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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
f7cc39de-8100-458a-92da-4f7cb9f9b177
|
adaptive-mask-sampling-and-manifold-to
| null | null |
https://ieeexplore.ieee.org/abstract/document/10097620
|
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10097620&tag=1
|
Adaptive Mask Sampling and Manifold to Euclidean Subspace Learning with Distance Covariance Representation for Hyperspectral Image Classification
|
For the abundant spectral and spatial information recorded in hyperspectral images (HSIs), fully exploring spectral-spatial relationships has attracted widespread attention in hyperspectral image classification (HSIC) community. However, there are still some intractable obstructs. For one thing, in the patch-based processing pattern, some spatial neighbor pixels are often inconsistent with the central pixel in land-cover class. For another thing, linear and nonlinear correlations between different spectral bands are vital yet tough for representing and excavating. To overcome these mentioned issues, an adaptive mask sampling and manifold to Euclidean subspace learning (AMS-M2ESL) framework is proposed for HSIC. Specifically, an adaptive mask based intra-patch sampling (AMIPS) module is firstly formulated for intra-patch sampling in an adaptive mask manner based on central spectral vector oriented spatial relationships. Subsequently, based on distance covariance descriptor, a dual channel distance covariance representation (DC-DCR) module is proposed for modeling unified spectral-spatial feature representations and exploring spectral-spatial relationships, especially linear and nonlinear interdependence in spectral domain. Furthermore, considering that distance covariance matrix lies on the symmetric positive definite (SPD) manifold, we implement a manifold to Euclidean subspace learning (M2ESL) module respecting Riemannian geometry of SPD manifold for high-level spectral-spatial feature learning. Additionally, we introduce an approximate matrix square-root (ASQRT) layer for efficient Euclidean subspace projection. Extensive experimental results on three popular HSI data sets with limited training samples demonstrate the superior performance of the proposed method compared with other state-of-the-art methods. The source code is available at https://github.com/lms-07/AMS-M2ESL.
|
['and Gongping Yang.', 'Yuwen Huang', 'Yikun Liu', 'Wei Li', 'Mingsong Li']
|
2023-04-07
| null | null | null |
ieee-transactions-on-geoscience-and-remote-14
|
['hyperspectral-image-segmentation']
|
['computer-vision']
|
[ 3.42205554e-01 -5.30824065e-01 2.08216589e-02 -2.03529254e-01
-5.05169451e-01 -4.76094306e-01 2.28208661e-01 -2.80595392e-01
-6.19618706e-02 3.20070952e-01 -3.02231900e-04 -2.27923408e-01
-7.40486264e-01 -6.63365126e-01 -3.20707619e-01 -1.08279335e+00
-2.44224042e-01 -2.63735592e-01 -1.28441051e-01 -1.71729714e-01
2.43571922e-01 6.08597457e-01 -1.41707337e+00 2.15603169e-02
1.30702555e+00 8.93952489e-01 4.94383454e-01 1.79868504e-01
4.71212640e-02 3.82932425e-02 1.94785461e-01 2.72575945e-01
5.74967325e-01 -3.39978635e-01 -3.91202360e-01 6.61228955e-01
3.64293694e-01 -6.23861849e-02 -3.60523015e-01 1.48168945e+00
2.98407018e-01 3.55097353e-01 8.35904658e-01 -1.06508541e+00
-7.70161450e-01 1.82755992e-01 -1.05251932e+00 -1.67833924e-01
-1.78763596e-03 1.69592574e-01 7.41462409e-01 -1.52071178e+00
1.83815300e-01 1.04757142e+00 5.29371560e-01 -1.70629546e-01
-1.26459026e+00 -5.47412992e-01 1.54595077e-01 5.08867979e-01
-2.06171894e+00 -1.57177120e-01 1.05782390e+00 -5.97534060e-01
4.05895531e-01 5.21425843e-01 6.74860835e-01 2.17384189e-01
-2.30704397e-01 5.83731592e-01 1.19349194e+00 -1.91661611e-01
-5.92112206e-02 1.43884689e-01 1.18638508e-01 4.82661366e-01
1.95311964e-01 4.94306833e-02 -1.65011391e-01 -2.93968022e-02
6.16770089e-01 2.65526623e-01 -8.12750399e-01 -5.80734730e-01
-1.19335067e+00 6.24772370e-01 7.35433221e-01 2.85570234e-01
-4.03789759e-01 -6.40611529e-01 -5.50160110e-02 6.99362382e-02
2.69497246e-01 3.13703623e-03 -1.29223898e-01 4.73085403e-01
-9.58435237e-01 -1.99858353e-01 3.47223490e-01 9.63049412e-01
1.32946289e+00 1.26767501e-01 1.40583307e-01 1.16326213e+00
5.30302227e-01 8.52225602e-01 3.31489801e-01 -7.43490219e-01
5.43871939e-01 8.79516065e-01 -4.73701442e-03 -1.55065000e+00
-4.50508237e-01 -6.44524574e-01 -1.39921248e+00 7.98539200e-04
-1.90361485e-01 1.54299363e-01 -2.99782872e-01 1.22214711e+00
5.12111604e-01 5.21519542e-01 4.65080701e-02 1.29128981e+00
4.66308773e-01 9.17795062e-01 -1.19309939e-01 -5.53656936e-01
1.15016580e+00 -5.79657078e-01 -3.89405698e-01 -5.62121607e-02
5.76320350e-01 -7.73713291e-01 1.05836403e+00 1.72861427e-01
-3.81430209e-01 -5.99686921e-01 -1.09942842e+00 1.49911836e-01
-3.13661337e-01 6.44493103e-01 4.12305385e-01 5.10438383e-01
-6.53920054e-01 4.36622024e-01 -5.70373356e-01 -2.44090855e-01
4.38602298e-01 2.34732106e-02 -5.15964329e-01 -3.48165870e-01
-9.23527539e-01 4.13589478e-01 5.36779583e-01 6.09941423e-01
-3.66580367e-01 -7.09495425e-01 -7.81603098e-01 -1.56556755e-01
3.69224668e-01 -3.07656210e-02 2.37431720e-01 -7.03063190e-01
-1.17411578e+00 5.32000065e-01 -1.57753929e-01 2.98403502e-01
3.09872348e-02 9.03679878e-02 -8.31581652e-01 2.04293445e-01
1.66306123e-01 2.21862361e-01 8.53390753e-01 -1.25287390e+00
-3.94372851e-01 -8.21047187e-01 -5.57866633e-01 6.64765894e-01
-7.04108238e-01 -3.11096221e-01 -2.50544012e-01 -6.47801042e-01
8.45670283e-01 -9.69742715e-01 -1.81098029e-01 1.26559630e-01
-5.01289666e-01 4.60794717e-02 1.05131209e+00 -7.44262934e-01
1.45297480e+00 -2.49402857e+00 3.56527299e-01 5.13115764e-01
-1.89060986e-01 3.66732091e-01 -3.72059435e-01 3.77688497e-01
-5.03158510e-01 -1.68436524e-02 -8.20821762e-01 2.00811550e-01
-2.34083474e-01 -2.15238720e-01 -1.22395195e-01 9.29747403e-01
3.16890806e-01 4.39906359e-01 -7.94532895e-01 -2.97171921e-01
4.65820521e-01 3.79852980e-01 -1.25960037e-01 -1.96463205e-02
2.64320344e-01 5.09554327e-01 -5.56182683e-01 8.18065405e-01
1.42508817e+00 -3.23892944e-02 -1.20404139e-01 -8.06242943e-01
-6.16297364e-01 -4.70783591e-01 -1.72461665e+00 1.80834401e+00
-1.38072342e-01 1.73296496e-01 3.37616563e-01 -1.23662603e+00
9.44927156e-01 6.09230101e-02 5.10782063e-01 -1.48914739e-01
-1.62128672e-01 5.07928848e-01 -5.67271076e-02 -5.93453884e-01
1.93586871e-01 4.92140390e-02 6.23754084e-01 2.41358727e-02
-2.83210963e-01 -2.22773150e-01 -3.18852216e-02 -5.29657677e-02
4.59609717e-01 1.07110590e-01 2.70854294e-01 -7.28766322e-01
1.26537406e+00 8.33962709e-02 5.81927359e-01 4.26569171e-02
-1.53335273e-01 5.72538257e-01 -6.30462766e-02 -1.10414162e-01
-8.38004172e-01 -7.53343344e-01 -6.08454168e-01 4.25074160e-01
4.58478868e-01 -1.65457517e-01 -4.76169854e-01 -3.34043175e-01
-8.82482529e-03 6.67681277e-01 -2.23542973e-01 -5.74712083e-02
-4.27802615e-02 -9.86720443e-01 9.95858312e-02 7.33815357e-02
7.96985149e-01 -5.52162647e-01 1.14980228e-01 9.52985957e-02
3.60420868e-02 -6.99022055e-01 -6.66683078e-01 -2.33120099e-01
-9.80796158e-01 -1.12215042e+00 -7.78104186e-01 -6.99351370e-01
8.13944221e-01 1.18959785e+00 1.99462891e-01 -1.37409866e-01
-4.75536466e-01 3.90109360e-01 -4.55050081e-01 -2.84505617e-02
1.54935613e-01 -1.76335886e-01 1.16746917e-01 6.20586634e-01
4.64500755e-01 -8.01278770e-01 -8.34054947e-01 6.21923327e-01
-1.13684464e+00 2.85477340e-01 8.06835055e-01 9.23692465e-01
7.57766068e-01 5.00395834e-01 8.41713324e-02 -3.50476742e-01
1.30865917e-01 -7.49324799e-01 -7.74182975e-01 3.28342766e-01
-6.37736440e-01 -4.16291118e-01 6.48535907e-01 -2.28241146e-01
-9.32308078e-01 3.32024634e-01 3.56844962e-01 -5.42555749e-01
-1.77688077e-01 8.66776586e-01 -5.58313847e-01 -2.87910312e-01
6.14181280e-01 9.00234401e-01 2.88928598e-01 -5.36120713e-01
2.45146051e-01 9.60083902e-01 3.40057969e-01 -3.49740237e-01
1.47127640e+00 5.62078834e-01 3.26935023e-01 -1.34395432e+00
-6.78415298e-01 -8.75131190e-01 -8.68177116e-01 -1.19288713e-01
6.35467529e-01 -1.18237126e+00 -3.81934106e-01 6.06293440e-01
-7.31607556e-01 3.71971689e-02 8.29206258e-02 8.41616571e-01
-1.60246193e-01 7.85924077e-01 -1.27864808e-01 -7.93991923e-01
-1.38028249e-01 -1.02321815e+00 9.86085355e-01 3.26701850e-01
6.10684574e-01 -8.34270954e-01 -1.69522509e-01 4.00669158e-01
3.21559757e-01 -9.00685415e-03 8.07846725e-01 -1.11098578e-02
-6.40423000e-01 -1.21902935e-01 -5.87138295e-01 6.71219409e-01
5.20953655e-01 -4.72795255e-02 -6.53943896e-01 -4.73256260e-01
1.36462107e-01 1.32040873e-01 5.91582656e-01 3.63608867e-01
1.39595318e+00 -7.93583021e-02 -1.53581604e-01 9.98418033e-01
1.80410266e+00 1.13339633e-01 5.02661645e-01 2.92720407e-01
9.73312318e-01 8.56475413e-01 8.72249782e-01 5.77135026e-01
1.79730520e-01 4.30703372e-01 4.49774683e-01 -1.63888052e-01
2.94086576e-01 -1.03462435e-01 3.17412406e-01 9.09710646e-01
-1.59174636e-01 2.90911078e-01 -6.88502073e-01 2.70209581e-01
-1.79371166e+00 -9.95278537e-01 -5.64651549e-01 2.29494977e+00
4.27920312e-01 -6.45922780e-01 -3.10561389e-01 2.10697532e-01
1.00517917e+00 3.79052520e-01 -8.29693019e-01 4.69154418e-01
-5.96962810e-01 -3.80999088e-01 6.22594118e-01 4.44376051e-01
-1.30157506e+00 7.03878522e-01 3.79003978e+00 1.19441354e+00
-1.05051053e+00 6.20713783e-03 4.18978751e-01 3.18093181e-01
-3.63747090e-01 1.86462879e-01 -5.03768086e-01 2.97901213e-01
2.45801657e-01 -1.08414762e-01 9.12186742e-01 6.87598050e-01
5.70668101e-01 -3.56204249e-02 -5.14101624e-01 1.43421686e+00
5.76587915e-02 -9.67453718e-01 1.41795933e-01 2.20036477e-01
8.02219212e-01 -2.93954946e-02 1.40117913e-01 1.33761764e-02
-4.74004835e-01 -6.99794292e-01 3.68350089e-01 5.20449758e-01
1.04401815e+00 -6.16901457e-01 4.28721726e-01 4.38111186e-01
-1.65084898e+00 -1.75783426e-01 -9.21296358e-01 3.10622394e-01
-1.45442382e-01 8.13261092e-01 -1.74575239e-01 1.15719783e+00
5.21273732e-01 1.40944183e+00 -7.29590952e-01 9.71774101e-01
-5.11364453e-02 4.06328470e-01 -2.79208660e-01 3.62360477e-01
3.48846853e-01 -1.24778783e+00 8.41634810e-01 9.06152129e-01
8.59175742e-01 6.64397001e-01 2.66108245e-01 9.49975491e-01
4.03317869e-01 6.29514098e-01 -6.25995159e-01 -1.03600182e-01
6.08044207e-01 1.65478814e+00 -4.18271869e-01 5.77890463e-02
-5.99604130e-01 9.63976741e-01 -1.92488074e-01 5.65040469e-01
-4.78278190e-01 -3.60322773e-01 5.78546345e-01 -5.79832913e-03
7.66874328e-02 -4.30804491e-01 2.63060275e-02 -1.34803462e+00
2.47112975e-01 -9.33549583e-01 2.18413875e-01 -8.26985061e-01
-1.25573313e+00 2.81657368e-01 -9.63946730e-02 -1.97367704e+00
6.64151430e-01 -8.06936264e-01 -5.99994481e-01 1.24827552e+00
-1.67098188e+00 -1.22872722e+00 -7.38187551e-01 8.54499519e-01
3.73670280e-01 -3.99533272e-01 7.98368692e-01 2.52847850e-01
-9.44582582e-01 8.89682472e-02 6.53058469e-01 -2.20365636e-02
4.11142975e-01 -9.76096630e-01 -3.91608953e-01 9.94597197e-01
-5.81433885e-02 6.04655147e-01 1.51239038e-01 -4.37898904e-01
-1.91151369e+00 -1.45848095e+00 5.61863482e-02 2.13814840e-01
8.43895197e-01 1.77199081e-01 -1.11267781e+00 2.88584083e-01
-2.03311190e-01 1.52214050e-01 9.20439959e-01 -2.32485741e-01
-3.14520657e-01 -4.74574119e-01 -7.81805754e-01 6.10351741e-01
9.54859555e-01 -7.53017008e-01 -2.86455080e-02 7.17952430e-01
3.17787409e-01 -3.96807082e-02 -1.08925951e+00 6.81834996e-01
2.42338866e-01 -8.62014353e-01 8.66076350e-01 -4.56096195e-02
4.42069024e-02 -1.16216195e+00 -6.44445539e-01 -1.34095359e+00
-6.78766251e-01 -3.37315142e-01 3.61749738e-01 1.16454935e+00
2.76133984e-01 -6.94500268e-01 6.14707649e-01 2.58530676e-01
-3.78453016e-01 -5.84286273e-01 -8.28196883e-01 -9.87587094e-01
-9.82147828e-02 -3.23741913e-01 5.57041526e-01 1.17066896e+00
-1.03772826e-01 1.37501255e-01 -1.96193382e-01 9.44945931e-01
9.29448605e-01 3.81706476e-01 6.41653359e-01 -1.22082210e+00
-1.25697672e-01 -4.98392820e-01 -4.65880603e-01 -9.26908076e-01
1.31135255e-01 -1.21227038e+00 -1.72492042e-01 -1.29753280e+00
2.80724883e-01 -4.31077868e-01 -3.19541425e-01 1.78334922e-01
-3.28426123e-01 5.06239347e-02 1.54917669e-02 7.40585446e-01
-8.28604326e-02 9.91493165e-01 1.25116503e+00 -4.14535493e-01
-3.34430784e-01 -1.21374898e-01 -5.25311172e-01 4.04910445e-01
6.67791963e-01 -4.78848554e-02 -5.28036594e-01 -2.77508676e-01
-8.30688179e-02 -3.04637644e-02 4.02211845e-01 -1.13691115e+00
2.91488528e-01 -4.95460272e-01 3.09300840e-01 -7.66287327e-01
1.89477205e-01 -1.06841123e+00 5.21779180e-01 2.62612551e-01
1.69833556e-01 -5.31246066e-01 5.91763183e-02 7.17168927e-01
-4.25415903e-01 -2.40942717e-01 8.85084689e-01 8.90393108e-02
-8.03791702e-01 7.75724292e-01 -8.37417915e-02 -4.72087979e-01
9.72551584e-01 -4.82338101e-01 -1.41115308e-01 6.11007884e-02
-4.43487227e-01 4.18778688e-01 4.66277391e-01 1.00489147e-01
8.30990493e-01 -1.54027033e+00 -1.02156544e+00 4.51690704e-01
5.51004171e-01 1.34509638e-01 8.20252538e-01 1.18911946e+00
-5.58440804e-01 2.64853418e-01 1.86759029e-02 -8.30775797e-01
-1.10642445e+00 4.33113903e-01 4.15063709e-01 3.81999642e-01
-3.92464668e-01 6.96243107e-01 3.24744999e-01 -6.52715266e-01
-2.48006225e-01 2.05723748e-01 -3.40818435e-01 1.49032816e-01
4.53942180e-01 5.09691536e-01 -1.22473039e-01 -1.15670490e+00
-2.92609394e-01 1.01454830e+00 3.70041370e-01 1.69644371e-01
1.36109245e+00 -2.91958541e-01 -6.66956365e-01 3.35251331e-01
1.41168702e+00 -1.57057926e-01 -1.19685364e+00 -5.53214014e-01
-4.33038324e-02 -7.38205492e-01 3.34644645e-01 -2.23942161e-01
-9.16697800e-01 9.42572534e-01 9.64350343e-01 1.88902337e-02
1.43759692e+00 -5.63513398e-01 2.18383923e-01 4.46470350e-01
3.16765636e-01 -1.05947256e+00 -3.62761766e-01 2.85475641e-01
1.07297421e+00 -1.39347816e+00 3.08684647e-01 -7.65525281e-01
-5.32601833e-01 1.26908910e+00 5.45634151e-01 4.17771526e-02
1.11801231e+00 -5.24897754e-01 -1.00093916e-01 -1.69230193e-01
3.67301553e-02 -1.64494827e-01 5.24833024e-01 4.80128914e-01
3.69814962e-01 3.85351360e-01 -3.02543968e-01 1.88153893e-01
8.70336443e-02 -5.17462492e-01 2.61201382e-01 4.60523814e-01
-5.33267796e-01 -8.12784076e-01 -7.04705536e-01 4.08241004e-01
3.04814219e-01 -2.41871282e-01 -3.06925159e-02 4.91483301e-01
1.68559611e-01 1.06278253e+00 -2.20644072e-01 -5.34617960e-01
2.07527354e-01 -9.56680179e-02 5.24502955e-02 -6.46968484e-01
3.48804504e-01 4.66833383e-01 -5.32503724e-01 -4.58681971e-01
-5.85951507e-01 -9.74870741e-01 -1.06637025e+00 4.47652265e-02
-6.10435665e-01 3.07559460e-01 7.05653608e-01 6.44090652e-01
3.26225489e-01 -5.15909977e-02 1.29844475e+00 -1.07714438e+00
-6.85704827e-01 -1.03427529e+00 -1.44710314e+00 2.24911332e-01
2.32193060e-02 -5.29597640e-01 -6.37285709e-01 -5.20398840e-02]
|
[10.062873840332031, -1.8647524118423462]
|
5f02b1c3-2056-421e-bb91-3bfb038e6236
|
eventplus-a-temporal-event-understanding
|
2101.04922
| null |
https://arxiv.org/abs/2101.04922v2
|
https://arxiv.org/pdf/2101.04922v2.pdf
|
EventPlus: A Temporal Event Understanding Pipeline
|
We present EventPlus, a temporal event understanding pipeline that integrates various state-of-the-art event understanding components including event trigger and type detection, event argument detection, event duration and temporal relation extraction. Event information, especially event temporal knowledge, is a type of common sense knowledge that helps people understand how stories evolve and provides predictive hints for future events. EventPlus as the first comprehensive temporal event understanding pipeline provides a convenient tool for users to quickly obtain annotations about events and their temporal information for any user-provided document. Furthermore, we show EventPlus can be easily adapted to other domains (e.g., biomedical domain). We make EventPlus publicly available to facilitate event-related information extraction and downstream applications.
|
['Nanyun Peng', 'Rujun Han', 'Shikhar Singh', 'Nuan Wen', 'Kung-Hsiang Huang', 'Mu Yang', 'Jiao Sun', 'Mingyu Derek Ma']
|
2021-01-13
| null |
https://aclanthology.org/2021.naacl-demos.7
|
https://aclanthology.org/2021.naacl-demos.7.pdf
|
naacl-2021-4
|
['temporal-relation-extraction']
|
['natural-language-processing']
|
[ 5.66234738e-02 2.28356346e-01 -4.50567693e-01 -3.29337418e-01
-5.76177001e-01 -8.07440042e-01 5.66177130e-01 1.36779225e+00
-3.40919614e-01 8.31309617e-01 7.39162266e-01 -3.28823656e-01
-2.70147860e-01 -1.01606941e+00 -4.86642867e-01 5.73870577e-02
-5.01170874e-01 3.57771158e-01 6.75076127e-01 -1.03066161e-01
9.63188335e-02 4.49105054e-01 -1.39052010e+00 6.38532639e-01
5.27459741e-01 6.45394206e-01 9.49983820e-02 7.20388412e-01
-1.51174992e-01 1.09580493e+00 -6.97247028e-01 -1.64976820e-01
-5.52786410e-01 -5.25010586e-01 -9.82070088e-01 -6.63343489e-01
-8.30417454e-01 -2.97105670e-01 -2.07273424e-01 3.92687470e-01
3.27559918e-01 3.24066281e-01 3.16742063e-01 -1.39354360e+00
-1.13176763e-01 9.34369028e-01 -2.08467856e-01 8.95184159e-01
9.40338373e-01 2.84165815e-02 7.82176614e-01 -4.89775211e-01
1.31985080e+00 9.46538925e-01 6.02485895e-01 3.16145331e-01
-7.90267706e-01 -5.92005432e-01 2.13668749e-01 6.22134566e-01
-1.22105384e+00 -1.30402610e-01 2.87422389e-01 -4.56805646e-01
1.61915636e+00 4.00004596e-01 1.03584123e+00 1.17137861e+00
2.79823303e-01 9.18928146e-01 4.36068237e-01 -3.40325475e-01
3.13672066e-01 -3.43795657e-01 3.64982903e-01 4.97687876e-01
-1.41012818e-01 1.53784171e-01 -1.00247478e+00 -3.71025026e-01
5.10820806e-01 1.21118248e-01 -1.33083448e-01 7.22699523e-01
-1.47446263e+00 3.37097794e-01 9.85760242e-02 3.09499145e-01
-6.53307498e-01 1.87116906e-01 9.21090007e-01 -1.21966012e-01
3.94848764e-01 1.57082677e-01 -9.64330614e-01 -6.45383477e-01
-5.92168093e-01 6.10022783e-01 9.62116599e-01 8.95700336e-01
1.63122118e-01 -5.95759869e-01 -4.64822114e-01 3.48757923e-01
1.61237955e-01 -2.87037104e-01 4.27802503e-01 -5.34155488e-01
1.31309062e-01 9.34423149e-01 3.55796337e-01 -6.48742914e-01
-8.14412177e-01 -2.10619979e-02 -1.72143415e-01 -4.46979314e-01
3.68263036e-01 -1.90183192e-01 -5.94568729e-01 1.78446770e+00
5.82109272e-01 7.77595460e-01 9.49400738e-02 4.35661614e-01
1.25787532e+00 6.24704361e-01 5.80588996e-01 -6.62681639e-01
2.02416801e+00 -2.28451565e-01 -1.25566554e+00 -7.18115270e-02
8.58687580e-01 -7.04333782e-01 4.82544482e-01 2.09013820e-01
-8.99643183e-01 1.27713159e-01 -8.78903747e-01 -1.26812279e-01
-1.02785611e+00 -3.60034317e-01 9.54826057e-01 2.08605230e-02
-2.96595484e-01 6.79633975e-01 -1.41307163e+00 -9.26274359e-01
4.42319691e-01 -4.17975299e-02 -2.00168476e-01 2.72287309e-01
-1.67844582e+00 1.11396837e+00 1.03937709e+00 -4.46015656e-01
-8.74793053e-01 -1.34950674e+00 -9.71922159e-01 -2.64017489e-02
7.50711083e-01 -7.02382028e-01 1.70447338e+00 7.22546801e-02
-8.80799830e-01 6.85939431e-01 -5.71934879e-01 -6.58102095e-01
8.33096504e-02 -2.91885167e-01 -1.11367750e+00 1.16435207e-01
1.87938929e-01 1.58082321e-01 -1.41825870e-01 -2.04376787e-01
-7.71054506e-01 -3.23473513e-01 -1.99003324e-01 3.58249880e-02
1.32828191e-01 9.25354004e-01 -5.18135905e-01 -8.11363578e-01
-1.99862868e-01 -1.91407889e-01 -4.20313239e-01 1.51833398e-02
-4.12395000e-01 -5.19065857e-01 7.20091224e-01 -7.54354119e-01
1.94345951e+00 -1.85992217e+00 -3.79928499e-01 -1.54145300e-01
1.11256257e-01 -3.17399353e-01 6.15148604e-01 9.92198765e-01
-5.49623609e-01 2.10606098e-01 5.25243022e-02 3.83072436e-01
-8.10259730e-02 2.09744141e-01 -3.02055895e-01 1.82670802e-02
4.16654170e-01 9.79749501e-01 -1.40623593e+00 -5.68036377e-01
3.21564436e-01 3.72002751e-01 -1.73289552e-01 9.61553901e-02
-6.92785084e-01 5.43185115e-01 -8.09896886e-01 6.08468175e-01
-1.52478172e-02 -5.64858317e-01 1.13863930e-01 -5.92387207e-02
-3.49064589e-01 9.07150507e-01 -1.19393957e+00 1.88532305e+00
-4.44039628e-02 4.99764472e-01 -7.51374781e-01 -5.39222598e-01
4.32926923e-01 9.50601220e-01 7.46141434e-01 -4.29761648e-01
-3.02496506e-03 -1.69004560e-01 -4.59412575e-01 -7.44488001e-01
4.49949920e-01 -6.70674071e-02 -3.39194179e-01 7.37396240e-01
-5.75941280e-02 5.15469611e-01 8.32532763e-01 4.13411796e-01
1.56684351e+00 3.29543173e-01 1.28382099e+00 1.34581968e-01
9.28297788e-02 3.82231891e-01 9.06681120e-01 4.20993030e-01
-2.84858868e-02 2.78356791e-01 7.28297770e-01 -4.77027744e-01
-5.25897622e-01 -1.12126803e+00 -3.10311317e-01 9.70374525e-01
-2.05792993e-01 -1.26818478e+00 -1.83718145e-01 -4.42254305e-01
-2.92064518e-01 1.03328085e+00 -7.83294618e-01 -3.14549208e-02
-5.88404119e-01 -9.82289851e-01 7.08503723e-01 1.06830025e+00
1.57032281e-01 -1.12913370e+00 -9.70153153e-01 7.98637092e-01
-5.97218752e-01 -1.01057720e+00 -3.52405667e-01 2.38515198e-01
-5.26618361e-01 -1.42373693e+00 -2.11040318e-01 -3.78214121e-01
2.80394942e-01 -6.44552231e-01 1.32918131e+00 -3.13141346e-01
-6.48925722e-01 2.32368127e-01 -3.99233222e-01 -1.08072376e+00
-3.20439547e-01 -4.02552485e-01 -3.08802158e-01 -5.36745906e-01
5.33628523e-01 -5.26938975e-01 -7.09126949e-01 2.90276289e-01
-9.96039808e-01 1.89344421e-01 -2.50124365e-01 3.40310961e-01
7.02723861e-01 2.59964883e-01 6.23987675e-01 -8.58139277e-01
5.58282733e-01 -9.36691403e-01 -4.48276162e-01 4.58321452e-01
-3.95118982e-01 -1.92697376e-01 2.19058990e-01 -4.63747531e-01
-1.31900191e+00 -7.03737065e-02 -2.28299364e-01 4.57869798e-01
-4.40137982e-01 1.08083165e+00 -9.29419100e-02 1.12079442e+00
8.53628814e-01 1.01806939e-01 -7.29680061e-01 -3.61278176e-01
5.18283188e-01 1.99365109e-01 8.21921289e-01 -3.20990711e-01
-2.85360329e-02 4.34779495e-01 -1.20569579e-01 -2.39304185e-01
-6.76585555e-01 -6.33195758e-01 -4.53745902e-01 -1.77890942e-01
1.05533051e+00 -7.30712235e-01 -1.14797223e+00 1.50946781e-01
-1.35810304e+00 -2.53003150e-01 -5.44625223e-01 5.89844823e-01
-3.94874692e-01 -1.31206185e-01 -6.72508538e-01 -6.74288869e-01
-1.73928767e-01 -3.63471866e-01 9.42732215e-01 5.54350019e-01
-1.05604327e+00 -1.15962648e+00 1.08271137e-01 -3.37886810e-01
-4.24756072e-02 8.07004094e-01 8.47974479e-01 -1.11049092e+00
-4.22379106e-01 -2.49488622e-01 1.19387694e-01 -1.02850008e+00
1.45252213e-01 1.18579499e-01 -4.95084733e-01 5.12116730e-01
-4.35660452e-01 1.21416330e-01 6.20444596e-01 4.63016778e-01
1.13103878e+00 -4.88048404e-01 -1.13243210e+00 1.67031616e-01
9.79601324e-01 6.95207596e-01 5.82733870e-01 3.28488111e-01
6.85171112e-02 4.42900985e-01 8.97898197e-01 1.01500714e+00
5.53930640e-01 6.30323768e-01 9.65255219e-03 -1.24694772e-01
1.89355984e-01 -4.26620901e-01 4.74150851e-03 -2.59747021e-02
-1.58240423e-02 -6.82996988e-01 -1.07534742e+00 8.83423269e-01
-2.25505590e+00 -1.24539053e+00 -3.31230998e-01 2.00122070e+00
1.20705724e+00 9.47237387e-02 3.57446492e-01 1.71840101e-01
6.17736101e-01 -1.89684078e-01 -6.13771021e-01 -9.41922441e-02
-1.27527416e-01 1.36201859e-01 2.34283060e-01 9.72978771e-02
-1.03897595e+00 8.59734058e-01 7.20850801e+00 5.55258811e-01
-5.80645084e-01 2.39752367e-01 4.07596171e-01 -2.67682433e-01
-1.22819930e-01 1.44958839e-01 -7.16438293e-01 4.25215781e-01
1.36971843e+00 -9.83189642e-01 -1.02630053e-02 4.36393976e-01
7.73982763e-01 -3.27636123e-01 -1.41771400e+00 6.95849955e-01
-6.16130650e-01 -1.92573321e+00 -2.88566262e-01 -4.49405730e-01
5.89049757e-02 -1.40298069e-01 -6.19430244e-01 -2.43251249e-02
8.47308636e-01 -7.93279171e-01 5.09397030e-01 7.79235601e-01
5.83201528e-01 -6.32101655e-01 4.26039666e-01 1.63065717e-01
-1.52054429e+00 2.02091947e-01 4.17826802e-01 -2.25089230e-02
1.02500057e+00 9.41622913e-01 -1.36810350e+00 9.64030206e-01
9.19495463e-01 1.15085566e+00 -2.71429807e-01 1.08612120e+00
-4.89524305e-01 7.21744299e-01 -6.25710547e-01 1.73410892e-01
-5.66102147e-01 5.91409028e-01 7.48597860e-01 1.56624055e+00
3.54588449e-01 8.06417167e-01 1.01353459e-01 8.10903728e-01
2.75252819e-01 -3.70518044e-02 -5.11305690e-01 -4.57315743e-01
6.91431165e-01 9.23368037e-01 -1.24647355e+00 -7.30343699e-01
-3.85239810e-01 8.62769663e-01 -1.48228079e-01 1.73014194e-01
-1.02241647e+00 -4.67134714e-01 6.20243847e-01 6.17283136e-02
1.31676912e-01 1.10198662e-01 -1.12864427e-01 -9.27614629e-01
-1.56469807e-01 -3.64884585e-01 1.45790482e+00 -1.15739274e+00
-1.00420761e+00 2.63633192e-01 5.16281366e-01 -1.03377080e+00
-6.30821347e-01 -7.90670738e-02 -8.29197526e-01 6.15595937e-01
-8.26218724e-01 -1.02130079e+00 -1.50701612e-01 7.02740133e-01
5.04036367e-01 5.54400980e-01 1.00527740e+00 3.57902408e-01
-7.40732491e-01 6.08929805e-02 -6.05829656e-01 1.87951624e-01
7.05961287e-01 -1.25746047e+00 5.87962449e-01 8.94292116e-01
-1.72281548e-01 7.65477717e-01 9.14338231e-01 -1.33287632e+00
-9.70070601e-01 -1.19144332e+00 1.42027068e+00 -6.79689705e-01
9.62889254e-01 -1.17593683e-01 -8.76340449e-01 1.18593216e+00
7.12387115e-02 -1.94980100e-01 1.01328146e+00 3.04207146e-01
-1.64745480e-01 2.30820045e-01 -9.18773055e-01 6.86618686e-01
1.31061172e+00 -2.94295728e-01 -1.07089460e+00 6.32375300e-01
9.45701718e-01 -8.20498466e-01 -1.26607406e+00 2.66567856e-01
4.59041387e-01 -3.54531229e-01 9.84093785e-01 -9.15672600e-01
3.57437700e-01 -6.20187640e-01 3.45261067e-01 -8.65386426e-01
-3.70111354e-02 -9.67506826e-01 -6.82882845e-01 1.50234377e+00
6.87869430e-01 -4.39903617e-01 1.96333438e-01 8.24639022e-01
-1.52701423e-01 -3.43697757e-01 -6.65087640e-01 -5.86310804e-01
-8.89367640e-01 -9.84379053e-01 9.24769461e-01 1.11773539e+00
8.98009181e-01 2.89917946e-01 1.88994780e-02 3.68362427e-01
-4.85990718e-02 1.08306944e-01 5.84849194e-02 -1.21529293e+00
-2.23983109e-01 -2.86150515e-01 -2.30084985e-01 -4.23033625e-01
-3.54243606e-01 -9.33684766e-01 -2.58170873e-01 -1.90643668e+00
2.59274065e-01 1.24746822e-01 -5.11181653e-01 1.09468794e+00
-3.27186793e-01 -1.67314157e-01 -5.63176394e-01 -2.60589927e-01
-7.28125155e-01 5.06472662e-02 7.46665418e-01 1.85627133e-01
-5.30412257e-01 -6.91654161e-02 -6.47500753e-01 8.55172038e-01
5.70213318e-01 -7.11845934e-01 -4.08554614e-01 1.38760030e-01
6.55977488e-01 5.09943426e-01 3.97014827e-01 -5.89089990e-01
4.53505009e-01 -4.53004509e-01 5.00928521e-01 -1.20430017e+00
-4.84359525e-02 -4.18629676e-01 7.67454803e-01 4.63006109e-01
-4.94841278e-01 1.80002585e-01 6.78418934e-01 7.26362467e-01
-1.35200813e-01 9.80381817e-02 1.54856950e-01 -2.51256764e-01
-9.75539088e-01 1.89249203e-01 -8.44930828e-01 1.48962930e-01
1.28954327e+00 3.03760096e-02 -7.79102087e-01 -2.73139805e-01
-1.33607948e+00 4.76783872e-01 -2.62408634e-03 5.95256329e-01
5.65832853e-01 -1.19173384e+00 -5.78241050e-01 -4.06367213e-01
5.02794445e-01 -1.98580682e-01 3.07677686e-01 1.07563806e+00
-1.57552853e-01 4.13613588e-01 6.08414151e-02 -2.47552633e-01
-1.40078604e+00 7.63343751e-01 3.13604623e-03 -5.48572719e-01
-1.07483804e+00 7.38795877e-01 2.48766821e-02 9.28497314e-02
1.49114564e-01 -6.55804694e-01 -4.16457087e-01 2.19670340e-01
1.30925095e+00 3.57778311e-01 5.31686004e-03 7.41618052e-02
-9.97701168e-01 -1.68502569e-01 -3.99844646e-02 -4.45083946e-01
1.46426296e+00 -5.31076975e-02 -1.10974677e-01 6.98288560e-01
4.67787117e-01 -2.25804031e-01 -7.71765530e-01 -8.36492404e-02
4.28088367e-01 8.38131160e-02 -3.03324193e-01 -1.33338881e+00
-3.31269354e-01 2.66634464e-01 2.15207636e-01 2.35107481e-01
1.40187037e+00 5.40183365e-01 7.52013505e-01 -5.16425669e-02
1.41730264e-01 -9.70055759e-01 -3.30459744e-01 3.70486498e-01
9.52565253e-01 -8.68745863e-01 2.49460295e-01 -7.47948825e-01
-4.88963395e-01 1.01257634e+00 3.18621486e-01 5.90986133e-01
9.24933553e-01 6.16804183e-01 -3.16799104e-01 -5.39287329e-01
-1.30939782e+00 -5.21564186e-01 3.25325608e-01 6.36586368e-01
8.30575168e-01 -4.50666212e-02 -7.06019580e-01 1.30825329e+00
-7.40893111e-02 5.92650950e-01 2.54886270e-01 1.21685266e+00
6.55292347e-02 -1.25534642e+00 -3.33020449e-01 3.80174011e-01
-8.22453022e-01 -2.46319473e-01 -4.14348871e-01 6.90396547e-01
2.21330300e-01 9.71335828e-01 2.63816237e-01 5.67833558e-02
4.52481717e-01 3.05106431e-01 2.46475115e-01 -6.34503841e-01
-7.09758580e-01 1.49753258e-01 7.36185372e-01 -9.60166693e-01
-3.39551777e-01 -9.00132895e-01 -2.20168948e+00 1.56732649e-02
2.06553042e-02 2.04977561e-02 2.89954990e-01 1.12686396e+00
6.52952671e-01 1.03979349e+00 -2.97535270e-01 3.76989692e-02
8.15952539e-01 -7.86724329e-01 -2.16226995e-01 3.18103373e-01
-2.24895645e-02 -6.28704309e-01 2.70120591e-01 6.36124671e-01]
|
[8.830374717712402, 9.132889747619629]
|
f8a29f68-40d1-43d1-8a90-071923d33bed
|
a-comparative-study-for-single-image-blind
| null | null |
http://openaccess.thecvf.com/content_cvpr_2016/html/Lai_A_Comparative_Study_CVPR_2016_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2016/papers/Lai_A_Comparative_Study_CVPR_2016_paper.pdf
|
A Comparative Study for Single Image Blind Deblurring
|
Numerous single image blind deblurring algorithms have been proposed to restore latent sharp images under camera motion. However, these algorithms are mainly evaluated using either synthetic datasets or few selected real blurred images. It is thus unclear how these algorithms would perform on images acquired "in the wild" and how we could gauge the progress in the field. In this paper, we aim to bridge this gap. We present the first comprehensive perceptual study and analysis of single image blind deblurring using real-world blurred images. First, we collect a dataset of real blurred images and a dataset of synthetically blurred images. Using these datasets, we conduct a large-scale user study to quantify the performance of several representative state-of-the-art blind deblurring algorithms. Second, we systematically analyze subject preferences, including the level of agreement, significance tests of score differences, and rationales for preferring one method over another. Third, we study the correlation between human subjective scores and several full-reference and no-reference image quality metrics. Our evaluation and analysis indicate the performance gap between synthetically blurred images and real blurred image and sheds light on future research in single image blind deblurring.
|
['Ming-Hsuan Yang', 'Jia-Bin Huang', 'Wei-Sheng Lai', 'Narendra Ahuja', 'Zhe Hu']
|
2016-06-01
| null | null | null |
cvpr-2016-6
|
['single-image-blind-deblurring']
|
['computer-vision']
|
[ 2.71464020e-01 -6.19758308e-01 2.15204224e-01 -2.91361302e-01
-7.09012330e-01 -8.51076305e-01 5.75786710e-01 -4.67710525e-01
-4.52053934e-01 7.16588616e-01 7.28334963e-01 -3.58193547e-01
-2.82816708e-01 1.46437809e-01 -4.56525445e-01 -4.64706481e-01
9.01031345e-02 -2.75091141e-01 -2.51772883e-03 3.16858217e-02
6.48201704e-01 1.42011970e-01 -1.50958610e+00 1.73095062e-01
1.21155727e+00 5.38207412e-01 6.36524498e-01 9.75052655e-01
6.78055048e-01 7.85153210e-01 -7.54563689e-01 -3.86147648e-01
3.60613257e-01 -4.38279241e-01 -8.38964760e-01 2.79688776e-01
9.30813015e-01 -8.59350860e-01 -5.55412650e-01 1.43864644e+00
9.03051674e-01 7.13666528e-02 3.38995397e-01 -7.60445237e-01
-1.54713607e+00 2.44042680e-01 -4.56518829e-01 6.81251347e-01
7.41457999e-01 7.08813667e-01 5.90853870e-01 -8.29308450e-01
4.35990036e-01 1.02069819e+00 5.12355387e-01 5.18564880e-01
-1.14387262e+00 -3.30714017e-01 -3.66909504e-01 4.66421127e-01
-1.16420126e+00 -1.03687561e+00 3.23493868e-01 -6.40177906e-01
5.43743789e-01 4.82857734e-01 2.43928596e-01 1.07525480e+00
2.43613459e-02 4.15506333e-01 1.83507836e+00 -4.59931821e-01
2.89781064e-01 7.06407949e-02 3.20200831e-01 2.03169793e-01
6.58830881e-01 4.61164385e-01 -4.90925878e-01 -1.10254504e-01
8.27104330e-01 -2.52810627e-01 -1.19304955e+00 -4.49987888e-01
-1.43329418e+00 3.29036057e-01 4.01637852e-01 2.36010566e-01
-3.48559290e-01 -4.06933576e-02 -2.56015919e-02 4.39341903e-01
4.96250987e-01 7.97824025e-01 -2.38162339e-01 -2.27679700e-01
-1.24425042e+00 1.81697518e-01 5.36747575e-01 5.68657339e-01
4.35228765e-01 -1.83301196e-01 -4.40890789e-01 1.15481162e+00
1.26601025e-01 5.48395813e-01 7.49996126e-01 -1.35638666e+00
9.82656330e-02 -3.19534868e-01 9.68511999e-01 -8.09603393e-01
1.60667360e-01 -1.77404448e-01 -5.44113815e-01 5.33524454e-01
5.17172217e-01 -1.65887371e-01 -1.01871598e+00 1.40305245e+00
-5.51859438e-01 3.61943953e-02 -4.78350930e-02 1.72848177e+00
4.95835394e-01 1.59847796e-01 -2.15800703e-01 -3.65380496e-01
1.41636980e+00 -1.02496934e+00 -8.04757178e-01 -4.75770772e-01
-1.50727913e-01 -1.17243171e+00 1.24513745e+00 4.26848739e-01
-1.40137720e+00 -8.02633762e-01 -1.00934994e+00 4.45215106e-02
1.52511466e-02 3.64600718e-01 3.89812559e-01 1.01682889e+00
-1.62536693e+00 3.51246536e-01 -5.17958045e-01 -5.64453602e-01
2.29334831e-01 7.43823349e-02 -2.21759543e-01 -5.51592350e-01
-1.18297470e+00 1.35960591e+00 -1.03863552e-01 3.96356210e-02
-9.91743803e-01 -5.77810764e-01 -7.26768672e-01 -1.44533306e-01
-8.54460150e-02 -9.60101724e-01 1.51757741e+00 -9.74118531e-01
-1.18720686e+00 1.02262068e+00 -4.65580642e-01 -4.47714955e-01
6.80070698e-01 -4.95736599e-01 -6.58215344e-01 1.25904262e-01
1.48850232e-01 5.66551387e-01 1.21166945e+00 -1.67410350e+00
-2.23364532e-01 -2.25891083e-01 9.80471596e-02 4.10671324e-01
-7.56658763e-02 3.37474257e-01 -3.29304636e-01 -9.61589277e-01
-1.45245805e-01 -7.58647025e-01 -3.48860659e-02 -1.17136247e-01
-3.41009572e-02 4.87238914e-01 5.67775965e-01 -1.00873911e+00
1.23092365e+00 -2.27135110e+00 -1.06633613e-02 -4.40201312e-01
3.34983230e-01 3.72279704e-01 -6.67557716e-02 5.76490015e-02
-2.27158323e-01 -1.31467342e-01 -3.09170663e-01 -2.07970500e-01
-1.39902994e-01 -3.74298543e-01 -4.19205070e-01 8.05732429e-01
-3.69769663e-01 8.30376387e-01 -1.01799369e+00 -1.57201231e-01
3.72104824e-01 3.09078693e-01 -3.72798026e-01 4.22162533e-01
2.37483099e-01 4.07879740e-01 2.48747900e-01 6.32629752e-01
1.03865218e+00 -2.46281400e-01 -1.93148360e-01 -5.41370034e-01
-2.82203734e-01 -9.59635824e-02 -7.93088973e-01 1.79578388e+00
-4.18760121e-01 1.25373542e+00 3.46922696e-01 -2.22699553e-01
2.94926763e-01 3.98078263e-01 -8.76681656e-02 -4.48818713e-01
-1.00637823e-01 3.14350367e-01 -2.08168905e-02 -9.12487388e-01
1.03912687e+00 -4.93329689e-02 3.95839334e-01 5.38607895e-01
-2.69641012e-01 -4.02866036e-01 1.06254490e-02 7.85536245e-02
1.01463854e+00 -1.88326180e-01 1.86143517e-01 -3.27632725e-01
3.71982545e-01 -2.63470653e-02 -2.65760750e-01 1.04221523e+00
-6.91927016e-01 1.28413403e+00 -2.40418643e-01 -1.67232931e-01
-9.02770042e-01 -1.31476545e+00 -1.18456282e-01 8.55721354e-01
7.71982729e-01 3.70000191e-02 -1.06806469e+00 -2.39059851e-01
-2.67097503e-01 8.00567865e-01 -5.99639654e-01 6.48540631e-02
-9.69325751e-02 -1.05743170e+00 3.05161625e-01 7.72399008e-02
6.87461793e-01 -8.78958941e-01 -7.49543488e-01 -3.72152358e-01
-5.60031533e-01 -1.00318134e+00 -1.03237903e+00 -5.63102782e-01
-6.87712014e-01 -1.04548550e+00 -1.36522532e+00 -7.06006050e-01
7.98908293e-01 1.12714803e+00 1.01302457e+00 -2.25299060e-01
-9.96148586e-02 5.60754716e-01 -2.96208054e-01 3.15737277e-01
-4.74994242e-01 -5.56573749e-01 3.47809017e-01 5.43667981e-03
2.45769933e-01 -2.60573924e-01 -1.14652014e+00 6.32435679e-01
-9.86264706e-01 -5.05564809e-02 8.33320677e-01 7.57702827e-01
-3.40279579e-01 3.57340835e-02 2.22594559e-01 -2.46339530e-01
1.48194313e+00 -3.16534698e-01 -4.78212237e-01 3.24549198e-01
-9.08154011e-01 -2.94337235e-02 3.83199751e-02 -5.79452991e-01
-1.38994944e+00 -5.33745468e-01 4.98890489e-01 -4.29764569e-01
-2.50283957e-01 3.00602466e-01 1.32176816e-01 -2.45519519e-01
1.12643397e+00 3.62876266e-01 -2.06281105e-03 -6.61340117e-01
6.45098746e-01 1.23010600e+00 1.14079511e+00 -3.09333712e-01
6.54079199e-01 3.32568914e-01 -8.42777729e-01 -6.70288146e-01
-4.34070408e-01 -4.79256749e-01 -2.57991314e-01 -2.58822590e-01
7.65342772e-01 -1.14811635e+00 -5.07535160e-01 9.90160167e-01
-1.18530715e+00 -4.71694529e-01 4.95248251e-02 6.87509120e-01
-5.05128741e-01 8.79584968e-01 -7.59840906e-01 -5.96265852e-01
-1.78400174e-01 -1.55494833e+00 7.67958462e-01 3.69457483e-01
-4.68414783e-01 -6.49652421e-01 1.70209631e-01 8.82938802e-01
9.71793473e-01 -5.70925295e-01 2.64626503e-01 1.84009880e-01
-5.31579018e-01 1.27537502e-02 -7.82514989e-01 5.43992639e-01
6.98158860e-01 -5.90496778e-01 -1.15209889e+00 -7.05681205e-01
3.41681153e-01 -1.84975877e-01 8.05092335e-01 8.64980161e-01
9.16832745e-01 -3.19292396e-01 1.56865865e-02 5.30280828e-01
1.27215219e+00 -2.21022144e-02 1.02092290e+00 4.01108474e-01
5.00344515e-01 3.94965857e-01 4.07135069e-01 5.85622191e-02
1.54358491e-01 7.27583349e-01 1.06950372e-01 -1.88204758e-02
-6.96000516e-01 9.15224031e-02 5.47587812e-01 4.54350829e-01
-2.23242402e-01 -2.96679556e-01 -5.68359315e-01 8.08231592e-01
-1.31469977e+00 -9.93976891e-01 1.33320550e-02 2.35698700e+00
1.02206981e+00 -1.96131632e-01 -1.36739582e-01 -1.85735479e-01
1.09520841e+00 4.20336664e-01 -3.82253498e-01 4.16576415e-02
-3.67972374e-01 -1.15759380e-01 6.89184129e-01 8.67814243e-01
-9.94504154e-01 7.36546278e-01 7.39864635e+00 4.56454784e-01
-1.03168118e+00 2.04138607e-01 5.93612492e-01 -3.46583240e-02
-2.89400164e-02 1.20824926e-01 1.59610361e-01 7.71721423e-01
7.50860870e-01 -1.50096819e-01 1.08391356e+00 3.61884981e-01
7.17793047e-01 -5.35323918e-01 -9.52412367e-01 1.39136791e+00
3.50237757e-01 -1.04211771e+00 -1.83920145e-01 -4.73870374e-02
9.86784399e-01 1.85052067e-01 4.84800130e-01 -4.94178146e-01
4.47873980e-01 -1.19785738e+00 8.02180171e-01 6.86725438e-01
1.09561121e+00 1.75138205e-01 6.22230649e-01 -1.05507158e-01
-2.67092139e-01 6.93155751e-02 -1.24355741e-01 -2.06586540e-01
1.41429737e-01 4.27037358e-01 -5.87201476e-01 3.57929707e-01
9.17574942e-01 5.47620773e-01 -9.07280862e-01 1.43693912e+00
-1.12271555e-01 5.92409372e-01 5.03766894e-01 4.71087933e-01
-2.59543985e-01 -1.50452018e-01 6.55353129e-01 1.19684911e+00
4.58480507e-01 1.76662520e-01 -6.26026988e-01 8.39308441e-01
1.33411512e-01 -4.81934488e-01 -3.19928974e-01 1.86879113e-01
4.83132303e-01 8.24607909e-01 -2.28419513e-01 -3.69980901e-01
-4.39691395e-01 1.65293062e+00 -2.91273564e-01 7.87307501e-01
-5.83616197e-01 -3.29478920e-01 1.00474524e+00 -1.93939712e-02
5.61223514e-02 -4.02799338e-01 -5.41110933e-01 -1.57945776e+00
-7.89134577e-02 -1.22497666e+00 4.00740094e-03 -1.77003789e+00
-1.43752575e+00 6.32227659e-01 1.14613578e-01 -1.36547554e+00
5.88258281e-02 -4.47706223e-01 -3.49641949e-01 1.38237643e+00
-1.48273098e+00 -5.15768528e-01 -7.73779452e-01 3.94551009e-01
7.67471850e-01 -6.19027242e-02 4.24312115e-01 2.20612243e-01
-2.23823115e-01 3.39193970e-01 2.28609294e-01 -1.32369474e-01
1.40133607e+00 -1.08723438e+00 5.79142392e-01 1.25961077e+00
-3.01174253e-01 1.08066761e+00 1.25212228e+00 -5.93479037e-01
-1.22988629e+00 -5.59520543e-01 4.45661664e-01 -7.41076112e-01
4.89753991e-01 2.55797923e-01 -8.87644947e-01 3.15800488e-01
7.59248316e-01 -3.87457050e-02 2.02493325e-01 -4.14605260e-01
-3.30823481e-01 1.54162362e-01 -1.28299749e+00 7.38769889e-01
9.07557249e-01 -6.30796671e-01 -8.95338416e-01 -8.38458762e-02
5.75309932e-01 -4.29359347e-01 -5.27922094e-01 2.36000657e-01
6.73902571e-01 -1.39752388e+00 1.23836136e+00 -6.36696741e-02
5.43228626e-01 -6.61725879e-01 -2.34926984e-01 -1.87884951e+00
-5.74348152e-01 -7.99888849e-01 5.36191277e-02 7.81495690e-01
5.08129857e-02 -6.25255048e-01 2.92339772e-01 6.60541773e-01
1.06369853e-02 1.10545561e-01 -4.04579133e-01 -6.29070282e-01
-3.02054465e-01 -7.26657212e-02 3.14354956e-01 1.07741272e+00
-6.30174652e-02 3.21860909e-02 -7.70184994e-01 2.93433845e-01
9.35291290e-01 1.07964173e-01 5.12684286e-01 -5.88943779e-01
-3.03178012e-01 -4.32242244e-01 -1.84048876e-01 -1.34454226e+00
-2.84772038e-01 -2.83832401e-01 2.18591794e-01 -1.52662373e+00
4.74261761e-01 -5.47130313e-03 -1.27868220e-01 -7.57189691e-02
-6.48998201e-01 4.14617479e-01 -1.94803029e-02 7.63553560e-01
-3.51829678e-01 1.67964578e-01 1.33892083e+00 -1.70105278e-01
-9.53138173e-02 -1.28475323e-01 -1.06334591e+00 3.84098649e-01
6.39381289e-01 8.40505287e-02 -4.79206562e-01 -1.00089097e+00
-1.81116387e-01 1.41287446e-01 8.42959702e-01 -8.62669706e-01
4.15121377e-01 -2.94435978e-01 4.04942185e-01 3.09724193e-02
2.19220728e-01 -5.64365625e-01 2.11800769e-01 2.77420282e-01
-4.31697994e-01 7.77028129e-02 5.46471551e-02 5.58917999e-01
-1.50858045e-01 -1.86587930e-01 9.97198761e-01 -2.52896190e-01
-8.73867154e-01 -3.46280098e-01 -2.60170966e-01 9.17359442e-02
4.90527064e-01 -3.95774126e-01 -7.92303503e-01 -8.19421470e-01
-3.91972572e-01 -2.79659599e-01 1.12017441e+00 5.80670595e-01
7.13186264e-01 -8.79985154e-01 -8.91783178e-01 7.04424083e-02
3.34281512e-02 -6.78070068e-01 4.49518591e-01 8.22223961e-01
-6.97954535e-01 3.42079580e-01 -4.09429222e-01 -2.31683254e-01
-1.24451399e+00 7.74651945e-01 4.41629738e-01 5.17074883e-01
-1.52827442e-01 6.22120917e-01 2.54404455e-01 3.27334464e-01
-1.50203360e-02 -2.62642145e-01 1.01291150e-01 -3.42997074e-01
6.58929527e-01 6.30358040e-01 6.97404444e-02 -7.22108185e-01
-3.02642465e-01 4.63752180e-01 -1.92868039e-02 -5.81061006e-01
9.01508152e-01 -7.67509282e-01 -3.37334007e-01 -2.74716411e-02
8.79705250e-01 4.37559456e-01 -1.53392589e+00 -8.34906027e-02
-2.02526197e-01 -1.23771572e+00 2.47093409e-01 -1.23690701e+00
-8.73010635e-01 4.25230622e-01 9.98652935e-01 1.97539523e-01
1.48871398e+00 5.47042228e-02 4.84817684e-01 -2.00826809e-01
2.83811897e-01 -5.16407311e-01 8.83340165e-02 1.09933568e-02
1.09157968e+00 -1.42511523e+00 1.39021903e-01 -1.62983477e-01
-7.36048937e-01 6.39340341e-01 3.11253428e-01 9.76618081e-02
2.38183513e-01 -6.53328523e-02 4.29776609e-01 -4.03706506e-02
-3.44083279e-01 -1.16531588e-01 4.29868817e-01 7.11959302e-01
4.27675277e-01 1.18351933e-02 -3.21327150e-01 2.59067088e-01
-3.02867711e-01 4.27064806e-01 9.13324475e-01 5.73433220e-01
-4.47469801e-01 -6.56157970e-01 -9.74944115e-01 1.96148306e-01
-4.88361329e-01 -3.95492136e-01 -3.35093796e-01 1.24818251e-01
-1.06483638e-01 1.56258690e+00 -2.62160182e-01 -3.25012505e-01
1.68906152e-01 -3.90434176e-01 6.62827015e-01 -2.42294237e-01
-1.43760443e-01 -2.18080759e-01 -6.49404749e-02 -1.39127806e-01
-6.01310909e-01 -6.50042415e-01 -1.04801498e-01 -3.26790482e-01
-2.63496548e-01 1.05779648e-01 5.56043923e-01 5.64859033e-01
4.08882588e-01 1.27889648e-01 3.82400185e-01 -1.02262425e+00
-6.44638658e-01 -1.25677252e+00 -4.05328631e-01 6.44809544e-01
9.46642399e-01 -3.75938207e-01 -7.15222478e-01 5.57130337e-01]
|
[11.663654327392578, -2.7905895709991455]
|
3c9ec195-05b5-44e5-9fec-bd5b6691b5ac
|
dual-mode-adaptive-svd-ghost-imaging
|
2302.07269
| null |
https://arxiv.org/abs/2302.07269v1
|
https://arxiv.org/pdf/2302.07269v1.pdf
|
Dual-mode adaptive-SVD ghost imaging
|
In this paper, we present a dual-mode adaptive singular value decomposition ghost imaging (A-SVD GI), which can be easily switched between the modes of imaging and edge detection. It can adaptively localize the foreground pixels via a threshold selection method. Then only the foreground region is illuminated by the singular value decomposition (SVD) - based patterns, consequently retrieving high-quality images with fewer sampling ratios. By changing the selecting range of foreground pixels, the A-SVD GI can be switched to the mode of edge detection to directly reveal the edge of objects, without needing the original image. We investigate the performance of these two modes through both numerical simulations and experiments. We also develop a single-round scheme to halve measurement numbers in experiments, instead of separately illuminating positive and negative patterns in traditional methods. The binarized SVD patterns, generated by the spatial dithering method, are modulated by a digital micromirror device (DMD) to speed up the data acquisition. This dual-mode A-SVD GI can be applied in various applications, such as remote sensing or target recognition, and could be further extended for multi-modality functional imaging/detection.
|
['Fan Wang', 'Xuchen Shan', 'Yao Wang', 'Jiaqi Song', 'Baolei Liu', 'Dajing Wang']
|
2023-02-14
| null | null | null | null |
['edge-detection']
|
['computer-vision']
|
[ 7.40524888e-01 -3.92816603e-01 1.91308156e-01 1.85783252e-01
-2.41955116e-01 -5.22064209e-01 2.20202103e-01 -5.93840063e-01
-5.14667749e-01 5.43878973e-01 -1.87752575e-01 -2.59018183e-01
-5.38291931e-02 -7.30218828e-01 -2.25390106e-01 -1.47238660e+00
2.12551832e-01 -1.48269400e-01 7.12850511e-01 6.53387159e-02
3.87150615e-01 5.23965180e-01 -1.41558373e+00 3.34270865e-01
6.12598300e-01 1.00522888e+00 5.38411200e-01 4.01927114e-01
5.44330925e-02 3.13504189e-01 -4.34152991e-01 4.23371345e-01
2.75220871e-01 -7.42308557e-01 -6.78833574e-03 2.65733182e-01
-1.46317214e-01 -4.96333897e-01 -3.49636316e-01 1.21652544e+00
5.93947828e-01 1.94088314e-02 5.95066428e-01 -7.33024478e-01
-7.73619831e-01 -2.36530490e-02 -1.10613155e+00 4.58658814e-01
3.53634655e-01 2.71292925e-01 1.99947760e-01 -1.03572690e+00
6.05542779e-01 1.02034783e+00 2.14001268e-01 5.37467539e-01
-1.28325891e+00 -6.94652915e-01 -5.24409637e-02 4.09456283e-01
-1.37302470e+00 -5.08322656e-01 1.11243963e+00 -5.05911231e-01
2.86529988e-01 5.45581758e-01 5.94194412e-01 6.08106256e-01
3.07553947e-01 2.66361535e-01 1.55672908e+00 -5.48172534e-01
1.22199252e-01 8.24429616e-02 7.89410397e-02 6.62937403e-01
3.47256780e-01 2.49898538e-01 -2.97954410e-01 -1.90493867e-01
1.22953057e+00 2.55301923e-01 -1.03904223e+00 -2.31478944e-01
-1.76928401e+00 2.60193795e-01 1.52943641e-01 3.99512172e-01
-4.69847083e-01 -3.34634244e-01 -2.07374827e-03 2.44190395e-01
2.18452513e-01 3.18528622e-01 1.19276926e-01 3.91080618e-01
-7.63288081e-01 -3.03459644e-01 4.48435932e-01 4.08007622e-01
4.80327427e-01 4.93587404e-02 -5.19072115e-01 4.74060714e-01
4.92071718e-01 7.31539011e-01 4.49227154e-01 -9.70362246e-01
2.86568440e-02 4.72817838e-01 5.26807725e-01 -8.45665514e-01
-1.43328920e-01 -7.38291889e-02 -8.54902267e-01 4.75959927e-01
3.30690175e-01 -1.44868508e-01 -8.53969812e-01 1.24059248e+00
7.41295338e-01 2.30736092e-01 1.04396395e-01 1.48031187e+00
8.27799499e-01 6.91402197e-01 -6.31747127e-01 -8.68258595e-01
1.60261881e+00 -5.01746595e-01 -9.12651420e-01 1.74668819e-01
-1.70278475e-02 -6.76202416e-01 1.12456989e+00 6.88555777e-01
-9.08803463e-01 -2.10244298e-01 -1.23012173e+00 7.91490078e-02
1.38624832e-01 3.84155512e-01 1.44853890e-01 4.10424411e-01
-6.10777318e-01 3.38074744e-01 -8.65931451e-01 9.93870124e-02
-1.12887574e-02 2.07492650e-01 -1.86956778e-01 -8.20058584e-02
-9.01213288e-01 3.81208062e-01 -3.41147184e-02 3.69250596e-01
-6.02063298e-01 -3.23035449e-01 -4.94509488e-01 -1.52453467e-01
2.71867096e-01 -3.81685406e-01 4.90286559e-01 -6.12576962e-01
-1.76054740e+00 9.37819779e-01 -4.36343670e-01 2.69786865e-01
3.61691296e-01 2.07417697e-01 -5.01303136e-01 5.21764636e-01
1.17797576e-01 7.38572478e-02 1.15671444e+00 -1.23262441e+00
-9.22061503e-02 -5.73893547e-01 -4.21284974e-01 2.87828147e-01
-2.64349520e-01 1.49614632e-01 -3.28561127e-01 -3.97278130e-01
7.79268742e-01 -7.41169095e-01 -1.49024636e-01 3.68301719e-01
-3.67730260e-01 2.95583904e-01 1.24946916e+00 -4.70338225e-01
1.13169456e+00 -2.58570075e+00 1.64149612e-01 1.57908693e-01
3.39065820e-01 1.70311362e-01 2.35148996e-01 3.12021106e-01
1.67839721e-01 -3.16194147e-01 -4.19733316e-01 1.76397026e-01
-7.50860989e-01 -5.49759790e-02 -2.71681219e-01 8.81857455e-01
-1.53330281e-01 4.92357075e-01 -7.83208311e-01 -4.83482003e-01
2.29285151e-01 4.32869047e-01 -1.99850008e-01 1.37272909e-01
1.05290629e-01 9.95690286e-01 -5.60221195e-01 6.86032534e-01
9.77868021e-01 -2.58677870e-01 1.81429535e-01 -6.83923483e-01
-7.99205959e-01 -2.49838516e-01 -1.39437902e+00 9.33665812e-01
1.71279591e-02 4.23307985e-01 7.02711880e-01 -8.65341306e-01
8.75164866e-01 1.97062194e-01 4.63640779e-01 -7.11300194e-01
1.02451682e-01 1.80545196e-01 -1.43482233e-03 -8.26140404e-01
1.76443994e-01 -3.60518545e-01 3.12951714e-01 3.38611573e-01
-6.68553889e-01 3.67263794e-01 -9.89125520e-02 -2.06671923e-01
7.25579083e-01 -2.17769057e-01 1.62606478e-01 -5.42180657e-01
7.24409163e-01 3.91495153e-02 6.39429331e-01 5.35521924e-01
-2.21375227e-01 6.20590746e-01 2.28482395e-01 -2.22052380e-01
-7.00788677e-01 -9.38124299e-01 -4.38738704e-01 6.75286889e-01
8.50372493e-01 1.94252610e-01 -5.30701935e-01 -2.31484347e-03
-5.72696850e-02 2.22708166e-01 -3.00631732e-01 -9.92068872e-02
-5.19092619e-01 -1.03010106e+00 1.42370999e-01 8.29701945e-02
8.67107689e-01 -7.80138254e-01 -8.28903437e-01 9.78772491e-02
-1.68128818e-01 -8.58095109e-01 -4.75818187e-01 -1.13150395e-01
-8.20030272e-01 -1.06403351e+00 -7.07402825e-01 -7.39426613e-01
9.06015754e-01 8.43225718e-01 2.91761726e-01 5.04123345e-02
-2.65867203e-01 2.96536922e-01 -2.68907338e-01 2.84577105e-02
-1.01286739e-01 -7.20036149e-01 2.61039883e-01 5.71669817e-01
1.26225024e-01 -5.74472010e-01 -1.17938566e+00 7.55785286e-01
-9.70507026e-01 2.28919461e-01 6.55456007e-01 8.74850333e-01
8.74022722e-01 2.99361367e-02 5.38737699e-02 -5.66198111e-01
3.70037705e-01 -8.37294459e-02 -7.39890516e-01 1.13242917e-01
-4.46317554e-01 -1.44478045e-02 4.15871263e-01 -8.16550672e-01
-1.12313378e+00 -2.02267319e-01 1.05471253e-01 -6.06750071e-01
7.60329217e-02 3.71899545e-01 -3.25907379e-01 -5.69624484e-01
5.98957360e-01 5.68621457e-01 3.76994133e-01 -5.21310389e-01
-1.88820008e-02 9.89932060e-01 6.55300319e-01 -2.02343524e-01
5.36631465e-01 9.95114803e-01 2.38340482e-01 -1.29918337e+00
-2.82478571e-01 -4.46225226e-01 -2.61492580e-01 -4.78711069e-01
9.95637417e-01 -7.49110043e-01 -8.50891650e-01 7.91576624e-01
-8.95058572e-01 -1.09804764e-01 1.31924525e-01 6.92940831e-01
2.05511168e-01 6.72826827e-01 -7.79020607e-01 -7.59240091e-01
-2.38987491e-01 -1.17683101e+00 1.02393937e+00 3.31099540e-01
4.37991142e-01 -5.64581752e-01 -2.44480014e-01 2.54639894e-01
3.19938093e-01 3.44567090e-01 8.55095029e-01 9.27123651e-02
-7.57290959e-01 6.39334694e-02 -2.32756555e-01 9.96377692e-02
3.40482891e-01 -1.54856190e-01 -8.86054397e-01 -4.69551384e-01
7.40983725e-01 2.02422485e-01 7.95283139e-01 4.68693703e-01
1.12329483e+00 -2.21225455e-01 -7.30718911e-01 8.04428935e-01
1.44070423e+00 4.80282336e-01 8.11124086e-01 3.45037341e-01
7.95174301e-01 2.74308264e-01 7.00948894e-01 4.05106306e-01
-7.20962584e-02 7.76028872e-01 2.50083059e-01 -2.23633692e-01
9.00750048e-03 1.40676245e-01 4.61907119e-01 6.65608644e-01
-2.64277339e-01 -1.45076796e-01 -4.78303522e-01 1.70150965e-01
-1.41161144e+00 -8.85677934e-01 -6.24758124e-01 2.51365328e+00
6.91132963e-01 -1.79316700e-01 -1.39572740e-01 6.63022175e-02
1.19049656e+00 1.47975057e-01 -8.50810707e-01 2.83444226e-01
-6.18685931e-02 -2.15675473e-01 5.56738913e-01 5.01421571e-01
-8.04408729e-01 3.26459885e-01 6.40732670e+00 5.90832889e-01
-1.60544586e+00 1.59804031e-01 3.21206599e-01 -1.41981557e-01
-4.63547260e-01 2.44866237e-02 -7.93922484e-01 7.27318347e-01
1.95670143e-01 1.46942690e-01 3.78125608e-01 3.86424929e-01
5.22955537e-01 -5.07496357e-01 -6.54240191e-01 1.18056071e+00
-1.44382194e-01 -1.07494664e+00 7.54521787e-02 2.36614272e-01
3.37625027e-01 -3.55615318e-01 -5.26674744e-03 -3.84334475e-01
-4.86884236e-01 -3.45899999e-01 4.55683798e-01 5.64696550e-01
1.07475591e+00 -1.83771297e-01 2.71301538e-01 4.53863949e-01
-1.06480932e+00 -8.19601268e-02 -4.20260608e-01 -2.00250596e-01
3.32326919e-01 1.05072951e+00 -2.15880588e-01 3.34410369e-01
7.22553372e-01 3.29176307e-01 -5.77597916e-02 6.66497290e-01
9.92065817e-02 5.77554524e-01 -4.50802296e-01 4.54866067e-02
-1.82375118e-01 -9.81103063e-01 1.06896079e+00 6.09632730e-01
4.49710101e-01 6.88773692e-01 1.47087455e-01 1.05010509e+00
3.36335361e-01 -2.72815526e-01 -5.96607745e-01 4.32217717e-02
5.53160846e-01 1.29474759e+00 -9.76830125e-01 -2.23092973e-01
-5.17973781e-01 1.22096074e+00 -2.73135573e-01 5.85200429e-01
-7.03244328e-01 -6.67429328e-01 4.17019486e-01 5.60378611e-01
3.69271994e-01 -4.30168658e-01 -1.58263221e-01 -1.46164620e+00
2.04706043e-01 -5.46750903e-01 1.92655027e-01 -9.08383012e-01
-8.81850183e-01 1.95113152e-01 8.65371898e-02 -1.42531848e+00
4.40736979e-01 -7.54207790e-01 -7.56736696e-01 7.94477761e-01
-1.41468871e+00 -4.87782776e-01 -4.80987817e-01 7.00419068e-01
-8.60541910e-02 3.50713909e-01 3.26533705e-01 3.12388331e-01
-6.73052073e-01 1.18364826e-01 5.67314386e-01 3.22537534e-02
6.70472741e-01 -7.57045984e-01 -2.44909406e-01 1.01492786e+00
-3.90954286e-01 4.97118622e-01 6.74113929e-01 -6.53147995e-01
-1.74698198e+00 -6.68084204e-01 1.04866549e-01 1.19956210e-01
3.57602000e-01 -3.14964950e-01 -1.09007823e+00 2.32698545e-01
8.65553692e-02 8.60404447e-02 5.40593147e-01 -7.87033141e-01
1.50102571e-01 -3.08977276e-01 -1.50930202e+00 7.02486336e-01
1.07455909e+00 -3.13113898e-01 -3.01495850e-01 2.52281547e-01
6.65774584e-01 -5.77450693e-01 -8.26596916e-01 6.18162215e-01
5.46320200e-01 -1.02137733e+00 9.00724351e-01 8.81678313e-02
8.59090015e-02 -8.73935401e-01 -4.44180612e-03 -8.60009134e-01
-6.48696959e-01 -6.73189461e-01 -3.95769961e-02 1.07548761e+00
-5.83712868e-02 -1.09053946e+00 4.69740838e-01 3.01850587e-01
-1.86703682e-01 -5.91439843e-01 -1.06219661e+00 -7.08581567e-01
-5.57707906e-01 3.32158417e-01 2.34260589e-01 8.14034343e-01
-4.00876626e-02 2.01563537e-01 -2.56882638e-01 6.38512731e-01
9.08985555e-01 5.29390335e-01 2.52043545e-01 -8.94422054e-01
-5.07359266e-01 -1.23091407e-01 -2.49837562e-01 -1.33051300e+00
-5.61170101e-01 -5.25034070e-01 7.43105188e-02 -1.19908059e+00
1.24865500e-02 -1.77767366e-01 -2.26880863e-01 1.58894017e-01
-1.70482606e-01 4.33907688e-01 -2.16234282e-01 8.26885998e-01
3.43917473e-03 4.94875640e-01 1.64016759e+00 1.74978212e-01
-4.43626434e-01 -2.37741619e-01 -3.87211591e-01 3.24388802e-01
4.16912973e-01 -2.73674816e-01 -2.65711337e-01 -2.28628352e-01
1.20929413e-01 2.67800927e-01 6.18832588e-01 -9.58395600e-01
2.05547288e-01 -3.14235181e-01 5.10033906e-01 -3.50926399e-01
8.79165009e-02 -6.43622398e-01 3.57637525e-01 5.28283179e-01
2.32829690e-01 -1.73160210e-01 2.28732694e-02 6.65343285e-01
-4.09886204e-02 -1.99199393e-01 1.04560685e+00 -3.06646883e-01
-7.17970550e-01 3.77431838e-03 -5.63141286e-01 -3.14893872e-01
1.20239055e+00 -6.17538929e-01 -4.29286033e-01 5.75210191e-02
-6.45710111e-01 7.40305404e-04 7.66515911e-01 -2.96709031e-01
8.13414693e-01 -1.37466252e+00 -1.68031111e-01 8.17811608e-01
-1.14026904e-01 -2.83349976e-02 4.29710478e-01 1.40171885e+00
-6.40619218e-01 -7.96635151e-02 -9.77960601e-02 -9.57727134e-01
-1.21271193e+00 5.96175015e-01 4.37515110e-01 1.94746688e-01
-1.02996016e+00 4.20605361e-01 2.97230184e-01 1.57901227e-01
-2.51475811e-01 -9.68483761e-02 -3.76692146e-01 -5.92052713e-02
8.97666514e-01 4.82110798e-01 -1.37289360e-01 -3.42741460e-01
-3.43139201e-01 8.10808420e-01 1.40760764e-01 -3.41302633e-01
1.06710553e+00 -3.73288542e-01 -4.86086428e-01 4.29198533e-01
9.11875129e-01 2.13801190e-01 -1.39615333e+00 -7.74165243e-02
-5.58041215e-01 -6.68163896e-01 4.48634207e-01 -2.28474826e-01
-1.14578021e+00 6.88428521e-01 8.37619066e-01 3.35006982e-01
1.37648761e+00 1.21433614e-02 6.41508877e-01 1.67080268e-01
4.15589184e-01 -7.62510240e-01 -2.46606041e-02 -6.67005926e-02
6.08756602e-01 -8.64835024e-01 5.13762236e-02 -5.73711872e-01
-5.72148383e-01 1.06730270e+00 4.73958731e-01 -1.21851236e-01
5.92015326e-01 4.53844488e-01 4.73665670e-02 -4.89059001e-01
-3.96732032e-01 -5.03346026e-02 3.82925235e-02 5.62338650e-01
1.28383487e-01 -6.40927851e-02 -5.16341627e-01 3.31551671e-01
6.72201574e-01 1.16165884e-01 6.23098016e-01 9.19895947e-01
-6.69889212e-01 -7.34044492e-01 -8.85891974e-01 6.06891751e-01
-2.96807736e-01 -6.25508605e-03 -1.64189190e-01 3.65367949e-01
-1.44108161e-01 7.46778190e-01 1.31245270e-01 -3.59757572e-01
3.53920341e-01 -5.12857810e-02 5.64667940e-01 -3.83268297e-01
6.29277229e-02 3.86226892e-01 -2.73127705e-01 -4.64872450e-01
-5.15311658e-01 -7.89824307e-01 -1.21600962e+00 -5.73831610e-02
-5.58710158e-01 3.30469459e-02 2.41393909e-01 7.98935235e-01
5.67979634e-01 1.99637502e-01 7.49293447e-01 -1.07642078e+00
-4.12271708e-01 -7.14628577e-01 -8.66611183e-01 1.34949803e-01
5.15344739e-01 -8.42757225e-01 -8.27453077e-01 -1.01390548e-01]
|
[11.361671447753906, -2.648261308670044]
|
ea0f7b90-e1ff-4624-a040-78cff8b4ff92
|
farmer-s-assistant-a-machine-learning-based
|
2204.11340
| null |
https://arxiv.org/abs/2204.11340v1
|
https://arxiv.org/pdf/2204.11340v1.pdf
|
Farmer's Assistant: A Machine Learning Based Application for Agricultural Solutions
|
Farmers face several challenges when growing crops like uncertain irrigation, poor soil quality, etc. Especially in India, a major fraction of farmers do not have the knowledge to select appropriate crops and fertilizers. Moreover, crop failure due to disease causes a significant loss to the farmers, as well as the consumers. While there have been recent developments in the automated detection of these diseases using Machine Learning techniques, the utilization of Deep Learning has not been fully explored. Additionally, such models are not easy to use because of the high-quality data used in their training, lack of computational power, and poor generalizability of the models. To this end, we create an open-source easy-to-use web application to address some of these issues which may help improve crop production. In particular, we support crop recommendation, fertilizer recommendation, plant disease prediction, and an interactive news-feed. In addition, we also use interpretability techniques in an attempt to explain the prediction made by our disease detection model.
|
['Aparna Bhonde', 'Nishit Jain', 'Akshay Chopade', 'Shloka Gupta']
|
2022-04-24
| null | null | null | null |
['disease-prediction']
|
['medical']
|
[ 3.30545083e-02 3.88524169e-03 -3.59227598e-01 -1.25488192e-01
1.21239282e-01 -7.64938414e-01 -1.77164003e-01 7.35976577e-01
3.34696263e-01 6.15075469e-01 4.99899406e-03 -9.73237514e-01
-1.94320008e-01 -1.28341508e+00 -6.11608565e-01 -4.74595428e-01
1.43703207e-01 2.22682565e-01 -5.52839898e-02 -5.82424700e-01
1.79584593e-01 3.84229362e-01 -1.43233442e+00 2.78482229e-01
1.10540628e+00 4.98963773e-01 8.51089001e-01 4.61923569e-01
-2.79018819e-01 4.80288237e-01 -2.72011489e-01 3.70886810e-02
-4.28714417e-02 -2.11611331e-01 -5.03929734e-01 8.50224793e-02
-4.16708529e-01 -8.09174657e-01 3.13648999e-01 9.49570596e-01
2.82601595e-01 -3.83999854e-01 6.91386044e-01 -9.47129965e-01
-9.51063931e-01 5.45092940e-01 -8.31918955e-01 -1.68467328e-01
3.33299726e-01 4.99378592e-02 6.78663075e-01 -4.51768577e-01
3.21783245e-01 1.08232236e+00 5.12641728e-01 1.29914522e-01
-9.73713100e-01 -7.61425495e-01 1.38906717e-01 -2.75085121e-02
-7.36387789e-01 -5.04559800e-02 2.52176195e-01 -6.00078523e-01
8.54635477e-01 4.69670966e-02 7.30298758e-01 5.44752836e-01
5.21209657e-01 5.08440733e-01 8.81845415e-01 -4.76138681e-01
1.51124611e-01 3.68849576e-01 -3.59183438e-02 6.06020451e-01
6.03782654e-01 2.92740129e-02 -4.63166274e-02 5.45843132e-02
7.67387986e-01 2.17596248e-01 -2.46842559e-02 -2.50870228e-01
-9.78411794e-01 9.81258690e-01 7.09221959e-01 2.31504172e-01
-7.38652349e-01 -1.79532990e-01 2.18222544e-01 2.78651834e-01
6.76258445e-01 5.07596970e-01 -1.09329367e+00 3.37925494e-01
-6.11098289e-01 2.62591928e-01 1.04149628e+00 6.92066789e-01
6.18297219e-01 9.23197195e-02 3.67222548e-01 4.42181230e-01
5.94104409e-01 6.75216675e-01 3.36219836e-03 -5.59292912e-01
8.90527368e-02 7.37472832e-01 2.58094698e-01 -1.46864927e+00
-4.75238413e-01 -2.50553578e-01 -1.06081843e+00 2.83232331e-01
3.75141829e-01 -4.27721858e-01 -1.02981579e+00 1.40262115e+00
1.22600958e-01 -1.72133490e-01 1.45249099e-01 6.95358336e-01
6.90889299e-01 7.61101246e-01 6.03698909e-01 -3.20510119e-02
1.45473146e+00 -4.25542802e-01 -9.62292016e-01 -2.71898597e-01
9.36545491e-01 -9.35606837e-01 6.80160224e-01 3.25471044e-01
-6.15601480e-01 -3.45589131e-01 -1.03230011e+00 2.59340435e-01
-9.81258392e-01 4.22693640e-01 1.13256335e+00 5.61834931e-01
-7.40602851e-01 5.50904512e-01 -1.02794659e+00 -9.84067976e-01
3.95802140e-01 3.39436173e-01 -3.49907309e-01 -2.34283254e-01
-1.08381212e+00 1.42786837e+00 5.84323943e-01 4.25597310e-01
-4.24569845e-01 -6.57963395e-01 -6.37744308e-01 8.82633030e-02
2.12748826e-01 -4.66526121e-01 9.78229523e-01 -8.74848008e-01
-1.18926144e+00 6.72098398e-01 -2.05820464e-02 -1.92702726e-01
1.75904613e-02 -4.75253552e-01 -3.49331975e-01 -4.70479190e-01
2.10224271e-01 6.82852864e-01 2.47770652e-01 -1.09895253e+00
-9.22315776e-01 -5.41031063e-01 1.19354442e-01 1.65080011e-01
-1.82992399e-01 1.10028341e-01 3.07547420e-01 -4.21458960e-01
3.51348042e-01 -9.14803028e-01 -4.16539341e-01 2.69278049e-01
-2.44490966e-01 1.73566774e-01 9.37031686e-01 -1.05465233e+00
1.07676220e+00 -2.10190487e+00 -1.22587875e-01 -6.49180822e-03
-1.09842278e-01 5.25180697e-01 -9.45681781e-02 5.22376597e-01
1.38774186e-01 5.47492921e-01 2.65210145e-03 7.84499228e-01
-3.28264505e-01 5.08743346e-01 -1.53315932e-01 1.01424329e-01
7.94532299e-01 4.41211611e-01 -1.05301690e+00 -2.54337266e-02
4.86740917e-01 4.58980858e-01 -2.69709855e-01 2.16529772e-01
-4.67151225e-01 2.55019814e-01 -8.73139322e-01 1.05141282e+00
1.00180185e+00 -2.57136166e-01 3.59707534e-01 9.78004886e-04
-4.11633611e-01 1.68870702e-01 -9.44246292e-01 1.08661091e+00
-2.97639161e-01 3.77224267e-01 1.43891215e-01 -1.33426189e+00
9.50808585e-01 5.07169902e-01 1.92848265e-01 -1.69709116e-01
-2.24117264e-01 4.89419788e-01 9.34466347e-02 -8.85796607e-01
4.20798659e-01 2.46304467e-01 4.96173859e-01 2.98753440e-01
-3.65024805e-01 -1.01861499e-01 2.65532136e-01 -1.81562856e-01
7.67817259e-01 6.03319407e-01 6.17960691e-01 -3.48109603e-01
3.23676295e-03 4.15170163e-01 5.48007488e-01 3.60088795e-01
-2.02801600e-02 2.16257647e-01 5.00155568e-01 -4.64503020e-01
-8.74989867e-01 -4.09215480e-01 -1.32236019e-01 1.35460401e+00
4.87503270e-03 1.30572483e-01 -3.90758425e-01 -3.00388157e-01
3.66569817e-01 7.20689416e-01 -4.77463484e-01 6.93781599e-02
-1.09201267e-01 -1.18038845e+00 1.43053472e-01 6.06160343e-01
4.68664616e-01 -1.28019333e+00 -7.36927330e-01 6.45790279e-01
-1.21896498e-01 -7.31179118e-01 2.41704389e-01 7.80855119e-01
-1.05682409e+00 -9.17324245e-01 -6.54221356e-01 -9.17206287e-01
6.66187465e-01 7.41386235e-01 1.01342678e+00 2.60732412e-01
-2.90076196e-01 -4.40315843e-01 -6.06305420e-01 -1.08174515e+00
-6.09171748e-01 4.60874647e-01 -3.13107401e-01 -9.09871042e-01
8.49056005e-01 -2.04955667e-01 -4.68450159e-01 1.71779215e-01
-8.72705698e-01 2.98898876e-01 8.75666738e-01 6.42465651e-01
1.91946909e-01 3.63739550e-01 9.68344092e-01 -8.98130953e-01
3.79170954e-01 -1.06567705e+00 -5.61714232e-01 6.09267175e-01
-6.38999224e-01 -1.12821169e-01 9.11454931e-02 -2.64594734e-01
-8.86824131e-01 3.52200478e-01 -5.66258058e-02 7.94319987e-01
-5.40472209e-01 1.34434175e+00 -4.13980857e-02 1.11041263e-01
4.51424062e-01 -3.27239841e-01 1.41049802e-01 -4.32606012e-01
7.56102875e-02 7.12408423e-01 7.72869885e-02 7.73974648e-03
6.26809299e-01 -2.65613258e-01 -9.59340259e-02 -8.79826188e-01
-9.27346170e-01 -2.41277948e-01 -6.71853483e-01 1.35371536e-01
7.31097698e-01 -9.91556406e-01 -3.80241305e-01 5.51429093e-01
-1.13877153e+00 -1.31848007e-01 4.21463877e-01 6.29272938e-01
1.25425933e-02 6.27373010e-02 -4.29108799e-01 -8.80385339e-01
-4.74805027e-01 -1.01739323e+00 6.09793901e-01 6.07398629e-01
-3.99413466e-01 -8.52138460e-01 -4.75275189e-01 3.47206503e-01
6.83377087e-01 5.49388587e-01 1.36583340e+00 -3.78850847e-01
-4.50741261e-01 -1.76930711e-01 -6.09061360e-01 1.26815051e-01
7.61643052e-01 4.72108841e-01 -8.43855977e-01 -3.09470408e-02
-4.93339270e-01 -4.44918990e-01 5.36126077e-01 9.18197453e-01
7.61540532e-01 1.45058081e-01 -4.95167136e-01 3.10390651e-01
1.32722104e+00 4.50904399e-01 2.69137293e-01 4.40805554e-01
3.17319602e-01 1.00524640e+00 1.04400587e+00 4.31656271e-01
4.20068741e-01 1.49209291e-01 6.97912455e-01 -4.96406287e-01
3.52525711e-01 -4.86671180e-02 1.45409152e-01 4.21714395e-01
-1.49353653e-01 -5.09634674e-01 -1.02352560e+00 6.75161898e-01
-1.99521708e+00 -7.51901031e-01 -3.57551306e-01 1.88256192e+00
7.28958905e-01 -2.27229372e-02 -2.30944693e-01 2.68057048e-01
4.43793297e-01 -3.07055593e-01 -5.37422776e-01 -4.97687697e-01
-2.25397587e-01 -5.19838631e-02 8.00071478e-01 1.25783741e-01
-1.23494983e+00 1.17380643e+00 6.25838947e+00 1.12919845e-01
-1.24549878e+00 -4.46161360e-01 8.38175416e-01 6.69160008e-01
-1.98636204e-01 1.28480002e-01 -6.66088402e-01 6.51108176e-02
5.99975348e-01 2.06147805e-01 2.89728791e-01 8.42960179e-01
6.01644754e-01 -5.61357856e-01 -7.60816813e-01 1.00679949e-01
-4.84275401e-01 -1.02238572e+00 -3.10102284e-01 1.96842737e-02
5.72003126e-01 -1.41491011e-01 -9.15562212e-02 1.12773113e-01
5.37672102e-01 -1.15068161e+00 2.06479847e-01 9.24525037e-02
4.03436542e-01 -6.19577646e-01 1.05737424e+00 3.99261475e-01
-1.07562792e+00 -6.69182464e-02 -7.56392300e-01 -5.16214907e-01
-2.01858997e-01 6.24225378e-01 -8.66613984e-01 5.39909840e-01
8.22454512e-01 6.08553708e-01 -3.89284492e-01 8.91494811e-01
-1.42887548e-01 7.23858356e-01 -4.46269572e-01 -1.39756948e-01
2.67337084e-01 -1.64082348e-01 -2.51206100e-01 9.57301199e-01
7.83326864e-01 3.14439863e-01 4.18215513e-01 6.10849917e-01
4.11971807e-01 3.12399387e-01 -7.59960473e-01 -4.76319134e-01
4.55665290e-01 1.08004832e+00 -9.76820052e-01 2.65413046e-01
-5.00030339e-01 6.76270723e-01 -9.63501930e-02 2.92261899e-01
-2.73659319e-01 -2.98802197e-01 6.59253776e-01 1.18909977e-01
1.09615453e-01 -1.31646886e-01 -4.70041186e-01 -8.23481619e-01
-2.37460554e-01 -1.15598834e+00 -1.85315356e-01 -7.78008580e-01
-1.09858286e+00 -3.84069383e-02 -2.34150425e-01 -5.54755211e-01
-2.81252712e-01 -8.32198799e-01 -4.79451686e-01 1.22572374e+00
-1.73328876e+00 -1.34273839e+00 -3.05964053e-01 -1.61851615e-01
4.89279240e-01 -1.33587375e-01 1.45692027e+00 1.39776617e-01
-5.00544608e-01 -7.32606202e-02 3.15777898e-01 -2.16294855e-01
6.87175810e-01 -9.84435439e-01 3.80820155e-01 4.47829813e-01
-5.24561644e-01 2.56441146e-01 8.62318218e-01 -9.29546535e-01
-1.39722133e+00 -1.07398105e+00 1.14682841e+00 -6.21755496e-02
5.45851469e-01 9.67809111e-02 -1.10716534e+00 5.23489356e-01
7.09008053e-02 -3.52584243e-01 8.03329110e-01 4.19325650e-01
2.34317422e-01 -4.63811494e-03 -1.23210967e+00 3.69225681e-01
2.88469911e-01 -2.74902508e-02 -5.93715571e-02 4.91739810e-01
3.91173631e-01 -2.09380493e-01 -7.83033431e-01 4.38265145e-01
9.21230912e-01 -5.65327048e-01 7.83194721e-01 -5.58166265e-01
6.65655375e-01 -9.94448662e-02 -1.67307243e-01 -1.41605031e+00
-6.71006858e-01 -7.69200251e-02 2.55802929e-01 1.35861266e+00
7.31246650e-01 -3.79727781e-01 7.66162574e-01 6.50280833e-01
3.19160372e-01 -7.25724101e-01 2.27356568e-01 -1.18434794e-01
9.87678617e-02 -3.87345962e-02 8.71651173e-01 1.03977382e+00
-1.01598807e-01 -6.76452518e-02 -5.60078561e-01 5.92436969e-01
2.84935564e-01 1.02313660e-01 5.62088311e-01 -1.47400224e+00
4.49201502e-02 -1.59671739e-01 -7.75644183e-02 -5.77222109e-01
-2.27924019e-01 -3.62920165e-01 1.11699417e-01 -1.84762609e+00
1.85541227e-01 -6.95167840e-01 -2.34186128e-01 8.81873190e-01
-5.43167353e-01 -3.20852667e-01 -2.39595887e-03 -3.24703306e-02
3.60772163e-01 -1.06855869e-01 1.12958109e+00 -5.27123064e-02
-6.07935727e-01 3.80829632e-01 -1.07075131e+00 7.89478898e-01
1.30931687e+00 -5.11822641e-01 -4.29515392e-01 -9.60363925e-01
5.47262549e-01 1.91169336e-01 1.50148138e-01 -4.27761853e-01
-2.80187190e-01 -7.39200771e-01 7.24053621e-01 -8.75614226e-01
-2.25828081e-01 -9.02804852e-01 3.34594131e-01 7.91221440e-01
-2.59454191e-01 1.54555261e-01 4.35142219e-01 2.47441292e-01
4.65932786e-02 -3.64808500e-01 3.64973664e-01 -2.21872598e-01
-5.89996696e-01 -5.43141700e-02 -6.49507582e-01 -6.35792136e-01
9.72035527e-01 -6.35505468e-02 -2.07604557e-01 -3.15083385e-01
-3.78643036e-01 6.19379222e-01 3.04406852e-01 5.88195622e-01
2.38440558e-01 -8.46440554e-01 -9.41132724e-01 1.71374395e-01
9.17681158e-02 5.43992408e-02 -1.05045415e-01 4.38337326e-01
-1.23161626e+00 6.82178438e-01 -7.12641001e-01 -2.14663997e-01
-1.02417171e+00 4.31603998e-01 -2.31731191e-01 -2.96951592e-01
-2.75757492e-01 5.68660200e-01 8.21558535e-02 -4.98275310e-01
1.80994868e-01 -6.52829647e-01 -6.06956720e-01 1.64263517e-01
3.04897666e-01 1.69500604e-01 -1.04030877e-01 -1.78305343e-01
-3.06410879e-01 2.97412157e-01 -1.53011546e-01 4.67620850e-01
1.67635739e+00 -7.05866516e-02 -7.41398633e-02 4.01249170e-01
2.31907070e-01 -5.46479702e-01 -1.02839231e+00 -3.68682737e-03
3.78228486e-01 -3.14849019e-01 2.62186944e-01 -1.16259253e+00
-1.12463772e+00 8.76070797e-01 8.60559106e-01 5.67169666e-01
1.17411304e+00 -5.91521382e-01 4.69611406e-01 7.49659419e-01
7.99198076e-02 -9.13563371e-01 -5.61258256e-01 3.72230530e-01
7.13676631e-01 -1.93217289e+00 1.26498237e-01 -4.93465185e-01
-4.14957315e-01 1.14359176e+00 5.32506049e-01 2.31694773e-01
8.46941829e-01 4.47895616e-01 5.08714199e-01 9.83097106e-02
-7.30017006e-01 -1.96633086e-01 -2.27780730e-01 9.81986701e-01
1.18003964e+00 4.12525177e-01 -3.25562507e-01 2.40838319e-01
1.27581134e-01 5.56142986e-01 4.19726878e-01 1.14523578e+00
-7.72156894e-01 -1.22775233e+00 -4.16817635e-01 7.06542790e-01
-7.68909454e-01 -1.11613944e-01 -2.57290184e-01 6.22101128e-01
2.63483316e-01 1.21526730e+00 -4.19143975e-01 -7.42077604e-02
1.01266719e-01 -8.92851800e-02 1.00076705e-01 -8.14582050e-01
-5.08713841e-01 1.16737328e-01 4.11916961e-04 2.51673609e-01
-5.16352177e-01 -5.64415216e-01 -9.95429397e-01 -6.77142262e-01
-7.99969435e-01 -3.59740227e-01 1.14289331e+00 9.29366469e-01
2.67489612e-01 4.89270627e-01 4.62404400e-01 -6.75290823e-01
-5.73110223e-01 -1.08553517e+00 -7.74318874e-01 -9.62391123e-02
1.02644652e-01 -6.19622707e-01 2.47097820e-01 1.70529380e-01]
|
[9.323701858520508, -1.570221185684204]
|
0eea20be-dabe-4439-9eef-ff6c6b35dd6c
|
a-graph-multi-separator-problem-for-image
|
2307.04592
| null |
https://arxiv.org/abs/2307.04592v1
|
https://arxiv.org/pdf/2307.04592v1.pdf
|
A Graph Multi-separator Problem for Image Segmentation
|
We propose a novel abstraction of the image segmentation task in the form of a combinatorial optimization problem that we call the multi-separator problem. Feasible solutions indicate for every pixel whether it belongs to a segment or a segment separator, and indicate for pairs of pixels whether or not the pixels belong to the same segment. This is in contrast to the closely related lifted multicut problem where every pixel is associated to a segment and no pixel explicitly represents a separating structure. While the multi-separator problem is NP-hard, we identify two special cases for which it can be solved efficiently. Moreover, we define two local search algorithms for the general case and demonstrate their effectiveness in segmenting simulated volume images of foam cells and filaments.
|
['Bjoern Andres', 'Jannik Presberger', 'Shengxian Zhao', 'Jannik Irmai']
|
2023-07-10
| null | null | null | null |
['semantic-segmentation', 'combinatorial-optimization']
|
['computer-vision', 'methodology']
|
[ 8.56865048e-01 4.53476161e-01 -3.26546669e-01 -1.46108335e-02
-7.08802581e-01 -1.02654231e+00 -4.74655367e-02 4.27349597e-01
-2.40926355e-01 8.95763993e-01 -5.19028544e-01 -4.94073719e-01
-3.14511240e-01 -8.35582912e-01 -8.87798190e-01 -9.93888438e-01
-8.79593566e-02 8.32911491e-01 6.24507964e-01 2.77401328e-01
4.12488848e-01 8.07305276e-01 -1.18943620e+00 1.13254443e-01
8.69145393e-01 1.04823196e+00 2.36465141e-01 8.20090652e-01
-4.22345400e-01 -1.07604176e-01 -4.04183596e-01 6.46076798e-02
5.42215466e-01 -5.53188145e-01 -1.32183135e+00 7.09465683e-01
5.24101853e-01 1.97774470e-01 4.66151357e-01 1.07552862e+00
-7.12988805e-03 -1.60682827e-01 7.79230773e-01 -1.55761337e+00
-2.26523161e-01 3.06713611e-01 -1.02895188e+00 1.29789308e-01
3.10934395e-01 -2.16866806e-01 9.57819223e-01 -1.58190757e-01
8.01315427e-01 8.97047818e-01 6.11304104e-01 1.61648110e-01
-1.72795272e+00 2.44651467e-01 4.07596648e-01 -2.63527066e-01
-9.92032766e-01 2.16803364e-02 3.55384588e-01 -4.91434187e-01
6.30550742e-01 7.44728804e-01 9.50433075e-01 1.10933743e-01
-6.83847740e-02 6.99016690e-01 1.32963002e+00 -4.91284043e-01
4.70160037e-01 -9.30594280e-03 6.21507466e-01 8.20054650e-01
3.73090655e-01 -1.47251859e-01 1.03152901e-01 -2.41644889e-01
9.48397279e-01 -3.70835692e-01 -4.20692533e-01 -6.70596540e-01
-1.23507845e+00 7.76560426e-01 2.75685251e-01 2.68723011e-01
-4.13307454e-04 1.52146295e-01 3.28954868e-02 1.02596350e-01
7.80165941e-02 5.85907996e-01 -2.13929370e-01 3.34124714e-01
-1.02945542e+00 3.98025930e-01 1.13047528e+00 7.57015169e-01
9.46818173e-01 -4.05139178e-01 9.27551650e-03 4.14177239e-01
-8.67963657e-02 1.50603473e-01 -1.24685414e-01 -1.27434278e+00
4.15226638e-01 7.21136451e-01 4.04342592e-01 -7.22473800e-01
-7.18759716e-01 -1.80215091e-01 -4.16732043e-01 3.95686746e-01
8.34046125e-01 -8.87529701e-02 -8.32242668e-01 1.39062846e+00
5.99421501e-01 3.18204433e-01 -1.27082899e-01 9.05307889e-01
5.26213467e-01 8.04044127e-01 -2.99710184e-01 -6.75602913e-01
1.38158095e+00 -1.28520334e+00 -3.21078420e-01 -1.50131106e-01
3.34335536e-01 -6.38433397e-01 4.43933189e-01 4.75124985e-01
-1.42391884e+00 -1.43314108e-01 -1.05768073e+00 2.40670089e-02
-2.71925598e-01 -1.93020433e-01 3.62983674e-01 5.70849538e-01
-8.73581767e-01 5.88149488e-01 -6.08961344e-01 -2.12839335e-01
7.86696896e-02 5.31152546e-01 -2.30226189e-01 1.91662684e-01
-4.40995395e-01 5.36595643e-01 4.68401253e-01 1.05048157e-01
-4.18650597e-01 -3.76796365e-01 -6.87434375e-01 -1.38909917e-03
5.63779771e-01 -5.30294657e-01 9.46846187e-01 -1.16157031e+00
-9.69161570e-01 1.28563094e+00 -5.39311051e-01 -5.81703544e-01
7.18634248e-01 3.22948337e-01 8.50050710e-03 4.94892538e-01
9.38485786e-02 7.58299410e-01 5.96472740e-01 -1.63230801e+00
-7.57467866e-01 -1.44074693e-01 5.17174184e-01 1.94385052e-01
2.06435621e-01 1.66816890e-01 -6.47542655e-01 -5.25147676e-01
5.70742726e-01 -9.62652028e-01 -6.30367279e-01 2.10011438e-01
-1.00352788e+00 -7.39328861e-02 8.88088048e-01 -4.37854856e-01
1.20041358e+00 -1.86534369e+00 3.45723540e-01 5.35664856e-01
1.94937110e-01 -1.41122859e-04 -2.94265486e-02 2.57187366e-01
-4.68627810e-02 4.25622076e-01 -8.04613531e-01 -7.45765939e-02
-7.83986822e-02 5.98158479e-01 5.70984259e-02 5.67029297e-01
1.35432974e-01 6.58704877e-01 -7.46799529e-01 -7.56831110e-01
-6.43695593e-02 -1.30367830e-01 -3.23714972e-01 -2.00985283e-01
-4.10762459e-01 3.50726813e-01 -4.50413883e-01 3.24646264e-01
1.06939054e+00 -4.18229222e-01 3.43863308e-01 -5.13617285e-02
-4.74260330e-01 -1.77144721e-01 -1.75036705e+00 9.31868076e-01
2.53858380e-02 3.23285967e-01 6.09494984e-01 -1.32591045e+00
6.96422815e-01 1.18616544e-01 7.01451302e-01 -3.97736728e-01
-1.01878224e-02 2.96382219e-01 -3.61559570e-01 -4.84419346e-01
5.57877600e-01 -2.05112264e-01 -1.05594613e-01 6.29444778e-01
-5.30988812e-01 -2.16674343e-01 5.62750041e-01 -4.04981934e-02
8.79845917e-01 -6.66081682e-02 1.27976269e-01 -6.94199860e-01
5.04249036e-01 2.34086990e-01 8.56562614e-01 8.16383660e-01
-1.84486717e-01 9.04072344e-01 1.13897741e+00 -1.04751177e-01
-7.26822257e-01 -1.23278630e+00 -5.44227183e-01 4.85547364e-01
8.73499572e-01 -1.58810973e-01 -1.05181742e+00 -5.89401782e-01
4.30143587e-02 4.21988545e-03 -5.89222670e-01 5.84696651e-01
-7.19228506e-01 -6.37991130e-01 -8.30760822e-02 3.83850336e-01
4.36770260e-01 -6.04696751e-01 -9.73062694e-01 1.83122978e-01
-2.55025804e-01 -9.76307750e-01 -5.69241285e-01 5.32671452e-01
-9.80031192e-01 -1.45577848e+00 -8.71022999e-01 -1.26064479e+00
1.07002103e+00 3.66725177e-01 1.11564183e+00 2.38709167e-01
-3.73989254e-01 1.03046075e-01 -1.64000332e-01 1.63204119e-01
-5.30785993e-02 1.90485164e-03 -6.29111230e-01 -1.25936851e-01
-3.37737679e-01 -8.84212647e-03 -5.54799080e-01 7.07263529e-01
-8.57534230e-01 8.81146267e-02 -1.53898209e-01 5.55794060e-01
1.45490432e+00 3.03968787e-01 1.84918121e-01 -1.00260687e+00
2.88498014e-01 -4.72985864e-01 -9.67012346e-01 5.95052540e-01
-9.13089700e-03 -1.43131129e-02 4.28815067e-01 -2.01026157e-01
-5.95320702e-01 6.21696889e-01 2.66322732e-01 1.23363147e-02
-1.73706770e-01 4.44199324e-01 -1.57341197e-01 -3.43051761e-01
-2.68099308e-02 -1.91869885e-01 -1.43657252e-01 -3.27924252e-01
1.16512895e-01 2.64973342e-01 5.91950893e-01 -5.42041600e-01
5.20667732e-01 7.42238343e-01 6.47234559e-01 -8.79763424e-01
-4.71757412e-01 -5.36354959e-01 -6.68715000e-01 -1.59077391e-01
1.16236711e+00 -2.90185392e-01 -6.05568409e-01 2.66102910e-01
-1.21989775e+00 -3.87608379e-01 -3.90822321e-01 -8.03949833e-02
-7.51651943e-01 7.58702755e-01 -4.79045808e-01 -7.46102273e-01
1.38394922e-01 -1.56255043e+00 9.00587440e-01 4.66706187e-01
-6.72436804e-02 -1.05953860e+00 -7.91861191e-02 3.91217619e-01
-1.77791223e-01 8.04628372e-01 1.14974475e+00 -4.58919317e-01
-7.96793938e-01 -5.70124499e-02 -2.84689426e-01 1.75366309e-02
-2.10911974e-01 2.90711164e-01 -2.45701253e-01 -2.05260485e-01
-1.89752132e-01 -1.25699388e-02 8.50195706e-01 6.77001595e-01
1.11175346e+00 -4.21977431e-01 -7.15276062e-01 4.89026666e-01
1.72209799e+00 3.40904236e-01 4.69401389e-01 3.39802563e-01
2.73531228e-01 7.85345733e-01 4.02875036e-01 6.17472157e-02
4.68970425e-02 7.83942997e-01 4.87796366e-01 -5.03634274e-01
1.40758157e-01 3.66795689e-01 -1.84711009e-01 2.05940887e-01
1.06376827e-01 -4.78409320e-01 -6.41464531e-01 5.80964088e-01
-1.94968843e+00 -6.70742154e-01 -8.85273278e-01 2.27036023e+00
5.48726559e-01 2.91479733e-02 5.17423689e-01 3.22288632e-01
1.10459793e+00 -1.11226879e-01 -5.38221419e-01 -8.95798743e-01
-4.12286460e-01 2.55306900e-01 7.58456588e-01 8.19228113e-01
-1.24595785e+00 6.21399522e-01 7.34964561e+00 7.55047023e-01
-7.31658638e-01 -1.93936393e-01 9.07051682e-01 5.85306920e-02
-4.76485133e-01 2.11201817e-01 -7.03936398e-01 3.88395727e-01
1.48905009e-01 8.84236619e-02 4.20876116e-01 4.23386961e-01
6.52721748e-02 -7.35269189e-01 -1.03000653e+00 4.75263447e-01
-1.17719755e-01 -1.32789862e+00 -2.43782848e-01 3.05776685e-01
1.16465259e+00 -5.12108266e-01 -5.11431694e-02 -7.42707551e-01
6.08721115e-02 -8.87784004e-01 7.90465891e-01 7.39557147e-02
4.45380181e-01 -6.16106689e-01 9.11543667e-02 1.56423807e-01
-1.47422218e+00 1.25818580e-01 -1.23627715e-01 1.94312364e-01
5.87109566e-01 5.55694103e-01 -3.10313672e-01 6.16130829e-01
4.46053475e-01 1.10417366e-01 -1.87046424e-01 1.78501391e+00
-1.06971428e-01 2.97976851e-01 -5.90170145e-01 3.16853285e-01
4.71066236e-01 -8.45210850e-01 7.10718870e-01 1.06971776e+00
1.61411896e-01 9.22631323e-02 5.36444664e-01 9.51787591e-01
1.83898196e-01 2.46579591e-02 -2.44581640e-01 1.94992110e-01
2.87593693e-01 1.14815390e+00 -1.60497248e+00 -2.59820580e-01
-4.05059159e-01 9.33551669e-01 1.22274980e-01 5.28673589e-01
-7.78447509e-01 -5.32615364e-01 4.33769792e-01 7.98906311e-02
7.60110259e-01 -2.15411544e-01 -6.81869030e-01 -6.64836049e-01
4.60522994e-02 -4.48926061e-01 3.98480386e-01 -5.36161125e-01
-8.93075466e-01 4.74623919e-01 1.19591746e-02 -8.09212625e-01
8.54179263e-02 -7.88629711e-01 -8.17657530e-01 8.33487272e-01
-1.50781274e+00 -7.36475348e-01 -1.02674276e-01 4.23638940e-01
9.97335687e-02 7.26571739e-01 4.79687929e-01 -1.82024956e-01
-7.66754806e-01 7.56071731e-02 3.15771013e-01 -1.48744836e-01
-1.50964037e-01 -1.35167348e+00 1.37845762e-02 9.26092207e-01
-1.72458023e-01 2.97293991e-01 6.71556652e-01 -5.68364501e-01
-1.07725763e+00 -9.10729766e-01 7.99588263e-01 2.70512670e-01
2.98452288e-01 -1.10605732e-01 -7.23677874e-01 5.65739751e-01
2.80499220e-01 7.59877264e-02 4.80966717e-01 -3.23816031e-01
9.50154737e-02 3.60424548e-01 -1.35163569e+00 4.28512067e-01
1.02558839e+00 -8.75456408e-02 -2.88212061e-01 6.93524897e-01
4.49934870e-01 -3.95799130e-01 -6.86496615e-01 4.63829696e-01
2.83493102e-01 -9.62636352e-01 1.07560563e+00 -5.31478763e-01
3.96584839e-01 -7.12781847e-01 -3.46700586e-02 -8.62075448e-01
-2.31900364e-01 -7.40546644e-01 6.00704551e-02 1.03465056e+00
5.53535461e-01 -4.85461533e-01 1.03128147e+00 5.33986628e-01
-2.05171973e-01 -1.12821615e+00 -1.30282366e+00 -1.14292073e+00
2.78313339e-01 6.66856915e-02 6.23571217e-01 7.51342952e-01
1.10310294e-01 -1.09526962e-02 1.50292113e-01 8.43983814e-02
5.66383362e-01 1.04255021e+00 2.30222747e-01 -1.17300272e+00
-6.33823633e-01 -7.94779003e-01 -1.56558737e-01 -1.33511460e+00
1.04864538e-01 -1.00851512e+00 2.52088398e-01 -1.93740165e+00
2.77572274e-01 -7.68156946e-01 2.24524170e-01 4.34879869e-01
1.22413374e-01 4.72883582e-01 1.82395712e-01 -2.80440263e-02
-4.90676731e-01 -3.02189797e-01 1.51928687e+00 -1.03465579e-01
-3.44333231e-01 1.90907747e-01 -5.38469076e-01 7.28267848e-01
6.91817045e-01 -1.29182577e-01 -1.50921317e-02 -2.80422866e-01
3.34131598e-01 4.76328164e-01 3.59172672e-01 -7.94447601e-01
-9.57240164e-02 -4.47593182e-01 -1.20766483e-01 -6.11181438e-01
2.53671974e-01 -9.89049971e-01 3.92146349e-01 5.57773829e-01
-5.17913960e-02 -2.09648162e-02 2.70932347e-01 4.91055965e-01
-2.91242376e-02 -9.18458998e-01 9.51153338e-01 -1.88873380e-01
-5.21377385e-01 1.31364718e-01 -5.98798931e-01 -6.60066120e-03
1.64904976e+00 -1.03270769e+00 -5.51496327e-01 1.29223391e-01
-1.00697446e+00 5.44679046e-01 7.90173769e-01 -2.84105808e-01
3.99230868e-01 -8.73480618e-01 -1.78100258e-01 1.88788609e-03
-1.43483192e-01 2.47549102e-01 7.43872225e-02 1.09818065e+00
-8.57538700e-01 3.47015381e-01 -7.05984607e-02 -7.09853351e-01
-1.49428570e+00 6.91602945e-01 4.28327978e-01 -3.14660102e-01
-4.38333362e-01 9.15687680e-01 4.22399998e-01 -1.83853395e-02
1.39194906e-01 -4.88906652e-01 -7.02920777e-04 8.15687925e-02
8.26701596e-02 7.65619099e-01 -1.13264330e-01 -7.68453777e-01
-2.09947661e-01 1.03873479e+00 2.67014623e-01 -1.04665518e-01
1.04103911e+00 -1.88103262e-02 -6.30006850e-01 1.17190719e-01
1.09186840e+00 -1.53765738e-01 -1.37421548e+00 1.14118628e-01
-6.45937696e-02 -3.37441534e-01 -2.06766129e-01 -7.32766569e-01
-1.31792390e+00 4.29444283e-01 1.73541695e-01 5.55319428e-01
1.31243706e+00 3.45917314e-01 9.15479720e-01 8.97401497e-02
3.01902860e-01 -1.20607567e+00 -2.00548366e-01 5.11679411e-01
6.68979704e-01 -5.96713901e-01 4.91639674e-02 -1.10780823e+00
-4.15296078e-01 1.23944128e+00 3.33627492e-01 -2.64281005e-01
6.21273041e-01 5.58188438e-01 -3.10374588e-01 -5.54818287e-02
-1.33161858e-01 -3.43555748e-01 1.24374799e-01 3.56601864e-01
-3.99760157e-02 4.54094782e-02 -7.41551220e-01 2.75966048e-01
-8.33875500e-03 -1.20758690e-01 4.65492159e-01 1.16961384e+00
-8.30851972e-01 -1.19520915e+00 -5.80407500e-01 4.10492420e-01
-2.32760146e-01 2.74167150e-01 -6.28613830e-01 5.15895247e-01
4.72432047e-01 9.38671410e-01 4.49272364e-01 -1.13740833e-02
3.75964753e-02 -1.99009329e-01 8.40533733e-01 -5.81009746e-01
-7.62807906e-01 4.08774614e-01 1.50511637e-01 -3.41040134e-01
-5.95637858e-01 -8.59891415e-01 -1.60350096e+00 -2.30551306e-02
-6.20614827e-01 2.39714384e-01 3.45867991e-01 9.47236836e-01
-6.81005567e-02 1.78063110e-01 5.21709621e-01 -6.18896484e-01
4.91667390e-02 -7.95673579e-02 -8.90701473e-01 9.65722743e-03
3.30513746e-01 -4.48695928e-01 -2.72703797e-01 2.26918012e-01]
|
[14.369312286376953, -3.149040937423706]
|
0a0a1ddb-c0ba-4a46-8369-1a2cbb64f21e
|
data-driven-simulation-of-inelastic-materials
|
2101.10730
| null |
https://arxiv.org/abs/2101.10730v2
|
https://arxiv.org/pdf/2101.10730v2.pdf
|
Model-free Data-Driven simulation of inelastic materials using structured data sets, tangent space information and transition rules
|
Model-free data-driven computational mechanics replaces phenomenological constitutive functions by numerical simulations based on data sets of representative samples in stress-strain space. The distance of strain and stress pairs from the data set is minimized, subject to equilibrium and compatibility constraints. Although this method operates well for non-linear elastic problems, there are challenges dealing with history-dependent materials, since one and the same point in stress-strain space might correspond to different material behaviour.In recent literature, this issue has been treated by including local histories into the data set. However, there is still the necessity to include models for the evolution of specific internal variables. Thus, a mixed formulation of classical and data-driven modeling is obtained. In the presented approach, the data set is augmented with directions in the tangent space of points in stress-strain space. Moreover, the data set is divided into subsets corresponding to different material behaviour. Based on this classification, transition rules map the modeling points to the various subsets. The approach will be applied to non-linear elasticity and elasto-plasticity with isotropic hardening.
|
['Klaus Hackl', 'Kerem Ciftci']
|
2021-01-26
| null | null | null | null |
['non-linear-elasticity']
|
['miscellaneous']
|
[ 3.81421372e-02 -2.69265890e-01 -9.98543203e-02 -1.22488022e-01
-3.17228809e-02 -1.99817330e-01 2.58195996e-01 3.27618450e-01
-5.69475114e-01 7.20692039e-01 -3.35014910e-01 4.90388632e-01
-8.25525820e-01 -8.71081412e-01 -4.78203237e-01 -9.82906640e-01
2.01851167e-02 8.84186149e-01 5.54097056e-01 -2.16839671e-01
2.64812589e-01 7.30727017e-01 -1.29621291e+00 -1.98160440e-01
7.65208542e-01 6.24023795e-01 1.61646888e-01 4.19592828e-01
1.68128401e-01 -3.67586799e-02 -7.74811134e-02 -3.70143652e-02
-4.19168472e-02 -4.19199258e-01 -7.90649831e-01 2.64188439e-01
-1.49743229e-01 -1.95923448e-01 1.31302297e-01 9.55003381e-01
2.84434915e-01 5.23155451e-01 8.22546363e-01 -8.09588909e-01
-3.91034454e-01 3.11920434e-01 -3.70673507e-01 -1.39872760e-01
2.69920528e-01 -1.78318530e-01 6.38243794e-01 -1.02726114e+00
1.03231084e+00 8.13171029e-01 4.68849748e-01 4.78767812e-01
-1.59452271e+00 3.07083189e-01 -2.61878759e-01 3.01259577e-01
-1.15560830e+00 6.90774433e-03 1.35114503e+00 -8.60633075e-01
5.34796476e-01 4.18266773e-01 8.44147444e-01 8.37477267e-01
5.99612296e-01 1.77371316e-02 1.21811807e+00 -4.72053319e-01
7.45330811e-01 3.73573638e-02 5.74412942e-01 -8.56905803e-02
5.85588694e-01 2.17051759e-01 -2.15058774e-01 -1.84388027e-01
5.69811225e-01 -1.18762203e-01 -2.45945185e-01 -8.52633476e-01
-6.65524244e-01 4.50414956e-01 3.95997874e-02 8.55157971e-01
-5.70932984e-01 -3.89151983e-02 3.57547432e-01 1.55521249e-02
6.24803603e-01 3.39601696e-01 -2.70102531e-01 -4.41279821e-02
-1.09885061e+00 6.63189113e-01 7.61264205e-01 2.09055260e-01
9.17678058e-01 1.69220313e-01 5.15063524e-01 5.97531199e-01
3.97222608e-01 2.85188138e-01 3.39216113e-01 -7.31090069e-01
3.30992267e-02 6.43091738e-01 1.09607801e-01 -1.06895220e+00
-6.07706130e-01 -5.02064645e-01 -7.76273370e-01 5.10875583e-01
6.02524221e-01 -2.21645366e-02 -5.64918220e-01 1.46729827e+00
6.62808120e-01 1.47072719e-02 2.53496785e-02 1.08202720e+00
1.27681643e-01 3.96653563e-01 -3.15872952e-02 -7.59273410e-01
1.04113102e+00 -1.24281615e-01 -5.96944451e-01 1.96815863e-01
3.19721162e-01 -6.39391363e-01 7.38326609e-01 3.84086162e-01
-1.45726562e+00 -5.16561806e-01 -9.49987292e-01 2.52709091e-01
-1.78057402e-01 -1.51536807e-01 -3.14126998e-01 1.68502972e-01
-6.07923329e-01 1.16382694e+00 -1.11874688e+00 -2.25524366e-01
-4.51041341e-01 2.00053543e-01 -1.86305210e-01 2.66252071e-01
-1.08221376e+00 1.08407891e+00 4.19171661e-01 6.66435778e-01
-2.91128486e-01 -6.73156261e-01 -4.51633215e-01 -3.49160768e-02
3.45354497e-01 -5.19526958e-01 6.05692029e-01 -4.84690070e-01
-1.63433528e+00 6.89950228e-01 1.54130742e-01 -3.14184986e-02
9.95475411e-01 -1.07178748e-01 -1.77939326e-01 1.20040603e-01
-3.14571351e-01 -3.82081807e-01 6.79520130e-01 -1.67812634e+00
4.48721170e-01 -3.33743185e-01 -3.14968586e-01 -5.95587716e-02
-2.48075604e-01 -7.31798336e-02 -1.38797343e-01 -5.30060649e-01
5.71545362e-01 -1.14919341e+00 -2.99917668e-01 -3.01962823e-01
-5.16580760e-01 1.05621301e-01 9.06475008e-01 -6.44179881e-01
1.12737024e+00 -1.94362354e+00 8.38975847e-01 6.18971527e-01
1.26190662e-01 2.15394069e-02 4.08372939e-01 7.66480744e-01
-4.72278520e-02 -1.11855313e-01 -7.86698341e-01 -7.86109567e-02
-1.24880940e-01 3.38623315e-01 6.05617836e-02 6.55251443e-01
4.60292064e-02 4.01972860e-01 -7.29131460e-01 -6.43918216e-01
1.91911086e-01 4.72638249e-01 -5.89381337e-01 -2.84198839e-02
-3.21325362e-01 6.82331204e-01 -6.49408698e-01 2.31756151e-01
8.87687802e-01 3.67127866e-01 2.91685075e-01 -2.18688324e-01
-4.55961317e-01 -3.47188264e-01 -1.37362194e+00 1.26593435e+00
-2.71664232e-01 6.93102484e-04 2.04113349e-01 -1.28962255e+00
1.39810967e+00 5.56517184e-01 8.96749914e-01 -2.04949856e-01
2.86307037e-01 7.28542209e-01 2.98318774e-01 -8.17948759e-01
4.91282284e-01 -3.35960954e-01 5.92196025e-02 3.26692522e-01
-1.63958624e-01 -4.14669067e-01 4.51286316e-01 -6.07383311e-01
5.90605795e-01 6.28463864e-01 -7.36322775e-02 -5.27452409e-01
8.08091938e-01 4.25192006e-02 5.83958447e-01 7.69985244e-02
1.73362017e-01 5.15351295e-01 5.40937364e-01 -2.35010862e-01
-1.31924295e+00 -8.92590165e-01 -6.79294527e-01 4.52383608e-01
2.94838279e-01 3.04167736e-02 -7.57415652e-01 1.69155791e-01
8.23757425e-03 5.73835611e-01 -6.62343025e-01 -3.11678052e-01
-1.19879544e+00 -9.98058021e-01 -3.63597125e-01 1.88957125e-01
2.02619389e-01 -1.14087868e+00 -8.17759633e-01 5.62797785e-01
3.69527638e-01 -7.58049369e-01 4.67957482e-02 9.97484475e-02
-1.40306973e+00 -1.04273403e+00 -6.50996327e-01 -4.56191272e-01
7.72104502e-01 -4.99291658e-01 7.29098976e-01 6.72011077e-02
-2.38051906e-01 2.87455589e-01 -3.90224695e-01 2.76490688e-01
-9.30784881e-01 3.95878591e-02 1.52860209e-01 2.06054181e-01
-2.39231348e-01 -7.71180511e-01 -5.11134565e-01 5.42089880e-01
-1.17822993e+00 -3.38652968e-01 3.46467458e-02 5.16635120e-01
8.73018682e-01 1.12469003e-01 5.25113285e-01 -6.89556003e-01
4.89553183e-01 -4.85972196e-01 -5.47827125e-01 1.50721788e-01
-5.97078502e-01 8.91644359e-02 8.64309192e-01 -6.92607343e-01
-1.07898402e+00 1.33426897e-02 -1.60909668e-01 -4.35039550e-01
-1.00993849e-01 8.53702664e-01 -2.26545706e-01 -5.57451099e-02
4.90665555e-01 1.32513061e-01 1.57285929e-01 -8.24361980e-01
-8.09069350e-02 1.24664441e-01 2.52476484e-01 -1.03048575e+00
7.08821297e-01 1.45813495e-01 7.44322598e-01 -8.34422827e-01
9.45806056e-02 -1.66824199e-02 -1.05299330e+00 -5.90261221e-01
5.74680805e-01 3.78930956e-01 -6.93794668e-01 6.68514431e-01
-9.34184909e-01 -9.66684520e-02 -7.95010448e-01 6.86298668e-01
-8.45895052e-01 5.87700605e-01 -5.54952562e-01 -8.48625898e-01
-2.20128834e-01 -1.46003878e+00 4.86001670e-01 8.22678134e-02
-2.28532523e-01 -1.32370472e+00 5.45761645e-01 -1.45178095e-01
3.79604489e-01 7.52470970e-01 1.06779671e+00 -4.69309747e-01
-1.91100091e-01 -6.39823139e-01 6.91241324e-01 4.47254032e-01
-5.15978923e-03 4.75522369e-01 -4.57958966e-01 -2.58960158e-01
5.87318420e-01 1.15514040e-01 5.66244543e-01 6.12854123e-01
6.83437109e-01 -5.69097176e-02 -2.29054406e-01 1.71036810e-01
1.92167532e+00 2.88821250e-01 3.62392813e-01 1.49873823e-01
4.61049139e-01 1.01704252e+00 4.60866213e-01 3.47753406e-01
-3.30067843e-01 1.10714757e+00 3.86741459e-01 2.96027303e-01
1.10725686e-01 2.37627149e-01 6.58549890e-02 8.95879745e-01
-6.59789622e-01 -1.49871930e-01 -7.77795136e-01 4.69911218e-01
-1.81823301e+00 -8.81809771e-01 -6.03089094e-01 2.39553475e+00
7.79609442e-01 3.23805988e-01 2.17696771e-01 5.71517706e-01
9.13630903e-01 3.97183783e-02 -5.91513276e-01 -6.56502843e-01
-8.84761289e-02 8.11764151e-02 1.56448901e-01 8.64865601e-01
-5.04775763e-01 7.83976093e-02 5.71302509e+00 2.81296551e-01
-1.52641368e+00 -6.27037790e-03 1.48821101e-01 1.95154101e-01
-3.18578660e-01 2.29834706e-01 -4.53730673e-01 8.46572220e-01
1.02673197e+00 -4.07231063e-01 1.63262054e-01 7.13367283e-01
3.91710550e-01 -3.44301283e-01 -8.98812950e-01 2.66557127e-01
-2.99212962e-01 -9.96967375e-01 -4.67738122e-01 1.01902828e-01
5.07061362e-01 -5.12779474e-01 -9.70778540e-02 -4.44283009e-01
-5.51744819e-01 -2.60739952e-01 7.78194904e-01 1.07953870e+00
6.18408322e-01 -3.36841404e-01 6.55397236e-01 3.76343757e-01
-1.02087831e+00 2.07831353e-01 -2.89169282e-01 1.44525960e-01
7.06939220e-01 8.37266147e-01 -4.68571573e-01 8.63144040e-01
2.33034819e-01 5.61558902e-01 -1.16683245e-01 9.80366826e-01
3.84834439e-01 4.84368980e-01 -2.98507899e-01 -9.23253000e-02
-2.48972580e-01 -8.62595737e-01 9.32892680e-01 7.14205205e-01
5.07558465e-01 -5.01205847e-02 -1.83386356e-01 1.04333317e+00
4.40634668e-01 2.27648854e-01 -4.11146283e-01 3.80641282e-01
1.24168791e-01 9.42666471e-01 -9.67660964e-01 7.27858096e-02
1.25123039e-01 4.99290109e-01 -2.35769987e-01 4.86991197e-01
-6.06591761e-01 5.81855774e-02 2.30293557e-01 7.64078975e-01
-2.32285768e-01 -7.47149825e-01 -2.62623966e-01 -7.81279564e-01
1.59816533e-01 -5.77900559e-02 8.57659876e-02 -5.06271839e-01
-1.11338663e+00 5.85452139e-01 8.92519176e-01 -1.16223001e+00
-4.81693417e-01 -6.33422792e-01 -5.46058953e-01 7.95198977e-01
-8.64972472e-01 -7.28075206e-01 -4.23040576e-02 2.56124675e-01
1.79538935e-01 3.42058867e-01 4.63616610e-01 3.24920237e-01
-5.31420648e-01 -4.70803631e-03 4.69074845e-01 -3.26126307e-01
3.75134170e-01 -1.11615980e+00 -2.10665017e-01 7.29304969e-01
-7.29623020e-01 5.44934571e-01 1.26996684e+00 -1.05243063e+00
-1.21255410e+00 -4.95551437e-01 8.21560085e-01 2.30266646e-01
7.17906475e-01 -2.04389751e-01 -1.59009969e+00 2.43601918e-01
-2.11974114e-01 3.06126982e-01 1.85571834e-01 -2.33819798e-01
4.12515074e-01 -1.70899674e-01 -1.15873253e+00 3.98208678e-01
6.65368438e-01 -1.88500851e-01 -3.73599142e-01 5.11451215e-02
9.57675651e-02 -2.59618878e-01 -1.36309028e+00 5.92327595e-01
3.97741675e-01 -7.67423928e-01 9.28344786e-01 -4.52289701e-01
5.40178478e-01 -1.78949967e-01 2.43089810e-01 -1.16270900e+00
-1.94607690e-01 -5.14500499e-01 -3.45923230e-02 1.17398345e+00
-1.17454007e-01 -7.43351936e-01 7.37325549e-01 8.46130908e-01
-3.37661684e-01 -9.90937650e-01 -1.28624058e+00 -8.07264149e-01
4.98166054e-01 -1.88354596e-01 2.46628001e-01 4.82540101e-01
-1.48953930e-01 -2.98579544e-01 -1.59502208e-01 -1.34596899e-01
7.32893229e-01 1.98986232e-01 1.14699863e-01 -1.54023278e+00
-2.60552794e-01 -3.57412308e-01 -2.90244013e-01 -2.89291412e-01
3.89242247e-02 -5.58853567e-01 5.56250289e-02 -1.35981739e+00
-1.15840964e-01 -7.44553387e-01 -2.19965223e-02 -1.56879947e-01
1.24651514e-01 8.30855668e-02 -8.95156711e-02 7.61397898e-01
4.01668727e-01 4.27327663e-01 1.44493902e+00 3.67392242e-01
-4.88823324e-01 1.36794701e-01 3.42545927e-01 5.38005829e-01
7.03885555e-01 -3.35187644e-01 -2.65943229e-01 -5.83818592e-02
3.12844127e-01 4.09132302e-01 2.87879378e-01 -9.18669581e-01
1.16140977e-01 -5.35020947e-01 -2.22588599e-01 -4.26002294e-01
3.37074161e-01 -1.01514971e+00 9.94626522e-01 5.80171227e-01
-2.81961918e-01 6.23413268e-03 -2.60163605e-01 2.88856030e-01
-9.57869589e-02 -8.55900049e-01 1.00545061e+00 1.66149452e-01
-2.53744543e-01 2.15933956e-02 -3.69970679e-01 -2.97944576e-01
1.11010528e+00 -7.72110820e-01 -2.86093969e-02 3.84302884e-01
-1.36863160e+00 -5.48334360e-01 7.53281176e-01 -1.58138558e-01
5.51810265e-01 -1.13273895e+00 -6.21101081e-01 2.62409866e-01
-2.37528145e-01 1.42675668e-01 7.68378496e-01 1.07286870e+00
-7.95712769e-01 6.52786810e-04 -4.12963599e-01 -5.89082539e-01
-8.97731960e-01 5.74106038e-01 6.22274697e-01 -2.12506726e-01
-4.02933598e-01 2.54802763e-01 -3.18659872e-01 8.60348120e-02
-5.21122873e-01 -4.67746884e-01 -2.58737087e-01 2.21509293e-01
-3.23908329e-01 9.23333943e-01 3.36882412e-01 -1.06869769e+00
-3.10187876e-01 1.16186130e+00 4.58071589e-01 -2.35795230e-01
1.44543159e+00 -4.49994542e-02 -3.54207575e-01 7.54419327e-01
1.17191553e+00 1.55531019e-01 -1.22952521e+00 -1.51879400e-01
2.01144859e-01 -3.30565721e-01 -1.91792697e-01 -2.88838774e-01
-9.27854359e-01 5.09957731e-01 2.22153977e-01 5.04944205e-01
1.06452489e+00 -5.50222993e-02 5.26152551e-01 -3.06038946e-01
1.71525478e-01 -1.16463435e+00 -1.29529610e-01 1.65729597e-01
1.19985938e+00 -5.05602479e-01 1.90654188e-01 -5.95608234e-01
2.57709231e-02 1.53278685e+00 4.38030481e-01 -5.09238422e-01
9.32747245e-01 4.28535640e-01 -3.81436229e-01 -1.36401439e-02
-2.00936034e-01 3.10694486e-01 3.22618335e-01 1.92294911e-01
4.60700393e-01 -6.62563518e-02 -1.29790151e+00 2.72691786e-01
1.30255464e-02 1.73502475e-01 5.41860402e-01 1.22521985e+00
-2.80421078e-01 -1.53391540e+00 -7.83278167e-01 3.70159373e-02
-1.75822198e-01 4.94551778e-01 1.09653614e-01 9.79493916e-01
1.97688654e-01 5.25985181e-01 2.10943207e-01 -1.62658319e-01
5.54507315e-01 1.97392508e-01 4.61144060e-01 -3.74268532e-01
-4.16485012e-01 3.04129153e-01 -1.26094252e-01 2.50138659e-02
-5.74424565e-01 -1.05494547e+00 -1.35579467e+00 -3.84085268e-01
-3.88428807e-01 3.21591616e-01 7.75473058e-01 9.07649457e-01
-6.13237396e-02 3.44878197e-01 7.58141339e-01 -1.01010656e+00
-5.66317618e-01 -7.50547767e-01 -8.86708260e-01 7.09794760e-01
4.47041355e-02 -9.59464669e-01 -3.46136123e-01 2.03741595e-01]
|
[6.2734375, 3.2708609104156494]
|
e0a42e12-007d-4da8-94c1-0ac61649e90c
|
associatively-segmenting-instances-and
|
1902.09852
| null |
http://arxiv.org/abs/1902.09852v2
|
http://arxiv.org/pdf/1902.09852v2.pdf
|
Associatively Segmenting Instances and Semantics in Point Clouds
|
A 3D point cloud describes the real scene precisely and intuitively.To date
how to segment diversified elements in such an informative 3D scene is rarely
discussed. In this paper, we first introduce a simple and flexible framework to
segment instances and semantics in point clouds simultaneously. Then, we
propose two approaches which make the two tasks take advantage of each other,
leading to a win-win situation. Specifically, we make instance segmentation
benefit from semantic segmentation through learning semantic-aware point-level
instance embedding. Meanwhile, semantic features of the points belonging to the
same instance are fused together to make more accurate per-point semantic
predictions. Our method largely outperforms the state-of-the-art method in 3D
instance segmentation along with a significant improvement in 3D semantic
segmentation. Code has been made available at:
https://github.com/WXinlong/ASIS.
|
['Jiaya Jia', 'Chunhua Shen', 'Xiaoyong Shen', 'Shu Liu', 'Xinlong Wang']
|
2019-02-26
|
associatively-segmenting-instances-and-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Wang_Associatively_Segmenting_Instances_and_Semantics_in_Point_Clouds_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Wang_Associatively_Segmenting_Instances_and_Semantics_in_Point_Clouds_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['3d-instance-segmentation-1']
|
['computer-vision']
|
[-4.88531627e-02 2.65739143e-01 -2.81286359e-01 -6.30643129e-01
-9.29972172e-01 -5.95373094e-01 4.88506943e-01 1.72093809e-01
4.80200201e-02 2.95168906e-01 -4.07111384e-02 8.46511498e-03
-7.30742440e-02 -9.04432535e-01 -9.79947805e-01 -3.74659389e-01
2.40885288e-01 8.20018530e-01 5.50530016e-01 -1.27699850e-02
3.26344728e-01 7.93829799e-01 -1.68370497e+00 9.85635966e-02
9.04714108e-01 9.95986283e-01 5.93609214e-01 1.96027249e-01
-7.45067358e-01 -2.27030199e-02 -2.11096704e-01 -2.42843360e-01
4.77750748e-01 7.21693635e-02 -9.82932508e-01 4.33581114e-01
3.45580935e-01 -6.94389790e-02 -7.73582533e-02 9.61807430e-01
2.05710515e-01 1.58291042e-01 6.70689702e-01 -1.20666182e+00
-2.85543770e-01 3.98084611e-01 -7.79129386e-01 -2.12510616e-01
1.73537865e-01 2.78153419e-02 1.11277997e+00 -8.60680580e-01
6.63089991e-01 1.17937493e+00 5.33229232e-01 3.99401128e-01
-1.11414266e+00 -5.22256374e-01 4.68282074e-01 -1.85271204e-02
-1.29428458e+00 -2.85661966e-02 1.21298766e+00 -4.95100647e-01
8.10815096e-01 1.53061897e-01 9.19539750e-01 6.82889760e-01
-4.35831606e-01 1.17423451e+00 7.46963263e-01 -1.86930001e-02
2.19176546e-01 5.83332255e-02 2.81552583e-01 5.12892127e-01
1.81053549e-01 -2.21295372e-01 -3.01562428e-01 1.49176553e-01
8.06512356e-01 3.69812191e-01 -1.79196671e-02 -7.73727417e-01
-1.19495463e+00 6.93525136e-01 8.75093400e-01 2.91267157e-01
-6.14449918e-01 2.43863955e-01 1.86759010e-01 -1.70850918e-01
7.55347967e-01 4.33987886e-01 -6.52303398e-01 -2.99742483e-02
-1.01937270e+00 4.16666955e-01 5.44115961e-01 1.21350169e+00
1.20168054e+00 -3.07166576e-01 1.53361887e-01 6.95307553e-01
4.11172271e-01 4.32507843e-01 -4.90661040e-02 -1.15810645e+00
2.54370272e-01 1.06670773e+00 -4.77065652e-04 -6.72968388e-01
-3.74193788e-01 -3.86523932e-01 -2.50224143e-01 2.08202451e-01
1.74903810e-01 2.90098011e-01 -1.22384095e+00 1.17748857e+00
7.74617612e-01 5.50471425e-01 -2.26549745e-01 9.89571154e-01
8.97094905e-01 7.80351996e-01 1.23200588e-01 4.97157365e-01
1.21676373e+00 -9.90936458e-01 -2.05731332e-01 -2.69843668e-01
4.93859619e-01 -6.54721916e-01 1.15490675e+00 6.56251535e-02
-1.04256833e+00 -5.98284960e-01 -9.26843345e-01 -3.68517995e-01
-4.23658907e-01 -2.60361135e-01 7.08947480e-01 3.52229595e-01
-7.79335558e-01 5.81363142e-01 -1.05121100e+00 -2.42044955e-01
9.04715121e-01 3.26902777e-01 -2.08131373e-01 -7.88014978e-02
-7.29479074e-01 5.25522828e-01 5.44066608e-01 -2.79287219e-01
-7.36016691e-01 -1.04102576e+00 -8.59664738e-01 -1.27782255e-01
5.86031437e-01 -9.24307108e-01 1.30996203e+00 -7.47012138e-01
-1.31294799e+00 1.20942485e+00 -2.65866876e-01 -1.49440780e-01
4.17518109e-01 -4.56878155e-01 1.50233611e-01 2.24476203e-01
3.79733056e-01 1.01493275e+00 4.65364993e-01 -1.83441532e+00
-7.59951174e-01 -7.40830123e-01 2.73829728e-01 3.87015164e-01
2.07337618e-01 -4.44671810e-01 -9.48419929e-01 -3.25584292e-01
4.53673631e-01 -8.10898185e-01 -4.57004368e-01 1.60282880e-01
-6.31616712e-01 -3.85918587e-01 8.12286019e-01 -4.17828202e-01
6.11104131e-01 -2.19694018e+00 1.67381242e-01 1.14254974e-01
2.57920295e-01 1.04887277e-01 7.78805912e-02 2.74184883e-01
2.11891279e-01 2.57511407e-01 -5.48479617e-01 -5.86065829e-01
2.61681437e-01 3.30359638e-01 -1.68600976e-01 3.03344905e-01
2.53455043e-01 1.12896299e+00 -9.39548910e-01 -5.04792988e-01
6.54671490e-01 5.84378242e-01 -5.39588273e-01 1.30714089e-01
-6.39348090e-01 6.38501823e-01 -1.07216489e+00 9.14878488e-01
8.60398710e-01 -3.99575472e-01 -2.63884872e-01 -1.66858330e-01
-1.12522304e-01 2.26012498e-01 -1.05086780e+00 2.17935491e+00
-3.83411080e-01 1.98661312e-01 -2.11625442e-01 -1.09226739e+00
1.04466999e+00 1.00289450e-04 7.66698360e-01 -4.62284267e-01
4.56493571e-02 2.75443166e-01 -6.95661187e-01 -2.50558525e-01
5.16609132e-01 -1.78197518e-01 -1.90034181e-01 3.26751806e-02
4.59180921e-02 -8.05998087e-01 -1.95191324e-01 7.25381076e-02
5.25026023e-01 5.04718244e-01 2.68703431e-01 -9.35821310e-02
2.56835401e-01 4.76097107e-01 5.15542984e-01 5.32308519e-01
-6.88747093e-02 8.48225892e-01 2.53773272e-01 -3.08484554e-01
-1.03457713e+00 -1.34491181e+00 -2.40866795e-01 5.76836050e-01
9.88586068e-01 -2.47899204e-01 -6.79771781e-01 -8.02506328e-01
3.42396498e-01 8.84985030e-01 -4.00869906e-01 2.38902912e-01
-4.05873358e-01 -2.62237579e-01 5.56717776e-02 5.49294472e-01
4.27740306e-01 -8.17384839e-01 -5.43613613e-01 -2.43416727e-02
-1.41607951e-02 -1.13500321e+00 -3.05127382e-01 1.44361136e-02
-1.19346941e+00 -9.84337687e-01 -6.66675806e-01 -6.16532624e-01
5.43304980e-01 5.40176988e-01 1.31239760e+00 -3.47598642e-02
-1.69038370e-01 3.99371922e-01 -4.89191860e-01 -4.93517905e-01
5.94105991e-03 3.37411046e-01 -3.15894783e-01 -2.38865748e-01
6.16701365e-01 -6.29572809e-01 -6.64491415e-01 2.31135368e-01
-8.18566024e-01 3.37731957e-01 4.27709371e-01 2.50533938e-01
1.31951463e+00 -1.60913646e-01 1.95244521e-01 -9.93703306e-01
-1.65565405e-02 -6.15420401e-01 -5.68908870e-01 3.43037993e-02
-2.80386537e-01 -1.54389804e-02 2.15382636e-01 3.61631066e-02
-8.73452067e-01 3.27883303e-01 -3.94952267e-01 -7.47577071e-01
-6.42952144e-01 1.98921561e-01 -4.21939015e-01 8.08907822e-02
3.48336399e-01 1.14033341e-01 -1.79461718e-01 -7.53826559e-01
7.20746517e-01 5.68841875e-01 2.62372494e-01 -6.56564772e-01
8.46529841e-01 7.44380295e-01 -7.36643672e-02 -7.59469092e-01
-1.06814528e+00 -8.99316251e-01 -9.60921586e-01 -1.95019558e-01
1.06094968e+00 -9.81719911e-01 -2.73922086e-01 2.29858413e-01
-1.11616683e+00 -3.94509643e-01 -5.55266798e-01 3.91053349e-01
-8.66806626e-01 2.45631486e-01 -3.48544389e-01 -4.67376620e-01
-1.02989443e-01 -1.32205844e+00 1.64291787e+00 3.87065053e-01
5.21583334e-02 -8.87648523e-01 -2.82929502e-02 5.87122560e-01
-1.13756813e-01 4.17194098e-01 5.52994609e-01 -6.27168596e-01
-1.12547910e+00 -2.23497137e-01 -3.95512253e-01 1.82971701e-01
1.46420106e-01 8.58982429e-02 -9.80577707e-01 2.06476569e-01
-1.71484441e-01 3.42695974e-02 8.04175794e-01 6.56536996e-01
1.46688294e+00 8.81981328e-02 -6.41926408e-01 8.91146600e-01
1.51203156e+00 -8.74856636e-02 4.42834854e-01 3.17381084e-01
1.00504100e+00 6.21533036e-01 7.78468668e-01 3.60749125e-01
6.29191577e-01 8.25380087e-01 6.51918530e-01 -6.80646822e-02
-1.90734327e-01 -4.71246749e-01 -2.40587071e-01 5.77072442e-01
-4.10642615e-03 -1.39121309e-01 -1.07787418e+00 7.86915779e-01
-1.87092590e+00 -6.80981338e-01 -2.79024005e-01 1.95740390e+00
5.41510880e-01 2.22722203e-01 1.37962490e-01 -1.50550470e-01
7.18465865e-01 2.87480772e-01 -6.85195506e-01 -1.01492301e-01
7.64180422e-02 1.13705441e-01 6.07920587e-01 4.76527333e-01
-1.16328835e+00 1.20284295e+00 5.03106070e+00 9.80450273e-01
-9.02870595e-01 1.96562827e-01 6.49497449e-01 -9.72294137e-02
-6.45688951e-01 1.42320946e-01 -8.74896944e-01 4.06391829e-01
5.14635861e-01 -9.04233828e-02 1.25274524e-01 1.04330993e+00
7.33745769e-02 9.65969712e-02 -9.07396793e-01 1.07640696e+00
-2.24059507e-01 -1.54015982e+00 2.63282031e-01 1.21671192e-01
7.45370567e-01 3.04894775e-01 -1.33924142e-01 9.48069170e-02
1.59991205e-01 -8.06565344e-01 7.53998816e-01 5.18876374e-01
5.46981692e-01 -7.53857613e-01 4.36644286e-01 4.89099473e-01
-1.24128008e+00 2.93278515e-01 -2.69559503e-01 1.40601397e-01
4.88311678e-01 9.04734969e-01 -6.75726473e-01 9.21589434e-01
8.39721143e-01 1.00738537e+00 -3.11981678e-01 1.29573119e+00
-2.48002037e-01 4.68707412e-01 -4.67700362e-01 1.82620153e-01
5.92240453e-01 -3.07625294e-01 7.91223824e-01 1.02310050e+00
3.69273543e-01 2.13190824e-01 4.41657424e-01 1.13068473e+00
-1.50462195e-01 1.98839288e-02 -6.52389348e-01 6.54614121e-02
5.42359293e-01 1.17235637e+00 -9.35196936e-01 -3.51455510e-01
-3.45314413e-01 9.42842245e-01 2.18138739e-01 1.86901942e-01
-8.01096261e-01 -1.64711520e-01 9.49368119e-01 2.64587373e-01
4.52433467e-01 -3.06049019e-01 -8.80389988e-01 -1.04284823e+00
7.59889707e-02 -3.76214385e-02 9.66263264e-02 -8.26733589e-01
-1.31919611e+00 3.88371438e-01 2.01870620e-01 -1.31671882e+00
1.52181998e-01 -5.25605619e-01 -4.09004539e-01 7.72495329e-01
-1.68662071e+00 -1.31748998e+00 -4.06440556e-01 3.44607770e-01
8.75722587e-01 3.51246148e-01 4.22664732e-01 1.97201237e-01
-2.76707023e-01 1.26100793e-01 -2.82358732e-02 -1.70506209e-01
2.23171085e-01 -1.36445546e+00 6.68856204e-01 5.87956727e-01
2.91309744e-01 3.19344580e-01 4.75188017e-01 -7.53069043e-01
-9.66059566e-01 -1.19997168e+00 7.46691406e-01 -7.31653154e-01
3.49968672e-01 -3.79763246e-01 -1.04810596e+00 6.58172548e-01
-3.00488144e-01 -7.80499950e-02 4.54496503e-01 9.17806849e-02
-2.82073557e-01 1.51877431e-02 -1.16087651e+00 5.45739174e-01
1.37784660e+00 -4.05378014e-01 -7.78286397e-01 4.14122850e-01
1.15606010e+00 -5.67932785e-01 -9.05405879e-01 5.85726440e-01
1.82692021e-01 -8.72038603e-01 1.43007410e+00 -3.34711075e-01
5.61402142e-01 -4.78203923e-01 -3.33440065e-01 -1.15204537e+00
-1.89667046e-01 -2.61892974e-01 -2.89937332e-02 9.07070935e-01
3.18030179e-01 -4.84152555e-01 1.04535282e+00 4.70521122e-01
-6.45606756e-01 -8.63357186e-01 -8.59569132e-01 -7.57148504e-01
9.48713198e-02 -7.40088880e-01 1.08665097e+00 7.97446191e-01
-4.26789910e-01 3.37495059e-02 1.12850547e-01 4.25742418e-01
7.48335779e-01 6.49336696e-01 8.90591443e-01 -1.36722314e+00
1.80866301e-01 -7.49508858e-01 -6.68801188e-01 -1.47227836e+00
2.51473755e-01 -1.15390348e+00 2.41767988e-02 -1.90171802e+00
1.09267995e-01 -7.57807255e-01 -1.64707169e-01 3.89063179e-01
-3.03495914e-01 2.85151660e-01 3.57265145e-01 3.42898250e-01
-6.36006474e-01 6.60245419e-01 1.37872791e+00 -9.16080251e-02
-3.28771472e-01 2.53919184e-01 -7.49140739e-01 6.81675196e-01
1.12049270e+00 -4.85637367e-01 -4.14565116e-01 -6.77111089e-01
-2.29716197e-01 -7.46875070e-03 5.49996674e-01 -9.77386475e-01
-1.98839568e-02 -2.39434898e-01 2.78109461e-01 -1.04890895e+00
6.85117185e-01 -9.68394220e-01 2.42029786e-01 1.07857501e-02
-1.21797405e-01 -3.88526440e-01 2.24177673e-01 6.33299172e-01
-2.20425919e-01 -2.09007978e-01 7.53179252e-01 -3.98279071e-01
-1.11971974e+00 7.38829553e-01 3.59978259e-01 1.73803326e-02
1.25889874e+00 -6.27663434e-01 8.44318233e-03 2.45512165e-02
-5.89739740e-01 4.34339553e-01 8.71742070e-01 5.52329123e-01
6.54978275e-01 -1.12492359e+00 -5.28418064e-01 1.76473688e-02
3.29294801e-01 7.04535067e-01 5.79992414e-01 5.53257525e-01
-6.33878171e-01 4.20797646e-01 -2.31086891e-02 -9.92983699e-01
-9.14965987e-01 4.83110964e-01 2.55455375e-01 1.59568518e-01
-9.89709020e-01 9.82148767e-01 5.67781329e-01 -9.04077470e-01
-5.93586117e-02 -3.48883629e-01 4.29552384e-02 -1.27799988e-01
1.15767702e-01 1.96921080e-01 -1.17217995e-01 -7.25775898e-01
-4.71105069e-01 9.88054514e-01 1.90233022e-01 1.21359192e-01
1.44932675e+00 -1.65037826e-01 7.82635808e-02 6.63249731e-01
1.27957428e+00 -1.74214482e-01 -1.58266926e+00 -2.87345290e-01
-1.13128070e-02 -7.64555037e-01 1.43008456e-01 -6.82265997e-01
-1.16095603e+00 9.27453876e-01 3.54980230e-01 8.72156918e-02
9.01082098e-01 6.96771264e-01 1.02720582e+00 -5.36361784e-02
4.54940885e-01 -7.88716376e-01 -1.78363398e-01 3.15622449e-01
6.53138578e-01 -1.36083615e+00 -2.22466886e-02 -7.15146899e-01
-8.52798104e-01 8.52609098e-01 4.91991013e-01 -3.25408697e-01
7.04789579e-01 -1.27612531e-01 2.74689961e-02 -5.78993499e-01
-2.32493594e-01 -4.76245344e-01 5.27963281e-01 7.03234792e-01
5.73243089e-02 3.61517489e-01 9.78837311e-02 4.94941890e-01
-2.28839308e-01 1.03769440e-03 6.55014142e-02 6.51831329e-01
-6.51349366e-01 -1.04335856e+00 -2.91204929e-01 4.69342798e-01
-6.00492284e-02 1.36003882e-01 -2.62917846e-01 8.46273959e-01
3.92788202e-02 5.18703520e-01 2.51289517e-01 -3.45427394e-01
5.00691414e-01 3.32547650e-02 3.44268918e-01 -8.54360878e-01
-2.10587695e-01 2.09018197e-02 -2.31771573e-01 -8.01751196e-01
-5.14607072e-01 -7.71610200e-01 -1.67650723e+00 -1.72364935e-01
-5.94050884e-02 -2.13585831e-02 8.58865023e-01 7.00445473e-01
5.59714377e-01 3.86680007e-01 5.75622559e-01 -1.40158272e+00
-9.20381024e-02 -4.32527125e-01 -6.69651270e-01 5.08894265e-01
1.39472693e-01 -7.86301672e-01 -2.96091884e-01 -1.03493288e-01]
|
[8.061857223510742, -3.1707301139831543]
|
b154e35e-8e9c-4cf6-a5e5-8ca506fd7361
|
abstract-to-executable-trajectory-translation
|
2210.07658
| null |
https://arxiv.org/abs/2210.07658v2
|
https://arxiv.org/pdf/2210.07658v2.pdf
|
Abstract-to-Executable Trajectory Translation for One-Shot Task Generalization
|
Training long-horizon robotic policies in complex physical environments is essential for many applications, such as robotic manipulation. However, learning a policy that can generalize to unseen tasks is challenging. In this work, we propose to achieve one-shot task generalization by decoupling plan generation and plan execution. Specifically, our method solves complex long-horizon tasks in three steps: build a paired abstract environment by simplifying geometry and physics, generate abstract trajectories, and solve the original task by an abstract-to-executable trajectory translator. In the abstract environment, complex dynamics such as physical manipulation are removed, making abstract trajectories easier to generate. However, this introduces a large domain gap between abstract trajectories and the actual executed trajectories as abstract trajectories lack low-level details and are not aligned frame-to-frame with the executed trajectory. In a manner reminiscent of language translation, our approach leverages a seq-to-seq model to overcome the large domain gap between the abstract and executable trajectories, enabling the low-level policy to follow the abstract trajectory. Experimental results on various unseen long-horizon tasks with different robot embodiments demonstrate the practicability of our methods to achieve one-shot task generalization.
|
['Hao Su', 'Yuzhe Qin', 'Zhiao Huang', 'Tongzhou Mu', 'Xiaochen Li', 'Stone Tao']
|
2022-10-14
| null | null | null | null |
['few-shot-imitation-learning']
|
['methodology']
|
[ 1.64077714e-01 4.07002240e-01 -1.59458995e-01 -1.59675270e-01
-7.21961200e-01 -9.02714074e-01 8.45720470e-01 -5.29609323e-02
-3.10758501e-01 8.50163937e-01 3.08856577e-01 -3.68867248e-01
-7.94160962e-02 -5.44859946e-01 -1.11507535e+00 -3.65792930e-01
-1.41902000e-01 5.49938023e-01 2.02955469e-01 -4.67468381e-01
1.14674866e-02 5.22530615e-01 -1.29398501e+00 2.57637531e-01
8.20194006e-01 4.44196165e-01 5.66755414e-01 8.16717148e-01
9.42407474e-02 7.66276717e-01 -4.62321371e-01 2.01958060e-01
5.82549512e-01 -3.43728632e-01 -9.58236635e-01 3.59897092e-02
1.50574535e-01 -6.68645382e-01 -6.64967060e-01 9.10484672e-01
7.91254267e-02 6.83824658e-01 4.77387011e-01 -1.40406156e+00
-4.52095985e-01 5.34498155e-01 2.15599760e-02 -2.54055887e-01
5.21105289e-01 7.33453333e-01 6.23161316e-01 -4.09788638e-01
8.23062420e-01 1.46974921e+00 4.62044597e-01 7.50558317e-01
-1.09564531e+00 -3.48401278e-01 5.08675635e-01 -2.59387136e-01
-9.92503464e-01 -2.88039178e-01 3.18825632e-01 -5.57429731e-01
1.05752075e+00 -5.70791401e-02 5.91774583e-01 1.39768481e+00
5.26546180e-01 7.50185132e-01 8.13191533e-01 6.32485002e-02
4.49252099e-01 -3.39526266e-01 -3.39430422e-01 7.35294521e-01
-2.40837224e-03 4.09874618e-01 -2.59490907e-01 -3.43636274e-02
8.74419928e-01 2.76719928e-01 -3.03062409e-01 -5.35522282e-01
-1.70203996e+00 4.67665523e-01 4.02525514e-01 -1.14465199e-01
-3.90744120e-01 5.94402790e-01 4.89143431e-01 3.17259699e-01
-5.54684550e-02 8.56869936e-01 -4.92623836e-01 -3.91008079e-01
-5.14010727e-01 9.50798869e-01 9.42807376e-01 1.71674228e+00
6.07247770e-01 1.72116607e-01 -4.23532456e-01 1.03721112e-01
-1.71250626e-01 5.98054290e-01 3.43150407e-01 -1.49028480e+00
6.74281120e-01 3.17321748e-01 6.03450418e-01 -5.32329440e-01
-4.49357063e-01 -7.28622377e-02 -3.45385641e-01 4.78110105e-01
4.23887759e-01 -3.61519009e-01 -1.03660309e+00 1.72896028e+00
4.62659121e-01 1.05211981e-01 4.52802181e-01 1.04891598e+00
1.70107231e-01 9.03999805e-01 2.02447847e-01 1.20660953e-01
1.18094754e+00 -1.37304866e+00 -3.53589028e-01 -4.55392510e-01
9.72193241e-01 -4.28332150e-01 1.44018424e+00 7.04530180e-02
-1.07258344e+00 -5.92627943e-01 -8.57513547e-01 -1.17936254e-01
4.65601310e-03 5.36794811e-02 3.93413395e-01 -3.07367593e-01
-7.17788815e-01 1.07653606e+00 -1.16631532e+00 -3.57547760e-01
2.32001677e-01 2.69780993e-01 -4.20530945e-01 -1.61037102e-01
-7.65936613e-01 9.29162800e-01 8.93708587e-01 -1.66577354e-01
-1.37136400e+00 -7.69298613e-01 -1.03552973e+00 4.93006185e-02
7.05628932e-01 -7.99140036e-01 1.88697982e+00 -5.63709795e-01
-1.96359324e+00 8.18807781e-02 4.94768061e-02 -4.08645183e-01
5.23822129e-01 -4.62574154e-01 1.62854921e-02 5.07209785e-02
1.65274739e-01 8.17364991e-01 8.23988855e-01 -1.14269686e+00
-8.45158577e-01 -5.83991176e-03 6.26029909e-01 4.63568538e-01
1.24590114e-01 -4.37868327e-01 -2.63034761e-01 -4.93419975e-01
-7.65319318e-02 -1.52991033e+00 -4.56148475e-01 4.96329628e-02
-2.30487630e-01 -1.21270634e-01 9.21430290e-01 -5.86390674e-01
6.07312322e-01 -2.19232607e+00 6.95191920e-01 -3.28605831e-01
1.41258791e-01 8.15793797e-02 -2.97736287e-01 6.34410977e-01
2.94743627e-01 -2.71200687e-01 -3.43444496e-01 -3.02261114e-01
2.35185221e-01 4.91603702e-01 -8.54258418e-01 2.67197758e-01
1.45263299e-01 1.13468587e+00 -1.57944238e+00 -1.74846843e-01
1.08055070e-01 1.88121572e-02 -6.30891740e-01 4.09806758e-01
-1.01205385e+00 9.61065412e-01 -8.73956501e-01 2.18363956e-01
1.20339200e-01 -1.76871046e-02 1.07613333e-01 2.12492183e-01
-2.88143247e-01 4.28105950e-01 -6.54156983e-01 2.28215885e+00
-7.61738002e-01 4.50283617e-01 2.12440073e-01 -6.47013187e-01
6.74590528e-01 1.84144154e-01 3.46524239e-01 -3.68095040e-01
-1.61500409e-01 2.53962457e-01 -1.09682992e-01 -6.29052639e-01
8.84198368e-01 -2.89979130e-01 -6.55217767e-01 4.34580564e-01
-1.58471733e-01 -8.82045209e-01 5.46547472e-02 1.82339281e-01
1.12212801e+00 1.04647851e+00 3.01199704e-01 -1.41233861e-01
2.49231104e-02 7.19546020e-01 6.50613844e-01 8.16649556e-01
-2.02853709e-01 3.83739114e-01 4.80532348e-01 -4.99601603e-01
-1.35584700e+00 -1.16003048e+00 7.20839858e-01 9.27225053e-01
3.19990516e-01 -3.78824770e-01 -8.42242122e-01 -6.75677299e-01
5.50116934e-02 1.11410677e+00 -2.95559853e-01 -3.48320514e-01
-1.07518244e+00 2.31473655e-01 5.68602860e-01 5.62652767e-01
2.77784556e-01 -1.09953165e+00 -1.22128510e+00 3.37950647e-01
-1.36800274e-01 -1.35483432e+00 -8.46529365e-01 -1.40099987e-01
-8.25699627e-01 -8.29141557e-01 -5.93061268e-01 -6.87763155e-01
5.31774580e-01 3.23109925e-01 7.65450239e-01 -2.77706444e-01
-3.65151055e-02 5.18935263e-01 -3.90691727e-01 -3.99128199e-01
-8.05994868e-01 1.45308021e-02 2.30004430e-01 -5.31127214e-01
-3.57889056e-01 -5.17700016e-01 -4.30477560e-01 -9.13723954e-04
-7.66127527e-01 4.35013711e-01 5.77128708e-01 7.68794537e-01
4.80112582e-01 -6.90343902e-02 3.17613542e-01 -3.43446612e-01
6.73002124e-01 -4.85403836e-01 -7.38305509e-01 -2.50584027e-03
-1.75183550e-01 3.44904929e-01 1.12932837e+00 -7.99960971e-01
-1.20169175e+00 2.70252973e-01 2.94644088e-01 -7.26083934e-01
-3.03797811e-01 2.70443887e-01 1.74121946e-01 2.92382449e-01
6.90925002e-01 2.54626065e-01 7.43291229e-02 -3.13499898e-01
7.36808717e-01 3.47863078e-01 8.20933819e-01 -1.15056324e+00
8.06645989e-01 4.43017066e-01 9.48533565e-02 -5.55979908e-01
-7.97039151e-01 -1.22828811e-01 -5.05370796e-01 6.26808181e-02
9.73552406e-01 -8.99848104e-01 -8.48469138e-01 3.05749983e-01
-1.41464579e+00 -1.16747355e+00 -7.11145878e-01 6.48370445e-01
-1.23840892e+00 2.10518286e-01 -6.54752851e-01 -5.73323607e-01
-1.11887380e-01 -1.57244670e+00 1.36619174e+00 5.62731586e-02
-4.22721773e-01 -6.99148357e-01 -4.27735336e-02 -2.78716385e-01
1.66310281e-01 3.67933035e-01 9.80010808e-01 -3.28269720e-01
-7.87104487e-01 4.14236449e-02 6.07270300e-02 -3.41651998e-02
1.33413255e-01 -4.14477319e-01 -4.31601733e-01 -4.80060220e-01
-9.26439166e-02 -5.31455994e-01 3.44205886e-01 -6.89306259e-02
1.03188765e+00 -4.89047348e-01 -5.09006977e-01 5.86949944e-01
9.57055628e-01 2.34106690e-01 3.52821797e-01 2.73139387e-01
8.31676602e-01 6.31411612e-01 1.03100395e+00 2.22091392e-01
4.34269041e-01 7.63403058e-01 3.07844788e-01 4.26538169e-01
-1.86364010e-01 -7.15303838e-01 7.07161725e-01 4.27093327e-01
2.50586212e-01 -6.10702634e-02 -9.86732960e-01 5.67542195e-01
-2.19111753e+00 -9.12606657e-01 1.98186383e-01 1.99208117e+00
6.98454916e-01 1.85806528e-01 8.63151252e-02 -5.40031731e-01
2.11390942e-01 6.37143627e-02 -8.52166414e-01 -4.05801237e-01
5.00500262e-01 -1.67783767e-01 5.07969558e-01 5.66191316e-01
-8.09568048e-01 1.44169903e+00 5.74905634e+00 6.23472333e-01
-1.17422104e+00 2.58179475e-02 -4.13881354e-02 -3.57140899e-01
-9.45054591e-02 2.90869653e-01 -6.98070407e-01 2.67273337e-01
8.95172060e-01 -3.90938520e-01 8.40280950e-01 1.09069693e+00
3.91078532e-01 6.77928254e-02 -1.60749578e+00 5.94001293e-01
-4.47070152e-01 -1.28736472e+00 -1.18598482e-02 -4.86698486e-02
7.35681117e-01 9.36992094e-02 -8.87529105e-02 8.04843307e-01
7.13034391e-01 -9.62786317e-01 1.21001124e+00 3.73963624e-01
8.83222580e-01 -3.40720326e-01 3.66734304e-02 8.24313402e-01
-1.26277459e+00 -3.38985056e-01 -3.08143497e-01 -3.66054803e-01
5.80146253e-01 -6.60809164e-04 -9.78244245e-01 7.27622807e-01
4.77401644e-01 6.19361758e-01 2.55293846e-01 4.93671715e-01
-3.27139258e-01 -4.20626551e-02 -2.26267517e-01 -1.60806198e-02
6.53300941e-01 -2.96945632e-01 1.00811231e+00 8.50789964e-01
5.24210274e-01 2.97414541e-01 7.87451088e-01 1.08773160e+00
1.53420523e-01 -5.45086920e-01 -1.11810124e+00 -3.62711430e-01
4.11888212e-01 8.55533957e-01 -4.66247976e-01 -4.73320007e-01
-2.13074923e-01 1.20504606e+00 5.08316517e-01 6.22065127e-01
-9.33492005e-01 -3.01371723e-01 9.86774325e-01 -1.92389101e-01
1.86574891e-01 -8.50138545e-01 -1.35083467e-01 -1.03918123e+00
1.88438058e-01 -9.86599863e-01 -8.50462168e-02 -9.30303693e-01
-6.91908002e-01 4.69044745e-01 5.52465558e-01 -1.34439373e+00
-5.69612145e-01 -4.45292950e-01 -5.14397979e-01 8.69034469e-01
-1.24116802e+00 -1.08945096e+00 -2.90487528e-01 3.65594298e-01
9.76268351e-01 -2.05655564e-02 6.95771694e-01 -3.53265017e-01
-2.21212983e-01 8.29572007e-02 -1.84541685e-03 -2.41393313e-01
4.66309786e-01 -1.00328135e+00 9.56116617e-01 6.73839927e-01
-3.43966126e-01 7.60432065e-01 9.02892292e-01 -1.01945674e+00
-1.72137427e+00 -1.44376528e+00 2.70247042e-01 -5.53506434e-01
6.91266358e-01 -3.08199257e-01 -7.74539053e-01 1.03017962e+00
-1.05223276e-01 -2.01846421e-01 -2.53033906e-01 -3.37232113e-01
-2.98020095e-01 3.19690466e-01 -9.33567584e-01 1.28027439e+00
1.44557071e+00 -4.68380839e-01 -7.98629105e-01 6.41608596e-01
1.47281063e+00 -9.86218512e-01 -8.12463820e-01 3.13641638e-01
5.13396382e-01 -4.12427187e-01 8.88675272e-01 -9.41643417e-01
6.06073141e-01 -3.85943919e-01 -1.17538340e-01 -1.71165848e+00
-1.83770537e-01 -1.13431180e+00 -3.92346025e-01 5.24206758e-01
2.10829392e-01 -6.24493420e-01 5.78043401e-01 5.52302897e-01
-8.31054688e-01 -8.65841389e-01 -6.20019317e-01 -1.34665668e+00
2.82204658e-01 -2.98267514e-01 8.68861616e-01 4.26415384e-01
3.08971792e-01 2.17722565e-01 -2.29502216e-01 2.48242974e-01
3.34355652e-01 3.10934812e-01 1.10355747e+00 -6.43333495e-01
-4.76643860e-01 -3.53149176e-01 9.05558243e-02 -1.64005387e+00
6.16157472e-01 -7.80598879e-01 6.91349983e-01 -1.50759089e+00
-2.13669404e-01 -7.14575410e-01 3.61512691e-01 4.81019288e-01
-3.21418531e-02 -6.19336426e-01 6.25209272e-01 4.82289135e-01
-4.32688475e-01 9.46624875e-01 1.71013868e+00 -1.15570329e-01
-5.46528161e-01 -1.18444696e-01 -3.49036783e-01 7.86889672e-01
8.69454861e-01 -5.09344935e-01 -9.11167920e-01 -7.22151041e-01
-2.12552220e-01 5.07254243e-01 4.39823389e-01 -9.55800474e-01
1.22031733e-01 -8.33484411e-01 -1.96080491e-01 -2.52723724e-01
6.92534626e-01 -6.55050993e-01 4.51332852e-02 6.94733322e-01
-3.95262778e-01 3.19516450e-01 4.27047908e-01 9.23059225e-01
2.11122796e-01 -1.49348632e-01 5.48478544e-01 -4.32134032e-01
-8.50287914e-01 4.22964871e-01 -2.93166667e-01 8.42929333e-02
1.46321774e+00 -3.10457814e-02 -4.10261035e-01 -3.37562978e-01
-5.78991830e-01 4.80220556e-01 8.98304999e-01 7.31909335e-01
5.11745512e-01 -1.05139959e+00 -3.78360361e-01 -1.17125541e-01
1.18608035e-01 6.57769442e-01 8.29603598e-02 6.78835511e-01
-5.49120784e-01 5.71833014e-01 -2.18490794e-01 -5.41880786e-01
-6.51179075e-01 8.03926826e-01 2.17204466e-01 -1.90076590e-01
-1.32227170e+00 5.44487774e-01 5.83374500e-01 -8.35106611e-01
1.38033673e-01 -6.19086146e-01 3.54970127e-01 -8.32066000e-01
3.92714083e-01 3.72539878e-01 -4.62650985e-01 -2.65712708e-01
-6.38012066e-02 2.66277909e-01 1.66776329e-02 -4.21660095e-01
1.08161080e+00 -2.65159365e-02 1.01307362e-01 4.93584007e-01
9.31231976e-01 -3.25635523e-01 -2.03180766e+00 -1.67046506e-02
-3.69900130e-02 -2.69482076e-01 -4.19787556e-01 -4.67546284e-01
-3.29333901e-01 7.37928808e-01 -1.42393455e-01 -2.73823261e-01
4.94663775e-01 -7.27060810e-02 1.28687322e+00 9.79471207e-01
8.72579992e-01 -1.03473854e+00 3.32389086e-01 9.01689649e-01
1.12683034e+00 -9.16603804e-01 -2.19696492e-01 -3.65724176e-01
-8.75261188e-01 1.03293145e+00 7.80233443e-01 -9.93684232e-02
-1.56091936e-02 2.45348662e-01 -2.99653292e-01 1.25563843e-03
-7.54412532e-01 4.06295322e-02 -3.80497053e-02 7.74690270e-01
-1.78465024e-01 2.12710932e-01 1.82171062e-01 3.91579241e-01
-3.06718379e-01 1.51861817e-01 8.48778069e-01 1.41031003e+00
-4.73713279e-01 -7.81780064e-01 -1.97378233e-01 7.20554292e-02
2.02130809e-01 2.31858000e-01 -7.80505463e-02 9.75287437e-01
-2.63673693e-01 7.29788303e-01 2.30674930e-02 -4.15797651e-01
5.11725843e-01 5.79062589e-02 6.56131148e-01 -1.09300077e+00
-2.86126882e-01 -4.43946600e-01 2.19145194e-01 -9.72830772e-01
2.22153395e-01 -6.77211881e-01 -1.88204718e+00 -3.09430987e-01
3.69934916e-01 4.60203923e-02 6.04370892e-01 1.17494416e+00
6.85044110e-01 7.39605069e-01 1.45182922e-01 -1.40447950e+00
-1.25339615e+00 -6.73637867e-01 -5.90165332e-02 5.65350831e-01
7.37828016e-01 -6.52414143e-01 1.64994404e-01 1.57423005e-01]
|
[4.540831565856934, 0.81562739610672]
|
adb4b8c2-d7a0-4940-a17d-8ec03721f063
|
a-probabilistic-hard-attention-model-for
|
2111.07534
| null |
https://arxiv.org/abs/2111.07534v1
|
https://arxiv.org/pdf/2111.07534v1.pdf
|
A Probabilistic Hard Attention Model For Sequentially Observed Scenes
|
A visual hard attention model actively selects and observes a sequence of subregions in an image to make a prediction. The majority of hard attention models determine the attention-worthy regions by first analyzing a complete image. However, it may be the case that the entire image is not available initially but instead sensed gradually through a series of partial observations. In this paper, we design an efficient hard attention model for classifying such sequentially observed scenes. The presented model never observes an image completely. To select informative regions under partial observability, the model uses Bayesian Optimal Experiment Design. First, it synthesizes the features of the unobserved regions based on the already observed regions. Then, it uses the predicted features to estimate the expected information gain (EIG) attained, should various regions be attended. Finally, the model attends to the actual content on the location where the EIG mentioned above is maximum. The model uses a) a recurrent feature aggregator to maintain a recurrent state, b) a linear classifier to predict the class label, c) a Partial variational autoencoder to predict the features of unobserved regions. We use normalizing flows in Partial VAE to handle multi-modality in the feature-synthesis problem. We train our model using a differentiable objective and test it on five datasets. Our model gains 2-10% higher accuracy than the baseline models when both have seen only a couple of glimpses.
|
['James J. Clark', 'Samrudhdhi B. Rangrej']
|
2021-11-15
| null | null | null | null |
['hard-attention']
|
['methodology']
|
[ 3.01466048e-01 4.18262720e-01 -2.76645929e-01 -3.08343470e-01
-9.01230514e-01 -3.32188547e-01 5.74432790e-01 -3.03490251e-01
-3.21677566e-01 6.26652360e-01 3.00266892e-01 -1.18335672e-02
-4.12937254e-02 -5.29050648e-01 -8.98495257e-01 -1.07556617e+00
1.27759114e-01 3.86458158e-01 7.68456161e-02 2.07973883e-01
2.12467372e-01 3.64915431e-01 -1.68430674e+00 4.22091395e-01
5.98822713e-01 1.17032194e+00 8.75733435e-01 7.55018950e-01
1.37257919e-01 1.12191260e+00 -3.55301827e-01 4.18368913e-02
4.01454985e-01 -6.21894956e-01 -8.37322414e-01 6.68035507e-01
2.04441190e-01 -5.92209876e-01 -3.15051883e-01 1.08266997e+00
1.40037730e-01 4.11889255e-01 7.76503503e-01 -1.12697935e+00
-6.05705023e-01 5.76645374e-01 -5.45891166e-01 4.93892878e-01
1.45559520e-01 3.54907185e-01 1.17466056e+00 -9.56949770e-01
7.12369502e-01 1.11732686e+00 -2.67692655e-01 4.12947834e-01
-1.25580406e+00 -2.35429391e-01 7.40033209e-01 4.68275398e-01
-1.20048642e+00 -4.53963965e-01 9.03075397e-01 -4.25338268e-01
8.35041761e-01 2.41753176e-01 6.39052272e-01 1.21611655e+00
3.93494338e-01 9.36466455e-01 9.84643519e-01 -3.38509530e-01
3.96455795e-01 4.39842910e-01 9.73934308e-02 8.12943220e-01
-4.62411463e-01 2.36604154e-01 -4.50074226e-01 9.32806451e-03
5.90973020e-01 3.01991463e-01 -4.04198498e-01 -1.39907762e-01
-1.23762906e+00 8.05390537e-01 7.20577240e-01 1.61677271e-01
-8.72049034e-01 -1.62913483e-02 -1.94182158e-01 4.41535592e-01
4.18480188e-01 5.02953529e-01 -3.83515269e-01 1.86481178e-01
-8.61319005e-01 -6.60870746e-02 3.29187602e-01 8.04082632e-01
9.59882617e-01 4.96718958e-02 -4.76948678e-01 6.01445138e-01
3.38654160e-01 2.40265146e-01 3.48740131e-01 -1.28120112e+00
2.90844232e-01 4.24367666e-01 1.86787307e-01 -8.34334016e-01
-8.89831632e-02 -5.52627563e-01 -7.16543853e-01 3.81356239e-01
2.63255417e-01 -1.53771237e-01 -1.18854964e+00 1.79486680e+00
2.32525617e-01 2.90385801e-02 -1.15234785e-01 1.17474413e+00
3.06956530e-01 9.81723249e-01 -4.08541374e-02 -3.46448004e-01
1.22566795e+00 -1.00885499e+00 -6.63951278e-01 -3.93191099e-01
1.06411974e-03 -4.89693046e-01 9.31951284e-01 4.73455131e-01
-1.23389089e+00 -7.87072539e-01 -7.76291907e-01 -1.06210351e-01
-1.96977243e-01 4.36330169e-01 1.57175198e-01 -7.61692524e-02
-1.07765985e+00 4.57397342e-01 -9.45276976e-01 -2.02424437e-01
4.74635452e-01 2.07139194e-01 -1.58831194e-01 -4.85944226e-02
-8.31860363e-01 7.49517798e-01 1.49132773e-01 1.50078416e-01
-1.65282822e+00 -4.34891313e-01 -8.91095459e-01 4.97815579e-01
4.88841295e-01 -6.92428708e-01 9.30312514e-01 -1.47460079e+00
-1.47646320e+00 6.27461791e-01 -5.47344565e-01 -3.28519493e-01
4.26677912e-01 2.10419912e-02 -5.61291911e-02 2.77999192e-01
1.82925701e-01 1.01087356e+00 1.15322280e+00 -1.32071304e+00
-8.82160008e-01 -2.47592345e-01 1.68653637e-01 4.41806972e-01
-3.24321501e-02 -1.21623322e-01 -4.86414045e-01 -4.10233259e-01
1.24728158e-01 -8.34687352e-01 -3.24669093e-01 3.32892016e-02
-5.77157438e-01 -6.42978698e-02 7.63469636e-01 -5.79364657e-01
7.46092200e-01 -2.39196157e+00 4.91443932e-01 6.30921498e-02
3.23944002e-01 -3.56178343e-01 -2.02298671e-01 1.00369446e-01
5.89924492e-02 -1.48133337e-01 -1.42209366e-01 -5.92045546e-01
-3.10975611e-01 2.18423769e-01 -6.29145622e-01 5.68686008e-01
3.90046060e-01 8.55353117e-01 -6.73755050e-01 -5.15738010e-01
4.07295078e-01 2.82089502e-01 -6.18139446e-01 5.59727788e-01
-3.65983129e-01 5.10799587e-01 -5.83000481e-01 5.85770011e-01
3.14228952e-01 -7.28116810e-01 -1.97808463e-02 3.57889906e-02
-1.79294333e-01 -3.74850519e-02 -1.13399422e+00 1.24672878e+00
-2.47103348e-01 8.65938365e-01 -4.62770052e-02 -1.08197916e+00
5.28259635e-01 2.53731549e-01 3.90544057e-01 -4.11891401e-01
2.40839526e-01 -2.73272276e-01 -3.36481482e-02 -6.97960198e-01
3.40675831e-01 1.51316524e-01 1.51406914e-01 3.93060595e-01
2.08768070e-01 3.66490036e-01 -1.25834733e-01 2.80719131e-01
9.34600472e-01 3.87958027e-02 3.24825644e-01 -1.68979004e-01
3.80801171e-01 -4.50442806e-02 5.10916352e-01 9.39476192e-01
-3.55985582e-01 8.79953980e-01 6.33448780e-01 -3.76682132e-01
-9.67340410e-01 -1.18579733e+00 8.58031586e-02 1.26329565e+00
3.65186393e-01 2.44687974e-01 -5.83173573e-01 -7.13192761e-01
-3.59010935e-01 7.60906577e-01 -1.05008233e+00 -2.73071051e-01
-8.02335441e-02 -3.91165406e-01 -3.22889686e-01 4.24640656e-01
5.54432809e-01 -1.57131267e+00 -8.80925596e-01 -1.60745252e-02
-4.26437169e-01 -8.58678043e-01 -5.14964223e-01 5.30799747e-01
-4.92589146e-01 -8.09601367e-01 -5.75032890e-01 -8.08678448e-01
8.86105895e-01 1.75638616e-01 7.37282455e-01 -8.12657401e-02
-5.32841422e-02 2.43506297e-01 -1.75608695e-01 -2.12795570e-01
-5.88112921e-02 8.29007402e-02 -1.45763814e-01 5.94077587e-01
5.01499102e-02 -2.37373024e-01 -6.12713218e-01 1.58635974e-01
-5.73201597e-01 6.96602389e-02 7.10509241e-01 1.01876950e+00
8.77851248e-01 -4.23803963e-02 3.49166542e-01 -5.30705631e-01
1.61035508e-01 -7.36753166e-01 -6.58917964e-01 2.98138380e-01
-3.00497502e-01 3.06298673e-01 5.14203489e-01 -6.89355373e-01
-1.30887306e+00 2.52733648e-01 2.04874650e-01 -7.72644460e-01
-3.50602925e-01 3.90094072e-01 -1.84663519e-01 4.33706075e-01
3.65708500e-01 4.58999842e-01 -1.02117412e-01 -1.27254829e-01
1.79878831e-01 5.24827361e-01 2.93469846e-01 -2.36148164e-01
4.52363580e-01 5.95684350e-01 -3.55584681e-01 -8.25988173e-01
-1.12445021e+00 -2.99052060e-01 -5.25563538e-01 -4.51668024e-01
1.22854066e+00 -9.07006741e-01 -6.71369076e-01 1.48911625e-01
-9.24237669e-01 -5.84766328e-01 -5.27692199e-01 6.22476995e-01
-6.06299698e-01 -7.88242668e-02 -3.95640671e-01 -9.72442746e-01
5.74363098e-02 -1.44435394e+00 1.18597686e+00 3.35220397e-01
-5.37271202e-02 -7.55076468e-01 -2.00896382e-01 1.74510807e-01
1.41480803e-01 -5.78205753e-03 6.18735373e-01 -3.89058590e-01
-1.14159191e+00 5.71225844e-02 -1.07431844e-01 2.45656684e-01
-1.58059411e-02 3.64868678e-02 -1.22330117e+00 -2.21063018e-01
1.01099469e-01 -3.92267108e-01 1.08625090e+00 9.44199324e-01
1.48334599e+00 -3.42088431e-01 -4.41047817e-01 6.04341924e-01
1.24234521e+00 4.01687711e-01 4.99367714e-01 1.22429959e-01
3.15433681e-01 6.26055479e-01 5.23922384e-01 4.65427965e-01
3.67919862e-01 4.09021586e-01 8.67179513e-01 -1.37803823e-01
6.74403533e-02 -1.96495980e-01 6.09236419e-01 4.07847852e-01
6.91341385e-02 -5.92561901e-01 -4.42723095e-01 8.10915172e-01
-1.74747884e+00 -1.15020835e+00 3.26088250e-01 1.99175322e+00
4.10838753e-01 8.48481134e-02 -2.09832042e-01 -2.89922982e-01
5.98511994e-01 3.13034862e-01 -8.35197508e-01 -1.87492311e-01
-7.57577643e-02 -3.37424725e-01 2.27405623e-01 8.06178033e-01
-1.00331736e+00 9.36902761e-01 6.05320072e+00 3.94564986e-01
-1.08816206e+00 4.24160026e-02 1.03822136e+00 -3.95286411e-01
-2.09113806e-01 2.43841186e-01 -7.75191903e-01 5.54081082e-01
6.34058356e-01 1.33990929e-01 7.47956634e-01 7.81768262e-01
2.38649130e-01 -4.71664160e-01 -1.06287360e+00 7.23212898e-01
2.37176627e-01 -1.08044851e+00 1.02424640e-02 4.29679960e-01
7.76130676e-01 2.22656310e-01 3.01210254e-01 2.55151302e-01
5.02909124e-01 -9.05872464e-01 9.31070387e-01 9.09882724e-01
4.46562678e-01 -5.66442132e-01 4.85730082e-01 6.30715430e-01
-7.75328219e-01 -4.49571520e-01 -4.16994035e-01 -1.30549103e-01
1.30287409e-01 3.67479384e-01 -6.61553383e-01 1.71934724e-01
7.25277901e-01 5.96533895e-01 -5.48102915e-01 8.28869820e-01
-2.80819476e-01 6.19296551e-01 -2.39408210e-01 4.19331677e-02
4.83607143e-01 -1.13086604e-01 6.17158353e-01 6.66606307e-01
3.39143008e-01 3.55529010e-01 3.45747560e-01 1.08669591e+00
9.40594450e-02 -3.07102531e-01 -5.20342886e-01 4.01086926e-01
1.74348027e-01 1.17306912e+00 -6.70963466e-01 -5.82982242e-01
-3.07073563e-01 1.16623259e+00 5.54594219e-01 8.41965199e-01
-8.46989334e-01 -1.38623804e-01 5.50869763e-01 -1.00311585e-01
5.18568575e-01 1.95271641e-01 -4.76088859e-02 -1.23825538e+00
-2.06024855e-01 -5.47491133e-01 5.40770113e-01 -1.37239122e+00
-9.88775790e-01 8.06407869e-01 -3.56365554e-02 -1.08284903e+00
-3.78602177e-01 -2.89101452e-01 -5.63185096e-01 9.18487191e-01
-1.33276165e+00 -7.34681845e-01 -3.17585886e-01 5.76987088e-01
1.04270029e+00 -1.09014861e-01 3.78355622e-01 -1.28724948e-01
-6.38251543e-01 2.32184440e-01 -4.12774868e-02 7.21050799e-02
4.04199868e-01 -1.30823469e+00 -1.57866716e-01 1.00112426e+00
3.38443071e-01 2.13056222e-01 7.92004526e-01 -4.70181972e-01
-8.12704444e-01 -1.03252017e+00 8.27381611e-01 -4.94738936e-01
3.34876537e-01 -3.89138371e-01 -9.29878712e-01 1.16472626e+00
5.02755523e-01 2.53181934e-01 8.69596452e-02 -2.79970825e-01
1.38764739e-01 -9.06913802e-02 -1.02575183e+00 6.36374295e-01
6.80417776e-01 -4.37186122e-01 -5.28230309e-01 2.01800793e-01
7.31974125e-01 -2.33520404e-01 -4.71750885e-01 -3.89259160e-02
4.46522355e-01 -8.32965374e-01 5.77290893e-01 -5.95750093e-01
6.00142419e-01 -2.63695359e-01 -2.28465170e-01 -1.44884276e+00
-6.69880986e-01 -3.40716779e-01 -1.08438507e-01 9.54641283e-01
6.53060138e-01 -5.74597716e-01 5.90619624e-01 5.36512911e-01
-4.52167392e-02 -7.06274152e-01 -8.15160275e-01 -2.80700684e-01
-4.24974144e-01 -2.29839340e-01 3.05005729e-01 8.03560972e-01
-2.12363005e-01 2.77881533e-01 -5.39699316e-01 5.68566680e-01
7.19129145e-01 4.52798545e-01 4.54081655e-01 -9.14334893e-01
-3.90803188e-01 -2.66493648e-01 -6.74915388e-02 -1.30890107e+00
2.28239462e-01 -7.08333552e-01 2.83873647e-01 -1.33356428e+00
5.05571485e-01 1.32650688e-01 -6.21923327e-01 5.22846282e-01
-2.00681016e-01 -5.30854464e-02 2.10955441e-01 8.93860236e-02
-7.34089375e-01 6.76331818e-01 1.35795271e+00 -2.42514655e-01
-3.57946217e-01 1.28006697e-01 -7.83628166e-01 5.51113963e-01
6.73179567e-01 -5.81981182e-01 -5.31264305e-01 -4.23799127e-01
-4.69452888e-02 5.16376615e-01 5.67248702e-01 -7.18165874e-01
3.02478254e-01 -6.06008589e-01 7.96330810e-01 -7.01695561e-01
4.47678626e-01 -9.16689038e-01 1.38032744e-02 2.54118711e-01
-7.67364323e-01 -1.05862647e-01 -2.41931677e-01 7.12493420e-01
-1.77581698e-01 -1.20652616e-01 8.55763912e-01 -2.47914627e-01
-7.33838022e-01 4.63951617e-01 -6.41609848e-01 -2.10124299e-01
1.13934863e+00 -1.45129412e-01 -3.16905119e-02 -6.71599269e-01
-1.18359137e+00 4.72714871e-01 3.42935413e-01 4.29229736e-01
7.54582465e-01 -1.07746649e+00 -6.08817399e-01 5.22244155e-01
1.75125785e-02 -5.81097007e-02 4.94299710e-01 7.02216685e-01
1.26078740e-01 1.60012200e-01 -1.37209952e-01 -8.23853374e-01
-1.03350878e+00 8.44326019e-01 4.78316396e-01 -1.79315954e-01
-5.42514682e-01 7.01420784e-01 6.90520942e-01 4.97576632e-02
3.16869110e-01 -3.97201806e-01 -4.03596997e-01 1.59534112e-01
5.74132264e-01 1.31516308e-02 -3.67448866e-01 -6.37989521e-01
-1.28548741e-01 3.01612705e-01 -5.32375872e-02 -4.15588200e-01
1.48913312e+00 -5.95274806e-01 1.20232828e-01 6.24470711e-01
1.35234892e+00 -3.22651029e-01 -2.03492951e+00 -2.32225806e-01
-4.68246639e-01 -4.29227501e-01 3.43469709e-01 -8.02322567e-01
-1.25027454e+00 8.48846138e-01 5.01935065e-01 2.69839048e-01
1.28973985e+00 4.37492013e-01 2.14684084e-01 2.03747377e-01
8.78048390e-02 -8.66926968e-01 1.63229987e-01 3.25358301e-01
9.98401284e-01 -1.41683495e+00 -1.93637341e-01 1.08033016e-01
-1.16647685e+00 7.33708739e-01 7.16001749e-01 -2.07405999e-01
7.96968102e-01 5.29255122e-02 -8.76975879e-02 -3.18269491e-01
-1.36016035e+00 -3.38605940e-01 3.57424617e-01 3.46967369e-01
-1.72931254e-01 -3.27787781e-03 5.11272252e-01 2.19455346e-01
1.74662042e-02 -1.69177994e-01 5.53242147e-01 5.52959621e-01
-6.68078303e-01 -3.61717314e-01 -3.23338509e-01 6.32525563e-01
-4.22394454e-01 -1.02683427e-02 -1.50592148e-01 5.20870626e-01
-2.43430045e-02 9.29752886e-01 4.91356373e-01 -1.90236360e-01
4.70673703e-02 1.19060476e-03 2.01638222e-01 -5.16822219e-01
-2.07879946e-01 2.41694674e-01 -4.29595381e-01 -7.04897225e-01
-2.52416790e-01 -9.19045806e-01 -9.11587536e-01 2.34675199e-01
-3.11631918e-01 -1.06077485e-01 3.66000503e-01 1.07536185e+00
3.94746661e-01 7.28423119e-01 1.01554382e+00 -9.67331767e-01
-4.92133856e-01 -9.44067538e-01 -6.62976444e-01 2.75253683e-01
9.30681586e-01 -7.24906266e-01 -6.31198049e-01 3.75746816e-01]
|
[9.51644515991211, 0.26236751675605774]
|
7f34dfc8-df7b-4af4-b34c-b45ac3882f48
|
accelerating-system-level-debug-using-rule
|
2207.00622
| null |
https://arxiv.org/abs/2207.00622v1
|
https://arxiv.org/pdf/2207.00622v1.pdf
|
Accelerating System-Level Debug Using Rule Learning and Subgroup Discovery Techniques
|
We propose a root-causing procedure for accelerating system-level debug using rule-based techniques. We describe the procedure and how it provides high quality debug hints for reducing the debug effort. This includes the heuristics for engineering features from logs of many tests, and the data analytics techniques for generating powerful debug hints. As a case study, we used these techniques for root-causing failures of the Power Management (PM) design feature Package-C8 and showed their effectiveness. Furthermore, we propose an approach for mining the root-causing experience and results for reuse, to accelerate future debug activities and reduce dependency on validation experts. We believe that these techniques are beneficial also for other validation activities at different levels of abstraction, for complex hardware, software and firmware systems, both pre-silicon and post-silicon.
|
['Zurab Khasidashvili']
|
2022-07-02
| null | null | null | null |
['subgroup-discovery']
|
['methodology']
|
[-4.23581690e-01 1.89933553e-01 -4.22438622e-01 -3.15483928e-01
-5.90103030e-01 -5.24621487e-01 -1.72946274e-01 2.78237402e-01
4.65564549e-01 7.45594680e-01 -2.88940430e-01 -8.57211649e-01
-3.36082995e-01 -5.82320631e-01 -5.10868430e-01 1.36704758e-01
-4.80194479e-01 3.24305862e-01 5.22636712e-01 -2.31288180e-01
7.12963223e-01 5.99255383e-01 -1.67653966e+00 2.16487974e-01
8.26286197e-01 5.41161716e-01 6.84403330e-02 6.15992665e-01
2.19095349e-01 8.26141894e-01 -1.05691111e+00 -3.15960869e-02
-4.12603132e-02 -5.60545921e-01 -5.74213266e-01 -3.73005942e-02
-3.24775368e-01 -2.45127782e-01 2.63465166e-01 1.06266844e+00
2.06964374e-01 -7.29796827e-01 1.10835768e-01 -1.41814852e+00
4.64583516e-01 1.08214223e+00 -6.81613863e-01 1.49668798e-01
7.85738230e-01 1.86003000e-01 6.40980959e-01 -4.62846160e-01
6.53007865e-01 8.54400635e-01 5.40984333e-01 2.82240748e-01
-1.13005912e+00 -7.20020533e-01 -3.02491605e-01 1.80753574e-01
-1.42859840e+00 -2.29942039e-01 7.00232565e-01 -6.38132274e-01
1.85017598e+00 2.02064350e-01 6.06144130e-01 1.08552599e+00
7.60428429e-01 1.69141114e-01 1.05550897e+00 -8.06012750e-01
8.64945412e-01 1.62445113e-01 8.92913699e-01 9.60989296e-01
9.45068002e-01 4.00420308e-01 -5.06865203e-01 -8.48688006e-01
6.14945292e-01 -2.43121743e-01 6.58359658e-03 8.64046961e-02
-6.58964396e-01 3.82113218e-01 -1.40521556e-01 6.28055632e-01
-1.19399175e-01 5.21733582e-01 5.85192859e-01 4.97455180e-01
-2.19012365e-01 8.17795157e-01 -9.24988747e-01 -6.60369515e-01
-1.31179118e+00 -4.39933792e-04 1.18509912e+00 1.19697344e+00
8.60511303e-01 3.90981853e-01 1.11712374e-01 -3.83515991e-02
4.83024538e-01 1.29516512e-01 1.56608403e-01 -5.99471390e-01
8.12854990e-02 9.79509056e-01 8.81443173e-02 -5.37467182e-01
-6.89183533e-01 -4.34067726e-01 -2.71049678e-01 7.54293740e-01
-3.47053856e-01 -3.79210383e-01 -9.30240214e-01 1.30882549e+00
-1.25915393e-01 1.43538743e-01 -4.70804930e-01 3.61812115e-01
3.93259525e-03 5.32187939e-01 -2.34109387e-01 -4.11779016e-01
1.32358277e+00 -4.20400918e-01 -7.12006748e-01 -6.54662624e-02
1.02955651e+00 -6.55187249e-01 8.31919193e-01 1.19894302e+00
-1.00685704e+00 -7.29053676e-01 -1.88906443e+00 7.82545865e-01
1.31827663e-03 4.42868561e-01 5.85369229e-01 1.08982742e+00
-1.25255167e+00 1.18559802e+00 -1.27314866e+00 -2.58240581e-01
-1.17396601e-02 6.57486975e-01 2.70957559e-01 3.51304412e-01
-5.22161126e-01 7.17588842e-01 3.53314042e-01 -3.19033861e-01
-1.33696806e+00 -9.94607806e-01 -6.10406876e-01 3.70176107e-01
4.67717737e-01 -5.06417632e-01 1.16430795e+00 -5.35703421e-01
-1.29999197e+00 3.35904717e-01 1.37660772e-01 -1.53550908e-01
-2.63326883e-01 -3.39099407e-01 -9.65320170e-01 -5.54645658e-01
1.50765693e-02 -1.51549399e-01 7.76902020e-01 -8.61963153e-01
-6.39959991e-01 -9.43438932e-02 -1.13583863e-01 -1.20826054e+00
4.26761322e-02 6.99192658e-02 1.03430547e-01 -9.05561894e-02
-3.59825045e-01 -4.86880064e-01 -4.68697071e-01 -9.55447435e-01
-6.15451872e-01 2.36453414e-01 7.08508313e-01 -5.65249503e-01
1.74304652e+00 -2.13059807e+00 -5.50967790e-02 9.51931417e-01
1.16033457e-01 -1.38260707e-01 3.79047304e-01 5.94115019e-01
-2.54009873e-01 1.85668126e-01 3.71397734e-01 1.87517196e-01
1.76606420e-02 1.43549396e-02 -3.07292759e-01 1.73670635e-01
4.08532977e-01 4.62845266e-01 -7.24876881e-01 4.13783230e-02
1.47571728e-01 -2.17721745e-01 -5.16582787e-01 1.15482435e-01
-4.17458296e-01 -8.89271832e-05 -6.08276606e-01 1.10048330e+00
4.11844224e-01 -2.34660059e-01 6.48681879e-01 -6.12848252e-02
-3.52863252e-01 3.07223946e-01 -1.24027503e+00 1.38845468e+00
-6.86177373e-01 6.31486535e-01 -7.05702528e-02 -2.81046838e-01
1.24501169e+00 6.80724755e-02 4.48904373e-02 -6.63258135e-01
4.07969356e-01 6.38751209e-01 1.01307504e-01 -3.74151289e-01
4.41787481e-01 3.59480977e-01 -5.49467504e-01 6.17382705e-01
5.72573878e-02 2.20375195e-01 3.39090735e-01 5.20064458e-02
1.97346604e+00 1.30446628e-01 5.33657014e-01 -8.99031103e-01
4.40967500e-01 3.18827271e-01 8.00843060e-01 3.09737951e-01
3.15458387e-01 -5.68546131e-02 1.11160767e+00 -3.81923355e-02
-9.87388492e-01 -7.73600280e-01 8.03872272e-02 2.90073067e-01
-2.55829036e-01 -1.38804245e+00 -8.30284595e-01 -7.13378787e-01
-1.16441041e-01 1.06252241e+00 -3.69079918e-01 -3.15633029e-01
-3.45242053e-01 -3.63705188e-01 4.46050376e-01 8.04981947e-01
-2.08299756e-01 -7.82872140e-01 -1.21479082e+00 5.03884256e-01
9.09271240e-01 -6.71155035e-01 1.22703589e-01 1.10087264e+00
-1.28155136e+00 -1.14193702e+00 3.80370915e-01 -3.61447126e-01
8.24138045e-01 -3.91670346e-01 1.60651803e+00 7.68740594e-01
-1.08841228e+00 2.07128331e-01 -5.93505085e-01 -9.54273716e-02
-8.13935459e-01 -4.12287474e-01 -9.62527618e-02 -8.09712708e-01
1.71600953e-01 -7.56367862e-01 -1.69459730e-01 5.63613355e-01
-3.13721508e-01 -5.20744801e-01 6.35562122e-01 7.78548539e-01
3.93025964e-01 5.67750812e-01 3.13111216e-01 -1.12319112e+00
5.65855443e-01 -4.25499380e-01 -1.31744945e+00 1.70946613e-01
-9.83540118e-01 4.07777160e-01 7.23114848e-01 -1.59784526e-01
-6.76849902e-01 2.14073226e-01 -2.98406124e-01 -2.52967656e-01
7.59216323e-02 5.05237043e-01 -4.79486585e-01 -3.33675951e-01
1.12890816e+00 -4.66923565e-01 -5.03175437e-01 -5.75632632e-01
-2.69616872e-01 1.66957080e-01 1.18566215e-01 -7.62856662e-01
6.03747606e-01 -1.54540315e-01 2.98806518e-01 -2.75538594e-01
5.10524921e-02 -5.09627201e-02 -2.44962096e-01 -2.08509699e-01
2.89140821e-01 -5.75218320e-01 -8.03274810e-01 -2.94617325e-01
-1.07793093e+00 -3.49846900e-01 -2.42494896e-01 -6.89036921e-02
-3.39735001e-01 1.10617705e-01 -4.10035402e-01 -8.92346680e-01
-1.39643058e-01 -1.39628601e+00 9.79111850e-01 4.65676486e-01
-9.97632623e-01 -8.45785260e-01 2.61345536e-01 -5.31344831e-01
3.72477204e-01 4.77351069e-01 1.30888128e+00 -6.36809766e-01
-9.79634225e-01 -1.66823685e-01 3.91567916e-01 1.84363797e-01
2.60922518e-02 6.09404802e-01 -8.69052649e-01 -3.21615964e-01
9.75753888e-02 6.93754703e-02 1.26889214e-01 2.69592673e-01
1.02178872e+00 2.07938537e-01 -9.01831567e-01 2.52674907e-01
1.76998031e+00 5.02481937e-01 8.35860252e-01 -2.75728554e-02
2.05737367e-01 1.83803558e-01 1.22059131e+00 1.03219640e+00
-4.13126618e-01 7.20061302e-01 4.85267460e-01 6.76084980e-02
3.61651421e-01 -1.21299393e-01 9.30947006e-01 6.32893324e-01
2.54092783e-01 1.52466223e-01 -1.04040754e+00 4.30353343e-01
-1.57011521e+00 -4.25864130e-01 -5.26204884e-01 2.27778125e+00
5.62122524e-01 8.70485485e-01 2.77261883e-01 4.88905638e-01
8.00221935e-02 -7.12327182e-01 -3.24405760e-01 -9.03896213e-01
6.55696213e-01 1.04875171e+00 5.38537085e-01 5.31656861e-01
-2.07163051e-01 5.47967315e-01 7.04538965e+00 9.96563554e-01
-8.93369019e-01 1.21622652e-01 2.87583917e-01 1.55104503e-01
-3.23214203e-01 7.07192302e-01 -1.29523098e+00 2.69090474e-01
1.67314172e+00 -4.14131075e-01 -5.77136800e-02 1.49325943e+00
1.49294436e-01 -6.27016842e-01 -1.82279539e+00 7.19145775e-01
-4.31314915e-01 -1.46040726e+00 -7.73497164e-01 1.84730470e-01
5.48936605e-01 -4.86959100e-01 -5.01915634e-01 4.74672675e-01
3.35004985e-01 -8.59004438e-01 6.49533212e-01 2.17099339e-01
5.91433346e-01 -1.20564651e+00 8.98496151e-01 5.83886616e-02
-9.09308553e-01 -1.11491427e-01 -1.74781397e-01 -3.18551451e-01
2.14364037e-01 1.27519965e+00 -1.22474146e+00 6.68568373e-01
6.49903953e-01 3.35635513e-01 -8.54077637e-01 8.86098206e-01
-2.95112997e-01 7.05173433e-01 -3.34915578e-01 -4.36959296e-01
-4.61162239e-01 2.91864872e-01 2.12247252e-01 1.16046011e+00
6.59770846e-01 -6.13242149e-01 -2.85746790e-02 1.15502954e+00
4.81516331e-01 -4.55469400e-01 -5.05404711e-01 -1.31588832e-01
7.55701005e-01 1.53420103e+00 -1.16323292e+00 -1.58400491e-01
4.35709208e-03 5.89416027e-01 -2.58317709e-01 -1.55707285e-01
-1.10886812e+00 -6.19359493e-01 7.30110824e-01 5.53663492e-01
3.71110260e-01 -3.90533864e-01 -7.53366709e-01 -4.46856618e-01
2.36057699e-01 -8.96087170e-01 1.45301849e-01 -4.95788306e-01
-8.29061449e-01 5.88484585e-01 1.69184223e-01 -1.15872264e+00
-7.12881505e-01 -7.61262476e-01 -1.12228334e+00 6.38287842e-01
-9.00449514e-01 -2.61230528e-01 -2.05149785e-01 5.88309541e-02
2.74337858e-01 -1.96530506e-01 7.67352819e-01 3.14592242e-01
-4.62666512e-01 8.64577115e-01 -2.75483280e-01 -8.57807398e-01
2.49575719e-01 -1.08554184e+00 5.70294857e-01 9.43217814e-01
-3.01285505e-01 8.89279783e-01 1.01462269e+00 -1.21647680e+00
-1.98513460e+00 -7.57220864e-01 5.10983586e-01 -2.05280542e-01
1.02231455e+00 -6.18597925e-01 -6.12804830e-01 5.35788178e-01
1.98810533e-01 -5.89369476e-01 7.51165807e-01 4.88951117e-01
7.88854733e-02 -2.56114632e-01 -1.19323862e+00 4.43591237e-01
7.53803313e-01 -2.45424032e-01 -7.17599094e-02 2.88926810e-01
4.75180954e-01 -3.04517686e-01 -1.00282347e+00 2.22903222e-01
2.12187041e-02 -1.30093300e+00 5.15175223e-01 9.32537317e-02
3.55468512e-01 -6.32379532e-01 -7.41324127e-02 -1.01430559e+00
-3.08915526e-01 -1.15659738e+00 -4.65104371e-01 1.31150794e+00
6.77319109e-01 -3.70428383e-01 7.32356012e-01 9.63568538e-02
-5.62537968e-01 -5.25739551e-01 -6.25960052e-01 -1.13118756e+00
-7.50882566e-01 -5.58527589e-01 5.45310855e-01 5.77755213e-01
4.52306777e-01 4.43805456e-01 -1.68794498e-01 1.31561413e-01
3.98472637e-01 -2.34802768e-01 5.61022043e-01 -1.19478106e+00
-1.27562428e+00 -1.84329867e-01 -7.12291181e-01 -3.05397838e-01
6.23620208e-03 -3.96856755e-01 -2.05576848e-02 -5.98846018e-01
-8.56324472e-03 -4.08007294e-01 1.12655908e-02 4.70510870e-01
7.36022741e-02 -3.97139937e-01 -3.02423745e-01 -4.44086105e-01
-4.48852926e-01 -2.39027888e-01 4.37520802e-01 2.45060951e-01
-4.19288903e-01 -2.96615548e-02 -6.13847435e-01 5.15287995e-01
6.24577820e-01 -8.81650448e-01 -5.48791766e-01 3.63038152e-01
8.78204048e-01 5.22354245e-01 2.24036694e-01 -1.40177333e+00
2.43878961e-01 1.62772387e-01 9.69747230e-02 -9.26679075e-01
-1.73013926e-01 -1.11048126e+00 8.60376894e-01 9.97106969e-01
5.35850823e-01 4.26940590e-01 7.42852092e-01 2.39313006e-01
-9.16001722e-02 -8.63515019e-01 5.09867430e-01 3.39860529e-01
-1.01107013e+00 -3.80483210e-01 -5.41697800e-01 -4.60751951e-01
1.32288337e+00 -1.83870926e-01 -2.35311881e-01 3.51017624e-01
-7.63159215e-01 2.26292267e-01 7.85484970e-01 -9.74884629e-02
6.24421358e-01 -1.10340631e+00 -2.45162562e-01 7.08369553e-01
1.12890959e-01 -7.89447188e-01 7.74711072e-02 7.51733363e-01
-7.70218372e-01 9.77228284e-02 -6.25570476e-01 -7.21841872e-01
-1.31537700e+00 6.89797580e-01 9.99893644e-04 -5.77967048e-01
-4.15275186e-01 5.42816341e-01 -4.05155450e-01 4.45827633e-01
-5.80880791e-02 -9.00665462e-01 3.06178719e-01 -3.04604977e-01
5.46631634e-01 5.33477426e-01 6.07043922e-01 8.19334447e-01
-9.36499119e-01 4.74774867e-01 -1.80529401e-01 -4.34732102e-02
1.21905160e+00 3.49704474e-01 -2.61473089e-01 6.36137486e-01
7.66001880e-01 2.50854909e-01 -5.99155366e-01 5.60268342e-01
1.00297976e+00 -2.42714822e-01 4.98437062e-02 -9.50218856e-01
-8.40590119e-01 5.82476676e-01 6.08520091e-01 2.49533266e-01
1.59468925e+00 7.55097643e-02 2.12090343e-01 2.77637869e-01
1.16367805e+00 -1.05841160e+00 4.51802788e-03 2.92823136e-01
2.89087802e-01 -4.26260680e-01 3.36000353e-01 -6.28625572e-01
8.63640904e-02 1.49234080e+00 6.10615849e-01 -4.19694185e-01
7.96544850e-01 1.60136890e+00 -6.02000475e-01 -5.00575066e-01
-1.11386395e+00 5.02258301e-01 -1.44115210e-01 7.87857890e-01
5.44918597e-01 1.20287769e-01 -3.63361359e-01 9.78477597e-01
2.37779319e-02 2.59505272e-01 8.91594589e-01 1.60394585e+00
-6.65903449e-01 -1.76882327e+00 -5.14606237e-01 4.59902346e-01
-1.88349232e-01 1.86083972e-01 -5.96513093e-01 1.27475309e+00
9.29494798e-02 8.00443470e-01 -4.18791592e-01 -1.10438514e+00
5.84251106e-01 -7.85583034e-02 9.63556349e-01 -8.12399328e-01
-1.01446855e+00 1.38063625e-01 6.89830124e-01 -1.17290688e+00
5.42237997e-01 -4.47410315e-01 -1.41422832e+00 -4.87883210e-01
-5.31974912e-01 2.11092532e-01 7.36705184e-01 7.12105393e-01
7.69087195e-01 1.36915898e+00 9.49028313e-01 -5.41104615e-01
-3.75355750e-01 -6.26326978e-01 -8.72647524e-01 -4.43255663e-01
-4.04783115e-02 -9.77039933e-01 -9.11764354e-02 -8.51025730e-02]
|
[7.557188987731934, 7.521271705627441]
|
395d7779-d870-4b36-92bc-f8662cbcdd16
|
belt-blockwise-missing-embedding-learning
|
2105.10360
| null |
https://arxiv.org/abs/2105.10360v3
|
https://arxiv.org/pdf/2105.10360v3.pdf
|
Multi-source Learning via Completion of Block-wise Overlapping Noisy Matrices
|
Matrix completion has attracted attention in many fields, including statistics, applied mathematics, and electrical engineering. Most of the works focus on the independent sampling models under which the observed entries are sampled independently. Motivated by applications in the integration of knowledge graphs derived from multi-source biomedical data such as those from Electronic Health Records (EHR) and biomedical text, we propose the {\bf B}lock-wise {\bf O}verlapping {\bf N}oisy {\bf M}atrix {\bf I}ntegration (BONMI) to treat blockwise missingness of symmetric matrices representing relatedness between entity pairs. Our idea is to exploit the orthogonal Procrustes problem to align the eigenspace of the two sub-matrices, then complete the missing blocks by the inner product of the two low-rank components. Besides, we prove the statistical rate for the eigenspace of the underlying matrix, which is comparable to the rate under the independently missing assumption. Simulation studies show that the method performs well under a variety of configurations. In the real data analysis, the method is applied to two tasks: (i) the integrating of several point-wise mutual information matrices built by English EHR and Chinese medical text data, and (ii) the machine translation between English and Chinese medical concepts. Our method shows an advantage over existing methods.
|
['Junwei Lu', 'Tianxi Cai', 'Doudou Zhou']
|
2021-05-21
| null | null | null | null |
['electrical-engineering']
|
['miscellaneous']
|
[ 6.11091852e-01 1.09752499e-01 -1.28962263e-01 -5.63070215e-02
-5.34350872e-01 -3.40784848e-01 2.24104747e-01 3.77403259e-01
-4.97242749e-01 8.15522134e-01 3.22971106e-01 -4.35783058e-01
-6.28916323e-01 -5.47489643e-01 -6.42078638e-01 -9.41871881e-01
-3.70565653e-01 4.69644248e-01 -4.71019983e-01 -7.14548007e-02
-5.59117645e-02 -6.95402771e-02 -8.98239553e-01 7.80128613e-02
1.16746044e+00 5.47640443e-01 2.48219632e-03 3.04328948e-01
4.37785052e-02 6.19313538e-01 -1.68562546e-01 -4.78890806e-01
3.03188324e-01 -3.21696728e-01 -7.24713743e-01 6.37152866e-02
-1.63110092e-01 1.83878079e-01 -4.11966711e-01 1.45163321e+00
3.57516408e-01 -7.58963674e-02 8.00029635e-01 -1.33654046e+00
-4.04776067e-01 7.97177315e-01 -1.11427355e+00 2.63230945e-03
3.97495270e-01 -4.41494167e-01 9.98385727e-01 -1.13608444e+00
8.29466224e-01 9.95117188e-01 6.18846536e-01 9.60503221e-02
-1.36437213e+00 -7.70766079e-01 -1.02216050e-01 2.20988557e-01
-1.87205887e+00 -2.50461936e-01 4.80753064e-01 -6.12186730e-01
4.03637081e-01 4.60016102e-01 4.06742334e-01 7.95630932e-01
2.78692812e-01 7.26939082e-01 1.07752800e+00 -3.89191896e-01
1.04349196e-01 9.65230763e-02 4.41742122e-01 5.47020733e-01
8.37623715e-01 -2.67464489e-01 -4.47861880e-01 -7.83197820e-01
5.88004529e-01 3.88573527e-01 -5.56746721e-01 -3.56895715e-01
-1.79482269e+00 7.16409266e-01 -2.50042994e-02 4.03745860e-01
-5.81618607e-01 -4.01188970e-01 2.40620524e-01 2.51502365e-01
2.04958320e-01 7.53800347e-02 -2.99471825e-01 3.18782985e-01
-7.13332057e-01 -1.75860941e-01 9.51942682e-01 1.29577458e+00
6.98952138e-01 -3.19949716e-01 -4.47788835e-02 7.56216466e-01
4.29175019e-01 7.75305152e-01 4.45370585e-01 -3.77771616e-01
8.07672620e-01 5.09230793e-01 1.08177178e-01 -1.24895918e+00
-5.33036590e-01 -4.51030076e-01 -1.68981040e+00 -7.54841030e-01
4.30874079e-01 -3.27675343e-01 -3.71197432e-01 1.92064261e+00
4.14286733e-01 3.71269315e-01 2.62841791e-01 6.21544540e-01
6.14093244e-01 3.96159768e-01 -2.77282625e-01 -6.05951667e-01
1.65147853e+00 -3.76947641e-01 -9.44099903e-01 -1.67011507e-02
6.58283710e-01 -7.55713046e-01 3.01721126e-01 3.81027639e-01
-7.52111197e-01 -1.54597193e-01 -9.07354593e-01 2.51265019e-01
1.29200876e-01 2.87575334e-01 5.14073133e-01 6.57633543e-01
-6.80210769e-01 2.85515070e-01 -7.14414895e-01 -1.58037379e-01
2.11307600e-01 3.89166951e-01 -9.25192118e-01 -5.36261499e-01
-1.28000355e+00 4.04177994e-01 3.67438138e-01 3.33096266e-01
-1.82283655e-01 -5.77004552e-01 -6.96398437e-01 -1.27461165e-01
4.88110214e-01 -7.77234674e-01 3.57430637e-01 -6.30909503e-01
-7.00183868e-01 5.82378805e-01 -1.86002061e-01 -1.95920497e-01
3.77417535e-01 -1.48835734e-01 -3.79701138e-01 -7.27699995e-02
4.05323267e-01 -1.92607269e-01 6.99875534e-01 -8.77352238e-01
-1.65743232e-01 -8.33759427e-01 -5.44971585e-01 1.57821789e-01
-6.09991372e-01 -2.60836840e-01 -5.40602624e-01 -8.56083274e-01
5.85583627e-01 -1.11813271e+00 -4.73304182e-01 -3.93073529e-01
-8.29591691e-01 1.93047151e-01 1.65078133e-01 -1.04415333e+00
1.33881533e+00 -2.25767255e+00 6.91859961e-01 6.97652936e-01
5.66614389e-01 -1.58800766e-01 -1.44329920e-01 7.89319336e-01
-3.87109339e-01 -1.34038538e-01 -5.03973901e-01 2.15859264e-02
-3.87703717e-01 1.28495172e-01 -7.74172321e-02 7.47271597e-01
-2.82376796e-01 4.50132072e-01 -1.05724645e+00 -3.63521248e-01
-3.08523566e-01 2.71020055e-01 -4.08895016e-01 5.90455383e-02
4.73661095e-01 5.02138317e-01 -5.41665137e-01 1.66176274e-01
8.68509889e-01 -6.19317412e-01 8.51948321e-01 -4.99382019e-01
3.08595210e-01 -1.15007885e-01 -1.77031398e+00 1.63438892e+00
-1.51566984e-02 4.66900729e-02 2.49886543e-01 -1.02936065e+00
6.83811843e-01 3.80566090e-01 8.70620251e-01 -9.48170349e-02
8.59865546e-02 2.27371916e-01 3.23981345e-01 -3.90845090e-01
1.75987408e-01 3.16079035e-02 3.25293429e-02 3.81164670e-01
-1.29573762e-01 4.76206332e-01 2.97820717e-01 5.95969498e-01
1.31676030e+00 -3.72131914e-01 7.55442560e-01 -4.91544396e-01
7.82655597e-01 -3.68369311e-01 8.41485083e-01 5.67632854e-01
3.09860170e-01 4.78570312e-01 5.77031553e-01 1.14233926e-01
-8.88657749e-01 -8.65435839e-01 -2.14540705e-01 4.07253206e-01
9.44785699e-02 -6.24053180e-01 -5.76748013e-01 -4.73411679e-01
-1.68566778e-02 4.51777220e-01 -7.13443875e-01 -6.22690357e-02
-1.56788126e-01 -1.19937825e+00 6.28908515e-01 2.40366772e-01
3.28528553e-01 -2.58421421e-01 -9.32412979e-04 1.72769845e-01
-7.30559051e-01 -1.19073713e+00 -7.21223831e-01 -4.12727483e-02
-9.27998483e-01 -1.31953335e+00 -8.93429279e-01 -4.13392782e-01
9.84290421e-01 3.77194375e-01 6.21046782e-01 -7.14929923e-02
-2.58474559e-01 3.81737590e-01 -3.39845836e-01 -2.01313570e-01
-2.66250461e-01 -8.24402049e-02 4.52394187e-01 6.32837594e-01
2.51042366e-01 -5.69538116e-01 -5.55135190e-01 4.56711501e-01
-1.17781043e+00 2.96622455e-01 8.19260657e-01 1.06488466e+00
5.91355622e-01 -6.67068288e-02 4.46328402e-01 -1.12294102e+00
6.98910832e-01 -8.40041459e-01 -1.96725711e-01 3.91365826e-01
-6.80139661e-01 2.07134157e-01 4.20375854e-01 -3.35467964e-01
-6.59791410e-01 8.82898346e-02 2.57918417e-01 -3.66944194e-01
3.86200845e-01 1.05629778e+00 -3.74313176e-01 4.59583998e-01
4.38480884e-01 3.78982067e-01 1.54404834e-01 -4.91815627e-01
2.71272302e-01 8.27528954e-01 3.62537205e-01 -2.86046535e-01
8.16252112e-01 6.00231171e-01 2.95828909e-01 -1.01136220e+00
-5.00030935e-01 -8.67819190e-01 -5.85341811e-01 2.92275697e-01
7.18857169e-01 -1.18751097e+00 -7.16273487e-01 2.19499052e-01
-9.99978483e-01 4.36323673e-01 7.26169944e-02 9.87804890e-01
-1.85851470e-01 1.02161920e+00 -5.59882045e-01 -6.49220169e-01
-5.14187753e-01 -9.27957535e-01 7.52601504e-01 -2.59159476e-01
-2.54404277e-01 -7.81469584e-01 4.15041804e-01 3.92170459e-01
-2.31088996e-01 2.48194095e-02 1.21654022e+00 -9.60467637e-01
-2.40793988e-01 -4.24280375e-01 -2.64526099e-01 2.36889228e-01
4.15336788e-01 -4.39306468e-01 -3.85275096e-01 -5.54252148e-01
9.58997309e-02 2.70975858e-01 5.20235121e-01 2.57248759e-01
7.69516647e-01 -6.15653217e-01 -6.56347930e-01 3.50517511e-01
1.23106909e+00 -3.07470523e-02 5.93785465e-01 -2.29763597e-01
9.12702739e-01 5.26050329e-01 4.63650316e-01 7.57811606e-01
3.57291043e-01 5.89269578e-01 -8.00165534e-02 -8.06046575e-02
5.39187610e-01 -1.01597533e-01 1.71267733e-01 1.51171601e+00
-1.63298130e-01 -1.14254683e-01 -9.35928822e-01 6.16807520e-01
-2.22170639e+00 -9.08388793e-01 -5.63253939e-01 2.68231344e+00
8.44312668e-01 -2.93922931e-01 -1.12797678e-01 1.46676093e-01
8.38457465e-01 -2.61649162e-01 -4.88771826e-01 3.72361332e-01
-1.68196797e-01 1.03777587e-01 7.61160016e-01 3.24555844e-01
-8.06839883e-01 2.03821152e-01 5.12164831e+00 9.18198466e-01
-4.25757885e-01 1.78330511e-01 4.36703295e-01 2.47645602e-01
-4.51571226e-01 1.84192300e-01 -5.41322589e-01 3.82817775e-01
6.61653638e-01 -1.96980342e-01 2.93480545e-01 3.91206741e-01
-6.28216751e-03 -1.74916878e-01 -1.11417425e+00 1.23171484e+00
2.19454527e-01 -1.01058936e+00 4.44712751e-02 3.49044383e-01
8.70226383e-01 -5.79172410e-02 -1.47813872e-01 -4.93588764e-03
3.00920308e-01 -8.56176138e-01 1.45226285e-01 6.17936313e-01
9.88356173e-01 -5.31571329e-01 7.86316574e-01 5.98520815e-01
-1.18687570e+00 1.30652651e-01 -3.75257671e-01 1.69404507e-01
9.26716104e-02 1.14189434e+00 -9.15409744e-01 1.26951075e+00
3.93544644e-01 7.70711243e-01 -2.96934903e-01 8.48979831e-01
1.56347752e-01 5.57414591e-01 -3.21592510e-01 1.17792524e-01
-2.65481830e-01 -8.02420437e-01 6.52309895e-01 8.77145171e-01
2.75846988e-01 2.19363540e-01 1.35733336e-01 4.31553960e-01
-7.22605810e-02 6.39624119e-01 -6.98665142e-01 -3.76588516e-02
5.05930960e-01 1.19619083e+00 -6.81227267e-01 -3.15055609e-01
-6.10241473e-01 9.82066631e-01 2.63049632e-01 3.12538922e-01
-6.16534710e-01 -3.94691825e-01 5.20410895e-01 -1.38674423e-01
1.27610490e-01 -1.27040952e-01 -3.23228598e-01 -1.46437335e+00
2.25480303e-01 -1.22749221e+00 4.73034531e-01 -3.25526476e-01
-1.42439759e+00 5.55119276e-01 6.58266246e-02 -1.36000645e+00
-1.74801543e-01 -3.64551485e-01 4.28969879e-03 9.97489035e-01
-9.46009934e-01 -6.91878915e-01 -1.11375265e-01 7.64543056e-01
-9.91367698e-02 -3.37115049e-01 9.24081683e-01 7.29475141e-01
-8.47191274e-01 4.51327175e-01 7.93609500e-01 4.01046842e-01
7.09959030e-01 -8.21801782e-01 1.04578346e-01 7.86169529e-01
2.66180098e-01 1.15393877e+00 6.73115611e-01 -1.00103664e+00
-1.67981315e+00 -8.55142236e-01 8.88378620e-01 -1.35670140e-01
6.27824187e-01 -2.75512099e-01 -7.53269434e-01 7.55160213e-01
-6.00326918e-02 -1.83585986e-01 1.07398391e+00 1.96902469e-01
-4.22216654e-01 3.03349346e-02 -8.72664869e-01 6.43771231e-01
9.22094285e-01 -3.13321233e-01 -2.84244210e-01 6.24610662e-01
2.69894987e-01 -9.46551487e-02 -1.21553695e+00 5.12322724e-01
5.70776463e-01 -5.98887563e-01 1.00250733e+00 -9.02505040e-01
2.89212286e-01 -4.85993117e-01 -3.11347723e-01 -1.12865162e+00
-4.62784439e-01 -7.10444629e-01 1.75070614e-02 8.67758036e-01
4.58558917e-01 -7.25830019e-01 5.63586295e-01 6.63496614e-01
4.62570369e-01 -5.17030239e-01 -1.13031995e+00 -5.95393419e-01
-2.95165539e-01 -1.00409828e-01 3.68032634e-01 1.24781835e+00
1.63783297e-01 7.38339841e-01 -7.54687726e-01 2.36743510e-01
8.35952997e-01 1.51086170e-02 8.77444327e-01 -1.39956582e+00
-6.66715860e-01 1.34422660e-01 -4.68067080e-01 -8.74859333e-01
-2.12879419e-01 -1.17842555e+00 -2.27273867e-01 -1.38915527e+00
8.59663129e-01 -5.87109506e-01 -3.31860214e-01 3.09715688e-01
-4.50140506e-01 -1.00737572e-01 1.23540601e-02 5.00830948e-01
-3.76800895e-01 5.91873705e-01 9.15961921e-01 -1.84917092e-01
-1.11997917e-01 2.72230562e-02 -7.50374258e-01 6.24113083e-01
2.92470068e-01 -6.38225913e-01 -4.31789219e-01 -1.78085923e-01
5.66762984e-01 5.47875226e-01 -2.95891315e-02 -6.07215583e-01
2.52712160e-01 -5.81179559e-02 5.65283895e-02 -4.50994760e-01
1.76710382e-01 -9.14172351e-01 7.28396833e-01 6.12399042e-01
-1.34761080e-01 3.19748968e-01 -1.04259759e-01 1.10355270e+00
-7.35665411e-02 -3.02984685e-01 3.30174208e-01 1.87679008e-02
4.24093939e-02 3.73609364e-01 -2.53812224e-01 1.56379133e-01
8.10714960e-01 1.21139497e-01 -7.52754062e-02 -3.78468841e-01
-7.33644068e-01 1.81323886e-01 3.57879102e-02 2.59459168e-01
6.12092197e-01 -1.42553616e+00 -1.13945329e+00 3.20762187e-01
2.02256769e-01 -9.56221521e-02 6.14065886e-01 1.54032338e+00
-8.65396187e-02 3.23999196e-01 2.16906704e-03 -6.43827796e-01
-1.43334794e+00 6.61695957e-01 -3.29490781e-01 -6.72578454e-01
-4.09194559e-01 4.35268164e-01 4.85742450e-01 -4.26115304e-01
-1.68178920e-02 1.52555823e-01 -2.64144897e-01 8.14283863e-02
4.76758391e-01 5.40118277e-01 1.85106754e-01 -7.80580401e-01
-4.83059198e-01 3.52669567e-01 -2.61766046e-01 -2.44847804e-01
1.26681006e+00 -1.98596328e-01 -7.18321681e-01 4.98548180e-01
1.28634739e+00 3.60448837e-01 -2.72309780e-01 -6.45330071e-01
1.08943768e-01 -2.96983719e-01 -2.90932804e-01 -2.62855679e-01
-8.02106380e-01 4.39613640e-01 3.27838391e-01 -1.49198756e-01
9.13637638e-01 -4.70880747e-01 3.63618016e-01 6.42029524e-01
5.15632391e-01 -7.97353387e-01 -3.57290387e-01 2.38392889e-01
6.97393775e-01 -8.80801797e-01 4.14171457e-01 -8.05295765e-01
-5.85205734e-01 7.12793589e-01 -5.50360195e-02 6.40540943e-02
9.68208551e-01 1.11966543e-02 -2.58937210e-01 -1.51058763e-01
-5.17738640e-01 -6.04005493e-02 4.95226622e-01 3.29222143e-01
5.07784545e-01 3.26436549e-01 -9.15177107e-01 7.11835504e-01
1.19145505e-01 7.49112247e-03 4.67157781e-01 6.84759259e-01
1.88833937e-01 -1.35535681e+00 -6.60896420e-01 7.89738834e-01
-4.82738167e-01 -3.94334108e-01 -2.70822197e-01 4.28613514e-01
-1.79608762e-01 9.75031137e-01 -4.19269294e-01 -3.56916130e-01
1.46267205e-01 3.64145525e-02 4.41479832e-01 -5.27734637e-01
-2.00498477e-01 2.17060626e-01 -2.00518146e-02 -1.08520664e-01
-4.76943970e-01 -1.02368796e+00 -1.04426289e+00 1.74049381e-02
-5.10704398e-01 3.70139807e-01 3.51525754e-01 8.96468580e-01
5.72347283e-01 3.18003476e-01 6.95677817e-01 -1.52899265e-01
-7.72836149e-01 -1.06046200e+00 -9.89236712e-01 7.49927521e-01
-1.98905468e-02 -4.71668065e-01 -1.97841823e-01 -8.29339847e-02]
|
[7.172196865081787, 4.747644901275635]
|
5fc2a444-2240-42de-890c-9c106e43537b
|
lavender-unifying-video-language
|
2206.07160
| null |
https://arxiv.org/abs/2206.07160v1
|
https://arxiv.org/pdf/2206.07160v1.pdf
|
LAVENDER: Unifying Video-Language Understanding as Masked Language Modeling
|
Unified vision-language frameworks have greatly advanced in recent years, most of which adopt an encoder-decoder architecture to unify image-text tasks as sequence-to-sequence generation. However, existing video-language (VidL) models still require task-specific designs in model architecture and training objectives for each task. In this work, we explore a unified VidL framework LAVENDER, where Masked Language Modeling (MLM) is used as the common interface for all pre-training and downstream tasks. Such unification leads to a simplified model architecture, where only a lightweight MLM head, instead of a decoder with much more parameters, is needed on top of the multimodal encoder. Surprisingly, experimental results show that this unified framework achieves competitive performance on 14 VidL benchmarks, covering video question answering, text-to-video retrieval and video captioning. Extensive analyses further demonstrate the advantage of LAVENDER over existing VidL methods in: (i) supporting all downstream tasks with just a single set of parameter values when multi-task finetuned; (ii) few-shot generalization on various downstream tasks; and (iii) enabling zero-shot evaluation on video question answering tasks. Code is available at https://github.com/microsoft/LAVENDER.
