.66423519e-02
1.44145161e-01 -4.86385554e-01 1.21027410e+00 7.73143589e-01
6.66170657e-01 -7.46342182e-01 1.03386605e+00 6.34265363e-01
-1.22973454e+00 2.72555470e-01 -5.94586968e-01 -1.37619361e-01
4.66520488e-01 8.32202911e-01 -7.57320344e-01 1.05076873e+00
6.94998980e-01 8.80802572e-01 -4.66561854e-01 1.28597462e+00
-3.88490856e-01 -1.38040125e-01 -9.26445782e-01 2.41397142e-01
-5.36765903e-03 -4.31002617e-01 1.10826468e+00 5.22134304e-01
7.60828793e-01 -5.07511459e-02 1.57422140e-01 5.75118244e-01
2.48623326e-01 -1.59059376e-01 -7.10273504e-01 6.56893313e-01
4.70957160e-01 1.23953199e+00 -4.74346846e-01 1.57206394e-02
-1.34332195e-01 8.58669579e-01 -7.90656805e-02 -1.51099861e-01
-9.05588984e-01 -1.65779367e-01 9.39213157e-01 1.52007341e-01
3.14865530e-01 -1.04298210e+00 -2.47639537e-01 -1.31632614e+00
3.05843890e-01 -4.85002339e-01 -2.03926519e-01 -1.08382702e+00
-6.95737720e-01 5.93065441e-01 2.60308504e-01 -1.88440943e+00
-4.80391771e-01 -7.74017453e-01 8.40061754e-02 9.03717875e-01
-1.31236434e+00 -9.69526947e-01 -4.86905545e-01 4.03237522e-01
4.83879894e-01 2.85219789e-01 6.79143131e-01 2.65853815e-02
-2.42914647e-01 6.81612045e-02 2.15445831e-01 -2.37660110e-01
8.93978238e-01 -1.09936762e+00 5.67284465e-01 9.56187427e-01
1.56274736e-01 5.18409491e-01 1.05740452e+00 -7.46982753e-01
-2.01081061e+00 -8.48283052e-01 5.09180307e-01 -8.14195991e-01
5.66590726e-01 -4.40610200e-01 -5.45855880e-01 1.05492496e+00
-1.92170814e-01 4.22914058e-01 -2.03613132e-01 -1.75771385e-01
-2.54028767e-01 -7.51693249e-02 -1.43632936e+00 4.36655670e-01
1.12392390e+00 -7.28334561e-02 -4.60506588e-01 3.18928838e-01
8.11838746e-01 -1.24109399e+00 -1.02238250e+00 6.71108246e-01
5.98421812e-01 -9.22431111e-01 1.11356831e+00 4.71459746e-01
-3.95632803e-01 -8.86061132e-01 -4.56912190e-01 -1.50202370e+00
-1.21783324e-01 -4.91225839e-01 -4.80353646e-02 9.07283008e-01
1.82664752e-01 -1.02184689e+00 7.21114874e-01 3.54343086e-01
-3.11088264e-01 -2.96985298e-01 -1.23039281e+00 -1.16010380e+00
-2.54951954e-01 -6.83827996e-01 3.98794979e-01 5.98795474e-01
-2.43598655e-01 1.21400073e-01 -2.86249489e-01 8.80653083e-01
9.92346168e-01 6.88763708e-03 1.33486819e+00 -1.35225892e+00
-1.05929919e-01 -6.75661415e-02 -9.80261087e-01 -1.12922883e+00
2.92382669e-02 -3.66897583e-01 2.02966422e-01 -1.44358182e+00
-4.63148981e-01 -4.44728494e-01 5.53175390e-01 7.57214427e-02
3.69540572e-01 2.71866292e-01 3.75886321e-01 3.69050026e-01
-4.67007428e-01 6.96163654e-01 7.78944552e-01 1.33442238e-01
-4.94255833e-02 1.68869764e-01 3.83497477e-02 1.12989819e+00
5.57203233e-01 -4.38145220e-01 -1.43104821e-01 -5.42592466e-01
4.54231948e-01 2.09170088e-01 4.12155449e-01 -1.43917894e+00
4.45974588e-01 -7.03101829e-02 2.59529173e-01 -9.68704820e-01
9.78627920e-01 -1.14175093e+00 7.27984488e-01 4.45927024e-01
5.05551994e-01 3.12637269e-01 3.81536692e-01 5.79988956e-01
-1.17890909e-01 -2.09908172e-01 5.84032357e-01 -1.31652117e-01
-8.83037925e-01 3.96916658e-01 -4.76222709e-02 -4.63344783e-01
1.13572538e+00 -3.86276871e-01 -2.73568511e-01 -5.13054729e-01
-5.41074812e-01 1.68151394e-01 1.27148068e+00 3.18565905e-01
6.79485321e-01 -1.31898475e+00 -5.41193664e-01 1.83157697e-01
4.88324285e-01 8.72487366e-01 1.49385095e-01 1.23205841e+00
-9.31379735e-01 5.45819163e-01 -1.89486623e-01 -1.34622157e+00
-1.19156516e+00 2.98064083e-01 3.93067956e-01 5.87462224e-02
-5.97630799e-01 6.54793203e-01 -1.40736371e-01 -1.12050903e+00
-2.92287190e-02 -6.81973815e-01 2.44934916e-01 -3.08336198e-01
5.65023012e-02 6.97799087e-01 1.91791311e-01 -1.14557934e+00
-6.66375279e-01 1.35042477e+00 2.47119099e-01 -3.04459184e-01
1.20591044e+00 -4.11207438e-01 4.50835638e-02 5.29940188e-01
9.47331429e-01 4.00850475e-01 -1.56035292e+00 -2.13452503e-01
-8.24974626e-02 -5.85429370e-01 2.40691528e-02 -4.09256369e-01
-8.57503355e-01 5.86871862e-01 5.65450847e-01 -3.69016379e-01
5.60527265e-01 -1.46618202e-01 3.15019757e-01 4.40695047e-01
1.22944999e+00 -6.11463606e-01 -5.88087618e-01 8.12339306e-01
1.22630298e+00 -1.25954604e+00 6.67153716e-01 -9.02547479e-01
-5.53522110e-01 1.02639949e+00 4.84609216e-01 -5.17116845e-01
4.06294346e-01 1.76090434e-01 -1.14344237e-02 7.64673129e-02
-2.30042621e-01 -5.93825951e-02 2.37422824e-01 5.13114393e-01
-1.49978042e-01 2.82615107e-02 -1.12021282e-01 -4.10620049e-02
-2.47605979e-01 -8.82911533e-02 7.54512608e-01 1.07140219e+00
-3.50828886e-01 -1.08604133e+00 -8.37440670e-01 -1.33313566e-01
3.50303084e-01 2.48768285e-01 -1.14195831e-01 1.16984010e+00
-1.19615588e-02 4.86148745e-01 1.25950918e-01 -5.32561779e-01
7.11976349e-01 -2.44169369e-01 8.44830871e-01 -6.30118370e-01
-1.53202087e-01 5.19804433e-02 8.95051286e-02 -1.12284184e+00
-2.25104541e-01 -1.17248535e+00 -1.50896895e+00 -3.20003927e-01
-4.32370186e-01 4.10879999e-02 1.21896386e+00 7.35406280e-01
6.42902434e-01 -1.91312861e-02 4.13238704e-01 -1.52637160e+00
-6.41470551e-01 -8.99343491e-01 -6.08983755e-01 -1.93807244e-01
4.03343350e-01 -1.27168787e+00 -5.45839906e-01 -4.86038089e-01]
|
[7.3489580154418945, -2.161762237548828]
|
a143122a-71f1-4423-8255-514a6169c34d
|
shct-a-successively-hierarchical-conditional
| null | null |
https://openreview.net/forum?id=ZCmUqcIjuGc
|
https://openreview.net/pdf?id=ZCmUqcIjuGc
|
SHCT: A Successively Hierarchical Conditional Transformer for Controllable Paraphrase Generation
|
Paraphrase generation has consistently been a challenging area in the field of NLP. Despite the considerable achievements made by previous work, existing methods lack a flexible way to include multiple controllable attributes to enhance the diversity of paraphrased sentences. To overcome this challenge, we propose a Successively Hierarchical Conditional Transformer(SHCT) to tackle this task. SHCT is based on a combination of Conditional Variational AutoEncoder(CVAE) with Transformer framework to benefit from the advantage of generating diversified words. More specifically, our SHCT deploys multi-head attention and dynamic memory mechanism to keep the interaction between each of the attributes and the corresponding encoder layer hidden state. To address the problem of absorbing flexible attributes, we apply a hierarchical structure to our SHCT which enables the framework to couple the CVAE latent variables with encoder layer hidden states successively. In addition, Our SHCT is trained by minimizing a tailor-designed loss for producing paraphrased sentences as required. Finally, We conduct extensive experiments to substantiate the validity and effectiveness of our proposed model. The results show that SHCT significantly outperforms the existing state-of-the-art approaches and generates more diverse paraphrased sentences.
|
['Anonymous']
|
2021-11-16
| null | null | null |
acl-arr-november-2021-11
|
['paraphrase-generation', 'paraphrase-generation']
|
['computer-code', 'natural-language-processing']
|
[ 2.64567174e-02 -1.26697332e-01 -3.28463241e-02 -3.57998163e-01
-8.11586082e-01 -3.73836011e-01 6.03887975e-01 -4.38073784e-01
-1.04874149e-01 9.05478418e-01 4.86551166e-01 -1.34876788e-01
5.91351911e-02 -7.86027908e-01 -8.81329954e-01 -7.47752488e-01
6.91296518e-01 3.89379084e-01 -8.59817676e-03 -2.51384854e-01
2.55671859e-01 4.38122265e-02 -1.40117800e+00 2.90632367e-01
1.27125502e+00 7.09075928e-01 6.45962417e-01 1.71113744e-01
-2.30495170e-01 6.90792799e-01 -5.10547042e-01 -7.13916123e-01
4.06686775e-02 -7.16603994e-01 -6.93326533e-01 1.22067772e-01
1.06012024e-01 -2.55628854e-01 -3.05023283e-01 9.44396853e-01
6.84781373e-01 2.05269948e-01 6.11735463e-01 -1.19709527e+00
-1.22647452e+00 8.16468418e-01 -3.83083433e-01 1.63314953e-01
3.18248570e-01 1.96631417e-01 1.23931634e+00 -1.10826790e+00
4.02131975e-01 1.35794437e+00 3.79968137e-01 6.81950927e-01
-1.29772615e+00 -7.85374224e-01 3.16352427e-01 3.36897016e-01
-1.20450437e+00 -3.73676568e-01 1.06484771e+00 -2.88903505e-01
8.62843573e-01 1.22323655e-01 5.08975804e-01 1.55789495e+00
1.54541939e-01 1.03723907e+00 9.82749701e-01 -2.98823923e-01
1.76529095e-01 3.76524568e-01 5.87897934e-02 5.58968961e-01
-1.39095798e-01 -2.45263070e-01 -4.77928907e-01 -9.71325710e-02
5.76181412e-01 2.33925477e-01 -3.63686860e-01 -3.20276320e-01
-9.97973979e-01 1.04229224e+00 4.53201175e-01 1.90134585e-01
-4.59802181e-01 -8.08104798e-02 3.97691458e-01 2.30934128e-01
3.15914780e-01 4.84398633e-01 -2.71847665e-01 2.29327362e-02
-1.07005060e+00 1.73411816e-01 4.51481879e-01 1.09843445e+00
5.97921133e-01 1.69906557e-01 -7.56778181e-01 1.06280529e+00
1.68455467e-01 2.83069581e-01 7.71202087e-01 -8.11862350e-01
7.43100107e-01 6.12461865e-01 5.67742670e-03 -6.69114888e-01
2.62294024e-01 -7.89188802e-01 -9.94940341e-01 -4.02990758e-01
-4.65037048e-01 -7.22188801e-02 -7.88308561e-01 2.04312158e+00
1.14611581e-01 1.56296551e-01 1.85188860e-01 7.30208278e-01
6.81792080e-01 1.07948768e+00 1.18636385e-01 -2.48547971e-01
1.06829989e+00 -1.27063155e+00 -7.55838692e-01 -1.62369728e-01
1.93379223e-01 -5.34858465e-01 1.47949195e+00 6.32763701e-03
-1.32015336e+00 -7.13023901e-01 -1.05354440e+00 -3.13276798e-01
8.65917653e-02 2.08050236e-01 3.62293720e-01 2.89005458e-01
-7.88293302e-01 4.18392956e-01 -7.32758224e-01 -7.76360917e-04
3.73784989e-01 2.18139783e-01 -1.20338283e-01 -1.99974026e-03
-1.56234789e+00 8.25661659e-01 4.51855600e-01 2.52681106e-01
-8.79127324e-01 -6.05928957e-01 -9.78954971e-01 5.19827068e-01
1.05025887e-01 -1.25158465e+00 1.18634963e+00 -8.84420753e-01
-1.83236182e+00 3.79688084e-01 -4.63308215e-01 -3.68201911e-01
4.98872757e-01 -2.61355102e-01 4.69102412e-02 -1.25675360e-02
2.66983807e-01 5.58334470e-01 9.30226862e-01 -1.14797747e+00
-3.01884502e-01 -1.89762846e-01 6.50649741e-02 4.12948996e-01
-7.19825208e-01 -1.99920997e-01 -4.74140465e-01 -7.65005767e-01
-1.55134574e-01 -8.69141817e-01 -9.89878699e-02 -5.17970026e-01
-5.27556479e-01 -4.12359297e-01 6.43174469e-01 -7.40932345e-01
1.47170484e+00 -2.09852862e+00 8.24750602e-01 -3.39889884e-01
-3.96309309e-02 3.52023453e-01 -8.59690085e-02 6.05324447e-01
1.14615723e-01 8.31502303e-02 -4.77484375e-01 -8.05544853e-01
1.47833839e-01 3.01680237e-01 -5.49649954e-01 -8.37136582e-02
3.03630382e-01 9.81069207e-01 -7.01149702e-01 -6.86283112e-01
2.22564474e-01 5.25879204e-01 -7.21794844e-01 4.79464412e-01
-2.70153373e-01 4.34825540e-01 -6.84545755e-01 3.58409882e-01
6.59237564e-01 -3.74736845e-01 -6.87933415e-02 -6.45615533e-02
6.75023794e-02 3.06126773e-01 -6.87365234e-01 1.72857225e+00
-7.17972696e-01 2.88056552e-01 -1.35704711e-01 -9.34770703e-01
9.29670930e-01 2.93011487e-01 6.05446883e-02 -5.44789433e-01
5.60986437e-02 5.89238331e-02 -3.53204608e-01 -5.75432479e-01
6.11401439e-01 -4.39579844e-01 -1.54950753e-01 2.72913814e-01
1.24499030e-01 -3.75669189e-02 1.35243929e-03 4.30292755e-01
7.74373233e-01 2.14494944e-01 1.99516620e-02 3.86533253e-02
8.66544068e-01 -3.61143380e-01 7.94223785e-01 6.42244220e-01
-4.93523031e-02 6.69105291e-01 4.28312123e-01 6.73089409e-03
-1.24428046e+00 -1.12959313e+00 4.07583453e-02 8.67441833e-01
1.02074347e-01 -2.75285304e-01 -8.24274361e-01 -6.00955725e-01
-1.60974845e-01 1.00814366e+00 -5.84220827e-01 -3.83863866e-01
-5.61411202e-01 -6.64012551e-01 2.94057161e-01 6.53236270e-01
7.90980637e-01 -1.35518074e+00 -2.58050829e-01 2.54912674e-01
-6.74486458e-01 -9.44663763e-01 -5.93819857e-01 -1.26151219e-01
-7.00110972e-01 -4.40748543e-01 -8.95517468e-01 -9.49339569e-01
4.31608707e-01 2.05353171e-01 1.01462293e+00 -1.34883657e-01
1.68624371e-01 -1.01864003e-01 -4.92392361e-01 -1.33206218e-01
-5.00249803e-01 5.22643685e-01 -1.90715402e-01 1.77001014e-01
2.21043825e-04 -8.39363217e-01 -7.30780482e-01 -5.17358957e-03
-1.02340555e+00 4.25644368e-01 8.27919960e-01 1.11241388e+00
4.62345630e-01 -1.82116538e-01 7.63788223e-01 -7.98622251e-01
9.99763072e-01 -7.68040657e-01 -2.71963120e-01 3.80963653e-01
-4.68569368e-01 4.34977591e-01 1.05767298e+00 -3.49829495e-01
-1.30417681e+00 -1.39701366e-01 -3.57137144e-01 -6.85983777e-01
2.53030032e-01 5.48779190e-01 -3.50766510e-01 5.99858880e-01
8.52924660e-02 8.00178289e-01 -1.34291485e-01 -4.40759927e-01
3.44747663e-01 8.64025831e-01 3.83540303e-01 -5.19348979e-01
6.49354279e-01 9.03782025e-02 -3.04527313e-01 -4.23092097e-01
-9.49501157e-01 -4.50517721e-02 -2.75683850e-01 6.49039596e-02
8.97497714e-01 -9.76609766e-01 -4.83092427e-01 4.41008747e-01
-1.24532652e+00 -1.51509479e-01 -6.04773611e-02 1.62594929e-01
-4.99348581e-01 5.36269546e-01 -7.09346592e-01 -6.80408120e-01
-6.94355726e-01 -1.30654883e+00 1.21293962e+00 2.39654511e-01
-4.74797636e-02 -7.52334714e-01 1.01082049e-01 6.33333147e-01
5.53936541e-01 -1.85530391e-02 1.15014219e+00 -4.54965711e-01
-7.26002395e-01 -3.83258089e-02 -1.07958794e-01 5.15137434e-01
7.21397698e-02 -2.26116225e-01 -6.53191924e-01 -3.80268365e-01
1.61854416e-01 -5.21698117e-01 1.06605220e+00 2.03470096e-01
1.21832347e+00 -5.22140265e-01 -2.94749081e-01 5.21175504e-01
1.24772131e+00 4.08360176e-02 7.39907444e-01 3.31771791e-01
6.64037704e-01 3.52170408e-01 3.58609080e-01 4.21930879e-01
5.27898669e-01 7.38815546e-01 2.08729878e-01 7.58768618e-02
6.04158789e-02 -5.87960303e-01 4.80159253e-01 1.21951091e+00
2.83785760e-01 -3.76202494e-01 -3.66766661e-01 5.95439613e-01
-1.81300604e+00 -1.19848633e+00 1.63126022e-01 1.95243526e+00
1.10637391e+00 4.33570072e-02 -1.55903652e-01 -1.88577697e-02
8.57442498e-01 3.42305869e-01 -5.71361899e-01 -3.69361997e-01
-8.82200897e-02 1.01324424e-01 -1.28406212e-01 4.95957971e-01
-7.08537161e-01 1.06167185e+00 5.24354124e+00 9.95778024e-01
-9.43203866e-01 1.36559010e-01 3.85264546e-01 -9.49959308e-02
-7.29672134e-01 6.34909123e-02 -7.94422150e-01 1.00480652e+00
6.76335514e-01 -2.20537275e-01 4.65870082e-01 8.57784390e-01
2.79163510e-01 3.46574336e-01 -9.36275840e-01 6.53874815e-01
1.67228609e-01 -1.18518424e+00 5.69924116e-01 -1.57015845e-01
8.01844656e-01 -3.53862643e-01 2.86787421e-01 6.97266757e-01
2.29569182e-01 -8.56037080e-01 7.33410656e-01 5.52029490e-01
4.75527585e-01 -8.16546142e-01 6.53501511e-01 5.77207446e-01
-1.00251448e+00 -3.18216801e-01 -5.94075918e-01 3.86798531e-02
3.56653839e-01 4.43072915e-01 -5.23582518e-01 6.66057944e-01
4.84833121e-01 6.42902374e-01 -5.17445803e-01 6.96992517e-01
-3.08986336e-01 6.16892397e-01 9.81075391e-02 -2.14886442e-01
2.83003747e-01 -2.23502696e-01 5.53042471e-01 9.78109717e-01
3.87868315e-01 5.87226963e-03 5.23139313e-02 1.21552360e+00
-2.37216368e-01 8.59400183e-02 -4.67045963e-01 9.87771228e-02
7.77961791e-01 9.85370874e-01 2.96906959e-02 -4.18165267e-01
-2.73911148e-01 1.35637522e+00 6.93594337e-01 4.09701973e-01
-1.00785196e+00 -4.83835638e-01 3.36075723e-01 -5.11556938e-02
5.88683248e-01 -4.24895845e-02 -1.50127500e-01 -1.50549626e+00
3.22076410e-01 -8.42842340e-01 1.47980466e-01 -1.01552176e+00
-1.44385886e+00 8.98285568e-01 -5.36572523e-02 -1.06503129e+00
-2.88313866e-01 -4.81800409e-03 -7.59763777e-01 1.09043336e+00
-1.54346848e+00 -1.35718226e+00 -2.24208996e-01 6.15596473e-01
1.05167949e+00 -1.27788588e-01 6.14519119e-01 1.81569889e-01
-9.82130349e-01 7.18118548e-01 1.22097284e-01 -1.70744136e-01
5.50770581e-01 -9.93311703e-01 4.43875998e-01 9.17853534e-01
-1.35592237e-01 8.56264353e-01 7.58982837e-01 -6.22639418e-01
-1.14457321e+00 -1.24942005e+00 1.11690700e+00 -3.69950861e-01
4.58828241e-01 -4.46541220e-01 -9.61255133e-01 8.16018283e-01
5.82189202e-01 -5.35268247e-01 4.86785978e-01 -8.09782520e-02
-1.96310103e-01 -8.08714181e-02 -7.47751474e-01 7.01265156e-01
9.00123179e-01 -5.18064916e-01 -9.09365237e-01 1.25420898e-01
1.08895922e+00 -2.31846601e-01 -5.41610956e-01 4.55184251e-01
3.08725804e-01 -9.53405917e-01 8.77071619e-01 -5.74402928e-01
1.03803337e+00 -7.59884119e-02 -8.07234943e-02 -1.46504486e+00
-6.01661623e-01 -4.18322861e-01 -8.39236528e-02 1.62632251e+00
4.15984303e-01 -5.33313453e-01 7.23658383e-01 5.70980310e-01
-3.97460431e-01 -1.11028349e+00 -7.94316232e-01 -6.71103418e-01
3.11564267e-01 1.10474721e-01 6.35760903e-01 6.90515518e-01
-9.33677033e-02 8.86673868e-01 -7.88302839e-01 -4.03684564e-02
4.67981607e-01 4.31654632e-01 6.75034940e-01 -8.31160605e-01
-5.57335436e-01 -3.43609422e-01 1.46813527e-01 -1.29549336e+00
5.64694464e-01 -9.73359168e-01 -3.37524014e-03 -1.71384203e+00
7.22106814e-01 -2.04472959e-01 -3.33240777e-01 2.86658585e-01
-7.74489403e-01 -1.52334914e-01 3.33936125e-01 3.05848062e-01
-5.22935629e-01 1.38073146e+00 1.34910834e+00 -1.42576024e-01
-2.43091330e-01 9.05265957e-02 -8.44206929e-01 2.38696441e-01
9.45148051e-01 -4.22751993e-01 -7.22386479e-01 -6.56052947e-01
7.53169060e-02 3.03808630e-01 4.11459655e-01 -7.01272428e-01
1.79747522e-01 -1.43587276e-01 2.75929481e-01 -6.89815760e-01
4.23757613e-01 -5.21736741e-01 1.06875502e-01 2.59591907e-01
-5.46042383e-01 2.31827229e-01 -1.11252904e-01 6.01855695e-01
-3.39373559e-01 -2.35101610e-01 7.14120746e-01 -2.15841427e-01
-2.81989127e-01 2.42455870e-01 -2.95868188e-01 8.29345211e-02
8.34412396e-01 -2.58723572e-02 -7.35171139e-02 -3.81637752e-01
-3.82148743e-01 5.18388987e-01 3.52276325e-01 5.76870084e-01
7.39182532e-01 -1.36932790e+00 -7.65911996e-01 2.87273884e-01
6.61631813e-03 -8.83291066e-02 5.54321647e-01 4.91529018e-01
-1.43711254e-01 4.82808918e-01 -2.20367879e-01 -2.91104347e-01
-9.64080870e-01 8.30277860e-01 1.76659271e-01 -5.37828207e-01
-6.14837766e-01 7.84880817e-01 4.01586086e-01 -2.99963385e-01
9.98701379e-02 1.13023901e-02 -1.98400393e-01 -1.41550347e-01
9.72193927e-02 1.97926611e-01 -1.86323449e-01 -3.60868394e-01
-7.25867748e-02 2.64994234e-01 -4.42250818e-01 -1.84989646e-01
1.27524114e+00 -4.27019000e-01 -6.84993789e-02 3.31427127e-01
1.28208899e+00 -6.46719784e-02 -1.24016845e+00 -1.73268974e-01
-4.95230496e-01 -2.82802016e-01 -1.41090006e-01 -6.18156552e-01
-9.76164520e-01 1.01933992e+00 3.54660265e-02 -9.20659397e-03
1.13323247e+00 -5.81948124e-02 1.34567153e+00 1.63713649e-01
1.82460725e-01 -8.12487304e-01 3.83513480e-01 4.41629499e-01
1.04553139e+00 -9.67924953e-01 -3.23304951e-01 -2.34672159e-01
-1.04794621e+00 5.46227157e-01 7.83692420e-01 -1.89672083e-01
1.76293939e-01 -2.53528655e-01 -1.93336919e-01 9.15141627e-02
-9.10334647e-01 1.77139178e-01 5.00666611e-02 7.93744847e-02
2.58283466e-01 -1.56693593e-01 -5.17156065e-01 8.61734092e-01
-3.60973388e-01 1.46697676e-02 3.50014567e-01 7.13352799e-01
-3.67780030e-01 -1.29448771e+00 3.66412401e-02 2.63952672e-01
-3.40229601e-01 -4.22840029e-01 -3.19842666e-01 4.63033587e-01
-9.01754126e-02 7.60807872e-01 -2.30978087e-01 -3.24527800e-01
2.80246943e-01 2.51565844e-01 3.25766504e-01 -5.54448545e-01
-6.25719011e-01 -1.48155436e-01 -3.48063439e-01 -2.12676585e-01
-2.52258003e-01 -5.95677912e-01 -8.71122241e-01 -1.23093836e-01
-4.54575062e-01 4.52901661e-01 3.24589163e-01 9.36311066e-01
6.26588047e-01 6.38551116e-01 9.07922626e-01 -5.73026419e-01
-8.89932811e-01 -1.07110488e+00 -3.43615025e-01 6.23886883e-01
3.85703951e-01 -6.76318765e-01 -3.63755524e-01 -2.50627622e-02]
|
[11.743955612182617, 9.271878242492676]
|
677c0560-13a3-4276-b263-dc352a969d89
|
a-hybrid-approach-combining-statistical
| null | null |
https://aclanthology.org/W18-3728
|
https://aclanthology.org/W18-3728.pdf
|
A Hybrid Approach Combining Statistical Knowledge with Conditional Random Fields for Chinese Grammatical Error Detection
|
This paper presents a method of combining Conditional Random Fields (CRFs) model with a post-processing layer using Google n-grams statistical information tailored to detect word selection and word order errors made by learners of Chinese as Foreign Language (CFL). We describe the architecture of the model and its performance in the shared task of the ACL 2018 Workshop on Natural Language Processing Techniques for Educational Applications (NLPTEA). This hybrid approach yields comparably high false positive rate (FPR = 0.1274) and precision (Pd= 0.7519; Pi= 0.6311), but low recall (Rd = 0.3035; Ri = 0.1696 ) in grammatical error detection and identification tasks. Additional statistical information and linguistic rules can be added to enhance the model performance in the future.
|
['Chilin Shih', 'Yiyi Wang']
|
2018-07-01
| null | null | null |
ws-2018-7
|
['grammatical-error-detection']
|
['natural-language-processing']
|
[-1.06555469e-01 1.16723776e-01 7.78719559e-02 -6.32674694e-01
-8.95253718e-01 -4.87655908e-01 4.86228257e-01 7.94573963e-01
-1.03777373e+00 8.57614160e-01 2.57827342e-01 -8.04654419e-01
5.67175522e-02 -6.19761705e-01 -7.37678170e-01 2.90794726e-02
2.91958265e-02 2.57467687e-01 3.57278854e-01 -5.04959747e-02
6.82649553e-01 1.68203086e-01 -1.63221359e+00 4.37497973e-01
1.73887658e+00 4.63979155e-01 7.52332330e-01 9.29718733e-01
-3.18783730e-01 8.00689220e-01 -8.37229550e-01 -5.93087077e-01
-4.46318537e-01 1.31093040e-01 -1.12677777e+00 -5.14845967e-01
6.35798573e-01 4.30569798e-02 4.37234402e-01 1.26215100e+00
4.68623936e-01 4.63634819e-01 4.72626597e-01 -5.11520207e-01
-7.67789721e-01 9.01476681e-01 -2.49049179e-02 4.00719553e-01
8.94428551e-01 -1.43468827e-01 4.94201720e-01 -1.15399277e+00
5.33727169e-01 1.27920353e+00 8.39792430e-01 6.86387658e-01
-7.50217736e-01 -9.86056387e-01 1.77623153e-01 1.34087026e-01
-1.29472053e+00 -4.58056480e-01 -2.13460758e-01 -4.24296737e-01
1.64136887e+00 2.23744273e-01 4.21238273e-01 1.02782249e+00
3.61041039e-01 9.61492360e-01 1.41649544e+00 -1.21475327e+00
2.57054031e-01 3.65941614e-01 5.11198640e-01 8.42031062e-01
1.96200982e-01 -1.13400193e-02 -9.74484742e-01 2.58915156e-01
1.18242249e-01 -5.91988742e-01 2.74650566e-02 8.65170419e-01
-7.40953267e-01 6.76084399e-01 -1.79277152e-01 5.11001348e-01
-1.22782946e-01 -4.96496409e-01 1.35422140e-01 4.39289451e-01
5.30902922e-01 4.20123011e-01 -9.17825162e-01 -7.06913710e-01
-8.96674931e-01 4.15035307e-01 6.88957334e-01 1.54401660e+00
2.47532323e-01 -3.24817151e-02 -3.36045861e-01 9.40560222e-01
8.51380110e-01 5.10850310e-01 8.57889295e-01 -3.73396665e-01
6.09302998e-01 4.61930603e-01 6.20415285e-02 -4.62469399e-01
-3.14429402e-01 -1.38299063e-01 -2.58570582e-01 -3.63929242e-01
6.45205975e-01 -3.31212342e-01 -9.97384489e-01 1.50885546e+00
1.41021132e-01 1.16327837e-01 -3.14970091e-02 3.30258757e-01
1.21372521e+00 5.48095584e-01 9.84082043e-01 -4.85678250e-03
1.47429311e+00 -6.45465732e-01 -8.42480958e-01 -4.61853951e-01
1.21557963e+00 -1.26710856e+00 1.15634525e+00 6.79873943e-01
-1.19263399e+00 -5.06833196e-01 -6.41658843e-01 -2.06171468e-01
-5.21231711e-01 3.82918030e-01 4.86199528e-01 1.10134566e+00
-9.85619783e-01 3.74978662e-01 -9.09321964e-01 -4.25552428e-01
1.58131063e-01 2.22648248e-01 -2.20842361e-01 -2.72259533e-01
-1.22155154e+00 9.24749076e-01 3.39139402e-01 -3.07506859e-01
-3.50787193e-01 -9.43851709e-01 -9.10007954e-01 -1.59069344e-01
-1.86435133e-01 1.52006656e-01 1.46621358e+00 -4.60355788e-01
-1.67802596e+00 1.07805109e+00 -4.30681884e-01 -3.35930735e-01
1.90201327e-01 -8.34872127e-01 -7.54063010e-01 -3.06071311e-01
4.15177852e-01 4.75944579e-01 -1.98201798e-02 -4.50419724e-01
-1.24075258e+00 -3.45342487e-01 -4.57987309e-01 3.24891992e-02
-2.17103750e-01 8.65514159e-01 -1.03522152e-01 -7.45695949e-01
-3.30447331e-02 -6.92712367e-01 -1.69667155e-01 -8.91515255e-01
-2.30271623e-01 -1.10315478e+00 -2.62133688e-01 -1.23460448e+00
1.70483696e+00 -1.78082681e+00 -5.02366781e-01 2.71456748e-01
-3.60675365e-01 8.69977653e-01 -6.29265755e-02 3.95334810e-01
1.01043127e-01 4.62294072e-01 1.33475363e-02 -3.88678342e-01
-2.13868663e-01 -1.30030453e-01 8.62897411e-02 5.02675697e-02
3.10398102e-01 6.51889086e-01 -1.16889429e+00 -5.40138841e-01
1.13329276e-01 3.29091460e-01 -5.65255940e-01 3.51518542e-01
-1.89763531e-02 2.37867489e-01 -1.75020561e-01 4.35587734e-01
6.50565684e-01 5.62787235e-01 1.31670386e-01 8.90032828e-01
-7.45559573e-01 1.27551711e+00 -1.35919940e+00 1.56034553e+00
-7.17507720e-01 5.04333317e-01 -1.72847450e-01 -4.67190564e-01
9.99938786e-01 3.36189449e-01 -1.90436900e-01 -7.60173738e-01
-3.03944740e-02 2.15196177e-01 -7.45287165e-02 -6.43734872e-01
7.53796101e-01 1.71233758e-01 -1.20645039e-01 1.77507743e-01
6.37266755e-01 2.52567440e-01 4.24037784e-01 1.06013656e-01
1.00325751e+00 2.11215883e-01 3.92251343e-01 -8.11893702e-01
7.09620178e-01 -5.33790886e-02 6.57805502e-01 9.68904257e-01
-2.28806645e-01 7.55061805e-02 -2.45582834e-02 -8.04217011e-02
-4.55986202e-01 -8.68278205e-01 -5.82363427e-01 1.28253794e+00
-6.19293571e-01 -8.23877513e-01 -8.41309905e-01 -7.81313241e-01
-1.95106000e-01 1.26646864e+00 1.45314471e-03 -8.46710522e-04
-5.65699577e-01 -3.82824600e-01 6.34622157e-01 4.44056541e-01
2.86588997e-01 -1.40608311e+00 -3.94345760e-01 4.07461047e-01
-1.45857753e-02 -1.24239159e+00 -1.42594665e-01 2.16724485e-01
-6.50801301e-01 -7.59162784e-01 -2.20891610e-01 -1.17789686e+00
6.47953928e-01 -2.99313843e-01 1.29625237e+00 4.69255835e-01
1.11924987e-02 1.22479312e-01 -9.65186775e-01 -9.23482597e-01
-4.86945689e-01 5.66285849e-02 1.68268085e-01 -6.54416025e-01
1.22918606e+00 2.22628340e-02 2.88602952e-02 -2.97808737e-01
-5.05426228e-01 -1.12999514e-01 2.69461989e-01 7.04395592e-01
2.03261688e-01 -1.34647042e-01 4.19400781e-01 -1.08939373e+00
7.40501285e-01 -3.63514036e-01 -8.39630842e-01 5.01214385e-01
-8.57158720e-01 -1.48977712e-01 5.42500854e-01 -3.55571687e-01
-1.42647803e+00 5.00263721e-02 -8.78363907e-01 5.13273835e-01
-8.89994860e-01 4.85501111e-01 -5.62001765e-02 -4.18085195e-02
6.00839436e-01 1.01843365e-01 -6.69427931e-01 -9.06449854e-01
-7.70224407e-02 1.10025704e+00 4.05692875e-01 -6.86241269e-01
5.29689454e-02 -7.64022827e-01 -6.06165946e-01 -1.00878584e+00
-9.98048365e-01 -6.07534289e-01 -8.33967268e-01 -6.03945963e-02
7.92500675e-01 -1.24319065e+00 -7.99793720e-01 6.20574176e-01
-1.18336654e+00 -3.95667613e-01 1.25753909e-01 8.94612312e-01
3.09115201e-02 1.82469860e-01 -8.50300789e-01 -1.12133121e+00
-4.11109418e-01 -8.22847128e-01 7.98680067e-01 5.52684486e-01
-3.36709917e-01 -9.87390637e-01 5.36093954e-03 3.91714454e-01
2.71413416e-01 -4.80788738e-01 7.15102077e-01 -1.14157903e+00
2.32861470e-02 -4.17067930e-02 1.21271774e-01 4.46744263e-01
-4.36499357e-01 -7.64707755e-03 -8.30850422e-01 2.98568923e-02
-3.29497218e-01 -2.01451495e-01 7.04388142e-01 3.24350446e-01
1.07392001e+00 -5.20756006e-01 -7.22314566e-02 1.05542354e-01
1.38562751e+00 2.90693521e-01 4.70737010e-01 3.04294527e-01
6.32510543e-01 6.90352201e-01 8.69406343e-01 2.41797894e-01
6.85928464e-01 3.72113228e-01 -3.62581521e-01 7.36723483e-01
-1.19374275e-01 -7.28357852e-01 7.58341253e-01 1.14382553e+00
1.70814395e-01 -2.95658886e-01 -1.52823162e+00 7.62332201e-01
-1.33240771e+00 -5.62355518e-01 -6.83934033e-01 2.37529707e+00
1.28202379e+00 3.55447060e-03 -1.14653647e-01 3.49161774e-02
7.15087175e-01 -5.85970104e-01 4.11975533e-01 -1.17289460e+00
2.56648548e-02 9.60429549e-01 3.90580058e-01 9.85281229e-01
-9.32239413e-01 1.62535906e+00 5.72489595e+00 9.40372348e-01
-5.57546318e-01 2.69813567e-01 2.06906989e-01 4.94884998e-01
-2.70321310e-01 -1.03080817e-01 -1.61734271e+00 7.79862642e-01
1.76573861e+00 1.42932892e-01 1.61525965e-01 5.17804861e-01
-1.25419691e-01 -5.68126261e-01 -4.34844136e-01 5.69059074e-01
6.64074440e-03 -9.03538883e-01 -1.71371683e-01 -3.37406218e-01
6.92533731e-01 2.32175663e-01 -3.47131580e-01 7.17270613e-01
9.22605872e-01 -1.00719631e+00 7.70672321e-01 5.79159915e-01
7.23199248e-01 -8.56255472e-01 7.60622859e-01 7.57003248e-01
-8.20308626e-01 6.20864220e-02 -5.06752193e-01 -5.46783984e-01
-2.89637089e-01 7.60410905e-01 -1.10373330e+00 2.01750085e-01
1.06622314e+00 4.85006869e-01 -9.57142174e-01 1.11330676e+00
-9.46656406e-01 1.26211178e+00 -3.67512077e-01 -7.43108511e-01
-1.23335473e-01 1.11618891e-01 5.07381320e-01 1.79563224e+00
3.49790007e-01 1.40741214e-01 3.53144333e-02 4.03937012e-01
9.76276118e-03 7.09570706e-01 -2.01508790e-01 8.21891949e-02
8.80871892e-01 1.08646441e+00 -4.01825994e-01 -2.23186865e-01
-5.17355204e-01 5.75622678e-01 7.43588686e-01 9.02817845e-02
-1.63912281e-01 -4.77067232e-01 3.63216966e-01 5.12713306e-02
1.60496965e-01 -3.83854330e-01 -6.01655781e-01 -1.21760213e+00
-1.78215936e-01 -8.07865381e-01 4.07684594e-01 -5.13114631e-01
-1.36449015e+00 3.39189470e-01 -2.21074715e-01 -6.62005484e-01
-3.04049253e-01 -8.35512996e-01 -5.38984656e-01 1.28350222e+00
-1.26703465e+00 -8.57202828e-01 2.31991142e-01 3.35243464e-01
6.04953647e-01 -2.81684935e-01 1.20684648e+00 4.19123560e-01
-5.50170779e-01 1.18994343e+00 5.90681331e-03 2.98896044e-01
8.16279292e-01 -1.61666393e+00 4.88183588e-01 1.16974330e+00
1.61950886e-01 8.89341414e-01 4.94283199e-01 -1.13538218e+00
-8.72635722e-01 -1.17232358e+00 2.45567822e+00 -5.59623420e-01
4.69032854e-01 -3.72898549e-01 -8.98975074e-01 7.12977111e-01
2.81757295e-01 -3.42052698e-01 9.50382829e-01 6.62299275e-01
3.02740186e-02 4.85151678e-01 -1.22824001e+00 3.73279065e-01
8.12212467e-01 -4.38542694e-01 -7.06451178e-01 4.27409261e-01
6.02029979e-01 -5.93168914e-01 -1.04015470e+00 3.67604464e-01
2.99432606e-01 -5.87090015e-01 3.70789915e-01 -9.27747905e-01
4.81715411e-01 3.43093351e-02 1.24569535e-01 -1.29510939e+00
-3.15187961e-01 -5.88390231e-01 1.66796565e-01 1.67884111e+00
6.81808472e-01 -3.34484071e-01 3.25531602e-01 7.97517538e-01
-4.14589107e-01 -6.11419380e-01 -9.45329845e-01 -5.45927405e-01
5.53645015e-01 -9.56244528e-01 4.43668306e-01 9.60200846e-01
2.34155878e-01 9.67427939e-02 1.10750988e-01 3.41173410e-01
1.33828148e-01 -6.17126644e-01 2.30114952e-01 -1.26953638e+00
2.91400850e-01 -2.18865834e-02 -1.68752387e-01 -9.04828072e-01
4.61269826e-01 -9.48939860e-01 1.48737490e-01 -1.12658870e+00
-2.60627061e-01 -5.39262831e-01 -2.98391700e-01 6.46136940e-01
-5.39025903e-01 -1.77917346e-01 6.86050206e-02 -4.44738537e-01
-6.98314250e-01 2.45609373e-01 6.34410620e-01 6.14798963e-01
-2.61413991e-01 2.62952238e-01 -5.74968517e-01 7.69707024e-01
9.56272900e-01 -7.27880478e-01 2.91006148e-01 -6.15873754e-01
4.29433703e-01 -2.18532011e-01 -1.57476708e-01 -1.02804935e+00
4.37039584e-01 3.67286317e-02 6.00995541e-01 -3.30881476e-01
-6.97689056e-01 -3.38770866e-01 -5.64181209e-01 3.24176610e-01
-5.53487539e-01 4.92749870e-01 5.36170304e-01 -7.17268586e-02
-1.14327997e-01 -8.00613046e-01 5.43155551e-01 -1.05291747e-01
-6.15526557e-01 -1.75823629e-01 -9.74103630e-01 2.27857336e-01
7.03289509e-01 1.72003254e-01 -1.03234440e-01 1.51185438e-01
-4.14048612e-01 2.78958321e-01 -9.88710150e-02 8.23902011e-01
5.70416451e-01 -8.78005624e-01 -8.78631115e-01 6.86063766e-01
8.19869246e-03 -6.86428323e-02 2.52533914e-03 6.24698162e-01
-5.87106347e-01 6.71474397e-01 -1.05234385e-01 -1.68399438e-01
-1.38580155e+00 -4.21309397e-02 -4.72024409e-03 -4.87787873e-01
-2.38062978e-01 1.34510958e+00 -8.79001319e-01 -8.54451716e-01
3.65395248e-01 -2.77603716e-01 -6.72265470e-01 -1.98930949e-01
8.36792707e-01 6.59928024e-01 5.75208127e-01 -6.16277516e-01
-4.18256402e-01 7.35672787e-02 -2.91710496e-01 -5.84871508e-02
1.37430215e+00 -4.87246960e-02 -1.04136005e-01 3.25013936e-01
5.89858890e-01 5.62165797e-01 -3.74616325e-01 -3.02729309e-01
6.58621252e-01 -2.85482824e-01 1.32832870e-01 -1.33501697e+00
-2.80983090e-01 7.97666252e-01 5.93336701e-01 -1.54213607e-01
8.91470850e-01 -2.59577781e-01 6.04086936e-01 1.63985565e-01
3.21759552e-01 -1.48808396e+00 -9.43559587e-01 1.23186958e+00
3.06300014e-01 -1.34774983e+00 -1.59298912e-01 -5.48029959e-01
-5.03429174e-01 1.08892345e+00 8.80427182e-01 -1.15205400e-01
9.58815336e-01 3.79951715e-01 -2.71585770e-02 1.48406655e-01
-7.95657516e-01 -1.50404811e-01 6.85762525e-01 5.36603034e-01
1.43633735e+00 3.56805235e-01 -1.19624674e+00 1.10888600e+00
-7.57569969e-01 -1.27943620e-01 2.98091978e-01 1.06997418e+00
-5.83716452e-01 -1.45770645e+00 -1.33787930e-01 4.39006954e-01
-1.03484809e+00 -6.92048490e-01 -2.98950613e-01 4.26218837e-01
2.54217267e-01 1.43274784e+00 1.12460434e-01 -3.13136786e-01
3.42502773e-01 5.31154752e-01 3.09345484e-01 -1.08480704e+00
-1.32051027e+00 -3.05507928e-01 2.33601809e-01 -5.71769178e-01
-2.61005878e-01 -8.28641474e-01 -1.39317024e+00 -8.07430297e-02
-3.13826114e-01 3.18417042e-01 7.80630291e-01 1.17233622e+00
3.60390395e-01 1.56931505e-01 1.55889438e-02 1.24331765e-01
-2.99916267e-01 -1.40653610e+00 -4.07154351e-01 9.77143422e-02
-1.95891317e-02 -2.26254195e-01 -1.15200914e-01 -5.51407486e-02]
|
[11.024953842163086, 10.728608131408691]
|
6ffee3b1-9382-4dcb-80c4-fdd3552899ac
|
mpe4g-multimodal-pretrained-encoder-for-co
|
2305.15740
| null |
https://arxiv.org/abs/2305.15740v1
|
https://arxiv.org/pdf/2305.15740v1.pdf
|
MPE4G: Multimodal Pretrained Encoder for Co-Speech Gesture Generation
|
When virtual agents interact with humans, gestures are crucial to delivering their intentions with speech. Previous multimodal co-speech gesture generation models required encoded features of all modalities to generate gestures. If some input modalities are removed or contain noise, the model may not generate the gestures properly. To acquire robust and generalized encodings, we propose a novel framework with a multimodal pre-trained encoder for co-speech gesture generation. In the proposed method, the multi-head-attention-based encoder is trained with self-supervised learning to contain the information on each modality. Moreover, we collect full-body gestures that consist of 3D joint rotations to improve visualization and apply gestures to the extensible body model. Through the series of experiments and human evaluation, the proposed method renders realistic co-speech gestures not only when all input modalities are given but also when the input modalities are missing or noisy.
|
['Hanseok Ko', 'Insung Ham', 'Seonghyeok Noh', 'Gwantae Kim']
|
2023-05-25
| null | null | null | null |
['gesture-generation']
|
['robots']
|
[ 4.60563570e-01 2.38617420e-01 -2.46480912e-01 -3.71087551e-01
-5.39491355e-01 -3.98316562e-01 1.02743328e+00 -8.46729279e-01
-2.61186212e-01 4.40814465e-01 8.05893123e-01 7.24698380e-02
3.97795945e-01 -4.85233754e-01 -6.53982043e-01 -6.75093830e-01
2.16708884e-01 3.68061215e-01 -1.18998982e-01 -2.73062676e-01
-9.65080336e-02 1.85947224e-01 -1.79854345e+00 5.76054096e-01
4.64016497e-01 6.17012322e-01 3.71052057e-01 1.05564213e+00
-3.45838547e-01 7.20940292e-01 -7.25562751e-01 -3.55898291e-01
1.20463438e-01 -6.97743237e-01 -6.62194490e-01 1.92724794e-01
2.00216129e-01 -1.03111088e+00 -6.67744339e-01 7.97745287e-01
8.24937403e-01 3.29807788e-01 6.84935093e-01 -1.37900078e+00
-5.19573808e-01 7.01936841e-01 -1.57000691e-01 -6.09479666e-01
1.01126766e+00 6.09478474e-01 7.33974218e-01 -7.24675298e-01
8.55486274e-01 1.50506365e+00 9.15041789e-02 1.20280027e+00
-6.40694916e-01 -6.08597577e-01 3.22814077e-01 1.57681331e-01
-1.16263425e+00 -7.12103069e-01 8.23181093e-01 -2.63582140e-01
9.92384374e-01 5.44581950e-01 8.00139010e-01 1.82833505e+00
-3.36247265e-01 1.13251173e+00 5.92597425e-01 -3.04505140e-01
2.41843518e-03 -3.09892386e-01 -4.31325972e-01 5.11716008e-01
-2.76187867e-01 4.71526533e-01 -8.50804150e-01 -1.58870760e-02
1.11287165e+00 2.52484351e-01 -2.29608312e-01 -1.81016624e-01
-1.70643365e+00 4.64736581e-01 3.59778285e-01 3.44573617e-01
-5.96484542e-01 4.05220330e-01 3.37213218e-01 9.86768752e-02
-9.93512273e-02 -2.06379667e-02 -1.30637690e-01 -4.59751874e-01
-6.56576037e-01 1.42017305e-01 5.76887429e-01 1.24310684e+00
1.62612781e-01 1.92535877e-01 -3.50233555e-01 6.37653589e-01
6.96159124e-01 9.38529074e-01 7.42014647e-01 -8.78993928e-01
8.09645474e-01 5.51980853e-01 1.77766249e-01 -6.90320015e-01
-3.83877784e-01 1.96646392e-01 -9.61569428e-01 2.82390177e-01
3.22739124e-01 -3.35877061e-01 -1.25764048e+00 1.76423478e+00
4.76585686e-01 2.79440612e-01 3.66307676e-01 1.38081074e+00
1.49672580e+00 5.66460371e-01 3.41034532e-01 -2.34248191e-01
1.10916150e+00 -1.02176619e+00 -1.22724712e+00 2.52863355e-02
2.37765104e-01 -7.08778679e-01 1.21565461e+00 4.60666381e-02
-1.02202356e+00 -7.01271355e-01 -8.96850407e-01 -7.25352168e-02
7.36802584e-03 1.08643286e-01 5.81543207e-01 4.48720723e-01
-6.72899723e-01 1.51414052e-01 -1.08397436e+00 -4.40084249e-01
1.06392108e-01 3.00444812e-01 -5.42052448e-01 1.41209647e-01
-9.12892640e-01 7.34246373e-01 2.33313128e-01 1.83379054e-01
-9.81452882e-01 1.13417849e-01 -1.13588929e+00 -2.14291904e-02
2.37873018e-01 -9.83982503e-01 1.35773516e+00 -9.84253109e-01
-2.23151207e+00 3.05985391e-01 -3.62597436e-01 2.14000165e-01
8.02073658e-01 -2.49350727e-01 -3.02139878e-01 2.65208066e-01
-3.16357106e-01 8.94664168e-01 1.11312294e+00 -1.49912143e+00
-4.01256412e-01 -4.65407372e-01 8.07300359e-02 8.50602686e-01
-1.34891495e-01 -7.46634230e-02 -6.76815033e-01 -6.44572377e-01
3.37060601e-01 -9.85610127e-01 -4.58319187e-02 -1.77287653e-01
-6.97811484e-01 5.35771102e-02 8.40573847e-01 -6.76233113e-01
8.31456065e-01 -1.98370743e+00 6.83511198e-01 1.59455121e-01
9.38886963e-03 2.33715326e-02 -4.23146129e-01 5.16019166e-01
2.04592600e-01 -2.16110125e-02 -5.28320856e-02 -6.59439087e-01
2.75599211e-01 5.06248534e-01 -1.42772228e-01 2.07075670e-01
-1.64403785e-02 1.14778972e+00 -7.43951678e-01 -5.52577019e-01
7.99294710e-01 9.95964289e-01 -4.06609386e-01 7.53477693e-01
-2.63018787e-01 1.01621366e+00 -5.41449189e-01 6.73108876e-01
4.42909211e-01 9.87958629e-03 3.05711657e-01 -2.02650189e-01
2.32247889e-01 1.26330689e-01 -1.20725322e+00 2.23662996e+00
-5.34524798e-01 3.42927307e-01 3.97038423e-02 -3.10273975e-01
5.87988973e-01 8.04749250e-01 3.97583693e-01 -5.95481336e-01
4.14707333e-01 -8.72444548e-03 -7.89807886e-02 -1.18015003e+00
4.03828681e-01 -7.00783581e-02 1.11977741e-01 5.08507669e-01
2.02340633e-01 -2.36521333e-01 -2.24863768e-01 -3.09439842e-03
9.20029819e-01 6.58034325e-01 7.65110627e-02 9.22934175e-01
3.22739422e-01 -2.56816894e-01 1.12683162e-01 4.36978966e-01
2.18428951e-02 9.02710736e-01 4.75021675e-02 -1.18997976e-01
-8.38747442e-01 -9.16369617e-01 4.10364568e-01 1.30889630e+00
2.37961873e-01 -2.73583144e-01 -5.57025611e-01 -7.60747850e-01
-3.69713217e-01 5.72340012e-01 -3.82417828e-01 7.40357116e-02
-6.01644814e-01 -2.73558497e-01 6.81228459e-01 6.84392393e-01
3.77441585e-01 -1.48155117e+00 -9.72016752e-01 -5.59189729e-02
-4.39498544e-01 -1.14789927e+00 -4.55737948e-01 -2.06349820e-01
-6.33444428e-01 -1.05280960e+00 -1.23434556e+00 -7.72635937e-01
7.80587077e-01 -1.18022906e-02 3.23898673e-01 2.38365561e-01
1.45015851e-01 4.67407614e-01 -7.59408534e-01 -4.89370385e-03
-4.36358780e-01 -1.49927974e-01 -9.95357782e-02 5.30702621e-02
1.17478944e-01 -5.04232943e-01 -3.12426209e-01 2.20343098e-01
-8.03234696e-01 4.74629849e-01 6.18368030e-01 1.01857758e+00
1.71871185e-01 -6.99023366e-01 2.03313947e-01 -2.85488784e-01
5.87429166e-01 -1.45494133e-01 9.02981013e-02 1.72306761e-01
2.48002902e-01 -1.61335561e-02 2.25129813e-01 -9.24344957e-01
-1.37166429e+00 5.32135665e-01 -3.67439568e-01 -6.26195431e-01
-6.62602425e-01 3.13550472e-01 -5.92169225e-01 2.53231317e-01
2.88972408e-01 3.66373360e-01 7.32228905e-02 -4.34487909e-01
8.28957438e-01 9.60556448e-01 6.10129118e-01 -3.51410478e-01
6.93378448e-01 2.94024646e-01 -2.62668163e-01 -8.52241576e-01
3.71089965e-01 -1.67410731e-01 -8.91194344e-01 -4.56911594e-01
9.13840055e-01 -8.11729848e-01 -1.01795971e+00 6.20485663e-01
-1.57254148e+00 -5.37109613e-01 2.31259875e-02 9.20016170e-01
-7.27443516e-01 3.47460508e-01 -5.80114007e-01 -1.10776222e+00
-1.15479052e-01 -1.53796685e+00 1.59284520e+00 1.92871630e-01
-4.21545297e-01 -5.09373724e-01 -1.50619850e-01 3.06441218e-01
1.90251976e-01 1.46733835e-01 5.23707390e-01 -5.52591622e-01
-4.05941457e-01 -1.12222508e-01 6.87768683e-02 -1.29612491e-01
3.44396710e-01 -5.37397303e-02 -8.77463222e-01 8.19735005e-02
-5.40928543e-01 -4.53365088e-01 5.55021107e-01 1.58669427e-01
5.86782277e-01 -5.08214116e-01 -2.66039431e-01 7.02735424e-01
6.83828950e-01 5.69222212e-01 6.20507240e-01 -9.12363380e-02
1.02825737e+00 7.24055767e-01 4.24149066e-01 4.87322718e-01
5.62124372e-01 8.55197728e-01 6.09612703e-01 -3.68665829e-02
-4.85124975e-01 -6.25224471e-01 5.48268795e-01 9.32346404e-01
-5.25666118e-01 -4.08590406e-01 -5.25323749e-01 2.90584713e-01
-1.81225896e+00 -1.05039871e+00 4.81138825e-02 1.92963934e+00
7.56552815e-01 -4.66988176e-01 1.78631887e-01 -7.32512623e-02
5.61975121e-01 1.79938361e-01 -5.97275376e-01 -7.30551630e-02
-1.67716131e-01 9.29618552e-02 -1.60625368e-01 7.04958260e-01
-9.17727113e-01 1.27349973e+00 6.07542849e+00 2.27984756e-01
-1.39442468e+00 5.31025454e-02 -1.28219932e-01 -3.93663734e-01
-5.12791753e-01 -4.20956373e-01 -2.62122512e-01 3.63728583e-01
4.04463738e-01 4.31398213e-01 4.37343955e-01 5.92049479e-01
2.08701566e-01 1.42802417e-01 -1.24229348e+00 1.34256959e+00
3.37093949e-01 -8.17927301e-01 1.45466849e-01 -6.63047433e-02
4.78576928e-01 -3.24240267e-01 9.44691896e-02 3.69516879e-01
2.38062367e-01 -1.19882381e+00 8.25378060e-01 5.66384852e-01
1.24565721e+00 -3.73362124e-01 6.72289848e-01 4.91029650e-01
-1.12034893e+00 1.24857463e-01 7.68930167e-02 -8.06640461e-02
7.31396317e-01 -4.58560914e-01 -9.05759633e-01 4.95139122e-01
3.04820806e-01 4.88134861e-01 -7.02026263e-02 5.70706844e-01
-5.83335757e-01 2.40038857e-01 -4.95803714e-01 -2.36045927e-01
-6.05051406e-02 6.62388355e-02 6.78422511e-01 9.89294171e-01
5.20921886e-01 4.52484727e-01 9.65892524e-02 5.89294076e-01
1.92134619e-01 2.19951436e-01 -7.91620016e-01 -2.07518280e-01
4.13943350e-01 8.08312356e-01 -3.74929339e-01 -5.80544353e-01
-3.81041706e-01 1.58995032e+00 -1.71880469e-01 8.60771894e-01
-6.67755842e-01 -1.40744284e-01 7.59152174e-01 -3.06084961e-01
-1.79892797e-02 -4.05748457e-01 -2.67382771e-01 -1.28715193e+00
-6.78997189e-02 -9.49736357e-01 -1.60390213e-02 -1.01296985e+00
-8.61223102e-01 7.96312392e-01 1.01121075e-01 -1.34094286e+00
-9.52399015e-01 -4.06103760e-01 -4.09245193e-01 7.00050354e-01
-9.34981704e-01 -1.72817016e+00 -5.70187569e-01 7.93247521e-01
7.46546090e-01 -3.30358654e-01 1.01403260e+00 1.92389712e-01
-2.36849785e-01 6.54356122e-01 -6.17484987e-01 1.84322357e-01
6.51622772e-01 -6.70764923e-01 2.11619213e-01 5.37565529e-01
3.22406739e-01 7.48932600e-01 6.69785559e-01 -9.31079388e-01
-1.61098266e+00 -6.01245403e-01 6.68443739e-01 -2.98651218e-01
5.29786153e-03 -2.18980864e-01 -5.92848241e-01 9.22812402e-01
4.07875866e-01 -1.70608193e-01 6.19926870e-01 -1.73720911e-01
-1.69813842e-01 5.64656675e-01 -1.01728463e+00 9.21148241e-01
1.46153927e+00 -5.88999748e-01 -8.89584363e-01 -3.77341313e-03
8.07384312e-01 -9.15152490e-01 -4.75036263e-01 4.22130704e-01
1.03408206e+00 -6.88948572e-01 9.09646094e-01 -8.43868196e-01
5.77388942e-01 -2.91119069e-01 -3.96677881e-01 -1.36936128e+00
2.93795437e-01 -5.36509514e-01 -3.16255480e-01 1.01786268e+00
2.08929330e-01 -2.24762291e-01 7.34644830e-01 6.00848019e-01
-9.63293836e-02 -1.52280122e-01 -9.55406904e-01 -2.43212298e-01
-3.45747650e-01 -7.46895492e-01 1.14003813e+00 9.40781057e-01
5.17157614e-01 3.46661985e-01 -8.15594256e-01 1.75158530e-01
3.16595227e-01 -9.79082137e-02 1.31276166e+00 -7.98605025e-01
-2.86100507e-01 -3.64526749e-01 -3.71979803e-01 -1.54496288e+00
2.96726823e-01 -7.20813930e-01 3.07844639e-01 -1.73473430e+00
9.70977545e-02 -8.28676522e-02 2.34494671e-01 7.35365689e-01
-1.50486052e-01 -4.73185405e-02 5.83062947e-01 2.08307967e-01
-3.21451426e-01 9.56018388e-01 1.72622013e+00 -3.11225146e-01
-4.44426894e-01 1.38914110e-02 -5.32295518e-02 6.84474111e-01
3.98159087e-01 -3.32209952e-02 -5.15463769e-01 -6.30969226e-01
-2.19000086e-01 5.88666379e-01 5.81307113e-01 -7.03035593e-01
2.55484879e-01 -4.78030264e-01 5.23816288e-01 -6.60450280e-01
9.27147985e-01 -9.39877808e-01 3.15112174e-01 3.33754838e-01
-6.61742985e-01 -1.99775130e-01 -7.51130357e-02 2.40057305e-01
-2.01461986e-01 1.22198097e-01 -7.76467891e-03 -4.99135256e-02
-6.64316118e-01 1.93397805e-01 -3.65401626e-01 -5.98485768e-01
8.45709682e-01 -2.12539330e-01 -1.63866520e-01 -1.00946689e+00
-1.14991117e+00 5.56438677e-02 3.27860892e-01 8.12014997e-01
1.03607285e+00 -1.74457514e+00 -5.62578678e-01 5.16411185e-01
1.39311761e-01 2.30478514e-02 4.27244455e-01 5.79501808e-01
-2.96435475e-01 3.21631253e-01 -3.32374454e-01 -5.72316051e-01
-1.56674731e+00 3.09435368e-01 3.09358593e-02 4.52050209e-01
-7.00371504e-01 8.10189605e-01 6.48577213e-02 -6.95155919e-01
6.02057993e-01 -4.68319058e-01 -3.95272911e-01 -1.36891395e-01
6.29521430e-01 5.72697371e-02 -4.89814460e-01 -1.33011258e+00
-2.29252696e-01 6.08174324e-01 6.11165822e-01 -8.76185536e-01
1.05196583e+00 -1.44063115e-01 3.73087466e-01 4.67488080e-01
1.01568270e+00 -8.67544413e-02 -1.31868589e+00 -1.33334562e-01
-6.93209171e-01 -6.25563264e-01 -3.34050298e-01 -9.51988876e-01
-1.17778587e+00 1.18720746e+00 5.59231997e-01 -4.69718724e-01
9.50764477e-01 -6.23360602e-03 8.18475127e-01 3.69576246e-01
6.29440367e-01 -8.53406668e-01 3.41246516e-01 5.00800967e-01
1.27163732e+00 -1.29313016e+00 -4.13969606e-01 -3.99591208e-01
-1.20375729e+00 1.03140378e+00 7.84258068e-01 4.03736919e-01
3.32818031e-01 3.34778428e-01 4.07433212e-01 -1.86345652e-02
-6.29609644e-01 -5.48665702e-01 3.38649005e-01 1.01899123e+00
4.77359056e-01 2.80778497e-01 -3.02579671e-01 8.87553215e-01
-4.17515129e-01 3.05689685e-02 1.66710392e-01 9.28593099e-01
-9.01844129e-02 -1.09442329e+00 -5.44240475e-01 -2.55790316e-02
1.16292030e-01 2.51678750e-02 -6.37003303e-01 6.61274254e-01
7.21867159e-02 1.06572378e+00 2.08103587e-03 -9.09940839e-01
3.88302386e-01 4.41468924e-01 8.05376828e-01 -5.11270046e-01
-5.42736530e-01 4.86326307e-01 1.20910592e-01 -7.47607052e-01
-5.56203902e-01 -6.36832356e-01 -1.74445057e+00 -4.43830602e-02
-1.88121170e-01 -3.97378176e-01 8.52243006e-01 1.04682052e+00
1.76418751e-01 6.79378211e-01 2.17610478e-01 -1.52232850e+00
-4.73826855e-01 -1.35274744e+00 -2.28629857e-01 7.69062996e-01
4.95428354e-01 -8.40612054e-01 -1.33297071e-01 3.04320395e-01]
|
[5.6316728591918945, -0.12270453572273254]
|
02b7d2f0-8468-40fd-8f8d-52b2e4c36d97
|
scattering-transform-based-image-clustering
|
2011.11586
| null |
https://arxiv.org/abs/2011.11586v2
|
https://arxiv.org/pdf/2011.11586v2.pdf
|
Scattering Transform Based Image Clustering using Projection onto Orthogonal Complement
|
In the last few years, large improvements in image clustering have been driven by the recent advances in deep learning. However, due to the architectural complexity of deep neural networks, there is no mathematical theory that explains the success of deep clustering techniques. In this work we introduce Projected-Scattering Spectral Clustering (PSSC), a state-of-the-art, stable, and fast algorithm for image clustering, which is also mathematically interpretable. PSSC includes a novel method to exploit the geometric structure of the scattering transform of small images. This method is inspired by the observation that, in the scattering transform domain, the subspaces formed by the eigenvectors corresponding to the few largest eigenvalues of the data matrices of individual classes are nearly shared among different classes. Therefore, projecting out those shared subspaces reduces the intra-class variability, substantially increasing the clustering performance. We call this method Projection onto Orthogonal Complement (POC). Our experiments demonstrate that PSSC obtains the best results among all shallow clustering algorithms. Moreover, it achieves comparable clustering performance to that of recent state-of-the-art clustering techniques, while reducing the execution time by more than one order of magnitude. In the spirit of reproducible research, we publish a high quality code repository along with the paper.
|
['Veniamin I. Morgenshtern', 'Angel Villar-Corrales']
|
2020-11-23
| null | null | null | null |
['image-clustering']
|
['computer-vision']
|
[-4.09548692e-02 -3.42418253e-01 3.06951553e-01 -2.14294776e-01
-5.17857552e-01 -5.90535820e-01 4.24913198e-01 -2.57060640e-02
-2.59844005e-01 1.46125872e-02 4.89500538e-02 -8.43003616e-02
-5.33597708e-01 -5.70707738e-01 -5.42468548e-01 -1.33993530e+00
-1.51215523e-01 5.40388525e-01 3.27865303e-01 4.17553857e-02
1.25675946e-01 5.02770543e-01 -1.65134084e+00 5.13275385e-01
6.18995786e-01 9.84954119e-01 1.34537533e-01 1.92301959e-01
-1.48349345e-01 2.89799690e-01 -2.55907118e-01 -4.17062007e-02
2.37668648e-01 -5.20405710e-01 -5.36871076e-01 1.03747740e-01
1.63632765e-01 3.17491323e-01 -3.78541172e-01 1.08259106e+00
3.81159753e-01 1.76260546e-02 7.23416924e-01 -1.19685733e+00
-5.79550266e-01 6.19144976e-01 -8.02548885e-01 -2.17713058e-01
-1.17804758e-01 -3.57911885e-01 1.00913894e+00 -7.99633980e-01
6.01949155e-01 1.00513065e+00 7.75929987e-01 4.51874346e-01
-1.26281321e+00 -6.17811978e-01 -2.54064739e-01 4.00028795e-01
-1.73011267e+00 -2.04131991e-01 8.89207661e-01 -4.69595909e-01
6.74131215e-01 2.92760313e-01 6.98937476e-01 7.31469989e-01
-1.93304807e-01 5.98598063e-01 1.11529195e+00 -5.92352867e-01
4.76192445e-01 -1.42404318e-01 2.36810327e-01 6.06217325e-01
1.92294449e-01 -1.67151883e-01 -3.92941266e-01 -1.80196881e-01
3.52094948e-01 1.35682449e-01 -3.15473855e-01 -8.47874522e-01
-1.27581573e+00 9.28979635e-01 4.92247015e-01 4.71173346e-01
-1.96723074e-01 1.64703369e-01 2.57495761e-01 -2.14529838e-02
2.57718801e-01 2.80898809e-01 -1.52219713e-01 1.19306125e-01
-1.05909908e+00 -8.50933716e-02 7.36592591e-01 5.85691452e-01
7.46542752e-01 -1.73233345e-01 3.85721028e-01 6.95317805e-01
3.49678427e-01 5.74124098e-01 2.44858846e-01 -1.25915956e+00
3.29792909e-02 7.84034669e-01 -3.30857694e-01 -1.23550856e+00
-6.26788437e-01 -6.48610353e-01 -1.29375970e+00 3.69239151e-02
2.32742861e-01 1.42462552e-01 -6.44449294e-01 1.60012805e+00
2.62462258e-01 -5.22619300e-02 1.14139408e-01 8.38531375e-01
5.49106777e-01 5.22559822e-01 -3.85416836e-01 -2.40297154e-01
1.05820668e+00 -5.63772082e-01 -4.41175550e-01 3.29840183e-01
4.38421249e-01 -8.63198340e-01 8.07590783e-01 4.00830448e-01
-5.54024398e-01 -4.20557767e-01 -1.00075746e+00 3.99626166e-01
-2.71406263e-01 6.58351555e-02 6.02744758e-01 8.06496382e-01
-1.17505658e+00 7.00495541e-01 -1.15888059e+00 -6.40049279e-01
4.83038604e-01 3.74817550e-01 -1.91855505e-01 -4.41220254e-02
-6.83752358e-01 2.29901806e-01 2.23555773e-01 -1.61567718e-01
-3.77911866e-01 -6.22637808e-01 -1.90934107e-01 4.53216843e-02
3.15485865e-01 -3.55569243e-01 6.18267357e-01 -8.95288527e-01
-1.17848396e+00 7.72464693e-01 -2.47894615e-01 -2.78092772e-01
1.57628462e-01 6.16584253e-03 -2.62622178e-01 5.58759332e-01
-5.54384955e-04 6.45009339e-01 7.73192167e-01 -1.58828402e+00
-3.12047780e-01 -5.48393071e-01 -4.96767998e-01 -1.96879674e-02
-6.57428026e-01 -4.36502881e-02 -6.20593190e-01 -2.67204165e-01
6.04961514e-01 -1.30587912e+00 -3.83393049e-01 -2.10618734e-01
-3.82030964e-01 -2.40459383e-01 9.92388070e-01 -1.65741056e-01
1.14283919e+00 -2.59955406e+00 3.58985275e-01 5.55714369e-01
5.55078089e-01 1.12660579e-01 7.64309242e-02 5.91449976e-01
-2.98688322e-01 -1.70018617e-02 -5.00033796e-01 -2.21714556e-01
-1.56913307e-02 5.59402704e-02 -1.42606765e-01 7.57414818e-01
-3.66014451e-01 6.24426782e-01 -6.87165499e-01 -3.39859605e-01
2.71938831e-01 5.07195473e-01 -3.99220794e-01 -2.84938991e-01
1.40890613e-01 4.33565110e-01 -1.28412887e-01 2.80633986e-01
9.30801928e-01 -5.58926702e-01 3.99495959e-01 -4.41614389e-01
-2.14875787e-01 -2.88647741e-01 -1.17887652e+00 1.61634386e+00
1.08733870e-01 8.67635190e-01 9.81424078e-02 -1.31758869e+00
7.45228291e-01 2.11542353e-01 1.15380204e+00 -4.64334130e-01
8.01819861e-02 2.47300550e-01 2.60800302e-01 -2.41618454e-01
2.02518404e-02 -8.37313235e-02 2.37455860e-01 7.32182980e-01
-7.38796815e-02 2.40037903e-01 2.36531779e-01 5.03987253e-01
9.53261197e-01 -4.55832571e-01 -4.43285368e-02 -6.15034878e-01
4.53032792e-01 7.91331008e-02 3.03622097e-01 6.66033328e-01
-1.13969848e-01 9.33493972e-01 2.96546459e-01 -4.55287874e-01
-1.08751595e+00 -1.21289086e+00 -2.25077569e-01 7.20565379e-01
3.10638212e-02 -5.96766412e-01 -1.14595068e+00 -2.13663474e-01
-2.05716684e-01 1.95445940e-01 -3.90870601e-01 -6.84875548e-02
-4.44238365e-01 -1.15337873e+00 5.08851290e-01 4.53240812e-01
6.80167913e-01 -7.92453766e-01 -5.18443763e-01 -4.78128977e-02
-4.17229325e-01 -1.26723707e+00 -5.80084026e-02 2.31624633e-01
-8.84668589e-01 -1.25185144e+00 -6.01021171e-01 -9.32677507e-01
7.05123603e-01 7.28980422e-01 1.00629246e+00 1.99615568e-01
-4.19620097e-01 3.83708805e-01 -5.82654834e-01 -1.55760303e-01
-2.24990517e-01 -3.04036792e-02 1.81123435e-01 3.52777123e-01
7.18707740e-01 -6.78641021e-01 -5.48765481e-01 1.61563501e-01
-8.38278353e-01 2.61645541e-02 5.76302111e-01 4.23656940e-01
6.33981049e-01 6.46884978e-01 -1.90981757e-02 -6.39987707e-01
3.29400271e-01 -1.93473458e-01 -4.73109901e-01 1.00327671e-01
-5.68266213e-01 1.33328155e-01 7.35873163e-01 -5.50244637e-02
-7.25134254e-01 4.53626186e-01 1.24721147e-01 -4.83498007e-01
-4.32054758e-01 5.67208230e-01 4.38820459e-02 -1.04038149e-01
6.73377156e-01 4.76445526e-01 1.17311366e-01 -3.92889440e-01
4.40676719e-01 6.23627782e-01 4.85868663e-01 -5.00969768e-01
9.77522433e-01 1.16441524e+00 3.17073554e-01 -1.23733771e+00
-8.03391993e-01 -8.96707952e-01 -1.15148938e+00 -3.48209202e-01
1.18237352e+00 -7.96050310e-01 -7.94904709e-01 5.47284245e-01
-9.24471498e-01 -1.34604543e-01 1.11201704e-01 4.88374263e-01
-3.64327282e-01 6.15064621e-01 -4.58971679e-01 -5.96692502e-01
-1.08254343e-01 -1.10274696e+00 8.76739383e-01 -1.58000022e-01
-2.14787293e-02 -7.79071450e-01 -1.43154734e-03 4.82436508e-01
1.86849818e-01 3.49852920e-01 1.19082820e+00 -3.67043912e-01
-5.98794520e-01 -3.02475318e-02 -2.53139198e-01 3.36965919e-01
-1.09841779e-01 1.42632708e-01 -8.18082750e-01 -3.79224509e-01
1.52123138e-01 1.68200880e-01 1.10351992e+00 6.74451828e-01
1.32128966e+00 1.83061615e-01 -4.65461999e-01 7.64740705e-01
1.58897078e+00 1.98474661e-01 6.24016225e-01 2.23385736e-01
9.41752017e-01 7.28117824e-01 -1.01690464e-01 2.10703120e-01
2.13676944e-01 5.37005782e-01 1.84590384e-01 -2.13620767e-01
-2.22243533e-01 3.28867435e-01 2.04728171e-01 1.29690218e+00
-3.25828403e-01 -3.59022096e-02 -1.14974523e+00 4.22674298e-01
-2.04915929e+00 -1.10153222e+00 -6.96712732e-01 2.13921762e+00
2.56419986e-01 -7.22953975e-02 -2.22203285e-02 3.85169566e-01
8.06290448e-01 -1.28185602e-05 -3.68594885e-01 2.12166533e-01
-3.28224361e-01 2.13696346e-01 4.52032447e-01 1.51611656e-01
-1.26520896e+00 6.80932760e-01 6.33762026e+00 7.99032271e-01
-9.82304931e-01 2.50508403e-03 4.71081078e-01 -1.72773093e-01
-1.32233156e-02 -1.14103869e-01 -3.30638170e-01 3.89100224e-01
9.56440151e-01 1.30644947e-01 5.70179343e-01 5.71962893e-01
2.29106039e-01 1.51428491e-01 -9.27410960e-01 1.25031400e+00
1.96547255e-01 -1.51577950e+00 2.42175069e-02 3.85179788e-01
9.20042694e-01 3.19039136e-01 2.66778708e-01 -2.32380897e-01
1.48574531e-01 -8.12417090e-01 5.76465070e-01 2.43080720e-01
4.90259111e-01 -1.12350965e+00 7.48457611e-01 1.62023634e-01
-1.29279661e+00 -2.02684686e-01 -5.60362160e-01 1.35114461e-01
-1.59416661e-01 1.03890789e+00 -2.13134214e-01 5.17418385e-01
1.17571902e+00 8.23892534e-01 -6.65854037e-01 8.36078823e-01
1.55739129e-01 7.88626373e-01 -3.23753208e-01 7.43607059e-02
5.13324380e-01 -7.82757878e-01 3.18408877e-01 1.03304040e+00
3.03672880e-01 2.62339592e-01 2.52670795e-01 8.27495098e-01
-1.51131526e-02 -1.52179413e-02 -4.65804309e-01 -1.37141779e-01
4.38563704e-01 1.39445055e+00 -1.40628600e+00 -1.84738472e-01
-5.19646883e-01 8.86748374e-01 1.56512871e-01 3.42538059e-01
-6.85598314e-01 -2.14889824e-01 5.28585672e-01 9.46963578e-02
4.95110482e-01 -6.03383720e-01 -4.73195285e-01 -8.13836336e-01
1.13821417e-01 -7.85681963e-01 2.69419968e-01 -4.24470276e-01
-1.33047152e+00 5.57465971e-01 -1.69917434e-01 -1.23534572e+00
2.52447098e-01 -6.89706266e-01 -4.58445340e-01 3.04532915e-01
-1.01283681e+00 -8.83105874e-01 -4.48640555e-01 7.64131486e-01
7.12112859e-02 -4.19518888e-01 8.35735023e-01 3.72381806e-01
-5.25675833e-01 1.22033492e-01 1.01447833e+00 3.42341691e-01
5.77712119e-01 -1.23279369e+00 1.58225104e-01 8.02568913e-01
4.49942946e-01 7.51035094e-01 4.53062534e-01 -3.43859971e-01
-1.33278906e+00 -1.02638149e+00 5.07475555e-01 -3.25177103e-01
7.66379893e-01 -6.17760658e-01 -8.02945912e-01 1.51494175e-01
1.93985596e-01 -1.49749845e-01 1.10924661e+00 2.65133798e-01
-5.61481357e-01 -2.09964529e-01 -6.50278389e-01 4.56022322e-01
8.42889726e-01 -5.59009314e-01 -2.12374136e-01 5.48264444e-01
6.09084547e-01 1.70364425e-01 -6.55801475e-01 1.67130247e-01
4.45579261e-01 -1.42639399e+00 1.20343900e+00 -1.38976544e-01
2.88883567e-01 -4.90494698e-01 -5.00676692e-01 -1.13195062e+00
-7.41834342e-01 -4.24612224e-01 9.19265896e-02 8.81086111e-01
2.55418032e-01 -2.94745654e-01 1.05123520e+00 8.25002417e-02
-2.34997526e-01 -5.14165998e-01 -8.04324567e-01 -1.00445282e+00
1.86397180e-01 -3.58236969e-01 4.16902095e-01 1.09463000e+00
-1.25731364e-01 3.54287446e-01 -2.35212184e-02 2.96042025e-01
1.22054875e+00 4.47496504e-01 4.03016686e-01 -1.74506760e+00
-1.37823403e-01 -5.60283422e-01 -3.93143892e-01 -5.69142640e-01
1.55316055e-01 -1.09975851e+00 6.26610592e-02 -1.43218029e+00
6.25067830e-01 -3.18471968e-01 -2.96900809e-01 1.57981887e-01
6.87081441e-02 6.00538671e-01 3.39638650e-01 7.57675946e-01
-8.67238998e-01 5.30683935e-01 8.43200445e-01 -1.01622306e-01
-7.07402602e-02 -2.38903329e-01 -6.22598231e-01 8.27724278e-01
8.22549999e-01 -5.16588211e-01 -2.54816562e-01 -5.12595236e-01
1.38030797e-01 -4.97571677e-01 3.39855909e-01 -1.58318233e+00
5.75570762e-01 2.12331459e-01 4.28938031e-01 -8.57573450e-01
2.01108620e-01 -9.82152104e-01 2.35261023e-01 6.42713428e-01
-1.83612809e-01 -1.21379554e-01 -1.22240037e-01 7.21259058e-01
-2.63074160e-01 8.78859870e-03 1.05652750e+00 -2.80610770e-01
-5.82691312e-01 1.85416877e-01 -5.87551296e-01 -2.06798211e-01
9.51456547e-01 -1.86372489e-01 -5.85289150e-02 -3.59281987e-01
-5.94126999e-01 -1.09106570e-01 4.54741925e-01 9.05903801e-02
4.56721753e-01 -1.30904102e+00 -6.44824505e-01 9.25169140e-02
4.61947508e-02 -1.27596706e-01 2.94053495e-01 9.62699592e-01
-5.98537982e-01 5.02483308e-01 -3.68961208e-02 -1.05094123e+00
-1.51130044e+00 6.88286185e-01 1.77404255e-01 1.31949827e-01
-8.35446417e-01 6.86328530e-01 2.51217067e-01 -6.32230103e-01
1.64657965e-01 1.34324536e-01 4.14327867e-02 1.21573415e-02
4.99679148e-01 5.69413602e-01 2.53539443e-01 -7.16136396e-01
-5.07992804e-01 9.55399513e-01 1.10807836e-01 -6.17265068e-02
1.63764489e+00 -1.38215810e-01 -6.30578399e-01 3.42314392e-01
1.52736664e+00 -2.34436825e-01 -9.28453207e-01 -1.20800808e-01
4.42581289e-02 -1.36585578e-01 1.30512744e-01 -4.73296463e-01
-1.45724666e+00 1.04881620e+00 7.55700529e-01 4.10991043e-01
1.22032595e+00 2.81758249e-01 7.35931039e-01 6.35968924e-01
3.29946071e-01 -1.18736792e+00 2.18180612e-01 5.40479958e-01
3.78866285e-01 -1.10984647e+00 1.44965723e-01 -5.44330359e-01
-4.84526217e-01 1.19934130e+00 1.58450544e-01 -9.36792046e-02
1.00034916e+00 2.58963078e-01 3.02658081e-01 -6.83906615e-01
-3.85292441e-01 -2.05263168e-01 2.74320006e-01 7.85044670e-01
4.53590423e-01 2.53034592e-01 3.27829160e-02 6.50311932e-02
-3.32636595e-01 -6.32322609e-01 3.74605656e-01 4.40309376e-01
-6.50400341e-01 -9.61661160e-01 -6.87102377e-01 3.06818068e-01
-1.46297649e-01 -4.52140905e-02 -7.76532769e-01 5.92813432e-01
1.49140745e-01 1.20239604e+00 1.60631508e-01 -4.87201005e-01
-1.69085220e-01 -2.19947174e-02 3.43657255e-01 -2.92501420e-01
-1.95299327e-01 4.11923826e-01 -5.82364798e-01 -5.83076656e-01
-8.40220749e-01 -7.87341356e-01 -1.56169915e+00 -4.14787054e-01
-9.62115973e-02 3.35000366e-01 6.36731863e-01 9.95323658e-01
4.70957607e-01 2.76186675e-01 6.44917667e-01 -7.78039694e-01
-1.31241664e-01 -4.76625055e-01 -8.40514481e-01 6.04476511e-01
-9.20551247e-04 -5.52004158e-01 -4.71692890e-01 3.98113281e-02]
|
[8.81336498260498, 3.515444755554199]
|
7f229dc4-e5ec-4e24-aa72-2a4b675e6e31
|
investigating-post-pretraining-representation
|
2109.12028
| null |
https://arxiv.org/abs/2109.12028v1
|
https://arxiv.org/pdf/2109.12028v1.pdf
|
Investigating Post-pretraining Representation Alignment for Cross-Lingual Question Answering
|
Human knowledge is collectively encoded in the roughly 6500 languages spoken around the world, but it is not distributed equally across languages. Hence, for information-seeking question answering (QA) systems to adequately serve speakers of all languages, they need to operate cross-lingually. In this work we investigate the capabilities of multilingually pre-trained language models on cross-lingual QA. We find that explicitly aligning the representations across languages with a post-hoc fine-tuning step generally leads to improved performance. We additionally investigate the effect of data size as well as the language choice in this fine-tuning step, also releasing a dataset for evaluating cross-lingual QA systems. Code and dataset are publicly available here: https://github.com/ffaisal93/aligned_qa
|
['Antonios Anastasopoulos', 'Fahim Faisal']
|
2021-09-24
| null |
https://aclanthology.org/2021.mrqa-1.14
|
https://aclanthology.org/2021.mrqa-1.14.pdf
|
emnlp-mrqa-2021-11
|
['cross-lingual-question-answering']
|
['natural-language-processing']
|
[-5.43148339e-01 1.48461643e-03 -2.19804361e-01 -6.66265965e-01
-1.57824731e+00 -9.78334665e-01 6.88166440e-01 1.96515560e-01
-6.38217688e-01 6.19674981e-01 6.48780227e-01 -9.22372818e-01
1.05750591e-01 -5.76632142e-01 -6.58458710e-01 -1.11763198e-02
2.95458078e-01 8.25166881e-01 -5.69221424e-03 -6.06755257e-01
-3.12326457e-02 1.37347102e-01 -1.10212433e+00 3.99733663e-01
1.47310901e+00 4.78880137e-01 1.73289105e-01 5.63701868e-01
-3.91007930e-01 7.31663406e-01 -3.88310730e-01 -7.81681359e-01
1.89724505e-01 -3.71716827e-01 -1.35816514e+00 -1.65238842e-01
8.31141472e-01 3.55120003e-02 -1.35313228e-01 8.90555441e-01
4.11871165e-01 1.96303893e-02 5.33325076e-01 -7.34568000e-01
-1.13334155e+00 6.17428184e-01 -1.82128817e-01 1.96232662e-01
6.34219050e-01 7.63085261e-02 1.41987252e+00 -8.22640538e-01
7.31334388e-01 1.37958038e+00 3.58627915e-01 4.65810120e-01
-1.04588604e+00 -5.56717575e-01 2.38028821e-02 3.44119817e-01
-1.63351858e+00 -8.95257652e-01 6.26908362e-01 -3.89160097e-01
9.49438274e-01 8.76634419e-02 1.15784347e-01 7.75126517e-01
-1.24604948e-01 6.74847543e-01 1.08295226e+00 -7.47020960e-01
-2.24608153e-01 3.31790328e-01 9.86573398e-02 9.05825555e-01
1.86578885e-01 -3.62957478e-01 -5.27158797e-01 -2.86787361e-01
2.53740937e-01 -7.01902032e-01 -3.41107279e-01 -2.95907527e-01
-1.04460323e+00 1.10617042e+00 4.50005382e-01 4.64063972e-01
-2.50144631e-01 -1.64046377e-01 3.78352433e-01 7.62223840e-01
3.15549403e-01 8.66035342e-01 -9.28939521e-01 1.36042722e-02
-6.37479126e-01 2.01460674e-01 9.37387288e-01 8.03782463e-01
1.02381980e+00 -2.67853200e-01 5.19220950e-03 1.32178617e+00
5.25909245e-01 5.66224456e-01 6.13800406e-01 -1.07024109e+00
7.99841940e-01 6.91011310e-01 2.17536509e-01 -6.38964415e-01
-2.61393815e-01 -2.03772962e-01 -3.14519346e-01 -6.46300614e-01
8.41469228e-01 -3.60239506e-01 -8.47977698e-01 2.02840519e+00
2.40893692e-01 -7.06362247e-01 3.26818019e-01 7.76597261e-01
8.75200331e-01 6.94515646e-01 4.26694065e-01 1.80640802e-01
1.72068596e+00 -9.60649550e-01 -5.69332480e-01 -4.85251009e-01
1.08565617e+00 -1.02931654e+00 1.48614669e+00 3.81994024e-02
-1.08448327e+00 -3.67377996e-01 -6.30967200e-01 -7.30772495e-01
-6.33301854e-01 2.42853269e-01 6.20455563e-01 6.87295079e-01
-1.14425981e+00 -3.09769839e-01 -5.61537921e-01 -4.58901256e-01
1.18883401e-01 -1.39587596e-01 -4.13481355e-01 -5.93996406e-01
-1.58411562e+00 1.26236403e+00 1.60299107e-01 -8.98392946e-02
-7.63832152e-01 -5.68328679e-01 -1.05883694e+00 -8.01492333e-02
2.84511507e-01 -6.49653494e-01 1.52190256e+00 -1.00598967e+00
-1.25633657e+00 1.11557138e+00 -6.19861841e-01 -1.63229004e-01
2.33256921e-01 -2.47014850e-01 -6.10206902e-01 -8.14248100e-02
4.21023697e-01 6.43001437e-01 2.48531222e-01 -1.13748848e+00
-4.59834516e-01 -4.97030944e-01 2.12445527e-01 5.56987524e-01
-4.83083799e-02 2.33670786e-01 -7.05562830e-01 -2.81884611e-01
-5.17501235e-02 -8.50581050e-01 -1.03081875e-01 -3.94549131e-01
-5.01907915e-02 -5.83519936e-01 1.41271679e-02 -1.23506141e+00
1.16810715e+00 -1.87343442e+00 -1.43177345e-01 9.23509523e-02
-3.30247045e-01 1.29218817e-01 -4.25968021e-01 7.07252800e-01
2.42858589e-01 1.97145596e-01 -1.10535897e-01 -1.98858619e-01
1.43327251e-01 3.35593939e-01 -2.78960258e-01 3.25370222e-01
3.15787554e-01 9.38443065e-01 -8.16263676e-01 -4.82924819e-01
-1.92765176e-01 4.18133140e-01 -6.75243080e-01 6.39078245e-02
-3.91378313e-01 4.78786170e-01 -5.69060087e-01 7.66468823e-01
4.20828313e-01 -3.26364636e-01 3.17240894e-01 7.68935978e-02
-6.49378523e-02 1.11311245e+00 -6.42315745e-01 2.10005760e+00
-8.05869281e-01 5.28664052e-01 8.54295865e-02 -6.16782129e-01
7.48612881e-01 4.78728414e-01 8.16222131e-02 -1.15677369e+00
5.48446691e-03 5.96058369e-01 2.50919849e-01 -3.68618637e-01
6.86379015e-01 -1.33713201e-01 -3.29409063e-01 3.83443058e-01
1.57190502e-01 -2.56401986e-01 3.60833913e-01 2.70179749e-01
7.33272612e-01 -1.65012240e-01 1.88796431e-01 -6.43140554e-01
8.12254548e-01 5.52321732e-01 3.15929621e-01 4.70990241e-01
-2.51971900e-01 1.29963934e-01 1.43709898e-01 -1.51936486e-01
-8.92286062e-01 -9.92232442e-01 -4.84815925e-01 1.56728709e+00
-1.04138196e-01 -4.10425663e-01 -7.02940166e-01 -5.69981694e-01
4.89481585e-03 1.01230288e+00 -2.24741280e-01 2.29512248e-02
-5.21170139e-01 -4.27749723e-01 7.95585215e-01 -2.08343025e-02
4.73959744e-01 -9.03150856e-01 -8.77876654e-02 3.59994057e-03
-5.73290765e-01 -9.90917563e-01 -4.91849571e-01 -1.78327337e-02
-4.10433531e-01 -8.33740711e-01 -6.82475805e-01 -1.01489449e+00
4.33076650e-01 4.79019918e-02 1.62775290e+00 1.62506789e-01
1.21961430e-01 6.88291371e-01 -3.77984881e-01 -2.84946084e-01
-5.41645050e-01 5.62074125e-01 -1.16041698e-01 -2.37076506e-01
7.76384711e-01 4.85583730e-02 -3.56837422e-01 1.20601594e-01
-6.16841972e-01 -1.86874524e-01 4.42750424e-01 5.78508973e-01
3.72971117e-01 -2.92516798e-01 7.39921987e-01 -8.78283381e-01
9.52558100e-01 -7.11427450e-01 -6.20726824e-01 8.04012001e-01
-3.33472133e-01 3.82093936e-01 4.28420752e-01 -6.82373196e-02
-1.28612924e+00 -2.22241402e-01 -4.40230668e-01 2.87513167e-01
-1.60846204e-01 9.17446196e-01 -1.70439214e-01 7.58450255e-02
7.85835683e-01 -2.28649564e-02 -3.18207800e-01 -6.58555746e-01
7.30092824e-01 9.48882759e-01 2.98196375e-01 -9.34876680e-01
5.51144123e-01 -1.64748318e-02 -8.71523082e-01 -8.75813484e-01
-1.08847558e+00 -5.63306689e-01 -6.44486964e-01 -9.04775504e-03
8.90181720e-01 -1.35493267e+00 -3.74848396e-01 1.30682454e-01
-1.08209431e+00 -5.15698373e-01 1.32121310e-01 3.77297252e-01
-1.86016232e-01 9.61406007e-02 -4.40190375e-01 -4.58363056e-01
-2.95921266e-01 -1.17267990e+00 8.34244967e-01 1.12043321e-01
-3.05249184e-01 -1.26400423e+00 4.54640567e-01 1.04283428e+00
4.61401433e-01 -4.76260513e-01 1.09362233e+00 -7.93902457e-01
-4.21341419e-01 1.86115339e-01 -1.72662050e-01 1.85232341e-01
2.34755293e-01 -2.93789953e-01 -9.27182972e-01 -3.20880502e-01
-2.58623004e-01 -8.01983416e-01 6.92398846e-01 1.30274385e-01
6.36204898e-01 -3.65216583e-01 1.20386288e-01 2.73432821e-01
1.34418440e+00 -1.76430643e-01 3.28721553e-01 4.57252115e-01
5.97963154e-01 9.07916367e-01 3.75253052e-01 -1.96262434e-01
1.24770737e+00 5.04027784e-01 -2.71697581e-01 1.56193107e-01
-2.23369151e-01 -3.56164157e-01 4.13645416e-01 1.20875227e+00
3.39021504e-01 -2.01741010e-01 -1.56691813e+00 1.05544496e+00
-1.33996654e+00 -5.96651495e-01 7.57357329e-02 2.21351671e+00
1.44988298e+00 -3.88901740e-01 -1.23053133e-01 -6.25133991e-01
3.01851928e-01 -2.16211509e-02 -4.64973450e-01 -6.50135756e-01
-2.17528686e-01 2.35101148e-01 4.79534417e-01 1.11619806e+00
-8.27161908e-01 1.39844298e+00 6.44055462e+00 7.53264844e-01
-1.12619388e+00 3.07065189e-01 6.39820933e-01 3.84784579e-01
-8.77550542e-01 2.88439728e-02 -8.18316340e-01 8.11415389e-02
1.32205069e+00 -4.42676485e-01 5.22046685e-01 5.32916725e-01
-1.19157687e-01 -2.36950275e-02 -8.17481458e-01 5.57353377e-01
7.53450096e-02 -9.43155468e-01 2.58590430e-01 -1.87101007e-01
7.68677533e-01 7.07074702e-01 -1.37391716e-01 5.63837826e-01
8.85756254e-01 -1.11037838e+00 5.37012935e-01 1.84806854e-01
6.04595184e-01 -6.58390999e-01 4.83252317e-01 3.60412508e-01
-9.55565214e-01 1.57861188e-01 -3.11097115e-01 1.32923812e-01
2.12810636e-01 2.96855062e-01 -8.75202656e-01 5.41072428e-01
7.11714029e-01 2.51204520e-01 -9.93938327e-01 6.57710075e-01
-4.31121677e-01 7.40966022e-01 -2.39030972e-01 6.38019145e-02
4.07697469e-01 -2.32406855e-01 1.00473225e-01 1.23275721e+00
2.30563864e-01 -1.71970844e-01 8.20796490e-02 5.99650085e-01
-2.50830382e-01 6.44448280e-01 -4.65662956e-01 -3.21614653e-01
6.94706380e-01 8.69591475e-01 4.79562134e-02 -3.74441087e-01
-7.81559646e-01 7.74547458e-01 8.15633118e-01 5.70696235e-01
-2.87180215e-01 -2.02203944e-01 8.55581999e-01 -3.16649862e-03
-5.86226396e-02 -5.08818805e-01 -8.57830420e-02 -1.49367476e+00
8.20671842e-02 -1.21790469e+00 8.29080582e-01 -7.35034525e-01
-1.64838636e+00 6.66285574e-01 -2.78908908e-01 -6.16181850e-01
-5.92924237e-01 -6.35010362e-01 1.97789557e-02 1.28567612e+00
-1.86961389e+00 -1.34341586e+00 4.86908443e-02 7.97812223e-01
4.04326737e-01 -6.48069233e-02 1.05396271e+00 4.99977291e-01
-3.87322396e-01 7.77831912e-01 1.43222734e-01 3.80090892e-01
1.13349664e+00 -1.12378252e+00 4.35955018e-01 7.74750531e-01
6.22101665e-01 1.09153068e+00 4.75211769e-01 -4.93890315e-01
-1.36601269e+00 -8.14943016e-01 1.65752578e+00 -8.03419173e-01
9.54294801e-01 -4.07519817e-01 -1.31378794e+00 9.88443315e-01
8.38607490e-01 -5.03415167e-01 9.17160511e-01 7.33698905e-01
-7.81102300e-01 -1.62359197e-02 -8.85322332e-01 5.74922144e-01
5.89302123e-01 -1.03084958e+00 -8.63163769e-01 4.10031408e-01
7.60351598e-01 -3.48599941e-01 -1.12084639e+00 1.80599511e-01
2.99885660e-01 -4.90650207e-01 8.92031908e-01 -7.19994187e-01
7.54680634e-02 -2.23894790e-01 -3.72670352e-01 -1.38479424e+00
-1.87550768e-01 -2.97065884e-01 4.91171539e-01 1.25521159e+00
9.55455780e-01 -9.20960009e-01 2.18242154e-01 7.09021926e-01
-1.23618141e-01 -4.62754995e-01 -1.06572413e+00 -5.41537702e-01
8.14218998e-01 -4.38486665e-01 6.32105410e-01 1.44765115e+00
-1.87497512e-02 7.21744180e-01 1.04551025e-01 4.37532634e-01
2.40920439e-01 2.77863666e-02 6.67022705e-01 -8.68244827e-01
-1.65337190e-01 -4.25195336e-01 1.87373549e-01 -1.10896170e+00
2.76216656e-01 -1.16122890e+00 8.11391473e-02 -1.64064109e+00
-1.03856949e-02 -8.42339993e-01 -1.84974909e-01 7.56546557e-01
-3.51709008e-01 -2.08860058e-02 -1.62854344e-01 2.36257210e-01
-6.10233843e-01 5.18470883e-01 1.15463889e+00 7.64559656e-02
-1.18544810e-01 -3.29275250e-01 -1.02096760e+00 3.94523621e-01
9.93969142e-01 -2.14429289e-01 -4.26648527e-01 -1.35129762e+00
3.37906808e-01 -1.74268216e-01 -6.14422224e-02 -5.08478940e-01
2.39557505e-01 -8.38520676e-02 1.68377403e-02 -2.31414139e-01
3.17687720e-01 -5.30678868e-01 -2.72265047e-01 1.80524364e-01
-5.29708207e-01 3.64486754e-01 4.99869794e-01 -1.07864149e-01
-5.74620545e-01 -2.07790017e-01 5.61122119e-01 -2.09919557e-01
-7.19262719e-01 8.58354867e-02 -4.47038531e-01 7.53089130e-01
3.35999519e-01 4.30366099e-01 -5.17280340e-01 -5.45825064e-01
-1.92822307e-01 5.97710192e-01 5.47024429e-01 6.71731591e-01
-9.28883031e-02 -1.22855318e+00 -1.04138136e+00 2.24006698e-01
4.95128423e-01 -2.55877614e-01 1.57154232e-01 4.49713767e-01
-6.61875308e-01 1.03118026e+00 2.58235224e-02 -2.69307554e-01
-8.32617462e-01 2.12082893e-01 4.67343301e-01 -2.33708769e-01
2.29197353e-01 9.99805927e-01 2.15869043e-02 -1.33367193e+00
-7.43787130e-03 -9.45486650e-02 -2.27700055e-01 1.72957052e-02
3.78994346e-01 -1.05554029e-01 2.00194463e-01 -9.96477067e-01
-4.35090601e-01 4.97742295e-01 -1.83206812e-01 -4.39017087e-01
9.27244246e-01 -4.00178909e-01 -1.84441730e-01 4.71833974e-01
1.08599782e+00 3.97320986e-01 -4.72886205e-01 -5.90577126e-01
3.49124879e-01 -2.15143770e-01 4.27730754e-02 -1.09924638e+00
-7.60215580e-01 8.21662664e-01 2.34690830e-01 6.74168319e-02
8.80620003e-01 4.54796910e-01 6.11645460e-01 6.23987973e-01
5.21080375e-01 -1.01150048e+00 -3.63257319e-01 8.72269332e-01
9.08285141e-01 -1.56285727e+00 -2.41672724e-01 -1.30067691e-01
-8.64515603e-01 4.15069848e-01 5.48420548e-01 1.84969798e-01
6.59825087e-01 -3.88203382e-01 7.80025899e-01 -3.62981260e-01
-8.59840333e-01 -3.98567140e-01 6.48664832e-01 2.73705453e-01
1.09437370e+00 3.28078628e-01 -3.46421093e-01 3.47894967e-01
-8.84874940e-01 -2.61188954e-01 7.56895468e-02 7.47577906e-01
-3.48049164e-01 -1.43815506e+00 -2.13758498e-01 2.06884906e-01
-4.66056168e-01 -5.39055169e-01 -6.10691190e-01 8.05927396e-01
-6.79929629e-02 1.28310657e+00 9.50223505e-02 6.79299533e-02
2.70881385e-01 5.03525019e-01 3.11994731e-01 -7.26018727e-01
-5.63692629e-01 -2.98531830e-01 6.24360561e-01 -4.08932835e-01
-2.97995687e-01 -7.73852468e-01 -1.07861030e+00 -2.51383811e-01
-1.09469049e-01 5.71766794e-01 7.18339324e-01 8.86583865e-01
5.53558230e-01 -4.83789407e-02 1.95106879e-01 -1.99583126e-03
-3.62679631e-01 -1.14377165e+00 -1.65019095e-01 3.83808404e-01
2.83143342e-01 -2.74747431e-01 -3.83224070e-01 9.46875568e-03]
|
[11.129227638244629, 9.3582124710083]
|
ac3102c6-7083-41a3-968b-2a52df1eb955
|
do-neural-topic-models-really-need-dropout
|
2303.15973
| null |
https://arxiv.org/abs/2303.15973v1
|
https://arxiv.org/pdf/2303.15973v1.pdf
|
Do Neural Topic Models Really Need Dropout? Analysis of the Effect of Dropout in Topic Modeling
|
Dropout is a widely used regularization trick to resolve the overfitting issue in large feedforward neural networks trained on a small dataset, which performs poorly on the held-out test subset. Although the effectiveness of this regularization trick has been extensively studied for convolutional neural networks, there is a lack of analysis of it for unsupervised models and in particular, VAE-based neural topic models. In this paper, we have analyzed the consequences of dropout in the encoder as well as in the decoder of the VAE architecture in three widely used neural topic models, namely, contextualized topic model (CTM), ProdLDA, and embedded topic model (ETM) using four publicly available datasets. We characterize the dropout effect on these models in terms of the quality and predictive performance of the generated topics.
|
['Debarshi Kumar Sanyal', 'Avishek Lahiri', 'Suman Adhya']
|
2023-03-28
| null | null | null | null |
['topic-models']
|
['natural-language-processing']
|
[-5.06040715e-02 4.51736599e-01 -1.10973090e-01 -5.86845219e-01
-6.87265575e-01 -3.26704144e-01 5.25696218e-01 -8.98306258e-03
-3.23150456e-01 6.86736763e-01 2.78716713e-01 -2.71935761e-01
-1.28647640e-01 -6.97582006e-01 -1.08963907e+00 -7.61059821e-01
2.89614499e-01 5.48485219e-01 1.58299297e-01 4.13984895e-01
-3.34721178e-01 -1.22262493e-01 -1.25312626e+00 1.44292573e-02
9.55325961e-01 7.01610327e-01 5.63982539e-02 1.37254640e-01
-1.33399546e-01 6.65212631e-01 -8.39061618e-01 -5.64412653e-01
-2.05381364e-01 -4.32226360e-01 -4.99000937e-01 1.12270065e-01
7.15765238e-01 -3.86383772e-01 -6.03126109e-01 9.03593719e-01
2.30386227e-01 1.47318155e-01 9.40650642e-01 -1.31432140e+00
-5.98737597e-01 1.13903975e+00 -2.17757672e-01 -4.88723740e-02
-6.82531297e-01 1.75610073e-02 1.09993041e+00 -8.72884989e-01
9.01216507e-01 1.19337165e+00 8.02589357e-01 5.00415385e-01
-1.44052649e+00 -7.78503299e-01 2.57379740e-01 -2.38171637e-01
-1.13741779e+00 -3.16357404e-01 7.74365067e-01 -5.46433687e-01
7.67132521e-01 -3.73206854e-01 3.99696678e-01 1.71691000e+00
4.05981272e-01 9.68518615e-01 8.38788688e-01 -1.49921343e-01
3.98664176e-01 5.12625813e-01 9.68319356e-01 3.10593367e-01
5.29618680e-01 -1.51900470e-01 -7.91843534e-01 -3.17057371e-01
5.86542368e-01 -3.83948572e-02 -2.07094118e-01 -3.16162676e-01
-5.91688693e-01 1.30431116e+00 2.31021345e-01 8.15010890e-02
-3.75561088e-01 3.45543593e-01 4.43448514e-01 2.14755699e-01
1.19417560e+00 2.75133640e-01 -6.45611167e-01 -1.53751299e-01
-1.35434544e+00 3.15212429e-01 7.86027908e-01 9.60774839e-01
6.43316686e-01 4.94358003e-01 -3.53885353e-01 9.75085974e-01
3.79215330e-01 2.04875410e-01 5.84635079e-01 -4.12635952e-01
4.92082566e-01 4.55253541e-01 -4.07777220e-01 -4.91312623e-01
-2.79566765e-01 -7.72514224e-01 -9.31881964e-01 -5.93324080e-02
1.74849093e-01 -6.60733938e-01 -1.12205029e+00 1.97405696e+00
-1.61088958e-01 2.23100245e-01 9.36663747e-02 4.82749373e-01
1.17259336e+00 7.58545995e-01 3.18520367e-01 1.13379225e-01
9.02113140e-01 -1.06972814e+00 -8.91659677e-01 -3.47943306e-01
5.98469436e-01 -3.20781291e-01 1.07617295e+00 2.85150081e-01
-1.09226525e+00 -3.20707649e-01 -9.97226179e-01 -2.22118333e-01
-3.94735515e-01 4.62577134e-01 6.38431787e-01 6.48043931e-01
-1.22169566e+00 5.57193875e-01 -9.40469980e-01 -3.37066501e-01
8.03507388e-01 4.15905923e-01 -1.32691979e-01 -1.45161584e-01
-1.25718677e+00 7.89249241e-01 3.48174274e-01 -2.40883622e-02
-1.27465427e+00 -1.01766002e+00 -6.80642545e-01 5.33369064e-01
9.21757445e-02 -5.41713953e-01 1.18889487e+00 -9.25123096e-01
-1.36789680e+00 5.29766917e-01 -1.21806957e-01 -8.72262895e-01
2.98209965e-01 -5.12310624e-01 7.05617592e-02 -3.14172566e-01
-1.63540661e-01 9.40713644e-01 7.62154102e-01 -9.44373012e-01
-1.15038380e-01 -3.29057574e-01 -4.67294037e-01 2.90598907e-03
-6.31803095e-01 -1.98998332e-01 -3.27385396e-01 -8.23561668e-01
-1.28929228e-01 -8.45163882e-01 -3.89859197e-03 -2.66842455e-01
-6.44120157e-01 -5.43262482e-01 9.68463957e-01 -7.86445260e-01
9.85842764e-01 -2.28662014e+00 6.44785305e-03 2.98924721e-03
1.69965208e-01 2.81863213e-01 -1.25863537e-01 2.53610015e-01
6.36640415e-02 2.77189493e-01 -2.50136733e-01 -9.27812397e-01
-7.65049458e-02 8.38770196e-02 -7.66651511e-01 3.11248630e-01
4.16560203e-01 6.83694243e-01 -3.76120508e-01 -1.21362604e-01
-9.55117270e-02 8.75852466e-01 -6.00653470e-01 1.86724052e-01
-5.82225442e-01 1.00386627e-02 -3.23548585e-01 1.94888636e-01
3.95269722e-01 -2.84197956e-01 -8.87098312e-02 3.15567464e-01
1.52947456e-01 7.35340476e-01 -4.80546117e-01 1.60644889e+00
-2.97780633e-01 1.32884347e+00 -2.97020346e-01 -5.52542448e-01
9.70587611e-01 7.59751737e-01 1.06365733e-01 -2.10163936e-01
1.81613803e-01 -4.88651684e-04 -2.26041079e-02 -5.23244515e-02
7.30007470e-01 -7.55323023e-02 2.55393893e-01 4.82316643e-01
7.44307339e-01 2.70355046e-01 -4.52768095e-02 2.49076650e-01
9.60744739e-01 1.31445587e-01 -3.90120625e-01 -4.00892466e-01
-5.02262890e-01 5.85070252e-02 4.31761771e-01 8.35319042e-01
1.90324470e-01 9.42014694e-01 9.26862299e-01 -8.94666165e-02
-1.10136032e+00 -8.53352785e-01 -5.12076497e-01 1.00014675e+00
-5.73842108e-01 -5.42498648e-01 -9.12447870e-01 -5.14221072e-01
-1.01135008e-01 9.98543262e-01 -7.87117481e-01 -6.89299166e-01
-2.10551754e-01 -1.10264874e+00 6.05480552e-01 6.02449358e-01
5.00563979e-01 -1.02752197e+00 -4.09391195e-01 1.06382452e-01
1.64380800e-02 -9.00464475e-01 -1.26519620e-01 6.35241032e-01
-1.42625701e+00 -7.63117373e-01 -7.57675290e-01 -5.45275509e-01
4.18748289e-01 -1.53809577e-01 1.24089992e+00 -4.62131172e-01
8.70695412e-02 1.65829167e-01 -1.00133710e-01 -6.84304178e-01
-2.45073348e-01 6.18539572e-01 -2.56036550e-01 -8.25774074e-02
3.98726583e-01 -5.08113921e-01 -1.47861019e-01 1.44211352e-02
-9.90740061e-01 4.00934406e-02 5.18073022e-01 8.69812369e-01
3.74467492e-01 2.03230843e-01 5.96654058e-01 -1.40569043e+00
7.58684397e-01 -7.81236470e-01 -5.29439926e-01 9.73213166e-02
-7.60226071e-01 1.44864336e-01 2.41514713e-01 -6.25347018e-01
-1.19368541e+00 -3.94881189e-01 -3.60082611e-02 -7.54212439e-01
-6.49657920e-02 8.01014900e-01 -8.73157680e-02 5.05647361e-01
4.29167300e-01 7.27778152e-02 -9.24653113e-02 -7.07778931e-01
1.68277770e-01 5.46181321e-01 9.86368209e-02 -3.11653465e-01
4.10353750e-01 1.49540693e-01 -3.52137238e-01 -9.46325302e-01
-1.03292692e+00 -1.34433016e-01 -1.73620969e-01 1.07398517e-01
8.68082583e-01 -1.22207761e+00 -5.46344444e-02 5.96234977e-01
-1.25553763e+00 -7.27643132e-01 -3.43372643e-01 5.94730377e-01
-3.32863927e-01 -5.09412348e-01 -7.26956904e-01 -5.95510364e-01
-3.99316102e-01 -1.18302691e+00 8.13375950e-01 2.47119173e-01
-3.35544109e-01 -1.27881455e+00 1.02721199e-01 -1.04844034e-01
7.62765110e-01 1.12996707e-02 1.37914228e+00 -1.07138383e+00
-4.85143393e-01 -2.06149384e-01 -2.44735759e-02 4.50544715e-01
-1.51400581e-01 2.32660204e-01 -1.39554799e+00 -2.13819236e-01
1.94410160e-01 -5.03458798e-01 1.54916549e+00 9.98813987e-01
9.09003556e-01 -1.12857588e-01 -3.10665578e-01 6.09505296e-01
1.33360386e+00 -1.30942777e-01 6.82769179e-01 2.23116666e-01
4.56657797e-01 5.40456295e-01 1.25856146e-01 -7.33021125e-02
2.98985355e-02 5.04572868e-01 3.16814631e-01 -9.15344618e-03
-1.93184108e-01 -3.29411507e-01 4.76408839e-01 6.27024233e-01
3.82134736e-01 -5.39009213e-01 -9.69445705e-01 6.07620358e-01
-1.80035615e+00 -4.53832299e-01 -6.12533316e-02 2.02737761e+00
8.38646531e-01 2.92697370e-01 -1.69426277e-01 -2.59767711e-01
5.67750275e-01 2.54724979e-01 -5.18605411e-01 -2.24263489e-01
-3.52355868e-01 1.20149709e-01 3.17112416e-01 1.05661310e-01
-1.27014375e+00 1.05143738e+00 6.68634558e+00 8.29378843e-01
-1.05080366e+00 4.94376272e-01 9.27698851e-01 -3.38773072e-01
-2.21940875e-01 2.00407170e-02 -1.17692161e+00 4.10513729e-01
1.50228179e+00 1.14993811e-01 -2.74015337e-01 1.27684522e+00
1.58924773e-01 -3.18278298e-02 -1.16408408e+00 4.97868121e-01
4.65900451e-03 -1.25863993e+00 1.60550669e-01 1.59059048e-01
9.69369113e-01 2.76695758e-01 6.08279407e-01 6.41258717e-01
3.97332013e-01 -1.04027665e+00 6.95397854e-01 3.60331863e-01
5.44417083e-01 -7.66522765e-01 8.44884932e-01 1.31605685e-01
-2.57649720e-01 2.72079319e-01 -7.45681524e-01 3.41580212e-01
-8.78033116e-02 9.36915934e-01 -7.06599832e-01 -2.70274505e-02
5.45559227e-01 6.56223238e-01 -6.06047630e-01 1.17668855e+00
-2.59546489e-01 1.54325461e+00 -1.53872862e-01 1.73847198e-01
4.97500747e-01 -1.97011799e-01 5.30097067e-01 8.50449204e-01
1.26157790e-01 -5.86531162e-01 -3.58176321e-01 1.26952493e+00
-4.52664196e-01 -1.81124970e-01 -5.84892094e-01 -2.10251123e-01
1.80457368e-01 8.82394135e-01 -6.42952561e-01 -1.57249063e-01
-3.68220747e-01 4.91891414e-01 5.21185756e-01 8.08236539e-01
-8.06100368e-01 -1.31926939e-01 6.09379709e-01 1.39121696e-01
3.89499992e-01 -1.31591514e-01 -4.14734066e-01 -1.04532993e+00
5.14229052e-02 -5.59603393e-01 7.95385167e-02 -8.01917076e-01
-1.16717827e+00 8.06987405e-01 1.91977158e-01 -5.45364618e-01
-2.26580799e-01 -4.18261141e-01 -1.02963686e+00 9.91222441e-01
-1.36139977e+00 -9.84275937e-01 -8.63781795e-02 2.86846250e-01
8.01083982e-01 -4.31197822e-01 6.49373949e-01 1.87251925e-01
-9.19775546e-01 5.56321740e-01 6.62436545e-01 1.69163674e-01
7.60769546e-01 -9.88530874e-01 5.16435623e-01 6.02398813e-01
1.52774155e-01 8.62840295e-01 7.51660228e-01 -6.76823676e-01
-8.63924801e-01 -1.66608131e+00 7.66576111e-01 -3.10315818e-01
3.08264822e-01 -6.62762105e-01 -1.24500704e+00 1.19621813e+00
4.32716429e-01 -4.38194841e-01 7.18606889e-01 6.22071505e-01
-3.78016055e-01 1.10014252e-01 -6.61126494e-01 4.14913714e-01
3.46114933e-01 -3.60414833e-01 -4.97749656e-01 3.75598997e-01
1.18871975e+00 -1.74640998e-01 -6.25951648e-01 6.35466054e-02
2.23247558e-01 -8.13191831e-01 5.80627084e-01 -8.90457749e-01
8.75861466e-01 5.75275838e-01 4.64562774e-02 -1.51513863e+00
-1.51604004e-02 -3.49518210e-01 -2.10850149e-01 1.72727466e+00
7.64539480e-01 -6.03278935e-01 1.14237118e+00 7.49852836e-01
-2.49774113e-01 -9.10088241e-01 -8.92587125e-01 -6.83842421e-01
5.55188596e-01 -3.23729932e-01 2.74474919e-01 7.24375546e-01
-5.11717141e-01 5.15421987e-01 -3.08053285e-01 -9.72246081e-02
5.05537510e-01 -3.54603171e-01 4.58326787e-01 -1.47866797e+00
-3.72401811e-02 -4.03546542e-01 -4.73879501e-02 -6.32264078e-01
5.89166403e-01 -7.69857943e-01 -4.52876370e-03 -1.50160706e+00
3.49230796e-01 -4.90441233e-01 -1.45762905e-01 3.31332862e-01
-1.88467756e-01 -1.61746353e-01 -4.24153768e-02 4.25210297e-01
-3.54520291e-01 1.03268957e+00 3.97121280e-01 -5.39302044e-02
-2.12283164e-01 3.44523638e-01 -5.45353830e-01 7.56764472e-01
9.26603019e-01 -9.54880059e-01 -6.00066841e-01 -7.66236484e-01
1.57239228e-01 -2.15245411e-01 4.56059426e-01 -1.12640166e+00
2.20071837e-01 3.96099716e-01 2.91912943e-01 -7.45225668e-01
4.77897465e-01 -6.12481058e-01 -4.22340520e-02 7.95148127e-03
-9.27976966e-01 -1.68748945e-01 3.04217070e-01 6.84874594e-01
-3.87296736e-01 -5.64024210e-01 5.46457827e-01 8.54326338e-02
-1.41811445e-01 4.23679024e-01 -4.08208728e-01 3.45085323e-01
5.16478717e-01 1.55388653e-01 -5.77593446e-01 -3.87079775e-01
-5.67834377e-01 1.45289823e-01 2.47025892e-01 4.17715132e-01
5.26405931e-01 -1.07085800e+00 -7.62453854e-01 3.91753316e-02
-2.15339229e-01 2.89569676e-01 2.89886802e-01 6.91193521e-01
2.72352174e-02 8.42568874e-01 1.07208855e-01 -5.54400742e-01
-8.65061343e-01 1.05904855e-01 2.40450606e-01 -4.26798910e-01
-6.57167196e-01 9.39436793e-01 5.41904569e-01 -2.64674127e-01
7.33398080e-01 -4.93059188e-01 -4.48716395e-02 1.08520621e-02
3.18866521e-02 2.12412968e-01 1.29506558e-01 -1.49976000e-01
1.02182753e-01 6.23781942e-02 -4.46581215e-01 -7.10610747e-02
1.49310827e+00 2.52976686e-01 -1.10734114e-03 9.73265052e-01
1.37022352e+00 -6.33283496e-01 -1.56039298e+00 -3.19671869e-01
3.94985043e-02 2.00539574e-01 4.21120942e-01 -5.23995340e-01
-1.07919538e+00 1.20984590e+00 4.49725091e-01 8.13034084e-03
6.37396872e-01 -1.95002370e-02 5.24458826e-01 2.99128562e-01
-1.60067871e-01 -8.65652621e-01 6.44133762e-02 7.95550227e-01
6.68923557e-01 -1.16468251e+00 -9.51223150e-02 -2.46366233e-01
-7.07337081e-01 7.90110528e-01 6.28841877e-01 -3.08377773e-01
1.04974651e+00 1.23575196e-01 -2.05677420e-01 -3.40478957e-01
-1.31583858e+00 3.09286237e-01 2.79669464e-01 5.19224226e-01
6.83695674e-01 -2.18973160e-01 2.56954581e-01 1.07837629e+00
-2.51746207e-01 -1.53032079e-01 6.48488104e-01 7.12145388e-01
-3.36968154e-01 -6.03478849e-01 1.25098526e-01 7.69341946e-01
-5.49394071e-01 -4.89813179e-01 -5.12087286e-01 9.86150622e-01
-2.39508465e-01 6.63298905e-01 2.39133835e-01 -1.16807662e-01
7.95099977e-03 7.60847270e-01 -1.64105937e-01 -9.36224639e-01
-9.75528777e-01 2.11762078e-02 7.29540437e-02 -1.62347704e-01
-8.63152221e-02 -5.94490409e-01 -7.57471144e-01 1.41395167e-01
-6.21022642e-01 1.92615941e-01 8.52373719e-01 9.06085610e-01
5.76908410e-01 8.96482229e-01 6.92351386e-02 -3.77690911e-01
-6.59466624e-01 -1.48525739e+00 -7.87933648e-01 2.42534220e-01
2.05386296e-01 -7.51349509e-01 -4.91441280e-01 7.72093516e-03]
|
[10.449676513671875, 6.956477165222168]
|
9c5bb0de-8d91-4cd5-8bf1-6f3a69332f91
|
mist-towards-improved-adversarial-examples
|
2305.12683
| null |
https://arxiv.org/abs/2305.12683v1
|
https://arxiv.org/pdf/2305.12683v1.pdf
|
Mist: Towards Improved Adversarial Examples for Diffusion Models
|
Diffusion Models (DMs) have empowered great success in artificial-intelligence-generated content, especially in artwork creation, yet raising new concerns in intellectual properties and copyright. For example, infringers can make profits by imitating non-authorized human-created paintings with DMs. Recent researches suggest that various adversarial examples for diffusion models can be effective tools against these copyright infringements. However, current adversarial examples show weakness in transferability over different painting-imitating methods and robustness under straightforward adversarial defense, for example, noise purification. We surprisingly find that the transferability of adversarial examples can be significantly enhanced by exploiting a fused and modified adversarial loss term under consistent parameters. In this work, we comprehensively evaluate the cross-method transferability of adversarial examples. The experimental observation shows that our method generates more transferable adversarial examples with even stronger robustness against the simple adversarial defense.
|
['Xiaoyu Wu', 'Chumeng Liang']
|
2023-05-22
| null | null | null | null |
['adversarial-defense']
|
['adversarial']
|
[ 5.27519226e-01 3.33783507e-01 5.78527004e-02 4.16622870e-02
-6.32227421e-01 -1.31892085e+00 9.33932304e-01 -6.39828384e-01
-1.44397676e-01 8.59007895e-01 -2.78721172e-02 -2.31672302e-01
-3.46311539e-01 -9.99697626e-01 -7.08926141e-01 -6.32248223e-01
1.65676430e-01 2.97751218e-01 -1.48307800e-01 -2.79505581e-01
3.22101712e-01 8.37288916e-01 -7.78461635e-01 -2.34651677e-02
1.13763595e+00 8.14105272e-01 -3.14712524e-01 7.35117733e-01
-2.48917878e-01 9.75746751e-01 -1.37562633e+00 -1.46920109e+00
8.07247102e-01 -4.04152572e-01 -5.40292680e-01 -7.12611154e-02
5.57669103e-01 -3.96006107e-01 -7.10273027e-01 1.45502841e+00
6.18943989e-01 -2.73306698e-01 8.94439518e-01 -1.74087858e+00
-1.55360496e+00 6.10121548e-01 -5.23816228e-01 -2.18176231e-01
2.36434281e-01 6.80815160e-01 6.16993666e-01 -1.07846260e-01
8.52852285e-01 1.43871379e+00 3.81689012e-01 1.01138580e+00
-1.13589084e+00 -1.14148521e+00 -7.18122125e-02 -2.12744266e-01
-9.55598891e-01 2.26744842e-02 1.10656428e+00 -3.64981860e-01
1.46895692e-01 7.20609665e-01 2.42810875e-01 1.65536618e+00
2.01522738e-01 9.75550175e-01 1.42024541e+00 -4.60051605e-03
1.89145386e-01 3.19271863e-01 -5.20637691e-01 3.25755954e-01
3.64547968e-01 3.39696616e-01 -1.28461748e-01 -3.82221013e-01
1.15230417e+00 -1.05717175e-01 -1.36287093e-01 9.03702527e-03
-8.93832803e-01 7.38896132e-01 5.20707011e-01 2.09394187e-01
-1.96967214e-01 2.96732724e-01 2.48292983e-01 4.75413382e-01
3.34922105e-01 1.12773359e+00 -1.96770653e-01 5.56720831e-02
-4.20657098e-01 4.27486777e-01 7.40350842e-01 1.25556970e+00
3.52371544e-01 3.51363808e-01 -5.49443781e-01 6.92680836e-01
-7.67170861e-02 9.05955672e-01 1.61730587e-01 -1.40053976e+00
4.91083622e-01 2.52207816e-01 1.07194543e-01 -1.09780157e+00
3.38125527e-01 -4.39288229e-01 -1.17399430e+00 6.76151156e-01
4.29060340e-01 -4.71976668e-01 -8.94683003e-01 1.59233499e+00
4.23376821e-02 2.09014282e-01 1.32531866e-01 6.07002914e-01
4.81038153e-01 2.72417188e-01 2.24745780e-01 9.05762017e-02
1.04742515e+00 -9.10330653e-01 -9.88431513e-01 -1.12398252e-01
-8.48580077e-02 -8.97767425e-01 1.02697194e+00 3.40777934e-01
-1.07240117e+00 -2.95511961e-01 -9.65428770e-01 2.73815393e-01
-3.72094870e-01 -4.05843437e-01 9.94442344e-01 1.36025977e+00
-5.10478020e-01 9.08306479e-01 8.74081701e-02 2.75804579e-01
1.31236732e+00 2.50584543e-01 -2.78698355e-01 -4.31938805e-02
-1.33437526e+00 6.61474109e-01 -1.52558461e-01 -2.48707071e-01
-1.29451287e+00 -1.02955627e+00 -2.99303919e-01 -1.71268165e-01
4.40352887e-01 -8.42012525e-01 9.55801666e-01 -1.25293696e+00
-1.75159955e+00 7.96794951e-01 7.04718053e-01 -5.81502676e-01
1.24749672e+00 -3.01379532e-01 -6.20773077e-01 1.54522121e-01
5.45891076e-02 5.53994834e-01 1.30921698e+00 -1.50768256e+00
-1.18920438e-01 -1.75451875e-01 5.77333868e-01 -1.72389939e-01
-7.55239666e-01 6.69975877e-02 -1.74895495e-01 -1.48814774e+00
-7.14816570e-01 -1.01437986e+00 -3.69533449e-01 4.86205429e-01
-8.20303559e-01 1.01610206e-01 9.86144960e-01 -4.59553927e-01
7.45430171e-01 -2.26888108e+00 7.27978796e-02 2.44852543e-01
1.97761446e-01 6.07794166e-01 -4.57814157e-01 1.84738412e-01
1.86152175e-01 9.08459485e-01 -2.94246376e-01 3.64707410e-02
4.53960210e-01 1.51599377e-01 -7.35004246e-01 1.71215668e-01
4.92050827e-01 1.24898708e+00 -8.51672113e-01 -3.48917246e-01
-1.09782785e-01 4.36690122e-01 -2.96577781e-01 2.28333503e-01
-3.19019228e-01 5.34873128e-01 -7.84711778e-01 7.51000106e-01
1.04286540e+00 1.26173392e-01 -2.73413539e-01 2.19495341e-01
7.09729671e-01 -6.47146881e-01 -8.29755902e-01 1.45040333e+00
-1.73104107e-01 7.77538598e-01 9.66058373e-02 -3.14324856e-01
8.11311066e-01 9.01892558e-02 7.33619705e-02 -4.80969816e-01
9.67471749e-02 9.69947502e-02 -4.73541282e-02 -3.56160223e-01
3.59592259e-01 -3.60361516e-01 -1.21925987e-01 4.75468457e-01
-3.09135735e-01 -6.97131336e-01 -2.20876858e-01 5.19548178e-01
1.22130346e+00 2.50792317e-02 -3.97075146e-01 1.45103171e-01
3.73774797e-01 -2.18385711e-01 4.26821381e-01 9.81118500e-01
-2.81407893e-01 6.15834475e-01 7.21436560e-01 -2.14758329e-02
-1.15221417e+00 -1.37740123e+00 1.77169312e-02 4.89973366e-01
1.19413167e-01 1.05282977e-01 -9.48863149e-01 -1.17968082e+00
4.34380144e-01 8.32396805e-01 -6.91271365e-01 -3.17232370e-01
-2.67914057e-01 -4.13474172e-01 1.58218789e+00 4.03906614e-01
8.75273108e-01 -1.02477682e+00 3.19256991e-01 -1.54645965e-01
3.51367533e-01 -9.42303419e-01 -4.61570323e-01 -4.19654757e-01
-6.10530317e-01 -7.91592360e-01 -1.34779060e+00 -3.34999382e-01
6.12775385e-01 -1.16019279e-01 6.94844246e-01 -1.21822534e-02
-5.12562931e-01 5.31366467e-01 -9.72133949e-02 -9.38342214e-01
-8.10436130e-01 -3.28143716e-01 7.19845481e-03 -4.79190797e-02
-1.08427353e-01 -6.21241689e-01 -4.82934594e-01 4.88924593e-01
-1.28068817e+00 -6.91467524e-01 6.17091775e-01 5.48242390e-01
2.01938629e-01 3.99529576e-01 8.52932036e-01 -1.03436315e+00
1.15523171e+00 -3.42873991e-01 -5.36251009e-01 2.17731208e-01
-6.30866647e-01 -1.06851801e-01 7.44274497e-01 -1.05017936e+00
-1.37348938e+00 -3.82463157e-01 1.70513332e-01 -7.28959441e-01
-1.25719264e-01 -2.31620044e-01 -5.46348751e-01 -6.14351988e-01
8.00825536e-01 -2.10966527e-01 1.77154392e-02 -3.51647913e-01
7.46708512e-01 5.39548874e-01 6.18417323e-01 -8.75568867e-01
1.72165537e+00 6.09852135e-01 2.24699825e-01 -5.02367139e-01
-4.11189258e-01 4.33814824e-01 -5.42712472e-02 -2.37190962e-01
8.26633990e-01 -4.64892030e-01 -8.03866804e-01 7.64418006e-01
-1.13650584e+00 -2.94973217e-02 -6.72866642e-01 7.89506137e-02
-4.14541334e-01 5.60011804e-01 -6.51349366e-01 -8.99758101e-01
-4.43019755e-02 -9.84775484e-01 7.00984776e-01 3.57858241e-01
-1.12794608e-01 -8.78542364e-01 -1.73476696e-01 6.90316260e-01
6.30060315e-01 7.27397323e-01 8.69598389e-01 -6.56819522e-01
-1.05475700e+00 -6.10699236e-01 -6.25552312e-02 8.86110067e-01
1.10653847e-01 1.09970547e-01 -1.24841118e+00 3.41863632e-02
4.50877585e-02 -1.97242647e-01 5.89721918e-01 3.15123238e-02
1.25118542e+00 -7.04374552e-01 -2.20329061e-01 5.83588779e-01
1.18775082e+00 2.51564860e-01 1.23005712e+00 5.10588586e-01
6.32022262e-01 6.29858255e-01 4.43982780e-01 2.55992681e-01
-6.48053348e-01 1.58659726e-01 5.10063589e-01 -2.27319330e-01
-1.78790331e-01 -3.17212462e-01 2.37160057e-01 3.46040614e-02
-3.66668791e-01 -6.10825002e-01 -4.16570604e-01 1.90174848e-01
-1.42272544e+00 -1.24985540e+00 5.04281186e-03 1.87638295e+00
8.51884246e-01 3.26730579e-01 -1.92496888e-02 -3.74053009e-02
8.88477802e-01 1.87306687e-01 -8.90700638e-01 -3.82894695e-01
-6.73224390e-01 2.72024393e-01 7.58943617e-01 1.35546044e-01
-9.25926924e-01 9.52203631e-01 7.18420267e+00 1.41616130e+00
-3.85886699e-01 1.71329871e-01 8.16777945e-01 -3.40645909e-01
-8.37433338e-01 -1.00434132e-01 -2.25835413e-01 4.82287079e-01
3.86158437e-01 -8.05295944e-01 5.91898382e-01 1.04651916e+00
-3.30883920e-01 5.70531070e-01 -8.76753092e-01 5.30511081e-01
-9.21460092e-02 -1.68719184e+00 5.21945238e-01 4.39406842e-01
1.25333059e+00 -6.69359922e-01 8.18511248e-01 9.39583704e-02
7.24016607e-01 -1.22796166e+00 5.50968885e-01 6.98983610e-01
7.16028988e-01 -1.03857803e+00 5.84018230e-01 3.28309834e-02
-4.17369395e-01 -1.44885376e-01 -3.84970337e-01 4.59810998e-03
1.08330563e-01 4.28954333e-01 -2.56407738e-01 6.29801214e-01
3.70657712e-01 2.93869376e-01 -4.33100879e-01 6.89844131e-01
-5.97272754e-01 4.49859411e-01 4.70060632e-02 -1.39191762e-01
1.91736311e-01 -3.17018479e-01 9.36506212e-01 7.88941264e-01
1.45269617e-01 -1.86812356e-02 -5.26883006e-01 1.35006118e+00
-6.53407037e-01 -2.06446990e-01 -1.10184729e+00 -6.15921021e-01
3.58691812e-01 1.14805937e+00 -3.07610363e-01 -2.79126436e-01
1.67965367e-02 1.39462662e+00 -1.42118454e-01 6.52792692e-01
-1.12342751e+00 -7.41397083e-01 9.54087019e-01 -4.96178307e-02
1.04701705e-01 6.89538494e-02 -5.04615843e-01 -8.02569807e-01
3.65579836e-02 -1.16748095e+00 -3.78851183e-02 -9.62725759e-01
-2.03669715e+00 3.99109453e-01 -3.36593062e-01 -1.29693890e+00
2.62910694e-01 -6.46593392e-01 -7.94627786e-01 9.39708173e-01
-1.05531383e+00 -1.33024454e+00 6.25256598e-02 6.10558569e-01
1.53722003e-01 -7.53636003e-01 6.50158763e-01 9.13451836e-02
-3.88910890e-01 1.11188841e+00 2.76215822e-01 3.42330188e-01
8.15674543e-01 -1.13738370e+00 4.84093130e-01 6.94084525e-01
2.56851390e-02 8.12016010e-01 6.66179717e-01 -9.68716383e-01
-1.38679588e+00 -1.13091958e+00 9.53070149e-02 -7.92930365e-01
1.13676584e+00 -1.97658285e-01 -6.49015427e-01 6.13502622e-01
5.51374555e-01 -3.88396144e-01 6.46406829e-01 -3.72914582e-01
-5.99879622e-01 2.23556496e-02 -1.77541971e+00 1.05576479e+00
1.28535795e+00 -8.08719933e-01 -3.70048970e-01 6.53779864e-01
1.17025900e+00 -6.78280964e-02 -1.17479539e+00 1.42268986e-01
4.44891721e-01 -5.79856157e-01 1.30623138e+00 -9.12812293e-01
9.74356055e-01 -1.88125357e-01 1.45569537e-02 -1.01993752e+00
-3.23959887e-01 -1.44110894e+00 7.79726282e-02 1.79102337e+00
4.04174715e-01 -7.71121800e-01 7.94435263e-01 1.10730577e+00
4.07733619e-01 -3.10745001e-01 -8.21083307e-01 -1.45527554e+00
5.69133818e-01 -3.81640613e-01 9.62065816e-01 1.17027855e+00
-4.19254065e-01 9.31929518e-03 -7.00728238e-01 1.40014932e-01
9.20320749e-01 -4.44297791e-01 1.01354980e+00 -9.58198786e-01
-3.93597603e-01 -5.79523623e-01 -5.76403439e-01 -6.20748281e-01
3.38638216e-01 -7.92684495e-01 -5.08070529e-01 -1.08617580e+00
-2.63510849e-02 -4.87655729e-01 -2.18346775e-01 2.02657640e-01
-2.83304155e-01 4.01676506e-01 4.91992086e-01 2.63168514e-01
-3.83143313e-02 4.44572359e-01 1.96813571e+00 -7.75266588e-01
2.94835299e-01 5.37222587e-02 -1.05416250e+00 7.12389946e-01
8.21598351e-01 -6.80432856e-01 -4.40755129e-01 -3.55268389e-01
3.06652337e-01 -3.26717556e-01 6.18192911e-01 -7.60212421e-01
1.18739776e-01 -3.79067570e-01 3.72715116e-01 1.06467739e-01
1.93586022e-01 -9.10394132e-01 3.98221463e-01 3.91854227e-01
-5.14564097e-01 -3.52218062e-01 3.22044551e-01 1.05809414e+00
8.21972415e-02 -4.17252421e-01 4.99871522e-01 -1.71470627e-01
-2.62732565e-01 4.07130510e-01 -2.22536147e-01 1.19689509e-01
1.33780909e+00 -1.15352757e-01 -8.82072866e-01 -5.78066528e-01
-6.02089643e-01 -1.23019770e-01 6.51730239e-01 5.33624768e-01
4.55924898e-01 -1.61879528e+00 -7.65877783e-01 -7.64142275e-02
-2.30574802e-01 -4.53294575e-01 2.26952791e-01 1.04676843e-01
-3.27256382e-01 -2.21588388e-01 -4.59486395e-01 2.01760866e-02
-1.10095096e+00 9.95105863e-01 1.10067226e-01 -1.92429990e-01
-1.46784037e-01 1.13475263e+00 3.44310731e-01 -2.03270122e-01
2.34233961e-01 3.22028220e-01 2.06590056e-01 -3.30843329e-01
4.22669947e-01 7.45134413e-01 -7.57431209e-01 -1.83048040e-01
-1.38394922e-01 6.11664414e-01 -1.98916033e-01 -1.90654337e-01
1.10125530e+00 4.24594909e-01 -1.11817829e-02 -1.84831113e-01
1.00582993e+00 4.70110655e-01 -1.36206722e+00 1.57839701e-01
-4.95967537e-01 -9.52392638e-01 -2.87156433e-01 -1.13049150e+00
-1.33822119e+00 7.35389292e-01 4.66465175e-01 5.69994748e-01
1.04304171e+00 -2.12685838e-01 8.01982582e-01 4.75502849e-01
4.14303482e-01 -1.11678267e+00 4.80995804e-01 -1.97023302e-01
1.31376028e+00 -9.29906368e-01 1.55083761e-01 -5.85142076e-01
-1.03962672e+00 6.42996430e-01 4.47811127e-01 -3.53996903e-01
3.47040445e-01 4.42514330e-01 1.07535571e-01 -6.76406398e-02
-1.38698310e-01 2.70538956e-01 2.36577004e-01 1.16702068e+00
-2.11401537e-01 8.54074955e-02 -1.31793812e-01 6.09316111e-01
-1.09901488e-01 -5.40238246e-02 4.77582693e-01 8.44274700e-01
8.43304545e-02 -1.38910174e+00 -6.12969756e-01 2.48599023e-01
-7.71087408e-01 -1.41230524e-01 -1.17449045e+00 8.50191653e-01
1.03285715e-01 9.92167652e-01 -3.70830417e-01 -4.19748366e-01
4.07019407e-01 -3.67449760e-01 4.90224153e-01 5.60076013e-02
-9.90120828e-01 -1.89840421e-01 1.04354747e-01 -3.30939054e-01
-3.51460040e-01 -2.47874081e-01 -6.94290757e-01 -7.00629115e-01
-4.26623404e-01 -8.55167359e-02 4.99901652e-01 5.13045847e-01
5.01807988e-01 7.14835405e-01 7.70170450e-01 -5.10904789e-01
-7.30028808e-01 -6.94492221e-01 -7.59291947e-01 8.04510057e-01
-8.81367177e-02 -3.01124990e-01 -7.40216315e-01 7.96259120e-02]
|
[5.63887882232666, 7.9793195724487305]
|
7d11843f-6b13-4eaa-b7a0-dada8eedec97
|
practical-and-configurable-network-traffic
|
2107.06080
| null |
https://arxiv.org/abs/2107.06080v1
|
https://arxiv.org/pdf/2107.06080v1.pdf
|
Practical and Configurable Network Traffic Classification Using Probabilistic Machine Learning
|
Network traffic classification that is widely applicable and highly accurate is valuable for many network security and management tasks. A flexible and easily configurable classification framework is ideal, as it can be customized for use in a wide variety of networks. In this paper, we propose a highly configurable and flexible machine learning traffic classification method that relies only on statistics of sequences of packets to distinguish known, or approved, traffic from unknown traffic. Our method is based on likelihood estimation, provides a measure of certainty for classification decisions, and can classify traffic at adjustable certainty levels. Our classification method can also be applied in different classification scenarios, each prioritizing a different classification goal. We demonstrate how our classification scheme and all its configurations perform well on real-world traffic from a high performance computing network environment.
|
['Jacobus Van der Merwe', 'Jeff M. Phillips', 'Joe Breen', 'Jiahui Chen']
|
2021-07-10
| null | null | null | null |
['traffic-classification']
|
['miscellaneous']
|
[ 1.43189520e-01 -7.85798371e-01 -6.49851382e-01 -6.13977313e-01
-2.55353391e-01 -6.89144671e-01 2.33960465e-01 8.74197334e-02
-1.79106817e-01 1.02696860e+00 -9.08828259e-01 -1.09298730e+00
-4.38638628e-01 -1.06424320e+00 1.44871876e-01 -5.38080752e-01
-3.01220804e-01 8.74903381e-01 8.03699911e-01 -4.77672741e-02
2.83098072e-01 1.05048645e+00 -1.55044270e+00 3.14440250e-01
4.11231786e-01 1.45531714e+00 -1.36264369e-01 6.36144400e-01
-2.20470041e-01 4.61821049e-01 -9.09096241e-01 -4.76164013e-01
3.43216926e-01 1.31420523e-01 -6.84201360e-01 -7.14927912e-02
1.41602427e-01 -1.52201682e-01 -1.19033828e-01 6.94250584e-01
1.79870233e-01 -2.10298657e-01 5.66791236e-01 -1.88898492e+00
1.95805117e-01 5.85542023e-01 -2.75599509e-01 8.13695073e-01
-1.13327019e-02 1.83817357e-01 1.01263535e+00 -1.47730723e-01
-1.26867658e-02 1.05699253e+00 3.82326871e-01 4.40777659e-01
-1.42508543e+00 -1.13930595e+00 1.42504379e-01 3.84344041e-01
-1.26760924e+00 -4.71297055e-01 6.23508632e-01 -4.61787879e-01
5.30095637e-01 7.02411234e-01 1.61304310e-01 8.94089699e-01
3.80167991e-01 4.31921750e-01 9.65058923e-01 -2.37079278e-01
4.06765193e-01 4.69614953e-01 4.03293699e-01 4.33778435e-01
4.37803507e-01 1.58872634e-01 9.78871584e-02 -7.23002791e-01
4.12580252e-01 2.25342721e-01 -1.35738954e-01 -2.13191941e-01
-9.26145911e-01 8.02262187e-01 -2.60702044e-01 -3.72650810e-02
-1.81837633e-01 2.42684528e-01 5.65621495e-01 5.66312075e-01
2.62069553e-01 1.57839239e-01 -8.18260968e-01 -3.01489800e-01
-8.97687495e-01 -1.49118572e-01 1.16539061e+00 9.32997167e-01
8.23679626e-01 2.56242365e-01 -2.11561203e-01 5.22830546e-01
1.72018021e-01 5.62647223e-01 7.36372173e-02 -9.70150113e-01
2.82695442e-01 4.74491417e-02 -1.43593177e-01 -8.96269202e-01
-3.11290383e-01 -4.64594632e-01 -7.47614324e-01 5.72356701e-01
3.77667814e-01 -1.96778208e-01 -6.42000914e-01 1.65076256e+00
1.91177446e-02 5.34973323e-01 -2.19731897e-01 3.59509945e-01
7.27078840e-02 5.92477679e-01 2.50729769e-01 -2.53306478e-01
1.16601789e+00 -3.25854778e-01 -1.81438714e-01 2.12444276e-01
2.44105712e-01 -7.83310831e-01 6.26333952e-01 5.24141669e-01
-5.91601074e-01 -3.71603251e-01 -8.74909937e-01 7.09699988e-01
-3.80358189e-01 -2.07697630e-01 9.80576456e-01 1.48567390e+00
-8.53506505e-01 4.07829434e-01 -6.76039279e-01 -3.75216603e-01
4.36917245e-01 6.35841846e-01 7.64639452e-02 1.39761120e-01
-1.05666757e+00 5.20796061e-01 5.31196237e-01 -4.08186853e-01
-7.98628688e-01 -3.68417054e-01 -4.44146723e-01 3.35220128e-01
5.72180212e-01 -3.57824206e-01 9.83114481e-01 -5.56782305e-01
-1.37614322e+00 3.68107766e-01 6.76150946e-03 -3.83841723e-01
2.42822409e-01 5.10152161e-01 -9.78478551e-01 -7.09952489e-02
-1.94961540e-02 4.43715677e-02 1.18980718e+00 -9.43260491e-01
-1.02084875e+00 3.99105847e-02 1.49098709e-01 -6.81009531e-01
-3.09042335e-01 3.96845281e-01 -1.09664209e-01 -6.03280485e-01
-2.17557862e-01 -8.41989398e-01 -3.18086952e-01 7.23269880e-02
-1.29617304e-01 -1.17778078e-01 1.31977761e+00 1.81107387e-01
1.29028296e+00 -1.77101219e+00 -6.76224709e-01 9.31163907e-01
1.85402408e-01 5.16996741e-01 2.69659888e-03 1.22157656e-01
9.62614343e-02 5.22178352e-01 4.92590219e-02 1.59152955e-01
-5.80520891e-02 3.26108634e-01 -5.06128073e-01 1.54355481e-01
3.72918785e-01 2.24800646e-01 -6.89981818e-01 -4.45777565e-01
3.57802749e-01 1.24659978e-01 -5.10981262e-01 1.71905339e-01
-1.25236064e-01 2.02431649e-01 -5.92798889e-01 7.71332562e-01
7.73667872e-01 -3.93471301e-01 4.01429206e-01 -1.16056256e-01
3.65605205e-02 1.87161028e-01 -1.27878582e+00 3.09350699e-01
-7.87187397e-01 7.25181341e-01 -5.49356565e-02 -1.13413644e+00
1.09560180e+00 2.58687407e-01 4.74679708e-01 -1.70880958e-01
2.32079089e-01 1.78281561e-01 7.27498382e-02 -7.55969211e-02
1.14269882e-01 1.20460875e-01 -6.07765764e-02 9.24187362e-01
-4.58234586e-02 2.66371667e-01 4.38420504e-01 1.69179626e-02
1.39378273e+00 -8.59387994e-01 4.31392372e-01 -3.02099556e-01
9.16422546e-01 -4.54547435e-01 8.05574954e-01 1.02909613e+00
-5.27961195e-01 -1.36918992e-01 7.22272754e-01 -5.07726669e-01
-7.78800309e-01 -1.36929834e+00 -4.28868264e-01 1.16467035e+00
1.06625460e-01 -2.05491513e-01 -1.57061160e-01 -1.06685507e+00
1.86415717e-01 5.70566058e-01 -9.17623565e-03 1.30930826e-01
-6.77856565e-01 -8.44441831e-01 5.58922112e-01 2.04163641e-01
2.27390826e-01 -7.06263244e-01 -2.44435355e-01 3.97470862e-01
8.81062448e-02 -1.51729155e+00 -1.64831623e-01 2.20815554e-01
-6.35611713e-01 -1.28321266e+00 1.67669803e-01 -3.66745651e-01
3.53947312e-01 5.05961478e-01 1.23339486e+00 4.46003497e-01
-3.02238792e-01 3.27144951e-01 -1.91014379e-01 -3.01577210e-01
-7.88853347e-01 2.41462022e-01 3.55813354e-01 3.52806985e-01
5.20904064e-01 -7.92279780e-01 -1.24974176e-01 9.64979231e-01
-7.31951714e-01 -6.17329895e-01 3.20423752e-01 6.80511594e-01
1.49915338e-01 8.32948327e-01 8.69938016e-01 -1.05263340e+00
7.48919964e-01 -9.62938249e-01 -9.41110671e-01 3.33385438e-01
-9.22732234e-01 4.61256243e-02 8.67444932e-01 -5.00600398e-01
-7.05177665e-01 -3.56324613e-01 -1.60280868e-01 -1.47062987e-01
-3.01555008e-01 -1.88247487e-02 -2.42086723e-01 -4.16395515e-01
5.54370105e-01 -2.23436356e-01 -5.47609217e-02 -2.78126925e-01
-1.11988738e-01 9.60881650e-01 1.30325377e-01 -9.95086968e-01
1.01105559e+00 4.68013763e-01 4.42599267e-01 -7.91406035e-01
-3.03509831e-01 -3.38895202e-01 -4.68946248e-01 -3.49150926e-01
1.25663681e-02 -3.54223013e-01 -1.17938852e+00 2.91791022e-01
-8.33271146e-01 4.67611738e-02 1.63058162e-01 5.28711200e-01
-2.57361263e-01 5.62395751e-01 -6.33651137e-01 -8.57130885e-01
-1.73394933e-01 -1.27714562e+00 4.45965797e-01 1.18035048e-01
-6.21848851e-02 -1.25241828e+00 -4.17998493e-01 4.28003669e-02
8.82955194e-01 -7.24181155e-05 1.14324641e+00 -1.08832812e+00
-9.16696012e-01 -3.58431011e-01 -5.21665692e-01 4.93592978e-01
4.53243315e-01 6.65866017e-01 -6.41481459e-01 -4.86921251e-01
-2.57168710e-01 -1.59008518e-01 7.75704801e-01 -9.12207365e-02
1.73153043e+00 -4.46749270e-01 -4.53791648e-01 3.87636930e-01
1.28111482e+00 2.92100847e-01 5.55456877e-01 1.74222454e-01
2.05671459e-01 3.69651347e-01 5.68998992e-01 5.94693184e-01
-3.29679511e-02 7.68720567e-01 5.01009107e-01 3.62183601e-01
3.85077178e-01 3.96354198e-01 2.55430639e-01 3.33222419e-01
1.96743652e-01 -7.56493270e-01 -9.54284489e-01 3.24053839e-02
-1.62606943e+00 -1.35909116e+00 1.09524332e-01 2.23809075e+00
5.38478553e-01 9.06786621e-01 3.06531429e-01 6.51543617e-01
1.04421008e+00 3.75806801e-02 -2.84743577e-01 -6.07652247e-01
2.69969583e-01 6.62806273e-01 6.94726825e-01 2.61951298e-01
-1.03001380e+00 6.63889468e-01 7.53640985e+00 1.31336164e+00
-1.47027862e+00 -1.75911561e-01 6.66543603e-01 3.40945572e-01
1.71494950e-02 1.36261448e-01 -9.25966382e-01 7.99948454e-01
1.31561041e+00 -3.72643411e-01 3.59620154e-01 1.04585719e+00
9.29333568e-02 1.67380020e-01 -9.81627464e-01 8.95653725e-01
-4.51536506e-01 -1.60266411e+00 3.33466202e-01 -2.35832129e-02
8.07561055e-02 -7.47080073e-02 1.89557120e-01 3.55472952e-01
4.47666168e-01 -8.14758003e-01 2.40068778e-01 2.89482594e-01
7.45732665e-01 -8.57270479e-01 8.47455144e-01 1.51825592e-01
-1.20470655e+00 -3.37767631e-01 3.16098109e-02 4.80069928e-02
1.54744044e-01 8.98264289e-01 -1.00650322e+00 3.73112619e-01
4.08018470e-01 3.04531723e-01 -5.20292521e-01 1.14124668e+00
1.97233006e-01 9.72686768e-01 -5.26171982e-01 -1.33536011e-01
-1.06241688e-01 2.16290429e-01 5.38660705e-01 1.34825456e+00
2.27905765e-01 -1.42883912e-01 5.71146309e-01 3.99330467e-01
7.44738104e-03 -1.26024410e-01 -1.81763947e-01 3.43664259e-01
1.02918005e+00 1.42216218e+00 -9.65102494e-01 -2.57049471e-01
-2.30214089e-01 2.57797480e-01 5.94805218e-02 2.17725366e-01
-9.15256977e-01 -7.38539815e-01 1.17823851e+00 3.30069065e-01
2.25425497e-01 -4.99500483e-01 -1.08310260e-01 -1.11362338e+00
-1.29858509e-01 -6.58049464e-01 6.70489073e-01 -4.73690853e-02
-1.56910467e+00 8.98253739e-01 1.33827314e-01 -1.40912199e+00
-2.16133803e-01 -9.87309158e-01 -8.26275468e-01 5.57477832e-01
-1.59456682e+00 -6.00145698e-01 -2.81692684e-01 7.19671905e-01
2.21065223e-01 -7.12916374e-01 7.67504871e-01 6.93955719e-01
-9.37112451e-01 8.81800354e-01 -2.25564232e-03 1.79690138e-01
6.40039861e-01 -8.24952841e-01 2.35723838e-01 6.28754377e-01
-8.04865826e-03 4.43200052e-01 6.22726023e-01 -2.39680603e-01
-8.38614523e-01 -1.34455240e+00 3.79729539e-01 -3.74976903e-01
8.66227984e-01 -9.89469290e-02 -7.81442702e-01 5.55547655e-01
-5.26917160e-01 2.64788598e-01 1.05375123e+00 1.35017842e-01
-6.37065589e-01 -7.24036396e-01 -1.56729221e+00 3.65991384e-01
6.38506293e-01 -3.69728446e-01 1.43717099e-02 1.50934383e-01
3.67542893e-01 9.74220484e-02 -9.03498530e-01 5.32098830e-01
4.70284671e-01 -8.84527922e-01 7.52288699e-01 -8.77852440e-01
-7.27399945e-01 -4.01945651e-01 -3.54742825e-01 -6.93549454e-01
-4.00862992e-01 -8.89001727e-01 -2.86767185e-01 1.36344755e+00
5.26696384e-01 -1.12377167e+00 8.27506244e-01 5.41441977e-01
4.95028913e-01 -5.37613451e-01 -1.11634040e+00 -1.20963454e+00
-4.61957693e-01 -8.07071090e-01 1.08758736e+00 8.61765146e-01
-2.03342125e-01 1.05429821e-01 -2.63062149e-01 2.69970238e-01
9.67173040e-01 1.05512261e-01 8.74755859e-01 -1.67205727e+00
-1.32606402e-01 -6.09913111e-01 -1.06785619e+00 -8.56031001e-01
4.38601166e-01 -8.00977767e-01 -5.93968630e-01 -4.09341365e-01
-1.52672842e-01 -1.36909187e+00 -7.65328825e-01 5.28992355e-01
2.50336975e-01 2.46242285e-01 4.05009575e-02 3.01111281e-01
-5.88581383e-01 5.60008697e-02 2.27820009e-01 -1.97969656e-02
1.79132484e-02 9.52028334e-01 -5.88463426e-01 5.09866118e-01
1.37646341e+00 -5.78240097e-01 -3.49918902e-01 2.05443501e-01
-3.28345031e-01 -3.04268654e-02 3.68131936e-01 -1.28419197e+00
1.69123501e-01 -6.67025685e-01 2.17417791e-01 -5.13257623e-01
2.13229075e-01 -1.14716566e+00 1.88009977e-01 4.36320633e-01
-1.19863927e-01 4.28893976e-02 3.87870222e-02 7.37456322e-01
1.15892254e-01 -5.59401885e-02 1.08789122e+00 2.04740644e-01
-7.99208462e-01 8.09001148e-01 -7.25676596e-01 1.13096140e-01
1.26968372e+00 -2.05022573e-01 -4.41132873e-01 -4.17832553e-01
-6.44349933e-01 2.89248824e-01 3.40852737e-01 4.38872844e-01
4.93684560e-01 -1.33606386e+00 -6.35345280e-01 4.03737813e-01
2.11482152e-01 -1.09012330e+00 -3.96508783e-01 3.22979301e-01
-4.77301091e-01 5.96743941e-01 -4.59622562e-01 -8.25212896e-01
-1.35013068e+00 6.42655134e-01 2.08739042e-01 -3.69246542e-01
-2.52348166e-02 5.31057902e-02 -5.34012675e-01 -2.49148905e-01
2.46103674e-01 -6.48646876e-02 -1.16530061e-02 -3.39455128e-01
6.94733083e-01 7.22950637e-01 1.58254460e-01 -4.90817696e-01
-5.99330425e-01 3.12813878e-01 -8.26027244e-02 9.99620482e-02
8.07780504e-01 -2.38610774e-01 -1.73566163e-01 1.37425587e-01
1.10577500e+00 -2.04954091e-02 -7.00141370e-01 -4.23434317e-01
1.71531871e-01 -9.60567355e-01 -4.54679467e-02 -4.05620635e-01
-1.37676823e+00 7.41962790e-01 4.67250228e-01 8.38368535e-01
1.22384465e+00 -4.01390553e-01 7.34138250e-01 6.11322701e-01
9.86616194e-01 -6.25917673e-01 -1.01720877e-01 5.61873734e-01
-6.47782758e-02 -1.27401817e+00 3.26591656e-02 -6.75654769e-01
-3.94193292e-01 1.57578874e+00 6.72253788e-01 2.66811579e-01
1.25764143e+00 3.87068868e-01 -4.29775976e-02 2.17286214e-01
-1.21999609e+00 -2.79135168e-01 -1.58030272e-01 8.87371182e-01
1.22925036e-01 1.73840210e-01 -1.78860962e-01 1.62508324e-01
1.07614614e-01 -1.61466435e-01 7.52248526e-01 8.70039940e-01
-7.67382145e-01 -1.60555291e+00 -2.35080570e-01 8.14906657e-01
-6.79761708e-01 1.24104157e-01 2.01143920e-01 5.18023610e-01
-6.09388091e-02 1.31022727e+00 2.36828223e-01 -6.25499964e-01
-4.54938784e-02 -1.88454151e-01 -8.64697695e-02 -5.66536546e-01
-2.27674782e-01 -6.75089002e-01 3.48100454e-01 -5.15771806e-01
-2.04571903e-01 -3.98656607e-01 -9.60853279e-01 -1.14456582e+00
-4.73822236e-01 5.54006875e-01 6.75328970e-01 7.75859833e-01
4.10582513e-01 2.32107997e-01 1.43829465e+00 -4.41930324e-01
-4.47031707e-01 -1.86602667e-01 -6.55822158e-01 9.07174349e-02
3.04159313e-01 -1.01323688e+00 -6.05423629e-01 -4.69360828e-01]
|
[5.070288181304932, 7.217759609222412]
|
6b6c6fe6-4ddb-40d5-891b-3e803e59d439
|
ice-inter-instance-contrastive-encoding-for
|
2103.16364
| null |
https://arxiv.org/abs/2103.16364v2
|
https://arxiv.org/pdf/2103.16364v2.pdf
|
ICE: Inter-instance Contrastive Encoding for Unsupervised Person Re-identification
|
Unsupervised person re-identification (ReID) aims at learning discriminative identity features without annotations. Recently, self-supervised contrastive learning has gained increasing attention for its effectiveness in unsupervised representation learning. The main idea of instance contrastive learning is to match a same instance in different augmented views. However, the relationship between different instances has not been fully explored in previous contrastive methods, especially for instance-level contrastive loss. To address this issue, we propose Inter-instance Contrastive Encoding (ICE) that leverages inter-instance pairwise similarity scores to boost previous class-level contrastive ReID methods. We first use pairwise similarity ranking as one-hot hard pseudo labels for hard instance contrast, which aims at reducing intra-class variance. Then, we use similarity scores as soft pseudo labels to enhance the consistency between augmented and original views, which makes our model more robust to augmentation perturbations. Experiments on several large-scale person ReID datasets validate the effectiveness of our proposed unsupervised method ICE, which is competitive with even supervised methods. Code is made available at https://github.com/chenhao2345/ICE.