|
['Lijuan Wang', 'Ce Liu', 'Zicheng Liu', 'Chung-Ching Lin', 'Kevin Lin', 'Zhe Gan', 'Linjie Li']
|
2022-06-14
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Li_LAVENDER_Unifying_Video-Language_Understanding_As_Masked_Language_Modeling_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_LAVENDER_Unifying_Video-Language_Understanding_As_Masked_Language_Modeling_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['video-question-answering']
|
['computer-vision']
|
[ 1.82745263e-01 -1.99764743e-01 -2.72286773e-01 -3.82857949e-01
-1.23757875e+00 -5.26524961e-01 8.14049184e-01 -2.98250049e-01
-5.00415146e-01 3.97146046e-01 2.31574520e-01 -5.23321986e-01
4.14726943e-01 -2.12106571e-01 -9.03554440e-01 -4.59585428e-01
1.88118309e-01 2.37567574e-01 3.77053231e-01 -1.51437268e-01
3.60862687e-02 -1.52431428e-01 -1.49358559e+00 6.99736297e-01
7.12706923e-01 9.89176631e-01 6.38780713e-01 1.00705636e+00
-1.93323910e-01 1.09928644e+00 -3.81744534e-01 -6.11682594e-01
2.10593292e-03 -4.71348673e-01 -7.70643055e-01 1.80246904e-01
6.21503592e-01 -7.13249266e-01 -4.83058780e-01 8.26412618e-01
5.66971421e-01 7.33043775e-02 5.19395232e-01 -1.43931258e+00
-1.04252684e+00 5.14181495e-01 -4.86317635e-01 4.22521383e-02
4.58398223e-01 6.25408113e-01 1.08822560e+00 -1.15302575e+00
5.05602717e-01 1.36213589e+00 4.18366313e-01 8.44384789e-01
-9.94473934e-01 -5.62474549e-01 3.66746783e-01 5.46571374e-01
-1.41158319e+00 -7.48386860e-01 3.78240705e-01 -4.56786573e-01
1.05614579e+00 8.23102370e-02 1.00412108e-01 1.48302603e+00
1.16111875e-01 1.14280891e+00 5.48574746e-01 -2.28268459e-01
-2.72398163e-02 1.23070851e-02 1.15300216e-01 7.35949159e-01
2.55956757e-03 -3.00547868e-01 -5.83237290e-01 8.27322677e-02
5.37086070e-01 -1.30928278e-01 -3.82518440e-01 -2.25087270e-01
-1.29167187e+00 8.21196377e-01 2.04313025e-02 2.19148723e-03
-1.14667431e-01 5.47176719e-01 7.12726474e-01 2.91568547e-01
3.22063714e-01 -6.21698622e-04 -3.02631259e-01 -2.67652482e-01
-9.67906415e-01 1.70081735e-01 7.12115049e-01 1.30984867e+00
6.63822472e-01 1.03364557e-01 -7.48734474e-01 8.91389430e-01
3.78690183e-01 5.49589932e-01 6.05072021e-01 -1.09133303e+00
8.15436006e-01 2.73975760e-01 -4.42738868e-02 -4.91809398e-01
-6.03413321e-02 -1.04129449e-01 -7.56747186e-01 -3.03537369e-01
9.94983613e-02 -1.29882336e-01 -1.16514742e+00 1.76068902e+00
8.00128877e-02 2.66131163e-01 1.78398132e-01 1.01909804e+00
1.12055790e+00 1.02597475e+00 1.79580897e-01 6.27110973e-02
1.43261933e+00 -1.57936203e+00 -6.25193655e-01 -3.54387283e-01
7.46154606e-01 -7.63315678e-01 1.44287395e+00 1.00812521e-02
-1.26764011e+00 -8.06201100e-01 -7.72226989e-01 -5.69970191e-01
-1.77245438e-01 2.77268529e-01 4.28828984e-01 4.11134005e-01
-1.43141723e+00 -5.11766002e-02 -6.68654084e-01 -2.97894686e-01
2.90305555e-01 1.26713052e-01 -2.05599710e-01 -3.05168778e-01
-1.21391237e+00 5.05915940e-01 3.59300256e-01 6.90359101e-02
-1.26712954e+00 -6.89332843e-01 -1.20011044e+00 1.60558373e-01
5.69218218e-01 -1.01693022e+00 1.69236958e+00 -9.83720660e-01
-1.50271070e+00 8.12272191e-01 -3.98764729e-01 -5.29292941e-01
5.49073756e-01 -3.41163546e-01 -2.98778802e-01 4.89332199e-01
1.55662268e-01 1.19635189e+00 1.13929999e+00 -1.21687841e+00
-5.66085279e-01 2.02841356e-01 3.85329723e-01 2.97584534e-01
-3.17864895e-01 9.56807137e-02 -1.31428766e+00 -6.60014331e-01
-5.44114351e-01 -8.39049637e-01 -6.59200996e-02 1.12671062e-01
-2.41719425e-01 -2.83175439e-01 9.26870167e-01 -7.68856645e-01
1.25229621e+00 -2.10575175e+00 7.64078870e-02 -5.11545181e-01
1.06507547e-01 4.00372237e-01 -7.61494577e-01 5.93712866e-01
6.30255193e-02 6.31832890e-03 -2.20035136e-01 -6.96668625e-01
1.16242900e-01 1.30982324e-01 -4.36332077e-01 1.08998872e-01
2.58391470e-01 1.43780661e+00 -6.94553792e-01 -7.45227814e-01
2.63295144e-01 4.00874436e-01 -7.85904229e-01 3.85996252e-01
-5.86066663e-01 1.01927102e-01 -4.27521646e-01 7.47123718e-01
4.20543343e-01 -6.93256676e-01 -4.83507710e-03 -1.97912440e-01
-1.26988785e-02 9.15545225e-02 -7.57726252e-01 2.04625797e+00
-5.18450379e-01 7.38007247e-01 2.10100472e-01 -8.56803060e-01
5.52703142e-01 5.16094387e-01 3.36762130e-01 -7.73825943e-01
8.88088893e-04 -3.59995407e-04 -2.86523223e-01 -8.65846515e-01
5.58683991e-01 3.35858941e-01 -1.11411087e-01 4.01620954e-01
3.08643490e-01 8.56449902e-02 5.27985930e-01 5.53576708e-01
8.43052089e-01 2.86638319e-01 5.16129248e-02 7.75597692e-02
8.14988673e-01 -1.04500823e-01 2.62771696e-01 9.18719172e-01
-1.89096153e-01 7.67992675e-01 4.63347316e-01 -3.68281864e-02
-1.04814470e+00 -1.03567004e+00 2.36373335e-01 1.49624884e+00
1.79619789e-01 -5.64070702e-01 -8.02579820e-01 -5.59226513e-01
-1.14310615e-01 7.88278878e-01 -3.97687703e-01 -1.76118657e-01
-4.46549952e-01 -4.12660867e-01 7.60868371e-01 4.23444420e-01
5.54052889e-01 -9.82722163e-01 -3.99717957e-01 3.48559278e-03
-4.98995513e-01 -1.62403750e+00 -1.00297487e+00 -2.99212217e-01
-6.25853002e-01 -7.68997192e-01 -1.05187058e+00 -1.00491786e+00
3.96504819e-01 7.27572441e-01 1.19748473e+00 8.79269391e-02
-2.08272830e-01 8.70919049e-01 -5.83593369e-01 -4.88939360e-02
-4.06407505e-01 5.20760752e-02 -3.22004139e-01 1.82099387e-01
1.12853177e-01 -1.80154994e-01 -6.69682026e-01 3.61958414e-01
-1.15707374e+00 5.08088648e-01 7.11722851e-01 7.74150908e-01
3.62844646e-01 -7.33215213e-01 7.38930404e-01 -4.61108178e-01
6.10826969e-01 -6.09719098e-01 -4.92466301e-01 6.24454141e-01
-2.88370669e-01 -7.85227679e-03 5.41968822e-01 -4.34889168e-01
-1.18029356e+00 -1.52242124e-01 -3.16604823e-01 -7.04853892e-01
-9.74116325e-02 4.64585781e-01 -2.13333800e-01 2.19611809e-01
1.94037914e-01 6.43211007e-01 1.72660530e-01 -4.56889242e-01
6.99568450e-01 8.14133048e-01 6.58854544e-01 -6.07705057e-01
5.79138935e-01 3.73018354e-01 -5.01915514e-01 -8.92753541e-01
-7.73564994e-01 -5.79833627e-01 -3.14967930e-01 -3.45394492e-01
1.09911799e+00 -1.43812239e+00 -7.51999617e-01 4.67583716e-01
-1.33454609e+00 -6.97615743e-01 9.41864774e-02 2.67699003e-01
-6.31860673e-01 6.63372159e-01 -8.34503531e-01 -5.68105757e-01
-5.65605819e-01 -1.55884337e+00 1.41567588e+00 3.66150998e-02
8.24858472e-02 -9.18212235e-01 -2.86814004e-01 8.22759926e-01
3.73203963e-01 -3.82332236e-01 8.23559701e-01 -4.08747524e-01
-7.33101785e-01 -2.75100768e-02 -5.32083631e-01 5.76905251e-01
-2.50789374e-01 -1.63024247e-01 -9.95696068e-01 -4.70582187e-01
-9.68030691e-02 -7.83777714e-01 1.12826264e+00 4.02216136e-01
1.51301777e+00 -1.94714159e-01 -1.48407936e-01 7.45900810e-01
1.33711779e+00 1.33408070e-01 5.99213481e-01 1.17339738e-01
8.58503580e-01 3.68662030e-01 4.57616210e-01 2.44151294e-01
8.25861990e-01 7.82140791e-01 3.62934768e-01 -2.79201586e-02
-4.38581109e-01 -3.74103159e-01 9.75370467e-01 9.63267505e-01
2.34581381e-01 -5.03828168e-01 -9.09594774e-01 5.07649779e-01
-2.11613321e+00 -9.50074375e-01 -7.18267411e-02 1.88130474e+00
6.79327786e-01 -1.34165674e-01 5.77442124e-02 -6.17476165e-01
6.56631231e-01 4.43564385e-01 -6.72056675e-01 -1.75272569e-01
-1.24396481e-01 -1.94326982e-01 2.74561882e-01 5.30908465e-01
-1.06404984e+00 1.18175924e+00 5.79736567e+00 1.06022513e+00
-9.93889153e-01 3.93382519e-01 5.96691310e-01 -2.52704650e-01
-3.70488942e-01 -8.46438929e-02 -9.14860189e-01 5.41980863e-01
9.05597627e-01 -2.58680344e-01 2.32581705e-01 7.56929994e-01
4.70521688e-01 -7.78350187e-03 -1.04998517e+00 1.29156089e+00
4.45990831e-01 -1.49385154e+00 5.80054939e-01 -2.19055042e-01
5.93788087e-01 3.09106588e-01 -3.45379896e-02 6.29582107e-01
-4.76541370e-02 -8.94199789e-01 9.00191247e-01 4.31865185e-01
1.06300747e+00 -3.82107139e-01 4.79165971e-01 3.07636648e-01
-1.25858963e+00 -8.33831131e-02 -2.77919352e-01 1.26585498e-01
6.26394928e-01 1.91793159e-01 -5.32051623e-01 6.22590423e-01
6.08473063e-01 7.48004377e-01 -6.13291204e-01 8.50684941e-01
-8.08577761e-02 6.19974852e-01 7.07725286e-02 1.65605113e-01
5.60424328e-01 -9.17360634e-02 4.20994312e-01 1.61561644e+00
2.49451771e-01 -2.11341918e-01 3.55611920e-01 6.67958260e-01
-2.13401899e-01 8.75490587e-05 -4.96572405e-01 -1.09534800e-01
3.37293297e-01 1.15934098e+00 -3.12756896e-01 -5.21023333e-01
-9.04947937e-01 1.31641495e+00 1.49883866e-01 7.62352228e-01
-1.19893944e+00 -2.06305370e-01 8.35812926e-01 -2.00231075e-01
4.73060638e-01 -4.06321853e-01 1.70116216e-01 -1.46510684e+00
1.21878758e-01 -1.04613030e+00 4.38887924e-01 -1.00083852e+00
-1.06331086e+00 5.12138069e-01 4.06163000e-02 -1.10688221e+00
-4.02760208e-01 -6.22694433e-01 -4.24856752e-01 6.96346045e-01
-1.70966434e+00 -1.34548175e+00 -3.61221373e-01 7.63541102e-01
1.24197793e+00 -2.71550506e-01 4.72616345e-01 5.07968962e-01
-7.50240922e-01 7.37164497e-01 -7.66219646e-02 1.58293709e-01
8.59620094e-01 -8.27045977e-01 4.59212840e-01 9.26129937e-01
1.03336260e-01 3.28067929e-01 3.35609496e-01 -3.80414367e-01
-1.72118938e+00 -1.29130018e+00 8.18325102e-01 -3.62985641e-01
8.51162732e-01 -5.70042551e-01 -9.34522867e-01 7.31249928e-01
5.76604664e-01 -1.95579141e-01 5.13100922e-01 -3.40629846e-01
-3.63685250e-01 3.95537131e-02 -3.99505675e-01 8.56975377e-01
9.13176119e-01 -8.16544354e-01 -2.98942924e-01 4.40277219e-01
1.15156305e+00 -3.30109954e-01 -5.78531444e-01 2.72843510e-01
3.96046430e-01 -8.60738993e-01 1.07640100e+00 -5.63516080e-01
7.91482985e-01 -3.94639641e-01 -2.15172395e-01 -7.17532456e-01
-2.04471782e-01 -7.68583059e-01 -4.35604542e-01 1.17360878e+00
4.16642308e-01 -2.88151234e-01 4.30643827e-01 3.94160450e-01
-4.40621376e-01 -9.05857325e-01 -6.87566102e-01 -7.81611621e-01
-1.88320667e-01 -8.02052021e-01 2.06990540e-01 5.24879396e-01
-4.63248640e-01 7.26206839e-01 -6.61998749e-01 6.52309880e-02
5.26898026e-01 1.07417330e-02 9.36370134e-01 -3.97204936e-01
-4.52307284e-01 -5.83900809e-01 7.55363330e-02 -1.67776120e+00
3.41274619e-01 -1.00726676e+00 1.14854321e-01 -1.73371768e+00
4.78023022e-01 1.31080478e-01 -1.45330820e-02 4.82921571e-01
-4.11347449e-01 1.59577712e-01 5.59407055e-01 3.28331858e-01
-1.18366265e+00 8.06059241e-01 1.20686233e+00 -2.11013198e-01
7.14255050e-02 -2.50696599e-01 -6.60602331e-01 6.00939989e-01
6.85177565e-01 -6.96466863e-02 -7.24114239e-01 -1.08355594e+00
-4.89423238e-02 1.47822171e-01 6.50511265e-01 -8.23754013e-01
2.43863896e-01 -3.32622752e-02 -6.79761320e-02 -5.09884894e-01
5.61286330e-01 -4.67721581e-01 -2.03610510e-01 3.90412182e-01
-4.49689120e-01 1.04609676e-01 2.58771986e-01 5.63210785e-01
-3.84385675e-01 -2.52490103e-01 5.55249453e-01 2.25741211e-02
-1.30891311e+00 5.62171817e-01 -2.97260016e-01 2.03629017e-01
9.86580610e-01 -1.59815773e-01 -4.40800488e-01 -7.65349925e-01
-4.15818274e-01 6.74563229e-01 3.85599196e-01 8.86323810e-01
7.41099298e-01 -1.17013574e+00 -9.28873718e-01 -7.40926042e-02
4.32121933e-01 -1.30390704e-01 5.58035910e-01 9.26372230e-01
-4.27676767e-01 7.98917234e-01 1.74706951e-01 -6.87219203e-01
-1.31267834e+00 6.16846383e-01 1.58961177e-01 -1.47931099e-01
-6.47129536e-01 9.31747437e-01 7.07712531e-01 -5.82131594e-02
4.26755339e-01 -1.35884166e-01 3.39884348e-02 7.42981508e-02
7.13499427e-01 8.63514096e-02 -2.14261487e-01 -5.22521436e-01
-1.56004757e-01 3.95212173e-01 -2.34980851e-01 -1.23482957e-01
9.56549108e-01 -5.19371688e-01 8.77827629e-02 4.54366922e-01
1.26796424e+00 -5.70722938e-01 -1.56494904e+00 -2.11221069e-01
-1.35352820e-01 -1.03909619e-01 -5.35289720e-02 -5.24393320e-01
-9.35397029e-01 1.03768206e+00 1.70179039e-01 -2.23482594e-01
1.18904495e+00 1.07795686e-01 1.04013348e+00 5.11878073e-01
2.00656027e-01 -1.00350404e+00 3.18211824e-01 7.31034219e-01
1.02411342e+00 -1.34918249e+00 -3.29679847e-01 -1.50331035e-01
-1.04763031e+00 9.65110362e-01 7.32938349e-01 2.26101279e-01
3.01358193e-01 1.13454409e-01 1.17212653e-01 4.62575965e-02
-1.31305552e+00 -3.83737206e-01 4.23404664e-01 3.45278114e-01
5.74845910e-01 -2.29626298e-01 -1.85900256e-01 5.04348755e-01
2.49992669e-01 1.79506704e-01 3.43821466e-01 7.75841475e-01
-3.65097284e-01 -1.02651262e+00 -1.84563383e-01 3.55764687e-01
-3.60220551e-01 -4.16768312e-01 -2.54311855e-03 7.41512895e-01
-1.66340530e-01 9.64784980e-01 8.59506130e-02 -2.14970395e-01
3.01463678e-02 1.75876200e-01 3.06020588e-01 -6.26888812e-01
-2.94087380e-01 1.22897699e-01 1.10730991e-01 -8.05432737e-01
-2.89108068e-01 -3.70783180e-01 -9.60612237e-01 -1.11766029e-02
-4.43737060e-02 -1.11499531e-02 5.35311341e-01 1.00128877e+00
7.02488184e-01 5.03656030e-01 1.87891960e-01 -9.62126493e-01
-5.94529510e-01 -8.63960564e-01 -1.46081910e-01 4.00005609e-01
4.32832450e-01 -4.32862580e-01 1.92389451e-02 5.36017835e-01]
|
[10.567750930786133, 1.162280559539795]
|
6140aa15-3e62-4617-b734-6d81e6f2e59d
|
efficient-video-representation-learning-via
|
2211.10636
| null |
https://arxiv.org/abs/2211.10636v3
|
https://arxiv.org/pdf/2211.10636v3.pdf
|
Efficient Video Representation Learning via Motion-Aware Token Selection
|
Recently emerged Masked Video Modeling techniques demonstrated their potential by significantly outperforming previous methods in self-supervised learning for video. However, they require an excessive amount of computations and memory while predicting uninformative tokens/frames due to random masking strategies, requiring excessive computing power for training. (e.g., over 16 nodes with 128 NVIDIA A100 GPUs). To resolve this issue, we exploit the unequal information density among the patches in videos and propose a new token selection method, MATS: Motion-Aware Token Selection, that finds tokens containing rich motion features and drops uninformative ones during both self-supervised pre-training and fine-tuning. We further present an adaptive frame selection strategy that allows the model to focus on informative and causal frames with minimal redundancy. Our method significantly reduces computation and memory requirements, enabling the pre-training and fine-tuning on a single machine with 8 GPUs while achieving comparable performance to computation- and memory-heavy state-of-the-art methods on multiple benchmarks and on the uncurated Ego4D dataset. We are hopeful that the efficiency of our MATS will contribute to reducing the barrier to conducting further research on self-supervised learning for videos.
|
['Sung Ju Hwang', 'Youngwan Lee', 'Jaehong Yoon', 'Sunil Hwang']
|
2022-11-19
| null | null | null | null |
['self-supervised-action-recognition']
|
['computer-vision']
|
[ 2.97917336e-01 -1.27246723e-01 -4.31322932e-01 -1.81259796e-01
-5.88756859e-01 -2.74456590e-01 2.96557724e-01 1.56327877e-02
-5.71218133e-01 5.25727689e-01 1.11659408e-01 -1.85767233e-01
3.70659560e-01 -8.14784050e-01 -8.05662215e-01 -9.33277130e-01
-5.00457227e-01 7.43632540e-02 6.78533733e-01 3.30300122e-01
1.55482933e-01 3.28962542e-02 -2.10470414e+00 5.77831566e-01
4.93266672e-01 1.09169483e+00 3.45099479e-01 5.83583117e-01
8.98987129e-02 1.29287720e+00 -3.96275282e-01 -7.01564699e-02
2.78793633e-01 -2.78135300e-01 -7.40778923e-01 2.51515806e-01
8.02496195e-01 -5.57600319e-01 -6.10860646e-01 8.15569162e-01
6.38475180e-01 1.72264948e-01 3.55597228e-01 -1.04165447e+00
-1.36804774e-01 5.27158022e-01 -7.33574808e-01 5.23230314e-01
6.83137998e-02 2.99419701e-01 7.75753498e-01 -9.44426656e-01
6.19361162e-01 1.04726565e+00 6.15105033e-01 7.17819333e-01
-1.08443177e+00 -6.51514649e-01 1.07904002e-01 6.15815938e-01
-1.49191904e+00 -7.02535272e-01 6.59503162e-01 -3.94151509e-01
1.17635405e+00 3.01489055e-01 8.05792511e-01 9.27384138e-01
4.39219512e-02 1.10112619e+00 8.66696775e-01 -4.69952315e-01
3.43088806e-01 -1.00318426e-02 -1.05407089e-01 1.12998879e+00
1.60554856e-01 1.34249460e-02 -1.17458332e+00 -1.90303341e-01
8.63535941e-01 -1.03573203e-01 3.64542790e-02 -3.31809521e-01
-1.46162701e+00 7.80548871e-01 -2.09652334e-02 9.91062075e-02
-2.60854840e-01 4.19354796e-01 8.12345624e-01 1.55633584e-01
6.55738950e-01 -4.57218736e-02 -4.54075366e-01 -4.25308675e-01
-1.39716649e+00 9.87961236e-03 4.88720953e-01 8.83307636e-01
1.10888493e+00 3.70238125e-01 -2.35298201e-01 6.16172731e-01
-6.47501647e-02 1.57512635e-01 6.62124157e-01 -1.14415252e+00
2.60788381e-01 4.25829083e-01 -1.98774427e-01 -9.64245677e-01
-4.03929472e-01 -3.24915349e-01 -8.62242103e-01 1.66354045e-01
1.69143423e-01 -1.96290627e-01 -8.30931365e-01 1.64964139e+00
6.47384524e-01 6.76809490e-01 -1.19351983e-01 8.78484190e-01
7.55231440e-01 6.13843799e-01 1.01229087e-01 -3.30656052e-01
1.42242277e+00 -1.31451941e+00 -4.56350267e-01 -2.03056768e-01
9.55961585e-01 -6.26702607e-01 1.04076672e+00 3.72211337e-01
-1.13503945e+00 -6.90438092e-01 -1.01854539e+00 1.53047824e-02
1.01088665e-01 1.82073295e-01 1.06773376e+00 6.83748543e-01
-1.09493935e+00 7.68659532e-01 -1.28008461e+00 -1.68935195e-01
8.16409528e-01 5.56731582e-01 -2.92129487e-01 3.64919417e-02
-8.54958773e-01 3.43613952e-01 4.01018620e-01 -1.92294002e-01
-1.00891256e+00 -8.47062945e-01 -8.06949317e-01 1.02277137e-01
4.86035168e-01 -6.63372576e-01 1.03255010e+00 -1.27795112e+00
-1.29253078e+00 9.40493345e-01 -4.56057012e-01 -7.42143571e-01
6.23770654e-01 -1.55332744e-01 -1.70407474e-01 4.77268219e-01
1.87327683e-01 1.07947052e+00 1.30689061e+00 -8.59377205e-01
-7.70273805e-01 -2.13731647e-01 -2.68304348e-01 1.89662650e-01
-7.67025232e-01 -3.06928009e-02 -6.69629335e-01 -9.31889355e-01
7.64389336e-02 -1.03425837e+00 -1.14265867e-01 -2.08202889e-03
-2.65029520e-01 -2.48439446e-01 1.01090622e+00 -2.25580037e-01
1.26363611e+00 -2.17714977e+00 -1.32075444e-01 -8.24874490e-02
4.10239786e-01 4.59411412e-01 -5.81216477e-02 2.87303291e-02
6.27123937e-02 -1.79321408e-01 1.14949532e-01 -5.83522975e-01
-1.94738016e-01 2.40253463e-01 -2.15802386e-01 6.10557973e-01
1.62426189e-01 5.92870772e-01 -1.02564013e+00 -1.07213128e+00
4.01122242e-01 4.33151007e-01 -8.57894063e-01 5.38911372e-02
-1.25778303e-01 3.07665855e-01 -2.44992152e-01 7.70269692e-01
6.76478803e-01 -3.94933045e-01 2.30393901e-01 -2.88380146e-01
-1.42615080e-01 4.05335844e-01 -1.11913013e+00 1.70600438e+00
-2.23792121e-01 7.29462683e-01 -1.51012644e-01 -1.03304565e+00
5.35881042e-01 2.23879993e-01 8.02927315e-01 -5.36616445e-01
3.25621404e-02 3.10430378e-02 -3.58656138e-01 -5.35257041e-01
6.56899393e-01 2.33346611e-01 2.95208216e-01 5.30291378e-01
1.17191888e-01 4.85719949e-01 3.90566349e-01 2.44570360e-01
1.30232632e+00 3.16841424e-01 4.71825041e-02 -2.95650661e-01
3.51469338e-01 1.71489283e-01 7.72040009e-01 8.91945899e-01
-2.63645560e-01 4.04998273e-01 3.64394337e-01 -9.03931558e-01
-9.75316286e-01 -7.36250103e-01 1.66146066e-02 1.46738231e+00
1.11419834e-01 -8.88680041e-01 -7.58093953e-01 -7.17695951e-01
-8.35721046e-02 2.62955815e-01 -5.02979398e-01 -1.05633803e-01
-7.70379126e-01 -8.81770909e-01 5.27929068e-01 5.83708644e-01
6.44482970e-01 -9.27431345e-01 -1.03731954e+00 2.55484492e-01
-1.94366068e-01 -1.29934525e+00 -3.32224607e-01 3.48404825e-01
-1.11908233e+00 -9.78308201e-01 -4.11216289e-01 -9.50184822e-01
6.38963282e-01 7.36238420e-01 1.19121587e+00 2.56523073e-01
-4.45633024e-01 2.56266743e-01 -3.27117085e-01 6.37249202e-02
-2.00868264e-01 3.46038416e-02 3.21720064e-01 -4.36213277e-02
4.75141823e-01 -5.65590799e-01 -9.00363624e-01 4.89087284e-01
-8.99617434e-01 4.22814488e-01 3.77911985e-01 9.99471068e-01
7.42940664e-01 1.06757909e-01 2.60901511e-01 -8.38479459e-01
-1.95271760e-01 -4.97963578e-01 -4.49723750e-01 -1.11113710e-03
-3.90026182e-01 -8.55092034e-02 5.58877885e-01 -7.01808512e-01
-8.77601027e-01 4.47036356e-01 1.78870007e-01 -6.96471214e-01
7.49406144e-02 4.43469435e-02 3.17180157e-01 -3.97460550e-01
6.01227283e-01 4.16923344e-01 -2.97115296e-02 -1.83620036e-01
-3.52629879e-03 3.60902816e-01 5.06913126e-01 -5.03340364e-01
5.64257562e-01 8.20148170e-01 9.99180004e-02 -1.07391512e+00
-6.01427436e-01 -4.47893918e-01 -5.28999686e-01 -3.67901593e-01
6.94451392e-01 -1.40809262e+00 -8.13664913e-01 5.32544196e-01
-8.17578673e-01 -4.60551918e-01 -3.03543597e-01 4.78655279e-01
-6.36438072e-01 7.61581898e-01 -6.63302958e-01 -6.65739179e-01
-4.30351079e-01 -1.01919031e+00 1.01633859e+00 -2.75921579e-02
-2.96714067e-01 -8.71114075e-01 -5.65981381e-02 4.93611038e-01
3.27277482e-01 1.94715317e-02 5.55301845e-01 -1.93851337e-01
-8.44116986e-01 -1.50384875e-02 -1.20464809e-01 2.48456672e-01
3.85512747e-02 -1.65596902e-01 -9.81707692e-01 -4.57294226e-01
-4.45070490e-02 -6.17627501e-01 1.25527608e+00 3.88660878e-01
1.51726222e+00 -5.11843443e-01 -3.85166109e-01 7.93384135e-01
1.30556703e+00 -3.23209256e-01 4.91853297e-01 4.21368569e-01
8.85966182e-01 3.13586026e-01 7.26528645e-01 8.47564638e-01
4.44832474e-01 7.63397276e-01 4.69229639e-01 -2.13813737e-01
-2.88079619e-01 -1.74106181e-01 5.27858198e-01 6.26044989e-01
-1.01713516e-01 -9.12336782e-02 -7.01551855e-01 6.79536402e-01
-2.05907202e+00 -1.20565605e+00 1.74785573e-02 2.21159530e+00
9.35227811e-01 2.01459572e-01 2.78282225e-01 2.69485921e-01
6.71898007e-01 4.49369162e-01 -6.30096078e-01 1.22555383e-02
-2.73950994e-01 1.89634785e-01 7.66552687e-01 2.43389785e-01
-1.55877209e+00 1.31230021e+00 6.22517300e+00 1.23064160e+00
-1.11893666e+00 2.28974104e-01 8.38015676e-01 -5.60952008e-01
9.65849459e-02 -6.03905991e-02 -9.46287692e-01 5.30469596e-01
7.93345153e-01 1.68580517e-01 3.65300387e-01 1.12116599e+00
4.10324782e-01 -4.27873880e-01 -1.05413735e+00 1.22838902e+00
1.39618739e-01 -1.83379471e+00 -4.12260331e-02 -1.57172933e-01
8.59954238e-01 2.15947077e-01 -7.35633448e-02 1.52629334e-02
5.25695235e-02 -6.90456092e-01 8.41576099e-01 2.28368072e-03
8.12728047e-01 -5.99244773e-01 3.83185416e-01 2.57616252e-01
-1.36030102e+00 -8.81463289e-02 -5.59200466e-01 -2.98931599e-01
-1.54369667e-01 7.05272853e-01 -8.36962223e-01 1.54169366e-01
8.86739135e-01 7.72826254e-01 -6.07505441e-01 7.69205332e-01
1.45055741e-01 8.95848036e-01 -4.82409358e-01 -6.61147684e-02
3.01910460e-01 2.64869690e-01 4.90028381e-01 1.52458107e+00
1.76685706e-01 -5.54713458e-02 3.00667286e-01 3.53788525e-01
2.97378562e-02 -6.79198056e-02 -2.91109532e-01 3.63939911e-01
6.05392098e-01 1.21084917e+00 -1.04156232e+00 -7.14321852e-01
-5.00252545e-01 1.13907480e+00 3.02334905e-01 3.78733687e-02
-9.99239564e-01 1.27922865e-02 6.92900836e-01 2.85581142e-01
6.34046614e-01 -4.06525940e-01 -3.28268409e-01 -1.37121665e+00
-6.03616089e-02 -8.33636880e-01 4.89084780e-01 -2.22482145e-01
-8.97736967e-01 5.35153508e-01 -2.27812320e-01 -1.43595445e+00
-2.51699746e-01 -3.42795789e-01 -4.08223599e-01 5.55485226e-02
-1.52489984e+00 -9.76341605e-01 -4.28572893e-01 8.80812168e-01
7.04153419e-01 -2.10767552e-01 6.29084349e-01 4.67065215e-01
-6.29147410e-01 8.55181694e-01 7.26592774e-03 2.00400781e-02
7.62895823e-01 -9.75953639e-01 4.16101038e-01 8.30511630e-01
2.47068837e-01 2.96878248e-01 5.04762471e-01 -6.52267158e-01
-1.78329384e+00 -1.30112016e+00 6.32670164e-01 1.80121548e-02
5.61905265e-01 -3.84843171e-01 -7.30600059e-01 5.39360583e-01
4.84243743e-02 4.78391349e-01 7.99658656e-01 -2.22755209e-01
-2.42664739e-01 -1.13324888e-01 -9.45361137e-01 6.83312297e-01
1.39568758e+00 -3.35552841e-01 -1.17995940e-01 6.63378835e-01
4.84553039e-01 -5.23228824e-01 -5.30692697e-01 2.39294276e-01
3.97106558e-01 -1.19469500e+00 1.02183032e+00 -2.83022285e-01
3.78487676e-01 -3.22793901e-01 -6.66366741e-02 -5.39933324e-01
-3.84257585e-01 -1.04840767e+00 -6.84978306e-01 1.00967097e+00
-3.86348809e-03 -2.67226666e-01 1.46490204e+00 2.51696110e-01
-1.46086095e-04 -7.62369454e-01 -1.21823168e+00 -7.99727380e-01
-4.88597423e-01 -6.26783609e-01 2.04551503e-01 9.99052465e-01
1.58984110e-01 6.18396141e-02 -8.36246550e-01 -5.69311194e-02
8.94124627e-01 -3.43856178e-02 8.49586904e-01 -7.03902960e-01
-2.98184842e-01 -1.12670071e-01 -6.47611260e-01 -1.25409901e+00
1.93673417e-01 -6.63708746e-01 -1.70811668e-01 -8.93379450e-01
3.07561725e-01 -5.86558342e-01 -7.63980076e-02 7.95272648e-01
-2.22153112e-01 7.58862734e-01 1.21548012e-01 4.22564328e-01
-1.05093324e+00 4.60977525e-01 9.95739043e-01 -1.39350751e-02
-1.61499023e-01 -1.42406240e-01 -2.68344581e-01 9.10991251e-01
5.97403705e-01 -5.82121015e-01 -2.64117390e-01 -6.25828087e-01
2.38204151e-02 -1.09060057e-01 5.33257544e-01 -1.37208879e+00
4.62209612e-01 1.38691351e-01 6.38899744e-01 -7.83945560e-01
3.89548749e-01 -5.33548832e-01 -7.79210255e-02 6.63023651e-01
-1.43955112e-01 1.02474779e-01 1.93786904e-01 4.81274098e-01
-2.23576650e-01 -1.10478386e-01 7.61787057e-01 -2.20113367e-01
-1.07217610e+00 3.83503705e-01 -6.99645102e-01 -3.71287540e-02
1.10493040e+00 -5.02961814e-01 -7.09997490e-02 -2.52610832e-01
-4.86123323e-01 1.24278709e-01 5.47344625e-01 2.15134010e-01
6.88230217e-01 -1.26963031e+00 -5.18236637e-01 3.35041225e-01
-2.43006408e-01 -3.77827249e-02 6.37642801e-01 7.03273416e-01
-7.43656456e-01 2.37086818e-01 -1.45507753e-01 -9.82440412e-01
-1.63760734e+00 5.22875905e-01 -1.98540129e-02 -2.65735179e-01
-8.15856040e-01 9.58982706e-01 3.04023251e-02 2.90287167e-01
4.65629876e-01 1.07761603e-02 1.17991373e-01 3.44775952e-02
6.74851418e-01 7.12689519e-01 3.29877660e-02 -5.20719230e-01
-5.57798028e-01 5.22902131e-01 -1.78448096e-01 1.85356677e-01
1.30575085e+00 -7.93796629e-02 -2.71333233e-02 8.32671002e-02
1.16881239e+00 -2.60625426e-02 -1.70655894e+00 -4.08296227e-01
-1.69892296e-01 -6.91857338e-01 3.63602936e-01 -1.82519257e-01
-1.29624701e+00 5.86008906e-01 6.98370993e-01 3.54150161e-02
1.26713586e+00 -1.99533850e-01 9.56294477e-01 4.55220431e-01
3.94653499e-01 -1.37184262e+00 3.50164503e-01 3.94676268e-01
1.23867519e-01 -1.34328628e+00 3.15670192e-01 -4.99864966e-01
-6.70644820e-01 1.18507659e+00 6.83911324e-01 -2.22757578e-01
4.51557606e-01 5.16975701e-01 -1.39127746e-01 -5.91743179e-02
-9.67680037e-01 -1.75271952e-03 6.23676591e-02 5.32852292e-01
3.40040386e-01 -1.24864161e-01 -1.07924260e-01 9.32017639e-02
-2.00263988e-02 2.32346877e-01 3.82634372e-01 1.15500331e+00
-4.61458117e-01 -9.49427009e-01 -3.12728971e-01 4.48678672e-01
-5.35086811e-01 -2.60401279e-01 2.17637509e-01 5.73562860e-01
2.68193185e-01 8.81259859e-01 3.90923738e-01 -5.73544621e-01
-2.97191769e-01 -1.61910519e-01 5.99764585e-01 -5.07832766e-01
-6.27424419e-01 2.06575409e-01 7.44397789e-02 -7.68999994e-01
-6.98470116e-01 -7.60154009e-01 -1.10907233e+00 -5.85398912e-01
-3.23801339e-01 -6.59994110e-02 4.33001399e-01 7.98349500e-01
7.32047439e-01 3.40833306e-01 7.85474241e-01 -1.29850423e+00
-4.14973706e-01 -6.96094930e-01 -2.36891344e-01 1.32738173e-01
2.66812116e-01 -6.88773215e-01 -2.46534348e-01 2.66445398e-01]
|
[9.063090324401855, 0.12572456896305084]
|
3ce4dd50-486b-4a6e-bb2a-96779fb07eda
|
explainable-outfit-recommendation-with-joint
|
1806.08977
| null |
http://arxiv.org/abs/1806.08977v3
|
http://arxiv.org/pdf/1806.08977v3.pdf
|
Explainable Outfit Recommendation with Joint Outfit Matching and Comment Generation
|
Most previous work on outfit recommendation focuses on designing visual
features to enhance recommendations. Existing work neglects user comments of
fashion items, which have been proved to be effective in generating
explanations along with better recommendation results. We propose a novel
neural network framework, neural outfit recommendation (NOR), that
simultaneously provides outfit recommendations and generates abstractive
comments. NOR consists of two parts: outfit matching and comment generation.
For outfit matching, we propose a convolutional neural network with a mutual
attention mechanism to extract visual features. The visual features are then
decoded into a rating score for the matching prediction. For abstractive
comment generation, we propose a gated recurrent neural network with a
cross-modality attention mechanism to transform visual features into a concise
sentence. The two parts are jointly trained based on a multi-task learning
framework in an end-to-end back-propagation paradigm. Extensive experiments
conducted on an existing dataset and a collected real-world dataset show NOR
achieves significant improvements over state-of-the-art baselines for outfit
recommendation. Meanwhile, our generated comments achieve impressive ROUGE and
BLEU scores in comparison to human-written comments. The generated comments can
be regarded as explanations for the recommendation results. We release the
dataset and code to facilitate future research.
|
['Zhaochun Ren', 'Jun Ma', 'Zhumin Chen', 'Yujie Lin', 'Pengjie Ren', 'Maarten de Rijke']
|
2018-06-23
| null | null | null | null |
['comment-generation']
|
['natural-language-processing']
|
[ 2.76215672e-01 3.46463844e-02 -3.85416418e-01 -5.96365631e-01
-7.67112970e-01 -3.23382378e-01 6.18316531e-01 -2.38084570e-01
3.78733426e-02 4.16783661e-01 9.47072566e-01 -5.49993634e-01
3.04085165e-01 -5.81537366e-01 -7.89918125e-01 -4.23571289e-01
5.49825907e-01 1.23510204e-01 -2.72294164e-01 -2.11189449e-01
4.70354915e-01 -1.09556459e-01 -1.41848743e+00 1.23479486e+00
8.01305532e-01 8.25868487e-01 4.57824051e-01 9.39531147e-01
-2.52863199e-01 7.55494177e-01 -6.32383943e-01 -8.25896502e-01
-1.02189757e-01 -4.71600145e-01 -4.58240092e-01 2.68300951e-01
4.40554023e-01 -6.42672777e-01 -4.70535159e-01 4.80439603e-01
6.35619223e-01 3.20676923e-01 9.90466774e-01 -9.72095072e-01
-1.93914104e+00 1.05026138e+00 -4.33499038e-01 -1.08008094e-01
3.28238934e-01 6.93027750e-02 1.39256680e+00 -1.35402250e+00
3.71085793e-01 1.29450786e+00 3.41442525e-01 7.29807258e-01
-8.55667055e-01 -6.20630383e-01 5.36787570e-01 9.84643996e-02
-6.06323421e-01 -1.93419874e-01 7.14528084e-01 -3.35927218e-01
1.11978257e+00 3.36339206e-01 4.19596583e-01 1.29254460e+00
1.61992684e-01 1.18695569e+00 3.42420816e-01 -1.25826702e-01
-2.34582633e-01 2.18926191e-01 2.76961848e-02 5.00217676e-01
8.15855525e-03 1.86361803e-03 -3.79425198e-01 3.45431179e-01
8.77968967e-01 7.20980763e-01 -3.04060847e-01 9.72596407e-02
-1.22355235e+00 1.21232831e+00 1.11034727e+00 -1.21110551e-01
-4.13708180e-01 1.88042969e-01 4.04394060e-01 1.55623227e-01
7.89744139e-01 3.36821854e-01 -2.04249904e-01 4.23959970e-01
-6.66887760e-01 2.25039586e-01 2.81976879e-01 7.95422733e-01
2.14906290e-01 2.64252752e-01 -9.33686733e-01 1.09221864e+00
6.82537794e-01 7.06853867e-01 5.78674138e-01 -3.82042378e-01
6.49569690e-01 5.71431518e-01 2.83253729e-01 -1.18742609e+00
-2.72583008e-01 -7.88200796e-01 -8.72568190e-01 -7.71015137e-02
-5.89048080e-02 -2.71730274e-01 -8.45834732e-01 1.30546856e+00
-1.72731593e-01 4.95133027e-02 2.02711254e-01 1.34025204e+00
1.68587863e+00 1.13378668e+00 6.70433491e-02 1.56779010e-02
1.18304110e+00 -1.82893252e+00 -6.61528468e-01 -2.14733794e-01
4.19913203e-01 -1.00421882e+00 1.46903920e+00 3.14532906e-01
-1.05240571e+00 -1.01029634e+00 -9.97785091e-01 -3.41074049e-01
-1.53240934e-01 9.12815928e-01 6.45165026e-01 1.47663131e-01
-9.34463322e-01 2.85307974e-01 -2.48340458e-01 -1.66412503e-01
3.83828968e-01 2.29990348e-01 -6.14102781e-02 -4.80687916e-02
-9.60987508e-01 4.49027628e-01 -9.75423753e-02 3.66968185e-01
-6.52889609e-01 -4.13614362e-01 -9.09883380e-01 5.53303584e-02
6.44484013e-02 -9.74635601e-01 1.43923092e+00 -1.25235116e+00
-1.41980600e+00 3.17984521e-01 -2.05041707e-01 -2.31203780e-01
1.29631639e-01 -5.12653887e-01 -6.21840596e-01 -2.08520800e-01
-5.35177626e-02 8.05811882e-01 9.27368283e-01 -1.33851516e+00
-4.29965228e-01 -4.29172777e-02 2.96538085e-01 3.36337656e-01
-6.73755586e-01 9.16211307e-02 -7.57916868e-01 -1.21839356e+00
-3.67662787e-01 -6.90613389e-01 -4.04229105e-01 -2.33646527e-01
-5.51630378e-01 -3.61563832e-01 5.39521217e-01 -8.30078423e-01
1.38830686e+00 -1.94372272e+00 6.60196617e-02 -1.06896773e-01
3.08476627e-01 2.80900151e-01 -6.52868986e-01 4.62066799e-01
-4.45264988e-02 3.62858316e-03 4.30121511e-01 -6.12080157e-01
2.09996060e-01 -1.10795461e-01 -7.70449817e-01 -6.15009665e-02
4.86464463e-02 1.38214374e+00 -8.51898491e-01 7.77365938e-02
2.15446770e-01 8.12820196e-01 -8.88310432e-01 6.24622762e-01
-4.46970612e-01 1.82577431e-01 -5.17340481e-01 1.89934880e-01
4.46766466e-01 -7.11177528e-01 -1.94903329e-01 -4.32620734e-01
1.32220760e-01 4.72183049e-01 -4.65551734e-01 1.60231054e+00
-6.69301987e-01 3.79361689e-01 -6.36337817e-01 -4.62894648e-01
1.22532880e+00 2.97741115e-01 -2.47265100e-01 -9.33705509e-01
1.86812937e-01 -6.19256087e-02 -4.40293103e-01 -6.28536165e-01
1.08582735e+00 5.56510687e-02 -4.62162681e-02 8.73364866e-01
-7.14314505e-02 4.07167852e-01 -5.98638579e-02 5.09552121e-01
6.55219078e-01 4.03974563e-01 -7.43325502e-02 3.84522438e-01
3.57311308e-01 -2.96539426e-01 1.12916254e-01 6.47996724e-01
5.03948331e-01 1.04188490e+00 1.08151302e-01 -4.80217844e-01
-1.05330980e+00 -8.41911912e-01 5.37561417e-01 1.34171391e+00
1.30173028e-01 -5.68771124e-01 -3.76732796e-01 -1.09702885e+00
-1.05337389e-01 9.13860023e-01 -8.04703534e-01 -2.70060152e-01
-2.34912917e-01 -3.34999889e-01 -1.08690381e-01 1.06056178e+00
8.68641287e-02 -1.53661132e+00 -5.92086427e-02 1.80703193e-01
-2.27111220e-01 -6.36043668e-01 -1.05783510e+00 -2.04128072e-01
-6.80552065e-01 -5.98006785e-01 -1.24795902e+00 -9.47336733e-01
1.00372231e+00 8.61757457e-01 1.19931138e+00 4.55104649e-01
2.52638876e-01 5.87804839e-02 -8.79462719e-01 -2.76370168e-01
-1.33489504e-01 -2.13619638e-02 -3.74271274e-01 -2.89457920e-03
3.15073758e-01 -2.43585527e-01 -9.72369373e-01 2.74185896e-01
-8.25230479e-01 6.47176564e-01 8.47487688e-01 1.03405452e+00
5.28254867e-01 -8.35229933e-01 8.05892825e-01 -1.22249949e+00
1.02868092e+00 -6.36423945e-01 5.45361452e-02 3.30817968e-01
-5.19783437e-01 9.79999080e-03 9.45240676e-01 -3.99537444e-01
-1.20531917e+00 -2.90924072e-01 -3.42248201e-01 -5.13909161e-01
-8.65767226e-02 7.24751234e-01 1.76859587e-01 5.23688614e-01
5.39943755e-01 4.87193763e-02 -4.18260753e-01 -5.78785241e-01
7.59184480e-01 9.83418763e-01 5.88619471e-01 -7.97830746e-02
6.03162110e-01 1.44776274e-02 -7.93021142e-01 -1.75125003e-01
-1.40852904e+00 -3.51646334e-01 -2.14756578e-01 -4.28629816e-01
7.81271636e-01 -9.96653974e-01 -4.97626692e-01 5.67043852e-03
-1.25420904e+00 -2.45739058e-01 5.35340793e-02 4.64924276e-01
-3.11480671e-01 1.56223416e-01 -7.89667189e-01 -6.92963958e-01
-1.00418365e+00 -1.07852519e+00 1.16811991e+00 4.55873370e-01
-8.97868499e-02 -9.62069154e-01 -1.08752966e-01 8.04733753e-01
5.39139211e-01 -2.71284521e-01 6.46373391e-01 -7.15201020e-01
-2.53920346e-01 -1.77484706e-01 -7.06261456e-01 1.32131666e-01
-8.26327428e-02 1.68802943e-02 -8.11750352e-01 -2.47734666e-01
-6.91272676e-01 -4.42698002e-01 1.32814598e+00 4.12973523e-01
1.46319294e+00 -7.23444700e-01 -1.72673717e-01 5.33866465e-01
1.14871371e+00 1.41887709e-01 6.76149011e-01 1.02556944e-01
1.08045769e+00 3.03036869e-01 7.18731225e-01 6.13281846e-01
7.32432306e-01 6.61881149e-01 6.04501963e-01 -5.08719623e-01
-3.95860344e-01 -6.50473535e-01 6.54888213e-01 1.12730205e+00
-1.57934606e-01 -6.95835173e-01 -1.30135417e-01 5.17873943e-01
-2.23211694e+00 -1.11084175e+00 -2.12685049e-01 1.80083251e+00
3.76905560e-01 -1.09342858e-02 2.25244105e-01 -2.45962813e-01
6.52798295e-01 1.39088526e-01 -6.16921604e-01 -8.02368283e-01
6.60847798e-02 -5.59863895e-02 -3.01544443e-02 3.11404139e-01
-9.27482545e-01 7.79289603e-01 5.51157951e+00 5.69484830e-01
-1.38662732e+00 -1.81963369e-02 6.93600893e-01 -2.16176644e-01
-1.09867775e+00 -3.70366603e-01 -6.10073984e-01 4.84837681e-01
7.37956405e-01 1.52021199e-01 2.37269148e-01 9.32624400e-01
4.20455784e-01 7.47469246e-01 -6.38054013e-01 8.88983369e-01
7.98466086e-01 -1.65497327e+00 6.10900283e-01 -1.19635865e-01
1.02718890e+00 -1.89988807e-01 4.57365036e-01 5.00164568e-01
4.21565384e-01 -1.16492200e+00 9.13839757e-01 6.78236783e-01
8.32936823e-01 -8.69185507e-01 1.00614822e+00 -8.24861377e-02
-1.15999854e+00 -1.82413831e-01 -7.31091201e-01 -1.67050034e-01
3.43847424e-01 5.02721190e-01 -7.88164079e-01 5.59274614e-01
4.47494566e-01 1.49362314e+00 -5.66434503e-01 1.05466604e+00
-5.86143196e-01 6.65284276e-01 5.66444278e-01 -4.37228531e-01
2.77877569e-01 -1.29506752e-01 5.97222485e-02 1.37857544e+00
5.52062631e-01 -3.71596850e-02 1.88817188e-01 7.68350780e-01
-4.24393475e-01 4.58819926e-01 -1.56188875e-01 -2.42297634e-01
3.22296917e-02 1.55496514e+00 -5.19028962e-01 -3.50272626e-01
-7.01040447e-01 1.27708948e+00 4.42839026e-01 5.93449354e-01
-9.39323783e-01 -4.84222829e-01 3.19084913e-01 -1.91013888e-01
8.22231412e-01 3.28721493e-01 -4.12488550e-01 -1.46064842e+00
-2.09632859e-01 -7.55216658e-01 3.17785084e-01 -1.46418595e+00
-1.53103769e+00 8.77417624e-01 -5.96739292e-01 -1.46680164e+00
-1.17324248e-01 -5.51604927e-01 -1.07073617e+00 7.28915036e-01
-1.57844138e+00 -1.67648649e+00 -3.85658085e-01 5.37811935e-01
1.14626968e+00 -2.99236000e-01 7.63892531e-01 3.95315200e-01
-6.73761368e-01 9.55809534e-01 4.02971692e-02 -7.23064393e-02
9.15389895e-01 -1.18067217e+00 6.01332843e-01 8.76350284e-01
3.10511380e-01 7.19679058e-01 5.77916801e-01 -5.91496587e-01
-1.16644168e+00 -1.43213630e+00 9.67530310e-01 -1.26743361e-01
3.20992559e-01 -2.02062964e-01 -5.61301827e-01 8.42139900e-01
4.70981121e-01 -2.50292718e-01 8.79360795e-01 2.31851220e-01
-5.86071312e-01 1.03565432e-01 -5.33813715e-01 7.79291451e-01
9.58659232e-01 -3.40294927e-01 -5.86297035e-01 2.47671694e-01
1.07404423e+00 -2.23984540e-01 -5.09043157e-01 1.21846534e-01
8.63833785e-01 -8.92530143e-01 9.42728162e-01 -8.30425799e-01
1.27757680e+00 -3.57947916e-01 5.80752753e-02 -1.58287597e+00
-7.39906847e-01 -4.40122575e-01 -1.77926719e-01 1.31926131e+00
9.32955742e-01 -4.50464822e-02 5.52760959e-01 2.40200296e-01
-6.15382731e-01 -9.92969036e-01 1.62787735e-01 3.39084528e-02
-2.60160238e-01 -4.41992313e-01 6.96474552e-01 5.72444499e-01
2.20101565e-01 1.03320873e+00 -1.04416025e+00 -2.40695864e-01
1.81007057e-01 7.45207548e-01 8.94644320e-01 -8.38252068e-01
-4.71087962e-01 -2.91945487e-01 2.19841927e-01 -1.70375621e+00
3.69035006e-02 -1.07784331e+00 1.56560108e-01 -2.34246564e+00
6.87491953e-01 -1.63882852e-01 -4.79881793e-01 6.03872955e-01
-5.22539735e-01 7.75422931e-01 3.66777539e-01 1.26847267e-01
-9.48472619e-01 7.11926579e-01 1.75491655e+00 -2.38998875e-01
-2.27543581e-02 1.00330293e-01 -1.41602659e+00 4.90769148e-01
5.72252572e-01 -1.27702668e-01 -5.62079549e-01 -8.13366711e-01
4.76022691e-01 1.12700485e-01 4.49527279e-02 -5.04311979e-01
-1.16881877e-01 -1.42543823e-01 7.22714186e-01 -8.36041331e-01
1.73408508e-01 -4.54684973e-01 -1.94716141e-01 1.16320744e-01
-1.04977775e+00 2.98985153e-01 -9.55608934e-02 6.69652104e-01
-7.36139044e-02 2.56575868e-02 1.28416762e-01 1.31199300e-01
-6.19473696e-01 4.49684054e-01 -1.72838435e-01 -5.53950667e-01
5.06295025e-01 -7.99100995e-02 -6.30780160e-01 -1.00742793e+00
-5.98296583e-01 1.66683272e-01 1.24724142e-01 1.01037276e+00
1.15717351e+00 -1.63354874e+00 -1.03864729e+00 7.83601776e-02
3.96609962e-01 -5.39752781e-01 5.23999929e-01 4.80384499e-01
-1.11194409e-01 5.28222501e-01 -1.66568235e-01 -1.25458017e-01
-1.32197070e+00 7.71742105e-01 -1.53396100e-01 -3.49225521e-01
-5.83140075e-01 1.10931587e+00 5.70339262e-01 -4.59805667e-01
2.71071553e-01 -3.49507183e-01 -8.75194788e-01 -1.21427588e-01
1.03873742e+00 -5.93386181e-02 -2.75389105e-01 -6.39989316e-01
2.01205298e-01 3.94162863e-01 -3.41882676e-01 2.67481625e-01
1.58802390e+00 -4.00934041e-01 3.11831266e-01 2.28465274e-01
1.10830939e+00 2.00109079e-01 -1.24995840e+00 -2.24736080e-01
-7.36771524e-01 -4.52915221e-01 4.62613776e-02 -1.14432633e+00
-1.49259698e+00 1.10519969e+00 1.34162888e-01 -6.63364073e-03
1.04711080e+00 5.86872287e-02 1.28050029e+00 1.29333377e-01
-2.87676096e-01 -5.22129536e-01 5.38383961e-01 4.25656915e-01
1.39512658e+00 -1.29108655e+00 -1.47958800e-01 -1.15911670e-01
-1.18057096e+00 1.31735075e+00 8.48762691e-01 -2.81076819e-01
4.23527032e-01 -1.35359585e-01 3.31307828e-01 -7.99166504e-03
-1.06828594e+00 -6.14459477e-02 9.69307721e-01 4.24047291e-01
9.71797168e-01 3.57800126e-02 -3.86831552e-01 1.39988101e+00
-1.48487508e-01 1.36479333e-01 3.65836561e-01 2.14116678e-01
-5.50855160e-01 -9.39801693e-01 -5.19612953e-02 7.23929822e-01
-3.14003110e-01 -2.97897518e-01 -3.92665237e-01 3.46269049e-02
-1.14921831e-01 1.21669018e+00 1.00052595e-01 -8.87718558e-01
3.37271601e-01 -3.98740739e-01 4.11136895e-02 -7.63644934e-01
-9.01473820e-01 2.77053773e-01 1.87284797e-01 -3.38767141e-01
-3.77325743e-01 -1.96484461e-01 -1.33090746e+00 -2.64394313e-01
-4.51598316e-01 2.06577495e-01 4.71781045e-01 9.92982090e-01
5.54772973e-01 9.47753310e-01 6.67186916e-01 -1.11244786e+00
-2.15290278e-01 -1.03237927e+00 -2.98909634e-01 7.28571117e-01
2.10844576e-01 -3.82903099e-01 -4.24700230e-02 2.44825974e-01]
|
[10.210273742675781, 5.624606132507324]
|
43d946ce-2179-4151-b0b2-01993c8cf4df
|
srn-side-output-residual-network-for-object
|
1807.06621
| null |
http://arxiv.org/abs/1807.06621v2
|
http://arxiv.org/pdf/1807.06621v2.pdf
|
SRN: Side-output Residual Network for Object Reflection Symmetry Detection and Beyond
|
In this paper, we establish a baseline for object reflection symmetry
detection in complex backgrounds by presenting a new benchmark and an
end-to-end deep learning approach, opening up a promising direction for
symmetry detection in the wild. The new benchmark, Sym-PASCAL, spans challenges
including object diversity, multi-objects, part-invisibility, and various
complex backgrounds that are far beyond those in existing datasets. The
end-to-end deep learning approach, referred to as a side-output residual
network (SRN), leverages the output residual units (RUs) to fit the errors
between the object ground-truth symmetry and the side-outputs of multiple
stages. By cascading RUs in a deep-to-shallow manner, SRN exploits the 'flow'
of errors among multiple stages to address the challenges of fitting complex
output with limited convolutional layers, suppressing the complex backgrounds,
and effectively matching object symmetry at different scales. SRN is further
upgraded to a multi-task side-output residual network (MT-SRN) for joint
symmetry and edge detection, demonstrating its generality to image-to-mask
learning tasks. Experimental results validate both the challenging aspects of
Sym-PASCAL benchmark related to real-world images and the state-of-the-art
performance of the proposed SRN approach.
|
['Guoying Zhao', 'Jie Chen', 'Jianbin Jiao', 'Qixiang Ye', 'Wei Ke']
|
2018-07-17
| null | null | null | null |
['symmetry-detection']
|
['computer-vision']
|
[ 5.22553980e-01 4.63276468e-02 1.90878823e-01 -4.67734843e-01
-7.73604751e-01 -4.10419732e-01 3.48583937e-01 -5.21368980e-01
-4.46910597e-02 4.18070331e-02 1.65188834e-01 -2.93419063e-02
7.18906820e-02 -4.98601705e-01 -1.09379709e+00 -4.67418253e-01
1.76333144e-01 1.03866227e-01 8.54486942e-01 -4.31715310e-01
1.67872950e-01 8.53660822e-01 -1.60215926e+00 8.63491058e-01
5.29612839e-01 1.29063261e+00 -2.24485859e-01 5.23245275e-01
2.67391920e-01 7.21227646e-01 -3.04606616e-01 -7.12459147e-01
9.03887749e-01 -2.45036826e-01 -3.58076543e-01 1.49040103e-01
1.35380876e+00 -1.80096090e-01 -4.46986914e-01 1.01298475e+00
8.08966041e-01 -9.85927582e-02 4.47397798e-01 -1.35830045e+00
-8.80978346e-01 1.06213219e-01 -1.03084850e+00 1.67695239e-01
-9.19435844e-02 4.47791457e-01 9.10208941e-01 -1.10994816e+00
8.22950423e-01 1.31693363e+00 1.29417634e+00 3.39219451e-01
-1.16668546e+00 -6.23915374e-01 3.15474212e-01 2.78881609e-01
-1.21712041e+00 -4.45324957e-01 7.58371472e-01 -4.40383792e-01
9.99032736e-01 1.81975901e-01 3.03781092e-01 1.08713019e+00
9.98422876e-02 1.12156808e+00 9.45467591e-01 -5.97555302e-02
-1.94843069e-01 -1.53020293e-01 -3.20370048e-02 5.37568629e-01
3.82103205e-01 1.26920491e-01 -5.96482694e-01 3.72566223e-01
7.72760749e-01 1.70428947e-01 -3.98844779e-01 -7.97736228e-01
-1.04232562e+00 3.48713815e-01 9.45681810e-01 -1.76162764e-01
-2.16581881e-01 -5.74450828e-02 7.84047365e-01 2.35203281e-01
3.59765351e-01 3.18365216e-01 -7.48028696e-01 4.54083353e-01
-9.30035770e-01 4.32690382e-01 3.83235604e-01 1.07458663e+00
4.30502892e-01 7.33953938e-02 -7.53826141e-01 8.65902960e-01
-7.23962486e-02 2.63115972e-01 -7.17188641e-02 -5.28712988e-01
8.24985802e-01 8.84322584e-01 -1.81516841e-01 -9.02860284e-01
-6.25966132e-01 -1.05585134e+00 -9.56379473e-01 6.31012201e-01
3.19150627e-01 2.66749740e-01 -1.06900120e+00 1.52459538e+00
5.55522263e-01 3.45894992e-01 -5.18201012e-03 1.03450096e+00
1.17405808e+00 1.94916546e-01 -4.79056448e-01 5.07596254e-01
1.35425472e+00 -1.53692162e+00 -1.17347516e-01 -3.86899173e-01
4.05044377e-01 -8.64097416e-01 9.75736439e-01 2.43877441e-01
-1.18881369e+00 -8.06769431e-01 -9.01698112e-01 -4.72430319e-01
-3.90265346e-01 4.22892153e-01 4.40459758e-01 3.89672071e-01
-1.13516963e+00 4.87227827e-01 -5.14421165e-01 -3.58606353e-02
8.53718936e-01 2.69964576e-01 -5.07913530e-01 -1.77971393e-01
-6.69684470e-01 7.04243839e-01 2.26505995e-01 5.29672086e-01
-5.35115540e-01 -1.12526691e+00 -1.10877883e+00 1.72099948e-01
6.32997572e-01 -9.84263599e-01 1.08637822e+00 -9.24798608e-01
-1.24988163e+00 1.04952896e+00 1.29188612e-01 -3.97121012e-01
1.10465944e+00 -3.11157763e-01 -2.58709610e-01 -6.69723330e-03
2.14742854e-01 7.89206088e-01 8.33093464e-01 -1.14215064e+00
-6.59615576e-01 -4.81923282e-01 -2.03243434e-01 1.39842212e-01
3.52350771e-02 3.76523972e-01 -7.30986953e-01 -7.35376894e-01
3.55851829e-01 -7.07630694e-01 4.84680384e-02 5.16904533e-01
-4.54903036e-01 -1.18571535e-01 1.06020284e+00 -7.33028412e-01
7.49640226e-01 -2.26568055e+00 -2.99416184e-02 -1.11393921e-01
1.53089225e-01 4.49630231e-01 -4.84157413e-01 8.26374143e-02
-5.25410712e-01 -3.33785772e-01 -2.33391911e-01 -5.59872866e-01
2.38016948e-01 -2.33658865e-01 -8.29236209e-02 5.44608533e-01
7.78365850e-01 1.32235229e+00 -5.50765276e-01 -2.60202289e-01
2.06811056e-01 4.82982308e-01 -7.11782575e-01 8.46842453e-02
1.01458952e-01 2.06581950e-01 1.63229242e-01 1.05115938e+00
1.24454963e+00 -2.03822806e-01 -5.03298402e-01 -6.04136586e-01
-3.07715666e-02 1.72367528e-01 -1.54694808e+00 1.39575267e+00
-2.26370454e-01 7.82496035e-01 2.44600311e-01 -7.36975551e-01
8.06221545e-01 -2.13007361e-01 3.10638458e-01 -1.06154776e+00
-2.02675283e-01 3.02574456e-01 1.75071016e-01 -5.04557431e-01
3.39812040e-01 4.16926056e-01 3.34925689e-02 -2.12614723e-02
1.61468133e-01 -1.82516277e-02 2.53555983e-01 -4.33184579e-02
9.91921127e-01 4.50829297e-01 -8.80651698e-02 -1.24469168e-01
5.81524611e-01 -4.86785889e-01 8.70540619e-01 7.29515314e-01
-2.69323587e-01 1.30110431e+00 5.20276010e-01 -7.60071695e-01
-1.23002911e+00 -1.22918522e+00 -2.70893574e-01 1.11980522e+00
2.00580448e-01 -1.66528344e-01 -4.81924534e-01 -7.79698730e-01
2.21559182e-01 2.02945471e-01 -8.33865047e-01 -1.63385957e-01
-7.73620903e-01 -6.33084059e-01 6.02643907e-01 9.07818913e-01
9.67765629e-01 -9.88640487e-01 -6.85203731e-01 -7.56171122e-02
-6.21100701e-02 -1.60124123e+00 -7.58137286e-01 1.96558997e-01
-4.88865286e-01 -1.17117763e+00 -8.15564156e-01 -7.91126370e-01
6.41230643e-01 4.63751405e-01 1.33949471e+00 -1.75353110e-01
-7.55575657e-01 8.95951912e-02 9.44159739e-03 -3.14034015e-01
-1.01442397e-01 -3.03571731e-01 -2.75659949e-01 3.50832075e-01
-4.41737436e-02 -4.71780807e-01 -1.01389277e+00 7.19850123e-01
-8.67942274e-01 2.62927979e-01 9.15290415e-01 8.66236985e-01
4.29709256e-01 7.96317868e-03 2.91074216e-01 -5.64480186e-01
-1.78052202e-01 5.53633571e-02 -8.37082624e-01 3.93589377e-01
-2.19733804e-01 -6.06709607e-02 3.45371157e-01 -3.86767566e-01
-1.06101561e+00 1.46614909e-01 -1.89877570e-01 -6.76566422e-01
1.25254458e-02 -3.02485287e-01 -5.04524946e-01 -5.62845111e-01
6.98648989e-01 2.45075032e-01 -1.66036367e-01 -3.39719027e-01
1.38304979e-01 4.46551114e-01 1.02646124e+00 -2.00770557e-01
9.38084602e-01 5.62749743e-01 2.18187034e-01 -4.30370718e-01
-1.02942467e+00 -7.90258706e-01 -6.45349622e-01 -4.95031849e-02
7.67480731e-01 -1.20812964e+00 -5.49846411e-01 9.35706973e-01
-1.09098876e+00 -3.88244241e-01 -4.70936507e-01 1.35141522e-01
-4.44237292e-01 4.45615679e-01 -4.73777205e-01 -4.92638826e-01
-5.31658590e-01 -1.07268989e+00 1.70649004e+00 4.42324251e-01
1.94864884e-01 -4.61751550e-01 -4.75313395e-01 4.99498874e-01
4.83337164e-01 3.55019718e-01 5.70082247e-01 -6.05881929e-01
-8.31651390e-01 -2.34295115e-01 -9.40986216e-01 5.08226812e-01
-1.97802633e-01 -7.06432911e-04 -9.73130643e-01 -2.54548550e-01
-5.19575663e-02 -5.02422392e-01 1.28301704e+00 2.29683846e-01
1.17165267e+00 -1.33072417e-02 1.26306280e-01 1.25091231e+00
1.37039924e+00 -5.67146957e-01 9.01847422e-01 6.12631917e-01
1.05103314e+00 5.14532208e-01 4.02026415e-01 6.15365468e-02
3.82626921e-01 8.74928534e-01 5.58308959e-01 -7.74195075e-01
-7.32092738e-01 -2.00239886e-02 3.49079639e-01 -1.27996236e-01
3.99384163e-02 2.92433370e-02 -7.17585385e-01 3.00881296e-01
-1.86401141e+00 -1.05697298e+00 -4.98249441e-01 2.11857343e+00
4.28919226e-01 4.58182305e-01 2.26050586e-01 6.87178820e-02
7.87099242e-01 1.30447417e-01 -6.93590879e-01 -2.54027456e-01
-5.55927694e-01 -1.29919518e-02 6.97761178e-01 -6.71151206e-02
-1.40904450e+00 7.23242521e-01 5.54968166e+00 7.99735844e-01
-1.21214151e+00 -7.12767318e-02 6.86030269e-01 1.28521249e-01
1.20977819e-01 -2.28496522e-01 -1.18800986e+00 2.91224003e-01
2.11560354e-01 4.00171965e-01 8.07405114e-02 1.05450332e+00
8.46044645e-02 1.04155287e-01 -1.13677537e+00 1.15466130e+00
4.19133753e-01 -1.47492087e+00 -6.68107048e-02 -3.11995834e-01
1.09977043e+00 3.88317227e-01 1.44752219e-01 4.22911912e-01
-7.76301399e-02 -7.64003336e-01 9.93738711e-01 1.14667237e-01
7.81151354e-01 -4.81988996e-01 8.56663525e-01 1.88478544e-01
-1.42184949e+00 -4.56858993e-01 -3.30486268e-01 2.62272954e-01
8.60769451e-02 4.28041577e-01 -8.12428057e-01 8.67316186e-01
1.14324081e+00 7.88593113e-01 -1.03674436e+00 1.44230223e+00
-9.48263630e-02 8.23164545e-03 -4.25126821e-01 4.05946791e-01
2.06013411e-01 2.38819104e-02 5.63760757e-01 1.78074825e+00
1.47725604e-02 -4.50525492e-01 1.33220553e-01 1.15441251e+00
-2.09849760e-01 -1.75309226e-01 -2.65956640e-01 3.63515317e-01
-1.07502431e-01 1.37935627e+00 -9.88731444e-01 -2.41972208e-01
-5.10269523e-01 1.24084830e+00 3.89056593e-01 3.36756200e-01
-9.55580533e-01 -3.11755687e-01 5.14702260e-01 3.28613758e-01
6.56719685e-01 1.50479138e-01 -4.47698832e-01 -1.23152471e+00
6.07626915e-01 -1.01796079e+00 4.71077293e-01 -8.43348444e-01
-1.58573163e+00 3.52032483e-01 -8.79399925e-02 -1.31675875e+00
3.28423440e-01 -1.06500423e+00 -9.98995781e-01 8.82774353e-01
-1.87051940e+00 -1.71605277e+00 -6.19930148e-01 5.57616591e-01
7.43976593e-01 -1.42065644e-01 2.32178316e-01 4.45459247e-01
-5.85207283e-01 8.88254106e-01 -1.09005533e-01 4.48487133e-01
8.02788675e-01 -1.19549978e+00 9.62819695e-01 1.32967818e+00
-1.37173846e-01 1.51487365e-01 3.73976022e-01 -5.30396521e-01
-1.25555122e+00 -1.43209362e+00 6.73807025e-01 -4.38854545e-01
4.37031180e-01 -8.28893006e-01 -8.25475574e-01 5.41154861e-01
-1.88055888e-01 7.37115145e-01 1.28957316e-01 -2.07414314e-01
-8.38474333e-01 -2.39895806e-01 -8.76217484e-01 6.86076224e-01
1.68137050e+00 -4.21105295e-01 -3.16855013e-01 5.15638113e-01
6.44038737e-01 -9.22744572e-01 -3.17378074e-01 1.25854445e+00
4.29971725e-01 -1.47780359e+00 1.28440225e+00 -6.35057926e-01
6.17228687e-01 -4.95096296e-01 -5.29809203e-03 -8.26904535e-01
-5.65555036e-01 -7.84065187e-01 1.30694374e-01 1.12102401e+00
3.40609431e-01 -5.18273711e-01 9.54337120e-01 3.39200974e-01
-7.00926781e-01 -9.03288782e-01 -8.61771226e-01 -9.93837059e-01
-2.64597356e-01 -2.90671557e-01 4.04282302e-01 6.98273480e-01
-1.14162636e+00 2.03248739e-01 -2.27256298e-01 4.74888682e-01
6.58771157e-01 3.57189447e-01 1.14927089e+00 -1.01202428e+00
-5.88586390e-01 -8.20865154e-01 -9.87709701e-01 -1.13047600e+00
-1.07810959e-01 -9.48147297e-01 6.06783330e-02 -1.44470954e+00
2.08576411e-01 -1.05967540e-02 -1.61376402e-01 3.38119894e-01
-5.19411683e-01 7.40089893e-01 3.13434124e-01 -9.00676847e-02
-8.90769541e-01 5.58900237e-01 1.39703834e+00 -1.68487683e-01
-8.55503082e-02 3.86281192e-01 -6.90971076e-01 8.22552085e-01
3.10183197e-01 -2.89022624e-01 -1.05400518e-01 -4.05600399e-01
3.55765373e-01 -6.47712708e-01 8.76232743e-01 -1.10761166e+00
2.95752436e-01 2.73124337e-01 9.31281865e-01 -9.79528427e-01
2.33773082e-01 -7.57037044e-01 -1.09373391e-01 3.39772433e-01
-5.33361584e-02 1.18009567e-01 3.68810266e-01 4.91652459e-01
-1.40592009e-01 1.66343480e-01 1.08650875e+00 9.74163637e-02
-8.21085513e-01 4.18421119e-01 4.58090007e-01 4.24654335e-01
9.99489009e-01 -8.06649029e-01 -3.01512420e-01 1.11892723e-01
-5.87824583e-01 5.59236228e-01 5.83405554e-01 6.00083292e-01
6.84761226e-01 -1.11597383e+00 -1.11272204e+00 6.34968042e-01
3.81442577e-01 4.53403860e-01 5.69709420e-01 1.05288696e+00
-6.94169223e-01 1.07578533e-02 -3.77778232e-01 -1.12747097e+00
-1.37520885e+00 4.53187943e-01 7.71961391e-01 -2.42072374e-01
-1.06697297e+00 1.25234318e+00 6.27943099e-01 -7.37731457e-01
6.45033181e-01 -4.48995769e-01 7.05702752e-02 -2.91594028e-01
5.20043254e-01 4.16846693e-01 5.35059333e-01 -4.69664782e-01
-3.42865288e-01 8.59438241e-01 -5.25683211e-03 3.52320731e-01
1.46405804e+00 1.05598226e-01 -4.07770881e-03 -1.45735946e-02
1.28940582e+00 9.60789621e-02 -1.74343801e+00 -4.18969363e-01
-1.46126613e-01 -5.26735902e-01 -2.67837822e-01 -7.99060225e-01
-1.12844980e+00 7.75332212e-01 6.75735652e-01 -3.29619944e-01
1.21318161e+00 -2.64090300e-01 5.95198810e-01 3.78141165e-01
-2.13827997e-01 -1.13621247e+00 4.14087206e-01 4.87454742e-01
1.16531873e+00 -1.55675006e+00 -1.24396876e-01 -7.64672995e-01
-6.90779626e-01 1.25806570e+00 1.00027072e+00 -2.17595667e-01
3.93866003e-01 5.36417246e-01 8.84724930e-02 -7.56779015e-02
-5.04578888e-01 -2.49607831e-01 7.20554352e-01 5.49971223e-01
4.84966822e-02 -4.88851339e-01 3.43621731e-01 3.96948844e-01
6.30585302e-04 -1.45805731e-01 2.01460585e-01 6.42979801e-01
-3.61326747e-02 -6.52980804e-01 -4.31438863e-01 1.74570814e-01
-3.21854323e-01 -6.97776228e-02 -6.78803205e-01 7.52417088e-01
4.16765988e-01 4.84611571e-01 7.43664280e-02 1.55543014e-01
8.44232976e-01 -6.98034242e-02 4.73635912e-01 -5.35884082e-01
-8.60653460e-01 -8.39256793e-02 5.35722589e-03 -9.38156605e-01
-7.40215331e-02 -4.96427596e-01 -9.96641219e-01 1.06358007e-01
-3.16589236e-01 -5.65855443e-01 5.15323639e-01 5.50358176e-01
6.17478609e-01 8.72556508e-01 6.52787089e-01 -1.44100356e+00
-9.20642734e-01 -8.72711539e-01 -2.04464689e-01 7.89792597e-01
4.49357212e-01 -4.84187663e-01 -3.55048984e-01 -2.14024305e-01]
|
[8.619140625, -1.7410895824432373]
|
a1a77b3f-6208-4925-b317-29b0f4ec366c
|
only-pay-for-what-is-uncertain-variance
|
2303.09033
| null |
https://arxiv.org/abs/2303.09033v1
|
https://arxiv.org/pdf/2303.09033v1.pdf
|
Only Pay for What Is Uncertain: Variance-Adaptive Thompson Sampling
|
Most bandit algorithms assume that the reward variance or its upper bound is known. While variance overestimation is usually safe and sound, it increases regret. On the other hand, an underestimated variance may lead to linear regret due to committing early to a suboptimal arm. This motivated prior works on variance-aware frequentist algorithms. We lay foundations for the Bayesian setting. In particular, we study multi-armed bandits with known and \emph{unknown heterogeneous reward variances}, and develop Thompson sampling algorithms for both and bound their Bayes regret. Our regret bounds decrease with lower reward variances, which make learning easier. The bound for unknown reward variances captures the effect of the prior on learning reward variances and is the first of its kind. Our experiments show the superiority of variance-aware Bayesian algorithms and also highlight their robustness.
|
['Branislav Kveton', 'Aadirupa Saha']
|
2023-03-16
| null | null | null | null |
['thompson-sampling', 'multi-armed-bandits']
|
['methodology', 'miscellaneous']
|
[ 3.80321927e-02 2.68202901e-01 -9.43130672e-01 -2.91579574e-01
-1.09150755e+00 -7.05687523e-01 2.44222820e-01 -4.01890911e-02
-3.96916330e-01 1.26808345e+00 1.47121295e-01 -5.94569743e-01
-8.56103301e-01 -6.87285900e-01 -9.73368704e-01 -8.67259800e-01
-9.33898613e-03 7.58030117e-01 -1.14265837e-01 2.93448865e-01
2.83357650e-01 1.75907016e-01 -1.12562001e+00 -1.33917153e-01
7.32509017e-01 1.20614648e+00 4.49050181e-02 5.06625175e-01
-1.00037433e-01 7.77122378e-01 -4.12937731e-01 -6.52746141e-01
5.19935906e-01 -4.64556158e-01 -5.80832005e-01 -7.59742409e-02
-6.43845126e-02 -7.97221541e-01 -1.97945431e-01 1.14242232e+00
2.74712265e-01 1.88735113e-01 6.30245447e-01 -1.12082756e+00
-3.84562582e-01 1.52959740e+00 -1.08687806e+00 2.64039308e-01
-8.03746358e-02 -1.79491222e-01 1.24044156e+00 2.27884829e-01
2.24355981e-01 1.15849888e+00 5.08202314e-01 5.14042497e-01
-1.15496397e+00 -5.89510441e-01 6.71928167e-01 1.06230617e-01
-8.71059358e-01 -3.61150473e-01 5.40795147e-01 -3.11070532e-01
3.53946984e-01 4.44878221e-01 7.33377397e-01 1.01095438e+00
-2.08814815e-01 1.33036208e+00 1.17649198e+00 -3.83421242e-01
5.59552908e-01 7.41806924e-02 5.42417884e-01 8.97110999e-02
1.11307359e+00 6.21283948e-01 -4.21096325e-01 -3.07261646e-01
6.53549850e-01 2.52085567e-01 -3.36000443e-01 -5.31154811e-01
-5.18815458e-01 9.15301144e-01 5.66717088e-02 -2.32791737e-01
-5.87748885e-01 8.09707761e-01 1.69461012e-01 3.69928360e-01
3.92344892e-01 1.02552392e-01 -3.64493757e-01 -5.70085168e-01
-1.03927672e+00 3.23489070e-01 8.69177461e-01 1.05521774e+00
4.00712401e-01 -2.06759557e-01 -7.00083971e-01 5.91608942e-01
1.79194108e-01 6.09707832e-01 -6.72104657e-02 -1.01937032e+00
7.11014509e-01 -2.50961810e-01 9.59166288e-01 -3.46333951e-01
-3.37021381e-01 -6.86644971e-01 -3.17109764e-01 -3.98601815e-02
9.52311039e-01 -4.13714170e-01 -6.71826899e-01 1.86105382e+00
2.75640965e-01 -1.43627807e-01 -1.73908114e-01 9.78417218e-01
-1.82214245e-01 3.25865299e-01 -3.42638075e-01 -8.19143593e-01
9.41819251e-01 -8.94701898e-01 -9.19924736e-01 -3.82565200e-01
4.18760955e-01 -4.20657992e-01 6.75450087e-01 6.30279541e-01
-1.17477274e+00 4.61604029e-01 -6.58746839e-01 5.99016607e-01
4.96645272e-01 -1.54416353e-01 1.09946799e+00 1.37437344e+00
-3.97870898e-01 7.73160040e-01 -1.01081336e+00 8.27361047e-02
7.40481675e-01 4.99409400e-02 5.53898454e-01 -2.12572962e-01
-5.61197758e-01 5.89987576e-01 1.57382473e-01 1.37488395e-01
-9.98191953e-01 -8.17943037e-01 -1.70245826e-01 1.77553102e-01
9.97836947e-01 -6.05776429e-01 1.80938721e+00 -1.06486666e+00
-1.73184299e+00 5.22326566e-02 -9.84878764e-02 -8.67827654e-01
1.18598425e+00 -6.80173457e-01 4.06087041e-01 -2.43062124e-01
-1.55321062e-01 -3.59252274e-01 7.97810316e-01 -1.01371336e+00
-6.88907027e-01 -4.23066676e-01 2.65144914e-01 1.25063509e-01
-1.57586426e-01 -3.04224342e-01 4.32754345e-02 -3.23740333e-01
-4.15279157e-02 -9.38199103e-01 -3.65623474e-01 -4.29054320e-01
-5.88479638e-01 9.53425020e-02 -1.27578184e-01 -9.57606062e-02
1.25758219e+00 -2.04063320e+00 -3.26487839e-01 3.98827016e-01
-4.41433974e-02 -5.48049271e-01 9.10204202e-02 4.18675095e-01
3.01659763e-01 1.70780554e-01 1.68731168e-01 1.10431187e-01
3.52992862e-01 1.55646548e-01 -6.04828000e-01 7.17400432e-01
-5.36400855e-01 5.96925795e-01 -9.05544043e-01 1.79658551e-02
-1.27116352e-01 -2.41512790e-01 -6.66334450e-01 -4.53198738e-02
-2.87924737e-01 -2.56888252e-02 -7.43291676e-01 3.58185798e-01
7.31831789e-01 -3.63319576e-01 5.28537154e-01 2.53851563e-01
-9.83009953e-03 3.88504684e-01 -1.35782540e+00 9.32262361e-01
-4.32285935e-01 1.50136203e-01 1.31066933e-01 -9.00301456e-01
2.24191651e-01 -3.29235829e-02 3.79579514e-01 -3.20109427e-01
3.15841407e-01 3.12883914e-01 1.12137824e-01 -1.22420192e-01
3.54630291e-01 -3.72680515e-01 1.22092597e-01 9.50292587e-01
-5.25939822e-01 5.68647385e-02 3.05067688e-01 1.92009568e-01
8.87877882e-01 1.04607575e-01 3.17107499e-01 -3.38639081e-01
-4.32453007e-01 -3.45369786e-01 5.80993116e-01 1.72860920e+00
-6.73169345e-02 2.39342168e-01 9.61529553e-01 -2.19332561e-01
-8.26830685e-01 -9.49701130e-01 -1.56536907e-01 1.20717621e+00
1.20195679e-01 1.14876137e-03 -3.02123547e-01 -7.17218280e-01
5.06605744e-01 1.07858980e+00 -9.00625169e-01 7.13693351e-02
2.22605944e-01 -1.06658340e+00 -9.03708301e-03 5.88695884e-01
8.77153948e-02 -1.14284918e-01 -7.84380376e-01 2.95850307e-01
3.13825272e-02 -5.67381501e-01 -4.30724353e-01 3.55117500e-01
-9.91673052e-01 -1.19077706e+00 -8.31513584e-01 3.82176071e-01
4.65276778e-01 2.57460535e-01 8.76259625e-01 -3.91028374e-01
2.35882849e-01 5.98912001e-01 -3.61869633e-01 -9.50513363e-01
1.46942347e-01 -7.70217925e-02 1.30332017e-03 -1.83592871e-01
3.17008942e-02 -4.71703857e-01 -6.45774841e-01 1.96973622e-01
-4.70060319e-01 -1.97150841e-01 6.12952054e-01 8.76372337e-01
4.44511503e-01 -1.31500289e-01 6.54116929e-01 -1.12515998e+00
7.19538689e-01 -4.08563823e-01 -1.09413910e+00 4.48538721e-01
-7.76825488e-01 4.05309498e-01 1.60753563e-01 -6.59554839e-01
-1.31409395e+00 -9.11592171e-02 4.82870579e-01 -7.46807754e-02
4.73059744e-01 4.59349930e-01 2.61966974e-01 3.59493524e-01
6.07439280e-01 -8.15095156e-02 -2.73915827e-01 -4.03173655e-01
4.83668894e-01 4.13671553e-01 -6.84433058e-02 -1.20401633e+00
3.44832361e-01 4.49128032e-01 1.58672929e-01 -2.47271791e-01
-1.53691649e+00 7.23031610e-02 3.77655238e-01 -1.82791010e-01
-1.49495052e-02 -6.43978119e-01 -1.05132020e+00 4.10447689e-03
-7.12141871e-01 -5.86851716e-01 -6.09303534e-01 9.85296607e-01
-1.03441477e+00 3.13325137e-01 -4.01553720e-01 -1.97020304e+00
-1.16684131e-01 -8.92872751e-01 4.08230126e-01 2.46293485e-01
4.85848449e-02 -6.19835973e-01 -2.82161892e-03 2.90871382e-01
4.03394401e-01 -2.43928388e-01 5.02880096e-01 -6.11745298e-01
-7.25731790e-01 -1.51708707e-01 -2.31687933e-01 -9.39057320e-02
-3.76944579e-02 -1.03192799e-01 -6.35730922e-01 -2.34073386e-01
-2.06173062e-01 -4.40776855e-01 1.14826798e+00 1.16202748e+00
1.16247022e+00 -7.12414026e-01 -3.82234663e-01 2.06435740e-01
1.19183648e+00 1.99019328e-01 2.47415438e-01 6.09348834e-01
-6.69540837e-02 3.48540932e-01 7.40006506e-01 1.22731769e+00
1.33942246e-01 4.53541666e-01 4.00473386e-01 6.43201232e-01
4.91092533e-01 -7.41655976e-02 2.87082374e-01 -1.27856255e-01
-3.63755107e-01 -1.94159210e-01 -5.35798788e-01 5.50316453e-01
-2.20032120e+00 -1.20342672e+00 -1.53244147e-02 2.68586254e+00
1.25563312e+00 2.84692079e-01 5.79254627e-01 -3.52133065e-02
6.90166831e-01 -8.80110338e-02 -8.51132095e-01 -4.94152725e-01
6.66569546e-02 -9.75174382e-02 1.15389609e+00 5.54080606e-01
-7.35319376e-01 5.44056535e-01 6.90938902e+00 9.72188950e-01
-4.85328823e-01 2.04958826e-01 7.37624645e-01 -1.12888467e+00
-6.07294798e-01 2.20730454e-01 -1.01811242e+00 5.40141344e-01
8.27870011e-01 -5.78775883e-01 7.62938321e-01 1.14756739e+00
1.18474767e-01 -5.26531696e-01 -1.29001999e+00 8.44113886e-01
-5.25290668e-01 -1.19198012e+00 -4.62264270e-01 1.96394905e-01
9.93671477e-01 -1.02841139e-01 9.78239700e-02 3.40374649e-01
1.26039588e+00 -7.05912709e-01 1.11620212e+00 8.20684254e-01
3.95776927e-01 -1.19794571e+00 7.73811281e-01 5.28273225e-01
-4.76359248e-01 -5.15675426e-01 -4.41782862e-01 -4.27677602e-01
1.79911837e-01 1.17044854e+00 -4.79602009e-01 4.46304262e-01
5.48420787e-01 3.12157661e-01 2.15915471e-01 1.41959465e+00
-4.39882785e-01 7.80492783e-01 -8.79704535e-01 -4.89557445e-01
2.83707589e-01 -3.69237512e-01 2.95080274e-01 7.71364570e-01
3.49739939e-01 3.86015996e-02 5.08248471e-02 6.91742599e-01
-1.46021377e-02 -1.00961834e-01 -1.24700695e-01 -3.20109189e-01
8.45450640e-01 8.13463628e-01 -6.30324185e-01 -2.30515495e-01
-2.10988045e-01 3.04143667e-01 3.43652815e-01 4.58892733e-01
-8.63897979e-01 -1.13973536e-01 6.05884135e-01 -1.63148478e-01
5.91496110e-01 2.29528442e-01 -5.68428159e-01 -9.50401723e-01
5.17318137e-02 -3.88014585e-01 6.01334333e-01 -1.76880881e-01
-1.38943732e+00 -2.79218823e-01 1.88200638e-01 -6.58210635e-01
-2.74632365e-01 -1.99766845e-01 -1.94280982e-01 4.54862654e-01
-1.54990804e+00 -4.44902450e-01 3.13929439e-01 1.89361617e-01
2.41067350e-01 3.90419871e-01 2.19571874e-01 -1.37312070e-01
-7.49948621e-01 8.10675800e-01 8.85863543e-01 -4.64296758e-01
3.94088179e-01 -1.11255050e+00 -3.58473748e-01 5.69724619e-01
-7.99190626e-02 6.72673643e-01 1.07091939e+00 -5.99113226e-01
-1.48626602e+00 -6.87176347e-01 -9.41640977e-03 -3.04275006e-01
9.65743542e-01 7.53370821e-02 -1.35071889e-01 9.46363032e-01
-2.18336433e-01 -2.69131362e-01 6.35199964e-01 7.13923156e-01
-4.96761620e-01 -2.16824844e-01 -1.06067467e+00 6.11217797e-01
1.08742213e+00 -2.99152806e-02 -2.29741111e-01 5.00552118e-01
3.81518543e-01 -3.65973204e-01 -6.43019497e-01 1.38640299e-01
1.03153563e+00 -1.10334361e+00 6.22850358e-01 -7.59237468e-01
2.18409523e-01 2.84640163e-01 -2.50327647e-01 -1.20155787e+00
-1.92775086e-01 -1.04027843e+00 -4.47631270e-01 1.11515391e+00
5.37581801e-01 -7.15736449e-01 9.76030409e-01 8.65079761e-01
2.42765278e-01 -6.95320427e-01 -1.04104578e+00 -1.36155009e+00
2.45742843e-01 -8.87259364e-01 6.34356499e-01 5.11859715e-01
9.84696150e-02 -1.17975734e-01 -7.12412357e-01 -5.96389323e-02
9.60483611e-01 6.50641799e-01 6.69151366e-01 -1.09419918e+00
-8.02168250e-01 -7.92210460e-01 4.25129294e-01 -1.44518352e+00
-1.42637312e-01 -2.98698336e-01 4.98318821e-02 -1.37425005e+00
7.42587388e-01 -6.33697808e-01 -3.70385885e-01 4.26166594e-01
-9.99567583e-02 -3.69481623e-01 2.08377346e-01 3.36751230e-02
-8.99218500e-01 5.16897500e-01 1.01916230e+00 -9.71779823e-02
-2.96880931e-01 6.12391412e-01 -1.13206422e+00 5.68775237e-01
8.54560256e-01 -7.84271419e-01 -4.38353479e-01 -2.20664009e-01
8.15004647e-01 3.20682198e-01 7.46873617e-02 -2.74322122e-01
-6.20566159e-02 -8.29289496e-01 1.01150058e-01 -6.61216617e-01
-1.00538437e-03 -6.85157895e-01 2.09225506e-01 5.02009392e-01
-7.11054385e-01 -7.44517863e-01 -1.09871224e-01 1.16065013e+00
5.88440061e-01 -6.12489581e-01 5.71158290e-01 -1.12209365e-01
3.00398111e-01 1.17201000e-01 -3.46909553e-01 2.27986589e-01
8.97192955e-01 -2.24251915e-02 -3.89238894e-01 -8.74193668e-01
-5.98851860e-01 3.53043616e-01 1.53597221e-01 -1.84760049e-01
1.21153399e-01 -9.77802813e-01 -5.57211220e-01 -4.50438350e-01
-2.39137448e-02 -2.81962067e-01 4.11534905e-01 1.16766322e+00
1.54934108e-01 3.22167039e-01 2.36274257e-01 -1.71555325e-01
-8.58707845e-01 5.79009950e-01 2.95137405e-01 -3.33425015e-01
-2.15906784e-01 9.83392894e-01 -3.58142033e-02 2.46401474e-01
5.64209163e-01 -2.79930592e-01 2.25240812e-01 6.61077425e-02
8.16688299e-01 7.80967951e-01 -2.79816329e-01 6.44149065e-01
-2.34642953e-01 6.96428120e-02 -4.55055714e-01 -4.52509940e-01
1.50568891e+00 -3.39456409e-01 2.23913804e-01 5.35222292e-01
2.02498749e-01 1.49972022e-01 -1.56634998e+00 -2.89101988e-01
2.02131808e-01 -8.18750262e-01 3.30291957e-01 -9.21398461e-01
-9.98417139e-01 4.42896932e-01 3.44811261e-01 5.51208377e-01
7.81833053e-01 -2.38521677e-02 3.06972444e-01 6.72298193e-01
7.05079734e-01 -1.27797914e+00 -3.23106140e-01 3.23147893e-01
5.10113597e-01 -9.38246191e-01 5.15771508e-01 -9.39040557e-02
-6.07233524e-01 9.10416126e-01 1.71500266e-01 2.30375789e-02
3.97446692e-01 3.00944418e-01 -4.82886404e-01 2.38574356e-01
-9.29299295e-01 -3.80170524e-01 -6.86204657e-02 4.87540960e-01
1.75274938e-01 3.10738951e-01 -7.60465205e-01 1.19612718e+00
-3.48055184e-01 1.63618565e-01 6.58304751e-01 9.42065239e-01
-5.49512208e-01 -9.45308626e-01 -5.90083361e-01 8.72112215e-01
-8.49013269e-01 -1.08808808e-01 -1.64238945e-01 6.22987926e-01
-7.61177719e-01 1.05377030e+00 6.59598708e-02 1.95769385e-01
1.05429575e-01 -1.81854755e-01 1.10093665e+00 -2.45046481e-01
-1.79411724e-01 3.69125515e-01 3.87970150e-01 -4.95836288e-01
-2.97985226e-01 -8.82607102e-01 -7.01239824e-01 -5.53368390e-01
-6.87695205e-01 4.70347226e-01 5.36728442e-01 9.25959945e-01
9.22577083e-02 3.74797583e-01 7.06316769e-01 -4.14902061e-01
-1.51765633e+00 -8.19746971e-01 -1.04412794e+00 -2.16051042e-01
4.05576408e-01 -8.03180218e-01 -5.32995522e-01 -5.78543842e-01]
|
[4.495402812957764, 3.258636474609375]
|
9ccfec5d-7457-4c23-9b24-1709ec33eab1
|
instance-level-sketch-based-retrieval-by-deep
|
1811.11375
| null |
https://arxiv.org/abs/1811.11375v2
|
https://arxiv.org/pdf/1811.11375v2.pdf
|
Instance-level Sketch-based Retrieval by Deep Triplet Classification Siamese Network
|
Sketch has been employed as an effective communicative tool to express the abstract and intuitive meanings of object. Recognizing the free-hand sketch drawing is extremely useful in many real-world applications. While content-based sketch recognition has been studied for several decades, the instance-level Sketch-Based Image Retrieval (SBIR) tasks have attracted significant research attention recently. The existing datasets such as QMUL-Chair and QMUL-Shoe, focus on the retrieval tasks of chairs and shoes. However, there are several key limitations in previous instance-level SBIR works. The state-of-the-art works have to heavily rely on the pre-training process, quality of edge maps, multi-cropping testing strategy, and augmenting sketch images. To efficiently solve the instance-level SBIR, we propose a new Deep Triplet Classification Siamese Network (DeepTCNet) which employs DenseNet-169 as the basic feature extractor and is optimized by the triplet loss and classification loss. Critically, our proposed DeepTCNet can break the limitations existed in previous works. The extensive experiments on five benchmark sketch datasets validate the effectiveness of the proposed model. Additionally, to study the tasks of sketch-based hairstyle retrieval, this paper contributes a new instance-level photo-sketch dataset - Hairstyle Photo-Sketch dataset, which is composed of 3600 sketches and photos, and 2400 sketch-photo pairs.
|
['xiangyang xue', 'Yu-Gang Jiang', 'Shaogang Gong', 'Yanwei Fu', 'Peng Lu', 'Hangyu Lin']
|
2018-11-28
| null | null | null | null |
['sketch-based-image-retrieval', 'sketch-recognition']
|
['computer-vision', 'computer-vision']
|
[ 4.71471027e-02 -7.49729037e-01 -3.46972197e-01 -2.05544814e-01
-6.96937263e-01 -2.91355759e-01 6.43267274e-01 -3.92231792e-01
-1.16608195e-01 4.86164689e-01 -1.78877488e-01 7.19139427e-02
-5.17250717e-01 -7.95846522e-01 -5.30375481e-01 -6.37629926e-01
3.57946783e-01 2.93540418e-01 1.15375243e-01 -3.68324906e-01
6.57692850e-01 6.03027582e-01 -1.36794543e+00 3.01099867e-01
5.55000961e-01 1.52024746e+00 1.11110076e-01 1.98449343e-01
-6.16131246e-01 3.34478974e-01 -5.60269535e-01 -7.88225114e-01
3.50116193e-01 -1.50979996e-01 -1.21486813e-01 -4.41297367e-02
7.86338449e-01 -5.57123363e-01 -5.72084665e-01 9.03230548e-01
7.88633347e-01 8.67370516e-03 7.02184737e-01 -1.67407155e+00
-1.16516829e+00 1.83372796e-01 -6.48534060e-01 -3.37631106e-01
4.07844603e-01 8.39930549e-02 1.11432207e+00 -1.33386576e+00
7.97793984e-01 1.44874668e+00 5.44793367e-01 4.46859658e-01
-6.73063695e-01 -1.32856882e+00 1.61147639e-01 3.46529812e-01
-1.75922835e+00 -9.64701697e-02 1.32355082e+00 -4.87839803e-03
4.15130466e-01 1.02253132e-01 9.04928505e-01 1.15052140e+00
-1.19677477e-01 1.41274655e+00 9.30812955e-01 -1.27690732e-01
-1.51495844e-01 -7.18139932e-02 -2.66047448e-01 9.66117203e-01
-4.21331404e-03 -8.74503702e-02 -5.89446485e-01 -1.12624906e-01
1.36779165e+00 6.00237191e-01 -6.10294007e-02 -5.95178485e-01
-1.31469667e+00 6.62587583e-01 6.70647681e-01 2.36531541e-01
-1.32111952e-01 4.27580059e-01 6.73902154e-01 4.59599137e-01
2.18204111e-01 1.82133108e-01 -2.24513304e-03 9.54352096e-02
-1.06365275e+00 4.50229615e-01 4.28086430e-01 1.37051392e+00
5.21127582e-01 -4.84598391e-02 -1.43704414e-01 1.21698511e+00
2.35828996e-01 7.52101958e-01 2.45511577e-01 -6.72224164e-01
5.80221593e-01 7.83181131e-01 -1.40767246e-01 -1.46917403e+00
3.61391753e-01 -1.09695889e-01 -1.25185323e+00 -1.02591142e-01
2.85636615e-02 5.87333679e-01 -7.37174332e-01 1.43045783e+00
1.42040448e-02 1.29200146e-01 -4.58376050e-01 1.19547749e+00
1.09194207e+00 5.86799979e-01 -2.67310385e-02 1.87426999e-01
1.28994894e+00 -1.09901822e+00 -7.84011126e-01 2.59121150e-01
-1.59363478e-01 -9.87317979e-01 1.46115947e+00 5.10734081e-01
-1.11471951e+00 -8.08974206e-01 -1.31849945e+00 -4.36320066e-01
-7.08354115e-01 4.55503047e-01 8.00961614e-01 4.13939923e-01
-6.83752358e-01 5.05084395e-01 3.60152125e-03 -3.68187875e-01
8.39126408e-01 2.50407785e-01 -5.15846074e-01 -5.50421655e-01
-1.19262648e+00 4.46138591e-01 -7.25549310e-02 3.72454166e-01
-7.05257058e-01 -5.60381591e-01 -5.94238997e-01 1.45476282e-01
4.84681249e-01 -6.39591157e-01 6.13594532e-01 -7.79147208e-01
-1.43927276e+00 8.75432432e-01 1.32225335e-01 3.44270170e-01
7.30783045e-01 -2.61555091e-02 -4.49215770e-01 2.82059222e-01
3.56319845e-02 8.47854257e-01 1.10628021e+00 -1.06203580e+00
-2.79627770e-01 -4.29987878e-01 3.76201384e-02 1.12099677e-01
-4.56604540e-01 -1.67425230e-01 -9.74007666e-01 -1.05844641e+00
2.61992604e-01 -7.73860574e-01 4.05572116e-01 7.81774879e-01
-1.10209875e-01 -6.99174583e-01 9.63268101e-01 -3.18725467e-01
1.25084949e+00 -2.16557837e+00 1.17955185e-01 2.12121606e-01
-3.95576879e-02 2.85676897e-01 -4.82399017e-01 8.05548072e-01
1.85783967e-01 2.27291673e-01 -2.16101915e-01 -3.60578656e-01
3.34785730e-01 2.23693371e-01 -5.89399993e-01 1.84020489e-01
3.63532096e-01 1.13054955e+00 -7.77643025e-01 -9.10815954e-01
3.79964262e-01 5.15345633e-01 -2.23769933e-01 1.77946463e-01
-7.74377733e-02 -1.43711299e-01 -7.11427987e-01 1.20670712e+00
9.75894094e-01 -1.45641893e-01 -5.27729839e-02 -4.38741982e-01
2.12681636e-01 -3.38521987e-01 -1.25547540e+00 2.50362563e+00
-4.22042221e-01 5.07164240e-01 7.36300554e-03 -8.25773895e-01
1.12316871e+00 9.97295156e-02 3.52131307e-01 -1.02227354e+00
-7.76096284e-02 4.17515785e-01 -5.39942145e-01 -5.37618816e-01
5.73968351e-01 -1.61245003e-01 -1.80962205e-01 2.76481003e-01
-1.56581491e-01 -3.25146377e-01 3.76605056e-02 4.21041608e-01
7.08671212e-01 3.79959524e-01 -8.41210485e-02 -8.95103961e-02
7.23702490e-01 -3.28981847e-01 1.84279993e-01 4.19677943e-01
-2.77433135e-02 7.81757176e-01 3.31861496e-01 -7.05119371e-01
-9.59434986e-01 -1.18622780e+00 -1.16577588e-01 8.90459955e-01
7.13130832e-01 -3.68674874e-01 -2.36587957e-01 -6.02836370e-01
3.47633988e-01 4.17861268e-02 -6.09770656e-01 1.40602896e-02
-5.02590835e-01 -4.77906093e-02 7.74540544e-01 5.05436957e-01
9.51602757e-01 -1.43870986e+00 -1.12788036e-01 1.14531079e-02
5.02133882e-03 -9.05014217e-01 -8.22746992e-01 -6.62494898e-01
-6.18132591e-01 -1.13647592e+00 -1.22969210e+00 -8.79560530e-01
7.23266602e-01 4.98844475e-01 1.04240847e+00 7.27666497e-01
-7.30957508e-01 4.50785577e-01 -2.87726581e-01 -1.87215269e-01
4.72295076e-01 4.69058231e-02 -2.94695556e-01 6.10156246e-02
2.61136472e-01 -5.97629964e-01 -8.96076679e-01 4.80170548e-01
-1.18958008e+00 -2.96794605e-02 9.38917220e-01 1.22564507e+00
6.05653882e-01 -2.92256713e-01 7.51809657e-01 -5.09856045e-01
1.00117314e+00 -1.86114594e-01 -2.63750732e-01 8.22124541e-01
-4.41990227e-01 -6.78714961e-02 6.15361094e-01 -5.96729755e-01
-7.92003334e-01 -2.73122966e-01 -4.36107665e-02 -1.00764346e+00
1.71853930e-01 3.09589982e-01 -3.47263604e-01 -3.36743385e-01
-7.94449374e-02 4.62399036e-01 -2.23041233e-02 -4.22190517e-01
4.35377330e-01 6.66454971e-01 3.78855318e-01 -1.05837977e+00
7.19558537e-01 3.35886031e-01 2.16719866e-01 -8.86305690e-01
-4.04520392e-01 -4.24346775e-01 -3.12495440e-01 -3.05480748e-01
5.25367796e-01 -6.77115440e-01 -1.22975945e+00 4.11975563e-01
-1.22932816e+00 2.08462894e-01 8.16390093e-04 -1.28059357e-01
-3.27469826e-01 7.21189201e-01 -3.34192425e-01 -1.03248048e+00
-7.08775342e-01 -1.15759337e+00 1.63687158e+00 2.06896126e-01
4.15620863e-01 -5.71044624e-01 -3.17967772e-01 2.07031637e-01
5.15102804e-01 6.31127730e-02 9.32319760e-01 -1.97668113e-02
-1.15392053e+00 -4.00247335e-01 -8.92689109e-01 1.21574067e-01
-5.93988374e-02 1.65834778e-03 -7.22827792e-01 -2.06620663e-01
-7.29211807e-01 -7.76612401e-01 1.03384757e+00 -1.63820460e-01
1.63776505e+00 -5.20254187e-02 -1.79235339e-01 6.28147960e-01
1.66069937e+00 1.79368630e-01 8.40439200e-01 -1.13715805e-01
6.04868472e-01 2.32888594e-01 7.55030870e-01 3.50184381e-01
4.17746931e-01 6.67846560e-01 2.21155331e-01 -6.88332692e-03
-2.33651191e-01 -7.99794614e-01 -1.43483624e-01 9.28086758e-01
-1.12657763e-01 -1.73736021e-01 -2.43417501e-01 4.53104645e-01
-1.85969591e+00 -9.60802436e-01 4.21975493e-01 1.95128226e+00
5.65866947e-01 -1.10999778e-01 -1.69278850e-04 1.21267870e-01
5.04232347e-01 4.40854162e-01 -6.88893855e-01 3.02995648e-02
-1.91975817e-01 4.43209916e-01 6.94542332e-03 -5.34400381e-02
-9.43546951e-01 1.05395067e+00 4.91798258e+00 1.51768470e+00
-1.12036216e+00 -2.12556869e-01 1.91286623e-01 4.29309756e-02
-3.88706356e-01 -1.45492628e-01 -4.06388909e-01 6.05451465e-01
-3.78124744e-01 2.46248856e-01 8.38251531e-01 8.38901162e-01
-3.85767132e-01 1.23799510e-01 -1.00651753e+00 1.66571403e+00
3.84833783e-01 -1.33093238e+00 6.11664057e-01 -2.25218743e-01
4.10194397e-01 -5.62193751e-01 2.50226110e-01 5.68057656e-01
-3.21205378e-01 -9.36199963e-01 5.71495771e-01 8.84780467e-01
1.37654841e+00 -6.93441331e-01 5.03520250e-01 1.99321136e-02
-1.72153044e+00 6.20031059e-02 -7.32554257e-01 1.55559972e-01
-1.08776934e-01 1.94079921e-01 -1.50149271e-01 8.79722238e-01
5.65531015e-01 1.00963378e+00 -4.99107063e-01 9.00807023e-01
-4.59303409e-02 -4.80983406e-02 -2.94474691e-01 -4.59774315e-01
3.33256632e-01 -5.26775420e-01 1.45884678e-01 1.06669867e+00
2.95767277e-01 2.53546834e-01 2.37118945e-01 1.11909044e+00
-3.96303445e-01 3.33903462e-01 -7.41185606e-01 -3.59613955e-01
6.15568280e-01 1.49503446e+00 -5.45151532e-01 -4.84844983e-01
-1.63013443e-01 1.15112257e+00 1.92202687e-01 3.52882743e-01
-5.96851170e-01 -8.95464301e-01 5.49466074e-01 -1.19503297e-01
2.76950896e-01 -1.67056903e-01 -1.22635417e-01 -1.21266866e+00
3.34284812e-01 -7.17174649e-01 1.84247464e-01 -1.09332108e+00
-1.75573182e+00 2.58371621e-01 -2.04226777e-01 -1.30084598e+00
3.18475962e-01 -7.44217634e-01 -7.17160821e-01 7.89311051e-01
-1.67921984e+00 -1.77044392e+00 -7.20219016e-01 6.65472925e-01
8.20051849e-01 -3.59375447e-01 6.65277660e-01 6.96232319e-01
-4.51695681e-01 8.78762305e-01 -1.76791728e-01 3.65648717e-01
8.75857592e-01 -8.85342896e-01 3.60635012e-01 8.20800588e-02
2.00714022e-01 1.03008366e+00 -7.25379139e-02 -5.68841636e-01
-1.97441328e+00 -6.32786751e-01 5.10352612e-01 -7.80492574e-02
5.11585951e-01 -4.15441364e-01 -6.58706188e-01 7.69650415e-02
1.64826244e-01 1.44127652e-01 1.35884121e-01 -1.96328327e-01
-6.09939754e-01 -5.52238464e-01 -9.67555583e-01 6.84340596e-01
1.56805253e+00 -8.31366479e-01 -4.50534791e-01 2.22018123e-01
3.23364615e-01 -2.16634765e-01 -8.69284153e-01 3.54194641e-01
1.40205407e+00 -8.57569337e-01 1.37768531e+00 -6.57029927e-01
7.08609104e-01 -1.13053977e-01 -1.61630422e-01 -8.29315901e-01
-5.90771660e-02 -2.56079704e-01 3.10641885e-01 1.12898946e+00
-2.79531419e-01 -3.59947503e-01 7.53475487e-01 3.25847566e-01
1.54320315e-01 -1.14242899e+00 -7.42347717e-01 -6.57296240e-01
-1.79876000e-01 -1.75854802e-01 9.37498808e-01 9.00824010e-01
-4.08956945e-01 2.86628634e-01 -5.97284317e-01 -4.61656421e-01
6.78818703e-01 5.74224591e-01 1.07781494e+00 -1.27224457e+00
2.47550040e-01 -7.72000432e-01 -5.09483933e-01 -1.37327230e+00
2.55987674e-01 -8.69470119e-01 -3.18764359e-01 -1.52726150e+00
2.23426417e-01 -8.42200220e-01 -4.06979859e-01 3.66244256e-01
-9.42470282e-02 3.89220834e-01 5.14267683e-01 2.75000781e-01
-8.43680799e-01 9.05475497e-01 1.70048881e+00 -6.23890638e-01
4.34644699e-01 -3.88708413e-01 -3.05378497e-01 2.65769333e-01
2.24811062e-01 -1.27479419e-01 -5.37847936e-01 -4.54934001e-01
4.13509846e-01 2.41490647e-01 6.77695394e-01 -6.92825258e-01
5.32611907e-01 -1.44748434e-01 5.35170436e-01 -1.15799427e+00
8.09109151e-01 -1.06654763e+00 1.04474694e-01 1.91779956e-01
-4.13895786e-01 2.20814675e-01 -1.26404315e-01 7.25349903e-01
-3.58108670e-01 -1.29500583e-01 3.87091815e-01 -4.42193925e-01
-8.05069864e-01 7.47917056e-01 4.66341138e-01 3.15540731e-02
6.92374170e-01 -3.19015861e-01 -1.86694548e-01 -2.75193900e-01
-3.78500894e-02 2.24834159e-01 4.34284002e-01 6.85886800e-01
1.24282789e+00 -1.88171279e+00 -4.65333849e-01 3.82821918e-01
4.05277938e-01 -1.67216972e-01 5.27574062e-01 4.44872230e-01
-5.03863096e-01 4.04502690e-01 -4.79761630e-01 -3.46604854e-01
-1.05524909e+00 6.52853966e-01 -4.66931723e-02 -5.62003255e-02
-5.91579318e-01 7.79532433e-01 2.17229411e-01 -4.66941744e-01
5.37558734e-01 -1.54010102e-01 1.71607316e-01 -2.44081393e-02
2.66388863e-01 3.00406575e-01 -2.96638936e-01 -1.34195596e-01
-2.77881473e-01 1.08059347e+00 3.22764851e-02 4.74558324e-02
1.06584847e+00 1.85673863e-01 -2.59250432e-01 3.22697073e-01
1.36133540e+00 -1.97572231e-01 -8.21945608e-01 -2.05773354e-01
-2.68379629e-01 -9.18651700e-01 -1.38507813e-01 -8.00687790e-01
-1.26602268e+00 1.32357824e+00 5.03290474e-01 -1.42417878e-01
1.01508188e+00 -5.40025532e-01 1.22568190e+00 7.68350005e-01
6.68704033e-01 -1.11233509e+00 5.32152653e-01 2.57706568e-02
1.38902116e+00 -1.30719912e+00 1.68651327e-01 -1.84127912e-01
-5.49499869e-01 1.24203587e+00 5.37764251e-01 -4.25566316e-01
5.97817123e-01 -1.26048893e-01 -2.42229775e-01 -3.50637347e-01
-4.11305875e-01 2.78181639e-02 5.34817338e-01 1.71607509e-01
1.56770051e-01 4.58623171e-02 -4.00094926e-01 7.67237961e-01
2.77689070e-01 4.12174851e-01 -3.34531397e-01 7.93667793e-01
-7.07829595e-02 -1.22153056e+00 -1.46861523e-01 4.72054660e-01
1.24965496e-01 -1.73290566e-01 -7.88630247e-01 9.45342183e-01
3.80880982e-02 4.32682633e-01 -1.50225222e-01 -3.31489742e-01
4.45473254e-01 2.33323108e-02 7.28035450e-01 8.55671391e-02
-4.90924805e-01 -1.28617674e-01 -3.57928485e-01 -4.73519236e-01
-5.02885818e-01 -7.65074342e-02 -8.10305536e-01 -3.42508376e-01
-3.98100525e-01 -6.24926910e-02 7.30619073e-01 6.10820174e-01
3.79116118e-01 2.90531129e-01 5.92747211e-01 -9.50585365e-01
-5.32410741e-01 -8.02417874e-01 -8.19615722e-01 6.19876742e-01
-1.21310562e-01 -1.11582935e+00 5.49749807e-02 -3.75971824e-01]
|
[11.656868934631348, 0.6279942989349365]
|
83c78ecb-c91b-43e9-9809-6c626970d251
|
vsr-a-unified-framework-for-document-layout
|
2105.06220
| null |
https://arxiv.org/abs/2105.06220v1
|
https://arxiv.org/pdf/2105.06220v1.pdf
|
VSR: A Unified Framework for Document Layout Analysis combining Vision, Semantics and Relations
|
Document layout analysis is crucial for understanding document structures. On this task, vision and semantics of documents, and relations between layout components contribute to the understanding process. Though many works have been proposed to exploit the above information, they show unsatisfactory results. NLP-based methods model layout analysis as a sequence labeling task and show insufficient capabilities in layout modeling. CV-based methods model layout analysis as a detection or segmentation task, but bear limitations of inefficient modality fusion and lack of relation modeling between layout components. To address the above limitations, we propose a unified framework VSR for document layout analysis, combining vision, semantics and relations. VSR supports both NLP-based and CV-based methods. Specifically, we first introduce vision through document image and semantics through text embedding maps. Then, modality-specific visual and semantic features are extracted using a two-stream network, which are adaptively fused to make full use of complementary information. Finally, given component candidates, a relation module based on graph neural network is incorported to model relations between components and output final results. On three popular benchmarks, VSR outperforms previous models by large margins. Code will be released soon.
|
['Fei Wu', 'Yi Niu', 'ShiLiang Pu', 'Zhanzhan Cheng', 'Liang Qiao', 'Can Li', 'Peng Zhang']
|
2021-05-13
| null | null | null | null |
['document-layout-analysis']
|
['computer-vision']
|
[ 3.26203465e-01 -2.71262556e-01 -3.81927818e-01 -1.96891680e-01
-2.75373161e-01 -7.76891589e-01 6.16923034e-01 4.46996123e-01
-2.23252028e-02 1.92994207e-01 4.08736795e-01 -3.70024025e-01
-7.03286603e-02 -7.80064225e-01 -4.59373415e-01 -5.14417410e-01
3.82823408e-01 2.13054433e-01 1.62935898e-01 -5.85506558e-02
5.73100567e-01 5.73858559e-01 -1.38043964e+00 6.52452886e-01
8.53257716e-01 9.10278022e-01 4.56213593e-01 8.02102566e-01
-1.05658400e+00 9.13363874e-01 -4.87834543e-01 -1.82720482e-01
-3.00455540e-01 -4.39037532e-01 -6.92725003e-01 4.02218550e-01
3.32581639e-01 5.06242097e-04 -5.15969396e-01 1.16635561e+00
2.81016111e-01 -2.18628570e-02 8.42428505e-01 -1.40702713e+00
-1.28038538e+00 8.79452288e-01 -7.54314661e-01 -3.26805055e-01
7.16687441e-01 -2.93515604e-02 1.23473132e+00 -9.83371258e-01
6.63831353e-01 1.42236888e+00 3.83515984e-01 3.18675458e-01
-1.27743685e+00 3.88143994e-02 6.16886914e-01 3.61646682e-01
-1.20538557e+00 -1.84548929e-01 1.31960225e+00 -4.35775667e-01
9.06305015e-01 4.35948938e-01 6.67992711e-01 1.06645417e+00
-7.29667172e-02 1.35531747e+00 6.70599222e-01 -7.06298590e-01
1.10056154e-01 2.10816488e-01 5.24265528e-01 8.62756610e-01
5.73650114e-02 -5.69520473e-01 -5.73561072e-01 2.87629217e-01
5.83023310e-01 1.23093538e-01 -3.60959053e-01 -6.02074802e-01
-1.24573493e+00 5.68737328e-01 6.23799264e-01 4.49480474e-01
-3.01030934e-01 8.93207639e-02 5.09090781e-01 -2.34417524e-02
-3.26589718e-02 3.43255073e-01 -1.40880391e-01 2.43883461e-01
-1.03765154e+00 -5.58231808e-02 6.59996986e-01 1.16910064e+00
7.06260502e-01 -8.87922943e-02 -5.25551915e-01 9.14552391e-01
6.45808041e-01 5.96414566e-01 3.24708760e-01 -6.48169875e-01
6.16381288e-01 1.12018740e+00 -3.54378462e-01 -1.41979206e+00
-4.22990143e-01 -2.72794694e-01 -9.20981109e-01 -1.14154004e-01
8.83192867e-02 3.69478524e-01 -1.11415267e+00 1.25809169e+00
-4.32338603e-02 -2.29632437e-01 8.92985240e-02 7.82684386e-01
1.11083281e+00 6.74661517e-01 -4.11288105e-02 -2.21636239e-02
1.58279359e+00 -1.36930788e+00 -1.09609103e+00 -2.39726499e-01
6.63793802e-01 -1.02551794e+00 1.31677628e+00 4.47650820e-01
-1.05798972e+00 -6.69936240e-01 -1.12477100e+00 -3.36674362e-01
-8.88015807e-01 4.64091778e-01 5.12973726e-01 6.17592335e-01
-1.12262356e+00 1.80646732e-01 -4.13345665e-01 -6.25920653e-01
4.86719906e-01 6.51605800e-03 -2.49984056e-01 -3.45412374e-01
-8.21907640e-01 6.08960688e-01 5.84284604e-01 4.50602233e-01
-4.48479772e-01 -3.65281999e-01 -1.10990262e+00 3.14734429e-01
5.33538222e-01 -6.99825048e-01 7.44100988e-01 -7.13052094e-01
-1.11992991e+00 4.97947842e-01 -2.86170155e-01 -1.99299321e-01
2.86408752e-01 -3.97947021e-02 -5.75477660e-01 2.60840803e-01
-8.52292180e-02 6.31685972e-01 7.65827417e-01 -1.71630287e+00
-5.04368007e-01 -3.42188179e-01 8.95949900e-02 2.67653614e-01
-6.09479964e-01 -2.47411877e-01 -1.11503410e+00 -3.88549656e-01
3.94613594e-01 -5.97687364e-01 9.29591507e-02 8.07459950e-02
-8.55157733e-01 -7.92556852e-02 1.19925332e+00 -8.40486467e-01
1.52337229e+00 -1.98587000e+00 3.55641693e-01 2.90184617e-01
3.09735149e-01 2.55018026e-01 -5.89677572e-01 7.44350791e-01
9.20334756e-02 2.05072552e-01 -1.23637609e-01 -3.24462414e-01
2.26721376e-01 1.57834858e-01 -2.99056709e-01 1.03275560e-01
1.17253274e-01 1.23202956e+00 -6.90292120e-01 -7.58888185e-01
6.43038392e-01 4.17341650e-01 -3.42467010e-01 7.51707032e-02
-5.53825974e-01 -1.14181325e-01 -5.15448928e-01 1.08649194e+00
7.88484335e-01 -4.14945155e-01 3.96974832e-01 -7.06925333e-01
1.72691625e-02 -3.40333670e-01 -1.22042763e+00 2.07451606e+00
-4.45476502e-01 7.50615478e-01 -8.47070664e-02 -1.14208174e+00
1.04152966e+00 -1.04367577e-01 2.76880801e-01 -8.51804614e-01
1.80135220e-01 -1.31210461e-01 -2.30189651e-01 -7.88025439e-01
7.15232730e-01 5.27070343e-01 4.62002642e-02 1.06381640e-01
-1.33793931e-02 -1.20923005e-01 3.58048588e-01 5.69295764e-01
1.12647188e+00 4.05510396e-01 1.69568080e-02 2.19697684e-01
8.97176385e-01 5.24496287e-02 9.63214636e-02 7.08388627e-01
3.44052073e-03 7.12072015e-01 7.38337457e-01 -8.17778409e-02
-8.76644909e-01 -1.10531461e+00 3.84857118e-01 9.97073472e-01
4.64141309e-01 -8.54705811e-01 -7.26977050e-01 -8.85732770e-01
-1.44262575e-02 7.93092370e-01 -6.97352111e-01 -4.30847146e-02
-4.12597924e-01 -4.17594314e-01 4.01364386e-01 8.38599205e-01
2.95944661e-01 -1.17408514e+00 -2.08809078e-01 -1.15126371e-01
-2.39643455e-01 -1.12055612e+00 -3.58078122e-01 4.00171801e-02
-6.06252968e-01 -1.12755251e+00 -7.19823182e-01 -9.25672352e-01
9.00225282e-01 5.37768722e-01 8.42563272e-01 2.45551184e-01
-5.69707155e-01 7.35951483e-01 -4.20192361e-01 -1.54769868e-01
-1.27804130e-01 2.69059651e-02 -5.00027895e-01 -5.23066409e-02
1.91311464e-01 -1.50104880e-01 -4.79071409e-01 -1.26251101e-01
-9.85098422e-01 9.04936865e-02 7.68992245e-01 8.01587641e-01
5.76599300e-01 1.19844355e-01 9.07479674e-02 -8.90425682e-01
9.00461555e-01 -1.64658695e-01 -4.50489461e-01 8.97697568e-01
-5.90074122e-01 3.00211072e-01 7.17434824e-01 -1.93157956e-01
-1.16026485e+00 9.97666270e-02 1.03255883e-01 -6.28971219e-01
-3.01546276e-01 7.53998876e-01 -6.69133008e-01 2.27963656e-01
3.52058530e-01 4.72269952e-01 -1.56664073e-01 -5.11716604e-01
9.29535329e-01 5.53942025e-01 5.27252674e-01 -5.01777828e-01
7.40302444e-01 3.83619398e-01 4.09117676e-02 -9.63510752e-01
-7.49608040e-01 -5.75791180e-01 -9.13817167e-01 -3.35529327e-01
9.84559417e-01 -3.94269168e-01 -7.72268593e-01 1.81904539e-01
-1.30446184e+00 6.48568794e-02 -1.23105362e-01 2.06200331e-01
-3.73969078e-01 8.08764160e-01 -4.66678053e-01 -8.62167299e-01
-1.95494100e-01 -1.17144811e+00 1.20113742e+00 3.90391380e-01
1.06957636e-03 -1.18874562e+00 -2.46269286e-01 3.93704504e-01
3.82090397e-02 2.61469707e-02 1.34832478e+00 -3.55638325e-01
-6.42543435e-01 -3.48834097e-01 -7.75496900e-01 2.40568444e-01
2.38584638e-01 4.45108920e-01 -9.27082598e-01 9.11981612e-02
-5.76553226e-01 -9.28434208e-02 1.23051155e+00 5.80023229e-01
1.43580198e+00 1.52081689e-02 -6.54019237e-01 5.51704347e-01
1.59917951e+00 3.29151988e-01 6.80424869e-01 3.19927931e-01
1.28608406e+00 8.84357393e-01 4.80838537e-01 2.38068819e-01
5.21628678e-01 2.89481193e-01 5.24310410e-01 -2.26290122e-01
-4.26830560e-01 -4.50254679e-01 2.16187492e-01 9.74222541e-01
2.06717640e-01 -7.08073616e-01 -9.84047890e-01 3.06671858e-01
-2.23312664e+00 -8.18912625e-01 -3.69875401e-01 1.68546426e+00
1.98199809e-01 2.50777274e-01 -7.41446987e-02 1.63732082e-01
7.15630829e-01 4.23842281e-01 -3.63559932e-01 -4.14588928e-01
-4.08956677e-01 -4.53274846e-01 2.91566372e-01 2.32675582e-01
-1.03521049e+00 1.19426751e+00 5.83105564e+00 7.54248202e-01
-8.57959569e-01 -2.82347322e-01 2.37919569e-01 3.22850615e-01
-7.17656851e-01 2.55229652e-01 -5.42141259e-01 1.84720889e-01
2.47520372e-01 2.29028285e-01 5.24966955e-01 6.94774508e-01
-4.23792973e-02 -1.27272487e-01 -1.14316249e+00 1.24687624e+00
5.53334236e-01 -1.49475622e+00 3.00233632e-01 -6.96401671e-02
3.99329931e-01 -6.81664407e-01 1.84639171e-01 2.14914456e-01
-5.13439588e-02 -1.02437603e+00 7.91941822e-01 8.51366878e-01
4.16871041e-01 -7.16321111e-01 5.79242229e-01 1.83052599e-01
-1.58917415e+00 -1.87335789e-01 -4.05522734e-01 2.91583091e-01
1.92844123e-01 4.20907855e-01 -6.17767513e-01 1.02217019e+00
2.98058927e-01 9.63149786e-01 -9.83937740e-01 8.91632318e-01
-4.35230553e-01 2.88632363e-01 2.07665145e-01 -1.78788990e-01
3.29259664e-01 -3.82097363e-01 2.13243857e-01 1.41495121e+00
1.35173649e-01 -5.17940819e-01 2.97946513e-01 1.15276349e+00
-9.87525843e-03 2.73514450e-01 -6.50234401e-01 -4.86204773e-01
3.82659078e-01 1.60296452e+00 -1.20581865e+00 -3.47321212e-01
-5.75101018e-01 1.18842471e+00 3.14896017e-01 6.62756503e-01
-8.52705717e-01 -4.94569689e-01 9.51439813e-02 -1.87989846e-01
2.36768097e-01 -3.62817019e-01 -5.43972433e-01 -1.22434783e+00
1.38447717e-01 -5.94963253e-01 3.61078441e-01 -1.06180656e+00
-1.05532193e+00 2.70151675e-01 -1.81542978e-01 -1.10860980e+00
3.30776304e-01 -8.25794756e-01 -5.51609099e-01 6.79807484e-01
-1.45926356e+00 -1.71440458e+00 -4.71467376e-01 3.49889904e-01
8.74205828e-01 -1.76941946e-01 6.97170556e-01 5.44917099e-02
-8.15443873e-01 3.65465790e-01 -2.26226822e-03 2.60799438e-01
4.97862816e-01 -1.37896991e+00 2.45455965e-01 9.43287134e-01
5.43124199e-01 7.16150403e-01 3.78654867e-01 -8.51812780e-01
-1.85495877e+00 -1.00683272e+00 6.28738523e-01 -3.28216940e-01
5.59856117e-01 -5.64375639e-01 -1.00414419e+00 4.17811900e-01
4.34824497e-01 -2.13760048e-01 5.24944365e-01 2.87378460e-01
-5.91949880e-01 1.33445084e-01 -6.32434189e-01 9.03921783e-01
1.07029760e+00 -7.56335139e-01 -6.34663820e-01 2.45610684e-01
8.06169391e-01 -6.99594989e-02 -5.74496388e-01 1.21770680e-01
6.27359867e-01 -7.05879629e-01 1.09920561e+00 -4.15094167e-01
5.50899148e-01 -5.65449238e-01 -3.32935095e-01 -1.05874574e+00
-3.16958070e-01 -9.42500830e-02 -3.39045465e-01 1.56538033e+00
3.25150013e-01 -6.77224770e-02 6.82201684e-01 1.86961487e-01
-8.77699777e-02 -4.68868345e-01 -8.71020481e-02 -7.34213471e-01
-4.32494640e-01 -6.38755739e-01 5.84823072e-01 8.74002397e-01
9.22271833e-02 6.15330875e-01 -2.74578840e-01 1.93745643e-01
6.39194191e-01 2.99895078e-01 5.19120276e-01 -1.01875925e+00
-2.79410779e-02 -8.09667885e-01 -2.63263911e-01 -9.82800722e-01
3.54898304e-01 -1.09000182e+00 -5.59007041e-02 -2.41730785e+00
2.98569828e-01 -2.83309501e-02 -3.11131895e-01 4.36791301e-01
-7.08733276e-02 -5.91925681e-02 4.11825716e-01 3.60365063e-02
-8.76047313e-01 6.02147937e-01 1.31785393e+00 -7.24385977e-01
-2.65968889e-01 -5.24894714e-01 -5.80711961e-01 7.57932365e-01
6.07550025e-01 1.51987672e-01 -7.17716753e-01 -5.50541699e-01
2.50761479e-01 9.51957032e-02 3.62776607e-01 -8.31878662e-01
5.62967181e-01 -1.33095622e-01 8.08392882e-01 -1.16911793e+00
3.69887114e-01 -8.40525568e-01 -2.80619979e-01 3.17366123e-01
-6.29256725e-01 -1.15195721e-01 3.31765376e-02 7.64496744e-01
-3.10625136e-01 -3.74650508e-01 2.66035885e-01 -7.27084577e-02
-1.02607834e+00 4.11540531e-02 -3.33358169e-01 -2.70823568e-01
7.40190804e-01 -5.28952658e-01 -5.32188535e-01 -1.59809351e-01
-7.70379066e-01 3.22394073e-01 4.83961135e-01 6.84418023e-01
1.02241778e+00 -1.30239367e+00 -1.33887604e-01 2.43231520e-01
5.02969146e-01 -2.05696151e-01 5.47336161e-01 6.58393919e-01
-6.11705124e-01 6.49670243e-01 -7.65137821e-02 -6.24319911e-01
-1.42063189e+00 1.17792165e+00 -1.65161490e-01 -5.14027774e-02
-4.82951999e-01 6.89697504e-01 2.87988365e-01 -3.49190474e-01
4.97379214e-01 -5.15226483e-01 -5.07800758e-01 4.02676195e-01
4.90370542e-01 2.82853991e-01 -1.50242239e-01 -5.45824647e-01
-4.79950428e-01 8.35075855e-01 -1.13081634e-01 -6.48744926e-02
8.57588291e-01 -2.85599321e-01 -2.18727827e-01 4.51591611e-01
1.15228629e+00 -4.54681888e-02 -8.39614391e-01 -2.34036312e-01
2.43067756e-01 -3.37951183e-01 2.27027964e-02 -8.45275939e-01
-8.91080141e-01 1.26603723e+00 5.36953151e-01 4.16164994e-01
1.30138922e+00 -4.75815088e-02 5.81563592e-01 2.92143375e-01
-1.73796982e-01 -1.27182329e+00 4.46153313e-01 3.78599256e-01
7.85292745e-01 -1.20111895e+00 3.31099257e-02 -6.41795874e-01
-7.98798680e-01 1.48581588e+00 7.13064194e-01 2.14812770e-01
4.03175920e-01 3.65338266e-01 -1.20164089e-01 -2.09392130e-01
-4.87142503e-01 -5.07871568e-01 6.38427198e-01 7.64561474e-01
5.03994763e-01 2.91940719e-02 -8.71049762e-02 4.98824239e-01
2.76169717e-01 -2.93032140e-01 3.05315137e-01 1.00343347e+00
-4.41369623e-01 -1.20598066e+00 -4.34390336e-01 3.02389413e-01
1.42460138e-01 -2.36024827e-01 -6.87006652e-01 6.28080845e-01
4.14981842e-02 9.16105628e-01 4.43375185e-02 -4.12779003e-01
3.71868938e-01 7.61342272e-02 6.11080050e-01 -5.06577790e-01
-2.19308257e-01 4.40071225e-01 -7.09985569e-02 -5.13111591e-01
-3.71876895e-01 -4.14308846e-01 -1.29207230e+00 3.55395935e-02
-3.43522578e-01 -2.39946708e-01 8.16088796e-01 7.80777872e-01
2.14595467e-01 1.11921322e+00 3.17877710e-01 -5.02207458e-01
-6.41249418e-02 -5.47443211e-01 -7.05058575e-01 2.92561352e-01
1.17708653e-01 -4.49705988e-01 4.36063507e-04 2.73178190e-01]
|
[11.608991622924805, 2.3663485050201416]
|
87931b0b-8ab1-4f13-bf1f-dcc395efca08
|
training-neural-networks-for
|
1911.00405
| null |
https://arxiv.org/abs/1911.00405v2
|
https://arxiv.org/pdf/1911.00405v2.pdf
|
Training Neural Networks for Likelihood/Density Ratio Estimation
|
Various problems in Engineering and Statistics require the computation of the likelihood ratio function of two probability densities. In classical approaches the two densities are assumed known or to belong to some known parametric family. In a data-driven version we replace this requirement with the availability of data sampled from the densities of interest. For most well known problems in Detection and Hypothesis testing we develop solutions by providing neural network based estimates of the likelihood ratio or its transformations. This task necessitates the definition of proper optimizations which can be used for the training of the network. The main purpose of this work is to offer a simple and unified methodology for defining such optimization problems with guarantees that the solution is indeed the desired function. Our results are extended to cover estimates for likelihood ratios of conditional densities and estimates for statistics encountered in local approaches.
|
['Kalliopi Basioti', 'George V. Moustakides']
|
2019-11-01
| null | null | null | null |
['density-ratio-estimation']
|
['methodology']
|
[ 1.90046102e-01 3.55748609e-02 -2.08654448e-01 -5.99227428e-01
-6.09282672e-01 -1.69130608e-01 2.88742900e-01 9.96827856e-02
-3.86090100e-01 1.13481486e+00 -4.36883628e-01 -3.72051626e-01
-4.22361702e-01 -8.93916488e-01 -7.35321522e-01 -6.84235752e-01
-2.81399667e-01 6.81002438e-01 3.55074406e-02 1.72591418e-01
2.15780884e-01 9.08302963e-01 -1.71426702e+00 -5.30283868e-01
7.81797230e-01 1.24438548e+00 1.51124373e-01 9.99647021e-01
-4.64653671e-02 3.06876242e-01 -7.62143433e-01 3.98510396e-02
3.58301878e-01 -3.02662790e-01 -5.69279194e-01 2.60070384e-01
2.89301783e-01 -2.68820226e-01 1.75989315e-01 1.21081257e+00
4.58819300e-01 3.67092848e-01 1.15616679e+00 -1.30407345e+00
-3.08892250e-01 2.07429722e-01 -2.83930600e-01 9.93207544e-02
1.31024450e-01 -4.70303059e-01 4.98574555e-01 -8.50540996e-01
1.45757556e-01 8.85238826e-01 5.81844032e-01 3.10910881e-01
-1.31374168e+00 -1.98558614e-01 -4.15353477e-01 -4.11687931e-03
-1.88495374e+00 -5.62275529e-01 3.60952735e-01 -3.88914973e-01
8.13151717e-01 2.03536063e-01 3.19684088e-01 4.27617848e-01
1.26717001e-01 2.51429975e-01 8.62758815e-01 -7.80354977e-01
5.13497531e-01 7.63574719e-01 -1.86557740e-01 5.75026512e-01
4.58814710e-01 8.43279064e-02 -2.16464207e-01 -3.79858278e-02
1.28960586e+00 -3.82818073e-01 -3.46315503e-01 -6.23731494e-01
-6.12520635e-01 9.48255718e-01 3.36169787e-02 5.24100542e-01
-4.37018961e-01 1.04963541e-01 -5.64976260e-02 1.47710934e-01
3.91667694e-01 3.17448646e-01 -3.75832111e-01 8.14673007e-02
-9.11272228e-01 1.24060228e-01 1.33295989e+00 1.02305138e+00
9.52565372e-01 2.54808843e-01 2.06382990e-01 7.97413230e-01
3.13013971e-01 3.95966202e-01 2.60060877e-01 -6.95301175e-01
4.45705131e-02 2.54857361e-01 4.72131729e-01 -7.77965724e-01
-2.33917803e-01 -4.90668476e-01 -6.57095373e-01 3.31502616e-01
8.54134381e-01 -3.19397509e-01 -8.65706563e-01 1.55734730e+00
5.62477887e-01 1.11481354e-01 7.25610182e-02 5.83056092e-01
2.09317490e-01 7.40231335e-01 -3.39968979e-01 -4.51154172e-01
8.71721268e-01 -1.04282461e-01 -8.17382216e-01 -4.03803550e-02
3.91667128e-01 -7.26679802e-01 7.04552174e-01 4.46532249e-01
-1.00580323e+00 -4.95282561e-01 -1.27330840e+00 2.69721329e-01
-4.29064095e-01 3.05709183e-01 1.05413303e-01 9.12276924e-01
-1.09060895e+00 7.36475587e-01 -5.48317075e-01 -2.95024872e-01
1.24415554e-01 6.89307570e-01 -2.37638474e-01 2.92381942e-01
-8.42552185e-01 1.25520957e+00 6.69339001e-01 4.07783031e-01
-7.40012407e-01 -4.65268523e-01 -8.59969437e-01 1.09282568e-01
2.36895725e-01 -2.31005758e-01 1.36866570e+00 -9.54563737e-01
-1.43081415e+00 5.55241764e-01 -1.91701893e-02 -5.25095224e-01
4.00377452e-01 -1.60456598e-01 -2.88257241e-01 -8.85951519e-02
5.48141599e-02 3.45293432e-01 9.19705749e-01 -9.61571813e-01
-6.91927433e-01 -7.99182057e-03 -2.19680831e-01 -1.25478640e-01
-3.01515996e-01 -1.11888841e-01 -2.29163408e-01 -5.52944466e-02
-8.65600035e-02 -3.10688078e-01 -1.81378916e-01 3.15027758e-02
-5.65415502e-01 -2.27872774e-01 7.35988975e-01 -6.85773969e-01
8.89801621e-01 -1.92431772e+00 -1.38325483e-01 6.79396510e-01
-8.41120332e-02 4.46527191e-02 2.04164356e-01 2.22894669e-01
-2.96515316e-01 -9.81642902e-02 -3.18765074e-01 -2.68068403e-01
2.28327755e-02 2.56606668e-01 -1.41158387e-01 9.89203930e-01
5.91099620e-01 2.94364810e-01 -3.99026841e-01 -5.99676907e-01
3.81361812e-01 5.86890697e-01 -3.33772600e-01 6.32858455e-01
-1.08548164e-01 1.96162134e-01 -3.64949465e-01 3.31063211e-01
7.07092941e-01 4.45001293e-03 1.33175716e-01 -8.36610124e-02
-2.33581454e-01 1.42918468e-01 -1.66325116e+00 8.29051197e-01
-4.13947970e-01 5.01972914e-01 1.68705016e-01 -1.48635793e+00
1.38437486e+00 3.85835558e-01 4.73423481e-01 -1.62212417e-01
4.78337526e-01 4.16534573e-01 -7.27879554e-02 -3.46201062e-01
2.69851506e-01 -6.75386906e-01 1.49318844e-01 1.92318708e-01
3.02143991e-01 -3.31526935e-01 3.04185688e-01 -4.05264288e-01
4.93779182e-01 9.63241160e-02 7.21212268e-01 -5.72753251e-01
6.09866023e-01 -3.43965024e-01 3.14489126e-01 8.02724659e-01
1.68932289e-01 5.73361218e-01 5.63149095e-01 -1.71016693e-01
-1.15215182e+00 -1.14952159e+00 -6.70774579e-01 4.88833994e-01
-1.52553499e-01 3.49289358e-01 -7.07506061e-01 -4.45570558e-01
-1.07172012e-01 7.67165899e-01 -5.55146039e-01 -8.58248621e-02
-4.15208906e-01 -6.86859429e-01 2.07308739e-01 5.78093410e-01
2.87531435e-01 -6.27830803e-01 -5.14933467e-01 2.24429563e-01
4.56749618e-01 -9.33810592e-01 -8.65035653e-02 7.00931013e-01
-7.82650173e-01 -1.02188015e+00 -7.80019581e-01 -6.83590591e-01
8.09985757e-01 -3.41243476e-01 8.89958501e-01 -1.04462788e-01
-2.57933497e-01 4.27512556e-01 1.19179532e-01 -5.68829656e-01
-5.56991518e-01 -9.12366286e-02 2.77904451e-01 -4.10946598e-03
2.69626945e-01 -7.53319561e-01 5.70425801e-02 4.48988616e-01
-1.01748371e+00 -5.91362119e-01 5.23567021e-01 5.71739674e-01
5.11611581e-01 3.27746540e-01 7.24881530e-01 -4.16408479e-01
6.87258422e-01 -4.18034703e-01 -1.23110056e+00 3.12375963e-01
-2.72093594e-01 3.55788946e-01 7.61616468e-01 -5.21257818e-01
-8.66722226e-01 1.77451804e-01 -1.67569235e-01 -2.48693317e-01
-5.36501527e-01 4.64528948e-01 -3.73778701e-01 -1.87205359e-01
7.14583814e-01 5.80080263e-02 5.98557815e-02 -2.87583947e-01
-1.15562603e-02 7.12389708e-01 7.08372951e-01 -6.05211735e-01
8.32177758e-01 1.91950947e-01 3.08704138e-01 -1.07815576e+00
-6.83062792e-01 -5.15429497e-01 -6.13699913e-01 -3.27391624e-01
6.29010797e-01 -2.51131207e-01 -7.19752669e-01 1.13202706e-01
-1.11934948e+00 -6.17310517e-02 -6.13545418e-01 6.40571952e-01
-8.50121319e-01 1.73461199e-01 1.87141728e-02 -1.49770653e+00
1.22219659e-01 -9.11335468e-01 8.16450715e-01 4.61138904e-01
-8.47984925e-02 -1.35351980e+00 1.36211008e-01 -4.16003555e-01
4.06104088e-01 -3.89629230e-02 6.69446349e-01 -9.31165338e-01
-3.58302087e-01 -7.59190202e-01 -1.54487818e-01 7.72838473e-01
3.57889324e-01 1.30619019e-01 -8.85355353e-01 -2.36964282e-02
3.73477429e-01 -1.36940211e-01 4.05345142e-01 7.92349398e-01
9.75621998e-01 -3.58378261e-01 -2.18590975e-01 3.87789309e-01
1.64445019e+00 3.86876643e-01 6.14178061e-01 3.19947973e-02
2.38954827e-01 7.15192854e-01 4.40228939e-01 5.55940151e-01
-2.27081254e-01 5.96124351e-01 2.43887231e-01 1.04767777e-01
6.21524513e-01 7.58678764e-02 1.21553421e-01 2.74156749e-01
6.45159036e-02 -3.09865534e-01 -8.94101799e-01 6.43753350e-01
-1.68881106e+00 -7.39984393e-01 1.30711053e-03 2.68580413e+00
7.79825330e-01 8.36633891e-02 1.81671336e-01 2.31462881e-01
1.09310687e+00 -3.80656689e-01 -3.74976277e-01 -4.30662721e-01
1.77563652e-01 4.07872915e-01 5.94632924e-01 7.43605018e-01
-1.10440862e+00 2.14053601e-01 7.54160023e+00 8.44201684e-01
-8.91260445e-01 -2.22651437e-01 5.74212313e-01 9.01114568e-02
2.30700001e-02 -4.48116362e-02 -1.08135152e+00 1.45217240e-01
1.17926800e+00 -1.02984034e-01 1.95017427e-01 9.34783936e-01
2.23763391e-01 -6.50457263e-01 -1.21163905e+00 5.96093059e-01
-8.64595398e-02 -1.04700732e+00 -3.10535461e-01 1.53924078e-01
5.79296052e-01 -5.29767275e-01 -4.17521819e-02 4.96682748e-02
1.79422542e-01 -1.28542829e+00 5.08637965e-01 7.61837602e-01
5.98567605e-01 -9.01338935e-01 9.56150174e-01 4.54513937e-01
-9.26492572e-01 1.83256716e-01 -5.83170831e-01 -1.63095236e-01
4.11528826e-01 8.57611060e-01 -1.30012739e+00 3.98328245e-01
2.35882327e-01 3.32297236e-01 -3.01654749e-02 1.51274550e+00
-1.08334946e-03 4.66840655e-01 -8.76401544e-01 -4.25345182e-01
-2.42166221e-03 -2.93007314e-01 5.48929691e-01 1.04193580e+00
5.87899804e-01 -5.48384547e-01 1.90456480e-01 1.16488576e+00
3.26710224e-01 8.20604116e-02 -6.80946231e-01 -5.84322400e-02
4.85071242e-01 1.05661798e+00 -7.95414805e-01 -1.96088523e-01
-2.41534427e-01 3.65259796e-01 2.81653017e-01 3.32964420e-01
-7.81434655e-01 -5.01275063e-01 4.21143830e-01 2.14359015e-01
4.45406377e-01 -3.40758026e-01 -1.66970760e-01 -6.96714878e-01
1.57557368e-01 -3.73122841e-01 1.30345240e-01 -3.99235398e-01
-1.15023494e+00 4.13321346e-01 6.35341287e-01 -1.01875973e+00
-7.71711886e-01 -1.00963259e+00 -6.31262660e-01 1.17607975e+00
-1.17992663e+00 -5.00462651e-01 1.11788332e-01 3.82915705e-01
7.91150182e-02 -3.12265128e-01 6.66848540e-01 3.59966844e-01
-5.46608329e-01 1.29822701e-01 1.50301263e-01 -4.23109904e-02
3.50374907e-01 -1.42326450e+00 -9.93283689e-02 9.31552887e-01
-4.66716066e-02 5.53620279e-01 1.26226461e+00 -5.92973530e-01
-9.20121789e-01 -5.86945415e-01 7.81340420e-01 -8.14402383e-03
7.24997878e-01 -2.07926244e-01 -9.14535820e-01 5.03448904e-01
-1.46717668e-01 -1.04977712e-02 2.70278752e-01 -7.83703923e-02
3.01577568e-01 -1.33884907e-01 -1.15185404e+00 1.72171071e-01
2.81910479e-01 -3.91944259e-01 -3.07750612e-01 3.04216474e-01
1.48332983e-01 -2.25734949e-01 -8.03984821e-01 3.85055214e-01
4.37869728e-01 -9.40926433e-01 7.78970718e-01 -5.01781464e-01
2.28696298e-02 -4.26005125e-01 -3.74457359e-01 -9.97740090e-01
2.58943170e-01 -4.34240639e-01 -1.42063368e-02 1.16849101e+00
3.58331293e-01 -7.94430733e-01 7.85946727e-01 7.20903695e-01
1.37662530e-01 -6.62495434e-01 -1.41231251e+00 -8.93847585e-01
-1.31164700e-01 -5.83732963e-01 2.12441117e-01 5.98301232e-01
-2.51964003e-01 1.57781839e-01 -4.50164467e-01 3.49066019e-01
5.72657466e-01 -3.18538278e-01 5.85875630e-01 -1.20407581e+00
-4.40030575e-01 -2.38746911e-01 -5.84847748e-01 -1.10589290e+00
9.19453353e-02 -3.17949116e-01 5.80500066e-01 -1.39354932e+00
-3.00091803e-01 -3.11516911e-01 1.11202322e-01 4.86527272e-02
3.10502201e-01 -1.15827538e-01 -4.64487791e-01 -2.49525785e-01
-3.95214278e-03 5.82281828e-01 7.14600444e-01 3.40182483e-01
-2.02960491e-01 5.47211885e-01 -3.62509966e-01 8.16600442e-01
9.84246075e-01 -5.19855022e-01 -4.21196461e-01 2.38473266e-01
2.62877733e-01 6.60884604e-02 4.93049473e-01 -1.15918922e+00
2.37509102e-01 -2.14721829e-01 4.28289324e-01 -5.80062389e-01
4.03815091e-01 -1.02078760e+00 3.65474597e-02 4.14824262e-02
-1.27289802e-01 4.51852046e-02 1.17111728e-01 4.51680750e-01
-2.08777383e-01 -1.05303597e+00 8.83906960e-01 1.64602831e-01
-3.63199890e-01 -1.17828315e-02 -5.36238909e-01 -1.99821487e-01
9.73293602e-01 -4.99509245e-01 1.78886250e-01 -7.13732421e-01
-7.77068973e-01 -5.38012870e-02 8.63410160e-02 -1.47840008e-01
6.97506964e-01 -1.08384633e+00 -4.88406360e-01 2.93916464e-01
-3.34950447e-01 7.40175396e-02 -2.51240075e-01 9.25566554e-01
-4.95338351e-01 3.31967503e-01 -1.97436646e-01 -5.55221379e-01
-8.04761708e-01 2.24864125e-01 9.51720834e-01 6.62270654e-03
4.76705655e-02 7.11416841e-01 -1.06164459e-02 -4.80876446e-01
2.69891649e-01 -3.17396283e-01 -1.67811945e-01 -2.29533866e-01
4.88489002e-01 3.55447382e-01 -1.46650374e-02 -4.55045611e-01
-1.92981422e-01 1.99305490e-01 1.18715405e-01 -2.80742407e-01
1.30307221e+00 1.11297108e-01 -1.98684216e-01 6.61150753e-01
1.34821367e+00 -7.45091066e-02 -1.23087287e+00 5.64187905e-03
6.92010671e-02 -2.61036366e-01 -2.61643976e-02 -3.52842420e-01
-6.73655808e-01 6.01254046e-01 6.13528550e-01 5.87020397e-01
1.06378138e+00 3.48916762e-02 -4.69692126e-02 5.64587295e-01
1.85595721e-01 -1.06166685e+00 -2.80112356e-01 3.41354460e-01
8.34300518e-01 -8.12994659e-01 8.17277953e-02 -4.15123075e-01
-9.82528031e-02 1.42261076e+00 6.08551204e-01 -4.62253481e-01
8.00663352e-01 6.57804370e-01 -1.77669376e-01 1.07255615e-01
-2.82335281e-01 -2.54081637e-01 6.29706085e-01 7.65497208e-01
4.93620396e-01 -3.34928125e-01 -2.54000306e-01 7.74761885e-02
-1.96454316e-01 -3.25649418e-02 3.83942872e-01 9.56538498e-01
-8.74878228e-01 -1.04301226e+00 -6.98626637e-01 5.45211792e-01
-3.87338191e-01 1.10727414e-01 -1.22835144e-01 1.02667427e+00
4.50423965e-03 7.55626082e-01 2.17680663e-01 1.32105544e-01
2.29226783e-01 9.21096951e-02 5.09514213e-01 -5.19564092e-01
4.91357595e-02 2.28453636e-01 3.04134876e-01 -1.95226982e-01
-5.03736079e-01 -6.64106369e-01 -9.87246335e-01 -1.63622014e-02
-8.66017342e-01 3.22404534e-01 7.05734909e-01 1.20311093e+00
-3.44219863e-01 3.87952417e-01 3.34841162e-01 -8.78095627e-01
-8.04928303e-01 -9.21273589e-01 -1.01876128e+00 -2.30229527e-01
4.58417028e-01 -7.91447937e-01 -3.51295531e-01 3.43526974e-02]
|
[6.825510501861572, 3.8953490257263184]
|
d0c5f5ef-0e62-488c-aa49-43227df058c5
|
classifying-multi-channel-uwb-sar-imagery-via
|
1810.02812
| null |
http://arxiv.org/abs/1810.02812v1
|
http://arxiv.org/pdf/1810.02812v1.pdf
|
Classifying Multi-channel UWB SAR Imagery via Tensor Sparsity Learning Techniques
|
Using low-frequency (UHF to L-band) ultra-wideband (UWB) synthetic aperture
radar (SAR) technology for detecting buried and obscured targets, e.g. bomb or
mine, has been successfully demonstrated recently. Despite promising recent
progress, a significant open challenge is to distinguish obscured targets from
other (natural and manmade) clutter sources in the scene. The problem becomes
exacerbated in the presence of noisy responses from rough ground surfaces. In
this paper, we present three novel sparsity-driven techniques, which not only
exploit the subtle features of raw captured data but also take advantage of the
polarization diversity and the aspect angle dependence information from
multi-channel SAR data. First, the traditional sparse representation-based
classification (SRC) is generalized to exploit shared information of classes
and various sparsity structures of tensor coefficients for multi-channel data.
Corresponding tensor dictionary learning models are consequently proposed to
enhance classification accuracy. Lastly, a new tensor sparsity model is
proposed to model responses from multiple consecutive looks of objects, which
is a unique characteristic of the dataset we consider. Extensive experimental
results on a high-fidelity electromagnetic simulated dataset and radar data
collected from the U.S. Army Research Laboratory side-looking SAR demonstrate
the advantages of proposed tensor sparsity models.
|
['Vishal Monga', 'Tiep Vu', 'Lam Nguyen']
|
2018-10-04
| null | null | null | null |
['sparse-representation-based-classification']
|
['computer-vision']
|
[ 5.42644322e-01 -5.68741381e-01 2.08484456e-01 -3.24632436e-01
-1.00122857e+00 -4.45366591e-01 3.92255634e-01 -3.70596588e-01
6.30484298e-02 7.19045401e-01 3.53087157e-01 -1.31095797e-01
-7.91375399e-01 -6.36777818e-01 -1.09454520e-01 -1.07954741e+00
-5.18541098e-01 2.32456569e-02 -7.78328031e-02 -4.11747336e-01
4.99793366e-02 7.48538673e-01 -1.26805282e+00 5.16655684e-01
7.04255998e-01 1.33932757e+00 1.21442907e-01 3.06768596e-01
2.73994237e-01 6.96510017e-01 -3.64774376e-01 2.38629598e-02
5.06748378e-01 1.52848745e-02 -2.76428282e-01 2.92531043e-01
3.58933359e-01 -8.36416110e-02 -4.13941026e-01 1.16697979e+00
2.84579277e-01 6.02301620e-02 4.64189559e-01 -6.36166990e-01
-5.10723233e-01 3.01172525e-01 -9.37820971e-01 5.62420905e-01
1.53529057e-02 -2.04631865e-01 7.14069486e-01 -1.21744120e+00
2.61784464e-01 1.05254996e+00 6.20768905e-01 -8.26014504e-02
-9.21795249e-01 -6.16677165e-01 -1.13415785e-01 1.83552831e-01
-1.34629834e+00 -3.85914534e-01 1.26241314e+00 -5.61632991e-01
3.44114810e-01 4.20839727e-01 4.39191848e-01 1.09412348e+00
3.69430900e-01 3.72965485e-01 1.41167676e+00 -2.68760026e-01
3.93056720e-02 -2.76922584e-01 5.30083537e-01 4.85487610e-01
7.28530884e-01 3.80775303e-01 -3.69673491e-01 -5.80924392e-01
5.67133904e-01 2.15767592e-01 -7.60647118e-01 -3.90947253e-01
-1.38503492e+00 9.89016712e-01 3.69358003e-01 5.27126968e-01
-7.39753842e-01 -3.36631805e-01 1.10047939e-03 2.77067214e-01
4.80736315e-01 4.87378359e-01 -1.16163857e-01 2.76786625e-01
-8.36114049e-01 2.13794842e-01 5.10208070e-01 4.56836402e-01
7.29993463e-01 9.81924713e-01 1.22052051e-01 8.87427211e-01
4.37071890e-01 1.13418913e+00 2.38430336e-01 -3.30942899e-01
4.52413380e-01 3.72143127e-02 2.05499604e-01 -1.48996747e+00
-4.25173640e-01 -1.30671549e+00 -1.13913703e+00 -1.48729235e-01
1.72679067e-01 -2.77276158e-01 -8.66725326e-01 1.17007387e+00
4.25577581e-01 1.20210633e-01 4.76042777e-01 1.33581066e+00
8.21732759e-01 6.28638744e-01 -4.11780417e-01 -4.48416412e-01
1.36789989e+00 -1.83518663e-01 -7.29092181e-01 -5.02014995e-01
3.63260835e-01 -1.05598342e+00 1.64330930e-01 6.89149976e-01
-4.41086918e-01 -4.82739955e-01 -1.23920453e+00 7.68136621e-01
1.62693188e-01 1.28258362e-01 1.01136100e+00 6.70114338e-01
-7.99858421e-02 8.25916752e-02 -7.04385459e-01 6.82089031e-02
3.20447445e-01 -1.29957438e-01 -2.42729187e-01 -6.41097248e-01
-1.12828684e+00 5.21634996e-01 -1.18146606e-01 6.60050929e-01
-8.01493585e-01 -7.28461802e-01 -6.87239707e-01 -3.12237889e-01
3.19949746e-01 -5.79787791e-02 6.16552234e-01 -7.13605285e-01
-7.44089603e-01 1.42148241e-01 2.01344401e-01 -3.77118915e-01
-2.65343815e-01 -2.23399252e-01 -1.06955230e+00 4.73918527e-01
6.05045408e-02 -4.33386058e-01 1.26198137e+00 -1.36368096e+00
-5.78923114e-02 -6.41662955e-01 -2.02909857e-01 -1.85547203e-01
-7.73134306e-02 -6.97910488e-02 6.55339599e-01 -1.04116118e+00
9.82865334e-01 -7.35843956e-01 -4.15304303e-01 -4.22107726e-01
-2.92889804e-01 6.19317770e-01 1.07289028e+00 -7.18356609e-01
8.80178809e-01 -2.26887083e+00 5.94817810e-02 5.07382512e-01
1.27414733e-01 1.22529387e-01 -1.51187137e-01 5.10090470e-01
-2.49622449e-01 -6.83717430e-01 -4.19365674e-01 5.01154184e-01
-6.35842025e-01 9.74823609e-02 -7.19969273e-01 9.13330615e-01
1.77976355e-01 3.22887719e-01 -5.93469739e-01 5.27155213e-02
-1.10053569e-01 5.53359091e-01 -2.66451716e-01 -6.22896384e-03
2.52225071e-01 8.19272637e-01 -9.68682170e-01 1.30684221e+00
1.20902932e+00 -1.08725682e-01 -8.75892341e-02 -9.51394439e-01
-3.73441994e-01 -3.20565730e-01 -1.36440098e+00 1.33198738e+00
-2.14572132e-01 1.87477425e-01 4.80442077e-01 -1.52561247e+00
1.28432357e+00 2.33870521e-01 6.09862626e-01 -8.47689450e-01
4.72347364e-02 4.02101636e-01 3.19367886e-01 -5.15725315e-01
3.41575950e-01 -4.99403030e-01 -1.13928758e-01 3.03034842e-01
-9.86893773e-02 2.92848796e-02 -2.61469871e-01 -7.50594139e-02
1.02949584e+00 -4.35585737e-01 3.14097017e-01 -5.09132862e-01
5.20569086e-01 4.01908964e-01 7.87374735e-01 6.73965216e-01
1.14304751e-01 4.23220694e-01 -2.88611233e-01 -7.42076159e-01
-6.45619452e-01 -1.03540134e+00 -4.99986798e-01 3.98169905e-01
-1.59498602e-02 1.37100622e-01 2.51383752e-01 -1.45772710e-01
2.48304885e-02 3.35669041e-01 -3.49645019e-01 -6.03501052e-02
-8.05423141e-01 -1.36613178e+00 2.49557555e-01 1.52262047e-01
4.83397275e-01 -2.48328105e-01 -7.45778501e-01 3.23886603e-01
-2.80038238e-01 -1.44845521e+00 3.14028680e-01 1.27900749e-01
-9.76098537e-01 -1.03803062e+00 -5.88580251e-01 -4.14427102e-01
3.97630930e-01 1.03427482e+00 6.69970155e-01 -4.51175928e-01
-7.79383302e-01 5.94488025e-01 -7.45936513e-01 -3.22544247e-01
2.36178011e-01 -8.44740868e-01 3.26874375e-01 9.15282726e-01
-2.96198949e-02 -6.86844528e-01 -3.39327127e-01 3.65042806e-01
-7.37067461e-01 -2.76908785e-01 1.09517443e+00 1.23293030e+00
5.77375948e-01 3.39705884e-01 5.65664053e-01 -5.47155619e-01
2.03878164e-01 -6.63019240e-01 -4.45644617e-01 3.43106613e-02
1.33204371e-01 -1.28689349e-01 4.77013588e-01 -5.13775587e-01
-1.29125190e+00 -2.95495927e-01 2.44797572e-01 -3.93088162e-01
-6.24325185e-04 1.02049339e+00 -1.65468454e-02 -7.43834496e-01
6.08457506e-01 5.26711047e-01 -1.57068133e-01 -6.80134535e-01
-1.47840798e-01 5.66892385e-01 4.81275707e-01 -6.79322779e-01
1.34966457e+00 9.29202259e-01 3.36485833e-01 -1.63018680e+00
-1.30256784e+00 -6.21035218e-01 -3.44083846e-01 -1.71649009e-01
3.76057535e-01 -1.17077124e+00 -2.40539033e-02 2.50625849e-01
-9.95701134e-01 3.47923487e-01 -2.20913589e-02 1.24929929e+00
2.40050573e-02 5.96852660e-01 -2.17642516e-01 -1.11981881e+00
-2.41504446e-01 -6.14704549e-01 7.55007803e-01 -2.09927887e-01
2.86597818e-01 -6.03912473e-01 2.78042145e-02 6.66823506e-01
6.28162086e-01 5.67696154e-01 8.44432831e-01 -4.53940302e-01
-7.28668630e-01 -4.01278764e-01 -2.11016104e-01 3.95677060e-01
2.57968813e-01 -6.79480553e-01 -5.05878508e-01 -6.53829992e-01
8.38126004e-01 -3.50021064e-01 7.32342899e-01 4.25948262e-01
7.85187125e-01 -2.83575356e-01 -2.00271413e-01 8.11946034e-01
1.70593858e+00 2.61133552e-01 4.72257465e-01 2.97704358e-02
7.18980670e-01 6.01034224e-01 1.10738945e+00 7.40511537e-01
-2.65597761e-01 5.49702823e-01 3.95991385e-01 6.58667088e-02
1.16634369e-01 4.52526927e-01 1.35000601e-01 9.38014567e-01
-2.96824366e-01 1.80668831e-01 -8.18354905e-01 4.82119322e-01
-1.36522663e+00 -1.03023827e+00 -4.22186911e-01 2.03552222e+00
7.11250752e-02 -1.77114785e-01 -5.22303700e-01 3.83889884e-01
3.69438440e-01 5.54566264e-01 -2.13882118e-01 1.01991676e-01
-5.51467121e-01 5.49594223e-01 5.02962828e-01 2.81975806e-01
-1.00204670e+00 2.00337052e-01 5.30720758e+00 6.59848154e-01
-1.28393114e+00 1.57028139e-01 2.07566880e-02 1.86284706e-01
-3.42191994e-01 -4.79182694e-03 -4.41280901e-01 2.08456870e-02
5.14208436e-01 1.19123116e-01 5.87061606e-02 5.29405415e-01
-2.44379994e-02 -7.64086396e-02 -4.64329392e-01 1.04615521e+00
2.62176275e-01 -1.15681040e+00 2.96384662e-01 4.20107506e-02
5.18658340e-01 -1.37749976e-02 2.06619486e-01 3.71209942e-02
-1.20265171e-01 -7.45483577e-01 4.61110085e-01 6.94321990e-01
4.94717002e-01 -6.87936485e-01 7.09372997e-01 2.89952546e-01
-1.20301473e+00 -3.65385145e-01 -5.34546256e-01 -2.77782291e-01
2.14679912e-01 1.34489703e+00 -5.15349448e-01 1.04409778e+00
6.78657472e-01 7.56530523e-01 -1.69598609e-01 9.42127049e-01
1.63769498e-01 7.17354953e-01 -1.81973010e-01 3.04650575e-01
5.50316453e-01 -2.45539308e-01 1.04287493e+00 1.05630016e+00
8.12252760e-01 7.08129644e-01 4.47137028e-01 5.64401090e-01
7.41515875e-01 -1.84791148e-01 -7.49554873e-01 -9.97885615e-02
1.12617739e-01 1.49902999e+00 -3.21082115e-01 1.14002846e-01
-5.27514577e-01 2.17017367e-01 -4.10252094e-01 5.64797282e-01
-6.65871024e-01 -2.10185423e-01 7.41549432e-01 2.43101537e-01
6.83391988e-01 -8.46631765e-01 -1.26417622e-01 -1.24544859e+00
1.22235589e-01 -9.58550453e-01 1.94989994e-01 -4.99236703e-01
-1.49145269e+00 7.21691132e-01 1.27922685e-03 -1.82420981e+00
2.78116912e-01 -5.31262755e-01 -4.55850780e-01 7.95711279e-01
-1.81285942e+00 -1.26336432e+00 -4.90244657e-01 7.65549362e-01
3.13382864e-01 -5.02086580e-01 7.71606028e-01 2.45418891e-01
-2.76619762e-01 -8.09066743e-02 2.19239190e-01 1.73092663e-01
2.56928951e-01 -3.32830071e-01 -3.50004882e-01 9.62856948e-01
4.80278069e-03 6.04842246e-01 9.00999367e-01 -7.72018135e-01
-2.18337870e+00 -1.15935266e+00 2.76587188e-01 3.04998845e-01
9.45413768e-01 -1.66634664e-01 -8.77491415e-01 3.73672664e-01
-3.51403236e-01 4.83285517e-01 1.07302463e+00 8.18993077e-02
-6.85223877e-01 -3.17898840e-01 -9.99756992e-01 -5.76287769e-02
6.77599192e-01 -1.45615935e-01 -8.34392250e-01 5.17384827e-01
3.01125616e-01 -8.16979483e-02 -8.81329536e-01 1.02076650e+00
5.16958654e-01 -8.40506077e-01 1.38255060e+00 -6.00672483e-01
5.03095873e-02 -4.19379354e-01 -1.04863846e+00 -1.30786455e+00
-6.49787605e-01 -3.81398618e-01 -1.42580383e-02 5.92589557e-01
5.17691523e-02 -7.00211704e-01 5.00103235e-01 -4.43905115e-01
-3.85530502e-01 -5.56970835e-01 -1.15619648e+00 -1.02364707e+00
-4.33621943e-01 -3.76213223e-01 1.92498207e-01 1.11954796e+00
-2.73145497e-01 3.61232638e-01 -8.31537247e-01 9.58735406e-01
1.51606977e+00 8.25523674e-01 4.15150732e-01 -1.50322223e+00
-6.18226111e-01 4.73584592e-01 -4.40121025e-01 -8.47348034e-01
-2.70947337e-01 -6.72198713e-01 -2.28267118e-01 -1.16332424e+00
-7.91718140e-02 -8.07349384e-01 -2.65931010e-01 1.15264462e-04
3.55049044e-01 4.82010067e-01 1.38332555e-02 4.06939447e-01
-5.32941520e-02 7.21524417e-01 1.29061103e+00 -4.35872167e-01
3.17143649e-01 1.23104811e-01 -6.88001871e-01 7.09109604e-01
5.00067055e-01 -4.53072220e-01 -1.87244281e-01 -6.74124420e-01
7.44274780e-02 5.93824267e-01 7.31937349e-01 -1.36051583e+00
1.14896871e-01 -3.81728709e-01 4.27229077e-01 -7.41720736e-01
7.82817900e-01 -8.55719745e-01 3.00821304e-01 5.49403429e-01
3.46714646e-01 -3.19147050e-01 7.14839483e-03 9.58502829e-01
-5.66701591e-01 4.87508066e-02 9.11219478e-01 -1.43194616e-01
-7.51633584e-01 3.41010213e-01 -3.22536945e-01 -1.77968904e-01
8.02486479e-01 -2.72898227e-01 -3.07982534e-01 -1.83223143e-01
-5.99611402e-01 -1.94458559e-01 -3.94044608e-01 1.95862073e-02
9.98949289e-01 -1.21207237e+00 -1.18808746e+00 5.29356003e-01
3.42992336e-01 -4.73096818e-01 7.99396992e-01 1.20638347e+00
-1.82161927e-02 5.11204183e-01 -4.31758970e-01 -6.99336112e-01
-1.13379121e+00 2.14003503e-01 1.18582919e-02 -1.24498419e-01
-6.99984610e-01 8.02460790e-01 2.19226226e-01 -3.86703387e-02
-3.81985188e-01 -1.07937984e-01 -3.39370131e-01 7.12059885e-02
8.21244657e-01 2.76566029e-01 2.78788418e-01 -9.09703851e-01
-4.50573504e-01 8.86099517e-01 -2.33429112e-02 1.17417097e-01
1.75610030e+00 2.09003285e-01 -2.00881243e-01 3.02967191e-01
1.00931239e+00 3.81093681e-01 -6.66401088e-01 -7.21869707e-01
-2.01954842e-01 -7.65818894e-01 4.19931829e-01 -5.43035507e-01
-1.07416999e+00 7.88900316e-01 4.52138603e-01 1.03562854e-01
1.13575339e+00 -1.38037622e-01 7.29709446e-01 7.42258072e-01
9.01962876e-01 -5.50351799e-01 2.81653881e-01 4.84946162e-01
1.04805899e+00 -1.02885675e+00 3.97842258e-01 -8.01250815e-01
-5.34848690e-01 1.23900867e+00 9.13692862e-02 -3.79998088e-01
7.90211678e-01 2.74889737e-01 2.86192801e-02 -6.45718038e-01
-3.30669940e-01 -1.34914175e-01 2.87063479e-01 7.84833670e-01
8.22742507e-02 2.24867225e-01 -2.50993013e-01 7.81669974e-01
-6.93440661e-02 -5.07920563e-01 6.65495634e-01 1.23802269e+00
-8.38449478e-01 -7.03716040e-01 -1.13848829e+00 6.39842033e-01
-3.89492363e-01 -2.22828351e-02 9.18462202e-02 4.98436987e-01
-1.46378651e-01 9.87703562e-01 -2.83780277e-01 -3.88064593e-01
2.59201348e-01 -3.81701469e-01 5.00124097e-01 -5.43945432e-01
2.86800461e-03 2.74246097e-01 3.03912997e-01 -4.57759827e-01
-6.35091662e-01 -7.75424063e-01 -7.07727313e-01 2.90905237e-01
-4.12187189e-01 4.74340111e-01 7.42939115e-01 7.62063444e-01
3.06871589e-02 4.71380353e-01 1.00025964e+00 -7.46361673e-01
-8.35867107e-01 -9.21974719e-01 -1.20008445e+00 -5.30217215e-02
6.10880315e-01 -1.21294558e+00 -6.85763240e-01 -3.39696705e-01]
|
[6.824368476867676, 1.0212428569793701]
|
c5bed6fa-31e1-4c34-a6d3-f8e9cf0d4db9
|
supervised-nonnegative-matrix-factorization
|
1809.10680
| null |
http://arxiv.org/abs/1809.10680v2
|
http://arxiv.org/pdf/1809.10680v2.pdf
|
Supervised Nonnegative Matrix Factorization to Predict ICU Mortality Risk
|
ICU mortality risk prediction is a tough yet important task. On one hand, due
to the complex temporal data collected, it is difficult to identify the
effective features and interpret them easily; on the other hand, good
prediction can help clinicians take timely actions to prevent the mortality.
These correspond to the interpretability and accuracy problems. Most existing
methods lack of the interpretability, but recently Subgraph Augmented
Nonnegative Matrix Factorization (SANMF) has been successfully applied to time
series data to provide a path to interpret the features well. Therefore, we
adopted this approach as the backbone to analyze the patient data. One
limitation of the raw SANMF method is its poor prediction ability due to its
unsupervised nature. To deal with this problem, we proposed a supervised SANMF
algorithm by integrating the logistic regression loss function into the NMF
framework and solved it with an alternating optimization procedure. We used the
simulation data to verify the effectiveness of this method, and then we applied
it to ICU mortality risk prediction and demonstrated its superiority over other
conventional supervised NMF methods.
|
['Yuan Zhao', 'Yuan Luo', 'Fei Wang', 'Chengsheng Mao', 'Guoqing Chao']
|
2018-09-27
| null | null | null | null |
['icu-mortality']
|
['medical']
|
[ 1.64335504e-01 -2.37026840e-01 -1.18395709e-01 -2.64130563e-01
-6.81844875e-02 -4.26044650e-02 5.48435077e-02 2.05873892e-01
-4.55337256e-01 8.56291533e-01 1.56846464e-01 -4.70631331e-01
-6.71695054e-01 -5.50973594e-01 -1.83061715e-02 -7.75783896e-01
-2.28729352e-01 3.43662590e-01 -7.47718140e-02 -1.52308539e-01
2.59097368e-01 2.54946291e-01 -1.24493396e+00 5.39760925e-02
1.34759665e+00 8.36071134e-01 2.53220588e-01 1.29436582e-01
-2.26267934e-01 1.03559172e+00 -2.78782725e-01 5.07997349e-02
1.90551758e-01 -4.66452926e-01 -6.04633033e-01 -4.87898961e-02
-6.63090050e-01 -2.41150349e-01 -1.23762332e-01 7.72453368e-01
4.51074779e-01 2.20678344e-01 6.88595116e-01 -1.53484440e+00
-2.19930530e-01 4.60550636e-01 -5.54438174e-01 3.05878282e-01
1.31688178e-01 -2.71639436e-01 8.02018464e-01 -6.37585938e-01
1.41756773e-01 1.11450148e+00 6.22798026e-01 5.89747608e-01
-1.01429617e+00 -6.01905346e-01 3.30720454e-01 3.13779503e-01
-1.01936316e+00 2.84691229e-02 7.57799387e-01 -5.66069901e-01
5.54037690e-01 2.10989833e-01 6.93072557e-01 6.89971805e-01
3.36125612e-01 6.44123197e-01 1.18333650e+00 -2.63210833e-01
4.86505888e-02 1.89560950e-01 3.72953415e-01 5.76347589e-01
3.56562376e-01 -1.46129709e-02 -7.30492026e-02 -3.22899222e-01
6.15169168e-01 7.57935405e-01 -4.79956985e-01 -7.91936889e-02
-1.38021457e+00 7.13576138e-01 3.05913597e-01 2.85506964e-01
-2.85105705e-01 -2.59423107e-01 4.19596493e-01 2.14746669e-01
3.97031188e-01 2.95514435e-01 -6.79742217e-01 -2.98899382e-01
-5.71753800e-01 -2.17589706e-01 5.84284186e-01 3.18176061e-01
4.65985179e-01 -8.21469054e-02 1.94918402e-02 7.19339430e-01
3.54197234e-01 3.00068736e-01 8.15430701e-01 -6.40832782e-01
5.88872731e-01 1.01929748e+00 1.26786783e-01 -1.26859152e+00
-6.46964133e-01 -3.85701358e-01 -1.31078959e+00 1.18197195e-01
3.18438202e-01 -2.70731717e-01 -6.98619723e-01 1.82648647e+00
1.50865972e-01 2.93877125e-01 1.35421351e-01 1.06072628e+00
4.72870976e-01 5.78848600e-01 1.32490486e-01 -9.38485026e-01
1.17406082e+00 -7.69823432e-01 -1.05540860e+00 1.67750672e-01
7.40278840e-01 -6.58969223e-01 9.51565385e-01 5.69175780e-01
-4.58398521e-01 -3.00040513e-01 -7.70704627e-01 4.36168224e-01
-9.44331288e-02 3.52240741e-01 8.87698531e-01 2.60724932e-01
-5.43684602e-01 8.64759803e-01 -1.03261220e+00 -3.27905446e-01
2.13248417e-01 5.09972632e-01 -4.05826956e-01 7.97308311e-02
-1.40913737e+00 7.95906246e-01 5.87179184e-01 5.54115534e-01
-3.48938733e-01 -3.86376381e-01 -5.24393559e-01 2.92208474e-02
3.60448748e-01 -8.43533337e-01 1.00302875e+00 -1.12279308e+00
-1.23015797e+00 4.91893478e-02 -3.13213617e-01 -2.02101737e-01
6.36459053e-01 -3.75041723e-01 -5.13742685e-01 -1.26296692e-02
3.21537978e-03 -4.76814508e-02 7.32026756e-01 -8.99635077e-01
-5.85565031e-01 -3.81002784e-01 -3.20670232e-02 1.80449918e-01
-6.19921088e-01 -1.90373078e-01 1.34755105e-01 -7.04306543e-01
4.83805090e-01 -8.97912681e-01 -5.43040991e-01 -3.16757351e-01
-2.18823418e-01 -9.63812023e-02 8.29856515e-01 -6.60293579e-01
1.73095584e+00 -2.14899015e+00 2.72440106e-01 2.06435636e-01
3.32289457e-01 1.70231879e-01 2.48841763e-01 5.42215228e-01
-5.09881973e-01 1.64148420e-01 -4.05500025e-01 -2.35684931e-01
-4.56786126e-01 2.85282582e-01 -2.35855103e-01 2.70893633e-01
1.97056040e-01 3.73533458e-01 -9.96975958e-01 -7.03589082e-01
2.15394869e-01 3.43818843e-01 -4.56895381e-01 4.78714049e-01
1.16330214e-01 7.07243860e-01 -8.70893836e-01 2.85107404e-01
4.85867709e-01 -3.72583807e-01 1.72695637e-01 -2.95482188e-01
-1.08841181e-01 -1.87832266e-01 -1.34483945e+00 1.29762721e+00
-3.31279367e-01 1.85355037e-01 -2.23763362e-01 -1.27495301e+00
8.52543533e-01 5.56181967e-01 1.03642261e+00 -2.94834197e-01
2.13614121e-01 3.95430744e-01 1.91533625e-01 -9.02931333e-01
-1.59393158e-02 -4.25659060e-01 2.52828121e-01 5.91269732e-01
-5.47052860e-01 4.07477915e-01 -1.88703258e-02 2.22235486e-01
8.52503181e-01 3.68679948e-02 5.06264389e-01 -2.66658515e-01
8.94115090e-01 1.45013586e-01 1.05257642e+00 2.08899528e-01
-1.47521989e-02 5.27046442e-01 5.00247598e-01 -7.97948718e-01
-4.45983142e-01 -7.76132345e-01 -2.35646233e-01 4.67113107e-01
7.63792247e-02 -4.93052840e-01 -3.31530571e-01 -8.20153415e-01
-9.76539999e-02 3.85455370e-01 -5.89255869e-01 -3.36713314e-01
-4.23853606e-01 -1.27278709e+00 4.29114364e-02 5.47891796e-01
4.16162729e-01 -8.81487548e-01 -5.62849998e-01 4.17349398e-01
-3.54859740e-01 -6.43811524e-01 -2.34408796e-01 2.88443089e-01
-1.24348581e+00 -1.28259826e+00 -4.95402753e-01 -5.60655773e-01
1.02539980e+00 4.46508229e-01 7.99794495e-01 3.99492174e-01
-2.11181462e-01 -1.20106433e-02 -5.48397124e-01 -5.22200465e-01
-2.72148401e-01 1.57071557e-02 5.54963827e-01 1.79721594e-01
1.06091261e-01 -6.87549531e-01 -7.57645607e-01 4.02110070e-01
-1.14164400e+00 1.80962682e-01 6.43615425e-01 1.09500396e+00
3.83661091e-01 4.71215069e-01 1.07205880e+00 -1.03430259e+00
8.10303628e-01 -5.44054508e-01 -3.13253790e-01 2.11958647e-01
-1.19377339e+00 3.87113653e-02 9.70400333e-01 -4.25668955e-01
-1.00395226e+00 1.15448698e-01 1.14693500e-01 -3.23429644e-01
1.85492426e-01 7.79550612e-01 -1.64248183e-01 3.05856884e-01
1.36523008e-01 2.02564031e-01 3.41228247e-01 -5.70704520e-01
-1.55400187e-01 7.64193416e-01 -4.86586578e-02 -5.14218628e-01
6.73462629e-01 3.22772026e-01 3.17559749e-01 -4.55864161e-01
-8.30319762e-01 -4.40997064e-01 -6.71733737e-01 -1.35868952e-01
8.09632003e-01 -6.94469333e-01 -9.09394801e-01 2.84387916e-01
-1.03740752e+00 1.96264699e-01 -2.11667328e-04 9.04914737e-01
-4.12039757e-01 8.51140141e-01 -5.75058043e-01 -1.08415377e+00
-4.91097718e-01 -9.91548240e-01 4.93982136e-01 6.12301519e-03
-9.87794325e-02 -1.03876519e+00 1.04750000e-01 2.56899059e-01
2.46414080e-01 2.86639124e-01 1.29107308e+00 -5.52341282e-01
-1.30924910e-01 -1.38571084e-01 -2.75752574e-01 5.21849632e-01
6.21508241e-01 1.17959455e-01 -6.39662862e-01 -3.00054610e-01
4.17456031e-01 5.39682917e-02 6.48377657e-01 1.63208783e-01
1.24063623e+00 -3.30748081e-01 -3.26621234e-01 4.48705405e-01
1.36050427e+00 5.07730484e-01 4.15806651e-01 4.07686234e-01
6.44497216e-01 4.30239260e-01 9.66656685e-01 7.38267958e-01
2.53999352e-01 4.56595957e-01 4.10514504e-01 -2.87076980e-01
5.92132151e-01 -6.41728565e-02 1.58532038e-01 1.37469912e+00
-4.80983317e-01 -1.31836668e-01 -8.72503698e-01 6.89242184e-02
-2.36813092e+00 -7.90618002e-01 -4.48789477e-01 2.25702143e+00
6.59522474e-01 1.00256771e-01 7.19584748e-02 5.67958057e-01
5.58359385e-01 -2.33825684e-01 -3.75296414e-01 -2.76701421e-01
7.05787241e-02 -3.88064802e-01 5.80588169e-02 -1.93743445e-02
-1.01558208e+00 2.89375901e-01 5.96660280e+00 6.39168859e-01
-1.13273048e+00 -4.63157110e-02 6.13932729e-01 2.14043126e-01
-1.60803244e-01 8.78513902e-02 -3.43634993e-01 6.43626928e-01
5.56315064e-01 -1.66144058e-01 4.39973742e-01 6.53489470e-01
8.27557206e-01 8.12558550e-03 -9.34729815e-01 1.08037031e+00
-2.71574229e-01 -6.13047123e-01 9.11381468e-02 -7.46317357e-02
4.90443647e-01 -5.42257845e-01 -3.32461119e-01 1.31811172e-01
-1.70786351e-01 -1.07243490e+00 8.50043893e-02 7.82178223e-01
4.19844687e-01 -5.99735796e-01 1.16329122e+00 7.61967301e-01
-1.10993326e+00 -3.50112259e-01 -4.03535128e-01 -5.30623019e-01
2.53511280e-01 7.73466229e-01 -7.23654330e-01 8.89424205e-01
4.35280323e-01 9.93417263e-01 -3.76865178e-01 9.62974131e-01
-7.72963017e-02 5.46190739e-01 -3.25595170e-01 1.26231372e-01
-4.21419367e-02 -5.40035725e-01 4.33720529e-01 6.78248882e-01
5.37066162e-01 2.72560149e-01 3.73789757e-01 5.10497630e-01
4.10492092e-01 5.99883556e-01 -4.14155245e-01 -1.77856356e-01
6.41115010e-02 1.22596884e+00 -7.55087376e-01 -8.42426941e-02
-4.89304513e-01 6.94672644e-01 1.45875722e-01 2.56445020e-01
-9.20792937e-01 -2.15739429e-01 2.98188388e-01 1.44234821e-01
-4.72299904e-01 -1.87211812e-01 -3.05853993e-01 -1.62318015e+00
5.63100688e-02 -7.97643423e-01 7.16599882e-01 -4.37831283e-01
-1.41362786e+00 7.07400262e-01 -1.32874057e-01 -1.83838391e+00
-7.72310346e-02 -3.29731166e-01 -6.26110673e-01 8.64312828e-01
-1.49581671e+00 -7.54723012e-01 -5.29707551e-01 6.69726849e-01
4.18132186e-01 -1.82130978e-01 9.19087887e-01 5.95589757e-01
-1.02430308e+00 1.67259037e-01 2.34292373e-01 7.45917764e-03
5.88196278e-01 -1.06176364e+00 -5.62481225e-01 7.09631145e-01
-3.75058562e-01 7.76240110e-01 5.90344071e-01 -6.89377964e-01
-1.28872168e+00 -9.86584127e-01 7.50046611e-01 -1.66341677e-01
6.25439048e-01 1.72500610e-01 -9.86342728e-01 4.67714310e-01
-1.94149315e-01 -2.12625057e-01 8.33778620e-01 1.99457854e-02
2.16195881e-01 -2.81245440e-01 -9.17876601e-01 5.68710029e-01
9.02910888e-01 8.99338648e-02 -8.79004240e-01 4.20418769e-01
7.23120451e-01 1.30140141e-01 -9.93268371e-01 7.31299937e-01
4.50404376e-01 -8.24034989e-01 6.47557378e-01 -8.61163020e-01
4.59414929e-01 -5.54532468e-01 1.31684273e-01 -1.33463025e+00
-5.11296868e-01 -3.70254219e-01 5.13983630e-02 1.34230053e+00
3.45363528e-01 -1.06542718e+00 5.63982308e-01 6.57338738e-01
1.07710853e-01 -1.05912089e+00 -8.07376266e-01 -6.28380597e-01
-2.49698341e-01 -3.47328752e-01 5.57650924e-01 9.91583943e-01
2.92439818e-01 4.00160491e-01 -6.27052486e-01 -1.22946864e-02
4.93327111e-01 3.08528394e-01 4.12337303e-01 -1.49819314e+00
-3.52014512e-01 -1.36609390e-01 -4.27626282e-01 -5.96180141e-01
-5.27515076e-02 -6.26408160e-01 -1.53899312e-01 -1.63992703e+00
3.06076854e-01 -8.37482810e-01 -8.19391251e-01 4.85741585e-01
-5.81866622e-01 -3.04677516e-01 7.71609172e-02 6.59638047e-01
-3.44441891e-01 6.60439551e-01 1.31088305e+00 6.51305616e-02
-4.79879916e-01 3.40634853e-01 -5.81163466e-01 6.51335657e-01
8.41908038e-01 -5.74473143e-01 -8.40839505e-01 -3.20435256e-01
3.76892209e-01 4.27750707e-01 6.84447736e-02 -9.57115114e-01
-4.27901037e-02 -5.38198233e-01 2.54636705e-01 -4.88575935e-01
1.06059276e-01 -1.43421173e+00 3.72007281e-01 7.20304847e-01
-2.00230367e-02 2.90587872e-01 -7.99683630e-02 6.82747364e-01
-4.89867568e-01 -3.04947998e-02 4.22397345e-01 -6.86183050e-02
-6.47146046e-01 3.41709465e-01 -2.26154819e-01 -4.02955785e-02
1.14014482e+00 -9.96094868e-02 -6.08137548e-02 -3.49137723e-01
-8.63831580e-01 4.33507949e-01 2.64561653e-01 3.59022945e-01
7.41405964e-01 -1.28873193e+00 -5.18321514e-01 3.46734524e-01
-1.62783457e-04 4.82312404e-02 3.99582535e-01 1.43523979e+00
-5.34547448e-01 2.54357755e-01 -1.21674635e-01 -4.59307253e-01
-1.18832195e+00 8.43931913e-01 2.48502135e-01 -3.46463263e-01
-6.05522215e-01 1.84531391e-01 2.16571033e-01 -1.57482848e-01
1.13976210e-01 -3.36319476e-01 -7.25332081e-01 1.22001216e-01
4.52511668e-01 4.35499191e-01 -6.04167804e-02 -4.44431961e-01
-4.44933981e-01 7.34606385e-01 2.22498924e-01 3.08522791e-01
1.41990709e+00 -2.26800486e-01 -4.01900619e-01 6.19568229e-01
1.10131180e+00 -2.08605886e-01 -6.89882576e-01 4.25130501e-02
1.65041089e-02 -4.44968849e-01 -1.06088184e-01 -4.97695893e-01
-1.17922783e+00 9.68983054e-01 5.04893839e-01 3.22531521e-01
1.51004481e+00 -6.59685791e-01 8.50451052e-01 4.02887732e-01
4.48586643e-01 -7.95672596e-01 -1.51479825e-01 2.52189696e-01
4.41734254e-01 -1.30670846e+00 8.45677257e-02 -6.73086941e-01
-6.86165631e-01 1.52324760e+00 5.28638601e-01 1.10751711e-01
9.85524297e-01 -9.00897309e-02 2.95142710e-01 1.70773253e-01
-7.47481346e-01 8.92988741e-02 1.78872064e-01 8.48341957e-02
5.12020886e-01 3.85197066e-02 -8.76406014e-01 1.03332531e+00
2.17310786e-01 5.05582750e-01 3.94312948e-01 8.62352133e-01
-2.62004465e-01 -1.17526519e+00 -3.05242151e-01 8.17629397e-01
-6.63363993e-01 3.87240872e-02 1.11716330e-01 4.92786974e-01
1.91609457e-01 1.15793335e+00 -3.82037669e-01 -6.99638903e-01
3.86116505e-01 6.58114627e-02 -1.55025378e-01 -4.94219512e-01
-3.97573411e-01 2.15021074e-02 -2.56691337e-01 -3.14980477e-01
-5.16777575e-01 -5.59533477e-01 -1.55918813e+00 2.07058433e-02
-3.62071246e-01 6.96846128e-01 3.15965891e-01 9.82496083e-01
3.23367804e-01 6.52710378e-01 1.05426872e+00 -2.73291111e-01
-8.11627448e-01 -8.64529788e-01 -6.58879280e-01 8.10536027e-01
1.97990030e-01 -8.48065972e-01 -7.21469104e-01 -2.68289521e-02]
|
[7.73414421081543, 3.830409288406372]
|
1360fca2-8031-477c-bd7f-f1684115170d
|
unituebingencl-at-semeval-2020-task-7-humor
| null | null |
https://aclanthology.org/2020.semeval-1.139
|
https://aclanthology.org/2020.semeval-1.139.pdf
|
UniTuebingenCL at SemEval-2020 Task 7: Humor Detection in News Headlines
|
This paper describes the work done by the team UniTuebingenCL for the SemEval 2020 Task 7: {``}Assessing the Funniness of Edited News Headlines{''}. We participated in both sub-tasks: sub-task A, given the original and the edited headline, predicting the mean funniness of the edited headline; and sub-task B, given the original headline and two edited versions, predicting which edited version is the funnier of the two. A Ridge Regression model using Elmo and Glove embeddings as well as Truncated Singular Value Decomposition was used as the final model. A long short term memory model recurrent network (LSTM) served as another approach for assessing the funniness of a headline. For the first sub-task, we experimented with the extraction of multiple features to achieve lower Root Mean Squared Error. The lowest Root Mean Squared Error achieved was 0.575 for sub-task A, and the highest Accuracy was 0.618 for sub-task B.
|
['Lea Gr{\\"u}ner', 'Charlotte Ammer']
|
2020-12-01
| null | null | null |
semeval-2020
|
['humor-detection']
|
['natural-language-processing']
|
[-2.39152595e-01 4.67625380e-01 1.67844057e-01 -3.87117594e-01
-9.39354777e-01 -3.94522935e-01 8.58122647e-01 4.43110913e-01
-8.17782819e-01 6.28897250e-01 4.97823626e-01 -2.34824359e-01
-1.31147146e-01 -5.89472830e-01 -5.48441410e-01 -3.95625114e-01
-1.76692054e-01 2.26090387e-01 -1.08216681e-01 -2.79310405e-01
5.67634165e-01 3.88903879e-02 -1.42994702e+00 4.87936020e-01
5.79447627e-01 1.32017112e+00 1.30633071e-01 7.86186814e-01
-1.08709767e-01 1.20421791e+00 -8.39244604e-01 -5.80652475e-01
3.02644726e-02 -4.87258166e-01 -8.21092486e-01 -2.02454671e-01
4.42379981e-01 -8.16263780e-02 -2.43452176e-01 8.17327619e-01
3.10619026e-01 4.68338579e-01 7.21581042e-01 -9.15386975e-01
-6.16410375e-01 6.97287917e-01 -2.59177864e-01 3.03407758e-01
4.19822991e-01 -2.17761531e-01 1.34564042e+00 -9.39947367e-01
5.93080521e-01 9.38393235e-01 8.95400524e-01 3.53312314e-01
-1.20926082e+00 -4.17710632e-01 -1.05807424e-01 2.16905236e-01
-9.43403363e-01 -4.50754046e-01 2.68588513e-01 -7.70943105e-01
1.10893476e+00 2.35014170e-01 3.47075433e-01 1.31614268e+00
4.77435291e-01 5.65677285e-01 1.22949076e+00 -2.79758155e-01
2.38654047e-01 5.61425447e-01 6.02460980e-01 4.74124849e-01
-3.58755022e-01 1.11510485e-01 -5.36503077e-01 -1.62824586e-01
1.29538909e-01 -2.15721488e-01 -1.66634008e-01 4.95487511e-01
-1.17160356e+00 1.13336527e+00 5.21536887e-01 5.34024358e-01
-7.11726606e-01 1.07675456e-01 6.76445246e-01 6.46340489e-01
7.43841112e-01 8.29624057e-01 -4.83583897e-01 -5.69380164e-01
-1.33382869e+00 3.90684366e-01 1.08636177e+00 5.44930339e-01
4.55163002e-01 -1.99707672e-01 -4.49477911e-01 1.00088513e+00
7.66672567e-02 -8.84011090e-02 5.79363048e-01 -8.23267817e-01
5.04715264e-01 3.11909616e-01 4.06571925e-01 -1.14100921e+00
-5.31165957e-01 -6.32109582e-01 -4.49371576e-01 3.33201528e-01
5.26243448e-01 -5.37245810e-01 -6.93109393e-01 1.37786102e+00
-2.93271661e-01 -2.64988393e-01 -2.84590602e-01 8.17176163e-01
8.27447653e-01 9.65074360e-01 -3.33552510e-02 -4.16102946e-01
1.30331123e+00 -1.07705474e+00 -8.69964898e-01 -1.95825875e-01
7.44023919e-01 -1.02808523e+00 1.05742860e+00 5.53510070e-01
-1.03719234e+00 -4.83536631e-01 -1.23837483e+00 -1.06706999e-01
-5.41800439e-01 1.83559507e-01 -3.76415513e-02 3.14425170e-01
-9.02224600e-01 9.97948527e-01 -4.83305901e-01 -1.53229728e-01
2.65673995e-02 -2.06805170e-01 -3.61544937e-01 3.73767436e-01
-1.33624637e+00 1.27662981e+00 1.62463173e-01 -6.36874810e-02
-6.81847990e-01 -7.63459086e-01 -6.07017338e-01 9.49443281e-02
1.49255306e-01 -6.40252084e-02 1.33399177e+00 -9.40071881e-01
-1.28956342e+00 1.06241453e+00 1.01406418e-01 -5.86686909e-01
7.49256670e-01 -6.34821475e-01 -6.44100904e-01 -4.44902301e-01
2.68853366e-01 2.49102578e-01 7.03112841e-01 -7.65923798e-01
-6.63726151e-01 -4.03042614e-01 -2.30073616e-01 3.74499522e-02
-3.26248586e-01 3.54499012e-01 1.99894104e-02 -8.29788566e-01
-2.79128224e-01 -6.93196356e-01 4.07235250e-02 -4.79644418e-01
-3.23620170e-01 -4.14969712e-01 3.95074248e-01 -1.29778111e+00
1.69371283e+00 -2.30650473e+00 6.91662282e-02 -4.05065753e-02
2.87454337e-01 6.92668334e-02 1.77811310e-02 6.05188191e-01
-7.21346810e-02 1.40357718e-01 2.12569520e-01 -4.59666967e-01
3.66876274e-02 -2.77462631e-01 -2.24039748e-01 3.06194067e-01
6.08266741e-02 3.79251122e-01 -7.88710177e-01 9.56776813e-02
-2.27799207e-01 3.05298209e-01 -1.14391170e-01 2.36342549e-01
-1.90205142e-01 -3.54509354e-02 -1.24881886e-01 -1.21104442e-01
1.95067935e-02 8.94838944e-02 -3.38735849e-01 -1.13461860e-01
-4.82763886e-01 4.49849546e-01 -8.89065742e-01 1.28217435e+00
-6.49495244e-01 1.24626052e+00 4.32287864e-02 -4.63189900e-01
1.33119690e+00 3.96278083e-01 1.38240859e-01 -8.65485132e-01
2.92238772e-01 1.97851509e-01 -3.21633130e-01 -6.13357663e-01
9.29853737e-01 -3.09194237e-01 -3.13082308e-01 6.20968401e-01
-6.68888562e-04 2.75599420e-01 1.28031135e-01 2.51076341e-01
1.18144202e+00 3.09556089e-02 2.12863952e-01 -3.00071597e-01
1.31060347e-01 -2.70632654e-01 3.87747407e-01 7.81932056e-01
-2.61278272e-01 7.20561564e-01 8.98217678e-01 -6.60845995e-01
-1.06101382e+00 -6.76280260e-01 7.06525594e-02 1.50693250e+00
-3.28661859e-01 -6.39083087e-01 -7.36419797e-01 -5.69890976e-01
-1.18909456e-01 1.66714156e+00 -9.60119605e-01 6.91991374e-02
-3.23853582e-01 -3.45982254e-01 5.67951083e-01 2.44780496e-01
2.33777121e-01 -1.07970762e+00 -7.85601079e-01 4.11655873e-01
-4.16925073e-01 -7.63610959e-01 -5.73684931e-01 4.96596366e-01
-4.27429587e-01 -6.76931679e-01 -6.66370392e-01 -4.81461376e-01
6.62991256e-02 -4.52316105e-01 9.40751731e-01 -2.26959199e-01
-1.31951883e-01 -1.63207754e-01 -5.20518839e-01 -3.60341877e-01
-4.51342285e-01 -2.02481207e-02 1.17834508e-02 2.54988641e-01
3.29199702e-01 -2.37392887e-01 -1.39235511e-01 1.91833586e-01
-5.72473764e-01 -1.47628114e-01 3.08863074e-01 8.29808295e-01
7.93570206e-02 -2.42430612e-01 5.01665354e-01 -9.45985138e-01
1.08868015e+00 -6.59752905e-01 -1.37178913e-01 9.21043605e-02
-7.05177605e-01 1.29833862e-01 7.71169007e-01 -1.15987778e-01
-9.34030592e-01 -5.44166863e-01 -2.64273405e-01 1.72450561e-02
5.71621284e-02 7.72890747e-01 3.03846210e-01 6.15228295e-01
1.12703252e+00 -5.55087999e-02 3.74446437e-02 -7.20725536e-01
1.82530940e-01 9.83096659e-01 4.94493872e-01 -2.14470513e-02
3.86063904e-01 -3.03928494e-01 -4.87837046e-01 -9.28016961e-01
-1.10386789e+00 -4.68678474e-01 -3.88183355e-01 -5.80377758e-01
9.12088156e-01 -6.44607306e-01 -4.49540466e-01 3.38337004e-01
-1.23515821e+00 -3.13495874e-01 -1.81534812e-01 5.61639905e-01
-5.59176803e-01 -9.16558430e-02 -5.01452506e-01 -6.52015567e-01
-4.73780066e-01 -8.13534200e-01 4.49877113e-01 -2.12854166e-02
-8.71003270e-01 -8.66318166e-01 2.88681626e-01 5.60827613e-01
4.78918135e-01 5.28623700e-01 9.42265570e-01 -1.21018898e+00
2.54325360e-01 -7.04370677e-01 -2.66374230e-01 5.11635125e-01
-1.44179583e-01 3.85652073e-02 -9.65925872e-01 -1.91713020e-01
9.46631655e-02 -4.06230390e-01 8.17696095e-01 4.02190298e-01
7.14447320e-01 -5.70445836e-01 1.56110421e-01 2.48061553e-01
1.10145426e+00 -7.53144175e-02 7.77161777e-01 8.09548676e-01
3.50191087e-01 5.75652480e-01 8.46620977e-01 7.08692372e-01
1.69303000e-01 7.19497979e-01 1.01619773e-01 2.59396374e-01
-5.11673577e-02 -3.58788282e-01 8.05786431e-01 6.56984568e-01
3.56451645e-02 -1.28541619e-01 -8.13826680e-01 5.36047637e-01
-1.70062876e+00 -1.08622169e+00 -4.05326158e-01 2.22564673e+00
5.82619011e-01 5.68067074e-01 3.58598292e-01 1.64308563e-01
5.46987534e-01 4.60382402e-01 -1.29093081e-01 -1.00735939e+00
2.00560108e-01 -1.98986992e-01 3.61641616e-01 6.31125093e-01
-1.07309198e+00 5.31155944e-01 6.36858892e+00 9.05230641e-01
-8.31081808e-01 1.50893494e-01 5.96119821e-01 -3.55562776e-01
-1.14350483e-01 -4.47382815e-02 -7.93944895e-01 8.27117920e-01
1.64775622e+00 -1.89839929e-01 4.24588025e-01 8.29163194e-01
4.20506030e-01 -2.97742456e-01 -1.07757473e+00 6.80487037e-01
3.71086180e-01 -1.17736876e+00 -3.51689786e-01 -1.22466445e-01
4.58310425e-01 2.58169055e-01 -1.34250671e-01 6.29592419e-01
8.74884576e-02 -1.00874531e+00 9.97416675e-01 8.97384048e-01
5.25366843e-01 -8.66483867e-01 9.62862790e-01 6.62265956e-01
-5.40231049e-01 -1.47592813e-01 -7.23344088e-02 -3.82558614e-01
2.55117595e-01 8.18500042e-01 -7.45171964e-01 2.07461223e-01
7.29106665e-01 6.02324963e-01 -6.18740022e-01 9.07301784e-01
-1.72523499e-01 6.34371877e-01 3.90572171e-03 -4.65570271e-01
4.67052311e-01 -1.53841063e-01 8.84165347e-01 1.57276666e+00
1.02166250e-01 -3.29943568e-01 -1.21824317e-01 6.30504072e-01
-1.84775844e-01 1.14082538e-01 -3.46283466e-01 -2.61267930e-01
3.68870407e-01 1.24420714e+00 -4.22280282e-01 -1.51854917e-01
-1.47170857e-01 9.71457005e-01 5.47650337e-01 1.94370627e-01
-8.12094331e-01 -1.03259623e+00 2.22552165e-01 3.76842260e-01
1.91142201e-01 7.51894414e-02 -5.75146079e-01 -8.20105672e-01
-1.47174761e-01 -7.49671638e-01 3.26301515e-01 -7.91600585e-01
-1.16409218e+00 8.80822599e-01 -2.68477261e-01 -9.40405011e-01
-1.14716582e-01 -3.16900432e-01 -8.59614432e-01 9.43497121e-01
-7.99028099e-01 -6.74490988e-01 -2.57165879e-01 -1.19767025e-01
6.31908596e-01 -4.06902939e-01 8.61700296e-01 2.86687493e-01
-4.39480036e-01 6.65297091e-01 3.70263815e-01 3.34166028e-02
7.32077241e-01 -1.21701860e+00 2.25069553e-01 4.86707121e-01
-6.03686087e-02 3.54497343e-01 1.23277688e+00 -5.25486887e-01
-7.57604420e-01 -1.12746227e+00 1.66129458e+00 -4.60888922e-01
9.91656184e-01 -1.91525191e-01 -6.73094690e-01 6.91691458e-01
1.16199791e-01 -6.40337765e-01 7.25853086e-01 4.27947104e-01
-2.39703491e-01 1.61874555e-02 -9.55830336e-01 4.44235951e-01
2.81368911e-01 -6.83116198e-01 -9.38552618e-01 2.21850350e-01
6.68937445e-01 -1.43349603e-01 -1.09185064e+00 -1.00490429e-01
6.26657307e-01 -9.93199348e-01 4.22568887e-01 -6.63127959e-01
1.02708399e+00 1.40858918e-01 -1.59327090e-01 -1.53740203e+00
-4.46186125e-01 -6.08247936e-01 6.83803558e-02 1.25299323e+00
8.07645440e-01 -3.43090713e-01 3.39072496e-01 5.89933872e-01
-1.67708233e-01 -8.15891266e-01 -8.44211042e-01 -7.73828149e-01
1.90803930e-02 -4.04320210e-01 -1.14779502e-01 7.61955798e-01
9.92150977e-02 6.46555305e-01 -6.76367760e-01 -5.05862415e-01
3.48788440e-01 -3.02660435e-01 6.29951537e-01 -1.02573872e+00
-2.62022465e-01 -5.88063240e-01 -2.88159460e-01 -6.92014337e-01
-9.81272832e-02 -9.42217827e-01 2.45935872e-01 -1.45296633e+00
5.01771644e-02 1.26155689e-01 -3.68384123e-01 1.59134537e-01
-1.08213603e-01 -1.08743072e-01 2.53209889e-01 3.04931235e-02
-4.30693507e-01 3.98168236e-01 6.04357362e-01 1.35445446e-01
-3.87399584e-01 3.60537678e-01 -7.10729122e-01 6.10705733e-01
6.30231619e-01 -7.22370684e-01 8.40677321e-02 -2.35378295e-01
5.58852851e-01 1.82658970e-01 3.21559399e-01 -8.10770273e-01
2.29762286e-01 1.50458336e-01 2.47467041e-01 -5.82762778e-01
2.83199936e-01 -4.39320773e-01 1.48605928e-01 2.75268137e-01
-9.44229007e-01 1.61278129e-01 -2.69214977e-02 5.52312493e-01
-2.51094759e-01 -6.45953953e-01 7.24438846e-01 -1.56487357e-02
-3.74959260e-01 -1.54955044e-01 -7.59142280e-01 9.67456847e-02
9.85329449e-01 -2.18696624e-01 -8.46096650e-02 -4.61379826e-01
-1.10130966e+00 1.49470195e-01 1.07248157e-01 5.74332237e-01
5.11349201e-01 -1.07658517e+00 -9.99333680e-01 -9.55795217e-03
-3.95224355e-02 -7.42975473e-01 4.58627045e-02 8.50579381e-01
-4.76443291e-01 1.52168483e-01 -2.65539795e-01 1.05670698e-01
-1.32564926e+00 1.46391630e-01 1.67641968e-01 -3.30105275e-01
-5.78777909e-01 9.50246453e-01 -4.70049262e-01 -4.11966354e-01
5.05513608e-01 1.44225031e-01 -5.21778941e-01 7.74803817e-01
8.72280359e-01 1.03980935e+00 2.74357617e-01 -7.20477581e-01
-1.41828209e-01 1.59884304e-01 -4.68640506e-01 -3.82252574e-01
1.64817905e+00 -4.27627042e-02 -4.95433137e-02 1.08516693e+00
1.63377857e+00 -2.20653757e-01 -9.00886774e-01 -1.44353196e-01
4.32066530e-01 -2.99436063e-01 4.29036826e-01 -1.07214737e+00
-4.39252079e-01 7.96777189e-01 3.32775891e-01 5.31942606e-01
6.67180061e-01 -3.25905204e-01 7.64901221e-01 4.03523922e-01
3.05230524e-02 -1.49948251e+00 -9.18262675e-02 8.62116635e-01
1.33342826e+00 -9.41518784e-01 6.30492344e-02 4.47729200e-01
-1.05470526e+00 1.32893014e+00 1.33621648e-01 -1.73903137e-01
5.83230317e-01 -2.45532915e-01 6.24854416e-02 -3.99161190e-01
-9.22184408e-01 4.50656980e-01 4.39249367e-01 8.57472867e-02
6.34762049e-01 2.29032829e-01 -5.32736182e-01 8.19997311e-01
-6.28441274e-01 6.24900199e-02 7.30619907e-01 6.31826043e-01
-6.50766790e-01 -3.61203849e-01 -1.82090297e-01 9.16879356e-01
-6.98618114e-01 6.40440211e-02 -5.46695590e-01 3.48391682e-01
-1.46461144e-01 1.11237478e+00 -1.38572138e-02 -6.71293259e-01
4.06453460e-01 2.55549580e-01 -6.88930005e-02 -5.90276062e-01
-1.16008854e+00 -1.78540498e-01 6.91562653e-01 -5.91058969e-01
2.35045284e-01 -7.14671373e-01 -9.24423158e-01 -5.76175511e-01
-2.82527745e-01 2.85858840e-01 9.25751150e-01 1.03379238e+00
4.70645390e-02 5.45989335e-01 8.30852449e-01 -7.50735044e-01
-1.04434466e+00 -1.30300331e+00 -6.71273232e-01 6.33922219e-01
4.77763891e-01 -3.16486448e-01 -6.62011147e-01 5.19524291e-02]
|
[8.744149208068848, 10.844122886657715]
|
65e2c2ae-678e-43da-bde4-b56d54da2f46
|
deep-operator-learning-based-surrogate-models
|
2306.00810
| null |
https://arxiv.org/abs/2306.00810v1
|
https://arxiv.org/pdf/2306.00810v1.pdf
|
Deep Operator Learning-based Surrogate Models with Uncertainty Quantification for Optimizing Internal Cooling Channel Rib Profiles
|
This paper designs surrogate models with uncertainty quantification capabilities to improve the thermal performance of rib-turbulated internal cooling channels effectively. To construct the surrogate, we use the deep operator network (DeepONet) framework, a novel class of neural networks designed to approximate mappings between infinite-dimensional spaces using relatively small datasets. The proposed DeepONet takes an arbitrary continuous rib geometry with control points as input and outputs continuous detailed information about the distribution of pressure and heat transfer around the profiled ribs. The datasets needed to train and test the proposed DeepONet framework were obtained by simulating a 2D rib-roughened internal cooling channel. To accomplish this, we continuously modified the input rib geometry by adjusting the control points according to a simple random distribution with constraints, rather than following a predefined path or sampling method. The studied channel has a hydraulic diameter, Dh, of 66.7 mm, and a length-to-hydraulic diameter ratio, L/Dh, of 10. The ratio of rib center height to hydraulic diameter (e/Dh), which was not changed during the rib profile update, was maintained at a constant value of 0.048. The ribs were placed in the channel with a pitch-to-height ratio (P/e) of 10. In addition, we provide the proposed surrogates with effective uncertainty quantification capabilities. This is achieved by converting the DeepONet framework into a Bayesian DeepONet (B-DeepONet). B-DeepONet samples from the posterior distribution of DeepONet parameters using the novel framework of stochastic gradient replica-exchange MCMC.
|
['Guillermo Paniagua', 'Guang Lina', 'Amirhossein Mollaali', 'Christian Moya', 'Izzet Sahin']
|
2023-06-01
| null | null | null | null |
['operator-learning']
|
['miscellaneous']
|
[-1.69503123e-01 1.69795260e-01 2.21912786e-01 -4.05124091e-02
-3.58751178e-01 -3.97755742e-01 6.12433851e-01 -6.41428307e-02
-5.49206376e-01 1.19710910e+00 -1.13696016e-01 -3.71548444e-01
-5.26463985e-01 -9.83591557e-01 -8.30968380e-01 -1.12090862e+00
-2.09925711e-01 9.57927763e-01 2.06278116e-02 8.14409107e-02
3.73460561e-01 7.80186415e-01 -1.43508327e+00 -3.75857741e-01
8.65414560e-01 9.59483862e-01 1.69903174e-01 5.41698813e-01
6.93545267e-02 1.31495863e-01 -4.07676518e-01 -9.92775112e-02
3.89912635e-01 -3.57747763e-01 -5.64631701e-01 -3.37149411e-01
-5.35930693e-01 -3.69970798e-01 -1.33128986e-01 7.95555949e-01
5.94350398e-01 4.35992062e-01 1.07778084e+00 -7.80543923e-01
-2.66746253e-01 6.07307732e-01 -3.97613436e-01 -2.48191416e-01
-3.67126793e-01 2.34667197e-01 5.53369462e-01 -6.12309694e-01
4.68875945e-01 1.18999398e+00 7.61616290e-01 4.05364990e-01
-1.27808237e+00 -4.38166112e-01 -3.21684718e-01 -3.62256944e-01
-1.53170609e+00 -6.02298379e-02 5.65291882e-01 -4.31544453e-01
4.10561502e-01 2.67690957e-01 8.40769291e-01 9.44735289e-01
7.10237086e-01 2.32477244e-02 1.12718308e+00 -3.07889730e-01
7.89196670e-01 1.01371119e-02 -2.34646216e-01 3.72853458e-01
4.14063454e-01 3.03244352e-01 -2.04955727e-01 -2.56430686e-01
1.08348703e+00 -2.88457066e-01 -8.36208686e-02 -3.44044715e-01
-8.48965883e-01 7.01856136e-01 3.20404142e-01 1.80334225e-01
-3.33930016e-01 3.84177208e-01 3.58053863e-01 -6.24784790e-02
1.58921137e-01 5.99394441e-01 -5.36811173e-01 -2.64225632e-01
-7.97298074e-01 3.21279794e-01 9.24991965e-01 8.22866082e-01
7.04399645e-01 8.43296424e-02 -3.03576261e-01 5.72689056e-01
4.81009036e-01 7.27371752e-01 3.24636430e-01 -1.16425073e+00
2.86864519e-01 1.65699244e-01 4.69168484e-01 -6.35336876e-01
-3.94044369e-01 -5.65902829e-01 -1.22817445e+00 7.67950267e-02
4.07509983e-01 -4.05547291e-01 -9.84922409e-01 1.61425591e+00
4.69548345e-01 -1.27724320e-01 6.77578598e-02 9.69840348e-01
9.40337330e-02 8.23099375e-01 -1.79420635e-01 -4.20538813e-01
9.24039125e-01 -4.86924678e-01 -5.33713758e-01 2.04543501e-01
3.40068161e-01 -3.63148242e-01 1.07739687e+00 3.39319050e-01
-9.40249503e-01 -3.89317393e-01 -1.03346229e+00 4.73392904e-01
-1.89043894e-01 1.65003344e-01 1.37079209e-01 6.12587810e-01
-6.74757659e-01 1.11664915e+00 -1.09325802e+00 7.05164596e-02
3.89719196e-02 8.72780755e-02 -4.10984922e-03 2.77854800e-01
-1.38565707e+00 8.14637899e-01 4.73005801e-01 9.49302375e-01
-1.21690619e+00 -6.85161471e-01 -5.88567376e-01 -3.51914950e-02
4.67680693e-02 -7.80986369e-01 1.10178673e+00 -3.55732709e-01
-2.28811717e+00 3.25837314e-01 2.61843920e-01 -2.58655697e-01
9.53841388e-01 -1.99263483e-01 9.03115422e-02 -5.59211075e-02
-2.42168345e-02 1.06211297e-01 7.86033869e-01 -1.44792378e+00
1.66373521e-01 -2.66681612e-01 -3.81162256e-01 2.25053549e-01
4.33896855e-02 -5.53117454e-01 -2.00662658e-01 -3.49315315e-01
4.97115850e-01 -1.12393522e+00 -3.86890769e-01 -1.49962649e-01
-6.14876151e-01 2.81402856e-01 6.81373954e-01 -5.18598795e-01
8.73991966e-01 -1.80747688e+00 3.93959910e-01 6.66269779e-01
-1.81876928e-01 6.44569024e-02 3.52052093e-01 6.96265161e-01
2.41014928e-01 1.09977089e-01 -9.76370752e-01 -3.62927973e-01
1.15144327e-01 4.59636360e-01 2.20890231e-02 6.34709179e-01
-7.91777000e-02 4.85797375e-01 -9.64133203e-01 -2.95119256e-01
4.11420137e-01 5.80988646e-01 -3.28929365e-01 2.98362613e-01
-1.45533755e-01 5.40190995e-01 -8.44326258e-01 4.23928589e-01
9.60856318e-01 2.63864189e-01 1.29828334e-01 -9.01705921e-02
-5.13194144e-01 -2.48508692e-01 -1.41072059e+00 1.46868050e+00
-8.28412116e-01 2.48713881e-01 3.52909237e-01 -7.27367997e-01
1.52239764e+00 9.40920487e-02 4.39203799e-01 -4.55619574e-01
3.46933156e-01 3.96169782e-01 -1.19014457e-01 -3.66614610e-01
5.52171469e-01 -6.15689874e-01 -4.82590385e-02 4.09731835e-01
-4.58037853e-01 -9.02104616e-01 -2.80279759e-02 -1.69174910e-01
8.02622497e-01 3.59561652e-01 -1.17984511e-01 -5.36078572e-01
6.84131384e-01 -2.16894239e-01 6.87174141e-01 5.82993507e-01
3.12592648e-02 7.39080906e-01 6.56714439e-01 -3.78962129e-01
-1.49898565e+00 -1.26162648e+00 -7.59194911e-01 2.04121396e-01
3.14432353e-01 -1.73640549e-02 -7.62395680e-01 1.55576795e-01
2.67031550e-01 7.62798429e-01 -7.67593503e-01 -1.90932110e-01
-5.71615696e-01 -9.77085233e-01 6.21776700e-01 3.95650089e-01
7.90417314e-01 -7.70046353e-01 -8.40845287e-01 2.91091591e-01
1.22924760e-01 -5.47391593e-01 1.29581004e-01 6.04941845e-01
-1.20647776e+00 -9.82643247e-01 -7.21500516e-01 -1.93885446e-01
6.11324191e-01 -8.93706501e-01 9.38013554e-01 -1.34667426e-01
-3.09162080e-01 -1.17046803e-01 -2.83064067e-01 4.78514694e-02
-6.47766948e-01 2.18104959e-01 3.64187211e-02 -2.13036001e-01
-4.46451455e-01 -5.03442109e-01 -6.70614541e-01 6.23453200e-01
-9.73196328e-01 -3.57512326e-04 4.23406243e-01 1.15752983e+00
6.97512209e-01 2.66122699e-01 2.86212981e-01 -5.82640350e-01
6.82628155e-01 -4.39695716e-01 -7.70561278e-01 8.39477107e-02
-6.74511731e-01 3.78616959e-01 8.45291913e-01 -1.70096308e-01
-1.22336340e+00 -3.67040634e-02 -1.29049554e-01 -3.99752855e-01
9.38378572e-02 5.80298364e-01 -1.49653032e-01 3.20258826e-01
4.51092660e-01 4.19524778e-03 6.39490187e-02 -4.40328985e-01
2.12061182e-01 6.65069044e-01 3.85581225e-01 -1.25513482e+00
6.75239444e-01 2.39941850e-01 4.65150535e-01 -7.61454046e-01
-1.54454410e-01 3.52943927e-01 -6.91923499e-01 -3.15456212e-01
6.78809762e-01 -4.33143705e-01 -9.88209367e-01 8.02681744e-01
-7.72933483e-01 -7.80110598e-01 -4.82584625e-01 5.72946489e-01
-6.49012983e-01 5.83868101e-02 -7.80069113e-01 -1.17141473e+00
-4.07456994e-01 -1.11387122e+00 8.42360616e-01 2.39321157e-01
1.15241088e-01 -8.37707341e-01 2.18408629e-01 -1.37062490e-01
7.06804216e-01 5.96450090e-01 9.32766199e-01 1.02427639e-01
-3.74195278e-01 -7.28769675e-02 1.39944315e-01 5.34181774e-01
1.11046135e-01 4.35684055e-01 -6.76843166e-01 -4.14061278e-01
1.32407650e-01 -2.44963184e-01 6.46747053e-01 6.34371877e-01
1.23614800e+00 3.52271982e-02 -3.60785335e-01 7.11404383e-01
1.44474947e+00 2.33684748e-01 6.08004153e-01 5.22523642e-01
2.10527107e-01 4.16011304e-01 5.66968322e-01 8.63439381e-01
-8.29797089e-02 3.79093021e-01 7.33647525e-01 1.96145922e-01
2.47526571e-01 -2.46950313e-01 1.02045268e-01 6.41069651e-01
-3.45479667e-01 -2.00871378e-01 -1.03961766e+00 4.24035728e-01
-1.49111104e+00 -5.73988914e-01 -1.67892009e-01 2.49991155e+00
7.77581930e-01 2.61455119e-01 -4.51938123e-01 5.04143573e-02
8.80455315e-01 -4.42155041e-02 -7.00489283e-01 -7.02508152e-01
8.30639452e-02 6.69414848e-02 8.41782570e-01 6.54070914e-01
-5.90495765e-01 3.02688897e-01 5.45073700e+00 6.12683594e-01
-1.13407183e+00 -3.85596067e-01 6.57205164e-01 3.71464007e-02
-5.07095575e-01 1.34183764e-01 -7.15106964e-01 7.03897238e-01
1.09872401e+00 -1.74200952e-01 6.07551455e-01 4.07635808e-01
5.49585879e-01 -3.77745926e-01 -7.38542438e-01 4.39262331e-01
-6.73010290e-01 -1.01951241e+00 -1.88716188e-01 -8.48710537e-02
8.25006366e-01 -1.82414144e-01 5.37550375e-02 1.87368989e-01
2.03224137e-01 -8.92370343e-01 7.37933397e-01 8.81249607e-01
9.77406979e-01 -8.25003564e-01 1.25925326e+00 3.70734543e-01
-1.18402147e+00 -1.85499221e-01 -4.30918932e-01 -3.10880318e-02
3.73708904e-01 9.14292514e-01 -8.84317458e-01 8.63275051e-01
8.29025090e-01 2.93514669e-01 3.90691422e-02 7.63238251e-01
1.12615870e-02 4.73112315e-01 -8.20988894e-01 -2.51354158e-01
2.03285664e-01 -9.02266920e-01 5.37447810e-01 6.45665526e-01
6.12565756e-01 -1.77674457e-01 -3.78355086e-01 1.33216441e+00
7.74959028e-02 -1.88394800e-01 -4.01071906e-01 2.08747059e-01
9.48822141e-01 9.74936604e-01 -7.48520732e-01 8.54237005e-02
5.40909469e-01 4.50677812e-01 -1.75413370e-01 3.92769158e-01
-1.06275034e+00 -5.58223486e-01 4.30026770e-01 2.51851350e-01
1.85175195e-01 -2.19819501e-01 -2.49829352e-01 -4.74748403e-01
-6.83383876e-03 -2.01370895e-01 -6.21743537e-02 -8.33271444e-01
-1.29606688e+00 5.50886631e-01 4.90977854e-01 -1.02858222e+00
-3.36147815e-01 -6.33298993e-01 -5.33800840e-01 1.23949766e+00
-1.03835297e+00 -7.80042648e-01 -3.09127986e-01 1.46564141e-01
2.91179940e-02 1.44838735e-01 7.74751127e-01 -5.99131696e-02
-7.81786025e-01 3.27538043e-01 9.19609845e-01 -1.79063976e-01
2.52790093e-01 -1.17932284e+00 2.45443717e-01 4.41248327e-01
-1.02971566e+00 6.79554045e-01 9.92524683e-01 -7.62058556e-01
-1.56678569e+00 -1.04445338e+00 6.08579740e-02 -8.94885510e-02
4.75017369e-01 -3.31970185e-01 -8.56585443e-01 3.39743048e-01
-1.29991934e-01 5.16291708e-02 2.51810928e-03 -1.92045838e-01
4.82488424e-01 -1.93084344e-01 -1.49297869e+00 3.87704909e-01
6.52477384e-01 -3.63851160e-01 -3.63695979e-01 -2.41670627e-02
4.52594608e-01 -6.78681254e-01 -1.16776979e+00 6.57068491e-01
7.84503520e-01 -9.29854572e-01 7.04508185e-01 -1.08075127e-01
2.72601008e-01 -4.61467266e-01 -2.51234710e-01 -1.45306921e+00
1.42008871e-01 -6.14762366e-01 1.37352273e-01 1.16151404e+00
2.84928709e-01 -8.08399975e-01 8.77569497e-01 6.70507193e-01
-4.92382795e-01 -1.08706379e+00 -1.43045092e+00 -7.12930858e-01
6.26927495e-01 -2.76988477e-01 7.42426455e-01 3.73204380e-01
-4.48860168e-01 -1.64586484e-01 -1.57366768e-01 2.51207322e-01
6.76247299e-01 -1.04648352e-01 5.32526910e-01 -1.28265131e+00
-2.55963326e-01 -6.65540248e-02 1.43687382e-01 -6.43516481e-01
7.37432912e-02 -4.83156085e-01 3.35482240e-01 -1.47498691e+00
-1.91333100e-01 -8.62259865e-01 1.98467784e-02 -1.94141101e-02
3.65962088e-01 -2.56327957e-01 -2.77766764e-01 1.60383210e-01
5.61710417e-01 1.20624256e+00 1.55780840e+00 6.27664477e-02
-4.92405266e-01 1.57683671e-01 1.82510242e-01 4.54744041e-01
8.43956351e-01 -3.51428360e-01 -4.40242052e-01 -2.06347570e-01
3.66927624e-01 5.95562994e-01 1.83339968e-01 -1.01073933e+00
-1.59890410e-02 -2.16879711e-01 3.09182793e-01 -6.15286946e-01
2.87311912e-01 -9.25099134e-01 7.57035315e-01 6.40464246e-01
-4.59438935e-02 8.19257572e-02 3.53614897e-01 6.55647635e-01
-9.49213430e-02 -2.57466793e-01 7.86937773e-01 -1.95052460e-01
-3.03585213e-02 -1.27633765e-01 -3.63833904e-01 -2.19388604e-01
9.11131859e-01 -2.55062610e-01 -9.81784612e-02 8.07902515e-02
-7.12709844e-01 1.37147054e-01 6.62198186e-01 -1.64506167e-01
4.15803522e-01 -1.16673970e+00 -4.38362628e-01 5.43216586e-01
-2.82181621e-01 5.64211965e-01 4.35842454e-01 7.24870622e-01
-1.03436434e+00 9.78469178e-02 -2.74641305e-01 -7.50497758e-01
-3.83552313e-01 2.05086395e-01 9.74352896e-01 -7.38732740e-02
-6.63588703e-01 6.25073850e-01 -4.09528702e-01 -9.18269992e-01
-7.85500258e-02 -5.33523262e-01 2.15643898e-01 -1.67787075e-01
-1.70899898e-01 6.85629189e-01 1.73500419e-01 -2.61969477e-01
-1.81243554e-01 6.62765682e-01 6.15123987e-01 -3.28492165e-01
1.47107327e+00 -7.46724010e-02 -2.29889676e-01 5.36924660e-01
8.91767979e-01 -1.21582240e-01 -1.48649800e+00 4.56732184e-01
-3.17701310e-01 -4.09350067e-01 1.51571319e-01 -7.58369148e-01
-9.19032812e-01 5.83237052e-01 5.82133532e-01 -2.07117628e-02
8.08072388e-01 -3.45937997e-01 4.27873820e-01 3.90941888e-01
4.29940790e-01 -1.17405546e+00 -3.15747112e-01 6.18618965e-01
8.09775651e-01 -5.14142454e-01 1.24516815e-01 7.29574114e-02
-2.81130940e-01 1.27727294e+00 5.92120886e-01 1.92969702e-02
6.98584020e-01 5.04918396e-01 -1.52670473e-01 2.84859519e-02
-4.91918296e-01 3.65870178e-01 -3.05155098e-01 1.59502625e-01
3.28984439e-01 1.79006949e-01 -3.65571588e-01 2.94806063e-01
-1.64606482e-01 1.74091578e-01 6.16569638e-01 7.71374047e-01
-2.20908180e-01 -9.69014466e-01 -7.03198493e-01 3.11292887e-01
1.28380388e-01 1.66143686e-01 2.80866325e-01 1.05042040e+00
1.23847209e-01 4.28856224e-01 3.78363371e-01 -2.61527747e-01
3.81888062e-01 5.41405454e-02 1.48026213e-01 -1.32147551e-01
-2.52286732e-01 -2.25507826e-01 -9.14774910e-02 -1.81852430e-01
2.58381832e-02 -6.23021185e-01 -1.54467261e+00 -5.31188130e-01
-2.72184879e-01 6.82086229e-01 8.20396066e-01 7.59968638e-01
-4.22463194e-02 5.20360172e-01 7.61161268e-01 -1.04437935e+00
-8.46907616e-01 -1.17978621e+00 -1.08997655e+00 -7.83503354e-02
2.17440173e-01 -1.04980052e+00 -8.04382145e-01 -5.58271825e-01]
|
[6.439406394958496, 3.3788089752197266]
|
3a418f73-03a3-44ae-9011-731e1bdf601d
|
deep-fusion-of-gray-level-co-occurrence
|
2205.05123
| null |
https://arxiv.org/abs/2205.05123v2
|
https://arxiv.org/pdf/2205.05123v2.pdf
|
Deep fusion of gray level co-occurrence matrices for lung nodule classification
|
Lung cancer is a severe menace to human health, due to which millions of people die because of late diagnoses of cancer; thus, it is vital to detect the disease as early as possible. The Computerized chest analysis Tomography of scan is assumed to be one of the efficient solutions for detecting and classifying lung nodules. The necessity of high accuracy of analyzing C.T. scan images of the lung is considered as one of the crucial challenges in detecting and classifying lung cancer. A new long-short-term-memory (LSTM) based deep fusion structure, is introduced, where, the texture features computed from lung nodules through new volumetric grey-level-co-occurrence-matrices (GLCM) computations are applied to classify the nodules into: benign, malignant and ambiguous. An improved Otsu segmentation method combined with the water strider optimization algorithm (WSA) is proposed to detect the lung nodules. Otsu-WSA thresholding can overcome the restrictions present in previous thresholding methods. Extended experiments are run to assess this fusion structure by considering 2D-GLCM computations based 2D-slices fusion, and an approximation of this 3D-GLCM with volumetric 2.5D-GLCM computations-based LSTM fusion structure. The proposed methods are trained and assessed through the LIDC-IDRI dataset, where 94.4%, 91.6%, and 95.8% Accuracy, sensitivity, and specificity are obtained, respectively for 2D-GLCM fusion and 97.33%, 96%, and 98%, accuracy, sensitivity, and specificity, respectively, for 2.5D-GLCM fusion. The yield of the same are 98.7%, 98%, and 99%, for the 3D-GLCM fusion. The obtained results and analysis indicate that the WSA-Otsu method requires less execution time and yields a more accurate thresholding process. It is found that 3D-GLCM based LSTM outperforms its counterparts.
|
['AhmadReza Naghsh Nilchi', 'Hossein Karshenas', 'Ahmed Saihood']
|
2022-05-10
| null | null | null | null |
['lung-nodule-classification']
|
['medical']
|
[ 2.40269557e-01 -1.59764051e-01 -7.06351548e-02 1.03991076e-01
-9.07225311e-01 9.40287486e-02 4.37538534e-01 1.31746367e-01
-6.03287458e-01 4.54988956e-01 -7.30924904e-02 -4.25975442e-01
-8.84665698e-02 -7.82068908e-01 -9.98382568e-02 -1.09342158e+00
-5.07901125e-02 5.75636327e-01 4.94848758e-01 3.83641183e-01
-1.03035927e-01 5.70381939e-01 -1.26271677e+00 6.26446784e-01
5.91443539e-01 1.48534417e+00 3.66426229e-01 9.62096989e-01
-3.13530326e-01 7.06484079e-01 -1.49961323e-01 4.56793159e-02
2.61531442e-01 -3.59164119e-01 -6.46566212e-01 3.60001951e-01
1.73523828e-01 -3.74334693e-01 1.48496125e-02 9.57200110e-01
2.43672743e-01 -1.89444255e-02 1.00078177e+00 -6.34063303e-01
5.06401509e-02 1.11319847e-01 -8.19605291e-01 3.45467597e-01
-1.79545715e-01 -1.87900774e-02 3.26987624e-01 -8.09954882e-01
1.36001274e-01 8.05564404e-01 7.30073452e-01 3.65214527e-01
-5.29077649e-01 -3.60087752e-01 -5.34167886e-01 8.48090798e-02
-1.18494022e+00 2.14575440e-01 8.93354118e-02 -5.22898018e-01
1.03011656e+00 6.88436627e-01 7.88472474e-01 6.26862824e-01
1.04805803e+00 3.24827909e-01 1.40162575e+00 -5.91379523e-01
-2.16924753e-02 2.39661247e-01 -5.79707511e-02 1.14915967e+00
5.79843402e-01 1.94498658e-01 6.07994944e-02 -1.94278732e-01
9.21539068e-01 3.82177740e-01 8.25888738e-02 -1.34036085e-02
-1.33427477e+00 5.67890584e-01 5.85942030e-01 1.09855497e+00
-6.08193457e-01 3.12837332e-01 4.31587428e-01 -5.27120903e-02
4.34430510e-01 -6.24462776e-02 -5.99435996e-03 2.44629592e-01
-1.40965331e+00 -1.83565751e-01 5.62218070e-01 5.13764322e-01
2.53405303e-01 1.47462904e-01 -6.29522443e-01 5.24971724e-01
4.88260537e-01 8.32614422e-01 1.08902383e+00 -7.90333211e-01
4.52917479e-02 5.01091421e-01 -3.11662227e-01 -8.45091879e-01
-3.92490476e-01 -5.72775424e-01 -1.42588532e+00 6.65224195e-02
1.28527388e-01 3.46684635e-01 -1.48527443e+00 1.03594041e+00
2.22576514e-01 8.35844055e-02 -2.82766912e-02 6.46002531e-01
8.49641562e-01 5.89072943e-01 1.96509629e-01 -4.37411785e-01
1.58004498e+00 -7.70444095e-01 -6.81157708e-01 1.33826822e-01
7.02139974e-01 -8.42298687e-01 5.59279740e-01 1.53298229e-01
-1.11682773e+00 -7.01905727e-01 -7.29264081e-01 3.27363372e-01
-2.09919021e-01 4.80771989e-01 2.60496825e-01 7.25126565e-01
-1.23457086e+00 4.91316795e-01 -1.19722772e+00 -5.65358639e-01
1.60435349e-01 5.15466571e-01 -1.16965517e-01 2.39956621e-02
-1.06755447e+00 9.79845464e-01 4.68508154e-01 2.56948322e-01
-6.34360611e-01 -1.65344462e-01 -3.45262885e-01 7.40708336e-02
2.47652218e-01 -8.90240669e-01 1.36959493e+00 -5.97840130e-01
-1.04944968e+00 9.81970131e-01 -2.64090985e-01 -5.46930075e-01
5.99064410e-01 4.37072605e-01 -1.02457628e-01 3.81385505e-01
6.99168965e-02 6.06170297e-01 7.59884953e-01 -1.04497576e+00
-1.00399256e+00 -4.13539380e-01 -8.15433800e-01 2.36127481e-01
-3.12202066e-01 1.18038058e-02 -3.37158352e-01 -5.18837631e-01
5.44197917e-01 -8.16885948e-01 -4.83400643e-01 -1.40249684e-01
-4.33477610e-01 -1.87893435e-02 1.04294705e+00 -1.09771979e+00
1.18223870e+00 -1.73916876e+00 -2.92893052e-01 5.83313763e-01
5.21878481e-01 3.11683178e-01 6.83449209e-01 -3.55561256e-01
8.24323222e-02 3.51990789e-01 -5.69833100e-01 -1.93999827e-01
-2.78115749e-01 2.55289406e-01 3.06642473e-01 3.36439043e-01
-1.21590510e-01 8.21759820e-01 -3.61596137e-01 -1.13893914e+00
5.37552357e-01 5.23271739e-01 1.43330887e-01 -2.23292299e-02
1.83773950e-01 2.08788261e-01 -6.91819251e-01 8.69613886e-01
5.14963806e-01 -1.64635450e-01 -3.40090171e-02 -2.97326386e-01
-3.83359820e-01 -3.95392478e-01 -7.93192565e-01 9.21251357e-01
-3.94585520e-01 3.87472510e-01 7.55747259e-02 -8.02913904e-01
7.58925796e-01 1.00484395e+00 8.35302234e-01 -7.72645891e-01
5.57075560e-01 7.71947742e-01 2.88737059e-01 -6.96870029e-01
3.77686694e-02 -4.43729192e-01 1.39423743e-01 1.43546671e-01
-3.36248040e-01 -2.49788344e-01 1.69518217e-01 -1.65572446e-02
9.84548450e-01 -2.72656411e-01 6.89083576e-01 -2.73479342e-01
8.20629478e-01 7.40465224e-02 2.06031755e-01 5.79571426e-01
-4.01936471e-01 5.13328254e-01 1.35171279e-01 -2.55817682e-01
-9.94265079e-01 -8.89242232e-01 -2.66597480e-01 3.41895282e-01
-4.47861224e-01 4.85670745e-01 -8.97930026e-01 -4.09720808e-01
-8.96275938e-02 4.01618183e-01 -6.67039752e-01 7.75919482e-02
-7.51135647e-01 -9.93050277e-01 4.40142572e-01 5.16228259e-01
1.00536847e+00 -9.07201469e-01 -1.22240257e+00 1.99942112e-01
-2.34647214e-01 -1.01820803e+00 -1.26764104e-01 3.19239706e-01
-1.44093597e+00 -8.68977547e-01 -1.14677155e+00 -8.57671857e-01
5.76852381e-01 3.22871387e-01 9.01425421e-01 4.71903473e-01
-5.92783511e-01 3.21353704e-01 -4.83803563e-02 -8.30376819e-02
-6.22681260e-01 -1.69827342e-01 -2.21232921e-01 -2.09646240e-01
4.61956449e-02 -1.96756303e-01 -3.51009667e-01 1.61071926e-01
-7.99351573e-01 1.83385581e-01 1.29594529e+00 7.15575397e-01
1.06370378e+00 2.69206107e-01 1.39506087e-01 -7.94824123e-01
1.91649824e-01 -1.95643604e-01 -2.75060326e-01 1.78972289e-01
-6.44307315e-01 -1.94515735e-01 5.35705626e-01 -7.97088072e-02
-1.03561115e+00 9.89500135e-02 -9.49780867e-02 -5.92611074e-01
-2.86234796e-01 4.23606783e-01 3.49403024e-01 -3.53127301e-01
3.33402693e-01 2.62106001e-01 -1.12563036e-01 -8.96899924e-02
-2.74758577e-01 6.53444231e-01 5.04116058e-01 -8.99227783e-02
7.14919448e-01 5.56741416e-01 3.46589804e-01 -9.23340380e-01
-5.67053735e-01 -8.33428681e-01 -7.79037297e-01 -5.34673572e-01
1.43193293e+00 -5.03655076e-01 -3.38579744e-01 5.54867625e-01
-8.36968958e-01 -2.50702612e-02 -2.11596996e-01 8.02697837e-01
-3.39982271e-01 4.91016477e-01 -7.30354428e-01 -8.73274326e-01
-8.72272253e-01 -1.32753313e+00 8.56295168e-01 1.90634802e-01
-5.16033322e-02 -1.07836461e+00 -2.98361570e-01 4.43755299e-01
8.39382946e-01 5.30678272e-01 1.09878993e+00 -6.60849690e-01
-4.01931673e-01 -4.42280740e-01 -4.12389547e-01 4.49457377e-01
2.53089994e-01 1.87106863e-01 -7.72427917e-01 -1.55305400e-01
7.32908189e-01 4.29689847e-02 1.08668709e+00 9.08953547e-01
1.13832307e+00 -4.80036763e-03 -6.16281331e-01 3.53248388e-01
1.46728969e+00 6.53370619e-01 3.93649906e-01 2.20188931e-01
6.09557927e-01 1.34916663e-01 5.11716485e-01 1.79882020e-01
-1.27683967e-01 1.45252645e-01 5.40982723e-01 -4.60469872e-01
-3.90329093e-01 4.34336185e-01 6.48381338e-02 1.19946241e+00
-2.82443792e-01 -2.70885170e-01 -1.18575060e+00 3.53007615e-01
-1.27239394e+00 -8.04618895e-01 -6.53150201e-01 2.19391346e+00
4.18022126e-01 3.99451554e-01 -1.63352966e-01 5.22489548e-01
8.39249551e-01 -4.58021797e-02 -1.94413692e-01 -3.71129364e-01
3.04205060e-01 4.07191902e-01 8.98828924e-01 3.19794357e-01
-1.39317453e+00 2.73538262e-01 5.21850204e+00 1.03957772e+00
-1.40116572e+00 3.75544637e-01 1.07414329e+00 2.30226859e-01
1.05828412e-01 -3.68760914e-01 -4.87628818e-01 3.55999887e-01
8.46858978e-01 1.72512546e-01 -2.39299893e-01 5.21677554e-01
3.82093042e-01 -7.26289928e-01 -6.31704628e-01 8.36608410e-01
4.79142629e-02 -8.84530962e-01 -6.18268596e-03 1.79715037e-01
6.52252138e-01 -1.32980924e-02 2.55980194e-01 1.98782042e-01
-2.73005098e-01 -1.03514564e+00 3.38392228e-01 5.94364047e-01
9.50343549e-01 -4.49817479e-01 1.31851876e+00 4.88584965e-01
-1.54911542e+00 1.12955786e-01 -2.52395689e-01 4.76504266e-01
-1.39335230e-01 8.23191345e-01 -1.13578308e+00 7.60527432e-01
9.30775881e-01 2.19487205e-01 -7.02201664e-01 1.03449178e+00
3.02271932e-01 6.68742776e-01 -4.99045849e-01 -2.29343802e-01
5.62952220e-01 -9.99416485e-02 3.29644799e-01 1.27846789e+00
8.53150308e-01 5.13952635e-02 2.67474055e-01 7.29874432e-01
1.70352384e-01 2.29070634e-01 -3.64293844e-01 1.15414441e-01
-1.90559775e-02 1.46670842e+00 -1.25695944e+00 -7.73627162e-01
-2.22870827e-01 7.19896674e-01 -4.62667793e-01 -4.52380255e-02
-1.07657218e+00 5.28345183e-02 -5.13438165e-01 2.23645598e-01
4.91197817e-02 1.02400221e-01 -3.89241666e-01 -6.16765618e-01
-1.58470303e-01 -5.89589119e-01 5.35748005e-01 -6.43585503e-01
-8.69870067e-01 7.74461508e-01 7.92494491e-02 -1.22098279e+00
-2.71575183e-01 -7.01092303e-01 -7.91095912e-01 8.57936859e-01
-1.21629298e+00 -1.04909241e+00 -6.30352199e-01 3.29147875e-01
5.20743072e-01 -4.64583077e-02 5.70481658e-01 6.48266673e-02
-4.57184374e-01 1.41098037e-01 2.45362341e-01 -5.10631725e-02
3.74614298e-01 -1.29527831e+00 -1.17474034e-01 6.48219526e-01
-3.54781419e-01 -1.05784424e-01 3.78146917e-01 -7.25378036e-01
-9.38258588e-01 -1.37389338e+00 8.48756135e-01 -1.90871637e-02
2.60396898e-01 3.49163085e-01 -9.98964846e-01 3.47473681e-01
1.16357282e-01 3.04546859e-03 3.44707221e-01 -9.56318080e-01
4.36898738e-01 9.84957144e-02 -1.45115304e+00 2.59829849e-01
2.32147232e-01 -1.67179689e-01 -3.33094776e-01 4.55701292e-01
2.32320994e-01 -4.36486453e-01 -1.09165156e+00 7.72404850e-01
5.04158020e-01 -1.27375877e+00 9.22983170e-01 3.67359966e-01
4.89781238e-02 -1.13608442e-01 -2.74126470e-01 -6.59245610e-01
-4.21863556e-01 4.47364032e-01 1.80316985e-01 5.28967202e-01
5.66655278e-01 -5.76724529e-01 9.50413585e-01 2.57396698e-01
-2.72586763e-01 -1.18598366e+00 -1.27165878e+00 -5.99457562e-01
3.19071561e-02 -2.67008305e-01 -2.20872030e-01 6.64808929e-01
-7.45531917e-01 -2.49580458e-01 8.47088248e-02 -1.21034719e-01
7.41671979e-01 -1.44302636e-01 9.44039319e-03 -1.22072208e+00
-1.36221454e-01 -6.10194862e-01 -2.89761990e-01 -2.54595578e-01
-4.78491545e-01 -9.31988418e-01 -8.80238637e-02 -2.03180957e+00
5.50356627e-01 -1.42691866e-01 -5.23912072e-01 4.28813815e-01
7.27716759e-02 2.71328568e-01 8.79478529e-02 5.31971931e-01
-1.33289576e-01 -1.64478109e-03 1.46000695e+00 -1.57405287e-01
3.67991924e-02 3.24938864e-01 1.08431354e-01 8.73370051e-01
9.21385169e-01 -5.45591950e-01 -3.01534757e-02 -5.63046969e-02
-3.54614913e-01 5.21611631e-01 4.58958864e-01 -1.56330967e+00
3.02413851e-01 -8.72274414e-02 8.46819043e-01 -1.12386644e+00
4.18173999e-01 -9.30019140e-01 2.45834798e-01 1.48733497e+00
-4.90492545e-02 9.33266357e-02 4.80302833e-02 2.76422620e-01
-3.83174300e-01 -4.77853507e-01 1.12965953e+00 -7.19388247e-01
-4.62851048e-01 2.26965576e-01 -9.33997571e-01 -4.78435248e-01
1.14740539e+00 -6.67615712e-01 1.99986219e-01 -1.86801419e-01
-8.21908414e-01 -1.02308653e-01 4.11947928e-02 -3.97219986e-01
7.80678988e-01 -1.27726996e+00 -6.07097507e-01 1.19229540e-01
-5.09224474e-01 5.78867607e-02 3.48245263e-01 1.52931523e+00
-8.64466667e-01 7.86712587e-01 -9.26612914e-02 -9.00733590e-01
-1.68699884e+00 3.25679958e-01 8.22252929e-01 -9.08737659e-01
-4.14786309e-01 9.15432036e-01 9.38792378e-02 -5.65950722e-02
8.02168325e-02 -1.10934603e+00 -1.81218982e-01 -2.11278543e-01
-8.63927975e-02 5.95429420e-01 5.45076013e-01 -6.18491292e-01
-2.05096215e-01 8.13262999e-01 8.71457160e-02 -1.38040036e-01
5.08562446e-01 8.26954469e-02 -4.35111403e-01 4.95825052e-01
9.71700311e-01 -2.91905433e-01 -3.20765585e-01 -2.58058533e-02
1.55058622e-01 2.02158969e-02 4.35054392e-01 -8.38116407e-01
-1.10633051e+00 1.15521181e+00 1.17541301e+00 3.46724957e-01
1.34115231e+00 -2.50012666e-01 1.03417492e+00 3.55615973e-01
3.07122488e-02 -8.18157554e-01 4.30824831e-02 3.29674155e-01
5.04003167e-01 -1.12002611e+00 1.01325437e-01 -3.70829105e-01
-3.66046727e-01 1.39490879e+00 5.76926470e-01 4.47235890e-02
6.38052404e-01 3.30076754e-01 -9.71633568e-03 -8.57005566e-02
-7.66014099e-01 -2.09753662e-01 4.35254425e-01 3.60906012e-02
6.57323420e-01 2.49623388e-01 -3.37684870e-01 2.36888990e-01
1.80394128e-01 1.39671937e-03 2.92053133e-01 1.03739667e+00
-1.03117347e+00 -5.74794352e-01 -1.01235628e+00 1.07994854e+00
-7.19337225e-01 1.10385016e-01 -2.52754122e-01 1.18774295e+00
3.66902560e-01 6.05503321e-01 7.86597356e-02 -3.20327580e-01
-1.04589174e-02 8.78215805e-02 4.20177668e-01 -4.32140142e-01
-7.50341892e-01 5.99092722e-01 -2.94062018e-01 -1.44059300e-01
-5.17487347e-01 -5.35733819e-01 -1.58040178e+00 -1.79052576e-01
-4.65922564e-01 2.34372169e-01 6.92489743e-01 9.29748535e-01
-2.59258538e-01 8.35862875e-01 4.42021161e-01 -7.34061897e-01
-6.31772995e-01 -1.04639578e+00 -5.18967867e-01 2.58179964e-03
1.59388795e-01 -5.29989541e-01 -4.89298284e-01 -6.61964640e-02]
|
[15.255053520202637, -2.217008352279663]
|
4d398456-9c38-4714-94e6-2b918775130a
|
pix2nerf-unsupervised-conditional-p-gan-for
|
2202.13162
| null |
https://arxiv.org/abs/2202.13162v1
|
https://arxiv.org/pdf/2202.13162v1.pdf
|
Pix2NeRF: Unsupervised Conditional $π$-GAN for Single Image to Neural Radiance Fields Translation
|
We propose a pipeline to generate Neural Radiance Fields~(NeRF) of an object or a scene of a specific class, conditioned on a single input image. This is a challenging task, as training NeRF requires multiple views of the same scene, coupled with corresponding poses, which are hard to obtain. Our method is based on $\pi$-GAN, a generative model for unconditional 3D-aware image synthesis, which maps random latent codes to radiance fields of a class of objects. We jointly optimize (1) the $\pi$-GAN objective to utilize its high-fidelity 3D-aware generation and (2) a carefully designed reconstruction objective. The latter includes an encoder coupled with $\pi$-GAN generator to form an auto-encoder. Unlike previous few-shot NeRF approaches, our pipeline is unsupervised, capable of being trained with independent images without 3D, multi-view, or pose supervision. Applications of our pipeline include 3d avatar generation, object-centric novel view synthesis with a single input image, and 3d-aware super-resolution, to name a few.
|
['Luc van Gool', 'Dengxin Dai', 'Anton Obukhov', 'Shengqu Cai']
|
2022-02-26
| null | null | null | null |
['3d-aware-image-synthesis']
|
['computer-vision']
|
[ 6.21274769e-01 2.30460003e-01 2.72910446e-01 -4.24880445e-01
-1.08705425e+00 -5.72172463e-01 7.78308213e-01 -7.90684819e-01
1.73926115e-01 5.83922923e-01 3.48724693e-01 1.69439182e-01
3.92859131e-01 -1.15324938e+00 -1.22600937e+00 -7.33293295e-01
6.03556871e-01 5.60982168e-01 -8.39140862e-02 -1.09013349e-01
3.77551140e-03 6.74457371e-01 -1.67947543e+00 1.57388881e-01
5.57802856e-01 8.99070323e-01 7.06705451e-01 9.91429985e-01
8.65251347e-02 8.64766002e-01 -5.17620742e-01 -3.54477406e-01
3.96060437e-01 -8.29133928e-01 -5.58732808e-01 3.32232744e-01
7.30030477e-01 -6.18052721e-01 -3.32552969e-01 8.18291843e-01
4.75828797e-01 2.17380181e-01 8.75828147e-01 -9.11708415e-01
-9.11113918e-01 2.13816166e-01 -6.32372260e-01 -2.45620295e-01
4.40835476e-01 4.08991337e-01 9.12753820e-01 -8.25924873e-01
9.36534464e-01 1.19216216e+00 2.67471433e-01 8.12228382e-01
-1.59545898e+00 -3.98181021e-01 -7.21473247e-02 -3.86554509e-01
-1.23102522e+00 -3.86238366e-01 1.07201314e+00 -6.01988018e-01
9.42809880e-01 -1.08177187e-02 7.22052932e-01 1.32867062e+00
1.02661207e-01 4.41440046e-01 1.07836831e+00 -2.18797594e-01
3.68524939e-01 7.57205412e-02 -7.54632294e-01 7.50094235e-01
-3.01804155e-01 8.53219703e-02 -6.69791043e-01 1.98208570e-01
1.29689133e+00 3.14254239e-02 -2.82514632e-01 -4.27500039e-01
-1.25952339e+00 7.74572730e-01 6.37489498e-01 -2.71136880e-01
-4.20146465e-01 5.88303685e-01 -2.87131459e-01 -1.00994393e-01
5.59964001e-01 4.94890451e-01 -3.55699033e-01 1.19054660e-01
-6.62840188e-01 4.08616155e-01 3.33776265e-01 1.18213582e+00
1.09938323e+00 4.90196824e-01 -3.31292413e-02 6.53198123e-01
1.56893075e-01 1.01301813e+00 2.11808495e-02 -1.19908929e+00
3.47391069e-01 3.06076169e-01 7.84806311e-02 -5.18060327e-01
1.15342595e-01 -3.63627106e-01 -8.09506118e-01 6.20791435e-01
-1.24490611e-01 -1.03075109e-01 -1.20302522e+00 2.07926631e+00
4.74607944e-01 1.11235842e-01 1.88736409e-01 9.81749177e-01
9.00249898e-01 1.06004322e+00 -4.39643055e-01 -5.62941022e-02
9.65983808e-01 -1.08406675e+00 -3.27573895e-01 -4.25877124e-01
7.95468464e-02 -7.14635372e-01 1.17287898e+00 2.27200851e-01
-1.41433835e+00 -6.67201817e-01 -8.01253796e-01 -4.78602320e-01
-5.82384169e-02 -8.39006230e-02 6.13843501e-01 3.28917980e-01
-9.86703396e-01 2.52939165e-01 -5.68860054e-01 -4.74414155e-02
4.07334030e-01 -9.29599851e-02 -4.75234956e-01 -4.21976894e-01
-7.19719529e-01 7.87609637e-01 2.26837234e-03 -1.70047462e-01
-1.56888914e+00 -8.92467380e-01 -1.02678418e+00 -1.33420944e-01
2.20252961e-01 -1.30449295e+00 1.18231702e+00 -8.42254996e-01
-1.74710691e+00 9.85335827e-01 -1.62638992e-01 -3.92809808e-02
2.56497294e-01 -1.27093330e-01 -4.11416180e-02 2.76508868e-01
3.66109908e-01 9.82378602e-01 1.03540540e+00 -1.63595223e+00
-2.42484510e-01 -1.95675924e-01 2.40831718e-01 5.54303765e-01
3.90777975e-01 -3.44144881e-01 -3.41803521e-01 -5.89896142e-01
1.40251163e-02 -6.89226627e-01 -3.66858900e-01 6.28306493e-02
-5.18133640e-01 3.67010862e-01 7.40162551e-01 -5.94725907e-01
4.46145236e-01 -2.06912136e+00 6.51657104e-01 -1.25875607e-01
-1.66406836e-02 -3.45413595e-01 -1.96979225e-01 3.74859035e-01
-5.61068580e-02 -4.47999127e-02 -4.20654893e-01 -5.29652536e-01
-1.20247833e-01 6.59114346e-02 -3.43408853e-01 2.22339824e-01
4.26750660e-01 9.34295774e-01 -9.31260169e-01 -1.24629691e-01
4.31065977e-01 8.07279527e-01 -8.11143160e-01 6.78883553e-01
-7.96787918e-01 9.27849293e-01 -3.26440752e-01 4.83968228e-01
6.52919054e-01 -4.45266485e-01 -1.51096657e-01 -5.02731204e-01
-3.28266956e-02 2.10165024e-01 -9.57947850e-01 2.23413277e+00
-7.89611638e-01 3.73370826e-01 -9.11393240e-02 -6.86230600e-01
9.63215709e-01 2.03894928e-01 5.15150666e-01 -4.65275913e-01
5.47625832e-02 4.03900519e-02 -6.14719450e-01 -3.50333035e-01
4.57173467e-01 -4.21168238e-01 -3.23237151e-01 5.92319787e-01
6.02098525e-01 -9.72219169e-01 -1.47972882e-01 3.22612524e-01
1.06738865e+00 6.87546134e-01 1.23294480e-01 1.77676365e-01
-2.41386350e-02 -2.07158178e-01 4.22096312e-01 4.85497236e-01
7.07140326e-01 1.37006760e+00 2.58618444e-01 -1.84017509e-01
-1.53701246e+00 -1.49571800e+00 1.68550104e-01 5.63429058e-01
1.50049299e-01 -1.38051003e-01 -6.93200409e-01 -5.86542308e-01
-3.48092347e-01 9.55953956e-01 -6.38450563e-01 -5.30841760e-02
-4.99077410e-01 -3.31912160e-01 8.94036964e-02 4.09024239e-01
4.57161158e-01 -1.02984715e+00 -7.97191560e-01 1.74152963e-02
-3.24794233e-01 -1.21118045e+00 -6.18800104e-01 2.20958143e-01
-6.52535319e-01 -7.91661382e-01 -8.57712150e-01 -6.51065767e-01
8.90772998e-01 4.16955084e-01 1.48615587e+00 -3.90325457e-01
-3.94921184e-01 5.66314936e-01 -2.75303066e-01 -2.04997092e-01
-4.83254462e-01 -2.26238042e-01 -2.44024232e-01 5.61803356e-02
-4.02431309e-01 -1.04094291e+00 -8.16970170e-01 1.84682637e-01
-9.75115836e-01 6.59832895e-01 5.67914724e-01 8.05393159e-01
1.03621626e+00 -1.43446445e-01 2.45475858e-01 -8.92037213e-01
-1.62026614e-01 -4.14460003e-01 -6.19104922e-01 6.86998069e-02
-9.98091474e-02 1.48926945e-02 7.51476824e-01 -2.79883146e-01
-1.40175998e+00 3.97358268e-01 -2.12800920e-01 -8.88762236e-01
-3.40366572e-01 8.51572771e-03 -6.08736217e-01 1.90597340e-01
9.49961841e-01 4.14865971e-01 -2.93689132e-01 -4.55760926e-01
8.85942817e-01 2.07473576e-01 8.23447466e-01 -4.71918851e-01
1.07201612e+00 6.19243443e-01 1.16418049e-01 -6.54876471e-01
-1.16695702e+00 3.00040864e-03 -6.53423369e-01 -2.98246622e-01
1.36150932e+00 -1.24401617e+00 -2.82804966e-01 3.98073941e-01
-1.18285811e+00 -7.49164045e-01 -7.47532845e-01 4.32634801e-01
-1.05873561e+00 -3.38838935e-01 -3.24331850e-01 -5.92028558e-01
-2.03157708e-01 -1.24452484e+00 1.61590970e+00 4.16266471e-01
1.54585585e-01 -6.83667839e-01 1.32001176e-01 6.76675200e-01
2.35300511e-01 6.18234515e-01 7.45137751e-01 4.34715599e-01
-1.11359227e+00 1.02668792e-01 -1.03672363e-01 4.82715279e-01
1.38311461e-01 -1.06057145e-01 -1.21736896e+00 -1.60426915e-01
-2.71394067e-02 -7.09317565e-01 5.98997712e-01 4.78094906e-01
1.31425679e+00 -3.25019240e-01 1.96298826e-02 1.04563475e+00
1.63820004e+00 5.69421686e-02 7.16876566e-01 -2.75181741e-01
1.01707971e+00 3.17111701e-01 2.29685143e-01 4.53284889e-01
4.01143938e-01 7.64987826e-01 8.32160592e-01 -2.05359101e-01
-5.56690156e-01 -6.66068852e-01 4.64388907e-01 5.67318738e-01
-2.45960668e-01 -4.10777837e-01 -4.06602144e-01 4.40384179e-01
-1.27754724e+00 -1.08052218e+00 1.62360877e-01 2.03661799e+00
6.41136050e-01 -2.11550817e-01 -1.41812786e-01 -3.35741371e-01
3.59223545e-01 4.53833818e-01 -8.50120366e-01 -1.79216400e-01
-1.92449823e-01 4.66197938e-01 1.80550441e-01 5.54404020e-01
-6.60331070e-01 9.23814595e-01 5.85493803e+00 4.16398257e-01
-1.19486189e+00 2.23966599e-01 6.82763457e-01 -3.59164417e-01
-1.09330893e+00 8.47451091e-02 -7.15354204e-01 2.83114016e-01
5.93794703e-01 1.74456388e-01 6.53994679e-01 9.08319712e-01
-1.07155673e-01 -1.44964069e-01 -1.08260334e+00 1.08604038e+00
4.55805779e-01 -1.61203587e+00 1.35873288e-01 2.08630309e-01
1.22511661e+00 3.60293277e-02 1.52707890e-01 -3.60551756e-03
6.00477993e-01 -1.07119441e+00 9.44093108e-01 7.87066340e-01
1.39105010e+00 -7.98588574e-01 -2.91603170e-02 4.21462387e-01
-1.02811098e+00 2.87440211e-01 -2.44577616e-01 5.36272585e-01
6.71210885e-01 6.03691638e-01 -5.76100290e-01 6.84198976e-01
8.63380611e-01 6.63978338e-01 -1.05534501e-01 3.10118496e-01
-5.65976501e-01 1.62856296e-01 -2.33314827e-01 4.34587896e-01
-5.77704310e-02 -2.36801162e-01 5.38376808e-01 7.46758938e-01
6.61856532e-01 3.50707799e-01 -4.85418625e-02 1.38326752e+00
-2.89466143e-01 -3.79241824e-01 -8.79830003e-01 1.40496165e-01
1.86319500e-01 1.34266388e+00 -4.29055601e-01 -1.52658567e-01
-3.31434965e-01 1.45967770e+00 3.36788088e-01 4.21298414e-01
-9.22683716e-01 -1.98262855e-01 6.04387224e-01 3.13767612e-01
5.84692657e-01 -2.24174887e-01 -1.40015721e-01 -1.48233271e+00
-4.07523029e-02 -6.70302391e-01 -7.08186105e-02 -1.66220844e+00
-1.22801113e+00 6.92175984e-01 2.03808155e-02 -1.35601187e+00
-5.24120748e-01 -3.35283905e-01 -5.86237133e-01 1.17610395e+00
-1.37093687e+00 -1.47850990e+00 -5.72815597e-01 7.00358927e-01
6.79835141e-01 -1.74645096e-01 9.81176436e-01 2.54378561e-02
-1.42045140e-01 2.23042399e-01 -2.54567027e-01 -1.96239457e-01
5.24082005e-01 -1.20962238e+00 5.73351800e-01 9.18882012e-01
2.49025434e-01 9.72351730e-02 5.51687896e-01 -5.96060455e-01
-1.52883089e+00 -1.30810869e+00 4.20671135e-01 -6.16619051e-01
-1.24025960e-02 -6.58919513e-01 -5.20272732e-01 7.93719590e-01
2.15298191e-01 2.63616085e-01 4.77235347e-01 -2.64722377e-01
-4.16972637e-01 -1.35179862e-01 -1.26430774e+00 5.58091402e-01
1.32781696e+00 -6.56082034e-01 -1.11184396e-01 3.17529619e-01
9.96256411e-01 -7.45339632e-01 -1.05808151e+00 2.70719796e-01
2.96999842e-01 -1.15522754e+00 1.20254302e+00 -1.94290593e-01
1.12692261e+00 -5.27773321e-01 -5.12309015e-01 -1.49061084e+00
-2.84474790e-01 -4.26881731e-01 -1.40220985e-01 9.95409906e-01
4.03832465e-01 -1.13822348e-01 7.24966705e-01 2.63671398e-01
-4.43562716e-01 -4.49967772e-01 -5.43300271e-01 -3.69445473e-01
-1.72911108e-01 -3.38926554e-01 6.82382584e-01 7.82910705e-01
-7.52466857e-01 7.86282957e-01 -7.49466360e-01 3.29638660e-01
9.24547195e-01 4.41657335e-01 1.23065031e+00 -7.29069829e-01
-8.12089086e-01 -1.28171533e-01 -1.24055699e-01 -1.34417522e+00
-5.19304425e-02 -8.42722833e-01 3.21532011e-01 -1.78391874e+00
2.03612983e-01 -3.77165616e-01 4.59881246e-01 1.96288526e-01
1.26498997e-01 4.84294176e-01 6.03169762e-02 4.40278798e-02
-2.49862716e-01 9.70034659e-01 1.60114193e+00 4.87101562e-02
4.18562489e-03 -1.19044445e-01 -7.62940884e-01 5.85465193e-01
4.37717408e-01 -2.85074055e-01 -7.71121740e-01 -7.83859432e-01
5.11889458e-01 5.14309704e-01 6.30916357e-01 -8.97886217e-01
-2.06195056e-01 -3.62544030e-01 7.04414487e-01 -7.37597764e-01
8.91330779e-01 -6.37831569e-01 7.42852092e-01 -1.79834679e-01
-2.54320055e-01 -2.03079864e-01 -2.43881628e-01 5.13211370e-01
-5.94827533e-02 -3.37767191e-02 9.25308168e-01 -6.45493627e-01
-5.43999255e-01 6.42969131e-01 1.37487249e-02 9.06872228e-02
8.54761004e-01 -1.38158783e-01 -1.24436907e-01 -5.98307192e-01
-5.97931921e-01 -1.81830928e-01 8.33401442e-01 3.23852599e-01
8.51987422e-01 -1.59451187e+00 -7.85210013e-01 3.86160910e-01
1.99445739e-01 9.41120982e-01 6.43818915e-01 6.90888837e-02
-5.22865951e-01 -1.46984965e-01 -1.99970976e-01 -8.07802737e-01
-7.87485659e-01 4.70197529e-01 4.29344445e-01 -3.41683142e-02
-8.35815966e-01 1.03033924e+00 7.88747013e-01 -7.50399888e-01
-2.49929219e-01 8.57979059e-02 2.20651910e-01 -4.91091132e-01
4.28120762e-01 -8.07324424e-02 -1.96880177e-01 -6.12142205e-01
7.18633235e-02 7.96680212e-01 2.80483544e-01 -6.40277803e-01
1.55554211e+00 -7.67264515e-02 1.55823380e-01 5.55756986e-01
1.36614501e+00 -4.53090779e-02 -1.90299797e+00 -2.98018754e-01
-1.13233268e+00 -6.69936895e-01 6.32463098e-02 -7.62260795e-01
-1.36393952e+00 8.83745193e-01 1.84779584e-01 -3.28582704e-01
1.24591780e+00 3.30993056e-01 6.84282780e-01 -5.19933291e-02
6.20898843e-01 -7.15860903e-01 5.10606110e-01 3.19911867e-01
1.07242382e+00 -1.00567865e+00 -4.63865399e-02 -3.08561862e-01
-7.53427327e-01 8.64646077e-01 8.53918195e-01 -2.68040627e-01
5.44632196e-01 2.81816989e-01 -1.39282793e-01 -2.23980621e-01
-8.82111907e-01 -9.19272527e-02 2.71342546e-01 6.55582070e-01
2.96120495e-01 -5.17850295e-02 5.92222035e-01 1.63847029e-01
-4.86189723e-01 -1.62428975e-01 5.57469428e-01 8.19705307e-01
-6.74350262e-02 -8.90645325e-01 -1.11967668e-01 2.44662151e-01
-3.98890488e-02 -5.28108254e-02 -9.90773588e-02 3.16928148e-01
2.03494355e-01 4.83002216e-01 2.70957917e-01 -5.13337433e-01
3.43399733e-01 -1.83302641e-01 9.45307970e-01 -9.06992257e-01
-1.91329330e-01 2.17425972e-01 -6.66864738e-02 -6.49254203e-01
-5.28580070e-01 -7.00079441e-01 -9.47165728e-01 -1.79275706e-01
-8.38596672e-02 -1.63760141e-01 6.70717478e-01 6.25826240e-01
3.28205019e-01 6.71513140e-01 1.11463439e+00 -1.31757414e+00
-2.36559063e-02 -7.42020309e-01 -5.89813113e-01 4.26898539e-01
3.10345203e-01 -5.28611064e-01 -2.68339127e-01 4.58860189e-01]
|
[9.265690803527832, -3.150806427001953]
|
d7b0cd10-1828-4b90-95b2-feea925f7037
|
reliable-prediction-intervals-with-directly
|
2302.00872
| null |
https://arxiv.org/abs/2302.00872v1
|
https://arxiv.org/pdf/2302.00872v1.pdf
|
Reliable Prediction Intervals with Directly Optimized Inductive Conformal Regression for Deep Learning
|
By generating prediction intervals (PIs) to quantify the uncertainty of each prediction in deep learning regression, the risk of wrong predictions can be effectively controlled. High-quality PIs need to be as narrow as possible, whilst covering a preset proportion of real labels. At present, many approaches to improve the quality of PIs can effectively reduce the width of PIs, but they do not ensure that enough real labels are captured. Inductive Conformal Predictor (ICP) is an algorithm that can generate effective PIs which is theoretically guaranteed to cover a preset proportion of data. However, typically ICP is not directly optimized to yield minimal PI width. However, in this study, we use Directly Optimized Inductive Conformal Regression (DOICR) that takes only the average width of PIs as the loss function and increases the quality of PIs through an optimized scheme under the validity condition that sufficient real labels are captured in the PIs. Benchmark experiments show that DOICR outperforms current state-of-the-art algorithms for regression problems using underlying Deep Neural Network structures for both tabular and image data.
|
['Anthony Bellotti', 'Haocheng Lei']
|
2023-02-02
| null | null | null | null |
['prediction-intervals']
|
['miscellaneous']
|
[ 3.32274884e-01 5.68892658e-01 -3.11523795e-01 -5.55931687e-01
-1.33091557e+00 -3.54253173e-01 4.35534537e-01 1.88918412e-01
-1.49801150e-01 9.92747366e-01 -1.50080711e-01 -1.27085492e-01
-4.16970700e-01 -1.13941562e+00 -1.13725102e+00 -7.35268950e-01
1.28959700e-01 7.94155300e-01 1.07551105e-01 6.04660809e-02
2.74100453e-01 4.41504955e-01 -1.43626773e+00 3.28187853e-01
1.08216465e+00 1.32121873e+00 -3.24356377e-01 2.23467454e-01
-1.02604434e-01 8.52493644e-01 -7.18437791e-01 -4.84562516e-01
2.65259415e-01 -3.03972721e-01 -5.42407572e-01 -5.26352346e-01
4.28792000e-01 -1.24068797e-01 1.54241666e-01 8.92917454e-01
3.54800254e-01 -1.54366130e-02 1.19614387e+00 -1.40209615e+00
-5.77883124e-01 9.24965799e-01 -7.74982154e-01 -9.06768069e-02
-9.50514600e-02 -3.71776462e-01 1.21117127e+00 -7.74954915e-01
4.23571497e-01 1.05801606e+00 9.36825633e-01 4.70587790e-01
-1.38747621e+00 -9.90469813e-01 -1.73800103e-02 -9.24191847e-02
-1.36579323e+00 -3.51160616e-02 7.02916741e-01 -4.21682596e-01
6.36770368e-01 2.32904539e-01 4.43025917e-01 7.54426062e-01
6.67886615e-01 4.46316391e-01 1.04382992e+00 -5.76666534e-01
1.91301301e-01 3.46773207e-01 -8.83512646e-02 5.58938444e-01
1.35398069e-02 3.69257212e-01 -5.50573826e-01 -1.37346342e-01
5.77128887e-01 -1.18123189e-01 -2.32836664e-01 -5.71419656e-01
-8.52702439e-01 1.19890094e+00 5.39490581e-01 -6.36617616e-02
2.68898141e-02 2.63708711e-01 4.00763720e-01 4.92145913e-03
7.92675555e-01 7.25285828e-01 -3.80680323e-01 1.99693352e-01
-1.09650338e+00 3.52036685e-01 5.46737432e-01 1.01676619e+00
6.89011395e-01 -2.52265364e-01 -6.57285690e-01 9.06244457e-01
3.95268537e-02 3.69571120e-01 -1.47293106e-01 -1.07830656e+00
4.49413210e-01 7.37464666e-01 1.63046598e-01 -1.01210451e+00
-2.42854729e-01 -6.12799346e-01 -8.89434099e-01 3.34606081e-01
4.83848900e-01 -1.10613815e-01 -8.93205404e-01 1.74536407e+00
2.33155474e-01 -1.19805142e-01 -2.36391276e-01 6.36540413e-01
6.23313069e-01 9.05033112e-01 -2.09927708e-02 -1.92332193e-01
9.08428609e-01 -7.23189235e-01 -5.01660764e-01 -1.08304463e-01
6.09035134e-01 -3.51300865e-01 1.14898252e+00 5.13666213e-01
-9.68111515e-01 -3.43713701e-01 -1.24630344e+00 7.92155974e-03
-2.97751933e-01 1.27695650e-01 2.95374244e-01 5.96571743e-01
-8.08244586e-01 7.79226899e-01 -4.59306180e-01 4.02427047e-01
5.98621011e-01 4.29605842e-01 -1.47938997e-01 -9.35666487e-02
-1.32485878e+00 1.01919270e+00 1.96076512e-01 5.16641475e-02
-6.96484268e-01 -1.20939910e+00 -8.12596738e-01 1.77649245e-01
3.33351672e-01 -1.90205619e-01 1.04832363e+00 -1.07600236e+00
-1.06261742e+00 6.70470774e-01 1.78708762e-01 -6.06220663e-01
8.16593111e-01 -8.65617171e-02 9.93574187e-02 -1.41102076e-01
-2.91634519e-02 1.11192751e+00 7.41867006e-01 -1.51302743e+00
-6.65336311e-01 -2.38022670e-01 -1.59471314e-02 1.91723555e-01
-2.87612051e-01 -3.43115807e-01 -3.53018165e-01 -3.33447665e-01
-2.02353075e-01 -9.48342502e-01 -3.00106704e-01 -6.71977475e-02
-3.37383181e-01 -3.24431688e-01 3.88821006e-01 -5.79237580e-01
1.13759100e+00 -1.95725572e+00 5.55542000e-02 4.55457717e-01
1.62138224e-01 -1.53916940e-01 4.22062166e-02 -2.58197784e-02
9.86967310e-02 1.45742595e-01 -6.22410059e-01 -2.80880988e-01
7.81233248e-05 2.96725601e-01 -5.60326695e-01 3.43043119e-01
3.88600111e-01 6.39452875e-01 -5.44530213e-01 -6.18885636e-01
3.98986135e-03 6.33532345e-01 -6.61245763e-01 1.61792636e-01
-3.97561818e-01 1.68225259e-01 -2.55920887e-01 3.78486425e-01
6.89394832e-01 -2.02088028e-01 -1.74657762e-01 -5.56361936e-02
1.08090326e-01 1.36308745e-01 -9.12412107e-01 9.53388453e-01
-4.72893089e-01 5.81177533e-01 -4.13337529e-01 -1.06377840e+00
1.37291348e+00 -3.85579653e-02 7.45469511e-01 -6.73635900e-01
-1.06096957e-02 1.43758297e-01 -3.12884659e-01 3.60383660e-01
3.65005672e-01 -1.57976806e-01 -1.93003163e-01 3.31449449e-01
-3.29108298e-01 -2.82866240e-01 -5.78063577e-02 -1.11070223e-01
8.59176874e-01 2.58113295e-01 1.04231872e-01 -3.03885221e-01
2.98548073e-01 3.00403893e-01 8.17318022e-01 8.02745819e-01
-5.29508619e-03 9.58458126e-01 9.85715508e-01 -4.20562059e-01
-1.50484776e+00 -9.74656463e-01 -6.45074189e-01 8.82151961e-01
-5.67765944e-02 -7.61646852e-02 -1.13596547e+00 -8.77823174e-01
-6.65898388e-03 9.88259494e-01 -9.10984576e-01 -1.99535057e-01
-4.96142626e-01 -8.57624233e-01 4.56188917e-01 6.86098814e-01
1.31581604e-01 -1.30092037e+00 -5.97779989e-01 1.38537481e-01
-4.84517440e-02 -8.91519845e-01 -3.42876911e-01 5.41462243e-01
-5.30082226e-01 -9.61583555e-01 -5.72929859e-01 -4.77932692e-01
8.60087156e-01 -2.52454609e-01 1.42793489e+00 -6.43640980e-02
-9.91420373e-02 -1.00463010e-01 -1.20570257e-01 -5.26703835e-01
-4.66602415e-01 1.00990474e-01 -4.74519700e-01 -4.93058205e-01
3.18453789e-01 -7.21591488e-02 -4.93403405e-01 4.28570926e-01
-7.58486152e-01 2.50034273e-01 3.46310288e-01 1.06075585e+00
1.16063058e+00 2.19579503e-01 7.84669220e-01 -1.09381902e+00
5.37742853e-01 -3.12076986e-01 -1.00424635e+00 3.95198196e-01
-1.03347576e+00 2.49024630e-01 7.23850369e-01 -4.91617858e-01
-9.11604166e-01 -1.03882506e-01 -1.62634030e-02 -4.36225563e-01
1.90701142e-01 4.76135552e-01 1.29390016e-01 1.44508585e-01
7.93579102e-01 -2.26235911e-01 -1.66307926e-01 1.04192562e-01
2.99274206e-01 4.40679431e-01 2.42527619e-01 -5.87163746e-01
2.53524601e-01 2.81738222e-01 3.95745456e-01 -1.15147950e-02
-1.22450209e+00 6.20806939e-04 -5.21263599e-01 -4.19449657e-01
7.08908200e-01 -5.76746583e-01 -6.58843637e-01 5.70509676e-03
-8.57817650e-01 -4.77559298e-01 -3.69517624e-01 9.61852670e-02
-6.98461711e-01 -1.99582219e-01 -2.05726519e-01 -8.08530390e-01
-7.02100694e-01 -1.24837387e+00 1.08318770e+00 2.71493457e-02
-3.13653111e-01 -7.37057090e-01 -1.20227054e-01 1.24640435e-01
1.28267869e-01 7.10533381e-01 1.20639217e+00 -4.71998304e-01
-4.12823349e-01 -3.91774833e-01 -4.07415956e-01 5.05993187e-01
-2.09191322e-01 1.53151050e-01 -8.37913930e-01 -9.35913473e-02
-2.57428527e-01 -3.58865589e-01 9.90888000e-01 7.95153022e-01
1.75243008e+00 -3.37417394e-01 -3.29361469e-01 5.64236343e-01
1.38722193e+00 3.13035816e-01 8.00645649e-01 3.57048273e-01
5.45904040e-01 9.17760134e-01 9.06500280e-01 3.63718808e-01
2.83696949e-01 6.88866198e-01 4.82678741e-01 -1.41567811e-01
1.88679218e-01 -3.24567229e-01 1.85414493e-01 1.60420850e-01
2.30575234e-01 -2.21598223e-01 -1.03100002e+00 4.56954449e-01
-1.82926905e+00 -6.60947084e-01 -7.53455311e-02 2.45179319e+00
1.08816898e+00 5.65051913e-01 -1.83620557e-01 2.64181048e-01
5.72209716e-01 -6.34046420e-02 -6.50023997e-01 -6.83512628e-01
1.23241559e-01 1.05937362e-01 8.43937635e-01 3.96757632e-01
-1.00008535e+00 6.83486640e-01 6.21179914e+00 1.08090818e+00
-8.63830626e-01 -7.56202191e-02 1.27106607e+00 -2.43338674e-01
-5.91444671e-01 -1.51390314e-01 -1.11578512e+00 3.79989088e-01
9.81384993e-01 1.25542268e-01 2.65698045e-01 1.17872000e+00
-1.56558245e-01 -5.72904572e-02 -1.21620166e+00 5.05970240e-01
1.83844026e-02 -1.47509587e+00 -2.17566326e-01 7.98157230e-02
8.91341805e-01 -1.70963630e-01 3.09724927e-01 3.50977361e-01
3.72357845e-01 -1.52117574e+00 7.71665454e-01 7.71642148e-01
1.15563488e+00 -1.33252943e+00 8.70980084e-01 3.26720178e-01
-6.96065485e-01 -1.42496750e-01 -6.07350707e-01 5.30220568e-01
-1.82174906e-01 8.50322545e-01 -9.13458824e-01 1.57728940e-01
9.52874541e-01 4.58794475e-01 -3.34759533e-01 7.90829360e-01
-3.85176986e-02 5.64722419e-01 -4.08662111e-01 -1.38415992e-01
2.58817732e-01 -2.42266908e-01 1.98827311e-02 1.08408630e+00
4.15323496e-01 -1.34750664e-01 -1.56530410e-01 1.09849811e+00
-9.67194065e-02 1.11986540e-01 -4.92724955e-01 3.43761653e-01
6.03159606e-01 9.39070165e-01 -5.32444656e-01 8.62070024e-02
-2.19204977e-01 2.38280639e-01 6.30819023e-01 -5.39260209e-02
-1.04253459e+00 -4.05973010e-02 2.58402199e-01 3.26775849e-01
2.57665396e-01 3.39821041e-01 -9.04672205e-01 -3.47064942e-01
-2.04355177e-02 -8.59151781e-01 5.59805870e-01 -8.36846948e-01
-1.37992942e+00 5.52726507e-01 1.79301694e-01 -1.19106483e+00
-2.44078979e-01 -5.85124671e-01 -3.26610684e-01 7.60500073e-01
-1.48811257e+00 -1.11589253e+00 -2.48970002e-01 1.84781700e-01
3.77600312e-01 -5.67782447e-02 8.35385442e-01 -1.24118151e-02
-4.17102933e-01 8.19910765e-01 2.78066993e-01 -2.37910807e-01
7.37032652e-01 -1.34563458e+00 -5.09653501e-02 3.65237534e-01
-1.59734398e-01 1.16633989e-01 7.28379965e-01 -5.61799586e-01
-6.06476545e-01 -1.16259062e+00 7.22877622e-01 -4.33040857e-01
1.88995540e-01 -3.25778097e-01 -9.93086278e-01 5.08812428e-01
-1.28580675e-01 1.20387346e-01 4.57551926e-01 1.41281739e-01
-3.42640638e-01 -5.86206138e-01 -1.29164243e+00 5.53717315e-01
5.70695281e-01 -1.02989361e-01 -1.11397967e-01 4.13151085e-01
8.07736814e-01 -5.89450955e-01 -1.18559420e+00 8.78193200e-01
6.85834706e-01 -8.92406881e-01 9.29683745e-01 -1.80451050e-01
1.05009687e+00 -7.00138435e-02 -2.99153347e-02 -1.27320480e+00
-1.03907578e-01 -1.49939284e-02 -6.61651567e-02 1.17729676e+00
8.61332059e-01 -3.16621542e-01 8.09527278e-01 8.05984080e-01
-6.56174570e-02 -1.37795055e+00 -9.59572315e-01 -3.87958258e-01
5.45030296e-01 -4.41198051e-01 8.41424167e-01 6.19362772e-01
-1.28931060e-01 -9.76189673e-02 -4.86660987e-01 -8.38285610e-02
7.46290326e-01 -2.15085372e-02 5.64980090e-01 -1.32047546e+00
1.32321820e-01 -4.48475331e-01 -3.47284675e-02 -3.65960062e-01
3.19311291e-01 -4.97452229e-01 2.26731077e-01 -1.56367791e+00
3.12641025e-01 -1.03904104e+00 -4.43162084e-01 4.63313460e-01
-8.14510286e-02 2.14857340e-01 -8.66811499e-02 7.43633658e-02
-1.61665574e-01 5.40812433e-01 1.35690570e+00 -2.00356454e-01
-4.32283431e-02 7.50586465e-02 -6.79864824e-01 5.44842303e-01
8.24150145e-01 -7.89832234e-01 -4.65374261e-01 -1.23079633e-03
6.50823534e-01 2.49706671e-01 1.21289544e-01 -6.73970938e-01
8.81137922e-02 -3.40802848e-01 7.88496256e-01 -9.46837068e-01
2.30381936e-01 -9.18956935e-01 2.30462730e-01 9.14393961e-02
-8.52030337e-01 -1.70916945e-01 2.06010919e-02 5.34667552e-01
-2.55374778e-02 -3.73947889e-01 1.07885385e+00 1.17808923e-01
-7.32307583e-02 2.17636868e-01 7.78666288e-02 -5.66846654e-02
1.13867593e+00 -5.86292744e-02 -2.15947717e-01 -2.90116012e-01
-2.76235908e-01 4.02489036e-01 2.51454204e-01 3.64688456e-01
6.46853685e-01 -1.31838596e+00 -7.11360335e-01 3.56083177e-02
1.78931117e-01 3.71472955e-01 -1.34379379e-02 4.65980053e-01
-4.59255874e-01 4.64719951e-01 -5.27945161e-02 -6.23629034e-01
-9.37174261e-01 6.01885259e-01 6.01850748e-01 -6.64441288e-01
-7.31115818e-01 8.17539752e-01 2.73072422e-01 -6.93140745e-01
5.18331289e-01 -1.53005451e-01 -4.66605663e-01 7.33959302e-02
2.71472335e-01 2.61549890e-01 1.27690092e-01 -3.19976598e-01
-1.29076261e-02 4.98861462e-01 -4.95104603e-02 -9.01242644e-02
1.44101715e+00 1.09713189e-01 -6.85587153e-02 2.58422196e-01
9.37313855e-01 -2.02390060e-01 -1.76456630e+00 -1.99131146e-02
-7.65058249e-02 -2.79486448e-01 3.16855401e-01 -9.85405803e-01
-1.24590766e+00 8.54870558e-01 5.04186749e-01 8.42238441e-02
1.23265505e+00 -9.10583511e-02 6.46652162e-01 -3.71852368e-02
2.25252122e-01 -1.22495770e+00 -2.90383678e-02 2.07073703e-01
1.07285058e+00 -1.25594795e+00 1.13794081e-01 -3.58433574e-01
-9.36766565e-01 9.90327001e-01 8.72418165e-01 -1.58507973e-01
5.57144105e-01 6.77382350e-01 -2.32536852e-01 -1.10575683e-01
-7.96232402e-01 4.36597496e-01 6.53753638e-01 5.37515879e-01
4.31994677e-01 1.86145782e-01 -3.81941080e-01 6.49359584e-01
-4.58687365e-01 -2.39152849e-01 3.34667981e-01 4.42066282e-01
-3.49091232e-01 -7.61929989e-01 -5.26580215e-01 7.81630754e-01
-5.91947794e-01 -4.06383239e-02 -1.64035305e-01 8.00163805e-01
7.65894651e-02 7.82014847e-01 3.01303267e-01 -1.99520588e-01
1.84002653e-01 -1.34764597e-01 4.29823190e-01 -2.76262105e-01
-3.68646264e-01 -1.05746374e-01 1.02702536e-01 -3.01130712e-01
-1.37482345e-01 -5.55223703e-01 -1.32718456e+00 -3.21968943e-01
-5.88178575e-01 1.06647294e-02 5.84660053e-01 1.00125515e+00
-1.26656517e-01 5.82178295e-01 6.28284752e-01 -4.69145060e-01
-7.34165132e-01 -8.29768300e-01 -3.90999734e-01 -4.60291607e-03
2.59126931e-01 -9.96347249e-01 -3.63189310e-01 -1.86397865e-01]
|
[7.957846164703369, 4.148825645446777]
|
3ca7e99b-37ab-461e-a174-4e30f221c973
|
e2v-sde-from-asynchronous-events-to-fast-and-1
|
2206.07578
| null |
https://arxiv.org/abs/2206.07578v2
|
https://arxiv.org/pdf/2206.07578v2.pdf
|
E2V-SDE: From Asynchronous Events to Fast and Continuous Video Reconstruction via Neural Stochastic Differential Equations
|
Event cameras respond to brightness changes in the scene asynchronously and independently for every pixel. Due to the properties, these cameras have distinct features: high dynamic range (HDR), high temporal resolution, and low power consumption. However, the results of event cameras should be processed into an alternative representation for computer vision tasks. Also, they are usually noisy and cause poor performance in areas with few events. In recent years, numerous researchers have attempted to reconstruct videos from events. However, they do not provide good quality videos due to a lack of temporal information from irregular and discontinuous data. To overcome these difficulties, we introduce an E2V-SDE whose dynamics are governed in a latent space by Stochastic differential equations (SDE). Therefore, E2V-SDE can rapidly reconstruct images at arbitrary time steps and make realistic predictions on unseen data. In addition, we successfully adopted a variety of image composition techniques for improving image clarity and temporal consistency. By conducting extensive experiments on simulated and real-scene datasets, we verify that our model outperforms state-of-the-art approaches under various video reconstruction settings. In terms of image quality, the LPIPS score improves by up to 12% and the reconstruction speed is 87% higher than that of ET-Net.
|
['Sungroh Yoon', 'Jeonghee Jo', 'Seongsik Park', 'Byunggook Na', 'Dongjin Lee', 'Jongwan Kim']
|
2022-06-15
|
e2v-sde-from-asynchronous-events-to-fast-and
|
http://openaccess.thecvf.com//content/CVPR2022/html/Kim_E2V-SDE_From_Asynchronous_Events_to_Fast_and_Continuous_Video_Reconstruction_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Kim_E2V-SDE_From_Asynchronous_Events_to_Fast_and_Continuous_Video_Reconstruction_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['video-reconstruction']
|
['computer-vision']
|
[ 1.51678756e-01 -6.40553415e-01 6.52447268e-02 -3.31709653e-01
-3.53152752e-01 -2.31565386e-01 3.67194444e-01 -4.51301903e-01
-3.59349608e-01 6.59448445e-01 1.53641716e-01 1.77896336e-01
-2.63355896e-02 -6.22598946e-01 -7.81615853e-01 -9.16977465e-01
1.26753241e-01 -1.19290635e-01 6.82656586e-01 2.22245052e-01
-6.71977699e-02 3.25923413e-01 -1.61781752e+00 1.72523424e-01
8.06295455e-01 9.95602965e-01 6.01824701e-01 5.41680098e-01
6.32520914e-02 1.26527023e+00 -3.13923031e-01 -2.49939710e-01
2.14320630e-01 -4.49926764e-01 -2.28281423e-01 4.87572074e-01
2.14807957e-01 -8.51477504e-01 -9.11473393e-01 1.11397016e+00
3.36613685e-01 1.61565647e-01 2.18676731e-01 -1.18328774e+00
-4.39405084e-01 9.09427851e-02 -7.59180844e-01 3.99006993e-01
4.78620291e-01 4.14582610e-01 4.66834158e-01 -7.60679007e-01
8.19716752e-01 1.13310385e+00 4.29281503e-01 5.92206180e-01
-1.24710047e+00 -7.49973178e-01 2.16609195e-01 5.12328565e-01
-1.25084639e+00 -6.71554983e-01 6.94911659e-01 -1.14502653e-01
7.98149586e-01 9.25692096e-02 7.06015110e-01 1.37566447e+00
3.09803098e-01 9.14956033e-01 1.01188886e+00 9.34677497e-02
2.07549036e-01 -1.60830393e-01 -3.55730116e-01 4.58887964e-01
1.32543638e-01 7.63198137e-02 -7.66033888e-01 1.72385886e-01
1.02742732e+00 5.11198640e-01 -6.97485268e-01 -1.20903969e-01
-1.50053287e+00 3.96789640e-01 1.08833358e-01 -4.57497537e-02
-5.57328701e-01 2.32729182e-01 3.32451403e-01 1.10722937e-01
2.40112796e-01 -2.33576253e-01 -1.31196976e-01 -3.39170814e-01
-8.08901429e-01 -1.36871943e-02 5.63637733e-01 1.03063059e+00
3.91595662e-01 2.93927372e-01 -8.06787908e-02 7.01962292e-01
2.94846177e-01 7.76018322e-01 2.68352300e-01 -1.30290639e+00
4.79568481e-01 8.58903956e-03 2.26681679e-01 -1.20889080e+00
6.82773516e-02 3.01383305e-02 -1.36097550e+00 5.12029156e-02
2.83315957e-01 -3.09995878e-02 -7.45759785e-01 1.62373137e+00
2.16270983e-01 6.82004452e-01 1.46387652e-01 1.15615392e+00
6.66462243e-01 1.26950967e+00 4.96529229e-03 -8.79172862e-01
1.10679591e+00 -7.80685008e-01 -1.15831697e+00 -8.53093565e-02
-2.05822140e-01 -7.84921587e-01 6.96110785e-01 8.37863803e-01
-1.29580903e+00 -7.46059060e-01 -7.54078507e-01 1.52239010e-01
3.86420399e-01 -4.75043841e-02 4.16511029e-01 1.75534263e-01
-9.32245493e-01 4.19736415e-01 -1.08731163e+00 -3.52668017e-01
3.54630530e-01 2.98690069e-02 -3.67991358e-01 -3.93693417e-01
-9.57548857e-01 4.11769062e-01 2.30846241e-01 1.47481605e-01
-1.14167154e+00 -3.59724045e-01 -6.86137319e-01 -6.05473369e-02
4.16819274e-01 -5.39513588e-01 1.08940029e+00 -1.02673948e+00
-1.56596375e+00 4.13622528e-01 -4.15717751e-01 -3.79938841e-01
6.85755968e-01 -8.06761086e-02 -6.48063004e-01 4.04075116e-01
-6.54891282e-02 5.08197188e-01 8.34877670e-01 -1.11336386e+00
-7.60100067e-01 -5.71105815e-02 -1.12738274e-01 1.95312589e-01
-2.56104052e-01 9.88034979e-02 -1.03581035e+00 -6.81910276e-01
1.92108855e-01 -8.65181148e-01 -1.73621953e-01 2.73537755e-01
-3.15348245e-02 1.68067828e-01 1.10709965e+00 -6.63909674e-01
1.08795357e+00 -2.37464476e+00 1.04443476e-01 -3.58214527e-01
2.51216888e-01 2.97015250e-01 -1.33112185e-02 2.69645721e-01
1.33807704e-01 -2.81246960e-01 -3.51198800e-02 -2.90382594e-01
-4.26645488e-01 5.24719656e-01 -6.35048926e-01 5.43457627e-01
3.72169837e-02 6.32314444e-01 -9.98468757e-01 -7.03253627e-01
6.88006461e-01 6.57634795e-01 -3.18100154e-01 3.21169138e-01
-1.21408768e-01 5.53414404e-01 -3.97651523e-01 5.24078548e-01
8.40299904e-01 -5.14176309e-01 2.53238171e-01 -3.55395526e-01
-9.10394862e-02 -1.36616498e-01 -1.51310050e+00 1.58704054e+00
-3.64285886e-01 1.01572609e+00 -6.61856309e-02 -7.01588929e-01
5.76940656e-01 4.16246533e-01 6.36123896e-01 -9.35626566e-01
-4.96988408e-02 -3.11749452e-03 -3.36572438e-01 -7.35160291e-01
5.94945788e-01 9.73814875e-02 4.16020185e-01 1.42232567e-01
-9.99130681e-02 3.20126206e-01 2.15726271e-01 2.41354495e-01
1.08923531e+00 1.82710081e-01 1.94515958e-01 3.33776921e-01
3.23399961e-01 -3.60161901e-01 1.07422578e+00 6.02842569e-01
-3.12465221e-01 9.66268241e-01 9.23975930e-02 -4.29485559e-01
-1.12357140e+00 -1.23244929e+00 -1.84900388e-01 3.87529403e-01
7.92175055e-01 -3.45148265e-01 -4.20265049e-01 -7.88485929e-02
-5.40045977e-01 5.25970817e-01 -4.28690501e-02 4.09965441e-02
-4.75141168e-01 -7.32091367e-01 3.21455240e-01 5.53871810e-01
9.17053819e-01 -9.72588062e-01 -7.25015640e-01 5.20445287e-01
-6.65888488e-01 -1.75281739e+00 -5.37783325e-01 -2.85855055e-01
-9.78679478e-01 -1.00537872e+00 -6.21955574e-01 -6.50460541e-01
5.74904323e-01 8.33980441e-01 9.94264245e-01 -2.10710809e-01
-2.33545303e-01 3.00180584e-01 -3.85695666e-01 3.52091007e-02
-2.05551401e-01 -6.52541757e-01 2.08398521e-01 4.30044472e-01
9.87104103e-02 -6.27488077e-01 -8.23492289e-01 5.41780770e-01
-1.39063346e+00 4.80455816e-01 4.44120109e-01 7.14270711e-01
8.84193361e-01 5.03081620e-01 1.18437305e-01 -4.51482266e-01
9.23609827e-03 -3.82107496e-01 -7.85626113e-01 2.32743740e-01
-3.71029675e-01 -1.92668214e-01 8.63924384e-01 -7.81215072e-01
-1.48566532e+00 2.26458356e-01 1.35946810e-01 -8.47434938e-01
-1.18027329e-01 3.01201898e-03 -2.06430584e-01 2.23891109e-01
1.92409113e-01 6.79213166e-01 -5.41077219e-02 -1.23196982e-01
1.56272929e-02 6.40498638e-01 6.59483552e-01 -3.43322575e-01
6.93211317e-01 9.20819700e-01 -4.89045233e-02 -9.34200764e-01
-5.39164126e-01 -3.18933815e-01 -2.22254321e-02 -5.28987825e-01
9.30702090e-01 -1.37464905e+00 -8.06101561e-01 9.77876365e-01
-1.22461510e+00 -2.30811641e-01 -1.47846714e-01 7.55713940e-01
-4.38181579e-01 7.35028207e-01 -1.00676751e+00 -7.52009034e-01
-2.43050512e-02 -1.17184317e+00 9.77928042e-01 5.16807497e-01
3.50822002e-01 -8.08545411e-01 -2.95661896e-01 1.48365378e-01
2.84688413e-01 1.97906762e-01 2.80352414e-01 3.59672904e-01
-1.17354596e+00 1.59137979e-01 -3.39254200e-01 4.01795477e-01
1.36612505e-01 3.14938784e-01 -8.98834229e-01 -2.51960725e-01
3.97558361e-01 -4.37875055e-02 8.12806904e-01 6.44783318e-01
1.38228369e+00 -1.65677354e-01 -2.06759721e-01 9.46515977e-01
1.68353319e+00 3.51082236e-01 1.03979087e+00 1.55179664e-01
5.77988684e-01 2.32639253e-01 7.51524627e-01 7.84951806e-01
3.50743324e-01 7.21607864e-01 5.69010556e-01 1.03132792e-01
-1.85053945e-01 -3.10820818e-01 8.36613655e-01 9.53884363e-01
-1.53683424e-01 -6.76738203e-01 -4.76692289e-01 6.31097913e-01
-2.11485410e+00 -1.43069935e+00 -4.85859513e-01 2.08883715e+00
7.32496321e-01 7.06433430e-02 -1.90478846e-01 -6.73498586e-02
8.01486373e-01 4.78422463e-01 -6.67155206e-01 2.61780977e-01
-6.02036417e-01 -3.90763462e-01 6.47681713e-01 1.08557567e-01
-8.25391710e-01 6.80455804e-01 6.10911226e+00 7.84479022e-01
-1.04549491e+00 4.97652851e-02 6.20668054e-01 -3.89763385e-01
4.37702332e-03 -7.32792402e-03 -7.52162516e-01 9.84440148e-01
7.56227911e-01 -1.09625541e-01 6.40619159e-01 4.76182729e-01
5.64839065e-01 -2.81318963e-01 -8.89851749e-01 1.48602915e+00
9.98087898e-02 -1.33783889e+00 4.33044285e-02 -3.11637390e-02
9.10143197e-01 1.14442982e-01 1.09501847e-03 -1.88902050e-01
1.26271471e-01 -6.43769622e-01 7.36913323e-01 7.04965413e-01
8.68810654e-01 -5.01633584e-01 6.04539692e-01 3.89070094e-01
-1.28695917e+00 6.76457733e-02 -5.79425871e-01 -2.50292011e-02
6.26305819e-01 7.16842949e-01 -1.17832631e-01 5.50791979e-01
1.03303170e+00 1.19498193e+00 -2.47602761e-01 1.00164711e+00
-1.30832747e-01 6.88098013e-01 -4.42975253e-01 3.43405515e-01
-1.31595684e-02 -4.34693098e-01 5.55394650e-01 9.61501300e-01
5.19811332e-01 5.04521608e-01 3.27968188e-02 6.82037592e-01
6.56167865e-02 -3.67021441e-01 -6.04287624e-01 2.30418816e-01
4.03388292e-01 1.10233426e+00 -6.22513473e-01 -3.30116928e-01
-7.30381966e-01 1.40716529e+00 -1.95644781e-01 6.03919923e-01
-1.24252701e+00 -5.82559705e-02 7.42533863e-01 -6.17982633e-03
4.73093897e-01 -2.48000085e-01 2.93295175e-01 -1.69827449e+00
2.50432938e-01 -8.48146081e-01 1.14089444e-01 -1.31409347e+00
-1.28698480e+00 5.03120363e-01 -1.24359161e-01 -1.67355287e+00
-1.52197570e-01 -2.80844182e-01 -3.04025441e-01 3.87029141e-01
-1.54017079e+00 -6.25387371e-01 -7.36559212e-01 9.68694568e-01
9.01009977e-01 -1.41862687e-02 3.18508983e-01 5.40756166e-01
-6.15234196e-01 4.68147472e-02 5.43525338e-01 3.46373916e-02
7.73186803e-01 -8.07942927e-01 2.20542103e-01 1.33002114e+00
1.51892036e-01 1.43832609e-01 6.97802126e-01 -5.20219266e-01
-1.77184343e+00 -1.30721688e+00 4.10532534e-01 -9.71617475e-02
3.94674242e-01 -6.86517283e-02 -9.71703053e-01 6.27705038e-01
2.37607434e-01 4.41352189e-01 1.75421044e-01 -6.21407807e-01
-8.64546001e-02 -4.27104831e-01 -8.18469226e-01 6.28997684e-01
1.13037264e+00 -4.64595526e-01 -3.09865981e-01 1.67328790e-01
6.93037152e-01 -5.53321838e-01 -7.29182303e-01 1.60338730e-01
2.82756478e-01 -1.25398552e+00 9.56035316e-01 6.07498139e-02
5.99557698e-01 -6.09769464e-01 -2.95929164e-01 -8.55563283e-01
-2.30870456e-01 -7.40374923e-01 -4.35538769e-01 1.30385590e+00
-3.59464407e-01 -4.70629394e-01 5.08085608e-01 5.32668531e-01
2.20900044e-01 -3.14276040e-01 -7.72502542e-01 -8.34365904e-01
-8.11295569e-01 -6.00859523e-01 2.48942196e-01 7.07600892e-01
-6.40058696e-01 1.52460560e-01 -1.02900696e+00 4.05194521e-01
1.00012541e+00 1.17640503e-01 6.75081432e-01 -8.89396310e-01
-3.73092204e-01 6.21379204e-02 -4.65572357e-01 -1.52096879e+00
-3.36106005e-03 -2.38183975e-01 1.42524153e-01 -1.45162022e+00
5.33323526e-01 -2.08513755e-02 -1.95003480e-01 1.89665593e-02
-2.41779670e-01 2.41467968e-01 2.93143511e-01 4.74315643e-01
-1.00775290e+00 7.19504058e-01 1.16348839e+00 7.07965195e-02
2.52015516e-02 -2.94670492e-01 -1.26594305e-01 7.31771767e-01
4.16974604e-01 -4.45187390e-01 -4.42333639e-01 -8.23160231e-01
1.00446135e-01 6.14738405e-01 5.55313349e-01 -1.10626018e+00
5.27242064e-01 -4.47553128e-01 5.13662636e-01 -6.58628345e-01
4.63263303e-01 -1.03249586e+00 8.09832096e-01 2.80928671e-01
8.60190317e-02 5.16919792e-03 -8.42544343e-03 1.07147944e+00
-5.34714162e-01 1.58054322e-01 8.28813672e-01 7.19448924e-02
-1.12770808e+00 5.49114287e-01 -4.73370314e-01 -5.31607531e-02
1.23439264e+00 -4.34781790e-01 -2.63952106e-01 -7.10776508e-01
-2.01150581e-01 1.87966481e-01 8.40574026e-01 2.95037419e-01
9.66318607e-01 -1.29571986e+00 -6.35217190e-01 2.11664110e-01
-1.31955398e-02 1.13020666e-01 7.75813103e-01 8.34686399e-01
-6.62111878e-01 1.05388388e-01 -2.36836463e-01 -9.72174048e-01
-1.29006553e+00 6.10744298e-01 1.57877401e-01 -1.36713532e-03
-1.06470251e+00 4.18445617e-01 4.47353035e-01 3.24944913e-01
2.90135175e-01 -1.99057341e-01 4.98449430e-02 -2.55130112e-01
7.18762696e-01 3.99839371e-01 -4.32258546e-01 -6.35142446e-01
-1.99068204e-01 5.74817181e-01 9.24023017e-02 2.89895888e-02
1.47915876e+00 -6.28787875e-01 1.79316804e-01 3.35067272e-01
1.01651907e+00 -1.46446824e-01 -1.94432974e+00 -4.17269707e-01
-6.63441122e-01 -8.92133236e-01 2.21036360e-01 -3.41482162e-01
-1.46908140e+00 6.92163765e-01 6.74949646e-01 1.29972667e-01
1.68919528e+00 -3.04763824e-01 1.10103679e+00 1.62613049e-01
4.70098674e-01 -9.41329062e-01 1.30552799e-01 2.18837351e-01
3.98189396e-01 -1.37068689e+00 3.58345620e-02 -3.35079670e-01
-8.36563528e-01 1.13890123e+00 4.49207872e-01 -2.44141389e-02
3.64293784e-01 3.30156147e-01 -1.07643269e-01 2.07532331e-01
-1.12694395e+00 6.39281720e-02 -1.44982934e-01 5.17943323e-01
-3.22983861e-02 -2.94404447e-01 -5.16213290e-02 1.30183384e-01
4.96007711e-01 3.06682378e-01 8.50145280e-01 6.41348004e-01
-4.75442037e-02 -7.51153409e-01 -4.27841246e-01 1.52533516e-01
-5.59074759e-01 1.30432755e-01 3.92241061e-01 6.38264835e-01
-1.19273029e-01 1.19570720e+00 1.42044097e-01 -3.09703827e-01
2.30965152e-01 -5.77654183e-01 4.18827206e-01 6.03029598e-03
6.26706406e-02 2.55077720e-01 -2.68985838e-01 -9.73529637e-01
-8.62125754e-01 -9.16547239e-01 -1.13927674e+00 -7.32844949e-01
-2.01412275e-01 -1.75985217e-01 4.24964756e-01 6.86111152e-01
4.04257327e-01 6.12602592e-01 8.06892395e-01 -7.08984137e-01
-2.95403749e-01 -5.31909704e-01 -7.29995847e-01 7.05613315e-01
2.84759641e-01 -3.88012320e-01 -4.47751522e-01 5.76551974e-01]
|
[10.769248962402344, -1.985932469367981]
|
a8c30171-6568-44c2-ae21-64c3efd38be3
|
basicvsr-improving-video-super-resolution
|
2104.13371
| null |
https://arxiv.org/abs/2104.13371v1
|
https://arxiv.org/pdf/2104.13371v1.pdf
|
BasicVSR++: Improving Video Super-Resolution with Enhanced Propagation and Alignment
|
A recurrent structure is a popular framework choice for the task of video super-resolution. The state-of-the-art method BasicVSR adopts bidirectional propagation with feature alignment to effectively exploit information from the entire input video. In this study, we redesign BasicVSR by proposing second-order grid propagation and flow-guided deformable alignment. We show that by empowering the recurrent framework with the enhanced propagation and alignment, one can exploit spatiotemporal information across misaligned video frames more effectively. The new components lead to an improved performance under a similar computational constraint. In particular, our model BasicVSR++ surpasses BasicVSR by 0.82 dB in PSNR with similar number of parameters. In addition to video super-resolution, BasicVSR++ generalizes well to other video restoration tasks such as compressed video enhancement. In NTIRE 2021, BasicVSR++ obtains three champions and one runner-up in the Video Super-Resolution and Compressed Video Enhancement Challenges. Codes and models will be released to MMEditing.
|
['Chen Change Loy', 'Xiangyu Xu', 'Shangchen Zhou', 'Kelvin C. K. Chan']
|
2021-04-27
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Chan_BasicVSR_Improving_Video_Super-Resolution_With_Enhanced_Propagation_and_Alignment_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Chan_BasicVSR_Improving_Video_Super-Resolution_With_Enhanced_Propagation_and_Alignment_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['video-enhancement', 'video-restoration']
|
['computer-vision', 'computer-vision']
|
[ 4.10804778e-01 -1.99151799e-01 -3.17833185e-01 2.39771474e-02
-8.37048471e-01 -3.15006763e-01 2.68152118e-01 -4.27011609e-01
-2.16191813e-01 7.14291513e-01 7.41344869e-01 3.56329083e-02
-1.74750879e-01 -4.89592373e-01 -6.65796340e-01 -4.97044027e-01
-2.74184555e-01 -3.85917902e-01 2.90836662e-01 -6.13594949e-01
3.69057894e-01 2.77792156e-01 -1.32457244e+00 5.09195387e-01
9.03912306e-01 8.37368488e-01 6.85267627e-01 7.23047256e-01
2.49976397e-01 1.04195225e+00 -6.22670986e-02 -2.31325075e-01
3.62956554e-01 -4.11019057e-01 -7.60239720e-01 3.63177471e-02
5.58460236e-01 -5.56186378e-01 -8.87867510e-01 1.07489741e+00
6.22991860e-01 2.92941660e-01 3.53797860e-02 -7.62810230e-01
-9.06200945e-01 6.82159543e-01 -9.37824965e-01 8.25899839e-01
4.94971335e-01 -3.64998244e-02 9.18722332e-01 -1.04346895e+00
9.04604256e-01 1.18712080e+00 7.56487370e-01 4.96232003e-01
-1.22312975e+00 -4.93690282e-01 2.67773449e-01 5.53370953e-01
-1.34423852e+00 -6.53390050e-01 7.06028759e-01 -2.72186428e-01
8.43350053e-01 4.15961236e-01 4.48484629e-01 9.65873420e-01
1.00170754e-01 7.65556157e-01 9.72641706e-01 -4.54609171e-02
-6.94211870e-02 -3.79443288e-01 -2.10108869e-02 4.40238327e-01
-1.00322828e-01 1.65948808e-01 -9.55217659e-01 1.34397030e-01
1.26693225e+00 -7.24735716e-03 -8.31636786e-01 9.21162814e-02
-1.27809203e+00 5.12665689e-01 4.64783907e-01 4.04547542e-01
-4.97748882e-01 1.83263376e-01 3.41005772e-01 2.90269852e-01
5.61404109e-01 3.33587795e-01 -2.85378210e-02 -2.03043327e-01
-1.27600694e+00 2.83378094e-01 8.80170688e-02 8.80830705e-01
4.92832154e-01 4.95144725e-01 -4.34725940e-01 8.59922230e-01
1.10378049e-01 3.45703632e-01 5.44091821e-01 -1.63370359e+00
5.62405467e-01 -4.62469198e-02 2.40746841e-01 -1.20150757e+00
-1.49467930e-01 -7.77711928e-01 -1.28862727e+00 1.49620429e-01
-1.44536778e-01 5.95680848e-02 -7.09743500e-01 1.58212936e+00
1.12725943e-02 6.07885897e-01 1.09544553e-01 1.28780067e+00
8.28518748e-01 8.97524834e-01 -3.03948730e-01 -5.21010518e-01
1.14020681e+00 -1.10570514e+00 -1.03413534e+00 -4.57105637e-02
8.41632932e-02 -8.37567687e-01 6.97039783e-01 3.91458124e-01
-1.59950221e+00 -7.07104802e-01 -1.05775023e+00 -1.06301211e-01
4.38155711e-01 -2.00078711e-01 4.29101348e-01 2.68080413e-01
-1.55923223e+00 9.22499239e-01 -7.74027228e-01 -1.27925351e-01
3.92915487e-01 2.13359252e-01 -3.93854827e-01 -2.94513643e-01
-1.19789183e+00 5.67555964e-01 -2.57950798e-02 1.55352488e-01
-8.45106423e-01 -9.99052465e-01 -8.68975401e-01 -2.05578525e-02
4.62015629e-01 -7.81984150e-01 9.37511921e-01 -7.38638878e-01
-1.51527464e+00 5.44551432e-01 -4.88907218e-01 -6.95943415e-01
7.20509350e-01 -4.51206088e-01 -4.97701108e-01 2.67983049e-01
3.32252234e-02 3.65695626e-01 1.01331079e+00 -1.18356681e+00
-6.49864316e-01 -2.16652736e-01 4.33584005e-02 4.63346332e-01
-3.45155150e-01 2.00800210e-01 -6.66675091e-01 -1.15666580e+00
2.67575860e-01 -6.15554452e-01 -4.14111704e-01 -1.01332076e-01
-1.85276404e-01 3.25953841e-01 8.36815834e-01 -1.13412201e+00
1.49640429e+00 -2.17084575e+00 6.82378709e-01 -1.25271156e-01
7.09512055e-01 2.89999217e-01 -2.51232356e-01 1.27553493e-01
-6.14505149e-02 1.16575204e-01 -1.95963129e-01 -2.64873385e-01
-3.67153138e-01 -4.70541418e-02 -4.02252972e-01 4.46649015e-01
-1.68850943e-01 9.02779818e-01 -8.40829372e-01 -3.14923018e-01
1.28959313e-01 9.28057551e-01 -8.63463640e-01 1.27227694e-01
2.65405655e-01 6.96378469e-01 -2.40249380e-01 4.14363712e-01
7.39118338e-01 -4.60105717e-01 6.90024123e-02 -5.27226567e-01
-3.46219540e-01 -3.89139429e-02 -1.25808191e+00 1.95647287e+00
-6.06625319e-01 8.64361048e-01 4.01543468e-01 -7.44835436e-01
8.11387897e-01 4.24518168e-01 6.20966971e-01 -8.97338986e-01
-3.93924028e-01 4.20495942e-02 -3.92315805e-01 -5.10002434e-01
1.08437538e+00 1.44383579e-01 4.92013663e-01 -1.92921124e-02
-9.21426713e-02 2.71107793e-01 2.04184949e-01 4.43761528e-01
1.10917079e+00 2.38931417e-01 8.97435769e-02 -1.84312016e-01
6.02782369e-01 -4.61744905e-01 8.32166791e-01 6.91663563e-01
-1.48785129e-01 8.23082864e-01 -1.44842193e-02 -1.72839284e-01
-1.27683556e+00 -1.13078201e+00 7.02088401e-02 9.48975623e-01
3.96943867e-01 -7.09692836e-01 -5.72303593e-01 -9.18066502e-02
-3.95022720e-01 3.28489929e-01 -4.19532090e-01 2.12468386e-01
-1.05050993e+00 -5.54413974e-01 2.39622355e-01 5.07250965e-01
8.96539032e-01 -6.65828526e-01 -4.46707070e-01 3.30186933e-01
-7.35779524e-01 -1.48462594e+00 -8.32883358e-01 -5.80082536e-01
-9.92442310e-01 -5.96047819e-01 -1.31315935e+00 -7.39370108e-01
1.94111049e-01 6.08949006e-01 1.03851819e+00 -2.46698149e-02
-1.72469378e-01 2.31625572e-01 -4.77306068e-01 6.39491856e-01
-3.45474273e-01 -1.29116312e-01 6.62885159e-02 1.29961878e-01
-4.88863319e-01 -9.13629413e-01 -9.07016218e-01 3.73383701e-01
-7.29764879e-01 3.66536528e-01 4.32923079e-01 7.39032924e-01
7.16310978e-01 8.77693892e-02 3.66290748e-01 -5.51461041e-01
3.80436450e-01 -3.56858999e-01 -4.13677245e-01 2.96171885e-02
-5.47718108e-01 -4.64963503e-02 6.38025343e-01 -2.69027323e-01
-1.20375299e+00 -4.42578524e-01 -2.35312387e-01 -7.29696274e-01
3.04368615e-01 2.88893223e-01 5.00505194e-02 -2.25139454e-01
4.75335032e-01 4.69512224e-01 -2.17174917e-01 -6.52754843e-01
3.59925807e-01 3.57477665e-01 9.33896244e-01 -1.50596216e-01
9.30611253e-01 6.68500960e-01 -6.35267794e-02 -8.87883961e-01
-4.46724623e-01 -3.73534292e-01 -3.33559185e-01 -3.25727433e-01
9.91330504e-01 -1.17343915e+00 -5.94179332e-01 4.34281319e-01
-1.04285562e+00 -3.11522812e-01 -2.42416978e-01 4.78112936e-01
-6.29485786e-01 7.14684129e-01 -1.00037444e+00 -4.90238696e-01
-4.37034786e-01 -1.22395015e+00 8.48438501e-01 2.29123682e-01
2.04497799e-01 -7.82331467e-01 -4.65297326e-02 5.77659130e-01
9.72867191e-01 1.83644630e-02 4.13108081e-01 2.21799880e-01
-7.99162805e-01 3.71857077e-01 -4.00376320e-01 3.77511114e-01
1.59195960e-01 -4.04661447e-01 -6.57929718e-01 -6.11054599e-01
8.52295607e-02 1.86183840e-01 1.17604423e+00 6.00998044e-01
1.11596525e+00 -3.89891833e-01 -5.02067916e-02 1.19134641e+00
1.45091963e+00 1.63245156e-01 9.09040570e-01 4.00622517e-01
7.73742676e-01 1.83514327e-01 5.52550614e-01 5.51976562e-01
3.65749270e-01 1.11901927e+00 4.57718521e-01 -2.01566741e-01
-5.43358684e-01 -1.12406991e-01 5.22973776e-01 1.03785253e+00
-7.32361674e-01 5.86706772e-02 -5.07796645e-01 3.63958418e-01
-1.89929152e+00 -1.35715413e+00 4.89610024e-02 2.00989890e+00
7.23161280e-01 -1.62876248e-01 -8.95983353e-02 1.84037108e-02
9.52150702e-01 6.21786475e-01 -3.86315554e-01 9.45091061e-03
-4.64984089e-01 1.23171709e-01 5.55639267e-01 7.82589138e-01
-1.00145102e+00 9.70368981e-01 6.39135742e+00 9.88094747e-01
-1.01069093e+00 2.22929806e-01 4.41565692e-01 -8.62906724e-02
-4.43321973e-01 -5.17786779e-02 -7.13903725e-01 3.53177518e-01
6.48864865e-01 -3.24665040e-01 8.06716681e-01 3.16466540e-01
6.55558825e-01 2.02433631e-01 -6.14203751e-01 1.37268758e+00
1.79245174e-01 -1.86098599e+00 5.77381179e-02 -4.85523678e-02
8.57019544e-01 -3.28384154e-02 3.67536247e-01 -1.75662711e-01
2.58689504e-02 -1.10068691e+00 6.13643587e-01 5.25696933e-01
1.02398682e+00 -5.71531892e-01 5.28262258e-01 -5.39119616e-02
-1.49529600e+00 -2.32974410e-01 -1.79375827e-01 1.47677779e-01
7.42284894e-01 4.36767220e-01 -5.94951138e-02 8.69084418e-01
9.40698802e-01 1.32150209e+00 -3.15135449e-01 1.01382089e+00
-2.56071053e-02 4.62828308e-01 9.47961733e-02 8.39175820e-01
2.88292877e-02 -1.38115466e-01 1.07099080e+00 1.30895269e+00
5.36198914e-01 2.68750906e-01 1.49869695e-01 6.54707253e-01
-7.19657764e-02 -7.57293850e-02 -2.39560857e-01 2.83877581e-01
3.69576126e-01 1.14027643e+00 -2.63544679e-01 -3.20421010e-01
-4.01504815e-01 1.33768821e+00 3.07186879e-03 7.45750666e-01
-9.03011858e-01 -1.21857882e-01 8.04483652e-01 4.25188601e-01
4.73645598e-01 -3.45087618e-01 -2.00746581e-02 -1.47270513e+00
-9.68238804e-03 -9.93428648e-01 4.80021536e-02 -9.18786466e-01
-8.36586058e-01 9.27518070e-01 -1.90092534e-01 -1.29786074e+00
-1.72700614e-01 -1.21792831e-01 -2.09335491e-01 7.81196058e-01
-1.78581488e+00 -9.85713065e-01 -5.31515837e-01 8.91955733e-01
9.96595800e-01 -2.65966773e-01 4.12816525e-01 5.84713757e-01
-4.72856283e-01 5.18737555e-01 9.41766277e-02 9.19933021e-02
6.67876542e-01 -9.57846403e-01 5.05534410e-01 1.31164801e+00
-1.48397312e-01 4.41759259e-01 1.02080870e+00 -5.70286989e-01
-1.47369099e+00 -1.15051842e+00 4.67921764e-01 1.96520641e-01
6.03422344e-01 4.84620295e-02 -9.59801018e-01 7.18103647e-01
4.25973117e-01 2.08846971e-01 2.03088984e-01 -3.43566835e-01
-5.30338287e-01 -1.73128948e-01 -1.06774020e+00 7.33574510e-01
1.54552281e+00 -6.37856185e-01 -3.09575468e-01 1.48208141e-01
9.78575528e-01 -5.99671304e-01 -1.05923152e+00 5.05378783e-01
3.28657061e-01 -1.16075182e+00 1.24336421e+00 -3.64455819e-01
8.14415157e-01 -3.87114763e-01 -6.65441036e-01 -1.10892141e+00
-7.00801849e-01 -1.34134328e+00 -4.30310518e-01 1.08417213e+00
1.53275700e-02 -5.39538205e-01 5.94077587e-01 2.36451969e-01
-2.03813449e-01 -5.82468331e-01 -8.17909360e-01 -7.23093808e-01
-2.58016855e-01 -2.28848815e-01 2.33348742e-01 9.37218785e-01
-2.10387528e-01 1.70490034e-02 -9.45446193e-01 2.57532597e-01
1.00322640e+00 1.50538329e-02 3.87377560e-01 -6.32662475e-01
-6.88856125e-01 -2.84738123e-01 -4.27146614e-01 -1.54979646e+00
-1.07299395e-01 -7.99550176e-01 -3.74497205e-01 -1.51505947e+00
4.67575610e-01 -3.11004966e-01 -3.23953032e-01 9.92809534e-02
-1.07608981e-01 7.53268600e-01 6.82017505e-01 3.54799539e-01
-6.78738952e-01 4.48227167e-01 1.43634915e+00 4.62513380e-02
-3.14639747e-01 -2.44542807e-01 -7.56306052e-01 6.67792439e-01
6.34909272e-01 7.30609521e-02 -2.98923165e-01 -7.38376796e-01
2.20497131e-01 6.29802048e-01 4.45421875e-01 -9.01394427e-01
2.73453802e-01 7.82906078e-03 2.34216437e-01 -2.76418805e-01
5.41342795e-01 -4.19219047e-01 3.69146228e-01 1.80193737e-01
-4.94536251e-01 4.07789230e-01 7.46949837e-02 5.75659692e-01
-3.71477902e-01 2.01058269e-01 9.37880874e-01 -8.84583294e-02
-9.29701686e-01 4.37157720e-01 -3.37612778e-01 1.80534288e-01
6.65861189e-01 -3.13633502e-01 -4.62654561e-01 -6.98233783e-01
-1.03540754e+00 9.70646143e-02 4.99175668e-01 4.99100357e-01
1.08057177e+00 -1.31389046e+00 -1.15627301e+00 3.60028911e-03
-5.72977483e-01 -3.91207457e-01 9.17208195e-01 1.07819498e+00
-4.00057971e-01 1.69737548e-01 -3.64424527e-01 -5.88250875e-01
-1.56685221e+00 4.13070291e-01 2.90798157e-01 -3.69721621e-01
-1.14908433e+00 7.79952586e-01 3.63106161e-01 3.20180058e-01
1.12907486e-02 1.06919155e-01 -4.14192796e-01 -2.45919943e-01
9.85628486e-01 7.50985503e-01 -2.14850634e-01 -8.48815620e-01
-1.66008428e-01 8.90707016e-01 -2.41562843e-01 -1.10758819e-01
1.46086884e+00 -6.77907944e-01 -7.00900555e-02 2.26257835e-03
1.12379122e+00 2.24976793e-01 -1.51524889e+00 -4.01077628e-01
-3.51419181e-01 -8.27469468e-01 3.00623775e-01 -5.47056377e-01
-1.58072448e+00 5.90687513e-01 7.63823748e-01 -1.75266430e-01
1.50740743e+00 -2.66443223e-01 1.00616431e+00 -2.67067909e-01
6.07613623e-01 -6.65981829e-01 2.13292211e-01 3.08907658e-01
1.12059867e+00 -1.06451392e+00 8.08536634e-02 -5.29194772e-01
-6.77883863e-01 9.69368160e-01 2.83128381e-01 -3.10285985e-01
4.75929648e-01 4.39256936e-01 -2.49389827e-01 2.80266434e-01
-7.65499294e-01 1.28641073e-02 2.11749643e-01 6.37695193e-01
4.95425135e-01 -2.54521668e-01 -3.12162399e-01 4.24874872e-01
-4.66412045e-02 2.44500548e-01 8.00654948e-01 4.40204561e-01
-3.47237706e-01 -8.47462356e-01 -3.69699597e-01 8.46408010e-02
-7.41125226e-01 -3.79917532e-01 4.61809099e-01 3.59953642e-01
-2.11558700e-01 9.14570630e-01 -7.02229291e-02 -4.55638081e-01
1.62547663e-01 -6.10969007e-01 4.95035887e-01 -7.39558414e-02
-4.23043817e-01 2.72084832e-01 -1.98821910e-03 -1.13300371e+00
-5.88015437e-01 -5.01024067e-01 -1.15960824e+00 -5.51343918e-01
5.93884438e-02 -1.21521465e-02 2.83045113e-01 6.05768979e-01
6.31063342e-01 8.28689873e-01 6.35624707e-01 -1.16555631e+00
-3.68557066e-01 -6.10352576e-01 -5.82199156e-01 2.02331915e-01
7.63671398e-01 -3.50453794e-01 -2.46722370e-01 2.13804871e-01]
|
[11.107913970947266, -1.9415841102600098]
|
c78f4a51-82c0-4e2c-b2d1-9d2a143a6a1c
|
e-lpips-robust-perceptual-image-similarity
|
1906.03973
| null |
https://arxiv.org/abs/1906.03973v2
|
https://arxiv.org/pdf/1906.03973v2.pdf
|
E-LPIPS: Robust Perceptual Image Similarity via Random Transformation Ensembles
|
It has been recently shown that the hidden variables of convolutional neural networks make for an efficient perceptual similarity metric that accurately predicts human judgment on relative image similarity assessment. First, we show that such learned perceptual similarity metrics (LPIPS) are susceptible to adversarial attacks that dramatically contradict human visual similarity judgment. While this is not surprising in light of neural networks' well-known weakness to adversarial perturbations, we proceed to show that self-ensembling with an infinite family of random transformations of the input --- a technique known not to render classification networks robust --- is enough to turn the metric robust against attack, while retaining predictive power on human judgments. Finally, we study the geometry imposed by our our novel self-ensembled metric (E-LPIPS) on the space of natural images. We find evidence of "perceptual convexity" by showing that convex combinations of similar-looking images retain appearance, and that discrete geodesics yield meaningful frame interpolation and texture morphing, all without explicit correspondences.
|
['Erik Härkönen', 'Markus Kettunen', 'Jaakko Lehtinen']
|
2019-06-10
| null | null | null | null |
['image-similarity-search']
|
['computer-vision']
|
[ 5.39525807e-01 3.15459639e-01 3.75970066e-01 -4.93028998e-01
-4.14278895e-01 -9.44678724e-01 8.23025227e-01 -1.06578238e-01
-4.38713670e-01 3.63761306e-01 1.95434049e-01 -2.58630961e-01
-1.90647572e-01 -6.75065279e-01 -8.11865330e-01 -6.01537108e-01
-2.35117957e-01 -1.87386394e-01 1.76934570e-01 -6.13203526e-01
5.29368401e-01 7.19150960e-01 -1.55654418e+00 1.40223771e-01
5.45391500e-01 8.79659832e-01 -4.31750566e-01 1.06812012e+00
5.88819802e-01 9.62644696e-01 -4.55857903e-01 -8.71857643e-01
8.38536322e-01 -6.22490764e-01 -9.97349381e-01 -2.23268345e-01
1.10292828e+00 -2.35425055e-01 -5.33579826e-01 1.39398825e+00
4.23816204e-01 2.77365476e-01 8.43805730e-01 -1.43546832e+00
-1.39910173e+00 9.05586481e-02 -9.29806679e-02 3.06709111e-01
5.60484946e-01 3.84010702e-01 9.01149988e-01 -5.34651399e-01
5.74135184e-01 1.24583220e+00 1.06619728e+00 6.77076817e-01
-1.36178553e+00 -3.82784337e-01 -2.59446353e-01 1.58767402e-01
-1.32618558e+00 -5.88750958e-01 8.41654003e-01 -4.21710283e-01
6.35895967e-01 7.40762711e-01 4.42899406e-01 1.06240797e+00
5.20568073e-01 -1.46036446e-02 1.35901999e+00 -3.95331085e-01
2.12903649e-01 -8.08314085e-02 -2.42957279e-01 7.82930553e-01
4.93375286e-02 4.46656257e-01 -2.61508048e-01 -5.04267775e-02
8.96289647e-01 -2.76649266e-01 -3.79081041e-01 -5.46956658e-01
-1.30153251e+00 7.41031170e-01 7.56793320e-01 2.08788559e-01
6.31067762e-03 2.02517182e-01 2.18331069e-01 7.56213546e-01
2.57106125e-01 9.83098269e-01 -5.94268069e-02 1.33163437e-01
-4.85396743e-01 1.18658490e-01 8.39698434e-01 7.60030031e-01
7.38306403e-01 1.75818980e-01 1.26995802e-01 4.86014634e-01
-1.93142444e-01 4.95862424e-01 4.84978259e-01 -1.56387138e+00
-4.98473421e-02 4.33672927e-02 8.20097476e-02 -1.69985378e+00
-3.76833886e-01 -9.81201828e-02 -9.44251657e-01 8.87365282e-01
8.58572721e-01 9.93229523e-02 -5.07146537e-01 2.27044821e+00
-1.00865766e-01 4.95742671e-02 1.80519819e-01 9.56009805e-01
4.10893738e-01 1.52960852e-01 1.04546994e-01 -5.33886515e-02
8.67611587e-01 -3.14467430e-01 -2.24000320e-01 9.53391045e-02
3.52950364e-01 -8.11267257e-01 1.31860840e+00 2.09578335e-01
-1.23125923e+00 -7.08510995e-01 -1.47712362e+00 -1.18882440e-01
-4.75251466e-01 -7.79746771e-01 5.01463532e-01 9.70734954e-01
-1.44137454e+00 1.12449229e+00 -3.23519766e-01 -5.76973855e-01
3.54123265e-01 3.48446995e-01 -6.13131523e-01 2.51599878e-01
-1.16755724e+00 1.33974385e+00 -2.40815226e-02 1.12166966e-03
-7.22559929e-01 -6.29790604e-01 -7.29756773e-01 -1.53764457e-01
-2.13075131e-01 -8.07677567e-01 8.18610549e-01 -1.51436210e+00
-1.40620160e+00 1.19392622e+00 1.58987254e-01 -5.12414515e-01
7.84753621e-01 7.82538578e-02 -6.93057895e-01 3.82831097e-01
-1.42644271e-01 9.39073086e-01 1.17082930e+00 -1.39843869e+00
-1.21543109e-01 -2.30343297e-01 4.85849917e-01 3.30498695e-01
-3.58710557e-01 1.73139811e-01 5.23777664e-01 -7.33270168e-01
7.12368563e-02 -9.92261529e-01 -2.20042497e-01 5.07441223e-01
-4.63047296e-01 2.53148317e-01 4.97342944e-01 -5.68198323e-01
5.31116962e-01 -2.26771021e+00 1.87186766e-02 4.77770895e-01
3.98426801e-01 2.07440987e-01 -4.43608999e-01 2.28096992e-01
-6.04030192e-01 4.07012045e-01 -3.64240348e-01 8.19770023e-02
2.45763376e-01 2.23541632e-01 -5.57909369e-01 1.02622092e+00
2.10236356e-01 1.12839305e+00 -8.20705593e-01 -5.22075534e-01
2.70811737e-01 4.74732399e-01 -5.52592337e-01 6.22290447e-02
1.39886558e-01 3.60051602e-01 3.53665426e-02 1.46554127e-01
6.34182334e-01 1.34456217e-01 -3.87164950e-01 -2.87808418e-01
2.01457769e-01 -1.53379008e-01 -7.73711622e-01 1.50901413e+00
-1.32300422e-01 1.04238772e+00 -2.50062078e-01 -9.28202331e-01
8.27517509e-01 5.88428155e-02 3.30718875e-01 -7.23080754e-01
3.25560063e-01 2.01983396e-02 2.13067845e-01 -4.36653316e-01
3.57280433e-01 -3.22858751e-01 8.95011276e-02 5.51664948e-01
-9.41296443e-02 -4.16848749e-01 -2.38284484e-01 1.27164081e-01
1.09956944e+00 -2.79717911e-02 5.25477082e-02 -5.46376407e-01
5.37564695e-01 -1.29600823e-01 3.08579564e-01 8.08644712e-01
-5.92994928e-01 1.04445851e+00 4.84324604e-01 -6.26346231e-01
-1.56503642e+00 -1.53938448e+00 -1.05690911e-01 8.97785008e-01
4.32790458e-01 1.03440881e-01 -1.11327207e+00 -3.95799011e-01
-1.08871855e-01 6.75262153e-01 -1.04706562e+00 -6.23482406e-01
-3.72302830e-01 -4.66065437e-01 9.92123961e-01 3.67323369e-01
5.70256352e-01 -1.18744636e+00 -5.70259035e-01 -2.64634073e-01
1.56688631e-01 -1.01003075e+00 -4.80118245e-01 -6.11553080e-02
-4.35743958e-01 -1.23301566e+00 -6.98943913e-01 -8.89693022e-01
7.33915150e-01 1.22114047e-01 1.14085853e+00 1.74642354e-01
-3.42845887e-01 7.40342319e-01 -1.71545431e-01 -1.32399216e-01
-9.08098757e-01 -5.74082434e-01 3.66835088e-01 6.63379729e-02
1.78385362e-01 -1.07272625e+00 -7.84556925e-01 5.68656147e-01
-9.66394186e-01 -3.14104378e-01 2.33761773e-01 4.75996107e-01
2.33825564e-01 -8.91195685e-02 3.06855917e-01 -4.36174303e-01
7.21832216e-01 -1.35695204e-01 -9.59099680e-02 3.20799083e-01
-3.90389740e-01 3.21989367e-03 8.59773517e-01 -6.03513002e-01
-6.03308558e-01 -2.16868713e-01 -1.31023992e-02 -4.67021912e-01
-3.68138403e-01 -2.61407614e-01 -2.13351622e-02 -8.87713790e-01
1.34477782e+00 1.22598477e-01 1.68934286e-01 2.76200753e-02
7.05834627e-01 1.77400634e-01 1.15776241e+00 -4.79502469e-01
1.22303784e+00 6.80230141e-01 3.12886596e-01 -8.76163602e-01
-5.98756015e-01 1.24442853e-01 -8.35297108e-01 -9.88352746e-02
1.11584365e+00 -4.26509410e-01 -9.64879274e-01 4.19812053e-01
-1.19224310e+00 -2.90539771e-01 -6.61394417e-01 2.46932417e-01
-1.09221196e+00 6.76796913e-01 -5.39714217e-01 -3.88183206e-01
4.61701937e-02 -8.35836291e-01 3.08390230e-01 -5.12768030e-02
-4.17301148e-01 -1.14213598e+00 4.34990264e-02 3.69223095e-02
4.52524841e-01 6.07922733e-01 9.50373292e-01 -7.33931720e-01
-2.36139506e-01 -4.23268527e-02 -3.79302740e-01 7.53396511e-01
1.30928561e-01 1.09176464e-01 -1.19067609e+00 -3.25504959e-01
3.42407078e-01 -3.27201307e-01 5.29596329e-01 1.48721665e-01
1.14829934e+00 -5.56917906e-01 3.42980683e-01 1.03524959e+00
1.31386244e+00 -7.14997053e-02 8.34679306e-01 3.45583081e-01
8.07705700e-01 6.50387406e-01 -6.78292289e-02 -6.03258163e-02
-9.15560946e-02 3.99266154e-01 6.14740908e-01 -3.33621860e-01
-2.59771109e-01 -1.67269751e-01 3.70037645e-01 4.80123013e-01
-2.64275461e-01 2.57501006e-02 -6.70000911e-01 2.41454065e-01
-1.34452713e+00 -1.32982552e+00 1.23399973e-01 2.24034715e+00
7.13589966e-01 2.89054275e-01 1.34165391e-01 2.24543661e-01
7.84121752e-01 2.79703200e-01 -5.59058070e-01 -1.00887942e+00
-4.86524612e-01 3.20508748e-01 6.94109261e-01 6.68056905e-01
-1.03447282e+00 7.41101682e-01 7.36012936e+00 5.20208299e-01
-9.04802620e-01 2.86680963e-02 6.13797784e-01 1.39687017e-01
-5.26961505e-01 -7.63307139e-02 2.02897459e-01 3.01925212e-01
7.62383997e-01 -4.52042848e-01 6.42758489e-01 5.62433124e-01
-1.38852850e-01 2.39038587e-01 -1.43141639e+00 8.98447335e-01
3.64962071e-01 -1.16963184e+00 3.15887272e-01 5.27377166e-02
7.85001874e-01 -1.52003691e-01 6.83280349e-01 -2.91789681e-01
7.76540577e-01 -1.45861912e+00 7.70739138e-01 4.86821383e-01
1.02917957e+00 -8.69945586e-01 3.49923313e-01 -3.70848253e-02
-7.28950143e-01 -1.16071552e-01 -7.45152116e-01 -9.86162499e-02
-2.56067187e-01 -3.34787369e-03 -3.98624957e-01 1.41846493e-01
4.66335416e-01 3.96959901e-01 -9.56498563e-01 8.42389524e-01
4.48798314e-02 6.42041638e-02 -1.09773278e-01 2.60517061e-01
2.16104344e-01 -1.42099991e-01 6.87717497e-01 1.02891409e+00
1.67631015e-01 2.82545954e-01 -3.05958539e-01 1.00952947e+00
8.31459165e-02 -1.91030130e-02 -1.03620696e+00 3.06120932e-01
1.08449548e-01 8.90710533e-01 -6.47204399e-01 1.10463113e-01
-1.67739168e-01 1.22235656e+00 1.48360193e-01 5.27290821e-01
-7.27907300e-01 -5.42602956e-01 1.03335619e+00 -3.36782262e-02
-1.28465697e-01 -3.36607158e-01 -4.76585686e-01 -9.89683390e-01
-5.67340031e-02 -9.69668567e-01 2.20311582e-02 -1.13146055e+00
-1.46676993e+00 7.84019172e-01 -3.06236595e-01 -1.33577764e+00
-2.38581747e-01 -5.77981949e-01 -8.13834012e-01 8.35345268e-01
-9.19192016e-01 -1.09505451e+00 -2.07110569e-01 1.05452144e+00
5.13792746e-02 -2.62574703e-01 1.01667905e+00 -2.18020111e-01
-3.25857922e-02 9.67596471e-01 -7.23849013e-02 2.91416377e-01
6.14070117e-01 -1.34709179e+00 8.35475683e-01 1.08646274e+00
5.38953722e-01 6.44623101e-01 1.08017826e+00 -1.75430939e-01
-9.57931340e-01 -7.59928644e-01 6.21256292e-01 -9.22100246e-01
8.02554429e-01 -1.22281931e-01 -8.98945570e-01 5.64928830e-01
3.27681780e-01 1.14657789e-01 5.56964755e-01 -2.24607185e-01
-9.65558052e-01 2.46653389e-02 -1.37431383e+00 1.09126854e+00
1.35789299e+00 -1.10466981e+00 -8.35607409e-01 3.39516431e-01
8.30441594e-01 -5.38488626e-02 -8.59793663e-01 2.44022489e-01
6.76532686e-01 -1.40367627e+00 1.34667170e+00 -8.84456515e-01
5.90594947e-01 -1.25162333e-01 -5.27145803e-01 -1.38425148e+00
-4.77022469e-01 -1.04775274e+00 6.46110833e-01 8.88726890e-01
1.52207926e-01 -7.54271567e-01 5.91800392e-01 8.93643498e-01
-2.65742652e-03 -3.20369750e-01 -1.04095113e+00 -1.00571227e+00
4.80854869e-01 -3.06293279e-01 3.89399916e-01 1.25048864e+00
1.16443612e-01 -1.78757280e-01 -2.44163856e-01 1.43482760e-01
7.47056663e-01 -4.11194086e-01 6.61917508e-01 -1.06071746e+00
-2.00907081e-01 -7.36349285e-01 -1.04369116e+00 -5.48458874e-01
3.21419746e-01 -8.84171903e-01 -1.78152993e-01 -8.46295059e-01
-2.29464453e-02 -3.50561768e-01 -4.10841644e-01 1.74377680e-01
1.64962694e-01 8.29244494e-01 3.10271204e-01 3.28785598e-01
-3.68243605e-01 2.85026819e-01 1.41396749e+00 -2.42538359e-02
2.39343300e-01 -2.09972873e-01 -8.39867294e-01 9.69582975e-01
8.01925182e-01 -2.60800451e-01 -5.55900931e-01 -3.91371667e-01
4.58868474e-01 -2.74008721e-01 9.80075777e-01 -1.33258832e+00
1.74380630e-01 -1.60524741e-01 5.83245993e-01 3.31894815e-01
3.46791029e-01 -7.60515273e-01 7.86414836e-04 4.16615009e-01
-8.98756981e-01 2.54121572e-01 -1.42894043e-02 4.44829732e-01
1.18158400e-01 -1.81841299e-01 1.09900451e+00 -8.52684528e-02
-6.95942342e-01 3.37984949e-01 -1.04171604e-01 4.19707686e-01
8.56579304e-01 -5.45669198e-01 -5.64135611e-01 -7.32383788e-01
-8.21467459e-01 -4.35513645e-01 1.09219241e+00 3.93249184e-01
5.65899491e-01 -1.39010608e+00 -7.40370810e-01 3.51558119e-01
2.95157898e-02 -6.29725695e-01 5.10333804e-04 3.78766209e-01
-9.01122451e-01 -1.35945782e-01 -9.23020720e-01 -3.94443363e-01
-1.05090976e+00 1.06638575e+00 6.89778388e-01 6.43364191e-01
-6.35806084e-01 1.05508602e+00 4.00203228e-01 -1.54669911e-01
1.07521705e-01 -5.85350171e-02 1.66555643e-01 -3.95673990e-01
3.41205716e-01 2.19069526e-01 -1.95140496e-01 -1.01776397e+00
-2.19948083e-01 7.89415956e-01 2.26524845e-01 -2.77731150e-01
9.75509048e-01 -2.27367744e-01 -5.58022410e-02 1.33835778e-01
1.61527216e+00 -7.76185514e-03 -1.44829023e+00 -5.68541996e-02
-3.18852782e-01 -6.58780456e-01 -4.02457058e-01 -5.89583755e-01
-8.03468585e-01 1.02399230e+00 6.89946234e-01 6.61035955e-01
9.95967388e-01 -3.01305681e-01 5.12924671e-01 5.45871258e-01
1.77448571e-01 -8.15154135e-01 1.82258800e-01 2.82864153e-01
1.18216658e+00 -1.10383773e+00 -9.19970870e-02 -2.52997994e-01
-5.08174002e-01 8.45335126e-01 4.55502450e-01 -6.59875393e-01
5.61498046e-01 -3.08373887e-02 2.60389984e-01 4.69764136e-02
-3.52511764e-01 8.34309030e-03 3.26998860e-01 1.26620245e+00
-9.73087549e-02 -1.48517817e-01 1.67987019e-01 -1.40564337e-01
-8.23285997e-01 -4.46767747e-01 6.44940794e-01 4.59499270e-01
-3.76844227e-01 -7.10951567e-01 -4.15896475e-01 3.19838934e-02
-4.00122046e-01 -1.85929075e-01 -6.75459921e-01 8.31067383e-01
1.17119268e-01 9.07779217e-01 2.49490604e-01 -6.90947115e-01
1.94021776e-01 -1.37761593e-01 8.23899806e-01 -1.12829059e-01
-5.27085721e-01 -7.20936179e-01 -2.01896310e-01 -8.41112494e-01
-3.68159145e-01 -4.24551100e-01 -7.42702127e-01 -8.30334663e-01
2.49996603e-01 -1.53487131e-01 4.27929819e-01 8.21754992e-01
-3.72737013e-02 5.12521863e-02 9.02535796e-01 -1.00547552e+00
-8.17797303e-01 -5.04568398e-01 -5.90251744e-01 1.07082796e+00
5.73115706e-01 -3.65365505e-01 -8.67389202e-01 3.26773226e-01]
|
[10.098861694335938, 2.319463014602661]
|
f66bc32f-c722-45b0-957b-34b23303a377
|
terpret-a-probabilistic-programming-language
|
1608.04428
| null |
http://arxiv.org/abs/1608.04428v1
|
http://arxiv.org/pdf/1608.04428v1.pdf
|
TerpreT: A Probabilistic Programming Language for Program Induction
|
We study machine learning formulations of inductive program synthesis; given
input-output examples, we try to synthesize source code that maps inputs to
corresponding outputs. Our aims are to develop new machine learning approaches
based on neural networks and graphical models, and to understand the
capabilities of machine learning techniques relative to traditional
alternatives, such as those based on constraint solving from the programming
languages community.
Our key contribution is the proposal of TerpreT, a domain-specific language
for expressing program synthesis problems. TerpreT is similar to a
probabilistic programming language: a model is composed of a specification of a
program representation (declarations of random variables) and an interpreter
describing how programs map inputs to outputs (a model connecting unknowns to
observations). The inference task is to observe a set of input-output examples
and infer the underlying program. TerpreT has two main benefits. First, it
enables rapid exploration of a range of domains, program representations, and
interpreter models. Second, it separates the model specification from the
inference algorithm, allowing like-to-like comparisons between different
approaches to inference. From a single TerpreT specification we automatically
perform inference using four different back-ends. These are based on gradient
descent, linear program (LP) relaxations for graphical models, discrete
satisfiability solving, and the Sketch program synthesis system.
We illustrate the value of TerpreT by developing several interpreter models
and performing an empirical comparison between alternative inference
algorithms. Our key empirical finding is that constraint solvers dominate the
gradient descent and LP-based formulations. We conclude with suggestions for
the machine learning community to make progress on program synthesis.
|
['Pushmeet Kohli', 'Nate Kushman', 'Daniel Tarlow', 'Rishabh Singh', 'Marc Brockschmidt', 'Jonathan Taylor', 'Alexander L. Gaunt']
|
2016-08-15
| null | null | null | null |
['program-induction']
|
['computer-code']
|
[ 3.71195287e-01 4.39167947e-01 -7.47951984e-01 -6.09568417e-01
-9.35564339e-01 -7.29809642e-01 7.22069502e-01 1.12675682e-01
2.15313315e-01 4.88366663e-01 -3.96775790e-02 -1.09105766e+00
-9.68020875e-03 -1.16977513e+00 -9.90648985e-01 -1.12702578e-01
-9.08845440e-02 8.05073678e-01 -1.08425722e-01 2.73282588e-01
2.66462088e-01 2.97908217e-01 -1.54111743e+00 5.47123313e-01
6.90344810e-01 5.57569325e-01 -1.54966637e-01 9.80512381e-01
-5.33143759e-01 1.04370224e+00 -3.54653627e-01 -3.36362153e-01
-1.86179966e-01 -3.55813026e-01 -8.02919447e-01 -1.90187648e-01
2.60191411e-01 -1.71383843e-01 6.32337108e-02 1.13572752e+00
-2.25101441e-01 -3.74806076e-01 7.20247090e-01 -1.72149944e+00
-6.02882266e-01 1.13838613e+00 -2.70194054e-01 -4.44784045e-01
6.47784352e-01 2.36657202e-01 1.26801729e+00 -6.48666203e-01
4.36135560e-01 1.50702155e+00 6.84610426e-01 5.41674733e-01
-2.07882810e+00 -2.50763386e-01 1.06960900e-01 -2.09782854e-01
-1.28231597e+00 -2.68897086e-01 4.95297045e-01 -8.37795496e-01
1.32740617e+00 4.97243434e-01 3.37018073e-01 8.22602451e-01
-7.35378712e-02 9.94629502e-01 1.04834509e+00 -9.55867708e-01
4.42237675e-01 7.04458356e-01 4.01700169e-01 1.25379038e+00
3.45587805e-02 3.92414808e-01 -1.47095650e-01 -8.81577790e-01
6.39383852e-01 -3.91765535e-01 2.38919538e-02 -5.24711490e-01
-9.33177054e-01 1.10585213e+00 -9.81843323e-02 6.11278054e-04
2.06494465e-01 6.12132907e-01 3.71148437e-01 2.43728489e-01
-5.53297289e-02 5.87247014e-01 -6.33608758e-01 -4.75259647e-02
-1.12467647e+00 6.07610822e-01 1.49612808e+00 1.21901572e+00
9.26132858e-01 1.51562825e-01 6.04983093e-03 4.29696083e-01
7.57008314e-01 5.61943650e-01 1.16457224e-01 -1.03771377e+00
5.27428150e-01 6.02133036e-01 -4.80133966e-02 -8.68447781e-01
2.12079454e-02 -1.98198352e-02 -2.79048294e-01 5.60250461e-01
3.40498507e-01 -3.91546190e-01 -6.47357285e-01 1.76980007e+00
-2.49498457e-01 -6.01770170e-02 5.50958663e-02 2.90162861e-01
5.87809682e-01 1.02481294e+00 2.11021174e-02 -9.05453190e-02
1.08338130e+00 -7.96116352e-01 -2.70256281e-01 -4.38450158e-01
1.17045069e+00 -3.10522765e-01 1.14391661e+00 4.69037116e-01
-1.22210169e+00 -3.79776418e-01 -1.03078949e+00 3.72445323e-02
-3.50104600e-01 3.50594521e-01 9.45135772e-01 9.87199605e-01
-1.08306301e+00 3.66251826e-01 -1.00090718e+00 1.04077868e-01
2.42268965e-01 4.42681611e-01 1.24020772e-02 -5.34945540e-03
-5.96619308e-01 7.74314404e-01 6.36472940e-01 -8.21003690e-02
-9.26317453e-01 -7.08554864e-01 -1.28844166e+00 2.26752028e-01
3.89867902e-01 -7.83381999e-01 1.63171411e+00 -1.19775414e+00
-1.59363830e+00 9.19881880e-01 -6.57012224e-01 -2.70590246e-01
2.00413659e-01 2.64506251e-01 -2.34447390e-01 -6.26507878e-01
-1.36013240e-01 3.33843857e-01 4.94178385e-01 -1.41755557e+00
-5.59661984e-01 -1.52595803e-01 2.73627967e-01 -4.11516756e-01
1.72894508e-01 2.86794782e-01 -3.41733962e-01 -1.68668017e-01
-6.42786175e-02 -8.91249239e-01 -3.17102879e-01 -1.27985746e-01
-8.13543975e-01 -2.24948689e-01 5.57204545e-01 -3.42982441e-01
1.39991856e+00 -2.05183148e+00 4.22017932e-01 6.71039581e-01
1.11787364e-01 -2.03784123e-01 -1.56905986e-02 5.06205916e-01
-3.03298652e-01 4.80873734e-01 -5.86391866e-01 -1.29473493e-01
6.54353142e-01 4.58432078e-01 -8.00592184e-01 2.03664511e-01
5.59888780e-01 1.03570771e+00 -7.15531468e-01 -5.34979880e-01
2.02814892e-01 -2.11173281e-01 -8.89397264e-01 4.60854113e-01
-1.16244066e+00 -1.31902725e-01 -4.62619722e-01 6.36099219e-01
4.55279857e-01 -1.53325289e-01 6.15011930e-01 9.77504179e-02
-2.25216657e-01 4.76678193e-01 -1.34584296e+00 1.35173941e+00
-7.60346115e-01 7.46402979e-01 -9.04563442e-02 -1.05112791e+00
9.51276362e-01 1.14851795e-01 -4.21751976e-01 5.08388095e-02
-4.39837687e-02 2.69656666e-02 -3.04100066e-01 -7.95924783e-01
2.42875442e-01 -1.56508550e-01 -5.55046678e-01 8.00143838e-01
1.22340489e-02 -7.22020805e-01 2.35725790e-01 1.57929048e-01
8.17044973e-01 5.74873328e-01 3.74559581e-01 -2.42239192e-01
3.96342933e-01 3.62883180e-01 3.40978891e-01 1.09156978e+00
6.76563978e-01 3.29422802e-01 1.41025448e+00 -5.28346598e-01
-9.43579972e-01 -1.17169774e+00 2.51582656e-02 1.13571799e+00
-4.89734501e-01 -6.50655806e-01 -6.64080918e-01 -5.74706614e-01
-2.90907416e-02 1.35789764e+00 -4.90128130e-01 6.36583269e-02
-6.83018565e-01 -6.90516770e-01 8.89812589e-01 7.62227476e-01
-9.17977542e-02 -8.97224724e-01 -5.93574226e-01 -3.77368443e-02
8.38613585e-02 -6.79220676e-01 3.80753502e-02 5.78015685e-01
-1.02539957e+00 -1.05070281e+00 1.40662983e-01 -8.58497262e-01
9.70293820e-01 -7.43115067e-01 1.46308649e+00 7.16495886e-02
-1.84862748e-01 2.42016107e-01 2.36217856e-01 -3.37560713e-01
-1.12089074e+00 -8.35564174e-03 -3.56550425e-01 -6.45792782e-01
3.99362594e-01 -5.74519455e-01 6.21511698e-01 -1.19022176e-01
-1.02039349e+00 1.75024405e-01 4.64364797e-01 9.22841311e-01
3.80222529e-01 2.43771717e-01 -2.24282563e-01 -1.47818053e+00
7.56715417e-01 -5.55733383e-01 -1.41346860e+00 6.38645530e-01
-6.00394487e-01 8.08872819e-01 7.80740976e-01 -4.29529309e-01
-1.11224902e+00 3.69637698e-01 6.82025105e-02 -3.49072218e-02
-2.98295110e-01 1.07448030e+00 -3.65616381e-01 3.25334609e-01
1.01354754e+00 2.45708272e-01 -1.40803739e-01 -4.79137637e-02
6.70827150e-01 3.88228983e-01 6.43119872e-01 -1.64194155e+00
1.01141191e+00 -1.31792888e-01 -7.96603709e-02 -4.59448069e-01
-3.11054230e-01 2.83116460e-01 -4.88332152e-01 3.26259762e-01
4.36705858e-01 -5.68275571e-01 -7.75976360e-01 -2.30359025e-02
-1.34741235e+00 -7.59424627e-01 -1.76134929e-01 1.07257508e-01
-8.61479640e-01 1.30086496e-01 -3.84333462e-01 -1.10156059e+00
2.24382669e-01 -1.56876194e+00 9.38529849e-01 5.22919893e-02
-9.23078537e-01 -1.23661768e+00 3.53315443e-01 -1.76055863e-01
1.15978144e-01 1.68804139e-01 1.85530019e+00 -6.11729324e-01
-7.40503609e-01 -2.85016805e-01 -2.02882767e-01 2.55451888e-01
-2.40739167e-01 6.42324924e-01 -9.45016801e-01 2.01277688e-01
-1.30970642e-01 -4.66223508e-01 3.19651514e-01 3.67920578e-01
1.45609140e+00 -6.15059435e-01 -5.66535950e-01 6.89792395e-01
1.58561718e+00 4.04674485e-02 6.33667886e-01 -2.74129938e-02
4.65553284e-01 5.92951357e-01 -2.97459401e-02 2.15262324e-01
2.69785285e-01 4.60546613e-01 3.15356880e-01 1.78192090e-02
4.37875539e-01 -6.57244921e-01 5.32997906e-01 2.63026327e-01
3.35127562e-01 9.31927785e-02 -1.26061988e+00 4.96649295e-01
-1.96438134e+00 -8.13273013e-01 -2.09202543e-01 2.25125480e+00
1.19371402e+00 2.62054563e-01 2.36170903e-01 -5.71413524e-02
2.88875043e-01 -8.37945193e-02 -2.39443868e-01 -9.36522186e-01
3.43313098e-01 6.27923369e-01 2.68960953e-01 9.95184600e-01
-8.63443494e-01 7.25112557e-01 7.14030457e+00 5.55546820e-01
-8.87592852e-01 -2.71456718e-01 4.92350310e-01 3.12376350e-01
-8.94299388e-01 5.56762815e-01 -8.79152298e-01 7.01149032e-02
1.28187537e+00 -4.28628713e-01 8.06362927e-01 1.31403172e+00
-1.10373341e-01 -2.06337750e-01 -2.09313512e+00 5.23853421e-01
-2.46639643e-02 -1.59396708e+00 -2.38973238e-02 -2.83351719e-01
6.25445485e-01 -3.42109412e-01 4.91231903e-02 6.85219467e-01
1.05237567e+00 -1.47877228e+00 9.38190341e-01 4.03783977e-01
6.46545529e-01 -6.54210150e-01 4.53560174e-01 4.91113693e-01
-8.85031223e-01 -1.96401894e-01 5.91024198e-02 -4.14701909e-01
-2.04882324e-01 5.92300236e-01 -9.99014378e-01 1.76393241e-01
1.01508103e-01 3.67291719e-01 -2.98700541e-01 6.48837745e-01
-6.37617111e-01 6.15508080e-01 -3.28626841e-01 -2.60875940e-01
5.42916358e-02 1.22727884e-03 2.77614325e-01 1.71671677e+00
1.00464329e-01 -1.45612523e-01 3.14960182e-01 2.05996037e+00
2.15671808e-01 -4.45732594e-01 -8.84731054e-01 -2.70447344e-01
4.68518078e-01 9.19974387e-01 -4.11189973e-01 -5.16377866e-01
-4.98476714e-01 2.64439166e-01 2.81017035e-01 6.16550803e-01
-8.37074935e-01 -4.93641227e-01 3.78565699e-01 -9.29630622e-02
1.08564764e-01 -2.47144237e-01 -7.76673138e-01 -1.09804976e+00
1.02457806e-01 -1.31286585e+00 2.19035760e-01 -7.61320651e-01
-7.33537793e-01 2.07506448e-01 6.39155388e-01 -4.60025400e-01
-8.50646019e-01 -8.21184933e-01 -8.29128146e-01 1.31426060e+00
-9.71441746e-01 -9.43908334e-01 1.85450643e-01 2.02542953e-02
2.25104108e-01 1.12555083e-02 1.28691781e+00 -2.66502202e-01
-5.49454331e-01 6.26383722e-01 -2.94924438e-01 2.27596328e-01
-8.29503834e-02 -1.44427133e+00 4.64139044e-01 8.81401539e-01
2.08267793e-01 1.07222140e+00 8.36472929e-01 -4.20442998e-01
-1.95214224e+00 -1.00883734e+00 7.47671008e-01 -7.52393782e-01
9.44385290e-01 -3.66403788e-01 -6.83375299e-01 1.32331419e+00
-1.32553965e-01 -1.20424919e-01 5.28130829e-01 4.82980698e-01
-7.90880203e-01 2.35661834e-01 -9.18355227e-01 6.87729537e-01
3.78147364e-01 -1.01251042e+00 -7.13787019e-01 2.55882770e-01
3.85346949e-01 -6.04561031e-01 -7.19603896e-01 3.74900438e-02
7.36157835e-01 -7.80021429e-01 7.60353446e-01 -8.05043101e-01
9.32317019e-01 -5.27313590e-01 -4.60134387e-01 -1.00247681e+00
-2.89214235e-02 -6.13788843e-01 -3.77737731e-01 1.13731289e+00
1.01768064e+00 -5.18998325e-01 8.52971256e-01 1.18095279e+00
2.51219682e-02 -6.18468881e-01 -3.83800030e-01 -4.06592548e-01
2.83116966e-01 -1.05406630e+00 5.54518700e-01 7.98140287e-01
4.84883815e-01 3.16002429e-01 3.16636488e-02 4.08035129e-01
4.84942943e-01 5.18761873e-01 9.73806143e-01 -1.12077200e+00
-1.19795001e+00 -7.28493392e-01 -1.12808784e-02 -9.63232577e-01
6.43135428e-01 -1.23059821e+00 2.04123408e-01 -1.13863599e+00
3.29658061e-01 -6.49328291e-01 4.23877776e-01 9.15441692e-01
2.43108213e-01 -5.64810157e-01 -1.88526064e-01 -1.93050280e-01
-1.20493248e-01 -1.31311044e-01 4.06834364e-01 -4.94460911e-01
-3.93342257e-01 3.24551612e-01 -7.57327139e-01 9.30140555e-01
6.01125598e-01 -7.10390329e-01 -4.96735096e-01 -5.86022675e-01
7.99433768e-01 5.14037132e-01 6.47137284e-01 -5.77537239e-01
2.78292269e-01 -4.60818321e-01 7.54873874e-03 -2.33821914e-01
-1.74735293e-01 -5.77589393e-01 4.50786501e-01 4.53971773e-01
-8.82777929e-01 1.10446718e-02 5.22305906e-01 1.44307047e-01
-1.43619701e-01 -8.80555749e-01 4.50627774e-01 -3.67096066e-01
-5.62154055e-01 -2.53912300e-01 -6.56055391e-01 3.40078101e-02
6.12700820e-01 6.33800924e-02 -1.38812289e-01 -2.16256052e-01
-6.70315146e-01 1.22214340e-01 5.17239928e-01 3.88089754e-02
4.48546588e-01 -9.40317273e-01 -4.95120764e-01 5.39521813e-01
1.61089033e-01 1.04506567e-01 -4.74094152e-01 4.53397632e-01
-4.56503123e-01 3.86550635e-01 2.21973658e-01 -6.41284525e-01
-1.17361975e+00 4.27688301e-01 4.04762894e-01 -3.60704303e-01
-1.52957052e-01 7.13666916e-01 1.97022408e-01 -9.33855295e-01
4.36834216e-01 -9.19277310e-01 4.45350647e-01 -6.73691154e-01
5.77364504e-01 -1.71243940e-02 -2.61349887e-01 2.16368809e-01
-2.43705362e-01 2.51705229e-01 8.16078186e-02 -3.49324346e-01
1.23390687e+00 6.56090021e-01 -6.84710741e-01 7.05869675e-01
1.06245160e+00 -3.46407481e-02 -7.32996583e-01 -2.73092449e-01
2.48703539e-01 -1.82978287e-01 4.52784784e-02 -9.12105143e-01
-5.59520364e-01 8.79577518e-01 9.58625674e-02 3.76672387e-01
8.21429670e-01 8.38277861e-02 -1.13837764e-01 7.52988696e-01
3.17503005e-01 -5.15645802e-01 -3.11710209e-01 5.48534453e-01
6.35696113e-01 -9.66058671e-01 7.60030821e-02 -3.24233174e-01
-1.55421376e-01 1.48797357e+00 3.94347638e-01 9.70918965e-03
2.91322470e-01 1.21289682e+00 -3.93162161e-01 -1.34205177e-01
-8.85162830e-01 3.17466676e-01 4.67395298e-02 7.90957093e-01
5.55243254e-01 2.58097112e-01 3.69891077e-01 6.32472754e-01
-3.61206174e-01 4.78894830e-01 5.17982543e-01 1.05049515e+00
-1.39424354e-01 -1.56497037e+00 -5.15681565e-01 4.80071634e-01
-1.20697200e-01 -3.09285551e-01 -2.29225442e-01 9.30506647e-01
-4.77798395e-02 7.02743173e-01 -9.15650204e-02 -2.87275434e-01
4.00797464e-02 3.20836842e-01 8.48201692e-01 -1.01583958e+00
-3.26992273e-01 -3.45417351e-01 4.98788089e-01 -3.69062632e-01
-4.62398902e-02 -5.03057182e-01 -1.15521550e+00 -2.55157083e-01
-2.19689399e-01 2.35185534e-01 7.50160873e-01 9.83950853e-01
-5.11689261e-02 3.35126162e-01 2.76800960e-01 -7.11169302e-01
-7.06158817e-01 -3.33378673e-01 -2.60156929e-01 -3.10835004e-01
3.30524266e-01 -1.11875594e-01 -3.48808110e-01 4.47326124e-01]
|
[8.411370277404785, 7.203490734100342]
|
32098368-1a08-4c5c-8db8-418e50181595
|
fast-vehicle-detection-and-tracking-on
|
2207.01183
| null |
https://arxiv.org/abs/2207.01183v2
|
https://arxiv.org/pdf/2207.01183v2.pdf
|
Fast Vehicle Detection and Tracking on Fisheye Traffic Monitoring Video using CNN and Bounding Box Propagation
|
We design a fast car detection and tracking algorithm for traffic monitoring fisheye video mounted on crossroads. We use ICIP 2020 VIP Cup dataset and adopt YOLOv5 as the object detection base model. The nighttime video of this dataset is very challenging, and the detection accuracy (AP50) of the base model is about 54%. We design a reliable car detection and tracking algorithm based on the concept of bounding box propagation among frames, which provides 17.9 percentage points (pp) and 6.2 pp. accuracy improvement over the base model for the nighttime and daytime videos, respectively. To speed up, the grayscale frame difference is used for the intermediate frames in a segment, which can double the processing speed.
|
['Wen-Huang Cheng', 'Hsueh-Ming Hang', 'Sandy Ardianto']
|
2022-07-04
| null | null | null | null |
['fast-vehicle-detection']
|
['computer-vision']
|
[-4.72265482e-01 -5.93243778e-01 -2.25603253e-01 9.82796866e-03
-4.94176745e-01 -4.75726426e-01 2.40621924e-01 -3.63193065e-01
-5.76959312e-01 5.37744582e-01 -5.38946509e-01 -5.55520833e-01
5.34390628e-01 -9.30053830e-01 -6.31918252e-01 -8.01225305e-01
-1.02731578e-01 -3.71216565e-01 1.51642573e+00 -1.20421261e-01
1.65636986e-01 6.24924779e-01 -1.62061703e+00 1.16434418e-01
5.98258853e-01 1.24415720e+00 3.02959502e-01 9.56839323e-01
-8.81607533e-02 8.53535354e-01 -7.50434339e-01 -4.40645039e-01
4.41655874e-01 3.70850638e-02 -5.56312734e-03 -2.23309360e-02
6.26300752e-01 -7.37134933e-01 -8.31192374e-01 1.33365560e+00
9.55213830e-02 2.02105999e-01 3.28976125e-01 -1.69055760e+00
-2.29346573e-01 -2.10026894e-02 -9.13800478e-01 8.87508333e-01
-1.38308909e-02 1.55975834e-01 2.24903405e-01 -6.91283345e-01
3.27686518e-01 1.21156514e+00 6.69419348e-01 4.28126693e-01
-4.19479817e-01 -1.23042476e+00 6.18157201e-02 9.86623764e-01
-1.76928186e+00 -4.20138717e-01 5.34651816e-01 -2.48814091e-01
3.80705059e-01 2.60280192e-01 6.48467481e-01 3.72212410e-01
5.01407683e-01 8.58096540e-01 5.79133987e-01 -2.01098323e-02
-6.56553730e-02 2.10551307e-01 2.11526051e-01 7.77088225e-01
6.42852604e-01 3.90180945e-01 -3.54832411e-03 4.12300855e-01
7.64818728e-01 2.55721271e-01 4.80082585e-03 2.32068628e-01
-7.63626575e-01 6.82128131e-01 4.73885804e-01 -1.90449089e-01
1.92242458e-01 3.03901106e-01 3.68294328e-01 7.67642707e-02
1.63078651e-01 -3.84145141e-01 -8.24832916e-02 -6.10448904e-02
-8.71811450e-01 2.09216550e-01 2.87129045e-01 1.66471899e+00
5.64313531e-01 1.71194986e-01 9.20861214e-02 3.49699408e-01
4.65840518e-01 1.08137965e+00 -2.14282423e-01 -1.24127376e+00
5.51883280e-01 2.55240411e-01 2.98520714e-01 -1.32950377e+00
-2.43599042e-01 2.49279719e-02 -3.95718694e-01 4.80613619e-01
6.89511538e-01 -4.13597524e-01 -8.14318717e-01 9.33757424e-01
4.92467970e-01 4.85404670e-01 -1.08179741e-01 9.73917365e-01
9.52575862e-01 1.13266373e+00 8.64052102e-02 -4.80222881e-01
1.46513391e+00 -1.01356554e+00 -8.69757712e-01 5.43300249e-02
4.91650075e-01 -7.01013088e-01 5.01112700e-01 3.57880503e-01
-7.86797881e-01 -6.82829380e-01 -1.20616019e+00 9.47797745e-02
-4.16238517e-01 1.86409891e-01 7.97982886e-02 8.56757402e-01
-5.57378411e-01 -6.35652319e-02 -6.16649568e-01 2.88126376e-02
2.55293131e-01 -1.04615852e-01 -6.60981983e-02 -2.02593938e-01
-1.20477951e+00 8.40795219e-01 2.05049440e-01 1.88309297e-01
-8.32391679e-01 -6.95946872e-01 -6.71914935e-01 -1.20315030e-01
6.04919910e-01 2.28906408e-01 1.01246822e+00 -3.47306252e-01
-1.24077404e+00 4.30778444e-01 -1.96999744e-01 -6.70840859e-01
6.51554286e-01 -1.43407241e-01 -9.59893286e-01 5.76976418e-01
4.03591394e-02 5.61901867e-01 7.59293497e-01 -9.46720600e-01
-1.43769050e+00 6.61343858e-02 1.81159258e-01 -1.43544853e-01
1.68434381e-01 4.75291550e-01 -8.85376811e-01 -2.81322628e-01
-2.70697773e-01 -1.03360391e+00 -6.04665503e-02 3.48315537e-01
6.50256465e-04 -1.83638528e-01 1.69745064e+00 -6.81534648e-01
1.35920572e+00 -2.36217666e+00 -9.92740273e-01 -4.64337654e-02
1.55954987e-01 5.31960070e-01 3.02980363e-01 -2.47635499e-01
4.31256622e-01 -1.74427614e-01 3.41024071e-01 1.37111932e-01
-3.23570758e-01 8.10174271e-02 -3.19524795e-01 7.22070396e-01
5.38023114e-02 3.01382035e-01 -9.43877876e-01 -9.06447172e-01
6.51034355e-01 3.39672148e-01 -2.32922360e-01 -7.78972358e-02
4.07941610e-01 -1.06339743e-02 -2.76272953e-01 9.32178557e-01
1.42162848e+00 3.76295269e-01 -5.12322366e-01 -3.11049461e-01
-7.66578853e-01 -2.73919165e-01 -1.13726115e+00 6.43566310e-01
-2.21458837e-01 1.39912581e+00 5.05067557e-02 -3.60742092e-01
9.73396301e-01 4.41997685e-02 3.31381530e-01 -8.41325879e-01
3.21850598e-01 -1.15204558e-01 -2.23955080e-01 -6.84353828e-01
7.43257523e-01 1.18186504e-01 -3.72264311e-02 -3.80988061e-01
-3.40239137e-01 2.30274677e-01 5.62834620e-01 6.19861037e-02
9.60717857e-01 -1.89499229e-01 -5.55611551e-02 -1.79867923e-01
8.19605708e-01 3.33096117e-01 7.91933656e-01 6.03437364e-01
-8.93035710e-01 4.45492566e-01 3.75808150e-01 -4.70163584e-01
-8.59931827e-01 -1.14765835e+00 -4.21244174e-01 8.50672603e-01
9.31056321e-01 -1.62149116e-01 -6.81016147e-01 -6.87728465e-01
-7.68758804e-02 5.15385747e-01 -1.98103771e-01 -8.30772892e-02
-9.41332698e-01 -4.23208952e-01 5.85798621e-01 7.82097042e-01
1.00198674e+00 -4.91108209e-01 -6.42322540e-01 5.03458008e-02
7.34303966e-02 -1.55934310e+00 -6.01214170e-01 -6.76346421e-01
-4.41821456e-01 -1.37724960e+00 -4.69566137e-01 -6.98541880e-01
5.89930654e-01 1.08929729e+00 5.66291809e-01 1.86870173e-01
-2.80334771e-01 9.40608140e-03 -3.55687112e-01 -5.63772202e-01
-2.47415677e-01 -5.73282957e-01 1.26424029e-01 -1.40581861e-01
5.63237131e-01 2.83495575e-01 -6.07609272e-01 9.29067612e-01
-3.98505002e-01 -1.19729310e-01 4.03829455e-01 2.95951396e-01
3.53510886e-01 4.68657136e-01 2.48751774e-01 -2.56053001e-01
-1.09891526e-01 -4.48390901e-01 -1.46505952e+00 -3.15145254e-02
-4.37825024e-01 -7.86817551e-01 6.08790040e-01 -3.80309552e-01
-1.07936907e+00 1.16536029e-01 -8.56456533e-02 -7.59052098e-01
-4.32332829e-02 -3.32880974e-01 -7.11511597e-02 -4.75703239e-01
2.33126774e-01 9.89876315e-02 1.61504615e-02 -2.65888065e-01
2.99701422e-01 6.74847245e-01 8.64477158e-01 2.11882479e-02
9.62471545e-01 6.62526548e-01 1.29429251e-01 -1.17755103e+00
-4.33269352e-01 -6.23827696e-01 -3.03929061e-01 -8.22135925e-01
1.08305585e+00 -1.18655431e+00 -1.28012335e+00 3.61157119e-01
-1.20808005e+00 1.23310626e-01 3.83907318e-01 8.27216685e-01
-1.14140116e-01 3.87270689e-01 -5.92899263e-01 -9.77173030e-01
-1.42683620e-02 -1.01033604e+00 6.99594975e-01 6.69062674e-01
7.46505141e-01 -6.48281455e-01 -3.50995928e-01 1.26801640e-01
1.60167277e-01 2.49117509e-01 3.73492055e-02 5.91290765e-04
-9.84527767e-01 -7.02626944e-01 -7.80831039e-01 4.93717611e-01
-3.04565877e-01 6.27719820e-01 -7.62532949e-01 5.79948444e-03
-1.02365352e-01 6.74725473e-01 9.24440920e-01 4.93360788e-01
1.17460787e+00 -7.44750053e-02 -6.14490390e-01 5.17758071e-01
1.25732112e+00 7.65784621e-01 8.40101123e-01 3.13401252e-01
6.46406651e-01 3.49380612e-01 1.32060480e+00 6.73474930e-03
4.83772576e-01 7.23190904e-01 4.38161254e-01 1.35977983e-01
-2.26873979e-01 -6.57099187e-02 6.74021482e-01 5.55510342e-01
-1.84591845e-01 -1.87480494e-01 -7.67731726e-01 4.27996010e-01
-1.55731654e+00 -1.43974280e+00 -8.00969541e-01 2.11498237e+00
2.52862245e-01 6.12973452e-01 3.34292233e-01 6.65895194e-02
1.07401586e+00 6.22097636e-04 -9.31885689e-02 -8.84972736e-02
2.69695193e-01 -6.61994040e-01 1.44594526e+00 5.09925544e-01
-1.34087408e+00 6.78624988e-01 7.11749983e+00 1.02188396e+00
-1.06623423e+00 7.17830956e-02 2.63331562e-01 -1.65099442e-01
5.29831707e-01 -1.11294769e-01 -1.68893325e+00 1.07651448e+00
1.16763830e+00 -2.73510605e-01 -5.19928262e-02 1.20199096e+00
4.32104081e-01 -3.28526884e-01 -4.08664495e-01 9.88960803e-01
1.26024202e-01 -1.36792314e+00 -2.27995485e-01 -1.32950306e-01
3.70591432e-01 -4.18166779e-02 -1.69892311e-02 5.72265327e-01
-9.76887047e-02 -4.42109436e-01 7.83618867e-01 3.04178864e-01
7.76958287e-01 -1.00853407e+00 7.47410417e-01 2.07693666e-01
-1.79609859e+00 -2.29118571e-01 -8.02211225e-01 1.74653322e-01
4.70706433e-01 2.05493242e-01 -5.95576048e-01 1.55332208e-01
8.73511374e-01 6.94683015e-01 -5.87895632e-01 1.43836546e+00
2.34310031e-02 6.69281662e-01 -5.14921963e-01 -1.19960837e-01
2.98422575e-01 -2.38424167e-01 7.37163365e-01 1.44293904e+00
3.98289442e-01 1.72519714e-01 1.74284637e-01 3.19901258e-01
9.31775123e-02 -2.68688530e-01 -5.23440182e-01 6.43353522e-01
7.63558447e-01 1.33731425e+00 -6.89847887e-01 -4.73678380e-01
-7.31071889e-01 2.24782050e-01 -5.14071167e-01 7.10575357e-02
-1.75435889e+00 -8.64035845e-01 7.43540585e-01 3.08885664e-01
6.35055542e-01 -4.55360711e-01 4.77521755e-02 -8.75655115e-01
-1.52494729e-01 -4.79823574e-02 1.67261913e-01 -7.57210732e-01
-6.54249191e-01 2.24793136e-01 2.77780801e-01 -1.89697063e+00
2.63793439e-01 -8.86559844e-01 -9.14283931e-01 1.40614703e-01
-1.62411177e+00 -8.27124536e-01 -6.81662619e-01 4.40713406e-01
7.06314504e-01 -7.32575804e-02 -1.29174128e-01 9.56394672e-01
-9.60353374e-01 7.63799667e-01 2.28702471e-01 3.48955333e-01
4.68023568e-01 -8.01654458e-01 2.55746931e-01 1.26012146e+00
-4.62800354e-01 1.39456496e-01 8.19824457e-01 -5.08008540e-01
-1.45269549e+00 -1.44056714e+00 2.88910180e-01 -5.00638843e-01
6.73506796e-01 -1.45429060e-01 -6.52496159e-01 4.40394640e-01
-5.08896522e-02 4.80925202e-01 2.02561006e-01 -7.40743458e-01
2.44328361e-02 -5.36874294e-01 -1.22277224e+00 5.04341602e-01
7.30827510e-01 -8.50016326e-02 -2.83410102e-01 1.59007028e-01
8.36545944e-01 -5.71013391e-01 -7.61676013e-01 1.34057030e-01
8.51905107e-01 -7.28186011e-01 1.03257394e+00 -2.56859008e-02
-1.08344726e-01 -1.04536557e+00 -1.30983919e-01 -4.71028656e-01
-2.50706315e-01 -5.12663245e-01 -3.06314379e-01 1.22581935e+00
1.21320866e-01 -4.09839481e-01 7.65069366e-01 4.48175907e-01
-3.92850906e-01 -4.03225392e-01 -1.08695459e+00 -1.24305820e+00
-3.92212629e-01 -5.42112052e-01 4.31701720e-01 3.84629250e-01
-2.88455456e-01 -6.68561086e-02 -5.56138813e-01 5.74676156e-01
8.43144596e-01 -1.91256523e-01 1.02500188e+00 -9.16152775e-01
4.31655616e-01 -1.57568127e-01 -8.48970950e-01 -1.31811178e+00
-2.54522294e-01 -9.10996944e-02 1.98782220e-01 -9.82725918e-01
-1.41841561e-01 -1.94231838e-01 -2.21540421e-01 -1.08428806e-01
-1.48080543e-01 5.36926150e-01 3.93479258e-01 1.26963839e-01
-7.65897632e-01 2.39350304e-01 1.09510219e+00 -8.44854787e-02
6.04374446e-02 3.18234414e-01 -1.17201664e-01 9.70457137e-01
6.57641709e-01 -5.71233392e-01 -1.48189828e-01 -1.84285060e-01
-3.22204709e-01 7.17957243e-02 3.37305158e-01 -1.24062562e+00
5.77743411e-01 -3.63724947e-01 5.54334641e-01 -1.33228838e+00
3.85126054e-01 -1.21837068e+00 -9.46840644e-02 7.70721972e-01
3.56057584e-01 1.42972589e-01 4.14793789e-01 8.75968218e-01
-1.39698416e-01 -1.10712789e-01 1.13661015e+00 1.78182647e-01
-1.42679226e+00 3.73964995e-01 -6.54096961e-01 -1.00380480e-01
1.65097427e+00 -4.74998772e-01 -9.23215330e-01 -9.43697393e-02
-2.44734854e-01 4.99031723e-01 3.26015979e-01 5.17306924e-01
7.30343759e-01 -1.58307743e+00 -5.25568724e-01 2.62487084e-01
6.34161979e-02 -3.35419357e-01 3.27045649e-01 1.03614950e+00
-1.13489711e+00 4.62413698e-01 -3.44095618e-01 -6.93040609e-01
-1.46735489e+00 7.25663662e-01 2.39839271e-01 4.90399867e-01
-7.20641077e-01 5.92461944e-01 9.19790491e-02 5.42304099e-01
5.22648655e-02 -3.84057105e-01 -5.01259327e-01 -9.66973528e-02
1.25993884e+00 1.13535988e+00 -3.75747472e-01 -8.48957419e-01
-6.22696340e-01 6.70667887e-01 -9.88014489e-02 2.86930084e-01
6.67277932e-01 -4.15895730e-01 3.95456225e-01 1.19105771e-01
1.19593227e+00 2.13843077e-01 -1.57392120e+00 1.73367530e-01
-3.45181823e-01 -9.69046235e-01 2.75970638e-01 3.47641930e-02
-1.18714964e+00 6.81204617e-01 9.24173713e-01 3.80164295e-01
8.32845747e-01 -2.20575362e-01 1.24292827e+00 3.70104730e-01
3.25238585e-01 -1.20625746e+00 -4.64876562e-01 4.27943051e-01
4.18869585e-01 -1.31121540e+00 2.73427486e-01 -7.28476346e-01
-5.52629054e-01 1.29500973e+00 9.66359258e-01 -4.72919196e-01
6.94131672e-01 4.55196798e-01 -4.84234430e-02 9.19769704e-02
-6.32429004e-01 -3.33287388e-01 1.16644554e-01 6.59429014e-01
-2.77824759e-01 -2.87542567e-02 -3.96300316e-01 2.42747545e-01
1.38977736e-01 -1.46426648e-01 7.67820954e-01 6.73033953e-01
-1.10481894e+00 -2.66626567e-01 -6.54244900e-01 2.72369057e-01
-3.94191831e-01 3.48430991e-01 3.64713043e-01 9.62533474e-01
2.32784674e-01 1.25001287e+00 4.40475374e-01 -6.96099579e-01
4.34650779e-01 -6.19133174e-01 7.37004355e-02 3.69777791e-02
2.02387780e-01 -8.66088197e-02 1.35230705e-01 -6.57965243e-01
-2.01644242e-01 -5.19631743e-01 -1.43444335e+00 -9.09980237e-01
-6.29005253e-01 1.51777133e-01 6.56164169e-01 6.79228544e-01
5.04083112e-02 2.81521171e-01 9.32031631e-01 -7.63742268e-01
-2.82140188e-02 -5.67134321e-01 -6.16024196e-01 -5.99541590e-02
5.17144203e-01 -1.11204243e+00 -5.10981977e-01 2.41342157e-01]
|
[8.050516128540039, -1.0316027402877808]
|
f640b47a-6790-4562-aee0-0175b938fcdd
|
combining-generative-and-discriminative-1
|
1708.00790
| null |
http://arxiv.org/abs/1708.00790v2
|
http://arxiv.org/pdf/1708.00790v2.pdf
|
Combining Generative and Discriminative Approaches to Unsupervised Dependency Parsing via Dual Decomposition
|
Unsupervised dependency parsing aims to learn a dependency parser from
unannotated sentences. Existing work focuses on either learning generative
models using the expectation-maximization algorithm and its variants, or
learning discriminative models using the discriminative clustering algorithm.
In this paper, we propose a new learning strategy that learns a generative
model and a discriminative model jointly based on the dual decomposition
method. Our method is simple and general, yet effective to capture the
advantages of both models and improve their learning results. We tested our
method on the UD treebank and achieved a state-of-the-art performance on thirty
languages.
|
['Wenjuan Han', 'Kewei Tu', 'Yong Jiang']
|
2017-08-02
|
combining-generative-and-discriminative-2
|
https://aclanthology.org/D17-1177
|
https://aclanthology.org/D17-1177.pdf
|
emnlp-2017-9
|
['dependency-grammar-induction', 'unsupervised-dependency-parsing']
|
['natural-language-processing', 'natural-language-processing']
|
[-2.71730989e-01 3.00801694e-01 -1.45033956e-01 -8.08746874e-01
-1.26569986e+00 -6.21878266e-01 3.83513719e-01 -2.82991499e-01
-2.74496049e-01 7.11231112e-01 2.83625335e-01 -3.59248132e-01
2.03730151e-01 -6.25776470e-01 -3.52311403e-01 -9.13413525e-01
-6.49363324e-02 8.43412995e-01 1.56043515e-01 5.83936945e-02
-1.50198057e-01 3.63015413e-01 -9.15997446e-01 1.45599797e-01
7.72803068e-01 1.99630320e-01 6.04544699e-01 7.79254854e-01
-3.18756461e-01 1.07222915e+00 -5.88238776e-01 -5.74566126e-01
-1.08156554e-01 -8.20398271e-01 -8.77670407e-01 2.90307552e-02
-1.67636275e-01 -6.48592934e-02 -2.60544151e-01 9.60926950e-01
3.93196672e-01 -1.24318376e-02 4.76632178e-01 -7.69037962e-01
-5.57232618e-01 1.18359435e+00 -3.54681313e-01 2.73873091e-01
1.58995315e-01 -5.19747257e-01 1.25441551e+00 -7.11265564e-01
5.28216600e-01 1.34326279e+00 3.66495848e-01 8.15832078e-01
-1.38337564e+00 -5.53405404e-01 3.65555465e-01 -1.38601428e-02
-1.08680069e+00 -5.83615363e-01 9.28544998e-01 -2.78757155e-01
1.20432055e+00 -2.22052097e-01 2.79083580e-01 1.02985227e+00
1.47972390e-01 1.13166499e+00 1.21856403e+00 -7.29346514e-01
2.96433181e-01 -2.19111070e-02 3.92518908e-01 7.73047745e-01
5.80869950e-02 -3.68414074e-02 -3.39275926e-01 -2.39497557e-01
7.08086431e-01 -2.35371947e-01 1.43916175e-01 -2.47386277e-01
-6.01718307e-01 1.25229442e+00 -4.98282984e-02 5.39653659e-01
-2.02565938e-01 -1.75588839e-02 1.84129536e-01 1.08305387e-01
5.87897420e-01 -2.46522687e-02 -7.43258536e-01 -2.36585990e-01
-9.99981225e-01 -5.31281531e-02 1.31082869e+00 9.91560102e-01
8.64532709e-01 -3.34350429e-02 1.52854770e-01 9.13514972e-01
6.53652668e-01 3.04067463e-01 5.31560302e-01 -8.21879804e-01
5.41881502e-01 3.16900939e-01 -3.40661705e-01 -2.04963356e-01
-4.99780685e-01 -1.77379951e-01 -5.88895977e-01 -1.15968995e-01
2.55480915e-01 -6.88105524e-01 -1.04879820e+00 1.82458961e+00
1.89626172e-01 1.22844376e-01 4.06681716e-01 5.19985735e-01
7.29886591e-01 5.74908316e-01 3.19369107e-01 -6.17752254e-01
9.70547974e-01 -9.33811247e-01 -9.11289990e-01 -4.50429887e-01
9.09291387e-01 -9.00531173e-01 5.84304631e-01 2.80890763e-01
-1.02773750e+00 -5.13897181e-01 -7.82961607e-01 -7.19229579e-02
5.35254031e-02 3.73651415e-01 1.00416827e+00 9.54713345e-01
-1.02832997e+00 5.68779171e-01 -1.58521342e+00 -1.74686790e-01
3.47120129e-02 4.41870034e-01 -4.49229360e-01 -1.68030277e-01
-8.43888283e-01 7.60869443e-01 6.07906401e-01 -1.67545572e-01
-6.52087688e-01 -5.57316728e-02 -1.12451231e+00 2.63232249e-03
7.54360482e-02 -5.34565926e-01 1.35659981e+00 -6.91923618e-01
-1.79841876e+00 7.71654963e-01 -4.73108172e-01 -3.21867585e-01
-2.00328864e-02 -4.14834201e-01 -2.19594449e-01 -6.52231723e-02
1.11658707e-01 3.32980961e-01 5.04373670e-01 -1.14545751e+00
-5.80177784e-01 -5.13845444e-01 -6.71960134e-03 -7.18317227e-03
-1.00476019e-01 2.40182534e-01 -5.91770649e-01 -7.06908584e-01
5.43166637e-01 -9.17106032e-01 -5.00286400e-01 -1.26073778e+00
-3.42127204e-01 -5.10626972e-01 7.51983523e-01 -9.77510095e-01
1.44437754e+00 -2.06732082e+00 3.87579352e-01 -1.60167485e-01
-1.46548897e-01 1.32042319e-01 -4.49863337e-02 6.57177866e-01
-9.49890390e-02 -3.66610661e-02 -4.39203769e-01 -8.64304423e-01
-1.23475596e-01 9.74619389e-01 1.97192393e-02 2.50005186e-01
2.74083287e-01 5.96536040e-01 -9.79490459e-01 -7.44035602e-01
3.33449319e-02 4.23938215e-01 -6.65490329e-01 5.45343041e-01
-4.03294563e-02 3.47487241e-01 -4.88569617e-01 7.32797325e-01
6.74413502e-01 2.46130362e-01 1.08301580e+00 1.43737406e-01
-1.83802411e-01 6.13235235e-01 -1.14532256e+00 1.93455052e+00
-5.37479162e-01 4.11634564e-01 1.16752066e-01 -1.13609695e+00
1.08607209e+00 4.34316874e-01 1.52357027e-01 -1.64718911e-01
-1.02812916e-01 1.57190815e-01 2.96168774e-03 -5.86871982e-01
6.44273497e-03 -5.16353548e-01 -4.14187610e-01 5.32756925e-01
7.58101583e-01 9.23493654e-02 3.56509686e-01 3.06224406e-01
1.17794693e+00 5.16643584e-01 4.95500386e-01 -3.95913422e-01
2.28993550e-01 -1.54965475e-01 9.32235360e-01 7.09312379e-01
9.20428336e-02 4.72010523e-01 6.96252286e-01 -2.31649384e-01
-6.41437769e-01 -1.27479315e+00 -1.12532236e-01 1.30709016e+00
-3.99309367e-01 -6.49993122e-01 -9.06887770e-01 -1.23404765e+00
-4.68552858e-01 9.50031161e-01 -3.93778503e-01 1.57849893e-01
-9.66522217e-01 -1.02564144e+00 3.86005640e-01 1.00012994e+00
3.54593843e-01 -1.04567230e+00 -2.94418454e-01 5.13414860e-01
-3.43115151e-01 -1.13928521e+00 -1.70164749e-01 8.79998446e-01
-1.03720784e+00 -8.93562078e-01 -1.87254086e-01 -1.20770943e+00
6.39060378e-01 -4.89049368e-02 1.34233356e+00 -2.52167046e-01
1.12152636e-01 1.67096108e-01 -6.88572884e-01 -1.67568490e-01
-6.95232153e-01 3.50305051e-01 -5.75178079e-02 -2.51219839e-01
7.02940404e-01 -6.66037500e-01 1.59084365e-01 -1.79025218e-01
-7.12551892e-01 -2.52238750e-01 9.22506511e-01 1.05742669e+00
4.85460371e-01 -1.11289732e-01 3.78472537e-01 -1.39959025e+00
5.30913532e-01 -4.28112388e-01 -4.86962587e-01 2.90717572e-01
-5.16091883e-01 5.63111007e-01 5.85030973e-01 -9.65361223e-02
-1.54404604e+00 7.47059643e-01 -7.16832519e-01 -1.16261937e-01
-4.39442724e-01 6.19491339e-01 -7.14159310e-01 3.74941796e-01
1.25638187e-01 9.28192139e-02 -5.59897304e-01 -1.03744984e+00
5.78279257e-01 6.79606676e-01 4.40084398e-01 -7.80275285e-01
5.06531298e-01 7.81067833e-02 -2.97373682e-01 -6.83294594e-01
-8.99818242e-01 -5.77844739e-01 -1.33138967e+00 1.52615026e-01
1.01051784e+00 -1.01316166e+00 4.89746630e-02 3.05982918e-01
-1.33126009e+00 -1.75720349e-01 8.34813714e-02 8.23654056e-01
-4.32302445e-01 6.41150832e-01 -9.68857944e-01 -7.46019423e-01
-1.71814218e-01 -9.74430859e-01 9.07505095e-01 5.97468540e-02
4.92091216e-02 -1.30856657e+00 8.33848953e-01 1.70280501e-01
-1.50324985e-01 -3.13721560e-02 8.76840353e-01 -1.00564444e+00
-2.50260592e-01 -5.02102934e-02 1.56682089e-01 6.55309200e-01
2.66074002e-01 5.19281290e-02 -9.68676150e-01 -2.83551544e-01
2.67135471e-01 -2.46527359e-01 1.09504139e+00 3.88919830e-01
4.88820642e-01 -2.01155603e-01 -3.80550832e-01 4.91548032e-01
1.43112051e+00 2.85322905e-01 5.14272273e-01 -2.60835383e-02
6.73864305e-01 5.41387081e-01 5.06887019e-01 3.13405067e-01
5.63309014e-01 4.68812644e-01 1.14088088e-01 1.67574525e-01
1.35975793e-01 -2.91555792e-01 6.27430856e-01 1.37230229e+00
-1.88827291e-01 4.48721694e-03 -9.68798339e-01 5.68907678e-01
-2.06928205e+00 -8.98870826e-01 -9.64284465e-02 1.66702461e+00
8.61645758e-01 2.39545941e-01 8.07448328e-02 -1.39501944e-01
7.42202103e-01 7.41957650e-02 1.60829388e-02 -6.55477941e-01
-1.65337510e-02 7.39763439e-01 1.83579192e-01 5.95068693e-01
-1.27896988e+00 1.45786500e+00 7.42854738e+00 4.71531391e-01
-6.69602811e-01 3.82437050e-01 4.02038872e-01 1.60802349e-01
-1.85359091e-01 4.24291670e-01 -1.07312214e+00 1.27860561e-01
1.23517323e+00 1.78962037e-01 3.49740759e-02 1.18221867e+00
-1.11287892e-01 3.93823087e-02 -1.08289087e+00 6.51456058e-01
1.16222337e-01 -1.03295827e+00 -2.21033782e-01 9.20600742e-02
5.64865410e-01 1.37596413e-01 -4.55313623e-01 5.12431860e-01
9.51838911e-01 -7.15304196e-01 3.58079791e-01 1.45620301e-01
2.52234101e-01 -9.04321790e-01 8.76829863e-01 5.85736394e-01
-1.17791009e+00 1.16513975e-01 -5.10602593e-01 -3.23679864e-01
2.95365989e-01 6.18773520e-01 -6.64438128e-01 6.86538398e-01
5.67526937e-01 6.17935836e-01 -3.39726746e-01 5.24437308e-01
-8.61271203e-01 1.10350406e+00 -2.50465333e-01 2.98177288e-03
2.76994109e-01 -3.28728706e-01 2.94607222e-01 1.74523234e+00
5.21692932e-02 1.76420227e-01 4.36471432e-01 3.67691994e-01
8.92048180e-02 5.64439446e-02 -4.41803575e-01 2.72181816e-02
4.91724163e-01 1.34217417e+00 -8.32168937e-01 -2.86391467e-01
-7.36049771e-01 8.61698210e-01 8.39027464e-01 1.47907346e-01
-6.80046260e-01 -2.44083196e-01 4.08429593e-01 -4.30120140e-01
7.67854333e-01 -7.09511638e-01 -2.18841687e-01 -1.37934637e+00
-1.05795033e-01 -4.89303499e-01 6.55983627e-01 -8.91341344e-02
-1.18609357e+00 8.37165833e-01 3.21744055e-01 -7.26791441e-01
-8.16183150e-01 -7.19944537e-01 -7.29675412e-01 8.39485765e-01
-1.31571078e+00 -1.25226700e+00 2.28785694e-01 6.33532584e-01
7.78628170e-01 -1.58568278e-01 1.24877131e+00 6.41281381e-02
-8.58605146e-01 4.65957671e-01 2.32571572e-01 5.13557374e-01
6.22851610e-01 -1.62232757e+00 3.98869663e-01 1.23076570e+00
6.73298240e-01 6.06225491e-01 5.58543265e-01 -4.59537953e-01
-1.16483080e+00 -8.60781729e-01 1.41493809e+00 -4.52972680e-01
6.04876518e-01 -4.94124979e-01 -7.68192828e-01 1.19579589e+00
4.67047453e-01 -2.17202350e-01 1.08120108e+00 5.06696641e-01
-3.43906432e-01 1.92115873e-01 -8.45770180e-01 1.17365774e-02
8.76212597e-01 -3.77126545e-01 -1.09080732e+00 1.60606623e-01
4.93158549e-01 -2.68642247e-01 -9.39840674e-01 2.40562990e-01
3.88850510e-01 -1.06071293e+00 5.68797410e-01 -6.67795897e-01
2.85553873e-01 1.12067439e-01 -4.39358175e-01 -1.41048789e+00
-6.59555197e-01 -4.60786641e-01 -2.91402698e-01 1.68000114e+00
5.34862399e-01 -4.18454260e-01 6.50130093e-01 4.03439373e-01
-3.91759247e-01 -5.07626772e-01 -1.04910541e+00 -8.04310501e-01
3.07310343e-01 -4.18307275e-01 1.20422691e-01 9.98056173e-01
1.88489154e-01 8.59552741e-01 -3.75392199e-01 3.36827457e-01
6.92933500e-01 3.44701618e-01 5.80715895e-01 -1.01103938e+00
-7.24474490e-01 3.93576287e-02 -1.37598857e-01 -1.22514129e+00
6.32293165e-01 -7.82684922e-01 2.29618236e-01 -1.50904310e+00
3.26256782e-01 -1.87283322e-01 -2.20006540e-01 6.11169100e-01
-3.23567182e-01 -2.28367746e-01 1.80963799e-01 1.59131810e-01
-4.93709326e-01 2.16064841e-01 6.77409589e-01 1.27719477e-01
-3.21326166e-01 2.37303987e-01 -6.58527672e-01 8.21183264e-01
8.90934885e-01 -9.26701367e-01 -2.66243309e-01 -6.70125127e-01
-2.22335294e-01 2.07844809e-01 -9.90713239e-02 -7.52461612e-01
1.64379820e-01 7.57297724e-02 2.52658337e-01 -7.40241706e-01
1.18166663e-01 -3.64244401e-01 -7.80228376e-02 4.23702389e-01
8.13344792e-02 1.11097142e-01 -4.35810722e-02 3.66681904e-01
-4.90396619e-01 -6.60152316e-01 7.10948169e-01 -4.60669219e-01
-7.92369962e-01 -1.54779553e-01 -4.38685507e-01 1.08211353e-01
6.88853264e-01 3.77179503e-01 8.78900886e-02 -8.85781925e-03
-1.14512289e+00 1.96568612e-02 1.44829437e-01 3.77713829e-01
4.84839022e-01 -1.07006085e+00 -8.58791471e-01 2.93544739e-01
-3.47420663e-01 -1.42656028e-01 -5.97788543e-02 4.76541460e-01
-3.18427861e-01 5.01399219e-01 7.10655153e-02 -4.50911820e-01
-1.36660874e+00 4.50091779e-01 -2.87610274e-02 -8.38340402e-01
-4.78203148e-01 8.76779318e-01 8.21402110e-03 -5.94835162e-01
7.41287768e-02 -1.46992907e-01 -3.36804807e-01 3.67459841e-02
2.53559262e-01 1.01404004e-01 -6.04834110e-02 -6.47119224e-01
-5.66841781e-01 3.42845112e-01 -1.09954379e-01 -1.52080983e-01
1.45783973e+00 6.23036586e-02 -1.97543100e-01 4.95369822e-01
1.20572412e+00 1.32812664e-01 -1.16474617e+00 -2.60696381e-01
3.78788799e-01 -1.44894868e-01 4.39130142e-02 -4.83698487e-01
-9.65229571e-01 9.00696874e-01 1.61436886e-01 3.02431583e-01
1.14830518e+00 6.37617290e-01 6.44263148e-01 4.34856653e-01
3.73551130e-01 -1.08937979e+00 -6.97510242e-02 8.12811852e-01
2.52609640e-01 -1.26780331e+00 -2.08166003e-01 -5.43827772e-01
-6.21115267e-01 1.23098290e+00 4.64426756e-01 -2.50436187e-01
9.16090786e-01 7.67631829e-01 3.46984208e-01 2.97833490e-03
-8.75684977e-01 -5.01915753e-01 -5.30549400e-02 6.88941419e-01
9.38069463e-01 3.37477863e-01 -5.33189654e-01 1.07887709e+00
-1.91192582e-01 -4.47076678e-01 2.35594168e-01 1.29475105e+00
-4.68719661e-01 -1.90146172e+00 -1.59392238e-01 -5.44685721e-02
-5.91734409e-01 -2.41816670e-01 -4.52508181e-01 9.90307987e-01
2.15486847e-02 1.12594259e+00 -5.89273423e-02 -4.03403848e-01
2.22130284e-01 4.93831694e-01 7.58172154e-01 -1.03940272e+00
-3.60135555e-01 6.66812718e-01 2.13553280e-01 -2.96327919e-01
-9.36373055e-01 -1.11724758e+00 -1.25275242e+00 1.62254393e-01
-4.59534198e-01 4.74119246e-01 6.14180207e-01 1.22525942e+00
-6.10980541e-02 3.77020836e-01 8.25816154e-01 -7.49201298e-01
-6.03447676e-01 -1.20785868e+00 -6.95145428e-01 7.82489404e-02
-1.11737534e-01 -3.44354779e-01 -2.50262320e-01 3.28632712e-01]
|
[10.347737312316895, 9.722945213317871]
|
0a53623c-ce3d-4da6-b2d6-c9b47bab4c1f
|
sparse-gaussian-process-audio-source
|
1810.12679
| null |
http://arxiv.org/abs/1810.12679v3
|
http://arxiv.org/pdf/1810.12679v3.pdf
|
Sparse Gaussian Process Audio Source Separation Using Spectrum Priors in the Time-Domain
|
Gaussian process (GP) audio source separation is a time-domain approach that
circumvents the inherent phase approximation issue of spectrogram based
methods. Furthermore, through its kernel, GPs elegantly incorporate prior
knowledge about the sources into the separation model. Despite these compelling
advantages, the computational complexity of GP inference scales cubically with
the number of audio samples. As a result, source separation GP models have been
restricted to the analysis of short audio frames. We introduce an efficient
application of GPs to time-domain audio source separation, without compromising
performance. For this purpose, we used GP regression, together with spectral
mixture kernels, and variational sparse GPs. We compared our method with
LD-PSDTF (positive semi-definite tensor factorization), KL-NMF
(Kullback-Leibler non-negative matrix factorization), and IS-NMF (Itakura-Saito
NMF). Results show that the proposed method outperforms these techniques.
|
['Dan Stowell', 'Mauricio A. Álvarez', 'Pablo A. Alvarado']
|
2018-10-30
| null | null | null | null |
['audio-source-separation']
|
['audio']
|
[ 9.02678445e-02 -3.18701297e-01 2.04615161e-01 1.60946682e-01
-1.51219833e+00 -7.13033915e-01 4.33566719e-01 -2.07959652e-01
2.49162968e-03 5.70251882e-01 2.69719988e-01 -1.28412306e-01
-5.39786816e-01 -2.79866785e-01 -3.40110421e-01 -1.13808632e+00
-2.45798379e-01 2.88137943e-01 1.14016928e-01 2.12235615e-01
4.36788872e-02 9.37786326e-02 -1.31733859e+00 -1.45668074e-01
9.47842419e-01 9.23809826e-01 1.24353170e-01 1.05416954e+00
1.40072897e-01 6.98874354e-01 -4.44772691e-01 -1.40553206e-01
5.88564314e-02 -3.20813894e-01 -2.67375410e-01 -8.63001868e-02
1.21364795e-01 2.06034333e-01 -3.10209125e-01 1.10338497e+00
5.73091686e-01 6.41694188e-01 9.33392346e-01 -1.50188982e+00
-5.65656304e-01 3.58855873e-01 -8.25859308e-01 1.64843336e-01
3.64720792e-01 -4.94837761e-01 9.42292690e-01 -1.06834865e+00
-5.87743074e-02 1.33353758e+00 1.08747947e+00 9.78345200e-02
-1.59838510e+00 -5.42921185e-01 -2.39525080e-01 -7.06160963e-02
-1.50015736e+00 -6.46219373e-01 9.71866667e-01 -6.37699008e-01
9.23710883e-01 3.73431474e-01 3.11024398e-01 9.88282800e-01
1.66244861e-02 8.69302392e-01 9.76931095e-01 -5.29493809e-01
3.77229065e-01 -1.24085754e-01 2.96189010e-01 1.92959115e-01
-6.58175722e-02 -9.15013552e-02 -8.23660374e-01 -9.55923140e-01
6.15303218e-01 -2.84480035e-01 -4.70815331e-01 -1.20797060e-01
-9.27533031e-01 7.03913331e-01 -4.29384977e-01 1.65378377e-01
-4.30199564e-01 4.43761230e-01 2.88201392e-01 1.78361878e-01
7.25624979e-01 1.96268409e-02 -3.92516375e-01 -5.50929248e-01
-1.48563933e+00 3.88870507e-01 8.95220935e-01 8.20054710e-01
5.33280373e-01 6.51293516e-01 -2.65400428e-02 9.57709491e-01
9.06238019e-01 1.10796154e+00 4.75151598e-01 -1.25586247e+00
3.84038895e-01 -3.24726760e-01 1.12052754e-01 -1.11949420e+00
-4.67441566e-02 -4.20747310e-01 -7.84588397e-01 -6.90703765e-02
4.07743186e-01 -3.58423024e-01 -6.64511561e-01 1.66688430e+00
2.85436451e-01 6.20256364e-01 2.24621087e-01 6.08396053e-01
3.75792980e-01 8.51625204e-01 -1.77822351e-01 -5.99587798e-01
1.21411777e+00 -8.16034019e-01 -1.07240844e+00 1.10784985e-01
-1.73929818e-02 -1.20301962e+00 7.49947548e-01 9.29063916e-01
-1.03926253e+00 -6.26328826e-01 -8.11280012e-01 2.27096364e-01
1.50417656e-01 2.54992843e-01 5.94535112e-01 1.12672389e+00
-1.07211602e+00 7.10344493e-01 -1.06056213e+00 7.86063224e-02
1.42154265e-02 4.38890755e-01 -2.08950907e-01 3.50420028e-01
-1.04703677e+00 2.10451365e-01 4.92101200e-02 2.23686099e-01
-8.21806729e-01 -8.75857651e-01 -6.91139698e-01 8.30621570e-02
2.17963248e-01 -4.90386814e-01 1.42475569e+00 -7.37228453e-01
-1.87632847e+00 1.98974967e-01 -7.05349743e-01 -4.89894509e-01
2.05581933e-02 -5.32299459e-01 -7.72766888e-01 5.04363239e-01
2.26372015e-02 1.78537428e-01 1.86974716e+00 -9.73441660e-01
-3.87292236e-01 -1.53608829e-01 -5.59545755e-01 4.10968959e-02
-4.58066985e-02 3.61730248e-01 -9.03127566e-02 -7.96212435e-01
3.77112389e-01 -1.08238113e+00 -1.30271897e-01 -5.43455064e-01
-2.35608473e-01 -5.85396029e-02 8.34402680e-01 -1.01489317e+00
1.20114148e+00 -2.67761660e+00 2.44863451e-01 2.89947450e-01
1.47414804e-01 3.29472646e-02 1.32663026e-01 5.97438097e-01
-2.55507708e-01 -3.58055606e-02 -2.27392361e-01 -8.17373335e-01
1.38432071e-01 1.46035269e-01 -7.67005026e-01 6.64707005e-01
1.04360558e-01 4.29365605e-01 -8.50915194e-01 -4.72385138e-01
1.81261078e-01 7.99493670e-01 -5.73241770e-01 -2.10307375e-01
1.80358157e-01 3.99865896e-01 -1.65838078e-01 4.63244617e-01
7.10533559e-01 7.13898689e-02 -8.54678676e-02 -2.75768310e-01
-1.62910044e-01 2.86028445e-01 -1.64551151e+00 1.73852801e+00
-2.45698184e-01 7.03387320e-01 4.91418540e-01 -7.14046180e-01
6.11408055e-01 1.00186241e+00 7.27900624e-01 2.39108995e-01
-9.11153704e-02 4.20036703e-01 -1.70727417e-01 -2.15891466e-01
5.36869764e-01 -3.53987038e-01 2.78305262e-01 1.95318729e-01
5.40074468e-01 -1.94335952e-01 2.89980233e-01 9.34206918e-02
5.90912759e-01 2.87669480e-01 2.18430817e-01 -4.02667075e-01
5.55395424e-01 -3.93848956e-01 6.23992980e-01 4.05261219e-01
-2.42126226e-01 8.61797750e-01 3.68336439e-01 4.34431374e-01
-5.76672435e-01 -1.49248064e+00 -9.71871614e-02 8.93916070e-01
-5.61890602e-01 -8.17210615e-01 -7.49891877e-01 1.11628491e-02
-1.11596823e-01 6.60375655e-01 -1.69185922e-01 1.43270016e-01
-4.81018096e-01 -1.01001596e+00 8.99756670e-01 3.38842779e-01
-4.37640250e-02 -2.91427404e-01 3.92997302e-02 3.59850943e-01
-4.47349638e-01 -9.98196840e-01 -5.86594999e-01 3.75698924e-01
-1.05366206e+00 -7.46480346e-01 -9.07228649e-01 -4.12787855e-01
1.22975178e-01 4.10978466e-01 5.58479071e-01 -1.15080845e+00
1.48773953e-01 8.03381681e-01 -1.74637288e-01 -3.74686867e-01
-2.54369527e-01 -4.14569169e-01 6.11738205e-01 2.83902347e-01
3.17558855e-01 -1.18393826e+00 -1.91736400e-01 4.96929884e-02
-5.92759132e-01 -4.97849345e-01 1.38627157e-01 7.05319285e-01
6.91574395e-01 7.77222753e-01 5.00355959e-01 -3.14738125e-01
1.10312080e+00 -4.32390183e-01 -4.40778434e-01 -1.10051885e-01
-3.33880931e-01 -1.24164842e-01 3.50017905e-01 -8.41693282e-01
-1.32569933e+00 -5.80389313e-02 -1.82586201e-02 -9.82643902e-01
6.26377538e-02 5.92856824e-01 -1.21950410e-01 -6.75078332e-02
6.73343897e-01 3.97029489e-01 -5.05604483e-02 -7.76924849e-01
4.83419329e-01 6.58308089e-01 6.18093550e-01 -7.88157225e-01
8.40759933e-01 5.55710495e-01 -9.87586193e-03 -1.34230506e+00
-3.91760260e-01 -8.58020961e-01 -4.85832691e-01 4.55321595e-02
8.20587099e-01 -1.20507801e+00 -5.97041905e-01 4.00166094e-01
-1.04376626e+00 6.10863939e-02 -3.48541379e-01 1.06412303e+00
-7.78079212e-01 6.88111782e-01 -8.14825594e-01 -1.61338568e+00
-3.15352261e-01 -7.68084884e-01 1.16390049e+00 5.34121469e-02
-5.32758057e-01 -1.07026243e+00 4.00260031e-01 2.41075858e-01
1.66272134e-01 9.09557939e-02 5.28927267e-01 -5.23842812e-01
-6.98941648e-02 -7.17248544e-02 1.17922820e-01 6.85728610e-01
3.14963967e-01 3.31189692e-01 -1.49982440e+00 -5.41363098e-02
8.23166192e-01 4.21747714e-01 3.50996763e-01 9.95184302e-01
4.12678301e-01 -1.64075673e-01 -1.70843288e-01 4.57224071e-01
1.15271723e+00 3.17987293e-01 5.01517951e-01 -1.88633814e-01
8.95901799e-01 3.41119379e-01 4.30722475e-01 4.38887149e-01
-2.66045090e-02 5.26877403e-01 -2.22874850e-01 3.40355486e-01
-4.73937839e-02 -2.74134785e-01 8.19329083e-01 1.37337518e+00
-3.59178573e-01 -2.04165913e-02 -7.28548884e-01 4.99729037e-01
-1.98475838e+00 -1.08659220e+00 -7.53358543e-01 2.21997023e+00
6.64072335e-01 -7.42063597e-02 3.19981039e-01 7.72481680e-01
4.96527016e-01 7.45254606e-02 9.33257043e-02 -1.97091788e-01
-1.39899671e-01 2.52671808e-01 4.53376532e-01 6.14793897e-01
-1.33203101e+00 5.30959189e-01 6.47866154e+00 1.39186680e+00
-8.86643767e-01 4.67413038e-01 -2.57225096e-01 -8.81710127e-02
-1.10529631e-01 3.48820426e-02 -5.83688736e-01 5.66423297e-01
1.46508336e+00 -2.29452521e-01 5.34390390e-01 7.11794913e-01
3.67394060e-01 -1.36706680e-01 -9.15330768e-01 1.40811527e+00
-9.47247520e-02 -6.35982871e-01 -4.70537484e-01 3.77246626e-02
4.81078893e-01 -3.29964496e-02 3.79422069e-01 2.12440416e-01
4.53743944e-03 -6.72598600e-01 8.31468999e-01 4.68249679e-01
3.02653730e-01 -8.36174905e-01 2.52614439e-01 3.30571949e-01
-1.35984468e+00 6.14498928e-02 -3.23394805e-01 -1.12715408e-01
5.60516834e-01 1.17696762e+00 -6.73752725e-01 8.63450646e-01
5.83149254e-01 6.21079564e-01 1.49685582e-02 1.04231179e+00
-2.79567212e-01 1.03041577e+00 -5.70378840e-01 6.33299649e-01
1.22990690e-01 -5.49355507e-01 1.16679168e+00 1.30858696e+00
7.17616200e-01 -1.99872553e-01 6.23588450e-02 7.80713558e-01
5.25438786e-01 3.18831913e-02 -3.37958962e-01 -4.37180936e-01
1.59648716e-01 1.10932493e+00 -6.75292075e-01 -1.10270754e-01
-3.48847330e-01 8.31986010e-01 -3.95740300e-01 7.31843352e-01
-7.53964901e-01 -2.57397145e-01 8.04875970e-01 -2.24054009e-02
3.33787173e-01 -6.17546141e-01 7.72971287e-02 -1.20982206e+00
-1.64879367e-01 -9.00558054e-01 2.62362324e-03 -8.16785455e-01
-1.40207326e+00 3.16934288e-01 1.80152550e-01 -1.38660324e+00
-5.59210658e-01 -4.53273237e-01 -3.90696883e-01 1.23752630e+00
-1.18783391e+00 -9.96348798e-01 5.21259189e-01 8.28946710e-01
4.32340205e-01 -3.71046104e-02 1.02127469e+00 6.42212987e-01
-3.64616662e-01 1.16391927e-01 5.65612972e-01 -2.79020727e-01
8.31704259e-01 -1.37598085e+00 3.11730742e-01 9.49527442e-01
5.52018642e-01 9.19935524e-01 1.07119691e+00 -6.31039202e-01
-1.33000517e+00 -8.23895335e-01 7.28168428e-01 -4.48287934e-01
1.11529696e+00 -2.82975376e-01 -8.83248687e-01 5.42548716e-01
1.46824896e-01 -2.39377588e-01 1.35681736e+00 3.12571585e-01
-3.75413269e-01 6.25005364e-02 -8.73618662e-01 4.35535043e-01
4.37260032e-01 -9.74052668e-01 -8.60114276e-01 2.41180256e-01
7.94222772e-01 -2.40424782e-01 -1.00681484e+00 -1.43349636e-02
4.01747197e-01 -7.01992810e-01 1.27037871e+00 -2.12607607e-01
-2.61771351e-01 -7.55105913e-01 -5.75972438e-01 -1.24135697e+00
-5.23622930e-01 -1.34309530e+00 -7.25062132e-01 1.57200384e+00
2.58475125e-01 -6.21390760e-01 4.15629089e-01 4.64270473e-01
-2.10489109e-01 -1.32509157e-01 -9.82001722e-01 -1.10491836e+00
-2.38006815e-01 -9.49261606e-01 9.89889055e-02 8.16894054e-01
1.95138365e-01 3.34818304e-01 -7.46144235e-01 6.13354385e-01
9.92451727e-01 -4.19173241e-02 5.67874849e-01 -1.48031902e+00
-9.34927046e-01 -1.79975212e-01 -2.47528523e-01 -1.00033307e+00
8.85269940e-02 -6.24823570e-01 9.48083252e-02 -1.03229964e+00
-2.14379236e-01 -2.63637662e-01 -2.76229024e-01 -8.94035306e-03
-1.21617928e-01 6.72222972e-02 2.49265909e-01 3.82830501e-01
-2.04373926e-01 5.92084646e-01 7.46744275e-01 8.32536817e-02
-4.74835515e-01 4.10449743e-01 -4.48770076e-01 1.09002316e+00
7.18403041e-01 -7.23720968e-01 -9.38640535e-01 -1.48248613e-01
3.31977189e-01 2.41870537e-01 1.83975458e-01 -1.18247521e+00
2.95715809e-01 9.48760062e-02 1.08253799e-01 -5.80768943e-01
8.48317504e-01 -7.58478820e-01 6.07827127e-01 -6.42899722e-02
1.21763632e-01 -3.61905545e-01 3.33316743e-01 1.03784049e+00
-6.19118690e-01 -3.29204738e-01 3.63490909e-01 3.37752402e-01
-1.38882831e-01 1.51225571e-02 -8.83796692e-01 2.75822990e-02
5.96572638e-01 -2.29432017e-01 1.52512744e-01 -5.49305737e-01
-9.80737209e-01 -3.18104565e-01 1.52841362e-03 7.83216357e-02
4.73014921e-01 -1.28346038e+00 -4.15339649e-01 1.47049978e-01
-4.02494133e-01 -1.91107377e-01 6.06690824e-01 1.40337503e+00
-8.43494311e-02 4.86353338e-01 3.16926509e-01 -6.93048358e-01
-1.42755818e+00 6.23621464e-01 -9.58508775e-02 -1.26123592e-01
-3.60099077e-01 1.03069651e+00 2.95433193e-01 -3.16095352e-01
1.19705424e-01 -5.03307283e-01 -4.12439480e-02 3.07548404e-01
6.38976336e-01 9.49685037e-01 -2.02288553e-02 -8.63543868e-01
-3.04487616e-01 6.50225699e-01 4.45000350e-01 -7.45338500e-01
1.04967940e+00 -2.80956805e-01 -3.59814465e-01 9.61885810e-01
1.30538964e+00 6.06380403e-01 -1.11444402e+00 -5.39747737e-02
3.68170775e-02 -2.89227903e-01 1.54477626e-01 -7.69318193e-02
-4.97042835e-01 1.16311324e+00 4.30301785e-01 3.48908305e-01
1.13848841e+00 -4.80222195e-01 6.50015950e-01 1.04157947e-01
3.11339617e-01 -9.20651376e-01 -2.84281343e-01 4.44031090e-01
5.85341692e-01 -7.51251996e-01 -1.94415137e-01 -8.64274502e-01
-4.94648576e-01 1.03588486e+00 -2.48735413e-01 -2.65212774e-01
1.13934731e+00 3.08070391e-01 -1.45666912e-01 1.47070900e-01
-4.71332788e-01 6.30214214e-02 5.72918594e-01 9.70214665e-01
4.26682800e-01 2.78986782e-01 1.13326944e-01 9.36247408e-01
-3.38326067e-01 -2.31946539e-02 2.71087497e-01 6.62879288e-01
-3.50510806e-01 -1.25485015e+00 -9.23604727e-01 1.46047771e-01
-8.78763318e-01 -3.85400683e-01 6.20862767e-02 2.97187954e-01
-2.58556038e-01 1.43563688e+00 -3.52809697e-01 -1.61535218e-01
7.47130662e-02 4.91575867e-01 4.89206463e-01 -4.59177673e-01
-3.92196298e-01 1.10832298e+00 1.85829148e-01 -2.06238970e-01
-6.56204402e-01 -8.33676934e-01 -1.14290833e+00 -1.05154596e-01
-4.66372371e-01 5.87327659e-01 6.23724461e-01 6.69166803e-01
1.89494818e-01 5.82549274e-01 2.59176999e-01 -1.00319886e+00
-5.50104737e-01 -9.86083269e-01 -1.18630564e+00 -1.87347576e-01
4.38307762e-01 -6.20602310e-01 -6.47356629e-01 4.36716110e-01]
|
[15.400691986083984, 5.602859020233154]
|
a8d7c610-8272-4554-bf04-531453c4641d
|
rethinking-textual-adversarial-defense-for
|
2208.10251
| null |
https://arxiv.org/abs/2208.10251v1
|
https://arxiv.org/pdf/2208.10251v1.pdf
|
Rethinking Textual Adversarial Defense for Pre-trained Language Models
|
Although pre-trained language models (PrLMs) have achieved significant success, recent studies demonstrate that PrLMs are vulnerable to adversarial attacks. By generating adversarial examples with slight perturbations on different levels (sentence / word / character), adversarial attacks can fool PrLMs to generate incorrect predictions, which questions the robustness of PrLMs. However, we find that most existing textual adversarial examples are unnatural, which can be easily distinguished by both human and machine. Based on a general anomaly detector, we propose a novel metric (Degree of Anomaly) as a constraint to enable current adversarial attack approaches to generate more natural and imperceptible adversarial examples. Under this new constraint, the success rate of existing attacks drastically decreases, which reveals that the robustness of PrLMs is not as fragile as they claimed. In addition, we find that four types of randomization can invalidate a large portion of textual adversarial examples. Based on anomaly detector and randomization, we design a universal defense framework, which is among the first to perform textual adversarial defense without knowing the specific attack. Empirical results show that our universal defense framework achieves comparable or even higher after-attack accuracy with other specific defenses, while preserving higher original accuracy at the same time. Our work discloses the essence of textual adversarial attacks, and indicates that (1) further works of adversarial attacks should focus more on how to overcome the detection and resist the randomization, otherwise their adversarial examples would be easily detected and invalidated; and (2) compared with the unnatural and perceptible adversarial examples, it is those undetectable adversarial examples that pose real risks for PrLMs and require more attention for future robustness-enhancing strategies.
|
['Hai Zhao', 'Zhuosheng Zhang', 'Rongzhou Bao', 'Jiayi Wang']
|
2022-07-21
| null | null | null | null |
['adversarial-defense']
|
['adversarial']
|
[ 3.57804179e-01 4.89300229e-02 1.17859274e-01 2.36283746e-02
-6.45194769e-01 -1.47752798e+00 7.72064388e-01 -9.27531943e-02
8.57764557e-02 5.35924077e-01 1.13286167e-01 -6.46968365e-01
3.38111341e-01 -1.16061211e+00 -7.47401476e-01 -5.38477004e-01
-1.18526414e-01 3.89933959e-02 3.81054312e-01 -7.68877149e-01
2.73304194e-01 7.15267599e-01 -8.83285940e-01 4.77855295e-01
8.91958475e-01 4.10824746e-01 -4.58824813e-01 7.72542238e-01
-3.85240465e-02 8.86136949e-01 -1.11663973e+00 -8.16061378e-01
6.24237120e-01 -3.67482960e-01 -6.46213293e-01 -3.10288429e-01
3.28683347e-01 -5.51271319e-01 -6.57821238e-01 1.55543137e+00
4.89056110e-01 -1.88567296e-01 5.87708712e-01 -1.43033814e+00
-1.34590030e+00 7.68425107e-01 -2.64815062e-01 4.39275950e-02
5.98534822e-01 6.89936340e-01 6.30087495e-01 -6.51425838e-01
1.07488930e-01 1.62790608e+00 5.20691752e-01 1.04565859e+00
-7.48628020e-01 -1.14439845e+00 3.87356997e-01 -2.15162620e-01
-1.31612301e+00 -3.02117020e-01 8.35395515e-01 -2.64373571e-01
5.13037443e-01 8.59802485e-01 2.78629586e-02 1.87423861e+00
4.89234269e-01 6.21867836e-01 1.01255715e+00 -3.29272360e-01
7.94981420e-02 3.36328298e-01 -2.15622053e-01 6.11176014e-01
3.44000369e-01 4.60124373e-01 1.45347625e-01 -7.61586845e-01
5.49431860e-01 6.08131550e-02 -3.47796261e-01 1.63619980e-01
-9.39145565e-01 1.00591743e+00 3.09257984e-01 3.43449295e-01
-6.35826122e-03 -9.24133383e-06 6.17736459e-01 5.48534095e-01
1.57671228e-01 8.94802690e-01 -5.07639647e-01 2.30571255e-01
-2.80158311e-01 2.26597205e-01 7.47580409e-01 8.20334792e-01
2.83195287e-01 6.35465324e-01 -2.39249095e-01 4.70919967e-01
1.84015602e-01 9.81856525e-01 5.03100991e-01 -3.03684801e-01
5.09370983e-01 5.37461102e-01 -6.02629259e-02 -1.41158831e+00
1.32808819e-01 -3.70279461e-01 -9.38711762e-01 4.15842026e-01
2.89834142e-01 -3.35008442e-01 -7.84285009e-01 1.89176965e+00
4.29604240e-02 1.52105972e-01 2.98174679e-01 6.14467323e-01
3.22874069e-01 6.58450484e-01 6.78307414e-02 -4.28750068e-02
1.05668950e+00 -7.22404718e-01 -6.08577788e-01 -4.89664167e-01
5.71425498e-01 -8.16924095e-01 1.50550318e+00 3.06686372e-01
-6.71925902e-01 -3.63211006e-01 -1.38756955e+00 7.61483073e-01
-6.62040114e-01 -7.20697880e-01 6.04706049e-01 1.41038287e+00
-5.57271242e-01 5.79619467e-01 -3.64938915e-01 5.03929257e-02
1.79153636e-01 2.30038956e-01 -2.98288971e-01 7.82615617e-02
-2.03268981e+00 8.50187063e-01 3.73192489e-01 -7.06844181e-02
-1.16918397e+00 -4.04692709e-01 -8.60243142e-01 -2.84290880e-01
2.83736557e-01 -3.80028427e-01 1.00779855e+00 -1.23635983e+00
-1.40412784e+00 5.30544281e-01 2.35057592e-01 -5.26470006e-01
8.20848167e-01 -1.10071927e-01 -1.16009474e+00 -1.11302547e-02
-1.25666603e-01 -1.89183466e-02 1.15488911e+00 -1.52388942e+00
-9.91692767e-02 -5.63682476e-03 4.47247952e-01 -1.38931602e-01
-8.67161870e-01 3.13229084e-01 4.13172357e-02 -1.28989804e+00
-3.83962035e-01 -8.70446563e-01 -4.54266816e-01 -2.34482348e-01
-8.22564542e-01 1.90608561e-01 1.17713404e+00 -3.53550613e-01
1.48395610e+00 -2.23589730e+00 -3.08291972e-01 3.65273595e-01
2.39933059e-01 8.61525893e-01 -3.70492995e-01 7.26487160e-01
-4.10288006e-01 8.53739798e-01 -3.52316767e-01 2.09116578e-01
7.82024190e-02 1.78673923e-01 -1.22991228e+00 3.82265896e-01
3.62995386e-01 1.00653732e+00 -1.00241673e+00 -1.07817821e-01
8.91197026e-02 1.27500653e-01 -5.42659998e-01 2.80836403e-01
-2.01450661e-01 3.50642234e-01 -7.54209101e-01 8.24231327e-01
7.61017561e-01 1.59986913e-01 -1.12199537e-01 9.09785479e-02
4.65106636e-01 -4.12017927e-02 -1.12880671e+00 7.27549016e-01
-1.15396135e-01 3.07456791e-01 -2.03510195e-01 -7.08814681e-01
9.97884035e-01 3.73954386e-01 -1.96175799e-01 -4.25684780e-01
-1.33889569e-02 1.81389704e-01 1.10724732e-01 -3.63753617e-01
4.10480261e-01 -3.73531692e-02 -6.14877999e-01 5.19209862e-01
-5.07249117e-01 2.54513975e-02 -2.77346909e-01 4.16338474e-01
1.40022624e+00 -5.08316815e-01 2.69142628e-01 -2.93585677e-02
8.99316013e-01 -2.79796004e-01 6.53668940e-01 1.29817533e+00
-3.32568616e-01 4.45346773e-01 4.04277414e-01 -5.13050795e-01
-1.04609096e+00 -1.41386569e+00 1.09321602e-01 8.67218971e-01
1.71001554e-01 -4.82758731e-01 -7.51502514e-01 -1.30218458e+00
-1.16778295e-02 8.18622172e-01 -5.71491063e-01 -7.70788372e-01
-6.72286272e-01 -7.01935053e-01 1.44624019e+00 3.92023891e-01
4.28306848e-01 -1.03100753e+00 3.12398642e-01 8.50654766e-02
3.07960864e-02 -1.10838294e+00 -6.35810971e-01 -2.89838910e-01
-3.94952506e-01 -1.00563812e+00 -2.91813582e-01 -4.90320832e-01
8.26331854e-01 6.21241294e-02 7.33343303e-01 4.37519819e-01
-2.91729663e-02 1.67806908e-01 -5.64628541e-01 -5.58634102e-01
-1.33863878e+00 -2.65684098e-01 5.22866964e-01 -3.32462229e-02
1.37527004e-01 -6.54160321e-01 -3.08385849e-01 3.50243717e-01
-1.34547222e+00 -4.92266208e-01 5.23800671e-01 5.77930748e-01
-6.78739622e-02 3.35125446e-01 9.32034612e-01 -1.08361113e+00
1.04837394e+00 -5.75452566e-01 -2.28903875e-01 2.71014631e-01
-4.04189885e-01 -4.89604957e-02 1.55944133e+00 -1.09520042e+00
-6.46886408e-01 -4.72215474e-01 -4.69772309e-01 -4.94448662e-01
-3.46056789e-01 1.42176628e-01 -5.99992156e-01 -4.59009707e-01
1.09581828e+00 4.44021404e-01 -2.31935844e-01 -1.68288857e-01
2.76762038e-01 5.43181896e-01 4.83349502e-01 -7.18198597e-01
1.86839449e+00 2.20393553e-01 -2.97827661e-01 -4.72827137e-01
-6.11137450e-01 4.29597110e-01 -1.54927120e-01 -2.96038240e-02
3.25371861e-01 -5.44609845e-01 -5.26528537e-01 5.79384148e-01
-1.19964886e+00 -7.93524757e-02 1.98359601e-02 -3.35743837e-02
-6.79606870e-02 1.05894887e+00 -8.28467190e-01 -8.40328276e-01
-4.55009252e-01 -1.12230730e+00 4.88025814e-01 -8.61454159e-02
-2.61506110e-01 -9.66349721e-01 -8.79298002e-02 1.85388699e-01
6.49380624e-01 6.27868831e-01 1.07943118e+00 -1.22440302e+00
-3.50235164e-01 -8.20520520e-01 2.28366479e-01 6.26078904e-01
3.36831242e-01 2.97731042e-01 -7.78666735e-01 -4.12568241e-01
2.12291583e-01 -2.57894874e-01 2.79827714e-01 -5.62487960e-01
1.25581408e+00 -1.07754433e+00 -6.04327582e-02 4.03199047e-01
1.17528677e+00 4.42001969e-01 9.24543262e-01 3.19320649e-01
6.94779336e-01 2.57626534e-01 4.55490112e-01 2.05188081e-01
-1.44763142e-01 3.63668293e-01 7.73970306e-01 -2.30004475e-01
2.91580647e-01 -5.70927382e-01 1.07295763e+00 5.07640481e-01
3.90923709e-01 -5.58597028e-01 -7.81818390e-01 1.53427705e-01
-1.45805073e+00 -1.22741652e+00 5.77537082e-02 2.14371276e+00
9.76539671e-01 5.76101542e-01 -1.65176675e-01 3.74947041e-01
9.00700450e-01 3.94950420e-01 -5.39600015e-01 -9.01139200e-01
-2.68762320e-01 2.61345565e-01 4.77538079e-01 5.74054539e-01
-1.24577963e+00 1.22537732e+00 6.77987242e+00 1.12480962e+00
-1.11834013e+00 -1.82094947e-01 4.93750066e-01 2.71255821e-01
-5.51298738e-01 7.27355182e-02 -6.02869511e-01 7.27716446e-01
8.31726968e-01 -4.14441347e-01 4.52878594e-01 1.00494933e+00
5.22723375e-03 8.32852006e-01 -8.89453828e-01 2.83458918e-01
6.58001676e-02 -1.04439044e+00 7.31868625e-01 -2.62512024e-02
5.60442567e-01 -5.05634189e-01 3.79516065e-01 6.07196450e-01
6.98385835e-01 -1.34781587e+00 5.75714469e-01 1.95852935e-01
5.37187457e-01 -9.54177797e-01 7.59139061e-01 6.20197237e-01
-9.96034503e-01 -1.42288983e-01 -2.75291145e-01 -1.53416499e-01
-1.14772283e-01 2.59183168e-01 -6.03086710e-01 6.54148579e-01
2.02331319e-01 1.52463272e-01 -8.00216913e-01 3.28554511e-01
-4.89302427e-01 7.14804292e-01 1.01631835e-01 -6.69082031e-02
3.59770179e-01 2.10321933e-01 9.12440240e-01 1.27673352e+00
8.81535411e-02 -1.36648566e-02 4.05014247e-01 7.91527987e-01
-1.12078138e-01 4.66402471e-02 -1.14443350e+00 -3.30153048e-01
7.74796844e-01 9.73831236e-01 -2.83883870e-01 -1.65821522e-01
-1.48136511e-01 1.13089108e+00 2.66247299e-02 3.96364003e-01
-1.16139281e+00 -6.03671789e-01 8.35492671e-01 -7.62093440e-02
-3.01886857e-01 -2.42572233e-01 -1.65217176e-01 -1.28782141e+00
1.02607660e-01 -1.64945614e+00 2.90544778e-01 -3.70581478e-01
-1.75647938e+00 8.07960749e-01 -3.31730783e-01 -1.44863904e+00
-1.10130578e-01 -5.50967038e-01 -1.02035475e+00 7.91820884e-01
-8.97283375e-01 -1.16789830e+00 2.10572630e-01 9.18301225e-01
4.02505428e-01 -5.97814798e-01 1.21530616e+00 1.32951766e-01
-6.81065738e-01 1.37080085e+00 -2.37851411e-01 8.04750204e-01
7.85464346e-01 -9.26666439e-01 8.93044353e-01 1.39640903e+00
-3.54754017e-03 1.02702069e+00 8.13176811e-01 -8.16993237e-01
-1.43648314e+00 -1.38896716e+00 4.60368097e-01 -8.58810663e-01
9.91813004e-01 -5.91906607e-01 -1.13268030e+00 7.03831613e-01
1.65382959e-02 -5.99098317e-02 7.91052639e-01 -3.29767406e-01
-8.04317355e-01 3.24400574e-01 -1.40811205e+00 1.21969497e+00
8.79387379e-01 -7.75242686e-01 -7.10429668e-01 5.92151046e-01
1.25391591e+00 -2.51079917e-01 -5.46412408e-01 7.29553580e-01
2.02428609e-01 -7.80903220e-01 1.21202803e+00 -9.83645678e-01
4.50073540e-01 -4.58103031e-01 -2.60775447e-01 -1.05784595e+00
-2.52688408e-01 -8.21519613e-01 -2.80104429e-01 1.48219657e+00
4.52290773e-01 -1.07767344e+00 3.32550347e-01 4.76550490e-01
3.28222029e-02 -6.53531253e-01 -5.91445327e-01 -1.09863091e+00
5.34815013e-01 -6.10401034e-01 7.85674870e-01 1.46074855e+00
-2.73721218e-01 -1.15182400e-01 -7.53567159e-01 8.51341367e-01
4.24376011e-01 -4.20466095e-01 8.71178091e-01 -6.23515427e-01
-3.55909258e-01 -4.03292209e-01 -4.81174976e-01 -6.62081420e-01
2.85664439e-01 -8.08267713e-01 -2.95178980e-01 -6.62479520e-01
-1.70677081e-01 -4.05732632e-01 -3.09085160e-01 5.72771311e-01
-5.00790417e-01 3.30179244e-01 1.14752226e-01 1.13546759e-01
-1.67163119e-01 4.05929834e-01 1.04236448e+00 -3.73425484e-01
1.16475381e-01 4.04952317e-02 -1.20753205e+00 9.99748111e-01
8.94255102e-01 -6.39255941e-01 -3.63806307e-01 -2.08448738e-01
8.85328352e-02 -2.43472591e-01 4.15523231e-01 -8.93352151e-01
-1.13143213e-02 -6.29136801e-01 3.42352569e-01 -7.02588186e-02
-1.45857051e-01 -7.24543452e-01 -9.03196782e-02 7.30809629e-01
-2.65219569e-01 2.36753315e-01 2.50529796e-01 6.10759079e-01
-4.02574055e-02 -2.26584613e-01 7.96766222e-01 -1.75343663e-01
-4.40900296e-01 4.58670735e-01 -6.55109584e-01 8.39186236e-02
1.24609947e+00 -6.48277104e-02 -5.73451340e-01 -4.94991362e-01
-4.44265544e-01 6.94859251e-02 6.48859680e-01 7.79107153e-01
7.29078531e-01 -1.45401645e+00 -8.47545266e-01 3.24731380e-01
-1.88816693e-02 -4.88428891e-01 1.96819246e-01 2.57479446e-03
-3.99534971e-01 -1.27682807e-02 -1.02416323e-02 -8.18734150e-03
-1.22781110e+00 1.13592744e+00 4.48346496e-01 -4.09658879e-01
-4.23437744e-01 7.19275057e-01 3.60594779e-01 -5.59992909e-01
1.46355122e-01 3.92891943e-01 -2.96858456e-02 -5.96921742e-01
7.55762279e-01 -6.82535768e-02 -2.73195475e-01 -4.73628849e-01
-4.58281845e-01 2.81040758e-01 -4.24942762e-01 2.60314107e-01
7.24170208e-01 2.96643764e-01 -1.07447930e-01 -7.34812617e-02
9.66643572e-01 7.94227540e-01 -7.05761850e-01 -1.33355722e-01
-1.87049836e-01 -6.67610109e-01 -6.85595155e-01 -8.62067699e-01
-8.41375589e-01 7.57359207e-01 3.36383402e-01 6.74439788e-01
1.07330692e+00 -3.57521474e-01 9.65123057e-01 4.35213625e-01
3.23968142e-01 -5.32920897e-01 4.96884704e-01 7.83700347e-01
1.09320891e+00 -1.03168476e+00 -2.41686642e-01 -4.49187249e-01
-8.41509819e-01 1.02396321e+00 8.73969078e-01 -3.33487093e-01
2.53602624e-01 5.22579253e-01 1.99005574e-01 2.72320360e-01
-5.27377367e-01 5.45825064e-01 2.83754528e-01 7.30266631e-01
1.02959216e-01 1.21370479e-01 -8.53933617e-02 8.67852807e-01
-4.47263300e-01 -8.32892895e-01 6.49876654e-01 8.98093045e-01
-3.02006513e-01 -1.51175547e+00 -7.58818507e-01 8.97105113e-02
-7.58479416e-01 -3.29555124e-01 -8.97305131e-01 6.97953701e-01
5.56947198e-04 1.27881420e+00 -6.33887112e-01 -9.90983486e-01
4.14499283e-01 4.80300821e-02 -9.11742225e-02 -5.93807280e-01
-8.94287825e-01 -5.24959743e-01 -3.34713012e-02 -4.50432211e-01
3.37785780e-01 -4.02627476e-02 -1.11830592e+00 -7.81061172e-01
-2.59902924e-01 2.09137380e-01 6.63512722e-02 9.18083131e-01
2.40484655e-01 4.58542436e-01 1.28910816e+00 -3.25001031e-01
-1.04453588e+00 -6.66874051e-01 -2.74587363e-01 6.63035810e-01
3.35801482e-01 -2.69788921e-01 -8.48036766e-01 -3.79817456e-01]
|
[5.949075698852539, 8.031827926635742]
|
5f035982-950e-4ce7-9981-6cf6b7e3bb0a
|
removing-word-level-spurious-alignment
|
2104.13872
| null |
https://arxiv.org/abs/2104.13872v2
|
https://arxiv.org/pdf/2104.13872v2.pdf
|
Removing Word-Level Spurious Alignment between Images and Pseudo-Captions in Unsupervised Image Captioning
|
Unsupervised image captioning is a challenging task that aims at generating captions without the supervision of image-sentence pairs, but only with images and sentences drawn from different sources and object labels detected from the images. In previous work, pseudo-captions, i.e., sentences that contain the detected object labels, were assigned to a given image. The focus of the previous work was on the alignment of input images and pseudo-captions at the sentence level. However, pseudo-captions contain many words that are irrelevant to a given image. In this work, we investigate the effect of removing mismatched words from image-sentence alignment to determine how they make this task difficult. We propose a simple gating mechanism that is trained to align image features with only the most reliable words in pseudo-captions: the detected object labels. The experimental results show that our proposed method outperforms the previous methods without introducing complex sentence-level learning objectives. Combined with the sentence-level alignment method of previous work, our method further improves its performance. These results confirm the importance of careful alignment in word-level details.