|
['Francois Bremond', 'Benoit Lagadec', 'Hao Chen']
|
2021-03-30
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Chen_ICE_Inter-Instance_Contrastive_Encoding_for_Unsupervised_Person_Re-Identification_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Chen_ICE_Inter-Instance_Contrastive_Encoding_for_Unsupervised_Person_Re-Identification_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['unsupervised-person-re-identification']
|
['computer-vision']
|
[ 2.33228743e-01 -1.80646572e-02 -2.40643710e-01 -6.92646444e-01
-6.34076297e-01 -5.05982459e-01 7.50341833e-01 2.21318692e-01
-4.15859818e-01 6.18155837e-01 4.84276623e-01 6.06123090e-01
-4.81994636e-02 -4.58397090e-01 -6.45712495e-01 -6.19534373e-01
1.42181292e-01 2.82010496e-01 -2.01974466e-01 4.12533395e-02
2.04420716e-01 2.52853688e-02 -1.59385324e+00 1.13756791e-01
1.03458500e+00 5.87508738e-01 -5.63785285e-02 3.32727693e-02
6.12137429e-02 6.60593510e-01 -4.11463469e-01 -8.16805661e-01
4.68780667e-01 -5.54379523e-01 -7.48255670e-01 4.96637784e-02
9.94903684e-01 -1.57366157e-01 -4.88374352e-01 1.26475227e+00
5.78579605e-01 3.17204684e-01 5.89381933e-01 -1.43059599e+00
-8.21994722e-01 5.45214593e-01 -8.76892924e-01 2.26518005e-01
4.83313620e-01 7.95542449e-02 1.05852914e+00 -8.94716799e-01
5.02466500e-01 1.23314917e+00 6.70851111e-01 6.56631529e-01
-1.34660602e+00 -9.79430616e-01 2.97940463e-01 5.37598610e-01
-1.58530295e+00 -5.28545976e-01 9.82794702e-01 -2.81342775e-01
4.29479569e-01 2.44146809e-01 4.36939925e-01 1.11839056e+00
-4.22407299e-01 1.03761971e+00 1.30505610e+00 -3.00038606e-01
-1.22354418e-01 2.11019441e-01 2.78990090e-01 5.67998290e-01
3.14935386e-01 2.05996111e-01 -6.56436741e-01 -2.60068998e-02
4.96423811e-01 2.38453269e-01 -3.27388734e-01 -4.00292933e-01
-1.20111442e+00 4.22133714e-01 7.92611837e-01 2.37142146e-02
-8.76073763e-02 -1.18637748e-01 4.66977984e-01 1.65591523e-01
4.18877095e-01 3.83329540e-01 -8.41768831e-02 -2.11900193e-02
-6.82240903e-01 3.57235610e-01 2.17689782e-01 9.33982015e-01
9.09898400e-01 -3.36395204e-01 -2.66208261e-01 1.17461383e+00
1.29358515e-01 3.37955028e-01 5.96823692e-01 -7.49841154e-01
5.52049458e-01 7.81894505e-01 -1.92883819e-01 -1.02661264e+00
-1.51631400e-01 -6.14786565e-01 -1.10817659e+00 -1.21192582e-01
2.94052482e-01 2.32290626e-01 -5.62369108e-01 2.03358197e+00
3.45290005e-01 6.66655898e-01 -1.14905149e-01 1.00050426e+00
9.07628775e-01 3.49242687e-01 1.23205103e-01 -2.42109671e-02
1.27622128e+00 -1.03601062e+00 -4.88951743e-01 -2.42396221e-01
4.69562411e-01 -4.72067505e-01 8.92490208e-01 -6.82038739e-02
-8.56500566e-01 -8.35494518e-01 -1.02259409e+00 -7.27441236e-02
-2.27250755e-01 1.73654526e-01 3.70629221e-01 5.28711736e-01
-5.99847853e-01 3.49711806e-01 -4.73250777e-01 -1.95377752e-01
6.13049686e-01 2.77814895e-01 -8.09079230e-01 -2.31102109e-01
-1.06986916e+00 5.24544299e-01 3.45976830e-01 4.42841053e-02
-3.37479115e-01 -6.59543872e-01 -1.11174166e+00 -7.99948424e-02
3.81330192e-01 -4.90133375e-01 7.70055354e-01 -9.99171317e-01
-1.09757793e+00 1.15538752e+00 -4.31317568e-01 -2.41660431e-01
5.49765110e-01 -3.64328057e-01 -3.75785887e-01 4.01146850e-03
4.16100502e-01 8.18038762e-01 7.91865528e-01 -1.56423986e+00
-4.21378911e-01 -6.21352851e-01 -3.78118008e-02 4.00797248e-01
-4.71710742e-01 -9.77503434e-02 -5.86836874e-01 -9.78799820e-01
1.43022001e-01 -1.07887125e+00 6.48898855e-02 -1.43253475e-01
-4.65095013e-01 -3.43621075e-01 3.51373762e-01 -6.71547949e-01
9.19380009e-01 -2.19521928e+00 1.21619403e-01 3.08624208e-01
2.26628527e-01 1.32435888e-01 -2.52998471e-01 1.62509501e-01
-3.29203963e-01 -5.98272681e-02 -3.63459617e-01 -7.50611484e-01
-6.92949742e-02 -1.20502844e-01 -8.21787566e-02 5.46362877e-01
3.17691386e-01 9.00712192e-01 -9.52629507e-01 -4.99877512e-01
2.46634949e-02 3.52195263e-01 -4.85619158e-01 3.01327080e-01
4.23162639e-01 8.15063834e-01 -1.41450033e-01 6.19340539e-01
9.55420017e-01 -1.12465568e-01 1.52656555e-01 -4.39860374e-01
1.09943457e-01 1.13362938e-01 -1.24109638e+00 1.69594789e+00
-1.01865791e-01 4.02065516e-01 -4.57046449e-01 -1.10951555e+00
1.00228918e+00 -1.04241267e-01 3.00996214e-01 -9.65679705e-01
-1.80769578e-01 3.44663933e-02 -3.61124724e-01 -1.99041545e-01
5.59392929e-01 1.10491090e-01 -8.71429965e-02 3.05413812e-01
-1.06690548e-01 4.97998685e-01 1.64169312e-01 4.24485207e-01
6.56856298e-01 3.73445958e-01 2.54668325e-01 -9.06249359e-02
7.29729116e-01 -4.64652777e-01 1.02362835e+00 7.05235898e-01
-3.64332736e-01 9.55843270e-01 1.61084756e-01 -2.92132944e-01
-8.70856225e-01 -1.08623970e+00 -1.44896686e-01 8.93821239e-01
7.54746616e-01 -5.82293272e-01 -6.75360739e-01 -8.89474154e-01
1.30003452e-01 2.20775723e-01 -7.15605319e-01 -2.47726321e-01
-6.57122970e-01 -6.31792665e-01 4.65117544e-01 7.01411963e-01
8.30796421e-01 -7.34296620e-01 1.19167648e-01 -6.12465106e-02
-5.44796765e-01 -9.73071933e-01 -9.02194858e-01 -2.96297342e-01
-5.74423730e-01 -1.12986851e+00 -9.86819088e-01 -9.63193119e-01
1.09703875e+00 5.75947881e-01 8.38205218e-01 1.35325462e-01
-3.74302506e-01 4.90287483e-01 -4.30149943e-01 1.14124455e-02
3.38326916e-02 -1.74915511e-02 4.67992127e-01 3.00752223e-01
5.20209968e-01 -5.69957495e-01 -7.34389544e-01 5.23256540e-01
-6.65523589e-01 3.17544267e-02 4.74607259e-01 1.07935083e+00
7.39634693e-01 -1.17775992e-01 7.05196440e-01 -9.42921400e-01
3.92803162e-01 -3.55290204e-01 -2.15132445e-01 3.10683370e-01
-6.16590977e-01 5.51698320e-02 5.34291923e-01 -4.33297098e-01
-1.13953042e+00 -5.67719452e-02 1.91749737e-01 -4.17647183e-01
-1.45406231e-01 2.81569034e-01 -4.64722633e-01 1.01947762e-01
4.65636194e-01 3.91027421e-01 9.99928545e-03 -3.73687804e-01
2.16439858e-01 6.34565234e-01 8.74119401e-01 -7.28962898e-01
1.01551569e+00 3.86162430e-01 -3.14848840e-01 -4.55960810e-01
-8.07873249e-01 -6.82658315e-01 -7.40293264e-01 -1.54100671e-01
5.17111063e-01 -1.28137326e+00 -6.38208807e-01 7.21772373e-01
-7.08681703e-01 -1.19284928e-01 -2.46852115e-01 3.33067894e-01
-2.62171537e-01 7.68034697e-01 -4.39807415e-01 -6.52371466e-01
-2.90453613e-01 -9.25306022e-01 9.43783820e-01 5.74321568e-01
-1.40997276e-01 -7.33413219e-01 1.48889869e-01 7.61889637e-01
2.59098168e-02 3.96556295e-02 3.95839393e-01 -6.16246700e-01
-3.79396677e-01 -3.38531554e-01 -3.95350724e-01 3.75797480e-01
3.48142534e-01 -4.55888212e-01 -1.04683244e+00 -6.09176576e-01
-4.67948347e-01 -4.21694160e-01 1.10022664e+00 -3.88149060e-02
1.27531672e+00 -3.86826068e-01 -4.03606981e-01 6.94876730e-01
1.08911932e+00 -3.24505448e-01 7.17803419e-01 5.01559913e-01
1.07593775e+00 6.68606639e-01 7.17921436e-01 3.27611119e-01
7.14563847e-01 1.01349235e+00 5.58822043e-02 8.71146247e-02
-1.67838112e-01 -4.68930095e-01 2.79229760e-01 5.83095729e-01
-2.84497112e-01 7.58992359e-02 -6.65283859e-01 4.69521850e-01
-1.93055701e+00 -1.27438855e+00 1.72525957e-01 2.49236655e+00
7.94482350e-01 -1.71076488e-02 4.34816271e-01 3.71488594e-02
1.21048987e+00 1.13197260e-01 -6.23636186e-01 2.96364188e-01
-2.52234101e-01 -1.61577642e-01 2.04016611e-01 2.20421717e-01
-1.31758237e+00 6.91462576e-01 4.21207190e+00 6.51586115e-01
-6.23794198e-01 7.79847801e-02 5.53012908e-01 -1.13054691e-02
-1.88352749e-01 1.15233297e-02 -7.35442102e-01 8.65946352e-01
2.41558060e-01 -1.54560611e-01 2.57572740e-01 7.49580801e-01
-9.99637917e-02 1.26790792e-01 -1.14610136e+00 1.43212652e+00
3.81942570e-01 -1.04252815e+00 9.72267464e-02 9.21391547e-02
8.19055259e-01 -4.41312253e-01 2.24857226e-01 4.10347551e-01
-6.78350031e-02 -7.32487261e-01 4.89814788e-01 5.90569854e-01
6.65975273e-01 -1.03936040e+00 8.88916969e-01 8.74564871e-02
-1.40239882e+00 -6.40127361e-02 -3.62031072e-01 1.24345779e-01
-4.62264605e-02 3.57136935e-01 -4.29895937e-01 6.51835144e-01
7.91869044e-01 1.04837370e+00 -8.36505651e-01 1.19556308e+00
-2.28229403e-01 4.05576259e-01 -7.71639794e-02 4.73414540e-01
-2.10292429e-01 -1.31073907e-01 5.62914670e-01 1.21265841e+00
-9.43184420e-02 6.77504465e-02 3.40935051e-01 6.68027163e-01
-4.31601614e-01 1.14250638e-01 -2.47046024e-01 3.52437764e-01
7.20660627e-01 1.08125019e+00 -2.66573846e-01 -2.94664681e-01
-3.72776926e-01 1.43397284e+00 6.09133482e-01 2.74859458e-01
-6.80178463e-01 -3.05674911e-01 8.18779349e-01 7.16675892e-02
3.51079442e-02 3.89108770e-02 -3.94296199e-02 -1.48334086e+00
3.84409398e-01 -9.56444502e-01 7.03676045e-01 -3.78960550e-01
-1.85937822e+00 3.29386443e-01 1.04168519e-01 -1.61617732e+00
-3.18181291e-02 -1.07789464e-01 -6.04797244e-01 7.43444741e-01
-1.56920397e+00 -1.35846663e+00 -7.05905735e-01 5.92393756e-01
4.96769041e-01 -3.54985386e-01 6.14763677e-01 3.29166055e-01
-9.45324779e-01 1.45318830e+00 5.06258868e-02 4.35310543e-01
1.31071866e+00 -1.27147365e+00 4.47989523e-01 9.46671069e-01
1.27052858e-01 9.55676913e-01 4.52701151e-01 -6.72645867e-01
-1.05616963e+00 -1.19533706e+00 6.02900326e-01 -3.41002941e-01
1.60967290e-01 -2.67143428e-01 -1.19244909e+00 5.78822732e-01
-6.61178976e-02 2.53326505e-01 9.76038933e-01 1.94528535e-01
-1.03339016e+00 -3.28372359e-01 -1.21085119e+00 5.49743593e-01
1.45928288e+00 -7.22129226e-01 -5.57623029e-01 9.81016606e-02
1.76084861e-01 -2.17334196e-01 -6.82027817e-01 5.63728154e-01
5.82127512e-01 -8.72889996e-01 1.07055008e+00 -3.76617283e-01
4.20778304e-01 -4.82834011e-01 9.76122245e-02 -1.27994192e+00
-5.30695081e-01 -3.32164317e-01 -1.00889027e-01 1.69298828e+00
-3.03405002e-02 -8.77298176e-01 7.57469833e-01 6.68294072e-01
1.48266256e-01 -3.64083588e-01 -7.59521425e-01 -9.98315513e-01
-1.42999232e-01 1.01469420e-01 5.92829347e-01 1.34754121e+00
9.57270935e-02 1.85968995e-01 -6.05715036e-01 4.15676206e-01
1.14405048e+00 1.93027973e-01 1.02759981e+00 -1.08682585e+00
-4.51755583e-01 -3.61878216e-01 -8.60342503e-01 -1.01987684e+00
4.19658720e-01 -1.19160688e+00 -1.24102861e-01 -1.16921711e+00
8.85164440e-01 -6.32449031e-01 -5.53198457e-01 6.67487979e-01
-8.97293091e-01 6.48939550e-01 2.58144498e-01 6.66991711e-01
-7.85055280e-01 7.82721639e-01 8.21572483e-01 -3.77574623e-01
-1.38206318e-01 -2.15920687e-01 -8.76180768e-01 4.10385072e-01
8.15780282e-01 -3.65566283e-01 -3.19315225e-01 -2.15808243e-01
-1.07701458e-01 -5.85197806e-01 6.25007391e-01 -1.18212485e+00
3.55301321e-01 1.58322304e-01 6.70790076e-01 -3.32260460e-01
2.41853759e-01 -4.35065240e-01 5.55175394e-02 1.09700687e-01
-6.34184837e-01 -4.04375046e-02 -5.48280738e-02 7.02413976e-01
-3.43055099e-01 -8.97615403e-02 7.72147417e-01 -1.26037048e-02
-8.52076471e-01 5.08573890e-01 3.36351424e-01 1.70673087e-01
7.26046383e-01 -3.09584320e-01 -3.41079712e-01 -2.62570143e-01
-5.27875602e-01 4.07384872e-01 7.28029072e-01 6.38604820e-01
6.72186196e-01 -1.58017898e+00 -1.01746082e+00 1.32629931e-01
6.35040760e-01 -1.02049179e-01 5.33359349e-01 6.37753785e-01
1.00946635e-01 -1.68543503e-01 -2.74125159e-01 -6.02379858e-01
-1.49854720e+00 4.67258930e-01 2.18384326e-01 -1.05065235e-03
-7.29679942e-01 9.87907887e-01 4.92906034e-01 -5.80730557e-01
2.09801123e-01 5.81886351e-01 -2.38114402e-01 5.06060347e-02
9.56790745e-01 4.06534165e-01 -2.20298767e-01 -9.25314367e-01
-5.24072826e-01 7.26425231e-01 -6.92331672e-01 1.56882674e-01
1.13326180e+00 -2.63965935e-01 -6.58364147e-02 2.35391781e-01
1.36079574e+00 6.98825866e-02 -1.31635332e+00 -5.22309780e-01
-1.54657811e-01 -8.02292049e-01 -2.99633890e-01 -5.87784946e-01
-1.01322281e+00 4.51352686e-01 8.69682968e-01 -3.06554675e-01
9.53222215e-01 6.46595284e-03 8.45374823e-01 3.36018503e-01
4.22289848e-01 -9.91979182e-01 3.05457920e-01 1.35912701e-01
6.94544852e-01 -1.63813400e+00 1.54205218e-01 -5.49092591e-01
-7.36098945e-01 6.34757161e-01 9.21451092e-01 -1.10864460e-01
1.58740103e-01 -1.69801146e-01 -7.37364963e-02 1.25302836e-01
-6.85987845e-02 -3.38245422e-01 3.72390538e-01 6.04155123e-01
3.76379639e-01 1.10060632e-01 -3.09188783e-01 5.38674712e-01
-1.41635492e-01 -2.42874101e-01 1.43857509e-01 8.68043005e-01
7.16399923e-02 -1.24410915e+00 -2.51472652e-01 3.46296877e-01
-1.72594234e-01 1.89462602e-02 -4.61613119e-01 3.89393687e-01
1.52964219e-01 8.15481246e-01 7.18812644e-02 -4.90293592e-01
2.94023544e-01 -5.18875495e-02 6.21316493e-01 -4.60144043e-01
-4.30203944e-01 -3.73969704e-01 -1.02656551e-01 -3.32259685e-01
-5.85311532e-01 -9.58820283e-01 -1.18441999e+00 -1.82043061e-01
-2.88254201e-01 1.05745375e-01 2.06464395e-01 9.43396330e-01
4.96856004e-01 6.75107241e-02 9.65990841e-01 -8.24325204e-01
-3.19334120e-01 -7.26982832e-01 -3.16196263e-01 1.01477492e+00
2.36049443e-01 -8.55819881e-01 -3.82289797e-01 1.07828818e-01]
|
[14.708587646484375, 1.0096876621246338]
|
e8f7c3d1-3991-499f-ab09-10121ee77034
|
fadman-federated-anomaly-detection-across
|
2205.14196
| null |
https://arxiv.org/abs/2205.14196v1
|
https://arxiv.org/pdf/2205.14196v1.pdf
|
FadMan: Federated Anomaly Detection across Multiple Attributed Networks
|
Anomaly subgraph detection has been widely used in various applications, ranging from cyber attack in computer networks to malicious activities in social networks. Despite an increasing need for federated anomaly detection across multiple attributed networks, only a limited number of approaches are available for this problem. Federated anomaly detection faces two major challenges. One is that isolated data in most industries are restricted share with others for data privacy and security. The other is most of the centralized approaches training based on data integration. The main idea of federated anomaly detection is aligning private anomalies from local data owners on the public anomalies from the attributed network in the server through public anomalies to federate local anomalies. In each private attributed network, the detected anomaly subgraph is aligned with an anomaly subgraph in the public attributed network. The significant public anomaly subgraphs are selected for federated private anomalies while preventing local private data leakage. The proposed algorithm FadMan is a vertical federated learning framework for public node aligned with many private nodes of different features, and is validated on two tasks correlated anomaly detection on multiple attributed networks and anomaly detection on an attributeless network using five real-world datasets. In the first scenario, FadMan outperforms competitive methods by at least 12% accuracy at 10% noise level. In the second scenario, by analyzing the distribution of abnormal nodes, we find that the nodes of traffic anomalies are associated with the event of postgraduate entrance examination on the same day.
|
['Qiang Yang', 'Lixin Fan', 'Wenjun Wang', 'Ning Zhang', 'Nannan Wu']
|
2022-05-27
| null | null | null | null |
['data-integration']
|
['knowledge-base']
|
[ 6.65812287e-03 2.36210540e-01 -1.78804010e-01 -2.35245511e-01
-2.70951629e-01 -5.16513228e-01 3.52332383e-01 5.97415090e-01
-8.53983685e-02 4.84629422e-01 -1.09514810e-01 -4.06722307e-01
-5.33295989e-01 -9.30557907e-01 -4.28759784e-01 -7.90096939e-01
-3.45294207e-01 4.61101025e-01 2.25263268e-01 -1.86142758e-01
1.83373764e-01 8.10364187e-01 -1.39040840e+00 1.87195167e-01
7.75893211e-01 1.17599499e+00 -1.02328539e+00 3.11951548e-01
-5.78585565e-01 8.05914283e-01 -5.59165955e-01 -6.06584907e-01
6.64057851e-01 -3.05768937e-01 -8.51891160e-01 6.96143135e-02
3.41561109e-01 6.91423193e-02 -4.42481935e-01 1.26765811e+00
2.81173378e-01 -1.48565602e-02 2.41006672e-01 -2.08629584e+00
-1.01664059e-01 4.42494869e-01 -7.30152249e-01 5.34228683e-01
3.76935229e-02 1.70619860e-01 1.05953455e+00 -4.42420632e-01
6.44021928e-01 9.47652936e-01 3.17125916e-01 3.26453954e-01
-9.04508531e-01 -9.66250598e-01 6.15536094e-01 2.71380901e-01
-1.10125864e+00 -1.98043287e-02 8.82541716e-01 -3.50335032e-01
3.66567314e-01 5.02871990e-01 5.24040639e-01 8.94315004e-01
3.76974881e-01 2.86029071e-01 7.29209602e-01 4.77920696e-02
2.98161447e-01 2.56096631e-01 5.17354071e-01 5.26468456e-01
5.92333496e-01 -1.08877495e-01 -4.72507447e-01 -8.57661426e-01
-6.66964725e-02 7.49415636e-01 1.52674675e-01 -4.76180047e-01
-7.01067984e-01 7.29356706e-01 3.59532684e-01 2.94728011e-01
-5.34020662e-01 -3.94249350e-01 8.35663855e-01 8.26045096e-01
5.82635105e-01 1.00934848e-01 -6.05632722e-01 3.05103749e-01
-2.27764919e-01 9.42946374e-02 9.24409449e-01 6.47299826e-01
8.76513124e-01 1.30062640e-01 5.70249632e-02 3.56952220e-01
3.17274928e-01 2.90127963e-01 4.27203864e-01 -4.63748187e-01
3.82561177e-01 1.37300587e+00 -3.57282490e-01 -1.48581958e+00
-2.44588703e-01 -6.58221602e-01 -9.39709306e-01 1.79758757e-01
4.74668503e-01 -2.99977094e-01 -4.41468537e-01 1.34413457e+00
8.06909680e-01 4.21288013e-01 1.75022837e-02 5.63383579e-01
3.81370276e-01 2.42330238e-01 2.33085118e-02 -1.53261617e-01
1.18960154e+00 -6.50500238e-01 -8.18375349e-01 6.73552509e-03
8.94449413e-01 -4.24140960e-01 2.78883159e-01 4.88629669e-01
-2.03542635e-01 1.98803976e-01 -6.96972132e-01 6.77926064e-01
-8.13687265e-01 -6.96710169e-01 4.79871064e-01 7.61885107e-01
-5.89858592e-01 5.43620348e-01 -5.50151646e-01 -5.29933035e-01
6.44378364e-01 5.55498779e-01 -7.50455260e-01 -1.45663202e-01
-9.28690910e-01 2.71406084e-01 1.91849589e-01 -2.40495458e-01
-6.53150558e-01 -4.84541535e-01 -6.86144650e-01 3.55455689e-02
5.85756540e-01 -1.20377831e-01 7.90895164e-01 -1.26344395e+00
-5.16791821e-01 7.72498488e-01 1.84914440e-01 -6.52849138e-01
5.02393723e-01 -6.55440167e-02 -1.34398258e+00 -1.43488735e-01
1.89889491e-01 -5.49562514e-01 7.60948718e-01 -8.65973771e-01
-1.02509165e+00 -1.04644454e+00 -5.14325500e-01 -2.17415288e-01
-8.34529698e-01 -1.20066166e-01 2.28310287e-01 -3.43451709e-01
3.51979256e-01 -6.52751863e-01 -3.64347249e-01 -9.57055688e-02
-6.09313071e-01 -3.46852660e-01 1.89653504e+00 -4.87980634e-01
1.19783902e+00 -2.26840401e+00 -3.60079348e-01 1.09689867e+00
6.00292861e-01 1.40494838e-01 3.24898735e-02 3.51663709e-01
-4.26144391e-01 9.68068615e-02 -1.74541041e-01 1.05168559e-01
-2.49517992e-01 3.28312069e-01 -4.26837385e-01 8.65561724e-01
-1.61199123e-01 2.76420414e-01 -6.95090055e-01 -4.69439358e-01
2.30192766e-02 -2.37234667e-01 -4.91550863e-01 2.71257430e-01
1.57767460e-01 7.13096142e-01 -8.19381833e-01 1.05636501e+00
7.11535096e-01 -6.33003563e-02 2.58539200e-01 3.10134321e-01
6.95645884e-02 -3.21336031e-01 -1.58876050e+00 1.08842075e+00
5.41988462e-02 1.87722281e-01 4.35055196e-01 -1.16537285e+00
1.25332010e+00 5.25512457e-01 1.04416525e+00 -4.97752994e-01
1.38954431e-01 6.08745039e-01 1.53239399e-01 -4.04565156e-01
5.66302910e-02 4.27842706e-01 -5.72299138e-02 8.19866359e-01
1.57104775e-01 8.92014802e-01 -6.83864802e-02 5.30376434e-01
1.58441019e+00 -6.70626462e-01 2.39284158e-01 -2.13429164e-02
9.95100319e-01 -1.38242647e-01 8.36309791e-01 6.04068041e-01
-5.08830547e-01 2.49287058e-02 8.55741918e-01 -9.14248824e-01
-7.35176682e-01 -1.01935947e+00 6.94222227e-02 1.07752824e+00
4.59344126e-02 -2.97539860e-01 -3.63266915e-01 -1.55697227e+00
3.76606613e-01 4.84404325e-01 -3.48348618e-01 -4.06943530e-01
-2.75165141e-01 -4.34806556e-01 5.82626939e-01 -2.38599684e-02
4.73500133e-01 -1.09985530e+00 2.32683644e-02 9.35798287e-02
3.18071395e-02 -9.84588385e-01 -1.27725825e-01 1.47500053e-01
-5.97017884e-01 -1.77282214e+00 1.13447279e-01 -3.15210283e-01
8.31145942e-01 1.67263560e-02 8.91025543e-01 3.06449443e-01
-4.71407354e-01 6.62983358e-01 -2.02779770e-02 -7.69850373e-01
-4.89967942e-01 6.91043288e-02 3.95691544e-01 9.57739234e-01
7.52622306e-01 -7.77505398e-01 -3.80929112e-01 4.92173970e-01
-9.21826065e-01 -1.03044713e+00 4.32593554e-01 4.88068610e-01
3.92090708e-01 1.39246270e-01 7.55390465e-01 -1.41991878e+00
4.61318910e-01 -1.29494822e+00 -6.23248219e-01 -6.13900907e-02
-9.26248968e-01 -2.48206288e-01 7.65928745e-01 -7.65685439e-02
-5.87685645e-01 2.09751036e-02 3.66006326e-03 -5.38920164e-01
-5.86037755e-01 -7.63448253e-02 -3.60617489e-01 -1.22478150e-01
7.51561105e-01 7.47654960e-02 3.71055514e-01 -3.62106353e-01
-1.51787877e-01 7.33544171e-01 3.46927881e-01 -3.13314557e-01
1.06776464e+00 4.84231502e-01 4.50316459e-01 -7.21020222e-01
-5.94475150e-01 -7.79594600e-01 -3.29158217e-01 -1.48668230e-01
5.25169015e-01 -6.19210958e-01 -8.61303031e-01 5.02020121e-01
-7.64545858e-01 7.23005533e-01 -6.09215617e-01 2.76511222e-01
8.72849766e-03 4.12617773e-01 -1.49986193e-01 -8.36859167e-01
-4.74099517e-01 -7.47662723e-01 1.58441037e-01 6.32958412e-02
-1.78622499e-01 -9.39084947e-01 1.36476055e-01 1.84794918e-01
4.76630241e-01 6.37779951e-01 9.21840608e-01 -1.96971786e+00
-5.03469169e-01 -8.62010241e-01 7.85098895e-02 3.66487354e-01
4.78621781e-01 -4.53257523e-02 -1.03200567e+00 -4.00941223e-01
-6.87589049e-02 3.35417949e-02 2.27457851e-01 -1.69478402e-01
1.49662483e+00 -7.04712987e-01 -5.63911140e-01 4.49015558e-01
1.21029150e+00 3.73412855e-02 3.45985472e-01 4.24729705e-01
8.26376617e-01 6.90732598e-01 4.86759275e-01 6.12088799e-01
-1.19316533e-01 1.16000073e-02 1.02252150e+00 -5.38522005e-02
7.21908689e-01 -1.20006211e-01 9.42590758e-02 3.19834232e-01
1.22480310e-01 -1.35448977e-01 -8.47662151e-01 6.01696312e-01
-1.94792187e+00 -9.86599743e-01 -3.24369133e-01 2.25720739e+00
-3.52059342e-02 1.08677901e-01 3.58468890e-01 2.57919043e-01
9.27736342e-01 1.41852722e-01 -7.99157679e-01 -6.74872220e-01
-7.16900826e-02 -2.45715547e-02 5.03093183e-01 -7.69564956e-02
-1.01811469e+00 2.55020946e-01 4.60760355e+00 3.14884067e-01
-8.62164736e-01 1.28623337e-01 8.17795753e-01 1.04583120e-02
-1.58537284e-01 1.03491008e-01 -6.55608594e-01 4.76445347e-01
9.34812307e-01 -3.86429280e-01 3.51566300e-02 1.25771809e+00
-1.19589180e-01 3.79639417e-01 -9.71181035e-01 7.04084158e-01
-1.89992011e-01 -1.02622986e+00 2.26118177e-01 5.57109594e-01
6.35306180e-01 -1.46112666e-02 1.05026970e-02 1.53885752e-01
2.19037130e-01 -1.00159669e+00 -1.61694095e-01 4.11121607e-01
2.63447315e-01 -1.26772964e+00 8.91540051e-01 3.62238467e-01
-1.03449392e+00 -7.19670415e-01 3.06870061e-04 2.37467825e-01
-4.89260405e-01 6.76148117e-01 -5.01058757e-01 7.68684328e-01
8.87413681e-01 6.78335547e-01 -5.31659365e-01 8.61579657e-01
2.76849091e-01 6.71260655e-01 -2.31344044e-01 4.90655452e-01
3.17809463e-01 -3.30460131e-01 9.75297749e-01 6.77071452e-01
3.47625911e-01 -1.70383513e-01 3.71410340e-01 5.64055681e-01
-1.63656622e-01 5.90990067e-01 -1.46025109e+00 6.40180185e-02
4.02672529e-01 1.63562548e+00 -3.35898906e-01 9.05305594e-02
-6.78692639e-01 6.51511669e-01 6.05467930e-02 1.72641963e-01
-3.76065403e-01 -5.66095114e-01 1.17832434e+00 4.43373621e-01
-1.58512488e-01 4.53584969e-01 2.53758896e-02 -8.45122337e-01
1.57940183e-02 -1.17138004e+00 1.31938446e+00 2.08209187e-01
-1.66518223e+00 6.13069594e-01 -4.02164608e-01 -1.52006400e+00
-2.28251487e-01 -4.14313465e-01 -1.09731340e+00 7.26518273e-01
-1.10103071e+00 -1.08146381e+00 -4.26223427e-01 1.40762627e+00
5.44224717e-02 -9.08693016e-01 9.99632061e-01 3.69785070e-01
-7.79544830e-01 6.25477314e-01 1.35345748e-02 5.31121790e-01
6.69704437e-01 -9.92273629e-01 2.30793431e-01 1.12068081e+00
1.00941844e-01 2.79901803e-01 2.64689445e-01 -8.08127165e-01
-1.13376951e+00 -1.37668729e+00 6.60799325e-01 -2.77247071e-01
7.56972611e-01 -7.19331279e-02 -1.17831743e+00 9.26357210e-01
1.32105082e-01 9.77926672e-01 9.40927505e-01 1.89637735e-01
-3.34318101e-01 -4.92495447e-01 -1.71196663e+00 3.35727513e-01
7.10793793e-01 -1.42014503e-01 5.68303168e-02 5.39001465e-01
6.21853113e-01 -1.24278158e-01 -7.75750637e-01 1.88361838e-01
-7.11294487e-02 -1.01865077e+00 5.97845316e-01 -1.16818988e+00
-4.36154008e-02 -2.15376347e-01 -1.13771193e-01 -1.19231772e+00
-2.24977612e-01 -7.93658435e-01 -4.90301251e-01 1.41718197e+00
1.76267698e-01 -1.36195207e+00 1.03260183e+00 5.98162770e-01
8.73766914e-02 -6.23593330e-01 -1.22239161e+00 -5.55414200e-01
-3.08300555e-01 -2.15126395e-01 8.86682808e-01 1.26514626e+00
-2.11896479e-01 1.64994001e-02 -8.51514041e-02 4.77776587e-01
9.61694300e-01 -3.97716761e-02 9.02700782e-01 -1.78299916e+00
1.40384495e-01 -1.43353388e-01 -9.86450195e-01 2.28717878e-01
3.13550204e-01 -1.05965424e+00 -9.16961670e-01 -8.52370322e-01
-2.45191827e-01 -4.22513127e-01 -8.38245749e-01 6.76641405e-01
2.72597671e-01 -4.87158149e-02 -3.20834368e-01 -4.83622439e-02
-4.77009654e-01 2.18312502e-01 6.80854023e-01 -2.25126281e-01
-1.65680975e-01 5.18449724e-01 -7.90485084e-01 9.81197298e-01
8.39005053e-01 -7.14862704e-01 -2.79646993e-01 3.37408543e-01
-1.58853471e-01 -8.28414038e-02 5.65837502e-01 -1.01364744e+00
3.39055806e-01 -1.97021127e-01 3.64968121e-01 -6.24900699e-01
-3.46935660e-01 -1.51332605e+00 -2.50922870e-02 6.08000875e-01
-6.97270259e-02 4.81608510e-01 -3.65291275e-02 1.12220335e+00
-2.69554377e-01 2.14213520e-01 6.73526645e-01 -1.48392886e-01
-7.12171018e-01 8.01086545e-01 -2.57170171e-01 2.18659699e-01
1.58889711e+00 -8.81813467e-02 -2.90118814e-01 -4.23441440e-01
-8.63517165e-01 5.49733937e-01 4.31520566e-02 5.04677236e-01
5.49111903e-01 -1.20646393e+00 -6.19468272e-01 6.11912847e-01
3.48892152e-01 -1.89019053e-03 3.14128518e-01 8.36167157e-01
-4.10647616e-02 5.25532365e-02 -3.87752116e-01 -6.55963659e-01
-1.57589006e+00 5.67793608e-01 5.08049488e-01 -3.48591030e-01
-7.24306881e-01 3.01412523e-01 -2.39039674e-01 -7.46252060e-01
4.48752850e-01 6.18100762e-01 -2.40887746e-01 -2.11133864e-02
3.86676103e-01 8.10606182e-01 3.74148190e-01 -6.61583245e-01
-5.46850502e-01 -4.96505424e-02 -4.50753599e-01 4.44585562e-01
1.13532162e+00 8.74806941e-02 -5.72122693e-01 2.80579418e-01
1.11143529e+00 2.15055883e-01 -4.09875602e-01 -5.77826917e-01
2.66591072e-01 -6.04052365e-01 -3.76604706e-01 -4.58476275e-01
-1.68087363e+00 3.09825003e-01 9.00125742e-01 4.24394190e-01
1.13722587e+00 -1.91429362e-01 6.18265271e-01 4.75859284e-01
1.25422597e-01 -8.38565886e-01 8.48086178e-02 3.44649255e-01
1.98122516e-01 -1.17610109e+00 -2.97465990e-03 -2.78820157e-01
-6.63447618e-01 9.75471437e-01 1.21773934e+00 -1.30552649e-01
9.83941615e-01 3.35003971e-03 1.53475419e-01 -6.23974502e-01
-6.26332045e-01 4.30241466e-01 7.60478601e-02 5.13746381e-01
4.09277305e-02 -1.70542464e-01 -7.23481476e-02 6.13282323e-01
2.12243080e-01 -5.52455306e-01 4.62785333e-01 8.59981477e-01
-2.74520129e-01 -1.19338274e+00 -5.17733276e-01 9.24917579e-01
-9.15616095e-01 4.43541080e-01 -5.16304314e-01 7.00133920e-01
1.93150595e-01 9.27291095e-01 4.25091952e-01 -2.55571783e-01
4.15111899e-01 6.09703183e-01 -4.47512299e-01 -4.39404786e-01
-9.05498385e-01 -3.82903069e-01 -2.05921412e-01 -9.13708627e-01
1.04749471e-01 -5.43971956e-01 -1.18777132e+00 -5.12158990e-01
-6.53449968e-02 4.85686451e-01 4.57452983e-01 8.66423130e-01
6.45579398e-01 4.68968928e-01 1.21134090e+00 2.47692898e-01
-6.50618970e-01 -5.76179087e-01 -1.08935976e+00 8.07344794e-01
4.82465804e-01 -3.54598165e-01 -6.62116647e-01 -7.14643955e-01]
|
[6.627865791320801, 5.75784158706665]
|
28d83e22-2549-4969-9b58-a402022e7c9f
|
the-role-of-visual-saliency-in-the-automation
|
1812.11960
| null |
http://arxiv.org/abs/1812.11960v1
|
http://arxiv.org/pdf/1812.11960v1.pdf
|
The role of visual saliency in the automation of seismic interpretation
|
In this paper, we propose a workflow based on SalSi for the detection and
delineation of geological structures such as salt domes. SalSi is a seismic
attribute designed based on the modeling of human visual system that detects
the salient features and captures the spatial correlation within seismic
volumes for delineating seismic structures. Using SalSi, we can not only
highlight the neighboring regions of salt domes to assist a seismic interpreter
but also delineate such structures using a region growing method and
post-processing. The proposed delineation workflow detects the salt-dome
boundary with very good precision and accuracy. Experimental results show the
effectiveness of the proposed workflow on a real seismic dataset acquired from
the North Sea, F3 block. For the subjective evaluation of the results of
different salt-dome delineation algorithms, we have used a reference salt-dome
boundary interpreted by a geophysicist. For the objective evaluation of
results, we have used five different metrics based on pixels, shape, and
curvedness to establish the effectiveness of the proposed workflow. The
proposed workflow is not only fast but also yields better results as compared
to other salt-dome delineation algorithms and shows a promising potential in
seismic interpretation.
|
['Tariq Alshawi', 'Ghassan AlRegib', 'Zhiling Long', 'Muhammad Amir Shafiq']
|
2018-12-31
| null | null | null | null |
['seismic-interpretation']
|
['miscellaneous']
|
[-1.37259126e-01 -2.37167608e-02 1.01749933e+00 -1.52559966e-01
-5.25522172e-01 -7.18570709e-01 6.91888213e-01 6.06876791e-01
-5.20193160e-01 1.70865685e-01 3.87293398e-01 -3.00955415e-01
-3.14299613e-01 -7.82136202e-01 -1.63173661e-01 -8.10297608e-01
-5.46294034e-01 4.98640507e-01 8.83317709e-01 -3.79590780e-01
7.80086219e-01 9.91915762e-01 -1.29700291e+00 1.60179988e-01
6.47821963e-01 4.60496873e-01 5.01814485e-01 5.47637761e-01
-1.58183843e-01 2.95631617e-01 -5.32218754e-01 3.34849834e-01
3.31222057e-01 3.81962284e-02 -4.63957340e-01 -5.89380637e-02
3.11778873e-01 -7.67761245e-02 -3.82515192e-02 8.65205646e-01
5.77884853e-01 2.65462250e-01 8.78887296e-01 -6.07474208e-01
1.18574262e-01 6.32109940e-01 -7.80393541e-01 5.19296348e-01
2.37943009e-01 2.34469235e-01 7.33250022e-01 -1.31337786e+00
6.18508577e-01 1.22514069e+00 7.11726964e-01 -3.85104924e-01
-8.11797202e-01 -4.67299908e-01 -1.46453857e-01 1.70546714e-02
-1.67414367e+00 -1.88013270e-01 7.27576733e-01 -1.02306688e+00
4.94237423e-01 4.53241378e-01 5.69715917e-01 -1.52880073e-01
9.29852575e-02 5.11955559e-01 1.15595126e+00 -4.82410997e-01
4.76271331e-01 -4.35787827e-01 1.98260412e-01 7.56739795e-01
3.73498976e-01 -5.25686294e-02 -5.19129753e-01 -2.25982830e-01
6.93034708e-01 -1.56924739e-01 -5.84051847e-01 -1.86329722e-01
-1.18134928e+00 5.95170319e-01 3.94678324e-01 5.65834761e-01
-4.63845968e-01 1.44056231e-01 4.04141188e-01 -2.23446846e-01
2.58449197e-01 3.57870549e-01 3.65530819e-01 -6.99527264e-02
-1.50935388e+00 3.33334416e-01 3.50323081e-01 1.51521474e-01
4.62659568e-01 2.66190171e-01 1.48343772e-01 3.23259175e-01
6.31818950e-01 7.58551896e-01 8.53205398e-02 -4.01437402e-01
5.37902176e-01 6.21824920e-01 4.24590468e-01 -1.14011323e+00
-7.51548290e-01 -1.81438521e-01 -5.64465284e-01 1.04816341e+00
3.62001032e-01 -2.97717974e-02 -1.03081799e+00 7.81914234e-01
2.44715095e-01 -1.81606084e-01 2.53073782e-01 1.02290785e+00
1.00085592e+00 8.65001738e-01 4.17166986e-02 2.68609017e-01
1.39252579e+00 -2.27720574e-01 -5.60999095e-01 -9.09939595e-03
3.92771035e-01 -7.11550057e-01 8.97492170e-01 1.78471506e-01
-9.62253451e-01 -3.59909356e-01 -1.12462723e+00 5.05503953e-01
-1.87252194e-01 2.38753468e-01 1.22298084e-01 5.16374290e-01
-1.02705669e+00 4.52182144e-01 -9.76147056e-01 -3.03032458e-01
-7.36623444e-03 -1.79611057e-01 -2.80713081e-01 4.84060436e-01
-9.59365964e-01 7.07768381e-01 4.67188865e-01 5.07938266e-01
-1.00804019e+00 -4.05072719e-01 -6.89040363e-01 4.31309670e-01
7.24256486e-02 7.36129060e-02 5.91889501e-01 -3.30455303e-01
-6.24247134e-01 7.75681853e-01 2.14127362e-01 -1.94270566e-01
7.61689484e-01 -1.80495102e-02 -3.11992526e-01 4.75569963e-01
9.42833945e-02 5.10181747e-02 9.68500227e-02 -1.72376680e+00
-8.24717879e-01 -2.70120263e-01 -3.52529943e-01 3.29530984e-01
9.72545817e-02 1.66073322e-01 -1.80382311e-01 -7.10202098e-01
7.95144975e-01 -5.47806442e-01 -1.28908142e-01 -2.68209595e-02
-2.40550146e-01 2.28439495e-01 1.10394239e+00 -1.01112068e+00
1.33456826e+00 -2.39309239e+00 -4.48411971e-01 8.88206899e-01
1.79418921e-01 3.68868381e-01 5.44954121e-01 6.55856609e-01
9.12125632e-02 1.75189748e-01 -1.00627434e+00 1.83961332e-01
-1.71496913e-01 -6.27758950e-02 -5.42471893e-02 4.83464003e-01
-3.36357921e-01 1.28560603e-01 -7.31020510e-01 -5.55398881e-01
3.19865346e-01 2.18357831e-01 2.00018689e-01 1.62004545e-01
3.04561436e-01 3.49455595e-01 -4.35777098e-01 7.52969027e-01
1.10126066e+00 4.13566738e-01 -5.20373844e-02 -1.83931231e-01
-9.73158956e-01 -4.39668536e-01 -1.67118609e+00 1.00688016e+00
-5.29114380e-02 8.28378201e-01 2.32026771e-01 -6.47617877e-01
1.66882896e+00 4.38697606e-01 1.95349216e-01 -5.47682762e-01
-2.13966817e-01 6.92534089e-01 -2.77297124e-02 -1.09464025e+00
7.75900602e-01 1.80945769e-02 1.64044097e-01 3.28720510e-01
-7.57903874e-01 -2.19321162e-01 3.61312568e-01 2.99005747e-01
7.15567410e-01 1.58573359e-01 1.14571974e-01 -8.62393796e-01
4.61787969e-01 1.37236372e-01 1.39376551e-01 6.59478843e-01
3.38377524e-03 7.89130688e-01 2.38672391e-01 -6.19821489e-01
-1.04813337e+00 -1.25980270e+00 -2.07281619e-01 6.37399733e-01
4.16141301e-01 1.63841680e-01 -5.31793118e-01 -2.18839839e-01
-6.59284964e-02 4.50978935e-01 -6.83690906e-01 4.68248188e-01
-7.39422560e-01 -8.17822933e-01 7.20583081e-01 5.20836234e-01
6.00802124e-01 -1.13262415e+00 -1.53167772e+00 1.57161549e-01
-1.08075410e-01 -6.07113659e-01 -1.03663214e-01 1.55049667e-01
-8.78990114e-01 -7.89524734e-01 -9.05142903e-01 -7.80972540e-01
7.52156675e-01 2.10434884e-01 8.36459398e-01 1.90185815e-01
-2.10638553e-01 -8.95800348e-03 -5.56523144e-01 -1.62841037e-01
-4.59582329e-01 -3.65562797e-01 -6.86441720e-01 6.46445677e-02
-3.26832891e-01 -3.41432810e-01 -7.96875834e-01 4.03442889e-01
-1.01103413e+00 8.87518525e-02 3.30740273e-01 1.85465470e-01
1.47608340e-01 1.78726222e-02 1.62334070e-01 -6.39105618e-01
4.67939109e-01 -4.69243169e-01 -7.32627690e-01 3.93024713e-01
-2.89873183e-01 2.32872769e-01 5.52667826e-02 2.96360403e-01
-1.28056800e+00 -4.29520197e-02 -9.00767818e-02 3.34650338e-01
-1.49734885e-01 9.28262830e-01 1.46221757e-01 -1.79670841e-01
7.35488772e-01 2.35751331e-01 -3.81698161e-01 -6.41870677e-01
-8.18917379e-02 8.83370042e-01 8.07271004e-01 -2.68857509e-01
7.55128205e-01 8.32222760e-01 2.06149235e-01 -9.56677318e-01
1.20421804e-01 -7.91943014e-01 -6.34789348e-01 -8.41063678e-01
7.95806766e-01 -5.01384139e-01 -5.47989845e-01 3.40412706e-01
-8.54653597e-01 -1.98130663e-02 1.95457056e-01 4.46963102e-01
2.76153069e-02 6.76442266e-01 -3.05982441e-01 -1.30253875e+00
-7.17633128e-01 -9.54799235e-01 7.91309297e-01 4.70789135e-01
-2.19822805e-02 -9.89130020e-01 1.76775664e-01 -8.26445222e-02
5.25822401e-01 8.97362769e-01 7.16787159e-01 -3.79441708e-01
-4.04340446e-01 -3.87466289e-02 -1.86793685e-01 -4.83239770e-01
8.36092681e-02 4.48277175e-01 -8.81291032e-01 -4.35980223e-02
-3.45284700e-01 3.33541870e-01 1.07645571e+00 3.53869647e-01
2.53321499e-01 1.97666094e-01 -3.10471237e-01 4.57086384e-01
1.82397664e+00 8.42845261e-01 5.99887490e-01 8.90050054e-01
5.18881977e-01 6.52466178e-01 8.39126647e-01 8.69730651e-01
8.42435956e-02 3.56570333e-01 6.21006429e-01 -6.94443703e-01
-9.42546036e-03 9.53775644e-02 8.79223272e-02 3.92771930e-01
-4.75323588e-01 -1.44083187e-01 -1.65616333e+00 1.00929177e+00
-1.74066591e+00 -8.39995503e-01 -7.33447969e-01 2.20415616e+00
2.09138945e-01 1.59460321e-01 -4.11402173e-02 4.31756675e-01
8.33934605e-01 2.82159686e-01 1.46782205e-01 -4.05270398e-01
-3.42682451e-01 1.35703906e-01 7.17407107e-01 7.85112083e-01
-1.06763840e+00 6.60164475e-01 5.18021107e+00 3.85143250e-01
-1.36153305e+00 -1.04190558e-01 3.48353177e-01 5.94140232e-01
-4.45471525e-01 1.22669250e-01 -4.09963369e-01 3.66199970e-01
3.86592090e-01 7.90187903e-03 -3.91895890e-01 4.67078060e-01
1.00399613e+00 -7.04311848e-01 -3.09300125e-01 5.59853911e-01
-9.53679010e-02 -1.45036232e+00 -8.32891613e-02 -3.82027239e-01
5.96891820e-01 -1.43417984e-01 -4.66681540e-01 -5.10120451e-01
1.45623341e-01 -7.34536052e-01 1.38822985e+00 8.10857415e-01
4.24785972e-01 -7.93180227e-01 8.86106253e-01 -4.76556309e-02
-1.69590449e+00 1.29009187e-01 3.02201044e-02 4.36369702e-02
4.17093068e-01 3.06574285e-01 -1.23139048e+00 6.08083546e-01
6.65383756e-01 1.42238662e-01 -6.58197761e-01 1.81251478e+00
-1.91439733e-01 6.55581832e-01 -3.22685808e-01 1.06745092e-02
7.57350266e-01 -3.13854992e-01 5.05522847e-01 1.81276250e+00
7.30728030e-01 8.72637853e-02 6.13328852e-02 6.89238608e-01
6.40397191e-01 4.30450231e-01 -3.78607839e-01 3.29516351e-01
7.06487656e-01 1.08507121e+00 -1.37041092e+00 -4.92857277e-01
-3.83272916e-02 2.51131862e-01 -4.14126694e-01 3.18684846e-01
-4.86775547e-01 -6.36848211e-01 -1.02346711e-01 5.95569074e-01
2.81014919e-01 -5.30726612e-01 -8.41647744e-01 -2.49798223e-01
2.38826312e-02 -3.48532677e-01 5.76120377e-01 -1.08695745e+00
-2.26029530e-01 6.88528478e-01 2.57441431e-01 -1.35822833e+00
1.85946465e-01 -2.92258650e-01 -9.68420267e-01 1.01674688e+00
-1.02172816e+00 -9.52969849e-01 -7.94242084e-01 7.58618936e-02
3.15621734e-01 7.14780390e-02 4.22011256e-01 1.17520176e-01
-1.63468525e-01 -1.77065134e-01 5.90998411e-01 2.01828718e-01
3.09613585e-01 -1.36838746e+00 6.18091524e-01 1.33173347e+00
-1.15559362e-01 1.82696134e-01 1.08350682e+00 -9.10910428e-01
-6.61860108e-01 -6.06119990e-01 6.61872387e-01 2.86582857e-01
4.09883291e-01 -1.15990371e-01 -1.35995150e+00 1.80114686e-01
1.06024690e-01 -2.95173973e-01 1.58877701e-01 -6.12853289e-01
1.10784858e-01 1.07623786e-02 -1.05926383e+00 4.55044985e-01
3.48776042e-01 4.20029694e-03 -7.83593714e-01 -1.40822828e-01
-3.40705454e-01 -3.11238080e-01 -7.27204859e-01 3.63721013e-01
5.05507290e-01 -1.26158929e+00 7.57806897e-01 3.96555424e-01
2.09644333e-01 -9.73100781e-01 1.49822503e-01 -1.12495685e+00
-7.85086751e-02 -2.89338052e-01 8.47949028e-01 1.00736976e+00
5.10109365e-01 -3.20413232e-01 6.97165787e-01 3.46512496e-01
-5.27364671e-01 -2.18506217e-01 -8.87596846e-01 -5.40955603e-01
-2.79203266e-01 -5.17811365e-02 4.67481107e-01 7.90132105e-01
4.24439088e-02 -5.03898799e-01 -4.28833887e-02 6.65267646e-01
7.06779003e-01 1.69281617e-01 4.67227966e-01 -1.27206159e+00
1.53004438e-01 -4.30948764e-01 -5.28420389e-01 -3.44909698e-01
-8.36455941e-01 -5.13395548e-01 4.19446737e-01 -1.91786051e+00
1.07678913e-01 -6.49753988e-01 -4.27215174e-02 4.76334631e-01
-2.74004668e-01 1.10910743e-01 1.17438860e-01 4.48827267e-01
9.60334316e-02 2.12790802e-01 9.59479928e-01 1.01681158e-01
-2.65014619e-01 -2.48304024e-01 1.58322603e-01 8.06034684e-01
5.81674635e-01 -5.36728740e-01 -6.38361275e-02 -4.12766576e-01
1.13539107e-01 2.50671685e-01 2.37096936e-01 -1.34368634e+00
2.77687103e-01 -1.27778754e-01 1.48576379e-01 -9.32863235e-01
-7.67541006e-02 -5.76590419e-01 3.60291094e-01 9.51298177e-01
-1.16646677e-01 3.14628154e-01 7.33741596e-02 2.38910273e-01
-4.02908504e-01 -8.14927876e-01 8.67378533e-01 -2.00628534e-01
-1.25880980e+00 -3.98566395e-01 -6.84437513e-01 -1.81449354e-01
9.31123555e-01 -6.48444057e-01 -2.49096945e-01 -1.53997168e-01
-7.41558433e-01 3.73828746e-02 5.08484602e-01 -1.54205963e-01
8.83155704e-01 -7.13002145e-01 -1.07668626e+00 5.86077943e-02
2.65094727e-01 7.11851045e-02 3.95163387e-01 6.58626139e-01
-1.78606439e+00 -1.71021536e-01 -3.88438940e-01 -5.79143524e-01
-1.49209297e+00 -1.41618520e-01 4.71727312e-01 9.23445374e-02
-1.06060839e+00 6.99686170e-01 2.71311045e-01 2.25554824e-01
3.06343902e-02 -5.09770632e-01 -8.09422731e-01 1.06133416e-01
4.47940469e-01 6.84820354e-01 6.73388541e-02 -7.54125357e-01
-5.62734663e-01 8.67692709e-01 5.06428599e-01 -5.87810874e-01
1.30619144e+00 -4.68348712e-02 2.17888013e-01 4.71859157e-01
3.97833586e-01 3.01782668e-01 -1.25283420e+00 -4.83719669e-02
5.17393708e-01 -6.80218220e-01 1.14538394e-01 -8.34177434e-01
-1.06704712e+00 8.50354850e-01 1.06298971e+00 1.29823998e-01
8.78301919e-01 -1.51922926e-01 2.29198635e-01 -7.82612339e-03
1.06888421e-01 -9.28968608e-01 -2.85814643e-01 3.10098380e-01
1.18227971e+00 -9.73050654e-01 2.54348516e-02 -2.83239454e-01
-8.08985233e-01 1.72102511e+00 1.22355767e-01 3.10867699e-03
5.80374897e-01 6.85858250e-01 5.20566702e-01 -6.31434083e-01
4.35119942e-02 -1.42002091e-01 9.30021778e-02 3.42225641e-01
4.96321350e-01 1.56014666e-01 -7.07737386e-01 7.38700852e-02
2.23569735e-03 -2.97016859e-01 8.56966853e-01 1.30212903e+00
-1.23428535e+00 -4.42570776e-01 -1.25528312e+00 3.72403744e-03
-1.45008624e-01 -4.38811742e-02 -1.16050020e-01 8.28110337e-01
2.30992556e-01 6.90364957e-01 3.45533192e-01 -7.11229295e-02
4.00638729e-01 -3.67820591e-01 -1.71425492e-01 -2.48596206e-01
-8.60909164e-01 3.01644474e-01 2.32271686e-01 2.33104184e-01
-3.58582318e-01 -9.64033067e-01 -1.89025187e+00 1.01899125e-01
9.74107385e-02 5.91074288e-01 8.82528901e-01 8.44816923e-01
-1.96047023e-01 3.59199822e-01 5.54080248e-01 -7.68114030e-01
2.01316774e-02 -1.12878966e+00 -8.07397604e-01 6.32647514e-01
-6.46178126e-02 -7.01116443e-01 -4.52270687e-01 2.76351869e-01]
|
[9.418212890625, -0.7180619239807129]
|
6d45da0a-4314-4f8a-aea4-b35820c14ca2
|
open-world-compositional-zero-shot-learning
|
2101.12609
| null |
https://arxiv.org/abs/2101.12609v3
|
https://arxiv.org/pdf/2101.12609v3.pdf
|
Open World Compositional Zero-Shot Learning
|
Compositional Zero-Shot learning (CZSL) requires to recognize state-object compositions unseen during training. In this work, instead of assuming prior knowledge about the unseen compositions, we operate in the open world setting, where the search space includes a large number of unseen compositions some of which might be unfeasible. In this setting, we start from the cosine similarity between visual features and compositional embeddings. After estimating the feasibility score of each composition, we use these scores to either directly mask the output space or as a margin for the cosine similarity between visual features and compositional embeddings during training. Our experiments on two standard CZSL benchmarks show that all the methods suffer severe performance degradation when applied in the open world setting. While our simple CZSL model achieves state-of-the-art performances in the closed world scenario, our feasibility scores boost the performance of our approach in the open world setting, clearly outperforming the previous state of the art.
|
['Zeynep Akata', 'Yongqin Xian', 'Muhammad Ferjad Naeem', 'Massimiliano Mancini']
|
2021-01-29
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Mancini_Open_World_Compositional_Zero-Shot_Learning_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Mancini_Open_World_Compositional_Zero-Shot_Learning_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['compositional-zero-shot-learning']
|
['computer-vision']
|
[ 2.77937561e-01 -3.29977348e-02 -2.05402330e-01 2.69226998e-01
-5.83182156e-01 -8.34169209e-01 8.70234013e-01 -5.03914654e-02
-3.83625895e-01 3.69437069e-01 2.38656759e-01 -1.36546567e-01
1.68677211e-01 -6.76318824e-01 -8.36705267e-01 -8.93436134e-01
7.72494897e-02 5.99778712e-01 5.12690127e-01 -1.36573002e-01
1.89399421e-01 1.34844631e-01 -1.87826407e+00 4.22172070e-01
5.06478608e-01 1.10081756e+00 1.74851239e-01 6.75084293e-01
-1.35495022e-01 6.38843834e-01 -2.92347640e-01 -4.36150223e-01
6.21353924e-01 -2.47844577e-01 -6.26128972e-01 1.93684101e-01
7.69511521e-01 -1.14433892e-01 -2.43835762e-01 1.32506740e+00
2.98148066e-01 1.90882564e-01 6.53822005e-01 -1.35014188e+00
-8.33307922e-01 5.35518169e-01 -1.43526375e-01 1.21028006e-01
3.94947380e-01 4.25133556e-01 1.53540707e+00 -1.17998278e+00
1.04935217e+00 9.26897466e-01 4.68067735e-01 5.57627678e-01
-1.45448816e+00 -5.46647072e-01 4.47430849e-01 3.97377104e-01
-1.23929799e+00 -7.28164256e-01 6.75503790e-01 -6.65580571e-01
9.78605747e-01 1.35773495e-01 5.79019010e-01 1.12873137e+00
-4.74823087e-01 8.98976326e-01 7.55092323e-01 -5.87845147e-01
5.30603945e-01 8.92918706e-02 -3.28341685e-02 9.48270798e-01
1.09858215e-01 1.69352293e-01 -7.24734783e-01 -1.59008905e-01
4.46357816e-01 1.28481284e-01 -3.35616082e-01 -1.21199620e+00
-1.58304369e+00 7.97744572e-01 6.59644425e-01 2.25707337e-01
-9.74535346e-02 1.47392988e-01 2.84581751e-01 4.95352417e-01
2.31136009e-01 6.56090975e-01 -2.32817650e-01 -3.36370580e-02
-7.88806438e-01 2.62350857e-01 7.92141438e-01 9.63177681e-01
7.95267761e-01 -1.41733348e-01 -8.38844180e-02 4.93594527e-01
-1.84534416e-02 6.93893358e-02 5.20792067e-01 -6.88148320e-01
6.96320355e-01 6.92840159e-01 7.26967752e-02 -1.96057066e-01
1.74905643e-01 -1.88516915e-01 -2.85397500e-01 4.21595693e-01
7.35766411e-01 6.08559139e-02 -1.09791529e+00 1.76351225e+00
3.57328504e-01 3.90599251e-01 7.03571737e-02 9.77441609e-01
3.33644241e-01 5.76086223e-01 -1.49611577e-01 -3.33349896e-03
1.19450867e+00 -1.49105465e+00 -4.52723026e-01 -2.24649116e-01
3.60881090e-01 -8.22867572e-01 1.28960359e+00 3.80080849e-01
-8.61990929e-01 -4.15672153e-01 -1.45773077e+00 -7.15166479e-02
-6.37028813e-01 -1.34410247e-01 4.03412431e-01 5.23897350e-01
-8.88358772e-01 6.64535344e-01 -9.17407751e-01 -4.28145558e-01
3.75578672e-01 2.74862498e-01 -4.22318906e-01 -3.06342095e-01
-7.17559040e-01 6.88497841e-01 4.01085645e-01 -2.87391037e-01
-1.23899209e+00 -8.23050976e-01 -9.68527079e-01 1.88236237e-01
8.06227028e-01 -5.96712470e-01 1.25107908e+00 -8.02992880e-01
-1.41011393e+00 6.38633788e-01 -1.63613126e-01 -2.63540536e-01
6.51543558e-01 -1.55240640e-01 -1.08750165e-01 1.38090536e-01
-1.79906920e-01 5.65454185e-01 9.39129829e-01 -1.29351616e+00
-4.50172663e-01 -1.46369562e-01 1.43153027e-01 1.49035081e-01
-4.01799470e-01 -1.87766030e-01 -3.62087220e-01 -4.62845713e-01
1.14388250e-01 -9.60664272e-01 -7.12224543e-02 4.94551569e-01
-1.99509576e-01 -2.33632028e-01 8.98548126e-01 -8.81318524e-02
8.23181450e-01 -2.34886742e+00 4.06097978e-01 6.85041621e-02
2.59258747e-01 2.85773098e-01 -2.54148334e-01 5.49777746e-01
-5.29388003e-02 -3.10741309e-02 -2.20679641e-01 -3.70488852e-01
4.15778011e-01 1.18808098e-01 -3.64243984e-01 6.04680657e-01
2.76509494e-01 9.41250980e-01 -1.13171589e+00 -2.78701723e-01
3.50937992e-01 2.17078909e-01 -7.17927217e-01 4.18856949e-01
-5.26374280e-01 9.60811973e-02 -1.01631530e-01 5.76917350e-01
2.34337911e-01 -4.92083549e-01 2.78894037e-01 -2.58379634e-02
4.11484055e-02 1.87828243e-01 -1.26594591e+00 1.99780118e+00
-4.60104495e-01 4.61831003e-01 -2.04826310e-01 -9.03185070e-01
3.79811436e-01 3.95109057e-01 1.47720590e-01 -5.44967651e-01
-1.00914262e-01 1.56521514e-01 1.47138551e-01 -4.71096367e-01
1.92157269e-01 -3.65093619e-01 -1.02084614e-02 6.05695903e-01
4.50007349e-01 6.59338459e-02 2.55454212e-01 1.16587663e-02
1.16498148e+00 2.43305251e-01 4.44847375e-01 -1.62329152e-01
4.24314052e-01 -1.57821909e-01 3.52631807e-01 4.69500154e-01
-9.04249027e-02 7.96926022e-01 5.46636105e-01 -5.08697808e-01
-1.30851626e+00 -1.68277264e+00 2.86329865e-01 1.26658630e+00
3.32062632e-01 -6.54774368e-01 -4.58232492e-01 -9.94447947e-01
-3.58133204e-02 4.57900912e-01 -5.69777668e-01 -1.11450851e-01
-5.74854314e-01 -2.52312809e-01 1.27733603e-01 5.17257214e-01
7.44912699e-02 -8.20795715e-01 -5.79383314e-01 2.39842925e-02
1.53130457e-01 -1.09682846e+00 -6.65382445e-01 1.96386531e-01
-5.31652510e-01 -1.14541388e+00 -5.82473338e-01 -9.83913720e-01
7.26420701e-01 1.19669177e-01 9.80210304e-01 -8.32524672e-02
-4.70283508e-01 -8.31570923e-02 -2.32959792e-01 -6.94889948e-03
-2.56413907e-01 -1.51659073e-02 -1.33133918e-01 1.22356497e-01
2.11988404e-01 -6.01678789e-01 -7.05732048e-01 1.00221597e-01
-8.10187161e-01 -3.57428044e-02 3.58473659e-01 9.45538104e-01
3.43856096e-01 -1.73191205e-01 2.81121850e-01 -7.42733419e-01
3.29414278e-01 -3.27149689e-01 -6.19956076e-01 5.30921519e-01
-6.48298323e-01 4.43463773e-01 8.75747681e-01 -8.98888648e-01
-6.51742637e-01 2.49985635e-01 2.02347592e-01 -8.86501551e-01
-3.77770811e-02 -9.49784517e-02 -2.68179029e-01 2.97879130e-02
6.62897706e-01 9.07270312e-02 -1.86148901e-02 -5.93260169e-01
5.43563426e-01 3.69468600e-01 5.88177681e-01 -6.72121525e-01
1.15237296e+00 6.14242852e-01 -1.53985754e-01 -4.64189738e-01
-7.27923870e-01 -6.13363862e-01 -6.10390604e-01 1.81024924e-01
8.39962780e-01 -9.37232018e-01 -6.55224383e-01 5.96257448e-02
-6.97879493e-01 -3.47898513e-01 -5.35196543e-01 1.79612324e-01
-5.78515351e-01 4.64592546e-01 -5.25691390e-01 -6.87352180e-01
-2.71487951e-01 -1.48796785e+00 1.08566821e+00 -1.01787187e-01
-3.31266403e-01 -8.73691261e-01 1.17520370e-01 2.12109402e-01
1.76400423e-01 1.79801181e-01 1.28816688e+00 -8.26860249e-01
-8.31616402e-01 -2.34559014e-01 -1.16730392e-01 2.58838654e-01
8.46696496e-02 -2.11056009e-01 -1.09431601e+00 -3.48244220e-01
-3.99119824e-01 -5.04258871e-01 9.30259526e-01 -4.47657228e-01
8.78824234e-01 -2.60700285e-01 -1.56853780e-01 4.38660026e-01
1.60287249e+00 -5.36774993e-02 3.66927862e-01 2.17208549e-01
7.27866650e-01 3.22595417e-01 3.34930152e-01 3.15488547e-01
1.62204206e-01 6.72199965e-01 5.82224071e-01 2.27913380e-01
-4.01828200e-01 -7.52782106e-01 4.35278654e-01 6.28202975e-01
2.20487669e-01 -3.09388250e-01 -6.59881353e-01 7.75393188e-01
-1.99485385e+00 -8.56880009e-01 4.72911954e-01 2.42740464e+00
6.37209415e-01 8.79926905e-02 1.94624260e-01 3.13154876e-01
5.69142878e-01 4.56033438e-01 -5.58173656e-01 -2.78266430e-01
7.58324042e-02 3.73756975e-01 2.33715415e-01 4.09018427e-01
-1.02174747e+00 8.96573305e-01 6.25555277e+00 5.62635005e-01
-1.14125121e+00 2.94309705e-01 1.33417040e-01 -5.20355701e-01
-3.13975871e-01 2.22254097e-01 -5.72849095e-01 5.07995605e-01
5.79064846e-01 4.58149910e-02 1.00256836e+00 8.27246130e-01
-4.61777359e-01 -4.42777574e-02 -1.58943331e+00 8.68926227e-01
2.62039512e-01 -1.43417776e+00 -2.99261939e-02 4.32676375e-02
1.07289982e+00 7.51798078e-02 2.24943995e-01 3.60457450e-01
4.74088103e-01 -8.69150460e-01 1.05060947e+00 7.64061436e-02
7.31054068e-01 -3.89071703e-01 3.66756767e-01 4.16154414e-01
-1.26245594e+00 -3.28228563e-01 -2.24244237e-01 -2.86376089e-01
2.42355075e-02 2.11399391e-01 -9.03625786e-01 3.09294045e-01
2.82050371e-01 4.62249249e-01 -4.85708356e-01 1.02660644e+00
-4.31432396e-01 3.02249044e-01 -2.85604239e-01 -3.82461809e-02
3.48358423e-01 -6.34960160e-02 4.76015955e-01 8.51694345e-01
2.97397971e-01 -4.65233415e-01 3.86212260e-01 9.21693385e-01
-3.49712014e-01 -1.08276941e-01 -6.66796982e-01 -2.60790974e-01
2.11841747e-01 9.75413024e-01 -5.24818599e-01 -5.51653564e-01
-5.56153119e-01 1.11834037e+00 6.49450719e-01 5.22000074e-01
-6.39725924e-01 -1.82649150e-01 1.00384796e+00 8.37269798e-02
9.83757794e-01 -3.06131225e-02 -1.47549510e-01 -1.51861131e+00
2.85288543e-01 -8.31748068e-01 5.23139000e-01 -5.33298075e-01
-1.31394660e+00 3.95168692e-01 -2.25405380e-01 -1.25320518e+00
-2.46673092e-01 -7.13623464e-01 -7.00665772e-01 5.21831572e-01
-1.48528051e+00 -1.05868089e+00 -3.37539576e-02 3.76992941e-01
8.40670168e-01 -1.63336154e-02 8.42333376e-01 1.84155330e-01
-4.68070835e-01 4.84946340e-01 1.25184193e-01 -2.30734646e-02
7.25527942e-01 -1.31275904e+00 4.74917650e-01 9.49461460e-01
5.99937618e-01 5.44098198e-01 8.67455304e-01 -3.82946491e-01
-1.56945670e+00 -9.16121304e-01 6.86129987e-01 -4.35237259e-01
1.03179789e+00 -9.19099331e-01 -6.20666921e-01 7.06514835e-01
1.57335445e-01 6.25769734e-01 6.70501173e-01 1.14684902e-01
-1.01804209e+00 8.38886872e-02 -9.78158534e-01 8.13491404e-01
1.42838573e+00 -7.67268360e-01 -7.54925013e-01 3.26328546e-01
7.21150577e-01 -2.19999373e-01 -6.87118113e-01 2.73004800e-01
8.52217436e-01 -7.34803438e-01 1.05708659e+00 -7.58565247e-01
4.59764004e-01 -5.45929670e-01 -4.28334266e-01 -1.31113100e+00
-2.00356066e-01 -5.93400240e-01 -4.99082953e-01 8.57816696e-01
5.41286945e-01 -5.09377420e-01 9.03780162e-01 3.83353710e-01
-1.70714427e-02 -6.51992857e-01 -9.44052398e-01 -1.01947784e+00
4.78652725e-03 -1.61353111e-01 7.66972840e-01 6.51306987e-01
2.62877554e-01 3.69498044e-01 -1.57139644e-01 1.67190254e-01
6.11304283e-01 4.88206029e-01 6.85745537e-01 -1.04390478e+00
-5.29053926e-01 -5.20912588e-01 -5.63823640e-01 -8.60130966e-01
4.68494266e-01 -1.00376081e+00 1.10839911e-01 -1.32491434e+00
3.64105076e-01 -1.12468064e-01 -5.26429653e-01 3.94371003e-01
-2.75896758e-01 4.53873724e-01 5.37081599e-01 7.45840371e-02
-8.78278255e-01 5.25082409e-01 1.16638839e+00 -4.31405395e-01
6.72061965e-02 -4.18339401e-01 -3.83100271e-01 5.51075161e-01
4.74244744e-01 -3.51617008e-01 -6.73183918e-01 -4.51392174e-01
1.47193195e-02 -3.63947153e-01 4.12595659e-01 -1.05187988e+00
2.87068605e-01 -1.91274047e-01 2.74237007e-01 -2.83592671e-01
5.83668172e-01 -9.30115044e-01 2.20954105e-01 4.19797659e-01
-3.52883488e-01 -8.54450986e-02 -2.49597147e-01 7.66722023e-01
1.90771818e-02 -2.77059436e-01 7.00699627e-01 -1.50834307e-01
-7.58522511e-01 3.08831602e-01 -9.72298384e-02 4.18479033e-02
1.28328931e+00 -3.51952791e-01 -6.47345066e-01 -1.38739385e-02
-6.97772741e-01 8.25601593e-02 1.02444160e+00 4.61606115e-01
4.55653101e-01 -1.34436190e+00 -7.56190196e-02 4.78806287e-01
6.51638091e-01 -1.09721988e-01 -1.02027208e-02 6.18514359e-01
-3.51127714e-01 2.28970885e-01 -1.34074554e-01 -3.60620230e-01
-1.17629170e+00 1.21071970e+00 5.93052059e-03 -3.62754822e-01
-7.83231020e-01 8.90485764e-01 5.67758858e-01 -4.40053970e-01
5.45533597e-01 -3.08104038e-01 2.96457767e-01 -3.50430049e-02
5.53209245e-01 2.95090437e-01 6.60091862e-02 -4.53534454e-01
-1.88779891e-01 4.25960809e-01 -4.39958237e-02 -2.28281602e-01
1.29233074e+00 2.02119946e-01 1.65554717e-01 6.04903877e-01
1.53894126e+00 -4.90425080e-02 -1.61712372e+00 -5.43793976e-01
1.60760954e-01 -6.31222546e-01 -2.32259795e-01 -5.20772994e-01
-9.03728247e-01 1.08279169e+00 5.62738180e-01 3.81249376e-02
8.12046051e-01 2.72568762e-01 8.14065397e-01 4.32272464e-01
2.47833028e-01 -8.41919243e-01 3.47117335e-01 2.35063806e-01
6.02402866e-01 -1.17462146e+00 -8.55885521e-02 -2.64805198e-01
-6.04120910e-01 1.10755384e+00 5.36354601e-01 -2.96811312e-01
3.58777612e-01 2.91799694e-01 -8.29052106e-02 -1.56290486e-01
-1.08693683e+00 -5.02976835e-01 2.62679428e-01 4.74840015e-01
7.94421434e-02 -1.56760607e-02 2.48487279e-01 9.29877982e-02
-2.45501816e-01 -2.77827770e-01 1.82308272e-01 1.00144935e+00
-4.15802091e-01 -1.18582356e+00 -2.13978186e-01 2.43107602e-01
-1.96419835e-01 -5.36313765e-02 -3.98884773e-01 5.61546266e-01
2.31183127e-01 7.05945492e-01 2.10013568e-01 -3.42641026e-01
2.68500298e-01 5.56957185e-01 6.33387566e-01 -8.06127727e-01
-5.02914131e-01 -3.92769836e-02 9.72361341e-02 -4.89672422e-01
-9.61232558e-02 -5.86556077e-01 -1.03309381e+00 8.90216138e-03
-3.17019761e-01 -1.73957393e-01 5.94257116e-01 8.76271427e-01
9.49399471e-02 2.37159088e-01 4.49423552e-01 -7.87744045e-01
-8.26354802e-01 -7.55221605e-01 -3.17701072e-01 7.36000836e-01
7.00718224e-01 -8.28065217e-01 -4.33125466e-01 1.43571064e-01]
|
[10.282177925109863, 2.2178797721862793]
|
74cfaf50-4d19-449f-bc6d-7ffab3edbe41
|
variational-sequential-optimal-experimental
|
2306.10430
| null |
https://arxiv.org/abs/2306.10430v1
|
https://arxiv.org/pdf/2306.10430v1.pdf
|
Variational Sequential Optimal Experimental Design using Reinforcement Learning
|
We introduce variational sequential Optimal Experimental Design (vsOED), a new method for optimally designing a finite sequence of experiments under a Bayesian framework and with information-gain utilities. Specifically, we adopt a lower bound estimator for the expected utility through variational approximation to the Bayesian posteriors. The optimal design policy is solved numerically by simultaneously maximizing the variational lower bound and performing policy gradient updates. We demonstrate this general methodology for a range of OED problems targeting parameter inference, model discrimination, and goal-oriented prediction. These cases encompass explicit and implicit likelihoods, nuisance parameters, and physics-based partial differential equation models. Our vsOED results indicate substantially improved sample efficiency and reduced number of forward model simulations compared to previous sequential design algorithms.
|
['Xun Huan', 'Jiayuan Dong', 'Wanggang Shen']
|
2023-06-17
| null | null | null | null |
['experimental-design']
|
['methodology']
|
[-2.95208339e-02 7.70239830e-02 -4.72733349e-01 -2.01236531e-01
-8.33026767e-01 -4.33494627e-01 3.08637321e-01 -3.23023498e-01
-6.36183977e-01 1.19001269e+00 1.03470773e-01 -7.47510791e-01
-6.92996502e-01 -2.14543656e-01 -6.33340657e-01 -6.18820488e-01
-2.69799978e-01 2.67990530e-01 -2.22190559e-01 3.38174939e-01
5.55923223e-01 4.87721324e-01 -1.29225039e+00 -5.16071439e-01
7.60577500e-01 9.86045897e-01 4.18183118e-01 7.51186430e-01
5.55193365e-01 3.94268334e-01 -1.80049360e-01 -1.48947284e-01
2.02736735e-01 -4.01471585e-01 -3.29821020e-01 -1.26767188e-01
-1.07830778e-01 -8.31503868e-01 -3.33380997e-01 9.78307843e-01
7.94189513e-01 5.99079847e-01 1.21367776e+00 -1.35724103e+00
-4.19077545e-01 1.40631169e-01 -5.27146876e-01 2.49012351e-01
2.34377429e-01 6.39658988e-01 1.27007449e+00 -4.66450989e-01
5.21998405e-01 1.50076926e+00 3.86999965e-01 5.42669713e-01
-1.89533770e+00 -3.70584279e-01 2.54874974e-01 -7.18809888e-02
-1.43996739e+00 -4.15663958e-01 4.86148447e-01 -5.99643528e-01
5.53172827e-01 8.94298330e-02 6.64187968e-01 1.21992683e+00
7.91384757e-01 7.22696662e-01 9.69872594e-01 -2.87500560e-01
9.18502927e-01 1.50892898e-01 3.78748700e-02 6.43626153e-01
4.97873873e-01 9.01899517e-01 -2.49418050e-01 -5.97310960e-01
9.91881490e-01 -2.64655143e-01 -9.46872532e-02 -7.52086520e-01
-8.20090473e-01 9.17615473e-01 -3.80461931e-01 -6.13793552e-01
-8.75233352e-01 5.52182257e-01 -2.77078953e-02 2.01996699e-01
1.83335796e-01 6.29373014e-01 -6.67470038e-01 -5.87416580e-03
-6.36640728e-01 1.04345405e+00 9.84213948e-01 8.73081446e-01
1.89987972e-01 8.17591697e-02 -7.86050439e-01 4.40838635e-01
1.01385200e+00 9.24803376e-01 -2.06771985e-01 -1.84770167e+00
1.61785886e-01 -4.01685178e-01 1.18366385e+00 -4.81482893e-01
-2.12110758e-01 -4.03292447e-01 -3.27769369e-01 3.39317739e-01
4.59998727e-01 -7.33559787e-01 -7.64185846e-01 2.09883332e+00
5.29848397e-01 3.73574309e-02 -7.69681111e-02 8.26921582e-01
-3.62914838e-02 6.34415090e-01 3.38054240e-01 -7.99401581e-01
1.07351267e+00 -1.24612667e-01 -8.43660057e-01 -1.11239530e-01
2.97528356e-01 -4.67162371e-01 8.90993953e-01 3.00999254e-01
-1.18622971e+00 -1.65643021e-02 -8.07242930e-01 3.36100638e-01
2.88823873e-01 -6.18211217e-02 4.68837202e-01 9.04801607e-01
-9.88107443e-01 9.65706050e-01 -6.03952646e-01 -4.58355621e-02
4.73161310e-01 3.76245260e-01 5.82760036e-01 1.35505721e-01
-8.69491696e-01 6.75308526e-01 9.68542397e-02 -6.22292385e-02
-1.49038160e+00 -1.20891166e+00 -6.01707935e-01 3.14561039e-01
4.69641894e-01 -1.05143750e+00 1.78500998e+00 1.10664874e-01
-2.01391006e+00 1.17128357e-01 -2.19379246e-01 -4.46144164e-01
7.21319914e-01 -1.79880455e-01 1.34730890e-01 -9.70790163e-02
3.56707796e-02 5.23641169e-01 7.62948275e-01 -1.06449497e+00
-3.28438014e-01 -4.67332788e-02 -1.81243360e-01 3.32776040e-01
1.41058505e-01 -4.66713049e-02 1.82289749e-01 -2.97462910e-01
-1.71199828e-01 -9.40678060e-01 -7.81696677e-01 3.58438253e-01
-3.59776974e-01 -8.49869028e-02 6.15183830e-01 -7.76617110e-01
1.27418423e+00 -1.73306537e+00 7.74822831e-02 3.10473144e-01
-8.33615512e-02 -4.66854066e-01 -3.91778871e-02 1.78408533e-01
4.46708024e-01 1.81403413e-01 -3.81851457e-02 -5.12745865e-02
5.80517888e-01 -3.52189280e-02 -2.52938390e-01 4.73666579e-01
-3.36919963e-01 6.11979544e-01 -8.36024880e-01 -5.23403108e-01
2.94313550e-01 -1.26198744e-02 -1.20101690e+00 4.53879744e-01
-6.19544029e-01 5.14548600e-01 -1.06469238e+00 2.24465355e-01
5.80131590e-01 -2.41647229e-01 5.36052585e-01 3.38419303e-02
-3.97970945e-01 -7.66793313e-03 -1.42795825e+00 1.19800127e+00
-3.74351114e-01 2.33574778e-01 3.91199887e-01 -9.21343267e-01
3.56114626e-01 2.78326958e-01 7.00072765e-01 -2.92847842e-01
4.17430758e-01 2.87258364e-02 -1.30275950e-01 -6.41133666e-01
3.08180332e-01 -3.65942657e-01 -1.80865124e-01 5.73097885e-01
9.66842845e-02 -3.75215709e-01 -7.57877855e-03 8.50129724e-02
7.02379763e-01 3.70127231e-01 5.49170613e-01 -8.24928641e-01
-2.38353148e-01 -3.19683611e-01 5.42187035e-01 1.32851636e+00
-5.83421469e-01 -1.30323665e-02 7.93023109e-01 1.94318816e-01
-1.32435966e+00 -1.23868215e+00 -2.70614505e-01 7.31579900e-01
1.54712200e-02 4.94240485e-02 -6.39683425e-01 -1.72013342e-01
4.73420113e-01 1.31747115e+00 -4.68148351e-01 -2.21565098e-01
-6.93966821e-02 -8.82041097e-01 -2.33795062e-01 1.98552772e-01
2.24128366e-01 -3.89928162e-01 -5.26277542e-01 2.77812958e-01
1.79325864e-01 -7.22699940e-01 -6.68876171e-01 -2.05436926e-02
-9.66802657e-01 -8.61715972e-01 -6.75178707e-01 -5.98146543e-02
2.41684914e-01 -1.31540865e-01 7.21748292e-01 -7.33475208e-01
-2.67778307e-01 7.29423881e-01 5.48724711e-01 -4.40725148e-01
-3.21130633e-01 -5.30713677e-01 4.73463744e-01 -1.57742083e-01
-1.22405402e-01 -4.46105212e-01 -9.05341625e-01 3.97870392e-01
-1.97123796e-01 -9.86895934e-02 3.41518223e-01 8.41289222e-01
6.23785794e-01 -1.67710334e-01 5.09869814e-01 -3.65450025e-01
9.38810945e-01 -5.46552837e-01 -1.34935594e+00 2.56645799e-01
-8.14508080e-01 5.12831450e-01 1.04194678e-01 -4.65617716e-01
-1.58586884e+00 -3.29356223e-01 2.02440470e-01 -3.96790355e-01
5.63686900e-02 3.55497718e-01 -2.22649232e-01 -6.62897108e-03
4.61181849e-01 -1.94875300e-01 1.78416476e-01 -5.24322271e-01
2.98190475e-01 5.83954871e-01 -5.98799549e-02 -1.23692358e+00
2.47547850e-01 -8.13359469e-02 3.59927297e-01 -8.51487637e-01
-8.14695776e-01 1.43195316e-01 -4.73054610e-02 -3.56307298e-01
7.61099815e-01 -7.13960886e-01 -1.55599022e+00 9.20434818e-02
-8.62127662e-01 -7.85170674e-01 -3.47595513e-01 1.05595410e+00
-1.06702089e+00 1.22314706e-01 -4.60051090e-01 -1.57667470e+00
6.41314387e-02 -1.06836402e+00 8.32811892e-01 4.35309708e-01
-3.52068275e-01 -1.04461110e+00 2.11996138e-01 -9.29232035e-03
1.81392461e-01 1.93699270e-01 9.93323922e-01 -1.39570823e-02
-8.27979624e-01 1.03860134e-02 8.08872655e-02 -7.21196905e-02
-4.85020429e-01 1.00425743e-01 -5.84338903e-01 -3.72480065e-01
-3.56245637e-02 -1.67373598e-01 3.73703569e-01 1.51780415e+00
1.26854324e+00 -4.70615149e-01 -5.57222784e-01 3.88523847e-01
1.31904531e+00 7.10738838e-01 6.63462877e-02 -2.76519626e-01
7.13018328e-02 3.72385591e-01 6.49118781e-01 1.17631638e+00
-6.90330192e-02 5.82008362e-01 1.21678028e-03 6.85242832e-01
5.89358211e-01 -6.46611810e-01 2.87507743e-01 7.76449293e-02
1.31104723e-01 -5.37213564e-01 -3.98596644e-01 3.24481249e-01
-1.85235453e+00 -9.10832465e-01 2.15318561e-01 2.52672410e+00
5.54341674e-01 5.89030124e-02 1.87341154e-01 -6.16397858e-01
8.50614250e-01 -1.11587070e-01 -1.22361016e+00 -2.96537399e-01
4.25559342e-01 -2.89570261e-02 9.36916828e-01 6.36710584e-01
-7.13943243e-01 4.05162543e-01 8.46133041e+00 9.24692810e-01
-8.92558098e-02 7.88879618e-02 7.71290481e-01 -4.78617489e-01
-5.00716746e-01 2.36752033e-01 -1.07569349e+00 5.02487600e-01
1.18646646e+00 -7.24677801e-01 7.05268085e-01 7.78381050e-01
1.05742133e+00 -4.99397516e-01 -1.15762889e+00 6.47600353e-01
-9.32225525e-01 -1.30413640e+00 -4.09791797e-01 4.69446957e-01
8.19968104e-01 -3.51198912e-01 2.65934020e-01 4.34203595e-01
1.05232537e+00 -5.98977923e-01 8.79711509e-01 7.66142786e-01
7.01096714e-01 -5.03061712e-01 1.94186792e-01 3.51538301e-01
-7.15691626e-01 -4.59379286e-01 -3.36856097e-01 -9.41863731e-02
5.18477619e-01 5.54869592e-01 -4.57754076e-01 -9.84467044e-02
4.57906067e-01 2.44070143e-01 5.53881764e-01 1.25262654e+00
2.18457188e-02 6.33906305e-01 -7.38571227e-01 -5.89656651e-01
8.58924165e-02 -5.93698978e-01 9.45689797e-01 6.64811313e-01
5.26115656e-01 3.64300966e-01 1.39107242e-01 1.39781821e+00
1.85757294e-01 -1.95888147e-01 -4.07855213e-01 -2.97625810e-01
7.93301105e-01 5.86296260e-01 -4.65202749e-01 -9.97566991e-03
2.36331988e-02 3.63521039e-01 1.89048219e-02 9.16781008e-01
-9.33475912e-01 -1.75125867e-01 1.10784781e+00 -2.08672136e-01
3.27916533e-01 -2.43564129e-01 -3.12831104e-01 -8.68948042e-01
-4.90968198e-01 -3.54980320e-01 4.82618332e-01 -7.43323207e-01
-9.71556425e-01 -6.82047367e-01 7.72968113e-01 -6.68940067e-01
-4.16097641e-01 -8.51788044e-01 -5.02332866e-01 9.70245779e-01
-8.18507671e-01 -2.12006897e-01 6.36079192e-01 1.16944216e-01
3.07521939e-01 2.14282479e-02 4.24300432e-01 -1.01526357e-01
-1.00398040e+00 4.31143552e-01 7.81347632e-01 -6.28244758e-01
1.72266468e-01 -1.08970165e+00 -7.19377697e-02 5.08734584e-01
-8.49179924e-01 4.93856937e-01 1.48440933e+00 -8.01140666e-01
-1.62615120e+00 -5.57011068e-01 1.75343871e-01 -9.23061222e-02
8.10849667e-01 -1.63493127e-01 -3.23879272e-01 6.53960049e-01
-5.69083057e-02 -3.83060753e-01 3.56395453e-01 1.68508813e-01
4.11724150e-01 2.67833173e-01 -1.45040190e+00 1.08388603e+00
1.15381348e+00 -1.30836844e-01 -1.16060346e-01 3.36063147e-01
8.23420465e-01 -3.04927498e-01 -1.03639984e+00 2.49971911e-01
8.92583191e-01 -3.15896094e-01 9.92261648e-01 -9.74640906e-01
7.81950131e-02 2.15715572e-01 -3.97917598e-01 -1.41704774e+00
-4.21658486e-01 -1.41043603e+00 -2.96876878e-01 7.55483031e-01
4.30810839e-01 -5.68615854e-01 6.97650909e-01 1.15986848e+00
1.52783230e-01 -6.47436380e-01 -8.76246154e-01 -8.93075109e-01
1.31320670e-01 -5.50100327e-01 3.66878211e-01 2.17363149e-01
-4.18479815e-02 1.53874695e-01 -6.74333870e-01 1.43121272e-01
1.48077023e+00 -1.95030794e-02 3.92699003e-01 -1.01103771e+00
-6.12974942e-01 -5.29533565e-01 2.69849092e-01 -1.43240440e+00
1.54817998e-01 -2.72458464e-01 2.21282661e-01 -1.23703063e+00
3.69552165e-01 -1.22240461e-01 -2.71963686e-01 -3.39921057e-01
-1.68507367e-01 -8.59212399e-01 -1.83942094e-01 -3.60427350e-01
-4.94501442e-01 1.03358543e+00 1.41587663e+00 2.39261344e-01
-2.96847910e-01 3.45284075e-01 -5.44538021e-01 3.48121196e-01
6.72366202e-01 -6.87606573e-01 -7.73941517e-01 1.84772998e-01
4.57537398e-02 8.35343838e-01 5.37217975e-01 -3.70617270e-01
-9.07446072e-02 -9.59095418e-01 3.42729926e-01 -4.72526699e-01
1.14728659e-01 -3.64321619e-01 3.53525192e-01 6.51544452e-01
-7.98425615e-01 -4.87100154e-01 2.48936236e-01 1.16067541e+00
7.47987390e-01 -3.84005308e-01 1.07730794e+00 -3.44802551e-02
-5.05737364e-01 3.72327477e-01 -1.06789780e+00 3.20318073e-01
8.38159621e-01 1.29505873e-01 9.41989943e-02 -6.26455069e-01
-1.02251995e+00 5.17247558e-01 1.63722653e-02 -2.83502460e-01
4.93913144e-01 -1.26177287e+00 -4.07078862e-01 -7.58386776e-02
-4.17699754e-01 -6.27745330e-01 6.53642893e-01 6.37153447e-01
-6.35661036e-02 4.43791062e-01 7.81139359e-02 -3.35954875e-01
-6.62140369e-01 4.76510465e-01 7.15627491e-01 -2.90397346e-01
-1.26537442e-01 6.78640783e-01 1.94854915e-01 -2.54678786e-01
1.01200379e-01 -1.40383840e-01 2.22212330e-01 -2.91867852e-01
2.27853149e-01 7.87541151e-01 -7.28412330e-01 3.05587918e-01
1.06393462e-02 1.54442713e-01 9.97632146e-02 -7.52428949e-01
1.08487999e+00 -5.37983179e-01 4.46836114e-01 2.94998050e-01
1.06903553e+00 -6.35088742e-01 -2.12222600e+00 -1.24427818e-01
-7.36377761e-02 -4.74803090e-01 8.34809422e-01 -5.92658997e-01
-4.90872204e-01 4.41897273e-01 7.38319993e-01 -2.23757606e-02
3.71920198e-01 -3.43685716e-01 2.75449783e-01 5.74781001e-01
3.79619598e-01 -1.39136314e+00 -9.33334902e-02 1.30268320e-01
6.14576161e-01 -9.77745533e-01 2.68557221e-01 -5.96638359e-02
-4.51482505e-01 4.85681206e-01 5.10827780e-01 -4.30019945e-03
1.20861149e+00 3.17087889e-01 -7.45021462e-01 1.60903647e-03
-9.58361089e-01 8.81146193e-02 1.45491272e-01 5.53432226e-01
1.30494118e-01 1.66882813e-01 -5.36383450e-01 5.39823294e-01
3.16121876e-01 4.68133777e-01 2.56429702e-01 1.03760922e+00
-6.95295215e-01 -6.18344665e-01 -3.12445045e-01 7.14128137e-01
-2.65873551e-01 9.12371501e-02 3.69043499e-01 7.30079174e-01
-7.99896061e-01 9.60705161e-01 1.09108202e-01 1.34023651e-01
3.27526987e-01 1.36890531e-01 7.73128808e-01 -1.12064488e-01
3.41828257e-01 3.36993307e-01 1.90163121e-01 -6.88664138e-01
9.56728086e-02 -1.03699028e+00 -5.69135547e-01 -6.07824147e-01
-2.63928890e-01 2.88195729e-01 4.30090338e-01 9.71653223e-01
5.08095264e-01 4.48009670e-01 7.58353353e-01 -9.46359992e-01
-1.43598413e+00 -7.17009783e-01 -9.03289795e-01 -2.75104791e-01
1.61689758e-01 -1.05617821e+00 -5.27385831e-01 -4.85366225e-01]
|
[6.480053901672363, 3.9418070316314697]
|
848bffdd-f320-486a-96b1-b01eb6480047
|
perceptual-image-enhancement-for-smartphone
|
2210.13552
| null |
https://arxiv.org/abs/2210.13552v1
|
https://arxiv.org/pdf/2210.13552v1.pdf
|
Perceptual Image Enhancement for Smartphone Real-Time Applications
|
Recent advances in camera designs and imaging pipelines allow us to capture high-quality images using smartphones. However, due to the small size and lens limitations of the smartphone cameras, we commonly find artifacts or degradation in the processed images. The most common unpleasant effects are noise artifacts, diffraction artifacts, blur, and HDR overexposure. Deep learning methods for image restoration can successfully remove these artifacts. However, most approaches are not suitable for real-time applications on mobile devices due to their heavy computation and memory requirements. In this paper, we propose LPIENet, a lightweight network for perceptual image enhancement, with the focus on deploying it on smartphones. Our experiments show that, with much fewer parameters and operations, our model can deal with the mentioned artifacts and achieve competitive performance compared with state-of-the-art methods on standard benchmarks. Moreover, to prove the efficiency and reliability of our approach, we deployed the model directly on commercial smartphones and evaluated its performance. Our model can process 2K resolution images under 1 second in mid-level commercial smartphones.
|
['Radu Timofte', 'Javier Vazquez-Corral', 'Florin Vasluianu', 'Marcos V. Conde']
|
2022-10-24
| null | null | null | null |
['hdr-reconstruction']
|
['computer-vision']
|
[ 2.85917580e-01 -5.92524290e-01 2.79771924e-01 9.93952975e-02
-4.73788053e-01 -1.86540455e-01 1.05514325e-01 -2.17219457e-01
-4.22547370e-01 5.03030598e-01 -8.56836811e-02 -3.33359182e-01
1.35883927e-01 -5.06849289e-01 -8.02157760e-01 -5.22620261e-01
9.09230337e-02 -5.20357311e-01 3.18711609e-01 1.85695030e-02
2.27542341e-01 3.36786777e-01 -1.65918744e+00 2.78657913e-01
1.02844536e+00 1.12778115e+00 4.87977386e-01 8.30232620e-01
3.38579565e-01 9.33584809e-01 -7.26768851e-01 -4.85091358e-01
2.93906510e-01 6.04260229e-02 -3.39706033e-01 3.97014350e-01
5.89056134e-01 -9.90267873e-01 -7.05225408e-01 1.24907553e+00
6.24445677e-01 -1.38200358e-01 1.77554749e-02 -1.20511246e+00
-7.32414722e-01 7.40012676e-02 -9.21795249e-01 2.68153757e-01
3.74500841e-01 3.90873402e-01 3.20808381e-01 -9.45686936e-01
1.15065090e-01 1.02454698e+00 8.52473855e-01 2.89011538e-01
-1.16293299e+00 -7.31860995e-01 -1.64669156e-01 6.05256259e-01
-1.18621933e+00 -8.97355318e-01 4.86816406e-01 5.82769811e-02
9.10505176e-01 2.42398620e-01 5.72821200e-01 1.16000545e+00
3.39859635e-01 5.89278340e-01 1.30877352e+00 -2.68657297e-01
1.43535003e-01 3.17898095e-02 -7.53727928e-02 4.42315936e-01
3.12252730e-01 -1.47826806e-01 -5.48321545e-01 1.45465985e-01
8.73994589e-01 2.88536668e-01 -6.46444499e-01 1.15080953e-01
-1.04170322e+00 2.02845156e-01 2.89776504e-01 7.38895908e-02
-4.22635496e-01 3.68962824e-01 2.32599333e-01 1.37813210e-01
4.24771637e-01 -2.44041961e-02 -2.87700891e-01 -2.78175473e-01
-1.08508313e+00 -2.94978440e-01 4.19000298e-01 8.88490081e-01
5.60901165e-01 -5.70030659e-02 -1.94603559e-02 8.85511696e-01
9.79359522e-02 6.42905176e-01 4.27086413e-01 -1.23719513e+00
4.52955335e-01 9.90816429e-02 3.38290721e-01 -1.27912188e+00
-1.99807271e-01 -2.60084778e-01 -1.29846299e+00 2.75970846e-01
2.06272662e-01 -1.05005689e-01 -6.16594017e-01 1.20018923e+00
-1.98648535e-02 6.34556890e-01 -8.00320506e-02 1.01514316e+00
7.77869225e-01 7.79862702e-01 -2.32407004e-01 -4.87530559e-01
1.38832414e+00 -1.21936369e+00 -8.58922064e-01 -2.48670444e-01
2.09857058e-03 -1.11759472e+00 1.51800537e+00 9.58169341e-01
-1.30591547e+00 -8.01370859e-01 -1.19773412e+00 -2.55034208e-01
9.30353031e-02 4.63199109e-01 5.88165939e-01 8.35575938e-01
-1.29919028e+00 6.50842249e-01 -8.85223389e-01 -2.20611855e-01
5.49548864e-01 2.83882171e-01 -2.35838026e-01 -3.31746608e-01
-7.99049079e-01 4.41834539e-01 -1.93611041e-01 2.29062080e-01
-7.90169716e-01 -6.58166111e-01 -4.76141900e-01 3.03089648e-01
4.90022987e-01 -5.11332035e-01 1.18961608e+00 -8.59639883e-01
-1.86349678e+00 6.04827046e-01 -1.90529674e-01 -4.42808956e-01
5.00932157e-01 -6.15412474e-01 -5.10317683e-01 3.26877743e-01
-3.38431478e-01 2.62152165e-01 1.27385950e+00 -1.11878681e+00
-4.82558429e-01 -1.29576415e-01 2.04932958e-01 1.18267283e-01
-8.58496547e-01 1.60138234e-01 -1.00868452e+00 -4.62087005e-01
-4.62493710e-02 -8.54993701e-01 -1.49750352e-01 1.54027805e-01
-3.20007682e-01 3.24916333e-01 1.09292996e+00 -8.68487358e-01
1.08534884e+00 -2.48540759e+00 -3.85285467e-01 -2.96592504e-01
3.73882532e-01 7.52001464e-01 -7.80506581e-02 3.96480039e-02
2.18567938e-01 5.37004769e-02 6.25227112e-04 -6.80642664e-01
-2.90440887e-01 2.02541915e-03 -3.71869504e-01 5.11226654e-01
-2.32298762e-01 5.84204912e-01 -6.32834494e-01 -3.54249030e-01
5.17814755e-01 8.96504283e-01 -4.16586965e-01 3.50482911e-01
2.04767287e-01 2.85366774e-01 9.46979970e-02 4.65071738e-01
1.20815611e+00 -4.82647657e-01 1.29279792e-01 -6.37283981e-01
-3.13114263e-02 1.66910097e-01 -1.23816001e+00 1.58459949e+00
-9.18305278e-01 9.21522677e-01 4.45668072e-01 -6.11360371e-01
4.65256304e-01 7.91739225e-02 7.93157965e-02 -9.22786236e-01
-2.00400576e-02 2.24295393e-01 -3.68188202e-01 -6.97871029e-01
7.05663621e-01 4.41549152e-01 4.40960914e-01 2.89560765e-01
-4.24915880e-01 1.09406687e-01 1.04850709e-01 1.61822677e-01
1.24616790e+00 -1.71998248e-01 5.56621365e-02 1.50194213e-01
4.62635189e-01 -6.75294340e-01 3.48324686e-01 7.39941895e-01
-1.54175043e-01 7.64529586e-01 2.74784058e-01 -3.95603299e-01
-9.84029531e-01 -1.05742562e+00 1.32321715e-01 9.03927624e-01
5.94562948e-01 -5.81749201e-01 -8.32592845e-01 -1.72302634e-01
-5.49405515e-01 2.14374140e-01 -1.15728760e-02 3.28080356e-03
-5.51809907e-01 -9.69206274e-01 4.24316674e-01 3.46363097e-01
1.03912783e+00 -9.26084042e-01 -9.15748060e-01 1.29452243e-01
-3.27590346e-01 -1.65581894e+00 -6.07921004e-01 -4.46296811e-01
-8.40865016e-01 -1.06108677e+00 -7.84858406e-01 -6.24177396e-01
6.42788351e-01 1.00987923e+00 9.66579080e-01 3.46031666e-01
-2.08412692e-01 2.43987799e-01 -1.35409683e-01 9.66840237e-02
-2.40226358e-01 -3.14343780e-01 7.97479376e-02 2.09181830e-01
-2.02126607e-01 -7.22868264e-01 -1.10250330e+00 3.49377185e-01
-1.16093421e+00 1.83695078e-01 5.96722305e-01 6.73375368e-01
4.80460495e-01 7.79705167e-01 1.06284924e-01 -5.24339736e-01
6.45999789e-01 -2.49218475e-02 -8.50416124e-01 4.89383638e-02
-5.05245447e-01 -4.12727594e-01 9.39897954e-01 -5.94944477e-01
-1.30616176e+00 -2.08989158e-01 -4.02838178e-02 -4.39332783e-01
-1.42689914e-01 1.79765537e-01 -2.95075566e-01 -2.80603081e-01
2.93728888e-01 2.87052184e-01 -9.45137069e-02 -6.14658594e-01
7.01568574e-02 9.79531407e-01 8.44033241e-01 -8.88240784e-02
6.20843351e-01 7.05533326e-01 -4.39792946e-02 -1.17749178e+00
-6.12447023e-01 -1.05400637e-01 -6.76357746e-03 -1.18601099e-01
5.08557916e-01 -1.32164156e+00 -1.07035434e+00 1.05104268e+00
-1.23620462e+00 -3.73882830e-01 2.19836965e-01 3.13155442e-01
-2.82444477e-01 8.70342255e-01 -1.07684720e+00 -5.79327464e-01
-5.72981119e-01 -1.42814410e+00 9.33344305e-01 4.76986110e-01
4.33095366e-01 -5.11983216e-01 -3.47736716e-01 5.49346089e-01
7.86867380e-01 -2.03546345e-01 4.20108199e-01 3.69998336e-01
-8.04455638e-01 3.36315483e-02 -6.94910765e-01 6.49747550e-01
1.96168706e-01 -1.52354622e-02 -1.09470701e+00 -5.63861907e-01
2.39561886e-01 -1.40040845e-01 8.03673029e-01 4.68176991e-01
1.83618248e+00 -3.29783887e-01 -1.40750825e-01 1.00982070e+00
1.53910100e+00 8.13973844e-02 1.23438656e+00 3.47423732e-01
6.44609571e-01 3.14453654e-02 5.03802001e-01 4.88752812e-01
2.00858787e-01 9.03223336e-01 5.99540830e-01 -4.30748731e-01
-3.53793114e-01 1.02207273e-01 6.64575815e-01 8.19084466e-01
-6.12365045e-02 -4.20875221e-01 -5.26249826e-01 3.05874646e-01
-1.59620821e+00 -7.17904270e-01 -4.07695323e-01 2.30194473e+00
7.39523828e-01 1.62910521e-01 -2.13286310e-01 2.74633646e-01
6.77515030e-01 2.06934258e-01 -4.78488952e-01 -9.05422047e-02
-2.63946384e-01 1.44895986e-01 6.11860871e-01 1.89223379e-01
-1.11012197e+00 6.43413663e-01 6.14010477e+00 9.63352442e-01
-1.25875878e+00 2.47981504e-01 9.13293660e-01 -3.37580234e-01
1.74304217e-01 -3.62141132e-01 -5.09209991e-01 8.29568923e-01
9.73006904e-01 1.97077900e-01 8.56751204e-01 5.82806706e-01
5.62715054e-01 -2.72063524e-01 -7.60975182e-01 1.59705734e+00
1.31591573e-01 -1.23375344e+00 -3.85363936e-01 1.34159073e-01
6.81380451e-01 -2.78788488e-02 4.50624436e-01 -9.34373513e-02
-2.00195938e-01 -1.01215410e+00 4.36803401e-01 3.79869163e-01
1.00962472e+00 -7.22672224e-01 8.40394676e-01 9.03795436e-02
-8.43374193e-01 -7.16524720e-02 -5.34851015e-01 -1.98729500e-01
1.43338546e-01 1.01873636e+00 -3.41031998e-01 1.67379618e-01
1.20641482e+00 7.25301743e-01 -7.79121935e-01 1.12255681e+00
-2.71556765e-01 5.29795945e-01 -3.13948780e-01 3.90041947e-01
-1.60063565e-01 -1.34736404e-01 2.77413309e-01 1.16409302e+00
7.53341556e-01 -2.35728808e-02 -2.03570113e-01 5.09572685e-01
-3.80143672e-01 -1.63883403e-01 -2.77405769e-01 2.50640482e-01
3.84917796e-01 1.41782296e+00 -6.64702475e-01 -3.08674514e-01
-6.53827548e-01 1.52009404e+00 -1.15778469e-01 3.88530076e-01
-1.15452909e+00 -4.33806479e-01 8.65055978e-01 2.19667554e-01
3.11759204e-01 -2.59955913e-01 -1.31874427e-01 -1.45728290e+00
4.85284269e-01 -1.21123016e+00 -2.18528718e-01 -1.06524587e+00
-1.04553497e+00 5.60427368e-01 -5.07609546e-01 -1.11227167e+00
4.58934903e-01 -6.40973628e-01 -5.32650054e-01 4.76238102e-01
-1.68277884e+00 -7.35993207e-01 -8.29608679e-01 7.85302997e-01
5.74892223e-01 4.32333574e-02 4.81364101e-01 7.28964210e-01
-8.02057683e-01 4.79100138e-01 4.40399319e-01 -1.00786760e-01
8.97983193e-01 -9.36852038e-01 5.93408823e-01 1.23146331e+00
-3.27715501e-02 5.01039743e-01 6.62880659e-01 -2.74987370e-01
-1.65705550e+00 -9.44893003e-01 3.75361741e-01 1.66641623e-01
4.71177101e-01 -3.55938345e-01 -9.41112459e-01 3.16090405e-01
5.44987321e-01 3.57644349e-01 4.16975409e-01 -2.97365636e-01
-2.18326196e-01 -4.58422244e-01 -9.88241553e-01 8.00904691e-01
9.50465679e-01 -4.46057677e-01 4.47890647e-02 3.51289570e-01
8.27621341e-01 -5.35644948e-01 -6.38393342e-01 1.31825611e-01
4.52852070e-01 -1.57370293e+00 1.18883967e+00 1.99774578e-01
5.28408229e-01 -5.23454368e-01 -2.86555886e-02 -1.17697549e+00
-1.19808950e-01 -1.00487185e+00 -3.80463898e-01 1.22933936e+00
-1.11360438e-01 -4.63939577e-01 5.81135750e-01 4.32510227e-01
3.27777602e-02 -4.42627132e-01 -7.53321826e-01 -6.47442460e-01
-6.29017293e-01 -5.57057858e-01 7.07900465e-01 4.42063808e-01
-3.91034782e-01 1.56251371e-01 -9.83085752e-01 4.09319013e-01
8.76599193e-01 5.74845076e-02 8.72131228e-01 -7.79948771e-01
-6.19446695e-01 -3.50025780e-02 -2.09724516e-01 -1.37461638e+00
-1.74533904e-01 1.01789832e-01 -6.85293227e-02 -1.28777850e+00
3.53022367e-01 -1.31125242e-01 -2.21289977e-01 1.95906550e-01
-2.45103046e-01 8.38231266e-01 4.57515150e-01 1.79805592e-01
-9.01337564e-01 3.96761686e-01 9.78618860e-01 -6.81986883e-02
-2.41767243e-01 -1.18413396e-01 -7.30257511e-01 9.08498645e-01
7.10456073e-01 -9.31167081e-02 -3.23063046e-01 -8.40644717e-01
4.01766270e-01 2.57778391e-02 6.64009273e-01 -1.28778136e+00
3.26859534e-01 1.56712398e-01 4.47948724e-01 -3.43553036e-01
5.10597944e-01 -8.74660552e-01 2.53712058e-01 4.28955913e-01
2.00855192e-02 -7.56293908e-03 3.00923944e-01 5.18655658e-01
-1.93658635e-01 8.50768238e-02 9.55559552e-01 1.24577075e-01
-6.07437789e-01 1.54378638e-01 -4.26958025e-01 -2.65189171e-01
7.96530366e-01 -8.34009200e-02 -6.22303367e-01 -6.63067400e-01
-4.11907524e-01 -2.23177463e-01 7.69928634e-01 1.54742151e-01
9.82147634e-01 -1.05100822e+00 -4.44415092e-01 3.28374982e-01
-3.32518071e-01 -2.01554418e-01 6.82390094e-01 9.76256847e-01
-8.08307409e-01 1.51877448e-01 -2.53086329e-01 -5.61406255e-01
-1.61646068e+00 7.14040875e-01 1.85602054e-01 -2.38368973e-01
-8.16957057e-01 5.46048284e-01 2.19783902e-01 2.72011787e-01
3.28083903e-01 -3.12174886e-01 3.23178284e-02 -3.96541506e-01
9.90828514e-01 7.10122108e-01 2.50022829e-01 -3.59766841e-01
-2.75268137e-01 5.45306027e-01 -5.27905934e-02 2.77220905e-01
1.20230567e+00 -6.33850396e-01 -1.51978493e-01 -1.17678843e-01
1.06134760e+00 2.06582084e-01 -1.54637277e+00 -1.50815278e-01
-5.23838639e-01 -8.19664478e-01 3.50519180e-01 -5.36095381e-01
-1.48235500e+00 8.63167405e-01 9.84631121e-01 2.60696173e-01
1.80660605e+00 -5.43090820e-01 1.30912924e+00 2.68990219e-01
3.79774302e-01 -9.48526442e-01 2.41285965e-01 1.29148185e-01
5.89021325e-01 -1.26674986e+00 1.86590284e-01 -5.62633634e-01
-3.23151469e-01 9.82475877e-01 3.83190244e-01 -7.71904225e-03
4.02472854e-01 4.43295628e-01 6.06234632e-02 3.05059969e-01
-6.07835472e-01 1.88877940e-01 -7.79253468e-02 6.83163166e-01
2.72885710e-01 -1.62780225e-01 -3.39410305e-02 2.76982248e-01
3.98907848e-02 2.29859024e-01 1.06762481e+00 4.20089066e-01
-1.08063757e-01 -7.97118366e-01 -6.74654961e-01 2.07622394e-01
-8.47705483e-01 -3.70139837e-01 2.26562843e-01 4.02724117e-01
2.26046458e-01 1.58417881e+00 4.83645312e-02 -3.25144351e-01
2.58025765e-01 -7.61191547e-01 4.44942147e-01 -4.34532091e-02
-3.07111144e-01 1.52963683e-01 -2.30849102e-01 -1.07892537e+00
-4.93496478e-01 -3.40075850e-01 -7.53732741e-01 -6.83881938e-01
-1.38547778e-01 -3.65762502e-01 6.90178931e-01 6.53721750e-01
6.76878333e-01 5.86790085e-01 6.67066753e-01 -1.02657044e+00
-4.11367625e-01 -8.39698195e-01 -5.65844715e-01 4.49743956e-01
5.03715694e-01 -2.99041629e-01 -4.64268416e-01 2.72652209e-01]
|
[10.933300971984863, -2.168522596359253]
|
13f2bb10-1ec7-40c7-84a0-86e75922a907
|
don-t-stop-pretraining-adapt-language-models
|
2004.10964
| null |
https://arxiv.org/abs/2004.10964v3
|
https://arxiv.org/pdf/2004.10964v3.pdf
|
Don't Stop Pretraining: Adapt Language Models to Domains and Tasks
|
Language models pretrained on text from a wide variety of sources form the foundation of today's NLP. In light of the success of these broad-coverage models, we investigate whether it is still helpful to tailor a pretrained model to the domain of a target task. We present a study across four domains (biomedical and computer science publications, news, and reviews) and eight classification tasks, showing that a second phase of pretraining in-domain (domain-adaptive pretraining) leads to performance gains, under both high- and low-resource settings. Moreover, adapting to the task's unlabeled data (task-adaptive pretraining) improves performance even after domain-adaptive pretraining. Finally, we show that adapting to a task corpus augmented using simple data selection strategies is an effective alternative, especially when resources for domain-adaptive pretraining might be unavailable. Overall, we consistently find that multi-phase adaptive pretraining offers large gains in task performance.
|
['Doug Downey', 'Ana Marasović', 'Kyle Lo', 'Noah A. Smith', 'Iz Beltagy', 'Swabha Swayamdipta', 'Suchin Gururangan']
|
2020-04-23
|
don-t-stop-pretraining-adapt-language-models-1
|
https://aclanthology.org/2020.acl-main.740
|
https://aclanthology.org/2020.acl-main.740.pdf
|
acl-2020-6
|
['citation-intent-classification']
|
['natural-language-processing']
|
[ 5.03371477e-01 5.69588393e-02 -5.27274132e-01 -5.62149227e-01
-1.09651625e+00 -7.93576658e-01 7.28962898e-01 2.79260099e-01
-1.02674091e+00 9.19987440e-01 4.33896035e-01 -5.77191770e-01
-9.28600281e-02 -2.97290355e-01 -4.61638898e-01 -1.52065337e-01
2.83051789e-01 9.91323471e-01 2.32105747e-01 -1.86265811e-01
2.91822135e-01 3.94185841e-01 -1.05846775e+00 5.70411861e-01
1.00546122e+00 4.98320758e-01 5.12815058e-01 3.56084585e-01
-3.65907013e-01 3.91477615e-01 -5.50179064e-01 -3.14623833e-01
2.68011898e-01 -3.13506305e-01 -1.03059304e+00 -1.30993184e-02
2.67984688e-01 2.80132592e-02 9.92766619e-02 8.05282414e-01
5.79580009e-01 2.88840145e-01 9.11389410e-01 -8.34658563e-01
-5.98924220e-01 5.18844664e-01 -3.59793246e-01 5.25038362e-01
2.73382783e-01 3.45493667e-02 9.43164945e-01 -1.06552482e+00
1.02919710e+00 1.12217450e+00 6.61371946e-01 6.42833710e-01
-1.45881581e+00 -7.06248105e-01 3.52321714e-01 -8.66157934e-02
-1.00612533e+00 -6.82054818e-01 6.54785156e-01 -5.37910759e-01
1.34467053e+00 -1.51697636e-01 1.27391741e-02 1.29034710e+00
3.18723500e-01 6.56068981e-01 1.21346962e+00 -7.37588286e-01
1.62163347e-01 5.08120775e-01 3.15267950e-01 2.31597066e-01
4.42809582e-01 -6.76007047e-02 -5.43680131e-01 -4.20319319e-01
4.08749402e-01 -3.29465359e-01 -5.82238548e-02 -2.44924620e-01
-1.23439097e+00 7.84362018e-01 -1.04923643e-01 5.80135942e-01
-4.46046591e-01 -4.77191120e-01 6.94050789e-01 6.77737176e-01
8.61960828e-01 1.17904019e+00 -1.10146987e+00 -1.40637517e-01
-9.92730260e-01 1.84393153e-01 9.05781150e-01 9.82460678e-01
8.57857883e-01 -2.07774919e-02 -6.16660379e-02 1.19952643e+00
7.78448135e-02 2.85833567e-01 9.80784416e-01 -6.87289715e-01
7.76416779e-01 5.65827489e-01 5.79794087e-02 -5.02638042e-01
-6.83383286e-01 -5.16465902e-01 -4.11717772e-01 -2.44943708e-01
6.85914636e-01 -7.03774452e-01 -1.11335981e+00 1.84583902e+00
8.46148431e-02 -1.98303282e-01 2.74877697e-01 5.84887028e-01
6.41987264e-01 4.65661556e-01 7.03748047e-01 -3.87061715e-01
1.35346556e+00 -7.20184624e-01 -6.13689005e-01 -1.01664698e+00
1.15984321e+00 -8.09494495e-01 1.25124288e+00 4.25988615e-01
-8.22766900e-01 -5.59110463e-01 -7.12412298e-01 -2.44976535e-01
-4.93185431e-01 7.63603896e-02 4.27235603e-01 5.66514432e-01
-7.99490750e-01 3.29884291e-01 -4.95741129e-01 -7.48567343e-01
7.99042434e-02 3.34265947e-01 -5.05685449e-01 -4.73448128e-01
-1.32673812e+00 1.26276267e+00 6.58919454e-01 -4.51436907e-01
-7.15630114e-01 -7.14092135e-01 -8.03158581e-01 -4.96604778e-02
5.82527101e-01 -6.45801306e-01 1.43732691e+00 -1.25784266e+00
-1.17501974e+00 8.99484158e-01 -2.20068142e-01 -3.37786704e-01
3.79204661e-01 -2.38082141e-01 -4.42669988e-01 -1.38761420e-02
2.82099217e-01 6.73650205e-01 7.32434690e-01 -9.99115169e-01
-6.43604040e-01 -2.19903588e-01 -1.71924397e-01 6.01524293e-01
-6.87579215e-01 2.64312595e-01 -3.32313508e-01 -6.18582189e-01
-2.30048001e-01 -8.62799525e-01 -4.49241221e-01 -5.37648618e-01
-4.54822034e-02 -4.00486887e-01 4.42747235e-01 -6.70840740e-01
1.08788645e+00 -2.10244679e+00 -7.13322982e-02 1.04769371e-01
-1.70785766e-02 3.37289006e-01 -4.43508089e-01 4.64353621e-01
-2.02254936e-01 2.88915068e-01 -2.05984101e-01 -1.72564894e-01
-7.63646811e-02 2.43580431e-01 -6.64308481e-03 1.63892344e-01
3.49570096e-01 7.93960631e-01 -8.62266660e-01 -5.53266466e-01
-1.44283175e-01 2.73130983e-02 -5.67701697e-01 -7.55010173e-02
-3.90846729e-01 3.56908798e-01 -6.82333112e-01 3.38227957e-01
2.34481066e-01 -3.94465506e-01 5.27963698e-01 1.35650054e-01
2.04870448e-01 6.02298141e-01 -8.69493663e-01 1.67675138e+00
-5.43108761e-01 6.09191239e-01 8.31322819e-02 -1.17117453e+00
8.61456037e-01 4.56115007e-01 3.98154229e-01 -8.60739052e-01
2.21226104e-02 3.15884203e-01 4.03621078e-01 -4.87824976e-01
4.39666837e-01 -5.19345105e-01 -1.16776034e-01 6.02909267e-01
3.27298313e-01 -6.22212663e-02 3.37368608e-01 1.13974310e-01
1.32656348e+00 6.67674690e-02 5.60396552e-01 -4.44140166e-01
3.61676484e-01 4.95336294e-01 5.46082079e-01 7.89877534e-01
-2.00166047e-01 2.91797012e-01 1.90976262e-01 -2.55472869e-01
-1.09262621e+00 -5.14979661e-01 -2.84845412e-01 1.72505820e+00
-5.42699993e-01 -3.22992504e-01 -5.02584577e-01 -1.07824516e+00
1.25262141e-01 8.68740916e-01 -5.04223108e-01 -5.61338663e-02
-5.75602114e-01 -9.44595098e-01 3.97411704e-01 5.04682243e-01
2.88998008e-01 -1.20963085e+00 -3.32869470e-01 3.83957148e-01
-1.84555128e-01 -1.23047125e+00 -4.38040227e-01 7.05418587e-01
-1.00299573e+00 -7.98817813e-01 -8.96154940e-01 -7.83983111e-01
7.17968285e-01 1.34275794e-01 1.30315471e+00 -2.67417014e-01
2.24517658e-01 4.48110759e-01 -4.15662944e-01 -5.76872945e-01
-6.22074842e-01 6.23557866e-01 8.37938860e-02 -4.80155110e-01
7.34502554e-01 -1.53731003e-01 -8.75141285e-03 2.71664351e-01
-8.00469637e-01 -1.46694615e-01 9.44241703e-01 8.77619088e-01
4.25237954e-01 -2.78589111e-02 1.14663339e+00 -1.62498105e+00
1.07552719e+00 -7.72681594e-01 -1.41758561e-01 3.24696481e-01
-8.95853698e-01 1.17006555e-01 6.16516709e-01 -7.51555383e-01
-1.30091822e+00 7.95018151e-02 -1.18358433e-01 -2.81969942e-02
-3.82239461e-01 1.08992696e+00 9.85042192e-03 1.08786911e-01
1.27843487e+00 1.52713116e-02 -1.26705229e-01 -6.98855698e-01
1.38798654e-01 5.93018532e-01 5.58180474e-02 -8.42974961e-01
5.87706864e-01 1.35887144e-02 -4.07550901e-01 -8.07397604e-01
-9.55872893e-01 -6.50973499e-01 -7.99920022e-01 2.50847220e-01
6.24871492e-01 -9.53388393e-01 1.88253567e-01 2.37094443e-02
-9.75917161e-01 -8.19388926e-01 -2.90909141e-01 6.97607338e-01
-3.08351338e-01 1.35123894e-01 -2.16258943e-01 -4.10398751e-01
-1.18787244e-01 -9.90677595e-01 7.52469957e-01 2.15515606e-02
-6.09023988e-01 -1.45639586e+00 9.06195790e-02 2.47262180e-01
3.05198640e-01 -2.64484763e-01 1.15579164e+00 -1.62917686e+00
1.54920846e-01 -1.60226926e-01 -1.59070775e-01 3.44864756e-01
3.82023305e-01 -5.47055066e-01 -8.83936167e-01 -2.92316735e-01
-2.16752570e-02 -5.41324019e-01 6.96678877e-01 3.44147891e-01
8.59222114e-01 -1.49515674e-01 -5.88148057e-01 2.25345671e-01
1.22990835e+00 3.29813480e-01 1.41876042e-01 4.80011970e-01
3.51264119e-01 7.45777130e-01 7.60085225e-01 7.91865885e-02
1.85604081e-01 4.77819622e-01 -5.12829840e-01 -1.00199215e-01
-8.49330891e-03 -1.33545920e-01 3.29046279e-01 5.86328149e-01
1.80707142e-01 -3.25093836e-01 -1.43041706e+00 8.03658485e-01
-1.62601125e+00 -6.62635505e-01 2.68248916e-01 1.99919260e+00
1.20494092e+00 4.71154660e-01 2.77252764e-01 -3.37771207e-01
4.83508348e-01 -1.67206854e-01 -7.24806786e-01 -4.31863546e-01
8.99280161e-02 2.65825719e-01 6.09635651e-01 2.89766073e-01
-9.49543417e-01 1.05693913e+00 7.57004261e+00 7.95821190e-01
-1.07368577e+00 3.03425014e-01 6.13049686e-01 1.19417533e-01
-3.84972751e-01 -2.18252838e-01 -1.06035733e+00 2.85850435e-01
1.38262343e+00 -6.18951261e-01 7.84167871e-02 9.96971786e-01
9.46233496e-02 -1.24454811e-01 -1.28126955e+00 4.84105200e-01
1.19895212e-01 -1.10984313e+00 -3.76029164e-02 9.10070688e-02
6.30295753e-01 3.61665845e-01 -3.45150866e-02 7.78769612e-01
6.13930285e-01 -8.18337142e-01 2.68201679e-01 -1.48367852e-01
7.58534431e-01 -5.18580496e-01 8.48350167e-01 8.32041442e-01
-6.83703601e-01 -8.59259721e-03 -4.19926763e-01 -5.31668589e-02
3.07834335e-02 4.48988467e-01 -1.44443512e+00 5.65268397e-01
3.80932152e-01 6.22416675e-01 -7.33511746e-01 7.67018318e-01
-2.21836995e-02 8.63230407e-01 -2.91823089e-01 7.97332302e-02
2.04573020e-01 2.76758671e-01 3.25947404e-01 1.65695155e+00
7.11411312e-02 1.46244317e-01 4.84015465e-01 3.11449885e-01
-1.36804909e-01 4.89683092e-01 -9.24689531e-01 -4.13026154e-01
4.72051054e-01 1.05410635e+00 -6.91286743e-01 -5.55886805e-01
-5.79995036e-01 5.50193191e-01 4.57634747e-01 5.40714741e-01
-1.54253781e-01 -2.31410965e-01 2.72030741e-01 1.92050830e-01
1.86819863e-02 -3.30719531e-01 -5.21266878e-01 -1.05421484e+00
-3.27418715e-01 -1.25381851e+00 6.09225154e-01 -5.96975923e-01
-1.57728994e+00 5.82533360e-01 2.92694360e-01 -9.83906209e-01
-5.66503286e-01 -6.86821580e-01 -1.34983838e-01 1.00925946e+00
-1.59088790e+00 -9.27061915e-01 2.36938000e-01 4.75499243e-01
8.82316530e-01 -4.69966859e-01 8.82013500e-01 3.81939977e-01
-3.60714227e-01 5.86097181e-01 1.80395484e-01 1.25269100e-01
1.25196755e+00 -1.17898870e+00 3.43400836e-01 6.46339893e-01
1.80055395e-01 8.70800495e-01 7.37133145e-01 -9.64316010e-01
-1.11931920e+00 -9.91662443e-01 1.28425980e+00 -5.70768356e-01
7.24418283e-01 -3.30693394e-01 -1.14227760e+00 1.00125086e+00
2.86640197e-01 -3.66881222e-01 8.82359147e-01 7.37693131e-01
-2.48220667e-01 1.13238640e-01 -1.14140129e+00 3.55251968e-01
9.14870441e-01 -5.17599046e-01 -9.97807741e-01 6.90649092e-01
4.69570071e-01 -4.08138067e-01 -7.90435970e-01 3.34655523e-01
3.37530702e-01 -1.99886888e-01 6.91611409e-01 -1.06312382e+00
3.37482095e-01 2.74909645e-01 1.71691757e-02 -1.59122372e+00
-4.79765981e-01 -3.87156457e-01 4.85756665e-01 1.09216464e+00
9.64215577e-01 -7.29386568e-01 6.94212615e-01 8.79345000e-01
-3.13787758e-01 -3.29431593e-01 -6.54692531e-01 -8.84872377e-01
5.39496303e-01 -4.73820388e-01 8.38239342e-02 1.26075661e+00
1.02442481e-01 7.94870257e-01 -2.64116973e-02 -2.07234308e-01
1.59897223e-01 -2.92269588e-01 6.30085707e-01 -1.41776299e+00
-1.66772544e-01 -2.03759342e-01 2.21908242e-01 -1.04615927e+00
6.13566875e-01 -1.07976198e+00 1.89392850e-01 -1.69541788e+00
1.58219337e-01 -6.99040711e-01 -5.25055468e-01 9.51254368e-01
-3.08294713e-01 -6.82028979e-02 3.14936899e-02 2.48806998e-01
-4.41047162e-01 -3.43161747e-02 1.17454457e+00 5.66406138e-02
-4.47034866e-01 -4.11809050e-02 -1.06949985e+00 7.55403340e-01
7.55016923e-01 -7.80027092e-01 -7.22628534e-01 -6.19399369e-01
2.21082717e-01 -1.37904495e-01 -2.93449968e-01 -6.68711483e-01
2.84682363e-01 -4.25762624e-01 4.51709211e-01 -2.41485402e-01
1.78253874e-01 -8.33575308e-01 -1.37335241e-01 2.18297988e-01
-7.09662259e-01 1.83246240e-01 7.68357754e-01 5.24767578e-01
-1.69768780e-01 -5.54016650e-01 8.10132563e-01 -3.03263396e-01
-9.03014719e-01 -2.43125912e-02 -6.37904823e-01 6.07676983e-01
7.00732946e-01 -1.57633781e-01 -4.10687268e-01 -1.52960360e-01
-8.22430551e-01 2.21439376e-01 2.90188313e-01 4.09292847e-01
2.72527397e-01 -9.89110410e-01 -7.36143589e-01 1.89798698e-01
1.87964872e-01 -1.36307061e-01 -3.53408456e-02 6.34374201e-01
-1.95447430e-02 7.95922518e-01 -2.09410056e-01 -3.93272191e-01
-1.16077590e+00 5.07312298e-01 -6.61601126e-02 -6.13923430e-01
-4.35880810e-01 8.06089759e-01 3.73279601e-01 -5.97683311e-01
9.56928357e-02 -2.87359357e-01 -2.80910909e-01 4.17287201e-01
2.69426793e-01 -5.18632792e-02 2.74526715e-01 -3.13318461e-01
-3.87566894e-01 1.49639949e-01 -6.30491436e-01 -4.89763319e-01
1.38909769e+00 9.46559384e-03 4.97804880e-01 5.80410898e-01
8.81676018e-01 2.26046145e-02 -8.56217444e-01 -7.45932579e-01
5.25654435e-01 -1.42203003e-01 3.83863710e-02 -1.13157523e+00
-5.31622589e-01 7.40960062e-01 5.63626699e-02 -1.78345684e-02
1.05810714e+00 -1.12156853e-01 3.73401493e-01 8.12429249e-01
3.94555897e-01 -1.30683339e+00 4.72045913e-02 6.88238680e-01
6.58021271e-01 -1.28689587e+00 2.87000865e-01 -2.22348884e-01
-1.02996540e+00 7.71498144e-01 6.78586841e-01 3.46152782e-02
8.29697192e-01 1.74258381e-01 1.08439490e-01 -4.30606566e-02
-9.98965681e-01 -1.42295823e-01 4.62590992e-01 6.56892836e-01
6.96162760e-01 -2.65538871e-01 -5.31959236e-01 5.90622425e-01
1.36713535e-01 1.57222569e-01 2.61157781e-01 1.07814157e+00
-4.64932680e-01 -1.43817520e+00 -1.23692557e-01 8.20729911e-01
-5.82492590e-01 -3.23750794e-01 -6.44733191e-01 1.03594410e+00
6.04334213e-02 7.63404250e-01 -1.48827732e-01 -1.24661559e-02
3.17480505e-01 7.70035207e-01 3.06859732e-01 -1.44228709e+00
-8.22407007e-01 4.01436180e-01 6.30449831e-01 -4.28999290e-02
-4.45673943e-01 -8.44161272e-01 -1.03291392e+00 1.13744631e-01
-2.53808945e-01 3.16559225e-01 5.80405772e-01 1.25404799e+00
4.82021779e-01 3.49573314e-01 1.20037287e-01 -3.81961286e-01
-6.07358992e-01 -1.23653293e+00 -3.80965203e-01 2.60683447e-01
8.94374922e-02 -7.63663709e-01 -1.89513609e-01 4.09432799e-01]
|
[10.60505485534668, 8.158111572265625]
|
e10560d2-87bc-47a0-83df-f5747f5bef93
|
cov3d-detection-of-the-presence-and-severity
|
2207.12218
| null |
https://arxiv.org/abs/2207.12218v1
|
https://arxiv.org/pdf/2207.12218v1.pdf
|
Cov3d: Detection of the presence and severity of COVID-19 from CT scans using 3D ResNets
|
Deep learning has been used to assist in the analysis of medical imaging. One such use is the classification of Computed Tomography (CT) scans when detecting for COVID-19 in subjects. This paper presents Cov3d, a three dimensional convolutional neural network for detecting the presence and severity of COVID19 from chest CT scans. Trained on the COV19-CT-DB dataset with human expert annotations, it achieves a macro f1 score of 0.9476 on the validation set for the task of detecting the presence of COVID19. For the task of classifying the severity of COVID19, it achieves a macro f1 score of 0.7552. Both results improve on the baseline results of the `AI-enabled Medical Image Analysis Workshop and Covid-19 Diagnosis Competition' (MIA-COV19D) in 2022.
|
['Robert Turnbull']
|
2022-07-05
| null | null | null | null |
['covid-19-detection']
|
['medical']
|
[ 4.07091193e-02 1.32440820e-01 -1.54511541e-01 -3.07840556e-01
-1.18275785e+00 -3.39079797e-01 1.42084450e-01 3.29011619e-01
-4.44671273e-01 2.70735860e-01 2.98122495e-01 -5.36121964e-01
-2.21425325e-01 -3.61061215e-01 -4.78178591e-01 -5.73520839e-01
-4.46089447e-01 9.55314100e-01 1.91258237e-01 3.85730118e-01
-5.59007525e-01 4.74934965e-01 -6.95024133e-01 7.52893865e-01
-1.69202313e-01 1.50061905e+00 1.89087585e-01 1.32550037e+00
7.26034224e-01 9.35589194e-01 -4.28717047e-01 -2.94826664e-02
9.56118852e-02 -1.82501361e-01 -1.00382459e+00 -2.07159892e-01
4.72754866e-01 -5.44123888e-01 -4.40241605e-01 5.34667075e-01
7.92680502e-01 -5.31980753e-01 1.12021017e+00 -8.09948921e-01
4.85865697e-02 2.98846692e-01 -6.76564336e-01 1.17028332e+00
1.31856248e-01 3.11496526e-01 8.72995794e-01 -7.37249911e-01
7.77956963e-01 7.95858860e-01 1.28478527e+00 4.47396487e-01
-7.13486195e-01 -5.86550355e-01 -7.12682962e-01 7.68497363e-02
-1.26092029e+00 1.93938896e-01 -1.01721995e-01 -9.05774057e-01
1.05932462e+00 3.85848910e-01 5.14321208e-01 1.06652856e+00
9.18357491e-01 7.13429451e-01 4.90202427e-01 -6.94203600e-02
-5.44675961e-02 -3.94540787e-01 4.56826895e-01 9.30070400e-01
1.83771923e-01 1.17292650e-01 6.53409734e-02 -3.36784631e-01
7.50545442e-01 -1.58508211e-01 1.53642863e-01 6.45932779e-02
-1.36627769e+00 9.35533464e-01 6.41395628e-01 3.85165274e-01
-6.60823882e-01 2.89764732e-01 9.90683794e-01 -8.30019712e-02
4.00499642e-01 3.95072222e-01 -3.55133504e-01 1.06526442e-01
-6.15723312e-01 4.94321197e-01 1.81629136e-01 4.51382488e-01
-4.90370095e-01 -3.44542935e-02 -4.77669746e-01 5.99596858e-01
5.75840734e-02 6.47419870e-01 7.15988755e-01 -8.45850229e-01
3.53357673e-01 3.68081063e-01 -1.13084681e-01 -6.10405743e-01
-1.36578524e+00 -5.48838139e-01 -1.26488054e+00 2.28025503e-02
1.46588981e-01 -5.89372337e-01 -1.24569392e+00 1.20966601e+00
1.23031596e-02 -1.73222180e-02 -1.71105891e-01 9.95424330e-01
1.16250002e+00 2.07510129e-01 1.56625256e-01 1.31401420e-01
1.50273311e+00 -3.58368725e-01 -3.66738439e-01 2.19811976e-01
8.98791015e-01 -3.55874807e-01 5.65074325e-01 5.56253195e-01
-9.27108586e-01 -4.70271617e-01 -1.10170805e+00 2.93493241e-01
-1.12414472e-01 1.25338659e-01 4.15013075e-01 6.63120627e-01
-9.18354988e-01 3.14026862e-01 -9.88191128e-01 8.88453275e-02
6.79959059e-01 3.28977853e-01 -1.89584687e-01 -1.12760499e-01
-1.34021652e+00 1.06731522e+00 2.22987115e-01 -2.10153848e-01
-1.47245979e+00 -1.13595605e+00 -5.24382114e-01 -1.15492135e-01
4.78959791e-02 -9.16841209e-01 1.58420682e+00 -4.13825721e-01
-1.62186503e-01 1.28703392e+00 4.80058193e-01 -9.51049030e-01
9.80280161e-01 -1.09619148e-01 -6.70585871e-01 3.74707699e-01
3.41369599e-01 7.04068184e-01 6.35941982e-01 -7.03430414e-01
-7.21710622e-01 -2.97467440e-01 -1.79862022e-01 -6.37810081e-02
3.67570996e-01 2.81290144e-01 -3.12194228e-01 -7.32812226e-01
-3.85696858e-01 -1.08994019e+00 -5.17864048e-01 4.42467108e-02
-4.70743299e-01 -3.96417648e-01 8.47818434e-01 -7.74726152e-01
8.40446889e-01 -2.14156818e+00 -2.78399348e-01 2.32766047e-01
1.08037961e+00 4.57658291e-01 2.08425745e-01 -4.17692065e-01
-6.33051455e-01 1.71157077e-01 -1.82426035e-01 1.18795671e-01
-2.75625885e-01 -1.02232015e-02 4.10428524e-01 6.96434736e-01
3.75836611e-01 9.54199135e-01 -5.65832734e-01 -5.28421640e-01
1.59692302e-01 3.94257307e-01 -4.33104128e-01 1.03469208e-01
-1.73998341e-01 3.43457192e-01 -4.01166797e-01 5.93437195e-01
3.42555314e-01 -8.15895081e-01 2.85479948e-02 -2.03152642e-01
4.40975934e-01 -3.00171196e-01 -6.82236969e-01 1.29101956e+00
-4.07001585e-01 9.47538495e-01 6.34676665e-02 -7.05732286e-01
2.11372510e-01 8.80609930e-01 1.30682445e+00 -5.64931750e-01
5.13911664e-01 1.25927210e-01 5.56823373e-01 -8.82476091e-01
-9.22617391e-02 -1.58327162e-01 -3.00425291e-01 7.36405432e-01
-1.65692866e-01 -9.32836533e-02 -6.71372842e-03 2.22468644e-01
1.69172788e+00 -6.46660089e-01 2.98438191e-01 -3.40225309e-01
2.53979564e-01 2.28389785e-01 2.36483201e-01 8.31500590e-01
-4.91627365e-01 1.01201487e+00 3.01669151e-01 -8.90764654e-01
-1.01115787e+00 -1.17627347e+00 -5.78713596e-01 5.98543346e-01
-4.06415403e-01 -7.70243257e-02 -6.65827692e-01 -9.11898434e-01
4.56218980e-02 3.43954921e-01 -1.00555027e+00 2.92936694e-02
-8.47717643e-01 -1.01537454e+00 9.39566672e-01 1.07993245e+00
2.06903651e-01 -1.12620461e+00 -1.17966676e+00 2.18649402e-01
-3.01965415e-01 -1.23778725e+00 -2.77714372e-01 2.38781780e-01
-4.30246383e-01 -1.49685097e+00 -1.10443699e+00 -6.65366828e-01
1.34947583e-01 -2.60362089e-01 1.34384274e+00 2.61080950e-01
-1.17368627e+00 4.12239045e-01 -1.89678922e-01 -1.01607335e+00
-5.40055215e-01 -8.07000883e-03 1.27015352e-01 -6.64685011e-01
1.85351163e-01 -2.13193931e-02 -7.69854009e-01 4.71813083e-02
-7.50670969e-01 3.59776579e-02 5.36029756e-01 7.73586512e-01
5.84142923e-01 2.42681727e-02 4.21947300e-01 -9.79243279e-01
6.34709716e-01 -6.38452113e-01 -1.24459647e-01 -1.26212984e-01
-5.43707192e-01 -3.63625705e-01 2.71295846e-01 2.95003317e-02
-3.75337422e-01 2.18392566e-01 -4.19087648e-01 -7.71736503e-01
-2.31768936e-01 3.91218990e-01 7.19197571e-01 2.85820752e-01
9.57445621e-01 -1.86066613e-01 -1.44600540e-01 -2.45606273e-01
-1.35713622e-01 5.96128404e-01 8.71980727e-01 -1.17095768e-01
2.95683563e-01 2.93030083e-01 4.59202379e-01 -5.30752420e-01
-1.01566756e+00 -7.38025427e-01 -6.98039949e-01 -4.37192887e-01
1.56352913e+00 -9.15552557e-01 -5.38158417e-01 3.44221205e-01
-1.04320478e+00 -1.26333132e-01 -2.37301230e-01 3.97052050e-01
-3.75393748e-01 -1.53247779e-02 -6.94288671e-01 -1.52740538e-01
-8.42312276e-01 -1.48077178e+00 1.21005857e+00 -2.89906889e-01
-5.94386280e-01 -8.02737236e-01 1.23432457e-01 2.27672219e-01
3.81864280e-01 8.05381954e-01 1.23725998e+00 -8.28440666e-01
8.14703479e-02 -5.60890973e-01 -7.03672111e-01 2.53289521e-01
-1.12420902e-01 -3.75946254e-01 -8.82832706e-01 -2.14339539e-01
-3.53212327e-01 -4.82310593e-01 8.56900036e-01 9.10198212e-01
1.56385732e+00 1.62390724e-01 -4.51261878e-01 4.20252949e-01
1.33799195e+00 4.37612146e-01 1.62506953e-01 1.52662292e-01
7.16181755e-01 -5.57166263e-02 3.54645342e-01 5.70124745e-01
3.03955544e-02 4.73168582e-01 6.65943801e-01 -4.25513774e-01
-4.06804651e-01 6.35595202e-01 -3.91446590e-01 2.83205658e-01
-2.63670295e-01 -2.93712139e-01 -1.80155134e+00 6.32125795e-01
-1.32446659e+00 -4.83078659e-01 -4.70934898e-01 1.25499594e+00
4.66504514e-01 2.78887957e-01 3.73770684e-01 2.11588666e-01
5.90890765e-01 -1.21586077e-01 -5.56904256e-01 -5.84706783e-01
3.30665559e-01 6.21319234e-01 5.72364807e-01 6.07063286e-02
-1.56736100e+00 -4.03101109e-02 7.00267363e+00 5.61484098e-01
-1.18511355e+00 5.16106248e-01 9.62683916e-01 -1.01469271e-01
4.47174042e-01 -1.10472262e+00 -5.14807165e-01 4.59699839e-01
1.16751385e+00 9.86973494e-02 -3.40310395e-01 7.12579370e-01
1.82644859e-01 -6.10961206e-02 -1.11568892e+00 9.57745373e-01
1.19490467e-01 -1.58017838e+00 -2.20647708e-01 -2.54877917e-02
7.44916320e-01 6.03372335e-01 3.16350132e-01 3.28997105e-01
3.15024585e-01 -1.28833818e+00 2.16383561e-01 2.24813282e-01
1.50168979e+00 -7.91161299e-01 1.17761087e+00 1.14348568e-01
-9.34048474e-01 7.39075541e-02 5.21704741e-03 5.41673005e-01
9.37772989e-02 5.75264931e-01 -1.74601746e+00 3.67094517e-01
9.04352307e-01 4.48692620e-01 -6.54263794e-01 9.73186135e-01
3.05188805e-01 8.34676504e-01 -1.62283793e-01 3.00921589e-01
4.43246037e-01 8.81650150e-01 6.27409816e-01 1.61052549e+00
5.37342532e-03 4.23834056e-01 1.30592346e-01 3.66507709e-01
-2.24173784e-01 -1.92117959e-01 -5.27707279e-01 2.07767949e-01
-2.39892557e-01 1.00529778e+00 -7.01794028e-01 -5.79228342e-01
-5.88650666e-02 4.85788643e-01 -4.93275046e-01 -2.48097435e-01
-1.16646516e+00 -6.78373799e-02 1.13175690e-01 2.44631395e-01
2.52731025e-01 2.72746027e-01 -5.81588447e-01 -4.99236047e-01
-5.14607787e-01 -9.29712713e-01 8.89689088e-01 -9.67785835e-01
-1.18120968e+00 8.46639752e-01 -6.39070272e-02 -1.20893776e+00
-6.96350276e-01 -9.50747848e-01 -5.56567371e-01 6.05026960e-01
-9.04823661e-01 -9.67967927e-01 -4.10636336e-01 4.80255663e-01
5.55372477e-01 -2.85476118e-01 1.01835823e+00 3.39954525e-01
-2.81112373e-01 6.44514561e-01 -1.71964522e-02 5.98474383e-01
4.58930671e-01 -1.45421231e+00 5.38169861e-01 5.09187758e-01
-4.49360430e-01 -1.66826904e-01 3.81089956e-01 -6.36624813e-01
-9.22059357e-01 -1.41939032e+00 5.37999332e-01 -7.79038548e-01
4.06903833e-01 1.48000181e-01 -4.40235674e-01 6.33515656e-01
5.67849318e-04 2.46087983e-01 5.81518650e-01 -3.76657367e-01
-2.03267828e-01 4.33141321e-01 -1.43487597e+00 -5.46624810e-02
6.69555545e-01 -3.05472076e-01 -4.85672623e-01 7.15608835e-01
8.43429565e-01 -7.70966828e-01 -1.35689867e+00 9.57471609e-01
7.04627633e-01 -5.50493658e-01 1.29357040e+00 -9.21136379e-01
8.22320819e-01 3.57056916e-01 -1.46442488e-01 -9.62303877e-01
-4.91419941e-01 1.54300556e-01 1.44340143e-01 4.10009660e-02
4.34652895e-01 1.24146059e-01 9.08824325e-01 2.06542656e-01
-1.62973091e-01 -9.70555246e-01 -1.09246171e+00 -4.52953160e-01
2.66526878e-01 -7.57707477e-01 1.00982487e-01 1.07931209e+00
-6.05695128e-01 1.53002411e-01 6.97148144e-02 9.54084098e-02
3.76299977e-01 -1.91141635e-01 2.18840599e-01 -1.29130161e+00
-4.23513234e-01 -5.44843674e-01 -6.87216401e-01 -9.30489153e-02
-4.12849993e-01 -8.97206843e-01 -2.12765466e-02 -1.47356451e+00
6.60280526e-01 -2.91283727e-01 -6.78441644e-01 3.23387772e-01
1.91070978e-02 6.83783710e-01 1.45416200e-01 1.76414907e-01
-4.85963613e-01 -5.59810996e-01 1.25418103e+00 -3.72083843e-01
3.10805649e-01 1.84306711e-01 -1.11338250e-01 9.35838401e-01
8.28513503e-01 -5.92935860e-01 -4.08835150e-02 -4.31274623e-01
1.71933070e-01 7.08786368e-01 4.69525725e-01 -1.24385655e+00
-3.86823177e-01 2.05517322e-01 9.16893423e-01 -1.17248917e+00
1.92147478e-01 -7.28412211e-01 -2.47614365e-02 1.16974258e+00
-5.58494031e-01 5.37026227e-01 4.41931427e-01 5.11881649e-01
2.79250424e-02 -5.35458364e-02 1.02846062e+00 -4.43152040e-01
-5.72401881e-01 4.48224902e-01 -9.32119131e-01 4.87308413e-01
9.92123425e-01 1.68540180e-01 -1.50182813e-01 -2.06843838e-01
-1.08887923e+00 1.52883649e-01 -4.37681407e-01 2.32232988e-01
6.22498691e-01 -1.06943011e+00 -1.15627241e+00 -7.37880496e-03
1.15504354e-01 2.92205811e-01 3.74853879e-01 1.09096682e+00
-9.77762341e-01 7.39289403e-01 -1.45449400e-01 -1.06084847e+00
-1.59023678e+00 3.16282183e-01 8.61616433e-01 -8.58098745e-01
-8.42687488e-01 1.09417212e+00 3.49781334e-01 -1.04083149e-02
2.78792083e-01 -4.59658772e-01 -2.00558990e-01 -9.92437154e-02
5.83493650e-01 2.49184653e-01 5.60568154e-01 -5.02878368e-01
-6.19715333e-01 2.59015918e-01 -4.57964063e-01 1.03186384e-01
1.46662831e+00 4.27679330e-01 3.18898976e-01 -1.04271537e-02
1.54441869e+00 -6.85700536e-01 -3.95497978e-01 -1.82699338e-01
-4.51421887e-02 1.93667501e-01 3.09207469e-01 -1.22072232e+00
-1.15490246e+00 1.01091743e+00 1.52808464e+00 -1.21369317e-01
8.84263694e-01 2.08542019e-01 1.02230132e+00 2.70936787e-01
-2.55393445e-01 -6.59457266e-01 3.77539515e-01 4.11486328e-01
7.93897212e-01 -1.31477308e+00 8.29095468e-02 -2.67354064e-02
-7.74064004e-01 1.12101316e+00 4.18351591e-01 -2.65002817e-01
8.67753506e-01 3.81579101e-01 2.53031224e-01 -9.32011962e-01
-6.52612269e-01 2.25213453e-01 4.22665805e-01 6.61519229e-01
4.73381668e-01 7.17122018e-01 1.48145676e-01 5.91878831e-01
-6.16239570e-02 9.39472988e-02 4.86119121e-01 8.49436402e-01
-2.97503322e-01 -3.70906740e-01 -2.50692248e-01 1.00481188e+00
-1.15026486e+00 -2.43270844e-02 -3.93439621e-01 9.87032056e-01
4.29388314e-01 3.82213414e-01 7.98061714e-02 -4.02146459e-01
2.65682369e-01 -9.37971398e-02 3.16959083e-01 -6.76386595e-01
-7.59232461e-01 4.28770818e-02 3.32576931e-01 -5.94201565e-01
-2.78346717e-01 -6.46462440e-01 -1.32259440e+00 8.51054341e-02
2.67682802e-02 -3.45565915e-01 5.26442528e-01 8.22402775e-01
-8.21612924e-02 1.26947522e+00 3.53176594e-01 -3.32925379e-01
-5.03698111e-01 -8.26144934e-01 -4.35853690e-01 3.43221694e-01
7.01036274e-01 -5.98638296e-01 1.26739860e-01 4.06887103e-03]
|
[15.329541206359863, -1.8997148275375366]
|
98e5117c-066f-4023-98d0-34d2fd4d2583
|
evaluating-the-capability-of-large-scale
|
2307.03972
| null |
https://arxiv.org/abs/2307.03972v1
|
https://arxiv.org/pdf/2307.03972v1.pdf
|
Evaluating the Capability of Large-scale Language Models on Chinese Grammatical Error Correction Task
|
Large-scale language models (LLMs) has shown remarkable capability in various of Natural Language Processing (NLP) tasks and attracted lots of attention recently. However, some studies indicated that large language models fail to achieve promising result beyond the state-of-the-art models in English grammatical error correction (GEC) tasks. In this report, we aim to explore the how large language models perform on Chinese grammatical error correction tasks and provide guidance for future work. We conduct experiments with 3 different LLMs of different model scale on 4 Chinese GEC dataset. Our experimental results indicate that the performances of LLMs on automatic evaluation metrics falls short of the previous sota models because of the problem of over-correction. Furthermore, we also discover notable variations in the performance of LLMs when evaluated on different data distributions. Our findings demonstrates that further investigation is required for the application of LLMs on Chinese GEC task.
|
['Yunfang Wu', 'Fanyi Qu']
|
2023-07-08
| null | null | null | null |
['grammatical-error-correction']
|
['natural-language-processing']
|
[-2.86535740e-01 -1.38440698e-01 2.51422644e-01 -6.12406194e-01
-1.06195295e+00 -1.02103837e-01 3.03447753e-01 4.86375272e-01
-8.00036728e-01 8.22032213e-01 1.25187278e-01 -4.15300041e-01
1.45730615e-01 -4.53678429e-01 -6.90810740e-01 -1.72506586e-01
-4.08033021e-02 4.32598203e-01 3.73831302e-01 -2.54693389e-01
7.47139871e-01 5.86619452e-02 -1.11746168e+00 3.81443828e-01
1.54638422e+00 3.20720941e-01 5.28550088e-01 6.74791515e-01
-4.36732531e-01 7.83442676e-01 -8.52735639e-01 -6.97562575e-01
-3.37158352e-01 -3.53695363e-01 -7.35998392e-01 -4.80811298e-01
3.58686984e-01 3.05975288e-01 3.13408643e-01 1.32185221e+00
5.72375894e-01 -1.38497157e-02 4.57142264e-01 -8.75220358e-01
-1.11491156e+00 8.65306854e-01 -2.96513617e-01 3.58795673e-01
1.27949014e-01 -3.11068654e-01 6.94464624e-01 -1.12948918e+00
5.85284412e-01 1.46561861e+00 9.41457570e-01 7.06020355e-01
-5.59951305e-01 -6.44836426e-01 3.05830598e-01 2.47578636e-01
-1.60863471e+00 -1.98259473e-01 2.70285666e-01 -3.30640599e-02
1.65898287e+00 6.72148094e-02 -5.22483811e-02 6.50103211e-01
5.04451632e-01 6.60792410e-01 1.10446870e+00 -9.75038826e-01
-3.02310586e-02 2.71986365e-01 5.81895411e-01 7.42319942e-01
4.46239769e-01 -2.59548903e-01 -3.34952474e-01 -6.16650954e-02
6.68764189e-02 -4.77736533e-01 -3.98801975e-02 5.45296609e-01
-9.10410762e-01 7.27810323e-01 3.26333493e-02 8.31142306e-01
-6.19999655e-02 3.50405216e-01 5.09443045e-01 4.29105699e-01
9.42647874e-01 4.51724917e-01 -8.60241532e-01 -5.18103063e-01
-8.08832765e-01 2.20601216e-01 6.33046210e-01 1.24490714e+00
4.54762608e-01 1.38358727e-01 -1.45918474e-01 1.11369979e+00
3.89565706e-01 5.59626639e-01 6.57357156e-01 -5.77690899e-01
9.58604991e-01 6.57441437e-01 7.92195052e-02 -1.01724744e+00
-5.56944788e-01 -5.23803890e-01 -6.95749342e-01 -4.36772704e-01
3.68681818e-01 -8.76915306e-02 -5.27587235e-01 1.58916736e+00
-2.69024253e-01 -3.65359634e-02 6.10862076e-02 5.35185218e-01
8.53659630e-01 6.68234527e-01 7.31725097e-01 -7.44072869e-02
1.00639844e+00 -9.62107301e-01 -9.61291671e-01 -2.65149295e-01
1.49927008e+00 -1.03252518e+00 1.47834408e+00 1.63590074e-01
-1.05137181e+00 -6.40342116e-01 -6.58730149e-01 -1.92087248e-01
-3.67536664e-01 6.67454004e-01 5.40668607e-01 9.13531065e-01
-1.08938074e+00 5.81636846e-01 -8.97721767e-01 -6.98883533e-01
-1.26592770e-01 -2.17101220e-02 -1.88048497e-01 -1.72544524e-01
-1.24290657e+00 1.16702712e+00 5.03819227e-01 2.51753837e-01
-3.69992733e-01 -3.97381753e-01 -7.96733439e-01 -4.91021536e-02
7.38908276e-02 -1.06411494e-01 1.14411998e+00 -5.19488871e-01
-1.02077293e+00 9.19782043e-01 -5.49985826e-01 -3.66654396e-01
1.92026541e-01 -4.51653391e-01 -8.11597824e-01 -4.26135510e-01
1.35287449e-01 3.03280354e-01 1.84325591e-01 -1.02118301e+00
-8.18708777e-01 -4.37724769e-01 -3.87684017e-01 -7.85762817e-02
-2.40155235e-01 8.78445745e-01 -5.51451981e-01 -8.13177466e-01
9.31959134e-03 -1.00262535e+00 -9.36429128e-02 -6.94889128e-01
-1.36115506e-01 -7.43697524e-01 3.47695112e-01 -8.98485482e-01
2.19979811e+00 -1.97930300e+00 -9.45284888e-02 -9.18161124e-02
-3.04586083e-01 6.39726043e-01 -3.74702901e-01 4.85651702e-01
1.25989228e-01 8.16432655e-01 -2.66711980e-01 -8.25818419e-01
-1.03979021e-01 1.78641900e-01 4.39445078e-02 1.25775084e-01
1.39106631e-01 1.03865445e+00 -7.60139585e-01 -5.12982070e-01
-2.89450020e-01 3.39820608e-02 -5.72108984e-01 1.01843746e-02
7.85126686e-02 1.64623037e-01 -4.02532548e-01 7.88723767e-01
8.09690058e-01 -7.13943616e-02 9.75759979e-03 5.07113099e-01
-2.75803626e-01 4.98040676e-01 -9.67969060e-01 1.56067121e+00
-3.75766993e-01 3.39987725e-01 -2.24912509e-01 -9.08655286e-01
9.69060123e-01 3.24805856e-01 -3.18641871e-01 -7.58578897e-01
-5.00051454e-02 7.82760203e-01 3.63979578e-01 -5.97542942e-01
9.26829576e-01 1.53384106e-02 -3.05055708e-01 4.02141601e-01
1.40946522e-01 -2.62092203e-02 3.58233958e-01 2.00466588e-01
7.20685482e-01 -1.20912462e-01 1.94218576e-01 -5.65219462e-01
7.35926926e-01 -1.08085787e-02 7.17499137e-01 1.03438425e+00
-4.53584462e-01 5.96459448e-01 2.61699319e-01 -1.21133812e-01
-8.29478145e-01 -5.10167897e-01 -1.38250828e-01 1.02933741e+00
-2.21783772e-01 -7.97769248e-01 -1.08620441e+00 -8.08057487e-01
-1.91896677e-01 1.02060008e+00 -2.97300518e-01 -3.03324871e-03
-8.20966959e-01 -1.23191690e+00 1.00336051e+00 5.53770840e-01
5.52079916e-01 -1.26113677e+00 4.35964465e-02 2.88201362e-01
-3.92314285e-01 -1.26559007e+00 -4.67758954e-01 -2.47399881e-01
-1.10139763e+00 -9.85099077e-01 -3.80097389e-01 -1.07194531e+00
7.29275823e-01 1.00238778e-01 1.17550182e+00 5.78882933e-01
8.54500160e-02 2.04918995e-01 -8.17342401e-01 -7.85969853e-01
-9.25266683e-01 3.60187709e-01 7.68346861e-02 -6.52666271e-01
9.05923784e-01 5.38894534e-02 7.03924373e-02 1.44148707e-01
-5.47590137e-01 -1.99320689e-01 3.31576169e-01 6.35424793e-01
3.68394017e-01 -3.88508141e-02 9.64648247e-01 -1.26025903e+00
1.16101325e+00 -3.02065164e-01 -3.83767039e-01 8.20674419e-01
-9.15544331e-01 1.20125249e-01 6.01435781e-01 2.50061080e-02
-1.33542621e+00 -8.63099456e-01 -4.58626539e-01 3.41531754e-01
-6.09971024e-02 9.02739048e-01 2.59662196e-02 -1.51110485e-01
4.56333756e-01 8.30214098e-02 -4.83475715e-01 -8.31806719e-01
-1.64896369e-01 9.33433652e-01 1.20547011e-01 -8.45051527e-01
2.20100880e-01 -3.15339774e-01 -4.49240267e-01 -5.95992148e-01
-8.70866954e-01 -3.31120312e-01 -5.85984349e-01 1.07616052e-01
7.22505093e-01 -1.01899755e+00 -1.50152415e-01 8.32455695e-01
-1.52347088e+00 -1.64735705e-01 3.98447931e-01 5.47998488e-01
3.72860283e-02 6.71671093e-01 -1.00343132e+00 -8.45555842e-01
-6.35668397e-01 -1.05696416e+00 9.59392071e-01 1.48055270e-01
-1.15392141e-01 -1.29977381e+00 4.13704943e-03 3.76366615e-01
6.37444139e-01 -3.65701884e-01 1.22756946e+00 -6.75685406e-01
-2.39512712e-01 -4.38329369e-01 -1.57772437e-01 5.23678839e-01
-1.42632708e-01 2.93608122e-02 -5.24354875e-01 -3.44648838e-01
-1.26626030e-01 -2.11903170e-01 8.10819149e-01 3.10932606e-01
1.37293124e+00 3.73055637e-02 -1.05065405e-01 3.30419332e-01
1.52174556e+00 1.83820814e-01 7.17073202e-01 3.88568610e-01
6.27024233e-01 3.91293406e-01 7.93358862e-01 1.27278954e-01
6.51154220e-01 3.76413882e-01 -7.05360174e-02 2.70916820e-01
-1.44404948e-01 -4.84933197e-01 4.59185123e-01 1.56258655e+00
-3.24269116e-01 -6.71546996e-01 -1.09156215e+00 5.56952834e-01
-1.81477237e+00 -5.49129605e-01 -6.42940819e-01 2.20413661e+00
6.97160542e-01 1.01765625e-01 -5.75261235e-01 -2.59218007e-01
7.47453034e-01 -1.72749192e-01 5.06193489e-02 -9.59621131e-01
-4.81368065e-01 2.55562544e-01 2.51865983e-01 5.19578576e-01
-8.54615271e-01 1.49887002e+00 6.50037479e+00 1.00992966e+00
-7.19862461e-01 5.44792235e-01 5.59870780e-01 4.55269635e-01
-3.60565960e-01 1.51617274e-01 -1.36823344e+00 5.55098832e-01
1.38227320e+00 -4.91934232e-02 1.75125167e-01 5.19650340e-01
4.33026701e-01 -2.14244783e-01 -6.07990980e-01 7.78363705e-01
2.92701751e-01 -8.48736465e-01 1.21461190e-01 -3.14151913e-01
9.52726781e-01 3.59599620e-01 -4.79422092e-01 8.83006155e-01
1.10956766e-01 -9.24434006e-01 5.82335949e-01 4.69566464e-01
6.24277174e-01 -6.25009716e-01 1.13178623e+00 7.81027496e-01
-8.41243684e-01 -5.04646748e-02 -9.54485595e-01 -2.41278693e-01
3.74868512e-01 3.33041728e-01 -4.82470244e-01 5.03414154e-01
7.67322421e-01 6.47389948e-01 -1.20859694e+00 1.00148332e+00
-4.71060604e-01 1.08588886e+00 2.26726793e-02 -4.84253764e-01
1.17052592e-01 -1.72028780e-01 3.31902742e-01 1.60396397e+00
5.65218806e-01 8.40638112e-03 -9.27705318e-02 7.24663258e-01
-1.28121480e-01 6.33965969e-01 -2.29784742e-01 -2.72572994e-01
6.15450859e-01 8.12868536e-01 -4.41997230e-01 -4.50995862e-01
-6.57267213e-01 8.13961625e-01 9.03473079e-01 2.74008065e-01
-7.14230120e-01 -5.10322869e-01 2.08176374e-01 -1.99562889e-02
-8.44503939e-02 -4.31069702e-01 -5.28655767e-01 -1.51598406e+00
1.12133436e-01 -8.50046039e-01 4.04468507e-01 -7.14536488e-01
-1.28124630e+00 6.95475996e-01 -9.32807475e-02 -9.43415284e-01
2.80105434e-02 -6.30474865e-01 -6.80590391e-01 1.03612125e+00
-1.69721484e+00 -1.16739297e+00 3.52087268e-03 2.14306891e-01
7.90381253e-01 -3.33428055e-01 1.01204920e+00 4.92889315e-01
-6.94917262e-01 1.09704673e+00 3.57777417e-01 1.23169690e-01
9.32966053e-01 -1.06818736e+00 6.39700830e-01 1.25082409e+00
8.24011043e-02 9.11741018e-01 5.03886640e-01 -9.38672543e-01
-8.31349611e-01 -1.09109235e+00 2.17230892e+00 -6.63455665e-01
4.30594951e-01 -2.44993463e-01 -1.22705770e+00 6.56900823e-01
2.13696435e-01 -2.55968571e-01 5.13719618e-01 4.49167818e-01
7.09894076e-02 1.85967520e-01 -1.00695240e+00 2.46386960e-01
1.02032483e+00 -4.03836399e-01 -5.51285386e-01 1.11806445e-01
7.36966848e-01 -3.78759503e-01 -6.54412091e-01 4.85825956e-01
1.26610234e-01 -6.30268753e-01 4.75437075e-01 -9.73343790e-01
4.51738954e-01 -1.90377310e-01 -2.73560226e-01 -1.44990575e+00
-2.80589342e-01 -1.05081446e-01 2.79232204e-01 1.59388888e+00
7.06831157e-01 -6.82319820e-01 1.05085999e-01 8.64934325e-01
-5.66204906e-01 -5.28636873e-01 -6.83100343e-01 -7.93369532e-01
6.28532588e-01 -9.06716824e-01 4.02921647e-01 8.41772318e-01
4.04659212e-02 8.25592279e-02 -2.91837871e-01 1.93089828e-01
2.55703390e-01 -2.44712576e-01 4.11568940e-01 -1.00086939e+00
5.16416766e-02 -1.56244934e-01 -8.53701234e-02 -6.36571169e-01
5.22868156e-01 -1.07943976e+00 6.74332827e-02 -1.26786315e+00
3.71695757e-01 -6.46214485e-01 -3.06214482e-01 4.53657240e-01
-7.79364944e-01 5.14020547e-02 4.03534383e-01 6.66783601e-02
-8.20162654e-01 5.66988826e-01 9.46023226e-01 1.57484189e-01
-1.21069349e-01 -2.49172151e-02 -7.39751756e-01 6.45048380e-01
8.78350019e-01 -7.50615120e-01 -9.47294990e-04 -1.11796975e+00
4.22077209e-01 -2.42464304e-01 -1.40938357e-01 -7.86290348e-01
1.47399247e-01 5.71323326e-03 1.22291818e-02 -3.38084638e-01
-6.16745353e-01 -2.13295475e-01 -2.04243794e-01 2.88224220e-01
-3.66598666e-01 6.85485780e-01 3.74870270e-01 3.19759548e-01
-4.69705611e-01 -7.29300976e-01 5.94141304e-01 -3.74872148e-01
-9.68501568e-01 -2.42487323e-02 -4.75093067e-01 3.38952541e-01
4.11034137e-01 1.60590678e-01 -3.15202594e-01 -4.44005616e-02
-3.93543929e-01 3.01600933e-01 2.94536978e-01 8.19690466e-01
5.35871148e-01 -1.07475317e+00 -9.90269244e-01 1.40280396e-01
3.73087943e-01 -2.47611567e-01 2.27056310e-01 7.61863232e-01
-7.10940659e-01 9.12045717e-01 2.89003432e-01 -2.60904908e-01
-1.30869079e+00 1.55093968e-01 2.03630403e-01 -6.07500196e-01
-2.20835194e-01 9.48903263e-01 -2.50013083e-01 -9.86806095e-01
7.61541501e-02 -5.21418393e-01 -2.71810293e-01 -5.04849136e-01
5.01678467e-01 5.33795595e-01 4.81467634e-01 -6.29972696e-01
-4.05902892e-01 5.39950550e-01 -2.47271910e-01 2.54879475e-01
1.20583069e+00 -5.14663815e-01 -4.27465618e-01 4.34183657e-01
9.00563896e-01 1.72237381e-01 -3.95504057e-01 -2.89616704e-01
5.85328579e-01 -4.33661312e-01 -1.79308355e-01 -1.08500457e+00
-6.12389684e-01 1.05502975e+00 5.40867984e-01 -3.82490277e-01
9.41606283e-01 -2.01661572e-01 6.95878506e-01 4.03333932e-01
8.41450870e-01 -1.52262390e+00 -4.16991442e-01 1.12811208e+00
8.19497764e-01 -1.59197474e+00 -3.95135343e-01 -2.19972789e-01
-8.31030130e-01 1.02954197e+00 9.42877293e-01 3.48630771e-02
7.69020021e-01 6.55917078e-02 1.79507822e-01 -1.56542316e-01
-8.15112829e-01 6.25470579e-02 1.92908093e-01 2.26251468e-01
1.18714547e+00 2.95700967e-01 -1.40580261e+00 1.03098857e+00
-1.43002778e-01 2.63362546e-02 6.32312000e-01 7.54893541e-01
-4.86015439e-01 -1.46888793e+00 -8.40547830e-02 2.49552995e-01
-7.96473503e-01 -5.71016610e-01 -1.73991397e-01 9.14907575e-01
1.48721352e-01 1.12563491e+00 -1.02980360e-01 -7.94862211e-02
3.41524720e-01 3.59344244e-01 3.67934763e-01 -8.64566386e-01
-7.48725176e-01 -1.68147713e-01 1.89764947e-01 -3.87540936e-01
-3.90607506e-01 -6.77272379e-01 -1.44592428e+00 -2.50682980e-01
-6.76490426e-01 4.56447661e-01 7.42836952e-01 1.04489064e+00
4.80586857e-01 1.87886417e-01 -1.63549911e-02 -4.46845815e-02
-6.11925364e-01 -1.69393289e+00 -7.66859710e-01 5.68757832e-01
-1.52165189e-01 -2.30481207e-01 -3.06300402e-01 -1.69099733e-01]
|
[11.059443473815918, 10.738810539245605]
|
6fd58f30-f916-482a-af0f-466407bfeff6
|
gated-ensemble-of-spatio-temporal-mixture-of
|
2012.15408
| null |
https://arxiv.org/abs/2012.15408v2
|
https://arxiv.org/pdf/2012.15408v2.pdf
|
Gated Ensemble of Spatio-temporal Mixture of Experts for Multi-task Learning in Ride-hailing System
|
Designing spatio-temporal forecasting models separately in a task-wise and city-wise manner pose a burden for the expanding transportation network companies. Therefore, a multi-task learning architecture is proposed in this study by developing gated ensemble of spatio-temporal mixture of experts network (GESME-Net) with convolutional recurrent neural network (CRNN), convolutional neural network (CNN), and recurrent neural network (RNN) for simultaneously forecasting spatio-temporal tasks in a city as well as across different cities. Furthermore, an input agnostic feature weighting layer is integrated with the architecture for learning joint representation in multi-task learning and revealing the contribution of the input features utilized in prediction. The proposed architecture is tested with data from Didi Chuxing for: (i) simultaneously forecasting demand and supply-demand gap in Beijing, and (ii) simultaneously forecasting demand across Chengdu and Xian. In both scenarios, models from our proposed architecture outperformed the single-task and multi-task deep learning benchmarks and ensemble-based machine learning algorithms.
|
['D. Wang', 'M. Abrar', 'S. N. Sadeek', 'S. M. Rifaat', 'M. H. Rahman']
|
2020-12-31
| null | null | null | null |
['spatio-temporal-forecasting']
|
['time-series']
|
[-2.94408113e-01 -5.39399385e-01 -1.05673611e-01 -5.04693687e-01
-9.32214618e-01 -3.08461726e-01 6.90119147e-01 -5.61524034e-01
-4.63993996e-02 6.46959722e-01 4.97283757e-01 -6.73174918e-01
-3.40100050e-01 -7.86506712e-01 -4.64615881e-01 -8.71419907e-01
2.80059408e-02 4.54170585e-01 -9.49851647e-02 -2.43001878e-01
-1.48801565e-01 4.82809752e-01 -1.43035340e+00 4.05327708e-01
1.19355822e+00 1.12171113e+00 3.57537180e-01 4.97193187e-01
-7.73248225e-02 1.17092812e+00 -2.77883559e-01 -1.65735692e-01
2.83331066e-01 2.21866682e-01 -4.55501258e-01 1.43975556e-01
1.43224195e-01 -1.08971402e-01 -5.83172739e-01 1.45761803e-01
5.12638211e-01 5.65367341e-01 9.13303673e-01 -1.47831810e+00
-6.44762158e-01 3.97801757e-01 -4.99379247e-01 2.28119671e-01
-6.42596424e-01 1.94786176e-01 6.84635580e-01 -1.04305816e+00
-1.24115333e-01 8.57134581e-01 6.57497466e-01 1.00119725e-01
-7.64382482e-01 -8.71093512e-01 6.61219954e-01 2.31290802e-01
-1.23598003e+00 -3.62653196e-01 8.07831883e-01 -7.66150057e-01
1.22425628e+00 4.99587208e-02 8.27358067e-02 1.15168524e+00
2.96894282e-01 8.71783018e-01 8.12964678e-01 2.21577853e-01
-9.61116403e-02 9.23175290e-02 -6.01723343e-02 1.82772577e-01
-3.35111707e-01 2.18818352e-01 -1.09168617e-02 1.39956668e-01
4.20206666e-01 4.51427937e-01 4.79916722e-01 6.03708982e-01
-1.11847830e+00 7.98364103e-01 5.31978548e-01 3.50236833e-01
-9.32420850e-01 3.90088618e-01 5.51212013e-01 2.34035924e-01
1.36604857e+00 -2.19281182e-01 -8.33568037e-01 1.53872192e-01
-1.18591440e+00 1.98276699e-01 2.02542841e-01 7.60446846e-01
5.09180069e-01 8.51387560e-01 -3.27247262e-01 9.75254774e-01
1.47779197e-01 7.08927870e-01 2.83152550e-01 -4.99225438e-01
1.05635571e+00 5.93140543e-01 3.09373915e-01 -1.03781283e+00
-9.23152804e-01 -7.73705602e-01 -1.37732780e+00 -1.93189271e-03
1.65799379e-01 -7.80320227e-01 -8.69111061e-01 1.48037052e+00
1.89252332e-01 5.59452832e-01 8.18952322e-02 7.40931928e-01
6.98433876e-01 1.13480461e+00 4.76910353e-01 2.44073033e-01
1.16880202e+00 -1.52171397e+00 -5.88261068e-01 -2.77660489e-02
9.35605466e-01 -6.67825401e-01 3.54060620e-01 1.27103589e-02
-8.68336976e-01 -1.02868760e+00 -4.06861365e-01 -4.51303553e-03
-9.04234946e-01 9.55141902e-01 5.17319679e-01 2.28989065e-01
-8.71881843e-01 2.31420249e-02 -4.48056877e-01 -1.75903544e-01
2.12260261e-01 1.77093908e-01 -3.90882678e-02 8.23429823e-02
-1.47813261e+00 1.08250546e+00 2.08350986e-01 6.36388898e-01
-8.78623366e-01 -1.00649202e+00 -6.67040408e-01 2.75760770e-01
-2.39392564e-01 -6.11301124e-01 8.98343742e-01 -9.78978038e-01
-1.42863786e+00 1.29066408e-01 -3.61178160e-01 -3.00233275e-01
3.66124034e-01 -3.51067223e-02 -1.27623224e+00 -6.57682121e-01
1.53486088e-01 5.14684379e-01 7.90866256e-01 -8.84292662e-01
-1.22669327e+00 -2.84516603e-01 -3.55862290e-01 4.86399531e-02
-1.00165665e-01 2.14778423e-01 1.36600718e-01 -1.02073967e+00
-2.12056443e-01 -9.70736682e-01 -5.26789844e-01 -7.49481142e-01
-3.72940272e-01 -8.58144581e-01 1.13475847e+00 -1.10218084e+00
1.41811693e+00 -2.10637927e+00 -2.56912857e-01 2.55987734e-01
-3.48959118e-02 4.19591278e-01 -5.19104600e-01 5.80398500e-01
-3.16208243e-01 -1.60509825e-01 1.87364101e-01 -6.87348723e-01
4.70131822e-02 1.38166323e-01 -6.37213886e-01 3.19288820e-01
3.70717973e-01 1.19784498e+00 -6.36747539e-01 1.32981865e-02
4.70383972e-01 6.06819689e-01 2.05821663e-01 6.79133013e-02
-1.09254166e-01 7.36160874e-01 -5.60235500e-01 6.45291269e-01
9.09546018e-01 -3.34198594e-01 -7.42967352e-02 -1.00949354e-01
-5.29850483e-01 3.72901052e-01 -8.98089647e-01 1.14497054e+00
-1.17912138e+00 7.93606400e-01 -2.45566726e-01 -1.04989207e+00
1.17818773e+00 7.69740582e-01 7.47590482e-01 -1.08654976e+00
-2.26301745e-01 4.17070448e-01 -1.48871064e-01 -5.64672172e-01
6.10809088e-01 2.24544495e-01 -1.52371272e-01 8.18128884e-01
-2.20436692e-01 6.15442157e-01 -1.81445926e-01 -4.25813228e-01
5.09928763e-01 7.67157748e-02 -5.01212955e-01 -2.55501866e-01
5.52462399e-01 5.57416081e-02 6.24513328e-01 2.72114754e-01
2.44057886e-02 3.64906698e-01 -1.65391430e-01 -1.23319817e+00
-1.11476254e+00 -5.33885002e-01 -7.81582668e-04 1.55130291e+00
-3.53954703e-01 1.50977612e-01 1.15779728e-01 -6.15163684e-01
2.18261555e-01 9.61519659e-01 -8.15595508e-01 1.88571543e-01
-7.49474585e-01 -7.93098211e-01 5.58757722e-01 8.62581432e-01
6.59294903e-01 -8.44454348e-01 -3.94674897e-01 4.80449140e-01
-3.52654397e-01 -1.11005509e+00 -3.87181461e-01 1.42393678e-01
-6.30473375e-01 -5.74703693e-01 -1.03919399e+00 -5.68886399e-01
3.19014221e-01 5.45254290e-01 1.07383275e+00 -5.62036037e-01
4.52332497e-01 -7.18326643e-02 1.79827784e-03 -5.07584274e-01
1.64468899e-01 7.38695979e-01 -1.73579916e-01 5.05896986e-01
3.63407165e-01 -7.74769425e-01 -6.81857765e-01 8.62840652e-01
-5.07425606e-01 3.13642949e-01 5.25060892e-01 5.23334622e-01
4.14679646e-02 2.76459694e-01 1.27852666e+00 -4.76330787e-01
4.82285827e-01 -1.33290040e+00 -6.41238391e-01 5.60209215e-01
-4.45697814e-01 -2.26273894e-01 7.62367964e-01 -3.77770960e-01
-1.44713283e+00 2.04264708e-02 -3.15438956e-02 -5.23981512e-01
-3.11909407e-01 6.67322159e-01 2.73229331e-01 3.96983564e-01
2.03566670e-01 4.95083153e-01 -4.63807762e-01 -5.56591511e-01
2.77046591e-01 7.22001135e-01 1.50591999e-01 -3.57206434e-01
6.07266188e-01 2.72095382e-01 3.94725241e-02 -3.64541858e-01
-7.03378439e-01 -5.65099835e-01 -8.01242530e-01 -3.85261595e-01
9.64922667e-01 -1.53448367e+00 -7.22330749e-01 6.82965398e-01
-1.30506551e+00 -5.17181456e-01 1.94350645e-01 8.02106142e-01
-1.42179742e-01 -5.65618515e-01 -3.09253424e-01 -1.12749290e+00
-3.05339366e-01 -1.11729085e+00 1.07657480e+00 -7.26497322e-02
7.50733539e-02 -1.50645173e+00 1.07313186e-01 4.24668908e-01
1.32917655e+00 2.48662964e-01 8.71194839e-01 -6.82377100e-01
-4.36552644e-01 -1.43537596e-01 -6.65746033e-01 2.00667500e-01
2.48965412e-01 8.05860981e-02 -1.11051607e+00 8.72488692e-02
-6.58500314e-01 -9.95528474e-02 1.06597245e+00 8.15431952e-01
1.18481481e+00 -3.77240598e-01 -4.19651031e-01 3.68705094e-01
1.19554877e+00 3.51272315e-01 5.44087470e-01 1.40923217e-01
7.95366824e-01 9.30511773e-01 1.22966550e-01 6.00282311e-01
1.14108431e+00 6.99668109e-01 2.62329906e-01 -4.82939571e-01
9.17307436e-02 -3.63268256e-02 2.57264197e-01 7.92754829e-01
-3.60452861e-01 -4.19126779e-01 -1.04896319e+00 9.81973112e-01
-2.30253315e+00 -8.66476715e-01 -5.01496732e-01 1.64437509e+00
-1.19908154e-02 -4.63233650e-01 4.86770183e-01 -2.15081140e-01
4.50601995e-01 3.88610691e-01 -6.99369907e-01 -4.40370977e-01
-3.58641863e-01 -2.00740516e-01 7.74564266e-01 1.12085097e-01
-1.31685066e+00 7.93376803e-01 5.51400089e+00 1.19338775e+00
-1.55960679e+00 4.26044732e-01 1.12557197e+00 -6.96536601e-02
-3.60433191e-01 -4.46161032e-01 -7.87412822e-01 7.64530480e-01
1.37941134e+00 2.74098307e-01 2.28992790e-01 7.96910286e-01
6.61178827e-01 2.35400558e-01 -3.86051506e-01 8.20525825e-01
-3.51213872e-01 -1.37535405e+00 -1.10963620e-01 1.70102358e-01
1.28388321e+00 7.42698312e-01 5.43706894e-01 7.70014763e-01
6.26353681e-01 -1.30087936e+00 4.18875426e-01 8.87228847e-01
5.07029295e-01 -8.81398976e-01 8.97400618e-01 4.96294141e-01
-1.63715672e+00 -4.72484231e-01 -2.12437764e-01 -1.81153789e-02
4.85258132e-01 7.61697233e-01 -1.91463619e-01 1.03007793e+00
5.45952439e-01 1.03433073e+00 -1.56177104e-01 7.46453762e-01
2.84203857e-01 8.50678921e-01 -3.49251926e-01 3.80626082e-01
9.22476351e-01 -3.21634889e-01 1.35074303e-01 1.28423822e+00
7.27878273e-01 1.78639051e-02 2.29053199e-01 6.26802146e-01
-8.71777683e-02 -8.62820372e-02 -7.10209489e-01 2.99424380e-01
1.99142158e-01 1.39414740e+00 -2.02542916e-01 -1.78243563e-01
-6.08515501e-01 5.77377558e-01 1.71984881e-01 8.71639073e-01
-1.11271441e+00 1.65470969e-02 8.64341438e-01 -1.17350534e-01
4.51556325e-01 -4.67834324e-01 -6.49845064e-01 -1.01773286e+00
-2.17946991e-01 -1.06771521e-01 2.76851863e-01 -7.61401176e-01
-1.37936401e+00 8.25101554e-01 -9.83774439e-02 -1.41674960e+00
-2.86613911e-01 -3.75079215e-01 -1.22743475e+00 1.55908930e+00
-2.12207532e+00 -1.92856562e+00 -1.38794020e-01 7.86926806e-01
8.88159752e-01 -6.15415871e-01 5.10641217e-01 8.16492856e-01
-1.10630107e+00 3.82372171e-01 4.11475599e-01 1.19577132e-01
3.16545069e-01 -7.18851209e-01 6.98855579e-01 6.40748799e-01
-3.41065973e-01 3.19075659e-02 -8.88475329e-02 -3.11507732e-01
-7.47819304e-01 -2.09545398e+00 1.59381628e+00 -2.83379078e-01
5.08553267e-01 -1.97272673e-01 -7.05361664e-01 8.45292866e-01
2.84946114e-01 1.27396360e-01 6.28612399e-01 2.40574360e-01
-3.58293176e-01 -3.98465693e-01 -8.71756434e-01 1.75091282e-01
5.66937327e-01 -5.83857715e-01 3.45085472e-01 7.57816851e-01
4.90331233e-01 5.97999617e-02 -8.88857424e-01 5.26030600e-01
6.82842553e-01 -4.88543838e-01 8.48871112e-01 -9.28307235e-01
5.68127453e-01 -1.65223628e-01 -3.99491549e-01 -1.66934133e+00
-8.86166096e-01 -2.44571656e-01 -3.07441801e-02 1.10461926e+00
8.93001735e-01 -9.33285594e-01 4.70699966e-01 7.83388853e-01
-5.19226491e-01 -8.65924060e-01 -1.08997619e+00 -7.40185261e-01
4.34815049e-01 -8.57884288e-01 1.11204338e+00 1.04639459e+00
-6.70888543e-01 1.84371620e-01 -1.02385223e+00 3.02779764e-01
6.63997456e-02 1.33934975e-01 6.18443370e-01 -1.16787219e+00
3.45295578e-01 -7.78639555e-01 2.44542152e-01 -1.12089932e+00
5.68560839e-01 -7.04778790e-01 -5.32230675e-01 -1.59311056e+00
-3.14519256e-01 -8.94938469e-01 -5.53333879e-01 5.57935894e-01
2.89514344e-02 3.14210206e-02 -8.84228572e-03 2.78382629e-01
-4.91002172e-01 8.47450793e-01 1.09748316e+00 -2.55226314e-01
2.59632152e-02 7.13621616e-01 -2.71261632e-01 2.90430337e-01
9.13153529e-01 -2.41487682e-01 -4.21459883e-01 -8.39761794e-01
3.74439269e-01 5.70961870e-02 3.65302831e-01 -8.65840673e-01
5.17546713e-01 -2.93422103e-01 5.34183383e-01 -1.29542613e+00
3.01732123e-01 -1.27736092e+00 4.56600010e-01 5.23246825e-02
-3.74703616e-01 5.20895958e-01 3.81369084e-01 2.18429893e-01
-2.77891159e-01 6.21525049e-01 9.57260132e-02 2.04120547e-01
-8.34628165e-01 6.45909309e-01 -7.02916026e-01 -4.58904445e-01
1.08864748e+00 -2.45225027e-01 -4.32509184e-01 -3.82004917e-01
-6.48715675e-01 8.60229194e-01 -6.92012846e-01 6.20254099e-01
4.56015736e-01 -1.60645926e+00 -1.30502200e+00 3.64048392e-01
1.06294721e-01 -2.42733151e-01 9.85780418e-01 1.10763097e+00
1.96643934e-01 9.78340268e-01 -7.93676376e-02 -4.04188603e-01
-4.43171114e-01 2.76255071e-01 4.19340253e-01 -7.69419789e-01
-8.22343007e-02 5.94088376e-01 1.45633996e-01 -1.18776381e+00
-1.41532645e-02 -4.69497234e-01 -3.48308682e-01 4.42142099e-01
1.95476964e-01 9.49530303e-01 3.46262753e-01 -9.36522543e-01
-1.46722972e-01 5.00147581e-01 1.55418694e-01 1.98177442e-01
1.50977635e+00 -4.29686695e-01 2.70419449e-01 6.06778443e-01
1.29233360e+00 -7.10246623e-01 -1.24903333e+00 -5.20717800e-01
7.03969598e-02 -1.05699152e-02 3.64619493e-01 -1.11883020e+00
-1.55932605e+00 1.01794100e+00 3.88497055e-01 8.04737806e-02
1.32520115e+00 -3.69505674e-01 1.30506301e+00 6.85920343e-02
-2.09579011e-04 -1.03641093e+00 -4.06752735e-01 7.90372312e-01
7.99224794e-01 -1.60510385e+00 -5.47181606e-01 2.13431984e-01
-9.66293335e-01 1.24168038e+00 5.05074501e-01 -2.75864601e-01
1.36559045e+00 -8.78533125e-02 6.30919039e-02 -1.95229933e-01
-1.39025235e+00 -4.38770235e-01 8.48286092e-01 4.06762064e-01
3.25250745e-01 4.32278723e-01 4.72330630e-01 5.65771759e-01
3.44080389e-01 5.98218180e-02 -2.71687716e-01 3.43444824e-01
4.04378437e-02 -5.96875072e-01 -1.01466978e-03 4.74742264e-01
-9.63845775e-02 -1.67299300e-01 3.97890568e-01 7.08591580e-01
3.35707843e-01 1.16786599e+00 5.23314536e-01 -7.18930483e-01
1.57762170e-01 1.09017983e-01 -5.08125067e-01 -1.36363758e-02
-1.12025464e+00 4.24320549e-02 1.49118096e-01 -3.45403105e-01
-5.69456816e-01 -4.61557597e-01 -5.46703815e-01 -5.02520561e-01
-3.55568938e-02 2.97908555e-04 9.14423048e-01 1.24016619e+00
8.90053153e-01 7.56571293e-01 1.32280171e+00 -9.43476379e-01
-4.36776355e-02 -1.38648605e+00 -6.16328955e-01 -8.97816643e-02
5.55054843e-01 -6.83037877e-01 -1.05659455e-01 -3.25557441e-01]
|
[6.4812235832214355, 2.1677863597869873]
|
5a361efe-331d-4641-9a4d-c2a72ec8d4a9
|
solving-the-rubiks-cube-without-human
|
1805.07470
| null |
http://arxiv.org/abs/1805.07470v1
|
http://arxiv.org/pdf/1805.07470v1.pdf
|
Solving the Rubik's Cube Without Human Knowledge
|
A generally intelligent agent must be able to teach itself how to solve
problems in complex domains with minimal human supervision. Recently, deep
reinforcement learning algorithms combined with self-play have achieved
superhuman proficiency in Go, Chess, and Shogi without human data or domain
knowledge. In these environments, a reward is always received at the end of the
game, however, for many combinatorial optimization environments, rewards are
sparse and episodes are not guaranteed to terminate. We introduce Autodidactic
Iteration: a novel reinforcement learning algorithm that is able to teach
itself how to solve the Rubik's Cube with no human assistance. Our algorithm is
able to solve 100% of randomly scrambled cubes while achieving a median solve
length of 30 moves -- less than or equal to solvers that employ human domain
knowledge.
|
['Alexander Shmakov', 'Forest Agostinelli', 'Pierre Baldi', 'Stephen McAleer']
|
2018-05-18
| null | null | null | null |
['rubik-s-cube']
|
['graphs']
|
[-7.63875842e-02 4.26750273e-01 1.38312951e-01 -4.29585315e-02
-3.83013308e-01 -9.24750626e-01 2.17848793e-01 4.47363630e-02
-6.19204581e-01 1.42673647e+00 -5.73759854e-01 -4.84147817e-01
-2.80157238e-01 -9.94966209e-01 -7.06847668e-01 -6.17209911e-01
-1.90757468e-01 1.03784287e+00 2.09993318e-01 -6.26194239e-01
3.98849249e-01 2.97707230e-01 -1.35594726e+00 -2.56443560e-01
1.20989895e+00 6.44872487e-01 4.73643303e-01 9.66803372e-01
6.24176674e-02 1.11212242e+00 -5.41462004e-01 -1.38578326e-01
5.13383090e-01 -6.21584713e-01 -1.15445733e+00 1.94893897e-01
-3.88333686e-02 -4.92686749e-01 -2.95255601e-01 9.56498146e-01
2.18767852e-01 2.23129749e-01 5.66791236e-01 -1.23420262e+00
-2.18569458e-01 7.48793066e-01 -4.31587398e-01 1.32470518e-01
6.58677638e-01 6.19916320e-01 9.69436467e-01 -9.13281813e-02
6.86534822e-01 7.96451211e-01 2.10101560e-01 6.96736097e-01
-1.00456977e+00 -5.12676120e-01 -1.95486508e-02 3.63142997e-01
-1.14578795e+00 1.20266818e-01 4.44403648e-01 -1.19872108e-01
1.21447015e+00 -1.23343870e-01 1.17112279e+00 8.03191125e-01
1.13786839e-01 7.43474483e-01 1.09114337e+00 -3.85328680e-01
7.32926250e-01 -1.04571462e-01 -4.51520741e-01 8.52893591e-01
2.13047594e-01 5.33420980e-01 -2.92872310e-01 2.46473923e-01
1.02520525e+00 -4.00411725e-01 1.61991566e-01 -7.94915676e-01
-1.15951824e+00 9.75922108e-01 4.16941702e-01 1.39042139e-01
-6.42562747e-01 3.29384923e-01 1.69486240e-01 7.53566861e-01
-4.29746360e-01 1.30112708e+00 -4.93243843e-01 -5.42645216e-01
-7.03559220e-01 7.77903795e-01 1.17931080e+00 9.27893102e-01
6.84477329e-01 3.94100875e-01 5.77303886e-01 1.14587456e-01
-2.96706200e-01 3.97010982e-01 3.51392716e-01 -1.49933910e+00
4.28089201e-01 7.04154789e-01 5.75656176e-01 -6.43043339e-01
-6.34106457e-01 -3.46678317e-01 -4.96448129e-01 9.96180415e-01
6.51949286e-01 -7.14884937e-01 -6.87123716e-01 1.39648497e+00
6.02978766e-01 -1.12892084e-01 3.29843849e-01 1.17025542e+00
4.24071789e-01 6.52304113e-01 -1.78885937e-01 -1.84317723e-01
1.09668159e+00 -1.00515997e+00 -2.43417710e-01 -5.00242889e-01
7.20325649e-01 -2.22506225e-01 8.20789099e-01 9.43336308e-01
-1.62604773e+00 -2.03105226e-01 -1.19052529e+00 3.07836115e-01
-2.18563840e-01 -4.87085700e-01 9.71286535e-01 6.10503376e-01
-9.32595551e-01 7.52867162e-01 -9.22266543e-01 1.23042930e-02
3.45040888e-01 7.21843302e-01 -4.68000919e-01 -1.12748832e-01
-9.77971971e-01 1.14836502e+00 7.59556174e-01 -2.92723715e-01
-1.36701179e+00 -4.19022828e-01 -8.00055981e-01 2.65030742e-01
9.34404075e-01 -6.74571455e-01 1.57397568e+00 -1.11364090e+00
-1.78293812e+00 7.91877091e-01 3.35068017e-01 -7.09738493e-01
6.32197976e-01 8.53681415e-02 1.38561711e-01 3.46485466e-01
-1.69289634e-02 8.39159369e-01 6.54617131e-01 -1.02051818e+00
-1.01799548e+00 -1.55515105e-01 7.54276991e-01 6.47372127e-01
6.66032508e-02 -3.64437044e-01 2.82886177e-02 -2.66777903e-01
-1.78727764e-03 -8.69556963e-01 -6.10667109e-01 -3.95500124e-01
1.02455867e-02 -4.27617967e-01 2.09106535e-01 -3.04372668e-01
6.67664528e-01 -1.88361049e+00 5.67411900e-01 2.88013607e-01
3.66053194e-01 3.13830785e-02 -4.51888531e-01 4.08974081e-01
4.46838848e-02 -2.24211231e-01 -7.79136419e-02 5.76204717e-01
2.30387181e-01 3.34886760e-01 1.72398061e-01 2.45646492e-01
-1.98755916e-02 9.20122802e-01 -1.30915856e+00 -1.67632714e-01
1.73403770e-02 -2.94749647e-01 -9.21096981e-01 4.32696491e-01
-5.79415321e-01 3.67306978e-01 -5.32395542e-01 5.05303860e-01
3.50540161e-01 -3.05656672e-01 5.15089214e-01 8.55531216e-01
2.29402184e-02 2.14585334e-01 -1.38414013e+00 1.65638793e+00
-3.91815156e-01 3.78649831e-01 2.91999608e-01 -1.14368927e+00
6.44319594e-01 2.57040679e-01 4.34972137e-01 -1.11105728e+00
1.57674491e-01 2.47458041e-01 4.97028828e-01 -3.80470157e-01
3.90700221e-01 -1.99174181e-01 -2.18738280e-02 7.52433956e-01
-1.05311640e-01 -9.19428110e-01 7.64773309e-01 1.16400540e-01
1.67527735e+00 4.96945046e-02 2.95597166e-01 -1.26663148e-01
3.12210590e-01 6.79598033e-01 6.50916338e-01 1.04461133e+00
-5.71312681e-02 7.84818903e-02 7.18577206e-01 -7.73223579e-01
-1.29397988e+00 -1.09856725e+00 6.30666614e-01 1.12660217e+00
3.04629862e-01 -7.57749602e-02 -1.01790273e+00 -5.58110356e-01
-1.23519637e-01 8.63288999e-01 -4.89772856e-01 -5.65841943e-02
-7.75073886e-01 -2.51755476e-01 6.46333098e-02 3.93427193e-01
5.33330441e-01 -1.64810443e+00 -1.25773799e+00 5.44750631e-01
1.17465951e-01 -8.70376885e-01 -6.65162504e-02 6.66894615e-01
-7.91197300e-01 -1.48178911e+00 -6.15126908e-01 -1.08353293e+00
8.48726153e-01 -6.08264543e-02 1.30599225e+00 3.61009032e-01
-5.79832852e-01 4.61956263e-01 -3.70937407e-01 -2.31971383e-01
-2.40146607e-01 2.82612443e-01 -3.63993049e-02 -1.03308129e+00
2.29328930e-01 -7.06610739e-01 -5.44647872e-01 3.05719972e-01
-5.32436430e-01 1.07815199e-01 4.45085883e-01 1.11909330e+00
1.55683100e-01 6.32218540e-01 3.97020310e-01 -5.54017663e-01
7.69006252e-01 -3.23228896e-01 -1.06476450e+00 5.78583628e-02
-5.15208364e-01 1.84472814e-01 9.68390763e-01 -4.23152417e-01
-6.67977393e-01 1.29607290e-01 3.32050949e-01 -1.40078804e-02
-1.60457358e-01 3.02025855e-01 8.18413049e-02 -2.14757800e-01
8.12088370e-01 3.44451517e-01 -6.45329133e-02 1.03679694e-01
1.08953856e-01 8.17382708e-02 5.09166837e-01 -1.06423163e+00
8.88318837e-01 -1.50703296e-01 -8.47880766e-02 -3.73674840e-01
-5.54679751e-01 4.50517051e-02 -2.43370011e-01 -9.32383463e-02
6.19507432e-01 -6.54411077e-01 -1.58196437e+00 3.42833042e-01
-7.66979992e-01 -1.05741429e+00 -4.62935746e-01 2.69362599e-01
-1.14896727e+00 4.36805338e-02 -5.18220484e-01 -8.35548520e-01
8.51847902e-02 -1.07058692e+00 3.74988467e-01 6.53947890e-01
-2.22815096e-01 -6.76640868e-01 -2.28985585e-02 4.53208178e-01
3.77716780e-01 2.43931681e-01 8.37832212e-01 -4.97681141e-01
-6.88459873e-01 5.60446009e-02 8.25646147e-02 -1.00581661e-01
-5.75499758e-02 -4.77072716e-01 -1.26000538e-01 -4.12353009e-01
-2.48690173e-01 -9.24925745e-01 -1.35970972e-02 1.80964500e-01
1.02473855e+00 -3.16727251e-01 4.13694568e-02 3.93222421e-01
1.18273580e+00 6.24257684e-01 5.95391750e-01 8.21896017e-01
8.69592503e-02 3.97155941e-01 7.32878029e-01 7.99575627e-01
3.87006253e-01 3.33177775e-01 7.44689107e-01 3.08359355e-01
4.66344535e-01 -1.80716619e-01 7.13716224e-02 1.98378056e-01
-3.98468167e-01 -1.59804419e-01 -1.08594859e+00 6.71307206e-01
-1.72053921e+00 -9.56471264e-01 2.35633194e-01 1.81049824e+00
9.90702212e-01 4.99662191e-01 3.78492385e-01 2.96394199e-01
3.55437458e-01 -3.05638403e-01 -8.93825293e-01 -6.96246326e-01
2.38104343e-01 5.77236891e-01 5.39185345e-01 5.75734556e-01
-7.81826079e-01 1.50199175e+00 7.17206192e+00 5.71641147e-01
-5.69704056e-01 -4.63951379e-01 2.65846908e-01 -2.92732745e-01
1.41053051e-02 6.32549599e-02 -2.99362659e-01 1.65396333e-01
8.09629440e-01 -3.39096457e-01 1.40355170e+00 1.00177228e+00
7.79431760e-02 -6.40300810e-01 -1.10488331e+00 8.09699535e-01
-3.76717657e-01 -1.22612870e+00 -5.20821273e-01 9.66948792e-02
8.85190427e-01 -5.12007356e-01 5.94029240e-02 6.59021795e-01
1.29955268e+00 -1.51574528e+00 4.68425602e-01 1.14818275e-01
3.63835305e-01 -1.31348014e+00 4.83143091e-01 8.32798779e-01
-6.56423926e-01 -4.77257729e-01 -3.99186790e-01 -5.75573921e-01
-1.19814180e-01 -6.63615838e-02 -1.23464048e+00 2.44828328e-01
4.62538719e-01 1.54145211e-01 -1.20186545e-01 1.08207524e+00
-6.50493503e-01 2.57885367e-01 -3.91554564e-01 -5.20685375e-01
7.79007971e-01 -2.71600425e-01 2.49432221e-01 4.52902049e-01
3.04585576e-01 9.06719625e-01 3.47073942e-01 7.25550532e-01
2.73262858e-01 -1.94738373e-01 -3.47449452e-01 -3.18382233e-01
3.28087866e-01 8.74123275e-01 -7.76627600e-01 -1.86044380e-01
1.78487152e-01 1.06897306e+00 4.73511785e-01 2.64004797e-01
-8.16361427e-01 -5.57614625e-01 4.68298256e-01 3.85160409e-02
5.40005147e-01 -4.43038881e-01 -5.74056387e-01 -7.61193991e-01
-1.83437526e-01 -1.50561357e+00 2.33778685e-01 -8.66423428e-01
-8.89358640e-01 4.24775213e-01 -3.78959984e-01 -8.14244032e-01
-8.26647699e-01 -7.96852767e-01 -5.99893153e-01 5.77834308e-01
-1.21466684e+00 -4.13938731e-01 -2.97086656e-01 8.17866385e-01
3.96471560e-01 -7.32419908e-01 7.64421463e-01 -2.88670421e-01
-2.68849313e-01 4.42931622e-01 -8.35432336e-02 2.29877997e-02
9.72310528e-02 -1.68963659e+00 2.00334147e-01 1.99632317e-01
-2.28755265e-01 2.74362326e-01 1.13575554e+00 -3.74180585e-01
-1.69576800e+00 -2.38724321e-01 2.41919294e-01 -1.31261408e-01
6.49426222e-01 -9.40610282e-03 -5.49026310e-01 4.63128924e-01
3.19080859e-01 -1.60939053e-01 2.45047718e-01 2.63785899e-01
-3.83369587e-02 -4.84370477e-02 -1.29338014e+00 6.83467627e-01
1.05285490e+00 -1.24447430e-02 -9.38081324e-01 3.56843740e-01
2.74170876e-01 -7.89430976e-01 -6.46624804e-01 -6.00025915e-02
2.51698583e-01 -9.22664702e-01 8.55414391e-01 -9.20618534e-01
7.14830995e-01 -2.70736188e-01 1.87133923e-01 -1.56541717e+00
-3.13612729e-01 -7.72770703e-01 -1.35491952e-01 3.15366477e-01
3.21858287e-01 -4.17067260e-01 1.34911215e+00 7.33440220e-01
1.82183981e-01 -5.17054617e-01 -8.53376865e-01 -8.97609174e-01
4.15364683e-01 -4.38511781e-02 4.98590082e-01 9.99192953e-01
5.97764373e-01 1.80699483e-01 -2.23385841e-01 1.26810357e-01
7.13774741e-01 2.50339448e-01 9.56593454e-01 -1.01373219e+00
-4.98153031e-01 -4.97875363e-01 -1.66150898e-01 -8.75443757e-01
2.72666216e-01 -6.52153134e-01 7.72253796e-02 -1.51658762e+00
-6.47232682e-02 -4.61006254e-01 -5.92064746e-02 6.31306827e-01
2.51185328e-01 -1.83225229e-01 2.59932429e-01 -4.42099065e-01
-9.72492158e-01 2.58738250e-01 1.72206140e+00 -1.59040079e-01
-3.01032841e-01 -1.51082352e-01 -7.91480541e-01 7.19647169e-01
8.76775980e-01 -4.20691609e-01 -4.07522231e-01 -2.95010269e-01
6.38293922e-01 7.57649660e-01 1.18920647e-01 -1.24125516e+00
4.61175531e-01 -8.98348212e-01 4.93766874e-01 -2.21025601e-01
9.77337360e-02 -9.37683582e-01 4.28027809e-02 8.24036479e-01
-1.93670556e-01 2.42522061e-01 2.05977261e-01 2.14223459e-01
-3.61821726e-02 -5.50464511e-01 6.90922737e-01 -6.95952892e-01
-7.85986423e-01 -8.56235325e-02 -8.14482152e-01 3.16764206e-01
1.61299205e+00 -2.43651077e-01 -5.03638834e-02 -5.54618001e-01
-9.28897798e-01 8.47716153e-01 5.12227356e-01 -3.24235447e-02
5.26796997e-01 -1.04484785e+00 -5.95537364e-01 -7.61187673e-02
-3.75280201e-01 1.72901884e-01 2.03010872e-01 2.27229834e-01
-1.04959404e+00 3.19872171e-01 -7.44909465e-01 -3.65297832e-02
-9.75963831e-01 7.36991882e-01 4.67635036e-01 -6.02528811e-01
-6.45945787e-01 7.76591063e-01 3.09943203e-02 -4.96439606e-01
2.77689815e-01 -5.40171377e-02 1.51276879e-03 -2.40814403e-01
5.75295389e-01 2.22249866e-01 -1.90128475e-01 4.74439919e-01
-2.58386761e-01 1.60912007e-01 -2.55188644e-01 -2.77664453e-01
1.66232169e+00 2.64264345e-01 2.26667166e-01 -2.44076952e-01
3.61694306e-01 -4.99941468e-01 -1.47787118e+00 1.64937735e-01
-4.88892756e-02 -3.88896704e-01 -1.63940549e-01 -1.25128782e+00
-8.78772199e-01 6.87976122e-01 2.07681209e-01 2.92321205e-01
1.15265477e+00 -4.86433983e-01 6.62330866e-01 1.16124606e+00
8.93848360e-01 -1.59052420e+00 3.94182175e-01 8.60460460e-01
6.97766900e-01 -1.23267543e+00 9.36504528e-02 1.23403203e-02
-9.26021218e-01 1.20846343e+00 1.04127860e+00 -5.99272430e-01
-1.11171767e-01 4.32456583e-01 -2.65860885e-01 -1.18201718e-01
-9.77938175e-01 -2.08647996e-01 -2.90762335e-01 7.46481001e-01
-9.98815745e-02 2.52248341e-04 -2.64557570e-01 3.89672339e-01
-7.30643570e-01 7.82719478e-02 8.05043042e-01 1.17370641e+00
-9.51134562e-01 -9.99540269e-01 -6.28504515e-01 1.34053811e-01
-5.15013151e-02 1.79828852e-01 -3.27854395e-01 8.80017102e-01
-4.07559238e-02 7.78318882e-01 2.53004022e-02 9.95672271e-02
1.98632538e-01 -1.43032551e-01 9.95040059e-01 -4.83387232e-01
-7.09825099e-01 -2.01982811e-01 5.35302386e-02 -6.35329306e-01
7.78677538e-02 -5.16050637e-01 -1.67506576e+00 -8.11994791e-01
2.06333354e-01 4.53932971e-01 3.78185749e-01 1.03352380e+00
-1.68278784e-01 5.60413003e-01 6.02095187e-01 -7.39106238e-01
-6.45861149e-01 -3.37764233e-01 -8.19321513e-01 -7.20506236e-02
1.47639588e-01 -6.16474569e-01 -2.77709305e-01 -2.55987108e-01]
|
[3.7382218837738037, 1.5492373704910278]
|
bcc8fb90-7845-48c2-b96a-c971778b96ea
|
long-range-graph-benchmark
|
2206.08164
| null |
https://arxiv.org/abs/2206.08164v3
|
https://arxiv.org/pdf/2206.08164v3.pdf
|
Long Range Graph Benchmark
|
Graph Neural Networks (GNNs) that are based on the message passing (MP) paradigm generally exchange information between 1-hop neighbors to build node representations at each layer. In principle, such networks are not able to capture long-range interactions (LRI) that may be desired or necessary for learning a given task on graphs. Recently, there has been an increasing interest in development of Transformer-based methods for graphs that can consider full node connectivity beyond the original sparse structure, thus enabling the modeling of LRI. However, MP-GNNs that simply rely on 1-hop message passing often fare better in several existing graph benchmarks when combined with positional feature representations, among other innovations, hence limiting the perceived utility and ranking of Transformer-like architectures. Here, we present the Long Range Graph Benchmark (LRGB) with 5 graph learning datasets: PascalVOC-SP, COCO-SP, PCQM-Contact, Peptides-func and Peptides-struct that arguably require LRI reasoning to achieve strong performance in a given task. We benchmark both baseline GNNs and Graph Transformer networks to verify that the models which capture long-range dependencies perform significantly better on these tasks. Therefore, these datasets are suitable for benchmarking and exploration of MP-GNNs and Graph Transformer architectures that are intended to capture LRI.
|
['Dominique Beaini', 'Anh Tuan Luu', 'Guy Wolf', 'Ali Parviz', 'Mikhail Galkin', 'Ladislav Rampášek', 'Vijay Prakash Dwivedi']
|
2022-06-16
| null | null | null | null |
['graph-regression']
|
['graphs']
|
[ 2.92504340e-01 5.01362979e-01 -2.09941670e-01 -3.65293115e-01
-7.56931454e-02 -5.73356926e-01 9.90597129e-01 7.01650202e-01
-1.46330073e-01 7.69556582e-01 -1.46502657e-02 -7.28795946e-01
-5.28914869e-01 -1.27493656e+00 -1.25260222e+00 -5.29389679e-01
-8.18182707e-01 9.10694182e-01 4.95528251e-01 -4.87009138e-01
-9.51051340e-02 5.74763596e-01 -1.17805707e+00 4.60115075e-01
5.99019587e-01 5.84308386e-01 4.32044715e-02 7.91016400e-01
-1.41572878e-01 1.01629019e+00 -3.20675373e-01 -4.61042583e-01
7.27212355e-02 -2.04569966e-01 -1.04601991e+00 -6.61522686e-01
5.45853972e-01 2.79944003e-01 -5.18366992e-01 6.85181975e-01
3.88722539e-01 2.28651296e-02 6.55359924e-01 -1.27288711e+00
-4.78682756e-01 1.29358542e+00 -3.15871894e-01 1.37876987e-01
4.24108952e-01 2.14052156e-01 1.63062882e+00 -2.72693515e-01
1.06729901e+00 1.48640203e+00 1.04396439e+00 3.52059901e-01
-1.49224460e+00 -4.19078678e-01 3.92547816e-01 1.09565273e-01
-1.07497633e+00 7.27585554e-02 6.91754937e-01 -2.78653111e-02
1.74318981e+00 2.99841642e-01 8.30682039e-01 1.29168892e+00
7.36850321e-01 4.95721340e-01 7.21597791e-01 9.15286541e-02
-2.40322910e-02 -4.94628847e-01 4.05452758e-01 8.96880388e-01
4.92332071e-01 6.16170689e-02 -6.04956448e-01 -1.90041155e-01
5.96711695e-01 -5.93136959e-02 -3.19913208e-01 -5.90584576e-01
-1.14434063e+00 9.09264803e-01 1.31399262e+00 5.37920177e-01
-3.45927864e-01 5.32064199e-01 6.30299389e-01 7.35133111e-01
4.26515460e-01 7.83888102e-01 -6.27039194e-01 1.57766193e-01
-4.30662602e-01 3.20164442e-01 1.21627665e+00 7.82699406e-01
7.69820035e-01 -1.97002932e-01 1.02785444e-02 4.77077872e-01
2.50093281e-01 -1.50777310e-01 2.67664008e-02 -3.24889332e-01
4.24132943e-01 1.03066862e+00 -6.37926698e-01 -1.27653980e+00
-9.07929718e-01 -7.63313234e-01 -1.02979481e+00 -1.46113217e-01
3.63774389e-01 2.69437641e-01 -9.74229157e-01 1.77898645e+00
1.41787469e-01 1.58981815e-01 6.17450774e-02 4.96705532e-01
1.40394187e+00 5.14084041e-01 1.47532642e-01 2.53013939e-01
9.80080724e-01 -7.25297034e-01 -6.87733665e-02 -3.80501032e-01
1.23133802e+00 -1.47992104e-01 7.68409729e-01 4.24581707e-01
-8.91262889e-01 -3.82473290e-01 -1.00996411e+00 -1.64602041e-01
-7.49756992e-01 -7.51923025e-01 1.24913502e+00 2.90480524e-01
-1.55741644e+00 1.20924294e+00 -6.91434503e-01 -6.86817825e-01
3.57767582e-01 6.49231136e-01 -6.51377022e-01 -2.14291140e-01
-1.39058411e+00 8.92563403e-01 5.78937769e-01 6.92660362e-02
-9.74395037e-01 -8.39386761e-01 -8.09075773e-01 2.58485794e-01
1.04466103e-01 -5.74009717e-01 5.01821399e-01 -7.01821148e-01
-8.20307791e-01 8.66639972e-01 2.83374786e-01 -9.79172230e-01
3.11068952e-01 3.30280453e-01 -9.33263451e-02 1.18693605e-01
-2.18413174e-01 1.06802726e+00 3.55487555e-01 -8.82946491e-01
-1.15390874e-01 -3.61901551e-01 5.80751121e-01 1.33184060e-01
-1.56979804e-05 -3.69357586e-01 -2.00185105e-01 -1.88675359e-01
4.79699560e-02 -1.04159009e+00 -2.49619648e-01 -2.86522120e-01
-7.01971054e-01 -5.70074320e-01 7.61706293e-01 -1.01073317e-01
7.49239147e-01 -1.67005932e+00 4.22195733e-01 5.45125604e-01
7.53349066e-01 2.57319450e-01 -6.59373224e-01 9.60142016e-01
-4.79674608e-01 3.49177331e-01 7.88707286e-03 -2.88104191e-02
-1.66644871e-01 5.41649878e-01 -7.82407969e-02 4.62232023e-01
1.79367527e-01 1.43813896e+00 -8.11737955e-01 -1.93027005e-01
4.89657111e-02 6.87367976e-01 -5.87368131e-01 -3.98467146e-02
-6.87636554e-01 2.43218899e-01 -1.70843855e-01 3.21966141e-01
5.61951458e-01 -8.25920820e-01 7.11895525e-01 -2.97502548e-01
4.21157360e-01 6.00362599e-01 -5.92946053e-01 1.45090866e+00
-1.42864376e-01 5.10656238e-01 -2.08945153e-03 -1.34467077e+00
8.64562154e-01 -8.55676755e-02 5.65247297e-01 -7.63208807e-01
-1.46077096e-01 -5.37308417e-02 5.37569106e-01 2.49764115e-01
1.24264061e-01 3.45718503e-01 2.32139423e-01 2.20640600e-01
3.33565801e-01 6.16700016e-02 4.76983249e-01 7.88904130e-01
1.89209759e+00 -1.98893063e-02 3.93362530e-02 -5.55656552e-01
3.28768432e-01 -7.00872913e-02 1.92979455e-01 9.55221534e-01
2.96869904e-01 2.90600419e-01 1.15857053e+00 -8.09383929e-01
-6.94824219e-01 -8.16542566e-01 1.37316018e-01 1.47916341e+00
-7.11660013e-02 -9.39216971e-01 -3.39341730e-01 -8.97326171e-01
1.11741923e-01 4.47541475e-01 -7.35391021e-01 -2.95546502e-01
-4.75811481e-01 -8.78014386e-01 8.55961144e-01 3.24896693e-01
1.66741759e-02 -1.13516057e+00 -1.21672481e-01 5.21390736e-01
1.29347265e-01 -9.61249411e-01 5.46601601e-02 6.10109389e-01
-7.90669143e-01 -1.52234435e+00 -2.91872650e-01 -6.08458698e-01
6.55860484e-01 1.23760909e-01 1.63450289e+00 6.04334295e-01
-2.69092649e-01 3.77446353e-01 -4.73187089e-01 -8.38627815e-02
-4.10771638e-01 7.01757193e-01 -4.33709919e-01 -4.96144295e-01
2.34762698e-01 -9.63474035e-01 -4.45223302e-01 1.81797683e-01
-8.24516654e-01 1.66113496e-01 6.82003677e-01 9.00516152e-01
3.55980664e-01 -4.85171452e-02 3.47114861e-01 -1.41140890e+00
6.88243270e-01 -5.86678028e-01 -5.27190983e-01 3.56905311e-01
-6.41384125e-01 2.96216279e-01 7.99669385e-01 -2.12322295e-01
-4.37353671e-01 -2.51337439e-01 -3.66219580e-01 -3.18567976e-02
1.22367397e-01 1.02311456e+00 1.20564409e-01 -5.80522716e-01
9.02084887e-01 7.19752386e-02 1.24313816e-01 -2.16237858e-01
3.86795163e-01 -1.59647629e-01 -4.31677364e-02 -6.90316677e-01
5.46707511e-01 2.22219706e-01 7.07874417e-01 -8.65879714e-01
-6.72116280e-01 -2.17537090e-01 -3.37525189e-01 1.34584069e-01
7.55978763e-01 -7.12567568e-01 -1.14327347e+00 4.29182500e-01
-1.06382298e+00 -6.45121396e-01 -4.20352351e-03 3.05312593e-02
-3.12598139e-01 2.69463390e-01 -1.00048912e+00 -2.97517926e-01
-4.78695393e-01 -1.16413355e+00 1.04397547e+00 -1.60241202e-01
-1.66410848e-01 -1.45008421e+00 1.41610540e-02 -2.13353653e-02
6.35400593e-01 5.07010281e-01 1.42457008e+00 -1.12818623e+00
-8.09924424e-01 8.90052244e-02 -4.05173689e-01 -2.11790845e-01
-1.46071821e-01 -1.78414758e-03 -6.69045448e-01 -5.87579370e-01
-8.88534546e-01 -6.24588668e-01 1.21495938e+00 2.93943077e-01
9.78042245e-01 -3.83099556e-01 -6.50279999e-01 6.58771098e-01
1.29989028e+00 -4.66659039e-01 7.46825159e-01 3.40943635e-02
1.05932987e+00 6.21314466e-01 -5.12892269e-02 -2.09027261e-01
6.16670132e-01 4.29524869e-01 9.41963732e-01 -1.32261783e-01
-1.29648596e-01 -5.80033243e-01 2.54355401e-01 7.16967523e-01
-1.12999350e-01 -6.31270826e-01 -1.11368394e+00 2.32773364e-01
-1.88937986e+00 -6.67743981e-01 -5.47908723e-01 1.79614449e+00
6.34935677e-01 5.21547794e-01 -1.63033269e-02 -3.33067290e-02
3.17108333e-01 5.07486463e-01 -5.62911272e-01 -4.09576565e-01
-3.15506220e-01 4.78302598e-01 6.88763142e-01 3.63808602e-01
-7.75941730e-01 1.02351952e+00 6.09500551e+00 6.15081549e-01
-1.12146556e+00 -3.13125670e-01 7.33085275e-01 3.35544586e-01
-5.95037818e-01 2.24575579e-01 -7.44039893e-01 1.07162434e-03
1.37618768e+00 1.40105933e-01 6.84191108e-01 6.37016535e-01
-4.11117464e-01 3.17322373e-01 -1.48900497e+00 6.19580746e-01
-3.07613790e-01 -1.84373045e+00 3.43886733e-01 2.53337085e-01
3.96766275e-01 7.83939242e-01 -1.71336696e-01 6.36757553e-01
8.70473266e-01 -1.54785573e+00 5.75866513e-02 3.81964743e-01
5.09767592e-01 -5.91940880e-01 6.72911465e-01 2.89165020e-01
-1.33274508e+00 1.97049141e-01 -6.39661193e-01 -5.38307130e-02
-3.62329394e-01 6.65091693e-01 -1.11980903e+00 8.61582696e-01
5.38575053e-01 1.13147831e+00 -1.02951920e+00 6.25651777e-01
-1.84288219e-01 6.13567472e-01 -4.81052697e-01 -2.04604015e-01
6.58894777e-01 -1.90612540e-01 4.16503966e-01 1.05836284e+00
6.14996515e-02 -3.21945250e-01 2.91233838e-01 7.86906064e-01
-4.01330739e-01 2.10234076e-02 -1.03199005e+00 -4.66118664e-01
2.22675323e-01 1.22858059e+00 -8.80947173e-01 -7.06679970e-02
-3.94613266e-01 5.12010276e-01 9.20513153e-01 2.19820738e-01
-7.16084540e-01 -1.26883999e-01 5.51563323e-01 3.75996709e-01
3.10504913e-01 -1.85208112e-01 3.95856589e-01 -8.28746617e-01
-3.53211552e-01 -1.16339457e+00 6.50137067e-01 -6.54466391e-01
-1.55645394e+00 6.15513504e-01 -2.37916276e-01 -3.89664561e-01
-1.72337219e-01 -8.72984767e-01 -6.84834599e-01 5.42886376e-01
-1.50129855e+00 -1.57640541e+00 -1.60010219e-01 6.25982046e-01
-4.40263450e-02 1.44258976e-01 8.42127621e-01 3.63165028e-02
-1.56263143e-01 4.85296041e-01 -2.50325382e-01 9.00036916e-02
3.47620785e-01 -1.37381613e+00 8.42383087e-01 3.83796215e-01
4.00338203e-01 9.37931836e-01 6.55059338e-01 -8.35678458e-01
-1.82949424e+00 -1.39054632e+00 7.11132765e-01 -3.80896956e-01
7.53903985e-01 -7.77821124e-01 -1.12042582e+00 1.05235946e+00
8.38817507e-02 2.52817154e-01 2.23390177e-01 5.90907514e-01
-6.60648286e-01 -9.93914604e-02 -9.30857837e-01 3.67681533e-01
1.74394369e+00 -4.89183426e-01 -2.26708859e-01 7.95984805e-01
9.43331957e-01 -3.96314323e-01 -1.27355587e+00 5.19414008e-01
2.01566502e-01 -1.09119272e+00 1.07977903e+00 -9.06742692e-01
2.06229568e-01 -1.74095910e-02 9.70790982e-02 -1.46016741e+00
-4.18175727e-01 -6.58629119e-01 7.51700178e-02 8.30630183e-01
7.63809204e-01 -1.05717134e+00 8.58513474e-01 1.74547166e-01
-1.77211147e-02 -7.84491062e-01 -8.87794733e-01 -6.79319203e-01
1.71241105e-01 -2.40288004e-01 6.81310713e-01 9.68783379e-01
-8.20118636e-02 7.86577821e-01 -9.83128101e-02 -5.65290228e-02
6.97676063e-01 -3.66542339e-02 8.60351384e-01 -1.68421149e+00
-4.18481886e-01 -5.59261441e-01 -7.67575622e-01 -9.86416698e-01
4.92267996e-01 -1.45587528e+00 -3.31873119e-01 -1.89392543e+00
8.86637047e-02 -5.92848003e-01 -3.40682060e-01 6.59532726e-01
7.21160844e-02 9.57158767e-03 -1.33144185e-01 -4.84390184e-03
-8.09942007e-01 3.41124684e-01 1.14029133e+00 -3.25872779e-01
1.11549519e-01 -3.33907038e-01 -7.60102093e-01 4.76149648e-01
5.81466198e-01 -4.60806847e-01 -6.52559459e-01 -3.37517560e-01
1.05383480e+00 -8.43305700e-03 4.36700910e-01 -8.35466683e-01
3.79776359e-01 1.90547943e-01 3.61602843e-01 -4.43872541e-01
3.25839072e-01 -6.38430655e-01 4.67810035e-01 6.58960879e-01
-5.95060050e-01 1.40923917e-01 -5.36345877e-02 7.97470331e-01
1.36619178e-03 2.46899247e-01 4.87283736e-01 -4.26209837e-01
-4.96409595e-01 7.04817116e-01 2.41125040e-02 9.71680656e-02
7.12275743e-01 -5.80065474e-02 -7.61053741e-01 -4.27563339e-01
-6.92104936e-01 3.78375143e-01 4.60473508e-01 3.71689439e-01
5.68233728e-01 -8.27282190e-01 -7.77796805e-01 9.55526680e-02
3.05078208e-01 -1.92945655e-02 -4.52319048e-02 7.12698817e-01
-6.99937701e-01 5.96849084e-01 -2.34351814e-01 -7.33359337e-01
-1.13872123e+00 5.45869708e-01 4.47890848e-01 -9.37487185e-01
-7.94646442e-01 1.01800418e+00 3.69949818e-01 -1.08788729e+00
1.59653034e-02 -4.25975055e-01 -1.93054140e-01 -2.64044464e-01
-4.79791015e-02 -3.56500521e-02 2.66392648e-01 -4.92014825e-01
-5.04016042e-01 1.50379151e-01 -2.79492259e-01 6.75003946e-01
1.55677211e+00 3.99599463e-01 -5.63584149e-01 1.50765210e-01
1.22812963e+00 -4.08461034e-01 -7.36370564e-01 -1.86256766e-01
2.38161296e-01 7.75486082e-02 -2.42602333e-01 -7.33460307e-01
-1.06966138e+00 8.27194691e-01 8.88224989e-02 4.88274723e-01
5.24864495e-01 1.61218718e-01 4.30634439e-01 7.84101903e-01
6.21839046e-01 -4.56543356e-01 -1.77080333e-02 7.94980407e-01
9.88159180e-01 -8.59196186e-01 9.89216380e-03 -5.27288198e-01
-2.52592236e-01 9.75447118e-01 6.93912625e-01 -3.05765897e-01
7.17335284e-01 1.72455683e-02 -5.63521087e-01 -7.84690559e-01
-1.32766533e+00 -1.59751520e-01 3.16887558e-01 7.89743781e-01
5.60379267e-01 6.32085875e-02 -5.70707880e-02 -1.86213441e-02
-1.84321225e-01 -4.82122421e-01 3.89827430e-01 5.88222861e-01
-2.86947310e-01 -1.25618708e+00 4.67648447e-01 8.98389995e-01
-3.43271762e-01 -3.67951989e-01 -9.86182630e-01 1.00701344e+00
-2.90887088e-01 7.50652969e-01 -1.27536029e-01 -3.23600471e-01
1.55747116e-01 -1.15475915e-01 6.08483434e-01 -6.91013098e-01
-9.10445690e-01 -3.42374027e-01 6.28813088e-01 -9.15030003e-01
-2.90255487e-01 -6.76490441e-02 -1.31446278e+00 -7.38598228e-01
-2.65547663e-01 2.61881232e-01 3.54965448e-01 7.06416726e-01
5.78043163e-01 7.86751747e-01 -7.67529234e-02 -6.62602067e-01
-3.62317741e-01 -1.02838731e+00 -4.96170163e-01 5.28326154e-01
3.35663445e-02 -4.08075273e-01 -3.12110692e-01 -7.63934135e-01]
|
[6.888023853302002, 6.205175399780273]
|
b9428949-09cf-4a78-9071-afa30e395baa
|
a-novel-1d-state-space-for-efficient-music
|
2111.00704
| null |
https://arxiv.org/abs/2111.00704v2
|
https://arxiv.org/pdf/2111.00704v2.pdf
|
A Novel 1D State Space for Efficient Music Rhythmic Analysis
|
Inferring music time structures has a broad range of applications in music production, processing and analysis. Scholars have proposed various methods to analyze different aspects of time structures, such as beat, downbeat, tempo and meter. Many state-of-the-art (SOFA) methods, however, are computationally expensive. This makes them inapplicable in real-world industrial settings where the scale of the music collections can be millions. This paper proposes a new state space and a semi-Markov model for music time structure analysis. The proposed approach turns the commonly used 2D state spaces into a 1D model through a jump-back reward strategy. It reduces the state spaces size drastically. We then utilize the proposed method for causal, joint beat, downbeat, tempo, and meter tracking, and compare it against several previous methods. The proposed method delivers similar performance with the SOFA joint causal models with a much smaller state space and a more than 30 times speedup.
|
['Zhiyao Duan', 'Andreas Ehmann', 'Matthew McCallum', 'Mojtaba Heydari']
|
2021-11-01
| null | null | null | null |
['inference-optimization']
|
['audio']
|
[-5.10459905e-03 -5.64480484e-01 -3.01738113e-01 2.06716999e-01
-5.88089943e-01 -8.86250198e-01 4.33200389e-01 -3.62048894e-02
8.52780975e-03 7.07764626e-01 1.39004067e-01 -2.85279602e-01
-5.42065442e-01 -5.17543554e-01 -1.84257075e-01 -6.18729353e-01
-8.13156590e-02 4.78267968e-01 3.89977127e-01 -1.82352021e-01
6.22118831e-01 1.54464439e-01 -1.43579161e+00 -2.01765716e-01
4.45996940e-01 7.56799221e-01 2.12598771e-01 1.04235065e+00
5.71343154e-02 6.78832293e-01 -8.17111731e-01 3.24364826e-02
1.81217015e-01 -8.81148100e-01 -6.84800208e-01 -1.96563765e-01
-1.06842704e-01 -1.93583563e-01 -3.71381998e-01 8.35955262e-01
6.27886295e-01 2.33557299e-01 2.46638879e-01 -1.34200501e+00
1.33428365e-01 1.06145883e+00 -7.70045519e-01 1.84217677e-01
6.06446803e-01 -1.92258000e-01 9.05752182e-01 -6.32808208e-02
2.19666258e-01 1.15742934e+00 9.23371673e-01 2.09969640e-01
-1.06111407e+00 -1.01353872e+00 8.48277882e-02 2.92246878e-01
-1.31484270e+00 -2.78512090e-01 9.03221071e-01 -3.58617961e-01
7.44062006e-01 4.83787894e-01 9.85138237e-01 8.14392388e-01
2.18000695e-01 8.05186749e-01 1.23566830e+00 -4.38493937e-01
1.94616273e-01 -5.34721732e-01 -9.93632711e-03 3.92531872e-01
1.37049839e-01 -6.54984117e-02 -8.86111319e-01 -4.82032627e-01
1.05748272e+00 -2.49794889e-02 7.95577392e-02 4.66183871e-02
-1.60890579e+00 4.23965931e-01 -3.66320878e-01 1.45695657e-01
-1.83296457e-01 5.49568236e-01 4.03047562e-01 2.09668517e-01
-9.80156511e-02 4.19444948e-01 -3.36364210e-01 -1.01025939e+00
-1.31839490e+00 5.41695714e-01 9.50597346e-01 9.55174208e-01
2.04979837e-01 3.17359149e-01 -1.85138285e-01 4.65537161e-01
3.49228233e-01 5.13993561e-01 4.85355169e-01 -1.08388340e+00
3.05610687e-01 2.41540611e-01 4.31573331e-01 -1.00995076e+00
-3.55697900e-01 -4.45740968e-01 -9.23865378e-01 -4.96472418e-02
5.73463023e-01 -4.69152397e-03 -5.30780137e-01 1.76842070e+00
5.16836762e-01 7.62975752e-01 -2.19461709e-01 8.76024127e-01
1.84456259e-01 7.97427297e-01 -4.57041115e-01 -7.49766350e-01
1.27226877e+00 -8.29333484e-01 -1.11964750e+00 1.84339508e-01
1.17544733e-01 -1.38300002e+00 9.01657939e-01 9.70397234e-01
-1.23747206e+00 -5.66516101e-01 -1.06910372e+00 3.02510709e-01
3.62027168e-01 2.45343402e-01 8.35913062e-01 6.19852245e-01
-5.03991008e-01 1.07774270e+00 -1.11649728e+00 -5.99699169e-02
-2.77043879e-01 4.35827017e-01 3.60126764e-01 5.17266989e-01
-1.07519436e+00 4.31131989e-01 3.89959157e-01 1.71240374e-01
-1.05789793e+00 -4.23084974e-01 -2.74710655e-01 -1.32978559e-01
4.30278450e-01 -5.67128897e-01 1.68125737e+00 -3.81417781e-01
-1.98773098e+00 2.43748352e-01 -6.44663572e-02 -2.54833788e-01
3.34486812e-01 -4.79139119e-01 -4.71229762e-01 -2.41661310e-01
-1.11937203e-01 -2.53477424e-01 1.01490617e+00 -5.01407266e-01
-5.86907387e-01 -1.97190776e-01 4.20305170e-02 2.05311760e-01
-4.85406965e-02 8.63933638e-02 -4.35153365e-01 -9.75505352e-01
4.40472633e-01 -1.27381945e+00 -3.25507522e-01 -6.38782620e-01
-6.06582582e-01 -6.32789284e-02 5.59924185e-01 -4.01288301e-01
2.11407900e+00 -2.04166627e+00 2.98557103e-01 3.29855680e-02
-1.27601296e-01 -4.75719497e-02 2.58059621e-01 8.69830191e-01
2.96859667e-02 -1.74337909e-01 6.75648302e-02 -1.06766410e-01
1.46469295e-01 1.62700072e-01 -5.22696495e-01 3.40759903e-01
-5.52202225e-01 3.62052053e-01 -1.02042615e+00 -4.56453532e-01
1.31202117e-01 9.29396152e-02 -5.76923728e-01 1.73861966e-01
-1.67266488e-01 5.24509311e-01 -6.13213718e-01 5.38071334e-01
3.29221904e-01 -8.96067321e-02 5.35203576e-01 -4.42858078e-02
-4.58124399e-01 6.56147599e-01 -1.89815986e+00 2.11587286e+00
-2.86970019e-01 3.26097429e-01 -1.34646893e-01 -5.79455853e-01
9.92597401e-01 5.61102271e-01 8.07397902e-01 -8.01952407e-02
2.10086063e-01 1.55246317e-01 2.34147295e-01 -2.44295910e-01
6.73118532e-01 -3.09675962e-01 -4.54243273e-01 7.30380118e-01
-1.57733127e-01 -2.49451235e-01 4.07579035e-01 1.90141965e-02
1.31996024e+00 4.58551258e-01 5.92426240e-01 -8.93891975e-02
2.66536981e-01 -9.79399532e-02 9.95669842e-01 5.25451839e-01
-2.52768129e-01 2.98257887e-01 6.33614242e-01 -2.25972861e-01
-9.31046069e-01 -9.71919894e-01 3.85092705e-01 9.91035044e-01
2.76985317e-01 -1.08007741e+00 -5.07079542e-01 -1.86841041e-01
-8.24834779e-02 4.41761911e-01 -1.87214732e-01 -1.37158334e-01
-6.14746094e-01 -5.45113027e-01 7.73070812e-01 4.69453543e-01
3.27683657e-01 -9.30528402e-01 -7.84977734e-01 6.03658795e-01
-3.44536811e-01 -7.46959269e-01 -7.04621017e-01 8.35219678e-03
-1.30559611e+00 -8.39202106e-01 -4.42326933e-01 -4.67001110e-01
-2.11184978e-01 7.12964907e-02 9.38202083e-01 -4.59452540e-01
-2.59828418e-01 1.94908783e-01 -2.91844249e-01 -5.79851568e-01
-2.72390574e-01 -2.63563432e-02 5.10237336e-01 -7.09830374e-02
-1.67513981e-01 -9.79072154e-01 -7.08918810e-01 5.29350460e-01
-5.13439655e-01 2.14218959e-01 4.48863328e-01 5.83185375e-01
8.03507388e-01 3.57087523e-01 6.21104717e-01 -6.70157194e-01
7.31348932e-01 -1.76495790e-01 -7.87003279e-01 -1.72191449e-02
-8.44976366e-01 3.18376869e-02 4.79210556e-01 -7.39188313e-01
-8.41630101e-01 3.09301019e-01 1.80279613e-01 -6.31126642e-01
1.27944052e-01 4.04338777e-01 8.25271085e-02 4.52080131e-01
2.15389386e-01 1.92015186e-01 -2.50032037e-01 -6.80532157e-01
2.31231496e-01 6.04338527e-01 6.54344082e-01 -7.56491542e-01
6.80518746e-01 3.22741807e-01 3.58329505e-01 -4.75103289e-01
-8.85789454e-01 -8.07769775e-01 -5.30797660e-01 -2.49831527e-01
5.38730681e-01 -7.91302085e-01 -1.12570739e+00 4.33898896e-01
-9.62070882e-01 -1.74279094e-01 -2.61896938e-01 9.10831094e-01
-1.09722149e+00 4.52271551e-01 -7.65246451e-01 -1.22009623e+00
-2.73185134e-01 -8.80014122e-01 9.65124488e-01 1.65490448e-01
-6.09148562e-01 -5.52365124e-01 5.85380852e-01 1.34359449e-01
-1.28508046e-01 2.90839553e-01 7.41687179e-01 -1.09337032e-01
-5.46305239e-01 -1.79577544e-01 3.65769774e-01 -1.58273861e-01
1.86443195e-01 6.54551461e-02 -6.06540024e-01 -1.74077034e-01
1.88428315e-03 2.97462363e-02 3.25502455e-01 4.38017130e-01
9.57469165e-01 -2.61598438e-01 7.82799255e-03 3.87809098e-01
1.13896763e+00 4.84691828e-01 5.18997908e-01 2.10215405e-01
5.48885465e-01 3.38586457e-02 1.17732847e+00 1.13768816e+00
1.28973931e-01 8.95588279e-01 3.80229473e-01 3.92616034e-01
-1.01496615e-01 -5.13177991e-01 5.76176345e-01 1.73860395e+00
-5.63694358e-01 -3.70009057e-02 -7.62489438e-01 6.06485009e-01
-2.33291340e+00 -1.29842710e+00 -3.75451744e-01 2.37898493e+00
1.00826609e+00 4.91091572e-02 4.54447120e-01 7.09663570e-01
7.19258845e-01 1.42356575e-01 -5.66849053e-01 -4.52824712e-01
2.98070967e-01 2.71270484e-01 3.42474282e-01 7.95860663e-02
-9.78457272e-01 7.41077840e-01 7.10833693e+00 9.15138185e-01
-7.81461358e-01 5.07773869e-02 -1.75921872e-01 -3.06760043e-01
7.17783272e-02 4.02230918e-01 -7.56691337e-01 3.92884165e-01
1.02400899e+00 -3.69959354e-01 7.62723804e-01 6.53699577e-01
6.31202996e-01 -6.99106976e-02 -1.02145791e+00 1.46691477e+00
-2.59124905e-01 -1.06353462e+00 -1.87769696e-01 1.33615397e-02
8.13845575e-01 -5.06776869e-01 -2.09693417e-01 1.76480591e-01
1.73199445e-01 -6.24604225e-01 9.84667242e-01 6.75238609e-01
8.01541686e-01 -8.89107168e-01 3.24065417e-01 5.39120376e-01
-1.64242315e+00 -1.52778015e-01 -7.59315565e-02 -6.27155781e-01
5.88224888e-01 5.06941378e-01 -6.43733263e-01 6.51971698e-01
5.25641084e-01 8.17261279e-01 5.50993048e-02 1.13942599e+00
-2.53984392e-01 1.00182772e+00 -3.98099959e-01 -2.02109367e-01
9.41467881e-02 -2.58323401e-01 8.66343677e-01 8.46095145e-01
6.22267544e-01 1.56970739e-01 3.72985125e-01 5.40876746e-01
4.32064891e-01 -5.63107729e-02 -2.79133618e-01 -4.07699734e-01
5.76946139e-01 9.92543221e-01 -8.58700335e-01 -4.46948797e-01
1.53215360e-02 7.28677452e-01 -4.12931830e-01 -8.32747519e-02
-1.15807998e+00 -5.24728954e-01 5.53337395e-01 1.45992801e-01
1.14388287e-01 -6.22842669e-01 -2.94402838e-01 -1.15916157e+00
-2.99338788e-01 -1.08070314e+00 5.18037200e-01 -7.31124640e-01
-7.95832992e-01 1.55935988e-01 4.79469635e-02 -1.74882936e+00
-5.22467136e-01 -5.88581786e-02 -5.24646759e-01 6.57024026e-01
-7.91338503e-01 -6.52016997e-01 2.35446058e-02 7.39152014e-01
8.46522272e-01 -4.67770211e-02 9.05551553e-01 3.76285881e-01
-5.41348815e-01 1.79530203e-01 1.50946423e-01 -2.17791408e-01
7.59054661e-01 -1.33023059e+00 2.21101061e-01 7.38738716e-01
4.24103111e-01 6.93792343e-01 9.76019740e-01 -7.76087344e-01
-1.74147725e+00 -6.94309175e-01 6.84991896e-01 -2.31842399e-01
8.31020176e-01 -1.38794839e-01 -3.08334053e-01 4.76218134e-01
8.88096318e-02 -6.15842223e-01 5.72218835e-01 5.59620023e-01
-7.34042227e-02 -2.11691260e-01 -5.12809753e-01 5.29013872e-01
1.01942980e+00 -5.06539762e-01 -6.98322952e-01 1.33597746e-01
4.43670511e-01 -4.85030800e-01 -9.72299457e-01 1.78765982e-01
1.05324948e+00 -8.84526432e-01 7.83741474e-01 -3.90171081e-01
7.64597356e-02 -7.40151107e-01 3.06861592e-03 -1.00277698e+00
-5.99372208e-01 -1.58294296e+00 -6.79014564e-01 1.21396422e+00
-2.36969143e-02 -2.20043123e-01 6.53060973e-01 -1.01236895e-01
-9.08271596e-02 -4.69336718e-01 -8.59030545e-01 -1.06866062e+00
-6.29396260e-01 -6.48840070e-01 7.36721277e-01 9.52192426e-01
2.61644602e-01 6.18556321e-01 -7.49257803e-01 1.12776704e-01
7.13290155e-01 7.96986282e-01 9.84344780e-01 -1.32922828e+00
-8.89825106e-01 -3.88037354e-01 -3.31622481e-01 -9.59953249e-01
-4.57698524e-01 -5.58521390e-01 1.07776068e-01 -1.25369000e+00
2.50162929e-01 -5.38542390e-01 -6.91120028e-01 3.57337743e-01
1.83214337e-01 1.29173070e-01 3.29329461e-01 4.75776404e-01
-6.44643247e-01 2.88220197e-01 1.33785617e+00 9.61828604e-02
-5.00592649e-01 5.47367692e-01 -2.22547546e-01 1.03935921e+00
9.47389662e-01 -7.66284168e-01 -6.85101449e-01 -1.50190651e-01
5.51970184e-01 7.06049144e-01 2.41561458e-02 -1.15445185e+00
2.09845796e-01 -4.68801290e-01 -2.72990689e-02 -1.06271565e+00
4.66530502e-01 -4.44154710e-01 6.65464282e-01 6.44190192e-01
-1.91951811e-01 4.49975073e-01 1.27709195e-01 8.38624358e-01
-2.32466638e-01 -2.13976383e-01 4.26802367e-01 -8.20675567e-02
-3.52773190e-01 1.79660797e-01 -3.24922919e-01 -1.24323420e-01
8.72699201e-01 -1.77121013e-01 2.25288942e-01 -4.90170926e-01
-9.82898831e-01 -3.18172634e-01 -3.01458649e-02 4.35146391e-01
2.59363294e-01 -1.48351514e+00 -2.58734256e-01 -2.23262653e-01
-2.91961133e-01 -2.75414973e-01 3.33469838e-01 1.12072217e+00
-3.79392117e-01 4.06888038e-01 -2.45744184e-01 -6.50912404e-01
-1.44696987e+00 5.28321743e-01 -2.90994138e-01 -6.22304916e-01
-5.79717755e-01 3.89595926e-01 -1.29863068e-01 4.78857197e-02
2.49502599e-01 -6.06810987e-01 1.56012094e-02 8.08243603e-02
4.67933774e-01 8.42952609e-01 -3.83835346e-01 -2.64638036e-01
-2.81044781e-01 7.69512653e-01 4.71374094e-01 -5.28688490e-01
1.06439674e+00 -2.16594592e-01 -3.60460430e-02 1.30401957e+00
3.86511087e-01 3.48085076e-01 -9.26441073e-01 2.06817940e-01
-9.18312296e-02 -4.26828563e-01 4.94251288e-02 -6.08471394e-01
-5.20403683e-01 7.90018022e-01 5.41021109e-01 4.10754085e-01
1.30940366e+00 -4.55956876e-01 1.03025281e+00 2.63471782e-01
6.98908389e-01 -1.23214078e+00 -6.12634514e-03 5.27683020e-01
5.84068418e-01 -4.76335406e-01 2.47835875e-01 -1.67419761e-01
-4.38682675e-01 1.07860029e+00 1.78516403e-01 -2.58769542e-01
6.13206804e-01 4.49167103e-01 -1.45173579e-01 8.25991109e-02
-1.02577245e+00 -1.62682369e-01 8.08749944e-02 1.30061746e-01
4.76608187e-01 4.71725941e-01 -6.23777568e-01 9.37447548e-01
-4.66596216e-01 1.30253449e-01 6.96604431e-01 9.98375118e-01
-3.23704869e-01 -1.31585312e+00 -6.66096330e-01 2.00338587e-01
-6.06703877e-01 1.82044525e-02 -1.05431348e-01 5.79014480e-01
7.36854896e-02 1.00945604e+00 -1.72993198e-01 -5.46452343e-01
3.49905640e-01 2.02532243e-02 7.96036959e-01 -4.81435388e-01
-6.56274438e-01 8.15249145e-01 1.20257854e-01 -7.26121485e-01
-4.25226212e-01 -9.32262421e-01 -1.30267966e+00 -3.41697723e-01
-5.69678724e-01 2.76507795e-01 7.65464902e-01 6.00275397e-01
2.64907688e-01 8.42251837e-01 7.10080504e-01 -7.46023655e-01
-6.19041324e-01 -1.08057344e+00 -9.05866086e-01 1.12966090e-01
9.47866216e-02 -7.80056536e-01 3.06336600e-02 2.08191276e-01]
|
[15.900257110595703, 5.448034286499023]
|
1c78f99c-f548-44ff-802d-8f36cc58cac2
|
ecnu-at-semeval-2017-task-8-rumour-evaluation
| null | null |
https://aclanthology.org/S17-2086
|
https://aclanthology.org/S17-2086.pdf
|
ECNU at SemEval-2017 Task 8: Rumour Evaluation Using Effective Features and Supervised Ensemble Models
|
This paper describes our submissions to task 8 in SemEval 2017, i.e., Determining rumour veracity and support for rumours. Given a rumoured tweet and a lot of reply tweets, the subtask A is to label whether these tweets are support, deny, query or comment, and the subtask B aims to predict the veracity (i.e., true, false, and unverified) with a confidence (in range of 0-1) of the given rumoured tweet. For both subtasks, we adopted supervised machine learning methods, incorporating rich features. Since training data is imbalanced, we specifically designed a two-step classifier to address subtask A .
|
['Man Lan', 'Yuanbin Wu', 'Feixiang Wang']
|
2017-08-01
| null | null | null |
semeval-2017-8
|
['rumour-detection']
|
['natural-language-processing']
|
[-3.06914628e-01 3.65542322e-01 -5.17241061e-01 -4.28259730e-01
-3.55919957e-01 -3.17931086e-01 7.53429532e-01 5.68615735e-01
-9.68116298e-02 1.20827353e+00 4.54091311e-01 -4.68265951e-01
3.03529918e-01 -7.14536190e-01 -3.63361210e-01 -1.44991487e-01
1.84418503e-02 5.57190239e-01 6.87105134e-02 -2.92519033e-01
8.02610219e-01 2.22359728e-02 -1.23515821e+00 5.40701568e-01
6.81952238e-01 1.16979980e+00 -7.06178308e-01 6.37669861e-01
-1.91567272e-01 1.87432158e+00 -9.86545384e-01 -4.87813175e-01
-3.89646858e-01 -3.52999002e-01 -1.09158182e+00 3.36942472e-03
1.29180830e-02 -2.76422650e-01 -6.88195005e-02 8.75993788e-01
9.62651428e-03 -3.92190218e-02 8.00370216e-01 -1.78366828e+00
-5.74396849e-01 7.80993879e-01 -5.87473273e-01 7.03447580e-01
6.45249069e-01 -1.09189726e-01 1.09075165e+00 -1.10002553e+00
4.80464101e-01 7.66643405e-01 6.14775062e-01 2.09249824e-01
-1.01592112e+00 -9.29414272e-01 -2.75643528e-01 2.78774261e-01
-1.02225125e+00 -6.66902244e-01 5.37972748e-01 -9.59340572e-01
6.23485148e-01 4.18097615e-01 4.09260213e-01 1.28563881e+00
3.83222640e-01 4.97997701e-01 1.36972928e+00 1.56621411e-01
5.87991536e-01 7.42308140e-01 4.60324407e-01 2.19891086e-01
2.00768784e-01 -3.22885126e-01 -7.98227966e-01 -7.72057593e-01
3.20309289e-02 -9.39556584e-02 -1.90474063e-01 4.88823563e-01
-9.92135108e-01 1.28452265e+00 4.40460294e-01 5.70463948e-02
-7.73342133e-01 -5.46078146e-01 6.46818578e-01 8.19075227e-01
1.10429275e+00 5.94373643e-01 -8.70491713e-02 -1.30989820e-01
-1.26458478e+00 3.87113214e-01 1.49338412e+00 4.44050550e-01
8.48308504e-01 -1.07027330e-01 -2.96011627e-01 7.97968447e-01
-7.80906156e-03 2.04932854e-01 5.81562042e-01 -4.53303576e-01
3.89528066e-01 5.26677132e-01 7.41231322e-01 -1.40134192e+00
-6.43573463e-01 -4.81035709e-01 -7.10590184e-01 -2.05336720e-01
6.68651238e-02 -3.44904304e-01 -4.69942749e-01 1.18186224e+00
1.89416781e-01 1.36137098e-01 1.00498535e-02 1.13104236e+00
1.00813496e+00 6.91053867e-01 -2.29977861e-01 -5.33961773e-01
1.20574641e+00 -7.15198398e-01 -8.99219632e-01 -3.34754229e-01
6.31736398e-01 -1.10119331e+00 6.32679403e-01 4.73242581e-01
-9.54683959e-01 5.49007109e-06 -1.05109000e+00 1.88776240e-01
-1.52242839e-01 -3.22454363e-01 2.95915872e-01 3.85645896e-01
-7.33006060e-01 4.46343511e-01 -1.59369469e-01 3.69126424e-02
1.81906849e-01 -2.99533904e-01 -1.76307693e-01 6.02762327e-02
-1.59409273e+00 9.13453400e-01 2.07649350e-01 -1.82377487e-01
-8.17937255e-01 -3.53537261e-01 -5.51812530e-01 -4.59048636e-02
2.46143967e-01 -2.51934916e-01 1.42501271e+00 -8.37634861e-01
-1.10703242e+00 1.15009928e+00 -2.42289126e-01 -7.10567355e-01
8.21978211e-01 -1.20261470e-02 -7.94402540e-01 -1.07772663e-01
4.60565835e-01 -2.43807152e-01 1.10780311e+00 -9.34336901e-01
-7.40180552e-01 -3.73868525e-01 -2.18954712e-01 -1.27502114e-01
-1.98632196e-01 3.42596948e-01 4.72897828e-01 -2.96720713e-01
3.02902162e-01 -6.27421618e-01 4.33044225e-01 -6.89987481e-01
-8.28835547e-01 -4.30797875e-01 8.13137054e-01 -6.47417247e-01
1.44416595e+00 -1.83086336e+00 -5.98370433e-01 1.89439937e-01
7.94566989e-01 1.33126587e-01 1.97198123e-01 6.00613415e-01
-5.52233169e-03 3.19355458e-01 3.50597233e-01 -1.33513525e-01
-1.24318287e-01 -4.57647257e-02 -8.16067636e-01 7.29729950e-01
1.13406204e-01 3.64056289e-01 -8.82751286e-01 -2.92641670e-01
-4.44423974e-01 -1.23287044e-01 -1.91658258e-01 4.69327152e-01
-1.07167035e-01 2.78220087e-01 -5.80116808e-01 4.67014015e-01
8.78247559e-01 -6.07287586e-01 -2.57301182e-01 1.51533023e-01
-4.46152985e-01 1.04806495e+00 -7.06680715e-01 8.54800493e-02
-2.97947407e-01 7.20008552e-01 1.12326443e-01 -8.55642974e-01
1.37876773e+00 4.27749544e-01 2.17094883e-01 -5.82086325e-01
2.18278572e-01 4.74761963e-01 -3.91240090e-01 -7.10929155e-01
8.59547019e-01 -4.54037756e-01 -2.21993834e-01 1.09277177e+00
-5.92724800e-01 1.58837717e-02 -8.51837397e-02 4.58044201e-01
1.19946253e+00 -8.17635238e-01 5.45504689e-01 -6.80169165e-02
4.50083822e-01 2.05383241e-01 3.72488916e-01 8.08806777e-01
-4.67187375e-01 4.71910477e-01 1.22858453e+00 -3.84313762e-01
-9.90758717e-01 -5.70810854e-01 -1.54537484e-01 1.34295595e+00
-5.49355783e-02 -1.13965675e-01 -2.49078214e-01 -5.60793042e-01
3.01424235e-01 1.03178692e+00 -7.67132401e-01 7.80854970e-02
-2.19207719e-01 -6.15400493e-01 5.72671831e-01 -9.04388428e-02
2.95896381e-01 -9.56794918e-01 -3.08765054e-01 2.45070636e-01
-6.00861013e-01 -1.03166842e+00 -3.81617457e-01 1.03766233e-01
-4.58706111e-01 -1.12641799e+00 -2.83540159e-01 -4.94123727e-01
1.71512991e-01 4.33495492e-01 1.25489938e+00 3.89358878e-01
3.72412562e-01 -4.50705349e-01 -6.42423987e-01 -3.05358887e-01
-5.39416432e-01 -4.80316877e-02 1.05521217e-01 2.12179542e-01
4.16557729e-01 -3.41740996e-01 -3.33907098e-01 5.37800729e-01
-6.84656203e-01 -6.17538132e-02 -4.27660905e-03 7.13846743e-01
4.51869564e-03 -1.87860072e-01 1.21900892e+00 -1.08221412e+00
1.16724968e+00 -1.49082577e+00 -8.35013762e-02 -2.96586871e-01
-7.74362326e-01 -6.26395583e-01 6.18738651e-01 -2.21267134e-01
-5.91787100e-01 -7.90382683e-01 -1.94901079e-01 7.63710141e-02
2.19883304e-03 9.44128513e-01 7.14941561e-01 3.40742975e-01
8.93587708e-01 1.08901255e-01 1.71093345e-01 -2.33930275e-01
-1.56671479e-01 1.38166547e+00 2.70768046e-01 -1.40135035e-01
3.01385522e-01 3.10496628e-01 -5.57915688e-01 -6.88084960e-01
-1.55641317e+00 -1.02563035e+00 4.72058170e-02 -2.53209502e-01
2.88485974e-01 -9.62028921e-01 -7.48153865e-01 4.68712330e-01
-1.24063110e+00 7.62166530e-02 1.82646856e-01 3.89136970e-01
-3.19766730e-01 9.33685303e-02 -8.16108108e-01 -9.97292995e-01
-4.86983955e-01 -5.63970983e-01 1.79525733e-01 -4.88057062e-02
-7.53966451e-01 -8.02638412e-01 1.84695572e-01 7.46961236e-01
6.27049804e-01 3.61434311e-01 5.14065444e-01 -1.57382822e+00
7.25993812e-02 -7.86636889e-01 -5.41225433e-01 5.76449223e-02
-1.57792091e-01 -1.15253501e-01 -1.01483405e+00 -2.24235386e-01
4.40472424e-01 -1.09058571e+00 7.90012538e-01 -7.61853606e-02
8.50495040e-01 -1.26295459e+00 -5.43383299e-04 -4.90301847e-02
7.77278900e-01 -5.52353919e-01 5.42781413e-01 5.78901470e-01
7.21577704e-02 6.10561252e-01 6.84302568e-01 1.11810338e+00
6.57048821e-01 4.24760491e-01 5.94917655e-01 3.92459780e-01
3.54823589e-01 -2.93406039e-01 4.63025481e-01 7.57131159e-01
4.11764085e-01 -4.95465308e-01 -7.76369929e-01 5.66244245e-01
-1.72193766e+00 -1.14661348e+00 -6.80970848e-01 1.96988738e+00
1.04975545e+00 4.35857266e-01 3.68389785e-01 2.34440804e-01
8.90980303e-01 5.42648077e-01 -4.80337620e-01 -7.67276168e-01
-9.56966132e-02 -5.27030647e-01 1.24267220e-01 4.77147847e-01
-8.58064115e-01 6.06204212e-01 6.13137007e+00 3.73343349e-01
-1.41578102e+00 2.81029850e-01 7.44939566e-01 -2.92853832e-01
-3.55989754e-01 -8.07026774e-02 -6.42162859e-01 7.06369460e-01
1.15771043e+00 -3.66063237e-01 3.11549038e-01 7.74798274e-01
5.85811019e-01 -2.86934644e-01 -7.24351168e-01 4.33397651e-01
4.08466607e-01 -1.27713525e+00 -2.80569196e-01 -8.48053470e-02
7.88465619e-01 5.24104893e-01 -8.90432969e-02 6.32808626e-01
2.50839978e-01 -1.11070442e+00 1.00618482e+00 4.08311903e-01
3.77377927e-01 -4.51057941e-01 1.13108337e+00 1.05422521e+00
-9.42628384e-02 6.28213286e-02 -3.71827155e-01 -5.45594215e-01
2.22507477e-01 1.38199675e+00 -1.34378827e+00 -6.78113569e-03
4.18088973e-01 6.75119042e-01 -2.00376362e-01 9.55505192e-01
-4.59011167e-01 1.10962284e+00 9.44640040e-02 -3.39722037e-01
3.17814559e-01 2.96908408e-01 7.94548094e-01 1.23656976e+00
1.11890189e-01 2.95878425e-02 3.13246578e-01 8.89806390e-01
-4.72396135e-01 1.35848969e-01 -1.04068972e-01 -1.99460164e-01
7.50880480e-01 1.13186979e+00 -2.06925839e-01 -4.65753704e-01
-2.92916615e-02 7.28565216e-01 5.16227901e-01 2.16882735e-01
-7.38864839e-01 -3.29445392e-01 2.10255742e-01 6.20548427e-01
-5.46543077e-02 3.10703039e-01 -6.93886459e-01 -1.29116333e+00
-2.06765562e-01 -6.56173706e-01 4.83924925e-01 -7.35298097e-01
-1.54362118e+00 7.19377756e-01 -5.90324461e-01 -8.89736712e-01
-3.01007450e-01 -2.42709368e-02 -1.18454134e+00 9.93858755e-01
-1.85895336e+00 -6.72545254e-01 -4.21571225e-01 3.43522578e-01
1.79755390e-01 -1.01454377e-01 3.78462881e-01 -1.86572559e-02
-5.75672746e-01 3.83244962e-01 -8.45155492e-02 5.85780144e-02
9.70150948e-01 -9.06553864e-01 4.93021831e-02 1.47353530e-01
-5.63620567e-01 4.07240033e-01 1.35992026e+00 -7.47207046e-01
-8.69988799e-01 -1.15139174e+00 1.73073232e+00 -2.48576626e-01
1.25789261e+00 1.00226462e-01 -1.11157894e+00 5.59538007e-01
-2.13781446e-01 6.89739287e-02 7.43664384e-01 4.21844959e-01
-5.37533224e-01 4.48906779e-01 -1.45236516e+00 3.58130112e-02
1.43120661e-01 -5.21095097e-01 -7.99476206e-01 6.11454189e-01
4.49450403e-01 -3.39905709e-01 -7.96321273e-01 -9.83440205e-02
4.06479895e-01 -1.24770045e+00 3.53306472e-01 -9.28027153e-01
9.57043648e-01 2.59722508e-02 5.84171445e-04 -1.35224497e+00
-4.46299165e-01 -4.46876734e-01 -2.47100338e-01 9.72668111e-01
5.42171597e-01 -9.03865218e-01 7.11771190e-01 3.54216635e-01
6.46241903e-02 -6.86318040e-01 -8.23604047e-01 -5.16884267e-01
9.81881283e-03 -2.64142156e-01 5.15854418e-01 1.42317462e+00
7.06789434e-01 5.46369493e-01 -8.40466619e-01 -1.06868893e-01
3.76636356e-01 4.18334246e-01 7.72948921e-01 -1.26756859e+00
1.45199463e-01 -3.74869376e-01 -1.64420515e-01 -8.07460129e-01
1.74573511e-01 -9.58504438e-01 -2.64016613e-02 -1.30142796e+00
4.54418182e-01 -4.51597065e-01 -1.97940558e-01 5.15079379e-01
-1.48930699e-01 4.85233307e-01 -2.56953955e-01 9.89828706e-01
-6.70600057e-01 3.93684298e-01 8.56600404e-01 -7.78732672e-02
-1.52052522e-01 6.11033261e-01 -9.51619864e-01 6.37272298e-01
1.08353972e+00 -7.49105275e-01 1.40516549e-01 1.06780916e-01
8.52956533e-01 4.87717867e-01 5.25689125e-01 -3.84244740e-01
2.98318267e-01 -3.99611980e-01 -7.82575607e-02 -6.99650526e-01
1.72702596e-01 -3.13495994e-02 -3.76872718e-01 2.96820879e-01
-6.76931620e-01 -1.60941601e-01 -3.40529114e-01 5.37896514e-01
-3.41384023e-01 -2.85475403e-01 8.70231450e-01 8.91468301e-02
-1.66034669e-01 8.48064497e-02 -6.05666757e-01 5.18329799e-01
7.87663341e-01 1.19053401e-01 -8.44189882e-01 -8.48560512e-01
-7.53643930e-01 3.53094697e-01 3.75280112e-01 1.97907850e-01
7.10029662e-01 -1.01238286e+00 -1.52557075e+00 -2.13360470e-02
3.79519552e-01 -6.35572910e-01 1.07086249e-01 1.28927386e+00
-1.22562833e-01 3.34183097e-01 -5.00927269e-02 -1.73022613e-01
-9.15816844e-01 1.94399133e-01 -3.60047370e-02 -2.43770912e-01
-3.64255637e-01 6.53062761e-01 -4.18324500e-01 -2.21622884e-01
8.87778476e-02 3.83059204e-01 -6.13991737e-01 5.23834825e-01
9.87934411e-01 8.22156012e-01 3.68159622e-01 -8.23516965e-01
-2.67754942e-01 -5.52587271e-01 -3.58279496e-01 9.04686525e-02
1.10791945e+00 -2.27825880e-01 -6.13791764e-01 8.14932346e-01
1.23752260e+00 1.71563774e-01 -5.67738533e-01 -4.95678872e-01
9.73268300e-02 -5.23066759e-01 3.15418601e-01 -9.96976912e-01
-4.84298915e-01 4.52066481e-01 -4.04104739e-01 1.08907962e+00
3.67268920e-01 1.60299972e-01 1.00306308e+00 1.47036731e-01
1.09969012e-01 -1.08310151e+00 2.17881590e-01 1.14500296e+00
1.24166226e+00 -1.52698874e+00 2.14240745e-01 -1.67281330e-01
-1.04358697e+00 9.86101508e-01 4.16341990e-01 1.86506454e-02
6.20816529e-01 -3.52444276e-02 1.16937973e-01 -5.27090073e-01
-1.15458250e+00 4.88901913e-01 1.73339948e-01 -4.85338271e-02
6.55881166e-01 3.36910784e-01 -6.52789593e-01 7.21864164e-01
-6.96891606e-01 1.99014619e-01 1.13918447e+00 5.23408890e-01
-1.10029995e+00 9.90929455e-02 -5.29913127e-01 1.00703061e+00
-7.87122846e-01 6.52144924e-02 -6.32911682e-01 1.90662339e-01
-2.01446399e-01 1.53473890e+00 1.22875676e-01 -6.73194587e-01
2.06911620e-02 -1.72094136e-01 -4.50976610e-01 -6.18299425e-01
-6.56615198e-01 -5.04355371e-01 6.06416464e-01 -2.69457936e-01
3.52648534e-02 -6.01877570e-01 -1.06224835e+00 -1.06614447e+00
-4.51427251e-01 6.30515158e-01 7.05837071e-01 1.11714363e+00
8.27726945e-02 -4.17193063e-02 1.35002124e+00 -3.24830353e-01
-1.16316140e+00 -1.34414279e+00 -8.01331580e-01 5.00237226e-01
8.70413482e-01 -4.61165249e-01 -9.79102671e-01 -3.74368966e-01]
|
[8.20391845703125, 10.1229248046875]
|
7a2b66b5-a8a5-4d3f-912a-e4ae636c5858
|
incorporating-coincidental-water-data-into
|
2101.07190
| null |
https://arxiv.org/abs/2101.07190v1
|
https://arxiv.org/pdf/2101.07190v1.pdf
|
Incorporating Coincidental Water Data into Non-intrusive Load Monitoring
|
Non-intrusive load monitoring (NILM) as the process of extracting the usage pattern of appliances from the aggregated power signal is among successful approaches aiding residential energy management. In recent years, high volume datasets on power profiles have become available, which has helped make classification methods employed for the NILM purpose more effective and more accurate. However, the presence of multi-mode appliances and appliances with close power values have remained influential in worsening the computational complexity and diminishing the accuracy of these algorithms. To tackle these challenges, we propose an event-based classification process, in the first phase of which the $K$-nearest neighbors method, as a fast classification technique, is employed to extract power signals of appliances with exclusive non-overlapping power values. Then, two deep learning models, which consider the water consumption of some appliances as a novel signature in the network, are utilized to distinguish between appliances with overlapping power values. In addition to power disaggregation, the proposed process as well extracts the water consumption profiles of specific appliances. To illustrate the proposed process and validate its efficiency, seven appliances of the AMPds are considered, with the numerical classification results showing marked improvement with respect to the existing classification-based NILM techniques.
|
['Sadegh Bolouki', 'Hamidreza Momeni', 'Elnaz Azizi', 'Mohammad-Mehdi Keramati']
|
2021-01-18
| null | null | null | null |
['non-intrusive-load-monitoring', 'non-intrusive-load-monitoring', 'non-intrusive-load-monitoring']
|
['knowledge-base', 'miscellaneous', 'time-series']
|
[ 5.40685430e-02 -3.44805300e-01 -3.84100229e-02 -2.89823472e-01
-3.40334147e-01 -3.68884891e-01 4.71929818e-01 1.65349230e-01
-9.97362584e-02 7.51131535e-01 1.71439439e-01 2.09124032e-02
-4.88910407e-01 -1.09952867e+00 -1.31323963e-01 -1.27119160e+00
-3.39940846e-01 1.30822614e-01 -3.87529343e-01 -1.67288147e-02
-8.77974927e-02 5.24368227e-01 -1.68559039e+00 5.47281373e-03
9.13545668e-01 1.28906918e+00 -7.94609562e-02 -4.45788018e-02
-8.07298645e-02 7.66875327e-01 -8.17593575e-01 2.00204318e-03
1.74518973e-01 -2.55016476e-01 -6.06201410e-01 6.62568444e-03
-4.14432317e-01 -5.01508534e-01 -3.79119486e-01 1.09645569e+00
6.41330242e-01 3.55741769e-01 5.94387293e-01 -1.48666370e+00
-1.57693401e-01 8.71869802e-01 -7.06391513e-01 2.46821687e-01
3.92594337e-01 9.38877463e-02 1.02926409e+00 -2.91667189e-02
-1.04735039e-01 6.94564164e-01 5.37099540e-01 -9.29976627e-02
-1.12288892e+00 -8.74478340e-01 -3.97733152e-02 7.21411705e-01
-1.45764565e+00 -2.15358600e-01 1.42818284e+00 -1.73229024e-01
1.28522170e+00 4.08857226e-01 9.05074835e-01 9.60211158e-01
-6.88626990e-02 9.97869492e-01 9.73046780e-01 -3.80015612e-01
5.04457891e-01 1.27084717e-01 1.71162888e-01 1.11828884e-02
1.73105553e-01 -2.70382781e-03 -1.03046417e-01 -1.25767976e-01
-1.73904136e-01 1.93811402e-01 -3.19988966e-01 -6.55453536e-04
-4.89100188e-01 6.90215707e-01 4.81918097e-01 8.87821496e-01
-7.09970832e-01 -3.38321030e-01 5.45801520e-01 -7.76350172e-03
7.07716882e-01 1.23444974e-01 -3.76733094e-01 -2.43060678e-01
-1.03087997e+00 -1.05555512e-01 7.66259134e-01 5.13646305e-01
5.93206525e-01 2.78677374e-01 9.83417034e-02 5.90272367e-01
2.34050438e-01 3.12377959e-01 6.99100554e-01 -5.26051104e-01
3.33453149e-01 1.08457398e+00 -3.81646827e-02 -1.03374743e+00
-8.98823142e-01 -4.52651709e-01 -1.24031436e+00 1.22507468e-01
4.50123586e-02 -2.48015076e-01 -2.06460357e-01 1.61767328e+00
4.00388092e-01 2.32667491e-01 -5.21572046e-02 4.03891563e-01
4.26094204e-01 1.02561378e+00 2.92157143e-01 -6.37322903e-01
1.36176586e+00 -2.80978352e-01 -9.20666575e-01 4.39671814e-01
4.04875904e-01 -3.50021809e-01 4.89912659e-01 3.99301499e-01
-5.27531087e-01 -3.40992421e-01 -1.20791435e+00 5.02820849e-01
-7.63222456e-01 -6.01503141e-02 4.98854101e-01 5.09124100e-01
-6.05631411e-01 7.66112268e-01 -6.66362762e-01 -3.26773643e-01
7.54848003e-01 3.63360196e-01 -1.33724436e-01 3.87367070e-01
-1.26574183e+00 8.85112166e-01 8.03101599e-01 2.48800725e-01
-2.22144648e-01 -6.46176279e-01 -6.25582635e-01 4.39448029e-01
3.38325300e-03 -2.46450946e-01 8.95218551e-01 -8.38883221e-01
-1.49522865e+00 2.32525200e-01 2.24497512e-01 -5.72599113e-01
4.56936777e-01 -2.98914015e-02 -9.92319643e-01 1.37761086e-01
-5.13143502e-02 -8.05069506e-02 8.28423321e-01 -8.88380527e-01
-1.07279837e+00 -5.27714849e-01 -2.43645534e-01 4.86374507e-03
-6.68166041e-01 -3.82467091e-01 3.20795357e-01 -3.06525588e-01
-2.42748693e-01 -4.69520152e-01 2.31756151e-01 -7.21588552e-01
-5.01914084e-01 -8.25982988e-01 1.17623615e+00 -9.16067362e-01
1.39635551e+00 -2.40451646e+00 -2.93001264e-01 5.57522297e-01
2.03464460e-03 3.03084522e-01 2.73898423e-01 4.91124749e-01
-2.61703491e-01 -3.78015697e-01 -3.05430651e-01 -2.56494045e-01
4.83248293e-01 2.01425135e-01 -2.25376025e-01 6.14509940e-01
-3.03061493e-03 6.25257373e-01 -7.57429481e-01 -2.62520492e-01
9.38178301e-01 6.42817557e-01 1.85974836e-01 2.65622050e-01
-6.53859675e-02 3.99414748e-01 -4.10002053e-01 4.18703824e-01
6.54446244e-01 -5.94890229e-02 2.66854972e-01 -6.33333445e-01
-1.14910223e-01 1.88762933e-01 -1.22341859e+00 1.16463721e+00
-5.95153689e-01 5.02499521e-01 -2.37220097e-02 -1.47168803e+00
7.45372772e-01 6.59421086e-01 1.19124234e+00 -1.09211242e+00
5.06778300e-01 1.26507208e-01 -1.28859147e-01 -5.10500968e-01
-4.11719643e-02 2.61619002e-01 -5.72832599e-02 3.40104997e-01
2.52794195e-02 2.31043831e-01 4.78055030e-01 -3.34588796e-01
9.83826160e-01 -1.31492332e-01 5.44683099e-01 -3.26988816e-01
7.87904978e-01 -4.23847973e-01 6.06451392e-01 1.89094886e-01
-1.25204340e-01 -2.99960673e-01 4.39324677e-02 -5.26623845e-01
-7.06906796e-01 -7.71087706e-01 -4.06987041e-01 7.60812461e-01
6.02589622e-02 -5.00117429e-03 -7.17289567e-01 -4.95652944e-01
8.71859193e-02 1.19886577e+00 -3.31510276e-01 -2.63769060e-01
-6.28044069e-01 -1.47215807e+00 1.73764035e-01 4.00676191e-01
9.13979411e-01 -1.28911138e+00 -9.50735271e-01 5.36299229e-01
-1.58760563e-01 -1.00998461e+00 5.85869588e-02 7.25161076e-01
-4.33289826e-01 -1.23312116e+00 -3.11395198e-01 -5.95220566e-01
4.62875456e-01 -1.87578440e-01 9.75303829e-01 -2.81977117e-01
-3.29065204e-01 5.84093072e-02 -4.11039144e-01 -3.73015285e-01
-1.94673285e-01 3.97745848e-01 1.48903772e-01 3.62816572e-01
9.15675998e-01 -1.30313516e+00 -6.54494882e-01 -4.88641560e-02
-6.76943541e-01 -2.02198982e-01 5.78639984e-01 3.28775525e-01
1.30166978e-01 1.14926076e+00 7.90363073e-01 -4.19926465e-01
6.15927458e-01 -8.34054947e-01 -8.78487349e-01 -1.54652998e-01
-9.42829967e-01 -2.38991365e-01 1.15400863e+00 -2.21761838e-01
-8.82988572e-01 -2.83846408e-02 -2.44969055e-01 -9.02055204e-03
-4.29397881e-01 -7.23091438e-02 -5.73131561e-01 4.08372521e-01
-2.80921417e-03 4.93415415e-01 -4.08635825e-01 -7.31657922e-01
8.81957859e-02 8.28050971e-01 4.30577457e-01 -1.57241106e-01
9.28294599e-01 2.39242703e-01 1.15703166e-01 -9.02234614e-01
-5.52103996e-01 -4.84769821e-01 -5.51131904e-01 -2.44633928e-02
9.12859380e-01 -6.83060169e-01 -1.21232021e+00 9.02695239e-01
-9.32324231e-01 -1.11075770e-02 -6.54538870e-01 2.51636297e-01
-2.24062085e-01 2.22975731e-01 -4.77727234e-01 -1.11582148e+00
-7.39967048e-01 -9.94329214e-01 7.39463210e-01 5.52873731e-01
-4.61532533e-01 -9.09179628e-01 -1.24601029e-01 1.39265150e-01
4.47488964e-01 5.72252572e-01 1.34506238e+00 -1.03853202e+00
-2.73988187e-01 -3.29407841e-01 7.13069215e-02 3.82100165e-01
5.32834768e-01 -2.63123125e-01 -1.06222796e+00 -4.64184910e-01
2.44199306e-01 3.03890556e-01 4.27221358e-02 2.39504352e-01
1.09553790e+00 -4.60482925e-01 -4.40805972e-01 4.92260963e-01
1.50502861e+00 7.61113167e-01 3.35741311e-01 3.97798657e-01
4.15200561e-01 2.92513311e-01 -9.86159891e-02 7.19738245e-01
3.14304531e-01 5.30978024e-01 6.10174119e-01 -1.38949916e-01
3.21623087e-01 -2.18853187e-02 1.51809618e-01 1.09726107e+00
1.57207176e-01 -3.31056833e-01 -3.52220386e-01 6.09448612e-01
-1.68835616e+00 -1.08515310e+00 2.14016631e-01 1.75155056e+00
5.69925547e-01 1.36101201e-01 2.37776861e-01 1.02522755e+00
6.10027552e-01 5.46780467e-01 -7.20820665e-01 -1.88518450e-01
-7.09720850e-02 3.48913491e-01 2.35521808e-01 1.38376191e-01
-1.22959733e+00 -7.85726607e-02 5.36302662e+00 8.36922526e-01
-8.81289840e-01 1.92768633e-01 4.10059601e-01 -1.43917099e-01
1.48561731e-01 -6.73717618e-01 -4.90227580e-01 1.10212278e+00
9.65103924e-01 -1.52402639e-01 6.58626616e-01 8.55761707e-01
5.09023428e-01 -2.63426065e-01 -1.23993266e+00 1.26859856e+00
-2.68812776e-01 -7.09224522e-01 -3.34032476e-01 1.39080033e-01
5.78594983e-01 -2.53347959e-03 -3.01799357e-01 2.46555209e-01
3.33355278e-01 -5.96048653e-01 3.51542950e-01 4.39140052e-01
1.90596089e-01 -1.14755988e+00 1.06798303e+00 4.43130255e-01
-1.55296874e+00 -5.99510252e-01 2.21901134e-01 -1.24935694e-01
1.56626657e-01 9.55619574e-01 -4.61622924e-01 9.45420921e-01
1.18090773e+00 6.10325694e-01 -1.84984371e-01 6.85057342e-01
-1.37020886e-01 7.58415282e-01 -7.00319052e-01 -5.62607311e-02
-3.79607156e-02 -4.86930609e-01 1.93974704e-01 9.24584568e-01
3.56893480e-01 1.06868587e-01 9.32929590e-02 7.68399417e-01
-1.86592758e-01 9.49909166e-02 -4.39397246e-01 3.48899961e-01
6.79869175e-01 1.53426111e+00 -5.95387876e-01 -2.78508663e-01
-4.50889140e-01 7.64180005e-01 -1.88675523e-01 2.87996322e-01
-7.04479337e-01 -5.34477115e-01 9.20919478e-01 -6.37070239e-02
1.00689478e-01 5.34097478e-02 -4.68589701e-02 -8.48579586e-01
5.81268519e-02 -5.37399232e-01 4.09111649e-01 -4.37489837e-01
-1.64363003e+00 3.24139863e-01 1.03419103e-01 -1.17881012e+00
-6.15629494e-01 -1.66733220e-01 -9.69085038e-01 7.47255385e-01
-1.54823291e+00 -9.24462140e-01 -3.95874023e-01 6.49458289e-01
5.43923020e-01 -2.21803755e-01 8.40037704e-01 7.90593266e-01
-9.52606082e-01 4.76291031e-01 6.13937020e-01 3.52530837e-01
-2.07232073e-01 -1.23733389e+00 -6.73335046e-02 8.53193760e-01
1.24081016e-01 -1.28339902e-01 6.14278793e-01 -1.44333050e-01
-1.22060096e+00 -1.08048797e+00 5.28155327e-01 -9.95293781e-02
5.83077550e-01 -2.98900545e-01 -6.92607999e-01 4.16031420e-01
4.44010943e-01 -2.71308601e-01 7.70561993e-01 -2.69155353e-01
2.88674623e-01 -7.28218734e-01 -1.41141212e+00 2.43850634e-01
5.47541142e-01 -5.46755552e-01 -7.09523499e-01 1.28244564e-01
1.56701252e-01 2.86657304e-01 -1.15609801e+00 4.61527139e-01
2.71216124e-01 -8.63322854e-01 7.03390837e-01 1.24473289e-01
-2.45778114e-01 -4.96995330e-01 -3.27162206e-01 -1.50441194e+00
-3.72145981e-01 -5.25910258e-01 -6.32938504e-01 1.85349524e+00
-1.92353204e-01 -9.21720266e-01 4.39959526e-01 3.56204420e-01
3.25302422e-01 -5.64130783e-01 -1.25073719e+00 -4.42426234e-01
-2.85947055e-01 -4.13428485e-01 1.39552999e+00 1.01199353e+00
1.88953906e-01 2.67369479e-01 -8.19896907e-02 2.79282898e-01
6.83087409e-01 2.37061918e-01 2.80403435e-01 -1.48163712e+00
-1.97068099e-02 -6.31041110e-01 -4.57082391e-01 -3.95892501e-01
3.14275056e-01 -7.69605815e-01 -1.93249077e-01 -1.37056458e+00
-5.81027307e-02 -8.62012878e-02 -8.04746687e-01 6.08381271e-01
2.42268622e-01 2.49342501e-01 1.55988276e-01 1.05507970e-01
-2.14188114e-01 7.64971435e-01 3.24281842e-01 -5.65429032e-01
-1.89130753e-01 2.53507614e-01 -3.04167539e-01 8.82981956e-01
9.53401268e-01 -1.26047492e-01 -4.88707989e-01 1.34200463e-02
-8.55824500e-02 -2.81952411e-01 1.87083688e-02 -1.37937534e+00
3.49333398e-02 4.10779744e-01 7.42770553e-01 -8.59407365e-01
9.83300060e-02 -1.58963299e+00 5.14251828e-01 4.88251150e-01
1.51006371e-01 1.15828980e-02 1.03153847e-01 1.27431482e-01
-2.27483094e-01 -3.21037099e-02 5.55694461e-01 5.12089171e-02
-8.26988518e-01 1.69838518e-01 -4.79740798e-01 -5.44067860e-01
1.12478888e+00 -1.95157900e-02 -1.42698109e-01 -1.09112099e-01
-6.36598289e-01 4.21605200e-01 -8.69164914e-02 4.40540105e-01
-1.82413965e-01 -1.42538488e+00 -3.03820550e-01 2.91237295e-01
-1.91219330e-01 -4.74313041e-03 2.50793904e-01 6.55562520e-01
2.68588085e-02 3.50428194e-01 -1.73556693e-02 -5.63495398e-01
-8.44785869e-01 6.23660803e-01 4.61197019e-01 -3.99954826e-01
-7.26970732e-01 -1.25223875e-01 -3.08866829e-01 -6.33927286e-02
2.21782058e-01 -3.54854852e-01 -5.97197592e-01 7.29476511e-01
3.70067954e-01 9.39085364e-01 3.96626592e-01 -6.60025477e-01
-3.49466622e-01 6.30405009e-01 3.57536912e-01 5.55175900e-01
1.70216036e+00 -3.96635860e-01 -2.37362891e-01 5.23382843e-01
1.39989686e+00 -4.31980103e-01 -1.06009293e+00 -5.75382859e-02
1.62157357e-01 3.11562885e-02 1.87548459e-01 -8.25779378e-01
-1.42853034e+00 6.76411331e-01 1.10915136e+00 9.26519990e-01
1.76468265e+00 -2.90658325e-01 1.09906566e+00 -2.35117301e-02
6.61263347e-01 -1.31483829e+00 -7.22160995e-01 -9.88473818e-02
1.52784392e-01 -9.77581859e-01 -1.47485226e-01 9.56571326e-02
3.02790433e-01 9.62941945e-01 2.56684899e-01 1.71848312e-01
9.04108763e-01 4.47686344e-01 -2.29490012e-01 3.33782174e-02
-1.56862155e-01 -8.47192854e-02 -4.00441438e-02 5.42773902e-01
-3.86683755e-02 2.36040279e-01 -1.89899206e-01 7.13900566e-01
-3.04894090e-01 1.43707260e-01 -7.09204599e-02 6.98927581e-01
-8.99830088e-02 -7.64420033e-01 -3.23432773e-01 7.15845346e-01
-5.01671612e-01 3.69025737e-01 3.09010178e-01 7.59132981e-01
4.51981813e-01 1.26632130e+00 3.68751794e-01 -3.86548191e-01
5.12149155e-01 4.07001585e-01 5.29620200e-02 2.55694091e-01
-5.71661472e-01 -2.36953363e-01 -2.24113390e-01 -5.05056441e-01
-4.34225976e-01 -4.70965713e-01 -1.11358404e+00 -5.41319311e-01
-2.53800511e-01 3.43167841e-01 4.36989546e-01 1.23211670e+00
1.26877487e-01 8.17253411e-01 1.31193888e+00 -1.05583894e+00
-6.13297462e-01 -1.11333382e+00 -9.75047827e-01 8.23590934e-01
3.17631692e-01 -4.92787033e-01 -6.03911161e-01 -2.14469358e-01]
|
[6.028477191925049, 2.603626012802124]
|
2d46ff2e-0d57-4605-849c-50ce05531bcd
|
bayesian-reparameterization-of-reward
|
2305.11340
| null |
https://arxiv.org/abs/2305.11340v1
|
https://arxiv.org/pdf/2305.11340v1.pdf
|
Bayesian Reparameterization of Reward-Conditioned Reinforcement Learning with Energy-based Models
|
Recently, reward-conditioned reinforcement learning (RCRL) has gained popularity due to its simplicity, flexibility, and off-policy nature. However, we will show that current RCRL approaches are fundamentally limited and fail to address two critical challenges of RCRL -- improving generalization on high reward-to-go (RTG) inputs, and avoiding out-of-distribution (OOD) RTG queries during testing time. To address these challenges when training vanilla RCRL architectures, we propose Bayesian Reparameterized RCRL (BR-RCRL), a novel set of inductive biases for RCRL inspired by Bayes' theorem. BR-RCRL removes a core obstacle preventing vanilla RCRL from generalizing on high RTG inputs -- a tendency that the model treats different RTG inputs as independent values, which we term ``RTG Independence". BR-RCRL also allows us to design an accompanying adaptive inference method, which maximizes total returns while avoiding OOD queries that yield unpredictable behaviors in vanilla RCRL methods. We show that BR-RCRL achieves state-of-the-art performance on the Gym-Mujoco and Atari offline RL benchmarks, improving upon vanilla RCRL by up to 11%.
|
['Marco Pavone', 'Ding Zhao', 'Tong Che', 'Wenhao Ding']
|
2023-05-18
| null | null | null | null |
['offline-rl']
|
['playing-games']
|
[-1.62970290e-01 -8.82731080e-02 -7.38163471e-01 -3.68657559e-01
-1.06204879e+00 -5.97286105e-01 4.34315205e-01 -1.27456859e-01
-8.89774919e-01 1.04352844e+00 5.35899475e-02 -6.90444052e-01
-3.51801068e-01 -8.26023161e-01 -9.30275142e-01 -7.78639376e-01
-3.89286846e-01 4.71310347e-01 1.78778052e-01 -4.72676784e-01
2.20751330e-01 5.37251174e-01 -1.49132323e+00 1.15041658e-01
7.69413054e-01 8.98912549e-01 7.15297759e-02 9.89721239e-01
2.66673982e-01 1.22217190e+00 -5.59883893e-01 -6.01298325e-02
3.25874031e-01 -4.04577345e-01 -6.78634405e-01 -8.20223272e-01
1.01742804e-01 -6.93095684e-01 -4.76121783e-01 6.64423943e-01
5.51309049e-01 6.37437284e-01 6.27966583e-01 -1.34788954e+00
-7.09704280e-01 8.75683546e-01 -5.27091146e-01 3.15542310e-01
1.32764086e-01 3.62680018e-01 1.49995029e+00 -4.12685364e-01
3.59685093e-01 1.37350917e+00 3.57888997e-01 9.84022260e-01
-1.58361542e+00 -7.07373440e-01 6.76045895e-01 7.44188800e-02
-1.08413684e+00 -1.14076845e-01 3.37506324e-01 -5.97090228e-04
1.16756892e+00 3.22485149e-01 5.43287516e-01 1.47023141e+00
2.19912082e-01 1.34291005e+00 1.44843054e+00 -2.13818684e-01
7.45415330e-01 -2.01783821e-01 -2.27933470e-02 2.63046503e-01
2.04611167e-01 7.96047568e-01 -3.92026514e-01 -1.02582693e-01
8.74517500e-01 -3.45105380e-02 6.55257180e-02 -5.04704773e-01
-8.95408034e-01 1.13320076e+00 6.23442829e-01 -5.38715459e-02
-2.77441144e-01 9.62165356e-01 3.21604729e-01 4.52344030e-01
1.44985151e-02 5.67779660e-01 -5.06359816e-01 -4.30411696e-01
-4.98657823e-01 6.81294918e-01 6.47408903e-01 1.15537894e+00
5.32544374e-01 3.95235866e-01 -5.77661812e-01 6.97781086e-01
3.05125117e-01 1.05812347e+00 6.19621396e-01 -1.12033200e+00
5.98701954e-01 -2.49303095e-02 2.14977324e-01 -3.01999241e-01
-4.35204893e-01 -7.12449312e-01 -5.16616464e-01 1.87400490e-01
3.12867731e-01 -1.31919682e-01 -8.08176935e-01 1.99725127e+00
-1.00297853e-01 -2.26346314e-01 2.96173453e-01 8.59479904e-01
4.90700722e-01 4.16937977e-01 1.67131707e-01 5.43312691e-02
7.24370778e-01 -7.55242288e-01 -4.01956648e-01 -5.49961925e-01
6.72988892e-01 6.53518289e-02 1.51522589e+00 6.31803632e-01
-9.03408647e-01 -2.46353209e-01 -1.08789682e+00 1.26264647e-01
-2.26920053e-01 -9.36161280e-02 9.34822857e-01 7.07896471e-01
-8.82976115e-01 7.11806893e-01 -8.17404032e-01 1.97128952e-01
6.40357077e-01 4.86318558e-01 9.34175551e-02 -2.21147493e-01
-1.16295791e+00 8.61614585e-01 9.26456228e-02 -2.25657551e-03
-1.47647452e+00 -5.47981441e-01 -5.63651145e-01 -7.46043324e-02
8.99224222e-01 -4.50962447e-02 1.82207441e+00 -6.17279351e-01
-1.78837144e+00 2.36149773e-01 1.04231834e-01 -7.52569497e-01
6.49767101e-01 -4.06855404e-01 -4.03878927e-01 -3.27564031e-02
1.00303933e-01 7.46383905e-01 8.16830218e-01 -1.28287828e+00
-5.57269692e-01 -3.12999457e-01 5.46039306e-02 1.41990051e-01
2.32394397e-01 -5.75471282e-01 -9.44428965e-02 -6.32113039e-01
-1.97818220e-01 -1.08201492e+00 -5.24665892e-01 -5.58770478e-01
-1.56609327e-01 -4.66669977e-01 2.71422595e-01 1.26850223e-02
1.11419654e+00 -1.85625064e+00 1.05186151e-02 3.75234663e-01
-5.96235916e-02 8.75046104e-02 -4.37973052e-01 2.13063836e-01
1.07175663e-01 8.83700699e-02 -3.06155011e-02 -2.07239017e-02
3.24460238e-01 7.95602202e-01 -8.82095814e-01 3.09267670e-01
1.88831598e-01 1.14556479e+00 -1.43803835e+00 -2.24684954e-01
-6.73073307e-02 -4.93909940e-02 -9.69699025e-01 2.66639352e-01
-6.30410194e-01 3.69452566e-01 -5.50560296e-01 7.85409391e-01
3.98436129e-01 -9.75492410e-03 2.09765926e-01 4.43143904e-01
9.79725830e-03 4.40313101e-01 -9.41558957e-01 1.34562027e+00
-5.16637743e-01 3.05489779e-01 -5.03416836e-01 -8.64720643e-01
8.25056612e-01 -1.39239788e-01 3.75158221e-01 -1.24691391e+00
1.77254081e-01 3.15100163e-01 7.95386434e-02 -1.27162650e-01
7.72098124e-01 8.76296610e-02 -4.47649539e-01 6.22488320e-01
9.74326059e-02 -1.26859769e-01 2.79993415e-02 2.62337416e-01
1.10826945e+00 7.45158732e-01 4.78237197e-02 -1.34696290e-01
1.84308607e-02 -1.65775597e-01 6.19798541e-01 1.51623809e+00
-3.56828809e-01 1.69599295e-01 6.35231793e-01 -1.17054477e-01
-6.80842221e-01 -1.49096012e+00 4.87291701e-02 1.65790665e+00
2.77558714e-02 -1.15352497e-01 -2.07130790e-01 -9.96900320e-01
4.47630197e-01 1.27981091e+00 -8.41859579e-01 -3.53981853e-01
-5.63534617e-01 -6.65956259e-01 8.35608423e-01 8.36426914e-01
3.44094068e-01 -1.08556914e+00 -8.92752707e-01 3.12192291e-01
-6.41020387e-02 -8.84531796e-01 -1.80238202e-01 8.75446498e-01
-1.04415023e+00 -7.49290764e-01 -6.31444633e-01 3.85656878e-02
3.70488912e-01 2.44985729e-01 1.31020272e+00 -3.11795443e-01
-1.26077160e-01 6.21079981e-01 -3.54051083e-01 -5.67293763e-01
-2.57798702e-01 4.82088588e-02 1.61404252e-01 -4.59229767e-01
3.81951094e-01 -4.72309142e-01 -7.48368859e-01 4.43252206e-01
-7.52926230e-01 -5.40134966e-01 8.10303807e-01 8.64417255e-01
7.84816265e-01 -4.94927764e-01 9.57693815e-01 -8.46326232e-01
7.62277722e-01 -5.59376299e-01 -8.35125923e-01 1.48529395e-01
-1.08047271e+00 6.19442046e-01 6.72643602e-01 -7.67617285e-01
-7.70381808e-01 -3.40327770e-01 -1.87617674e-01 -5.45619905e-01
2.69460112e-01 6.24196976e-02 4.05834764e-01 1.53072745e-01
8.69725466e-01 3.38996053e-01 -7.81164542e-02 -2.34040052e-01
7.17721820e-01 4.81484473e-01 5.71331918e-01 -1.21746278e+00
5.26371419e-01 3.16188335e-01 6.20276034e-02 -3.67195278e-01
-9.82694566e-01 -3.12210917e-01 1.12890676e-01 -1.96601361e-01
6.81675673e-01 -9.25096273e-01 -1.37435687e+00 4.63303775e-02
-2.80771524e-01 -1.04516912e+00 -7.63458073e-01 5.60748458e-01
-9.99272943e-01 -1.47080012e-02 -4.76197302e-01 -1.28497910e+00
-2.35009104e-01 -1.19221115e+00 6.61651969e-01 2.95469940e-01
7.65930302e-03 -6.43589795e-01 1.96269512e-01 -2.24149767e-02
6.72801733e-01 -1.89246237e-01 9.26545620e-01 -7.16250777e-01
-6.17136657e-01 3.19739506e-02 -4.63492312e-02 4.65761006e-01
-3.75146508e-01 -3.65987897e-01 -1.06469584e+00 -5.39783359e-01
-4.29969639e-01 -1.00411034e+00 8.35877538e-01 3.35522801e-01
1.39362812e+00 -1.34602278e-01 3.34475748e-02 4.39527690e-01
1.39852381e+00 1.99927360e-01 7.14295268e-01 5.22480667e-01
3.59520018e-01 2.12640371e-02 1.03491449e+00 6.80693209e-01
4.46058899e-01 5.72120249e-01 8.61910403e-01 2.95294642e-01
2.57410139e-01 -7.22297013e-01 6.49746299e-01 2.97001779e-01
-2.94584632e-01 -2.24900723e-01 -4.41006124e-01 3.35231632e-01
-2.05873799e+00 -8.57816696e-01 1.33105069e-01 2.54966474e+00
9.62416291e-01 3.40191275e-01 3.92258823e-01 -1.65220812e-01
2.73151956e-02 1.38904050e-01 -1.07365394e+00 -7.62861967e-01
-7.44987465e-03 5.99496007e-01 1.10904336e+00 2.43715778e-01
-8.06920290e-01 1.04135752e+00 6.71057224e+00 9.31004822e-01
-8.84648025e-01 8.03977326e-02 4.80888844e-01 -3.18519771e-01
-3.96599054e-01 2.35266592e-02 -1.19811654e+00 1.77923664e-01
1.03719437e+00 1.94129705e-01 1.04130220e+00 1.19368422e+00
-6.64447844e-02 -3.56034160e-01 -1.28581202e+00 7.13396192e-01
-1.93061426e-01 -9.26885009e-01 -8.65295008e-02 1.67660087e-01
7.22820759e-01 4.43900764e-01 3.24920774e-01 1.30820429e+00
1.21446824e+00 -1.25147927e+00 7.18503475e-01 4.72507864e-01
6.90975428e-01 -9.85484660e-01 4.37152237e-01 4.15666968e-01
-7.67531574e-01 -5.11035144e-01 -6.15672410e-01 1.29506573e-01
-5.35568357e-01 -1.86608452e-02 -6.98466361e-01 5.48039854e-01
8.04004252e-01 2.27632433e-01 -4.71452296e-01 6.56816542e-01
-4.73195970e-01 6.19079351e-01 -2.55586982e-01 -5.31749308e-01
6.81302786e-01 1.45411761e-02 2.84216166e-01 9.67786372e-01
8.28173757e-02 -2.11000174e-01 1.19873777e-01 9.57659662e-01
-7.22503476e-03 -1.73437610e-01 -6.10698462e-01 -1.35515435e-02
5.56437433e-01 8.83435011e-01 -4.07426953e-01 -1.30898863e-01
8.54086056e-02 4.95820731e-01 6.43242955e-01 3.78217369e-01
-1.05339932e+00 -2.50426620e-01 5.74019790e-01 -1.92902237e-01
4.79524463e-01 -1.95860893e-01 -3.26291029e-03 -1.03221750e+00
-3.50628495e-01 -1.05307245e+00 5.64881623e-01 -6.82919979e-01
-1.21255076e+00 4.00640935e-01 2.40763843e-01 -1.24796414e+00
-6.43632889e-01 -6.42062128e-01 -2.16367573e-01 4.70555037e-01
-1.83979011e+00 -6.39739811e-01 1.07817546e-01 6.81844294e-01
3.56281668e-01 -1.60633802e-01 8.65703285e-01 -2.69775577e-02
-4.26353753e-01 1.00363517e+00 3.40343446e-01 -1.50648728e-01
6.50207341e-01 -1.56176805e+00 1.92338258e-01 4.67076361e-01
1.19707033e-01 4.64995563e-01 5.69082379e-01 -3.94622862e-01
-1.77693808e+00 -1.00724936e+00 1.17449544e-01 -4.44914848e-01
5.83006442e-01 -3.26981425e-01 -5.30071199e-01 8.72551024e-01
-3.11543614e-01 2.32406199e-01 4.95644778e-01 3.14293861e-01
-6.40013516e-01 -3.34544867e-01 -1.09472513e+00 1.09738910e+00
9.91642237e-01 -3.88537407e-01 -5.75063765e-01 2.08182648e-01
6.90591395e-01 -4.10970181e-01 -7.99112320e-01 3.13140094e-01
5.84035456e-01 -9.05204356e-01 1.12521040e+00 -9.00898933e-01
2.64476538e-01 5.03670648e-02 -3.88934165e-01 -1.37301338e+00
-1.36927396e-01 -7.84169853e-01 -4.02092040e-01 6.92099154e-01
3.20388615e-01 -8.23246181e-01 7.34868944e-01 3.04998994e-01
-4.87297401e-02 -9.64804411e-01 -9.59773362e-01 -1.20332861e+00
3.75266731e-01 -9.96196389e-01 5.21124065e-01 4.20999974e-01
-8.68199319e-02 1.25540718e-01 -4.93193656e-01 -1.06946133e-01
6.16298497e-01 1.79380625e-01 9.15856063e-01 -8.04072082e-01
-7.60555863e-01 -4.29729134e-01 2.53735155e-01 -1.59271765e+00
4.20658430e-03 -9.41384375e-01 3.69592547e-01 -1.12285066e+00
-1.27897456e-01 -8.25238705e-01 -5.74267626e-01 6.13175273e-01
-3.72512229e-02 3.79489101e-02 2.31398761e-01 7.15785921e-02
-9.63913560e-01 8.90013158e-01 1.46438563e+00 1.11001909e-01
-3.48986864e-01 3.71354163e-01 -7.71264493e-01 3.76253456e-01
8.05798829e-01 -7.22904027e-01 -6.66567266e-01 -9.78466570e-02
4.94760096e-01 1.33806482e-01 4.06692266e-01 -8.36568415e-01
-9.65756774e-02 -3.92539233e-01 3.61021280e-01 -7.48771966e-01
2.98299968e-01 -5.92268765e-01 -4.87771869e-01 5.98872483e-01
-7.47664571e-01 2.01509044e-01 1.38998911e-01 9.48140144e-01
4.71934527e-01 -3.28470320e-01 6.73699439e-01 -1.53576553e-01
-4.37592834e-01 1.72572553e-01 -7.12432027e-01 6.50326014e-01
7.92138278e-01 1.98328882e-01 -5.43300211e-01 -5.50920844e-01
-4.88719612e-01 6.61260843e-01 -8.71930569e-02 3.87760729e-01
6.98944569e-01 -1.25090361e+00 -3.56032282e-01 2.12341219e-01
2.62504816e-01 6.65966049e-02 1.42432958e-01 7.74540067e-01
-2.26979539e-01 5.17849028e-01 -9.68882069e-02 -6.00398362e-01
-5.87142944e-01 4.99856710e-01 2.17003763e-01 -7.83223450e-01
-6.33543432e-01 7.22154796e-01 -2.64569610e-01 -6.75100386e-01
5.98239958e-01 -6.92613423e-01 -1.60997873e-03 -3.43510240e-01
8.35201368e-02 3.51328820e-01 -8.05390924e-02 1.64849654e-01
-6.66025430e-02 2.44517066e-02 -3.45721275e-01 -4.06227380e-01
1.22842216e+00 1.86396062e-01 5.74975550e-01 6.90747678e-01
6.95040584e-01 -1.24057271e-01 -1.51580775e+00 -1.92196876e-01
4.83905412e-02 -4.91538525e-01 2.21502498e-01 -9.98795867e-01
-7.31004477e-01 7.36840904e-01 6.35619283e-01 2.82823928e-02
7.61954367e-01 -3.21131140e-01 5.26857018e-01 1.05487871e+00
6.71233118e-01 -1.43053401e+00 3.86572063e-01 7.43667364e-01
6.87058389e-01 -1.25025761e+00 2.22829476e-01 4.16648805e-01
-1.05768704e+00 8.10422122e-01 7.20401168e-01 -4.64069635e-01
3.47519130e-01 3.52346040e-02 -2.93222517e-01 1.57076433e-01
-1.08662307e+00 -4.16831315e-01 -1.29355088e-01 6.16631567e-01
4.19893116e-02 2.17316255e-01 -3.35952267e-02 5.32569110e-01
-1.14275940e-01 1.26208842e-01 3.60087037e-01 1.20767903e+00
-2.28494599e-01 -1.08121717e+00 -1.64963622e-02 5.02171814e-01
-5.05257726e-01 -6.25710264e-02 1.41710237e-01 1.28272831e+00
-5.20034134e-01 9.01107550e-01 1.26721084e-01 -4.97877836e-01
3.32127482e-01 -2.49519959e-01 6.87856257e-01 -3.34908575e-01
-6.59695625e-01 7.25594908e-02 3.86735126e-02 -7.79756308e-01
3.85425389e-02 -4.27002966e-01 -1.49547279e+00 -2.58105278e-01
-1.48122028e-01 2.08576351e-01 5.90009511e-01 9.28061426e-01
2.02069730e-01 6.81292772e-01 8.29110026e-01 -6.46963954e-01
-1.49724233e+00 -7.91392028e-01 -5.74605763e-01 3.11964042e-02
3.87564451e-01 -7.27828205e-01 -4.11078304e-01 -9.76141274e-01]
|
[4.091960906982422, 2.0774753093719482]
|
835199ca-ef1f-4f9d-bc0f-45faea67b43e
|
exploiting-structure-for-fast-kernel-learning
|
1808.03351
| null |
http://arxiv.org/abs/1808.03351v1
|
http://arxiv.org/pdf/1808.03351v1.pdf
|
Exploiting Structure for Fast Kernel Learning
|
We propose two methods for exact Gaussian process (GP) inference and learning
on massive image, video, spatial-temporal, or multi-output datasets with
missing values (or "gaps") in the observed responses. The first method ignores
the gaps using sparse selection matrices and a highly effective low-rank
preconditioner is introduced to accelerate computations. The second method
introduces a novel approach to GP training whereby response values are inferred
on the gaps before explicitly training the model. We find this second approach
to be greatly advantageous for the class of problems considered. Both of these
novel approaches make extensive use of Kronecker matrix algebra to design
massively scalable algorithms which have low memory requirements. We
demonstrate exact GP inference for a spatial-temporal climate modelling problem
with 3.7 million training points as well as a video reconstruction problem with
1 billion points.
|
['Trefor W. Evans', 'Prasanth B. Nair']
|
2018-08-09
| null | null | null | null |
['video-reconstruction']
|
['computer-vision']
|
[ 2.85911888e-01 -1.80959493e-01 3.14642489e-01 -1.33102283e-01
-1.06313491e+00 -5.52528858e-01 7.44271815e-01 1.23444004e-02
-5.16885996e-01 8.96099925e-01 5.78823611e-02 -6.20932460e-01
-2.54844099e-01 -8.76161695e-01 -1.11720335e+00 -9.88972068e-01
-3.04159492e-01 6.97995603e-01 5.40559413e-03 3.19944888e-01
2.37259775e-01 2.31550992e-01 -1.43663514e+00 3.21869701e-02
5.99249959e-01 1.02712762e+00 4.04881418e-01 1.06085372e+00
2.58650612e-02 8.36168587e-01 -4.42716526e-03 -3.31132263e-01
5.47183335e-01 6.29546642e-02 -3.45045686e-01 1.25506312e-01
3.71912688e-01 -1.63140774e-01 -3.47869039e-01 7.96773255e-01
7.21634984e-01 2.55495757e-01 6.03636563e-01 -1.10161972e+00
-2.82096148e-01 1.75856680e-01 -8.12757969e-01 2.98299398e-02
2.88178295e-01 3.75295460e-01 8.24280322e-01 -1.17134786e+00
3.59553307e-01 1.38746500e+00 1.07561171e+00 2.01521013e-02
-1.75470150e+00 -3.85327250e-01 -5.85864969e-02 -1.53776467e-01
-1.46812701e+00 -4.72926825e-01 1.35741115e-01 -8.82000506e-01
9.80408192e-01 2.62846708e-01 5.17852724e-01 1.00948811e+00
1.03934228e-01 4.32888031e-01 1.30520332e+00 -9.91538689e-02
5.50494075e-01 -1.44836828e-01 -8.13530833e-02 7.01330662e-01
6.26359209e-02 4.60796028e-01 -5.07575989e-01 -9.18615460e-01
9.39924121e-01 -5.46494946e-02 -1.81703642e-01 -1.84824020e-01
-1.29273796e+00 9.69654262e-01 -5.66753447e-02 -5.76818764e-01
-6.95185959e-01 4.94928896e-01 2.06440479e-01 2.14103252e-01
6.05381191e-01 5.82415834e-02 -5.43590188e-01 -2.40425810e-01
-9.97937322e-01 5.09285688e-01 1.11995947e+00 8.96815538e-01
1.05148697e+00 2.55045265e-01 -1.26191214e-01 7.28605568e-01
4.03117955e-01 1.02055728e+00 -1.24590509e-01 -1.31797349e+00
6.88191831e-01 -1.70494482e-01 4.13078219e-01 -1.13642728e+00
-2.82228291e-01 -2.03741819e-01 -1.07546902e+00 8.63510221e-02
4.90129381e-01 -6.16483986e-01 -8.94655764e-01 1.51937771e+00
6.22037351e-01 7.35013306e-01 4.36586812e-02 5.66826046e-01
4.49009091e-01 1.05338109e+00 1.29086792e-01 -2.93382853e-01
1.19742620e+00 -5.27924061e-01 -2.62649655e-01 -1.95042297e-01
3.44142854e-01 -5.84283531e-01 9.08258736e-01 3.71260375e-01
-9.28589523e-01 -4.29112673e-01 -5.21444321e-01 3.14476907e-01
-7.31167644e-02 1.93205655e-01 8.59041333e-01 4.80549902e-01
-1.25289786e+00 7.40192354e-01 -1.01142561e+00 -3.46804745e-02
2.80129880e-01 5.10408342e-01 -3.05074632e-01 -1.28408745e-01
-8.77587795e-01 4.38920617e-01 2.10620627e-01 1.70241803e-01
-1.00912762e+00 -9.45587635e-01 -6.87362134e-01 1.39452964e-01
4.13631350e-01 -1.02680755e+00 9.55257952e-01 -5.62988818e-01
-1.39666224e+00 5.86943209e-01 -2.81096369e-01 -5.85027575e-01
5.83141804e-01 -3.92991938e-02 -1.03474092e-02 8.05530101e-02
-7.67682567e-02 6.08556449e-01 1.16047227e+00 -9.46609139e-01
-6.47857308e-01 -3.63887817e-01 -4.33949530e-01 2.02433854e-01
2.45090172e-01 1.41186714e-02 -3.17478240e-01 -5.39041340e-01
2.76569784e-01 -1.19859803e+00 -7.08081186e-01 -2.50370000e-02
-1.16660394e-01 7.93676153e-02 4.99123245e-01 -9.31976438e-01
7.21231341e-01 -2.00169253e+00 2.36386612e-01 3.63662571e-01
5.34738302e-02 -1.15188487e-01 -2.12710023e-01 5.66259921e-01
2.86619067e-01 -3.22185419e-02 -4.07971889e-01 -7.15806246e-01
-2.31231824e-02 8.03274512e-01 -5.15367985e-01 5.66869020e-01
1.04933776e-01 6.22319460e-01 -7.68819332e-01 -3.04675281e-01
1.54417485e-01 4.93023276e-01 -6.88843787e-01 2.29826525e-01
-2.88492709e-01 7.53729522e-01 -3.11661631e-01 3.87420595e-01
7.91395843e-01 -3.75337332e-01 -5.12313209e-02 1.30744934e-01
-2.19601631e-01 -1.38670038e-02 -1.63757682e+00 1.35048056e+00
-3.35817546e-01 2.74031997e-01 4.32949692e-01 -7.40050912e-01
5.76630116e-01 5.31700611e-01 6.16361618e-01 -1.83040455e-01
-2.90251464e-01 1.21827900e-01 -5.58491588e-01 -3.65621984e-01
4.70851004e-01 -3.15361440e-01 1.40267402e-01 2.34456316e-01
-1.16504617e-02 -3.18513870e-01 1.45607427e-01 1.47530854e-01
1.02410567e+00 3.40599418e-01 9.07381997e-02 -2.88887650e-01
2.47935936e-01 9.27460715e-02 8.84266078e-01 1.28311348e+00
2.19588965e-01 6.65268660e-01 6.45300329e-01 -5.28506994e-01
-1.21741450e+00 -1.17102838e+00 -3.30824219e-02 1.04628277e+00
-5.57676017e-01 -3.62530798e-01 -3.04214716e-01 2.07138322e-02
3.34382743e-01 3.99869025e-01 -4.48639363e-01 4.25545186e-01
-2.89814144e-01 -1.24399877e+00 4.38875705e-01 4.47072357e-01
3.91186208e-01 -7.05495536e-01 -2.78602302e-01 2.09361151e-01
-2.35805690e-01 -1.03144622e+00 -1.01400353e-01 2.32458711e-01
-1.07970214e+00 -7.44325459e-01 -5.50312638e-01 -3.41542512e-01
5.62129855e-01 9.54036489e-02 9.95865703e-01 -5.22157550e-01
5.19965142e-02 5.92713594e-01 -3.67599219e-04 -3.47336531e-01
-2.53633916e-01 -3.42008322e-01 2.20790505e-01 -7.34207733e-03
1.08614296e-01 -7.96316803e-01 -3.53527337e-01 1.08842896e-02
-5.61324954e-01 9.09373686e-02 5.69487095e-01 1.19685864e+00
9.36252952e-01 1.41123310e-02 1.99167252e-01 -7.21542537e-01
4.95355248e-01 -7.32155323e-01 -1.12614691e+00 1.06617063e-01
-4.36998546e-01 1.52257830e-01 4.17619526e-01 -5.25363743e-01
-1.12663174e+00 5.62006652e-01 -1.93746597e-01 -6.44714653e-01
-9.32623670e-02 8.50604713e-01 1.36458561e-01 -1.96652055e-01
6.47017598e-01 3.54991555e-01 6.62613055e-03 -5.91463089e-01
2.58674115e-01 3.10751587e-01 7.08490193e-01 -9.92882311e-01
6.42358601e-01 5.48708975e-01 5.02414584e-01 -1.07786381e+00
-4.91549164e-01 -6.26995265e-01 -4.44074214e-01 1.81158200e-01
7.05195367e-01 -1.45593977e+00 -7.80581713e-01 4.71857160e-01
-9.89584267e-01 -7.41805017e-01 -3.29384178e-01 7.91248679e-01
-7.97179103e-01 5.57572126e-01 -8.38808894e-01 -1.05442250e+00
-3.36448014e-01 -8.94637883e-01 1.27838123e+00 -3.58616054e-01
1.51468888e-01 -9.99192894e-01 3.57814640e-01 1.16573587e-01
2.51908183e-01 1.05598018e-01 6.55610919e-01 -5.95763512e-02
-6.83447123e-01 2.42602713e-02 -2.94196844e-01 2.14733079e-01
-2.80614763e-01 2.58517601e-02 -9.50007558e-01 -4.05589968e-01
2.83385992e-01 -4.01730463e-02 8.93028796e-01 7.77517438e-01
1.12792957e+00 -4.60559666e-01 -1.76965356e-01 8.79026890e-01
1.60285735e+00 -2.45707110e-01 6.15874171e-01 -4.91942577e-02
8.53811562e-01 5.29974043e-01 3.66171688e-01 8.80507529e-01
3.49333704e-01 3.64492536e-01 2.49667004e-01 -2.05109902e-02
4.45735127e-01 -2.70626515e-01 3.40604335e-01 7.14480102e-01
-3.09041709e-01 -1.46824140e-02 -1.04121816e+00 5.00107467e-01
-2.14895225e+00 -1.20627689e+00 -5.01340568e-01 2.65729594e+00
7.40693748e-01 -3.57121319e-01 -1.84944775e-02 -2.99020916e-01
4.38941956e-01 1.26928523e-01 -3.18710387e-01 -1.30260214e-01
-2.38534659e-01 2.04336032e-01 9.48224127e-01 5.81593990e-01
-1.27303040e+00 6.64572775e-01 7.17622471e+00 7.47843683e-01
-6.82247043e-01 2.86600500e-01 5.99039733e-01 -1.91492945e-01
-3.01715136e-02 2.40219206e-01 -8.21267068e-01 5.92522681e-01
1.30914533e+00 1.94959134e-01 6.09390259e-01 6.64062679e-01
6.03701234e-01 -4.50365424e-01 -9.19231772e-01 1.29375494e+00
-4.30797130e-01 -1.25615060e+00 -2.75333792e-01 2.92670399e-01
9.17246342e-01 5.19545615e-01 6.28308430e-02 2.53224790e-01
9.56932187e-01 -7.92831779e-01 4.43131208e-01 7.06739485e-01
6.16742134e-01 -6.53313875e-01 4.07147765e-01 5.41275501e-01
-1.08250594e+00 -1.26523852e-01 -8.11823905e-01 -3.06263030e-01
3.48549902e-01 9.50817823e-01 -4.96757686e-01 3.16008478e-01
8.12413156e-01 5.36925852e-01 -3.43371063e-01 1.13376129e+00
-1.09015331e-01 9.80335772e-01 -8.85148704e-01 4.17779654e-01
2.97562599e-01 -5.91183901e-01 6.80453956e-01 1.22945786e+00
7.99153626e-01 1.67671472e-01 2.84630805e-01 6.72484100e-01
2.86210477e-01 -7.06483200e-02 -7.79964089e-01 1.95805416e-01
2.15867311e-01 1.04764318e+00 -3.33049089e-01 -4.82945621e-01
-6.83998048e-01 7.72734702e-01 2.41436735e-01 6.27779305e-01
-5.76938033e-01 2.97668904e-01 7.93713272e-01 -6.98357970e-02
6.06849313e-01 -5.70321321e-01 -1.41266033e-01 -1.30118799e+00
-1.59551919e-01 -8.20893764e-01 4.94688123e-01 -9.05769467e-01
-1.46368003e+00 -8.80204588e-02 5.40149957e-02 -8.69024992e-01
-5.76176226e-01 -5.72493851e-01 -3.55990261e-01 1.22506881e+00
-1.14229548e+00 -1.11320519e+00 -1.10692708e-02 6.00029826e-01
1.20823085e-01 4.63459902e-02 8.31609726e-01 2.47312099e-01
-3.28517139e-01 -1.86050817e-01 5.52654147e-01 -3.60332638e-01
3.81510884e-01 -1.38841033e+00 5.22486150e-01 8.32413077e-01
1.15504041e-01 5.72362244e-01 8.96731079e-01 -9.00944829e-01
-1.77599049e+00 -1.07403278e+00 7.63356507e-01 -4.28452075e-01
7.87642419e-01 -4.09378141e-01 -8.32349956e-01 8.64380360e-01
-7.50753656e-02 4.32868861e-02 7.60140002e-01 3.69143248e-01
-2.24134952e-01 6.64717108e-02 -9.44563031e-01 5.31012654e-01
6.48561180e-01 -5.87911785e-01 -3.43232632e-01 7.44586885e-01
4.80080545e-01 -4.87537831e-01 -1.01620543e+00 2.56752461e-01
4.16630745e-01 -5.14853001e-01 1.15387726e+00 -4.80701178e-01
1.33636817e-01 -3.68227065e-01 -5.14024019e-01 -1.31178761e+00
-4.12118971e-01 -7.29758084e-01 -4.87107366e-01 8.89343619e-01
3.84507895e-01 -7.60250390e-01 8.24238539e-01 7.82911003e-01
1.54958084e-01 -4.50652480e-01 -8.74544978e-01 -6.21809542e-01
1.27035165e-02 -5.81722140e-01 4.88318086e-01 7.81833768e-01
-4.57021177e-01 1.75583586e-01 -9.94688213e-01 5.07291317e-01
9.99189854e-01 1.17475666e-01 1.01873326e+00 -1.42768610e+00
-8.91975582e-01 9.43416059e-02 -2.77260453e-01 -1.20529974e+00
-6.29285425e-02 -4.86547798e-01 1.09817684e-01 -1.34899926e+00
2.85621494e-01 -6.10681355e-01 8.28630477e-02 3.66645455e-01
-1.90039515e-01 -1.58984754e-02 -7.20145702e-02 3.32317501e-01
-2.80902833e-01 4.94404942e-01 7.73069501e-01 8.50722417e-02
-1.19386926e-01 3.42080921e-01 -2.81370491e-01 4.93359983e-01
7.61044204e-01 -6.35553777e-01 -4.25354719e-01 -5.73744297e-01
5.35547912e-01 4.60753024e-01 8.04239392e-01 -9.75013375e-01
4.62991774e-01 -4.24646109e-01 4.65041488e-01 -8.54369581e-01
5.91199756e-01 -6.47444129e-01 8.13945770e-01 2.97361314e-01
-2.94946525e-02 2.62830257e-01 2.54141778e-01 1.06019354e+00
-1.02169169e-02 -1.13312803e-01 5.52694976e-01 -4.57333714e-01
-6.85419559e-01 4.95571494e-01 -5.09779453e-01 -1.62010625e-01
6.57703280e-01 6.02046885e-02 -3.62226218e-02 -5.46147048e-01
-1.12777412e+00 4.11820233e-01 3.46329272e-01 -1.63938865e-01
4.45544392e-01 -1.02305198e+00 -9.80409920e-01 2.41076782e-01
-2.92512149e-01 6.21255450e-02 4.86559182e-01 1.11882460e+00
-4.74743843e-01 3.26805860e-01 8.47622156e-02 -9.10041451e-01
-1.13750184e+00 5.88900268e-01 1.16786152e-01 -4.04880375e-01
-5.65256000e-01 8.23385060e-01 1.35546178e-01 -7.10682154e-01
7.94028416e-02 -1.54479846e-01 3.96751672e-01 -4.46877033e-02
4.41314429e-01 5.53667247e-01 -1.59988835e-01 -6.93765342e-01
2.05291063e-01 3.53463262e-01 5.11420071e-01 -4.99073565e-01
1.51528692e+00 -3.19678307e-01 -2.04851016e-01 5.49970865e-01
8.97005439e-01 -2.14372456e-01 -1.84969044e+00 -4.78698015e-01
-7.14180246e-02 -6.99462235e-01 1.07720047e-01 -3.23863119e-01
-7.40761101e-01 1.04430699e+00 3.98362309e-01 -7.09381923e-02
8.25748801e-01 -2.97283232e-01 2.81813711e-01 7.47850060e-01
5.91758549e-01 -1.17638707e+00 -2.58745700e-01 6.37572944e-01
6.79815650e-01 -1.30427670e+00 2.40356803e-01 -2.89031297e-01
-5.10132730e-01 6.36759460e-01 6.03482276e-02 7.77537450e-02
7.27169514e-01 3.78800750e-01 -4.29412663e-01 -1.73070252e-01
-1.02748346e+00 -2.14417055e-01 -1.52746946e-01 7.01326132e-01
-3.08894850e-02 1.87010169e-01 -1.08148851e-01 4.43120226e-02
4.66698520e-02 -3.61940265e-03 2.74664819e-01 8.69495511e-01
-4.32692379e-01 -8.63701046e-01 -6.69571877e-01 8.05307209e-01
-5.16933978e-01 -3.79039228e-01 2.42972538e-01 3.63775104e-01
-1.67325750e-01 8.48333538e-01 6.67620823e-02 -1.03822798e-01
-1.30472211e-02 1.94658488e-01 3.74642581e-01 -4.95879024e-01
-4.21256751e-01 2.98732191e-01 1.30169451e-01 -5.53057194e-01
-2.74856329e-01 -1.06983840e+00 -7.75281787e-01 -6.24741971e-01
-6.81219697e-02 1.22380771e-01 7.25886106e-01 6.45768821e-01
5.07434309e-01 -5.72113767e-02 4.92540538e-01 -1.15767348e+00
-9.43584979e-01 -8.24675798e-01 -8.17424774e-01 1.46101981e-01
1.58990845e-01 -5.54228127e-01 -5.03807127e-01 1.60175353e-01]
|
[6.9478440284729, 3.8501272201538086]
|
b49da377-eb91-4c3b-9df2-9e894e5091e9
|
laeo-net-revisiting-people-looking-at-each
|
2101.02136
| null |
https://arxiv.org/abs/2101.02136v1
|
https://arxiv.org/pdf/2101.02136v1.pdf
|
LAEO-Net++: revisiting people Looking At Each Other in videos
|
Capturing the 'mutual gaze' of people is essential for understanding and interpreting the social interactions between them. To this end, this paper addresses the problem of detecting people Looking At Each Other (LAEO) in video sequences. For this purpose, we propose LAEO-Net++, a new deep CNN for determining LAEO in videos. In contrast to previous works, LAEO-Net++ takes spatio-temporal tracks as input and reasons about the whole track. It consists of three branches, one for each character's tracked head and one for their relative position. Moreover, we introduce two new LAEO datasets: UCO-LAEO and AVA-LAEO. A thorough experimental evaluation demonstrates the ability of LAEO-Net++ to successfully determine if two people are LAEO and the temporal window where it happens. Our model achieves state-of-the-art results on the existing TVHID-LAEO video dataset, significantly outperforming previous approaches. Finally, we apply LAEO-Net++ to a social network, where we automatically infer the social relationship between pairs of people based on the frequency and duration that they LAEO, and show that LAEO can be a useful tool for guided search of human interactions in videos. The code is available at https://github.com/AVAuco/laeonetplus.