|
['Yuji Matsumoto', 'Taro Watanabe', 'Atsushi Hashimoto', 'Yoshitaka Ushiku', 'Ukyo Honda']
|
2021-04-28
| null |
https://aclanthology.org/2021.eacl-main.323
|
https://aclanthology.org/2021.eacl-main.323.pdf
|
eacl-2021-2
|
['image-sentence-alignment']
|
['natural-language-processing']
|
[ 8.56781006e-01 2.98530787e-01 1.16712525e-01 -6.12531364e-01
-1.00562000e+00 -3.91159803e-01 5.13021886e-01 1.28794909e-01
-7.03569949e-01 7.86004424e-01 7.66088590e-02 -7.39901420e-03
5.82554460e-01 -4.30114120e-01 -1.20307958e+00 -6.71308398e-01
3.93229276e-01 2.77474016e-01 5.05656123e-01 -1.01037778e-01
4.40684706e-01 5.63039305e-03 -1.40053260e+00 5.18174708e-01
6.66641176e-01 5.36807358e-01 8.64517808e-01 6.17136121e-01
-1.86638519e-01 7.02218950e-01 -6.90067589e-01 -5.54218888e-01
1.24085195e-01 -9.17162359e-01 -6.43270612e-01 6.21196866e-01
8.24055791e-01 -3.00141543e-01 3.96737233e-02 1.35565066e+00
3.54121059e-01 -2.20247254e-01 3.93811047e-01 -1.14044535e+00
-7.82534063e-01 4.72072870e-01 -6.74516559e-01 2.61069536e-01
2.25494832e-01 2.84351736e-01 9.20186341e-01 -9.46922421e-01
6.72419965e-01 1.14216399e+00 1.95854694e-01 7.33699143e-01
-1.27513862e+00 -5.81321418e-01 3.36570203e-01 1.10000528e-01
-1.22221601e+00 -6.14704728e-01 9.09412980e-01 -4.55267966e-01
6.62343979e-01 2.29761437e-01 2.27619335e-01 1.11152327e+00
2.79178321e-02 6.66372418e-01 1.10722744e+00 -7.57310033e-01
1.68524504e-01 4.16884333e-01 1.48111936e-02 6.50708020e-01
4.14974064e-01 -2.78242528e-01 -2.95318544e-01 6.75334707e-02
5.79960942e-01 -3.08315217e-01 -3.49024713e-01 -4.40186352e-01
-1.38683975e+00 7.25817800e-01 3.47086370e-01 5.31098127e-01
-4.59404171e-01 2.17638120e-01 1.95920914e-01 -1.19456634e-01
3.47215921e-01 5.06797254e-01 -3.08593035e-01 2.92609304e-01
-8.95428300e-01 -1.64449245e-01 3.63927037e-01 1.13545942e+00
8.65328729e-01 -1.23650923e-01 -5.93125761e-01 5.41272163e-01
2.28116482e-01 6.87053025e-01 4.27483618e-01 -6.37086451e-01
8.57178986e-01 4.04127538e-01 3.85227412e-01 -1.10565484e+00
1.85233168e-02 -4.07360464e-01 -4.71961796e-01 -1.92167446e-01
4.13318604e-01 1.84141304e-02 -1.08991444e+00 1.98343873e+00
1.69473067e-01 2.12240592e-01 3.16871166e-01 1.23075414e+00
8.57327521e-01 6.96974337e-01 3.39741379e-01 -3.24234962e-01
1.47311187e+00 -1.34483981e+00 -9.21735108e-01 -8.48258018e-01
3.78657073e-01 -1.10116363e+00 1.14220750e+00 -1.42930597e-01
-1.01500583e+00 -7.97065198e-01 -9.86083984e-01 1.42801672e-01
-5.62464185e-02 3.44486326e-01 9.56491679e-02 3.55067343e-01
-1.09977090e+00 7.04230443e-02 -4.56569940e-01 -4.93859529e-01
2.48914748e-01 1.72138661e-01 -3.45025033e-01 -1.25762373e-01
-1.12083805e+00 9.84981000e-01 4.37420428e-01 1.81426555e-01
-8.48068118e-01 -5.13888896e-03 -8.92560303e-01 4.35898900e-02
3.92095387e-01 -6.91073298e-01 1.37418258e+00 -1.79858470e+00
-1.17523170e+00 1.01358569e+00 -3.76235753e-01 -3.46187413e-01
3.30539227e-01 -6.40662238e-02 -8.87538418e-02 4.87993002e-01
4.05261338e-01 1.20061815e+00 1.03829396e+00 -1.66762197e+00
-4.86640006e-01 -1.38148516e-01 -9.52434540e-03 3.19085389e-01
-1.63939878e-01 7.44949207e-02 -7.77922690e-01 -6.22113287e-01
1.99661609e-02 -8.92422557e-01 -4.71184999e-01 -2.35857829e-01
-4.34104770e-01 6.13617375e-02 4.41715330e-01 -6.18937790e-01
8.11446548e-01 -2.15576243e+00 -5.47592379e-02 -2.42429659e-01
-2.35104412e-01 2.98876494e-01 -5.64344227e-01 3.61115605e-01
-1.28453001e-01 1.89143449e-01 -4.32206571e-01 -5.37031651e-01
-3.38415176e-01 3.06621224e-01 -3.56310576e-01 4.05329764e-01
7.11589158e-01 1.05796981e+00 -1.11682034e+00 -1.06400132e+00
2.27392063e-01 1.85455471e-01 -2.07899734e-01 4.98397976e-01
-4.71845627e-01 6.70482695e-01 -4.64875311e-01 1.77678987e-01
7.36648858e-01 -4.48772013e-01 5.77072091e-02 -3.48766655e-01
9.99240503e-02 1.16972372e-01 -6.50464058e-01 1.80959404e+00
-4.52860773e-01 6.07472122e-01 -9.58727375e-02 -1.01970696e+00
8.02034855e-01 2.77950138e-01 1.72655582e-01 -8.64176691e-01
1.40785053e-01 1.66652203e-01 1.71482936e-01 -9.48793828e-01
5.29006362e-01 -1.11221381e-01 -2.46641904e-01 3.73703271e-01
4.02843859e-03 -2.87865311e-01 4.56663996e-01 2.32432500e-01
7.58188605e-01 1.58575535e-01 9.75184441e-02 1.08395644e-01
6.98463261e-01 1.57513306e-01 4.87607807e-01 9.05821741e-01
-2.94820815e-01 1.22565770e+00 3.34992945e-01 -1.82450846e-01
-1.37998521e+00 -8.19841266e-01 7.73040131e-02 8.92583549e-01
4.09607083e-01 -1.95662856e-01 -9.57016647e-01 -7.39142299e-01
-4.88992214e-01 8.08951855e-01 -5.80120385e-01 5.60243540e-02
-6.15664423e-01 -5.96054733e-01 1.44718885e-01 1.94762006e-01
3.48194838e-01 -1.33828604e+00 -6.61385000e-01 2.47124538e-01
-5.57324469e-01 -1.68250692e+00 -7.66446650e-01 -8.18932056e-02
-8.34450603e-01 -9.40087318e-01 -8.75823140e-01 -1.24769092e+00
1.43139589e+00 4.19142187e-01 1.19716716e+00 1.43727437e-01
-2.93838345e-02 2.34398335e-01 -6.15860343e-01 -2.64727980e-01
-6.55794084e-01 -5.57163544e-02 -3.76716644e-01 3.86558592e-01
1.81920633e-01 -1.79667845e-01 -6.46795928e-01 4.18205261e-01
-1.24979103e+00 5.75964630e-01 1.18363070e+00 8.62196326e-01
5.87998807e-01 -3.69710773e-01 5.75488389e-01 -8.75108659e-01
4.46006626e-01 -1.32108882e-01 -3.82954270e-01 4.35690135e-01
-1.93188116e-01 2.04698488e-01 5.96729338e-01 -5.47336042e-01
-8.45940888e-01 5.67235470e-01 -1.38989463e-02 -2.93218136e-01
-2.46411756e-01 1.48314074e-01 -1.70558691e-01 2.28502408e-01
4.86494064e-01 6.30163848e-01 -4.69293073e-02 -6.95487410e-02
3.50170225e-01 8.11588824e-01 4.82511073e-01 -3.24025154e-01
7.07194507e-01 4.64466989e-01 -3.35721761e-01 -6.39048636e-01
-1.13853157e+00 -4.22356337e-01 -6.47544205e-01 -2.28584841e-01
1.11198473e+00 -9.54941154e-01 -6.86628073e-02 1.44725323e-01
-1.72706449e+00 6.53101876e-02 -1.05333887e-01 4.47495937e-01
-5.65467477e-01 5.85042119e-01 -2.78921992e-01 -8.13895106e-01
-4.09960926e-01 -1.38709748e+00 1.48141289e+00 2.75813818e-01
-7.52314329e-02 -4.73741531e-01 -1.36640534e-01 5.67497432e-01
2.51854628e-01 1.02623902e-01 5.72615802e-01 -6.25238121e-01
-7.06674695e-01 -1.67623356e-01 -4.59333003e-01 5.94320238e-01
2.56119460e-01 -3.18608761e-01 -9.55129206e-01 -2.91049033e-01
2.16734812e-01 -3.17637444e-01 7.07419395e-01 2.00187609e-01
8.81627977e-01 -3.60061735e-01 -2.71131545e-01 -2.30179410e-02
1.57119346e+00 1.74001738e-01 7.18308508e-01 2.94059217e-01
6.12716317e-01 7.83543289e-01 8.28152001e-01 -8.14897344e-02
7.44159296e-02 6.89188004e-01 6.04221463e-01 -5.38481593e-01
-2.05967948e-01 -3.55570376e-01 4.66306716e-01 7.97511518e-01
5.34945309e-01 -5.67699909e-01 -6.81110799e-01 9.09786046e-01
-2.04236078e+00 -8.90039504e-01 -4.42774296e-01 2.11321378e+00
9.43194687e-01 2.92897701e-01 -3.88192177e-01 -2.89828807e-01
1.22810543e+00 1.52178451e-01 -4.71471459e-01 -3.56345385e-01
-1.87299147e-01 -2.50811819e-02 4.96738166e-01 5.42305470e-01
-1.15288222e+00 1.06321239e+00 5.67987156e+00 5.95890939e-01
-1.06040537e+00 2.87514687e-01 8.55880857e-01 5.16209863e-02
-2.56763250e-01 1.11151814e-01 -6.35816991e-01 7.10040510e-01
7.02398598e-01 1.49363682e-01 -1.13916777e-01 6.65956140e-01
4.02315855e-01 -5.13378441e-01 -1.15420258e+00 9.03343856e-01
7.43173242e-01 -9.46373105e-01 2.34519944e-01 -2.67684221e-01
8.58029962e-01 -7.47581869e-02 5.83621599e-02 -5.97355403e-02
-1.64134040e-01 -8.21037471e-01 9.08049464e-01 4.37114060e-01
6.08585656e-01 -3.01967144e-01 1.10206878e+00 2.82314032e-01
-6.97466195e-01 3.26783150e-01 -3.77110183e-01 -6.77155182e-02
4.83544886e-01 5.38699508e-01 -1.08550596e+00 4.03074682e-01
3.66677076e-01 4.29425091e-01 -8.98398519e-01 9.40613866e-01
-6.11198962e-01 6.34891927e-01 -1.13749597e-02 -2.22046405e-01
3.47569197e-01 -5.31263053e-02 3.86431068e-01 1.18158150e+00
1.82068989e-01 -1.85997337e-01 1.65546089e-01 8.90349448e-01
8.84062052e-02 2.88245678e-01 -6.00692749e-01 -5.88489212e-02
-3.26831304e-02 1.27701867e+00 -9.09307897e-01 -5.26932061e-01
-5.14189661e-01 1.29184508e+00 1.49171278e-01 3.04611146e-01
-1.00297153e+00 -2.95985788e-01 1.06483258e-01 1.39984027e-01
6.04511619e-01 -4.40804549e-02 -1.05114557e-01 -1.10226822e+00
4.91934061e-01 -6.45605385e-01 -8.47693831e-02 -1.34294760e+00
-1.36860120e+00 9.41696584e-01 -2.09658653e-01 -1.38553679e+00
-1.58813164e-01 -3.31410199e-01 -4.92421925e-01 8.88848841e-01
-1.71413636e+00 -1.17461884e+00 -2.47049332e-01 1.59188539e-01
7.76793897e-01 3.17031175e-01 5.69014370e-01 2.37640128e-01
-4.19425488e-01 3.91001165e-01 -2.32093632e-02 2.67976105e-01
9.00475860e-01 -9.19306457e-01 4.29191738e-01 1.07141280e+00
4.00111824e-01 3.61648500e-01 1.06620550e+00 -7.09900200e-01
-9.05608773e-01 -1.19644904e+00 1.20006490e+00 -4.80107337e-01
4.22308356e-01 -7.31025994e-01 -8.34838867e-01 3.59001964e-01
7.50961423e-01 -8.38415474e-02 3.90651166e-01 -5.25047779e-01
1.31611135e-02 -1.64341897e-01 -8.49984586e-01 5.94904602e-01
7.14300454e-01 -3.36699009e-01 -8.64346504e-01 5.46092689e-01
9.45127547e-01 -3.32396507e-01 -6.24825433e-02 3.12587798e-01
1.61247402e-01 -7.15505540e-01 7.81127334e-01 -4.41994876e-01
7.43112147e-01 -7.05089509e-01 1.45678371e-01 -1.01486802e+00
-1.08114481e-02 -1.73193961e-01 5.72694957e-01 1.33576035e+00
6.19044423e-01 -1.44536287e-01 6.66261673e-01 3.44388574e-01
-1.13246188e-01 -3.45086515e-01 -8.44320357e-01 -7.09816873e-01
-3.33102077e-01 -1.29090354e-01 2.19944924e-01 6.53895497e-01
-3.10190141e-01 8.34382832e-01 -4.61180687e-01 3.49675387e-01
5.27006269e-01 3.75566006e-01 7.34438062e-01 -6.27441823e-01
-6.78855702e-02 -1.22820958e-01 -3.91892672e-01 -1.00644338e+00
2.12818548e-01 -7.46194422e-01 7.33488619e-01 -1.64370155e+00
5.29961765e-01 -9.23203006e-02 -1.96219236e-01 4.34251785e-01
-5.78796208e-01 5.77807248e-01 2.56275982e-01 2.76319325e-01
-9.52995300e-01 6.18318617e-01 1.50811934e+00 -3.87096256e-01
1.07421361e-01 -2.96350569e-01 -6.06471539e-01 4.81488556e-01
8.31639707e-01 -8.53091180e-01 -3.61443311e-01 -7.43516445e-01
-7.56890932e-03 -3.19936201e-02 4.53873485e-01 -8.87533844e-01
2.03054905e-01 -9.30804983e-02 3.21263850e-01 -5.05085647e-01
1.98613957e-01 -8.30063283e-01 -6.93138614e-02 4.93051380e-01
-4.47076172e-01 2.04600364e-01 2.09647611e-01 6.82341576e-01
-4.54691887e-01 -6.43568695e-01 6.95772350e-01 -2.14406267e-01
-6.73477471e-01 -1.27965912e-01 -3.31015557e-01 -6.61896989e-02
1.28381526e+00 -2.28498697e-01 -1.89228311e-01 -4.47280556e-01
-6.44969821e-01 2.54134655e-01 6.52148366e-01 6.00567520e-01
7.52673805e-01 -1.12029195e+00 -8.98521781e-01 2.40731705e-02
5.08239806e-01 1.93342611e-01 1.31452993e-01 6.37129247e-01
-4.33082908e-01 4.90555763e-01 -1.44283697e-01 -8.20078969e-01
-1.28889906e+00 8.58290792e-01 7.72474259e-02 -1.41356483e-01
-2.23491192e-01 7.27401376e-01 5.42498231e-01 -8.60526785e-02
1.62819609e-01 -3.44193935e-01 -1.81353465e-01 -1.40928164e-01
3.60626876e-01 -6.77062452e-01 -7.57635832e-02 -9.32916105e-01
-2.90947855e-01 6.24591887e-01 -2.93393940e-01 -2.37746581e-01
1.18960690e+00 -3.61109704e-01 -1.34490579e-01 4.22673702e-01
1.09788907e+00 -1.60262614e-01 -1.23523223e+00 -1.61982402e-01
-3.70904543e-02 -5.05510032e-01 -2.21263900e-01 -7.65498817e-01
-9.68982279e-01 8.24482918e-01 6.80811226e-01 -1.03020273e-01
1.10883546e+00 2.96205729e-01 7.58090675e-01 4.01508473e-02
2.61697024e-01 -9.98761117e-01 4.58518088e-01 1.52182177e-01
9.05880749e-01 -1.64211106e+00 -1.57828212e-01 -5.40081620e-01
-8.13659489e-01 7.07891762e-01 9.58536565e-01 1.31356195e-02
-7.68213496e-02 -2.23611265e-01 2.43986532e-01 -2.21105553e-02
-5.96562266e-01 -5.96811593e-01 1.35597512e-01 3.88581276e-01
3.16483796e-01 -4.12473977e-01 -6.43800080e-01 3.60928953e-01
1.95558578e-01 -4.69161198e-02 7.26161659e-01 1.04389656e+00
-4.05655086e-01 -1.20961797e+00 -5.08298814e-01 2.23531961e-01
-3.27758819e-01 -4.44011003e-01 -5.39817750e-01 4.23432559e-01
1.60557583e-01 1.05102336e+00 2.61660069e-01 -1.09469853e-01
2.41175190e-01 -2.33556349e-02 4.20756251e-01 -9.40634847e-01
-4.07411277e-01 1.26799410e-02 -5.19791804e-02 -1.20221816e-01
-8.48107994e-01 -4.49655712e-01 -1.06054401e+00 5.81301391e-01
-6.01047814e-01 5.25193393e-01 8.42623055e-01 9.15570676e-01
4.29163873e-01 4.39163685e-01 6.72162235e-01 -7.04406857e-01
-2.59981722e-01 -1.09320700e+00 4.97582601e-03 8.71533275e-01
3.16827327e-01 -2.70303816e-01 -4.54392374e-01 5.90968013e-01]
|
[10.929567337036133, 0.9922721982002258]
|
6a67fddc-f933-4f9c-99d4-e564399808f4
|
global-relational-models-of-source-code
| null | null |
https://openreview.net/forum?id=B1lnbRNtwr
|
https://openreview.net/pdf?id=B1lnbRNtwr
|
Global Relational Models of Source Code
|
Models of code can learn distributed representations of a program's syntax and semantics to predict many non-trivial properties of a program. Recent state-of-the-art models leverage highly structured representations of programs, such as trees, graphs and paths therein (e.g. data-flow relations), which are precise and abundantly available for code. This provides a strong inductive bias towards semantically meaningful relations, yielding more generalizable representations than classical sequence-based models. Unfortunately, these models primarily rely on graph-based message passing to represent relations in code, which makes them de facto local due to the high cost of message-passing steps, quite in contrast to modern, global sequence-based models, such as the Transformer. In this work, we bridge this divide between global and structured models by introducing two new hybrid model families that are both global and incorporate structural bias: Graph Sandwiches, which wrap traditional (gated) graph message-passing layers in sequential message-passing layers; and Graph Relational Embedding Attention Transformers (GREAT for short), which bias traditional Transformers with relational information from graph edge types. By studying a popular, non-trivial program repair task, variable-misuse identification, we explore the relative merits of traditional and hybrid model families for code representation. Starting with a graph-based model that already improves upon the prior state-of-the-art for this task by 20%, we show that our proposed hybrid models improve an additional 10-15%, while training both faster and using fewer parameters.
|
['David Bieber', 'Petros Maniatis', 'Rishabh Singh', 'Charles Sutton', 'Vincent J. Hellendoorn']
|
2020-05-01
| null | null | null |
iclr-2020-1
|
['program-repair', 'variable-misuse', 'program-repair']
|
['computer-code', 'computer-code', 'reasoning']
|
[ 1.22107066e-01 3.73117954e-01 -8.09057057e-01 -1.61529839e-01
-5.94908535e-01 -4.79564965e-01 5.14777780e-01 9.49428320e-01
1.53045505e-01 4.26631719e-02 4.12173778e-01 -1.02380121e+00
1.77959949e-02 -1.22617936e+00 -1.07564855e+00 -1.46560654e-01
-5.61668694e-01 -1.58708412e-02 3.69059980e-01 -4.06588286e-01
3.28525841e-01 2.00927034e-01 -1.47200370e+00 4.63394076e-01
6.56979263e-01 7.45074570e-01 -9.27118957e-02 8.28047335e-01
-3.51381242e-01 1.65391934e+00 -2.96324193e-01 -6.04155600e-01
-2.33594865e-01 -2.79709846e-01 -1.09040499e+00 -3.09628457e-01
3.49643975e-01 -2.35319853e-01 -7.56533027e-01 1.03299379e+00
-9.40460488e-02 -1.62336335e-03 3.76244128e-01 -1.16137016e+00
-1.09666622e+00 1.20733762e+00 -7.10273683e-01 2.54232287e-01
5.68458796e-01 1.55752569e-01 1.65072727e+00 -5.84642351e-01
6.38064861e-01 1.20206368e+00 1.21387076e+00 3.56043935e-01
-1.57805562e+00 -1.68859094e-01 3.22799176e-01 2.63766021e-01
-1.12652826e+00 -2.37133160e-01 6.05852485e-01 -6.58163369e-01
1.67228258e+00 2.40483239e-01 2.74410248e-01 9.36305344e-01
3.91546339e-01 6.66649520e-01 5.70381641e-01 -2.76417434e-01
-8.81213546e-02 -2.02333286e-01 8.21666777e-01 1.30280447e+00
4.77302283e-01 2.65623983e-02 -1.76660657e-01 -5.81746817e-01
4.40454900e-01 2.90149361e-01 -2.97647893e-01 -7.00569630e-01
-1.00497901e+00 9.91225421e-01 1.01644993e+00 4.08176482e-01
-6.50363639e-02 7.59729803e-01 8.40265930e-01 6.10938132e-01
4.33299214e-01 4.11183327e-01 -4.30914074e-01 -2.08055571e-01
-6.63367748e-01 3.16084117e-01 9.83288169e-01 1.08172989e+00
1.15380943e+00 2.42582545e-01 -2.12874055e-01 6.39056444e-01
5.11500180e-01 1.77123249e-01 5.34545541e-01 -2.43942454e-01
8.59282970e-01 1.23212767e+00 -6.53109252e-01 -1.28977275e+00
-4.04532790e-01 -4.79005307e-01 -5.94755471e-01 -1.17570907e-01
1.75810605e-01 3.70941609e-01 -8.65232468e-01 1.66922247e+00
2.51938943e-02 1.57281026e-01 -2.53068566e-01 3.35133344e-01
9.30139184e-01 5.53557277e-01 6.94385618e-02 2.73541540e-01
1.18068850e+00 -1.04819179e+00 -2.70492762e-01 -4.37769204e-01
1.47345424e+00 -1.67814568e-01 1.10834134e+00 -6.03667535e-02
-8.38040650e-01 -3.62940252e-01 -1.07126200e+00 -4.14829642e-01
-5.35693944e-01 -3.84377211e-01 1.13042533e+00 7.76769638e-01
-1.41963911e+00 7.01617301e-01 -9.64226961e-01 -2.64079690e-01
6.04420543e-01 3.38986605e-01 -4.65811074e-01 -2.86125600e-01
-7.59568989e-01 7.28975534e-01 1.72071636e-01 -2.72264868e-01
-1.16694701e+00 -9.43227351e-01 -1.49282205e+00 3.87210101e-01
5.82391858e-01 -6.29997909e-01 1.25086427e+00 -7.53950655e-01
-1.08683240e+00 9.49413776e-01 -2.92366296e-01 -7.30847538e-01
-1.47978052e-01 -1.22689955e-01 -1.60987854e-01 -3.15957099e-01
-2.52971351e-01 -3.15254629e-01 6.43564939e-01 -1.10124266e+00
-1.28577024e-01 -3.14501524e-01 6.58722103e-01 -4.56365436e-01
-3.06966811e-01 3.07466626e-01 -1.13395862e-01 -5.85259199e-01
-1.51362211e-01 -7.21900165e-01 -3.60350102e-01 -2.06379905e-01
-5.23598969e-01 -4.70755517e-01 6.92980468e-01 -6.75019741e-01
1.76997817e+00 -2.11342597e+00 4.00741667e-01 1.18935689e-01
9.59914625e-01 9.30202082e-02 -2.14549258e-01 7.95766652e-01
-4.19090360e-01 5.17339885e-01 -6.11532927e-01 -2.31817290e-01
1.47579506e-01 3.36489439e-01 -5.17765462e-01 6.55927539e-01
3.58486831e-01 1.44367254e+00 -1.01729000e+00 -1.55706033e-01
-7.76744559e-02 2.08093360e-01 -9.51670706e-01 4.12204921e-01
-6.45372808e-01 -3.49564970e-01 -3.60687613e-01 5.59882104e-01
2.72771537e-01 -6.78213716e-01 3.62488985e-01 1.35399193e-01
2.13394910e-01 9.40320611e-01 -6.76615536e-01 1.81106782e+00
-8.64043891e-01 5.38892329e-01 -1.26468226e-01 -1.35362601e+00
7.43855476e-01 1.29293501e-01 1.35199398e-01 -5.18323660e-01
-6.98951334e-02 4.74461839e-02 -2.10367367e-02 -5.53031206e-01
6.43825591e-01 2.21612588e-01 -2.61366725e-01 7.11429358e-01
2.22276822e-01 5.20695485e-02 1.71099436e-02 6.92123652e-01
1.75455678e+00 1.82947114e-01 5.64419031e-01 -3.37061137e-01
5.34899116e-01 -3.31655145e-01 3.20708096e-01 8.30689549e-01
2.79884726e-01 3.68353635e-01 1.28778148e+00 -3.96707028e-01
-7.74103105e-01 -7.08540499e-01 3.37533921e-01 1.42368305e+00
-1.45595044e-01 -1.33165991e+00 -5.42471707e-01 -1.13959694e+00
2.63245612e-01 4.76941288e-01 -8.57683539e-01 -5.68170667e-01
-8.57460916e-01 -5.39188921e-01 6.49449468e-01 9.05556858e-01
-1.57510161e-01 -6.80626810e-01 -2.73093879e-01 1.04968205e-01
2.22847059e-01 -7.58131266e-01 -2.63755828e-01 3.55647117e-01
-1.04024029e+00 -1.35142636e+00 -1.93714932e-01 -6.07799232e-01
6.61449432e-01 2.24215671e-01 1.66935682e+00 9.08621728e-01
-5.45066260e-02 4.52467829e-01 -4.98224229e-01 1.16830729e-01
-7.75145113e-01 3.73399049e-01 -6.00666940e-01 -1.57129928e-01
1.06328234e-01 -6.97888970e-01 -1.61036208e-01 -1.20351508e-01
-1.04520953e+00 -1.76521793e-01 4.60872352e-01 9.11956787e-01
1.07657321e-01 -1.93233728e-01 2.10516736e-01 -1.57484436e+00
5.02417505e-01 -1.06744349e+00 -5.95436096e-01 1.85403556e-01
-6.18180990e-01 5.35178304e-01 9.23156261e-01 -2.48136014e-01
-6.54279351e-01 -4.75056291e-01 -2.13864505e-01 -3.79143894e-01
2.56387264e-01 1.09034932e+00 1.11209944e-01 -4.44910198e-01
1.05849910e+00 1.23464800e-01 -4.92860712e-02 -4.37132955e-01
7.42647648e-01 2.25817099e-01 4.04486239e-01 -9.89173710e-01
1.01370621e+00 1.42184841e-02 4.15076353e-02 -4.14602160e-01
-4.23111320e-01 -3.29204082e-01 -3.48549455e-01 4.60750610e-01
4.71202433e-01 -6.59960151e-01 -6.93493664e-01 3.57596576e-01
-1.05475485e+00 -6.30770326e-01 -3.18298101e-01 -1.37267619e-01
-3.30917537e-01 6.17781043e-01 -1.07985663e+00 -5.86682260e-01
-9.01121870e-02 -1.25515854e+00 1.08287179e+00 -3.04340571e-01
-1.89436555e-01 -1.31529236e+00 8.05535465e-02 -1.39383346e-01
6.96752667e-01 3.72298777e-01 1.67911100e+00 -7.49063075e-01
-8.28848898e-01 -1.52191326e-01 -4.32118922e-01 2.56762385e-01
-7.09727630e-02 -1.55847132e-01 -8.54722321e-01 -3.54332238e-01
-1.96320027e-01 -3.36831778e-01 1.03137827e+00 -8.61645564e-02
1.32381797e+00 -6.32788599e-01 -4.85495836e-01 7.62006342e-01
1.54746401e+00 -3.24650466e-01 7.36542165e-01 6.55822381e-02
1.20518446e+00 2.86627650e-01 -1.32539600e-01 2.88757771e-01
7.97964752e-01 4.60150808e-01 7.72830367e-01 4.54879031e-02
-3.06518078e-01 -6.05299056e-01 5.03179193e-01 1.06411433e+00
1.44306733e-03 -1.03620894e-01 -1.18256688e+00 7.42502272e-01
-1.88723719e+00 -7.26005971e-01 -3.98818523e-01 2.23000836e+00
8.75106752e-01 1.62460525e-02 1.88428149e-01 2.38667220e-01
4.32536811e-01 6.80911958e-01 -2.31939957e-01 -6.66014135e-01
2.64014184e-01 6.69498682e-01 7.26161659e-01 4.41497415e-01
-9.64631438e-01 8.05925548e-01 6.12319326e+00 6.75500095e-01
-1.04620337e+00 3.00496161e-01 2.85909951e-01 4.42440957e-01
-9.39487517e-01 4.49124455e-01 -4.55197871e-01 1.71947569e-01
1.46875811e+00 -3.91198367e-01 7.74031520e-01 1.01288557e+00
-6.28942728e-01 3.17154288e-01 -1.69236755e+00 7.81938970e-01
-2.75325775e-03 -1.56943238e+00 -2.88533177e-02 7.19105974e-02
4.95887518e-01 3.53646159e-01 -1.32159919e-01 8.22900057e-01
7.80064702e-01 -1.26801169e+00 6.67801678e-01 2.35501617e-01
6.68002844e-01 -4.72446293e-01 5.28127909e-01 1.52555471e-02
-1.53304255e+00 -4.99516129e-01 -2.24118009e-01 -2.17196539e-01
-3.89911354e-01 5.81772387e-01 -5.26917875e-01 9.04649794e-01
3.77185285e-01 1.18932235e+00 -1.00323427e+00 7.73353577e-01
-4.09544319e-01 8.10815811e-01 2.71216892e-02 -2.51179021e-02
2.08654106e-01 2.61573315e-01 3.51510614e-01 1.53188252e+00
9.59413871e-02 -2.99673855e-01 2.04651996e-01 1.29690087e+00
-5.55308461e-01 -2.02707142e-01 -1.04178548e+00 -2.92580366e-01
3.49111438e-01 9.19059753e-01 -3.66589338e-01 -3.66229177e-01
-1.04550862e+00 4.29027647e-01 7.83626497e-01 4.10152525e-01
-7.76980817e-01 -4.71594095e-01 5.54875970e-01 2.67374575e-01
4.14901078e-01 -4.06501889e-01 -2.31005535e-01 -1.28860319e+00
1.41008183e-01 -1.21787405e+00 6.48734331e-01 -3.28497797e-01
-1.05607045e+00 5.26773572e-01 -4.00235802e-02 -6.90335155e-01
-4.11376864e-01 -6.72014832e-01 -4.98564392e-01 8.18571687e-01
-1.55766547e+00 -1.25953054e+00 -2.42692791e-03 4.81585175e-01
2.12471932e-01 2.35509104e-03 8.78073454e-01 3.08579147e-01
-4.50278014e-01 9.59999382e-01 -2.02631712e-01 3.00191432e-01
4.81531657e-02 -1.37167203e+00 1.05673957e+00 1.01250839e+00
3.21415275e-01 1.23939252e+00 3.60768408e-01 -4.83807087e-01
-2.25767183e+00 -1.20602047e+00 8.80333722e-01 -7.16488421e-01
1.15791893e+00 -5.98361611e-01 -1.35672462e+00 1.31262422e+00
-8.10801014e-02 5.43805361e-01 3.02571237e-01 6.57867253e-01
-1.32188213e+00 2.84410324e-02 -7.09408879e-01 4.22989637e-01
1.30484986e+00 -1.16644561e+00 -6.78446829e-01 2.66204476e-01
1.11297989e+00 -5.06912410e-01 -1.03617775e+00 2.85239041e-01
1.92388386e-01 -1.01912379e+00 9.59279239e-01 -9.68671262e-01
8.12694132e-01 -1.04853913e-01 -2.66862959e-01 -1.16946375e+00
-3.97075951e-01 -7.49646723e-01 -7.29671538e-01 1.13206780e+00
4.32851195e-01 -9.40829515e-01 5.40264547e-01 3.11363280e-01
-5.36629021e-01 -8.85641098e-01 -5.52989960e-01 -6.86847806e-01
1.81249037e-01 -7.03694999e-01 7.46680915e-01 1.09376919e+00
4.07186747e-01 3.08667243e-01 -6.60368660e-03 1.17529124e-01
2.12812662e-01 2.95732886e-01 7.95242846e-01 -1.12133050e+00
-6.75839901e-01 -7.88872957e-01 -8.63525808e-01 -1.20113027e+00
5.66589236e-01 -1.59806454e+00 -2.03433827e-01 -1.39870906e+00
2.98818082e-01 -4.30376619e-01 -2.87249804e-01 7.97830582e-01
-2.73221105e-01 -3.60649943e-01 6.52140379e-03 1.17528075e-02
-4.99088556e-01 4.41963762e-01 5.45424938e-01 -5.03550768e-01
1.61725834e-01 -1.69792205e-01 -9.77354884e-01 3.79899472e-01
2.66509771e-01 -6.60170972e-01 -5.91981471e-01 -7.23944843e-01
7.53093302e-01 1.99028581e-01 5.19665003e-01 -6.84939384e-01
1.77471668e-01 1.29413545e-01 -4.44245040e-01 1.64881572e-01
-1.89262614e-01 -5.10147870e-01 -8.18283707e-02 4.45770651e-01
-5.88278532e-01 2.81854928e-01 3.00651848e-01 8.72831047e-01
-2.13755995e-01 -1.96640894e-01 4.10329938e-01 -2.27157965e-01
-6.61763668e-01 4.67441112e-01 -1.66736856e-01 2.65780121e-01
6.45705342e-01 8.42645839e-02 -7.85089672e-01 -3.03648740e-01
-1.87410802e-01 -1.03760354e-01 6.03022397e-01 6.04078650e-01
4.79172498e-01 -1.08629155e+00 -4.55955386e-01 2.80336261e-01
4.44677293e-01 -5.52203096e-02 5.91612700e-03 9.19640779e-01
-4.04393077e-01 3.09359074e-01 2.74867564e-01 -4.90100324e-01
-9.99244809e-01 1.15507066e+00 2.82584757e-01 -5.34547687e-01
-9.22085762e-01 7.92218149e-01 5.12201071e-01 -6.89311862e-01
5.99070778e-03 -9.75181043e-01 -2.00966792e-03 -5.19370437e-01
3.94562721e-01 1.76891416e-01 3.96879077e-01 -4.44054216e-01
-4.50722903e-01 3.74557734e-01 -1.12156898e-01 6.47960305e-01
1.33486986e+00 4.11582172e-01 -7.55475163e-01 5.05156219e-01
1.50280523e+00 1.23913743e-01 -6.93699241e-01 -5.52753091e-01
5.03646076e-01 -5.62349081e-01 2.01283693e-02 -4.07151759e-01
-1.13233197e+00 1.02397561e+00 -1.03049919e-01 6.00362420e-01
7.85649776e-01 1.70927420e-01 7.08854437e-01 4.57505375e-01
5.89646459e-01 -1.22329138e-01 -5.85266687e-02 7.04403460e-01
7.40740061e-01 -7.97801137e-01 -3.91650312e-02 -4.70449924e-01
-6.23063557e-02 1.02730870e+00 3.60343844e-01 -3.18347335e-01
5.04785955e-01 4.98931885e-01 -5.60172737e-01 -4.43702281e-01
-1.02968550e+00 -2.92668361e-02 3.11834991e-01 7.66085923e-01
8.57406378e-01 1.75513048e-02 -1.35876064e-03 6.98568344e-01
1.36067033e-01 -2.17185795e-01 5.81909180e-01 1.11300969e+00
-1.32509932e-01 -1.32129955e+00 -1.10158287e-02 8.43080580e-01
-5.05178809e-01 -5.10605514e-01 -1.44254982e-01 8.62442374e-01
-4.10417706e-01 7.72657871e-01 -2.41404846e-01 -5.91944218e-01
2.87573785e-01 -4.42826040e-02 4.58540291e-01 -1.08201492e+00
-9.57415164e-01 -7.89437175e-01 2.79642403e-01 -9.70284820e-01
-2.36739218e-02 -2.15482980e-01 -1.15399563e+00 -5.82412124e-01
-2.66895622e-01 -1.39605738e-02 2.64462173e-01 6.40459716e-01
5.51690698e-01 8.12646389e-01 3.92804384e-01 -6.07998729e-01
-7.69479752e-01 -7.78326869e-01 -2.66339988e-01 5.37814736e-01
7.03360498e-01 -5.05390406e-01 -3.25936586e-01 -2.47476827e-02]
|
[7.533560276031494, 7.846220970153809]
|
f334a39a-ed8d-48d5-ac94-fc0862fb218a
|
visually-explaining-3d-cnn-predictions-for
|
2207.12859
| null |
https://arxiv.org/abs/2207.12859v1
|
https://arxiv.org/pdf/2207.12859v1.pdf
|
Visually explaining 3D-CNN predictions for video classification with an adaptive occlusion sensitivity analysis
|
This paper proposes a method for visually explaining the decision-making process of 3D convolutional neural networks (CNN) with a temporal extension of occlusion sensitivity analysis. The key idea here is to occlude a specific volume of data by a 3D mask in an input 3D temporal-spatial data space and then measure the change degree in the output score. The occluded volume data that produces a larger change degree is regarded as a more critical element for classification. However, while the occlusion sensitivity analysis is commonly used to analyze single image classification, it is not so straightforward to apply this idea to video classification as a simple fixed cuboid cannot deal with the motions. To this end, we adapt the shape of a 3D occlusion mask to complicated motions of target objects. Our flexible mask adaptation is performed by considering the temporal continuity and spatial co-occurrence of the optical flows extracted from the input video data. We further propose to approximate our method by using the first-order partial derivative of the score with respect to an input image to reduce its computational cost. We demonstrate the effectiveness of our method through various and extensive comparisons with the conventional methods in terms of the deletion/insertion metric and the pointing metric on the UCF-101. The code is available at: https://github.com/uchiyama33/AOSA.
|
['Kazuhiro Fukui', 'Koichiro Niinuma', 'Naoya Sogi', 'Tomoki Uchiyama']
|
2022-07-26
| null | null | null | null |
['video-classification']
|
['computer-vision']
|
[-3.94545794e-02 -3.60270739e-01 -8.46597105e-02 -3.12670052e-01
1.95634604e-01 -6.02381468e-01 4.08377826e-01 -2.60961413e-01
-5.73228359e-01 4.48570400e-01 -5.58060408e-02 -4.96057332e-01
-1.60927415e-01 -5.67727447e-01 -6.35775626e-01 -7.91265726e-01
-1.71133026e-01 -2.33821481e-01 3.23487520e-01 6.34050593e-02
2.14306176e-01 9.65872705e-01 -1.51580775e+00 2.34773830e-01
5.85399151e-01 1.26752579e+00 1.86238945e-01 4.63369995e-01
-1.85904607e-01 4.22324210e-01 -6.92286134e-01 1.23117547e-02
6.20953500e-01 -3.09854478e-01 -5.84933341e-01 3.57497990e-01
4.73135293e-01 -3.86836469e-01 -3.83931994e-01 1.06316125e+00
3.28175306e-01 3.74562383e-01 5.81105828e-01 -1.33952546e+00
-5.61645627e-01 -7.38113299e-02 -6.02768123e-01 6.48475945e-01
-1.09192403e-02 6.19092174e-02 5.14828086e-01 -8.72768641e-01
6.82441294e-01 1.33398330e+00 3.12265962e-01 5.85732520e-01
-1.06507456e+00 -4.93040919e-01 4.92516279e-01 3.56155932e-01
-1.28488719e+00 -2.68865824e-01 1.16709363e+00 -7.60975838e-01
6.33833945e-01 3.67144942e-01 8.06886494e-01 8.38726759e-01
1.77251801e-01 7.09340394e-01 9.39634204e-01 -2.64819831e-01
1.60312191e-01 3.76407057e-02 2.23346949e-01 7.47898757e-01
2.09954157e-01 -2.11588163e-02 -2.34683871e-01 2.02417344e-01
1.05275238e+00 2.13818774e-01 -4.51147765e-01 -6.58146322e-01
-1.16034865e+00 6.43462539e-01 6.64323866e-01 2.67589331e-01
-5.16744144e-02 -1.23779941e-02 3.01979184e-01 3.97768915e-01
6.68987095e-01 5.95836565e-02 -4.47082192e-01 2.31133759e-01
-6.91075802e-01 1.61254033e-01 4.76959795e-01 7.07164228e-01
6.88596070e-01 8.99497569e-02 -2.64862686e-01 4.99609679e-01
2.18777433e-01 1.44521609e-01 4.09577817e-01 -1.17174220e+00
4.95764583e-01 7.46537268e-01 1.21397935e-01 -1.15463376e+00
-4.53505069e-01 -5.85745752e-01 -9.59021926e-01 9.48393643e-01
7.59836614e-01 -1.15797222e-01 -8.23219836e-01 1.80999875e+00
6.13849401e-01 9.34074968e-02 -1.14237420e-01 1.22071993e+00
8.22455347e-01 5.15797675e-01 -2.00784132e-01 -4.22504276e-01
1.30135262e+00 -7.97159195e-01 -7.75065362e-01 1.42661452e-01
5.30277967e-01 -5.96060216e-01 1.11168158e+00 7.67689422e-02
-8.97400916e-01 -7.65029311e-01 -1.16024387e+00 -1.88543811e-01
-5.22928536e-01 2.93073565e-01 4.00157094e-01 4.31143731e-01
-8.37035656e-01 6.30841017e-01 -9.06556129e-01 -1.76639244e-01
5.02487779e-01 2.30711564e-01 -4.15514797e-01 2.78858602e-01
-1.08506954e+00 6.51107967e-01 2.45814189e-01 3.25193822e-01
-4.59298581e-01 -6.05189800e-01 -7.14076817e-01 -4.79065478e-02
2.96364844e-01 -4.92925137e-01 8.68432879e-01 -1.00241053e+00
-1.37085259e+00 7.77739108e-01 -3.27846140e-01 -2.05951929e-01
9.15729105e-01 -1.05685264e-01 -4.81279008e-02 1.13847502e-01
-1.29231483e-01 6.77821517e-01 7.67023981e-01 -9.75099027e-01
-3.50551099e-01 -3.15335512e-01 3.50007832e-01 4.72039312e-01
-3.81637514e-01 -2.07977340e-01 -4.69973922e-01 -7.96131611e-01
4.69457626e-01 -8.02831888e-01 -1.33330217e-02 6.43124521e-01
-2.11724654e-01 -1.40333146e-01 1.16809142e+00 -4.93438959e-01
1.14568150e+00 -2.39236903e+00 5.29268198e-02 -5.90612143e-02
3.38311732e-01 3.96130651e-01 -5.07323891e-02 -1.27920702e-01
-1.75564960e-01 2.07108870e-01 -2.65584022e-01 -2.87927061e-01
-2.39461824e-01 -5.20003997e-02 -1.22189417e-01 6.30912602e-01
3.42908442e-01 6.28981590e-01 -7.77509332e-01 -4.42359686e-01
4.31974262e-01 5.31273484e-01 -5.52878022e-01 -4.37003970e-02
-3.06473877e-02 6.42021894e-01 -5.41940868e-01 5.56502342e-01
8.06163788e-01 -1.12062119e-01 -2.64390171e-01 -3.07223678e-01
-3.67918879e-01 5.58931939e-02 -1.47545075e+00 1.59537554e+00
3.54300328e-02 9.97198761e-01 -1.46310121e-01 -1.04780400e+00
1.01133692e+00 2.15211809e-01 6.44385397e-01 -5.10470271e-01
2.07324624e-01 -7.95132443e-02 8.16926956e-02 -7.97629356e-01
4.70634438e-02 3.20509017e-01 3.88939649e-01 2.31826931e-01
-3.70506793e-01 1.60529330e-01 2.63311714e-01 -1.29295036e-01
8.39272082e-01 3.55394840e-01 3.62534851e-01 -3.23086500e-01
6.35185838e-01 -1.36249334e-01 7.60312498e-01 3.98326933e-01
-5.96561193e-01 6.27580702e-01 7.78980076e-01 -9.11842823e-01
-8.66446316e-01 -9.59267080e-01 -3.16697299e-01 5.73081672e-01
3.51396799e-01 5.78089207e-02 -6.76132321e-01 -7.23721445e-01
1.13552138e-01 1.87800497e-01 -7.36568034e-01 -6.90767765e-02
-6.51731670e-01 -6.09012544e-01 1.64390430e-01 3.64633113e-01
1.01643729e+00 -1.02081692e+00 -1.03871667e+00 -1.06373891e-01
-1.81604505e-01 -1.04668319e+00 -5.46256244e-01 -5.25793955e-02
-1.05076730e+00 -1.06722856e+00 -6.61661565e-01 -8.08167279e-01
7.88226306e-01 4.94296968e-01 5.70904136e-01 8.10966045e-02
-2.98180997e-01 -1.58410799e-02 -1.54570371e-01 -1.80156857e-01
-3.15929577e-02 -1.84649020e-01 6.89989552e-02 2.63999015e-01
2.65367061e-01 -4.16728973e-01 -8.70408893e-01 5.24749041e-01
-9.35691774e-01 3.19047272e-01 2.09959015e-01 6.03656590e-01
4.04599339e-01 -5.29991500e-02 8.31323117e-02 -3.93850923e-01
3.91193718e-01 -5.15326336e-02 -6.84937000e-01 5.34860305e-02
-2.07477868e-01 4.61646542e-02 4.77946401e-01 -8.28496873e-01
-9.03428257e-01 2.21282169e-01 2.47834280e-01 -8.48782063e-01
-3.16841304e-01 1.56991228e-01 -1.94929019e-01 -1.34323508e-01
5.39905667e-01 -2.28785485e-01 5.35899997e-02 -4.54066366e-01
1.16255313e-01 3.35704356e-01 2.91212201e-01 -2.06780851e-01
7.74030626e-01 6.94464326e-01 4.46006119e-01 -7.01754808e-01
-5.97450078e-01 -2.75064379e-01 -9.91405189e-01 -4.80957031e-01
9.53633726e-01 -5.28532207e-01 -1.03905189e+00 5.11568964e-01
-1.35570168e+00 -3.45944822e-01 -1.25722244e-01 6.78350627e-01
-3.28577667e-01 3.45677763e-01 -4.08972740e-01 -7.73244321e-01
-8.39935523e-03 -1.33804691e+00 7.07390547e-01 1.20852202e-01
6.13545403e-02 -9.16991591e-01 -2.58889496e-01 -8.69821291e-03
1.64698943e-01 5.49666286e-01 8.79011214e-01 -2.33654559e-01
-6.74246550e-01 -6.56334236e-02 -3.18754703e-01 4.41569895e-01
3.08541030e-01 1.89776450e-01 -1.03579903e+00 -1.71092540e-01
2.77232528e-01 1.64015353e-01 9.73035514e-01 6.57745719e-01
1.37030947e+00 -2.60984898e-01 -1.86410591e-01 8.65743399e-01
1.27939355e+00 5.26241004e-01 4.58886743e-01 3.43263716e-01
7.53388762e-01 7.31817305e-01 5.47620535e-01 2.72997886e-01
-1.01548679e-01 9.11284029e-01 5.90747952e-01 -2.11050376e-01
-3.08177382e-01 -1.49706481e-02 2.03724593e-01 4.23919976e-01
-3.46600413e-01 -2.48855159e-01 -7.03293443e-01 2.86928028e-01
-1.85132253e+00 -9.02955174e-01 -1.02331378e-01 2.26910019e+00
5.16957343e-01 2.14920506e-01 -1.55576825e-01 3.98636669e-01
8.61810386e-01 3.53338689e-01 -5.98446012e-01 -1.45367220e-01
-1.53588310e-01 -3.51662457e-01 3.73264819e-01 7.12780058e-01
-1.17978799e+00 7.15162575e-01 5.17859173e+00 5.72328031e-01
-1.32522571e+00 5.81134995e-03 7.48881876e-01 -2.31990919e-01
6.43454269e-02 -1.18444040e-01 -6.55093610e-01 4.84103411e-01
8.59445557e-02 7.52378702e-02 2.92708844e-01 4.43050563e-01
6.61270797e-01 -1.35618821e-01 -1.10735917e+00 1.03578758e+00
-8.79279673e-02 -1.22208464e+00 2.09162459e-01 9.48537514e-03
5.48340976e-01 -3.56936276e-01 2.59384722e-01 -2.61747748e-01
-3.53003532e-01 -6.77557051e-01 8.58825326e-01 6.30237401e-01
6.85431004e-01 -2.17568606e-01 4.45747703e-01 2.46128187e-01
-1.27727866e+00 -1.53175548e-01 -3.13298613e-01 -3.09285611e-01
-5.31663327e-03 5.53140759e-01 -5.52831471e-01 1.61086559e-01
8.59225810e-01 7.73312509e-01 -4.65250462e-01 1.18436182e+00
-8.66604894e-02 1.64349616e-01 -2.25450248e-01 -8.69270489e-02
1.62363604e-01 -4.45790857e-01 8.37187469e-01 9.19162333e-01
4.22179371e-01 1.73669994e-01 -1.57998174e-01 8.31314147e-01
6.92153058e-04 -6.69999979e-03 -7.76072979e-01 4.44274008e-01
2.89986849e-01 9.86800313e-01 -8.26580524e-01 -2.55136997e-01
-4.63058174e-01 8.47488105e-01 2.56806314e-01 8.96787524e-01
-8.00243139e-01 -2.26273403e-01 8.10942054e-01 1.03359446e-01
3.18573684e-01 -4.75298703e-01 -4.45100099e-01 -1.12900603e+00
4.64548588e-01 -3.55511248e-01 3.13124150e-01 -9.00690258e-01
-7.26620078e-01 5.89398444e-01 1.45631611e-01 -1.76750958e+00
1.52839720e-01 -8.20377707e-01 -5.76079547e-01 7.37129092e-01
-1.45989490e+00 -4.64780718e-01 -6.68519855e-01 6.54865205e-01
6.53303683e-01 -5.26237972e-02 3.91310751e-01 2.92956680e-01
-6.47438407e-01 4.64206100e-01 -7.24178627e-02 2.79143006e-01
5.22074282e-01 -8.78438294e-01 2.44127005e-01 9.37938154e-01
4.54283989e-04 4.29726183e-01 6.77941978e-01 -3.82243127e-01
-8.68917465e-01 -9.49871540e-01 8.28551590e-01 -3.26650828e-01
4.61800367e-01 -3.88895661e-01 -1.01634216e+00 5.13024092e-01
-4.60788943e-02 4.08658504e-01 5.24400547e-02 -4.00496870e-01
-2.23546192e-01 -5.09900033e-01 -1.08329809e+00 8.56485724e-01
1.44155669e+00 -2.58441210e-01 -4.52445686e-01 9.49768871e-02
7.85891056e-01 -5.04813313e-01 -6.56481922e-01 6.40863001e-01
6.30097449e-01 -9.94310319e-01 9.26682889e-01 -5.50893247e-01
2.91570902e-01 -7.40645409e-01 -1.03434540e-01 -1.09044361e+00
-3.94979000e-01 -5.45156538e-01 8.79908074e-03 8.21236074e-01
2.21533105e-01 -6.92476392e-01 7.50969529e-01 4.67772901e-01
-1.24221988e-01 -7.97483087e-01 -1.13119149e+00 -6.63353801e-01
-1.21526115e-01 -3.88298899e-01 4.13063049e-01 8.30269754e-01
-1.84710175e-01 -1.87904015e-01 -1.19058833e-01 3.07295591e-01
5.43386936e-01 1.84884537e-02 6.88962102e-01 -1.25943303e+00
4.06278260e-02 -7.08087683e-01 -7.46128678e-01 -1.21241403e+00
-2.01972984e-02 -6.35057330e-01 -3.32306832e-01 -1.29496694e+00
-9.14560258e-02 -3.26968789e-01 -3.69060695e-01 3.94866556e-01
-2.67706737e-02 3.53336960e-01 3.18978459e-01 4.14624661e-01
-2.19010726e-01 5.31090617e-01 1.79740393e+00 -1.19049415e-01
-4.73744839e-01 2.20503241e-01 -1.76195085e-01 7.01876640e-01
9.13278162e-01 -2.64834791e-01 -5.61271131e-01 -7.07270861e-01
-1.04805715e-01 -3.82909961e-02 7.43023336e-01 -9.82583404e-01
1.35672703e-01 -1.76927507e-01 5.71309686e-01 -5.85622489e-01
3.60271513e-01 -8.90565276e-01 9.96942967e-02 5.72646379e-01
-4.53309357e-01 2.77104139e-01 2.32233018e-01 4.13113296e-01
-1.42849013e-01 -2.04820395e-01 7.26311028e-01 -6.41165674e-02
-7.81280220e-01 4.41814989e-01 -1.51508436e-01 -2.41994068e-01
1.05406308e+00 -4.27779257e-01 -4.65581208e-01 -1.33821338e-01
-8.69612694e-01 -3.42990793e-02 4.44660962e-01 5.10239065e-01
6.42257988e-01 -1.50620103e+00 -3.52766156e-01 5.84049582e-01
-2.44430546e-02 -1.03404373e-01 2.36649245e-01 8.53976786e-01
-6.49260283e-01 4.25749063e-01 -4.69748259e-01 -7.84188747e-01
-1.33255196e+00 5.94345689e-01 6.82421744e-01 1.13481745e-01
-6.40357733e-01 4.89516884e-01 5.52322984e-01 1.25425458e-01
6.00442648e-01 -7.00303733e-01 -4.84586567e-01 3.07119787e-02
3.65085363e-01 3.34408790e-01 -3.51470038e-02 -4.69083935e-01
-3.18872958e-01 9.16469812e-01 3.35251987e-01 -1.00133069e-01
9.76562142e-01 -2.06592351e-01 2.38552410e-03 5.93552172e-01
1.46768057e+00 -3.09232920e-01 -1.84361887e+00 -2.85266280e-01
-3.75053257e-01 -6.12117887e-01 -4.95823137e-02 -2.97000200e-01
-1.36210155e+00 9.98098493e-01 8.95749986e-01 3.73859942e-01
1.18723989e+00 -9.86949429e-02 2.17640832e-01 3.26216877e-01
-1.72124788e-01 -7.91120350e-01 5.13094626e-02 3.52723628e-01
1.02045333e+00 -1.25151825e+00 -1.15732566e-01 -4.31077927e-01
-3.84564281e-01 1.11211956e+00 7.32870698e-01 -9.86701548e-02
8.95381629e-01 -6.08396009e-02 2.61721671e-01 -4.42975201e-02
-5.04600763e-01 -1.86385348e-01 4.54945952e-01 4.20229375e-01
2.77821779e-01 -1.00796171e-01 -5.53030789e-01 5.17923720e-02
7.08573386e-02 -9.40533429e-02 4.08378422e-01 7.54148483e-01
-1.90584153e-01 -6.03617609e-01 -3.36808056e-01 5.97002469e-02
-2.58180141e-01 1.35867238e-01 -3.20936516e-02 9.37721193e-01
4.70181942e-01 7.85421014e-01 3.73016715e-01 -2.78950602e-01
3.62934172e-01 1.14568420e-01 3.60156476e-01 -2.92161673e-01
3.57518382e-02 1.18167311e-01 -3.32175553e-01 -5.83998561e-01
-8.05717647e-01 -8.05149555e-01 -1.13311040e+00 -1.93083778e-01
4.69485223e-02 -1.51566058e-01 7.11773753e-01 8.21973503e-01
1.70670509e-01 5.08292556e-01 7.38253653e-01 -9.02357459e-01
-2.05069676e-01 -8.12796891e-01 -3.65353078e-01 6.99646354e-01
8.06755900e-01 -1.05282903e+00 -7.86099553e-01 2.58655548e-01]
|
[9.161250114440918, -0.6081537008285522]
|
2b8d304a-5d93-43e2-8e1a-5f32c24235d0
|
selfevolve-a-code-evolution-framework-via
|
2306.02907
| null |
https://arxiv.org/abs/2306.02907v1
|
https://arxiv.org/pdf/2306.02907v1.pdf
|
SelfEvolve: A Code Evolution Framework via Large Language Models
|
Large language models (LLMs) have already revolutionized code generation, after being pretrained on publicly available code data. However, while various methods have been proposed to augment LLMs with retrieved knowledge and enhance the quality of code generation, the performance of these retrieval-based methods is limited by the strength of the retrievers used. In addition, while LLMs show great emergent ability, they still struggle to produce the correct code in one turn. To address these challenges, we propose a novel two-step pipeline, called \autoknow, that leverages LLMs as both knowledge providers and self-reflective programmers. Unlike retrieval-based methods, \autoknow~obtains the knowledge from input prompts and generates intermediate code based on the generated knowledge. After that, \autoknow~asks LLM to act as an expert programmer to perform debugging for the generated code. This is achieved by receiving the error message from the interpreter, without requiring special test cases for correctness verification. We evaluate \autoknow~on three code generation datasets, including DS-1000 for data science code, HumanEval for software engineering code, and TransCoder for C++-to-Python translation. Our empirical experiments show that \autoknow~outperforms strong baselines by a significant margin on all datasets. We also conduct exhaustive analytical experiments to validate the effectiveness of the two stages of \autoknow, and find that both are superior to other prompting-based methods. Further scalability analysis demonstrates that \autoknow~can be adapted to other more advanced models, such as GPT-4, and bring consistent efficacy improvement.
|
['Yu Wang', 'Yuhao Wang', 'Shuyang Jiang']
|
2023-06-05
| null | null | null | null |
['code-generation']
|
['computer-code']
|
[-2.27701012e-02 6.82067079e-03 -4.36331809e-01 -1.85106352e-01
-1.24739826e+00 -7.13660121e-01 4.68836904e-01 9.18157548e-02
-5.93584068e-02 3.11648518e-01 1.90810531e-01 -6.80187881e-01
1.48402959e-01 -7.29730487e-01 -9.90077734e-01 7.53000528e-02
1.10664412e-01 1.57176986e-01 2.62313783e-01 -3.48168135e-01
6.02383375e-01 -2.43430644e-01 -1.47379506e+00 6.43144965e-01
1.50461102e+00 3.09451252e-01 5.22453010e-01 6.87097609e-01
-3.70816916e-01 1.11844063e+00 -7.57095456e-01 -4.61496413e-01
4.19318713e-02 -3.45382780e-01 -9.81082499e-01 -5.31601250e-01
1.20746307e-01 -4.89111781e-01 7.08845956e-03 1.09403801e+00
4.82402474e-01 -3.34285021e-01 2.28154942e-01 -1.25803363e+00
-9.86338139e-01 1.16724503e+00 -5.26764512e-01 3.67008261e-02
7.41878748e-01 3.63525659e-01 9.11942661e-01 -1.11708224e+00
6.11832142e-01 1.05843568e+00 6.97278619e-01 5.44029891e-01
-1.15072346e+00 -7.13686526e-01 -1.43011138e-01 -1.94982886e-02
-1.39361238e+00 -5.27363420e-01 5.03021896e-01 -7.17455268e-01
1.48544240e+00 7.90583566e-02 1.97214723e-01 1.05904663e+00
5.02611846e-02 9.10932183e-01 7.79967606e-01 -5.91892719e-01
-5.36833629e-02 2.82696724e-01 2.39610925e-01 9.81591702e-01
1.97524339e-01 1.39794126e-02 -5.65350115e-01 -5.50760686e-01
4.45802480e-01 -1.21844783e-01 -4.10459161e-01 9.27272886e-02
-1.31683064e+00 5.62010944e-01 2.54827321e-01 2.77438790e-01
-1.66192695e-01 4.22332674e-01 3.46433580e-01 5.43768227e-01
2.62072057e-01 8.60829473e-01 -5.38449585e-01 -6.60675466e-01
-1.07342815e+00 3.19820583e-01 9.66251493e-01 1.41047955e+00
8.84010136e-01 1.65993422e-01 -3.93757612e-01 7.61986077e-01
4.20711786e-01 6.96047962e-01 7.10360050e-01 -8.18184555e-01
9.50373173e-01 1.08485115e+00 3.49182375e-02 -8.60882223e-01
5.60056195e-02 -5.31397879e-01 -2.92626679e-01 -4.00907695e-02
-9.00564790e-02 -1.76494434e-01 -5.08853137e-01 1.63348997e+00
-1.47599354e-01 6.34791553e-02 2.14868382e-01 6.59720242e-01
8.20195138e-01 6.78819001e-01 -2.51843184e-01 1.28506109e-01
9.45295572e-01 -1.16116297e+00 -1.67287692e-01 -4.81219918e-01
1.07121539e+00 -8.65988433e-01 1.31231618e+00 4.17630613e-01
-1.01327908e+00 -5.40321589e-01 -1.06367755e+00 -7.99989700e-03
1.13902152e-01 5.28285921e-01 7.72545874e-01 5.09266675e-01
-1.33041990e+00 6.15561426e-01 -9.41867948e-01 -1.69343010e-01
1.51723742e-01 3.69841047e-02 -6.38923198e-02 -1.34826705e-01
-8.30416858e-01 5.64503491e-01 3.54313552e-01 -1.29411981e-01
-1.05563509e+00 -1.06524754e+00 -8.18682849e-01 1.83289386e-02
3.30558628e-01 -7.46594071e-01 1.68418145e+00 -8.28126490e-01
-1.42286599e+00 3.75205815e-01 -2.02803269e-01 -1.70019120e-01
2.20072284e-01 -4.66860354e-01 -3.70009512e-01 -2.49732196e-01
2.84407765e-01 4.68779206e-01 6.29654109e-01 -1.23215806e+00
-4.31862354e-01 2.18627810e-01 4.19402331e-01 -1.92948207e-01
-4.84779984e-01 2.46609956e-01 -8.17412496e-01 -4.90845352e-01
-3.49687785e-01 -8.40406835e-01 -1.31765353e-02 -3.86786431e-01
-3.82348508e-01 -1.99397564e-01 4.99223590e-01 -7.73914576e-01
1.71627319e+00 -2.25846696e+00 1.95239425e-01 4.23824005e-02
2.59981781e-01 4.09940541e-01 -5.68014264e-01 6.01928830e-01
6.22647144e-02 4.20099616e-01 -2.39885300e-01 -4.98388298e-02
2.64896393e-01 -1.06740408e-01 -6.40889764e-01 -1.49652228e-01
4.38740015e-01 1.29895055e+00 -1.11922860e+00 -3.48621637e-01
-3.14312220e-01 2.03540549e-01 -9.95810032e-01 3.97413373e-01
-5.97736955e-01 -2.07828507e-02 -5.36742389e-01 6.57398939e-01
2.39398867e-01 -6.07834518e-01 5.99444956e-02 3.06023270e-01
-6.68038726e-02 5.87858140e-01 -8.59144628e-01 1.94164431e+00
-8.00188780e-01 5.57984710e-01 -3.04068804e-01 -4.30336952e-01
9.65583503e-01 3.69611025e-01 -1.34136409e-01 -7.96648920e-01
-3.82386714e-01 4.61211890e-01 -3.18227336e-02 -6.94588721e-01
6.92698896e-01 4.39357251e-01 -2.44845480e-01 8.68836761e-01
-1.85480565e-01 -1.52195469e-01 3.19792211e-01 5.46214521e-01
1.54764330e+00 4.62394297e-01 1.97331104e-02 4.84039113e-02
3.66042107e-01 2.91681111e-01 5.26072204e-01 9.42450047e-01
4.29956615e-01 3.94624174e-01 4.74509239e-01 8.04265216e-03
-7.38844454e-01 -7.65302837e-01 3.77018481e-01 1.15219319e+00
-2.97747403e-02 -1.00168359e+00 -8.30755711e-01 -8.22734296e-01
-1.39300480e-01 1.00859654e+00 -2.19309494e-01 -4.97800201e-01
-6.92691624e-01 -6.95534468e-01 9.58078861e-01 7.05256224e-01
4.66391474e-01 -9.93185341e-01 -5.60637057e-01 3.34786057e-01
-5.16976535e-01 -7.70366609e-01 -5.86483240e-01 -7.50381947e-02
-7.17348576e-01 -1.03206313e+00 -2.37174660e-01 -5.86826682e-01
7.55070090e-01 2.62367427e-01 1.57011390e+00 7.33095646e-01
-1.43987879e-01 3.25714052e-01 -6.46903574e-01 -1.68059736e-01
-1.10147905e+00 4.51596022e-01 -3.55388582e-01 -6.99603021e-01
1.88687786e-01 -5.32182217e-01 -4.61854994e-01 2.36478478e-01
-9.37450826e-01 3.34476292e-01 9.85904813e-01 6.25043929e-01
1.31430894e-01 -1.35797009e-01 6.62802219e-01 -1.00576317e+00
8.39037657e-01 -7.20597744e-01 -8.16614985e-01 4.23381656e-01
-9.12637949e-01 3.98842037e-01 9.04299915e-01 -3.55809361e-01
-1.06629527e+00 -3.19287270e-01 -1.29302114e-01 -1.64096057e-01
3.25587839e-01 1.10567951e+00 9.72103551e-02 4.94538471e-02
1.09768510e+00 4.64237362e-01 -3.26953620e-01 -6.14360750e-01
2.68855721e-01 8.77512157e-01 5.52367806e-01 -1.16428161e+00
1.10257471e+00 -2.36712843e-01 -8.08973730e-01 -1.68438271e-01
-4.33026761e-01 -1.12838343e-01 1.66432895e-02 1.85395211e-01
3.14594209e-01 -1.02224779e+00 -3.50461006e-01 3.87117177e-01
-1.30967081e+00 -5.55701971e-01 -4.47264612e-02 1.54734671e-01
-2.66314745e-01 4.18274164e-01 -7.50135958e-01 -4.88608003e-01
-6.21131182e-01 -1.46481192e+00 1.17402899e+00 6.84341192e-02
-3.00879240e-01 -7.37962365e-01 1.30568147e-01 4.71727818e-01
7.50772953e-01 -2.10783735e-01 1.44848049e+00 -5.18680215e-01
-1.05531752e+00 -1.67789921e-01 -2.10797846e-01 3.94213110e-01
-4.83954437e-02 2.84279495e-01 -7.32082248e-01 -3.26683253e-01
-4.07144994e-01 -4.31013703e-01 4.65841830e-01 -3.83315206e-01
9.59701359e-01 -3.76390666e-01 -2.96054184e-01 5.87906241e-01
1.41020238e+00 6.32644743e-02 5.40184557e-01 2.43063509e-01
6.49031818e-01 9.36649069e-02 5.74634135e-01 3.80238205e-01
7.09822834e-01 5.42295933e-01 3.11642736e-01 1.58159330e-01
-2.16346607e-01 -3.68639559e-01 9.97748494e-01 1.49054027e+00
1.59517601e-01 -2.67026760e-03 -1.23538017e+00 7.15607166e-01
-1.75031567e+00 -5.61074793e-01 -2.31693134e-01 2.16165662e+00
1.43681431e+00 1.60862450e-02 -3.04205030e-01 -2.77598023e-01
2.92677283e-01 -2.35760257e-01 -5.32344043e-01 -2.97646880e-01
2.97690719e-01 3.40415537e-01 9.66633707e-02 2.77939677e-01
-4.75490898e-01 1.02116013e+00 6.06374693e+00 7.79040217e-01
-1.20948100e+00 4.80041690e-02 6.69241995e-02 2.11963326e-01
-8.36435676e-01 4.97921497e-01 -9.16337192e-01 4.07706380e-01
1.13347960e+00 -7.38960743e-01 7.68276334e-01 1.19246590e+00
-5.13009243e-02 -1.00832760e-01 -1.42919314e+00 6.93446517e-01
1.41523331e-01 -1.34994280e+00 -1.09110912e-02 -2.85442442e-01
9.84462321e-01 2.75442839e-01 -1.50473952e-01 9.37611699e-01
7.06811786e-01 -9.00584638e-01 7.42255330e-01 5.95531583e-01
7.90804744e-01 -5.07970810e-01 6.46926343e-01 6.61204696e-01
-1.12659228e+00 4.70586345e-02 -2.12714866e-01 -8.47186744e-02
-2.28734285e-01 5.82553685e-01 -1.15172148e+00 7.99761713e-01
4.81925786e-01 6.85808718e-01 -1.14676452e+00 9.60218191e-01
-4.83208746e-01 7.51926482e-01 8.99976566e-02 -4.50007431e-02
-1.14715412e-01 4.09596294e-01 2.74867177e-01 1.34040725e+00
5.48776686e-01 -2.91292876e-01 3.07483494e-01 1.51040792e+00
-3.81864190e-01 2.50701252e-02 -4.24453884e-01 -5.95220864e-01
7.26292670e-01 1.17880893e+00 -1.78458676e-01 -5.39904773e-01
-6.27845764e-01 8.21149647e-01 3.77225548e-01 4.70616281e-01
-8.50576997e-01 -7.85080254e-01 4.42172736e-01 6.68137893e-03
1.03776887e-01 -2.46764541e-01 4.06998470e-02 -1.36311007e+00
5.52346289e-01 -1.45310724e+00 -2.87860110e-02 -1.01462638e+00
-8.76158178e-01 6.74439728e-01 2.51506530e-02 -1.15134001e+00
-6.46525681e-01 -1.79554522e-01 -4.08500671e-01 9.69690979e-01
-1.67696464e+00 -9.27465558e-01 -3.73484254e-01 2.12294966e-01
5.44452131e-01 -1.21834517e-01 7.92145550e-01 3.77769589e-01
-4.16918874e-01 7.52759337e-01 -9.11537781e-02 2.23676234e-01
7.76833892e-01 -1.38720047e+00 8.10286343e-01 1.18895245e+00
1.36463763e-02 1.39671576e+00 3.61785144e-01 -9.18375373e-01
-2.03395152e+00 -1.40571463e+00 9.16109562e-01 -7.76047707e-01
7.66912282e-01 -2.49170095e-01 -1.10777259e+00 7.26249158e-01
1.55196086e-01 -1.60568401e-01 4.72269177e-01 -1.12120295e-02
-6.97969019e-01 3.28090464e-05 -5.62725961e-01 5.22320032e-01
1.02276003e+00 -7.25680113e-01 -6.50303841e-01 2.79797703e-01
1.03173745e+00 -6.37346148e-01 -1.08212531e+00 3.30892116e-01
2.64267623e-01 -8.10534239e-01 6.98484421e-01 -3.64296556e-01
1.04082024e+00 -6.65486574e-01 -1.35866985e-01 -1.23575544e+00
8.63488414e-04 -7.44407356e-01 -9.65722501e-02 1.45283639e+00
7.21741438e-01 -5.80117524e-01 2.28803962e-01 6.83166146e-01
-5.47325790e-01 -5.19412458e-01 -3.02732974e-01 -8.12149048e-01
-6.37940168e-02 -6.71120226e-01 9.18312430e-01 9.35624123e-01
2.48890862e-01 3.50198030e-01 -1.96657985e-01 2.02123612e-01
1.50941253e-01 4.72997218e-01 1.18013000e+00 -8.05856943e-01
-9.52721894e-01 -3.77037257e-01 1.18700311e-01 -1.21490502e+00
2.07601145e-01 -1.39976978e+00 2.27599725e-01 -1.53035140e+00
5.73230624e-01 -5.77816784e-01 1.08582182e-02 9.47418988e-01
-4.79544759e-01 -1.11645401e-01 2.06202883e-02 3.40713829e-01
-7.79910684e-01 3.72278899e-01 9.20100808e-01 -1.71137139e-01
-3.12044144e-01 -2.21856952e-01 -1.11317337e+00 5.13856113e-01
4.58129317e-01 -5.60498059e-01 -6.19983137e-01 -9.60136950e-01
9.54898894e-01 3.67980331e-01 3.43509614e-01 -9.81431842e-01
3.13685805e-01 -5.76867238e-02 -1.49321333e-01 -2.05349818e-01
-4.13302302e-01 -4.21001941e-01 2.89982587e-01 2.98144907e-01
-3.43719125e-01 4.59927887e-01 2.82284170e-01 2.81388283e-01
-2.03171194e-01 -4.88192141e-01 2.90505826e-01 -1.04506835e-01
-7.81261504e-01 -4.94268164e-02 -2.93080986e-01 3.71213526e-01
4.73523140e-01 1.23503670e-01 -1.00375986e+00 -3.33715528e-02
1.48123696e-01 2.30679899e-01 7.61642337e-01 8.24178457e-01
5.48163414e-01 -1.10872149e+00 -6.40312672e-01 2.08536908e-01
6.15751922e-01 -1.22606657e-01 -2.11356923e-01 8.49669635e-01
-4.42865074e-01 3.00812274e-01 3.44570100e-01 -6.32161140e-01
-9.75663185e-01 4.44250882e-01 6.45560026e-02 -2.90011942e-01
-4.40833718e-01 6.52501225e-01 -4.29320261e-02 -7.87942350e-01
-8.86447802e-02 -6.20385051e-01 2.43858203e-01 -6.03594303e-01
5.78741789e-01 1.37053400e-01 3.25510025e-01 -8.44442248e-02
-4.16714340e-01 1.97971180e-01 -3.22582334e-01 6.96939901e-02
1.34214985e+00 3.77730608e-01 -4.44132954e-01 2.63376534e-01
9.90142167e-01 2.97809184e-01 -8.69811356e-01 -4.35140252e-01
2.36753941e-01 -5.70723474e-01 -1.84438929e-01 -1.10923207e+00
-9.57614303e-01 6.94297612e-01 1.11175090e-01 -2.01232046e-01
1.05550766e+00 -8.80877450e-02 6.94690526e-01 6.60985589e-01
7.64140010e-01 -6.93686724e-01 4.04687017e-01 5.48385322e-01
8.94704819e-01 -1.20017397e+00 -2.46030122e-01 -2.33397961e-01
-4.19038057e-01 9.91549075e-01 1.00250196e+00 3.24613035e-01
-9.94152855e-03 5.39460540e-01 -5.48071973e-03 -5.35977222e-02
-1.21778417e+00 1.82352990e-01 1.64320305e-01 3.51709068e-01
8.31822157e-01 -1.95175260e-01 -1.44830048e-01 8.51842642e-01
-3.52549404e-01 4.74507451e-01 6.38158083e-01 1.29363036e+00
-3.02427083e-01 -1.44612861e+00 -2.26184323e-01 5.84668934e-01
-2.30673596e-01 -5.81533074e-01 -2.71159828e-01 5.92616439e-01
-1.01526298e-01 1.03551364e+00 -5.08521676e-01 -6.48452222e-01
3.40129882e-01 1.24477178e-01 3.36649477e-01 -1.10598791e+00
-9.72155929e-01 -2.40336731e-01 2.75113881e-01 -7.32260823e-01
1.17830686e-01 -4.01095033e-01 -1.43452561e+00 -3.18049371e-01
-5.56795776e-01 3.87974769e-01 5.99241078e-01 7.59663045e-01
8.51521432e-01 7.13310242e-01 4.43467826e-01 -2.74260283e-01
-6.97020471e-01 -8.58103156e-01 1.63832232e-01 2.09381342e-01
1.34259179e-01 -2.15299129e-01 -1.35890558e-01 2.71539956e-01]
|
[7.768386363983154, 7.846240997314453]
|
91377a65-f02b-4d54-a17c-a415e03c19c7
|
time-series-prediction-for-food
|
2209.06889
| null |
https://arxiv.org/abs/2209.06889v1
|
https://arxiv.org/pdf/2209.06889v1.pdf
|
Time Series Prediction for Food sustainability
|
With exponential growth in the human population, it is vital to conserve natural resources without compromising on producing enough food to feed everyone. Doing so can improve people's livelihoods, health, and ecosystems for the present and future generations. Sustainable development, a paradigm of the United Nations, is rooted in food, crop, livestock, forest, population, and even the emission of gases. By understanding the overall usage of natural resources in different countries in the past, it is possible to forecast the demand in each country. The proposed solution consists of implementing a machine learning system using a statistical regression model that can predict the top k products that would endure a shortage in each country in a specific period in the future. The prediction performance in terms of absolute error and root mean square error show promising results due to its low errors. This solution could help organizations and manufacturers understand the productivity and sustainability needed to satisfy the global demand.
|
['Fiona Victoria Stanley Jothiraj']
|
2022-09-14
| null | null | null | null |
['time-series-prediction']
|
['time-series']
|
[-1.93280607e-01 -1.25247210e-01 -4.90658194e-01 1.39139429e-01
3.21482807e-01 -2.31685624e-01 3.44323635e-01 4.17600006e-01
-4.61575836e-02 8.83013546e-01 7.36912787e-02 -5.38857877e-01
-9.52608362e-02 -1.31722236e+00 -1.41413257e-01 -7.17827439e-01
2.19325554e-02 3.17130387e-02 -1.52373955e-01 -3.62807244e-01
3.68277490e-01 7.53775537e-01 -1.37023795e+00 -2.52685249e-01
1.18567681e+00 1.13244414e+00 6.53220952e-01 1.81148469e-01
-3.00158530e-01 4.01995659e-01 -2.40950495e-01 7.98746943e-02
1.89659357e-01 -4.92971152e-01 -2.80194074e-01 -2.23175898e-01
-3.86277229e-01 -1.97355971e-01 3.38042736e-01 9.86541569e-01
-7.52597973e-02 -1.65016845e-01 6.56786561e-01 -9.61655736e-01
-6.96612895e-01 2.11568624e-01 -7.35138178e-01 -4.44320351e-01
-5.75301126e-02 1.64994851e-01 4.21015173e-01 -5.83642423e-01
1.09682441e-01 1.02294564e+00 4.53519076e-01 -2.49067750e-02
-9.19344008e-01 -8.32704425e-01 -7.95721486e-02 9.26457793e-02
-1.00452936e+00 -1.21449701e-01 3.92742962e-01 -4.84233707e-01
8.24286163e-01 3.72038662e-01 9.33914602e-01 5.40490486e-02
9.45557952e-01 -1.18462361e-01 9.55907643e-01 -6.98671877e-01
1.15692548e-01 1.03581905e-01 -3.62694800e-01 4.87979859e-01
8.42996538e-01 -1.31885335e-01 -1.63093194e-01 3.03774297e-01
5.79302728e-01 2.79543877e-01 1.60749868e-01 -1.56004041e-01
-1.09759963e+00 8.29397380e-01 4.96549547e-01 5.60254991e-01
-6.89478934e-01 -5.09059615e-03 5.78772044e-03 1.04985557e-01
4.40817446e-01 4.41405594e-01 -9.01234686e-01 2.65043169e-01
-8.83507609e-01 5.61015680e-02 7.86252022e-01 2.91620195e-01
7.41511762e-01 3.14386874e-01 4.11105603e-01 4.87605125e-01
3.69054705e-01 1.34613073e+00 1.76520526e-01 -1.11498737e+00
1.31101593e-01 8.61782193e-01 5.66456556e-01 -1.52833581e+00
-8.09006989e-01 -7.90687025e-01 -1.06421041e+00 8.67481008e-02
2.81600237e-01 -3.51753026e-01 -9.06444192e-01 1.52881646e+00
1.30678415e-01 -5.50665379e-01 9.13268328e-02 5.47293425e-01
1.48210436e-01 1.15783036e+00 3.33344758e-01 -5.59836268e-01
9.60741282e-01 -7.56593049e-01 -9.50662374e-01 -2.79572845e-01
4.94835109e-01 -6.29313767e-01 6.40881002e-01 4.82459396e-01
-6.02050960e-01 -4.99297649e-01 -8.39679003e-01 5.69010019e-01
-6.17758274e-01 3.01967055e-01 7.81440437e-01 5.87206900e-01
-5.31044841e-01 6.56459451e-01 -8.43985796e-01 -6.86542392e-01
1.64905265e-02 1.68907866e-01 -1.63621545e-01 6.29158765e-02
-1.00411224e+00 1.39763486e+00 4.28019464e-01 3.20705950e-01
-2.71301746e-01 -6.41108930e-01 -6.68745697e-01 3.06899250e-01
5.10616563e-02 -3.48894507e-01 6.58943653e-01 -5.93917370e-01
-9.35858846e-01 2.16406226e-01 -1.07701793e-01 -2.67811656e-01
6.29596338e-02 -2.77866036e-01 -5.85175037e-01 -4.68160957e-01
2.70923227e-01 3.84510875e-01 1.74889136e-02 -1.03609407e+00
-1.16360831e+00 -7.84849465e-01 -7.86837995e-01 -4.20168713e-02
-6.86220527e-01 -7.28804246e-02 2.14879900e-01 -2.76238292e-01
2.93631613e-01 -1.12172902e+00 -6.84479773e-01 -3.52665335e-01
-2.95164343e-02 -6.09341785e-02 8.84013772e-01 -1.21245098e+00
1.05885041e+00 -1.84112549e+00 -9.92818400e-02 2.58502156e-01
-5.04837632e-01 2.41411299e-01 1.70439601e-01 5.64934313e-01
3.19072485e-01 1.76163167e-01 -1.01863205e-01 6.33730948e-01
-3.78885448e-01 2.69698530e-01 1.51872914e-02 4.09557283e-01
3.28598529e-01 2.29358852e-01 -9.31257904e-01 1.68931499e-01
5.08761287e-01 3.70419681e-01 1.90076921e-02 -1.56435538e-02
-2.32805014e-01 4.89905536e-01 -4.53214198e-01 6.95496082e-01
7.65120506e-01 -9.94404703e-02 4.56086904e-01 -6.60407729e-03
-6.53501034e-01 -7.31641799e-02 -9.81079817e-01 1.14300919e+00
-7.14153528e-01 3.76305819e-01 4.92950790e-02 -9.29737091e-01
1.28368974e+00 1.37829512e-01 6.54554784e-01 -1.31630909e+00
-2.00030118e-01 7.59078026e-01 -3.04680839e-02 -5.94748259e-01
3.79090905e-01 -1.59003571e-01 -1.06818089e-03 -4.77753468e-02
-4.53709900e-01 6.82866648e-02 3.68151814e-01 -2.73263991e-01
4.03193265e-01 2.12725282e-01 5.37611485e-01 -6.46676481e-01
5.55086434e-01 2.21221760e-01 7.86700547e-01 -1.24422573e-01
-5.04725762e-02 -4.08730447e-01 1.75348908e-01 -7.13534772e-01
-1.26499200e+00 -8.71682882e-01 -6.93881959e-02 8.63005459e-01
1.67276070e-01 5.41874111e-01 -2.45995075e-01 1.61941260e-01
2.04250485e-01 1.04596710e+00 -8.63591805e-02 2.53911968e-02
-4.41933870e-01 -1.05319476e+00 -1.70497715e-01 1.96224213e-01
7.62229204e-01 -1.00961518e+00 -1.11903131e+00 2.73773611e-01
-3.05251479e-01 -7.82718837e-01 1.78476378e-01 3.03563088e-01
-8.19653332e-01 -7.81592727e-01 -8.31169069e-01 -6.36993885e-01
4.79860425e-01 3.35023373e-01 9.39965904e-01 -1.74973294e-01
-9.64058340e-02 -3.96228313e-01 -3.11322182e-01 -9.69416976e-01
-4.14467782e-01 1.51573613e-01 4.18289334e-01 -5.54327071e-01
3.68780643e-01 -5.18360794e-01 -7.75496364e-01 1.13466956e-01
-3.42306405e-01 2.36118421e-01 6.25777423e-01 2.27743551e-01
3.28411907e-01 8.26135755e-01 9.80975330e-01 -3.13023150e-01
2.96428613e-03 -8.98778677e-01 -7.15589404e-01 4.32199925e-01
-9.35260952e-01 -4.10956815e-02 5.28761864e-01 9.73190293e-02
-1.10490048e+00 2.28197470e-01 2.90659219e-01 5.22715688e-01
8.37322548e-02 6.80712223e-01 -2.03164041e-01 4.09171581e-01
1.17517866e-01 -3.43472175e-02 6.49183616e-02 -6.37420714e-01
9.70922261e-02 7.37531841e-01 6.91204607e-01 -1.98992938e-02
7.06792831e-01 1.14573129e-01 5.49162507e-01 -1.06086099e+00
-6.31626904e-01 -6.07095718e-01 -4.86584395e-01 -4.47415680e-01
7.46339738e-01 -9.04652476e-01 -6.89546287e-01 5.32729626e-01
-9.72623289e-01 9.80319902e-02 1.81858823e-01 6.72389686e-01
-3.26440413e-03 -1.05169162e-01 2.24752054e-01 -1.32414925e+00
-5.83086431e-01 -5.87589860e-01 2.50575334e-01 7.52840579e-01
-1.16156958e-01 -8.43067408e-01 7.08065554e-03 8.37782472e-02
6.64748013e-01 7.26531804e-01 1.05111206e+00 -6.87450990e-02
-2.27785289e-01 2.62750871e-02 -3.93599719e-01 4.84746069e-01
9.48123515e-01 1.74079388e-01 -3.24637353e-01 -2.21954226e-01
2.09774338e-02 1.34057567e-01 7.81664848e-01 6.14876866e-01
5.78724861e-01 -7.52849638e-01 -3.65912378e-01 2.68140554e-01
1.89353573e+00 6.78872883e-01 6.27866149e-01 4.77857798e-01
4.12286758e-01 1.18677723e+00 9.84240353e-01 4.61816847e-01
1.93562597e-01 2.43035033e-01 7.53021002e-01 -2.58883178e-01
1.36369944e-01 -2.70623416e-02 2.65730947e-01 6.93669617e-01
-2.24756107e-01 -8.19277987e-02 -1.16739845e+00 1.00140238e+00
-1.60765851e+00 -1.09230828e+00 -4.69825625e-01 2.29100227e+00
3.28515649e-01 -1.70420825e-01 9.03038085e-02 3.45675588e-01
3.92235160e-01 -3.42765152e-01 -4.35551524e-01 -8.06477487e-01
-3.13946865e-02 -5.79858012e-02 1.32055569e+00 8.84623900e-02
-7.46993959e-01 5.81684291e-01 6.41366577e+00 3.46517801e-01
-1.40925503e+00 -4.07111913e-01 8.32650483e-01 3.68865758e-01
-3.38944972e-01 -2.50977762e-02 -6.33341551e-01 4.30822939e-01
1.25857580e+00 -3.74515116e-01 3.78566831e-01 6.66897655e-01
1.06371617e+00 -5.82147062e-01 -3.01878959e-01 3.02188098e-01
-4.66031969e-01 -1.20646119e+00 -2.04024881e-01 3.15189511e-01
9.32314157e-01 -1.13734089e-01 4.89899740e-02 -1.93723798e-01
3.52247477e-01 -1.07772398e+00 4.27006513e-01 8.12147677e-01
4.65189517e-01 -1.00599504e+00 8.61792922e-01 8.36759090e-01
-1.17491889e+00 -5.42156398e-01 -2.23376438e-01 -4.00073618e-01
1.42717138e-01 1.03586173e+00 -6.24513447e-01 6.93566084e-01
8.14835668e-01 2.09849715e-01 -1.58055410e-01 7.91352868e-01
3.45961563e-02 6.08037710e-01 -4.73273873e-01 1.30070686e-01
1.26662686e-01 -6.69374049e-01 -1.28325243e-02 9.64897633e-01
8.53584588e-01 -2.76961960e-02 -1.14185568e-02 4.09268826e-01
2.47463003e-01 5.95289171e-01 -4.92465436e-01 -1.65677205e-01
5.26762068e-01 9.07461405e-01 -7.41050839e-01 2.05369309e-01
-4.13275748e-01 5.22601843e-01 -4.52158064e-01 -1.22708203e-02
-5.27868927e-01 -4.06269014e-01 5.29144883e-01 1.83475807e-01
7.99999610e-02 -5.11318624e-01 -5.81892371e-01 -4.38248336e-01
-8.98615867e-02 -6.22007132e-01 -5.84384836e-02 -3.30915332e-01
-7.19699800e-01 -1.46878725e-02 -2.12459639e-01 -8.65765274e-01
-1.82913557e-01 -5.36175728e-01 -4.43299800e-01 1.13734138e+00
-1.69942129e+00 -1.25283670e+00 -1.57240689e-01 -3.20748270e-01
4.28945601e-01 -8.49043429e-02 9.13276374e-01 3.40878889e-02
-6.05370104e-01 -9.29182917e-02 7.53039539e-01 -3.42819363e-01
2.78541684e-01 -7.79379785e-01 1.09720089e-01 8.10688794e-01
-4.13086265e-01 1.90212235e-01 9.30155396e-01 -9.11099374e-01
-1.13728178e+00 -1.16374838e+00 1.32852173e+00 3.10789019e-01
6.40743315e-01 3.43632549e-01 -5.95271230e-01 1.27149656e-01
6.79722354e-02 -5.77000260e-01 6.71548545e-01 2.07323968e-01
-2.10858826e-02 -5.93082726e-01 -1.25510383e+00 2.83887744e-01
2.24051327e-01 1.99591428e-01 1.61096320e-01 1.81521982e-01
8.24396074e-01 3.08516771e-01 -1.01720607e+00 5.58484375e-01
1.20459557e+00 -7.44814277e-01 6.58161819e-01 -5.07579803e-01
5.14851034e-01 -1.47711650e-01 -4.05201823e-01 -1.33208179e+00
-5.94779670e-01 8.14220682e-02 8.04201961e-02 1.30096698e+00
6.42535388e-01 -4.63352561e-01 6.96797490e-01 8.22946787e-01
1.46800160e-01 -6.96388543e-01 -5.32478154e-01 -1.03048587e+00
2.59760737e-01 4.83440347e-02 8.56172383e-01 7.60509491e-01
-2.06515640e-01 4.20599580e-02 -6.89848483e-01 3.62113833e-01
6.45272553e-01 1.53291672e-01 6.09612942e-01 -1.49362147e+00
3.53163004e-01 -1.64750382e-01 -8.24302137e-02 -8.80035087e-02
-3.76072019e-01 -3.25683564e-01 -2.53705353e-01 -2.18985367e+00
3.88260633e-01 -6.19797766e-01 -3.77678365e-01 6.42586827e-01
-1.01176519e-02 2.11176947e-01 3.84460777e-01 2.03541681e-01
2.88634568e-01 2.72646934e-01 1.21079469e+00 -1.64060414e-01
-6.34628773e-01 5.77460304e-02 -9.16252315e-01 7.56264091e-01
1.21151388e+00 -3.74242812e-01 -1.96536705e-01 -4.99236405e-01
6.47314847e-01 -3.86338052e-03 -6.59139082e-02 -1.17811608e+00
-1.77603468e-01 -1.29463029e+00 5.91088355e-01 -7.56300390e-01
-1.12931177e-01 -1.25654411e+00 8.28121424e-01 1.21360290e+00
2.58995503e-01 -1.65006042e-01 1.64867863e-01 8.53720903e-02
1.08980253e-01 -9.70477015e-02 8.09314251e-01 -2.11865809e-02
-6.06364846e-01 -1.34981737e-01 -4.77406800e-01 -8.68069708e-01
1.28681839e+00 -1.59603238e-01 -2.78027654e-01 -1.20910220e-01
-4.08887744e-01 5.76486886e-01 5.29949784e-01 4.01490033e-01
1.02426030e-01 -1.00978208e+00 -9.11017358e-01 -1.78476691e-01
-1.59465089e-01 -2.90639430e-01 1.41709954e-01 3.93291295e-01
-9.24420536e-01 8.39839995e-01 -8.15390527e-01 8.91529024e-02
-9.46764231e-01 4.85963553e-01 1.62341803e-01 -6.39271796e-01
1.26528874e-01 2.84196585e-01 -1.20242357e-01 -7.52205998e-02
-2.04805180e-01 -2.28869468e-01 -4.19511110e-01 1.18842378e-01
6.21727705e-01 7.90016234e-01 -7.04588369e-02 -6.96838379e-01
-4.18802351e-01 4.94390070e-01 4.08816427e-01 2.50163227e-01
1.58363771e+00 -4.11724508e-01 -3.94348383e-01 5.46301603e-01
7.63248563e-01 2.55980372e-01 -7.67693937e-01 3.79725903e-01
1.10075556e-01 -4.01888490e-01 2.54735827e-01 -1.13256431e+00
-1.11245930e+00 5.79243243e-01 8.77691567e-01 3.46689075e-01
1.48482776e+00 -4.62160707e-01 9.75037277e-01 2.15238556e-01
5.05821586e-01 -9.51144040e-01 -3.88331652e-01 1.89163923e-01
7.34492660e-01 -1.27212751e+00 2.60579854e-01 -3.44557375e-01
-2.25468725e-01 1.08557165e+00 1.95495978e-01 2.30692565e-01
7.15877593e-01 -2.70780884e-02 -6.91579804e-02 1.73364520e-01
-4.35519785e-01 -1.77784398e-01 -2.75729537e-01 7.68578053e-01
6.93290830e-01 6.41763210e-01 -7.03848779e-01 3.12891334e-01
3.36174481e-02 1.27317935e-01 2.76661158e-01 5.67264676e-01
-1.23967755e+00 -1.27707279e+00 -7.16068685e-01 7.99096763e-01
-5.41137099e-01 1.12217918e-01 -9.17783827e-02 4.79757130e-01
6.58506155e-01 1.38558781e+00 1.86302751e-01 -6.26307428e-02
2.59411752e-01 -1.13143563e-01 2.49211684e-01 -2.47926444e-01
-5.66876344e-02 -1.49462253e-01 8.85321945e-02 -2.01530010e-01
-6.17922425e-01 -7.40888476e-01 -1.45125282e+00 -8.60395312e-01
-2.58729428e-01 1.63342729e-01 1.41785347e+00 8.48347306e-01
1.26512989e-01 3.50502461e-01 1.12414551e+00 -5.94121456e-01
-4.03035283e-01 -1.00052309e+00 -7.54492283e-01 -5.46961352e-02
1.47020683e-01 -6.04600191e-01 -1.01788536e-01 1.55036807e-01]
|
[9.369535446166992, -1.5701687335968018]
|
66fc6f5f-d535-4f90-b29b-07b4007d6401
|
cross-modal-face-and-voice-style-transfer
|
2302.13838
| null |
https://arxiv.org/abs/2302.13838v2
|
https://arxiv.org/pdf/2302.13838v2.pdf
|
Cross-modal Face- and Voice-style Transfer
|
Image-to-image translation and voice conversion enable the generation of a new facial image and voice while maintaining some of the semantics such as a pose in an image and linguistic content in audio, respectively. They can aid in the content-creation process in many applications. However, as they are limited to the conversion within each modality, matching the impression of the generated face and voice remains an open question. We propose a cross-modal style transfer framework called XFaVoT that jointly learns four tasks: image translation and voice conversion tasks with audio or image guidance, which enables the generation of ``face that matches given voice" and ``voice that matches given face", and intra-modality translation tasks with a single framework. Experimental results on multiple datasets show that XFaVoT achieves cross-modal style translation of image and voice, outperforming baselines in terms of quality, diversity, and face-voice correspondence.
|
['Yuki Mitsufuji', 'Mayank K. Singh', 'Naoya Takahashi']
|
2023-02-27
| null | null | null | null |
['voice-conversion', 'open-question', 'voice-conversion']
|
['audio', 'natural-language-processing', 'speech']
|
[ 3.97837460e-01 1.03354387e-01 -7.09765479e-02 -4.86516684e-01
-1.03717935e+00 -8.17988575e-01 7.89613366e-01 -5.39169312e-01
1.72142629e-02 5.47986925e-01 4.25254047e-01 1.44105554e-01
4.38655943e-01 -5.84018171e-01 -1.06118107e+00 -5.46549678e-01
5.66671848e-01 3.83764058e-01 -2.48407930e-01 6.02102019e-02
-2.43474349e-01 3.46778512e-01 -1.79401982e+00 7.27300584e-01
5.16022980e-01 1.36012411e+00 1.91228077e-01 6.00704253e-01
-2.31062770e-01 4.27464545e-01 -1.66617915e-01 -8.21362078e-01
2.58201152e-01 -8.42363596e-01 -6.87292516e-01 5.84688902e-01
1.11158597e+00 -6.06418014e-01 -2.03586221e-01 8.92152309e-01
7.44400263e-01 -4.41283807e-02 6.69464290e-01 -1.48617339e+00
-9.95491982e-01 2.40199357e-01 -4.50551540e-01 -4.75375682e-01
7.19011486e-01 3.63487542e-01 8.03851128e-01 -1.32918882e+00
9.64443326e-01 1.66282988e+00 4.53329563e-01 9.53971744e-01
-1.49740911e+00 -8.82635653e-01 -2.85124481e-01 -2.49373689e-01
-1.41551673e+00 -1.21110582e+00 4.67208028e-01 -3.36201757e-01
3.52949768e-01 2.76396185e-01 5.47592580e-01 1.09671891e+00
4.90741730e-02 5.42897999e-01 1.10734403e+00 -2.59024650e-01
-1.23551808e-01 3.78408283e-01 -1.07087934e+00 7.17000246e-01
-4.83845472e-01 1.53847426e-01 -1.01826119e+00 9.86228809e-02
1.05340183e+00 -3.50097865e-01 -2.13634431e-01 -1.95950642e-01
-1.28173602e+00 6.68785870e-01 3.25264722e-01 -5.32545801e-03
-2.33126372e-01 2.35866860e-01 2.21941888e-01 4.31062311e-01
4.46073264e-01 2.36417547e-01 -1.60255015e-01 5.98127507e-02
-9.25745130e-01 2.25643009e-01 4.33792472e-01 1.28999329e+00
6.88533843e-01 1.65751949e-01 -3.64241064e-01 9.25286055e-01
1.34195954e-01 1.03947771e+00 3.58029336e-01 -1.45278466e+00
1.85560957e-01 1.55180603e-01 4.37662750e-02 -6.99841380e-01
1.31830007e-01 -1.11875623e-01 -6.59487605e-01 1.65224761e-01
1.46300402e-02 -1.33569032e-01 -8.53470743e-01 2.10514808e+00
5.69187641e-01 2.36288056e-01 -1.91322528e-03 8.71605217e-01
1.25190461e+00 6.81160510e-01 -6.44402504e-02 -4.07787591e-01
1.41660452e+00 -9.22952175e-01 -9.29334044e-01 -1.21480048e-01
-1.77866161e-01 -1.40084934e+00 1.27276742e+00 -1.05549343e-01
-1.57849479e+00 -9.42614019e-01 -3.89606118e-01 -2.43489340e-01
2.79969752e-01 1.24689795e-01 1.84186861e-01 3.38156044e-01
-1.52037120e+00 2.26941705e-01 -1.82751983e-01 -4.59449947e-01
4.52459872e-01 1.07143626e-01 -7.76279032e-01 -2.55256355e-01
-7.64919877e-01 4.03211594e-01 -3.43324095e-02 -4.03099358e-01
-1.17051625e+00 -9.69692767e-01 -9.19646144e-01 -2.34282851e-01
7.64478091e-03 -1.19705713e+00 1.46253502e+00 -1.41122782e+00
-1.73808360e+00 1.27376509e+00 -4.04860914e-01 1.92817245e-02
7.35265315e-01 -2.12267134e-02 -5.01468480e-01 2.86971688e-01
4.02547717e-01 1.52012217e+00 1.44535112e+00 -1.42066073e+00
-8.10407639e-01 -3.07792038e-01 -2.32143626e-01 5.02479196e-01
-2.40164101e-01 9.25016850e-02 -8.78380358e-01 -6.89326525e-01
-4.30839472e-02 -9.68993127e-01 4.36804205e-01 5.65529227e-01
-2.95702517e-01 3.82955298e-02 8.71827960e-01 -8.01853955e-01
5.51303685e-01 -2.53135371e+00 2.21543103e-01 -1.81018502e-01
-9.44436155e-03 -8.62613469e-02 -6.36019826e-01 2.68744469e-01
-2.86934488e-02 -9.59621444e-02 1.57572459e-02 -5.94261289e-01
-4.54629287e-02 2.45450158e-02 -5.58440804e-01 2.71274477e-01
2.85485297e-01 1.06090462e+00 -8.50890279e-01 -7.58052826e-01
-1.27274349e-01 7.31448948e-01 -7.99456716e-01 6.86705232e-01
-2.45920911e-01 8.93706918e-01 2.20588729e-04 7.17102230e-01
6.10810339e-01 -1.15631206e-03 -1.07933089e-01 -4.94872272e-01
2.28431448e-01 -8.60218853e-02 -8.45924199e-01 1.98194122e+00
-6.54568791e-01 5.75209856e-01 4.79130030e-01 -3.43675390e-02
8.11734438e-01 7.42014289e-01 5.19983351e-01 -8.18128765e-01
-1.98690984e-02 3.07371080e-01 -3.01491976e-01 -4.03446436e-01
4.22095895e-01 -5.68862975e-01 5.30222505e-02 6.82738364e-01
3.55667681e-01 -6.58670843e-01 -1.71814151e-02 1.20664351e-01
4.63584065e-01 2.19049633e-01 -3.06566805e-01 1.37004465e-01
3.41267854e-01 -3.76079589e-01 2.46949255e-01 2.69146979e-01
1.81361195e-02 1.08767402e+00 1.35930032e-01 1.34345084e-01
-1.20071340e+00 -1.58885479e+00 -1.25480801e-01 1.39498222e+00
-6.88832626e-02 -2.24786267e-01 -1.08220983e+00 -4.39976960e-01
-5.84868416e-02 5.03487349e-01 -3.77567679e-01 -5.35343178e-02
-1.93533301e-01 1.95537239e-01 6.29775107e-01 3.15849394e-01
4.03712124e-01 -1.27449977e+00 3.97515018e-03 -1.06526531e-01
-7.37386346e-01 -1.59775555e+00 -1.30813932e+00 -7.34126270e-01
-6.44718170e-01 -7.29320288e-01 -8.71724248e-01 -1.02577960e+00
8.83543789e-01 3.13891351e-01 1.09877908e+00 -1.66077852e-01
-3.59149307e-01 8.04406106e-01 -1.07128985e-01 -5.69387637e-02
-8.11680317e-01 -3.97011667e-01 2.18539372e-01 4.84421015e-01
-4.90589619e-01 -8.14930499e-01 -7.78341770e-01 3.45589995e-01
-1.03435981e+00 3.44324589e-01 6.54057622e-01 8.15285325e-01
6.54985666e-01 -2.88785338e-01 5.42931080e-01 -5.88843882e-01
6.95244014e-01 -1.35890052e-01 -2.23688155e-01 5.80147319e-02
-3.98573756e-01 -3.12434971e-01 3.75693172e-01 -5.19423187e-01
-1.13589883e+00 1.71760276e-01 -1.04954757e-01 -8.91598403e-01
-1.62325308e-01 5.90602774e-03 -5.10351956e-01 7.77986348e-02
5.21866024e-01 5.42739749e-01 4.45790470e-01 -2.93346673e-01
8.40848744e-01 8.16375911e-01 1.15281141e+00 -5.52527189e-01
7.91283786e-01 6.36724234e-01 -2.84173489e-01 -6.22241974e-01
-5.34769058e-01 -7.83144161e-02 -5.24909616e-01 -4.32131797e-01
9.16343868e-01 -1.28970706e+00 -6.32771313e-01 4.19979036e-01
-1.40650654e+00 -1.69378996e-01 -5.48578143e-01 1.58076182e-01
-9.88992453e-01 3.56162712e-02 -4.93635654e-01 -4.56915438e-01
-4.18646455e-01 -1.28317750e+00 1.62824726e+00 1.01138495e-01
-2.58892655e-01 -6.53406739e-01 -1.38150513e-01 6.56522334e-01
2.81840235e-01 -6.36244798e-03 8.23873997e-01 5.88758811e-02
-6.09973788e-01 1.17255971e-01 -3.49908113e-01 3.60050976e-01
3.97991776e-01 -1.49409071e-01 -1.23540747e+00 -5.58431864e-01
-3.83102596e-01 -5.10602713e-01 4.83596563e-01 1.60960793e-01
7.00731695e-01 -5.24266601e-01 -4.23454233e-02 7.24125803e-01
9.28590834e-01 1.44673290e-03 5.55800736e-01 -5.13519526e-01
5.37171423e-01 8.46913278e-01 4.60450113e-01 3.79090637e-01
3.30925226e-01 9.08390701e-01 3.02758366e-01 -2.89949715e-01
-9.37579155e-01 -7.60751486e-01 7.42964268e-01 6.37346029e-01
3.10490608e-01 -3.39620449e-02 -3.46881777e-01 6.82657480e-01
-1.45662773e+00 -9.17134523e-01 2.76461691e-01 2.01422644e+00
1.25950885e+00 -5.98062575e-01 1.43840775e-01 -3.39110911e-01
9.48201418e-01 -1.10790208e-01 -5.87543666e-01 -3.08178008e-01
-1.39994025e-01 1.62492335e-01 -1.57772496e-01 6.38087690e-01
-7.76596010e-01 1.15317595e+00 6.30537128e+00 8.34281623e-01
-1.46681499e+00 4.32582945e-01 6.37381673e-01 -2.38102436e-01
-5.52253842e-01 -3.00584406e-01 -4.70413953e-01 3.03740084e-01
6.34381771e-01 -1.37141541e-01 8.31047416e-01 4.26643550e-01
3.68348241e-01 2.37784013e-01 -1.39321125e+00 1.38429236e+00
2.94418186e-01 -1.24834466e+00 5.27867675e-01 1.98336374e-02
9.70106125e-01 -3.62724096e-01 5.54566562e-01 -6.03482872e-02
4.78921719e-02 -1.14784336e+00 1.07572830e+00 5.77635586e-01
1.78725326e+00 -5.66503763e-01 1.02415577e-01 -1.90106817e-02
-1.15154505e+00 2.37653986e-01 -4.35265992e-03 5.03758252e-01
2.09264889e-01 1.66044921e-01 -1.02987039e+00 3.52174670e-01
8.46199989e-01 3.49413514e-01 -3.20427448e-01 5.43313205e-01
-2.10986122e-01 2.03146592e-01 3.21914479e-02 7.75961637e-01
-2.10789621e-01 -4.94349077e-02 6.67218924e-01 9.16350305e-01
6.20967746e-01 -2.62666494e-01 1.05253331e-01 1.04779696e+00
-5.30903041e-01 4.53641206e-01 -6.75842285e-01 -1.77143991e-01
6.73647881e-01 1.31125355e+00 -3.18036973e-01 -1.40150547e-01
-2.40555629e-01 1.35546529e+00 -2.06005186e-01 3.33074450e-01
-7.34027684e-01 1.09567204e-02 9.66636717e-01 5.11421502e-01
1.53670251e-01 3.25315744e-01 -8.87416750e-02 -8.71241033e-01
1.28769010e-01 -1.06705523e+00 5.80639057e-02 -1.28771877e+00
-1.23692572e+00 8.65469933e-01 -3.29135180e-01 -1.25826180e+00
-5.11543214e-01 -2.28972480e-01 -4.44868088e-01 7.88722217e-01
-1.30892956e+00 -1.78776586e+00 -4.21754092e-01 1.05439639e+00
7.25474358e-01 -3.97593528e-01 8.31157565e-01 4.73314852e-01
3.48867066e-02 9.50317740e-01 -7.81462267e-02 7.50961527e-02
1.41520405e+00 -6.56694293e-01 2.10597306e-01 4.68908429e-01
1.62990481e-01 2.84807205e-01 6.82478309e-01 -5.56259215e-01
-1.44689381e+00 -1.49056864e+00 9.05452073e-01 -2.72604495e-01
2.13725448e-01 -2.69449055e-01 -5.09639740e-01 5.49162447e-01
5.66785455e-01 1.93189874e-01 6.50281250e-01 -3.34128022e-01
-4.85265046e-01 -3.20222020e-01 -1.27934062e+00 7.12481022e-01
1.11592460e+00 -1.01884449e+00 -9.71642807e-02 1.87411815e-01
9.10905361e-01 -5.76515257e-01 -8.66054237e-01 2.24832729e-01
8.19097400e-01 -8.59954357e-01 1.07251120e+00 -5.42552531e-01
7.87805200e-01 -3.32509339e-01 -4.24813300e-01 -1.29352140e+00
-9.56540257e-02 -1.01192427e+00 2.79617637e-01 1.84381437e+00
3.24736834e-01 -2.36647323e-01 2.78417051e-01 3.67400676e-01
-1.58789948e-01 -2.24336326e-01 -1.01176584e+00 -5.87611079e-01
-9.55955088e-02 -3.34928185e-01 8.29523683e-01 9.52424526e-01
-3.98745507e-01 5.25251746e-01 -5.60841918e-01 -8.18376020e-02
5.25141358e-01 3.14705789e-01 1.11559534e+00 -8.40850174e-01
-2.75067389e-01 -3.08948606e-01 -2.13256255e-01 -8.17476749e-01
4.94367778e-01 -1.10153317e+00 9.81449038e-02 -1.30796206e+00
3.72970372e-01 -6.42287731e-02 2.84439385e-01 5.02102673e-01
2.19953343e-01 8.62080574e-01 4.19334531e-01 3.77001286e-01
-3.61296207e-01 7.37694323e-01 1.80557215e+00 -8.67317542e-02
-1.29167689e-02 -1.75750047e-01 -8.35696816e-01 5.48234701e-01
2.16828316e-01 -2.20979646e-01 -5.03975093e-01 -6.08911693e-01
7.77523452e-03 4.57437336e-01 4.47462797e-01 -7.04296708e-01
1.01208724e-01 -2.29553178e-01 4.21397865e-01 -1.75550595e-01
6.98725522e-01 -6.63577080e-01 4.37593699e-01 1.34876534e-01
-4.02674168e-01 1.27506703e-01 2.76285648e-01 4.44439918e-01
-3.70624453e-01 3.68469000e-01 1.11220980e+00 2.63892300e-02
-3.48374248e-01 5.18853784e-01 -2.13483833e-02 1.88949168e-01
8.38685393e-01 -2.17253089e-01 -2.82374360e-02 -9.79468942e-01
-9.58013117e-01 -3.06315832e-02 7.00363159e-01 8.50522935e-01
8.06743741e-01 -1.95118356e+00 -1.02447116e+00 3.23224485e-01
2.47130528e-01 -1.58259630e-01 3.48508090e-01 8.38986516e-01
-1.67013288e-01 2.01442704e-01 -3.90734613e-01 -8.81643355e-01
-1.45510936e+00 2.95026004e-01 4.46539849e-01 4.43871081e-01
-2.94098854e-01 7.90155053e-01 7.72173405e-01 -6.07379138e-01
1.55651346e-01 3.51763010e-01 3.40879917e-01 1.00028433e-01
5.00209033e-01 -2.12083850e-02 -1.03676483e-01 -1.12534118e+00
2.87202429e-02 7.04418719e-01 1.84936360e-01 -6.25382900e-01
1.17595744e+00 -5.34522772e-01 -3.21431190e-01 2.10838154e-01
1.14027846e+00 1.88914999e-01 -1.44322777e+00 -5.50507486e-01
-8.29345644e-01 -7.84863770e-01 -8.67054313e-02 -9.46407020e-01
-1.26900828e+00 9.17470753e-01 8.20183575e-01 -4.12321687e-01
1.39861727e+00 3.54863346e-01 9.65009093e-01 -1.08420625e-01
1.69177994e-01 -8.33988965e-01 5.91324508e-01 2.83610493e-01
1.44965374e+00 -1.07733619e+00 -5.49560487e-01 -4.93228406e-01
-7.71393180e-01 8.28887761e-01 7.09896743e-01 4.06604916e-01
3.78799230e-01 1.13036886e-01 3.81326467e-01 2.06742093e-01
-7.46262193e-01 -2.32793450e-01 5.97047806e-01 8.24329674e-01
6.50617719e-01 -6.02848865e-02 2.23596320e-01 4.03664738e-01
-6.30122066e-01 4.51486111e-02 1.32596329e-01 4.78856921e-01
-1.51542991e-01 -1.12159264e+00 -5.66498220e-01 1.75940245e-02
-3.97154659e-01 -1.92494497e-01 -7.44019866e-01 4.41015005e-01
3.38251978e-01 9.54004884e-01 2.71380126e-01 -2.91448742e-01
2.78653026e-01 1.73676163e-01 7.61782110e-01 -7.12783575e-01
-4.08279836e-01 5.28360486e-01 -7.85404146e-02 -6.18745148e-01
-1.47588715e-01 -7.13167369e-01 -1.06991589e+00 -5.09373963e-01
-7.27434754e-02 -7.82135054e-02 5.81643581e-01 7.69931614e-01
7.64403462e-01 3.51108938e-01 9.15645301e-01 -8.49610627e-01
-2.26951957e-01 -8.30682456e-01 -6.15973055e-01 7.61245966e-01
3.50200415e-01 -3.50515574e-01 9.24767032e-02 7.49917030e-01]
|
[12.710094451904297, -0.23396185040473938]
|
bcb92066-6e58-4a2d-944f-88021f9508e9
|
learning-based-repetitive-precision-motion
|
2111.10246
| null |
https://arxiv.org/abs/2111.10246v1
|
https://arxiv.org/pdf/2111.10246v1.pdf
|
Learning-Based Repetitive Precision Motion Control with Mismatch Compensation
|
Learning-based control methods utilize run-time data from the underlying process to improve the controller performance under model mismatch and unmodeled disturbances. This is beneficial for optimizing industrial processes, where the dynamics are difficult to model, and the repetitive nature of the process can be exploited. In this work, we develop an iterative approach for repetitive precision motion control problems where the objective is to follow a reference geometry with minimal tracking error. Our method utilizes a nominal model of the process and learns the mismatch using Gaussian Process Regression (GPR). The control input and the GPR data are updated after each iteration to improve the performance in a run-to-run fashion. We provide a preliminary convergence analysis, implementation details of the proposed controller for minimizing different error types, and a case study where we demonstrate improved tracking performance with simulation and experimental results.
|
['John Lygeros', 'Alisa Rupenyan', 'Dawn M. Tilbury', 'Kira Barton', 'Efe C. Balta']
|
2021-11-19
| null | null | null | null |
['gpr', 'gpr']
|
['computer-vision', 'miscellaneous']
|
[ 3.04931700e-01 1.65101260e-01 -1.69940531e-01 2.87759691e-01
-3.63388658e-01 -5.75151205e-01 3.98124367e-01 1.50909677e-01
-2.62758005e-02 7.70623922e-01 -4.48669434e-01 -2.90807635e-01
-5.11573434e-01 -3.77298564e-01 -7.09918082e-01 -9.27608073e-01
6.97686300e-02 2.94437498e-01 -4.23452929e-02 -6.37432262e-02
4.65996474e-01 5.39449751e-01 -9.27235782e-01 -5.25521696e-01
7.42895246e-01 7.71476686e-01 6.54144585e-01 1.09616816e+00
3.78610551e-01 5.47865748e-01 -2.32116431e-01 4.05357301e-01
5.42690217e-01 -2.63318181e-01 -2.01530725e-01 5.76583564e-01
-1.41958594e-01 5.02335094e-02 -2.80111820e-01 1.08793116e+00
5.81628144e-01 7.82185256e-01 5.83597302e-01 -9.80913877e-01
-1.13498650e-01 8.29751566e-02 -6.44800723e-01 -1.71180084e-01
6.94543645e-02 7.00978160e-01 3.89580280e-01 -6.72709942e-01
3.41492742e-01 1.41519618e+00 6.36999190e-01 2.95550883e-01
-1.43193948e+00 -4.43550617e-01 2.93287963e-01 -4.13443089e-01
-1.31971133e+00 -1.62859574e-01 5.20539582e-01 -6.56434298e-01
5.30643165e-01 -3.91821414e-02 3.90563816e-01 7.09225118e-01
9.49568868e-01 3.36314648e-01 8.19658399e-01 -1.83214411e-01
4.45777535e-01 -1.97046638e-01 -1.41937688e-01 6.00433946e-01
5.06712615e-01 4.80299622e-01 6.40206262e-02 4.69095958e-03
1.52778184e+00 1.79328591e-01 -4.91379619e-01 -8.68960261e-01
-1.02460337e+00 7.84316361e-01 1.70402393e-01 -4.90740314e-02
-7.94445097e-01 2.80551583e-01 2.18034625e-01 4.30707365e-01
-1.08068101e-01 9.58316088e-01 -5.76291084e-01 -2.93751597e-01
-5.55200636e-01 5.71068108e-01 1.06771195e+00 1.33551574e+00
3.80061358e-01 5.24839461e-01 -8.40951353e-02 5.77994168e-01
5.80969691e-01 5.76590061e-01 4.17655587e-01 -1.21628523e+00
4.89562571e-01 2.68620670e-01 6.50413632e-01 -9.54295695e-01
-3.22084457e-01 -3.19343239e-01 -8.65333259e-01 7.07943022e-01
5.18929183e-01 -6.30155802e-01 -9.56298947e-01 1.21718192e+00
2.39790320e-01 2.81217694e-01 1.69177592e-01 7.49117851e-01
-4.26518977e-01 8.13010573e-01 -1.36098728e-01 -7.76700318e-01
9.36538339e-01 -8.94035220e-01 -1.09969938e+00 -7.09443837e-02
3.50136638e-01 -9.44730639e-01 9.63610649e-01 5.50482452e-01
-1.18379772e+00 -7.34212995e-01 -1.00417399e+00 6.74821973e-01
1.12599455e-01 2.78454363e-01 -1.97849691e-01 3.39200824e-01
-5.12345731e-01 1.10806859e+00 -1.19169879e+00 -3.41228366e-01
-1.13783717e-01 4.04746413e-01 2.12883368e-01 4.08455729e-01
-5.90089083e-01 1.12101138e+00 5.52875936e-01 4.07372952e-01
-1.00674653e+00 -1.02387869e+00 -7.99850762e-01 -1.81420773e-01
9.68297064e-01 -7.08010077e-01 1.63539255e+00 -4.89338577e-01
-2.35106826e+00 -2.41420671e-01 -9.92305130e-02 -2.36760080e-01
7.09148407e-01 -6.13921463e-01 -3.93724144e-02 -1.12143137e-01
-3.93199712e-01 -1.63487345e-01 1.26718664e+00 -1.31415522e+00
-7.70292163e-01 -1.63777005e-02 -3.31698477e-01 1.93591177e-01
3.31936240e-01 -5.28337777e-01 -2.66402215e-01 -5.90187132e-01
1.39298677e-01 -1.25871778e+00 -8.99548352e-01 -1.29814178e-01
-4.38790321e-01 2.84185708e-01 1.12760472e+00 -3.89434725e-01
1.10829151e+00 -2.03739214e+00 3.02530855e-01 4.68975127e-01
-2.15541080e-01 3.06459725e-01 1.09951444e-01 5.77533185e-01
-1.38042783e-02 -2.26394579e-01 -9.12957452e-03 1.27608240e-01
-9.08191279e-02 7.13739768e-02 -2.92988569e-01 4.72692162e-01
4.07802522e-01 5.66952527e-01 -8.93192589e-01 3.81421335e-02
6.04504585e-01 2.43852973e-01 -1.94278672e-01 3.94092947e-01
-2.52303094e-01 8.11158121e-01 -8.33361387e-01 3.05363387e-01
2.95997888e-01 -2.77551804e-02 1.51225314e-01 -3.47993970e-01
-5.50769210e-01 -5.02072573e-01 -1.79935181e+00 1.05776763e+00
-8.28828990e-01 1.78802237e-01 7.94982731e-01 -8.78952742e-01
1.22859943e+00 4.89750743e-01 6.11581206e-01 -1.02708116e-01
4.20174509e-01 1.05142631e-01 -3.34808305e-02 -3.73793334e-01
3.44157189e-01 -1.77152336e-01 2.71367252e-01 -2.84301229e-02
-3.14407587e-01 -9.59075570e-01 1.45166330e-02 -5.51966310e-01
9.48509872e-01 2.29069963e-01 6.03681266e-01 -4.13201690e-01
8.18399549e-01 7.28085712e-02 6.17334783e-01 5.05164504e-01
-1.40294433e-01 2.83970535e-01 1.89745009e-01 -1.00665346e-01
-1.24226403e+00 -8.04260612e-01 1.77551493e-01 3.77397865e-01
3.82899612e-01 1.08763687e-01 -3.09286028e-01 1.38214067e-01
1.97922096e-01 8.85771155e-01 -2.67090976e-01 -4.06904429e-01
-7.78977633e-01 -3.63902301e-01 -2.22200900e-01 6.45902574e-01
1.69010088e-01 -4.51347500e-01 -7.53439009e-01 8.41312826e-01
7.11862922e-01 -9.32552934e-01 -5.61389923e-01 1.84376284e-01
-1.16285908e+00 -1.34656787e+00 -5.59484601e-01 -8.52070689e-01
6.34182990e-01 -6.82843477e-02 3.50606561e-01 -4.44595605e-01
-3.09651643e-01 6.03739917e-01 2.60458946e-01 -6.22920930e-01
-5.52935779e-01 -3.19764584e-01 4.08441514e-01 -2.07603768e-01
-4.77713019e-01 -9.76093784e-02 -3.65516782e-01 6.96523726e-01
-4.01121140e-01 -2.13667929e-01 6.66310489e-01 8.46984029e-01
8.48255455e-01 6.58242822e-01 3.42280388e-01 -4.07725900e-01
9.73804235e-01 -4.07897756e-02 -1.24582422e+00 5.49819507e-02
-6.36578381e-01 -1.44061223e-01 9.22037840e-01 -8.87323439e-01
-1.21526921e+00 4.92346972e-01 5.02947152e-01 -9.86725032e-01
7.02889785e-02 5.12308061e-01 -1.70563146e-01 -5.42040952e-02
2.19997793e-01 4.27847542e-02 6.74697399e-01 -3.89682800e-01
2.77912289e-01 2.62388051e-01 5.08430660e-01 -7.63103724e-01
9.71751690e-01 -8.48053694e-02 6.11518204e-01 -9.63776529e-01
-3.00837606e-01 -5.89868963e-01 -7.17146337e-01 -1.93609983e-01
6.81078076e-01 -7.95867741e-01 -1.19001603e+00 3.07772785e-01
-7.63118923e-01 -7.08034158e-01 -5.34040868e-01 8.59232068e-01
-1.22458410e+00 -3.87986121e-03 -6.69231951e-01 -1.35868502e+00
-2.21377090e-01 -1.28585994e+00 8.77195358e-01 4.66931283e-01
-3.62487078e-01 -1.17521453e+00 7.01791421e-02 -4.13093507e-01
4.90108371e-01 4.99244750e-01 7.63874412e-01 -3.60159397e-01
-4.97138470e-01 -3.43801260e-01 3.22787106e-01 1.71456948e-01
5.87799191e-01 1.31290421e-01 -3.69537026e-01 -7.67320454e-01
4.62013870e-01 1.73462212e-01 1.05091393e-01 7.54232526e-01
7.17760265e-01 -2.64147967e-01 -5.67732573e-01 1.14762291e-01
1.58588743e+00 7.23624885e-01 1.46227688e-01 3.92891705e-01
9.29633439e-01 5.04028976e-01 1.11097527e+00 5.38290679e-01
-2.89129734e-01 4.97411817e-01 2.66607344e-01 3.03505093e-01
4.53184903e-01 -8.94038156e-02 4.31641370e-01 6.14387333e-01
-1.26425549e-01 4.33496721e-02 -7.22826660e-01 3.78245145e-01
-2.04296398e+00 -7.98038304e-01 -2.33382180e-01 2.52335382e+00
5.61308324e-01 -1.56707801e-02 -1.77417710e-01 6.05042055e-02
1.11492264e+00 -3.48077685e-01 -7.40034461e-01 -4.61338669e-01
5.10685980e-01 -9.43326354e-02 9.43613768e-01 7.96934843e-01
-1.17243505e+00 4.69295055e-01 6.21651745e+00 5.69714963e-01
-1.34759426e+00 -5.03924131e-01 3.19544286e-01 1.09466739e-01
5.94128370e-01 -2.28602991e-01 -8.67419600e-01 2.48301655e-01
1.11106443e+00 -6.88058019e-01 4.64098454e-01 1.02505314e+00
1.12553513e+00 3.80997844e-02 -1.24054754e+00 7.54812121e-01
-5.13352692e-01 -9.69241560e-01 -5.04662991e-01 5.17645292e-02
9.57250893e-01 -5.70596099e-01 1.07018888e-01 3.03460628e-01
5.89595318e-01 -7.17141628e-01 5.67702472e-01 8.22328508e-01
1.37549743e-01 -8.11538517e-01 7.32904077e-01 4.25963521e-01
-1.47768211e+00 -5.58217824e-01 -3.95348370e-01 1.21503491e-02
7.23501325e-01 3.52595508e-01 -8.67329299e-01 7.52767622e-01
1.19193802e-02 7.10323036e-01 6.58681169e-02 1.31516969e+00
2.87320949e-02 3.51045310e-01 -3.20803285e-01 -2.26648897e-01
3.81263673e-01 -6.52489126e-01 1.11333585e+00 8.09732795e-01
6.76826537e-01 -1.36751279e-01 8.23555946e-01 8.90236497e-01
6.60341024e-01 2.28126310e-02 -7.65315235e-01 -1.90068096e-01
2.91680843e-01 9.41929817e-01 -4.00746971e-01 -7.03011379e-02
5.33521846e-02 5.99861801e-01 -3.07743520e-01 4.99030918e-01
-6.65208161e-01 -5.43667793e-01 8.02080274e-01 -4.74923737e-02
5.69402099e-01 -7.94679761e-01 -3.35262924e-01 -4.32428241e-01
-1.44229993e-01 -8.00313830e-01 -8.11974183e-02 -3.52232456e-01
-1.25493658e+00 -1.54013187e-01 8.29594061e-02 -1.51554215e+00
-6.64525270e-01 -6.91149831e-01 -7.33223677e-01 9.66993868e-01
-1.14419758e+00 -5.56966007e-01 4.07525152e-02 2.17160910e-01
1.01491857e+00 -2.69357860e-01 5.11255205e-01 -1.78680927e-01
-7.27992594e-01 -1.68444246e-01 6.73030257e-01 -4.57157612e-01
5.70573926e-01 -1.37334311e+00 -1.47858962e-01 5.68603039e-01
-7.73334265e-01 8.03676665e-01 1.31059504e+00 -9.65282440e-01
-2.01572514e+00 -1.67102301e+00 -8.27260613e-02 -6.63659647e-02
1.16882229e+00 2.65628457e-01 -9.39054191e-01 5.67605376e-01
-2.73556449e-02 -2.01863781e-01 -1.12311110e-01 -5.10252357e-01
7.21414089e-01 -1.34509742e-01 -1.09455502e+00 8.58834445e-01
4.02641803e-01 4.45297211e-02 -4.53393996e-01 1.82413193e-03
6.51491880e-01 -7.01636136e-01 -1.19575047e+00 4.14841801e-01
4.04561073e-01 1.62660867e-01 7.08767176e-01 -4.09216195e-01
6.44909358e-03 -6.67899787e-01 4.24301364e-02 -1.97846377e+00
-3.83123875e-01 -1.29577792e+00 -5.49788535e-01 9.61679041e-01
3.49826574e-01 -5.96317291e-01 7.36645877e-01 6.87874019e-01
-1.63715854e-01 -8.28683376e-01 -6.30635619e-01 -1.23936093e+00
9.84949097e-02 -9.49718431e-02 1.66136295e-01 5.12438357e-01
-7.23994970e-02 3.85030270e-01 -2.89171606e-01 8.53532910e-01
6.25443757e-01 -4.38500866e-02 9.12882745e-01 -9.31870580e-01
-2.44381428e-01 -2.93244600e-01 -2.04062864e-01 -9.57565963e-01
7.82522187e-02 -1.71439610e-02 6.77678943e-01 -1.39283597e+00
-5.25862634e-01 -9.42130387e-02 3.52790728e-02 -2.25031123e-01
-3.08318675e-01 -5.54910719e-01 3.31824243e-01 5.83864786e-02
-1.05215117e-01 7.20323920e-01 1.62060690e+00 -2.65344411e-01
-9.38145995e-01 6.75602734e-01 -1.74302593e-01 7.10382521e-01
1.02418125e+00 -3.20977531e-02 -9.71388757e-01 -1.15040101e-01
-5.66803932e-01 2.90215671e-01 5.77417426e-02 -1.18393695e+00
3.26609194e-01 -5.86202502e-01 2.70288140e-01 -5.18782675e-01
3.65832955e-01 -1.39379323e+00 4.61879104e-01 9.31037426e-01
-3.53627384e-01 2.35771596e-01 2.58101225e-01 1.15515816e+00
-1.38657004e-01 -2.61094004e-01 1.04999328e+00 -2.55556628e-02
-4.67039466e-01 3.36173296e-01 -5.56235194e-01 -3.78491849e-01
1.38848662e+00 -2.85158813e-01 2.00363815e-01 -3.94427806e-01
-9.49025035e-01 6.20360672e-01 2.58944631e-01 3.74907583e-01
3.72462600e-01 -1.03257275e+00 -3.26827049e-01 1.56427862e-03
-3.55530262e-01 1.11615635e-01 -1.21569425e-01 7.53698468e-01
-3.93500835e-01 6.17385149e-01 -5.33188358e-02 -7.50944018e-01
-9.81774449e-01 7.55324423e-01 5.33303976e-01 -2.88157403e-01
-7.30537117e-01 2.16526553e-01 1.10823251e-01 -3.21406305e-01
1.28139466e-01 -6.80371106e-01 7.93808550e-02 -3.59886795e-01
4.77111995e-01 6.81272745e-01 -7.80892968e-02 -1.71469897e-01
2.46038333e-01 8.14989805e-01 1.18845120e-01 -1.15551248e-01
1.16741848e+00 -3.75538915e-01 3.76789123e-01 8.35810840e-01
7.07907677e-01 -4.32065725e-01 -2.02735686e+00 2.08801148e-03
1.89809948e-01 -4.97678757e-01 1.45234659e-01 -2.30124831e-01
-9.34508264e-01 3.38272244e-01 7.41330147e-01 6.56967536e-02
6.14329338e-01 -8.08077157e-01 4.31845009e-01 5.99779963e-01
4.27074283e-01 -1.37151372e+00 2.97725916e-01 8.85720253e-01
9.27587926e-01 -7.56718040e-01 2.54714880e-02 -6.71920598e-01
-7.35314667e-01 1.27233207e+00 7.21275032e-01 -5.31807899e-01
8.33896220e-01 4.52522337e-01 5.45603260e-02 2.14936063e-01
-8.18214655e-01 2.11020738e-01 -1.54504860e-02 6.43987954e-01
2.31103882e-01 -1.38383418e-01 -1.61252558e-01 1.04168020e-01
3.39392692e-01 1.13411359e-01 4.93725479e-01 1.44810951e+00
-6.66981459e-01 -9.52969372e-01 -8.05786788e-01 2.05769047e-01
-2.86782056e-01 6.24477923e-01 3.16340744e-01 9.81846988e-01
-5.50766468e-01 9.54260707e-01 5.76996841e-02 1.62451714e-01
1.16730499e+00 4.43479382e-02 4.82206613e-01 -6.96250021e-01
-3.28327358e-01 4.94467646e-01 -3.77065549e-03 -7.47441351e-01
1.08958498e-01 -5.09102643e-01 -1.35530484e+00 -1.40083078e-02
-4.92075562e-01 1.54993180e-02 6.75484359e-01 7.07099438e-01
2.02783212e-01 8.91456127e-01 7.94503331e-01 -1.10281551e+00
-1.28591728e+00 -8.69357467e-01 -6.46053076e-01 -4.86685261e-02
4.47493881e-01 -1.01157987e+00 -1.71707585e-01 3.03045213e-01]
|
[5.081645965576172, 2.319078207015991]
|
532f6987-8236-4d89-852a-e9332a40dd79
|
solving-the-rubiks-cube-with-approximate
| null | null |
https://openreview.net/forum?id=Hyfn2jCcKm
|
https://openreview.net/pdf?id=Hyfn2jCcKm
|
Solving the Rubik's Cube with Approximate Policy Iteration
|
Recently, Approximate Policy Iteration (API) algorithms have achieved super-human proficiency in two-player zero-sum games such as Go, Chess, and Shogi without human data. These API algorithms iterate between two policies: a slow policy (tree search), and a fast policy (a neural network). In these two-player games, a reward is always received at the end of the game. However, the Rubik’s Cube has only a single solved state, and episodes are not guaranteed to terminate. This poses a major problem for these API algorithms since they rely on the reward received at the end of the game. We introduce Autodidactic Iteration: an API algorithm that overcomes the problem of sparse rewards by training on a distribution of states that allows the reward to propagate from the goal state to states farther away. Autodidactic Iteration is able to learn how to solve the Rubik’s Cube and the 15-puzzle without relying on human data. 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.
|
['Pierre Baldi', 'Alexander Shmakov', 'Stephen McAleer', 'Forest Agostinelli']
|
2019-05-01
| null | null | null |
iclr-2019-5
|
['rubik-s-cube']
|
['graphs']
|
[-2.45659456e-01 2.75564790e-01 -1.04723752e-01 2.95619428e-01
-7.28899479e-01 -1.02824831e+00 2.39209324e-01 -3.37474234e-02
-5.29242694e-01 1.26648176e+00 -1.51000500e-01 -6.39159799e-01
-3.06081444e-01 -9.73040581e-01 -7.16071665e-01 -5.99515796e-01
-2.19185010e-01 1.18441415e+00 3.17981303e-01 -4.32654798e-01
4.54166442e-01 6.89944774e-02 -1.23051584e+00 2.69384354e-01
8.75982821e-01 8.28578353e-01 2.54862159e-01 9.99877930e-01
-1.48290500e-01 1.14249635e+00 -6.60127401e-01 -8.99685174e-02
4.58469212e-01 -5.91464758e-01 -1.10330856e+00 -3.04415971e-01
-3.42028886e-01 -5.24817467e-01 -2.44253725e-01 1.02409339e+00
2.43688166e-01 2.94531405e-01 3.89128536e-01 -1.29839301e+00
-1.29718855e-01 9.83867586e-01 -4.17170197e-01 9.41516161e-02
5.99521399e-01 6.23605549e-01 1.01226652e+00 -9.29908603e-02
7.45337844e-01 9.14764047e-01 3.66182655e-01 5.75577736e-01
-1.07267249e+00 -5.01740456e-01 2.75641352e-01 2.85473377e-01
-1.20017660e+00 -2.51028165e-02 3.20141852e-01 -1.85747385e-01
1.33132637e+00 7.80854672e-02 1.10473776e+00 7.96337903e-01
1.53000623e-01 7.87249386e-01 1.20802510e+00 -1.97661996e-01
9.31096613e-01 -4.63728994e-01 -3.39083999e-01 6.87536895e-01
9.91420671e-02 5.83124340e-01 -5.03287315e-01 -2.54118592e-01
9.95072722e-01 -4.49200362e-01 -1.27156958e-01 -5.15601695e-01
-9.58860636e-01 8.70850503e-01 3.61258984e-01 2.03126386e-01
-5.95888913e-01 4.45152640e-01 2.55637407e-01 5.68933129e-01
-2.22199813e-01 1.17522717e+00 -3.29259098e-01 -8.98803949e-01
-1.01048982e+00 7.55061209e-01 1.25674784e+00 6.15325272e-01
4.79223222e-01 1.41253382e-01 -3.79475541e-02 1.03415106e-03
-2.72837520e-01 3.31905067e-01 2.90868282e-01 -1.63852620e+00
6.63687587e-01 5.02813339e-01 7.71440923e-01 -5.73777854e-01
-4.91401464e-01 -4.37677264e-01 -4.46319580e-01 8.79698455e-01
9.39742088e-01 -4.21630919e-01 -7.67453611e-01 1.74559295e+00
2.59677857e-01 3.02437931e-01 1.14691138e-01 1.27678871e+00
2.73637831e-01 8.07743788e-01 -2.68611640e-01 -2.12128490e-01
9.20849204e-01 -9.81024444e-01 -2.19488397e-01 -4.90477055e-01
8.02075148e-01 -2.14254230e-01 8.23269010e-01 9.73743796e-01
-1.80440247e+00 8.29188526e-02 -9.63651836e-01 3.72819901e-01
-3.62879746e-02 -3.39391291e-01 9.61776018e-01 3.80264819e-01
-1.07841325e+00 7.81121373e-01 -1.05784369e+00 1.29918963e-01
1.17278062e-01 6.05846345e-01 -3.57780814e-01 -1.99348643e-01
-9.72564161e-01 1.12646401e+00 6.46630168e-01 -6.37602359e-02
-1.09656847e+00 -3.83253098e-01 -6.26699984e-01 3.95120174e-01
9.18315828e-01 -6.85042799e-01 1.59112573e+00 -9.71700370e-01
-1.77007139e+00 5.70620477e-01 1.13268457e-01 -5.94357550e-01
5.48664212e-01 1.27920002e-01 3.47583801e-01 2.92159468e-02
1.12085268e-01 4.09454852e-01 4.55953509e-01 -9.54928994e-01
-8.57190371e-01 -1.47328749e-01 6.62625074e-01 5.99536300e-01
2.56017864e-01 -3.30966383e-01 -2.69271433e-01 -1.04869857e-01
2.08262175e-01 -8.75454962e-01 -5.20871639e-01 -4.67128813e-01
-1.39238209e-01 -3.34447116e-01 -6.20881692e-02 -4.44055855e-01
1.13122225e+00 -1.69714069e+00 6.77963972e-01 4.70867217e-01
1.33850738e-01 8.28328952e-02 -5.06623328e-01 4.62342709e-01
6.01703078e-02 -1.65278286e-01 -3.95934768e-02 2.01973349e-01
1.94789588e-01 3.78794312e-01 -1.67471647e-01 2.36290634e-01
-3.27278942e-01 8.60364437e-01 -1.32481611e+00 -2.42506657e-02
-2.67060816e-01 -4.43300188e-01 -1.06287742e+00 2.11257830e-01
-5.89988649e-01 2.35882506e-01 -4.05937821e-01 4.45170194e-01
4.48501915e-01 -2.03790843e-01 4.55275208e-01 5.75325608e-01
-2.75977284e-01 5.27902007e-01 -1.43611562e+00 1.73593128e+00
-2.43154690e-01 2.16367152e-02 2.98748970e-01 -9.74741161e-01
4.44583833e-01 2.80082732e-01 5.35721958e-01 -1.06627333e+00
1.36181727e-01 3.96093577e-01 4.06400681e-01 -1.35056451e-01
6.49322510e-01 -1.12157673e-01 -2.34480381e-01 9.79359686e-01
-2.82203197e-01 -6.04911029e-01 7.77213395e-01 2.53251612e-01
1.66198182e+00 1.50278062e-01 3.40920053e-02 2.67846622e-02
7.18019009e-02 5.62942386e-01 8.70827436e-01 1.22170246e+00
2.86571775e-02 3.24899614e-01 1.14416575e+00 -7.15906382e-01
-1.16435754e+00 -1.18104458e+00 8.20646524e-01 9.66165185e-01
1.79808944e-01 -3.03203881e-01 -7.72468448e-01 -3.97428304e-01
-1.17621638e-01 9.46196496e-01 -5.84643543e-01 -2.54453212e-01
-8.22384477e-01 -1.37251884e-01 3.44155461e-01 4.98828501e-01
3.29864353e-01 -1.23348773e+00 -9.29597974e-01 5.56955397e-01
-2.81900704e-01 -6.23895109e-01 -3.45347196e-01 5.10442257e-01
-7.93629646e-01 -1.26445711e+00 -5.48557281e-01 -6.28717363e-01
6.32505119e-01 -2.65997499e-01 1.32790244e+00 1.11714289e-01
-1.43416256e-01 3.18361193e-01 -1.77478388e-01 7.92667270e-02
-2.70409882e-01 3.18906993e-01 5.62617369e-03 -8.92835915e-01
-2.64093969e-02 -7.00108230e-01 -5.54130018e-01 3.06477964e-01
-3.27898175e-01 1.06690958e-01 4.08277005e-01 1.00009668e+00
4.43295896e-01 2.82443196e-01 2.42723435e-01 -4.78567779e-01
9.49079931e-01 -4.81646895e-01 -1.09855199e+00 1.57237738e-01
-3.66508186e-01 3.90639216e-01 6.56952918e-01 -5.19210696e-01
-6.35379791e-01 8.24114233e-02 1.76462039e-01 -3.88830662e-01
3.58704925e-01 5.68545520e-01 4.17572170e-01 9.00155902e-02
7.86803067e-01 4.30484414e-01 -5.79084679e-02 -5.13570793e-02
3.42759907e-01 4.32851426e-02 7.67758310e-01 -1.25437689e+00
6.11660302e-01 -8.73355195e-03 -1.04490414e-01 9.08167362e-02
-5.74157476e-01 -3.06085795e-02 5.35062701e-02 -2.02703655e-01
5.05248845e-01 -6.14204049e-01 -1.62659657e+00 4.66887146e-01
-1.04101527e+00 -1.12717748e+00 -6.03159189e-01 3.29716265e-01
-1.01815391e+00 4.56338786e-02 -7.65941024e-01 -1.06627691e+00
-4.84760329e-02 -9.77735937e-01 4.43705082e-01 4.54155952e-01
-2.91872770e-01 -5.45237124e-01 1.69107407e-01 1.62354767e-01
3.97962093e-01 1.43773884e-01 8.73452187e-01 -3.27993155e-01
-7.67580807e-01 1.31406980e-02 1.05936088e-01 -2.61731029e-01
-2.92624056e-01 -4.57901388e-01 -6.93504810e-02 -5.05946279e-01
-1.41044393e-01 -7.27399051e-01 3.91952336e-01 4.24727947e-01
9.33044970e-01 -4.48268741e-01 -1.53366700e-01 4.28980947e-01
1.13100755e+00 3.75652879e-01 6.42529726e-01 6.56095862e-01
-5.76820374e-02 6.74019977e-02 6.47907197e-01 6.81030154e-01
4.56517190e-01 5.08434176e-01 8.19419742e-01 5.30356884e-01
3.55060041e-01 -3.54399920e-01 3.15199584e-01 1.73499361e-01
-4.81730193e-01 -1.10903881e-01 -9.06915069e-01 6.27214491e-01
-2.13216591e+00 -1.10825324e+00 2.54399151e-01 2.34430695e+00
1.11638224e+00 5.64186931e-01 3.33874285e-01 1.80460468e-01
2.45667920e-01 -2.43422985e-01 -1.02487922e+00 -9.51245964e-01
2.54597038e-01 5.83145201e-01 5.63609362e-01 7.09113836e-01
-5.27987838e-01 1.29980874e+00 6.79212999e+00 7.31229901e-01
-8.15204322e-01 -2.69222111e-01 3.96041095e-01 -6.54591560e-01
-2.06316963e-01 1.89177543e-01 -3.82247865e-01 4.88571733e-01
7.51136899e-01 -3.05394202e-01 1.46452403e+00 9.60349500e-01
6.41418546e-02 -7.42248178e-01 -1.06437242e+00 8.22889030e-01
-3.99597168e-01 -1.42091286e+00 -5.22568405e-01 1.45014867e-01
9.44066048e-01 -2.90766060e-01 1.76538289e-01 6.96750402e-01
1.22517455e+00 -1.38127303e+00 8.42402101e-01 4.21075821e-01
5.58811307e-01 -1.15989256e+00 5.49866796e-01 8.52793097e-01
-9.11196291e-01 -4.68232870e-01 -3.40389490e-01 -7.61205435e-01
1.41618893e-01 1.18099660e-01 -8.67320061e-01 1.76472485e-01
3.82899284e-01 -1.05271526e-01 1.84172213e-01 1.20763195e+00
-5.33860207e-01 2.71648318e-01 -6.56535387e-01 -1.99843526e-01
7.54848361e-01 -3.66009176e-01 4.61264789e-01 2.72382408e-01
4.74417835e-01 7.02868104e-01 3.62204850e-01 1.07579136e+00
2.96805739e-01 -4.85945135e-01 -2.50960976e-01 -1.96653038e-01
5.93170702e-01 7.82768726e-01 -7.60498166e-01 -1.13781534e-01
2.12702066e-01 9.88004923e-01 6.21690929e-01 3.63113374e-01
-7.60583997e-01 -1.96932361e-01 6.48030519e-01 7.46983960e-02
4.91417706e-01 -5.41394591e-01 -5.69854736e-01 -9.07065809e-01
-1.54773429e-01 -1.37741244e+00 4.33245033e-01 -9.26441729e-01
-7.52053618e-01 3.08999926e-01 -2.79208481e-01 -7.10706174e-01
-9.04987037e-01 -5.34581304e-01 -7.92996228e-01 9.21031475e-01
-1.05352163e+00 -4.57540900e-01 4.22779955e-02 5.84716618e-01
1.91615626e-01 -2.40819111e-01 8.20574641e-01 -3.12307656e-01
-3.31191838e-01 4.36704308e-01 8.52035955e-02 -5.88620119e-02
9.48415548e-02 -1.44470060e+00 3.84498984e-01 4.60956186e-01
-2.02161849e-01 3.94244730e-01 8.45105231e-01 -5.11582494e-01
-1.66692615e+00 -1.58936217e-01 3.43011707e-01 -2.30189607e-01
5.13584077e-01 3.50096859e-02 -5.57638526e-01 6.76434278e-01
-5.22103608e-02 -1.51133105e-01 8.84064957e-02 3.00244331e-01
-1.08529344e-01 2.04844832e-01 -1.14743400e+00 7.38321841e-01
1.00594258e+00 -1.54084459e-01 -5.80024600e-01 1.04572982e-01
2.23675936e-01 -1.10897160e+00 -4.52674299e-01 -2.09137186e-01
4.54587787e-01 -1.16374958e+00 7.78821409e-01 -8.33298564e-01
5.36397219e-01 -3.00443441e-01 2.03707948e-01 -1.69415963e+00
-4.28923041e-01 -8.48500192e-01 -5.24692237e-01 2.79653907e-01
2.93559372e-01 -6.67220652e-01 1.23341346e+00 9.27692115e-01
1.18085392e-01 -8.37783039e-01 -1.18450058e+00 -1.00229073e+00
4.89897370e-01 -2.78273016e-01 5.89822710e-01 6.28518343e-01
5.75942338e-01 7.58569837e-02 -3.26921850e-01 7.36877918e-02
5.97502112e-01 3.89863878e-01 7.56631851e-01 -8.27139258e-01
-7.61507511e-01 -7.17394471e-01 7.95069337e-02 -1.19730580e+00
2.91270353e-02 -6.91608191e-01 7.30609521e-02 -1.64737594e+00
1.25510752e-01 -7.12556958e-01 -6.05793409e-02 6.73359275e-01
1.62816420e-01 -1.31059736e-01 4.11721140e-01 -1.69805363e-01
-9.58479345e-01 2.45456010e-01 1.50217068e+00 -7.45978951e-02
-3.36708099e-01 -3.98834497e-02 -7.54649401e-01 5.98139822e-01
8.64099801e-01 -5.50122440e-01 -3.34626883e-01 -5.70588708e-01
8.77196252e-01 7.84496605e-01 5.32911681e-02 -1.14100945e+00
6.35698557e-01 -6.43963814e-01 1.05306327e-01 -3.70570064e-01
3.58101249e-01 -4.89135027e-01 1.56450510e-01 9.64489818e-01
-1.42534614e-01 2.34863311e-01 2.12067038e-01 2.42318004e-01
1.41487122e-01 -4.78937656e-01 5.07468343e-01 -5.93807280e-01
-5.29931366e-01 1.01759424e-02 -6.66138947e-01 4.04032201e-01
1.19778562e+00 -3.22563589e-01 -2.20290691e-01 -7.52343833e-01
-9.71220255e-01 7.25925863e-01 4.80633825e-01 -1.27783448e-01
3.86975437e-01 -1.10901475e+00 -4.96433139e-01 -8.09047446e-02
-6.03760302e-01 1.97344348e-01 1.76974550e-01 5.57824790e-01
-5.95263898e-01 3.08480322e-01 -4.43181187e-01 -1.23342693e-01
-8.36270094e-01 5.28695226e-01 5.84565341e-01 -9.93076205e-01
-4.85781759e-01 8.43436420e-01 -1.92278728e-01 -4.23416942e-01
1.72069162e-01 -3.77771765e-01 2.12477908e-01 -3.01023692e-01
5.07726729e-01 4.53898102e-01 -2.20446631e-01 2.59377271e-01
-2.37831697e-01 1.63527578e-01 -5.85001297e-02 -4.55513358e-01
1.44885457e+00 4.22759712e-01 -4.06216830e-02 -7.67903477e-02
2.65040547e-01 -3.73014957e-01 -1.50100327e+00 7.67316744e-02
-2.54095465e-01 -4.30765688e-01 -2.74699181e-02 -1.16534066e+00
-9.06116843e-01 5.13412058e-01 1.16223574e-01 3.49526823e-01
9.05325830e-01 -3.08367997e-01 8.72721255e-01 8.01918805e-01
9.94332135e-01 -1.38389552e+00 2.97808647e-01 1.08171427e+00
6.88288212e-01 -7.37780511e-01 -2.19493862e-02 1.89527199e-01
-7.31907189e-01 1.08217251e+00 8.23279619e-01 -3.91482145e-01
-1.44150391e-01 5.42228937e-01 -3.55893195e-01 -3.58731523e-02
-1.13777995e+00 -3.71031910e-01 -3.73552263e-01 6.35852575e-01
-1.99406415e-01 1.12285070e-01 -2.96050429e-01 8.24488103e-01
-6.44985378e-01 2.06138059e-01 6.78416133e-01 1.12988532e+00
-7.16355026e-01 -1.02299035e+00 -6.79167688e-01 3.39957654e-01
-7.24035278e-02 3.90269421e-02 -2.50014216e-01 5.75023293e-01
-2.17304721e-01 7.43379831e-01 2.57813960e-01 1.43295210e-02
2.09149331e-01 -3.56849991e-02 9.53121483e-01 -4.38291550e-01
-6.60365164e-01 -3.33129823e-01 1.48057371e-01 -1.01360834e+00
4.33874190e-01 -5.27177274e-01 -1.54032922e+00 -9.58115578e-01
-1.48710515e-02 5.00386655e-01 3.44740748e-01 9.47936058e-01
1.43883899e-01 4.87566322e-01 4.59727734e-01 -8.62854004e-01
-8.25025618e-01 -3.97926539e-01 -5.78564525e-01 -1.37642220e-01
-5.93959056e-02 -5.79267204e-01 -1.24329418e-01 -5.94942629e-01]
|
[3.6498239040374756, 1.5622774362564087]
|
1ca1abaa-b000-4cd5-bfcf-eeef82126f45
|
sc-depthv3-robust-self-supervised-monocular
|
2211.03660
| null |
https://arxiv.org/abs/2211.03660v1
|
https://arxiv.org/pdf/2211.03660v1.pdf
|
SC-DepthV3: Robust Self-supervised Monocular Depth Estimation for Dynamic Scenes
|
Self-supervised monocular depth estimation has shown impressive results in static scenes. It relies on the multi-view consistency assumption for training networks, however, that is violated in dynamic object regions and occlusions. Consequently, existing methods show poor accuracy in dynamic scenes, and the estimated depth map is blurred at object boundaries because they are usually occluded in other training views. In this paper, we propose SC-DepthV3 for addressing the challenges. Specifically, we introduce an external pretrained monocular depth estimation model for generating single-image depth prior, namely pseudo-depth, based on which we propose novel losses to boost self-supervised training. As a result, our model can predict sharp and accurate depth maps, even when training from monocular videos of highly-dynamic scenes. We demonstrate the significantly superior performance of our method over previous methods on six challenging datasets, and we provide detailed ablation studies for the proposed terms. Source code and data will be released at https://github.com/JiawangBian/sc_depth_pl
|
['Chunhua Shen', 'Ian Reid', 'Wei Yin', 'Huangying Zhan', 'Jia-Wang Bian', 'Libo Sun']
|
2022-11-07
| null | null | null | null |
['indoor-monocular-depth-estimation']
|
['computer-vision']
|
[ 1.70304418e-01 -1.35420188e-01 -2.72404671e-01 -5.54384589e-01
-5.13320148e-01 -5.10647535e-01 3.09331357e-01 -6.73614800e-01
-1.36039793e-01 8.38107526e-01 2.72002339e-01 1.78965613e-01
3.11418593e-01 -5.59497356e-01 -7.26804078e-01 -7.10328102e-01
4.46533918e-01 1.25774786e-01 3.17739546e-01 3.48789483e-01
1.37664378e-01 3.28275114e-01 -1.42640197e+00 3.59708697e-01
8.36673498e-01 1.09514725e+00 4.76162702e-01 6.16155446e-01
1.08393244e-01 1.03934562e+00 -2.28428364e-01 -1.98844448e-01
3.80361885e-01 -1.84733868e-01 -6.68649733e-01 1.89862505e-01
8.93793583e-01 -1.18594491e+00 -9.55606282e-01 1.10858023e+00
3.74550015e-01 -1.51464313e-01 3.88999581e-01 -1.21201861e+00
-4.62410569e-01 3.36138718e-02 -7.22806990e-01 4.77387682e-02
4.38623995e-01 2.28303269e-01 6.40176177e-01 -1.04319727e+00
7.38224089e-01 1.08866465e+00 3.51095140e-01 7.88050115e-01
-9.89937603e-01 -7.09997058e-01 4.20147687e-01 2.79958874e-01
-1.16163075e+00 -6.41149461e-01 9.30058002e-01 -4.03851300e-01
8.11847389e-01 -1.37126207e-01 6.40789926e-01 1.26936960e+00
1.54901192e-01 1.16129482e+00 9.93807197e-01 -3.32379178e-03
3.81956063e-02 -6.90456554e-02 -2.40254432e-01 6.05601251e-01
2.69121140e-01 5.19404709e-01 -7.06483603e-01 2.50464141e-01
1.18651772e+00 2.10857570e-01 -7.20079660e-01 -7.58084059e-01
-1.04862928e+00 4.68818128e-01 4.47346568e-01 -1.54738491e-02
-7.00770989e-02 3.34728092e-01 2.13546723e-01 1.26136601e-01
7.31017232e-01 -1.42249703e-01 -5.73687851e-01 -7.83712566e-02
-7.50228226e-01 1.50433108e-01 2.60294020e-01 1.17044222e+00
8.85559797e-01 3.83276865e-02 1.40319318e-01 8.92714500e-01
2.97738850e-01 3.95442396e-01 2.31805637e-01 -1.40728104e+00
6.14475429e-01 4.03124392e-01 1.33830279e-01 -7.77067959e-01
-3.66077214e-01 -4.17337447e-01 -9.15332794e-01 2.31673479e-01
3.09077084e-01 -1.23065464e-01 -8.89818370e-01 1.62819529e+00
5.15080810e-01 1.85714528e-01 -2.52771862e-02 1.20926571e+00
1.04650295e+00 5.99982381e-01 -4.77065146e-01 -1.52264297e-01
7.11817920e-01 -1.29400182e+00 -7.41018653e-01 -7.19042718e-01
2.43210703e-01 -6.41742229e-01 9.63906705e-01 6.23369753e-01
-1.37538409e+00 -5.11494696e-01 -1.06006265e+00 -4.25992429e-01
1.99908227e-01 1.36404216e-01 7.05192387e-01 1.99954510e-01
-1.09058392e+00 4.13623869e-01 -1.00138211e+00 -1.64733976e-02
5.99864602e-01 8.27873200e-02 -3.80379051e-01 -7.14067459e-01
-8.52838635e-01 4.30832475e-01 2.46225014e-01 1.74883559e-01
-1.20604384e+00 -7.19775975e-01 -1.14615631e+00 -2.98313588e-01
3.40950906e-01 -9.41054881e-01 1.32471788e+00 -8.33033741e-01
-1.46267247e+00 1.02502632e+00 -4.05412912e-01 -7.66103417e-02
7.98192680e-01 -5.82032382e-01 8.11780766e-02 5.02960682e-01
1.51474103e-01 9.66353536e-01 8.17761064e-01 -1.54915547e+00
-6.56523824e-01 -6.01798475e-01 3.51719320e-01 4.34687793e-01
-9.47197229e-02 -3.47836763e-01 -8.12098801e-01 -3.88097703e-01
6.34562969e-01 -6.69915974e-01 -6.44804630e-03 4.28371727e-01
-3.21716458e-01 1.89400509e-01 7.54841506e-01 -5.21361291e-01
8.63067210e-01 -2.14537144e+00 1.10224046e-01 -4.14759547e-01
4.55761343e-01 -6.27568290e-02 5.91050982e-02 1.15669027e-01
8.01083893e-02 -3.09552252e-01 -2.39580557e-01 -6.44421935e-01
-3.49686444e-01 2.11645409e-01 -1.91490456e-01 6.09921098e-01
-1.76684558e-02 9.11892951e-01 -9.52765048e-01 -3.66380244e-01
4.41563845e-01 6.53768897e-01 -6.58076048e-01 4.98186380e-01
-2.05927461e-01 8.58988225e-01 -2.69687951e-01 9.12987053e-01
1.10925019e+00 -4.38977450e-01 -2.41529569e-02 -3.76241863e-01
-3.16043757e-03 2.63632417e-01 -9.66622949e-01 2.14203644e+00
-4.70686853e-01 7.50794113e-01 1.98310599e-01 -6.85907900e-01
6.59479201e-01 7.86127076e-02 5.64929426e-01 -7.90549397e-01
5.23884632e-02 1.66770369e-01 -4.95356143e-01 -4.15713757e-01
3.21801186e-01 4.70157675e-02 4.28884983e-01 1.50809646e-01
-8.24202001e-02 -4.88121569e-01 9.32124630e-03 2.45313391e-01
8.77417982e-01 6.60451770e-01 7.98372999e-02 2.10015565e-01
5.52995563e-01 -4.26962137e-01 8.50841641e-01 3.67811143e-01
-4.22460526e-01 1.21841586e+00 3.80120695e-01 -4.82631743e-01
-9.57448661e-01 -1.27622008e+00 -3.33915740e-01 4.29141223e-01
6.89573765e-01 -1.85869828e-01 -4.01162565e-01 -6.36951566e-01
-1.01917014e-01 2.84080923e-01 -3.97069365e-01 -2.07802970e-02
-4.82308120e-01 -4.34650004e-01 3.57007086e-02 6.29473209e-01
8.60643566e-01 -7.61427343e-01 -4.77330893e-01 1.18575245e-01
-4.59491134e-01 -1.66934502e+00 -4.50032979e-01 4.10332624e-03
-1.14213359e+00 -1.13383877e+00 -9.81143355e-01 -8.16372752e-01
6.76989198e-01 6.98042154e-01 1.25944149e+00 1.07254554e-03
-6.87905923e-02 3.40222985e-01 -1.32359430e-01 -1.00077651e-01
1.68517046e-02 -2.07200885e-01 -1.77895382e-01 -1.21050984e-01
2.34104633e-01 -7.92231619e-01 -1.23457205e+00 4.57230687e-01
-8.92103791e-01 5.70044100e-01 2.99369931e-01 9.24017131e-01
6.34019077e-01 -1.37816697e-01 3.13848138e-01 -8.66620421e-01
-3.83483559e-01 -2.63486177e-01 -8.43161881e-01 -2.25329638e-01
-4.15865362e-01 -3.90509844e-01 3.92604560e-01 -5.11165224e-02
-1.34089673e+00 1.86408892e-01 -1.89116225e-01 -7.06946731e-01
-1.48163855e-01 6.06976822e-02 -3.90104562e-01 -1.72775805e-01
3.60138059e-01 4.02597159e-01 -1.81333795e-01 -3.09197783e-01
3.77167687e-02 4.11133885e-01 4.62830633e-01 -4.13173825e-01
7.36129403e-01 1.03618538e+00 -2.14278817e-01 -5.03208697e-01
-1.29389322e+00 -4.76277590e-01 -6.52930379e-01 -3.28154087e-01
6.91974759e-01 -1.56506801e+00 -4.18009520e-01 9.50013161e-01
-1.24130392e+00 -7.38066256e-01 5.19712046e-02 6.42469823e-01
-7.93892086e-01 5.63471675e-01 -9.38037694e-01 -6.15073025e-01
-2.00151309e-01 -9.83254254e-01 1.28785646e+00 2.59227991e-01
3.37014079e-01 -1.01385343e+00 -9.55157280e-02 7.17456341e-01
2.06178710e-01 2.24121794e-01 3.98736417e-01 4.21719164e-01
-1.19010580e+00 1.70509517e-01 -5.74202657e-01 6.59873366e-01
2.42630109e-01 -1.69823900e-01 -1.32594609e+00 -4.72330242e-01
2.17153668e-01 -5.63232422e-01 9.71460760e-01 6.47327542e-01
1.42487693e+00 5.04803192e-03 -2.55236804e-01 1.29258537e+00
1.64192986e+00 1.84078485e-01 7.88307786e-01 3.01295191e-01
1.10614979e+00 5.75355232e-01 8.39222074e-01 4.79302019e-01
5.92866898e-01 6.55160367e-01 6.82080925e-01 -1.62847459e-01
-2.61603087e-01 -2.57978410e-01 3.31624568e-01 5.63149154e-01
1.58009231e-01 -2.88861662e-01 -7.03016400e-01 5.97739935e-01
-1.70125270e+00 -7.80874789e-01 -1.99256055e-02 2.02312279e+00
7.93260515e-01 1.68669268e-01 -1.97432831e-01 -1.11823648e-01
3.59799892e-01 4.69902158e-01 -1.01431727e+00 1.87567130e-01
-3.37439090e-01 -2.45556265e-01 4.06753033e-01 6.90141201e-01
-1.04286432e+00 9.54187393e-01 5.81517935e+00 3.50429684e-01
-1.12691677e+00 1.46765456e-01 8.14402819e-01 -5.48511982e-01
-3.67984295e-01 -5.50003499e-02 -7.48998284e-01 4.34988618e-01
9.97547507e-02 -1.90995373e-02 2.40108967e-01 8.53995383e-01
2.68491685e-01 -4.84195679e-01 -1.20859432e+00 1.44778001e+00
2.30706260e-01 -1.15429115e+00 -1.13119923e-01 4.12475783e-03
1.33604848e+00 2.83073306e-01 1.09594442e-01 -5.87433465e-02
1.45635223e-02 -7.65325427e-01 5.54150224e-01 2.52004266e-01
1.03492641e+00 -5.77317059e-01 6.05147302e-01 3.35796863e-01
-1.07193077e+00 -4.99822572e-02 -5.91968179e-01 -1.29478097e-01
3.95729899e-01 9.23923969e-01 -1.93768755e-01 6.51889920e-01
8.37221682e-01 1.40951085e+00 -3.00493658e-01 1.04863656e+00
-4.85724479e-01 1.41495809e-01 -1.06177427e-01 6.75916374e-01
3.62502574e-03 -1.89069554e-01 2.68829286e-01 7.53355205e-01
2.37656713e-01 1.16202593e-01 -7.91221485e-02 9.80552852e-01
-8.46931636e-02 -2.95315921e-01 -7.36219823e-01 3.83869559e-01
2.40715653e-01 1.09904873e+00 -2.86303788e-01 -3.04908752e-01
-8.05562079e-01 1.28516567e+00 4.35520798e-01 5.67157507e-01
-8.12950432e-01 1.16518922e-01 8.55669916e-01 2.14462891e-01
1.80683792e-01 -3.27060252e-01 -4.60345089e-01 -1.77986479e+00
3.99908006e-01 -5.74921250e-01 1.84258193e-01 -1.18714607e+00
-1.23476267e+00 5.13773143e-01 -4.94752415e-02 -1.44583917e+00
-1.38159037e-01 -6.13313496e-01 -3.66247356e-01 6.42162800e-01
-2.05190969e+00 -9.33291316e-01 -9.94011521e-01 6.18799686e-01
7.15697289e-01 8.52819756e-02 1.64938197e-01 6.01582468e-01
-4.16267574e-01 4.49458390e-01 2.63436083e-02 -5.37733659e-02
9.30016160e-01 -1.06362998e+00 3.98495167e-01 8.74142051e-01
-1.38380855e-01 1.51097223e-01 2.93232530e-01 -6.05002105e-01
-1.26497650e+00 -1.09680712e+00 5.03580928e-01 -4.32239234e-01
2.59901404e-01 -3.50364268e-01 -1.00453353e+00 7.80479491e-01
5.37315793e-02 3.33168387e-01 1.68656215e-01 -3.51201594e-01
-2.27558553e-01 -1.89943373e-01 -9.25078034e-01 4.34917688e-01
1.53117573e+00 -5.60550511e-01 -9.65052322e-02 4.23746198e-01
7.87583351e-01 -8.95948768e-01 -6.73794031e-01 6.42929435e-01
6.42133176e-01 -1.60947442e+00 1.09967422e+00 7.98988342e-02
9.66126084e-01 -3.52746457e-01 -1.83360994e-01 -1.00853133e+00
-4.80396971e-02 -2.53236145e-01 -4.61089879e-01 8.90322745e-01
-5.26405685e-03 -6.30013406e-01 1.27133024e+00 3.54190320e-01
-2.99690723e-01 -8.49226236e-01 -7.62996137e-01 -7.57538438e-01
-5.81841953e-02 -4.11586612e-01 2.25058451e-01 7.99364686e-01
-4.37749237e-01 2.62157589e-01 -5.10512292e-01 2.49790028e-01
9.97796118e-01 1.96425095e-01 9.98915136e-01 -7.49701440e-01
-4.25374836e-01 -9.78550315e-02 -3.93815100e-01 -1.92678046e+00
1.37485340e-01 -4.88856882e-01 6.30795807e-02 -1.67591703e+00
4.08367783e-01 -3.65645647e-01 6.78752735e-02 1.00293785e-01
-1.93975970e-01 4.44952130e-01 -2.57488829e-03 2.80509949e-01
-5.00823975e-01 9.03264999e-01 1.81224549e+00 8.67010653e-03
-7.82034323e-02 -2.48876512e-02 -4.53065157e-01 8.51625383e-01
7.71813750e-01 -1.94417834e-01 -7.65042305e-01 -8.58708084e-01
1.55584380e-01 2.60546029e-01 5.21744609e-01 -1.08221006e+00
1.96684748e-01 -2.91075021e-01 6.83521450e-01 -1.01032281e+00
7.12906539e-01 -7.76064873e-01 9.05812904e-02 3.31243306e-01
3.78844216e-02 -1.98440343e-01 7.25861117e-02 5.31656802e-01
-4.56292331e-01 3.00285947e-02 9.30849016e-01 -1.34029895e-01
-8.78701091e-01 8.17040801e-01 1.81626663e-01 1.44327536e-01
9.00243938e-01 -3.77960265e-01 -4.96454239e-01 -6.99122190e-01
-5.48070967e-01 3.77748519e-01 9.87436950e-01 4.05578047e-01
1.08879495e+00 -1.28762472e+00 -6.41607583e-01 4.50430155e-01
2.43080065e-01 7.22216129e-01 6.74887478e-01 7.18895733e-01
-8.20666552e-01 3.27285558e-01 -1.97213113e-01 -9.33236003e-01
-1.06540143e+00 5.07083356e-01 5.07160842e-01 1.03166699e-01
-8.46157253e-01 8.93851936e-01 1.12092054e+00 -2.99514562e-01
3.84168953e-01 -3.41295183e-01 1.74518913e-01 -3.81753057e-01
5.92984259e-01 1.06613509e-01 -1.76156744e-01 -3.05353940e-01
-2.37299815e-01 8.66714001e-01 -2.19288826e-01 -2.86233630e-02
1.27406621e+00 -6.56561196e-01 1.23773932e-01 4.69181538e-01
1.36386013e+00 -1.51174024e-01 -2.12796926e+00 -4.26095158e-01
-6.25654221e-01 -1.04977894e+00 1.45043746e-01 -6.24400020e-01
-1.42238593e+00 9.86604273e-01 5.26627362e-01 -5.74450850e-01
1.32012045e+00 7.59185627e-02 9.29208934e-01 3.13245840e-02
5.64672112e-01 -8.49295199e-01 3.60746711e-01 6.20186508e-01
8.64680469e-01 -1.68587565e+00 1.08247481e-01 -8.19781303e-01
-5.49940407e-01 9.91556227e-01 1.20911026e+00 -1.47167323e-02
5.91859281e-01 1.64782450e-01 2.21640706e-01 -7.74138644e-02
-8.67881954e-01 4.12241630e-02 8.31568539e-02 6.66891694e-01
3.30470622e-01 -3.36522549e-01 1.14155464e-01 2.93299723e-02
8.13117400e-02 6.11746348e-02 6.46512330e-01 8.02569866e-01
-1.25419036e-01 -8.14146399e-01 -8.38446021e-02 1.55585527e-01
-2.74859011e-01 -1.33079916e-01 -1.54338688e-01 6.16667569e-01
2.81220768e-02 7.88347006e-01 4.75308038e-02 -2.52894729e-01
2.11897552e-01 -4.83865947e-01 7.69503295e-01 -6.19933069e-01
1.28782392e-01 1.41113192e-01 -6.19729720e-02 -9.25648212e-01
-5.14047027e-01 -6.33949757e-01 -1.07017255e+00 -3.71523768e-01
-3.73652205e-02 -5.35436273e-01 5.23859501e-01 6.79893374e-01
2.77327627e-01 4.22093838e-01 9.08805728e-01 -1.14010942e+00
-1.22540675e-01 -6.49971485e-01 -5.54755509e-01 3.75213027e-01
5.63289881e-01 -7.90577829e-01 -6.65045023e-01 1.87669341e-02]
|
[8.821861267089844, -2.4718008041381836]
|
f73849ca-c2eb-470c-b9e5-b7236e23e62a
|
low-bit-quantization-and-quantization-aware
| null | null |
https://openreview.net/forum?id=rJxVxiiDoX
|
https://openreview.net/pdf?id=rJxVxiiDoX
|
Low-bit quantization and quantization-aware training for small-footprint keyword spotting
|
We investigate low-bit quantization to reduce computational cost of deep neural network (DNN) based keyword spotting (KWS). We propose approaches to further reduce quantization bits via integrating quantization into keyword spotting model training, which we refer to as quantization-aware training. Our experimental results on large dataset indicate that quantization-aware training can recover performance models quantized to lower bits representations. By combining quantization-aware training and weight matrix factorization, we are able to significantly reduce model size and computation for small-footprint keyword spotting, while maintaining performance.
|
['Shiv Naga Prasad Vitaladevuni', 'Oleg Rybakov', 'Spyros Matsoukas', 'Chris Beauchene', 'Ming Sun', 'Yusuf Goren', 'Yuriy Mishchenko']
|
2018-10-19
| null | null | null | null |
['small-footprint-keyword-spotting']
|
['speech']
|
[ 1.77432373e-01 -3.09440941e-01 -5.64468801e-01 -3.59110206e-01
-1.14925110e+00 -2.31625080e-01 3.22229445e-01 3.87661578e-03
-9.29500937e-01 5.23970187e-01 4.35227722e-01 -7.23618805e-01
-4.60560992e-02 -7.82528698e-01 -8.37440550e-01 -2.44814485e-01
3.94204229e-01 5.22127748e-02 -1.14508942e-01 1.97667126e-02
2.80485451e-01 3.46037716e-01 -1.35882688e+00 5.26851118e-01
3.97622526e-01 1.23356378e+00 4.32340205e-01 8.61816406e-01
-2.77294695e-01 5.97207904e-01 -6.10583842e-01 -8.84495601e-02
3.35080922e-01 7.09151402e-02 -6.03455484e-01 -2.62527555e-01
6.27480507e-01 -8.60248089e-01 -8.57622445e-01 9.78253305e-01
5.24409115e-01 1.20725088e-01 4.30351257e-01 -1.03813481e+00
-1.18086529e+00 1.03143787e+00 -1.11582279e-01 2.83801824e-01
-1.87201813e-01 -1.98231265e-01 1.23474193e+00 -1.23146331e+00
3.24996978e-01 1.39232957e+00 5.67958713e-01 7.77225792e-01
-1.23178422e+00 -9.83693898e-01 -6.51811138e-02 6.94722533e-01
-2.06213069e+00 -7.89643824e-01 4.04509962e-01 1.56099722e-01
1.66888416e+00 4.49019581e-01 5.20443261e-01 8.71182323e-01
-1.72926530e-01 1.09953153e+00 3.18735123e-01 -6.88949466e-01
3.18049222e-01 -1.14354320e-01 -6.58214018e-02 4.90821898e-01
1.99215814e-01 2.86763511e-03 -9.03826296e-01 -2.29559168e-01
7.09875762e-01 3.50489132e-02 -3.16782475e-01 2.04708248e-01
-9.33620453e-01 1.02678430e+00 2.56606638e-01 5.38299419e-02
-4.45991576e-01 9.43687320e-01 6.28005564e-01 3.28627199e-01
2.46959180e-01 3.21384192e-01 -6.87203348e-01 -4.33756679e-01
-1.29424882e+00 3.76126736e-01 4.61337119e-01 1.13601613e+00
8.81023824e-01 4.32083219e-01 -5.09785831e-01 1.03712046e+00
2.59640425e-01 5.24769783e-01 1.08496487e+00 -1.12847388e+00
5.60648322e-01 2.30947763e-01 -5.65185584e-02 -7.58855879e-01
5.78987896e-02 -2.48149350e-01 -8.17590117e-01 -5.14911294e-01
-4.69954222e-01 -8.00388902e-02 -1.33741486e+00 1.57992351e+00
-1.19269714e-01 2.38273501e-01 6.95804926e-03 8.59047472e-01
7.81766772e-01 8.67362678e-01 -4.56054583e-02 -7.43106082e-02
1.21495903e+00 -1.04432249e+00 -9.93823051e-01 6.61878213e-02
1.27087903e+00 -4.83251125e-01 1.44851589e+00 4.26107883e-01
-8.18936348e-01 -2.78623492e-01 -1.05752146e+00 -4.32184279e-01
-5.82880855e-01 4.01070386e-01 6.50467694e-01 1.00071740e+00
-1.33071196e+00 5.19123316e-01 -9.84339714e-01 5.02142049e-02
6.11999154e-01 5.09028554e-01 -3.07779729e-01 -1.14006333e-01
-1.59906411e+00 5.74892521e-01 8.50426674e-01 -2.49382421e-01
-1.01047122e+00 -8.33788991e-01 -8.91298413e-01 4.63436306e-01
3.02320212e-01 -3.75351936e-01 1.42241371e+00 -3.15991879e-01
-1.24806237e+00 7.98791181e-03 -6.06025279e-01 -1.19541585e+00
-2.62269139e-01 -2.90813744e-01 -4.58847582e-01 2.00979918e-01
-3.81649762e-01 1.20511305e+00 8.47991467e-01 -8.61113846e-01
-5.76831937e-01 2.21087597e-02 -1.79364890e-01 1.06155090e-01
-1.48866534e+00 6.91411421e-02 -8.38226855e-01 -1.17145610e+00
-8.96674767e-02 -4.58336264e-01 -8.84979293e-02 1.29306570e-01
-4.78425890e-01 -5.96094012e-01 1.04168642e+00 -9.32023048e-01
1.95306098e+00 -1.95895350e+00 -2.48348847e-01 1.02448940e-01
5.10317922e-01 6.85686946e-01 -3.40422213e-01 3.19726050e-01
3.53228420e-01 6.14206970e-01 1.98074088e-01 -8.96019816e-01
4.24917489e-01 6.93290293e-01 -7.02929378e-01 1.48659691e-01
4.00401391e-02 1.40981269e+00 -5.30574918e-01 -5.63967824e-01
2.04154283e-01 6.53530478e-01 -7.38900602e-01 -4.67166230e-02
-4.16729122e-01 -6.22379601e-01 -6.15073331e-02 7.39202201e-01
4.99529332e-01 -1.25713095e-01 -8.95553827e-02 -3.11499178e-01
3.80272716e-01 6.67371809e-01 -9.07532334e-01 1.67941749e+00
-6.54863238e-01 8.96057129e-01 -1.80327967e-01 -8.53636086e-01
6.30789101e-01 4.55683917e-01 -5.36522791e-02 -9.97813523e-01
-1.21797808e-03 5.64118549e-02 -6.19804680e-01 2.35027522e-02
1.15853333e+00 1.91176206e-01 1.45399094e-01 4.72271204e-01
2.70684272e-01 2.32629657e-01 9.36338454e-02 4.49465334e-01
8.64860833e-01 -5.44605136e-01 -2.05022190e-02 5.03466418e-03
-1.13247804e-01 -4.18481141e-01 4.14766580e-01 8.45215261e-01
-3.38058546e-02 3.26248974e-01 4.71588783e-02 -4.64690894e-01
-1.28775918e+00 -5.96956372e-01 3.14311422e-02 1.33903670e+00
-2.53019482e-01 -1.23835731e+00 -7.69334018e-01 -2.59679824e-01
2.41309717e-01 7.43833542e-01 -5.65298259e-01 -5.32221973e-01
-4.01407897e-01 -5.22271752e-01 1.05258477e+00 7.03291476e-01
3.76427695e-02 -5.84167361e-01 -3.72914255e-01 1.88980192e-01
4.59709093e-02 -1.00954235e+00 -7.75544047e-01 5.69606364e-01
-9.18424010e-01 -2.74968833e-01 -9.53958213e-01 -6.30338252e-01
1.64269134e-01 4.33340788e-01 8.96999061e-01 1.97529033e-01
-1.46649629e-01 4.10759263e-02 -5.57556093e-01 -3.34605902e-01
-2.29567960e-01 4.28956568e-01 4.95144516e-01 -3.63657534e-01
6.10458910e-01 -6.16237342e-01 -6.28337920e-01 -8.84568319e-02
-1.37043571e+00 -4.11634631e-02 5.41163504e-01 8.55425000e-01
8.32451701e-01 1.25025243e-01 3.88094723e-01 -3.70262563e-01
1.08164966e+00 -1.83111221e-01 -6.26751125e-01 4.56438720e-01
-1.19719255e+00 3.83098811e-01 6.66035771e-01 -7.27245033e-01
-8.61979350e-02 -2.60108799e-01 -2.17306167e-01 -1.23099303e+00
2.21346959e-01 6.12477362e-01 8.07689130e-03 -4.00667936e-01
6.76673353e-01 6.42131925e-01 -4.00898784e-01 -9.15992439e-01
7.66824186e-01 9.51456070e-01 4.20819253e-01 -2.99703807e-01
5.98985255e-01 6.94864318e-02 -4.76054341e-01 -6.08109772e-01
-3.46263826e-01 -2.96414167e-01 -3.35829020e-01 2.16191322e-01
4.75905478e-01 -1.13148963e+00 -5.44792354e-01 5.96036017e-02
-1.21381652e+00 -3.23433220e-01 -2.75707453e-01 3.30653250e-01
-1.40972212e-01 4.21649605e-01 -8.06382954e-01 -8.58698726e-01
-7.30393469e-01 -1.06137693e+00 1.28558922e+00 -4.81459126e-02
-1.24371372e-01 -8.85876775e-01 -2.46470854e-01 5.98181747e-02
6.44103885e-01 -6.54582858e-01 9.76834476e-01 -6.64882004e-01
-4.66096818e-01 -3.28533351e-01 -5.71577191e-01 5.89452922e-01
2.44597383e-02 -5.55073202e-01 -9.18199003e-01 -3.55891377e-01
-4.05829459e-01 -6.49060130e-01 1.24769139e+00 7.08508343e-02
1.93651175e+00 -6.64396465e-01 -2.03504309e-01 8.55047703e-01
1.46819818e+00 -3.14854495e-02 4.61100876e-01 1.38232738e-01
1.03851402e+00 -1.90280035e-01 3.43551904e-01 6.84478223e-01
3.77281666e-01 8.08192670e-01 2.31790632e-01 7.70403147e-02
-4.53081846e-01 -6.42267644e-01 1.29563600e-01 1.18328881e+00
7.50361741e-01 -4.57046568e-01 -9.58507001e-01 9.14770007e-01
-1.68534744e+00 -4.17665988e-01 6.30946100e-01 1.72754669e+00
1.42606437e+00 2.07107544e-01 -1.52636379e-01 5.54622591e-01
3.76178265e-01 5.17790951e-02 -7.08978236e-01 -6.59010589e-01
-1.24909990e-01 4.96475607e-01 1.16645169e+00 6.07661307e-01
-9.06956673e-01 1.38447678e+00 7.56930780e+00 1.75628591e+00
-1.18214297e+00 3.24863613e-01 6.49370849e-01 -4.57535684e-01
-8.23708475e-01 -2.40159228e-01 -1.28958309e+00 4.90347236e-01
1.60859716e+00 -4.03078377e-01 7.72509456e-01 8.40530276e-01
1.86380014e-01 5.60522735e-01 -1.01294243e+00 1.51059127e+00
1.00047842e-01 -1.88108575e+00 9.74959791e-01 1.63495526e-01
4.79378521e-01 2.15794221e-01 2.03598872e-01 2.43575260e-01
3.45184416e-01 -1.41834283e+00 7.59347916e-01 3.41299206e-01
1.25438666e+00 -9.50607359e-01 4.35570240e-01 1.99811772e-01
-1.04780936e+00 -9.99581292e-02 -8.42877567e-01 7.86695331e-02
6.46319762e-02 6.81587696e-01 -1.35552669e+00 6.14057444e-02
6.10974729e-01 3.98109585e-01 -5.31982601e-01 7.37788498e-01
1.29997969e-01 9.48345304e-01 -6.54117286e-01 -3.04520547e-01
4.09251899e-01 4.55277175e-01 3.89315300e-02 1.34028208e+00
2.62814194e-01 -9.29432735e-02 -4.62543743e-04 9.41786706e-01
-6.10552967e-01 5.57980463e-02 -1.92600489e-01 -5.77835500e-01
1.08596158e+00 7.48728156e-01 -3.90863210e-01 -6.22815192e-01
-1.95786521e-01 1.19035184e+00 3.77278596e-01 3.78314018e-01
-5.37916005e-01 -7.29348838e-01 1.00079107e+00 -2.16250211e-01
7.47307777e-01 -5.35055041e-01 -2.48729035e-01 -1.05129027e+00
6.23088665e-02 -7.16716111e-01 -1.36510348e-02 -6.95410550e-01
-5.52826107e-01 2.85109967e-01 1.92898661e-02 -8.39881897e-01
-2.46534735e-01 -4.87457603e-01 2.10248709e-01 1.08379424e+00
-1.78086698e+00 -9.08621490e-01 2.84642726e-01 5.07830918e-01
6.48577034e-01 -2.27502272e-01 1.12311506e+00 7.23531187e-01
-3.71615559e-01 1.58127820e+00 4.63787287e-01 2.55286068e-01
3.50455433e-01 -1.15874469e+00 9.83810723e-01 7.71548152e-01
8.66750240e-01 8.16851020e-01 3.61113816e-01 -4.01855618e-01
-1.75391960e+00 -1.37294006e+00 1.19866860e+00 -4.46977578e-02
5.61881483e-01 -6.77273214e-01 -8.69871795e-01 3.63522917e-01
-3.63194086e-02 9.31379125e-02 8.61837685e-01 -2.95449257e-01
-7.86817551e-01 -3.54929753e-02 -8.87394071e-01 7.01738358e-01
8.52057576e-01 -1.28081644e+00 -2.59830058e-01 4.93976116e-01
1.67635262e+00 -2.59004861e-01 -1.04014480e+00 2.02944145e-01
4.43817884e-01 -1.27837956e-02 1.17893004e+00 -6.14456177e-01
8.88359826e-03 7.47760981e-02 -6.91494346e-01 -1.05911565e+00
-1.85268089e-01 -5.34657121e-01 -9.49352801e-01 8.84847999e-01
4.60069597e-01 -2.36792639e-01 1.05232346e+00 6.01940155e-01
1.73080191e-01 -9.36424077e-01 -1.27435362e+00 -7.48997808e-01
2.72018135e-01 -9.34003234e-01 8.92620742e-01 6.67555153e-01
-2.04342231e-01 -1.61161460e-02 -5.51643133e-01 3.62524167e-02
4.52238083e-01 -2.93676943e-01 3.21024477e-01 -6.79502904e-01
-2.13455528e-01 -3.82653087e-01 -4.44558561e-01 -1.59639668e+00
2.50961095e-01 -9.65557933e-01 -7.29769841e-02 -1.38873124e+00
-1.25355572e-01 -2.15975836e-01 -6.76243782e-01 1.07115364e+00
-9.94949266e-02 6.60173833e-01 1.80164784e-01 2.74893612e-01
-8.18292797e-01 1.00743759e+00 7.15159714e-01 -3.60307693e-01
2.02606902e-01 -7.05980718e-01 -7.07341731e-01 -5.02853915e-02
5.71313679e-01 -5.68556190e-01 -6.26036704e-01 -1.00941932e+00
2.53040820e-01 -2.84632593e-01 1.14171796e-01 -9.71198618e-01
4.23820704e-01 -4.48334776e-02 2.34771505e-01 -6.07653737e-01
5.26308894e-01 -6.41278088e-01 -3.89073789e-01 3.07524621e-01
-8.32018495e-01 6.84258267e-02 4.74835664e-01 5.37661195e-01
-1.80277556e-01 -1.56891957e-01 4.50372517e-01 3.88272405e-02
-9.44613755e-01 4.78915185e-01 -5.78002751e-01 -2.59333163e-01
3.72858196e-01 -1.19235933e-01 1.22597851e-01 -6.89595580e-01
-5.27492523e-01 2.23309964e-01 2.62718499e-01 6.14115894e-01
1.14543700e+00 -1.77369297e+00 -3.84810507e-01 3.35722834e-01
1.44329861e-01 -1.64058283e-01 -1.47241488e-01 3.13097537e-02
-4.40247953e-01 1.14625001e+00 3.68710369e-01 -2.32923001e-01
-1.37163985e+00 4.53430474e-01 1.76997021e-01 1.32029271e-02
-4.52742547e-01 1.44333971e+00 -3.57328951e-01 -3.46853942e-01
8.35855782e-01 -8.17574024e-01 6.24145605e-02 -2.34866276e-01
1.03580642e+00 2.08628848e-01 5.68304420e-01 -3.44369262e-01
-1.89966843e-01 -1.29387779e-02 -3.83440495e-01 -2.87930757e-01
9.65700567e-01 -2.24982366e-01 5.03377505e-02 2.32001171e-01
1.74810469e+00 -7.39892662e-01 -9.71815109e-01 -7.96764672e-01
6.65092096e-02 -5.40187240e-01 9.45909917e-01 -5.52131653e-01
-1.18402350e+00 1.02327788e+00 8.27258646e-01 -7.41363764e-02
1.18296254e+00 -3.62203836e-01 1.55251598e+00 1.05124450e+00
3.64713758e-01 -1.27890205e+00 -2.55706400e-01 5.71073532e-01
5.39600670e-01 -7.64412224e-01 -1.01152830e-01 1.70525789e-01
-5.45780696e-02 9.61288452e-01 2.41146132e-01 5.99906072e-02
8.01448524e-01 4.71996456e-01 -1.04806878e-01 6.61895350e-02
-1.19688666e+00 1.35514408e-01 4.19965416e-01 5.28823435e-01
-1.15182899e-01 1.59993216e-01 9.20313746e-02 7.03550994e-01
-4.68882769e-01 2.80098706e-01 1.08324565e-01 9.57070470e-01
-6.15272462e-01 -1.30265152e+00 -2.15118453e-01 8.47654462e-01
-4.01781559e-01 -9.32455957e-01 -3.15018237e-01 -1.09162353e-01
-1.04042955e-01 6.56833768e-01 1.97893694e-01 -9.75664616e-01
-1.89612061e-02 3.16083193e-01 1.64948925e-01 -5.77869952e-01
-2.33113021e-01 -1.12423953e-02 -1.01770796e-01 -6.93910241e-01
3.31296772e-02 8.38638172e-02 -1.02928257e+00 -5.45174479e-01
-6.16834760e-01 2.90191442e-01 9.50607657e-01 9.18054104e-01
8.57700646e-01 4.20660913e-01 3.73781264e-01 -7.75944591e-01
-8.58749747e-01 -1.11881649e+00 -5.64939857e-01 -1.37511730e-01
6.99427605e-01 -2.87331998e-01 -1.95841283e-01 2.13865675e-02]
|
[8.713595390319824, 3.374701738357544]
|
a2e90714-add9-416e-be37-993a827d7ee9
|
on-the-complexity-of-multi-agent-decision
|
2305.00684
| null |
https://arxiv.org/abs/2305.00684v1
|
https://arxiv.org/pdf/2305.00684v1.pdf
|
On the Complexity of Multi-Agent Decision Making: From Learning in Games to Partial Monitoring
|
A central problem in the theory of multi-agent reinforcement learning (MARL) is to understand what structural conditions and algorithmic principles lead to sample-efficient learning guarantees, and how these considerations change as we move from few to many agents. We study this question in a general framework for interactive decision making with multiple agents, encompassing Markov games with function approximation and normal-form games with bandit feedback. We focus on equilibrium computation, in which a centralized learning algorithm aims to compute an equilibrium by controlling multiple agents that interact with an unknown environment. Our main contributions are: - We provide upper and lower bounds on the optimal sample complexity for multi-agent decision making based on a multi-agent generalization of the Decision-Estimation Coefficient, a complexity measure introduced by Foster et al. (2021) in the single-agent counterpart to our setting. Compared to the best results for the single-agent setting, our bounds have additional gaps. We show that no "reasonable" complexity measure can close these gaps, highlighting a striking separation between single and multiple agents. - We show that characterizing the statistical complexity for multi-agent decision making is equivalent to characterizing the statistical complexity of single-agent decision making, but with hidden (unobserved) rewards, a framework that subsumes variants of the partial monitoring problem. As a consequence, we characterize the statistical complexity for hidden-reward interactive decision making to the best extent possible. Building on this development, we provide several new structural results, including 1) conditions under which the statistical complexity of multi-agent decision making can be reduced to that of single-agent, and 2) conditions under which the so-called curse of multiple agents can be avoided.
|
['Alexander Rakhlin', 'Noah Golowich', 'Dean P. Foster', 'Dylan J. Foster']
|
2023-05-01
| null | null | null | null |
['multi-agent-reinforcement-learning']
|
['methodology']
|
[ 5.44704869e-02 4.75675493e-01 -4.61054921e-01 6.87438026e-02
-9.37530160e-01 -7.49715388e-01 4.68753666e-01 2.44971618e-01
-8.71384501e-01 1.03907847e+00 -1.77821234e-01 -5.21047533e-01
-7.81590581e-01 -7.21506476e-01 -5.84954739e-01 -1.00430322e+00
-4.68067527e-01 7.38334417e-01 7.48045817e-02 -1.42799884e-01
-7.40378816e-03 1.72895938e-01 -1.27746868e+00 -2.05982089e-01
8.69974732e-01 8.10864389e-01 1.17820941e-01 1.14814353e+00
3.62646997e-01 1.11754572e+00 -5.20543396e-01 -2.33163536e-01
5.46459138e-01 -5.85572660e-01 -8.07852447e-01 4.37017500e-01
-3.20986092e-01 -7.21923232e-01 -1.29371718e-01 1.38973868e+00
4.88626420e-01 2.09393889e-01 8.03943276e-01 -1.48359549e+00
-3.28225166e-01 1.05511725e+00 -6.22204840e-01 1.33145183e-01
1.77908972e-01 3.28502893e-01 1.31325185e+00 2.62173831e-01
4.04474646e-01 1.40855253e+00 2.76368558e-01 7.30094016e-01
-1.41486013e+00 -3.11072826e-01 5.52219033e-01 -2.64469199e-02
-7.38204718e-01 -2.25941822e-01 2.44683892e-01 -5.03829539e-01
7.39831030e-01 2.51254052e-01 6.30416214e-01 7.88800240e-01
1.86502784e-01 1.19533038e+00 1.51506555e+00 -7.45787382e-01
6.22191131e-01 -1.43040329e-01 2.05228999e-01 6.73787713e-01
5.82307816e-01 5.91131985e-01 9.71990824e-03 -3.26102138e-01
8.17535102e-01 4.64744270e-02 -7.01189740e-03 -6.77180231e-01
-9.69537675e-01 1.23004580e+00 -2.55830318e-01 -2.77520269e-02
-6.22933149e-01 3.95139873e-01 1.67105108e-01 9.96505678e-01
2.48958781e-01 3.73853177e-01 -5.06143928e-01 -8.79201144e-02
-4.43059623e-01 6.86785698e-01 1.18125772e+00 7.15806842e-01
8.90443623e-01 1.70700163e-01 -1.54160857e-01 4.50927287e-01
3.70795220e-01 6.83945477e-01 2.02234596e-01 -1.51830375e+00
5.98798990e-01 -6.38424456e-02 8.46215785e-01 -3.01780283e-01
-5.33728778e-01 -5.19597769e-01 -4.94964927e-01 8.03048968e-01
7.18809962e-01 -9.65203881e-01 -1.83368817e-01 2.10583448e+00
1.95781514e-01 -2.55762637e-01 3.41747850e-01 5.90624690e-01
-3.82752836e-01 5.04026055e-01 -3.11524659e-01 -9.33740795e-01
9.90182638e-01 -7.98347175e-01 -5.02232134e-01 -1.76442653e-01
6.79480374e-01 -2.37783998e-01 6.61377788e-01 5.08084834e-01
-1.30541289e+00 1.56806678e-01 -9.17574108e-01 8.01575422e-01
1.58342749e-01 -4.83624190e-01 7.01484978e-01 8.27866554e-01
-1.13468838e+00 6.52222037e-01 -8.83110583e-01 -1.33390054e-01
-4.82975990e-02 4.93868589e-01 2.64982462e-01 2.66789794e-01
-9.32398379e-01 8.70375812e-01 2.15079442e-01 -2.51859576e-01
-1.38800013e+00 -3.05291414e-01 -4.95382071e-01 6.49651065e-02
1.28326917e+00 -8.53794158e-01 2.17455363e+00 -1.38024616e+00
-1.94817317e+00 2.82250077e-01 2.38074213e-01 -7.12546349e-01
8.80839527e-01 -2.26946324e-02 1.43231601e-01 7.58718699e-02
4.00334060e-01 -3.11194286e-02 6.63321316e-01 -1.22868204e+00
-1.17158961e+00 -4.07993942e-01 6.84984446e-01 5.93936443e-01
2.28048965e-01 -4.00758721e-03 3.85553151e-01 -3.26060951e-01
-5.89913487e-01 -1.19459581e+00 -7.40302563e-01 -7.06498206e-01
-3.10042560e-01 -4.06302661e-01 1.35953367e-01 2.87318788e-02
9.83554482e-01 -1.66788828e+00 1.64717048e-01 2.19875231e-01
3.23026538e-01 -3.05705994e-01 -2.94560790e-01 6.07747257e-01
8.23353678e-02 2.46208638e-01 -8.12184252e-03 -2.66232967e-01
7.07429051e-01 2.51372099e-01 -2.04718843e-01 5.97866535e-01
-2.26832971e-01 7.37862229e-01 -8.65277112e-01 -1.44693285e-01
-5.76619394e-02 -7.11406469e-01 -8.15514266e-01 9.98782590e-02
-5.78661978e-01 3.22098583e-02 -9.16664898e-01 1.96972296e-01
2.15544909e-01 -2.07692862e-01 4.71025467e-01 9.45029080e-01
-2.38584384e-01 1.00216873e-01 -1.65249383e+00 9.31309164e-01
-2.87130773e-01 1.56824321e-01 7.31898487e-01 -1.15548384e+00
1.21732336e-02 3.72042924e-01 7.00322688e-01 -2.32618332e-01
2.82308906e-01 6.84053823e-02 2.51910031e-01 -4.31823552e-01
3.74245107e-01 -3.58194083e-01 -4.06545520e-01 1.02685380e+00
-6.61071986e-02 6.23546494e-03 4.43172455e-01 1.17596000e-01
1.16075957e+00 -1.58807591e-01 5.98999560e-01 -4.58945513e-01
-3.81146446e-02 -5.53261861e-02 6.88632309e-01 1.55898309e+00
-3.41704249e-01 -4.02760208e-01 1.02903962e+00 -6.51677325e-02
-8.04975450e-01 -9.44078028e-01 4.89535272e-01 1.35057092e+00
-2.18174867e-02 -8.57576579e-02 -9.44785655e-01 -4.07927185e-01
3.51604700e-01 5.82111657e-01 -8.22846651e-01 3.06451857e-01
-1.16329722e-01 -1.03233564e+00 1.66224107e-01 -3.46293375e-02
5.36583424e-01 -8.50206316e-01 -7.95690179e-01 5.26489139e-01
-1.20924031e-02 -7.30256200e-01 -4.79902238e-01 2.42920712e-01
-7.80929267e-01 -1.34574842e+00 -5.93146324e-01 -1.93637639e-01
1.77654326e-01 1.27153084e-01 8.34042311e-01 -1.25159249e-01
3.17578137e-01 1.00719404e+00 -1.43773854e-01 -7.73878336e-01
-6.93697453e-01 1.44814134e-01 4.08775806e-01 4.87573147e-02
3.32264118e-02 -2.76752293e-01 -5.90104103e-01 1.68321609e-01
-8.45135093e-01 -1.52035505e-01 5.34286201e-01 7.97279179e-01
1.79108262e-01 2.54593849e-01 6.74359977e-01 -7.96191633e-01
9.95580733e-01 -4.43838060e-01 -1.31076372e+00 3.27839345e-01
-7.00255990e-01 3.59678358e-01 5.52002251e-01 -4.36575115e-01
-8.58825147e-01 -1.78689197e-01 4.81310248e-01 1.35658354e-01
5.54136187e-02 3.95570934e-01 -2.25366317e-02 1.03491455e-01
4.08245832e-01 1.92203060e-01 4.67537671e-01 -1.15406953e-01
2.38581747e-01 5.49534976e-01 -5.01746722e-02 -1.06169295e+00
5.49464166e-01 1.51758015e-01 1.78930134e-01 -5.37801802e-01
-6.55730546e-01 -9.78673398e-02 -1.08414762e-01 -2.08348826e-01
6.19253814e-01 -8.40804100e-01 -1.56815434e+00 5.62574327e-01
-8.92550588e-01 -9.91977751e-01 -5.65445125e-01 6.45711243e-01
-1.40074873e+00 4.39284384e-01 -7.44224727e-01 -1.63257957e+00
2.71952540e-01 -1.13724613e+00 3.83563668e-01 2.45130092e-01
2.66246319e-01 -9.58499372e-01 4.85846370e-01 2.77583778e-01
1.29849777e-01 -9.10879765e-03 7.43816137e-01 -6.70370758e-01
-7.93046296e-01 3.40936065e-01 3.62744689e-01 1.34044409e-01
-2.57991880e-01 -1.27214298e-01 -4.85555232e-01 -6.60425127e-01
8.70801657e-02 -3.57346714e-01 6.73394084e-01 9.10819113e-01
4.87570912e-01 -7.56001472e-01 -1.10099092e-01 -1.43785238e-01
1.47741306e+00 6.17242932e-01 2.00590659e-02 6.14211619e-01
-3.27142843e-05 6.14502847e-01 5.99539816e-01 8.85481298e-01
5.35339475e-01 4.40313548e-01 5.80195606e-01 4.45169508e-01
8.44500184e-01 -1.21617690e-01 7.20259309e-01 4.01639432e-01
-2.80735642e-01 -2.92231679e-01 -4.24158245e-01 3.75171423e-01
-2.39815640e+00 -1.39175892e+00 3.62037241e-01 2.53622913e+00
8.41395557e-01 1.28539786e-01 9.17332888e-01 -8.23564604e-02
6.93124115e-01 -9.16119665e-02 -9.58842933e-01 -5.28950930e-01
-3.43861058e-02 -1.43708810e-01 9.26351786e-01 1.15566349e+00
-9.92955625e-01 6.54467821e-01 7.13374424e+00 7.93972492e-01
-4.01645929e-01 4.19689685e-01 6.25718772e-01 -3.45815182e-01
-9.67648774e-02 6.66934252e-02 -7.85723329e-01 2.96760172e-01
9.94595110e-01 -4.80798960e-01 1.00713396e+00 9.06247914e-01
6.14161253e-01 -4.96219128e-01 -1.21358633e+00 6.26204133e-01
-6.02299273e-01 -1.03148055e+00 -5.13475716e-01 6.78946316e-01
1.06905222e+00 -9.04196408e-03 -7.13713095e-02 5.16217589e-01
1.59614825e+00 -8.40613365e-01 8.16902161e-01 3.70599031e-01
3.89034837e-01 -9.90146697e-01 4.61967766e-01 8.63277733e-01
-8.40159953e-01 -7.22705722e-01 -1.52580187e-01 -6.83000505e-01
-2.52165437e-01 2.23787844e-01 -3.71872663e-01 5.11020958e-01
2.21848220e-01 1.31544381e-01 2.12004885e-01 9.13287759e-01
-6.86093420e-02 4.48983848e-01 -3.90303522e-01 -4.79468226e-01
5.75027704e-01 -6.00167453e-01 6.35906339e-01 8.17214131e-01
1.09867722e-01 2.35413939e-01 6.61614418e-01 5.61289251e-01
1.05649419e-01 8.62388127e-03 -5.07241249e-01 1.29802385e-02
3.18794340e-01 9.28238392e-01 -5.41339576e-01 -2.56034225e-01
-4.04016018e-01 6.77373528e-01 4.69526082e-01 5.66438019e-01
-4.75641221e-01 -2.12814972e-01 9.81896102e-01 -2.50603050e-01
4.18585628e-01 -1.40843928e-01 1.12446705e-02 -1.20554733e+00
-1.17306381e-01 -1.02299047e+00 6.88700318e-01 -1.19417094e-01
-1.18515134e+00 -9.83657390e-02 2.03305140e-01 -8.10427964e-01
-1.04802716e+00 -4.47723806e-01 -4.37590927e-01 4.62528676e-01
-1.34303319e+00 -5.10195851e-01 7.27248311e-01 5.80470443e-01
5.36190391e-01 -1.76299036e-01 7.19330847e-01 -2.80729741e-01
-5.78970790e-01 3.91732156e-01 8.27070713e-01 -4.49017473e-02
7.13875964e-02 -1.84750104e+00 -3.67284387e-01 8.11852515e-01
-1.35849491e-01 8.39025900e-02 7.70162582e-01 -2.15976477e-01
-1.47724187e+00 -7.06944883e-01 1.15843229e-01 -2.64775276e-01
1.08079636e+00 -5.56155927e-02 -1.19816154e-01 9.54207957e-01
3.61425608e-01 -5.47947645e-01 4.91519541e-01 3.24201018e-01
1.04812093e-01 -1.19659245e-01 -1.00414908e+00 8.82424831e-01
7.98020065e-01 -2.49218315e-01 -4.56914067e-01 2.98091948e-01
6.21756911e-01 -8.48837942e-02 -7.02556968e-01 -1.73064530e-01
4.24571246e-01 -8.98990273e-01 4.86200929e-01 -8.87486279e-01
9.96285975e-02 -1.16548963e-01 -1.64574847e-01 -1.73810995e+00
-4.27037954e-01 -1.58309424e+00 -7.53909815e-03 7.01348543e-01
4.31346506e-01 -1.05832386e+00 6.95948839e-01 6.93464875e-01
1.81217775e-01 -5.45629144e-01 -1.09763134e+00 -1.02970779e+00
7.16204166e-01 -2.81487316e-01 3.70831966e-01 6.13850474e-01
2.60503739e-01 2.74452865e-01 -4.64885354e-01 2.39604324e-01
1.05302596e+00 2.38061100e-01 6.13092005e-01 -1.10634112e+00
-1.05203831e+00 -8.12697232e-01 6.05381913e-02 -1.31276453e+00
2.99574912e-01 -3.90340835e-01 4.39563096e-02 -1.19348526e+00
5.50583959e-01 -4.59125519e-01 -2.63757885e-01 1.39952049e-01
4.37032506e-02 -5.34442663e-01 6.40069008e-01 1.26185939e-01
-1.13887060e+00 3.84627700e-01 1.19041765e+00 -3.60469357e-03
-2.72559375e-01 6.81346357e-01 -9.16025162e-01 7.91179776e-01
8.85329723e-01 -4.97303665e-01 -4.10352319e-01 -6.26843721e-02
3.52268040e-01 8.04773629e-01 2.74269342e-01 -5.73231876e-01
3.32202047e-01 -9.32943940e-01 -3.26896727e-01 -2.77096659e-01
-1.08965382e-01 -5.97238719e-01 -4.58185226e-02 9.36698914e-01
-7.37084925e-01 2.12658644e-01 -3.60954911e-01 9.23572004e-01
3.28304291e-01 -8.00223410e-01 7.00435340e-01 -3.86801779e-01
-2.99656987e-01 3.10650349e-01 -1.24261045e+00 3.56074035e-01
1.22696018e+00 1.99266583e-01 -1.98459134e-01 -1.07827246e+00
-1.03714979e+00 5.95558345e-01 3.16373676e-01 -2.60957330e-01
1.49083491e-02 -9.36514556e-01 -7.55658329e-01 -3.27532023e-01
-2.83905268e-01 -2.72431582e-01 1.45728782e-01 6.60404801e-01
6.27392754e-02 3.12629551e-01 2.51709707e-02 -1.42414719e-01
-1.12938440e+00 5.69558024e-01 7.96071589e-01 -7.97934055e-01
-7.09487870e-02 2.61522263e-01 1.73525289e-01 -3.04484457e-01
2.40117937e-01 -5.69544196e-01 1.60691887e-01 7.03756660e-02
5.43245137e-01 4.78109926e-01 -4.82193410e-01 7.20936209e-02
2.05475166e-01 1.81970060e-01 -2.01420084e-01 -8.48632812e-01
1.29426718e+00 -5.61874270e-01 1.17720269e-01 6.82606280e-01
4.68110412e-01 -8.70732814e-02 -1.64279270e+00 -5.53295553e-01
7.42946491e-02 -3.91054992e-03 -1.26451686e-01 -7.07202137e-01
-7.07757056e-01 4.19032216e-01 4.24190164e-01 9.76890922e-01
7.53341317e-01 -9.59620252e-02 -3.24782468e-02 8.03219914e-01
7.93720722e-01 -1.54399836e+00 -2.25157533e-02 6.84503853e-01
5.81830502e-01 -1.12321913e+00 -1.66321561e-01 8.27631578e-02
-7.64841497e-01 9.24744368e-01 1.38173118e-01 -2.69906461e-01
6.05450273e-01 3.27093065e-01 -3.66587102e-01 1.10056326e-01
-1.27835345e+00 -7.37149954e-01 -5.48278689e-01 7.78168917e-01
-2.81787723e-01 6.53741837e-01 -4.81945910e-02 7.03638852e-01
-1.97890788e-01 8.51871222e-02 1.04575300e+00 9.25434053e-01
-7.70925641e-01 -1.08981061e+00 -4.69252765e-01 5.80322564e-01
-8.16479623e-01 1.89432904e-01 -1.63389578e-01 8.73899937e-01
-4.50710565e-01 1.38176095e+00 -1.66944191e-02 1.61296308e-01
1.54725671e-01 -1.18599795e-01 6.92714691e-01 -5.84090769e-01
-3.49490404e-01 2.82736510e-01 1.59011111e-01 -3.23675990e-01
-6.84033275e-01 -9.13314164e-01 -7.72993386e-01 -6.95738614e-01
-3.22514236e-01 3.54608595e-01 1.96351632e-01 1.15369368e+00
-7.14489147e-02 2.47750580e-01 9.67474759e-01 -5.67796052e-01
-1.76270676e+00 -8.73849571e-01 -1.15879571e+00 -2.12107182e-01
7.83431530e-01 -6.06437504e-01 -5.93529165e-01 -5.12189567e-01]
|
[4.220224857330322, 2.721081495285034]
|
3879ecae-7be2-4928-8c04-9f0aa552ea90
|
multi-rate-adaptive-transform-coding-for
|
2210.14308
| null |
https://arxiv.org/abs/2210.14308v2
|
https://arxiv.org/pdf/2210.14308v2.pdf
|
Multi-rate adaptive transform coding for video compression
|
Contemporary lossy image and video coding standards rely on transform coding, the process through which pixels are mapped to an alternative representation to facilitate efficient data compression. Despite impressive performance of end-to-end optimized compression with deep neural networks, the high computational and space demands of these models has prevented them from superseding the relatively simple transform coding found in conventional video codecs. In this study, we propose learned transforms and entropy coding that may either serve as (non)linear drop-in replacements, or enhancements for linear transforms in existing codecs. These transforms can be multi-rate, allowing a single model to operate along the entire rate-distortion curve. To demonstrate the utility of our framework, we augmented the DCT with learned quantization matrices and adaptive entropy coding to compress intra-frame AV1 block prediction residuals. We report substantial BD-rate and perceptual quality improvements over more complex nonlinear transforms at a fraction of the computational cost.
|
['Jingning Han', 'Cheng Chen', 'Bohan Li', 'Lyndon R. Duong']
|
2022-10-25
| null | null | null | null |
['data-compression']
|
['time-series']
|
[ 7.34648943e-01 8.91145095e-02 -5.25703132e-01 -3.77388358e-01
-7.22583652e-01 -1.80040985e-01 4.86024022e-01 -2.21707955e-01
-2.76406080e-01 5.27105868e-01 5.98804235e-01 -3.65570277e-01
1.00979827e-01 -7.23785102e-01 -7.36989737e-01 -4.41459507e-01
-3.03708702e-01 7.63576918e-06 1.73973322e-01 -1.24878071e-01
1.93483815e-01 4.69841897e-01 -1.45905614e+00 7.61247456e-01
5.80597699e-01 1.35908973e+00 2.13906452e-01 9.65355456e-01
1.35655627e-01 1.32055449e+00 -2.20265597e-01 -6.46090806e-01
5.85444868e-01 -4.88132179e-01 -5.66235721e-01 1.29315928e-01
6.02060139e-01 -1.31393230e+00 -1.21668983e+00 9.75672781e-01
3.54017526e-01 7.16989636e-02 4.73889917e-01 -9.30593729e-01
-9.64210570e-01 4.44752574e-01 -2.74278969e-01 2.06394047e-01
1.74542099e-01 2.24043906e-01 1.31166470e+00 -8.31245542e-01
5.86804628e-01 1.25449014e+00 7.34686375e-01 5.06632447e-01
-1.25696874e+00 -5.47190785e-01 -4.22432154e-01 5.49193263e-01
-1.20173275e+00 -1.07745326e+00 5.12301505e-01 -8.56483504e-02
1.65428627e+00 2.06809774e-01 6.94752991e-01 6.89370453e-01
4.72512931e-01 6.11338019e-01 2.37743840e-01 -2.08512291e-01
2.59155303e-01 -3.12162817e-01 -7.40724146e-01 4.50410247e-01
4.11348157e-02 4.83957410e-01 -5.71315348e-01 1.31809473e-01
9.68300223e-01 1.76102176e-01 -4.04721767e-01 -3.55345219e-01
-8.02276790e-01 8.19664299e-01 6.74089909e-01 2.23798882e-02
-4.62669730e-01 7.51340508e-01 6.62514865e-01 6.04735076e-01
4.15480018e-01 6.35539070e-02 -2.50455737e-01 -4.78893548e-01
-1.47978330e+00 1.38472781e-01 5.60553014e-01 9.45657849e-01
4.94817942e-01 4.65200663e-01 -6.28002062e-02 8.31363261e-01
2.28162512e-01 2.90566772e-01 5.02428174e-01 -1.56229532e+00
7.03407168e-01 1.71625182e-01 -2.68672436e-01 -9.90615785e-01
2.17831552e-01 -2.86206484e-01 -1.08227241e+00 3.00633103e-01
-2.55040616e-01 4.58581567e-01 -9.59801495e-01 1.35841846e+00
-2.15197682e-01 -4.17280309e-02 5.83141707e-02 6.29059732e-01
1.39730766e-01 1.06115413e+00 -2.17896979e-02 -2.36367643e-01
8.43226492e-01 -1.00145614e+00 -8.48325849e-01 -3.24061662e-02
7.85064757e-01 -7.73918450e-01 6.91938579e-01 4.42291468e-01
-1.78947508e+00 -6.25173151e-01 -1.30499721e+00 -7.07146227e-01
1.40942052e-01 -3.38989943e-01 3.21564257e-01 4.37699139e-01
-1.53052652e+00 9.27605391e-01 -9.25628960e-01 1.09560840e-01
7.20245183e-01 4.83067840e-01 -2.17178628e-01 -3.36834490e-01
-9.23085272e-01 6.78153932e-01 4.93332833e-01 -3.73245388e-01
-1.00214815e+00 -7.70551085e-01 -8.90160859e-01 6.24543965e-01
-2.76470650e-02 -5.31688869e-01 1.31497204e+00 -1.10996711e+00
-1.23458374e+00 4.03987736e-01 -2.09058017e-01 -1.03663337e+00
3.97083223e-01 -3.21557432e-01 -3.54748845e-01 6.55792832e-01
-2.19121367e-01 1.00719237e+00 1.12940836e+00 -8.58032525e-01
-7.42034912e-01 1.45066902e-01 -5.54912575e-02 2.16103479e-01
-5.54300725e-01 -6.98324069e-02 -2.68029839e-01 -1.06396651e+00
1.30198479e-01 -5.88729262e-01 2.47887727e-02 8.85480821e-01
3.73053923e-02 3.87303323e-01 1.24059951e+00 -1.13522458e+00
1.44842756e+00 -2.36969018e+00 1.07782051e-01 -8.96881744e-02
4.45404470e-01 3.76191109e-01 -2.88500816e-01 2.23961368e-01
-2.24741518e-01 3.23167861e-01 -4.79143262e-01 -3.76126945e-01
-7.67294094e-02 3.38147491e-01 -4.80435371e-01 2.80668139e-01
2.32855693e-01 9.75781024e-01 -5.63076913e-01 -3.94636303e-01
3.09824258e-01 8.84238482e-01 -1.16139233e+00 1.05010562e-01
1.54201398e-02 -1.72658443e-01 1.85290445e-02 6.62263274e-01
5.99790275e-01 -1.93033978e-01 2.15677172e-02 -5.58742940e-01
1.44102097e-01 7.14304566e-01 -5.36775708e-01 1.66076136e+00
-3.43091905e-01 1.08879280e+00 3.70066091e-02 -1.03425062e+00
5.86001873e-01 4.54570353e-01 7.22607434e-01 -1.01358402e+00
-2.41681710e-02 3.12173188e-01 9.06511992e-02 -2.60460794e-01
9.51743901e-01 -2.40384787e-01 4.82450217e-01 3.13736856e-01
8.42285231e-02 -2.29311332e-01 1.64782822e-01 1.14323527e-01
1.24022985e+00 -1.40687386e-02 4.22292531e-01 6.50707558e-02
3.11810791e-01 -4.48629647e-01 3.57009798e-01 5.02739131e-01
-3.41890156e-01 7.61829734e-01 1.28894880e-01 -6.13029838e-01
-2.02724123e+00 -1.24739921e+00 -1.74329996e-01 1.01646340e+00
-6.52825385e-02 -5.97626150e-01 -3.80990088e-01 -1.87964886e-01
-2.33439311e-01 6.43703580e-01 4.35340591e-03 -5.77418268e-01
-8.86590004e-01 -2.47887641e-01 7.47968078e-01 5.27269721e-01
8.20268035e-01 -7.71766841e-01 -7.63290584e-01 5.15474796e-01
-2.68813491e-01 -1.03612471e+00 -5.50816715e-01 3.99158835e-01
-1.25099480e+00 -6.12979352e-01 -7.36406147e-01 -7.95486867e-01
2.33996347e-01 2.24246085e-01 1.16165257e+00 1.95136458e-01
-2.00547442e-01 7.30035976e-02 -4.21181172e-01 2.25787789e-01
-8.22987914e-01 -1.02333844e-01 -2.11814478e-01 -4.75231558e-01
2.53811367e-02 -9.73153412e-01 -9.65208352e-01 -3.73895727e-02
-1.24787033e+00 1.20542519e-01 5.55727959e-01 9.34964776e-01
6.55838311e-01 2.27520943e-01 2.27034420e-01 -2.77724206e-01
4.92045105e-01 -3.16905320e-01 -2.90590048e-01 -2.34424975e-02
-7.87020683e-01 2.06245586e-01 8.30571115e-01 -1.43273488e-01
-7.45566964e-01 -1.11922950e-01 -5.18976569e-01 -6.18788600e-01
3.14650625e-01 2.38459826e-01 1.38540387e-01 -3.44512463e-01
5.03839970e-01 5.92435300e-01 -2.95487512e-03 -2.53139645e-01
3.47056985e-01 6.69585705e-01 7.07351625e-01 -2.08025396e-01
8.11275899e-01 3.61657292e-01 -3.91925760e-02 -4.85527366e-01
-2.21760690e-01 1.30417362e-01 -3.15131009e-01 3.46607007e-02
7.94854879e-01 -1.26546717e+00 -2.95433253e-01 6.07679747e-02
-1.06759095e+00 -2.66922683e-01 -7.46184826e-01 3.35393548e-01
-1.01507795e+00 5.00952244e-01 -1.15697992e+00 -4.54602152e-01
-4.11516637e-01 -1.31172550e+00 8.57989252e-01 -2.40108430e-01
-1.43814027e-01 -7.79676318e-01 -2.24983752e-01 1.36496872e-01
8.59229743e-01 -1.19407125e-01 1.06738138e+00 -5.19601144e-02
-9.34065402e-01 -1.17641971e-01 -4.30932105e-01 7.92269289e-01
2.11382762e-01 -9.16274339e-02 -7.96233177e-01 -6.67556822e-01
2.28188485e-01 -5.20424604e-01 1.07151628e+00 5.32719076e-01
1.51383650e+00 -8.76659513e-01 8.87631104e-05 1.32707906e+00
1.55919671e+00 5.68959415e-01 1.11502421e+00 2.84881264e-01
5.85865259e-01 -6.34141192e-02 1.75141022e-02 6.79586291e-01
1.51994660e-01 5.12779951e-01 5.05707026e-01 1.12536307e-02
-5.10882974e-01 -3.86870265e-01 4.77187783e-01 1.10165441e+00
-1.70414746e-01 -4.24133211e-01 -5.84819257e-01 3.60417157e-01
-1.32053733e+00 -1.22973645e+00 6.13937974e-01 1.93179071e+00
9.67034161e-01 3.09814185e-01 -3.62418383e-01 4.96926785e-01
4.36856270e-01 5.20718753e-01 -7.90700078e-01 -7.43442297e-01
-1.48235306e-01 4.28057253e-01 8.32035840e-01 4.98031497e-01
-9.58708584e-01 5.89607596e-01 7.54503536e+00 8.61298442e-01
-1.13308668e+00 -6.68974221e-02 9.79269922e-01 -6.33230627e-01
-3.34772050e-01 -8.96628648e-02 -2.91657805e-01 3.51982325e-01
1.42333686e+00 -2.84658223e-01 9.94903266e-01 8.61085117e-01
1.85753971e-01 2.32876092e-01 -1.33382857e+00 1.25230312e+00
-3.28269005e-02 -1.67181921e+00 5.30441582e-01 3.24711025e-01
7.50167787e-01 5.52976392e-02 4.53095019e-01 1.78647786e-01
1.08334869e-01 -1.13321602e+00 8.64431798e-01 1.03382006e-01
1.37075591e+00 -6.67489707e-01 3.71004701e-01 -4.59915474e-02
-1.35653389e+00 -4.37365204e-01 -5.57347000e-01 -1.12514071e-01
4.23001349e-01 1.45057723e-01 -4.62306798e-01 4.93341945e-02
7.74442852e-01 1.09078455e+00 -2.38135472e-01 8.52366209e-01
2.63249576e-01 5.52384734e-01 -1.24230765e-01 6.75563395e-01
2.37560287e-01 1.68420151e-01 4.02726114e-01 1.04947484e+00
9.29633498e-01 -3.81016769e-02 -3.85165542e-01 6.90316260e-01
-4.90622282e-01 -3.14253241e-01 -7.09779143e-01 -1.27333447e-01
4.92077142e-01 3.88749719e-01 -3.06883782e-01 -5.55696964e-01
-5.10123909e-01 1.37169671e+00 3.52909304e-02 4.60349828e-01
-8.67625177e-01 -3.09011370e-01 9.46009040e-01 3.44123542e-01
7.81559408e-01 -2.22425699e-01 -2.95172483e-01 -1.06822896e+00
1.45372689e-01 -1.09308600e+00 -3.27436160e-03 -7.79868007e-01
-8.46736014e-01 4.46614772e-01 -1.34559423e-01 -1.55425441e+00
-5.15145183e-01 -4.15336728e-01 -1.09002320e-02 6.06578171e-01
-1.69758773e+00 -6.60141945e-01 3.63710895e-02 5.95276356e-01
8.86896133e-01 -3.20083052e-01 6.00012362e-01 6.61468804e-01
4.99269329e-02 8.84393036e-01 5.75195491e-01 -7.70977363e-02
4.45426017e-01 -6.53064668e-01 5.96179962e-01 1.01206362e+00
-9.42869633e-02 2.46894270e-01 6.64350808e-01 -4.11854953e-01
-1.39746392e+00 -1.11264980e+00 6.73930883e-01 2.90441096e-01
4.93471205e-01 -1.15131602e-01 -1.00693178e+00 7.57066429e-01
2.36035109e-01 1.53550759e-01 4.76566821e-01 -6.17094636e-01
-4.86868709e-01 -2.05772564e-01 -1.30621290e+00 6.64804399e-01
1.10137630e+00 -8.81313741e-01 -2.58487433e-01 -8.97157937e-02
1.10253656e+00 -2.67371655e-01 -9.98417139e-01 4.14393365e-01
7.00274944e-01 -1.32706451e+00 1.37376344e+00 -1.91964641e-01
1.27365255e+00 -7.14079058e-03 -7.47142673e-01 -9.63579416e-01
-6.34040534e-01 -5.51664948e-01 -7.72601128e-01 6.79627597e-01
2.90081762e-02 -2.09426284e-01 8.27868700e-01 6.65358424e-01
-3.35536033e-01 -7.63570189e-01 -1.31634831e+00 -6.87691569e-01
7.34542534e-02 -5.88521838e-01 5.49575210e-01 4.84816700e-01
-1.25892490e-01 -1.74706519e-01 -7.39433825e-01 -2.19188094e-01
5.49389005e-01 -5.64431012e-01 1.75136745e-01 -6.51827574e-01
-5.68479598e-01 -6.47220314e-01 -7.07274675e-01 -1.65257096e+00
-1.45996660e-01 -9.22424018e-01 4.98446859e-02 -1.27975154e+00
1.53591424e-01 -2.15103686e-01 -3.04186583e-01 4.09042358e-01
2.63264120e-01 5.14190614e-01 4.94684219e-01 4.98462707e-01
-3.70847732e-01 1.03621256e+00 1.27093458e+00 -4.76915479e-01
-5.73591143e-02 -6.17246687e-01 -5.01440883e-01 4.19048369e-01
6.56085730e-01 -4.62415397e-01 -7.42955089e-01 -8.92717004e-01
2.57046491e-01 3.74007761e-01 2.89755285e-01 -1.45947623e+00
1.30385727e-01 1.53320625e-01 5.02377748e-01 -4.06096756e-01
5.92171073e-01 -1.07562172e+00 1.69480577e-01 6.89333498e-01
-7.37589955e-01 2.83355504e-01 1.51313826e-01 5.11369407e-01
-5.94403446e-01 -1.52060564e-03 9.71515179e-01 -4.69105206e-02
-6.59678102e-01 5.04483700e-01 -4.61139321e-01 -1.73348814e-01
8.28230679e-01 -8.10008585e-01 -1.44318670e-01 -7.77201951e-01
-5.47523320e-01 -3.48841220e-01 7.25661635e-01 2.54979253e-01
1.17837417e+00 -1.56255388e+00 -6.14579856e-01 4.54284549e-01
-2.67158747e-01 -1.42382681e-01 2.21701801e-01 3.77089888e-01
-1.11316848e+00 4.43921983e-01 -5.87681711e-01 -3.82253051e-01
-1.10791087e+00 6.09614789e-01 3.64469320e-01 -2.42421851e-01
-8.32351625e-01 7.33497739e-01 2.82280892e-01 3.95689994e-01
2.31261045e-01 -3.58219415e-01 4.57752913e-01 -5.86656809e-01
7.37802327e-01 3.48876476e-01 -1.44498482e-01 -6.45610154e-01
6.82674423e-02 2.83348411e-01 -1.59191981e-01 -1.76634919e-02
1.43631160e+00 -4.97768939e-01 -1.29019683e-02 2.99516432e-02
1.69157708e+00 -6.18084729e-01 -1.67617977e+00 -1.58833802e-01
-3.37370187e-01 -7.54233599e-01 3.29106003e-01 -4.61609125e-01
-1.40448916e+00 1.03520012e+00 8.88353169e-01 7.15565532e-02
1.72161496e+00 -3.53139043e-01 1.30043304e+00 4.66731846e-01
2.10863546e-01 -1.06040192e+00 -1.21863512e-02 4.85636830e-01
8.43177974e-01 -1.09820545e+00 2.45898604e-01 -1.16432354e-01
-2.37230599e-01 1.34006572e+00 1.56639427e-01 -2.40901664e-01
6.37648821e-01 6.12946153e-01 -4.31078702e-01 3.17377985e-01
-1.14917529e+00 3.56592298e-01 8.69603455e-02 6.65192246e-01
6.46581769e-01 -2.16172114e-01 -1.35091603e-01 -3.21575344e-01
-3.42888445e-01 3.56995553e-01 4.22530353e-01 8.99964690e-01
-6.25440001e-01 -9.47677672e-01 -1.36975184e-01 7.65695691e-01
-5.44175804e-01 -6.41410112e-01 3.94680083e-01 3.39862972e-01
9.78727937e-02 7.50878096e-01 3.57893705e-01 -6.49995983e-01
-7.92979673e-02 -1.09770671e-01 3.43440622e-01 1.35407886e-02
-3.03058863e-01 -2.66097318e-02 -1.54992923e-01 -9.05732691e-01
-2.33787462e-01 -2.81079978e-01 -1.09269464e+00 -7.91042149e-01
1.91123009e-01 -5.65628529e-01 3.80916655e-01 6.56331897e-01
4.79016483e-01 6.41545594e-01 5.97387314e-01 -1.04082906e+00
-6.70586109e-01 -6.47383809e-01 -4.21721309e-01 4.92907345e-01
8.43990147e-01 -1.54875726e-01 -2.94681519e-01 7.43745506e-01]
|
[11.37735652923584, -1.5717798471450806]
|
eacce34e-2c51-48f4-8f50-53ee8e5d5c57
|
a-deep-learning-based-native-language
| null | null |
https://aclanthology.org/W17-5047
|
https://aclanthology.org/W17-5047.pdf
|
A deep-learning based native-language classification by using a latent semantic analysis for the NLI Shared Task 2017
|
This paper proposes a deep-learning based native-language identification (NLI) using a latent semantic analysis (LSA) as a participant (ETRI-SLP) of the NLI Shared Task 2017 where the NLI Shared Task 2017 aims to detect the native language of an essay or speech response of a standardized assessment of English proficiency for academic purposes. To this end, we use the six unit forms of a text data such as character 4/5/6-grams and word 1/2/3-grams. For each unit form of text data, we convert it into a count-based vector, extract a 2000-rank LSA feature, and perform a linear discriminant analysis (LDA) based dimension reduction. From the count-based vector or the LSA-LDA feature, we also obtain the output prediction values of a support vector machine (SVM) based classifier, the output prediction values of a deep neural network (DNN) based classifier, and the bottleneck values of a DNN based classifier. In order to incorporate the various kinds of text-based features and a speech-based i-vector feature, we design two DNN based ensemble classifiers for late fusion and early fusion, respectively. From the NLI experiments, the F1 (macro) scores are obtained as 0.8601, 0.8664, and 0.9220 for the essay track, the speech track, and the fusion track, respectively. The proposed method has comparable performance to the top-ranked teams for the speech and fusion tracks, although it has slightly lower performance for the essay track.