|
['Andrew Zisserman', 'Pablo Medina-Suarez', 'Vicky Kalogeiton', 'Manuel J. Marin-Jimenez']
|
2021-01-06
| null | null | null | null |
['mutual-gaze']
|
['computer-vision']
|
[-3.65421832e-01 -3.17288011e-01 -1.02138273e-01 -3.62370998e-01
-1.30246812e-02 -5.75804174e-01 5.45443475e-01 -3.31037194e-02
-3.91183048e-01 3.02162111e-01 3.75912398e-01 2.43439861e-02
1.46791600e-02 -6.10048234e-01 -7.35525489e-01 -4.78509992e-01
-5.27615786e-01 5.24807751e-01 2.84285188e-01 -2.32505620e-01
-1.76741891e-02 2.17333093e-01 -1.51791954e+00 4.32299405e-01
1.74589485e-01 8.71153355e-01 -1.83760211e-01 1.08504438e+00
5.59068501e-01 1.18232858e+00 -6.37314320e-01 -4.35666770e-01
6.05308749e-02 -4.16249603e-01 -7.36197054e-01 1.25034899e-01
7.84730613e-01 -6.37080729e-01 -8.98468614e-01 7.42671609e-01
4.75686669e-01 6.87609762e-02 3.64550412e-01 -1.68317676e+00
-4.62896466e-01 6.74243689e-01 -5.81007838e-01 8.02580595e-01
9.86239374e-01 4.30019081e-01 9.63234127e-01 -7.60622919e-01
6.28409982e-01 1.33722699e+00 9.01682138e-01 6.02508247e-01
-5.19320488e-01 -7.87196338e-01 1.90560579e-01 3.85923833e-01
-1.42221141e+00 -3.33216190e-01 5.56517959e-01 -5.63074172e-01
7.32319772e-01 3.50650102e-01 1.00716853e+00 1.43889558e+00
2.08312813e-02 1.23499656e+00 6.23139739e-01 -2.21246138e-01
-3.42769265e-01 -3.66789997e-01 3.41729224e-01 9.33872044e-01
-1.41959563e-02 -1.52365804e-01 -1.01389086e+00 9.68849361e-02
5.74219227e-01 1.62294880e-01 -3.46083969e-01 7.46585056e-02
-1.44447279e+00 3.38982642e-01 5.01463830e-01 2.72436917e-01
-3.26630890e-01 4.82782573e-01 3.55298102e-01 1.31084919e-01
5.21779954e-01 5.85385598e-02 1.79261193e-02 -4.85558271e-01
-6.09624445e-01 5.94281077e-01 7.87313938e-01 9.53962624e-01
2.50702709e-01 -6.11016393e-01 -2.65755087e-01 5.36628544e-01
3.51823092e-01 3.25393975e-01 7.48585388e-02 -1.12001908e+00
5.74251294e-01 5.98003387e-01 3.95133734e-01 -1.38077164e+00
-6.02538466e-01 -6.06483296e-02 -5.67637861e-01 -1.43222526e-01
7.14890897e-01 -3.68265539e-01 -4.27494556e-01 1.73882806e+00
3.56742889e-01 3.20276409e-01 -2.79435664e-01 9.93722558e-01
1.11454284e+00 8.21136594e-01 -3.57990712e-01 2.03873105e-02
1.55597985e+00 -1.45086753e+00 -8.03522646e-01 -1.95349395e-01
7.31426597e-01 -4.52166140e-01 6.20395482e-01 2.33332708e-01
-1.06606734e+00 -5.24069846e-01 -4.27991331e-01 -2.32710373e-02
-1.86492443e-01 2.47990996e-01 6.04270101e-01 3.29880208e-01
-1.17881274e+00 4.35566187e-01 -9.12853241e-01 -7.58730233e-01
7.63105035e-01 3.83456439e-01 -2.63206184e-01 7.48350769e-02
-1.18576229e+00 2.55336791e-01 8.54504555e-02 5.28899729e-01
-8.97697389e-01 -3.90553445e-01 -6.77175045e-01 7.17597157e-02
6.32580340e-01 -5.52272201e-01 1.48013270e+00 -1.00570750e+00
-1.00811350e+00 1.06127310e+00 -4.67161208e-01 -6.78488612e-01
9.08691406e-01 -6.44589901e-01 -5.58005989e-01 3.21147621e-01
2.83686310e-01 8.26222301e-01 4.58391428e-01 -9.29549396e-01
-9.38155532e-01 -1.90004542e-01 4.34770614e-01 1.64964482e-01
-3.45556587e-01 4.73371804e-01 -1.05348516e+00 -5.44395506e-01
-5.91795862e-01 -1.12660170e+00 2.93742150e-01 2.62563646e-01
-6.10793829e-01 -8.86155367e-01 7.19682813e-01 -5.23131311e-01
1.67482114e+00 -2.03797460e+00 -1.07233293e-01 3.66252996e-02
8.76854360e-01 6.00902021e-01 1.23604096e-01 6.12414718e-01
1.29436940e-01 3.30506737e-05 2.81020492e-01 -5.79578519e-01
6.75456896e-02 -1.11167632e-01 1.85834259e-01 7.07434356e-01
-1.47267237e-01 1.03924751e+00 -1.13489461e+00 -5.22387743e-01
9.14372876e-02 5.69213986e-01 -3.57003510e-01 3.30794811e-01
-2.44929697e-02 4.51369554e-01 -2.52445519e-01 5.86525023e-01
3.64311516e-01 -6.09603524e-01 8.84650573e-02 1.06155992e-01
-1.70051113e-01 1.94672287e-01 -8.00856531e-01 1.02634621e+00
5.07502025e-03 1.41829169e+00 -2.05106661e-02 -4.18073565e-01
4.55545992e-01 3.17902595e-01 6.03457093e-01 -4.71834242e-01
5.20520568e-01 -1.63692310e-02 3.71849239e-02 -8.51269066e-01
3.59119207e-01 7.57269681e-01 6.88989647e-03 6.80886984e-01
-2.13655904e-01 9.32976544e-01 6.32626712e-01 4.42208260e-01
1.03340769e+00 6.45787269e-02 2.67145813e-01 -8.45402777e-02
6.14677429e-01 -3.48193437e-01 5.27279615e-01 9.21300709e-01
-6.98056400e-01 5.79342782e-01 6.87771142e-01 -9.48707044e-01
-7.01034844e-01 -5.75568914e-01 3.71515393e-01 1.32939458e+00
4.19407606e-01 -7.87795365e-01 -1.07447350e+00 -5.90774477e-01
-1.67865977e-01 2.36893892e-01 -1.18802416e+00 2.78905660e-01
-1.02284658e+00 -3.51733685e-01 6.31788731e-01 6.47847354e-01
6.68929398e-01 -1.32977581e+00 -7.71494389e-01 -1.20380454e-01
-6.81802213e-01 -1.35351908e+00 -1.23655081e+00 -5.92239678e-01
-6.54119775e-02 -1.47048903e+00 -8.05740654e-01 -7.12462008e-01
5.50173521e-01 6.86529398e-01 1.38063598e+00 5.29501438e-01
-1.22604631e-01 6.49171472e-01 -4.36613232e-01 -4.78367984e-01
-8.68343487e-02 1.39296055e-01 1.36718571e-01 5.90424240e-01
9.05504644e-01 -3.28967839e-01 -9.10398543e-01 7.59984434e-01
-3.47273439e-01 4.87101078e-02 -2.09988296e-01 2.08603472e-01
-4.29545492e-02 -4.86188643e-02 -7.23580793e-02 -6.74232483e-01
4.28755611e-01 -6.59713984e-01 -2.78303504e-01 1.44867957e-01
-2.11050250e-02 -6.74473822e-01 2.25929961e-01 -4.07844365e-01
-6.27302468e-01 -1.39823318e-01 3.51317190e-02 -5.99054515e-01
-3.09931010e-01 1.18195698e-01 1.18386321e-01 3.76338959e-01
3.72842789e-01 -4.77815047e-02 -1.71928361e-01 -1.70020849e-01
-2.13829562e-01 6.31420016e-01 8.10635328e-01 -7.26631954e-02
4.91111219e-01 8.47919047e-01 -2.99724132e-01 -8.94145846e-01
-1.24488807e+00 -6.88182831e-01 -8.76776278e-01 -9.43576932e-01
1.19352567e+00 -1.08479607e+00 -1.61191630e+00 1.12148213e+00
-1.18342960e+00 -5.95191002e-01 2.76757121e-01 3.86623651e-01
-1.33203268e-01 2.49378994e-01 -6.74813926e-01 -8.51156175e-01
-2.08444998e-01 -7.71475315e-01 1.00008094e+00 4.50089276e-01
-5.37642658e-01 -1.15569949e+00 -6.56667165e-03 7.46719956e-01
-8.57630819e-02 3.35552365e-01 -3.56782466e-01 -5.53221941e-01
-6.59117401e-01 -3.07778120e-01 -1.00602910e-01 -1.95115730e-01
-4.92575392e-02 2.51107872e-01 -6.56889319e-01 -4.95127141e-01
-4.54691082e-01 -2.02576980e-01 8.26071620e-01 4.61905301e-01
1.00677311e+00 -4.44077790e-01 -7.24557638e-01 5.69109321e-01
6.92222059e-01 1.56093106e-01 4.12647247e-01 4.97795552e-01
1.02263665e+00 6.10207200e-01 7.20588684e-01 5.50688982e-01
9.14102077e-01 8.84305894e-01 5.78929305e-01 1.48431346e-01
-4.93157245e-02 -1.43488750e-01 5.87941766e-01 6.43847585e-01
-5.12280643e-01 -9.97321546e-01 -1.06617439e+00 7.83325553e-01
-2.24117804e+00 -1.27613509e+00 -5.51595330e-01 1.71675289e+00
2.17719585e-01 1.29029408e-01 8.28870416e-01 -1.45261243e-01
1.17452729e+00 5.64410329e-01 -3.44749421e-01 1.11094348e-01
7.49614164e-02 -5.51430643e-01 3.08667392e-01 5.19999206e-01
-1.31754398e+00 7.17492282e-01 5.48016977e+00 3.32704514e-01
-8.45987082e-01 7.19049349e-02 4.67134327e-01 -4.78903174e-01
4.17120188e-01 -4.40303236e-01 -1.20321465e+00 7.77212143e-01
6.71274006e-01 2.20756873e-01 5.26354134e-01 3.51095438e-01
4.28658545e-01 -2.40865588e-01 -1.32007933e+00 1.22606564e+00
4.78266388e-01 -1.26969635e+00 -4.40800697e-01 1.70827970e-01
4.42791820e-01 1.62176266e-01 -4.82560731e-02 -1.73596397e-01
1.96215093e-01 -9.97851253e-01 9.87300754e-01 5.57449222e-01
3.45735401e-01 -6.42112792e-01 7.64982700e-01 3.01970661e-01
-1.43143725e+00 -1.79461807e-01 1.92072272e-01 -2.88459599e-01
4.25964564e-01 1.58984765e-01 -7.65203714e-01 1.99754030e-01
1.34912133e+00 1.36755073e+00 -7.82435060e-01 9.87481117e-01
-3.22608292e-01 6.37041330e-01 -3.76132548e-01 -3.04871053e-01
3.20456266e-01 2.11881772e-02 7.77906477e-01 1.30298662e+00
1.88697085e-01 1.26507863e-01 1.67348191e-01 5.02980292e-01
-2.15585276e-01 -5.00016332e-01 -4.78745610e-01 -9.10572559e-02
5.76014936e-01 9.30989861e-01 -8.33860159e-01 -3.89511794e-01
-4.38979536e-01 1.01369309e+00 4.04369652e-01 9.95499566e-02
-1.14770269e+00 -2.00299665e-01 5.27735174e-01 3.58562231e-01
5.80466211e-01 -1.66979700e-01 5.81119597e-01 -1.33474565e+00
1.32698655e-01 -8.34194660e-01 5.29262304e-01 -8.84568751e-01
-1.09174395e+00 7.03500509e-01 4.80258875e-02 -1.27317297e+00
-1.68943405e-01 -4.60270286e-01 -8.06406319e-01 3.59214455e-01
-1.04469001e+00 -1.22167361e+00 -6.88848495e-01 6.76891446e-01
8.15402985e-01 -1.51768342e-01 7.49890879e-02 5.08947194e-01
-9.36225057e-01 5.61120749e-01 -2.73255527e-01 8.32269192e-01
6.84368908e-01 -1.12490737e+00 8.66874278e-01 8.97830665e-01
4.96544838e-02 5.58429062e-01 8.39099765e-01 -5.07843673e-01
-1.02905917e+00 -9.91414368e-01 1.31053019e+00 -1.00963378e+00
6.35068953e-01 -6.34457469e-01 -6.33858562e-01 1.21638095e+00
4.47460651e-01 2.34456912e-01 7.22900867e-01 1.15026794e-01
-1.24214403e-01 6.39669746e-02 -5.48576713e-01 6.84059918e-01
1.51397252e+00 -4.85186517e-01 -4.35387760e-01 5.09531260e-01
4.30301249e-01 -6.47285879e-01 -5.65845609e-01 -8.14811811e-02
9.58156943e-01 -1.55407393e+00 9.30594504e-01 -4.46990699e-01
5.52244663e-01 -2.19739288e-01 2.88711190e-01 -9.66472447e-01
-2.17550293e-01 -9.07548428e-01 -5.39806902e-01 1.24259412e+00
-1.86286774e-02 -3.49433780e-01 7.85089016e-01 2.47391582e-01
1.73407331e-01 -8.32792640e-01 -6.80307448e-01 -5.39565265e-01
-6.14787638e-01 -3.80249918e-01 4.32760626e-01 9.51155126e-01
-3.72936666e-01 3.37316632e-01 -9.64707434e-01 2.81763434e-01
6.21076882e-01 -1.26689523e-01 1.30181658e+00 -1.17963338e+00
-2.99536526e-01 -3.15719873e-01 -3.08983445e-01 -1.49007523e+00
6.80479854e-02 -3.73847425e-01 4.87722829e-02 -1.37765491e+00
2.78791875e-01 7.53891841e-02 -6.81725442e-02 3.65267277e-01
-1.31980583e-01 5.71871400e-01 3.87606353e-01 3.83736759e-01
-1.28288579e+00 1.81677058e-01 1.13600814e+00 3.15403938e-03
-2.91440129e-01 2.76474267e-01 -2.25008026e-01 1.02911639e+00
6.10485971e-01 -4.02359396e-01 5.26031032e-02 -4.67089087e-01
5.68935454e-01 6.04894273e-02 7.40271866e-01 -1.14049041e+00
6.38044000e-01 3.05066854e-01 1.67899311e-01 -9.76803541e-01
3.06145608e-01 -5.29134750e-01 -5.52849099e-02 6.23464763e-01
-4.22085226e-01 3.55756849e-01 -3.20108265e-01 7.17689812e-01
-1.88074276e-01 -3.75956446e-02 4.56403494e-01 -1.91986859e-01
-8.32860887e-01 5.47445655e-01 -7.20376372e-01 7.37545788e-02
1.21455610e+00 -2.17523575e-01 -3.90755951e-01 -8.62231076e-01
-7.38997638e-01 8.83507311e-01 2.12130174e-01 6.62045181e-01
3.53307873e-01 -1.34927297e+00 -5.46906531e-01 -2.13349774e-01
1.26582429e-01 -8.29867348e-02 4.95386839e-01 1.05722475e+00
-7.13848710e-01 5.16068995e-01 -1.79740146e-01 -8.07373703e-01
-2.00092435e+00 6.60814941e-01 5.16955256e-01 -1.38430580e-01
-6.13222539e-01 9.83840466e-01 2.74839491e-01 -3.43659490e-01
5.29485643e-01 -1.59124583e-01 -5.91911137e-01 3.04758936e-01
8.07599962e-01 6.47534132e-01 -6.83102787e-01 -1.15463805e+00
-6.69645250e-01 5.00139594e-01 -1.13629937e-01 2.24874735e-01
1.35813200e+00 -5.62918782e-01 -2.25817785e-01 6.30656838e-01
1.22644925e+00 1.14382934e-02 -1.49369168e+00 -3.75874847e-01
-3.22409719e-01 -6.30374789e-01 -4.08681124e-01 -3.90994519e-01
-1.02592373e+00 8.41947496e-01 3.84901732e-01 4.03470039e-01
7.79641926e-01 3.38326007e-01 1.04261625e+00 4.10755396e-01
2.35742226e-01 -8.24700058e-01 3.50916743e-01 5.20064533e-01
7.33335137e-01 -1.42380869e+00 4.65070829e-04 -5.40722847e-01
-5.31160653e-01 1.11486518e+00 8.02512586e-01 -1.07237503e-01
7.71031678e-01 -3.31417739e-01 1.06826380e-01 -7.08690464e-01
-7.62877166e-01 -1.64394855e-01 3.26242685e-01 3.92870188e-01
2.71961868e-01 -4.20124643e-03 7.57355914e-02 2.28652224e-01
-3.18089753e-01 3.03572148e-01 5.87428749e-01 9.24561501e-01
-7.35413209e-02 -6.07553959e-01 -4.10827607e-01 1.00645699e-01
-6.03944957e-01 2.98604488e-01 -6.30735219e-01 7.92526603e-01
2.19743416e-01 9.88666654e-01 5.61609209e-01 -5.35405397e-01
1.92986682e-01 -4.57389712e-01 2.55048037e-01 -2.74905860e-01
-8.33660722e-01 3.40531580e-02 2.98062712e-01 -7.97128558e-01
-1.01887214e+00 -9.31705713e-01 -7.57392108e-01 -7.04707146e-01
-1.93680003e-02 1.31442770e-02 -1.94352120e-01 1.06312883e+00
3.13537806e-01 5.52523673e-01 7.17295170e-01 -1.16707253e+00
2.41745591e-01 -8.01940382e-01 -2.77154356e-01 4.99873459e-01
6.84714973e-01 -6.58096969e-01 -3.14599425e-01 5.20534441e-02]
|
[8.222574234008789, 0.5661910176277161]
|
ea9c982a-6edd-4ee6-bb27-0735c2327754
|
perceptual-attacks-of-no-reference-image
|
2210.00933
| null |
https://arxiv.org/abs/2210.00933v1
|
https://arxiv.org/pdf/2210.00933v1.pdf
|
Perceptual Attacks of No-Reference Image Quality Models with Human-in-the-Loop
|
No-reference image quality assessment (NR-IQA) aims to quantify how humans perceive visual distortions of digital images without access to their undistorted references. NR-IQA models are extensively studied in computational vision, and are widely used for performance evaluation and perceptual optimization of man-made vision systems. Here we make one of the first attempts to examine the perceptual robustness of NR-IQA models. Under a Lagrangian formulation, we identify insightful connections of the proposed perceptual attack to previous beautiful ideas in computer vision and machine learning. We test one knowledge-driven and three data-driven NR-IQA methods under four full-reference IQA models (as approximations to human perception of just-noticeable differences). Through carefully designed psychophysical experiments, we find that all four NR-IQA models are vulnerable to the proposed perceptual attack. More interestingly, we observe that the generated counterexamples are not transferable, manifesting themselves as distinct design flows of respective NR-IQA methods.
|
['Kede Ma', 'Xiaokang Yang', 'Guodong Guo', 'Guangtao Zhai', 'Xiongkuo Min', 'Dingquan Li', 'Weixia Zhang']
|
2022-10-03
| null | null | null | null |
['no-reference-image-quality-assessment']
|
['computer-vision']
|
[ 4.47570711e-01 1.76158436e-02 -2.52187485e-03 1.24272518e-01
-5.59498608e-01 -7.16306567e-01 8.43723059e-01 1.32616246e-02
-3.24172109e-01 4.25604343e-01 -5.32150120e-02 -5.02903104e-01
-4.07451779e-01 -3.85834634e-01 -7.18262196e-01 -6.97299600e-01
-3.38451833e-01 -2.41141364e-01 9.18110386e-02 -2.26776645e-01
7.23417222e-01 5.77024221e-01 -1.48755336e+00 -1.40848219e-01
9.61945951e-01 1.02734447e+00 -4.48617414e-02 1.06302953e+00
6.39086545e-01 7.06887782e-01 -7.02476680e-01 -7.91384876e-01
6.97732389e-01 -1.14413895e-01 -7.49218524e-01 2.50492960e-01
7.15833247e-01 -1.44519314e-01 -4.49523598e-01 1.65555692e+00
4.20114070e-01 5.07497229e-02 7.08808601e-01 -1.33240080e+00
-1.31746721e+00 -7.88634866e-02 -5.82076788e-01 5.72375894e-01
3.66167635e-01 6.58116221e-01 8.78777742e-01 -8.71356070e-01
4.43831414e-01 1.61809266e+00 4.57915843e-01 3.50444525e-01
-1.42172003e+00 -1.83209136e-01 -4.99340221e-02 5.48588037e-01
-1.44394851e+00 -4.17317122e-01 6.51027501e-01 -5.05680144e-01
7.12028801e-01 4.54984367e-01 4.36083674e-01 7.77605176e-01
5.83842814e-01 3.69640470e-01 1.68290603e+00 -5.42221606e-01
3.62867057e-01 8.97849575e-02 1.65136933e-01 6.16863191e-01
4.05143142e-01 7.98093319e-01 -4.90175217e-01 -7.37003684e-02
8.08584630e-01 -5.69227338e-01 -3.83460104e-01 -3.85252833e-01
-1.24892271e+00 3.84705186e-01 5.27511537e-01 -8.78147632e-02
-1.43131495e-01 -1.01083573e-02 1.06098533e-01 4.80379105e-01
5.87385409e-02 8.15894306e-01 -1.16091974e-01 6.19483106e-02
-5.38054407e-01 1.82938218e-01 3.36472839e-01 9.31429148e-01
5.27625680e-01 3.10292572e-01 -3.46058726e-01 6.01048827e-01
1.88124612e-01 8.33460331e-01 3.71264398e-01 -1.27377546e+00
2.75090039e-01 2.42013782e-01 3.74925852e-01 -1.31248903e+00
-3.33468407e-01 -5.43227971e-01 -1.02043152e+00 8.75070632e-01
5.36754251e-01 2.80326307e-01 -7.46862054e-01 1.53994691e+00
-1.00833237e-01 2.37471890e-02 3.12557489e-01 1.01711798e+00
3.59023303e-01 4.90631789e-01 -2.21457761e-02 -6.43560231e-01
1.17093265e+00 -6.32871389e-01 -4.43839639e-01 8.03192183e-02
-1.89926270e-02 -8.59480739e-01 1.41265166e+00 9.09679413e-01
-1.34150910e+00 -9.97183204e-01 -1.33965611e+00 1.51638351e-02
-2.21500888e-01 1.29688550e-02 4.83997166e-02 1.06298840e+00
-1.26941013e+00 6.07494116e-01 -2.49432579e-01 -3.93466540e-02
3.23909014e-01 1.73299000e-01 -1.51953682e-01 4.18238603e-02
-9.37626421e-01 1.21310067e+00 2.39302471e-01 1.44863486e-01
-1.15969682e+00 -6.11563385e-01 -5.35568774e-01 -2.56855994e-01
5.46399653e-01 -5.87644279e-01 1.09142554e+00 -1.21981943e+00
-1.50595033e+00 1.08530593e+00 -1.70423508e-01 -5.99890411e-01
7.31781304e-01 -2.39866629e-01 -7.01569736e-01 1.92854449e-01
-2.90329039e-01 5.79639792e-01 1.28847933e+00 -1.51385033e+00
-1.98109269e-01 -2.77628630e-01 2.21835896e-01 3.97785008e-02
-1.33025413e-03 9.98001844e-02 -1.60518512e-01 -9.20561314e-01
-1.01047918e-01 -8.32096934e-01 -2.18633950e-01 3.41780126e-01
-7.66120434e-01 2.19766840e-01 5.36059618e-01 -3.48643094e-01
8.70290279e-01 -2.14047956e+00 6.76570982e-02 2.85375744e-01
2.09582239e-01 7.15123475e-01 -4.48453218e-01 4.11228947e-02
-1.62610441e-01 2.83618689e-01 -1.14192963e-01 7.39222392e-02
2.01062292e-01 -9.24693272e-02 -5.34678638e-01 6.98752224e-01
3.84992927e-01 9.38910961e-01 -8.26418161e-01 -2.24335656e-01
3.41954917e-01 3.73362213e-01 -2.71575332e-01 1.90613613e-01
-7.20917284e-02 4.84008342e-01 -1.61001474e-01 6.14112079e-01
8.40552390e-01 -1.54384160e-02 -2.09767461e-01 -6.47230864e-01
-1.36752605e-01 -1.44632176e-01 -1.03172970e+00 9.86892700e-01
-2.23081023e-01 8.39093983e-01 -2.91960984e-01 -7.03254044e-01
7.44296670e-01 -5.24788387e-02 -1.13285102e-01 -1.27035737e+00
3.86218131e-02 -8.44191536e-02 3.06013077e-01 -3.99492443e-01
6.05608463e-01 -1.44373626e-01 2.08909169e-01 1.19071327e-01
-9.12029296e-02 -1.83559358e-01 1.75493523e-01 5.99534139e-02
8.50285232e-01 -3.06168437e-01 5.91987252e-01 -4.82459843e-01
8.36605132e-01 -3.00175309e-01 4.07957256e-01 1.14657080e+00
-6.18280947e-01 7.68907785e-01 4.44808364e-01 -2.81984776e-01
-1.30873287e+00 -1.65072274e+00 -2.23143518e-01 7.41151571e-01
5.16228855e-01 -6.22777343e-02 -7.29478002e-01 -2.02322900e-01
-2.94298232e-01 6.61296844e-01 -6.19185925e-01 -3.52328062e-01
-1.93288893e-01 -6.73605919e-01 7.33990312e-01 2.96733737e-01
6.21596396e-01 -8.88371289e-01 -8.71383429e-01 -2.01083556e-01
2.21818060e-01 -1.14568579e+00 -3.76329720e-01 -4.26234126e-01
-5.69446743e-01 -1.23979986e+00 -8.49179149e-01 -2.05643937e-01
6.06155336e-01 3.75448674e-01 1.23738694e+00 1.72347948e-01
-4.61437970e-01 6.71632528e-01 -2.66122848e-01 -3.57736319e-01
-7.42727816e-01 -8.03120255e-01 3.73155117e-01 1.71151519e-01
-2.29227897e-02 -3.99938852e-01 -7.63596773e-01 6.73483074e-01
-8.20720315e-01 -9.30930525e-02 5.79090595e-01 5.62079310e-01
6.87643707e-01 1.26050964e-01 3.22544098e-01 -3.38887155e-01
8.24570775e-01 5.61215542e-02 -1.07033312e+00 4.53694731e-01
-9.33975160e-01 1.60082594e-01 8.31650138e-01 -5.23703396e-01
-9.35592353e-01 -3.07078451e-01 1.25223026e-01 -5.31869113e-01
-2.02347025e-01 -1.16612948e-02 -3.45464736e-01 -5.10532141e-01
7.76521266e-01 2.84266979e-01 -4.24578935e-01 -1.39071047e-01
7.00855434e-01 2.93823987e-01 1.18636620e+00 -6.10395372e-01
1.18386853e+00 5.08670866e-01 3.76932561e-01 -1.01490271e+00
-6.21051848e-01 -1.50045052e-01 -5.52335382e-01 -3.01629484e-01
8.87617350e-01 -5.91551721e-01 -9.96165276e-01 6.87067747e-01
-1.24233353e+00 -9.35358182e-02 -4.11300361e-01 2.70786762e-01
-9.66350615e-01 5.97844243e-01 -1.31767511e-01 -9.77101922e-01
-4.48557548e-02 -1.36184275e+00 7.07776070e-01 4.06017393e-01
1.90843195e-01 -8.58463585e-01 5.89639060e-02 3.08082104e-01
3.01383615e-01 1.09270245e-01 1.08728111e+00 -2.09011599e-01
-6.54057205e-01 2.54920930e-01 -5.98645329e-01 6.23373270e-01
-3.29864696e-02 1.65465266e-01 -1.26343346e+00 -4.60829705e-01
9.54289660e-02 -2.50368446e-01 6.58405244e-01 4.43986297e-01
1.13243997e+00 -5.80430806e-01 3.80033582e-01 6.56595409e-01
1.59994221e+00 4.39364344e-01 9.27545309e-01 3.23248833e-01
5.51070035e-01 5.43848395e-01 5.69341898e-01 2.62991071e-01
-5.00391982e-03 8.50090384e-01 7.19380915e-01 -2.25300521e-01
-1.71925619e-01 -1.92512851e-02 3.79167080e-01 4.72955465e-01
-3.63546044e-01 -2.35740200e-01 -6.92942798e-01 3.60054702e-01
-1.26708400e+00 -9.50604141e-01 -1.36835620e-01 2.40243459e+00
5.66415846e-01 3.24456155e-01 1.83187291e-01 3.96952242e-01
6.49851561e-01 1.93247065e-01 -9.35631871e-01 -6.54805779e-01
-5.15897930e-01 9.01797041e-02 5.77121615e-01 4.15356308e-01
-1.06845820e+00 5.32532990e-01 7.48107433e+00 6.39398396e-01
-9.28089499e-01 -1.05106366e-04 8.42609644e-01 9.97336432e-02
-6.39434531e-02 -8.29611048e-02 -1.61794960e-01 2.68448889e-01
8.65095496e-01 -5.45783162e-01 5.30736089e-01 5.03697217e-01
4.61691976e-01 -2.51042187e-01 -1.12855351e+00 1.19858551e+00
-9.42447260e-02 -1.30395842e+00 2.83951849e-01 1.15244903e-01
8.78744841e-01 -3.28285605e-01 8.06192338e-01 -3.37582767e-01
2.09562585e-01 -1.28251827e+00 1.06217647e+00 5.81063211e-01
8.55772436e-01 -6.92349136e-01 4.59446698e-01 -5.88475764e-02
-8.01054239e-01 -2.83135772e-01 -6.75903499e-01 -2.27112234e-01
-4.06746119e-02 4.39525187e-01 -1.95363879e-01 5.68371773e-01
6.23429894e-01 4.85559225e-01 -9.77636576e-01 1.12646902e+00
-4.44190651e-02 4.46485490e-01 4.01413858e-01 4.61689025e-01
3.03435624e-02 -1.63949519e-01 8.93092752e-01 8.50467503e-01
1.08777493e-01 9.46398750e-02 -2.81231403e-01 1.14342248e+00
-3.63320634e-02 -2.08557427e-01 -5.51254570e-01 1.90528944e-01
2.24016622e-01 7.70182908e-01 -6.10804319e-01 -1.41431600e-01
-3.08051646e-01 8.76505077e-01 -1.35353506e-01 5.66763818e-01
-6.66863084e-01 -2.09109068e-01 7.56523907e-01 -4.76152748e-02
9.28598270e-02 -2.70074457e-01 -3.35215151e-01 -1.05665541e+00
7.66647980e-03 -1.17292440e+00 -1.05482720e-01 -1.21734381e+00
-1.40963650e+00 5.69721162e-01 2.31354572e-02 -1.51779926e+00
-2.26634480e-02 -8.37154746e-01 -5.22244155e-01 8.99973691e-01
-1.57082438e+00 -7.37541676e-01 -2.19341546e-01 7.51843631e-01
3.84115338e-01 -2.48118311e-01 5.58542728e-01 6.48319861e-03
-4.84463274e-01 7.83332765e-01 1.99348465e-01 4.27095853e-02
4.53977406e-01 -1.25287318e+00 7.01552808e-01 1.34204578e+00
3.22680593e-01 5.93753755e-01 1.08758128e+00 -2.18597457e-01
-1.59163141e+00 -9.47709203e-01 8.35964754e-02 -3.81570548e-01
6.55744791e-01 1.07487790e-01 -9.58619952e-01 1.27478108e-01
3.13747674e-01 1.71228454e-01 1.33136332e-01 -4.58713114e-01
-7.35265672e-01 -1.47103697e-01 -1.25553930e+00 6.87032163e-01
1.01793385e+00 -7.33235896e-01 -7.12474883e-01 -7.60963857e-02
6.69461131e-01 1.74110755e-02 -7.31185138e-01 2.81266361e-01
4.22933251e-01 -1.57733083e+00 1.39697838e+00 -4.05048221e-01
3.66338491e-01 -5.61058223e-01 -3.30360144e-01 -1.32135081e+00
-2.27450505e-01 -8.00067663e-01 1.00421607e-01 9.31900322e-01
2.74794959e-02 -6.44242167e-01 1.77297860e-01 3.74693930e-01
1.02376543e-01 -3.14968020e-01 -1.04560542e+00 -1.35252833e+00
2.65382469e-01 -4.28147048e-01 2.77175605e-01 6.17476165e-01
-3.05931956e-01 -1.01951785e-01 -4.68979061e-01 5.14002740e-01
1.00615692e+00 -1.85382098e-01 6.24315619e-01 -9.88412619e-01
-5.11386514e-01 -6.63432300e-01 -8.48101318e-01 -9.31596816e-01
-2.85980761e-01 -3.77287328e-01 -2.95028538e-02 -8.24254513e-01
2.89746851e-01 -1.40837535e-01 -5.18038809e-01 -9.94590297e-02
-7.75354803e-02 6.23513281e-01 6.77756608e-01 2.93419957e-01
-5.66038489e-01 4.63766515e-01 1.41631615e+00 -4.83788073e-01
-6.71645487e-03 -1.61334220e-02 -7.28423476e-01 8.73565853e-01
6.95265770e-01 -1.61735117e-01 -6.88651443e-01 -2.75218576e-01
4.99325931e-01 -2.39352006e-02 1.06153655e+00 -1.22873747e+00
3.06547154e-03 -4.20184463e-01 2.63113052e-01 -3.19578081e-01
1.37459919e-01 -5.26972055e-01 -5.80012426e-02 4.33101803e-01
-5.02361417e-01 3.12384218e-01 2.35746622e-01 7.06109464e-01
-2.01511011e-01 -1.37346283e-01 1.19284701e+00 -7.28453547e-02
-9.23006773e-01 1.05637545e-02 -5.09048283e-01 2.27376983e-01
9.11580026e-01 -4.75042164e-01 -8.07579637e-01 -3.38133901e-01
-6.13239825e-01 -3.93967211e-01 5.88571191e-01 2.69964248e-01
9.22619224e-01 -1.03803074e+00 -6.14003718e-01 2.00716734e-01
1.82822987e-01 -6.33380234e-01 2.79109567e-01 7.80805528e-01
-4.42289025e-01 4.38425630e-01 -6.65482402e-01 -6.31650150e-01
-1.21443975e+00 1.15711725e+00 5.64953625e-01 4.68686149e-02
-3.78382295e-01 5.63881457e-01 5.36837757e-01 2.00187311e-01
2.13728860e-01 -4.58402783e-01 2.46030204e-02 -5.10367930e-01
7.72740364e-01 7.93803394e-01 -1.39085921e-02 -8.80979359e-01
-2.79508978e-01 9.54931796e-01 1.25182897e-01 -2.68335789e-01
6.69946492e-01 -4.31176215e-01 2.49456063e-01 2.58904219e-01
1.04921556e+00 -5.25087453e-02 -1.51025271e+00 -1.20276012e-01
-1.76999569e-02 -7.34320641e-01 -4.94380705e-02 -9.31547880e-01
-9.17980254e-01 1.11223423e+00 1.08633292e+00 4.03569162e-01
1.68161261e+00 -1.92682266e-01 1.55903354e-01 3.93230051e-01
4.77821112e-01 -7.29668736e-01 4.22331095e-01 1.24929391e-01
1.20716798e+00 -1.13256145e+00 9.29781795e-02 -2.63276368e-01
-5.79889774e-01 7.44089246e-01 4.93914634e-01 -2.60549814e-01
4.47262794e-01 -1.73499987e-01 1.97216406e-01 1.90171480e-01
-6.65679038e-01 -2.87792802e-01 6.39981151e-01 1.17065465e+00
-2.30447769e-01 -1.72607645e-01 -6.26624003e-02 8.05075616e-02
-1.34191304e-01 -1.74725816e-01 9.82436001e-01 3.88341576e-01
-2.89294720e-01 -7.92583227e-01 -7.67923534e-01 9.42915156e-02
-3.75605255e-01 -9.08196047e-02 -3.02527189e-01 6.61777735e-01
8.40916038e-02 1.20077229e+00 -1.10227369e-01 -4.64610726e-01
4.70255077e-01 -5.26608169e-01 6.88644707e-01 8.16633552e-02
-2.65526146e-01 -1.60952434e-01 -4.06197488e-01 -7.81636775e-01
-6.78422749e-01 -3.15166712e-01 -7.49427855e-01 -3.82264048e-01
-7.10345209e-02 -3.99684072e-01 3.76808494e-01 8.48741770e-01
2.32706785e-01 3.96278918e-01 7.35835791e-01 -8.26952994e-01
-6.54483020e-01 -4.54140842e-01 -5.56604922e-01 4.89973485e-01
8.07455301e-01 -6.28484249e-01 -3.59472573e-01 3.42943311e-01]
|
[9.943950653076172, 2.0022573471069336]
|
0c24dbd8-e103-456a-971d-dea51192eb48
|
making-binary-classification-from-multiple
|
2306.07036
| null |
https://arxiv.org/abs/2306.07036v1
|
https://arxiv.org/pdf/2306.07036v1.pdf
|
Making Binary Classification from Multiple Unlabeled Datasets Almost Free of Supervision
|
Training a classifier exploiting a huge amount of supervised data is expensive or even prohibited in a situation, where the labeling cost is high. The remarkable progress in working with weaker forms of supervision is binary classification from multiple unlabeled datasets which requires the knowledge of exact class priors for all unlabeled datasets. However, the availability of class priors is restrictive in many real-world scenarios. To address this issue, we propose to solve a new problem setting, i.e., binary classification from multiple unlabeled datasets with only one pairwise numerical relationship of class priors (MU-OPPO), which knows the relative order (which unlabeled dataset has a higher proportion of positive examples) of two class-prior probabilities for two datasets among multiple unlabeled datasets. In MU-OPPO, we do not need the class priors for all unlabeled datasets, but we only require that there exists a pair of unlabeled datasets for which we know which unlabeled dataset has a larger class prior. Clearly, this form of supervision is easier to be obtained, which can make labeling costs almost free. We propose a novel framework to handle the MU-OPPO problem, which consists of four sequential modules: (i) pseudo label assignment; (ii) confident example collection; (iii) class prior estimation; (iv) classifier training with estimated class priors. Theoretically, we analyze the gap between estimated class priors and true class priors under the proposed framework. Empirically, we confirm the superiority of our framework with comprehensive experiments. Experimental results demonstrate that our framework brings smaller estimation errors of class priors and better performance of binary classification.
|
['Tongliang Liu', 'Masashi Sugiyama', 'Gang Niu', 'Bo Han', 'Jun Yu', 'Xiaobo Xia', 'Yuhao Wu']
|
2023-06-12
| null | null | null | null |
['pseudo-label']
|
['miscellaneous']
|
[ 4.49681878e-01 2.51231521e-01 -7.24152267e-01 -7.13811815e-01
-8.15959215e-01 -5.83392143e-01 3.28472346e-01 2.07933977e-01
-3.69919956e-01 1.06125689e+00 -4.32973057e-01 -4.16926950e-01
-1.98783100e-01 -8.09610665e-01 -7.19439447e-01 -9.77971017e-01
3.51407915e-01 7.26372302e-01 1.96580097e-01 1.80760473e-01
1.04340188e-01 2.21936211e-01 -1.50526893e+00 -3.94899473e-02
9.15604949e-01 1.27480817e+00 6.46460950e-02 1.34319276e-01
-2.29769386e-02 5.69931090e-01 -5.55061698e-01 -2.80991971e-01
5.47705114e-01 -2.82370061e-01 -8.02523851e-01 3.89484644e-01
2.06884220e-01 -3.01336765e-01 1.86683357e-01 1.45789242e+00
7.58144259e-02 2.52924357e-02 9.56023514e-01 -1.56252611e+00
-4.09500867e-01 6.08229101e-01 -1.09470224e+00 4.07075733e-02
-4.07265350e-02 -5.21852188e-02 9.22801137e-01 -6.05204821e-01
2.36732602e-01 1.02715850e+00 4.98234570e-01 3.06357920e-01
-9.88424778e-01 -8.83870661e-01 2.37461418e-01 9.52899531e-02
-1.72257161e+00 -4.80482355e-02 5.99372268e-01 -2.87731916e-01
5.07992283e-02 1.79910734e-01 1.71958789e-01 7.88904250e-01
-3.49596858e-01 6.94850028e-01 1.39666450e+00 -5.35970211e-01
3.86213303e-01 5.51061928e-01 7.53467202e-01 2.97931790e-01
5.51685452e-01 2.33559795e-02 -1.94685981e-01 -2.30916813e-01
5.61311185e-01 3.36332142e-01 -4.05556053e-01 -2.21511915e-01
-1.11932826e+00 8.12656105e-01 2.13117167e-01 1.58343181e-01
-8.30413178e-02 -4.76312757e-01 -9.47962515e-03 2.03033641e-01
1.60370842e-01 -8.41272995e-02 -6.82980478e-01 3.56647789e-01
-7.32774556e-01 -1.36791945e-01 9.17257011e-01 1.30387878e+00
1.14150643e+00 -4.94618177e-01 1.24281056e-01 8.19205999e-01
3.57993662e-01 5.45904458e-01 4.95074838e-01 -6.24065578e-01
7.78759301e-01 4.96430308e-01 3.86968970e-01 -8.51376951e-01
-3.27408537e-02 -2.67400444e-01 -8.76769841e-01 -1.47294402e-01
8.56228292e-01 -2.82723606e-01 -9.08033967e-01 1.68803954e+00
6.21104598e-01 3.26472849e-01 2.44434565e-01 9.40993309e-01
4.32714939e-01 7.07040787e-01 -1.11874513e-01 -6.49908483e-01
1.41096747e+00 -8.61209750e-01 -7.01820850e-01 -2.04273105e-01
6.34306550e-01 -6.40127957e-01 7.83757269e-01 5.04396796e-01
-3.71017814e-01 -4.49245840e-01 -1.30197906e+00 1.70173511e-01
-3.44009489e-01 5.03546059e-01 6.45694315e-01 9.01375175e-01
-1.97120458e-01 3.20548743e-01 -5.88343859e-01 8.67545903e-02
3.93017739e-01 5.94342768e-01 -4.80873376e-01 -3.86856377e-01
-1.15851378e+00 5.45892596e-01 9.31507826e-01 2.61631966e-01
-6.12272918e-01 -4.22804087e-01 -8.12954307e-01 1.66754320e-01
8.30136418e-01 -9.32389647e-02 1.12537706e+00 -1.24067950e+00
-1.08597898e+00 7.21582949e-01 -1.67465210e-01 -1.59278989e-01
3.91193241e-01 4.84281741e-02 -2.91297704e-01 1.03544071e-02
2.47515008e-01 5.75442016e-01 7.23146260e-01 -1.43898737e+00
-9.73602116e-01 -4.61610645e-01 1.60558105e-01 2.28938386e-01
-4.06523198e-01 -2.33291432e-01 -3.43998194e-01 -4.31647599e-01
3.97678077e-01 -7.98587024e-01 -1.79423138e-01 -5.86893260e-02
-5.58925271e-01 -4.38136399e-01 1.00314748e+00 -3.67322773e-01
9.15823936e-01 -2.22395468e+00 -3.93294752e-01 4.78139758e-01
9.11552012e-02 2.09603384e-01 2.87736923e-01 -9.11230147e-02
-1.40351519e-01 1.42728105e-01 -5.74966908e-01 -7.20526576e-02
-6.62443042e-02 4.98197109e-01 -4.35517132e-01 4.09399539e-01
2.21600294e-01 3.08126301e-01 -9.62709486e-01 -8.48748028e-01
-7.22545981e-02 1.56878885e-02 -2.19189137e-01 1.95702866e-01
-2.60997377e-02 4.07910615e-01 -5.99988282e-01 7.83765435e-01
1.17077982e+00 -5.77202320e-01 4.81765896e-01 -1.39769033e-01
2.09581226e-01 -1.18158579e-01 -1.91050839e+00 9.87329841e-01
-1.23463444e-01 -2.20339905e-04 -1.07635148e-01 -1.33086538e+00
9.63753223e-01 3.25734496e-01 1.87847927e-01 -8.33715051e-02
3.65637958e-01 2.59755373e-01 -7.37896860e-02 -3.13874066e-01
-3.28068994e-02 -5.05238295e-01 -5.58974519e-02 4.26290482e-01
1.07017860e-01 2.66814344e-02 3.51091236e-01 1.75169725e-02
5.79463780e-01 -1.65281236e-01 5.79299986e-01 -2.15948239e-01
6.06105804e-01 -2.26287842e-01 1.21419168e+00 7.82677472e-01
-2.72437483e-01 5.28380275e-01 5.16709745e-01 -1.82079554e-01
-6.30984485e-01 -8.19737434e-01 -5.25909007e-01 6.22112751e-01
5.67879260e-01 1.62607722e-03 -4.86459076e-01 -9.95249033e-01
-7.45538548e-02 4.05917913e-01 -5.75472772e-01 1.59087151e-01
-2.35119000e-01 -1.10310245e+00 2.18598947e-01 4.97926265e-01
6.37526035e-01 -6.07611895e-01 -1.00250751e-01 -1.42389134e-01
-2.46401563e-01 -1.13597584e+00 -4.57167506e-01 5.72969675e-01
-8.70661259e-01 -1.34911442e+00 -5.83419681e-01 -9.63898480e-01
1.09942102e+00 2.57250488e-01 6.32283688e-01 1.29479319e-01
1.60327509e-01 -1.39484495e-01 -3.82631987e-01 -4.36032534e-01
-1.38888955e-01 -1.40003517e-01 1.74978375e-01 3.61502141e-01
5.42819560e-01 -4.24149930e-01 -2.19254106e-01 7.98339188e-01
-1.14885271e+00 -7.83740655e-02 7.94994116e-01 1.11901021e+00
8.96839976e-01 5.34056842e-01 6.26220286e-01 -1.28907084e+00
1.31417498e-01 -7.93727279e-01 -8.49914670e-01 6.51242077e-01
-5.48483729e-01 -1.84328388e-02 7.73634851e-01 -6.94744289e-01
-1.13721800e+00 2.98790306e-01 2.23182559e-01 -4.42537636e-01
-3.48851085e-01 5.36053479e-01 -5.98987937e-01 5.88282868e-02
4.69594151e-01 -3.30827236e-02 -1.47090003e-01 -3.90786767e-01
1.10396609e-01 1.10058141e+00 5.37972629e-01 -8.74418676e-01
8.82765710e-01 4.83098388e-01 1.02844454e-01 -5.22819161e-01
-1.28530622e+00 -6.03317916e-01 -7.83993423e-01 1.66643694e-01
5.79947352e-01 -8.55296373e-01 -6.39413655e-01 3.69574100e-01
-7.75548995e-01 -3.03744413e-02 -8.89147893e-02 8.12255681e-01
-1.14360489e-01 7.46652126e-01 -3.02421659e-01 -8.36606860e-01
2.24032868e-02 -1.31836987e+00 8.44499767e-01 6.13842189e-01
3.62039477e-01 -8.84071708e-01 -2.86598265e-01 5.06303966e-01
-3.58328074e-01 1.34678796e-01 7.62705326e-01 -9.55262303e-01
-5.82180321e-01 -4.14968997e-01 -4.40229416e-01 5.26031494e-01
3.58388186e-01 -1.74632967e-01 -1.04034662e+00 -2.34277919e-01
2.08419576e-01 -5.53382635e-01 5.63962042e-01 1.85740292e-01
1.43317604e+00 -2.28648081e-01 -5.32243311e-01 3.66936743e-01
1.29459643e+00 3.44839901e-01 3.59156340e-01 -7.87480101e-02
5.24491727e-01 6.21238410e-01 1.30057907e+00 4.54272777e-01
2.49061778e-01 4.73093212e-01 3.75643224e-01 5.95804937e-02
3.78422588e-01 -1.98092639e-01 -9.59403887e-02 7.63028562e-01
1.35294944e-01 -3.36812645e-01 -8.53195071e-01 3.98990065e-01
-1.75592983e+00 -6.68134391e-01 -5.93589805e-02 2.55537343e+00
1.11341226e+00 1.50029302e-01 -1.42221838e-01 5.40093839e-01
1.29252517e+00 -3.40133458e-01 -6.44416571e-01 2.67995149e-01
-4.27665748e-03 4.14790921e-02 5.47868550e-01 4.71912473e-01
-1.30242658e+00 4.58201647e-01 5.79590559e+00 1.08598936e+00
-8.93222988e-01 4.12536711e-02 1.04098296e+00 4.21555698e-01
3.36953849e-02 3.63378495e-01 -1.29471982e+00 7.36956716e-01
4.98685509e-01 -4.25333455e-02 5.96393123e-02 1.08336329e+00
-2.73816139e-01 -4.32034999e-01 -1.30740011e+00 9.74656641e-01
-5.91676235e-02 -9.36064184e-01 -8.67341608e-02 1.03300273e-01
7.44011462e-01 -4.50094104e-01 -2.21209645e-01 3.82618606e-01
4.59871918e-01 -6.88672543e-01 4.83160526e-01 1.35635257e-01
8.78225803e-01 -6.88192725e-01 1.11441147e+00 9.61592495e-01
-1.10986412e+00 -1.68390527e-01 -5.70937693e-01 7.20279152e-03
-9.04389843e-02 1.02666092e+00 -6.77712321e-01 8.95974815e-01
6.02392137e-01 7.43841290e-01 -2.84643471e-01 1.01765847e+00
-3.02322984e-01 6.49294794e-01 -6.44928813e-01 1.56751961e-01
-7.46774003e-02 -4.01845664e-01 -9.02618915e-02 7.48106420e-01
1.17473081e-01 5.51733315e-01 6.22667849e-01 4.80784506e-01
-2.44472221e-01 9.65220258e-02 -3.46285582e-01 1.40944809e-01
8.27057779e-01 1.21337426e+00 -1.03539479e+00 -4.92548496e-01
-4.86667246e-01 4.49540019e-01 3.43159884e-01 2.41965279e-01
-8.72603834e-01 -3.50675315e-01 -1.36883408e-02 -3.23944777e-01
3.04035306e-01 2.92857081e-01 -2.62051433e-01 -1.31712556e+00
1.59661755e-01 -8.04506004e-01 8.35499227e-01 -5.12633979e-01
-1.65961802e+00 1.65937826e-01 2.54095405e-01 -1.41441238e+00
2.28677299e-02 -6.85516775e-01 -5.36830544e-01 8.23141575e-01
-1.70402515e+00 -8.94489229e-01 -2.68304706e-01 6.60432994e-01
1.60198689e-01 1.22865915e-01 5.91805816e-01 5.43615460e-01
-7.75366843e-01 8.99221778e-01 1.11004911e-01 2.77431279e-01
8.11566830e-01 -1.08974767e+00 -6.45631611e-01 7.67303348e-01
-5.45887575e-02 5.91966808e-01 3.16130579e-01 -5.38401425e-01
-9.24421489e-01 -9.85549808e-01 7.17061341e-01 -1.95865661e-01
5.17244995e-01 -3.83034977e-03 -8.19198728e-01 9.47898984e-01
-4.36869800e-01 3.94645989e-01 9.44486082e-01 1.10662781e-01
-3.52160007e-01 -3.51127744e-01 -1.43276119e+00 1.56317830e-01
4.93838459e-01 -2.06195667e-01 -7.23980665e-01 6.95605516e-01
6.40489936e-01 -4.07788843e-01 -8.98994207e-01 6.06945157e-01
3.66678774e-01 -5.80302000e-01 7.15315342e-01 -3.51605684e-01
2.75249541e-01 -8.06638062e-01 -2.37780929e-01 -8.77914667e-01
2.52230406e-01 -1.48795232e-01 1.19231276e-01 1.48079228e+00
4.55093622e-01 -9.85230863e-01 7.47991979e-01 7.62005270e-01
1.56131551e-01 -7.79637158e-01 -9.22018826e-01 -8.31481457e-01
-8.22372437e-02 -1.41230911e-01 6.25543833e-01 1.45214641e+00
-1.13250442e-01 2.64146596e-01 -3.55216026e-01 6.81009412e-01
5.99425316e-01 4.08214062e-01 6.10285461e-01 -1.40695131e+00
-5.37068844e-01 4.76120301e-02 -3.68408650e-01 -1.20075083e+00
1.56908229e-01 -6.57676220e-01 1.86437562e-01 -9.70896482e-01
5.61258554e-01 -9.77728307e-01 -3.57530832e-01 8.13128889e-01
-5.25905132e-01 1.32733926e-01 -5.07906415e-02 5.15205801e-01
-5.51816404e-01 3.28586698e-01 1.10596311e+00 -1.32821232e-01
5.54269589e-02 2.73411453e-01 -7.71319628e-01 7.86659062e-01
6.70273542e-01 -7.58100688e-01 -5.25321722e-01 -1.79102674e-01
-1.40187487e-01 2.69245028e-01 -8.47834907e-03 -8.53762686e-01
2.37378448e-01 -3.64443839e-01 4.06749189e-01 -6.75704598e-01
1.71990544e-01 -1.16155875e+00 6.93966523e-02 2.15340927e-01
-1.07461058e-01 -5.03611922e-01 -1.54362097e-01 7.81012654e-01
-3.61430168e-01 -7.17185199e-01 9.37901437e-01 6.72173277e-02
-4.59353417e-01 3.62185925e-01 9.38632935e-02 5.02586775e-02
1.23326993e+00 -2.06263438e-01 -3.69623810e-01 -4.28828970e-02
-7.91124701e-01 5.79774618e-01 2.56336570e-01 -3.54390629e-02
2.09363133e-01 -1.29421651e+00 -3.88766140e-01 6.84130117e-02
1.50515869e-01 4.39291775e-01 -1.56074595e-02 6.93396568e-01
-9.64325517e-02 3.08019340e-01 -5.19055389e-02 -8.04031670e-01
-1.18553007e+00 8.05857897e-01 1.94272861e-01 -4.33117360e-01
-1.11669779e-01 7.76798129e-01 4.17516649e-01 -6.67258561e-01
2.16987610e-01 -1.73458442e-01 -3.35522413e-01 9.58096832e-02
4.53449726e-01 1.57982379e-01 1.59886666e-02 -5.20087481e-01
-1.60586685e-01 4.96800661e-01 -3.72007906e-01 1.58687353e-01
9.21230912e-01 -1.76596679e-02 -1.09880395e-01 4.64285403e-01
1.01041412e+00 -1.08934738e-01 -1.25053191e+00 -4.92264122e-01
-5.05592078e-02 -6.66274309e-01 -1.87377170e-01 -6.15519404e-01
-1.10966575e+00 9.29204881e-01 4.59337801e-01 5.43198735e-02
1.27210462e+00 -1.50646448e-01 4.75534499e-01 5.06111383e-01
5.62967777e-01 -1.04538524e+00 -2.08024725e-01 2.64548540e-01
2.30937585e-01 -1.68222606e+00 1.75551489e-01 -9.25224900e-01
-6.36602581e-01 1.03075945e+00 8.12471211e-01 2.24888369e-01
9.42618370e-01 8.85608271e-02 2.43356060e-02 -8.22527148e-03
-4.50359017e-01 -1.06203511e-01 1.36069715e-01 4.48184848e-01
1.14525221e-01 1.52833179e-01 -4.53806281e-01 8.85555208e-01
-2.21218169e-03 3.33183119e-03 4.41957951e-01 1.10905504e+00
-3.05117518e-01 -1.24808252e+00 -6.65984333e-01 5.98395228e-01
-3.83173674e-01 4.46612127e-02 -8.60873237e-02 7.60011911e-01
4.36838269e-01 1.06977224e+00 -1.75369799e-01 -8.65705311e-02
-5.66512048e-02 4.89391237e-02 1.73812360e-01 -8.38259935e-01
-1.16226440e-02 -9.25405920e-02 -2.09930390e-01 2.68131141e-02
-6.40830755e-01 -4.48503584e-01 -1.34855700e+00 -7.01161847e-02
-1.05769396e+00 5.56775391e-01 3.62131923e-01 1.22462010e+00
-1.46536335e-01 1.81422252e-02 8.49141836e-01 -6.34066343e-01
-8.26975763e-01 -8.91415417e-01 -1.00016165e+00 2.72733510e-01
6.25251085e-02 -1.00594687e+00 -6.97462201e-01 3.04452419e-01]
|
[9.31001091003418, 3.98052978515625]
|
5a3ef850-3396-45d0-bc03-787463075320
|
openapepose-a-database-of-annotated-ape
|
2212.00741
| null |
https://arxiv.org/abs/2212.00741v1
|
https://arxiv.org/pdf/2212.00741v1.pdf
|
OpenApePose: a database of annotated ape photographs for pose estimation
|
Because of their close relationship with humans, non-human apes (chimpanzees, bonobos, gorillas, orangutans, and gibbons, including siamangs) are of great scientific interest. The goal of understanding their complex behavior would be greatly advanced by the ability to perform video-based pose tracking. Tracking, however, requires high-quality annotated datasets of ape photographs. Here we present OpenApePose, a new public dataset of 71,868 photographs, annotated with 16 body landmarks, of six ape species in naturalistic contexts. We show that a standard deep net (HRNet-W48) trained on ape photos can reliably track out-of-sample ape photos better than networks trained on monkeys (specifically, the OpenMonkeyPose dataset) and on humans (COCO) can. This trained network can track apes almost as well as the other networks can track their respective taxa, and models trained without one of the six ape species can track the held out species better than the monkey and human models can. Ultimately, the results of our analyses highlight the importance of large specialized databases for animal tracking systems and confirm the utility of our new ape database.
|
['Benjamin Hayden', 'Jan Zimmermann', 'Jessica Raper', 'Rebecca Richardson', 'Praneet Bala', 'Nisarg Desai']
|
2022-11-30
| null | null | null | null |
['pose-tracking']
|
['computer-vision']
|
[-3.72303903e-01 -4.32348810e-02 -1.47329971e-01 -1.45301580e-01
-3.99533749e-01 -6.19347394e-01 4.49185878e-01 -3.15272778e-01
-1.33371782e+00 8.66630018e-01 -4.22078110e-02 3.67085040e-01
1.82163641e-01 -5.35457134e-01 -8.25000584e-01 -1.65301576e-01
-9.47772145e-01 8.62603962e-01 6.39091730e-01 -1.83669433e-01
-2.77816594e-01 6.18380487e-01 -1.17895818e+00 -1.04682185e-01
3.30140859e-01 5.49730062e-01 9.83425602e-02 5.36256611e-01
6.85951889e-01 3.20704162e-01 -3.50554526e-01 -8.79783690e-01
4.97284979e-01 -2.25661527e-02 -4.08475310e-01 -7.97254980e-01
1.35045147e+00 -7.99868405e-01 -8.67521763e-01 8.21121871e-01
2.95097023e-01 2.59313315e-01 4.31723118e-01 -1.35409462e+00
-4.44008827e-01 7.34370708e-01 -5.37924528e-01 6.55796468e-01
-9.80136264e-03 6.71445668e-01 1.07856011e+00 -3.44695270e-01
8.43908250e-01 1.39117920e+00 1.37211394e+00 8.57733190e-01
-1.14441967e+00 -1.14961529e+00 -1.92028359e-01 3.35230142e-01
-1.34494126e+00 -5.32679081e-01 8.45632032e-02 -6.53633356e-01
1.02561855e+00 -3.46047819e-01 1.22408020e+00 1.38712144e+00
4.17611711e-02 5.92209399e-01 7.97444165e-01 4.58381206e-01
-7.02272356e-02 -7.01187551e-01 -5.45867383e-02 9.70230401e-01
4.35551763e-01 3.77824992e-01 -5.54323018e-01 -3.73118758e-01
7.97472596e-01 2.25671068e-01 -3.19891512e-01 -3.42587262e-01
-1.68398523e+00 6.03310466e-01 1.13935137e+00 -2.39268411e-02
-3.76562208e-01 6.58997416e-01 3.45279604e-01 -6.47946671e-02
1.03800157e-02 3.95154774e-01 -2.33670890e-01 -5.04263341e-01
-1.10089433e+00 4.41177934e-01 7.33806312e-01 8.82108390e-01
2.69607246e-01 7.69884810e-02 1.20488666e-01 1.04163706e+00
4.91512477e-01 9.72128987e-01 1.63186073e-01 -8.26855123e-01
6.20640993e-01 2.28688374e-01 3.56349051e-01 -6.48633718e-01
-8.08832228e-01 -3.66735179e-03 -5.82405806e-01 3.95585716e-01
7.92738080e-01 -1.44091412e-01 -9.04535294e-01 2.02828598e+00
3.45085353e-01 -1.26002476e-01 -4.29955870e-01 9.13361251e-01
6.37501359e-01 4.51368451e-01 6.39408708e-01 4.53750581e-01
1.57928169e+00 -1.04105413e+00 -6.48766756e-02 -7.36186504e-01
-5.34354448e-02 -2.06544414e-01 5.67596197e-01 -8.46799463e-02
-7.80245900e-01 -4.56453919e-01 -1.15188539e+00 7.43714496e-02
-4.61289078e-01 3.90459523e-02 7.38864899e-01 4.73740906e-01
-1.01140630e+00 8.97157729e-01 -1.37519586e+00 -9.27206039e-01
1.10353196e+00 6.51626527e-01 -9.87847447e-01 1.20954186e-01
-9.25700247e-01 1.35274231e+00 3.21796596e-01 2.42432639e-01
-1.15875864e+00 -5.31471193e-01 -7.62054384e-01 -4.69479933e-02
-3.66344029e-04 -4.54984903e-01 1.23611856e+00 -6.11837029e-01
-9.20517385e-01 1.20641255e+00 6.94797993e-01 -8.05384874e-01
7.92284608e-01 -3.37019771e-01 -4.74189103e-01 4.66594636e-01
2.15747640e-01 1.49531102e+00 4.47501868e-01 -5.55331707e-01
-8.05823445e-01 -7.06884265e-01 -7.09187388e-02 -4.35525142e-02
1.18657522e-01 2.05485031e-01 -3.25361013e-01 -5.91536582e-01
-3.47752213e-01 -1.28757656e+00 -1.07765578e-01 1.27860999e+00
9.45008621e-02 -3.80724669e-02 6.01247013e-01 -9.65600908e-01
6.10156357e-01 -2.17488861e+00 5.84646836e-02 -2.50954270e-01
5.19767404e-01 5.05024612e-01 -2.17046291e-01 3.77897084e-01
2.44554028e-01 -3.37460101e-01 -4.74604517e-01 -7.46963844e-02
1.28629968e-01 2.55605847e-01 1.96173429e-01 1.13298535e+00
-1.55941233e-01 8.98919880e-01 -1.11156046e+00 -5.85565507e-01
1.07365698e-01 4.23690796e-01 -7.01914847e-01 2.46867612e-02
-2.13599131e-02 1.85272947e-01 -1.29803732e-01 7.21343517e-01
3.77092451e-01 1.00019418e-01 1.01639982e-02 6.69256598e-02
-1.81339025e-01 3.27557549e-02 -4.43013400e-01 1.43923652e+00
-1.81432143e-01 9.43598628e-01 4.66446906e-01 -1.09121814e-01
4.23583359e-01 -1.71377305e-02 4.51978862e-01 -5.24695098e-01
1.67943329e-01 2.85077691e-01 7.61492431e-01 -2.93077864e-02
3.06319147e-01 -1.71835616e-01 1.10788628e-01 3.59586358e-01
4.07444775e-01 1.85908034e-01 3.91361028e-01 -7.42654726e-02
1.09442925e+00 2.44779766e-01 4.13067222e-01 -3.03487033e-01
8.66295844e-02 1.92594439e-01 4.69141036e-01 6.67143941e-01
-8.69169831e-01 4.26343411e-01 1.10132247e-01 -9.50192273e-01
-1.37331462e+00 -1.30160165e+00 -4.85131621e-01 1.28146183e+00
-3.00340001e-02 -2.40211740e-01 -5.96359432e-01 -5.63780189e-01
5.26113868e-01 2.87416250e-01 -9.38357711e-01 -1.97036892e-01
-9.76084113e-01 -6.63231432e-01 1.10257530e+00 8.18834901e-01
7.95312345e-01 -1.43424141e+00 -8.06037188e-01 3.31214517e-01
2.77405493e-02 -1.01714993e+00 -7.13170826e-01 -2.56206065e-01
-3.18018645e-01 -1.32919693e+00 -1.01968062e+00 -5.86631298e-01
2.91903764e-01 1.65643618e-01 7.05154657e-01 1.17264591e-01
-2.87414104e-01 5.08996546e-01 -1.45490110e-01 -2.77098492e-02
-3.26767206e-01 -7.45916590e-02 6.52303457e-01 -4.57805514e-01
5.21910012e-01 -7.97801375e-01 -6.90171719e-01 7.79746771e-01
-3.36453378e-01 -2.57356972e-01 4.02376205e-01 3.72281253e-01
2.85254180e-01 -1.15235186e+00 3.06518555e-01 -2.79464960e-01
-2.07783692e-02 -6.37001395e-01 -1.02702379e+00 2.56808132e-01
3.10125887e-01 -2.49071941e-01 5.43680608e-01 -5.51513016e-01
-4.82907653e-01 2.07489043e-01 -3.90404880e-01 -2.23079517e-01
1.24290427e-02 -2.52399385e-01 2.14131266e-01 -3.65988612e-01
8.30422938e-01 -1.65148601e-01 4.56513185e-03 -6.46327734e-01
2.17104599e-01 5.71127415e-01 1.14332962e+00 -5.81457734e-01
8.06123495e-01 4.80298787e-01 9.74624678e-02 -1.14814413e+00
-3.33042830e-01 -1.89651653e-01 -9.56479013e-01 -4.95667785e-01
1.38702929e+00 -1.01632023e+00 -1.08370185e+00 6.06330276e-01
-1.00638580e+00 -5.60778499e-01 6.68863505e-02 7.53589988e-01
-5.44715703e-01 3.21561158e-01 -6.81630552e-01 -4.83418226e-01
-5.02347291e-01 -9.31511402e-01 1.18086421e+00 2.27904886e-01
-4.68098521e-01 -3.99624318e-01 4.62252021e-01 3.03801566e-01
2.71570235e-01 3.67172867e-01 2.95565277e-01 -9.54027534e-01
-6.31454051e-01 -4.26764756e-01 -2.17122644e-01 1.10276062e-02
-1.86603412e-01 1.22365355e-01 -4.88752812e-01 -3.96061063e-01
-6.82683408e-01 -4.60435361e-01 8.65273774e-01 1.56550750e-01
5.71991086e-01 -2.56520271e-01 -5.90540349e-01 8.63881826e-01
8.16812336e-01 1.74214587e-01 4.09318432e-02 2.45602593e-01
6.85471892e-01 2.89748311e-01 1.24746270e-01 1.73405468e-01
5.50485969e-01 9.01797831e-01 4.22566950e-01 4.04609352e-01
-3.56156468e-01 -7.13128030e-01 3.85152072e-01 3.15871209e-01
-6.79648817e-01 -1.84287876e-01 -1.16835809e+00 4.91898835e-01
-1.73697209e+00 -1.25104010e+00 5.33535779e-02 2.31720543e+00
4.14013654e-01 -1.97316468e-01 7.84856737e-01 -7.92179227e-01
7.99379766e-01 2.11447045e-01 -6.45655036e-01 1.02777399e-01
8.24239776e-02 5.31082265e-02 9.53672945e-01 3.99475880e-02
-1.38102233e+00 9.63803530e-01 6.76503897e+00 3.63726854e-01
-7.28305757e-01 1.19759321e-01 -1.75445434e-02 -3.07199746e-01
6.00830674e-01 -4.86545056e-01 -8.88191342e-01 7.14556754e-01
9.64209437e-01 2.47540437e-02 7.49792635e-01 1.22641492e+00
-4.08980697e-01 8.26067999e-02 -1.54626095e+00 1.14385271e+00
7.28315786e-02 -9.21354473e-01 -5.47632754e-01 1.43096671e-01
4.98164415e-01 8.98759246e-01 -2.47929558e-01 3.58824611e-01
7.03156471e-01 -8.97922337e-01 8.87677789e-01 4.81723756e-01
6.88317358e-01 -4.75739390e-01 5.43292046e-01 3.49442035e-01
-1.10765803e+00 -9.02025327e-02 -6.24386728e-01 7.70211592e-02
3.73839974e-01 -5.68862557e-01 -4.77387100e-01 -4.17203695e-01
1.02273059e+00 7.53232121e-01 -8.82873356e-01 1.67106831e+00
-3.31164807e-01 4.16937590e-01 -1.13147390e+00 -1.52376875e-01
3.75594944e-01 1.04769543e-01 6.06138349e-01 1.11457491e+00
8.11846182e-02 2.71989048e-01 8.48466158e-02 8.19095492e-01
-4.08344686e-01 -1.65200666e-01 -3.77612859e-01 -1.10757858e-01
4.89048064e-01 1.24572730e+00 -7.78465152e-01 -3.37543964e-01
-4.60660756e-01 6.80677295e-01 5.68879664e-01 -2.76679575e-01
-9.91045237e-01 3.96332964e-02 9.79883969e-01 2.34229952e-01
4.72741067e-01 -4.36320901e-01 6.27806365e-01 -1.19183290e+00
-2.21899167e-01 -8.33086967e-01 6.34316027e-01 -7.75781333e-01
-1.65381181e+00 7.91439533e-01 1.86446100e-01 -1.39103293e+00
3.84470001e-02 -7.61157930e-01 -3.82758290e-01 4.18942183e-01
-7.41950810e-01 -1.53972948e+00 -1.69308305e-01 5.79486430e-01
2.39641652e-01 -1.07036352e-01 3.37397367e-01 6.77415252e-01
-4.26681459e-01 5.36833882e-01 7.98959211e-02 5.41931391e-01
6.31064057e-01 -9.05582011e-01 9.48008120e-01 7.65691161e-01
4.19382542e-01 6.70828581e-01 3.94801795e-01 -9.35736120e-01
-1.19960403e+00 -9.76546347e-01 3.00411999e-01 -9.25221622e-01
1.02978384e+00 -5.70055783e-01 -6.35140777e-01 1.34847021e+00
-1.46816388e-01 3.49954188e-01 4.84013945e-01 -1.71072818e-02
-5.22493243e-01 -3.92536633e-02 -1.27483809e+00 5.38846314e-01
1.58520532e+00 -3.02097350e-01 -1.18401539e+00 4.65861857e-01
2.55538762e-01 -3.93164217e-01 -8.99410307e-01 1.07915349e-01
1.34944069e+00 -6.59538925e-01 1.19072282e+00 -8.52356672e-01
5.22077322e-01 -4.00973737e-01 3.05004389e-04 -1.23739338e+00
-3.35847706e-01 3.60899791e-02 1.68958455e-01 7.64590085e-01
1.42244816e-01 -5.04087865e-01 5.28792799e-01 6.76495075e-01
-3.63580100e-02 6.28264993e-02 -1.43277109e+00 -1.06337023e+00
1.49343148e-01 -1.13149090e-02 4.64538217e-01 6.06763363e-01
-2.16206536e-01 6.74655512e-02 -8.45005989e-01 1.52779773e-01
1.01105487e+00 -1.22057468e-01 9.72145438e-01 -1.36271262e+00
-2.15629369e-01 -6.32768273e-01 -9.62111652e-01 -8.35411310e-01
1.54742271e-01 -9.00102019e-01 2.17849776e-01 -1.27245224e+00
-7.07802130e-03 -2.23969221e-01 1.25944957e-01 9.38300729e-01
-3.60541902e-02 7.09692180e-01 4.77810949e-01 3.99016321e-01
-7.23681808e-01 5.14738142e-01 1.11531842e+00 -7.73973688e-02
4.24980581e-01 -7.22165685e-03 1.11502185e-02 9.47710574e-01
2.25298464e-01 -8.30842674e-01 3.59211385e-01 -3.68504286e-01
3.44101787e-02 9.66450572e-02 7.49627650e-01 -1.50671506e+00
4.88565832e-01 1.96104646e-01 4.69649702e-01 -4.20623660e-01
5.53155959e-01 -1.03532445e+00 4.44825083e-01 8.16426694e-01
-6.15540743e-02 7.19603896e-02 2.91511863e-01 5.22482872e-01
2.94808805e-01 -4.56380136e-02 1.25198555e+00 -3.92088801e-01
-7.61452556e-01 7.71224082e-01 -3.83189529e-01 3.84449661e-01
8.73772860e-01 -7.97296762e-02 -5.11568248e-01 9.87192541e-02
-4.21862096e-01 3.90369296e-01 5.67583442e-01 6.33962870e-01
-1.13799470e-02 -1.20730817e+00 -6.33590877e-01 2.25569252e-02
2.41748556e-01 -4.86436874e-01 2.98580587e-01 6.51075542e-01
-1.19650817e+00 1.84275717e-01 -1.10905576e+00 -4.20279682e-01
-1.18943858e+00 4.49565291e-01 5.01203656e-01 3.26920420e-01
-8.09119523e-01 8.44542205e-01 -2.13254262e-02 -6.84514284e-01
1.84982643e-01 -2.41281256e-01 6.33440688e-02 1.98116736e-03
5.85268199e-01 5.61941385e-01 -4.93922770e-01 -1.34018588e+00
-6.51188135e-01 4.35706794e-01 1.60563424e-01 1.58263922e-01
1.84203517e+00 4.56157446e-01 -1.37210220e-01 -9.29684266e-02
8.70335221e-01 -9.97916013e-02 -1.36403966e+00 6.72158077e-02
-2.53639162e-01 -5.40988863e-01 -3.57248813e-01 -7.08324611e-01
-1.06909132e+00 8.99797857e-01 7.29666710e-01 -4.51076984e-01
1.76121294e-01 2.45481074e-01 9.87633348e-01 6.48494840e-01
9.32476997e-01 -8.32139015e-01 -5.97836196e-01 1.66952237e-01
9.23765838e-01 -1.29879856e+00 1.42218858e-01 3.90055835e-01
-4.85542059e-01 9.15525436e-01 1.13155615e+00 -2.32495964e-01
2.90674537e-01 -1.01420574e-01 -2.70153791e-01 -3.06382835e-01
-3.95713657e-01 -1.34319067e-01 3.38344276e-01 7.73040533e-01
-1.05251707e-01 2.85194844e-01 -1.04820460e-01 6.42246783e-01
-7.57331133e-01 -2.27609023e-01 1.92141518e-01 7.43448794e-01
-4.81752038e-01 -7.57555962e-01 -4.75147426e-01 5.85040748e-01
-2.73760110e-01 -1.02619529e-01 -5.30591190e-01 1.44549155e+00
3.15504402e-01 2.40612730e-01 2.61795729e-01 -2.16033589e-02
5.44544697e-01 -9.57513303e-02 6.84189618e-01 -4.82801437e-01
-8.35639298e-01 -5.99353731e-01 2.04266086e-01 -4.32754278e-01
-3.99910450e-01 -8.51501584e-01 -7.39434242e-01 -6.42391562e-01
1.33354872e-01 -1.39773324e-01 4.41702098e-01 6.22126400e-01
-1.96660876e-01 -3.08604002e-01 -2.51831412e-01 -1.10753536e+00
-3.51832986e-01 -1.05535603e+00 -6.03695035e-01 2.77889997e-01
3.09364319e-01 -9.15099144e-01 -3.40015203e-01 -2.92546391e-01]
|
[7.6716156005859375, -0.8754842877388]
|
bc9b3687-a944-428f-9c94-b72dcd53ced2
|
localize-group-and-select-boosting-text-vqa
|
2108.08965
| null |
https://arxiv.org/abs/2108.08965v1
|
https://arxiv.org/pdf/2108.08965v1.pdf
|
Localize, Group, and Select: Boosting Text-VQA by Scene Text Modeling
|
As an important task in multimodal context understanding, Text-VQA (Visual Question Answering) aims at question answering through reading text information in images. It differentiates from the original VQA task as Text-VQA requires large amounts of scene-text relationship understanding, in addition to the cross-modal grounding capability. In this paper, we propose Localize, Group, and Select (LOGOS), a novel model which attempts to tackle this problem from multiple aspects. LOGOS leverages two grounding tasks to better localize the key information of the image, utilizes scene text clustering to group individual OCR tokens, and learns to select the best answer from different sources of OCR (Optical Character Recognition) texts. Experiments show that LOGOS outperforms previous state-of-the-art methods on two Text-VQA benchmarks without using additional OCR annotation data. Ablation studies and analysis demonstrate the capability of LOGOS to bridge different modalities and better understand scene text.
|
['Carolyn P. Rose', 'Jean Oh', 'Yansen Wang', 'Zhen Fan', 'Xiaopeng Lu']
|
2021-08-20
| null | null | null | null |
['data-ablation', 'text-clustering']
|
['computer-vision', 'natural-language-processing']
|
[ 5.14964223e-01 -1.68072075e-01 -5.98943383e-02 -4.88845050e-01
-1.29642582e+00 -9.44952846e-01 8.15352440e-01 4.91139174e-01
-2.78965890e-01 2.14490429e-01 4.43626165e-01 -4.96423095e-01
5.02346829e-02 -3.16833705e-01 -8.01880360e-01 -3.23981971e-01
5.46516597e-01 6.86646581e-01 4.52704102e-01 -2.98036456e-01
6.47232234e-01 1.02306809e-02 -1.35650325e+00 1.12836277e+00
1.01299560e+00 9.29877341e-01 4.60794210e-01 9.97421265e-01
-8.07487071e-01 1.37475121e+00 -6.26830518e-01 -5.00571787e-01
-1.89063832e-01 -6.95528328e-01 -1.36207938e+00 3.66935343e-01
1.37840104e+00 -2.73764074e-01 -2.15731308e-01 8.19440484e-01
1.76399752e-01 6.61275163e-02 7.00461388e-01 -1.34618700e+00
-9.91580427e-01 4.49416578e-01 -4.05301660e-01 3.11484665e-01
7.90571094e-01 1.99012756e-01 1.49942124e+00 -1.21135461e+00
6.32583201e-01 1.52412701e+00 2.90934056e-01 5.34135878e-01
-1.05334353e+00 -8.65339041e-02 2.00855017e-01 4.06836599e-01
-1.23298109e+00 -3.39522481e-01 6.18163884e-01 -4.90310431e-01
9.83274162e-01 4.70363349e-01 3.67700666e-01 1.11893392e+00
-1.52532160e-02 1.56691146e+00 1.06263423e+00 -5.53039610e-01
4.93832827e-02 6.90069422e-02 2.35398650e-01 9.54949319e-01
-1.03816010e-01 -6.87294662e-01 -1.03452182e+00 5.41411974e-02
4.15458649e-01 -1.55942172e-01 -4.45957124e-01 -5.28678834e-01
-1.49146211e+00 5.98885655e-01 4.29283649e-01 1.05040871e-01
1.31885856e-01 6.12738013e-01 4.40155178e-01 1.79592147e-01
2.10804790e-01 6.76066637e-01 -3.81885171e-01 -4.09587473e-03
-9.05852497e-01 2.39885412e-02 6.13960624e-01 1.01968145e+00
9.66273665e-01 -1.00448042e-01 -5.37350059e-01 6.64851665e-01
4.24597025e-01 7.66797364e-01 1.03970177e-01 -1.03837860e+00
1.15038097e+00 1.01698172e+00 -1.70619071e-01 -6.81101501e-01
-1.31186560e-01 1.40281290e-01 -2.97993749e-01 -2.23545954e-01
5.98364353e-01 2.98868686e-01 -1.44950044e+00 1.19038415e+00
7.72058070e-02 -2.24675804e-01 1.66548923e-01 9.82698619e-01
1.24527144e+00 8.32695901e-01 2.12337062e-01 2.89300293e-01
1.71199262e+00 -1.29839671e+00 -7.72957265e-01 -6.76733255e-01
5.92290103e-01 -8.78279567e-01 1.59536850e+00 1.86239108e-01
-9.39535975e-01 -4.80297267e-01 -8.18434417e-01 -8.04816663e-01
-7.24678576e-01 3.68557930e-01 4.10150141e-01 5.45237243e-01
-1.27757025e+00 -1.97812945e-01 -3.49697471e-01 -6.49200976e-01
5.99481761e-01 -1.00962661e-01 -2.79942721e-01 -5.41346312e-01
-8.73486042e-01 6.08941138e-01 2.96904087e-01 -4.16238382e-02
-1.29261541e+00 -6.82325840e-01 -1.11318254e+00 1.98440608e-02
7.37250209e-01 -9.20762539e-01 1.17769754e+00 -1.20485318e+00
-1.05341315e+00 1.10502040e+00 -6.20548606e-01 -2.57313877e-01
3.13706309e-01 -4.48647797e-01 -6.10464588e-02 9.25456226e-01
3.32980752e-01 9.10435140e-01 1.16649282e+00 -1.59400952e+00
-5.41113734e-01 -4.62201893e-01 8.80951583e-02 5.14507413e-01
-2.53583044e-02 -2.31421843e-01 -9.44611371e-01 -4.97572690e-01
1.39445394e-01 -5.19696593e-01 9.48010758e-02 1.06385216e-01
-6.01285458e-01 -3.88367593e-01 1.16740704e+00 -7.44609058e-01
8.60419631e-01 -1.87801540e+00 3.01849693e-01 -7.89056644e-02
3.21464777e-01 -1.25465110e-01 -3.32022399e-01 7.04618394e-01
2.80679137e-01 3.09932113e-01 -3.00471038e-01 -4.81384784e-01
3.66981961e-02 2.86020398e-01 -7.15514839e-01 1.36813641e-01
4.88542765e-01 1.49124253e+00 -8.34966421e-01 -1.07817090e+00
3.34708035e-01 1.89233795e-01 -2.82550007e-01 2.86825430e-02
-9.17737961e-01 1.84120253e-01 -5.72672784e-01 1.15176451e+00
3.34890366e-01 -5.70085049e-01 -4.88771871e-02 -3.43477994e-01
2.36375958e-01 -2.89457850e-02 -6.67210639e-01 2.01528168e+00
-1.23794690e-01 1.25039387e+00 -6.07157946e-02 -9.71982360e-01
6.06064379e-01 7.71384984e-02 1.32821426e-01 -9.52981830e-01
1.47125766e-01 2.90936194e-02 -5.83291888e-01 -9.11571026e-01
7.71490157e-01 3.61959130e-01 -1.56794325e-01 4.11643833e-01
2.97530502e-01 -5.78953207e-01 1.93643928e-01 8.75988007e-01
7.98263371e-01 1.22482322e-01 1.23562820e-01 -2.00261936e-01
6.76366866e-01 4.26547438e-01 -3.13807786e-01 1.06686592e+00
-4.23910439e-01 7.74517357e-01 5.67210019e-01 -1.29204601e-01
-8.71771336e-01 -1.19090581e+00 1.29446417e-01 1.46449459e+00
5.89576423e-01 -5.11671782e-01 -7.51986086e-01 -8.81833732e-01
2.19612028e-02 7.62922168e-01 -9.82858062e-01 4.03118789e-01
-2.86906898e-01 -8.82665962e-02 6.88213944e-01 6.68966591e-01
5.01366317e-01 -9.37053025e-01 -3.51463050e-01 -3.74814063e-01
-6.88069642e-01 -1.54252839e+00 -5.99306285e-01 -2.00311281e-03
-6.98752999e-01 -1.26262367e+00 -5.75807154e-01 -9.77298439e-01
6.18046403e-01 6.92698479e-01 1.51074505e+00 2.16122925e-01
-4.69674051e-01 1.50975430e+00 -6.90598726e-01 -3.85831505e-01
-2.68478632e-01 9.05754045e-04 -6.73863351e-01 1.86653644e-01
3.55288625e-01 3.10966969e-01 -5.29916525e-01 2.10694030e-01
-1.08650017e+00 8.89148340e-02 5.74065685e-01 6.66844904e-01
6.68198049e-01 -3.54026079e-01 2.19600186e-01 -5.98011851e-01
4.84234393e-01 -1.91680357e-01 -3.21728110e-01 9.59252477e-01
-1.67920828e-01 1.88396245e-01 1.79267988e-01 -1.41161919e-01
-1.06938982e+00 9.33821872e-02 3.36263895e-01 -3.84772986e-01
-2.81252623e-01 2.61465132e-01 -2.09125191e-01 8.56787339e-02
6.49678886e-01 6.10045135e-01 -2.53473431e-01 -1.44807890e-01
9.69070017e-01 3.55682522e-01 6.04304254e-01 -7.27428317e-01
7.34831214e-01 8.38482022e-01 -1.40318930e-01 -1.17532361e+00
-1.15370154e+00 -8.76674891e-01 -8.36530924e-01 -4.96332347e-01
1.66159153e+00 -9.48081136e-01 -7.24430978e-01 2.17609838e-01
-1.26092327e+00 -4.38078046e-01 -1.71069086e-01 -3.20972316e-02
-5.86304009e-01 5.99151552e-01 -2.75401056e-01 -7.85073757e-01
-2.90365487e-01 -1.01090086e+00 1.78598666e+00 1.23905115e-01
2.27539018e-01 -1.01316237e+00 -3.55449587e-01 1.18973613e+00
-1.00853145e-01 1.40638515e-01 1.02782190e+00 -5.48657060e-01
-1.19119728e+00 2.08810627e-01 -5.89755177e-01 9.44841355e-02
-2.25766584e-01 -2.06990987e-01 -1.01191270e+00 -7.75224268e-02
-3.53912741e-01 -8.65776360e-01 1.21515191e+00 2.40132958e-01
1.25952959e+00 -2.29550973e-02 -2.25937217e-01 3.76017720e-01
1.48061037e+00 -2.83359718e-02 6.59276068e-01 3.23723197e-01
1.21430457e+00 8.74884963e-01 6.86357677e-01 1.27146700e-02
7.74778366e-01 3.28321874e-01 7.87706017e-01 -3.40372652e-01
-4.00887787e-01 -3.53110522e-01 3.47718865e-01 4.99333620e-01
3.70101452e-01 -5.68652034e-01 -1.35263765e+00 7.75274873e-01
-1.90177000e+00 -8.75702381e-01 -5.09641528e-01 1.59585905e+00
6.28770292e-01 -3.64481330e-01 -1.98247865e-01 -1.97911233e-01
4.14794654e-01 3.57168645e-01 -5.59098482e-01 -3.08241934e-01
-4.22883093e-01 -8.69278237e-02 2.65301049e-01 5.52801907e-01
-1.05021024e+00 1.51685631e+00 6.34805918e+00 6.98394299e-01
-6.24393642e-01 1.51535030e-02 5.77055633e-01 1.05255701e-01
-4.86036688e-01 1.69239178e-01 -6.10446870e-01 -1.86472535e-01
1.84676856e-01 2.64708102e-01 3.88988644e-01 6.20313764e-01
-2.07188040e-01 -4.33479100e-01 -1.27522087e+00 1.16153944e+00
9.02436137e-01 -1.44381952e+00 7.65739620e-01 -4.08638895e-01
7.35214710e-01 7.52108246e-02 1.53666362e-01 1.35011569e-01
2.40801960e-01 -1.32568097e+00 9.57735181e-01 5.48326135e-01
6.38936281e-01 -3.29758078e-01 4.40197825e-01 -1.00615369e-02
-1.43599653e+00 -1.03954881e-01 -1.13402471e-01 4.61640149e-01
1.76749200e-01 3.42641585e-02 -9.87555802e-01 6.59177959e-01
9.04379010e-01 6.76047623e-01 -1.48693907e+00 5.45634210e-01
-3.57206345e-01 5.38013935e-01 1.66850299e-01 -2.29092881e-01
5.29596090e-01 9.43644568e-02 4.49949384e-01 1.27285457e+00
-1.82354018e-01 9.25598964e-02 3.51298988e-01 9.10418987e-01
-1.63951978e-01 3.09538811e-01 -3.73997927e-01 -3.21510404e-01
1.19141713e-01 9.96617377e-01 -7.96839476e-01 -4.74373847e-01
-4.44534034e-01 1.46203363e+00 1.87112704e-01 8.26195300e-01
-5.03511071e-01 -3.50409448e-01 2.17016608e-01 -1.08746707e-01
4.52461839e-01 -4.58192647e-01 -3.55529279e-01 -1.32914555e+00
-8.56649280e-02 -1.12836158e+00 7.23140061e-01 -1.59813988e+00
-1.11511087e+00 3.56450021e-01 -1.65276960e-01 -9.15843308e-01
1.74040571e-01 -1.00476861e+00 -3.30806553e-01 5.31789601e-01
-1.65154719e+00 -1.82831466e+00 -7.08551109e-01 1.08989334e+00
1.21418333e+00 -3.92598808e-02 2.82614708e-01 -3.00446719e-01
-1.94118366e-01 3.00900191e-01 -5.93347363e-02 3.01706314e-01
9.28451538e-01 -1.64964879e+00 3.01367462e-01 9.28445816e-01
7.06276119e-01 4.49686289e-01 4.68817383e-01 -5.90354860e-01
-1.82310998e+00 -9.56804335e-01 6.05185747e-01 -1.26627398e+00
7.77639985e-01 -6.61208272e-01 -9.13336039e-01 6.94025755e-01
6.64949477e-01 -2.93953538e-01 6.62260652e-01 -7.98286870e-02
-7.84257770e-01 7.34519288e-02 -4.71708953e-01 7.48879671e-01
7.66543984e-01 -1.06347144e+00 -9.39359546e-01 4.21881586e-01
9.09229636e-01 -3.26739311e-01 -6.32563829e-01 -2.91932579e-02
2.58809090e-01 -7.77168810e-01 1.14465320e+00 -7.08748221e-01
7.33772933e-01 -5.83253980e-01 -5.73215306e-01 -7.47723997e-01
2.97436297e-01 -2.91273177e-01 7.13174641e-02 1.29790676e+00
5.26013792e-01 -5.83984442e-02 6.17185056e-01 2.93487430e-01
-5.29500954e-02 -3.40871215e-01 -9.02345777e-01 -3.55110824e-01
3.18734581e-03 -7.43113220e-01 1.66693851e-01 1.14696157e+00
-1.91884473e-01 7.78476954e-01 -2.96856225e-01 3.89446467e-01
6.93878829e-01 2.24709362e-01 1.03322160e+00 -9.61495876e-01
-5.42671308e-02 -3.64560694e-01 -2.80947745e-01 -1.37182331e+00
5.37587516e-02 -1.03745103e+00 1.49973676e-01 -2.21280670e+00
3.53456944e-01 4.75758873e-02 7.60356784e-02 5.81058443e-01
-5.27382851e-01 3.11330676e-01 5.19755423e-01 2.70422369e-01
-1.45115221e+00 6.35665059e-01 1.35732937e+00 -6.78261757e-01
2.16945801e-02 -6.38607502e-01 -6.57830477e-01 4.92695421e-01
4.33950216e-01 8.69861022e-02 -5.54427147e-01 -9.96606827e-01
4.31051195e-01 4.83687185e-02 7.67522037e-01 -6.73432648e-01
4.67522323e-01 -1.39628291e-01 5.73668897e-01 -1.16052794e+00
4.18398529e-01 -6.93760872e-01 -7.88726985e-01 9.74990278e-02
-4.99742240e-01 -4.38455446e-03 2.74817735e-01 9.01803195e-01
-2.98386455e-01 -1.41506359e-01 3.51674706e-01 -1.54216319e-01
-1.28008497e+00 1.89871229e-02 -5.33449054e-01 4.76936579e-01
8.35473895e-01 -2.50924319e-01 -1.06194651e+00 -6.23676419e-01
-4.25528944e-01 8.56804490e-01 4.41375107e-01 7.37096846e-01
9.23234046e-01 -9.12129819e-01 -5.95947802e-01 -3.11997026e-01
8.96731675e-01 -3.76578122e-02 3.94479871e-01 5.95577478e-01
-7.56710529e-01 6.66120172e-01 3.02913457e-01 -1.21664464e+00
-1.46628356e+00 6.09618247e-01 3.90740484e-01 -1.16779227e-02
-4.26430017e-01 8.21440637e-01 5.02613723e-01 -3.49919349e-01
2.40152285e-01 -3.84233564e-01 -2.56208569e-01 2.16117382e-01
5.78485787e-01 8.93806666e-02 -1.72188357e-01 -7.66152024e-01
-4.15088683e-01 1.01500380e+00 -3.16014364e-02 -3.83978188e-01
7.02729464e-01 -6.16664231e-01 -2.30023667e-01 5.11864841e-01
1.11056530e+00 -1.20181702e-01 -1.10364854e+00 -4.41717476e-01
2.35394523e-01 -5.08392394e-01 -3.59015875e-02 -1.09534252e+00
-5.84085345e-01 1.29266071e+00 4.57894385e-01 1.40240833e-01
1.06509042e+00 6.67256415e-01 5.05786777e-01 7.51447380e-01
-1.17869511e-01 -1.18102634e+00 1.09347105e+00 7.00290918e-01
1.02036166e+00 -1.72964942e+00 1.26117036e-01 -4.26692814e-01
-1.23594093e+00 1.24384427e+00 8.15304935e-01 4.67752010e-01
7.99503922e-02 -3.32843632e-01 4.83958781e-01 -6.92669809e-01
-6.18009269e-01 -7.42811382e-01 8.43560815e-01 6.67525232e-01
1.34648517e-01 -2.70914584e-01 3.39397788e-01 1.13140419e-01
6.38131276e-02 -7.06244051e-01 4.66385365e-01 8.97900283e-01
-5.95724583e-01 -7.92106569e-01 -5.95924437e-01 2.67800152e-01
-3.17197829e-01 -4.57937002e-01 -1.06215680e+00 8.97907615e-01
-2.81662703e-01 1.27770448e+00 1.50431588e-01 -7.00128973e-02
1.95312425e-01 2.49446228e-01 4.66999382e-01 -5.58375180e-01
-4.31316435e-01 -4.37729768e-02 -6.58837259e-02 -7.36876369e-01
-5.88432550e-01 -4.94741559e-01 -1.22507918e+00 3.38050723e-01
-3.19944054e-01 -8.94627422e-02 8.27887893e-01 1.09938538e+00
2.23786697e-01 5.46968818e-01 1.97777793e-01 -4.99565214e-01
1.27921864e-01 -5.78335345e-01 -1.18020751e-01 6.06695771e-01
6.24462187e-01 -2.14846253e-01 -2.91827470e-01 4.70580578e-01]
|
[11.122151374816895, 1.8400673866271973]
|
e9c1e4d3-ae5b-444d-85fd-f0ba3bd39fd8
|
respiratory-rate-estimation-from-face-videos
|
1909.03503
| null |
https://arxiv.org/abs/1909.03503v1
|
https://arxiv.org/pdf/1909.03503v1.pdf
|
Respiratory Rate Estimation from Face Videos
|
Vital signs, such as heart rate (HR), heart rate variability (HRV), respiratory rate (RR), are important indicators for a person's health. Vital signs are traditionally measured with contact sensors, and may be inconvenient and cause discomfort during continuous monitoring. Commercial cameras are promising contact-free sensors, and remote photoplethysmography (rPPG) have been studied to remotely monitor heart rate from face videos. For remote RR measurement, most prior art was based on small periodical motions of chest regions caused by breathing cycles, which are vulnerable to subjects' voluntary movements. This paper explores remote RR measurement based on rPPG obtained from face videos. The paper employs motion compensation, two-phase temporal filtering, and signal pruning to capture signals with high quality. The experimental results demonstrate that the proposed framework can obtain accurate RR results and can provide HR, HRV and RR measurement synergistically in one framework.
|
['Min Wu', 'Mingliang Chen', 'Qiang Zhu', 'Harrison Zhang', 'Quanzeng Wang']
|
2019-09-08
| null | null | null | null |
['heart-rate-variability', 'respiratory-rate-estimation']
|
['medical', 'medical']
|
[ 3.73100102e-01 -3.69585931e-01 -1.42150298e-01 -2.28698194e-01
-1.14153763e-02 -1.62354112e-01 -6.95504919e-02 -7.53313482e-01
-1.02368303e-01 7.63358831e-01 2.41829455e-01 3.72594476e-01
1.62623405e-01 -4.29794312e-01 2.94268668e-01 -8.60674798e-01
8.51710215e-02 -7.14006960e-01 -3.50521117e-01 4.06157598e-02
1.03531942e-01 5.29228687e-01 -1.30250454e+00 -2.88156152e-01
6.20311618e-01 1.13680661e+00 -4.24218178e-01 7.43205726e-01
3.64601880e-01 4.40585881e-01 -8.28848124e-01 5.97525202e-02
2.71361042e-03 -9.48725045e-01 -1.37462378e-01 -1.49319440e-01
1.42491922e-01 -6.72178030e-01 -4.53832567e-01 6.10659182e-01
9.88155901e-01 3.91673073e-02 2.86367416e-01 -1.01578772e+00
-4.11666453e-01 -4.71339077e-02 -8.34939420e-01 5.51850617e-01
7.72649944e-01 4.02371407e-01 -7.93856289e-03 -6.61998510e-01
3.08750659e-01 1.02462423e+00 1.13488078e+00 8.76238942e-01
-1.15424418e+00 -6.35375082e-01 -6.32234931e-01 2.57036597e-01
-1.48551714e+00 -8.68539274e-01 1.16591418e+00 -1.03634514e-01
5.85645854e-01 8.53245616e-01 9.50292826e-01 1.03409910e+00
4.71525699e-01 -2.08670467e-01 1.46730995e+00 -1.08105958e-01
-1.50888607e-01 2.03272149e-01 -5.45743369e-02 4.06886727e-01
3.36557090e-01 1.64921775e-01 -5.95292866e-01 -1.93034515e-01
1.25112462e+00 3.55647922e-01 -1.08092117e+00 3.20867717e-01
-1.19023132e+00 2.77699679e-01 7.85267055e-02 4.49944049e-01
-3.96960855e-01 1.85654894e-01 2.75060654e-01 3.20256829e-01
2.34630257e-01 -2.33602449e-02 -7.77582228e-02 -5.42285264e-01
-8.87704253e-01 -4.16437417e-01 9.20760274e-01 5.33736169e-01
1.52219459e-01 1.90984458e-01 -4.60954249e-01 7.69957006e-01
6.48509920e-01 1.00602579e+00 7.37378538e-01 -1.26959956e+00
8.37446935e-03 2.36763507e-01 1.62178621e-01 -1.22552001e+00
-4.46784437e-01 1.21068925e-01 -1.16577184e+00 -2.38463119e-01
1.41636655e-01 -2.23870724e-01 -2.40827948e-01 1.45244014e+00
7.07196951e-01 6.08907104e-01 -1.56056419e-01 1.29319811e+00
1.23988533e+00 5.79739630e-01 -4.59324918e-04 -1.37605059e+00
1.50655198e+00 -2.01515406e-01 -1.42990685e+00 1.63232252e-01
-6.05901033e-02 -7.87924469e-01 9.11739647e-01 2.96440780e-01
-8.91660273e-01 -8.86127114e-01 -7.88521469e-01 1.59409538e-01
3.61373246e-01 3.58360671e-02 2.96423249e-02 1.08843899e+00
-8.73113811e-01 9.66379821e-01 -6.77734137e-01 -3.08491379e-01
2.59955805e-02 -1.74155414e-01 -7.40134716e-02 1.72470793e-01
-1.32260513e+00 6.40152395e-01 -5.49024522e-01 8.82217228e-01
-2.10616857e-01 -6.46478653e-01 -5.84882319e-01 -2.79943913e-01
1.23226628e-01 -6.56678736e-01 9.08651888e-01 -4.54428732e-01
-2.23956490e+00 7.28698909e-01 -4.48087871e-01 1.50698259e-01
5.44904947e-01 -3.65736365e-01 -8.71982098e-01 7.62357056e-01
-6.29070759e-01 -1.42120406e-01 1.25497997e+00 -6.40087962e-01
5.99123001e-01 -4.55197722e-01 -8.83614957e-01 2.87269473e-01
-1.92726016e-01 3.62425864e-01 4.87586260e-02 -3.49388957e-01
2.65867501e-01 -7.30283022e-01 2.64326394e-01 1.41622692e-01
5.46826907e-02 1.79578468e-01 8.94796908e-01 -9.61704195e-01
1.56193018e+00 -1.93455327e+00 -3.29729915e-01 8.09119120e-02
1.96562946e-01 4.98752385e-01 3.58056605e-01 1.28995404e-01
-4.42100242e-02 2.68684238e-01 -1.54991418e-01 1.60650149e-01
-4.96940255e-01 -3.22419368e-02 -5.40129095e-02 1.12767172e+00
-3.06056857e-01 1.01882124e+00 -3.98844808e-01 -6.57060444e-01
5.12170553e-01 1.05730605e+00 2.23525628e-01 4.90356535e-01
8.18843246e-01 9.73127782e-01 -2.59331822e-01 7.74657547e-01
7.51301944e-01 1.02757856e-01 4.67958339e-02 -6.85447872e-01
-1.19480215e-01 -9.39814895e-02 -1.11024094e+00 1.11282647e+00
-2.46080339e-01 5.99702716e-01 3.46480101e-01 -5.71841180e-01
1.40056384e+00 9.03204501e-01 7.20980406e-01 -7.04947174e-01
2.99672246e-01 -1.62015870e-01 -2.92501211e-01 -1.21693063e+00
-3.01201433e-01 -4.40303266e-01 4.79211181e-01 4.20944422e-01
-5.57127535e-01 -3.98508571e-02 -4.20310199e-01 -4.31301951e-01
8.21437001e-01 2.76296139e-01 5.32501400e-01 -1.12792157e-01
9.19001937e-01 -7.41335869e-01 7.12930679e-01 1.02431506e-01
-7.47324944e-01 7.79643238e-01 1.05192021e-01 -4.47459877e-01
-4.44285333e-01 -6.59858227e-01 -4.11606401e-01 9.06172693e-02
1.83350876e-01 -5.32419458e-02 -5.82331836e-01 -4.55988087e-02
-1.41635880e-01 2.71087624e-02 -3.53620231e-01 -4.02824938e-01
-7.60046363e-01 -8.34254086e-01 5.40642798e-01 3.24130893e-01
7.42226541e-01 -1.27120864e+00 -1.10115433e+00 2.68367320e-01
-6.30898476e-01 -1.00612128e+00 -5.38497925e-01 -6.99074626e-01
-1.38040853e+00 -1.14576447e+00 -9.87716496e-01 -7.01106563e-02
2.75888026e-01 5.52587509e-01 8.98347020e-01 1.12964161e-01
-9.50143456e-01 7.55890727e-01 -1.26526311e-01 -1.29040182e-01
-1.69455365e-04 -6.73893869e-01 2.82570779e-01 3.93902600e-01
3.19120556e-01 -8.47506344e-01 -1.29844368e+00 6.35133326e-01
-2.94657022e-01 -5.74172318e-01 1.43861070e-01 1.72251999e-01
5.65095127e-01 -3.66885632e-01 7.27241158e-01 -4.81033206e-01
7.76024759e-01 -1.39666706e-01 -2.46606663e-01 -9.23029706e-02
-7.91653037e-01 -6.89428568e-01 4.09798086e-01 -5.38229883e-01
-1.26937532e+00 -4.97902483e-01 2.21843883e-01 -6.46236241e-01
-3.90666664e-01 1.72838084e-02 1.51284027e-03 -2.17672363e-01
7.66311467e-01 2.07213089e-01 5.23407698e-01 -4.19227362e-01
2.11317185e-02 7.84855902e-01 7.44718313e-01 -2.21826229e-02
5.85601747e-01 4.54274952e-01 3.44404727e-01 -1.41982603e+00
-2.92178869e-01 -6.35854602e-01 -5.03713548e-01 -8.07762563e-01
9.13663089e-01 -1.06095016e+00 -1.27914476e+00 5.90086401e-01
-8.56716275e-01 7.97288865e-02 -1.69981509e-01 9.58444595e-01
-3.19614321e-01 8.06170404e-01 -8.77360284e-01 -1.35831022e+00
-9.04351175e-01 -1.77494869e-01 6.10649943e-01 9.57239151e-01
-4.39353913e-01 -8.25528860e-01 2.54948944e-01 7.10257232e-01
8.66946876e-01 9.17086720e-01 -3.50886881e-02 5.75286448e-01
-4.21314277e-02 -3.68703365e-01 1.75400108e-01 3.37227583e-01
4.20233011e-01 2.38264829e-01 -1.35823238e+00 -2.29016870e-01
8.52286041e-01 -3.18060704e-02 3.66367370e-01 6.49143159e-01
1.04459405e+00 -5.48328519e-01 -1.30352780e-01 8.36099625e-01
1.39546216e+00 3.77520561e-01 1.23289478e+00 -4.85805213e-01
6.33965254e-01 6.91764891e-01 5.19730091e-01 6.21013701e-01
-8.36776868e-02 3.46120864e-01 1.69433683e-01 -3.21695417e-01
-1.82800218e-02 1.56105042e-01 7.50467300e-01 9.95275497e-01
-9.79061842e-01 1.61856443e-01 -2.34243616e-01 -3.67793664e-02
-1.25897050e+00 -1.32361567e+00 -7.05377817e-01 2.43250680e+00
9.08899605e-01 -5.85501611e-01 1.74544364e-01 3.82964015e-01
9.91261959e-01 2.08648473e-01 -6.43263280e-01 -4.67098862e-01
1.74598470e-01 4.13927317e-01 3.00373018e-01 8.43707174e-02
-6.35332882e-01 -8.78636241e-02 6.77443647e+00 -2.68425226e-01
-1.33843637e+00 1.07260123e-01 5.00413299e-01 -3.48125160e-01
-3.01829521e-02 -2.36379728e-01 -5.60876489e-01 7.03261495e-01
1.26399362e+00 -1.36302248e-01 2.62127221e-01 4.75903779e-01
9.22466636e-01 -3.81765932e-01 -7.12143958e-01 1.64313316e+00
1.77512690e-01 -5.39473116e-01 -6.73767388e-01 8.43342990e-02
1.80102989e-01 -7.37320960e-01 -1.65241033e-01 -1.78122446e-01
-1.00969326e+00 -1.16558492e+00 -4.66063797e-01 1.20402598e+00
1.20777059e+00 -5.45498669e-01 6.38669968e-01 6.58380985e-02
-1.24768806e+00 3.06446701e-01 -4.36563104e-01 -3.14777195e-02
1.38612881e-01 9.92819726e-01 -1.71907544e-01 3.27705950e-01
5.79016626e-01 8.48545551e-01 -1.22530177e-01 7.97629952e-01
-2.53738910e-01 7.48224556e-01 -3.13751101e-01 7.79717937e-02
-8.55061829e-01 -6.45673037e-01 6.72968626e-01 7.23145843e-01
5.07234871e-01 9.46877778e-01 -2.98287988e-01 1.07706964e+00
1.83160901e-01 1.63967367e-02 -7.26903260e-01 1.56554043e-01
3.45624328e-01 1.58297932e+00 -4.14102286e-01 -4.93324250e-02
-4.76441413e-01 7.62537003e-01 -8.24450612e-01 4.21454966e-01
-1.15560210e+00 -6.21333897e-01 5.84254086e-01 3.56084794e-01
-4.13341761e-01 -5.20751178e-02 -8.69361907e-02 -1.29045320e+00
2.79057384e-01 -5.69935560e-01 2.14241415e-01 -9.15915072e-01
-8.53109539e-01 1.62611052e-01 -2.11771920e-01 -1.39349353e+00
5.16294204e-02 1.84845254e-01 -1.08629358e+00 1.23364651e+00
-1.52336991e+00 -2.15017170e-01 -1.08706498e+00 8.15855443e-01
8.14979598e-02 5.41951537e-01 8.08699012e-01 4.03028011e-01
-9.28635120e-01 4.60665882e-01 -5.02713263e-01 -2.68537581e-01
9.00937915e-01 -7.39232063e-01 -2.50797182e-01 5.93951702e-01
-5.51150858e-01 6.16901994e-01 5.06182492e-01 -4.93838161e-01
-1.80017018e+00 -7.54033804e-01 5.89865506e-01 -2.54244715e-01
-1.81865901e-01 2.75702327e-01 -9.89581347e-01 2.92303767e-02
-9.77925118e-03 5.99346519e-01 8.22652340e-01 -5.61348438e-01
1.73382476e-01 -6.13790631e-01 -1.54375815e+00 1.14125982e-01
8.87112916e-01 -5.12704313e-01 -6.13866568e-01 -1.82975233e-01
2.49837756e-01 -1.90551102e-01 -1.69729233e+00 4.47466940e-01
1.04421532e+00 -9.98177946e-01 1.00262702e+00 1.90396234e-01
-6.23108856e-02 -3.78027260e-01 5.31444490e-01 -8.19293797e-01
-2.52791137e-01 -1.30010998e+00 -5.52355409e-01 1.57204509e+00
-3.90278518e-01 -9.91163850e-01 3.88387859e-01 8.99388969e-01
3.70498747e-01 -1.48180485e-01 -7.78919101e-01 -7.20039725e-01
-8.11296344e-01 1.59462959e-01 2.00062960e-01 9.93862689e-01
3.60468954e-01 3.39994013e-01 -7.10279882e-01 -1.13673806e-01
8.10367525e-01 -3.06144706e-03 6.61267340e-01 -1.38418436e+00
6.82241190e-03 7.15253921e-03 -1.51138261e-01 -5.08510768e-01
-6.20460927e-01 -1.33908406e-01 -7.38536492e-02 -1.32596743e+00
-1.48610212e-02 2.02658340e-01 -3.54497939e-01 3.75026315e-02
-2.15883166e-01 5.86588860e-01 -1.79350287e-01 3.97468656e-01
1.48510680e-01 4.95815963e-01 1.50105059e+00 4.34726745e-01
-7.62489080e-01 4.75848913e-02 -3.20144564e-01 4.84153986e-01
7.25269377e-01 -1.89403683e-01 -4.64823157e-01 4.65951860e-01
3.40999849e-03 9.09882843e-01 3.90702128e-01 -1.11587000e+00
-5.90834059e-02 -9.56977755e-02 9.36573505e-01 -1.66684076e-01
3.86168271e-01 -7.91855216e-01 7.48048365e-01 8.16660225e-01
2.74183806e-02 1.92281306e-02 -1.62401393e-01 4.97999609e-01
-1.80991158e-01 2.70902783e-01 1.15188408e+00 -3.66567224e-01
-1.71774160e-02 3.79221857e-01 -2.56822616e-01 -2.42017806e-02
9.34583127e-01 -7.38319576e-01 -5.33302844e-01 -4.44810271e-01
-7.86299586e-01 -1.42777041e-01 6.90349117e-02 1.33590162e-01
8.97113621e-01 -1.51909411e+00 -5.74045718e-01 4.39678729e-01
-2.59267658e-01 -3.46750617e-01 6.53073132e-01 1.60882056e+00
-6.25416934e-01 1.81197017e-01 -2.35923782e-01 -7.37528503e-01
-1.56525350e+00 2.97808558e-01 6.15513444e-01 2.75789082e-01
-9.23917651e-01 4.50411558e-01 -3.79369587e-01 4.16523606e-01
3.82404812e-02 -1.53425992e-01 -6.40409887e-01 3.01241636e-01
8.10692847e-01 9.83010828e-01 -2.60948449e-01 -4.60520148e-01
-4.84936178e-01 1.28198516e+00 6.84213221e-01 -3.60059738e-02
7.93581903e-01 -9.43696141e-01 -1.32799000e-01 5.42839587e-01
1.12601101e+00 -1.08014017e-01 -1.12656105e+00 1.61233351e-01
-5.98265767e-01 -6.28894627e-01 4.15265746e-02 -4.96017247e-01
-1.30492783e+00 9.22085822e-01 9.28322911e-01 2.66015172e-01
1.43811893e+00 -6.97048128e-01 9.39795554e-01 -1.44996226e-01
1.43240094e-01 -1.12420559e+00 1.52139381e-01 -4.42696929e-01
8.80219400e-01 -7.86132216e-01 2.86153316e-01 -5.31385124e-01
-4.39202011e-01 1.46833396e+00 4.95549351e-01 8.23884457e-02
7.11407840e-01 -4.24048528e-02 3.28677177e-01 1.54993936e-01
-6.94600999e-01 1.09721586e-01 1.75171062e-01 6.30291581e-01
8.58791888e-01 -2.34951764e-01 -7.13182032e-01 1.73713624e-01
1.48873284e-01 4.84304935e-01 6.98161304e-01 7.97928095e-01
-4.18474227e-01 -3.32770348e-01 -6.97803617e-01 3.75106722e-01
-8.46196055e-01 3.73283207e-01 -2.66034901e-01 4.80122626e-01
-2.20011175e-01 1.31079006e+00 -1.34681687e-01 -6.38622195e-02
4.79121387e-01 3.64759475e-01 7.38005936e-01 -1.24542870e-01
-4.86901760e-01 4.52613831e-01 -1.04900233e-01 -1.00931013e+00
-8.18177581e-01 -5.29084206e-01 -9.89852905e-01 -3.87535632e-01
-1.85363933e-01 -2.36448929e-01 5.52367330e-01 5.04394174e-01
3.02875608e-01 4.04596180e-01 1.15563321e+00 -4.55482394e-01
-4.35613602e-01 -1.06021285e+00 -8.63265812e-01 3.83560807e-01
4.95669246e-01 -2.20570639e-01 -8.28672826e-01 2.68968940e-01]
|
[13.898090362548828, 2.8249173164367676]
|
75944290-705e-4316-acd5-31ef064b4ffb
|
attentionhtr-handwritten-text-recognition
|
2201.09390
| null |
https://arxiv.org/abs/2201.09390v3
|
https://arxiv.org/pdf/2201.09390v3.pdf
|
AttentionHTR: Handwritten Text Recognition Based on Attention Encoder-Decoder Networks
|
This work proposes an attention-based sequence-to-sequence model for handwritten word recognition and explores transfer learning for data-efficient training of HTR systems. To overcome training data scarcity, this work leverages models pre-trained on scene text images as a starting point towards tailoring the handwriting recognition models. ResNet feature extraction and bidirectional LSTM-based sequence modeling stages together form an encoder. The prediction stage consists of a decoder and a content-based attention mechanism. The effectiveness of the proposed end-to-end HTR system has been empirically evaluated on a novel multi-writer dataset Imgur5K and the IAM dataset. The experimental results evaluate the performance of the HTR framework, further supported by an in-depth analysis of the error cases. Source code and pre-trained models are available at https://github.com/dmitrijsk/AttentionHTR.
|
['Ekta Vats', 'Dmitrijs Kass']
|
2022-01-23
| null | null | null | null |
['handwriting-recognition']
|
['computer-vision']
|
[ 5.04595935e-01 -2.64193505e-01 -2.83309788e-01 -5.29465199e-01
-7.12787569e-01 -4.26507890e-01 8.33256841e-01 -5.01507461e-01
-5.35064638e-01 4.52933699e-01 4.10877496e-01 -5.34333348e-01
2.22254142e-01 -3.61343086e-01 -6.76766872e-01 -5.87793827e-01
3.89143884e-01 4.56182033e-01 -8.93461630e-02 -1.75551802e-01
8.54353964e-01 5.06022871e-01 -9.67452765e-01 5.91086268e-01
5.12189746e-01 8.01764250e-01 5.40612996e-01 1.28300202e+00
8.02323222e-02 1.45071590e+00 -5.52398384e-01 -6.39697969e-01
9.49255973e-02 -5.00486910e-01 -8.60527635e-01 2.86003917e-01
3.56063813e-01 -7.37566590e-01 -9.29761946e-01 5.90374887e-01
9.21987593e-01 3.68038595e-01 6.13866568e-01 -5.77504754e-01
-1.21181774e+00 7.40580976e-01 -3.97551030e-01 3.31738859e-01
4.38338052e-03 4.47132200e-01 9.09818411e-01 -1.34597969e+00
6.63351893e-01 9.01326954e-01 4.62932408e-01 7.98730195e-01
-6.50542140e-01 -4.80179816e-01 -6.29165247e-02 6.31917059e-01
-9.48205769e-01 -6.01263940e-01 6.54204249e-01 -2.21909270e-01
1.64203370e+00 1.15771495e-01 3.11741740e-01 1.58066607e+00
3.65156621e-01 1.27286136e+00 9.81038213e-01 -7.01373518e-01
8.01750720e-02 -3.04790705e-01 5.18178642e-01 6.52977347e-01
-2.32118532e-01 1.27228022e-01 -7.94840455e-01 4.81542796e-01
9.01161313e-01 -7.63164908e-02 -1.31035447e-01 3.54145080e-01
-7.91724682e-01 7.01316297e-01 3.35602671e-01 2.74351776e-01
-3.53980392e-01 1.62160784e-01 7.93839395e-01 1.51146889e-01
3.49893659e-01 1.61270186e-01 -3.33002567e-01 -5.44105828e-01
-1.14777088e+00 -1.39053524e-01 5.79701006e-01 1.06845999e+00
3.50127965e-01 4.61190701e-01 -5.59606671e-01 1.24355006e+00
3.19144547e-01 4.08427805e-01 1.00358915e+00 -3.50772142e-01
7.89972663e-01 4.32979405e-01 -2.15894893e-01 -5.53278685e-01
6.18849099e-02 -4.32840198e-01 -7.85417616e-01 -5.48887514e-02
1.29969791e-01 -3.83167155e-02 -1.50466835e+00 9.33316588e-01
-3.85034978e-01 2.12040871e-01 1.50622979e-01 1.11648703e+00
7.17691123e-01 8.06432068e-01 -5.63876554e-02 2.28532553e-01
1.04777324e+00 -1.59168792e+00 -8.02813172e-01 -2.07565770e-01
5.15235126e-01 -8.66246343e-01 1.08930111e+00 3.11307430e-01
-1.18132985e+00 -6.62900329e-01 -1.12464201e+00 -6.11800849e-01
-2.04091296e-01 7.20948398e-01 -2.54866276e-02 3.92683357e-01
-8.76947224e-01 5.05655229e-01 -1.02260005e+00 -4.45241719e-01
5.95363677e-01 1.29711106e-01 -5.93181029e-02 -3.29757392e-01
-8.44269097e-01 1.19049466e+00 4.35033709e-01 5.53316355e-01
-1.39454365e+00 -3.83532435e-01 -5.71539342e-01 -1.15800342e-02
-4.91593080e-03 -1.47155777e-01 1.42559850e+00 -9.39469457e-01
-2.06356001e+00 7.29287446e-01 -1.44908398e-01 -6.76438868e-01
8.49258363e-01 -3.50764632e-01 -2.32479766e-01 1.32560775e-01
-4.97170061e-01 2.44777143e-01 9.88234758e-01 -9.96182024e-01
-4.69323844e-01 -2.88882792e-01 -4.79654431e-01 2.88187414e-01
-4.55077142e-01 2.56585926e-01 -5.44591308e-01 -9.82489586e-01
-4.71298516e-01 -6.97265327e-01 -1.09221496e-01 -7.63892174e-01
-6.25400543e-01 4.66633774e-02 9.30356741e-01 -1.36170959e+00
1.12338960e+00 -1.83830154e+00 1.48915455e-01 -1.10626500e-02
-2.79995263e-01 7.20846474e-01 -5.27541518e-01 7.59242833e-01
4.42417823e-02 -1.11342698e-01 -1.15390025e-01 -6.50187135e-01
-8.71632844e-02 9.07347947e-02 -7.90996134e-01 3.30427289e-01
4.10469502e-01 1.14235270e+00 -4.70732510e-01 -1.49789229e-01
4.03153270e-01 5.62121272e-01 -1.28731400e-01 5.12067258e-01
-2.53734559e-01 3.56776506e-01 -2.93496072e-01 6.98806047e-01
2.75717825e-01 -3.55413437e-01 8.07762891e-02 4.66454867e-03
-1.43802270e-01 3.10228527e-01 -4.36158687e-01 1.84360719e+00
-5.21296799e-01 9.74608064e-01 -6.39123738e-01 -7.79421568e-01
1.38248968e+00 3.64032030e-01 -8.19351822e-02 -9.72302139e-01
4.52670336e-01 3.26697975e-01 2.18293965e-02 -4.61041659e-01
1.10788023e+00 2.13789448e-01 2.55491406e-01 4.73579854e-01
4.13773924e-01 4.77479249e-01 1.09717503e-01 -9.43815485e-02
1.20477426e+00 4.45206016e-01 -1.97120141e-02 1.16821103e-01
4.90808994e-01 5.35933748e-02 1.65278852e-01 9.95147109e-01
-1.15197644e-01 8.75095963e-01 -1.71119496e-01 -5.00324905e-01
-1.67953849e+00 -6.06504381e-01 1.13423159e-02 1.23169839e+00
-1.98188841e-01 -1.49199933e-01 -6.30281150e-01 -5.73324084e-01
-3.15016836e-01 8.55295539e-01 -7.68037140e-01 -7.66854659e-02
-7.53785133e-01 -5.09693503e-01 9.96396720e-01 8.72158289e-01
8.05781424e-01 -1.38600898e+00 -5.59516549e-01 2.98434407e-01
-6.92608282e-02 -1.08180022e+00 -4.19454783e-01 2.80272424e-01
-8.30166757e-01 -6.13814771e-01 -1.20648670e+00 -7.31587231e-01
5.38919628e-01 -1.07862853e-01 6.13562763e-01 5.06172609e-03
-5.23293912e-01 2.50473469e-01 -7.53603518e-01 -2.18278587e-01
-5.40511131e-01 4.19155836e-01 -4.18079317e-01 5.86979873e-02
4.77190614e-01 -1.99875206e-01 -5.50473869e-01 2.97026575e-01
-6.89043820e-01 2.49644399e-01 9.78579462e-01 1.10402203e+00
3.34931254e-01 -7.09512830e-01 4.83639270e-01 -5.41089058e-01
5.65753162e-01 -1.67335108e-01 -5.67285120e-01 5.40527821e-01
-5.13636053e-01 -2.57891994e-02 7.26108432e-01 -4.95458663e-01
-1.33378291e+00 -7.51066953e-04 -2.68172801e-01 -6.30172968e-01
-5.80742732e-02 5.85604548e-01 1.62087634e-01 1.57744154e-01
4.72353846e-01 9.92166698e-01 -3.61698002e-01 -6.42412782e-01
3.93253565e-01 1.15843713e+00 6.60250902e-01 -3.73412013e-01
2.91096359e-01 6.19757511e-02 -4.46084470e-01 -9.99259353e-01
-5.24241805e-01 -2.71259159e-01 -7.07212210e-01 -3.17852557e-01
6.81824327e-01 -7.88668215e-01 -4.71633792e-01 9.92693722e-01
-1.16452920e+00 -1.03834045e+00 -4.05019633e-02 2.68733859e-01
-8.20172310e-01 2.97965318e-01 -1.25081432e+00 -8.10300946e-01
-7.40536451e-01 -1.04845393e+00 9.04790401e-01 1.37552708e-01
-5.48618734e-02 -1.02674317e+00 1.81424960e-01 6.49155259e-01
4.68320072e-01 -4.11261171e-01 6.84145808e-01 -1.01054132e+00
-6.64363086e-01 -2.71865427e-01 -1.99004844e-01 5.29842079e-01
-8.05975497e-02 1.41575515e-01 -1.15397716e+00 -5.17353833e-01
-3.15609246e-01 -7.03120410e-01 1.15952921e+00 2.89039731e-01
1.24368751e+00 -4.62769628e-01 1.48721576e-01 6.28138125e-01
1.58845615e+00 3.42139721e-01 1.11693025e+00 5.43295562e-01
9.12184417e-01 1.36304200e-01 4.79857594e-01 7.21464574e-01
1.26521796e-01 4.96496201e-01 1.58091467e-02 7.77841732e-02
-6.07490718e-01 -4.29010332e-01 5.46644628e-01 1.20650458e+00
-3.22226137e-01 -6.75867558e-01 -1.14008653e+00 5.12126505e-01
-1.80631912e+00 -9.46180522e-01 9.08224657e-02 1.81534207e+00
8.65676641e-01 -8.06361139e-02 -7.80731887e-02 3.95403169e-02
7.16997623e-01 2.61488587e-01 -6.51009917e-01 -7.52001584e-01
-1.23738386e-01 2.91880280e-01 6.44820452e-01 5.69626272e-01
-9.28601861e-01 1.39965653e+00 5.69346952e+00 1.03685653e+00
-1.46110833e+00 -2.54670102e-02 6.81486249e-01 -1.11490646e-02
9.79256779e-02 -2.05278769e-01 -1.18021786e+00 4.32399720e-01
1.30715668e+00 6.55683279e-02 4.45799798e-01 5.60803235e-01
3.23741406e-01 3.16380382e-01 -9.56341743e-01 8.32020462e-01
3.27886969e-01 -1.55050290e+00 3.96282494e-01 -4.43966826e-03
6.35824382e-01 4.19241399e-01 3.37280899e-01 2.73508132e-01
3.24242055e-01 -1.18138063e+00 8.27011049e-01 6.58101797e-01
9.45040286e-01 -5.91979682e-01 6.95911169e-01 1.67346612e-01
-7.68390834e-01 -2.36167625e-01 -5.06551087e-01 -6.21597376e-03
-1.46270143e-02 6.59400523e-02 -1.20484829e+00 6.18642628e-01
3.19054663e-01 1.17846167e+00 -6.94520891e-01 8.56341541e-01
-4.21869606e-01 1.09232724e+00 3.13416779e-01 -2.03447938e-01
3.99638444e-01 6.15104884e-02 3.46983761e-01 1.84283578e+00
2.41428688e-01 -2.07132790e-02 -1.96402684e-01 6.62504733e-01
-3.36006910e-01 -6.59147685e-04 -3.78509998e-01 -3.96513104e-01
2.75891960e-01 1.19475114e+00 -6.24102533e-01 -4.04668808e-01
-3.29796523e-01 1.42665255e+00 4.00951117e-01 6.52231812e-01
-7.69264102e-01 -5.41129231e-01 2.48844579e-01 -2.91799515e-01
7.22247958e-01 -3.00178230e-01 -5.29012561e-01 -1.37737060e+00
-9.13756639e-02 -1.13201880e+00 1.99651867e-01 -8.34161758e-01
-1.17665648e+00 7.55596817e-01 -6.88303828e-01 -9.09077704e-01
-3.90075266e-01 -8.68928969e-01 -5.11890948e-01 9.76779699e-01
-1.50480747e+00 -1.58752739e+00 -2.42476195e-01 4.71151173e-01
1.24883533e+00 -6.52113259e-01 7.61303842e-01 2.13532269e-01
-9.48317766e-01 9.03677881e-01 4.79638398e-01 6.72260344e-01
5.08938253e-01 -9.43242550e-01 8.97964478e-01 1.11116278e+00
1.01097986e-01 3.72607708e-01 3.96989137e-01 -7.80381799e-01
-1.56388092e+00 -1.27272689e+00 7.03324974e-01 -4.24411178e-01
6.79656804e-01 -4.08305287e-01 -1.02924335e+00 1.04084682e+00
5.88681221e-01 -1.96715936e-01 5.79125166e-01 -2.57295281e-01
-3.82572323e-01 2.56722033e-01 -5.92487514e-01 6.17534876e-01
7.33282328e-01 -6.79656923e-01 -6.50080562e-01 8.41440931e-02
1.02477372e-01 -4.81026053e-01 -9.16410208e-01 8.76333788e-02
7.20441759e-01 -5.91448486e-01 5.46858430e-01 -8.93038690e-01
1.20725060e+00 1.24875531e-01 -3.22975606e-01 -1.12031984e+00
-3.76935691e-01 -4.25491452e-01 -2.17304677e-01 1.21151769e+00
4.50500488e-01 -2.20489651e-01 8.09313118e-01 4.31263506e-01
-4.26245064e-01 -7.80103326e-01 -7.22631931e-01 -9.52470601e-01
1.33608729e-01 -3.34573179e-01 9.16043445e-02 8.31510663e-01
-7.93793201e-02 2.77618557e-01 -7.55123377e-01 -7.72205517e-02
4.68620509e-01 -1.92022771e-02 6.16768420e-01 -2.45980412e-01
-4.86273378e-01 -4.43476021e-01 -3.71675164e-01 -1.21237671e+00
2.29204029e-01 -1.01685905e+00 2.33021379e-01 -1.44857383e+00
4.42823559e-01 4.77343425e-02 -2.30363086e-01 7.80073941e-01
-4.68828231e-02 3.19969654e-01 4.50746596e-01 5.01912236e-01
-5.48427224e-01 8.48599017e-01 1.08517933e+00 -2.09521323e-01
-5.52389072e-04 -3.25433612e-01 6.77697212e-02 2.82069474e-01
1.02975273e+00 -9.46852118e-02 -1.52502492e-01 -9.14050579e-01
-3.38416725e-01 1.76236898e-01 2.04892084e-01 -8.27641547e-01
3.74379784e-01 1.98891178e-01 6.44109547e-01 -5.88836491e-01
3.77293915e-01 -4.42596734e-01 -3.92395109e-01 5.53995490e-01
-9.75923777e-01 4.68408614e-02 3.53049695e-01 2.93966949e-01
-1.46954700e-01 -4.13052410e-01 8.01135778e-01 -9.92597267e-03
-9.80531812e-01 1.95501193e-01 -6.34733379e-01 -2.05469280e-01
5.83053410e-01 -2.91592151e-01 -7.18442380e-01 -3.08977574e-01
-5.80358565e-01 -9.36805643e-03 2.90347189e-01 6.36387646e-01
1.09463620e+00 -1.01663697e+00 -1.07643771e+00 1.20865911e-01
7.97916874e-02 -6.09822512e-01 3.12629908e-01 5.10738373e-01
-6.99759901e-01 6.32317543e-01 -6.42948508e-01 -3.70634019e-01
-1.33124208e+00 3.29836935e-01 4.00294840e-01 -2.91757077e-01
-6.67946339e-01 8.39257717e-01 -2.82194406e-01 -1.52887926e-01
4.16184396e-01 -1.30565539e-01 2.26211939e-02 -3.43723983e-01
7.11550891e-01 5.14950991e-01 2.09095791e-01 -5.23218334e-01
-6.42544404e-02 3.50290835e-01 -7.71279156e-01 -3.98328364e-01
1.50143707e+00 -5.86280935e-02 2.13678628e-01 4.97061759e-01
1.15045452e+00 -6.93604469e-01 -1.57542384e+00 -3.40240479e-01
1.83380827e-01 -4.78171945e-01 -5.85286226e-03 -1.26298487e+00
-9.89376307e-01 1.24705255e+00 5.26214898e-01 -4.56927449e-01
1.12225044e+00 -5.25067985e-01 7.79891729e-01 6.10769689e-01
-3.21572237e-02 -1.21010232e+00 3.54707897e-01 1.08425856e+00
1.12717021e+00 -1.11283410e+00 -2.66337752e-01 3.23392481e-01
-8.98193359e-01 1.42931378e+00 6.15130842e-01 -3.60269874e-01
7.42655769e-02 3.40538919e-01 2.20755801e-01 2.66115189e-01
-1.03027225e+00 1.56059280e-01 2.21044123e-01 3.03610057e-01
6.02469862e-01 -3.00479621e-01 -1.61673009e-01 4.80529457e-01
-3.53474244e-02 3.56416017e-01 6.84164107e-01 1.03287697e+00
-1.16692640e-01 -1.09441555e+00 -6.11593388e-02 4.12880152e-01
-5.20968378e-01 -2.60175288e-01 -5.76428473e-01 4.07218158e-01
-6.83381975e-01 7.70273805e-01 1.26657942e-02 -5.82616210e-01
1.55595213e-01 2.49067828e-01 6.56775236e-01 -3.38631004e-01
-9.19518113e-01 -4.28559780e-02 1.44082941e-02 -2.12854311e-01
1.28402635e-02 -4.22569275e-01 -9.20902371e-01 -2.54348487e-01
-2.38374799e-01 -3.45303208e-01 7.52735555e-01 8.81124020e-01
4.51861531e-01 7.01386988e-01 5.67605555e-01 -8.49521756e-01
-9.40132618e-01 -1.35632002e+00 -3.37741166e-01 2.14744851e-01
3.60744953e-01 6.88813552e-02 2.33241156e-01 4.90110189e-01]
|
[11.892358779907227, 2.3911352157592773]
|
233ac4fb-ca27-47fd-a08d-5adaace75299
|
serf-interpretable-sleep-staging-using
|
2209.11174
| null |
https://arxiv.org/abs/2209.11174v2
|
https://arxiv.org/pdf/2209.11174v2.pdf
|
SERF: Interpretable Sleep Staging using Embeddings, Rules, and Features
|
The accuracy of recent deep learning based clinical decision support systems is promising. However, lack of model interpretability remains an obstacle to widespread adoption of artificial intelligence in healthcare. Using sleep as a case study, we propose a generalizable method to combine clinical interpretability with high accuracy derived from black-box deep learning. Clinician-determined sleep stages from polysomnogram (PSG) remain the gold standard for evaluating sleep quality. However, PSG manual annotation by experts is expensive and time-prohibitive. We propose SERF, interpretable Sleep staging using Embeddings, Rules, and Features to read PSG. SERF provides interpretation of classified sleep stages through meaningful features derived from the AASM Manual for the Scoring of Sleep and Associated Events. In SERF, the embeddings obtained from a hybrid of convolutional and recurrent neural networks are transposed to the interpretable feature space. These representative interpretable features are used to train simple models like a shallow decision tree for classification. Model results are validated on two publicly available datasets. SERF surpasses the current state-of-the-art for interpretable sleep staging by 2%. Using Gradient Boosted Trees as the classifier, SERF obtains 0.766 $\kappa$ and 0.870 AUC-ROC, within 2% of the current state-of-the-art black-box models.
|
['Cassie S. Mitchell', 'Irfan Al-Hussaini']
|
2022-09-21
| null | null | null | null |
['sleep-quality-prediction', 'sleep-staging']
|
['medical', 'medical']
|
[ 1.91452801e-02 4.75429803e-01 -3.72006625e-01 -8.35432172e-01
-6.18621171e-01 -2.27858365e-01 -1.78371504e-01 4.43606734e-01
-5.37179291e-01 8.06238592e-01 4.84021842e-01 -7.16111839e-01
-3.49452466e-01 -2.34268129e-01 1.16459532e-02 -4.42779660e-01
1.09401673e-01 6.66213810e-01 -2.74178684e-01 5.55116311e-02
-3.60246330e-01 1.72256067e-01 -1.23268378e+00 5.49564779e-01
1.08064270e+00 1.48615706e+00 -1.58936217e-01 6.81649566e-01
7.25838989e-02 7.35893667e-01 -6.95685446e-01 -3.51359814e-01
-8.81649107e-02 -6.75590813e-01 -8.00205648e-01 -2.40728587e-01
-8.88619293e-03 -2.91237116e-01 1.60874836e-02 5.80549836e-01
4.27891046e-01 -2.30573311e-01 5.07291257e-01 -1.01057303e+00
-9.18985605e-01 3.73344183e-01 4.63271260e-01 7.84073174e-01
2.36823648e-01 3.59407753e-01 1.06825173e+00 -6.12023115e-01
2.79808585e-02 6.27139628e-01 1.16212606e+00 9.76480246e-01
-1.17921472e+00 -3.79262537e-01 -4.02453452e-01 4.18594599e-01
-1.08406937e+00 -4.57690686e-01 2.94848736e-02 -2.44146287e-01
1.35628152e+00 7.64188230e-01 1.34705639e+00 1.03234875e+00
7.98340797e-01 1.73541725e-01 1.10230041e+00 -2.42178679e-01
5.76504886e-01 2.14686558e-01 6.61008477e-01 9.30845320e-01
6.52656674e-01 -7.12052686e-03 -5.73885322e-01 -1.66271642e-01
2.50444889e-01 7.25639164e-01 -1.12849975e-03 4.18723732e-01
-9.13164854e-01 8.47947776e-01 7.94440150e-01 1.65732995e-01
-2.68154562e-01 9.73078758e-02 2.93183357e-01 1.77821577e-01
6.29355788e-01 8.76555681e-01 -4.79251832e-01 -4.21711683e-01
-1.32552230e+00 -3.40140313e-01 8.23922276e-01 2.55610228e-01
3.34129244e-01 -1.85896754e-01 -3.84676993e-01 5.87007284e-01
5.64051151e-01 4.74899799e-01 8.45555544e-01 -8.79976153e-01
-1.00985110e-01 1.25518274e+00 2.61954814e-02 -5.48170984e-01
-1.10475433e+00 -6.83261514e-01 -8.88109803e-01 -2.25524530e-02
6.13309741e-02 4.10539992e-02 -8.69086385e-01 1.10710895e+00
-1.71024859e-01 -1.93614021e-01 1.02489153e-02 7.84439504e-01
1.15192223e+00 2.05584407e-01 1.14121042e-01 -9.20056179e-02
1.77717113e+00 -1.10617375e+00 -9.06048596e-01 -6.00915134e-01
6.05161667e-01 -1.42431721e-01 1.15659797e+00 4.67175066e-01
-1.08675098e+00 -4.24886525e-01 -1.18726897e+00 -5.03838718e-01
-3.79941285e-01 4.04319286e-01 6.39427245e-01 7.04338491e-01
-1.37835991e+00 7.27988958e-01 -1.63972163e+00 -4.71807539e-01
9.29862797e-01 8.99345458e-01 -2.55125284e-01 2.95213610e-01
-1.01852882e+00 1.24172020e+00 -3.71765136e-03 3.56805086e-01
-7.91948497e-01 -6.66839063e-01 -7.26459980e-01 3.52906168e-01
-3.14644217e-01 -1.31741631e+00 1.24993944e+00 -7.26957858e-01
-1.35260463e+00 9.39969838e-01 -4.83393103e-01 -7.63509393e-01
1.79784894e-01 -4.72272456e-01 -6.93831265e-01 1.93812832e-01
1.83188185e-01 3.77842247e-01 4.78868634e-01 -3.79901737e-01
-3.38179529e-01 -5.10238707e-01 4.98340931e-03 -1.05316818e-01
-5.50687671e-01 -1.05738945e-01 -8.66060704e-02 -5.29535934e-02
-8.29712823e-02 -1.00902760e+00 -4.20007944e-01 4.15862978e-01
-4.57235038e-01 -3.69165361e-01 2.40875244e-01 -6.55788720e-01
1.80679667e+00 -2.11731529e+00 -2.47053504e-01 -2.52126455e-01
1.10127091e+00 3.25829148e-01 4.35198456e-01 -3.42440307e-02
-4.89139631e-02 3.30341399e-01 -2.24185914e-01 -8.47160220e-01
-5.73383868e-02 6.47511840e-01 3.49608213e-01 4.38057363e-01
5.31799734e-01 1.09916568e+00 -1.01252806e+00 -3.22086632e-01
4.11012083e-01 2.92526573e-01 -7.77899921e-01 3.82692009e-01
4.06186521e-01 3.11378747e-01 -3.17105234e-01 5.38187921e-01
4.92531061e-02 -9.86151040e-01 7.34722316e-02 -1.00980215e-01
1.33819684e-01 8.16631615e-01 -2.55578727e-01 1.73910403e+00
-6.68226838e-01 6.64294183e-01 -6.01011992e-01 -6.21485353e-01
6.99366689e-01 8.97303596e-02 1.64528146e-01 -3.65285605e-01
3.19247127e-01 2.54189909e-01 3.52010787e-01 -7.21856117e-01
-7.52004310e-02 -5.24716675e-01 -1.52804673e-01 3.00954223e-01
1.96488455e-01 1.07941017e-01 -2.03155726e-01 -7.51222372e-02
1.79734552e+00 -4.17013675e-01 8.92344177e-01 -4.45643097e-01
3.19310695e-01 -1.10678017e-01 7.68745303e-01 6.93660915e-01
-4.92141664e-01 6.55131698e-01 4.19184268e-01 -9.52952445e-01
-7.40750194e-01 -1.35651326e+00 -5.15072227e-01 6.98556364e-01
-2.12403893e-01 -8.90622079e-01 -7.05496132e-01 -8.71116042e-01
-4.98909913e-02 6.48657441e-01 -1.18412292e+00 -5.87808311e-01
-2.60388330e-02 -8.97248328e-01 6.30585730e-01 9.37985420e-01
1.39333636e-01 -1.02367818e+00 -9.84432280e-01 2.08606407e-01
-1.03518769e-01 -9.86284494e-01 -1.99364215e-01 6.89621270e-01
-1.08748293e+00 -1.13660669e+00 -2.68138260e-01 -3.21692050e-01
7.02424049e-01 -2.99028099e-01 1.36798847e+00 4.40968812e-01
-4.85684246e-01 -7.53549486e-02 -2.35660121e-01 -6.44471109e-01
-5.08510947e-01 2.63519794e-01 4.50669348e-01 -1.30485699e-01
8.86070311e-01 -5.34784079e-01 -1.35865521e+00 1.95560023e-01
-5.71091533e-01 1.31465256e-01 7.38742769e-01 1.01621127e+00
5.49589038e-01 -6.38190031e-01 5.56269109e-01 -7.87703097e-01
5.56408226e-01 -6.43508554e-01 1.11219801e-01 -7.40946531e-02
-1.25919688e+00 1.50368646e-01 9.52043653e-01 1.09854199e-01
-4.43353951e-01 -2.34129861e-01 -4.13928926e-01 -3.43771547e-01
-1.79685235e-01 2.53397614e-01 3.67918491e-01 4.54098076e-01
9.70654726e-01 1.16993494e-01 1.52954921e-01 -5.05713403e-01
-1.28887638e-01 1.17294312e+00 4.18668151e-01 1.78536430e-01
2.63619721e-01 5.40699363e-01 -9.66207236e-02 -5.11380672e-01
-1.56304002e+00 -5.70465446e-01 -5.32920659e-01 2.37105712e-01
1.38261521e+00 -8.10874343e-01 -7.51551390e-01 -2.07790345e-01
-8.31236839e-01 -3.09366405e-01 -5.96770167e-01 3.31919909e-01
-3.27156186e-01 2.63475422e-02 -4.26079392e-01 -7.36933589e-01
-1.10362613e+00 -1.17170882e+00 1.28636193e+00 2.57094055e-01
-1.15949059e+00 -1.26313591e+00 1.15822859e-01 7.39099801e-01
5.25522888e-01 1.33644328e-01 8.52560699e-01 -1.08590555e+00
-4.38640639e-02 -1.66716784e-01 -1.20532446e-01 6.97298288e-01
5.69862843e-01 -3.45684975e-01 -1.27524328e+00 7.13024335e-03
1.97525218e-01 -1.22338869e-01 7.62383282e-01 4.11976188e-01
1.36619568e+00 -5.36048591e-01 -3.52186352e-01 9.52859759e-01
1.15254712e+00 1.83727443e-01 5.12785196e-01 4.07514840e-01
3.53962183e-01 -2.25762814e-01 -5.71322069e-02 4.84971195e-01
5.92318594e-01 2.98948973e-01 3.19969326e-01 -2.62796283e-01
-9.62712765e-02 1.04278944e-01 4.63017613e-01 9.95930493e-01
-8.02726224e-02 1.05625562e-01 -1.00422382e+00 3.68862063e-01
-1.67260408e+00 -5.13678849e-01 5.31215928e-02 1.77292991e+00
8.73376012e-01 3.88217598e-01 2.61940286e-02 2.86875308e-01
4.33416888e-02 -1.81329042e-01 -7.39813387e-01 -9.04987454e-01
2.51495451e-01 7.97075093e-01 1.09018825e-01 1.60267219e-01
-8.46387267e-01 3.68174553e-01 6.34308290e+00 1.92544773e-01
-1.07343996e+00 5.96356630e-01 8.67361963e-01 -5.80823421e-01
-5.71399927e-02 -4.04615641e-01 -6.60102010e-01 8.51084888e-01
1.81393421e+00 8.99404660e-02 4.93671536e-01 8.79221082e-01
8.34107220e-01 1.85856037e-02 -1.46039820e+00 1.20108712e+00
1.44201383e-01 -1.57538974e+00 -3.63781482e-01 -5.77581376e-02
3.41310024e-01 3.52331161e-01 7.09494427e-02 3.44624519e-01
-3.18166502e-02 -1.45920563e+00 3.48587990e-01 6.02458119e-01
1.24667633e+00 -8.75832736e-02 1.24859202e+00 -1.71157941e-01
-7.35792339e-01 -4.31537479e-01 -1.55640200e-01 -4.30645466e-01
-1.69056237e-01 5.27233958e-01 -1.24372602e+00 1.84346467e-01
1.03192925e+00 1.11150551e+00 -9.86186326e-01 9.01590049e-01
-3.20918441e-01 9.68626142e-01 -4.03914869e-01 -3.75223011e-01
2.02587590e-01 1.44256487e-01 3.51977125e-02 1.15858984e+00
3.11658800e-01 9.46330279e-02 -2.99606532e-01 1.15571904e+00
-4.66540679e-02 -2.38078609e-01 -3.70554090e-01 -9.49613675e-02
1.20323874e-01 1.49509120e+00 -7.34974682e-01 -4.88461524e-01
-2.75504231e-01 1.08823240e+00 3.00158858e-01 -1.75652936e-01
-9.26390350e-01 -1.00521959e-01 7.72734165e-01 3.21905017e-01
-1.03957370e-01 2.47901395e-01 -9.72545624e-01 -1.24017048e+00
-2.03287929e-01 -6.85723126e-01 4.43985194e-01 -7.32028902e-01
-1.37064481e+00 9.67893183e-01 -4.35461968e-01 -1.32808185e+00
3.23750940e-03 -6.85061872e-01 -7.96024084e-01 6.11060917e-01
-1.26407111e+00 -8.82168710e-01 -7.77841032e-01 1.99057072e-01
6.12743199e-01 -1.36085734e-01 1.24375689e+00 2.01313123e-01
-8.99147809e-01 7.13594139e-01 1.01443194e-01 1.00671925e-01
4.21359420e-01 -1.59109330e+00 4.31209177e-01 2.92193741e-01
-4.91969287e-02 9.78128135e-01 3.87990236e-01 -1.44762069e-01
-1.01208651e+00 -1.24496162e+00 1.15213585e+00 -1.11913574e+00
5.97473323e-01 -6.52492866e-02 -5.12932241e-01 6.52897179e-01
-3.37325111e-02 2.50286490e-01 1.59686303e+00 3.23372424e-01
2.58519441e-01 -5.73491335e-01 -1.09462214e+00 2.45807275e-01
8.98267627e-01 -5.36238015e-01 -1.06591225e+00 2.31918737e-01
4.80698466e-01 -2.19586149e-01 -9.44221973e-01 1.51032194e-01
7.16009438e-01 -1.05204213e+00 5.92231989e-01 -9.54761207e-01
6.90235734e-01 -1.44157231e-01 1.77039087e-01 -1.26216090e+00
-3.84466559e-01 -6.19443774e-01 -1.90547377e-01 3.92042011e-01
5.61769247e-01 -7.38497734e-01 8.49396169e-01 8.61012876e-01
-7.48587191e-01 -1.38775074e+00 -1.13925409e+00 -3.88032854e-01
-2.34549120e-01 -4.26486522e-01 5.54459870e-01 5.03654480e-01
3.55946809e-01 5.70688009e-01 7.96970800e-02 -2.26636250e-02
2.06252664e-01 -2.30251431e-01 1.76485240e-01 -1.41581273e+00
-2.80338287e-01 -2.27203444e-01 -7.20441937e-01 -4.53886390e-01
-9.93959904e-02 -1.26669073e+00 -1.49782524e-01 -2.10439134e+00
4.36401159e-01 -3.76645595e-01 -1.00815427e+00 9.65816736e-01
-3.69231284e-01 5.88712275e-01 1.12149417e-01 1.30821511e-01
-8.12695205e-01 3.99568737e-01 8.98349404e-01 -1.40873730e-01
-2.64192551e-01 1.99755922e-01 -1.09386075e+00 1.04507279e+00
9.18320417e-01 -5.96735179e-01 -4.28395241e-01 -3.28881443e-01
3.45637769e-01 -7.47770816e-03 3.75122815e-01 -1.38738048e+00
-6.52945414e-02 2.14348167e-01 6.47636652e-01 -3.54731977e-02
5.84032297e-01 -8.00713837e-01 1.08215719e-01 8.56798172e-01
-3.27925235e-01 2.78428942e-01 2.95427710e-01 5.12151957e-01
4.78087813e-02 -1.24285882e-02 6.26067698e-01 -6.10441389e-03
-2.51987904e-01 -8.00169725e-03 -5.33646822e-01 1.76263526e-01
5.43367684e-01 -3.58512700e-01 -3.20343196e-01 -2.07011417e-01
-1.28471661e+00 -1.46445902e-02 1.45357713e-01 1.70016944e-01
8.16687763e-01 -9.45493579e-01 -2.67551661e-01 4.45427805e-01
2.02547938e-01 -4.23061587e-02 1.25009075e-01 1.17632484e+00
-8.54386151e-01 6.32576108e-01 -2.15868175e-01 -8.15476477e-01
-1.18559885e+00 1.13801658e-01 6.19637609e-01 -4.53655273e-01
-5.68366587e-01 6.76128864e-01 3.88523489e-02 -5.66746779e-02
-1.49617344e-01 -1.22889400e+00 -7.19848722e-02 -9.68474820e-02
6.15733504e-01 3.14594179e-01 5.01849711e-01 7.24566281e-02
-8.46842587e-01 1.25929654e-01 1.31160960e-01 6.09793127e-01
1.43374443e+00 1.54057855e-03 -1.03496678e-01 5.74109137e-01
1.12436545e+00 -3.32085729e-01 -6.88369572e-01 4.02954429e-01
-4.48360592e-02 1.20281205e-01 1.65121481e-01 -1.30284011e+00
-7.71771073e-01 1.17567408e+00 1.09839427e+00 2.96043754e-01
1.20546257e+00 6.79429471e-02 9.76656854e-01 5.04875422e-01
-4.03765291e-02 -5.80785453e-01 -4.20036018e-02 1.34305552e-01
5.07682383e-01 -1.34630060e+00 1.26783207e-01 1.34011149e-01
-6.23109698e-01 1.31355262e+00 4.37091947e-01 -7.70511329e-02
8.47290754e-01 5.64001836e-02 2.63531983e-01 -3.18483651e-01
-8.64279449e-01 4.82368320e-02 5.53673744e-01 3.37043881e-01
3.39703143e-01 3.27012330e-01 -1.71760008e-01 1.37185168e+00
-6.25086606e-01 6.80746078e-01 3.73652399e-01 4.45329547e-01
-3.15924197e-01 -7.88155496e-01 1.05165713e-01 1.21299028e+00
-8.50968361e-01 -4.76722419e-01 -1.79671630e-01 4.22448546e-01
4.47572649e-01 1.12064803e+00 3.07244867e-01 -5.06000340e-01
1.80019259e-01 3.25861663e-01 -1.54630188e-03 -1.12683189e+00
-9.84759629e-01 -4.14635599e-01 3.07480603e-01 -6.87938333e-01
-2.69599140e-01 -2.44679257e-01 -1.32986152e+00 -5.90709150e-02
-7.51930242e-03 2.23083720e-01 1.58717647e-01 1.11723793e+00
8.28319430e-01 9.48626816e-01 2.25978538e-01 -4.92816597e-01
-4.23927158e-01 -9.57792401e-01 -4.21284109e-01 4.12451714e-01
9.19460893e-01 -5.65402687e-01 -5.87209165e-01 4.53240238e-02]
|
[13.504387855529785, 3.532036542892456]
|
fc99c478-b4be-4c74-92c2-b0508b6f5012
|
stateless-and-rule-based-verification-for
|
2204.07430
| null |
https://arxiv.org/abs/2204.07430v2
|
https://arxiv.org/pdf/2204.07430v2.pdf
|
Stateless and Rule-Based Verification For Compliance Checking Applications
|
Underlying computational model has an important role in any computation. The state and transition (such as in automata) and rule and value (such as in Lisp and logic programming) are two comparable and counterpart computational models. Both of deductive and model checking verification techniques are relying on a notion of state and as a result, their underlying computational models are state dependent. Some verification problems (such as compliance checking by which an under compliance system is verified against some regulations and rules) have not a strong notion of state nor transition. Behalf of it, these systems have a strong notion of value symbols and declarative rules defined on them. SARV (Stateless And Rule-Based Verification) is a verification framework that designed to simplify the overall process of verification for stateless and rule-based verification problems (e.g. compliance checking). In this paper, a formal logic-based framework for creating intelligent compliance checking systems is presented. We define and introduce this framework, report a case study and present results of an experiment on it. The case study is about protocol compliance checking for smart cities. Using this solution, a Rescue Scenario use case and its compliance checking are sketched and modeled. An automation engine for and a compliance solution with SARV are introduced. Based on 300 data experiments, the SARV-based compliance solution outperforms famous machine learning methods on a 3125-records software quality dataset.