|
['Yun-Keun Lee', 'Jeon-Gue Park', 'Yun-Kyung Lee', 'Hyung-Bae Jeon', 'Yoo Rhee Oh', 'Hwa Jeon Song']
|
2017-09-01
| null | null | null |
ws-2017-9
|
['native-language-identification']
|
['natural-language-processing']
|
[-5.01560457e-02 -4.16869223e-01 -4.35013860e-01 -3.94959748e-01
-1.04654038e+00 -5.90983450e-01 6.97673857e-01 2.57194698e-01
-5.67443967e-01 4.06542659e-01 5.46241641e-01 -6.25450611e-01
-2.94584155e-01 -7.28491366e-01 -1.02721080e-01 -6.33400440e-01
5.43781281e-01 3.13715607e-01 -2.70057857e-01 1.98034253e-02
2.77388752e-01 3.86919022e-01 -1.56418371e+00 4.98108208e-01
1.04702222e+00 1.26108646e+00 -1.02391969e-02 6.67544782e-01
-5.43933809e-01 7.27263808e-01 -6.48568273e-01 -2.85046160e-01
-3.72749828e-02 -4.17079777e-01 -7.21749425e-01 -2.92298585e-01
5.26293039e-01 -1.93654552e-01 -5.58948517e-01 1.26016581e+00
5.24500430e-01 3.06450069e-01 8.80766392e-01 -1.03852725e+00
-5.10515451e-01 7.45625019e-01 -2.93761402e-01 5.80942258e-02
6.27776504e-01 -1.69799924e-01 1.03108525e+00 -1.30520582e+00
3.16923350e-01 1.49426532e+00 5.24500012e-01 3.59995604e-01
-7.73210883e-01 -9.99910831e-01 -1.01642951e-01 3.13833833e-01
-1.13900125e+00 -6.17805421e-01 6.80764377e-01 -6.03564739e-01
8.13325405e-01 1.99303836e-01 3.44686776e-01 1.48842430e+00
2.50532418e-01 1.25220335e+00 1.34683836e+00 -6.00263298e-01
1.63061425e-01 2.23141134e-01 7.51094699e-01 6.42242372e-01
-1.01115085e-01 -8.74152035e-02 -8.51372480e-01 -8.48697722e-02
5.26212342e-02 1.13763601e-01 7.82147199e-02 3.67552310e-01
-1.28053916e+00 9.99569535e-01 -1.89003512e-01 4.93823320e-01
-1.51800618e-01 -4.92395371e-01 5.19112885e-01 4.75402266e-01
5.86997688e-01 9.63588208e-02 -5.03334999e-01 -4.02607650e-01
-1.08472335e+00 2.56603379e-02 9.06457126e-01 5.58002174e-01
5.51162601e-01 2.07292438e-01 -6.28132701e-01 1.02758694e+00
3.47194791e-01 3.88256222e-01 1.24753118e+00 -6.02508783e-01
6.33791327e-01 9.40478265e-01 -4.55804825e-01 -8.92573774e-01
-3.84496272e-01 -4.56635267e-01 -1.01011336e+00 -7.77414665e-02
3.59862089e-01 -1.03608996e-01 -7.35555708e-01 1.50915909e+00
-1.15065098e-01 -5.29622957e-02 3.93195689e-01 6.32687926e-01
1.24278200e+00 9.98784006e-01 7.11034536e-02 -4.19318616e-01
1.36454153e+00 -1.19853473e+00 -8.65375459e-01 -9.43624154e-02
1.08104134e+00 -9.87262368e-01 1.05952704e+00 5.46556115e-01
-7.84054697e-01 -9.84785736e-01 -1.05338955e+00 -8.91391262e-02
-5.40713608e-01 8.16928446e-01 1.69321299e-01 6.83024228e-01
-7.89191306e-01 3.48840117e-01 -3.13175559e-01 -3.08021694e-01
-1.01807658e-02 2.55691439e-01 -4.87271249e-01 -9.31777805e-02
-1.40781343e+00 5.43161690e-01 2.93540567e-01 -3.60907078e-01
-5.85761011e-01 -5.87184250e-01 -8.44403446e-01 2.28827998e-01
-5.59361093e-02 -9.98241529e-02 8.47141981e-01 -8.69524002e-01
-1.85498166e+00 1.04176104e+00 -1.86069131e-01 -2.69359261e-01
1.63771659e-01 4.37793434e-02 -5.58852732e-01 -1.86667755e-01
2.23624364e-01 3.20506126e-01 6.58634245e-01 -5.39476693e-01
-8.88403356e-01 -6.70610905e-01 -3.04623812e-01 3.19523364e-01
-1.11022139e+00 3.04247707e-01 5.45499939e-03 -8.13355803e-01
1.91072091e-01 -8.21517348e-01 4.12184119e-01 -5.94189346e-01
-4.97001708e-01 -1.11475742e+00 8.43032241e-01 -1.16844404e+00
1.51121402e+00 -2.54567456e+00 6.18469529e-02 6.14885911e-02
2.08138525e-01 4.92153287e-01 -2.43832693e-01 1.64601162e-01
-2.62854218e-01 1.07637331e-01 2.27980927e-01 -5.27584851e-01
1.17658265e-01 -1.95614815e-01 -3.04072797e-01 2.66583472e-01
-3.89806271e-01 6.47011280e-01 -7.42552280e-01 -2.49591604e-01
-2.93788854e-02 2.66421624e-02 -1.98662966e-01 2.89766431e-01
3.20319831e-01 1.15376927e-01 -1.65260985e-01 4.73596543e-01
4.38678563e-01 1.35372624e-01 3.05668972e-02 -1.36631235e-01
-4.51904118e-01 8.29367638e-01 -1.04629004e+00 1.57790089e+00
-5.24494290e-01 8.47736955e-01 -1.23186270e-02 -1.29202199e+00
1.42125273e+00 4.54911560e-01 2.40334049e-01 -7.13030815e-01
8.92539099e-02 4.21574086e-01 1.98918343e-01 -2.55340487e-01
3.17668200e-01 2.04783246e-01 -4.35849637e-01 6.00121140e-01
4.55556571e-01 4.77790207e-01 1.15770027e-02 2.34047756e-01
1.10034692e+00 -2.41682306e-01 3.22404236e-01 -4.91233945e-01
1.13696170e+00 -2.12719738e-01 4.84881520e-01 6.81924343e-01
-3.19395065e-01 2.09727645e-01 4.52096224e-01 -4.50202674e-01
-8.54668558e-01 -8.06341887e-01 -1.79406226e-01 1.52076328e+00
-4.44219530e-01 -5.24209738e-01 -6.32426918e-01 -7.88086712e-01
6.25100359e-02 8.13313127e-01 -1.19141214e-01 -5.21076441e-01
-2.82563120e-01 -3.64546359e-01 5.70464492e-01 4.79452163e-01
6.08790040e-01 -9.39201057e-01 3.66636664e-01 1.24202885e-01
-3.12378347e-01 -9.06075954e-01 -6.12270415e-01 2.12780550e-01
-3.53388578e-01 -4.70211595e-01 -7.80123472e-01 -1.18293881e+00
2.02147439e-01 9.75362435e-02 4.16725457e-01 -3.21879894e-01
6.37134612e-02 2.28386864e-01 -3.64238679e-01 -1.34271622e-01
-6.13254488e-01 4.42275494e-01 7.14620650e-01 2.50873268e-01
7.93040097e-01 -1.89995691e-01 -1.11894459e-01 2.33817771e-02
-3.39436382e-01 -1.48239045e-03 3.42399508e-01 9.07958627e-01
1.99983552e-01 1.30142406e-01 7.57237494e-01 -1.89513505e-01
1.10116196e+00 -2.84655035e-01 -4.26663339e-01 3.61225843e-01
-6.63431525e-01 -1.87516093e-01 7.68765569e-01 -7.21296549e-01
-9.68295515e-01 1.03308521e-01 -2.15204462e-01 -3.23292404e-01
-2.64716715e-01 8.80837619e-01 -3.00577760e-01 1.49131402e-01
5.88616431e-01 5.08140564e-01 -5.54937720e-02 -6.48162246e-01
-3.82584222e-02 1.65364158e+00 4.68930036e-01 -4.57417876e-01
5.13315797e-01 -2.91848987e-01 -4.77225959e-01 -8.63924265e-01
-1.27632165e+00 -7.31226683e-01 -6.71176136e-01 -1.62026018e-01
8.78269136e-01 -1.11998165e+00 -9.16698933e-01 7.78537929e-01
-1.32079232e+00 1.17205761e-01 -1.34828761e-01 8.66011381e-01
-1.60559267e-01 3.71107250e-01 -6.82613194e-01 -9.11237478e-01
-5.25916874e-01 -1.44954026e+00 8.54669511e-01 3.50852013e-01
-5.25102258e-01 -8.53480279e-01 -1.27916083e-01 8.17459881e-01
1.53105974e-01 -4.52702254e-01 1.23107719e+00 -1.56448746e+00
1.86408609e-01 -3.53126556e-01 -2.36971125e-01 8.68347764e-01
-4.71126847e-02 -2.56504506e-01 -1.05661261e+00 -3.69736254e-01
1.93814427e-01 -4.06436890e-01 1.01676333e+00 2.17778251e-01
1.26736271e+00 -4.81794715e-01 -4.92298603e-02 5.22267222e-01
8.90891790e-01 3.03582668e-01 1.67162970e-01 1.94759339e-01
7.79948115e-01 6.54401362e-01 2.76647151e-01 3.36905092e-01
4.36871976e-01 5.95068276e-01 -4.26447541e-01 3.87888759e-01
-1.59825429e-01 -2.17258200e-01 1.13133645e+00 1.70030797e+00
4.33593512e-01 -2.78302729e-01 -1.31465971e+00 5.47023118e-01
-1.59362507e+00 -7.46601939e-01 -2.89792180e-01 2.24069667e+00
8.66608679e-01 7.05944896e-02 2.50176527e-02 6.10457480e-01
7.81875253e-01 5.68701737e-02 -2.85547584e-01 -6.61567926e-01
-8.84319767e-02 3.08795810e-01 3.40199083e-01 5.68130195e-01
-1.22893548e+00 9.70802546e-01 5.18885994e+00 1.52491760e+00
-1.19696248e+00 1.76767096e-01 5.40421426e-01 1.34681925e-01
-1.42671570e-01 -1.62279278e-01 -1.42460215e+00 7.67103612e-01
1.38247061e+00 -3.99765164e-01 3.37697178e-01 7.93427944e-01
-7.38849863e-02 -7.71739855e-02 -1.07006383e+00 1.22415233e+00
3.08805108e-01 -1.18639982e+00 2.30692625e-01 2.45888472e-01
6.47268176e-01 -1.58721969e-01 3.59932423e-01 8.35252225e-01
9.93591547e-02 -1.04442000e+00 5.18682361e-01 5.57094395e-01
9.27926540e-01 -8.59267473e-01 7.86595464e-01 8.60485077e-01
-1.00979471e+00 -4.28285152e-01 -2.06391558e-01 -2.20495433e-01
-5.67441344e-01 6.74696684e-01 -6.61496997e-01 4.93148148e-01
4.46620405e-01 6.59582734e-01 -3.34086686e-01 4.77775455e-01
-1.36834592e-01 1.08835554e+00 -9.49274078e-02 -2.99115658e-01
2.78158844e-01 -1.95674613e-01 5.93992174e-01 1.24732614e+00
5.64465046e-01 -8.56931359e-02 3.59130621e-01 6.03250980e-01
-3.14437926e-01 4.96496767e-01 -5.08040607e-01 -3.73218209e-01
5.93060851e-01 1.36451542e+00 -3.09138924e-01 -5.34727156e-01
-4.18761700e-01 9.81012106e-01 1.85698017e-01 2.19139010e-01
-1.94283441e-01 -8.37855101e-01 5.24770379e-01 -1.76189497e-01
-2.46198416e-01 -2.76362985e-01 -5.24883449e-01 -1.28826654e+00
-2.06967965e-01 -1.07995236e+00 4.30322617e-01 -4.75325167e-01
-1.52521372e+00 4.34861839e-01 -4.11356747e-01 -1.13052642e+00
-3.13087106e-01 -9.31114852e-01 -7.33392775e-01 1.32544959e+00
-1.03145933e+00 -8.49727094e-01 -1.10838592e-01 5.33111453e-01
7.00691164e-01 -1.23009157e+00 9.96726692e-01 3.55988353e-01
-9.03980553e-01 1.05048537e+00 4.94455338e-01 6.34641290e-01
8.37032735e-01 -9.85389650e-01 9.35837179e-02 6.63252532e-01
1.29472792e-01 5.49348950e-01 3.68441306e-02 -5.21898627e-01
-1.24206960e+00 -1.00074732e+00 1.60849380e+00 -3.10516953e-01
8.07112277e-01 -4.02507007e-01 -7.35845983e-01 4.46082056e-01
-3.03293876e-02 -2.59317726e-01 8.95742893e-01 2.64919370e-01
-3.15484218e-02 -2.74513751e-01 -8.43311608e-01 3.27194691e-01
6.49981976e-01 -9.67439711e-01 -7.03202784e-01 4.45392370e-01
8.89558673e-01 -1.54683307e-01 -9.89543319e-01 3.47924381e-01
5.51830053e-01 -5.76387227e-01 8.39385867e-01 -5.74937880e-01
5.19527733e-01 3.25126588e-01 -2.76595175e-01 -1.29366946e+00
-4.86081183e-01 -1.78352848e-01 -2.50705574e-02 1.47380495e+00
2.36617401e-01 -5.59156001e-01 5.58359444e-01 2.18144894e-01
-1.49597853e-01 -7.18529165e-01 -1.16384315e+00 -8.94335747e-01
2.23117933e-01 -4.91886675e-01 2.64493048e-01 1.29929662e+00
1.87407777e-01 6.56736553e-01 -1.19745851e-01 -2.63741285e-01
4.75355178e-01 -1.10420346e-01 4.36976582e-01 -1.64981329e+00
2.84818292e-01 -7.95992553e-01 -3.59111726e-01 -8.63931298e-01
8.46262753e-01 -1.55697656e+00 -3.92599732e-01 -1.23869944e+00
1.93327636e-01 -3.06228809e-02 -5.10904670e-01 5.89481950e-01
5.90008944e-02 -2.14230940e-01 1.16930135e-01 3.02102715e-01
-4.14263368e-01 6.08377635e-01 7.82186329e-01 -3.73972595e-01
-2.67578185e-01 1.34297401e-01 -5.17003477e-01 8.00691664e-01
8.07841957e-01 -4.53295261e-01 -3.41903418e-02 -2.61391122e-02
8.39930326e-02 1.51554778e-01 7.52627924e-02 -1.12273777e+00
5.90847075e-01 2.07110699e-02 5.77222884e-01 -9.28858995e-01
1.55873910e-01 -2.46126637e-01 -6.66195035e-01 4.44096416e-01
-6.05874240e-01 -2.56234139e-01 1.50781767e-02 1.74344215e-03
-5.24862349e-01 -4.56964225e-01 6.00144207e-01 1.82506084e-01
-4.62218285e-01 2.02662140e-01 -8.09151947e-01 -1.35214597e-01
6.27520323e-01 -1.85180575e-01 -1.47075877e-01 -3.60536665e-01
-6.69796407e-01 8.33661947e-03 -1.86321631e-01 7.83918500e-01
6.81438029e-01 -1.73328245e+00 -8.94534469e-01 6.47876740e-01
1.01549931e-01 -4.02320117e-01 1.81586355e-01 7.69978583e-01
-1.25493124e-01 7.61497498e-01 -9.14127454e-02 -3.93093020e-01
-1.22640073e+00 6.84504956e-02 1.75069291e-02 -5.28293312e-01
-1.30199686e-01 8.69672000e-01 1.89226314e-01 -9.30136859e-01
5.65609157e-01 -2.43204720e-02 -5.63381672e-01 4.88059938e-01
6.17150605e-01 6.52927876e-01 2.04535723e-01 -7.62756526e-01
-4.26943004e-01 3.07476312e-01 -9.97009128e-02 -2.62736768e-01
1.09257519e+00 1.16584398e-01 -2.79402196e-01 7.63707936e-01
1.52059937e+00 6.38464615e-02 -3.24156821e-01 -5.45889676e-01
1.05022185e-01 7.88662061e-02 4.85641986e-01 -7.94983506e-01
-6.50124311e-01 1.14216435e+00 6.23741627e-01 9.58614275e-02
6.83251441e-01 -2.55403996e-01 9.80953813e-01 3.89493138e-01
-1.16013609e-01 -1.49881411e+00 5.66468127e-02 1.17587459e+00
8.45785618e-01 -1.11907864e+00 -2.41900325e-01 1.63170740e-01
-4.11778301e-01 1.48108566e+00 5.88703752e-01 3.21230739e-01
6.38331175e-01 1.89154282e-01 -1.86299458e-02 2.39326075e-01
-6.75714672e-01 5.04524298e-02 6.98755682e-01 1.41361207e-01
6.18808270e-01 2.77178288e-01 -4.66027975e-01 1.29645836e+00
-4.78841782e-01 -2.71385670e-01 2.26134494e-01 2.35607490e-01
-5.91766119e-01 -1.04403639e+00 -4.26575273e-01 6.98165715e-01
-3.78764182e-01 -1.65094838e-01 -3.17885488e-01 1.88779727e-01
1.01426750e-01 1.01608968e+00 1.82765424e-01 -8.58338535e-01
1.00461475e-03 7.93022215e-01 -1.93616711e-02 -6.44125044e-01
-6.57968640e-01 -1.55185685e-01 9.47574899e-03 -2.74355382e-01
-2.85280105e-02 -7.58729041e-01 -1.09111154e+00 -4.00994927e-01
-1.70128316e-01 6.12358227e-02 8.97461891e-01 1.26839185e+00
1.36820838e-01 3.25383186e-01 8.32351327e-01 -4.18743581e-01
-8.68119478e-01 -1.29420245e+00 -5.86240768e-01 1.90080196e-01
2.32559398e-01 -3.60745132e-01 -7.07608700e-01 -1.86139315e-01]
|
[10.349732398986816, 10.521793365478516]
|
0cadca5c-3980-42c3-b221-12eba82932ac
|
confidence-ranking-for-ctr-prediction
|
2307.01206
| null |
https://arxiv.org/abs/2307.01206v1
|
https://arxiv.org/pdf/2307.01206v1.pdf
|
Confidence Ranking for CTR Prediction
|
Model evolution and constant availability of data are two common phenomena in large-scale real-world machine learning applications, e.g. ads and recommendation systems. To adapt, the real-world system typically retrain with all available data and online learn with recently available data to update the models periodically with the goal of better serving performance. In this paper, we propose a novel framework, named Confidence Ranking, which designs the optimization objective as a ranking function with two different models. Our confidence ranking loss allows direct optimization of the logits output for different convex surrogate functions of metrics, e.g. AUC and Accuracy depending on the target task and dataset. Armed with our proposed methods, our experiments show that the introduction of confidence ranking loss can outperform all baselines on the CTR prediction tasks of public and industrial datasets. This framework has been deployed in the advertisement system of JD.com to serve the main traffic in the fine-rank stage.
|
['Jingping Shao', 'Zhangang Lin', 'Xiwei Zhao', 'Pei Wang', 'Congcong Liu', 'Jian Zhu']
|
2023-06-28
| null | null | null | null |
['click-through-rate-prediction']
|
['miscellaneous']
|
[-1.69999406e-01 -1.90908358e-01 -4.58488882e-01 -7.69037783e-01
-9.77568746e-01 -6.78247392e-01 4.31686759e-01 3.38030308e-01
-4.41527188e-01 5.93750894e-01 -2.74202049e-01 -3.56266022e-01
-3.28327000e-01 -6.02478266e-01 -8.65059376e-01 -3.59712809e-01
-3.35863084e-01 1.00146842e+00 4.36183363e-01 -3.12260032e-01
2.78765887e-01 3.35290253e-01 -1.44726884e+00 3.92117321e-01
7.73804188e-01 1.45560050e+00 1.28541052e-01 5.13111949e-01
3.20659488e-01 4.47908521e-01 -4.27789152e-01 -8.79678130e-01
5.97578704e-01 3.87696892e-01 -3.48263681e-01 -1.47744104e-01
2.13803023e-01 -2.04773799e-01 -3.62813354e-01 7.32695639e-01
4.49025571e-01 9.73626375e-02 4.19587970e-01 -1.48917520e+00
-3.14194471e-01 9.02608693e-01 -5.96662462e-01 1.65417492e-01
-1.69805512e-01 -1.28171459e-01 1.46038592e+00 -6.79074466e-01
2.18340248e-01 1.32026708e+00 3.15753132e-01 1.85959637e-01
-1.47441041e+00 -5.46011448e-01 5.57448924e-01 3.20686877e-01
-1.08441198e+00 -2.52694398e-01 5.67150533e-01 -3.79896700e-01
2.37355813e-01 3.87052506e-01 1.63044482e-01 9.80072200e-01
3.26602757e-01 7.92006850e-01 9.57680821e-01 -1.06917910e-01
2.21524686e-01 6.89644039e-01 4.36600834e-01 3.36228520e-01
1.55297235e-01 2.30234101e-01 -4.22957242e-01 -4.95306373e-01
2.68166214e-01 1.38448821e-02 2.28583172e-01 -5.97451746e-01
-8.48852456e-01 1.05216622e+00 3.38045001e-01 -1.26805022e-01
-3.46618354e-01 2.26252660e-01 4.55419987e-01 8.15550148e-01
6.43210530e-01 3.79925102e-01 -9.85534012e-01 -4.41586450e-02
-6.46184027e-01 2.44493559e-01 8.63500237e-01 8.42002690e-01
2.83759177e-01 -3.31432879e-01 -5.12055933e-01 1.14998460e+00
3.75472784e-01 3.07011068e-01 4.31351751e-01 -8.13502073e-01
5.65306127e-01 2.73764759e-01 3.33428651e-01 -7.07293153e-01
-3.17413539e-01 -9.42656398e-01 -3.85544240e-01 -3.40349972e-02
3.86857569e-01 -7.70936860e-03 -5.60736299e-01 1.81546783e+00
2.70905554e-01 2.51040488e-01 -3.45256239e-01 6.86233938e-01
6.80656210e-02 4.61135417e-01 2.36527137e-02 -2.56072253e-01
1.00204289e+00 -1.01222813e+00 -3.20968777e-01 -1.33076489e-01
5.04687548e-01 -6.34073853e-01 1.25971282e+00 8.24554026e-01
-9.80110645e-01 -7.11430013e-01 -1.05650246e+00 4.37895030e-01
-3.07763666e-01 1.47492737e-01 5.55949450e-01 5.72366357e-01
-8.09091389e-01 8.58368993e-01 -5.35723031e-01 -1.27460808e-01
1.26594156e-01 6.15250826e-01 1.44252226e-01 -6.52540028e-02
-1.16165042e+00 6.64606571e-01 9.01050568e-02 1.73688710e-01
-8.44487786e-01 -8.51031840e-01 -4.65131216e-02 1.47746980e-01
6.09825373e-01 -5.69069445e-01 1.53514183e+00 -8.10769677e-01
-1.46936488e+00 6.27866626e-01 5.03430247e-01 -7.40142345e-01
8.40000868e-01 -5.20201683e-01 -7.30260730e-01 -6.74702287e-01
-6.74073622e-02 1.88510567e-01 8.62860799e-01 -1.09397578e+00
-1.02860880e+00 -4.46307093e-01 1.45616814e-01 -1.82410300e-01
-2.81651467e-01 -1.89338163e-01 -5.08648634e-01 -4.25229996e-01
-3.25393975e-01 -1.17394388e+00 -3.25989693e-01 -3.18429083e-01
-3.20799232e-01 -1.25241518e-01 6.48837686e-01 -4.41281855e-01
1.62036550e+00 -2.18495512e+00 2.23723575e-02 6.07175469e-01
1.60792042e-02 -1.32391341e-02 -3.53736401e-01 2.71921158e-01
1.01345509e-01 3.32263438e-03 2.84390867e-01 -6.97096214e-02
2.67355889e-01 2.59262398e-02 -3.38874072e-01 1.98973089e-01
-2.68125609e-02 6.62202477e-01 -6.72239900e-01 -6.11947179e-02
-1.85587823e-01 6.26637489e-02 -6.40340447e-01 1.61975339e-01
-5.46431184e-01 3.43721718e-01 -7.96115637e-01 3.87465775e-01
4.77731407e-01 -6.05329931e-01 3.36977810e-01 -2.72974908e-01
-6.54815435e-02 1.72542736e-01 -1.06391454e+00 1.55830646e+00
-7.57949829e-01 1.26498893e-01 9.79985818e-02 -1.07020032e+00
8.26424837e-01 -2.01185673e-01 7.58603096e-01 -8.06279302e-01
6.93089068e-02 4.01227266e-01 1.99486688e-01 -8.26411787e-03
3.44257444e-01 3.88206154e-01 -2.97445387e-01 1.96072966e-01
-3.06428879e-01 4.10650402e-01 2.62815326e-01 2.95067251e-01
1.22287023e+00 -3.05298388e-01 -7.29241222e-02 -2.98398107e-01
5.49182296e-01 -1.46886021e-01 2.62254030e-01 8.95668209e-01
5.72883412e-02 1.76894173e-01 4.87914383e-01 -3.98216575e-01
-9.40377653e-01 -1.07937407e+00 -2.94834852e-01 1.62364686e+00
-1.98306218e-02 -1.93516269e-01 -9.99444276e-02 -9.83860433e-01
5.91193914e-01 7.85226107e-01 -3.87067348e-01 -2.90265828e-01
-5.37506998e-01 -7.53988564e-01 -1.31878570e-01 1.52492717e-01
6.63920939e-02 -6.84759974e-01 -1.00022100e-01 5.91590703e-01
1.12568595e-01 -1.07347405e+00 -5.62738359e-01 2.31133461e-01
-7.92484045e-01 -8.41136754e-01 -2.97032773e-01 -2.41493151e-01
1.47455260e-01 9.28503275e-02 1.40971959e+00 -1.41550288e-01
-1.06924988e-01 4.37873960e-01 -3.74359488e-01 -3.17385197e-01
-2.87633568e-01 4.07095313e-01 1.21015953e-02 3.55982602e-01
-3.44753042e-02 -5.79049647e-01 -7.03730345e-01 7.39304483e-01
-7.54609048e-01 -4.17719096e-01 8.68344665e-01 8.89068186e-01
7.59484410e-01 6.09572083e-02 8.68910134e-01 -1.18685067e+00
8.10116887e-01 -7.77607799e-01 -9.35178339e-01 4.60548103e-01
-1.24291003e+00 1.85347453e-01 5.66393733e-01 -6.51327729e-01
-7.11958528e-01 -1.90666690e-01 1.08671308e-01 -2.23918125e-01
4.44727868e-01 6.55810475e-01 -4.22980972e-02 2.97510445e-01
4.33973283e-01 -2.94280976e-01 -2.07486257e-01 -7.35606909e-01
4.03104484e-01 6.33381188e-01 3.51903051e-01 -7.36922562e-01
8.35697234e-01 1.53234348e-01 2.73956638e-02 -1.32551923e-01
-1.19759727e+00 -5.05753398e-01 -3.33630264e-01 -6.43503144e-02
2.56995350e-01 -8.69634569e-01 -7.29148626e-01 8.33278745e-02
-7.51324654e-01 -2.61573642e-01 -3.97229344e-01 3.55723202e-01
-4.99450535e-01 -7.01607615e-02 -4.49518651e-01 -6.58085585e-01
-4.00913596e-01 -9.41615641e-01 1.02756727e+00 8.33411291e-02
3.04906636e-01 -6.83074236e-01 1.31806433e-01 2.92941004e-01
6.33651435e-01 7.99264088e-02 1.03693402e+00 -1.06753302e+00
-5.57516038e-01 -5.73807240e-01 -8.55983198e-02 6.11624122e-01
-1.32344067e-01 -6.94936514e-02 -8.25766206e-01 -6.05064392e-01
-4.24525350e-01 -2.66417891e-01 7.16650605e-01 3.85091662e-01
1.66932130e+00 -4.22721058e-01 -3.52137715e-01 1.77082464e-01
1.08633196e+00 2.43113697e-01 2.44900718e-01 3.23652804e-01
1.65572956e-01 5.63239455e-01 1.20291281e+00 7.70144999e-01
2.09346399e-01 1.15259159e+00 5.64576209e-01 2.17242446e-02
3.61348540e-01 -1.78750545e-01 5.10551929e-01 6.91906214e-01
3.04898471e-01 -3.85129571e-01 -4.02773023e-01 -2.24107485e-02
-2.16265512e+00 -7.59803295e-01 6.80928826e-02 2.73320460e+00
7.47769117e-01 7.01847911e-01 3.95726800e-01 -1.92963719e-01
7.27450252e-01 -1.41168267e-01 -9.81359422e-01 -4.40384060e-01
1.63395673e-01 2.83220038e-02 7.29048729e-01 2.31751278e-01
-9.64171231e-01 6.61904871e-01 5.54023647e+00 8.70698333e-01
-1.11128795e+00 2.71154195e-01 9.61392343e-01 -4.10283923e-01
-6.15068190e-02 -3.57774459e-02 -9.34473455e-01 7.74019301e-01
1.33055294e+00 -4.62834626e-01 5.52364826e-01 9.49193597e-01
2.78388053e-01 3.65269303e-01 -1.41625977e+00 8.59435737e-01
-3.85558754e-01 -9.26272452e-01 -2.60784388e-01 3.63028258e-01
6.18549764e-01 4.44394529e-01 5.59161425e-01 8.06266963e-01
5.40580690e-01 -6.52125597e-01 7.66238868e-01 5.76793849e-01
6.57405674e-01 -6.33523643e-01 6.68254256e-01 2.09884927e-01
-8.39508712e-01 -6.24973834e-01 -1.97871059e-01 4.44095820e-01
2.98346758e-01 9.52531636e-01 -6.49296641e-01 3.62857372e-01
5.88211775e-01 4.69076067e-01 -6.31341338e-01 1.23193133e+00
4.26852524e-01 6.57863855e-01 -3.73640031e-01 1.09452508e-01
4.44891304e-02 -2.78829783e-01 5.19028068e-01 8.43971908e-01
3.22702736e-01 -3.58233482e-01 4.66609478e-01 4.37654912e-01
-3.25892091e-01 3.41292441e-01 -1.04539864e-01 -2.48805825e-02
4.81724918e-01 1.42102540e+00 -1.86095402e-01 -1.17468327e-01
-4.68533009e-01 5.88873088e-01 2.03761160e-01 2.38414392e-01
-1.21160257e+00 1.09959483e-01 6.31550252e-01 5.93779922e-01
3.88052523e-01 -1.36573896e-01 1.56372756e-01 -9.06354487e-01
2.37124339e-01 -7.91606843e-01 6.85214937e-01 -3.62663686e-01
-1.65326524e+00 6.47119343e-01 7.14007718e-03 -1.25665641e+00
-2.02737242e-01 -6.48180187e-01 -4.20175105e-01 4.42735523e-01
-1.63448071e+00 -9.33578968e-01 -3.69789847e-03 4.60789949e-01
6.86940312e-01 -4.79170859e-01 2.43619129e-01 7.20361888e-01
-6.54503405e-01 9.52854216e-01 6.89628661e-01 -4.69403774e-01
1.01806784e+00 -1.07111371e+00 2.43487537e-01 1.10702299e-01
1.95845053e-01 3.85664523e-01 8.77764285e-01 -3.89670670e-01
-1.21660793e+00 -1.19114900e+00 4.65046972e-01 -2.64029771e-01
1.13560843e+00 -5.12615144e-01 -4.80397135e-01 4.63912219e-01
-3.91111732e-01 9.14178118e-02 5.30582726e-01 4.55994904e-01
-4.50816453e-01 -8.57677877e-01 -1.04329693e+00 2.98869878e-01
8.65755141e-01 -4.94691953e-02 3.35741818e-01 7.12405920e-01
9.41184342e-01 -1.05061427e-01 -9.84782219e-01 5.03728569e-01
9.26747680e-01 -5.31149387e-01 9.43758547e-01 -1.14359009e+00
1.01585731e-01 -4.83831987e-02 -5.48113704e-01 -1.27740455e+00
-4.86443043e-01 -6.31756485e-01 -4.46359336e-01 1.22457242e+00
1.00247025e+00 -8.70649278e-01 6.09434247e-01 6.78780317e-01
5.92567958e-02 -7.56775200e-01 -9.47442412e-01 -9.93892014e-01
-4.45354581e-02 -3.26636374e-01 5.68632901e-01 3.67115498e-01
-5.56524217e-01 6.84905350e-01 -6.16115749e-01 6.99464604e-02
6.25315309e-01 2.57746279e-01 8.18079054e-01 -1.75377166e+00
-8.54839563e-01 -3.42878878e-01 -2.57828563e-01 -1.09510791e+00
-5.40653337e-03 -8.78976643e-01 -3.47918242e-01 -8.60815167e-01
2.78204411e-01 -1.00455666e+00 -9.77780938e-01 2.29713157e-01
1.63600564e-01 -2.25535288e-01 1.39200866e-01 2.21272871e-01
-9.54792261e-01 5.34678578e-01 8.25198472e-01 -2.94411302e-01
-2.65126854e-01 8.81022274e-01 -8.65789056e-01 2.32800305e-01
8.70112479e-01 -6.75600648e-01 -4.86833334e-01 -2.53153414e-01
4.69317168e-01 2.63540596e-02 2.25083023e-01 -7.04766273e-01
-2.50822399e-02 -1.88186660e-01 9.57466182e-05 -6.06018066e-01
1.98033601e-01 -1.27507281e+00 2.56930262e-01 2.75048226e-01
-7.70537198e-01 4.56406891e-01 -5.99111766e-02 1.06472671e+00
-9.42491293e-02 1.50111184e-01 8.11466932e-01 3.81954342e-01
-3.34136039e-01 5.17894149e-01 3.12326789e-01 2.87704598e-02
1.00000143e+00 3.30401152e-01 -4.95717198e-01 -4.35875416e-01
-8.52679253e-01 6.36213124e-01 -6.70384765e-02 8.86657655e-01
9.45123285e-02 -1.34908199e+00 -9.67554927e-01 -1.66431516e-01
2.63789326e-01 -6.94388270e-01 7.96138868e-02 9.07819629e-01
1.71987072e-01 5.61239183e-01 1.26766160e-01 -5.77336907e-01
-1.09734631e+00 6.15578711e-01 1.59141824e-01 -9.46273804e-01
-6.23988658e-02 4.92675990e-01 1.64634719e-01 -3.28730345e-01
4.44789737e-01 -1.50989711e-01 -2.15836957e-01 -9.20097902e-02
2.26245269e-01 4.86651272e-01 5.02791524e-01 -2.13561177e-01
-1.33284494e-01 4.26215306e-02 -6.40611470e-01 -1.45706624e-01
1.38112104e+00 -3.34651470e-01 3.42110842e-01 4.87486929e-01
1.24743843e+00 -3.74590494e-02 -1.03366005e+00 -5.51774025e-01
2.98445225e-01 -4.37259108e-01 7.57043809e-02 -1.36012232e+00
-1.32954013e+00 3.66160989e-01 1.08888257e+00 3.35736066e-01
1.02854025e+00 1.98022395e-01 6.97397649e-01 3.94286454e-01
5.02654135e-01 -1.32254422e+00 1.74966156e-01 2.86326945e-01
8.74356389e-01 -1.43665755e+00 -7.71444887e-02 -1.55629456e-01
-7.45394945e-01 6.72283471e-01 5.81858158e-01 2.07324177e-02
1.09383452e+00 -1.88849904e-02 -2.57684946e-01 -1.04857929e-01
-1.46136236e+00 1.47038726e-02 4.78941053e-01 1.28689140e-01
3.03504974e-01 2.76246488e-01 -6.79271400e-01 9.08257902e-01
-4.45173904e-02 -2.75220364e-01 1.59461081e-01 6.35118902e-01
-4.03253853e-01 -1.59176338e+00 6.69531450e-02 1.05098307e+00
-5.33226609e-01 2.67737042e-02 -1.65694863e-01 6.42684639e-01
-2.75033832e-01 1.06842303e+00 -5.86243793e-02 -6.33418441e-01
8.24559391e-01 -1.35955602e-01 1.69482738e-01 -5.09574056e-01
-5.82422674e-01 1.43498495e-01 2.32064754e-01 -7.55737960e-01
1.26158148e-01 -8.49676847e-01 -6.73109472e-01 -2.31530428e-01
-6.55023992e-01 3.96737754e-01 8.72113228e-01 4.20070291e-01
4.77004766e-01 3.67621005e-01 1.38048923e+00 -2.16343805e-01
-1.49269426e+00 -8.04981709e-01 -8.46497655e-01 4.07185078e-01
7.39699900e-02 -7.44808912e-01 -2.78300077e-01 -3.04344416e-01]
|
[9.918739318847656, 5.422080993652344]
|
7f0cc19e-abb2-4740-89a2-fed1bf77c4f9
|
streaming-speaker-attributed-asr-with-token
|
2203.16685
| null |
https://arxiv.org/abs/2203.16685v2
|
https://arxiv.org/pdf/2203.16685v2.pdf
|
Streaming Speaker-Attributed ASR with Token-Level Speaker Embeddings
|
This paper presents a streaming speaker-attributed automatic speech recognition (SA-ASR) model that can recognize ``who spoke what'' with low latency even when multiple people are speaking simultaneously. Our model is based on token-level serialized output training (t-SOT) which was recently proposed to transcribe multi-talker speech in a streaming fashion. To further recognize speaker identities, we propose an encoder-decoder based speaker embedding extractor that can estimate a speaker representation for each recognized token not only from non-overlapping speech but also from overlapping speech. The proposed speaker embedding, named t-vector, is extracted synchronously with the t-SOT ASR model, enabling joint execution of speaker identification (SID) or speaker diarization (SD) with the multi-talker transcription with low latency. We evaluate the proposed model for a joint task of ASR and SID/SD by using LibriSpeechMix and LibriCSS corpora. The proposed model achieves substantially better accuracy than a prior streaming model and shows comparable or sometimes even superior results to the state-of-the-art offline SA-ASR model.
|
['Takuya Yoshioka', 'Jinyu Li', 'Zhuo Chen', 'Yashesh Gaur', 'Xiaofei Wang', 'Zhong Meng', 'Xiong Xiao', 'Yu Wu', 'Jian Wu', 'Naoyuki Kanda']
|
2022-03-30
| null | null | null | null |
['speaker-identification']
|
['speech']
|
[ 3.52115065e-01 -8.10501203e-02 2.65609384e-01 -7.16903687e-01
-1.64522862e+00 -5.54447949e-01 5.77037334e-01 1.04635186e-01
-4.19205070e-01 1.25601575e-01 3.53482008e-01 -2.79889673e-01
5.17706275e-01 1.32812440e-01 -5.69688201e-01 -7.31789231e-01
-4.87557277e-02 6.95829690e-01 4.10019793e-02 -2.70591434e-02
-1.13615327e-01 3.81139576e-01 -1.62566018e+00 6.90776587e-01
4.12984908e-01 8.98766160e-01 3.12737614e-01 1.43562675e+00
-3.32627594e-01 7.35593677e-01 -1.28142762e+00 -2.35401183e-01
-4.01755497e-02 -4.49260980e-01 -7.54789948e-01 6.82586655e-02
4.56505477e-01 -3.36963177e-01 -2.95460790e-01 7.03193486e-01
8.43999624e-01 -1.15246198e-03 4.56630051e-01 -1.18367362e+00
-1.63114667e-01 1.03595901e+00 -2.85242915e-01 4.71495628e-01
6.30176067e-01 -1.10042147e-01 8.65366995e-01 -1.17707348e+00
1.67279229e-01 1.44699383e+00 3.24787557e-01 7.10818887e-01
-1.14833808e+00 -8.41959476e-01 4.52479860e-03 2.64162809e-01
-1.97516406e+00 -1.41885054e+00 6.50820911e-01 -8.08749422e-02
1.33065295e+00 7.23446071e-01 1.91345513e-01 1.11586630e+00
-2.15223014e-01 1.07819939e+00 7.35297620e-01 -4.43629950e-01
3.06157529e-01 4.48852539e-01 1.44075274e-01 2.43668258e-01
-5.61849415e-01 -2.42529765e-01 -1.22146332e+00 -2.36741737e-01
2.95679897e-01 -2.16478348e-01 -1.73290700e-01 7.04909742e-01
-1.46310294e+00 4.69110906e-01 -3.69238377e-01 2.84021109e-01
-5.47046363e-01 8.03421810e-02 7.56510437e-01 5.45890927e-01
5.98675907e-01 -3.49233717e-01 -1.54793248e-01 -3.96541893e-01
-1.54513156e+00 -1.52543753e-01 1.00312984e+00 1.14428163e+00
4.48208213e-01 6.75385714e-01 -1.35046035e-01 1.00933790e+00
4.91232663e-01 7.40892649e-01 8.42114925e-01 -4.48653191e-01
7.71613598e-01 -1.01104446e-01 2.76334845e-02 -2.32133046e-01
1.22014441e-01 -2.82926053e-01 -7.76039720e-01 -1.21075578e-01
-9.63846296e-02 -1.01671495e-01 -7.61873305e-01 1.28062797e+00
4.13949490e-01 5.11216044e-01 5.64923525e-01 8.75713408e-01
9.68958139e-01 1.28126538e+00 -2.91025221e-01 -3.43272388e-01
1.69598305e+00 -1.05402398e+00 -1.02317107e+00 -1.92865774e-01
3.74267340e-01 -1.00270462e+00 5.42109728e-01 4.62340534e-01
-1.08812523e+00 -5.70138931e-01 -8.82770240e-01 8.64655599e-02
4.88640293e-02 3.99215013e-01 -1.80612057e-01 1.00227749e+00
-1.33627486e+00 -8.83693323e-02 -8.54122818e-01 -2.13770553e-01
-5.67840263e-02 5.46492159e-01 -3.84304047e-01 2.71456718e-01
-1.17236483e+00 5.76231718e-01 -3.69016044e-02 2.91542351e-01
-1.38964343e+00 -3.12496215e-01 -8.83861840e-01 3.18235934e-01
1.41213667e-02 -1.10412337e-01 1.61686826e+00 -1.23015678e+00
-2.15276766e+00 7.10421860e-01 -1.10558867e+00 -7.83087730e-01
1.73354119e-01 -7.89145902e-02 -9.98967707e-01 3.84102523e-01
-1.19789906e-01 2.81125218e-01 1.28995955e+00 -1.00804186e+00
-5.62530398e-01 -4.99617606e-01 -7.25714922e-01 3.52007359e-01
-4.00545686e-01 9.29779708e-01 -1.21948935e-01 -6.56215787e-01
9.85866264e-02 -6.45199060e-01 2.98512399e-01 -5.48341930e-01
-6.05117679e-01 -3.90914202e-01 9.27195191e-01 -1.10866737e+00
1.21323228e+00 -2.53820801e+00 -6.85162283e-03 8.53784382e-02
-1.56721473e-01 4.57914561e-01 -1.96360141e-01 8.97437572e-01
-1.76414132e-01 -2.61954874e-01 -6.54475838e-02 -1.13484371e+00
5.32441661e-02 -1.14118531e-02 -7.26683915e-01 5.88289976e-01
1.39483824e-01 2.61703640e-01 -5.44420183e-01 -4.55139220e-01
8.64662528e-02 7.72338331e-01 3.09872627e-02 6.53429329e-01
2.87951827e-01 2.34544188e-01 7.39078820e-02 6.57117128e-01
6.48047030e-01 2.95520425e-01 1.70826614e-01 1.33573249e-01
-2.68016279e-01 9.05417919e-01 -1.53090751e+00 1.53759766e+00
-6.59875214e-01 8.77555966e-01 7.68248737e-01 -9.06240404e-01
1.16696477e+00 1.25492644e+00 4.56995936e-03 -1.17957883e-01
4.50133458e-02 4.09049392e-01 -1.15081243e-01 -4.49469209e-01
6.34907901e-01 -1.97738767e-01 2.88287681e-02 5.43975055e-01
3.69931936e-01 2.07778901e-01 -3.62229377e-01 3.09545219e-01
1.01391685e+00 -5.18732190e-01 2.05802232e-01 -5.05570462e-03
1.02632034e+00 -5.11281848e-01 3.22229594e-01 5.58495700e-01
-2.87242591e-01 5.86713314e-01 -1.08384103e-01 1.22923478e-01
-8.18788826e-01 -1.24924409e+00 1.89422876e-01 1.31093609e+00
-2.71051437e-01 -4.49012488e-01 -8.71186912e-01 -3.71827215e-01
-2.26636469e-01 8.87023211e-01 -1.74827445e-02 2.66943216e-01
-6.87496543e-01 -1.36952803e-01 1.13171673e+00 4.17878240e-01
9.67091322e-02 -8.59925807e-01 -1.41706869e-01 4.56086457e-01
-2.82017708e-01 -1.26305127e+00 -1.17956042e+00 2.44077370e-01
-3.97467732e-01 -1.64390147e-01 -8.62613440e-01 -1.12801862e+00
3.64048809e-01 3.50685716e-01 5.39921284e-01 -5.04297256e-01
5.76352654e-03 3.42584938e-01 -3.85234356e-01 -3.77845556e-01
-1.07989430e+00 4.91813645e-02 5.91252804e-01 8.94701958e-01
4.10489321e-01 -4.60379839e-01 -2.67349184e-01 4.02139962e-01
-7.07421005e-01 -1.38275862e-01 3.10612917e-01 6.14298522e-01
8.57146308e-02 -1.49010122e-01 1.08814228e+00 -4.06482130e-01
4.69769448e-01 -3.40215564e-01 -3.03675264e-01 2.86580026e-01
-1.42825335e-01 4.28981055e-03 9.02630150e-01 -5.97167194e-01
-1.09994757e+00 1.44472882e-01 -6.10866189e-01 -6.43662453e-01
-4.75797445e-01 1.33833408e-01 -3.11503083e-01 3.56525987e-01
2.60806918e-01 1.05853498e+00 8.22350830e-02 -6.07420027e-01
2.20490679e-01 2.01115274e+00 6.29561186e-01 -9.46997181e-02
5.67432344e-01 2.38327771e-01 -9.16299284e-01 -1.44090009e+00
-1.13003008e-01 -1.16639006e+00 -3.91846418e-01 -1.56156123e-01
6.09884918e-01 -1.47019506e+00 -1.03642726e+00 6.85072303e-01
-1.66091931e+00 2.50340328e-02 -1.02485567e-01 7.06305623e-01
-3.21292073e-01 3.69239360e-01 -8.24914992e-01 -1.62798893e+00
-8.34205627e-01 -1.13752759e+00 1.46201766e+00 -1.61340699e-01
-3.68210524e-01 -5.85104585e-01 2.13704593e-02 5.34665763e-01
4.74838257e-01 -6.28469408e-01 1.23585016e-01 -1.29167163e+00
-3.52911055e-01 -3.18289280e-01 1.74248293e-01 7.22322702e-01
2.11712793e-01 -2.41801918e-01 -1.41636920e+00 -5.64741254e-01
2.72010952e-01 -4.76062559e-02 6.53912544e-01 5.03814556e-02
7.23919332e-01 -8.53008568e-01 -1.11993611e-01 4.04501230e-01
1.00510645e+00 2.36236379e-01 2.68284470e-01 -3.34844410e-01
5.52107334e-01 5.93813241e-01 1.51918963e-01 5.38686335e-01
5.14013052e-01 8.47968876e-01 -1.65947363e-01 3.33905607e-01
-1.27315551e-01 -1.47562921e-01 1.36147106e+00 1.78203273e+00
3.28603804e-01 -4.31663960e-01 -7.15081513e-01 8.29533517e-01
-1.45722067e+00 -1.12144351e+00 4.04948220e-02 2.37174869e+00
9.13066745e-01 -2.31940553e-01 4.01306868e-01 4.22389835e-01
1.11578310e+00 2.94232696e-01 -2.15062752e-01 -8.34411681e-01
-3.31490114e-02 2.41127014e-01 4.33229506e-01 8.22819650e-01
-6.90768421e-01 7.68704534e-01 5.86312437e+00 9.06743050e-01
-1.38068056e+00 4.53922123e-01 2.42944598e-01 -3.02437156e-01
-7.72621632e-02 -3.03658366e-01 -1.14099073e+00 4.36918288e-01
2.03510809e+00 -3.07747692e-01 6.62370861e-01 7.54116476e-01
3.82190555e-01 3.68763059e-01 -1.35476768e+00 1.34710062e+00
7.29530871e-01 -1.02623034e+00 3.05378227e-03 -2.64507174e-01
6.51510656e-02 6.64210618e-02 -4.52734306e-02 1.56707704e-01
6.47095144e-02 -8.04947019e-01 1.08157682e+00 -1.07627884e-01
1.07701910e+00 -7.76913106e-01 6.91234887e-01 5.95193744e-01
-1.62316895e+00 -1.23726530e-02 3.12719941e-02 2.96946079e-01
4.81949538e-01 2.96303183e-01 -1.60721397e+00 5.69800496e-01
4.27852988e-01 3.68597507e-01 -1.43602133e-01 5.77448010e-01
2.97832750e-02 1.24501252e+00 -4.54247147e-01 -7.37242848e-02
-2.70212125e-02 2.31325388e-01 8.43523443e-01 1.77383721e+00
5.55802941e-01 -7.18262047e-02 -1.14391237e-01 3.86067837e-01
-1.11483537e-01 3.36559564e-01 -2.66263694e-01 -9.39737037e-02
9.77414131e-01 1.02511585e+00 -3.97161573e-01 -7.73329318e-01
-1.52896747e-01 1.51269567e+00 -1.35103017e-02 3.90064895e-01
-6.76225841e-01 -6.23388052e-01 7.65949607e-01 -1.37889422e-02
4.76537913e-01 -2.57285774e-01 1.45161390e-01 -1.19453013e+00
4.61753532e-02 -1.07487690e+00 8.51043910e-02 -4.65055048e-01
-1.05288601e+00 1.05639184e+00 -3.45685840e-01 -1.15607476e+00
-5.30585766e-01 5.39624784e-03 -6.92643821e-01 1.18852901e+00
-1.32929027e+00 -1.07732999e+00 2.66689450e-01 7.11796463e-01
1.09154093e+00 -5.83643198e-01 1.16916704e+00 4.04798925e-01
-6.87098444e-01 9.02238607e-01 3.19630861e-01 2.00972676e-01
8.42533469e-01 -1.19935083e+00 8.03832710e-01 8.90722692e-01
3.51753503e-01 5.90322018e-01 7.38473892e-01 -2.50856280e-01
-1.63687110e+00 -1.16015947e+00 1.38876033e+00 -2.24589393e-01
2.87743658e-01 -1.07175148e+00 -9.29493964e-01 8.23773384e-01
5.40294528e-01 -2.30910048e-01 1.02630699e+00 -1.82155326e-01
-4.06265736e-01 -5.77421546e-01 -1.02396870e+00 1.78236112e-01
3.53591710e-01 -1.12474573e+00 -6.91192627e-01 1.37329757e-01
9.58753407e-01 -1.63742870e-01 -6.99152648e-01 -5.18246531e-01
4.00502175e-01 -5.79595804e-01 8.40771914e-01 -6.75301105e-02
-2.20327795e-01 -4.13144439e-01 -4.18612152e-01 -1.16414189e+00
1.76654771e-01 -1.23683441e+00 -1.66408911e-01 1.77377856e+00
3.67889285e-01 -7.75662601e-01 4.72778529e-01 1.62403822e-01
-3.42577368e-01 3.67909297e-02 -1.74829626e+00 -1.00431180e+00
-6.30483031e-01 -6.90164685e-01 7.45837808e-01 7.05579162e-01
1.94564536e-01 5.42365432e-01 -6.25265419e-01 8.64611924e-01
7.53283560e-01 -3.90801251e-01 7.26335227e-01 -6.55733585e-01
-3.48809928e-01 5.09948619e-02 -4.42857981e-01 -1.33032489e+00
3.67179990e-01 -1.11353409e+00 5.73105156e-01 -1.13905346e+00
-2.33653083e-01 8.29168335e-02 -3.15118968e-01 1.92053765e-01
6.55372664e-02 -1.46731585e-01 4.81676683e-02 1.18326463e-01
-5.39161980e-01 5.90358078e-01 2.36752987e-01 -1.80273935e-01
-2.75167674e-01 1.69685259e-01 -1.23482808e-01 1.39964998e-01
4.74872857e-01 -7.32658446e-01 -8.40093717e-02 -3.95290285e-01
-5.38005531e-01 6.97263002e-01 1.53799072e-01 -1.06264484e+00
6.97222173e-01 4.26505774e-01 -1.55937195e-01 -7.22848356e-01
7.29597747e-01 -6.26949072e-01 1.10946804e-01 3.04101348e-01
-6.41804576e-01 1.11298129e-01 -1.39883999e-03 4.51869667e-01
-4.64251339e-01 -7.37531483e-02 4.45314705e-01 3.42966735e-01
-2.03888327e-01 3.32954228e-02 -9.76222873e-01 -3.97972822e-01
6.88207388e-01 -6.28203750e-02 6.83973432e-02 -5.75111449e-01
-7.22915530e-01 -5.16264932e-03 -2.19620973e-01 4.17149723e-01
1.01241899e+00 -1.13005316e+00 -1.25248909e+00 5.88948727e-01
1.06459208e-01 -1.49532527e-01 4.37386006e-01 6.80601776e-01
-2.34551489e-01 5.61168730e-01 4.52621907e-01 -7.77267158e-01
-1.90346360e+00 3.88848841e-01 5.19190319e-02 1.36085838e-01
-4.36251909e-01 9.87811983e-01 -1.47227153e-01 -3.63740981e-01
5.10744095e-01 -1.56862289e-01 1.28884122e-01 -1.37590868e-02
1.01464164e+00 4.22598958e-01 2.26294950e-01 -1.11839998e+00
-7.52970517e-01 -5.69295101e-02 -2.04412147e-01 -7.85964847e-01
1.11685812e+00 -4.27127689e-01 -1.96126644e-02 1.01891100e+00
1.38946974e+00 1.58668101e-01 -8.29208434e-01 -3.81131262e-01
-2.85470933e-01 -2.66989887e-01 6.45590201e-02 -3.67822409e-01
-5.48539340e-01 1.17841935e+00 6.90500736e-01 2.24711299e-01
8.49711776e-01 5.85689209e-02 1.13798714e+00 1.75104290e-01
2.29635313e-01 -1.01616931e+00 -1.53287604e-01 2.84914255e-01
9.06796396e-01 -1.02501583e+00 -6.87845409e-01 -1.73313171e-01
-9.02306855e-01 1.15608609e+00 7.01323226e-02 2.17741907e-01
5.33302605e-01 5.04812062e-01 4.27603543e-01 3.39086831e-01
-1.12075150e+00 1.35493502e-01 -2.25876458e-02 5.04106700e-01
4.71742541e-01 2.19220057e-01 4.60456014e-01 6.96987092e-01
-3.45477819e-01 -2.93147624e-01 4.52530295e-01 6.08419657e-01
-3.82027507e-01 -1.01174963e+00 -8.60303223e-01 2.22558249e-02
-5.49523175e-01 -3.44030470e-01 -1.63252994e-01 -1.79955244e-01
-4.30324763e-01 1.38131988e+00 2.40250707e-01 -5.28734267e-01
1.28359511e-01 5.01166523e-01 2.63214149e-02 -7.84366190e-01
-1.06744087e+00 4.84108180e-01 3.18648815e-01 -1.67701185e-01
-2.01259941e-01 -8.91370475e-01 -1.14787602e+00 -2.09516540e-01
-3.96072239e-01 5.68229556e-01 1.19430256e+00 9.60509837e-01
4.30515051e-01 5.19476771e-01 1.22632730e+00 -8.22094500e-01
-8.76141131e-01 -1.27734661e+00 -8.14548314e-01 7.62626249e-03
8.78291547e-01 2.56963164e-01 -5.92066586e-01 4.04995054e-01]
|
[14.597254753112793, 6.345139026641846]
|
3262e47d-a1fe-46f2-a9cf-9abdca1d2335
|
benchmarking-and-analyzing-3d-human-pose-and
|
2209.10529
| null |
https://arxiv.org/abs/2209.10529v1
|
https://arxiv.org/pdf/2209.10529v1.pdf
|
Benchmarking and Analyzing 3D Human Pose and Shape Estimation Beyond Algorithms
|
3D human pose and shape estimation (a.k.a. "human mesh recovery") has achieved substantial progress. Researchers mainly focus on the development of novel algorithms, while less attention has been paid to other critical factors involved. This could lead to less optimal baselines, hindering the fair and faithful evaluations of newly designed methodologies. To address this problem, this work presents the first comprehensive benchmarking study from three under-explored perspectives beyond algorithms. 1) Datasets. An analysis on 31 datasets reveals the distinct impacts of data samples: datasets featuring critical attributes (i.e. diverse poses, shapes, camera characteristics, backbone features) are more effective. Strategical selection and combination of high-quality datasets can yield a significant boost to the model performance. 2) Backbones. Experiments with 10 backbones, ranging from CNNs to transformers, show the knowledge learnt from a proximity task is readily transferable to human mesh recovery. 3) Training strategies. Proper augmentation techniques and loss designs are crucial. With the above findings, we achieve a PA-MPJPE of 47.3 mm on the 3DPW test set with a relatively simple model. More importantly, we provide strong baselines for fair comparisons of algorithms, and recommendations for building effective training configurations in the future. Codebase is available at http://github.com/smplbody/hmr-benchmarks
|
['Ziwei Liu', 'Tianwei Zhang', 'Lei Yang', 'Zhongang Cai', 'Hui En Pang']
|
2022-09-21
| null | null | null | null |
['3d-human-pose-and-shape-estimation', 'human-mesh-recovery']
|
['computer-vision', 'computer-vision']
|
[ 1.06981814e-01 2.97928788e-02 -2.99069434e-01 -2.37329021e-01
-1.15586030e+00 -4.02247578e-01 4.47471291e-01 -1.10628523e-01
-4.11421478e-01 5.43380082e-01 3.67472589e-01 1.22036815e-01
-1.19172834e-01 -5.21038294e-01 -9.00723755e-01 -6.16119802e-01
3.84340361e-02 5.79888582e-01 1.44654289e-01 -3.29702497e-01
1.49789959e-01 6.23744667e-01 -1.54789245e+00 3.48384567e-02
5.08131027e-01 1.05029309e+00 5.75503223e-02 2.28814781e-01
2.32037932e-01 1.69899195e-01 -3.40372860e-01 -7.17855752e-01
5.80537975e-01 2.45166626e-02 -7.07850575e-01 3.49864364e-02
8.93784404e-01 -4.66550261e-01 -2.04687446e-01 6.98825896e-01
1.03745461e+00 -8.35512707e-04 6.08313322e-01 -1.21151042e+00
-3.10082942e-01 2.24588558e-01 -5.99602818e-01 -1.44107014e-01
4.48880255e-01 4.66428578e-01 9.96990800e-01 -1.22984254e+00
8.27078044e-01 1.22740996e+00 1.06311560e+00 6.08095527e-01
-1.33689404e+00 -8.22728455e-01 2.73900107e-02 2.24066456e-03
-1.54739344e+00 -5.54679573e-01 7.33543336e-01 -4.04093415e-01
6.38573825e-01 3.43808234e-01 7.99012661e-01 1.60049927e+00
5.45006320e-02 7.28566587e-01 1.02029324e+00 -1.75647631e-01
-2.63735875e-02 1.64237306e-01 -2.26274028e-01 6.29061460e-01
5.10266185e-01 1.77593589e-01 -6.99684381e-01 -1.08873159e-01
8.23451042e-01 -2.64706522e-01 -3.76853853e-01 -8.09039831e-01
-1.19685555e+00 5.77632487e-01 3.70944560e-01 -4.75127958e-02
-3.28001261e-01 9.53699350e-02 6.04608238e-01 1.96634367e-01
5.41705906e-01 5.94399691e-01 -5.09812951e-01 -2.02639356e-01
-9.57210720e-01 7.11990833e-01 7.41958618e-01 1.15096450e+00
5.52968860e-01 -1.91466957e-01 -3.38064395e-02 8.02839935e-01
9.24356058e-02 6.97564542e-01 -1.33410871e-01 -1.10104370e+00
7.43596494e-01 4.11633551e-01 6.43088147e-02 -1.35189474e+00
-4.58265483e-01 -5.75051606e-01 -6.41124666e-01 2.20979333e-01
7.25157857e-01 -4.77683954e-02 -8.09095085e-01 1.59653139e+00
5.19351363e-01 -7.42560551e-02 -4.43245471e-01 1.08227372e+00
9.19674098e-01 1.29992232e-01 -2.79872846e-02 2.53718942e-01
1.44557130e+00 -7.77282834e-01 -3.76702338e-01 -1.99869320e-01
4.63608205e-01 -1.01741266e+00 1.33265734e+00 4.88120705e-01
-1.25606740e+00 -4.55599189e-01 -8.62674356e-01 -1.09536432e-01
-2.40928784e-01 1.38010532e-01 5.59539497e-01 4.88662660e-01
-9.15377021e-01 7.65015721e-01 -8.39802325e-01 -5.47590971e-01
5.27370155e-01 4.81116027e-01 -6.20269239e-01 -3.10437351e-01
-9.60405171e-01 8.85259926e-01 5.82806543e-02 2.18058094e-01
-7.28635311e-01 -9.22080874e-01 -6.35518134e-01 -3.53344530e-01
4.68770623e-01 -9.04676080e-01 1.22074902e+00 -5.94953418e-01
-1.28954375e+00 1.15194547e+00 3.27581167e-02 -2.10135952e-01
9.65719640e-01 -4.67336744e-01 2.36765090e-02 1.65028304e-01
1.62644163e-01 8.24820340e-01 7.31343985e-01 -1.39211237e+00
-4.81904685e-01 -3.89998585e-01 -1.09046571e-01 1.61020115e-01
-3.37650418e-01 4.38625226e-03 -8.38714063e-01 -8.87587726e-01
3.98410149e-02 -1.23958254e+00 -1.11308590e-01 2.39878312e-01
-5.30589581e-01 5.77928051e-02 3.77462447e-01 -8.33269775e-01
1.03137887e+00 -1.85330033e+00 3.44792008e-01 2.01811597e-01
3.32393765e-01 1.50887936e-01 -1.26937121e-01 4.97970700e-01
5.21081202e-02 1.14918925e-01 -3.15364376e-02 -7.32322395e-01
1.92464851e-02 -2.20365282e-02 3.29048745e-02 6.23874664e-01
2.98421681e-01 8.82853150e-01 -5.01754880e-01 -5.99743783e-01
2.42808759e-01 6.84494793e-01 -7.71527767e-01 1.54475987e-01
1.09241843e-01 7.24384665e-01 -4.21902776e-01 1.07324171e+00
6.54997230e-01 -3.96464676e-01 -1.46834761e-01 -6.91425741e-01
1.48274556e-01 6.55279532e-02 -1.22428370e+00 1.80966449e+00
-2.94249505e-01 4.94719267e-01 1.45906344e-01 -5.85046411e-01
7.25924790e-01 2.09157884e-01 6.35037601e-01 -7.17150807e-01
2.96726137e-01 3.35484922e-01 -1.73868150e-01 -4.74887252e-01
4.63705510e-01 1.43121421e-01 9.67162326e-02 7.09635243e-02
2.67322641e-02 4.02702726e-02 -4.51216884e-02 5.82629591e-02
9.77147281e-01 3.07066113e-01 9.81467366e-02 -3.57099056e-01
6.28698990e-02 1.33170143e-01 5.77942848e-01 4.64211702e-01
-2.83469170e-01 1.07675934e+00 2.23270237e-01 -3.83356988e-01
-1.23357403e+00 -1.09084225e+00 -3.75004500e-01 9.69823062e-01
1.97780401e-01 -6.33809507e-01 -7.74848104e-01 -4.31847543e-01
2.50419855e-01 1.87778249e-01 -8.00639629e-01 -3.70113272e-03
-8.43534887e-01 -6.28323495e-01 6.31583512e-01 7.26936221e-01
3.78361970e-01 -8.86456549e-01 -5.23563921e-01 -1.15830660e-01
-2.72876859e-01 -1.27560306e+00 -4.21441078e-01 -1.88519120e-01
-9.93484080e-01 -1.21126997e+00 -9.77072060e-01 -5.46532333e-01
6.90597296e-01 1.80950001e-01 1.28418899e+00 1.51187018e-01
-2.27761477e-01 5.56988418e-01 -3.60262364e-01 -2.87229747e-01
-1.12266928e-01 4.43048567e-01 1.14746056e-01 -2.91081607e-01
-3.47930640e-02 -6.45882487e-01 -1.07822084e+00 6.78628385e-01
-3.55531037e-01 1.78748801e-01 8.19322050e-01 6.49869382e-01
8.83464456e-01 -4.29195523e-01 2.59415388e-01 -9.27104473e-01
3.50274652e-01 -4.74586219e-01 -2.25258172e-01 8.78931805e-02
-7.17288613e-01 -2.05972075e-01 3.56957227e-01 -5.10351360e-01
-9.40719604e-01 -2.45767124e-02 -3.07124346e-01 -5.38614094e-01
-1.69117004e-01 1.57064468e-01 -2.51755953e-01 -2.65585124e-01
8.46329808e-01 -9.51905996e-02 2.21997350e-01 -7.33125508e-01
2.05313891e-01 2.34317005e-01 5.25589645e-01 -1.08171213e+00
8.82704318e-01 5.90828180e-01 -7.41877500e-03 -7.62433410e-01
-5.69906890e-01 -3.72906029e-01 -5.34103096e-01 -2.14042798e-01
5.67618847e-01 -1.08368242e+00 -3.88459533e-01 4.12890702e-01
-9.00293112e-01 -3.88054758e-01 -1.85347214e-01 3.28823090e-01
-5.54644167e-01 1.90540731e-01 -6.73310697e-01 -4.94038224e-01
-5.41552126e-01 -1.43066239e+00 1.44696474e+00 -8.91127959e-02
-5.40014625e-01 -8.05352628e-01 -1.46767125e-01 7.62881994e-01
3.87859613e-01 5.68994880e-01 5.18053412e-01 -5.09685814e-01
-6.94836497e-01 -3.42122018e-01 -1.59310743e-01 2.70456970e-01
-3.71001624e-02 4.11756299e-02 -1.07781637e+00 -5.64084172e-01
-3.35630447e-01 -4.62208599e-01 5.98866403e-01 3.89829576e-01
1.14381814e+00 -1.25773653e-01 -4.00399476e-01 8.32819998e-01
1.07867730e+00 -3.26326400e-01 6.96173728e-01 4.80943650e-01
1.00465524e+00 6.54054284e-01 7.78615057e-01 4.04434174e-01
4.49209124e-01 9.24280286e-01 4.01329428e-01 -2.43070006e-01
-4.77887958e-01 -3.59273344e-01 1.30686879e-01 7.71311343e-01
-5.16511321e-01 -2.22568959e-03 -1.18987715e+00 3.43587041e-01
-1.55648756e+00 -6.87557936e-01 -5.12575284e-02 2.21116090e+00
7.65645444e-01 2.58037984e-01 3.35218161e-01 7.01199919e-02
4.85835344e-01 9.19842348e-02 -5.24381399e-01 2.30278313e-01
3.79910320e-02 2.40619600e-01 4.81586456e-01 1.39150977e-01
-1.00649559e+00 8.10978949e-01 5.91412115e+00 1.05484581e+00
-1.08894002e+00 2.18424201e-02 6.62275732e-01 -3.68376821e-01
-2.73469180e-01 -2.17745304e-01 -1.00176358e+00 4.54646647e-01
5.62610447e-01 7.76086748e-02 3.19794118e-01 8.20145190e-01
2.28578106e-01 8.39019790e-02 -1.21555018e+00 1.18061590e+00
3.83674423e-03 -1.30339158e+00 8.66188668e-03 2.90476829e-01
6.00002766e-01 1.91706613e-01 1.41494751e-01 2.77358830e-01
-2.17521921e-01 -1.03682840e+00 9.95035946e-01 4.92644727e-01
9.88720834e-01 -6.54026270e-01 7.69202709e-01 7.81330615e-02
-1.22508287e+00 2.62196124e-01 -1.49569228e-01 1.40129194e-01
7.29864836e-02 5.19001067e-01 -8.04132998e-01 6.07489407e-01
9.30162549e-01 6.21205986e-01 -7.62861729e-01 1.05272889e+00
1.05897719e-02 6.45004153e-01 -4.08059776e-01 2.49214545e-01
-2.33015329e-01 9.98099148e-02 5.17529368e-01 1.23621035e+00
2.24871993e-01 -1.06288910e-01 1.69013336e-01 6.98417842e-01
-1.85206652e-01 3.70100677e-01 -5.78630984e-01 2.04820290e-01
6.60282075e-01 1.26215720e+00 -8.17453861e-01 1.37809813e-01
-4.57852334e-01 7.19168425e-01 3.96043360e-01 2.42659122e-01
-9.29587126e-01 1.20158426e-01 8.98267567e-01 7.96992600e-01
6.38834760e-02 -2.01615646e-01 -5.04890501e-01 -9.60651577e-01
2.36204281e-01 -1.19899809e+00 2.87435144e-01 -4.00970608e-01
-1.49316752e+00 4.08818662e-01 1.52968407e-01 -1.34901810e+00
4.67299372e-02 -6.33181274e-01 -3.49177927e-01 6.93555117e-01
-1.25121570e+00 -1.39196277e+00 -5.25164127e-01 4.55890507e-01
4.30258065e-01 4.12939973e-02 7.38757789e-01 6.86549187e-01
-6.54731750e-01 1.06171238e+00 -3.27315629e-01 1.62044197e-01
9.38242972e-01 -9.39854145e-01 6.13036871e-01 5.33364177e-01
-6.81610825e-03 6.79555118e-01 7.59037077e-01 -6.48750842e-01
-1.66597068e+00 -9.24996495e-01 3.71140867e-01 -9.56178248e-01
4.05122012e-01 -4.04363602e-01 -8.55879545e-01 6.03927732e-01
-1.58681214e-01 -2.97624730e-02 5.04695952e-01 3.81206304e-01
-2.40963653e-01 -8.96598697e-02 -1.03982270e+00 7.35851109e-01
1.46545064e+00 -3.64336431e-01 -1.23434566e-01 1.44702956e-01
6.01356268e-01 -8.50314856e-01 -1.24115562e+00 7.97751784e-01
9.00557637e-01 -9.32195663e-01 1.28952801e+00 -4.34948057e-01
5.86632967e-01 1.99748874e-02 -2.73872107e-01 -1.00166798e+00
-2.18375623e-01 -4.46923822e-01 -1.65778637e-01 1.20921135e+00
3.97497565e-01 -4.77577329e-01 1.10410261e+00 8.44111323e-01
-4.29387912e-02 -1.36046958e+00 -7.68070698e-01 -7.42957413e-01
2.67579854e-01 -5.54109693e-01 5.19709826e-01 9.13537920e-01
-4.57191139e-01 1.62930861e-01 -4.76980716e-01 3.38907018e-02
6.86410487e-01 -9.61662456e-02 1.29805052e+00 -1.20510256e+00
-1.91143081e-01 -4.60311353e-01 -3.32608819e-01 -8.17219555e-01
-4.07907665e-02 -7.91672051e-01 -1.59954011e-01 -1.35142028e+00
2.59142876e-01 -7.07315028e-01 1.95365231e-02 3.94338846e-01
-1.84787184e-01 4.47991192e-01 2.39314705e-01 3.15170735e-01
-4.52644229e-01 5.42164683e-01 1.50125515e+00 1.36491910e-01
3.37189026e-02 7.93626457e-02 -6.83063447e-01 7.32419431e-01
1.09217560e+00 -3.01679105e-01 -3.94948542e-01 -5.44901788e-01
2.42358133e-01 -1.22870043e-01 6.13844991e-01 -1.15801179e+00
1.07333377e-01 3.27595584e-02 3.28614950e-01 -4.11980659e-01
4.89193082e-01 -8.11633706e-01 3.79633665e-01 2.63077766e-01
-5.05201966e-02 2.90305942e-01 2.94676453e-01 4.24626917e-01
7.28146583e-02 1.39105618e-01 6.54308856e-01 -1.22755319e-01
-6.68294430e-01 5.05797982e-01 3.64173889e-01 5.34740567e-01
7.69623637e-01 -5.10199189e-01 -7.50643434e-03 -2.48557165e-01
-6.04854763e-01 1.68060943e-01 7.91735590e-01 4.78233725e-01
5.74083447e-01 -1.32571328e+00 -7.72173703e-01 1.55600095e-02
2.17192799e-01 1.82220995e-01 2.86172092e-01 1.01958930e+00
-5.68004966e-01 1.18874870e-01 -2.31622159e-01 -6.73256278e-01
-1.23934805e+00 1.82388604e-01 2.87559092e-01 -1.73538044e-01
-8.72087061e-01 8.07465792e-01 -5.98945562e-03 -5.15928447e-01
5.06298125e-01 -2.11278349e-01 1.71620518e-01 1.86558738e-01
3.10668617e-01 7.04906642e-01 3.92709762e-01 -7.78079689e-01
-4.13207173e-01 7.05826879e-01 -1.25174955e-01 1.48941517e-01
1.37263465e+00 4.71539237e-02 2.99512774e-01 1.56438828e-01
1.20341015e+00 1.21887274e-01 -1.29043376e+00 -1.32947326e-01
-9.04796124e-02 -6.56093121e-01 -1.33144245e-01 -8.04464877e-01
-1.32094634e+00 6.86370909e-01 6.31441653e-01 -3.35517287e-01
9.36433375e-01 1.89879239e-01 9.74250495e-01 8.08201805e-02
8.18711281e-01 -9.85948384e-01 1.21767312e-01 1.71471506e-01
1.13177216e+00 -1.37852967e+00 2.37248063e-01 -6.58479393e-01
-5.73465347e-01 7.49047160e-01 7.24317551e-01 -2.73826253e-03
5.42608976e-01 3.43322098e-01 5.19989766e-02 -5.07227600e-01
-3.80038410e-01 -5.46957478e-02 4.37041372e-01 4.23327506e-01
5.47511458e-01 3.76460925e-02 -2.08004788e-01 6.23043537e-01
-4.66750771e-01 -1.02468319e-01 -4.68856096e-02 7.41606414e-01
-1.01157893e-04 -1.06188786e+00 -5.41750729e-01 5.82821250e-01
-3.74295264e-01 1.05904847e-01 -3.14686269e-01 1.23411131e+00
1.45125702e-01 6.10721707e-01 -2.67950505e-01 -5.44697165e-01
8.93773019e-01 -1.73207700e-01 6.35167897e-01 -4.91215229e-01
-5.80740571e-01 -6.44175112e-02 3.15068871e-01 -9.34204102e-01
-3.12551647e-01 -8.06531250e-01 -9.80425775e-01 -8.11003029e-01
-3.21753383e-01 -3.33658785e-01 4.53374833e-01 5.27859509e-01
5.15628815e-01 3.48631203e-01 1.92085117e-01 -1.37763429e+00
-6.61207676e-01 -6.49245858e-01 -3.98101658e-01 6.92965090e-01
-2.81339791e-02 -1.13241982e+00 -3.77391130e-01 -2.40065828e-01]
|
[7.092249870300293, -1.0915299654006958]
|
bf01cd7c-5612-43f4-96da-3b70712b95bf
|
automatic-readability-assessment-of-german-1
|
2209.04299
| null |
https://arxiv.org/abs/2209.04299v1
|
https://arxiv.org/pdf/2209.04299v1.pdf
|
Automatic Readability Assessment of German Sentences with Transformer Ensembles
|
Reliable methods for automatic readability assessment have the potential to impact a variety of fields, ranging from machine translation to self-informed learning. Recently, large language models for the German language (such as GBERT and GPT-2-Wechsel) have become available, allowing to develop Deep Learning based approaches that promise to further improve automatic readability assessment. In this contribution, we studied the ability of ensembles of fine-tuned GBERT and GPT-2-Wechsel models to reliably predict the readability of German sentences. We combined these models with linguistic features and investigated the dependence of prediction performance on ensemble size and composition. Mixed ensembles of GBERT and GPT-2-Wechsel performed better than ensembles of the same size consisting of only GBERT or GPT-2-Wechsel models. Our models were evaluated in the GermEval 2022 Shared Task on Text Complexity Assessment on data of German sentences. On out-of-sample data, our best ensemble achieved a root mean squared error of 0.435.
|
['Stephan Bialonski', 'Niklas Grieger', 'Tobias Bornheim', 'Patrick Gustav Blaneck']
|
2022-09-09
|
automatic-readability-assessment-of-german
|
https://aclanthology.org/2022.germeval-1.10
|
https://aclanthology.org/2022.germeval-1.10.pdf
|
germeval-2022-9
|
['lexical-complexity-prediction']
|
['natural-language-processing']
|
[-4.83055748e-02 2.26552531e-01 3.19313526e-01 -3.65022600e-01
-1.07709014e+00 -5.75517058e-01 7.63861597e-01 6.77766919e-01
-4.96145964e-01 8.69274378e-01 6.05242610e-01 -3.88462096e-01
-2.57023007e-01 -9.15843189e-01 -3.83546203e-01 -2.60771632e-01
1.58467859e-01 6.26853406e-01 2.75172926e-02 -4.53133881e-01
2.68142909e-01 -1.12277001e-01 -1.46836364e+00 4.91650313e-01
1.51937580e+00 7.38357484e-01 2.41965428e-01 9.94460702e-01
-1.31342225e-02 6.15769982e-01 -5.61865509e-01 -6.38841212e-01
-6.35728007e-03 -4.93549049e-01 -9.80140746e-01 -5.52262008e-01
7.43330419e-01 1.61137581e-01 -3.79121810e-01 7.71142960e-01
7.05829740e-01 1.23855643e-01 9.98642206e-01 -4.40116197e-01
-6.24430358e-01 8.40627313e-01 -4.21296060e-03 3.15975428e-01
6.10674202e-01 2.07543135e-01 1.18170094e+00 -7.08548367e-01
6.26994431e-01 1.11349750e+00 9.50116932e-01 3.89717847e-01
-1.22541583e+00 -3.38722914e-01 -1.06005386e-01 3.22487444e-01
-1.16793883e+00 -4.04471129e-01 1.15890376e-01 -7.89895773e-01
1.59385061e+00 3.89585614e-01 5.12876391e-01 9.69904363e-01
4.96676922e-01 5.43621302e-01 1.65548897e+00 -8.22846174e-01
6.22168602e-03 -6.28148066e-03 4.20754910e-01 7.79543877e-01
3.25568974e-01 -6.07125908e-02 -6.03519499e-01 -9.42851305e-02
1.71814218e-01 -7.70129442e-01 -3.36152017e-01 4.13239241e-01
-1.27569234e+00 7.78006673e-01 2.93954939e-01 4.67372596e-01
-3.43300477e-02 -2.62288600e-01 4.09584701e-01 5.68233013e-01
9.05553520e-01 9.75409627e-01 -8.44681501e-01 -5.51058710e-01
-9.85479653e-01 3.27754050e-01 9.63637352e-01 7.81207383e-01
5.22337914e-01 -4.56211925e-01 -8.28747749e-01 1.12868929e+00
-3.15661952e-02 4.89774525e-01 7.45683670e-01 -5.05278051e-01
8.91143501e-01 8.68299007e-01 -9.07081217e-02 -7.77287960e-01
-7.15358198e-01 -6.15755200e-01 -9.04829979e-01 1.13643430e-01
6.47535861e-01 -2.46432349e-01 -7.19021559e-01 1.45526326e+00
-3.87259841e-01 -5.60682893e-01 -4.43698540e-02 3.81716520e-01
1.09957981e+00 5.74825108e-01 -7.44825751e-02 1.08664185e-01
1.13943350e+00 -8.80838692e-01 -4.17455912e-01 -3.44250739e-01
1.10740769e+00 -7.33993888e-01 1.30778301e+00 4.27031904e-01
-1.15443599e+00 -6.96448505e-01 -9.66265976e-01 -2.44492263e-01
-4.66821313e-01 3.21919829e-01 8.70190710e-02 9.13016737e-01
-1.30124795e+00 9.38007057e-01 -5.48767149e-01 -3.79898489e-01
2.31876835e-01 1.44969061e-01 -5.46326935e-01 4.11097333e-02
-1.23074901e+00 1.39082730e+00 5.39407969e-01 9.67882760e-03
-2.88756967e-01 -3.92200440e-01 -5.83515048e-01 1.39393106e-01
-2.50573546e-01 -8.74483109e-01 1.05018389e+00 -7.53860772e-01
-1.50825346e+00 1.16252279e+00 -8.06078091e-02 -5.06036818e-01
7.84840405e-01 -3.11335474e-01 -4.78124261e-01 -4.69831020e-01
-1.13290310e-01 1.83691069e-01 8.57052207e-02 -4.60646838e-01
-3.62465978e-01 -2.12778464e-01 -3.09687048e-01 2.38565072e-01
-5.93905389e-01 3.46763693e-02 1.94512844e-01 -4.26173300e-01
-1.87742844e-01 -8.52736473e-01 1.33521885e-01 -7.30261266e-01
-6.46458685e-01 -4.25588518e-01 -1.62999973e-01 -1.19468904e+00
1.74070203e+00 -1.60192728e+00 3.85627359e-01 1.48738846e-01
4.59084809e-01 5.03617346e-01 -5.21627605e-01 7.57512987e-01
1.60086855e-01 3.75199646e-01 2.55329814e-02 -2.48412117e-01
1.94986299e-01 -5.25405288e-01 2.83405364e-01 9.96425226e-02
3.07346910e-01 1.01693964e+00 -9.51413751e-01 -2.47207567e-01
4.70097959e-02 1.16934799e-01 -6.00804448e-01 2.72687972e-01
-3.37820232e-01 1.12737529e-01 -1.76433295e-01 1.18265525e-01
3.17468882e-01 -1.73138484e-01 -5.18342666e-02 2.12458864e-01
-1.54710665e-01 8.68654788e-01 -5.86758137e-01 1.31175506e+00
-7.80163527e-01 1.02203941e+00 -5.52706003e-01 -3.52164686e-01
1.13023782e+00 1.47622809e-01 -1.44623801e-01 -7.36872196e-01
-3.32363392e-03 5.22074401e-01 4.20125246e-01 -3.64362299e-01
6.11190140e-01 1.43258482e-01 -1.06231920e-01 3.07247460e-01
2.55623609e-01 -2.56794214e-01 5.57530820e-01 -3.38118151e-02
1.48506677e+00 1.40091907e-02 3.90997291e-01 -6.81134105e-01
7.59195089e-01 -3.77583772e-01 1.31882757e-01 8.26292932e-01
-7.92474151e-02 6.42207563e-01 4.92228687e-01 -3.66880596e-01
-1.30220580e+00 -8.54753017e-01 -2.27045879e-01 1.05946398e+00
-6.54082417e-01 -8.65775645e-01 -1.01702094e+00 -4.56761390e-01
-2.44556725e-01 1.16287804e+00 -4.59238917e-01 -2.99003661e-01
-2.26481229e-01 -9.38038170e-01 6.84254408e-01 3.21257502e-01
5.48674405e-01 -9.39825773e-01 -2.24513814e-01 3.76135647e-01
-4.74304348e-01 -9.54303443e-01 -4.35746640e-01 -5.84495347e-03
-6.77722216e-01 -8.51647913e-01 -6.24228954e-01 -6.88507080e-01
2.76010662e-01 -4.74859387e-01 1.86028016e+00 2.21539184e-01
-6.81924373e-02 1.50151119e-01 -5.66765130e-01 -5.40970325e-01
-8.94596994e-01 7.12744355e-01 -2.09542826e-01 -4.67385024e-01
4.39992189e-01 -2.90851206e-01 -3.45542014e-01 1.49987146e-01
-3.55973780e-01 3.37181121e-01 4.82643813e-01 1.05218911e+00
2.00240463e-01 -2.12097943e-01 5.85384190e-01 -7.81955004e-01
1.01379347e+00 -7.90292993e-02 -3.35470527e-01 4.76884842e-01
-8.15059841e-01 2.94420809e-01 5.17258346e-01 -1.99243397e-01
-7.84351110e-01 -4.63646173e-01 -4.71575469e-01 4.11104649e-01
-8.86828005e-02 7.73574591e-01 -2.03494042e-01 5.55049926e-02
9.27484334e-01 1.50965050e-01 -2.88013786e-01 -5.53891599e-01
1.89642280e-01 1.00638390e+00 2.37160653e-01 -5.83320320e-01
4.42440778e-01 -6.72537565e-01 -1.42812476e-01 -8.90705526e-01
-1.00043416e+00 -5.72636649e-02 -6.89133763e-01 -1.56684220e-01
8.01435173e-01 -1.01236379e+00 -2.74750084e-01 8.63083899e-01
-1.14560795e+00 -5.50144315e-01 -3.44281793e-02 2.79273689e-01
-5.72509766e-01 2.89547890e-01 -6.83539510e-01 -6.53351426e-01
-7.14621365e-01 -9.19781029e-01 9.48528290e-01 1.60524935e-01
-6.43401265e-01 -1.20420456e+00 4.32785958e-01 5.53114116e-01
6.00632548e-01 1.09108135e-01 1.14436316e+00 -1.06126702e+00
-1.73333362e-01 -3.70600969e-01 -9.81024057e-02 5.66335738e-01
-2.63524102e-03 -3.34830135e-02 -9.17739809e-01 -3.74527454e-01
-5.45877934e-01 -4.10902113e-01 9.89147186e-01 2.56010294e-01
1.09538496e+00 -3.10946792e-01 -1.03120860e-02 4.46052670e-01
1.10333979e+00 -3.71331453e-01 7.93540716e-01 6.87446117e-01
5.40792644e-01 4.46620256e-01 1.29025549e-01 2.33022377e-01
5.82801938e-01 6.25507772e-01 -3.18964347e-02 2.14936525e-01
-2.59840339e-01 -2.75982231e-01 4.28118914e-01 1.12808168e+00
-6.58250749e-01 -5.45428693e-01 -1.28071392e+00 1.81393474e-01
-1.32600594e+00 -9.50042248e-01 -4.01397139e-01 2.42420292e+00
9.90570843e-01 3.48999172e-01 5.51527773e-04 1.66713417e-01
5.67455769e-01 -2.89583188e-02 -4.96276543e-02 -7.53331602e-01
-5.21326184e-01 3.90470833e-01 1.52952701e-01 6.52383029e-01
-1.03855562e+00 8.02688658e-01 6.10235643e+00 8.66652250e-01
-4.96175200e-01 2.56735459e-02 9.21928585e-01 1.12684533e-01
-4.04900849e-01 -2.72805125e-01 -1.01859558e+00 5.65425158e-01
1.40574026e+00 -4.58948135e-01 4.51228768e-01 8.31442624e-02
1.83748201e-01 -1.17872149e-01 -1.26209009e+00 6.73296332e-01
6.80424049e-02 -1.01081049e+00 -1.47408038e-01 1.03558257e-01
1.14941323e+00 3.30058396e-01 -1.89587116e-01 5.44278204e-01
3.00053507e-01 -1.28995550e+00 6.06035233e-01 8.93504024e-01
8.66583467e-01 -5.36669612e-01 1.04167771e+00 6.47259772e-01
-8.51515174e-01 -1.23853505e-01 -4.25142795e-01 -3.24818462e-01
-2.10142717e-01 9.20022905e-01 -7.40574419e-01 6.01981997e-01
3.80567163e-01 5.28482378e-01 -1.14129698e+00 1.19485497e+00
-1.27957821e-01 7.88914263e-01 -1.89920291e-01 -4.92144078e-01
-2.88560260e-02 -1.23306423e-01 5.47095060e-01 1.47009540e+00
7.22723424e-01 -8.43825340e-02 -1.19176857e-01 7.10148335e-01
-2.42354125e-01 5.36083043e-01 -1.54161707e-01 -3.20203573e-01
3.93348098e-01 1.10067701e+00 -2.80720890e-01 -3.05615127e-01
-3.23467761e-01 8.37530196e-01 8.89655650e-01 -1.67301804e-01
-3.16740483e-01 -5.36230147e-01 3.82587671e-01 3.25458288e-01
1.34480437e-02 -1.72064215e-01 -3.73542756e-01 -1.37518322e+00
-1.25928894e-01 -1.10597575e+00 2.49882117e-01 -6.87891483e-01
-1.53375912e+00 1.06020200e+00 -2.62901068e-01 -1.01103234e+00
-3.64379734e-01 -8.86281133e-01 -6.31468356e-01 1.47530162e+00
-1.14777911e+00 -8.79793048e-01 -5.01495302e-01 2.37146746e-02
4.85644996e-01 -6.02006674e-01 1.14636528e+00 -7.26109967e-02
-5.60955942e-01 7.73427129e-01 7.20664501e-01 2.63813883e-02
8.56828868e-01 -1.57536697e+00 7.64327049e-01 6.58078074e-01
1.37023538e-01 3.51397157e-01 7.65153468e-01 -5.84293187e-01
-7.76364326e-01 -1.11010373e+00 1.58563030e+00 -8.73183548e-01
5.48487723e-01 -3.74385774e-01 -8.32391441e-01 3.02283943e-01
4.55221862e-01 -5.41297138e-01 6.68003798e-01 5.63297808e-01
-4.54321533e-01 5.19863255e-02 -9.33056712e-01 2.98396379e-01
1.03001916e+00 -6.86381161e-01 -6.68273270e-01 4.83976781e-01
3.76141250e-01 -4.85343993e-01 -1.09891760e+00 2.85436809e-01
5.91575205e-01 -1.33870101e+00 5.75903118e-01 -5.05476236e-01
8.54902804e-01 3.32703233e-01 -1.57941177e-01 -1.99758613e+00
-6.44177556e-01 -4.28362787e-01 1.46950901e-01 1.14108348e+00
8.29108715e-01 -7.31644869e-01 3.37960213e-01 5.18220663e-01
-3.34120303e-01 -7.02937961e-01 -9.14469838e-01 -8.27269077e-01
8.68328571e-01 -7.56593123e-02 4.19270933e-01 4.97450292e-01
2.25586176e-01 4.96613920e-01 5.09304740e-02 -3.83532315e-01
1.73615858e-01 -3.34320724e-01 5.35978496e-01 -1.56635380e+00
-4.42925274e-01 -1.01125860e+00 -5.17567337e-01 -6.31976366e-01
2.81730384e-01 -1.28651989e+00 6.46074489e-02 -1.86848438e+00
4.68246311e-01 -5.20965196e-02 -1.81088731e-01 3.80170196e-01
-7.34892786e-01 -3.89004052e-02 1.17531858e-01 -1.50403410e-01
-4.07151222e-01 4.78865862e-01 1.05853772e+00 -2.37231776e-01
-2.23636568e-01 1.34327114e-02 -5.94597697e-01 2.92833000e-01
8.96162391e-01 6.84245005e-02 -1.08440213e-01 -5.09799004e-01
5.58693767e-01 -1.52551353e-01 6.84582219e-02 -1.29934216e+00
-1.30134732e-01 2.69932330e-01 5.45343757e-01 -2.02332050e-01
1.12468377e-02 4.66566570e-02 -1.64091424e-03 3.47736418e-01
-7.02704489e-01 3.77187371e-01 2.32683390e-01 1.83980182e-01
-1.66497171e-01 -4.08732593e-01 4.77085173e-01 -6.65143579e-02
-1.46908775e-01 -1.17329312e-02 -5.30511916e-01 3.51612389e-01
4.60146189e-01 2.06475127e-02 -6.43295646e-01 -3.87676388e-01
-5.10162711e-01 1.11426719e-01 4.26208615e-01 3.65950346e-01
1.78104594e-01 -1.00447857e+00 -1.40686047e+00 1.64732680e-01
3.49721879e-01 -3.46236497e-01 1.92759946e-01 7.25560367e-01
-5.77570081e-01 7.13054001e-01 -9.57085341e-02 -2.18642116e-01
-1.35786712e+00 3.68684530e-02 5.85989296e-01 -9.48193789e-01
-2.86182284e-01 9.86523271e-01 -2.18744382e-01 -8.66840839e-01
-1.99102372e-01 -3.20496202e-01 -3.15828264e-01 9.30339620e-02
5.20974219e-01 7.42526889e-01 6.46901071e-01 -5.78615904e-01
-8.70924070e-02 4.47006285e-01 -1.25611976e-01 -5.52387387e-02
1.29291081e+00 5.48543893e-02 -3.34671676e-01 5.95714390e-01
9.68008399e-01 8.50385278e-02 -5.64570487e-01 -1.39645031e-02
2.95635253e-01 -1.38312265e-01 2.96478957e-01 -1.37824118e+00
-3.38563859e-01 1.06130242e+00 5.50830722e-01 2.05216751e-01
1.05950642e+00 -2.91763902e-01 4.01559174e-01 5.27630746e-01
4.41785723e-01 -1.12828195e+00 -2.10705400e-01 1.00032377e+00
1.06009650e+00 -1.25438797e+00 -1.28475532e-01 -1.58877090e-01
-4.54913348e-01 1.41344905e+00 4.96862233e-01 7.65938163e-02
4.51164424e-01 6.74728751e-02 -1.70647487e-01 2.59560764e-01
-1.01967895e+00 -9.80512574e-02 9.27805007e-01 4.76797700e-01
1.18244457e+00 4.07915801e-01 -6.00888014e-01 6.93351090e-01
-8.44429672e-01 -9.59371030e-02 6.63952708e-01 3.45915318e-01
-5.18950582e-01 -1.32275796e+00 -1.15759812e-01 1.12479353e+00
-1.74111769e-01 -4.95603353e-01 -5.72875559e-01 4.81380492e-01
-1.34433717e-01 1.12438977e+00 1.64570138e-02 -6.06240153e-01
4.65862125e-01 4.16887969e-01 8.24332595e-01 -6.21946394e-01
-8.07010770e-01 -2.97723860e-01 7.76514113e-01 -2.19164550e-01
4.60865796e-02 -9.50379014e-01 -4.72863436e-01 -3.76685500e-01
-3.79988074e-01 9.63582918e-02 3.62322092e-01 1.11545050e+00
2.80204058e-01 4.92602557e-01 2.71350354e-01 -5.83259165e-01
-8.47516835e-01 -1.70255160e+00 -4.34717327e-01 2.51937807e-01
2.37766102e-01 -1.21410549e-01 -3.85208488e-01 -2.43685171e-01]
|
[11.060815811157227, 10.20193099975586]
|
27662242-f885-4efb-bda8-15230d139cf8
|
blind-audio-source-separation-with-minimum
|
1907.02404
| null |
https://arxiv.org/abs/1907.02404v2
|
https://arxiv.org/pdf/1907.02404v2.pdf
|
Blind Audio Source Separation with Minimum-Volume Beta-Divergence NMF
|
Considering a mixed signal composed of various audio sources and recorded with a single microphone, we consider on this paper the blind audio source separation problem which consists in isolating and extracting each of the sources. To perform this task, nonnegative matrix factorization (NMF) based on the Kullback-Leibler and Itakura-Saito $\beta$-divergences is a standard and state-of-the-art technique that uses the time-frequency representation of the signal. We present a new NMF model better suited for this task. It is based on the minimization of $\beta$-divergences along with a penalty term that promotes the columns of the dictionary matrix to have a small volume. Under some mild assumptions and in noiseless conditions, we prove that this model is provably able to identify the sources. In order to solve this problem, we propose multiplicative updates whose derivations are based on the standard majorization-minimization framework. We show on several numerical experiments that our new model is able to obtain more interpretable results than standard NMF models. Moreover, we show that it is able to recover the sources even when the number of sources present into the mixed signal is overestimated. In fact, our model automatically sets sources to zero in this situation, hence performs model order selection automatically.
|
['Nicolas Gillis', 'Man Shun Ang', 'Valentin Leplat']
|
2019-07-04
| null | null | null | null |
['audio-source-separation']
|
['audio']
|
[ 3.81072700e-01 -9.77265239e-02 3.26750129e-01 9.23456699e-02
-8.41571391e-01 -6.13449574e-01 2.70116597e-01 7.83948302e-02
-4.51480359e-01 7.33476877e-01 1.08496606e-01 -2.06594676e-01
-5.55146158e-01 -3.82776111e-01 -5.27139723e-01 -9.69893336e-01
-2.35737965e-01 3.76386940e-01 -1.64326072e-01 -2.99777120e-01
1.34562403e-01 4.24865186e-01 -1.68212938e+00 -4.74428609e-02
9.10744071e-01 9.44921792e-01 3.45146388e-01 8.32265794e-01
8.79156310e-03 6.33374810e-01 -6.54001355e-01 -2.33875409e-01
5.19468606e-01 -6.31009102e-01 -4.79288459e-01 1.09390289e-01
1.34112462e-01 1.59623533e-01 2.37944156e-01 1.36672962e+00
5.39407790e-01 2.69019246e-01 7.08414972e-01 -1.05305529e+00
-1.14338443e-01 6.01195633e-01 -4.18498456e-01 2.43945628e-01
3.97942811e-01 -5.26976109e-01 9.84526455e-01 -1.07843804e+00
3.37767094e-01 9.72840846e-01 6.43420637e-01 3.54929924e-01
-1.29966235e+00 -4.47442800e-01 -1.16569504e-01 1.71451300e-01
-1.48703825e+00 -6.17485106e-01 1.06270301e+00 -5.11066139e-01
5.17626882e-01 7.46260524e-01 4.64912266e-01 7.39811063e-01
-2.24181294e-01 6.41821325e-01 9.41260874e-01 -8.69202673e-01
3.38202447e-01 1.87362909e-01 1.26778200e-01 3.06124330e-01
2.60804266e-01 -9.34969708e-02 -5.77816248e-01 -5.49304724e-01
4.43178058e-01 -2.09403232e-01 -7.18704343e-01 -4.08027917e-01
-1.14596593e+00 8.31006110e-01 -8.43244195e-02 6.59480691e-01
-4.70193535e-01 -9.72119495e-02 3.40074040e-02 2.55209059e-01
4.77704138e-01 4.21221375e-01 -1.26182377e-01 -4.48194034e-02
-1.05394578e+00 1.95628345e-01 9.14574146e-01 5.38340509e-01
5.59924483e-01 2.05334008e-01 2.97267735e-01 9.88163352e-01
3.46398681e-01 6.64973855e-01 4.43973154e-01 -9.16685283e-01
4.01766896e-01 2.49277696e-01 3.24956119e-01 -1.23022580e+00
-3.16479295e-01 -7.16505051e-01 -9.82713342e-01 2.18197376e-01
4.36145782e-01 -1.90203652e-01 -6.43092453e-01 1.89498091e+00
2.57057101e-01 5.37803054e-01 9.49201360e-02 9.36837137e-01
2.12354735e-01 7.57002771e-01 -5.31622946e-01 -1.02664363e+00
9.42627311e-01 -4.48109955e-01 -1.09904253e+00 -3.46940830e-02
1.29727736e-01 -9.39986885e-01 7.12857485e-01 9.30446148e-01
-1.18959248e+00 -4.15807366e-01 -1.20951176e+00 3.39873046e-01
-1.40478790e-01 2.38636434e-01 2.01111123e-01 8.17711174e-01
-9.94918168e-01 4.80025738e-01 -6.02783084e-01 1.32127911e-01
-2.82870203e-01 3.95513177e-01 -3.76088947e-01 2.83650488e-01
-1.11190283e+00 5.23142695e-01 6.95563555e-02 5.33298016e-01
-6.64229631e-01 -4.52522188e-01 -6.23453140e-01 -6.77941972e-03
4.19746011e-01 -5.48394501e-01 9.96384859e-01 -1.26143861e+00
-1.43087053e+00 4.33739096e-01 -5.24650395e-01 -4.31838691e-01
6.42197311e-01 -1.95502475e-01 -4.77725953e-01 3.93004775e-01
6.80523291e-02 -1.38101652e-01 1.52297735e+00 -1.52551258e+00
-4.25129026e-01 -4.81555641e-01 -3.59279037e-01 -5.55308238e-02
-5.03288329e-01 6.77885935e-02 -1.89804018e-01 -1.03885245e+00
3.75768334e-01 -9.34604406e-01 -2.97579020e-01 -4.06506658e-01
-3.74129385e-01 1.12173565e-01 3.77842426e-01 -7.50762582e-01
1.33928370e+00 -2.40836334e+00 6.63836122e-01 6.51658297e-01
2.58104563e-01 2.21987918e-01 4.64233644e-02 4.09190029e-01
-3.77504617e-01 -1.64150193e-01 -6.24233961e-01 -7.30489492e-01
-1.29770949e-01 1.69235822e-02 -3.93201351e-01 6.59798265e-01
-3.33782345e-01 -2.39455178e-02 -7.54391611e-01 -2.72682071e-01
1.87668607e-01 6.80841804e-01 -4.34601784e-01 2.22930089e-01
1.54812425e-01 4.59496677e-01 4.63095419e-02 1.49466276e-01
6.91966295e-01 2.30006278e-01 1.60336077e-01 -9.75427590e-03
-1.25740096e-01 1.13676069e-03 -2.03209162e+00 1.59460986e+00
-5.26950419e-01 4.79109794e-01 8.29077303e-01 -1.17373085e+00
8.18208277e-01 5.89514792e-01 6.85758293e-01 -8.26320201e-02
1.55894578e-01 6.77185178e-01 -2.21963469e-02 -4.18904334e-01
2.34216511e-01 -4.63071227e-01 2.48787940e-01 3.12760264e-01
1.28141150e-01 8.26880112e-02 4.34746295e-01 1.17968559e-01
7.23545969e-01 -5.61212063e-01 2.66473800e-01 -4.83930051e-01
8.44311059e-01 -5.63448608e-01 5.09298027e-01 5.65899730e-01
1.33623332e-01 6.93353474e-01 2.43389979e-01 1.29524514e-01
-5.32000959e-01 -9.49052036e-01 -6.06268980e-02 8.11144829e-01
-9.53851268e-02 -4.57619041e-01 -8.43796194e-01 -2.89962143e-01
-1.18299693e-01 7.24406421e-01 -5.91478348e-01 8.01443011e-02
-4.91082370e-01 -9.29193437e-01 2.62054443e-01 2.10797638e-02
7.01152440e-03 -6.37925684e-01 -4.84856844e-01 3.01411837e-01
-5.31890452e-01 -7.47368276e-01 -2.76818544e-01 5.06763875e-01
-6.27817214e-01 -9.75590825e-01 -1.11145389e+00 -6.04202807e-01
6.77610576e-01 3.00690383e-01 7.07034707e-01 -4.19370115e-01
-2.91531868e-02 4.28317219e-01 -4.11654323e-01 -4.04895186e-01
-5.16405880e-01 -3.72879118e-01 3.79403770e-01 6.50100768e-01
-7.15943426e-02 -9.17701304e-01 -2.41218090e-01 1.91229150e-01
-1.03358281e+00 -3.50701988e-01 2.48014212e-01 6.59239829e-01
5.59253991e-01 5.44989645e-01 5.83628058e-01 -4.98520195e-01
8.20399821e-01 -3.84447902e-01 -6.07768059e-01 2.75292154e-02
-5.22266269e-01 5.79728410e-02 8.82275462e-01 -4.51565355e-01
-8.20446610e-01 2.68068761e-01 -2.19546631e-01 -4.77132112e-01
7.61312693e-02 3.79549861e-01 -3.91974449e-01 -1.66066483e-01
6.67144001e-01 2.79199094e-01 -1.96869597e-01 -9.67508793e-01
3.25087577e-01 7.37450898e-01 5.57803571e-01 -2.79845327e-01
8.24001789e-01 4.96634722e-01 6.35017268e-03 -1.17076635e+00
-4.79789019e-01 -7.86425352e-01 -4.42988515e-01 -3.08681756e-01
5.48534155e-01 -5.96128762e-01 -6.25587046e-01 3.09406728e-01
-1.19911516e+00 1.38085231e-01 -3.00273955e-01 7.51117587e-01
-5.36954820e-01 5.32958508e-01 -3.19899231e-01 -1.49363339e+00
-1.92241251e-01 -1.02256620e+00 7.58761823e-01 -2.70483106e-01
-1.12761624e-01 -8.43211770e-01 3.27265054e-01 7.65880123e-02
1.96445763e-01 3.68665047e-02 7.14558005e-01 -5.82147956e-01
8.58882219e-02 -1.00800104e-01 2.68589497e-01 8.00531149e-01
2.16882899e-01 -2.46456891e-01 -1.01384044e+00 -2.14422137e-01
7.14595795e-01 5.01586556e-01 1.00202501e+00 5.46664476e-01
7.73649454e-01 -4.78639752e-01 -4.39701863e-02 5.12694001e-01
1.45840240e+00 2.98559397e-01 2.48203784e-01 3.52909826e-02
6.19279087e-01 7.62015045e-01 2.55563378e-01 5.45382142e-01
-1.64324597e-01 7.89944351e-01 5.31170964e-01 1.61058545e-01
1.73772752e-01 8.84250626e-02 4.89075631e-01 1.04957426e+00
-9.53090265e-02 -3.75521660e-01 -6.95037484e-01 7.01448023e-01
-1.83244455e+00 -9.14104044e-01 -3.89205873e-01 2.69317508e+00
7.01935351e-01 9.14243534e-02 3.84957016e-01 9.38948631e-01
7.45957613e-01 -1.57367904e-02 -3.68461832e-02 -2.18922317e-01
-3.17847103e-01 1.98070198e-01 4.06612486e-01 1.06507504e+00
-9.74396229e-01 1.16791040e-01 6.28961754e+00 8.73005569e-01
-1.04474759e+00 1.49086222e-01 -6.48656040e-02 -7.66431242e-02
-3.77017349e-01 -1.62276238e-01 -4.08004016e-01 6.02330983e-01
9.01450515e-01 -1.64324507e-01 6.33421540e-01 5.13374448e-01
3.36892009e-01 -1.09797738e-01 -7.75056064e-01 1.37807214e+00
4.12167460e-01 -8.07600319e-01 -7.96771571e-02 -6.67548040e-03
4.41037118e-01 -4.51187521e-01 5.32156713e-02 -4.17634249e-01
-2.39605114e-01 -8.04334998e-01 9.52008069e-01 3.82591605e-01
3.65062565e-01 -9.05811727e-01 6.18832469e-01 5.87356389e-01
-1.07367754e+00 -2.82054454e-01 -1.02662683e-01 -1.15921050e-01
4.47992414e-01 1.27200806e+00 -5.62795639e-01 7.73285627e-01
2.76009142e-01 4.61917728e-01 -1.70388430e-01 1.20119357e+00
-1.99486151e-01 7.02503145e-01 -6.97089195e-01 3.21219891e-01
-5.80318011e-02 -4.87791300e-01 1.11460817e+00 1.21301877e+00
7.55435765e-01 -9.51363742e-02 -3.89383622e-02 6.52469516e-01
1.94946349e-01 4.23155338e-01 -4.76058125e-01 2.44600013e-01
5.33562452e-02 9.88507390e-01 -7.01575994e-01 -2.09528357e-01
3.11895553e-03 9.65125263e-01 -1.84591576e-01 4.18731749e-01
-4.60071057e-01 -4.47006911e-01 5.58609188e-01 1.00294583e-01
2.64004052e-01 -2.77016491e-01 -1.81777462e-01 -1.36481893e+00
3.16055179e-01 -9.01298761e-01 2.97358066e-01 -4.86338794e-01
-1.06274974e+00 7.78869212e-01 5.97974733e-02 -1.38593006e+00
-5.24902999e-01 -4.48654473e-01 -2.60414511e-01 8.76568556e-01
-1.27219331e+00 -4.76875246e-01 2.37578526e-01 8.69236171e-01
3.41627955e-01 -5.12568429e-02 9.49452460e-01 5.95546484e-01
-5.10036767e-01 2.81486720e-01 4.67686862e-01 -1.41690999e-01
6.05857849e-01 -1.42759192e+00 -3.33589464e-01 1.22317839e+00
6.22438073e-01 7.50718474e-01 1.31015646e+00 -2.98483372e-01
-1.19971871e+00 -5.79399824e-01 1.03353047e+00 -1.34547651e-01
6.01392031e-01 -5.07926702e-01 -7.68321097e-01 3.64008665e-01
1.30019858e-01 -1.92966297e-01 8.89214456e-01 2.04566251e-02
-1.68758124e-01 -3.12730730e-01 -8.74584377e-01 7.89065361e-02
6.28964603e-01 -4.83929813e-01 -6.90091133e-01 4.03590262e-01
2.87445068e-01 -4.96351756e-02 -5.15963018e-01 2.85045892e-01
2.79938370e-01 -1.18019199e+00 9.01727438e-01 -2.12187469e-01
-6.94546476e-02 -5.70161819e-01 -3.03247184e-01 -1.43426108e+00
-1.47613212e-01 -1.13140881e+00 -1.62787557e-01 1.16542315e+00
4.88874078e-01 -6.24907851e-01 3.66797745e-01 8.65924060e-02
6.44331127e-02 -3.95568520e-01 -1.31414080e+00 -9.63146567e-01
-2.74636030e-01 -8.87963116e-01 1.36219099e-01 7.73438990e-01
2.73157448e-01 3.67702991e-01 -7.76581168e-01 2.89513171e-01
8.29688251e-01 6.80348575e-02 4.22483653e-01 -1.34419811e+00
-6.86880648e-01 -3.88754547e-01 -2.96733022e-01 -9.43742275e-01
2.58711338e-01 -6.51034534e-01 1.98094577e-01 -1.20422721e+00
-2.25691035e-01 -3.11473429e-01 -3.76139581e-01 4.99038883e-02
-7.62912333e-02 1.75703451e-01 4.23998535e-01 1.46928817e-01
-2.07738712e-01 3.48901838e-01 7.23935306e-01 -1.07229829e-01
-4.69378769e-01 4.34517473e-01 -5.85915148e-01 9.49978352e-01
4.68235254e-01 -4.82163280e-01 -4.34321016e-01 -3.11404049e-01
6.20033622e-01 1.94269985e-01 2.28593305e-01 -1.10354161e+00
1.75323665e-01 2.83892661e-01 2.71685570e-02 -2.62800962e-01
6.24009192e-01 -1.16597104e+00 4.32521820e-01 2.44787127e-01
-3.43114346e-01 -1.87176839e-01 6.32948205e-02 6.81351244e-01
-4.37894464e-01 -5.53182065e-01 6.30410671e-01 1.50123268e-01
-1.75357193e-01 -2.36671910e-01 -5.64783633e-01 -2.54970491e-01
6.93557143e-01 6.44332767e-02 3.79293621e-01 -8.43440413e-01
-1.03637266e+00 -2.10584745e-01 5.85020147e-02 8.98719504e-02
5.38055539e-01 -1.30387318e+00 -8.53585482e-01 3.86992186e-01
-1.89710766e-01 -2.86227614e-01 2.87012547e-01 1.09262300e+00
-1.86880186e-01 2.43505210e-01 2.11942151e-01 -4.36778158e-01
-1.44630301e+00 7.49238908e-01 2.87149698e-01 -1.35314405e-01
-1.26175046e-01 8.48491788e-01 4.22836989e-02 -2.74238195e-02
3.26319426e-01 -3.06181341e-01 -3.23017091e-01 4.45911139e-01
8.22939992e-01 7.43601620e-01 2.59955347e-01 -9.57522750e-01
-3.79927009e-01 6.89189970e-01 5.23358166e-01 -7.65504599e-01
1.18120539e+00 -2.11591214e-01 -4.69474852e-01 6.85696423e-01
1.38463867e+00 8.20394933e-01 -7.36197114e-01 7.64872059e-02
-1.69578612e-01 -3.96854222e-01 1.76473092e-02 -5.18210053e-01
-9.85682726e-01 9.52826142e-01 5.67980647e-01 6.98032856e-01
1.54073453e+00 -3.76299560e-01 6.02396727e-01 1.46490172e-01
3.52595270e-01 -8.83231878e-01 -4.57185179e-01 2.19520643e-01
9.53696430e-01 -8.65676343e-01 -2.29399249e-01 -6.32444859e-01
-1.42552689e-01 1.20605695e+00 -2.21183598e-01 -6.68359622e-02
8.91052008e-01 2.70879388e-01 1.45552501e-01 3.46068412e-01
-3.14842910e-01 -3.80562663e-01 3.52839082e-01 4.84271437e-01
1.44897670e-01 1.32919133e-01 -5.49487412e-01 7.94123530e-01
-2.83524960e-01 -2.13029683e-01 4.73658681e-01 5.23739994e-01
-4.19030130e-01 -1.26124763e+00 -9.46104705e-01 1.95969746e-01
-7.48366058e-01 -9.30842236e-02 -4.78526562e-01 2.00418979e-01
2.36144096e-01 1.51353693e+00 -4.97732311e-01 -3.39006007e-01
3.33835840e-01 1.03671633e-01 3.47300202e-01 -4.43872511e-01
-4.54232275e-01 8.01714301e-01 -1.25339523e-01 -2.86890596e-01
-7.74816036e-01 -5.99628091e-01 -9.92551267e-01 5.60349301e-02
-4.98557329e-01 8.08070958e-01 7.26665616e-01 9.38167751e-01
1.30515367e-01 3.61152947e-01 7.51777112e-01 -8.67991149e-01
-5.65362871e-01 -9.18127835e-01 -1.09855413e+00 3.48207921e-01
8.30739141e-01 -6.26762569e-01 -9.84595478e-01 2.50341505e-01]
|
[15.310540199279785, 5.633208274841309]
|
fd53123d-0f1b-4136-8e83-c6751b9c01be
|
differentiable-top-k-classification-learning-1
|
2206.07290
| null |
https://arxiv.org/abs/2206.07290v1
|
https://arxiv.org/pdf/2206.07290v1.pdf
|
Differentiable Top-k Classification Learning
|
The top-k classification accuracy is one of the core metrics in machine learning. Here, k is conventionally a positive integer, such as 1 or 5, leading to top-1 or top-5 training objectives. In this work, we relax this assumption and optimize the model for multiple k simultaneously instead of using a single k. Leveraging recent advances in differentiable sorting and ranking, we propose a differentiable top-k cross-entropy classification loss. This allows training the network while not only considering the top-1 prediction, but also, e.g., the top-2 and top-5 predictions. We evaluate the proposed loss function for fine-tuning on state-of-the-art architectures, as well as for training from scratch. We find that relaxing k does not only produce better top-5 accuracies, but also leads to top-1 accuracy improvements. When fine-tuning publicly available ImageNet models, we achieve a new state-of-the-art for these models.
|
['Oliver Deussen', 'Christian Borgelt', 'Hilde Kuehne', 'Felix Petersen']
|
2022-06-15
|
differentiable-top-k-classification-learning
|
https://openreview.net/forum?id=6PTUd_zPdHL
|
https://openreview.net/pdf?id=6PTUd_zPdHL
| null |
['classification']
|
['methodology']
|
[ 1.45474464e-01 -6.25855178e-02 -5.61499059e-01 -7.11542904e-01
-9.98615265e-01 -4.18690950e-01 2.70102888e-01 2.61643738e-01
-6.98106587e-01 4.58760381e-01 5.25522195e-02 -1.79077342e-01
-2.22948343e-01 -7.66536772e-01 -8.02972257e-01 -5.16919732e-01
-1.30497143e-01 2.66169399e-01 2.46953845e-01 5.92009304e-03
2.68949896e-01 3.13256949e-01 -1.50787413e+00 6.02325261e-01
9.31799650e-01 1.61281192e+00 -4.39826138e-02 5.77871084e-01
1.54230744e-01 6.06268585e-01 -2.17026934e-01 -7.93423355e-01
3.21728885e-01 3.86684947e-02 -8.23806822e-01 -3.58027726e-01
9.37007725e-01 -5.05080760e-01 -3.23842138e-01 1.01378846e+00
1.75467283e-01 2.07865406e-02 5.30926108e-01 -1.25843740e+00
-9.23232377e-01 8.11746240e-01 -3.45455289e-01 -8.71604085e-02
-3.18676919e-01 -2.19523422e-02 1.49993384e+00 -7.68174469e-01
1.27703756e-01 1.13411629e+00 6.74950719e-01 5.62268138e-01
-1.22899187e+00 -6.79893434e-01 4.64192986e-01 2.38837108e-01
-1.31469572e+00 -3.50039214e-01 6.23804033e-01 -5.05230010e-01
9.45893824e-01 1.24365725e-01 4.31493640e-01 7.93357074e-01
3.45949471e-01 9.62507129e-01 1.00898051e+00 -3.31506670e-01
1.92485034e-01 2.36766532e-01 4.42704439e-01 5.95751107e-01
1.98998705e-01 -2.10752748e-02 -4.76745367e-01 -2.66819522e-02
4.54129964e-01 1.51129410e-01 -1.44224122e-01 -2.57839024e-01
-1.28073978e+00 6.78033113e-01 7.27077842e-01 3.94450985e-02
-3.74056071e-01 5.86453319e-01 5.11799872e-01 2.02828452e-01
6.10406876e-01 4.68780816e-01 -8.23560596e-01 -8.04982111e-02
-1.21140695e+00 1.72248617e-01 5.20288169e-01 5.86119831e-01
9.19945657e-01 -3.07151616e-01 -5.23917258e-01 9.57975924e-01
1.15600377e-01 7.43393451e-02 3.23237270e-01 -9.12540376e-01
7.22551525e-01 6.99828267e-01 7.06700236e-02 -6.93139911e-01
-3.38591576e-01 -7.02158511e-01 -9.48588252e-01 3.19565237e-02
3.30104351e-01 2.05404192e-01 -1.09717476e+00 2.24008226e+00
5.59647791e-02 1.89404175e-01 -4.42664623e-02 7.34954774e-01
5.67029893e-01 5.16111612e-01 1.71793818e-01 1.23999175e-02
1.23628724e+00 -1.31378901e+00 -2.18264282e-01 -3.01197588e-01
8.02822948e-01 -4.85164165e-01 1.29303086e+00 5.01665592e-01
-9.32069838e-01 -4.90788192e-01 -9.18210268e-01 -1.56405985e-01
-4.83850211e-01 5.46689987e-01 6.29331470e-01 4.61652398e-01
-1.21084321e+00 1.04753995e+00 -8.29954386e-01 -3.18081640e-02
4.14125502e-01 4.15562749e-01 -3.04558247e-01 -5.77046536e-02
-1.21294355e+00 7.98015654e-01 4.73367095e-01 1.36615988e-02
-8.51525664e-01 -1.03132451e+00 -4.35225248e-01 3.44275296e-01
3.99569482e-01 -7.08245337e-01 1.32676935e+00 -7.19495416e-01
-1.41484404e+00 9.27623451e-01 -7.75564685e-02 -7.28606701e-01
6.47220790e-01 -5.17793298e-01 -2.43833929e-01 -1.50086075e-01
-1.41669378e-01 8.29404473e-01 5.49846530e-01 -1.03659511e+00
-8.25422168e-01 -3.65039080e-01 4.76095617e-01 1.41618863e-01
-8.12087715e-01 -1.54353514e-01 -4.94949669e-01 -4.64005560e-01
-1.08672872e-01 -9.63559031e-01 -2.71027923e-01 5.63646331e-02
-5.84271193e-01 -3.74057859e-01 2.15312257e-01 -6.00767314e-01
1.38761568e+00 -2.26042795e+00 -1.41364232e-01 1.65224671e-01
1.88674182e-01 2.58330286e-01 -2.23893598e-01 8.80361870e-02
7.04057217e-02 5.24474263e-01 -1.48490801e-01 -5.07781923e-01
1.55515969e-01 -4.51859161e-02 -1.23081893e-01 2.05904737e-01
3.49408835e-01 8.44897509e-01 -8.59686613e-01 -1.43230274e-01
1.02381930e-01 5.00391066e-01 -8.82665157e-01 -1.20662481e-01
-1.45676255e-01 -3.23250294e-02 -4.32182282e-01 2.07578838e-01
6.38660431e-01 -7.51991928e-01 -9.31337103e-02 -3.25612247e-01
4.66508456e-02 5.65262735e-01 -9.35453951e-01 1.52724147e+00
-6.84764266e-01 3.10834736e-01 -1.20309800e-01 -9.44236517e-01
6.49022281e-01 -5.29310182e-02 3.68606329e-01 -7.10411608e-01
-1.35491326e-01 3.68017703e-01 -7.36374035e-02 8.63921940e-02
6.52511656e-01 1.96408316e-01 -3.58720087e-02 4.52087037e-02
2.58950759e-02 4.90775734e-01 1.23618409e-01 6.06420636e-02
9.39210951e-01 -1.02795355e-01 2.27995470e-01 -3.46961260e-01
4.31024641e-01 -2.43081376e-01 6.20055616e-01 7.37001479e-01
-7.70365298e-02 4.40045238e-01 5.14717877e-01 -4.18031067e-01
-1.01943862e+00 -7.53009200e-01 -1.25249773e-01 1.42742646e+00
1.15958802e-01 -4.70927566e-01 -7.20092773e-01 -8.13579679e-01
2.67180443e-01 7.66393602e-01 -7.68720329e-01 -4.89912659e-01
-5.65454483e-01 -6.69381142e-01 4.69629496e-01 7.05219567e-01
5.68405688e-01 -7.41544366e-01 -3.27472657e-01 1.25885010e-01
-1.42103761e-01 -1.20025194e+00 -6.21206999e-01 4.20749694e-01
-1.05625844e+00 -7.52569795e-01 -7.26013720e-01 -6.05571926e-01
6.52142286e-01 1.60394415e-01 1.35465038e+00 1.44060016e-01
2.20245168e-01 -1.07478835e-01 -3.43080431e-01 -2.74318773e-02
-1.72141090e-01 5.30574501e-01 7.16929957e-02 -4.09377888e-02
1.11870103e-01 -3.21803570e-01 -8.97001386e-01 3.67988199e-01
-7.26692259e-01 1.68340981e-01 6.27522528e-01 8.87418866e-01
8.63255322e-01 1.45157687e-02 7.34685719e-01 -7.89690137e-01
3.29694837e-01 -3.64596136e-02 -6.16750121e-01 5.78009605e-01
-8.66406202e-01 1.12272330e-01 1.16959441e+00 -3.92508566e-01
-5.64783692e-01 -1.13621511e-01 -7.59459659e-02 -6.14165604e-01
4.53850962e-02 5.21797121e-01 5.71438447e-02 -1.52738661e-01
5.59409201e-01 1.88604832e-01 -5.03073096e-01 -5.32230198e-01
4.54129159e-01 6.37681484e-01 4.72780645e-01 -5.54887593e-01
4.19349849e-01 2.61441499e-01 -1.57317534e-01 -4.11882192e-01
-1.35731184e+00 -5.05842924e-01 -4.47775781e-01 1.18853264e-01
6.99118316e-01 -9.53809738e-01 -7.28656173e-01 5.33103824e-01
-8.33761454e-01 -4.41955328e-01 -1.15025222e-01 3.59636366e-01
-4.39331800e-01 5.82362004e-02 -6.71810985e-01 -5.91808140e-01
-7.00083256e-01 -1.34562707e+00 9.71945047e-01 1.72681272e-01
2.24351868e-01 -8.20640981e-01 -1.90767765e-01 3.87225628e-01
4.71569598e-01 -2.00451985e-01 9.64459777e-01 -8.90065789e-01
-6.85719430e-01 -2.52450585e-01 -4.29695845e-01 7.53190339e-01
-9.99998301e-02 -2.40832437e-02 -1.07303107e+00 -4.80207771e-01
-4.89679545e-01 -5.13239264e-01 1.44224906e+00 3.12707037e-01
1.75188458e+00 -4.87712413e-01 -2.94449091e-01 8.58584702e-01
1.50322831e+00 -1.13068588e-01 5.98073721e-01 4.03781116e-01
6.69922829e-01 2.86419898e-01 6.31772220e-01 3.76815945e-01
7.00180709e-01 8.16649497e-01 5.31024635e-01 1.27262041e-01
-8.29463638e-03 -4.69764143e-01 2.07171410e-01 7.23823786e-01
1.82831094e-01 -2.05254048e-01 -8.31599116e-01 7.14208305e-01
-1.80839145e+00 -5.85169435e-01 2.45107949e-01 2.47628403e+00
9.86195683e-01 2.81616509e-01 -2.95226891e-02 -6.54550195e-02
7.21127868e-01 1.03388876e-01 -8.18598688e-01 -3.07487458e-01
1.63048357e-01 1.26682580e-01 7.85372257e-01 3.96032035e-01
-1.31365728e+00 1.09298241e+00 5.93972492e+00 1.11135292e+00
-1.48289227e+00 -1.21078707e-01 1.15048802e+00 -2.86625594e-01
-2.95017689e-01 9.84921604e-02 -1.18204701e+00 5.79498470e-01
9.21262681e-01 -7.12892339e-02 7.41469800e-01 1.08798683e+00
-6.92948326e-02 2.30689555e-01 -1.42940891e+00 9.59456325e-01
-1.67965293e-01 -1.30501461e+00 2.11962998e-01 1.54280305e-01
8.30115080e-01 2.63443947e-01 1.20061941e-01 4.23626453e-01
2.20140725e-01 -1.05333769e+00 9.51776683e-01 2.82404035e-01
8.03235829e-01 -7.54424393e-01 7.00105846e-01 3.75578851e-01
-1.15316641e+00 -1.01841830e-01 -4.51305628e-01 3.03199947e-01
-2.18128245e-02 9.62144852e-01 -4.97769445e-01 3.99031103e-01
9.51824605e-01 7.81892002e-01 -4.01683986e-01 1.24930739e+00
-1.68107510e-01 6.99335694e-01 -5.12453377e-01 -6.77179620e-02
3.54699075e-01 -1.44022489e-02 3.85387093e-02 1.24906099e+00
3.11678082e-01 -1.10976815e-01 3.98138553e-01 5.80676913e-01
-6.98091567e-01 1.62878886e-01 -9.46043432e-02 -7.06674978e-02
6.17309988e-01 1.30636311e+00 -4.57461327e-01 -4.78905678e-01
-3.61837596e-01 8.59632075e-01 7.73165703e-01 2.58695364e-01
-7.60888755e-01 -4.54230994e-01 9.51888382e-01 -8.47937465e-02
5.61843872e-01 -9.27888080e-02 -4.56448108e-01 -1.21275651e+00
2.30973452e-01 -6.39232397e-01 4.19909239e-01 -3.23840469e-01
-1.51431227e+00 5.49837708e-01 -1.03025571e-01 -1.03135371e+00
1.71989396e-01 -9.14726377e-01 -3.93342882e-01 6.48553491e-01
-1.96652412e+00 -1.17573917e+00 -3.04714590e-01 1.94606557e-01
9.76345837e-02 7.54545555e-02 5.84101439e-01 5.66286564e-01
-5.39489865e-01 1.21939671e+00 4.37633485e-01 1.20842367e-01
7.13754117e-01 -1.15726328e+00 4.68041420e-01 5.11513293e-01
6.04390763e-02 5.41930795e-01 2.94859827e-01 -2.92200744e-01
-8.55895817e-01 -1.14802229e+00 9.47946608e-01 -5.86940274e-02
7.11331964e-01 -3.46438468e-01 -8.79324019e-01 5.91444850e-01
-2.12081492e-01 1.93017840e-01 6.01199448e-01 3.24670851e-01
-6.14461124e-01 -6.13318324e-01 -1.16010571e+00 5.15988886e-01
1.11210477e+00 -5.39160490e-01 -1.02137908e-01 2.62313932e-01
1.00860548e+00 -3.66094649e-01 -1.12663519e+00 6.41732931e-01
7.62968004e-01 -8.85802567e-01 1.09798479e+00 -8.21727633e-01
8.14575374e-01 -9.39239189e-03 -4.85878021e-01 -1.47120702e+00
-5.00250459e-01 -1.99996099e-01 -3.15341622e-01 1.02788854e+00
8.57591748e-01 -7.25336969e-01 1.04692674e+00 6.30661011e-01
-2.53427982e-01 -1.38091850e+00 -9.00655210e-01 -8.51976812e-01
4.62162346e-01 -4.36147779e-01 5.83751440e-01 7.88569570e-01
-2.75023580e-01 7.73223862e-02 -5.27998686e-01 1.22752786e-01
6.18364215e-01 2.10162342e-01 5.04012406e-01 -1.19111383e+00
-4.05635208e-01 -6.81823552e-01 -4.14373249e-01 -1.38203382e+00
1.30633757e-01 -9.97876525e-01 1.39600644e-02 -1.49023652e+00
5.03442943e-01 -7.34732032e-01 -9.63071346e-01 7.78082013e-01
-2.42135108e-01 2.68179804e-01 3.48404616e-01 2.75873899e-01
-7.48353541e-01 6.81882441e-01 1.25325191e+00 1.89135130e-02
-5.44509329e-02 -1.20967463e-01 -9.71085250e-01 6.17363989e-01
8.53649437e-01 -4.47031736e-01 -2.93937564e-01 -6.51746035e-01
4.10695940e-01 -3.10678303e-01 4.02822196e-01 -9.78208661e-01
1.62777305e-01 -2.84080237e-01 2.63930440e-01 -5.52998781e-01
4.50191617e-01 -6.07258320e-01 -3.75132561e-01 2.68126547e-01
-7.55050123e-01 -1.73500210e-01 6.55182227e-02 4.89437908e-01
-2.62402922e-01 -1.68052688e-01 1.06898248e+00 -7.08951950e-02
-6.99642360e-01 5.16926408e-01 4.87194568e-01 2.14568228e-01
8.46995831e-01 3.92559031e-03 -6.24349356e-01 -1.11877993e-01
-5.24648786e-01 5.47256112e-01 5.33736706e-01 3.69142592e-01
4.13627297e-01 -1.48265100e+00 -7.10372865e-01 -1.15728013e-01
3.87365341e-01 -4.52511432e-03 3.45415622e-01 8.39266598e-01
-3.15426826e-01 5.36526263e-01 -2.52286419e-02 -6.42808259e-01
-1.14061940e+00 2.86835641e-01 4.44950491e-01 -8.62806797e-01
-3.96267593e-01 1.04747164e+00 4.64891255e-01 -6.25827670e-01
4.67407137e-01 -5.52528501e-01 -1.01072021e-01 -1.69911385e-01
5.00595391e-01 2.02877998e-01 3.32086384e-01 -2.51738012e-01
-5.98440289e-01 5.90350509e-01 -5.94198883e-01 2.06056640e-01
1.30379283e+00 1.98337495e-01 7.72802010e-02 3.69654626e-01
1.38693798e+00 -3.47654819e-01 -1.28281641e+00 -3.59638870e-01
-1.71404183e-01 -5.44939756e-01 2.72518694e-01 -1.06180155e+00
-1.19871593e+00 9.27971005e-01 4.60177332e-01 -2.05631889e-02
1.19366705e+00 -6.11568242e-02 9.14054692e-01 6.91124678e-01
5.14663815e-01 -1.18294144e+00 1.38099357e-01 7.91650772e-01
6.49025440e-01 -1.23275483e+00 -1.36401311e-01 -1.31743655e-01
-4.80537117e-01 7.76341796e-01 7.87824035e-01 -2.02787936e-01
7.92170465e-01 -9.21637565e-02 -2.08112746e-01 1.36573583e-01
-1.02984381e+00 1.30524024e-01 5.11484444e-01 -1.09291356e-02
6.82955444e-01 3.17323655e-01 -1.21986203e-01 7.18660235e-01
-3.67242366e-01 2.39942491e-01 7.49198198e-02 6.13941729e-01
-4.37594920e-01 -9.54280376e-01 1.00327946e-01 9.85134184e-01
-6.91071332e-01 -4.50927705e-01 -2.60922611e-01 4.26658183e-01
-7.30199441e-02 6.28203988e-01 -3.90777364e-02 -8.53545189e-01
3.21819335e-01 -1.65748131e-02 2.71418601e-01 -2.93916464e-01
-6.53640807e-01 -5.75720727e-01 7.68983662e-02 -7.12893903e-01
-7.22900182e-02 -2.32643619e-01 -1.02847135e+00 -5.19681096e-01
-5.00517666e-01 2.72348225e-02 7.32031524e-01 8.05135965e-01
5.81311285e-01 4.21338022e-01 6.57000780e-01 -6.36693776e-01
-1.06741893e+00 -9.15422618e-01 -3.49904239e-01 3.95455122e-01
4.04686451e-01 -5.96264184e-01 -4.78555322e-01 -2.44843349e-01]
|
[9.321444511413574, 3.316835641860962]
|
642f8cef-1df1-4738-8bc4-f7dfbe46f4de
|
achieving-long-term-fairness-in-submodular
|
2304.04700
| null |
https://arxiv.org/abs/2304.04700v1
|
https://arxiv.org/pdf/2304.04700v1.pdf
|
Achieving Long-term Fairness in Submodular Maximization through Randomization
|
Submodular function optimization has numerous applications in machine learning and data analysis, including data summarization which aims to identify a concise and diverse set of data points from a large dataset. It is important to implement fairness-aware algorithms when dealing with data items that may contain sensitive attributes like race or gender, to prevent biases that could lead to unequal representation of different groups. With this in mind, we investigate the problem of maximizing a monotone submodular function while meeting group fairness constraints. Unlike previous studies in this area, we allow for randomized solutions, with the objective being to calculate a distribution over feasible sets such that the expected number of items selected from each group is subject to constraints in the form of upper and lower thresholds, ensuring that the representation of each group remains balanced in the long term. Here a set is considered feasible if its size does not exceed a constant value of $b$. Our research includes the development of a series of approximation algorithms for this problem.
|
['Twumasi Mensah-Boateng', 'Jing Yuan', 'Shaojie Tang']
|
2023-04-10
| null | null | null | null |
['data-summarization']
|
['miscellaneous']
|
[ 2.13504866e-01 3.92345011e-01 -8.30161870e-01 -6.55419707e-01
-3.96036237e-01 -3.93151850e-01 -1.50730342e-01 8.42068076e-01
-5.05747616e-01 1.09551775e+00 3.57306927e-01 3.76634635e-02
-4.99743640e-01 -8.26066256e-01 -4.39034730e-01 -6.83421493e-01
-1.51972800e-01 6.07779860e-01 -4.82751191e-01 -1.27884876e-02
4.73003238e-01 3.42721701e-01 -1.39503014e+00 -6.43144995e-02
1.21275043e+00 1.01839912e+00 -2.92743653e-01 2.70854477e-02
-5.43268919e-02 4.85306174e-01 -7.85324574e-01 -6.61911786e-01
6.05709314e-01 -5.17223120e-01 -5.78680098e-01 2.43427604e-01
3.68590176e-01 -4.35283333e-01 2.77808756e-01 1.31074297e+00
5.31221271e-01 4.40134287e-01 6.62380457e-01 -1.61176395e+00
-3.52071762e-01 8.11870694e-01 -9.80016172e-01 -1.04525730e-01
2.26369649e-01 -1.69466406e-01 1.27717531e+00 -1.19684562e-01
5.83184123e-01 1.21534967e+00 7.66482502e-02 6.49864316e-01
-1.36149561e+00 -6.60774112e-01 4.36659604e-01 -1.84487954e-01
-1.19630194e+00 -5.05493462e-01 5.34944773e-01 -1.14100382e-01
3.93707424e-01 8.34680617e-01 5.70651293e-01 1.73685282e-01
1.59450904e-01 4.98007834e-01 6.11025512e-01 -4.25249547e-01
6.45808160e-01 2.99961597e-01 2.70601749e-01 5.72002791e-02
1.14693499e+00 -3.98856729e-01 -6.71472311e-01 -6.46439075e-01
-3.40712257e-03 -1.74441431e-02 -3.33047330e-01 -6.99549854e-01
-8.38761866e-01 1.17566597e+00 2.40065262e-01 -1.44136757e-01
-7.77725458e-01 1.18784666e-01 3.74646932e-01 2.10524186e-01
5.33553123e-01 5.98554194e-01 -8.10788125e-02 3.61694425e-01
-1.09807837e+00 7.29106963e-01 7.97220290e-01 8.49449575e-01
4.99581635e-01 -1.73550844e-01 -5.29742837e-01 5.87054968e-01
1.23096861e-01 2.90099710e-01 1.18982904e-02 -1.22161531e+00
8.28694344e-01 8.32530260e-01 3.59137654e-01 -1.14712071e+00
-1.02279447e-01 -1.32020712e-01 -8.12638164e-01 1.12065651e-01
4.79275823e-01 -3.53587449e-01 -4.08213615e-01 2.00108218e+00
5.42169631e-01 -7.30601907e-01 -2.35302806e-01 9.68072593e-01
3.52517068e-01 6.49239063e-01 1.27870262e-01 -1.10524201e+00
1.08107209e+00 -1.50604889e-01 -9.21533346e-01 -1.10414840e-01
4.82663244e-01 -3.19049925e-01 5.31073809e-01 4.09332424e-01
-1.38883340e+00 2.80567259e-01 -8.57909381e-01 6.35697693e-02
3.49453017e-02 -3.69555980e-01 5.82906187e-01 9.83728349e-01
-6.05503261e-01 2.41516382e-01 -9.88754258e-02 -2.43749127e-01
8.02464902e-01 6.73989236e-01 -2.15599269e-01 -6.15071468e-02
-8.02981675e-01 7.33275592e-01 3.79234314e-01 4.03039344e-02
-3.38722110e-01 -7.64323413e-01 -7.89005935e-01 4.45553988e-01
6.74287140e-01 -8.59841704e-01 8.53588998e-01 -1.18283665e+00
-6.50362194e-01 8.36876214e-01 -3.13500822e-01 -5.70869625e-01
7.76326716e-01 2.54601389e-01 2.41363287e-01 -2.40063921e-01
2.80774832e-01 4.78778362e-01 4.27464753e-01 -1.16504574e+00
-7.66859651e-01 -8.98694992e-01 1.24632612e-01 5.60432553e-01
-6.40349746e-01 1.75701663e-01 4.76168618e-02 -3.56591910e-01
-1.31352648e-01 -7.04558074e-01 -6.53650224e-01 -8.46180096e-02
-4.85527307e-01 -1.34460911e-01 2.74491698e-01 -4.98049110e-01
1.43159711e+00 -1.84491658e+00 2.52520859e-01 6.67630851e-01
2.05961376e-01 -1.39949784e-01 1.36029959e-01 2.91399688e-01
3.13405901e-01 3.19133490e-01 -3.93437386e-01 -2.68728316e-01
1.02243140e-01 1.44935533e-01 -7.75715187e-02 7.82633007e-01
-1.98028982e-01 3.93408477e-01 -5.10709167e-01 -2.11278558e-01
-2.11673841e-01 -4.59519029e-01 -7.57144928e-01 1.07010357e-01
-2.28168473e-01 -3.11614275e-01 -4.10023689e-01 5.83133578e-01
1.00641775e+00 3.56086373e-01 3.94515216e-01 3.27518702e-01
-2.91952435e-02 -5.03591001e-02 -1.55195999e+00 8.46459627e-01
1.02778010e-01 1.94680959e-01 3.30680251e-01 -1.29037094e+00
9.04507816e-01 -3.84937450e-02 9.13757145e-01 -5.42956650e-01
3.06411892e-01 2.17024267e-01 1.65041871e-02 -3.89742889e-02
7.86117315e-01 -2.85744429e-01 -3.76200348e-01 6.63318038e-01
-6.65671647e-01 1.14453323e-01 5.22630334e-01 3.15346241e-01
5.32018363e-01 -8.83594453e-01 6.15418196e-01 -5.85578382e-01
4.04416054e-01 1.73560977e-01 1.01423120e+00 6.87418222e-01
-2.77286261e-01 4.23419118e-01 9.82996225e-01 -2.32142210e-01
-9.35644448e-01 -6.91653073e-01 -8.76164138e-02 9.42738354e-01
3.05691332e-01 9.17239860e-02 -6.16922379e-01 -6.48866594e-01
6.40223861e-01 8.97071123e-01 -7.15560973e-01 -2.81262845e-01
-2.09665060e-01 -1.12364733e+00 -1.83618590e-01 1.10415794e-01
2.69789994e-01 -6.12812221e-01 -8.23541820e-01 5.96215762e-02
-1.51242077e-01 -5.25342166e-01 -1.01079690e+00 1.71984546e-02
-7.72716403e-01 -1.19362855e+00 -7.75665462e-01 -4.92563725e-01
1.09930933e+00 2.45890945e-01 8.45116198e-01 -2.22128689e-01
-2.18578205e-01 9.61649418e-02 -1.60327420e-01 -1.03306568e+00
-1.59287918e-02 -5.35465181e-02 8.74445960e-02 1.38350710e-01
4.15243953e-01 1.08116500e-01 -5.71361303e-01 1.57653332e-01
-8.67486835e-01 -2.75588095e-01 -7.97896385e-02 5.13823986e-01
5.63725770e-01 2.54451662e-01 1.00629056e+00 -9.37162280e-01
1.05399597e+00 -8.07382524e-01 -7.14028358e-01 5.02595425e-01
-7.92479515e-01 -1.97870180e-01 3.94705594e-01 -3.66419762e-01
-8.43305051e-01 -2.70281136e-01 6.72106445e-01 1.38004869e-01
4.98441815e-01 5.25677919e-01 -6.48618162e-01 2.65229434e-01
3.36904645e-01 -9.64299142e-02 4.66663152e-01 3.73684727e-02
2.56086081e-01 7.04708993e-01 7.86353201e-02 -4.79500949e-01
2.47134522e-01 4.53047454e-01 1.51142448e-01 -4.60202426e-01
-6.00718677e-01 -3.90061140e-01 -1.34872928e-01 -3.05505812e-01
3.29106897e-01 -6.10763729e-01 -9.80394244e-01 3.32766622e-02
-6.38263226e-01 2.17266709e-01 -6.92605257e-01 2.96114862e-01
-5.03048360e-01 1.43209800e-01 2.63836682e-01 -1.23289299e+00
-4.93684322e-01 -6.97579622e-01 3.01840961e-01 4.39564556e-01
-5.19123673e-01 -5.66845477e-01 -1.23654231e-01 5.02221525e-01
2.47514755e-01 6.94866002e-01 9.25800025e-01 -6.64649069e-01
-4.61490780e-01 -3.93128604e-01 1.50882095e-01 1.75325677e-01
3.03210378e-01 -1.30858108e-01 -1.72524199e-01 -5.93247712e-01
4.45350222e-02 -2.25144699e-01 6.18176997e-01 9.25816715e-01
1.32758927e+00 -9.28085268e-01 -1.88855797e-01 3.82274449e-01
1.31635427e+00 3.77618492e-01 3.03318232e-01 3.08994949e-01
1.22827344e-01 1.18812633e+00 8.59913528e-01 1.03631628e+00
6.31349623e-01 5.67793012e-01 5.24846733e-01 9.76806656e-02
6.87704921e-01 1.15183815e-01 5.73450252e-02 -3.16621840e-01
2.52556711e-01 -7.05249131e-01 -4.95367736e-01 8.07096183e-01
-2.11196232e+00 -1.12074339e+00 -2.11135987e-02 2.87084031e+00
6.22319996e-01 -2.55968809e-01 7.35739350e-01 2.04071030e-01
9.80549335e-01 3.18891294e-02 -6.96574032e-01 -1.17545748e+00
-2.27228463e-01 -3.82157087e-01 8.29524219e-01 3.33413929e-01
-6.59318209e-01 8.14997032e-02 6.09243774e+00 4.79629368e-01
-8.32543790e-01 -5.39183021e-01 1.25315678e+00 -8.32003236e-01
-8.81689012e-01 -5.07368296e-02 -6.49614632e-01 6.04220986e-01
5.43150187e-01 -1.13426709e+00 3.67551595e-01 4.62064862e-01
7.24810183e-01 -6.48597777e-01 -9.90409017e-01 5.27494371e-01
5.02447337e-02 -1.15795708e+00 1.61942855e-01 4.90337968e-01
1.06034243e+00 -7.00307071e-01 1.67434454e-01 -1.18901841e-01
3.13612431e-01 -9.95216250e-01 7.44130433e-01 2.13704959e-01
5.66072822e-01 -1.50494885e+00 6.29015446e-01 5.13768256e-01
-5.21630764e-01 -4.23761666e-01 -4.42939669e-01 -2.29990989e-01
2.58934110e-01 8.94502282e-01 -5.00959992e-01 5.54310262e-01
5.23154616e-01 1.28884628e-01 8.64381418e-02 1.31019330e+00
4.29941177e-01 2.51333922e-01 -3.60992551e-01 -1.78543717e-01
9.77348015e-02 -3.90637904e-01 4.36714798e-01 5.18281400e-01
2.68597871e-01 2.44434297e-01 1.97920203e-01 7.79698193e-01
-4.80240881e-01 7.19380558e-01 -6.27150178e-01 2.07498018e-02
7.01486290e-01 9.24408257e-01 -4.34995174e-01 -1.14229992e-01
-9.09469277e-02 4.25173044e-01 7.97902644e-02 4.20982055e-02
-5.57518840e-01 -3.15216243e-01 7.56645381e-01 3.81861120e-01
-1.33118674e-01 2.83198863e-01 -1.03356266e+00 -7.34022558e-01
1.69489607e-01 -7.16421187e-01 1.00947344e+00 2.11059060e-02
-1.15968657e+00 -1.51686385e-01 3.07477564e-01 -8.14229012e-01
-1.98331997e-02 1.07172586e-01 -7.80776680e-01 9.50651288e-01
-1.19751740e+00 -5.35205126e-01 -8.99702683e-03 3.04474503e-01
7.79965371e-02 -1.30546078e-01 2.55792707e-01 8.36541951e-02
-7.18251526e-01 8.13596427e-01 3.32308322e-01 -4.60479230e-01
4.89947349e-01 -1.10395181e+00 -3.96695167e-01 8.71017396e-01
-4.52906609e-01 5.73362529e-01 8.51477981e-01 -7.62657464e-01
-1.13453448e+00 -1.02127194e+00 1.25288928e+00 7.95756876e-02
-1.39988780e-01 -2.09517851e-01 -6.70953631e-01 4.14694518e-01
-1.17245233e-02 -1.76942796e-01 8.79553080e-01 1.47370622e-01
1.53481767e-01 -4.82021153e-01 -1.89066947e+00 5.78282177e-01
8.13066840e-01 3.89462471e-01 -1.03660122e-01 1.28460005e-01
2.20493793e-01 -2.12886125e-01 -7.03538775e-01 3.66561532e-01
3.71382058e-01 -7.09375262e-01 5.76890826e-01 -8.49152803e-01
2.91159481e-01 -4.22993600e-02 -1.92670926e-01 -1.35627389e+00
-2.47747764e-01 -7.09240854e-01 3.09691448e-02 1.41706467e+00
2.57451564e-01 -6.57291532e-01 1.04377079e+00 1.27849019e+00
3.82519811e-01 -7.70780921e-01 -1.12115324e+00 -5.89986622e-01
1.47405237e-01 3.58812660e-01 8.10317457e-01 8.14847171e-01
1.96794286e-01 -4.90838569e-03 -6.50781155e-01 -1.04407743e-01
1.00593829e+00 4.56261337e-01 7.71903813e-01 -1.19634604e+00
4.34032887e-01 -5.14978051e-01 -8.40084180e-02 -2.55515903e-01
1.46388993e-01 -7.93397248e-01 -1.03673801e-01 -1.60730016e+00
6.66391551e-01 -5.08319795e-01 -1.44913256e-01 2.79161900e-01
-2.49349266e-01 -1.91172749e-01 4.41025108e-01 -1.07913971e-01
-4.85244334e-01 6.06069088e-01 9.32108104e-01 -2.05787361e-01
-4.73619193e-01 3.25737596e-01 -1.63166094e+00 2.31680453e-01
7.90138125e-01 -5.84184945e-01 -5.40808797e-01 -3.05879079e-02
1.39018193e-01 2.07934454e-01 -2.01107204e-01 -1.31036744e-01
1.15312256e-01 -9.86589313e-01 2.01860204e-01 -4.82762069e-01
-1.69968128e-01 -9.15891111e-01 4.19342756e-01 5.98235905e-01
-6.33071363e-01 -1.92393586e-02 -9.88095775e-02 3.87254596e-01
-6.17684722e-02 -3.30545664e-01 8.81382346e-01 5.33577353e-02
-1.10441335e-01 3.60025644e-01 -1.98785320e-01 2.66201288e-01
1.70223558e+00 -3.66357416e-01 -3.75895053e-01 -6.29794478e-01
-3.96423005e-02 1.04389346e+00 6.42277181e-01 3.09442431e-01
4.02159691e-01 -1.29708219e+00 -1.05112052e+00 -1.70257986e-01
1.97580308e-01 1.84996113e-01 2.74793386e-01 5.22550941e-01
-3.15293103e-01 3.51810306e-01 -3.81256819e-01 -4.92518842e-02
-1.42171896e+00 4.01720405e-01 2.60933369e-01 -1.28724977e-01
-2.88068373e-02 6.31493330e-01 2.63916217e-02 -3.79172824e-02
3.14085245e-01 1.26988590e-01 -2.50518262e-01 6.57542765e-01
5.79063237e-01 8.55489790e-01 -7.09342882e-02 -5.06745398e-01
-5.06826758e-01 6.04286864e-02 -1.74312532e-01 5.29442765e-02
1.48040235e+00 -2.82082319e-01 -4.46304202e-01 -3.49385329e-02
9.50556755e-01 2.59394139e-01 -8.95365477e-01 -3.45581360e-02
-5.45571884e-03 -1.03233528e+00 -8.17341655e-02 -5.34116626e-01
-1.06812418e+00 -2.23814715e-02 1.13251261e-01 2.64156878e-01
1.38846493e+00 -1.90154165e-01 4.00147349e-01 -1.48332655e-01
3.36642772e-01 -1.36761475e+00 -3.38390946e-01 -1.91543221e-01
8.87982488e-01 -1.19935024e+00 5.22412360e-01 -2.28554904e-01
-8.98996055e-01 6.45732403e-01 7.32992232e-01 -2.63740122e-02
2.19114125e-01 -5.13122277e-03 -3.85114551e-01 2.30643097e-02
-7.74575412e-01 7.24149868e-02 3.56105447e-01 3.81276339e-01
1.82193801e-01 4.52973366e-01 -1.34622395e+00 4.71233279e-01
-1.48753330e-01 -8.09174776e-02 1.05615532e+00 8.48873615e-01
-6.95790112e-01 -1.00194550e+00 -7.02483594e-01 1.08344066e+00
-6.56456113e-01 4.10202295e-01 -4.74076092e-01 4.29221362e-01
1.92368805e-01 9.74018097e-01 3.18462014e-01 4.00921553e-01
5.11184037e-01 -3.44912410e-01 3.21376890e-01 -5.86214960e-01
-5.34633398e-01 -2.23783717e-01 8.07089359e-02 -1.93867862e-01
-3.00453693e-01 -1.08703804e+00 -1.04618347e+00 -6.78652465e-01
-3.42920244e-01 3.74843776e-01 4.57373828e-01 4.77296293e-01
-2.41312589e-02 1.44403964e-01 9.41402793e-01 -2.43939012e-01
-9.14985001e-01 -2.93270767e-01 -1.01272631e+00 4.67278063e-01
2.72899568e-01 -3.02723825e-01 -1.56487495e-01 -5.13908625e-01]
|
[6.611871242523193, 4.9490132331848145]
|
45d4b083-812e-40bd-86ad-e2ac51d15a31
|
example-based-explanations-with-adversarial
|
2203.16141
| null |
https://arxiv.org/abs/2203.16141v1
|
https://arxiv.org/pdf/2203.16141v1.pdf
|
Example-based Explanations with Adversarial Attacks for Respiratory Sound Analysis
|
Respiratory sound classification is an important tool for remote screening of respiratory-related diseases such as pneumonia, asthma, and COVID-19. To facilitate the interpretability of classification results, especially ones based on deep learning, many explanation methods have been proposed using prototypes. However, existing explanation techniques often assume that the data is non-biased and the prediction results can be explained by a set of prototypical examples. In this work, we develop a unified example-based explanation method for selecting both representative data (prototypes) and outliers (criticisms). In particular, we propose a novel application of adversarial attacks to generate an explanation spectrum of data instances via an iterative fast gradient sign method. Such unified explanation can avoid over-generalisation and bias by allowing human experts to assess the model mistakes case by case. We performed a wide range of quantitative and qualitative evaluations to show that our approach generates effective and understandable explanation and is robust with many deep learning models
|
['Björn W. Schuller', 'Wolfgang Nejdl', 'Thanh Tam Nguyen', 'Zhao Ren', 'Yi Chang']
|
2022-03-30
| null | null | null | null |
['sound-classification']
|
['audio']
|
[ 1.47288442e-01 4.93195683e-01 7.74857309e-03 -5.96354008e-01
-6.63591623e-01 -3.39829952e-01 4.44211423e-01 2.29916692e-01
1.60775304e-01 6.74515188e-01 1.49203032e-01 -5.62103808e-01
-4.35080171e-01 -5.93403280e-01 -8.36270332e-01 -5.36510944e-01
1.46586552e-01 6.24338388e-01 -1.67577788e-01 3.60674001e-02
2.71916300e-01 6.23220980e-01 -1.56911707e+00 5.94722092e-01
1.22531652e+00 6.00281537e-01 -4.26152170e-01 7.11955309e-01
-1.89251620e-02 4.93101209e-01 -8.87597859e-01 -6.03843272e-01
8.50877240e-02 -6.00826979e-01 -7.57509291e-01 -4.83245365e-02
3.33165288e-01 -3.76551658e-01 3.17991018e-01 9.37286556e-01
7.14500129e-01 -4.47135977e-02 9.68488157e-01 -1.55322897e+00
-7.63861775e-01 6.11494124e-01 -9.20472760e-03 -1.33807495e-01
4.07696038e-01 2.37762406e-01 9.42301631e-01 -9.00898337e-01
4.05957818e-01 1.20134628e+00 1.02686489e+00 9.96371388e-01
-1.16014397e+00 -7.06517339e-01 -3.14263720e-03 4.12901402e-01
-9.10711110e-01 8.02304149e-02 8.08156371e-01 -3.11272949e-01
6.10483825e-01 7.58969784e-01 7.03965366e-01 1.48361886e+00
-7.28045329e-02 5.31641126e-01 8.68087947e-01 -3.65077674e-01
4.65013057e-01 5.09688377e-01 1.60428673e-01 4.14242864e-01
6.62760496e-01 2.78892785e-01 -1.87466502e-01 -4.78857964e-01
3.10414582e-01 3.41088057e-01 -4.67636257e-01 -3.51931483e-01
-9.27645206e-01 1.06371021e+00 6.09790027e-01 1.13308929e-01
-4.33911264e-01 1.68025792e-01 3.04292679e-01 1.30694389e-01
2.50967503e-01 7.77597547e-01 -6.20967150e-01 2.67671853e-01
-7.18319595e-01 4.30218488e-01 8.41973603e-01 4.64213192e-01
5.41129947e-01 1.64932370e-01 -1.54060870e-01 6.32554114e-01
4.51832443e-01 5.40615559e-01 7.27661788e-01 -8.70315313e-01
9.52227786e-02 7.58218229e-01 1.76723465e-01 -9.55166996e-01
-4.30239111e-01 -4.83714879e-01 -8.94268036e-01 3.53241771e-01
3.27583522e-01 -9.57406387e-02 -9.62155342e-01 1.55646276e+00
3.83224249e-01 1.66111410e-01 9.59745422e-02 9.33348835e-01
1.07808757e+00 1.56813294e-01 -2.49639247e-02 -6.01377152e-02
9.54239190e-01 -7.64603078e-01 -5.79120517e-01 -2.45526768e-02
6.63168728e-01 -4.60742325e-01 1.49961889e+00 6.39148831e-01
-6.59989357e-01 -5.85764229e-01 -8.52792680e-01 2.76688546e-01
-3.45140517e-01 -1.43278196e-01 3.65774453e-01 7.23710477e-01
-6.25202894e-01 9.93073046e-01 -6.21708512e-01 -2.92955160e-01
4.54362869e-01 4.61840957e-01 -3.52811933e-01 1.20868511e-01
-1.19709063e+00 8.21600378e-01 2.67526865e-01 5.50488196e-02
-7.60850847e-01 -6.94727600e-01 -4.97777194e-01 1.37520850e-01
-1.38098866e-01 -9.27110732e-01 1.21489608e+00 -1.08944917e+00
-1.10908246e+00 5.81908882e-01 4.48289551e-02 -5.91435790e-01
8.26569736e-01 -5.74156582e-01 -4.78331506e-01 7.35624954e-02
-1.70941249e-01 5.22003710e-01 1.02893686e+00 -1.64123011e+00
-2.58550763e-01 -3.85941863e-02 -2.04734311e-01 -2.56192923e-01
-2.09060475e-01 -2.04257250e-01 3.40440571e-01 -7.61127353e-01
1.69430867e-01 -9.29995537e-01 -3.33625436e-01 5.49948812e-02
-8.67853582e-01 1.06153116e-02 6.72131300e-01 -5.26346684e-01
1.14140797e+00 -2.06456089e+00 -2.76659787e-01 4.43691880e-01
3.89365703e-01 4.28260088e-01 3.26846428e-02 3.31711620e-01
-4.16839749e-01 7.63036728e-01 -4.81438816e-01 -1.74233153e-01
1.39242694e-01 3.86481225e-01 -7.02701092e-01 2.18798131e-01
3.66486967e-01 6.48886859e-01 -9.27993238e-01 -3.38037193e-01
3.78652662e-01 7.55320013e-01 -8.73970628e-01 4.66270298e-01
-2.06577957e-01 6.06122553e-01 -2.84028798e-01 3.92094284e-01
4.03365016e-01 -2.40780488e-01 -1.25606835e-01 1.45422737e-03
4.80774790e-01 3.68071914e-01 -1.16975224e+00 9.34477746e-01
-5.06890178e-01 6.25345349e-01 -6.83348119e-01 -8.71557117e-01
9.50788558e-01 5.37663877e-01 3.72376405e-02 -1.37968641e-02
2.66054161e-02 6.23183370e-01 2.41779983e-01 -7.99187303e-01
-1.42369298e-02 -3.10994089e-01 3.93863291e-01 6.13614500e-01
-3.53808045e-01 -2.78891772e-01 -3.01088274e-01 -2.36976147e-01
9.77453351e-01 -2.17486918e-01 5.18549621e-01 1.61230624e-01
4.07981664e-01 2.95934826e-02 4.23439860e-01 1.04217267e+00
1.29744178e-03 1.25188684e+00 4.68621045e-01 -8.72865379e-01
-9.28986251e-01 -1.07197726e+00 -2.24375993e-01 4.03583735e-01
-2.56238341e-01 -7.47512793e-03 -7.30982184e-01 -1.17729127e+00
1.22703075e-01 1.31674099e+00 -8.56548786e-01 -5.06816566e-01
-2.83162028e-01 -6.38101578e-01 4.46729720e-01 6.94970250e-01
1.95442483e-01 -1.49768603e+00 -7.61000156e-01 2.20821165e-02
-9.28219408e-02 -5.10713100e-01 3.65857519e-02 2.77551353e-01
-1.00113297e+00 -1.34128833e+00 -3.47353220e-01 -2.07746983e-01
9.44070697e-01 -8.00116211e-02 1.29400635e+00 6.95499182e-01
-2.36081421e-01 2.02535763e-01 -4.60872024e-01 -7.39687324e-01
-9.71562028e-01 -1.08042419e-01 1.45832971e-01 -1.61868930e-01
3.97419661e-01 -4.10067827e-01 -8.04669142e-01 3.25294346e-01
-1.03707051e+00 -1.90843418e-01 6.16661727e-01 1.02267396e+00
6.33577347e-01 -1.22700736e-01 7.47907698e-01 -1.21452880e+00
8.41545761e-01 -5.47658801e-01 2.94918418e-02 2.49405742e-01
-8.95499170e-01 1.48435131e-01 9.99625564e-01 -6.87157750e-01
-7.55272985e-01 -1.00539133e-01 -2.85488576e-01 -6.08152270e-01
-7.11517096e-01 1.76357627e-01 -1.09305128e-01 1.43701702e-01
1.13997602e+00 -6.79371282e-02 -1.95407972e-01 -4.95299727e-01
4.09073025e-01 7.36082137e-01 3.59685630e-01 -1.35986730e-01
9.06283498e-01 3.91183794e-01 -1.39244884e-01 -4.06974643e-01
-8.60564530e-01 -1.01404876e-01 -4.22405422e-01 -1.85313687e-01
7.30036616e-01 -3.10283989e-01 -4.15370971e-01 -4.54100408e-03
-1.35960078e+00 -1.09529778e-01 -6.98729217e-01 6.40216947e-01
-3.91351134e-01 2.08162218e-01 -5.28783537e-02 -7.47873783e-01
-4.73734736e-01 -1.09842312e+00 1.02425635e+00 4.62486632e-02
-1.01263368e+00 -1.14457607e+00 2.05291912e-01 2.88954526e-01
3.09948027e-01 5.01582265e-01 1.12941325e+00 -1.76101363e+00
-2.34429479e-01 -5.77675164e-01 5.86208962e-02 5.24440825e-01
1.64392069e-01 4.29459721e-01 -1.28504968e+00 -4.22564000e-02
1.28628284e-01 -2.27393433e-01 6.43130481e-01 3.29257250e-01
1.60982823e+00 -7.14212418e-01 -1.18550882e-01 4.99122977e-01
9.71312702e-01 -5.43948263e-02 5.54759622e-01 2.61415184e-01
6.15782797e-01 6.67012751e-01 4.36531901e-01 3.00649345e-01
-1.88693717e-01 3.84893715e-01 8.57688546e-01 -4.25379634e-01
9.76288617e-02 -2.19554245e-01 1.27921104e-01 3.78342181e-01
4.28637080e-02 -3.69742841e-01 -9.34104979e-01 5.15747607e-01
-1.73323655e+00 -1.07709575e+00 -4.59955156e-01 2.30838060e+00
4.58509147e-01 2.35673919e-01 -7.36815110e-02 6.42682910e-01
7.55462825e-01 -3.08126539e-01 -6.27656043e-01 -8.09403837e-01
2.29151279e-01 2.13777691e-01 1.67446192e-02 3.34762901e-01
-7.52498984e-01 2.13003442e-01 7.01895905e+00 5.36884904e-01
-1.11225247e+00 -1.18480407e-01 5.31293035e-01 7.56024709e-03
-8.85793746e-01 -2.53934741e-01 -3.33359331e-01 3.16823781e-01
7.56099343e-01 6.12027273e-02 1.08121350e-01 1.09868741e+00
2.93591350e-01 5.27684093e-01 -1.29702270e+00 6.09851956e-01
-9.88831818e-02 -1.26421893e+00 4.14476156e-01 -1.36154950e-01
7.74866462e-01 -3.50592434e-01 2.02591985e-01 5.51060997e-02
1.35052532e-01 -1.28350043e+00 4.50283378e-01 4.49564219e-01
4.05275643e-01 -5.88824570e-01 1.01947451e+00 2.55281210e-01
-5.23881197e-01 -2.29095384e-01 -2.15663299e-01 4.08509485e-02
-2.94310451e-01 6.87800050e-01 -1.32595599e+00 5.02268553e-01
6.22521460e-01 3.09062093e-01 -5.98671019e-01 1.08651447e+00
-7.49067843e-01 9.57312226e-01 -2.40238816e-01 -9.23333541e-02
-5.13771772e-02 3.32840234e-01 6.39945567e-01 1.07110751e+00
5.29728770e-01 -2.03264818e-01 -3.93749893e-01 1.11791420e+00
1.02586441e-01 1.65385172e-01 -8.71152878e-01 2.75246114e-01
5.28507292e-01 9.94522631e-01 -5.89551866e-01 -3.32655489e-01
-1.05391942e-01 6.22078776e-01 -1.14162266e-01 2.98314303e-01
-1.00662923e+00 -2.38303512e-01 3.91470641e-01 2.55286247e-01
1.06441736e-01 6.21540427e-01 -6.45331919e-01 -8.28385890e-01
-8.40299949e-02 -1.27753472e+00 4.20405388e-01 -9.70270336e-01
-1.34974515e+00 7.31396437e-01 3.64045389e-02 -1.50987124e+00
-5.94738841e-01 -5.14982343e-01 -1.17389286e+00 5.43224156e-01
-1.18647647e+00 -8.32268894e-01 -5.75838208e-01 3.76831055e-01
4.17455167e-01 -1.57182530e-01 9.46873724e-01 7.51555189e-02
-5.10797977e-01 6.12819076e-01 -2.93849297e-02 -1.80392593e-01
6.50828838e-01 -1.57440078e+00 3.00122082e-01 6.23581350e-01
3.89514953e-01 7.65041053e-01 1.31233931e+00 -5.36317647e-01
-4.94512171e-01 -1.22192001e+00 9.05037105e-01 -6.26978278e-01
2.41354272e-01 1.38632655e-01 -1.49824846e+00 4.70128834e-01
-1.01225406e-01 8.47191177e-03 8.89701545e-01 8.80978853e-02
-2.68999070e-01 -3.69840153e-02 -1.37166142e+00 7.41188645e-01
6.35949731e-01 -2.39457861e-01 -8.28760028e-01 5.13932049e-01
6.63947999e-01 -2.93534279e-01 -5.97053409e-01 7.15357542e-01
5.61851323e-01 -1.37592053e+00 9.44203436e-01 -9.74461138e-01
4.90427285e-01 -2.13337794e-01 7.94798210e-02 -1.36258709e+00
7.67859295e-02 -4.72955734e-01 -1.38963729e-01 1.02239108e+00
5.46487451e-01 -6.56981289e-01 7.21412003e-01 7.43274450e-01
-1.56864733e-01 -9.07231152e-01 -6.22010529e-01 -6.50774777e-01
-7.56914727e-04 -7.18892336e-01 9.48423684e-01 9.55934763e-01
-3.16059023e-01 -5.54873124e-02 -3.07335585e-01 4.34474319e-01
4.47957516e-01 1.18371896e-01 9.83782649e-01 -1.38799047e+00
-3.67541701e-01 -3.76005292e-01 -2.58826315e-01 -2.93088198e-01
2.18431838e-02 -6.10486448e-01 7.80103281e-02 -1.62627304e+00
1.26363123e-02 -1.17197864e-01 -3.13051879e-01 5.03248751e-01
-4.71408069e-01 7.78417140e-02 -1.12956814e-01 2.05329597e-01
-2.59906828e-01 4.03475702e-01 1.06230009e+00 1.12416029e-01
-1.26169890e-01 4.82713014e-01 -6.92983389e-01 1.15463603e+00
1.02215528e+00 -1.06176353e+00 -4.88412917e-01 -9.57663357e-02
3.38616431e-01 -3.04295331e-01 8.65700901e-01 -9.21226084e-01
-3.69282693e-01 -9.56559554e-02 3.45563829e-01 -4.06534970e-01
-4.49127220e-02 -9.70228732e-01 5.85842371e-01 1.01033151e+00
-7.33202338e-01 -6.46091029e-02 7.12102354e-02 6.83706701e-01
-1.35943443e-01 -4.71978784e-01 6.87884152e-01 -4.44991561e-03
-8.96277726e-02 2.06372347e-02 -1.85690373e-01 1.43728226e-01
7.50782549e-01 -3.36592555e-01 -3.02142233e-01 -6.01095557e-01
-8.50674272e-01 -5.22800200e-02 4.35736924e-01 2.58684784e-01
8.79715741e-01 -1.30705452e+00 -7.13995099e-01 2.99573481e-01
9.47385430e-02 7.73236230e-02 2.81886961e-02 6.76578104e-01
-5.98165154e-01 2.15459958e-01 -9.03480407e-03 -6.43715203e-01
-1.19813907e+00 5.12073636e-01 6.15474045e-01 -1.63138792e-01
-5.94346702e-01 7.15521753e-01 3.09127659e-01 -7.16825247e-01
3.88274431e-01 -5.35625398e-01 -2.25231737e-01 -2.68809021e-01
2.27211773e-01 5.52236974e-01 5.11819497e-02 -9.79585871e-02
-4.31515545e-01 3.94282043e-01 2.20202252e-01 1.82925582e-01
1.18856978e+00 2.94163793e-01 2.85403997e-01 4.63881940e-01
8.96297455e-01 1.23100122e-02 -1.06964147e+00 3.89193445e-01
-4.21944670e-02 -4.37437236e-01 -4.98539656e-01 -9.84477162e-01
-9.59330738e-01 1.38674212e+00 6.61151886e-01 6.75888836e-01
1.01935530e+00 -2.29072213e-01 5.09738743e-01 5.18029511e-01
-3.17610919e-01 -6.71335697e-01 2.92677581e-01 -1.75979197e-01
1.38428473e+00 -1.38918769e+00 2.25942675e-02 -6.35343939e-02
-7.62368619e-01 1.22326064e+00 5.29880881e-01 -5.12846895e-02
4.37349617e-01 -4.03244823e-01 4.07762200e-01 -3.27448994e-01
-7.78127432e-01 2.48008534e-01 7.13820696e-01 7.67180264e-01
3.64295691e-01 -8.66621267e-03 -2.45695084e-01 8.25579643e-01
-2.77318478e-01 -2.16747940e-01 4.20539409e-01 2.42616311e-01
-4.17138755e-01 -9.24873710e-01 -5.75453639e-01 5.82803309e-01
-5.88489413e-01 3.87747044e-04 -8.19512308e-01 1.16966093e+00
2.92426437e-01 8.69405925e-01 -2.48585880e-01 -5.04055619e-01
4.97369051e-01 1.73673302e-01 -9.71757174e-02 -7.47120619e-01
-8.42740059e-01 -2.69095212e-01 -6.27029268e-03 -3.48307818e-01
-2.89330363e-01 -4.70471799e-01 -1.41390955e+00 -1.95070371e-01
-6.25210047e-01 2.03548208e-01 5.19638777e-01 1.08836973e+00
3.05576116e-01 6.85160577e-01 6.20425344e-01 -5.61170697e-01
-9.22537625e-01 -7.69693613e-01 -2.07087338e-01 9.43666458e-01
4.80545908e-01 -5.91089129e-01 -1.05185914e+00 -4.99402955e-02]
|
[8.755579948425293, 5.6591081619262695]
|
d4eebd3b-a0b0-4d0d-a799-92116db95eeb
|
boxe-a-box-embedding-model-for-knowledge-base
|
2007.06267
| null |
https://arxiv.org/abs/2007.06267v2
|
https://arxiv.org/pdf/2007.06267v2.pdf
|
BoxE: A Box Embedding Model for Knowledge Base Completion
|
Knowledge base completion (KBC) aims to automatically infer missing facts by exploiting information already present in a knowledge base (KB). A promising approach for KBC is to embed knowledge into latent spaces and make predictions from learned embeddings. However, existing embedding models are subject to at least one of the following limitations: (1) theoretical inexpressivity, (2) lack of support for prominent inference patterns (e.g., hierarchies), (3) lack of support for KBC over higher-arity relations, and (4) lack of support for incorporating logical rules. Here, we propose a spatio-translational embedding model, called BoxE, that simultaneously addresses all these limitations. BoxE embeds entities as points, and relations as a set of hyper-rectangles (or boxes), which spatially characterize basic logical properties. This seemingly simple abstraction yields a fully expressive model offering a natural encoding for many desired logical properties. BoxE can both capture and inject rules from rich classes of rule languages, going well beyond individual inference patterns. By design, BoxE naturally applies to higher-arity KBs. We conduct a detailed experimental analysis, and show that BoxE achieves state-of-the-art performance, both on benchmark knowledge graphs and on more general KBs, and we empirically show the power of integrating logical rules.
|
['Tommaso Salvatori', 'İsmail İlkan Ceylan', 'Ralph Abboud', 'Thomas Lukasiewicz']
|
2020-07-13
| null |
http://proceedings.neurips.cc/paper/2020/hash/6dbbe6abe5f14af882ff977fc3f35501-Abstract.html
|
http://proceedings.neurips.cc/paper/2020/file/6dbbe6abe5f14af882ff977fc3f35501-Paper.pdf
|
neurips-2020-12
|
['knowledge-base-completion', 'knowledge-base-completion']
|
['graphs', 'knowledge-base']
|
[-2.25515038e-01 3.70268822e-01 -6.59679234e-01 -1.45083994e-01
-2.48404473e-01 -6.64761722e-01 7.23154247e-01 3.70005190e-01
-8.20902586e-02 8.67011964e-01 5.52964568e-01 -4.98532236e-01
-5.60511768e-01 -1.38387215e+00 -9.93011117e-01 -3.36655915e-01
-4.92810816e-01 4.51617777e-01 3.71758372e-01 -2.69523770e-01
-2.58823246e-01 5.86421788e-01 -1.40670359e+00 3.75617832e-01
7.71602035e-01 8.33338559e-01 -1.83699593e-01 2.43462875e-01
-2.00293541e-01 1.25078297e+00 -1.84264243e-01 -7.51876891e-01
-9.74280983e-02 1.41731337e-01 -9.14024889e-01 -4.58687544e-01
3.37755322e-01 -3.88473630e-01 -7.14498281e-01 7.75073767e-01
-5.25259003e-02 1.10544778e-01 8.38413358e-01 -1.39060402e+00
-1.16929173e+00 6.42626643e-01 -1.03732415e-01 -5.00951521e-03
5.50607383e-01 -9.93714035e-02 1.58563864e+00 -1.16282272e+00
8.67053151e-01 1.18571758e+00 7.91327775e-01 2.88570791e-01
-1.57428777e+00 -3.29559237e-01 2.19510928e-01 4.70932782e-01
-1.72979152e+00 -3.36895525e-01 7.25006759e-01 -3.50267202e-01
1.31470609e+00 2.81389236e-01 7.28784442e-01 9.99177575e-01
1.06583320e-01 7.94993043e-01 8.20385695e-01 -4.70274627e-01
5.05773902e-01 2.76808918e-01 3.07266176e-01 1.02791607e+00
6.81933343e-01 -1.35440096e-01 -7.55419612e-01 -4.66524005e-01
8.55471373e-01 -1.41849462e-02 -3.93794149e-01 -6.85135305e-01
-1.38975501e+00 8.81813526e-01 5.73241889e-01 1.44666895e-01
-2.46620625e-01 3.01914185e-01 2.45400712e-01 1.42869934e-01
2.05527514e-01 4.30138171e-01 -4.74387467e-01 -2.65220962e-02
-6.16781116e-01 4.28157210e-01 1.02982843e+00 1.27076876e+00
1.00791323e+00 -1.31910682e-01 -1.94185413e-02 5.73678434e-01
2.57589489e-01 4.69465971e-01 1.62795316e-02 -7.83172131e-01
6.62441194e-01 1.04733133e+00 1.74080923e-01 -1.41366148e+00
-2.84135491e-01 -5.30634895e-02 -7.91267395e-01 -9.32752863e-02
4.98868451e-02 2.81029671e-01 -7.12919652e-01 1.84261847e+00
1.40577853e-01 1.35728255e-01 2.97608376e-01 5.51801264e-01
8.79150867e-01 6.54524922e-01 -1.17735550e-01 2.64445059e-02
1.52238095e+00 -5.81007481e-01 -7.71837354e-01 -4.57548462e-02
7.33259916e-01 5.50591089e-02 1.20878804e+00 -2.49841977e-02
-9.30021644e-01 -2.87952662e-01 -1.10855663e+00 -4.24956560e-01
-1.05557466e+00 -3.43981609e-02 1.08039570e+00 4.65926379e-01
-9.88664567e-01 1.45939305e-01 -8.42906654e-01 -2.28416979e-01
2.85023659e-01 1.61157116e-01 -7.20600188e-01 -2.15581462e-01
-1.57265854e+00 9.26685214e-01 7.00421214e-01 8.85747373e-02
-4.36127156e-01 -1.01258981e+00 -1.36319149e+00 2.93714911e-01
9.00076687e-01 -8.38512838e-01 4.67654854e-01 1.16194129e-01
-1.08183038e+00 4.35701013e-01 -2.76650280e-01 -4.90038991e-01
9.44132656e-02 -3.25405121e-01 -7.20350981e-01 8.44266564e-02
1.75438300e-01 6.17270947e-01 2.96953052e-01 -1.33965373e+00
-3.45327735e-01 -3.40031028e-01 6.87304616e-01 -5.79251572e-02
-5.21248877e-01 -4.16460484e-01 -6.47868276e-01 -5.63944817e-01
4.52428125e-02 -7.39538670e-01 3.17593008e-01 3.88851739e-03
-5.95648646e-01 -4.57683146e-01 7.60795474e-01 -4.98022676e-01
1.66159499e+00 -2.13709426e+00 3.21728617e-01 3.46792847e-01
3.98243427e-01 4.70112301e-02 2.58579433e-01 7.01786041e-01
-1.17837824e-02 4.17217523e-01 -2.01043636e-01 -6.51671067e-02
5.06164789e-01 8.07172179e-01 -9.01638448e-01 1.44132778e-01
4.53040749e-01 1.47577536e+00 -9.01773572e-01 -6.98130846e-01
1.59548134e-01 3.30700010e-01 -6.47015452e-01 -3.11690629e-01
-5.88202417e-01 -4.37491387e-01 -3.86973143e-01 7.31715262e-01
3.82109463e-01 -6.20141089e-01 3.96037132e-01 -5.04341602e-01
4.79504354e-02 4.18776780e-01 -1.46031427e+00 1.51584780e+00
-3.51186335e-01 3.67182314e-01 -4.11666930e-01 -6.95543110e-01
6.22362852e-01 3.01787078e-01 1.88144192e-01 -1.60705924e-01
-6.31329417e-01 6.82527572e-02 -2.67372847e-01 -4.34842736e-01
5.29557407e-01 -1.52860641e-01 -2.63259619e-01 3.63209456e-01
1.85316280e-01 9.34384614e-02 3.26990008e-01 5.83046317e-01
1.32789636e+00 1.27641067e-01 5.21599352e-01 -1.84497505e-01
3.18832517e-01 3.20049301e-02 7.94112742e-01 6.32647634e-01
1.48765340e-01 -5.87037243e-02 7.26410449e-01 -6.12590134e-01
-7.46314585e-01 -1.41681123e+00 -2.59147465e-01 6.56149089e-01
1.91578239e-01 -9.86276686e-01 1.24479488e-01 -7.36938715e-01
5.27038693e-01 9.20717180e-01 -8.11901569e-01 -1.54307514e-01
-3.10498059e-01 -2.83101648e-01 8.46962988e-01 1.04585171e+00
4.88792777e-01 -8.09197783e-01 -4.02535766e-01 1.47613153e-01
-1.35172471e-01 -1.42981708e+00 1.20330686e-02 7.72853643e-02
-6.14421785e-01 -1.10797250e+00 -1.23782493e-02 -5.95229626e-01
4.90994722e-01 -7.49182254e-02 1.04681194e+00 -1.48211822e-01
-1.57282338e-01 6.21185184e-01 -2.96955526e-01 -1.66121528e-01
7.68470466e-02 -1.23147383e-01 3.58328402e-01 -1.35018483e-01
3.85755152e-01 -7.65821278e-01 -2.62094706e-01 4.96530421e-02
-1.06718123e+00 2.44168475e-01 5.52515447e-01 8.36727202e-01
6.88676238e-01 5.76524675e-01 4.13418204e-01 -9.29847479e-01
4.70609546e-01 -3.32599819e-01 -4.49713230e-01 7.78021872e-01
-6.08067036e-01 4.07045245e-01 6.67572379e-01 -5.87969497e-02
-1.05240738e+00 -4.03732985e-01 3.36356014e-01 -4.31129664e-01
-7.06696883e-02 1.05113041e+00 -3.50545913e-01 2.51885891e-01
5.09105802e-01 2.98176348e-01 -3.62497330e-01 -2.85846621e-01
8.80260646e-01 2.42012024e-01 5.26608467e-01 -1.10017514e+00
8.61688256e-01 6.94085360e-01 2.40234211e-01 -7.95265377e-01
-8.53907645e-01 -1.79586634e-01 -7.25930154e-01 2.44561732e-01
7.28317797e-01 -9.45226669e-01 -7.00192928e-01 -2.28658125e-01
-1.11326933e+00 -2.54740238e-01 -4.43989903e-01 4.31429416e-01
-5.57139635e-01 3.08263153e-01 -7.48105884e-01 -6.40847743e-01
1.83349233e-02 -6.64753616e-01 9.06793892e-01 -2.22914040e-01
-3.92053008e-01 -1.36101770e+00 7.37056732e-02 1.83364838e-01
4.57078852e-02 3.44036341e-01 1.72090900e+00 -4.04117227e-01
-8.05744469e-01 -1.68365642e-01 -2.63816744e-01 9.55664739e-02
3.63806099e-01 8.04935172e-02 -7.15795696e-01 -6.58813352e-03
-5.84178984e-01 -3.91568512e-01 8.35463583e-01 -2.17680737e-01
9.96876061e-01 -6.00736618e-01 -6.21507704e-01 4.68194246e-01
1.49752557e+00 -2.61584640e-01 7.09119380e-01 2.49424025e-01
6.87940001e-01 3.84796500e-01 2.75208950e-01 2.86010683e-01
7.71716475e-01 7.74254084e-01 1.61989719e-01 2.77090371e-01
-5.94014116e-02 -6.45387232e-01 2.67769635e-01 7.69952476e-01
-3.35628331e-01 3.11303400e-02 -1.16548610e+00 7.20144033e-01
-1.99968970e+00 -1.15560198e+00 6.42864257e-02 1.88612854e+00
1.22858369e+00 2.23898754e-01 -1.05875887e-01 1.33389756e-01
2.40265563e-01 2.05303252e-01 -4.09927130e-01 -1.14958532e-01
-3.23409647e-01 1.61260366e-01 3.43758374e-01 5.16572714e-01
-9.47348177e-01 1.01896274e+00 6.32340050e+00 6.56500876e-01
-7.34176934e-01 -3.92198078e-02 -4.25387174e-02 -8.38827118e-02
-8.03540528e-01 2.85759240e-01 -8.68379772e-01 3.05473208e-01
4.71842885e-01 -1.28904253e-01 4.75391686e-01 5.60624480e-01
-4.33652997e-01 1.11989260e-01 -1.41191006e+00 7.58270621e-01
-9.05862898e-02 -1.83486545e+00 4.31970626e-01 2.12540358e-01
7.90559113e-01 -2.81327128e-01 -1.28485277e-01 6.30828798e-01
4.90094960e-01 -1.01311958e+00 6.84544384e-01 6.49750292e-01
1.04154515e+00 -6.12683833e-01 5.42169511e-01 2.09556013e-01
-1.25945461e+00 -6.84941113e-02 -4.38731909e-01 -1.55823603e-01
-1.43422484e-02 7.33043432e-01 -5.78740060e-01 1.01419032e+00
6.00336373e-01 7.95318961e-01 -5.07678449e-01 3.32796723e-01
-6.37694895e-01 5.49332857e-01 -5.09752393e-01 2.67267544e-02
9.74577889e-02 -2.41237767e-02 2.41106465e-01 1.31137252e+00
-5.45088723e-02 1.61278695e-01 -7.82338008e-02 1.29835045e+00
-1.49172887e-01 -2.46297076e-01 -8.98844659e-01 -2.22399846e-01
8.19073200e-01 8.33628416e-01 -3.99326295e-01 -4.50304687e-01
-7.95622468e-01 7.24254012e-01 6.86587989e-01 7.75193036e-01
-8.71054351e-01 -4.20323938e-01 7.82738805e-01 -8.02908465e-02
4.76009637e-01 -3.81949604e-01 -3.55501622e-01 -1.39012194e+00
5.58575749e-01 -5.27262151e-01 4.72622812e-01 -7.82819569e-01
-1.26759338e+00 1.15712672e-01 4.53826457e-01 -7.22447455e-01
-8.68831947e-02 -9.11289573e-01 -3.39255005e-01 6.57607377e-01
-1.55728555e+00 -1.33398211e+00 -9.32168663e-02 7.26220250e-01
-9.74874496e-02 -2.85916030e-02 1.21564567e+00 1.22440785e-01
-4.93625641e-01 4.89993036e-01 -5.36174625e-02 3.31371754e-01
3.14028323e-01 -1.51431620e+00 1.94211200e-01 5.88968337e-01
4.66247082e-01 1.23717761e+00 4.93108869e-01 -7.64007449e-01
-1.78373086e+00 -1.06311417e+00 1.06119132e+00 -7.94143260e-01
1.02245700e+00 -6.59717321e-01 -9.80039716e-01 1.26373994e+00
-2.94957727e-01 4.09675866e-01 9.43014741e-01 8.05042088e-01
-9.84621704e-01 -2.49053568e-01 -7.19746232e-01 9.88801599e-01
1.13459098e+00 -1.00875652e+00 -1.04558361e+00 8.15685838e-02
9.08402145e-01 -1.00432426e-01 -1.24112010e+00 6.83503211e-01
7.16780663e-01 -7.20583558e-01 1.09317982e+00 -9.83313441e-01
5.63321590e-01 -5.07733583e-01 -4.76293236e-01 -1.00642252e+00
-5.48168957e-01 -2.75057793e-01 -1.02765095e+00 1.20171893e+00
6.19390488e-01 -6.78048909e-01 6.68413281e-01 8.48717153e-01
1.88195214e-01 -1.10100234e+00 -9.36385214e-01 -1.03384018e+00
-2.42382381e-02 -6.25297725e-01 9.06632781e-01 1.29297376e+00
5.24589419e-01 3.87895018e-01 -2.66557366e-01 4.66134489e-01
3.40330839e-01 5.09602785e-01 7.33922184e-01 -1.09023488e+00
-3.14204752e-01 -3.21489006e-01 -7.23394513e-01 -1.08830118e+00
3.27164263e-01 -1.01084292e+00 -4.01887119e-01 -1.99296343e+00
1.46646708e-01 -5.23738444e-01 -3.38003635e-01 1.00918639e+00
2.54922779e-03 -1.28114477e-01 -3.34687270e-02 1.68798864e-01
-6.40070915e-01 7.84854054e-01 9.97151196e-01 -3.79531831e-01
-5.55999177e-05 -6.06153965e-01 -7.13708639e-01 6.33812964e-01
4.71673697e-01 -1.14598401e-01 -5.80513656e-01 -3.38844001e-01
9.79074180e-01 -5.68875298e-02 8.36139560e-01 -7.18931317e-01
4.38185513e-01 -2.93004274e-01 2.54784167e-01 -4.57756966e-01
6.18170857e-01 -9.17492092e-01 9.54978541e-02 -7.45860673e-03
-1.68613598e-01 -2.14552179e-01 3.85445029e-01 9.13744628e-01
-4.02842760e-01 9.04165134e-02 -1.29815757e-01 1.84530482e-01
-1.07193911e+00 1.25331953e-01 4.96884026e-02 -2.20474545e-02
8.41454983e-01 -2.62444586e-01 -7.57979929e-01 -1.94834188e-01
-6.31005883e-01 3.53433967e-01 4.49226469e-01 3.97274911e-01
7.60531723e-01 -1.76951659e+00 -3.58442366e-01 1.64000154e-01
7.10857451e-01 1.51789621e-01 -1.34297144e-02 7.60123193e-01
-3.88808012e-01 8.36404741e-01 1.55038357e-01 -1.75840169e-01
-7.30814815e-01 8.78963470e-01 -2.44870670e-02 -3.33503336e-01
-9.10721123e-01 6.25554144e-01 2.03497306e-01 -5.68309128e-01
1.98555186e-01 -7.43704140e-01 1.58966333e-01 -1.42346576e-01
3.20904672e-01 3.00985932e-01 -1.06236272e-01 -4.24647093e-01
-6.43978119e-01 3.32575381e-01 -3.91002856e-02 3.32228094e-02
1.26149499e+00 3.24927747e-01 -3.46723497e-01 6.87252462e-01
8.60682726e-01 2.28316322e-01 -9.10325110e-01 -6.23794734e-01
1.30769208e-01 -2.85879165e-01 -1.06185026e-01 -7.80898631e-01
-3.65470737e-01 7.10796952e-01 -2.71531522e-01 2.65727788e-01
7.29742765e-01 3.19555759e-01 4.78346586e-01 8.59070241e-01
6.12684727e-01 -9.69666004e-01 -9.79890227e-02 5.84029078e-01
9.61172521e-01 -8.18602920e-01 2.63316602e-01 -5.68292737e-01
-4.83485371e-01 1.06444263e+00 4.45787907e-01 2.23011136e-01
7.78410792e-01 2.42195159e-01 -5.29366553e-01 -5.98348141e-01
-1.13291764e+00 -2.54549444e-01 4.06782836e-01 3.61316025e-01
1.17552720e-01 2.90825367e-01 3.03277168e-02 7.42484033e-01
-2.12335110e-01 1.06749840e-01 1.45380303e-01 1.08832908e+00
-2.21989930e-01 -7.74139881e-01 -1.93010941e-01 4.39285845e-01
3.17492485e-02 -1.59871742e-01 -4.22253251e-01 1.16330349e+00
3.19451898e-01 6.37379527e-01 -7.65246004e-02 -3.88913006e-01
3.66597861e-01 3.66721690e-01 5.88231325e-01 -7.02736199e-01
3.46104413e-01 -6.08641446e-01 5.40424027e-02 -5.02049506e-01
-2.58560210e-01 -4.36112463e-01 -1.33677530e+00 -3.09093416e-01
-3.01027179e-01 1.09674208e-01 1.36172742e-01 8.95853519e-01
6.14014924e-01 4.47861493e-01 -1.18395306e-01 1.66678652e-02
-4.13002938e-01 -7.41352201e-01 -8.41772914e-01 4.93122160e-01
2.24640518e-01 -1.02242494e+00 -8.58593360e-03 3.58526967e-02]
|
[8.837167739868164, 7.764684677124023]
|
a7b849d8-71a0-4638-ad24-9a1dc08f4dd5
|
dl-corrector-remapper-a-grid-free-bias
|
2210.12293
| null |
https://arxiv.org/abs/2210.12293v1
|
https://arxiv.org/pdf/2210.12293v1.pdf
|
DL-Corrector-Remapper: A grid-free bias-correction deep learning methodology for data-driven high-resolution global weather forecasting
|
Data-driven models, such as FourCastNet (FCN), have shown exemplary performance in high-resolution global weather forecasting. This performance, however, is based on supervision on mesh-gridded weather data without the utilization of raw climate observational data, the gold standard ground truth. In this work we develop a methodology to correct, remap, and fine-tune gridded uniform forecasts of FCN so it can be directly compared against observational ground truth, which is sparse and non-uniform in space and time. This is akin to bias correction and post-processing of numerical weather prediction (NWP), a routine operation at meteorological and weather forecasting centers across the globe. The Adaptive Fourier Neural Operator (AFNO) architecture is used as the backbone to learn continuous representations of the atmosphere. The spatially and temporally non-uniform output is evaluated by the non-uniform discrete inverse Fourier transform (NUIDFT) given the output query locations. We call this network the Deep-Learning-Corrector-Remapper (DLCR). The improvement in DLCR's performance against the gold standard ground truth over the baseline's performance shows its potential to correct, remap, and fine-tune the mesh-gridded forecasts under the supervision of observations.
|
['Karthik Kashinath', 'Akshay Subramaniam', 'Jaideep Pathak', 'Tao Ge']
|
2022-10-21
| null | null | null | null |
['weather-forecasting']
|
['miscellaneous']
|
[-2.81842262e-01 -3.69032443e-01 5.33242762e-01 -5.89349329e-01
-6.04843259e-01 -6.80910826e-01 8.97220254e-01 -1.65377006e-01
-2.11166710e-01 1.13984478e+00 3.69472474e-01 -6.76020205e-01
-2.97401488e-01 -1.16842580e+00 -5.33261657e-01 -1.06877673e+00
-3.42914134e-01 5.24595439e-01 -1.38708308e-01 -7.48727024e-01
-3.05799227e-02 8.25568199e-01 -1.71746993e+00 1.04920799e-02
1.04872215e+00 1.01904535e+00 -3.67739201e-02 9.21362579e-01
-1.04574494e-01 6.13495231e-01 -3.54785949e-01 3.72603178e-01
5.20228863e-01 -3.99643749e-01 -4.01513219e-01 -5.28731108e-01
9.51843262e-01 -3.86171788e-01 1.25756517e-01 9.79336023e-01
5.32177687e-01 6.97188973e-01 5.87218583e-01 -8.15276265e-01
-4.67607945e-01 4.31044362e-02 -3.33943844e-01 3.85180682e-01
-3.08138758e-01 5.61731718e-02 9.54883516e-01 -1.19744158e+00
3.92648458e-01 9.90160406e-01 1.21283329e+00 1.80146307e-01
-1.52191520e+00 -5.17805696e-01 -1.99232414e-01 -3.44749391e-01
-1.36841953e+00 -6.12590909e-01 1.76022589e-01 -9.37585413e-01
1.12638009e+00 6.17052019e-01 5.17757952e-01 5.67680895e-01
3.67467821e-01 -3.37475270e-01 1.22424519e+00 -2.20007673e-01
2.70895839e-01 -1.11722119e-01 -3.02059621e-01 1.54253125e-01
1.55487908e-02 1.13010931e+00 -4.44909871e-01 -2.71324813e-01
7.91986287e-01 3.53883207e-02 -6.10031128e-01 1.64257176e-02
-8.73809099e-01 1.08367324e+00 6.31046176e-01 2.43250981e-01
-7.85406291e-01 -2.67412532e-02 7.95311779e-02 6.77072465e-01
1.30387914e+00 8.16791892e-01 -1.04765880e+00 1.19459759e-02
-1.84842026e+00 8.40758681e-01 7.40696371e-01 2.88374335e-01
9.29227710e-01 9.39117193e-01 -7.50210369e-03 5.02610981e-01
3.48398030e-01 1.36082029e+00 3.36543232e-01 -1.17283678e+00
9.58068520e-02 4.06149030e-02 6.27806544e-01 -1.06926250e+00
-6.07440174e-01 -7.67901540e-01 -1.41229725e+00 8.49197388e-01
1.26694843e-01 -8.72860312e-01 -1.03438830e+00 1.43605995e+00
5.41312754e-01 3.61315995e-01 3.06552947e-01 1.33821237e+00
6.75338149e-01 1.40888488e+00 -1.80028811e-01 4.53885980e-02
7.97838807e-01 -7.40575969e-01 -7.37437487e-01 -6.26931712e-02
7.13569105e-01 -7.93330014e-01 6.93914115e-01 -5.56912310e-02
-6.27739727e-01 -8.15587938e-01 -8.96675289e-01 1.46573931e-01
-9.44545448e-01 -2.28953771e-02 3.51183861e-01 5.71569242e-02
-1.61499166e+00 9.94814634e-01 -6.37869775e-01 -1.42619759e-01
-3.44877660e-01 -1.44966111e-01 -3.30066592e-01 6.15446568e-01
-1.69176352e+00 1.00872087e+00 2.09069535e-01 5.81856310e-01
-8.93415451e-01 -1.36451912e+00 -9.16155457e-01 2.88759768e-01
-3.99575293e-01 -6.73167646e-01 1.09746909e+00 -9.87158954e-01
-1.43799078e+00 4.77845430e-01 -2.23771006e-01 -8.98853958e-01
2.72056162e-01 -2.25324616e-01 -8.31550300e-01 -3.67452919e-01
1.44533053e-01 4.58970010e-01 9.82124865e-01 -1.08670115e+00
-9.53467369e-01 -7.39218444e-02 -4.81969595e-01 9.00084972e-02
3.92831951e-01 -2.89255947e-01 6.08135819e-01 -9.34972823e-01
1.14700861e-01 -8.37331712e-01 -3.93064350e-01 -6.59229532e-02
-1.46954451e-02 1.73575670e-01 8.50379527e-01 -1.06176245e+00
8.72961462e-01 -2.20714164e+00 -9.37014893e-02 5.06732404e-01
9.75357443e-02 2.63507783e-01 -1.03808366e-01 6.54000700e-01
-2.64029741e-01 -1.37569997e-02 -5.29983938e-01 -3.73261988e-01
-1.64896473e-01 5.73139608e-01 -1.23687851e+00 6.51864707e-01
2.33827978e-01 6.66981578e-01 -5.96808374e-01 1.96791917e-01
2.98463255e-01 7.29796052e-01 -4.27795589e-01 5.18706560e-01
-5.33943951e-01 1.08798838e+00 6.38929531e-02 -7.19111264e-02
1.22312164e+00 -1.58255175e-01 -3.79788220e-01 8.83505046e-02
-9.47088063e-01 4.68257308e-01 -1.23373008e+00 1.40169740e+00
-5.37975907e-01 8.70108426e-01 5.01658261e-01 -4.77444857e-01
1.08564627e+00 6.03544593e-01 7.21076205e-02 -9.53893006e-01
-5.14931381e-01 3.88896972e-01 -1.69088259e-01 -3.72080952e-01
7.70744860e-01 -2.74804950e-01 5.02466679e-01 4.45962727e-01
-1.64623737e-01 -4.22388762e-01 -3.29090267e-01 4.45673950e-02
2.41777018e-01 4.00488585e-01 -4.18344699e-02 -7.84997761e-01
4.63181853e-01 3.47934753e-01 6.04328692e-01 8.38030219e-01
4.34062451e-01 8.32835436e-01 1.40865937e-01 -1.05197942e+00
-1.22910559e+00 -7.67333865e-01 -5.14079988e-01 1.46038043e+00
-6.13586187e-01 -3.54325399e-02 -3.41385722e-01 -5.77547913e-03
3.03429753e-01 9.07122135e-01 -9.87178028e-01 3.94245535e-01
-3.74114573e-01 -8.54233623e-01 4.91119146e-01 1.82166398e-01
4.07580942e-01 -9.15778637e-01 -6.14549696e-01 2.60948241e-01
1.09668627e-01 -6.87481046e-01 -3.20353247e-02 2.70231754e-01
-9.72085416e-01 -5.31659842e-01 -7.46488690e-01 -6.57768995e-02
2.75122106e-01 -8.47955421e-02 1.34571087e+00 -1.88531935e-01
2.50964969e-01 -1.36680841e-01 -2.35519290e-01 -4.43573534e-01
-2.72118300e-01 1.24722920e-01 2.11591080e-01 1.51156306e-01
-2.95989484e-01 -9.37380433e-01 -5.60938179e-01 8.40897039e-02
-8.14120770e-01 1.00266166e-01 6.61716759e-02 7.30054200e-01
6.51883125e-01 -2.27840722e-01 2.56907731e-01 -8.73164594e-01
3.54922444e-01 -6.55587256e-01 -1.43781281e+00 -3.97591554e-02
-9.44557011e-01 1.30978063e-01 5.82976580e-01 3.44408602e-01
-1.29912031e+00 -2.47333407e-01 -3.96192402e-01 -5.26658952e-01
-2.21394002e-01 9.37358379e-01 5.92767537e-01 -2.61922717e-01
1.11145210e+00 3.68680507e-01 -1.56959996e-01 -9.33382809e-01
3.62144768e-01 3.68381053e-01 7.94031203e-01 -3.04795206e-01
9.62775528e-01 4.57464516e-01 3.02460551e-01 -8.50476742e-01
-1.06494379e+00 -3.59598190e-01 -4.67507869e-01 1.56548526e-02
8.06785703e-01 -1.22686362e+00 -7.80818164e-02 4.06052440e-01
-1.12935579e+00 -7.21612573e-01 -5.47163606e-01 5.17802775e-01
5.58056459e-02 -3.44497919e-01 -2.25053728e-01 -9.31460738e-01
-7.76011169e-01 -4.39482182e-01 1.02841234e+00 1.67360321e-01
-3.47580388e-02 -1.34175825e+00 8.75573814e-01 -5.32651603e-01
1.27517366e+00 6.58178210e-01 6.41006052e-01 -2.90433943e-01
-2.39679679e-01 -2.83861868e-02 -3.64998877e-01 4.82022375e-01
1.09495856e-01 2.41198793e-01 -1.43748915e+00 -3.91315877e-01
2.78109998e-01 6.30911961e-02 9.40756381e-01 6.85899615e-01
7.59146035e-01 -5.41201711e-01 1.73134059e-01 1.45358157e+00
1.57464564e+00 -3.12487692e-01 1.53766215e-01 1.02852292e-01
5.49197435e-01 5.23945272e-01 3.12122256e-01 4.31697100e-01
1.81675285e-01 2.58401811e-01 6.03824615e-01 -6.38901412e-01
2.00562969e-01 4.15543690e-02 -1.43732056e-01 6.32599771e-01
-6.17527187e-01 -8.03052038e-02 -1.16494155e+00 5.31796575e-01
-1.77661872e+00 -1.02927864e+00 -5.08242846e-01 2.05073309e+00
7.26551354e-01 -3.24132174e-01 -6.05586767e-01 -2.57874429e-01
1.56510293e-01 6.83673143e-01 -3.92100424e-01 -6.91871941e-01
-4.03082937e-01 6.84447527e-01 1.05314887e+00 8.73920202e-01
-1.19723535e+00 6.79465592e-01 6.04507351e+00 2.30807632e-01
-1.62690616e+00 4.60336447e-01 3.93175155e-01 1.41479298e-01
-1.78770795e-01 -1.34316415e-01 -7.91001081e-01 3.28401029e-01
1.63970256e+00 6.47656918e-02 6.47484601e-01 6.65390134e-01
8.44185770e-01 6.99653998e-02 -6.86732590e-01 6.40728772e-01
-5.44344127e-01 -1.96293366e+00 -1.06815234e-01 -5.25989085e-02
1.32138538e+00 7.01207638e-01 1.50856506e-02 2.85176933e-01
7.08252013e-01 -1.22933340e+00 4.56636429e-01 1.12020898e+00
1.08439422e+00 -5.15213966e-01 9.43424642e-01 3.80143493e-01
-1.10871279e+00 2.98179120e-01 -4.01333660e-01 -4.06626612e-01
1.28278464e-01 1.00536847e+00 -4.48947489e-01 7.24042296e-01
1.13834357e+00 7.21029997e-01 1.25030251e-02 8.43253791e-01
-9.26685929e-02 9.97246087e-01 -4.51672286e-01 7.63305306e-01
6.72653437e-01 -3.44859511e-01 6.10767126e-01 1.06566978e+00
7.54382849e-01 3.54219824e-01 -6.40051439e-02 7.96751440e-01
1.64926752e-01 -3.84770427e-03 -5.36729634e-01 4.96501178e-01
1.28078759e-01 1.10320115e+00 1.46181986e-01 -4.33414102e-01
-1.08252250e-01 5.41944623e-01 -1.71271116e-02 7.19606161e-01
-4.89188224e-01 -2.24185139e-01 1.36044955e+00 1.97458789e-01
5.28758109e-01 -3.11478108e-01 -1.40698612e-01 -7.78860331e-01
-3.84114057e-01 -8.48349094e-01 2.12001532e-01 -1.12729073e+00
-1.36249161e+00 7.81590581e-01 -2.01804072e-01 -1.29208493e+00
-5.39214253e-01 -4.83895868e-01 -8.55532169e-01 1.92672467e+00
-2.00186563e+00 -7.79332101e-01 -5.06564558e-01 5.40234089e-01
-6.18696632e-03 -3.09155136e-02 1.25143528e+00 1.28467992e-01
-4.04684171e-02 -8.33689496e-02 1.02825058e+00 6.31320551e-02
7.78545320e-01 -1.50762033e+00 9.21837747e-01 8.31354797e-01
-1.81675434e-01 3.40388209e-01 1.04548705e+00 -6.28338218e-01
-9.90358293e-01 -1.76943135e+00 1.18573022e+00 -7.59599805e-02
6.33899689e-01 -8.04536790e-02 -1.44177008e+00 6.57459855e-01
2.95790464e-01 5.63295007e-01 2.23515883e-01 4.23069717e-03
-2.55110443e-01 -6.20508611e-01 -1.09240460e+00 -2.65205689e-02
2.37113029e-01 -5.54681957e-01 -5.76263309e-01 6.84666932e-01
7.69762039e-01 -8.86857450e-01 -8.57320011e-01 6.29049301e-01
3.72965425e-01 -1.17185318e+00 6.00192785e-01 -8.37182462e-01
5.42894304e-01 -5.46874046e-01 -2.92808473e-01 -1.95625842e+00
-4.81008321e-01 -5.72021842e-01 1.53921291e-01 9.59961653e-01
5.70851564e-01 -1.17865753e+00 2.16634989e-01 2.42102310e-01
-1.34820178e-01 -2.34197184e-01 -1.09693706e+00 -4.71205860e-01
4.39218044e-01 -3.08660597e-01 1.05903685e+00 1.40911531e+00
-9.49832916e-01 -1.78219348e-01 -6.05825901e-01 1.05707610e+00
4.59297299e-01 5.40274382e-01 5.48126876e-01 -1.63572228e+00
-1.05647139e-01 -1.48434564e-01 9.68354493e-02 -6.20070755e-01
-1.94554985e-01 -5.47516108e-01 3.37146968e-01 -1.36178637e+00
-9.52510953e-01 -5.46134651e-01 -2.89271474e-01 3.62160563e-01
1.85047820e-01 1.58507004e-01 -2.50992715e-01 3.66045296e-01
3.62595677e-01 5.43501258e-01 1.13040423e+00 1.86942190e-01
-2.12103993e-01 -8.61227512e-02 2.87917495e-01 4.20275539e-01
8.53292704e-01 -5.60516953e-01 -4.86706831e-02 -9.48794365e-01
6.06379449e-01 1.49063379e-01 4.66449916e-01 -1.20164454e+00
3.05429459e-01 -2.54802793e-01 6.51939929e-01 -6.64360702e-01
1.06323063e-01 -7.92500913e-01 6.65656149e-01 4.39395495e-02
-2.81442195e-01 4.33274537e-01 5.16959012e-01 7.85159767e-02
-5.21166742e-01 2.64440745e-01 7.65571952e-01 -1.57310933e-01
-6.64650261e-01 3.61566544e-01 -4.27190751e-01 1.35118484e-01
4.20989215e-01 2.94759870e-01 -1.98482364e-01 -4.44407821e-01
-7.26282775e-01 3.30048919e-01 2.97279477e-01 3.77744101e-02
2.00358421e-01 -1.07665241e+00 -1.31552029e+00 9.96749401e-01
-1.79987893e-01 2.52029836e-01 4.53154653e-01 7.68630743e-01
-8.19058418e-01 6.31258368e-01 -3.61482054e-02 -5.71916521e-01
-6.26488745e-01 1.28993792e-02 1.20283782e+00 -1.17203943e-01
-7.65310585e-01 8.49349618e-01 4.95253913e-02 -1.06827605e+00
-2.50923514e-01 -7.47896254e-01 -1.51023954e-01 2.15666488e-01
6.19878829e-01 1.82986706e-01 4.35998648e-01 -7.27555931e-01
6.13572970e-02 6.33680463e-01 6.64545000e-01 -2.36756086e-01
1.49080062e+00 -9.49780867e-02 -4.29536581e-01 7.28667378e-01
8.69521856e-01 -1.32551026e-02 -1.48879647e+00 -3.24041307e-01
-2.95108765e-01 -3.82045448e-01 7.53278434e-01 -9.37727690e-01
-1.40165639e+00 1.07177091e+00 8.11509490e-01 5.87624013e-01
8.45921814e-01 -7.21867502e-01 6.48493826e-01 3.60705942e-01
-2.55612373e-01 -7.19054282e-01 -1.30545139e+00 1.02234221e+00
1.15604353e+00 -1.26715624e+00 -8.76110569e-02 4.76279736e-01
-2.74825692e-01 1.06396019e+00 1.26059532e-01 -2.80918926e-01
1.34736454e+00 4.22868013e-01 6.60014689e-01 -3.03973526e-01
-7.73823142e-01 8.30648001e-03 5.16305506e-01 2.04169020e-01
4.03158247e-01 3.23336244e-01 4.41079587e-01 1.01668888e-03
-5.26649952e-01 1.39996111e-01 3.03125270e-02 4.99472022e-01
-4.41881388e-01 -3.77466738e-01 -8.65743339e-01 5.89266062e-01
-1.81067541e-01 -6.41402185e-01 3.92036527e-01 3.80567670e-01
2.69978672e-01 6.81281745e-01 6.10698283e-01 1.29749507e-01
2.30291471e-01 2.57306904e-01 -4.37981039e-01 -3.13021332e-01
-7.02740073e-01 -3.32493216e-01 -6.10231943e-02 -7.11104751e-01
-3.21044564e-01 -4.16545123e-01 -8.68899822e-01 -7.20268250e-01
2.80985326e-01 7.13474751e-01 7.11804271e-01 9.51523960e-01
5.32659709e-01 4.76189554e-01 6.85758233e-01 -1.50928056e+00
-5.23860395e-01 -1.11971200e+00 -8.51589024e-01 -1.35291070e-02
1.40433550e+00 -4.24483657e-01 -7.79310703e-01 -7.35569298e-02]
|
[6.581233978271484, 2.9632344245910645]
|
9e25523e-6550-4e5e-9713-41bbb072d870
|
long-range-3d-with-quadocular-thermal-lwir
|
1911.06975
| null |
https://arxiv.org/abs/1911.06975v2
|
https://arxiv.org/pdf/1911.06975v2.pdf
|
Long Range 3D with Quadocular Thermal (LWIR) Camera
|
Long Wave Infrared (LWIR) cameras provide images regardles of the ambient illumination, they tolerate fog and are not blinded by the incoming car headlights. These features make LWIR cameras attractive for autonomous navigation, security and military applications. Thermal images can be used similarly to the visible range ones, including 3D scene reconstruction with two or more such cameras mounted on a rigid frame. There are two additional challenges for this spectral range: lower image resolution and lower contrast of the textures. In this work, we demonstrate quadocular LWIR camera setup, calibration, image capturing and processing that result in long range 3D perception with 0.077 pix disparity error over 90% of the depth map. With low resolution (160 x 120) LWIR sensors we achieved 10% range accuracy at 28 m with 56 degrees horizontal field of view (HFoV) and 150 mm baseline. Scaled to the now-standard 640 x 512 resolution and 200 mm baseline suitable for head-mounted application the result would be 10% accuracy at 130 m.
|
['Andrey Filippov', 'Oleg Dzhimiev']
|
2019-11-16
| null | null | null | null |
['3d-scene-reconstruction']
|
['computer-vision']
|
[ 3.87061894e-01 -2.21376151e-01 3.98004740e-01 -3.30901235e-01
-3.84951025e-01 -8.30889404e-01 3.36176246e-01 -3.86388928e-01
-9.63750541e-01 6.07949495e-01 9.07840021e-03 -3.46251845e-01
1.22797370e-01 -6.10316753e-01 -4.45363253e-01 -9.06332493e-01
4.38530743e-01 -1.34777069e-01 4.08975631e-01 -1.09227166e-01
2.01698735e-01 5.52649021e-01 -1.96947670e+00 1.58326820e-01
3.33305031e-01 1.00143921e+00 3.47276896e-01 1.01175964e+00
4.79769617e-01 7.46704400e-01 -2.64494598e-01 1.03338175e-02
6.98329806e-01 1.12767495e-01 -1.89085193e-02 1.84240099e-02
1.20130837e+00 -9.22338843e-01 -6.07660636e-02 8.58887434e-01
5.86589813e-01 1.98298857e-01 2.74943918e-01 -7.22479045e-01
-3.69926125e-01 -3.69931549e-01 -8.59024405e-01 6.91236034e-02
7.19252825e-01 3.75962377e-01 3.20844084e-01 -8.86879802e-01
2.73987412e-01 8.15389037e-01 5.97335696e-01 5.04467666e-01
-8.83843720e-01 -6.18101478e-01 -2.93542534e-01 1.45509839e-01
-1.31240129e+00 -5.52586675e-01 2.10525736e-01 -3.93335164e-01
1.00836289e+00 4.14348006e-01 5.89615166e-01 5.23303509e-01
5.69858074e-01 -5.71130574e-01 1.69498062e+00 -4.38867807e-01
5.79689555e-02 5.66915751e-01 1.56207368e-01 2.90707171e-01
8.61339211e-01 5.19048035e-01 -6.65070295e-01 1.29512548e-01
5.57366669e-01 2.62062728e-01 -7.87645280e-01 9.33318958e-02
-9.38730359e-01 4.75620687e-01 4.23971355e-01 -1.13082893e-01
1.20042436e-01 7.66962320e-02 -3.46560240e-01 3.76185179e-01
2.81764746e-01 2.38294169e-01 -2.49518067e-01 1.69778958e-01
-2.26585880e-01 -1.58027396e-01 2.14553267e-01 8.38063359e-01
8.50073099e-01 5.39923757e-02 4.02811199e-01 4.22348320e-01
6.36859357e-01 1.32560384e+00 1.15471162e-01 -1.04307544e+00
2.55926311e-01 1.12059340e-01 4.33303356e-01 -2.16611177e-01
-2.21140012e-01 -2.17670277e-02 -2.35430315e-01 1.14710617e+00
2.94679374e-01 -2.40787998e-01 -1.01855993e+00 9.88532901e-01
3.12662303e-01 -8.61693919e-02 3.70044112e-01 1.17494226e+00
9.42352533e-01 6.79163575e-01 -5.88320971e-01 -3.63787860e-01
1.37536538e+00 -4.08014923e-01 -5.31565309e-01 -5.83176732e-01
1.98336184e-01 -1.07670438e+00 8.85109961e-01 6.47462130e-01
-9.87360775e-01 -5.48899710e-01 -1.13323867e+00 -1.31807342e-01
-8.72825906e-02 2.09100563e-02 -5.82628511e-02 1.13306689e+00
-1.33463061e+00 -5.48908114e-02 -5.27895629e-01 -3.83207560e-01
-3.06717426e-01 1.76708698e-01 -5.97262383e-01 -5.64853013e-01
-6.78796053e-01 1.04214323e+00 -6.24962747e-02 3.39115053e-01
-3.66658807e-01 -7.21896648e-01 -8.70772660e-01 -5.13382494e-01
1.17557244e-02 -4.39418197e-01 7.37406909e-01 -4.63584125e-01
-1.71302104e+00 1.23237979e+00 -3.29013795e-01 -3.58791083e-01
4.45620328e-01 -6.40258253e-01 -4.77537304e-01 2.91029245e-01
-2.30277270e-01 1.98247924e-01 8.17386508e-01 -1.17144990e+00
-5.74910223e-01 -7.65016079e-01 9.32611618e-03 5.19792497e-01
1.92876145e-01 2.64682978e-01 -5.15073761e-02 4.41081583e-01
1.46112874e-01 -9.93291855e-01 -4.45979461e-02 2.63084620e-01
1.23529844e-01 6.34954810e-01 1.18325329e+00 -2.18183294e-01
4.10236627e-01 -2.00272560e+00 -6.10958397e-01 -1.43276840e-01
3.39428373e-02 2.87214041e-01 1.37643635e-01 2.99717307e-01
-5.50291203e-02 -4.58616823e-01 5.49888872e-02 -7.34121576e-02
-6.68477178e-01 2.84139886e-02 -2.26230234e-01 1.16940916e+00
-6.28911316e-01 5.91306567e-01 -4.04480428e-01 1.08917058e-01
8.33957076e-01 9.16930258e-01 -2.72479244e-02 1.93528607e-01
4.22831297e-01 7.05384493e-01 2.45538726e-01 6.13536298e-01
1.12276506e+00 5.98497510e-01 -2.79321760e-01 -2.36877114e-01
-8.37824702e-01 -2.33073846e-01 -1.11907291e+00 1.22114933e+00
-4.42390233e-01 1.07536829e+00 4.94304925e-01 6.56916946e-02
1.20892656e+00 8.10389370e-02 -9.25483778e-02 -9.54802036e-01
2.75729626e-01 1.62752435e-01 -4.24550027e-01 -5.29605329e-01
6.59731269e-01 -6.16293490e-01 3.70658189e-01 2.50219941e-01
-7.33024478e-01 -3.63860220e-01 -4.91287291e-01 -3.72547358e-01
6.34426951e-01 -1.55487005e-02 1.45495562e-02 -2.33383253e-01
6.01880968e-01 -1.03386100e-02 1.56053767e-01 6.27400398e-01
1.09356746e-01 9.73191977e-01 -4.57435876e-01 -6.63688302e-01
-8.58837962e-01 -1.07776666e+00 -4.49548632e-01 4.88782823e-01
5.28695405e-01 1.47173986e-01 -3.72346967e-01 4.85737264e-01
-3.61923009e-01 6.74437940e-01 -2.41913900e-01 1.46355748e-01
-5.09925842e-01 -6.20381892e-01 -3.55181750e-03 1.48960918e-01
7.14122653e-01 -3.73184562e-01 -1.59913552e+00 -3.44310582e-01
7.36537203e-02 -1.47929585e+00 1.43971577e-01 -1.56437591e-01
-7.63233960e-01 -1.19113481e+00 -4.52956706e-01 -2.24177048e-01
4.78848994e-01 1.10460603e+00 9.90624189e-01 -3.71950060e-01
-6.86855197e-01 6.59837544e-01 -4.03027654e-01 -4.15081054e-01
1.56857938e-01 -9.03928339e-01 2.02399388e-01 -7.14329630e-02
3.85685176e-01 -1.04958117e-01 -9.64327991e-01 4.86918211e-01
-5.28724968e-01 -2.06413329e-01 3.68963510e-01 2.40549535e-01
2.47749031e-01 -3.11803184e-02 -3.12777698e-01 -5.16375244e-01
-1.48068503e-01 1.97833955e-01 -1.33348656e+00 -1.69786587e-01
-7.45793104e-01 -5.76668084e-01 2.32703805e-01 4.54894938e-02
-1.27201498e+00 7.48655898e-03 1.92341164e-01 -2.65722513e-01
-5.33884883e-01 -1.75723404e-01 1.43928587e-01 -3.65888536e-01
9.62490857e-01 -3.38034071e-02 2.08806738e-01 -1.12482771e-01
5.99665605e-02 9.67451453e-01 6.63069427e-01 1.09177075e-01
1.00413394e+00 1.13675833e+00 2.17597604e-01 -1.42520273e+00
-7.11772442e-01 -7.34822273e-01 -6.33200526e-01 -4.62217331e-01
1.39439952e+00 -1.52381134e+00 -9.63890135e-01 4.78069842e-01
-7.18229651e-01 -3.15953881e-01 -3.44055966e-02 9.92839754e-01
5.06316312e-02 3.68334621e-01 -3.38772506e-01 -1.21525288e+00
-4.41082358e-01 -8.84036064e-01 1.16921628e+00 4.65047896e-01
2.53197521e-01 -7.80280888e-01 -3.07754166e-02 8.55685174e-01
6.02954865e-01 3.34746063e-01 9.14268494e-02 8.99900019e-01
-7.49922037e-01 -4.00556803e-01 -1.98481426e-01 2.77381480e-01
9.47466493e-02 -2.18980685e-02 -1.75547230e+00 -4.60068405e-01
4.15583014e-01 -1.44968003e-01 8.96170855e-01 6.28192246e-01
3.45267296e-01 1.19852953e-01 -3.16813700e-02 9.78954673e-01
1.94567001e+00 2.79947191e-01 9.78254259e-01 5.71995318e-01
7.61883974e-01 8.90411496e-01 1.00385642e+00 2.54291147e-01
3.81132402e-02 7.13437438e-01 7.53341556e-01 -2.80593753e-01
-1.34117892e-02 4.41441178e-01 8.21168303e-01 -2.47672811e-01
-3.25796366e-01 -1.56688660e-01 -7.95439482e-01 -1.78028401e-02
-1.12327361e+00 -1.02213836e+00 -9.07367051e-01 2.85025764e+00
1.70659497e-01 -2.18285933e-01 -2.42722362e-01 4.07505929e-02
6.07007563e-01 -1.17926002e-01 -2.62259334e-01 -3.49562168e-01
-2.03407913e-01 8.26597363e-02 1.44723380e+00 1.17129564e+00
-7.11163759e-01 4.50123668e-01 6.10907078e+00 -1.55701116e-01
-1.35163140e+00 6.09905645e-02 2.29663864e-01 -4.30217832e-01
-2.40840569e-01 -1.39266059e-01 -1.26806700e+00 1.73943698e-01
1.01618874e+00 2.50355393e-01 2.32944518e-01 3.76537114e-01
5.66171169e-01 -7.04150617e-01 -7.41514266e-01 1.49055517e+00
3.94589663e-01 -6.80009842e-01 -5.26698291e-01 5.15459955e-01
5.19513011e-01 6.29082084e-01 2.16356501e-01 -6.15544796e-01
-9.49945897e-02 -1.14107835e+00 6.45238101e-01 1.48861393e-01
1.35544193e+00 -7.94342339e-01 6.76006138e-01 3.52578700e-01
-1.06414139e+00 -1.33335188e-01 -1.00854278e+00 -3.79493356e-01
1.78372428e-01 4.56997663e-01 -5.77322185e-01 2.08607867e-01
1.14295518e+00 4.44408178e-01 -5.47235131e-01 7.18057811e-01
-2.70757496e-01 -4.88461107e-02 -5.75930953e-01 3.36976111e-01
3.05512678e-02 -7.85961688e-01 4.69677597e-01 6.98597729e-01
5.66523314e-01 2.67033786e-01 -1.28636226e-01 4.55084026e-01
6.03971064e-01 -5.18223643e-01 -1.12731099e+00 9.44649994e-01
2.89644096e-02 1.27398670e+00 -3.25681508e-01 8.55304375e-02
-5.83176553e-01 8.17981899e-01 -6.56970024e-01 4.79313612e-01
-6.77965820e-01 -3.64127576e-01 9.38728154e-01 5.12968779e-01
-1.22706769e-02 -5.04299164e-01 -8.33881125e-02 -1.16837513e+00
1.31481186e-01 -1.33918345e-01 7.48048201e-02 -1.38729465e+00
-6.74476445e-01 9.14589942e-01 7.33249262e-02 -1.43081880e+00
1.40251487e-01 -1.09923911e+00 -2.52938658e-01 1.14948237e+00
-2.03608370e+00 -1.14261937e+00 -1.18562496e+00 9.08132613e-01
3.77875179e-01 1.60839021e-01 8.57521296e-01 -5.45437559e-02
-7.61891669e-03 6.28199726e-02 3.37702990e-01 -3.02926093e-01
9.04679894e-01 -9.78555024e-01 -9.90238488e-02 1.16071832e+00
-2.47589439e-01 3.93317968e-01 8.00584137e-01 -3.06582868e-01
-1.87258196e+00 -1.08767486e+00 5.21222949e-01 -8.09162021e-01
2.19372123e-01 -5.28467476e-01 -4.47754025e-01 6.06125116e-01
4.67280835e-01 3.98728102e-01 6.52130246e-01 -4.71586287e-01
-4.53545898e-01 -5.38306057e-01 -1.49638832e+00 2.93808192e-01
6.20835543e-01 -5.27033448e-01 -3.96791190e-01 3.59134853e-01
3.56039762e-01 -6.32660806e-01 -4.21028674e-01 2.62943536e-01
9.32465374e-01 -1.54132283e+00 1.11909378e+00 5.97880900e-01
-2.58985490e-01 -7.76515007e-01 -3.08938712e-01 -6.76385164e-01
-1.05148725e-01 -4.84095126e-01 8.11745465e-01 8.02611709e-01
1.85164273e-01 -1.33093596e+00 6.20593250e-01 7.53612518e-01
-1.69856891e-01 2.17274517e-01 -8.82637739e-01 -7.49725342e-01
-3.76360089e-01 -2.93694228e-01 -1.52362272e-01 4.09706295e-01
-3.92043054e-01 3.88350755e-01 -5.29957891e-01 8.13403249e-01
1.24827862e+00 2.72199541e-01 8.73789370e-01 -1.28965652e+00
-3.00065011e-01 3.37295413e-01 -4.57523644e-01 -6.67632818e-01
-3.12511861e-01 -2.04535007e-01 -7.73634836e-02 -1.66544497e+00
2.84145158e-02 -1.16985045e-01 3.68783712e-01 2.86929626e-02
2.10977212e-01 8.04902852e-01 -1.85929351e-02 2.09840178e-01
7.68679008e-03 8.20560195e-03 6.83108628e-01 2.70898074e-01
-9.40710828e-02 -2.07004189e-01 -4.62321162e-01 7.04456508e-01
6.66141212e-01 -1.66052893e-01 -5.90801418e-01 -8.63931775e-01
3.27528596e-01 -2.32159626e-02 5.22284627e-01 -1.32229424e+00
3.77061248e-01 -1.32664114e-01 6.17433250e-01 -7.54532516e-01
8.16074908e-01 -1.33368564e+00 7.14629531e-01 7.61606634e-01
2.49353215e-01 7.61837959e-02 1.51455313e-01 5.64420104e-01
1.07630044e-02 -2.24773750e-01 1.37008262e+00 -4.06536639e-01
-8.47132623e-01 2.47196816e-02 -5.32427907e-01 -4.42221403e-01
1.45957506e+00 -8.30119133e-01 -1.12871385e+00 -3.89336318e-01
-1.31224990e-01 -1.74338818e-01 1.21981609e+00 1.71880685e-02
9.38689888e-01 -6.07235432e-01 -6.40171349e-01 5.48514903e-01
3.95758450e-01 1.74088359e-01 2.47081727e-01 7.66150534e-01
-1.29231250e+00 4.28644836e-01 -1.09044001e-01 -9.31345940e-01
-1.98893297e+00 1.62225842e-01 5.63096642e-01 7.63352871e-01
-1.06743360e+00 9.01037455e-01 3.74016672e-01 7.06380606e-02
-2.11633548e-01 -1.76666662e-01 -2.39541128e-01 -2.63562083e-01
1.12478125e+00 4.23191994e-01 1.82278648e-01 -7.59463966e-01
-5.80238163e-01 1.63091147e+00 2.62322009e-01 -2.56740659e-01
1.27092552e+00 -8.95228207e-01 -1.54463887e-01 5.75936556e-01
8.70255947e-01 4.97666895e-01 -1.27795041e+00 2.10427493e-01
-8.26578379e-01 -1.06301296e+00 3.33293706e-01 -5.19284189e-01
-7.32986331e-01 9.78472412e-01 1.06801641e+00 -1.47245854e-01
1.20873880e+00 -1.21570818e-01 1.88890487e-01 4.91479456e-01
6.32806003e-01 -6.93845391e-01 -4.10478443e-01 4.22226757e-01
3.82013261e-01 -1.30165696e+00 3.02580357e-01 -5.23268461e-01
-6.61957979e-01 1.32124484e+00 3.77619416e-01 4.32881080e-02
2.58755982e-01 4.60929602e-01 7.79763103e-01 -2.06639856e-01
-6.75518036e-01 -4.52442080e-01 -2.18861680e-02 1.06769729e+00
4.69464272e-01 -2.11840719e-01 1.78267628e-01 -7.11773038e-01
-1.01627134e-01 -3.41896594e-01 1.10106015e+00 8.58105004e-01
-9.27057147e-01 -6.04824781e-01 -8.80525112e-01 -8.63805134e-03
-5.26889503e-01 5.16661815e-02 -1.57504119e-02 4.65107739e-01
-5.14737591e-02 1.57384348e+00 1.84653834e-01 -1.04986206e-01
3.58297735e-01 -3.10899943e-01 6.51165843e-01 -5.81912816e-01
-3.28630388e-01 2.92257011e-01 2.54223272e-02 -6.90392017e-01
-7.31204867e-01 -6.13353610e-01 -7.60892510e-01 -3.34258974e-01
-2.30363756e-01 -4.66466732e-02 1.08068013e+00 2.90571392e-01
8.24247673e-02 -1.23866603e-01 6.25748456e-01 -7.81225681e-01
4.16355170e-02 -7.76046813e-01 -1.03597248e+00 -1.78719163e-01
7.19086707e-01 -3.75026405e-01 -9.53588247e-01 7.34523162e-02]
|
[9.91514778137207, -2.7109646797180176]
|
949902ec-653a-409a-8862-519c1bd5c1a8
|
towards-3d-human-pose-estimation-in-the-wild
|
1704.02447
| null |
http://arxiv.org/abs/1704.02447v2
|
http://arxiv.org/pdf/1704.02447v2.pdf
|
Towards 3D Human Pose Estimation in the Wild: a Weakly-supervised Approach
|
In this paper, we study the task of 3D human pose estimation in the wild.