|
['Ehsaneddin Asgari', 'Mohammad Izadi', 'Mohammad Reza Besharati']
|
2022-04-14
| null | null | null | null |
['formal-logic']
|
['reasoning']
|
[ 7.90047422e-02 2.75385708e-01 -4.59628403e-01 -4.17232811e-01
-1.54258147e-01 -4.66032624e-01 9.75181818e-01 3.23870838e-01
7.86520392e-02 5.19525051e-01 -6.69426993e-02 -9.71476555e-01
-5.43573558e-01 -1.25290704e+00 -3.31774563e-01 3.48936729e-02
-7.78181255e-02 6.60200000e-01 6.09544218e-01 -6.01894438e-01
1.60754770e-01 2.70944357e-01 -1.98336554e+00 3.10885459e-01
6.25301659e-01 9.52985466e-01 -3.47889245e-01 4.38238412e-01
-3.29347998e-01 1.34664464e+00 -3.23106110e-01 1.76761094e-02
7.64162987e-02 -2.57857174e-01 -1.20010722e+00 6.26865849e-02
-6.35345504e-02 -4.40510362e-02 3.14368121e-02 1.26766574e+00
-3.59354556e-01 -3.84712398e-01 5.11063278e-01 -2.20877194e+00
-3.16978931e-01 9.79846418e-01 3.05152684e-01 -5.47521949e-01
9.17542279e-01 3.84328425e-01 8.94710422e-01 9.44985971e-02
6.33644640e-01 1.01435614e+00 6.30254447e-01 4.48182791e-01
-1.22181547e+00 -4.67818111e-01 -2.51325015e-02 4.63411599e-01
-1.39948463e+00 -2.08075270e-01 5.89138687e-01 -7.59122968e-01
1.27071965e+00 7.64410377e-01 8.77933562e-01 3.91221762e-01
3.45324337e-01 2.13768467e-01 1.47631598e+00 -8.76173317e-01
5.10947406e-01 5.37026882e-01 9.39771593e-01 6.76790476e-01
7.32257426e-01 4.44935739e-01 2.68184215e-01 -3.03533942e-01
1.60190821e-01 -9.00334939e-02 5.57106994e-02 -2.65283316e-01
-6.08713984e-01 6.10170901e-01 -3.48305702e-01 7.03314185e-01
1.43393293e-01 4.91059646e-02 6.70237303e-01 6.52608991e-01
-5.75473011e-01 1.01125270e-01 -7.63938367e-01 -6.00013882e-02
-9.07114863e-01 3.01096290e-01 1.22548497e+00 1.24990523e+00
6.85748756e-01 1.17829733e-01 -1.93556920e-01 -3.71939182e-01
8.34921479e-01 7.67477214e-01 3.53039831e-01 -7.88573265e-01
-6.82417154e-02 1.61178684e+00 4.69878107e-01 -8.63336563e-01
-4.48229879e-01 2.25384444e-01 -6.38630390e-01 6.98402703e-01
1.91446483e-01 4.04367268e-01 -6.90714836e-01 1.49611616e+00
2.37676427e-01 2.68473476e-01 3.70754153e-01 3.99375975e-01
6.73184395e-01 4.41664845e-01 1.14388578e-01 -7.47937024e-01
1.53999531e+00 -3.74646068e-01 -1.07288170e+00 4.13660586e-01
1.03712392e+00 -1.03744520e-02 8.78723681e-01 7.04917371e-01
-8.96563649e-01 -1.99727058e-01 -1.02746773e+00 4.75086540e-01
-8.18184137e-01 -1.07159205e-01 6.36825800e-01 9.20310020e-01
-1.07192624e+00 2.12390333e-01 -6.28036797e-01 -2.48226956e-01
-1.16250478e-01 5.41157663e-01 -4.20694798e-01 2.85006136e-01
-1.44046724e+00 1.08866644e+00 6.05546176e-01 -3.27172667e-01
-7.93045342e-01 -2.88725644e-01 -1.11084688e+00 6.24626428e-02
4.43113983e-01 -2.39552855e-01 1.42267013e+00 -7.32474327e-01
-1.43206596e+00 8.98571968e-01 1.08940914e-01 -8.88290823e-01
5.13837934e-01 4.53051656e-01 -1.30318415e+00 -4.35581028e-01
-8.57856404e-03 -3.68711829e-01 2.23145038e-01 -1.14392853e+00
-6.52392924e-01 -1.89809501e-01 3.18652540e-01 -1.20551240e+00
2.81574905e-01 5.15170753e-01 1.39597058e-01 3.68029118e-01
4.14281040e-02 -7.02749074e-01 -2.46341318e-01 -2.22528428e-01
-3.74185443e-02 -3.68520081e-01 6.20551765e-01 -3.07895571e-01
1.88363600e+00 -1.52497053e+00 -7.02515483e-01 9.08173561e-01
-1.18040547e-01 5.92447102e-01 1.47537991e-01 6.72926426e-01
-2.83255607e-01 4.21774030e-01 -2.36552700e-01 4.81949955e-01
6.30208492e-01 4.34829205e-01 -2.39833564e-01 4.47795004e-01
5.53739630e-02 7.52058744e-01 -6.76939309e-01 -1.02158976e+00
5.66015303e-01 3.33567150e-02 -5.42129517e-01 1.08398870e-02
-8.11373353e-01 1.23359598e-01 -4.15305078e-01 6.90946877e-01
7.69601822e-01 -1.44952312e-01 8.17547381e-01 -2.72990521e-02
-5.68965793e-01 1.84011608e-01 -1.80851948e+00 1.14974737e+00
-6.16929710e-01 -1.94476172e-01 -1.85178816e-01 -8.40011060e-01
1.22189367e+00 7.93642342e-01 4.04330403e-01 -7.95269787e-01
6.11917317e-01 1.42155215e-01 -5.27001694e-02 -9.08669651e-01
-2.70782784e-02 6.06364608e-02 -5.12057126e-01 9.54390585e-01
-4.68936533e-01 -4.79643583e-01 4.35987324e-01 -4.66193538e-03
1.22426164e+00 3.41176212e-01 1.11351645e+00 -2.58737415e-01
1.31790137e+00 3.68615627e-01 8.18909824e-01 5.02128065e-01
-4.03551906e-01 -3.53582591e-01 6.84879065e-01 -7.37861753e-01
-9.43016291e-01 -4.36015636e-01 8.30659866e-02 1.65987149e-01
-9.00022611e-02 -1.07941699e+00 -5.84927380e-01 -5.57629049e-01
-8.46944675e-02 9.51166630e-01 -2.64080137e-01 -3.00632894e-01
-3.96143377e-01 1.73085809e-01 8.95256698e-01 3.25164735e-01
4.52710956e-01 -1.00439167e+00 -8.81230175e-01 1.99041769e-01
-4.11987491e-02 -1.12713730e+00 3.55994016e-01 4.14799564e-02
-8.01650107e-01 -1.65053332e+00 1.17367744e+00 -5.70722580e-01
5.01628458e-01 -4.86227572e-01 1.34348536e+00 8.02805781e-01
1.80552408e-01 1.91020727e-01 -5.20189047e-01 -5.39165497e-01
-1.13755488e+00 -5.62441528e-01 2.24125728e-01 -2.18040168e-01
6.00372553e-01 -4.40075397e-01 3.36986363e-01 3.24863225e-01
-1.19541383e+00 -2.11971477e-01 1.37113214e-01 3.98199022e-01
5.16524851e-01 6.08116388e-01 1.80779576e-01 -8.89473021e-01
7.22056687e-01 -2.92965882e-02 -1.55056322e+00 9.09068823e-01
-1.40258694e+00 4.55779791e-01 6.85676873e-01 1.66375078e-02
-6.75195277e-01 3.36804807e-01 9.58709270e-02 -3.39057036e-02
-5.76847076e-01 5.08088648e-01 -6.66581154e-01 1.93910733e-01
6.04255438e-01 3.12723905e-01 3.63285057e-02 -1.62120104e-01
3.77472304e-03 7.96041191e-01 3.25392485e-01 -9.24281776e-01
1.01500916e+00 3.17539752e-01 5.06871462e-01 -1.01792417e-01
-1.67618096e-01 -5.96131943e-02 -5.79727769e-01 -3.99148703e-01
6.24699175e-01 -3.78631502e-01 -1.52374458e+00 3.03756714e-01
-1.15860808e+00 -3.37949306e-01 -6.12996757e-01 1.08035684e-01
-8.00707102e-01 2.44805694e-01 -2.77531058e-01 -1.62968731e+00
-4.43923980e-01 -1.10942948e+00 5.93824029e-01 -1.41506761e-01
-5.46065271e-01 -7.68212557e-01 4.82947856e-01 1.51473969e-01
3.05001110e-01 4.24327821e-01 1.09557950e+00 -6.79515898e-01
-2.66862363e-01 -6.22939050e-01 1.92171395e-01 4.78828222e-01
1.81491718e-01 5.60384810e-01 -6.40913010e-01 1.35999456e-01
-1.81901697e-02 3.20385396e-02 -3.33586276e-01 -6.33369833e-02
7.44815528e-01 -7.45803833e-01 -2.32521281e-01 -1.30277351e-01
1.67602742e+00 4.16665643e-01 9.28596735e-01 5.36170423e-01
-3.18654805e-01 2.53201962e-01 9.84226346e-01 5.07453501e-01
3.59252393e-01 9.15991187e-01 4.83866662e-01 5.17711520e-01
9.54149589e-02 -2.57449061e-01 4.51234370e-01 4.09181118e-01
-3.10227126e-01 4.00050342e-01 -1.37201250e+00 2.19785556e-01
-1.96690869e+00 -1.54071939e+00 -8.83517981e-01 2.15319014e+00
1.08458424e+00 2.86922783e-01 2.07281604e-01 1.30399871e+00
5.43097019e-01 -4.04150039e-01 3.07454675e-01 -8.93483818e-01
5.55749470e-03 3.33534569e-01 1.38105139e-01 8.37471664e-01
-8.22106302e-01 7.73425639e-01 5.89606190e+00 3.25342536e-01
-9.05791938e-01 3.84183139e-01 -5.31001389e-01 7.04987466e-01
-3.30551952e-01 5.73758662e-01 -7.18915105e-01 3.13992649e-01
1.28250766e+00 -1.28649920e-01 3.16064358e-01 9.76988316e-01
6.77545846e-01 -2.24611118e-01 -1.17027736e+00 7.43442893e-01
-3.18822801e-01 -1.36797643e+00 -1.09773763e-01 -2.15362385e-01
6.00642860e-01 -5.25416672e-01 -6.12579882e-01 2.81071275e-01
6.17187738e-01 -1.03736818e+00 1.30947936e+00 7.89392233e-01
7.99297035e-01 -5.16200781e-01 1.04008353e+00 4.36971784e-01
-1.24290800e+00 -4.56157148e-01 2.05642566e-01 -4.80128586e-01
1.13046005e-01 4.82004762e-01 -5.95432699e-01 9.72044170e-01
4.76223320e-01 3.86464208e-01 -4.54049230e-01 9.93882298e-01
-6.30555511e-01 5.39687514e-01 -2.03833982e-01 -1.35582790e-01
-1.94651768e-01 -3.67675245e-01 2.51515657e-01 1.12182713e+00
9.07777697e-02 2.74479508e-01 3.53130065e-02 1.03673792e+00
7.65715063e-01 -3.01664114e-01 -7.81721950e-01 2.38965824e-01
4.11614716e-01 9.99278724e-01 -2.61579961e-01 -6.25669360e-01
-5.69261491e-01 -1.91479802e-01 -2.83971786e-01 -1.03215883e-02
-1.13743162e+00 -4.89580445e-02 4.14228767e-01 5.56618810e-01
-3.60542350e-02 -7.38908425e-02 -4.11214501e-01 -9.05153394e-01
-7.70546049e-02 -1.13617361e+00 4.02696282e-01 -8.49769592e-01
-6.02923691e-01 5.40968955e-01 3.56953055e-01 -1.61873472e+00
-2.25741223e-01 -4.84465957e-01 -6.24640763e-01 5.64662516e-01
-1.49172056e+00 -1.57610917e+00 -4.20239002e-01 1.05212855e+00
-4.08129662e-01 -8.88305232e-02 1.17065644e+00 3.51394653e-01
-2.90736943e-01 2.49028340e-01 -8.68586183e-01 4.79091592e-02
3.81643698e-02 -8.28860402e-01 -3.54353487e-01 1.05415916e+00
-7.01932758e-02 5.57854950e-01 1.05425894e+00 -8.41843605e-01
-1.74984217e+00 -1.19160819e+00 1.31157184e+00 -6.37000322e-01
7.25127101e-01 2.41774563e-02 -5.11020660e-01 1.03391171e+00
-8.00052062e-02 1.23796791e-01 3.98590863e-01 -2.64603227e-01
-6.13743126e-01 -4.37839448e-01 -1.52135444e+00 3.35842937e-01
9.91532266e-01 -5.89659750e-01 -1.03050840e+00 3.02628100e-01
4.99401033e-01 1.55372560e-01 -1.05151641e+00 4.23678398e-01
4.98845726e-01 -1.13459730e+00 3.29178244e-01 -1.11424601e+00
-2.36476570e-01 -1.26253831e+00 -3.45280081e-01 -3.75773579e-01
-3.36006284e-01 -6.27462089e-01 -5.00178397e-01 1.37654018e+00
2.99444050e-01 -8.27243030e-01 3.68858099e-01 1.02773523e+00
-9.61517915e-02 -2.90692240e-01 -8.16721320e-01 -1.32340324e+00
-2.19791055e-01 -9.91955340e-01 1.40563524e+00 8.93083453e-01
6.07534707e-01 -3.71588096e-02 1.49635756e-02 3.71147722e-01
2.63396084e-01 3.96843255e-01 8.42921376e-01 -1.73989201e+00
-5.80942556e-02 -4.50893760e-01 -8.54907751e-01 2.81964064e-01
4.22010809e-01 -8.88660967e-01 -1.79698452e-01 -1.68842638e+00
-1.97824657e-01 -4.50143397e-01 -6.76850155e-02 1.16504848e+00
1.06989622e+00 -3.74302953e-01 -7.61868432e-02 -8.03572405e-03
-5.77580750e-01 -4.48744036e-02 8.36364329e-01 -6.98415220e-01
-1.61836237e-01 1.42918855e-01 -2.47938156e-01 4.34338272e-01
8.77901137e-01 -5.07997036e-01 -4.15290147e-01 3.87600064e-01
7.89265513e-01 2.89961576e-01 3.60398978e-01 -1.36117637e+00
3.60631019e-01 -8.92863035e-01 -8.33612025e-01 -2.99682021e-01
-2.63221145e-01 -1.74186778e+00 1.05341887e+00 1.29030085e+00
-1.40895620e-02 8.61057118e-02 -2.30694667e-01 -1.08944692e-01
-3.62344801e-01 -3.88937950e-01 8.81245315e-01 1.98099360e-01
-7.71372557e-01 -7.01314956e-02 -5.68626821e-01 -3.99785101e-01
1.60269058e+00 -3.24582458e-01 -5.13535500e-01 -2.69741982e-01
-6.99424326e-01 1.04475901e-01 6.52625263e-01 1.26877263e-01
4.98025835e-01 -1.16810787e+00 -3.44204754e-01 6.14097059e-01
3.72451723e-01 -3.44160020e-01 -2.36574113e-01 1.09983611e+00
-4.28042799e-01 6.79714024e-01 -2.87747115e-01 -3.74550670e-01
-1.44839001e+00 1.00262058e+00 6.39234602e-01 -5.26343048e-01
-3.02224904e-01 -3.74048978e-01 -9.15288866e-01 -6.19103253e-01
8.03512856e-02 -1.07602966e+00 -6.37567580e-01 -3.66424710e-01
7.24247038e-01 3.63906056e-01 3.09818745e-01 -6.11580670e-01
-8.12262535e-01 7.58616745e-01 8.35516930e-01 -5.77251725e-02
1.29305208e+00 1.91994265e-01 -5.44718266e-01 1.68688089e-01
2.99478441e-01 -4.06954139e-02 -1.68331370e-01 -6.37373403e-02
4.00468349e-01 -1.22684233e-01 -2.33951896e-01 -1.01265693e+00
-8.19802940e-01 3.58656526e-01 6.67224526e-01 6.41830325e-01
1.04617059e+00 -1.32675245e-01 3.35960358e-01 4.26952422e-01
1.04201090e+00 -1.25570595e+00 -6.91717744e-01 6.98267937e-01
6.75764799e-01 -9.38169777e-01 -5.27780652e-02 -4.42653120e-01
-3.04390788e-01 1.04763353e+00 5.19587040e-01 1.31268844e-01
9.26363468e-01 1.07137501e+00 -1.47912726e-01 -4.52365130e-01
-9.47354972e-01 -3.05515885e-01 -2.98846364e-01 8.49979818e-01
-2.19077477e-03 5.41623831e-01 -8.37851167e-01 1.43348169e+00
-2.50058770e-01 9.05492127e-01 7.47018099e-01 1.32479835e+00
-5.15045702e-01 -1.54618251e+00 -6.57779515e-01 1.63320437e-01
5.59927113e-02 3.37423444e-01 -3.17078114e-01 1.02966785e+00
5.53696811e-01 1.60396326e+00 -2.67551720e-01 -8.46629560e-01
7.69588768e-01 2.46259123e-01 4.70144540e-01 -4.12508935e-01
-8.95816803e-01 -8.47946048e-01 5.00910699e-01 -8.40491056e-01
-6.74922705e-01 -6.27017915e-01 -1.57304239e+00 -5.53926170e-01
-2.90113300e-01 5.63297212e-01 4.53358650e-01 1.03634298e+00
-1.81703329e-01 2.36132383e-01 4.20488328e-01 3.51478457e-01
-6.27801359e-01 -4.96192276e-01 -6.59029305e-01 4.96046543e-01
-1.07147656e-01 -4.71377611e-01 -2.93667495e-01 2.92251050e-01]
|
[8.65793514251709, 6.772846221923828]
|
363c1099-5280-4ae8-8a51-8579ba28add5
|
locally-interpretable-model-agnostic
|
2108.06907
| null |
https://arxiv.org/abs/2108.06907v2
|
https://arxiv.org/pdf/2108.06907v2.pdf
|
Select Wisely and Explain: Active Learning and Probabilistic Local Post-hoc Explainability
|
Albeit the tremendous performance improvements in designing complex artificial intelligence (AI) systems in data-intensive domains, the black-box nature of these systems leads to the lack of trustworthiness. Post-hoc interpretability methods explain the prediction of a black-box ML model for a single instance, and such explanations are being leveraged by domain experts to diagnose the underlying biases of these models. Despite their efficacy in providing valuable insights, existing approaches fail to deliver consistent and reliable explanations. In this paper, we propose an active learning-based technique called UnRAvEL (Uncertainty driven Robust Active Learning Based Locally Faithful Explanations), which consists of a novel acquisition function that is locally faithful and uses uncertainty-driven sampling based on the posterior distribution on the probabilistic locality using Gaussian process regression(GPR). We present a theoretical analysis of UnRAvEL by treating it as a local optimizer and analyzing its regret in terms of instantaneous regrets over a global optimizer. We demonstrate the efficacy of the local samples generated by UnRAvEL by incorporating different kernels such as the Matern and linear kernels in GPR. Through a series of experiments, we show that UnRAvEL outperforms the baselines with respect to stability and local fidelity on several real-world models and datasets. We show that UnRAvEL is an efficient surrogate dataset generator by deriving importance scores on this surrogate dataset using sparse linear models. We also showcase the sample efficiency and flexibility of the developed framework on the Imagenet dataset using a pre-trained ResNet model.
|
['Ranjitha Prasad', 'Aditya Saini']
|
2021-08-16
| null | null | null | null |
['gpr', 'gpr']
|
['computer-vision', 'miscellaneous']
|
[ 1.63270757e-01 8.94948065e-01 -3.35257828e-01 -6.16262615e-01
-1.21680415e+00 -4.01648730e-01 8.98397744e-01 1.86823174e-01
-1.22262128e-01 8.67369831e-01 1.22429296e-01 -2.11938933e-01
-6.58230901e-01 -3.87997836e-01 -1.23423898e+00 -8.58881593e-01
-7.30181038e-02 1.07577145e+00 -2.08996639e-01 3.13446581e-01
3.20996523e-01 2.64765829e-01 -1.33643401e+00 5.13281189e-02
1.30215359e+00 1.10720527e+00 -1.38050225e-02 5.34715950e-01
6.60321563e-02 8.73115242e-01 -2.29254723e-01 -7.12630630e-01
4.20044065e-01 -2.87923306e-01 -5.44804335e-01 8.87961611e-02
3.96496385e-01 -2.99011916e-01 2.49635994e-01 8.98126483e-01
2.15540364e-01 1.82046458e-01 7.71979630e-01 -1.52274287e+00
-8.49777818e-01 8.26004505e-01 -3.30919683e-01 -9.34825167e-02
-9.21705365e-02 2.22503632e-01 1.32601428e+00 -1.19041646e+00
3.45820218e-01 1.25877392e+00 8.63090336e-01 4.27098989e-01
-1.62904215e+00 -3.86901945e-01 1.80874497e-01 1.90253422e-01
-9.58809078e-01 -7.69861400e-01 5.83286166e-01 -4.72684860e-01
5.32735527e-01 1.93022847e-01 2.89668888e-01 1.36279762e+00
3.15208226e-01 1.05454075e+00 9.79877055e-01 -3.93886119e-01
9.67684984e-01 4.17961240e-01 2.80879587e-01 7.89397895e-01
3.87598664e-01 4.45013940e-01 -9.44470882e-01 -6.16692066e-01
6.08065248e-01 -1.98669881e-02 -1.34423986e-01 -8.55529785e-01
-9.94815052e-01 1.15862119e+00 5.58818221e-01 -4.30778503e-01
-6.81559384e-01 3.31082135e-01 -4.15733680e-02 -9.69437286e-02
9.28477645e-01 6.23719871e-01 -5.21729112e-01 1.12462096e-01
-1.04131436e+00 2.17834219e-01 8.78844917e-01 9.70796108e-01
7.09392071e-01 9.13137868e-02 -2.59071797e-01 4.60767239e-01
8.82622957e-01 3.14741224e-01 1.88654378e-01 -1.01145136e+00
2.62254864e-01 5.53842187e-01 3.16743344e-01 -8.12975526e-01
-1.79228112e-02 -7.13477910e-01 -7.11356997e-01 4.63639766e-01
2.99390793e-01 -8.03006217e-02 -8.59674156e-01 1.64007676e+00
2.95180172e-01 3.06022286e-01 1.26377866e-01 9.55074787e-01
3.90717804e-01 4.94195133e-01 5.27411252e-02 -1.07092805e-01
6.38193071e-01 -1.25720036e+00 -5.69026172e-01 -4.06785250e-01
1.62902042e-01 -2.49333501e-01 1.02848411e+00 5.59821308e-01
-1.07013011e+00 -3.52173090e-01 -1.07523870e+00 1.00366563e-01
1.30949616e-01 1.08219497e-01 8.50874424e-01 5.70950389e-01
-9.65675056e-01 6.62885845e-01 -1.19011807e+00 4.31960933e-02
9.00275290e-01 3.81870657e-01 -3.14769685e-01 2.12854490e-01
-7.47731268e-01 8.11020434e-01 2.87985861e-01 2.39037663e-01
-1.53429258e+00 -1.13766897e+00 -7.91044533e-01 1.29957244e-01
3.70671660e-01 -8.55018139e-01 1.13860738e+00 -1.50416911e+00
-1.61284208e+00 4.40445185e-01 -1.91974431e-01 -1.04770482e+00
1.03078175e+00 -5.00578701e-01 2.06819400e-01 4.45291810e-02
-1.24664996e-02 5.96075833e-01 1.29774833e+00 -1.49677253e+00
-2.44811833e-01 -2.87601799e-01 -9.67098176e-02 -8.33349768e-03
7.40202069e-02 -2.65115678e-01 4.75314595e-02 -4.61639971e-01
9.63028073e-02 -9.20740426e-01 -5.55044353e-01 3.08898926e-01
-5.29871464e-01 8.04601982e-02 4.08210516e-01 -5.44475853e-01
6.49692774e-01 -1.89223802e+00 2.01817393e-01 3.64527881e-01
4.36847895e-01 -4.39179279e-02 2.12586112e-02 7.57133439e-02
-5.63906357e-02 2.03427821e-01 -7.07939684e-01 -8.94292533e-01
3.19418609e-01 3.99220020e-01 -8.57253432e-01 6.21941984e-01
5.17189264e-01 1.14958417e+00 -9.72985446e-01 -1.92321494e-01
1.94759339e-01 6.29438400e-01 -5.81012428e-01 3.90116602e-01
-3.82619798e-01 5.51304638e-01 -4.17435169e-01 4.86950964e-01
5.73605359e-01 -4.95237648e-01 -1.93801925e-01 1.80422664e-01
2.67071962e-01 -1.88726583e-03 -1.00861776e+00 1.73729563e+00
-4.68859315e-01 4.50314075e-01 -1.31128237e-01 -1.06790960e+00
9.22920704e-01 3.38113874e-01 1.52812868e-01 -1.77448153e-01
-1.34588346e-01 3.25780183e-01 -3.14078331e-01 -2.24574849e-01
1.97554991e-01 -1.78181529e-01 2.25823477e-01 4.79251206e-01
2.71163464e-01 1.13057323e-01 -3.20630401e-01 1.59614518e-01
1.11983585e+00 4.14239883e-01 4.54963535e-01 -4.85882699e-01
2.83349633e-01 6.26793504e-02 5.02896667e-01 1.09289849e+00
-1.38458401e-01 9.87381756e-01 4.11348075e-01 -5.27514935e-01
-1.13426077e+00 -1.36415470e+00 -2.98982263e-01 7.63396919e-01
-1.01006292e-01 -1.41019404e-01 -6.54023409e-01 -1.01246309e+00
1.34699792e-01 1.32023787e+00 -9.22932982e-01 -2.05255628e-01
1.91861451e-01 -1.07507563e+00 3.49943221e-01 4.25952524e-01
3.57888132e-01 -8.37961972e-01 -6.48049474e-01 2.45084260e-02
1.61901429e-01 -7.80068099e-01 7.41975680e-02 4.32414800e-01
-1.03955960e+00 -8.54306042e-01 -4.66807723e-01 1.08699344e-01
1.04828930e+00 -2.82013148e-01 1.28250027e+00 -2.24704161e-01
2.82832817e-03 3.86503458e-01 -2.05194846e-01 -7.14199960e-01
-3.29930067e-01 1.21785523e-02 2.39544272e-01 3.42667103e-01
-1.72978994e-02 -5.42987525e-01 -5.91936529e-01 1.64628893e-01
-8.49009037e-01 2.10206985e-01 6.59155846e-01 1.07013547e+00
7.29511321e-01 -3.50126088e-01 6.22827947e-01 -1.21631765e+00
5.49333692e-01 -7.86316097e-01 -7.67311692e-01 3.02382916e-01
-1.03165162e+00 6.46409273e-01 6.55109107e-01 -3.56222391e-01
-1.34232271e+00 3.11340749e-01 2.94650674e-01 -6.39931023e-01
-9.06528011e-02 4.10891235e-01 -1.00552507e-01 -9.24498960e-03
9.76111591e-01 -9.27809253e-02 1.19440138e-01 -2.68442333e-01
5.25904238e-01 3.27723652e-01 2.54114240e-01 -7.80525923e-01
9.97424185e-01 6.86651707e-01 -8.00841227e-02 -3.15391749e-01
-1.28492713e+00 1.41484640e-03 -4.32120830e-01 -1.58132672e-01
4.21566814e-01 -8.96671295e-01 -4.33803469e-01 -7.53682926e-02
-1.04699671e+00 -3.30434889e-01 -6.88499689e-01 5.08846581e-01
-1.02789271e+00 1.30735748e-02 2.51434501e-02 -1.21295345e+00
-4.07206386e-01 -1.23857057e+00 1.17371869e+00 1.48285657e-01
-3.98002326e-01 -1.34800863e+00 1.66879848e-01 5.64821661e-01
5.43171346e-01 2.86215872e-01 7.54049718e-01 -1.15190661e+00
-8.81893754e-01 -1.48830041e-01 -1.86737962e-02 4.27004755e-01
-2.27609634e-01 7.90585503e-02 -1.36818790e+00 -7.72777423e-02
2.82394618e-01 -4.40063179e-01 8.98334205e-01 5.51076114e-01
1.17398989e+00 -6.38172269e-01 -1.01166196e-01 6.75398946e-01
1.48876035e+00 -2.59939581e-01 6.03478611e-01 1.27933145e-01
4.27000612e-01 6.62238181e-01 4.92628098e-01 4.54393178e-01
9.12758708e-02 2.77912736e-01 7.11305559e-01 -9.76052284e-02
3.25351328e-01 -4.20836151e-01 5.47417462e-01 3.60346228e-01
-1.38842151e-01 -5.56340325e-04 -9.07189608e-01 5.20681202e-01
-2.52443409e+00 -7.59061038e-01 -9.75900237e-03 2.51185656e+00
7.70292997e-01 1.88523874e-01 -2.87515640e-01 -2.29806319e-01
4.73033607e-01 -1.58890650e-01 -8.95752966e-01 -2.48575717e-01
-1.93399101e-01 5.65752611e-02 6.02713764e-01 7.25990772e-01
-9.02417541e-01 5.19367933e-01 6.46210194e+00 8.32851231e-01
-7.51455903e-01 3.33597839e-01 1.05003929e+00 -1.11319266e-01
-5.98468959e-01 4.53513354e-01 -6.06419623e-01 3.42013627e-01
1.02174664e+00 -2.57759452e-01 4.84194100e-01 1.21939409e+00
3.81723851e-01 -7.93556646e-02 -1.49795187e+00 6.81908965e-01
1.10802375e-01 -1.45677578e+00 5.54883592e-02 4.77018468e-02
9.09311295e-01 1.74084157e-02 2.59570092e-01 5.24617545e-02
6.25802636e-01 -1.31628025e+00 1.04669595e+00 9.41673160e-01
2.32289687e-01 -6.75833762e-01 6.87105358e-01 5.72450936e-01
-2.84532040e-01 -1.39016256e-01 -4.86587077e-01 7.04791397e-02
-2.12991089e-02 1.01213002e+00 -1.01297998e+00 5.58096707e-01
3.45793217e-01 6.25850379e-01 -4.51880693e-01 1.05295527e+00
-4.57633585e-01 9.79804397e-01 -4.45657641e-01 2.85447717e-01
1.07416980e-01 -3.01623583e-01 8.22553277e-01 7.07919002e-01
2.05411687e-01 -3.82090002e-01 -2.34936863e-01 1.70725596e+00
8.29327330e-02 -2.62895316e-01 -4.49575722e-01 4.58099321e-03
2.80110002e-01 1.29414070e+00 -5.30868173e-01 -1.45840690e-01
7.19700456e-02 8.61822724e-01 5.12763143e-01 3.81380230e-01
-9.87940967e-01 3.04019541e-01 4.28837866e-01 -9.41190720e-02
1.13119274e-01 1.35463253e-01 -7.24637806e-01 -1.26154673e+00
-2.53149699e-02 -8.40645194e-01 1.31145000e-01 -9.16283369e-01
-1.70987523e+00 4.67736810e-01 1.42426401e-01 -9.54039752e-01
-4.28278357e-01 -5.39605975e-01 -5.78765631e-01 9.37376916e-01
-1.37945330e+00 -1.11641455e+00 -1.11013271e-01 2.43526965e-01
5.81761718e-01 -3.98014784e-01 9.42265987e-01 -2.92822123e-01
-6.19006991e-01 6.21970952e-01 3.83247823e-01 -3.55427146e-01
5.04284322e-01 -1.53473628e+00 3.97093326e-01 8.66336823e-01
4.70975161e-01 9.30465221e-01 1.08189237e+00 -5.94986260e-01
-1.38452339e+00 -9.77145731e-01 5.00093162e-01 -8.22717607e-01
7.49053538e-01 -4.84570146e-01 -7.16667414e-01 9.04174030e-01
-3.33179459e-02 2.42330447e-01 5.22782981e-01 1.36534318e-01
-2.53508180e-01 -1.09037071e-01 -1.43312931e+00 3.99124533e-01
7.51616418e-01 -3.19249660e-01 -6.59981966e-01 4.18732792e-01
6.07287645e-01 -1.94785699e-01 -5.11691272e-01 4.30187017e-01
4.10848022e-01 -1.16628850e+00 9.29835856e-01 -7.86284149e-01
7.27557540e-01 -1.34190366e-01 -3.15174572e-02 -1.38587701e+00
-1.54860690e-01 -7.78335392e-01 -4.11732435e-01 1.13780916e+00
7.74266541e-01 -8.43428791e-01 8.50242198e-01 1.07338250e+00
-5.42239957e-02 -9.09155488e-01 -9.66591299e-01 -4.84416574e-01
-1.88917473e-01 -7.21660733e-01 3.66919667e-01 6.53113544e-01
-3.63272011e-01 2.50482500e-01 -3.32526118e-01 3.44282150e-01
1.22184896e+00 -9.01521221e-02 7.03361511e-01 -1.30215168e+00
-5.31120479e-01 3.45805399e-02 -2.56035030e-01 -6.62272274e-01
3.02097559e-01 -7.31664002e-01 2.90301889e-01 -1.26186073e+00
3.22211474e-01 -5.66050172e-01 -1.78322017e-01 4.19680655e-01
-2.42386773e-01 -1.57799870e-01 -4.43728976e-02 4.76220280e-01
-6.37894154e-01 8.76345813e-01 6.38549507e-01 4.43647392e-02
3.04428395e-02 1.19925283e-01 -7.84252167e-01 9.59632277e-01
5.55930316e-01 -6.56483889e-01 -7.23064899e-01 -3.21525335e-01
5.93241870e-01 -6.89431131e-02 8.40026677e-01 -7.90286422e-01
2.91990787e-01 6.65100000e-04 3.84934187e-01 -2.76240647e-01
2.86498159e-01 -8.94016862e-01 4.44985837e-01 7.15400428e-02
-8.85282636e-01 -1.99696124e-01 2.30292026e-02 1.01095724e+00
1.85661539e-02 -3.29075515e-01 6.32238090e-01 -6.93155918e-03
-3.83775920e-01 3.25764835e-01 6.44904524e-02 7.68839717e-02
9.39891219e-01 -1.37624964e-01 -2.00846344e-01 -3.70794535e-01
-5.85436046e-01 1.46454617e-01 4.31015909e-01 1.76990390e-01
6.93208218e-01 -1.12460887e+00 -7.78858066e-01 2.13792071e-01
1.77485526e-01 3.13497692e-01 1.83487814e-02 9.42368746e-01
-4.37759489e-01 2.35198677e-01 1.64537523e-02 -8.57496142e-01
-7.07014024e-01 3.20385128e-01 4.60420370e-01 -2.59942114e-01
-5.70240080e-01 9.12816525e-01 3.75509679e-01 -6.66154146e-01
3.03723961e-01 -1.80623010e-01 1.62635356e-01 -2.39610046e-01
4.25259233e-01 2.87025601e-01 8.80595446e-02 -1.82979181e-01
-2.10265934e-01 -6.36605769e-02 1.12599649e-01 -3.01127195e-01
1.50004673e+00 -5.16148880e-02 3.83414924e-02 7.79807091e-01
7.28842497e-01 -1.36718869e-01 -1.90288019e+00 -1.44363284e-01
1.59430832e-01 -5.12184143e-01 2.66902834e-01 -9.37797964e-01
-9.50494468e-01 7.41843462e-01 4.18515116e-01 7.31202355e-03
7.74898410e-01 -1.54125206e-02 4.67077158e-02 3.63017976e-01
4.12497163e-01 -8.00497651e-01 -9.43051428e-02 1.02689408e-01
1.10489523e+00 -1.47076130e+00 6.84044361e-02 -1.23951763e-01
-9.11361277e-01 9.27839994e-01 1.39863610e-01 -4.03078079e-01
5.20370305e-01 1.73044339e-01 -1.43201724e-02 -1.60996556e-01
-1.06752110e+00 2.73389697e-01 4.03968245e-01 5.38063467e-01
2.39997637e-03 2.65954360e-02 1.98248416e-01 9.26453531e-01
-1.04984120e-01 6.68114563e-03 3.11326534e-01 4.27641302e-01
-1.29825145e-01 -6.95575535e-01 -4.35760945e-01 2.93392539e-01
-2.88416475e-01 -2.16144785e-01 -4.50950682e-01 5.32517076e-01
-1.07622862e-01 8.82444501e-01 -9.35955495e-02 -7.46327713e-02
-1.84549958e-01 3.90558839e-02 2.00790688e-01 -5.02290249e-01
-4.86188978e-01 -3.34449530e-01 -1.61748566e-02 -6.37072802e-01
-2.93566197e-01 -6.43872440e-01 -1.10248482e+00 -1.45611569e-01
-3.29589248e-01 2.02481568e-01 7.66811907e-01 1.24958122e+00
5.76435447e-01 2.63726860e-01 4.78264064e-01 -8.31059933e-01
-1.19363427e+00 -8.22399616e-01 -3.69712681e-01 3.51882279e-01
3.69972020e-01 -7.91807652e-01 -9.37566161e-01 7.55495951e-03]
|
[8.71823787689209, 5.411235332489014]
|
7da661ef-d679-4ef5-a15b-68a7abdd8a5f
|
scaling-native-language-identification-with
|
2211.10117
| null |
https://arxiv.org/abs/2211.10117v1
|
https://arxiv.org/pdf/2211.10117v1.pdf
|
Scaling Native Language Identification with Transformer Adapters
|
Native language identification (NLI) is the task of automatically identifying the native language (L1) of an individual based on their language production in a learned language. It is useful for a variety of purposes including marketing, security and educational applications. NLI is usually framed as a multi-label classification task, where numerous designed features are combined to achieve state-of-the-art results. Recently deep generative approach based on transformer decoders (GPT-2) outperformed its counterparts and achieved the best results on the NLI benchmark datasets. We investigate this approach to determine the practical implications compared to traditional state-of-the-art NLI systems. We introduce transformer adapters to address memory limitations and improve training/inference speed to scale NLI applications for production.
|
['Gerold Schneider', 'Ahmet Yavuz Uluslu']
|
2022-11-18
| null | null | null | null |
['marketing', 'native-language-identification']
|
['miscellaneous', 'natural-language-processing']
|
[ 5.96551597e-01 -1.27638280e-01 -4.17816937e-01 -5.09553432e-01
-1.00986576e+00 -8.53933632e-01 8.48797560e-01 -5.67977540e-02
-2.19615817e-01 9.33137357e-01 3.67787182e-02 -4.47536737e-01
-1.16901165e-02 -5.44631600e-01 -5.58549345e-01 -2.83855557e-01
3.00105631e-01 9.63946342e-01 -3.55157375e-01 2.70942539e-01
-2.39765495e-02 3.62402707e-01 -1.30314159e+00 4.79566216e-01
8.80074441e-01 8.23335826e-01 4.48504463e-02 5.71132541e-01
-4.29438859e-01 7.21567929e-01 -6.89223289e-01 -7.90160954e-01
2.41488926e-02 -4.07796323e-01 -8.10008526e-01 -4.70071167e-01
5.86377442e-01 -7.78657719e-02 2.24744603e-02 1.19911897e+00
6.77001595e-01 -1.03724644e-01 8.25168252e-01 -1.26735985e+00
-9.85417426e-01 6.82846427e-01 -2.26893976e-01 -5.33534363e-02
7.10080683e-01 -1.04436889e-01 8.35192382e-01 -6.78987980e-01
5.09932101e-01 1.73882544e+00 7.38878608e-01 7.69502282e-01
-1.37577331e+00 -1.11479652e+00 -2.24548683e-01 -2.04296902e-01
-1.54134881e+00 -5.69873154e-01 4.91657555e-01 -4.88967210e-01
1.14215875e+00 3.18807587e-02 -4.21191677e-02 1.70656490e+00
2.99895525e-01 1.02982473e+00 1.49975753e+00 -5.07693291e-01
-1.87102422e-01 5.84417582e-01 2.46922716e-01 7.80581176e-01
1.28848866e-01 1.59982234e-01 -7.49124587e-01 -1.65911809e-01
3.72860610e-01 -2.99428493e-01 8.52688029e-02 2.49391258e-01
-1.38363874e+00 1.14009607e+00 -2.57721096e-01 5.89179099e-01
1.81479976e-02 6.28748611e-02 4.34730858e-01 2.59292573e-01
4.87974435e-01 4.73767489e-01 -5.72626173e-01 -4.55640197e-01
-1.02298903e+00 1.81415081e-01 9.46431100e-01 1.03052592e+00
6.08816683e-01 3.66007984e-02 -4.79596794e-01 1.18803084e+00
3.98775369e-01 6.93889022e-01 9.27595794e-01 -5.65869212e-01
4.33316678e-01 8.34554136e-01 -3.24965745e-01 -3.92732233e-01
-3.38966906e-01 -6.22084320e-01 -9.33570325e-01 -3.22029665e-02
4.47422832e-01 -1.59431860e-01 -7.95222998e-01 1.82853341e+00
-3.10224742e-02 2.10274518e-01 2.51421154e-01 3.45174104e-01
1.01882935e+00 7.68529713e-01 3.48043799e-01 1.20982736e-01
1.32385480e+00 -8.55606258e-01 -5.41188478e-01 -4.28904206e-01
7.50742912e-01 -8.60839367e-01 1.10443056e+00 4.01709050e-01
-6.06425285e-01 -9.43237662e-01 -6.61168873e-01 -1.42821431e-01
-7.21188068e-01 7.86260188e-01 9.40254331e-01 1.29928327e+00
-1.07922518e+00 4.34213459e-01 -2.23951861e-01 -5.15802562e-01
2.70982742e-01 6.36931241e-01 -5.51968336e-01 4.83160652e-02
-1.20601249e+00 5.99537253e-01 3.54310811e-01 -1.82422012e-01
-1.01009226e+00 -8.53939593e-01 -1.02306330e+00 -1.99283138e-01
9.13249552e-02 -5.69231391e-01 1.19534826e+00 -1.00058389e+00
-1.75996542e+00 1.30009794e+00 -3.25830847e-01 -3.89432788e-01
3.14413607e-01 -1.44600943e-01 -6.92521513e-01 -3.67971271e-01
4.38814223e-01 8.78795326e-01 6.37724400e-01 -8.01408410e-01
-7.54013896e-01 -3.23303521e-01 -1.91661596e-01 1.60461161e-02
-3.20798278e-01 3.89938593e-01 -7.01996963e-03 -5.07282555e-01
-4.73696828e-01 -1.17383814e+00 4.99944091e-01 -5.68302393e-01
-5.32377839e-01 -6.62171960e-01 8.08717191e-01 -8.44664454e-01
1.18082988e+00 -2.12470889e+00 -1.50245890e-01 -9.26899090e-02
-4.28843535e-02 5.55922389e-01 -1.96163312e-01 2.79957980e-01
-9.10345092e-03 2.78505176e-01 1.18052214e-01 -5.79581976e-01
2.47064799e-01 -1.69578433e-01 -2.36668646e-01 2.20481679e-01
-3.04979887e-02 1.15809512e+00 -7.38492250e-01 -4.19171214e-01
7.56791756e-02 5.20795405e-01 -1.78907678e-01 5.17770290e-01
-1.70875147e-01 6.14839852e-01 -1.08887702e-01 8.54460835e-01
2.24413618e-01 -1.92136258e-01 1.60047994e-03 1.84255783e-02
1.12953521e-01 4.24048901e-01 -8.14012527e-01 1.76147187e+00
-7.90136874e-01 5.92901170e-01 -1.23953693e-01 -8.63250256e-01
1.03698397e+00 4.39732105e-01 2.58552969e-01 -5.10222793e-01
2.95448042e-02 3.27597857e-01 2.04484180e-01 -8.70725811e-02
1.17975585e-01 4.83720899e-02 -5.17136455e-01 4.79667366e-01
5.49427509e-01 3.30539525e-01 1.93136185e-01 -3.09582204e-02
7.86581337e-01 3.56856972e-01 4.63527948e-01 -6.17977679e-01
8.28712523e-01 -3.35968077e-01 4.41043586e-01 9.14371908e-01
-2.53884166e-01 1.46532282e-01 1.61391750e-01 -1.94480106e-01
-7.35174119e-01 -1.17429864e+00 -1.85418084e-01 1.41621125e+00
-4.15986985e-01 -3.05743694e-01 -7.12501287e-01 -7.55150080e-01
7.05721229e-03 1.03556108e+00 -3.78309637e-01 -7.73003995e-02
-2.46813834e-01 -5.76544881e-01 9.45412874e-01 4.08192366e-01
7.58444488e-01 -1.29012549e+00 -4.12813984e-02 2.71471232e-01
-1.65998489e-01 -1.26778448e+00 -6.58962965e-01 -2.57696887e-03
-2.33069420e-01 -6.33203089e-01 -6.96360469e-01 -1.07472825e+00
3.88471484e-01 -2.46182114e-01 1.18215835e+00 -4.07197088e-01
-2.54935324e-01 3.81818652e-01 -1.07040271e-01 -4.24963713e-01
-1.03009999e+00 6.62790060e-01 2.65203089e-01 1.20994568e-01
6.29232049e-01 -4.37109321e-01 3.62410843e-02 3.61068249e-02
-4.05886948e-01 2.32220247e-01 5.87850869e-01 8.11415792e-01
3.84819716e-01 1.71805263e-01 6.83288217e-01 -1.26842535e+00
9.16390479e-01 -1.30036071e-01 -5.42510509e-01 8.33704650e-01
-6.49126351e-01 2.82981485e-01 6.69516802e-01 -8.74633789e-01
-1.11627007e+00 1.43664584e-01 -2.94269294e-01 -1.45920692e-02
-4.97063726e-01 2.12833047e-01 -4.07360256e-01 -3.28623265e-01
1.18130572e-01 5.29751062e-01 -4.07118708e-01 -5.69012046e-01
1.35373622e-01 1.11659706e+00 5.57294965e-01 -7.21317232e-01
4.03273433e-01 -4.51289386e-01 -8.76117051e-02 -7.50236630e-01
-1.08348310e+00 -4.52199638e-01 -8.20759892e-01 -6.32227510e-02
7.84449160e-01 -1.13932884e+00 -1.11122239e+00 5.51894188e-01
-1.12055850e+00 -4.40741926e-01 1.84931025e-01 3.70679110e-01
-4.48600382e-01 7.25419074e-02 -6.22720420e-01 -8.76430273e-01
-8.12457979e-01 -1.36892867e+00 1.38415217e+00 3.15615326e-01
-6.02710664e-01 -1.23720276e+00 1.00987487e-01 5.92697263e-01
4.41009432e-01 -8.40544924e-02 1.13985932e+00 -1.16628683e+00
-1.55134276e-01 8.86672288e-02 -2.37371042e-01 4.37754065e-01
8.12864229e-02 -4.58940715e-01 -1.11768436e+00 -4.96563017e-01
-2.73197919e-01 -5.64639390e-01 4.41300988e-01 7.63832107e-02
9.61216390e-01 -3.88293028e-01 -5.50071239e-01 6.65679455e-01
1.37046063e+00 2.61870712e-01 2.74363011e-01 -6.98470175e-02
9.35917974e-01 4.59753364e-01 1.35226339e-01 5.79710444e-03
2.94638723e-01 7.11811841e-01 -4.36954141e-01 2.21088618e-01
-4.32076931e-01 -6.80564821e-01 6.08145237e-01 6.34261608e-01
5.71801603e-01 -4.57152456e-01 -9.95985389e-01 5.26485682e-01
-1.50164735e+00 -6.85389459e-01 1.74030423e-01 2.38390660e+00
1.08401561e+00 2.24071145e-01 2.98326984e-02 -2.10540250e-01
6.75568104e-01 -2.75871307e-01 -4.47619975e-01 -7.08589971e-01
-6.13965616e-02 3.98376524e-01 5.40565550e-01 7.97029793e-01
-1.22555768e+00 1.27168167e+00 6.70533133e+00 1.26832461e+00
-1.09437966e+00 4.39893901e-01 9.68823135e-01 2.90419132e-01
-5.05541675e-02 -1.61487430e-01 -1.83651257e+00 6.10912919e-01
1.50516152e+00 -1.85526684e-01 7.38516152e-01 7.99367249e-01
-3.09308082e-01 2.18586475e-02 -1.42790079e+00 1.27930188e+00
3.38297814e-01 -8.86456192e-01 1.75225645e-01 2.69315898e-01
6.48387671e-01 -1.17979996e-01 2.44847879e-01 7.71096885e-01
4.43441659e-01 -1.29341352e+00 4.91473407e-01 3.50574315e-01
1.34109092e+00 -6.90589309e-01 6.69822931e-01 4.78979141e-01
-1.08140576e+00 -2.62409002e-02 -1.08387642e-01 -1.89176598e-03
1.05056511e-02 5.76635897e-01 -1.28736246e+00 2.16143578e-01
2.67197132e-01 4.02565151e-01 -8.45177829e-01 3.73533964e-01
-2.35623226e-01 1.08070016e+00 -4.47063953e-01 -1.98625088e-01
1.80678606e-01 -1.82122961e-01 1.82868451e-01 1.44508040e+00
4.84016269e-01 -5.14115989e-01 3.97040397e-01 1.02136707e+00
-2.55297571e-01 2.48706073e-01 -9.39471364e-01 -5.28531849e-01
4.72512960e-01 1.25536036e+00 -3.91695380e-01 -3.67128760e-01
-5.63794196e-01 1.42953205e+00 4.24417406e-01 7.30202869e-02
-6.18207514e-01 -3.21358770e-01 6.40561461e-01 -1.71545763e-02
-6.20299093e-02 -6.49109334e-02 -1.46793947e-01 -1.04510570e+00
-3.59447539e-01 -1.09312797e+00 3.29519778e-01 -4.54057038e-01
-1.37318170e+00 6.33753717e-01 1.83298960e-02 -8.14430296e-01
-8.84981453e-01 -9.13206875e-01 -1.95386052e-01 1.11342478e+00
-1.26782119e+00 -1.79424202e+00 5.68995252e-02 4.15660113e-01
5.82380414e-01 -9.00012434e-01 1.33593214e+00 3.99270833e-01
-6.44447207e-01 1.22521222e+00 2.44458690e-01 3.91388983e-01
8.76140118e-01 -1.27031064e+00 4.74930376e-01 8.36542308e-01
6.46087468e-01 7.19833314e-01 2.40574241e-01 -7.36506820e-01
-1.19690704e+00 -1.23100615e+00 1.65729868e+00 -4.28335547e-01
5.20595670e-01 -8.36029828e-01 -3.42459291e-01 8.85537505e-01
3.07143718e-01 -3.56434703e-01 1.07003117e+00 2.14056194e-01
-1.86000854e-01 -2.16820672e-01 -1.17897356e+00 3.58842343e-01
1.09290183e+00 -8.40693057e-01 -1.70781255e-01 6.48196459e-01
6.00253701e-01 -2.39347205e-01 -1.16336751e+00 2.37822115e-01
5.37823081e-01 -6.65056348e-01 1.02112067e+00 -3.09610665e-01
1.21411294e-01 1.00935809e-02 1.18154576e-02 -1.28955805e+00
-5.66393256e-01 -7.60935187e-01 6.82925135e-02 1.91481972e+00
5.69223106e-01 -8.42864215e-01 4.53972876e-01 4.48497027e-01
3.28525960e-01 -5.70476115e-01 -8.94551873e-01 -8.56130838e-01
-1.46265477e-02 -2.22942069e-01 6.66540086e-01 6.67006552e-01
-2.32392073e-01 9.92004573e-01 -5.02217770e-01 -2.94451956e-02
5.75488269e-01 2.45062485e-01 7.59817839e-01 -1.28112352e+00
-3.87565136e-01 -4.48364019e-01 -3.15654010e-01 -8.73020470e-01
9.10179853e-01 -1.37170827e+00 -3.15282375e-01 -1.08832693e+00
4.36199546e-01 -3.14964473e-01 -1.09719798e-01 6.98482335e-01
-1.01773314e-01 5.01175523e-01 1.48087502e-01 -4.33713384e-02
-5.27574480e-01 1.44115955e-01 6.66821837e-01 -4.93854642e-01
4.56664972e-02 1.97568059e-01 -9.13767099e-01 3.31355631e-01
8.29023421e-01 -4.38293308e-01 -5.88546515e-01 -3.50098908e-01
2.44588200e-02 -9.61312652e-02 -3.49994865e-03 -1.28645039e+00
2.03070208e-01 2.22469538e-01 4.94431555e-01 -3.89615476e-01
2.12277591e-01 -4.67057616e-01 2.48033106e-01 3.42955619e-01
-6.76442742e-01 6.77168295e-02 1.61558613e-01 3.90287600e-02
-1.35393828e-01 -5.11291027e-01 6.23663664e-01 -1.84252679e-01
-7.55359411e-01 2.94003814e-01 -3.74404520e-01 1.19740874e-01
9.43749845e-01 6.36040047e-03 -1.35612652e-01 -3.01001817e-01
-3.05407166e-01 -1.68229237e-01 7.74268657e-02 6.75797343e-01
2.78841436e-01 -1.31850088e+00 -9.57957149e-01 6.67594552e-01
2.41911784e-01 -5.62966526e-01 -1.99434176e-01 2.50790417e-01
-3.67071182e-01 9.66458201e-01 3.35952006e-02 -3.71719897e-01
-1.35768414e+00 4.63116199e-01 2.80241728e-01 -9.07452285e-01
-2.22790778e-01 8.64919424e-01 3.92276645e-01 -9.11663115e-01
3.13800544e-01 -2.26874333e-02 -1.11307099e-01 -3.03157866e-01
6.35266483e-01 1.14089362e-01 -2.03001574e-02 -8.89157057e-01
-5.10193467e-01 4.26336378e-01 -1.09653018e-01 -1.96573704e-01
7.91558087e-01 1.86096765e-02 -1.16065405e-01 6.03184462e-01
1.55223632e+00 -7.38179088e-02 -5.72230458e-01 -2.16034204e-01
-8.58179033e-02 -9.65477675e-02 5.90164363e-02 -1.34737647e+00
-6.43273532e-01 9.75383759e-01 8.00787210e-01 -3.87720913e-01
6.48494780e-01 -1.92695811e-01 1.01685166e+00 3.24735135e-01
5.34961998e-01 -1.08003902e+00 -3.85180444e-01 4.03394699e-01
5.06238997e-01 -1.44022012e+00 -2.59802103e-01 -2.74516672e-01
-6.77419305e-01 1.04730844e+00 4.42754626e-01 3.40435624e-01
5.89970469e-01 4.33913648e-01 9.18442905e-02 2.32912093e-01
-4.22221452e-01 -5.21336682e-02 5.85680246e-01 7.99960494e-01
9.11104202e-01 4.18100238e-01 -2.92337120e-01 8.84156227e-01
-3.41867715e-01 -3.09265181e-02 -2.28718117e-01 2.43559778e-01
4.74287421e-02 -1.53502262e+00 -2.41700321e-01 6.41457260e-01
-6.68506622e-01 -2.97239661e-01 -4.53871101e-01 5.07813752e-01
4.83768433e-01 1.02919865e+00 -2.71146208e-01 -3.73838067e-01
-8.31384361e-02 6.03216827e-01 5.46364546e-01 -6.73816502e-01
-9.53911960e-01 -2.51386017e-01 2.76341408e-01 -2.93719083e-01
-3.04294318e-01 -7.92045355e-01 -7.27010369e-01 -2.35157013e-01
6.45107031e-02 1.91336460e-02 6.24262989e-01 1.02560890e+00
3.44533712e-01 3.52823645e-01 3.10601205e-01 -3.10897827e-01
-6.16896451e-01 -1.22973561e+00 -5.97295821e-01 5.18986940e-01
-1.65419132e-01 -5.19076228e-01 -2.48792432e-02 1.21591888e-01]
|
[10.395227432250977, 10.523723602294922]
|
d9f4e37b-f60c-4622-905b-bd7335328420
|
using-under-trained-deep-ensembles-to-learn
|
2009.11128
| null |
https://arxiv.org/abs/2009.11128v2
|
https://arxiv.org/pdf/2009.11128v2.pdf
|
Using Under-trained Deep Ensembles to Learn Under Extreme Label Noise
|
Improper or erroneous labelling can pose a hindrance to reliable generalization for supervised learning. This can have negative consequences, especially for critical fields such as healthcare. We propose an effective new approach for learning under extreme label noise, based on under-trained deep ensembles. Each ensemble member is trained with a subset of the training data, to acquire a general overview of the decision boundary separation, without focusing on potentially erroneous details. The accumulated knowledge of the ensemble is combined to form new labels, that determine a better class separation than the original labels. A new model is trained with these labels to generalize reliably despite the label noise. We focus on a healthcare setting and extensively evaluate our approach on the task of sleep apnea detection. For comparison with related work, we additionally evaluate on the task of digit recognition. In our experiments, we observed performance improvement in accuracy from 6.7\% up-to 49.3\% for the task of digit classification and in kappa from 0.02 up-to 0.55 for the task of sleep apnea detection.
|
['Mohan Kankanhalli', 'Stein Kristiansen', 'Konstantinos Nikolaidis', 'Vera Goebel', 'Thomas Plagemann']
|
2020-09-23
| null | null | null | null |
['sleep-apnea-detection']
|
['medical']
|
[ 5.53281069e-01 4.08676445e-01 9.08128265e-03 -6.75131798e-01
-1.04776180e+00 -4.43404943e-01 5.89874648e-02 4.58324283e-01
-5.33118188e-01 1.09173405e+00 -9.50153321e-02 -3.05812657e-01
-1.00923507e-02 -4.64932591e-01 -3.94587040e-01 -9.61352348e-01
1.33854985e-01 3.97710860e-01 -8.68608207e-02 2.50256419e-01
8.52630585e-02 3.30604285e-01 -1.57952487e+00 5.39309859e-01
1.12533092e+00 1.30089188e+00 -1.74823254e-01 5.35547733e-01
3.34310383e-02 7.87331283e-01 -7.50660181e-01 -2.31235206e-01
9.55554992e-02 -3.96310925e-01 -7.38748372e-01 1.50074780e-01
6.40169203e-01 -9.89190266e-02 1.36636019e-01 1.05424178e+00
8.03345084e-01 1.26345865e-02 9.49360013e-01 -6.70837820e-01
-1.07387781e-01 4.63026941e-01 -1.24849431e-01 2.48189256e-01
-3.08326422e-03 9.79944468e-02 6.31013989e-01 -7.23051727e-01
1.56696558e-01 6.52687967e-01 1.05097616e+00 9.11182523e-01
-1.28019655e+00 -7.44729877e-01 -1.39821349e-02 1.34938955e-01
-1.34429991e+00 -7.10383415e-01 3.31941456e-01 -6.13446653e-01
6.18153989e-01 2.91993171e-01 2.53494322e-01 1.13074684e+00
1.76433578e-01 3.72432888e-01 1.54066050e+00 -4.78902757e-01
4.48173076e-01 3.56801778e-01 5.23783207e-01 5.12805343e-01
3.73706281e-01 7.71972761e-02 -1.95216984e-01 6.89471737e-02
6.98502511e-02 -1.53466761e-02 -1.07910566e-01 6.69564828e-02
-7.39092171e-01 6.05723202e-01 4.89380151e-01 4.93167341e-01
-4.76829857e-01 -2.42557302e-01 3.43008190e-01 1.88377112e-01
7.07100928e-01 5.47888756e-01 -3.73114526e-01 2.71590147e-02
-1.16993761e+00 -1.50358886e-01 9.75147426e-01 3.82152855e-01
5.18293202e-01 -1.78572208e-01 -4.60774451e-01 9.77216363e-01
1.30656078e-01 4.41925764e-01 4.98710930e-01 -9.97241378e-01
2.36712337e-01 7.01647639e-01 7.75350928e-02 -6.24066591e-01
-8.82698834e-01 -1.03819454e+00 -1.10027564e+00 4.45019037e-01
4.96351957e-01 -2.62886286e-01 -1.27481425e+00 1.53343177e+00
8.20306614e-02 1.38154374e-02 2.08959162e-01 4.86307234e-01
9.27568495e-01 1.09820955e-01 3.58041525e-01 -3.19409043e-01
1.25312698e+00 -7.19958544e-01 -6.98649585e-01 -3.59000862e-01
8.04080904e-01 -3.57830405e-01 6.69590116e-01 7.93039382e-01
-8.85919034e-01 -6.72410488e-01 -1.20333195e+00 4.11741942e-01
-3.76677364e-01 4.38544929e-01 4.03873362e-02 8.35674167e-01
-1.06249142e+00 8.10830832e-01 -6.74572051e-01 -2.81640738e-01
7.58572340e-01 5.06314754e-01 -2.87155479e-01 -2.45027483e-01
-8.90933275e-01 1.20677674e+00 3.57057542e-01 8.34636837e-02
-7.66602755e-01 -4.42473352e-01 -6.48046792e-01 -8.86368677e-02
1.71133816e-01 -7.76056111e-01 1.34574735e+00 -8.33974779e-01
-9.95858490e-01 1.11761272e+00 -1.58072948e-01 -5.94764829e-01
7.12534726e-01 -1.69440210e-01 -5.84221840e-01 -7.09810033e-02
1.47775173e-01 4.57094491e-01 6.95571601e-01 -1.23842859e+00
-7.01587737e-01 -6.33931041e-01 -3.56833458e-01 8.65996107e-02
-2.38519073e-01 -2.39584103e-01 1.56585142e-01 -5.32321632e-01
9.37129781e-02 -9.31382895e-01 -3.61546665e-01 -3.18653643e-01
-3.33360642e-01 -2.85996079e-01 3.85029137e-01 -8.91352057e-01
1.36376536e+00 -2.29854107e+00 -1.63922936e-01 2.62409419e-01
5.23013294e-01 3.09234679e-01 3.33034903e-01 -7.83917436e-04
-1.45418435e-01 6.29844815e-02 -7.22381592e-01 -7.48524666e-01
-3.03906232e-01 4.08120900e-01 5.93009777e-02 3.43489796e-01
2.03459084e-01 4.13393080e-01 -7.38859653e-01 -4.29451376e-01
1.21174619e-01 2.95214653e-01 -2.59829432e-01 2.26391658e-01
1.14653885e-01 7.25586653e-01 -1.57353625e-01 4.24194545e-01
6.19820058e-01 -3.04737210e-01 8.21967050e-02 -5.24553470e-02
7.97907487e-02 2.39427000e-01 -9.37859356e-01 1.45344424e+00
-6.06507003e-01 4.19620782e-01 -2.35985205e-01 -1.27454722e+00
1.06158268e+00 4.41782624e-01 2.90852606e-01 -4.48611051e-01
2.88091421e-01 4.82952446e-01 1.61088258e-01 -7.11196005e-01
-4.65570800e-02 -5.09395838e-01 -3.61092463e-02 3.09613973e-01
2.99108088e-01 2.29318559e-01 5.17609380e-02 -1.88266143e-01
1.08397722e+00 -2.74261355e-01 4.00594503e-01 -3.70588601e-01
5.10613203e-01 -2.57671893e-01 3.43276024e-01 9.21025515e-01
-2.97501743e-01 6.99925244e-01 2.72291481e-01 -4.35891807e-01
-7.41205394e-01 -9.26169276e-01 -6.08344972e-01 9.21226084e-01
-2.65508026e-01 -1.45222068e-01 -8.61814976e-01 -1.21521056e+00
-3.77955288e-02 1.08127248e+00 -8.32826793e-01 -5.38224578e-01
-3.16698760e-01 -1.07149470e+00 6.78991139e-01 7.73976505e-01
5.02510786e-01 -1.05674875e+00 -6.39014661e-01 1.34128377e-01
-1.67739362e-01 -1.05448425e+00 -1.30759701e-02 6.85851455e-01
-8.77659321e-01 -1.11690581e+00 -6.48036420e-01 -7.14084864e-01
6.65253758e-01 -4.98738706e-01 1.13969576e+00 1.79302573e-01
-2.65948713e-01 1.27316907e-01 -3.13024163e-01 -7.11111784e-01
-9.07395005e-01 1.80388570e-01 1.70574203e-01 2.21155271e-01
5.19176483e-01 -3.86972159e-01 -6.29600883e-01 3.50884169e-01
-6.05893373e-01 -2.98379362e-01 6.89624310e-01 8.78669620e-01
3.50496769e-01 -8.36442560e-02 1.03028977e+00 -9.56532061e-01
4.40549880e-01 -5.29525816e-01 -2.16637943e-02 2.02036649e-01
-8.87669742e-01 1.65070996e-01 4.40867096e-01 -2.98087627e-01
-8.34914207e-01 1.81973964e-01 -4.23657238e-01 -3.88968959e-02
-6.83885694e-01 1.84396356e-01 1.03570580e-01 3.63444015e-02
1.05752814e+00 2.65536830e-02 -3.33563685e-02 -6.49643958e-01
-2.52333172e-02 1.18752337e+00 5.02351046e-01 -2.22347870e-01
1.07620999e-01 2.97527343e-01 2.53741425e-02 -6.19303226e-01
-1.31072557e+00 -5.67193270e-01 -8.37345243e-01 -2.57274002e-01
1.04593766e+00 -7.24687397e-01 -2.13248104e-01 4.55186814e-01
-7.96276569e-01 -3.73514295e-01 -4.68471467e-01 5.47859311e-01
-3.57985616e-01 2.75841862e-01 -1.83666259e-01 -1.01946974e+00
-4.11616504e-01 -1.00474203e+00 8.43408763e-01 2.06632301e-01
-4.88157064e-01 -1.06465137e+00 1.59157634e-01 3.99457544e-01
3.56389225e-01 3.26290667e-01 7.06893325e-01 -1.30349874e+00
2.57981122e-01 -3.31313908e-01 -8.61655101e-02 1.11245203e+00
2.63829380e-01 -4.40901905e-01 -1.46592009e+00 -2.86735088e-01
1.94330156e-01 -3.33405137e-01 1.26546311e+00 2.45775312e-01
1.22842526e+00 2.30349172e-02 -5.57092369e-01 2.41729006e-01
1.04925060e+00 3.00896704e-01 5.06544471e-01 1.79531991e-01
3.27497512e-01 6.02666378e-01 4.39218968e-01 2.92319804e-01
1.68972895e-01 5.05675197e-01 2.95413345e-01 -1.95112064e-01
-4.38533157e-01 1.10877283e-01 6.97656125e-02 4.51585025e-01
-1.13270715e-01 2.40348447e-02 -1.08201063e+00 4.19239312e-01
-1.44545174e+00 -4.33742523e-01 -7.21078813e-02 2.41010523e+00
7.92100430e-01 2.54195482e-01 2.31880113e-01 5.85262239e-01
8.72601330e-01 -3.04096937e-01 -5.10962427e-01 -2.59455472e-01
1.35818139e-01 4.99012589e-01 3.04088950e-01 3.90120745e-01
-1.31317401e+00 4.41331834e-01 6.37809944e+00 6.36263192e-01
-1.20007610e+00 2.58547455e-01 9.56105709e-01 -1.87195092e-02
3.32073480e-01 -4.70272571e-01 -7.52699554e-01 6.87155902e-01
1.12964904e+00 5.07311165e-01 5.94527014e-02 5.87035537e-01
-1.44050986e-01 -2.57888854e-01 -1.20595145e+00 1.04682243e+00
3.02023292e-01 -7.58928716e-01 -3.50857913e-01 2.74077635e-02
7.56196678e-01 -2.99078622e-03 9.34971571e-02 4.22929883e-01
2.25729980e-02 -1.35067308e+00 3.90815049e-01 7.29159832e-01
1.03979111e+00 -6.54414177e-01 1.13003814e+00 5.40203691e-01
-4.72401381e-01 -3.29104543e-01 -1.32393539e-01 -9.26428735e-02
-1.09423645e-01 9.14286375e-01 -1.21287227e+00 5.28673947e-01
7.38754809e-01 6.52361214e-01 -7.69212782e-01 1.12437952e+00
-2.28745818e-01 7.67647147e-01 -3.31199110e-01 2.84429248e-02
1.63693372e-02 2.36071959e-01 2.29063883e-01 1.31388974e+00
5.00231206e-01 9.45251808e-02 1.99824609e-02 5.56710660e-01
-1.40214682e-01 1.61231026e-01 -5.48672259e-01 3.17640334e-01
1.16578460e-01 1.24314404e+00 -6.42517984e-01 -6.31597519e-01
-1.41318396e-01 8.83963287e-01 3.93257976e-01 1.04780085e-01
-5.55107117e-01 -2.03648806e-01 7.94535801e-02 1.15031479e-02
2.40213707e-01 2.66401172e-01 -6.58880353e-01 -8.71737301e-01
-1.13575133e-02 -8.41587603e-01 5.90103388e-01 -4.24233615e-01
-1.36588848e+00 7.67252624e-01 -2.43731633e-01 -1.21640289e+00
-1.72319487e-01 -7.37935841e-01 -5.46319246e-01 8.90114546e-01
-1.22246897e+00 -6.52015090e-01 -5.73475420e-01 2.83086866e-01
2.82619447e-01 -2.02388152e-01 1.05484080e+00 4.50882912e-01
-6.23830497e-01 7.60239303e-01 2.05985367e-01 -2.36811955e-02
9.76494074e-01 -1.33972490e+00 -6.52180193e-03 5.60952723e-01
-1.93322267e-04 1.64668426e-01 5.69729447e-01 -4.84740913e-01
-5.76950163e-02 -1.26264966e+00 9.45590556e-01 -8.99654984e-01
1.16458490e-01 -5.54540008e-02 -9.87183511e-01 3.68305594e-01
-7.78381154e-02 2.84894723e-02 1.14053833e+00 2.21725613e-01
-2.57378727e-01 -2.82978415e-01 -1.50202584e+00 2.00781956e-01
1.03558969e+00 -2.28164166e-01 -6.77680612e-01 1.79663524e-01
2.05470785e-01 -3.02074105e-01 -9.21653211e-01 7.57068932e-01
5.37553430e-01 -1.12650967e+00 6.71329021e-01 -6.65557563e-01
3.47677827e-01 -5.55019118e-02 -1.82949752e-01 -1.46313560e+00
-2.51623303e-01 -6.92554563e-02 -4.64045769e-03 1.00858033e+00
6.63342655e-01 -7.50085890e-01 5.59569657e-01 3.31916034e-01
-3.74767005e-01 -9.07404721e-01 -9.99666035e-01 -5.14171183e-01
1.78027675e-01 -5.22742212e-01 1.72702730e-01 8.18214476e-01
-8.13905746e-02 4.29623842e-01 -2.46425271e-01 1.70282945e-01
6.23618007e-01 -2.37716153e-01 3.00314486e-01 -1.56874192e+00
-1.80503950e-01 -3.10317129e-01 -5.05693674e-01 -4.88759398e-01
4.96613234e-02 -1.11781001e+00 2.28475839e-01 -1.56747818e+00
2.70122588e-01 -6.18062198e-01 -7.84978092e-01 5.31187057e-01
-3.67758751e-01 6.83459759e-01 -1.14105530e-01 1.44851699e-01
-6.89821541e-01 -1.41794223e-03 9.19130445e-01 -6.27430528e-02
7.94595927e-02 4.29280311e-01 -9.64592040e-01 7.42214382e-01
1.22810364e+00 -6.17700398e-01 -1.22023880e-01 -5.43306470e-02
-7.56996721e-02 -2.34937802e-01 3.08082223e-01 -1.41504037e+00
-1.34107336e-01 4.43165958e-01 6.40053034e-01 -2.51395881e-01
2.62920499e-01 -7.68157601e-01 -1.48909122e-01 7.45233893e-01
-5.72636724e-01 -3.78385454e-01 2.22879097e-01 4.61929917e-01
-1.11656720e-02 -5.23404300e-01 9.56420004e-01 -8.64957571e-02
-3.86421859e-01 -9.16598365e-02 -2.55067527e-01 1.15995809e-01
9.78635669e-01 -7.41359815e-02 -1.68618187e-01 -2.94296890e-01
-1.63444710e+00 5.59478216e-02 2.31675789e-01 9.40525606e-02
3.33306670e-01 -1.04910886e+00 -8.34883690e-01 4.61891472e-01
2.58874357e-01 -4.49492186e-02 4.34855521e-01 1.09479654e+00
-1.73036218e-01 2.96491534e-01 -6.75107166e-02 -7.97427833e-01
-1.39738584e+00 5.51036716e-01 6.96491122e-01 -4.27789569e-01
-4.24126327e-01 6.39428616e-01 1.58136874e-01 -2.75155008e-01
4.37879294e-01 -5.45884907e-01 -5.24868190e-01 1.74207553e-01
6.07397437e-01 6.09094024e-01 5.63152015e-01 -3.85079771e-01
-3.69609654e-01 6.02233589e-01 -1.00981398e-02 1.89569861e-01
9.92135644e-01 -2.99384817e-02 -4.03740034e-02 6.63986325e-01
1.12583709e+00 -9.97610539e-02 -9.39019918e-01 2.12300383e-02
8.41868371e-02 2.38416102e-02 -1.13992818e-01 -1.33955336e+00
-7.45689452e-01 1.05856919e+00 1.21467376e+00 4.53440249e-01
1.13000488e+00 1.86385855e-01 3.26984882e-01 3.31740022e-01
1.20003238e-01 -9.42794383e-01 -4.06060480e-02 3.14600915e-01
6.53027833e-01 -1.47839797e+00 -3.17592889e-01 -1.47531554e-01
-5.88341355e-01 1.07566786e+00 3.83155107e-01 -2.05132971e-03
5.90013504e-01 2.04152092e-01 3.32064480e-01 -6.16115779e-02
-4.33920205e-01 -2.48828754e-01 4.91074562e-01 6.87086582e-01
3.41739386e-01 1.98669299e-01 -3.48960280e-01 8.37631166e-01
-6.61712736e-02 1.32740363e-01 2.41772935e-01 4.75728959e-01
-6.71848714e-01 -1.04876661e+00 -3.81767541e-01 9.15860295e-01
-6.75165951e-01 -2.30681598e-02 -3.49585176e-01 4.68855917e-01
5.99535525e-01 1.08103728e+00 2.22036913e-01 -4.90569919e-01
4.19532299e-01 7.07319796e-01 5.04525602e-01 -8.28079462e-01
-5.00758588e-01 -2.49512613e-01 4.36674863e-01 -1.97634652e-01
-5.13966918e-01 -7.57621408e-01 -1.02301967e+00 2.26434276e-01
-1.52693704e-01 2.29405627e-01 5.12253225e-01 1.02207196e+00
1.06342241e-01 8.79939735e-01 5.01142323e-01 -5.12917995e-01
-7.96020389e-01 -1.31343603e+00 -4.78529364e-01 4.76712406e-01
6.80077255e-01 -7.03278601e-01 -7.41668105e-01 -5.36349900e-02]
|
[9.931060791015625, 3.453364849090576]
|
ee6e4a05-c31b-44b9-ba6d-eec1a3b0fe55
|
column-networks-for-collective-classification
|
1609.04508
| null |
http://arxiv.org/abs/1609.04508v2
|
http://arxiv.org/pdf/1609.04508v2.pdf
|
Column Networks for Collective Classification
|
Relational learning deals with data that are characterized by relational
structures. An important task is collective classification, which is to jointly
classify networked objects. While it holds a great promise to produce a better
accuracy than non-collective classifiers, collective classification is
computational challenging and has not leveraged on the recent breakthroughs of
deep learning. We present Column Network (CLN), a novel deep learning model for
collective classification in multi-relational domains. CLN has many desirable
theoretical properties: (i) it encodes multi-relations between any two
instances; (ii) it is deep and compact, allowing complex functions to be
approximated at the network level with a small set of free parameters; (iii)
local and relational features are learned simultaneously; (iv) long-range,
higher-order dependencies between instances are supported naturally; and (v)
crucially, learning and inference are efficient, linear in the size of the
network and the number of relations. We evaluate CLN on multiple real-world
applications: (a) delay prediction in software projects, (b) PubMed Diabetes
publication classification and (c) film genre classification. In all
applications, CLN demonstrates a higher accuracy than state-of-the-art rivals.
|
['Truyen Tran', 'Dinh Phung', 'Trang Pham', 'Svetha Venkatesh']
|
2016-09-15
| null | null | null | null |
['genre-classification']
|
['computer-vision']
|
[-6.99475333e-02 3.92941982e-02 -7.01880991e-01 -4.27667141e-01
-5.06245732e-01 -4.71653193e-01 6.46910250e-01 9.19589579e-01
3.40397768e-02 7.63787985e-01 1.47569403e-01 -4.82631534e-01
-9.65194941e-01 -1.04763627e+00 -9.04105604e-01 -4.79235679e-01
-6.50046289e-01 5.70716023e-01 3.43798131e-01 -7.97467679e-03
1.09804899e-01 7.01818407e-01 -1.48333848e+00 5.76959789e-01
4.29531008e-01 1.47460401e+00 -2.78326452e-01 6.29626930e-01
2.34345887e-02 1.62381947e+00 -4.96620625e-01 -3.94815207e-01
-1.16131298e-01 -7.60841668e-02 -1.23663282e+00 -2.74485588e-01
3.92176807e-01 7.00024888e-02 -5.03285825e-01 6.44233108e-01
2.10778341e-01 5.60001805e-02 7.25317776e-01 -1.35474849e+00
-1.00696683e+00 9.17417049e-01 -3.69911671e-01 3.11681598e-01
9.51967761e-02 -1.35589972e-01 1.76161146e+00 -2.92814076e-01
7.77173579e-01 1.26354408e+00 9.13136721e-01 -3.44770439e-02
-1.57946706e+00 -5.66452265e-01 1.74579158e-01 -4.04979996e-02
-1.31004119e+00 -2.69552767e-01 2.60183811e-01 -4.47394818e-01
1.40086794e+00 1.78568125e-01 5.96454024e-01 8.97638142e-01
6.00219548e-01 4.44598287e-01 6.58599377e-01 3.40467803e-02
1.49813846e-01 -2.67448425e-01 2.62213081e-01 8.19781542e-01
2.13103667e-01 -3.71325493e-01 -7.97328591e-01 -3.41580331e-01
7.46438622e-01 4.03555572e-01 1.92116022e-01 -3.25399684e-04
-1.52879596e+00 6.51967347e-01 6.50926292e-01 4.32996809e-01
-3.12469691e-01 6.29767239e-01 7.03178465e-01 8.24854672e-01
8.68415654e-01 5.60620129e-01 -8.72409582e-01 -1.66414112e-01
-2.69083411e-01 2.09924221e-01 1.03971994e+00 9.80210006e-01
4.81978685e-01 -4.84949946e-01 -6.40733987e-02 8.13537598e-01
1.28498584e-01 7.26461709e-02 5.08425161e-02 -8.04795802e-01
6.30572796e-01 1.06011331e+00 -4.85029072e-01 -1.27590942e+00
-6.47114754e-01 -6.16572559e-01 -1.33754575e+00 -3.84882540e-01
2.20465243e-01 3.84209715e-02 -2.49767005e-01 1.53940940e+00
1.15218885e-01 5.06553315e-02 1.46619409e-01 1.61300376e-01
1.09409857e+00 3.78402829e-01 -2.32263971e-02 -2.68557638e-01
1.20973682e+00 -7.62384593e-01 -6.12691700e-01 1.09463986e-02
9.63350296e-01 -4.03821796e-01 4.42530930e-01 4.48866963e-01
-9.15946424e-01 -3.80248845e-01 -8.23775411e-01 -3.46456200e-01
-5.36939561e-01 -3.10733706e-01 1.48538816e+00 -7.92692881e-03
-1.10217035e+00 7.95072436e-01 -6.44805789e-01 -1.44998267e-01
8.99635732e-01 8.12898219e-01 -7.29631603e-01 -7.93950632e-02
-1.08033490e+00 6.52262270e-01 1.06109036e-02 -3.14211622e-02
-4.51043457e-01 -9.67414677e-01 -5.31642497e-01 2.67489642e-01
5.40628910e-01 -6.98136866e-01 8.02436173e-01 -4.94663566e-01
-9.81565356e-01 7.02791154e-01 1.20129548e-01 -4.16713834e-01
2.23821536e-01 -1.96941882e-01 -3.96993756e-01 6.71337396e-02
8.36201385e-02 3.03076774e-01 3.45794320e-01 -7.30005562e-01
-5.65478444e-01 -3.30606878e-01 1.57959133e-01 7.23919459e-03
-5.89221776e-01 -1.22915685e-01 -3.08251113e-01 -5.85121810e-01
2.41074055e-01 -7.76329577e-01 -2.38153905e-01 4.62417714e-02
-6.34785473e-01 -1.07235420e+00 8.88751030e-01 8.64346325e-02
1.14829600e+00 -2.10900092e+00 3.73467922e-01 2.10837767e-01
9.80517924e-01 -1.94329768e-01 -1.03482708e-01 8.24784756e-01
-5.54147735e-02 3.88808161e-01 3.61144930e-01 -4.38300639e-01
-2.37670973e-01 2.86050886e-01 -2.75413483e-01 6.57806396e-01
2.92364091e-01 1.03801107e+00 -1.00515354e+00 -5.42154551e-01
-3.43443155e-01 2.57957697e-01 -5.39684832e-01 -4.58949804e-02
-3.37563366e-01 2.99875997e-02 -3.64121825e-01 8.95313859e-01
9.77291688e-02 -9.75809574e-01 5.31667471e-01 -4.37582374e-01
1.87926531e-01 5.60021520e-01 -9.48034823e-01 1.36592615e+00
-3.52488697e-01 8.05779338e-01 -2.57879913e-01 -1.33814538e+00
9.25321996e-01 3.41932684e-01 9.15010750e-01 -4.39464390e-01
-2.52748311e-01 2.16601267e-01 1.41901761e-01 -4.30584967e-01
1.69791028e-01 1.38442189e-01 -1.36737272e-01 7.68879890e-01
3.52096587e-01 5.99581711e-02 2.68519640e-01 4.71842766e-01
1.70208931e+00 -5.35754204e-01 2.15259239e-01 -4.17910725e-01
-2.33376231e-02 -6.16082430e-01 4.46366876e-01 9.08925712e-01
3.21198374e-01 2.03824155e-02 1.26938081e+00 -6.68908715e-01
-6.31422818e-01 -7.83430278e-01 -2.60799438e-01 1.33118975e+00
1.21643126e-01 -7.11048007e-01 -4.07646876e-03 -6.97428882e-01
5.53922534e-01 -1.46945789e-01 -8.14344883e-01 -7.74143785e-02
-3.19514692e-01 -6.72613442e-01 5.60449719e-01 7.77000129e-01
2.61971027e-01 -9.74426568e-01 -3.87473553e-02 1.58256426e-01
1.96490437e-01 -1.26067615e+00 -1.65351182e-01 7.29616225e-01
-1.13270509e+00 -1.27996600e+00 2.01458260e-01 -8.22527647e-01
3.35979491e-01 2.25920111e-01 1.50721991e+00 5.21213591e-01
-2.75004387e-01 3.61591548e-01 -1.79950684e-01 -2.80374736e-01
-1.54046193e-01 6.00491047e-01 2.23255605e-01 -1.32593006e-01
2.63645768e-01 -9.46528018e-01 -2.87294179e-01 2.07472205e-01
-7.60790169e-01 -1.84296682e-01 6.47953868e-01 8.46171558e-01
7.22112894e-01 5.21161973e-01 9.34532523e-01 -1.21183562e+00
7.07896709e-01 -8.62025917e-01 -4.11208063e-01 3.78652930e-01
-5.28781295e-01 -1.33895189e-01 6.31811380e-01 -5.80785275e-01
-3.44848663e-01 -3.77574176e-01 3.91408414e-01 -1.14165924e-01
9.54496786e-02 9.34425652e-01 1.97312966e-01 1.36743218e-01
6.36503816e-01 -2.24351734e-01 1.28282428e-01 -3.75979751e-01
3.83646935e-01 4.92115289e-01 1.65550157e-01 -7.89886475e-01
2.76188582e-01 2.81903535e-01 7.05202281e-01 -9.25416708e-01
-1.11201906e+00 -2.28738472e-01 -9.64301527e-01 1.46758612e-02
7.36842573e-01 -9.38421249e-01 -1.40580940e+00 2.95110881e-01
-1.15973747e+00 -4.89714354e-01 -2.53262401e-01 2.46522516e-01
-3.02763164e-01 -1.17065988e-01 -9.82556522e-01 -5.11679590e-01
-2.79582381e-01 -9.57987607e-01 9.83855188e-01 -1.89636186e-01
-3.48567754e-01 -1.43404400e+00 -3.68442059e-01 1.99367806e-01
3.52645069e-01 4.85262066e-01 1.47147775e+00 -1.00084031e+00
-8.72262597e-01 -2.77299315e-01 -4.20942903e-01 2.74347216e-01
3.15262020e-01 3.10543060e-01 -4.90403980e-01 -1.46864355e-01
-5.72471201e-01 -7.61114299e-01 7.45840788e-01 1.42340586e-01
1.58294857e+00 -3.15032542e-01 -6.35528386e-01 5.80158114e-01
1.11068749e+00 -1.28391296e-01 1.87666923e-01 -6.34968504e-02
9.10963595e-01 5.10392666e-01 1.73211396e-01 3.54563087e-01
6.13885403e-01 4.03878570e-01 6.28746629e-01 -3.52648869e-02
7.49364272e-02 3.61386733e-03 2.29349136e-02 1.02596998e+00
-1.96225479e-01 -3.87190908e-01 -1.10036087e+00 3.10911536e-01
-2.18520856e+00 -9.79931355e-01 -4.19647843e-01 1.60875535e+00
9.39915419e-01 5.13773739e-01 -3.37853171e-02 1.51879758e-01
2.68682629e-01 3.08738142e-01 -7.85785556e-01 -7.96880946e-02
-3.04325551e-01 7.22211674e-02 2.48187736e-01 4.94938530e-02
-1.26250362e+00 7.03348577e-01 6.34294128e+00 8.40838313e-01
-9.43626821e-01 8.51132274e-02 9.60220575e-01 -1.49760887e-01
-2.26867408e-01 -2.82534122e-01 -7.63758600e-01 1.90499067e-01
9.12197948e-01 1.68095261e-01 3.93956393e-01 6.62777901e-01
-6.12343669e-01 5.53540997e-02 -1.89759398e+00 7.26521552e-01
-4.15920854e-01 -2.09337926e+00 -3.64639945e-02 2.79138207e-01
7.18621492e-01 4.08352971e-01 1.67105183e-01 3.48789245e-01
6.10410452e-01 -1.40761220e+00 4.75800663e-01 7.35440731e-01
8.99795532e-01 -8.29532683e-01 7.86802709e-01 3.59895647e-01
-1.15815425e+00 -3.14392537e-01 -4.18029070e-01 -1.67470261e-01
-5.09438157e-01 8.95855069e-01 -3.52362722e-01 5.45275509e-01
9.23044503e-01 1.51369143e+00 -7.18703270e-01 5.40861666e-01
1.94064334e-01 5.22542179e-01 -4.88988429e-01 -3.69967937e-01
8.36173519e-02 7.71727338e-02 6.51930869e-02 1.17666304e+00
-9.29767340e-02 1.62618235e-01 1.92764595e-01 6.37425065e-01
-6.88852310e-01 -9.96090323e-02 -8.44635010e-01 -4.19318974e-01
6.47816479e-01 1.15530598e+00 -9.59138811e-01 -7.30690584e-02
-4.62995380e-01 3.84438932e-01 8.75376761e-01 1.60901383e-01
-4.36033905e-01 -3.07408273e-01 7.22775519e-01 8.23598802e-02
2.41880089e-01 -1.02746546e-01 -2.81863451e-01 -8.31870735e-01
-7.75086135e-02 -7.34366417e-01 4.64191467e-01 -3.76295745e-01
-1.97691059e+00 7.07790077e-01 -2.50019878e-01 -9.87209201e-01
-5.56051955e-02 -6.46409154e-01 -2.41057172e-01 5.37662685e-01
-8.39802325e-01 -1.37125802e+00 2.84614749e-02 5.00645220e-01
3.38466465e-02 -4.73264992e-01 1.07958317e+00 3.50616813e-01
-5.86722374e-01 5.77858806e-01 9.70829800e-02 2.74104744e-01
5.01055300e-01 -1.31113374e+00 1.68865114e-01 7.97024518e-02
1.43803045e-01 1.05771315e+00 2.56303966e-01 -5.39251208e-01
-1.72640443e+00 -1.11846530e+00 9.41965878e-01 -8.47446680e-01
1.23906207e+00 -6.20616198e-01 -9.12851810e-01 8.58608842e-01
4.80814977e-03 8.95031214e-01 1.14487028e+00 1.12051392e+00
-7.42489100e-01 -6.31722808e-01 -7.43300498e-01 2.09901631e-01
1.27433002e+00 -8.17498505e-01 2.80402843e-02 9.56351280e-01
1.14607918e+00 -3.20044547e-01 -1.79903924e+00 3.01674992e-01
5.70987046e-01 -8.52228045e-01 9.89579260e-01 -1.03010488e+00
9.85387802e-01 1.49285451e-01 -3.15995455e-01 -8.59807312e-01
-4.36971217e-01 -5.95940709e-01 -7.80485630e-01 1.31255555e+00
5.77093840e-01 -6.85023844e-01 4.18637991e-01 3.21709186e-01
-5.76103963e-02 -1.67692351e+00 -8.18793058e-01 -6.45772696e-01
-5.10328189e-02 -4.06493992e-01 6.52251065e-01 1.22942269e+00
1.08359002e-01 9.52861369e-01 -2.00587630e-01 -4.01297174e-02
3.80624115e-01 2.05976993e-01 6.79594219e-01 -1.69950235e+00
-4.58245695e-01 -5.87373972e-01 -4.76045549e-01 -9.69426274e-01
3.54545712e-01 -1.00400305e+00 -6.08275831e-01 -1.57078588e+00
2.02065259e-01 -1.10507572e+00 -5.03284633e-01 7.96980917e-01
1.43685907e-01 7.28505626e-02 -1.79784030e-01 4.98188883e-01
-1.32361531e+00 2.53235400e-01 1.11117232e+00 -1.62495658e-01
-6.62891716e-02 4.03468132e-01 -8.99254918e-01 3.59545678e-01
3.69194210e-01 -4.54919755e-01 -5.81859469e-01 -3.12977642e-01
1.09579492e+00 4.82210070e-01 2.13797599e-01 -6.82453275e-01
6.43102109e-01 -9.24225450e-02 3.41793150e-01 -6.64885461e-01
1.46496698e-01 -7.54366755e-01 4.62326594e-02 2.69010603e-01
-7.30875313e-01 1.98681593e-01 -3.40683758e-02 9.53275561e-01
-2.10649028e-01 2.79380113e-01 4.29001719e-01 1.76048800e-01
-1.97315469e-01 5.67556620e-01 -7.97144398e-02 5.28203733e-02
9.05581772e-01 4.12307739e-01 -8.44416678e-01 -2.63628453e-01
-4.15891498e-01 3.03655356e-01 -3.91725712e-02 5.21254241e-01
3.87737483e-01 -1.29500771e+00 -3.95401865e-01 -8.34003389e-02
9.80586261e-02 5.73436141e-01 6.52142465e-02 1.03014541e+00
-2.87494421e-01 3.40777814e-01 4.02177095e-01 -8.29543054e-01
-1.17834270e+00 6.69967711e-01 1.78843364e-01 -6.40417099e-01
-7.40576029e-01 1.09627652e+00 -2.60043442e-01 -4.86561149e-01
4.71816838e-01 -4.38316435e-01 -8.78914297e-02 4.56769317e-01
2.48907983e-01 2.70403117e-01 1.92578197e-01 -2.66541272e-01
-5.38354218e-01 1.39406294e-01 -2.98119962e-01 6.91546202e-01
2.10008335e+00 3.36271763e-01 -9.93722558e-01 8.55351329e-01
1.42642009e+00 -1.69747904e-01 -8.63023520e-01 -5.87367773e-01
2.19248429e-01 5.96079864e-02 -2.46886797e-02 -6.33733690e-01
-1.05728388e+00 6.88484609e-01 -6.06849566e-02 9.69217002e-01
7.81339467e-01 4.37468410e-01 3.68672192e-01 8.65259290e-01
2.46279836e-01 -8.12938154e-01 5.44668972e-01 7.37165272e-01
6.98692858e-01 -1.15566826e+00 5.25805771e-01 -3.55349302e-01
-2.60401905e-01 1.31411624e+00 5.38107514e-01 -2.49599908e-02
1.30460536e+00 5.14324546e-01 -5.74846864e-01 -7.81074107e-01
-1.55246794e+00 3.05896074e-01 4.21639800e-01 2.18331546e-01
7.59322524e-01 -6.36111796e-02 2.84815222e-01 6.61678255e-01
-4.68720607e-02 -2.33314350e-01 1.27936140e-01 7.07270622e-01
-1.09107010e-01 -7.88453698e-01 2.89517969e-01 1.16979837e+00
-6.00308895e-01 -1.52675003e-01 -2.50152558e-01 8.24264705e-01
2.73438245e-01 8.24149668e-01 3.70295942e-01 -6.20819092e-01
1.13717072e-01 -4.86803919e-01 3.50569695e-01 -9.66292143e-01
-7.20231235e-01 -2.31839418e-01 2.67791808e-01 -6.37214422e-01
-4.97978300e-01 -6.19638741e-01 -9.77481663e-01 -7.10900605e-01
-3.56334418e-01 2.84538008e-02 2.21021861e-01 1.01404691e+00
5.50343215e-01 9.45621371e-01 3.43691558e-01 -4.97520953e-01
-5.07381499e-01 -9.45425510e-01 -7.06575453e-01 1.93493530e-01
4.81467277e-01 -5.15825510e-01 -2.67143011e-01 -2.46615648e-01]
|
[7.127668380737305, 6.323044776916504]
|
e5c180e9-7b1d-48cd-881e-2a91d75397e1
|
causal-razors
|
2302.10331
| null |
https://arxiv.org/abs/2302.10331v2
|
https://arxiv.org/pdf/2302.10331v2.pdf
|
Causal Razors
|
When performing causal discovery, assumptions have to be made on how the true causal mechanism corresponds to the underlying joint probability distribution. These assumptions are labeled as causal razors in this work. We review numerous causal razors that appeared in the literature, and offer a comprehensive logical comparison of them. In particular, we scrutinize an unpopular causal razor, namely parameter minimality, in multinomial causal models and its logical relations with other well-studied causal razors. Our logical result poses a dilemma in selecting a reasonable scoring criterion for score-based casual search algorithms.
|
['Wai-Yin Lam']
|
2023-02-20
| null | null | null | null |
['causal-discovery']
|
['knowledge-base']
|
[ 2.27043957e-01 2.28177175e-01 -8.58222961e-01 -3.59652817e-01
-2.44713470e-01 -6.99568391e-01 9.18313622e-01 2.02111110e-01
4.93720500e-03 1.04112720e+00 5.99893212e-01 -8.67626071e-01
-1.04066467e+00 -6.41432166e-01 -5.55108190e-01 -5.11521876e-01
-4.84658509e-01 6.04708552e-01 1.47292361e-01 4.68799099e-02
6.57762110e-01 3.31497729e-01 -9.42333400e-01 8.06855485e-02
9.95082438e-01 1.97778866e-01 -3.58336568e-01 5.41871369e-01
4.40745980e-01 1.10508955e+00 -5.91543674e-01 -7.81278431e-01
-2.32387170e-01 -7.17539370e-01 -1.10671616e+00 -7.83395946e-01
4.83491570e-01 -2.91762650e-01 -5.69349885e-01 1.10585201e+00
3.09391290e-01 -2.71371484e-01 1.03031278e+00 -1.56801581e+00
-8.13643992e-01 1.23516798e+00 -5.48907459e-01 6.69799566e-01
6.30519331e-01 1.12548053e-01 1.56199646e+00 -7.17999101e-01
5.54962575e-01 1.72940040e+00 6.29617751e-01 2.77182549e-01
-1.50570273e+00 -7.80190051e-01 2.46422470e-01 5.56431651e-01
-1.51027286e+00 -2.86567360e-01 3.95560652e-01 -5.06150961e-01
6.38069689e-01 6.61858261e-01 5.60579121e-01 1.27079344e+00
7.52555966e-01 2.91168094e-01 1.12388313e+00 -4.59586322e-01
2.70324975e-01 -2.49587879e-01 1.90298587e-01 6.95039690e-01
8.77872467e-01 5.08362830e-01 -9.60103273e-01 -1.04087222e+00
8.82798135e-01 -5.15908599e-01 -2.88056582e-01 -1.60016388e-01
-7.58162141e-01 1.19647503e+00 2.86025226e-01 1.62247881e-01
-2.28293523e-01 7.48134136e-01 2.02572793e-02 1.01058289e-01
-5.92480041e-02 7.21835434e-01 -3.89051646e-01 6.41930699e-02
-6.15255535e-01 5.13003886e-01 7.18004048e-01 4.52209383e-01
-1.24221422e-01 -1.78134292e-01 -4.98312488e-02 3.95045340e-01
5.56769252e-01 4.03400540e-01 -1.37944326e-01 -8.08000922e-01
1.38127971e-02 2.39651188e-01 2.73985267e-01 -1.23598766e+00
-4.62531924e-01 -3.27551872e-01 -7.75353014e-01 8.42681080e-02
4.55204874e-01 1.36499882e-01 -5.59927106e-01 1.81172669e+00
2.00445399e-01 2.69201398e-02 -3.97027194e-01 7.28702128e-01
6.45433366e-01 1.54769108e-01 5.70557714e-01 -5.73331475e-01
1.32213187e+00 -1.04074843e-01 -8.67234766e-01 -5.62186241e-02
2.50578439e-03 -7.58103251e-01 1.08029354e+00 5.53052247e-01
-9.05169547e-01 2.13951379e-01 -9.04205322e-01 3.60330850e-01
2.71094680e-01 -3.71293426e-01 1.21204925e+00 7.38973379e-01
-4.98403817e-01 6.96526110e-01 -5.67700446e-01 -3.40217769e-01
2.79861778e-01 1.72721431e-01 -1.69557873e-02 1.39647663e-01
-1.59405601e+00 1.05763602e+00 -1.56877209e-02 5.45081869e-02
-1.35317647e+00 -8.73687506e-01 -2.00285152e-01 1.52349491e-02
7.26530135e-01 -1.04540622e+00 1.20788074e+00 -6.17809966e-03
-5.83950520e-01 5.91502368e-01 -3.07056725e-01 -1.53390884e-01
5.08043647e-01 -3.94686341e-01 -6.22674584e-01 -7.77758583e-02
2.31462300e-01 -1.60193056e-01 6.48040533e-01 -1.43751788e+00
-5.56447864e-01 -1.28745779e-01 3.80154610e-01 -2.50141304e-02
2.00243101e-01 5.19412458e-01 5.93548976e-02 -7.93261707e-01
1.62995830e-01 -7.13954449e-01 -4.25256431e-01 -4.44801688e-01
-8.48377645e-01 -7.16041148e-01 -1.42328322e-01 -1.23187564e-01
1.89615119e+00 -1.57756352e+00 5.19727133e-02 3.66397500e-01
4.75590497e-01 -8.42329860e-01 2.79067278e-01 6.45792902e-01
-5.12937784e-01 8.72825444e-01 -1.15817785e-02 4.35121655e-01
6.06170334e-02 8.86569023e-02 -6.95960760e-01 8.85227323e-01
-7.15835765e-03 7.70312071e-01 -1.11188245e+00 -7.31297851e-01
-1.05326444e-01 -5.26989140e-02 -5.53236902e-01 7.62887001e-02
-8.69119763e-02 9.81917083e-02 -4.02250290e-01 7.06682682e-01
3.08647573e-01 -4.39012200e-01 6.45615399e-01 -9.59415585e-02
-3.81410986e-01 9.00661349e-01 -1.03105938e+00 6.67954266e-01
1.62516594e-01 3.02040756e-01 -2.14490920e-01 -4.69412357e-01
3.10433984e-01 4.58638161e-01 1.92245960e-01 -2.06230819e-01
-1.64348911e-03 2.35789359e-01 3.29328656e-01 -4.12534326e-01
1.45383894e-01 -7.22243845e-01 -3.07572484e-01 5.77232540e-01
-1.10875711e-01 -4.00805533e-01 1.99745279e-02 4.66168433e-01
1.27156878e+00 -2.86468863e-01 8.40962112e-01 -7.47566998e-01
-6.09664097e-02 2.96717703e-01 7.72082746e-01 1.35804439e+00
6.87685311e-02 2.55627692e-01 8.34152222e-01 -4.33335513e-01
-7.26111710e-01 -1.42854679e+00 -6.13850296e-01 8.64684939e-01
1.62053853e-01 -7.26514220e-01 -1.41396999e-01 -7.19493628e-01
2.18297951e-02 1.17655778e+00 -1.11448133e+00 -2.60958701e-01
-4.24429268e-01 -1.29956293e+00 9.21064436e-01 2.68853545e-01
-4.26509589e-01 -5.30530393e-01 -6.41133010e-01 -6.72847182e-02
-1.48515701e-01 -2.44478539e-01 -7.81854242e-02 5.96726596e-01
-8.12111735e-01 -1.57683456e+00 1.59138843e-01 1.95252776e-01
3.54859650e-01 -4.74434569e-02 1.33665729e+00 2.38460407e-01
-2.36336380e-01 4.69858907e-02 -1.31006837e-01 -3.91751915e-01
-3.02987278e-01 -3.89831841e-01 3.84121865e-01 -8.14245343e-01
2.97413558e-01 -5.32720447e-01 -2.59318441e-01 3.47497851e-01
-6.10552549e-01 -4.53593165e-01 3.85951787e-01 8.32984447e-01
5.66230938e-02 5.86461008e-01 4.18490052e-01 -9.67108548e-01
8.23591650e-01 -8.97635639e-01 -4.55793470e-01 3.12627822e-01
-1.00240421e+00 4.74840291e-02 1.88300475e-01 -2.45980918e-01
-1.11586511e+00 -6.03215754e-01 1.44474223e-01 1.26863718e-01
6.97000325e-02 8.13907027e-01 -1.05912320e-01 2.40474671e-01
1.05228710e+00 -5.54625034e-01 -7.14643717e-01 -5.48197329e-01
4.59367305e-01 -1.13648009e-02 4.36617106e-01 -8.09669256e-01
8.14538062e-01 5.39287806e-01 2.19974712e-01 -3.14289391e-01
-1.03746641e+00 8.77765939e-03 -5.15284598e-01 -2.42285859e-02
5.62857091e-01 -4.42318559e-01 -1.04888809e+00 -2.08289981e-01
-1.37680137e+00 -1.82033285e-01 2.61645727e-02 6.65756106e-01
-3.98898363e-01 2.21826751e-02 -5.05743861e-01 -1.13403106e+00
2.45637059e-01 -7.50520766e-01 4.45251405e-01 2.11387742e-02
-1.05266392e+00 -1.11021602e+00 4.68455970e-01 2.41442043e-02
-2.17683557e-02 8.45746323e-02 1.48968780e+00 -4.47997183e-01
-2.38704517e-01 2.89532430e-02 -2.16096595e-01 -7.28132606e-01
1.18167058e-01 5.33694565e-01 -5.70655882e-01 3.09244752e-01
-1.25122249e-01 -1.53830335e-01 7.42577136e-01 7.05945909e-01
9.27574396e-01 -6.64533138e-01 -7.54496872e-01 1.40186727e-01
1.40443790e+00 3.12736183e-01 3.16517889e-01 1.76536366e-01
4.47902381e-01 7.76702523e-01 5.28835773e-01 4.87600595e-01
2.27399796e-01 5.94060898e-01 5.45787156e-01 1.99442178e-01
7.02378079e-02 -6.94244266e-01 3.26789841e-02 1.82551384e-01
-3.47011149e-01 -4.58208799e-01 -1.11927319e+00 4.65128809e-01
-1.85117686e+00 -1.18948519e+00 -9.29518104e-01 1.90911973e+00
1.22137177e+00 2.62897998e-01 1.89999744e-01 8.79974850e-03
5.77800214e-01 -1.03966305e-02 -2.99777716e-01 -3.70959044e-01
-3.56183410e-01 -2.39475723e-02 6.52882695e-01 7.97412455e-01
-9.75317121e-01 7.14579821e-01 9.24617195e+00 7.76619613e-01
-4.84854102e-01 2.53250360e-01 4.55729127e-01 -3.68372947e-01
-8.58723819e-01 4.68512625e-01 -5.69783151e-01 1.61268532e-01
9.03688192e-01 -5.17327905e-01 1.01109900e-01 5.38970411e-01
6.15937054e-01 -3.65901619e-01 -1.25136435e+00 3.66073817e-01
-4.20827568e-01 -1.32674146e+00 2.46221885e-01 9.06185210e-02
7.09361017e-01 -3.80266160e-01 -1.44241497e-01 -3.79250824e-01
1.36017394e+00 -1.76508176e+00 9.45411682e-01 5.08535028e-01
7.67003834e-01 -5.97175956e-01 6.29064620e-01 -1.92307800e-01
-5.72194338e-01 -2.87443489e-01 -2.25260198e-01 -5.25265098e-01
5.28113008e-01 1.07114303e+00 -6.22520864e-01 4.62130129e-01
8.54406416e-01 3.14879388e-01 -3.56059641e-01 1.12080169e+00
-9.20714021e-01 1.26642942e+00 -2.33019218e-01 -3.79478857e-02
-3.05730365e-02 2.76562601e-01 7.06953108e-01 1.11953878e+00
-1.26716644e-01 5.99750459e-01 -3.00833881e-01 1.17587626e+00
2.91226834e-01 -2.17740819e-01 -4.64464247e-01 -8.76912773e-02
9.43176210e-01 5.64924359e-01 -6.41346812e-01 -9.97070670e-02
-7.66374618e-02 2.45383903e-01 3.39708664e-03 1.17374077e-01
-1.00658393e+00 2.76727349e-01 7.48470306e-01 6.97997138e-02
-4.96922314e-01 -6.67678863e-02 -1.05131197e+00 -8.23346496e-01
-5.52830517e-01 -9.16639268e-01 9.65510964e-01 -7.80962706e-01
-1.60927582e+00 8.32511559e-02 5.80124795e-01 -4.92950857e-01
-1.80498622e-02 -2.27423310e-01 -7.83043444e-01 8.29171777e-01
-7.03964055e-01 -6.33615971e-01 1.45071879e-01 3.54450315e-01
1.35706946e-01 3.04551393e-01 7.22301483e-01 9.50855687e-02
-6.16466343e-01 3.54970485e-01 -5.62050402e-01 -5.13725758e-01
7.91008413e-01 -1.37024689e+00 2.07824230e-01 7.54299879e-01
-1.84347019e-01 1.35550165e+00 1.49012148e+00 -1.20917118e+00
-1.24779618e+00 -4.30214375e-01 1.23743784e+00 -8.01992893e-01
1.21016324e+00 1.19645290e-01 -6.21619284e-01 7.56982207e-01
1.57122508e-01 -7.55035281e-01 7.95274734e-01 9.76289272e-01
-6.65760159e-01 4.20593619e-01 -7.99847960e-01 9.10191357e-01
1.48711395e+00 -1.52810290e-01 -1.07530332e+00 1.93981275e-01
5.51771104e-01 2.86228448e-01 -6.22307003e-01 5.94022810e-01
8.43629479e-01 -7.81377196e-01 1.16838682e+00 -1.12087226e+00
7.70163178e-01 -3.19819897e-01 -1.43954486e-01 -1.14154100e+00
-9.33725774e-01 -4.92935449e-01 2.62615323e-01 8.60081971e-01
6.04479671e-01 -3.98158073e-01 3.38310838e-01 7.70540595e-01
2.69627571e-01 -6.09937906e-01 -9.67389584e-01 -6.70625627e-01
2.24812642e-01 -5.72205961e-01 3.98423374e-01 1.47702706e+00
5.58036208e-01 6.08599126e-01 -5.01837790e-01 3.77069741e-01
1.11764979e+00 1.29734710e-01 3.11583102e-01 -1.59382594e+00
-2.65241861e-01 -8.02943289e-01 1.86191693e-01 -5.71846187e-01
-1.72487363e-01 -5.91172695e-01 -1.83572575e-01 -1.43088508e+00
7.85524905e-01 -7.90838063e-01 -9.97710079e-02 5.99555433e-01
-4.18407708e-01 1.88414514e-01 -2.05818996e-01 4.33908463e-01
-2.51196295e-01 3.58349800e-01 8.85584891e-01 -1.16828099e-01
-1.18356712e-01 -1.93426937e-01 -1.16483104e+00 1.20649242e+00
7.94491529e-01 -9.82415974e-01 -3.94779086e-01 -9.75321755e-02
1.20189953e+00 3.25807482e-01 8.71401310e-01 -2.84775030e-02
2.22987369e-01 -1.25160623e+00 1.02791734e-01 -5.35376430e-01
-1.45399317e-01 -3.58383745e-01 6.75323904e-01 6.62415028e-01
-5.72132111e-01 6.36865795e-02 5.15178442e-02 4.82456475e-01
1.50307089e-01 -5.59289098e-01 5.65450668e-01 8.97510722e-03
-4.37486589e-01 -3.36924732e-01 -5.23772299e-01 -5.56675121e-02
4.77404773e-01 2.32301742e-01 -6.23784900e-01 -3.15204948e-01
-6.79493368e-01 8.77098218e-02 1.92495614e-01 3.70205939e-01
4.58395541e-01 -1.20706391e+00 -9.92253602e-01 -7.95820296e-01
1.40103757e-01 -7.19279051e-01 1.14846654e-01 1.11032283e+00
-1.04497649e-01 6.58087373e-01 1.30911320e-01 -1.02625191e-01
-1.17918682e+00 5.92933238e-01 2.03688949e-01 -1.21853799e-01
-2.71400124e-01 1.17043197e+00 5.27105927e-01 -1.57255884e-02
-1.15270559e-02 1.18844889e-01 -9.65319425e-02 4.34315465e-02
6.07702196e-01 7.76115954e-01 -3.81813288e-01 3.04002482e-02
-1.02042603e+00 1.54979490e-02 4.59902406e-01 -3.00433129e-01
1.29790974e+00 -2.14361697e-01 -9.32452083e-01 8.30760717e-01
4.25735921e-01 6.85410261e-01 -5.74616134e-01 3.28041613e-01
2.11018458e-01 -7.08314955e-01 -4.54521179e-02 -1.32851291e+00
-2.94714004e-01 2.04634532e-01 -1.59655899e-01 4.52499330e-01
6.39825225e-01 4.79575664e-01 -3.56629312e-01 -1.45031899e-01
4.61507529e-01 -5.71143329e-01 -1.94637954e-01 2.82717288e-01
1.29928827e+00 -6.79003358e-01 6.11750782e-01 -8.18825662e-01
-2.30430484e-01 7.96758711e-01 3.10503542e-01 -1.09935462e-01
6.74810052e-01 2.88855284e-01 -3.71296048e-01 -8.88240635e-01
-1.16060293e+00 2.04864696e-01 4.80759621e-01 3.28448981e-01
7.20982671e-01 4.64314789e-01 -1.20046270e+00 9.40487027e-01
-5.73190749e-01 -2.07670271e-01 7.12092698e-01 2.98120499e-01
-2.53357500e-01 -8.61879230e-01 -7.72165895e-01 5.77565551e-01
-6.65552080e-01 -6.29513204e-01 -8.76812637e-01 7.75183737e-01
1.07848808e-01 1.50068939e+00 -1.68607250e-01 -3.13760042e-01
1.34162232e-01 -1.25858411e-01 5.40809453e-01 -4.59693879e-01
-3.30311030e-01 9.19307545e-02 5.96741855e-01 -6.87077343e-01
-2.59601057e-01 -8.71765196e-01 -9.92614865e-01 -7.75526583e-01
-5.38165092e-01 3.85288775e-01 2.16447622e-01 8.78699422e-01
-2.13479176e-01 4.39858705e-01 9.20833722e-02 -1.66770726e-01
-5.76914668e-01 -8.36858034e-01 -5.26168942e-01 -9.49130207e-02
1.32325932e-01 -1.23852420e+00 -6.74373388e-01 3.06081562e-03]
|
[8.018372535705566, 5.479950904846191]
|
16b2b305-39ca-4e2a-8d72-97546226bad0
|
deep-learning-for-survival-analysis-a-review
|
2305.14961
| null |
https://arxiv.org/abs/2305.14961v1
|
https://arxiv.org/pdf/2305.14961v1.pdf
|
Deep Learning for Survival Analysis: A Review
|
The influx of deep learning (DL) techniques into the field of survival analysis in recent years, coupled with the increasing availability of high-dimensional omics data and unstructured data like images or text, has led to substantial methodological progress; for instance, learning from such high-dimensional or unstructured data. Numerous modern DL-based survival methods have been developed since the mid-2010s; however, they often address only a small subset of scenarios in the time-to-event data setting - e.g., single-risk right-censored survival tasks - and neglect to incorporate more complex (and common) settings. Partially, this is due to a lack of exchange between experts in the respective fields. In this work, we provide a comprehensive systematic review of DL-based methods for time-to-event analysis, characterizing them according to both survival- and DL-related attributes. In doing so, we hope to provide a helpful overview to practitioners who are interested in DL techniques applicable to their specific use case as well as to enable researchers from both fields to identify directions for future investigation. We provide a detailed characterization of the methods included in this review as an open-source, interactive table: https://survival-org.github.io/DL4Survival. As this research area is advancing rapidly, we encourage the research community to contribute to keeping the information up to date.
|
['Andreas Bender', 'Raphael Sonabend', 'Philipp Kopper', 'Simon Wiegrebe']
|
2023-05-24
| null | null | null | null |
['survival-analysis']
|
['miscellaneous']
|
[-2.64816415e-02 -4.82021600e-01 -5.52823126e-01 -2.92423487e-01
-9.47406232e-01 -5.83375871e-01 3.27908367e-01 7.41713166e-01
-3.69504094e-01 9.28962588e-01 3.14924181e-01 -4.60283339e-01
-3.78734648e-01 -7.55253971e-01 -1.96038872e-01 -9.04594541e-01
-5.40592134e-01 5.57294130e-01 -2.18039423e-01 1.66905478e-01
-1.76468939e-01 8.08506787e-01 -1.50570023e+00 1.52267575e-01
6.88700974e-01 8.21920633e-01 -2.31572017e-02 5.39986789e-01
-1.11353159e-01 5.69075406e-01 -2.05736786e-01 -4.03423518e-01
-1.48191735e-01 -4.43024546e-01 -4.73905504e-01 -3.47785771e-01
-2.80519389e-02 -4.14252609e-01 -4.72759247e-01 4.74950582e-01
8.32074821e-01 -4.04851697e-02 8.13921332e-01 -1.18453646e+00
-5.71109891e-01 1.27839961e-03 -2.02227995e-01 3.39879900e-01
1.26861945e-01 1.25886902e-01 8.74038637e-01 -8.66544604e-01
6.27038121e-01 1.02607667e+00 8.24504972e-01 5.34064054e-01
-1.13145959e+00 -3.62046689e-01 -2.30990667e-02 2.62314349e-01
-1.32036090e+00 -3.84133905e-01 6.43488824e-01 -6.64189696e-01
7.43425667e-01 3.48270446e-01 8.41901660e-01 1.35864210e+00
6.01361275e-01 8.03987026e-01 1.16546035e+00 -1.83093756e-01
4.13136482e-01 -2.30380952e-01 1.01565935e-01 3.70078802e-01
4.37466681e-01 5.67434549e-01 -3.26196700e-01 -3.40568930e-01
4.75757062e-01 6.02950275e-01 -6.88189715e-02 -5.44041336e-01
-1.13243902e+00 1.02122486e+00 1.70568734e-01 4.25714970e-01
-3.43164384e-01 -1.49890095e-01 5.68299651e-01 3.67582530e-01
7.68040061e-01 2.11355067e-03 -7.68348694e-01 -9.62074548e-02
-1.13400054e+00 3.89998496e-01 9.14502621e-01 4.80812788e-01
2.04809248e-01 -1.43978640e-01 -4.17740420e-02 7.61117220e-01
1.67654693e-01 2.45283857e-01 2.79033095e-01 -6.36123061e-01
-8.61100405e-02 6.35406137e-01 1.29092224e-02 -7.14637101e-01
-9.95148480e-01 -5.84846139e-01 -1.33445024e+00 1.89843494e-02
5.85286140e-01 -2.14233041e-01 -7.23675013e-01 1.79467690e+00
3.68737847e-01 -1.39978692e-01 -3.87500748e-02 8.09644103e-01
9.25310016e-01 3.81353378e-01 3.11342955e-01 -4.35317397e-01
1.42028439e+00 -4.19238240e-01 -7.12612629e-01 -4.86020334e-02
9.45709467e-01 -4.07763779e-01 8.24370563e-01 3.12873185e-01
-8.17672431e-01 1.03032969e-01 -6.76198304e-01 -1.19155921e-01
-7.26124644e-01 -2.96214614e-02 8.65479112e-01 3.91272128e-01
-8.92407775e-01 6.43076062e-01 -1.08796179e+00 -1.10574353e+00
6.33645236e-01 1.54653668e-01 -3.55878353e-01 -2.89543778e-01
-1.45885372e+00 8.85149717e-01 2.81165075e-02 -4.56418358e-02
-1.09096706e+00 -9.03676867e-01 -5.98736763e-01 -8.45294148e-02
2.49453440e-01 -1.24525940e+00 9.15686011e-01 -6.77761972e-01
-8.25071216e-01 1.15004385e+00 1.28487432e-02 -2.77245134e-01
6.54461563e-01 -7.66402781e-02 -4.13447797e-01 -1.71950627e-02
-5.47696985e-02 2.16417789e-01 1.34075254e-01 -8.87122989e-01
-6.64765060e-01 -8.10155690e-01 -2.49874234e-01 5.64123392e-02
-4.15718287e-01 3.81586075e-01 -1.64865837e-01 -8.54561388e-01
-1.75003901e-01 -7.80078590e-01 -3.72339129e-01 3.38152111e-01
-1.77061275e-01 -2.19432205e-01 4.40156072e-01 -5.59104323e-01
1.27736950e+00 -1.94431627e+00 3.51577461e-01 -5.13240278e-01
4.32612717e-01 -7.40006492e-02 1.54302940e-01 9.76949930e-01
-2.23396942e-01 7.21028075e-02 -4.76585269e-01 -7.04325616e-01
-2.24650756e-01 -4.28106897e-02 1.69682428e-02 9.81481493e-01
-1.38916187e-02 8.79265070e-01 -1.00130033e+00 -4.00795758e-01
2.97605783e-01 6.68380141e-01 2.23466866e-02 2.02121869e-01
-2.97409836e-02 7.67374516e-01 -4.32522744e-01 1.09454346e+00
4.03925270e-01 -4.17650580e-01 8.77482072e-02 3.56988162e-01
-2.68590063e-01 -7.97929764e-02 -5.48888326e-01 1.63901091e+00
-2.81313211e-01 5.91066062e-01 6.03989065e-02 -1.10322857e+00
6.13942623e-01 4.06783223e-01 9.18705583e-01 -2.96922147e-01
2.40273133e-01 3.30751717e-01 -9.78833362e-02 -3.88309360e-01
-1.81621358e-01 -7.74006009e-01 -1.73805907e-01 2.52573580e-01
-2.96256924e-03 3.17564964e-01 -5.10125011e-02 1.32332966e-01
1.19875336e+00 -8.34789649e-02 7.01504648e-01 -2.28096709e-01
4.22314137e-01 5.89458048e-02 5.58817148e-01 7.41873860e-01
-5.03982186e-01 5.44232309e-01 5.54830492e-01 -8.12183678e-01
-8.47980917e-01 -9.68871534e-01 -8.25077116e-01 7.87351489e-01
-4.11146075e-01 -2.12631404e-01 -2.68089414e-01 -5.10018587e-01
2.65706956e-01 5.65442026e-01 -8.50614071e-01 -1.09995253e-01
-1.52943000e-01 -1.43997419e+00 6.07047260e-01 5.30521631e-01
-3.39631140e-02 -8.07598650e-01 -3.39712232e-01 2.71233737e-01
-2.10642777e-02 -5.18359482e-01 1.49152577e-01 3.79460335e-01
-1.14278662e+00 -1.07778513e+00 -1.23875082e+00 -5.58436990e-01
2.17190966e-01 -7.79274106e-02 1.15545118e+00 1.31335557e-01
-5.20529568e-01 2.48124555e-01 -4.13980097e-01 -5.02832890e-01
-1.75964117e-01 1.75945610e-01 2.00186908e-01 -2.19902679e-01
7.26798534e-01 -6.14234090e-01 -9.59754705e-01 3.41529325e-02
-1.04749990e+00 -1.06237538e-01 4.47636664e-01 8.02679956e-01
6.98374748e-01 -2.49202207e-01 1.06890249e+00 -7.01916277e-01
4.14556414e-01 -1.10446084e+00 -5.95331669e-01 7.81232789e-02
-9.92561936e-01 -3.83855283e-01 6.24410450e-01 -1.61296427e-01
-5.21253347e-01 -4.05097678e-02 -3.24088365e-01 -1.41595066e-01
-5.05338848e-01 1.12493682e+00 -1.60612036e-02 2.22775638e-01
4.67556924e-01 1.95515350e-01 4.46531996e-02 -6.11974895e-01
-3.09584104e-02 7.23787248e-01 7.98184574e-02 -2.61277437e-01
4.70627427e-01 6.77794158e-01 4.09861743e-01 -5.50388157e-01
-8.74552846e-01 -4.19039905e-01 -5.63036144e-01 -2.46397823e-01
8.16017509e-01 -8.76488328e-01 -4.82700497e-01 8.34366560e-01
-6.45112514e-01 -4.58939552e-01 -1.72668383e-01 5.90368092e-01
-6.04424834e-01 3.09425920e-01 -6.71401024e-01 -8.31137896e-01
-4.16040182e-01 -9.55724776e-01 1.07823634e+00 1.18145458e-01
-6.76635951e-02 -1.52660465e+00 4.85251337e-01 8.88994560e-02
5.13266981e-01 7.63452232e-01 1.21013093e+00 -8.45662355e-01
-1.51893631e-01 -7.52739966e-01 -8.37602541e-02 -5.11899665e-02
2.65868813e-01 -9.61715728e-02 -9.11623001e-01 -4.68416005e-01
-6.20563105e-02 -3.67941976e-01 9.59093392e-01 8.31555188e-01
1.27092040e+00 2.10147947e-01 -6.49816334e-01 7.46375680e-01
1.34880888e+00 2.35236287e-01 3.20982635e-01 3.68419498e-01
4.16007370e-01 9.27947462e-01 5.33182144e-01 6.32891178e-01
6.53003037e-01 7.06338882e-01 4.57670182e-01 -5.08429646e-01
-1.02979034e-01 4.58548777e-02 -3.87698673e-02 4.28987682e-01
5.88484630e-02 -6.39272034e-01 -1.30402172e+00 5.80268860e-01
-1.82584035e+00 -7.90814877e-01 -2.85763741e-01 2.54501486e+00
6.19474173e-01 -2.62997180e-01 3.25096428e-01 -5.18477894e-02
4.22138840e-01 1.41899601e-01 -8.20832491e-01 3.06890924e-02
-4.54722822e-01 -2.78972507e-01 3.99731129e-01 2.45118931e-01
-1.15770566e+00 5.66104829e-01 6.48705006e+00 6.20978534e-01
-1.19241834e+00 1.27500832e-01 1.00475812e+00 -4.17340189e-01
-1.55993551e-01 -1.24006132e-02 -5.41766644e-01 4.34018373e-01
1.24224854e+00 -4.27950948e-01 1.33051291e-01 5.36401391e-01
5.69598198e-01 -4.61883396e-02 -1.19468665e+00 8.32246244e-01
-1.53347120e-01 -1.27869022e+00 -4.10547286e-01 3.01117957e-01
4.08000380e-01 2.73366600e-01 7.42916018e-02 3.38022828e-01
1.64925739e-01 -1.07543409e+00 2.04126954e-01 7.24206507e-01
1.15303016e+00 -7.39872098e-01 9.87420857e-01 3.50748956e-01
-9.17991340e-01 -1.28178626e-01 -1.13980025e-01 -1.40002742e-01
3.96659821e-01 1.03594184e+00 -3.31625700e-01 8.12421322e-01
8.31388474e-01 9.46992815e-01 -5.08806944e-01 1.23327613e+00
2.52076596e-01 7.15740860e-01 -1.70641899e-01 1.41923711e-01
-5.55121107e-03 -4.04647700e-02 4.23174202e-01 1.02827871e+00
5.40411294e-01 2.25226015e-01 7.66140074e-02 5.33224881e-01
9.29705054e-02 2.45169714e-01 -6.96323156e-01 -3.39754134e-01
1.61829174e-01 1.23730385e+00 -7.05035210e-01 -3.60608399e-02
-8.73350739e-01 7.31378496e-01 2.65144259e-01 2.45293170e-01
-5.35600424e-01 -1.44736066e-01 7.25008845e-01 3.90810996e-01
-2.61878818e-01 -2.79355824e-01 -4.55523133e-01 -1.23445821e+00
-4.98234272e-01 -7.23890245e-01 1.03408372e+00 -5.16013503e-01
-1.76058757e+00 3.42445552e-01 1.42104790e-01 -1.10324681e+00
9.92211979e-03 -4.91212547e-01 -2.21228227e-01 9.27673757e-01
-1.47168016e+00 -1.02315164e+00 -2.11710870e-01 3.65155905e-01
4.01239812e-01 -1.32639512e-01 1.11357784e+00 3.80615085e-01
-9.35697019e-01 3.68347108e-01 7.23965585e-01 -9.37470123e-02
1.05763853e+00 -1.05327725e+00 1.37376770e-01 4.23573792e-01
-3.85576844e-01 6.35596693e-01 6.07886612e-01 -7.60724008e-01
-1.50592768e+00 -1.14482749e+00 1.10737693e+00 -7.67596304e-01
7.44178176e-01 -4.52939123e-01 -8.55585456e-01 6.93915546e-01
-2.14100808e-01 1.16667159e-01 1.31883419e+00 3.44653040e-01
9.02938321e-02 -7.81853423e-02 -1.25320768e+00 6.30212188e-01
8.88320744e-01 -2.08985761e-01 -1.80628017e-01 5.34073830e-01
2.72867322e-01 -2.38223299e-01 -1.24327397e+00 5.43789387e-01
6.42367601e-01 -9.09653187e-01 1.00877440e+00 -8.70738328e-01
4.48400378e-01 -3.06820334e-03 -3.73062491e-02 -9.94498670e-01
-4.63711381e-01 -1.99188873e-01 -3.00069004e-01 1.01126468e+00
8.36649165e-02 -7.40017354e-01 7.00955033e-01 4.29382443e-01
-2.11721137e-01 -1.17702854e+00 -1.23690474e+00 -7.30039358e-01
6.71584964e-01 -3.96205455e-01 4.55868363e-01 1.02356446e+00
-6.33221418e-02 -4.89307158e-02 -4.81281102e-01 3.49329226e-02
7.49046803e-01 1.87205419e-01 4.67464954e-01 -1.71513891e+00
8.64979401e-02 -3.47792089e-01 -4.12864506e-01 -4.55691695e-01
5.52954786e-02 -9.36545789e-01 -3.60805809e-01 -1.73701394e+00
5.45125961e-01 -3.69732201e-01 -6.33946776e-01 3.21516395e-01
-2.43990585e-01 1.36334345e-01 -2.82652676e-01 4.87732619e-01
-3.50950569e-01 4.76027966e-01 9.81057584e-01 1.97121784e-01
-8.15406069e-02 1.91901252e-01 -8.18295777e-01 4.30597186e-01
9.92836952e-01 -7.10068882e-01 -1.22351572e-01 -2.93195490e-02
1.85112476e-01 5.38286805e-01 5.74028015e-01 -8.68730485e-01
-2.75617559e-02 -4.28632438e-01 5.54137111e-01 -6.73850715e-01
8.28939527e-02 -8.47224832e-01 4.86294776e-01 6.02251709e-01
-1.67439878e-01 9.39853787e-02 2.46380776e-01 7.06657410e-01
-1.20816961e-01 5.70940860e-02 7.46840656e-01 -1.80704221e-01
-4.05112565e-01 6.66996658e-01 -7.10473299e-01 -1.92652091e-01
1.11293876e+00 -9.33956429e-02 -3.94114971e-01 -6.63831711e-01
-9.20896411e-01 4.19541210e-01 7.77479231e-01 3.06217939e-01
3.29689533e-01 -1.28112125e+00 -8.12618494e-01 -1.06922835e-01
4.80897456e-01 -1.49014384e-01 7.49505103e-01 1.37711239e+00
-3.59475374e-01 7.48001397e-01 -7.89360851e-02 -3.63745481e-01
-1.04721224e+00 8.64019275e-01 4.52189445e-01 -3.93772513e-01
-8.12664628e-01 5.46918809e-01 4.55635995e-01 -3.50999832e-01
3.50649476e-01 6.32311106e-02 -1.77869156e-01 3.98298651e-01
4.81948167e-01 5.11985600e-01 1.60599560e-01 -3.87177497e-01
-7.34969318e-01 2.09392294e-01 3.31341401e-02 1.59839824e-01
1.53018737e+00 -3.21433485e-01 -7.54545331e-02 8.90686870e-01
1.17834687e+00 -3.20604891e-01 -1.14935946e+00 3.40900570e-02
1.95081308e-02 -2.05651626e-01 2.10127458e-01 -1.03707254e+00
-1.04560828e+00 9.16327655e-01 7.95047164e-01 2.42950767e-01
1.40312588e+00 2.81264782e-01 4.83560115e-01 -1.07995249e-01
3.17965060e-01 -5.45067489e-01 -2.98156619e-01 2.96253532e-01
8.33738923e-01 -1.32249415e+00 9.76396948e-02 4.72987145e-02
-3.38988930e-01 1.05596948e+00 9.74704996e-02 2.83150136e-01
7.70343482e-01 2.47797936e-01 1.00294717e-01 -3.24276358e-01
-8.50167572e-01 -2.25564256e-01 -2.51065101e-02 7.24117994e-01
8.84418547e-01 -2.04657968e-02 -7.07342982e-01 7.61437893e-01
2.14442879e-01 4.96696085e-01 4.65530992e-01 8.72348785e-01
-2.11848363e-01 -1.25036347e+00 -3.05559814e-01 8.87372911e-01
-7.15237081e-01 -1.23789065e-01 -4.22844499e-01 7.04594791e-01
-1.89098660e-02 9.32006299e-01 -9.66057628e-02 -2.63098963e-02
1.10751957e-01 1.99503094e-01 7.31921941e-02 -4.55148757e-01
-4.14146990e-01 5.77645972e-02 3.15492153e-02 -3.45321894e-01
-2.53191322e-01 -1.18136346e+00 -9.56307948e-01 -6.94453180e-01
-3.93402055e-02 -1.65348873e-01 4.59908098e-01 7.57351339e-01
3.50658327e-01 4.29992259e-01 3.22909653e-01 -6.91435993e-01
-3.53554964e-01 -7.57645130e-01 -8.70342970e-01 -2.47539710e-02
5.71388662e-01 -7.91143894e-01 -6.06063604e-01 -2.02889338e-01]
|
[7.696544170379639, 5.569269180297852]
|
61c5a8cf-7672-49cc-82d5-09299122945e
|
unbnlp-at-semeval-2021-task-1-predicting
| null | null |
https://aclanthology.org/2021.semeval-1.83
|
https://aclanthology.org/2021.semeval-1.83.pdf
|
UNBNLP at SemEval-2021 Task 1: Predicting lexical complexity with masked language models and character-level encoders
|
In this paper, we present three supervised systems for English lexical complexity prediction of single and multiword expressions for SemEval-2021 Task 1. We explore the use of statistical baseline features, masked language models, and character-level encoders to predict the complexity of a target token in context. Our best system combines information from these three sources. The results indicate that information from masked language models and character-level encoders can be combined to improve lexical complexity prediction.
|
['Paul Cook', 'Samin Fakharian', 'Ali Hakimi Parizi', 'Milton King']
|
2021-08-01
| null | null | null |
semeval-2021
|
['lexical-complexity-prediction']
|
['natural-language-processing']
|
[ 1.15459539e-01 4.43405174e-02 -6.18846655e-01 -8.39168727e-01
-9.89831030e-01 -3.63207012e-01 4.90610719e-01 6.27196729e-01
-1.04373753e+00 7.80276120e-01 5.40547490e-01 -5.25274456e-01
6.09578550e-01 -5.01659036e-01 -3.35805207e-01 3.65921147e-02
-1.02665685e-01 1.35816708e-01 1.50737599e-01 -3.32342982e-01
5.24372756e-01 4.20015246e-01 -1.48548353e+00 1.07039833e+00
6.33885920e-01 8.44655752e-01 4.39015299e-01 8.45877171e-01
-6.18821919e-01 1.23528385e+00 -5.11309803e-01 -4.55277443e-01
-1.21915840e-01 -2.47797444e-01 -9.01459694e-01 -4.09098238e-01
3.31326813e-01 2.95988023e-02 -1.57175422e-01 8.23678970e-01
4.44481850e-01 -1.87203929e-01 5.66882908e-01 -7.28802443e-01
-2.59459704e-01 1.35095251e+00 -1.55960053e-01 8.66854191e-01
6.37963593e-01 1.66389182e-01 1.38397825e+00 -1.26567173e+00
6.99390650e-01 1.50869310e+00 7.66050339e-01 5.46580493e-01
-1.12440252e+00 -6.06393397e-01 3.32169592e-01 4.56502914e-01
-1.60847676e+00 -9.48388755e-01 5.60362816e-01 -4.07205552e-01
2.25969958e+00 3.90922837e-02 3.30015749e-01 8.22495162e-01
4.21397924e-01 9.07940090e-01 1.26669848e+00 -1.10590243e+00
-3.75545770e-02 1.24836683e-01 3.74860615e-01 9.77795124e-01
2.20599677e-02 2.11386919e-01 -8.72162104e-01 -1.43720403e-01
-6.06107004e-02 -9.33020890e-01 4.23741549e-01 7.11196899e-01
-8.26811790e-01 8.58338177e-01 -1.42451167e-01 5.92275381e-01
-3.06655794e-01 8.36366788e-02 8.80249977e-01 4.96173173e-01
6.47060812e-01 7.03869104e-01 -9.30460453e-01 -4.08616871e-01
-1.07310975e+00 2.92545557e-01 8.04415405e-01 1.03142965e+00
7.71627426e-01 2.78253794e-01 -3.33729267e-01 8.89632463e-01
3.66437912e-01 1.93843320e-01 6.55136526e-01 -5.71406424e-01
7.87696183e-01 4.37272340e-01 -1.96034968e-01 -4.16878223e-01
-7.92806447e-01 2.98970848e-01 -1.92687679e-02 -2.35537365e-01
2.68469065e-01 -1.03717454e-01 -7.68853128e-01 1.86235273e+00
-2.78168947e-01 -3.78471643e-01 1.83700457e-01 9.67363939e-02
8.33582580e-01 7.77348042e-01 8.35167587e-01 -6.07718527e-01
1.35946500e+00 -7.54722655e-01 -9.64455545e-01 -4.53774840e-01
1.39995897e+00 -9.42809105e-01 1.10974169e+00 2.55586535e-01
-1.67661273e+00 -6.27633750e-01 -1.01876009e+00 -3.27952236e-01
-6.27470374e-01 1.67595983e-01 5.23728192e-01 7.81333745e-01
-9.52835143e-01 6.26066744e-01 -6.95666254e-01 -1.25628582e-03
1.00107051e-01 3.02943081e-01 -2.18284801e-02 1.41657308e-01
-1.67923343e+00 1.39607680e+00 5.86584747e-01 -1.43305078e-01
-4.15858597e-01 -4.96027499e-01 -1.07217741e+00 -7.43824616e-02
-6.16075583e-02 7.96084583e-04 1.52614248e+00 -9.41250145e-01
-1.40638530e+00 1.16063845e+00 -6.39584363e-01 -4.58139300e-01
-9.35401097e-02 -7.23484010e-02 -5.75097561e-01 -1.61736310e-01
1.45030782e-01 5.37643850e-01 4.22564864e-01 -5.80545127e-01
-9.32392836e-01 -1.68826625e-01 -3.46627951e-01 2.93201476e-01
-3.16105098e-01 8.91156495e-01 -4.66942489e-02 -6.71723843e-01
-3.50420028e-01 -7.98153579e-01 -2.37242267e-01 -6.49871528e-01
-2.19936728e-01 -7.31163502e-01 9.69401598e-02 -9.15241003e-01
1.86806524e+00 -2.08490825e+00 7.96923339e-02 6.59183636e-02
-2.09526911e-01 2.44109407e-01 -3.36320072e-01 3.52164656e-01
-2.93270528e-01 6.43748462e-01 5.39293438e-02 -4.41549629e-01
1.74197799e-03 1.64348185e-01 1.76945269e-01 -1.57322399e-02
6.73056901e-01 1.29021358e+00 -7.29483604e-01 -6.81217968e-01
2.87222806e-02 4.09573130e-02 -5.66399932e-01 2.48781621e-01
-4.62302297e-01 -1.34629130e-01 1.46727130e-01 6.25802338e-01
2.49863386e-01 4.55152482e-01 4.24271673e-01 -6.25560805e-02
-3.61373574e-01 1.25030279e+00 -9.20563877e-01 1.35364568e+00
-8.06074858e-01 5.86313844e-01 -2.25681707e-01 -7.09286511e-01
6.59070611e-01 4.55319881e-01 2.44353358e-02 -9.08571541e-01
1.88253358e-01 4.14486200e-01 4.61870164e-01 -4.13305432e-01
6.45710945e-01 -3.54721934e-01 -5.94193280e-01 1.81081131e-01
1.82161927e-01 -2.34364271e-01 4.44319367e-01 -1.94273725e-01
1.05038297e+00 -1.66667864e-01 7.83188641e-01 -5.89926481e-01
7.81952083e-01 -8.72991160e-02 4.78890538e-01 5.49768984e-01
-2.11256251e-01 -3.20348069e-02 3.63643020e-01 -4.59989101e-01
-1.17400360e+00 -5.80981433e-01 -1.68295905e-01 1.84141481e+00
-5.07741868e-01 -8.90711129e-01 -5.50402284e-01 -4.37020391e-01
-3.35514136e-02 1.10395110e+00 -4.55162525e-01 -1.29532740e-01
-1.07341516e+00 -4.69686836e-01 1.13761222e+00 8.77805531e-01
-2.41494834e-01 -1.37167776e+00 -2.03159109e-01 5.64700425e-01
-3.52292269e-01 -1.53091753e+00 -5.85576594e-01 8.61794770e-01
-7.36727238e-01 -6.98816776e-01 -5.86912632e-02 -1.11838663e+00
6.29598200e-02 -3.67405742e-01 1.38318384e+00 2.63543040e-01
-3.22145790e-01 -2.24869832e-01 -3.21211964e-01 -5.93830287e-01
-8.90322506e-01 3.89260054e-01 2.75855333e-01 -6.10931695e-01
1.04863870e+00 -8.87188092e-02 2.07648084e-01 -1.21423900e-01
-4.71924335e-01 -2.55994439e-01 5.57719946e-01 6.69721305e-01
4.26641017e-01 -1.05064042e-01 3.92762661e-01 -9.93980408e-01
8.80426109e-01 -2.37640738e-01 -3.43009204e-01 2.33238578e-01
-7.75137246e-01 4.95988965e-01 6.29142344e-01 -5.50562561e-01
-1.08638871e+00 3.73697370e-01 -5.55554867e-01 9.37991068e-02
-1.74866319e-01 9.32087600e-01 -4.65316087e-01 1.11110508e-01
8.85457098e-01 7.61120021e-02 -3.45759124e-01 -4.86633807e-01
5.50574660e-01 7.15597212e-01 2.26473764e-01 -8.35728943e-01
9.86726806e-02 -4.88092810e-01 -3.70137185e-01 -1.08317912e+00
-8.32798660e-01 -4.17241484e-01 -8.74866545e-01 2.45528817e-01
8.54296505e-01 -1.17742085e+00 -2.31398657e-01 4.89086479e-01
-1.58217430e+00 -2.76215434e-01 -4.26545769e-01 2.97385901e-01
-5.01886427e-01 4.34034258e-01 -1.12276733e+00 -7.85299003e-01
-3.25336188e-01 -9.86149013e-01 9.17845964e-01 -1.80593893e-01
-6.48499429e-01 -9.44231093e-01 -3.39100361e-02 -1.06027782e-01
3.35850507e-01 -3.63570809e-01 1.29003799e+00 -8.83952141e-01
-5.87516613e-02 -7.93517753e-02 6.94299117e-02 3.21292162e-01
2.06946973e-02 9.24016610e-02 -8.28076661e-01 1.26060948e-01
-2.77508408e-01 -7.90992796e-01 7.16110706e-01 2.33952895e-01
1.10309803e+00 -4.34111208e-01 -5.71871586e-02 3.22073489e-01
1.23884928e+00 8.69349465e-02 5.63630998e-01 1.07368290e-01
5.39990127e-01 6.30917013e-01 3.77731234e-01 4.22819227e-01
7.42367744e-01 6.00459993e-01 -3.06423277e-01 2.59889752e-01
-1.22634798e-01 -1.74620435e-01 6.38805568e-01 1.46618748e+00
7.01517835e-02 -3.55651267e-02 -1.30079305e+00 5.93146384e-01
-1.18966603e+00 -9.04798329e-01 -1.08948112e-01 1.67847407e+00
1.51103210e+00 6.93119228e-01 2.99703032e-01 2.19417557e-01
6.01874769e-01 1.65487364e-01 -1.06559478e-01 -1.31356919e+00
-4.75400448e-01 5.83442152e-01 5.63341558e-01 1.04924452e+00
-8.39835823e-01 1.53386700e+00 8.08116817e+00 1.00258970e+00
-6.51286066e-01 1.45888999e-01 5.44249475e-01 1.20715559e-01
-3.54291707e-01 4.42893989e-02 -1.47928572e+00 3.02321345e-01
1.82547283e+00 -4.92284983e-01 4.00178522e-01 7.31221437e-01
5.74073754e-02 -2.31584951e-01 -1.45073807e+00 8.05287540e-01
7.99964666e-02 -1.10234249e+00 4.55851518e-02 -2.24169746e-01
4.28721905e-01 1.91187903e-01 -3.70607346e-01 6.24288380e-01
1.89527526e-01 -1.08856547e+00 6.55746281e-01 4.80775386e-01
8.25952351e-01 -8.36927712e-01 6.32552147e-01 4.07717943e-01
-1.50525820e+00 2.61226837e-02 -2.27755651e-01 -5.52416742e-01
1.33260310e-01 3.76443297e-01 -7.67325580e-01 5.69065427e-03
1.09171174e-01 4.72314268e-01 -8.51412296e-01 3.89723212e-01
-2.03942761e-01 8.09670746e-01 -2.41561726e-01 -5.56129098e-01
9.01753306e-02 5.47975481e-01 1.50164723e-01 2.14390063e+00
-3.87633353e-01 2.90444314e-01 3.48597974e-01 4.09380049e-01
2.86912024e-02 8.19109499e-01 -4.89875644e-01 -3.66267681e-01
6.19368374e-01 8.41304541e-01 -4.38126057e-01 -7.03838825e-01
-7.37346470e-01 6.98234379e-01 8.11259508e-01 -1.95544168e-01
-6.11183226e-01 -4.94803309e-01 6.25578523e-01 3.21870446e-02
3.06722879e-01 -4.76990700e-01 -6.10862553e-01 -1.14458525e+00
-1.27882004e-01 -1.04145896e+00 3.29869181e-01 -3.37038606e-01
-1.46920741e+00 8.29879522e-01 9.26908627e-02 -5.62974155e-01
-4.43625093e-01 -1.02074063e+00 -4.14264113e-01 1.18200660e+00
-1.76452446e+00 -6.89491808e-01 4.00197268e-01 2.36866519e-01
8.47617626e-01 -4.17413861e-01 1.13710845e+00 1.69590190e-01
-3.21076870e-01 1.06282032e+00 -2.79819369e-01 1.28410667e-01
3.81444693e-01 -1.38075066e+00 8.26084197e-01 5.36317050e-01
2.45746061e-01 3.55939627e-01 3.61809880e-01 -7.51642525e-01
-8.98066938e-01 -8.60949457e-01 1.99154007e+00 -6.35518253e-01
7.74115860e-01 -5.85587263e-01 -9.23728287e-01 5.04452169e-01
2.04570696e-01 -1.60967037e-01 8.89268398e-01 4.79744285e-01
-4.19851571e-01 3.46554875e-01 -7.50080109e-01 3.31567556e-01
8.75728607e-01 -1.09790397e+00 -9.93025541e-01 2.44775176e-01
8.26569557e-01 -3.26615274e-01 -9.45542991e-01 5.06055772e-01
5.13959348e-01 -4.80442762e-01 6.00775242e-01 -1.01776493e+00
5.42455018e-01 2.79766172e-01 -5.36256850e-01 -1.24768960e+00
-3.94794732e-01 -2.59721845e-01 -1.84520125e-01 8.74101579e-01
9.95152712e-01 9.02613625e-04 5.40255010e-01 7.83441544e-01
-2.09889803e-02 -7.41621673e-01 -1.18495083e+00 -8.52692783e-01
7.30944157e-01 -8.83586884e-01 2.68916726e-01 7.45621026e-01
6.70835197e-01 1.04190254e+00 -1.63698494e-01 -2.25448504e-01
1.48325950e-01 -4.44883823e-01 8.83185342e-02 -9.36669409e-01
-1.90645978e-01 -6.85082436e-01 -5.81346989e-01 -8.57344389e-01
8.43075573e-01 -1.19715607e+00 2.78362721e-01 -9.11318600e-01
3.92892733e-02 -2.68537104e-01 -2.91480452e-01 7.61571229e-01
-4.58619118e-01 -1.20272875e-01 2.63621539e-01 -2.42118657e-01
-4.16876167e-01 4.31196451e-01 3.90314460e-01 4.37278934e-02
-3.05426508e-01 -3.96343797e-01 -4.96083468e-01 7.89180040e-01
1.02408755e+00 -6.11822367e-01 5.39551489e-02 -5.16582489e-01
3.51124674e-01 5.81512898e-02 -4.42905664e-01 -7.84073174e-01
1.01720609e-01 -1.95062473e-01 3.22297931e-01 -6.63565934e-01
9.43995491e-02 -3.68616581e-01 -5.60619771e-01 3.80788326e-01
-7.39617348e-01 6.63331211e-01 8.06374371e-01 5.04066646e-02
-1.75419688e-01 -6.05906188e-01 8.60070169e-01 -4.56188232e-01
-8.01399231e-01 -1.54148102e-01 -1.16300786e+00 5.52816927e-01
5.37152231e-01 -5.60468435e-02 7.26440325e-02 1.39703443e-02
-1.01414335e+00 2.56934506e-03 8.82671699e-02 6.01492822e-01
5.77749729e-01 -1.24327862e+00 -9.16563094e-01 2.62712657e-01
3.19197327e-01 -7.00026035e-01 -4.45658237e-01 4.11656857e-01
-3.56979281e-01 6.91871822e-01 -1.68001771e-01 -1.74183369e-01
-1.38682163e+00 7.52758026e-01 3.04015279e-01 -4.51042503e-01
2.63419673e-02 1.09563148e+00 -5.57593524e-01 -4.68899488e-01
3.89824718e-01 -6.28430188e-01 -1.16935037e-01 1.45830512e-01
8.07193995e-01 1.54843330e-01 3.90124291e-01 -9.68061149e-01
-6.75836086e-01 2.22910881e-01 -4.59465414e-01 -2.58294612e-01
1.04812622e+00 -1.98560312e-01 -1.76213071e-01 9.98716891e-01
1.42366827e+00 1.83109477e-01 -3.12174916e-01 -7.22909629e-01
9.60538149e-01 -4.74308096e-02 1.08966075e-01 -7.04178751e-01
-3.79146963e-01 8.77776444e-01 4.10536766e-01 -2.17750594e-01
8.79184246e-01 4.10145745e-02 5.77682495e-01 6.58467472e-01
1.93487242e-01 -1.68990433e+00 -3.56704853e-02 1.12419510e+00
5.74310362e-01 -1.18538296e+00 4.12597274e-03 -4.76175666e-01
-7.61619031e-01 1.19396651e+00 8.08915138e-01 -2.91748364e-02
9.26982820e-01 7.65206873e-01 9.97129083e-02 -5.03018759e-02
-1.46738923e+00 -4.14264709e-01 2.45870396e-01 8.39614272e-02
1.17017090e+00 8.81109610e-02 -8.46598148e-01 8.26391697e-01
-3.53265375e-01 -2.07297280e-01 4.22135681e-01 1.09767151e+00
-5.94728291e-01 -1.37910140e+00 -1.39581086e-02 6.86204433e-01
-7.02179253e-01 -9.44253922e-01 -5.62854588e-01 4.69729185e-01
2.65289575e-01 7.94554770e-01 8.53645150e-03 -4.11142737e-01
5.74277461e-01 8.47254932e-01 5.78893781e-01 -1.02566540e+00
-7.49023616e-01 5.56372441e-02 7.89142907e-01 -6.35987401e-01
-4.46965814e-01 -9.29082870e-01 -1.20137584e+00 -1.18603259e-01
-4.81890410e-01 7.77096227e-02 6.23705685e-01 1.07396305e+00
-1.31915668e-02 1.91370487e-01 6.85689628e-01 -6.34767413e-01
-5.20726144e-01 -1.20209634e+00 -4.03715223e-01 1.20433033e-01
4.29043658e-02 -3.14100057e-01 -2.94909924e-01 8.58463868e-02]
|
[10.648832321166992, 10.449344635009766]
|
8ddd88a4-ed43-46f5-beab-cdc53485194b
|
target-specific-de-novo-design-of-drug
|
2302.07868
| null |
https://arxiv.org/abs/2302.07868v5
|
https://arxiv.org/pdf/2302.07868v5.pdf
|
Target Specific De Novo Design of Drug Candidate Molecules with Graph Transformer-based Generative Adversarial Networks
|
Discovering novel drug candidate molecules is one of the most fundamental and critical steps in drug development. Generative deep learning models, which create synthetic data given a probability distribution, have been developed with the purpose of picking completely new samples from a partially known space. Generative models offer high potential for designing de novo molecules; however, in order for them to be useful in real-life drug development pipelines, these models should be able to design target-specific molecules, which is the next step in this field. In this study, we propose DrugGEN, for the de novo design of drug candidate molecules that interact with selected target proteins. The proposed system represents compounds and protein structures as graphs and processes them via serially connected two generative adversarial networks comprising graph transformers. DrugGEN is trained using a large dataset of compounds from ChEMBL and target-specific bioactive molecules, to design effective and specific inhibitory molecules against the AKT1 protein, which has critical importance for developing treatments against various types of cancer. On fundamental benchmarks, DrugGEN models have either competitive or better performance against other methods. To assess the target-specific generation performance, we conducted further in silico analysis with molecular docking and deep learning-based bioactivity prediction. Results indicate that de novo molecules have high potential for interacting with the AKT1 protein structure in the level of its native ligand. DrugGEN can be used to design completely novel and effective target-specific drug candidate molecules for any druggable protein, given target features and a dataset of experimental bioactivities. Code base, datasets, results and trained models of DrugGEN are available at https://github.com/HUBioDataLab/DrugGEN
|
['Tunca Doğan', 'Abdurrahman Olğaç', 'Ahmet Rifaioğlu', 'Deniz Cansen Kahraman', 'Altay Koyaş', 'Heval Ataş Güvenilir', 'Hayriye Çelikbilek', 'Ahmet Sarıgün', 'Elif Çevrim', 'Atabey Ünlü']
|
2023-02-15
| null | null | null | null |
['molecular-docking']
|
['medical']
|
[ 4.39638525e-01 1.44069239e-01 -3.71129394e-01 9.32304040e-02
-7.60346651e-01 -7.40176618e-01 5.32445312e-01 3.74744833e-01
3.60404365e-02 1.47012949e+00 -9.18508843e-02 -4.63547766e-01
3.30998190e-02 -1.05731177e+00 -9.77278590e-01 -1.00937057e+00
6.92701936e-02 8.54431152e-01 -1.01359211e-01 -1.88098416e-01
6.47728741e-02 8.84056211e-01 -8.17683220e-01 2.92213619e-01
1.01812577e+00 2.14852154e-01 1.48834378e-01 3.85271370e-01
1.75599530e-01 3.73916000e-01 -5.67141593e-01 -4.87194091e-01
-2.72057094e-02 -7.90497303e-01 -6.42121613e-01 -3.89533490e-01
1.48580045e-01 1.72759160e-01 -2.10549310e-01 9.58691835e-01
8.79119396e-01 -6.19090348e-02 9.45829570e-01 -7.41106510e-01
-5.95157802e-01 4.63939369e-01 7.55198300e-02 -1.41129643e-01
4.48234916e-01 4.82239515e-01 8.64875138e-01 -1.03939056e+00
7.32895792e-01 1.04366398e+00 2.95145243e-01 9.54970837e-01
-1.74973452e+00 -8.85720730e-01 -4.60969299e-01 1.78556535e-02
-1.25407577e+00 -2.38998637e-01 6.45991743e-01 -5.63118935e-01
9.13995981e-01 2.89615691e-01 7.33457863e-01 1.53525305e+00
6.97576284e-01 3.99275094e-01 8.18792820e-01 1.14667483e-01
5.49366951e-01 -6.38732538e-02 -2.90763021e-01 6.72990978e-01
3.64434808e-01 3.97043645e-01 -3.69575053e-01 -5.67232668e-01
5.71456194e-01 2.35243902e-01 -4.20781851e-01 -3.69273245e-01
-8.74582052e-01 1.11233854e+00 5.66587925e-01 2.66201764e-01
-5.43443263e-01 9.95807201e-02 -8.04870855e-03 -8.27007890e-02
1.31472319e-01 1.09078217e+00 -5.29407442e-01 2.57929862e-01
-5.50906062e-01 3.30768108e-01 7.52830625e-01 4.21885312e-01
5.97399890e-01 1.35551840e-01 -1.38366133e-01 3.77740771e-01
1.86076179e-01 3.65246236e-01 5.16111195e-01 -1.31552786e-01
-2.37466499e-01 7.12385297e-01 -6.85945973e-02 -6.47789419e-01
-5.09885728e-01 -6.85255587e-01 -8.50808322e-01 1.75276130e-01
1.76443130e-01 -2.40670249e-01 -1.10384798e+00 1.77384079e+00
4.85912889e-01 3.71185571e-01 3.93608242e-01 6.79949999e-01
9.76720750e-01 9.07723963e-01 6.40562057e-01 -3.50394577e-01
1.08732355e+00 -4.79831398e-01 -2.54319042e-01 1.94949120e-01
5.44710279e-01 -6.92175865e-01 8.09481859e-01 4.34849530e-01
-9.47987080e-01 -2.42870986e-01 -1.17174971e+00 2.64223784e-01
-5.34861207e-01 -3.06528434e-02 8.63525748e-01 6.50134504e-01
-5.67913473e-01 1.00997698e+00 -8.42438996e-01 -8.83888337e-04
8.70284736e-01 8.43580484e-01 -4.17573839e-01 -1.08653441e-01
-1.39808154e+00 8.34419847e-01 6.15859866e-01 -2.19234347e-01
-1.59362543e+00 -9.47201133e-01 -5.78698874e-01 1.70122460e-01
5.75950816e-02 -1.10898578e+00 7.15663791e-01 -6.92580223e-01
-1.59178591e+00 4.33762699e-01 8.83111134e-02 -6.94209337e-01
4.26112600e-02 1.34615704e-01 -2.55538255e-01 -1.29644692e-01
-1.56536758e-01 5.72096169e-01 5.16160250e-01 -9.24663961e-01
-1.39644876e-01 -4.04052645e-01 -1.74269229e-01 5.87396845e-02
1.37761369e-01 -4.67292190e-01 1.82394266e-01 -5.34629941e-01
-5.35634458e-01 -1.10453534e+00 -5.63300014e-01 -5.28099358e-01
-7.02063262e-01 -1.23078555e-01 5.23608327e-01 -2.96583444e-01
8.33487570e-01 -1.63021910e+00 5.37744284e-01 3.15743655e-01
3.68404001e-01 8.18259060e-01 -3.02457392e-01 7.93800592e-01
-5.15414119e-01 3.39430332e-01 -9.91657376e-02 4.75735724e-01
-5.09376049e-01 -3.84949088e-01 -1.56884059e-01 5.22484899e-01
3.98867339e-01 1.23375309e+00 -1.00377083e+00 1.85154557e-01
1.98964756e-02 8.15103114e-01 -5.16415834e-01 1.50221288e-01
-9.38407660e-01 8.94621909e-01 -8.82710874e-01 6.00564122e-01
4.72113520e-01 -3.05699050e-01 4.97347862e-01 -1.11342221e-01
4.59822595e-01 2.66669929e-01 -4.32026118e-01 1.38968575e+00
-9.22555774e-02 7.19026551e-02 -8.75864565e-01 -8.25594068e-01
9.60978329e-01 4.69786644e-01 7.18885362e-01 -4.52582866e-01
6.10204898e-02 2.25735471e-01 3.22735190e-01 6.06591515e-02
-2.85306543e-01 -4.24311608e-01 1.25816107e-01 -7.24943280e-02
9.66923907e-02 -1.29196048e-01 2.23238945e-01 1.44633874e-01
1.29338741e+00 1.47887191e-03 2.85945952e-01 -1.48528889e-02
5.82627237e-01 2.58204490e-01 4.05750155e-01 1.43076256e-01
3.08986783e-01 2.27377847e-01 6.05637550e-01 -5.55797517e-01
-1.04756010e+00 -1.10210538e+00 -1.65479809e-01 4.38515216e-01
-1.96038529e-01 -3.57818097e-01 -6.25597537e-01 -7.03539908e-01
-9.09785405e-02 7.30852902e-01 -5.67088306e-01 -6.64819062e-01
-1.98080048e-01 -1.09215021e+00 5.44284642e-01 -9.97125879e-02
-3.94741260e-02 -1.18919504e+00 1.44980192e-01 7.43639886e-01
2.71177828e-01 -4.81722802e-01 -2.14600384e-01 4.13464487e-01
-7.96382725e-01 -1.59851205e+00 -8.34199548e-01 -7.56268620e-01
7.18755424e-01 -3.46817076e-01 9.39453781e-01 -1.22024603e-01
-3.38438153e-01 -2.48896793e-01 -1.02494493e-01 -6.70978129e-01
-7.49560237e-01 1.96925864e-01 2.35749148e-02 -2.69820213e-01
2.33924732e-01 -7.57754385e-01 -8.56257975e-01 2.13822052e-01
-8.71724963e-01 8.42860639e-02 7.08448172e-01 1.17992032e+00
9.73897576e-01 4.91191112e-02 9.60888445e-01 -1.23853862e+00
8.32401097e-01 -5.76486826e-01 -7.23494351e-01 1.50035858e-01
-6.27067268e-01 3.02664787e-01 9.39417303e-01 -5.53066134e-01
-6.80664957e-01 5.91852248e-01 -4.92832839e-01 -1.55082578e-02
-2.91537922e-02 7.28546441e-01 -6.49040461e-01 -2.49395948e-02
9.94228363e-01 3.51120472e-01 -6.58393605e-03 -2.82092839e-01
2.77511954e-01 -2.51238309e-02 3.26126516e-02 -5.40672004e-01
6.25311494e-01 3.74945290e-02 5.86577892e-01 -5.22565544e-01
-4.28793222e-01 -1.32643715e-01 -1.39810786e-01 2.99479246e-01
7.23186851e-01 -8.14650655e-01 -9.16033804e-01 2.02540413e-01
-1.08885384e+00 -3.64547729e-01 -3.02686002e-02 4.08986449e-01
-4.98676151e-01 -7.17026740e-02 -4.27994967e-01 -4.10679728e-02
-7.58633614e-01 -1.51553559e+00 7.44868577e-01 3.45514059e-01
-2.65585423e-01 -9.82790649e-01 5.76772094e-01 1.83791742e-01
-4.07405831e-02 8.01554143e-01 1.19695711e+00 -1.09734929e+00
-8.67711425e-01 -2.95160115e-01 3.06302041e-01 1.89270675e-01
3.14626127e-01 5.34086302e-02 -6.44526303e-01 -3.72614324e-01
-4.62442875e-01 -2.75121003e-01 7.19381094e-01 5.36325693e-01
9.28879797e-01 -4.79024380e-01 -7.43035436e-01 5.52004695e-01
1.47394216e+00 8.23878467e-01 9.76902425e-01 -1.31869232e-02
6.57361627e-01 -5.87704359e-03 4.15825933e-01 2.74234831e-01
-3.18481505e-01 6.74742281e-01 5.39971292e-01 -3.42458487e-01
-2.99618375e-02 -5.01476467e-01 2.76491135e-01 -1.49888068e-01
-1.03046305e-01 -6.74362481e-01 -7.85714805e-01 1.62469134e-01
-1.43269300e+00 -1.18465626e+00 -2.27649644e-01 2.28387427e+00
1.21235049e+00 3.04601174e-02 2.60626137e-01 -2.22301275e-01
5.40033698e-01 -4.87864614e-01 -9.69744682e-01 -2.96817154e-01
-1.66806906e-01 1.08793747e+00 4.03739393e-01 4.27814424e-01
-8.64160597e-01 1.17023706e+00 5.54331875e+00 9.70966756e-01
-1.43044293e+00 -3.10055226e-01 1.07359338e+00 -3.92436571e-02
-3.56138676e-01 1.57671094e-01 -8.52756679e-01 5.21963179e-01
1.18958724e+00 -4.64218408e-01 2.44401053e-01 7.12980390e-01
5.08530796e-01 1.83907270e-01 -1.18781316e+00 5.80150604e-01
-5.32948196e-01 -2.14516878e+00 3.12477589e-01 2.18148768e-01
9.03499603e-01 -8.86833593e-02 1.94932610e-01 7.11219236e-02
5.38318992e-01 -1.62890553e+00 -1.54167607e-01 5.93169451e-01
8.04178953e-01 -1.16487849e+00 5.78215480e-01 1.88951612e-01
-6.10692441e-01 3.59134287e-01 -3.03504467e-01 4.15223002e-01
-1.83646321e-01 6.07875824e-01 -1.50294268e+00 3.14125717e-01
-2.56470442e-02 6.44180536e-01 -5.05793214e-01 1.01774895e+00
-3.75118881e-01 6.15165055e-01 -5.58066517e-02 -2.40424305e-01
1.09407343e-01 -2.03760043e-01 3.11861038e-01 7.41137505e-01
7.18118772e-02 1.07300557e-01 2.09118456e-01 1.00111294e+00
-3.59074295e-01 3.15226674e-01 -6.55373216e-01 -5.51181972e-01
3.15208018e-01 1.05923879e+00 -4.97714430e-01 -2.43687674e-01
1.57325193e-01 9.19124305e-01 4.45569307e-02 5.04215419e-01
-1.00201988e+00 -2.33193815e-01 8.25613320e-01 3.43578964e-01
-2.99115945e-02 1.02051914e-01 2.47440231e-03 -6.97446287e-01
-6.88701868e-01 -1.29905891e+00 1.57567307e-01 -4.05810773e-01
-1.07271111e+00 5.38752913e-01 -4.54853624e-01 -1.06511736e+00
5.29032499e-02 -6.00812376e-01 -5.87993443e-01 1.08425200e+00
-9.48996544e-01 -1.25918961e+00 1.30263031e-01 4.30578172e-01
4.15979534e-01 -2.95764863e-01 1.18524718e+00 2.52497196e-01
-5.89603722e-01 3.71827483e-01 4.33227807e-01 -2.74853528e-01
8.06041300e-01 -1.02578354e+00 4.04738009e-01 3.98770988e-01
1.90767214e-01 8.27598453e-01 9.02021766e-01 -9.75837767e-01
-1.19837260e+00 -1.27337480e+00 4.13982421e-01 -5.08030713e-01
3.71988356e-01 -3.11947078e-01 -7.14774489e-01 2.66847700e-01
-1.93857163e-01 -2.49838799e-01 1.15105820e+00 -3.07986170e-01
1.07342497e-01 3.13201576e-01 -1.26084030e+00 6.68051720e-01
6.51013613e-01 -8.15764144e-02 1.79288015e-02 9.19858634e-01
6.52673960e-01 -5.00809968e-01 -1.07106352e+00 4.17686135e-01
4.31211084e-01 -5.58980048e-01 1.23322427e+00 -1.12677252e+00
4.81033504e-01 -5.29753804e-01 3.83446842e-01 -1.52050626e+00
-3.29598993e-01 -7.57262886e-01 -4.01587896e-02 6.16576433e-01
1.00065219e+00 -5.50468922e-01 1.14344156e+00 3.62333596e-01
5.54405898e-02 -1.13873959e+00 -7.67084718e-01 -5.86335599e-01
3.71899188e-01 1.13874197e-01 6.48108840e-01 6.80602729e-01
-5.02650347e-03 9.99896228e-01 -3.41695040e-01 2.65588649e-02
2.21016183e-01 -2.43880041e-02 8.62647355e-01 -1.17841601e+00
-3.93386096e-01 -3.85012239e-01 -4.32152927e-01 -4.32476938e-01
2.03599125e-01 -1.14375293e+00 -5.27741432e-01 -1.66318357e+00
1.99214295e-01 -3.25392216e-01 -1.44141778e-01 5.83568990e-01
-6.34346381e-02 -2.22176360e-03 -1.98149100e-01 6.36092722e-02
9.89846885e-03 7.09891319e-01 1.42747653e+00 -5.75828910e-01
-5.29929221e-01 4.54158574e-01 -8.18588436e-01 2.37981737e-01
9.75993931e-01 -6.30236447e-01 -6.04012072e-01 6.71714902e-01
2.09841490e-01 1.26683369e-01 2.76159883e-01 -8.66857946e-01
-2.54323751e-01 -5.26506722e-01 7.39016533e-01 -2.15746433e-01
3.30561817e-01 -7.06041157e-01 9.38837588e-01 9.52144802e-01
-1.50907129e-01 -4.21347529e-01 3.10782552e-01 6.26273453e-01
-5.10654338e-02 2.58473493e-02 1.02008307e+00 -1.10868007e-01
-3.86130154e-01 7.57818103e-01 -4.97204065e-01 -3.02984715e-01
1.20678377e+00 -1.07928135e-01 -1.99514747e-01 -2.34257996e-01
-1.04150724e+00 -2.26366416e-01 5.01839995e-01 1.86762795e-01
8.24845374e-01 -1.09398806e+00 -6.67450607e-01 -3.31798457e-02
1.30984798e-01 -1.33127391e-01 8.39736015e-02 4.16849822e-01
-7.94666231e-01 6.37740731e-01 -1.77828461e-01 -2.27270201e-01
-1.28120124e+00 8.88633311e-01 6.64342344e-01 -4.68993306e-01
5.30030467e-02 8.09155881e-01 5.88432848e-01 -1.38907433e-01
-3.05092096e-01 6.56385571e-02 -1.64338455e-01 -3.31211239e-01
2.85165101e-01 -5.61266877e-02 1.24866575e-01 -3.72879893e-01
-2.84993678e-01 1.21717051e-01 -2.51722574e-01 6.02412701e-01
1.52255177e+00 7.34848022e-01 1.07669033e-01 -4.19673979e-01
1.06476223e+00 -4.28410098e-02 -9.82579768e-01 2.33410493e-01
-5.82396127e-02 1.83373149e-02 -1.30675375e-01 -1.26042759e+00
-8.16095412e-01 4.70700115e-01 8.27356935e-01 -4.75034982e-01
8.87771428e-01 7.66313374e-02 7.60147810e-01 2.45312855e-01
2.81029940e-01 -4.47210401e-01 2.17825681e-01 1.78812623e-01
8.64335477e-01 -1.07660949e+00 1.83587715e-01 -4.55697745e-01
-3.48448575e-01 9.93507624e-01 4.68715996e-01 -3.31007801e-02
4.36247349e-01 -1.50425300e-01 -3.18657517e-01 -6.02586985e-01
-6.96045101e-01 2.60725189e-02 4.48402822e-01 7.18672991e-01
8.09007108e-01 2.34531224e-01 -6.40814662e-01 4.73160893e-01
1.53625563e-01 -1.02174105e-02 3.05745721e-01 5.45185864e-01
-2.06493765e-01 -1.91800237e+00 -1.10822074e-01 2.18961313e-01
-5.68463504e-01 -3.91119599e-01 -8.83866549e-01 6.40817404e-01
4.18274067e-02 5.89186549e-01 -7.01299608e-01 -3.19747746e-01
2.67076999e-01 1.40636982e-02 4.56687599e-01 -8.49986851e-01
-5.27307391e-01 1.75085500e-01 1.07828654e-01 -1.37942404e-01
-8.54723006e-02 -2.31188536e-01 -1.41457796e+00 -2.05833182e-01
-4.32478935e-01 4.47073013e-01 5.72532594e-01 5.14212489e-01
7.92068005e-01 5.81219256e-01 6.47730768e-01 -6.20229661e-01
-1.27527103e-01 -6.71285272e-01 -3.59673351e-01 1.94552213e-01
-1.01739794e-01 -4.98105407e-01 1.55905798e-01 2.94238120e-01]
|
[5.011440753936768, 5.74326229095459]
|
53e26e02-0044-42cc-819c-538e94f8b2f4
|
analyzing-categorical-time-series-with-the-r
|
2304.12332
| null |
https://arxiv.org/abs/2304.12332v1
|
https://arxiv.org/pdf/2304.12332v1.pdf
|
Analyzing categorical time series with the R package ctsfeatures
|
Time series data are ubiquitous nowadays. Whereas most of the literature on the topic deals with real-valued time series, categorical time series have received much less attention. However, the development of data mining techniques for this kind of data has substantially increased in recent years. The R package ctsfeatures offers users a set of useful tools for analyzing categorical time series. In particular, several functions allowing the extraction of well-known statistical features and the construction of illustrative graphs describing underlying temporal patterns are provided in the package. The output of some functions can be employed to perform traditional machine learning tasks including clustering, classification and outlier detection. The package also includes two datasets of biological sequences introduced in the literature for clustering purposes, as well as three interesting synthetic databases. In this work, the main characteristics of the package are described and its use is illustrated through various examples. Practitioners from a wide variety of fields could benefit from the valuable tools provided by ctsfeatures.
|
['José Antonio Vilar Fernández', 'Ángel López Oriona']
|
2023-04-24
| null | null | null | null |
['outlier-detection']
|
['methodology']
|
[-4.03702892e-02 -3.35112959e-01 5.21381013e-02 -5.16955733e-01
-3.81884724e-01 -6.99491322e-01 5.56703568e-01 8.68006468e-01
-2.82802671e-01 5.22798479e-01 -2.91489542e-01 -4.29761231e-01
-5.26979268e-01 -6.57551885e-01 -1.61793604e-01 -1.00752258e+00
-8.93483102e-01 1.68910071e-01 1.77545890e-01 -1.03144974e-01
4.20482844e-01 7.27796197e-01 -1.96330631e+00 4.10474539e-02
7.80595303e-01 9.94139194e-01 -3.77954282e-02 3.43422174e-01
-2.28079259e-01 1.93853110e-01 -8.01312387e-01 -5.41385636e-02
1.31672874e-01 -4.08301979e-01 -1.53548971e-01 3.04294210e-02
-6.07566297e-01 3.93384993e-01 1.44987032e-02 6.70256376e-01
3.11104894e-01 2.85859346e-01 5.86696029e-01 -1.61895287e+00
-1.29282832e-01 2.36564651e-01 -4.93686318e-01 3.06595683e-01
3.85157615e-01 1.20979488e-01 6.85520113e-01 -5.41071296e-01
5.73583186e-01 9.17442620e-01 6.25546157e-01 -2.32143849e-01
-1.24220240e+00 -3.42054904e-01 -2.90068984e-01 3.45959187e-01
-1.20427358e+00 -4.11770418e-02 8.43610287e-01 -4.70180869e-01
9.60777223e-01 5.17386079e-01 7.39516795e-01 6.83292568e-01
2.51220465e-01 4.65939820e-01 1.22285235e+00 -2.72829860e-01
5.57258308e-01 2.86040287e-02 3.22614044e-01 2.23691553e-01
2.03948408e-01 1.29797801e-01 -2.59612888e-01 -5.50006270e-01
5.59379995e-01 4.31011796e-01 8.00028369e-02 -4.45953220e-01
-1.32353115e+00 6.91469073e-01 -7.16819093e-02 9.61927056e-01
-3.80960435e-01 -2.64157981e-01 8.93396795e-01 7.31057405e-01
6.93943381e-01 3.82076144e-01 -4.76349354e-01 -5.94787896e-01
-7.29946434e-01 2.76896894e-01 7.76803434e-01 7.21112728e-01
4.21273232e-01 1.26108192e-02 2.23701298e-01 7.59596467e-01
-2.39143074e-02 1.30271599e-01 5.72046638e-01 -6.89981103e-01
-6.65845200e-02 8.91471922e-01 6.63700104e-02 -1.13833702e+00
-7.87495315e-01 -2.05937922e-01 -9.09923196e-01 7.35058337e-02
7.87103415e-01 6.72901645e-02 -5.18213987e-01 1.31915545e+00
4.21238631e-01 -7.92457014e-02 -1.93395942e-01 4.72232103e-01
5.01983404e-01 5.93782425e-01 -1.38811335e-01 -7.29786277e-01
1.15005374e+00 6.25921115e-02 -8.08427691e-01 5.51163316e-01
6.36260331e-01 -6.00604236e-01 8.71826887e-01 6.54767156e-01
-8.15094173e-01 -2.75492698e-01 -6.28904998e-01 2.37555549e-01
-8.52516353e-01 -2.47106284e-01 8.34082782e-01 5.51711380e-01
-9.14751887e-01 8.38803411e-01 -1.16751373e+00 -7.17317045e-01
1.86265007e-01 3.03432465e-01 -4.42343205e-01 1.90142274e-01
-1.02479172e+00 5.60991049e-01 2.73155004e-01 -2.75731012e-02
-1.09919295e-01 -4.82004613e-01 -9.12951529e-01 -1.31068856e-01
2.71669656e-01 -3.08187574e-01 9.31716919e-01 -7.30927885e-01
-1.03360593e+00 8.24883580e-01 -1.33047149e-01 -2.26769805e-01
6.65297687e-01 2.34639853e-01 -6.57867014e-01 1.07180007e-01
1.09568365e-01 -2.37307057e-01 6.76622152e-01 -4.62004483e-01
-3.78970623e-01 -6.52403712e-01 -5.16488552e-01 -4.66036171e-01
-2.79706120e-01 3.12157631e-01 -2.16451317e-01 -1.00753665e+00
-1.77985393e-02 -6.30503178e-01 -3.81759197e-01 -4.08806689e-02
-2.49965429e-01 -4.66980070e-01 8.05858552e-01 -4.19862300e-01
1.43625069e+00 -2.51884127e+00 1.13545708e-01 5.25230587e-01
-2.07628191e-01 -3.19489181e-01 1.72236219e-01 1.10913587e+00
-6.01104796e-01 1.05865270e-01 -5.29188037e-01 -2.31347289e-02
-1.34349823e-01 2.89676875e-01 -1.55306861e-01 8.70815098e-01
1.09266274e-01 4.27815378e-01 -1.07900763e+00 -1.21225737e-01
4.80584383e-01 1.55423477e-01 -9.75605398e-02 1.75081879e-01
-9.74641591e-02 5.76384544e-01 -3.35206151e-01 6.69416726e-01
3.21493357e-01 -1.63076594e-01 2.67858297e-01 4.25068587e-01
-6.55586421e-01 2.52726991e-02 -1.04976046e+00 1.51871312e+00
8.54682699e-02 5.55103779e-01 -3.54818642e-01 -1.33520377e+00
1.19089222e+00 4.32887226e-01 1.03248620e+00 -5.37890315e-01
1.48519874e-01 1.52136505e-01 1.85227528e-01 -7.10622191e-01
2.77325422e-01 2.46309284e-02 -1.63691327e-01 5.81500232e-01
-3.21303546e-01 -1.18361063e-01 5.80584943e-01 -9.30247232e-02
1.45031226e+00 6.25713468e-02 7.99069703e-01 -2.89728224e-01
4.67458904e-01 2.78215319e-01 4.29942518e-01 2.99165130e-01
-5.46455868e-02 1.87270820e-01 7.69510269e-01 -6.08969748e-01
-1.19887066e+00 -8.75556171e-01 -4.61207896e-01 8.23615313e-01
-3.52214724e-01 -6.36850893e-01 -2.59765536e-01 -8.26684833e-02
2.65506893e-01 5.78639090e-01 -8.24381948e-01 1.31737813e-02
-2.94554919e-01 -8.27260375e-01 2.62389809e-01 3.59936357e-01
-8.61408040e-02 -1.41044605e+00 -8.42180133e-01 1.91935048e-01
9.66272131e-02 -6.53874934e-01 1.61112994e-01 3.63229454e-01
-1.38674414e+00 -1.14706242e+00 -4.41872209e-01 -3.70880693e-01
5.90674877e-01 3.53283882e-01 9.67237532e-01 6.66831657e-02
-8.39667559e-01 4.35201913e-01 -6.10859036e-01 -6.64825618e-01
-3.92104745e-01 2.75298161e-03 6.54829964e-02 -4.43300530e-02
6.32666886e-01 -8.47915351e-01 -2.50070363e-01 4.35636342e-01
-1.27767956e+00 -4.49546158e-01 4.67083082e-02 6.93610668e-01
6.04465842e-01 4.46615309e-01 9.14418757e-01 -6.40457213e-01
5.90396345e-01 -9.74804342e-01 -7.70631909e-01 -4.86114360e-02
-3.06600541e-01 -1.83748245e-01 9.39759374e-01 -3.10046703e-01
-4.22465086e-01 -5.89102395e-02 2.55634710e-02 -2.96197861e-01
-5.08186996e-01 8.59840870e-01 1.45005826e-02 4.54297572e-01
5.33838928e-01 3.64549488e-01 2.63470620e-01 -5.66428244e-01
7.51530677e-02 5.80826879e-01 2.95406371e-01 -3.91328186e-01
6.02707684e-01 5.11048913e-01 1.92640632e-01 -1.22569370e+00
-4.13919464e-02 -8.16852152e-01 -6.24041319e-01 -3.23909491e-01
3.69347751e-01 -3.21866751e-01 -8.70225668e-01 4.94562000e-01
-6.35017097e-01 -6.42654747e-02 -4.26051974e-01 3.16862136e-01
-7.77545333e-01 2.28890046e-01 -2.14374587e-01 -1.09217381e+00
1.19092539e-01 -7.69374967e-01 7.86793292e-01 -8.63530040e-02
-5.73284447e-01 -1.10499454e+00 1.74079418e-01 -4.55711931e-01
1.72656879e-01 9.86503124e-01 1.09322941e+00 -9.56620038e-01
-2.39445001e-01 -6.46356940e-01 4.04986739e-01 -1.14306152e-01
4.84051764e-01 4.53468651e-01 -7.62370288e-01 -2.87527174e-01
1.51305094e-01 1.22117668e-01 4.51511919e-01 5.33334494e-01
1.53561914e+00 9.97469351e-02 -3.86777610e-01 4.02815014e-01
1.26791835e+00 7.06302524e-01 6.04517221e-01 3.51666451e-01
9.02442262e-02 1.09945059e+00 9.94456708e-01 9.00988877e-01
-2.18583532e-02 6.33672953e-01 4.59640592e-01 -3.63710076e-02
1.06188738e+00 2.80930489e-01 1.64092958e-01 6.98806286e-01
-3.03233504e-01 1.36739492e-01 -1.03186488e+00 6.32996798e-01
-2.06629181e+00 -1.26085901e+00 -5.42157769e-01 2.39162111e+00
5.69097996e-01 5.14768995e-03 7.22234845e-01 7.77784169e-01
5.86211264e-01 5.72066568e-02 -6.46792352e-01 -4.35937166e-01
-7.13256449e-02 -4.93657552e-02 7.21415803e-02 -9.82141122e-02
-1.04856110e+00 2.86516249e-01 6.49851179e+00 4.89830166e-01
-1.04053485e+00 -4.56210315e-01 5.82790017e-01 3.04494649e-02
-1.04438858e-02 -1.76576227e-01 -5.41341044e-02 5.57290792e-01
1.22900748e+00 -6.57377660e-01 2.21026555e-01 9.00523484e-01
7.60549128e-01 -3.17930400e-01 -1.03430939e+00 1.06582451e+00
-4.33030367e-01 -9.93279040e-01 -5.71223140e-01 1.85386375e-01
2.37204134e-01 -1.95221081e-01 -1.11365184e-01 8.08072016e-02
7.67795183e-03 -7.62501657e-01 2.65139610e-01 6.20265543e-01
6.78210437e-01 -1.07275939e+00 5.53152144e-01 2.79189348e-01
-1.13264012e+00 -1.74790055e-01 -2.50965476e-01 -3.52093101e-01
9.18229371e-02 1.11716592e+00 -6.50380611e-01 9.95766819e-01
8.76070499e-01 1.31691396e+00 -6.32660210e-01 1.49280655e+00
1.79785758e-01 6.26863897e-01 -4.63222057e-01 -1.17860742e-01
-6.33100942e-02 -5.48196971e-01 5.96191406e-01 1.06266654e+00
4.65802938e-01 -2.08264333e-03 -3.30549926e-02 4.51437861e-01
6.06395364e-01 4.16792244e-01 -1.07478094e+00 -4.34846908e-01
3.72453988e-01 1.25395989e+00 -1.13562179e+00 -1.92347378e-01
-5.49160302e-01 4.93057579e-01 -4.89613786e-02 2.29410276e-01
-7.60463178e-01 -6.58186972e-01 8.69849086e-01 2.24177226e-01
1.29627017e-02 -5.25911570e-01 -3.18922132e-01 -9.17686880e-01
5.76400645e-02 -8.92560065e-01 7.52501488e-01 -5.22044361e-01
-1.40998054e+00 3.38348001e-01 3.34561914e-01 -1.45090878e+00
-5.24172187e-01 -4.85522836e-01 -6.60777152e-01 6.68481946e-01
-7.17794001e-01 -1.81347415e-01 -3.09208542e-01 6.79440141e-01
4.07298952e-01 -3.90157886e-02 9.12860572e-01 1.92578793e-01
-6.70222342e-01 -5.35128079e-02 5.88173628e-01 -1.24369591e-01
6.08154535e-01 -1.33094633e+00 4.99094605e-01 4.64049101e-01
-1.39934227e-01 9.31459427e-01 1.03554475e+00 -5.96458614e-01
-1.35357893e+00 -7.97707796e-01 8.62443924e-01 -1.71773642e-01
1.07464802e+00 -3.82321119e-01 -1.12078404e+00 3.71619403e-01
-1.16198733e-01 -2.04768747e-01 1.11451876e+00 1.08312204e-01
-1.09512091e-01 -6.97850883e-02 -1.05823803e+00 6.21784508e-01
8.15621197e-01 -1.59678295e-01 -5.25814056e-01 2.30561793e-01
1.21794477e-01 9.47184339e-02 -1.08764923e+00 1.46361217e-01
4.93271679e-01 -1.20416284e+00 7.92514682e-01 -6.88032866e-01
1.43611073e-01 -4.99349654e-01 2.05438733e-01 -1.26703525e+00
-5.86310737e-02 -7.90384710e-01 5.60514145e-02 1.00835097e+00
-4.98995036e-02 -9.45403159e-01 2.92421013e-01 3.08131188e-01
-6.29368275e-02 -6.70023203e-01 -9.16741610e-01 -9.65302825e-01
-2.75404394e-01 -6.95531666e-01 5.25046587e-01 1.10185635e+00
5.59722960e-01 -2.21717313e-01 -2.07712036e-02 -3.60401362e-01
5.10280073e-01 3.45336586e-01 8.63186955e-01 -1.65592694e+00
-7.62023106e-02 -4.83862787e-01 -7.40962327e-01 -2.49409169e-01
-1.40144810e-01 -8.04331005e-01 -3.85798156e-01 -1.19124126e+00
-3.15310061e-01 -3.74373257e-01 -2.47378603e-01 5.43968499e-01
6.64678290e-02 -6.65484145e-02 -2.00075164e-01 2.43884757e-01
-1.55204460e-01 3.21668983e-01 7.89005101e-01 3.54683012e-01
-4.25159186e-01 3.74347895e-01 -3.07521105e-01 6.23470485e-01
9.71006036e-01 -6.37392044e-01 -3.82927895e-01 3.95469427e-01
1.80583388e-01 2.89321303e-01 3.23381513e-01 -7.16172695e-01
2.88905040e-03 -1.68984517e-01 3.11943948e-01 -7.84635186e-01
9.16104415e-04 -1.19319940e+00 6.87221587e-01 5.69219232e-01
1.93617865e-02 6.92858577e-01 3.58781546e-01 6.06304765e-01
-3.86337429e-01 9.26364288e-02 6.10687077e-01 -4.32056822e-02
-4.67934489e-01 9.47593972e-02 -5.69659293e-01 -3.40946525e-01
1.36738229e+00 -4.68362361e-01 -1.79916263e-01 -4.87592876e-01
-9.62973654e-01 1.30673766e-01 6.79769576e-01 6.05703950e-01
3.71069819e-01 -1.09937155e+00 -2.92519629e-01 2.64724195e-01
3.62781495e-01 -3.29181880e-01 1.93641722e-01 1.28119791e+00
-4.13603574e-01 4.17398989e-01 -5.27558148e-01 -7.98722923e-01
-1.27729893e+00 9.27213311e-01 -8.99826437e-02 -6.35940880e-02
-8.39469671e-01 -1.71029195e-02 -3.11376691e-01 -1.27696171e-01
1.68144852e-01 -4.81431991e-01 -2.47893587e-01 5.48195958e-01
7.10108638e-01 6.95946038e-01 8.95513818e-02 -3.35855007e-01
-5.59593976e-01 2.53535509e-01 2.70661771e-01 8.15027803e-02
1.91645634e+00 -1.23085469e-04 -4.64676380e-01 1.34622729e+00
9.96584296e-01 -2.43290052e-01 -8.79702687e-01 4.82215434e-02
7.10541546e-01 -4.81919140e-01 -7.42656052e-01 -4.53226835e-01
-6.17156923e-01 7.17243612e-01 2.81187087e-01 1.12271142e+00
1.55809259e+00 -1.72532097e-01 4.00024503e-02 8.25294778e-02
6.46705449e-01 -7.39478528e-01 -2.53004998e-01 4.01043475e-01
8.89403522e-01 -9.27405238e-01 -1.67627603e-01 -2.40441859e-01
-3.37778002e-01 1.35313940e+00 -1.26729071e-01 -1.98921561e-01
6.85879290e-01 4.07555699e-01 -4.63872142e-02 -2.41676852e-01
-7.89658725e-01 -2.46168196e-01 2.06765104e-02 7.29168057e-01
8.00590754e-01 -2.34642476e-01 -6.71401024e-01 4.69216108e-01
-1.87690839e-01 3.29902060e-02 5.93884289e-01 1.08415043e+00
-3.46627414e-01 -1.21730864e+00 -5.63467026e-01 6.48360431e-01
-6.14765525e-01 1.49264500e-01 -4.26034540e-01 9.48462248e-01
-3.12317282e-01 9.65528011e-01 2.05367163e-01 -8.82018954e-02
3.43768686e-01 2.39073262e-01 1.17513344e-01 -4.54157025e-01
-9.51818943e-01 2.51957089e-01 -1.02546290e-01 -6.29757047e-01
-5.59374988e-01 -1.06419921e+00 -1.14817572e+00 -3.49834532e-01
7.50310123e-02 3.59759271e-01 7.59451151e-01 6.52129471e-01
2.03087330e-01 5.67290485e-01 6.60936713e-01 -7.53139794e-01
-3.92628275e-02 -9.64627206e-01 -1.04395533e+00 4.63057637e-01
1.55753464e-01 -6.89081430e-01 -4.96290028e-01 1.19727939e-01]
|
[7.251550674438477, 3.358781337738037]
|
cf1349ea-9383-41e0-9d33-310be6fa1164
|
mbse-analysis-for-energy-sustainability
|
2208.01514
| null |
https://arxiv.org/abs/2208.01514v1
|
https://arxiv.org/pdf/2208.01514v1.pdf
|
MBSE analysis for energy sustainability improvement in manufacturing industry
|
With the ever increasing complexity of Industry 4.0 systems, plant energy management systems developed to improve energy sustainability become equally complex. Based on a Model-Based Systems Engineering analysis, this paper aims to provide a general approach to perform holistic development of an autonomous energy management system for manufacturing industries. This Energy Management System (EMS) will be capable of continuously improving its ability to assess, predict, and act, in order to improve by monitoring and controlling the energy sustainability of manufacturing systems. The approach was implemented with the System Modeling Language (SysML).
|
['Jean-Luc Dion', 'Arkadiusz Kosecki', 'Martin Ghienne', 'Olivia Penas', 'Romain Delabeye']
|
2022-08-02
| null | null | null | null |
['energy-management']
|
['time-series']
|
[ 7.02062398e-02 -1.13137374e-02 1.50466561e-01 1.37429431e-01
5.31946421e-01 -6.11269295e-01 8.91881704e-01 3.52500856e-01
3.66989315e-01 5.52395582e-01 -5.51459551e-01 -4.44670886e-01
-5.81601202e-01 -1.15548611e+00 -5.64187355e-02 -4.05161232e-01
2.11965516e-01 4.73572314e-01 -1.67450801e-01 -2.21791536e-01
2.23003402e-01 9.78965461e-01 -1.69197512e+00 -6.66121840e-02
6.90179229e-01 9.01368856e-01 7.19959557e-01 7.62634337e-01
9.74488072e-03 5.54664791e-01 -4.04578358e-01 7.68478096e-01
-1.09460697e-01 -7.97041953e-01 -8.70533705e-01 6.02893829e-02
-7.73520648e-01 -7.47497147e-03 2.78715640e-01 1.03195667e+00
-1.34549618e-01 1.92851216e-01 8.12558055e-01 -1.17619383e+00
-4.55687165e-01 3.95406216e-01 3.20184648e-01 -3.34058642e-01
1.02056235e-01 3.24024886e-01 -1.13474075e-02 -4.88422483e-01
4.25525218e-01 7.85072148e-01 -1.56792745e-01 1.25688955e-01
-9.43211555e-01 -2.72962928e-01 -3.80996644e-01 6.10605538e-01
-1.22168136e+00 -1.31312042e-01 6.63747013e-01 -4.55490321e-01
1.24072683e+00 7.05816388e-01 1.23296428e+00 5.08882225e-01
7.74190247e-01 -2.50996172e-01 1.27794898e+00 -8.27213168e-01
6.79266155e-01 3.47702950e-01 1.54752538e-01 1.65403739e-01
8.20752800e-01 3.36941004e-01 1.28090158e-01 2.74371743e-01
6.90208077e-01 -2.90411949e-01 2.34566674e-01 -3.92233491e-01
-7.17159569e-01 2.20499918e-01 7.83205181e-02 9.96143997e-01
-6.63257241e-01 3.29413921e-01 7.37484097e-02 1.96694970e-01
-1.73129156e-01 5.33788621e-01 -4.41545367e-01 -4.79115814e-01
-4.20157433e-01 -1.09048665e-01 1.01882720e+00 8.54323924e-01
3.49210352e-01 5.77015281e-01 4.18904424e-01 1.62029266e-03
6.31801307e-01 4.49541241e-01 2.51153201e-01 -1.25129247e+00
-4.89456713e-01 1.07576013e+00 4.43296760e-01 -2.53335744e-01
-4.88738805e-01 -4.01296049e-01 -6.28544450e-01 7.02771842e-01
-3.71480644e-01 -1.47641271e-01 -8.20539892e-01 1.10980117e+00
1.95295557e-01 -4.67184991e-01 4.15686369e-01 3.03286076e-01
1.05352417e-01 8.98526847e-01 1.74923405e-01 -6.25457585e-01
1.14496756e+00 -1.22071438e-01 -1.39425385e+00 2.80413657e-01
4.56330925e-01 -6.08030498e-01 4.06229228e-01 6.88000262e-01
-1.37266362e+00 -6.41484857e-01 -1.20303321e+00 5.56527197e-01
-9.31745946e-01 8.59704316e-02 1.04135565e-01 6.65536404e-01
-9.98132765e-01 8.08997214e-01 -9.67337787e-01 -6.36424005e-01
-3.01912457e-01 3.42508167e-01 -9.48715862e-03 4.55216289e-01
-1.18893695e+00 1.70591879e+00 1.16282976e+00 3.14293385e-01
-5.84396720e-01 -4.09048110e-01 -6.13520324e-01 4.21592087e-01
-4.16723788e-02 -7.03042924e-01 1.28170168e+00 -3.65469843e-01
-1.82557321e+00 1.62043765e-01 2.93499082e-01 -4.68690336e-01
3.92289788e-01 3.74501455e-03 -8.40926170e-01 1.72901340e-02
-6.50856197e-01 -1.71891913e-01 1.26360923e-01 -1.45899856e+00
-7.93526053e-01 -3.31828654e-01 -3.65565538e-01 -2.64666647e-01
-4.16452736e-01 -2.73967028e-01 5.56707442e-01 2.12727785e-01
-1.19991995e-01 -8.83447349e-01 -2.17825159e-01 -8.15109432e-01
6.70001060e-02 -1.10013098e-01 1.22190893e+00 -3.61120522e-01
1.13316071e+00 -1.53381431e+00 3.01106155e-01 5.52835166e-01
-6.68933749e-01 3.59145731e-01 5.11184990e-01 1.01990950e+00
-2.05970183e-01 -1.13237761e-02 -5.65753765e-02 3.00137550e-01
3.26408744e-01 5.37123203e-01 2.05276802e-01 -1.64612979e-01
1.96083695e-01 3.18889916e-01 -7.15104520e-01 -1.92782536e-01
1.34289134e+00 4.42019969e-01 2.91585088e-01 4.41387035e-02
-3.39509785e-01 4.00467396e-01 -3.14364463e-01 5.29543340e-01
3.30825359e-01 5.09227104e-02 7.40843832e-01 -2.80723721e-01
-8.90987456e-01 -4.27603364e-01 -1.47725558e+00 1.20531237e+00
-8.37212980e-01 1.54130727e-01 2.84556448e-01 -8.86910677e-01
9.89916623e-01 8.80136073e-01 1.05949318e+00 -5.62552392e-01
4.24318373e-01 5.88636220e-01 -9.49040502e-02 -7.56137729e-01
3.21260750e-01 -2.02045292e-02 1.91550910e-01 -3.90914708e-01
-5.55106346e-03 -5.05238771e-01 4.81260329e-01 -4.67946470e-01
8.85577619e-01 5.89563668e-01 5.00433087e-01 -6.78471744e-01
1.25730741e+00 1.33797616e-01 1.65418178e-01 -2.88203537e-01
-7.91148245e-02 -5.16487539e-01 1.44245610e-01 -9.66488943e-02
-8.28838170e-01 -8.66333306e-01 -2.26635888e-01 -1.51047945e-01
2.69309253e-01 -7.10576493e-03 -9.02512133e-01 -5.88432699e-02
-2.33252272e-01 1.64155614e+00 1.59767956e-01 -3.86185855e-01
-3.85208666e-01 -4.60093617e-01 -4.91924435e-01 1.30912960e-01
2.66659647e-01 -7.74218559e-01 -1.16129255e+00 7.87006795e-01
4.29108709e-01 -9.18618441e-01 6.70146286e-01 3.95725012e-01
-8.77012491e-01 -1.18160379e+00 1.87534481e-01 -4.80857491e-01
5.15087843e-01 -1.31494984e-01 1.00289917e+00 1.48084626e-01
-4.39862847e-01 5.09497046e-01 -3.11946154e-01 -8.19836736e-01
-1.19812965e+00 -9.44362506e-02 7.79363289e-02 -6.06038392e-01
1.53581277e-01 -6.56647742e-01 -6.43307090e-01 1.92244366e-01
-9.23389196e-01 4.14781831e-02 6.53141975e-01 4.65469323e-02
5.23047090e-01 9.83208179e-01 9.08763170e-01 -4.92581099e-01
6.95404172e-01 -5.47716141e-01 -1.04913080e+00 3.77459943e-01
-1.57901597e+00 -2.05058500e-01 9.52029824e-01 2.83286422e-01
-1.18311548e+00 3.76678079e-01 -1.42769381e-01 -1.03813238e-01
-6.35789871e-01 1.70205101e-01 -5.98485887e-01 1.55392699e-02
-7.62793496e-02 2.73003131e-01 8.06750208e-02 -4.66642767e-01
1.41072556e-01 5.22140563e-01 4.98892188e-01 -1.89859584e-01
8.61357689e-01 -4.24695015e-02 9.32692468e-01 -7.23045766e-01
5.23276269e-01 -1.44214451e-01 -3.13502640e-01 -9.83965993e-01
1.10981679e+00 -2.97873348e-01 -9.86314774e-01 8.66288245e-02
-9.39480782e-01 -1.48217514e-01 -6.64857209e-01 6.50056303e-01
-6.38956428e-01 1.36962738e-02 -2.22575724e-01 -1.36092305e+00
-2.85404563e-01 -1.08061695e+00 2.96468109e-01 3.25834066e-01
-4.20673132e-01 -9.44724262e-01 2.69803524e-01 -1.19416311e-01
5.81135392e-01 7.68113494e-01 9.87101316e-01 -7.54668489e-02
-7.62757063e-01 -2.49222040e-01 5.54462850e-01 8.25083137e-01
5.91334522e-01 6.48747623e-01 -5.78256667e-01 -1.99757755e-01
3.20386946e-01 5.69695592e-01 -1.35001719e-01 1.67281047e-01
8.33896101e-01 1.55564710e-01 -3.84604931e-01 -2.22463518e-01
2.08132172e+00 1.01254511e+00 5.25975108e-01 2.94750571e-01
-6.61294833e-02 8.41104031e-01 1.09627306e+00 4.50734884e-01
-8.12202021e-02 5.69481075e-01 1.03354251e+00 2.42583007e-01
-1.88315772e-02 3.67632329e-01 3.99800658e-01 8.09228420e-01
-3.69206548e-01 -2.72087485e-01 -6.59829199e-01 2.20685244e-01
-1.76176882e+00 -9.48707879e-01 -8.50209236e-01 1.96939838e+00
2.43777782e-01 3.19121271e-01 -2.82834500e-01 5.64575791e-01
4.64842021e-01 -4.58359510e-01 -1.00545816e-01 -9.95127439e-01
4.54345793e-01 3.30232054e-01 6.07783914e-01 3.81422877e-01
-4.69747126e-01 1.67483762e-01 6.02671909e+00 3.23634595e-01
-5.45149565e-01 3.60346437e-02 -2.16479495e-01 1.43213943e-01
-9.39813703e-02 2.51992702e-01 -3.95974189e-01 6.82333469e-01
1.96523166e+00 -5.81500053e-01 7.02153981e-01 7.28077352e-01
9.58218813e-01 -2.93635219e-01 -9.93802071e-01 3.73004287e-01
-7.01877356e-01 -1.33944631e+00 -2.79010721e-02 4.68100995e-01
7.09351122e-01 -5.39000928e-01 -5.20098090e-01 2.38039680e-02
3.15924853e-01 -5.61158419e-01 7.82961369e-01 1.17302144e+00
4.43500243e-02 -1.24907184e+00 9.92408156e-01 4.30797607e-01
-1.36963618e+00 -3.25556189e-01 3.14771347e-02 -1.18768074e-01
7.45298207e-01 4.99981582e-01 -4.92799252e-01 1.35041666e+00
5.88380992e-01 -8.58677328e-02 -1.04698114e-01 4.79989827e-01
1.42230242e-01 2.55850554e-01 -8.54492709e-02 -3.01113576e-02
-9.36866775e-02 -8.09986413e-01 2.36550689e-01 7.82384098e-01
8.46105874e-01 -6.92540854e-02 -2.81068325e-01 9.81680155e-01
6.75343275e-01 1.37961016e-03 -6.75411820e-01 -3.76668215e-01
5.24736822e-01 1.43142974e+00 -7.59940624e-01 -3.24553907e-01
-1.04718678e-01 5.33815682e-01 -7.97258794e-01 -1.54144719e-01
-8.18134606e-01 -3.37405086e-01 7.06425130e-01 5.37734963e-02
-8.40014517e-02 -6.54861629e-01 -4.28500831e-01 -1.79530606e-01
-4.17538315e-01 -1.07282035e-01 -1.61510810e-01 -9.01839495e-01
-7.34346390e-01 1.19201213e-01 4.35237795e-01 -1.05870831e+00
-8.85983825e-01 -9.52048182e-01 -6.39182270e-01 7.90836632e-01
-1.29026508e+00 -1.10338295e+00 -3.94156933e-01 4.12775837e-02
4.81170505e-01 -5.59680015e-02 1.24122441e+00 1.85051724e-01
-6.10010386e-01 -8.63585472e-01 3.83609116e-01 -9.96775866e-01
2.17820257e-01 -1.81315494e+00 -2.26396084e-01 1.01971877e+00
-6.58328354e-01 7.13195428e-02 1.07186282e+00 -7.10227549e-01
-1.74938631e+00 -9.42405701e-01 5.64349771e-01 -2.44311377e-01
6.36366844e-01 2.26447180e-01 -6.76696777e-01 4.18486118e-01
9.79426980e-01 -6.63792491e-01 4.64830428e-01 -6.88494861e-01
8.56183171e-01 -3.56670171e-01 -1.42843628e+00 3.23061049e-01
7.25758195e-01 6.08353987e-02 -4.73574251e-01 2.71896213e-01
4.18010056e-01 -7.04724267e-02 -1.81441689e+00 6.96436167e-01
3.99331629e-01 -7.51198351e-01 6.76501930e-01 -1.14323914e-01
-9.16557200e-03 -4.37375456e-01 1.09977096e-01 -1.31053317e+00
-3.62593472e-01 -6.79360330e-01 -7.80375898e-01 1.55561018e+00
-1.68888748e-01 -1.00111890e+00 3.29728901e-01 5.70966482e-01
-6.01674139e-01 -5.04134238e-01 -1.05210960e+00 -9.24212337e-01
-2.13107064e-01 -9.53070074e-02 9.11809921e-01 6.86352551e-01
2.71008074e-01 -5.10816835e-02 4.58916813e-01 4.78659481e-01
5.85251391e-01 -1.23277895e-01 2.29485989e-01 -1.51632416e+00
2.60538727e-01 -5.93294322e-01 -3.03098261e-01 6.00968242e-01
-2.11205110e-01 -5.10641277e-01 -2.45088786e-01 -2.39003134e+00
-5.08853495e-01 2.21379012e-01 -6.60915673e-01 5.19316830e-02
5.51271975e-01 -1.95709899e-01 3.32287580e-01 1.41758725e-01
3.95429462e-01 6.00516200e-01 1.07649148e+00 3.69000673e-01
-1.01851366e-01 1.04688130e-01 -6.97788298e-02 6.37323618e-01
1.12973475e+00 -1.89520657e-01 -7.83740878e-01 4.63338286e-01
1.78575322e-01 -8.20015520e-02 3.84974301e-01 -1.71114755e+00
-3.32158282e-02 -6.34038925e-01 4.23469901e-01 -8.81561995e-01
7.07792044e-02 -2.02711511e+00 1.63132596e+00 1.42627394e+00
2.88299143e-01 1.61041185e-01 1.32137284e-01 2.48821676e-01
-1.22458436e-01 -5.06278813e-01 6.63185596e-01 -3.39568816e-02
-8.06177199e-01 -5.00310004e-01 -9.53577518e-01 -1.18987763e+00
2.01920438e+00 -4.70305145e-01 -5.01613736e-01 4.79183227e-01
-8.62363458e-01 3.15394431e-01 7.52145708e-01 1.58974752e-01
1.57893822e-01 -1.17297089e+00 -2.37604469e-01 1.65552452e-01
-1.21073626e-01 -5.22117078e-01 4.72553432e-01 3.96863014e-01
-9.68769372e-01 7.25596070e-01 -7.49158025e-01 -1.61391333e-01
-1.04639494e+00 7.88362563e-01 7.14989185e-01 -3.50648463e-01
-1.58936545e-01 -4.05361772e-01 -6.49079382e-01 3.61676402e-02
-5.53305745e-01 -6.15132451e-01 -4.04281855e-01 -1.71131134e-01
-5.61032072e-02 1.01536310e+00 2.74702489e-01 -1.13646455e-01
-4.48470861e-01 4.55395609e-01 9.67054904e-01 -4.27657105e-02
1.25225401e+00 -1.57177001e-01 -3.20975423e-01 5.20305932e-01
4.23240662e-01 -1.88774973e-01 -1.01186156e+00 9.02560651e-01
4.02156442e-01 1.55788008e-02 3.79882753e-01 -1.15800655e+00
-6.31044626e-01 2.43369579e-01 1.12296486e+00 1.23902249e+00
1.49182141e+00 -4.44989741e-01 3.16660911e-01 1.67177826e-01
4.66723561e-01 -1.78678274e+00 -4.21318740e-01 -3.59198563e-02
8.93253922e-01 -5.78355551e-01 1.03947870e-01 -6.59622490e-01
6.83462024e-02 1.73603034e+00 4.37349886e-01 2.14109451e-01
7.37211645e-01 7.51501977e-01 -5.85346043e-01 -2.62504488e-01
-6.19626939e-01 -4.10998702e-01 -2.77412921e-01 5.73224485e-01
2.67371148e-01 2.39115283e-01 -7.44120717e-01 3.28647316e-01
2.58753061e-01 4.17267501e-01 4.91864443e-01 1.22284293e+00
-7.93574452e-01 -1.70015979e+00 -6.80792272e-01 -1.13775313e-01
-7.00368881e-02 8.87974560e-01 -3.85166258e-01 1.38212252e+00
5.28170288e-01 1.16527057e+00 -3.89963947e-02 -2.71209210e-01
1.07615948e+00 2.74755687e-01 4.84076977e-01 -2.39825875e-01
-6.73872948e-01 -3.29273313e-01 3.91934514e-01 -3.75593245e-01
-4.40556288e-01 -6.98293447e-01 -1.57727265e+00 -3.48687261e-01
-1.75325751e-01 3.09990346e-01 1.64076555e+00 7.67630517e-01
4.73165900e-01 1.56202316e+00 9.02698159e-01 -5.13712108e-01
-7.55852938e-01 -9.93495524e-01 -7.74011314e-01 -2.62397557e-01
-4.58536357e-01 -5.88437617e-01 -2.45882675e-01 3.10715705e-01]
|
[5.8464741706848145, 2.4720022678375244]
|
2a38e2fe-af59-4026-9e31-0d406eccdd61
|
graph-ranking-for-collective-named-entity
| null | null |
https://aclanthology.org/P14-2013
|
https://aclanthology.org/P14-2013.pdf
|
Graph Ranking for Collective Named Entity Disambiguation
| null |
['Robert Gaizauskas', 'Ayman Alhelbawy']
|
2014-06-01
| null | null | null |
acl-2014-6
|
['graph-ranking']
|
['graphs']
|
[-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]
|
[-7.445267677307129, 3.6603121757507324]
|
2bceb98b-26f8-4530-bc14-3abbdb2bd5ec
|
3d-sis-3d-semantic-instance-segmentation-of
|
1812.07003
| null |
http://arxiv.org/abs/1812.07003v3
|
http://arxiv.org/pdf/1812.07003v3.pdf
|
3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans
|
We introduce 3D-SIS, a novel neural network architecture for 3D semantic
instance segmentation in commodity RGB-D scans. The core idea of our method is
to jointly learn from both geometric and color signal, thus enabling accurate
instance predictions. Rather than operate solely on 2D frames, we observe that
most computer vision applications have multi-view RGB-D input available, which
we leverage to construct an approach for 3D instance segmentation that
effectively fuses together these multi-modal inputs. Our network leverages
high-resolution RGB input by associating 2D images with the volumetric grid
based on the pose alignment of the 3D reconstruction. For each image, we first
extract 2D features for each pixel with a series of 2D convolutions; we then
backproject the resulting feature vector to the associated voxel in the 3D
grid. This combination of 2D and 3D feature learning allows significantly
higher accuracy object detection and instance segmentation than
state-of-the-art alternatives. We show results on both synthetic and real-world
public benchmarks, achieving an improvement in mAP of over 13 on real-world
data.
|
['Matthias Nießner', 'Ji Hou', 'Angela Dai']
|
2018-12-17
|
3d-sis-3d-semantic-instance-segmentation-of-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Hou_3D-SIS_3D_Semantic_Instance_Segmentation_of_RGB-D_Scans_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Hou_3D-SIS_3D_Semantic_Instance_Segmentation_of_RGB-D_Scans_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['3d-instance-segmentation-1', '3d-semantic-instance-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 4.15946633e-01 2.58277833e-01 -3.95500511e-02 -5.95821738e-01
-1.00501454e+00 -6.92962110e-01 4.10998702e-01 1.17816128e-01
-2.91245997e-01 6.22802265e-02 -5.03525473e-02 -2.23536745e-01
2.36346424e-01 -9.68057871e-01 -1.12483335e+00 -5.04502773e-01
-8.04729238e-02 7.58330703e-01 4.31167603e-01 1.90703064e-01
2.71062553e-02 9.53722835e-01 -1.51322424e+00 4.06183332e-01
8.49159509e-02 1.60380208e+00 2.96367314e-02 7.94285059e-01
-4.52485442e-01 4.51186717e-01 -2.60243304e-02 -3.14845406e-02
7.70118535e-01 8.19686055e-02 -8.66461158e-01 5.66579998e-01
6.33277774e-01 -5.90477407e-01 -2.77122617e-01 6.70607746e-01
3.84823799e-01 -9.45761204e-02 6.10000432e-01 -9.89509404e-01
-2.26501703e-01 1.53473228e-01 -9.43693638e-01 9.20782983e-02
3.96838367e-01 2.82558709e-01 8.85300457e-01 -9.92033839e-01
6.69668794e-01 1.26067531e+00 7.92169034e-01 3.17344695e-01
-1.33643484e+00 -4.51422662e-01 1.08453803e-01 -3.47060740e-01
-1.04826021e+00 -1.24028854e-01 9.62779701e-01 -4.41444695e-01
1.04101908e+00 1.28349498e-01 1.12570870e+00 6.09071493e-01
-9.52223875e-03 9.71815050e-01 1.21702826e+00 -9.53859091e-02
2.04125449e-01 -1.90979704e-01 -9.06469971e-02 9.76212561e-01
-9.69606340e-02 9.78021994e-02 -5.47680259e-01 2.67736018e-02
1.17841792e+00 3.16527754e-01 7.50588626e-02 -8.32482755e-01
-1.28452849e+00 5.94501615e-01 8.88508141e-01 -2.50916123e-01
-6.25002027e-01 5.30047119e-01 9.85196456e-02 -1.97165161e-01
7.42216468e-01 1.37835130e-01 -6.59025133e-01 1.57346368e-01
-8.61046135e-01 1.84148759e-01 5.59693515e-01 5.97258031e-01
1.03327405e+00 -2.88839936e-01 6.83228150e-02 4.01598066e-01
4.32499796e-01 7.10699081e-01 -1.94289997e-01 -1.41437173e+00
3.23068976e-01 8.57596457e-01 -2.60363761e-02 -5.93552411e-01
-7.45370805e-01 -2.29072168e-01 -4.90034521e-01 4.87480760e-01
4.93684858e-01 2.26237625e-01 -1.41144991e+00 1.12493634e+00
7.30387211e-01 2.79144228e-01 -1.43735334e-01 1.04797232e+00
7.76201725e-01 4.09285814e-01 -1.41011611e-01 3.22834402e-01
1.14369833e+00 -6.00008488e-01 2.00757027e-01 -3.04879904e-01
2.51078278e-01 -5.84245265e-01 9.12275732e-01 3.59338254e-01
-1.26489234e+00 -3.12282175e-01 -8.77877593e-01 -1.68079093e-01
-3.39282721e-01 -2.42021322e-01 9.21307862e-01 4.53751057e-01
-9.26373899e-01 5.98967433e-01 -1.22508061e+00 -2.12508917e-01
8.94913733e-01 4.78750676e-01 -4.41808194e-01 -3.20461363e-01
-5.42415082e-01 5.79962432e-01 3.12072158e-01 -3.48488092e-02
-9.44506526e-01 -1.04631448e+00 -8.19784045e-01 -3.62859815e-01
4.36857373e-01 -8.22435319e-01 1.26627278e+00 -7.23048985e-01
-1.21015429e+00 1.31781626e+00 7.26457164e-02 -3.44276488e-01
5.15483618e-01 -1.62882149e-01 2.24076226e-01 3.71710420e-01
2.17361972e-01 9.23836291e-01 7.60458648e-01 -1.65014482e+00
-8.00344646e-01 -7.82456517e-01 2.29798317e-01 3.09668094e-01
2.31623337e-01 -5.36787033e-01 -8.42659056e-01 -6.27166852e-02
7.15258539e-01 -7.17469096e-01 -4.65991020e-01 4.30275261e-01
-6.15070820e-01 1.01797823e-02 7.84466982e-01 -4.69593704e-01
1.59003809e-01 -2.01227951e+00 1.23025969e-01 3.99234951e-01
4.50295061e-01 -3.13477337e-01 1.00562438e-01 -3.11748028e-01
9.75145102e-02 -9.45302844e-02 -3.82893175e-01 -5.66438317e-01
-4.01176624e-02 3.62459391e-01 -1.51121527e-01 7.40536988e-01
3.56239647e-01 1.08155632e+00 -9.43297267e-01 -4.73636419e-01
7.31672049e-01 5.96051574e-01 -6.35313272e-01 2.33798251e-01
-5.46023607e-01 5.36271513e-01 -6.72387540e-01 9.71652448e-01
7.31721342e-01 -5.82814753e-01 -1.77352741e-01 -4.42055672e-01
1.14673274e-02 2.31588468e-01 -1.10198474e+00 2.21918678e+00
-5.52564740e-01 3.23392004e-01 6.17888384e-02 -9.50827897e-01
6.49642766e-01 -5.64734004e-02 1.13437438e+00 -7.28815675e-01
1.32647425e-01 1.23418495e-01 -6.10769868e-01 -2.96004355e-01
1.99037224e-01 -1.43571943e-02 -1.20731585e-01 6.84625924e-01
1.30819872e-01 -6.59071922e-01 -3.00594628e-01 1.58574283e-01
1.13085997e+00 4.52171355e-01 -4.22213376e-02 1.40106276e-01
1.12792850e-01 2.04476416e-01 1.11039355e-01 7.11248219e-01
1.33677408e-01 8.07442784e-01 5.09086311e-01 -9.14062142e-01
-1.18344820e+00 -1.55750060e+00 -8.41372311e-02 6.69436455e-01
3.98681968e-01 9.05314162e-02 -6.89426839e-01 -8.41469884e-01
4.69721884e-01 3.96739781e-01 -8.23224664e-01 1.88877314e-01
-6.55901551e-01 -5.68918824e-01 3.47784400e-01 6.69907928e-01
5.20372927e-01 -7.28231609e-01 -1.18100095e+00 2.17854172e-01
1.57456055e-01 -1.33818114e+00 -2.26743948e-02 7.39812255e-01
-1.00676882e+00 -1.28682554e+00 -3.96497011e-01 -3.88187736e-01
6.15290523e-01 2.85580307e-01 1.49849439e+00 -4.11787117e-03
-6.29689157e-01 8.05782616e-01 -8.32074285e-02 -2.27082640e-01
-2.26729155e-01 6.28155619e-02 -2.23386407e-01 -1.79035887e-01
2.78489143e-01 -5.47407210e-01 -7.99636781e-01 -4.83400002e-02
-8.34727824e-01 3.44274074e-01 6.16790414e-01 4.55615968e-01
1.15119994e+00 -2.80392528e-01 -1.46348715e-01 -7.60510981e-01
-2.52447218e-01 -3.36465180e-01 -8.32767069e-01 7.35419020e-02
-2.56273538e-01 9.02007893e-02 -2.56254114e-02 2.16979217e-02
-5.77234268e-01 7.70325720e-01 -2.27616832e-01 -8.21950018e-01
-4.61273253e-01 2.23382916e-02 4.49596122e-02 -1.44614324e-01
3.34981829e-01 2.72810608e-01 2.93477457e-02 -5.06050825e-01
7.62441516e-01 4.35634732e-01 5.64097822e-01 -6.48158252e-01
7.45634496e-01 8.93207371e-01 2.81381756e-01 -5.62896609e-01
-1.13028240e+00 -4.44984227e-01 -1.00659621e+00 -3.92895669e-01
1.31871665e+00 -9.37312186e-01 -9.01764572e-01 3.75853926e-01
-1.18944442e+00 -5.84551752e-01 -4.75228786e-01 2.55606085e-01
-8.15704882e-01 -1.02219850e-01 -4.78983581e-01 -6.48529410e-01
-1.62053660e-01 -1.36767817e+00 1.80818689e+00 5.67672551e-02
2.17331991e-01 -7.16466963e-01 -3.26306045e-01 4.05439138e-01
1.49886116e-01 6.53936744e-01 1.00926578e+00 -3.72886211e-01
-1.01170576e+00 -1.10336453e-01 -5.32281220e-01 1.66065976e-01
1.49687333e-02 -2.27328017e-01 -1.14204597e+00 1.78721905e-01
-1.26274094e-01 -4.31514144e-01 8.31151366e-01 6.47019565e-01
1.51007831e+00 2.78424859e-01 -4.22109842e-01 9.47479308e-01
1.58536220e+00 -3.63424212e-01 1.72922045e-01 2.75738358e-01
1.06952024e+00 3.42477947e-01 4.36227471e-01 4.07555759e-01
5.87956607e-01 5.63553751e-01 1.02152896e+00 -4.46783304e-01
-2.31486812e-01 -1.60755113e-01 -1.83578566e-01 1.32830545e-01
1.35995466e-02 1.74550176e-01 -1.22938812e+00 4.37522352e-01
-1.50801551e+00 -4.14881587e-01 -6.94656819e-02 2.04379511e+00
6.20478809e-01 3.31801385e-01 2.01055318e-01 -1.74718034e-02
3.69606197e-01 1.08288839e-01 -1.00495267e+00 3.62931304e-02
1.40846938e-01 4.61252302e-01 1.09130943e+00 3.56669962e-01
-1.20744300e+00 8.00444484e-01 6.09042406e+00 2.86321521e-01
-1.33056152e+00 9.48755890e-02 1.03963542e+00 -5.06391644e-01
-4.43310201e-01 -3.35313529e-01 -6.47578537e-01 5.12796305e-02
5.82486272e-01 4.26450282e-01 5.40543795e-01 8.73157084e-01
-7.56022334e-02 -2.59726495e-01 -1.33513391e+00 1.07875478e+00
-7.82577172e-02 -1.59600270e+00 -1.56558871e-01 2.64159620e-01
6.89384222e-01 7.15357244e-01 -2.02068295e-02 -1.70412049e-01
5.09989500e-01 -1.14421523e+00 9.34726179e-01 3.22005481e-01
8.26210797e-01 -7.01009035e-01 3.81525397e-01 2.00060546e-01
-1.19784808e+00 1.41042635e-01 -1.23306833e-01 2.51932353e-01
2.19451219e-01 7.79204547e-01 -9.05600309e-01 4.54294711e-01
9.10918653e-01 8.70053172e-01 -4.87532407e-01 7.64933348e-01
-4.33226973e-02 2.14808211e-01 -7.98756361e-01 3.35013270e-01
4.20660943e-01 8.96936208e-02 2.16322899e-01 9.54891205e-01
2.14003205e-01 1.52461171e-01 4.71902817e-01 1.07603347e+00
-2.27938339e-01 -3.56294751e-01 -6.95561469e-01 2.01694712e-01
3.70002836e-01 1.28323293e+00 -1.25945866e+00 -3.70911926e-01
-5.15162647e-01 1.06157136e+00 2.41241902e-01 1.82181567e-01
-7.39749134e-01 2.18041494e-01 7.62900114e-01 1.98492870e-01
6.33992493e-01 -4.30318624e-01 -6.70016170e-01 -7.32756793e-01
1.41429221e-02 -2.75661558e-01 1.68468431e-01 -7.77835131e-01
-1.28848147e+00 3.38350922e-01 -6.69812560e-02 -1.05997074e+00
-2.35986769e-01 -8.80129993e-01 -1.60677150e-01 8.85563135e-01
-1.68942845e+00 -1.32493138e+00 -4.62751269e-01 6.32747650e-01
4.86641109e-01 3.95893306e-01 7.45794475e-01 5.22669218e-02
-7.10820779e-02 -8.04425962e-03 -2.09215805e-01 1.81855097e-01
1.35883629e-01 -1.51356959e+00 7.45341778e-01 3.64658535e-01
3.11070740e-01 4.86840904e-02 2.80158877e-01 -3.84312123e-01
-1.88360107e+00 -1.24472249e+00 6.09729253e-02 -7.93635786e-01
3.93939584e-01 -5.22159159e-01 -4.81274337e-01 6.55631840e-01
-3.43707561e-01 7.65862405e-01 4.36997831e-01 -2.03425497e-01
-4.53923166e-01 -7.16853142e-02 -1.35016203e+00 3.67626771e-02
1.22782958e+00 -5.52754343e-01 -2.91490167e-01 4.09597129e-01
9.48818266e-01 -1.08052540e+00 -1.08820152e+00 3.38069111e-01
6.17172718e-01 -1.16096258e+00 1.51567817e+00 -5.04597127e-01
6.24150634e-01 -3.73595119e-01 -6.33089185e-01 -1.01446378e+00
1.68377161e-01 1.32367373e-01 -1.41745374e-01 6.37230337e-01
2.30328634e-01 -1.91465661e-01 1.22770405e+00 6.99470758e-01
-2.05217078e-01 -1.07084835e+00 -1.06323886e+00 -3.20174575e-01
-6.62028939e-02 -9.64914799e-01 7.55254745e-01 5.00606239e-01
-7.84975290e-01 7.95575157e-02 2.18019575e-01 4.62283939e-01
9.87051725e-01 5.58317959e-01 8.35535288e-01 -1.19457388e+00
-3.58938664e-01 -4.26964104e-01 -5.56118548e-01 -1.11468887e+00
2.27134023e-02 -1.03399229e+00 6.47270307e-02 -1.67426920e+00
4.72491980e-02 -7.17438638e-01 -2.32813701e-01 5.84461331e-01
1.27549455e-01 8.45699310e-01 1.09693974e-01 -3.11834365e-02
-5.92100084e-01 1.12857036e-01 1.21806204e+00 -2.03355163e-01
-2.59907097e-01 -1.55367866e-01 -4.33860034e-01 8.06524456e-01
4.28349674e-01 -3.67292434e-01 6.44198852e-03 -7.40122497e-01
-7.19469914e-04 1.83240861e-01 7.63442099e-01 -9.90664780e-01
-8.21525902e-02 -1.95556134e-01 1.03901243e+00 -1.03950906e+00
6.74297810e-01 -1.09791529e+00 -3.35402340e-02 4.07830209e-01
-2.07327440e-01 -2.12369785e-01 1.72995940e-01 4.99006152e-01
3.06002468e-01 4.24742818e-01 7.75085151e-01 -6.63085401e-01
-8.31525087e-01 6.20284855e-01 2.56516516e-01 -1.46221384e-01
1.02055645e+00 -3.54434758e-01 1.07839026e-01 -5.70519082e-03
-7.79344082e-01 1.59881696e-01 7.41040051e-01 3.34652662e-01
7.51025379e-01 -1.26894963e+00 -4.27202523e-01 4.89332467e-01
5.01850732e-02 8.14538002e-01 6.31299987e-02 6.66328788e-01
-5.72342694e-01 1.89478621e-01 -1.31071448e-01 -1.40335155e+00
-7.92137742e-01 3.08479965e-01 6.68012679e-01 -4.68826406e-02
-8.90937746e-01 9.94029462e-01 2.25007072e-01 -6.77557349e-01
3.29042226e-01 -4.68495607e-01 3.98488641e-01 -1.89256698e-01
3.56130302e-01 -7.43088126e-02 3.67014974e-01 -6.33714318e-01
-5.95068097e-01 7.99381912e-01 9.70313177e-02 -2.87813693e-01
1.79009998e+00 -2.98534986e-02 -4.46228385e-02 5.11419117e-01
1.43159306e+00 -4.77144718e-01 -1.71568763e+00 -3.87785107e-01
-2.18719319e-01 -6.08477890e-01 4.17708069e-01 -7.22849846e-01
-1.47548318e+00 8.39470446e-01 8.45789135e-01 9.87370983e-02
9.92260993e-01 6.11124277e-01 8.49688530e-01 2.51555443e-01
5.30178666e-01 -7.33164608e-01 1.94714844e-01 3.60608995e-01
3.96315247e-01 -1.31902277e+00 1.79674506e-01 -3.38566095e-01
-2.18711883e-01 1.18155360e+00 4.89356816e-01 -3.60076487e-01
6.34981811e-01 3.29055756e-01 1.20792247e-01 -7.50946164e-01
-2.19540358e-01 -1.97881564e-01 2.40532756e-01 5.83135009e-01
1.45991594e-01 9.18924436e-02 6.89829230e-01 3.13116789e-01
-1.02537751e-01 -2.58124080e-02 1.46531820e-01 1.01160979e+00
-3.91718864e-01 -8.22769344e-01 -5.32284260e-01 7.11947441e-01
-2.43063658e-01 2.22306073e-01 -2.72639841e-01 7.76563585e-01
2.26148501e-01 3.59005541e-01 5.71378529e-01 -4.00806367e-01
4.11018670e-01 2.43839137e-02 8.49686801e-01 -7.16011167e-01
-3.06484044e-01 1.40345916e-01 -2.76667476e-01 -1.23765457e+00
-4.91382867e-01 -7.58693278e-01 -1.57598305e+00 -1.07660286e-01
1.40743554e-01 -5.41829646e-01 1.19529688e+00 1.02451277e+00
2.61793733e-01 5.38777113e-01 7.38156915e-01 -1.67761445e+00
-1.33158430e-01 -3.82905066e-01 -5.18460631e-01 3.34787220e-01
5.87624788e-01 -6.85385942e-01 -1.59521431e-01 1.24346120e-02]
|
[8.38707447052002, -2.9787211418151855]
|
a09e4b32-bd7b-4b1c-8779-488077479a52
|
face-recognition-in-movie-trailers-via-mean
| null | null |
http://openaccess.thecvf.com/content_cvpr_2013/html/Ortiz_Face_Recognition_in_2013_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2013/papers/Ortiz_Face_Recognition_in_2013_CVPR_paper.pdf
|
Face Recognition in Movie Trailers via Mean Sequence Sparse Representation-Based Classification
|
This paper presents an end-to-end video face recognition system, addressing the difficult problem of identifying a video face track using a large dictionary of still face images of a few hundred people, while rejecting unknown individuals. A straightforward application of the popular n-minimization for face recognition on a frame-by-frame basis is prohibitively expensive, so we propose a novel algorithm Mean Sequence SRC (MSSRC) that performs video face recognition using a joint optimization leveraging all of the available video data and the knowledge that the face track frames belong to the same individual. By adding a strict temporal constraint to the ii-minimization that forces individual frames in a face track to all reconstruct a single identity, we show the optimization reduces to a single minimization over the mean of the face track. We also introduce a new Movie Trailer Face Dataset collected from 101 movie trailers on YouTube. Finally, we show that our method matches or outperforms the state-of-the-art on three existing datasets (YouTube Celebrities, YouTube Faces, and Buffy) and our unconstrained Movie Trailer Face Dataset. More importantly, our method excels at rejecting unknown identities by at least 8% in average precision.
|
['Enrique. G. Ortiz', 'Mubarak Shah', 'Alan Wright']
|
2013-06-01
| null | null | null |
cvpr-2013-6
|
['sparse-representation-based-classification']
|
['computer-vision']
|
[ 2.16603354e-01 -4.83287692e-01 -1.96355447e-01 -7.21267760e-01
-7.46047497e-01 -7.57629812e-01 3.35665911e-01 -6.38270795e-01
-3.75382006e-01 5.49653947e-01 -2.49854073e-01 2.56755710e-01
9.49190930e-02 -1.62847117e-01 -8.28175128e-01 -8.33641231e-01
-7.79600292e-02 3.85242462e-01 -1.85003951e-01 1.15181863e-01
1.93394169e-01 6.80480242e-01 -1.97463191e+00 5.17242908e-01
9.28967446e-02 1.50924587e+00 -3.30094188e-01 6.72410131e-01
1.91696808e-01 9.03566837e-01 -4.35937107e-01 -9.69463050e-01
6.99887872e-01 -3.95298928e-01 -5.03831148e-01 3.61136347e-01
1.54023492e+00 -7.57063746e-01 -6.75375640e-01 1.03184044e+00
5.90864599e-01 1.24269851e-01 3.38094205e-01 -1.50581181e+00
-4.66755152e-01 1.77126052e-03 -5.71310103e-01 3.59077156e-01
5.23409724e-01 -5.26492484e-02 6.09929204e-01 -1.29663956e+00
7.71357715e-01 1.36143470e+00 7.99544513e-01 7.90372729e-01
-1.04426944e+00 -1.06771660e+00 2.02318445e-01 4.54083800e-01
-1.94687808e+00 -1.40203583e+00 4.49624300e-01 -3.82212251e-01
5.79178393e-01 3.87965739e-01 5.15206397e-01 1.04348516e+00
-2.95515418e-01 6.96756899e-01 6.16256416e-01 -8.91548172e-02
3.62471528e-02 -9.02876705e-02 -6.88110739e-02 9.56761718e-01
-3.16049717e-02 7.85440020e-03 -1.18798661e+00 -3.55147302e-01
6.26272738e-01 1.84870940e-02 -3.95973295e-01 -1.24939963e-01
-7.80822396e-01 7.20574021e-01 -2.97096878e-01 -1.03810966e-01
-1.41911119e-01 -1.16695678e-02 2.33060345e-01 4.99057204e-01
5.34009814e-01 -4.78231400e-01 -2.24636823e-01 -8.91428441e-02
-1.39913487e+00 2.63259381e-01 1.09931791e+00 8.95636618e-01
7.06193507e-01 3.01003516e-01 6.27902001e-02 6.25665009e-01
2.76477158e-01 8.23731303e-01 2.05360278e-01 -1.29504943e+00
8.00625980e-02 1.00086592e-02 -1.60113424e-02 -1.21358812e+00
2.94136971e-01 7.83850402e-02 -4.36477035e-01 1.65367231e-01
5.84698319e-01 -1.13043100e-01 -7.17549205e-01 1.74239135e+00
5.77075362e-01 8.18961799e-01 -2.20852464e-01 1.05279267e+00
8.33856463e-01 3.32221538e-01 -2.58794010e-01 -7.41234303e-01
1.23127341e+00 -6.65805519e-01 -7.74826407e-01 7.22361580e-02
6.72618151e-02 -8.91563594e-01 2.07377836e-01 5.00080228e-01
-1.03719354e+00 -5.21931827e-01 -7.61220872e-01 5.11291474e-02
1.24960870e-01 5.10200679e-01 4.41637009e-01 9.93725121e-01
-1.10953009e+00 5.31215966e-01 -4.79656726e-01 -2.09113061e-01
7.29664624e-01 9.37740743e-01 -8.94647837e-01 -2.73618490e-01
-5.33342063e-01 3.23609889e-01 -2.51114994e-01 1.92009360e-01
-1.15502489e+00 -7.01550364e-01 -7.39168406e-01 -1.51572302e-01
5.80919087e-01 -2.89168060e-01 1.04497552e+00 -1.67722082e+00
-1.50377393e+00 1.14265466e+00 -7.47099340e-01 -2.72031426e-01
5.91319561e-01 -3.10031146e-01 -7.03429163e-01 5.30841291e-01
4.98585850e-02 3.95820796e-01 1.54971194e+00 -9.07407939e-01
-5.66070020e-01 -7.51119912e-01 -1.81112036e-01 -1.65451895e-02
-5.42443335e-01 5.68824887e-01 -8.87742698e-01 -8.33839297e-01
-1.38034791e-01 -1.08308458e+00 4.78852630e-01 4.95308757e-01
-6.49452880e-02 -1.22487567e-01 1.10379851e+00 -7.42462933e-01
1.06534064e+00 -2.36299157e+00 1.29472822e-01 2.54251212e-01
1.86296433e-01 3.58857691e-01 -3.02425623e-01 3.45448852e-02
-1.75967008e-01 -2.15373904e-01 4.04100791e-02 -7.28711247e-01
-2.84681380e-01 3.78069878e-02 -3.87002409e-01 1.08325565e+00
-9.22726020e-02 4.37035143e-01 -6.61615431e-01 -5.59851468e-01
-1.35937274e-01 7.02377856e-01 -5.98534226e-01 1.97340563e-01
1.35714889e-01 2.77135879e-01 -1.17869481e-01 1.07561028e+00
8.63430023e-01 2.08819304e-02 3.94131392e-01 -3.23879868e-01
7.44632930e-02 -4.15536731e-01 -1.45635462e+00 1.50179434e+00
1.93804771e-01 7.51183152e-01 4.50652272e-01 -7.77854443e-01
7.85066426e-01 3.87948185e-01 8.90161753e-01 -2.88178563e-01
7.48393238e-02 1.93961725e-01 -3.59361112e-01 -6.16610050e-01
2.34138429e-01 -9.69254063e-04 5.09530365e-01 4.90264505e-01
4.21180040e-01 6.03505909e-01 3.01088333e-01 1.72865823e-01
9.07145739e-01 9.44992453e-02 -1.77816153e-01 -2.31951073e-01
7.76280344e-01 -3.99681062e-01 7.99262702e-01 7.00964272e-01
-3.29821765e-01 7.09255636e-01 2.29052782e-01 -8.30117583e-01
-6.59378946e-01 -1.02704453e+00 -3.35736647e-02 1.10731781e+00
-1.01936787e-01 -6.06662691e-01 -8.62316489e-01 -9.20903802e-01
2.14623272e-01 -1.52387619e-01 -6.15754545e-01 1.84586793e-01
-6.92289233e-01 -5.03004968e-01 6.92257464e-01 2.28155479e-01
4.79326516e-01 -6.59974635e-01 -1.92544013e-01 -1.59170806e-01
-2.46321633e-01 -1.38488424e+00 -1.16602540e+00 -7.01578379e-01
-4.07537192e-01 -1.42657876e+00 -8.05447042e-01 -1.01526213e+00
8.47530544e-01 4.18531299e-01 9.72823739e-01 3.14308435e-01
-5.20015299e-01 6.95552588e-01 -1.38157517e-01 8.58112350e-02
4.73416522e-02 -5.58456182e-01 7.17324436e-01 9.58363295e-01
5.52129388e-01 -2.18881667e-01 -4.66223806e-01 6.08379126e-01
-5.09608448e-01 -5.72383821e-01 -1.86046630e-01 6.69149041e-01
6.25733972e-01 2.14274204e-03 4.81325746e-01 -6.03026092e-01
-1.46539807e-02 -3.48979771e-01 -6.85482979e-01 4.77695465e-01
-1.20874960e-02 -5.01963615e-01 4.37464476e-01 -8.22747052e-01
-8.72763216e-01 6.57268465e-01 7.06581697e-02 -1.26586461e+00
1.28202304e-01 -1.21556453e-01 -9.71405134e-02 -6.84776485e-01
2.95399159e-01 3.27270269e-01 4.31934744e-01 -4.29153591e-01
2.18943600e-02 6.19621336e-01 8.96813691e-01 -4.21427190e-01
7.87602186e-01 9.16514099e-01 2.21423302e-02 -1.14547503e+00
-6.52931511e-01 -5.29646754e-01 -6.84304297e-01 -6.78025067e-01
6.29131496e-01 -1.16947949e+00 -1.52840233e+00 6.51253164e-01
-1.07032871e+00 2.39425488e-02 -3.94835398e-02 3.86314869e-01
-4.81710434e-01 6.64975584e-01 -4.82021719e-01 -1.02033913e+00
-2.31992543e-01 -1.02540839e+00 1.23303103e+00 6.99035749e-02
-2.92251967e-02 -5.15514791e-01 -1.37464464e-01 3.15184116e-01
1.19128615e-01 1.01530604e-01 5.66777810e-02 -5.07510662e-01
-5.81779838e-01 -3.15192729e-01 -5.27014676e-03 3.80954981e-01
1.57083109e-01 2.81689376e-01 -1.00061870e+00 -5.83700240e-01
2.52280980e-01 -4.37907070e-01 8.84623289e-01 1.95810437e-01
1.23063457e+00 -5.25780737e-01 -2.54406780e-01 8.46555769e-01
1.26755941e+00 1.69414639e-01 5.19545376e-01 -2.69316435e-01
7.31585741e-01 7.38554060e-01 4.58087057e-01 7.70237803e-01
2.04635933e-01 1.00903606e+00 1.11203894e-01 3.71248394e-01
-8.11901838e-02 1.02344736e-01 7.33495891e-01 4.13502485e-01
-2.31621996e-01 -1.38490602e-01 -4.00169641e-01 3.86017263e-01
-1.86911893e+00 -1.49505806e+00 1.51912138e-01 2.51625490e+00
4.56043184e-01 -5.31361103e-01 4.72332239e-01 2.59098457e-03
9.71551359e-01 6.91815373e-03 -5.94224036e-01 1.15960509e-01
-4.11495060e-01 2.29834780e-01 3.32541287e-01 5.05155087e-01
-1.28578198e+00 8.38065982e-01 6.67114973e+00 8.64942372e-01
-1.21555138e+00 6.38705269e-02 7.78429747e-01 -6.63190603e-01
2.56603956e-01 -1.97591841e-01 -1.20026588e+00 5.81421733e-01
8.84490728e-01 -1.69178918e-01 9.31162894e-01 7.04104364e-01
2.05835607e-03 1.05177753e-01 -1.29166102e+00 1.68802631e+00
9.41314161e-01 -1.35277772e+00 4.96254750e-02 1.84017360e-01
5.37983119e-01 -2.59577602e-01 2.19910383e-01 -1.47999808e-01
-1.61800668e-01 -1.09834826e+00 9.72473621e-01 6.48792088e-01
1.13845170e+00 -6.26837194e-01 3.65979999e-01 -6.30924404e-02
-1.36408424e+00 -1.69723704e-01 -2.96485096e-01 2.63043135e-01
1.78649366e-01 3.05716574e-01 -2.68077910e-01 3.25529814e-01
9.32364821e-01 8.80413473e-01 -3.87545466e-01 7.63181627e-01
3.24961841e-01 3.51577193e-01 -4.96372908e-01 5.28001904e-01
-2.51100034e-01 -1.30658045e-01 5.04202008e-01 9.97419298e-01
3.95942986e-01 4.85865772e-01 3.34922135e-01 2.69888341e-01
-4.05598491e-01 1.04445517e-01 -5.55998385e-01 4.09590341e-02
4.00474042e-01 1.10519993e+00 -4.69902068e-01 -3.82133484e-01
-5.97276270e-01 1.21874750e+00 1.53315753e-01 3.42639983e-01
-8.00633550e-01 7.72747025e-02 1.08983684e+00 2.98024286e-02
6.79350555e-01 -8.97672847e-02 4.57980305e-01 -1.36311877e+00
3.56009692e-01 -1.28271139e+00 5.35717309e-01 -3.18119049e-01
-1.34955502e+00 7.84832835e-01 -3.44229966e-01 -1.15394413e+00
-2.93327332e-01 -6.10633671e-01 -3.33696514e-01 6.27383292e-01
-1.24110734e+00 -9.51538861e-01 -3.15106988e-01 1.28064013e+00
6.04066312e-01 -6.25080049e-01 7.20281780e-01 7.85166979e-01
-7.51169682e-01 9.46571469e-01 4.38984819e-02 3.13469440e-01
8.37069392e-01 -4.05570060e-01 1.77619159e-01 7.17253566e-01
4.10072833e-01 6.08593583e-01 3.64375323e-01 -7.07355320e-01
-1.95802081e+00 -9.50487256e-01 8.92263472e-01 -4.94985104e-01
2.80346692e-01 -4.75624263e-01 -6.41674459e-01 7.60524333e-01
-1.15038887e-01 6.78796411e-01 8.71003687e-01 -2.87888199e-01
-5.50752163e-01 -3.76027703e-01 -1.43661702e+00 2.30915353e-01
1.27694678e+00 -7.22072303e-01 7.50362277e-02 5.08265793e-01
-1.36927171e-02 -2.49322370e-01 -7.85565734e-01 3.35999638e-01
1.14419806e+00 -8.94820988e-01 1.08510172e+00 -7.04336822e-01
-8.07918012e-02 -3.88680637e-01 -5.12239099e-01 -4.53192800e-01
2.83217188e-02 -1.02201152e+00 -5.01648664e-01 1.40428412e+00
-1.88057348e-01 -3.28869224e-01 1.16713810e+00 7.27528036e-01
4.65275973e-01 -4.52566922e-01 -1.37032259e+00 -9.59395349e-01
-6.14207506e-01 -1.56848446e-01 4.30763304e-01 9.30824339e-01
-3.36077839e-01 -2.45452315e-01 -8.94438982e-01 1.85499966e-01
1.00355518e+00 -5.45452349e-02 8.61181438e-01 -1.13699913e+00
-2.20584795e-01 1.58208702e-02 -6.48677588e-01 -1.00829685e+00
7.28896976e-01 -7.18668699e-01 -1.34631231e-01 -5.57083607e-01
2.29928538e-01 -1.52972922e-01 1.02636166e-01 3.91365230e-01
1.18310243e-01 9.52206910e-01 4.19931173e-01 3.80848795e-01
-6.48080349e-01 3.18508267e-01 6.12445116e-01 -2.47771934e-01
1.70581117e-01 -4.56004925e-02 -3.22637409e-01 7.63951063e-01
3.39819372e-01 -5.47291696e-01 -1.52056590e-01 -5.68612516e-01
-3.43075573e-01 1.64853200e-01 3.86839598e-01 -8.85094464e-01
5.59175611e-01 9.12889913e-02 6.29201055e-01 -2.88929492e-01
8.31643879e-01 -8.20391059e-01 4.77851033e-01 2.15532050e-01
-4.23362851e-02 2.22235724e-01 1.82716385e-01 4.94348556e-01
-1.24281727e-01 -1.45624161e-01 8.99827838e-01 -3.32329131e-04
-8.44983578e-01 8.20553362e-01 -1.80075243e-01 4.76657599e-03
1.18861091e+00 -3.41877818e-01 -1.38791978e-01 -4.65196311e-01
-7.77586162e-01 -1.12686278e-02 4.47234988e-01 5.51424742e-01
7.87574589e-01 -1.32368946e+00 -9.47113574e-01 6.83388174e-01
-2.00035185e-01 -8.22614312e-01 4.11329359e-01 7.18887329e-01
-2.67270863e-01 4.33012843e-02 -2.35269770e-01 -7.04715133e-01
-2.11702633e+00 4.89786983e-01 2.96825022e-01 5.26904762e-01
-3.88114542e-01 1.16822922e+00 9.40437689e-02 1.38942614e-01
4.48664993e-01 6.85538352e-01 2.80069131e-02 2.60284334e-01
1.13636792e+00 4.46040183e-01 6.81819916e-02 -1.45807135e+00
-6.99588418e-01 8.77399623e-01 -1.87941834e-01 -4.21903171e-02
1.13342881e+00 -1.30991340e-01 -2.21061245e-01 -1.06917076e-01
1.44088495e+00 7.93540552e-02 -1.42902648e+00 -1.39418080e-01
-3.98981243e-01 -1.14957035e+00 -1.71671450e-01 -3.56055528e-01
-1.67194414e+00 3.25734079e-01 7.07962155e-01 -2.84842193e-01
1.20469832e+00 -2.37396270e-01 6.93159640e-01 4.43238318e-01
5.17867446e-01 -8.39266896e-01 5.85298166e-02 2.82903165e-01
7.49754965e-01 -1.28088188e+00 1.19305514e-01 -6.68967068e-01
-4.33192015e-01 1.21927202e+00 3.99942100e-01 -1.01514727e-01
8.06850612e-01 2.26350561e-01 -2.91631334e-02 -5.02481461e-02
-6.50065422e-01 5.81610994e-03 2.53803790e-01 5.05726278e-01
1.49644449e-01 -3.07303190e-01 1.85250238e-01 2.98018724e-01
3.54197174e-02 3.59396905e-01 2.30170459e-01 7.85360992e-01
-1.93767250e-01 -1.00292039e+00 -5.52082419e-01 4.82189626e-01
-8.97144854e-01 -3.49151604e-02 -3.97826493e-01 3.54189008e-01
1.85024917e-01 1.10938656e+00 1.87048405e-01 -3.33823115e-01
-4.78460751e-02 4.14884016e-02 7.61203289e-01 -2.73971170e-01
-5.07107675e-01 5.24279140e-02 7.90422484e-02 -8.47001076e-01
-7.09092081e-01 -1.04882538e+00 -6.83448553e-01 -8.22674572e-01
-1.97672516e-01 -2.30346862e-02 4.94203985e-01 7.99737513e-01
4.19312030e-01 -3.42275232e-01 9.25457835e-01 -1.04613221e+00
-3.65322858e-01 -3.35388213e-01 -7.66290605e-01 5.09889543e-01
5.65701544e-01 -6.86644137e-01 -4.18905854e-01 6.46406710e-01]
|
[13.299652099609375, 0.7752413153648376]
|
5b176237-5a2c-4084-b7b2-31937f997b87
|
transformer-training-strategies-for
|
2306.10891
| null |
https://arxiv.org/abs/2306.10891v1
|
https://arxiv.org/pdf/2306.10891v1.pdf
|
Transformer Training Strategies for Forecasting Multiple Load Time Series
|
Recent work uses Transformers for load forecasting, which are the state of the art for sequence modeling tasks in data-rich domains. In the smart grid of the future, accurate load forecasts must be provided on the level of individual clients of an energy supplier. While the total amount of electrical load data available to an energy supplier will increase with the ongoing smart meter rollout, the amount of data per client will always be limited. We test whether the Transformer benefits from a transfer learning strategy, where a global model is trained on the load time series data from multiple clients. We find that the global model is superior to two other training strategies commonly used in related work: multivariate models and local models. A comparison to linear models and multi-layer perceptrons shows that Transformers are effective for electrical load forecasting when they are trained with the right strategy.
|
['Veit Hagenmeyer', 'Ralf Mikut', 'Benjamin Schäfer', 'Oliver Neumann', 'Benedikt Heidrich', 'Maximilian Beichter', 'Matthias Hertel']
|
2023-06-19
| null | null | null | null |
['load-forecasting']
|
['miscellaneous']
|
[-5.83795235e-02 -2.21546948e-01 -2.60737091e-01 -3.78093004e-01
-5.66210926e-01 -5.24224937e-01 5.97110868e-01 5.71818799e-02
3.75202484e-02 6.61568940e-01 3.80252838e-01 -6.94839358e-01
5.84117062e-02 -1.08311903e+00 -2.39643499e-01 -8.78301799e-01
-7.39859715e-02 9.67171967e-01 -6.12947494e-02 -2.98731565e-01
1.96024075e-01 4.80980784e-01 -1.08780766e+00 4.52759594e-01
7.22124457e-01 1.25592768e+00 4.47315305e-01 5.81728220e-01
-3.09873044e-01 1.24148500e+00 -7.41010368e-01 1.74493700e-01
3.57870698e-01 -3.87169570e-01 -6.10295534e-01 -8.15586969e-02
-2.78050095e-01 -7.31891751e-01 -2.62386382e-01 6.77168608e-01
6.91874385e-01 -2.33920783e-01 6.43862128e-01 -1.34303737e+00
-2.35065117e-01 9.30002332e-01 -3.02128851e-01 4.92980808e-01
1.75372481e-01 1.65918052e-01 9.13407803e-01 -2.86362231e-01
-8.32005814e-02 7.32112408e-01 8.05975974e-01 -1.26330823e-01
-1.32593369e+00 -6.71860039e-01 1.20409288e-01 7.47603536e-01
-9.24907386e-01 -2.82624453e-01 9.46045995e-01 -5.11901736e-01
1.63525605e+00 2.18288928e-01 5.52885175e-01 6.55865192e-01
2.74792910e-01 6.28807545e-01 1.00685596e+00 -3.81324232e-01
5.52098334e-01 1.73337340e-01 -1.00557841e-01 -1.62108600e-01
-1.98397502e-01 5.99835031e-02 -4.50228840e-01 -4.20656115e-01
4.85817492e-01 9.37791690e-02 -1.74709693e-01 -2.19544277e-01
-9.95421886e-01 9.74283516e-01 1.46670686e-02 7.46373236e-01
-7.98428476e-01 -1.05312847e-01 8.48730445e-01 6.49301887e-01
6.52704239e-01 1.15890779e-01 -9.85543847e-01 -5.81887782e-01
-1.27031839e+00 -1.49889439e-01 1.38972187e+00 8.13187659e-01
4.73151356e-01 6.30313873e-01 1.65844917e-01 4.71451849e-01
5.14702015e-02 2.84730017e-01 8.60468626e-01 -5.71815670e-01
5.79481661e-01 4.99027699e-01 2.40204588e-01 -4.25744116e-01
-5.67268729e-01 -5.19344151e-01 -7.84897029e-01 3.42382818e-01
6.26768649e-01 -3.80950838e-01 -3.75197947e-01 1.10536849e+00
-1.32332444e-01 2.66471714e-01 7.70135075e-02 1.49586692e-01
-1.47377383e-02 9.61901546e-01 1.48613483e-01 -4.40280467e-01
8.29870224e-01 -7.70316422e-01 -6.36356354e-01 -1.57336146e-01
1.02895868e+00 -6.28330410e-01 5.11509240e-01 6.40075028e-01
-9.29928362e-01 -4.23047304e-01 -8.95930171e-01 4.47340012e-01
-4.50539261e-01 8.24450925e-02 1.24587126e-01 7.12322116e-01
-9.40333486e-01 9.02249098e-01 -1.04545975e+00 -3.44307989e-01
1.02230616e-01 2.65358925e-01 2.42150858e-01 1.63096905e-01
-1.13224757e+00 1.54309595e+00 4.88295734e-01 7.82249868e-02
-7.93254197e-01 -8.75123620e-01 -5.37291646e-01 4.80338186e-01
-1.68727472e-01 -8.34619775e-02 1.61660731e+00 -9.59941685e-01
-1.55624461e+00 -4.15061004e-02 -5.33059500e-02 -7.42186427e-01
4.95060980e-01 2.12353483e-01 -5.39388478e-01 -2.41735190e-01
-4.64663178e-01 -3.14072877e-01 6.17977798e-01 -7.96404541e-01
-9.61743832e-01 -2.12523192e-01 -3.71598512e-01 -2.84359492e-02
-3.61268669e-01 4.26664837e-02 7.37057149e-01 -3.80066752e-01
-3.84989887e-01 -6.20543957e-01 -2.75070101e-01 -6.33292675e-01
4.93912138e-02 -7.30651915e-01 1.35919595e+00 -1.49795938e+00
1.12036645e+00 -1.74205673e+00 -2.15930760e-01 3.58412027e-01
-2.68932492e-01 2.29826480e-01 2.47501284e-01 1.21867716e+00
-5.54249465e-01 -1.91481188e-01 1.75582811e-01 -9.14754346e-02
3.73346865e-01 2.59941518e-01 -4.83176559e-01 4.09979314e-01
-3.91885310e-01 9.84607339e-01 -7.00174689e-01 6.73522800e-02
5.70979774e-01 1.58852130e-01 1.11520089e-01 2.30081677e-01
-3.94339144e-01 3.47087801e-01 -2.59929687e-01 2.80560493e-01
3.13384533e-01 -5.25485337e-01 5.54072499e-01 -1.65180027e-01
2.84133758e-02 7.06117511e-01 -1.00038397e+00 1.25989544e+00
-7.32948482e-01 5.38282096e-01 -2.30796978e-01 -1.51254678e+00
9.60389972e-01 8.14427793e-01 9.25243020e-01 -9.22297180e-01
-9.96589437e-02 1.69918358e-01 3.13221067e-02 -3.89917105e-01
1.00433588e-01 -2.82162070e-01 2.90410072e-01 9.06115592e-01
-1.78561181e-01 1.80972889e-01 3.61679971e-01 -2.05066323e-01
1.19979072e+00 7.62229264e-02 2.46073455e-01 -2.63192564e-01
3.30118120e-01 -2.23091878e-02 5.06826699e-01 2.71730214e-01
-2.55893245e-02 -7.90590644e-02 4.08148617e-01 -5.30259490e-01
-1.59716201e+00 -6.87902749e-01 -7.19967410e-02 1.17050195e+00
-6.64774835e-01 -3.96264017e-01 -4.01662827e-01 -4.97756869e-01
1.23904385e-01 1.37284327e+00 -9.03598294e-02 2.43484318e-01
-6.15673959e-01 -8.91723812e-01 2.17697285e-02 9.56112385e-01
1.79138109e-01 -1.16880667e+00 -6.37670159e-01 8.87502670e-01
-1.41024515e-01 -9.18272018e-01 -3.12452853e-01 5.23984492e-01
-9.97832954e-01 -7.25387275e-01 -6.60259843e-01 -5.82816184e-01
1.90281361e-01 -2.97288239e-01 1.45554101e+00 -3.08988333e-01
8.44073519e-02 3.92878592e-01 -2.91548997e-01 -6.22244000e-01
-7.04468310e-01 1.78134158e-01 -1.05288982e-01 -2.31541798e-01
6.20237768e-01 -1.14448750e+00 -4.91900444e-01 2.68148512e-01
-5.57942808e-01 1.11646399e-01 3.74878109e-01 5.49443543e-01
-1.76357880e-01 6.89073622e-01 1.16546464e+00 -5.45621037e-01
5.37740648e-01 -6.66884184e-01 -8.00117314e-01 2.65715092e-01
-1.07465303e+00 -1.00782193e-01 9.66411471e-01 -3.97367567e-01
-9.40104902e-01 5.01374423e-04 -7.04191774e-02 -8.88892487e-02
-1.56278491e-01 6.82283342e-01 1.04921833e-02 2.56554335e-01
2.54102886e-01 7.11360097e-01 -1.53740617e-02 -7.43810356e-01
1.24929361e-01 9.19571221e-01 6.18362017e-02 -1.31180435e-01
6.66045487e-01 8.91659129e-03 -2.15109169e-01 -7.11464942e-01
-3.71503644e-02 -4.46846008e-01 -5.67257404e-01 -1.12727575e-01
4.27710593e-01 -9.21261728e-01 -9.02725935e-01 6.92329109e-01
-8.22162986e-01 -7.37076819e-01 -5.18896282e-01 4.54554617e-01
-6.58135653e-01 4.19956982e-01 -7.62538552e-01 -9.62663293e-01
-5.26581466e-01 -9.41929579e-01 6.25869989e-01 -1.17165871e-01
-4.27507132e-01 -1.49774373e+00 6.37705848e-02 1.97719365e-01
9.71626997e-01 -5.21820374e-02 1.34570384e+00 -1.13151908e+00
-4.10191268e-01 -4.37387556e-01 1.21934414e-01 6.71600044e-01
5.11957645e-01 -4.70432967e-01 -7.80526221e-01 -5.57095826e-01
2.17990696e-01 -1.05079733e-01 7.97167718e-02 2.86173463e-01
9.62940037e-01 -5.67635417e-01 -2.51788765e-01 -1.05288014e-01
1.55516410e+00 6.31272018e-01 6.02364063e-01 3.06099713e-01
3.47876161e-01 4.42351162e-01 -2.26222262e-01 7.04772294e-01
8.45217228e-01 4.89879668e-01 2.19738051e-01 2.26754118e-02
3.49057019e-01 -1.49502799e-01 4.31670964e-01 1.12737060e+00
1.96266174e-01 -1.56228647e-01 -1.09456289e+00 7.04787016e-01
-2.06898761e+00 -1.18901420e+00 1.65662497e-01 2.22597313e+00
6.78468049e-01 1.15052879e-01 4.63687360e-01 5.75039804e-01
3.62784386e-01 -5.40332459e-02 -7.14205086e-01 -3.61425877e-01
1.21811152e-01 3.19706872e-02 6.97025478e-01 2.95389980e-01
-5.03132880e-01 -3.33763137e-02 7.01339722e+00 6.96678936e-01
-1.04661584e+00 2.24477593e-02 6.52511656e-01 -4.50446606e-02
-1.31585717e-01 3.68041582e-02 -6.80710733e-01 1.05191648e+00
1.75977540e+00 -6.66139960e-01 6.87080860e-01 7.71678209e-01
6.64347529e-01 -2.25053698e-01 -1.48058116e+00 9.08423841e-01
-1.15803853e-01 -1.21529317e+00 -4.32712197e-01 2.79195249e-01
6.09391749e-01 4.60034788e-01 -4.64161724e-01 2.85399616e-01
6.23135269e-01 -1.05767810e+00 3.85411024e-01 7.09289014e-01
-7.81098232e-02 -7.55711019e-01 8.64738643e-01 9.98589814e-01
-1.32367921e+00 -4.99298632e-01 2.32236534e-02 -1.88869491e-01
5.54900050e-01 6.72551453e-01 -1.03396928e+00 8.10943961e-01
7.79382586e-01 7.21508622e-01 -2.58210331e-01 8.78790438e-01
1.31531820e-01 1.04845178e+00 -6.06097341e-01 7.32610077e-02
4.27831002e-02 -3.95410091e-01 -1.09280311e-01 9.36559975e-01
3.11803728e-01 -2.01172963e-01 5.43437839e-01 4.31646585e-01
4.67217207e-01 5.24175093e-02 -4.28439319e-01 -2.75084823e-02
2.91436613e-01 1.12973380e+00 -3.88693094e-01 -6.41849220e-01
-8.17058206e-01 5.87646782e-01 4.14068736e-02 4.00703698e-01
-2.99971819e-01 -1.14681043e-01 2.57918537e-01 1.12021923e-01
6.34980321e-01 -4.19446141e-01 -4.84621406e-01 -8.09456587e-01
-1.02385871e-01 -9.88647163e-01 4.88001764e-01 -9.71336365e-01
-1.72670996e+00 1.19699493e-01 -4.08858694e-02 -1.00062692e+00
-1.19448173e+00 -4.15117621e-01 -1.03873205e+00 1.25975907e+00
-1.60009623e+00 -9.14005637e-01 1.61315322e-01 4.97494072e-01
6.19437695e-01 -2.85482347e-01 9.04753923e-01 2.44755000e-01
-1.41643137e-01 3.16493288e-02 6.83253348e-01 1.94411099e-01
-5.91538497e-04 -1.55369747e+00 4.54672128e-01 4.79332864e-01
6.83203340e-02 -2.09109947e-01 8.68788779e-01 -5.94834089e-01
-1.42843139e+00 -6.63115561e-01 1.19730353e+00 -1.69731766e-01
9.47743833e-01 -1.54926747e-01 -1.12192249e+00 1.15971088e+00
8.01604390e-01 -5.12793839e-01 7.30985940e-01 3.96344550e-02
1.13035487e-02 -3.40802670e-01 -1.06488657e+00 1.90504245e-03
9.13488120e-02 -6.66852415e-01 -6.09577000e-01 5.57973683e-01
3.54762711e-02 4.15275171e-02 -1.38804436e+00 4.87100556e-02
3.50802958e-01 -6.59960330e-01 6.04637682e-01 -3.44488651e-01
-1.59617022e-01 -1.07180558e-01 -2.43369535e-01 -1.85412848e+00
-4.95415092e-01 -7.09788620e-01 -4.03036207e-01 1.20177865e+00
5.06520271e-01 -1.15652597e+00 8.86078060e-01 8.47333252e-01
1.55327141e-01 -4.08268154e-01 -9.62025046e-01 -9.45869803e-01
4.92968202e-01 -3.59148383e-01 1.09010661e+00 8.72751713e-01
5.16235650e-01 3.88514012e-01 -2.44732261e-01 8.13934058e-02
5.88670254e-01 3.53919029e-01 4.04147595e-01 -1.18906379e+00
-4.09386963e-01 -4.39901173e-01 -2.58873373e-01 -7.76510119e-01
8.91851559e-02 -7.64996648e-01 -3.79360437e-01 -1.81054127e+00
-9.11317244e-02 -3.21055263e-01 -3.46324027e-01 6.51185632e-01
4.06898499e-01 -3.42058510e-01 3.62187743e-01 1.71754047e-01
2.35221550e-01 4.35312361e-01 4.82948482e-01 -2.11749241e-01
-2.05921859e-01 3.61969173e-01 -1.51767552e-01 6.59169436e-01
1.26248538e+00 -1.51592568e-01 -5.79643250e-01 -2.10932761e-01
2.04735249e-01 3.71227115e-01 -1.99603941e-02 -8.16072941e-01
5.62174439e-01 -5.41091524e-02 7.29092300e-01 -9.07990515e-01
1.15730584e-01 -1.23170757e+00 6.90121174e-01 5.66433728e-01
-6.26836494e-02 5.56923091e-01 1.06710717e-01 2.99025118e-01
2.44763270e-02 -2.34500632e-01 4.40002978e-01 -2.95960754e-01
-5.60930789e-01 1.44245222e-01 -7.26688921e-01 -5.08476555e-01
9.36052024e-01 -9.22579691e-02 -1.05072677e-01 -8.17429125e-01
-8.02582264e-01 3.76904964e-01 4.53760400e-02 3.92024547e-01
-6.02453202e-02 -1.23557341e+00 -9.08256352e-01 1.92940265e-01
-2.44296625e-01 -5.18042862e-01 7.07509294e-02 8.07320654e-01
-3.93116623e-02 6.37170911e-01 -2.17304230e-02 -4.85728443e-01
-8.16831410e-01 6.89323425e-01 5.89321792e-01 -7.71488488e-01
-8.75037849e-01 -1.53740540e-01 -2.02900156e-01 -2.17029184e-01
2.70675898e-01 -4.32515740e-01 -2.28711039e-01 2.91644722e-01
6.71563387e-01 8.17054212e-01 5.85391641e-01 -4.29074615e-01
4.74927723e-02 1.03928551e-01 1.67513415e-02 6.77947775e-02
1.68382883e+00 -2.04252452e-01 -7.60809779e-02 8.79579902e-01
9.14592564e-01 -6.32149577e-01 -1.06056798e+00 -4.62072194e-01
3.75870705e-01 -5.34347109e-02 1.98163807e-01 -1.14612865e+00
-1.11227977e+00 7.91914999e-01 6.86988533e-01 9.76614475e-01
1.17366230e+00 -3.01978499e-01 9.17057753e-01 7.28374496e-02
4.63363469e-01 -1.02400339e+00 -5.15514016e-01 2.27147877e-01
7.21184194e-01 -8.20280373e-01 -9.34219211e-02 2.64049292e-01
-4.22983497e-01 1.17345178e+00 -6.18091188e-02 6.84521571e-02
1.02213347e+00 7.88831413e-01 1.76947676e-02 1.70843214e-01
-1.07658231e+00 8.70179906e-02 -2.14110330e-01 7.46227562e-01
2.68595755e-01 2.49665901e-01 1.32155374e-01 3.97192717e-01
-3.41862679e-01 3.60773891e-01 4.98257399e-01 1.15083230e+00
-3.59426171e-01 -1.33959198e+00 -3.56508106e-01 9.52672482e-01
-4.44728166e-01 -1.41221687e-01 2.39309907e-01 3.79253268e-01
-3.12954575e-01 1.25553191e+00 1.30588278e-01 1.26641378e-01
3.44695598e-01 6.49726152e-01 2.88058013e-01 -2.76576102e-01
-7.25592136e-01 1.15876086e-01 1.18513130e-01 -2.73117095e-01
-1.22853130e-01 -8.90233874e-01 -9.63732421e-01 -6.65183604e-01
-1.94937393e-01 2.25673079e-01 8.12605739e-01 1.27841163e+00
8.23163167e-02 4.09081608e-01 1.01872754e+00 -1.00413930e+00
-1.07723808e+00 -1.44593930e+00 -1.03670430e+00 9.25080758e-03
2.88422346e-01 4.83768545e-02 -4.07741606e-01 3.67798805e-01]
|
[6.162865161895752, 2.875365734100342]
|
068432ae-9f15-4c3f-8d2c-ac0d1cac6d8c
|
an-overview-of-structural-coverage-metrics
|
2208.03407
| null |
https://arxiv.org/abs/2208.03407v1
|
https://arxiv.org/pdf/2208.03407v1.pdf
|
An Overview of Structural Coverage Metrics for Testing Neural Networks
|
Deep neural network (DNN) models, including those used in safety-critical domains, need to be thoroughly tested to ensure that they can reliably perform well in different scenarios. In this article, we provide an overview of structural coverage metrics for testing DNN models, including neuron coverage (NC), k-multisection neuron coverage (kMNC), top-k neuron coverage (TKNC), neuron boundary coverage (NBC), strong neuron activation coverage (SNAC) and modified condition/decision coverage (MC/DC). We evaluate the metrics on realistic DNN models used for perception tasks (including LeNet-1, LeNet-4, LeNet-5, and ResNet20) as well as on networks used in autonomy (TaxiNet). We also provide a tool, DNNCov, which can measure the testing coverage for all these metrics. DNNCov outputs an informative coverage report to enable researchers and practitioners to assess the adequacy of DNN testing, compare different coverage measures, and to more conveniently inspect the model's internals during testing.
|
['Corina S. Pasareanu', 'Luca Manolache', 'Rishi Dange', 'Divya Gopinath', 'Youcheng Sun', 'Muhammad Usman']
|
2022-08-05
| null | null | null | null |
['dnn-testing']
|
['adversarial']
|
[ 2.02168137e-01 1.69618800e-01 -5.84877608e-03 -4.00620133e-01
-2.10778967e-01 -6.65446460e-01 3.23300391e-01 -2.49715261e-02
-3.87248904e-01 9.48731303e-01 -1.98330924e-01 -7.64907479e-01
-3.06863755e-01 -7.86131740e-01 -7.59563982e-01 -5.12467384e-01
-1.31737351e-01 5.10682344e-01 6.43452883e-01 -1.28867835e-01
-3.74617688e-02 7.26515055e-01 -1.95545948e+00 3.41968447e-01
7.01107085e-01 1.18729293e+00 7.25358054e-02 4.64916319e-01
5.30010104e-01 8.21387291e-01 -8.89014244e-01 -2.74512738e-01
2.01620892e-01 -1.74731940e-01 -6.29877746e-01 -3.89157742e-01
3.20411742e-01 -3.39727253e-01 -1.15238234e-01 1.26160479e+00
6.17827833e-01 5.95905893e-02 4.79537964e-01 -1.65825570e+00
2.45173037e-01 9.16153371e-01 2.30804488e-01 5.71320772e-01
-2.31826514e-01 4.95172650e-01 5.35331488e-01 -3.38108718e-01
6.31927490e-01 1.22611904e+00 7.30364799e-01 9.67704415e-01
-1.05292428e+00 -7.62270391e-01 2.53988393e-02 1.24590665e-01
-1.05855143e+00 -4.59741682e-01 6.99978918e-02 -4.15787369e-01
1.49500215e+00 2.61683702e-01 1.11174917e+00 1.62049234e+00
4.59566474e-01 6.31217659e-01 8.28469515e-01 -1.85599536e-01
8.00890565e-01 -3.93862687e-02 3.00660819e-01 4.58306223e-01
9.06639874e-01 4.08892661e-01 -1.03019081e-01 1.56860173e-01
9.10262287e-01 -2.84093022e-01 -1.59351423e-01 8.80943835e-02
-8.77291083e-01 5.70762932e-01 -1.60183907e-02 2.01936543e-01
-3.49229932e-01 4.42209661e-01 7.08500624e-01 4.22820985e-01
-7.14516267e-02 3.49837571e-01 -6.00325704e-01 -3.89429867e-01
-7.02407956e-01 4.61589962e-01 8.16277385e-01 1.26310813e+00
5.22987068e-01 6.06491506e-01 -3.32123637e-01 7.37374723e-01
2.18017027e-01 2.48811230e-01 4.65255588e-01 -1.21691537e+00
2.77983636e-01 6.16294920e-01 -3.46201360e-01 -1.64644912e-01
-6.05764627e-01 -7.44045317e-01 -9.51080203e-01 5.52412391e-01
3.03652678e-02 -4.75548983e-01 -1.13143706e+00 1.76091504e+00
-1.59559473e-01 1.04328088e-01 3.76374155e-01 4.54960406e-01
1.07974577e+00 2.73878455e-01 -3.32728066e-02 1.73465684e-01
9.56092536e-01 -7.80121267e-01 -3.72998565e-01 -4.85367030e-01
1.00889957e+00 9.91125181e-02 7.99732268e-01 5.68874836e-01
-1.24124300e+00 -3.99241984e-01 -1.39518571e+00 5.50424576e-01
-3.90880972e-01 -2.91223675e-01 2.49288827e-01 8.30038786e-01
-1.48154187e+00 5.39694726e-01 -9.58471835e-01 -3.88696760e-01
6.75390720e-01 6.87350869e-01 -3.63266349e-01 -2.56004870e-01
-1.24898791e+00 1.13738751e+00 7.23395705e-01 -1.52373582e-01
-1.97878861e+00 -5.89626610e-01 -8.21245074e-01 2.15012625e-01
5.32847792e-02 -4.42639470e-01 1.35009909e+00 -6.71811700e-01
-9.27040279e-01 5.13692975e-01 5.02909541e-01 -7.87999630e-01
2.54847050e-01 4.50502992e-01 -6.83219016e-01 -2.24543475e-02
-3.96133065e-01 1.23159456e+00 -2.33393997e-01 -1.20537722e+00
-8.15908372e-01 -6.69473633e-02 4.37261850e-01 6.13137428e-03
-2.90552825e-02 -8.64494219e-02 -2.52106935e-01 -4.64217544e-01
-1.48346841e-01 -7.13188589e-01 -1.62598521e-01 -1.22656651e-01
-6.65456772e-01 -1.58809766e-01 7.27541208e-01 -4.02163506e-01
9.95692492e-01 -2.13854122e+00 -2.31318876e-01 5.27876437e-01
5.81140071e-02 4.73053247e-01 -5.25022447e-01 1.14796266e-01
-9.67185870e-02 1.83736786e-01 -2.56392062e-01 -1.36156706e-02
-9.47308913e-02 6.61923647e-01 2.94691801e-01 1.05861984e-01
1.79854095e-01 8.55360091e-01 -4.58671361e-01 -1.75112545e-01
8.31769407e-02 4.48654026e-01 -5.89782655e-01 -4.35748607e-01
-4.15019393e-01 8.63580871e-03 -8.42975751e-02 1.01477420e+00
4.40951139e-01 2.10990638e-01 -1.44298524e-01 1.28099278e-01
1.57464996e-01 9.35953036e-02 -8.46488833e-01 1.00942183e+00
-7.25846067e-02 9.56333756e-01 -2.47587506e-02 -8.16074431e-01
9.67606544e-01 5.24913669e-01 6.27582669e-02 -9.62987185e-01
3.67475450e-01 2.20459029e-01 5.28526664e-01 -3.63731414e-01
1.12253269e-02 5.92003316e-02 1.25413910e-01 2.04117388e-01
5.05095720e-01 3.46856296e-01 4.17263836e-01 -3.02202731e-01
1.54304326e+00 -6.96145535e-01 -1.27587348e-01 -6.62591875e-01
2.80805051e-01 -6.53788820e-02 6.70181811e-01 7.19769061e-01
-4.81757611e-01 3.46261322e-01 1.19924212e+00 -1.76849067e-01
-1.23796546e+00 -8.85308504e-01 -3.24721128e-01 5.50466597e-01
-1.04083337e-01 1.62647247e-01 -1.41987085e+00 -5.56927323e-01
-1.84511647e-01 1.25533557e+00 -6.32592559e-01 -4.97456670e-01
-2.56381571e-01 -5.34462929e-01 1.34693754e+00 1.04552841e+00
6.97884619e-01 -1.57570601e+00 -9.80873168e-01 3.93318906e-02
8.87896717e-02 -9.96514678e-01 2.79267430e-01 7.36809492e-01
-1.18118644e+00 -1.14738333e+00 -3.24412316e-01 -9.62214410e-01
5.90941846e-01 -2.77471930e-01 1.04761064e+00 2.91693956e-01
-1.44119442e-01 1.67297050e-01 -1.75222963e-01 -6.45083547e-01
-6.52280152e-01 -8.89010727e-02 2.90441006e-01 -8.47635627e-01
2.92348146e-01 -5.62849104e-01 -1.42801136e-01 7.34077811e-01
-1.24814773e+00 -1.74188599e-01 5.89016497e-01 6.54755235e-01
6.36535287e-01 3.43742996e-01 6.44392014e-01 -6.61156476e-01
8.85349870e-01 -2.48398796e-01 -8.68472755e-01 4.19898689e-01
-9.48037148e-01 -1.88753024e-01 2.43931755e-01 -5.65804183e-01
-3.50279808e-01 -5.15876412e-01 -5.69748223e-01 -5.37210941e-01
-4.83665824e-01 7.64317393e-01 -4.07595783e-01 -9.65134874e-02
8.80565226e-01 3.55829857e-02 2.11966969e-02 -1.72991395e-01
-5.23839533e-01 2.47203276e-01 5.28202116e-01 -2.74908423e-01
1.98015347e-02 7.80031364e-03 -4.24674787e-02 -4.44966882e-01
-3.79977748e-02 1.11989789e-01 -2.42480338e-01 -4.67286110e-01
6.15548968e-01 -7.57342100e-01 -7.76664376e-01 6.50301456e-01
-1.05529630e+00 -1.00389719e+00 -4.60494310e-01 5.36219418e-01
-4.03059751e-01 -3.61841828e-01 -6.90388918e-01 -6.72827065e-01
-3.29788744e-01 -1.44315755e+00 3.84819508e-01 1.92688063e-01
-1.55118853e-01 -1.11532223e+00 9.87470299e-02 -1.30010217e-01
3.92485619e-01 4.66047943e-01 1.30436778e+00 -1.05923700e+00
-2.14429051e-01 -3.28093439e-01 -1.03921011e-01 9.31977808e-01
-5.31470895e-01 1.09136991e-01 -1.08281791e+00 -2.68919170e-01
-1.43614635e-01 -4.45680022e-01 8.08987379e-01 8.72294247e-01
1.18683136e+00 -4.60276425e-01 -3.62524658e-01 4.77440596e-01
1.43665266e+00 5.34659326e-01 1.22442806e+00 5.32732308e-01
2.83432394e-01 4.21935588e-01 1.79608703e-01 1.78603023e-01
-6.53095692e-02 1.58933744e-01 1.28908157e+00 6.95235729e-02
8.75736549e-02 2.47784317e-01 5.77580512e-01 4.70669240e-01
-1.14953287e-01 -1.00775707e+00 -1.14523458e+00 4.51650918e-01
-1.59917009e+00 -8.30330253e-01 6.24627508e-02 1.88440084e+00
4.62472260e-01 7.62493670e-01 1.25568556e-02 6.45360589e-01
7.02322304e-01 -3.81371498e-01 -7.96135068e-01 -7.38210022e-01
-4.50802118e-01 1.29191533e-01 5.09530127e-01 3.21038455e-01
-6.11589372e-01 6.11947715e-01 7.50671148e+00 8.82227778e-01
-7.90467978e-01 1.99832842e-01 7.89109170e-01 -2.69913107e-01
-3.40739399e-01 -3.08022380e-01 -1.00356984e+00 3.00698489e-01
1.36131215e+00 1.98686257e-01 4.73205298e-01 1.06173956e+00
1.90783441e-02 -1.90652937e-01 -1.32930052e+00 6.58765197e-01
-1.98582098e-01 -1.43447816e+00 -2.41642389e-02 2.07727432e-01
7.60196149e-01 5.75300336e-01 -2.39729751e-02 5.41611552e-01
5.43588519e-01 -1.27466118e+00 9.91158545e-01 4.03328180e-01
9.83947635e-01 -9.84440446e-01 1.37265611e+00 3.67975563e-01
-5.03520608e-01 -2.23487765e-01 -6.02216005e-01 1.62887782e-01
-1.23214796e-01 5.56345880e-01 -7.47432113e-01 1.26969172e-02
9.33426917e-01 4.15225327e-01 -5.43117821e-01 1.28805292e+00
-1.15299439e-02 7.77535915e-01 -2.93374896e-01 -3.11175436e-01
3.04913700e-01 4.28292304e-01 4.05417293e-01 1.08945239e+00
5.09499490e-01 -1.72691986e-01 -6.00465238e-01 9.90102112e-01
-8.61522257e-02 -7.07886517e-01 -4.73353803e-01 1.54312491e-01
6.48695946e-01 1.11516500e+00 -8.04004312e-01 -1.35441840e-01
-3.03370357e-02 3.03255498e-01 -1.78003892e-01 4.48754340e-01
-9.69020784e-01 -2.19120890e-01 8.84307981e-01 1.59262925e-01
1.51903912e-01 1.32337883e-01 -3.43944967e-01 -2.88716793e-01
-3.83985974e-02 -1.04831493e+00 1.26323596e-01 -8.88003230e-01
-7.54265308e-01 1.06705165e+00 4.12482172e-01 -1.18903601e+00
-7.62663484e-02 -1.08630335e+00 -6.33082509e-01 5.32183766e-01
-9.74572897e-01 -4.39419180e-01 -4.91110653e-01 5.19064188e-01
3.56032491e-01 -5.79011023e-01 7.93521047e-01 5.94781578e-01
-9.38107610e-01 8.29014122e-01 -1.45385548e-01 1.19821414e-01
-1.58898488e-01 -7.46517360e-01 4.81998682e-01 7.08691180e-01
-4.43100333e-01 2.70018727e-01 7.61861622e-01 -6.67578042e-01
-7.27309883e-01 -1.54186261e+00 3.89688313e-01 -1.75236061e-01
8.04105029e-02 -3.41294765e-01 -5.75468600e-01 8.14137101e-01
7.57618472e-02 -9.04000327e-02 3.91701251e-01 -1.89600170e-01
-3.87231559e-02 -3.32858503e-01 -1.49403000e+00 6.90673411e-01
1.18634570e+00 -2.26163566e-02 5.10782413e-02 2.07761213e-01
8.18534374e-01 -4.87587720e-01 -7.53311276e-01 7.60307312e-01
5.07220209e-01 -1.24136508e+00 4.70227033e-01 -4.57568377e-01
1.79200321e-01 -7.56618157e-02 -4.86619562e-01 -1.16264534e+00
-2.71831423e-01 -6.36748970e-02 -8.04888234e-02 8.56620312e-01
9.63842690e-01 -9.56370056e-01 7.90484190e-01 5.57120919e-01
-7.75466561e-01 -1.15998161e+00 -1.28908455e+00 -1.16659033e+00
-1.39964342e-01 -7.96980739e-01 8.81780326e-01 6.57895207e-01
-2.68283933e-01 -7.09987134e-02 2.83761740e-01 8.71720389e-02
1.58705086e-01 -1.24888217e+00 1.21340379e-01 -1.53223753e+00
1.06818944e-01 -7.72655249e-01 -9.54716146e-01 -4.07716691e-01
-6.80218786e-02 -6.90496087e-01 2.59220544e-02 -1.85766435e+00
1.36565238e-01 -4.21563834e-01 -4.47513282e-01 9.55445051e-01
7.76842415e-01 1.52450532e-01 -4.07532901e-01 -8.34859312e-02
-4.11805987e-01 1.24411814e-01 1.03322637e+00 -2.35897273e-01
1.37487143e-01 6.71551824e-02 -6.05511129e-01 7.33398020e-01
9.53486562e-01 -5.25442243e-01 -8.19690883e-01 -4.26576078e-01
3.99691135e-01 -1.00132890e-01 7.48654485e-01 -1.78881180e+00
3.20825905e-01 8.88213217e-02 3.56307030e-01 -8.51129055e-01
2.02563837e-01 -7.75163651e-01 3.79202664e-01 8.64457488e-01
-4.54377115e-01 6.05370760e-01 7.58668780e-01 1.62711412e-01
-1.72825038e-01 -7.12346256e-01 7.11956382e-01 6.76484331e-02
-8.23132873e-01 2.75882989e-01 -8.47428441e-01 -5.64906523e-02
1.17839324e+00 -7.61703491e-01 -8.10968101e-01 -1.94603488e-01
-6.63468122e-01 6.26258969e-01 3.80360991e-01 1.93045422e-01
9.95221794e-01 -1.36157584e+00 -5.03459871e-01 4.76225406e-01
1.86209023e-01 1.74951240e-01 3.75019073e-01 9.62615192e-01
-1.09689140e+00 5.60703397e-01 -6.17240727e-01 -6.51132822e-01
-1.04806399e+00 2.26341516e-01 8.58675897e-01 -1.75669357e-01
-4.89183098e-01 1.14920962e+00 4.81088758e-02 -5.59944510e-01
8.43778014e-01 -9.15995419e-01 -3.61445010e-01 -2.97698140e-01
2.77504116e-01 6.81836903e-01 5.81594050e-01 -3.85378785e-02
-4.68853980e-01 -1.26513943e-01 1.40464097e-01 -5.31250536e-01
1.07497573e+00 4.99993801e-01 8.54345411e-02 2.59040534e-01
7.38112152e-01 -9.95690584e-01 -1.39541793e+00 4.88398850e-01
-1.59882560e-01 5.55976748e-01 1.85682029e-01 -1.18214786e+00
-1.26627684e+00 8.05852413e-01 7.83885658e-01 3.01447392e-01
1.25242937e+00 -7.55352527e-02 4.78183836e-01 5.30863285e-01
4.09616977e-01 -1.21590984e+00 -1.76190346e-01 8.23562205e-01
8.25457871e-01 -4.73386794e-01 -4.03333008e-01 1.19708531e-01
-6.97249949e-01 9.06838238e-01 9.71261382e-01 5.63550331e-02
8.14282775e-01 7.58152306e-01 -1.59704000e-01 -2.85548270e-01
-1.32442355e+00 1.47466615e-01 -7.40339160e-02 8.49044800e-01
-1.35342494e-01 -2.52103228e-02 3.58736753e-01 7.27955818e-01
-2.16092020e-01 -1.01189815e-01 4.42204118e-01 8.48297238e-01
-7.62501180e-01 -5.45589268e-01 1.25073746e-01 8.95322502e-01
-1.73553362e-01 -1.87199920e-01 -3.80394608e-01 1.09064734e+00
4.41383898e-01 9.29420650e-01 3.49341303e-01 -1.08329117e+00
5.89746356e-01 -1.05671063e-01 5.63269198e-01 -4.17032152e-01
-7.09946215e-01 -2.93464269e-02 5.78207076e-01 -5.34337521e-01
-5.35039119e-02 -6.08241260e-01 -1.25747824e+00 -6.45767748e-01
-4.86091524e-01 -1.90483540e-01 7.47580469e-01 9.18703377e-01
5.54021299e-02 9.94085431e-01 -6.35275841e-02 -4.74946380e-01
-3.84903908e-01 -1.22507536e+00 -7.85394728e-01 -4.28846419e-01
1.80131346e-01 -6.16498232e-01 -4.39067930e-01 -4.10689831e-01]
|
[6.585072040557861, 7.646378040313721]
|
d27cc83b-0886-4f2c-9cb1-2d0210802074
|
consistent-and-complementary-graph
|
2004.03106
| null |
https://arxiv.org/abs/2004.03106v1
|
https://arxiv.org/pdf/2004.03106v1.pdf
|
Consistent and Complementary Graph Regularized Multi-view Subspace Clustering
|
This study investigates the problem of multi-view clustering, where multiple views contain consistent information and each view also includes complementary information. Exploration of all information is crucial for good multi-view clustering. However, most traditional methods blindly or crudely combine multiple views for clustering and are unable to fully exploit the valuable information. Therefore, we propose a method that involves consistent and complementary graph-regularized multi-view subspace clustering (GRMSC), which simultaneously integrates a consistent graph regularizer with a complementary graph regularizer into the objective function. In particular, the consistent graph regularizer learns the intrinsic affinity relationship of data points shared by all views. The complementary graph regularizer investigates the specific information of multiple views. It is noteworthy that the consistent and complementary regularizers are formulated by two different graphs constructed from the first-order proximity and second-order proximity of multiple views, respectively. The objective function is optimized by the augmented Lagrangian multiplier method in order to achieve multi-view clustering. Extensive experiments on six benchmark datasets serve to validate the effectiveness of the proposed method over other state-of-the-art multi-view clustering methods.
|
['Jun Wang', 'Shanmin Pang', 'Qinghai Zheng', 'Lei Chen', 'Jihua Zhu', 'Zhongyu Li']
|
2020-04-07
| null | null | null | null |
['multi-view-subspace-clustering']
|
['computer-vision']
|
[-3.55298460e-01 -2.44796515e-01 -1.19553231e-01 -2.23489434e-01
-6.59298122e-01 -6.98594153e-01 3.19751263e-01 2.43667420e-03
4.37544612e-03 1.86037630e-01 2.95910597e-01 3.11839759e-01
-4.62986887e-01 -4.48963583e-01 -4.27677870e-01 -1.13560915e+00
9.51402858e-02 4.16409254e-01 7.77347609e-02 6.68387413e-02
9.01963189e-02 1.73331872e-02 -1.21189260e+00 3.59048605e-01
8.69518638e-01 4.24890727e-01 3.68451893e-01 2.53560930e-01
2.03698292e-01 4.68856186e-01 4.39243764e-02 -1.42905682e-01
3.82150322e-01 -4.94530171e-01 -5.08455813e-01 6.51627541e-01
3.49873424e-01 2.57966071e-01 -1.28456309e-01 1.40168393e+00
5.44691920e-01 2.63730347e-01 3.54162216e-01 -1.36098409e+00
-6.80992246e-01 4.40048695e-01 -1.23062563e+00 1.10852063e-01
3.37709069e-01 -1.08861595e-01 1.15684843e+00 -1.01906049e+00
6.97298586e-01 1.31230032e+00 2.37709552e-01 1.60536706e-01
-1.39084268e+00 -3.50109905e-01 6.96876407e-01 2.24726304e-01
-1.59987104e+00 -2.32912272e-01 1.23551285e+00 -5.03700554e-01
2.67867625e-01 2.12652236e-01 7.11415589e-01 4.90099251e-01
-7.17848614e-02 7.93379724e-01 1.30781829e+00 -5.53130358e-02
9.26917344e-02 3.01877201e-01 2.52405971e-01 7.94472694e-01
3.91175121e-01 -2.54160672e-01 -4.55605447e-01 -3.96684229e-01
3.61764908e-01 4.47340161e-01 -6.64716959e-01 -1.23111272e+00
-1.37657750e+00 8.28798831e-01 4.50253665e-01 4.00157571e-01
-3.88816595e-01 -4.40501153e-01 2.08510280e-01 -3.50289755e-02
3.43559444e-01 2.30951644e-02 -1.95200052e-02 4.83612210e-01
-7.16190577e-01 -2.17750043e-01 5.95927954e-01 1.09986162e+00
9.82636631e-01 -1.10520706e-01 1.91705644e-01 8.21167290e-01
6.74650848e-01 4.53478903e-01 2.62628168e-01 -8.30686569e-01
6.37582362e-01 1.14432478e+00 -2.49487177e-01 -1.39388847e+00
-2.63009131e-01 -5.76184273e-01 -1.21248305e+00 -1.24638744e-01
1.60428494e-01 -9.60852951e-02 -4.87207502e-01 1.75409818e+00
7.25107253e-01 3.78021508e-01 2.25820228e-01 1.11802018e+00
9.31718886e-01 4.57990080e-01 -5.29020429e-01 -6.05573237e-01
1.26639318e+00 -1.00730085e+00 -6.25477672e-01 1.64959840e-02
2.81796038e-01 -7.45033503e-01 7.11454868e-01 3.51173162e-01
-1.01523197e+00 -4.92134273e-01 -1.12043929e+00 3.32336575e-01
-1.39430225e-01 1.23032995e-01 3.12829643e-01 3.35941136e-01
-9.51220393e-01 1.68294355e-01 -6.54424012e-01 -2.73055434e-01
5.85617721e-02 3.68605345e-01 -6.82892799e-01 -4.41992074e-01
-7.26328373e-01 2.17427507e-01 4.75630045e-01 1.53838158e-01
-5.67708015e-01 -3.93757313e-01 -7.91354001e-01 -1.52899325e-01
7.50988781e-01 -7.10816860e-01 2.29748979e-01 -7.09413946e-01
-9.21198964e-01 8.68766725e-01 -3.50833744e-01 2.76620686e-01
1.70760170e-01 4.74828556e-02 -4.05027300e-01 4.99923885e-01
3.79065514e-01 1.66884199e-01 8.13809633e-01 -2.11171031e+00
-4.41822857e-01 -8.31091940e-01 -1.17404193e-01 7.31094003e-01
-2.81062007e-01 -3.71384263e-01 -1.16103935e+00 -4.28370863e-01
6.89804256e-01 -1.07786596e+00 -2.37783507e-01 -3.99066329e-01
-6.18821442e-01 1.16586238e-01 1.08906257e+00 -6.37602210e-01
1.38384259e+00 -2.22407722e+00 9.12741661e-01 5.14504969e-01
6.18121743e-01 -1.09428555e-01 2.98453253e-02 5.62902033e-01
-2.55569160e-01 -8.59126970e-02 -2.94321686e-01 -3.19520652e-01
-3.17504555e-01 2.48117223e-01 2.90773302e-01 8.70485425e-01
-4.37950969e-01 5.66027045e-01 -1.06082225e+00 -6.59141660e-01
1.93790153e-01 4.68975782e-01 -4.66111034e-01 2.64777064e-01
3.20831299e-01 8.02443087e-01 -4.88344043e-01 5.87571084e-01
9.49971139e-01 -7.21722722e-01 5.48129022e-01 -6.64264143e-01
6.51131198e-02 -5.96327484e-01 -1.72390890e+00 1.85218346e+00
7.75143206e-02 -1.99822932e-02 5.06146073e-01 -9.71362114e-01
5.93446910e-01 3.77191812e-01 9.45464432e-01 -1.60920843e-01
-2.06693932e-02 -4.05054539e-03 -1.21561751e-01 -4.43018466e-01
1.82106733e-01 -1.44124508e-01 2.06282884e-01 6.00136936e-01
-7.16651306e-02 2.72906512e-01 2.35423326e-01 7.01529860e-01
5.62131643e-01 -1.66114822e-01 4.71050352e-01 -3.19254905e-01
1.08233452e+00 -2.98121542e-01 8.41498494e-01 1.88072369e-01
-1.56095535e-01 7.91027308e-01 4.30067062e-01 -2.02993289e-01
-7.10692406e-01 -1.01549709e+00 2.35688865e-01 6.30921006e-01
7.24923313e-01 -6.95711374e-01 -7.12984860e-01 -9.84958887e-01
-1.16352722e-01 2.20371366e-01 -5.35614014e-01 -1.00380093e-01
-3.35066527e-01 -8.52612376e-01 -1.47133276e-01 2.58058965e-01
5.47483563e-01 -3.06386143e-01 1.05166351e-02 -1.05281547e-01
-4.72296119e-01 -1.13981664e+00 -8.66520703e-01 -1.52991759e-02
-8.65961015e-01 -1.44073284e+00 -6.85876429e-01 -8.77577722e-01
1.16226315e+00 1.11912167e+00 8.87774765e-01 2.53955871e-01
1.77577555e-01 8.49018931e-01 -4.44856405e-01 2.18722850e-01
-8.52034800e-03 -2.14703023e-01 2.35095054e-01 6.85164928e-01
2.26016417e-01 -8.43357325e-01 -5.60680151e-01 5.60049415e-01
-8.56788695e-01 1.08428754e-01 4.59619403e-01 9.07475829e-01
1.13236082e+00 3.89798164e-01 1.20118581e-01 -1.14556181e+00
3.63137424e-01 -5.84994256e-01 -3.54071826e-01 5.79061389e-01
-6.61536813e-01 1.14073604e-02 5.83273470e-01 -1.02111779e-01
-9.35078919e-01 2.98254132e-01 5.44456780e-01 -9.85141277e-01
5.00444323e-02 6.26952767e-01 -6.49110079e-01 -1.38419017e-01
8.44614282e-02 4.51499850e-01 4.14526500e-02 -3.37081194e-01
6.19439363e-01 2.59280443e-01 3.32690060e-01 -4.71262395e-01
9.93903339e-01 9.94326353e-01 1.44961655e-01 -7.17354417e-01
-7.27151155e-01 -1.03213561e+00 -1.11953831e+00 -4.20656860e-01
9.65398788e-01 -1.19857824e+00 -6.67301476e-01 1.90120056e-01
-6.25489652e-01 4.83726412e-01 1.12892739e-01 5.92474043e-01
-4.15809989e-01 9.57435727e-01 -2.43612543e-01 -6.13873243e-01
2.01337179e-03 -1.17396986e+00 1.11496699e+00 7.27572218e-02
3.34786087e-01 -1.28388071e+00 1.25808418e-01 6.56304657e-01
-3.63129616e-01 4.39490467e-01 7.44593978e-01 -5.27972639e-01
-5.23970366e-01 -2.37504710e-02 -2.24676490e-01 2.05659568e-01
4.20170039e-01 -9.64511260e-02 -6.87415421e-01 -7.93485761e-01
3.91215920e-01 1.00196332e-01 7.57031798e-01 4.08186048e-01
6.53887987e-01 -1.77525997e-01 -4.79417384e-01 7.32627571e-01
1.80655050e+00 1.15431227e-01 1.52799621e-01 9.44804549e-02
1.40450609e+00 6.26468718e-01 5.54448843e-01 5.03972828e-01
7.94144392e-01 5.44095039e-01 4.51930434e-01 -1.53732032e-01
2.60625005e-01 -1.16138630e-01 1.48367822e-01 1.48471236e+00
-3.12848270e-01 -9.40010175e-02 -7.99854815e-01 4.28812385e-01
-2.22827029e+00 -1.06309676e+00 -4.67355579e-01 2.27267170e+00
1.18587956e-01 -2.20560983e-01 3.05191636e-01 7.54738376e-02
1.22127700e+00 4.24722254e-01 -5.05623341e-01 3.16516548e-01
-2.90231854e-01 -7.66131222e-01 3.19990456e-01 4.29803252e-01
-1.15991354e+00 5.36355913e-01 5.11303186e+00 5.69541693e-01
-5.70288002e-01 1.28992021e-01 3.88659239e-01 -2.16893777e-02
-5.62827945e-01 3.02351236e-01 -5.01666665e-01 3.66892904e-01
1.97484523e-01 -1.07935786e-01 7.10360885e-01 6.89417362e-01
1.91614226e-01 -2.16147229e-01 -1.00636637e+00 1.29132402e+00
4.19036448e-01 -9.24966633e-01 2.00253800e-01 3.55071008e-01
1.06449425e+00 -2.44091392e-01 7.10032433e-02 -2.35068470e-01
3.29469979e-01 -3.61189544e-01 4.19598132e-01 6.26779079e-01
4.02389288e-01 -9.39346135e-01 5.45342445e-01 4.53104913e-01
-1.84463775e+00 -2.05851961e-02 -2.28430510e-01 3.55764985e-01
2.49334186e-01 6.56792641e-01 -1.45533621e-01 1.37226176e+00
6.80991828e-01 1.38099480e+00 -7.52033770e-01 6.18035614e-01
-5.97789511e-02 1.62946016e-01 -1.24021351e-01 5.67139089e-01
3.19272429e-01 -9.72523451e-01 8.53443563e-01 6.45478249e-01
7.09314272e-02 3.26046169e-01 6.24269307e-01 6.30566120e-01
2.54682869e-01 2.01774627e-01 -7.57527471e-01 1.42884135e-01
3.72799158e-01 1.45025849e+00 -9.02768314e-01 -2.72123486e-01
-7.90078580e-01 1.05212915e+00 4.58892494e-01 5.42355776e-01
-5.58406293e-01 1.96542278e-01 3.52695674e-01 -1.46221951e-01
3.95575643e-01 -3.32500517e-01 -2.62620687e-01 -1.63475573e+00
1.52646855e-01 -9.53370690e-01 8.23249698e-01 -5.75496972e-01
-1.47474694e+00 3.83466274e-01 2.14950126e-02 -1.62927413e+00
1.95274100e-01 -8.34160000e-02 -6.23710990e-01 7.15111792e-01
-1.35697150e+00 -1.37064576e+00 -3.92082036e-01 1.18044591e+00
2.97838718e-01 -3.11515629e-01 4.40644205e-01 3.26014459e-01
-9.26502705e-01 2.33049706e-01 3.90720069e-01 -3.27617824e-02
6.71702504e-01 -1.42028058e+00 -3.18807393e-01 1.02097332e+00
2.91585207e-01 8.06054413e-01 2.97506452e-01 -7.03874767e-01
-1.83822322e+00 -1.00437069e+00 2.41506219e-01 -3.67697299e-01
4.54193801e-01 -2.30184361e-01 -9.61743116e-01 6.31268740e-01
3.04877937e-01 1.96681321e-01 1.06618059e+00 9.26602557e-02
-3.09022039e-01 -1.54298186e-01 -9.29215133e-01 4.09555584e-01
8.45877886e-01 -6.05902612e-01 -3.21469098e-01 3.46918523e-01
5.15111804e-01 -1.72952965e-01 -1.06299031e+00 3.80356938e-01
2.75453925e-01 -1.29050040e+00 1.07891440e+00 -3.59672993e-01
1.47290975e-01 -7.46541977e-01 -4.79644030e-01 -1.38583410e+00
-6.71918273e-01 -3.41130584e-01 -2.61020690e-01 1.50409663e+00
5.84290028e-02 -5.31041026e-01 5.86512089e-01 3.26147348e-01
3.53378966e-03 -8.25313687e-01 -7.53839374e-01 -7.39791393e-01
-3.33448172e-01 1.07877403e-01 3.76817405e-01 1.41444945e+00
9.95553061e-02 5.30838907e-01 -6.11330807e-01 6.51841283e-01
1.19647384e+00 6.40957713e-01 7.57506788e-01 -1.25841236e+00
-4.15916204e-01 -1.44900590e-01 -3.72706234e-01 -7.11207449e-01
6.08694814e-02 -1.02560282e+00 -2.19676986e-01 -1.65375912e+00
8.85834813e-01 -1.74667329e-01 -4.02682751e-01 1.14113353e-02
-6.93324327e-01 -1.27318993e-01 2.88151950e-01 5.79675972e-01
-9.37673450e-01 6.17058456e-01 1.44815505e+00 -8.87134671e-02
-3.26725900e-01 -1.27320364e-01 -9.97248113e-01 8.34483683e-01
4.31269974e-01 -2.91071743e-01 -8.27170074e-01 -3.07611614e-01
1.37470528e-01 1.89257473e-01 2.82080889e-01 -8.92692387e-01
6.40351892e-01 -2.16688082e-01 2.25172013e-01 -7.50830591e-01
3.32568705e-01 -1.15984499e+00 4.43845630e-01 1.21361785e-01
4.45802882e-02 2.15067461e-01 -3.16184014e-01 1.25579906e+00
-5.56608200e-01 2.42192760e-01 8.23767185e-01 -4.96544063e-01
-7.41028905e-01 4.81422514e-01 1.34690493e-01 2.53029525e-01
1.23264503e+00 -3.29554081e-01 -6.91406429e-02 -4.22274947e-01
-9.75494802e-01 6.57859623e-01 7.61767566e-01 2.92679995e-01
7.31230497e-01 -1.69883919e+00 -5.52097797e-01 1.59064695e-01
2.71337777e-01 8.51548091e-02 5.45707524e-01 1.11249340e+00
2.41625160e-02 1.36769965e-01 -9.29800197e-02 -9.75970626e-01
-1.65072060e+00 1.09416556e+00 5.49658723e-02 -3.62942487e-01
-5.43157995e-01 6.92174971e-01 5.74101269e-01 -4.63175833e-01
-2.04526372e-02 3.52428883e-01 -5.38701952e-01 3.07949036e-01
1.73997208e-01 6.15811229e-01 -3.67686749e-01 -1.17425752e+00
-5.24029016e-01 1.17226410e+00 2.81537361e-02 -1.84847098e-02
1.36734533e+00 -7.41895914e-01 -5.08046031e-01 7.10268676e-01
1.34531987e+00 7.66181052e-02 -1.06433749e+00 -4.89388943e-01
-2.80728221e-01 -6.37361228e-01 1.19239197e-03 -1.49729282e-01
-1.55971646e+00 7.19183147e-01 2.59144902e-01 2.50506997e-01
1.37158656e+00 -2.06232034e-02 4.53907907e-01 1.13170020e-01
3.97231370e-01 -1.00078797e+00 2.51658618e-01 9.51810181e-02
6.74940646e-01 -1.46509027e+00 4.59979892e-01 -8.80061209e-01
-9.48128045e-01 8.24282408e-01 7.27784753e-01 -6.27117008e-02
8.86509180e-01 -2.57269591e-01 -1.17687315e-01 -7.63761222e-01
-5.76077700e-01 -2.14317992e-01 5.58083475e-01 3.56912643e-01
1.32119626e-01 5.95898740e-02 -1.76622838e-01 4.08445776e-01
4.08949256e-01 -6.47911847e-01 2.86159307e-01 8.76373589e-01
-1.66835636e-01 -1.05629873e+00 -6.27241850e-01 2.60076374e-01
-1.18792661e-01 9.66258720e-02 -6.68453217e-01 5.72698772e-01
1.14216916e-01 1.20010042e+00 -3.60171705e-01 -6.21830046e-01
1.86929375e-01 -2.08018064e-01 1.01731241e-01 -6.34244561e-01
-4.12842691e-01 6.46220088e-01 -3.14134121e-01 -6.01149261e-01
-1.00114191e+00 -9.14600134e-01 -1.12546599e+00 -1.48796663e-01
-4.77324069e-01 4.24147040e-01 2.30730817e-01 7.39669085e-01
4.61040288e-01 3.74557972e-01 1.06728363e+00 -7.91134715e-01
-7.34871402e-02 -3.74740273e-01 -1.07978487e+00 5.87836564e-01
1.81022480e-01 -6.94237173e-01 -4.94426221e-01 1.72550768e-01]
|
[8.238896369934082, 4.633297920227051]
|
db336120-3478-46bb-ad98-5536a2394ec5
|
constraint-based-inference-of-heuristics-for
|
2105.14194
| null |
https://arxiv.org/abs/2105.14194v1
|
https://arxiv.org/pdf/2105.14194v1.pdf
|
Constraint-Based Inference of Heuristics for Foreign Exchange Trade Model Optimization
|
The Foreign Exchange (Forex) is a large decentralized market, on which trading analysis and algorithmic trading are popular. Research efforts have been focusing on proof of efficiency of certain technical indicators. We demonstrate, however, that the values of indicator functions are not reproducible and often reduce the number of trade opportunities, compared to price-action trading. In this work, we develop two dataset-agnostic Forex trading heuristic templates with high rate of trading signals. In order to determine most optimal parameters for the given heuristic prototypes, we perform a machine learning simulation of 10 years of Forex price data over three low-margin instruments and 6 different OHLC granularities. As a result, we develop a specific and reproducible list of most optimal trade parameters found for each instrument-granularity pair, with 118 pips of average daily profit for the optimized configuration.
|
['Qiben Yan', 'Nikolay Ivanov']
|
2021-05-11
| null | null | null | null |
['algorithmic-trading']
|
['time-series']
|
[-6.66430056e-01 -2.05021307e-01 -4.71944124e-01 -1.50263071e-01
-9.47679400e-01 -1.19606781e+00 8.23432267e-01 9.83100533e-02
-1.36488885e-01 9.72675979e-01 -5.01630567e-02 -6.45026624e-01
-6.22326672e-01 -6.89547181e-01 -3.95121306e-01 -4.48941052e-01
-3.41087699e-01 9.68343675e-01 -1.12332679e-01 1.51616916e-01
6.51554942e-01 4.68736172e-01 -1.00214839e+00 -4.20064591e-02
7.14524508e-01 1.62564647e+00 -3.58753622e-01 1.14685483e-01
-6.45141974e-02 4.51220781e-01 -7.34376907e-01 -6.58765078e-01
1.27772462e+00 -4.67846602e-01 -2.57487655e-01 -1.76540330e-01
-3.32626730e-01 -2.42557436e-01 4.42475170e-01 1.04603398e+00
1.76846847e-01 -2.87372649e-01 6.41582966e-01 -1.26596403e+00
-2.72683412e-01 1.30820096e+00 -6.60042405e-01 3.72485429e-01
-3.44709635e-01 3.94956082e-01 1.67969537e+00 -3.51404160e-01
4.06238556e-01 7.93650925e-01 5.17643988e-01 -1.58087313e-01
-1.31433403e+00 -9.12765622e-01 -2.68112093e-01 -2.69117236e-01
-9.85243142e-01 9.60655604e-03 7.21286893e-01 -4.82700497e-01
8.75409365e-01 3.97342771e-01 1.11540020e+00 6.31751537e-01
5.74911058e-01 2.78322607e-01 1.74183857e+00 -2.63980657e-01
4.59823012e-01 4.43633705e-01 6.08818568e-02 -1.89986713e-02
8.44816685e-01 5.95767677e-01 -1.83608830e-01 -6.94429040e-01
9.00677681e-01 -1.25035912e-01 7.80295953e-02 -2.97523737e-01
-1.17711294e+00 1.12453973e+00 -2.55815666e-02 4.31146950e-01
-6.50817454e-01 2.08276778e-01 2.90965080e-01 1.07688594e+00
4.12678540e-01 8.77949178e-01 -1.00075221e+00 -4.38583463e-01
-8.07588100e-01 4.50268477e-01 1.16959870e+00 7.43144512e-01
5.38042784e-01 6.01038635e-02 -2.93505579e-01 3.23302448e-01
2.39617139e-01 6.74884021e-01 4.54778820e-01 -1.19094467e+00
5.88139951e-01 5.45332491e-01 5.96667409e-01 -8.15730810e-01
-3.66972893e-01 -8.32052886e-01 -3.64395082e-01 4.05967176e-01
8.08409214e-01 -5.25889933e-01 -2.00692028e-01 1.12485158e+00
-5.87899983e-02 -3.37256163e-01 1.48049649e-02 5.44354856e-01
-6.39196754e-01 5.06626785e-01 -4.05549973e-01 -7.00738728e-01
1.37134027e+00 -8.68684471e-01 -5.60793579e-01 3.42158437e-01
3.50305080e-01 -8.83472025e-01 7.10726023e-01 7.44369566e-01
-1.08111489e+00 -7.58021232e-03 -7.29278266e-01 7.66290367e-01
-3.96891147e-01 -4.94695008e-02 9.00822461e-01 5.83432972e-01
-7.70951688e-01 9.40777957e-01 -4.59374636e-01 2.61333019e-01
3.43873054e-02 3.75790685e-01 2.72832572e-01 1.14320588e+00
-1.15649235e+00 9.08862650e-01 7.16536164e-01 1.39699772e-01
-5.44752300e-01 -9.69971538e-01 9.85318124e-02 4.04692531e-01
4.61684912e-01 -3.62639338e-01 1.38874781e+00 -1.05886745e+00
-1.70146251e+00 4.95083839e-01 1.12185359e+00 -1.27898180e+00
9.46172535e-01 -3.12544815e-02 -5.00150502e-01 -2.14607015e-01
-1.90009177e-01 1.49758279e-01 6.00863636e-01 -9.99223232e-01
-1.00590456e+00 -1.47034585e-01 -2.13639945e-01 -7.76340291e-02
-3.84664573e-02 1.08480990e-01 3.70198727e-01 -9.32104647e-01
-5.09036779e-02 -8.73649299e-01 -3.19062322e-01 -5.72197497e-01
-2.47577578e-01 1.55245617e-01 2.64779001e-01 -5.79415083e-01
1.32424259e+00 -1.83810329e+00 -4.26918834e-01 8.72332752e-01
-2.39669055e-01 -3.81597012e-01 4.58462119e-01 6.76819146e-01
-7.00114071e-02 3.37406248e-01 -1.48701325e-01 3.32698673e-01
6.30962789e-01 -1.14127211e-02 -6.51829422e-01 4.20820087e-01
-4.15177733e-01 8.65379512e-01 -3.37555885e-01 -2.20630065e-01
6.64092824e-02 -5.09033322e-01 -5.54948449e-01 1.33047536e-01
-5.72626114e-01 2.83713222e-01 -7.02393472e-01 8.01064253e-01
5.02153933e-01 -2.23420307e-01 3.05689603e-01 -2.60031726e-02
-6.15738630e-01 9.60407779e-03 -1.42133498e+00 9.16029274e-01
-3.89057755e-01 1.08586088e-01 5.72472960e-02 -6.87987328e-01
1.01282954e+00 1.32977024e-01 6.86719060e-01 -7.66534328e-01
3.93195421e-01 7.21119761e-01 1.72235370e-01 1.31404534e-01
1.93457663e-01 -3.64125997e-01 -2.81653106e-01 1.05645037e+00
-2.68786162e-01 -7.28496388e-02 1.71155706e-01 -4.07149643e-01
8.08198094e-01 -3.23111534e-01 6.32005394e-01 -8.84612679e-01
1.08898677e-01 9.12634954e-02 5.86345077e-01 7.10554659e-01
-6.21467605e-02 2.34818757e-01 8.99119318e-01 -5.39350212e-01
-1.02248406e+00 -7.90693939e-01 -1.83246642e-01 4.22980189e-01
-2.24781752e-01 1.35656148e-01 -5.84576488e-01 -4.28185850e-01
7.88783371e-01 8.48019660e-01 -5.92230380e-01 5.64843357e-01
-3.52224141e-01 -1.00796151e+00 2.75799096e-01 2.38377228e-02
6.05003059e-01 -1.11319256e+00 -1.15935349e+00 4.91306365e-01
5.05215943e-01 -5.44345796e-01 -4.86909658e-01 6.38950229e-01
-9.28571284e-01 -1.28368497e+00 -6.10843897e-01 8.97459313e-02
2.37113819e-01 -2.82613248e-01 1.25635719e+00 -3.01984161e-01
2.39743754e-01 8.28498453e-02 -2.23058045e-01 -5.82076907e-01
-3.90338361e-01 1.69426546e-01 -1.94347203e-01 3.10658187e-01
-1.13815558e-03 -5.78991950e-01 -7.48455644e-01 3.84800553e-01
-6.42302811e-01 -3.74523580e-01 7.74508655e-01 8.01917136e-01
5.87027133e-01 1.88830364e-02 6.40179396e-01 -9.75704253e-01
9.34736729e-01 -3.96967351e-01 -1.81685364e+00 5.09961367e-01
-1.39962864e+00 4.72303331e-01 4.66120511e-01 -3.16748708e-01
-9.74995494e-01 -2.72941351e-01 4.97468531e-01 -4.31334645e-01
4.56025124e-01 5.53470552e-01 1.26637533e-01 6.49888292e-02
2.15085775e-01 -5.80123737e-02 1.22628123e-01 -8.36210191e-01
2.71004975e-01 3.03856313e-01 2.40737915e-01 -8.80577922e-01
9.05945718e-01 1.85538158e-01 -1.81953356e-01 -6.25770167e-03
-2.72666663e-01 -5.15685305e-02 -1.06104679e-01 6.26493320e-02
2.91694880e-01 -6.20547414e-01 -1.02907610e+00 4.83352840e-01
-6.66564286e-01 -5.19108713e-01 -6.80454552e-01 7.18488336e-01
-7.49326706e-01 -1.71705976e-01 -6.64626181e-01 -9.78705704e-01
-5.45614779e-01 -1.04533124e+00 4.78042960e-01 1.01623975e-01
-3.04742515e-01 -1.11531270e+00 5.55819511e-01 6.59914911e-02
6.20356143e-01 5.06755769e-01 7.82271624e-01 -1.13471627e+00
-9.16282654e-01 4.42010313e-02 -6.09971955e-02 3.58375967e-01
3.65652382e-01 2.93502152e-01 -6.28647447e-01 -4.59667534e-01
5.12641072e-01 1.55316651e-01 3.13028514e-01 6.33511543e-01
6.22765064e-01 -7.84506202e-01 -1.39774039e-01 5.94838560e-01
1.60962689e+00 7.20749617e-01 2.20670536e-01 1.00529897e+00
-2.04889819e-01 5.16173124e-01 9.54944313e-01 9.29330528e-01
-4.08959776e-01 5.75208604e-01 2.42998704e-01 1.96852431e-01
7.64528990e-01 -2.14322835e-01 1.57006413e-01 4.76308405e-01
2.34708786e-02 9.38846543e-02 -7.79720008e-01 3.83346736e-01
-1.57941747e+00 -9.03386116e-01 2.91597784e-01 2.26650238e+00
7.28220999e-01 5.13194919e-01 5.16281307e-01 -7.37473145e-02
5.75214922e-01 2.38254100e-01 -5.11714280e-01 -3.93529326e-01
-3.00706476e-01 1.52403027e-01 1.25524127e+00 3.32961023e-01
-7.02568591e-01 4.92255896e-01 6.96490717e+00 7.34832287e-01
-1.03405631e+00 -1.14056416e-01 9.41595554e-01 -2.01350763e-01
-7.35459268e-01 4.93366003e-01 -7.80511856e-01 7.73115337e-01
1.19770181e+00 -7.75471926e-01 5.14602184e-01 8.34174037e-01
3.62860590e-01 -4.73205559e-02 -8.35644782e-01 9.59756017e-01
-7.54567742e-01 -1.71991134e+00 -1.06792718e-01 6.05000138e-01
8.38511646e-01 -7.32011646e-02 3.04386050e-01 -3.40593159e-02
4.90644485e-01 -6.18291140e-01 1.04793596e+00 5.51842511e-01
1.88866362e-01 -1.01657414e+00 8.22658420e-01 3.79275903e-02
-8.80256176e-01 -5.40034533e-01 4.02123518e-02 3.77572961e-02
3.74071188e-02 8.38633478e-01 -3.94937038e-01 6.75362885e-01
6.76766813e-01 1.08813621e-01 -2.54164398e-01 8.60927165e-01
2.69966424e-01 4.73738998e-01 -6.75949812e-01 -1.03091061e-01
5.78909397e-01 -1.01482582e+00 2.52668232e-01 7.18784511e-01
6.92919850e-01 -3.52655686e-02 -2.30551198e-01 1.16274023e+00
-2.15018749e-01 1.77787080e-01 -5.16186595e-01 -1.15677454e-01
7.01547086e-01 1.15332258e+00 -9.83320892e-01 -1.37025803e-01
-4.34350371e-01 3.43541563e-01 -2.67252266e-01 1.56666741e-01
-9.12173986e-01 -6.60237074e-02 5.71506977e-01 1.51143000e-01
4.10732180e-01 7.72826839e-04 -6.16874337e-01 -9.30707157e-01
1.04387283e-01 -1.27296126e+00 6.92070365e-01 -8.02393481e-02
-1.21816456e+00 5.53545617e-02 1.21264234e-01 -1.48396957e+00
-5.12374997e-01 -5.29888809e-01 -6.64864063e-01 6.75821185e-01
-1.35047495e+00 -4.83611077e-01 5.65040827e-01 3.28643084e-01
1.91251501e-01 -6.17907643e-01 3.03075552e-01 -4.85410355e-02
-2.61914253e-01 4.10799235e-01 8.70226264e-01 -4.34105024e-02
2.82854378e-01 -1.44771934e+00 3.02813590e-01 4.70462471e-01
3.71130645e-01 5.72755814e-01 8.47841799e-01 -6.65058196e-01
-1.34358525e+00 -5.63751101e-01 4.81239736e-01 -2.74100900e-01
1.41603076e+00 -1.00333527e-01 -6.34658933e-01 8.16277742e-01
6.25431657e-01 -4.26525623e-01 3.92068446e-01 -6.57395497e-02
-4.93116258e-03 -7.03838289e-01 -1.28303289e+00 3.03214967e-01
3.93498152e-01 -1.58498183e-01 -6.31114542e-01 2.01273143e-01
3.50181550e-01 -9.72681195e-02 -1.29529774e+00 9.50144157e-02
6.48157537e-01 -1.29045153e+00 5.80179930e-01 -9.34547707e-02
-5.24881601e-01 -3.60586010e-02 7.97655713e-03 -1.30319190e+00
1.57872856e-01 -1.49355865e+00 9.07007828e-02 1.26320863e+00
8.35997105e-01 -1.33441234e+00 6.17794454e-01 6.88585639e-01
3.09513271e-01 -5.59920251e-01 -1.29399586e+00 -1.14535522e+00
2.37386316e-01 -4.07606475e-02 1.03179073e+00 8.89260352e-01
-6.87380508e-02 -7.05412999e-02 -2.91767806e-01 -1.64804146e-01
8.49707663e-01 9.81118798e-01 6.08681619e-01 -1.18036735e+00
-6.24771714e-01 -9.55624223e-01 2.81010330e-01 -4.42677081e-01
-1.83581069e-01 -5.69691539e-01 -5.25692105e-01 -4.28265393e-01
7.03296810e-02 -6.45191669e-01 -6.63361132e-01 2.62055010e-01
5.33064902e-01 -3.98880452e-01 3.66859287e-01 4.71633196e-01
-6.14421293e-02 2.26987183e-01 9.28642154e-01 -2.09838197e-01
-3.47826064e-01 2.63049990e-01 -6.78317070e-01 4.45957631e-01
1.12297833e+00 -4.30611730e-01 -2.99291730e-01 1.15507863e-01
4.02432621e-01 4.10301983e-01 1.45296067e-01 -6.75379157e-01
3.55499312e-02 -6.14112854e-01 2.35802922e-02 -8.29668999e-01
-3.87192726e-01 -1.07139456e+00 1.05170524e+00 8.06994259e-01
-3.06971192e-01 7.54397035e-01 -7.55589977e-02 3.85880679e-01
-4.38008845e-01 -2.14184120e-01 6.77101314e-01 -2.60272235e-01
-2.24746972e-01 1.60023481e-01 9.10593569e-02 1.70515582e-01
1.31993103e+00 7.85134658e-02 -1.76395074e-01 -2.18914196e-01
-3.51209670e-01 3.92515510e-01 5.27993083e-01 1.70672446e-01
-1.53600872e-01 -1.14584816e+00 -7.19522297e-01 -9.73408762e-03
-3.48326594e-01 -5.72153091e-01 -2.88204968e-01 6.65914953e-01
-7.65521348e-01 8.24303865e-01 -3.48664284e-01 -1.23380311e-01
-4.60804045e-01 7.04264402e-01 8.02192986e-01 -8.99566114e-01
-5.32901108e-01 6.68118447e-02 -1.32102042e-01 -2.67736427e-02
-6.90041995e-03 -8.98130536e-01 2.81717598e-01 6.12388313e-01
2.97824085e-01 5.50279975e-01 7.93729648e-02 1.65779993e-01
-7.93429688e-02 5.81531167e-01 2.49602646e-01 -5.14625192e-01
1.34913123e+00 9.39018652e-02 -1.72523577e-02 4.11915898e-01
8.76700997e-01 2.34131888e-01 -1.45292807e+00 2.32420638e-01
8.36794138e-01 -4.30500120e-01 -4.13566500e-01 -8.88031721e-01
-1.51774240e+00 2.88178563e-01 5.53139925e-01 7.54800320e-01
8.94781828e-01 -2.97820479e-01 5.26261449e-01 3.93537551e-01
6.94576502e-01 -1.44907439e+00 -6.52551830e-01 1.81446802e-02
1.03792572e+00 -8.63781035e-01 -1.27900476e-02 2.96939790e-01
-5.26312053e-01 9.75538731e-01 -4.76497933e-02 -2.88206041e-01
8.74015808e-01 5.77090859e-01 3.02015781e-01 -1.55199662e-01
-9.03032303e-01 3.05519879e-01 -2.71422565e-01 1.09942909e-02
-7.59431794e-02 3.65562379e-01 -8.81159544e-01 8.15915883e-01
-5.00370324e-01 -1.01169206e-01 3.11532855e-01 7.46466219e-01
-2.53329933e-01 -1.18432570e+00 -5.71415782e-01 6.86922252e-01
-7.15182483e-01 8.64668470e-03 -4.39759970e-01 1.49031389e+00
-4.31242555e-01 4.30321693e-01 1.96373880e-01 -1.91262532e-02
4.45835352e-01 1.86493080e-02 1.17394552e-01 7.21411854e-02
-1.22598577e+00 3.67367655e-01 1.01000659e-01 -4.86361235e-01
-1.80767268e-01 -9.41346943e-01 -1.01025808e+00 -5.28932571e-01
-2.50233412e-01 7.61044860e-01 3.76341194e-01 6.00148141e-01
2.16209278e-01 -5.20766191e-02 1.17351878e+00 -3.51695091e-01
-1.53255081e+00 -8.21652114e-01 -1.33085763e+00 2.98848629e-01
6.90753534e-02 -8.07188153e-01 -8.73918235e-01 -4.55452323e-01]
|
[4.715696334838867, 4.0537333488464355]
|
a5581e68-4e2c-4d61-8b53-f44115b372a4
|
deep-speaker-feature-learning-for-text
|
1705.03670
| null |
http://arxiv.org/abs/1705.03670v1
|
http://arxiv.org/pdf/1705.03670v1.pdf
|
Deep Speaker Feature Learning for Text-independent Speaker Verification
|
Recently deep neural networks (DNNs) have been used to learn speaker
features. However, the quality of the learned features is not sufficiently
good, so a complex back-end model, either neural or probabilistic, has to be
used to address the residual uncertainty when applied to speaker verification,
just as with raw features. This paper presents a convolutional time-delay deep
neural network structure (CT-DNN) for speaker feature learning. Our
experimental results on the Fisher database demonstrated that this CT-DNN can
produce high-quality speaker features: even with a single feature (0.3 seconds
including the context), the EER can be as low as 7.68%. This effectively
confirmed that the speaker trait is largely a deterministic short-time property
rather than a long-time distributional pattern, and therefore can be extracted
from just dozens of frames.