This task is challenging due to lack of training data, as existing datasets are
either in the wild images with 2D pose or in the lab images with 3D pose.
We propose a weakly-supervised transfer learning method that uses mixed 2D
and 3D labels in a unified deep neutral network that presents two-stage
cascaded structure. Our network augments a state-of-the-art 2D pose estimation
sub-network with a 3D depth regression sub-network. Unlike previous two stage
approaches that train the two sub-networks sequentially and separately, our
training is end-to-end and fully exploits the correlation between the 2D pose
and depth estimation sub-tasks. The deep features are better learnt through
shared representations. In doing so, the 3D pose labels in controlled lab
environments are transferred to in the wild images. In addition, we introduce a
3D geometric constraint to regularize the 3D pose prediction, which is
effective in the absence of ground truth depth labels. Our method achieves
competitive results on both 2D and 3D benchmarks.
|
['Qi-Xing Huang', 'xiangyang xue', 'Xiao Sun', 'Yichen Wei', 'Xingyi Zhou']
|
2017-04-08
|
towards-3d-human-pose-estimation-in-the-wild-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Zhou_Towards_3D_Human_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Zhou_Towards_3D_Human_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['3d-multi-person-pose-estimation-absolute', '3d-multi-person-pose-estimation-root-relative', 'monocular-3d-human-pose-estimation']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[-1.00993611e-01 3.11804384e-01 -1.05091758e-01 -5.52584410e-01
-8.06466460e-01 -4.01732802e-01 2.82627910e-01 -2.70359337e-01
-7.36756146e-01 4.45861489e-01 2.28215888e-01 1.35095552e-01
4.77522850e-01 -5.50546706e-01 -1.01094639e+00 -3.00723344e-01
5.53356074e-02 8.25290024e-01 1.08439893e-01 -1.89096034e-01
-1.57538280e-02 4.92826939e-01 -1.30776298e+00 -2.66875569e-02
2.80555934e-01 1.30101264e+00 -1.01523802e-01 7.68483460e-01
3.15028846e-01 4.68906879e-01 -4.66382027e-01 -1.64448112e-01
7.77688563e-01 -1.48466706e-01 -7.79058099e-01 3.20130676e-01
8.70015025e-01 -9.94082212e-01 -4.73111928e-01 6.42297328e-01
9.76683795e-01 -9.83858854e-02 5.78970909e-01 -1.21919572e+00
-1.24976821e-01 -2.72025347e-01 -8.05781186e-01 -2.96248287e-01
7.38461077e-01 2.80558795e-01 5.90957105e-01 -8.97857189e-01
5.15644312e-01 1.52040291e+00 8.05875063e-01 7.36226678e-01
-1.13620675e+00 -7.12627351e-01 3.69388551e-01 -3.79376352e-01
-1.27748632e+00 -3.08401563e-04 1.01283288e+00 -7.63538241e-01
9.46452200e-01 -4.54435021e-01 8.25518429e-01 1.48927808e+00
1.13190122e-01 1.03876698e+00 9.45230246e-01 -3.56127501e-01
-3.67792062e-02 -1.54652148e-01 -2.91622251e-01 8.91809523e-01
4.99098636e-02 1.83620140e-01 -6.37278318e-01 2.08612204e-01
1.32369721e+00 2.45135561e-01 -2.94467192e-02 -9.78183210e-01
-1.03543496e+00 6.80934250e-01 6.85566068e-01 -2.36586466e-01
-1.24666639e-01 2.49012902e-01 3.35367978e-01 4.29476909e-02
8.70042562e-01 3.10205370e-01 -9.15405989e-01 5.28230332e-02
-6.77716911e-01 5.93880951e-01 6.26641512e-01 9.18994784e-01
7.65029490e-01 -5.25135994e-01 -1.43767312e-01 5.03592491e-01
3.36135834e-01 3.95458311e-01 3.41412187e-01 -9.02936637e-01
6.50124490e-01 6.71435237e-01 3.26597661e-01 -6.95791543e-01
-7.01561272e-01 -4.45319831e-01 -4.67359245e-01 4.41090107e-01
6.26737475e-01 -1.76105902e-01 -1.02024996e+00 1.90113604e+00
5.98042965e-01 -4.99310121e-02 -2.47117892e-01 1.20855272e+00
7.07100332e-01 2.55438566e-01 -3.14227305e-02 4.96872485e-01
9.98581827e-01 -1.40294540e+00 -4.05273765e-01 -5.99900723e-01
6.55665874e-01 -4.48239446e-01 9.98097420e-01 1.92387536e-01
-1.02737010e+00 -8.57819021e-01 -1.06502283e+00 -5.67144334e-01
-3.77744287e-01 2.42383912e-01 4.50390667e-01 3.72508734e-01
-9.22004282e-01 5.19032478e-01 -9.83045936e-01 -3.33142966e-01
4.96230811e-01 5.14061093e-01 -6.59721315e-01 -1.19178072e-01
-1.02087343e+00 1.03402841e+00 8.17345604e-02 2.65654415e-01
-1.02572870e+00 -6.45644546e-01 -1.34351373e+00 -4.62207764e-01
5.23782372e-01 -9.19553399e-01 1.34443676e+00 -4.62386638e-01
-1.61824417e+00 1.54742718e+00 5.29180504e-02 -8.24001357e-02
1.06754780e+00 -9.65963840e-01 5.10473251e-01 9.00539830e-02
3.59473735e-01 1.05978751e+00 6.31003678e-01 -1.27602494e+00
-2.05517128e-01 -9.22999918e-01 2.72951156e-01 4.47675526e-01
-2.51507014e-02 -6.15545273e-01 -8.65324378e-01 -5.00735819e-01
1.68524981e-01 -1.04713273e+00 -2.42236957e-01 6.62206948e-01
-4.17575210e-01 -1.45480901e-01 5.90064347e-01 -5.15573621e-01
5.08115649e-01 -1.93211246e+00 4.53799307e-01 5.25909616e-03
3.98109198e-01 -5.18746264e-02 -1.34518310e-01 1.13672413e-01
-7.39596644e-03 -3.05289179e-01 5.21505289e-02 -9.84273911e-01
8.51303339e-02 -4.04482447e-02 9.25738513e-02 7.29984522e-01
4.55178410e-01 9.52655613e-01 -9.56331849e-01 -5.09968579e-01
4.45095390e-01 5.79035938e-01 -7.77198851e-01 7.92480946e-01
-9.29583386e-02 9.55473781e-01 -4.48375255e-01 6.21248543e-01
6.56852901e-01 -3.57466847e-01 -9.96135846e-02 -3.97241712e-01
1.74191982e-01 2.45578736e-01 -8.43770564e-01 2.52315784e+00
-7.12206304e-01 2.83734173e-01 -7.49313831e-02 -8.83015454e-01
8.75509143e-01 2.14832276e-01 6.03823841e-01 -5.49534082e-01
3.72354805e-01 2.08696127e-01 -4.00745094e-01 -4.05495405e-01
9.17880461e-02 -6.18759133e-02 -3.71946365e-01 3.20021063e-01
3.59849751e-01 -5.17600954e-01 -3.43215108e-01 -1.32259488e-01
8.54220808e-01 9.54879880e-01 1.05292879e-01 1.58795387e-01
2.03753248e-01 -2.84474611e-01 3.97067070e-01 2.73601860e-01
-3.86508465e-01 1.04350924e+00 5.40926158e-01 -6.53675318e-01
-1.09834099e+00 -1.19524729e+00 1.30368546e-01 1.01250577e+00
2.69143850e-01 -2.47678727e-01 -7.11803138e-01 -9.73417342e-01
1.58801645e-01 -5.36230952e-02 -1.04223597e+00 -1.51215672e-01
-5.64706504e-01 -5.48028834e-02 4.17050660e-01 9.05404091e-01
8.00243437e-01 -6.27940893e-01 -8.89161468e-01 -6.92138895e-02
-9.71124172e-02 -1.42046487e+00 -6.89414680e-01 4.08681750e-01
-7.00548232e-01 -1.15835631e+00 -1.06534016e+00 -8.29588175e-01
8.15679789e-01 -7.32664391e-02 1.16137242e+00 -3.48474532e-01
-3.18010062e-01 4.85554039e-01 -2.70898372e-01 -4.17226285e-01
4.13450539e-01 2.94849753e-01 1.48877706e-02 -2.37249002e-01
4.41127807e-01 -5.18677533e-01 -8.03412437e-01 3.83991152e-01
-3.18811923e-01 6.97184578e-02 6.51484311e-01 8.54624271e-01
7.01413691e-01 -4.37809408e-01 2.60808244e-02 -7.81551242e-01
3.08118798e-02 -1.20186415e-02 -4.71734494e-01 -1.49399340e-01
-1.16892152e-01 1.41700923e-01 1.28381878e-01 -4.27354634e-01
-8.47881973e-01 6.50162578e-01 -1.56915724e-01 -7.32682526e-01
-2.81595677e-01 8.61577615e-02 -4.54273701e-01 -7.27501214e-02
5.80999136e-01 -3.52027118e-01 2.67071366e-01 -5.88158488e-01
2.50768941e-02 3.96189511e-01 5.59216321e-01 -6.39859140e-01
9.99247909e-01 4.71881568e-01 1.38142943e-01 -3.16206008e-01
-1.49035954e+00 -4.19801176e-01 -1.29077125e+00 -2.58692950e-01
1.09204805e+00 -1.49928796e+00 -6.33891284e-01 7.59587884e-01
-1.13644326e+00 -7.01854765e-01 -1.02089174e-01 5.36244631e-01
-8.80577743e-01 8.69040564e-02 -6.15458846e-01 -6.37501776e-01
-1.10953256e-01 -1.27735126e+00 1.94890022e+00 -1.52611405e-01
-4.81881499e-01 -7.90086746e-01 9.95412469e-02 5.39506614e-01
3.18091884e-02 8.46519232e-01 3.83088619e-01 -4.05257970e-01
-2.86983520e-01 -6.47339284e-01 -1.37187570e-01 3.34284991e-01
-2.36760825e-02 -6.88306868e-01 -1.10804069e+00 -3.68816674e-01
-1.12735391e-01 -1.05899203e+00 6.89925373e-01 4.43922698e-01
1.15765703e+00 2.75210261e-01 -2.10146099e-01 8.63338947e-01
9.09140944e-01 -4.19665158e-01 3.07927608e-01 2.95486778e-01
1.03522491e+00 8.74664605e-01 6.41838551e-01 4.49721187e-01
6.11038685e-01 8.71565938e-01 5.82734346e-01 -4.70383525e-01
-1.11195102e-01 -7.14687407e-01 2.05263466e-01 2.15163693e-01
-7.00230375e-02 6.21524677e-02 -8.52809429e-01 1.17321126e-01
-1.80817127e+00 -3.20549220e-01 4.23359573e-01 2.20480323e+00
8.69546235e-01 3.49735111e-01 3.60385597e-01 7.73778111e-02
4.99180734e-01 2.27908388e-01 -8.27515304e-01 8.31398144e-02
2.67865747e-01 2.17196003e-01 4.98002499e-01 4.40142840e-01
-1.43272376e+00 8.82661283e-01 5.46360159e+00 1.97308406e-01
-1.09155416e+00 -7.21185002e-03 6.92236602e-01 -5.12982607e-01
2.39921704e-01 -4.25582677e-01 -8.20857823e-01 1.29505321e-01
3.10578465e-01 6.44397318e-01 -5.76289482e-02 9.80326056e-01
-5.79883568e-02 -1.13845415e-01 -1.72196066e+00 1.38770390e+00
2.42821977e-01 -7.01483369e-01 -1.90109923e-01 2.93526232e-01
6.05167449e-01 -1.87016249e-01 1.03692159e-01 4.12756890e-01
8.83499682e-02 -1.17434514e+00 9.39470589e-01 3.69400293e-01
1.14250922e+00 -7.79266357e-01 8.13055694e-01 4.87416029e-01
-9.81748164e-01 1.13597952e-01 1.03386100e-02 -4.17755663e-01
1.80403754e-01 3.88039917e-01 -4.85372424e-01 2.46165842e-01
9.06069398e-01 8.62592041e-01 -5.06316483e-01 6.66655004e-01
-5.50243914e-01 -2.00977981e-01 -3.27881128e-01 1.79879770e-01
3.11219424e-01 1.83966354e-01 8.28116238e-02 8.11964512e-01
5.38873137e-04 -1.93958148e-01 5.44182897e-01 7.21125305e-01
-2.40683883e-01 -3.79713804e-01 -6.27494633e-01 3.52784634e-01
8.50478262e-02 9.54536498e-01 -4.59203005e-01 3.59187946e-02
-2.55621284e-01 1.34472191e+00 5.16244233e-01 1.71190247e-01
-7.84507275e-01 -2.23891422e-01 5.94689727e-01 4.25557226e-01
2.50762969e-01 -4.40539956e-01 -2.37052843e-01 -1.28777957e+00
2.19867140e-01 -4.52424347e-01 8.40711668e-02 -7.71467209e-01
-1.43029237e+00 4.88222569e-01 2.34516740e-01 -1.30605721e+00
-4.29355323e-01 -1.06150186e+00 -2.57904261e-01 9.57207203e-01
-1.49348533e+00 -1.39002347e+00 -6.24007583e-01 6.12662077e-01
4.08292592e-01 1.35719940e-01 7.49425650e-01 3.19144100e-01
-4.00372356e-01 8.01793158e-01 -6.55892670e-01 6.23085678e-01
1.17914093e+00 -1.29503882e+00 4.91953701e-01 3.88836920e-01
-4.00439918e-01 3.46178234e-01 3.45446587e-01 -5.19897401e-01
-1.19156528e+00 -1.06286573e+00 8.23680580e-01 -8.42545986e-01
1.45434648e-01 -9.71449614e-01 -2.97564924e-01 8.14090073e-01
-2.49619544e-01 5.26619673e-01 6.16957486e-01 2.69389182e-01
-6.12403452e-01 -2.81768236e-02 -1.18607211e+00 3.45415920e-01
1.41446078e+00 -6.32185757e-01 -4.10994053e-01 4.36024606e-01
8.45420897e-01 -9.56715405e-01 -6.38449788e-01 5.57923675e-01
7.53753662e-01 -8.11109126e-01 1.12941825e+00 -5.09518564e-01
6.99897051e-01 -1.18219636e-01 -2.88237244e-01 -1.06927884e+00
-4.31703180e-02 -3.44777673e-01 -1.77282058e-02 6.02471590e-01
2.32382968e-01 -2.17773274e-01 1.25614345e+00 5.73846519e-01
-4.13542278e-02 -1.00749338e+00 -7.73144603e-01 -6.93998158e-01
1.78981766e-01 -3.56513143e-01 3.01303744e-01 4.51025218e-01
-3.76750201e-01 6.28073156e-01 -5.32379806e-01 5.58255799e-03
8.47294986e-01 1.37758048e-04 1.28839147e+00 -1.25281203e+00
-2.63956815e-01 -8.25813264e-02 -6.33887768e-01 -1.62441051e+00
4.21465039e-01 -6.02621675e-01 2.65230149e-01 -1.18266773e+00
1.17001653e-01 -4.93508205e-02 1.36213452e-01 4.48132575e-01
-1.00022294e-01 4.28709626e-01 -4.19785678e-02 2.61309445e-02
-5.87996006e-01 7.70344913e-01 1.54005527e+00 -1.18514068e-01
-1.83650449e-01 -4.71867695e-02 -3.40771347e-01 8.45889151e-01
4.60047603e-01 -3.00082207e-01 -4.98285770e-01 -7.05749333e-01
-5.55647770e-04 -4.58288118e-02 6.97831094e-01 -1.10503137e+00
1.86142977e-03 2.56401062e-01 1.03899229e+00 -9.46920037e-01
8.44628870e-01 -7.31811225e-01 -4.99203384e-01 4.98153299e-01
-4.85896230e-01 -1.79070726e-01 9.17486399e-02 5.25894165e-01
-1.01458699e-01 3.03008348e-01 8.01259160e-01 -4.29230064e-01
-4.98454899e-01 7.92898595e-01 3.53677779e-01 2.30102509e-01
1.01579666e+00 -4.25468713e-01 2.42268890e-01 -4.96203989e-01
-7.68155694e-01 4.58890826e-01 5.33151090e-01 4.13969249e-01
6.62793517e-01 -1.37362909e+00 -4.99374866e-01 3.33368361e-01
2.56473899e-01 8.23603809e-01 2.01777399e-01 6.03870988e-01
-6.04968190e-01 1.86648861e-01 -3.47187459e-01 -8.81823957e-01
-7.54871428e-01 3.63019556e-01 6.22693837e-01 -4.60595608e-01
-3.94686282e-01 1.06342018e+00 4.70078170e-01 -1.07270777e+00
7.29669869e-01 -1.11281082e-01 2.12539569e-01 -1.63125232e-01
3.50272238e-01 9.47863609e-03 -2.54427921e-03 -6.18915200e-01
-3.96815360e-01 9.86393392e-01 -8.78912807e-02 -1.04912825e-01
1.48844075e+00 -2.43166052e-02 2.81247288e-01 5.94254375e-01
1.85541451e+00 -3.70262712e-01 -1.89774942e+00 -2.73275584e-01
-5.16508222e-01 -5.11795580e-01 1.22205997e-02 -7.95293689e-01
-1.14984834e+00 1.08954024e+00 6.32612586e-01 -8.41491044e-01
8.72250497e-01 7.44066238e-02 8.55960011e-01 3.90377194e-01
4.66743857e-01 -1.11963022e+00 7.73787856e-01 6.16390705e-01
9.46849287e-01 -1.55259454e+00 1.46056991e-02 -4.16672200e-01
-4.39834476e-01 9.03630435e-01 1.18691611e+00 -5.19396007e-01
7.65848279e-01 2.13769913e-01 9.13594216e-02 -2.05281287e-01
-3.56309414e-01 -2.16444075e-01 3.08987409e-01 7.00253844e-01
5.47513247e-01 -1.80736706e-01 2.38170832e-01 6.18560493e-01
-3.01934123e-01 4.06883508e-02 -7.63779581e-02 1.22830093e+00
-7.25918412e-02 -9.38026488e-01 -3.48944902e-01 1.63789168e-02
-1.43088669e-01 3.91558081e-01 -6.06930196e-01 9.76672947e-01
3.81911963e-01 5.81730425e-01 2.13175833e-01 -6.91146612e-01
6.64604247e-01 -4.74601164e-02 9.51630533e-01 -9.79157686e-01
-3.02629560e-01 2.12006792e-01 -2.12975293e-01 -9.40721273e-01
-3.64130735e-01 -3.53101969e-01 -1.03456891e+00 -4.61861677e-02
-7.46596679e-02 -3.28972280e-01 5.59084952e-01 9.92829859e-01
2.58157790e-01 4.50317025e-01 5.51190972e-01 -1.73354244e+00
-5.70945561e-01 -9.93993282e-01 -5.03820598e-01 5.39589345e-01
4.43025559e-01 -1.17480183e+00 -1.94354460e-01 -1.81334302e-01]
|
[6.9361572265625, -0.9661102294921875]
|
416b7faf-8157-4de8-a9a0-dc02337becab
|
a-practical-framework-for-unsupervised
|
2304.01864
| null |
https://arxiv.org/abs/2304.01864v1
|
https://arxiv.org/pdf/2304.01864v1.pdf
|
A Practical Framework for Unsupervised Structure Preservation Medical Image Enhancement
|
Medical images are extremely valuable for supporting medical diagnoses. However, in practice, low-quality (LQ) medical images, such as images that are hazy/blurry, have uneven illumination, or are out of focus, among others, are often obtained during data acquisition. This leads to difficulties in the screening and diagnosis of medical diseases. Several generative adversarial networks (GAN)-based image enhancement methods have been proposed and have shown promising results. However, there is a quality-originality trade-off among these methods in the sense that they produce visually pleasing results but lose the ability to preserve originality, especially the structural inputs. Moreover, to our knowledge, there is no objective metric in evaluating the structure preservation of medical image enhancement methods in unsupervised settings due to the unavailability of paired ground-truth data. In this study, we propose a framework for practical unsupervised medical image enhancement that includes (1) a non-reference objective evaluation of structure preservation for medical image enhancement tasks called Laplacian structural similarity index measure (LaSSIM), which is based on SSIM and the Laplacian pyramid, and (2) a novel unsupervised GAN-based method called Laplacian medical image enhancement (LaMEGAN) to support the improvement of both originality and quality from LQ images. The LaSSIM metric does not require clean reference images and has been shown to be superior to SSIM in capturing image structural changes under image degradations, such as strong blurring on different datasets. The experiments demonstrated that our LaMEGAN achieves a satisfactory balance between quality and originality, with robust structure preservation performance while generating compelling visual results with very high image quality scores. The code will be made available at https://github.com/AillisInc/USPMIE.
|
['Hitoshi Iyatomi', 'Atsushi Fukuda', 'Quan Huu Cap']
|
2023-04-04
| null | null | null | null |
['medical-image-enhancement', 'image-enhancement']
|
['computer-vision', 'computer-vision']
|
[ 5.99382401e-01 -1.52646631e-01 1.52732506e-01 -1.50197119e-01
-7.92878449e-01 -2.92805493e-01 1.99603721e-01 7.72956684e-02
-1.28689244e-01 7.22719431e-01 3.14063102e-01 -1.38477966e-01
-3.01955760e-01 -6.77277446e-01 -3.70186627e-01 -1.14565301e+00
1.47696018e-01 -1.64284691e-01 3.27583961e-02 -1.56838223e-01
-1.70711607e-01 3.54743510e-01 -1.37652040e+00 1.91604599e-01
1.44801331e+00 8.52048039e-01 3.10168505e-01 6.12059414e-01
3.05444986e-01 8.63880694e-01 -6.26902282e-01 -4.26551074e-01
1.81219324e-01 -7.96736777e-01 -6.09622717e-01 2.77828425e-01
3.01679045e-01 -3.82116973e-01 -3.55253190e-01 1.35127199e+00
7.15647340e-01 -3.30938213e-02 5.29632568e-01 -1.12474501e+00
-8.45371246e-01 1.21230423e-01 -6.00873053e-01 2.82483816e-01
1.14917234e-01 3.11909616e-01 5.33308983e-01 -6.48550451e-01
5.29852152e-01 9.42723155e-01 4.59446460e-01 4.51805949e-01
-1.23918378e+00 -4.75971520e-01 -4.58458602e-01 2.02285126e-01
-1.09026790e+00 -3.78797472e-01 8.82166624e-01 -3.17760676e-01
1.96179375e-01 5.10966241e-01 4.14574623e-01 8.07535589e-01
4.59285796e-01 6.94277704e-01 1.44394255e+00 -3.91205460e-01
1.25705644e-01 1.96611375e-01 -3.27232182e-01 7.52245784e-01
3.40141654e-01 3.61488700e-01 -2.40525007e-01 -3.86217944e-02
7.80287743e-01 2.35675633e-01 -7.46344209e-01 -1.89067453e-01
-1.23100019e+00 5.45766234e-01 6.30603135e-01 6.41778827e-01
-4.64696825e-01 -2.59063989e-01 1.18543088e-01 3.12193960e-01
3.43924046e-01 4.84168947e-01 1.34962082e-01 1.36121392e-01
-9.54756796e-01 -1.95961013e-01 2.07194805e-01 4.35446024e-01
4.68934208e-01 1.57910168e-01 -4.41460311e-01 8.66188407e-01
-2.93854587e-02 6.78294718e-01 6.01810038e-01 -1.00705719e+00
1.58782154e-01 4.39362854e-01 4.17608805e-02 -1.31093347e+00
-1.70865521e-01 -7.97084689e-01 -1.45724273e+00 4.77420092e-01
2.01479435e-01 1.06121160e-01 -8.90297353e-01 1.87130189e+00
2.97113687e-01 5.23893535e-02 2.76751548e-01 1.02961850e+00
9.88302052e-01 5.66938818e-01 -4.43690550e-03 -3.96852702e-01
1.21069121e+00 -8.79284203e-01 -1.13071668e+00 -1.37896284e-01
2.84401178e-01 -9.35265005e-01 1.15206742e+00 3.82954717e-01
-1.46344686e+00 -6.65654600e-01 -1.10691285e+00 1.20681360e-01
-2.42385315e-03 5.03629297e-02 3.31265628e-01 7.37092018e-01
-1.11201465e+00 5.55958867e-01 -8.92810941e-01 -6.78863078e-02
4.47592229e-01 1.02424838e-01 -3.92142981e-01 -5.36881447e-01
-1.13854635e+00 7.38162041e-01 1.90981850e-01 1.36595190e-01
-7.83603191e-01 -7.11139619e-01 -8.29183698e-01 -6.38617277e-02
1.62114784e-01 -7.38614202e-01 7.57351935e-01 -1.24804533e+00
-1.21117413e+00 7.73842990e-01 -1.79603603e-02 -2.09481671e-01
5.55375636e-01 6.72842413e-02 -6.78893387e-01 5.70236981e-01
4.45363559e-02 5.03800094e-01 9.41729009e-01 -1.43553841e+00
-2.58127898e-01 -2.93884218e-01 -9.34217572e-02 1.91498354e-01
-3.50417495e-01 -1.30914114e-02 -3.74110997e-01 -1.07130206e+00
5.74725531e-02 -7.31372714e-01 -1.63067922e-01 1.48491710e-01
-4.40230757e-01 5.57284117e-01 7.62431979e-01 -1.03032732e+00
1.15733433e+00 -2.30823898e+00 -3.71632725e-02 6.95845159e-03
3.66448283e-01 5.16298354e-01 -2.60941476e-01 1.20887756e-01
-6.40610382e-02 1.06061548e-01 -6.77668571e-01 5.16791083e-03
-4.06161904e-01 1.00625053e-01 1.50653899e-01 5.97725809e-01
1.16544738e-01 1.10356188e+00 -1.11174119e+00 -6.60378456e-01
3.56124580e-01 7.04827726e-01 -2.48892054e-01 3.39273214e-01
3.83617461e-01 8.90820682e-01 -2.41293073e-01 7.56351411e-01
9.30801928e-01 -4.07989234e-01 -1.08287968e-01 -5.24795830e-01
1.76173776e-01 -2.12868437e-01 -9.56927538e-01 1.33805954e+00
-3.28882307e-01 5.63868046e-01 2.38861024e-01 -8.85216773e-01
6.13211095e-01 5.77659309e-01 6.44157767e-01 -8.91753912e-01
1.15802467e-01 2.09506109e-01 9.01276320e-02 -6.42595172e-01
2.23343059e-01 -3.78570467e-01 3.50189924e-01 3.14689189e-01
-2.37606078e-01 -6.43425956e-02 2.80799747e-01 1.49354711e-01
1.04587328e+00 -3.98701906e-01 1.78535268e-01 -2.57529281e-02
5.37600875e-01 -3.72334599e-01 5.91735542e-01 5.22245109e-01
-3.54144365e-01 1.07414126e+00 1.13065556e-01 6.47769496e-02
-1.07660902e+00 -1.21952307e+00 -2.46647045e-01 3.17719966e-01
4.75060552e-01 2.04736441e-02 -6.70952857e-01 -3.86585683e-01
-4.57358181e-01 4.73578066e-01 -4.71988022e-01 -3.61072004e-01
-3.34231138e-01 -8.89676332e-01 4.97344375e-01 4.05965269e-01
7.37099648e-01 -1.07469773e+00 -5.98497391e-01 9.50967297e-02
-5.88659704e-01 -1.10194528e+00 -8.27229261e-01 -2.88582236e-01
-9.56077635e-01 -1.13922119e+00 -1.10632455e+00 -6.91187978e-01
1.06499588e+00 3.96223485e-01 9.76891577e-01 2.59357899e-01
-3.96929979e-01 3.10849249e-01 -4.19259489e-01 -2.92954922e-01
-7.79300809e-01 -6.00948155e-01 -1.32734105e-01 2.45539337e-01
-2.66465694e-01 -5.27069926e-01 -1.08458996e+00 4.83207136e-01
-1.56792188e+00 1.86165899e-01 8.55211854e-01 1.12001920e+00
6.92089736e-01 5.86719096e-01 4.17058587e-01 -7.67523110e-01
5.76212406e-01 -1.02773353e-01 -2.09230915e-01 3.05261225e-01
-8.07099819e-01 -1.39512017e-01 5.27984381e-01 -2.94446975e-01
-1.24888706e+00 -3.65005106e-01 -2.25150868e-01 -4.16343510e-01
-1.76854253e-01 5.50035715e-01 -1.24742948e-01 -1.64125517e-01
6.47159815e-01 4.50775653e-01 3.08941215e-01 -3.19626480e-01
3.50233763e-02 6.41060233e-01 8.15072596e-01 -4.00030836e-02
8.85887563e-01 6.27770901e-01 8.06183834e-03 -8.00632298e-01
-6.64796591e-01 -4.76987958e-01 -2.16488346e-01 -1.82048529e-01
8.68478417e-01 -7.76409030e-01 -2.03353643e-01 7.46491075e-01
-7.76547730e-01 -2.48097450e-01 -2.97986329e-01 4.85929519e-01
-3.46562266e-01 8.33056033e-01 -7.99788177e-01 -6.47262454e-01
-6.40393019e-01 -1.27997899e+00 6.87344730e-01 3.88879329e-01
1.19787060e-01 -1.07400632e+00 -1.25744328e-01 5.88259697e-01
6.86882317e-01 6.96598887e-01 1.06007648e+00 1.40117124e-01
-3.64706725e-01 -4.86394577e-02 -2.12473020e-01 7.46722996e-01
6.19433582e-01 -2.86934018e-01 -8.14848900e-01 -6.49407148e-01
2.90552884e-01 -1.23109259e-01 6.33076549e-01 7.27498412e-01
1.09605825e+00 -3.99497926e-01 5.74161522e-02 7.90553689e-01
1.49538517e+00 3.77511710e-01 1.09812462e+00 1.02110595e-01
4.69789386e-01 5.44167280e-01 6.38133585e-01 1.24878362e-01
-1.91822782e-01 7.11806357e-01 4.20448154e-01 -9.87797379e-01
-6.02404654e-01 -6.74766153e-02 3.10408980e-01 7.58723080e-01
-1.27498418e-01 -2.95504332e-01 -5.82973123e-01 6.20606065e-01
-1.45969594e+00 -8.39223564e-01 -1.10460117e-01 2.26470208e+00
9.51254308e-01 -2.15041175e-01 -1.66582391e-01 3.31985503e-01
8.56341600e-01 -4.67392541e-02 -6.10803604e-01 -1.32200085e-02
-3.79317850e-01 1.35846645e-01 4.05104518e-01 3.63563716e-01
-9.69764650e-01 2.79955745e-01 5.69455290e+00 7.87972867e-01
-1.16831052e+00 3.67664933e-01 9.63045359e-01 1.76911160e-01
-3.43614817e-01 -3.91736776e-01 -1.26312012e-02 6.29208922e-01
5.55725813e-01 -1.11074865e-01 2.22533673e-01 3.35922539e-01
4.76649791e-01 -1.43316865e-01 -6.04669452e-01 1.10157275e+00
9.65080187e-02 -9.78923500e-01 -9.01450142e-02 6.51982576e-02
9.62808013e-01 -3.90858263e-01 4.94627744e-01 -3.51025879e-01
-5.21238856e-02 -1.05799150e+00 3.89936984e-01 4.87725645e-01
1.17756510e+00 -7.20134258e-01 9.64606404e-01 7.90796354e-02
-8.47251892e-01 9.58452150e-02 -1.57526091e-01 5.95326185e-01
3.04064721e-01 9.36152816e-01 -4.16659206e-01 7.57333338e-01
7.36655593e-01 5.32669187e-01 -5.89977741e-01 1.26505470e+00
-3.72385591e-01 5.71055532e-01 1.94897011e-01 6.60751402e-01
7.35929608e-02 -3.00243288e-01 7.81833231e-01 1.00260925e+00
5.06184936e-01 2.34363720e-01 -1.52057081e-01 9.16713119e-01
1.04719199e-01 8.34816098e-02 -4.39420402e-01 3.69336270e-02
1.22111114e-02 1.16631913e+00 -7.87950933e-01 -2.32023567e-01
-2.80836850e-01 1.18297768e+00 -5.17604649e-01 5.43215930e-01
-7.72401512e-01 -2.98827559e-01 3.44003826e-01 2.58003354e-01
5.28832339e-02 1.75416619e-01 -1.93204135e-01 -1.15347791e+00
2.94746552e-02 -1.20640349e+00 3.64132583e-01 -8.96885753e-01
-1.22880495e+00 8.03773999e-01 -1.53570771e-01 -1.52392983e+00
7.90628046e-02 -2.71272033e-01 -5.06547570e-01 7.14348555e-01
-1.64232445e+00 -1.01479971e+00 -7.04843700e-01 6.43584430e-01
3.26333523e-01 5.18897437e-02 5.39144039e-01 5.47619879e-01
-4.16280091e-01 6.94865048e-01 4.26745445e-01 -6.66457787e-02
7.93364882e-01 -1.24489510e+00 -2.81991124e-01 1.16835558e+00
-1.22216612e-01 3.93082529e-01 7.83469260e-01 -5.73535025e-01
-1.05654585e+00 -1.23218286e+00 5.17358720e-01 -3.26441936e-02
1.23297840e-01 2.13632181e-01 -1.08793473e+00 1.26691923e-01
3.62690777e-01 -2.43403460e-03 6.47967815e-01 -5.96113801e-01
1.33611739e-01 -2.34049588e-01 -1.43759143e+00 5.35773218e-01
6.04605913e-01 -2.45753467e-01 -2.86641628e-01 1.47698238e-01
4.76415008e-01 -3.91250730e-01 -1.03099418e+00 6.21462405e-01
2.86987662e-01 -1.18173289e+00 1.11222982e+00 1.47028863e-02
7.39653945e-01 -6.04352117e-01 7.37935528e-02 -1.27995539e+00
-4.44286287e-01 -4.02045727e-01 3.69725004e-02 1.15964592e+00
2.46998355e-01 -5.94849646e-01 4.94161457e-01 3.46460402e-01
-7.16533586e-02 -7.26477563e-01 -5.24586141e-01 -8.45718086e-01
-2.66052753e-01 1.52786039e-02 3.73226404e-01 9.04375434e-01
-3.03114325e-01 -4.34180945e-02 -4.57746297e-01 2.57545650e-01
8.03077817e-01 4.34061848e-02 3.94246340e-01 -7.31283128e-01
-3.76891583e-01 -3.18425715e-01 -4.59077954e-01 -5.14226377e-01
-3.49927992e-01 -8.48242462e-01 4.03681621e-02 -1.65037537e+00
4.82674927e-01 -3.78331631e-01 -6.30927265e-01 4.27011997e-01
-4.24965620e-01 6.97405696e-01 9.08487290e-02 4.68061179e-01
-3.44649106e-01 5.71030319e-01 1.64509416e+00 -4.76335734e-01
-1.98072210e-01 -9.09951050e-03 -8.19474697e-01 4.57565606e-01
7.56040454e-01 -3.58192712e-01 -6.06056929e-01 -3.13086987e-01
-1.37900993e-01 2.14813486e-01 4.41256642e-01 -1.16641796e+00
8.35638214e-03 -7.44061396e-02 4.01449382e-01 -3.24226499e-01
1.05429627e-01 -6.60702169e-01 5.56409597e-01 7.82262444e-01
-2.03415319e-01 -4.63800021e-02 2.98880786e-02 4.07592386e-01
-6.75207257e-01 -1.43533185e-01 1.25767982e+00 -1.06145151e-01
-4.61369336e-01 2.67612964e-01 -1.52215138e-01 -1.04183055e-01
8.63059878e-01 -4.37377661e-01 -2.41477430e-01 -6.59651160e-01
-4.79046166e-01 -1.18852399e-01 6.91755772e-01 4.46147770e-02
9.67413068e-01 -1.31263888e+00 -9.56785142e-01 6.98889121e-02
5.57855591e-02 -9.88889486e-02 6.88032091e-01 1.33289957e+00
-6.02777839e-01 8.33543539e-02 -3.47588152e-01 -6.03369594e-01
-1.44046640e+00 7.78816402e-01 3.39453191e-01 -5.98618984e-01
-7.10363746e-01 4.83600765e-01 7.08542585e-01 3.46309096e-02
1.94176324e-02 -9.17375609e-02 -6.79921508e-02 -3.96658301e-01
7.89451301e-01 3.50392431e-01 2.93442100e-01 -7.08679199e-01
-1.15012594e-01 5.21320283e-01 6.08052611e-02 1.27038926e-01
1.19184160e+00 -4.14144754e-01 -1.66125491e-01 -6.04679845e-02
1.04961538e+00 7.48965144e-02 -1.27102923e+00 -2.57288098e-01
-4.83502746e-01 -7.18171537e-01 3.63076389e-01 -1.01227212e+00
-1.57232237e+00 7.28644133e-01 1.16004920e+00 1.75114796e-01
1.87232637e+00 -1.41251370e-01 9.87211108e-01 -2.68938273e-01
7.08949342e-02 -7.79084682e-01 4.10106093e-01 -1.45692527e-01
9.75210905e-01 -1.21404517e+00 -4.26200405e-02 -5.49943149e-01
-6.31637692e-01 7.41825223e-01 2.24022821e-01 2.98705399e-01
2.92062432e-01 3.01297754e-01 3.29689473e-01 -1.98807746e-01
-2.42422625e-01 -2.25182325e-01 4.42613512e-01 7.89640605e-01
3.80882025e-01 -5.36048859e-02 -3.96389157e-01 1.90577731e-01
1.29402995e-01 -4.00060304e-02 5.07119119e-01 8.12404096e-01
-1.16751477e-01 -9.74439621e-01 -6.50364339e-01 3.49173963e-01
-8.04717779e-01 -1.19128786e-01 2.28946228e-02 4.68916416e-01
2.19063312e-01 1.22780228e+00 -3.18239212e-01 -1.57659128e-01
2.52400011e-01 -4.77014571e-01 4.78397310e-01 -2.62825072e-01
-2.55936950e-01 2.49554440e-01 -2.68594086e-01 -4.33803707e-01
-6.84950233e-01 -3.36439282e-01 -9.06533062e-01 -2.64688045e-01
-3.07296067e-01 4.46852855e-02 3.87437314e-01 6.96071982e-01
2.59437382e-01 7.90701628e-01 7.88816571e-01 -4.70591605e-01
-5.07970937e-02 -8.03127646e-01 -8.53476703e-01 8.96424711e-01
4.68175769e-01 -4.68555838e-01 -4.01117951e-01 4.25700188e-01]
|
[13.460837364196777, -2.31612229347229]
|
904dbf92-ff58-477e-ad4d-251dbef09629
|
t5ql-taming-language-models-for-sql
|
2209.10254
| null |
https://arxiv.org/abs/2209.10254v1
|
https://arxiv.org/pdf/2209.10254v1.pdf
|
T5QL: Taming language models for SQL generation
|
Automatic SQL generation has been an active research area, aiming at streamlining the access to databases by writing natural language with the given intent instead of writing SQL. Current SOTA methods for semantic parsing depend on LLMs to achieve high predictive accuracy on benchmark datasets. This reduces their applicability, since LLMs requires expensive GPUs. Furthermore, SOTA methods are ungrounded and thus not guaranteed to always generate valid SQL. Here we propose T5QL, a new SQL generation method that improves the performance in benchmark datasets when using smaller LMs, namely T5-Base, by 13pp when compared against SOTA methods. Additionally, T5QL is guaranteed to always output valid SQL using a context-free grammar to constrain SQL generation. Finally, we show that dividing semantic parsing in two tasks, candidate SQLs generation and candidate re-ranking, is a promising research avenue that can reduce the need for large LMs.
|
['António Alegria', 'Hugo Veiga', 'David Aparício', 'Samuel Arcadinho']
|
2022-09-21
| null | null | null | null |
['text-to-sql']
|
['computer-code']
|
[ 2.53373563e-01 2.64319450e-01 -2.71231443e-01 -6.83716476e-01
-9.26558614e-01 -5.12191415e-01 3.34829569e-01 4.33700979e-01
-1.89331144e-01 6.27810419e-01 -4.09450904e-02 -7.05946147e-01
1.29226804e-01 -1.64249051e+00 -1.01843047e+00 1.48203731e-01
4.02476460e-01 7.99390912e-01 5.93927681e-01 -2.35218033e-01
3.66034418e-01 1.91861913e-01 -1.94750333e+00 7.33739138e-01
1.26893020e+00 1.08571494e+00 1.74392596e-01 6.08961821e-01
-1.04916954e+00 8.12075555e-01 -7.72464752e-01 -7.53395379e-01
7.40179494e-02 -4.40680355e-01 -1.04799128e+00 -5.21128774e-01
5.89058399e-01 -1.93377852e-01 4.36908960e-01 8.48034859e-01
1.90818429e-01 -2.75400966e-01 -1.73447821e-02 -1.09667993e+00
-1.83180362e-01 9.69416916e-01 -1.86150912e-02 -5.57363987e-01
6.17833912e-01 -2.63043106e-01 1.29105031e+00 -8.16844940e-01
6.82881176e-01 1.46677911e+00 4.22139913e-01 6.31581426e-01
-1.16265249e+00 -4.26415026e-01 1.69028416e-02 4.34262305e-02
-1.07100546e+00 -3.00190657e-01 4.14570630e-01 -3.51139046e-02
1.22041786e+00 9.41450536e-01 4.52879250e-01 7.35797882e-01
-6.21245578e-02 7.90637553e-01 1.00847042e+00 -6.99836671e-01
4.80383724e-01 2.16122374e-01 4.77013916e-01 8.44081819e-01
6.23777747e-01 -2.80263901e-01 -9.20974255e-01 -5.26734352e-01
3.28025758e-01 -1.68891340e-01 3.35550994e-01 -4.03870940e-01
-9.80075836e-01 9.31671739e-01 -6.82933778e-02 -1.49329916e-01
4.01303098e-02 1.52865872e-01 6.44801378e-01 2.14441419e-01
-4.16696444e-02 8.58264685e-01 -7.52746224e-01 -2.66378045e-01
-7.35895514e-01 6.71903312e-01 1.27275014e+00 1.39148450e+00
9.24197078e-01 -2.00500503e-01 -2.10520342e-01 6.91827774e-01
6.35895431e-02 6.98928118e-01 3.29356581e-01 -1.02258420e+00
6.73341572e-01 1.19941473e+00 -1.22295357e-02 -7.98580050e-01
-2.24329218e-01 -7.43136927e-02 -2.93024510e-01 -8.22378024e-02
3.68710250e-01 8.77576023e-02 -6.99392438e-01 1.57172787e+00
1.86938018e-01 -3.69593769e-01 3.90646636e-01 4.13001716e-01
6.59976840e-01 7.17534006e-01 8.78815800e-02 -1.50686622e-01
1.43293881e+00 -8.19124103e-01 -5.05501568e-01 -3.14547896e-01
8.94743323e-01 -7.85306752e-01 1.79213309e+00 5.42464197e-01
-1.15456569e+00 -4.81840670e-01 -9.55065906e-01 -2.39425033e-01
-5.72690666e-01 1.44329816e-02 1.18166947e+00 1.01664138e+00
-8.65432501e-01 4.14610654e-01 -9.37862098e-01 -4.41803098e-01
1.20411910e-01 3.81606191e-01 -1.18196728e-02 2.97423024e-02
-1.18938589e+00 4.73880321e-01 7.18349576e-01 -2.44045973e-01
-2.87418455e-01 -7.13162124e-01 -8.34407687e-01 1.29509196e-01
8.99481833e-01 -8.14099371e-01 1.15011394e+00 -3.26947093e-01
-1.65135765e+00 8.33049238e-01 -5.61936855e-01 -6.66844785e-01
4.55171168e-01 -4.97399002e-01 -2.54435509e-01 -2.19666973e-01
3.22994411e-01 5.95520914e-01 3.19232225e-01 -9.27318513e-01
-7.81170487e-01 -4.40888196e-01 1.79518849e-01 -7.74086341e-02
-4.19511795e-01 -8.40330422e-02 -8.39994609e-01 -4.84383166e-01
3.37724417e-01 -8.78438175e-01 -3.94380242e-01 -4.38777953e-01
-7.04823911e-01 -3.49637002e-01 5.13986528e-01 -2.90315330e-01
1.53559411e+00 -1.67408049e+00 -1.83546945e-01 5.20050883e-01
-1.07269555e-01 2.60117978e-01 2.17695951e-01 3.68317425e-01
3.58375907e-01 4.93830085e-01 -2.25083470e-01 1.59963131e-01
1.99422523e-01 3.28920960e-01 -7.45417058e-01 -6.65359199e-01
2.62087107e-01 8.44294369e-01 -5.77280760e-01 -6.92726612e-01
9.57621485e-02 -2.85662174e-01 -1.09426117e+00 5.07710695e-01
-9.97213721e-01 -2.53388673e-01 -5.17879426e-01 6.58077240e-01
8.35137784e-01 -1.95444301e-01 6.01521611e-01 -9.47068408e-02
-1.09552786e-01 6.79735005e-01 -1.35007966e+00 1.68334961e+00
-8.41100097e-01 -8.99480805e-02 -4.93136883e-01 -6.71786249e-01
1.25566065e+00 -3.20525885e-01 1.76547527e-01 -8.25306356e-01
-4.97260332e-01 6.52405918e-01 -5.48550546e-01 -4.88607615e-01
5.69326758e-01 3.02572638e-01 -6.24886155e-01 3.71783286e-01
-3.04999202e-01 -2.28123263e-01 6.99253380e-01 2.31189042e-01
1.07446146e+00 2.64177978e-01 1.41128197e-01 -1.71268404e-01
7.36000657e-01 5.07665575e-01 6.58496559e-01 1.13978887e+00
6.27485752e-01 3.84909540e-01 9.45130289e-01 -4.42118138e-01
-7.53238499e-01 -1.10777056e+00 -1.07407935e-01 1.33638835e+00
8.48377571e-02 -1.01665509e+00 -1.04062808e+00 -8.71935308e-01
1.16187073e-02 1.11936164e+00 -1.68695115e-02 -1.81067847e-02
-1.06177843e+00 -7.87500441e-01 6.81459665e-01 4.82671469e-01
4.83331710e-01 -9.86559868e-01 -9.45538342e-01 3.51544946e-01
-2.88213581e-01 -1.20628762e+00 -9.67070088e-02 3.65343869e-01
-1.00803161e+00 -1.21296418e+00 4.60957550e-02 -6.42274737e-01
5.92223942e-01 -2.78405190e-01 1.46614194e+00 2.93056846e-01
-3.53490800e-01 -6.68124482e-02 -2.87836939e-01 -5.95921159e-01
-6.98262334e-01 5.31605363e-01 -6.05267346e-01 -4.20052528e-01
6.13006949e-01 -1.16427138e-01 -1.14888698e-01 1.67150185e-01
-1.02044845e+00 3.90271902e-01 4.80890095e-01 9.11234736e-01
7.96764135e-01 -8.28440934e-02 5.35896659e-01 -1.75742352e+00
4.83595848e-01 -7.11939707e-02 -1.09856784e+00 6.58224881e-01
-1.01545322e+00 7.01954961e-01 9.22030568e-01 2.11284816e-01
-1.19868433e+00 1.64325431e-01 -3.29552591e-01 2.16407895e-01
-6.98946267e-02 4.93798673e-01 -5.23963213e-01 3.59582573e-01
7.23456979e-01 2.41995260e-01 3.87732917e-03 -5.28585732e-01
3.84983242e-01 3.87273192e-01 3.84997070e-01 -9.76660311e-01
4.95952308e-01 3.09220970e-01 1.11004166e-01 -3.94557923e-01
-7.98173010e-01 -1.24244414e-01 -2.84211457e-01 3.77237111e-01
5.07559717e-01 -4.92885947e-01 -8.72085810e-01 3.09946656e-01
-1.25835657e+00 -4.61518839e-02 -1.61256030e-01 1.67737133e-03
-6.05656207e-01 2.65379369e-01 -5.35692990e-01 -6.01873934e-01
-6.58665895e-01 -1.27057457e+00 1.16144872e+00 -6.36883602e-02
-5.15509009e-01 -6.24303341e-01 -3.23942393e-01 5.13090611e-01
3.32305342e-01 7.75398165e-02 1.59214473e+00 -7.38387644e-01
-1.03991234e+00 -1.90653995e-01 -2.11971670e-01 2.11740524e-01
3.04438751e-02 3.70727032e-02 -6.05317175e-01 2.31699705e-01
-1.34897918e-01 -3.28756392e-01 6.24510169e-01 -1.11845106e-01
1.56007743e+00 -4.82706428e-01 -3.79516840e-01 7.22020507e-01
1.48617351e+00 3.53527904e-01 5.89192629e-01 5.15379369e-01
5.59615433e-01 7.40043342e-01 1.15859377e+00 6.04110837e-01
5.41061759e-01 7.20815122e-01 3.15651387e-01 1.05754673e-01
-4.19521965e-02 -6.30554914e-01 3.20166141e-01 7.10093141e-01
4.20875996e-01 -2.07718268e-01 -1.24377215e+00 2.09269658e-01
-1.93438435e+00 -4.69101518e-01 -5.93389750e-01 2.43145466e+00
1.10989523e+00 5.01788020e-01 -6.44562915e-02 2.12124765e-01
3.59916896e-01 -1.86501369e-01 -4.84786958e-01 -6.25926077e-01
-2.09597200e-01 8.12981665e-01 5.64834952e-01 5.60728312e-01
-7.41209269e-01 1.34271836e+00 6.18459415e+00 7.46284604e-01
-1.05631340e+00 -1.66893348e-01 4.69859183e-01 9.76005569e-02
-7.80940354e-01 4.25088018e-01 -1.39168966e+00 3.46784264e-01
1.01961458e+00 -3.55911583e-01 1.66068912e-01 1.24241066e+00
-2.61967573e-02 -2.82784581e-01 -1.17477453e+00 5.42115748e-01
-1.51224375e-01 -1.63898683e+00 5.63257635e-01 -4.08673853e-01
2.69561023e-01 -5.97369492e-01 -2.18141243e-01 4.30916190e-01
2.53065467e-01 -9.01943803e-01 6.78218782e-01 3.48740369e-01
7.46523559e-01 -9.96836305e-01 7.64343917e-01 1.93495303e-01
-1.08528376e+00 6.44884557e-02 -4.33603823e-01 2.59817421e-01
-1.27495676e-01 7.28794217e-01 -1.20665193e+00 5.20397127e-01
8.13535810e-01 2.21357420e-01 -9.84902382e-01 4.36720371e-01
-5.82937859e-02 5.32754242e-01 -2.68611610e-01 -4.02649283e-01
-1.69566184e-01 4.72867489e-03 1.30699411e-01 1.22226834e+00
3.71462792e-01 -2.78199792e-01 3.06389779e-01 9.65870678e-01
1.46450087e-01 4.77145135e-01 -3.88890773e-01 -2.56810933e-02
5.24629056e-01 6.55748665e-01 -5.86885870e-01 -6.83622479e-01
-2.86762208e-01 7.76557207e-01 2.21433923e-01 -5.75733073e-02
-8.32220972e-01 -8.07929873e-01 3.82995784e-01 4.39399421e-01
2.38090884e-02 6.28476143e-02 -8.80579412e-01 -1.10020387e+00
4.71613288e-01 -1.17635536e+00 6.52085245e-01 -5.50354719e-01
-7.50414848e-01 6.53101146e-01 -1.25782490e-02 -8.47041428e-01
-4.78115529e-01 -6.01348460e-01 -1.32723674e-01 7.94926167e-01
-1.26836479e+00 -9.29915488e-01 -3.43383133e-01 4.26805079e-01
5.51504076e-01 -1.26900569e-01 1.14315236e+00 3.80330920e-01
-4.90684390e-01 9.41038728e-01 -3.74103189e-01 -2.10903808e-01
6.46414101e-01 -1.33073187e+00 6.25762641e-01 8.26268554e-01
-7.27235228e-02 1.11764169e+00 5.45386374e-01 -8.20560098e-01
-1.91328716e+00 -1.26559484e+00 1.12743866e+00 -3.73175472e-01
3.57713521e-01 -7.23124266e-01 -1.03785360e+00 4.30807084e-01
-2.33697221e-01 -3.64735007e-01 5.30644536e-01 2.75492847e-01
-4.70022619e-01 -5.36627114e-01 -1.04066110e+00 7.52272487e-01
1.17208529e+00 -3.02591950e-01 -3.42402428e-01 3.87215197e-01
1.27049398e+00 -5.19801021e-01 -8.58426869e-01 6.94450915e-01
5.10907829e-01 -1.19367516e+00 9.65399504e-01 -5.17244816e-01
3.98419410e-01 -2.86021084e-01 -3.77946913e-01 -4.61610198e-01
4.91643220e-01 -7.03511655e-01 -2.35811636e-01 1.26499593e+00
7.02537417e-01 -7.55989790e-01 1.36933017e+00 7.15198159e-01
2.71518845e-02 -7.16711760e-01 -4.07256991e-01 -8.54784310e-01
-2.19203517e-01 -8.15088391e-01 1.20827317e+00 5.57785392e-01
-9.77499038e-02 3.33908141e-01 -6.27277717e-02 1.22360075e-02
5.30117452e-01 6.89884722e-01 1.11579180e+00 -1.11354625e+00
-3.66381079e-01 -2.90393472e-01 -2.25090794e-02 -1.09712851e+00
7.64157111e-03 -1.08235371e+00 -5.98239265e-02 -1.28998923e+00
-3.20614874e-02 -9.76609588e-01 4.34112512e-02 5.52104235e-01
-2.63965905e-01 -3.92902613e-01 6.00174628e-02 5.26766889e-02
-4.46915835e-01 1.09795712e-01 7.82950997e-01 6.53390288e-02
-3.86875451e-01 2.51752913e-01 -7.07907736e-01 4.72632468e-01
6.41138792e-01 -6.16905272e-01 -6.60971045e-01 -3.63460571e-01
6.83089495e-01 4.34025198e-01 8.47664177e-02 -1.00292683e+00
1.47831798e-01 -5.81757724e-01 -1.15560532e-01 -8.65629375e-01
-8.36212710e-02 -4.70617503e-01 2.90121496e-01 6.58261418e-01
-6.28928721e-01 3.58238727e-01 1.35802239e-01 2.10641012e-01
-5.30185223e-01 -5.00761807e-01 4.77075458e-01 -3.30862641e-01
-6.93719566e-01 -1.19138481e-02 -4.36710753e-02 3.67477387e-01
7.55131543e-01 5.04875206e-04 -2.87201285e-01 9.84827206e-02
-4.53335106e-01 1.69522390e-01 4.23979014e-01 4.49293554e-01
5.89423239e-01 -1.03849769e+00 -3.24312031e-01 3.78578424e-01
2.95428991e-01 3.35951418e-01 -1.71380147e-01 2.28916347e-01
-1.12697184e+00 8.07883501e-01 1.37500152e-01 -5.95888734e-01
-1.29590130e+00 3.99999917e-01 3.82938120e-03 -7.18353212e-01
-2.58580774e-01 7.38776624e-01 -6.39486536e-02 -8.80751967e-01
2.51615912e-01 -6.10716224e-01 1.10706016e-01 -2.23241717e-01
2.08845690e-01 3.40308964e-01 4.60571647e-01 2.74373263e-01
-4.78927851e-01 3.19420040e-01 -4.94077578e-02 -8.82520303e-02
1.04879797e+00 3.68275493e-01 -6.15619719e-01 2.16944024e-01
9.86395657e-01 3.78681272e-01 -3.97384405e-01 -7.10854083e-02
8.16412687e-01 -4.81079072e-01 -4.20082092e-01 -8.73003304e-01
-7.57098556e-01 7.46161401e-01 2.86119342e-01 2.70285398e-01
1.20196354e+00 -1.41243204e-01 1.18866396e+00 9.09504712e-01
6.92705154e-01 -1.12628162e+00 3.45859490e-02 5.70678234e-01
6.92659438e-01 -1.14054120e+00 -5.24214916e-02 -1.02565897e+00
-5.14128149e-01 1.39497304e+00 1.04516387e+00 2.41408587e-01
3.18976402e-01 4.45846856e-01 -8.65765363e-02 -1.23185813e-01
-9.48099732e-01 -8.48862901e-02 1.70515552e-02 3.34889382e-01
5.62878549e-01 1.22316584e-01 -5.52156031e-01 8.57755244e-01
-6.65126562e-01 1.30102159e-02 4.92540091e-01 1.23097765e+00
-3.84085447e-01 -1.95108116e+00 -3.30616504e-01 7.84250140e-01
-4.72905874e-01 -3.79152775e-01 -2.88610071e-01 6.54404283e-01
7.99767971e-02 7.97760189e-01 -1.25790238e-01 -3.38310212e-01
6.73415124e-01 3.09874117e-01 2.32039824e-01 -1.01137781e+00
-6.62441790e-01 -1.82903811e-01 4.74842131e-01 -6.27919674e-01
1.49392501e-01 -5.59197009e-01 -1.69805980e+00 -3.12659979e-01
-2.27166444e-01 2.45943472e-01 6.83304846e-01 5.14426470e-01
8.12858880e-01 3.35474044e-01 4.68861192e-01 2.82349527e-01
-5.89451313e-01 -2.35701233e-01 1.00609049e-01 2.72013456e-01
-3.65913272e-01 -2.52314925e-01 1.02269724e-01 2.38972846e-02]
|
[9.876947402954102, 7.819868087768555]
|
59a2e7bc-106b-4177-82eb-670c473024ec
|
cryptgpu-fast-privacy-preserving-machine
|
2104.10949
| null |
https://arxiv.org/abs/2104.10949v1
|
https://arxiv.org/pdf/2104.10949v1.pdf
|
CryptGPU: Fast Privacy-Preserving Machine Learning on the GPU
|
We introduce CryptGPU, a system for privacy-preserving machine learning that implements all operations on the GPU (graphics processing unit). Just as GPUs played a pivotal role in the success of modern deep learning, they are also essential for realizing scalable privacy-preserving deep learning. In this work, we start by introducing a new interface to losslessly embed cryptographic operations over secret-shared values (in a discrete domain) into floating-point operations that can be processed by highly-optimized CUDA kernels for linear algebra. We then identify a sequence of "GPU-friendly" cryptographic protocols to enable privacy-preserving evaluation of both linear and non-linear operations on the GPU. Our microbenchmarks indicate that our private GPU-based convolution protocol is over 150x faster than the analogous CPU-based protocol; for non-linear operations like the ReLU activation function, our GPU-based protocol is around 10x faster than its CPU analog. With CryptGPU, we support private inference and private training on convolutional neural networks with over 60 million parameters as well as handle large datasets like ImageNet. Compared to the previous state-of-the-art, when considering large models and datasets, our protocols achieve a 2x to 8x improvement in private inference and a 6x to 36x improvement for private training. Our work not only showcases the viability of performing secure multiparty computation (MPC) entirely on the GPU to enable fast privacy-preserving machine learning, but also highlights the importance of designing new MPC primitives that can take full advantage of the GPU's computing capabilities.
|
['David J. Wu', 'Yuan Tian', 'Brian Knott', 'Sijun Tan']
|
2021-04-22
| null | null | null | null |
['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning']
|
['methodology', 'natural-language-processing']
|
[ 1.24960616e-01 -4.52619512e-03 -1.61439423e-02 -6.84527755e-01
-8.83129478e-01 -1.10491490e+00 5.45681477e-01 3.39271367e-01
-1.03821027e+00 4.93694603e-01 -6.19087368e-04 -1.05890799e+00
5.52396178e-01 -1.16844678e+00 -1.12021828e+00 -9.95898426e-01
-2.01266482e-01 5.99478073e-02 2.49949619e-02 -9.91696715e-02
2.83367373e-02 6.48579478e-01 -1.54864907e+00 6.48274064e-01
2.02700973e-01 1.06436849e+00 -6.73282802e-01 8.71603370e-01
1.76264077e-01 3.79280597e-01 -1.97526380e-01 -9.04457450e-01
9.08573925e-01 2.33134031e-01 -8.05279553e-01 -7.48561203e-01
6.62044346e-01 -9.40377235e-01 -6.10242903e-01 9.47033525e-01
4.08761978e-01 -1.37493163e-01 2.48999009e-03 -1.38853204e+00
-3.02513633e-02 4.69173998e-01 -2.93383837e-01 -1.59306705e-01
-1.86238959e-01 2.09603935e-01 1.03375673e+00 5.60160866e-03
5.75479031e-01 7.82754540e-01 8.21458459e-01 6.01425171e-01
-1.29320025e+00 -1.05787027e+00 -4.67791468e-01 -4.21120003e-02
-1.17829537e+00 -4.51229095e-01 1.17086969e-01 1.58863235e-02
1.34591436e+00 8.39654326e-01 6.00855231e-01 8.79791617e-01
6.78792596e-01 6.13667130e-01 1.11474335e+00 -2.55241811e-01
2.81354636e-01 1.11350320e-01 5.25421619e-01 7.06654131e-01
5.59516251e-01 3.68121356e-01 -3.20900202e-01 -9.82953846e-01
5.40126503e-01 1.90411992e-02 -2.69255459e-01 -2.74889767e-01
-1.03493178e+00 1.04216075e+00 2.41630390e-01 -2.11817652e-01
4.65445127e-03 8.02612424e-01 1.22947049e+00 7.65534520e-01
3.30778837e-01 2.67687649e-01 -8.25761497e-01 -7.40238875e-02
-5.67639947e-01 5.68864703e-01 1.50260043e+00 8.26029181e-01
9.49897170e-01 -5.20696461e-01 -8.33188742e-03 -9.68734771e-02
-1.04837418e-01 1.64665312e-01 5.84143810e-02 -1.09581983e+00
4.18579429e-01 1.88142449e-01 -1.84889287e-01 -6.98698342e-01
-1.28192887e-01 -1.18219450e-01 -9.79032218e-01 5.48613071e-01
4.77502674e-01 -3.21154624e-01 -2.55959183e-01 1.68649745e+00
5.85001826e-01 1.53254151e-01 4.16550726e-01 6.45581901e-01
4.03571606e-01 7.97048151e-01 9.41470414e-02 1.48313984e-01
1.76409078e+00 -8.33363175e-01 -9.80210677e-02 7.74191245e-02
1.33423674e+00 -5.37405074e-01 7.98596859e-01 3.64367336e-01
-7.98082232e-01 6.87921792e-02 -1.11517096e+00 -7.69567907e-01
-3.78847092e-01 -2.97215670e-01 1.56037235e+00 1.22348464e+00
-1.37434423e+00 7.69465327e-01 -1.10900366e+00 1.94017142e-01
7.62552917e-01 8.06725144e-01 -1.05382597e+00 7.70959407e-02
-9.32436347e-01 1.63429603e-01 5.11754632e-01 -3.39955062e-01
-5.10700583e-01 -1.14043748e+00 -6.56000972e-01 3.73811901e-01
-4.82276753e-02 -9.58946049e-01 1.11602175e+00 -7.04645276e-01
-1.48382282e+00 1.34164357e+00 2.03057677e-01 -1.14054143e+00
6.67876065e-01 -2.01300848e-02 3.10050309e-01 -3.02519687e-02
-5.43587267e-01 5.09362340e-01 5.06399453e-01 -6.27133965e-01
-8.45749676e-01 -5.29027104e-01 3.16580176e-01 -2.18664445e-02
-5.50241113e-01 1.79138064e-01 -2.39038378e-01 -4.26282972e-01
-1.80238336e-01 -1.25845945e+00 -2.66733229e-01 5.86899400e-01
-8.35195407e-02 9.68500972e-02 9.93572593e-01 -4.35388088e-01
3.96550417e-01 -2.42475700e+00 -5.49575210e-01 2.69936413e-01
5.67198098e-01 5.00137329e-01 2.24464417e-01 2.52603710e-01
1.94107026e-01 -1.08553983e-01 -3.09289992e-01 -6.57498419e-01
3.30014646e-01 4.88598645e-01 -7.38062561e-01 9.96934772e-01
-4.02437449e-01 9.20309067e-01 -6.65284157e-01 2.88503170e-02
-1.57644480e-01 6.01485372e-01 -1.17856240e+00 -6.94268793e-02
-1.12243272e-01 2.30155915e-01 -4.08339113e-01 1.89269111e-01
1.31729007e+00 -9.32024345e-02 4.08372879e-01 -3.77045907e-02
-1.83943003e-01 3.70685726e-01 -8.34899545e-01 1.63587785e+00
-5.25044858e-01 8.11586201e-01 5.30343831e-01 -5.44660687e-01
4.84733880e-01 3.33272994e-01 9.38182026e-02 -3.02609384e-01
4.97913778e-01 2.78454483e-01 -3.41540039e-01 2.40717053e-01
5.98208785e-01 1.21696807e-01 -2.70162046e-01 1.03684640e+00
-1.74270287e-01 -1.59981519e-01 -6.05411887e-01 2.55199429e-02
1.45559096e+00 -3.41679573e-01 8.86687040e-02 -5.83387256e-01
6.13118351e-01 -1.64351821e-01 3.92222732e-01 8.99411380e-01
-9.47334617e-02 2.08825976e-01 8.05319309e-01 -1.16513753e+00
-1.00940669e+00 -6.76707208e-01 -1.79512069e-01 1.53431964e+00
-2.61569053e-01 -5.84982038e-01 -9.53337312e-01 -6.26422524e-01
3.08136284e-01 6.11345112e-01 -3.71171504e-01 -1.35651618e-01
-6.20658398e-01 -9.93254960e-01 1.17747462e+00 3.86274219e-01
9.32918191e-01 -4.30206954e-01 -7.76592731e-01 -9.52454060e-02
4.71351326e-01 -9.49893653e-01 -5.28807998e-01 4.58233029e-01
-9.46663260e-01 -7.25609660e-01 -1.07252136e-01 -6.15099072e-01
6.44766867e-01 3.10553551e-01 7.00480282e-01 2.33617157e-01
-1.77244991e-01 9.40729305e-02 2.01212928e-01 -4.02665317e-01
-6.50959015e-01 2.19028905e-01 -2.30591491e-01 -1.13326170e-01
4.30274636e-01 -9.04738426e-01 -7.08511531e-01 -2.09241241e-01
-9.17769909e-01 1.71454698e-01 2.48897940e-01 9.32227731e-01
7.39696681e-01 -9.39908475e-02 -4.51685011e-01 -1.38473809e+00
4.92936373e-01 -1.50196582e-01 -1.24864030e+00 -3.00883017e-02
-6.10117078e-01 2.30690435e-01 9.59553599e-01 -2.55451739e-01
-8.05787683e-01 3.60303968e-01 -4.63001251e-01 -3.27910602e-01
-2.46290006e-02 -1.55968755e-01 -9.46958438e-02 -9.28626537e-01
6.30822957e-01 2.48617142e-01 3.13032061e-01 -2.57036150e-01
3.75726759e-01 5.01909733e-01 7.22140431e-01 -1.03154826e+00
4.12461370e-01 9.16534245e-01 4.93716389e-01 -3.58257085e-01
-1.06077433e-01 -1.36519998e-01 -2.00829893e-01 7.39037097e-01
7.16449797e-01 -1.12723279e+00 -1.65786397e+00 9.36649799e-01
-1.28558671e+00 -4.46456432e-01 -3.18749130e-01 1.98109791e-01
-4.31094080e-01 5.15759647e-01 -1.05339599e+00 -4.30701643e-01
-1.10207140e+00 -1.41545486e+00 1.06148207e+00 -7.61668235e-02
1.67095616e-01 -9.04955268e-01 -3.31804231e-02 3.86509031e-01
3.90030742e-01 3.07715327e-01 8.32147300e-01 -8.47065628e-01
-8.53005588e-01 -5.18996596e-01 -4.78339285e-01 4.67741102e-01
-6.48307681e-01 -4.12340224e-01 -1.48081958e+00 -7.23270237e-01
4.32506859e-01 -2.74762899e-01 9.58208263e-01 -1.64816454e-01
1.70569396e+00 -8.53296876e-01 -1.35278553e-01 1.69520664e+00
1.59992051e+00 -2.97351867e-01 7.58680940e-01 3.15148920e-01
7.30130672e-01 2.39358902e-01 2.24744026e-02 5.85826814e-01
4.39596951e-01 2.73436219e-01 5.06430149e-01 -6.14843853e-02
4.95835692e-01 -8.00079554e-02 7.79984742e-02 3.59839380e-01
-2.07983889e-02 1.77433833e-01 -7.63682365e-01 2.50235461e-02
-1.73490238e+00 -4.77140993e-01 -4.12907839e-01 2.42723441e+00
9.64596927e-01 -8.29200596e-02 -2.79471099e-01 -1.20179608e-01
2.45719366e-02 1.19169697e-01 -3.90150279e-01 -1.06781292e+00
8.84702522e-03 9.18101013e-01 1.52356553e+00 4.07685459e-01
-1.23206449e+00 9.73149002e-01 5.49030304e+00 8.72878551e-01
-1.25040174e+00 6.33715987e-01 9.22832668e-01 -1.30818054e-01
-4.36241865e-01 1.57569379e-01 -7.92530835e-01 2.43149996e-01
1.10688663e+00 -2.43077710e-01 5.78124404e-01 1.14654505e+00
-6.09687448e-01 1.30287006e-01 -1.29841030e+00 9.91660893e-01
-5.36974132e-01 -1.66845095e+00 -3.42609912e-01 6.22942567e-01
5.57362258e-01 4.17391539e-01 1.57794148e-01 5.29477485e-02
4.29091513e-01 -8.55719924e-01 4.85675246e-01 -2.92600960e-01
9.53396916e-01 -1.08753335e+00 7.95263350e-01 2.70744145e-01
-7.10937679e-01 1.67719334e-01 -6.50152087e-01 -3.84805262e-01
-4.00154769e-01 3.21494848e-01 -6.86737180e-01 9.52985436e-02
7.83268571e-01 4.13249806e-02 -2.17889667e-01 4.44326788e-01
-2.52255738e-01 6.14976466e-01 -6.40667379e-01 -5.38147092e-02
4.42629516e-01 -2.80389071e-01 2.43662402e-01 1.35984957e+00
1.87203214e-01 4.14555937e-01 -3.98534358e-01 4.59170669e-01
-5.48754513e-01 1.12196423e-01 -6.40857100e-01 1.16635308e-01
3.06327999e-01 1.28122306e+00 -5.40471673e-01 -1.85261488e-01
-3.57502222e-01 1.33050478e+00 3.05588871e-01 -1.51725248e-01
-7.23146558e-01 -4.29976463e-01 1.58407283e+00 -2.99948782e-01
2.53057569e-01 -3.99180502e-01 -8.11757088e-01 -1.28520906e+00
-8.90164822e-03 -9.11460280e-01 4.58544642e-01 1.58690125e-01
-1.06226480e+00 2.93369055e-01 -3.90652686e-01 -5.18309057e-01
7.36112073e-02 -7.66916633e-01 -4.94066656e-01 1.02144742e+00
-1.30952024e+00 -1.25371516e+00 1.16622880e-01 8.14342022e-01
-5.41512668e-01 -1.80771858e-01 1.51467061e+00 3.25180769e-01
-9.06271413e-02 1.41546631e+00 6.24101579e-01 1.11231588e-01
5.39518714e-01 -8.82832408e-01 1.25384080e+00 7.27361560e-01
4.34253924e-02 1.04643381e+00 5.53388357e-01 -2.48323053e-01
-2.25828815e+00 -1.10564482e+00 9.16026592e-01 -3.23740870e-01
6.17573678e-01 -9.20451343e-01 -9.11834419e-01 1.07285714e+00
-6.75834194e-02 6.89298272e-01 9.72545624e-01 1.18667977e-02
-9.98895645e-01 -4.12355334e-01 -1.59976721e+00 6.22384727e-01
9.49229598e-01 -8.69495213e-01 1.64456263e-01 6.63228691e-01
1.00840056e+00 -1.08325887e+00 -7.58973420e-01 -2.15578601e-01
8.99357140e-01 -1.02985132e+00 1.13554764e+00 -6.21787727e-01
1.13936327e-01 1.19286872e-01 -3.70552212e-01 -5.72265625e-01
-1.09579548e-01 -1.16142464e+00 -1.40056312e-01 6.13045216e-01
-2.62777414e-02 -1.55769086e+00 1.33095491e+00 1.06344545e+00
1.68837547e-01 -5.65716684e-01 -1.29844701e+00 -4.12002742e-01
3.57459635e-01 -6.60707176e-01 1.09500718e+00 1.06716907e+00
-3.94355118e-01 -4.53720808e-01 -1.38933286e-01 5.07682979e-01
8.63634884e-01 6.22555800e-02 1.09055686e+00 -8.26278389e-01
-7.15969443e-01 -2.06047446e-01 -7.64684975e-01 -6.70987964e-01
5.81267715e-01 -1.22548819e+00 -4.00281727e-01 -4.71574306e-01
1.53002679e-01 -6.22690797e-01 -1.19085193e-01 1.08648896e+00
3.80422175e-01 5.55958450e-01 3.80613841e-02 -5.87032437e-02
3.11405514e-03 7.59433657e-02 7.23088741e-01 4.08330783e-02
3.78686786e-02 7.96437785e-02 -8.43044281e-01 4.63574171e-01
6.33463979e-01 -6.50026679e-01 -1.61903709e-01 -7.49261022e-01
3.36797297e-01 -1.82713479e-01 6.08836532e-01 -9.42543030e-01
6.00422084e-01 1.57640219e-01 -3.47516425e-02 3.08669601e-02
3.41057360e-01 -7.76454628e-01 4.56600189e-01 8.13813508e-01
-1.16381787e-01 -6.66507035e-02 5.24421334e-01 3.49426478e-01
2.26198345e-01 1.43490016e-01 6.87364578e-01 -2.67727487e-02
-4.71245110e-01 6.31568551e-01 -1.90444946e-01 -2.39321277e-01
1.10100901e+00 2.14448020e-01 -5.94067454e-01 -9.04563889e-02
-1.33961126e-01 -4.17186394e-02 9.47710991e-01 -1.90495655e-01
1.77497864e-01 -8.39677989e-01 -5.58791459e-01 6.77363992e-01
-9.91325229e-02 -1.85058303e-02 2.46222332e-01 4.49307978e-01
-1.37558484e+00 4.31066036e-01 -3.49234015e-01 -2.55467534e-01
-1.73735809e+00 2.95026153e-01 9.41360742e-02 -3.34101021e-01
-9.68668580e-01 1.34179652e+00 4.35669512e-01 -5.98793626e-01
2.27633089e-01 -3.35608482e-01 9.30731595e-01 -4.56895590e-01
9.18184459e-01 3.01996380e-01 4.95256722e-01 -2.40779638e-01
-2.22853184e-01 -7.13953972e-02 -3.80563408e-01 -4.60249260e-02
1.47120750e+00 5.15003502e-01 -8.24224949e-01 -4.20815766e-01
1.88449025e+00 -4.00573574e-02 -1.24853837e+00 3.55227897e-03
-6.39253497e-01 -5.89511514e-01 3.78871769e-01 -2.51541853e-01
-1.19096851e+00 1.00425720e+00 5.57323754e-01 -3.05070460e-01
1.12960744e+00 -4.62087244e-01 1.34228063e+00 8.62064362e-01
1.00557852e+00 -6.33879900e-01 -1.01924682e+00 5.53427875e-01
1.45688504e-01 -1.04152513e+00 1.91251054e-01 -5.75328052e-01
-1.00556232e-01 1.20730567e+00 1.34925514e-01 -2.72298425e-01
7.20082104e-01 9.01930034e-01 -7.64259845e-02 2.04585269e-01
-8.25770557e-01 7.05764413e-01 -2.20538527e-01 3.91912609e-01
-7.21726269e-02 3.33457172e-01 -3.93714100e-01 6.74699843e-01
-4.81247991e-01 6.55824468e-02 3.30062509e-01 1.25507140e+00
6.20276518e-02 -1.57031250e+00 -2.33464077e-01 4.24295247e-01
-8.00884247e-01 -2.63866425e-01 7.88841024e-02 5.47621250e-01
1.27042055e-01 1.88006818e-01 2.03471929e-01 -4.07541662e-01
-2.35287234e-01 -1.00634992e-01 5.21383166e-01 -3.51339817e-01
-1.48475897e+00 -6.44021690e-01 6.56279176e-02 -1.03053379e+00
4.60731506e-01 -5.91547608e-01 -1.23180795e+00 -1.31839252e+00
8.38475376e-02 -2.05646783e-01 1.10506463e+00 5.14297664e-01
6.84835076e-01 -2.50626922e-01 4.76955414e-01 -8.11307728e-01
-9.31593895e-01 1.94460064e-01 -5.83916903e-01 8.91645104e-02
3.12959254e-01 3.49892884e-01 -4.98516411e-01 -3.76666337e-01]
|
[5.873862266540527, 6.848789691925049]
|
fb952fce-034f-4e12-a4da-5ed4a0dfde7e
|
ponder-point-cloud-pre-training-via-neural
|
2301.00157
| null |
https://arxiv.org/abs/2301.00157v1
|
https://arxiv.org/pdf/2301.00157v1.pdf
|
Ponder: Point Cloud Pre-training via Neural Rendering
|
We propose a novel approach to self-supervised learning of point cloud representations by differentiable neural rendering. Motivated by the fact that informative point cloud features should be able to encode rich geometry and appearance cues and render realistic images, we train a point-cloud encoder within a devised point-based neural renderer by comparing the rendered images with real images on massive RGB-D data. The learned point-cloud encoder can be easily integrated into various downstream tasks, including not only high-level tasks like 3D detection and segmentation, but low-level tasks like 3D reconstruction and image synthesis. Extensive experiments on various tasks demonstrate the superiority of our approach compared to existing pre-training methods.
|
['Wanli Ouyang', 'Xiaowei Zhou', 'Tong He', 'Sida Peng', 'Di Huang']
|
2022-12-31
| null | null | null | null |
['point-cloud-pre-training']
|
['computer-vision']
|
[ 2.24847853e-01 2.30525434e-01 4.11087126e-02 -6.06955230e-01
-9.41573679e-01 -4.32275355e-01 7.23831415e-01 3.26765701e-02
3.82228848e-03 1.79742604e-01 -2.55520791e-01 -4.70236272e-01
2.26960033e-01 -1.02367496e+00 -1.27885318e+00 -3.35955560e-01
-1.08869337e-01 6.73503220e-01 2.16970548e-01 -2.25906476e-01
3.02034527e-01 1.22759950e+00 -1.80086744e+00 1.79750115e-01
7.90367603e-01 1.28517330e+00 1.65706307e-01 6.09850168e-01
-4.41560328e-01 4.08078969e-01 -3.95494640e-01 -1.86860085e-01
5.47963917e-01 1.22409016e-01 -5.83340168e-01 5.65881073e-01
6.09074593e-01 -5.59376299e-01 -2.22739860e-01 9.44343984e-01
1.64105803e-01 2.54770607e-01 6.84223235e-01 -9.65018630e-01
-6.45196259e-01 -3.01063359e-01 -5.24571419e-01 -3.26875567e-01
4.63938683e-01 2.07439452e-01 6.43667400e-01 -1.12684679e+00
6.68486238e-01 1.43004239e+00 5.63385367e-01 4.12106425e-01
-1.12645662e+00 -5.06041586e-01 2.62245953e-01 -4.14982289e-01
-1.06138897e+00 -2.17572004e-01 1.34472072e+00 -5.24170816e-01
8.01916838e-01 2.76631474e-01 7.53374338e-01 7.70324826e-01
-1.65983126e-01 7.89496660e-01 1.08958340e+00 -2.93908566e-01
3.54169428e-01 -6.63080737e-02 -3.58265668e-01 1.00289094e+00
-2.77253538e-01 4.38222677e-01 -3.25455338e-01 -1.26222491e-01
1.68117368e+00 1.57347605e-01 -1.53238192e-01 -6.08577907e-01
-1.24828017e+00 8.25503826e-01 9.64265347e-01 -1.65214807e-01
-4.51472312e-01 6.08331025e-01 1.29860476e-01 1.67247385e-01
1.07587230e+00 4.02947247e-01 -4.28997070e-01 1.85921475e-01
-7.62429357e-01 1.94874197e-01 3.38946253e-01 1.18688989e+00
1.09558821e+00 2.94651240e-01 2.29016826e-01 5.40848911e-01
4.83744949e-01 4.70309585e-01 2.75532380e-02 -1.38929355e+00
3.97562802e-01 7.16779232e-01 1.67501509e-01 -7.32693911e-01
-9.53106955e-02 -2.13305160e-01 -7.45873690e-01 8.44703734e-01
-2.43159160e-01 2.16687694e-01 -1.07758880e+00 9.76524413e-01
5.52286267e-01 4.29334581e-01 -1.23964071e-01 9.31117356e-01
1.17000031e+00 8.79956424e-01 -2.14749545e-01 2.33146563e-01
8.84678662e-01 -6.26368821e-01 -8.28122944e-02 -1.77453756e-01
1.53211072e-01 -5.43531001e-01 1.14382291e+00 1.05478652e-01
-1.34310985e+00 -7.57510543e-01 -9.66420770e-01 -3.44675899e-01
-7.80510604e-02 -7.42856190e-02 1.04359710e+00 9.06600878e-02
-1.16394484e+00 8.20964158e-01 -8.94073606e-01 1.80610165e-01
7.62014151e-01 3.00412357e-01 -2.90514141e-01 5.35946079e-02
-4.98262167e-01 5.55707455e-01 1.86236903e-01 5.97146079e-02
-7.96411693e-01 -8.11955750e-01 -1.03599560e+00 -1.44086853e-01
-6.32416829e-02 -1.23334837e+00 1.20566595e+00 -8.56860518e-01
-1.76985490e+00 1.28348553e+00 -1.36711016e-01 -2.41046082e-02
5.79255462e-01 -1.27500683e-01 1.82428151e-01 3.31415087e-01
5.69877289e-02 9.29786563e-01 1.06933618e+00 -1.93434227e+00
-5.33133268e-01 -4.94319618e-01 1.84696317e-01 4.43625778e-01
4.24295723e-01 -2.41585001e-01 -5.66211402e-01 -3.63248408e-01
5.37796557e-01 -6.87207162e-01 -6.26786530e-01 7.46902108e-01
-3.98846567e-01 -1.83179170e-01 8.49402130e-01 -3.27226728e-01
-1.49507970e-01 -2.02629018e+00 8.15055221e-02 2.13208646e-01
2.70024151e-01 5.19410372e-02 -1.06727958e-01 4.75920290e-02
-1.12319291e-01 -1.56301484e-02 -2.90501773e-01 -7.43952930e-01
7.21141472e-02 3.90897959e-01 -6.31513178e-01 5.65557778e-01
7.40214348e-01 1.20965195e+00 -9.17089641e-01 -3.99752676e-01
9.78530526e-01 8.03532600e-01 -4.72384036e-01 4.54656065e-01
-7.59151161e-01 9.99682665e-01 -6.32777095e-01 8.60565543e-01
7.54132986e-01 -5.01951873e-01 -6.96079075e-01 -2.92563047e-02
-1.26792699e-01 3.24088722e-01 -7.07899213e-01 2.09312010e+00
-7.46200740e-01 6.42662346e-01 1.91439707e-02 -8.76766026e-01
1.13166904e+00 4.15144786e-02 4.99821901e-01 -5.82257211e-01
1.94373399e-01 6.66648522e-02 -7.75008976e-01 -2.80082822e-01
5.10993898e-01 -2.30440453e-01 6.91292733e-02 8.99756178e-02
-1.45220205e-01 -1.10806894e+00 -6.63525343e-01 2.47264989e-02
7.13931382e-01 8.58446717e-01 -1.03608556e-01 1.18425325e-01
2.59917706e-01 3.41747493e-01 1.10174052e-01 1.80589721e-01
2.81879514e-01 9.94022787e-01 1.09745003e-01 -6.21579230e-01
-1.17993188e+00 -1.26175654e+00 -3.19612831e-01 7.79950559e-01
4.46219891e-01 -6.24248050e-02 -3.61115694e-01 -3.66697818e-01
3.34328741e-01 5.46418786e-01 -4.94183838e-01 1.26675546e-01
-7.14755177e-01 -1.49718821e-01 9.02250856e-02 7.04103351e-01
4.26480830e-01 -9.75510418e-01 -7.02653885e-01 6.00906555e-03
2.50284344e-01 -1.36046624e+00 1.32821426e-01 2.10814670e-01
-1.45231569e+00 -8.50996614e-01 -6.42936170e-01 -8.13951731e-01
9.35305834e-01 4.87800330e-01 1.49309981e+00 2.64131755e-01
-1.12612605e-01 5.94274342e-01 -7.88577870e-02 -4.55096960e-01
-5.83008349e-01 -3.28077525e-01 -2.50828862e-01 -2.51069725e-01
-2.10466594e-01 -8.82205486e-01 -6.13220632e-01 4.58863191e-02
-8.36959422e-01 3.81393403e-01 4.68101621e-01 4.49733406e-01
1.17654133e+00 -3.67910832e-01 9.75974090e-03 -7.33079672e-01
3.02895993e-01 -6.62421733e-02 -8.61361086e-01 1.50397047e-02
-2.99569536e-02 4.02138121e-02 3.77252430e-01 -9.16090757e-02
-8.83075535e-01 5.34378052e-01 -5.33267617e-01 -1.13555682e+00
-3.20877731e-01 1.84567362e-01 -6.70141354e-02 -4.88696754e-01
5.25403559e-01 2.88333863e-01 -5.82360402e-02 -4.47384626e-01
7.30348825e-01 4.03487563e-01 8.21726561e-01 -7.52174795e-01
1.22377241e+00 8.53060782e-01 2.31319800e-01 -9.29772794e-01
-6.02684975e-01 -3.72976661e-01 -1.04316926e+00 -2.59986192e-01
8.84437382e-01 -1.14944899e+00 -8.21177781e-01 1.04820125e-01
-1.64499819e+00 -5.13549745e-01 -5.52006602e-01 1.05062135e-01
-1.08913231e+00 2.30205432e-01 -4.71965700e-01 -7.46646225e-01
-3.40328127e-01 -1.30352008e+00 2.08340740e+00 -5.91530539e-02
2.06133157e-01 -1.00384855e+00 -6.94320574e-02 1.40274823e-01
-1.03751831e-01 8.90619099e-01 9.30552661e-01 2.45847076e-01
-1.15150666e+00 -2.20738068e-01 -4.53521371e-01 3.16591680e-01
3.60597596e-02 2.55279213e-01 -1.25344598e+00 4.51449379e-02
1.52719487e-02 -3.80904704e-01 6.19174004e-01 3.74410361e-01
1.61098111e+00 6.62118644e-02 -3.84064883e-01 1.34380078e+00
1.55751622e+00 -2.51156121e-01 4.85954672e-01 1.56008303e-01
1.11731398e+00 3.93595219e-01 4.72609818e-01 2.76198715e-01
4.23472762e-01 5.67710161e-01 1.06286383e+00 -5.79147995e-01
-1.45715281e-01 -5.67792356e-01 -1.30330071e-01 7.61901975e-01
-4.15628105e-01 2.34965190e-01 -8.89327168e-01 8.03732201e-02
-1.62794149e+00 -4.73767966e-01 -2.95727283e-01 1.89844906e+00
4.91972774e-01 -2.40938198e-02 -9.72336307e-02 1.04385149e-02
3.54690760e-01 2.65545815e-01 -8.06976080e-01 -3.03739846e-01
-1.06662530e-02 4.33093995e-01 4.94658202e-01 4.29231644e-01
-1.13908470e+00 1.11616349e+00 6.74096870e+00 4.07016337e-01
-1.28309429e+00 -2.25114673e-02 6.21331632e-01 1.09317593e-01
-7.00096667e-01 -1.56294972e-01 -3.00503075e-01 1.61696866e-01
4.36985642e-01 -2.45788731e-02 2.33318552e-01 1.10744381e+00
-1.93076525e-02 2.55805969e-01 -1.43430388e+00 1.37285888e+00
-1.47262543e-01 -1.72070658e+00 1.81972444e-01 8.99896100e-02
8.04609954e-01 4.06088293e-01 1.81655779e-01 -1.90405780e-03
6.76927030e-01 -1.01905739e+00 9.53366935e-01 4.92636859e-01
1.11337066e+00 -7.94411361e-01 3.19835730e-02 5.49674332e-01
-1.07438910e+00 3.83631349e-01 -6.81479156e-01 -1.69053744e-03
2.07241222e-01 5.78668177e-01 -6.78847015e-01 6.29389942e-01
6.33002818e-01 1.00213468e+00 -3.28534275e-01 9.45443153e-01
-4.28685933e-01 -2.91825738e-03 -4.84373599e-01 2.71062762e-01
4.58484530e-01 -3.97984892e-01 4.76966798e-01 8.17943871e-01
2.58211553e-01 2.67381728e-01 5.94174601e-02 1.32608795e+00
-1.10866517e-01 -1.23698823e-01 -9.57653165e-01 3.16470772e-01
3.21637124e-01 1.05282414e+00 -7.77076840e-01 -3.79131824e-01
-2.08538249e-01 1.23545003e+00 4.65119213e-01 3.85037154e-01
-6.27886117e-01 -3.24099250e-02 7.77202427e-01 1.34966016e-01
3.36329490e-01 -7.11984575e-01 -5.14715016e-01 -1.11698127e+00
8.90866891e-02 -3.47640097e-01 -3.00581306e-01 -1.24626136e+00
-1.26586020e+00 8.68840635e-01 -1.18112303e-02 -1.57607341e+00
-3.51933450e-01 -8.26949954e-01 -6.27931714e-01 9.91824985e-01
-1.81277752e+00 -1.43582773e+00 -5.19152343e-01 5.81182241e-01
5.44357717e-01 1.73818186e-01 8.88826132e-01 -8.02348480e-02
-8.40305351e-03 -6.11660592e-02 -1.73377413e-02 -2.53397529e-03
-1.44521771e-02 -1.34017181e+00 1.10033500e+00 5.62790930e-01
3.17066491e-01 3.79167050e-01 1.75254658e-01 -5.01449883e-01
-1.66121173e+00 -1.30591285e+00 -5.83176240e-02 -6.91931427e-01
1.58161119e-01 -5.77650070e-01 -9.81257558e-01 6.61963463e-01
-2.53157198e-01 5.45479715e-01 2.43937701e-01 -1.90635934e-01
-2.84126788e-01 2.50378072e-01 -1.09155679e+00 4.46341842e-01
1.25721359e+00 -7.82597780e-01 -4.69822913e-01 7.79230833e-01
9.99819338e-01 -1.07539117e+00 -8.41003716e-01 4.21634376e-01
7.92311504e-02 -1.02153873e+00 1.61253524e+00 -5.79298675e-01
9.08935547e-01 -1.72387332e-01 -1.93738908e-01 -1.40257406e+00
-1.37932464e-01 -5.20517588e-01 -2.16999441e-01 7.18971133e-01
-3.84994149e-02 -4.07243878e-01 1.07044256e+00 6.85374558e-01
-6.50277197e-01 -8.49479079e-01 -9.40970957e-01 -5.90234995e-01
2.42037833e-01 -7.22337663e-01 9.65489209e-01 8.16595495e-01
-6.01150632e-01 6.76374286e-02 1.87395029e-02 4.69953537e-01
8.50153744e-01 6.25879884e-01 9.46470141e-01 -1.45233476e+00
-1.85238406e-01 -4.42664444e-01 -6.11946404e-01 -1.60144138e+00
4.16167110e-01 -1.06022012e+00 2.69558042e-01 -1.82962120e+00
-5.42611897e-01 -9.62738037e-01 2.02362061e-01 1.75499842e-01
-8.66157189e-02 4.12979126e-01 1.20756611e-01 4.36028630e-01
-3.47637534e-01 8.15220416e-01 1.65746486e+00 -7.49221444e-02
-1.73302025e-01 1.82672620e-01 -4.16018635e-01 9.08710539e-01
4.08312857e-01 -1.29097208e-01 -3.28041911e-01 -9.98319209e-01
1.88198686e-01 4.17161375e-01 7.18900979e-01 -8.75295281e-01
2.55815964e-02 -2.72521913e-01 7.79240608e-01 -1.00402474e+00
8.33762228e-01 -8.75393629e-01 -1.93916820e-02 9.93239358e-02
-8.33517089e-02 5.74001856e-02 2.34027952e-01 5.15180826e-01
-1.33308142e-01 7.72374570e-02 5.51836133e-01 -5.20315588e-01
-7.79549122e-01 7.57916212e-01 3.42492938e-01 -3.83556724e-01
9.32953238e-01 -4.63456303e-01 8.92179608e-02 -3.19391817e-01
-5.36745787e-01 -2.09751260e-03 8.61210525e-01 4.43901271e-01
1.22871387e+00 -1.41371918e+00 -5.66123009e-01 5.70162356e-01
1.69006124e-01 1.07881510e+00 -1.58721000e-01 3.14024501e-02
-1.07143915e+00 1.51238605e-01 -1.27719462e-01 -1.19983983e+00
-8.22966278e-01 4.36003178e-01 4.04558033e-01 3.49464715e-01
-1.23116195e+00 9.63905752e-01 4.68152344e-01 -8.72513652e-01
2.07183167e-01 -6.89439893e-01 3.54010880e-01 -6.93762004e-01
2.06094846e-01 -7.84054399e-03 2.54821897e-01 -7.51674056e-01
-1.27787426e-01 9.52029824e-01 4.86802101e-01 6.27358183e-02
1.59326768e+00 1.39662251e-01 -5.42391948e-02 5.48640788e-01
1.43899715e+00 -4.84762698e-01 -1.69360864e+00 -2.37115875e-01
-2.74464577e-01 -8.74128222e-01 3.45039159e-01 -2.58576393e-01
-1.18932486e+00 1.22606802e+00 4.52469975e-01 -7.17490390e-02
7.35008657e-01 2.94860333e-01 6.58777058e-01 3.70857865e-01
8.46906722e-01 -4.76566523e-01 1.28697231e-01 3.60984534e-01
1.18011904e+00 -1.38446450e+00 2.24289671e-01 -8.30414474e-01
-4.00784910e-01 1.04291821e+00 4.22263145e-01 -8.10079098e-01
5.22106171e-01 2.77400494e-01 1.18818045e-01 -4.94330078e-01
-4.41292524e-01 -1.84636116e-01 5.01617789e-01 8.90412629e-01
1.37851164e-01 2.89917532e-02 6.69848621e-01 -3.27622108e-02
-4.46280330e-01 -3.63973975e-02 1.22832164e-01 8.50011349e-01
-2.28025258e-01 -9.79810297e-01 -3.83136809e-01 3.00642252e-01
2.07209393e-01 2.40571231e-01 -3.38655859e-01 7.51858056e-01
2.61920206e-02 1.79562062e-01 5.57737708e-01 -3.18090528e-01
4.20859963e-01 -2.56908178e-01 8.12347174e-01 -8.39918733e-01
-3.84012401e-01 -1.09184729e-02 -2.67412663e-01 -7.62552440e-01
-5.59922457e-01 -4.82148051e-01 -1.42832065e+00 -1.04556605e-01
-1.43368870e-01 -2.48299703e-01 1.11497676e+00 7.13048756e-01
3.47646505e-01 6.49324834e-01 9.63802457e-01 -1.87192285e+00
-3.38887960e-01 -3.70295465e-01 -3.40145320e-01 4.81692642e-01
5.18465281e-01 -6.51011765e-01 -2.14975566e-01 3.03691085e-02]
|
[8.487683296203613, -3.4736037254333496]
|
5fecc78a-aa2b-43c1-b7e6-e2df585ea582
|
static-background-removal-in-vehicular-radar
|
2307.01444
| null |
https://arxiv.org/abs/2307.01444v1
|
https://arxiv.org/pdf/2307.01444v1.pdf
|
Static Background Removal in Vehicular Radar: Filtering in Azimuth-Elevation-Doppler Domain
|
A significant challenge in autonomous driving systems lies in image understanding within complex environments, particularly dense traffic scenarios. An effective solution to this challenge involves removing the background or static objects from the scene, so as to enhance the detection of moving targets as key component of improving overall system performance. In this paper, we present an efficient algorithm for background removal in automotive radar applications, specifically utilizing a frequency-modulated continuous wave (FMCW) radar. Our proposed algorithm follows a three-step approach, encompassing radar signal preprocessing, three-dimensional (3D) ego-motion estimation, and notch filter-based background removal in the azimuth-elevation-Doppler domain. To begin, we model the received signal of the FMCW multiple-input multiple-output (MIMO) radar and develop a signal processing framework for extracting four-dimensional (4D) point clouds. Subsequently, we introduce a robust 3D ego-motion estimation algorithm that accurately estimates radar ego-motion speed, accounting for Doppler ambiguity, by processing the point clouds. Additionally, our algorithm leverages the relationship between Doppler velocity, azimuth angle, elevation angle, and radar ego-motion speed to identify the spectrum belonging to background clutter. Subsequently, we employ notch filters to effectively filter out the background clutter. The performance of our algorithm is evaluated using both simulated data and extensive experiments with real-world data. The results demonstrate its effectiveness in efficiently removing background clutter and enhacing perception within complex environments. By offering a fast and computationally efficient solution, our approach effectively addresses challenges posed by non-homogeneous environments and real-time processing requirements.
|
['Lyutianyang Zhang', 'Sumit Roy', 'Xiangyu Gao']
|
2023-07-04
| null | null | null | null |
['autonomous-driving', 'motion-estimation']
|
['computer-vision', 'computer-vision']
|
[ 5.55273652e-01 -7.17225373e-01 4.24855381e-01 -1.58895060e-01
-8.25930417e-01 -6.18901193e-01 5.47976553e-01 -3.77413481e-01
-4.55959558e-01 4.65228707e-01 -1.53707847e-01 -4.91920292e-01
-3.19556385e-01 -7.86065578e-01 -2.46315047e-01 -9.75068271e-01
-1.52196541e-01 -9.05035436e-02 3.97331446e-01 -2.17125908e-01
2.32288077e-01 1.06863225e+00 -1.62467861e+00 -3.14997435e-01
8.64957809e-01 1.01295722e+00 1.49175122e-01 1.04647946e+00
1.67777196e-01 5.49556971e-01 -8.21136296e-01 -7.21076876e-02
5.82586825e-01 -6.58574849e-02 5.72146356e-01 9.82367769e-02
6.43392920e-01 -3.71243417e-01 -5.60036898e-01 1.13346207e+00
3.63435447e-01 2.17890576e-01 4.75471795e-01 -1.22379792e+00
2.70535469e-01 -5.58388650e-01 -7.55118251e-01 5.97177446e-01
-2.56290376e-01 1.90301284e-01 1.66425660e-01 -1.08013988e+00
1.17303960e-01 1.00268221e+00 5.20729959e-01 -9.80405957e-02
-8.74551237e-01 -9.87724960e-01 7.25945309e-02 3.78840208e-01
-1.47704589e+00 -5.79958320e-01 8.34017575e-01 -5.92452109e-01
4.28093106e-01 4.23004299e-01 4.66017038e-01 5.84288418e-01
6.74860716e-01 2.52942175e-01 9.22478795e-01 -4.35904503e-01
2.31886834e-01 -3.12406868e-01 4.34908271e-01 2.56298512e-01
9.02044058e-01 5.56880832e-01 -4.32431191e-01 -1.93771452e-01
4.48880792e-01 -8.10966566e-02 -3.01735640e-01 -4.27296788e-01
-8.95064950e-01 6.02522731e-01 -8.78290981e-02 -1.32714048e-01
-6.74346745e-01 1.41650870e-01 6.75229123e-03 5.23818173e-02
4.16782767e-01 -1.38145760e-02 -3.07082292e-02 2.01359913e-02
-9.95874643e-01 4.87694263e-01 6.01914048e-01 9.20044124e-01
6.72888815e-01 6.24982595e-01 -1.04144178e-01 4.15076822e-01
3.83710563e-01 1.42735946e+00 -2.31414214e-01 -8.95997226e-01
3.78263324e-01 -5.60959103e-03 4.79209125e-01 -1.20545304e+00
-2.39119366e-01 -1.07684016e+00 -7.28182912e-01 6.00779831e-01
1.80673122e-01 -3.43628258e-01 -1.03508568e+00 1.52476740e+00
6.77040756e-01 5.77328444e-01 4.31301177e-01 8.28609705e-01
2.29929715e-01 4.71458614e-01 -1.09555885e-01 -4.33815241e-01
1.49185109e+00 -2.96809018e-01 -1.02923703e+00 -7.84066021e-01
8.09298381e-02 -1.01850820e+00 2.65523177e-02 4.07142878e-01
-5.19017696e-01 -6.70259595e-01 -1.30367947e+00 8.27008426e-01
-1.24471150e-02 4.56111357e-02 1.77510232e-01 9.64417934e-01
-4.09425169e-01 -2.48819306e-01 -7.00474143e-01 5.41836172e-02
1.44238248e-01 -6.87093288e-03 5.79572991e-02 -4.90021825e-01
-1.10548723e+00 1.05521047e+00 -3.38013954e-02 4.51843321e-01
-5.79726994e-01 -8.46984684e-01 -8.86804402e-01 -2.51642048e-01
5.36860108e-01 -5.45444191e-01 1.05838454e+00 -4.77604479e-01
-8.97885501e-01 1.78243294e-01 -4.83925581e-01 -8.07621479e-01
2.71523982e-01 -6.19305372e-01 -1.10971701e+00 3.15649092e-01
5.89470193e-02 -1.07211486e-01 1.32205534e+00 -1.41410756e+00
-1.12455726e+00 -3.88290465e-01 -5.50882339e-01 1.96016565e-01
3.96688044e-01 -2.22011417e-01 -3.24773610e-01 -6.51362717e-01
3.25918257e-01 -8.16132188e-01 -3.31885487e-01 -5.10883212e-01
1.78004041e-01 7.58125842e-01 1.05414402e+00 -6.40141845e-01
9.80244339e-01 -2.52689028e+00 -6.36634946e-01 4.71555501e-01
2.88914979e-01 3.89553010e-01 2.44942699e-02 2.03767359e-01
2.57955432e-01 -7.64973700e-01 -1.85740247e-01 1.88745469e-01
-1.86991021e-01 -6.45268783e-02 -6.70130074e-01 7.27888763e-01
5.50363302e-01 5.75289786e-01 -6.12276793e-01 -6.36436939e-02
5.16446710e-01 6.52448118e-01 -1.58463314e-01 -2.54378557e-01
3.36274773e-01 3.59317422e-01 -4.30357724e-01 7.16568828e-01
1.37947202e+00 4.95016903e-01 -2.03223959e-01 -3.70026976e-01
-5.62448800e-01 -3.95678550e-01 -1.47886848e+00 6.67216122e-01
-1.90868199e-01 1.04697275e+00 8.84551644e-01 -6.90551937e-01
1.07260084e+00 -1.93226278e-01 4.81464058e-01 -1.04884243e+00
1.57911912e-01 1.42080724e-01 2.18352139e-01 -2.88503528e-01
7.20111787e-01 -3.10042351e-01 -8.66545513e-02 -2.22584933e-01
-5.37939250e-01 -1.42537475e-01 3.95077094e-02 9.70928073e-02
1.46896112e+00 -3.20521116e-01 4.08832222e-01 -7.98937306e-03
6.61062717e-01 3.78765076e-01 7.65189052e-01 9.25404906e-01
-4.84324962e-01 -1.08344611e-02 -3.33758533e-01 -2.30818521e-02
-4.44804043e-01 -1.39105535e+00 -1.46502331e-01 4.89983320e-01
3.58780473e-01 1.10434964e-01 -2.50802547e-01 1.07360408e-01
3.65862489e-01 1.00969887e+00 -1.68977275e-01 -2.61318922e-01
-8.16825747e-01 -8.53839636e-01 1.78644955e-01 1.31043807e-01
4.18206364e-01 -2.69923419e-01 -1.21743536e+00 4.91880596e-01
-1.05037786e-01 -1.55331981e+00 -9.51239392e-02 1.28893018e-01
-6.33706331e-01 -9.58013773e-01 -1.74799800e-01 -2.58573085e-01
3.51484030e-01 1.32122576e+00 7.86842465e-01 -3.97361815e-01
-9.28912699e-01 8.57974887e-01 -8.82726982e-02 -1.00326645e+00
-1.04982732e-02 -9.76961255e-01 2.08440542e-01 4.68844295e-01
6.38246834e-01 -3.18030685e-01 -5.05580425e-01 2.93205023e-01
-6.82491839e-01 -2.45521545e-01 1.15094030e+00 4.57031816e-01
3.51128787e-01 5.65144479e-01 4.95504647e-01 -4.07580495e-01
3.30847353e-01 -2.63317496e-01 -1.06826234e+00 -3.41925740e-01
-4.98453639e-02 -5.04148781e-01 2.73517221e-01 -2.33258665e-01
-1.20992863e+00 1.67194724e-01 3.39746207e-01 -5.45274973e-01
-2.40636557e-01 4.57570851e-01 -2.70681173e-01 -4.51618761e-01
4.50775087e-01 3.95693958e-01 3.11608566e-03 2.41604652e-02
3.82343233e-01 5.34945428e-01 1.10622585e+00 -7.90842175e-02
1.71096575e+00 9.99972165e-01 4.59971547e-01 -1.56808937e+00
-6.38128579e-01 -9.18354273e-01 -5.22629738e-01 -5.74460089e-01
6.61178946e-01 -1.30821383e+00 -4.24820274e-01 3.27092171e-01
-8.32820237e-01 1.84660047e-01 6.42226413e-02 1.03617430e+00
-2.02224761e-01 4.47752416e-01 8.46076682e-02 -1.56049585e+00
-1.16278127e-01 -5.65792978e-01 7.63777316e-01 1.87474728e-01
8.95995833e-03 -5.71644783e-01 -3.13527100e-02 2.98618972e-01
6.17051840e-01 5.27663529e-01 6.37438059e-01 -2.23172203e-01
-1.11049509e+00 -6.01997495e-01 -1.73942834e-01 1.62707075e-01
1.01622723e-01 -3.05758923e-01 -8.28381717e-01 -4.12920117e-01
4.48276818e-01 5.35109460e-01 6.94888353e-01 7.91762173e-01
9.42005962e-02 3.14193904e-01 -5.82467914e-01 7.58512497e-01
1.47512686e+00 5.57664216e-01 5.26904523e-01 3.50175798e-01
3.34276259e-01 6.14030659e-01 1.39403713e+00 4.51410949e-01
1.44893583e-02 5.54490507e-01 4.61256146e-01 -7.23674893e-02
-4.06683348e-02 5.36892354e-01 3.65227133e-01 4.50175703e-01
2.30358988e-01 2.95808110e-02 -7.11063564e-01 4.63568479e-01
-1.56921768e+00 -1.26641309e+00 -4.81308222e-01 2.33947706e+00
-9.43783000e-02 5.24628699e-01 -3.03516716e-01 5.01325615e-02
7.40278244e-01 -7.83937201e-02 -6.86384678e-01 1.21499933e-01
-7.70296305e-02 3.69556636e-01 1.07104647e+00 6.30875289e-01
-1.19008708e+00 5.09291112e-01 5.38180637e+00 4.06447500e-01
-1.00643063e+00 -2.26356864e-01 -2.13007584e-01 -1.62051871e-01
1.45174801e-01 -1.33770138e-01 -1.09796655e+00 1.83204219e-01
1.08909619e+00 -4.88523424e-01 2.32376624e-02 5.16965091e-01
4.70252693e-01 -4.10220325e-01 -4.08635110e-01 1.02455139e+00
1.42236412e-01 -1.10953593e+00 -3.12255949e-01 4.00300980e-01
-3.55144665e-02 -4.29230109e-02 5.24202846e-02 4.96356338e-01
1.85585976e-01 -5.59428871e-01 7.56621599e-01 6.18961692e-01
3.74268562e-01 -1.09555125e+00 5.28795540e-01 5.11041462e-01
-1.41241038e+00 -1.76531374e-01 -1.91677511e-01 -1.56949967e-01
4.55984086e-01 1.00386214e+00 -7.15711534e-01 9.35081005e-01
4.08705115e-01 1.08545944e-01 -1.57417029e-01 1.33051169e+00
1.89073801e-01 4.54352558e-01 -2.62991935e-01 3.39515746e-01
2.40439668e-01 -4.07876521e-01 1.13224101e+00 1.45341027e+00
6.39971018e-01 6.10630870e-01 4.00393099e-01 4.40430552e-01
6.40619874e-01 -4.47589785e-01 -6.95668936e-01 1.67347714e-01
7.61107266e-01 1.44918466e+00 -5.56573331e-01 -9.81490836e-02
-5.08484304e-01 1.47012159e-01 -5.28598785e-01 7.33477056e-01
-1.08397913e+00 -8.44805956e-01 1.21094501e+00 2.42175702e-02
6.54121518e-01 -9.30750191e-01 -6.10032380e-02 -5.88577569e-01
8.96731485e-03 -7.89370000e-01 3.45187075e-02 -5.42543828e-01
-9.74986613e-01 3.80890280e-01 1.08247273e-01 -1.64025497e+00
-1.56903565e-02 -6.14014447e-01 -5.59558213e-01 1.34549880e+00
-1.97371531e+00 -8.80990922e-01 -7.40803421e-01 4.40593094e-01
3.85502160e-01 -2.73729682e-01 4.32450444e-01 3.62982363e-01
-3.16740125e-01 -5.60473017e-02 1.42976061e-01 -2.60048695e-02
6.65207267e-01 -5.66069305e-01 5.43666840e-01 1.54211247e+00
-2.91560709e-01 6.18378282e-01 1.11482465e+00 -1.01801693e+00
-1.94507265e+00 -1.42505074e+00 1.80264309e-01 -2.14003891e-01
7.56057262e-01 -4.95997667e-01 -6.83820307e-01 3.38316411e-01
-1.68227434e-01 -2.37982906e-02 8.02420497e-01 -4.81709361e-01
-4.28034030e-02 -1.45836368e-01 -8.52943540e-01 6.53720260e-01
6.79644883e-01 -1.16863158e-02 -7.07015634e-01 -5.38365915e-02
4.02486682e-01 -5.02784848e-01 -2.67625481e-01 8.21888208e-01
5.16781509e-01 -6.86671317e-01 1.30872726e+00 -1.84769660e-01
-4.37895149e-01 -9.16003644e-01 -6.23433650e-01 -1.09624350e+00
-7.15299189e-01 -7.67843604e-01 -3.13300490e-01 8.24458957e-01
-1.48963615e-01 -8.47082019e-01 7.35045493e-01 -8.06908906e-02
-3.17973703e-01 -3.26958150e-02 -1.02543771e+00 -1.02478433e+00
-5.55285037e-01 -8.12137485e-01 -2.81495228e-02 5.10294199e-01
-8.35437238e-01 3.20861220e-01 -2.18978658e-01 1.21041846e+00
1.50642061e+00 2.68780589e-01 1.32569385e+00 -1.42791963e+00
-1.56357780e-01 -2.74601616e-02 -5.94266891e-01 -9.00455534e-01
-7.82457218e-02 -3.56649518e-01 4.01819795e-01 -1.38309789e+00
-5.08864701e-01 -1.38547808e-01 -9.87806246e-02 -4.78629231e-01
-2.57996440e-01 2.19435662e-01 1.66537613e-01 1.15881510e-01
-7.13010952e-02 3.91950101e-01 7.43435800e-01 -5.81406280e-02
-6.46693707e-02 3.38411599e-01 -5.92230678e-01 7.63976395e-01
5.23202658e-01 -4.40202832e-01 -3.69209826e-01 -2.23438025e-01
-3.15669298e-01 2.26757880e-02 6.21227920e-01 -1.55643296e+00
4.16920155e-01 -3.83755893e-01 8.15113783e-01 -1.36263180e+00
7.68852592e-01 -9.99600589e-01 1.64893180e-01 6.26737595e-01
6.63161337e-01 -6.94174990e-02 6.64630294e-01 1.03085673e+00
-2.36754984e-01 6.53174445e-02 9.47313607e-01 2.50468671e-01
-1.15271628e+00 7.54561573e-02 -9.53033447e-01 -1.50183663e-01
1.32956219e+00 -5.05981803e-01 -4.73129570e-01 -4.44236189e-01
-3.95693660e-01 2.27227941e-01 2.70563085e-02 1.90944344e-01
6.83056831e-01 -1.12205791e+00 -8.59205008e-01 5.55078387e-01
1.90511137e-01 -5.31373620e-01 4.91615742e-01 9.12910104e-01
-4.17208135e-01 6.15503728e-01 -9.24280733e-02 -7.68567502e-01
-1.46776080e+00 2.63386726e-01 2.26447523e-01 1.50902569e-01
-7.33693302e-01 6.11227572e-01 4.09548700e-01 1.77702799e-01
3.16762701e-02 -1.18925072e-01 3.49209737e-03 -2.65148804e-02
9.85196829e-01 3.82069170e-01 2.71100879e-01 -7.74719715e-01
-5.67294896e-01 7.40312159e-01 2.99584176e-02 -3.97082418e-01
8.88313591e-01 -2.91952163e-01 2.56805867e-01 2.11762249e-01
6.31038010e-01 6.38438284e-01 -1.47439492e+00 -2.72992134e-01
1.04076847e-01 -7.25410938e-01 4.47566152e-01 -4.08165067e-01
-5.23759663e-01 6.39576972e-01 7.09754169e-01 5.67124449e-02
1.24307990e+00 -6.39817119e-01 5.66065729e-01 4.33521509e-01
3.70611221e-01 -8.58907104e-01 -1.51222482e-01 8.96452129e-01
3.43721718e-01 -7.22949862e-01 2.30932936e-01 -7.41737962e-01
-4.34883296e-01 9.84563887e-01 3.18282753e-01 -1.43695220e-01
5.32280684e-01 7.98182070e-01 4.91453439e-01 -1.72824591e-01
-6.03627443e-01 -5.75549245e-01 1.66034922e-01 8.96185696e-01
-1.52954981e-01 -9.56141204e-03 2.16287002e-01 3.12021583e-01
-1.62924811e-01 -3.46040785e-01 5.68076909e-01 1.43463612e+00
-1.20581126e+00 -4.98997957e-01 -1.22855806e+00 4.06146556e-01
-1.23559058e-01 2.00133119e-02 -6.63641021e-02 9.93051648e-01
-1.25005066e-01 1.31210232e+00 2.45452449e-01 -2.67194331e-01
8.36413801e-01 6.82511553e-02 4.05976981e-01 -3.73025775e-01
1.13410600e-01 4.43409115e-01 1.11392133e-01 -2.87098795e-01
-1.13122918e-01 -7.31120408e-01 -1.06382465e+00 6.72451919e-03
-2.74995953e-01 1.13639541e-01 9.58494425e-01 8.52970183e-01
4.15003717e-01 9.54023600e-01 6.25832975e-01 -7.98980176e-01
-5.41573048e-01 -4.63653237e-01 -5.73549509e-01 -1.54557273e-01
6.52028799e-01 -9.90739167e-01 -6.26677036e-01 -1.32826239e-01]
|
[6.7047014236450195, 0.9209728240966797]
|
8ff239c6-485a-423c-90af-e6028fe4b20a
|
crystal-graph-neural-networks-for-data-mining
| null | null |
https://www.researchgate.net/publication/333667001_Crystal_Graph_Neural_Networks_for_Data_Mining_in_Materials_Science
|
https://storage.googleapis.com/rimcs_cgnn/cgnn_matsci_May_27_2019.pdf
|
Crystal Graph Neural Networks for Data Mining in Materials Science
|
Machine learning methods have been employed for materials prediction in various ways. It has recently been proposed that a crystalline material is represented by a multigraph called a crystal graph. Convolutional neural networks adapted to those graphs have successfully predicted bulk properties of materials with the use of equilibrium bond distances as spatial information. An investigation into graph neural networks for small molecules has recently shown that the no distance model performs almost as well as the distance model. This paper proposes crystal graph neural networks (CGNNs) that use no bond distances, and introduces a scale-invariant graph coordinator that makes up crystal graphs for the CGNN models to be trained on the dataset based on a theoretical materials database. The CGNN models predict the bulk properties such as formation energy, unit cell volume, band gap, and total magnetization for every testing material, and the average errors are less than the corresponding ones of the database. The predicted band gaps and total magnetizations are used for the metal-insulator and nonmagnet-magnet binary classifications, which result in success. This paper presents discussions about high- throughput screening of candidate materials with the use of the predicted formation energies, and also about the future progress of materials data mining on the basis of the CGNN architectures.
|
['Takenori Yamamoto']
|
2019-05-27
| null | null | null |
technical-report-rimcs-llc-2019-5
|
['formation-energy']
|
['miscellaneous']
|
[ 8.14539194e-02 2.33342111e-01 -4.17409152e-01 -4.40222651e-01
2.55144946e-02 2.03018993e-01 4.54111308e-01 4.83380944e-01
-1.10340536e-01 9.62151647e-01 -1.68142021e-01 -3.82648438e-01
-3.86965990e-01 -1.38151336e+00 -9.13089156e-01 -1.05272400e+00
-3.33546370e-01 9.79678392e-01 2.81667799e-01 -3.75429541e-01
5.52273273e-01 6.97294772e-01 -1.80225301e+00 2.76172936e-01
1.00051844e+00 1.27763438e+00 3.03109318e-01 3.68453979e-01
-1.19057320e-01 6.39430106e-01 -3.67007226e-01 -5.29181361e-02
-1.43249379e-02 -2.60855138e-01 -6.37868464e-01 -3.24340940e-01
5.51830232e-01 1.37198478e-01 -4.71644074e-01 9.86194909e-01
4.24900115e-01 1.24857120e-01 1.04638314e+00 -8.54393959e-01
-1.31201541e+00 9.57816541e-01 -1.78979319e-02 -3.29397470e-02
2.59348810e-01 -4.64532264e-02 1.16330600e+00 -5.79267621e-01
8.21345866e-01 8.09987307e-01 4.85971540e-01 5.30942738e-01
-7.33825147e-01 -5.73368907e-01 -3.77392650e-01 7.52578795e-01
-1.11926365e+00 1.23881906e-01 9.56201434e-01 -4.18159068e-01
1.78161621e+00 1.99467108e-01 9.92424071e-01 5.58052778e-01
9.63797331e-01 2.91344732e-01 8.15051377e-01 -6.72746003e-01
1.69764951e-01 -3.02624345e-01 6.26083434e-01 1.03990436e+00
7.23725200e-01 1.43838063e-01 -4.45092201e-01 1.89130768e-01
4.22720432e-01 -6.72414824e-02 -9.42842960e-02 -3.15337032e-01
-7.28664339e-01 9.48257327e-01 1.09284925e+00 6.73986912e-01
-2.84934819e-01 4.59347218e-02 1.92782521e-01 1.95056975e-01
4.75969881e-01 8.24987590e-01 -1.30002096e-01 5.51420510e-01
-6.03110015e-01 2.36495003e-01 6.87731922e-01 7.37472475e-01
8.32211494e-01 2.33913645e-01 1.55784264e-01 5.57549536e-01
1.86367422e-01 4.19934988e-01 5.58957279e-01 -6.09598160e-02
2.04469740e-01 1.10125411e+00 -3.49042445e-01 -1.19990075e+00
-1.05812061e+00 -2.89572984e-01 -1.11757755e+00 -3.93461660e-02
1.01135962e-01 3.97190243e-01 -1.15617692e+00 1.19745517e+00
1.38091922e-01 -3.42390716e-01 -1.33886859e-01 7.62940645e-01
1.57532644e+00 8.08774889e-01 -2.22822726e-01 -1.37525916e-01
6.86811745e-01 -9.80767727e-01 -5.45118392e-01 2.03344971e-02
8.80270481e-01 -6.77360520e-02 7.06500292e-01 4.37403053e-01
-1.01467645e+00 -7.34727383e-01 -1.51238060e+00 1.29637554e-01
-9.61724520e-01 -1.63982570e-01 1.15708113e+00 6.11094713e-01
-9.73805428e-01 1.28959990e+00 -7.75721848e-01 -2.18237162e-01
1.52714148e-01 1.02468503e+00 -3.29659820e-01 6.51395097e-02
-1.34213686e+00 8.92686665e-01 8.01329553e-01 1.14853106e-01
-4.46757734e-01 -1.86671838e-01 -5.98207474e-01 -1.92979202e-01
-1.56222448e-01 -3.74121368e-01 4.31525558e-01 -7.02807963e-01
-1.13476789e+00 8.01980257e-01 2.42799908e-01 -6.31156802e-01
-1.89243034e-01 4.48389530e-01 -8.10459554e-01 -3.62127051e-02
3.19668874e-02 5.39527714e-01 3.43044698e-01 -6.29741251e-01
-5.01905642e-02 -3.38250130e-01 -2.82248527e-01 -1.29121870e-01
-4.67701435e-01 -5.01591563e-01 3.97055075e-02 -1.57963350e-01
4.78574812e-01 -8.30977023e-01 -3.33107978e-01 -7.77816296e-01
-7.59852946e-01 -7.44373977e-01 4.69412714e-01 -6.08600676e-01
1.10547054e+00 -1.40858686e+00 2.31723785e-01 7.01155782e-01
5.66425979e-01 1.57417849e-01 5.76369558e-03 5.10436356e-01
-3.13628793e-01 -3.47261243e-02 -7.32598454e-02 7.32013345e-01
-3.48547012e-01 -1.93844244e-01 4.06798899e-01 4.10701782e-01
-4.47835252e-02 1.03962243e+00 -5.49581647e-01 -1.39835593e-03
3.01875651e-01 2.58842498e-01 -4.72704828e-01 -8.49004909e-02
-6.84083164e-01 2.85314471e-01 -4.14916933e-01 6.40739918e-01
7.94428051e-01 -6.22392833e-01 2.88630605e-01 -3.26241851e-01
-7.35806599e-02 5.09155631e-01 -6.50005698e-01 1.15045881e+00
2.33277053e-01 4.52944905e-01 -6.52761817e-01 -1.33752394e+00
1.49826324e+00 -1.57292206e-02 6.66646421e-01 -1.27917683e+00
4.25683707e-02 6.01696134e-01 5.61002612e-01 -4.74964231e-01
3.55990052e-01 9.69156995e-02 3.93584490e-01 2.01109186e-01
2.15436265e-01 -8.84077251e-02 5.47352374e-01 -3.75800356e-02
9.24604774e-01 -1.48030117e-01 9.67589766e-02 -5.77658594e-01
5.60709774e-01 1.61376461e-01 -9.11611542e-02 7.21372008e-01
5.32522500e-01 2.21153513e-01 1.71186030e-01 -1.12551951e+00
-1.23977184e+00 -8.23034883e-01 -2.59040713e-01 6.63626909e-01
1.98328122e-01 -5.46521306e-01 -7.49667525e-01 -1.37733430e-01
7.78918788e-02 2.41534457e-01 -5.30493081e-01 -5.90608537e-01
-6.27663910e-01 -1.17569661e+00 1.13762720e-02 4.71394211e-01
4.50599134e-01 -1.56428766e+00 -1.10913485e-01 3.11386406e-01
4.29539263e-01 -7.47139931e-01 2.46263027e-01 7.45490491e-01
-8.83978605e-01 -1.21920335e+00 -1.68416336e-01 -1.14228463e+00
6.38412118e-01 -1.90651566e-01 1.05719435e+00 4.32292610e-01
-4.63022590e-01 -4.10230815e-01 -2.87415415e-01 -3.55261862e-01
-6.19937062e-01 3.66740286e-01 3.08461457e-01 -4.08642471e-01
5.98881662e-01 -6.38885975e-01 -4.79488343e-01 -7.57143497e-02
-5.81041336e-01 2.89278120e-01 5.13085485e-01 7.55244792e-01
8.19216728e-01 1.69267207e-01 6.78826272e-01 -1.10142779e+00
7.16320157e-01 -4.03485924e-01 -5.82381070e-01 2.96010464e-01
-1.18613887e+00 4.81028080e-01 8.88000667e-01 -1.63636133e-01
-3.78719956e-01 -1.41717345e-02 -3.36583644e-01 7.43038878e-02
1.43792406e-01 8.69025111e-01 -1.11397132e-01 -5.13291597e-01
7.50262380e-01 1.60800934e-01 -2.93034881e-01 -2.35697716e-01
6.78249747e-02 7.43605375e-01 2.20154166e-01 -4.30821002e-01
3.10667843e-01 2.78088879e-02 7.06622839e-01 -1.09495533e+00
-6.23312652e-01 -1.34802330e-02 -8.32495987e-01 -4.09269750e-01
9.53003287e-01 -3.98387820e-01 -9.52765346e-01 5.25512993e-01
-8.85914028e-01 -2.77426913e-02 1.30501643e-01 5.52213848e-01
-4.04681444e-01 2.10193813e-01 -7.41776526e-01 -4.57974166e-01
-8.79657269e-01 -1.24188554e+00 4.70744133e-01 1.97000012e-01
1.99535694e-02 -1.10802126e+00 -4.43224870e-02 3.52753758e-01
3.32759142e-01 3.96843582e-01 1.60191905e+00 -8.50156724e-01
-7.89284289e-01 -3.85284483e-01 3.98400612e-02 1.88591808e-01
2.56350309e-01 -4.95014079e-02 -6.01434112e-01 -3.16967487e-01
-3.30119610e-01 -1.67873755e-01 1.21985245e+00 9.81887102e-01
1.24682069e+00 -3.21487933e-02 -4.82370943e-01 7.00512290e-01
1.44326150e+00 6.30670369e-01 7.24626482e-01 5.73342979e-01
1.22733831e+00 2.15498641e-01 1.15512975e-01 -2.18407009e-02
-5.33532053e-02 3.93413723e-01 7.09631562e-01 9.63822603e-02
1.28533065e-01 -3.82301919e-02 7.53793716e-02 1.25375640e+00
-6.03551447e-01 -4.60543841e-01 -1.10007453e+00 4.59287912e-02
-1.55146670e+00 -8.23180139e-01 -6.72854424e-01 2.05327630e+00
3.87885362e-01 6.50444925e-01 4.42634970e-02 1.59229547e-01
7.84707665e-01 1.70039386e-01 -8.73532414e-01 -6.17500126e-01
-4.26456392e-01 7.36852467e-01 7.94951379e-01 2.77824730e-01
-9.33742344e-01 9.00048852e-01 6.54676485e+00 6.33865118e-01
-1.38821530e+00 -4.39879656e-01 5.55960655e-01 1.64441377e-01
-3.01198453e-01 -8.22758079e-02 -8.60340178e-01 3.78327459e-01
1.29771173e+00 1.54788971e-01 4.91103768e-01 8.19023192e-01
-2.50530958e-01 1.57259777e-01 -1.07160461e+00 9.99151587e-01
7.27075860e-02 -2.14436650e+00 4.28015471e-01 1.88280955e-01
8.02366793e-01 3.37754756e-01 -9.43586007e-02 -4.61181849e-02
-6.17992021e-02 -1.30524123e+00 4.46428269e-01 6.49018347e-01
7.45042503e-01 -7.98115671e-01 6.89765453e-01 -4.78338934e-02
-1.26256907e+00 4.94143143e-02 -7.15313494e-01 -2.33316302e-01
-2.11580291e-01 7.29194582e-01 -1.01917517e+00 6.48896098e-01
5.97757220e-01 9.21887755e-01 -7.85774410e-01 8.94948959e-01
1.07445784e-01 3.25215220e-01 -2.01628089e-01 -8.38411510e-01
2.15301886e-01 -7.31684327e-01 6.87502250e-02 6.43008828e-01
3.26190233e-01 -3.65149021e-01 5.14790080e-02 9.82473373e-01
-3.50898266e-01 3.48257035e-01 -7.50917673e-01 -4.61557686e-01
2.85761416e-01 9.34318960e-01 -9.04259741e-01 -2.00039595e-01
-2.97701418e-01 3.95141721e-01 3.78402770e-01 -4.70201336e-02
-4.28726077e-01 -4.60102558e-01 -4.25614864e-02 4.69080299e-01
1.80625558e-01 -2.21917685e-03 -5.99415004e-02 -6.67191088e-01
1.16616659e-01 -4.99923348e-01 1.88644305e-01 -8.13238502e-01
-1.26818204e+00 7.25895643e-01 -1.97321489e-01 -7.83311725e-01
-1.57860830e-01 -1.36276734e+00 -6.32955015e-01 5.81091166e-01
-1.03076220e+00 -9.69669044e-01 -7.85338804e-02 2.37372056e-01
-1.48879774e-02 -4.73885268e-01 7.90503621e-01 7.63131753e-02
-4.86657292e-01 2.19897792e-01 6.48198485e-01 1.37589984e-02
1.68461338e-01 -1.19224763e+00 6.37583196e-01 1.91944927e-01
9.40270349e-02 3.91001791e-01 6.64166987e-01 -9.93584991e-01
-1.51745653e+00 -1.06945252e+00 7.39929914e-01 -1.04081146e-02
5.33608019e-01 -3.91247392e-01 -9.11809802e-01 3.72734398e-01
2.78751999e-02 -1.82271674e-01 4.21315521e-01 -1.64563134e-02
2.13477071e-02 -1.18304938e-01 -9.36963320e-01 3.13488096e-01
1.11786795e+00 -3.51971209e-01 -1.44240865e-02 9.68247831e-01
7.15981483e-01 -2.52429426e-01 -1.04461098e+00 5.78872025e-01
3.99158806e-01 -1.02891207e+00 8.18141103e-01 -8.23021352e-01
4.39897239e-01 1.11158967e-01 -1.53318569e-01 -1.15629089e+00
-5.60765028e-01 5.82947917e-02 -1.78549245e-01 5.17981887e-01
6.89585149e-01 -5.95059156e-01 1.15619695e+00 2.75985122e-01
-4.19184357e-01 -1.19572663e+00 -9.71622050e-01 -7.90477455e-01
1.35825306e-01 6.72780117e-03 8.52755010e-01 7.80175805e-01
-3.29135247e-02 7.45623767e-01 -2.60642827e-01 -1.27852365e-01
4.10698593e-01 4.18188065e-01 1.68638766e-01 -1.95110142e+00
-1.17068619e-01 -3.24234366e-01 -9.16219592e-01 -5.15198708e-01
3.53578269e-01 -1.53198719e+00 -4.16077316e-01 -1.80342114e+00
2.06292316e-01 -3.41204435e-01 -4.52851921e-01 7.63325393e-02
4.70504105e-01 -1.20320227e-02 -3.80846888e-01 1.65994778e-01
-4.11847621e-01 4.08386588e-01 1.52938509e+00 -6.18784964e-01
-2.45390147e-01 -1.37975082e-01 -2.65921444e-01 4.24490839e-01
9.59450483e-01 -2.19372466e-01 -1.41302690e-01 -3.76839824e-02
7.87915349e-01 1.00840395e-03 -1.61406428e-01 -1.59116066e+00
1.37292728e-01 2.68569868e-02 6.62218928e-01 -9.28942859e-01
3.46185416e-01 -4.57345277e-01 3.38768125e-01 8.45100880e-01
-2.06129521e-01 8.04689750e-02 -2.65295833e-01 3.86380911e-01
1.60487685e-02 -4.01981562e-01 7.28407800e-01 -3.71146977e-01
-5.36198676e-01 5.82811654e-01 -2.80869335e-01 -6.61165416e-01
8.52113664e-01 -6.00833416e-01 -3.11849952e-01 -7.41052777e-02
-9.00764406e-01 -1.96941614e-01 3.22965860e-01 1.96018964e-01
7.73330629e-01 -1.43528831e+00 -3.10611159e-01 4.61581409e-01
9.60609168e-02 -1.40636772e-01 1.04091570e-01 4.65336829e-01
-1.02788687e+00 8.99883568e-01 -6.62256360e-01 -5.98856091e-01
-1.03469706e+00 8.20155203e-01 6.02562904e-01 -3.57607901e-01
-4.04097825e-01 6.93690360e-01 -1.97510451e-01 -5.01534164e-01
-1.28516406e-01 -4.99289632e-01 -4.86709774e-01 -3.46110642e-01
7.47155771e-02 2.28428811e-01 7.87297785e-01 -6.67242765e-01
-3.07291180e-01 6.32250667e-01 -2.43557081e-01 7.64069855e-01
1.93291283e+00 5.90930879e-01 -7.18791783e-01 4.03816462e-01
1.14455795e+00 -4.98076141e-01 -3.40612173e-01 2.42809448e-02
6.21053278e-02 2.82410234e-01 1.15814600e-02 -5.30731440e-01
-1.27197683e+00 7.51009762e-01 9.27638710e-01 4.63629365e-01
7.85952926e-01 1.01620167e-01 8.24353456e-01 8.88452768e-01
3.73622656e-01 -1.26295578e+00 3.24270576e-02 5.57446778e-01
6.04169190e-01 -1.10037470e+00 1.50085941e-01 -3.73684853e-01
4.18016762e-02 1.65411949e+00 9.67997611e-01 -3.13510358e-01
9.00973618e-01 -1.80437237e-01 -6.19962871e-01 -8.96453857e-01
-4.98614550e-01 6.30068704e-02 4.84171361e-01 6.99701548e-01
7.47909665e-01 5.05460203e-01 -4.29318637e-01 2.39700377e-01
-5.02517581e-01 -3.32422942e-01 3.07106882e-01 6.99433506e-01
-9.04076934e-01 -1.21054220e+00 -1.48320764e-01 1.23321009e+00
-8.65568295e-02 -2.97876120e-01 -8.08242083e-01 6.57229543e-01
8.00626203e-02 6.84404194e-01 1.57203630e-01 -9.30857956e-01
-1.84719861e-02 6.19017379e-03 8.15838695e-01 -4.21684206e-01
-3.18599612e-01 -5.28774202e-01 2.23960262e-02 -2.43713021e-01
-5.74016511e-01 7.36841410e-02 -1.61181939e+00 -4.13733870e-01
-8.32314372e-01 5.49297154e-01 7.16747999e-01 9.60677862e-01
1.18265212e-01 5.89617848e-01 4.94087458e-01 -9.76296663e-01
5.42585291e-02 -1.17364442e+00 -9.92309093e-01 4.42866325e-01
9.32908878e-02 -6.20276868e-01 -1.00837417e-01 -5.80249250e-01]
|
[5.335961818695068, 5.524656295776367]
|
814867a8-26ab-426a-bb3b-9b85f54935f0
|
replacing-language-model-for-style-transfer
|
2211.07343
| null |
https://arxiv.org/abs/2211.07343v1
|
https://arxiv.org/pdf/2211.07343v1.pdf
|
Replacing Language Model for Style Transfer
|
We introduce replacing language model (RLM), a sequence-to-sequence language modeling framework for text style transfer. Our method autoregressively replaces each token in the original sentence with a text span in the target style. In contrast, the new span is generated via a non-autoregressive masked language model. The RLM generation scheme gathers the flexibility of autoregressive models and the accuracy of non-autoregressive models, which bridges the gap between sentence-level and word-level style transfer methods. To further control the style of generated sentences, we conduct a style-content disentanglement on the hidden representations of RLM. Empirical results on real-world text style transfer tasks demonstrate the effectiveness of RLM compared with other baselines.
|
['Ruineng Li', 'Pengyu Cheng']
|
2022-11-14
| null | null | null | null |
['text-style-transfoer']
|
['natural-language-processing']
|
[ 4.68681723e-01 1.42596871e-01 -1.14307381e-01 -6.10690176e-01
-8.00900042e-01 -5.40429354e-01 9.23081636e-01 -5.07342935e-01
-3.30810308e-01 7.84922183e-01 7.71732628e-01 -4.56092834e-01
7.82239079e-01 -7.43792713e-01 -5.52839100e-01 -3.23311657e-01
5.52227139e-01 5.12504458e-01 -2.11621732e-01 -4.93873835e-01
1.65714592e-01 1.24156855e-01 -5.79340279e-01 6.18632436e-01
8.72451782e-01 3.81055981e-01 3.32562715e-01 8.19963276e-01
-6.52222574e-01 1.00016081e+00 -8.22233915e-01 -4.51946348e-01
-3.94946709e-02 -8.77938509e-01 -7.24567890e-01 -1.08434252e-01
3.08818012e-01 -4.57866669e-01 -6.08688176e-01 7.32980013e-01
4.28243548e-01 1.47606388e-01 1.05546641e+00 -1.05917513e+00
-1.27179396e+00 7.60587931e-01 -5.69659173e-01 -8.90680328e-02
4.38331217e-01 2.04541013e-01 8.22074711e-01 -1.00000668e+00
5.38436830e-01 1.97837067e+00 3.91138136e-01 1.08145905e+00
-1.68437350e+00 -7.50980735e-01 3.88794750e-01 -4.61770952e-01
-8.91412258e-01 -4.73674864e-01 8.95569026e-01 -4.35780615e-01
9.43404555e-01 -7.18737673e-03 2.46958986e-01 1.60325766e+00
6.28813744e-01 8.70458782e-01 1.24523163e+00 -7.32123315e-01
1.32314757e-01 1.62384331e-01 1.12128384e-01 4.77179676e-01
-1.73378021e-01 5.05636744e-02 -7.57762015e-01 -5.64386956e-02
8.98477674e-01 -2.35547915e-01 1.32378623e-01 -9.22578275e-02
-9.78911161e-01 1.06618857e+00 9.11327899e-02 1.53509527e-01
-2.04031572e-01 1.82415083e-01 5.85690141e-01 6.37673020e-01
8.74298215e-01 4.32701617e-01 -4.22045141e-01 -2.36975193e-01
-9.92822170e-01 7.80506358e-02 6.83954597e-01 1.40775084e+00
6.34225845e-01 6.40076220e-01 -8.36772263e-01 1.00612342e+00
3.09579402e-01 6.79207087e-01 6.78444505e-01 -5.95839024e-01
6.89213574e-01 2.99262702e-01 -4.12374623e-02 -5.21783590e-01
1.71732321e-01 -6.85553178e-02 -9.61573243e-01 2.56489873e-01
1.44127354e-01 -2.92566061e-01 -8.04757655e-01 2.03345084e+00
-2.94585258e-01 -5.32286726e-02 7.78052658e-02 4.01832730e-01
4.77733999e-01 8.59377027e-01 5.78977704e-01 -2.15719491e-02
1.24616909e+00 -1.10361362e+00 -8.01392853e-01 -3.97722632e-01
6.60016119e-01 -7.57202923e-01 1.71635759e+00 -8.41767043e-02
-1.33468783e+00 -8.47730160e-01 -7.95328319e-01 -2.99432814e-01
-7.84067288e-02 1.77432343e-01 2.22860634e-01 5.90740025e-01
-1.22088802e+00 4.15009350e-01 -5.53287804e-01 -4.98452671e-02
2.37185448e-01 9.75082070e-03 -1.70541614e-01 1.05759285e-01
-1.51209557e+00 8.72712195e-01 1.90557972e-01 -1.97496533e-01
-7.35191166e-01 -1.02500570e+00 -1.11457825e+00 5.21101393e-02
-1.55362561e-01 -1.06699753e+00 1.47550845e+00 -1.32578659e+00
-2.05749512e+00 1.01579106e+00 -7.04285741e-01 -3.46934289e-01
6.85886860e-01 -6.52444661e-01 -2.31913969e-01 -2.28322759e-01
7.18944296e-02 8.59298408e-01 1.18707335e+00 -1.07486928e+00
-1.06795691e-01 1.31864250e-01 -1.64085433e-01 2.05760151e-01
-4.03283924e-01 2.94602007e-01 -1.13477208e-01 -1.32076597e+00
-7.20126927e-01 -9.50721622e-01 -2.42388621e-01 -2.80077368e-01
-2.31073827e-01 -2.32493833e-01 5.67047894e-01 -8.89704943e-01
1.30339801e+00 -2.21551681e+00 4.43421125e-01 -2.39306495e-01
-2.91206148e-02 2.27448568e-01 -7.50858188e-01 7.39522815e-01
-5.00491112e-02 2.58451432e-01 -2.05198735e-01 -7.87503064e-01
6.09294279e-03 1.86968327e-01 -8.86428058e-01 -1.80642933e-01
4.60305274e-01 1.32741809e+00 -8.53017211e-01 -3.59287947e-01
4.15699407e-02 3.96446079e-01 -7.48368859e-01 6.63622081e-01
-3.73725593e-01 5.50821483e-01 -3.35985482e-01 -1.47002324e-01
3.96923304e-01 4.66421340e-03 6.14125840e-02 1.42741486e-01
1.13320880e-01 7.82103479e-01 -4.54693347e-01 1.98128211e+00
-1.02577507e+00 5.54176033e-01 -2.38975555e-01 -2.47299075e-01
1.16520309e+00 3.87554735e-01 -2.99031436e-01 -4.71354723e-01
-9.97065976e-02 -1.88993052e-01 -1.27256677e-01 -4.37004082e-02
9.12492275e-01 -4.94226754e-01 -4.35007304e-01 7.40514040e-01
2.21341535e-01 -3.16008866e-01 -1.60174251e-01 4.22795862e-01
6.70719862e-01 3.46143186e-01 3.12058896e-01 -4.27012354e-01
5.01667142e-01 -4.54106987e-01 3.40642869e-01 1.06923437e+00
-3.34319361e-02 5.73470592e-01 4.81585354e-01 -1.12209648e-01
-1.24120069e+00 -1.30694377e+00 4.78533685e-01 1.57038653e+00
-4.06608194e-01 -3.93606752e-01 -9.34334815e-01 -7.98702776e-01
-1.49393275e-01 1.30443239e+00 -7.83854663e-01 -5.54068089e-01
-7.94224262e-01 -1.06505655e-01 7.21284926e-01 7.88824856e-01
2.58271098e-01 -1.38743472e+00 1.46271661e-01 2.74353743e-01
-2.33314037e-01 -8.73333752e-01 -1.23926985e+00 -4.35343236e-01
-8.23650420e-01 -2.24472851e-01 -7.76902854e-01 -8.36540163e-01
6.89054966e-01 5.49422316e-02 1.56629348e+00 -1.88705400e-01
1.26496524e-01 2.83902615e-01 -2.52230138e-01 -4.53735352e-01
-1.06148148e+00 3.44773859e-01 2.52895001e-02 -5.56540117e-02
3.94831032e-01 -5.68948507e-01 -2.43976638e-01 -7.08435774e-02
-8.93364131e-01 3.10059994e-01 5.35479724e-01 9.99055862e-01
6.66774958e-02 -6.47457719e-01 9.04611290e-01 -1.11792719e+00
1.06283414e+00 -2.57815629e-01 -3.68587434e-01 1.91853076e-01
-3.41306806e-01 3.73630077e-01 6.97708249e-01 -7.90795386e-01
-1.58560383e+00 -1.88244089e-01 -1.71723917e-01 -1.90951452e-01
-1.97368652e-01 2.27482289e-01 -4.51156527e-01 4.69043344e-01
4.07248259e-01 4.80322897e-01 1.29485086e-01 -4.95599657e-01
9.72660482e-01 6.76960230e-01 3.55780303e-01 -7.52970219e-01
9.47473407e-01 1.02674194e-01 -4.18726861e-01 -8.43556643e-01
-7.97278881e-01 6.65444806e-02 -7.59298325e-01 1.64573997e-01
8.72797132e-01 -1.08285832e+00 -2.50433683e-01 5.54796636e-01
-1.58201027e+00 -7.82694280e-01 -4.71385568e-01 1.73337176e-01
-7.98776150e-01 3.66983980e-01 -1.15442157e+00 -9.15421307e-01
-5.33440769e-01 -8.63491952e-01 1.09576738e+00 -1.23807393e-01
-9.93179739e-01 -1.34859800e+00 1.43058658e-01 6.63334727e-02
6.93239391e-01 -3.21912855e-01 1.38540995e+00 -6.31305039e-01
-1.31129429e-01 9.90814120e-02 -4.87713963e-02 4.10673618e-01
2.60035515e-01 -3.97632509e-01 -8.79981399e-01 -3.60511750e-01
5.16058356e-02 -3.59089971e-01 8.07223678e-01 1.93022877e-01
8.83745909e-01 -5.59937656e-01 1.26518622e-01 4.04463351e-01
9.16811705e-01 1.45657539e-01 8.63402188e-01 -3.51327844e-02
9.47705746e-01 5.96197605e-01 1.45455956e-01 2.06511497e-01
2.59593397e-01 5.76866925e-01 -4.52831358e-01 -2.79808819e-01
-3.13484430e-01 -9.33304250e-01 8.61588717e-01 1.14998388e+00
3.42062235e-01 -2.96378285e-01 -4.21396166e-01 2.27851108e-01
-1.45566940e+00 -9.36551750e-01 4.67349403e-02 1.93550909e+00
1.30596769e+00 2.55756825e-01 1.75943568e-01 -3.08316618e-01
6.15680218e-01 3.08969110e-01 -3.39607686e-01 -8.80169094e-01
-1.61621764e-01 2.66066164e-01 1.14699332e-02 9.38657641e-01
-8.18400860e-01 1.55943716e+00 7.20794678e+00 9.07248139e-01
-8.92624497e-01 2.81897057e-02 6.29252613e-01 -1.82746295e-02
-7.16071665e-01 -4.41110060e-02 -9.69383657e-01 6.47707880e-01
1.13181150e+00 -4.06317770e-01 6.15475237e-01 7.39517391e-01
5.66121042e-01 3.97999138e-01 -1.41533566e+00 6.89863682e-01
2.41857708e-01 -1.04915142e+00 7.46459544e-01 -8.37810114e-02
7.38774955e-01 -3.58112693e-01 8.75659660e-02 7.66443312e-01
8.04248095e-01 -1.09272015e+00 7.85501301e-01 7.45022595e-01
1.02984750e+00 -6.08590007e-01 2.32913777e-01 3.12476873e-01
-9.20691550e-01 2.57342905e-01 -2.47982055e-01 -3.26663375e-01
4.48328763e-01 3.16654295e-01 -6.09361529e-01 3.80896717e-01
3.75201670e-03 6.33184731e-01 -5.34521043e-01 -2.19094269e-02
-4.95264560e-01 1.07652307e+00 5.15930235e-01 4.90731858e-02
1.27742495e-02 -4.84984845e-01 6.48739159e-01 1.78286421e+00
-2.94781057e-04 -2.18582511e-01 8.28116536e-02 1.18882227e+00
-3.93362910e-01 1.62668750e-01 -9.09915209e-01 -2.05192402e-01
4.64005113e-01 8.82749438e-01 -1.54398218e-01 -7.39432514e-01
-5.32636940e-01 1.55126691e+00 5.13044894e-01 7.85020113e-01
-4.75054294e-01 -4.29896206e-01 7.65780926e-01 -2.61017680e-02
3.52108665e-02 -4.96320426e-01 -6.73295557e-01 -1.30660152e+00
-1.16807409e-01 -1.07983363e+00 7.81081617e-02 -9.06799793e-01
-1.82938910e+00 7.15733707e-01 -4.13182527e-02 -9.71765995e-01
-6.81401730e-01 -4.59869295e-01 -9.18318272e-01 1.60483897e+00
-1.28413367e+00 -1.43718290e+00 2.02307925e-01 1.90491378e-01
1.23189533e+00 -4.86882776e-01 9.69588220e-01 -2.10206315e-01
-2.07183942e-01 9.32818353e-01 -4.84619513e-02 4.38419938e-01
9.88648236e-01 -1.26871717e+00 1.20660555e+00 6.87264621e-01
-1.59815788e-01 1.00500917e+00 5.72201610e-01 -9.05962408e-01
-1.00068688e+00 -1.45351970e+00 1.25236762e+00 -7.49992251e-01
7.42764175e-01 -1.01315796e+00 -1.28011763e+00 8.64223480e-01
7.82809496e-01 -7.17265129e-01 8.20847273e-01 5.99171929e-02
-5.74576318e-01 2.53370672e-01 -6.79278851e-01 1.05288672e+00
8.88790607e-01 -9.79726076e-01 -9.54210699e-01 -1.31040037e-01
1.21342969e+00 2.95903496e-02 -5.68798900e-01 -4.82480898e-02
5.38934410e-01 -4.91340935e-01 7.81816781e-01 -1.27435994e+00
8.31853628e-01 8.78849030e-02 -2.40621716e-02 -1.78867424e+00
-6.70207798e-01 -9.65301037e-01 -2.54417728e-06 1.79868841e+00
4.35347646e-01 -5.77216625e-01 4.81611460e-01 7.24685252e-01
5.08991219e-02 -3.27679247e-01 -4.94120002e-01 -9.24460232e-01
9.35985744e-01 -1.33502141e-01 6.33076429e-01 7.32387483e-01
-1.42820477e-01 9.54259276e-01 -7.07876325e-01 -4.35013533e-01
3.12266588e-01 -1.64961383e-01 7.81884432e-01 -8.65814507e-01
-7.09707975e-01 -6.19711220e-01 1.92840725e-01 -1.37863624e+00
8.30828011e-01 -9.53584969e-01 9.41451937e-02 -1.35727751e+00
2.56310314e-01 -2.93590128e-03 -1.07279189e-01 2.14986444e-01
-6.68662727e-01 -9.15191695e-02 5.37998497e-01 9.83598977e-02
-2.53907055e-01 1.03045607e+00 1.12084961e+00 -1.23455673e-01
-4.33534622e-01 -3.84028815e-02 -7.94877291e-01 7.06075370e-01
9.03545797e-01 -3.47441345e-01 -5.33386409e-01 -6.44306958e-01
-1.19984098e-01 1.09282471e-01 -3.88143887e-03 -2.54844278e-01
-3.25806916e-01 -3.14067334e-01 4.12776500e-01 -3.86188686e-01
2.03703955e-01 -2.24578753e-01 -3.29527766e-01 3.46148431e-01
-1.23414791e+00 2.48023689e-01 1.77128360e-01 4.59428489e-01
-8.76500458e-03 -7.70388171e-02 8.15328777e-01 -1.03869878e-01
-2.24455953e-01 1.92590907e-01 -9.29057300e-01 3.52984875e-01
3.43128085e-01 6.39299452e-02 -5.34678958e-02 -6.93011463e-01
-3.84827971e-01 2.34883502e-02 5.54223716e-01 9.02651548e-01
4.89768833e-01 -1.58886635e+00 -9.82738793e-01 3.66232306e-01
4.81715687e-02 -4.16890919e-01 1.44873649e-01 1.99565664e-01
1.11217916e-01 2.84761041e-01 -6.21343106e-02 -1.91425070e-01
-9.72519815e-01 7.73853302e-01 7.68144727e-02 -5.61083853e-01
-5.05224705e-01 5.14347076e-01 9.00466323e-01 -5.39009213e-01
2.30994374e-02 5.20343520e-02 -1.59938224e-02 -3.41018021e-01
8.37947607e-01 2.85044163e-01 -2.92421728e-01 -4.73497301e-01
8.32121149e-02 1.73109889e-01 -3.64717245e-01 -6.33155107e-01
9.01662350e-01 -4.56958681e-01 -1.30051106e-01 1.05971217e+00
1.15618229e+00 2.09257722e-01 -1.37377036e+00 -3.93462837e-01
-2.67142691e-02 -1.79211482e-01 -2.63115436e-01 -8.06920409e-01
-3.69777471e-01 1.19256985e+00 4.63706888e-02 -2.55807370e-01
8.63365769e-01 -3.02634299e-01 8.75714660e-01 1.79429680e-01
2.89632455e-02 -9.31723535e-01 3.49199444e-01 1.03121912e+00
1.27429461e+00 -8.57193589e-01 -6.12344325e-01 -2.71401376e-01
-1.06290746e+00 8.74601066e-01 7.80599475e-01 -3.43030632e-01
4.15969044e-01 4.48108107e-01 4.16596144e-01 6.41584873e-01
-9.52953517e-01 3.17266077e-01 2.48218238e-01 7.02828228e-01
9.74491596e-01 6.66109100e-03 -2.43426427e-01 6.87829375e-01
-4.47031230e-01 9.44960862e-02 3.48487467e-01 7.05996454e-01
-1.94830090e-01 -1.47491646e+00 -2.39018291e-01 2.12567851e-01
-2.32656524e-01 -7.12707639e-01 -5.86696506e-01 3.34711820e-01
-6.78482354e-01 9.65524554e-01 1.76787540e-01 -1.65086448e-01
2.98482388e-01 5.39006710e-01 3.32617313e-01 -7.71680593e-01
-6.21809781e-01 1.87982991e-01 7.96122774e-02 -2.02028468e-01
2.20113680e-01 -3.71813327e-01 -8.87417853e-01 -1.03117943e-01
2.24890754e-01 9.60552022e-02 2.79022068e-01 9.55822766e-01
5.35982251e-01 6.86282873e-01 8.64615977e-01 -9.17786241e-01
-8.39146793e-01 -1.38079548e+00 -4.42583889e-01 6.63481593e-01
2.92402416e-01 -2.11344063e-01 3.17997001e-02 4.98656183e-01]
|
[11.665252685546875, 9.494989395141602]
|
2a005b3c-5167-4281-9429-91bbac9bb0c4
|
self-supervised-non-uniform-kernel-estimation
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Fang_Self-Supervised_Non-Uniform_Kernel_Estimation_With_Flow-Based_Motion_Prior_for_Blind_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Fang_Self-Supervised_Non-Uniform_Kernel_Estimation_With_Flow-Based_Motion_Prior_for_Blind_CVPR_2023_paper.pdf
|
Self-Supervised Non-Uniform Kernel Estimation With Flow-Based Motion Prior for Blind Image Deblurring
|
Many deep learning-based solutions to blind image deblurring estimate the blur representation and reconstruct the target image from its blurry observation. However, these methods suffer from severe performance degradation in real-world scenarios because they ignore important prior information about motion blur (e.g., real-world motion blur is diverse and spatially varying). Some methods have attempted to explicitly estimate non-uniform blur kernels by CNNs, but accurate estimation is still challenging due to the lack of ground truth about spatially varying blur kernels in real-world images. To address these issues, we propose to represent the field of motion blur kernels in a latent space by normalizing flows, and design CNNs to predict the latent codes instead of motion kernels. To further improve the accuracy and robustness of non-uniform kernel estimation, we introduce uncertainty learning into the process of estimating latent codes and propose a multi-scale kernel attention module to better integrate image features with estimated kernels. Extensive experimental results, especially on real-world blur datasets, demonstrate that our method achieves state-of-the-art results in terms of both subjective and objective quality as well as excellent generalization performance for non-uniform image deblurring. The code is available at https://see.xidian.edu.cn/faculty/wsdong/Projects/UFPNet.htm.
|
['Guangming Shi', 'Jinjian Wu', 'Xin Li', 'Weisheng Dong', 'Fangfang Wu', 'Zhenxuan Fang']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['deblurring', 'blind-image-deblurring']
|
['computer-vision', 'computer-vision']
|
[-2.29041159e-01 -8.75804186e-01 -7.32556581e-02 -2.61736661e-01
-4.94765669e-01 -5.72420716e-01 1.76041439e-01 -7.67633438e-01
-6.37528114e-03 8.17297399e-01 7.53166974e-01 -1.91825807e-01
-1.96120605e-01 -2.52534926e-01 -6.55759394e-01 -8.51621270e-01
1.38738811e-01 -3.08655232e-01 -5.02572767e-02 3.36454988e-01
2.46061951e-01 1.66901439e-01 -1.02097809e+00 6.47595972e-02
1.44094324e+00 1.00693178e+00 5.32709777e-01 7.24463522e-01
4.01735932e-01 1.16383314e+00 -5.06938279e-01 -1.20547101e-01
1.50297970e-01 -5.07706225e-01 -6.13937378e-01 2.22501040e-01
5.42518854e-01 -1.02551520e+00 -1.20987213e+00 1.62742090e+00
3.68000746e-01 -2.03165077e-02 5.86095333e-01 -9.16207254e-01
-1.69663107e+00 3.46309811e-01 -6.90134764e-01 4.68883932e-01
-3.54911052e-02 5.46289146e-01 5.05087674e-01 -9.08631206e-01
2.08360441e-02 1.09558952e+00 5.22305369e-01 3.30098927e-01
-1.02958262e+00 -7.31252611e-01 -1.13704287e-01 6.94244623e-01
-1.42420578e+00 -7.58284926e-01 6.20965064e-01 -5.24705708e-01
2.26562276e-01 3.58590275e-01 1.26056910e-01 1.05760336e+00
2.30556220e-01 8.16278458e-01 1.19721580e+00 1.24602430e-01
1.11291207e-01 -1.89605325e-01 1.10826492e-01 6.55930161e-01
4.77539957e-01 1.89355671e-01 -2.48755038e-01 1.52302850e-02
1.43954206e+00 1.86818689e-01 -1.20867670e+00 -2.17212737e-01
-1.56128812e+00 5.93024135e-01 9.02915597e-01 2.34313309e-01
-3.18739414e-01 4.33592290e-01 1.77846506e-01 1.72860883e-02
6.04625762e-01 4.12115961e-01 -3.23612720e-01 -2.23228872e-01
-1.04430711e+00 1.43222958e-01 3.30918580e-01 9.29019570e-01
8.02729905e-01 1.16305657e-01 -5.32507420e-01 9.63086247e-01
2.21227363e-01 5.22944868e-01 6.46962821e-01 -1.13944817e+00
3.07850480e-01 2.75364704e-02 7.22805738e-01 -9.01730895e-01
2.13189408e-01 -5.51712096e-01 -1.30466342e+00 -6.46015769e-03
4.66942877e-01 -2.45366529e-01 -1.11783624e+00 1.57123661e+00
-1.37981594e-01 9.96294498e-01 -9.50343013e-02 1.68964255e+00
5.90882361e-01 6.87994003e-01 -2.52478600e-01 -2.23497391e-01
1.32624924e+00 -1.38463151e+00 -1.03289425e+00 -3.99853677e-01
-8.69089514e-02 -9.52770412e-01 9.28536773e-01 9.25098285e-02
-9.52507615e-01 -8.04385006e-01 -9.46890593e-01 -1.97392881e-01
2.23401949e-01 6.18800759e-01 7.66700864e-01 2.94742227e-01
-1.05202472e+00 3.23458672e-01 -8.61480176e-01 1.91266183e-02
6.79620802e-01 -5.56623340e-02 -7.25060180e-02 -4.79426742e-01
-1.44790089e+00 8.64954352e-01 2.12454766e-01 4.52258974e-01
-1.11832762e+00 -6.77022099e-01 -9.26402450e-01 3.00744712e-01
1.31412372e-01 -9.11362886e-01 1.32251203e+00 -9.73171890e-01
-1.27646446e+00 2.12944075e-01 -3.82623672e-01 -2.16931939e-01
6.03026032e-01 -6.79915607e-01 -4.44556922e-01 8.62300247e-02
9.93136689e-02 2.67864227e-01 1.33579659e+00 -1.21086526e+00
-2.30469763e-01 6.60768598e-02 9.27142203e-02 3.81140225e-02
-2.80088425e-01 -7.65034184e-02 -4.59934831e-01 -1.10985470e+00
-9.83416215e-02 -6.64497554e-01 -3.56655754e-02 1.83024317e-01
-2.56179094e-01 2.11589962e-01 9.19943213e-01 -1.13148355e+00
1.38792872e+00 -2.22063637e+00 8.56789425e-02 -6.18966162e-01
5.37687659e-01 5.17506540e-01 -3.80212441e-02 -8.60825088e-03
-1.15812428e-01 -2.52817690e-01 -4.31670189e-01 -9.50599536e-02
-1.26202032e-01 -5.68399355e-02 -7.38978326e-01 8.02033007e-01
1.35594651e-01 1.06888926e+00 -1.12858868e+00 -4.77807559e-02
4.79961127e-01 6.23718500e-01 -2.21475378e-01 5.55836737e-01
-1.09865153e-02 6.45004272e-01 -3.28838348e-01 4.80087519e-01
1.20125544e+00 -6.50942862e-01 -4.16021258e-01 -6.07061625e-01
-8.32702890e-02 2.87477802e-02 -8.94881070e-01 1.90102696e+00
-5.39409578e-01 1.02811253e+00 9.13960785e-02 -6.48098111e-01
3.23126167e-01 3.43042761e-01 6.95323348e-02 -2.95231789e-02
1.13257989e-01 2.96676368e-01 -4.79957350e-02 -7.94654846e-01
4.74986553e-01 1.87923349e-02 3.97557199e-01 2.49791101e-01
-9.03456807e-02 7.45272562e-02 -3.13468874e-01 7.00738057e-02
9.61944938e-01 4.77916785e-02 1.26043230e-01 -2.63706297e-01
5.38749933e-01 -4.62104082e-01 5.65316558e-01 6.78387940e-01
-5.64134300e-01 1.04770017e+00 3.34985182e-02 -2.58063912e-01
-9.78868186e-01 -1.20930767e+00 -2.14802191e-01 4.01161551e-01
7.52096534e-01 8.59766304e-02 -9.97281611e-01 -4.15894747e-01
-5.81026599e-02 6.27522767e-01 -4.84332979e-01 -3.83666307e-01
-4.25561488e-01 -9.75379586e-01 3.18179876e-01 4.86330032e-01
1.01523590e+00 -5.52685797e-01 -1.69592053e-01 1.31634134e-03
-5.39691806e-01 -1.23288357e+00 -1.34828079e+00 -5.29618919e-01
-6.11183882e-01 -1.00894666e+00 -1.43360126e+00 -8.33896756e-01
8.35887194e-01 8.67785871e-01 6.41463518e-01 -3.15996371e-02
-1.93200439e-01 -1.55202392e-02 -1.41473949e-01 1.98590145e-01
-1.06730290e-01 -3.18367153e-01 7.45800734e-02 3.68396521e-01
2.14145869e-01 -4.49208915e-01 -1.16533768e+00 3.64576519e-01
-9.53535020e-01 2.57324278e-01 7.76939869e-01 1.08373618e+00
-2.10758388e-01 2.76479423e-01 3.96159708e-01 -2.39252836e-01
8.25958490e-01 -5.85869849e-01 -5.72129965e-01 1.42189860e-01
-4.64410841e-01 1.47156551e-01 4.34586674e-01 -6.98792934e-01
-1.38356328e+00 -4.20111984e-01 2.58075416e-01 -9.98829842e-01
-2.61872470e-01 4.36254263e-01 -9.56556350e-02 -5.14707193e-02
4.71930265e-01 4.76781756e-01 -2.19553977e-01 -6.06289566e-01
3.98861319e-01 1.05684650e+00 7.91400194e-01 -3.12600404e-01
9.71319318e-01 3.69911700e-01 -4.66434568e-01 -4.89382565e-01
-1.00252235e+00 -4.28835064e-01 -2.64489323e-01 -8.63387659e-02
8.00865471e-01 -1.41165686e+00 -5.40352821e-01 1.19163609e+00
-1.40628958e+00 -3.66941690e-01 3.37823600e-01 8.26199889e-01
-3.83238584e-01 7.83970237e-01 -1.17224061e+00 -4.76895899e-01
-1.02359377e-01 -1.40035033e+00 8.58128011e-01 5.71779966e-01
1.75183028e-01 -8.91415834e-01 -1.43302768e-01 3.55878711e-01
9.57020938e-01 -3.72826695e-01 5.28627217e-01 3.76920134e-01
-1.19001853e+00 1.56461194e-01 -1.05538058e+00 6.36967659e-01
7.54960954e-01 -4.15644586e-01 -1.18485379e+00 -3.97883624e-01
4.22028750e-01 -8.82890355e-03 1.16213131e+00 7.20736980e-01
1.44130874e+00 -5.49831986e-01 -3.85108516e-02 9.19688940e-01
1.37098324e+00 -1.50473803e-01 7.44631827e-01 1.08546108e-01
1.07117772e+00 8.48502517e-02 3.01052392e-01 2.19916567e-01
2.64026344e-01 6.53611839e-01 4.15456057e-01 2.06755083e-02
-6.25651062e-01 -1.16293989e-01 4.32229221e-01 8.25874150e-01
4.85316887e-02 -7.08968863e-02 -5.58298647e-01 7.94303894e-01
-1.98279166e+00 -9.49991643e-01 -1.52459607e-01 1.98054612e+00
1.11514902e+00 -1.74292505e-01 -4.13511276e-01 -4.77317840e-01
1.08211923e+00 4.58108932e-01 -9.09605324e-01 3.76525491e-01
-1.68104619e-01 -2.97563106e-01 6.60334647e-01 7.58787811e-01
-1.25078142e+00 9.23917830e-01 4.96600866e+00 9.22939241e-01
-1.32942772e+00 3.19763899e-01 8.56792152e-01 9.67627689e-02
-1.61582515e-01 4.57769297e-02 -1.19022325e-01 1.12940657e+00
6.28643990e-01 -1.74311146e-01 1.11198902e+00 6.11272991e-01
4.51900065e-01 -3.74951363e-02 -6.76411569e-01 1.36180925e+00
-2.00582407e-02 -1.37447214e+00 -2.33380929e-01 -1.38311923e-01
9.87631083e-01 1.78714335e-01 2.63096571e-01 3.63408402e-02
3.11459899e-01 -1.03228128e+00 7.17617035e-01 8.87470186e-01
1.17363405e+00 -3.06876987e-01 6.97560966e-01 3.24176759e-01
-9.58557367e-01 5.77699095e-02 -6.18485689e-01 -2.82387614e-01
1.31010756e-01 1.04770935e+00 -3.20135027e-01 4.46125656e-01
7.16459274e-01 1.08495474e+00 -5.31830966e-01 1.50333452e+00
-5.40883362e-01 7.77354598e-01 3.74675363e-01 4.28835928e-01
6.01895526e-02 -3.29593688e-01 5.69416702e-01 1.13820732e+00
6.61203086e-01 1.16655324e-03 -2.96463251e-01 1.31638551e+00
-2.49333531e-01 -6.65956020e-01 -1.87224329e-01 9.39879343e-02
4.03503180e-01 1.17060912e+00 -1.67313427e-01 -3.79096299e-01
-4.49485987e-01 1.56120741e+00 1.94412991e-01 1.02221274e+00
-1.15108156e+00 -5.28008103e-01 1.36953676e+00 -3.38441849e-01
3.73756379e-01 -4.65466768e-01 -2.51529723e-01 -1.98412216e+00
-9.89793986e-03 -6.55778527e-01 -1.92855522e-01 -1.35375023e+00
-1.67460871e+00 4.95415479e-01 -2.15975776e-01 -1.30667126e+00
2.22848743e-01 -6.46656930e-01 -6.69824183e-01 1.43984127e+00
-1.72237515e+00 -9.94870782e-01 -7.13966250e-01 3.50884855e-01
7.78917968e-01 8.35323110e-02 2.81174958e-01 3.68105531e-01
-5.39292157e-01 2.55164415e-01 5.34581244e-01 3.18093210e-01
1.16267371e+00 -1.23750103e+00 6.09049618e-01 1.26759064e+00
-2.67087728e-01 8.58602524e-01 7.82531381e-01 -5.57743311e-01
-1.35709977e+00 -1.34446967e+00 3.34648371e-01 -4.07359898e-01
9.66521680e-01 -9.36165825e-02 -1.17343223e+00 5.54899633e-01
5.86716294e-01 4.30607945e-01 1.37666119e-02 -5.42520165e-01
-2.82065868e-01 -1.54885963e-01 -8.24886024e-01 6.37028813e-01
9.15101647e-01 -8.52902651e-01 -4.64986235e-01 2.15286732e-01
8.95585120e-01 -7.05656111e-01 -5.43575764e-01 3.78310204e-01
3.33182186e-01 -9.85308945e-01 1.00319934e+00 -2.09524602e-01
6.42126262e-01 -7.92131484e-01 1.55911267e-01 -1.68398714e+00
-7.82623649e-01 -5.49332738e-01 -6.38690352e-01 1.02237725e+00
-3.35147053e-01 -7.85274506e-01 4.22955453e-01 4.22841132e-01
-2.67483771e-01 -5.59670925e-01 -5.51238894e-01 -5.67957401e-01
-2.56433725e-01 -1.16653472e-01 6.93243682e-01 1.13710380e+00
-4.78292733e-01 1.12624198e-01 -9.74120557e-01 7.40593314e-01
8.11757088e-01 3.95767480e-01 4.04641896e-01 -4.46345329e-01
-3.84923190e-01 -4.77870405e-01 -2.70285964e-01 -1.73552966e+00
1.46990463e-01 -3.83453131e-01 2.66205132e-01 -1.52830327e+00
3.70299131e-01 -1.52709007e-01 -4.48258281e-01 -1.35195749e-02
-8.39145958e-01 2.88365304e-01 -1.26941979e-01 7.74376750e-01
-3.57625455e-01 7.67333031e-01 1.60739827e+00 -3.66780907e-01
3.58766645e-01 -8.68037045e-02 -7.70090520e-01 5.61726809e-01
6.63577437e-01 -3.92504334e-02 -4.69570845e-01 -1.02096713e+00
-4.45541620e-01 7.56094828e-02 8.12868655e-01 -9.59762633e-01
3.08443397e-01 -4.61919606e-01 5.99191546e-01 -9.88120437e-02
1.87835380e-01 -6.36660159e-01 1.26376018e-01 3.15991402e-01
-2.83672631e-01 -2.43180335e-01 6.19502142e-02 7.48491764e-01
-3.63190323e-01 -1.51338279e-01 7.98720837e-01 -7.81396553e-02
-7.74972677e-01 5.09173930e-01 -1.38558522e-01 -1.34470820e-01
5.94415545e-01 1.86104611e-01 -9.86212075e-01 -6.60805881e-01
-1.83288440e-01 3.70188281e-02 8.10222685e-01 5.24993539e-01
6.70843482e-01 -1.37521327e+00 -7.11002588e-01 1.85057342e-01
-9.20849945e-03 -2.99942177e-02 6.99787378e-01 9.56402540e-01
-7.81585097e-01 4.55090314e-01 -1.33580044e-01 -3.57881725e-01
-7.58387387e-01 7.66398370e-01 5.46233773e-01 3.13994616e-01
-3.01554263e-01 9.93685544e-01 7.29297578e-01 2.72075623e-01
1.66712534e-02 -5.43748319e-01 2.32006446e-01 -6.21271312e-01
7.63744116e-01 3.12009990e-01 -2.71803796e-01 -6.21161759e-01
-8.96129012e-02 3.74200106e-01 -5.75906932e-02 2.71935284e-01
1.00979292e+00 -5.25703907e-01 -4.14551616e-01 1.29278168e-01
1.27741349e+00 1.02375045e-01 -1.95694590e+00 -3.81744564e-01
-3.14554363e-01 -1.16498029e+00 3.88399631e-01 -8.34621549e-01
-1.24640989e+00 9.18963075e-01 6.91480935e-01 -1.98788345e-02
1.34251165e+00 -6.24644645e-02 1.06011629e+00 -7.09855705e-02
1.63975701e-01 -3.72665077e-01 1.45758033e-01 3.41370076e-01
9.78606284e-01 -1.34312081e+00 -4.32259142e-02 -2.75971264e-01
-5.73299587e-01 1.06590271e+00 6.00035250e-01 -2.33208537e-01
5.83955705e-01 -1.23370595e-01 2.45316699e-01 1.73692971e-01
-4.01260138e-01 1.95602048e-02 4.90239173e-01 4.17295396e-01
2.97683567e-01 8.46848562e-02 5.44647723e-02 3.99035782e-01
2.31140539e-01 2.82514691e-01 7.19823718e-01 3.37032497e-01
-2.75492400e-01 -5.48897564e-01 -6.18782163e-01 3.36686403e-01
-3.65801424e-01 -4.32355523e-01 2.13294953e-01 5.53951040e-02
5.89581728e-02 8.94185066e-01 -1.35842457e-01 -2.85764754e-01
-2.06295311e-01 -4.75078881e-01 4.22709078e-01 -2.64706522e-01
2.55473316e-01 1.51080236e-01 -3.45506191e-01 -2.50347823e-01
-3.57689649e-01 -5.03252149e-01 -6.08552277e-01 -3.75956178e-01
-4.57973689e-01 1.50585890e-01 2.69700736e-01 7.70393729e-01
4.81779337e-01 5.76389313e-01 6.28228128e-01 -9.42725539e-01
-7.52222300e-01 -1.29368258e+00 -5.50017953e-01 3.71561259e-01
1.06139731e+00 -6.45728230e-01 -5.74184418e-01 4.01050359e-01]
|
[11.542720794677734, -2.6800520420074463]
|
8c2081c4-2145-472a-b116-8ca0f43ab6f0
|
towards-end-to-end-text-spotting-in-natural
|
1906.06013
| null |
https://arxiv.org/abs/1906.06013v6
|
https://arxiv.org/pdf/1906.06013v6.pdf
|
Towards End-to-End Text Spotting in Natural Scenes
|
Text spotting in natural scene images is of great importance for many image understanding tasks. It includes two sub-tasks: text detection and recognition. In this work, we propose a unified network that simultaneously localizes and recognizes text with a single forward pass, avoiding intermediate processes such as image cropping and feature re-calculation, word separation, and character grouping. In contrast to existing approaches that consider text detection and recognition as two distinct tasks and tackle them one by one, the proposed framework settles these two tasks concurrently. The whole framework can be trained end-to-end and is able to handle text of arbitrary shapes. The convolutional features are calculated only once and shared by both detection and recognition modules. Through multi-task training, the learned features become more discriminate and improve the overall performance. By employing the $2$D attention model in word recognition, the irregularity of text can be robustly addressed. It provides the spatial location for each character, which not only helps local feature extraction in word recognition, but also indicates an orientation angle to refine text localization. Our proposed method has achieved state-of-the-art performance on several standard text spotting benchmarks, including both regular and irregular ones.
|
['Peng Wang', 'Chunhua Shen', 'Hui Li']
|
2019-06-14
| null | null | null | null |
['text-spotting', 'image-cropping']
|
['computer-vision', 'computer-vision']
|
[ 5.82604408e-01 -5.82616508e-01 -1.46965742e-01 -3.19262207e-01
-6.02424383e-01 -4.85895634e-01 7.87863791e-01 3.08169037e-01
-7.83650517e-01 1.44414976e-01 -8.43052343e-02 -3.53232503e-01
1.95792437e-01 -6.32044554e-01 -4.70596313e-01 -7.91001618e-01
7.42133856e-01 4.72255528e-01 3.37915301e-01 5.82547374e-02
8.38228405e-01 5.54041028e-01 -1.41060078e+00 4.37189758e-01
8.40163827e-01 9.70914960e-01 5.12451589e-01 8.19880068e-01
-5.88185608e-01 4.98745263e-01 -5.66815853e-01 -2.10223928e-01
4.76898141e-02 -1.42997101e-01 -5.82471192e-01 4.96091843e-01
5.66097558e-01 -2.45483443e-01 -3.49956065e-01 1.00567353e+00
6.41440213e-01 2.37771887e-02 7.57698417e-01 -6.42923951e-01
-8.06116581e-01 4.87820387e-01 -1.00399876e+00 9.69011411e-02
2.74678469e-01 4.66212183e-02 1.06048083e+00 -1.29276395e+00
1.55880168e-01 1.16942596e+00 4.05115098e-01 3.14892441e-01
-1.01830363e+00 -5.52197695e-01 3.19732219e-01 9.40028653e-02
-1.50756299e+00 -4.28777575e-01 6.40444398e-01 -4.45625991e-01
1.06806850e+00 3.78303915e-01 2.65807301e-01 5.77176452e-01
1.49065629e-01 1.29989231e+00 6.09667420e-01 -7.06190884e-01
-1.88234955e-01 -5.70293553e-02 3.55590016e-01 7.53474236e-01
2.51452535e-01 -5.37197888e-01 -5.77923894e-01 3.02614361e-01
8.36982012e-01 4.20610964e-01 -2.63258398e-01 -1.01652443e-01
-1.62098706e+00 6.09324217e-01 1.68258786e-01 3.99801403e-01
-1.44487485e-01 6.90152273e-02 4.26844060e-01 3.96775594e-03
2.19206214e-01 7.70722032e-02 -1.64850608e-01 -4.06323075e-02
-1.21315849e+00 -4.56856601e-02 3.96070778e-01 7.61066258e-01
6.87581956e-01 -2.80608106e-02 -4.79203284e-01 1.05433953e+00
3.13607126e-01 8.17105412e-01 7.87151754e-01 2.47163326e-01
7.71736920e-01 9.23239172e-01 6.43464252e-02 -1.13678277e+00
-4.57728922e-01 -2.43551776e-01 -9.50578392e-01 -1.01654348e-03
3.90760958e-01 7.27745742e-02 -1.31194139e+00 9.81166244e-01
9.79423672e-02 -1.06803991e-01 -2.54457861e-01 9.53148365e-01
8.15825939e-01 7.92443633e-01 -1.46694735e-01 1.65504172e-01
1.70808780e+00 -1.35547376e+00 -7.80792832e-01 -5.91056705e-01
5.36082506e-01 -1.36789989e+00 1.03238237e+00 3.87511820e-01
-8.82334352e-01 -6.28908098e-01 -1.00624025e+00 -4.99300092e-01
-6.35960221e-01 9.85824108e-01 2.81289309e-01 6.29413188e-01
-9.21906114e-01 1.07287884e-01 -7.12244153e-01 -4.78770137e-01
3.35197896e-01 5.51977754e-01 -2.31530890e-01 -3.89843062e-02
-6.68311536e-01 6.50952160e-01 3.67323697e-01 4.63647008e-01
-3.42192411e-01 7.84014165e-02 -6.99064016e-01 2.54824966e-01
3.52737367e-01 -3.29536200e-01 8.71311188e-01 -9.19866085e-01
-1.47607946e+00 7.45903492e-01 -5.13572454e-01 -8.82033035e-02
3.97411853e-01 -3.01708847e-01 -3.99066746e-01 1.56106383e-01
2.09223047e-01 4.96416450e-01 1.47219837e+00 -8.71340632e-01
-7.85468698e-01 -3.63877714e-01 -5.53631127e-01 4.75949198e-01
-5.63351393e-01 2.02803850e-01 -1.00996494e+00 -9.64081287e-01
3.55108619e-01 -4.01189536e-01 6.48262948e-02 1.74710721e-01
-5.36281407e-01 -4.41272676e-01 1.10760820e+00 -6.04596555e-01
1.36712074e+00 -2.09080577e+00 1.23139866e-01 1.28258914e-01
2.52168000e-01 4.79524583e-01 -2.17615306e-01 3.40432495e-01
3.28526497e-02 9.50002670e-02 -2.00439966e-03 -6.57683432e-01
4.21480648e-02 -2.18558177e-01 -2.94234067e-01 5.51847160e-01
4.08499926e-01 1.02451503e+00 -3.11382473e-01 -7.03828156e-01
6.24190390e-01 3.89379144e-01 -9.98595133e-02 -2.71528717e-02
-9.83891934e-02 -3.06595396e-02 -5.61278939e-01 7.14947283e-01
7.92174697e-01 -3.58174175e-01 -3.48132886e-02 -1.42033622e-01
-2.73006678e-01 -4.04429436e-02 -1.45594895e+00 1.46325696e+00
-4.26350653e-01 8.58778536e-01 6.92767352e-02 -1.10451269e+00
1.18513131e+00 5.29929884e-02 1.11028358e-01 -7.13879824e-01
2.80539781e-01 1.96923748e-01 -3.21392149e-01 -5.69312811e-01
7.45647430e-01 3.82921427e-01 -2.43768450e-02 6.79981768e-01
-6.58442229e-02 7.32348040e-02 2.80821413e-01 -2.37966608e-02
6.33213818e-01 -8.59270245e-02 1.84105858e-01 -1.83972716e-01
9.57924426e-01 -1.53890103e-01 -2.57374104e-02 7.44160295e-01
9.82754007e-02 6.39190197e-01 2.57119358e-01 -6.68445706e-01
-9.05332506e-01 -6.23035192e-01 -1.29153669e-01 1.46386361e+00
3.89208913e-01 -3.19830716e-01 -5.05023181e-01 -7.25762904e-01
3.99963520e-02 3.06454718e-01 -6.12993360e-01 1.43775687e-01
-7.36643314e-01 -8.14293385e-01 5.78191221e-01 5.86075544e-01
6.46124005e-01 -1.07327223e+00 -5.64792573e-01 1.06074482e-01
-4.61015329e-02 -1.01865458e+00 -1.03729713e+00 4.56805140e-01
-6.32642329e-01 -9.58103120e-01 -1.10287046e+00 -1.30860603e+00
9.62381959e-01 6.40101671e-01 6.82918489e-01 2.28183284e-01
-4.91473526e-01 1.51263803e-01 -4.80843514e-01 -3.85069214e-02
1.13012977e-01 2.04273477e-01 -3.58522207e-01 5.16019106e-01
5.20596027e-01 1.42663464e-01 -7.06183374e-01 5.27703106e-01
-1.13047886e+00 9.10162032e-02 9.05807137e-01 1.11735642e+00
4.27350283e-01 -2.08534673e-02 3.13131243e-01 -4.67566699e-01
7.14814126e-01 1.55790344e-01 -4.59110647e-01 5.28030634e-01
-4.72783327e-01 9.35064107e-02 7.93676734e-01 -4.25200790e-01
-8.96000147e-01 4.29695874e-01 -1.39500946e-01 -9.10033658e-03
-3.76154840e-01 3.28627348e-01 -4.82434444e-02 -3.31626952e-01
3.08698505e-01 9.45384145e-01 -2.10792124e-01 -5.42692482e-01
2.30507016e-01 1.13910770e+00 2.63451815e-01 -2.67523170e-01
8.04394186e-01 3.99990052e-01 -3.05160463e-01 -1.28912258e+00
-4.87785727e-01 -8.45099986e-01 -9.34705615e-01 1.51024964e-02
8.51390004e-01 -7.97584116e-01 -8.24097693e-01 9.55370069e-01
-1.22701657e+00 -1.64536774e-01 2.38594562e-01 2.06753209e-01
-1.01487413e-01 8.83371294e-01 -3.81844640e-01 -7.46762276e-01
-6.59557164e-01 -1.28275847e+00 1.68978298e+00 2.64878064e-01
1.70129746e-01 -9.78486717e-01 -2.97760516e-01 2.00284630e-01
3.95889401e-01 -4.16217685e-01 8.76085639e-01 -7.48115540e-01
-5.33936799e-01 -5.37479699e-01 -6.75484300e-01 6.99515417e-02
1.09792665e-01 1.42641783e-01 -7.46512353e-01 -3.57138962e-01
-2.81267017e-01 -2.84753084e-01 1.24434853e+00 1.63351402e-01
1.09485817e+00 -3.36094871e-02 -4.13478106e-01 5.54040968e-01
1.35812950e+00 1.96909785e-01 5.19198298e-01 3.51692498e-01
9.11070049e-01 2.84313947e-01 4.81337219e-01 4.46677506e-01
2.24508718e-01 5.93502223e-01 1.98268458e-01 -4.62933958e-01
-7.08146319e-02 9.08174217e-02 1.41057745e-01 5.44863462e-01
3.39761078e-01 -5.91834664e-01 -1.03119171e+00 4.64755386e-01
-1.98549438e+00 -7.06994712e-01 -1.99788958e-01 1.92801929e+00
5.95491111e-01 1.39836803e-01 7.75002828e-03 2.44649842e-01
1.06746757e+00 4.67749238e-01 -5.22543073e-01 -3.19696903e-01
-2.01157659e-01 3.13029200e-01 4.81659859e-01 3.94861698e-01
-1.31727576e+00 1.20759797e+00 5.87314892e+00 1.18188787e+00
-1.50439727e+00 -3.51082593e-01 7.11836815e-01 1.22263081e-01
2.19186500e-01 -3.91080499e-01 -1.06609881e+00 3.32244575e-01
1.41392127e-01 1.65070623e-01 3.51355880e-01 6.51059270e-01
5.69270402e-02 -2.31968299e-01 -9.30342257e-01 1.31172431e+00
4.37620342e-01 -1.27425396e+00 4.58850771e-01 -3.46671879e-01
4.66856271e-01 -1.83640122e-01 1.04536861e-01 9.65339616e-02
-3.08750391e-01 -1.15884042e+00 7.64966488e-01 3.80553901e-01
9.38104510e-01 -7.01461077e-01 6.80160344e-01 4.21526164e-01
-1.59378099e+00 -1.05085179e-01 -2.80918419e-01 1.10600054e-01
-1.58630654e-01 4.97035235e-01 -6.58939302e-01 3.57181251e-01
4.33411330e-01 7.72536933e-01 -8.51604044e-01 1.05501866e+00
-1.00795418e-01 1.70037314e-01 -2.81621516e-01 -6.34879708e-01
3.84242088e-01 -1.30870536e-01 1.80698887e-01 1.52685106e+00
2.40330741e-01 -2.77035207e-01 4.00858790e-01 6.02661610e-01
-8.17159638e-02 5.46646893e-01 -2.10709155e-01 -2.66774029e-01
4.19242121e-02 1.37449074e+00 -1.25029385e+00 -3.97209674e-01
-4.59167361e-01 1.36009419e+00 2.57978857e-01 3.66251081e-01
-5.85061967e-01 -1.09402955e+00 1.34462357e-01 -1.62276998e-01
6.88125134e-01 -4.76780593e-01 -6.14172518e-01 -1.31278455e+00
2.95122862e-01 -8.02318215e-01 1.33645415e-01 -5.61474264e-01
-8.87716711e-01 5.63117921e-01 -6.43984556e-01 -9.75723028e-01
1.51760668e-01 -1.07043660e+00 -8.07663321e-01 1.08759892e+00
-1.62698936e+00 -1.27194357e+00 -4.71722335e-01 7.76787162e-01
1.02211106e+00 -1.05454279e-02 6.54589176e-01 2.40853086e-01
-9.29557800e-01 8.46131682e-01 4.23154682e-01 5.49147308e-01
9.65805292e-01 -1.12203705e+00 7.22674966e-01 1.11120915e+00
2.80793667e-01 5.26056886e-01 2.07422450e-01 -6.72085226e-01
-1.60918868e+00 -9.91401136e-01 8.17424297e-01 -1.26873434e-01
6.47204936e-01 -6.60316229e-01 -8.12432289e-01 3.24294806e-01
1.84972733e-01 -1.64287761e-01 2.77054638e-01 -1.77773044e-01
-3.04982126e-01 -8.91134422e-03 -7.41600394e-01 5.67267418e-01
5.45708895e-01 -4.53103364e-01 -3.33419412e-01 5.11928201e-01
2.97028154e-01 -4.04173493e-01 -2.87627071e-01 -2.66742352e-02
5.49531519e-01 -7.99408436e-01 7.94082046e-01 -1.68020159e-01
1.63653418e-01 -4.01226074e-01 1.52346745e-01 -7.13291347e-01
-3.01494360e-01 -5.41226923e-01 1.49885029e-01 1.07312477e+00
5.03692687e-01 -5.81703365e-01 7.69072354e-01 -1.82190798e-02
3.32684666e-02 -7.13373899e-01 -8.51541758e-01 -2.79281706e-01
-9.81210247e-02 -2.46742338e-01 3.38293731e-01 6.83861554e-01
9.74192247e-02 5.31305492e-01 -3.73897731e-01 2.79134661e-02
3.89629722e-01 4.97836351e-01 6.61171496e-01 -1.11373127e+00
2.64099222e-02 -8.29219460e-01 -3.10821712e-01 -1.85852015e+00
-9.33430791e-02 -7.61664808e-01 3.48121405e-01 -1.64418304e+00
1.81736737e-01 -1.69763595e-01 8.05640593e-02 5.25848627e-01
-4.63076532e-01 3.02689701e-01 1.50871217e-01 2.19217300e-01
-7.99665749e-01 5.08841753e-01 1.22045827e+00 -3.35891962e-01
-1.47063896e-01 -1.52752707e-02 -4.57874537e-01 6.13273680e-01
7.63364375e-01 -6.86433464e-02 8.24393108e-02 -8.66723716e-01
8.59999731e-02 -1.78725913e-01 3.19872797e-01 -9.21866894e-01
6.33426368e-01 1.81044638e-02 7.05330491e-01 -1.09445369e+00
1.50788546e-01 -8.27289879e-01 -7.64352560e-01 2.85213232e-01
-3.04719329e-01 8.07207152e-02 2.64438331e-01 5.93989015e-01
-1.93441510e-01 -5.48664331e-01 7.76940465e-01 1.04823798e-01
-6.88215554e-01 1.37287989e-01 -5.07633924e-01 -3.38755369e-01
9.66816187e-01 -4.07942235e-01 -4.44037348e-01 -5.28203771e-02
-4.82198566e-01 2.87417054e-01 2.92109579e-01 4.14558530e-01
8.61944973e-01 -9.23357964e-01 -6.38487458e-01 5.56712568e-01
1.13831572e-01 1.59635291e-01 1.90690249e-01 7.91690111e-01
-7.02468276e-01 7.27382064e-01 5.35608567e-02 -8.77108097e-01
-1.30383623e+00 5.49294889e-01 2.58139342e-01 -2.81695127e-01
-5.55941820e-01 7.87158370e-01 1.67232931e-01 -1.97731391e-01
6.12818360e-01 -3.68155420e-01 -3.09363216e-01 1.22175947e-01
7.78705060e-01 1.48091346e-01 1.61316574e-01 -6.53563738e-01
-3.03459436e-01 1.25440454e+00 -4.49599087e-01 1.78615838e-01
1.07715189e+00 -3.90883863e-01 -1.97443575e-01 1.96064368e-01
1.05236816e+00 8.28399956e-02 -1.02757299e+00 -3.89056891e-01
7.37740844e-02 -4.58910555e-01 1.37632981e-01 -8.31666231e-01
-9.90783095e-01 1.22893631e+00 5.58492005e-01 2.74000645e-01
1.14944208e+00 -3.92024189e-01 7.47138202e-01 6.28048658e-01
-2.07972556e-01 -1.22560740e+00 3.84101570e-01 7.09449172e-01
7.04579115e-01 -1.34728754e+00 4.69744913e-02 -3.11444551e-01
-5.69317937e-01 1.69693589e+00 5.11210680e-01 -1.25187382e-01
3.72040600e-01 4.18659717e-01 1.28979106e-02 -5.63441068e-02
-3.84242028e-01 -3.18313211e-01 5.27103484e-01 2.25253716e-01
5.58544457e-01 -2.54580826e-01 -1.85221970e-01 3.67235184e-01
4.09808338e-01 -3.51308942e-01 2.61351913e-01 9.29270744e-01
-9.44107890e-01 -1.01477695e+00 -6.31842792e-01 6.18188620e-01
-3.73867065e-01 -4.35526192e-01 -5.49184382e-01 5.59497774e-01
-1.55414551e-01 8.80887628e-01 2.19428375e-01 -2.82152772e-01
2.06143990e-01 1.89578414e-01 1.07629351e-01 -6.31985426e-01
-5.67749560e-01 5.11280060e-01 -3.05689514e-01 -8.77218395e-02
-1.91949308e-01 -4.66697514e-01 -1.17468035e+00 -1.38599038e-01
-8.61054301e-01 -5.88110052e-02 7.88571715e-01 8.89934242e-01
3.05032700e-01 6.18070006e-01 7.43755400e-01 -9.66493487e-01
-5.92079043e-01 -9.99745131e-01 -4.38576162e-01 4.52977568e-02
3.94324571e-01 -3.25805366e-01 -2.39016399e-01 2.62591809e-01]
|
[11.966407775878906, 2.255176067352295]
|
98be0e4b-af97-451f-b329-72febeb1ab97
|
times-are-changing-investigating-the-pace-of
| null | null |
https://aclanthology.org/W19-4718
|
https://aclanthology.org/W19-4718.pdf
|
Times Are Changing: Investigating the Pace of Language Change in Diachronic Word Embeddings
|
We propose Word Embedding Networks, a novel method that is able to learn word embeddings of individual data slices while simultaneously aligning and ordering them without feeding temporal information a priori to the model. This gives us the opportunity to analyse the dynamics in word embeddings on a large scale in a purely data-driven manner. In experiments on two different newspaper corpora, the New York Times (English) and die Zeit (German), we were able to show that time actually determines the dynamics of semantic change. However, there is by no means a uniform evolution, but instead times of faster and times of slower change.
|
['Br', 'David Lassner', 'Stephanie l']
|
2019-08-01
| null | null | null |
ws-2019-8
|
['diachronic-word-embeddings']
|
['natural-language-processing']
|
[-3.32505524e-01 -8.89529139e-02 -2.94866145e-01 -2.46321797e-01
1.33481935e-01 -7.58239865e-01 1.16955066e+00 6.37977660e-01
-1.04917359e+00 5.35122275e-01 5.49625218e-01 -5.17001510e-01
-2.80578643e-01 -9.60778475e-01 -2.37178370e-01 -6.21261716e-01
-5.63260555e-01 5.40255547e-01 4.15604621e-01 -5.95831096e-01
1.53548539e-01 5.39839089e-01 -1.24813247e+00 -3.67467627e-02
3.41617495e-01 2.31357455e-01 7.77506977e-02 8.18358004e-01
-4.20586556e-01 4.28630263e-01 -4.12326664e-01 -3.09502274e-01
1.63377553e-01 -4.42400366e-01 -8.92505288e-01 -9.95474905e-02
-2.74300762e-03 1.76468760e-01 -7.10984409e-01 9.03094590e-01
6.67374283e-02 1.43123880e-01 6.47249758e-01 -8.89025509e-01
-8.91801655e-01 8.23048770e-01 -2.85298049e-01 7.03071773e-01
9.82455984e-02 -1.72804650e-02 1.40040660e+00 -4.87540752e-01
1.17075419e+00 1.19511306e+00 5.06923556e-01 3.37226361e-01
-1.33687687e+00 -1.33647233e-01 4.66169834e-01 4.57145333e-01
-1.06642210e+00 -6.80359378e-02 9.43090439e-01 -6.67187691e-01
9.64241922e-01 7.85896778e-02 1.07082927e+00 1.35599089e+00
3.46009761e-01 3.74153197e-01 9.80634987e-01 -5.78361452e-01
1.18409798e-01 9.61826146e-02 5.32426894e-01 4.48150426e-01
2.22641274e-01 3.32916558e-01 -4.92863983e-01 6.85201809e-02
5.18597007e-01 1.26439989e-01 -2.67698705e-01 -4.36143726e-01
-1.41660488e+00 9.40486193e-01 4.60584074e-01 1.15686083e+00
-2.21162781e-01 3.67292494e-01 5.99595547e-01 6.84823096e-01
7.14092493e-01 4.95049030e-01 -8.26156855e-01 -4.56942618e-01
-6.55285954e-01 2.20318839e-01 7.49539018e-01 2.08231449e-01
6.02132022e-01 -9.64303166e-02 3.72995347e-01 4.36675072e-01
2.31605381e-01 8.95567760e-02 1.04469478e+00 -3.00506413e-01
2.06750572e-01 3.39860559e-01 2.12437600e-01 -1.18458819e+00
-4.64087099e-01 -6.59980923e-02 -3.23480874e-01 -6.21595420e-03
7.04504848e-01 5.69877587e-03 -9.43731010e-01 2.20025349e+00
7.61597678e-02 2.28771582e-01 -1.51845580e-02 5.34383535e-01
2.51280189e-01 6.82980418e-01 5.52261621e-02 -2.90897816e-01
1.45133471e+00 -5.74179053e-01 -8.85375500e-01 -1.90432876e-01
6.70349240e-01 -4.23710704e-01 1.11968005e+00 1.43732592e-01
-7.44317055e-01 -4.46724504e-01 -1.08376896e+00 -6.49364591e-02
-1.03606808e+00 -7.96133518e-01 6.57117128e-01 3.12984914e-01
-1.14259148e+00 8.74705911e-01 -1.09749424e+00 -9.02630806e-01
4.77995761e-02 8.13707337e-02 -3.93161863e-01 3.32554728e-01
-1.47665131e+00 1.23023105e+00 4.87978697e-01 -2.33018652e-01
-4.68247265e-01 -6.30966187e-01 -8.25573564e-01 -4.72031906e-02
5.12544066e-02 -1.61751911e-01 1.04454863e+00 -9.77689266e-01
-1.30833185e+00 9.56178129e-01 -6.39772192e-02 -5.19115567e-01
4.30004776e-01 4.39111479e-02 -9.40275669e-01 -1.73630938e-01
-3.05547893e-01 3.64448905e-01 6.10222042e-01 -9.90644395e-01
-3.77395868e-01 -4.69651878e-01 1.72520995e-01 -1.24474317e-01
-8.91646206e-01 4.38832082e-02 1.65219866e-02 -6.38060808e-01
-6.07772693e-02 -6.04261816e-01 -2.88502008e-01 3.04292189e-03
2.18175724e-01 -4.43660975e-01 5.09247065e-01 -4.97835368e-01
1.48164344e+00 -2.40415144e+00 5.60018480e-01 -1.04699776e-01
4.75166351e-01 9.15687829e-02 -3.69558185e-01 8.47861290e-01
-4.27745283e-01 2.99540311e-01 -1.48981050e-01 -1.86379403e-01
2.42868319e-01 7.20227301e-01 -3.37113112e-01 7.99143434e-01
1.37327179e-01 9.63841796e-01 -1.21546555e+00 -2.62666970e-01
1.16682827e-01 3.01357716e-01 -5.39289653e-01 -9.95224416e-02
-2.05330700e-01 -7.71764219e-02 -2.65913874e-01 6.37213364e-02
7.41685405e-02 1.44326910e-02 5.06137252e-01 3.07585597e-02
-5.65034330e-01 3.48771036e-01 -8.53654742e-01 1.80524480e+00
-5.45331776e-01 1.02337861e+00 -2.34396830e-01 -1.22773361e+00
7.93753147e-01 3.82327616e-01 5.75195909e-01 -9.60793853e-01
1.79910019e-01 1.10901348e-01 3.96065056e-01 -6.14039600e-01
7.40358353e-01 -5.72229445e-01 -2.39883736e-01 7.15272963e-01
2.17304051e-01 2.49111392e-02 4.35271651e-01 -4.02505472e-02
1.22690213e+00 -2.06379831e-01 3.09937030e-01 -6.16885722e-01
1.48282098e-02 9.45688284e-04 4.93439734e-01 1.39840320e-01
-3.89842033e-01 3.23737383e-01 7.39188194e-01 -9.63530779e-01
-1.47206521e+00 -1.19085169e+00 -3.40906978e-01 9.17053819e-01
2.67638620e-02 -8.14741313e-01 -1.74412012e-01 -4.84918922e-01
1.00305192e-01 7.41700709e-01 -1.21689510e+00 -1.69296980e-01
-6.69761062e-01 -7.89812922e-01 2.60319650e-01 3.84670615e-01
-2.47148529e-01 -1.13889694e+00 -8.11034739e-01 5.32625020e-01
3.04763168e-01 -7.85216212e-01 -4.12603348e-01 4.40864116e-01
-8.83358717e-01 -8.76422107e-01 -4.93757844e-01 -6.30483985e-01
3.92569661e-01 -2.29938462e-01 1.34362423e+00 -2.93291211e-01
-5.57328105e-01 2.33023822e-01 -3.40609908e-01 -4.32120293e-01
-2.40413114e-01 1.82562605e-01 2.93008298e-01 -1.98257208e-01
5.41479826e-01 -9.05047774e-01 -4.39628303e-01 -1.02519691e-01
-1.31408548e+00 -4.96356636e-01 1.56097725e-01 8.21369529e-01
-2.35539619e-02 5.10542504e-02 3.13598752e-01 -7.69957602e-01
8.12049925e-01 -6.46808386e-01 -5.30817091e-01 1.82929505e-02
-9.75994587e-01 3.22632015e-01 6.64254904e-01 -7.40396976e-01
-4.54824328e-01 -5.20840883e-01 -7.12828562e-02 -3.77359018e-02
-3.54886316e-02 7.23967314e-01 3.96894783e-01 6.18887126e-01
4.84474927e-01 1.10640280e-01 -3.92766744e-02 -6.66878700e-01
9.36914384e-01 3.91421050e-01 3.66634518e-01 -3.75525206e-01
8.92458797e-01 5.43069243e-01 -4.45702761e-01 -8.04876745e-01
-3.29321980e-01 -3.60310555e-01 -1.10783696e+00 -3.63250114e-02
1.07161331e+00 -4.83761311e-01 -7.10847527e-02 2.10774064e-01
-1.24265873e+00 -3.89890730e-01 -8.61476600e-01 3.42936307e-01
-3.05866659e-01 7.58118480e-02 -6.06396496e-01 -3.89341831e-01
3.16082120e-01 -4.56516743e-01 3.13000649e-01 -6.48519490e-03
-7.22641408e-01 -1.76115048e+00 9.03677166e-01 -5.52880526e-01
4.64333415e-01 3.79300147e-01 1.27907920e+00 -8.76796246e-01
5.36609292e-02 -2.70092845e-01 8.85953680e-02 1.05417594e-01
4.61237103e-01 3.61055404e-01 -6.49583459e-01 -2.02963591e-01
-8.21705163e-02 2.62363851e-01 7.92967618e-01 7.37438649e-02
6.72565520e-01 -3.46987844e-01 -2.63460666e-01 4.17819202e-01
1.74243546e+00 3.47377181e-01 5.90066314e-01 6.64198697e-01
5.35319746e-01 6.30264997e-01 5.22996224e-02 2.40456492e-01
3.12370062e-01 6.70360804e-01 1.44987240e-01 2.36336496e-02
-1.21563785e-02 -1.49113372e-01 4.30967242e-01 1.36547875e+00
-4.35396060e-02 -3.66799593e-01 -1.09829640e+00 1.24533892e+00
-1.75675702e+00 -1.07982874e+00 -4.60111387e-02 2.05838752e+00
9.88162339e-01 2.68214345e-01 1.17363021e-01 1.02840580e-01
3.46529841e-01 8.36442053e-01 -2.03711629e-01 -9.34369564e-01
1.74492508e-01 6.19408250e-01 4.45588022e-01 6.79527581e-01
-7.27257192e-01 8.66936326e-01 7.72235060e+00 3.25360447e-01
-1.15491068e+00 2.93999404e-01 1.14377849e-01 -3.47086340e-02
-8.11232150e-01 1.18431062e-01 -1.69528842e-01 5.93160391e-01
1.43516195e+00 -6.64551735e-01 3.09911460e-01 2.46320173e-01
9.41483676e-02 2.43461013e-01 -1.21178687e+00 6.15032613e-01
-1.71735868e-01 -1.25641930e+00 -2.05784645e-02 6.06162213e-02
4.91921097e-01 1.96944311e-01 -1.15436010e-01 9.81595889e-02
3.76648307e-01 -1.04300356e+00 7.64627099e-01 4.40292984e-01
4.41181600e-01 -5.61158597e-01 5.68810821e-01 1.28402457e-01
-9.82818961e-01 -8.69301111e-02 -3.30916941e-01 -4.81828481e-01
3.52712542e-01 6.30532682e-01 -4.10559446e-01 4.81838495e-01
4.50374752e-01 9.41552281e-01 -3.98352385e-01 4.61729586e-01
-3.27027887e-01 5.08261800e-01 -1.16285294e-01 -1.72004685e-01
4.38917696e-01 -3.00388664e-01 5.68603516e-01 1.21858120e+00
2.84794807e-01 -1.64489403e-01 -2.56288618e-01 7.17115879e-01
5.93635738e-02 -2.68927217e-02 -8.49077642e-01 -6.88516855e-01
2.08170250e-01 1.03319633e+00 -9.15723920e-01 -1.48689583e-01
-5.39849281e-01 9.09936428e-01 3.48297983e-01 2.77648419e-01
-8.33770037e-01 -4.19271022e-01 1.19918907e+00 2.23142385e-01
5.64481437e-01 -8.83277476e-01 4.26047184e-02 -1.27226901e+00
-1.44877926e-01 -3.43242884e-01 3.03537399e-01 -3.44065845e-01
-1.29925501e+00 7.02791393e-01 1.35925293e-01 -8.12729836e-01
-3.22692305e-01 -8.20279360e-01 -7.39362597e-01 7.11391389e-01
-1.46085560e+00 -5.36835730e-01 2.20779702e-01 3.56733233e-01
4.03136462e-01 4.74875681e-02 9.44341600e-01 3.24828982e-01
-4.27815199e-01 1.58661559e-01 4.17199373e-01 1.35355338e-01
5.42244554e-01 -1.56371331e+00 6.99131072e-01 7.51283109e-01
7.87753582e-01 6.04614079e-01 1.09999168e+00 -1.89474851e-01
-1.16363204e+00 -4.99558449e-01 1.52828455e+00 -6.42284870e-01
1.59035873e+00 -5.95901430e-01 -9.88269448e-01 6.68475986e-01
6.32372856e-01 6.62771314e-02 5.63366413e-01 4.83686239e-01
-4.77982879e-01 -7.17105418e-02 -7.07727373e-01 5.87940037e-01
9.74521279e-01 -8.62395644e-01 -1.16153729e+00 2.26419702e-01
1.12024426e+00 1.73502728e-01 -8.65405679e-01 1.56096891e-02
6.08299613e-01 -6.93254411e-01 8.66497159e-01 -1.16384089e+00
5.90147853e-01 3.39788310e-02 -1.20546468e-01 -1.63388455e+00
-4.41758454e-01 -4.91067052e-01 -1.08538434e-01 1.18606961e+00
4.67696846e-01 -9.33080018e-01 3.94289911e-01 2.78570503e-01
1.79339960e-01 -6.38486624e-01 -1.08020318e+00 -1.01102412e+00
5.22127807e-01 -3.93546194e-01 6.63473845e-01 1.24140298e+00
2.25469917e-01 2.11256266e-01 5.91624826e-02 -2.65219331e-01
1.55587137e-01 -1.92951541e-02 8.95837620e-02 -1.29715538e+00
-6.82900697e-02 -8.57218683e-01 -9.81470227e-01 -5.16782522e-01
1.63982555e-01 -8.95711720e-01 -1.46237805e-01 -1.37645030e+00
2.99061853e-02 -1.11502722e-01 -9.25180554e-01 1.52620643e-01
2.76047383e-02 5.97659918e-03 -7.85526037e-02 8.02573860e-02
-1.98212445e-01 6.23807251e-01 8.86069417e-01 -5.03774770e-02
8.36472027e-03 -6.10303700e-01 -4.81085271e-01 5.75892091e-01
6.77272916e-01 -6.71017766e-01 -5.45907855e-01 -6.72162533e-01
4.83884990e-01 -3.97475749e-01 1.64430737e-01 -7.61604846e-01
-1.69380102e-02 -2.15233445e-01 -5.21658175e-02 -5.62124215e-02
-3.63656618e-02 -8.41417968e-01 1.19937405e-01 5.75937808e-01
-4.58693206e-01 7.18443811e-01 3.29747021e-01 7.56105363e-01
-3.52802306e-01 -2.90632069e-01 6.50979221e-01 -1.03789732e-01
-9.28894341e-01 4.32272196e-01 -4.60319012e-01 8.31149518e-02
1.05983198e+00 -1.45815602e-02 -9.44250375e-02 -7.16065243e-02
-8.43027949e-01 -1.70969423e-02 6.23446941e-01 7.82930434e-01
1.60059005e-01 -1.58370221e+00 -4.92140383e-01 1.23870909e-01
2.13060766e-01 -5.76794505e-01 -9.66934413e-02 5.35677671e-01
-4.64743525e-01 3.77639055e-01 -3.55114698e-01 -1.89618170e-01
-8.67741525e-01 8.44105184e-01 2.03721672e-01 -3.87156516e-01
-6.16182923e-01 6.79416478e-01 -1.46013215e-01 -4.13973659e-01
-9.43421647e-02 -6.81388795e-01 -3.90646040e-01 8.46573353e-01
3.27217162e-01 -3.02817728e-02 -1.37391776e-01 -4.66176569e-01
-4.21842277e-01 5.41488051e-01 -5.06534614e-03 -5.38451672e-01
1.87325203e+00 -7.63719231e-02 -3.21755379e-01 1.36695659e+00
1.39863503e+00 -1.12277158e-01 -9.34988797e-01 -2.16630623e-01
4.90447372e-01 -5.87044179e-01 -1.01833031e-01 -3.09597522e-01
-8.88026059e-01 8.50535810e-01 5.76901913e-01 9.78304625e-01
6.35756552e-01 2.11289823e-01 7.69855678e-01 1.11175232e-01
1.73855871e-01 -1.19518828e+00 1.12751789e-01 5.94699383e-01
6.88245535e-01 -8.46471190e-01 -1.16029426e-01 4.18574512e-01
-4.48004790e-02 1.33887661e+00 8.02417845e-02 -4.28647310e-01
1.00325179e+00 2.72232175e-01 3.93482745e-02 -4.46777850e-01
-1.13155198e+00 -1.34323105e-01 1.67630211e-01 4.14115816e-01
4.70657319e-01 5.37704341e-02 -7.64797568e-01 2.76248604e-01
-1.39383540e-01 -2.95808882e-01 5.37059724e-01 9.27689791e-01
-2.80917794e-01 -1.62955081e+00 9.70615819e-02 1.60667196e-01
-3.59401166e-01 -3.49864960e-02 -3.14606071e-01 1.22601891e+00
1.35452211e-01 2.78023511e-01 5.75985789e-01 -4.76180911e-01
3.40085715e-01 4.55519259e-01 4.98710275e-01 -6.26079261e-01
-2.42217392e-01 -3.18392634e-01 4.52775918e-02 -5.41756272e-01
-4.80078399e-01 -8.12555730e-01 -1.08366811e+00 -4.24047858e-01
6.48782030e-02 1.07631728e-01 7.35485017e-01 9.47220564e-01
4.89097163e-02 7.73191988e-01 5.32190323e-01 -5.44919372e-01
-3.27887505e-01 -9.08344567e-01 -8.56290519e-01 7.85305500e-01
5.14865339e-01 -5.33442438e-01 -6.36577189e-01 4.19115163e-02]
|
[10.212876319885254, 8.923996925354004]
|
d702a890-ddf2-48a2-8a34-91740d1d75f4
|
saroco-detecting-satire-in-a-novel-romanian
|
2105.06456
| null |
https://arxiv.org/abs/2105.06456v3
|
https://arxiv.org/pdf/2105.06456v3.pdf
|
SaRoCo: Detecting Satire in a Novel Romanian Corpus of News Articles
|
In this work, we introduce a corpus for satire detection in Romanian news. We gathered 55,608 public news articles from multiple real and satirical news sources, composing one of the largest corpora for satire detection regardless of language and the only one for the Romanian language. We provide an official split of the text samples, such that training news articles belong to different sources than test news articles, thus ensuring that models do not achieve high performance simply due to overfitting. We conduct experiments with two state-of-the-art deep neural models, resulting in a set of strong baselines for our novel corpus. Our results show that the machine-level accuracy for satire detection in Romanian is quite low (under 73% on the test set) compared to the human-level accuracy (87%), leaving enough room for improvement in future research.
|
['Radu Tudor Ionescu', 'Mihaela Gaman', 'Ana-Cristina Rogoz']
|
2021-05-13
| null |
https://aclanthology.org/2021.acl-short.136
|
https://aclanthology.org/2021.acl-short.136.pdf
|
acl-2021-5
|
['satire-detection']
|
['natural-language-processing']
|
[ 6.56942418e-03 7.31981173e-02 -5.76883018e-01 -1.32916257e-01
-1.06579125e+00 -8.38306069e-01 1.11016369e+00 2.73175895e-01
-6.73829854e-01 4.06614184e-01 1.11090815e+00 -2.81616688e-01
3.54090631e-01 -6.38045847e-01 -3.28124195e-01 -5.05771279e-01
3.45517278e-01 7.55268037e-01 3.06098815e-02 -8.88266563e-01
4.89244878e-01 2.02817783e-01 -1.11325848e+00 8.44066560e-01
4.70913649e-01 5.99687755e-01 -3.09662968e-01 2.93214649e-01
1.15437239e-01 1.36300826e+00 -9.33863282e-01 -9.01215017e-01
1.73707753e-01 -7.42763400e-01 -8.79321992e-01 -2.54732698e-01
4.87406582e-01 -4.17609364e-01 -5.28097510e-01 1.06162393e+00
5.33212960e-01 -1.12503842e-01 5.91132939e-01 -6.95379436e-01
-8.01275730e-01 1.25448585e+00 -4.60822701e-01 3.57702523e-01
3.36393028e-01 -2.20391527e-01 1.41139543e+00 -6.89212263e-01
1.06181884e+00 1.39285195e+00 7.33545542e-01 6.53705478e-01
-8.14724684e-01 -6.68314815e-01 -1.48407176e-01 1.09933786e-01
-8.88858259e-01 -6.50551677e-01 9.01442111e-01 -3.43893558e-01
5.81174850e-01 2.09765568e-01 5.18802166e-01 1.59374416e+00
2.40999505e-01 1.03739107e+00 1.28538895e+00 -2.55688906e-01
1.15056820e-01 1.66954622e-01 6.03118166e-02 3.02048504e-01
1.52297720e-01 -2.78216362e-01 -4.69696909e-01 -5.54696500e-01
5.17515302e-01 -3.99892926e-01 -1.99462444e-01 6.48426935e-02
-1.30239725e+00 1.38540435e+00 4.01135474e-01 8.61132860e-01
-3.83236885e-01 -1.33443043e-01 9.76931691e-01 4.83082205e-01
8.21690500e-01 7.97350824e-01 -4.53349888e-01 -2.93295652e-01
-1.11167085e+00 7.59523094e-01 9.66042876e-01 2.59766191e-01
-4.16628301e-01 3.91278192e-02 -7.60773942e-02 9.21621442e-01
-2.12071538e-01 4.71795887e-01 7.56297052e-01 -6.85308278e-01
6.31536663e-01 4.35901821e-01 2.72996068e-01 -1.35983503e+00
-5.33727467e-01 -5.99421263e-01 -5.43690801e-01 -1.81475393e-02
8.08831096e-01 -1.34614795e-01 -3.38603288e-01 1.82276845e+00
-1.87481657e-01 -5.42563081e-01 2.20556542e-01 9.91534114e-01
1.19656420e+00 6.95002854e-01 -9.91396531e-02 -2.52851546e-01
1.58345616e+00 -7.80560374e-01 -5.73000729e-01 -2.40165845e-01
6.97447777e-01 -1.15942240e+00 1.28722394e+00 7.76822805e-01
-1.12906408e+00 -6.86013652e-03 -1.22089803e+00 -2.11601898e-01
-1.14280656e-01 1.26478374e-01 6.08962417e-01 3.16938728e-01
-5.09579122e-01 6.58154070e-01 -4.39876169e-01 -3.06695968e-01
4.88479257e-01 -3.36416572e-01 -5.87936752e-02 2.08204806e-01
-1.35891008e+00 1.43757737e+00 1.14960402e-01 -2.79965103e-01
-7.85693467e-01 -3.21521550e-01 -6.22692943e-01 -1.25705987e-01
3.48862499e-01 7.82814324e-02 1.55376780e+00 -1.38758183e+00
-1.25891244e+00 1.52428257e+00 3.03403348e-01 -5.88970721e-01
6.92062438e-01 -2.66609371e-01 -6.96914911e-01 7.47385919e-02
1.83607578e-01 1.66876018e-01 4.60550368e-01 -1.11685956e+00
-4.40637708e-01 -3.29952426e-02 2.75477290e-01 1.81717470e-01
-1.62631452e-01 8.42415988e-01 1.74452022e-01 -7.93851674e-01
1.66836068e-01 -8.32970858e-01 1.31816238e-01 -5.36639273e-01
-2.09081158e-01 -2.94277668e-01 4.24036890e-01 -8.95016193e-01
1.07734084e+00 -1.97667992e+00 -3.68972472e-03 -3.58885318e-01
2.53170043e-01 -7.42975250e-02 -1.53073505e-01 5.30983925e-01
6.41686320e-02 3.55949886e-02 1.47819385e-01 -2.12070122e-01
1.54638320e-01 2.25988533e-02 -7.55030751e-01 9.44401443e-01
-1.93246663e-01 8.45562041e-01 -1.10856616e+00 -1.36471346e-01
-1.69697896e-01 2.56469250e-01 -6.08962238e-01 -5.06457746e-01
-2.21928537e-01 3.21910173e-01 -3.36005986e-01 5.69117785e-01
1.35847881e-01 -7.86960823e-04 2.56298065e-01 -1.18400596e-01
-9.09985304e-02 1.41582668e+00 -5.16898572e-01 1.49644244e+00
-5.79302371e-01 8.80249858e-01 2.29069278e-01 -8.00469756e-01
7.46320069e-01 3.94434392e-01 4.24029201e-01 -8.93959999e-01
7.38736808e-01 3.75567049e-01 4.64468956e-01 -6.35314509e-02
7.39469230e-01 -5.04584253e-01 -7.13928163e-01 7.88700461e-01
-1.30517676e-01 -7.79727325e-02 1.86793387e-01 3.03162962e-01
1.06627274e+00 -2.32790768e-01 3.98863345e-01 -7.71173179e-01
2.49597788e-01 3.01379353e-01 5.31616032e-01 7.81228423e-01
-1.17782854e-01 5.17695844e-01 8.26293588e-01 -7.03828871e-01
-1.25705922e+00 -6.78183377e-01 -2.74580956e-01 1.23628068e+00
-1.38624594e-01 -5.61618865e-01 -5.62186956e-01 -7.90415049e-01
-3.67119610e-01 1.14729476e+00 -6.91635072e-01 1.33471176e-01
-7.30810642e-01 -9.07497883e-01 8.34724069e-01 1.95731178e-01
2.79504836e-01 -1.38212729e+00 -9.17166293e-01 1.78002700e-01
-5.92933297e-01 -9.43187773e-01 -2.81413555e-01 1.49770364e-01
-5.23822665e-01 -1.21413374e+00 -5.17709792e-01 -6.46697283e-01
-4.16657375e-03 1.30415246e-01 1.57660842e+00 4.74366285e-02
1.92296654e-01 -3.00571263e-01 -5.48285306e-01 -3.88991326e-01
-1.12286413e+00 -2.59686291e-01 1.22861966e-01 -5.61694980e-01
6.10536873e-01 -3.68183166e-01 -2.41529226e-01 5.37732467e-02
-6.25306308e-01 1.21049199e-03 2.19173238e-01 9.25227404e-01
-8.52268711e-02 -2.04325557e-01 5.89119732e-01 -8.75701487e-01
8.60281289e-01 -4.70078111e-01 -5.29311180e-01 -3.05510521e-01
-2.86949337e-01 -2.49350220e-01 8.35769415e-01 -7.13749051e-01
-7.43701577e-01 -5.15062511e-01 -2.86165416e-01 1.19509019e-01
1.23989813e-01 5.61391234e-01 2.27051482e-01 5.06837189e-01
1.22198474e+00 -7.56796598e-02 -1.37877718e-01 -6.74184322e-01
5.69553375e-01 7.27268159e-01 8.81312013e-01 -4.42963928e-01
4.93973553e-01 4.41847384e-01 -5.28402269e-01 -4.41385090e-01
-1.53854322e+00 -4.61412251e-01 -3.05720776e-01 1.85160011e-01
5.36930442e-01 -1.03808987e+00 -5.39252222e-01 2.98703969e-01
-1.25177860e+00 -3.71164083e-01 -2.49718919e-01 4.10982788e-01
-4.87796903e-01 2.62607306e-01 -1.27109456e+00 -6.62650347e-01
-6.60268962e-01 -8.89295757e-01 6.73994720e-01 -2.94577360e-01
-7.90540040e-01 -1.01522362e+00 2.92915434e-01 5.04489303e-01
3.07524294e-01 4.67045218e-01 8.33160937e-01 -1.36204600e+00
5.81146121e-01 -3.81792307e-01 -1.21984854e-01 1.76954597e-01
-5.34939133e-02 -4.48496789e-01 -9.31829035e-01 -1.89711064e-01
6.93077922e-01 -9.24194276e-01 9.76874352e-01 1.88904449e-01
4.14135486e-01 -7.45194435e-01 7.42005110e-02 -8.15834701e-02
1.02494550e+00 -4.26250733e-02 7.61636615e-01 7.76386917e-01
3.20389062e-01 3.37717444e-01 3.59756559e-01 5.26243448e-01
1.47056863e-01 6.20179057e-01 1.98205501e-01 -1.52856648e-01
-5.60127050e-02 -3.06669861e-01 6.71503246e-01 5.89983463e-01
1.09207086e-01 -2.86815166e-01 -6.48155868e-01 6.54102385e-01
-1.74704611e+00 -1.56806684e+00 -3.13665956e-01 2.07444286e+00
1.06714702e+00 4.37529147e-01 4.05882537e-01 1.98136494e-01
4.15352821e-01 4.65816200e-01 6.01333678e-02 -6.84258938e-01
-4.21626151e-01 3.81366014e-01 4.82434988e-01 6.02117121e-01
-1.34098220e+00 1.06699443e+00 7.29594421e+00 6.91564739e-01
-1.08648694e+00 3.13266367e-01 5.66519260e-01 -4.00473982e-01
-3.59358132e-01 -9.85793024e-02 -5.66598713e-01 3.45564574e-01
8.02890897e-01 -7.33404830e-02 4.44486648e-01 1.12191641e+00
2.55434096e-01 -1.46620180e-02 -7.90632010e-01 9.32448804e-01
5.79266906e-01 -1.37326586e+00 -3.37487280e-01 -4.31820452e-02
8.07431757e-01 6.77566290e-01 -2.43514463e-01 4.88928586e-01
5.07601023e-01 -9.25354838e-01 1.18837535e+00 -3.91070634e-01
4.85599905e-01 -7.74049759e-01 8.93633962e-01 4.63131815e-01
-1.85261071e-01 1.29866585e-01 -2.72085369e-01 -5.44530094e-01
2.69652158e-01 4.38372493e-01 -6.32443011e-01 -5.10507114e-02
3.16308171e-01 7.04465032e-01 -4.88441646e-01 4.91416514e-01
-4.49673086e-01 8.79540682e-01 -3.80011529e-01 -3.00107002e-01
6.00442410e-01 1.95414037e-01 6.38732910e-01 1.33087242e+00
-7.40200430e-02 6.88464940e-02 2.71124333e-01 8.38101327e-01
-3.50710124e-01 4.35295373e-01 -5.51293969e-01 -1.93208843e-01
1.06075659e-01 1.54220414e+00 -8.99545074e-01 -5.53635836e-01
-5.12380660e-01 9.34143066e-01 2.27430671e-01 -2.03433648e-01
-9.24576342e-01 -4.38605733e-02 2.49518111e-01 2.72839844e-01
-6.45681843e-02 -1.45286158e-01 -5.82210839e-01 -1.34528100e+00
-2.65731990e-01 -1.36204457e+00 5.76110899e-01 -6.06110454e-01
-1.57859266e+00 5.71486712e-01 -5.96575774e-02 -9.69004750e-01
-2.56888241e-01 -7.40101397e-01 -6.11896873e-01 5.28057218e-01
-8.62725317e-01 -9.92919862e-01 3.09190929e-01 2.85590649e-01
6.75397038e-01 -2.04283848e-01 9.06291485e-01 2.83183426e-01
6.16503395e-02 4.37778980e-01 1.15049981e-01 5.16081750e-01
7.89591134e-01 -9.39806640e-01 5.18476844e-01 7.09271789e-01
6.16570234e-01 6.14980757e-01 1.18501306e+00 -5.64190805e-01
-9.74202514e-01 -3.63955110e-01 1.49199414e+00 -7.79792190e-01
1.26256979e+00 -4.53921020e-01 -7.09979653e-01 6.35461509e-01
4.78425562e-01 -8.80121827e-01 6.17181957e-01 5.09231448e-01
-7.45791852e-01 5.50952375e-01 -1.17769396e+00 7.71002114e-01
8.04986835e-01 -3.17396224e-01 -1.08288372e+00 7.05263257e-01
2.49450326e-01 -4.50909555e-01 -3.88834506e-01 2.47569352e-01
6.12878978e-01 -9.21351671e-01 6.34835124e-01 -7.00877666e-01
1.15697837e+00 7.25308582e-02 -2.54024625e-01 -1.18818104e+00
-4.67824697e-01 -3.42858702e-01 1.96075007e-01 9.88876700e-01
4.74975944e-01 -2.89750546e-01 5.74889004e-01 1.29652455e-01
-1.66793123e-01 -5.63267827e-01 -8.29226792e-01 -4.62751925e-01
5.31781316e-01 -3.75239849e-01 -4.71109450e-02 1.50684464e+00
4.93648916e-01 9.47598338e-01 -5.80455840e-01 -6.08655334e-01
1.39910474e-01 4.04828906e-01 5.88656366e-01 -1.16790116e+00
-4.76280302e-01 -9.04527664e-01 -1.73534647e-01 -8.64303946e-01
3.42511743e-01 -1.07366621e+00 -1.40714347e-02 -1.49494028e+00
7.38945961e-01 -3.22609171e-02 -4.32473123e-02 7.04245865e-01
2.95245737e-01 8.30348492e-01 -5.08989766e-03 3.51990551e-01
-4.23703313e-01 4.00471210e-01 1.11673915e+00 -1.61941588e-01
-2.45602533e-01 -2.94374347e-01 -1.16202295e+00 1.07218099e+00
9.69249427e-01 -5.55785775e-01 -1.00716695e-01 -1.49661914e-01
3.78942043e-01 -4.29685771e-01 3.52142602e-01 -6.11086130e-01
-4.64986473e-01 -2.62672454e-01 3.29212248e-01 -5.94325185e-01
1.34936363e-01 -1.57675534e-01 -2.72145182e-01 6.41939461e-01
-5.75249016e-01 1.12952225e-01 2.05433443e-01 -6.94568530e-02
5.52662611e-02 -1.70045465e-01 1.07643783e+00 -2.93272734e-01
-3.42066139e-01 -4.32382941e-01 -6.33289635e-01 5.26323795e-01
2.25360692e-01 4.56233859e-01 -7.86521912e-01 -7.27277875e-01
-2.68746078e-01 -3.80502284e-01 6.65679693e-01 6.98616445e-01
-8.24144692e-04 -1.39116621e+00 -1.22311997e+00 -4.59917217e-01
6.99572563e-02 -7.02018023e-01 -3.97237360e-01 9.10062611e-01
-5.94032049e-01 3.00995082e-01 -1.09038234e-01 5.36256209e-02
-1.13366807e+00 7.15415537e-01 1.13058276e-02 -3.78506273e-01
-1.09587514e+00 4.51990277e-01 2.88641620e-02 -1.96235895e-01
-4.43323664e-02 -1.22920178e-01 -2.68302500e-01 8.78691226e-02
8.86718392e-01 1.58371422e-02 -8.77691898e-03 -1.14775968e+00
-4.03151512e-01 -1.54307485e-01 -1.74012899e-01 -4.77032781e-01
1.22966385e+00 2.79677302e-01 -1.31781176e-01 7.00232863e-01
9.52977300e-01 6.66437626e-01 -2.73067564e-01 -6.71160221e-02
-4.99613583e-02 -2.44337901e-01 9.44265053e-02 -1.19813192e+00
-5.38136184e-01 2.73990899e-01 -5.02770096e-02 2.93995410e-01
6.16353154e-01 3.30393821e-01 8.90224516e-01 3.99842411e-01
2.78946787e-01 -1.20266175e+00 -2.39537116e-02 8.59347820e-01
1.28472817e+00 -9.78401423e-01 3.34373653e-01 8.50030184e-02
-8.99661005e-01 9.84581172e-01 4.98468205e-02 -7.42560625e-01
2.55538702e-01 3.14444274e-01 5.44497311e-01 -4.77461904e-01
-6.19621336e-01 1.07519343e-01 3.69481474e-01 -8.35379884e-02
1.01860332e+00 -2.35410593e-02 -1.20376444e+00 1.01959884e+00
-8.14847767e-01 -3.78428817e-01 5.71728706e-01 5.21971524e-01
-6.67789757e-01 -8.02066326e-01 -2.90175229e-01 2.49768078e-01
-1.08628559e+00 -3.38203698e-01 -1.13247442e+00 8.87815475e-01
-1.85622618e-01 1.06594682e+00 -5.61413802e-02 -3.18795174e-01
1.52984336e-01 5.24565876e-02 6.38514042e-01 -3.66197228e-01
-9.88036394e-01 3.12450647e-01 8.37816119e-01 -3.05395484e-01
-3.14906776e-01 -4.69392329e-01 -1.13890529e+00 -6.02540195e-01
-2.35599518e-01 2.58876145e-01 4.44521964e-01 1.15396988e+00
-2.06330851e-01 -1.17053287e-02 4.01387155e-01 -8.65623236e-01
-8.29287946e-01 -1.28293979e+00 -4.57020849e-01 8.97765398e-01
-3.99172492e-02 -5.84582448e-01 -4.78245050e-01 -2.20786154e-01]
|
[8.593968391418457, 10.462284088134766]
|
c2c25b29-8f3d-4c6d-8ea6-9b994f0137fe
|
learning-to-segment-rigid-motions-from-two
|
2101.03694
| null |
https://arxiv.org/abs/2101.03694v1
|
https://arxiv.org/pdf/2101.03694v1.pdf
|
Learning to Segment Rigid Motions from Two Frames
|
Appearance-based detectors achieve remarkable performance on common scenes, but tend to fail for scenarios lack of training data. Geometric motion segmentation algorithms, however, generalize to novel scenes, but have yet to achieve comparable performance to appearance-based ones, due to noisy motion estimations and degenerate motion configurations. To combine the best of both worlds, we propose a modular network, whose architecture is motivated by a geometric analysis of what independent object motions can be recovered from an egomotion field. It takes two consecutive frames as input and predicts segmentation masks for the background and multiple rigidly moving objects, which are then parameterized by 3D rigid transformations. Our method achieves state-of-the-art performance for rigid motion segmentation on KITTI and Sintel. The inferred rigid motions lead to a significant improvement for depth and scene flow estimation. At the time of submission, our method ranked 1st on KITTI scene flow leaderboard, out-performing the best published method (scene flow error: 4.89% vs 6.31%).