|
['Zhiyuan Tang', 'Lantian Li', 'Dong Wang', 'Yixiang Chen', 'Ying Shi']
|
2017-05-10
| null | null | null | null |
['text-independent-speaker-verification']
|
['speech']
|
[-1.76804841e-01 7.12997327e-03 1.22072026e-01 -1.07705355e+00
-1.23483956e+00 -4.60049152e-01 6.40518606e-01 -2.14791238e-01
-2.88458019e-01 5.06093919e-01 9.39031988e-02 -3.05930048e-01
-7.18024597e-02 -2.73272514e-01 -6.48276746e-01 -9.49304163e-01
-4.84432250e-01 1.90388039e-01 -9.43698883e-02 1.61322430e-02
-6.45233169e-02 4.53913033e-01 -1.71048856e+00 6.07398048e-04
4.52978104e-01 1.35748208e+00 -1.40701249e-01 7.86566198e-01
-2.68902272e-01 3.90344292e-01 -1.09060192e+00 -2.74000883e-01
-5.66730136e-03 -2.04182521e-01 -4.91626769e-01 -3.07832599e-01
4.84422088e-01 -5.74660301e-01 -5.05622804e-01 9.98260498e-01
9.00125086e-01 1.24822944e-01 6.78794503e-01 -1.15006268e+00
-5.54948926e-01 1.12107706e+00 3.43893208e-02 2.92898804e-01
3.04679811e-01 1.16625927e-01 1.05065060e+00 -9.64333653e-01
1.97401028e-02 1.37637615e+00 7.84887969e-01 7.40748942e-01
-9.23134089e-01 -1.01454341e+00 1.05885386e-01 2.82323390e-01
-1.59614515e+00 -9.50712800e-01 7.91678429e-01 -2.97566652e-01
8.45085382e-01 1.93745028e-02 4.67000842e-01 1.60304034e+00
2.96482384e-01 7.58823872e-01 8.78128052e-01 -2.92145550e-01
1.76284224e-01 1.17887303e-01 1.34884357e-01 3.84896725e-01
-2.32702896e-01 7.06684351e-01 -9.19913888e-01 -5.32861287e-03
2.97120214e-01 -1.58607736e-01 -1.28724068e-01 3.15668076e-01
-8.85995507e-01 8.24589610e-01 3.83188456e-01 4.88089651e-01
-1.99527755e-01 3.79837126e-01 4.85893548e-01 3.78307372e-01
3.33663851e-01 -6.34795427e-02 -6.21997952e-01 -5.23016810e-01
-1.27861273e+00 3.91214378e-02 7.28923082e-01 8.38826478e-01
5.19247413e-01 7.60957122e-01 -2.06050277e-01 5.94163120e-01
5.32236218e-01 8.79386127e-01 5.24659932e-01 -6.08666897e-01
1.05982646e-01 -1.52275562e-02 -1.59486886e-02 -6.88517272e-01
-3.64169419e-01 -5.10400474e-01 -8.48010540e-01 1.53270170e-01
4.91405100e-01 -4.33012605e-01 -1.04957545e+00 1.92734051e+00
3.57858054e-02 2.30159149e-01 -4.49245051e-02 6.91822112e-01
1.18941987e+00 6.73637450e-01 -2.73591697e-01 6.78996295e-02
1.30318165e+00 -3.18355262e-01 -9.52924728e-01 1.26252370e-02
9.30356532e-02 -6.86215222e-01 7.84826875e-01 3.66881579e-01
-7.80689538e-01 -6.61226928e-01 -1.07344317e+00 1.77525058e-01
-2.28419185e-01 -7.43776485e-02 4.42489535e-01 1.14341700e+00
-1.24845612e+00 6.29286885e-01 -7.72134900e-01 1.33145228e-01
3.57940763e-01 5.02319574e-01 -2.61087179e-01 1.53020054e-01
-1.64010775e+00 6.30857348e-01 9.21584144e-02 5.67935407e-01
-1.34841049e+00 -6.49575353e-01 -8.12635601e-01 3.45817387e-01
-2.13234961e-01 -1.79799020e-01 1.52722085e+00 -8.41586411e-01
-2.09691095e+00 3.46460372e-01 -4.70088482e-01 -5.81391811e-01
3.40114295e-01 -1.40411183e-01 -8.87486935e-01 2.53655668e-02
-3.54245543e-01 5.74152648e-01 1.14876568e+00 -9.73686457e-01
-5.07280171e-01 -2.06828102e-01 -3.60160977e-01 -4.31010753e-01
-8.12536329e-02 4.01462585e-01 -2.70731524e-02 -4.47412878e-01
8.91900212e-02 -8.73669147e-01 2.41816416e-01 -1.85203373e-01
-4.59191322e-01 -4.06809628e-01 7.26535559e-01 -8.44016850e-01
1.05597615e+00 -2.37981820e+00 -5.54973066e-01 1.35881990e-01
-1.50166705e-01 1.70162439e-01 -9.29630548e-02 2.54869610e-01
-1.60643011e-01 1.45088181e-01 -1.67521268e-01 -5.69918811e-01
3.62345129e-01 -1.02206968e-01 -3.82914662e-01 6.00279450e-01
3.15412819e-01 8.51275980e-01 -5.09727418e-01 -8.56933892e-02
7.42244441e-03 9.58981752e-01 -1.60870552e-01 1.66145250e-01
3.43569368e-02 2.57169336e-01 -1.98695678e-02 8.05450082e-01
8.15735817e-01 3.20512474e-01 -2.79764414e-01 -5.49274683e-03
7.71209449e-02 6.68165386e-01 -1.03347957e+00 1.60544229e+00
-4.41148311e-01 1.22979152e+00 2.58221570e-02 -6.18701160e-01
1.09715188e+00 6.90651834e-01 4.18981105e-01 -5.94159067e-01
3.87297392e-01 1.67607829e-01 2.45908275e-01 -3.47545028e-01
3.32139909e-01 -4.03257430e-01 -1.73039645e-01 4.99201447e-01
3.45014870e-01 -1.05475940e-01 -4.15249258e-01 -1.24300614e-01
8.07321846e-01 -2.73043633e-01 -2.14969411e-01 -2.53219813e-01
3.79513919e-01 -6.14855349e-01 7.81602085e-01 6.88114464e-01
-4.99437720e-01 7.43242443e-01 3.78775626e-01 -2.83056885e-01
-8.09051514e-01 -9.96230125e-01 -4.73332167e-01 1.05237281e+00
-2.05170035e-01 -2.15623617e-01 -7.32733369e-01 -4.15943712e-01
-4.81716134e-02 7.47774839e-01 -4.97477740e-01 -3.64039272e-01
-3.55424523e-01 -4.45850164e-01 1.10690355e+00 5.22662282e-01
2.76297688e-01 -9.62486267e-01 -1.59758359e-01 5.16765893e-01
-8.93296581e-03 -1.05658662e+00 -6.77068591e-01 5.89237988e-01
-5.93302369e-01 -4.47968841e-01 -4.21502173e-01 -6.39358938e-01
1.28575251e-01 -1.80036306e-01 9.09011483e-01 -3.90196264e-01
-8.19750130e-02 9.43598226e-02 -1.94240764e-01 -6.76655054e-01
-5.45880437e-01 -2.24790141e-01 4.97783750e-01 1.47847399e-01
7.56797194e-01 -6.78480804e-01 -4.11184639e-01 1.10440277e-01
-7.17305124e-01 -8.62322569e-01 4.05611247e-01 1.21325326e+00
-2.25069020e-02 3.42550844e-01 1.09924567e+00 -1.48586631e-01
5.37182748e-01 -1.03646815e-01 -6.39361084e-01 -5.21481708e-02
-5.07808447e-01 2.72361040e-01 3.35729659e-01 -6.96711898e-01
-1.09770286e+00 1.50533572e-01 -6.14067793e-01 -3.88824552e-01
-3.98158282e-01 4.58289236e-01 -3.72224033e-01 2.14109272e-01
5.14567316e-01 3.62288654e-01 1.01768166e-01 -3.68194669e-01
1.96597904e-01 9.90391672e-01 4.43776935e-01 -5.30135095e-01
7.20959902e-01 2.02062622e-01 -4.51916307e-01 -8.63689721e-01
-6.54623568e-01 -2.04042956e-01 -3.89979929e-01 -8.11971501e-02
6.01836920e-01 -1.13096726e+00 -1.03558850e+00 7.65277922e-01
-1.11746800e+00 -1.96523160e-01 -2.88879216e-01 8.76295567e-01
-2.00995252e-01 1.11615807e-01 -7.81180203e-01 -1.26443529e+00
-4.85483378e-01 -1.18290997e+00 1.03030586e+00 2.04372093e-01
-6.31579608e-02 -7.57656038e-01 -1.61233515e-01 -3.38054150e-02
6.98352396e-01 -1.15919396e-01 6.71175539e-01 -9.29130495e-01
-4.51633811e-01 -4.56803888e-01 -2.27042884e-02 6.25701189e-01
2.14460462e-01 3.08779567e-01 -1.85589528e+00 -3.61843169e-01
2.97596246e-01 -1.86878070e-01 7.55157828e-01 6.85941756e-01
1.12269247e+00 -1.79249361e-01 8.73633623e-02 5.17258108e-01
9.91933346e-01 2.67125458e-01 4.11306441e-01 -4.57105376e-02
3.48902434e-01 3.56433004e-01 1.03434883e-01 5.16840458e-01
3.12580824e-01 5.03759086e-01 4.19261009e-01 3.71100515e-01
-1.63333699e-01 -2.74463054e-02 9.83968616e-01 1.01506937e+00
2.58986861e-01 -2.53298253e-01 -7.82031000e-01 4.69826400e-01
-1.24963033e+00 -1.16072953e+00 9.81229916e-02 2.16559911e+00
8.02067578e-01 2.71733433e-01 1.57368228e-01 3.15085232e-01
7.60608912e-01 2.06977099e-01 -6.87402725e-01 -4.99784976e-01
-2.31794119e-01 8.62201080e-02 9.15819034e-02 4.25605506e-01
-8.72220635e-01 6.05634212e-01 7.35733986e+00 6.59389257e-01
-1.72598910e+00 -3.44445668e-02 4.73558396e-01 -1.99747324e-01
-2.82567620e-01 -4.76638615e-01 -1.18812847e+00 6.17820919e-01
1.73856151e+00 -1.12547822e-01 1.91226900e-01 8.80642593e-01
2.24919990e-01 2.56366849e-01 -1.29147863e+00 1.15798354e+00
-1.77296549e-02 -8.87929559e-01 -2.53186405e-01 2.24079832e-01
3.95377159e-01 3.79453540e-01 4.83987570e-01 6.44229591e-01
2.76267320e-01 -1.34648240e+00 1.07528806e+00 4.37563449e-01
9.17438388e-01 -1.04000902e+00 1.06662345e+00 3.34634393e-01
-1.02965498e+00 -6.23903424e-02 -3.98942083e-01 2.62958825e-01
2.30530888e-01 1.06324840e+00 -1.23894393e+00 3.31183195e-01
7.80787051e-01 3.58927190e-01 -2.74162412e-01 9.72070515e-01
-4.40343916e-02 1.09781170e+00 -6.57436132e-01 -1.90818742e-01
1.57993644e-01 4.11542088e-01 5.32996953e-01 1.37006164e+00
7.40772665e-01 -3.76003385e-01 -2.42844507e-01 7.25018024e-01
-1.56368583e-01 -3.31111103e-01 -3.50602806e-01 -9.82578993e-02
6.64996028e-01 1.00973308e+00 -1.47759423e-01 -1.62402749e-01
-1.99675158e-01 6.37997091e-01 9.19488966e-02 4.11750704e-01
-9.47191417e-01 -4.37105566e-01 9.39826906e-01 -3.21381807e-01
7.16539681e-01 -2.60254562e-01 -7.13308016e-03 -1.05708611e+00
9.39930677e-02 -8.18581939e-01 4.58077714e-03 -3.71389449e-01
-1.47270060e+00 9.84697104e-01 -3.10691684e-01 -1.26555693e+00
-8.18754017e-01 -6.03445649e-01 -7.19731212e-01 1.01489234e+00
-1.66594756e+00 -9.64529097e-01 9.69621912e-02 6.11996949e-01
2.94468135e-01 -3.03057104e-01 1.09622014e+00 3.62041652e-01
-5.38742959e-01 1.01554728e+00 3.70729208e-01 2.34060094e-01
7.17524230e-01 -1.24282086e+00 4.44066793e-01 7.85485268e-01
2.73380071e-01 5.51940143e-01 7.28022039e-01 2.15873122e-03
-1.45201492e+00 -8.46941888e-01 1.01353240e+00 -4.69099164e-01
5.88202298e-01 -5.34989297e-01 -1.00560391e+00 4.15097982e-01
3.96456003e-01 2.70997703e-01 8.22375774e-01 3.74831617e-01
-5.86646855e-01 -3.67418885e-01 -1.35696566e+00 8.67945030e-02
7.19323575e-01 -1.05594063e+00 -7.03114808e-01 -1.82565227e-01
8.30266476e-01 -3.04767847e-01 -7.92536616e-01 3.07760239e-01
6.20468378e-01 -1.00456822e+00 8.50666106e-01 -3.33560944e-01
-1.72195956e-01 -2.61648536e-01 -4.24635798e-01 -1.46982598e+00
-1.32794306e-01 -6.99855983e-01 -2.57889271e-01 1.73532593e+00
6.39237225e-01 -5.93205869e-01 6.10308409e-01 5.91381669e-01
-1.79366052e-01 -3.71601582e-01 -1.44985354e+00 -1.15329587e+00
7.08931386e-02 -1.10365641e+00 1.15244222e+00 6.15709960e-01
-2.21625999e-01 3.80610406e-01 -3.85505110e-01 4.78909463e-01
5.35301805e-01 -1.00434921e-03 3.33237648e-01 -1.45392787e+00
2.16088053e-02 -5.55080891e-01 -5.86243808e-01 -1.00379634e+00
5.66040754e-01 -6.50120854e-01 6.46908879e-01 -1.00606787e+00
-2.22005114e-01 -3.57279807e-01 -6.40287340e-01 1.95637807e-01
7.20746210e-03 -1.89652637e-01 -2.10686430e-01 -4.14131522e-01
-2.22205222e-02 8.60514224e-01 6.91475689e-01 -4.44758534e-01
-1.18683159e-01 3.39725047e-01 -6.12646103e-01 4.46828067e-01
8.33446026e-01 -5.06031573e-01 -2.05633521e-01 -2.68861890e-01
-2.00555757e-01 3.05675231e-02 1.86550930e-01 -1.05781293e+00
3.75496745e-01 2.68221676e-01 4.96000499e-01 -9.53048766e-01
6.77714825e-01 -8.70071590e-01 -4.55821911e-03 2.48224363e-01
-3.51029277e-01 -3.07121098e-01 3.34419221e-01 6.02406323e-01
-6.47877455e-01 -1.63677603e-01 6.17697179e-01 2.90706843e-01
-5.17768681e-01 3.98373216e-01 -5.35324454e-01 -2.14703918e-01
4.28913951e-01 2.53747199e-02 3.48063596e-02 -6.45180523e-01
-6.52202010e-01 -2.52093792e-01 -2.05466866e-01 4.27072227e-01
6.08533919e-01 -1.46277964e+00 -8.60713363e-01 4.03848141e-01
7.56139830e-02 -8.59474856e-03 4.20763403e-01 3.67689967e-01
5.84491603e-02 6.78744733e-01 1.80122212e-01 -1.18143702e+00
-1.05040240e+00 4.57201064e-01 3.96067202e-01 2.80817479e-01
-5.21881342e-01 1.17218399e+00 -1.00296691e-01 -4.75085557e-01
7.39461541e-01 -7.20481813e-01 -1.30123561e-02 3.01481187e-01
8.44169378e-01 -6.15337752e-02 3.34572703e-01 -7.28996992e-01
-7.14437485e-01 3.52997929e-01 2.55756658e-02 -4.16542679e-01
1.61102116e+00 -2.97130853e-01 3.27533692e-01 6.69721842e-01
1.41550899e+00 2.72150263e-02 -1.44341326e+00 -4.10911202e-01
-3.10836900e-02 -2.22326040e-01 3.79698992e-01 -6.25652015e-01
-1.04902041e+00 1.20901561e+00 9.50101316e-01 4.00003880e-01
8.74847233e-01 -8.01978335e-02 7.96577036e-01 5.48709810e-01
2.04524398e-01 -9.08723176e-01 -1.27311319e-01 7.83147097e-01
9.35275912e-01 -1.54578817e+00 -5.27706861e-01 1.71185300e-01
-4.96439189e-01 1.31640196e+00 1.88053086e-01 2.41964951e-01
1.19058001e+00 4.36226517e-01 2.72847682e-01 -4.52136770e-02
-8.45160663e-01 -7.25919455e-02 4.11863476e-01 5.78484714e-01
5.73733687e-01 2.73060828e-01 6.73161805e-01 7.79311597e-01
-6.54120088e-01 -2.02044338e-01 3.49323452e-01 4.92372841e-01
-3.07070524e-01 -8.56459141e-01 -3.72523725e-01 2.69431114e-01
-5.87810576e-01 -1.94201782e-01 2.35979781e-02 3.55176240e-01
-1.38166919e-01 1.28159320e+00 5.85893318e-02 -6.44768059e-01
1.27496541e-01 5.20718277e-01 1.81784526e-01 -2.94158161e-01
-6.18454039e-01 1.77525640e-01 1.62717387e-01 -4.18688267e-01
-4.00984019e-01 -7.18819022e-01 -1.14264333e+00 -5.76216519e-01
-5.99486113e-01 1.47045657e-01 1.16448331e+00 8.87526751e-01
2.82710791e-01 6.37201428e-01 1.00134099e+00 -8.61845195e-01
-7.58987606e-01 -1.31780124e+00 -8.79717350e-01 -1.56390041e-01
1.00075185e+00 -4.67372864e-01 -8.02062809e-01 -8.99668690e-03]
|
[14.380437850952148, 6.058828353881836]
|
a31efd49-e09d-4215-9250-a02274893a31
|
surrogate-neural-networks-for-efficient
|
2303.17468
| null |
https://arxiv.org/abs/2303.17468v1
|
https://arxiv.org/pdf/2303.17468v1.pdf
|
Surrogate Neural Networks for Efficient Simulation-based Trajectory Planning Optimization
|
This paper presents a novel methodology that uses surrogate models in the form of neural networks to reduce the computation time of simulation-based optimization of a reference trajectory. Simulation-based optimization is necessary when there is no analytical form of the system accessible, only input-output data that can be used to create a surrogate model of the simulation. Like many high-fidelity simulations, this trajectory planning simulation is very nonlinear and computationally expensive, making it challenging to optimize iteratively. Through gradient descent optimization, our approach finds the optimal reference trajectory for landing a hypersonic vehicle. In contrast to the large datasets used to create the surrogate models in prior literature, our methodology is specifically designed to minimize the number of simulation executions required by the gradient descent optimizer. We demonstrated this methodology to be more efficient than the standard practice of hand-tuning the inputs through trial-and-error or randomly sampling the input parameter space. Due to the intelligently selected input values to the simulation, our approach yields better simulation outcomes that are achieved more rapidly and to a higher degree of accuracy. Optimizing the hypersonic vehicle's reference trajectory is very challenging due to the simulation's extreme nonlinearity, but even so, this novel approach found a 74% better-performing reference trajectory compared to nominal, and the numerical results clearly show a substantial reduction in computation time for designing future trajectories.
|
['Jonathan P. How', 'Piero Miotto', 'Matthew Stoeckle', 'Rebecca Russell', 'Evelyn Ruff']
|
2023-03-30
| null | null | null | null |
['trajectory-planning']
|
['robots']
|
[-1.93135932e-01 -3.35979134e-01 2.71953847e-02 -3.74158472e-02
-5.50411582e-01 -7.63162553e-01 4.78590041e-01 -1.12061657e-01
-4.55343992e-01 9.38384354e-01 -2.20487803e-01 -9.63447630e-01
-3.97939831e-01 -8.28489244e-01 -7.91750908e-01 -5.97902894e-01
-2.86662340e-01 4.97168779e-01 -1.34712443e-01 -5.98188102e-01
3.94535154e-01 9.12173986e-01 -1.63426328e+00 -4.87796813e-01
9.47137296e-01 7.38953769e-01 1.64713729e-02 7.58446813e-01
4.35942233e-01 1.71927303e-01 -4.89859194e-01 5.35251051e-02
6.03403032e-01 -5.89635551e-01 -2.58005142e-01 -4.25837964e-01
5.51764704e-02 -3.09962839e-01 -1.62222460e-01 8.28436494e-01
5.02002954e-01 7.65347302e-01 6.67917907e-01 -1.08082831e+00
2.36929640e-01 5.64877391e-02 1.97790518e-01 -2.17536706e-02
3.15568335e-02 7.39131451e-01 3.45300406e-01 -6.08735919e-01
5.62068522e-01 8.80519569e-01 9.80009377e-01 1.82666421e-01
-1.28659415e+00 -6.72870636e-01 -2.44723126e-01 -1.41493276e-01
-1.53805912e+00 -3.25348675e-01 5.15725613e-01 -6.00675941e-01
1.09490180e+00 4.53540504e-01 1.11816955e+00 4.55948114e-01
4.09340799e-01 -1.12050809e-01 8.16359162e-01 -2.82338351e-01
3.64370346e-01 2.15132803e-01 -4.10407901e-01 5.58665574e-01
4.89322692e-01 8.71415615e-01 2.88643122e-01 -2.06734374e-01
7.27645338e-01 -1.65791661e-01 -4.04644370e-01 -3.55532050e-01
-9.10161674e-01 7.85561025e-01 3.21120828e-01 5.89675307e-02
-5.25807679e-01 3.28415990e-01 2.61910379e-01 3.52832198e-01
-7.58416206e-02 1.24408543e+00 -3.06218386e-01 -5.36016405e-01
-1.39554656e+00 6.87661886e-01 1.18122029e+00 5.26144624e-01
4.83412623e-01 6.91258907e-01 2.34054953e-01 1.69741839e-01
3.49640697e-01 6.90147817e-01 1.90240264e-01 -1.16257906e+00
4.12094295e-01 4.64377970e-01 6.60960674e-01 -1.07105196e+00
-5.61395764e-01 -6.90635443e-01 -3.46986920e-01 9.25460994e-01
6.18838847e-01 -8.41446400e-01 -7.63418913e-01 1.39127409e+00
3.10209870e-01 -4.83645573e-02 3.54771823e-01 1.16373038e+00
5.70050068e-02 9.36723709e-01 -1.18135750e-01 -1.83945209e-01
5.21501899e-01 -6.27664208e-01 -3.98775756e-01 1.64549872e-02
7.62884378e-01 -5.89404881e-01 9.73709822e-01 2.20245779e-01
-1.08862352e+00 -2.59171963e-01 -1.59283614e+00 6.09438360e-01
-5.20266771e-01 2.79443245e-03 3.88978004e-01 7.70746768e-01
-9.44737613e-01 1.14912152e+00 -7.63532400e-01 -1.52357310e-01
-9.02344659e-02 5.94295323e-01 1.39865905e-01 4.98024762e-01
-1.24989700e+00 1.40055704e+00 5.01949489e-01 2.96436667e-01
-9.55095828e-01 -9.37158167e-01 -7.32915521e-01 9.90000218e-02
2.68027633e-01 -5.01540899e-01 1.22959340e+00 -7.89801657e-01
-1.83008194e+00 -1.53342099e-03 -1.43337235e-01 -4.75332946e-01
6.56590104e-01 9.55429375e-02 -3.33040804e-01 -7.31328279e-02
-4.88471627e-01 2.78742731e-01 5.17496645e-01 -1.16223109e+00
-3.73582393e-01 3.08654785e-01 4.01793160e-02 3.34723353e-01
1.37709752e-01 -1.67937160e-01 -1.28497273e-01 -2.10706323e-01
-2.22536504e-01 -1.21399975e+00 -7.33175874e-01 -2.36373812e-01
7.75835514e-02 4.24436718e-01 8.87620926e-01 -5.98547578e-01
1.36230421e+00 -1.63593554e+00 -5.60981259e-02 5.72785735e-01
-2.40621492e-01 6.22933090e-01 2.53775269e-01 8.97377431e-01
-2.84207612e-02 2.16410518e-01 -2.50897080e-01 2.54681110e-01
1.83003850e-03 -8.60485509e-02 -5.10322571e-01 5.14436901e-01
3.25858407e-02 7.30927110e-01 -9.03109550e-01 -2.26129722e-02
5.00884533e-01 3.48584712e-01 -5.58801234e-01 1.28531620e-01
-1.52524024e-01 4.21990037e-01 -3.80404949e-01 3.68848652e-01
3.01305711e-01 2.84069534e-02 1.27872616e-01 -3.36671956e-02
-6.20094001e-01 4.27873097e-02 -1.29524887e+00 9.68628049e-01
-7.95200050e-01 8.37895691e-01 2.63631374e-01 -9.57667947e-01
1.18845165e+00 2.31002539e-01 4.35818970e-01 -5.51186264e-01
4.39007670e-01 5.59413970e-01 1.52621448e-01 -4.34907615e-01
7.34890282e-01 -4.03136343e-01 -1.61852408e-02 2.99273133e-01
-4.50818539e-01 -9.52789545e-01 2.82006353e-01 -2.45661736e-01
5.76174378e-01 2.81344235e-01 1.56753846e-02 -5.01818717e-01
5.43580711e-01 7.26165891e-01 3.85041714e-01 4.61403489e-01
1.30803496e-01 3.05134833e-01 4.18772697e-01 -4.47943598e-01
-1.39368451e+00 -6.00859225e-01 -1.60048500e-01 1.65515572e-01
3.52374583e-01 -1.09380998e-01 -7.10544825e-01 -2.00062823e-02
3.17542940e-01 1.18731213e+00 -2.93213189e-01 -3.62854838e-01
-7.01510549e-01 -7.17947721e-01 5.06134689e-01 3.43047202e-01
1.39085457e-01 -6.41265392e-01 -1.02283919e+00 4.15277362e-01
5.14728189e-01 -5.11650801e-01 -2.92044044e-01 1.89280450e-01
-1.12028337e+00 -9.75358129e-01 -3.70465636e-01 -2.80734777e-01
8.07306468e-01 -2.90684551e-01 8.14673424e-01 3.02532196e-01
-1.09372333e-01 -5.52269034e-02 1.40421435e-01 -4.10729676e-01
-5.70796311e-01 -2.53858954e-01 2.53110796e-01 -4.60677058e-01
-1.52243748e-01 -3.60864758e-01 -4.48417753e-01 4.95593548e-01
-6.17147326e-01 -1.28331840e-01 3.07590902e-01 7.72783637e-01
5.16432166e-01 3.10088307e-01 5.20576775e-01 -5.34823835e-01
1.03616917e+00 -5.49895346e-01 -1.43335354e+00 -1.15423493e-01
-1.05488396e+00 1.13571547e-01 1.28345501e+00 -4.42605287e-01
-7.55393386e-01 1.48720548e-01 -4.47005257e-02 -6.58978224e-01
5.10943159e-02 8.41890216e-01 2.55001247e-01 -5.99638820e-01
6.75648093e-01 -4.16240282e-02 4.92167264e-01 -2.08511323e-01
-7.86870494e-02 4.11474615e-01 3.53333414e-01 -4.45917368e-01
1.01613390e+00 1.48061275e-01 5.10852039e-01 -7.65985668e-01
-5.06462902e-02 7.51604438e-02 -4.28740084e-01 -6.00796759e-01
3.39668542e-01 -6.37921751e-01 -1.07055748e+00 8.46231580e-02
-6.58145845e-01 -7.01763391e-01 -2.82314628e-01 8.67015541e-01
-6.19067490e-01 -2.00434312e-01 6.51261434e-02 -9.72655773e-01
-2.76558131e-01 -1.44494045e+00 4.53175396e-01 4.68281507e-01
-5.31994104e-01 -1.17665505e+00 1.06206805e-01 -2.37728000e-01
8.28756511e-01 8.33505511e-01 7.26789832e-01 -2.51593441e-01
-7.12562323e-01 -6.40620530e-01 2.08615020e-01 1.05154313e-01
-3.27864170e-01 3.74124557e-01 -2.89564341e-01 -5.44756949e-01
3.61894034e-02 -1.53752826e-02 2.69780040e-01 4.03040498e-01
6.92266166e-01 -4.86891240e-01 -4.39256459e-01 1.03748393e+00
1.83044612e+00 7.27955163e-01 4.35960382e-01 5.26281238e-01
3.45812827e-01 4.87669319e-01 9.80881453e-01 2.27695063e-01
1.16132293e-02 6.17143393e-01 3.30062747e-01 -1.67562485e-01
2.61538923e-01 -2.94859827e-01 3.18037808e-01 4.71954137e-01
-2.27397591e-01 -1.64525598e-01 -1.22057760e+00 5.10035574e-01
-1.53135848e+00 -1.03525424e+00 -2.23816156e-01 2.56952977e+00
4.50634539e-01 6.13279082e-02 -1.30403629e-02 -7.06592351e-02
4.61122870e-01 -1.56301647e-01 -6.42437756e-01 -9.64057982e-01
3.44797403e-01 -7.37000555e-02 1.08183396e+00 6.08587801e-01
-7.46473968e-01 4.72981602e-01 7.41049433e+00 5.05003273e-01
-1.60698485e+00 -4.35799360e-01 1.94190383e-01 -5.44536173e-01
-3.23485434e-01 3.64359587e-01 -8.38508785e-01 5.46411574e-01
1.60269725e+00 -7.13047862e-01 7.37426519e-01 6.85567677e-01
9.23820496e-01 -3.27830762e-01 -8.47790599e-01 6.10850990e-01
-3.10319066e-01 -1.66936076e+00 -2.33387977e-01 8.55750442e-02
8.49816203e-01 -7.58650750e-02 -1.85721591e-01 3.30464005e-01
2.56390989e-01 -1.28425360e+00 9.59162235e-01 8.78371894e-01
7.91396499e-01 -1.18807161e+00 6.71968937e-01 6.09528720e-01
-9.01781321e-01 -1.87463254e-01 -2.40085460e-02 -3.80981624e-01
4.48321700e-01 2.11086422e-01 -1.03408730e+00 3.95653188e-01
3.29258174e-01 1.22237563e-01 -9.98322889e-02 1.33705068e+00
9.97867137e-02 8.30541313e-01 -6.60743594e-01 -7.92046845e-01
5.89485645e-01 -6.15536153e-01 9.18716967e-01 1.00595915e+00
4.97132152e-01 -2.32279152e-02 1.83348447e-01 1.02444303e+00
4.67324138e-01 -1.20561823e-01 -8.82109821e-01 -3.13093901e-01
5.11424780e-01 9.74793613e-01 -5.11570156e-01 -9.76955369e-02
3.41582261e-02 -3.10734808e-02 -2.59086460e-01 6.02469385e-01
-9.99827564e-01 -1.06754684e+00 6.68515563e-01 2.52570391e-01
2.77758002e-01 -7.75217175e-01 -4.56985921e-01 -4.41912293e-01
-1.22489221e-01 -7.76209474e-01 -2.42089570e-01 -5.84868252e-01
-4.21199799e-01 5.81546605e-01 3.02654952e-01 -1.52206910e+00
-1.00033307e+00 -5.47879040e-01 -9.27461684e-01 1.48396301e+00
-1.08369982e+00 -4.45778936e-01 -7.54035190e-02 -2.66594663e-02
1.25712097e-01 -2.17973605e-01 6.04666352e-01 1.36388227e-01
-4.65714037e-01 3.86446536e-01 6.15113437e-01 -4.18303549e-01
1.58018798e-01 -9.14019406e-01 3.10131431e-01 9.15604472e-01
-6.59318447e-01 7.75813103e-01 1.28194368e+00 -9.45103526e-01
-1.91221786e+00 -8.49302292e-01 4.76387292e-01 -2.09124282e-01
8.79201233e-01 7.94831850e-03 -7.78307617e-01 3.00026953e-01
-2.21671630e-02 -4.61935043e-01 7.96089917e-02 -2.93820024e-01
8.17600548e-01 -8.54333863e-02 -1.03216457e+00 9.40815449e-01
5.13716578e-01 -2.47391775e-01 -2.98313588e-01 8.69661123e-02
4.56822842e-01 -7.51004159e-01 -9.56954122e-01 5.38542867e-01
6.42421544e-01 -5.65700054e-01 8.34095120e-01 -4.49266702e-01
9.81494784e-02 -6.82101488e-01 1.99960962e-01 -1.57527065e+00
1.13516442e-01 -1.04129326e+00 4.58504632e-02 8.10888469e-01
8.61109257e-01 -8.62969935e-01 7.44058013e-01 9.29583371e-01
-3.34040105e-01 -1.22780657e+00 -7.94034839e-01 -1.13839233e+00
2.78202087e-01 -4.47342515e-01 7.82952607e-01 6.04483724e-01
-1.12619653e-01 -3.21923941e-01 -1.77814942e-02 3.48371357e-01
4.88370091e-01 3.26300859e-01 9.19604182e-01 -8.62400174e-01
-9.46791247e-02 -5.34539521e-01 -1.32971272e-01 -5.63212335e-01
2.50219926e-02 -7.83735454e-01 3.90633374e-01 -1.37283492e+00
-7.28313029e-01 -7.31333554e-01 2.27592215e-01 5.76596186e-02
1.25439301e-01 -1.52116805e-01 1.99367598e-01 6.57767281e-02
1.73340514e-01 5.60000062e-01 1.24986506e+00 2.27076128e-01
-6.42060459e-01 2.41745725e-01 -3.14507335e-01 7.10326493e-01
7.91113853e-01 -4.56878752e-01 -5.07776678e-01 -4.66268025e-02
2.56409347e-01 4.13308173e-01 5.00626028e-01 -1.25047934e+00
3.83272469e-01 -4.60860312e-01 4.88936067e-01 -6.70699358e-01
2.91423708e-01 -9.88957167e-01 7.51170754e-01 6.87620997e-01
2.35303864e-02 2.09084079e-01 7.98354685e-01 2.30166331e-01
-2.05708385e-01 -6.01031899e-01 8.24520469e-01 1.72032192e-01
-6.13013923e-01 -2.18538474e-02 -7.42128253e-01 -9.07536745e-02
1.17545581e+00 -4.61774796e-01 -4.95698815e-03 -5.36881208e-01
-4.13281739e-01 4.84042197e-01 8.54322314e-01 1.01667479e-01
3.69400531e-01 -1.01904500e+00 -3.53217989e-01 2.04956368e-01
-5.69802761e-01 -2.24451676e-01 -1.00455709e-01 8.51676226e-01
-9.93150592e-01 6.28902376e-01 -1.90751731e-01 -3.66001040e-01
-8.15164208e-01 3.14646214e-01 8.54212582e-01 -1.99965332e-02
-6.07549071e-01 3.40520799e-01 -5.50236344e-01 -3.83954018e-01
-1.08116955e-01 -2.56954193e-01 1.67286217e-01 -1.15500525e-01
1.50600225e-01 6.14334762e-01 1.38675958e-01 -5.26073813e-01
-2.41607100e-01 5.74160635e-01 6.00818872e-01 -4.12696540e-01
1.21098578e+00 1.36835948e-01 2.73078084e-01 3.00951540e-01
1.19140995e+00 3.43429460e-03 -1.59549034e+00 7.03822494e-01
-1.83371782e-01 -4.47829247e-01 3.51509511e-01 -8.41016531e-01
-8.48640800e-01 4.38161135e-01 3.26201558e-01 5.01736673e-03
9.71212029e-01 -7.58801341e-01 6.87641919e-01 7.14116871e-01
3.87052566e-01 -1.28451896e+00 -7.20188260e-01 7.45247781e-01
9.82390344e-01 -7.48990059e-01 1.87308460e-01 1.88550711e-01
-5.22026062e-01 1.20674992e+00 5.75464427e-01 -3.83914888e-01
5.11911333e-01 5.42015791e-01 9.79227498e-02 -3.66299450e-02
-6.20725274e-01 2.40846768e-01 2.33621255e-01 1.36105761e-01
-5.97801991e-02 -9.43136439e-02 -4.48552102e-01 3.26860666e-01
-5.18975735e-01 1.99952304e-01 6.07718229e-01 9.06771362e-01
-4.32783455e-01 -7.92972803e-01 -6.42639518e-01 5.26588082e-01
-1.40385451e-02 1.55378088e-01 3.77478376e-02 1.21929395e+00
-2.89059937e-01 8.89442623e-01 6.10815063e-02 -4.75672781e-01
7.31918871e-01 -1.75061841e-02 8.44002068e-02 -8.87096822e-02
-9.17982817e-01 -4.36511636e-01 4.65418249e-01 -6.82755113e-01
2.02022046e-01 -6.90772593e-01 -1.59988439e+00 -5.46958625e-01
-3.52115899e-01 5.76746285e-01 1.04400694e+00 8.47242057e-01
3.67278039e-01 5.20887315e-01 7.22987473e-01 -1.29527259e+00
-6.87929869e-01 -4.13622707e-01 -2.85606235e-01 4.67918329e-02
3.72978091e-01 -8.15893114e-01 -6.30938590e-01 -2.78519005e-01]
|
[5.261869430541992, 2.1408259868621826]
|
a228a8e9-ab0d-4af2-9d60-0ea67c309bc9
|
look-harder-a-neural-machine-translation
| null | null |
https://aclanthology.org/P19-1290
|
https://aclanthology.org/P19-1290.pdf
|
Look Harder: A Neural Machine Translation Model with Hard Attention
|
Soft-attention based Neural Machine Translation (NMT) models have achieved promising results on several translation tasks. These models attend all the words in the source sequence for each target token, which makes them ineffective for long sequence translation. In this work, we propose a hard-attention based NMT model which selects a subset of source tokens for each target token to effectively handle long sequence translation. Due to the discrete nature of the hard-attention mechanism, we design a reinforcement learning algorithm coupled with reward shaping strategy to efficiently train it. Experimental results show that the proposed model performs better on long sequences and thereby achieves significant BLEU score improvement on English-German (EN-DE) and English-French (ENFR) translation tasks compared to the soft attention based NMT.
|
['Sathish Reddy Indurthi', 'Sangha Kim', 'Insoo Chung']
|
2019-07-01
| null | null | null |
acl-2019-7
|
['hard-attention']
|
['methodology']
|
[ 3.20119470e-01 -8.06995761e-03 -4.52266216e-01 -2.45291308e-01
-1.14055014e+00 -3.47771585e-01 5.81134379e-01 -2.42474630e-01
-4.28750515e-01 1.09509766e+00 3.56995553e-01 -7.23672569e-01
4.41602498e-01 -4.18580860e-01 -8.78073931e-01 -5.10210812e-01
4.98256862e-01 7.64325440e-01 -2.56004840e-01 -4.25979316e-01
3.24809849e-02 1.47057660e-02 -7.11669683e-01 3.26201767e-01
1.31726158e+00 4.10337478e-01 7.52445221e-01 5.34297347e-01
-2.69669473e-01 6.23076677e-01 -6.16898835e-01 -6.62457466e-01
2.13441074e-01 -9.11576033e-01 -8.60469878e-01 -2.74882346e-01
1.69724420e-01 -2.88074046e-01 -3.29376340e-01 1.04846859e+00
7.66208589e-01 1.21417455e-01 6.42294228e-01 -8.29210937e-01
-1.20426083e+00 9.49202299e-01 -5.82814693e-01 4.14162725e-01
2.89220184e-01 1.66680917e-01 1.27504945e+00 -1.34884346e+00
4.44353670e-01 1.21913254e+00 3.27950418e-01 8.55357587e-01
-1.01764619e+00 -4.85828429e-01 7.08347559e-02 2.82847494e-01
-1.06093097e+00 -4.03527647e-01 3.83126438e-01 -1.37856454e-02
1.63500655e+00 2.54552126e-01 4.63125765e-01 1.23979712e+00
7.56692231e-01 1.07168245e+00 9.30691063e-01 -5.99372983e-01
9.36047956e-02 -1.07689731e-01 -1.64849177e-01 2.34278575e-01
-2.63371676e-01 1.08322084e-01 -4.82135564e-01 1.28954602e-02
7.67389774e-01 -9.91902575e-02 -1.02150969e-01 1.56145766e-01
-1.50077474e+00 8.92424643e-01 3.45364273e-01 2.64762729e-01
-7.31027007e-01 2.30345726e-01 4.24237370e-01 6.61585271e-01
5.43740332e-01 5.62576950e-01 -7.04744101e-01 -5.63029408e-01
-6.24338269e-01 2.95316447e-02 3.66596937e-01 1.35959768e+00
5.80612302e-01 3.12077999e-01 -6.80954158e-01 1.10883689e+00
-4.18569818e-02 7.63864338e-01 7.25641847e-01 -3.65777314e-01
8.25829983e-01 2.87232786e-01 1.87339827e-01 -3.37632298e-01
1.20822184e-01 -5.73455036e-01 -8.36195230e-01 -3.31622869e-01
-8.23821127e-03 -3.39241654e-01 -1.06566167e+00 1.75334549e+00
-9.91013497e-02 -1.47712559e-01 2.56901056e-01 9.75116670e-01
3.16751420e-01 1.13285279e+00 3.72544378e-02 -4.86432195e-01
1.05527604e+00 -1.39142358e+00 -1.00004780e+00 -4.84057724e-01
5.15818119e-01 -9.41468656e-01 1.59470856e+00 -1.60480201e-01
-1.33589518e+00 -6.52397692e-01 -7.54770398e-01 2.82984637e-02
-1.43915471e-02 1.34621903e-01 2.89979398e-01 2.49199033e-01
-8.55586112e-01 5.68620622e-01 -7.58537650e-01 -2.97193378e-01
1.77457258e-01 5.66247582e-01 -1.80471782e-02 -2.70975411e-01
-1.53357542e+00 1.15940475e+00 2.60191917e-01 2.76963115e-01
-9.30191159e-01 -2.26994917e-01 -6.59229636e-01 3.12313050e-01
9.56465825e-02 -7.38078833e-01 1.47559917e+00 -1.40146816e+00
-2.02896214e+00 3.79876822e-01 -4.20859843e-01 -5.21355212e-01
4.02158022e-01 -4.64261025e-01 -1.56438112e-01 -4.75057773e-02
7.65361935e-02 8.14014435e-01 9.72182751e-01 -6.47036135e-01
-3.87247801e-01 1.00501545e-01 -1.42271966e-01 4.88181710e-01
-4.20339376e-01 5.19045234e-01 -2.09158868e-01 -8.83336306e-01
-3.85780960e-01 -1.03823936e+00 -3.75134289e-01 -7.94254541e-01
-1.78094953e-01 -4.65602696e-01 5.94696045e-01 -7.41073430e-01
1.19689512e+00 -1.78438175e+00 6.15391135e-01 -3.82887155e-01
-3.11646163e-01 5.05200863e-01 -6.34620845e-01 6.78908288e-01
4.47368510e-02 5.09409085e-02 -2.52769768e-01 -2.25563467e-01
-1.52609115e-02 4.59642678e-01 -3.92363966e-01 4.47694398e-02
6.42120242e-01 1.53390527e+00 -1.03061700e+00 -3.17376852e-01
-3.41107463e-03 2.46861935e-01 -5.55048764e-01 5.79649568e-01
-5.12163341e-01 4.71067578e-01 -4.44958150e-01 5.42631805e-01
3.61906141e-01 -2.03202754e-01 7.28975385e-02 4.06614572e-01
3.62095572e-02 5.97291708e-01 -2.27129340e-01 1.69904387e+00
-6.52438760e-01 5.28974593e-01 -2.50036776e-01 -6.35197580e-01
1.01576257e+00 6.23744547e-01 3.46400589e-02 -8.90561998e-01
2.19434261e-01 4.05135810e-01 5.39025068e-01 -3.95580500e-01
6.16863668e-01 -2.42712557e-01 -4.09147553e-02 5.93793511e-01
9.97935981e-02 5.35699725e-02 4.12738025e-02 -3.65075245e-02
9.52696741e-01 4.63800192e-01 1.38857812e-01 -3.10911834e-01
4.81517583e-01 9.06990841e-03 7.91712046e-01 6.38959050e-01
-2.52052695e-01 6.16319001e-01 1.87066197e-01 -3.52387309e-01
-1.44171512e+00 -8.74885619e-01 3.39934677e-01 1.53689611e+00
-1.45041108e-01 -2.95760512e-01 -9.62197602e-01 -8.72902155e-01
-3.66000801e-01 7.42269278e-01 -4.00950581e-01 -2.79144734e-01
-1.06202352e+00 -7.28874683e-01 2.44932070e-01 7.33738542e-01
2.12583676e-01 -1.56600702e+00 -2.13695750e-01 6.10100925e-01
-6.62821829e-01 -8.76372516e-01 -1.18795872e+00 3.12841862e-01
-7.36135542e-01 -2.80344307e-01 -1.14655757e+00 -1.02793241e+00
5.49770534e-01 1.04841933e-01 1.13930976e+00 -1.44211724e-01
2.68070251e-01 -3.59089524e-01 -6.63177252e-01 -2.85954684e-01
-7.14427531e-01 5.30305803e-01 1.73799798e-01 4.93114479e-02
5.69020331e-01 -2.32254073e-01 -2.71334887e-01 2.76753724e-01
-5.07820964e-01 1.76669538e-01 1.09522533e+00 1.39469993e+00
5.18562555e-01 -7.58418739e-01 1.02012002e+00 -4.14548039e-01
1.00464594e+00 -4.09040809e-01 -3.51569623e-01 4.59219545e-01
-6.25787020e-01 1.20519847e-01 1.08986831e+00 -7.43569493e-01
-8.79515827e-01 -2.22830400e-01 -2.13957071e-01 -5.45989871e-01
2.39048526e-01 6.15558326e-01 -2.00018600e-01 3.71770352e-01
4.62256253e-01 7.14087903e-01 -3.00208610e-02 -4.60014939e-01
1.57821119e-01 1.07561874e+00 2.84651220e-01 -5.65145850e-01
6.13199770e-01 -4.02960449e-01 -4.91833389e-01 -4.26989347e-01
-4.89771873e-01 -1.66168928e-01 -5.59815228e-01 1.63148418e-01
8.09606493e-01 -8.20667803e-01 -2.93866307e-01 2.39685133e-01
-1.24262917e+00 -6.00425839e-01 -6.64317682e-02 7.83145130e-01
-8.73329580e-01 2.39285037e-01 -1.05802572e+00 -6.73910558e-01
-9.21054542e-01 -1.43799531e+00 9.29991663e-01 -4.35441248e-02
-3.61830503e-01 -6.71881735e-01 -2.18515471e-02 2.08899319e-01
6.88836575e-01 -4.24870342e-01 1.12245548e+00 -6.08534575e-01
-6.11468494e-01 -6.57778094e-03 -2.00468421e-01 4.00755107e-01
8.28854367e-02 -3.07158798e-01 -3.97317678e-01 -5.74345410e-01
-1.78146780e-01 -4.04639482e-01 7.51782000e-01 3.77607673e-01
6.42519534e-01 -3.70476872e-01 2.49712151e-02 3.22689116e-01
1.13571250e+00 3.56321603e-01 7.39404082e-01 2.60515392e-01
6.50601268e-01 1.79614812e-01 9.12102997e-01 1.24582425e-01
1.04206383e-01 7.83800662e-01 2.32201219e-01 -2.02813089e-01
-1.19586796e-01 -3.23257655e-01 9.43397760e-01 1.23075581e+00
-5.67438751e-02 -5.97511172e-01 -7.46797025e-01 6.90443277e-01
-2.08367658e+00 -7.23864615e-01 4.97064181e-02 1.97981310e+00
1.15529704e+00 1.84327155e-01 -2.70579401e-02 -3.54069620e-01
7.77950764e-01 -7.04714134e-02 -5.16936064e-01 -1.06101894e+00
-1.26656041e-01 3.84919018e-01 4.02349561e-01 6.15944326e-01
-5.67004323e-01 1.53698587e+00 6.59730387e+00 9.60809350e-01
-1.01064622e+00 3.72321904e-01 4.03882772e-01 -9.78318527e-02
-3.44348401e-01 -7.30813667e-02 -6.55862212e-01 5.07199287e-01
1.30166650e+00 -2.88343966e-01 6.64812744e-01 4.82421875e-01
4.64577854e-01 4.08078730e-01 -1.11545289e+00 5.68705976e-01
-1.32043675e-01 -1.06411719e+00 2.96173722e-01 8.69989172e-02
1.00067842e+00 3.02855819e-01 1.15870483e-01 5.57044685e-01
3.50152344e-01 -1.01682127e+00 6.14921689e-01 1.29819155e-01
1.00332284e+00 -1.01928651e+00 8.87696683e-01 4.96906340e-01
-6.55891240e-01 3.35534513e-02 -8.29098880e-01 -2.21805468e-01
1.13667101e-01 2.89132416e-01 -1.04531717e+00 5.43081045e-01
2.22756982e-01 6.99143231e-01 -8.61168653e-03 6.66730940e-01
-5.04840314e-01 1.00346196e+00 9.82863307e-02 -3.73202205e-01
6.78286552e-01 -2.30029687e-01 5.44626713e-01 1.32241869e+00
4.74280268e-01 -2.61787713e-01 1.76512927e-01 8.64978790e-01
-4.18623239e-01 4.38356519e-01 -3.85260224e-01 -3.38828623e-01
2.99728662e-01 8.82366538e-01 -2.70598292e-01 -4.40709233e-01
-4.87312526e-01 1.37587988e+00 5.82173824e-01 5.60109615e-01
-9.58001018e-01 -5.96280098e-01 7.98362672e-01 -1.37601227e-01
5.20563781e-01 -2.29345217e-01 -1.15122899e-01 -1.20036232e+00
2.44587613e-03 -1.09659684e+00 -1.02086216e-01 -7.56721079e-01
-1.26592684e+00 1.01307023e+00 -3.45204085e-01 -1.29207826e+00
-5.35774708e-01 -2.16148913e-01 -5.22384703e-01 1.36202872e+00
-1.61129451e+00 -1.18162549e+00 5.19761622e-01 3.79502654e-01
1.07809448e+00 -4.40063864e-01 9.28759098e-01 2.17831075e-01
-7.57620275e-01 9.55833316e-01 5.04182577e-01 1.79670215e-01
8.08316588e-01 -1.20402718e+00 1.05311334e+00 8.92945230e-01
1.50398508e-01 7.55143583e-01 5.63356698e-01 -7.88401961e-01
-1.47582853e+00 -1.19276583e+00 1.40752077e+00 -2.79579699e-01
3.84874076e-01 -4.38838750e-01 -8.78015518e-01 7.83671439e-01
9.08127248e-01 -3.49625587e-01 5.25969088e-01 -1.29942540e-02
-1.49845272e-01 2.73682475e-01 -7.32670009e-01 7.92698026e-01
8.14034998e-01 -4.42981988e-01 -6.73800170e-01 4.40204591e-01
1.00133586e+00 -3.35914224e-01 -7.17685699e-01 2.54255295e-01
2.85690993e-01 -3.19657564e-01 5.73547482e-01 -8.05800140e-01
7.46743500e-01 -5.69909438e-02 -5.30204847e-02 -1.83217728e+00
-6.80575371e-01 -1.03887534e+00 -3.19970220e-01 9.06447172e-01
8.26839149e-01 -5.24342000e-01 5.32949924e-01 9.81987342e-02
-5.55571377e-01 -9.41798270e-01 -9.46626723e-01 -9.78255749e-01
4.96436030e-01 7.61201531e-02 7.59012461e-01 7.81506002e-01
2.20998511e-01 7.77183950e-01 -9.36852813e-01 -2.18230218e-01
2.39679441e-01 2.53350884e-01 4.78331745e-01 -5.43318987e-01
-4.86239731e-01 -4.55784708e-01 1.63388461e-01 -1.43684781e+00
3.25795442e-01 -1.08747721e+00 3.25169861e-01 -1.59771299e+00
4.14652079e-01 -9.16006323e-03 -3.98724258e-01 6.31227493e-01
-7.41276801e-01 1.91565305e-01 1.33810386e-01 3.04116637e-01
-4.08340245e-01 9.03828859e-01 1.54227078e+00 -2.34941691e-01
-1.50221467e-01 3.12926085e-03 -4.84290153e-01 -1.49965091e-02
9.23666298e-01 -6.04375660e-01 -1.61871687e-01 -8.40901613e-01
-4.69190106e-02 2.68639445e-01 -3.90930921e-01 -4.61234361e-01
-1.29969358e-01 -5.40641487e-01 3.17921966e-01 -5.59565485e-01
1.09100930e-01 -4.88220811e-01 -4.19115663e-01 5.66972792e-01
-7.05578446e-01 5.37505805e-01 7.28355050e-02 4.13935125e-01
-1.88905835e-01 -1.10901937e-01 7.55527616e-01 -2.20441237e-01
-3.16770285e-01 3.83303255e-01 -5.24984658e-01 3.78540792e-02
7.92502403e-01 8.05333629e-03 -3.43949758e-02 -5.15369415e-01
-5.15371978e-01 3.47431958e-01 3.28522712e-01 7.54143000e-01
6.49487078e-01 -1.55593550e+00 -1.37641203e+00 3.15958858e-01
6.53434023e-02 -3.34736824e-01 -1.26178101e-01 8.44945908e-01
-1.99776381e-01 6.48967147e-01 -4.69692647e-01 -2.96727061e-01
-1.15057003e+00 7.20411360e-01 2.06264690e-01 -3.67114663e-01
-5.12586117e-01 8.63623023e-01 2.46865880e-02 -3.98090512e-01
1.06717214e-01 -1.91412181e-01 1.93981871e-01 -4.85341907e-01
4.68829393e-01 1.10525362e-01 8.49780142e-02 -5.77664495e-01
-1.92623109e-01 2.51864433e-01 -6.01801813e-01 -1.67253092e-01
1.11145055e+00 -2.66684353e-01 -8.94028172e-02 1.46466672e-01
1.16098499e+00 -1.16180718e-01 -1.19804585e+00 -4.22337890e-01
3.43065755e-03 -4.65899855e-01 -1.38550475e-01 -9.89794612e-01
-6.22328460e-01 1.17311299e+00 1.80839568e-01 -2.07979038e-01
9.52430308e-01 -1.71890020e-01 1.29077399e+00 3.42031300e-01
2.24171281e-01 -1.14650667e+00 2.50548702e-02 1.00160873e+00
8.91541302e-01 -1.27099335e+00 -6.43073738e-01 3.32076289e-02
-7.46001661e-01 1.03458405e+00 7.57020533e-01 4.47016731e-02
-7.41250589e-02 1.19419836e-01 1.77705511e-01 4.42189038e-01
-1.16480649e+00 -1.85854018e-01 1.13968410e-01 5.19236803e-01
7.71793306e-01 1.91861078e-01 -5.65265775e-01 2.76459366e-01
-7.15384111e-02 -3.65126170e-02 4.59475309e-01 8.74592960e-01
-5.37335694e-01 -1.59207916e+00 -2.44286865e-01 2.13338196e-01
-7.50766516e-01 -5.74376583e-01 -6.38260186e-01 1.89560398e-01
-3.80455732e-01 9.21799421e-01 -1.09770693e-01 -4.15617138e-01
1.96457773e-01 3.24835330e-01 4.97259825e-01 -7.23304570e-01
-9.99971628e-01 5.01328290e-01 -2.41194963e-02 -1.68261290e-01
-4.07138956e-04 -5.76169074e-01 -1.19028974e+00 -2.60866165e-01
-4.46949184e-01 3.84395063e-01 4.40203369e-01 8.17953408e-01
5.54534018e-01 6.03649914e-01 9.46111083e-01 -6.36206031e-01
-1.11396074e+00 -1.51481938e+00 -7.85049945e-02 2.82547623e-01
3.55611742e-01 -2.17730656e-01 1.05379157e-01 -2.24786058e-01]
|
[11.71902847290039, 10.043811798095703]
|
fcb7fdd4-6be7-4b30-b9da-25a50976cb0e
|
towards-large-scale-simulations-of-open-ended
|
2304.05639
| null |
https://arxiv.org/abs/2304.05639v1
|
https://arxiv.org/pdf/2304.05639v1.pdf
|
Towards Large-Scale Simulations of Open-Ended Evolution in Continuous Cellular Automata
|
Inspired by biological and cultural evolution, there have been many attempts to explore and elucidate the necessary conditions for open-endedness in artificial intelligence and artificial life. Using a continuous cellular automata called Lenia as the base system, we built large-scale evolutionary simulations using parallel computing framework JAX, in order to achieve the goal of never-ending evolution of self-organizing patterns. We report a number of system design choices, including (1) implicit implementation of genetic operators, such as reproduction by pattern self-replication, and selection by differential existential success; (2) localization of genetic information; and (3) algorithms for dynamically maintenance of the localized genotypes and translation to phenotypes. Simulation results tend to go through a phase of diversity and creativity, gradually converge to domination by fast expanding patterns, presumably a optimal solution under the current design. Based on our experimentation, we propose several factors that may further facilitate open-ended evolution, such as virtual environment design, mass conservation, and energy constraints.
|
['Bert Wang-Chak Chan']
|
2023-04-12
| null | null | null | null |
['artificial-life']
|
['miscellaneous']
|
[-1.09354839e-01 -1.06765099e-01 3.95320803e-01 2.19790190e-01
7.42895842e-01 -4.22763258e-01 7.56113291e-01 4.38172817e-02
-2.43569225e-01 1.07976711e+00 -4.19456251e-02 -1.08552173e-01
-2.36937791e-01 -1.05334902e+00 -2.28168771e-01 -9.27203536e-01
-4.74123716e-01 4.85497892e-01 2.84828424e-01 -4.52621967e-01
6.51645362e-01 5.47454655e-01 -2.12921119e+00 -3.11315656e-01
1.12235200e+00 3.83687437e-01 2.34923556e-01 7.04145372e-01
-3.20211262e-01 4.10689324e-01 -5.44077098e-01 -3.32659371e-02
2.11743563e-01 -1.02071321e+00 -5.64142704e-01 5.57931885e-02
-1.06068444e+00 4.72946554e-01 2.40027994e-01 8.30234528e-01
6.27290487e-01 2.46141642e-01 8.06320488e-01 -1.14261401e+00
-1.04199541e+00 4.53113914e-01 -4.74847943e-01 5.89626729e-02
2.35393733e-01 5.43666124e-01 2.67975807e-01 -5.66005647e-01
8.53256226e-01 1.07726038e+00 6.31471157e-01 5.95423222e-01
-9.64055002e-01 -2.04068378e-01 -3.61044854e-01 -1.68259427e-01
-1.96452665e+00 -2.84501642e-01 5.01279771e-01 -2.05679853e-02
1.04415131e+00 7.12278128e-01 1.48914123e+00 8.18165958e-01
7.42359638e-01 1.76557347e-01 1.10197282e+00 -8.79239440e-01
8.39636028e-01 4.45681870e-01 -4.14609641e-01 7.42965400e-01
6.54723942e-01 1.57386765e-01 -2.52866954e-01 -5.71928561e-01
7.65269876e-01 -4.16342556e-01 -5.74433170e-02 -4.24097598e-01
-1.09991860e+00 5.15607059e-01 -1.22037735e-02 7.68383026e-01
-6.26240313e-01 -8.17081612e-03 -7.40927681e-02 2.76451051e-01
-5.10679781e-02 1.16037297e+00 -3.88273805e-01 -3.85287583e-01
-3.73702079e-01 2.72890121e-01 8.97687376e-01 5.27097523e-01
6.61332726e-01 7.58812204e-02 1.24592029e-01 8.02519619e-01
3.18245053e-01 3.03530604e-01 1.03992248e+00 -9.69494939e-01
-5.48641562e-01 9.59072948e-01 2.20459938e-01 -1.01813245e+00
-3.76884699e-01 -6.99979842e-01 -7.27617621e-01 4.36840892e-01
2.93621775e-02 -4.60964978e-01 -6.49687350e-01 1.63671124e+00
4.91193324e-01 -7.93580711e-02 3.20688069e-01 7.23264396e-01
-1.56616513e-02 7.97727525e-01 -5.08081801e-02 -7.37862468e-01
9.71735358e-01 -8.79614115e-01 -7.09980905e-01 4.11131769e-01
5.59952021e-01 -5.06655276e-01 8.83039713e-01 1.10586070e-01
-1.09886169e+00 -3.80550385e-01 -1.07185304e+00 7.29635179e-01
-6.99436486e-01 -2.61003733e-01 7.32409358e-01 1.18427742e+00
-1.32387877e+00 6.36045754e-01 -7.32332945e-01 -1.03547573e+00
-9.66944471e-02 2.02182025e-01 1.38463959e-01 7.71678209e-01
-8.42569888e-01 8.06136072e-01 3.93483728e-01 -1.78597141e-02
-4.29158866e-01 -1.68720394e-01 7.71063427e-03 1.18682630e-01
-1.33551791e-01 -1.20484006e+00 3.33672017e-01 -1.39014041e+00
-1.96428609e+00 7.43087769e-01 5.11535443e-02 -6.68162033e-02
4.58389103e-01 6.40183389e-01 -4.69050139e-01 -3.18560421e-01
-2.35980794e-01 6.68312788e-01 2.14598030e-01 -1.68910050e+00
-4.10973698e-01 -2.45347068e-01 -4.66876864e-01 5.50003529e-01
-5.43260396e-01 3.74835841e-02 6.99353591e-02 -5.13155341e-01
8.26787725e-02 -1.08045840e+00 -5.10412633e-01 -2.48246476e-01
5.99121489e-03 -1.80014044e-01 4.60597277e-01 -1.55969918e-01
1.38693821e+00 -1.97248936e+00 2.97915667e-01 5.14786780e-01
-1.81911662e-02 1.35908052e-01 -5.99771962e-02 7.83456743e-01
3.61121327e-01 3.19200397e-01 -2.03433156e-01 6.77310005e-02
-1.42759889e-01 3.02122802e-01 1.88696846e-01 1.99374005e-01
3.72046344e-02 8.59906316e-01 -7.43893504e-01 -5.94034374e-01
-2.30502695e-01 1.52651563e-01 -5.52738190e-01 5.46411425e-02
-9.86621603e-02 3.01445305e-01 -6.51131272e-01 7.63107598e-01
3.76508862e-01 -2.01167330e-01 2.92287111e-01 5.93786597e-01
-8.00972223e-01 -4.83372718e-01 -1.21463251e+00 1.29153562e+00
-1.12418406e-01 1.75882742e-01 1.42267063e-01 -5.16034424e-01
1.38965344e+00 8.56295303e-02 5.08329511e-01 -6.87066734e-01
5.36997437e-01 4.11522955e-01 3.94185245e-01 -4.94770199e-01
6.89330220e-01 1.97442010e-01 2.44349271e-01 8.00468147e-01
-2.61271030e-01 -2.58161128e-01 2.11857840e-01 -2.91809261e-01
9.63948369e-01 1.23520747e-01 4.24551725e-01 -8.63150477e-01
5.07789850e-01 2.71346122e-01 6.30106926e-01 8.19064915e-01
-3.62625986e-01 3.55720431e-01 8.56373757e-02 -5.76577067e-01
-1.28035855e+00 -8.26305985e-01 -2.64445711e-02 7.36493886e-01
5.63569784e-01 -4.86003049e-02 -9.04396236e-01 3.33766341e-01
-2.94909865e-01 6.65959179e-01 -7.65517890e-01 -2.83068627e-01
-4.03719097e-01 -1.30052578e+00 5.00569701e-01 -2.69523293e-01
5.91909468e-01 -1.38311183e+00 -1.20312679e+00 4.31387752e-01
2.34006345e-01 -1.34742320e-01 8.08413103e-02 2.72704244e-01
-8.78931522e-01 -4.72476691e-01 -6.30798638e-01 -7.35743165e-01
8.19900513e-01 8.66286084e-02 8.38725865e-01 8.18889320e-01
-5.52908719e-01 3.19767818e-02 -4.24251020e-01 -4.13764298e-01
-6.84471905e-01 9.92623791e-02 2.76830405e-01 -1.33065999e-01
-1.30603403e-01 -1.08418560e+00 -6.60905123e-01 5.56362987e-01
-7.94726729e-01 3.53994668e-02 6.10802054e-01 7.53797591e-01
4.24438924e-01 5.22219241e-01 6.39907479e-01 -1.43641710e-01
1.14071035e+00 -5.61185479e-01 -3.26828927e-01 6.79520547e-01
-8.67909014e-01 1.54312253e-01 6.82694912e-01 -6.08421683e-01
-1.07161891e+00 -2.07063019e-01 1.45515189e-01 1.47990296e-02
2.77543836e-03 2.55647659e-01 -1.09838899e-02 -3.31375659e-01
8.25590432e-01 7.90733874e-01 4.19007272e-01 8.19971934e-02
1.70771003e-01 6.99134767e-01 6.06376603e-02 -7.78884709e-01
3.62179369e-01 2.87776798e-01 -3.52918208e-02 -9.25620139e-01
4.82765585e-01 3.64842564e-01 -3.91780466e-01 -4.63979959e-01
7.61680663e-01 -3.01073730e-01 -6.99821651e-01 6.50601625e-01
-6.98767245e-01 -2.14757800e-01 -5.61502397e-01 1.63061902e-01
-4.22211111e-01 1.73784524e-01 -1.78711832e-01 -1.28445339e+00
-4.08430785e-01 -6.85004711e-01 2.74991035e-01 8.24017704e-01
-3.91928434e-01 -7.79489577e-01 6.39861166e-01 -2.75099814e-01
8.81071687e-01 2.12151378e-01 1.03079677e+00 -2.40778327e-01
-3.72862816e-01 3.35698962e-01 3.83567423e-01 -4.60564345e-01
1.59625247e-01 6.43356502e-01 -3.95767063e-01 -1.26782551e-01
-2.52524894e-02 3.62767503e-02 -9.44021437e-03 2.42670655e-01
5.04089117e-01 -1.92804411e-01 -6.59535527e-01 6.51281118e-01
1.34738803e+00 1.18177962e+00 8.51145625e-01 7.44754612e-01
-1.27230391e-01 6.82020366e-01 2.42267832e-01 7.06606746e-01
2.21778572e-01 3.92013311e-01 1.96473539e-01 5.03907874e-02
1.22791618e-01 -1.61563754e-01 -1.28857195e-01 1.02907932e+00
-5.30521691e-01 -4.57126617e-01 -9.05862451e-01 5.83911598e-01
-1.92090511e+00 -9.77443099e-01 1.86274827e-01 2.08362794e+00
5.94781697e-01 -9.84071717e-02 3.08091670e-01 -8.59152153e-02
9.07678425e-01 -3.37081134e-01 -4.48390514e-01 -7.70176768e-01
-5.74222505e-01 -1.50057152e-01 2.74998218e-01 1.66334897e-01
-3.73780608e-01 7.69537807e-01 7.35475111e+00 5.25784910e-01
-1.27653456e+00 1.15688415e-02 7.92510748e-01 1.44043926e-03
-6.48897529e-01 2.57754922e-01 -1.94785953e-01 7.88141191e-01
8.22283149e-01 -6.29575729e-01 8.03885281e-01 3.62721741e-01
3.44738156e-01 -2.13330656e-01 -5.20554930e-02 5.72874010e-01
-1.55953050e-01 -1.42874360e+00 4.55893464e-02 2.44581640e-01
1.09592772e+00 -4.36154217e-01 -2.67871529e-01 2.44009998e-02
5.20275772e-01 -9.05328810e-01 8.74940813e-01 6.67969823e-01
1.66390285e-01 -8.30822468e-01 4.68325019e-01 5.10632455e-01
-9.01365221e-01 -3.21255118e-01 -4.59197849e-01 -3.82757038e-01
2.29286864e-01 6.08313717e-02 -4.63131905e-01 4.83163387e-01
6.18377924e-01 -1.59847379e-01 -5.12018979e-01 1.20729828e+00
3.04083318e-01 2.84452200e-01 -5.30269444e-01 -9.95001674e-01
8.50765929e-02 -7.31285691e-01 7.78433919e-01 7.22382665e-01
8.14615488e-01 3.74724776e-01 -2.57284552e-01 1.15758920e+00
4.57456678e-01 2.91810870e-01 -6.09512508e-01 -1.69777781e-01
8.16119075e-01 9.39812839e-01 -1.53119600e+00 -5.32845147e-02
3.36192042e-01 7.70431757e-01 -1.38576934e-02 1.84841707e-01
-9.19808209e-01 -5.03505290e-01 5.58078170e-01 1.06988505e-01
2.87899394e-02 -4.36144173e-01 -7.89887547e-01 -7.61171877e-01
-4.03150946e-01 -7.42633998e-01 -9.08611864e-02 -8.22312593e-01
-6.85637057e-01 9.03290570e-01 -3.30870271e-01 -6.75441742e-01
-1.48397446e-01 2.89937779e-02 -9.46045995e-01 6.65253222e-01
-6.03211880e-01 -7.22463250e-01 -1.93452150e-01 3.20845783e-01
2.08525464e-01 -7.14541495e-01 8.22304070e-01 -5.08975610e-02
-6.68733001e-01 3.82758141e-01 4.66090977e-01 -7.22691476e-01
3.24006826e-02 -7.37421393e-01 2.01148704e-01 7.85778105e-01
-1.08611062e-01 7.72648692e-01 9.95068908e-01 -8.55358303e-01
-1.67859888e+00 -5.26562095e-01 6.09950542e-01 -5.00681549e-02
3.95084649e-01 -2.79265136e-01 -7.35236645e-01 -5.10787927e-02
4.06286836e-01 -6.50880516e-01 3.83413613e-01 -6.85583577e-02
5.54786742e-01 1.84037745e-01 -1.28612566e+00 1.15126324e+00
1.37900221e+00 3.33365291e-01 -1.81763127e-01 -2.89603546e-02
4.91685718e-01 1.50632843e-01 -5.18200278e-01 2.32367530e-01
8.43823075e-01 -1.01453972e+00 6.28401935e-01 -1.37618795e-01
7.15237260e-02 -5.76888144e-01 1.63913190e-01 -1.21733415e+00
-8.80614400e-01 -1.02907681e+00 5.15744984e-01 1.17756617e+00
5.70087671e-01 -1.13968146e+00 4.85487550e-01 3.70878041e-01
-1.35979265e-01 -8.46195161e-01 -8.39773715e-01 -8.34939659e-01
1.83859039e-02 6.07461035e-01 9.62807834e-01 1.05843401e+00
2.37065461e-02 -9.57156718e-02 -1.65624082e-01 -6.49954975e-02
2.45305255e-01 1.12085812e-01 6.16184413e-01 -1.03618169e+00
-3.31937492e-01 -8.21939170e-01 -3.92825425e-01 -2.42237240e-01
-2.77635813e-01 -2.53446281e-01 3.57225947e-02 -1.18513930e+00
1.94277555e-01 -8.16651285e-01 -6.25707805e-02 1.42822221e-01
1.16797179e-01 1.77396163e-01 -1.03661805e-01 5.36560953e-01
-4.18119341e-01 6.99510396e-01 1.23493135e+00 3.92033070e-01
-8.06884468e-01 -4.65151757e-01 -8.59167457e-01 3.65373880e-01
8.25931847e-01 -3.08023959e-01 -4.92822349e-01 -1.23603947e-01
4.34380412e-01 3.45482007e-02 -2.23081365e-01 -1.17621291e+00
2.89403141e-01 -7.04546690e-01 2.21875131e-01 8.23260844e-02
9.91884023e-02 -5.93229413e-01 1.17741644e+00 1.04005146e+00
-5.16674221e-02 5.07483423e-01 2.19615344e-02 3.86875838e-01
1.28472462e-01 -3.77351493e-01 7.26868808e-01 -1.33652046e-01
-5.45877159e-01 -3.79447758e-01 -9.37745988e-01 -4.43405986e-01
1.64217389e+00 -1.00795162e+00 -3.11531425e-01 -6.14223108e-02
-5.64776838e-01 1.66815504e-01 1.22235918e+00 4.39250499e-01
1.46542594e-01 -1.00810719e+00 -6.07055902e-01 3.41419905e-01
-2.09477007e-01 -6.68672740e-01 6.79330155e-02 4.42232221e-01
-1.18530297e+00 6.51159883e-02 -9.40689266e-01 -3.03889781e-01
-1.05556667e+00 4.29825068e-01 5.81263483e-01 1.38461873e-01
-3.46305847e-01 6.52838588e-01 -2.22134978e-01 -3.24799895e-01
-2.78594881e-01 1.47538573e-01 -2.88888633e-01 -3.26296151e-01
1.70140386e-01 4.15202439e-01 -1.27609983e-01 -3.18432808e-01
-4.15163815e-01 5.03070116e-01 6.19629145e-01 -3.77213180e-01
1.28271055e+00 -4.35722172e-01 -5.61897337e-01 3.14522833e-01
3.80996346e-01 3.43398340e-02 -7.93424666e-01 6.62731767e-01
-1.80401191e-01 -4.86351162e-01 -3.84444356e-01 -9.70217347e-01
-5.82633257e-01 -5.03036492e-02 6.19072497e-01 6.65533900e-01
1.24771547e+00 -3.66851240e-01 2.95902014e-01 3.37517381e-01
8.63590837e-01 -1.21661448e+00 -1.20653078e-01 5.01510382e-01
6.66993737e-01 -4.12541747e-01 -1.62237659e-01 -3.03940778e-03
-5.09143591e-01 9.88400400e-01 6.22474074e-01 -1.59861609e-01
4.20358628e-01 5.16032219e-01 -8.79895240e-02 -1.58524290e-01
-1.33580244e+00 7.89887160e-02 -4.93350595e-01 7.26169825e-01
2.87570596e-01 2.34197285e-02 -1.11765409e+00 2.81431437e-01
-2.90151477e-01 8.41129795e-02 6.00180387e-01 1.23738873e+00
-7.31043160e-01 -1.06490982e+00 -5.71889102e-01 2.84327455e-02
-1.34607583e-01 3.17383796e-01 -7.03428626e-01 7.53943384e-01
4.69196558e-01 7.42990732e-01 2.21483797e-01 -2.47121900e-01
-1.02123849e-01 -5.62667847e-02 4.08705175e-01 -1.76597771e-03
-9.60797727e-01 -2.78758705e-01 -3.92501466e-02 6.48211688e-02
-1.62674949e-01 -8.27613831e-01 -1.19387507e+00 -5.46970606e-01
-3.03784490e-01 7.50032365e-01 8.06565106e-01 5.81478834e-01
8.77992690e-01 4.19418186e-01 5.68871975e-01 -5.79775095e-01
-1.08112648e-01 -7.11072206e-01 -6.59651637e-01 1.77951396e-01
-5.45327961e-01 -6.43678963e-01 -2.95273393e-01 -8.46949592e-02]
|
[5.600257873535156, 4.104466915130615]
|
4de0a5e9-26d3-4ad3-b42d-d459f76cba5e
|
max-pooling-with-vision-transformers
|
2210.17400
| null |
https://arxiv.org/abs/2210.17400v1
|
https://arxiv.org/pdf/2210.17400v1.pdf
|
Max Pooling with Vision Transformers reconciles class and shape in weakly supervised semantic segmentation
|
Weakly Supervised Semantic Segmentation (WSSS) research has explored many directions to improve the typical pipeline CNN plus class activation maps (CAM) plus refinements, given the image-class label as the only supervision. Though the gap with the fully supervised methods is reduced, further abating the spread seems unlikely within this framework. On the other hand, WSSS methods based on Vision Transformers (ViT) have not yet explored valid alternatives to CAM. ViT features have been shown to retain a scene layout, and object boundaries in self-supervised learning. To confirm these findings, we prove that the advantages of transformers in self-supervised methods are further strengthened by Global Max Pooling (GMP), which can leverage patch features to negotiate pixel-label probability with class probability. This work proposes a new WSSS method dubbed ViT-PCM (ViT Patch-Class Mapping), not based on CAM. The end-to-end presented network learns with a single optimization process, refined shape and proper localization for segmentation masks. Our model outperforms the state-of-the-art on baseline pseudo-masks (BPM), where we achieve $69.3\%$ mIoU on PascalVOC 2012 $val$ set. We show that our approach has the least set of parameters, though obtaining higher accuracy than all other approaches. In a sentence, quantitative and qualitative results of our method reveal that ViT-PCM is an excellent alternative to CNN-CAM based architectures.
|
['Fiora Pirri', 'Marco Schaerf', 'Marta Sanzari', 'Damiano Zappia', 'Simone Rossetti']
|
2022-10-31
| null | null | null | null |
['weakly-supervised-object-detection', 'weakly-supervised-object-localization']
|
['computer-vision', 'computer-vision']
|
[ 5.70928514e-01 6.01348877e-01 -1.19670495e-01 -6.31093919e-01
-8.38598371e-01 -5.63690484e-01 5.54273069e-01 -1.28889769e-01
-7.28688717e-01 5.62149405e-01 -2.50950634e-01 -1.30740225e-01
1.20312601e-01 -6.09636545e-01 -1.10255206e+00 -7.93965876e-01
2.81545013e-01 4.03767735e-01 7.42570698e-01 -2.22301155e-01
1.62937462e-01 1.78034425e-01 -1.59887838e+00 5.73776722e-01
7.21394897e-01 1.12959313e+00 3.52950215e-01 4.10886317e-01
-2.46401533e-01 6.19616866e-01 -5.16375303e-01 -6.86008334e-01
4.49751318e-01 -3.48702937e-01 -1.11839044e+00 3.87159884e-02
9.03811693e-01 5.10859750e-02 2.27649525e-01 1.02354038e+00
3.75325352e-01 -3.47648785e-02 4.35553223e-01 -9.66827095e-01
-3.91912133e-01 8.62279058e-01 -6.77420557e-01 -2.47016549e-02
-5.07679433e-02 3.82292569e-01 1.13310695e+00 -7.87239492e-01
7.13380456e-01 1.22067714e+00 9.80218589e-01 7.07551956e-01
-1.28903949e+00 -5.06362319e-01 3.78526330e-01 2.21944466e-01
-1.28541160e+00 -1.24898590e-01 6.08544588e-01 -1.19934365e-01
1.04520166e+00 2.81183749e-01 7.76384473e-01 1.02503848e+00
-1.95724413e-01 1.12938976e+00 1.57631111e+00 -5.77463031e-01
2.33543396e-01 4.92910087e-01 1.80026427e-01 8.67111564e-01
7.29778334e-02 1.38565535e-02 -7.00772583e-01 3.83996785e-01
3.94982666e-01 -3.45194519e-01 -6.50993511e-02 -4.01523978e-01
-1.03822970e+00 6.01478398e-01 6.35083139e-01 4.00380999e-01
3.98134440e-02 4.05236721e-01 1.82153061e-01 1.12128502e-03
6.29945278e-01 4.74211395e-01 -7.19867826e-01 3.59951518e-02
-1.47274661e+00 2.57489592e-01 6.53685391e-01 1.01697457e+00
1.00937164e+00 -6.96165934e-02 -4.63980943e-01 7.51567900e-01
3.08308423e-01 3.71220946e-01 1.23621792e-01 -1.09088600e+00
2.48190001e-01 7.13596344e-01 -1.54283911e-01 -5.43631017e-01
-3.66833895e-01 -7.67193615e-01 -3.92610520e-01 3.98869544e-01
5.89535236e-01 1.14780940e-01 -1.39583325e+00 1.68396044e+00
1.93670839e-01 1.81032911e-01 -1.26915082e-01 8.27675462e-01
6.91098809e-01 4.52177286e-01 3.22221905e-01 2.13535815e-01
1.35455191e+00 -1.24013627e+00 -5.02496004e-01 -5.03992200e-01
5.67957342e-01 -5.67807019e-01 1.28003538e+00 5.33624470e-01
-1.24566066e+00 -7.33448863e-01 -1.12103701e+00 -1.08649433e-01
-6.85254335e-01 6.53975010e-02 6.85718715e-01 1.09429467e+00
-1.55393207e+00 8.31623375e-01 -9.27448452e-01 -4.50193286e-01
1.17176747e+00 5.56673288e-01 -5.04332446e-02 1.01730332e-01
-8.93773615e-01 9.05363441e-01 3.00589412e-01 1.97342798e-01
-9.30768132e-01 -7.60532320e-01 -8.86521995e-01 -8.43558274e-03
3.74234289e-01 -5.28961718e-01 1.08525741e+00 -1.27755356e+00
-1.63887072e+00 1.32921767e+00 -1.84105262e-01 -9.04570520e-01
6.37626886e-01 -1.34911224e-01 2.15111136e-01 2.74836093e-01
2.19472721e-01 1.47682726e+00 8.91610146e-01 -1.54235387e+00
-8.08486998e-01 -1.63327813e-01 1.00043386e-01 1.00038685e-01
-2.90886372e-01 -4.04580422e-02 -6.28250301e-01 -4.05949801e-01
1.19635038e-01 -6.52879596e-01 -3.69630963e-01 -4.19008881e-02
-5.41985929e-01 -3.41492981e-01 7.21430063e-01 -3.94121200e-01
8.34716201e-01 -1.97893608e+00 -5.49797304e-02 1.06526174e-01
-6.40566051e-02 5.07552087e-01 -1.42960131e-01 -2.46012434e-02
-3.38800028e-02 3.95907193e-01 -9.19574916e-01 -8.88712764e-01
2.05910787e-01 2.11597845e-01 -7.68763795e-02 4.55302507e-01
5.62807262e-01 1.19796872e+00 -6.80846095e-01 -6.70101643e-01
3.61933947e-01 3.06651026e-01 -5.37599862e-01 -3.45927547e-03
-4.85265136e-01 2.79968739e-01 -5.58790565e-02 7.01732039e-01
8.85497868e-01 -9.49056819e-02 -1.26198635e-01 -2.02380762e-01
-1.67559013e-01 9.91172623e-03 -1.05948853e+00 2.03541780e+00
-3.47782150e-02 6.22413993e-01 1.40225008e-01 -1.34672308e+00
6.91829503e-01 4.27782498e-02 1.72751456e-01 -8.82151484e-01
1.97327077e-01 3.49942654e-01 -2.06729859e-01 -3.63456666e-01
3.32041472e-01 -5.02879694e-02 1.47722945e-01 -6.76054507e-02
6.02883935e-01 -5.36938369e-01 3.39494199e-01 1.64769769e-01
8.92983973e-01 7.13795781e-01 -2.54334420e-01 -6.43347144e-01
4.46311980e-01 2.95587778e-01 4.68120456e-01 1.02558112e+00
-4.03470576e-01 1.06604517e+00 5.96759856e-01 -4.92264442e-02
-9.25657213e-01 -1.03064394e+00 -3.23205829e-01 9.77182567e-01
2.48516530e-01 -1.43904224e-01 -1.45534360e+00 -1.02801442e+00
-2.68231362e-01 6.31241798e-01 -7.47529626e-01 1.35048553e-01
-5.63630939e-01 -9.47190106e-01 8.54227364e-01 5.59744716e-01
7.02226102e-01 -1.27074289e+00 -6.65350139e-01 6.45011291e-02
9.49409604e-02 -1.30415750e+00 -1.33952662e-01 5.20548582e-01
-6.81689560e-01 -1.10151017e+00 -9.32180762e-01 -1.03222752e+00
6.81887031e-01 -2.87857316e-02 9.85490680e-01 8.65049511e-02
-3.54469001e-01 4.21024680e-01 -4.01587576e-01 -4.34113801e-01
-1.64854050e-01 2.49912694e-01 -5.21476328e-01 4.99462001e-02
2.83983439e-01 -3.80387068e-01 -8.74885976e-01 2.70831257e-01
-9.13233876e-01 2.46023685e-01 6.77324414e-01 8.52293551e-01
7.93706357e-01 -2.12240160e-01 4.08922613e-01 -1.26542747e+00
5.19927219e-02 -1.11454867e-01 -4.10089701e-01 2.00860903e-01
-8.80048931e-01 -1.46883428e-01 3.50543201e-01 -6.69048131e-02
-1.23210537e+00 3.87073576e-01 -4.69032258e-01 -2.07516342e-01
-5.60590148e-01 1.32336766e-01 -1.92646682e-01 -2.12286249e-01
7.20612168e-01 1.01691231e-01 -1.02605812e-01 -3.50800455e-01
5.92602313e-01 4.20512408e-01 5.46403825e-01 -5.61936319e-01
6.40557110e-01 8.42567444e-01 -2.36910701e-01 -7.69633949e-01
-1.15074265e+00 -6.36975229e-01 -8.20831180e-01 -6.84663653e-02
1.24412119e+00 -9.23098624e-01 -4.35144931e-01 6.17528856e-01
-1.02127171e+00 -7.82293558e-01 -7.14920044e-01 -6.96456898e-03
-6.87573969e-01 2.60472387e-01 -6.77183270e-01 -8.03528488e-01
-1.73259690e-01 -1.24929678e+00 1.25367534e+00 4.18669999e-01
-7.89420009e-02 -8.70522320e-01 -3.04089993e-01 8.13560784e-01
4.59568232e-01 3.45631897e-01 4.96373594e-01 -6.28244936e-01
-6.99005723e-01 5.34030609e-02 -5.80735147e-01 6.75370097e-01
-2.79453516e-01 -1.89042851e-01 -1.49966049e+00 -1.62233844e-01
-7.53461942e-02 -2.84831047e-01 1.33367288e+00 6.04429841e-01
1.29911447e+00 4.17230241e-02 -2.48326913e-01 7.98814058e-01
1.56915641e+00 -4.19091806e-02 8.05685163e-01 4.64443922e-01
8.18896949e-01 8.32783341e-01 4.73937571e-01 -1.78490609e-01
4.21035260e-01 4.56028759e-01 5.99349082e-01 -6.77350223e-01
-6.08179212e-01 -1.06479146e-01 3.70414019e-01 2.58982778e-01
-5.45792580e-02 -1.88865095e-01 -8.04780245e-01 7.38317192e-01
-1.78505325e+00 -6.14316761e-01 -3.65374506e-01 1.75158858e+00
9.74320352e-01 5.27887166e-01 -1.62642412e-02 1.83280841e-01
4.55226898e-01 2.51361549e-01 -3.32892299e-01 -4.24326092e-01
-5.80822945e-01 9.79210198e-01 8.34309220e-01 4.66857404e-01
-1.27682400e+00 1.54329383e+00 5.81446314e+00 1.18298817e+00
-8.75021398e-01 4.14110035e-01 9.89680529e-01 5.73465824e-02
-2.01683670e-01 1.54865712e-01 -9.66180742e-01 2.65325129e-01
7.53372490e-01 7.46946633e-01 1.87203199e-01 7.28815794e-01
8.09290558e-02 -4.39482361e-01 -1.02794909e+00 7.20332444e-01
2.65753627e-01 -1.56421232e+00 -4.73045819e-02 -2.17031270e-01
9.23609972e-01 1.62286162e-01 1.94659337e-01 3.64158452e-01
2.16799080e-01 -1.10208762e+00 1.26152480e+00 2.37455472e-01
7.78021336e-01 -3.95443618e-01 8.99879813e-01 5.54765612e-02
-1.00009322e+00 8.72332007e-02 -1.72716141e-01 1.63270950e-01
7.42120668e-02 3.87139946e-01 -7.06529200e-01 5.06119192e-01
1.16317844e+00 6.76608741e-01 -9.15641785e-01 9.31191266e-01
-3.71590614e-01 9.83572662e-01 -2.81278849e-01 6.43599182e-02
5.73088706e-01 3.32204588e-02 3.43270779e-01 1.69600046e+00
-1.13587588e-01 -2.08683923e-01 -6.87109753e-02 1.30836487e+00
9.99699607e-02 1.38400001e-02 -1.42016575e-01 1.75590858e-01
-1.17071480e-01 1.25437677e+00 -1.40664232e+00 -2.90153205e-01
-2.04199851e-01 1.15938687e+00 2.50624508e-01 4.13898855e-01
-7.28356242e-01 -1.09828226e-01 1.54424295e-01 1.92567095e-01
5.82768798e-01 4.81369197e-02 -8.39758575e-01 -7.73248971e-01
4.78762090e-02 -6.67172611e-01 2.33143404e-01 -6.71604514e-01
-1.20669425e+00 5.08366823e-01 8.19033980e-02 -8.22842538e-01
3.30787480e-01 -9.40458834e-01 -4.44521695e-01 7.08730161e-01
-1.97000897e+00 -1.58358896e+00 -7.33111054e-02 5.37532985e-01
7.63692796e-01 1.41371772e-01 6.08246088e-01 2.23776549e-01
-4.93503571e-01 7.09210694e-01 -3.40069801e-01 3.36962752e-02
5.83180785e-01 -1.64865291e+00 2.54026860e-01 1.07232094e+00
3.42438817e-01 3.71260613e-01 5.78871012e-01 -4.60686058e-01
-8.62452805e-01 -1.02115273e+00 5.97348690e-01 -5.66117167e-01
3.43061358e-01 -7.03980029e-01 -7.92894602e-01 4.19017136e-01
6.19863093e-01 1.08551361e-01 3.02859217e-01 -1.16826251e-01
-2.19745964e-01 -1.30422860e-01 -1.17606091e+00 4.92988497e-01
1.24262857e+00 -3.08846623e-01 -4.28155631e-01 2.40997568e-01
9.42137241e-01 -5.19351125e-01 -4.13006335e-01 5.80966830e-01
2.78768629e-01 -1.29641950e+00 1.04010439e+00 -1.03702687e-01
4.10286039e-01 -4.05286521e-01 -1.51906803e-01 -9.66507196e-01
6.27998710e-02 -4.52376068e-01 3.35960567e-01 1.46031618e+00
8.03245962e-01 -5.47665060e-01 1.17636847e+00 4.76181537e-01
-5.95968187e-01 -9.94078755e-01 -1.08266366e+00 -6.59028888e-01
2.29966760e-01 -7.76457965e-01 3.22535068e-01 8.01331520e-01
-3.50606740e-01 -3.78109254e-02 6.12002946e-02 2.31875908e-02
5.76513350e-01 -2.13699058e-01 4.99597728e-01 -9.60693955e-01
-3.22449803e-01 -8.06192815e-01 -2.17197657e-01 -9.48634863e-01
1.53634891e-01 -1.01861918e+00 3.91460925e-01 -1.70557034e+00
1.34311825e-01 -6.69097722e-01 -3.79317939e-01 6.98959887e-01
-1.98271632e-01 7.36426294e-01 2.36248985e-01 -1.38340533e-01
-8.32707942e-01 3.99720281e-01 1.29650915e+00 -3.49934101e-01
-4.83074188e-02 -1.03630781e-01 -6.70865357e-01 8.10529411e-01
7.65863180e-01 -4.92480993e-01 -2.21885189e-01 -4.52408731e-01
2.22279057e-02 -6.24353766e-01 6.19364858e-01 -1.08073294e+00
2.57907242e-01 1.40235156e-01 3.26904178e-01 -5.28671443e-01
3.11529249e-01 -6.56609118e-01 -3.75008553e-01 4.30220038e-01
-2.51573861e-01 -5.57541490e-01 3.70882303e-01 4.85588551e-01
-1.93633437e-01 -2.93598741e-01 8.75642002e-01 -2.97795057e-01
-9.00869191e-01 1.91365123e-01 -1.09686226e-01 1.16007723e-01
8.46542656e-01 -7.36152112e-01 -2.73661971e-01 2.38643005e-01
-7.59857416e-01 3.14819336e-01 3.56032044e-01 2.23957285e-01
4.40053701e-01 -6.39520288e-01 -5.86960375e-01 1.03237107e-01
-6.28388301e-03 5.40895343e-01 2.58174688e-01 9.72067118e-01
-6.73370063e-01 2.72371233e-01 6.19755778e-03 -9.65031207e-01
-9.52728391e-01 2.12818295e-01 4.93312359e-01 -2.81605303e-01
-5.64318955e-01 1.45168602e+00 1.95098057e-01 -5.09907842e-01
4.43863839e-01 -4.39290553e-01 -2.15954468e-01 7.41828755e-02
7.54118040e-02 -6.41106116e-03 2.37577692e-01 -4.94447470e-01
-2.67609268e-01 6.97327733e-01 -1.59221925e-02 -2.87216872e-01
1.37340021e+00 -1.21862948e-01 -9.24884975e-02 2.89398402e-01
8.98842633e-01 -1.90748379e-01 -1.68719614e+00 -8.17244127e-02
1.74258709e-01 -1.79954246e-01 -5.02091534e-02 -1.13906455e+00
-1.25995183e+00 1.02121222e+00 9.39020991e-01 7.28779137e-02
9.88274395e-01 2.30281979e-01 7.32321918e-01 -2.51220614e-02
3.61685485e-01 -1.25711906e+00 3.78477909e-02 3.14711839e-01
5.20982862e-01 -1.36646426e+00 -2.41955280e-01 -7.11133540e-01
-6.63096309e-01 7.59956837e-01 7.57251740e-01 -2.93057472e-01
6.42373621e-01 2.16621250e-01 1.81993157e-01 -3.20555925e-01
-2.71257460e-01 -7.14239180e-01 3.86885613e-01 7.70609081e-01
3.05912614e-01 -9.75282677e-03 -3.73748481e-01 7.36120999e-01
-3.23804647e-01 -3.46390247e-01 1.89150766e-01 7.78907359e-01
-3.67457569e-01 -1.19767570e+00 -1.56482920e-01 2.22469419e-01
-6.47664964e-01 -3.19641888e-01 -3.47640485e-01 8.14357877e-01
7.87258387e-01 1.07133698e+00 3.34121324e-02 -2.17547759e-01
3.77057821e-01 1.40822038e-01 5.69114566e-01 -8.64257693e-01
-1.02204072e+00 5.63942008e-02 -9.35142562e-02 -7.53209412e-01
-8.11618626e-01 -6.43581271e-01 -1.43246353e+00 2.25320593e-01
-3.58588070e-01 -1.60992295e-01 7.04777062e-01 1.10698223e+00
6.14866614e-02 7.01001108e-01 1.87028006e-01 -1.03088450e+00
-9.66063067e-02 -7.90863693e-01 -3.73333544e-01 5.20172536e-01
1.17499940e-01 -4.95260268e-01 -3.59080404e-01 3.25901777e-01]
|
[9.482600212097168, 0.41934022307395935]
|
4a6b0375-f9bf-4786-9f99-3746ea315ae9
|
contrastive-model-adaptation-for-cross
|
2303.05194
| null |
https://arxiv.org/abs/2303.05194v1
|
https://arxiv.org/pdf/2303.05194v1.pdf
|
Contrastive Model Adaptation for Cross-Condition Robustness in Semantic Segmentation
|
Standard unsupervised domain adaptation methods adapt models from a source to a target domain using labeled source data and unlabeled target data jointly. In model adaptation, on the other hand, access to the labeled source data is prohibited, i.e., only the source-trained model and unlabeled target data are available. We investigate normal-to-adverse condition model adaptation for semantic segmentation, whereby image-level correspondences are available in the target domain. The target set consists of unlabeled pairs of adverse- and normal-condition street images taken at GPS-matched locations. Our method -- CMA -- leverages such image pairs to learn condition-invariant features via contrastive learning. In particular, CMA encourages features in the embedding space to be grouped according to their condition-invariant semantic content and not according to the condition under which respective inputs are captured. To obtain accurate cross-domain semantic correspondences, we warp the normal image to the viewpoint of the adverse image and leverage warp-confidence scores to create robust, aggregated features. With this approach, we achieve state-of-the-art semantic segmentation performance for model adaptation on several normal-to-adverse adaptation benchmarks, such as ACDC and Dark Zurich. We also evaluate CMA on a newly procured adverse-condition generalization benchmark and report favorable results compared to standard unsupervised domain adaptation methods, despite the comparative handicap of CMA due to source data inaccessibility. Code is available at https://github.com/brdav/cma.
|
['Luc van Gool', 'Tim Brödermann', 'Christos Sakaridis', 'David Bruggemann']
|
2023-03-09
| null | null | null | null |
['source-free-domain-adaptation']
|
['computer-vision']
|
[ 5.14011264e-01 -1.00784093e-01 -5.41432083e-01 -6.99695647e-01
-1.15260422e+00 -9.16413426e-01 4.84274328e-01 -2.12730803e-02
-3.37810457e-01 4.63097662e-01 6.19231761e-02 -7.86403269e-02
6.30544052e-02 -6.59780741e-01 -8.24923217e-01 -7.53674388e-01
3.78497481e-01 7.02298343e-01 2.00539410e-01 -4.80969511e-02
5.62308989e-02 5.65232158e-01 -1.04616523e+00 2.18518376e-02
1.00586987e+00 9.08269882e-01 2.35485077e-01 5.42557240e-01
1.06515475e-01 -7.59260580e-02 -3.95421624e-01 -7.86262453e-02
5.81434786e-01 -3.98544222e-01 -7.23544359e-01 4.33625847e-01
5.71701348e-01 -1.14437304e-01 -1.46740183e-01 9.94119823e-01
3.93417120e-01 2.95354694e-01 9.83914256e-01 -1.39184785e+00
-6.75066054e-01 -2.24826172e-01 -7.12183893e-01 1.74643040e-01
1.55619547e-01 3.90045822e-01 8.02946091e-01 -7.31930614e-01
9.64169681e-01 8.54820132e-01 4.41802680e-01 6.51297152e-01
-1.53886986e+00 -9.12048817e-01 3.93339515e-01 1.45023838e-01
-1.26974452e+00 -3.45206082e-01 8.73607874e-01 -5.85653305e-01
7.32297182e-01 -6.49957806e-02 4.15208966e-01 1.38718331e+00
-1.70886852e-02 6.16353989e-01 1.18591368e+00 -3.08785200e-01
5.24141550e-01 1.67724609e-01 -3.06945462e-02 3.38134468e-01
7.35690398e-03 2.80854344e-01 -4.81458783e-01 9.02803838e-02
6.73804939e-01 -1.92901492e-01 -1.62959881e-02 -8.33655238e-01
-1.08888268e+00 7.86625445e-01 5.34818470e-01 4.81923036e-02
-3.27176332e-01 -3.21901649e-01 2.76904315e-01 2.20950484e-01
6.68403089e-01 4.35227871e-01 -8.42919707e-01 4.78604771e-02
-8.98960829e-01 1.57170892e-01 4.85749692e-01 1.17324769e+00
9.80041504e-01 2.99922954e-02 -1.30865108e-02 9.74567294e-01
1.76574439e-01 8.06946576e-01 4.35440212e-01 -1.10645270e+00
6.83292747e-01 3.33543807e-01 2.61430610e-02 -6.93829596e-01
-2.66880959e-01 -5.27885318e-01 -5.10753512e-01 1.78937376e-01
5.15203476e-01 -1.05479181e-01 -1.49463725e+00 1.93550897e+00
5.27176380e-01 4.25004154e-01 1.96753591e-01 8.55239093e-01
3.41581166e-01 5.22134364e-01 3.32324594e-01 9.24056023e-02
9.38318610e-01 -9.39857960e-01 -2.66765296e-01 -6.37582302e-01
5.99238694e-01 -6.78113043e-01 1.35937619e+00 3.19511443e-02
-6.62253141e-01 -5.50538421e-01 -9.89730060e-01 1.94585800e-01
-7.06705749e-01 -7.63296261e-02 1.03700779e-01 5.73218584e-01
-8.28667700e-01 3.37524503e-01 -8.57491851e-01 -6.49492741e-01
6.08019471e-01 2.28494465e-01 -5.16960263e-01 -3.47974330e-01
-1.03374219e+00 6.82769299e-01 4.65755224e-01 -4.22705650e-01
-1.06906581e+00 -9.51680779e-01 -1.01578975e+00 -2.39515454e-01
3.37804437e-01 -5.49409449e-01 1.19161141e+00 -1.11113715e+00
-1.30360329e+00 1.11409307e+00 -2.75167882e-01 -4.19007093e-01
3.18509251e-01 -1.70418650e-01 -7.38439083e-01 3.65364969e-01
5.31942964e-01 9.79910493e-01 1.05725551e+00 -1.55574608e+00
-6.31104231e-01 -4.28734064e-01 -2.21887067e-01 4.13535357e-01
-2.58819282e-01 -2.28582501e-01 -6.66816175e-01 -7.78656304e-01
2.49325365e-01 -1.09646702e+00 -2.23237544e-01 1.52828217e-01
-3.79209906e-01 3.78334075e-01 1.13748527e+00 -8.48193288e-01
8.13423872e-01 -2.36568809e+00 3.18442620e-02 4.66747940e-01
-2.09357649e-01 1.63624465e-01 -4.86065865e-01 2.54399478e-02
-3.87272656e-01 7.12664351e-02 -8.12254131e-01 -2.14399219e-01
-1.50374427e-01 4.36967432e-01 -2.66160786e-01 5.69547892e-01
4.00014013e-01 6.98712826e-01 -1.00551629e+00 -5.59630930e-01
5.66878676e-01 1.36586055e-01 -5.94895244e-01 2.08471522e-01
-3.16632003e-01 9.87299860e-01 -6.35681987e-01 8.35379481e-01
7.52843976e-01 -5.96419983e-02 -2.92578012e-01 -1.16696522e-01
3.82870674e-01 -6.40024990e-02 -9.19291914e-01 2.03935766e+00
-6.60809100e-01 4.47943747e-01 -6.33300319e-02 -1.17490542e+00
8.94351721e-01 1.67667136e-01 6.43065929e-01 -8.85275304e-01
-1.96742877e-01 3.57405812e-01 -2.43128344e-01 -3.45227182e-01
2.27134183e-01 -5.75765371e-02 -3.51715088e-01 8.99598897e-02
3.15927446e-01 -6.00890219e-01 -8.61465186e-02 1.21704251e-01
8.45851481e-01 4.06411797e-01 3.14311445e-01 -1.34851456e-01
2.96292692e-01 4.08881038e-01 7.48611450e-01 5.17725050e-01
-5.20080388e-01 9.85294878e-01 9.12615210e-02 8.69025737e-02
-1.17019236e+00 -1.65894639e+00 -2.46109158e-01 1.00699127e+00
2.74250060e-01 2.40268409e-01 -9.02880788e-01 -1.06635082e+00
-4.63143773e-02 1.10261238e+00 -7.09134281e-01 -4.53905374e-01
-3.72406065e-01 -4.14665878e-01 4.14166898e-01 7.92157888e-01
7.05975413e-01 -7.46006072e-01 -2.38536090e-01 3.06213927e-02
-1.29092127e-01 -1.28376269e+00 -6.82379901e-01 1.98407620e-01
-8.36953700e-01 -9.52661872e-01 -8.01207483e-01 -7.55916238e-01
8.50765944e-01 3.80612575e-02 9.77761209e-01 -5.58916092e-01
-3.94352451e-02 6.61492050e-01 -3.73200059e-01 -2.86845803e-01
-3.17101508e-01 1.09091140e-01 1.19721116e-02 1.53818011e-01
4.16804820e-01 -5.74912608e-01 -4.68351781e-01 7.14177310e-01
-8.88754904e-01 -2.02491254e-01 3.85776877e-01 8.09180915e-01
1.00083244e+00 -2.35243887e-01 5.66830814e-01 -8.06885898e-01
4.08575088e-02 -6.98162615e-01 -6.05398834e-01 1.32379562e-01
-5.61397433e-01 -1.87574312e-01 5.08819699e-01 -4.34077829e-01
-1.37707877e+00 4.00700480e-01 -2.26465687e-02 -6.42656326e-01
-8.07997942e-01 2.64238179e-01 -5.00505745e-01 2.29294859e-02
9.59530294e-01 7.16557950e-02 -3.03542346e-01 -3.92986953e-01
4.90560800e-01 6.65756643e-01 9.26668465e-01 -7.25047708e-01
1.16040015e+00 6.43354118e-01 -1.98217615e-01 -6.73209965e-01
-7.53775418e-01 -8.81959319e-01 -8.50472689e-01 -3.98859419e-02
1.03636229e+00 -1.01774633e+00 5.18800139e-01 5.88233232e-01
-7.69900382e-01 -6.78655148e-01 -5.71480572e-01 5.66123843e-01
-9.18193400e-01 9.87504944e-02 -9.52028185e-02 -1.79957286e-01
1.28492683e-01 -1.04430187e+00 1.14109361e+00 3.53140175e-01
-2.49247864e-01 -1.30468988e+00 1.87717266e-02 5.29890239e-01
2.02102035e-01 3.85628492e-01 8.56625140e-01 -1.07350969e+00
-1.93849444e-01 -2.12106362e-01 -1.88060567e-01 6.65798604e-01
4.04046088e-01 -2.11713567e-01 -1.11435843e+00 -2.30907068e-01
-8.71815160e-02 -1.67461291e-01 5.88824987e-01 6.75360620e-01
1.22467732e+00 1.07417539e-01 -4.36393648e-01 8.29005122e-01
1.43914914e+00 2.54625976e-01 4.22350496e-01 5.22570550e-01
6.76388979e-01 5.55797160e-01 1.05873287e+00 1.23326525e-01
2.03386217e-01 7.56243646e-01 2.60837764e-01 -4.63092297e-01
-3.75542164e-01 -4.87379998e-01 1.70790210e-01 2.98182189e-01
3.11819583e-01 -2.82577723e-01 -1.16817248e+00 1.05056691e+00
-1.53837800e+00 -4.98424917e-01 1.22453123e-01 2.37351727e+00
7.68451810e-01 1.73820689e-01 1.28012383e-02 -2.81141520e-01
8.45233679e-01 1.61151528e-01 -1.03381121e+00 -2.37413183e-01
-2.14237705e-01 2.94069707e-01 9.01821792e-01 4.89929497e-01
-1.32056534e+00 1.16685462e+00 5.44733858e+00 1.01562405e+00
-1.19999170e+00 4.69141781e-01 7.89472580e-01 -1.77419975e-01
-1.85701862e-01 -2.47363616e-02 -5.36063194e-01 4.27914649e-01
9.44895446e-01 2.54391544e-02 2.48605460e-01 9.38664019e-01
2.27975026e-01 -1.82379484e-01 -1.10171211e+00 7.26614952e-01
-3.07387812e-03 -9.84051466e-01 -1.87223285e-01 -5.79737797e-02
1.18964803e+00 2.30165660e-01 3.95576715e-01 3.07777941e-01
4.16741312e-01 -7.91043937e-01 5.19057512e-01 1.45179197e-01
1.12741733e+00 -5.85752547e-01 4.73451972e-01 7.35263973e-02
-1.14508498e+00 8.42067599e-03 4.74318350e-03 3.93369883e-01
3.03800762e-01 3.59060973e-01 -8.38155866e-01 6.78819537e-01
8.77814174e-01 9.83218610e-01 -5.93197525e-01 8.65201890e-01
-2.68591642e-01 9.53105390e-01 -3.43671679e-01 7.92383373e-01
3.53249431e-01 -1.99866846e-01 7.04031289e-01 1.07109725e+00
3.26972127e-01 -1.28135785e-01 2.63671935e-01 8.22061419e-01
9.04039070e-02 -4.36837040e-02 -7.64265656e-01 2.13554949e-01
5.50273359e-01 8.72753680e-01 -8.44468474e-01 -4.27040547e-01
-3.36634457e-01 1.21210396e+00 -1.15564831e-01 9.10662770e-01
-8.56788933e-01 -1.20896474e-01 8.73337567e-01 8.57872814e-02
4.12347466e-01 -1.47616938e-01 -7.96841025e-01 -1.02057636e+00
3.63696069e-02 -7.68603623e-01 5.00510871e-01 -8.79841387e-01
-1.42930460e+00 2.25684077e-01 4.23600942e-01 -1.56883001e+00
-3.27618986e-01 -4.63378906e-01 -5.34382284e-01 1.08882844e+00
-1.55239773e+00 -1.32249832e+00 -2.64263451e-01 9.12528217e-01
8.75819027e-01 -2.46624783e-01 8.37540269e-01 1.85289189e-01
-5.09211779e-01 6.99499428e-01 4.07144040e-01 1.79957628e-01
1.05648243e+00 -1.13739753e+00 5.99748135e-01 1.00923121e+00
5.28617725e-02 1.67129517e-01 6.40747547e-01 -7.39554584e-01
-5.84790826e-01 -1.50348163e+00 4.60907668e-01 -5.81725419e-01
6.69759393e-01 -2.88781404e-01 -1.06305814e+00 8.32182467e-01
-8.74630213e-02 3.03145468e-01 6.10671580e-01 -1.52073920e-01
-5.29951811e-01 -2.17700243e-01 -1.47775555e+00 6.13214374e-01
1.00776076e+00 -6.64215803e-01 -5.30485630e-01 3.62963498e-01
7.37281680e-01 -5.22342563e-01 -8.98877203e-01 3.34626883e-01
1.66191131e-01 -5.83287716e-01 1.11480618e+00 -5.50166249e-01
3.92472506e-01 -3.73292714e-01 -4.23170298e-01 -1.69612896e+00
-2.26926237e-01 -1.02718800e-01 2.73648947e-01 1.30064428e+00
5.73640823e-01 -8.95684958e-01 8.53745997e-01 6.21337771e-01
-3.74500811e-01 -4.73412842e-01 -1.06048119e+00 -1.18747282e+00
4.75011170e-01 -5.30252457e-01 5.58131158e-01 1.14201391e+00
-5.73179364e-01 5.18724211e-02 1.25411302e-01 6.33224189e-01
5.62227190e-01 -5.77572174e-03 5.64476430e-01 -9.19436574e-01
-1.91596314e-01 -2.66395867e-01 -3.39907348e-01 -7.34195769e-01
4.42007422e-01 -8.71147871e-01 1.76930532e-01 -1.40671349e+00
-1.16175391e-01 -5.96118927e-01 -4.43175226e-01 5.93234181e-01
-6.78787082e-02 3.10659766e-01 8.00796077e-02 2.31774420e-01
-2.57543206e-01 4.49423462e-01 1.16793013e+00 -3.71526778e-01
-3.15916061e-01 -1.05392933e-03 -4.98178065e-01 6.56457305e-01
1.13283336e+00 -5.52093923e-01 -6.05854809e-01 -5.37759125e-01
-5.22754252e-01 -3.31085056e-01 4.64318365e-01 -1.14104724e+00
-1.59763217e-01 -4.75116193e-01 4.07713950e-01 -2.44064406e-01
3.97506356e-01 -1.12615359e+00 -1.02643073e-02 7.65495449e-02
-3.22073162e-01 -3.31074059e-01 4.43579525e-01 7.14980900e-01
-2.61826575e-01 -1.95789054e-01 1.06131124e+00 1.58860177e-01
-1.21182048e+00 3.65114689e-01 -3.13714370e-02 4.18204904e-01
1.18390346e+00 -4.99902368e-01 -1.61893427e-01 -2.97188610e-01
-7.15144813e-01 2.22096533e-01 9.33132231e-01 4.89099562e-01
3.57371032e-01 -1.19247198e+00 -6.25094354e-01 2.76002169e-01
6.34297669e-01 3.30312788e-01 4.62682247e-01 6.78157568e-01
-2.25976527e-01 2.35772789e-01 -3.46532732e-01 -9.04743314e-01
-1.03143144e+00 5.37090719e-01 3.42720032e-01 -1.67739183e-01
-3.73120785e-01 7.93086350e-01 5.81535637e-01 -8.86768699e-01
-6.31951764e-02 -2.40021050e-01 2.54882663e-01 -1.99210718e-01
9.34763160e-03 8.34133178e-02 1.18442833e-01 -9.01534021e-01
-3.25353295e-01 9.27360415e-01 8.83044768e-03 -2.64112890e-01
1.06760550e+00 -2.16061488e-01 5.99443138e-01 2.72126973e-01
1.41369343e+00 -1.93718180e-01 -1.83768535e+00 -4.74303842e-01
-1.70192897e-01 -5.12076735e-01 -1.73434187e-02 -1.12561429e+00
-1.25219846e+00 9.14580524e-01 1.03418374e+00 -4.80284065e-01
1.41336679e+00 1.84936211e-01 6.41420543e-01 8.46339911e-02
1.59401685e-01 -1.53886724e+00 5.77179417e-02 3.37533325e-01
6.96513176e-01 -1.57296813e+00 -2.93422163e-01 -4.38934416e-01
-9.74199772e-01 5.42641163e-01 7.67499268e-01 -1.30191401e-01
6.09423876e-01 -4.04174402e-02 2.90573567e-01 6.65763905e-03
-1.27431646e-01 -1.92053020e-01 2.95596600e-01 1.34396625e+00
-1.93827793e-01 2.76392341e-01 3.12223405e-01 3.99716824e-01
-2.82768924e-02 -1.02876350e-01 1.43880576e-01 8.74499261e-01
-7.20329136e-02 -9.88665581e-01 -5.74454665e-01 4.28918272e-01
1.19119346e-01 1.00735566e-02 -2.59020925e-01 8.37887645e-01
1.71083629e-01 9.50921118e-01 2.18551442e-01 -3.03863436e-01
4.94081855e-01 1.75684005e-01 6.57692403e-02 -8.72335076e-01
-1.45155519e-01 5.19032124e-03 1.46818813e-02 -6.73350573e-01
-3.04768831e-01 -9.01219070e-01 -1.35773933e+00 2.25590870e-01
-3.46743898e-03 -2.61995584e-01 7.19195902e-01 9.40748632e-01
4.26729262e-01 4.00899470e-01 6.71481192e-01 -9.29606855e-01
-3.59631836e-01 -7.63395548e-01 -6.20564818e-01 7.45049536e-01
3.18878919e-01 -7.83309937e-01 -2.49640331e-01 4.82086390e-01]
|
[9.743281364440918, 1.3745609521865845]
|
4c56df19-2b0d-4d2c-a550-9686d98f8b3a
|
learning-representations-from-product-titles
|
1811.01166
| null |
https://arxiv.org/abs/1811.01166v3
|
https://arxiv.org/pdf/1811.01166v3.pdf
|
Learning Representations from Product Titles for Modeling Shopping Transactions
|
Shopping transaction analysis is important for understanding the shopping behaviors of customers. Existing models such as association rules are poor at modeling products that have short purchase histories and cannot be applied to new products (the cold-start problem). In this paper, we propose BASTEXT, an efficient model of shopping baskets and the texts associated with the products (e.g., product titles). The model's goal is to learn the product representations from the textual contents to capture the relationships between the products in the baskets. Given the products already in a basket, a classifier identifies whether a potential product is relevant to the basket based on their vector representations. This relevancy enables us to learn high-quality representations of the products. The experiments demonstrate that BASTEXT can efficiently model millions of baskets and that it outperforms the state-of-the-art methods in the next product recommendation task. We also show that BASTEXT is a strong baseline for keyword-based product search.
|
['Binh Nguyen', 'Atsuhiro Takasu']
|
2018-11-03
| null | null | null | null |
['product-recommendation']
|
['miscellaneous']
|
[-3.33798558e-01 -4.52728003e-01 -9.52779293e-01 -8.43707919e-01
-3.57345879e-01 -7.10475445e-01 3.55035484e-01 4.65935051e-01
-3.07556063e-01 1.50875643e-01 4.46816117e-01 -3.39647442e-01
-1.72980338e-01 -1.06199634e+00 -7.82952726e-01 -2.42938802e-01
-1.91362098e-01 1.01544654e+00 7.89932609e-02 -5.49402356e-01
2.92307705e-01 4.13320512e-02 -1.67929971e+00 8.77846241e-01
5.81454694e-01 1.20548308e+00 3.94156098e-01 3.31271231e-01
-5.63902974e-01 5.82508266e-01 7.86782205e-02 -9.41784739e-01
4.46739614e-01 2.53505930e-02 -8.55151296e-01 -1.37496600e-02
1.19072594e-01 -5.34006894e-01 -4.70961720e-01 7.36072242e-01
-3.22769195e-01 4.06169355e-01 8.65680754e-01 -1.06046999e+00
-9.46416974e-01 1.24465275e+00 -6.26045585e-01 2.60901839e-01
4.67617780e-01 -2.07951725e-01 1.85463583e+00 -1.10518205e+00
6.21987402e-01 1.23905087e+00 5.40331006e-01 6.33272976e-02
-1.15476763e+00 -5.41974843e-01 7.52651095e-01 3.55289280e-01
-1.11438465e+00 -8.77628382e-03 3.84960473e-01 -3.73632342e-01
1.31779218e+00 1.75157949e-01 8.91094923e-01 1.16482055e+00
3.51165272e-02 1.41987872e+00 2.35280514e-01 -3.61123621e-01
5.56786656e-02 3.75414521e-01 1.05746734e+00 2.57113934e-01
5.01902938e-01 5.70786782e-02 -7.29325473e-01 -2.39983842e-01
4.71452415e-01 6.74708664e-01 3.24832171e-01 -4.42482769e-01
-8.46095681e-01 1.28535521e+00 3.82538468e-01 2.91025251e-01
-6.94392979e-01 -9.06211361e-02 5.67506969e-01 3.32767367e-01
2.80166477e-01 5.95104456e-01 -9.97781515e-01 1.27689779e-01
-5.99709511e-01 8.07220280e-01 1.14464831e+00 1.36263263e+00
6.16876066e-01 -5.49130976e-01 1.63698450e-01 8.28685701e-01
8.81676674e-01 2.06635877e-01 5.63868523e-01 -2.03895614e-01
6.25022352e-01 5.42134345e-01 2.64219820e-01 -8.67087007e-01
-2.06656709e-01 -4.76389498e-01 -2.55774915e-01 -6.02064848e-01
1.59545884e-01 1.60943925e-01 -9.64632750e-01 1.13484621e+00
2.05481291e-01 3.16105038e-02 -2.05451436e-02 6.05125189e-01
8.38000119e-01 8.51031780e-01 3.25862765e-01 -1.42184734e-01
1.47409487e+00 -1.06472850e+00 -4.82890815e-01 -5.20462275e-01
8.42123687e-01 -1.04102933e+00 8.96650374e-01 7.37928510e-01
-8.98746192e-01 -7.60014415e-01 -1.00164044e+00 -7.34235793e-02
-7.07799137e-01 -1.90602005e-01 1.22984254e+00 4.84088987e-01
-1.93304405e-01 7.24275529e-01 -7.85068870e-01 -5.08151948e-01
1.96118280e-01 2.16819808e-01 2.86380529e-01 -5.05336404e-01
-1.41010213e+00 8.60182166e-01 6.65035069e-01 -7.10401610e-02
-6.91017687e-01 -7.69476175e-01 -1.00960469e+00 3.34956616e-01
4.79915202e-01 -5.89270055e-01 1.55639434e+00 -7.03170717e-01
-8.43387783e-01 4.06510442e-01 -7.30414316e-02 -6.09737217e-01
-1.59377873e-01 -7.25770593e-01 -9.44483757e-01 -5.90592384e-01
3.26999098e-01 2.65418828e-01 6.30665958e-01 -1.14495349e+00
-1.39951086e+00 -4.31721777e-01 -3.49646388e-03 1.65044576e-01
-2.59437054e-01 -9.97896269e-02 -1.08684254e+00 -5.55716991e-01
1.82823464e-01 -9.62582290e-01 -3.55876923e-01 -7.60743022e-01
-3.96153182e-01 -5.01766562e-01 2.35089332e-01 -2.18242303e-01
1.38569331e+00 -2.27242756e+00 -4.68674660e-01 6.28577650e-01
-1.26199782e-01 -6.89178035e-02 -2.93776780e-01 7.56035864e-01
9.59739909e-02 1.88497808e-02 8.40452254e-01 -1.46383941e-01
3.70182991e-01 4.65680271e-01 -7.54518151e-01 1.81033537e-02
-3.75946879e-01 1.16107655e+00 -9.71944094e-01 -2.42551327e-01
9.23099816e-02 1.36229306e-01 -4.46130663e-01 1.74034193e-01
-5.56022942e-01 -4.26548183e-01 -8.05114448e-01 7.07132876e-01
4.57220763e-01 -4.87837553e-01 4.50855941e-01 -4.03360933e-01
3.45364511e-01 8.79405797e-01 -1.15369165e+00 1.50666583e+00
-4.70959365e-01 -1.00140251e-01 -4.00737792e-01 -6.73223495e-01
7.11732209e-01 -9.24619064e-02 4.80574638e-01 -9.88872945e-01
-3.26915570e-02 1.80687949e-01 -7.83021748e-02 -4.26058292e-01
7.93374121e-01 2.49599162e-02 -3.26719195e-01 5.72645485e-01
2.45416850e-01 2.99655885e-01 6.86191261e-01 3.81446362e-01
5.80172658e-01 4.98141907e-02 3.24593902e-01 -1.05667681e-01
-5.05144410e-02 1.19378559e-01 3.89178276e-01 9.93920684e-01
4.10629660e-01 5.39903268e-02 -1.59762457e-01 -8.91597331e-01
-9.25390065e-01 -1.00644517e+00 -2.29061744e-03 1.73132217e+00
3.94592971e-01 -1.02531636e+00 1.05654439e-02 -9.75504100e-01
7.06370652e-01 9.89615381e-01 -7.49118447e-01 -1.33652568e-01
-4.36770111e-01 -7.21575022e-01 -3.02432835e-01 9.57947731e-01
-1.45803317e-01 -8.07803929e-01 6.57845587e-02 6.20353162e-01
-1.44465402e-01 -6.28137946e-01 -7.33006001e-01 4.55769688e-01
-8.84044766e-01 -1.07953215e+00 -1.19004898e-01 -1.00863349e+00
2.08779305e-01 5.26039481e-01 1.67288101e+00 -9.12380293e-02
2.92199403e-02 -1.95773598e-02 -6.07159019e-01 -4.99701917e-01
-2.38350198e-01 4.13256198e-01 1.82056233e-01 9.44019482e-02
1.25223863e+00 -1.82112738e-01 -8.21125150e-01 6.14948630e-01
-6.40763283e-01 -2.61490285e-01 5.60221076e-01 7.68352032e-01
7.01069534e-01 5.08614719e-01 4.81771290e-01 -1.51415336e+00
6.26034319e-01 -7.97312081e-01 -4.53100443e-01 4.50863510e-01
-1.21088386e+00 7.91484937e-02 4.36373979e-01 -7.02553570e-01
-1.09813929e+00 2.34649852e-01 -3.82756621e-01 1.24766015e-01
-1.08894713e-01 1.03455055e+00 1.60142750e-01 7.79726148e-01
5.74301064e-01 7.77580664e-02 -6.99931145e-01 -9.73802805e-01
7.60573387e-01 4.00023609e-01 2.87409902e-01 -2.67335981e-01
3.37997347e-01 5.64920828e-02 -5.39195001e-01 -3.71356606e-01
-1.29915392e+00 -1.34166396e+00 -6.11521840e-01 3.59255701e-01
1.95145503e-01 -1.05252779e+00 -8.42209995e-01 -1.81799129e-01
-5.69580674e-01 1.23566158e-01 -3.75091940e-01 7.89395213e-01
-2.64457017e-01 2.10085034e-01 -1.02336764e+00 -6.69742644e-01
-4.80051845e-01 -6.87548339e-01 8.72117579e-01 9.44150984e-02
-6.72951579e-01 -9.28725719e-01 1.55358419e-01 3.83683860e-01
8.85040592e-03 -5.27709603e-01 1.21604276e+00 -1.18922806e+00
-4.67353910e-01 -4.97655302e-01 1.81347594e-01 4.16716933e-02
1.58762366e-01 -2.26415843e-01 -6.50149643e-01 -3.97586554e-01
-1.04770996e-01 -2.21870840e-01 1.29185963e+00 5.77214003e-01
6.47095025e-01 -4.20332283e-01 -9.08713579e-01 3.54984641e-01
1.39673674e+00 5.36558926e-01 3.32326561e-01 3.36549938e-01
4.90822107e-01 5.74177504e-01 1.24560523e+00 5.47693372e-01
5.35639822e-01 5.65891743e-01 2.65984535e-01 3.42517830e-02
4.70688939e-01 -8.79370272e-01 1.68630213e-01 8.15295458e-01
3.27299088e-01 -3.70329052e-01 -4.17468905e-01 8.48633349e-01
-2.18771863e+00 -8.05856824e-01 -2.87316293e-01 2.01092505e+00
7.02788532e-01 5.69705486e-01 3.41235965e-01 -2.47100294e-01
2.12423265e-01 -1.88530892e-01 -7.74839938e-01 -4.93046075e-01
1.99186638e-01 2.95719095e-02 5.83091259e-01 2.23421440e-01
-1.39243567e+00 8.62462223e-01 7.02114487e+00 6.78873718e-01
-2.74686903e-01 -8.37944672e-02 5.28578699e-01 -2.01276898e-01
-4.34361190e-01 -1.21716738e-01 -1.62061226e+00 3.85381609e-01
1.01345265e+00 -2.00198427e-01 1.84208617e-01 1.30771255e+00
-2.37005785e-01 1.10625602e-01 -1.60959327e+00 9.01669383e-01
9.87222195e-02 -1.27032185e+00 3.48978639e-01 2.29242116e-01
7.81257808e-01 2.34701291e-01 2.04232588e-01 8.37023199e-01
1.06487203e+00 -7.82775700e-01 5.51719964e-01 2.41960779e-01
3.57322469e-02 -8.71943772e-01 7.86900222e-01 2.78679699e-01
-1.49240267e+00 -4.10878301e-01 -6.21667504e-01 1.73736989e-01
4.18034345e-01 4.10515219e-01 -1.03367972e+00 4.32406694e-01
8.90589833e-01 1.17143190e+00 -3.45513463e-01 9.43316221e-01
3.29181366e-02 7.04473734e-01 -2.76882917e-01 -2.83623457e-01
5.13561666e-01 -1.79562911e-01 -3.17023337e-01 1.23140371e+00
2.98933655e-01 3.89498323e-02 5.31293452e-01 7.20974863e-01
-8.42010826e-02 5.62688708e-01 -6.39797091e-01 -2.76068598e-01
2.25398824e-01 1.07290530e+00 -5.78023016e-01 -4.98538554e-01
-9.44158077e-01 6.66581273e-01 1.74843565e-01 3.03909600e-01
-3.98184329e-01 -6.15690313e-02 1.02816451e+00 1.66999772e-01
9.36542094e-01 9.56097022e-02 -2.06497312e-01 -1.11250699e+00
-8.28530788e-02 -8.45062792e-01 8.10073674e-01 -4.02123332e-01
-2.00384760e+00 4.96941835e-01 -1.18268333e-01 -1.06581855e+00
-5.48066020e-01 -5.14065683e-01 -2.40556002e-01 8.55388939e-01
-1.26394761e+00 -1.25238454e+00 4.18896466e-01 6.03160977e-01
9.57953811e-01 -1.48680374e-01 8.68803084e-01 2.62919813e-01
-6.97034299e-02 4.75935280e-01 5.61310053e-01 1.93674490e-01
4.25147235e-01 -1.23169065e+00 1.12144554e+00 3.85199755e-01
7.55392373e-01 1.21977425e+00 6.27903044e-01 -1.04980719e+00
-1.67918098e+00 -1.06572855e+00 1.29422629e+00 -7.07406700e-01
7.55037844e-01 -4.83741462e-01 -9.15898621e-01 1.02284205e+00
6.31924346e-03 -4.69167769e-01 1.17343175e+00 1.14395559e+00
-8.00458491e-01 -3.94368291e-01 -6.29011869e-01 4.61956918e-01
7.86770582e-01 -3.66490841e-01 -8.31929207e-01 5.52468598e-01
7.91368186e-01 1.81589037e-01 -9.13287997e-01 4.71719392e-02
1.04397011e+00 -3.53421539e-01 1.26815808e+00 -1.11454248e+00
1.58087313e-01 2.23298892e-01 -2.10382521e-01 -1.33486319e+00
-9.60707128e-01 -1.93345577e-01 -3.47535342e-01 1.05866945e+00
9.84648407e-01 -1.01858310e-01 1.01749921e+00 8.57754648e-01
1.94600344e-01 -6.95979238e-01 -1.24161527e-01 -7.48665392e-01
-2.41701275e-01 -5.19331872e-01 1.05207682e+00 7.50562370e-01
5.56709766e-01 9.14156199e-01 -4.81553584e-01 -5.19533351e-04
3.97157162e-01 8.73487294e-01 5.94960392e-01 -1.49115694e+00
-6.47245646e-01 -2.89255559e-01 7.77416527e-02 -1.70752811e+00
-2.20131531e-01 -9.56696630e-01 5.43258935e-02 -1.52391410e+00
4.39475000e-01 -5.58365643e-01 -8.27891648e-01 2.91216612e-01
-9.42630768e-02 -5.60400523e-02 8.56944397e-02 2.54217684e-01
-8.30894232e-01 -9.80295707e-03 1.04793561e+00 -4.52004403e-01
-5.32807410e-01 5.68946242e-01 -1.08557320e+00 4.31430161e-01
6.67126417e-01 -4.14322644e-01 -7.14627445e-01 -3.05032253e-01
8.82386804e-01 -2.61289775e-01 -5.52513599e-01 -1.42430663e-01
2.49990284e-01 -1.86014727e-01 5.79889536e-01 -1.28485417e+00
3.58469635e-01 -9.10113096e-01 3.29295605e-01 2.73466051e-01
-8.19168091e-01 1.14592753e-01 -9.57112089e-02 9.15669322e-01
-1.06733806e-01 -5.04131138e-01 1.26195505e-01 -5.19630253e-01
-9.36800539e-01 6.16810918e-01 -3.25335562e-01 -3.54792655e-01
5.75095832e-01 1.82465717e-01 5.97427078e-02 -2.18792960e-01
-9.74719644e-01 4.89616871e-01 -3.61675732e-02 9.91069853e-01
6.58248305e-01 -1.33201432e+00 -5.56083083e-01 3.85682225e-01
6.38061702e-01 -4.20411378e-01 4.77036834e-02 8.04277733e-02
-4.03509922e-02 9.13613856e-01 1.61410481e-01 -1.99493602e-01
-1.23407710e+00 1.34860420e+00 -2.13395521e-01 -6.43110693e-01
-5.53816497e-01 1.03135800e+00 4.58501786e-01 -1.02067649e-01
4.33446556e-01 -7.04520047e-01 -6.79666519e-01 2.57418364e-01
7.55183518e-01 -5.77118136e-02 3.60143304e-01 -4.96961027e-01
-1.62110567e-01 1.73731983e-01 -1.32372677e+00 2.14002326e-01
1.55054557e+00 -3.49186450e-01 3.08028013e-01 5.26772678e-01
1.17002738e+00 -4.64885473e-01 -9.18425739e-01 -7.76012063e-01
4.73305374e-01 -7.77080238e-01 2.81843897e-02 -8.88813734e-01
-1.12021124e+00 3.95399779e-01 5.24717152e-01 4.79761153e-01
6.28874481e-01 3.45894009e-01 1.23848915e+00 6.38052821e-01
6.72405541e-01 -1.17201507e+00 -1.56413540e-01 4.27907199e-01
5.19199967e-01 -1.24018562e+00 4.44331132e-02 -5.16203344e-01
-8.26877475e-01 8.84463787e-01 4.02596831e-01 -4.37977351e-02
1.21600008e+00 1.37845814e-01 -1.86213423e-02 -4.64013219e-01
-1.13050377e+00 -3.72163922e-01 4.83815134e-01 4.79556471e-01
5.36490798e-01 2.63369918e-01 -1.78395510e-01 9.67474520e-01
-2.04810902e-01 -7.49013126e-02 -3.18433456e-02 9.57389355e-01
-4.49857652e-01 -1.42166996e+00 2.55586743e-01 1.02842021e+00
-4.95171815e-01 -3.54808629e-01 -2.21519977e-01 4.38465208e-01
1.29122555e-01 1.14626229e+00 2.68783215e-02 -6.38828874e-01
5.81468880e-01 1.87669232e-01 2.50625968e-01 -1.04843867e+00
-8.18930209e-01 5.32852471e-01 3.21709156e-01 -6.04435265e-01
-4.05887961e-02 -7.69213796e-01 -9.08485651e-01 -4.56674278e-01
-7.31901586e-01 3.48832101e-01 5.63108504e-01 7.78692245e-01
2.76167274e-01 6.06110394e-02 6.24952972e-01 -3.22186351e-01
-5.70799291e-01 -9.34986949e-01 -1.17739773e+00 8.09421122e-01
1.53760135e-01 -7.50575364e-01 2.47403622e-01 2.02675655e-01]
|
[10.076894760131836, 5.900309085845947]
|
459d869b-71bc-4833-9ba4-7f138a860485
|
find-beauty-in-the-rare-contrastive
|
2302.08662
| null |
https://arxiv.org/abs/2302.08662v1
|
https://arxiv.org/pdf/2302.08662v1.pdf
|
Find Beauty in the Rare: Contrastive Composition Feature Clustering for Nontrivial Cropping Box Regression
|
Automatic image cropping algorithms aim to recompose images like human-being photographers by generating the cropping boxes with improved composition quality. Cropping box regression approaches learn the beauty of composition from annotated cropping boxes. However, the bias of annotations leads to quasi-trivial recomposing results, which has an obvious tendency to the average location of training samples. The crux of this predicament is that the task is naively treated as a box regression problem, where rare samples might be dominated by normal samples, and the composition patterns of rare samples are not well exploited. Observing that similar composition patterns tend to be shared by the cropping boundaries annotated nearly, we argue to find the beauty of composition from the rare samples by clustering the samples with similar cropping boundary annotations, ie, similar composition patterns. We propose a novel Contrastive Composition Clustering (C2C) to regularize the composition features by contrasting dynamically established similar and dissimilar pairs. In this way, common composition patterns of multiple images can be better summarized, which especially benefits the rare samples and endows our model with better generalizability to render nontrivial results. Extensive experimental results show the superiority of our model compared with prior arts. We also illustrate the philosophy of our design with an interesting analytical visualization.
|
['Weicai Zhong', 'Zhiguo Cao', 'Hao Lu', 'Jiale Zhang', 'Yinpeng Chen', 'Zhiyu Pan']
|
2023-02-17
| null | null | null | null |
['image-cropping', 'philosophy']
|
['computer-vision', 'miscellaneous']
|
[ 5.39758146e-01 1.45152127e-02 -2.40873545e-01 -1.36781007e-01
-1.53163657e-01 -5.38529217e-01 6.55148923e-01 -1.88192755e-01
1.86230958e-01 3.74073297e-01 2.78746128e-01 1.59896523e-01
2.64849178e-02 -5.66660702e-01 -8.38925242e-01 -1.01405263e+00
1.17189877e-01 1.79871112e-01 1.62255064e-01 -2.54723310e-01
3.40950698e-01 4.03238863e-01 -1.93058908e+00 6.82375491e-01
9.66230333e-01 9.24571633e-01 3.28836173e-01 4.53760564e-01
-2.17894942e-01 7.18747854e-01 -5.49930632e-01 -5.32862365e-01
7.56404042e-01 -6.83055818e-01 -3.22931319e-01 6.60841644e-01
1.00127840e+00 -1.88278571e-01 4.51084301e-02 1.41828167e+00
3.17841768e-02 1.04750581e-02 6.59158051e-01 -1.51220989e+00
-9.61426258e-01 7.33875215e-01 -1.09464467e+00 -4.08696651e-01
1.17928810e-01 2.76487082e-01 1.08671939e+00 -7.87388921e-01
8.23632717e-01 1.27936697e+00 6.65781915e-01 6.38880074e-01
-1.35040402e+00 -7.92694867e-01 5.73603868e-01 1.55523820e-02
-1.45337927e+00 -3.65178674e-01 9.49345469e-01 -5.88175237e-01
-5.90557158e-02 6.90157056e-01 8.60789716e-01 9.59038258e-01
-9.25919041e-02 8.13806057e-01 1.32726657e+00 -2.78432459e-01
2.20024168e-01 3.78392845e-01 -2.64564127e-01 5.67216218e-01
3.87944788e-01 -1.11733861e-01 -5.12305021e-01 2.88709942e-02
7.58587778e-01 3.81038457e-01 -6.60164058e-01 -8.48526537e-01
-1.38431406e+00 3.98414642e-01 4.79982734e-01 1.70275509e-01
-2.27948770e-01 -7.05960989e-02 1.41677514e-01 6.11648671e-02
5.20095050e-01 5.41733682e-01 -2.93280363e-01 3.35273176e-01
-1.12037861e+00 4.06428069e-01 5.22109926e-01 1.39074671e+00
1.06628108e+00 -9.43934917e-02 -1.05466060e-01 7.78638542e-01
-7.31599852e-02 4.30426925e-01 4.22648460e-01 -9.48737741e-01
9.64161158e-02 8.38343740e-01 1.47294616e-02 -1.10856545e+00
3.77648771e-02 -1.36739522e-01 -1.22159684e+00 5.42286813e-01
6.30707145e-01 5.39486296e-02 -8.44258249e-01 1.57986891e+00
2.85176784e-01 2.84988843e-02 -3.73123825e-01 1.05531657e+00
2.76535183e-01 3.67266834e-01 1.61232688e-02 -1.59430072e-01
1.49555409e+00 -1.17576575e+00 -7.71138847e-01 9.86238644e-02
7.89495707e-02 -7.75245488e-01 1.43078876e+00 4.40618515e-01
-8.66880119e-01 -6.25342727e-01 -1.11946893e+00 1.06982194e-01
-1.73546672e-01 -2.46690679e-02 6.50725424e-01 4.91212696e-01
-7.32532918e-01 9.62717772e-01 -2.05974132e-01 -2.24219128e-01
7.05561399e-01 -2.51549780e-01 -2.39291310e-01 7.95736164e-03
-3.88054520e-01 3.63691002e-01 3.99732053e-01 -1.14713885e-01
-7.38803744e-01 -1.25309920e+00 -5.56826174e-01 -1.57254338e-01
5.69224119e-01 -5.45056343e-01 7.78873742e-01 -1.79421890e+00
-1.32636631e+00 9.66357470e-01 -2.25791618e-01 -1.93059504e-01
8.35947454e-01 -1.14353120e-01 -4.19424981e-01 9.36403647e-02
4.83886339e-02 6.64451003e-01 1.32487988e+00 -2.08917451e+00
-8.72993350e-01 -1.59758478e-01 -3.15621823e-01 2.81399488e-01
-2.29480758e-01 -1.98373973e-01 -3.53439927e-01 -8.55690837e-01
2.74370193e-01 -7.16659188e-01 -2.21588016e-01 5.99817038e-01
-4.48512107e-01 1.95111021e-01 7.02171504e-01 -5.21196544e-01
1.37331116e+00 -2.22285867e+00 3.97949200e-03 4.27066058e-01
6.31184638e-01 -3.20248008e-01 -4.64698635e-02 2.21926793e-01
-3.22440952e-01 3.60451251e-01 -5.12211800e-01 -8.88289437e-02
-8.50704499e-03 5.51648848e-02 -6.30217731e-01 5.91931581e-01
3.10332894e-01 7.42917538e-01 -1.09305191e+00 -5.42053223e-01
7.19412416e-02 2.00246889e-02 -3.68591577e-01 2.32184604e-01
-4.58847821e-01 4.84608322e-01 8.73004831e-03 8.52161586e-01
8.79937291e-01 -3.38447750e-01 4.11981583e-01 -4.72176105e-01
-4.61206809e-02 -2.94139594e-01 -1.22016072e+00 1.48292851e+00
9.86923128e-02 6.34638965e-01 3.23436819e-02 -7.45252609e-01
8.80052149e-01 -9.32428688e-02 5.07398546e-01 -5.90943098e-01
-9.58600044e-02 -4.04549763e-03 2.45463345e-02 -5.59302509e-01
6.89071178e-01 -4.33350623e-01 1.78576097e-01 5.79978228e-01
-3.52037787e-01 -3.89908612e-01 1.29203841e-01 5.61034568e-02
6.10253513e-01 4.55095589e-01 4.30807978e-01 -4.80580777e-01
1.02785759e-01 6.17004465e-03 6.91886187e-01 5.86469173e-01
-2.06086338e-01 1.01397872e+00 4.97607648e-01 -4.30955589e-01
-1.67879236e+00 -9.34619308e-01 -2.12138578e-01 9.24445093e-01
6.58675015e-01 -4.37148511e-01 -8.82260203e-01 -6.71375692e-01
-2.39608921e-02 4.36301559e-01 -1.06696272e+00 1.44826367e-01
-6.38100922e-01 -6.07045949e-01 2.28327408e-01 2.89419383e-01
4.01011974e-01 -9.49194014e-01 -4.89167809e-01 -3.85802388e-01
-1.18843317e-01 -8.42901289e-01 -7.42144585e-01 -1.24478891e-01
-6.84377372e-01 -1.21858442e+00 -1.00009573e+00 -7.63538361e-01
1.00051928e+00 7.08065510e-01 1.31516588e+00 4.17616606e-01
-2.07101703e-01 6.59876615e-02 -5.01535118e-01 -3.82503659e-01
-4.56718177e-01 -3.92490774e-01 2.79928409e-02 5.20465493e-01
1.43853873e-01 -7.67867744e-01 -8.99477243e-01 5.12054622e-01
-1.12059069e+00 5.77824593e-01 6.26692355e-01 6.10335827e-01
7.57855237e-01 2.19211593e-01 1.17525592e-01 -1.12632322e+00
4.16475356e-01 -4.00466859e-01 -2.42961437e-01 6.09032452e-01
-6.92339718e-01 3.17859687e-02 8.38105321e-01 -9.25751925e-01
-9.52768862e-01 -1.76515460e-01 6.82678699e-01 -8.16860259e-01
-2.21407413e-01 -1.87667415e-01 -4.01333094e-01 9.31410566e-02
6.67893946e-01 2.04364136e-01 2.24651918e-01 -4.60029453e-01
6.95229948e-01 2.84610897e-01 6.20209217e-01 -7.35432565e-01
1.13637638e+00 9.51160133e-01 -2.78347254e-01 -7.79687643e-01
-6.55104458e-01 -3.05278152e-01 -6.43787384e-01 -4.68928486e-01
7.26006806e-01 -8.27502489e-01 -5.30948937e-01 3.67082834e-01
-7.53557980e-01 -4.36107248e-01 -7.94391751e-01 -2.61421800e-01
-4.15242702e-01 7.99903274e-01 -2.93373704e-01 -7.99271584e-01
-7.79290199e-02 -7.95434356e-01 1.12653732e+00 3.11668664e-01
-4.35590476e-01 -5.85236788e-01 -1.58849239e-01 1.03786401e-01
1.20529458e-01 4.74413484e-01 1.17138577e+00 -9.53961164e-02
-7.76077807e-01 4.87186573e-02 -3.93687546e-01 2.76798159e-01
2.30221674e-01 5.45051098e-01 -1.02043247e+00 4.26607318e-02
-1.00131012e-01 1.68104261e-01 7.28571653e-01 1.30884841e-01
1.57751703e+00 -5.04424632e-01 -2.28663668e-01 8.24980974e-01
1.44330072e+00 -4.57538031e-02 7.89569795e-01 2.26122424e-01
9.50932980e-01 1.02437627e+00 5.36900580e-01 3.57932448e-01
-5.63894026e-02 4.92324561e-01 3.58319253e-01 -1.99963912e-01
-3.28780472e-01 -7.31887937e-01 2.21419469e-01 6.03756666e-01
-1.79441944e-01 2.69066636e-02 -5.53080142e-01 4.19742256e-01
-1.85463500e+00 -1.16760647e+00 -1.85835734e-01 2.23147154e+00
1.04974413e+00 -2.39534929e-01 3.14730287e-01 1.76289678e-01
9.49110210e-01 1.97253391e-01 -5.04596233e-01 1.31588861e-01
-5.11359692e-01 1.63375139e-02 5.46756625e-01 9.22482535e-02
-9.86888945e-01 7.68210351e-01 6.45735884e+00 1.24187851e+00
-1.00820744e+00 -1.68878183e-01 8.98001254e-01 -2.05706999e-01
-6.95473850e-01 2.46786907e-01 -5.22776604e-01 8.50119293e-01
7.32839108e-02 -1.02045856e-01 6.70806527e-01 8.58601749e-01
1.81679443e-01 -2.06663996e-01 -1.17799938e+00 1.09653914e+00
2.90758371e-01 -1.34070969e+00 4.30269063e-01 -4.50762510e-02
1.15321720e+00 -6.39815032e-01 2.12082565e-01 -1.93983361e-01
4.27242994e-01 -9.57695067e-01 1.34110880e+00 6.70284867e-01
9.77330089e-01 -3.13816160e-01 1.06145643e-01 1.31105676e-01
-1.10300219e+00 -2.22073674e-01 -4.38706338e-01 -2.26129871e-02
8.14348608e-02 8.28437924e-01 -5.60136497e-01 5.79284191e-01
8.47914994e-01 7.82850623e-01 -7.51063645e-01 9.84055638e-01
-1.84384838e-01 2.25671649e-01 1.05834052e-01 -1.66761559e-02
-2.10471511e-01 -7.95996904e-01 3.52369666e-01 1.03777182e+00
2.33242080e-01 -1.54864356e-01 -1.74551476e-02 1.27553785e+00
-6.87159300e-02 2.26093188e-01 -5.72332561e-01 2.10105032e-02
2.70262986e-01 1.37785566e+00 -9.98567224e-01 -6.08312011e-01
-1.72562599e-01 9.94908690e-01 2.58437663e-01 5.10966957e-01
-7.91847885e-01 4.99275811e-02 7.63382554e-01 5.46997309e-01
3.18865269e-01 7.58676827e-02 -8.00874114e-01 -1.26213014e+00
2.33162805e-01 -1.27380383e+00 2.34656055e-02 -9.79070306e-01
-1.74348211e+00 3.59449744e-01 -8.20966884e-02 -1.79100466e+00
4.86288488e-01 -4.14767206e-01 -6.70183361e-01 5.23389459e-01
-1.18591785e+00 -1.17130947e+00 -6.74660742e-01 3.12245041e-01
5.35576761e-01 9.77206379e-02 4.49847758e-01 1.09581128e-01
-4.89647299e-01 4.70355690e-01 1.45718947e-01 6.11658990e-02
9.33899164e-01 -1.36347055e+00 9.36459973e-02 9.95134413e-01
2.27569774e-01 6.66504800e-01 7.61410356e-01 -7.79779196e-01
-1.04068100e+00 -1.09064770e+00 3.53799433e-01 -5.18378198e-01
6.67780161e-01 -6.78331554e-01 -9.78504241e-01 2.77134269e-01
2.67373830e-01 -1.63831517e-01 6.00683689e-01 -1.46499619e-01
-7.07758427e-01 -2.41150677e-01 -8.21564317e-01 1.19673204e+00
1.29498672e+00 -2.47364625e-01 -2.70142019e-01 1.64023712e-01
7.10587561e-01 -1.03997579e-02 -6.17364764e-01 3.37235898e-01
8.23186874e-01 -1.21440911e+00 8.04725707e-01 -3.89890999e-01
9.39454556e-01 -6.50632799e-01 -7.35017806e-02 -1.14376938e+00
-2.57735968e-01 -7.28168488e-01 -1.34624904e-02 1.37527525e+00
1.40318677e-01 -2.15267643e-01 7.58629799e-01 6.84007943e-01
1.43551022e-01 -7.13192821e-01 -3.83469522e-01 -5.79120159e-01
1.02407552e-01 -2.00434327e-01 1.07630324e+00 1.08939993e+00
-2.24073827e-01 3.58694158e-02 -5.73602676e-01 -3.16716842e-02
7.13782787e-01 5.51637232e-01 1.10755825e+00 -1.06245506e+00
-2.62329727e-01 -8.78870189e-01 -1.67396337e-01 -1.09782064e+00
-1.05351560e-01 -6.66433871e-01 2.13199332e-01 -8.15558791e-01
6.26459479e-01 -7.76687503e-01 3.67314294e-02 6.44991398e-02
-6.05079889e-01 3.70437294e-01 2.73238212e-01 5.23683131e-01
-6.45570636e-01 6.03151143e-01 1.73746002e+00 -3.65204275e-01
-4.54475135e-02 -2.23923236e-01 -1.06721282e+00 8.26321721e-01
5.53798676e-01 -1.68799758e-01 -3.42202306e-01 -1.32584959e-01
2.87355274e-01 -6.79690838e-01 4.79867935e-01 -9.20296788e-01
3.34709398e-02 -4.03556138e-01 6.25629485e-01 -4.79714602e-01
-9.35469717e-02 -1.04580891e+00 4.50410932e-01 6.16746135e-02
-3.58215511e-01 -1.68539345e-01 -2.59828836e-01 7.21787274e-01
3.85220982e-02 -1.83081672e-01 7.68919230e-01 -3.20743740e-01
-7.50667930e-01 4.33915555e-01 -4.56501022e-02 3.44560519e-02
1.21508861e+00 -7.25506127e-01 -3.17894846e-01 -3.66714001e-01
-1.76378861e-01 -2.38086432e-02 1.19103360e+00 4.08481479e-01
5.27511060e-01 -1.44170070e+00 -5.35626709e-01 3.88154089e-01
3.11930090e-01 3.32093179e-01 5.11775851e-01 7.38106549e-01
-5.42123258e-01 -4.46369618e-01 -2.79483944e-01 -7.39057004e-01
-1.32130229e+00 1.01577473e+00 1.90220699e-01 9.74193364e-02
-9.93467391e-01 7.70881593e-01 8.50334764e-01 -2.04630181e-01
2.02014774e-01 -3.87833387e-01 -5.58993109e-02 1.13694653e-01
7.26652682e-01 2.83161342e-01 -3.61982614e-01 -3.96544933e-01
1.31699726e-01 7.91691780e-01 -1.48540959e-01 1.08552612e-01
1.20963705e+00 -2.49415949e-01 -3.06768447e-01 5.53977430e-01
7.89796233e-01 4.12986130e-01 -1.81051409e+00 -2.80082047e-01
1.73284125e-03 -9.78666902e-01 -6.36049509e-01 -5.92718720e-01
-1.15828454e+00 7.44368553e-01 3.63637656e-01 2.94273794e-01
1.17876387e+00 8.10682848e-02 3.96086216e-01 -2.69502521e-01
3.33707213e-01 -1.07621074e+00 2.25770518e-01 -1.91851422e-01
1.09523785e+00 -1.19312966e+00 3.08970958e-01 -6.68514252e-01
-9.51320767e-01 1.12353182e+00 7.26949394e-01 -3.72773170e-01
3.89521539e-01 4.27508235e-01 5.47305010e-02 -2.24371612e-01
-5.28545916e-01 -1.04785353e-01 3.14336956e-01 8.18735003e-01
6.36643022e-02 2.70920157e-01 1.14161512e-02 5.73429585e-01
-4.22307730e-01 -5.01331925e-01 3.55505019e-01 5.06181002e-01
-2.35346481e-01 -9.38421249e-01 -5.64295053e-01 4.63328034e-01
-1.00382492e-01 -1.46290123e-01 -6.93670630e-01 8.42787623e-01
5.05400538e-01 5.02666235e-01 3.16352755e-01 -5.33689857e-01
1.43168598e-01 3.70939597e-02 4.58006501e-01 -4.45930451e-01
-5.00681281e-01 2.07716793e-01 -4.33680564e-01 -5.67466319e-01
-4.68746483e-01 -5.95032215e-01 -7.45171189e-01 -4.93495226e-01
-1.96907178e-01 -3.22903633e-01 3.16161841e-01 4.54965740e-01
2.46772081e-01 6.15066826e-01 7.47764409e-01 -9.26984906e-01
-3.21140915e-01 -8.47635984e-01 -8.46369624e-01 9.94817138e-01
3.09034586e-01 -5.41570902e-01 -5.23915708e-01 5.34912944e-01]
|
[11.364068984985352, -0.8224357962608337]
|
3403481d-dfb7-4680-a7e6-c73b84552789
|
temporally-layered-architecture-for-efficient
|
2305.18701
| null |
https://arxiv.org/abs/2305.18701v1
|
https://arxiv.org/pdf/2305.18701v1.pdf
|
Temporally Layered Architecture for Efficient Continuous Control
|
We present a temporally layered architecture (TLA) for temporally adaptive control with minimal energy expenditure. The TLA layers a fast and a slow policy together to achieve temporal abstraction that allows each layer to focus on a different time scale. Our design draws on the energy-saving mechanism of the human brain, which executes actions at different timescales depending on the environment's demands. We demonstrate that beyond energy saving, TLA provides many additional advantages, including persistent exploration, fewer required decisions, reduced jerk, and increased action repetition. We evaluate our method on a suite of continuous control tasks and demonstrate the significant advantages of TLA over existing methods when measured over multiple important metrics. We also introduce a multi-objective score to qualitatively assess continuous control policies and demonstrate a significantly better score for TLA. Our training algorithm uses minimal communication between the slow and fast layers to train both policies simultaneously, making it viable for future applications in distributed control.
|
['Hava Siegelmann', 'Terrence Sejnowski', 'Devdhar Patel']
|
2023-05-30
| null | null | null | null |
['continuous-control']
|
['playing-games']
|
[ 6.16022479e-03 -2.06746131e-01 -5.34229755e-01 1.21386513e-01
-4.41948295e-01 -5.97487330e-01 7.04349875e-01 6.87201396e-02
-5.06552517e-01 8.01172793e-01 2.75353193e-01 -3.37283343e-01
-3.00700605e-01 -4.61501658e-01 -6.51215255e-01 -7.53005743e-01
-5.98120630e-01 9.30312555e-03 1.90154344e-01 -6.78109080e-02
1.76378384e-01 5.18281698e-01 -1.26756167e+00 -3.54799293e-02
5.07873237e-01 1.05273664e+00 3.58712822e-01 8.72779489e-01
4.88068849e-01 1.04996562e+00 -3.17064464e-01 6.94606543e-01
4.62809473e-01 -4.94688004e-01 -6.56665623e-01 -1.89835340e-01
-8.90486389e-02 -3.68088245e-01 -1.70456946e-01 4.47455227e-01
4.80154842e-01 7.04011798e-01 3.53318512e-01 -1.14749992e+00
-2.39662558e-01 4.01512116e-01 -4.05051529e-01 2.95554012e-01
6.64500967e-02 8.36903572e-01 8.60305130e-01 -3.60962808e-01
3.53730708e-01 1.30349052e+00 6.55041575e-01 7.98361242e-01
-1.48646796e+00 -6.21385992e-01 6.52129591e-01 2.30248496e-02
-8.68197501e-01 -8.36962998e-01 4.60761994e-01 -1.33125350e-01
1.56297684e+00 9.43713337e-02 9.45817590e-01 1.03756177e+00
5.61513603e-01 7.99493670e-01 1.21369791e+00 -1.87421158e-01
9.30178225e-01 -5.86583853e-01 -4.34420675e-01 7.65134513e-01
6.45803809e-02 6.29914105e-01 -6.67050302e-01 -6.42162338e-02
8.92309904e-01 -1.77106425e-01 -8.51897821e-02 -6.85110331e-01
-1.31794679e+00 3.63623798e-01 3.12912285e-01 -6.80174306e-02
-7.12467551e-01 1.06304324e+00 5.04515171e-01 2.61943907e-01
-6.13277927e-02 9.79848385e-01 -5.77006102e-01 -5.10593832e-01
-6.42122924e-01 2.66546071e-01 6.24695003e-01 8.45294297e-01
2.96414465e-01 4.13369745e-01 -4.08958226e-01 3.60919774e-01
1.03328250e-01 2.89095730e-01 4.38470751e-01 -1.77711093e+00
2.14314058e-01 2.94773519e-01 5.51048577e-01 -5.41486323e-01
-4.72874075e-01 -2.27112338e-01 -6.92562521e-01 9.70358253e-01
6.63467422e-02 -4.03370976e-01 -9.43560898e-01 2.20898414e+00
2.58769542e-01 9.89863649e-02 -1.62546225e-02 7.82518983e-01
-5.24275780e-01 7.30857491e-01 3.10729831e-01 -3.63771379e-01
8.71618688e-01 -1.16771197e+00 -8.09219837e-01 -4.39354479e-01
3.33500206e-01 -3.77866216e-02 1.18578815e+00 3.53172451e-01
-1.42056513e+00 -3.22649390e-01 -1.15201294e+00 2.19566911e-01
-2.00022683e-01 -1.12027377e-01 8.22483838e-01 1.98322713e-01
-1.33812833e+00 1.01940966e+00 -1.66569269e+00 -3.53641868e-01
3.25086355e-01 4.44008410e-01 1.15966916e-01 4.77667093e-01
-7.60484457e-01 1.15081847e+00 3.99653345e-01 -1.31452695e-01
-1.44566953e+00 -7.19404101e-01 -7.16775894e-01 3.64123136e-01
6.64079070e-01 -8.66813242e-01 1.92756093e+00 -5.90121984e-01
-2.13286257e+00 8.07528570e-02 -2.97173043e-03 -7.29647815e-01
6.10473514e-01 -2.63256341e-01 4.15637866e-02 9.53209102e-02
-1.30052030e-01 8.36155653e-01 6.02088094e-01 -8.17333341e-01
-7.08194733e-01 4.07036468e-02 1.86556384e-01 4.41204160e-01
-4.39106196e-01 -1.78762689e-01 -1.20424099e-01 -6.04117453e-01
-4.49308962e-01 -1.04130828e+00 -6.03039265e-01 3.58937591e-01
1.03904419e-01 -2.37109676e-01 7.93256342e-01 -2.03109682e-01
1.34231758e+00 -1.86543238e+00 4.44585860e-01 -1.48343623e-01
1.25885382e-01 2.57808417e-02 -1.80313602e-01 5.42171478e-01
3.24367732e-02 2.66299158e-01 -2.86367595e-01 -2.95869708e-01
1.44748688e-01 2.14770436e-01 -2.20669255e-01 2.44654343e-01
1.37600362e-01 8.81778777e-01 -1.13564694e+00 -2.80986935e-01
2.65045017e-01 1.81130707e-01 -5.66841185e-01 1.55968472e-01
-7.40701258e-01 5.19137979e-01 -6.83462977e-01 4.78548795e-01
-1.63147599e-01 -3.14408243e-01 5.39979517e-01 2.98874378e-01
-6.00222468e-01 3.68766576e-01 -8.55051935e-01 1.66188741e+00
-7.61385381e-01 5.21984696e-01 5.47414720e-01 -6.67038262e-01
3.05923134e-01 5.11542737e-01 8.01612258e-01 -1.11186099e+00
6.75213113e-02 2.41130751e-04 -1.60599306e-01 -2.95335799e-01
3.17877769e-01 2.25371987e-01 -9.05773640e-02 7.68353164e-01
-2.89583623e-01 -1.37205109e-01 2.41499931e-01 -3.31887626e-03
1.57432568e+00 6.14284456e-01 4.53113139e-01 -5.78545630e-01
-2.72803325e-02 8.37729201e-02 7.73066282e-01 6.98157668e-01
-6.47466898e-01 -2.39087358e-01 4.90905970e-01 -4.79173690e-01
-1.15840685e+00 -8.79649401e-01 4.98224556e-01 1.38225985e+00
4.62434776e-02 -4.34512585e-01 -4.25565600e-01 -3.08387101e-01
1.36358276e-01 8.43298852e-01 -7.03714550e-01 -2.55852818e-01
-8.92540276e-01 -2.80088246e-01 1.45813823e-01 9.84264851e-01
5.45689762e-01 -1.17617714e+00 -1.67872763e+00 2.76490450e-01
1.41418368e-01 -7.77789712e-01 -8.51146877e-01 5.02839684e-01
-1.03619885e+00 -9.38418448e-01 -3.37888271e-01 -5.18345952e-01
4.44937766e-01 3.05330873e-01 1.07149303e+00 -1.80213943e-01
-2.63683856e-01 6.46490753e-01 1.33524865e-01 -1.79222345e-01
-2.89359897e-01 -1.61648005e-01 3.78189892e-01 -6.21322453e-01
-3.10057074e-01 -7.76703715e-01 -9.20410693e-01 3.43532354e-01
-5.81103086e-01 3.49641144e-01 5.43935239e-01 5.97640932e-01
6.40014946e-01 1.20803326e-01 4.23892498e-01 -5.08785844e-02
8.53893280e-01 -3.71768087e-01 -9.27754462e-01 6.69586882e-02
-8.29032779e-01 4.27619129e-01 7.63554633e-01 -6.37319922e-01
-9.85048473e-01 2.48320490e-01 5.09220779e-01 -6.83809340e-01
3.84033859e-01 1.16581045e-01 1.71497732e-01 -7.85798728e-02
4.12631392e-01 2.20362827e-01 1.04085892e-01 -2.39916921e-01
3.85470390e-01 -6.00867085e-02 3.08386952e-01 -9.34995472e-01
2.63507813e-01 2.78518647e-01 1.50933951e-01 -2.90362835e-01
-2.55609691e-01 -3.75196547e-03 -2.47561902e-01 -3.24249119e-01
8.94376576e-01 -7.90734828e-01 -1.30539644e+00 3.06390077e-01
-7.56708026e-01 -1.28158200e+00 -4.94254798e-01 3.81575644e-01
-1.04967511e+00 -2.13275224e-01 -5.45541584e-01 -9.80767071e-01
-3.72654110e-01 -9.54056978e-01 7.31331885e-01 3.84703994e-01
-4.02340204e-01 -8.94252062e-01 3.94735336e-01 -2.89475530e-01
8.45630646e-01 4.14211541e-01 9.20675457e-01 -7.45587498e-02
-6.60400689e-01 4.23262984e-01 2.49966934e-01 8.80426913e-02
1.79652900e-01 -1.78834736e-01 -5.86963892e-01 -8.07753801e-01
-1.10646285e-01 -6.69026732e-01 8.53943169e-01 4.40822423e-01
1.37377334e+00 -6.29448652e-01 -5.23742259e-01 4.36214536e-01
1.37958717e+00 7.07371116e-01 1.16470963e-01 5.13689160e-01
2.66626805e-01 1.15366533e-01 5.58941007e-01 5.78122318e-01
2.67150581e-01 4.87663418e-01 5.28977334e-01 8.58580228e-03
-5.70904128e-02 -1.04589894e-01 7.02339947e-01 5.81220746e-01
-2.02877894e-01 -2.83977896e-01 -7.81218171e-01 8.41173470e-01
-2.12195730e+00 -1.02669263e+00 8.56084406e-01 2.31955600e+00
9.48739350e-01 2.45017782e-01 2.40486488e-01 -2.58754253e-01
2.68467724e-01 4.26089734e-01 -1.34302855e+00 -6.37736976e-01
4.65808213e-01 3.03621590e-01 6.03993297e-01 5.61861575e-01
-1.12916481e+00 7.66563773e-01 8.01600647e+00 3.62433225e-01
-1.06716180e+00 -1.26642525e-01 5.69324851e-01 -8.54701281e-01
1.87623966e-02 -1.47100687e-01 -2.64232457e-01 4.76125002e-01
1.24225271e+00 -5.10063231e-01 9.67395008e-01 9.02258873e-01
7.70083129e-01 -1.00931503e-01 -1.42367661e+00 4.34739381e-01
-5.69579601e-01 -1.40419424e+00 -4.24194574e-01 1.17550984e-01
9.08136666e-01 9.85686928e-02 1.03666395e-01 5.05478859e-01
1.03160763e+00 -1.10175669e+00 7.56991208e-01 4.25029576e-01
6.55424953e-01 -6.90981567e-01 -1.12086385e-01 2.66951710e-01
-1.54220533e+00 -5.39098084e-01 -1.59184390e-03 -2.96341985e-01
5.95168471e-02 -4.84279767e-02 -5.14838278e-01 5.68747036e-02
7.33981431e-01 5.63496947e-01 -6.42937347e-02 7.99080074e-01
-7.96297342e-02 4.08064425e-01 -4.54526782e-01 -3.02993983e-01
4.90730047e-01 -2.08420982e-03 5.77315271e-01 9.90771890e-01
1.97456196e-01 2.37768754e-01 4.33232337e-01 9.17625308e-01
8.01610276e-02 -5.49529970e-01 -6.11514688e-01 -2.85995156e-01
7.58478940e-01 9.51976240e-01 -6.15144372e-01 -2.63211519e-01
-7.66924769e-02 7.60259092e-01 3.74941915e-01 4.18618917e-01
-9.42222774e-01 -2.43920013e-01 1.11746764e+00 -3.27127934e-01
3.13798398e-01 -9.60740209e-01 -1.31614298e-01 -7.48249054e-01
-2.66648848e-02 -8.00148964e-01 2.38081351e-01 -6.81690693e-01
-6.80190146e-01 1.61416888e-01 8.66325498e-02 -1.07894516e+00
-5.23314714e-01 -2.95376569e-01 -6.81084871e-01 5.56459785e-01
-1.37849581e+00 -5.55373490e-01 -1.34982929e-01 4.00301814e-01
8.75575662e-01 2.00998783e-01 8.75326812e-01 -1.70930121e-02
-8.57875943e-01 2.40002185e-01 1.39398381e-01 -3.72377843e-01
4.25753146e-01 -1.44984329e+00 5.85396171e-01 8.41319025e-01
-6.85138702e-01 5.39058924e-01 6.93078578e-01 -5.06239712e-01
-1.82769871e+00 -1.11403894e+00 1.77323520e-01 -2.73708522e-01
6.17541671e-01 -2.27515072e-01 -5.61699331e-01 7.98998773e-01
4.58071262e-01 -1.78812385e-01 2.02853739e-01 -1.10039860e-02
3.44799496e-02 -9.23208594e-02 -9.54536557e-01 7.99558342e-01
1.13621116e+00 -2.47647762e-01 -2.97787577e-01 2.32291251e-01
8.45029652e-01 -3.91941965e-01 -7.96821713e-01 2.54180312e-01
6.87667310e-01 -9.02731121e-01 8.18982184e-01 -7.54047692e-01
2.82635033e-01 -2.76707709e-01 -5.71323782e-02 -1.47839034e+00
-7.16871321e-01 -1.03672171e+00 -7.61531949e-01 5.27310312e-01
2.32655078e-01 -6.35874212e-01 6.58175766e-01 8.26924980e-01
-2.34234363e-01 -1.00566888e+00 -9.81891036e-01 -1.05112696e+00
3.99662293e-02 -1.10437401e-01 3.49102736e-01 7.33127713e-01
2.09698915e-01 4.40867245e-02 -2.93636024e-01 1.86800763e-01
4.29610550e-01 1.28684133e-01 1.79552913e-01 -6.25877857e-01
-3.52755904e-01 -9.02647614e-01 2.07492828e-01 -9.51842368e-01
-1.12909324e-01 -3.92340690e-01 4.23128575e-01 -1.49647093e+00
1.04997888e-01 -4.06882077e-01 -7.29859829e-01 1.05421388e+00
7.49665946e-02 -4.15404260e-01 3.38635027e-01 1.63395911e-01
-8.69908571e-01 9.46247876e-01 1.22238326e+00 -4.23523784e-02
-6.03578329e-01 -3.01878452e-01 -5.23632228e-01 5.58571279e-01
1.04401219e+00 -1.94878191e-01 -8.41165721e-01 -6.67451620e-01
-4.26547155e-02 2.62303382e-01 1.27681792e-01 -1.38701391e+00
3.94064754e-01 -8.87340903e-01 3.15959632e-01 1.83737129e-02
4.20910954e-01 -7.06142187e-01 6.14950433e-02 9.95888948e-01
-6.55524492e-01 5.08876204e-01 5.72641015e-01 8.66581202e-01
4.09984112e-01 6.93781316e-01 1.04496908e+00 -3.13638210e-01
-8.33044767e-01 2.47791782e-01 -7.97917783e-01 -1.06591463e-01
1.34590387e+00 -1.24854364e-01 -1.94532201e-01 -1.88535705e-01
-6.07008338e-01 9.00415957e-01 5.54646015e-01 4.90808308e-01
3.02285373e-01 -1.24968922e+00 -1.66379973e-01 -1.08106703e-01
-3.28349024e-01 -1.81542411e-01 -5.26379496e-02 7.14544773e-01
-4.04168427e-01 5.73252738e-01 -5.22635937e-01 -3.34729910e-01
-8.83876264e-01 5.96639216e-01 7.66291380e-01 -4.65053022e-01
-8.16732764e-01 5.47194719e-01 -1.39390025e-02 -2.94504110e-02
5.17791748e-01 -6.53441966e-01 2.69406199e-01 -3.72471064e-01
4.90444928e-01 7.19307899e-01 -2.63188004e-01 3.51496577e-01
-5.17007232e-01 4.36041385e-01 2.73709446e-01 -4.22967017e-01
1.40574718e+00 -9.74899158e-02 1.24836177e-01 4.07556593e-01
5.48738837e-01 -4.69174504e-01 -2.14616823e+00 6.79689646e-02
-9.42347273e-02 -1.54400796e-01 3.78265858e-01 -1.15598261e+00
-8.68268430e-01 5.27194023e-01 6.75693929e-01 1.89833835e-01
1.33187580e+00 -5.23410439e-01 7.24013090e-01 6.49747431e-01
5.69422007e-01 -1.50905871e+00 3.86387378e-01 5.56573808e-01
9.15521383e-01 -9.40746963e-01 4.48906422e-02 3.60618174e-01
-6.48307204e-01 9.22291398e-01 8.30731094e-01 -9.45317671e-02
2.63015181e-01 4.77851093e-01 -2.77788997e-01 5.09708151e-02
-1.63983512e+00 -2.71160007e-02 -3.42917472e-01 3.96343976e-01
9.57227722e-02 6.68782294e-02 -8.75387639e-02 7.96331652e-03
3.28642547e-01 1.71686411e-01 2.82492518e-01 1.46036363e+00
-6.14489257e-01 -7.67602682e-01 6.41466901e-02 1.30433679e-01
-1.89936161e-01 3.51348072e-01 -2.34799176e-01 8.22008491e-01
-3.69807899e-01 8.74136925e-01 1.64302871e-01 -3.72307360e-01
1.49369523e-01 2.75351759e-02 5.00141442e-01 -3.27231258e-01
-4.66826826e-01 -9.62500423e-02 1.00557491e-01 -1.50972462e+00
-3.15999597e-01 -5.53989053e-01 -1.49543333e+00 -4.51212168e-01
2.79309481e-01 -7.52136409e-02 6.85233295e-01 6.48612082e-01
8.09248269e-01 1.06703424e+00 7.38201320e-01 -1.23937309e+00
-7.97538459e-01 -6.04860783e-01 -8.11627731e-02 -1.36436284e-01
8.52165401e-01 -6.24142885e-01 -1.88751340e-01 -5.69132157e-02]
|
[4.233820915222168, 1.6537929773330688]
|
d6e4956e-1560-49a1-b017-473a245c4811
|
how-much-and-when-do-we-need-higher-order
|
2001.11181
| null |
https://arxiv.org/abs/2001.11181v3
|
https://arxiv.org/pdf/2001.11181v3.pdf
|
How Much and When Do We Need Higher-order Information in Hypergraphs? A Case Study on Hyperedge Prediction
|
Hypergraphs provide a natural way of representing group relations, whose complexity motivates an extensive array of prior work to adopt some form of abstraction and simplification of higher-order interactions. However, the following question has yet to be addressed: How much abstraction of group interactions is sufficient in solving a hypergraph task, and how different such results become across datasets? This question, if properly answered, provides a useful engineering guideline on how to trade off between complexity and accuracy of solving a downstream task. To this end, we propose a method of incrementally representing group interactions using a notion of n-projected graph whose accumulation contains information on up to n-way interactions, and quantify the accuracy of solving a task as n grows for various datasets. As a downstream task, we consider hyperedge prediction, an extension of link prediction, which is a canonical task for evaluating graph models. Through experiments on 15 real-world datasets, we draw the following messages: (a) Diminishing returns: small n is enough to achieve accuracy comparable with near-perfect approximations, (b) Troubleshooter: as the task becomes more challenging, larger n brings more benefit, and (c) Irreducibility: datasets whose pairwise interactions do not tell much about higher-order interactions lose much accuracy when reduced to pairwise abstractions.