|
['Deva Ramanan', 'Gengshan Yang']
|
2021-01-11
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Yang_Learning_To_Segment_Rigid_Motions_From_Two_Frames_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Yang_Learning_To_Segment_Rigid_Motions_From_Two_Frames_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['motion-segmentation', 'scene-flow-estimation']
|
['computer-vision', 'computer-vision']
|
[ 9.59801376e-02 -6.32925406e-02 -1.58611700e-01 -1.60707906e-01
-6.30865693e-01 -7.66441703e-01 6.41918004e-01 -4.31841582e-01
-4.37435627e-01 3.12607706e-01 5.37911952e-02 -3.26958783e-02
3.20464194e-01 -4.03341830e-01 -6.17748082e-01 -6.29080892e-01
6.65021464e-02 7.71415412e-01 7.71863461e-01 -1.36115164e-01
1.90624520e-01 5.42768955e-01 -1.29164720e+00 1.47955149e-01
5.13469934e-01 5.53002119e-01 1.59177169e-01 1.05157542e+00
-1.12381920e-01 1.07416236e+00 -1.78584889e-01 -1.85337931e-01
4.06182975e-01 -4.98178571e-01 -1.19947696e+00 3.95068854e-01
1.01225328e+00 -5.92527151e-01 -7.59141862e-01 7.30532110e-01
2.42271632e-01 4.56462443e-01 6.64256513e-01 -1.03523767e+00
-7.55926520e-02 3.07102472e-01 -6.80666506e-01 4.53279585e-01
2.49544680e-01 4.65467274e-01 7.26293385e-01 -8.48397315e-01
1.18689406e+00 1.30633187e+00 5.15118480e-01 8.51067245e-01
-1.26519537e+00 -2.12079689e-01 4.24390525e-01 7.91694522e-02
-9.53072429e-01 -6.63637757e-01 6.45240426e-01 -7.88819253e-01
1.11076498e+00 6.70513213e-02 6.76503897e-01 1.05736279e+00
9.93609726e-02 1.10542822e+00 6.58046305e-01 -8.02253857e-02
1.59465685e-01 -1.64007947e-01 -1.53471231e-01 8.25491011e-01
3.23785506e-02 -1.19296443e-02 -3.20117533e-01 2.74928898e-01
9.20002103e-01 -3.75704527e-01 -3.20990920e-01 -6.90142095e-01
-1.32616401e+00 5.91102719e-01 3.86613518e-01 2.15503722e-01
-1.30868748e-01 6.48081243e-01 2.04886124e-01 -1.61940977e-01
4.80834067e-01 1.25475422e-01 -4.19879913e-01 -1.86536774e-01
-1.27580500e+00 4.38649148e-01 7.57932067e-01 8.86918008e-01
8.18217576e-01 4.14842546e-01 -2.43850984e-02 3.76744688e-01
3.74730587e-01 5.47518253e-01 2.03496695e-01 -1.56094730e+00
4.76481795e-01 2.81052381e-01 8.24943483e-02 -8.50072026e-01
-6.20766282e-01 -1.40439466e-01 -5.89412987e-01 3.69681418e-01
8.94439816e-01 -1.45292148e-01 -1.26038635e+00 1.49182081e+00
5.32991886e-01 4.02811885e-01 -1.47814244e-01 1.15795445e+00
9.17959571e-01 5.04163563e-01 8.29833597e-02 -9.53150447e-03
8.28278780e-01 -1.39811158e+00 -3.46876502e-01 -6.00305319e-01
6.94050372e-01 -8.41721594e-01 4.67409015e-01 3.54639083e-01
-1.40354848e+00 -5.85033298e-01 -7.51152694e-01 -2.47415751e-01
-9.47752036e-04 1.14770979e-02 6.58090234e-01 6.50599182e-01
-1.29907310e+00 7.33156145e-01 -1.22792220e+00 -4.32433188e-01
4.79841053e-01 3.41368586e-01 -4.35032248e-01 -2.53080018e-02
-5.59460104e-01 8.29298019e-01 2.61122227e-01 2.14489818e-01
-1.13223624e+00 -7.99629748e-01 -1.00309777e+00 -3.26651752e-01
3.15051407e-01 -1.23499691e+00 1.27172410e+00 -1.01517093e+00
-1.73682678e+00 1.14399254e+00 -3.25243086e-01 -5.03099382e-01
1.19506240e+00 -2.63957888e-01 5.50662540e-02 4.14239138e-01
4.72721048e-02 1.19806635e+00 8.70187998e-01 -1.15385163e+00
-7.27641523e-01 -7.34502077e-02 -1.52760595e-02 2.84041166e-01
1.90221146e-01 -5.83903007e-02 -8.76985967e-01 -3.87819380e-01
2.23489180e-01 -1.09618461e+00 -6.00119948e-01 2.89237946e-02
-4.27926123e-01 1.14884958e-01 8.84123862e-01 -7.81922340e-01
6.65503323e-01 -1.78459060e+00 5.35371125e-01 -2.34736323e-01
2.10067496e-01 2.99799860e-01 -1.06628403e-01 -7.22439438e-02
6.26727790e-02 -7.29213357e-02 -5.47774851e-01 -7.88604021e-01
-2.42580444e-01 1.89477086e-01 -2.58697867e-01 7.94767976e-01
3.01172674e-01 1.08905172e+00 -9.94982302e-01 -5.56465805e-01
8.82344425e-01 3.94853473e-01 -6.17529631e-01 7.29474099e-03
-4.20220673e-01 9.53864694e-01 -3.93568695e-01 4.95286226e-01
7.62129903e-01 -1.30419493e-01 2.50957292e-02 1.40166402e-01
-2.48320371e-01 1.71922863e-01 -1.28808510e+00 2.19606709e+00
-1.13924325e-01 1.00125933e+00 1.61259055e-01 -8.02606523e-01
5.06195784e-01 6.17542341e-02 8.87882352e-01 -3.59482616e-01
1.84517577e-01 7.23588243e-02 1.18489601e-02 -6.44521415e-01
4.82928008e-01 6.40412420e-02 2.08665028e-01 6.97571263e-02
1.75081268e-01 -5.95503807e-01 2.86204845e-01 1.90938652e-01
1.14684772e+00 6.97728634e-01 -8.17363709e-02 -2.15867609e-01
6.56706452e-01 3.68010908e-01 5.80312490e-01 7.02150583e-01
-4.54676390e-01 1.10896873e+00 2.98673809e-01 -8.80894780e-01
-1.11269724e+00 -1.05532658e+00 5.83660901e-02 7.55308926e-01
4.90689486e-01 -1.29346967e-01 -8.30771148e-01 -6.85914040e-01
-2.66569942e-01 4.86555308e-01 -5.59008181e-01 2.61978537e-01
-1.06119525e+00 -6.96810663e-01 4.06524450e-01 5.72492957e-01
4.17845875e-01 -1.08812201e+00 -8.17623556e-01 4.25195307e-01
-3.22541445e-01 -1.71687937e+00 -3.68288368e-01 -1.65886298e-01
-1.12289655e+00 -1.11600590e+00 -8.09709132e-01 -5.17474413e-01
3.49675298e-01 3.48668069e-01 1.26355517e+00 -1.65942922e-01
-6.55253589e-01 4.87995028e-01 -4.01415788e-02 7.26911277e-02
-5.05011618e-01 1.88110828e-01 -8.33430439e-02 3.11248470e-02
-1.55153915e-01 -4.57559764e-01 -9.37723994e-01 2.84317136e-01
-8.13608050e-01 1.49558231e-01 3.66036028e-01 3.23687732e-01
4.44364816e-01 -5.06910443e-01 -3.53862494e-01 -6.65780127e-01
-3.95149201e-01 -2.67365813e-01 -7.30701029e-01 -5.85478172e-02
-9.06343386e-02 1.96213443e-02 2.58272886e-01 -2.21772254e-01
-1.20674694e+00 6.45644844e-01 -2.66204518e-03 -6.96683764e-01
-5.36930025e-01 -3.87834668e-01 1.43608466e-01 -3.26799601e-01
5.68905652e-01 1.02551721e-01 -3.41751903e-01 -2.92104244e-01
7.62094021e-01 -1.15569696e-01 7.90197730e-01 -3.72077614e-01
8.80325675e-01 1.17004812e+00 2.45093703e-01 -1.03271043e+00
-6.67886019e-01 -7.63423443e-01 -1.10313344e+00 -3.52675050e-01
1.36474133e+00 -9.61004674e-01 -5.12971580e-01 7.38358498e-01
-1.38970542e+00 -7.95146763e-01 -2.14352891e-01 5.58565497e-01
-8.14916193e-01 6.43401265e-01 -7.33312428e-01 -7.72820473e-01
-1.25986397e-01 -1.29325318e+00 1.18995214e+00 2.29721397e-01
-1.77898854e-01 -1.19694805e+00 2.52692163e-01 5.45390725e-01
2.20142290e-01 5.62855899e-01 2.55906284e-01 -5.19620590e-02
-1.11516178e+00 1.57481171e-02 -6.80832937e-02 1.64894283e-01
-1.39279336e-01 4.65122551e-01 -1.20894349e+00 -2.12585747e-01
-1.20226793e-01 -1.07533939e-01 1.22423851e+00 8.78365993e-01
7.81067908e-01 2.44008273e-01 -5.16282976e-01 1.09025335e+00
1.30609429e+00 3.89518812e-02 6.85251355e-01 3.60647827e-01
1.25375104e+00 7.35773265e-01 3.49572420e-01 1.94273904e-01
3.31382543e-01 9.06032085e-01 6.20528758e-01 -1.71741962e-01
-5.77732265e-01 4.93692942e-02 4.55301017e-01 5.47426224e-01
-3.75978649e-01 -3.23052317e-01 -1.02803266e+00 6.31675720e-01
-1.81819463e+00 -9.76277053e-01 -7.36826599e-01 1.98517418e+00
1.71450242e-01 2.04100370e-01 2.58842319e-01 -4.09194648e-01
5.15534699e-01 3.56996864e-01 -6.68980300e-01 -2.28906006e-01
-2.44163737e-01 1.30240778e-02 6.89504981e-01 8.06360364e-01
-1.44096971e+00 1.45726252e+00 6.63565493e+00 5.55664957e-01
-1.03944516e+00 1.95727721e-01 7.59597659e-01 -1.69074044e-01
-1.00531643e-02 3.27780873e-01 -8.30877185e-01 1.90041125e-01
6.35276258e-01 2.42429867e-01 2.13995621e-01 6.00811422e-01
3.79677176e-01 -2.28812963e-01 -1.02488899e+00 1.07983661e+00
1.42823726e-01 -1.43391931e+00 -1.46310031e-01 6.45516515e-02
1.14388728e+00 5.01079142e-01 -6.22652955e-02 -9.86192003e-02
4.19495136e-01 -9.53842878e-01 9.54568744e-01 5.05817831e-01
3.63779962e-01 -3.52062166e-01 4.72593486e-01 2.78273821e-01
-1.28042316e+00 2.75870323e-01 -2.19468936e-01 -6.68643638e-02
5.85942566e-01 2.62502283e-01 -6.26360118e-01 6.09267235e-01
5.80451906e-01 9.73688543e-01 -4.51578170e-01 1.19772255e+00
-2.95195103e-01 3.93046111e-01 -4.05869573e-01 4.21114355e-01
5.72003663e-01 -2.90933669e-01 8.80002677e-01 1.44529343e+00
6.63762400e-03 1.01926522e-02 3.26246470e-01 9.97515380e-01
8.61637518e-02 -1.31205350e-01 -6.17943525e-01 4.36525404e-01
-1.05178252e-01 1.45883060e+00 -1.12218070e+00 -4.41243380e-01
-3.32088143e-01 1.28076231e+00 2.35870287e-01 4.33615923e-01
-8.69400322e-01 3.03232431e-01 7.30780721e-01 1.66714609e-01
3.53219330e-01 -6.00081861e-01 -1.39278784e-01 -1.41696799e+00
-7.09877312e-02 -3.19839686e-01 2.08799124e-01 -6.60975993e-01
-8.33665192e-01 4.68508601e-01 8.71200487e-02 -1.13319159e+00
-4.20527250e-01 -7.36559808e-01 -6.72589481e-01 5.98333359e-01
-1.26879215e+00 -1.09403241e+00 -4.51504111e-01 2.98027545e-01
9.74395335e-01 1.64368436e-01 3.88552994e-01 2.18781784e-01
-5.64676166e-01 1.59525529e-01 -4.19035964e-02 2.48874515e-01
5.94044089e-01 -1.29684949e+00 9.54842925e-01 1.21064615e+00
4.97054696e-01 -3.93849798e-02 6.85914934e-01 -4.74074304e-01
-1.29999876e+00 -1.08553898e+00 5.96588492e-01 -1.01068127e+00
5.92058182e-01 -2.81893283e-01 -8.15719306e-01 6.44292176e-01
7.95203149e-02 3.71064276e-01 3.38970637e-03 -5.61501682e-01
-1.56791106e-01 2.94821858e-01 -7.94043779e-01 6.37626410e-01
1.41713703e+00 -1.76764011e-01 -2.99486220e-01 2.81988412e-01
6.18620038e-01 -9.33673322e-01 -4.43349361e-01 2.99844414e-01
3.67784977e-01 -1.22021186e+00 1.20163131e+00 -6.98060930e-01
5.87092817e-01 -5.38224459e-01 3.71768102e-02 -8.54515314e-01
-2.49849558e-01 -1.05528533e+00 -1.61941737e-01 8.65431130e-01
1.56879604e-01 -1.91949233e-01 1.16621697e+00 7.24687219e-01
-2.88053542e-01 -3.00720096e-01 -8.59033525e-01 -7.57540405e-01
1.81976795e-01 -5.93471169e-01 -1.86551586e-01 9.46503103e-01
-6.02540493e-01 3.41853589e-01 -3.96120876e-01 -7.00213853e-03
7.39363730e-01 1.09164357e-01 1.12557852e+00 -9.01375830e-01
-4.14299726e-01 -7.75361419e-01 -6.67781293e-01 -1.35072505e+00
2.87781179e-01 -8.39221418e-01 2.05801293e-01 -1.55173826e+00
6.61615655e-02 -2.04050854e-01 2.46766329e-01 2.79225200e-01
-1.53212383e-01 5.09521365e-01 3.67207944e-01 1.87799543e-01
-7.48113394e-01 3.48501265e-01 1.39309812e+00 -2.04625472e-01
-4.61800575e-01 -6.10380135e-02 -7.87010342e-02 1.09971142e+00
6.27045333e-01 -3.09351295e-01 -3.46557438e-01 -7.70753145e-01
-1.92538053e-01 2.04073712e-01 6.21564269e-01 -1.18829823e+00
2.54792124e-01 -2.15392545e-01 4.43578690e-01 -5.89438677e-01
4.24179822e-01 -4.04424936e-01 2.47348472e-01 5.59177637e-01
3.94289605e-02 -1.42408628e-03 3.74926418e-01 5.80107808e-01
-9.14610997e-02 -8.58161598e-02 8.80530655e-01 -3.30171764e-01
-1.15929365e+00 5.73014379e-01 -4.97212380e-01 4.02136981e-01
8.92441750e-01 -4.62377220e-01 -1.12006843e-01 -2.29199976e-01
-8.78827572e-01 6.40322939e-02 5.31315446e-01 5.83584189e-01
5.18640637e-01 -8.40993047e-01 -8.80229294e-01 -1.06444553e-01
-1.93479627e-01 4.03226644e-01 5.26478589e-01 1.01766968e+00
-1.14139664e+00 3.98619175e-01 -1.24666199e-01 -1.08087361e+00
-1.09531760e+00 3.44879955e-01 6.65430665e-01 -1.86284512e-01
-1.00157297e+00 1.03346586e+00 4.93767887e-01 -1.68530032e-01
2.13270411e-02 -5.56996405e-01 1.24487318e-01 -1.84894800e-01
3.39253336e-01 5.20134330e-01 -6.61278260e-04 -9.47459698e-01
-5.45766652e-01 1.09933770e+00 2.85998471e-02 -3.85759890e-01
9.81463671e-01 -1.60979241e-01 1.12492107e-01 3.41207772e-01
1.17617130e+00 -2.10944459e-01 -1.87444592e+00 6.75195828e-02
-1.43784791e-01 -5.97069740e-01 -2.51594037e-02 -3.93394023e-01
-1.22019219e+00 1.08270991e+00 4.69907433e-01 -2.03767568e-01
7.89311945e-01 4.34384011e-02 8.23475242e-01 1.77872866e-01
1.83120817e-01 -9.03444529e-01 6.85570017e-02 7.48853445e-01
4.19153690e-01 -1.35245562e+00 -1.50025981e-02 -5.74781895e-01
-4.80934262e-01 1.24836433e+00 6.71492517e-01 -4.06846434e-01
2.50306219e-01 1.42960578e-01 2.52803504e-01 -7.96278268e-02
-5.89677453e-01 -3.19920868e-01 4.56381351e-01 6.01485014e-01
2.31473789e-01 -8.41328874e-02 1.82469681e-01 -2.81608433e-01
-9.57334414e-03 -2.53631890e-01 4.46151286e-01 5.83987832e-01
-2.62151867e-01 -9.24664974e-01 -2.46411443e-01 -1.64686367e-01
-3.91204417e-01 1.10905029e-01 -4.72977936e-01 1.07057047e+00
9.87869576e-02 8.02651286e-01 2.15716332e-01 -4.74379994e-02
2.81102002e-01 -7.36622736e-02 7.65807688e-01 -3.69592726e-01
-4.80907768e-01 3.16062748e-01 2.44978461e-02 -1.01174366e+00
-7.26244152e-01 -8.85964572e-01 -1.24568295e+00 -2.35577270e-01
1.49212331e-01 -4.39402193e-01 5.69535315e-01 1.05882049e+00
5.52053079e-02 4.93606389e-01 3.37110698e-01 -1.47666323e+00
5.25519364e-02 -6.06138051e-01 -1.84820116e-01 6.49900436e-01
4.12996769e-01 -4.61335748e-01 -4.19659555e-01 5.38512647e-01]
|
[8.549690246582031, -1.7266886234283447]
|
8a4be2f3-f7ce-4d51-b950-9cfcee755eb1
|
stochastic-dimension-reduced-second-order
|
2301.12174
| null |
https://arxiv.org/abs/2301.12174v1
|
https://arxiv.org/pdf/2301.12174v1.pdf
|
Stochastic Dimension-reduced Second-order Methods for Policy Optimization
|
In this paper, we propose several new stochastic second-order algorithms for policy optimization that only require gradient and Hessian-vector product in each iteration, making them computationally efficient and comparable to policy gradient methods. Specifically, we propose a dimension-reduced second-order method (DR-SOPO) which repeatedly solves a projected two-dimensional trust region subproblem. We show that DR-SOPO obtains an $\mathcal{O}(\epsilon^{-3.5})$ complexity for reaching approximate first-order stationary condition and certain subspace second-order stationary condition. In addition, we present an enhanced algorithm (DVR-SOPO) which further improves the complexity to $\mathcal{O}(\epsilon^{-3})$ based on the variance reduction technique. Preliminary experiments show that our proposed algorithms perform favorably compared with stochastic and variance-reduced policy gradient methods.
|
['Yinyu Ye', 'Dongdong Ge', 'Qi Deng', 'Chenghan Xie', 'Jinsong Liu']
|
2023-01-28
| null | null | null | null |
['policy-gradient-methods']
|
['methodology']
|
[-1.81295842e-01 -6.34650060e-04 -3.17451954e-01 7.81973004e-02
-7.07250357e-01 -6.29512548e-01 2.89758027e-01 -4.90140766e-02
-8.89264703e-01 1.09802973e+00 1.55870140e-01 -9.38691735e-01
-3.71918738e-01 -3.01024675e-01 -6.25715077e-01 -6.60716116e-01
-2.87473857e-01 2.27996022e-01 2.47810572e-01 -2.49825820e-01
6.01341665e-01 4.37259585e-01 -1.18142068e+00 -3.11976641e-01
1.26104128e+00 1.26538599e+00 1.01663478e-01 5.62385738e-01
1.96935743e-01 4.41404313e-01 -2.05921695e-01 9.62630957e-02
6.10773742e-01 -5.62654793e-01 -5.27607560e-01 -1.40016422e-01
7.89316148e-02 -3.84275615e-01 -3.55585933e-01 1.39529967e+00
6.01571083e-01 5.95031023e-01 7.50508487e-01 -9.70399559e-01
-1.82233945e-01 1.83392763e-02 -7.94431090e-01 2.57048666e-01
3.39355230e-01 8.05044621e-02 7.30396748e-01 -8.28190088e-01
4.55416292e-01 1.19802582e+00 4.84039754e-01 3.67261201e-01
-1.08289063e+00 -4.75736916e-01 6.45160079e-01 -8.93684551e-02
-1.12984264e+00 -2.47801438e-01 5.83823323e-01 -4.02463466e-01
7.66005933e-01 5.14318407e-01 6.51747048e-01 4.55961049e-01
2.49608472e-01 9.50033188e-01 1.63242853e+00 -3.65265369e-01
5.68785191e-01 3.94504815e-01 -1.78701326e-01 9.59103942e-01
3.41347784e-01 4.58489954e-01 -2.33650655e-02 -4.16190982e-01
9.41223979e-01 -1.03392959e-01 -4.26328778e-01 -4.57690448e-01
-9.63271201e-01 1.02165926e+00 3.05379741e-02 1.42010808e-01
-4.79840070e-01 1.58888489e-01 1.88200429e-01 2.65951216e-01
6.08930707e-01 3.85868609e-01 -3.56985033e-01 -6.04496598e-01
-1.06384254e+00 5.48776150e-01 8.96613002e-01 9.84133601e-01
4.48810637e-01 3.02583814e-01 -3.20147127e-01 5.44538856e-01
4.87737834e-01 8.83344710e-01 2.19735116e-01 -1.21102381e+00
4.86114085e-01 2.37727538e-01 8.98254573e-01 -1.05859137e+00
-3.37099254e-01 -5.68430901e-01 -7.04150975e-01 3.87530118e-01
4.98030484e-01 -6.84102714e-01 -5.48066974e-01 1.51622891e+00
4.99613643e-01 -1.11361012e-01 -5.48224002e-02 1.03334558e+00
-1.63165748e-01 8.44901860e-01 -4.53531891e-01 -8.91771436e-01
9.86351788e-01 -8.70235324e-01 -6.81651294e-01 -2.35415325e-01
4.91375238e-01 -6.36387289e-01 1.30817211e+00 3.63656580e-01
-1.10263658e+00 -6.55613989e-02 -1.00869536e+00 5.69265962e-01
-3.33978832e-02 3.00879329e-01 5.73622465e-01 6.16250038e-01
-9.99891818e-01 8.37109923e-01 -6.55361712e-01 9.37904865e-02
1.44094229e-01 3.90741915e-01 1.60464436e-01 3.74555558e-01
-9.57325876e-01 7.49790311e-01 2.15975687e-01 1.65979192e-01
-9.00627434e-01 -5.68146288e-01 -6.34272754e-01 -1.47315398e-01
7.28706419e-01 -1.19992986e-01 1.29321182e+00 -2.06909865e-01
-2.10020900e+00 1.81936592e-01 -4.99303669e-01 -4.23562616e-01
8.58981490e-01 -3.78529578e-01 -1.29121393e-01 1.64047509e-01
1.24397222e-03 3.84748541e-02 8.33328962e-01 -1.27640271e+00
-8.72865379e-01 -5.19483507e-01 1.57215729e-01 4.87784982e-01
-4.93834883e-01 3.84431407e-02 -3.08440030e-01 -5.84118307e-01
2.62275487e-01 -1.20318913e+00 -7.68806756e-01 -1.84105530e-01
-2.42767215e-01 6.62971064e-02 6.51539207e-01 -7.71750808e-01
1.67694354e+00 -1.87126911e+00 2.42468923e-01 4.88243699e-01
4.17389460e-02 4.59056228e-01 3.82851422e-01 1.92307606e-01
2.81293750e-01 1.40977845e-01 -1.57403648e-01 -1.81579083e-01
9.30074453e-02 9.74360034e-02 -3.33972007e-01 8.63200843e-01
-3.45378906e-01 4.25238550e-01 -1.07067585e+00 -2.42272839e-01
9.07014087e-02 1.82633057e-01 -7.34519601e-01 -2.22418249e-01
-7.35759363e-02 5.71685255e-01 -8.46937895e-01 2.67900139e-01
6.89645767e-01 4.06889655e-02 1.12999566e-01 2.02849861e-02
-5.87601244e-01 -2.03001462e-02 -1.67351294e+00 1.28123641e+00
-5.31552672e-01 2.67676353e-01 4.58604425e-01 -1.02091062e+00
7.38978744e-01 2.30371580e-01 8.32495213e-01 -5.21279514e-01
3.37723196e-01 3.21319193e-01 -5.72048604e-01 -1.21076763e-01
4.13240433e-01 -2.03032404e-01 1.26355037e-01 4.76732284e-01
-3.26832563e-01 -1.13467291e-01 3.10806751e-01 1.11994125e-01
7.66876280e-01 1.72335878e-01 2.87124544e-01 -6.80841506e-01
7.22222567e-01 -1.89744696e-01 9.01712060e-01 8.96424413e-01
-6.34678900e-01 -8.51898268e-02 8.30184102e-01 -2.21295193e-01
-9.03673708e-01 -7.14991629e-01 3.21768485e-02 9.30714548e-01
1.33142740e-01 -2.59225070e-01 -7.92063415e-01 -7.96231389e-01
2.10669294e-01 8.53295088e-01 -5.69341004e-01 -1.35821655e-01
-3.92384052e-01 -3.58318508e-01 1.66397080e-01 3.17334652e-01
5.71661949e-01 -4.12401348e-01 -6.26225650e-01 3.05947840e-01
1.19588338e-01 -6.80830061e-01 -9.07972395e-01 9.47825685e-02
-1.08290958e+00 -6.83907330e-01 -1.02456129e+00 -3.37004274e-01
9.42144156e-01 2.27888703e-01 2.49813825e-01 -5.63736260e-01
1.45564571e-01 3.90908301e-01 -2.29706541e-01 -5.33352137e-01
4.21293080e-02 -1.31264925e-01 7.22425044e-01 9.94908363e-02
5.27096689e-02 -3.87813032e-01 -6.54616773e-01 3.78120273e-01
-5.10088742e-01 -3.48604560e-01 3.85939389e-01 7.57523000e-01
8.65775943e-01 -7.08859488e-02 2.21415922e-01 -6.97635949e-01
1.23667991e+00 -5.59011288e-02 -1.60713899e+00 8.55713710e-02
-1.07897866e+00 5.03809035e-01 8.78790975e-01 -4.96602207e-01
-1.09215319e+00 2.36897156e-01 2.47888193e-01 -7.80479610e-01
3.52266908e-01 3.36125433e-01 2.96909422e-01 -3.00308138e-01
6.89645946e-01 6.16034925e-01 1.63922325e-01 -6.42000377e-01
5.09051800e-01 5.62411129e-01 3.27974200e-01 -8.23075056e-01
8.09506536e-01 5.43032706e-01 3.09285112e-02 -5.31157613e-01
-8.38769972e-01 -4.67188597e-01 -2.25847274e-01 -1.07492678e-01
6.28010809e-01 -6.41840160e-01 -1.25184214e+00 3.61661278e-02
-5.66347241e-01 -3.78757000e-01 -3.77530336e-01 8.90379250e-01
-8.22718382e-01 4.52944636e-01 -3.88925940e-01 -1.51911116e+00
-3.62464696e-01 -1.20767474e+00 5.02632737e-01 3.25894713e-01
2.86119491e-01 -9.38306510e-01 1.32950559e-01 -1.80683192e-02
4.44366664e-01 8.22503939e-02 2.86168426e-01 -1.75270438e-01
-1.13117378e-02 -2.46997163e-01 -1.98323250e-01 4.08508033e-01
-2.61348765e-02 -3.75815809e-01 -3.15085322e-01 -7.80230641e-01
3.14609677e-01 2.23855232e-03 6.59891665e-01 5.95383883e-01
1.02281582e+00 -7.97921777e-01 -3.92761111e-01 5.31507313e-01
1.29180968e+00 5.86819053e-01 2.32053563e-01 3.81330043e-01
2.83037305e-01 2.20643327e-01 1.04315448e+00 9.63481605e-01
1.36702225e-01 4.39860106e-01 1.95892781e-01 -2.59064198e-01
5.92120528e-01 -3.21161658e-01 4.60474044e-01 5.95182776e-01
-2.06206515e-01 1.85347989e-01 -5.31078279e-01 6.02938473e-01
-2.10045886e+00 -8.42554510e-01 -4.29584160e-02 2.42911243e+00
8.53333890e-01 1.32717684e-01 2.95881808e-01 -6.57991618e-02
5.78684270e-01 7.34143555e-02 -7.19539046e-01 -6.62843585e-01
2.15756014e-01 2.06727669e-01 9.06929374e-01 8.82217944e-01
-9.85519230e-01 1.15447545e+00 6.11269999e+00 9.95809555e-01
-9.36736226e-01 1.80929508e-02 3.60876530e-01 -3.39738309e-01
-1.65996030e-01 3.39050263e-01 -8.33782971e-01 5.29753208e-01
6.44974411e-01 -4.06743854e-01 7.97486484e-01 1.03218055e+00
7.11836934e-01 -4.85094815e-01 -5.98755717e-01 9.61223304e-01
-2.32055902e-01 -9.49625731e-01 -3.97769630e-01 3.12288672e-01
1.09594655e+00 -4.40890670e-01 3.48773539e-01 2.89352596e-01
5.52146196e-01 -5.08986712e-01 5.39444268e-01 1.76151007e-01
7.82536268e-01 -9.09819424e-01 4.55636591e-01 4.15138721e-01
-1.19918096e+00 -4.98027831e-01 -4.33270961e-01 -9.02013332e-02
2.86226213e-01 7.08586812e-01 -4.15156215e-01 4.96157646e-01
6.38208926e-01 1.29901335e-01 2.65021205e-01 9.40534472e-01
-3.34331453e-01 3.58332396e-01 -6.70729876e-01 -3.80538583e-01
6.98447227e-01 -6.19962037e-01 8.48516047e-01 8.37568402e-01
3.17116678e-01 4.46702570e-01 5.04979491e-01 4.43677217e-01
3.57381225e-01 3.39022160e-01 -2.93771505e-01 -9.33195874e-02
3.30642134e-01 8.96659255e-01 -5.27148962e-01 -5.51273108e-01
-1.01310693e-01 1.16465151e+00 9.74289551e-02 6.43088877e-01
-9.05056417e-01 -7.85985649e-01 7.45673776e-01 -1.12932902e-02
5.41114271e-01 -5.09831369e-01 -5.44981696e-02 -1.27020764e+00
2.68585801e-01 -7.61149764e-01 1.47668257e-01 -5.94174974e-02
-5.99756777e-01 3.40529740e-01 2.29658902e-01 -1.25036025e+00
-3.27463657e-01 -5.19911468e-01 -1.54524937e-01 7.77962804e-01
-1.22900701e+00 -3.13506544e-01 2.29447082e-01 7.41511524e-01
2.39096358e-01 -2.48129949e-01 5.59914827e-01 2.51359791e-01
-6.99113250e-01 6.52624667e-01 9.08310592e-01 -3.54869127e-01
3.85881126e-01 -1.13804126e+00 -1.62108392e-01 8.43241036e-01
-3.81936133e-01 5.98306298e-01 9.83009636e-01 -5.94620049e-01
-1.36247563e+00 -8.38322222e-01 5.30934691e-01 2.42585480e-01
5.64466715e-01 -1.47292241e-01 -3.05717349e-01 4.74713266e-01
-1.36997923e-01 4.06154105e-03 2.20525473e-01 -1.61184415e-01
1.87019080e-01 -1.79802909e-01 -1.27648532e+00 9.35492635e-01
9.51206744e-01 -5.41545749e-01 -2.57929415e-01 5.67885637e-01
4.98405218e-01 -5.17120600e-01 -8.11462402e-01 3.41137201e-01
4.45318341e-01 -6.66158140e-01 6.27719104e-01 -7.24953115e-01
-4.01552558e-01 -4.82029349e-01 -1.46442905e-01 -1.38427377e+00
-2.84369022e-01 -1.39535427e+00 -4.79320794e-01 4.71095830e-01
5.92871666e-01 -1.02206492e+00 7.89371073e-01 5.55720925e-01
1.43720642e-01 -1.19277632e+00 -9.73950326e-01 -1.12055814e+00
1.84760422e-01 -2.52167284e-01 -9.35155302e-02 7.59578347e-01
3.86471301e-01 -1.25154659e-01 -6.42190993e-01 1.16820231e-01
6.01165533e-01 -9.00228396e-02 4.33715969e-01 -7.23004162e-01
-5.97201169e-01 -3.82079363e-01 6.48786426e-02 -1.49330711e+00
-1.36785373e-01 -2.96243966e-01 1.16462559e-01 -1.52354145e+00
-1.21606536e-01 -5.24204910e-01 -5.80397487e-01 3.96097183e-01
-3.07471365e-01 -4.16655779e-01 2.04544514e-01 -2.86569335e-02
-5.39230585e-01 8.33721638e-01 1.21976769e+00 1.43912375e-01
-7.49789536e-01 2.48691499e-01 -6.04218662e-01 7.57563293e-01
9.70908046e-01 -4.21031415e-01 -7.13423014e-01 -2.98542291e-01
-5.31038456e-02 4.26478565e-01 -1.97820991e-01 -8.27552259e-01
1.42079696e-01 -4.87293631e-01 1.87773257e-02 -5.78581333e-01
2.65772790e-01 -4.65198487e-01 -2.59948939e-01 8.24745953e-01
-1.89679906e-01 1.30214348e-01 3.41654748e-01 6.70507669e-01
1.80060729e-01 -3.01919878e-01 7.64267683e-01 -2.01076806e-01
-2.65833706e-01 1.79326251e-01 -7.38955438e-01 6.70116022e-02
1.16217792e+00 -7.41615966e-02 1.95303679e-01 -7.51369655e-01
-8.04910660e-01 3.88601840e-01 1.15588456e-01 -3.69219296e-03
6.26780748e-01 -1.20898390e+00 -4.17808324e-01 -1.75213143e-01
-6.11718297e-01 -3.44443589e-01 2.55322810e-02 1.04583108e+00
-4.19908136e-01 8.36459339e-01 1.13457471e-01 -2.35605910e-01
-9.59999681e-01 6.41685605e-01 1.73238039e-01 -5.43988168e-01
-1.50420934e-01 1.09315646e+00 -1.82241619e-01 -4.15527254e-01
3.02652985e-01 -2.72747695e-01 1.60113767e-01 -1.99238315e-01
4.30248290e-01 8.56450319e-01 -3.47365677e-01 -2.68713415e-01
-3.90566945e-01 3.97182882e-01 -2.36243799e-01 -8.99303854e-01
1.06081200e+00 -9.46309417e-02 -6.47023916e-02 9.25219730e-02
1.29094470e+00 7.24982694e-02 -1.62891364e+00 -1.07547335e-01
-1.99806783e-02 -5.81647038e-01 4.58078563e-01 -5.22184670e-01
-8.87125671e-01 6.52151227e-01 1.03655064e+00 -1.74757540e-01
9.85955596e-01 -6.96003258e-01 5.39388776e-01 4.91728693e-01
5.85718453e-01 -1.68473852e+00 4.47817966e-02 7.54124224e-01
7.27640927e-01 -9.19780731e-01 4.61755365e-01 -1.67352021e-01
-6.79967999e-01 6.58722878e-01 4.95532483e-01 -3.15493554e-01
7.92229533e-01 -1.39280930e-01 -2.29063213e-01 1.81755573e-01
-4.05336648e-01 -2.38278538e-01 2.78038144e-01 2.62410760e-01
2.79824764e-01 1.11360840e-01 -1.17277694e+00 1.47391006e-01
6.43662512e-02 -2.36695372e-02 1.52823776e-01 1.36035252e+00
-6.15619421e-01 -1.01298237e+00 -2.87405431e-01 4.16666090e-01
-4.40765291e-01 2.18806248e-02 -1.92361385e-01 4.59957182e-01
-4.56247240e-01 1.16280770e+00 -3.44296962e-01 -2.48539358e-01
3.30169320e-01 -8.20616260e-02 5.40267229e-01 -6.69746920e-02
-1.79349139e-01 3.81294012e-01 1.11792475e-01 -1.00613618e+00
2.11894572e-01 -6.00100338e-01 -1.34170127e+00 -3.06598455e-01
-3.53985935e-01 6.11852884e-01 7.97802567e-01 8.09454381e-01
5.45104563e-01 2.37884820e-01 9.56160963e-01 -8.37771118e-01
-1.23008645e+00 -7.83828676e-01 -4.80468512e-01 -6.12499788e-02
2.75722712e-01 -9.07329917e-01 -3.36873204e-01 -5.63207150e-01]
|
[4.2526679039001465, 2.6476738452911377]
|
fa1fbd52-b7f6-478f-ac4c-d0a72df52fb1
|
gpt-4-a-review-on-advancements-and
|
2305.03195
| null |
https://arxiv.org/abs/2305.03195v1
|
https://arxiv.org/pdf/2305.03195v1.pdf
|
Gpt-4: A Review on Advancements and Opportunities in Natural Language Processing
|
Generative Pre-trained Transformer 4 (GPT-4) is the fourth-generation language model in the GPT series, developed by OpenAI, which promises significant advancements in the field of natural language processing (NLP). In this research article, we have discussed the features of GPT-4, its potential applications, and the challenges that it might face. We have also compared GPT-4 with its predecessor, GPT-3. GPT-4 has a larger model size (more than one trillion), better multilingual capabilities, improved contextual understanding, and reasoning capabilities than GPT-3. Some of the potential applications of GPT-4 include chatbots, personal assistants, language translation, text summarization, and question-answering. However, GPT-4 poses several challenges and limitations such as computational requirements, data requirements, and ethical concerns.
|
['Mursal Dawodi', 'Jawid Ahmad Baktash']
|
2023-05-04
| null | null | null | null |
['text-summarization']
|
['natural-language-processing']
|
[ 1.28433496e-01 7.70101786e-01 -3.56385484e-02 -1.83078676e-01
-1.22616637e+00 -7.79991329e-01 7.36924648e-01 1.33957192e-01
-1.34957582e-01 1.00833726e+00 6.03918672e-01 -6.58427119e-01
1.00590862e-01 -5.24569392e-01 -4.59167361e-01 -3.30105424e-03
1.26316756e-01 9.20498252e-01 -5.75733483e-02 -5.65039992e-01
9.64282528e-02 -9.82374549e-02 -6.90082908e-01 5.94816029e-01
1.45495772e+00 4.57737893e-01 3.89638066e-01 5.72579026e-01
-5.82164466e-01 1.12122273e+00 -6.80917084e-01 -8.55051935e-01
-1.74510926e-02 -4.48271513e-01 -1.38098192e+00 -3.86435628e-01
3.22964549e-01 -1.37472674e-01 -1.62984021e-02 6.87669694e-01
5.69808900e-01 -1.69904847e-02 4.09317583e-01 -1.55283785e+00
-1.10539854e+00 1.09271276e+00 -1.11996271e-01 3.05095930e-02
8.69164169e-01 1.88043132e-01 1.13058841e+00 -8.83976638e-01
9.28400815e-01 1.65555096e+00 8.38931203e-01 7.75502264e-01
-8.98628473e-01 -5.20286143e-01 -3.73063050e-02 2.54268587e-01
-1.18082166e+00 -5.08680224e-01 2.53602028e-01 -1.91249013e-01
1.70361531e+00 2.69503146e-01 3.92057836e-01 1.27478480e+00
5.83774626e-01 1.17011678e+00 1.00642693e+00 -7.60439992e-01
-2.83099003e-02 1.91915080e-01 -1.89348578e-01 5.47789693e-01
-2.18020529e-01 -5.41512847e-01 -6.91470742e-01 -3.87775987e-01
4.84174907e-01 -6.22499228e-01 -1.41709864e-01 3.18213910e-01
-1.44534004e+00 8.48565638e-01 -5.99834770e-02 6.41434968e-01
-3.80428553e-01 -1.27235711e-01 5.28389573e-01 4.38770473e-01
7.75623977e-01 7.52452195e-01 -6.15445971e-01 -5.55657208e-01
-6.84521317e-01 2.77669936e-01 1.19428027e+00 1.48768544e+00
4.39242989e-01 3.63022313e-02 -5.29165328e-01 8.35434556e-01
2.07664743e-01 4.34986949e-01 5.24479926e-01 -1.14809108e+00
1.08469737e+00 5.53311050e-01 -1.42277673e-01 -6.06324375e-01
-1.80037007e-01 1.76349804e-02 -6.11780941e-01 -8.14094722e-01
1.34022862e-01 -4.76774305e-01 -5.24716020e-01 1.61366439e+00
-2.42551155e-02 -3.90029043e-01 6.31505549e-01 2.41385549e-01
1.01270044e+00 1.14180768e+00 1.42995000e-01 -2.51632273e-01
1.35426152e+00 -1.13476002e+00 -8.83603990e-01 -6.87710047e-01
8.54767263e-01 -1.07831478e+00 1.01878202e+00 7.72047369e-03
-1.17557251e+00 -2.26508930e-01 -5.09062588e-01 -4.40419436e-01
-3.68805230e-01 2.55445361e-01 6.37469590e-01 5.17453969e-01
-1.25144470e+00 3.05918247e-01 -5.70864856e-01 -1.08528316e+00
-5.85502274e-02 -5.83661571e-02 -3.20045531e-01 -2.28860289e-01
-1.40058172e+00 1.44317925e+00 2.62394190e-01 4.80370559e-02
-2.25653440e-01 -6.23056352e-01 -1.04403460e+00 1.13293625e-01
3.45731169e-01 -8.22354019e-01 1.62275732e+00 -2.14766011e-01
-1.90263140e+00 6.41869962e-01 -4.99869555e-01 -4.64167982e-01
2.07034528e-01 -3.90845507e-01 -2.84329236e-01 -9.97509360e-02
6.89571857e-01 6.63309276e-01 6.94997087e-02 -4.94806617e-01
-4.94777888e-01 -1.50796205e-01 -1.08865993e-02 4.02740568e-01
-1.05693135e-02 4.71150637e-01 -2.39328071e-01 -4.56896365e-01
-2.37098500e-01 -9.36419308e-01 -2.32397437e-01 -5.31140745e-01
-3.97397339e-01 -7.16350079e-01 6.47615016e-01 -9.95932102e-01
9.97449458e-01 -1.80846035e+00 9.69946682e-02 -4.25913066e-01
-1.59508958e-01 5.08433640e-01 -3.68575305e-01 1.20877326e+00
3.87264073e-01 4.81752068e-01 -1.12505592e-01 -1.69490382e-01
1.96300521e-01 4.61851627e-01 -4.78035212e-01 -3.15057933e-01
3.50942373e-01 1.38331425e+00 -1.08076560e+00 -6.76072717e-01
1.13329329e-01 2.03589767e-01 -4.22552854e-01 7.41580278e-02
-6.24675512e-01 4.42395627e-01 -5.96910357e-01 5.70265830e-01
1.85461417e-01 1.36611098e-02 4.74779636e-01 1.67279199e-01
-3.38216215e-01 8.72786343e-01 -3.25680822e-01 1.77806425e+00
-6.75122201e-01 8.24046612e-01 -4.11257371e-02 -5.52108943e-01
8.76800895e-01 8.04589987e-01 2.53751308e-01 -7.04850256e-01
2.16724481e-02 2.94101477e-01 -1.06636070e-01 -8.42489004e-01
7.84986436e-01 -5.43361492e-02 -3.48788887e-01 8.56665313e-01
4.06466484e-01 -4.62927818e-01 4.99211133e-01 4.58664566e-01
9.62849498e-01 3.33671242e-01 4.50131923e-01 -3.27187568e-01
2.60936499e-01 2.69427538e-01 6.68873489e-01 7.00325370e-01
-1.12626739e-01 1.81055710e-01 3.81313413e-01 1.62543058e-02
-9.68200326e-01 -6.31516099e-01 2.70130187e-01 9.72112596e-01
-3.84576380e-01 -8.43191922e-01 -8.63298893e-01 -4.56769168e-01
-4.26470786e-01 1.35177541e+00 -1.27732009e-01 1.11142375e-01
-8.86427164e-01 -3.21714401e-01 1.16729069e+00 4.27492142e-01
7.32742906e-01 -1.20581555e+00 -3.20634931e-01 4.91549492e-01
-1.12923253e+00 -1.54154038e+00 -5.79235494e-01 -1.96766913e-01
-7.76341975e-01 -7.72067666e-01 -4.66493696e-01 -1.02568710e+00
2.52186447e-01 1.32891268e-01 1.14708614e+00 -4.68062490e-01
2.24232718e-01 4.43734795e-01 -6.18825495e-01 -6.57616675e-01
-9.21334743e-01 4.78678077e-01 -1.73650935e-01 -5.00528932e-01
4.00623709e-01 -4.48337168e-01 1.54864565e-01 6.99949190e-02
-4.96750653e-01 3.59784126e-01 6.72693968e-01 7.13947296e-01
6.67412654e-02 -2.48628095e-01 7.62145519e-01 -9.82244730e-01
1.28394794e+00 -2.85852820e-01 1.80393420e-02 9.90077317e-01
-4.97407734e-01 1.35224789e-01 5.18389642e-01 -9.46874842e-02
-1.51039457e+00 -6.62361562e-01 -2.79978305e-01 1.12635367e-01
1.27398551e-01 8.69225025e-01 -2.85594910e-01 3.28952014e-01
6.07282221e-01 1.86966851e-01 -1.08418167e-01 -5.25382280e-01
7.11073995e-01 8.07084560e-01 4.72621709e-01 -8.62427890e-01
5.21156430e-01 -2.74653882e-01 -6.22722208e-01 -7.70793438e-01
-7.73184061e-01 -1.65264726e-01 -4.52697337e-01 -4.75458801e-02
7.26561368e-01 -7.95289755e-01 -3.57247800e-01 3.53535324e-01
-1.62230003e+00 -4.54937220e-01 -3.85092974e-01 2.93930829e-01
-3.06489706e-01 4.47662920e-01 -1.05852962e+00 -6.89941764e-01
-1.11041737e+00 -8.81315589e-01 1.10740912e+00 9.65794101e-02
-8.07046294e-01 -1.10247946e+00 6.80235401e-02 8.24704707e-01
5.62718272e-01 7.17923716e-02 1.28838038e+00 -7.11723089e-01
-2.39987805e-01 -4.60354798e-02 -8.35285634e-02 2.99891710e-01
2.77054608e-01 -6.32041842e-02 -5.59171259e-01 -1.24643900e-01
6.89818040e-02 -4.09109950e-01 5.72177116e-03 7.64750168e-02
2.74932623e-01 -9.47063088e-01 -2.96714962e-01 6.51161447e-02
8.65026534e-01 4.20726895e-01 5.31661272e-01 3.47987205e-01
5.45773029e-01 7.20870793e-01 6.65242016e-01 -6.39855787e-02
9.23948526e-01 3.73977274e-01 -1.29112020e-01 2.42771983e-01
-6.46060985e-03 -7.73240983e-01 6.86180472e-01 1.57089186e+00
2.05001738e-02 -4.45702761e-01 -9.99615371e-01 5.87379038e-01
-1.91121840e+00 -8.34211409e-01 -3.14118475e-01 1.42791951e+00
9.23053980e-01 -2.30009675e-01 -3.33065361e-01 -3.98452669e-01
7.49720037e-01 4.90405522e-02 -3.06769997e-01 -9.62273002e-01
-2.81257957e-01 2.69944489e-01 1.86886936e-02 5.23728549e-01
-3.50314140e-01 1.39292300e+00 7.37307787e+00 6.96488500e-01
-9.04730439e-01 2.86604583e-01 2.68274039e-01 3.14371884e-01
-5.11028588e-01 5.25150180e-01 -6.10103428e-01 3.39027867e-02
9.76312637e-01 -9.01253998e-01 4.87072110e-01 6.06420100e-01
1.63978726e-01 -7.96222128e-03 -1.17922592e+00 5.66610515e-01
4.00145441e-01 -1.31287265e+00 3.29765886e-01 -8.83439034e-02
5.12602448e-01 3.07453275e-01 -3.06632549e-01 8.91923904e-01
4.99238461e-01 -7.35544622e-01 7.86677659e-01 3.89613770e-02
5.63164651e-01 -4.25043106e-01 7.90272355e-01 6.39524400e-01
-1.00465846e+00 1.04855582e-01 -3.12464267e-01 -2.56924778e-01
5.73513687e-01 3.26374799e-01 -1.30649102e+00 1.03969920e+00
5.58797359e-01 6.56657517e-01 -3.90869915e-01 4.95546371e-01
-6.09551907e-01 7.52302527e-01 -2.66299397e-01 -1.38619989e-01
2.83453435e-01 -3.01436245e-01 6.12572789e-01 1.25418770e+00
5.45314372e-01 1.88634008e-01 1.74533799e-01 7.19840825e-01
-1.55190736e-01 2.78031856e-01 -5.84425628e-01 -3.14770401e-01
7.04487979e-01 9.83694136e-01 -2.07241520e-01 -5.15038669e-01
-3.05220783e-01 8.65646720e-01 3.43315244e-01 2.81695306e-01
-4.96841013e-01 -3.62604797e-01 2.77963519e-01 -1.51143402e-01
-1.34764733e-02 -1.99480623e-01 -8.84615034e-02 -1.36555815e+00
1.08455464e-01 -1.30783951e+00 4.24515158e-01 -1.25408351e+00
-1.27595758e+00 9.72549438e-01 1.50501043e-01 -8.95742834e-01
-6.72930777e-01 -2.62044251e-01 -6.47644520e-01 8.52508724e-01
-1.05289114e+00 -1.52877927e+00 3.64927799e-01 4.19520408e-01
8.97896945e-01 -1.03637464e-01 1.20712686e+00 1.41019747e-01
-4.59920228e-01 5.07203817e-01 -9.15113837e-03 1.41463295e-01
6.96509242e-01 -9.39076602e-01 1.00413454e+00 1.01805997e+00
-7.08554685e-02 7.67774999e-01 6.19922996e-01 -7.18663812e-01
-1.39696872e+00 -1.18792486e+00 1.99806774e+00 -5.73754668e-01
9.45583999e-01 -4.74939167e-01 -7.13965237e-01 1.35873973e+00
8.41547251e-01 -9.58688140e-01 7.56694794e-01 2.71639973e-01
-2.12996572e-01 1.10120140e-01 -1.20374608e+00 8.08182895e-01
9.07534003e-01 -8.01192224e-01 -1.07533085e+00 5.46862960e-01
9.38998997e-01 -6.56001627e-01 -1.12728047e+00 1.80027843e-01
3.96658778e-01 -4.48698372e-01 4.22358841e-01 -3.13634992e-01
2.32531324e-01 1.90129817e-01 -7.56061962e-03 -1.51308191e+00
-2.46175721e-01 -1.31252599e+00 3.11404109e-01 1.67875707e+00
7.44710445e-01 -9.80604410e-01 2.21843928e-01 9.52411652e-01
-5.67607880e-01 -5.45908809e-01 -1.03906620e+00 -7.31544495e-01
5.23097932e-01 -5.42298496e-01 7.51384079e-01 1.04803014e+00
7.21989512e-01 1.09027743e+00 -5.53201735e-01 -2.98683584e-01
2.33499065e-01 1.05153114e-01 7.64180362e-01 -9.67354000e-01
-8.66243541e-02 -1.35878354e-01 1.25575647e-01 -1.16141760e+00
2.46292025e-01 -9.85508382e-01 1.96824744e-02 -2.29914665e+00
-3.06477062e-02 -2.36420155e-01 5.40223897e-01 9.83313203e-01
-6.35029823e-02 -2.74848849e-01 5.21137416e-01 2.61266828e-01
-3.44885707e-01 6.32164896e-01 1.35128224e+00 -2.87435770e-01
-2.54229754e-01 -2.71782905e-01 -1.03929937e+00 3.79937500e-01
1.00314796e+00 -4.07049924e-01 -3.48323286e-01 -1.01958537e+00
2.52607018e-01 3.65976542e-01 -1.26727477e-01 -4.83312041e-01
4.38157380e-01 -2.46773884e-01 -3.10114086e-01 -5.35658062e-01
2.37884939e-01 -1.81553081e-01 7.62249306e-02 7.77717084e-02
-3.67447823e-01 3.73515427e-01 3.84588242e-01 -1.84551120e-01
-2.78986126e-01 -2.35709757e-01 -5.55561529e-03 -4.40249056e-01
-3.72190922e-01 -1.05318837e-01 -1.20709920e+00 3.31970125e-01
6.78014994e-01 4.04327288e-02 -6.57644331e-01 -5.47896862e-01
-2.19498381e-01 6.56526089e-01 2.38362234e-02 8.49832296e-01
4.63239670e-01 -1.06787968e+00 -1.00556898e+00 -1.62367344e-01
1.64391056e-01 2.89439373e-02 4.35806736e-02 6.21787846e-01
-4.39466774e-01 9.76173043e-01 -1.42984986e-01 -2.76228458e-01
-1.09957707e+00 1.27035931e-01 7.47034773e-02 -6.87121868e-01
-7.14909256e-01 5.01846969e-01 -1.66588426e-01 -9.84853387e-01
-2.13225991e-01 -4.19063032e-01 -1.29646972e-01 -3.04517508e-01
2.55268067e-01 3.06924641e-01 3.62212770e-02 -6.79426610e-01
-3.22347671e-01 1.48288131e-01 -2.22302109e-01 -4.73236501e-01
1.22787845e+00 -2.51736939e-01 -6.27419174e-01 4.50129747e-01
6.73878193e-01 -1.85275167e-01 -3.26681107e-01 -5.59872389e-01
1.71250537e-01 -2.19249469e-03 -3.08247030e-01 -9.86123621e-01
-4.39220101e-01 7.41269827e-01 -2.40288571e-01 1.04413651e-01
7.19219565e-01 9.69364941e-02 1.36846828e+00 6.79452837e-01
6.70177639e-01 -1.13750088e+00 -2.16204256e-01 1.28128362e+00
1.09479976e+00 -7.63171971e-01 -2.07628205e-01 -3.17609668e-01
-8.63293052e-01 8.24130654e-01 6.48717880e-01 5.38477182e-01
1.69237345e-01 8.88919234e-02 4.09957975e-01 -1.19437695e-01
-1.14451420e+00 9.73941162e-02 -1.25357911e-01 7.21172988e-01
6.81284070e-01 8.06734338e-02 -4.37981904e-01 4.11391914e-01
-8.08279812e-01 4.19138512e-03 4.37592715e-01 1.26531172e+00
5.73354401e-02 -1.47241247e+00 -2.49785647e-01 2.35990331e-01
-4.21594113e-01 -5.17774343e-01 -7.76922286e-01 7.15665400e-01
-1.27629146e-01 1.45317626e+00 -3.45076352e-01 -3.07398796e-01
4.37761307e-01 3.64755392e-01 5.42298615e-01 -8.97749007e-01
-9.45508897e-01 -1.43872827e-01 7.55311787e-01 -2.90981680e-01
-4.39867049e-01 -6.76472068e-01 -9.53111172e-01 -5.12743056e-01
-3.62859637e-01 8.40414524e-01 6.36310518e-01 1.20899510e+00
6.78620934e-01 2.17776552e-01 1.88979894e-01 -4.31911290e-01
-2.98450053e-01 -1.45055676e+00 -3.51363719e-02 -2.39533529e-01
-2.55714595e-01 1.64213441e-02 -9.94721707e-03 7.56255537e-02]
|
[11.441564559936523, 9.202988624572754]
|
71762e1b-ffe1-4815-9490-f02dd49c37e1
|
finrl-podracer-high-performance-and-scalable
|
2111.05188
| null |
https://arxiv.org/abs/2111.05188v1
|
https://arxiv.org/pdf/2111.05188v1.pdf
|
FinRL-Podracer: High Performance and Scalable Deep Reinforcement Learning for Quantitative Finance
|
Machine learning techniques are playing more and more important roles in finance market investment. However, finance quantitative modeling with conventional supervised learning approaches has a number of limitations. The development of deep reinforcement learning techniques is partially addressing these issues. Unfortunately, the steep learning curve and the difficulty in quick modeling and agile development are impeding finance researchers from using deep reinforcement learning in quantitative trading. In this paper, we propose an RLOps in finance paradigm and present a FinRL-Podracer framework to accelerate the development pipeline of deep reinforcement learning (DRL)-driven trading strategy and to improve both trading performance and training efficiency. FinRL-Podracer is a cloud solution that features high performance and high scalability and promises continuous training, continuous integration, and continuous delivery of DRL-driven trading strategies, facilitating a rapid transformation from algorithmic innovations into a profitable trading strategy. First, we propose a generational evolution mechanism with an ensemble strategy to improve the trading performance of a DRL agent, and schedule the training of a DRL algorithm onto a GPU cloud via multi-level mapping. Then, we carry out the training of DRL components with high-performance optimizations on GPUs. Finally, we evaluate the FinRL-Podracer framework for a stock trend prediction task on an NVIDIA DGX SuperPOD cloud. FinRL-Podracer outperforms three popular DRL libraries Ray RLlib, Stable Baseline 3 and FinRL, i.e., 12% \sim 35% improvements in annual return, 0.1 \sim 0.6 improvements in Sharpe ratio and 3 times \sim 7 times speed-up in training time. We show the high scalability by training a trading agent in 10 minutes with $80$ A100 GPUs, on NASDAQ-100 constituent stocks with minute-level data over 10 years.
|
['Jian Guo', 'Anwar Walid', 'Zhaoran Wang', 'Jiahao Zheng', 'Xiao-Yang Liu', 'Zechu Li']
|
2021-11-07
| null | null | null | null |
['stock-trend-prediction']
|
['time-series']
|
[-1.23891497e+00 -6.15561664e-01 -1.81147441e-01 -2.46049643e-01
-6.67645037e-01 -7.64214456e-01 7.07607150e-01 8.19872692e-02
-4.09494996e-01 6.72306120e-01 -2.93294638e-01 -6.57257617e-01
-1.41303698e-02 -1.18282962e+00 -7.05265760e-01 -4.77066904e-01
-2.84177780e-01 8.94603610e-01 1.86278760e-01 -3.87587011e-01
4.51410413e-01 5.97730458e-01 -1.20002627e+00 5.77663518e-02
7.60578871e-01 1.47773027e+00 -2.32832804e-01 6.91252112e-01
-2.46797785e-01 1.45441413e+00 -6.10579431e-01 -6.06632471e-01
6.95527911e-01 -1.22693367e-01 -3.29681993e-01 -3.05615425e-01
1.83256925e-03 -6.39496505e-01 -1.01517551e-01 6.85692251e-01
4.14590299e-01 -1.42168924e-01 3.56893271e-01 -1.31757879e+00
-3.33800942e-01 6.47390246e-01 -1.13384831e+00 5.68691671e-01
-4.51779723e-01 4.13957655e-01 1.22117364e+00 -6.76526189e-01
2.04622552e-01 9.95992303e-01 7.11840749e-01 2.33104914e-01
-9.38555717e-01 -9.77727473e-01 -9.50837433e-02 -2.98924595e-01
-5.47489285e-01 1.12297162e-01 4.29901212e-01 -6.25989735e-01
1.44054794e+00 7.54507929e-02 1.08849549e+00 6.51338875e-01
6.73893452e-01 7.76862323e-01 1.42846286e+00 4.22578976e-02
5.81401765e-01 1.49313286e-01 -4.39328179e-02 6.65701330e-01
2.88667560e-01 7.84039736e-01 -6.76899314e-01 -4.02365357e-01
1.16990268e+00 1.66468754e-01 5.92804372e-01 6.93614259e-02
-7.30447233e-01 1.37640512e+00 3.80889058e-01 3.19231786e-02
-5.27572811e-01 5.94221234e-01 5.74779630e-01 6.42052054e-01
9.00199831e-01 8.42197299e-01 -9.23036456e-01 -7.94958591e-01
-1.22182560e+00 7.00630307e-01 8.67278636e-01 5.82804620e-01
6.09019578e-01 7.97479749e-01 6.68500662e-02 4.95592862e-01
1.39233291e-01 6.63795769e-01 7.33532190e-01 -1.23755896e+00
2.77359039e-01 6.80491984e-01 8.32504928e-02 -6.29465938e-01
-4.76414621e-01 -8.87756407e-01 -6.72556996e-01 9.26582694e-01
2.47056648e-01 -5.26715875e-01 -2.49947950e-01 1.14919817e+00
3.18162322e-01 3.13812554e-01 1.37381420e-01 5.57727993e-01
9.35493037e-02 7.97357500e-01 -2.18835488e-01 -7.60536119e-02
1.29865992e+00 -1.33991599e+00 -4.83616479e-02 4.02383469e-02
8.25168669e-01 -6.87890232e-01 9.95008886e-01 6.79860353e-01
-1.21389067e+00 -5.69645703e-01 -9.85471547e-01 5.26960373e-01
-2.61092842e-01 -1.07894972e-01 1.28632784e+00 7.07945645e-01
-8.68982732e-01 1.01701593e+00 -1.15192497e+00 6.32372558e-01
4.54041511e-01 4.51002747e-01 2.97190636e-01 5.39750040e-01
-8.86128664e-01 4.75050807e-01 1.86769888e-01 -2.79839873e-01
-8.46042037e-01 -1.36805427e+00 -2.45575368e-01 1.45951837e-01
2.43379086e-01 -5.50225496e-01 1.49578929e+00 -1.04371023e+00
-1.82177234e+00 6.22246206e-01 7.78079808e-01 -1.22583365e+00
9.95855629e-01 -4.11800891e-01 -3.62266421e-01 -2.58539587e-01
-2.77429167e-02 3.05866510e-01 7.24270403e-01 -4.13941175e-01
-1.15076745e+00 -3.42695355e-01 -9.28437710e-02 -1.79237761e-02
-2.73844659e-01 1.72507480e-01 1.80760995e-02 -1.04333293e+00
-6.59448266e-01 -8.08741748e-01 -2.46444821e-01 -5.75293303e-01
5.03721595e-01 -2.76847005e-01 5.75561047e-01 -5.00292838e-01
1.04083204e+00 -1.88749075e+00 -4.94308889e-01 1.38054609e-01
2.68092483e-01 2.99747229e-01 7.92579725e-02 2.03622848e-01
1.92833930e-01 -4.79966551e-02 5.11935018e-02 -1.64295629e-01
1.66100144e-01 7.79787973e-02 -8.53661060e-01 1.40119627e-01
2.81497449e-01 1.25097346e+00 -7.70921350e-01 -2.20371094e-02
-1.64867014e-01 -6.46825656e-02 -9.64331210e-01 4.78438079e-01
-4.72082257e-01 2.61412263e-01 -4.73900139e-01 8.15649569e-01
7.98628926e-01 -3.00792575e-01 -5.09932600e-02 5.07745028e-01
-5.44719458e-01 8.83740187e-02 -1.00407636e+00 1.12065732e+00
-7.30342090e-01 3.08707207e-01 -3.23483765e-01 -8.00870180e-01
1.37597823e+00 -1.20000832e-01 7.28058219e-01 -1.28358388e+00
-3.16544548e-02 7.77735531e-01 4.86594960e-02 2.29470164e-01
5.03539979e-01 -1.21798962e-01 -2.09101498e-01 1.09836805e+00
-6.10276870e-02 -2.82643259e-01 3.11674118e-01 -6.10247850e-02
1.13806832e+00 3.81315172e-01 -1.10096969e-01 -4.65880066e-01
8.42534378e-02 3.35452616e-01 6.75543725e-01 6.98143959e-01
-1.91029962e-02 -8.09498653e-02 8.20951223e-01 -1.05526900e+00
-1.21080470e+00 -9.86247599e-01 4.04123142e-02 1.36620009e+00
-5.77075779e-01 -2.40044758e-01 -5.82498074e-01 -6.27247393e-01
6.11963153e-01 6.81048572e-01 -5.95277786e-01 4.15680185e-02
-8.48233461e-01 -1.31996441e+00 5.61544180e-01 6.66056633e-01
8.60508442e-01 -1.27438855e+00 -1.08726370e+00 6.86123312e-01
8.62573683e-01 -6.85000956e-01 -3.00516069e-01 2.90893376e-01
-1.13781440e+00 -7.34051287e-01 -4.94648784e-01 -3.68677258e-01
1.09647298e-02 -1.96640387e-01 1.65649486e+00 3.84371867e-03
-2.78988600e-01 -6.92291483e-02 -1.93298254e-02 -7.34286606e-01
-3.25455457e-01 1.74314812e-01 9.77053344e-02 -2.69714326e-01
2.17246309e-01 -5.10680437e-01 -7.39730418e-01 1.04713902e-01
-5.41795611e-01 -3.04876640e-02 4.21280742e-01 1.12330747e+00
5.27970374e-01 1.33595258e-01 8.11590552e-01 -9.40446377e-01
7.59880781e-01 -4.88038212e-01 -1.63264787e+00 3.42378840e-02
-1.27029717e+00 1.85072511e-01 8.51204038e-01 -2.96461761e-01
-9.80980456e-01 -3.76316220e-01 -1.20679535e-01 -4.78024781e-01
6.51115119e-01 5.90149939e-01 6.74902201e-01 1.67045847e-01
5.04416764e-01 2.53843397e-01 1.51143089e-01 -3.47463161e-01
2.99664978e-02 2.47205615e-01 2.12804079e-01 -8.78036976e-01
7.40659118e-01 2.54120201e-01 -1.82257473e-01 -9.70118567e-02
-6.98818862e-01 5.75563731e-03 -2.07417570e-02 -2.64859721e-02
4.18746978e-01 -1.18272257e+00 -1.00749516e+00 8.52260768e-01
-5.44498324e-01 -9.90341842e-01 -6.29777253e-01 4.22897637e-01
-5.70319116e-01 -2.98938006e-01 -1.06374872e+00 -7.14846492e-01
-9.22311306e-01 -1.08314419e+00 6.89814985e-01 4.08290923e-01
1.33029222e-01 -1.19029796e+00 6.39309585e-01 1.56177163e-01
7.18438983e-01 2.64304996e-01 6.69855416e-01 -7.07694709e-01
-5.81089139e-01 -5.96386790e-02 -2.43136913e-01 5.11713803e-01
-1.41655073e-01 1.06600963e-01 -6.26859069e-01 -4.77971435e-01
2.00862244e-01 -6.71466053e-01 6.08208239e-01 3.33663166e-01
1.01891458e+00 -3.60255927e-01 3.78796220e-01 9.85290170e-01
1.51496828e+00 4.89153057e-01 2.36138359e-01 9.47903335e-01
3.80120188e-01 1.78478897e-01 1.07999873e+00 1.02822435e+00
1.96667269e-01 3.51966649e-01 3.48373055e-01 -2.76443124e-01
4.22308266e-01 -1.65649131e-01 6.38794959e-01 9.77393746e-01
-2.86156893e-01 3.46623152e-01 -1.07164979e+00 -6.64353371e-02
-1.80956638e+00 -7.97686696e-01 2.70691589e-02 1.92227030e+00
1.03581786e+00 6.77367568e-01 5.12143791e-01 -4.40672219e-01
-5.11564203e-02 -1.00243022e-03 -1.05293608e+00 -8.81142259e-01
-5.99844307e-02 7.11910367e-01 7.45701075e-01 2.24457771e-01
-8.27460945e-01 1.25720334e+00 5.45502186e+00 1.03025508e+00
-1.44428146e+00 1.07382305e-01 1.28914797e+00 -2.98703253e-01
-2.84298688e-01 -1.56786263e-01 -1.04868269e+00 5.71772337e-01
1.33822834e+00 -5.20564735e-01 6.67828560e-01 1.31574380e+00
1.36950836e-01 6.79913461e-02 -7.66824007e-01 1.00474918e+00
-6.53662145e-01 -2.07018042e+00 -1.60428405e-01 4.03512478e-01
1.03179860e+00 4.89474893e-01 3.66357952e-01 8.59619498e-01
9.50019956e-01 -9.76807892e-01 9.10262465e-01 3.79866272e-01
4.47737396e-01 -1.47353911e+00 7.76537538e-01 3.35051447e-01
-1.07807076e+00 -2.07286879e-01 -5.87892056e-01 -2.59667695e-01
-3.79135638e-01 5.18967032e-01 -5.13473928e-01 3.57063651e-01
1.01485169e+00 7.44576871e-01 -5.81679225e-01 5.31530321e-01
1.09827928e-01 8.81396830e-01 -1.47240832e-01 -1.95238516e-01
7.93868780e-01 -6.83960021e-01 -1.00582823e-01 1.06412685e+00
3.75495464e-01 -7.50478730e-02 3.88795920e-02 9.04332995e-01
-2.00820938e-01 1.67935222e-01 -1.99039206e-01 -1.78424031e-01
2.15141714e-01 1.33254445e+00 -6.70750737e-01 -5.06263137e-01
-5.13655126e-01 6.03697836e-01 2.88150311e-01 -2.18940198e-01
-1.23016667e+00 -1.44470662e-01 9.29461837e-01 1.07935622e-01
2.84494191e-01 -2.67736554e-01 -5.63241005e-01 -1.10718834e+00
-1.96955249e-01 -1.28792548e+00 3.80271792e-01 -3.37248713e-01
-1.34003031e+00 6.70368016e-01 -5.38165629e-01 -1.18589818e+00
-6.41513705e-01 -7.59242058e-01 -7.98910081e-01 7.52860427e-01
-1.81909144e+00 -7.63685167e-01 7.75685012e-02 4.88726318e-01
5.65460205e-01 -1.09078515e+00 6.03575110e-01 6.11672699e-02
-5.07424593e-01 5.37921131e-01 6.32866204e-01 1.14664592e-01
3.67680639e-01 -1.60406101e+00 9.71896291e-01 4.45965290e-01
1.72909841e-01 2.88713276e-01 3.72336715e-01 -7.11080432e-01
-1.53027868e+00 -1.19706655e+00 3.68371308e-01 -1.93724409e-01
1.42956293e+00 -2.64765799e-01 -8.18723798e-01 5.37009776e-01
4.30268168e-01 2.65635122e-02 5.42842746e-01 -1.46500960e-01
-3.48035067e-01 -5.85055947e-01 -9.56515014e-01 4.24824893e-01
3.83989334e-01 -1.40525565e-01 -2.45952159e-01 2.68843800e-01
6.60861850e-01 -5.17631471e-01 -1.16588938e+00 7.78848976e-02
5.00329316e-01 -1.33348918e+00 7.55630732e-01 -5.58812380e-01
5.91736615e-01 1.88590303e-01 1.23378761e-01 -1.24747229e+00
-9.65292677e-02 -9.90982652e-01 -3.48369122e-01 1.07244360e+00
2.59633183e-01 -1.00649929e+00 1.08060229e+00 3.00411493e-01
4.38698195e-03 -1.11345541e+00 -7.73948491e-01 -1.04333997e+00
8.02003086e-01 -4.08607244e-01 9.72030044e-01 8.94683421e-01
-3.52446854e-01 4.22835983e-02 -4.05740231e-01 -3.46996129e-01
7.04892218e-01 6.32726550e-01 9.67677116e-01 -1.13055670e+00
-1.01065528e+00 -7.90601730e-01 1.30399778e-01 -5.74837506e-01
6.17211871e-02 -7.45453477e-01 -6.79471374e-01 -6.65618002e-01
4.21535037e-02 -7.67981887e-01 -4.77135867e-01 4.46572453e-01
2.99242884e-01 4.49697562e-02 2.12165505e-01 4.59535182e-01
-2.54113019e-01 5.92424214e-01 1.01820314e+00 -5.51562011e-02
-2.89064258e-01 2.38739133e-01 -4.89004612e-01 6.00056350e-01
1.03417361e+00 -5.23001909e-01 -6.32192492e-02 -2.76764452e-01
7.24735737e-01 2.83574730e-01 1.35360911e-01 -9.02205527e-01
-3.12556699e-02 -3.03663760e-01 4.03823942e-01 -6.32954836e-01
-8.95383507e-02 -3.44032824e-01 8.41831863e-02 1.12576127e+00
1.22938752e-01 9.22891974e-01 3.33430409e-01 6.04343824e-02
-4.06232953e-01 -3.90032195e-02 8.92320693e-01 -3.14917147e-01
-5.86876631e-01 6.14483535e-01 -2.47668535e-01 3.16496432e-01
1.03298473e+00 1.55964911e-01 -2.94470876e-01 4.49495316e-02
-3.32506716e-01 4.56760228e-01 3.32484305e-01 3.42895627e-01
3.81558776e-01 -1.30637586e+00 -9.62010920e-01 3.66816252e-01
-3.73908669e-01 -2.71945268e-01 -1.46079743e-02 5.79550922e-01
-1.23279977e+00 2.57633656e-01 -5.67045033e-01 -3.78777266e-01
-7.19905376e-01 3.30142140e-01 6.59170687e-01 -1.17323375e+00
-5.53624392e-01 8.33203137e-01 8.07293877e-02 -4.56346571e-01
-1.35616839e-01 -3.30654830e-01 2.20624566e-01 2.91277438e-01
6.28570676e-01 6.21738195e-01 2.49439731e-01 1.13025986e-01
-1.04072310e-01 3.51657420e-01 -1.95071921e-01 -1.91694722e-01
2.01828241e+00 5.43803751e-01 -9.07119960e-02 4.15507287e-01
8.68216515e-01 1.39668941e-01 -1.80525386e+00 -1.53297009e-02
3.99066687e-01 -2.61120349e-01 1.50852054e-01 -8.60556841e-01
-1.58492863e+00 8.32513452e-01 7.27932811e-01 2.49651849e-01
9.38310504e-01 -4.29313868e-01 1.01509726e+00 1.43554196e-01
6.02591097e-01 -1.42231774e+00 4.95438725e-01 6.72862232e-01
7.46988893e-01 -1.15820980e+00 7.37212747e-02 5.67421496e-01
-9.53539193e-01 1.40157640e+00 7.33338535e-01 -7.67829716e-01
7.83686817e-01 8.17728758e-01 1.58613756e-01 -1.74474895e-01
-1.25055361e+00 3.05525303e-01 -2.49279469e-01 -4.42236885e-02
2.40333885e-01 1.75700024e-01 -4.39549461e-02 7.33281314e-01
-6.55594230e-01 5.23283370e-02 3.84732932e-01 9.70295846e-01
-3.36139798e-01 -1.27998710e+00 -1.22177698e-01 4.87398952e-01
-8.07197213e-01 -1.87471539e-01 1.59074634e-01 1.04567289e+00
-1.35964245e-01 3.03154290e-01 6.39492214e-01 -2.29247957e-01
1.74196109e-01 -3.69613647e-01 2.51365632e-01 -3.23683500e-01
-1.47962844e+00 4.27268267e-01 -2.91399747e-01 -4.79182392e-01
2.39770696e-01 -1.03904617e+00 -1.35038447e+00 -9.67608571e-01
1.86359853e-01 3.47560585e-01 4.91021395e-01 5.44869602e-01
3.03141505e-01 5.11858940e-01 1.24821532e+00 -4.77186739e-01
-1.13490939e+00 -6.21186137e-01 -1.07063830e+00 -1.57509565e-01
-8.99738222e-02 -4.71003801e-01 -2.21422315e-01 -4.80627984e-01]
|
[4.441044330596924, 4.012338638305664]
|
4ae76b34-acf6-47f0-968c-79b26e5a43f9
|
person-image-generation-with-semantic
|
2008.07884
| null |
https://arxiv.org/abs/2008.07884v1
|
https://arxiv.org/pdf/2008.07884v1.pdf
|
Person image generation with semantic attention network for person re-identification
|
Pose variation is one of the key factors which prevents the network from learning a robust person re-identification (Re-ID) model. To address this issue, we propose a novel person pose-guided image generation method, which is called the semantic attention network. The network consists of several semantic attention blocks, where each block attends to preserve and update the pose code and the clothing textures. The introduction of the binary segmentation mask and the semantic parsing is important for seamlessly stitching foreground and background in the pose-guided image generation. Compared with other methods, our network can characterize better body shape and keep clothing attributes, simultaneously. Our synthesized image can obtain better appearance and shape consistency related to the original image. Experimental results show that our approach is competitive with respect to both quantitative and qualitative results on Market-1501 and DeepFashion. Furthermore, we conduct extensive evaluations by using person re-identification (Re-ID) systems trained with the pose-transferred person based augmented data. The experiment shows that our approach can significantly enhance the person Re-ID accuracy.
|
['Meichen Liu', 'Shuzhi Sam Ge', 'Kejun Wang', 'Juihang Ji']
|
2020-08-18
| null | null | null | null |
['pose-guided-image-generation']
|
['computer-vision']
|
[ 1.45540148e-01 -1.44170105e-01 1.93399444e-01 -5.78614354e-01
-3.02500963e-01 -3.08956057e-01 5.01406848e-01 -4.95503724e-01
-4.60414201e-01 5.50300539e-01 3.54465127e-01 5.58242679e-01
2.45309219e-01 -6.70364499e-01 -7.49053180e-01 -6.29371464e-01
5.29809296e-01 5.57661653e-01 2.11362559e-02 -1.69881091e-01
-1.24835767e-01 2.07279965e-01 -1.49997973e+00 3.52600100e-03
9.37751889e-01 8.36908996e-01 -5.86204289e-04 4.52588648e-01
1.09303631e-01 1.52223930e-01 -6.53836071e-01 -1.02906084e+00
4.01581436e-01 -3.95144492e-01 -6.80333376e-01 5.57895064e-01
9.56864119e-01 -6.36043727e-01 -2.73656160e-01 1.40369821e+00
7.75709808e-01 2.43131682e-01 3.58182222e-01 -1.31256509e+00
-1.06144977e+00 3.80788088e-01 -8.81665051e-01 -2.29215473e-02
5.20283103e-01 8.62456262e-02 3.74946564e-01 -8.37430179e-01
3.57917905e-01 1.67905819e+00 1.02038586e+00 9.40656364e-01
-1.15336740e+00 -7.70213306e-01 4.24556673e-01 1.41704023e-01
-1.57033026e+00 -3.57610345e-01 9.23674643e-01 -2.22774982e-01
2.63924956e-01 2.70049334e-01 8.65725756e-01 1.38782132e+00
-2.61876434e-01 8.35532308e-01 1.05489564e+00 -4.39320564e-01
-3.63144785e-01 1.89702973e-01 1.76365122e-01 7.86185265e-01
5.71114182e-01 1.67976514e-01 -3.32982421e-01 1.15928248e-01
9.84245420e-01 1.36123851e-01 -2.74414092e-01 -3.50530952e-01
-1.22373557e+00 4.28242475e-01 4.67051357e-01 -8.81639197e-02
-5.86536303e-02 3.16005290e-01 3.55409682e-01 -1.83985949e-01
4.65501785e-01 2.87273079e-02 1.24937678e-02 4.06314701e-01
-6.87672257e-01 4.35768098e-01 3.72492433e-01 1.10717320e+00
3.43897730e-01 1.08053520e-01 -5.75315654e-01 1.12008679e+00
4.82572287e-01 1.05765069e+00 4.96474594e-01 -7.20642209e-01
3.74245554e-01 5.97556472e-01 3.53403538e-01 -1.30612016e+00
-3.45862210e-01 -5.53795576e-01 -1.05314064e+00 -2.02278957e-01
4.58228260e-01 -1.64356008e-01 -1.19095969e+00 1.90436637e+00
4.10076171e-01 1.68430552e-01 -1.69444606e-01 1.23384607e+00
1.14526248e+00 3.12814593e-01 3.40371042e-01 1.98057815e-01
1.60053360e+00 -1.24302006e+00 -7.87142158e-01 -4.45083380e-01
-9.62413773e-02 -6.41575575e-01 8.37367773e-01 5.43758161e-02
-1.06169558e+00 -1.25340307e+00 -9.98017490e-01 3.06978379e-03
-1.81996539e-01 6.09063506e-01 3.71981651e-01 1.11563277e+00
-9.47444499e-01 3.14552516e-01 -5.68797052e-01 -6.90937817e-01
3.35528284e-01 4.79150623e-01 -4.70390648e-01 -7.92520046e-02
-1.16437483e+00 5.62108457e-01 2.82259852e-01 3.42978239e-01
-7.17239320e-01 -4.17679459e-01 -9.84236658e-01 -2.35756278e-01
1.61425754e-01 -9.95855808e-01 9.47988510e-01 -1.35183537e+00
-1.49427605e+00 1.28713810e+00 -1.87413797e-01 -2.17991158e-01
7.95947373e-01 -3.42580140e-01 -5.98221302e-01 9.82989836e-03
3.53737861e-01 9.45161104e-01 9.88781393e-01 -1.62858760e+00
-6.86158538e-01 -6.31957293e-01 -1.40870109e-01 3.54234815e-01
-4.39222604e-01 1.82900980e-01 -1.11246586e+00 -1.17082560e+00
4.03499529e-02 -1.13311470e+00 -1.36924952e-01 -6.34448603e-02
-6.86652124e-01 -3.12664993e-02 4.99715567e-01 -1.29854405e+00
8.23335290e-01 -1.85329866e+00 2.73304135e-01 3.04398119e-01
7.82473087e-02 2.21073732e-01 -8.87859687e-02 -2.41332993e-01
-1.21839046e-01 7.02831373e-02 -2.82770097e-02 -8.75551403e-01
3.77056673e-02 -7.61029720e-02 1.38428081e-02 5.46913028e-01
-1.88165694e-01 1.16265929e+00 -5.52063227e-01 -6.05739772e-01
3.52917790e-01 5.68026960e-01 -3.75658691e-01 2.97620058e-01
1.64128333e-01 7.19711661e-01 -3.39822978e-01 7.06991255e-01
9.19422209e-01 -1.27849609e-01 5.79278097e-02 -8.23893070e-01
3.58753234e-01 -5.30881464e-01 -1.24268568e+00 1.68831742e+00
7.79955983e-02 1.46045208e-01 1.16899749e-02 -5.99846601e-01
9.57762420e-01 9.84416306e-02 4.28382009e-01 -8.60187232e-01
3.66696656e-01 -1.51123628e-01 -3.45916539e-01 -2.87169307e-01
5.26769698e-01 2.36543685e-01 -8.37317780e-02 3.20714831e-01
-7.10310191e-02 5.63166261e-01 -1.32611487e-02 -8.26736912e-02
-3.06236353e-02 2.60695010e-01 -2.09638685e-01 -4.06833977e-01
7.94603527e-01 -2.86340326e-01 7.17128038e-01 6.62919104e-01
-4.92052972e-01 8.49097192e-01 -2.85497874e-01 -6.78183794e-01
-1.23226917e+00 -9.98573184e-01 6.96992427e-02 1.09273946e+00
8.78094077e-01 -1.54179692e-01 -1.25969267e+00 -6.89087033e-01
-3.45124267e-02 2.44201824e-01 -7.63890088e-01 -2.15434760e-01
-6.81579828e-01 -1.05175388e+00 6.14665449e-01 8.59812677e-01
1.30689478e+00 -7.99905539e-01 1.77727371e-01 -1.17978612e-02
-7.94400394e-01 -1.14195788e+00 -1.12592709e+00 -8.02534580e-01
-4.68240768e-01 -9.43980157e-01 -1.37711298e+00 -1.07053649e+00
1.09243047e+00 3.89258802e-01 7.83526361e-01 1.78580284e-01
-2.34268785e-01 6.28757060e-01 -2.51058936e-01 -2.38110244e-01
-1.19806603e-01 -3.24878506e-02 3.58044565e-01 6.03484869e-01
3.28717679e-01 -1.49160698e-01 -9.23848927e-01 6.48049116e-01
-4.43817645e-01 3.42975467e-01 2.58186549e-01 7.54485309e-01
6.78763449e-01 3.21802907e-02 3.63003254e-01 -5.57420671e-01
4.98419702e-01 2.66575158e-01 -1.87455758e-01 6.10003889e-01
-5.53502679e-01 -3.05681735e-01 2.40227237e-01 -5.29863238e-01
-1.29377544e+00 2.58105546e-01 -1.98771238e-01 -2.64397293e-01
-2.46635929e-01 -3.58583719e-01 -7.04390705e-01 -3.89434844e-01
2.85022467e-01 3.59033108e-01 6.96026608e-02 -6.77623272e-01
3.34110379e-01 6.18842244e-01 1.00569248e+00 -7.61711597e-01
9.37179565e-01 6.94126248e-01 -5.40572882e-01 -4.04516250e-01
-7.87210226e-01 -2.81925231e-01 -7.16969728e-01 -5.24006486e-01
1.23716176e+00 -1.20439470e+00 -7.43274748e-01 1.08738422e+00
-1.14898169e+00 -2.92935446e-02 -2.86297910e-02 1.63672552e-01
-2.66221136e-01 6.87994480e-01 -7.71106124e-01 -6.25359058e-01
-6.04340255e-01 -1.10516465e+00 1.19799864e+00 6.27146780e-01
-1.25971690e-01 -7.13730395e-01 -1.23057984e-01 7.62287199e-01
1.78991973e-01 2.43826717e-01 3.01106334e-01 -3.42163831e-01
-2.53387958e-01 -3.12231243e-01 -5.44724464e-01 2.69693494e-01
2.85157412e-01 -4.23976421e-01 -9.57544327e-01 -6.14398718e-01
-4.10554856e-01 -3.61930765e-02 8.29946816e-01 2.59124696e-01
1.23534453e+00 -2.65873492e-01 -5.15513480e-01 9.23418760e-01
1.09124815e+00 2.75169536e-02 6.08183682e-01 3.82922620e-01
1.39703774e+00 8.16616595e-01 4.60767746e-01 2.03313380e-01
7.66469061e-01 9.43308055e-01 2.00263448e-02 -6.78049088e-01
-5.57673097e-01 -5.75879216e-01 2.03443319e-01 4.03913617e-01
-5.61350703e-01 -1.47228643e-01 -6.08247697e-01 3.69936705e-01
-1.82795227e+00 -9.06396031e-01 -9.27735027e-03 2.15671968e+00
6.62754834e-01 -2.30239555e-01 5.09612679e-01 -1.19201452e-01
1.34934890e+00 -7.96398073e-02 -6.24339223e-01 2.51457483e-01
-2.84726560e-01 -1.92917779e-01 7.49171257e-01 5.37219644e-01
-1.42728436e+00 1.12857497e+00 6.01281548e+00 7.53176630e-01
-7.34155834e-01 1.40940413e-01 8.42565238e-01 2.01419249e-01
-6.20807968e-02 -7.10874379e-01 -1.03858829e+00 7.32608140e-01
2.54511684e-01 4.99832630e-02 4.33305532e-01 7.98168361e-01
3.27903256e-02 3.22383583e-01 -8.71754646e-01 1.55965495e+00
6.79568410e-01 -8.23508084e-01 3.73999953e-01 -2.76702666e-03
8.21215510e-01 -7.76410758e-01 1.82798147e-01 1.05734229e-01
2.92782634e-01 -9.86526787e-01 1.10448384e+00 6.63946211e-01
9.77599561e-01 -9.22828734e-01 8.45454276e-01 -2.62936860e-01
-1.45854092e+00 -8.31432454e-03 -2.74256527e-01 3.89456451e-01
2.73237258e-01 3.79184894e-02 -2.83885866e-01 6.39535069e-01
1.22055578e+00 6.62105620e-01 -9.55831587e-01 8.40350747e-01
-7.93229938e-02 1.27260998e-01 -2.64193788e-02 3.47320199e-01
-4.61720765e-01 -1.12261489e-01 3.87681782e-01 1.15783226e+00
1.92216992e-01 -8.59170556e-02 5.21898806e-01 9.31548536e-01
4.26784111e-03 -2.84588728e-02 2.52054483e-02 2.23122925e-01
2.02499419e-01 1.04707360e+00 -7.04359889e-01 -4.25683379e-01
-2.10075736e-01 1.63681161e+00 -1.52695803e-02 4.95200008e-01
-1.13627100e+00 -4.71164919e-02 6.14865124e-01 1.95412859e-01
9.98266563e-02 -4.86670714e-03 -1.95486531e-01 -1.25409138e+00
5.02751917e-02 -9.32764173e-01 4.30666804e-01 -8.52655709e-01
-1.69787514e+00 6.45171583e-01 -1.72483269e-02 -9.40244138e-01
1.66440308e-01 -5.55471361e-01 -2.44524866e-01 9.84884083e-01
-1.15215588e+00 -1.94577157e+00 -8.72701466e-01 8.67177665e-01
4.80834007e-01 -3.09995264e-01 5.56444466e-01 6.51749730e-01
-9.90118980e-01 1.21579504e+00 -2.90620834e-01 6.28322482e-01
9.25351620e-01 -1.14916587e+00 6.55766785e-01 1.08456838e+00
-1.25011176e-01 7.31670678e-01 4.58911538e-01 -1.06206346e+00
-9.33091879e-01 -1.39288497e+00 4.95475382e-01 -3.99006218e-01
-1.73310578e-01 -4.15499926e-01 -5.37679911e-01 7.26700902e-01
-8.88247341e-02 -2.29107454e-01 6.06856525e-01 -4.83535193e-02
-2.64475346e-01 -4.02377516e-01 -1.18982971e+00 6.94687665e-01
1.44317222e+00 -3.39503497e-01 -3.80226672e-01 1.60831437e-01
7.29188144e-01 -5.17471969e-01 -7.22188950e-01 6.54942811e-01
7.59230793e-01 -7.17048049e-01 1.59762502e+00 -3.71817946e-01
-6.23199996e-03 -5.13888299e-01 -6.31461442e-02 -1.03484666e+00
-7.15592980e-01 -3.08423340e-01 1.13454983e-01 1.73029792e+00
-6.55782968e-02 -5.38990557e-01 8.93725038e-01 1.00668764e+00
3.07830006e-01 -1.37776032e-01 -4.67940420e-01 -6.76078558e-01
-2.87084490e-01 -2.15217061e-02 1.00392711e+00 8.07512343e-01
-7.27811515e-01 6.05265982e-02 -9.99970317e-01 4.25336778e-01
1.15109921e+00 1.11504957e-01 1.05210578e+00 -1.07159114e+00
-1.56158313e-01 -2.53953725e-01 -3.26288790e-01 -1.19105470e+00
3.56295668e-02 -7.61071861e-01 -3.65677066e-02 -1.44545484e+00
7.53746092e-01 -3.68467242e-01 -1.97354183e-01 4.47657079e-01
-5.71433544e-01 6.58865035e-01 3.05132270e-01 3.49259526e-01
-6.98303640e-01 7.15022206e-01 1.48936260e+00 -4.29130018e-01
-8.52262080e-02 -1.97827630e-02 -9.26641107e-01 7.33579874e-01
6.59010530e-01 -6.06260486e-02 -2.78066784e-01 -5.56461573e-01
-2.80778289e-01 -5.19124150e-01 9.48803723e-01 -1.14694881e+00
2.20862329e-01 8.48684460e-02 1.12087059e+00 -6.14659607e-01
3.82747769e-01 -6.99965775e-01 2.95311421e-01 4.69639778e-01
-2.00781927e-01 4.64288928e-02 1.43994540e-01 7.16557384e-01
9.58226323e-02 2.07189411e-01 8.78159881e-01 -1.84535652e-01
-9.15238738e-01 7.10923910e-01 2.24624380e-01 -1.13326237e-01
9.18933928e-01 -5.21813929e-01 -1.48002371e-01 -2.97361135e-01
-7.68754125e-01 2.95915306e-01 6.45027816e-01 7.63794422e-01
6.15352631e-01 -1.69739103e+00 -8.45437825e-01 3.66837561e-01
1.42200798e-01 -2.82389075e-01 6.07270539e-01 4.31609541e-01
-5.04573584e-01 1.07329853e-01 -4.00992364e-01 -6.22552574e-01
-1.53103089e+00 5.29738963e-01 7.86618590e-01 -5.72526008e-02
-6.68572783e-01 9.50519323e-01 5.37544370e-01 -5.50432503e-01
4.82302964e-01 1.73312247e-01 -2.93688715e-01 -2.75080800e-01
7.22164214e-01 3.91510069e-01 -4.21374470e-01 -1.11377454e+00
-4.34198558e-01 9.56375182e-01 -1.72097653e-01 -6.31803274e-02
8.45501602e-01 -6.15549982e-01 -4.63005267e-02 -1.58411473e-01
8.02443326e-01 -1.27281353e-01 -1.40723050e+00 -2.72331446e-01
-5.60771823e-01 -5.94857931e-01 -3.52206051e-01 -9.60832357e-01
-1.38857818e+00 5.38224936e-01 1.19085133e+00 -2.89949328e-01
9.62092638e-01 -2.38943830e-01 1.19234228e+00 -3.59217413e-02
5.00302136e-01 -1.29957044e+00 3.62441063e-01 8.83290321e-02
9.12312090e-01 -1.36452258e+00 -8.38748664e-02 -7.06239879e-01
-6.84903681e-01 6.05953038e-01 1.08099020e+00 1.30438125e-02
2.71728873e-01 -1.71338364e-01 1.42325759e-01 8.15350041e-02
4.82287645e-01 -3.14155757e-01 6.77281260e-01 1.10523760e+00
7.40412101e-02 1.90331772e-01 5.78738488e-02 1.16713703e+00
-3.64820570e-01 -3.34656537e-01 -2.43382081e-01 2.48938888e-01
-2.17611387e-01 -1.18187356e+00 -8.06675851e-01 -1.94767416e-02
-3.41932178e-01 1.94421515e-01 -6.11274421e-01 7.01117992e-01
5.99693596e-01 1.00423253e+00 1.05200112e-02 -5.87371469e-01
3.60317469e-01 -2.25644454e-01 6.93367958e-01 -3.21406901e-01
-5.37982583e-01 2.34941617e-02 1.12454295e-01 -4.91632938e-01
-6.01666451e-01 -3.91856223e-01 -8.75443339e-01 -6.23278260e-01
-8.55749696e-02 -1.97582603e-01 3.15892875e-01 7.65136003e-01
1.87421381e-01 6.44477785e-01 3.90978694e-01 -9.77553666e-01
-2.13120997e-01 -7.44788170e-01 -4.67053354e-01 1.06714308e+00
7.99743086e-02 -7.28721142e-01 1.12890333e-01 5.20633459e-01]
|
[14.631427764892578, 0.8882884383201599]
|
c0de0590-2877-44a7-b0db-f2b0e8ecbe00
|
mgpsn-motion-guided-pseudo-siamese-network
|
2110.03302
| null |
https://arxiv.org/abs/2110.03302v5
|
https://arxiv.org/pdf/2110.03302v5.pdf
|
MPSN: Motion-aware Pseudo Siamese Network for Indoor Video Head Detection in Buildings
|
Head detection in the indoor video is an essential component of building occupancy detection. While deep models have achieved remarkable progress in general object detection, they are not satisfying enough in complex indoor scenes. The indoor surveillance video often includes cluttered background objects, among which heads have small scales and diverse poses. In this paper, we propose Motion-aware Pseudo Siamese Network (MPSN), an end-to-end approach that leverages head motion information to guide the deep model to extract effective head features in indoor scenarios. By taking the pixel-wise difference of adjacent frames as the auxiliary input, MPSN effectively enhances human head motion information and removes the irrelevant objects in the background. Compared with prior methods, it achieves superior performance on the two indoor video datasets. Our experiments show that MPSN successfully suppresses static background objects and highlights the moving instances, especially human heads in indoor videos. We also compare different methods to capture head motion, which demonstrates the simplicity and flexibility of MPSN. To validate the robustness of MPSN, we conduct adversarial experiments with a mathematical solution of small perturbations for robust model selection. Finally, for confirming its potential in building control systems, we apply MPSN to occupancy counting. Code is available at https://github.com/pl-share/MPSN.
|
['Peng Liu', 'Qianchuan Zhao', 'Xiaoteng Ma', 'Kailai Sun']
|
2021-10-07
| null | null | null | null |
['head-detection']
|
['computer-vision']
|
[ 1.45479605e-01 -4.01711553e-01 1.89733282e-01 -2.25628361e-01
-7.25174487e-01 -2.83256769e-01 2.23753810e-01 -2.82728404e-01
-4.93640810e-01 7.03117907e-01 4.94412601e-01 2.15239003e-02
-6.25443161e-02 -5.89108646e-01 -7.58584380e-01 -1.11081839e+00
-3.09959292e-01 -3.88215482e-02 3.90896022e-01 -4.35071774e-02
-1.81644231e-01 3.24352205e-01 -1.59399903e+00 -4.98390868e-02
5.33160627e-01 6.74350441e-01 3.32895011e-01 1.02570069e+00
6.26621425e-01 1.11333525e+00 -6.73205674e-01 1.93281919e-01
3.35757643e-01 -9.82078537e-02 -2.49817848e-01 2.07181007e-01
8.56156290e-01 -7.02458978e-01 -9.93160903e-01 8.96910250e-01
1.16031742e+00 4.84697312e-01 1.79798037e-01 -1.35115826e+00
-3.37674677e-01 2.73950607e-01 -7.11748421e-01 6.31500125e-01
6.27447367e-01 4.04543519e-01 6.74333394e-01 -9.28402841e-01
9.40915719e-02 1.34132457e+00 7.25203454e-01 7.12166727e-01
-8.98074090e-01 -8.84153426e-01 4.04315919e-01 3.10818940e-01
-1.60112143e+00 -7.29034662e-01 7.40823090e-01 -9.01462138e-02
4.09614623e-01 8.00321281e-01 8.81351829e-01 1.26039112e+00
-6.83374628e-02 1.35822165e+00 5.91441751e-01 -9.87197645e-03
3.13596815e-01 -4.07531857e-01 2.53955796e-02 6.61117196e-01
2.89460868e-01 3.05117685e-02 -4.84870821e-01 -1.18899636e-01
5.68361580e-01 4.43076372e-01 -6.97535276e-01 -3.87444198e-01
-1.12112081e+00 6.55209959e-01 6.34950340e-01 8.89862422e-03
-3.57266247e-01 3.83751094e-01 7.05765784e-02 -4.92255062e-01
2.38221124e-01 -1.12851381e-01 -1.39890999e-01 2.04210535e-01
-1.15438926e+00 4.56932008e-01 6.03008151e-01 1.03699970e+00
2.57634163e-01 4.82155494e-02 -6.23608410e-01 6.99842453e-01
3.56950939e-01 1.15692306e+00 7.74386302e-02 -1.30296123e+00
3.00294846e-01 1.14908829e-01 2.30869114e-01 -1.01035225e+00
-5.45025527e-01 -4.38287556e-01 -1.09122169e+00 -2.28775755e-01
2.57915407e-01 -1.73558876e-01 -1.07044995e+00 1.88749731e+00
7.91691899e-01 3.12394798e-01 -3.67290705e-01 1.01507986e+00
8.73897254e-01 7.64876544e-01 8.75153691e-02 -2.32549056e-01
1.11505473e+00 -1.04433334e+00 -7.69162774e-01 -6.69446290e-01
3.39850932e-02 -3.99481088e-01 8.14677477e-01 8.75487551e-02
-1.12094784e+00 -4.39170420e-01 -8.25205207e-01 5.64074330e-02
-6.67297223e-04 -1.37499288e-01 2.44138524e-01 5.39117932e-01
-1.05103290e+00 1.28078282e-01 -1.15659642e+00 -4.80856419e-01
6.66132331e-01 7.05590785e-01 -3.45953554e-02 -5.36483884e-01
-9.56976652e-01 4.33545887e-01 -3.06887012e-02 2.94026375e-01
-1.18796539e+00 -5.97852170e-01 -1.04357851e+00 2.28581857e-02
4.91956413e-01 -9.29390490e-01 1.41743708e+00 -4.89745617e-01
-8.60120952e-01 4.01522964e-01 -3.97091120e-01 -4.36293453e-01
6.96861923e-01 -3.84790272e-01 -1.36568666e-01 -2.45589595e-02
4.82507229e-01 8.54303479e-01 8.55855048e-01 -1.17100978e+00
-8.70874047e-01 -4.20757294e-01 -3.25635192e-03 2.01002344e-01
-2.84132242e-01 -9.51046422e-02 -9.09620523e-01 -5.99843025e-01
1.94561630e-02 -1.15184391e+00 -3.94636154e-01 9.19796340e-03
-6.10074580e-01 1.92554846e-01 9.16600823e-01 -7.89465785e-01
1.49191749e+00 -2.29041624e+00 -1.35214075e-01 1.40478283e-01
3.25642079e-01 1.46483183e-01 6.41628057e-02 6.28807247e-02
2.52725452e-01 -3.23089480e-01 -5.24751723e-01 -7.12187648e-01
1.55544892e-01 2.70524204e-01 2.93726157e-02 9.08987522e-01
-2.31366023e-01 8.25032413e-01 -1.02227318e+00 -6.46910012e-01
5.93588710e-01 8.75755012e-01 -5.83925366e-01 -2.62533990e-03
2.05105603e-01 5.07308006e-01 -4.14593279e-01 9.12754178e-01
9.84508276e-01 -1.83876336e-01 -8.64378065e-02 -6.36070594e-02
-1.72868874e-02 -1.12272672e-01 -1.29531670e+00 1.20342469e+00
-2.21284211e-01 7.08186269e-01 5.50519407e-01 -4.61750060e-01
1.19008191e-01 6.92771673e-02 5.30999303e-01 -6.45572186e-01
2.02565745e-01 -1.84646055e-01 -2.07868323e-01 -3.27079713e-01
3.72927248e-01 2.28503510e-01 -1.50953740e-01 -1.94855690e-01
-3.95047486e-01 7.43515193e-02 3.00306916e-01 2.55314946e-01
1.54163671e+00 -2.77016014e-01 3.30199271e-01 -1.97062373e-01
5.17067313e-01 -5.04601657e-01 7.25834429e-01 1.10974634e+00
-7.45344579e-01 8.67579699e-01 -3.27194810e-01 -3.74459326e-01
-6.28982663e-01 -1.25963664e+00 -7.96975419e-02 1.37069452e+00
1.78794950e-01 -2.31970921e-01 -9.24369752e-01 -4.68575656e-01
-4.28133234e-02 6.02249622e-01 -5.88273346e-01 2.79685408e-02
-8.75927866e-01 -9.64071870e-01 3.78544450e-01 7.47983158e-01
5.98807275e-01 -1.02760601e+00 -8.24339092e-01 1.67990178e-01
-5.38895547e-01 -1.24622679e+00 -7.84320652e-01 1.66778013e-01
-2.95698583e-01 -1.04643869e+00 -9.76936221e-01 -6.58074319e-01
6.79112434e-01 8.10268402e-01 9.93219376e-01 2.31346980e-01
-5.31680882e-01 6.73161089e-01 9.21136513e-02 -4.63823497e-01
6.32232055e-02 -1.95394587e-02 2.81217009e-01 -1.90695703e-01
2.15765476e-01 -4.84975338e-01 -1.01024830e+00 3.06251198e-01
-8.36182833e-01 -1.55225649e-01 3.24006319e-01 4.89074945e-01
4.15033638e-01 2.65900075e-01 -1.16841830e-02 -3.22911084e-01
1.96975902e-01 -3.17512274e-01 -5.65905869e-01 -1.40193895e-01
1.98381931e-01 -4.93372679e-01 5.68915009e-01 -2.21508265e-01
-7.20349252e-01 3.52662683e-01 -3.21592033e-01 -4.02611077e-01
-3.02893668e-01 -2.37112671e-01 -4.84464794e-01 2.69605238e-02
2.82977909e-01 2.78794497e-01 -4.14151013e-01 -1.87177092e-01
9.01343580e-03 2.44855747e-01 7.74931371e-01 -1.67576626e-01
1.03725886e+00 9.02405262e-01 -1.26593575e-01 -1.28351343e+00
-1.25246727e+00 -8.40119839e-01 -3.05832446e-01 -2.89770633e-01
1.03602362e+00 -1.22400141e+00 -8.68741155e-01 5.67090333e-01
-9.32031453e-01 -4.84782457e-01 -2.44343176e-01 2.61620671e-01
-2.57964969e-01 2.97706485e-01 -7.09111214e-01 -1.19504452e+00
-3.78948182e-01 -1.01283526e+00 1.36174011e+00 4.38260764e-01
-1.25052914e-01 -7.65724361e-01 1.99958566e-04 5.20221412e-01
2.70680755e-01 3.09046060e-01 5.73268421e-02 -2.46181697e-01
-8.63575399e-01 -3.90875906e-01 1.26840606e-01 1.51926950e-01
1.74105689e-01 7.70064965e-02 -1.22286391e+00 -6.42301500e-01
6.72894716e-02 1.14656985e-01 1.06479168e+00 1.13443613e+00
1.21383595e+00 -4.20455366e-01 -5.93623519e-01 7.25656509e-01
9.02162850e-01 1.30661249e-01 5.61994493e-01 5.12041628e-01
9.72899735e-01 3.50618422e-01 5.31192899e-01 7.12288857e-01
5.58077753e-01 5.66825747e-01 5.38782001e-01 -3.84725720e-01
-5.24230935e-02 -2.02330023e-01 5.40946662e-01 5.82162261e-01
9.57411453e-02 -5.12109697e-01 -9.05961275e-01 5.32478750e-01
-1.90493262e+00 -1.22456491e+00 2.91132666e-02 2.15259123e+00
3.26490611e-01 2.54268408e-01 2.42065951e-01 2.26448894e-01
7.90513933e-01 5.14678717e-01 -5.97351789e-01 6.36555910e-01
-1.98524803e-01 4.98335809e-02 7.93853402e-01 6.06270790e-01
-1.66872013e+00 6.34819925e-01 5.49065018e+00 5.91235578e-01
-6.42702997e-01 2.69113302e-01 4.93220627e-01 -6.59514129e-01
2.52944410e-01 -5.72512209e-01 -1.09305918e+00 6.68895423e-01
5.65084815e-01 3.10630620e-01 2.87046701e-01 9.63928819e-01
6.20675027e-01 -2.51819402e-01 -9.52179313e-01 9.80054080e-01
1.89575404e-01 -9.52524662e-01 -4.06642228e-01 1.67330787e-01
7.25775659e-01 1.92185462e-01 2.37343282e-01 4.24459606e-01
3.45069349e-01 -7.98112929e-01 8.06875110e-01 3.28540802e-01
4.14703600e-02 -8.30705225e-01 6.48766994e-01 4.82241869e-01
-1.49028003e+00 -4.37665939e-01 -3.86142880e-01 -5.55122420e-02
4.42451417e-01 4.09510940e-01 -6.30468071e-01 3.26036550e-02
1.08338726e+00 4.33999211e-01 -7.74577916e-01 1.28692210e+00
-3.40165585e-01 5.97147048e-01 -7.22002208e-01 9.27591249e-02
2.03050107e-01 2.24046126e-01 6.61298215e-01 1.39667404e+00
1.66867405e-01 2.02813104e-01 5.40660143e-01 5.40324867e-01
1.42794524e-04 -3.72578859e-01 -6.27771914e-01 6.78341329e-01
4.69861925e-01 1.08776438e+00 -7.26069927e-01 -2.89837092e-01
-2.71219254e-01 1.03604877e+00 -1.58461481e-01 4.53347206e-01
-1.36321175e+00 -4.79219817e-02 8.26042771e-01 3.43277007e-01
5.18413723e-01 -2.34196916e-01 2.48676732e-01 -8.78391027e-01
1.23989750e-02 -8.55476141e-01 2.73776889e-01 -5.72460353e-01
-9.01229620e-01 1.36407316e-01 5.49139231e-02 -8.77384007e-01
2.29950368e-01 -2.55259067e-01 -6.77478611e-01 2.12472558e-01
-1.12795675e+00 -1.05344093e+00 -7.67153680e-01 7.76728153e-01
9.66131866e-01 4.76464182e-01 3.34457785e-01 7.51697361e-01
-1.11788762e+00 6.00649774e-01 9.70597193e-02 3.40226024e-01
5.26733518e-01 -1.05450344e+00 6.05800390e-01 1.29708886e+00
-1.81125537e-01 5.51179469e-01 9.77906227e-01 -6.19719744e-01
-1.31652343e+00 -1.53915942e+00 3.68219554e-01 -6.57442510e-01
1.81561306e-01 -6.19084835e-01 -6.75242424e-01 6.56796277e-01
3.36975269e-02 2.94049144e-01 5.33327281e-01 -4.88746822e-01
3.85151386e-01 -3.04348413e-02 -9.91889775e-01 7.48636961e-01
1.19016910e+00 -1.23171628e-01 -2.17136234e-01 5.58337152e-01
7.58660495e-01 -5.41258693e-01 -2.32307777e-01 5.77502728e-01
4.47230577e-01 -1.08324659e+00 1.43322659e+00 -1.77139938e-02
2.96516009e-02 -6.59533441e-01 -5.73356330e-01 -7.81771004e-01
-6.63718045e-01 -5.06219506e-01 -5.57026625e-01 9.93574560e-01
-2.29091458e-02 -1.47390753e-01 1.03860950e+00 7.56033719e-01
-9.83799845e-02 -4.84467357e-01 -1.02656758e+00 -6.17174447e-01
-5.27539015e-01 -5.46876073e-01 3.69526982e-01 5.27696371e-01
-7.16856956e-01 2.52372205e-01 -6.55547738e-01 7.23108828e-01
1.07130468e+00 -4.29183543e-01 1.01870024e+00 -8.43683004e-01
-9.92622524e-02 -8.64995718e-02 -3.61269951e-01 -1.23727334e+00
2.02823952e-01 -2.84903228e-01 5.21428883e-01 -1.74720097e+00
3.67593199e-01 1.01488583e-01 -3.91111434e-01 2.89387226e-01
-5.14242113e-01 3.25439334e-01 3.30448478e-01 -7.93017894e-02
-1.03316259e+00 7.32860625e-01 9.58649158e-01 -4.49301213e-01
-2.53431588e-01 4.38607544e-01 -4.31298167e-01 1.19057941e+00
7.87513077e-01 -4.68611002e-01 -2.13676915e-01 -3.22073072e-01
-4.22751456e-01 -2.23000303e-01 7.42636263e-01 -1.58494878e+00
4.29409653e-01 -8.21057931e-02 9.28395331e-01 -9.64002788e-01
6.90588236e-01 -8.93734872e-01 -2.52959747e-02 8.24981093e-01
9.74554271e-02 6.33139759e-02 2.15284422e-01 7.65724063e-01
1.45977452e-01 2.48881534e-01 9.91963148e-01 -1.25023440e-01
-7.04779923e-01 5.53842962e-01 -5.20320058e-01 7.86672533e-02
9.08771992e-01 -2.57712692e-01 -1.09830134e-01 -6.62360728e-01
-5.61564088e-01 5.75322688e-01 5.51281750e-01 2.47833312e-01
7.34418452e-01 -1.20457983e+00 -4.88197386e-01 2.01924369e-01
-2.27434888e-01 3.18650752e-01 5.73519409e-01 1.04819536e+00
-4.64630276e-01 3.71993184e-01 3.54758829e-01 -8.27912748e-01
-1.57619846e+00 6.76596582e-01 3.79074723e-01 5.47816567e-02
-6.13171101e-01 1.14638376e+00 8.32780123e-01 -3.67111504e-01
7.80072272e-01 -5.51982641e-01 1.70067903e-02 -2.70129323e-01
8.99739742e-01 7.29268670e-01 -2.60457397e-01 -8.02714944e-01
-7.71040142e-01 3.01539630e-01 7.24414811e-02 1.00031689e-01
1.27697515e+00 -4.17027086e-01 2.58814692e-01 1.36198997e-01
1.05912042e+00 2.33471930e-01 -1.64725339e+00 7.10063279e-02
-3.82822275e-01 -5.40600598e-01 6.61755055e-02 -2.28143916e-01
-1.19425809e+00 5.77832878e-01 9.91090238e-01 9.56395920e-03
1.25590670e+00 -5.82465604e-02 9.91429925e-01 4.68419760e-01
2.77782261e-01 -9.30720448e-01 7.84503147e-02 5.05173624e-01
7.12569416e-01 -1.41744566e+00 1.43026188e-01 -2.79574752e-01
-3.73142421e-01 4.82755542e-01 7.17096567e-01 8.30431506e-02
4.99805361e-01 5.21912575e-01 -9.13798660e-02 -1.22991152e-01
-4.26369578e-01 -3.66674423e-01 -6.18243627e-02 6.56392157e-01
1.17495649e-01 8.81241709e-02 4.22012627e-01 3.12487781e-01
-3.32579643e-01 -2.22463489e-01 3.63206565e-01 1.34644544e+00
-5.82819104e-01 -3.54368538e-01 -9.50490892e-01 2.41874084e-01
-5.99889755e-01 -1.39684722e-01 -1.20409727e-01 8.46915841e-01
2.62338251e-01 1.10050368e+00 -1.87028438e-01 -2.14886963e-01
6.30107939e-01 -4.41601098e-01 3.42442930e-01 -3.81328940e-01
-3.12867045e-01 2.24070296e-01 -2.28239760e-01 -7.64476657e-01
-4.04671729e-01 -7.81103015e-01 -1.06404686e+00 -6.11859381e-01
-2.99518317e-01 -2.06754580e-01 2.34121069e-01 6.28216326e-01
-3.09398267e-02 8.32664073e-01 5.52191854e-01 -1.50704861e+00
-4.02378589e-01 -6.70539796e-01 -5.29477775e-01 3.07137161e-01
9.38261032e-01 -5.85017085e-01 -6.57801151e-01 -6.05738983e-02]
|
[7.660022258758545, -0.7855173945426941]
|
a269f6d6-a8ee-4971-9a58-d74cba007085
|
afnet-m-adaptive-fusion-network-with-masks
|
2205.11785
| null |
https://arxiv.org/abs/2205.11785v1
|
https://arxiv.org/pdf/2205.11785v1.pdf
|
AFNet-M: Adaptive Fusion Network with Masks for 2D+3D Facial Expression Recognition
|
2D+3D facial expression recognition (FER) can effectively cope with illumination changes and pose variations by simultaneously merging 2D texture and more robust 3D depth information. Most deep learning-based approaches employ the simple fusion strategy that concatenates the multimodal features directly after fully-connected layers, without considering the different degrees of significance for each modality. Meanwhile, how to focus on both 2D and 3D local features in salient regions is still a great challenge. In this letter, we propose the adaptive fusion network with masks (AFNet-M) for 2D+3D FER. To enhance 2D and 3D local features, we take the masks annotating salient regions of the face as prior knowledge and design the mask attention module (MA) which can automatically learn two modulation vectors to adjust the feature maps. Moreover, we introduce a novel fusion strategy that can perform adaptive fusion at convolutional layers through the designed importance weights computing module (IWC). Experimental results demonstrate that our AFNet-M achieves the state-of-the-art performance on BU-3DFE and Bosphorus datasets and requires fewer parameters in comparison with other models.
|
['Feng Zhao', 'Zhaoqing Zhu', 'Hanting Li', 'Mingzhe Sui']
|
2022-05-24
| null | null | null | null |
['3d-facial-expression-recognition', 'facial-expression-recognition']
|
['computer-vision', 'computer-vision']
|
[ 1.09725699e-01 -2.13279337e-01 4.53024469e-02 -6.66565776e-01
-3.77075523e-01 -5.60100283e-03 4.47077543e-01 -4.00168657e-01
-4.51016039e-01 3.74720961e-01 1.78729147e-01 2.26520732e-01
-4.94173095e-02 -6.28431261e-01 -3.52850497e-01 -8.74017596e-01
3.47299548e-03 -2.18933895e-01 6.68689311e-02 -4.84245688e-01
1.45436481e-01 1.06254542e+00 -1.83337295e+00 3.58621359e-01
5.58428586e-01 1.66525006e+00 1.41358480e-01 2.28229344e-01
-2.64634281e-01 5.32190561e-01 -5.72287679e-01 -4.76219833e-01
1.29769832e-01 -1.27603441e-01 -2.84843177e-01 3.63307327e-01
6.40207410e-01 -4.32403862e-01 -3.22348207e-01 1.18749297e+00
9.91406143e-01 -9.50241387e-02 5.52643359e-01 -1.22652316e+00
-5.34097433e-01 -2.53221318e-02 -1.05285001e+00 2.72442013e-01
2.33040959e-01 1.41395897e-01 4.90298390e-01 -1.36868274e+00
4.06830102e-01 1.59295285e+00 3.89893383e-01 6.65386617e-01
-7.93663681e-01 -7.73155034e-01 3.68096888e-01 3.51946712e-01
-1.63841748e+00 -6.73919559e-01 1.24492550e+00 -2.12828249e-01
7.65908539e-01 1.76376045e-01 5.53422511e-01 9.73670900e-01
4.75804992e-02 8.76261771e-01 1.09509456e+00 -2.66653240e-01
-1.42425731e-01 -1.10495657e-01 -1.74598992e-01 1.09331882e+00
-2.21395195e-01 -1.82792664e-01 -7.73005009e-01 -2.76190788e-02
8.88190985e-01 1.00548998e-01 -3.84402871e-01 -2.05866560e-01
-7.38838375e-01 5.96645474e-01 4.06011641e-01 4.16128904e-01
-5.43138981e-01 4.31254953e-02 1.93183273e-01 1.61999688e-01
6.39064193e-01 -1.19371809e-01 -6.13599718e-01 2.43141964e-01
-7.61098027e-01 1.01906091e-01 1.53862223e-01 7.00253010e-01
9.82532859e-01 1.43734798e-01 -4.09093529e-01 9.68513787e-01
4.59950536e-01 4.02303189e-01 5.03268957e-01 -7.80019701e-01
8.04990456e-02 8.73048723e-01 -1.06321476e-01 -1.18263054e+00
-5.57703435e-01 -4.10918236e-01 -9.68319893e-01 3.18916321e-01
-6.12789020e-02 -2.69663095e-01 -1.01429391e+00 2.00616765e+00
7.00310230e-01 1.35991305e-01 -1.21962570e-01 1.18626416e+00
1.28466213e+00 2.57843614e-01 -1.70852561e-02 -2.93602318e-01
1.36709809e+00 -8.55422497e-01 -1.05448341e+00 4.08716872e-02
2.00075820e-01 -8.46870363e-01 6.56300843e-01 2.36736163e-01
-1.20016885e+00 -8.54949713e-01 -1.02368033e+00 -1.76161647e-01
-4.18049872e-01 2.97251821e-01 7.75457025e-01 5.31657875e-01
-1.08790922e+00 2.27689922e-01 -5.46057880e-01 -5.61674796e-02
7.33681858e-01 6.98930144e-01 -6.96213543e-01 2.58288905e-02
-1.29858291e+00 7.77943254e-01 1.85774833e-01 6.14957333e-01
-7.90457726e-01 -4.14776415e-01 -1.13904929e+00 -3.22554074e-02
2.12065578e-01 -4.87168431e-01 9.12382066e-01 -1.20545900e+00
-1.81251144e+00 1.08277118e+00 -3.29095900e-01 3.09772521e-01
2.47375354e-01 -3.27213742e-02 -4.86810267e-01 2.94994265e-01
-3.09303969e-01 9.44624782e-01 1.15956199e+00 -1.04327869e+00
-5.91852188e-01 -8.11165690e-01 8.21452066e-02 4.63724554e-01
-5.71271062e-01 2.42482856e-01 -7.07537174e-01 -6.23225391e-01
1.62478685e-01 -2.50653684e-01 -2.34921481e-02 4.95500743e-01
-5.52269891e-02 -3.72690976e-01 1.07527649e+00 -5.13872743e-01
1.23664999e+00 -2.29078746e+00 3.89565349e-01 1.52334407e-01
2.94536114e-01 3.96057248e-01 -3.52090508e-01 -3.53779882e-01
-1.65897831e-02 -2.36408100e-01 -2.77409405e-02 -5.58553576e-01
2.94335745e-02 1.31133020e-01 2.54387438e-01 5.97599626e-01
7.67183721e-01 8.77311170e-01 -5.84862471e-01 -6.61111116e-01
4.10458505e-01 7.85307705e-01 -5.98819196e-01 2.97981620e-01
9.22589749e-02 4.00994390e-01 -5.99535406e-01 9.59657729e-01
1.31947875e+00 -2.27325875e-02 1.42244231e-02 -6.50122166e-01
-3.20998617e-02 -4.15684879e-01 -1.13867819e+00 1.95134664e+00
-3.17835182e-01 2.00161859e-01 5.79524815e-01 -9.62922752e-01
1.22378516e+00 2.33027533e-01 4.92374450e-01 -8.91524732e-01
6.25411749e-01 1.63640559e-01 -2.49758691e-01 -7.05715835e-01
2.06559211e-01 -1.58385381e-01 1.73727982e-02 -1.18459731e-01
6.00478470e-01 1.78194970e-01 -3.79068285e-01 -4.91730332e-01
5.67014217e-01 1.45916462e-01 1.04157649e-01 -1.60298914e-01
1.07594812e+00 -8.12256098e-01 7.93188632e-01 1.00351922e-01
-3.96409273e-01 5.03123105e-01 4.95908558e-01 -6.07228458e-01
-4.78745550e-01 -7.31749237e-01 -3.25752139e-01 1.05794036e+00
3.76134694e-01 -1.46702111e-01 -7.17431366e-01 -8.46104801e-01
8.61094519e-02 -3.90919596e-02 -1.00018132e+00 -2.86811918e-01
-4.05297428e-01 -8.10797870e-01 3.38716030e-01 3.33230495e-01
9.70938325e-01 -7.76747048e-01 -4.82956350e-01 -5.63604981e-02
-1.81746464e-02 -9.76235509e-01 -4.50126380e-01 1.49889499e-01
-3.63074332e-01 -8.13248217e-01 -9.27340329e-01 -7.76336133e-01
6.86430275e-01 2.05849081e-01 6.59571409e-01 1.11356623e-01
-3.51286650e-01 8.43035281e-02 -2.54942566e-01 -3.60677481e-01
3.49315017e-01 -1.35408252e-01 -1.40819043e-01 7.13088751e-01
5.95916808e-01 -5.66413164e-01 -6.22675300e-01 3.51441801e-01
-9.27225888e-01 8.70100260e-02 5.38177967e-01 8.99325311e-01
6.72220945e-01 -1.31666347e-01 4.27536279e-01 -2.97114938e-01
5.09245932e-01 -2.91689515e-01 -4.05982524e-01 2.28460982e-01
5.78245819e-02 -1.84489310e-01 1.97293237e-01 -4.41208690e-01
-1.20103168e+00 1.08476378e-01 -3.97137582e-01 -8.80730391e-01
-1.98510885e-01 2.09080622e-01 -7.74881065e-01 -4.87734616e-01
2.15463236e-01 1.07858628e-01 1.92599401e-01 -4.96588290e-01
2.36043587e-01 8.51768255e-01 3.75793010e-01 -4.93942857e-01
4.11026895e-01 5.72065413e-01 2.66012043e-01 -7.06375122e-01
-9.28804994e-01 -1.22504488e-01 -7.66373515e-01 -4.99731332e-01
9.31623697e-01 -1.08590269e+00 -9.27506149e-01 8.87637734e-01
-1.10820723e+00 3.05390568e-03 -1.18964715e-02 2.56772935e-01
-4.10167068e-01 2.10552439e-01 -4.37856764e-01 -7.21807420e-01
-3.62867951e-01 -1.30289435e+00 1.61882079e+00 5.95499337e-01
2.11415529e-01 -7.67104566e-01 -3.20663691e-01 1.14097431e-01
4.93450850e-01 5.79379141e-01 6.31922245e-01 -2.01647669e-01
-2.37638757e-01 5.67517709e-03 -4.78466332e-01 5.78160107e-01
3.41486633e-01 -1.08980261e-01 -1.38757086e+00 -1.22644879e-01
7.84159377e-02 -4.38228577e-01 1.01041198e+00 3.05174798e-01
1.50746536e+00 -9.36843678e-02 -3.07675842e-02 9.63184178e-01
1.10925233e+00 7.59957805e-02 5.99906027e-01 -8.06864798e-02
6.76550686e-01 7.32032776e-01 6.60646081e-01 7.88950920e-01
2.91689575e-01 8.29820573e-01 6.05955184e-01 -4.72882003e-01
-1.87316477e-01 1.09792151e-01 2.25111350e-01 4.93383348e-01
-1.32870421e-01 5.88407405e-02 -2.68978953e-01 2.56159663e-01
-1.73933291e+00 -7.22139299e-01 4.17223096e-01 1.72296858e+00
7.81516135e-01 -1.97093174e-01 -1.52484074e-01 5.40920310e-02
8.02349448e-01 3.28538567e-01 -5.75335145e-01 -3.19047183e-01
-6.65050983e-01 5.56640744e-01 7.22244307e-02 4.04319853e-01
-1.27920198e+00 9.16474521e-01 5.47860622e+00 1.28923512e+00
-1.53491950e+00 9.40594301e-02 8.07159185e-01 -1.76002920e-01
-2.21095458e-01 -5.87559998e-01 -8.71829450e-01 4.85854745e-01
1.95705041e-01 4.33690906e-01 1.25427231e-01 5.94980955e-01
3.19138654e-02 9.13802758e-02 -6.99501038e-01 1.45166063e+00
3.37650448e-01 -1.08518589e+00 2.40396351e-01 -1.02259807e-01
6.49649262e-01 -3.92626226e-01 2.69486576e-01 1.10743925e-01
-3.53208035e-01 -1.04178298e+00 7.64074087e-01 9.55556095e-01
1.02962899e+00 -1.00749338e+00 8.24368179e-01 -6.39550611e-02
-1.17312431e+00 -6.03633523e-02 -4.48993444e-01 9.55833048e-02
-3.91141213e-02 4.69750345e-01 -1.54363364e-01 5.85947692e-01
7.73218513e-01 8.33451509e-01 -5.26938915e-01 6.71293378e-01
-1.90771386e-01 -7.53660798e-02 -3.69632274e-01 1.02356099e-01
1.62721649e-01 1.14402756e-01 3.55490655e-01 1.13921547e+00
4.78723049e-01 7.30464831e-02 -1.18638940e-01 5.80495954e-01
-1.44732967e-01 1.15080252e-01 -2.51831800e-01 2.57146269e-01
1.75338537e-01 1.64864564e+00 -2.87683398e-01 -1.64867923e-01
-4.82960701e-01 1.19877732e+00 3.87849540e-01 3.37932676e-01
-7.94892251e-01 -4.11596179e-01 1.11957884e+00 -1.73254520e-01
6.20085776e-01 -3.94079136e-03 2.01088279e-01 -1.19043243e+00
2.61873472e-03 -7.59753108e-01 3.75091940e-01 -9.31521833e-01
-1.29738951e+00 7.39625096e-01 -2.15138033e-01 -9.52500820e-01
2.03086153e-01 -9.53161895e-01 -2.92016387e-01 9.96295989e-01
-2.05100536e+00 -1.39035022e+00 -5.62929273e-01 1.06520140e+00
1.93673715e-01 -2.06453770e-01 7.15568244e-01 5.89691162e-01
-7.38500714e-01 8.69748712e-01 -4.53742921e-01 5.28128371e-02
7.84005880e-01 -8.12194824e-01 -1.77353650e-01 4.20872301e-01
-3.00822526e-01 2.69786209e-01 2.18987823e-01 -1.61524236e-01
-1.40746236e+00 -9.72715855e-01 6.84386790e-01 2.48151761e-03
2.44562954e-01 -3.44275028e-01 -7.53997803e-01 1.96141884e-01
1.32327899e-01 3.95702213e-01 6.69430733e-01 -2.20810011e-01
-2.59885758e-01 -4.60165292e-01 -1.40283191e+00 4.05580699e-01
1.27715290e+00 -4.48061049e-01 -1.88703492e-01 1.10561168e-02
6.00177884e-01 -5.46827555e-01 -9.87535477e-01 9.92486835e-01
5.71963191e-01 -1.08246148e+00 7.98475146e-01 -4.89049405e-01
1.47357881e-01 -5.22148311e-01 -4.88168120e-01 -8.80996466e-01
-3.53506953e-01 -6.50329471e-01 -1.95956126e-01 1.21282315e+00
-8.36554170e-02 -3.87620986e-01 5.43395996e-01 2.05703273e-01
-9.25738364e-02 -1.10690033e+00 -1.17316914e+00 -9.03367326e-02
-4.63099301e-01 -2.38746524e-01 1.05340362e+00 1.03305519e+00
-3.53821188e-01 2.33440936e-01 -4.15037394e-01 1.22732840e-01
3.33857805e-01 6.70770034e-02 6.46590889e-01 -1.11792791e+00
9.28178206e-02 -7.01170206e-01 -8.19755554e-01 -1.05542612e+00
4.20105368e-01 -7.29863107e-01 -3.55106801e-01 -9.95629072e-01
1.47727117e-01 1.03572272e-02 -6.22962832e-01 7.12103963e-01
-3.26667517e-01 5.66342592e-01 -7.39917485e-03 -2.50260472e-01
-6.20854676e-01 1.09437644e+00 1.66928601e+00 -7.03720599e-02
9.56496447e-02 -4.64308232e-01 -8.67380619e-01 7.98690736e-01
4.19920146e-01 8.27052295e-02 -2.42551073e-01 -6.58324540e-01
-1.13107212e-01 -1.03969030e-01 3.95471871e-01 -8.10951173e-01
1.42762348e-01 -2.01966345e-01 1.02922475e+00 -5.67798018e-01
6.21393204e-01 -9.20199156e-01 -2.20213830e-01 3.07844230e-03
-4.03725505e-02 -3.05653047e-02 4.49219018e-01 2.21196443e-01
-5.11943579e-01 2.17457443e-01 9.77831185e-01 3.55357379e-02
-9.44740593e-01 7.29744256e-01 -1.97225720e-01 -4.46942270e-01
1.00630355e+00 -2.76418120e-01 -5.94734140e-02 -1.83545366e-01
-7.53563762e-01 2.31649518e-01 2.27356419e-01 4.75578099e-01
9.48326826e-01 -1.72712600e+00 -6.03761077e-01 8.31118762e-01
1.90333158e-01 -8.25654063e-03 8.31513643e-01 8.36841643e-01
-2.43435115e-01 8.81852880e-02 -6.40105546e-01 -6.42034650e-01
-1.37179101e+00 3.63992780e-01 7.47094691e-01 5.96343726e-02
8.24638084e-03 1.25370634e+00 4.46607769e-01 -4.02339548e-01
2.95622140e-01 3.81278321e-02 -6.63068056e-01 3.18544120e-01
6.50411308e-01 -7.25624636e-02 8.04307759e-02 -9.91333902e-01
-4.92370397e-01 9.81661499e-01 -5.65777496e-02 2.11111814e-01
1.31494856e+00 -1.53922021e-01 -3.18407178e-01 1.94654707e-02
1.51014304e+00 -2.38130987e-01 -1.45908177e+00 -5.02384841e-01
-6.10708177e-01 -6.88786685e-01 4.05276775e-01 -5.88127613e-01
-1.66724110e+00 1.14304256e+00 9.90488827e-01 -4.65627432e-01
1.84483635e+00 -1.08669981e-01 5.15279710e-01 -1.35633620e-02
1.81630611e-01 -9.77601409e-01 3.04954648e-01 3.20803016e-01
1.05260313e+00 -1.03627169e+00 -1.54016867e-01 -2.82688230e-01
-4.00913209e-01 1.20919693e+00 9.43874180e-01 4.45667915e-02
9.52560008e-01 3.11071873e-01 1.05124705e-01 -3.80587906e-01
-6.50350511e-01 -4.45997566e-01 4.94945556e-01 3.98989677e-01
3.08143586e-01 -2.52741277e-01 -2.32351974e-01 7.85955369e-01
2.18767717e-01 7.50304386e-02 -2.83393145e-01 9.84633327e-01
-4.33372796e-01 -9.56144571e-01 -2.25443617e-01 1.53385192e-01
-5.23146808e-01 1.75205618e-01 -3.33229333e-01 6.54244542e-01
7.82001317e-01 7.20716059e-01 2.72910446e-01 -6.73571527e-01
3.12320411e-01 -2.66333073e-02 8.74877691e-01 -2.00439900e-01
-3.57732952e-01 4.22897965e-01 -2.07411647e-01 -6.88661575e-01
-8.48967910e-01 -4.63560373e-01 -1.00182450e+00 -2.06133455e-01
-4.04449016e-01 -2.52311260e-01 5.89147866e-01 8.45729411e-01
6.08131349e-01 5.71838915e-01 9.51447725e-01 -1.02087617e+00
-1.19811960e-01 -1.05481195e+00 -6.46424472e-01 3.44804287e-01
4.42065388e-01 -1.13798964e+00 -1.53937548e-01 -2.02146932e-01]
|
[13.58326244354248, 1.5468226671218872]
|
2639da95-c8ac-43c8-8cee-36df694eccd3
|
reco-retrieve-and-co-segment-for-zero-shot-1
|
2206.07045
| null |
https://arxiv.org/abs/2206.07045v1
|
https://arxiv.org/pdf/2206.07045v1.pdf
|
ReCo: Retrieve and Co-segment for Zero-shot Transfer
|
Semantic segmentation has a broad range of applications, but its real-world impact has been significantly limited by the prohibitive annotation costs necessary to enable deployment. Segmentation methods that forgo supervision can side-step these costs, but exhibit the inconvenient requirement to provide labelled examples from the target distribution to assign concept names to predictions. An alternative line of work in language-image pre-training has recently demonstrated the potential to produce models that can both assign names across large vocabularies of concepts and enable zero-shot transfer for classification, but do not demonstrate commensurate segmentation abilities. In this work, we strive to achieve a synthesis of these two approaches that combines their strengths. We leverage the retrieval abilities of one such language-image pre-trained model, CLIP, to dynamically curate training sets from unlabelled images for arbitrary collections of concept names, and leverage the robust correspondences offered by modern image representations to co-segment entities among the resulting collections. The synthetic segment collections are then employed to construct a segmentation model (without requiring pixel labels) whose knowledge of concepts is inherited from the scalable pre-training process of CLIP. We demonstrate that our approach, termed Retrieve and Co-segment (ReCo) performs favourably to unsupervised segmentation approaches while inheriting the convenience of nameable predictions and zero-shot transfer. We also demonstrate ReCo's ability to generate specialist segmenters for extremely rare objects.
|
['Samuel Albanie', 'Weidi Xie', 'Gyungin Shin']
|
2022-06-14
|
reco-retrieve-and-co-segment-for-zero-shot
|
https://arxiv.org/abs/2206.07045
|
https://arxiv.org/pdf/2206.07045
| null |
['unsupervised-semantic-segmentation-with', 'unsupervised-semantic-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 7.16164052e-01 5.36864460e-01 -3.24460506e-01 -4.13919449e-01
-1.24093568e+00 -8.99016023e-01 7.79146016e-01 -8.01787004e-02
-4.68357831e-01 6.18064225e-01 -9.49451774e-02 -2.38857567e-01
-1.99318640e-02 -7.54209936e-01 -7.59326041e-01 -3.79863828e-01
2.51909137e-01 8.34717691e-01 4.55578953e-01 -1.28740132e-01
7.91140646e-02 3.53817612e-01 -1.84100986e+00 1.19184993e-01
8.51885974e-01 7.20610380e-01 1.54763132e-01 5.12680948e-01
-4.03919131e-01 4.75529671e-01 -4.81094509e-01 -3.48517448e-01
3.44782174e-01 -5.51283717e-01 -8.81988823e-01 5.04692376e-01
5.65411687e-01 -2.05029041e-01 -5.54638915e-02 9.17614639e-01
3.11117083e-01 1.29301548e-01 8.76029968e-01 -1.20816755e+00
-8.63574743e-01 6.92536712e-01 -4.16742921e-01 -3.30453813e-02
3.79408181e-01 2.18541592e-01 1.35846126e+00 -7.24736452e-01
1.04496777e+00 8.45427394e-01 7.35660315e-01 9.32103574e-01
-1.48568404e+00 -6.07243121e-01 2.08082087e-02 -3.08090359e-01
-1.50501490e+00 -5.25680780e-01 4.60701376e-01 -6.16946876e-01
7.98683763e-01 2.54239470e-01 5.61366796e-01 9.62211668e-01
-7.12079644e-01 9.77875412e-01 9.37651098e-01 -5.20398915e-01
2.17186272e-01 4.15469080e-01 1.88142024e-02 6.26044631e-01
1.20680686e-02 -2.20353931e-01 -4.20067430e-01 4.48897630e-02
6.78063452e-01 -2.18173027e-01 -4.09732133e-01 -7.78744340e-01
-1.15018106e+00 7.79917955e-01 5.09503186e-01 3.80658358e-01
1.34849036e-02 1.93791822e-01 2.31976688e-01 1.85760260e-01
3.77776653e-01 8.96113753e-01 -4.62793112e-01 1.81905672e-01
-1.38753748e+00 1.18327186e-01 8.70862067e-01 1.52237248e+00
1.07380724e+00 -1.42848492e-01 -1.65834166e-02 7.99865842e-01
3.76214907e-02 3.34433109e-01 6.45428419e-01 -9.42569792e-01
1.03197061e-01 4.59002435e-01 8.70637968e-02 -5.79363227e-01
-7.10200742e-02 -2.80077010e-01 -2.34010190e-01 -4.96209562e-02
3.80814821e-01 1.08706601e-01 -1.47118402e+00 1.80714297e+00
2.75397032e-01 9.01667401e-02 1.44420221e-01 5.91396987e-01
5.29927909e-01 5.17487466e-01 4.01745439e-01 1.51246801e-01
1.23309410e+00 -8.75022054e-01 -3.36093828e-02 -3.11844766e-01
7.67355978e-01 -5.12950301e-01 1.29818320e+00 -5.21969497e-02
-9.80568647e-01 -4.22548980e-01 -1.04423165e+00 -4.14374992e-02
-7.05911100e-01 -3.46674919e-01 6.84900343e-01 6.78365767e-01
-1.14867067e+00 5.16749859e-01 -6.08195484e-01 -5.71491778e-01
9.34009552e-01 4.93339390e-01 -2.22438589e-01 -1.58247441e-01
-9.44888055e-01 6.78753376e-01 6.67681992e-01 -5.85162997e-01
-1.02286708e+00 -9.60256994e-01 -9.28822637e-01 -3.00012045e-02
6.85428381e-01 -7.06708312e-01 1.42380166e+00 -1.61000288e+00
-1.02341795e+00 1.24012959e+00 1.47861779e-01 -4.45600241e-01
4.65327531e-01 1.08386010e-01 -1.09867424e-01 4.01357621e-01
5.99265277e-01 1.55248415e+00 9.55874801e-01 -1.39710057e+00
-7.16497838e-01 3.31422053e-02 2.06366107e-01 2.51455098e-01
-3.00665349e-01 -6.42206818e-02 -7.06307590e-01 -5.86588323e-01
4.27320376e-02 -9.40588772e-01 -2.76666075e-01 3.08914054e-02
-4.96024072e-01 -8.84511918e-02 6.11079633e-01 -3.10617775e-01
7.82223761e-01 -2.15786862e+00 -4.95447479e-02 2.01833814e-01
1.16395772e-01 1.46613508e-01 -2.07704440e-01 2.58957982e-01
6.16714321e-02 3.93796772e-01 -7.20393300e-01 -2.50851005e-01
1.02822624e-01 3.74488235e-01 -4.22778845e-01 2.04340369e-01
4.97902721e-01 1.03328562e+00 -1.04817533e+00 -7.39571214e-01
-2.76234411e-02 3.29350412e-01 -6.73014104e-01 2.72752978e-02
-8.18298876e-01 2.68104792e-01 -5.28663814e-01 7.48970389e-01
1.71345919e-01 -4.70600754e-01 4.18308116e-02 7.03428537e-02
1.34817779e-01 -2.96192989e-02 -9.22029316e-01 1.73171234e+00
-3.18204075e-01 4.75770026e-01 -1.40776619e-01 -1.02134633e+00
6.88128889e-01 3.58179629e-01 5.89597225e-01 -3.30229938e-01
1.45741373e-01 5.35752416e-01 -1.98980719e-01 -3.95482749e-01
5.44167578e-01 -5.23826420e-01 -4.12853688e-01 4.93821502e-01
4.16079581e-01 -5.54369569e-01 3.37055713e-01 4.81247097e-01
9.62538779e-01 5.69492877e-01 1.60197586e-01 -1.71212494e-01
2.06983641e-01 5.85916519e-01 3.20536852e-01 8.89508426e-01
-2.57253051e-01 9.49473917e-01 3.16018015e-01 -6.87104464e-02
-1.43360746e+00 -1.14865756e+00 -1.77427068e-01 1.30485702e+00
1.98244572e-01 -7.12041035e-02 -9.05155241e-01 -8.43247950e-01
7.71885738e-03 7.64695406e-01 -5.64500630e-01 1.26663670e-01
-3.09798181e-01 -3.20884556e-01 7.89701045e-01 6.23908401e-01
1.99364364e-01 -1.13392818e+00 -7.15713382e-01 1.80722535e-01
1.53424412e-01 -1.03559697e+00 -5.25046408e-01 2.04245791e-01
-5.15446126e-01 -9.61722493e-01 -1.07295585e+00 -1.10745847e+00
9.29373205e-01 1.36100054e-01 1.33401620e+00 3.81257161e-02
-5.53702235e-01 7.74220347e-01 -4.26219165e-01 -2.92335451e-01
-5.24373353e-01 2.65385419e-01 -2.83190370e-01 -4.38963771e-02
4.97444004e-01 -6.19282126e-01 -3.83913904e-01 2.85309881e-01
-1.30254459e+00 1.91066802e-01 6.62433624e-01 7.85468161e-01
7.12099075e-01 -4.71627861e-01 8.83826852e-01 -1.46043301e+00
2.44476750e-01 -6.01033866e-01 -5.44742823e-01 2.98352897e-01
-5.56783020e-01 7.16559067e-02 3.76075596e-01 -5.77660859e-01
-9.19549525e-01 4.44449067e-01 1.14388503e-01 -5.33659339e-01
-3.44245374e-01 3.99065018e-01 -1.94547996e-01 1.11762002e-01
8.31648052e-01 1.91647232e-01 -1.53030321e-01 -3.28446299e-01
1.13214612e+00 6.38615370e-01 8.68736744e-01 -7.04754293e-01
9.32354450e-01 4.71045792e-01 -4.90601331e-01 -6.78008139e-01
-1.00939095e+00 -7.70839095e-01 -8.19849551e-01 8.78429785e-02
1.04790950e+00 -1.04999423e+00 2.09841937e-01 2.06642766e-02
-8.26605558e-01 -3.68320167e-01 -8.49643350e-01 -1.57543812e-02
-8.96273434e-01 1.29880562e-01 -3.71171296e-01 -3.99389178e-01
-1.65846214e-01 -9.10147727e-01 1.05764079e+00 3.36531997e-01
-3.26535940e-01 -9.05719638e-01 -2.58794874e-01 3.91436934e-01
2.96682447e-01 2.35837460e-01 9.21882153e-01 -1.05838394e+00
-8.69892538e-01 -2.85596102e-01 -3.71659130e-01 3.06217879e-01
-1.14251394e-02 -1.93729386e-01 -1.18464327e+00 -1.95830405e-01
-5.00673890e-01 -6.75271153e-01 7.77067780e-01 8.86227712e-02
9.03225183e-01 -2.44880076e-02 -5.37575305e-01 5.72479129e-01
1.62578940e+00 -4.03042361e-02 4.62649614e-01 2.76906431e-01
7.66755044e-01 7.43342519e-01 4.45976973e-01 1.53320730e-02
1.69370815e-01 4.64428395e-01 -7.81127065e-02 -2.02612728e-01
-3.53955925e-01 -4.72317040e-01 -9.66248587e-02 7.04329371e-01
2.38300726e-01 -1.32645965e-01 -1.15411043e+00 1.15449309e+00
-1.59542322e+00 -7.93123901e-01 3.03764462e-01 1.97833776e+00
1.16942990e+00 2.77936131e-01 1.27498969e-01 -3.64395142e-01
7.02823937e-01 6.19097427e-02 -6.90814495e-01 -1.15918301e-01
3.60083245e-02 5.00217676e-01 7.53677607e-01 3.53781044e-01
-1.19862425e+00 1.31201303e+00 6.57192421e+00 8.93609703e-01
-7.38370717e-01 2.14069635e-01 6.32564843e-01 2.72905268e-02
-6.38830543e-01 2.73293942e-01 -7.23562896e-01 1.89730719e-01
9.91458952e-01 -3.61771584e-01 2.63793558e-01 9.67887938e-01
-3.52704018e-01 -1.67474675e-03 -1.30365312e+00 7.39687800e-01
4.41421568e-01 -1.33880591e+00 1.40906155e-01 -3.73817049e-02
1.06575823e+00 1.75620526e-01 5.05471192e-02 4.38032269e-01
5.82655668e-01 -1.06888688e+00 7.17416108e-01 2.95522809e-01
1.24471962e+00 -4.79462862e-01 4.06023622e-01 2.30177581e-01
-9.07010674e-01 -6.62886398e-03 -2.85769820e-01 1.88759103e-01
1.98617145e-01 1.32571310e-01 -1.19360018e+00 3.20194900e-01
2.71444440e-01 5.56727946e-01 -5.69093406e-01 1.04168129e+00
-2.23337680e-01 4.62968916e-01 -3.27149719e-01 1.51072621e-01
5.12240767e-01 -3.12176906e-02 2.30581462e-01 1.21699536e+00
1.10382691e-01 -2.61352696e-02 4.43321884e-01 1.14051747e+00
-4.81661588e-01 1.90518156e-01 -7.06824660e-01 -3.30222338e-01
5.93756437e-01 1.06002641e+00 -1.13676870e+00 -8.03291023e-01
-5.59311867e-01 9.92435217e-01 2.09752038e-01 3.21900606e-01
-6.64030910e-01 -5.42223275e-01 3.99741918e-01 1.97143704e-01
6.19680226e-01 5.12091257e-02 -3.24716866e-01 -1.05079353e+00
-2.21958920e-01 -7.72949636e-01 3.21777254e-01 -8.65534008e-01
-1.18506241e+00 3.93456966e-01 2.17997998e-01 -9.91141021e-01
-2.24200100e-01 -4.30492640e-01 -2.91937858e-01 8.48778427e-01
-1.44576311e+00 -1.61202395e+00 -7.40924254e-02 3.98133397e-01
8.21720541e-01 -4.29991446e-02 8.96582067e-01 2.37694457e-01
-2.92454332e-01 4.89215761e-01 -5.71514526e-03 2.50711381e-01
6.02076650e-01 -1.42732286e+00 5.15135884e-01 7.30606854e-01
5.22583246e-01 5.33103704e-01 5.77458441e-01 -6.40521646e-01
-8.69369030e-01 -1.16945016e+00 6.94423676e-01 -8.56796682e-01
7.61012256e-01 -4.24834102e-01 -7.94836760e-01 1.00235593e+00
-1.11564144e-01 3.21124159e-02 7.78060377e-01 -3.27479057e-02
-6.73573256e-01 2.97300249e-01 -1.00750089e+00 6.07496858e-01
1.00128293e+00 -7.46338844e-01 -9.26614344e-01 3.84384006e-01
1.03130305e+00 -7.04091415e-02 -7.09758759e-01 1.41884774e-01
3.30307066e-01 -5.49299300e-01 8.93127263e-01 -6.78541481e-01
5.54137468e-01 -2.45120272e-01 -1.85458571e-01 -8.64096403e-01
1.14366123e-02 -5.38187861e-01 3.86179566e-01 1.65224230e+00
8.18346083e-01 -3.91942888e-01 1.01670325e+00 8.91747832e-01
-1.87974706e-01 -5.94351351e-01 -6.87618852e-01 -8.34085882e-01
1.54385164e-01 -4.61772412e-01 4.37424183e-01 1.08525002e+00
-1.87741578e-01 4.59778965e-01 4.07491531e-03 -1.90687142e-02
4.15487409e-01 1.55911773e-01 8.13038886e-01 -1.04565775e+00
-4.77457255e-01 -3.29276443e-01 -5.57540357e-01 -1.01026857e+00
2.10040823e-01 -1.16802049e+00 4.53142703e-01 -1.45327938e+00
3.19136560e-01 -1.02701735e+00 -3.40415299e-01 7.24148810e-01
-5.47238216e-02 7.20728278e-01 2.77870357e-01 6.61850214e-01
-7.92255044e-01 1.21333316e-01 1.03997612e+00 -2.56279141e-01
-3.13171387e-01 -2.35950708e-01 -9.79413629e-01 7.52632022e-01
5.26599884e-01 -5.82471132e-01 -7.65887022e-01 -4.19622034e-01
-8.15726351e-03 -3.76186579e-01 2.31284097e-01 -1.03926075e+00
1.60835102e-01 9.23664961e-03 1.10977739e-01 -1.77399248e-01
2.60907263e-01 -6.51350915e-01 3.74788269e-02 -2.51929485e-03
-5.86220086e-01 -3.94833416e-01 1.26647865e-02 7.35662758e-01
-1.16625793e-01 -4.62182671e-01 8.03994000e-01 -5.38203597e-01
-1.05717409e+00 2.31284499e-01 -1.71912938e-01 5.48181415e-01
1.24819040e+00 -5.28454721e-01 5.19870734e-03 -1.37561828e-01
-8.07892978e-01 1.76844448e-01 1.03488266e+00 1.96880594e-01
2.91014940e-01 -8.41409683e-01 -3.68509263e-01 1.20595753e-01
4.52457935e-01 1.73040420e-01 -6.52593076e-02 6.11144722e-01
-6.03043020e-01 2.06719697e-01 -4.30353731e-02 -6.62700236e-01
-8.57869983e-01 8.79148364e-01 2.22775623e-01 4.29695323e-02
-8.94292235e-01 1.20490944e+00 3.86768788e-01 -5.55282354e-01
1.81619167e-01 -1.78811029e-02 1.36726111e-01 1.93023711e-01
3.63403916e-01 -3.51658404e-01 -2.35965103e-01 -6.72226846e-01
-6.89602420e-02 5.10636449e-01 -2.24241093e-01 -3.59690011e-01
1.35388243e+00 4.49516475e-02 2.62478322e-01 3.48355800e-01
1.10956156e+00 -8.50968882e-02 -1.47258306e+00 -2.67403394e-01
4.61097002e-01 -3.70805085e-01 -2.37141520e-01 -8.23390961e-01
-7.78476417e-01 6.73973680e-01 3.03373933e-01 1.94172740e-01
9.22482550e-01 5.56366265e-01 9.68942761e-01 2.55447865e-01
5.20172417e-01 -1.08299899e+00 -4.18967456e-02 1.06100917e-01
1.93262309e-01 -1.19030952e+00 -1.78807691e-01 -7.05957532e-01
-7.28642523e-01 7.69471109e-01 3.75887334e-01 -1.66569173e-01
3.68025005e-01 4.20202464e-02 1.85551316e-01 -3.14903051e-01
-3.85184765e-01 -6.91054761e-01 1.54730439e-01 1.02019286e+00
1.62063286e-01 1.23218864e-01 3.31371985e-02 3.31844836e-01
-1.50650024e-01 -5.80454357e-02 5.21170676e-01 1.04921365e+00
-6.79670751e-01 -1.08851564e+00 -1.33283928e-01 5.15617013e-01
-3.55020314e-01 -3.08487594e-01 -3.99399251e-01 8.98763061e-01
3.38330358e-01 5.62019527e-01 1.39095128e-01 -9.49687809e-02
1.46997407e-01 4.53640997e-01 2.47561157e-01 -1.20303619e+00
-3.69068772e-01 7.36920461e-02 1.34879366e-01 -2.37622663e-01
-6.00216985e-01 -6.40415132e-01 -1.19598794e+00 3.02952319e-01
-4.63431120e-01 1.25211403e-01 5.91864407e-01 9.58329141e-01
3.07680130e-01 3.46156955e-01 1.86186150e-01 -8.83622646e-01
-5.29641569e-01 -5.75934231e-01 -5.65006495e-01 8.30079257e-01
1.02371261e-01 -6.07028484e-01 -2.80276239e-01 5.77069402e-01]
|
[9.701573371887207, 0.8656459450721741]
|
f5e61680-02b4-425d-9347-4e4c741c74f7
|
skin-lesion-classification-using-hybrid-deep
|
1702.08434
| null |
http://arxiv.org/abs/1702.08434v2
|
http://arxiv.org/pdf/1702.08434v2.pdf
|
Skin Lesion Classification Using Hybrid Deep Neural Networks
|
Skin cancer is one of the major types of cancers with an increasing incidence
over the past decades. Accurately diagnosing skin lesions to discriminate
between benign and malignant skin lesions is crucial to ensure appropriate
patient treatment. While there are many computerised methods for skin lesion
classification, convolutional neural networks (CNNs) have been shown to be
superior over classical methods. In this work, we propose a fully automatic
computerised method for skin lesion classification which employs optimised deep
features from a number of well-established CNNs and from different abstraction
levels. We use three pre-trained deep models, namely AlexNet, VGG16 and
ResNet-18, as deep feature generators. The extracted features then are used to
train support vector machine classifiers. In the final stage, the classifier
outputs are fused to obtain a classification. Evaluated on the 150 validation
images from the ISIC 2017 classification challenge, the proposed method is
shown to achieve very good classification performance, yielding an area under
receiver operating characteristic curve of 83.83% for melanoma classification
and of 97.55% for seborrheic keratosis classification.
|
['Amirreza Mahbod', 'Isabella Ellinger', 'Rupert Ecker', 'Gerald Schaefer', 'Chunliang Wang']
|
2017-02-27
| null | null | null | null |
['skin-lesion-classification']
|
['medical']
|
[ 6.01330757e-01 -3.25315334e-02 -1.32308483e-01 -1.42543316e-01
-8.35499406e-01 -1.94116458e-01 8.22826385e-01 4.60350364e-01
-7.13255644e-01 6.90618455e-01 -3.67388502e-03 -3.65051329e-01
-1.95935085e-01 -7.37194598e-01 -4.23082486e-02 -8.94323945e-01
7.84959830e-03 -1.58883601e-01 1.48173943e-01 -2.37015009e-01
2.26545930e-01 8.00012052e-01 -1.63994598e+00 4.52761441e-01
9.39454079e-01 1.46471500e+00 -3.24532777e-01 1.16536415e+00
5.55576533e-02 5.57908654e-01 -5.32136142e-01 -4.88359272e-01
7.25357160e-02 -2.96713769e-01 -9.32575285e-01 1.38592452e-01
3.85294557e-01 2.19622806e-01 1.55961150e-02 8.64302278e-01
6.21389687e-01 -3.79515618e-01 8.25074017e-01 -6.56020880e-01
-7.42120594e-02 -4.44005802e-02 -3.61488461e-01 1.78145975e-01
1.71231344e-01 2.18778551e-01 7.43475795e-01 -6.05737686e-01
6.58100307e-01 5.81138611e-01 9.51287448e-01 6.86484814e-01
-9.91657674e-01 -2.25587621e-01 -5.10453820e-01 2.23467365e-01
-1.32434249e+00 -2.96690315e-01 1.43750504e-01 -4.59098577e-01
1.01604652e+00 5.05066633e-01 8.68161559e-01 1.04262948e+00
3.34295601e-01 4.41239715e-01 1.49997687e+00 -5.98751545e-01
2.45990440e-01 3.61839473e-01 -8.10841769e-02 9.01749909e-01
3.54076654e-01 1.24567896e-01 -1.53722435e-01 -6.14350960e-02
5.81742167e-01 1.11544468e-02 -2.07491800e-01 1.34235829e-01
-7.10899949e-01 8.67491841e-01 9.72979724e-01 4.38285559e-01
-6.70537114e-01 5.94168007e-02 5.63872755e-01 2.35134184e-01
4.12011802e-01 3.56396735e-01 -1.36559635e-01 1.72203705e-01
-6.88086808e-01 8.22469871e-03 6.77662730e-01 1.54718995e-01
3.75934660e-01 -2.35582918e-01 -2.06181332e-01 1.05302727e+00
7.25552216e-02 1.13558136e-01 7.95166969e-01 -1.17923245e-02
-9.76037327e-03 9.95744944e-01 -3.65186363e-01 -6.56111181e-01
-6.52373374e-01 -6.42937422e-01 -1.44785440e+00 4.89983350e-01
4.24160123e-01 -9.37911216e-03 -1.45872104e+00 1.14218307e+00
1.35597333e-01 -3.43440250e-02 3.84229422e-01 7.79386938e-01
7.94264138e-01 1.61350325e-01 5.14164865e-01 3.60033423e-01
1.45687687e+00 -6.33222878e-01 -2.88301647e-01 9.35083851e-02
8.43610942e-01 -6.73444748e-01 4.58770484e-01 5.19691527e-01
-8.24149370e-01 -3.59549403e-01 -1.11257946e+00 5.34845814e-02
-7.78234541e-01 4.77392375e-01 6.47718787e-01 9.07123566e-01
-1.20188057e+00 4.60837483e-01 -6.67251945e-01 -7.90600479e-01
7.86509812e-01 5.08447587e-01 -7.07879543e-01 -7.91117847e-02
-1.16428208e+00 1.08266699e+00 4.95440990e-01 2.43961453e-01
-5.80286682e-01 -4.84143585e-01 -7.66238987e-01 -1.57128200e-01
-1.21990956e-01 -7.32557476e-01 9.77265298e-01 -1.36716878e+00
-1.37653875e+00 1.19277239e+00 1.43155113e-01 -8.45352232e-01
6.40688896e-01 1.22030318e-01 -5.79256952e-01 3.37081224e-01
-3.08281332e-01 5.36105454e-01 7.39954531e-01 -7.40222752e-01
-9.79587495e-01 -3.17536414e-01 -1.16067335e-01 1.68396384e-01
-5.21212339e-01 -1.37563810e-01 -1.36913478e-01 -2.91432768e-01
-1.98547125e-01 -8.58963490e-01 -5.40723205e-01 1.59186348e-01
-6.87782526e-01 -1.25953406e-01 4.35381681e-01 -8.85696054e-01
1.01669359e+00 -1.86918855e+00 7.87774622e-02 6.21915221e-01
2.60738105e-01 8.48409176e-01 -3.91259566e-02 3.61676842e-01
-2.04584286e-01 1.46836162e-01 -2.61558056e-01 -1.56716198e-01
-3.67858052e-01 -1.21455677e-01 4.89263386e-01 5.55666327e-01
7.04792917e-01 8.66368234e-01 -7.12415993e-01 -2.65790731e-01
5.88685215e-01 8.40085506e-01 1.21192724e-01 9.15765762e-03
4.14224304e-02 -6.11271933e-02 -2.06194192e-01 7.45011032e-01
4.40804899e-01 -1.72502920e-01 1.03771850e-03 -1.56104833e-01
2.82164305e-01 -5.80534041e-02 -7.97738254e-01 1.29113436e+00
-6.92336261e-01 8.19433391e-01 -1.59021933e-04 -8.18732500e-01
8.25448513e-01 4.71940607e-01 2.25918844e-01 -6.87338710e-01
3.98277402e-01 3.82620513e-01 1.71568796e-01 -6.45559311e-01
1.37614191e-01 -1.38007551e-01 1.95638850e-01 -1.54601946e-01
1.36292905e-01 -3.08226291e-02 2.54180372e-01 -1.02896012e-01
1.26612103e+00 -4.15748984e-01 7.85966575e-01 -1.76055849e-01
9.56953585e-01 1.14148252e-01 1.56815499e-02 2.64216900e-01
-1.48351088e-01 6.32568896e-01 3.43705744e-01 -7.42542684e-01
-7.70157158e-01 -7.46196330e-01 -3.80555242e-01 4.17449117e-01
-4.28440064e-01 -1.39348865e-01 -1.00429249e+00 -7.70685017e-01
-8.56496543e-02 1.87297449e-01 -9.76794541e-01 -2.16984034e-01
-8.85437056e-02 -9.20467973e-01 6.91479146e-01 4.85796332e-01
7.17592120e-01 -1.04431009e+00 -6.25301540e-01 8.77195224e-02
3.11736375e-01 -8.21162701e-01 3.06758642e-01 2.34932452e-01
-5.26153862e-01 -1.51331055e+00 -1.04487550e+00 -7.85064816e-01
9.39065754e-01 -1.02857210e-01 7.43808568e-01 3.51099938e-01
-1.27833915e+00 1.40743079e-02 -3.88302654e-01 -5.36290050e-01
-6.90211058e-01 4.47181374e-01 -2.91424870e-01 3.38172466e-01
4.93122160e-01 6.64502457e-02 -7.69966066e-01 -1.87614366e-01
-1.16618812e+00 1.44316122e-01 1.12649524e+00 9.89790261e-01
4.48864877e-01 1.41912684e-01 4.14995581e-01 -1.05781615e+00
7.06375480e-01 -3.65639269e-01 -1.68035284e-01 1.68452010e-01
-4.28424329e-01 -2.35490292e-01 7.27849483e-01 -4.26150970e-02
-7.73448646e-01 3.51381719e-01 -6.87973559e-01 2.50425220e-01
-7.06001759e-01 6.97671294e-01 2.30056047e-01 -4.84990656e-01
1.10858250e+00 1.84891388e-01 2.52075434e-01 -1.80771217e-01
-3.66991051e-02 9.03180778e-01 4.47933376e-01 2.52638280e-01
6.02222323e-01 3.11657101e-01 4.05022711e-01 -1.19555998e+00
-6.96875334e-01 -7.05413461e-01 -5.76491535e-01 -2.36511156e-01
9.67974961e-01 -7.39759624e-01 -3.85901839e-01 1.05548632e+00
-7.36113191e-01 -1.64744437e-01 1.91677734e-02 9.34285223e-02
-2.22311988e-01 1.04363307e-01 -5.12162626e-01 -6.70373738e-01
-7.16845691e-01 -9.10699308e-01 1.01494193e+00 7.08877504e-01
-4.36283052e-01 -1.46642148e+00 -6.14341684e-02 1.29157230e-01
7.19588697e-01 8.46224964e-01 7.99820185e-01 -6.66598082e-01
2.70420909e-02 -1.01423931e+00 -2.50421643e-01 5.59491217e-01
4.12415594e-01 2.77964532e-01 -1.17874157e+00 -3.94435734e-01
-6.80673182e-01 -4.04284954e-01 1.25812685e+00 2.87582874e-01
1.32362306e+00 -7.17050061e-02 -4.26374525e-01 7.22357273e-01
1.95467401e+00 -5.08916266e-02 9.52291191e-01 3.91192019e-01
3.59321922e-01 5.82772493e-01 1.83348283e-01 5.44022694e-02
-1.07552253e-01 2.07296774e-01 7.27203071e-01 -6.30573273e-01
-3.14072043e-01 1.05321370e-01 -2.19085068e-01 3.23499925e-02
-3.98171216e-01 -3.61051448e-02 -1.08228695e+00 6.00915790e-01
-1.24534452e+00 -6.14591956e-01 -2.49613360e-01 2.17985201e+00
6.32697344e-01 2.18824863e-01 1.77454222e-02 7.27009237e-01
6.62766576e-01 -1.41062185e-01 -3.24107260e-01 -5.82400680e-01
-1.71200901e-01 8.64962161e-01 6.54985845e-01 2.03768536e-01
-1.39820671e+00 7.47770071e-01 5.35481501e+00 7.27742493e-01
-1.56128609e+00 -3.65981400e-01 8.16524029e-01 3.52977365e-01
3.23071986e-01 -5.54090023e-01 -4.68163699e-01 1.84544951e-01
1.01412475e+00 4.28256067e-03 -1.12157620e-01 4.71502841e-01
-8.83419588e-02 -4.14110720e-01 -7.91000426e-01 8.25286508e-01
1.74162313e-01 -1.42812467e+00 1.23077640e-02 1.74769312e-01
7.76320636e-01 -1.89072162e-01 2.96984345e-01 -1.84421316e-02
-5.61502911e-02 -1.57592523e+00 -8.94159973e-02 7.40685046e-01
1.17431414e+00 -8.15193832e-01 1.10373449e+00 1.87004328e-01
-8.35261822e-01 -1.51576206e-01 -2.14109987e-01 1.83356434e-01
-3.34553689e-01 6.01294816e-01 -1.09567881e+00 6.58376276e-01
4.89911854e-01 7.30789840e-01 -1.04617906e+00 1.42184925e+00
-1.81061730e-01 5.64799190e-01 -2.38775983e-01 -3.77393961e-01
5.05782306e-01 4.11671817e-01 8.62516984e-02 1.34795225e+00
1.51680380e-01 -2.85180926e-01 -4.62565929e-01 2.46862397e-01
1.70992434e-01 2.54503071e-01 -4.71755564e-01 -1.98293731e-01
-9.82630029e-02 1.83271003e+00 -7.25433052e-01 -1.55203655e-01
-1.50098547e-01 9.73481178e-01 9.08134058e-02 1.25810370e-01
-4.30355400e-01 -7.48191476e-01 5.67203760e-01 2.11902499e-01
6.22596592e-02 3.35950553e-01 -2.03130528e-01 -7.73937285e-01
-2.36789927e-01 -7.75037467e-01 5.17662644e-01 -3.68171722e-01
-1.28720224e+00 9.12911057e-01 -6.13452375e-01 -1.08096981e+00
-2.53313810e-01 -1.19472861e+00 -9.57997203e-01 1.28144205e+00
-1.77788079e+00 -1.38025093e+00 -8.55765581e-01 3.36209029e-01
3.73245895e-01 -3.44959527e-01 1.33728492e+00 -1.55289412e-01
-6.25155210e-01 5.84580660e-01 5.67432940e-02 3.83544505e-01
4.99565721e-01 -1.47622168e+00 3.75644833e-01 5.85245371e-01
-2.88118750e-01 1.89853862e-01 1.19581416e-01 -2.78702796e-01
-1.05099106e+00 -1.43310237e+00 7.42635250e-01 1.14322796e-01
4.78144228e-01 -1.35957330e-01 -6.48020506e-01 -7.20972791e-02
2.29818061e-01 1.78495139e-01 1.02167130e+00 -2.24873409e-01
-3.38424891e-01 -1.05929874e-01 -1.41173422e+00 4.08574969e-01
3.70130002e-01 -3.13009888e-01 3.03379856e-02 2.72591442e-01
-1.99664906e-01 -3.11588377e-01 -1.08873940e+00 4.99993801e-01
6.57773435e-01 -1.08031237e+00 8.87790084e-01 -6.90494835e-01
8.08974087e-01 1.05427921e-01 1.51543602e-01 -1.52546835e+00
-2.53208429e-01 -1.76339656e-01 2.37018391e-01 7.78948784e-01
5.46465576e-01 -7.71383047e-01 1.07020700e+00 3.00613999e-01
1.72663033e-01 -1.39464724e+00 -7.49589086e-01 -3.73281181e-01
7.90026858e-02 -1.61752310e-02 3.01256776e-01 6.03187680e-01
-3.30928266e-01 2.02841744e-01 1.88210681e-01 -9.29870009e-02
5.28872490e-01 -4.84654456e-01 4.80373323e-01 -1.29011381e+00
1.78705886e-01 -8.39598715e-01 -1.00851703e+00 1.93818714e-02
-1.26610860e-01 -9.56759095e-01 -2.52749860e-01 -1.85641241e+00
9.82757434e-02 -3.76717001e-01 -6.60389543e-01 4.93616730e-01
-2.83722132e-01 6.76294327e-01 -2.02152088e-01 -2.93168008e-01
-9.88043025e-02 -1.62466392e-01 9.42358434e-01 -2.00120643e-01
1.60780235e-03 3.47234011e-01 -7.12279856e-01 6.06899321e-01
1.07164598e+00 2.40581423e-01 -4.15532179e-02 1.17876930e-02
-6.56902194e-02 -2.25676164e-01 6.73764288e-01 -1.45089626e+00
1.39010325e-01 3.78617048e-02 9.41428542e-01 -1.60977408e-01
3.03423882e-01 -6.45919502e-01 2.34486610e-02 9.99649465e-01
-5.75035810e-01 -5.62414587e-01 3.63726705e-01 4.08640504e-01
-3.62460911e-01 -2.83380955e-01 1.12096369e+00 -2.70650983e-02
-8.55400562e-01 4.71140712e-01 -4.89452124e-01 -5.41522801e-01
1.37453842e+00 -3.97798449e-01 -2.47170612e-01 -4.05064449e-02
-7.78189898e-01 -1.26255125e-01 3.54124665e-01 3.20419580e-01
6.82609618e-01 -1.10943747e+00 -8.67317319e-01 2.26458073e-01
4.61918980e-01 -5.06261811e-02 4.24578846e-01 7.44048655e-01
-9.35361326e-01 5.20829797e-01 -3.90873671e-01 -5.62112272e-01
-1.62459671e+00 -8.12597424e-02 7.86950231e-01 -3.52137864e-01
-3.14556003e-01 1.04577780e+00 -5.22484004e-01 -1.01883546e-01
1.94438279e-01 -2.70066470e-01 -5.50813377e-01 -1.23723326e-02
5.86404443e-01 2.42668450e-01 5.53129911e-01 -5.77215970e-01
-2.05615237e-01 4.94447529e-01 -3.83008957e-01 2.12471768e-01
1.21347547e+00 5.50591409e-01 -1.04878850e-01 -9.96036455e-02
1.37143159e+00 -4.69936430e-01 -7.70234346e-01 -3.61365862e-02
4.31005359e-02 -2.51958519e-01 1.31188139e-01 -1.33400059e+00
-9.10147727e-01 1.08706224e+00 9.65399206e-01 4.85549361e-01
1.54297376e+00 -3.95168483e-01 6.41039431e-01 1.91478133e-01
2.90401150e-02 -1.00474155e+00 -2.74136484e-01 9.27448198e-02
7.18326628e-01 -1.27160060e+00 -3.61433737e-02 -5.53687572e-01
-5.19513190e-01 1.47477019e+00 5.79933286e-01 -3.43381763e-01
5.16063929e-01 1.68910295e-01 3.04762602e-01 -1.72635183e-01
-7.21698165e-01 -6.03021502e-01 4.38522369e-01 7.42594540e-01
5.24811029e-01 2.01960906e-01 -3.37425590e-01 2.32492983e-01
-2.13304628e-02 4.89423759e-02 3.78309757e-01 8.23621273e-01
-3.37295592e-01 -1.02882326e+00 -1.09101541e-01 1.07590222e+00
-8.17531228e-01 8.88148546e-02 -8.78801584e-01 9.82965171e-01
1.42134547e-01 6.26910925e-01 -7.75572881e-02 -2.51333237e-01
2.16281101e-01 9.94059071e-02 4.60912973e-01 -6.31518304e-01
-9.62506413e-01 -4.52184528e-01 2.68555522e-01 -2.74691939e-01
-3.62816811e-01 -2.94982105e-01 -8.37697089e-01 8.21672007e-02
-3.01304609e-01 -1.10147819e-01 1.17924380e+00 7.84090936e-01
8.89036655e-02 6.27883434e-01 6.36307418e-01 -5.83715141e-01
-5.43805420e-01 -1.05055833e+00 -4.73229885e-01 5.53293645e-01
3.12092930e-01 -3.59596074e-01 -1.46376640e-01 -7.80691439e-03]
|
[15.677809715270996, -2.994845390319824]
|
aced68e1-4350-47cc-b1db-91be5780f822
|
learning-knowledge-graph-based-world-models
|
2106.09608
| null |
https://arxiv.org/abs/2106.09608v2
|
https://arxiv.org/pdf/2106.09608v2.pdf
|
Learning Knowledge Graph-based World Models of Textual Environments
|
World models improve a learning agent's ability to efficiently operate in interactive and situated environments. This work focuses on the task of building world models of text-based game environments. Text-based games, or interactive narratives, are reinforcement learning environments in which agents perceive and interact with the world using textual natural language. These environments contain long, multi-step puzzles or quests woven through a world that is filled with hundreds of characters, locations, and objects. Our world model learns to simultaneously: (1) predict changes in the world caused by an agent's actions when representing the world as a knowledge graph; and (2) generate the set of contextually relevant natural language actions required to operate in the world. We frame this task as a Set of Sequences generation problem by exploiting the inherent structure of knowledge graphs and actions and introduce both a transformer-based multi-task architecture and a loss function to train it. A zero-shot ablation study on never-before-seen textual worlds shows that our methodology significantly outperforms existing textual world modeling techniques as well as the importance of each of our contributions.
|
['Mark O. Riedl', 'Prithviraj Ammanabrolu']
|
2021-06-17
| null |
http://proceedings.neurips.cc/paper/2021/hash/1e747ddbea997a1b933aaf58a7953c3c-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/1e747ddbea997a1b933aaf58a7953c3c-Paper.pdf
|
neurips-2021-12
|
['text-based-games']
|
['playing-games']
|
[ 4.76208270e-01 5.29961765e-01 2.11708307e-01 2.10605055e-01
-6.91376686e-01 -8.18391562e-01 1.23305738e+00 7.80195091e-03
-4.02126819e-01 1.01922750e+00 6.84995234e-01 -2.12836027e-01
-1.06856629e-01 -1.50108516e+00 -8.75423551e-01 -2.60860413e-01
-2.30865747e-01 1.19177008e+00 5.56219876e-01 -7.61962593e-01
1.84397236e-01 -1.10812001e-01 -1.30260980e+00 5.04754305e-01
5.64726532e-01 2.20310941e-01 3.61177832e-01 1.03537035e+00
-1.76046729e-01 2.00239086e+00 -9.36183035e-01 -6.86012626e-01
1.47113502e-01 -7.63231099e-01 -1.06885481e+00 3.99327427e-02
-2.24367902e-01 -3.28018248e-01 -7.47485638e-01 7.05987334e-01
3.01652342e-01 7.50992775e-01 5.97812355e-01 -1.21834826e+00
-8.69247079e-01 9.91162419e-01 -1.54809415e-01 -1.14049651e-02
1.00897598e+00 7.16629505e-01 9.09712970e-01 -2.27651492e-01
1.25650454e+00 1.26061273e+00 3.72803628e-01 6.51869953e-01
-1.12298977e+00 -2.34328449e-01 2.54780829e-01 4.35498655e-01
-9.27056968e-01 -3.51949036e-01 7.46073306e-01 -4.45939064e-01
1.88609302e+00 -1.34645598e-02 9.01037872e-01 1.90427303e+00
4.01319295e-01 7.36031830e-01 5.26295543e-01 -4.34822857e-01
4.30229008e-01 -4.25537944e-01 -6.34178817e-01 1.01576829e+00
-9.73368958e-02 3.74234736e-01 -1.10364115e+00 -1.27267152e-01
1.11135614e+00 -5.62845409e-01 2.92197794e-01 -6.72215223e-01
-1.62739432e+00 7.84212530e-01 1.75577357e-01 3.80257592e-02
-4.76537108e-01 6.39382243e-01 5.65060377e-01 2.51338810e-01
4.55579340e-01 1.14037168e+00 -2.20754758e-01 -7.76330769e-01
-1.62763298e-01 7.83772290e-01 9.18708682e-01 1.13737381e+00
2.84951359e-01 2.91997880e-01 -1.43983275e-01 5.07510424e-01
3.53761315e-02 2.75892437e-01 7.16644466e-01 -8.16286385e-01
9.80126619e-01 5.77543855e-01 5.49823046e-01 -8.41509879e-01
-4.34949696e-01 -5.58246486e-02 -1.94637284e-01 2.80014008e-01
4.30477411e-01 -5.15762448e-01 -1.02913177e+00 1.89216506e+00
2.52829373e-01 6.18854403e-01 4.19914484e-01 4.36061889e-01
9.06793594e-01 8.16604435e-01 3.31101239e-01 1.68857485e-01
1.23115408e+00 -1.15348577e+00 -6.02293134e-01 -8.60143661e-01
7.11570680e-01 1.00359209e-01 1.14969337e+00 1.70691073e-01
-1.47488225e+00 -3.57432067e-01 -1.13131320e+00 -1.04175314e-01
-8.25822175e-01 -5.56395113e-01 8.33161473e-01 3.41520369e-01
-1.18542814e+00 3.96967471e-01 -8.93644869e-01 -2.36561656e-01
3.34245205e-01 6.73807859e-02 -3.36802810e-01 1.90353736e-01
-1.49254298e+00 1.39565194e+00 9.84460354e-01 -2.98918784e-01
-1.40627801e+00 -3.06989342e-01 -1.53445077e+00 5.24169430e-02
9.81190979e-01 -1.23220026e+00 1.50917661e+00 -6.16636813e-01
-1.61823928e+00 7.00124919e-01 3.06130826e-01 -5.47665417e-01
6.96539223e-01 -7.17089772e-02 -2.42106616e-01 -2.82621950e-01
4.59912390e-01 3.38907212e-01 4.41490829e-01 -1.03986549e+00
-8.72208953e-01 -2.40451142e-01 5.48680246e-01 7.60427713e-01
2.84432858e-01 -1.32897377e-01 -3.70637447e-01 -5.85681736e-01
-2.72301584e-01 -6.82959735e-01 -6.87943399e-01 -7.14264691e-01
-4.91104186e-01 -1.26394063e-01 3.36147577e-01 -5.94070077e-01
8.31229448e-01 -1.70323670e+00 5.90669394e-01 -3.75750549e-02
4.21088576e-01 -1.37409225e-01 -6.97902679e-01 9.94152188e-01
1.23219937e-01 6.49105087e-02 1.21131830e-01 -2.09638283e-01
2.56750494e-01 1.04756512e-01 -3.58400732e-01 -1.28586516e-01
1.70226395e-01 1.51747370e+00 -1.51461697e+00 -2.69483626e-01
4.79000926e-01 1.73045337e-01 -3.91901046e-01 1.87677801e-01
-1.04208827e+00 2.77924895e-01 -7.15190947e-01 7.98274502e-02
-3.25124025e-01 -1.73599944e-01 4.71652240e-01 4.88837063e-01
1.98808879e-01 5.41530252e-01 -9.60017622e-01 2.11599588e+00
-6.27123296e-01 4.95503098e-01 -6.84776127e-01 -5.07711053e-01
6.73613727e-01 4.84658539e-01 2.57804632e-01 -9.64459479e-01
-1.24247938e-01 -2.44645074e-01 4.21504583e-03 -8.38784516e-01
7.38120914e-01 -2.98592478e-01 -5.71745276e-01 7.00334013e-01
1.21071741e-01 -6.74742639e-01 5.31017542e-01 4.34581697e-01
1.65712285e+00 8.81579161e-01 6.18566930e-01 4.33368295e-01
-1.79612368e-01 3.47953767e-01 3.23930949e-01 1.16996753e+00
1.63798124e-01 1.45020023e-01 7.00898886e-01 -8.30837429e-01
-1.27855670e+00 -1.26068962e+00 9.54437256e-01 1.42575204e+00
1.57844529e-01 -4.76490319e-01 -5.31907558e-01 -4.42544639e-01
-4.21043903e-01 1.23080480e+00 -9.96781170e-01 -4.67921585e-01
-7.73776352e-01 -4.73898232e-01 7.67165780e-01 6.99131429e-01
4.77023691e-01 -1.98013127e+00 -1.24914634e+00 6.61146700e-01
-5.59351802e-01 -1.19156873e+00 -1.37049228e-01 2.10569128e-01
-4.25680459e-01 -1.12497449e+00 -6.68184236e-02 -8.44162583e-01
2.72057533e-01 -3.42788368e-01 1.68801677e+00 -3.14288348e-01
-2.13108912e-01 4.98077124e-01 -5.91226339e-01 -5.27967691e-01
-6.32799149e-01 1.52795970e-01 -1.93188012e-01 -3.96866590e-01
2.17182368e-01 -8.59219491e-01 9.99108404e-02 -2.45566860e-01
-7.88003206e-01 7.58005619e-01 3.14474672e-01 8.98592353e-01
1.45906180e-01 2.31347665e-01 5.22517264e-01 -1.09417117e+00
1.12656116e+00 -6.21612847e-01 -4.17454511e-01 3.65770131e-01
7.08082989e-02 1.41932487e-01 6.60463631e-01 -4.29929197e-01
-1.47275782e+00 1.71258189e-02 1.70378327e-01 3.07940811e-01
-1.35820165e-01 7.95202553e-01 -1.37996286e-01 4.76975173e-01
1.18078005e+00 5.94170272e-01 -5.94052553e-01 3.12081397e-01
8.00541878e-01 -5.41586950e-02 8.47077787e-01 -1.13311934e+00
9.31497216e-01 1.57456264e-01 -1.61311075e-01 -5.49522996e-01
-4.83663410e-01 -2.10618246e-02 -6.50906026e-01 -3.07579577e-01
8.46746325e-01 -7.39461660e-01 -8.01979423e-01 4.61379588e-01
-1.11648738e+00 -1.37688434e+00 -6.89761341e-01 3.27234482e-03
-1.39060986e+00 -7.30525926e-02 -5.70658565e-01 -8.80569816e-01
9.03626531e-02 -8.52404416e-01 1.01427865e+00 1.48957238e-01
-5.64896464e-01 -1.31120431e+00 4.64435726e-01 4.17485595e-01
1.41060576e-01 8.34300578e-01 1.05367398e+00 -6.01945758e-01
-5.68016887e-01 -3.86640392e-02 1.82443023e-01 -7.41276085e-01
3.12293142e-01 -3.10925126e-01 -5.49857318e-01 1.28768504e-01
-3.32776129e-01 -8.59194934e-01 3.18892419e-01 1.36161476e-01
8.49847376e-01 -4.74986047e-01 -3.37294161e-01 3.21787506e-01
1.18072402e+00 7.89614677e-01 1.02343774e+00 7.23361194e-01
5.70361376e-01 5.17712057e-01 3.79452825e-01 7.25356221e-01
6.88207269e-01 6.80369556e-01 6.04188263e-01 1.74447194e-01
-2.92328354e-02 -1.07630837e+00 3.05145532e-01 -6.04427829e-02
-3.57110530e-01 -6.60039246e-01 -1.25336683e+00 9.28769946e-01
-2.20010805e+00 -1.49042428e+00 3.01140547e-01 1.69064319e+00
1.06673276e+00 4.24320102e-01 5.16277887e-02 -4.77168202e-01
1.94202721e-01 6.10695958e-01 -9.20164168e-01 -3.03025484e-01
-1.51035292e-02 3.93127143e-01 8.77318718e-03 7.92678058e-01
-1.01891065e+00 1.62326825e+00 6.37112951e+00 6.27994716e-01
-3.30730110e-01 -5.66645637e-02 3.84198874e-01 -3.16726208e-01
-2.54596323e-01 -2.74343878e-01 -4.65143435e-02 7.94452205e-02
7.77118385e-01 -4.64733243e-01 9.38860416e-01 6.07183754e-01
2.01154098e-01 -6.06091053e-04 -1.25143135e+00 7.05645502e-01
6.46483600e-02 -1.62534928e+00 2.50105262e-01 -7.31871203e-02
7.56884396e-01 -6.39609843e-02 8.86381865e-02 7.98699975e-01
1.50592065e+00 -1.39603329e+00 1.05347705e+00 3.90671402e-01
8.52138460e-01 -5.56148350e-01 2.73981065e-01 5.81348360e-01
-1.25366819e+00 -1.76295087e-01 7.51736155e-03 -6.96914315e-01
2.58582056e-01 -4.44569468e-01 -1.21612930e+00 3.89263541e-01
2.15321988e-01 5.86188376e-01 -4.45954293e-01 6.94850504e-01
-6.53933167e-01 1.77210823e-01 1.44864425e-01 -3.78608316e-01
5.07400870e-01 -1.03119843e-01 5.89807332e-01 8.84859502e-01
1.05751283e-01 4.12922084e-01 1.83279261e-01 1.07927167e+00
-8.28666836e-02 -3.14992368e-01 -1.17502141e+00 -2.30654135e-01
4.44614917e-01 7.34435916e-01 -7.13225543e-01 -5.34789681e-01
-1.47309244e-01 1.15303791e+00 5.31606436e-01 6.06202543e-01
-8.52516472e-01 -3.24940473e-01 6.75246298e-01 -1.70863062e-01
-5.45986146e-02 -2.87429184e-01 -3.26445699e-01 -1.13636732e+00
-1.09459408e-01 -1.14322007e+00 3.02108496e-01 -1.32324088e+00
-8.78721476e-01 6.31755531e-01 -4.64886054e-02 -8.19667637e-01
-8.80515397e-01 -4.34543908e-01 -7.26606131e-01 7.24489629e-01
-7.58829236e-01 -1.49877059e+00 -1.89430788e-01 6.84944570e-01
1.13452113e+00 -4.86413300e-01 1.20196176e+00 -5.45438528e-01
-1.72604412e-01 2.00220365e-02 -2.28561506e-01 4.83907759e-01
8.43039248e-03 -1.61188233e+00 1.49714637e+00 8.85608852e-01
5.29063523e-01 3.42883945e-01 8.62643242e-01 -1.06712341e+00
-1.15843642e+00 -1.02644050e+00 6.72885776e-01 -1.12437868e+00
8.57452691e-01 -5.72995603e-01 -3.92645001e-01 1.38507140e+00
1.85963497e-01 -5.37299693e-01 5.49421787e-01 1.60637945e-01
-3.95397246e-01 6.17748022e-01 -1.00597966e+00 1.43813801e+00
1.55686736e+00 -5.90173841e-01 -9.70462143e-01 5.25558293e-01
8.53204012e-01 -9.16117966e-01 -4.32733446e-01 -2.42366165e-01
4.38739240e-01 -7.73669600e-01 1.05967462e+00 -1.39485312e+00
9.91618156e-01 -1.21870913e-01 2.09177628e-01 -1.93783033e+00
-4.15604562e-01 -9.16214049e-01 -1.75950825e-01 4.83665317e-01
6.30918384e-01 -5.06578565e-01 1.06823838e+00 8.27430189e-01
-1.08078659e-01 -2.34088942e-01 -8.42215061e-01 -3.87153536e-01
8.22113268e-03 -5.50113678e-01 7.48579741e-01 9.43480372e-01
6.46153748e-01 6.78696215e-01 -4.79540825e-01 -1.03894286e-01
3.88898492e-01 -2.39860639e-01 8.67804587e-01 -1.01600313e+00
-4.75838572e-01 -3.83601964e-01 -2.30403915e-01 -8.00273418e-01
2.17975467e-01 -5.01178682e-01 2.99325913e-01 -2.08366013e+00
1.56375304e-01 -1.61075830e-01 3.08779836e-01 7.58395731e-01
-6.60922974e-02 -1.61356047e-01 2.10971609e-01 -3.24074090e-01
-1.04039156e+00 6.48902118e-01 1.46105456e+00 -4.39281076e-01
-4.70869541e-01 -3.46756577e-01 -6.42028034e-01 7.36854196e-01
7.24879086e-01 -1.13929763e-01 -1.10024214e+00 -8.29962373e-01
1.08143675e+00 6.39552534e-01 2.28208840e-01 -9.45583642e-01
5.38134515e-01 -7.41158545e-01 3.43775600e-01 1.16397381e-01
7.66238093e-01 -4.81213957e-01 2.34361291e-01 3.16860914e-01
-6.18527234e-01 1.91326946e-01 3.93612027e-01 6.24547839e-01
-2.73413081e-02 8.00276250e-02 6.21938929e-02 -7.66499817e-01
-1.17539525e+00 1.12465352e-01 -7.94286549e-01 3.05790246e-01
1.34701550e+00 -4.72202659e-01 -6.44378662e-01 -9.09649432e-01
-9.04658198e-01 4.29116219e-01 2.56223708e-01 6.72876894e-01
8.70308578e-01 -1.21413243e+00 -9.17247772e-01 -5.68742752e-02
2.49164328e-01 1.84505537e-01 2.42995262e-01 -3.15848798e-01
-8.18890750e-01 1.46242738e-01 -5.54828882e-01 3.62716824e-01
-7.65492678e-01 6.24578595e-01 7.00214386e-01 -7.59074271e-01
-7.73129046e-01 9.31051016e-01 1.98741212e-01 -6.12100482e-01
6.76301196e-02 -2.24105328e-01 -5.98782003e-01 -2.40400150e-01
5.52136958e-01 1.70345262e-01 -3.04240465e-01 -3.40049505e-01
8.93555060e-02 -5.26393950e-02 3.51095609e-02 -7.48462617e-01
1.56580198e+00 2.40746439e-01 1.56148136e-01 5.34927130e-01
1.20240375e-01 -2.60219961e-01 -1.38425779e+00 -1.95854396e-01
7.77229965e-02 -3.47036123e-01 -5.80843389e-01 -1.29896903e+00
-3.05804551e-01 4.27337855e-01 -3.25768620e-01 2.82433510e-01
7.32411385e-01 7.55880028e-02 7.35345840e-01 6.50681436e-01
1.00709474e+00 -1.16557825e+00 8.25097322e-01 9.73683059e-01
1.00458324e+00 -8.39461684e-01 -4.06877398e-01 -1.41448170e-01
-9.84441876e-01 8.18209827e-01 9.67414796e-01 1.18008284e-02
-1.60643399e-01 4.04348612e-01 -1.89800888e-01 -4.38874930e-01
-1.23415279e+00 -3.47713441e-01 -2.41705939e-01 1.49222445e+00
-4.74430295e-03 2.46754497e-01 4.08467531e-01 5.70853531e-01
-7.47803390e-01 -1.02022253e-01 8.49914432e-01 8.07892025e-01
-2.13279858e-01 -9.15493965e-01 -1.47706538e-01 4.71023262e-01
8.56246427e-02 -1.56899214e-01 -7.60915518e-01 7.14933991e-01
9.02381167e-03 1.11106610e+00 -5.46178184e-02 -5.91534793e-01
4.50083166e-01 2.81781971e-01 6.62486970e-01 -1.11877811e+00
-6.12390757e-01 -6.72906935e-01 6.41856909e-01 -7.37177372e-01
4.11044760e-03 -5.27836621e-01 -1.21621609e+00 -1.80938587e-01
2.67047435e-01 -7.09701106e-02 1.55599549e-01 1.01099622e+00
-3.95497009e-02 1.04556406e+00 -7.24736825e-02 -7.99011886e-01
-1.74064904e-01 -9.41113949e-01 -4.51643646e-01 5.89133382e-01
6.73507303e-02 -6.23731613e-01 2.66077250e-01 2.05445409e-01]
|
[3.8079473972320557, 1.2841827869415283]
|
2ca3c165-6cae-4bad-b68b-0b4c369705c1
|
a-global-constraint-for-mining-sequential
|
1511.08350
| null |
http://arxiv.org/abs/1511.08350v1
|
http://arxiv.org/pdf/1511.08350v1.pdf
|
A global Constraint for mining Sequential Patterns with GAP constraint
|
Sequential pattern mining (SPM) under gap constraint is a challenging task.