|
['HyungSeok Song', 'Se-eun Yoon', 'Yung Yi', 'Kijung Shin']
|
2020-01-30
| null | null | null | null |
['hyperedge-prediction']
|
['graphs']
|
[ 2.55963147e-01 4.88836169e-01 -2.40136191e-01 -2.65785992e-01
-3.77884209e-01 -6.86540842e-01 5.95360696e-01 4.06942099e-01
1.54447686e-02 7.92951822e-01 2.90652871e-01 -6.10687256e-01
-5.41822195e-01 -8.98070753e-01 -8.23887587e-01 -4.33117270e-01
-6.28263414e-01 6.70659006e-01 3.50799143e-01 -3.51350635e-01
8.55543017e-02 4.69455421e-01 -1.35569227e+00 1.19765945e-01
9.75111723e-01 5.35381854e-01 -3.48384082e-01 5.87435067e-01
3.76274623e-02 6.01027846e-01 -3.45378399e-01 -7.54901171e-01
4.13816422e-01 -5.29563785e-01 -1.21176803e+00 9.31538641e-02
4.16150093e-01 5.85666969e-02 -3.22432607e-01 9.37933207e-01
1.43675506e-01 1.47344181e-02 6.05806828e-01 -1.49734569e+00
-3.92378747e-01 1.03565717e+00 -4.51980054e-01 1.48483336e-01
5.21598220e-01 2.75668532e-01 1.54644752e+00 -3.90024692e-01
6.43649161e-01 1.29999673e+00 8.70429993e-01 -9.96974390e-03
-1.61084437e+00 -3.09853524e-01 3.28055054e-01 -9.46295559e-02
-1.53013384e+00 -8.04616287e-02 4.34656531e-01 -4.50185508e-01
9.36923265e-01 7.02890933e-01 6.62155032e-01 5.64792216e-01
1.79050341e-02 3.03077161e-01 8.17696929e-01 -2.60800868e-01
-3.99577022e-02 5.23320623e-02 6.49265230e-01 6.41875863e-01
8.69178772e-01 -9.12516713e-02 -3.72530967e-01 -4.90444154e-01
3.45357448e-01 -2.25844041e-01 -4.69520628e-01 -5.27894080e-01
-8.53781939e-01 7.50017226e-01 5.81268668e-01 2.09378719e-01
-3.20251361e-02 2.11101174e-01 2.69429892e-01 6.12628102e-01
3.34921211e-01 7.65002549e-01 -3.54394764e-01 1.25805279e-02
-4.62747127e-01 5.92552900e-01 1.17347336e+00 1.01713181e+00
9.63776946e-01 -4.63065058e-01 1.42310455e-01 6.37114286e-01
-1.05347216e-01 -1.14483744e-01 -1.93280324e-01 -8.11648309e-01
5.11502802e-01 9.55723166e-01 -3.17873731e-02 -1.13915575e+00
-5.69289863e-01 -5.15021503e-01 -7.97502339e-01 -8.72837529e-02
7.10756063e-01 -4.90709357e-02 -5.59196830e-01 2.03552723e+00
2.85120517e-01 1.58037338e-02 -2.54386961e-01 7.53847122e-01
5.68172038e-01 3.47084671e-01 9.30081308e-03 -2.64172554e-01
1.30088472e+00 -7.41250515e-01 -2.92281330e-01 -2.69289523e-01
1.19914150e+00 -4.80506092e-01 1.04987228e+00 2.41767332e-01
-1.14367759e+00 -2.39172235e-01 -1.09382236e+00 -1.86928064e-01
-4.32372510e-01 -6.60812497e-01 1.06320405e+00 6.70805693e-01
-1.20421708e+00 9.34369743e-01 -5.81933975e-01 -5.32046616e-01
6.98517784e-02 6.14571810e-01 -5.26060104e-01 -2.33346060e-01
-1.22369266e+00 9.87696350e-01 4.04288560e-01 -8.52721483e-02
-1.26472995e-01 -9.28760827e-01 -7.09568143e-01 3.08330595e-01
6.67005956e-01 -9.74614203e-01 7.32479632e-01 -6.47402823e-01
-7.17840910e-01 6.39126301e-01 -1.48236215e-01 -3.79836798e-01
3.32840621e-01 1.39473289e-01 -3.43488127e-01 -9.12439525e-02
-1.87361941e-01 3.07587326e-01 1.17647670e-01 -1.47655952e+00
-3.56099159e-01 -6.06841743e-01 5.93137801e-01 3.40934247e-01
-4.00510520e-01 -1.82026386e-01 -6.26451313e-01 -4.24408674e-01
1.97947457e-01 -1.36384845e+00 -3.30882192e-01 -1.31098703e-01
-7.15187192e-01 -3.40920150e-01 3.82046431e-01 -3.79723042e-01
1.47214377e+00 -1.84395683e+00 3.83484125e-01 6.37151837e-01
7.47708857e-01 9.13377181e-02 -2.69042313e-01 9.01127458e-01
-2.90552318e-01 5.71867466e-01 -3.33917677e-01 4.95470427e-02
6.83008879e-02 2.54755974e-01 -2.10659787e-01 2.76447058e-01
1.20202221e-01 9.32885051e-01 -8.86142254e-01 -3.66114765e-01
-1.82708248e-01 2.44598627e-01 -7.82133341e-01 -1.14999279e-01
-1.59146026e-01 2.53916141e-02 -3.83376539e-01 2.24970222e-01
6.40505135e-01 -7.38954067e-01 6.78464055e-01 -2.41675183e-01
2.77911782e-01 3.22045326e-01 -1.27319336e+00 1.08462083e+00
1.46803856e-02 4.66647774e-01 -1.01878308e-01 -9.05707538e-01
5.93488932e-01 -1.72882438e-01 3.94954950e-01 -3.00062627e-01
-7.60749206e-02 2.98032518e-02 3.84252638e-01 -3.07071000e-01
4.47952032e-01 3.27127874e-02 -4.01434638e-02 6.28518522e-01
-3.36753309e-01 -4.13733236e-02 5.02619922e-01 6.27255380e-01
1.76402354e+00 -4.18842912e-01 3.95383090e-01 -4.82695580e-01
1.82408318e-01 2.56593917e-02 4.04237419e-01 6.50234580e-01
2.42267311e-01 3.71437371e-01 1.07663417e+00 -5.45704901e-01
-9.01223004e-01 -8.72115195e-01 -8.66840184e-02 8.72397840e-01
2.35883445e-01 -9.48052347e-01 -7.49238610e-01 -6.55253828e-01
4.00248587e-01 3.95154208e-01 -6.77487373e-01 -2.57840335e-01
-4.77519751e-01 -9.67018187e-01 3.13101113e-01 4.53412592e-01
-3.00852228e-02 -4.47486192e-01 -7.92007148e-02 -8.92342031e-02
-1.12568341e-01 -1.23902404e+00 -4.39342856e-01 1.53866038e-01
-9.48705673e-01 -1.41414475e+00 -2.35334948e-01 -4.78048176e-01
8.46259832e-01 4.46962804e-01 1.38737512e+00 7.11643755e-01
-3.91842052e-02 9.66473892e-02 -2.21966594e-01 -5.93391098e-02
-3.17695022e-01 2.15653062e-01 -1.85413122e-01 -3.44368875e-01
3.25973570e-01 -9.73003328e-01 -4.98281032e-01 4.02601868e-01
-7.97889173e-01 -1.44025803e-01 6.51982546e-01 6.27281427e-01
2.41483346e-01 2.55072564e-01 1.46651447e-01 -1.58222604e+00
9.27848339e-01 -4.73031491e-01 -4.21989143e-01 3.54865372e-01
-7.19429255e-01 3.98148060e-01 7.83492923e-01 -1.64730042e-01
-4.75082576e-01 -1.30678073e-01 1.48520306e-01 -9.73801240e-02
3.45748186e-01 8.09979439e-01 -1.69504941e-01 -2.55612552e-01
7.66419947e-01 -2.43158415e-01 -2.98562665e-02 -3.51485938e-01
3.95541966e-01 1.99135050e-01 3.29897225e-01 -7.47589469e-01
9.33107615e-01 3.77880394e-01 4.88539249e-01 -7.57306397e-01
-7.25304842e-01 -3.74113113e-01 -6.11730695e-01 7.63209015e-02
4.21381742e-01 -5.71800470e-01 -9.89683867e-01 -5.63445278e-02
-8.64038467e-01 -4.77412522e-01 -4.72214408e-02 1.14109777e-01
-3.73618841e-01 6.96688175e-01 -5.05092561e-01 -5.45484960e-01
6.91724494e-02 -9.55426931e-01 7.67610967e-01 -2.51690239e-01
-6.54033899e-01 -9.27147448e-01 2.78386362e-02 3.79692137e-01
8.53816979e-03 3.71731699e-01 1.52461219e+00 -6.67875767e-01
-8.82846177e-01 -1.97507471e-01 -3.17120880e-01 -4.79311794e-02
-4.82788272e-02 2.10890174e-02 -6.05592608e-01 -3.79260808e-01
-5.03212035e-01 -1.52327552e-01 7.13930190e-01 1.10189654e-01
1.19462645e+00 -4.70572710e-01 -6.54344559e-01 5.86455524e-01
1.32997429e+00 -2.58773446e-01 6.68839753e-01 -1.79143488e-01
8.49896669e-01 8.69132817e-01 3.56564164e-01 2.56506115e-01
6.42097235e-01 8.77016842e-01 2.20614567e-01 -8.76330584e-02
-1.79280654e-01 -5.13380647e-01 -2.71940473e-02 8.46840143e-01
-3.28109503e-01 -4.41188544e-01 -1.15000153e+00 1.94214299e-01
-1.90493917e+00 -7.72828579e-01 -5.32718241e-01 2.39682889e+00
7.22504139e-01 4.31128055e-01 5.36860943e-01 3.21208119e-01
5.55667579e-01 1.28372774e-01 -4.75182354e-01 -2.54458964e-01
-4.82689999e-02 -1.82636250e-02 5.39659619e-01 6.43906653e-01
-6.62249029e-01 6.55386448e-01 6.88115025e+00 6.62545800e-01
-6.50866032e-01 -2.65529692e-01 7.24550903e-01 1.30951628e-01
-7.02480495e-01 5.64075172e-01 -5.49207807e-01 4.08375442e-01
9.42404151e-01 -4.98921782e-01 6.67653024e-01 4.67738152e-01
-1.45432934e-01 -8.16226527e-02 -1.63845181e+00 7.09191918e-01
-1.49270952e-01 -1.14013481e+00 1.59288138e-01 4.03488219e-01
6.25866115e-01 -2.68751591e-01 -2.84497261e-01 2.70175606e-01
8.48809004e-01 -1.20026100e+00 1.07996419e-01 4.58763361e-01
4.43392247e-01 -7.12416291e-01 4.71768320e-01 4.13217634e-01
-1.28778327e+00 -3.27240266e-02 -2.63414115e-01 -3.33683014e-01
6.88370615e-02 7.33273566e-01 -7.18365490e-01 8.02382112e-01
4.02284831e-01 3.55433285e-01 -6.87048137e-01 8.64265442e-01
-8.62949267e-02 4.76425618e-01 -5.17801046e-01 -1.53492577e-02
1.84741486e-02 -2.49358803e-01 4.85629708e-01 9.71280694e-01
5.43814078e-02 5.01747727e-01 1.75350010e-01 6.02685571e-01
-2.97905326e-01 2.90383771e-02 -8.11206520e-01 -1.76752701e-01
6.68011129e-01 1.09770870e+00 -8.53111029e-01 -1.52030904e-02
-3.54930729e-01 4.26334888e-01 7.59119987e-01 4.49174374e-01
-5.28371811e-01 -2.07895160e-01 6.99329734e-01 5.74873567e-01
-1.16357477e-02 -1.52400523e-01 -4.52118605e-01 -9.68328595e-01
1.37465164e-01 -9.31870699e-01 7.43643284e-01 -7.05255806e-01
-1.33430696e+00 4.13621128e-01 1.20450854e-01 -1.01669991e+00
-2.46364534e-01 -4.98837680e-01 -5.16625524e-01 6.67418897e-01
-1.06987917e+00 -8.40128660e-01 -4.13708657e-01 1.15528971e-01
-1.95644379e-01 3.83425385e-01 7.39651501e-01 2.78153300e-01
-3.93314898e-01 7.56490529e-01 -2.56217480e-01 -1.22423850e-01
4.53822494e-01 -1.25408292e+00 4.44296449e-01 6.75410628e-01
6.04626499e-02 7.94355035e-01 1.06733692e+00 -6.03310347e-01
-1.70326626e+00 -8.99048388e-01 9.77018952e-01 -8.28562558e-01
6.69948041e-01 -3.53040010e-01 -1.10881293e+00 8.89594555e-01
-2.34079838e-01 1.18592493e-01 6.06386125e-01 8.62519264e-01
-4.58137929e-01 -1.35981828e-01 -8.78543377e-01 8.79661798e-01
1.72342408e+00 -4.88960028e-01 -1.91438451e-01 4.55745399e-01
8.59933019e-01 -1.56180456e-01 -1.24142456e+00 5.76691091e-01
4.16381359e-01 -9.66475904e-01 9.76443350e-01 -8.19083571e-01
5.01528800e-01 -1.52911380e-01 -6.01195060e-02 -1.25182462e+00
-4.60162759e-01 -8.78168643e-01 -8.45340490e-02 1.03544998e+00
6.67547643e-01 -6.97036207e-01 1.01549828e+00 9.94698763e-01
8.07103813e-02 -1.07357144e+00 -4.10413831e-01 -8.79850090e-01
7.70877674e-02 -1.91066042e-01 7.98975527e-01 1.21415615e+00
3.90528411e-01 6.34204268e-01 -2.19878212e-01 3.03184986e-01
6.80623710e-01 2.10417733e-01 1.18038869e+00 -1.44626391e+00
-4.04252738e-01 -5.24014592e-01 -7.24877477e-01 -1.07430112e+00
1.70637116e-01 -1.14774024e+00 -4.75943863e-01 -1.55430555e+00
5.12615323e-01 -5.92064619e-01 -4.34247814e-02 4.84561801e-01
-3.11582327e-01 1.49647102e-01 1.76806331e-01 2.10623786e-01
-5.63555419e-01 2.74545819e-01 1.13104510e+00 3.20841260e-02
-9.72014368e-02 7.16504781e-03 -1.14308107e+00 6.10021412e-01
3.75953555e-01 -2.15090588e-01 -6.98638439e-01 -2.54633456e-01
6.81163430e-01 3.44318330e-01 2.80152261e-01 -5.91159880e-01
3.69113147e-01 -2.10351497e-01 -1.90490577e-02 -3.94362807e-01
2.97472924e-01 -7.66534984e-01 6.26627505e-01 5.18708110e-01
-4.34740484e-01 1.43314719e-01 5.25974147e-02 6.05488360e-01
-4.44078967e-02 -1.18420413e-02 4.27681386e-01 2.87712514e-02
-3.65805924e-01 4.01739508e-01 1.82788819e-01 3.33667696e-01
9.13356483e-01 -3.42699081e-01 -5.23818791e-01 -5.75912774e-01
-7.12669730e-01 4.28103358e-01 7.17544794e-01 2.95286149e-01
1.79819435e-01 -1.14948070e+00 -5.72653532e-01 -1.36834756e-02
2.39345506e-01 -1.07571483e-01 8.73719081e-02 8.28518808e-01
-2.93123275e-01 3.86885017e-01 1.93787366e-01 -4.77073371e-01
-1.38989198e+00 5.79074562e-01 1.36258468e-01 -4.29609299e-01
-5.70611119e-01 7.48545408e-01 3.91657382e-01 -4.09331769e-01
5.32816313e-02 -2.22382516e-01 8.10003653e-02 -8.28872621e-02
1.57107040e-01 4.33162749e-01 1.74757063e-01 -3.81267577e-01
-2.87922025e-01 4.59100753e-01 -1.93731859e-01 3.66734087e-01
1.45991910e+00 -5.03912754e-02 -2.88413167e-01 2.33517513e-01
1.11195385e+00 5.41209430e-02 -9.13056731e-01 -8.67199376e-02
2.09364995e-01 -5.50451219e-01 -3.35146904e-01 -5.91133416e-01
-7.66553044e-01 5.35238981e-01 -1.39760688e-01 8.66252065e-01
9.62773979e-01 1.98550060e-01 5.28465331e-01 5.34180820e-01
4.18783635e-01 -6.50072098e-01 -2.74424553e-01 3.92158419e-01
8.44702601e-01 -9.54465985e-01 4.42713559e-01 -1.13774443e+00
-4.61606920e-01 7.07908452e-01 5.95161498e-01 -1.07010482e-02
7.47421086e-01 2.70837218e-01 -5.44742644e-01 -5.25587022e-01
-1.09297025e+00 -1.88607946e-01 2.56022692e-01 3.28405589e-01
3.23867202e-01 1.56181782e-01 -3.64340454e-01 4.38177943e-01
-6.27739489e-01 -2.03461617e-01 4.54507709e-01 6.07611656e-01
-2.34115288e-01 -1.21977878e+00 -4.30038339e-03 9.29527283e-01
-1.47781789e-01 -7.23260865e-02 -8.88188004e-01 1.04911518e+00
-2.50110328e-01 7.62417674e-01 -1.17673345e-01 -5.65659225e-01
3.88111293e-01 -2.89010197e-01 5.32732904e-01 -6.55818760e-01
-4.56074506e-01 -4.87673819e-01 6.43905163e-01 -5.66353798e-01
-1.26942426e-01 -4.44258749e-01 -1.10683668e+00 -9.29518104e-01
-3.88750643e-01 2.83968329e-01 1.91028863e-01 9.20120239e-01
5.27961612e-01 3.14638615e-01 4.89346355e-01 -4.86468524e-01
-6.39711797e-01 -5.68397105e-01 -6.05033159e-01 6.42455459e-01
8.02470744e-02 -6.08632267e-01 -6.46501184e-01 -3.66596043e-01]
|
[7.0719122886657715, 5.883218765258789]
|
7081d631-de25-4d87-b09c-758013c0b082
|
attention-based-graph-neural-network-for-semi
|
1803.03735
| null |
http://arxiv.org/abs/1803.03735v1
|
http://arxiv.org/pdf/1803.03735v1.pdf
|
Attention-based Graph Neural Network for Semi-supervised Learning
|
Recently popularized graph neural networks achieve the state-of-the-art
accuracy on a number of standard benchmark datasets for graph-based
semi-supervised learning, improving significantly over existing approaches.
These architectures alternate between a propagation layer that aggregates the
hidden states of the local neighborhood and a fully-connected layer. Perhaps
surprisingly, we show that a linear model, that removes all the intermediate
fully-connected layers, is still able to achieve a performance comparable to
the state-of-the-art models. This significantly reduces the number of
parameters, which is critical for semi-supervised learning where number of
labeled examples are small. This in turn allows a room for designing more
innovative propagation layers. Based on this insight, we propose a novel graph
neural network that removes all the intermediate fully-connected layers, and
replaces the propagation layers with attention mechanisms that respect the
structure of the graph. The attention mechanism allows us to learn a dynamic
and adaptive local summary of the neighborhood to achieve more accurate
predictions. In a number of experiments on benchmark citation networks
datasets, we demonstrate that our approach outperforms competing methods. By
examining the attention weights among neighbors, we show that our model
provides some interesting insights on how neighbors influence each other.
|
['Li-Jia Li', 'Sewoong Oh', 'Kiran K. Thekumparampil', 'Chong Wang']
|
2018-03-10
|
attention-based-graph-neural-network-for-semi-1
|
https://openreview.net/forum?id=rJg4YGWRb
|
https://openreview.net/pdf?id=rJg4YGWRb
|
iclr-2018-1
|
['graph-regression']
|
['graphs']
|
[ 8.31241384e-02 6.84224069e-01 -6.10158682e-01 -5.29270947e-01
-2.09015772e-01 -3.83897811e-01 7.47579992e-01 4.04466540e-01
-1.39577746e-01 6.74306035e-01 2.85789460e-01 -5.57757080e-01
-2.52747148e-01 -1.01570368e+00 -9.70708370e-01 -5.30240595e-01
-4.18018609e-01 5.30604720e-01 5.06803811e-01 -1.45336449e-01
2.61565953e-01 4.10837114e-01 -1.22072577e+00 2.54333556e-01
9.01104689e-01 8.34145606e-01 1.84607450e-02 4.79669094e-01
-3.48197907e-01 1.19130135e+00 -3.12581062e-01 -5.85340977e-01
9.78292227e-02 -3.11697662e-01 -9.62430596e-01 -1.97017595e-01
6.75964832e-01 -5.34345210e-02 -5.99655449e-01 1.09275568e+00
1.52179062e-01 1.94771379e-01 6.32325351e-01 -1.07298470e+00
-1.12618828e+00 1.18824279e+00 -3.70452464e-01 3.19730312e-01
-1.50945053e-01 2.16863491e-02 1.43806553e+00 -5.07922173e-01
7.67466605e-01 1.23203325e+00 8.25032830e-01 2.67374754e-01
-1.30052304e+00 -4.87306237e-01 7.19788849e-01 2.56613106e-01
-1.00538945e+00 -2.79332828e-02 1.06078529e+00 -3.45981658e-01
1.15930820e+00 -6.16208687e-02 6.74708247e-01 9.55197394e-01
3.22649479e-01 5.26238084e-01 7.91480780e-01 -4.58556980e-01
1.19703561e-01 -8.75312462e-03 8.41064453e-01 1.08950794e+00
5.83368003e-01 -2.85181645e-02 -3.71088862e-01 -2.32334375e-01
6.24328792e-01 1.32198915e-01 -2.96773255e-01 -4.86172557e-01
-1.05964398e+00 8.46310079e-01 1.06717360e+00 4.40759301e-01
-4.07244533e-01 3.12720001e-01 9.69529599e-02 3.20549786e-01
7.13951230e-01 6.74322188e-01 -4.48760599e-01 3.99077028e-01
-9.11396086e-01 -3.43901962e-01 8.81697834e-01 7.60987699e-01
9.44720268e-01 1.20382113e-02 -2.74103284e-01 6.17537022e-01
3.95529032e-01 -1.35305777e-01 1.89950481e-01 -8.45595896e-01
4.31963831e-01 1.27801788e+00 -3.68881226e-01 -1.05729997e+00
-4.64153051e-01 -1.02947450e+00 -1.06593859e+00 1.66681707e-01
3.93438339e-01 -1.37261257e-01 -1.03895056e+00 1.84504378e+00
-2.85820588e-02 3.55872750e-01 -1.24127783e-01 5.40664911e-01
8.96753013e-01 5.83613098e-01 1.67131312e-02 -2.93925107e-02
8.75837028e-01 -1.46550632e+00 -4.31568921e-01 -4.43405151e-01
7.17278898e-01 -1.14465103e-01 9.69264388e-01 -7.31601119e-02
-1.00185609e+00 -4.88908052e-01 -1.14071512e+00 5.05170040e-03
-7.66033113e-01 -1.30573750e-01 8.62804413e-01 1.57165647e-01
-1.51392531e+00 1.16039634e+00 -7.96358824e-01 -4.41314936e-01
6.20169938e-01 5.28463840e-01 -3.97102743e-01 6.22553788e-02
-1.25159526e+00 8.63366127e-01 3.28249216e-01 6.99831769e-02
-5.57579875e-01 -5.60116231e-01 -7.80691922e-01 5.15822887e-01
3.03994179e-01 -7.61001110e-01 7.89386988e-01 -1.03829384e+00
-1.28738821e+00 5.79595089e-01 -1.65070757e-01 -6.04112387e-01
2.47667149e-01 2.30265707e-01 -1.81340531e-01 2.60466039e-02
-1.40242919e-01 5.50130486e-01 5.37075996e-01 -1.14692438e+00
-2.71179974e-01 -3.25052917e-01 1.94970936e-01 -2.11324058e-02
-6.62388504e-01 -2.86676764e-01 -6.94377244e-01 -4.54056740e-01
-6.48475438e-02 -9.34095323e-01 -5.26283562e-01 -2.06263244e-01
-6.68409109e-01 -4.47780460e-01 6.15632176e-01 -1.83739603e-01
1.34405482e+00 -1.71505177e+00 2.65711248e-01 4.28354919e-01
7.13523448e-01 2.72684544e-01 -3.14183295e-01 3.78785431e-01
-2.25929335e-01 4.28611308e-01 -9.85410511e-02 -4.77132738e-01
-1.77645341e-01 1.02224737e-01 -1.08859561e-01 2.34472170e-01
2.53214985e-01 1.39647019e+00 -9.44078088e-01 -3.54136080e-01
-9.94631276e-02 5.33552229e-01 -4.02095526e-01 4.60788757e-02
-2.69064158e-01 2.23576084e-01 -4.16124165e-01 3.03873837e-01
2.41783366e-01 -1.19614124e+00 4.27437603e-01 -3.47014479e-02
3.29496354e-01 5.69047153e-01 -8.67369950e-01 1.33844626e+00
-4.86434959e-02 7.08016753e-01 -8.56689587e-02 -1.23825848e+00
9.10979688e-01 -1.41927868e-01 6.35244250e-02 -4.96185571e-01
9.55683365e-02 -9.03705359e-02 1.78577498e-01 1.21832840e-01
1.28468320e-01 4.30537552e-01 3.26552719e-01 7.33504534e-01
2.55328864e-01 4.10279512e-01 2.17227399e-01 6.97652519e-01
1.39488840e+00 -2.67014235e-01 1.42324820e-01 -4.77825105e-01
5.03814220e-01 -1.59477293e-01 3.74850422e-01 1.10823798e+00
-1.40569806e-01 3.56778562e-01 9.46323991e-01 -7.26984978e-01
-7.69839287e-01 -8.71353686e-01 3.00681621e-01 1.25124586e+00
4.03015539e-02 -6.22956455e-01 -7.43217826e-01 -9.78218973e-01
2.48033497e-02 5.32117605e-01 -1.10690165e+00 -3.75753224e-01
-4.97163445e-01 -5.91392934e-01 3.02110314e-01 6.30801558e-01
3.43165070e-01 -1.15534294e+00 1.47088632e-01 1.10853113e-01
3.68117243e-01 -9.68068838e-01 -3.43987584e-01 3.76226157e-01
-1.16709721e+00 -1.09638000e+00 -5.14570713e-01 -8.71305346e-01
9.82505739e-01 2.11466119e-01 1.37676454e+00 4.90428895e-01
2.23380134e-01 1.37489224e-02 -1.16319269e-01 -2.05367535e-01
-3.93235177e-01 6.98070407e-01 -2.97032923e-01 -8.98859054e-02
2.50754774e-01 -7.89520442e-01 -5.05373418e-01 2.35285778e-02
-4.34596628e-01 1.89104065e-01 7.13517904e-01 9.42638695e-01
4.43517745e-01 -8.07618797e-02 6.16877317e-01 -1.43376017e+00
7.96444952e-01 -5.57649434e-01 -6.20004892e-01 4.44409549e-01
-1.12521708e+00 6.28234863e-01 8.91907871e-01 -2.96711475e-01
-5.99169016e-01 -6.43393919e-02 3.21231008e-01 -2.96881467e-01
1.48664817e-01 7.93077648e-01 2.30167434e-01 -3.07828307e-01
5.36450088e-01 7.45150149e-02 -3.63472141e-02 -3.80426973e-01
4.87821907e-01 4.39974874e-01 3.24937612e-01 -3.01803589e-01
6.66960537e-01 2.92715997e-01 1.99069932e-01 -4.91244316e-01
-1.01528645e+00 -1.70957148e-01 -8.55541945e-01 -1.96102917e-01
5.81912637e-01 -6.41832292e-01 -6.21039093e-01 2.82598823e-01
-1.11981356e+00 -6.73537731e-01 -1.11344665e-01 2.52506256e-01
-1.51908785e-01 2.44535655e-01 -9.97406840e-01 -3.66023868e-01
-4.36264545e-01 -1.06110179e+00 6.04900956e-01 2.33561769e-01
-4.43170965e-02 -1.47638142e+00 1.41165569e-01 1.53310463e-01
6.77310169e-01 1.46179780e-01 1.19317698e+00 -1.05268562e+00
-7.27273285e-01 -1.78314999e-01 -5.04184425e-01 1.41614839e-01
-2.98035201e-02 1.29981056e-01 -7.65803754e-01 -3.37219626e-01
-6.35710895e-01 -9.13164765e-02 1.46927142e+00 4.45188433e-01
1.14452863e+00 -4.77802783e-01 -6.49207890e-01 6.30463123e-01
1.16191542e+00 -2.57193148e-01 3.87340963e-01 2.76640773e-01
1.01729727e+00 4.47307944e-01 -1.61009029e-01 -1.67266190e-01
7.46387482e-01 1.40642300e-01 5.96778035e-01 -3.20250481e-01
-2.17537090e-01 -2.85909146e-01 1.57809615e-01 9.07230437e-01
-2.49386415e-01 -4.49822187e-01 -1.02899253e+00 5.12118220e-01
-2.08759522e+00 -8.86423826e-01 -1.46607190e-01 1.98379350e+00
6.90844476e-01 5.32872498e-01 7.18842000e-02 -2.35988826e-01
9.12561655e-01 5.94329715e-01 -7.76864707e-01 -4.14038539e-01
-7.92996138e-02 2.49814227e-01 6.20088518e-01 6.93762660e-01
-9.80709732e-01 1.16489768e+00 7.22267008e+00 3.95190924e-01
-1.02569413e+00 -2.91275624e-02 8.31044912e-01 1.22265674e-01
-4.42246079e-01 1.48752987e-01 -6.18688881e-01 3.31135452e-01
1.21841919e+00 -4.46822867e-02 6.03783548e-01 8.81230474e-01
-1.42737716e-01 3.72639596e-01 -1.22722256e+00 4.57197905e-01
4.13648337e-02 -1.94493949e+00 3.40106457e-01 1.81102082e-01
9.28460598e-01 7.14883327e-01 -1.41501635e-01 2.83674777e-01
8.31995130e-01 -1.14509487e+00 3.25939685e-01 6.15879297e-01
2.88506806e-01 -5.55769861e-01 7.21577525e-01 3.84730905e-01
-1.12864554e+00 -1.29370421e-01 -3.92396927e-01 -2.78496951e-01
-2.30033159e-01 7.74268746e-01 -6.41471207e-01 4.43457335e-01
6.33516014e-01 9.82610524e-01 -9.57017541e-01 7.98181593e-01
-4.00066346e-01 7.55205333e-01 -1.40027642e-01 -3.39826792e-01
4.60782558e-01 -1.13919228e-01 3.75902027e-01 1.22287476e+00
-1.71387997e-02 -2.92273369e-02 3.36390994e-02 1.13638115e+00
-6.95987403e-01 -7.10391104e-02 -7.15489149e-01 -2.72256553e-01
4.86670166e-01 1.20922124e+00 -8.29677939e-01 -4.88502562e-01
-5.19204617e-01 7.18567193e-01 1.26813698e+00 6.54449284e-01
-5.38142562e-01 -5.07831633e-01 3.16392899e-01 2.57274300e-01
4.77818519e-01 -2.88807988e-01 -1.98814660e-01 -1.08830130e+00
-1.03658274e-01 -4.87830967e-01 5.03715158e-01 -5.45081854e-01
-1.41795671e+00 8.86573911e-01 -3.26595724e-01 -5.94632506e-01
-1.49473503e-01 -7.09625125e-01 -1.09676564e+00 6.71848893e-01
-1.65966928e+00 -1.14181125e+00 -3.09136093e-01 4.51077431e-01
2.37322003e-02 -2.50254333e-01 6.68470740e-01 -3.82644050e-02
-9.08578038e-01 7.57431328e-01 1.13444291e-01 4.63204592e-01
5.49492002e-01 -1.46317720e+00 7.95629144e-01 8.50246668e-01
4.13912803e-01 9.07543540e-01 3.46854568e-01 -6.50055587e-01
-1.21895981e+00 -1.29365098e+00 1.00799453e+00 -3.26152563e-01
8.59555840e-01 -4.05468315e-01 -1.16181457e+00 1.04334962e+00
4.24587190e-01 4.66706753e-01 4.96923000e-01 3.99282932e-01
-5.11369467e-01 -8.16085413e-02 -6.32181466e-01 5.32022536e-01
1.28224230e+00 -5.60295284e-01 -4.98901904e-01 3.86686683e-01
8.87514591e-01 4.68266606e-02 -8.37330103e-01 2.38420889e-01
2.94629753e-01 -1.08189142e+00 7.81978250e-01 -9.63770747e-01
4.62273508e-01 -4.98341210e-02 2.73690760e-01 -1.47956681e+00
-8.20273399e-01 -6.97906971e-01 -6.32569849e-01 8.90955269e-01
8.85981739e-01 -9.95061159e-01 1.03446627e+00 5.34768462e-01
-8.45556557e-02 -1.06511903e+00 -5.02975225e-01 -6.27501309e-01
1.85488388e-01 -8.22159424e-02 5.29924989e-01 1.20615363e+00
2.05903098e-01 7.91900218e-01 -5.93757145e-02 4.27551195e-02
7.30358720e-01 1.49257347e-01 4.68929946e-01 -1.74716973e+00
-2.97249258e-01 -7.54292786e-01 -4.13772523e-01 -1.06349218e+00
6.16473317e-01 -1.23823810e+00 -4.25581366e-01 -2.00307798e+00
3.79004329e-01 -2.81724751e-01 -7.93374002e-01 7.50664353e-01
-3.71912897e-01 9.68516618e-02 -1.77677032e-02 2.79882818e-01
-9.37846959e-01 3.66828084e-01 1.20321548e+00 -3.24502021e-01
-2.54615992e-01 -1.65306211e-01 -1.02076983e+00 7.51465976e-01
7.25797057e-01 -5.14656723e-01 -4.37969685e-01 -4.92683053e-01
3.50727975e-01 -3.58837932e-01 2.02627301e-01 -7.77957737e-01
7.71564901e-01 2.53527127e-02 3.00898165e-01 -3.72372150e-01
-7.85549656e-02 -4.49786812e-01 -1.87288225e-01 4.16949689e-01
-7.00569808e-01 -4.06830646e-02 -5.69427721e-02 8.98204684e-01
-9.75041315e-02 1.47628576e-01 6.47124708e-01 -2.00197726e-01
-4.71252680e-01 5.54387569e-01 -1.29618704e-01 2.91958880e-02
7.95218050e-01 -2.87501141e-02 -7.93150425e-01 -5.59536219e-01
-6.91699564e-01 4.25076723e-01 4.63155627e-01 4.09249187e-01
2.27323070e-01 -1.20571637e+00 -5.51820993e-01 2.32141942e-01
3.08907107e-02 -6.76422119e-02 -1.50638983e-01 7.58404374e-01
-3.91651928e-01 5.07123649e-01 -1.70241594e-02 -4.13589418e-01
-8.12860429e-01 6.44524336e-01 4.96435463e-01 -7.04608262e-01
-5.56020498e-01 7.45365798e-01 -5.35927303e-02 -6.49052143e-01
3.34499300e-01 -4.95432377e-01 -4.01368201e-01 -2.48770475e-01
2.77852595e-01 3.43696207e-01 -1.05143264e-02 -4.46677625e-01
-3.47218335e-01 4.61810648e-01 -2.59354800e-01 3.95028681e-01
1.58632827e+00 8.76438618e-02 -4.00315851e-01 3.65189880e-01
1.16345513e+00 -2.26164088e-01 -1.17363393e+00 -4.94636446e-01
1.34455040e-01 1.41665399e-01 4.32994902e-01 -7.96612263e-01
-1.30507612e+00 9.13014352e-01 2.33730190e-02 5.41145504e-01
7.94388115e-01 2.12810218e-01 5.80721259e-01 7.63109624e-01
1.69575557e-01 -8.85390401e-01 9.88018350e-04 7.39446461e-01
5.97766638e-01 -1.36379504e+00 8.66034403e-02 -2.70142317e-01
-2.86409974e-01 1.15380359e+00 6.36762440e-01 -3.85337859e-01
9.64963436e-01 1.66360572e-01 -1.91599309e-01 -5.53916872e-01
-1.17643762e+00 -4.16771360e-02 5.91828048e-01 3.35559547e-01
5.17661273e-01 -3.17803696e-02 -4.53207828e-03 6.29002869e-01
9.43991318e-02 -3.30675244e-01 4.23432678e-01 5.25184810e-01
-5.98954856e-01 -1.02988100e+00 2.38702402e-01 8.39191616e-01
-3.39388222e-01 -3.86152655e-01 -9.84830141e-01 8.46729398e-01
-3.94107908e-01 7.36563265e-01 1.24629989e-01 -5.33009052e-01
1.57640308e-01 1.61181062e-01 2.09332913e-01 -6.39707208e-01
-6.96895361e-01 -4.02708948e-01 5.65445945e-02 -6.88067257e-01
-3.12905490e-01 -6.19200133e-02 -1.15944684e+00 -4.72648978e-01
-5.87739170e-01 2.82791853e-01 3.40025097e-01 8.12240183e-01
8.68010044e-01 7.38234937e-01 4.41867352e-01 -7.96288967e-01
-4.20258313e-01 -9.68917787e-01 -4.50128824e-01 2.27764308e-01
3.69362831e-01 -5.45593381e-01 -5.64537227e-01 -3.72168094e-01]
|
[6.970187187194824, 6.216841220855713]
|
bd154d7e-7856-413c-8993-100b720751c2
|
a-multi-granularity-matching-attention
|
2303.15870
| null |
https://arxiv.org/abs/2303.15870v1
|
https://arxiv.org/pdf/2303.15870v1.pdf
|
A Multi-Granularity Matching Attention Network for Query Intent Classification in E-commerce Retrieval
|
Query intent classification, which aims at assisting customers to find desired products, has become an essential component of the e-commerce search. Existing query intent classification models either design more exquisite models to enhance the representation learning of queries or explore label-graph and multi-task to facilitate models to learn external information. However, these models cannot capture multi-granularity matching features from queries and categories, which makes them hard to mitigate the gap in the expression between informal queries and categories. This paper proposes a Multi-granularity Matching Attention Network (MMAN), which contains three modules: a self-matching module, a char-level matching module, and a semantic-level matching module to comprehensively extract features from the query and a query-category interaction matrix. In this way, the model can eliminate the difference in expression between queries and categories for query intent classification. We conduct extensive offline and online A/B experiments, and the results show that the MMAN significantly outperforms the strong baselines, which shows the superiority and effectiveness of MMAN. MMAN has been deployed in production and brings great commercial value for our company.
|
['Sulong Xu', 'Songlin Wang', 'Haiqing Hu', 'Mingming Li', 'Yiming Qiu', 'Chunyuan Yuan']
|
2023-03-28
| null | null | null | null |
['intent-classification']
|
['natural-language-processing']
|
[-2.19288543e-02 -2.90014178e-01 -7.83582091e-01 -7.67821312e-01
-6.16878808e-01 -5.07496059e-01 4.60292846e-01 1.62803113e-01
-1.88402295e-01 -1.87828630e-01 3.41558099e-01 -3.36443216e-01
-1.63220644e-01 -7.41767645e-01 -3.05714399e-01 -1.55249581e-01
1.44657806e-01 4.24980134e-01 4.40352224e-02 -4.46952909e-01
2.87973970e-01 3.86488289e-02 -1.44905818e+00 6.02601111e-01
9.47559357e-01 1.49633956e+00 1.66130990e-01 1.14963323e-01
-5.97898483e-01 7.94568062e-01 -2.40020052e-01 -6.11097157e-01
7.45133311e-02 -3.36667210e-01 -6.58742070e-01 -9.26573351e-02
1.99116006e-01 -3.04951549e-01 -5.78170240e-01 9.90946233e-01
4.03536052e-01 1.88063309e-02 4.32384998e-01 -1.61010742e+00
-1.12424839e+00 5.41591942e-01 -4.52720672e-01 2.59013288e-02
5.03978431e-01 -3.05172638e-03 1.82646739e+00 -8.36257756e-01
2.11228490e-01 1.34023178e+00 4.77195710e-01 1.55799031e-01
-1.06155324e+00 -8.90848815e-01 4.88404334e-01 2.72490591e-01
-1.41270411e+00 -7.28903860e-02 1.11516953e+00 -2.44567841e-01
9.49823678e-01 5.78622282e-01 4.83033180e-01 9.85891521e-01
6.46243319e-02 1.38776755e+00 4.98227954e-01 -1.11143716e-01
-6.78359494e-02 3.42722893e-01 7.02727973e-01 5.78792334e-01
-3.87857133e-03 9.86191034e-02 -3.48088473e-01 -4.05669123e-01
5.99028051e-01 7.34036207e-01 4.97970823e-03 -4.57829684e-01
-8.15540850e-01 1.25134993e+00 9.45408583e-01 3.65002841e-01
-1.95032433e-01 7.46996626e-02 3.87332559e-01 5.01661420e-01
4.16966766e-01 3.75841796e-01 -5.01621902e-01 3.34448904e-01
-3.17855418e-01 1.17272615e-01 8.93539846e-01 1.36708939e+00
8.93807709e-01 -4.01687950e-01 -4.78352338e-01 1.04237211e+00
7.72278786e-01 2.66578346e-01 7.48792052e-01 -5.54080486e-01
4.10272747e-01 1.33236027e+00 -1.45250276e-01 -1.46385849e+00
-2.49884874e-01 -6.41837418e-01 -7.56359935e-01 -6.66054547e-01
-3.96840900e-01 2.71904498e-01 -7.68709421e-01 1.70829332e+00
1.08231850e-01 -2.59136170e-01 -2.38004372e-01 9.88397539e-01
1.14028394e+00 4.26163226e-01 4.34336066e-01 5.15750647e-02
1.54377246e+00 -1.42935884e+00 -7.16393352e-01 -6.20825708e-01
9.65820670e-01 -8.02596986e-01 1.33664811e+00 -1.30064562e-01
-5.17863512e-01 -8.32240701e-01 -8.30470860e-01 -5.18056273e-01
-6.46175325e-01 2.16941401e-01 1.43283737e+00 4.81783986e-01
-4.68223512e-01 1.57647058e-01 -2.94221520e-01 -1.47367239e-01
4.68343258e-01 6.23938084e-01 -1.72274355e-02 -2.77208567e-01
-1.62024891e+00 3.03546727e-01 1.63384944e-01 3.78097855e-02
-3.41299534e-01 -6.19164467e-01 -9.65993226e-01 4.37296271e-01
2.95313507e-01 -6.87109292e-01 1.27781212e+00 -9.85500515e-01
-9.69053805e-01 8.42574179e-01 -2.38506660e-01 -2.76479013e-02
1.29260615e-01 -8.87287259e-02 -7.09558308e-01 -3.60553086e-01
3.32921237e-01 6.65314674e-01 5.24642348e-01 -9.54431474e-01
-8.29388440e-01 -6.76870286e-01 2.56721586e-01 3.47104639e-01
-4.18287188e-01 -8.59524086e-02 -1.01984692e+00 -7.60851026e-01
3.69291127e-01 -6.86351180e-01 -2.12161526e-01 -2.04932526e-01
-4.74706233e-01 -8.07918191e-01 6.87903047e-01 -3.83095771e-01
1.48494577e+00 -2.35215354e+00 -3.52655441e-01 2.00527847e-01
3.98483992e-01 -5.11397608e-02 -3.93116832e-01 5.66324174e-01
6.23933133e-03 2.04815462e-01 3.79721373e-01 -1.93153948e-01
4.73281443e-01 2.90019885e-02 -4.94616121e-01 5.20393401e-02
-1.94520935e-01 1.60842264e+00 -6.82195008e-01 -5.24850547e-01
-1.54394777e-02 1.31423220e-01 -6.17850363e-01 4.67651963e-01
-4.81576920e-01 -1.56643212e-01 -1.01930821e+00 9.99114454e-01
5.37098169e-01 -7.15497792e-01 2.58429106e-02 -5.60244262e-01
4.15278912e-01 5.11230350e-01 -7.63050497e-01 1.72115028e+00
-4.83554333e-01 9.26222578e-02 8.28894898e-02 -9.63159323e-01
9.20987368e-01 -5.40685318e-02 6.43848598e-01 -1.33438361e+00
6.62209988e-02 1.71849191e-01 -1.78402245e-01 -4.83058631e-01
4.26316679e-01 2.30552167e-01 -4.64617401e-01 4.43512231e-01
-7.21603408e-02 2.82372236e-01 2.25224998e-03 3.00304264e-01
8.11556280e-01 -3.42289120e-01 2.02006072e-01 -1.62674055e-01
4.85226005e-01 -6.94037229e-03 6.65025890e-01 7.60236979e-01
-1.27801284e-01 1.85524747e-01 3.74035269e-01 -5.85222960e-01
-4.49096441e-01 -5.93899071e-01 1.06807895e-01 1.72297537e+00
7.80318081e-01 -5.66788673e-01 -9.89546329e-02 -9.91312087e-01
4.38987643e-01 6.11765027e-01 -6.16428435e-01 -7.44824409e-01
-3.21172595e-01 -5.58365881e-01 -1.39228059e-02 5.96724510e-01
7.86724746e-01 -1.00687873e+00 2.34566525e-01 1.11037433e-01
-2.56094009e-01 -7.74783909e-01 -1.26241374e+00 1.68745480e-02
-5.86122155e-01 -9.27358389e-01 -3.38484704e-01 -1.18413115e+00
4.17120099e-01 6.58886552e-01 1.36941528e+00 3.31412554e-01
-1.80408508e-01 2.28380606e-01 -2.50868917e-01 -3.56810629e-01
2.06080377e-01 3.52874011e-01 -4.12488163e-01 2.32422173e-01
1.18935013e+00 -3.39777797e-01 -9.75067616e-01 8.10756087e-01
-9.02238309e-01 -4.90005687e-02 9.23393369e-01 8.69716644e-01
5.04767179e-01 -8.32374766e-02 6.68390036e-01 -1.11556482e+00
9.19147015e-01 -7.52395153e-01 -4.89357024e-01 5.29954612e-01
-1.12528622e+00 -2.01019049e-02 3.45237583e-01 -4.98099655e-01
-7.72455335e-01 -1.97955165e-02 -3.72816056e-01 -3.05308938e-01
1.12339891e-01 7.27502763e-01 -6.20691776e-01 -3.40292640e-02
1.75462395e-01 2.36028627e-01 -2.75084764e-01 -6.74307942e-01
4.83416140e-01 9.56482828e-01 5.19753769e-02 -4.26387154e-02
4.19186354e-01 6.58897385e-02 -4.23669904e-01 -1.98590547e-01
-1.15311658e+00 -1.19868290e+00 -6.47697821e-02 2.89143056e-01
6.47633612e-01 -9.69898224e-01 -1.05993724e+00 1.90823153e-01
-9.16313112e-01 7.17877969e-02 -8.49783868e-02 3.10527354e-01
-2.38849759e-01 2.42724851e-01 -8.58844519e-01 -4.33394909e-01
-4.08925265e-01 -1.12835801e+00 1.24586332e+00 3.91703814e-01
3.86674027e-03 -1.13912404e+00 -1.58455104e-01 6.89309418e-01
6.29562855e-01 -5.53649843e-01 1.27917790e+00 -1.26594937e+00
-8.92971754e-01 -3.61337274e-01 -6.37114286e-01 -5.81177026e-02
3.56762797e-01 -6.76745594e-01 -8.57999980e-01 -2.50365466e-01
1.44756034e-01 -4.01096702e-01 1.07963240e+00 4.96997088e-02
1.23477805e+00 -5.48172593e-01 -7.17481017e-01 6.43630505e-01
1.24063253e+00 3.65843058e-01 4.32112604e-01 6.66113421e-02
7.39128888e-01 5.15557528e-01 6.60380423e-01 5.87796308e-02
8.22329640e-01 8.01777601e-01 4.25894409e-01 -3.43665093e-01
8.34812447e-02 -7.98582256e-01 -1.53133590e-02 7.00106084e-01
7.62818456e-01 -2.92942911e-01 -4.22422916e-01 3.38628620e-01
-1.91281581e+00 -6.90086007e-01 1.13219962e-01 1.76642442e+00
6.46897078e-01 9.71718356e-02 -5.14558032e-02 -5.02624035e-01
5.71226656e-01 2.11551309e-01 -9.45597291e-01 -4.29581705e-04
1.67555213e-01 -3.28509331e-01 2.00045064e-01 3.03926349e-01
-1.19306326e+00 9.10911739e-01 5.88442421e+00 1.18068910e+00
-7.86604166e-01 2.38829833e-02 7.21947968e-01 4.23402637e-01
-9.36438203e-01 7.49324113e-02 -9.03944016e-01 5.46371341e-01
3.42804104e-01 -1.28759593e-01 2.86477119e-01 1.35900772e+00
-3.33030492e-01 5.67195535e-01 -1.27721930e+00 1.22776997e+00
1.27846420e-01 -1.17296553e+00 3.61433268e-01 1.67671040e-01
3.61795604e-01 -4.59201494e-03 5.83473854e-02 1.09039176e+00
3.90404433e-01 -7.60380507e-01 1.22045204e-01 4.61508334e-01
4.50984895e-01 -5.51509559e-01 8.37394655e-01 3.87805343e-01
-1.55740905e+00 -4.09463137e-01 -4.41414177e-01 1.66270912e-01
-2.89968159e-02 4.30128157e-01 -3.26041967e-01 5.26238918e-01
5.11635184e-01 8.21812510e-01 -7.18910813e-01 7.27826953e-01
1.51684612e-01 2.15037525e-01 -9.17826742e-02 -3.72844994e-01
2.64651269e-01 -4.78385925e-01 -1.40308868e-02 9.78437006e-01
-2.04187762e-02 -6.25653267e-02 6.49670839e-01 1.02106893e+00
-5.20397782e-01 4.61279064e-01 -5.19571900e-01 -4.69135851e-01
3.60115647e-01 1.36749375e+00 -4.45297569e-01 -2.62440175e-01
-9.04333591e-01 1.10834920e+00 3.49430233e-01 4.00678605e-01
-7.02952266e-01 -6.95991218e-01 7.71096766e-01 -1.11968048e-01
2.59359956e-01 2.94645399e-01 -1.49668559e-01 -1.25482666e+00
8.99056569e-02 -9.07651067e-01 8.87207448e-01 -6.36871457e-01
-1.71676135e+00 4.66936797e-01 -4.07319337e-01 -9.82035756e-01
-1.97131842e-01 -3.82208288e-01 -5.53611875e-01 8.80518198e-01
-1.44456434e+00 -1.56046724e+00 -4.61940259e-01 5.09790123e-01
5.79947710e-01 -2.78462529e-01 7.33539224e-01 6.16080880e-01
-5.06203055e-01 9.89017427e-01 -8.53599459e-02 3.63541216e-01
5.04794538e-01 -9.62637007e-01 4.79109347e-01 2.65584677e-01
4.53619808e-01 1.06518161e+00 3.61649282e-02 -5.76788008e-01
-1.55445576e+00 -1.04245234e+00 1.23464036e+00 -4.20373082e-01
5.55103958e-01 -6.83144569e-01 -9.03847456e-01 8.85150313e-01
1.02711141e-01 -4.20571789e-02 9.76114392e-01 4.64860827e-01
-6.37962759e-01 -4.78544593e-01 -6.79437160e-01 4.90090698e-01
1.20298326e+00 -8.44178438e-01 -5.27274907e-01 6.35972500e-01
1.36284435e+00 7.18206242e-02 -8.28850150e-01 6.08877420e-01
4.50832695e-01 -9.14137006e-01 1.07003868e+00 -8.89792502e-01
-3.53167094e-02 1.39511764e-01 -2.81078666e-01 -8.77293944e-01
-6.34363115e-01 -3.23466033e-01 -1.22611314e-01 1.43680120e+00
4.97341394e-01 -7.61536062e-01 8.53574514e-01 9.49754238e-01
-1.13267519e-01 -1.03679359e+00 -3.56036752e-01 -4.21924740e-01
-3.49687696e-01 -3.73451531e-01 1.03734398e+00 9.23640907e-01
-3.59722064e-03 1.07701313e+00 -8.25901553e-02 -2.25121900e-01
3.21036726e-01 1.03513968e+00 5.23038447e-01 -1.24834919e+00
-3.48243415e-01 -7.64940321e-01 -1.47960380e-01 -1.90519977e+00
2.05673322e-01 -1.02764690e+00 -3.99297953e-01 -1.49556482e+00
5.88097930e-01 -8.06034982e-01 -6.18581474e-01 4.51483667e-01
-3.16720217e-01 -8.99919271e-02 -1.39349788e-01 6.13829076e-01
-1.00729954e+00 7.24386632e-01 1.18495417e+00 -5.29966652e-01
-2.39388421e-01 4.75712985e-01 -1.33624136e+00 3.63849521e-01
5.23315489e-01 -3.07236940e-01 -7.98974097e-01 -3.10111165e-01
4.86399442e-01 -9.61890668e-02 1.53989047e-01 -2.15044841e-01
3.83889407e-01 -1.64893135e-01 2.82121569e-01 -7.60600507e-01
8.49223509e-02 -9.43808377e-01 -1.38701081e-01 4.42652822e-01
-8.21391523e-01 2.73070503e-02 -1.93051189e-01 8.11131597e-01
-5.25984704e-01 -6.72393739e-02 1.41413078e-01 -1.09843232e-01
-9.25003111e-01 7.93718457e-01 3.29912633e-01 1.27256110e-01
6.31765008e-01 2.04440281e-01 -3.17990631e-01 -5.70058286e-01
-3.30243289e-01 8.13752890e-01 2.07464248e-01 9.93680418e-01
4.43300217e-01 -1.71474659e+00 -2.96866328e-01 5.87804735e-01
6.03543639e-01 -3.36973101e-01 3.45873803e-01 4.85630184e-01
5.31977415e-02 9.88679886e-01 2.53341347e-01 -5.04713535e-01
-9.36874866e-01 1.10851252e+00 2.26879299e-01 -5.62790871e-01
-6.47403002e-02 9.14979517e-01 7.28712738e-01 -7.05379486e-01
4.01670367e-01 -1.37640059e-01 -4.39293265e-01 -6.69987127e-02
3.46769154e-01 -4.19187881e-02 1.66054498e-02 -4.80328202e-01
-4.32235658e-01 3.90030056e-01 -5.33608317e-01 3.84649068e-01
8.30668688e-01 -2.84551620e-01 -1.08145326e-01 1.27421632e-01
1.68357360e+00 -2.65634805e-01 -6.98594689e-01 -4.79303986e-01
1.23684198e-01 -4.85583663e-01 1.11476876e-01 -7.16746330e-01
-1.17126751e+00 7.56390810e-01 4.49964017e-01 5.15598714e-01
1.19951248e+00 3.55209291e-01 1.08408773e+00 5.66355109e-01
2.57188886e-01 -9.50939536e-01 3.17525178e-01 2.11637318e-01
9.50633705e-01 -1.56677246e+00 -3.00520629e-01 -6.35040104e-01
-6.15506887e-01 5.47893584e-01 8.81069481e-01 2.15969875e-01
8.53672802e-01 -2.99580425e-01 1.70273036e-01 -3.69613409e-01
-7.71965206e-01 -3.67231637e-01 7.92528987e-01 2.64852554e-01
4.82019991e-01 -1.53349355e-01 -3.15457672e-01 1.11088705e+00
1.27829179e-01 -2.87087232e-01 -5.56446135e-01 7.68166959e-01
-3.04411173e-01 -1.30533910e+00 4.06564951e-01 8.56873095e-01
-2.22038984e-01 -2.32878730e-01 -5.62802136e-01 6.96785927e-01
-6.45167977e-02 1.09187603e+00 -5.05157523e-02 -8.69038999e-01
4.47910696e-01 -6.33402541e-03 -2.27104858e-01 -5.42589843e-01
-6.69565320e-01 2.48731673e-01 -1.35464966e-02 -9.69357073e-01
-1.42439872e-01 -1.82626411e-01 -9.92125571e-01 -4.37476784e-02
-6.86897159e-01 4.82951492e-01 4.57084894e-01 7.51260102e-01
9.25820589e-01 2.80235887e-01 9.06388164e-01 -3.73813435e-02
-8.20123911e-01 -1.04544806e+00 -6.79059386e-01 9.29829478e-01
1.27514452e-01 -5.62409043e-01 -3.20360929e-01 -3.87227178e-01]
|
[10.228209495544434, 5.732882022857666]
|
b519b8f5-80fb-4b19-882d-8fc158aa0ec5
|
recognizing-people-by-body-shape-using-deep
|
2305.19160
| null |
https://arxiv.org/abs/2305.19160v1
|
https://arxiv.org/pdf/2305.19160v1.pdf
|
Recognizing People by Body Shape Using Deep Networks of Images and Words
|
Common and important applications of person identification occur at distances and viewpoints in which the face is not visible or is not sufficiently resolved to be useful. We examine body shape as a biometric across distance and viewpoint variation. We propose an approach that combines standard object classification networks with representations based on linguistic (word-based) descriptions of bodies. Algorithms with and without linguistic training were compared on their ability to identify people from body shape in images captured across a large range of distances/views (close-range, 100m, 200m, 270m, 300m, 370m, 400m, 490m, 500m, 600m, and at elevated pitch in images taken by an unmanned aerial vehicle [UAV]). Accuracy, as measured by identity-match ranking and false accept errors in an open-set test, was surprisingly good. For identity-ranking, linguistic models were more accurate for close-range images, whereas non-linguistic models fared better at intermediary distances. Fusion of the linguistic and non-linguistic embeddings improved performance at all, but the farthest distance. Although the non-linguistic model yielded fewer false accepts at all distances, fusion of the linguistic and non-linguistic models decreased false accepts for all, but the UAV images. We conclude that linguistic and non-linguistic representations of body shape can offer complementary identity information for bodies that can improve identification in applications of interest.
|
["Alice J. O'Toole", 'Carlos D. Castillo', 'Veda Nandan Gandi', 'Matthew Q. Hill', 'Thomas M. Metz', 'Lucas Jaggernauth', 'Blake A. Myers']
|
2023-05-30
| null | null | null | null |
['person-identification']
|
['computer-vision']
|
[-4.36741933e-02 -9.02417526e-02 1.02054335e-01 -4.47849274e-01
-4.00707811e-01 -9.80208755e-01 7.04065740e-01 1.86327517e-01
-6.16250277e-01 4.78181571e-01 1.52574927e-01 -5.31224571e-02
-4.25597340e-01 -7.79595375e-01 -1.47784859e-01 -5.48770249e-01
-2.10470349e-01 5.71078897e-01 -1.89121455e-01 -4.66037631e-01
2.69378582e-03 8.42758477e-01 -1.88077593e+00 1.15372576e-01
3.42400372e-01 9.70850825e-01 -5.42762160e-01 6.24420464e-01
1.72730163e-01 6.71155099e-03 -7.58556545e-01 -1.08844471e+00
6.44083202e-01 -2.06251338e-01 -5.71736395e-01 9.75628793e-02
1.08558178e+00 -1.89382404e-01 9.36547592e-02 1.00356317e+00
8.29865396e-01 5.44007123e-02 7.93429852e-01 -1.09034240e+00
-8.64933550e-01 1.57508835e-01 -5.56497395e-01 1.96292941e-02
9.80939865e-01 7.19411969e-02 6.71738565e-01 -7.78845608e-01
3.26954186e-01 1.50205004e+00 1.29039538e+00 6.58147871e-01
-1.27864051e+00 -4.91958827e-01 -1.98316559e-01 -2.90830553e-01
-1.92190111e+00 -5.89537799e-01 4.04029608e-01 -4.65696663e-01
7.38106608e-01 4.47539061e-01 7.93783605e-01 7.13912010e-01
1.08198769e-01 -1.06391922e-01 9.45513010e-01 -5.50709188e-01
-3.18759322e-01 3.43797177e-01 -2.46006861e-01 7.25386262e-01
7.66604185e-01 4.03066754e-01 -3.98944020e-01 -2.59856433e-01
6.71176374e-01 -7.43463859e-02 -2.01467261e-01 -2.05490693e-01
-1.04553390e+00 7.15923905e-01 3.53495240e-01 3.72510314e-01
-2.78788716e-01 -2.49396548e-01 1.73006773e-01 3.68976533e-01
4.12172049e-01 5.79896569e-01 -1.88342512e-01 1.58085048e-01
-1.01646256e+00 1.46133751e-01 8.52094531e-01 7.46153593e-01
6.27173901e-01 2.56976515e-01 6.95196837e-02 8.93899500e-01
3.21421444e-01 8.79350007e-01 4.58590358e-01 -7.54349351e-01
1.99811935e-01 6.37601495e-01 2.42194295e-01 -1.48670805e+00
-7.01416850e-01 -1.99404925e-01 -8.00892591e-01 4.00708735e-01
7.28610814e-01 -1.56416133e-01 -6.49945021e-01 1.47062182e+00
4.26287770e-01 -3.80281121e-01 2.00209126e-01 1.15572715e+00
1.13331139e+00 1.21875584e-01 -5.84121794e-02 9.61934403e-02
1.61301577e+00 -3.34348202e-01 -3.58458281e-01 -2.36176297e-01
4.18352962e-01 -8.35496902e-01 5.92171550e-01 1.67804033e-01
-1.12297606e+00 -1.01951361e+00 -1.01008558e+00 2.00603068e-01
-6.83499098e-01 2.37937629e-01 1.72354117e-01 1.20451188e+00
-1.37106657e+00 6.02336705e-01 -1.88412383e-01 -6.78622186e-01
4.75239530e-02 7.17695475e-01 -8.46257925e-01 7.13696182e-02
-1.08280826e+00 1.16219413e+00 1.41361281e-01 1.56650707e-01
-2.09901050e-01 -1.64700136e-01 -1.27008498e+00 -5.23500144e-01
-2.31112421e-01 -6.41308904e-01 6.12545788e-01 -1.13180208e+00
-1.09367085e+00 1.58146369e+00 1.64514519e-02 -4.73914266e-01
4.20806050e-01 -9.88692269e-02 -9.18361247e-01 1.55410603e-01
1.12405770e-01 1.02004325e+00 8.20560873e-01 -1.05460525e+00
-4.38026011e-01 -7.79531121e-01 1.77326128e-01 3.67993981e-01
-1.53690651e-01 1.48120552e-01 6.05794303e-02 -4.97532636e-01
1.68845966e-01 -9.49207485e-01 3.25237453e-01 1.01584420e-01
1.68639570e-02 -6.22364730e-02 4.52102721e-01 -8.29626143e-01
6.97950661e-01 -2.03773284e+00 -9.65346918e-02 2.22978443e-01
-4.08588015e-02 4.37106371e-01 -1.60379689e-02 3.89022708e-01
-1.36964411e-01 3.53954285e-01 1.56950787e-01 -1.68429434e-01
-7.61503279e-02 1.08060069e-01 2.45424584e-01 9.42875206e-01
-2.35045385e-02 7.79705942e-01 -6.79823935e-01 -6.35789216e-01
6.02170467e-01 6.00057364e-01 -1.32859334e-01 -2.19202042e-01
6.64017975e-01 2.83540875e-01 2.58512110e-01 9.82443810e-01
5.84120810e-01 2.42139086e-01 -1.89621761e-01 -3.56032878e-01
-1.01485394e-01 -3.00740629e-01 -1.25156438e+00 1.32144189e+00
-3.04567486e-01 5.11352956e-01 8.44661817e-02 -5.39157510e-01
1.30416429e+00 3.98667574e-01 3.54488939e-01 -4.87010747e-01
1.21829577e-01 1.77044824e-01 1.75306842e-01 -3.93275410e-01
5.39042652e-01 -5.75068355e-01 -1.05068654e-01 1.04636654e-01
-1.26046941e-01 -2.26742476e-01 -4.51398902e-02 -4.56127018e-01
3.32240373e-01 -1.16254222e-02 3.91026437e-01 -2.91714489e-01
8.60291183e-01 -1.92290321e-01 1.90556616e-01 5.15232325e-01
-3.93180698e-01 8.11183810e-01 -2.55665332e-01 -8.14716876e-01
-7.48931766e-01 -1.11041439e+00 -3.31834286e-01 1.14548111e+00
4.13234830e-01 -7.97971115e-02 -6.61818981e-01 -3.62786651e-01
5.55483997e-01 3.68434697e-01 -7.49868393e-01 -3.01130235e-01
-2.08138943e-01 -5.24346173e-01 1.03622425e+00 6.12934470e-01
4.08738256e-01 -8.85798752e-01 -6.83838189e-01 -2.03321710e-01
6.17772564e-02 -8.52598548e-01 -5.16037703e-01 -2.92006850e-01
-5.99214196e-01 -9.81226802e-01 -9.30563033e-01 -7.04405963e-01
7.44901359e-01 4.07728143e-02 1.13176191e+00 1.22045930e-02
-4.33578044e-01 7.92836130e-01 -2.43456483e-01 -5.92666924e-01
-1.77189812e-01 -5.21992266e-01 6.52886868e-01 2.46929646e-01
6.34750247e-01 -7.76179731e-02 -4.33416218e-01 6.13335967e-01
-3.65080267e-01 -6.07509792e-01 2.14357331e-01 7.96552658e-01
3.18764031e-01 -5.12787467e-03 4.70817149e-01 -3.59010398e-02
4.34249282e-01 -4.37125936e-03 -3.10509741e-01 1.64287344e-01
-2.99734652e-01 -4.02422130e-01 3.72073323e-01 -6.37509167e-01
-5.23755610e-01 -1.97051212e-01 -2.86112446e-02 -4.03980881e-01
-4.93922263e-01 1.04780696e-01 -1.38102099e-01 -5.22356987e-01
1.00653267e+00 -1.20860957e-01 3.45247686e-01 -2.83096164e-01
1.59955174e-01 8.40781033e-01 7.31618643e-01 -4.75322694e-01
7.91265547e-01 6.45548344e-01 6.75467998e-02 -1.15987694e+00
-3.65661472e-01 -5.99291921e-01 -1.06197476e+00 -4.23230529e-01
9.88871515e-01 -9.13622856e-01 -8.14950645e-01 4.90052998e-01
-7.71057427e-01 1.89697742e-01 -4.85054433e-01 7.24632204e-01
-2.15282202e-01 4.49518502e-01 -1.78821176e-01 -1.23967171e+00
-5.34723997e-01 -7.26784766e-01 1.40022552e+00 2.57904857e-01
-7.79010296e-01 -1.22632408e+00 -3.78948659e-01 3.65036607e-01
1.89872026e-01 4.78049606e-01 4.18218642e-01 -8.40222776e-01
4.08410877e-01 -7.66090810e-01 -1.60627261e-01 4.17581469e-01
6.20509565e-01 -6.11473508e-02 -9.82491374e-01 -5.86628437e-01
-2.89731920e-01 -2.56365716e-01 5.22748172e-01 4.98748302e-01
4.93311971e-01 -3.37851495e-01 -1.94644466e-01 7.55983591e-01
1.14719725e+00 2.83503294e-01 4.87351269e-01 2.51430571e-01
5.07786810e-01 1.09154475e+00 4.59567696e-01 3.04117322e-01
4.16879892e-01 7.64000177e-01 3.62949580e-01 -2.70054102e-01
2.61430182e-02 5.17349578e-02 3.98240328e-01 -1.04570352e-02
-4.66832221e-01 -4.03127214e-03 -1.06998885e+00 5.99256575e-01
-1.01977336e+00 -9.98042941e-01 -2.49355864e-02 2.76366878e+00
3.73038173e-01 -1.48430958e-01 6.74545586e-01 3.67607057e-01
9.86098647e-01 -6.25679316e-03 -3.67931932e-01 -7.57233262e-01
-3.22622269e-01 -1.44349545e-01 5.94132245e-01 4.77618247e-01
-1.32925999e+00 6.05647743e-01 6.63657808e+00 2.27113366e-01
-1.13094568e+00 -1.18269272e-01 3.52961808e-01 -4.99761552e-02
2.41841435e-01 -5.83433211e-01 -8.46560180e-01 2.74609685e-01
9.10580337e-01 6.98838755e-02 3.42981368e-01 7.15377510e-01
-2.74776608e-01 4.16036248e-02 -1.22933781e+00 1.43292236e+00
7.83701658e-01 -8.23548973e-01 1.01106334e-02 2.02235058e-01
4.49163973e-01 -2.90574819e-01 2.14309767e-01 -3.26710492e-02
1.38523802e-01 -1.30117357e+00 7.74803638e-01 5.75734615e-01
1.14133549e+00 -8.08034778e-01 1.09152043e+00 1.80612683e-01
-1.27777338e+00 -9.66324285e-02 -4.96383399e-01 -2.09814966e-01
-5.44631332e-02 6.79251850e-02 -6.72447443e-01 6.50036454e-01
8.54371011e-01 4.67818499e-01 -7.67649174e-01 6.94991827e-01
2.19287813e-01 -1.43116504e-01 -4.92625177e-01 7.38735944e-02
-3.17849987e-03 -1.72986001e-01 4.80945587e-01 1.10049045e+00
4.26572353e-01 1.22902110e-01 2.90428728e-01 4.76059288e-01
2.87124753e-01 7.96451494e-02 -1.17716157e+00 1.13486640e-01
4.15275574e-01 1.11035562e+00 -5.90701699e-01 -1.85405746e-01
-3.43588293e-01 7.90801466e-01 -3.47513855e-01 7.39630461e-02
-5.85885644e-01 -2.05989912e-01 7.97215044e-01 4.35123682e-01
-1.01863943e-01 2.13375445e-02 -6.99219108e-02 -8.27176034e-01
-1.03021257e-01 -8.30895543e-01 6.47545159e-01 -9.12489057e-01
-1.37947893e+00 6.59532964e-01 2.02406302e-01 -1.40090966e+00
-4.13189024e-01 -8.31685483e-01 -2.54726797e-01 1.28175461e+00
-8.23877752e-01 -1.58424139e+00 -3.04702967e-01 4.16546434e-01
3.98722757e-03 -4.90836948e-01 1.11164749e+00 1.06997222e-01
4.65485360e-03 8.55650663e-01 -2.25561693e-01 5.32624185e-01
8.22389483e-01 -1.14297831e+00 3.82000864e-01 3.52451265e-01
2.99309611e-01 7.93190837e-01 4.75742698e-01 -5.74016154e-01
-1.15080070e+00 -1.04272151e+00 9.85082567e-01 -7.96385765e-01
1.47892488e-02 -7.62686599e-04 -4.80962396e-01 4.61048722e-01
-1.47136018e-01 5.92254288e-02 1.07218504e+00 2.20209822e-01
-3.29853356e-01 -2.09355026e-01 -1.80289853e+00 3.40869844e-01
1.03596759e+00 -7.19513357e-01 -7.97388017e-01 1.74651444e-01
-2.52834205e-02 -3.02379251e-01 -1.27261746e+00 5.50443411e-01
1.27488196e+00 -9.91423130e-01 1.54736292e+00 -3.26593459e-01
-1.24408610e-01 -3.52779567e-01 -3.82150412e-01 -1.11051488e+00
-3.87476355e-01 -2.29018569e-01 4.31469470e-01 1.24289668e+00
1.70582235e-01 -1.02370667e+00 6.48259819e-01 4.19944197e-01
2.14767620e-01 -3.18931431e-01 -9.98547733e-01 -1.09199214e+00
1.59261405e-01 -2.15285182e-01 6.96827710e-01 7.54475713e-01
-1.26142710e-01 3.67723070e-02 -3.71829003e-01 3.89733344e-01
6.70912683e-01 1.83220059e-01 7.62303472e-01 -1.63827658e+00
1.15516149e-01 -5.39834559e-01 -1.21325898e+00 -3.92958075e-01
1.30733341e-01 -8.49589407e-01 -3.83855790e-01 -1.57806337e+00
-2.98905879e-01 -1.55424520e-01 2.09613323e-01 4.51415241e-01
1.43426463e-01 8.70569408e-01 1.89270705e-01 2.47011334e-01
-7.13856295e-02 -1.46278366e-01 8.34838629e-01 -2.91180879e-01
3.25541571e-02 1.50194368e-03 -6.51246071e-01 1.07054234e+00
5.44440746e-01 1.34921027e-02 -2.14045644e-01 -3.95330250e-01
-5.92791066e-02 -9.08554997e-03 6.55910552e-01 -1.33384538e+00
1.11846916e-01 2.05495909e-01 1.13257921e+00 -6.03756249e-01
9.15837109e-01 -8.18156540e-01 4.69893932e-01 4.64335531e-01
-8.99886563e-02 2.81178981e-01 4.24572319e-01 1.12288475e-01
-5.01361042e-02 -1.79703623e-01 7.68763185e-01 -2.04114482e-01
-5.87279737e-01 4.91091758e-02 -1.21922262e-01 -1.39140546e-01
9.44707811e-01 -1.07299089e+00 -4.70649078e-02 -5.99773765e-01
-1.02794242e+00 -1.58624172e-01 1.01519418e+00 5.77895582e-01
7.18182147e-01 -1.45048678e+00 -8.60165834e-01 7.70626724e-01
2.91868061e-01 -5.18366694e-01 -1.51188485e-02 7.69467235e-01
-4.02519435e-01 3.61568391e-01 -3.62012416e-01 -7.76277542e-01
-1.74257886e+00 5.72257400e-01 7.50385225e-01 2.94894010e-01
-1.42625272e-01 1.09553504e+00 5.98953944e-03 -7.29779959e-01
1.83663905e-01 -1.77292868e-01 -4.15117085e-01 4.98679429e-01
5.58978975e-01 4.41888005e-01 2.46748724e-03 -1.74398220e+00
-6.17215097e-01 1.35853636e+00 4.63446200e-01 -1.19345069e-01
6.78271651e-01 -1.82643324e-01 7.20508993e-02 5.86023450e-01
9.63676393e-01 9.40819383e-02 -4.81259704e-01 6.88573048e-02
-2.99551606e-01 -6.53705001e-01 -4.12570924e-01 -8.02339792e-01
-7.26140916e-01 1.02794921e+00 1.00599432e+00 3.66118729e-01
9.42993402e-01 -2.58191139e-01 1.34495258e-01 1.50731981e-01
8.80625069e-01 -9.71994400e-01 -3.32495570e-01 3.32177907e-01
1.00979507e+00 -1.27627230e+00 2.75765955e-01 -1.12669528e-01
-6.63601041e-01 1.14915645e+00 4.25711900e-01 2.33647034e-01
5.34012794e-01 -8.48132744e-02 3.46890092e-01 -2.75142998e-01
5.47896139e-02 -5.35491586e-01 8.27930629e-01 1.13805223e+00
5.00877321e-01 1.71342030e-01 9.70389992e-02 3.24266255e-01
-5.33907354e-01 -4.45223093e-01 -2.44759675e-02 6.54251933e-01
-2.58258641e-01 -7.92489052e-01 -1.17180181e+00 4.66975302e-01
-5.19516468e-01 1.40489042e-01 -8.70343208e-01 1.05272949e+00
7.81416059e-01 1.15863800e+00 4.44918036e-01 -6.31635904e-01
4.41325724e-01 2.65594482e-01 6.22423232e-01 -5.77102602e-01
-7.70419061e-01 -4.87994738e-02 4.33637291e-01 -1.98854923e-01
-7.41798818e-01 -8.34656715e-01 -1.07376528e+00 -4.53670055e-01
-5.76279640e-01 1.72711104e-01 5.44747472e-01 5.85877776e-01
-7.22299330e-03 -4.74168621e-02 2.85849929e-01 -1.17982697e+00
-5.46265543e-01 -1.00841153e+00 -8.88247192e-01 4.89562780e-01
3.28276813e-01 -8.66505086e-01 -5.45262754e-01 2.23257169e-01]
|
[13.68407917022705, 0.9857144951820374]
|
4557f96a-0ae7-4ec1-abae-635e5da63eee
|
l3cube-mahasent-md-a-multi-domain-marathi
|
2306.13888
| null |
https://arxiv.org/abs/2306.13888v1
|
https://arxiv.org/pdf/2306.13888v1.pdf
|
L3Cube-MahaSent-MD: A Multi-domain Marathi Sentiment Analysis Dataset and Transformer Models
|
The exploration of sentiment analysis in low-resource languages, such as Marathi, has been limited due to the availability of suitable datasets. In this work, we present L3Cube-MahaSent-MD, a multi-domain Marathi sentiment analysis dataset, with four different domains - movie reviews, general tweets, TV show subtitles, and political tweets. The dataset consists of around 60,000 manually tagged samples covering 3 distinct sentiments - positive, negative, and neutral. We create a sub-dataset for each domain comprising 15k samples. The MahaSent-MD is the first comprehensive multi-domain sentiment analysis dataset within the Indic sentiment landscape. We fine-tune different monolingual and multilingual BERT models on these datasets and report the best accuracy with the MahaBERT model. We also present an extensive in-domain and cross-domain analysis thus highlighting the need for low-resource multi-domain datasets. The data and models are available at https://github.com/l3cube-pune/MarathiNLP .
|
['Raviraj Joshi', 'Rahul Tangsali', 'Isha Joshi', 'Aditya Vyawahare', 'Aabha Pingle']
|
2023-06-24
| null | null | null | null |
['sentiment-analysis']
|
['natural-language-processing']
|
[-4.27676558e-01 -4.70136732e-01 -4.93912935e-01 -7.85648286e-01
-1.25619566e+00 -1.05251086e+00 8.70196998e-01 4.23048049e-01
-5.65742135e-01 6.98031485e-01 5.07465184e-01 -6.09992668e-02
1.16085425e-01 -5.31694293e-01 -5.67183733e-01 -3.42476934e-01
1.28090277e-01 6.55901432e-01 9.16475579e-02 -1.18241990e+00
1.59497157e-01 -1.25895694e-01 -1.20553148e+00 8.36107850e-01
5.47806084e-01 9.37886059e-01 -3.01637828e-01 5.35404563e-01
-2.17327982e-01 9.30836201e-01 -6.84877336e-01 -8.99220884e-01
1.57256320e-01 -3.39069158e-01 -6.64515257e-01 -4.68362957e-01
4.33150470e-01 3.12395006e-01 9.46168825e-02 7.07367659e-01
5.85004628e-01 -4.15611491e-02 5.55599570e-01 -1.39526701e+00
-6.10138059e-01 6.22825265e-01 -6.19327843e-01 2.96144933e-01
4.29568768e-01 -2.56679624e-01 1.42317080e+00 -9.46396232e-01
1.13851893e+00 1.23307550e+00 7.34527111e-01 2.67809004e-01
-7.70953238e-01 -7.92022884e-01 1.89346850e-01 1.20984755e-01
-1.06549931e+00 -2.92401761e-01 7.36457825e-01 -5.31733692e-01
7.97671378e-01 1.76163271e-01 6.71542645e-01 1.41743898e+00
2.79411793e-01 1.11255872e+00 1.72451115e+00 -2.92845964e-01
3.66787389e-02 4.96769220e-01 3.66485864e-01 3.87970023e-02
-9.04070735e-02 -3.49461228e-01 -1.11945629e+00 -1.25237480e-01
-2.58333266e-01 -3.10269833e-01 2.10046992e-01 5.09787491e-03
-1.13203168e+00 1.02431297e+00 1.38351053e-01 1.65124372e-01
-1.08255073e-01 -4.64092612e-01 9.37061131e-01 8.89192641e-01
1.29990733e+00 1.36806771e-01 -1.09001338e+00 -5.30071676e-01
-7.68628359e-01 5.87804198e-01 9.19407725e-01 9.37598884e-01
7.13797271e-01 -1.72953725e-01 3.62888604e-01 1.33796751e+00
2.03096882e-01 9.78023052e-01 4.58517164e-01 -2.92550832e-01
7.47851074e-01 4.77010250e-01 -7.19669089e-02 -9.59049582e-01
-5.72192907e-01 -2.25876510e-01 -2.94521898e-01 -2.52730787e-01
4.31559533e-01 -4.69343215e-01 -6.13035738e-01 1.48336267e+00
5.37542403e-01 -7.14446664e-01 3.58651251e-01 9.91149783e-01
1.28908718e+00 7.02989697e-01 -2.91957986e-02 3.87533277e-01
1.59607661e+00 -9.95067954e-01 -5.87278605e-01 -7.54569590e-01
9.72931325e-01 -1.23472953e+00 1.27681696e+00 6.37877166e-01
-5.80363870e-01 -7.72085190e-02 -8.68106842e-01 -1.52990863e-01
-9.55385983e-01 2.39521056e-01 5.26614845e-01 6.32902086e-01
-6.58088684e-01 -7.81019703e-02 -3.10914189e-01 -6.45282626e-01
2.98350215e-01 -4.73371409e-02 -5.79052031e-01 -2.56234676e-01
-1.56989110e+00 1.01508319e+00 3.08312271e-02 -1.05126210e-01
-7.21935451e-01 -6.39983118e-01 -8.88212562e-01 -8.55519950e-01
1.16939284e-02 2.47701541e-01 1.27197289e+00 -1.31423879e+00
-1.20469511e+00 1.36581266e+00 -1.49451584e-01 -2.01435193e-01
3.98286432e-01 -3.24001521e-01 -9.36067462e-01 -1.34942830e-01
4.98199254e-01 4.61270988e-01 4.60369796e-01 -9.20185149e-01
-8.11488569e-01 -4.84362870e-01 1.80353895e-01 1.54233158e-01
-4.32817221e-01 6.59215331e-01 -4.32320774e-01 -6.71668291e-01
-3.43020111e-01 -1.27639878e+00 -4.66463016e-03 -7.74588108e-01
-2.89927190e-03 -1.16052061e-01 6.89828575e-01 -6.67084336e-01
1.10557330e+00 -2.36125469e+00 -1.61427513e-01 6.27892837e-02
-1.81707412e-01 -1.87179223e-01 -3.48226637e-01 8.43373597e-01
-1.88997686e-01 -1.10795088e-01 1.76218972e-01 -4.17880625e-01
5.35723418e-02 2.81497717e-01 -3.35069895e-01 6.93306506e-01
2.72616744e-01 7.21614897e-01 -8.42104256e-01 -2.57013619e-01
-2.23926529e-01 3.79805684e-01 -6.08425736e-01 -3.88910383e-01
-1.83723167e-01 3.09740037e-01 -4.43999469e-01 8.79506886e-01
6.86365843e-01 9.82549563e-02 3.10643077e-01 -3.38175818e-02
-4.92309719e-01 6.63203716e-01 -6.44227445e-01 1.63220465e+00
-6.12021446e-01 9.75292683e-01 2.14637205e-01 -7.04290032e-01
1.18564844e+00 -7.15036765e-02 3.75456095e-01 -1.17150402e+00
3.66251200e-01 5.53241014e-01 -1.74488485e-01 -2.69780666e-01
9.50395465e-01 -5.09087145e-01 -9.98243630e-01 4.60351288e-01
3.54720652e-02 -2.56752580e-01 5.78957081e-01 3.69707823e-01
6.60917819e-01 -1.58569012e-02 2.04166546e-02 -5.79348803e-01
5.00587761e-01 7.51556933e-01 5.23966789e-01 2.70761341e-01
-3.33601505e-01 6.87562704e-01 9.30212200e-01 -4.90416616e-01
-8.85504127e-01 -6.28165781e-01 -3.67277056e-01 1.69989812e+00
-2.57853627e-01 -7.35475004e-01 -1.88375875e-01 -5.99056184e-01
2.12017614e-02 5.09739518e-01 -8.79409015e-01 3.12574357e-01
-2.29941621e-01 -1.21355331e+00 6.54426157e-01 9.05257240e-02
1.96237519e-01 -7.19536662e-01 -1.08165676e-02 -1.51878044e-01
-5.76604486e-01 -1.17515814e+00 -1.89687550e-01 3.54492158e-01
-3.35666597e-01 -8.97804677e-01 -6.30895853e-01 -8.58796716e-01
1.57507882e-01 -1.58347905e-01 1.45588434e+00 -6.92009747e-01
1.29948333e-01 1.32709846e-01 -1.03873122e+00 -7.42008924e-01
-4.55284029e-01 4.16464984e-01 1.14182122e-01 -1.32409066e-01
8.38224769e-01 3.38810533e-02 -2.82778263e-01 4.37110215e-01
-6.77773058e-01 -3.04603547e-01 2.01675490e-01 6.39230788e-01
5.00056863e-01 -3.58376950e-01 8.56623173e-01 -1.25388432e+00
5.77527642e-01 -1.09912312e+00 -2.55980790e-01 -1.63207173e-01
-1.73491374e-01 -5.29030144e-01 5.58302224e-01 -2.50837088e-01
-1.01867151e+00 -4.17340159e-01 -4.49940801e-01 2.51733333e-01
-7.91546702e-02 1.01684439e+00 1.61530063e-01 2.02273384e-01
8.64053190e-01 -1.96584210e-01 -1.20847791e-01 -4.36411321e-01
5.09000003e-01 1.10139489e+00 1.18650474e-01 -6.98184788e-01
3.97851944e-01 6.60654366e-01 -6.05175078e-01 -1.04695094e+00
-1.37740803e+00 -7.33343661e-01 -5.05510271e-01 -5.13744414e-01
5.14663517e-01 -1.52579570e+00 -1.88138783e-01 6.56410217e-01
-6.04636550e-01 -5.16724586e-01 -1.92787677e-01 3.68965566e-01
-1.95040882e-01 1.12002462e-01 -6.76434278e-01 -5.02365232e-01
-3.10336679e-01 -8.90100658e-01 9.93319094e-01 -1.99519143e-01
-6.18362367e-01 -1.09549046e+00 6.60112143e-01 7.88146555e-01
2.49101132e-01 2.95063853e-01 5.32528043e-01 -1.08096027e+00
4.01486844e-01 -4.29435104e-01 3.20937186e-02 4.72421229e-01
1.18219526e-02 1.81284055e-01 -9.03728664e-01 -3.17088276e-01
-2.17264831e-01 -1.03603375e+00 6.16248250e-01 1.78196445e-01
1.77737638e-01 5.13225310e-02 3.21173161e-01 1.55645773e-01
1.17279935e+00 -1.53470680e-01 3.49514365e-01 1.05505180e+00
3.18432093e-01 8.58284950e-01 1.45194888e+00 7.23186672e-01
1.00143695e+00 2.56619900e-01 6.92501515e-02 -1.04510084e-01
1.80094704e-01 -5.15133440e-02 9.17595029e-01 1.31231856e+00
2.84534931e-01 -3.12470376e-01 -1.24736750e+00 1.09796393e+00
-1.65259063e+00 -5.99214733e-01 -6.17298305e-01 1.71070421e+00
8.46565127e-01 1.50898337e-01 4.89234716e-01 -7.25565031e-02
1.39796257e-01 5.62698662e-01 -2.24003911e-01 -7.98762798e-01
-8.53657782e-01 1.13737211e-01 4.03578401e-01 3.12801421e-01
-1.18815279e+00 1.16521513e+00 6.06528521e+00 1.01437855e+00
-1.19335413e+00 4.62405533e-01 6.29376888e-01 -5.99268019e-01
-4.57389802e-01 8.71026143e-02 -9.91259634e-01 3.61156732e-01
1.31020534e+00 3.51801775e-02 -8.16004127e-02 9.94392812e-01
2.28404388e-01 -2.77825743e-01 -4.30483043e-01 6.06817603e-01
3.15295994e-01 -1.08629358e+00 -3.75740021e-01 -1.86924264e-01
8.54119062e-01 1.15201831e+00 3.03181916e-01 6.46366179e-01
4.32988703e-01 -4.76386458e-01 1.20981479e+00 -1.32811204e-01
7.62872159e-01 -1.05654347e+00 8.36900473e-01 1.51115343e-01
-6.86493933e-01 5.83505183e-02 -2.42235079e-01 -1.03700161e-01
1.53499424e-01 7.44420290e-01 -2.59318203e-01 5.47931135e-01
1.28148425e+00 1.50347841e+00 -6.92791343e-01 7.60771334e-02
-1.74158290e-02 8.47116828e-01 -4.84556288e-01 -3.72274160e-01
5.98621726e-01 -4.32601571e-01 3.41204077e-01 1.61115265e+00
1.97103992e-01 -1.88060895e-01 7.25058988e-02 -4.39612661e-03
1.11163303e-01 6.36840522e-01 -4.74630058e-01 -2.83088982e-01
1.36688817e-02 1.57517278e+00 -5.63630342e-01 -3.03298235e-01
-9.48833466e-01 5.90196669e-01 3.33920360e-01 7.06905946e-02
-7.96405971e-01 -2.65561819e-01 9.15879905e-01 1.01515919e-01
7.54954144e-02 -1.84040889e-01 -2.11019441e-01 -1.44454527e+00
-1.50116324e-01 -1.42827082e+00 6.53738976e-01 -7.10200071e-01
-1.64177477e+00 7.67014563e-01 -3.85895232e-03 -1.44161487e+00
-4.09400575e-02 -9.03363466e-01 -8.31946731e-02 7.11850822e-01
-1.79327452e+00 -1.50618327e+00 3.59565206e-02 6.81901038e-01
5.08156776e-01 -3.71865720e-01 6.00299239e-01 8.31749082e-01
-4.93388504e-01 4.69590217e-01 4.42890525e-01 1.57968000e-01
1.52669430e+00 -8.68853927e-01 4.26737607e-01 3.78418058e-01
-1.78838134e-01 4.01634902e-01 7.59344459e-01 -6.41884327e-01
-1.33272731e+00 -8.56552660e-01 1.21160948e+00 -9.22184885e-01
1.50187659e+00 -7.82946408e-01 -4.48273957e-01 8.17253232e-01
5.10399401e-01 -4.45145011e-01 1.32034528e+00 6.81920946e-01
-4.62113202e-01 -1.41427472e-01 -9.52398419e-01 4.96572316e-01
3.52905005e-01 -5.83269596e-01 -3.37088019e-01 5.66168308e-01
3.25309515e-01 -4.59594309e-01 -1.23800147e+00 3.32173035e-02
6.80643022e-01 -9.34261680e-01 6.53582752e-01 -5.48353732e-01
8.18178654e-01 -7.72276446e-02 -5.36211729e-01 -1.61054826e+00
1.30915821e-01 -1.51385874e-01 6.66851521e-01 1.41861904e+00
9.21002865e-01 -7.99247861e-01 6.63148999e-01 2.01459914e-01
-1.30739555e-01 -4.49717343e-01 -8.56101394e-01 -5.07376850e-01
5.96954763e-01 -1.04501045e+00 3.81034166e-01 1.36910176e+00
3.40188771e-01 5.41642964e-01 -4.57341641e-01 -2.35493481e-01
2.86984354e-01 3.24960560e-01 1.02733672e+00 -9.06993091e-01
8.36074203e-02 -1.33737817e-01 -2.08656713e-01 -7.24378765e-01
2.78607607e-01 -1.05133128e+00 -1.26904815e-01 -1.19746566e+00
2.15156842e-03 -5.78930020e-01 -1.22888163e-01 6.79742575e-01
3.30552399e-01 6.80233598e-01 3.75272259e-02 3.84259224e-02
-6.92776203e-01 5.17768264e-01 9.93009448e-01 -1.23459883e-01
-3.69087942e-02 -2.42732868e-01 -9.47780132e-01 6.87544942e-01
1.09345257e+00 -7.72923112e-01 -1.37557924e-01 -4.49639082e-01
1.13310742e+00 -3.88406336e-01 -1.61021844e-01 -4.74048764e-01
-2.30401650e-01 -1.39238238e-01 2.53077466e-02 -9.52304125e-01
5.13922215e-01 -3.93192798e-01 -8.78869146e-02 7.97909591e-03
-3.40873122e-01 4.72225040e-01 4.78747904e-01 1.94564015e-01
-7.43130267e-01 4.40248847e-02 6.68527782e-01 7.74689913e-02
-9.47175264e-01 -1.69255515e-03 -7.38111436e-01 7.22628117e-01
6.75255239e-01 1.95007324e-01 -6.54474735e-01 -3.96104306e-01
-5.61761200e-01 4.96950746e-01 6.14411533e-01 9.81795728e-01
1.79699913e-01 -1.22734523e+00 -1.10619771e+00 1.11106552e-01
7.68091023e-01 -4.02314216e-01 4.77854908e-01 9.72777903e-01
-6.02150798e-01 2.87241548e-01 -2.69770473e-01 -4.11911845e-01
-1.25658381e+00 -5.65218925e-03 -1.05177633e-01 -1.94820017e-01
-1.78107005e-02 7.16978431e-01 -1.15346737e-01 -1.29440224e+00
-7.11328387e-01 -1.21821679e-01 -3.98472607e-01 8.78696620e-01
3.18704784e-01 2.44201481e-01 1.67686567e-01 -1.27269840e+00
-7.11521924e-01 4.95042801e-01 -2.61610091e-01 -3.45397323e-01
1.50865841e+00 -3.24584067e-01 -4.56973821e-01 8.68038595e-01
1.42492652e+00 4.01071101e-01 -5.11634886e-01 -1.26659691e-01
1.08796908e-02 -3.88231456e-01 -9.33446642e-03 -1.01452816e+00
-8.73997211e-01 5.63909590e-01 1.63709983e-01 1.30672529e-01
9.86089647e-01 9.63368565e-02 6.92561984e-01 1.61842540e-01
2.67304689e-01 -1.61640763e+00 -9.89424363e-02 1.08175039e+00
7.76334584e-01 -1.46706378e+00 1.01714119e-01 -6.74128011e-02
-1.52544773e+00 6.53979421e-01 4.52798188e-01 -3.06912903e-02
8.82263243e-01 2.30819285e-01 9.21594262e-01 -5.30274451e-01
-9.38105524e-01 -9.69311073e-02 2.01771230e-01 2.93481171e-01
8.51647794e-01 1.89217448e-01 -4.07943964e-01 9.46895957e-01
-8.59809518e-01 -2.54080445e-01 7.43843794e-01 1.03686643e+00
1.69567845e-03 -1.16535544e+00 -2.69320399e-01 3.24421585e-01
-7.14099348e-01 -1.01782784e-01 -7.71015882e-01 1.03265488e+00
-1.54396400e-01 1.16932952e+00 -5.49335778e-02 -3.69675279e-01
6.07133925e-01 -1.50862619e-01 -2.45628450e-02 -4.31425035e-01
-9.80582297e-01 2.03870907e-01 8.30684483e-01 -2.89226562e-01
-6.58612072e-01 -9.84608650e-01 -8.82706821e-01 -6.23312354e-01
1.04589947e-01 3.16579580e-01 9.86952424e-01 5.70768833e-01
2.48154759e-01 1.64109636e-02 5.97455502e-01 -5.70650697e-01
1.23001613e-01 -9.47941601e-01 -7.78296053e-01 4.28015888e-01
3.32922578e-01 -2.97308475e-01 -5.27559459e-01 -2.30427593e-01]
|
[11.189703941345215, 6.974056243896484]
|
c744d46d-a9b1-4c3d-b238-850fc1a23aed
|
a-survey-on-incomplete-multi-view-clustering
|
2208.08040
| null |
https://arxiv.org/abs/2208.08040v1
|
https://arxiv.org/pdf/2208.08040v1.pdf
|
A Survey on Incomplete Multi-view Clustering
|
Conventional multi-view clustering seeks to partition data into respective groups based on the assumption that all views are fully observed. However, in practical applications, such as disease diagnosis, multimedia analysis, and recommendation system, it is common to observe that not all views of samples are available in many cases, which leads to the failure of the conventional multi-view clustering methods. Clustering on such incomplete multi-view data is referred to as incomplete multi-view clustering. In view of the promising application prospects, the research of incomplete multi-view clustering has noticeable advances in recent years. However, there is no survey to summarize the current progresses and point out the future research directions. To this end, we review the recent studies of incomplete multi-view clustering. Importantly, we provide some frameworks to unify the corresponding incomplete multi-view clustering methods, and make an in-depth comparative analysis for some representative methods from theoretical and experimental perspectives. Finally, some open problems in the incomplete multi-view clustering field are offered for researchers.
|
['Jinxing Li', 'Zhao Zhang', 'Yong Xu', 'Bob Zhang', 'Lunke Fei', 'Zheng Zhang', 'Jie Wen']
|
2022-08-17
| null | null | null | null |
['incomplete-multi-view-clustering']
|
['computer-vision']
|
[-1.25114232e-01 -2.59815216e-01 -3.42922747e-01 -2.58255333e-01
-6.02320492e-01 -6.81174934e-01 1.62236571e-01 7.87358359e-02
2.13612616e-01 1.50444940e-01 8.96677673e-02 1.33960634e-01
-3.09034675e-01 -4.40076351e-01 -7.12449849e-02 -1.14317501e+00
1.91258937e-01 4.48872328e-01 -9.42557864e-03 2.02205434e-01
1.38377339e-01 1.21443041e-01 -1.57120728e+00 5.90036869e-01
6.76003873e-01 5.00335217e-01 2.83350080e-01 4.72945273e-01
5.04184514e-02 3.34089369e-01 -3.86398226e-01 -4.23567593e-01
-8.59601796e-02 -4.89102542e-01 -5.26821494e-01 8.09529245e-01
-5.36103547e-02 -8.60869288e-02 2.97657144e-03 1.21888733e+00
5.16774714e-01 8.28095078e-02 5.84573328e-01 -1.56396139e+00
-7.15224743e-01 3.00665766e-01 -1.06033134e+00 -9.72153544e-02
3.11657399e-01 -5.42622149e-01 1.00053394e+00 -1.12170458e+00
7.07981408e-01 1.07300329e+00 3.13250929e-01 3.71981502e-01
-8.48221898e-01 -2.53065437e-01 6.82242274e-01 6.28679097e-01
-1.39048290e+00 -2.86390275e-01 9.95309711e-01 -3.65608692e-01
2.00505316e-01 3.79950702e-01 6.75889313e-01 7.52409697e-01
2.79568341e-02 9.60256755e-01 1.29400802e+00 -1.88733190e-01
3.24402243e-01 1.58034757e-01 1.20203361e-01 3.95553708e-01
3.72345507e-01 -4.22312051e-01 -2.53277898e-01 -3.80848825e-01
3.40376049e-01 6.20850623e-01 -2.68931001e-01 -1.03153944e+00
-1.25630724e+00 7.54917681e-01 1.30352087e-03 4.18345422e-01
1.59978122e-02 -5.84773183e-01 4.30777788e-01 9.71831977e-02
5.08959770e-01 -2.47171521e-01 -1.41038761e-01 3.57170790e-01
-9.36710715e-01 -2.04362914e-01 5.81821978e-01 1.26119983e+00
6.50881886e-01 -2.07381040e-01 5.55078328e-01 8.10265481e-01
2.99569160e-01 4.56036359e-01 1.84497342e-01 -1.09226620e+00
2.30488107e-01 8.32602561e-01 -2.72249907e-01 -1.14445722e+00
-2.88387001e-01 -3.80947798e-01 -1.36565936e+00 -1.88938633e-01
1.25468135e-01 5.16168401e-03 -3.27227563e-01 1.25918615e+00
6.71934068e-01 1.57333001e-01 1.76285684e-01 9.20635581e-01
1.09186924e+00 3.20000738e-01 -5.85289180e-01 -9.08818722e-01
1.44296861e+00 -9.13257957e-01 -9.30889189e-01 3.71619552e-01
3.18643600e-01 -9.52072084e-01 6.70327961e-01 7.46428609e-01
-1.07198083e+00 -5.46015143e-01 -9.52528715e-01 3.64022076e-01
-3.74205321e-01 1.33712724e-01 3.64531040e-01 6.22878194e-01
-9.39254284e-01 1.30307302e-02 -1.01340318e+00 -7.50242889e-01
1.60214379e-01 2.14513913e-01 -5.13653696e-01 -6.10536098e-01
-6.45852208e-01 2.01244548e-01 1.63351014e-01 1.43866450e-01
-6.30670667e-01 -3.87406915e-01 -4.84666824e-01 -2.27656871e-01
6.93720520e-01 -7.05078423e-01 6.40249789e-01 -5.88706434e-01
-9.33547795e-01 9.50655639e-01 -5.91806293e-01 2.86472619e-01
2.42188245e-01 2.41899580e-01 -6.94347620e-01 5.20764053e-01
1.60442159e-01 5.93526708e-03 5.16851783e-01 -1.97234750e+00
-7.44506001e-01 -8.75830233e-01 1.12260487e-02 5.68336248e-01
-4.81787294e-01 -5.93935177e-02 -1.05496264e+00 -2.87846982e-01
6.05725288e-01 -1.06392360e+00 -5.17930567e-01 -3.72088850e-01
-7.19621301e-01 -1.21406116e-01 1.29885983e+00 -2.31226936e-01
1.42460728e+00 -2.30120873e+00 5.53198874e-01 1.21946841e-01
7.18860149e-01 -2.13745102e-01 2.73782194e-01 9.45419788e-01
-2.15499520e-01 5.60845733e-02 -1.81357250e-01 -4.51877683e-01
-2.91911304e-01 1.97604492e-01 4.67330143e-02 9.28192914e-01
-7.20971107e-01 4.64050889e-01 -9.08675969e-01 -8.36979270e-01
3.93005341e-01 2.90462583e-01 -4.13030386e-01 7.90067390e-02
2.94993371e-01 6.77028894e-01 -6.69968009e-01 7.64377236e-01
9.20233011e-01 -7.11884320e-01 5.60797393e-01 -2.58733422e-01
-1.41446128e-01 -6.97583139e-01 -1.41244566e+00 1.69131529e+00
1.20343357e-01 2.00238302e-01 5.05166054e-01 -1.29704380e+00
4.50576216e-01 8.18383038e-01 1.13480353e+00 -4.04425971e-02
-1.64542511e-01 7.65791312e-02 -9.20108706e-02 -4.86339867e-01
4.25727934e-01 -1.79163054e-01 -9.14285332e-02 7.48630464e-01
-3.19682747e-01 1.53324679e-01 1.82882309e-01 5.28925836e-01
3.68872821e-01 -3.13740909e-01 5.62540710e-01 -5.80103435e-02
7.64981210e-01 7.44664222e-02 5.50644755e-01 1.51503816e-01
-5.01918018e-01 1.00471127e+00 3.73726726e-01 -1.56883627e-01
-9.09194767e-01 -1.09171283e+00 8.66224803e-03 6.91881359e-01
4.53768015e-01 -7.03873634e-01 -8.32053959e-01 -6.69659913e-01
-4.18261707e-01 -1.40717164e-01 -5.54729760e-01 1.76339582e-01
-1.91634163e-01 -1.01283503e+00 -3.59759815e-02 3.39422256e-01
3.42537016e-01 -3.68914217e-01 -1.15900040e-01 -2.02245906e-01
-4.80098099e-01 -1.18207133e+00 -3.94402176e-01 -2.50943124e-01
-1.14077997e+00 -1.56998968e+00 -9.12070572e-01 -1.01851416e+00
9.48553801e-01 1.27956402e+00 1.06499684e+00 3.46499413e-01
-2.10072417e-02 8.88694465e-01 -7.78426886e-01 -3.20696086e-02
-9.01615769e-02 -1.09579735e-01 2.69238055e-01 2.46815473e-01
6.40178919e-01 -6.00052178e-01 -7.87007749e-01 7.34747589e-01
-1.02317584e+00 2.94536930e-02 2.50373572e-01 7.08074808e-01
1.13673615e+00 5.84836304e-01 5.58810353e-01 -1.17169738e+00
1.95854560e-01 -6.06584728e-01 -2.20789298e-01 4.75402445e-01
-5.50704300e-01 -8.33112359e-01 7.98878670e-01 8.52976963e-02
-1.00878692e+00 1.12685142e-02 2.60750562e-01 -6.99482441e-01
-4.20725405e-01 6.38167977e-01 -5.57829082e-01 3.20652723e-01
-3.04648019e-02 3.53520215e-01 5.36995288e-03 -4.87309128e-01
6.19005799e-01 7.57264018e-01 1.76180646e-01 -1.80575907e-01
6.84913635e-01 1.17267966e+00 -5.93805276e-02 -1.06728721e+00
-9.05902624e-01 -9.79875743e-01 -9.81727779e-01 -4.52966422e-01
1.12486064e+00 -1.06840169e+00 -6.94198906e-01 2.36627743e-01
-6.83450282e-01 3.19340795e-01 1.56852946e-01 6.19190991e-01
-6.38466775e-01 8.74439597e-01 -5.56112170e-01 -6.15911126e-01
7.17724338e-02 -1.30618632e+00 8.75991046e-01 1.06749438e-01
1.98680654e-01 -1.32724047e+00 -3.69335711e-02 7.33245671e-01
-2.05239609e-01 2.09383950e-01 9.82432485e-01 -6.95850194e-01
-4.93644267e-01 6.74445240e-04 1.09748095e-01 -2.68573575e-02
3.92828465e-01 1.06355608e-01 -9.26470876e-01 -7.80570805e-01
6.22078419e-01 7.40886182e-02 4.12676930e-01 8.15674603e-01
1.08534491e+00 2.53941894e-01 -7.54264355e-01 4.85370487e-01
1.67001915e+00 3.52460593e-01 2.32097328e-01 -1.82728812e-01
8.39622974e-01 9.16103005e-01 6.41842842e-01 5.93593121e-01
6.69249952e-01 4.42510515e-01 4.65891361e-01 -1.37329966e-01
2.82419443e-01 1.05338059e-01 -1.25384018e-01 1.75030947e+00
-1.32681578e-01 -4.56962287e-01 -5.42213440e-01 5.03054917e-01
-1.86534214e+00 -1.33927023e+00 -4.92215812e-01 2.14420724e+00
-6.07132874e-02 -4.02157664e-01 3.86009783e-01 3.46804947e-01
1.30310571e+00 3.01559597e-01 -5.75323343e-01 1.81236148e-01
-2.87970990e-01 -8.39582860e-01 -4.96546440e-02 1.79347381e-01
-1.15697253e+00 3.63185197e-01 6.35545683e+00 6.06846571e-01
-5.50394297e-01 2.53046393e-01 8.08656335e-01 -1.54394239e-01
-3.90883178e-01 8.66049007e-02 -5.31615257e-01 3.40141773e-01
3.57990384e-01 -2.41759509e-01 3.15195829e-01 7.17574179e-01
4.21014607e-01 -1.84083521e-01 -1.07175112e+00 1.33792388e+00
4.67694610e-01 -9.34939504e-01 1.22453859e-02 4.11453158e-01
1.17739737e+00 -2.02114597e-01 1.84839025e-01 -2.79277265e-01
6.51064813e-02 -5.24356723e-01 -2.49715745e-02 2.10019574e-01
7.12942839e-01 -1.12522030e+00 5.00136673e-01 6.14996314e-01
-1.68395221e+00 3.22134644e-02 -4.47884321e-01 4.15622480e-02
4.84889984e-01 7.09263682e-01 -1.36649698e-01 1.34903467e+00
6.98035657e-01 1.27880323e+00 -3.97276640e-01 8.86794090e-01
3.22582364e-01 2.72198349e-01 1.29438207e-01 4.98362899e-01
4.65374850e-02 -7.35542655e-01 4.52450871e-01 5.19834638e-01
3.57921511e-01 4.67966050e-01 6.19074881e-01 2.82241344e-01
2.82203197e-01 2.07649961e-01 -7.84879625e-01 2.33680919e-01
5.12398183e-01 1.34283602e+00 -1.24860024e+00 -5.27921319e-01
-1.10858345e+00 9.35366094e-01 -5.47737181e-02 4.64440674e-01
-3.86262029e-01 5.83552942e-02 3.33206564e-01 -1.55886039e-01
2.40094021e-01 -1.33899420e-01 -3.49503249e-01 -1.51760566e+00
-1.19522430e-01 -9.72692013e-01 8.47603917e-01 -5.83763480e-01
-1.57608283e+00 3.81078809e-01 1.63268179e-01 -1.88576674e+00
-4.77854125e-02 -1.82335898e-01 -4.01239783e-01 3.57173830e-01
-1.06292248e+00 -1.02704906e+00 -2.07386926e-01 8.40108097e-01
7.26228178e-01 -2.63938040e-01 7.62937605e-01 3.34612519e-01
-7.45734572e-01 1.19580880e-01 6.74736917e-01 2.77012661e-02
8.73095512e-01 -1.27043307e+00 -2.16162771e-01 7.23793209e-01
5.95896952e-02 7.56455719e-01 5.17776310e-01 -4.95196730e-01
-1.62653756e+00 -7.97693610e-01 5.95627725e-01 -4.84273285e-01
2.93892175e-01 -2.05186099e-01 -8.03271592e-01 6.50846541e-01
6.12091005e-01 5.11466898e-02 1.50543976e+00 2.47517526e-02
5.89132197e-02 -3.82432877e-03 -9.60226119e-01 5.90841770e-01
8.03684056e-01 -4.10330713e-01 -2.85117239e-01 3.89927059e-01
3.45704436e-01 -9.36819911e-02 -1.20412409e+00 1.98435396e-01
5.09464085e-01 -1.54591334e+00 1.06621385e+00 -3.42713624e-01
2.76518941e-01 -5.91139078e-01 -5.69776714e-01 -1.26362121e+00
-4.63294894e-01 -1.87044412e-01 -1.45646334e-01 1.30673575e+00
-2.36164138e-01 -4.92717505e-01 1.13127232e+00 2.41816461e-01
-1.68975636e-01 -8.48394275e-01 -6.81512475e-01 -5.37212849e-01
-7.34290257e-02 -1.45481795e-01 3.82930368e-01 1.16955614e+00
2.35409752e-01 3.62355739e-01 -4.01446462e-01 4.08355951e-01
1.12658036e+00 8.75719428e-01 6.19615436e-01 -1.39020729e+00
-1.70935094e-01 -1.46467388e-01 -1.04854338e-01 -9.25150156e-01
4.29337546e-02 -6.72904909e-01 -4.13402945e-01 -1.90640485e+00
9.39989805e-01 -1.78205982e-01 -2.85636038e-01 -3.78149509e-01
-2.61455417e-01 3.09396654e-01 2.85202116e-01 7.30000794e-01
-1.13747180e+00 3.85274500e-01 1.61781895e+00 2.21544243e-02
-1.19218871e-03 3.09561610e-01 -9.26468670e-01 9.59932208e-01
4.71299350e-01 -1.50862977e-01 -9.27343309e-01 -2.72923529e-01
5.63549437e-02 5.31659305e-01 -1.67000383e-01 -6.49870932e-01
4.16543126e-01 -3.00236821e-01 2.91304231e-01 -1.41271341e+00
4.12020653e-01 -1.17916501e+00 2.89041549e-01 2.99751401e-01
1.83910102e-01 4.67215419e-01 -5.21896303e-01 1.18889189e+00
-6.80749655e-01 9.70959216e-02 7.09099472e-01 -6.40603602e-01
-7.24881768e-01 5.27097225e-01 -5.35688698e-01 8.54749084e-02
1.40576327e+00 -4.89174277e-01 -1.32068276e-01 -5.60852051e-01
-1.12736773e+00 5.98631024e-01 9.66685116e-01 1.65272877e-01
6.55278504e-01 -1.51214123e+00 -4.11300123e-01 7.91475251e-02
3.13137084e-01 -4.76796590e-02 1.04602325e+00 1.18208158e+00
-6.14157580e-02 4.73465115e-01 1.11024596e-01 -9.51681316e-01
-1.62832761e+00 1.30355561e+00 -5.56705706e-02 -1.17563419e-01
-4.00181204e-01 1.62108332e-01 7.50550508e-01 -2.57939160e-01
1.23744942e-01 3.98697883e-01 -5.79975843e-01 3.63292545e-01
5.61684132e-01 7.31447101e-01 -6.67787716e-02 -9.24359620e-01
-4.99585360e-01 9.69736278e-01 1.54886708e-01 -1.22367470e-02
1.27877223e+00 -9.47437525e-01 -3.79513472e-01 8.61570597e-01
1.35553348e+00 2.03241669e-02 -8.36512029e-01 -1.90461621e-01
-3.08914214e-01 -3.82467002e-01 -3.97923350e-01 -1.10850431e-01
-1.47412884e+00 1.03365159e+00 3.06729496e-01 5.48156559e-01
1.40776074e+00 1.30930662e-01 4.48662341e-01 1.40798211e-01
6.08920574e-01 -1.14051402e+00 1.94730267e-01 -2.10849680e-02
3.21978092e-01 -1.25612044e+00 3.63860101e-01 -7.43621647e-01
-7.98190713e-01 9.07409847e-01 5.72155774e-01 7.62315467e-02
1.06781864e+00 6.92378804e-02 1.71233013e-01 -4.80407327e-01
-8.92311096e-01 1.86200775e-02 -2.00990979e-02 7.33338952e-01
4.76718307e-01 6.12310395e-02 -2.59380370e-01 3.73502016e-01
3.27506483e-01 -3.90339434e-01 7.17819214e-01 8.45130026e-01
-3.23870540e-01 -1.18080783e+00 -8.22402060e-01 4.81734186e-01
-5.38450003e-01 4.11655992e-01 -3.41217726e-01 7.05795467e-01
5.95542975e-03 1.44806421e+00 -4.19425778e-02 -2.55314559e-01
9.54647139e-02 -1.58947036e-01 2.01300114e-01 -7.66880631e-01
-1.58899009e-01 7.67942846e-01 -4.19045061e-01 -2.74644583e-01
-1.09903121e+00 -8.97958815e-01 -9.13690984e-01 -4.02770311e-01
-4.26772743e-01 3.89587104e-01 1.85484454e-01 8.64422798e-01
2.14236408e-01 2.04056859e-01 9.49789226e-01 -7.21655428e-01
-2.02263296e-02 -3.98197919e-01 -1.19129825e+00 5.25559843e-01
1.66766480e-01 -5.86020947e-01 -2.04692468e-01 4.33847249e-01]
|
[8.275586128234863, 4.59217643737793]
|
a81546ab-78e1-46e0-89e3-6c4234c39aa9
|
take-more-positives-a-contrastive-learning
|
2101.04340
| null |
https://arxiv.org/abs/2101.04340v2
|
https://arxiv.org/pdf/2101.04340v2.pdf
|
Take More Positives: An Empirical Study of Contrastive Learing in Unsupervised Person Re-Identification
|
Unsupervised person re-identification (re-ID) aims at closing the performance gap to supervised methods. These methods build reliable relationship between data points while learning representations. However, we empirically show that the reason why they are successful is not only their label generation mechanisms, but also their unexplored designs. By studying two unsupervised person re-ID methods in a cross-method way, we point out a hard negative problem is handled implicitly by their designs of data augmentations and PK sampler respectively. In this paper, we find another simple solution for the problem, i.e., taking more positives during training, by which we generate pseudo-labels and update models in an iterative manner. Based on our findings, we propose a contrastive learning method without a memory back for unsupervised person re-ID. Our method works well on benchmark datasets and outperforms the state-of-the-art methods. Code will be made available.
|
['Lin Ma', 'Qian Zhang', 'Xiangyuan Lan', 'Ran Song', 'Wei zhang', 'Xuanyu He']
|
2021-01-12
| null | null | null | null |
['unsupervised-person-re-identification']
|
['computer-vision']
|
[ 2.43477121e-01 2.97080755e-01 -3.22403699e-01 -4.30098414e-01
-2.91986078e-01 -3.29755723e-01 8.15593123e-01 1.69048294e-01
-6.54029906e-01 8.25043440e-01 3.51334155e-01 5.40770851e-02
-9.43389609e-02 -6.86532080e-01 -4.84005272e-01 -5.37803471e-01
2.64841050e-01 8.49317133e-01 9.63211358e-02 -2.45576706e-02
2.58902729e-01 2.45317280e-01 -1.64643192e+00 -1.20530643e-01
7.99930394e-01 1.73334017e-01 -2.49480382e-01 2.64731914e-01
-8.41797963e-02 7.09416866e-01 -3.30480009e-01 -7.44813800e-01
1.97754234e-01 -6.23668671e-01 -1.02950752e+00 1.64156213e-01
3.92382890e-01 -3.74217868e-01 -2.54984915e-01 9.33160365e-01
6.36388481e-01 3.82611096e-01 1.07699323e+00 -1.27889454e+00
-8.29951763e-01 7.52914071e-01 -6.38932705e-01 -5.79960644e-02
3.81261975e-01 -1.93757206e-01 8.64282608e-01 -8.97895575e-01
2.96786755e-01 1.16231227e+00 1.04251266e+00 1.21881938e+00
-1.54987049e+00 -7.19507456e-01 2.90779680e-01 1.37536690e-01
-1.59066415e+00 -5.88655472e-01 7.23475277e-01 -5.43032765e-01
5.53252935e-01 1.66707560e-01 3.76912355e-01 1.27487278e+00
-6.84767962e-01 7.80196726e-01 1.24900854e+00 -8.01063776e-01
2.28643894e-01 6.09822989e-01 6.04108274e-01 5.35959840e-01
4.56343085e-01 2.26907715e-01 -4.31929648e-01 -2.70160764e-01
6.52401030e-01 1.86698318e-01 2.30200849e-02 -3.93283397e-01
-9.78916466e-01 8.41713846e-01 5.25819920e-02 3.04165304e-01
-1.65142212e-02 -4.97346967e-02 3.11193764e-01 1.93205386e-01
4.57556069e-01 4.71349329e-01 -3.25980932e-01 -5.17063141e-02
-7.35049605e-01 3.68388116e-01 7.68557906e-01 7.06669450e-01
8.46100509e-01 -3.02569717e-01 -2.50566244e-01 1.02826703e+00
2.20891073e-01 1.51557595e-01 9.26853359e-01 -5.89557588e-01
1.74051136e-01 7.59912312e-01 2.60562479e-01 -7.13450909e-01
-5.29339850e-01 -3.69549364e-01 -8.47902358e-01 -4.82975021e-02
6.35955334e-01 -1.71382129e-01 -7.89187551e-01 1.98806167e+00
3.05291504e-01 4.72426623e-01 1.05675362e-01 6.28436744e-01
6.94103777e-01 1.50612369e-01 2.80602455e-01 -2.46323824e-01
1.28696430e+00 -1.09630334e+00 -6.95137084e-01 -1.51320279e-01
8.39526296e-01 -2.18785271e-01 9.84033704e-01 3.14824581e-01
-7.90051460e-01 -9.20222521e-01 -8.87581587e-01 1.22533098e-01
-4.85566020e-01 3.55191678e-01 6.94385052e-01 1.19389701e+00
-8.56730580e-01 8.65077198e-01 -5.15525758e-01 -7.88627446e-01
2.04259485e-01 5.26803970e-01 -3.93764287e-01 2.10932657e-01
-9.43140984e-01 7.24388540e-01 5.20782292e-01 -2.26010188e-01
-6.48273766e-01 -4.47132111e-01 -6.00087643e-01 -1.16218738e-01
3.48742574e-01 -6.49215043e-01 1.12011099e+00 -1.01477885e+00
-1.44694686e+00 1.07528305e+00 -3.02354336e-01 -5.84512830e-01
6.08903885e-01 -3.66838276e-01 -4.26207095e-01 -2.03698143e-01
1.14528567e-01 6.69582427e-01 7.87265956e-01 -1.63048482e+00
-6.61700547e-01 -4.48324293e-01 -5.28868958e-02 6.16407916e-02
-7.91862905e-01 -2.11842582e-02 -4.01301116e-01 -6.65934563e-01
-3.33847702e-02 -1.06128359e+00 -3.11850846e-01 -5.34972012e-01
-4.23425078e-01 -6.20078981e-01 3.88880700e-01 -5.35465658e-01
1.38698912e+00 -2.07412696e+00 -1.32426620e-01 3.42288792e-01
3.87988269e-01 3.78871471e-01 -3.21961641e-02 3.04969341e-01
-3.60804021e-01 3.52602243e-01 -1.42999038e-01 -9.55897510e-01
2.70371772e-02 8.27890560e-02 -2.90325344e-01 4.17593747e-01
-6.11591488e-02 8.56092632e-01 -9.56551850e-01 -5.56565523e-01
-6.25946820e-02 2.32634529e-01 -4.99985188e-01 3.38647157e-01
2.12074891e-01 6.06931090e-01 -1.54884771e-01 3.98365557e-01
5.71348071e-01 -2.80699909e-01 3.24787229e-01 -2.39859298e-01
-7.86816254e-02 1.52274981e-01 -1.44679117e+00 1.40233767e+00
-9.98552795e-03 1.58770531e-01 -6.64095819e-01 -1.16854930e+00
1.00947356e+00 1.73402727e-01 4.66075540e-01 -4.58236933e-01
4.99828868e-02 1.05694368e-01 -2.74583519e-01 -2.87045598e-01
6.02536440e-01 -1.01391383e-01 9.47485715e-02 8.12120318e-01
4.62009795e-02 6.97246611e-01 1.95235044e-01 4.96173427e-02
7.54186809e-01 2.79077172e-01 4.27458286e-01 -1.16888866e-01
6.30357325e-01 -1.67635173e-01 5.85262597e-01 1.20806015e+00
-2.44490340e-01 7.16675639e-01 1.14131138e-01 -4.92719501e-01
-1.08066702e+00 -9.04262245e-01 -5.79428934e-02 1.16367590e+00
9.35003608e-02 -6.37863874e-01 -8.96293402e-01 -9.77692008e-01
2.66306177e-02 5.12648284e-01 -9.32055652e-01 -2.23932907e-01
-5.95125675e-01 -8.79117191e-01 6.78722024e-01 6.89994633e-01
5.40372372e-01 -8.41986537e-01 -8.05525109e-02 1.37301624e-01
-2.78856337e-01 -8.89250398e-01 -3.56333852e-01 1.00319177e-01
-9.71021116e-01 -9.34295833e-01 -9.34903681e-01 -8.36642087e-01
1.02265453e+00 4.34801072e-01 1.02260923e+00 2.51611680e-01
1.63990576e-02 5.50034583e-01 -5.36773205e-01 -3.27595770e-01
-4.05154169e-01 3.88741523e-01 6.43163145e-01 2.02670217e-01
9.03654695e-01 -6.29816651e-01 -4.81026918e-01 5.91113567e-01
-6.63708329e-01 1.45735875e-01 3.80403489e-01 9.77761984e-01
4.94549483e-01 1.35347977e-01 7.98525572e-01 -1.36062479e+00
6.12594128e-01 -4.30882186e-01 -2.92344958e-01 3.50757718e-01
-1.09410715e+00 2.31020913e-01 4.52590197e-01 -7.10356653e-01
-1.04372311e+00 2.80480593e-01 2.38002427e-02 -5.64773865e-02
-4.44913298e-01 5.83286397e-02 -7.29602203e-02 7.68641084e-02
7.79149532e-01 2.76421547e-01 -4.09861729e-02 -8.22422326e-01
3.18298131e-01 7.85930514e-01 5.52103519e-01 -7.37010181e-01
9.86117780e-01 6.02021277e-01 -3.95694554e-01 -5.36827266e-01
-9.52482164e-01 -6.85908556e-01 -9.12416995e-01 -6.89184815e-02
7.15543449e-01 -9.93097425e-01 -7.34890819e-01 5.23703218e-01
-8.85447204e-01 -2.37042069e-01 -4.59841788e-01 4.35598135e-01
-4.56446737e-01 5.55458426e-01 -4.29714203e-01 -1.05063856e+00
-1.86550289e-01 -6.11851454e-01 6.52681589e-01 4.63850170e-01
-4.94357646e-01 -9.60014760e-01 3.74745697e-01 3.76158595e-01
1.84143826e-01 -1.65765449e-01 6.23351336e-01 -1.11576462e+00
-8.79530236e-02 -5.99369146e-02 -2.67360181e-01 2.81919956e-01
4.23561126e-01 -6.08918548e-01 -1.26428711e+00 -4.89791840e-01
-2.80719906e-01 -2.64890075e-01 9.60957944e-01 -3.84075753e-02
1.16722417e+00 -4.75397944e-01 -6.65953338e-01 4.09567297e-01
1.25676274e+00 -8.57089460e-02 6.91159785e-01 4.74435419e-01
8.75584245e-01 8.70884061e-01 2.86177903e-01 5.52688658e-01
6.78557634e-01 8.08687091e-01 -1.20647758e-01 -1.67644471e-01
-1.75655201e-01 -5.96819580e-01 1.27273872e-01 5.36687732e-01
-4.46080863e-01 -6.11047745e-02 -7.52443850e-01 6.15447998e-01
-2.17336440e+00 -1.06581581e+00 -1.10904276e-01 2.57003951e+00
9.86414611e-01 -2.40241941e-02 7.91295230e-01 3.91804367e-01
8.06630135e-01 -3.83152694e-01 -4.05333906e-01 7.38306940e-02
4.65911739e-02 4.54230094e-03 5.14347792e-01 3.31524462e-01
-1.22457695e+00 9.52372134e-01 6.48168039e+00 5.95165908e-01
-5.15062928e-01 2.22833082e-01 6.18751943e-01 2.10481375e-01
-2.91265417e-02 4.62594144e-02 -1.28862762e+00 5.46877682e-01
8.09708655e-01 9.51879546e-02 3.89422476e-01 7.49958158e-01
-5.45131788e-02 3.64197344e-02 -1.40421188e+00 1.34588051e+00
2.90753186e-01 -9.83894348e-01 4.35048081e-02 1.06978878e-01
7.39496887e-01 -4.39447552e-01 -1.91612154e-01 4.16242570e-01
4.79731500e-01 -9.11103368e-01 4.71874684e-01 7.25303710e-01
5.20469129e-01 -7.26854801e-01 7.53259003e-01 3.90863001e-01
-9.31622982e-01 -2.96428591e-01 -2.64919460e-01 -1.50943279e-01
-2.97394558e-03 4.17193025e-01 -6.60571694e-01 4.51962918e-01
7.40306914e-01 7.10930943e-01 -9.92660999e-01 1.07100439e+00
-1.34596080e-01 7.81320453e-01 -9.35275182e-02 1.58357784e-01
-4.25242811e-01 -1.06462643e-01 1.78323209e-01 1.26118851e+00
1.22670062e-01 -1.79481125e-04 2.11582288e-01 9.05897141e-01
3.01789474e-02 8.70311931e-02 -5.56580484e-01 1.21889245e-02
4.19198573e-01 1.05688882e+00 -6.42944813e-01 -3.53926450e-01
-4.03512686e-01 1.19553125e+00 6.32742286e-01 3.53582621e-01
-5.62279105e-01 -1.17296137e-01 3.30858171e-01 3.51440251e-01
-1.38141876e-02 -1.65603101e-01 -2.58189470e-01 -1.15531576e+00
-1.29496798e-01 -8.58418941e-01 5.38393140e-01 -3.15772146e-01
-1.83787751e+00 2.58917689e-01 2.12240234e-01 -1.17742956e+00
-4.54533309e-01 -3.87276024e-01 -3.07459623e-01 6.69758976e-01
-1.54682183e+00 -1.13547337e+00 -4.46148932e-01 6.31152332e-01
2.69176364e-01 -3.70483160e-01 9.52470005e-01 4.89040732e-01
-7.62177527e-01 1.06003153e+00 2.11105440e-02 3.78759861e-01
8.70528340e-01 -1.24360669e+00 4.92485255e-01 7.08266199e-01
3.10357779e-01 8.59311819e-01 5.58363974e-01 -7.30682552e-01
-9.04230893e-01 -9.05162692e-01 1.02289021e+00 -6.54760242e-01
2.82773226e-01 -4.86364722e-01 -9.20886159e-01 7.16537476e-01
-1.52296156e-01 -3.57750267e-01 1.00822759e+00 5.20793617e-01
-4.02408719e-01 -2.85198335e-02 -1.19476676e+00 6.34880722e-01
1.36618567e+00 -3.13042909e-01 -6.39985859e-01 2.48765796e-01
4.86316323e-01 2.56235362e-03 -5.15323460e-01 3.53873640e-01
4.77933079e-01 -9.32337880e-01 1.09755480e+00 -6.48885727e-01
2.64631137e-02 -3.33885729e-01 1.66645959e-01 -1.07588732e+00
-6.09902322e-01 -6.16367221e-01 -3.38216275e-01 1.83939850e+00
2.78681815e-01 -8.91562581e-01 9.75322545e-01 7.33772218e-01
4.82844353e-01 -3.81021470e-01 -5.60003877e-01 -9.08329308e-01
-9.64736715e-02 -1.33015692e-01 7.56923616e-01 1.15649796e+00
-2.25511938e-02 4.68759835e-01 -7.20972478e-01 9.70102102e-02
8.35265398e-01 -2.74516314e-01 1.13041890e+00 -1.64009273e+00
-4.60304916e-01 -2.22569868e-01 -1.76486135e-01 -1.16736615e+00
2.93210924e-01 -7.09656477e-01 -1.76669464e-01 -1.19035244e+00
6.31566882e-01 -8.23602319e-01 -3.54653060e-01 7.39651740e-01
-3.49311948e-01 2.98337042e-01 1.30317569e-01 5.82282245e-01
-5.77310145e-01 3.67374241e-01 6.43798411e-01 -1.12556517e-01
-5.36186874e-01 1.40353695e-01 -1.03417432e+00 7.07887113e-01
1.07957423e+00 -6.26126587e-01 -4.41209197e-01 -1.61326364e-01
1.91505715e-01 -6.99763238e-01 4.98428017e-01 -1.26863003e+00
3.09134156e-01 1.85807105e-02 4.47892874e-01 -4.56751466e-01
2.24176154e-01 -6.75555825e-01 2.69853752e-02 3.08257043e-01
-5.81439614e-01 2.13466547e-02 -2.32192159e-01 6.75123572e-01
1.93183392e-01 -7.23308206e-01 6.77993059e-01 -2.48926178e-01
-6.85098648e-01 1.68019980e-01 -2.16503084e-01 1.55186523e-02
7.69645452e-01 -4.68211681e-01 -1.34368926e-01 -2.96418935e-01
-7.26277173e-01 6.97224289e-02 4.50840235e-01 5.09265482e-01
3.79139572e-01 -1.43460655e+00 -7.49963522e-01 3.17542732e-01
3.53534490e-01 -4.06335980e-01 1.08735124e-02 5.45282006e-01
1.34668663e-01 1.56268969e-01 -6.79844618e-02 -3.21268976e-01
-1.20270753e+00 7.06574738e-01 3.46852660e-01 -3.89236361e-01
-5.91227889e-01 5.62809706e-01 1.26098692e-01 -6.87745869e-01
6.36299670e-01 2.94412762e-01 -6.57384455e-01 2.08322108e-01
7.96983778e-01 5.20139396e-01 -2.33870372e-01 -6.10475540e-01
-2.74753869e-01 5.80078304e-01 -4.41745728e-01 -1.04620926e-01
1.18275619e+00 -2.83232629e-01 1.15440905e-01 4.99625534e-01
9.20654893e-01 5.49731590e-03 -1.01296902e+00 -5.61594009e-01
1.32937714e-01 -3.72821450e-01 -5.15931547e-01 -6.99403226e-01
-6.89848542e-01 4.95881110e-01 1.06238294e+00 1.81994617e-01
8.86181056e-01 2.03758061e-01 6.23772144e-01 4.60113972e-01
3.00643951e-01 -1.44024003e+00 2.81143695e-01 3.77283245e-01
3.49575222e-01 -1.51681030e+00 2.58018672e-01 -3.31446260e-01
-5.50679624e-01 7.93972850e-01 7.85178483e-01 3.62391993e-02
5.05528867e-01 -1.45852774e-01 -1.07041061e-01 8.61464664e-02
-2.04151034e-01 -5.31530261e-01 1.40609950e-01 1.02094781e+00
4.17218596e-01 7.76612759e-02 -4.37638491e-01 9.89625335e-01
-1.60363406e-01 1.12827554e-01 3.26835930e-01 7.39487529e-01
-1.99616462e-01 -1.51916075e+00 -4.03242469e-01 2.88546294e-01
-2.23840892e-01 1.48677276e-02 -5.59577227e-01 6.51396155e-01
3.72274905e-01 1.00666738e+00 -6.30215555e-02 -5.30081034e-01
2.90007204e-01 1.73793972e-01 4.34760630e-01 -6.15803301e-01
-6.33860707e-01 -4.20279145e-01 1.29739180e-01 -6.21500537e-02
-6.18065298e-01 -6.21709287e-01 -1.07346141e+00 -2.13977799e-01
-3.87324423e-01 2.15835258e-01 3.49958837e-01 1.00215459e+00
2.94586152e-01 1.94801658e-01 7.18671679e-01 -6.59467638e-01
-6.61204040e-01 -9.98039901e-01 -5.31834483e-01 7.70859659e-01
1.29081249e-01 -8.30778718e-01 -4.32009637e-01 2.48841316e-01]
|
[14.81615161895752, 1.1041351556777954]
|
f93fe532-b025-44a4-865f-d524702eb66b
|
evrnet-efficient-video-restoration-on-edge
|
2012.02228
| null |
https://arxiv.org/abs/2012.02228v1
|
https://arxiv.org/pdf/2012.02228v1.pdf
|
EVRNet: Efficient Video Restoration on Edge Devices
|
Video transmission applications (e.g., conferencing) are gaining momentum, especially in times of global health pandemic. Video signals are transmitted over lossy channels, resulting in low-quality received signals. To restore videos on recipient edge devices in real-time, we introduce an efficient video restoration network, EVRNet. EVRNet efficiently allocates parameters inside the network using alignment, differential, and fusion modules. With extensive experiments on video restoration tasks (deblocking, denoising, and super-resolution), we demonstrate that EVRNet delivers competitive performance to existing methods with significantly fewer parameters and MACs. For example, EVRNet has 260 times fewer parameters and 958 times fewer MACs than enhanced deformable convolution-based video restoration network (EDVR) for 4 times video super-resolution while its SSIM score is 0.018 less than EDVR. We also evaluated the performance of EVRNet under multiple distortions on unseen dataset to demonstrate its ability in modeling variable-length sequences under both camera and object motion.
|
['Vikas Chandra', 'Rakesh Ranjan', 'Vikram Mulukutla', 'Varun Nasery', 'Fitsum Reda', 'Amit Kumar', 'Sachin Mehta']
|
2020-12-03
| null | null | null | null |
['video-restoration']
|
['computer-vision']
|
[ 6.07127786e-01 -3.23145628e-01 6.76750541e-02 -1.42444998e-01
-6.67415559e-01 -3.42925876e-01 1.39850648e-02 -5.82996488e-01
-5.10990798e-01 7.29551196e-01 6.01674139e-01 -2.38303691e-01
-4.57783565e-02 -4.14864272e-01 -7.70937085e-01 -7.34943688e-01
-5.41819513e-01 -1.44849837e-01 1.62684351e-01 -1.22012340e-01
8.79081860e-02 2.83196330e-01 -1.00680137e+00 3.77870232e-01
7.17996359e-01 8.90445352e-01 4.55583513e-01 1.20395756e+00
4.75892693e-01 8.98356557e-01 -7.27247477e-01 -4.08163875e-01
2.20008105e-01 -1.97082728e-01 -5.04031599e-01 1.65169984e-01
3.09898198e-01 -1.13232088e+00 -1.23911810e+00 9.97297466e-01
8.10054421e-01 3.15851450e-01 3.76399904e-01 -1.16051912e+00
-7.05691934e-01 6.38564467e-01 -6.98365569e-01 7.61718392e-01
3.77708793e-01 3.99301611e-02 3.90028119e-01 -5.97913802e-01
5.97039282e-01 1.33962393e+00 8.89133573e-01 7.04471946e-01
-1.01974070e+00 -8.22001159e-01 -2.23407865e-01 5.56394160e-01
-1.32932425e+00 -1.14449883e+00 1.93226561e-01 1.72264963e-01
9.39764202e-01 4.42773849e-01 1.52708888e-01 1.16014695e+00
2.87536561e-01 5.31599760e-01 4.53216255e-01 1.19867861e-01
1.02047585e-01 -4.02160108e-01 -3.74065727e-01 2.50277221e-01
2.37306163e-01 1.35060742e-01 -7.37589359e-01 -1.21511281e-01
1.18636918e+00 -1.84903853e-02 -9.87745881e-01 4.89985466e-01
-1.11615038e+00 3.80095661e-01 -1.44227296e-01 6.04219623e-02
-5.32201409e-01 6.13938212e-01 4.60344225e-01 7.18251824e-01
5.92535913e-01 -3.36737901e-01 -1.81334138e-01 -3.26779902e-01
-1.01209068e+00 -2.03701645e-01 4.41605389e-01 1.09935713e+00
8.74159709e-02 2.28533074e-01 -2.25413233e-01 9.37224329e-01
1.07792974e-01 5.00059485e-01 2.27695480e-01 -1.58836234e+00
6.25356019e-01 -5.39320290e-01 6.27277866e-02 -1.10630882e+00
-1.70554116e-01 -3.33393306e-01 -1.44923913e+00 -8.21723565e-02
-7.91388825e-02 -3.63071203e-01 -7.25532055e-01 1.81035519e+00
-2.90823560e-02 9.45917547e-01 3.60197216e-01 1.09225655e+00
7.96791196e-01 9.24800992e-01 -2.78184265e-01 -6.32872701e-01
1.14343071e+00 -8.03570449e-01 -1.01862049e+00 4.75952551e-02
1.83949932e-01 -8.42976153e-01 1.23886786e-01 5.27326465e-01
-1.48730659e+00 -3.42351526e-01 -9.24856901e-01 9.73220691e-02
6.56241953e-01 -2.52315015e-01 3.22504818e-01 6.56023920e-01
-1.76457596e+00 6.94639325e-01 -7.41755664e-01 -1.68579832e-01
6.34445071e-01 5.04604995e-01 -5.66325963e-01 -6.40475750e-01
-1.08733475e+00 3.79261136e-01 -3.38616893e-02 1.05152801e-01
-1.14966261e+00 -7.67578900e-01 -8.35487783e-01 2.39584804e-01
1.93940684e-01 -7.60418952e-01 1.02331805e+00 -8.32936466e-01
-1.32202995e+00 2.68803746e-01 -1.99156985e-01 -5.57381332e-01
6.02510929e-01 -4.52732965e-02 -8.69987547e-01 7.53953695e-01
-2.72025555e-01 5.81655562e-01 9.93374228e-01 -1.06468630e+00
-5.65224588e-01 -5.72425453e-03 -7.64065236e-02 1.83922961e-01
-4.06640112e-01 3.45153242e-01 -8.04488480e-01 -1.02002609e+00
1.88861966e-01 -6.06132627e-01 -1.19607158e-01 3.39763492e-01
-1.92679971e-01 3.47571075e-01 1.05623281e+00 -1.29288137e+00
1.09939921e+00 -2.28803420e+00 1.51744053e-01 -7.01290444e-02
5.45669079e-01 2.95711368e-01 -5.07502437e-01 8.16125646e-02
2.93961782e-02 5.79880960e-02 -2.21468568e-01 -2.81280547e-01
-3.88703614e-01 2.61783868e-01 -1.47521228e-01 6.79640710e-01
-4.62260693e-01 4.38005358e-01 -7.37102807e-01 -3.30273360e-01
4.86552007e-02 1.18263984e+00 -7.78623462e-01 2.46056393e-02
3.83283734e-01 4.10237521e-01 -1.15642555e-01 6.47426546e-01
1.20844197e+00 -2.12500319e-01 4.21988547e-01 -5.49190998e-01
2.82375097e-01 -2.08763838e-01 -9.34411347e-01 1.69976544e+00
-3.55962813e-01 1.20692408e+00 7.04594851e-01 -7.86603451e-01
3.21852356e-01 7.70066023e-01 5.44581234e-01 -9.11853611e-01
7.98972920e-02 -2.46321643e-03 -4.42269772e-01 -6.77534223e-01
8.23596954e-01 9.06410217e-02 3.47443908e-01 3.46724004e-01
-2.02161014e-01 6.09270573e-01 -2.70436853e-02 5.16018569e-01
1.72762048e+00 -3.41484725e-01 -3.20304751e-01 5.51349260e-02
2.04304293e-01 -6.20072067e-01 7.43280113e-01 7.70025074e-01
-2.96541065e-01 7.39690363e-01 6.64598867e-02 -2.19821095e-01
-1.17214632e+00 -1.21370566e+00 1.12793528e-01 7.40963578e-01
6.36312544e-01 -2.29439139e-01 -8.55379403e-01 -1.23600110e-01
-3.27766478e-01 5.04028261e-01 8.02571848e-02 -5.71113899e-02
-6.96013093e-01 -9.60703731e-01 7.62509465e-01 3.17720443e-01
6.47206068e-01 -7.98524857e-01 -3.59292120e-01 3.88334095e-01
-7.91361630e-01 -1.57418895e+00 -1.07540727e+00 -5.22246838e-01
-1.04057944e+00 -7.89385796e-01 -8.38501692e-01 -7.71768391e-01
6.89483643e-01 9.39955354e-01 1.02284443e+00 2.91852087e-01
-1.98867500e-01 5.66329539e-01 -4.71786529e-01 5.43228030e-01
-4.99619395e-01 -5.78584373e-01 2.84345716e-01 -1.11861952e-01
-1.68450013e-01 -9.38690603e-01 -8.12255800e-01 4.70192432e-01
-1.25326586e+00 1.77798979e-02 2.36370817e-01 6.82161272e-01
3.04833114e-01 6.08766675e-01 6.79033041e-01 -1.90251708e-01
6.33763492e-01 -4.94735003e-01 -3.52433026e-01 2.39568502e-01
-2.91059852e-01 -2.43190572e-01 4.24171299e-01 -4.89991724e-01
-1.20846248e+00 -5.92432082e-01 -1.72144175e-01 -6.57558799e-01
1.59462854e-01 -5.25278226e-03 -3.64736356e-02 -2.62509227e-01
1.71879470e-01 3.35277587e-01 1.02827512e-01 -4.83749568e-01
9.20059159e-02 9.51987267e-01 9.25731599e-01 -1.48280501e-01
7.44189858e-01 7.82941282e-01 -2.73010939e-01 -1.00853980e+00
5.73663041e-02 -3.14394049e-02 1.54862523e-01 -3.50339532e-01
6.68718278e-01 -1.35627794e+00 -1.12245476e+00 7.50733972e-01
-1.28623772e+00 -3.28745157e-01 1.84630781e-01 6.57216966e-01
-3.62332731e-01 8.07572126e-01 -1.36274171e+00 -5.64013958e-01
-6.85133100e-01 -1.15224326e+00 6.65029764e-01 9.43675041e-02
2.76078403e-01 -7.80947804e-01 -4.00397897e-01 6.47633731e-01
8.11188579e-01 -8.08371678e-02 4.20077801e-01 1.38609810e-02
-8.23669136e-01 2.34652236e-01 -6.26802444e-01 5.45801520e-01
2.25447342e-01 -4.39357072e-01 -7.95874059e-01 -8.57805848e-01
1.81156537e-03 5.36011457e-02 9.95443404e-01 7.99870908e-01
1.39543355e+00 -5.88064551e-01 -2.68797338e-01 1.10596597e+00
1.35028136e+00 3.98207396e-01 1.21012676e+00 7.64172152e-02
4.99203116e-01 4.94550392e-02 3.37755889e-01 6.51795089e-01
3.44999790e-01 7.71900594e-01 5.20806313e-01 -4.22374979e-02
-3.64109159e-01 3.49586815e-01 8.73722553e-01 9.98016775e-01
-1.84848994e-01 -9.36821103e-01 -1.19385771e-01 3.92199337e-01
-1.57461166e+00 -1.30268252e+00 3.56128216e-02 2.11328340e+00
6.15937650e-01 -6.58123717e-02 -2.41306737e-01 -1.09218784e-01
1.15132856e+00 1.12001076e-01 -5.95204532e-01 -1.86928183e-01
-4.77425098e-01 -6.21518157e-02 8.53307664e-01 4.97426271e-01
-7.11174905e-01 5.49484670e-01 6.64400244e+00 1.10151505e+00
-7.17757702e-01 3.80253583e-01 7.29947567e-01 -3.69009614e-01
-2.74832666e-01 -4.08181876e-01 -2.16670528e-01 7.03368664e-01
1.08782208e+00 1.75789706e-02 1.01849079e+00 -5.54339448e-03
7.23890901e-01 2.99810879e-02 -8.23906124e-01 1.42729521e+00
3.26429784e-01 -1.66866994e+00 -5.38513474e-02 1.17189452e-01
5.65641582e-01 1.34956107e-01 1.32831074e-02 -2.62675524e-01
2.17398718e-01 -1.01378918e+00 4.76456761e-01 4.04996008e-01
1.27797699e+00 -7.61147916e-01 8.06179881e-01 -6.20850138e-02
-1.06288767e+00 -3.45847905e-02 -5.74825883e-01 2.88548559e-01
8.57218087e-01 6.17064059e-01 -3.36097181e-01 6.22933984e-01
1.02408922e+00 7.88037658e-01 1.55487224e-01 1.09364951e+00
3.22080016e-01 5.98291814e-01 -2.65014768e-01 8.91745389e-01
-2.24992082e-01 -1.85247898e-01 9.62078214e-01 1.26348710e+00
8.60238373e-01 5.65515041e-01 -2.09191665e-01 2.02889487e-01
-5.27826667e-01 -6.02808714e-01 -1.77546412e-01 3.01513165e-01
6.83359206e-01 9.24867570e-01 -4.35539752e-01 -3.84368747e-01
-2.54817516e-01 1.59984362e+00 -4.15386230e-01 6.41957939e-01
-1.10524559e+00 -2.51070917e-01 1.23183966e+00 -9.97963920e-02
4.53616709e-01 -1.57933190e-01 3.15978259e-01 -1.28973579e+00
-1.03361327e-02 -1.06832957e+00 2.00152278e-01 -1.08021343e+00
-9.91787136e-01 7.66379356e-01 -3.62517029e-01 -1.21223247e+00
7.98130408e-02 -1.48215696e-01 -2.63941020e-01 4.93069142e-01
-1.63584483e+00 -3.05644393e-01 -5.21922946e-01 8.52884650e-01
7.34201491e-01 -1.37744144e-01 3.85624975e-01 1.14597380e+00
-6.46448851e-01 8.81447136e-01 4.52297300e-01 1.28226370e-01
8.04773152e-01 -3.44154596e-01 4.66587692e-01 1.05453789e+00
-4.81362581e-01 6.35348111e-02 9.37466502e-01 -6.47865415e-01
-1.65957940e+00 -1.18298829e+00 2.87911296e-01 2.82669812e-01
3.28424960e-01 2.03086138e-02 -9.24741626e-01 5.88038683e-01
4.27044332e-01 2.34522358e-01 4.34609324e-01 -8.34774256e-01
-2.33148739e-01 -7.93724805e-02 -1.61561978e+00 5.48628032e-01
1.43285060e+00 -4.06898618e-01 7.14195520e-02 2.98028260e-01
1.11349583e+00 -5.94758868e-01 -8.65599155e-01 2.22764507e-01
5.10543108e-01 -1.05828917e+00 1.22065210e+00 -1.18280023e-01
3.11022609e-01 -3.07753056e-01 -4.49666411e-01 -1.06682491e+00
-3.13340783e-01 -1.15516293e+00 -1.71459466e-01 1.00992537e+00
-1.19578429e-01 -5.75931013e-01 7.06243217e-01 2.45301768e-01
-2.13260904e-01 -2.11831555e-01 -1.20969641e+00 -8.73977900e-01
-7.97721207e-01 -4.90898967e-01 3.78819853e-01 8.18380892e-01
-4.30445969e-01 -1.75267100e-01 -1.07197487e+00 5.22654593e-01
1.08337033e+00 -6.34641111e-01 2.78667271e-01 -6.57595336e-01
-4.00600553e-01 6.84176758e-02 -6.00946426e-01 -1.51225901e+00
1.95401702e-02 -6.07693791e-01 -3.85792069e-02 -1.51068401e+00
4.41692293e-01 -3.55328619e-01 -1.90959811e-01 1.57720432e-01
1.83100328e-01 6.68443859e-01 1.16791375e-01 2.08989814e-01
-6.66426837e-01 3.94643575e-01 1.15470552e+00 -7.64479265e-02
-7.67261684e-02 -1.95426509e-01 -5.92119575e-01 5.21638632e-01
6.35415673e-01 -4.41939056e-01 -3.71593505e-01 -9.73100483e-01
-4.28972282e-02 8.59241843e-01 4.67734963e-01 -7.85102069e-01
4.31124061e-01 1.09932743e-01 3.68543625e-01 -2.78322071e-01
5.37722707e-01 -8.70400727e-01 7.01976955e-01 3.35516602e-01
-6.31240681e-02 2.62419492e-01 1.54305622e-01 7.16927826e-01
2.08597314e-02 1.96974143e-01 6.16030693e-01 2.55065084e-01
-7.14091957e-01 4.43814695e-01 -6.56646311e-01 -9.31781828e-02
6.94477975e-01 -5.47381461e-01 -7.78243065e-01 -8.62439215e-01
-6.83483660e-01 2.03561664e-01 5.07269979e-01 3.00185442e-01
1.25095522e+00 -1.06868958e+00 -1.08222055e+00 2.99621504e-02
-7.12267101e-01 -3.62902582e-01 1.03699589e+00 9.39405799e-01
-7.69339025e-01 -1.53427854e-01 -2.57882118e-01 -3.94499272e-01
-1.78449225e+00 3.11655134e-01 3.31446916e-01 9.80999172e-02
-1.20865309e+00 7.59605587e-01 1.42330974e-01 2.27642313e-01
3.40536982e-01 1.63208380e-01 2.80145835e-02 -3.43445241e-01
1.09043396e+00 1.03255200e+00 -8.49735066e-02 -6.07163131e-01
-2.19589025e-01 2.88693249e-01 -9.00742412e-02 -8.64267051e-02
1.52581954e+00 -7.74441659e-01 -2.35898554e-01 -7.19914317e-01
1.20491409e+00 -5.98237589e-02 -1.57187414e+00 -1.15890056e-01
-6.06605291e-01 -7.95999110e-01 5.31252563e-01 -5.44920325e-01
-1.77772176e+00 2.48768046e-01 9.31081593e-01 -1.66097477e-01
1.68251204e+00 -3.23798984e-01 1.38409305e+00 -7.17506558e-02
4.95045751e-01 -7.49831915e-01 8.33493099e-03 6.26998544e-02
6.37699187e-01 -8.12775254e-01 4.31448184e-02 -5.55002451e-01
-4.17199284e-01 1.01739371e+00 6.76097572e-02 1.73392132e-01
4.34826493e-01 6.65257692e-01 -1.92979932e-01 1.52163714e-01
-8.63741636e-01 5.93217850e-01 -3.49421322e-01 7.10823894e-01
9.09981057e-02 7.92430155e-03 6.08183397e-03 1.72072411e-01
2.30144471e-01 1.97881028e-01 1.09942985e+00 6.05817139e-01
-4.61010784e-01 -7.98956513e-01 -6.28741324e-01 3.15517187e-01
-8.10881734e-01 -3.96832317e-01 6.08093619e-01 2.25132942e-01
-3.12148482e-01 1.53696644e+00 3.23420554e-01 -5.29298067e-01
-6.93482086e-02 -9.14648473e-01 4.70671773e-01 1.71971887e-01
-2.92579859e-01 2.31499046e-01 2.59555727e-01 -9.00464535e-01
-5.31959236e-01 -4.34860736e-01 -1.13534331e+00 -1.18187642e+00
-5.88204190e-02 -2.08095461e-02 5.53591490e-01 7.56034315e-01
6.03717327e-01 9.26538944e-01 6.89626753e-01 -8.44800770e-01
-2.61803776e-01 -5.12546301e-01 -6.58467948e-01 1.11697055e-01
7.65593648e-01 -5.79731651e-02 -6.17265582e-01 2.91513711e-01]
|
[11.152521133422852, -2.068045139312744]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.