Many efficient specialized methods have been developed but they are all
suffering from a lack of genericity. The Constraint Programming (CP) approaches
are not so effective because of the size of their encodings. In[7], we have
proposed the global constraint Prefix-Projection for SPM which remedies to this
drawback. However, this global constraint cannot be directly extended to
support gap constraint. In this paper, we propose the global constraint GAP-SEQ
enabling to handle SPM with or without gap constraint. GAP-SEQ relies on the
principle of right pattern extensions. Experiments show that our approach
clearly outperforms both CP approaches and the state-of-the-art cSpade method
on large datasets.
|
['Thierry Charnois', 'Samir Loudni', 'Amina Kemmar', 'Yahia Lebbah', 'Patrice Boizumault']
|
2015-11-26
| null | null | null | null |
['sequential-pattern-mining']
|
['natural-language-processing']
|
[ 5.24066031e-01 -4.91993222e-03 -3.58772844e-01 -2.21061885e-01
-7.05742463e-02 -2.67942220e-01 1.87369302e-01 5.45859262e-02
-1.82967186e-01 7.96024799e-01 -8.61601457e-02 -4.15572137e-01
-4.56728220e-01 -9.60708201e-01 -3.38886112e-01 -4.77534235e-01
-9.70708951e-02 4.23599750e-01 9.27765071e-01 -1.11034967e-01
6.05480969e-01 1.93455100e-01 -1.72110951e+00 5.09382308e-01
8.17909420e-01 9.23308909e-01 4.57040846e-01 -4.88332622e-02
-6.86983109e-01 4.91605364e-02 -1.80247232e-01 -2.46927634e-01
5.47709405e-01 -6.49198711e-01 -7.69228458e-01 1.31202668e-01
-2.24611282e-01 3.31357867e-01 1.71734914e-01 9.07734692e-01
5.85100092e-02 -8.87138098e-02 1.04148962e-01 -1.48683810e+00
-1.35895565e-01 5.41745663e-01 -9.08064663e-01 1.41758576e-01
6.07206821e-01 -4.08247024e-01 1.17033327e+00 -8.32244158e-01
8.68474245e-01 8.08549345e-01 6.30247235e-01 4.10688162e-01
-1.13253856e+00 -5.67274749e-01 4.21707004e-01 4.45780009e-01
-1.39391840e+00 8.05240646e-02 8.17464232e-01 -1.74770728e-01
1.32272696e+00 4.86285806e-01 6.76366746e-01 7.37161994e-01
-3.03773601e-02 6.25577450e-01 1.49582255e+00 -6.91536903e-01
3.49536508e-01 9.02911350e-02 5.17591357e-01 4.39949542e-01
4.91159290e-01 8.70663207e-03 -6.43928945e-01 -3.39648962e-01
2.48425260e-01 -8.29895288e-02 -2.62815118e-01 -4.24226850e-01
-9.21635091e-01 7.50344098e-01 -3.56851488e-01 5.89671195e-01
-7.50164688e-02 -4.79766548e-01 4.33817506e-01 4.70146984e-01
-9.96492058e-03 4.80644524e-01 -7.35738039e-01 -2.54252762e-01
-9.74697530e-01 6.90364420e-01 1.12228489e+00 1.27903521e+00
4.97129351e-01 -2.96963602e-01 1.37300819e-01 5.30171752e-01
4.23750132e-02 3.99173014e-02 3.30405176e-01 -4.23812777e-01
4.45413888e-01 1.01855564e+00 2.11824365e-02 -1.26687479e+00
-4.74097818e-01 -4.73625332e-01 -5.24716794e-01 2.28804108e-02
3.08457732e-01 1.01573259e-01 -6.20245457e-01 1.55050886e+00
2.99057871e-01 4.85191718e-02 -7.02798367e-02 5.77435017e-01
3.01597267e-01 6.66720629e-01 -1.61733583e-01 -8.26167285e-01
1.19567108e+00 -1.10719585e+00 -7.52115667e-01 -4.83646840e-02
6.06104195e-01 -6.47765696e-01 8.22117865e-01 9.96677577e-01
-9.21099901e-01 -2.45478630e-01 -1.13940787e+00 5.64167202e-01
-4.19019461e-01 -3.15678835e-01 7.69767761e-01 9.80124116e-01
-5.83329260e-01 4.48165327e-01 -7.34655321e-01 -5.41068316e-01
2.29951605e-01 5.07362604e-01 -4.73612189e-01 -1.04345836e-01
-7.55109310e-01 8.51603687e-01 7.63532281e-01 -4.14324291e-02
-7.09651634e-02 -5.11585414e-01 -4.56806034e-01 1.19214118e-01
1.12697661e+00 -3.12492460e-01 8.12353790e-01 -5.98607600e-01
-1.33232677e+00 7.11019516e-01 -2.71484673e-01 -5.69823444e-01
4.78226006e-01 -1.04538703e-04 -4.77413088e-01 -2.21793398e-01
-1.57042950e-01 7.65167177e-02 3.97538751e-01 -1.02553213e+00
-9.35113192e-01 -3.71169746e-01 4.83778724e-03 -3.08106672e-02
-3.50648433e-01 4.15698677e-01 -4.89906400e-01 -5.68811655e-01
5.50729573e-01 -8.57464015e-01 -4.77540761e-01 -2.57341266e-01
-3.59518826e-01 -3.16671818e-01 1.05191422e+00 -2.72678435e-01
2.04534602e+00 -1.96524930e+00 2.34208688e-01 5.46572149e-01
-3.34170192e-01 4.47690666e-01 1.67165950e-01 9.38885212e-01
2.36280784e-02 2.33603984e-01 -7.67784715e-01 -1.27752900e-01
9.65338498e-02 7.06542492e-01 -3.22453588e-01 1.54862732e-01
7.88974538e-02 3.20959896e-01 -6.17421448e-01 -6.55181706e-01
-1.65676489e-01 -1.29970849e-01 -8.86747956e-01 -1.72731727e-02
-5.52391350e-01 1.52962163e-01 -2.73520231e-01 8.04811001e-01
1.07608128e+00 1.94987968e-01 7.35481322e-01 2.66285658e-01
-4.41484541e-01 2.07517102e-01 -1.96446812e+00 1.72359312e+00
6.64834157e-02 -1.23066172e-01 9.03683379e-02 -1.36019051e+00
1.21156323e+00 1.37140304e-01 6.13324881e-01 -6.39139235e-01
-1.43882856e-01 6.43798351e-01 1.80217341e-01 -6.09685481e-01
3.37112069e-01 -2.06550553e-01 -5.55839539e-02 5.03249347e-01
-4.11131620e-01 2.86193460e-01 6.42170191e-01 -1.30502269e-01
1.17202950e+00 4.16097343e-01 7.00999260e-01 -6.44534171e-01
9.62139428e-01 2.33640775e-01 1.28094447e+00 7.20829308e-01
9.77638885e-02 5.95663667e-01 6.70885324e-01 -5.20618856e-01
-7.80527771e-01 -6.52822196e-01 -2.34707803e-01 6.12887025e-01
7.64741562e-03 -1.01106060e+00 -3.19054365e-01 -8.35238397e-01
2.22816002e-02 3.31356883e-01 -3.06718230e-01 3.08834791e-01
-9.43989813e-01 -9.04056370e-01 2.82829195e-01 4.31664556e-01
5.11099100e-01 -1.08747411e+00 -1.00001991e+00 5.65174699e-01
4.19814326e-03 -1.16638052e+00 -9.54995751e-02 3.95375937e-01
-7.91921794e-01 -1.08739376e+00 -8.23168159e-02 -7.58491933e-01
5.67124605e-01 2.22781941e-01 7.95290887e-01 3.66784185e-01
-2.51484215e-01 -3.08421612e-01 -8.85644138e-01 -5.11482239e-01
-1.05978124e-01 1.89796925e-01 -8.12629163e-02 8.51824507e-02
7.35005200e-01 -9.73966300e-01 -1.38612464e-01 6.48772001e-01
-8.82419109e-01 2.58650538e-02 4.65457410e-01 8.41572821e-01
8.45388114e-01 3.16406310e-01 7.45731235e-01 -1.24171209e+00
5.57580173e-01 -4.35225248e-01 -6.71533048e-01 3.04398090e-01
-1.17021620e+00 4.30407114e-02 4.48921412e-01 -2.65718073e-01
-9.86688793e-01 1.66034669e-01 -1.38448417e-01 3.55930701e-02
-1.52997345e-01 8.78949404e-01 -3.91033322e-01 1.20718770e-01
1.85694456e-01 3.06575328e-01 -2.33248264e-01 -7.96556592e-01
-1.83377787e-01 6.11853600e-01 3.01235944e-01 -8.36412728e-01
6.63806438e-01 3.43885720e-01 4.35327441e-01 -4.94019270e-01
-4.88101870e-01 -6.31604433e-01 -4.42775160e-01 3.11468691e-01
5.22593379e-01 -1.55978605e-01 -3.61927897e-01 1.16675049e-01
-9.73852932e-01 9.01217610e-02 -1.49081871e-01 1.04503110e-01
-5.31537831e-01 8.80260050e-01 -1.21513218e-01 -1.03780103e+00
-2.10866168e-01 -8.79238307e-01 3.13591897e-01 -6.74419999e-02
-2.56632656e-01 -3.95523667e-01 1.36268541e-01 -4.47533391e-02
5.06385744e-01 5.70634365e-01 9.22194004e-01 -7.87272930e-01
-4.46545362e-01 2.62449924e-02 -2.08207637e-01 2.76110303e-02
1.40031185e-02 -7.70171657e-02 -3.32662761e-01 -7.32861832e-02
1.32569283e-01 1.43017292e-01 5.61405957e-01 -4.85155545e-02
1.20245516e+00 -2.66270638e-01 -6.36038601e-01 5.09472072e-01
1.76266897e+00 5.03086567e-01 7.34179080e-01 6.44625962e-01
9.95792821e-02 7.92520046e-01 1.15005398e+00 6.72481120e-01
3.59996930e-02 1.10236204e+00 3.75015765e-01 4.65492755e-01
1.75505340e-01 -8.25974420e-02 1.11506812e-01 5.75398326e-01
-1.86490685e-01 -2.64898539e-01 -9.99336958e-01 7.05084741e-01
-2.14957500e+00 -1.01990986e+00 -6.82496667e-01 2.14149880e+00
7.18991756e-01 5.33646941e-01 3.29351276e-01 8.43291044e-01
4.18735772e-01 -1.69229224e-01 1.59258693e-01 -9.90894735e-01
-2.08433121e-01 4.67363268e-01 2.98607379e-01 3.70144963e-01
-7.77689874e-01 7.16887951e-01 5.94988823e+00 7.38786280e-01
-5.47762811e-01 1.47311434e-01 -2.91858286e-01 1.77090406e-01
-3.56139690e-01 4.32761252e-01 -1.08218122e+00 6.45188570e-01
5.85614204e-01 -1.95337817e-01 1.13094270e-01 8.31836045e-01
-1.81886226e-01 -2.99077392e-01 -9.82996762e-01 8.06742132e-01
1.42845009e-02 -1.14506805e+00 -1.46416932e-01 2.83678949e-01
6.75947070e-01 -4.19815421e-01 -2.99075872e-01 1.46113649e-01
-4.23561841e-01 -8.57824206e-01 4.44518656e-01 1.41578957e-01
3.60944867e-01 -7.65936315e-01 8.89232814e-01 6.01477265e-01
-1.23103833e+00 -3.41126412e-01 -3.98183674e-01 -4.21548456e-01
4.08852428e-01 7.80685246e-01 -4.76618260e-01 1.04298747e+00
6.64042771e-01 3.89556617e-01 -2.26892903e-01 1.16860580e+00
-1.88240930e-01 5.41459024e-01 -7.36076295e-01 -3.00635137e-02
2.31032893e-01 -3.44059795e-01 7.94319928e-01 1.52740145e+00
4.64858115e-01 1.11722432e-01 3.28230917e-01 8.42507422e-01
5.59791148e-01 3.55052650e-01 -5.64824462e-01 5.02693951e-02
5.32693148e-01 8.38944077e-01 -8.68412793e-01 4.52894010e-02
-8.07879925e-01 6.33402705e-01 -3.64729911e-02 -2.33545348e-01
-6.07883811e-01 -3.52860540e-01 3.63999248e-01 2.89866745e-01
7.11410284e-01 -3.99386495e-01 -5.62553704e-01 -8.80746365e-01
6.42054498e-01 -1.05842400e+00 7.29564488e-01 -3.40132900e-02
-1.07766724e+00 6.24739766e-01 3.20151627e-01 -1.25155842e+00
4.98560369e-02 -5.89111328e-01 -5.60611665e-01 6.56996548e-01
-1.60011470e+00 -1.05367482e+00 -9.07519385e-02 7.83969760e-01
4.64644074e-01 -3.06474697e-02 9.58409488e-01 4.29041654e-01
-5.61735153e-01 6.47351444e-01 -3.82853955e-01 -5.43270051e-01
3.27812284e-01 -1.00624180e+00 -4.98774881e-03 1.24580669e+00
-1.51943108e-02 7.26441681e-01 1.00762165e+00 -6.92894280e-01
-1.54872537e+00 -5.07875144e-01 1.33744502e+00 -1.71427429e-01
4.42822307e-01 -2.38966584e-01 -1.04030478e+00 5.56714594e-01
1.25830799e-01 -4.07779515e-02 7.49161959e-01 1.46018401e-01
-4.46576267e-01 -1.41934782e-01 -1.30855918e+00 2.97308177e-01
1.60286129e+00 1.15780979e-01 -9.18107271e-01 7.53344968e-03
5.60523510e-01 -1.96222886e-01 -7.08357811e-01 6.71881199e-01
5.84466457e-01 -1.33566749e+00 4.96551514e-01 -4.09548193e-01
2.58504122e-01 -6.08273149e-01 -3.19264352e-01 -6.17651880e-01
-2.16613695e-01 -6.52171493e-01 -2.45598674e-01 1.33377337e+00
5.27792215e-01 -8.02028537e-01 8.90779793e-01 3.72588605e-01
-3.84176314e-01 -1.14189124e+00 -1.06824791e+00 -1.26149523e+00
-2.01455981e-01 -5.30724049e-01 8.43872070e-01 9.96257365e-01
7.63063610e-01 -1.93988949e-01 -5.31190276e-01 -6.30924031e-02
3.64036709e-01 6.52768731e-01 6.54969394e-01 -1.40513825e+00
-9.54883575e-01 -5.06860495e-01 -3.25153112e-01 -6.61732435e-01
-2.72736758e-01 -8.89605582e-01 -2.79988915e-01 -1.49012089e+00
2.79783249e-01 -7.23974645e-01 -3.45676959e-01 6.23872936e-01
1.80981055e-01 1.14032356e-02 2.29335681e-01 6.03552572e-02
-4.48928028e-01 -3.47192064e-02 8.64656270e-01 2.35892773e-01
-4.79955584e-01 -1.04666017e-01 -6.47901952e-01 4.33444619e-01
1.04174447e+00 -8.40950429e-01 -4.05315638e-01 -7.67050385e-02
7.96238720e-01 -8.00856575e-02 -2.90895969e-01 -1.08077776e+00
2.59517342e-01 -6.35017216e-01 -4.40917104e-01 -8.48902345e-01
-4.97133434e-02 -1.13975739e+00 6.02156460e-01 7.06021786e-01
1.39080748e-01 3.16851258e-01 7.76853189e-02 5.84722102e-01
-3.34655315e-01 -5.19980609e-01 4.34570402e-01 -2.07509235e-01
-7.21102774e-01 1.17408641e-01 -2.65880793e-01 -2.59297758e-01
1.22957504e+00 -6.17900133e-01 -8.98919255e-02 3.56978685e-01
-7.02467263e-01 2.83477604e-01 4.47942734e-01 2.98880517e-01
5.53657055e-01 -1.02402914e+00 -5.12203157e-01 3.53029639e-01
1.50056511e-01 5.14075086e-02 -7.90870786e-02 1.20112884e+00
-3.19153637e-01 6.03288233e-01 -5.32119930e-01 -2.84977704e-01
-1.40343273e+00 1.06813204e+00 -1.51415899e-01 -7.66910970e-01
-7.56885767e-01 5.81822217e-01 -3.80253911e-01 -1.56886116e-01
2.99566716e-01 -3.57784867e-01 -9.37242508e-02 -1.93403766e-01
5.51490068e-01 3.19081098e-01 6.94155693e-02 -4.09834050e-02
-6.30580485e-01 7.88941443e-01 -1.28983660e-02 -7.52104120e-03
1.50612438e+00 -2.87811067e-02 -3.11221063e-01 -6.72631562e-02
6.59793079e-01 2.67298788e-01 -5.55655420e-01 -2.66888648e-01
6.11257970e-01 -6.65631354e-01 -5.12242675e-01 -7.82016873e-01
-8.20650101e-01 5.69003761e-01 9.76788178e-02 3.36939722e-01
1.45613325e+00 -2.63255924e-01 7.89598167e-01 2.56151259e-01
9.49752390e-01 -1.09641147e+00 -4.34084833e-01 4.14048761e-01
7.82214522e-01 -7.60171592e-01 1.82880729e-01 -1.14281583e+00
-3.59487325e-01 1.11995673e+00 7.25486696e-01 -1.51602849e-01
6.81376934e-01 6.54246330e-01 -6.60446942e-01 -7.39162713e-02
-7.67262638e-01 -3.18031341e-01 -2.01535538e-01 6.77335143e-01
1.39019012e-01 -2.73482762e-02 -1.60682452e+00 9.55654263e-01
-6.61849529e-02 3.25155169e-01 5.04742324e-01 1.43248785e+00
-4.92346972e-01 -2.03677106e+00 -2.41455108e-01 2.08043322e-01
-4.79621083e-01 1.37362421e-01 -4.48962718e-01 9.37392414e-01
8.58003139e-01 9.44163859e-01 -3.03653270e-01 -4.78381097e-01
3.96252483e-01 2.99457788e-01 6.98495746e-01 -6.80290520e-01
-6.79877460e-01 -1.12452932e-01 4.88931209e-01 -6.87817395e-01
-5.91204762e-01 -9.55816269e-01 -1.25413239e+00 -6.28906041e-02
-3.55439514e-01 3.36720675e-01 5.01812577e-01 8.41862202e-01
1.93442598e-01 3.81921716e-02 3.81552160e-01 6.14642724e-02
-5.14230013e-01 -7.33792603e-01 -5.08027017e-01 1.62950590e-01
-3.94134283e-01 -8.34463537e-01 -2.10779950e-01 -2.91199982e-01]
|
[8.285184860229492, 6.293193340301514]
|
85b4477e-b13c-4738-8e52-c19a3480cdc9
|
ppmn-pixel-phrase-matching-network-for-one
|
2208.05647
| null |
https://arxiv.org/abs/2208.05647v1
|
https://arxiv.org/pdf/2208.05647v1.pdf
|
PPMN: Pixel-Phrase Matching Network for One-Stage Panoptic Narrative Grounding
|
Panoptic Narrative Grounding (PNG) is an emerging task whose goal is to segment visual objects of things and stuff categories described by dense narrative captions of a still image. The previous two-stage approach first extracts segmentation region proposals by an off-the-shelf panoptic segmentation model, then conducts coarse region-phrase matching to ground the candidate regions for each noun phrase. However, the two-stage pipeline usually suffers from the performance limitation of low-quality proposals in the first stage and the loss of spatial details caused by region feature pooling, as well as complicated strategies designed for things and stuff categories separately. To alleviate these drawbacks, we propose a one-stage end-to-end Pixel-Phrase Matching Network (PPMN), which directly matches each phrase to its corresponding pixels instead of region proposals and outputs panoptic segmentation by simple combination. Thus, our model can exploit sufficient and finer cross-modal semantic correspondence from the supervision of densely annotated pixel-phrase pairs rather than sparse region-phrase pairs. In addition, we also propose a Language-Compatible Pixel Aggregation (LCPA) module to further enhance the discriminative ability of phrase features through multi-round refinement, which selects the most compatible pixels for each phrase to adaptively aggregate the corresponding visual context. Extensive experiments show that our method achieves new state-of-the-art performance on the PNG benchmark with 4.0 absolute Average Recall gains.
|
['Si Liu', 'Xiaolin Wei', 'Xiaoming Wei', 'Junshi Huang', 'Tianrui Hui', 'Zi-han Ding', 'Zihan Ding']
|
2022-08-11
| null | null | null | null |
['panoptic-segmentation']
|
['computer-vision']
|
[ 4.16504681e-01 4.19018343e-02 -5.44578493e-01 -2.41562322e-01
-1.11274624e+00 -5.40706038e-01 6.37392282e-01 -1.58409663e-02
-4.12750363e-01 3.76803607e-01 2.85199195e-01 4.92719375e-02
3.37149650e-01 -9.65639412e-01 -6.53853118e-01 -6.00176990e-01
3.46725792e-01 3.00384313e-01 7.16817379e-01 -1.80201501e-01
2.37030849e-01 2.59332478e-01 -1.37165987e+00 7.15216458e-01
9.15823579e-01 1.20190775e+00 5.30386865e-01 2.05343664e-01
-4.24226940e-01 3.00408870e-01 -1.69202209e-01 -4.88352507e-01
4.89448994e-01 -3.59723330e-01 -6.11050785e-01 4.71182525e-01
6.19416833e-01 -1.69809535e-01 -2.73504376e-01 1.34306490e+00
1.82680905e-01 1.68162674e-01 4.71147269e-01 -9.39808071e-01
-6.95823312e-01 7.14883924e-01 -1.11681616e+00 2.71489918e-01
1.57745793e-01 5.57303369e-01 1.55282998e+00 -1.22131562e+00
7.50615537e-01 1.37673604e+00 4.93763834e-01 3.14388037e-01
-1.35430539e+00 -8.23524952e-01 5.17250538e-01 -8.16460401e-02
-1.56483316e+00 -6.43110573e-02 7.79204667e-01 -3.22295189e-01
8.67776930e-01 2.12066710e-01 9.51021612e-01 8.67731392e-01
2.34848022e-01 9.94521916e-01 1.09794211e+00 -1.47426920e-02
9.75688398e-02 -1.21825058e-02 -2.52123605e-02 9.37193096e-01
-1.18405879e-01 -8.99241939e-02 -5.02956569e-01 2.25231186e-01
1.06737745e+00 1.73959583e-01 -1.91017210e-01 -3.99240434e-01
-1.44594800e+00 7.87302077e-01 9.42669690e-01 2.44783401e-01
-6.11179709e-01 2.39652544e-01 1.39630109e-01 -2.64815688e-01
3.91993880e-01 3.97255749e-01 -2.25312322e-01 3.48267376e-01
-1.36458099e+00 4.01557952e-01 2.13846445e-01 9.41928148e-01
1.03729904e+00 -3.98879707e-01 -6.55588567e-01 7.73490489e-01
2.88612843e-01 3.15174639e-01 3.55920158e-02 -7.06590474e-01
8.92562687e-01 8.71136725e-01 -1.01519592e-01 -1.02191579e+00
-2.91446090e-01 -6.56941891e-01 -7.22777903e-01 -3.72687355e-02
4.20191646e-01 1.84003785e-01 -1.39776230e+00 1.60069335e+00
3.22058111e-01 1.12172209e-01 -9.86820385e-02 1.26793778e+00
1.03357327e+00 9.98337269e-01 4.57482725e-01 -1.00358970e-01
1.92824864e+00 -1.47397983e+00 -5.39610028e-01 -7.48475671e-01
1.59604251e-01 -7.05673516e-01 1.41885304e+00 -1.00300178e-01
-1.18608773e+00 -5.44587553e-01 -8.86247933e-01 -4.21203971e-01
-2.94175416e-01 -8.26196447e-02 5.55216312e-01 2.34195039e-01
-7.40423918e-01 1.25425652e-01 -5.38350582e-01 -2.65989721e-01
8.69234025e-01 1.93355948e-01 -1.91501632e-01 -2.67967314e-01
-1.10618103e+00 4.56054091e-01 6.46036685e-01 1.08173132e-01
-6.96937501e-01 -8.90364051e-01 -9.29884315e-01 3.00223261e-01
6.77616835e-01 -7.71971285e-01 8.63921881e-01 -9.69488144e-01
-1.05312002e+00 1.04882467e+00 -5.29591292e-02 -2.40157604e-01
3.31208616e-01 -4.73037623e-02 -2.58638769e-01 3.93981189e-01
4.60124105e-01 1.49928582e+00 6.28052294e-01 -1.25661540e+00
-9.96579289e-01 -2.38914192e-01 7.95588493e-02 4.62519109e-01
1.48841143e-02 3.30281258e-02 -1.24672151e+00 -9.27073598e-01
6.02998614e-01 -7.22204328e-01 -5.38412571e-01 1.76502466e-01
-5.48775673e-01 -1.41072705e-01 6.09429896e-01 -6.68424249e-01
1.17333794e+00 -2.31867409e+00 3.67160887e-02 3.36059004e-01
1.60922810e-01 -2.25593731e-01 -2.39411592e-01 -1.23291627e-01
2.12069303e-01 5.05511388e-02 -6.81120932e-01 -3.37824762e-01
6.48140758e-02 6.40731454e-02 -3.89613181e-01 1.48768485e-01
4.91677314e-01 1.31064594e+00 -9.26847994e-01 -8.50790143e-01
1.06283449e-01 2.25460470e-01 -6.58288717e-01 -2.28518806e-02
-7.27358341e-01 4.09593999e-01 -5.89748919e-01 1.02480435e+00
7.39126623e-01 -3.80644053e-01 -1.68538168e-01 -4.37836289e-01
-1.60309643e-01 2.20129117e-01 -1.07464767e+00 2.10070300e+00
-2.72811830e-01 4.71519023e-01 -6.76691681e-02 -6.67432427e-01
8.49800169e-01 -1.77052096e-02 3.51366580e-01 -8.23676765e-01
8.74071419e-02 2.74304599e-01 -2.62394011e-01 -3.94041598e-01
6.79901838e-01 -2.02574342e-01 -6.31569803e-01 1.22058466e-01
-1.26616910e-01 -3.81429434e-01 4.91319373e-02 1.26317918e-01
8.35697591e-01 1.98319435e-01 2.28349134e-01 -2.60823131e-01
4.32250232e-01 3.51764947e-01 7.76690066e-01 5.61559081e-01
-1.64814785e-01 1.13669169e+00 4.74291533e-01 -4.63775218e-01
-1.07299960e+00 -1.31123614e+00 -2.90222540e-02 1.14144158e+00
6.51294053e-01 -2.92136103e-01 -6.47961199e-01 -8.24915588e-01
-1.87919155e-01 6.44480824e-01 -5.58428109e-01 1.15820773e-01
-7.64530420e-01 -7.39954591e-01 1.95927098e-01 7.10805178e-01
9.49836969e-01 -1.37639713e+00 -6.03448689e-01 3.15496176e-01
-3.40990007e-01 -1.33957517e+00 -9.84853685e-01 1.63447604e-01
-5.42877913e-01 -6.58119142e-01 -9.12531972e-01 -1.12157309e+00
7.14786530e-01 2.95718372e-01 1.16187549e+00 -1.47861063e-01
-7.44659975e-02 -2.21893057e-01 -1.59409806e-01 2.02496991e-01
5.01531810e-02 1.44177049e-01 -5.47209084e-01 9.29263011e-02
2.97372788e-01 -4.20556366e-01 -9.72984612e-01 4.37392086e-01
-8.47769678e-01 5.76947808e-01 9.94430721e-01 7.44691133e-01
1.21019053e+00 -8.46467465e-02 1.47887677e-01 -8.03564847e-01
1.77784592e-01 -4.26905364e-01 -5.88805795e-01 3.22118998e-01
-2.70247459e-01 -1.41642168e-01 4.05613273e-01 -3.89442801e-01
-1.01973164e+00 2.99672544e-01 -1.11772157e-01 -5.30342340e-01
-6.13065511e-02 3.15184534e-01 -4.87581939e-01 3.27973425e-01
3.20643991e-01 3.23737711e-01 -4.80151504e-01 -2.88743675e-01
8.33416224e-01 4.03370559e-01 9.96807873e-01 -3.93651396e-01
8.28638136e-01 4.92663652e-01 -3.37029964e-01 -2.23404676e-01
-1.07684493e+00 -5.52634001e-01 -5.99602222e-01 -1.55639470e-01
1.38254237e+00 -1.17345655e+00 -9.92341787e-02 4.82416935e-02
-1.08146834e+00 -2.19913051e-01 -2.91789591e-01 2.81925142e-01
-4.47046697e-01 1.10436454e-01 -5.96280396e-01 -3.40352267e-01
-5.09397924e-01 -1.39989877e+00 1.31719506e+00 5.50683737e-01
-6.01557083e-02 -3.75107139e-01 -3.68765056e-01 6.64953351e-01
2.06129290e-02 1.50925353e-01 8.25903893e-01 -3.17391098e-01
-1.06939209e+00 6.95867538e-02 -8.50523889e-01 -7.10317912e-03
-1.96704820e-01 -3.34962279e-01 -8.77841711e-01 -1.45987824e-01
-3.47846389e-01 -7.42861778e-02 1.29448628e+00 4.34208810e-01
1.15379083e+00 -2.50005037e-01 -4.40152943e-01 8.24995399e-01
1.65528893e+00 -2.82593071e-02 6.64081693e-01 3.70850414e-01
1.12289333e+00 6.01881325e-01 9.01375890e-01 8.52143764e-02
3.09279114e-01 7.09066451e-01 3.96702409e-01 -3.90600413e-01
-3.51966351e-01 -4.20623988e-01 3.36237520e-01 1.14044458e-01
3.17098916e-01 -2.56088257e-01 -8.43882084e-01 8.77542913e-01
-1.82250273e+00 -7.69461751e-01 -7.22395107e-02 1.73525202e+00
8.20849240e-01 3.97900581e-01 1.77951485e-01 -3.14231217e-01
9.22306299e-01 5.42451262e-01 -5.53373992e-01 -1.68682709e-01
-2.78955013e-01 1.88142523e-01 5.61393619e-01 2.55557388e-01
-1.30371284e+00 1.41127491e+00 5.19551659e+00 1.34079421e+00
-9.45891619e-01 2.41497457e-01 1.01934004e+00 -7.34426752e-02
-5.32174051e-01 4.84523885e-02 -8.23661387e-01 5.00847340e-01
2.19322145e-01 4.67039794e-01 2.47789264e-01 7.81744242e-01
1.64236397e-01 -2.82602191e-01 -8.02551866e-01 9.21695173e-01
-1.83082558e-02 -1.61477458e+00 1.80798650e-01 3.25106829e-02
1.13768637e+00 6.69751987e-02 1.57794937e-01 1.97670549e-01
3.26730683e-02 -8.84499609e-01 1.09526360e+00 1.65396467e-01
8.99512529e-01 -7.89306641e-01 5.01682043e-01 1.86240505e-02
-1.56359923e+00 -2.78896004e-01 -2.68918037e-01 3.76866877e-01
2.96296865e-01 5.28048217e-01 -5.14450431e-01 3.46754998e-01
7.66275465e-01 6.08367801e-01 -5.16751945e-01 1.10187387e+00
-3.45665842e-01 4.67145532e-01 -4.04857635e-01 1.15540005e-01
9.39614952e-01 -1.84036747e-01 5.53641558e-01 1.44669116e+00
2.10236505e-01 3.08492780e-01 4.28284377e-01 1.37691963e+00
-1.16792977e-01 2.29547769e-01 -1.63990974e-01 3.08432668e-01
3.95585120e-01 1.50428152e+00 -1.33724809e+00 -4.47598934e-01
-5.55992365e-01 1.01829040e+00 2.22410604e-01 2.68061250e-01
-9.50462818e-01 -1.28013253e-01 4.72492486e-01 3.46426070e-01
6.80962086e-01 1.95315462e-02 -7.98665524e-01 -9.72782373e-01
2.13583335e-01 -5.95466614e-01 5.73931038e-01 -8.82784665e-01
-1.11887622e+00 5.43987155e-01 1.23086404e-02 -1.29544520e+00
3.11186343e-01 -1.45714864e-01 -7.09094644e-01 8.59885395e-01
-1.51963305e+00 -1.50784075e+00 -3.71343702e-01 3.80901605e-01
9.83148634e-01 1.78150624e-01 3.09570163e-01 2.37678051e-01
-6.36531293e-01 5.45819581e-01 -2.52102792e-01 1.65210411e-01
4.16840941e-01 -1.06516635e+00 5.12153983e-01 1.02313268e+00
1.31009951e-01 3.12482148e-01 5.02665043e-01 -6.84420049e-01
-7.67938495e-01 -1.45015347e+00 7.91758239e-01 -1.60392389e-01
5.68817556e-01 -5.26838779e-01 -9.08816993e-01 4.01454896e-01
1.60561770e-01 1.24164492e-01 1.76577091e-01 -2.39703462e-01
-3.84930879e-01 -1.11898787e-01 -1.05982089e+00 9.55417216e-01
1.13944173e+00 -4.00250107e-01 -6.76822841e-01 2.53031701e-01
9.53134358e-01 -3.73392254e-01 -5.80321550e-01 3.16484928e-01
4.16973144e-01 -7.37769067e-01 1.13266063e+00 5.16046658e-02
7.13748336e-01 -6.40022516e-01 -2.18184575e-01 -8.07768643e-01
-4.56488520e-01 -5.10070920e-01 3.09667706e-01 1.48400557e+00
6.56957865e-01 -1.27579764e-01 9.17360306e-01 4.47485566e-01
-3.73942524e-01 -9.46394444e-01 -8.79417896e-01 -4.38357353e-01
-1.68857694e-01 -4.36455011e-01 6.20390654e-01 6.92255318e-01
-1.47522241e-01 4.49950218e-01 -1.13224894e-01 2.87292212e-01
4.35161799e-01 5.43197274e-01 4.70504701e-01 -7.23486304e-01
-1.54218838e-01 -7.23320007e-01 -2.06052974e-01 -1.38890409e+00
-2.24749725e-02 -9.48065162e-01 5.64394891e-01 -1.67955732e+00
5.77700436e-01 -5.74084759e-01 -5.06785095e-01 5.33520162e-01
-5.59110999e-01 8.51632178e-01 3.19137037e-01 3.24295521e-01
-8.57498825e-01 4.97015029e-01 1.72221088e+00 -3.88335288e-01
-4.54609424e-01 -3.23941499e-01 -8.34427595e-01 6.80620015e-01
5.65438628e-01 -4.99064803e-01 -3.00944954e-01 -3.59677166e-01
1.08407937e-01 -3.13057899e-02 6.24496937e-01 -9.94633555e-01
1.52462125e-01 -3.55270177e-01 5.23997724e-01 -1.04063809e+00
3.24195832e-01 -7.14938581e-01 -3.35271694e-02 2.04882652e-01
-2.00778186e-01 -1.53388912e-02 1.91156134e-01 7.16558576e-01
-1.14051804e-01 2.98308432e-02 8.79236400e-01 -3.85785311e-01
-1.16318929e+00 4.17770237e-01 -1.42175123e-01 1.49820922e-02
1.02085412e+00 -4.20375437e-01 -2.71605581e-01 1.81020081e-01
-5.20939827e-01 4.34568226e-01 6.30442142e-01 4.33339983e-01
5.72810352e-01 -1.22867787e+00 -7.01717138e-01 1.40987009e-01
2.96541572e-01 4.91259009e-01 3.55366498e-01 8.84926260e-01
-5.52944899e-01 2.11728379e-01 -2.31427774e-01 -9.04070616e-01
-9.75463450e-01 6.29607439e-01 1.44743040e-01 -4.07795340e-01
-1.02191269e+00 1.17287552e+00 9.37503636e-01 -6.95992410e-02
-4.11376357e-02 -5.50667524e-01 -2.16950104e-01 8.94254148e-02
2.95822620e-01 -2.84822941e-01 -3.56208891e-01 -8.63090038e-01
-3.30962390e-01 9.05138373e-01 -2.49006212e-01 -3.44582409e-01
1.16132212e+00 -2.20846385e-01 -2.89137643e-02 7.09660724e-02
1.11416054e+00 -1.60125867e-01 -1.55680847e+00 -4.53045994e-01
-1.70393348e-01 -5.20393372e-01 2.30583012e-01 -8.66939962e-01
-1.29409528e+00 6.20655894e-01 5.07807732e-01 -2.94782221e-01
1.29063940e+00 4.24315333e-01 1.14707768e+00 -1.03917256e-01
6.95416778e-02 -1.02887225e+00 2.04484925e-01 1.25825524e-01
7.81847119e-01 -1.06815517e+00 1.95798390e-02 -8.54803860e-01
-7.99407721e-01 6.99934065e-01 8.01909924e-01 -2.68861890e-01
2.89490312e-01 -1.28866974e-02 -7.79083446e-02 -3.64223927e-01
-4.30719316e-01 -5.33013701e-01 6.97669208e-01 2.76325762e-01
-4.06761803e-02 7.41502866e-02 -3.81211162e-01 8.37831140e-01
-7.31115714e-02 -4.03073877e-01 2.82992087e-02 5.28308272e-01
-5.78561664e-01 -7.03649461e-01 -5.91174483e-01 3.81109834e-01
-3.59242976e-01 -4.34835911e-01 -1.68579787e-01 8.09182048e-01
5.52246511e-01 7.00325072e-01 4.41375732e-01 -3.19193095e-01
3.66361439e-01 -1.15791842e-01 1.85816616e-01 -7.52241790e-01
-8.82343769e-01 6.06042743e-01 -1.53809398e-01 -6.41478956e-01
-3.47962737e-01 -7.17676401e-01 -1.51539648e+00 1.04068391e-01
-1.49896368e-01 -2.06900880e-01 2.46391699e-01 9.43565667e-01
1.34700388e-01 6.23832524e-01 4.15963233e-01 -9.10842717e-01
1.93371385e-01 -8.19633603e-01 -4.65269893e-01 4.81314629e-01
7.66800269e-02 -6.22587979e-01 -4.70065977e-03 4.52266894e-02]
|
[9.712986946105957, 0.5432993769645691]
|
a99fc49c-b97e-42ca-92d1-6ade487cd41e
|
long-tail-visual-relationship-recognition
|
2004.00436
| null |
https://arxiv.org/abs/2004.00436v7
|
https://arxiv.org/pdf/2004.00436v7.pdf
|
Exploring Long Tail Visual Relationship Recognition with Large Vocabulary
|
Several approaches have been proposed in recent literature to alleviate the long-tail problem, mainly in object classification tasks. In this paper, we make the first large-scale study concerning the task of Long-Tail Visual Relationship Recognition (LTVRR). LTVRR aims at improving the learning of structured visual relationships that come from the long-tail (e.g., "rabbit grazing on grass"). In this setup, the subject, relation, and object classes each follow a long-tail distribution. To begin our study and make a future benchmark for the community, we introduce two LTVRR-related benchmarks, dubbed VG8K-LT and GQA-LT, built upon the widely used Visual Genome and GQA datasets. We use these benchmarks to study the performance of several state-of-the-art long-tail models on the LTVRR setup. Lastly, we propose a visiolinguistic hubless (VilHub) loss and a Mixup augmentation technique adapted to LTVRR setup, dubbed as RelMix. Both VilHub and RelMix can be easily integrated on top of existing models and despite being simple, our results show that they can remarkably improve the performance, especially on tail classes. Benchmarks, code, and models have been made available at: https://github.com/Vision-CAIR/LTVRR.
|
['Jun Chen', 'Aniket Agarwal', 'Kenneth Church', 'Jiaji Huang', 'Sherif Abdelkarim', 'Panos Achlioptas', 'Mohamed Elhoseiny', 'Boyang Li']
|
2020-03-25
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Abdelkarim_Exploring_Long_Tail_Visual_Relationship_Recognition_With_Large_Vocabulary_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Abdelkarim_Exploring_Long_Tail_Visual_Relationship_Recognition_With_Large_Vocabulary_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['visual-relationship-detection']
|
['computer-vision']
|
[-1.83579504e-01 -5.97186163e-02 -1.99304715e-01 -5.01448274e-01
-3.63072485e-01 -5.23785830e-01 7.71111012e-01 8.97650048e-02
-3.22919816e-01 4.91673410e-01 4.38417457e-02 -3.79188746e-01
-1.08682122e-02 -4.78747278e-01 -9.35398757e-01 -6.25138819e-01
-3.43727730e-02 6.38607085e-01 4.05586839e-01 -3.74692261e-01
-3.23144883e-01 4.41483945e-01 -1.50506806e+00 4.47237849e-01
3.94308329e-01 9.83667612e-01 2.67530024e-01 6.84799433e-01
1.45135492e-01 1.14940655e+00 -2.78135568e-01 -8.62344027e-01
7.51592368e-02 -2.24445343e-01 -8.82807255e-01 -1.54105410e-01
8.83166313e-01 -4.19157073e-02 -5.80712378e-01 7.36671805e-01
5.37304342e-01 3.85546595e-01 8.40410948e-01 -1.60052156e+00
-9.77874637e-01 8.06215823e-01 -9.94017065e-01 2.04671651e-01
1.20225415e-01 -7.10268691e-02 1.46035111e+00 -1.12590396e+00
7.76917994e-01 1.52941000e+00 7.34525859e-01 3.67306620e-01
-1.27939284e+00 -5.85372210e-01 3.40526491e-01 5.58105886e-01
-1.57823825e+00 -2.01948971e-01 5.80734849e-01 -5.40108442e-01
9.39697027e-01 2.59354651e-01 4.06824499e-01 1.21367252e+00
-1.52956828e-01 9.61111426e-01 1.10098529e+00 -2.71590114e-01
-2.73654133e-01 -6.88733608e-02 5.57849765e-01 7.43412495e-01
1.37987196e-01 4.55884412e-02 -4.83640820e-01 1.23851202e-01
6.05854273e-01 -2.28377283e-01 -4.28520977e-01 -7.32461572e-01
-1.07163370e+00 7.31572211e-01 1.01662064e+00 -2.70496812e-02
2.19019912e-02 2.68635213e-01 3.69147807e-01 3.29354227e-01
6.66516125e-01 6.80791214e-02 -3.69602621e-01 3.46618503e-01
-5.44912100e-01 1.67056724e-01 8.31931889e-01 1.23842645e+00
6.28646493e-01 -3.84141862e-01 -4.88832861e-01 1.19146025e+00
4.74679083e-01 3.42208534e-01 -4.22824845e-02 -6.58790231e-01
5.73224723e-01 2.74821043e-01 -1.59977958e-01 -9.22931671e-01
-4.22492176e-01 -6.06028140e-01 -8.77556920e-01 1.38565674e-01
4.72985685e-01 3.27024668e-01 -1.05031383e+00 1.62755370e+00
1.35871291e-01 2.04408035e-01 6.49076328e-02 9.19968784e-01
1.61879468e+00 7.46888101e-01 9.46821272e-02 1.18567027e-01
1.47458267e+00 -1.59539688e+00 -4.97930437e-01 -4.26447451e-01
5.26974440e-01 -6.59724414e-01 1.24044216e+00 1.90458089e-01
-8.69200587e-01 -5.89057565e-01 -6.92079961e-01 -5.17298222e-01
-6.51906490e-01 3.53336126e-01 6.47360265e-01 -6.94110170e-02
-1.05963206e+00 2.71960795e-01 -5.81465006e-01 -6.13409936e-01
5.92450023e-01 1.92763004e-02 -5.91534674e-01 -2.96952993e-01
-9.29466248e-01 1.13102102e+00 2.77085274e-01 3.93740416e-01
-1.00784564e+00 -6.36898994e-01 -8.92377734e-01 -1.56125575e-01
5.36604285e-01 -6.40779555e-01 1.11639941e+00 -5.76720715e-01
-7.71330476e-01 1.54786003e+00 -1.12783781e-03 -4.48742270e-01
6.01755083e-01 -4.71702129e-01 -1.25641271e-01 -8.56974199e-02
-1.88907888e-02 7.19211638e-01 6.33170605e-01 -1.62523532e+00
-2.83008844e-01 -3.41218084e-01 1.11328028e-02 1.17946573e-01
4.57514934e-02 3.23975265e-01 -8.74706864e-01 -7.37010896e-01
-4.46252555e-01 -9.64039266e-01 5.19145951e-02 2.68696964e-01
-4.90669966e-01 -4.27915514e-01 7.46096253e-01 -6.16865516e-01
9.96647358e-01 -2.17219281e+00 4.76691008e-01 -1.36080146e-01
3.82624835e-01 5.01568079e-01 -5.04923701e-01 4.17446911e-01
-4.10238296e-01 -1.27548456e-01 -3.31665725e-01 -8.07401597e-01
-9.11435410e-02 4.96643335e-01 -4.33083892e-01 3.85285437e-01
-4.70616110e-02 1.32740629e+00 -7.76421189e-01 -5.63058019e-01
1.25587150e-01 5.43835461e-01 -1.05430812e-01 5.77653885e-01
-3.77837598e-01 3.17508668e-01 -7.18169063e-02 6.42626345e-01
6.36878729e-01 -4.96884435e-01 -1.29992589e-01 -4.53890830e-01
2.88415682e-02 -1.35076106e-01 -5.80168605e-01 1.50574708e+00
-3.67599368e-01 8.46339047e-01 -1.02658309e-01 -8.43847036e-01
1.06613815e+00 -1.24463938e-01 -2.42519323e-02 -6.21198952e-01
1.91896081e-01 -1.29479980e-02 -8.15188047e-03 -3.98814350e-01
5.71388841e-01 1.56109363e-01 2.52250284e-01 -1.02052130e-01
2.06173167e-01 3.14450152e-02 2.88613051e-01 3.79633427e-01
7.73878515e-01 5.33977628e-01 4.21356320e-01 -1.02553852e-02
2.79841304e-01 -3.01705450e-01 2.34900489e-01 6.99451566e-01
-1.14947669e-01 9.75041091e-01 6.27849698e-01 -2.59579390e-01
-7.55534947e-01 -1.17620599e+00 -7.52513632e-02 1.50541818e+00
3.18725616e-01 -5.75652421e-01 -4.65686806e-02 -8.99019778e-01
1.93226919e-01 7.06075251e-01 -9.14444745e-01 -3.20091844e-02
-3.61812979e-01 -8.67782831e-01 5.63496292e-01 8.53292882e-01
4.04754728e-01 -1.17614961e+00 -1.67884380e-01 -2.27076054e-01
-2.39748955e-01 -1.62510240e+00 -2.97429919e-01 2.12988421e-01
-3.26488137e-01 -1.07655752e+00 -9.41165984e-01 -7.69702494e-01
4.12497759e-01 4.19193804e-01 1.72519362e+00 1.16890505e-01
-4.21133310e-01 3.58836859e-01 -7.00392365e-01 -3.14706206e-01
-5.12140356e-02 -3.82635137e-03 -5.68785191e-01 3.95780355e-02
1.61340684e-01 -4.61703867e-01 -4.53810900e-01 4.63941008e-01
-5.36867976e-01 1.83226869e-01 4.33837771e-01 7.25750089e-01
8.34092379e-01 -4.23850387e-01 2.86373526e-01 -9.33101058e-01
2.17778295e-01 -5.88468671e-01 -5.92636108e-01 6.11947179e-01
-3.36930990e-01 -2.41284475e-01 3.37968141e-01 -3.77651811e-01
-8.50936055e-01 -2.07566947e-01 -1.22954041e-01 -7.38987625e-01
5.35121597e-02 6.23464942e-01 -1.76422969e-01 -1.41257286e-01
5.55877924e-01 -1.17303491e-01 -3.13476980e-01 -6.05606556e-01
7.53383517e-01 5.25689602e-01 7.67421484e-01 -4.72887427e-01
6.35680735e-01 3.47069860e-01 1.12865835e-01 -9.54396784e-01
-1.26750612e+00 -8.00438583e-01 -4.70403135e-01 -1.74959153e-01
1.04719663e+00 -1.15744174e+00 -5.43118119e-01 6.63854182e-01
-1.03447461e+00 -9.47080135e-01 -1.32439882e-01 2.55188316e-01
-5.42472124e-01 2.70740002e-01 -5.20804584e-01 -6.07808590e-01
-3.10017347e-01 -7.56490827e-01 1.12042165e+00 2.09186509e-01
2.41640843e-02 -1.10679388e+00 -1.07708229e-02 5.55674851e-01
2.13697255e-01 3.62823457e-01 8.58663797e-01 -5.69870234e-01
-5.10622680e-01 2.16562957e-01 -8.00269961e-01 4.24542785e-01
-8.81038532e-02 2.62172699e-01 -9.66474593e-01 -3.35081547e-01
-7.77603686e-01 -7.18471944e-01 1.57770073e+00 2.39376664e-01
1.22656012e+00 1.72177806e-01 -3.99251938e-01 9.49209094e-01
1.30389369e+00 -2.20229894e-01 7.11282611e-01 2.51962751e-01
1.32441425e+00 6.61017716e-01 8.89169455e-01 2.14870065e-01
8.78995836e-01 8.84929121e-01 7.78752923e-01 -4.57292527e-01
-7.48921633e-01 -1.18472636e-01 1.46027029e-01 5.56973100e-01
-4.82226193e-01 -7.40441918e-01 -1.12032962e+00 6.09845638e-01
-2.28918219e+00 -6.48144722e-01 -5.56597471e-01 1.91902971e+00
7.27903426e-01 -1.76602200e-01 1.46144196e-01 -4.01503384e-01
5.64812422e-01 4.25216466e-01 -2.59862691e-01 -1.99670002e-01
-5.02306819e-01 1.31769076e-01 4.91652250e-01 2.99741417e-01
-1.38921988e+00 1.41665554e+00 5.40906763e+00 7.10758984e-01
-7.71520674e-01 4.25630016e-03 4.79681611e-01 1.53760940e-01
-1.17357545e-01 -1.21732935e-01 -9.74511921e-01 -1.43698171e-01
5.50385892e-01 1.77586064e-01 3.49704653e-01 7.27925062e-01
-1.68979511e-01 4.32009948e-03 -1.31372082e+00 1.16547310e+00
2.91303843e-01 -1.00711775e+00 1.56330958e-01 -2.33903751e-01
5.42880476e-01 4.07431513e-01 1.45389140e-01 3.92441750e-01
6.35698915e-01 -1.19025135e+00 7.30721653e-01 4.90028024e-01
8.09633791e-01 -5.48628509e-01 8.23561847e-01 -3.67562808e-02
-1.42344666e+00 1.95223227e-01 -4.85424280e-01 1.18179604e-01
-7.01219141e-02 4.15673345e-01 -4.30955023e-01 8.78442228e-01
1.14338160e+00 1.32578158e+00 -1.11971831e+00 1.20742571e+00
-5.54162383e-01 5.85749567e-01 -7.07530975e-02 2.45113656e-01
1.03954710e-01 -2.04385906e-01 5.02461195e-01 1.31325829e+00
-3.97025794e-02 -7.33824074e-02 1.77930146e-01 8.79849374e-01
-2.94097215e-01 6.03052042e-02 -6.25561595e-01 -4.63436283e-02
5.17616346e-02 1.44472122e+00 -6.42649889e-01 -1.20660506e-01
-5.74464738e-01 7.95842290e-01 7.94135273e-01 5.19079387e-01
-1.06705296e+00 -8.75450522e-02 5.67761779e-01 -1.23088248e-01
5.81354618e-01 -1.50329396e-01 6.10681362e-02 -1.22165012e+00
-1.49520054e-01 -6.95805252e-01 7.12926388e-01 -1.29448128e+00
-1.67722857e+00 7.93161750e-01 3.23518723e-01 -8.77391398e-01
1.75362363e-01 -7.29204118e-01 -3.38848263e-01 6.84812963e-01
-1.88470781e+00 -1.94121945e+00 -7.00716913e-01 7.57196069e-01
3.24759573e-01 -5.42272851e-02 6.71307564e-01 2.57316619e-01
-6.28482521e-01 7.17436433e-01 -1.42327085e-01 2.34062493e-01
1.01092041e+00 -1.44701469e+00 5.34591198e-01 6.95736229e-01
3.93574268e-01 3.19523811e-01 6.09532177e-01 -6.01974487e-01
-9.92898941e-01 -1.54326582e+00 6.99185252e-01 -6.14417017e-01
9.08594847e-01 -6.60743654e-01 -1.17492843e+00 1.11313927e+00
2.79449642e-01 5.87170064e-01 4.01707828e-01 2.28778929e-01
-9.34299171e-01 -1.17639348e-01 -5.86594105e-01 6.83185756e-01
1.36349368e+00 -5.08902073e-01 -6.50358975e-01 3.28459918e-01
7.83808768e-01 -3.99739325e-01 -7.40547836e-01 6.01325631e-01
4.39668506e-01 -8.13080788e-01 1.18023849e+00 -6.14871919e-01
6.00167453e-01 -4.54429001e-01 -3.69384825e-01 -1.27886152e+00
-4.24487799e-01 -3.74833643e-01 -2.79241651e-01 1.39709747e+00
4.41127181e-01 -5.42168617e-01 4.28199738e-01 1.08290970e-01
-1.43905312e-01 -6.92283988e-01 -5.95182180e-01 -9.64693606e-01
5.23377992e-02 -2.56348431e-01 6.81761280e-02 8.54494512e-01
-7.70639718e-01 7.21532583e-01 -6.19071364e-01 2.50537962e-01
8.44027817e-01 4.48097765e-01 9.26098943e-01 -1.18834043e+00
-4.59903985e-01 -3.25652421e-01 -3.51498514e-01 -8.87003064e-01
3.12575817e-01 -1.23887920e+00 1.14637718e-01 -1.93112791e+00
5.13150454e-01 -4.01365727e-01 -2.94865698e-01 9.11229193e-01
-2.83839136e-01 5.53307712e-01 4.11570787e-01 9.44075435e-02
-8.10902715e-01 8.14298689e-01 1.38021266e+00 -2.30578884e-01
4.09378037e-02 8.36672336e-02 -5.11137903e-01 6.88893497e-01
5.92770457e-01 -2.94633150e-01 -4.67417777e-01 -4.76083964e-01
4.40083265e-01 -3.14397991e-01 9.77552116e-01 -5.46527326e-01
9.91082042e-02 1.39779702e-01 9.02016368e-03 -7.26069093e-01
4.80993390e-01 -6.89502180e-01 -1.06786229e-01 -1.03394836e-01
-4.79255915e-01 9.31439996e-02 2.01042607e-01 5.98540843e-01
-2.47714013e-01 1.53125614e-01 8.28990400e-01 9.86482576e-02
-9.98088658e-01 5.02209067e-01 2.20464662e-01 3.77137154e-01
1.04003024e+00 1.80125639e-01 -8.44144046e-01 -3.97438169e-01
-8.44771981e-01 5.43425143e-01 2.73903549e-01 7.13447571e-01
6.83974206e-01 -1.08447468e+00 -1.07065320e+00 -2.75715828e-01
7.48239279e-01 1.57187164e-01 1.57726705e-01 9.86209810e-01
-4.03464973e-01 8.08769464e-02 -1.91795379e-01 -5.95799029e-01
-1.74514878e+00 8.25940609e-01 3.62375647e-01 -2.73296624e-01
-8.57828200e-01 1.25286007e+00 7.20407784e-01 -4.30983484e-01
5.04130542e-01 -3.42476785e-01 -5.84003508e-01 3.26907575e-01
3.79119277e-01 1.43056855e-01 2.89032124e-02 -9.32014644e-01
-5.31674445e-01 8.52535009e-01 -1.40415713e-01 3.14225018e-01
1.48839796e+00 -1.11530326e-01 -3.75771016e-01 6.31226242e-01
1.26622987e+00 -2.62716174e-01 -1.18048835e+00 -4.41757858e-01
-1.29619151e-01 -2.74517387e-01 -8.44844952e-02 -1.00875175e+00
-1.27296209e+00 9.39280510e-01 4.72951353e-01 -8.31988800e-05
1.10047114e+00 6.15974903e-01 3.96072775e-01 2.66098082e-01
1.84211597e-01 -5.24581254e-01 1.02019235e-01 6.92803442e-01
1.44190896e+00 -1.32776654e+00 -4.16106991e-02 -8.53233159e-01
-9.60583627e-01 7.29893446e-01 7.58589149e-01 -1.92472693e-02
5.90903997e-01 1.66083515e-01 1.06269531e-01 -2.73038745e-01
-9.29258227e-01 -9.03298259e-01 5.72984219e-01 8.39843214e-01
4.24617350e-01 1.72640160e-01 2.80669760e-02 3.46015751e-01
-1.37214646e-01 -1.75311059e-01 3.36336523e-01 5.22996962e-01
1.14259593e-01 -9.79696155e-01 1.50025534e-02 2.75454283e-01
-2.60815769e-01 -3.27107638e-01 -6.86120272e-01 9.84334528e-01
7.80584216e-02 8.79057109e-01 -2.37995349e-02 -2.88801938e-01
5.52053869e-01 -4.07803237e-01 5.99332809e-01 -5.54367959e-01
-5.16897261e-01 -1.63443103e-01 4.43621248e-01 -6.00799978e-01
-6.08764052e-01 -3.31544995e-01 -1.16427970e+00 -2.12794878e-02
-1.83035672e-01 -1.70902401e-01 2.78284371e-01 6.17056847e-01
5.40145226e-02 8.13428640e-01 9.44756493e-02 -6.98563635e-01
-1.62656099e-01 -8.91455531e-01 -5.69444060e-01 5.68251789e-01
3.21234316e-01 -7.79228091e-01 -1.89649731e-01 -2.33226717e-02]
|
[10.472249984741211, 1.6637970209121704]
|
600e1806-1866-435a-a526-5b5b2086d50c
|
generative-models-for-graph-based-protein
| null | null |
http://papers.nips.cc/paper/9711-generative-models-for-graph-based-protein-design
|
http://papers.nips.cc/paper/9711-generative-models-for-graph-based-protein-design.pdf
|
Generative Models for Graph-Based Protein Design
|
Engineered proteins offer the potential to solve many problems in biomedicine, energy, and materials science, but creating designs that succeed is difficult in practice. A significant aspect of this challenge is the complex coupling between protein sequence and 3D structure, with the task of finding a viable design often referred to as the inverse protein folding problem. We develop relational language models for protein sequences that directly condition on a graph specification of the target structure. Our approach efficiently captures the complex dependencies in proteins by focusing on those that are long-range in sequence but local in 3D space. Our framework significantly improves in both speed and robustness over conventional and deep-learning-based methods for structure-based protein sequence design, and takes a step toward rapid and targeted biomolecular design with the aid of deep generative models.
|
['Vikas Garg', 'Regina Barzilay', 'Tommi Jaakkola', 'John Ingraham']
|
2019-12-01
| null |
https://openreview.net/forum?id=SJgxrLLKOE
|
https://openreview.net/pdf?id=SJgxrLLKOE
|
iclr-workshop-deepgenstruct-2019
|
['protein-design']
|
['medical']
|
[ 3.59500319e-01 -4.60088477e-02 -3.15042175e-02 -2.00949147e-01
-4.38029438e-01 -8.01336348e-01 3.46023500e-01 4.06578988e-01
-8.02653953e-02 9.09259737e-01 3.78464848e-01 -6.83259785e-01
-1.60580799e-01 -7.17511475e-01 -1.11985159e+00 -9.10958171e-01
-8.02397728e-02 5.13186157e-01 -2.56381303e-01 -3.81153166e-01
2.68362612e-01 8.37193310e-01 -8.35267067e-01 2.59767920e-01
7.54402995e-01 3.37216228e-01 4.68974948e-01 6.36698186e-01
-2.68222928e-01 5.68923831e-01 -2.26727754e-01 -1.11959778e-01
-2.11431995e-01 -8.09513330e-01 -7.34462976e-01 -2.20460907e-01
-4.67952527e-02 4.41819787e-01 -1.29989132e-01 5.59111178e-01
6.34081542e-01 4.03383784e-02 8.89951527e-01 -4.06183600e-01
-1.06684446e+00 3.49317752e-02 -2.69769907e-01 -1.40044525e-01
4.76865411e-01 4.66621161e-01 1.09676182e+00 -8.79481077e-01
8.98586154e-01 1.14868689e+00 6.46427870e-01 6.55992806e-01
-1.88426220e+00 -2.09081873e-01 -4.85393181e-02 6.43393397e-02
-1.07618487e+00 -3.58853132e-01 6.24907136e-01 -7.02877045e-01
1.65138483e+00 8.81951582e-03 6.76381826e-01 9.60839033e-01
8.57083321e-01 1.30402699e-01 7.41565228e-01 -2.15041876e-01
3.50196391e-01 -7.40828097e-01 5.81009574e-02 8.19418550e-01
1.17563382e-01 5.04418872e-02 -5.73533237e-01 -5.54093063e-01
4.28054899e-01 2.05824912e-01 -2.02795640e-01 -7.73541629e-01
-1.07942080e+00 8.52109134e-01 4.42732990e-01 2.05570012e-01
-5.06324053e-01 3.08984846e-01 -8.56198072e-02 3.81424092e-02
7.68265203e-02 8.51535141e-01 -7.40661740e-01 1.84277613e-02
-4.60415870e-01 5.98116279e-01 8.91151071e-01 7.57565260e-01
8.20591748e-01 -3.30706805e-01 9.69449431e-02 4.11412954e-01
3.59014452e-01 2.43188649e-01 -5.42850718e-02 -5.62873006e-01
-9.53176916e-02 6.41079664e-01 1.56941056e-01 -7.46048808e-01
-6.55612230e-01 -3.27855945e-01 -5.10810256e-01 1.08482637e-01
2.81832963e-01 1.12132065e-01 -1.04788840e+00 1.89935088e+00
4.67390805e-01 -3.02353203e-01 1.48622943e-02 6.38198018e-01
7.05524206e-01 8.13243151e-01 3.68541747e-01 -3.78587544e-01
1.23648202e+00 -4.62870657e-01 -4.98461455e-01 -1.51382998e-01
6.18824244e-01 -6.17630839e-01 8.33701253e-01 8.21323395e-02
-9.88056958e-01 -1.96776494e-01 -9.68329847e-01 -4.48479235e-01
-3.38477910e-01 -4.17314649e-01 7.33234584e-01 4.30413783e-01
-8.61131012e-01 6.85087919e-01 -7.13581562e-01 -2.92855382e-01
3.28807950e-01 7.07691133e-01 -4.99563068e-01 -1.10618018e-01
-8.93225908e-01 1.24838853e+00 2.37403154e-01 -2.27719289e-03
-7.52478957e-01 -9.93719518e-01 -6.33758247e-01 9.70403180e-02
2.41865441e-01 -9.17637587e-01 8.00656676e-01 -1.88706502e-01
-1.42110193e+00 8.29646051e-01 -4.70394701e-01 -2.84746855e-01
9.64164138e-02 -6.88836630e-03 1.22247897e-01 -2.67149001e-01
-2.48570517e-01 5.59828937e-01 2.58454293e-01 -8.14366162e-01
2.57578194e-01 -5.39671719e-01 -2.33441174e-01 8.02655444e-02
2.50311673e-01 -1.03921182e-01 7.14494810e-02 -4.39775378e-01
7.88424462e-02 -1.07862878e+00 -7.73759186e-01 1.20170258e-01
-3.69763315e-01 -1.47572935e-01 4.51225847e-01 -5.61874986e-01
8.28472018e-01 -1.59260488e+00 8.29746544e-01 3.48651022e-01
5.19510090e-01 3.16515326e-01 -1.26829684e-01 1.01678360e+00
-2.28557006e-01 1.64077014e-01 -1.61968485e-01 4.24547195e-01
-1.63197696e-01 1.35054171e-01 -7.37831742e-02 4.68244553e-01
4.02802467e-01 1.35755968e+00 -8.60577285e-01 2.50358917e-02
3.62681113e-02 8.16002727e-01 -8.35189164e-01 2.36138642e-01
-9.07204509e-01 7.83442080e-01 -6.82768941e-01 3.69187891e-01
3.70108277e-01 -6.25052869e-01 9.15159225e-01 -3.96695524e-01
-1.60327867e-01 4.30475473e-01 -3.45307976e-01 1.81378138e+00
-1.11185960e-01 9.61501971e-02 -2.34998852e-01 -9.54895020e-01
1.03008246e+00 8.19933936e-02 8.07667255e-01 -4.09132123e-01
-1.32354796e-01 -1.24246664e-02 3.12956393e-01 -4.92337674e-01
1.01711880e-02 -4.34838265e-01 3.12096272e-02 4.61603045e-01
-1.21543512e-01 -8.57557505e-02 -6.25987723e-02 -1.16583519e-02
1.31518376e+00 6.95278823e-01 4.03669059e-01 -4.25691426e-01
2.43906081e-01 1.26918465e-01 6.62266791e-01 4.02359396e-01
1.86557293e-01 3.65074456e-01 8.90152276e-01 -6.26771688e-01
-1.50627184e+00 -9.26772714e-01 1.37875676e-01 9.39718962e-01
-8.15024078e-02 -6.77899241e-01 -6.86283231e-01 -4.92392391e-01
1.58645317e-01 2.38439694e-01 -5.24383545e-01 -4.73037630e-01
-7.62723446e-01 -1.06252635e+00 2.24056885e-01 1.34483889e-01
-4.10933405e-01 -9.87772346e-01 -2.88356960e-01 6.17103338e-01
6.12038895e-02 -7.04207838e-01 -7.34565079e-01 6.22832179e-01
-6.66905165e-01 -1.16690361e+00 -5.02940416e-01 -7.12696791e-01
7.18054593e-01 4.56414670e-02 1.12131679e+00 3.43936458e-02
-7.84303129e-01 -2.14684308e-01 1.00140698e-01 -2.35627368e-01
-7.38125026e-01 3.04250829e-02 -5.36945276e-02 -3.52109790e-01
4.38324839e-01 -7.57995367e-01 -8.01643014e-01 5.17455749e-02
-8.85924518e-01 1.79699525e-01 4.81502503e-01 1.01202309e+00
8.62110913e-01 -4.53787178e-01 7.52690971e-01 -8.98890436e-01
6.72876060e-01 -3.13888371e-01 -5.69944203e-01 5.77800274e-01
-6.93156481e-01 8.12374353e-01 6.45301223e-01 -2.50825077e-01
-5.95282137e-01 6.00777984e-01 -3.37987930e-01 1.26032397e-01
5.50598167e-02 5.99422812e-01 -5.21124184e-01 -2.01061398e-01
6.54481888e-01 1.78508729e-01 3.60559672e-01 -6.15059614e-01
5.00350237e-01 1.45286471e-01 9.32345614e-02 -8.01880002e-01
3.14890593e-01 8.08823332e-02 5.47508419e-01 -8.17921162e-01
-4.14911777e-01 1.34185543e-02 -5.42359352e-01 1.70751065e-01
1.01969767e+00 -6.10826552e-01 -1.23106873e+00 1.64043516e-01
-1.13055491e+00 -2.43274093e-01 5.41286469e-02 5.48573546e-02
-7.18893349e-01 4.07677412e-01 -6.02968514e-01 -4.39396948e-01
-3.22166413e-01 -1.47977173e+00 1.11413610e+00 -1.52651265e-01
-5.47297597e-01 -9.12730098e-01 3.32928568e-01 2.07904458e-01
2.34989941e-01 5.48172355e-01 1.72576845e+00 -2.76813239e-01
-9.21323657e-01 1.36199862e-01 4.48013358e-02 -1.09690599e-01
3.62092108e-01 -1.48805216e-01 -3.77292544e-01 -1.08912654e-01
-2.57515430e-01 -2.62776583e-01 8.62594903e-01 4.90904212e-01
7.15334892e-01 -2.74789691e-01 -5.05880177e-01 4.48235750e-01
1.30015457e+00 3.64152044e-01 5.09371996e-01 -1.64822564e-01
9.41368639e-01 5.89742601e-01 1.66487932e-01 1.07612409e-01
8.93277079e-02 8.55038822e-01 5.58079593e-02 -1.69279158e-01
-7.25837564e-03 -6.63776577e-01 2.08297536e-01 4.48119015e-01
3.55625786e-02 -3.06900114e-01 -9.91198957e-01 7.79015869e-02
-1.95030320e+00 -8.61431360e-01 -6.70462698e-02 1.96885705e+00
1.15248823e+00 -1.52497008e-01 8.24026167e-02 -4.26612139e-01
3.85865837e-01 -1.61583453e-01 -1.01385307e+00 -4.28715914e-01
-3.87713820e-01 4.98054832e-01 4.59288508e-01 6.93203688e-01
-6.75564408e-01 9.09124970e-01 7.55860233e+00 3.74002218e-01
-9.39196706e-01 -2.89090037e-01 6.79052413e-01 -7.91249768e-05
-7.05996752e-01 2.62496322e-01 -6.82602763e-01 3.54833722e-01
9.98490751e-01 -4.35659550e-02 5.90681553e-01 3.20791721e-01
5.87125599e-01 3.78464818e-01 -1.26857948e+00 6.19620919e-01
-2.91027337e-01 -2.03340816e+00 -3.43691860e-03 3.33013862e-01
6.31202817e-01 -4.58585937e-03 -8.25104043e-02 -4.46222186e-01
4.73612338e-01 -1.50895524e+00 2.83652872e-01 6.76081479e-01
6.34155571e-01 -8.55961263e-01 8.03600922e-02 3.53932172e-01
-7.78250635e-01 2.76365429e-01 -2.30055079e-01 1.49287954e-01
7.60588050e-02 6.86130285e-01 -8.61105561e-01 4.49768417e-02
1.97249040e-01 6.86786532e-01 8.64843801e-02 4.28700328e-01
-8.82179141e-02 1.76205635e-01 -1.83594525e-01 -3.47300410e-01
-7.87172243e-02 -5.60586393e-01 2.84792632e-01 1.04457963e+00
9.26426351e-02 2.68351346e-01 2.32070893e-01 1.30956841e+00
-1.30821094e-01 -3.83439586e-02 -7.26996422e-01 -5.53504646e-01
1.90670416e-01 7.85496831e-01 -5.31910419e-01 1.84326515e-01
-2.41294041e-01 6.52460992e-01 4.52539980e-01 5.75002074e-01
-7.00917721e-01 -1.41694052e-02 1.06801879e+00 2.04108998e-01
4.62838322e-01 -7.15046585e-01 -1.19807750e-01 -8.74367774e-01
-1.08462088e-01 -1.15206742e+00 -1.12232566e-01 -5.69037318e-01
-1.16443288e+00 7.45852217e-02 -5.79385638e-01 -3.42965722e-01
-1.45321488e-01 -8.51565778e-01 -6.51203394e-02 1.12445641e+00
-1.02542305e+00 -1.16144657e+00 4.80336905e-01 2.40512043e-02
4.27702963e-02 8.43545720e-02 1.04413497e+00 5.05983494e-02
-3.67958993e-01 2.41128460e-01 3.62487435e-01 -5.53555667e-01
3.83856297e-01 -9.87593949e-01 9.41569865e-01 3.59891534e-01
-8.33148956e-02 1.02403128e+00 1.03245771e+00 -9.32197809e-01
-2.20833182e+00 -9.76773798e-01 1.13184559e+00 -7.62197137e-01
3.83404046e-01 -6.63457334e-01 -8.95802796e-01 3.60806584e-01
-2.00294733e-01 -2.74388939e-01 8.92644942e-01 1.35465816e-01
-4.55588192e-01 3.54714066e-01 -1.02970386e+00 5.13456047e-01
1.36026084e+00 -7.38120675e-01 -1.60783485e-01 6.87399149e-01
9.78781819e-01 -2.84392744e-01 -1.05230200e+00 2.93636948e-01
5.36548913e-01 -5.51987469e-01 1.25599444e+00 -1.28232479e+00
3.30363929e-01 -4.60532278e-01 -4.95946258e-02 -1.02757943e+00
-7.43228018e-01 -1.04557502e+00 -1.72527552e-01 5.47359884e-01
7.94189155e-01 -3.00991237e-01 1.04430091e+00 7.69180477e-01
-9.27692056e-02 -9.57730651e-01 -6.38412833e-01 -4.39823955e-01
3.92999291e-01 8.53136852e-02 5.89696169e-01 6.38066053e-01
1.76740348e-01 7.39289939e-01 -3.59874517e-01 -2.42445946e-01
4.10262913e-01 3.93608332e-01 5.13849437e-01 -9.72855210e-01
-5.99853396e-01 -3.64883602e-01 -3.92255247e-01 -1.19669664e+00
7.63317645e-02 -9.25589323e-01 1.02647819e-01 -1.67586267e+00
3.27293694e-01 -1.93066746e-01 -1.62032440e-01 3.54276836e-01
-1.29169419e-01 -2.35638887e-01 -1.55869603e-01 3.03620822e-04
-3.58522326e-01 6.70441449e-01 1.28477156e+00 -1.25443041e-01
-2.26361826e-01 -2.94048250e-01 -8.62212777e-01 1.16244540e-01
5.50559878e-01 -4.94129300e-01 -2.98982769e-01 -1.22217536e-01
6.32966757e-01 2.35643968e-01 1.74634635e-01 -2.72475749e-01
-1.59101151e-02 -4.93756413e-01 3.87356102e-01 -3.70375901e-01
3.20603967e-01 -6.06670201e-01 6.45522177e-01 5.86021543e-01
-6.23350203e-01 1.82532802e-01 -7.89972302e-03 7.51401007e-01
3.82757694e-01 2.41522565e-01 8.18414748e-01 -2.24587977e-01
-4.33628820e-02 4.54653710e-01 -5.22014141e-01 -4.80711460e-02
8.80745530e-01 -9.58748255e-03 -2.24550769e-01 -1.45424277e-01
-9.02760565e-01 -1.67224146e-02 8.32931340e-01 3.23208064e-01
6.76866829e-01 -9.69778776e-01 -5.06900787e-01 2.20386580e-01
2.62318440e-02 -1.56889915e-01 -3.04863714e-02 4.50131148e-01
-7.47862041e-01 7.37726629e-01 2.57182084e-02 -4.96150851e-01
-1.17942739e+00 7.79204965e-01 5.67794919e-01 -2.02714667e-01
-5.03955960e-01 6.55236900e-01 5.09667158e-01 -4.20939237e-01
-1.52779207e-01 -1.22007690e-01 2.44513527e-01 -5.00738084e-01
3.10531437e-01 -9.66577381e-02 7.72443116e-02 -5.48502862e-01
-5.36193073e-01 5.82848787e-01 -1.64163262e-01 3.99467528e-01
1.66150939e+00 2.30604082e-01 -4.55727667e-01 -5.29674639e-04
1.21892929e+00 -3.58356655e-01 -1.26385939e+00 -2.31748283e-01
2.98301518e-01 8.77798945e-02 -2.10294425e-01 -8.11894834e-01
-5.59890211e-01 5.95399201e-01 4.82006133e-01 -3.80641103e-01
6.18366063e-01 2.53121287e-01 8.57861042e-01 7.61830389e-01
3.34471643e-01 -7.15950489e-01 -2.01087091e-02 4.89777774e-01
6.79707348e-01 -8.28435361e-01 1.80311263e-01 -2.57969081e-01
-1.62726775e-01 9.96666312e-01 1.63668036e-01 -5.25641348e-03
6.28739059e-01 5.02557814e-01 -1.91018909e-01 -4.85145897e-01
-9.96895432e-01 7.95833915e-02 2.85290748e-01 5.85886419e-01
1.01822460e+00 6.88123032e-02 -4.10108984e-01 3.09696853e-01
2.50383496e-01 9.54887569e-02 -3.03585455e-02 1.24384356e+00
-5.66960752e-01 -1.83883095e+00 -5.70745282e-02 1.88070461e-01
-4.82936442e-01 -2.10472584e-01 -8.76215518e-01 1.18024215e-01
-1.79648288e-02 7.94945717e-01 -4.13580894e-01 -3.30955833e-01
2.95902044e-01 1.32725477e-01 9.08813953e-01 -5.54256499e-01
-5.10786355e-01 2.50657409e-01 3.37630510e-02 -4.32070613e-01
-2.64769942e-01 -5.83003819e-01 -1.55638611e+00 -4.79418337e-01
-2.45640814e-01 3.52541983e-01 8.16802979e-01 9.19243217e-01
1.08058012e+00 5.04476726e-01 4.02170092e-01 -4.97146696e-01
-4.42951322e-01 -3.10273230e-01 -8.37278739e-02 3.11045736e-01
4.21513170e-01 -4.58293110e-01 3.54834348e-01 2.34241009e-01]
|
[4.752908706665039, 5.610057353973389]
|
eb3acb4f-3ee2-4a1f-8bf0-0d4285e4461e
|
using-textual-interface-to-align-external
|
2305.13710
| null |
https://arxiv.org/abs/2305.13710v1
|
https://arxiv.org/pdf/2305.13710v1.pdf
|
Using Textual Interface to Align External Knowledge for End-to-End Task-Oriented Dialogue Systems
|
Traditional end-to-end task-oriented dialogue systems have been built with a modularized design. However, such design often causes misalignment between the agent response and external knowledge, due to inadequate representation of information. Furthermore, its evaluation metrics emphasize assessing the agent's pre-lexicalization response, neglecting the quality of the completed response. In this work, we propose a novel paradigm that uses a textual interface to align external knowledge and eliminate redundant processes. We demonstrate our paradigm in practice through MultiWOZ-Remake, including an interactive textual interface built for the MultiWOZ database and a correspondingly re-processed dataset. We train an end-to-end dialogue system to evaluate this new dataset. The experimental results show that our approach generates more natural final responses and achieves a greater task success rate compared to the previous models.
|
['Zhou Yu', 'Derek Chen', 'Deema Alnuhait', 'Qingyang Wu']
|
2023-05-23
| null | null | null | null |
['task-oriented-dialogue-systems']
|
['natural-language-processing']
|
[-6.06548041e-02 4.97528762e-01 2.37123802e-01 -5.58054447e-01
-8.84557426e-01 -6.55934453e-01 8.33952785e-01 2.39175960e-01
-8.06906402e-01 8.14073086e-01 5.48116148e-01 -8.26464593e-02
6.13301322e-02 -7.04579294e-01 4.48638089e-02 -7.12651312e-02
6.20107949e-01 1.14546871e+00 3.88062805e-01 -7.12252140e-01
4.82855946e-01 -3.21578383e-01 -1.11429048e+00 5.74507415e-01
9.70823586e-01 3.92157614e-01 4.02002066e-01 6.59749687e-01
-4.16328341e-01 9.93819356e-01 -1.05838799e+00 -8.18487644e-01
9.00937468e-02 -4.82787192e-01 -1.56520104e+00 -4.75902893e-02
9.67717171e-02 -4.46535736e-01 1.29565224e-02 8.90067935e-01
6.59599185e-01 2.94205904e-01 4.84569997e-01 -9.25354004e-01
-3.88616025e-01 7.85982668e-01 6.26791792e-04 -1.21589504e-01
1.03367174e+00 4.28937823e-01 9.05439675e-01 -8.33870590e-01
8.82225037e-01 1.31919098e+00 5.11922419e-01 8.37356031e-01
-1.32720113e+00 -2.22635657e-01 3.19390744e-02 1.29001930e-01
-8.86001408e-01 -6.14694893e-01 6.09924793e-01 -4.66177851e-01
1.29841053e+00 5.01554191e-01 2.37528756e-01 1.12825882e+00
-1.00348018e-01 8.24648201e-01 1.12994909e+00 -6.25214636e-01
-7.26402625e-02 5.40739357e-01 4.41745490e-01 5.56256592e-01
-4.27673072e-01 -3.66532147e-01 -9.05237556e-01 -2.50203699e-01
3.23869556e-01 -5.10115981e-01 -2.73435056e-01 -7.99281299e-02
-1.15685928e+00 7.32088745e-01 -1.81515291e-01 2.76715338e-01
-4.83400524e-01 -4.59356159e-01 6.66600049e-01 7.59535193e-01
4.89121586e-01 8.95209193e-01 -5.98485112e-01 -4.82354075e-01
-2.93540627e-01 5.53234577e-01 1.44196463e+00 9.54421163e-01
6.76740766e-01 -2.70102769e-01 -5.32878339e-01 1.13887930e+00
1.58701688e-01 2.72545397e-01 8.90072167e-01 -8.04750443e-01
8.19321752e-01 8.95602345e-01 4.07946259e-01 -8.34026814e-01
-6.42735004e-01 -3.11131086e-02 -4.30162251e-01 -4.86986004e-02
7.37975001e-01 -2.62485355e-01 -1.42365769e-01 1.81018651e+00
4.63944674e-01 -7.87063062e-01 5.05333602e-01 8.86632621e-01
1.04938936e+00 4.24053878e-01 2.90109098e-01 -3.57016504e-01
1.48410821e+00 -1.27609003e+00 -1.25990772e+00 -2.07482889e-01
9.45110798e-01 -1.05058670e+00 1.61906958e+00 4.56563592e-01
-1.26709151e+00 -6.24682128e-01 -6.68193519e-01 -1.03432789e-01
-2.12231129e-01 1.41910329e-01 3.46966207e-01 4.06027466e-01
-9.84096706e-01 3.53682101e-01 -1.48543075e-01 -4.13539261e-01
-5.63764870e-01 1.59420267e-01 -3.71856630e-01 5.07835686e-01
-1.48008871e+00 1.36429632e+00 4.61298317e-01 -1.84263065e-01
-5.56711674e-01 -4.81877953e-01 -6.65420234e-01 -1.28308699e-01
7.13859379e-01 -8.48056614e-01 2.16964078e+00 -7.88905799e-01
-2.31665039e+00 8.37136328e-01 1.96073595e-02 -1.28574952e-01
8.32974732e-01 -4.95714545e-01 -2.01810915e-02 -1.81070715e-01
1.87000394e-01 2.68426090e-01 4.84779447e-01 -1.04012084e+00
-8.15923810e-01 -4.07053798e-01 2.30938748e-01 6.49231434e-01
-2.42651209e-01 4.49127853e-02 -3.20449919e-01 -4.18082863e-01
-8.59324783e-02 -6.27932429e-01 -1.14717357e-01 -8.88179302e-01
-3.64914745e-01 -7.76038110e-01 3.93950731e-01 -6.15420043e-01
1.46824169e+00 -1.70893097e+00 3.53811085e-01 -1.16228491e-01
4.69237834e-01 3.18542928e-01 -2.39781901e-01 1.11558914e+00
2.46922940e-01 -5.67598129e-03 1.05740070e-01 -4.06751096e-01
1.94322422e-01 -2.90960342e-01 -1.31878123e-01 -2.27712110e-01
3.39732096e-02 7.69416809e-01 -1.01500475e+00 -4.01947916e-01
-3.01390775e-02 -1.93478808e-01 -4.55391288e-01 7.72749245e-01
-4.62107986e-01 2.55257279e-01 -5.63555300e-01 9.77628380e-02
2.84024268e-01 8.79393704e-03 3.47234279e-01 -1.27098382e-01
-1.89337000e-01 6.60747051e-01 -8.44271719e-01 1.98353863e+00
-5.75262368e-01 2.75633693e-01 1.10395730e-01 -2.68121272e-01
1.09005737e+00 5.43736577e-01 3.62353064e-02 -9.19389069e-01
2.07454026e-01 3.67756337e-02 7.51521215e-02 -9.93900120e-01
9.47099209e-01 1.93455815e-03 -2.80560076e-01 8.62561643e-01
1.84581056e-01 -2.57698923e-01 3.10438335e-01 5.49675405e-01
9.84052122e-01 1.67375430e-01 4.33292300e-01 -2.02463958e-02
6.85988188e-01 4.24553424e-01 2.28171349e-01 8.32379282e-01
-2.25784123e-01 3.96110341e-02 4.75696415e-01 -5.16951442e-01
-8.91591549e-01 -5.77825189e-01 3.74863595e-01 1.37818098e+00
9.37842429e-02 -7.52007186e-01 -1.08505106e+00 -8.52167010e-01
-4.42031235e-01 9.08487022e-01 -2.33714193e-01 -1.99576288e-01
-4.42471743e-01 -4.04016674e-01 8.57787609e-01 5.22736236e-02
7.74967849e-01 -1.15827477e+00 -8.39530408e-01 6.35435283e-01
-9.38951373e-01 -8.87606382e-01 -5.20010650e-01 -1.14202842e-01
-5.34418583e-01 -1.09740829e+00 -6.80886582e-02 -6.71646237e-01
1.51818171e-01 2.03054383e-01 1.37363756e+00 1.45013735e-01
2.28195831e-01 3.35143983e-01 -6.01546764e-01 -2.17073232e-01
-7.89128065e-01 3.33250642e-01 -3.01437043e-02 -1.26926169e-01
6.58029377e-01 -9.93071645e-02 -2.91787595e-01 4.69185889e-01
-6.58614933e-01 5.75549781e-01 3.55095923e-01 1.05314124e+00
-1.99575156e-01 -4.44150507e-01 9.71511006e-01 -1.21122217e+00
1.81642258e+00 -1.04541302e-01 -2.72369385e-01 4.01274234e-01
-6.05976462e-01 2.22277433e-01 6.27588451e-01 -3.81400347e-01
-1.69203591e+00 -1.47556394e-01 -2.50037253e-01 3.08809847e-01
-2.13326141e-01 7.15724051e-01 -2.48660240e-02 3.25888991e-01
9.60536957e-01 1.37289479e-01 3.08553934e-01 -4.73507732e-01
4.07782227e-01 1.02157116e+00 5.48805416e-01 -9.01007593e-01
3.93353134e-01 -2.71772206e-01 -9.27296579e-01 -4.78341043e-01
-5.81485093e-01 -5.53637028e-01 -4.78469700e-01 -5.50812006e-01
5.72837591e-01 -6.30918980e-01 -1.17242813e+00 5.95459878e-01
-1.64003098e+00 -6.91277862e-01 -1.39270231e-01 5.29525757e-01
-7.12069929e-01 2.65838623e-01 -8.18762481e-01 -8.11078131e-01
-6.69712365e-01 -1.20621872e+00 6.88252330e-01 2.16562271e-01
-9.03571546e-01 -9.06381249e-01 5.06523013e-01 6.85515285e-01
5.68756402e-01 -2.96797127e-01 9.72263873e-01 -1.25768483e+00
-9.21216160e-02 -4.57104743e-02 -5.52571900e-02 -2.13157479e-02
6.63451776e-02 -1.85851470e-01 -9.60483015e-01 2.06117472e-03
4.13938850e-01 -8.87460411e-01 2.28044987e-01 -4.46797132e-01
2.15149432e-01 -5.11323392e-01 -1.08184971e-01 -2.05819860e-01
9.90034401e-01 3.61180902e-01 4.79347289e-01 5.04453003e-01
2.13933259e-01 1.25094116e+00 8.21693301e-01 4.28042263e-01
7.84296691e-01 9.43670571e-01 -1.06848842e-02 -2.68444046e-02
-6.86123818e-02 -2.50663728e-01 4.18378651e-01 1.08140457e+00
1.82464167e-01 -2.74535894e-01 -9.14604902e-01 4.28102493e-01
-2.22570467e+00 -1.04478729e+00 -2.14605644e-01 2.08971596e+00
1.44976640e+00 -1.63129289e-02 1.49949163e-01 -3.30332011e-01
5.33582211e-01 -1.06212437e-01 -2.01284155e-01 -6.29677594e-01
2.11654931e-01 -6.68655634e-02 -1.33653194e-01 9.79586184e-01
-6.70215428e-01 1.40484786e+00 6.53321600e+00 7.19271958e-01
-8.01007211e-01 2.82832950e-01 1.72321871e-01 1.16290547e-01
-2.25341871e-01 -1.02589861e-01 -6.91906631e-01 1.01069950e-01
7.28554666e-01 -4.32885796e-01 1.79885074e-01 7.23394752e-01
2.97781706e-01 -2.60694176e-01 -1.29949677e+00 7.46848941e-01
1.72032043e-01 -8.65481496e-01 6.22586235e-02 -3.10058653e-01
1.18548483e-01 -3.01234603e-01 -4.13371503e-01 7.60847390e-01
6.12718582e-01 -5.75395107e-01 7.66791284e-01 4.80745584e-01
3.77159148e-01 -4.39788640e-01 8.54593933e-01 9.10127759e-01
-7.05688298e-01 3.35142970e-01 -5.95019422e-02 -4.39127684e-01
1.33855417e-01 -2.26221129e-01 -1.57309747e+00 6.76057637e-01
3.29798281e-01 -7.05009624e-02 -6.10172987e-01 5.56049466e-01
-2.44977504e-01 2.77186260e-02 -2.70882789e-02 -4.29577768e-01
1.08366169e-01 -2.53539085e-01 6.02794826e-01 1.34117091e+00
-3.08747888e-01 1.94209456e-01 5.48374832e-01 6.19231582e-01
-5.63602336e-02 6.59259260e-01 -7.45686471e-01 2.69485712e-01
6.21480346e-01 1.31517351e+00 -2.10114300e-01 -6.17572606e-01
-3.13256830e-01 8.97003174e-01 7.62587309e-01 2.95444667e-01
-5.29225290e-01 -4.66596961e-01 2.89300203e-01 -2.09199250e-01
-5.21705747e-01 -3.09052859e-02 -1.88013285e-01 -1.17054677e+00
6.22683950e-02 -1.47670043e+00 3.67679924e-01 -7.79814482e-01
-1.40189731e+00 1.04265904e+00 -6.06516674e-02 -8.48400295e-01
-6.62760794e-01 -3.28071862e-01 -3.35942239e-01 1.05075526e+00
-1.07571852e+00 -9.23074245e-01 -2.72666931e-01 6.78537667e-01
1.18220770e+00 -2.61540562e-01 1.24441624e+00 1.81322992e-01
-4.24257904e-01 5.73544383e-01 -3.51483345e-01 1.75317124e-01
1.33711827e+00 -1.30238163e+00 1.79309830e-01 5.11993706e-01
-3.00732851e-01 8.76456857e-01 9.29626107e-01 -9.30458963e-01
-1.29234946e+00 -5.86605549e-01 1.10188234e+00 -8.12661290e-01
5.74004889e-01 -1.38739184e-01 -9.57231402e-01 6.72727048e-01
9.87573326e-01 -1.14099634e+00 8.83128583e-01 4.04197931e-01
-4.18893397e-01 2.46937394e-01 -1.03176808e+00 8.77060115e-01
8.67253542e-01 -4.66780603e-01 -1.31154215e+00 3.64838690e-01
7.70317793e-01 -6.32889986e-01 -7.23340452e-01 2.09734872e-01
4.70698208e-01 -8.62957120e-01 4.78960276e-01 -8.38989079e-01
3.46209854e-01 -2.07076386e-01 7.10711181e-02 -1.69207478e+00
-2.35116512e-01 -7.88677216e-01 4.32224780e-01 1.30300975e+00
6.35801494e-01 -6.58641636e-01 3.42864841e-01 1.08982599e+00
-1.62043929e-01 -2.32497215e-01 -5.83957195e-01 -4.48096603e-01
-2.92002916e-01 -2.11373195e-01 4.73986447e-01 1.01210904e+00
8.12142134e-01 1.26564240e+00 -4.86480445e-01 -2.84952223e-01
2.05339342e-01 -2.41633877e-02 1.39898539e+00 -1.23070431e+00
-2.00029194e-01 -5.64923584e-01 3.19132954e-01 -1.27777064e+00
1.15493663e-01 -7.59266734e-01 2.31623337e-01 -1.42937255e+00
2.09574103e-01 -3.39516997e-01 3.26756239e-01 4.19545531e-01
-3.46447259e-01 -1.89323887e-01 2.87418991e-01 3.78718704e-01
-8.34754407e-01 6.29875720e-01 1.33149505e+00 -2.16384143e-01
-6.60727620e-01 -1.59598619e-01 -6.93326473e-01 6.59870625e-01
8.61929059e-01 -4.65902418e-01 -5.64314961e-01 -6.36271894e-01
2.68397272e-01 3.55032980e-01 1.35065345e-02 -6.23685539e-01
7.54261553e-01 -2.56620169e-01 -1.29815638e-01 -4.64277983e-01
2.70619929e-01 -5.72824359e-01 4.24407050e-02 1.97843730e-01
-8.36956322e-01 3.71755600e-01 8.55582431e-02 2.08911285e-01
-3.65157932e-01 -6.22160614e-01 2.82684475e-01 -4.02304411e-01
-4.42135125e-01 -2.92024374e-01 -7.88043261e-01 9.04857293e-02
8.73143852e-01 -6.72392473e-02 -7.47991264e-01 -5.13711035e-01
-5.59163868e-01 5.06198406e-01 3.45902205e-01 4.63545740e-01
5.46914399e-01 -1.09653068e+00 -8.71305943e-01 2.24690549e-02
4.25827533e-01 -1.43359438e-01 9.91515443e-02 6.45029008e-01
-4.36586112e-01 4.84981000e-01 -2.79671401e-01 -3.85798812e-01
-1.50602520e+00 2.90704310e-01 2.34332174e-01 -7.50261188e-01
-4.83286530e-01 5.08796632e-01 1.87895656e-01 -9.12788689e-01
4.04152781e-01 -1.87814441e-02 -6.16224289e-01 3.53046089e-01
6.66253209e-01 7.97587037e-02 2.55067885e-01 -3.08366597e-01
1.08456314e-01 1.36373714e-01 -5.15217066e-01 -7.11865842e-01
8.75092089e-01 -4.09330964e-01 -1.42675906e-01 5.31867445e-01
4.97189850e-01 1.72801629e-01 -6.76557839e-01 -6.91034853e-01
3.16821337e-01 -4.96515781e-01 -5.37470341e-01 -1.19184971e+00
-3.05177569e-01 4.92130369e-01 8.45739245e-02 6.52340591e-01
7.00458527e-01 -2.35394657e-01 6.11764252e-01 1.02233696e+00
5.62630117e-01 -1.44982326e+00 4.26709414e-01 9.96228814e-01
1.10952687e+00 -1.29411125e+00 -3.72642428e-01 -2.68157333e-01
-1.10626364e+00 1.20436680e+00 1.28151500e+00 5.26870668e-01
-7.00312704e-02 -1.83125474e-02 7.55143285e-01 -3.54396313e-01
-1.43101478e+00 -1.47797406e-01 -3.26719731e-01 4.44514036e-01
5.16224504e-01 4.16475395e-03 -8.03048909e-01 8.45194697e-01
-4.21306789e-01 -8.70966688e-02 6.04732692e-01 9.22777176e-01
-3.51243913e-01 -1.39343572e+00 -2.67456293e-01 -1.17244665e-02
-2.53517538e-01 -7.65770152e-02 -9.51630890e-01 7.16681004e-01
-6.00296855e-01 1.28899336e+00 -3.67254049e-01 -5.05181491e-01
1.07123840e+00 6.46575630e-01 2.87523508e-01 -7.87563205e-01
-1.48173559e+00 9.09454077e-02 7.67340362e-01 -4.14903909e-01
-2.42318511e-01 -2.08133653e-01 -1.12121034e+00 -3.89146298e-01
-5.24843931e-01 6.07408345e-01 5.49134135e-01 8.83749664e-01
3.33205044e-01 4.77881491e-01 5.59109211e-01 -4.40775722e-01
-9.90126908e-01 -1.59225225e+00 9.84396040e-02 9.59131956e-01
-7.11374208e-02 -5.18719971e-01 1.90589987e-02 1.24264374e-01]
|
[12.849360466003418, 8.076214790344238]
|
b23e58b4-231e-4fd6-95f7-b1dfbb65d70b
|
learning-monocular-depth-in-dynamic
|
2305.07397
| null |
https://arxiv.org/abs/2305.07397v1
|
https://arxiv.org/pdf/2305.07397v1.pdf
|
Learning Monocular Depth in Dynamic Environment via Context-aware Temporal Attention
|
The monocular depth estimation task has recently revealed encouraging prospects, especially for the autonomous driving task. To tackle the ill-posed problem of 3D geometric reasoning from 2D monocular images, multi-frame monocular methods are developed to leverage the perspective correlation information from sequential temporal frames. However, moving objects such as cars and trains usually violate the static scene assumption, leading to feature inconsistency deviation and misaligned cost values, which would mislead the optimization algorithm. In this work, we present CTA-Depth, a Context-aware Temporal Attention guided network for multi-frame monocular Depth estimation. Specifically, we first apply a multi-level attention enhancement module to integrate multi-level image features to obtain an initial depth and pose estimation. Then the proposed CTA-Refiner is adopted to alternatively optimize the depth and pose. During the refinement process, context-aware temporal attention (CTA) is developed to capture the global temporal-context correlations to maintain the feature consistency and estimation integrity of moving objects. In particular, we propose a long-range geometry embedding (LGE) module to produce a long-range temporal geometry prior. Our approach achieves significant improvements over state-of-the-art approaches on three benchmark datasets.
|
['Xianzhi Li', 'Jian Pu', 'Yuanzhu Gan', 'Yunzhe Wu', 'Zhi-Gang Fan', 'Zhuozheng Li', 'Zizhang Wu']
|
2023-05-12
| null | null | null | null |
['monocular-depth-estimation']
|
['computer-vision']
|
[-2.43473817e-02 -2.25369349e-01 6.45647421e-02 -7.31728435e-01
-5.85274220e-01 -3.13820004e-01 6.21514320e-01 -2.65737027e-01
-4.84230071e-01 4.10420060e-01 6.20177574e-02 5.44392392e-02
-4.76158522e-02 -6.46499991e-01 -8.14096570e-01 -7.83712327e-01
3.24150532e-01 1.86385885e-01 3.35709423e-01 3.25358026e-02
5.08259833e-01 6.95650995e-01 -1.51035070e+00 2.84515284e-02
7.62387156e-01 1.23489618e+00 5.21134794e-01 4.09425586e-01
4.75188456e-02 8.81312966e-01 -1.08059093e-01 -1.47037566e-01
3.16248357e-01 -7.90988281e-02 -4.94967341e-01 1.42724454e-01
7.64743984e-01 -7.84980893e-01 -7.18290627e-01 1.05595958e+00
3.68266732e-01 4.02848899e-01 2.99369782e-01 -1.20707095e+00
-2.93087333e-01 -4.19362076e-02 -7.47145593e-01 3.26187164e-01
3.55618805e-01 4.97089714e-01 8.37363899e-01 -1.23365521e+00
7.30803967e-01 1.42509711e+00 1.03117682e-01 3.50717843e-01
-7.13470459e-01 -5.92766941e-01 6.94669187e-01 6.78068817e-01
-1.35045803e+00 -3.26743156e-01 1.09876931e+00 -3.44183087e-01
8.98148656e-01 -1.06860742e-01 7.33995974e-01 6.75427914e-01
4.63720441e-01 8.12450767e-01 8.43911588e-01 6.21663146e-02
-3.42177763e-03 -1.75690413e-01 -1.29115626e-01 7.39814043e-01
1.20276853e-01 4.19013292e-01 -6.58853650e-01 4.16103274e-01
8.38794768e-01 2.43179724e-01 -2.20773771e-01 -6.94846511e-01
-1.28020263e+00 6.31115615e-01 7.25626171e-01 1.38142899e-01
-3.96856159e-01 3.31315160e-01 8.75895564e-03 -8.04978162e-02
5.19321382e-01 3.32809687e-01 -5.16921222e-01 -4.47767936e-02
-6.13778412e-01 3.44111860e-01 4.10914607e-02 1.05275714e+00
9.59235847e-01 1.13254547e-01 -1.86177313e-01 4.19303238e-01
5.26105642e-01 3.46184134e-01 1.40437782e-01 -1.23371005e+00
7.60944247e-01 7.21669734e-01 2.14049742e-01 -1.28341877e+00
-5.75379908e-01 -6.10149741e-01 -6.47346616e-01 1.00858271e-01
2.98539609e-01 2.18854696e-01 -7.67533779e-01 1.66045117e+00
7.63371885e-01 3.01787376e-01 -1.81262881e-01 1.37758124e+00
7.13794589e-01 5.75640619e-01 -2.52536416e-01 -8.51286110e-03
1.15530479e+00 -1.15765619e+00 -5.35423398e-01 -5.81296742e-01
5.24050355e-01 -7.05427110e-01 6.54255331e-01 1.89501956e-01
-1.16420245e+00 -7.63970792e-01 -1.10641086e+00 -4.89835173e-01
-2.28560045e-01 -2.27509579e-03 3.48812193e-01 1.64357498e-01
-6.44361377e-01 2.13133022e-01 -9.46019471e-01 1.09395534e-01
3.80598366e-01 2.63047785e-01 -3.74602407e-01 -6.37319803e-01
-9.67444420e-01 9.01925206e-01 2.56466657e-01 4.59556937e-01
-8.27758610e-01 -7.26315975e-01 -1.28995073e+00 -1.43292725e-01
6.04896486e-01 -8.84010136e-01 9.31621373e-01 -3.34904760e-01
-1.52445924e+00 7.39405155e-01 -3.98891658e-01 -2.29787931e-01
6.02028906e-01 -5.19381762e-01 1.66199133e-02 3.06399763e-01
1.99381754e-01 1.08489788e+00 7.00458944e-01 -1.21667230e+00
-8.13125253e-01 -6.59907281e-01 3.26755494e-01 6.68721855e-01
2.51883149e-01 -4.24454778e-01 -7.25821376e-01 -3.90148133e-01
6.70945048e-01 -9.26382184e-01 -3.51972133e-01 3.18296283e-01
-2.60289907e-01 3.35229263e-02 8.09115410e-01 -5.28773367e-01
8.14106345e-01 -2.07683921e+00 4.61972624e-01 -2.28999481e-01
2.32778177e-01 -1.16511211e-01 3.08127366e-02 -2.18634382e-01
1.50410444e-01 -4.10533965e-01 -7.09998533e-02 -6.64471149e-01
-1.47260219e-01 1.74963534e-01 -5.37969954e-02 8.26641023e-01
4.46011215e-01 1.10938740e+00 -9.96634543e-01 -4.37077016e-01
8.94152164e-01 5.05633175e-01 -8.72119308e-01 3.70597899e-01
-3.38936567e-01 8.50620329e-01 -6.13676846e-01 6.27575636e-01
1.03923643e+00 -2.22700462e-01 -3.94472152e-01 -5.44916749e-01
-4.39438283e-01 2.56036103e-01 -9.71705794e-01 2.03254986e+00
-5.55044472e-01 6.78075373e-01 -1.03970207e-01 -7.91401207e-01
7.93681741e-01 -1.71841726e-01 5.74091911e-01 -8.14060509e-01
3.99671108e-01 5.89776300e-02 -1.31805465e-01 -4.23671305e-01
6.87429786e-01 1.85866266e-01 2.92060561e-02 1.25795789e-02
-2.95555174e-01 -6.25400960e-01 -1.50490165e-01 2.04406697e-02
7.08909333e-01 4.59873915e-01 1.94442660e-01 9.32296813e-02
9.27359998e-01 -1.54303506e-01 7.09037244e-01 2.02706903e-01
-2.13834018e-01 8.39230299e-01 1.20987505e-01 -6.81772709e-01
-9.80154514e-01 -7.81452835e-01 -2.61367895e-02 5.24184763e-01
8.40482473e-01 1.34641081e-02 -2.56700069e-01 -4.52668101e-01
-1.50924046e-02 5.95088124e-01 -5.87282658e-01 -2.22289667e-01
-6.88851714e-01 -5.33737183e-01 -1.90490052e-01 5.98360717e-01
8.29908669e-01 -7.01336741e-01 -1.01544666e+00 3.52882922e-01
-3.33680183e-01 -1.71313417e+00 -7.31502175e-01 4.35557310e-03
-7.40960479e-01 -9.28043306e-01 -6.01823747e-01 -4.57176864e-01
5.59710264e-01 6.35105908e-01 7.72362411e-01 -1.01302490e-01
-1.09683797e-01 -8.94708280e-03 -1.09342404e-01 -5.51746450e-02
2.46432617e-01 -4.05787490e-02 -1.31621227e-01 2.11240411e-01
3.58919531e-01 -5.52407801e-01 -9.83556986e-01 5.63088179e-01
-6.42659664e-01 4.07333165e-01 5.77785969e-01 8.25461328e-01
6.10897541e-01 -1.79838955e-01 1.65334120e-01 -3.23099643e-01
-3.59861583e-01 -3.10588151e-01 -1.00707567e+00 -1.51909888e-01
-3.39561641e-01 2.31496561e-02 3.75328928e-01 -2.26768643e-01
-1.11183858e+00 2.69367933e-01 -1.04931086e-01 -8.97438467e-01
-7.77522922e-02 2.98335969e-01 -5.13155818e-01 -2.12588057e-01
7.12318271e-02 3.23456377e-01 -3.18087816e-01 -1.90813228e-01
3.06838840e-01 1.92310020e-01 5.73355973e-01 -2.86372513e-01
9.09211338e-01 7.49560654e-01 1.94314197e-01 -5.88304758e-01
-1.14655495e+00 -5.14579535e-01 -8.95015299e-01 -3.05452108e-01
1.31911016e+00 -1.31177258e+00 -5.97769856e-01 7.05846250e-01
-1.42412066e+00 -2.88948894e-01 2.31524125e-01 7.44943142e-01
-6.17193460e-01 3.92516464e-01 -3.84649992e-01 -6.81908250e-01
-3.09426170e-02 -1.40037966e+00 1.41982555e+00 1.60442069e-01
2.45730981e-01 -7.28366911e-01 -1.93371534e-01 5.02279460e-01
8.60391334e-02 3.69014442e-01 5.99926233e-01 1.98445231e-01
-1.38165450e+00 1.57094657e-01 -5.91486573e-01 2.80163158e-02
1.06992330e-02 -2.81857610e-01 -9.72234964e-01 -2.25200444e-01
1.66164055e-01 -3.83814946e-02 6.87536895e-01 5.80952823e-01
1.13825500e+00 7.72603899e-02 -2.59423763e-01 1.20597434e+00
1.25913775e+00 4.23299134e-01 5.05007803e-01 4.07876909e-01
1.25435185e+00 6.74350023e-01 1.09297311e+00 4.41365182e-01
7.84134805e-01 1.00015748e+00 6.74747705e-01 1.26263678e-01
8.15421343e-02 -2.82562554e-01 1.54991567e-01 6.43231153e-01
2.58112848e-01 -1.94186836e-01 -6.36615098e-01 5.40742755e-01
-1.77156639e+00 -7.51874685e-01 3.41258831e-02 1.88290203e+00
4.44799840e-01 3.54888886e-01 -3.20131123e-01 -1.11354619e-01
4.57924068e-01 3.89330983e-01 -9.44156766e-01 6.75895065e-02
-1.38755783e-01 -3.30776691e-01 5.09593844e-01 7.17353284e-01
-9.91311669e-01 1.06193793e+00 4.60720062e+00 5.52386761e-01
-1.19245172e+00 8.65713134e-02 7.02187002e-01 -2.17547506e-01
-4.30094182e-01 -3.92318889e-03 -9.51201677e-01 2.95351893e-01
1.97627872e-01 1.70142017e-02 1.78989381e-01 7.64965713e-01
3.13162655e-01 -3.12224150e-01 -1.22300863e+00 1.20436060e+00
7.33918920e-02 -1.17296875e+00 -8.66057053e-02 1.86967835e-01
1.01254940e+00 1.66426808e-01 1.44395083e-01 1.65922046e-02
-1.20109484e-01 -7.43871629e-01 9.62609410e-01 4.79322642e-01
7.60019362e-01 -9.10159647e-01 6.53794169e-01 3.59645903e-01
-1.56781983e+00 -2.46061981e-01 -3.50496620e-01 -1.03773758e-01
5.35498500e-01 5.91699719e-01 -4.20949191e-01 8.62007022e-01
6.83845937e-01 1.23483372e+00 -4.89337355e-01 9.26922679e-01
-7.38388896e-02 -2.62934923e-01 -3.06413174e-01 1.85859188e-01
5.97813904e-01 -2.03484058e-01 5.69648802e-01 5.41138291e-01
2.70990014e-01 4.23011124e-01 -2.88305171e-02 9.72996414e-01
2.23873928e-01 -2.69402266e-01 -4.88243103e-01 3.90030444e-01
3.92245591e-01 1.14070463e+00 -5.01622558e-01 -2.47038096e-01
-6.52894497e-01 8.87304962e-01 3.27294379e-01 2.48321176e-01
-9.85893786e-01 1.42158493e-01 8.49827588e-01 1.04400739e-01
4.78892803e-01 -5.88176668e-01 -4.31060165e-01 -1.32064843e+00
1.94215685e-01 -5.31244278e-01 9.73628759e-02 -1.03806961e+00
-8.35952818e-01 6.43252969e-01 1.86947271e-01 -1.40083826e+00
-1.53636143e-01 -4.82773006e-01 -4.61422622e-01 7.19055831e-01
-1.97540629e+00 -1.00556290e+00 -8.38238478e-01 6.05884790e-01
8.99133742e-01 1.40846342e-01 -1.68728326e-02 4.16939646e-01
-5.03900468e-01 4.02693391e-01 -4.69152361e-01 -2.34762773e-01
5.36214709e-01 -8.84007931e-01 5.81602454e-01 8.68617415e-01
-1.51838601e-01 3.81018758e-01 4.92233247e-01 -4.96022731e-01
-1.51538587e+00 -1.30067706e+00 6.54072642e-01 -5.82163334e-01
2.22592786e-01 -2.13414133e-01 -8.28046501e-01 5.20887196e-01
-1.94638610e-01 2.24134564e-01 -2.09933057e-01 -6.24139309e-01
-9.97566134e-02 -1.23607770e-01 -8.87732863e-01 6.63888574e-01
1.32349455e+00 -5.00409186e-01 -4.32792306e-01 -4.62262444e-02
9.35874104e-01 -8.75005543e-01 -5.15939236e-01 5.65396965e-01
6.30863667e-01 -1.05371189e+00 1.15429926e+00 2.81728432e-02
6.07780218e-01 -5.48441470e-01 -3.20297003e-01 -8.50540876e-01
-1.08986169e-01 -4.74789739e-01 -1.49895579e-01 7.69991457e-01
-2.03634389e-02 -5.20072341e-01 8.74054193e-01 5.79281151e-01
-4.92926121e-01 -9.78581429e-01 -1.11033392e+00 -4.35158938e-01
-1.78142399e-01 -5.35106599e-01 5.96359670e-01 6.20298684e-01
-5.43985426e-01 3.19302082e-01 -3.93028200e-01 4.97238427e-01
6.91251576e-01 2.93050349e-01 1.06897748e+00 -8.81671965e-01
-1.94361106e-01 -3.35053056e-01 -6.30883932e-01 -1.97923517e+00
2.09071025e-01 -4.74008083e-01 1.25087157e-01 -1.19467640e+00
-5.51105067e-02 -3.13069850e-01 -7.18501117e-03 -1.59829482e-01
-2.94581294e-01 2.87464947e-01 1.76982999e-01 -4.56025153e-02
-5.67706645e-01 1.08130491e+00 1.68364215e+00 -6.12714440e-02
-1.53423741e-01 -1.12083793e-01 -2.68423736e-01 6.87545359e-01
4.05778021e-01 -3.47044855e-01 -4.85954314e-01 -7.14541316e-01
2.49720663e-01 3.71220261e-01 6.91225052e-01 -1.01049232e+00
3.93503129e-01 -3.23116332e-01 4.81586009e-01 -1.33210373e+00
9.05239940e-01 -8.94200325e-01 -8.48620534e-02 3.79122913e-01
-7.14071691e-02 3.01616073e-01 4.70403433e-02 6.62250042e-01
-3.18850249e-01 1.44242808e-01 8.50998461e-01 -7.78570399e-02
-9.47858155e-01 9.03786957e-01 -8.04065261e-03 3.12868059e-02
1.00223470e+00 -2.79389739e-01 -1.03202567e-01 -3.54148805e-01
-4.55881953e-01 5.42051375e-01 6.68824553e-01 5.61009347e-01
1.01306212e+00 -1.34284616e+00 -5.62997222e-01 4.09060478e-01
2.93161422e-01 6.59294963e-01 6.59657359e-01 1.01627994e+00
-6.20155871e-01 5.93813896e-01 -2.32777268e-01 -1.06589687e+00
-9.09576058e-01 5.87155521e-01 5.44057310e-01 -5.33553958e-02
-7.04225361e-01 9.72639382e-01 8.19275260e-01 -3.19644660e-01
4.09844443e-02 -5.87654412e-01 -7.45765865e-02 -2.67985672e-01
3.70962292e-01 2.36464545e-01 -2.71841865e-02 -8.00087929e-01
-5.37219524e-01 1.25793529e+00 -1.05215311e-01 -2.34563664e-01
1.22492778e+00 -7.25584030e-01 2.39605054e-01 2.96901792e-01
1.41203821e+00 -2.96780288e-01 -1.95173514e+00 -3.67283463e-01
-3.49812090e-01 -9.35125113e-01 3.15003216e-01 -1.94943398e-01
-1.22564816e+00 1.11833167e+00 4.75224167e-01 -5.97718179e-01
9.63580847e-01 -1.48782745e-01 8.59050989e-01 2.71119505e-01
5.57738900e-01 -8.01698923e-01 3.33633870e-01 6.62439823e-01
8.51834536e-01 -1.53333831e+00 9.24638584e-02 -6.15186453e-01
-5.34648836e-01 1.02611256e+00 1.09065366e+00 -8.61917436e-02
6.27719581e-01 -1.45249337e-01 -1.37783155e-01 -2.82105863e-01
-8.88561070e-01 -1.11198410e-01 5.62063575e-01 3.57125163e-01
7.80853792e-04 -3.64349008e-01 1.08144060e-01 2.07252517e-01
-9.89835784e-02 -2.90247232e-01 3.44566554e-01 7.07044899e-01
-2.84234643e-01 -6.12786174e-01 -1.58030406e-01 -1.17399663e-01
1.06527461e-02 5.29725924e-02 -5.75571358e-02 7.67822444e-01
3.58725995e-01 8.80902529e-01 1.81126043e-01 -4.74282652e-01
3.44362527e-01 -2.89109051e-01 6.79536641e-01 -6.11433446e-01
-9.80174989e-02 1.56154573e-01 -2.04992533e-01 -9.37906504e-01
-4.79333699e-01 -7.88895249e-01 -1.08528030e+00 -2.03126907e-01
-2.90878415e-01 -4.02393192e-01 5.41246116e-01 1.17994022e+00
2.40515634e-01 5.19415438e-01 7.93249130e-01 -1.32159126e+00
-8.21394548e-02 -6.76320016e-01 -8.11977237e-02 1.85731158e-01
5.60124516e-01 -1.06723952e+00 -3.97079229e-01 -2.79725969e-01]
|
[8.3619966506958, -2.2247402667999268]
|
589c6852-4e84-4a58-814b-2e4b62672887
|
comparison-of-semantic-segmentation
|
1805.08105
| null |
http://arxiv.org/abs/1805.08105v1
|
http://arxiv.org/pdf/1805.08105v1.pdf
|
Comparison of Semantic Segmentation Approaches for Horizon/Sky Line Detection
|
Horizon or skyline detection plays a vital role towards mountainous visual
geo-localization, however most of the recently proposed visual geo-localization
approaches rely on \textbf{user-in-the-loop} skyline detection methods.
Detecting such a segmenting boundary fully autonomously would definitely be a
step forward for these localization approaches. This paper provides a
quantitative comparison of four such methods for autonomous horizon/sky line
detection on an extensive data set. Specifically, we provide the comparison
between four recently proposed segmentation methods; one explicitly targeting
the problem of horizon detection\cite{Ahmad15}, second focused on visual
geo-localization but relying on accurate detection of skyline \cite{Saurer16}
and other two proposed for general semantic segmentation -- Fully Convolutional
Networks (FCN) \cite{Long15} and SegNet\cite{Badrinarayanan15}. Each of the
first two methods is trained on a common training set \cite{Baatz12} comprised
of about 200 images while models for the third and fourth method are fine tuned
for sky segmentation problem through transfer learning using the same data set.
Each of the method is tested on an extensive test set (about 3K images)
covering various challenging geographical, weather, illumination and seasonal
conditions. We report average accuracy and average absolute pixel error for
each of the presented formulation.
|
['George Bebis', 'Martin Čadík', 'Touqeer Ahmad', 'Pavel Campr']
|
2018-05-21
| null | null | null | null |
['line-detection']
|
['computer-vision']
|
[ 5.54698333e-02 2.37170860e-01 1.99057475e-01 -4.37351108e-01
-1.03739190e+00 -7.60170460e-01 6.78378463e-01 1.39696568e-01
-5.30911148e-01 8.04185152e-01 -3.53167623e-01 -3.37695301e-01
2.27778312e-02 -9.62141454e-01 -8.91790986e-01 -6.86459780e-01
-2.37994552e-01 4.95910197e-01 6.11634135e-01 -3.50661278e-01
2.49242693e-01 3.61738801e-01 -1.50759220e+00 -2.03627989e-01
1.05397129e+00 1.30724561e+00 3.44873965e-01 1.04062676e+00
-9.12168175e-02 5.90003490e-01 -6.15755081e-01 -5.64361587e-02
4.89613920e-01 -2.59766817e-01 -8.66381288e-01 1.22077085e-01
8.61708283e-01 6.97233062e-03 -1.04091885e-02 1.05818510e+00
4.25598264e-01 2.25037664e-01 6.70167565e-01 -1.25884736e+00
-1.06559880e-01 2.79345930e-01 -8.26066911e-01 6.33099973e-01
4.90366891e-02 1.71543285e-01 6.98529065e-01 -7.67091095e-01
3.26532304e-01 6.29737377e-01 1.00861239e+00 -2.82392383e-01
-6.62982345e-01 -4.12616283e-01 3.24040726e-02 2.26867512e-01
-1.72902989e+00 -1.26529619e-01 5.41368067e-01 -5.22964001e-01
7.21698105e-01 4.05480653e-01 5.22759676e-01 3.36060733e-01
-3.04619998e-01 6.49425030e-01 1.35401297e+00 -4.19676870e-01
2.25650743e-01 1.24047950e-01 2.39351586e-01 7.58963168e-01
2.29255199e-01 -1.19700223e-01 -2.38371521e-01 1.11148737e-01
6.90371513e-01 -4.50302064e-01 -4.04858619e-01 -2.06548318e-01
-1.04866183e+00 9.01706100e-01 8.39610577e-01 3.31667542e-01
-2.54308760e-01 3.22740048e-01 2.45684981e-01 -6.53320476e-02
6.62847519e-01 1.54765591e-01 -2.88785487e-01 2.40776539e-01
-1.61900747e+00 4.83434737e-01 5.94491422e-01 9.16148543e-01
1.11738670e+00 3.79189849e-01 5.17330598e-03 5.69556057e-01
2.17754483e-01 8.28705966e-01 2.89995253e-01 -7.35701501e-01
5.48629999e-01 4.69346911e-01 5.14095485e-01 -8.19564223e-01
-7.91829646e-01 -6.12336755e-01 -6.32151783e-01 4.38576490e-01
6.85091436e-01 -2.55990058e-01 -1.48480129e+00 1.12411845e+00
5.32965243e-01 1.45424336e-01 7.90392689e-04 1.02871621e+00
1.14037430e+00 8.10887456e-01 5.30166700e-02 2.62538284e-01
1.43358052e+00 -1.20899832e+00 -2.10245207e-01 -5.31446755e-01
5.43841720e-01 -6.87403083e-01 8.83131981e-01 2.71618724e-01
-7.81888306e-01 -5.05944431e-01 -1.15207970e+00 -6.59956783e-02
-8.78901839e-01 5.81470847e-01 5.33543050e-01 6.89907789e-01
-1.47262084e+00 2.48819724e-01 -3.50825697e-01 -8.21628988e-01
1.38906658e-01 1.74019173e-01 -6.00479133e-02 2.00352028e-01
-1.33710241e+00 6.14506721e-01 6.73537016e-01 3.83329183e-01
-1.00230861e+00 -3.24147791e-01 -9.23548996e-01 -1.16590457e-02
2.52058625e-01 -6.43226385e-01 9.17502463e-01 -1.14138496e+00
-1.02724636e+00 1.43886769e+00 1.42601803e-01 -7.72656918e-01
8.43367696e-01 -3.28979552e-01 -5.42060971e-01 2.66352534e-01
4.99074370e-01 8.93004000e-01 7.60489285e-01 -1.33578348e+00
-9.31204736e-01 -2.66946822e-01 -6.24137186e-03 5.13494968e-01
3.48711491e-01 -7.93926045e-02 -6.81804001e-01 -5.61165988e-01
4.08528410e-02 -8.86704683e-01 -1.64898187e-01 -1.93222076e-01
-5.64674616e-01 -1.18464410e-01 7.22921252e-01 -9.77857351e-01
9.40835834e-01 -1.80506885e+00 -2.46618539e-01 3.14881027e-01
-6.06161654e-02 4.90698308e-01 4.01746243e-01 5.95379412e-01
1.12598635e-01 1.46284448e-02 -5.99230707e-01 -1.63558662e-01
-1.35605052e-01 -8.13855454e-02 -1.90514222e-01 8.40318143e-01
-1.79589838e-01 7.16140568e-01 -5.45597374e-01 -6.48778319e-01
6.45560563e-01 3.20914865e-01 2.12637797e-01 1.64079919e-01
-2.74565518e-01 5.30969679e-01 -4.65418339e-01 6.69008136e-01
1.21269417e+00 -2.23507434e-01 -5.10042727e-01 -7.82240741e-03
-4.56300229e-01 -1.86391607e-01 -1.38367784e+00 1.33561432e+00
-3.16806316e-01 6.79239273e-01 1.47241846e-01 -8.38265896e-01
9.40483272e-01 4.65603508e-02 7.48311356e-02 -8.81723404e-01
3.88443209e-02 3.03628415e-01 -6.58215106e-01 -4.69191670e-01
7.21106172e-01 2.18539268e-01 -3.58814150e-02 -1.06970787e-01
-1.97052583e-01 -4.60312903e-01 2.39992008e-01 1.30520731e-01
5.94393730e-01 1.31999210e-01 2.42663383e-01 -3.65198076e-01
6.99451506e-01 5.29244900e-01 1.96420953e-01 1.15156198e+00
-4.29684222e-01 1.08288431e+00 1.01054184e-01 -3.52628618e-01
-1.08182120e+00 -1.03061867e+00 -7.69771934e-02 1.01270676e+00
5.15766144e-01 9.96117201e-03 -9.08739984e-01 -5.92204094e-01
-1.43987253e-01 7.30374217e-01 -7.64350057e-01 4.83144253e-01
-2.08852485e-01 -9.33302104e-01 9.51992750e-01 4.06958580e-01
1.29832470e+00 -9.17210221e-01 -1.02072978e+00 -1.88553959e-01
-2.50986278e-01 -1.14821088e+00 4.28204574e-02 1.46542206e-01
-3.13934982e-01 -1.18843007e+00 -1.00207400e+00 -6.13409638e-01
1.61925584e-01 3.97774100e-01 1.16129231e+00 -5.61984107e-02
-3.55056763e-01 3.55776429e-01 -3.08312148e-01 -4.72471982e-01
3.76453437e-02 1.79658115e-01 -5.17233312e-01 4.88963388e-02
1.41429394e-01 -2.58669436e-01 -1.09993410e+00 3.51765752e-01
-7.08695054e-01 1.63356289e-01 4.28971052e-01 3.19466323e-01
5.04702151e-01 -3.04375943e-02 1.70732096e-01 -7.75639534e-01
-3.44826989e-02 -6.69646680e-01 -9.05332327e-01 1.53247908e-01
-4.72359061e-01 -4.54849392e-01 3.25912416e-01 3.93254071e-01
-1.17500377e+00 2.80972779e-01 -3.09689701e-01 -1.45040140e-01
-7.06476927e-01 2.55067855e-01 3.08389701e-02 -4.39702034e-01
8.29240501e-01 5.18196046e-01 -7.33360827e-01 -3.62463981e-01
6.67854309e-01 6.41959786e-01 9.01466012e-01 2.14035399e-02
1.02262080e+00 1.01428735e+00 -1.01664416e-01 -1.15278125e+00
-1.03580117e+00 -1.17906404e+00 -6.49884939e-01 -3.74094903e-01
1.26209390e+00 -1.24497342e+00 -1.40684754e-01 7.94994950e-01
-6.83926463e-01 -4.62288588e-01 1.35011882e-01 1.70871779e-01
-5.15364349e-01 2.92143673e-01 -1.44234911e-01 -1.07872772e+00
-5.81355870e-01 -7.22352028e-01 1.38000047e+00 6.70225203e-01
2.84530312e-01 -1.12464046e+00 1.80879325e-01 5.39363682e-01
4.92031693e-01 6.31308734e-01 2.78851181e-01 -3.87735039e-01
-5.95257223e-01 -2.66622990e-01 -6.60574198e-01 1.53214037e-01
-2.18238786e-01 -3.06674927e-01 -1.38929522e+00 -2.14951754e-01
-3.24913502e-01 -2.21338451e-01 1.16584361e+00 6.88650012e-01
7.35236049e-01 2.66021211e-03 -3.97906482e-01 9.40584064e-01
1.91168594e+00 -4.46628034e-02 6.33747041e-01 8.88523638e-01
1.07606983e+00 3.64813387e-01 9.36986446e-01 3.04245144e-01
5.81601977e-01 5.05511820e-01 8.55213284e-01 -6.81870699e-01
-3.52648646e-03 8.85767192e-02 -3.14862281e-02 -2.92192459e-01
-1.92936569e-01 -4.48719025e-01 -1.06416905e+00 1.10052049e+00
-1.82502222e+00 -6.16720021e-01 -7.02315450e-01 2.08846760e+00
2.25731403e-01 3.18276137e-02 3.73745501e-01 5.46546876e-02
7.14632869e-01 5.88535488e-01 -3.65633160e-01 -2.22147882e-01
-3.00832570e-01 6.82549030e-02 1.47990417e+00 3.96098912e-01
-1.85700989e+00 1.25742388e+00 5.12519598e+00 7.86165655e-01
-1.26929593e+00 1.22722223e-01 7.47674704e-01 3.97906035e-01
5.79786599e-02 1.57006145e-01 -9.91886735e-01 4.56236273e-01
8.74199092e-01 1.97671130e-01 3.60876024e-02 1.04961550e+00
5.31161308e-01 -1.05244064e+00 -1.76015511e-01 9.93129134e-01
1.47463605e-01 -1.42205346e+00 -2.97734827e-01 -2.96704710e-01
9.09418702e-01 6.51242375e-01 1.63190154e-04 -1.98429804e-02
2.38300622e-01 -1.13725197e+00 9.43570435e-01 6.04158700e-01
6.93898380e-01 -6.49697125e-01 6.40862703e-01 3.59470606e-01
-1.46987998e+00 1.29656345e-01 -2.72105753e-01 2.19318807e-01
2.82052487e-01 6.55195713e-01 -8.88784289e-01 8.70082319e-01
1.32563746e+00 4.63091493e-01 -9.80434358e-01 1.50160372e+00
-2.90876091e-01 9.62046981e-01 -6.27877772e-01 3.06306362e-01
8.87103856e-01 -3.00823897e-01 5.64891219e-01 1.51797390e+00
2.45279238e-01 -2.88801044e-01 1.48518994e-01 7.79400051e-01
3.35452706e-01 9.32362303e-02 -5.67515194e-01 3.55753839e-01
5.57781234e-02 1.49524987e+00 -1.26380277e+00 -3.57779711e-01
-2.35629439e-01 1.09163237e+00 1.41753390e-01 4.94766742e-01
-1.03219116e+00 -5.44247389e-01 5.31683788e-02 3.77125859e-01
6.32348061e-01 -3.73824239e-01 -2.34573528e-01 -7.64222980e-01
-2.86405832e-02 -3.25727314e-01 6.37632370e-01 -1.12020230e+00
-6.17704213e-01 3.94086868e-01 3.09414715e-01 -9.62855756e-01
7.78837055e-02 -3.78744125e-01 -8.57888341e-01 1.09048045e+00
-1.85710001e+00 -1.60929334e+00 -9.97934043e-01 5.77961266e-01
6.13157272e-01 3.67184311e-01 2.05002770e-01 3.34434420e-01
-3.01289856e-01 2.77147055e-01 4.57786262e-01 -9.14785415e-02
5.03158271e-01 -1.65882075e+00 6.08260870e-01 1.11741579e+00
1.06961876e-01 -8.59476477e-02 9.52992260e-01 -5.62389016e-01
-5.88738501e-01 -1.59561586e+00 5.37058115e-01 -1.97719842e-01
4.58466500e-01 -2.79571325e-01 -9.44391668e-01 8.03707063e-01
2.52904177e-01 -1.30617365e-01 -9.10786614e-02 -6.25893712e-01
1.58852622e-01 -3.42176482e-02 -1.18643093e+00 2.71002024e-01
5.88142514e-01 -2.08770886e-01 -3.98725122e-01 7.15497077e-01
4.08836961e-01 -6.80963099e-01 -5.53200781e-01 3.75522614e-01
-1.58057824e-01 -1.49018359e+00 9.54192340e-01 3.00003320e-01
-9.06936228e-02 -7.92544186e-01 9.76796374e-02 -1.10195744e+00
1.92456141e-01 -4.18991536e-01 4.11347032e-01 1.20076382e+00
2.78641611e-01 -4.92262840e-01 9.05703366e-01 -1.97965980e-01
-2.99199551e-01 -1.43517062e-01 -9.24470365e-01 -6.99912965e-01
1.43828348e-03 -4.00210410e-01 2.69289166e-01 7.83337772e-01
-8.79056096e-01 -9.32804421e-02 -3.67705941e-01 7.80034125e-01
6.99348867e-01 2.14189649e-01 9.30936396e-01 -8.33109379e-01
-5.82540892e-02 -1.71820089e-01 -4.75091994e-01 -8.09192538e-01
-4.15700018e-01 -5.96496165e-01 9.43998471e-02 -1.99638712e+00
-1.81776345e-01 -5.50839067e-01 3.34673226e-02 2.11941049e-01
-1.27787486e-01 6.12281740e-01 -2.89933920e-01 2.40065306e-01
-7.47558057e-01 4.10362512e-01 6.39400065e-01 1.53160065e-01
-1.89021304e-02 1.98370695e-01 1.90770123e-02 8.79237592e-01
8.33920360e-01 -1.91799462e-01 -4.45399404e-01 -1.70591757e-01
4.13810134e-01 -9.93097201e-02 8.06025147e-01 -1.62962687e+00
2.67823666e-01 -4.88330200e-02 3.63771737e-01 -1.10462260e+00
3.13969225e-01 -3.58704507e-01 -1.61933340e-02 1.14703521e-01
2.87205607e-01 -4.99613173e-02 2.73250580e-01 7.64961898e-01
-2.76189327e-01 -3.07011604e-01 1.11750591e+00 -4.93312925e-01
-1.64839470e+00 3.15131173e-02 -3.15054327e-01 3.13397020e-01
1.30526078e+00 -5.32556951e-01 -3.15415978e-01 -5.72540283e-01
-8.03778410e-01 6.29085720e-01 5.51482737e-01 9.76973549e-02
3.43792439e-01 -5.85042715e-01 -6.17652476e-01 -9.81008485e-02
4.24156368e-01 3.27640831e-01 3.31993699e-01 9.94501054e-01
-1.19350517e+00 4.82298255e-01 2.93949060e-02 -7.88825452e-01
-1.30292630e+00 3.03565025e-01 9.54849303e-01 -1.84919477e-01
-7.29933739e-01 1.04960215e+00 3.76980305e-01 -3.52322042e-01
5.47795258e-02 -2.86671788e-01 -4.05433953e-01 -3.02709192e-02
1.96513124e-02 4.60086495e-01 9.98446047e-02 -9.82711971e-01
-4.05471385e-01 7.41878092e-01 2.56156534e-01 -1.30162820e-01
9.97993171e-01 -5.56273818e-01 1.41422674e-01 5.92540622e-01
8.13249826e-01 -3.04366946e-01 -1.29454780e+00 1.06417444e-02
1.28553391e-01 -2.66375303e-01 3.04707944e-01 -8.31935823e-01
-9.30571198e-01 8.81133437e-01 1.11402428e+00 1.77273646e-01
9.09922957e-01 -2.76633597e-04 5.51541626e-01 6.03547804e-02
2.99176693e-01 -1.29428566e+00 -4.08785939e-01 3.97903562e-01
5.69710791e-01 -1.59946716e+00 -3.07008922e-02 -4.87099499e-01
-7.98455834e-01 9.59892452e-01 4.85129952e-01 -2.27827519e-01
5.90668261e-01 -3.66261333e-01 2.48038828e-01 -5.86022317e-01
2.81651169e-01 -7.74316013e-01 3.41365159e-01 5.98785818e-01
-4.30014171e-02 1.18306525e-01 -2.76001871e-01 -9.85202044e-02
-2.25138605e-01 -5.37692122e-02 5.37112474e-01 8.87364209e-01
-7.41954148e-01 -1.57961845e-01 -8.31380427e-01 2.21389219e-01
-2.44650826e-01 -2.60823876e-01 -3.83774191e-01 1.19387162e+00
4.86189514e-01 8.40271533e-01 2.19450593e-01 2.09979907e-01
5.48639186e-02 -1.60129115e-01 9.19418633e-02 -2.89318621e-01
-5.98348916e-01 6.56249896e-02 3.13611954e-01 -4.06738490e-01
-4.32904065e-01 -5.71837306e-01 -1.30785239e+00 1.59255639e-02
-1.76702857e-01 2.30633989e-01 7.84620941e-01 1.02096915e+00
-4.60715219e-02 1.99515179e-01 4.16760027e-01 -1.02660513e+00
9.09912772e-03 -9.62632835e-01 -9.15639937e-01 1.46904722e-01
3.21649849e-01 -6.35981619e-01 -4.79820400e-01 -3.80895436e-02]
|
[8.800299644470215, -1.4879510402679443]
|
e62b7a45-d31a-428f-9615-2b6c5046b7b2
|
on-mutual-information-maximization-for
|
1907.13625
| null |
https://arxiv.org/abs/1907.13625v2
|
https://arxiv.org/pdf/1907.13625v2.pdf
|
On Mutual Information Maximization for Representation Learning
|
Many recent methods for unsupervised or self-supervised representation learning train feature extractors by maximizing an estimate of the mutual information (MI) between different views of the data. This comes with several immediate problems: For example, MI is notoriously hard to estimate, and using it as an objective for representation learning may lead to highly entangled representations due to its invariance under arbitrary invertible transformations. Nevertheless, these methods have been repeatedly shown to excel in practice. In this paper we argue, and provide empirical evidence, that the success of these methods cannot be attributed to the properties of MI alone, and that they strongly depend on the inductive bias in both the choice of feature extractor architectures and the parametrization of the employed MI estimators. Finally, we establish a connection to deep metric learning and argue that this interpretation may be a plausible explanation for the success of the recently introduced methods.
|
['Paul K. Rubenstein', 'Josip Djolonga', 'Michael Tschannen', 'Sylvain Gelly', 'Mario Lucic']
|
2019-07-31
| null |
https://openreview.net/forum?id=rkxoh24FPH
|
https://openreview.net/pdf?id=rkxoh24FPH
|
iclr-2020-1
|
['self-supervised-image-classification']
|
['computer-vision']
|
[ 4.69838142e-01 2.90382087e-01 -1.33949697e-01 -4.34346706e-01
-5.33449113e-01 -5.04946768e-01 1.11118996e+00 1.58296347e-01
-4.23847169e-01 6.74273431e-01 3.19066554e-01 6.30123764e-02
-6.27371192e-01 -7.38882422e-01 -5.12178957e-01 -8.62558901e-01
-5.33619747e-02 6.12683117e-01 -2.24862233e-01 -2.12348804e-01
4.46484119e-01 5.25178373e-01 -1.59603035e+00 -4.25880775e-02
5.18049836e-01 9.28334653e-01 -6.18918911e-02 5.90340018e-01
1.43832788e-01 9.40718055e-01 -3.60982478e-01 -4.22423363e-01
2.13295758e-01 -6.22914255e-01 -9.88347471e-01 -3.52684297e-02
1.15181275e-01 5.84961176e-02 -3.47641230e-01 1.09183037e+00
1.00488164e-01 2.08372459e-01 1.08143544e+00 -1.20018470e+00
-4.02900428e-01 4.48212624e-01 -2.79153228e-01 1.82897270e-01
1.64185613e-01 -1.65745988e-01 1.45705998e+00 -7.78437376e-01
6.41339839e-01 8.50330532e-01 3.51312190e-01 3.93880874e-01
-1.54548466e+00 -2.42890403e-01 -3.87034118e-01 2.71439195e-01
-1.18256819e+00 -7.14672446e-01 7.97594070e-01 -5.16503334e-01
6.96104288e-01 2.29640573e-01 4.21121597e-01 1.22863030e+00
1.01701744e-01 5.55597782e-01 1.27977943e+00 -5.68155825e-01
1.85153767e-01 4.46971238e-01 1.19564004e-01 8.31419170e-01
5.90778530e-01 4.04113680e-01 -6.42875612e-01 -1.88373238e-01
6.98190689e-01 -2.67616302e-01 -2.37233251e-01 -9.38213527e-01
-1.37365675e+00 1.06630051e+00 5.01703858e-01 6.06431186e-01
-1.07303336e-01 2.22498596e-01 3.79759967e-01 5.47831178e-01
2.81042099e-01 7.93101251e-01 -2.49619305e-01 -2.19711259e-01
-6.21141672e-01 -2.63419468e-02 7.01074123e-01 4.48484153e-01
9.71970975e-01 -2.30662435e-01 5.10087073e-01 4.13856059e-01
3.26608062e-01 1.53240070e-01 5.59648275e-01 -9.08983946e-01
3.31797630e-01 4.70557481e-01 -8.73572286e-03 -1.00764751e+00
-4.31941807e-01 -3.61329347e-01 -8.38487804e-01 4.75103170e-01
6.06896460e-01 1.05180508e-02 -2.15409935e-01 1.89121997e+00
-1.55604616e-01 -1.54057413e-01 1.20815493e-01 6.89580381e-01
3.85252923e-01 1.06363297e-01 -1.17849782e-01 -1.95055500e-01
6.88842416e-01 -3.65600020e-01 -5.19441664e-01 -1.39346823e-01
6.75874531e-01 -4.40212041e-01 5.91156900e-01 2.88278311e-01
-7.23495543e-01 -3.81278694e-01 -1.22025681e+00 -1.18212600e-03
-3.20680350e-01 9.88483950e-02 1.02429044e+00 4.92248029e-01
-5.42921782e-01 1.00083590e+00 -8.81488800e-01 -4.50759619e-01
2.69749075e-01 6.56471550e-01 -7.67312407e-01 2.80212492e-01
-8.54479849e-01 1.32452166e+00 3.71032119e-01 1.88397989e-01
-4.53237444e-01 5.22787683e-02 -6.71822608e-01 1.69350848e-01
-4.85969521e-02 -7.12437212e-01 6.71446800e-01 -1.11953342e+00
-1.48034167e+00 9.07824278e-01 -2.91959476e-02 -4.23205823e-01
3.48903596e-01 -5.29547483e-02 -1.83093846e-01 1.72988266e-01
-1.84085041e-01 3.53058577e-01 8.96314085e-01 -9.63237762e-01
-6.60008192e-02 -4.91877198e-01 1.79082006e-01 1.19885206e-01
-2.20651492e-01 -2.92960674e-01 5.61870158e-01 -3.32315415e-01
5.92138708e-01 -9.13456678e-01 -2.21899331e-01 -2.32677683e-01
-1.56698748e-01 -1.98617235e-01 2.25953430e-01 -1.68413252e-01
7.28245020e-01 -2.21273446e+00 6.70045435e-01 4.66167629e-01
2.43042797e-01 -1.64282218e-01 5.01378328e-02 6.10195696e-01
-2.73723006e-01 1.28259305e-02 -2.98199296e-01 -6.74546361e-02
1.80319339e-01 2.51036674e-01 -2.91704208e-01 9.00461435e-01
3.81392121e-01 8.84691954e-01 -8.03655803e-01 -3.29778880e-01
3.90663654e-01 4.07200515e-01 -3.85825336e-01 1.27450705e-01
2.12528571e-01 5.06906331e-01 -5.00614583e-01 -9.87398177e-02
2.72832125e-01 -2.80578971e-01 3.62770349e-01 -2.60919541e-01
1.04885211e-03 6.94339573e-01 -1.17738676e+00 1.54677856e+00
-2.92725384e-01 9.29699004e-01 -4.81389403e-01 -1.47621310e+00
7.70809114e-01 3.94348025e-01 6.13655567e-01 -3.94521803e-01
1.68389916e-01 3.76492321e-01 2.56768137e-01 -5.03511071e-01
2.77051151e-01 -5.51098228e-01 7.93620851e-03 6.97060645e-01
6.23444796e-01 -9.12650600e-02 3.25815044e-02 1.33729160e-01
9.77877438e-01 2.30871975e-01 4.97449189e-01 -2.63639241e-01
3.68887693e-01 -3.85409653e-01 3.07800174e-01 6.38551772e-01
4.51967604e-02 7.99052000e-01 6.18805468e-01 -2.84815311e-01
-9.46421981e-01 -1.12480700e+00 -4.07149881e-01 5.82521260e-01
-6.49645403e-02 -4.99005497e-01 -4.25375372e-01 -8.35512161e-01
-3.12563598e-01 4.23649281e-01 -7.07642198e-01 -4.90638286e-01
-3.15343142e-01 -7.74558008e-01 3.45886022e-01 3.55832934e-01
1.80310741e-01 -9.12827730e-01 -7.40866601e-01 4.05156873e-02
1.73292421e-02 -9.44738567e-01 2.41365671e-01 6.81369841e-01
-1.07660902e+00 -1.23895812e+00 -1.86196104e-01 -2.68897563e-01
6.58858716e-01 1.38895765e-01 9.70343173e-01 -3.62219885e-02
3.38022746e-02 3.83149266e-01 -2.93112248e-01 -9.64593142e-02
-5.44533432e-01 3.33766431e-01 1.43496215e-01 1.78520605e-01
4.24106807e-01 -8.99791360e-01 -2.97962695e-01 1.41712084e-01
-1.01204133e+00 -1.89810365e-01 5.99083185e-01 8.50713313e-01
2.46212408e-02 -2.03959361e-01 5.60227573e-01 -9.63595629e-01
3.12857181e-01 -4.74331498e-01 -4.06363487e-01 1.28448963e-01
-6.21110380e-01 5.94494283e-01 3.66062343e-01 -8.99097770e-02
-8.29394102e-01 5.27008921e-02 -4.10383008e-02 -6.23386540e-03
-1.69476807e-01 5.51814497e-01 -5.55380806e-02 -1.87921092e-01
7.51988471e-01 6.76260442e-02 2.18718290e-01 -3.49276900e-01
4.20837849e-01 4.01091576e-01 1.76794440e-01 -4.93005365e-01
8.39268684e-01 5.99448919e-01 3.99734288e-01 -7.89301157e-01
-8.45716834e-01 -2.89341390e-01 -8.97998631e-01 1.44033641e-01
6.38949811e-01 -7.05378592e-01 -5.50186098e-01 3.72043997e-02
-8.94895196e-01 3.67892534e-02 -4.85608369e-01 7.49674737e-01
-9.80236113e-01 3.55478048e-01 -1.84540644e-01 -7.31164932e-01
1.04937337e-01 -8.63284171e-01 5.58840513e-01 -8.88059754e-03
-4.07780915e-01 -1.21456480e+00 7.57666752e-02 1.37577057e-01
4.30942237e-01 2.77608365e-01 9.15523529e-01 -6.38838351e-01
-3.85954380e-01 -2.75207728e-01 -1.09443776e-01 4.74262208e-01
2.50251472e-01 5.27572520e-02 -1.18181694e+00 -1.75533667e-01
3.94443601e-01 -3.41162115e-01 9.94472265e-01 2.30547376e-02
9.00634348e-01 -1.65705070e-01 -5.40183224e-02 4.71459746e-01
1.46128464e+00 -4.23405200e-01 5.63731909e-01 2.74634093e-01
5.48492491e-01 8.05593431e-01 1.29373357e-01 3.12731445e-01
6.72783256e-02 7.67263055e-01 3.62109631e-01 3.38355780e-01
7.04580024e-02 -2.09720075e-01 4.59126741e-01 1.04759419e+00
-3.58984560e-01 2.18458265e-01 -4.84449834e-01 2.18593836e-01
-1.86565220e+00 -1.19602466e+00 -5.88140711e-02 2.55998516e+00
4.85584617e-01 2.30945572e-01 8.60389322e-02 3.79374146e-01
2.65030503e-01 2.03932062e-01 -2.24052951e-01 -3.26419055e-01
-3.53534818e-01 5.08118212e-01 3.54963958e-01 3.52998376e-01
-8.84966671e-01 4.84319806e-01 6.87887573e+00 3.43929589e-01
-1.09213150e+00 1.40431598e-01 2.65631020e-01 5.75736389e-02
-4.20828462e-01 3.45431149e-01 -1.48883164e-01 2.12878928e-01
9.52730536e-01 -2.13424981e-01 4.09220904e-01 5.11828363e-01
-2.93947101e-01 -2.51255065e-01 -1.57832611e+00 8.56021523e-01
3.42636108e-02 -1.22351348e+00 -8.67219418e-02 3.84028822e-01
5.23601294e-01 2.84607321e-01 1.18637927e-01 -2.15442292e-02
-1.97451338e-02 -1.15487850e+00 5.70891201e-01 5.87346613e-01
4.02936965e-01 -6.94458842e-01 6.34878874e-01 3.64334613e-01
-6.64676428e-01 -8.61165579e-03 -5.69537103e-01 -3.28984708e-01
-2.99738497e-01 5.86213529e-01 -4.24634933e-01 4.04104799e-01
-1.01903766e-01 6.79023206e-01 -6.95440471e-01 9.17724848e-01
-3.44130158e-01 5.05418718e-01 -4.82592851e-01 6.25216141e-02
1.52314171e-01 -5.76992273e-01 6.20755315e-01 9.01622176e-01
1.99397936e-01 -2.57393241e-01 -4.62717146e-01 9.45080996e-01
6.97850510e-02 -5.31007908e-02 -1.02714598e+00 -2.64929950e-01
-4.67089377e-02 1.31637371e+00 -5.80823958e-01 2.22373405e-03
-4.09375846e-01 8.58869195e-01 6.18914306e-01 2.90835023e-01
-5.70241034e-01 -1.79892898e-01 5.59512079e-01 -6.08549714e-02
4.47786689e-01 -5.13334394e-01 -1.75357103e-01 -1.58188903e+00
1.40784755e-01 -7.18644798e-01 1.10541865e-01 -3.05317670e-01
-1.18362546e+00 2.76933998e-01 -5.07699810e-02 -1.14654291e+00
-5.49068511e-01 -7.68809915e-01 -4.95897114e-01 5.00047863e-01
-1.27688098e+00 -6.57078147e-01 2.52184600e-01 3.59650433e-01
-1.86827108e-02 -1.11597665e-01 1.13938391e+00 1.53545598e-02
-3.13621968e-01 2.81880409e-01 3.24685961e-01 1.27082681e-02
4.04403180e-01 -1.36808324e+00 7.04529881e-02 4.71895337e-01
1.11064541e+00 7.56854951e-01 8.84636283e-01 1.21322438e-01
-1.57829392e+00 -4.97562975e-01 8.64434302e-01 -7.72383451e-01
7.27995813e-01 -3.69519740e-01 -7.07145154e-01 6.94316089e-01
2.89547089e-02 2.51314282e-01 8.22648406e-01 3.87013435e-01
-5.61557293e-01 5.24072647e-02 -9.30373013e-01 1.93558395e-01
8.74725163e-01 -9.27859366e-01 -6.01770997e-01 3.91195387e-01
-7.29764067e-03 6.75618201e-02 -8.45739543e-01 2.45771885e-01
6.87464952e-01 -1.37684560e+00 6.99311316e-01 -7.21624017e-01
4.00913626e-01 -8.60287398e-02 -3.38610828e-01 -1.24154329e+00
-3.01846892e-01 -4.39628690e-01 -1.51118666e-01 1.01258957e+00
3.97228718e-01 -7.22690344e-01 5.49001694e-01 4.40232575e-01
6.21560514e-01 -6.07805192e-01 -1.20648396e+00 -6.88952208e-01
1.41370162e-01 -3.93037021e-01 2.61375636e-01 1.02806175e+00
3.32288861e-01 7.23687887e-01 -2.99158126e-01 -7.55919144e-02
6.81267381e-01 5.83762787e-02 5.15561342e-01 -1.45655060e+00
-5.29127359e-01 -4.38792586e-01 -9.04092431e-01 -6.67871237e-01
3.56905222e-01 -1.03208959e+00 -1.72292456e-01 -1.03883350e+00
2.40668297e-01 -5.13157606e-01 -6.51821554e-01 -7.64347538e-02
1.28407151e-01 2.01420173e-01 1.73109263e-01 3.49432647e-01
-3.77074510e-01 5.25145113e-01 6.94107592e-01 3.20434272e-02
1.99684635e-01 2.28157774e-01 -6.91818595e-01 7.72830069e-01
6.75517559e-01 -5.87097824e-01 -1.82920456e-01 -2.08006471e-01
8.26620102e-01 -1.72759309e-01 5.55597067e-01 -9.84716475e-01
1.60805956e-02 2.53530070e-02 3.56733710e-01 1.33764267e-01
3.08178395e-01 -8.45023572e-01 5.93330115e-02 1.50064051e-01
-6.30284667e-01 -2.24144280e-01 -4.03128237e-01 5.54980576e-01
-2.07919583e-01 -7.54850566e-01 5.87008119e-01 -1.82966143e-01
-3.79695803e-01 -7.06942603e-02 -3.16249579e-01 3.28654598e-04
4.97929931e-01 -5.58403246e-02 5.29640913e-02 -4.86557603e-01
-6.75378025e-01 -4.00253266e-01 4.57235157e-01 3.60236108e-01
3.50459397e-01 -1.25982690e+00 -5.42696595e-01 2.37414539e-01
1.57496214e-01 -5.18790305e-01 -3.06260198e-01 9.69025195e-01
-3.46144475e-02 3.54187548e-01 -1.86481714e-01 -4.48195189e-01
-8.10915232e-01 4.54445064e-01 3.50795001e-01 -1.42077520e-01
-5.39416432e-01 4.92289245e-01 1.31504193e-01 -2.11803049e-01
-2.08053440e-02 -2.19442457e-01 1.19403820e-03 1.15530819e-01
1.84014618e-01 2.65559226e-01 -4.08623405e-02 -8.04593503e-01
-3.37924868e-01 5.54994464e-01 7.48042017e-02 -2.55976111e-01
1.41176414e+00 -8.96097943e-02 -1.32084861e-01 8.34462881e-01
1.54772913e+00 -1.49341315e-01 -8.19412231e-01 -3.49076062e-01
3.06296438e-01 -2.50789285e-01 -8.81035533e-03 -2.29497343e-01
-8.53227794e-01 1.16643178e+00 4.37378317e-01 4.08054739e-01
7.34358072e-01 1.75033540e-01 2.70986348e-01 6.91849470e-01
3.59246284e-01 -9.54230368e-01 4.06655371e-02 2.49889031e-01
7.59241283e-01 -1.44462252e+00 2.91776180e-01 -2.34365035e-02
-3.44052434e-01 1.52825403e+00 1.54174049e-03 -5.58422565e-01
4.99085158e-01 -1.98337018e-01 -2.35850155e-01 -3.71087521e-01
-6.78536296e-01 -3.06201845e-01 4.11714941e-01 4.25992876e-01
7.40979731e-01 9.29834172e-02 -3.37751180e-01 -9.04291123e-02
-3.17408502e-01 -2.23089606e-01 5.71538687e-01 6.28812134e-01
-3.55096847e-01 -1.31977546e+00 1.02650769e-01 4.10881549e-01
-3.95522326e-01 1.87067688e-01 -4.52319831e-01 7.13054419e-01
9.57956761e-02 6.92496002e-01 -1.02454990e-01 -2.40106389e-01
-6.37878850e-02 2.09703624e-01 1.04599726e+00 -6.31715536e-01
-3.71932209e-01 -3.49216193e-01 -2.08044499e-02 -4.10466492e-01
-8.24060440e-01 -7.10759461e-01 -9.14579928e-01 5.80036305e-02
-5.96519589e-01 2.54708439e-01 7.80729830e-01 1.42981863e+00
1.50613084e-01 1.28357932e-01 7.21647441e-01 -6.42136276e-01
-1.00034153e+00 -9.68704402e-01 -5.86992979e-01 5.73615670e-01
4.33943868e-01 -8.36974204e-01 -5.60540259e-01 -2.15975061e-01]
|
[9.013495445251465, 3.1408417224884033]
|
0eeaeb37-7bd9-4309-b377-33d68b99416a
|
deep-learning-methods-for-drug-response
|
2211.10442
| null |
https://arxiv.org/abs/2211.10442v1
|
https://arxiv.org/pdf/2211.10442v1.pdf
|
Deep learning methods for drug response prediction in cancer: predominant and emerging trends
|
Cancer claims millions of lives yearly worldwide. While many therapies have been made available in recent years, by in large cancer remains unsolved. Exploiting computational predictive models to study and treat cancer holds great promise in improving drug development and personalized design of treatment plans, ultimately suppressing tumors, alleviating suffering, and prolonging lives of patients. A wave of recent papers demonstrates promising results in predicting cancer response to drug treatments while utilizing deep learning methods. These papers investigate diverse data representations, neural network architectures, learning methodologies, and evaluations schemes. However, deciphering promising predominant and emerging trends is difficult due to the variety of explored methods and lack of standardized framework for comparing drug response prediction models. To obtain a comprehensive landscape of deep learning methods, we conducted an extensive search and analysis of deep learning models that predict the response to single drug treatments. A total of 60 deep learning-based models have been curated and summary plots were generated. Based on the analysis, observable patterns and prevalence of methods have been revealed. This review allows to better understand the current state of the field and identify major challenges and promising solution paths.
|
['Rick L. Stevens', 'Jamie Overbeek', 'Austin Clyde', 'Oleksandr Narykov', 'Yitan Zhu', 'Thomas S. Brettin', 'Alexander Partin']
|
2022-11-18
| null | null | null | null |
['drug-response-prediction']
|
['medical']
|
[ 3.90768856e-01 -3.05459231e-01 -1.08271646e+00 -1.25205785e-01
-8.70224595e-01 -3.13241363e-01 5.92574477e-01 5.42834938e-01
-2.02584922e-01 9.19803798e-01 2.45141655e-01 -4.48189199e-01
-5.06283820e-01 -7.74823964e-01 -7.56765306e-02 -1.07093751e+00
9.63472575e-02 5.26815295e-01 -2.71714211e-01 -1.71618253e-01
2.89896041e-01 8.06539536e-01 -9.75925326e-01 6.49354279e-01
8.37173820e-01 1.05194461e+00 1.35077402e-01 4.60623384e-01
-1.18453354e-01 9.34098661e-01 -4.78567272e-01 -9.95670334e-02
-4.23525900e-01 -4.45795894e-01 -8.00367773e-01 -3.22253823e-01
1.50745973e-01 6.60864413e-02 -6.10163450e-01 5.49559176e-01
9.31422532e-01 -3.62698674e-01 9.17534411e-01 -6.46921992e-01
-8.77059102e-01 2.35453427e-01 -3.36823136e-01 2.76807934e-01
1.89302191e-01 1.51820421e-01 6.61598265e-01 -7.25734770e-01
5.27464092e-01 6.79514110e-01 8.72662663e-01 1.17133415e+00
-1.15324163e+00 -5.16474962e-01 -3.84920746e-01 2.36100748e-01
-1.10788763e+00 -3.83846939e-01 5.07251084e-01 -6.35002553e-01
1.09800351e+00 5.74257016e-01 6.38608515e-01 1.29821837e+00
7.44332254e-01 7.90261328e-01 8.71619403e-01 -1.65136680e-01
1.61962047e-01 1.40169010e-01 1.56165048e-01 6.95667088e-01
1.09603383e-01 2.11895466e-01 -5.00205457e-01 -3.28152359e-01
2.63820380e-01 3.96389812e-01 -4.44042623e-01 -9.45114791e-02
-9.05828238e-01 9.58587825e-01 6.39020920e-01 6.66879654e-01
-3.03635389e-01 1.42040998e-02 7.01216280e-01 -4.99181412e-02
5.20788014e-01 6.24761999e-01 -6.51559114e-01 2.04606026e-01
-8.16658318e-01 1.65026695e-01 4.41139489e-01 2.82778263e-01
2.71874785e-01 1.43646717e-01 -3.44618052e-01 8.13355565e-01
-1.67725310e-01 1.88499078e-01 7.26552784e-01 -2.82712787e-01
-6.50973842e-02 9.51127470e-01 -1.82688221e-01 -1.05021870e+00
-1.12658215e+00 -8.58000994e-01 -1.40236545e+00 -5.87109774e-02
1.83113649e-01 1.10563338e-01 -7.59247899e-01 1.18241453e+00
-5.62851205e-02 -1.35904729e-01 5.91885783e-02 4.87551153e-01
1.27271640e+00 3.48293990e-01 6.21813953e-01 -3.63124549e-01
1.11323142e+00 -8.02591264e-01 -7.26144969e-01 -8.82223062e-03
1.31846774e+00 -3.83925557e-01 5.65069437e-01 3.13215554e-01
-8.13216805e-01 -1.56514347e-01 -7.92131305e-01 -2.22408205e-01
-6.06433094e-01 3.91471654e-01 1.03559136e+00 6.32688105e-01
-7.88296878e-01 8.04175436e-01 -9.24556434e-01 -6.26512825e-01
1.02802050e+00 6.75947130e-01 -3.86969149e-01 -2.08818808e-01
-1.07260883e+00 1.11857414e+00 1.20762989e-01 -1.46088779e-01
-1.18016863e+00 -1.25723386e+00 -5.52608848e-01 -9.79487076e-02
-2.21162468e-01 -1.11157012e+00 9.24425840e-01 -4.83348787e-01
-1.31730211e+00 1.00797737e+00 -2.35222444e-01 -5.42841554e-01
1.64831862e-01 1.06422275e-01 -3.17997426e-01 -1.63597181e-01
-3.93535525e-01 3.21840644e-01 1.56989396e-01 -6.52477384e-01
-6.39281929e-01 -7.14049518e-01 -6.28897309e-01 7.89956823e-02
-7.61847019e-01 -4.25736122e-02 -4.27283123e-02 -3.84173840e-01
-2.23817036e-01 -7.49343574e-01 -6.49212658e-01 -5.41362315e-02
-2.96695262e-01 -3.63834083e-01 7.78003037e-01 -3.72817934e-01
1.47307158e+00 -1.75070298e+00 3.69586021e-01 -2.87207216e-01
6.43851936e-01 4.54354912e-01 -6.20724112e-02 6.95608079e-01
-2.07523942e-01 3.97588521e-01 -1.68595225e-01 -3.74183916e-02
-4.92758304e-01 -6.33903473e-05 -3.22845817e-01 8.40560079e-01
1.47228584e-01 1.25728714e+00 -8.67516577e-01 6.59803599e-02
4.86577272e-01 7.96992064e-01 -3.08709621e-01 6.83364198e-02
-3.27169806e-01 6.58267856e-01 -9.86650825e-01 1.16512072e+00
3.71003330e-01 -5.26085079e-01 2.08228290e-01 -1.94349721e-01
2.94658300e-02 2.89769650e-01 3.44799347e-02 1.52281260e+00
-2.81144738e-01 6.89796805e-01 -4.63418752e-01 -1.31774569e+00
9.05094385e-01 3.84741098e-01 9.54312801e-01 -8.67065966e-01
3.75545442e-01 4.59033579e-01 -5.08414358e-02 -4.97291714e-01
-1.19325228e-01 -1.61042899e-01 1.68675438e-01 -1.26412570e-01
-1.81201920e-01 2.75252927e-02 -2.31772825e-01 -3.31657022e-01
1.43037128e+00 -3.45940560e-01 5.32428920e-01 -1.87687501e-01
5.74204922e-01 3.43683362e-01 3.26163977e-01 4.40791696e-01
-3.73057634e-01 3.69176835e-01 3.94949704e-01 -1.18076301e+00
-6.32154047e-01 -4.77628410e-01 -5.90145230e-01 8.00687730e-01
-3.79536599e-01 -3.32505971e-01 -4.49105620e-01 -4.37156409e-01
-2.79121920e-02 3.49952340e-01 -1.02911353e+00 -3.61179113e-01
-3.08518946e-01 -1.49139309e+00 6.27385855e-01 4.97595131e-01
1.69707403e-01 -8.27653885e-01 -2.72913545e-01 2.60118276e-01
5.05123772e-02 -7.01358676e-01 2.38340393e-01 3.93994808e-01
-1.14130199e+00 -1.28187764e+00 -7.29286492e-01 -8.44936490e-01
4.14848387e-01 1.08825557e-01 1.09667158e+00 1.23578981e-01
-7.01145113e-01 -1.92474782e-01 -5.31921396e-03 -6.63998008e-01
-4.37271655e-01 3.04305077e-01 3.79489735e-02 -5.65834880e-01
6.47900939e-01 -2.26223230e-01 -8.03296745e-01 5.90870343e-02
-6.78124487e-01 -1.03441134e-01 8.10326517e-01 1.00458312e+00
8.04016650e-01 -1.07120395e-01 8.91040683e-01 -1.06623280e+00
6.03507519e-01 -7.59294391e-01 -1.94983944e-01 2.52184838e-01
-9.61580753e-01 -2.27894649e-01 6.57337308e-01 -1.36102825e-01
-6.04539037e-01 7.01042488e-02 -4.08518195e-01 4.29299800e-03
-1.98386490e-01 7.55238891e-01 2.43659928e-01 -3.03201169e-01
9.31066275e-01 3.65297824e-01 8.26137736e-02 -3.30409110e-01
1.54666398e-02 7.10933447e-01 9.99612361e-02 -9.53534096e-02
4.32944968e-02 4.45594102e-01 3.89667869e-01 -8.35945487e-01
-9.54810917e-01 -3.10536385e-01 -2.43869305e-01 -5.47513030e-02
7.83590913e-01 -6.38444781e-01 -8.91224504e-01 5.57346404e-01
-8.32645237e-01 -2.45170146e-01 9.84615460e-02 2.85077423e-01
-4.25190002e-01 -2.93660983e-02 -6.95154965e-01 -2.96968490e-01
-8.63825738e-01 -1.30450380e+00 7.45379090e-01 3.20805758e-01
-4.98519361e-01 -1.47308600e+00 4.84981567e-01 3.29994410e-01
7.65117586e-01 5.74906886e-01 1.31040251e+00 -7.89347053e-01
-2.34159768e-01 -6.59855187e-01 -1.03063479e-01 -1.30759835e-01
5.19807577e-01 1.88779552e-02 -1.19806159e+00 -3.53928059e-01
-2.29915991e-01 -5.21948159e-01 1.03127551e+00 9.31001842e-01
1.71498179e+00 -1.28019139e-01 -1.21382523e+00 8.17454576e-01
1.43334353e+00 5.04763365e-01 6.81861937e-01 2.41054967e-01
4.61980641e-01 3.79838139e-01 -3.01977154e-02 3.18893492e-01
-1.18336035e-02 5.46833277e-01 5.92661977e-01 -3.34697008e-01
-1.43993080e-01 9.86760035e-02 -1.44126266e-01 3.56280684e-01
-1.13587372e-01 -5.79011500e-01 -1.01306450e+00 4.13521916e-01
-1.38420010e+00 -9.03755128e-01 -3.01526994e-01 1.92971265e+00
8.38235140e-01 -7.69138262e-02 -1.34511337e-01 -1.23200476e-01
2.61322498e-01 -5.15835248e-02 -8.14364195e-01 -2.49543682e-01
-4.06944752e-01 3.66256177e-01 3.40684265e-01 1.90713987e-01
-1.11402977e+00 7.64898777e-01 7.69182301e+00 1.12766767e+00
-1.61756659e+00 -2.64391661e-01 1.24994373e+00 -5.43082319e-03
-2.12181091e-01 -3.93138528e-01 -7.02203155e-01 1.27999231e-01
1.02335835e+00 -1.90942451e-01 -2.93054953e-02 6.73229754e-01
4.45028216e-01 2.45217487e-01 -1.26306856e+00 8.32311571e-01
-1.86371148e-01 -2.24871540e+00 1.86296359e-01 2.01828584e-01
8.22703302e-01 3.38750690e-01 4.57105488e-01 1.63419589e-01
8.79481211e-02 -1.53264499e+00 -4.09065515e-01 8.48454833e-01
1.04135060e+00 -6.64309144e-01 7.26323664e-01 1.82322189e-01
-6.27413273e-01 -4.20257568e-01 -2.75048941e-01 -9.11329091e-02
-4.33050424e-01 6.38723195e-01 -8.54610145e-01 5.68760097e-01
4.45416033e-01 1.26147127e+00 -5.51020682e-01 1.04153204e+00
2.78813243e-01 6.82540596e-01 1.27637342e-01 -2.50024974e-01
1.57745227e-01 1.09339960e-01 2.79748756e-02 1.21002007e+00
3.06698501e-01 3.24803554e-02 1.17673784e-01 6.43573225e-01
-1.17556252e-01 2.63298452e-01 -5.64665556e-01 -5.99174500e-01
2.41405860e-01 1.23781431e+00 -4.57998812e-01 1.18725605e-01
-4.12775099e-01 6.46314800e-01 2.94076592e-01 2.06348836e-01
-6.11701667e-01 2.54472811e-02 8.19375277e-01 2.39602521e-01
-5.36021173e-01 2.28056982e-01 -6.93315566e-01 -6.50768936e-01
-7.85750628e-01 -8.83146942e-01 7.51892686e-01 -4.45247889e-01
-1.30536962e+00 6.05128467e-01 -5.28341115e-01 -9.93035138e-01
6.67063668e-02 -9.61584926e-01 -5.92540443e-01 8.63527358e-01
-1.45831263e+00 -1.12028968e+00 -1.93990171e-01 3.20895970e-01
6.58774912e-01 -5.22532225e-01 1.51126301e+00 2.62882590e-01
-9.79835510e-01 5.11614442e-01 6.16041601e-01 -2.17453316e-01
7.07475185e-01 -8.25091362e-01 -1.59299746e-01 -2.07007781e-01
-4.52357084e-01 4.18803751e-01 6.14010632e-01 -4.28961307e-01
-1.49964797e+00 -1.00995743e+00 8.40261400e-01 -6.21907949e-01
6.39471948e-01 1.78184852e-01 -8.32391977e-01 2.55841434e-01
1.94692969e-01 9.91902035e-03 1.29222047e+00 8.92081931e-02
-7.04217795e-03 -2.11097449e-01 -9.72551107e-01 5.37252486e-01
3.65377158e-01 -2.26889148e-01 3.15142930e-01 6.61349535e-01
2.89264172e-01 -3.22315037e-01 -9.62007105e-01 8.57123375e-01
5.65641999e-01 -7.90859342e-01 9.62822556e-01 -1.21823895e+00
7.96316624e-01 1.19023107e-01 6.39923513e-02 -1.26861537e+00
-7.11675167e-01 -2.33882427e-01 -9.89161134e-02 5.06642520e-01
6.22895122e-01 -3.26865196e-01 1.14767146e+00 6.96596980e-01
-3.89399350e-01 -1.75372756e+00 -8.32309604e-01 -1.11342244e-01
6.83803856e-01 -1.29042938e-01 2.28073910e-01 1.01768529e+00
2.02703178e-01 2.82247514e-01 -3.11076075e-01 -3.16865712e-01
2.37821326e-01 -7.61982650e-02 3.76301944e-01 -1.20796800e+00
8.86890143e-02 -9.52040255e-01 -4.93371993e-01 -4.42317247e-01
1.28078207e-01 -1.12210810e+00 -8.01347017e-01 -1.83319259e+00
6.87499702e-01 -4.23854113e-01 -6.29154384e-01 6.73741341e-01
2.61234771e-02 2.11941481e-01 -5.36874712e-01 2.46747643e-01
-1.18419066e-01 3.75085860e-01 1.26368809e+00 -7.19778538e-01
-1.21915467e-01 1.50481418e-01 -9.46867943e-01 7.01970994e-01
1.01931083e+00 -3.57416719e-01 -2.58308172e-01 -2.90867716e-01
2.64424294e-01 1.48454785e-01 3.72450389e-02 -9.16051209e-01
1.58958852e-01 -5.41271925e-01 8.51871252e-01 -5.06506681e-01
1.49160683e-01 -5.49072564e-01 2.70829529e-01 9.74378109e-01
-5.77253044e-01 -2.84529567e-01 4.60296899e-01 6.53425157e-01
-1.68626532e-01 -1.19219488e-03 9.01131988e-01 -1.78738218e-03
-5.82955182e-01 6.19920552e-01 -6.70894682e-01 -3.82745415e-01
1.13261151e+00 -8.41613486e-02 -5.08841872e-01 -2.42264252e-02
-7.85320044e-01 1.35227218e-01 1.14746116e-01 2.62458801e-01
5.98772526e-01 -1.23597836e+00 -7.80336618e-01 -1.31471097e-01
3.29288214e-01 -3.37097973e-01 6.30744040e-01 9.39313293e-01
-6.46842957e-01 8.50953996e-01 -1.77187607e-01 -4.08711344e-01
-1.21579957e+00 7.73178220e-01 9.34883714e-01 -5.87176323e-01
-3.62949610e-01 9.03462648e-01 4.03377235e-01 -2.13719681e-01
3.59211355e-01 2.87349541e-02 -6.41756535e-01 -9.44812000e-02
6.16641223e-01 4.28706825e-01 3.95900905e-01 -2.36848265e-01
-5.66127658e-01 4.26528454e-01 -4.70531553e-01 9.78352010e-01
1.57487833e+00 3.31111401e-01 -2.05111742e-01 4.08509411e-02
1.24030995e+00 -3.80474329e-01 -5.92593849e-01 6.71904162e-03
4.04402567e-03 -5.06964587e-02 3.48769516e-01 -1.30534160e+00
-1.14886928e+00 9.71948206e-01 8.35661829e-01 -5.33887371e-02
1.39073968e+00 1.46141537e-02 7.39420414e-01 3.39087963e-01
2.36405227e-02 -5.89761138e-01 1.22272149e-01 3.08795065e-01
8.05520773e-01 -1.34248590e+00 2.32483953e-01 -3.63201499e-01
-2.71235049e-01 1.38727605e+00 4.36003387e-01 2.56859034e-01
7.62982130e-01 2.71197408e-01 9.83123947e-03 -3.90738070e-01
-8.64099860e-01 2.35451058e-01 2.44824767e-01 6.46701396e-01
1.10352743e+00 1.51034296e-01 -5.67010224e-01 7.64011502e-01
2.60044515e-01 2.16508090e-01 2.83700317e-01 6.51487768e-01
-4.47019607e-01 -1.39529514e+00 9.87947807e-02 7.13036597e-01
-7.73410618e-01 -1.04453065e-01 -7.24123001e-01 5.23231447e-01
-5.91439158e-02 7.91795433e-01 -2.36401767e-01 -4.42553431e-01
1.24178678e-01 -5.07396422e-02 4.57348078e-01 -4.03689146e-01
-5.10768116e-01 -1.28316898e-02 -1.16475947e-01 -2.56711632e-01
-4.09511596e-01 -2.77019978e-01 -1.05512476e+00 -3.66943836e-01
-1.86930269e-01 -1.48706481e-01 6.22339308e-01 9.80853856e-01
5.97120106e-01 7.29754746e-01 5.21276772e-01 -4.73412603e-01
-4.17524219e-01 -8.41027021e-01 -4.69817221e-01 -8.82206485e-02
3.21336836e-01 -4.81610179e-01 1.06752068e-02 -1.29950708e-02]
|
[5.73466682434082, 5.717627048492432]
|
a57542c0-e33e-432e-b5bd-a0884bf3de16
|
from-association-to-generation-text-only
|
2304.13273
| null |
https://arxiv.org/abs/2304.13273v3
|
https://arxiv.org/pdf/2304.13273v3.pdf
|
From Association to Generation: Text-only Captioning by Unsupervised Cross-modal Mapping
|
With the development of Vision-Language Pre-training Models (VLPMs) represented by CLIP and ALIGN, significant breakthroughs have been achieved for association-based visual tasks such as image classification and image-text retrieval by the zero-shot capability of CLIP without fine-tuning. However, CLIP is hard to apply to generation-based tasks. This is due to the lack of decoder architecture and pre-training tasks for generation. Although previous works have created generation capacity for CLIP through additional language models, a modality gap between the CLIP representations of different modalities and the inability of CLIP to model the offset of this gap, which fails the concept to transfer across modalities. To solve the problem, we try to map images/videos to the language modality and generate captions from the language modality. In this paper, we propose the K-nearest-neighbor Cross-modality Mapping (Knight), a zero-shot method from association to generation. With text-only unsupervised training, Knight achieves State-of-the-Art performance in zero-shot methods for image captioning and video captioning. Our code is available at https://github.com/junyangwang0410/Knight.
|
['Jitao Sang', 'Yi Zhang', 'Ming Yan', 'Junyang Wang']
|
2023-04-26
| null | null | null | null |
['video-captioning']
|
['computer-vision']
|
[ 4.34957474e-01 1.88683763e-01 -3.76644135e-01 -2.64538109e-01
-1.13683319e+00 -2.97336996e-01 8.38984787e-01 -2.85665542e-01
-6.24433570e-02 6.59722090e-01 5.97711980e-01 -1.03968158e-01
3.12955052e-01 -4.74646777e-01 -1.04571414e+00 -3.20074767e-01
4.68851209e-01 3.50977093e-01 1.14969648e-01 -1.82831198e-01
-7.33824866e-03 -2.43766621e-01 -1.56698549e+00 1.01213503e+00
6.78822339e-01 8.01903129e-01 6.07611895e-01 7.91547418e-01
-3.24335843e-01 8.67981791e-01 -2.19101444e-01 -4.24031049e-01
1.33628190e-01 -9.40831542e-01 -5.71718514e-01 -5.48261292e-02
7.58336723e-01 -5.85854173e-01 -6.88443065e-01 8.17660689e-01
6.33939922e-01 -8.11647251e-02 7.15630889e-01 -1.74572289e+00
-1.54827833e+00 6.04507685e-01 -5.21401703e-01 -1.94534972e-01
4.94180053e-01 1.49697170e-01 7.31974721e-01 -1.17491508e+00
9.42227900e-01 1.14140999e+00 3.81297290e-01 1.19596398e+00
-1.09521079e+00 -5.77823222e-01 -1.08450577e-01 4.76846546e-01
-1.40670943e+00 -7.38195240e-01 4.73677039e-01 -5.94964445e-01
9.42971528e-01 1.52594447e-01 3.84322822e-01 1.53206170e+00
-1.72381431e-01 9.19851959e-01 8.72289956e-01 -6.31739438e-01
-2.54994016e-02 3.40151727e-01 -3.58973056e-01 4.04059500e-01
-2.09867656e-01 4.23646457e-02 -1.02739501e+00 1.74432546e-01
7.43427038e-01 -1.71060249e-01 -5.23478448e-01 -3.44876856e-01
-1.42936909e+00 7.67118394e-01 4.72181737e-01 1.50242493e-01
-1.80842087e-01 3.53271723e-01 3.44378442e-01 2.09570438e-01
3.14238697e-01 3.93475950e-01 -4.21796069e-02 -6.33571371e-02
-1.20922315e+00 2.48965733e-02 4.96291995e-01 1.43336368e+00
5.07597506e-01 6.66344166e-02 -8.35928321e-01 7.37702668e-01
2.70191848e-01 6.61544621e-01 5.11558235e-01 -8.68003249e-01
7.04237103e-01 1.95208684e-01 2.40871068e-02 -6.53084159e-01
7.55969957e-02 -4.71622571e-02 -7.55583346e-01 -1.08914621e-01
1.48865789e-01 -5.48629537e-02 -1.29295361e+00 1.89252830e+00
-1.66986093e-01 5.30098200e-01 2.90317953e-01 1.16540980e+00
1.23156273e+00 1.10102332e+00 2.42676660e-01 -6.05958216e-02
1.41438091e+00 -1.28129959e+00 -7.45788157e-01 -2.69421458e-01
4.23897564e-01 -8.46727550e-01 1.16795218e+00 -1.88210517e-01
-1.09723783e+00 -7.65575945e-01 -8.64966989e-01 -3.42914999e-01
-3.52153689e-01 1.31051736e-02 3.54413182e-01 2.28580475e-01
-1.25283670e+00 -5.12025766e-02 -5.12984395e-01 -6.74980164e-01
4.65440392e-01 -1.39822647e-01 -5.45541465e-01 -4.26660299e-01
-1.31891906e+00 8.22695196e-01 4.36826289e-01 -1.99849978e-01
-1.09577203e+00 -9.14563239e-01 -1.04095387e+00 -1.83608849e-02
1.53223112e-01 -9.44948077e-01 1.13910687e+00 -1.26899624e+00
-1.14097822e+00 8.77684712e-01 -3.54579002e-01 -4.92356896e-01
3.96834642e-01 -1.17443517e-01 -3.66919935e-01 3.79747063e-01
2.61552751e-01 1.55381703e+00 9.66765165e-01 -1.49711907e+00
-3.93426090e-01 5.55216745e-02 -1.71836428e-02 3.47399861e-01
-5.23102283e-01 6.85915500e-02 -9.31567311e-01 -5.60816705e-01
-3.17400098e-01 -9.07794237e-01 1.13931142e-01 2.40941897e-01
-3.16033363e-01 -1.01245847e-03 7.76482165e-01 -8.61768126e-01
1.10045671e+00 -2.32017231e+00 6.10729717e-02 -3.75015885e-01
-5.82134202e-02 2.81988651e-01 -6.95846677e-01 7.57396221e-01
-5.21293208e-02 6.89951032e-02 -3.01660538e-01 -4.99102443e-01
3.85424756e-02 8.67643356e-02 -6.67050242e-01 -6.37150789e-03
3.95695567e-01 1.30932033e+00 -9.97799695e-01 -7.11732030e-01
2.31828168e-01 7.60724902e-01 -4.35634673e-01 3.55412215e-01
-4.62402105e-01 3.81235003e-01 -2.12183222e-02 6.14409924e-01
6.22007847e-01 -4.07527983e-01 -2.21581385e-02 -3.99016589e-01
6.59670979e-02 -1.55659690e-01 -6.75074041e-01 2.31768775e+00
-3.90590549e-01 1.00957441e+00 -4.02627885e-01 -7.06908047e-01
5.88929236e-01 7.23835528e-01 3.19110125e-01 -1.04529917e+00
-2.26068497e-01 2.03793511e-01 -3.16937596e-01 -7.42069483e-01
6.82545304e-01 -2.84673716e-03 -8.85220692e-02 2.62308359e-01
4.62792307e-01 -8.45653787e-02 3.32926124e-01 6.09189212e-01
7.34076619e-01 4.23677742e-01 -1.15910269e-01 2.78383762e-01
9.12484527e-02 2.08599508e-01 1.81284621e-01 7.82299161e-01
4.58227918e-02 1.36543226e+00 2.53622532e-01 1.68049201e-01
-1.38907039e+00 -1.22537231e+00 6.38276860e-02 1.11721981e+00
2.22026274e-01 -4.29880321e-01 -8.19406569e-01 -3.45760465e-01
-2.50336736e-01 9.34821904e-01 -5.97509205e-01 -2.82370239e-01
-2.70196974e-01 -2.56702334e-01 6.28699839e-01 5.90676785e-01
4.29399371e-01 -1.09424794e+00 -3.58791918e-01 -9.48040688e-04
-6.30946755e-01 -1.45768952e+00 -7.67695248e-01 -4.06100005e-01
-4.22444850e-01 -7.12342143e-01 -1.35728729e+00 -1.10592258e+00
8.43986928e-01 4.53775495e-01 1.03407300e+00 -1.98123187e-01
-2.16322005e-01 7.47614563e-01 -6.18164718e-01 -3.07912260e-01
-5.41890562e-01 -1.37309656e-01 -1.47003874e-01 5.55454241e-03
2.68796861e-01 -3.30197543e-01 -6.63683593e-01 8.06843787e-02
-1.06694794e+00 7.31371820e-01 6.90365255e-01 8.24499667e-01
4.42604005e-01 -7.63330638e-01 6.44383490e-01 -4.45614696e-01
5.76042175e-01 -6.95567012e-01 -2.58224666e-01 5.86431384e-01
-2.32126847e-01 2.00103782e-02 3.98061424e-01 -5.83785534e-01
-1.01728320e+00 2.18876600e-02 1.96016893e-01 -9.49710011e-01
-4.48419787e-02 4.93615240e-01 2.60750744e-02 2.93568790e-01
6.61000133e-01 5.36990285e-01 8.29574689e-02 -2.72321135e-01
8.59341979e-01 9.01007235e-01 8.70847940e-01 -3.85219485e-01
5.98583281e-01 3.32721084e-01 -3.51721734e-01 -7.75435090e-01
-7.20691085e-01 -3.37775022e-01 -2.68757403e-01 -3.57997388e-01
1.25558364e+00 -1.35128748e+00 -1.39555350e-01 2.55175680e-01
-1.43156040e+00 -3.60344052e-01 -2.12370977e-01 3.72556597e-01
-7.39357114e-01 3.66653025e-01 -5.45580566e-01 -5.98905981e-01
-4.04159337e-01 -1.00636506e+00 1.18011868e+00 3.00731093e-01
-5.66337593e-02 -6.96178198e-01 -6.16756268e-02 5.17458320e-01
5.83293915e-01 -2.68468354e-03 6.95627987e-01 -2.66074508e-01
-6.78318918e-01 -1.80880297e-02 -6.11211121e-01 3.44720691e-01
-1.55850366e-01 -2.15498403e-01 -1.04259098e+00 -3.10886979e-01
-4.40209925e-01 -7.34769344e-01 9.26279604e-01 3.36701244e-01
1.05596876e+00 -2.00192258e-01 -2.96584845e-01 5.84917009e-01
1.51034188e+00 9.71872434e-02 1.04476225e+00 7.35688508e-02
8.18853199e-01 5.66171944e-01 4.50339109e-01 2.69842118e-01
6.58846200e-01 8.34836841e-01 2.89798468e-01 -1.86785847e-01
-7.50335634e-01 -8.60058546e-01 4.98811960e-01 7.60427475e-01
1.16220035e-01 -4.77253288e-01 -8.31358552e-01 8.56311381e-01
-2.02173758e+00 -1.14131844e+00 -2.50534322e-02 2.13476610e+00
9.13657963e-01 -2.60823697e-01 -2.32258707e-01 -6.69950545e-01
8.15296531e-01 -5.72234876e-02 -4.53614891e-01 -2.34048486e-01
-3.04141760e-01 -1.76863864e-01 3.12591225e-01 4.21105236e-01
-8.84519994e-01 1.03779066e+00 5.72231483e+00 9.90567684e-01
-1.11122847e+00 3.61972511e-01 5.04646242e-01 -1.79099411e-01
-4.27807838e-01 1.36952549e-01 -7.11054504e-01 6.43157721e-01
9.86759782e-01 -2.32343942e-01 4.14725691e-01 5.69080055e-01
1.08145095e-01 3.50980386e-02 -1.26183569e+00 1.33305216e+00
7.20546961e-01 -1.55118632e+00 4.79938954e-01 -2.37979457e-01
1.07321882e+00 1.30595416e-01 2.08271757e-01 4.37370986e-01
-2.18131170e-01 -1.08674920e+00 8.28334928e-01 7.46285319e-01
1.29380190e+00 -1.78520769e-01 4.83125836e-01 1.83163300e-01
-9.27750528e-01 1.30360082e-01 -3.91880393e-01 3.19783628e-01
5.18356562e-01 2.07531750e-01 -8.19858074e-01 4.67687100e-01
4.37611282e-01 5.84997118e-01 -4.93366361e-01 1.11579943e+00
-1.24087840e-01 3.45069826e-01 4.42928262e-02 3.16290319e-01
1.56524226e-01 2.10175082e-01 4.76745576e-01 1.22488344e+00
6.81077242e-01 -1.15396038e-01 2.69695986e-02 1.06433654e+00
-2.70648718e-01 -7.86833540e-02 -7.14778423e-01 -3.44100267e-01
4.84938383e-01 1.01744771e+00 -3.30934405e-01 -5.79810500e-01
-7.06252337e-01 1.42270315e+00 6.11630566e-02 5.93346357e-01
-1.04316437e+00 -3.11411053e-01 4.35467869e-01 1.94983706e-01
2.49296442e-01 -2.07828566e-01 3.92843597e-02 -1.33720160e+00
1.34224683e-01 -7.47969329e-01 1.76989466e-01 -1.42008519e+00
-1.34656048e+00 6.59779429e-01 1.91351950e-01 -1.53283072e+00
-4.68722790e-01 -3.94713998e-01 -4.35057044e-01 8.87041807e-01
-1.52098596e+00 -1.66695523e+00 -4.37142402e-01 7.30068088e-01
8.41058195e-01 -2.58200377e-01 8.66636217e-01 3.96328479e-01
-3.05472445e-02 7.79920518e-01 1.26572430e-01 1.61626682e-01
1.19729269e+00 -8.14981163e-01 3.68739396e-01 8.25219214e-01
2.74769962e-01 2.20467895e-01 7.07815707e-01 -6.52140796e-01
-1.41260326e+00 -1.14309692e+00 9.65225935e-01 -5.01590908e-01
5.56395173e-01 -5.66688895e-01 -8.00132573e-01 5.33660471e-01
7.92391896e-01 6.59993365e-02 6.78358793e-01 -4.50726599e-01
-6.20048165e-01 1.70756310e-01 -6.91445410e-01 7.33322382e-01
1.07348430e+00 -8.05564940e-01 -4.90855545e-01 4.58986372e-01
1.05451822e+00 -2.95143664e-01 -6.17513597e-01 1.63036466e-01
5.35386860e-01 -6.34666145e-01 8.58683705e-01 -5.45187593e-01
1.05764127e+00 -3.77308756e-01 -3.51555079e-01 -1.18330145e+00
-1.15477853e-01 -2.96975493e-01 -2.21216574e-01 1.44005048e+00
6.74657226e-01 -1.49839699e-01 5.56214869e-01 6.37312353e-01
-2.64130473e-01 -4.28370684e-01 -7.98748851e-01 -8.37348878e-01
-5.77190146e-02 -4.10451561e-01 3.10609579e-01 9.79414761e-01
1.83535144e-01 4.71774757e-01 -8.04617465e-01 -6.58755302e-02
5.68401098e-01 7.52505511e-02 7.20341742e-01 -4.16273683e-01
-5.15866220e-01 -1.93679690e-01 -3.72796267e-01 -1.05342221e+00
6.83548823e-02 -1.13699675e+00 2.45063916e-01 -2.07994294e+00
6.02900207e-01 -2.42417902e-02 -1.85110345e-01 6.48243368e-01
-6.53956756e-02 6.96857810e-01 7.06417799e-01 4.87558722e-01
-8.44736636e-01 6.87657654e-01 1.33030427e+00 -2.94197917e-01
-4.15971279e-02 -6.48148060e-01 -7.64435410e-01 1.53470054e-01
7.58369446e-01 -2.05694035e-01 -6.54459178e-01 -9.84203219e-01
9.99716595e-02 2.77657270e-01 6.07453823e-01 -1.02864480e+00
3.38355511e-01 2.09046658e-02 3.37742597e-01 -3.99195224e-01
7.01116621e-01 -5.00183940e-01 2.09888667e-01 1.16903789e-01
-6.02342844e-01 -1.38606936e-01 2.51552880e-01 4.88590419e-01
-4.10232425e-01 -1.26840144e-01 4.71676916e-01 -1.11222789e-01
-1.06456661e+00 3.08643699e-01 -2.16236174e-01 1.00247458e-01
9.79041994e-01 -2.19476089e-01 -8.22725236e-01 -6.79792345e-01
-7.06351995e-01 3.88406008e-01 5.09514213e-01 9.60515082e-01
9.00474310e-01 -1.57741177e+00 -1.09048557e+00 7.30314851e-02
6.71597540e-01 -3.72087330e-01 6.35123312e-01 8.47016096e-01
-2.91749597e-01 5.28876662e-01 -4.27773267e-01 -6.00420773e-01
-1.12410879e+00 6.56731367e-01 3.50500382e-02 1.30069882e-01
-4.87494469e-01 7.78709829e-01 4.75335211e-01 -3.55440266e-02
8.34832639e-02 2.58395821e-01 1.26382802e-02 -5.36855944e-02
8.26902986e-01 -1.04848631e-01 -3.26782495e-01 -7.48434007e-01
-1.39671355e-01 4.14424598e-01 -7.78542161e-02 -4.33361739e-01
9.16618109e-01 -2.79023886e-01 3.56901586e-02 3.93324703e-01
1.21234107e+00 -5.49974740e-01 -1.27685654e+00 -1.47417679e-01
-4.81623948e-01 -4.83773768e-01 -1.83673367e-01 -9.33875799e-01
-8.70760202e-01 1.00938046e+00 7.23133683e-01 -1.92366540e-01
1.07327569e+00 2.90939063e-01 9.32726443e-01 3.40075567e-02
1.85596585e-01 -1.10781467e+00 2.35273436e-01 4.73701417e-01
1.17991805e+00 -1.41631818e+00 -3.98049116e-01 -2.70529389e-01
-1.02612042e+00 8.69933486e-01 8.51438224e-01 2.42907703e-01
3.32448214e-01 -5.41609786e-02 7.69166052e-02 1.90698072e-01
-9.12841380e-01 -4.24050063e-01 3.86815190e-01 9.21899796e-01
4.68795240e-01 -3.47718783e-02 -1.39881134e-01 4.59026098e-01
4.08591777e-02 2.84114152e-01 5.13733745e-01 8.04357648e-01
-2.51786977e-01 -9.60906267e-01 -2.77110189e-01 3.51543367e-01
-5.14702797e-02 -4.96575385e-01 -2.51156390e-01 4.06213969e-01
8.83073881e-02 9.35898066e-01 2.50452429e-01 -3.40854526e-01
6.97316229e-02 2.04297706e-01 5.88169515e-01 -6.26486480e-01
4.30119829e-03 9.38924088e-04 1.56744733e-03 -4.44302648e-01
-5.39254367e-01 -2.56763756e-01 -1.00468278e+00 1.79181946e-03
-7.56698996e-02 5.30739687e-02 7.79266596e-01 6.17297888e-01
7.97607541e-01 3.88629794e-01 2.02253535e-01 -9.38106358e-01
-1.17189072e-01 -9.77760911e-01 -1.79302856e-01 5.61717749e-01
1.16595045e-01 -3.93866539e-01 -7.68111348e-02 4.91856992e-01]
|
[11.034482955932617, 1.0886456966400146]
|
a9f3c241-c50c-4323-bb81-51b90e0af189
|
all-optical-neural-network-quantum-state
|
2103.06457
| null |
https://arxiv.org/abs/2103.06457v2
|
https://arxiv.org/pdf/2103.06457v2.pdf
|
All-optical neural network quantum state tomography
|
Quantum state tomography (QST) is a crucial ingredient for almost all aspects of experimental quantum information processing. As an analog of the "imaging" technique in the quantum settings, QST is born to be a data science problem, where machine learning techniques, noticeably neural networks, have been applied extensively. In this work, we build an integrated all-optical setup for neural network QST, based on an all-optical neural network (AONN). Our AONN is equipped with built-in nonlinear activation function, which is based on electromagnetically induced transparency. Experiment results demonstrate the validity and efficiency of the all-optical setup, indicating that AONN can mitigate the state-preparation-and-measurement error and predict the phase parameter in the quantum state accurately. Given that optical setups are highly desired for future quantum networks, our all-optical setup of integrated AONN-QST may shed light on replenishing the all-optical quantum network with the last brick.
|
['Shengwang Du', 'Bei Zeng', 'Xuanying Lai', 'Ningping Cao', 'Chenfeng Cao', 'Ying Zuo']
|
2021-03-11
| null | null | null | null |
['quantum-state-tomography']
|
['medical']
|
[ 4.17987049e-01 1.24310516e-01 1.44219652e-01 -3.59107077e-01
-2.47996956e-01 -2.87082821e-01 3.22617531e-01 -3.18151534e-01
-6.84836328e-01 7.72346020e-01 -2.49067754e-01 -5.47947407e-01
-1.38169080e-01 -1.01770568e+00 -7.71591246e-01 -1.16740239e+00
2.45459095e-01 1.89918295e-01 -2.63097845e-02 -4.58891749e-01
3.28056157e-01 4.01792437e-01 -9.61088240e-01 3.39490138e-02
8.06687951e-01 1.08439481e+00 -7.01702908e-02 4.11150843e-01
1.82167679e-01 7.33214080e-01 -3.23032081e-01 -3.71713310e-01
4.28536355e-01 -6.11455917e-01 -7.10416734e-01 -7.35867381e-01
2.70293683e-01 -4.11281109e-01 -1.21558464e+00 1.62582898e+00
4.13918048e-01 2.74754614e-01 3.51431608e-01 -8.38165462e-01
-8.69172573e-01 5.97824991e-01 1.51896581e-01 2.93547362e-01
-4.21996005e-02 5.53077936e-01 9.73441243e-01 -4.27787632e-01
6.54051065e-01 8.38360250e-01 2.47714266e-01 8.39647472e-01
-1.55288053e+00 -7.63963699e-01 -1.00831401e+00 6.80473149e-01
-1.23972368e+00 -5.11693001e-01 7.71861792e-01 -1.66048706e-01
9.91180122e-01 -1.03242636e-01 9.58831966e-01 8.49147320e-01
3.54315758e-01 2.69769937e-01 1.53113747e+00 -8.07648301e-01
2.44360730e-01 3.23324084e-01 2.61922956e-01 1.01833391e+00
2.58246005e-01 7.79814005e-01 -7.91464269e-01 1.40583128e-01
8.80819201e-01 -1.13588706e-01 -3.40294600e-01 -7.27210492e-02
-1.31206775e+00 8.06705177e-01 9.67473447e-01 4.16445106e-01
-1.65182754e-01 4.90004063e-01 -1.71809085e-02 5.61001599e-01
1.20084971e-01 1.02961242e+00 1.11085154e-01 1.38138741e-01
-8.05066764e-01 -3.43861371e-01 6.83912218e-01 5.19102097e-01
1.28375554e+00 1.93873778e-01 -2.01836213e-01 2.06826791e-01
-3.40380110e-02 5.92109621e-01 -3.27709335e-04 -1.24565041e+00
-8.25503170e-02 3.94710600e-01 2.57055797e-02 -4.61708874e-01
-4.73315686e-01 -3.76494378e-01 -1.04454672e+00 1.45885944e-01
2.96195686e-01 -3.93035799e-01 -7.57111847e-01 1.79786503e+00
2.34805211e-01 1.89851999e-01 3.38658810e-01 1.06911111e+00
6.18292391e-01 7.59799182e-01 -6.53868735e-01 -3.16651583e-01
1.00273347e+00 -7.13164985e-01 -7.77392983e-01 1.54591560e-01
7.47326016e-01 -2.84984946e-01 6.73371911e-01 3.13284785e-01
-8.30209374e-01 -3.65328372e-01 -1.24961257e+00 -2.46302947e-01
-2.11051226e-01 -2.70429224e-01 1.04579997e+00 8.49702597e-01
-1.13929391e+00 1.26469231e+00 -7.62495160e-01 -3.30598176e-01
1.90350473e-01 7.09427774e-01 -6.57945156e-01 -4.37656082e-02
-1.33810961e+00 1.07895315e+00 8.52525771e-01 6.19111717e-01
-7.17767358e-01 -3.98016006e-01 -1.39537662e-01 -5.26954196e-02
3.16058129e-01 -9.92904186e-01 9.24932718e-01 -3.10319781e-01
-2.11911726e+00 7.58407831e-01 -1.68014914e-01 -5.51313937e-01
-2.46615514e-01 2.45439038e-01 -3.83690387e-01 4.61696982e-01
-2.74352074e-01 4.25559103e-01 5.69674373e-01 -6.68573558e-01
4.75924164e-02 -6.55225277e-01 8.60661268e-02 -4.27436948e-01
-2.91749597e-01 -1.77673072e-01 2.70119965e-01 5.36865115e-01
7.34409809e-01 -1.21928263e+00 -3.53892058e-01 -2.31778875e-01
-5.66009462e-01 2.90668779e-03 6.01225913e-01 -2.40126830e-02
7.14559913e-01 -1.92169547e+00 -1.92229196e-01 1.61042318e-01
3.79465222e-01 5.21129727e-01 3.77491079e-02 6.30163193e-01
-2.01111957e-01 -1.01884231e-01 -2.34328602e-02 -3.75280008e-02
-5.85494563e-03 1.05367534e-01 -1.76941857e-01 6.80166662e-01
1.39598042e-01 1.20716846e+00 -7.84700036e-01 -3.60435881e-02
2.67428994e-01 3.27058405e-01 -8.89748573e-01 1.51868507e-01
-1.37978226e-01 1.15651429e+00 -4.85338807e-01 5.78078151e-01
6.80545211e-01 -5.61160445e-01 2.31334623e-02 -6.40496850e-01
-5.86977243e-01 6.59292936e-01 -6.69315279e-01 1.86703002e+00
1.05801068e-01 8.04463267e-01 1.08907729e-01 -1.19545722e+00
6.04233086e-01 3.13944817e-01 3.67163897e-01 -1.30347764e+00
3.75595182e-01 4.31628466e-01 5.16473532e-01 -7.25918889e-01
5.92637241e-01 -6.52268887e-01 1.04359053e-01 4.72479939e-01
5.41020751e-01 -5.33831239e-01 6.40184656e-02 3.00872356e-01
9.42269325e-01 7.27303186e-03 5.01657911e-02 -3.98373067e-01
2.76131749e-01 -1.33337691e-01 3.05330724e-01 1.08681393e+00
-4.00578916e-01 2.76733458e-01 3.74006569e-01 -6.40258372e-01
-1.37421513e+00 -8.67450356e-01 -5.12491286e-01 3.89170319e-01
3.97095770e-01 -2.41459951e-01 -7.07176924e-01 1.35071248e-01
-4.16178882e-01 3.99870902e-01 -2.69821078e-01 -5.58070242e-01
-3.16786945e-01 -9.12107229e-01 6.25686169e-01 -1.76224023e-01
8.39851558e-01 -9.42100227e-01 -7.71649033e-02 3.53137225e-01
5.59184961e-02 -1.44231069e+00 2.74422914e-01 5.15839756e-01
-6.94352746e-01 -7.60322273e-01 -1.35931924e-01 -1.89580500e-01
4.32958215e-01 1.29332736e-01 5.59519470e-01 -8.77290815e-02
-1.45210236e-01 1.32757619e-01 1.57109015e-02 -2.37636253e-01
-6.79859996e-01 -5.58232702e-02 5.31052709e-01 1.29302830e-01
4.56663430e-01 -1.05434418e+00 -8.71006310e-01 -3.91176999e-01
-7.03129113e-01 3.03945392e-01 7.27800369e-01 8.57200444e-01
1.76274493e-01 -2.12819695e-01 1.30997583e-01 -7.58418381e-01
2.59438246e-01 -9.93402824e-02 -9.52379286e-01 1.34392917e-01
-4.92460161e-01 5.85542858e-01 6.22120559e-01 1.12851538e-01
-7.94164777e-01 -3.48190874e-01 -4.12207603e-01 -9.61474143e-03
-1.46753922e-01 6.10124171e-01 1.55916959e-01 -8.33701432e-01
7.80635297e-01 3.96268457e-01 -5.75563163e-02 5.21932282e-02
2.38014817e-01 6.53450429e-01 5.05722582e-01 -6.68065608e-01
1.00073946e+00 7.38371789e-01 8.56844664e-01 -1.15082610e+00
-1.23318553e+00 -3.77853841e-01 -7.11207569e-01 -7.22079650e-02
9.91060436e-01 -7.54424751e-01 -1.46240890e+00 5.67008138e-01
-1.39165246e+00 -9.63924751e-02 -2.27050662e-01 9.57148850e-01
-2.97224313e-01 3.82361293e-01 -9.03694034e-01 -1.16742110e+00
-3.66617620e-01 -1.26077950e+00 6.03719115e-01 6.38947427e-01
6.89493477e-01 -8.10001314e-01 7.46458545e-02 5.71729302e-01
6.30185604e-01 3.85023281e-02 1.01429808e+00 -1.33255899e-01
-1.28319716e+00 -2.37669811e-01 -4.91124868e-01 5.40385246e-01
-4.15150255e-01 -2.95761496e-01 -1.60296237e+00 -3.00969064e-01
3.42271209e-01 -6.88707888e-01 1.16964304e+00 1.79871991e-01
1.12088406e+00 -1.92276165e-02 -1.27137959e-01 8.60516429e-01
1.46538997e+00 2.14431584e-01 7.33783960e-01 5.58173880e-02
8.08628380e-01 3.99675906e-01 -3.16900104e-01 1.31052494e-01
2.51391917e-01 4.74001288e-01 7.59469211e-01 9.02968943e-02
2.16587290e-01 -2.92765982e-02 4.80127692e-01 1.23655176e+00
-5.11557400e-01 3.71917413e-04 -5.59277117e-01 -8.77374262e-02
-1.45780230e+00 -1.28148961e+00 -6.12262189e-01 2.43787766e+00
7.91960895e-01 4.03393209e-02 -4.92737889e-01 -2.02702820e-01
5.07654786e-01 1.44518003e-01 -5.02438366e-01 -3.71958822e-01
-2.76507825e-01 7.34891117e-01 4.28467363e-01 3.05710256e-01
-7.04583943e-01 1.10206223e+00 6.19414711e+00 4.30243105e-01
-1.59472406e+00 2.35427976e-01 1.89818311e-02 3.26164275e-01
-2.87381858e-01 5.72529197e-01 -7.94929445e-01 4.01855230e-01
1.27780509e+00 1.54443592e-01 9.21889424e-01 2.75008976e-01
6.49245307e-02 -4.21897084e-01 -1.06989932e+00 9.08328772e-01
-5.12898564e-01 -1.67723811e+00 -6.99165836e-02 4.19030339e-01
8.06380689e-01 5.05585313e-01 1.28571048e-01 3.50449592e-01
-3.78645599e-01 -8.62922072e-01 4.06474590e-01 6.66834712e-01
1.03955281e+00 -3.25144619e-01 5.55404603e-01 5.65105557e-01
-5.95378995e-01 -1.87376708e-01 -6.43684089e-01 -4.10764754e-01
1.88388199e-01 9.40983593e-01 -5.57118654e-01 4.82496560e-01
5.24851382e-01 5.59356034e-01 -2.56442279e-01 8.75120521e-01
-1.54795393e-01 7.31880605e-01 -3.77426237e-01 -3.01483601e-01
2.62540609e-01 -1.04373884e+00 5.91745317e-01 6.46614134e-01
3.35711837e-01 2.49977872e-01 -2.29247734e-01 1.66336811e+00
-2.82844633e-01 -5.10793686e-01 -7.22816885e-01 -7.01794565e-01
2.47814253e-01 1.49648762e+00 -3.67678165e-01 -1.97511911e-01
-6.20493703e-02 8.35743904e-01 2.78626680e-01 4.39590365e-01
-4.61999089e-01 -4.39848602e-01 4.16948020e-01 -1.53520808e-01
-1.51968375e-01 -5.64409196e-01 -4.58884379e-03 -1.75664270e+00
-1.20408639e-01 -8.49019736e-02 -4.21236813e-01 -7.76729286e-01
-1.08496702e+00 3.37630689e-01 -5.37188947e-01 -8.66878867e-01
2.93703198e-01 -1.09520018e+00 -5.14036000e-01 8.00101876e-01
-1.95914221e+00 -8.06106627e-01 -1.40183002e-01 6.53201103e-01
-7.58920193e-01 1.67779550e-01 1.22605836e+00 3.04028660e-01
-8.14351141e-01 1.38812765e-01 5.65846443e-01 3.82248014e-01
5.48209667e-01 -1.17530012e+00 6.25535101e-02 9.04379070e-01
2.97037274e-01 1.24959767e+00 6.55105650e-01 -1.81682318e-01
-2.02066803e+00 -5.45675993e-01 4.44957256e-01 -2.34177589e-01
9.98751223e-01 -6.11689806e-01 -7.69646585e-01 5.37210643e-01
4.26472574e-01 4.79642481e-01 7.60463715e-01 -2.76138424e-04
-3.11353296e-01 -3.23358595e-01 -1.06406808e+00 2.62480259e-01
1.01772130e+00 -1.36281180e+00 -3.24777812e-01 6.82421982e-01
7.44501650e-01 -2.74916351e-01 -7.72122025e-01 3.13944310e-01
6.81478679e-01 -1.32860780e+00 6.87036633e-01 -5.03958285e-01
2.32397199e-01 -2.32038915e-01 -2.27209121e-01 -1.28351998e+00
-2.77665794e-01 -1.21799386e+00 1.08271940e-02 4.59000409e-01
3.59061211e-01 -1.10948110e+00 1.04300988e+00 4.65087354e-01
-5.70007682e-01 -2.61070758e-01 -1.32138658e+00 -6.41413867e-01
1.67916752e-02 -3.63084644e-01 2.65015941e-02 8.65518749e-01
-1.25859678e-02 5.32343328e-01 -3.45578253e-01 3.03895682e-01
9.08845246e-01 -4.92338836e-02 3.82093191e-01 -1.34860981e+00
-3.96055818e-01 -2.67387658e-01 -4.35475409e-01 -1.01989090e+00
2.00809032e-01 -1.18031085e+00 -1.12925582e-01 -1.00477314e+00
2.73637325e-01 -3.60008419e-01 -4.83636528e-01 9.65640023e-02
2.38992080e-01 5.00572383e-01 4.94354824e-03 2.28577539e-01
-5.56632936e-01 7.56524920e-01 1.61934519e+00 -8.55482891e-02
3.68350185e-02 -7.50831366e-02 -2.47079492e-01 4.24226046e-01
6.08493567e-01 -6.17394149e-01 1.41357020e-01 -3.13805282e-01
7.79122770e-01 1.52452484e-01 7.95874536e-01 -1.36954761e+00
8.53913605e-01 -1.93095510e-03 2.82046851e-02 -1.90834224e-01
9.55786943e-01 -4.83965546e-01 -1.67127684e-01 7.67195225e-01
-3.22141461e-02 -6.42940879e-01 -4.68501598e-01 3.50981563e-01
-1.08173259e-01 -3.96301508e-01 1.10604680e+00 -4.46846128e-01
-6.20508552e-01 4.60505426e-01 -1.54659912e-01 -2.83845425e-01
5.11850297e-01 -2.29909971e-01 -8.32880557e-01 -6.21531680e-02
-4.76189375e-01 -3.40423286e-01 2.24468857e-01 -6.07685924e-01
4.45246547e-01 -1.01782513e+00 -1.20887317e-01 5.30661702e-01
2.93541308e-02 -1.84137687e-01 4.28764731e-01 1.22033584e+00
-6.89939082e-01 9.29252565e-01 -2.43326604e-01 -5.95336020e-01
-4.08471733e-01 3.53024691e-01 6.86020136e-01 -8.67031217e-02
-3.44852418e-01 8.94322991e-01 -2.54473209e-01 -7.49105632e-01
-4.87674385e-01 -1.53892294e-01 2.61388034e-01 -4.79555190e-01
3.30411434e-01 7.47273639e-02 1.10575788e-01 -4.75662500e-01
6.69402853e-02 3.88532311e-01 9.14015472e-02 -1.63062692e-01
1.49545085e+00 2.04305142e-01 -7.12981999e-01 4.89421129e-01
1.18473864e+00 -2.19876632e-01 -8.63621712e-01 -4.87297237e-01
-7.30378106e-02 -6.32614223e-03 4.55524892e-01 -5.26661098e-01
-7.72566736e-01 1.44940436e+00 5.99189639e-01 4.37652558e-01
6.46211684e-01 -4.94847745e-02 9.62136686e-01 1.25933635e+00
8.79412413e-01 -1.07381463e+00 3.42356488e-02 7.20114410e-01
1.24815598e-01 -1.49127042e+00 -3.21718603e-01 4.59197387e-02
1.08451441e-01 1.37235141e+00 4.01950240e-01 -1.91504672e-01
4.40619916e-01 -1.14188537e-01 -2.98432052e-01 -4.97015476e-01
-4.53564376e-01 -2.20307648e-01 1.34167433e-01 3.81578326e-01
3.54643792e-01 2.04209104e-01 1.36844248e-01 5.32554947e-02
-2.69394636e-01 -3.18168178e-02 9.52039957e-01 6.66075766e-01
-5.85993946e-01 -1.25077903e+00 -2.87949853e-02 2.16716945e-01
-3.13799560e-01 -2.98999012e-01 -8.68649110e-02 4.52043772e-01
1.90679580e-01 7.22828329e-01 -3.07759881e-01 -5.56006730e-01
1.56812549e-01 1.42643034e-01 1.05778861e+00 -6.55584574e-01
-2.49300435e-01 -3.57578844e-01 -3.69209409e-01 -6.45188391e-01
-7.28271723e-01 -6.87893927e-02 -1.29174328e+00 -6.72345400e-01
-9.07042742e-01 3.09939772e-01 1.03260076e+00 1.42730498e+00
3.69514436e-01 1.68908045e-01 6.58826411e-01 -7.88227618e-01
-6.09906673e-01 -9.51786518e-01 -9.03926551e-01 -2.01653957e-01
6.80978775e-01 -3.29224199e-01 -3.00084263e-01 -6.25383615e-01]
|
[5.577231407165527, 4.944746494293213]
|
3e38686b-9d88-4e97-b4fb-df993e3fdc23
|
analysis-of-the-operation-of-industrial
|
2112.08258
| null |
https://arxiv.org/abs/2112.08258v1
|
https://arxiv.org/pdf/2112.08258v1.pdf
|
Analysis of the Operation of Industrial Trucks based on Position Data
|
Indoor positioning systems (IPSs) can make an important contribution to the analysis and optimization of internal transport processes. The overall aim of this work is to examine how position data can be used to analyze the operation of industrial trucks in warehouses. This is achieved by developing a concept for the analysis of industrial truck operations based merely on position data. The concept consists of a signal processing scheme to derive kinematic data and three analysis methods - Monitoring, Area analysis, and Motion analysis. Schemes for the signal processing and detection of motion events were developed and implemented as part of the TrOpLocerApp (Truck Operation Localization Analyzer-Application) for recording, displaying, and processing position data, according to the proposed system concept. The TrOpLocer-App source code is published on GitLab. Different filter algorithms were examined, as part of the signal processing scheme, from which the low pass Butterworth filter has shown the best results in static experiments. Validation of the motion detection scheme shows good detection quality for distinct events in a realistic movement experiment.
|
['Jochen Kreutzfeldt', 'Johannes Hinckeldeyn', 'Hendrik Rose', 'Jakob Schyga']
|
2021-12-15
| null | null | null | null |
['motion-detection']
|
['computer-vision']
|
[-8.27371851e-02 -4.21811700e-01 5.82073152e-01 -2.58855611e-01
-3.58148158e-01 -3.69416893e-01 4.21811491e-01 3.86087090e-01
-3.67783040e-01 1.76560283e-01 -6.96087554e-02 -5.44677138e-01
-7.84424245e-01 -6.88362360e-01 -9.81480181e-02 -9.59903598e-01
-1.75500527e-01 4.90304291e-01 6.09731197e-01 -4.74875301e-01
2.18676150e-01 1.04402280e+00 -1.26972067e+00 2.13734895e-01
3.13694924e-01 7.70082176e-01 5.83837092e-01 9.14019108e-01
-1.08132269e-02 2.64993638e-01 -9.04105723e-01 4.52604741e-01
9.10244137e-02 -4.71432179e-01 -4.05043036e-01 -2.32613698e-01
-6.94893897e-01 4.45256196e-02 -9.93276909e-02 6.89788759e-01
6.53611302e-01 2.48969316e-01 5.76466382e-01 -1.21921241e+00
3.89723897e-01 4.34810668e-01 8.20628256e-02 4.10394758e-01
5.56440175e-01 -7.72427842e-02 1.55123070e-01 -5.39642453e-01
4.85465854e-01 9.14140999e-01 7.43892312e-01 -3.70098144e-01
-1.29369915e+00 -3.79234642e-01 -5.28013110e-01 5.09583056e-01
-1.40798640e+00 -2.11801544e-01 8.54378939e-01 -4.32300478e-01
7.24796295e-01 6.22031152e-01 4.99910712e-01 7.50058711e-01
6.67849481e-01 3.93070996e-01 9.42694128e-01 -6.03873789e-01
2.82021075e-01 2.72262871e-01 4.02446508e-01 1.81232154e-01
2.29090899e-01 6.67907763e-03 -1.31444171e-01 4.12020162e-02
6.70780003e-01 -2.40237921e-01 -1.86639935e-01 -3.26111823e-01
-1.15379882e+00 5.29716372e-01 2.33391106e-01 1.00401294e+00
-6.75529540e-01 7.25840405e-02 4.94397789e-01 3.24585766e-01
2.66805410e-01 1.12700760e-01 -5.37899852e-01 -4.75868553e-01
-7.67220616e-01 2.99993843e-01 8.72399151e-01 7.94620216e-01
2.85500228e-01 6.46064207e-02 -1.50337582e-02 3.80555063e-01
5.83352625e-01 6.69602096e-01 3.87979388e-01 -5.14889061e-01
4.16237235e-01 2.34742537e-01 1.64804235e-01 -1.09260035e+00
-1.03101301e+00 -1.18468545e-01 -1.98178068e-01 2.34240562e-01
5.63331604e-01 -5.19699454e-01 -4.32203203e-01 9.08004105e-01
3.53864461e-01 -3.10305446e-01 -9.59775075e-02 5.31890750e-01
5.14665544e-01 9.38417792e-01 -2.33023778e-01 -3.34296852e-01
1.21952319e+00 -1.57625332e-01 -1.42833281e+00 3.71440381e-01
7.20560730e-01 -1.21397734e+00 3.96006674e-01 5.69054246e-01
-7.95862019e-01 -8.35887372e-01 -9.30853605e-01 6.18577421e-01
-6.19553149e-01 5.11913776e-01 2.56609201e-01 8.75198901e-01
-8.24542999e-01 5.79041064e-01 -1.11478090e+00 -7.75201738e-01
-4.94778723e-01 4.69512582e-01 -1.94499731e-01 4.35771883e-01
-1.00910664e+00 1.12994599e+00 3.82249027e-01 8.01920891e-01
-2.56862640e-01 -4.03487772e-01 -7.67980754e-01 -4.99653704e-02
-2.87454184e-02 -1.94045663e-01 1.00254357e+00 -1.18138865e-01
-1.61044192e+00 1.00250974e-01 -1.05139501e-01 -4.34351206e-01
7.02962458e-01 3.54250968e-02 -9.77774322e-01 7.35643730e-02
2.76902914e-01 -2.77785480e-01 5.57020187e-01 -1.01607037e+00
-8.72857034e-01 -1.73472896e-01 -5.55421948e-01 -2.25314960e-01
3.44486803e-01 1.57831833e-01 -1.25993207e-01 -3.89158189e-01
6.21181071e-01 -8.02071691e-01 -1.18095323e-01 -8.08363616e-01
-1.48524687e-01 1.44260705e-01 9.57504511e-01 -8.44757438e-01
1.22417283e+00 -2.18773890e+00 -1.96410805e-01 8.50288332e-01
-6.25685036e-01 1.00308113e-01 4.08894628e-01 8.57839346e-01
-2.79117316e-01 -8.05382669e-01 3.49847823e-01 1.74884975e-01
1.17049575e-01 1.22692153e-01 -4.18482162e-02 6.71992838e-01
-2.46070281e-01 3.69938433e-01 -6.20936632e-01 -9.98648778e-02
1.05577862e+00 3.97553802e-01 -5.07108048e-02 -2.80580789e-01
4.94264990e-01 5.26539266e-01 -4.39026028e-01 4.86392736e-01
5.76592743e-01 5.86748779e-01 2.57129252e-01 -5.21080732e-01
-5.80396235e-01 5.17653227e-02 -1.78754187e+00 1.21770287e+00
-5.82535207e-01 8.60382795e-01 4.48720574e-01 -9.49004114e-01
1.14561868e+00 5.86909711e-01 9.35002208e-01 -6.58792734e-01
5.14306843e-01 4.25159395e-01 4.45210002e-02 -9.67461050e-01
6.23822510e-01 7.49229938e-02 4.21880223e-02 -1.64011911e-01
-2.95945536e-02 -2.17188969e-02 5.53342402e-01 -1.85627714e-01
9.66342747e-01 2.57024258e-01 1.43192843e-01 -5.22556543e-01
7.81044126e-01 1.25854611e-01 1.37698233e-01 2.44973555e-01
-8.63701776e-02 8.91407877e-02 4.22248721e-01 -2.12624148e-01
-5.82189322e-01 -7.79757500e-01 -5.02705216e-01 5.61929643e-01
3.08963478e-01 -2.36279324e-01 -5.81549883e-01 1.37760071e-02
1.62783060e-02 8.45728159e-01 -1.59490928e-01 8.53916444e-03
-6.52495384e-01 -9.05937970e-01 2.03355968e-01 3.81271303e-01
1.93994239e-01 -8.56038272e-01 -7.69443810e-01 7.54703522e-01
-2.49323510e-02 -8.79365027e-01 3.19922119e-01 6.22949481e-01
-8.86329770e-01 -1.04886210e+00 -3.67066622e-01 -5.70087135e-01
5.07580280e-01 2.60367841e-01 2.97477216e-01 -3.88842821e-01
-3.06633055e-01 5.19211113e-01 -3.92739922e-01 -6.52714789e-01
-6.22672439e-01 -1.83650795e-02 -1.21366919e-03 -8.72616540e-04
2.32803702e-01 -3.93879950e-01 -2.91673392e-01 8.03998828e-01
-5.18729329e-01 -5.92239082e-01 4.60192442e-01 1.59425452e-01
-1.05635123e-02 6.18941247e-01 4.10451800e-01 -6.50253832e-01
7.20668256e-01 -3.42457354e-01 -9.84975338e-01 -1.64778128e-01
-4.21119362e-01 -3.25449973e-01 3.81544620e-01 1.25145555e-01
-1.04599464e+00 1.10980414e-01 -5.82030833e-01 1.14830427e-01
-6.06363595e-01 2.93778151e-01 -4.09266084e-01 -2.90988088e-01
5.80865145e-01 4.69674058e-02 1.28644928e-01 -6.83419704e-01
-5.11568785e-02 6.58240020e-01 4.70264226e-01 -1.79109275e-01
7.62322307e-01 3.05261791e-01 3.11427534e-01 -1.35914063e+00
3.97354633e-01 -1.05150568e+00 -6.52473986e-01 -7.10547626e-01
8.74752343e-01 -5.09177387e-01 -9.62505698e-01 3.32990944e-01
-1.17587483e+00 -1.36306062e-01 -2.49790251e-01 9.09377694e-01
-3.92603189e-01 2.29103759e-01 -5.08714616e-01 -1.01360226e+00
2.15854496e-01 -1.13538742e+00 8.15843284e-01 2.86079608e-02
-4.60590869e-01 -1.01495278e+00 3.74338813e-02 1.31563678e-01
5.57287335e-01 3.43887299e-01 4.03132558e-01 -6.01945758e-01
-4.63146955e-01 -7.55899072e-01 4.88394111e-01 1.89868182e-01
1.57280937e-01 7.57650658e-02 -8.65124702e-01 -3.57648470e-02
3.77798915e-01 8.81709337e-01 3.16983968e-01 7.58366287e-01
4.32271570e-01 3.71402234e-01 -6.21766806e-01 3.12109977e-01
1.52594841e+00 8.74155581e-01 7.45237350e-01 5.32588005e-01
2.00056702e-01 6.10076547e-01 1.11813152e+00 3.55613202e-01
-2.28157148e-01 1.14675879e+00 3.82558286e-01 -5.89295849e-02
2.16941997e-01 3.66096824e-01 4.72202063e-01 6.38417304e-01
-3.66364777e-01 -1.58989310e-01 -7.49958813e-01 4.14967865e-01
-1.77819240e+00 -1.00504673e+00 -1.34267950e+00 2.05619931e+00
-1.87246412e-01 5.27230740e-01 7.99807683e-02 7.44670093e-01
6.99689269e-01 4.22855541e-02 6.10750496e-01 -6.77579463e-01
5.20483851e-01 1.42248496e-01 1.06283045e+00 7.07993627e-01
-1.07860947e+00 1.06736399e-01 5.90701628e+00 6.68188095e-01
-9.99837339e-01 2.71420270e-01 -4.69530731e-01 2.13768423e-01
4.49965268e-01 -2.40193859e-01 -8.63140762e-01 8.30828309e-01
1.22762024e+00 2.10026905e-01 -2.10667029e-02 8.10429990e-01
9.20065343e-01 -5.66678703e-01 -8.48198116e-01 6.55470133e-01
-2.25005999e-01 -1.02374160e+00 -6.49536252e-01 2.78902709e-01
1.56308904e-01 -1.96370706e-01 -4.01092857e-01 5.54421730e-02
-3.72573555e-01 -1.89982101e-01 9.97010887e-01 5.35407305e-01
-1.33583963e-01 -8.38391125e-01 1.28301501e+00 1.84500918e-01
-1.43798053e+00 -2.45365143e-01 -1.15802236e-01 -6.69103023e-03
9.58241999e-01 6.97903156e-01 -1.04604185e+00 1.30052519e+00
3.84635955e-01 5.56335412e-02 -1.98941261e-01 1.25864804e+00
-8.12463835e-02 4.89177197e-01 -6.55012131e-01 -1.50384143e-01
1.38109848e-01 -3.47760260e-01 8.03901315e-01 1.42902172e+00
5.49836159e-01 -2.69801706e-01 -2.73570746e-01 6.09864652e-01
1.07676649e+00 1.47834988e-02 -4.72639322e-01 5.00857651e-01
2.15030774e-01 1.22099614e+00 -1.15250027e+00 1.22771664e-02
-8.50326866e-02 6.07900262e-01 -1.05886757e+00 2.86881119e-01
-1.01313019e+00 -8.49199951e-01 4.01242584e-01 4.22297567e-01
3.42762738e-01 -6.82423949e-01 -2.21492425e-01 -1.96318954e-01
-8.69602412e-02 -2.20456779e-01 1.26689717e-01 -7.34535456e-01
-4.50929374e-01 3.07040870e-01 4.62542415e-01 -1.40257537e+00
-4.30168182e-01 -1.09560645e+00 -7.50112712e-01 9.69779611e-01
-8.15497756e-01 -8.26341271e-01 -2.77411222e-01 5.04997730e-01
5.80680072e-01 -8.84830356e-02 6.64331436e-01 6.45904124e-01
-4.08526450e-01 -2.54754692e-01 4.72790629e-01 -1.96904555e-01
3.16737950e-01 -1.26253641e+00 4.05268595e-02 9.93260086e-01
-3.68248895e-02 4.67015028e-01 1.19680309e+00 -6.49966300e-01
-1.28286791e+00 -6.12498462e-01 1.04101157e+00 -5.38081169e-01
7.17553854e-01 -3.38141710e-01 -6.13661766e-01 5.82775950e-01
8.28646496e-02 -3.46768379e-01 3.54803979e-01 -1.50728896e-01
9.37766790e-01 -4.46797639e-01 -1.05025125e+00 1.43457457e-01
2.97997147e-01 -2.19773144e-01 -7.71743119e-01 4.16656435e-01
3.00191287e-02 -7.02311933e-01 -8.24311197e-01 1.21564217e-01
3.85755867e-01 -9.86384809e-01 9.74060774e-01 2.72271037e-01
-3.62553358e-01 -7.72193313e-01 1.90425009e-01 -1.10771561e+00
-3.61576468e-01 -7.34609187e-01 4.32266086e-01 1.26858950e+00
3.04844260e-01 -9.51624691e-01 3.75684232e-01 1.38462573e-01
-2.58405119e-01 -2.47690510e-02 -1.01910245e+00 -7.50362039e-01
-9.64017749e-01 -7.47802734e-01 1.96023792e-01 5.17320633e-01
1.91758841e-01 1.57169029e-01 1.61132917e-01 7.32275248e-01
2.27605209e-01 -3.05418849e-01 7.67413974e-01 -1.10513675e+00
-2.75484234e-01 -1.93285733e-01 -1.00639725e+00 -4.37124908e-01
-5.82351863e-01 -4.24655199e-01 1.00269005e-01 -1.73895550e+00
-9.56119657e-01 -1.51565284e-01 -1.42592177e-01 -1.56478241e-01
5.19315958e-01 -6.42539635e-02 1.80923611e-01 1.89020596e-02
1.18828930e-01 -8.44826400e-02 5.84821999e-01 2.74688900e-01
-3.64640683e-01 8.54290783e-01 2.86462933e-01 5.87222517e-01
6.12164497e-01 -3.55485916e-01 -4.13550198e-01 -4.70423773e-02
1.98963791e-01 -1.70732625e-02 5.34668028e-01 -1.54281890e+00
4.04427767e-01 3.76628757e-01 4.26618665e-01 -1.14763772e+00
3.38012040e-01 -1.64324284e+00 1.00234246e+00 1.03068900e+00
1.90308124e-01 3.92110288e-01 2.90053874e-01 4.70676363e-01
-2.14825347e-01 -4.97916788e-01 3.22203338e-01 1.79674238e-01
-9.58128870e-01 -7.07045734e-01 -9.00798917e-01 -8.92983854e-01
1.32690322e+00 -5.77803016e-01 -9.44734886e-02 -2.33621061e-01
-1.04603672e+00 -2.43604137e-03 -5.73050370e-03 2.07354501e-02
5.04899696e-02 -1.20962417e+00 -2.05588162e-01 5.30695140e-01
-2.34069124e-01 -3.66669148e-01 4.04750884e-01 1.35983205e+00
-1.29706216e+00 7.28813946e-01 -2.80084968e-01 -7.17575431e-01
-1.34877360e+00 5.02474964e-01 3.83350074e-01 -3.74612622e-02
-5.01625299e-01 3.02494556e-01 -4.47912127e-01 -3.01606301e-02
1.23840041e-01 -9.83563304e-01 -4.43873078e-01 4.36089754e-01
3.55972439e-01 9.81053114e-01 7.34072447e-01 -6.95355356e-01
-6.26513481e-01 7.69380987e-01 6.14551723e-01 -3.04299623e-01
1.17306495e+00 -2.46633142e-01 -5.74834048e-05 5.59120953e-01
1.09278059e+00 3.62908244e-01 -6.47060573e-01 6.25362039e-01
4.10138279e-01 -3.77304763e-01 6.17484637e-02 -5.93572974e-01
-6.11105442e-01 4.68310714e-01 8.77925158e-01 9.94937301e-01
1.02126598e+00 -4.76261705e-01 3.59663278e-01 1.37888491e-01
7.03474939e-01 -1.28180492e+00 -7.48767912e-01 3.44178557e-01
5.76509953e-01 -5.52121043e-01 -5.22361845e-02 -7.15299249e-01
-3.02074969e-01 1.43156910e+00 -2.94316441e-01 1.94305982e-02
8.63728642e-01 6.08981013e-01 -6.19809143e-02 -2.42872924e-01
-1.16840629e-02 -3.11766624e-01 -8.57670307e-02 8.14453840e-01
3.43766510e-01 1.54951245e-01 -7.68095911e-01 4.96848464e-01
3.64080444e-02 8.04393962e-02 3.58208805e-01 1.26840901e+00
-5.76151550e-01 -1.06361020e+00 -1.00894737e+00 2.63621174e-02
-4.39833999e-01 5.85709155e-01 8.00517574e-02 1.26808977e+00
3.48852009e-01 1.00238740e+00 1.04469202e-01 -3.31763953e-01
1.08756065e+00 1.17661215e-01 4.03643847e-01 -1.74287751e-01
-8.59851718e-01 2.80921578e-01 4.40412045e-01 -8.60466421e-01
-2.48820290e-01 -8.71939063e-01 -1.24988198e+00 -1.27512395e-01
-3.84567291e-01 4.19225603e-01 1.54581189e+00 8.03566039e-01
-1.36895105e-01 1.25466764e+00 6.15459740e-01 -1.08513296e+00
-2.40397543e-01 -8.95578325e-01 -1.04061091e+00 -7.26288036e-02
3.47243249e-02 -5.66638172e-01 -3.52556139e-01 -2.36326153e-03]
|
[6.558043003082275, 1.7863552570343018]
|
4685edec-b3c0-4a92-a5a2-993dfd1b107b
|
connecting-levels-of-analysis-in-the
|
2305.06037
| null |
https://arxiv.org/abs/2305.06037v2
|
https://arxiv.org/pdf/2305.06037v2.pdf
|
Connecting levels of analysis in the computational era
|
Neuroscience and artificial intelligence are closely intertwined, but so are the physics of dynamical system, philosophy and psychology. Each of these fields try in their own way to relate observations at the level of molecules, synapses, neurons or behavior, to a function. An influential conceptual approach to this end was popularized by David Marr, which focused on the interaction between three theoretical 'levels of analysis'. With the convergence of simulation-based approaches, algorithm-oriented Neuro-AI and high-throughput data, we currently see much research organized around four levels of analysis: observations, models, algorithms and functions. Bidirectional interaction between these levels influences how we undertake interdisciplinary science.
|
['André Longtin', 'Richard Naud']
|
2023-05-10
| null | null | null | null |
['philosophy']
|
['miscellaneous']
|
[-2.07229145e-03 -1.81201130e-01 -4.82801124e-02 1.14521459e-01
3.47572267e-01 -5.50663769e-01 1.08905733e+00 3.32893610e-01
-4.63958919e-01 8.57894897e-01 8.82069990e-02 -3.64646912e-01
-4.64713335e-01 -6.37021780e-01 -2.79410720e-01 -7.78619766e-01
-8.30350593e-02 2.88468599e-01 3.05771768e-01 -4.40044969e-01
6.58037364e-01 7.22740829e-01 -1.59998894e+00 -2.15135425e-01
6.65633917e-01 6.33857191e-01 -5.92131056e-02 7.54929602e-01
-4.28606629e-01 5.14149129e-01 -2.85090387e-01 -2.46495754e-01
-1.64054453e-01 -6.46115780e-01 -6.36162758e-01 -3.92167807e-01
-5.65133274e-01 7.28074312e-01 -2.51359046e-01 1.03515816e+00
3.96315008e-01 -1.99105963e-01 6.90707505e-01 -9.53252435e-01
-8.07130933e-01 1.29940867e-01 -3.33971709e-01 6.89245403e-01
2.19107464e-01 3.44356924e-01 3.49430054e-01 -6.65863156e-01
7.34984398e-01 1.35388207e+00 5.28396249e-01 5.61509788e-01
-1.39529049e+00 -3.48643988e-01 5.76410666e-02 1.59988329e-01
-1.10029256e+00 -5.75107217e-01 4.46408451e-01 -6.27841711e-01
8.62610102e-01 2.28049816e-03 9.70887899e-01 8.21789384e-01
7.18799293e-01 2.15240009e-02 1.50683165e+00 -4.83612955e-01
5.10853350e-01 3.64065439e-01 3.11278731e-01 2.55827188e-01
6.14443839e-01 4.43841755e-01 -5.60544312e-01 1.54899284e-02
9.06591296e-01 3.37181278e-02 -2.38294117e-02 -3.91765498e-02
-1.11742365e+00 5.46182871e-01 1.22308172e-02 8.27150881e-01
-4.24820483e-01 1.96762219e-01 3.65600824e-01 4.04757798e-01
3.70657384e-01 5.59538662e-01 -4.41606969e-01 -2.79983729e-01
-4.42214847e-01 3.49787444e-01 1.03597152e+00 1.28960431e-01
3.39137912e-01 -2.41090208e-02 5.30055344e-01 3.41550648e-01
5.76522291e-01 2.58509189e-01 6.95178449e-01 -9.59437430e-01
-2.00837418e-01 6.70382619e-01 -1.09329537e-01 -7.82399595e-01
-6.29651546e-01 -1.97819769e-01 -5.66589236e-01 4.20091867e-01
4.68483478e-01 -3.07984799e-01 -7.58103669e-01 1.60663688e+00
3.81274283e-01 -2.09802147e-02 2.41349503e-01 7.31578052e-01
7.11902142e-01 6.28821254e-01 1.71627134e-01 -4.14582580e-01
1.28898263e+00 -4.89028364e-01 -8.52728128e-01 1.60647318e-01
2.77961612e-01 -4.51074839e-01 5.25933385e-01 5.71747065e-01
-1.32438612e+00 -1.66932225e-01 -8.98191214e-01 2.17318803e-01
-9.85928833e-01 -4.68078494e-01 7.23520577e-01 5.36906362e-01
-1.29306698e+00 7.06036150e-01 -9.75646198e-01 -7.90101826e-01
3.35303009e-01 4.36303616e-01 -6.31990507e-02 5.49644947e-01
-1.03603888e+00 1.53471982e+00 2.16308594e-01 -5.40066101e-02
-4.45584476e-01 -4.87091959e-01 -2.01639339e-01 -2.91652858e-01
-4.48154006e-03 -9.30195510e-01 7.65635669e-01 -7.64076352e-01
-1.56159687e+00 1.16198814e+00 -3.87827098e-01 -5.76338291e-01
1.19842812e-01 2.64596373e-01 -6.82524502e-01 3.94469202e-02
-4.83762413e-01 4.32313234e-01 1.41327977e-01 -1.34501529e+00
-4.31515992e-01 -8.62330377e-01 -1.92513973e-01 3.16289999e-02
-1.84188355e-02 2.11155981e-01 2.44206071e-01 -1.83954716e-01
3.04002911e-01 -8.04201305e-01 -4.08137202e-01 1.08692043e-01
1.39727443e-01 -4.32215422e-01 6.73286438e-01 4.93893819e-03
9.94270802e-01 -1.74263990e+00 4.68990326e-01 1.36189133e-01
6.80225730e-01 3.26064527e-01 3.90028775e-01 9.85499203e-01
-2.02212632e-01 2.67730743e-01 -3.67356241e-02 2.41363183e-01
-1.93428397e-01 4.78450544e-02 1.94296949e-02 7.63765275e-01
4.61176559e-02 1.04164791e+00 -9.45710361e-01 -6.75076693e-02
3.93007785e-01 6.33253098e-01 -1.48475751e-01 -2.06569746e-01
4.93802540e-02 5.85108161e-01 -6.47369385e-01 6.58241987e-01
1.06156461e-01 -3.51952761e-01 4.35393512e-01 2.10156709e-01
-7.78148413e-01 2.03267738e-01 -6.95301890e-01 1.44277644e+00
-2.57397536e-02 9.62237656e-01 1.06254876e-01 -1.42047226e+00
9.38207269e-01 5.99820197e-01 5.67843735e-01 -8.39532316e-01
5.07102609e-01 2.39617214e-01 8.49468648e-01 -7.19558060e-01
-2.38509402e-01 -2.65902698e-01 4.66901511e-01 5.51091373e-01
1.09743454e-01 -6.25637695e-02 8.90077427e-02 -7.99902603e-02
1.17696023e+00 3.11887205e-01 5.74949741e-01 -6.83066189e-01
5.47151685e-01 1.50435254e-01 1.49986818e-01 4.56662804e-01
-6.64211631e-01 -2.56020948e-02 3.27821165e-01 -6.62223279e-01
-1.20141149e+00 -1.13413477e+00 -6.13573074e-01 6.09687269e-01
3.18407565e-01 -8.99302140e-02 -9.50331867e-01 5.07736683e-01
-5.30186743e-02 1.97434202e-01 -9.86378610e-01 -2.20351607e-01
-3.06119621e-01 -9.67250586e-01 2.83137977e-01 5.01622744e-02
4.48747240e-02 -1.39768243e+00 -9.85582471e-01 3.88345182e-01
7.10655570e-01 -8.59421790e-01 6.78827286e-01 5.04316211e-01
-1.17326939e+00 -8.71546149e-01 -3.89982611e-01 -4.25049067e-01
4.57139015e-01 1.72087699e-01 1.06727278e+00 2.89259195e-01
-4.87572849e-01 2.81345397e-01 -2.35867307e-01 -9.42915738e-01
-4.26102459e-01 -2.64186442e-01 4.64352310e-01 -3.35385770e-01
4.80893314e-01 -1.45910978e+00 -6.10931277e-01 -1.18877046e-01
-9.26361561e-01 -1.37217656e-01 5.89070201e-01 2.89877176e-01
1.74742505e-01 -4.31556523e-01 8.78876269e-01 -7.60558426e-01
8.57541502e-01 -6.35685623e-01 -1.47497714e-01 2.67153502e-01
-8.21895361e-01 4.27294485e-02 5.88529646e-01 -4.07733142e-01
-5.16947925e-01 -4.64835286e-01 -1.35117453e-02 -1.28035665e-01
-5.01480997e-01 3.90406489e-01 -4.49720211e-02 -2.87919760e-01
7.96275556e-01 5.52282989e-01 5.07238746e-01 -2.08427846e-01
2.78203458e-01 4.10853714e-01 5.10677814e-01 -5.17995894e-01
6.90376580e-01 6.52738035e-01 1.18686914e-01 -1.08141375e+00
-2.69745886e-01 -6.09055758e-02 -7.00893879e-01 -5.65252125e-01
8.04642618e-01 -2.65366077e-01 -1.04743576e+00 2.63765424e-01
-1.14576972e+00 1.77532788e-02 -3.55755776e-01 5.95720768e-01
-4.50330019e-01 -1.05244555e-01 -3.51157635e-01 -1.26962805e+00
-1.74497187e-01 -8.52396965e-01 5.16969025e-01 6.86669350e-01
-3.04933608e-01 -1.42815876e+00 6.39493048e-01 -6.09387867e-02
5.44155896e-01 4.72950757e-01 9.02981043e-01 -6.63695753e-01
-4.70452420e-02 -1.53560653e-01 2.39744894e-02 -2.73021430e-01
6.14627898e-02 3.54486823e-01 -9.71517324e-01 2.16860741e-01
2.88148016e-01 -3.27833295e-01 4.51140970e-01 6.15111589e-01
7.31308281e-01 8.87133107e-02 -6.32666767e-01 3.14549327e-01
1.48365605e+00 9.29334939e-01 8.23992312e-01 3.80575508e-01
1.68228492e-01 9.61288810e-01 -3.38413008e-02 1.58856615e-01
8.39612037e-02 2.23666921e-01 6.17779829e-02 7.61747062e-02
1.53045997e-01 1.70342714e-01 1.61716044e-01 1.11119866e+00
-6.27304733e-01 -1.00540459e-01 -1.02051580e+00 1.86249539e-01
-1.87719095e+00 -1.14884269e+00 -3.79259318e-01 1.96075213e+00
5.44949174e-01 5.07458270e-01 4.15698498e-01 1.76191628e-01
7.01086164e-01 -3.32505018e-01 -8.01302493e-01 -7.26492584e-01
-1.17936529e-01 1.67914107e-01 2.11293504e-01 3.84068519e-01
-2.48918414e-01 9.00312603e-01 8.31912994e+00 5.72359324e-01
-1.17462981e+00 -1.58944130e-02 6.61574006e-01 -9.02937609e-04
-2.80610055e-01 5.66341355e-02 -6.16909564e-01 7.27599621e-01
1.58304226e+00 -4.07225877e-01 8.04263175e-01 6.67857826e-02
8.17639947e-01 -5.18356860e-01 -8.16564858e-01 7.51704216e-01
-2.11058140e-01 -1.48809075e+00 -1.39573347e-02 4.08623576e-01
3.71469110e-01 1.49216056e-01 4.35398184e-02 -1.64049968e-01
3.68148863e-01 -1.31985044e+00 5.00229776e-01 1.06638551e+00
9.17973816e-02 -3.21799785e-01 1.75125539e-01 4.82531101e-01
-7.78475583e-01 3.85909304e-02 -2.11469054e-01 -7.56903052e-01
9.79214981e-02 4.81818467e-01 -4.84278388e-02 1.46338344e-01
4.71363753e-01 5.03153205e-01 -2.07575515e-01 9.92195666e-01
3.57976884e-01 4.43363458e-01 -2.47559428e-01 -5.90649903e-01
1.31559342e-01 -6.40021920e-01 6.87702656e-01 9.71917868e-01
-2.94323087e-01 4.77585971e-01 -4.60406542e-01 1.10015059e+00
5.77623248e-01 -2.32438400e-01 -7.76168346e-01 -5.02732992e-01
5.91263175e-01 1.48732758e+00 -1.23854411e+00 -2.88893878e-01
-4.70878363e-01 1.39156058e-01 1.18437178e-01 2.92019516e-01
-6.77193820e-01 7.41712078e-02 8.98014426e-01 1.31166786e-01
-1.03021450e-01 -5.10722637e-01 -7.04106629e-01 -9.27708924e-01
-4.88158226e-01 -7.60931253e-01 -3.25500607e-01 -5.52262187e-01
-1.02128279e+00 2.92180240e-01 -2.74443358e-01 -6.45946741e-01
3.09667867e-02 -8.10857654e-01 -9.09323394e-01 6.97526693e-01
-1.00293076e+00 -5.38702250e-01 2.02802360e-01 2.60714531e-01
1.46418765e-01 -9.48192775e-02 9.27453876e-01 1.01175614e-01
-7.14345276e-01 -2.18369275e-01 4.71758574e-01 -3.36891294e-01
-5.67171760e-02 -1.08943737e+00 4.38994318e-01 2.52647221e-01
-5.85855916e-03 7.57217109e-01 1.04353440e+00 -3.24973166e-01
-1.69403517e+00 -2.20604807e-01 7.66078949e-01 -5.80837667e-01
9.85581577e-01 -3.98532182e-01 -7.94156551e-01 2.07698047e-01
3.30252677e-01 -1.32242396e-01 7.47881055e-01 7.65699372e-02
2.49936551e-01 1.65812135e-01 -1.03605878e+00 7.39662528e-01
1.12137938e+00 -3.90851706e-01 -7.45398641e-01 2.04080567e-01
4.42889422e-01 5.89205250e-02 -1.05862403e+00 2.23827735e-01
9.40072775e-01 -1.12046456e+00 8.81239891e-01 -8.68689477e-01
2.95457721e-01 -1.49132267e-01 2.86576867e-01 -1.10412169e+00
-2.46376127e-01 -6.07641518e-01 -1.19759731e-01 8.50367188e-01
4.89060581e-01 -1.02600920e+00 5.13693452e-01 5.92607617e-01
-1.61115378e-02 -1.21690774e+00 -6.85070336e-01 -5.49597263e-01
4.57559079e-01 -2.23552957e-01 1.40426159e-01 8.77065301e-01
6.46610260e-01 5.12871683e-01 4.02458876e-01 -5.56954205e-01
3.08312356e-01 -1.56162888e-01 4.24863726e-01 -1.58697665e+00
-3.18977796e-02 -1.04711962e+00 -8.55002582e-01 -3.03439826e-01
-2.91931182e-01 -5.42170405e-01 -1.85670242e-01 -1.45938742e+00
2.43171409e-01 2.84493733e-02 -3.09807509e-01 -1.19625650e-01
2.29962766e-01 1.70707956e-01 -2.14468330e-01 5.07536709e-01
-4.88205731e-01 1.56313032e-01 1.24999130e+00 5.72520494e-01
-3.74400377e-01 -4.88703966e-01 -9.71210241e-01 9.22432244e-01
8.75442863e-01 -3.11170459e-01 -2.21151829e-01 4.49114554e-02
4.94732589e-01 -9.59261134e-03 2.98307687e-01 -1.09981513e+00
4.82268006e-01 -3.32360655e-01 4.14075792e-01 -1.19050726e-01
3.70042145e-01 -5.54147840e-01 2.53555864e-01 8.63774836e-01
-2.53417194e-01 3.50581944e-01 8.21736827e-02 4.18010354e-01
-2.65423860e-03 -6.48911819e-02 1.00511062e+00 -4.62484539e-01
-2.50555187e-01 2.89494302e-02 -9.04163599e-01 4.43400778e-02
1.20295286e+00 -7.23896265e-01 -5.26783764e-01 -2.34289765e-01
-1.14732897e+00 6.32471815e-02 4.38364446e-01 3.98342125e-02
3.30386043e-01 -7.28721142e-01 -3.12396765e-01 -2.33218193e-01
-4.75108594e-01 -5.78924179e-01 -1.10468082e-01 1.29417956e+00
-4.61818844e-01 9.84910429e-01 -4.41914588e-01 -3.80922377e-01
-6.82871759e-01 5.74103355e-01 6.02272630e-01 4.39624749e-02
-3.39093477e-01 4.76063550e-01 1.74402401e-01 -2.22363025e-01
-7.42758140e-02 -7.54461661e-02 -5.51314175e-01 -6.65727854e-02
5.45315325e-01 4.97730821e-01 -2.67904282e-01 -4.48255539e-01
-3.61414075e-01 6.38982654e-01 9.46559533e-02 -4.08390582e-01
1.56801176e+00 -2.03004748e-01 -4.16502327e-01 1.14352190e+00
8.73153687e-01 -4.12794560e-01 -8.33230495e-01 3.06991130e-01
2.67336339e-01 1.78684428e-01 4.84057441e-02 -8.60285819e-01
-5.78312457e-01 7.74523556e-01 5.97153723e-01 9.58802938e-01
8.46864939e-01 6.88679665e-02 4.19278532e-01 2.40407303e-01
2.34923325e-02 -1.02607143e+00 -1.83409601e-01 3.65949452e-01
4.01046306e-01 -8.41004312e-01 -5.98418787e-02 -1.70897827e-01
-1.66903645e-01 1.05911171e+00 5.44595063e-01 -6.24760449e-01
1.12344003e+00 6.14491045e-01 -2.02021092e-01 -6.48629665e-01
-1.18119884e+00 -4.01372671e-01 1.89868867e-01 6.00606441e-01
9.36366498e-01 -2.63922215e-01 -1.01849878e+00 3.05884868e-01
-1.58113763e-01 2.54104316e-01 3.04255724e-01 9.50996041e-01
-9.83374000e-01 -1.13099444e+00 -4.85066593e-01 4.88484651e-01
-6.11014962e-01 -3.75621617e-02 -9.38023269e-01 8.44685733e-01
1.47868037e-01 7.11727917e-01 2.06714034e-01 -2.68409491e-01
-5.12160063e-02 5.40519617e-02 8.25585723e-01 -3.05320859e-01
-6.53509557e-01 5.36894016e-02 -2.77558893e-01 -4.52476144e-01
-9.17478085e-01 -8.89719665e-01 -1.45279312e+00 -7.73801208e-01
-1.20623969e-01 4.03744459e-01 9.16598797e-01 1.44511974e+00
3.14182967e-01 5.17659485e-01 3.17713886e-01 -8.22084367e-01
7.41900504e-02 -7.29313433e-01 -8.31081033e-01 -2.64767319e-01
2.39590257e-01 -6.61624908e-01 -3.91871721e-01 -1.07937949e-02]
|
[5.673561096191406, 4.149896621704102]
|
53dc0609-6034-44b2-9591-4d44e1a97796
|
pingan-vcgroup-s-solution-for-icdar-2021-1
|
2105.01846
| null |
https://arxiv.org/abs/2105.01846v1
|
https://arxiv.org/pdf/2105.01846v1.pdf
|
PingAn-VCGroup's Solution for ICDAR 2021 Competition on Scientific Table Image Recognition to Latex
|
This paper presents our solution for the ICDAR 2021 Competition on Scientific Table Image Recognition to LaTeX. This competition has two sub-tasks: Table Structure Reconstruction (TSR) and Table Content Reconstruction (TCR). We treat both sub-tasks as two individual image-to-sequence recognition problems. We leverage our previously proposed algorithm MASTER \cite{lu2019master}, which is originally proposed for scene text recognition. We optimize the MASTER model from several perspectives: network structure, optimizer, normalization method, pre-trained model, resolution of input image, data augmentation, and model ensemble. Our method achieves 0.7444 Exact Match and 0.8765 Exact Match @95\% on the TSR task, and obtains 0.5586 Exact Match and 0.7386 Exact Match 95\% on the TCR task.
|
['Rong Xiao', 'Xin Tang', 'Bingcong Li', 'Yihao Chen', 'Peng Gao', 'Jiaquan Ye', 'Xianbiao Qi', 'Yelin He']
|
2021-05-05
| null | null | null | null |
['scene-text-recognition']
|
['computer-vision']
|
[ 4.15151775e-01 -3.49159122e-01 -1.58280149e-01 -2.41313949e-01
-1.23145401e+00 -7.84781277e-01 6.44410908e-01 1.99875548e-01
-4.62147772e-01 5.52622318e-01 3.02936584e-01 -2.58760780e-01
2.94432223e-01 -5.53591371e-01 -1.22128081e+00 -2.04633653e-01
3.48591149e-01 5.08979499e-01 -2.08673641e-01 1.31115943e-01
4.72529858e-01 6.51492715e-01 -1.42228281e+00 9.90298271e-01
6.19150877e-01 1.43828833e+00 1.12951612e-02 1.11552691e+00
-4.65679675e-01 1.45507312e+00 -8.48435402e-01 -8.22919726e-01
1.32184967e-01 -9.46661085e-02 -1.03452444e+00 1.07457869e-01
9.18582916e-01 -1.57333776e-01 -7.79681265e-01 1.07817876e+00
5.03691375e-01 8.88035595e-02 9.33130741e-01 -9.87515986e-01
-7.98820794e-01 8.72582614e-01 -7.78036118e-01 3.26594383e-01
4.53476042e-01 -1.89160705e-01 8.90872240e-01 -1.14063156e+00
9.96728718e-01 1.27148247e+00 5.19799471e-01 4.18390065e-01
-1.19989789e+00 -6.85217679e-01 -1.05352610e-01 4.73555446e-01
-1.63486767e+00 -6.92222178e-01 2.61800200e-01 -5.43431580e-01
1.00797021e+00 3.75725895e-01 -5.92295341e-02 1.20395410e+00
1.78609997e-01 1.22130382e+00 1.05976820e+00 -2.21912101e-01
6.75768852e-02 8.60246345e-02 1.50759637e-01 7.80123174e-01
1.14374168e-01 -4.68716234e-01 -1.10942876e+00 1.79531872e-01
8.43151331e-01 -3.85340273e-01 -5.79556078e-02 -2.54771728e-02
-1.38667047e+00 3.96306902e-01 1.03572890e-01 1.72676146e-01
-1.36742190e-01 -2.03861166e-02 6.48083806e-01 2.07512707e-01
5.04064523e-02 5.51312089e-01 -5.37262738e-01 -1.30659983e-01
-9.86760914e-01 3.89925063e-01 8.76364172e-01 1.47476912e+00
3.39626908e-01 2.82730579e-01 -8.46252441e-01 1.04380214e+00
-5.91409020e-02 6.22900426e-01 1.90845773e-01 -9.56457555e-01
1.06048298e+00 5.34484148e-01 -1.63244322e-01 -7.98869014e-01
-2.66313553e-01 -6.49395764e-01 -1.51365936e+00 -3.95506740e-01
3.72124344e-01 3.37969750e-01 -8.58781219e-01 1.28002596e+00
-8.33227783e-02 -1.61732748e-01 8.01320598e-02 3.48596215e-01
1.33405983e+00 7.40609229e-01 -1.62266731e-01 -4.92032692e-02
1.42867291e+00 -1.17551875e+00 -7.87947476e-01 -1.43016547e-01
8.09248388e-01 -1.20923150e+00 1.20312583e+00 7.05141783e-01
-1.39310396e+00 -6.69162989e-01 -1.11272478e+00 -3.70831579e-01
-4.51231122e-01 8.23994935e-01 2.03372583e-01 3.86014223e-01
-8.45006943e-01 5.50449193e-01 -2.07560673e-01 -2.15094015e-02
6.30936384e-01 3.15113813e-01 -4.09258664e-01 -2.27998227e-01
-7.17154145e-01 6.08026624e-01 2.03012586e-01 -8.15344304e-02
-6.74014330e-01 -7.70600259e-01 -6.46534681e-01 1.87213138e-01
6.11210942e-01 -5.75307190e-01 1.18462300e+00 -3.67536575e-01
-1.34268212e+00 1.47429347e+00 -2.88176209e-01 -6.74388170e-01
4.30649459e-01 -2.74537414e-01 -4.73416120e-01 1.05298543e-02
2.89389845e-02 4.71669257e-01 5.38420320e-01 -1.04325676e+00
-5.51447213e-01 -4.92089063e-01 -6.05526567e-01 6.81808665e-02
-4.18726318e-02 1.35312736e-01 -1.13570547e+00 -1.08789206e+00
-9.15620178e-02 -6.19455457e-01 1.65412471e-01 -2.96811312e-01
-7.27739215e-01 -1.89867839e-01 1.66632965e-01 -1.04760623e+00
1.25634086e+00 -1.90196323e+00 1.24401391e-01 1.36847630e-01
4.10072029e-01 1.47352636e-01 -3.81982803e-01 1.95038289e-01
-1.48906127e-01 1.68703243e-01 -1.87412590e-01 -5.58459640e-01
1.72373280e-01 -2.28245601e-01 -6.12098813e-01 3.85902405e-01
-1.06264949e-01 9.57277894e-01 -1.31984502e-01 -5.84612787e-01
2.12385133e-02 2.50437766e-01 -4.48430568e-01 4.25638020e-01
-1.66905090e-01 6.96179122e-02 -1.10101953e-01 9.21046317e-01
9.11651433e-01 -5.34803569e-01 4.36581001e-02 -4.24330771e-01
1.80989690e-02 1.87895119e-01 -1.42502582e+00 1.74673641e+00
-2.46372581e-01 7.21937597e-01 1.82353213e-01 -8.62565219e-01
1.05101466e+00 3.91085483e-02 4.94984597e-01 -1.05559170e+00
2.17252627e-01 1.08128779e-01 -4.60655510e-01 -2.97663182e-01
8.09310913e-01 5.82342684e-01 -8.03662166e-02 1.91594794e-01
9.20304283e-02 9.40427259e-02 4.69259083e-01 4.38971370e-01
1.31070960e+00 2.67842561e-01 6.78296238e-02 -3.85391414e-01
8.84127438e-01 -1.08345024e-01 2.08144546e-01 1.12798584e+00
1.02428950e-01 8.98178995e-01 4.87425297e-01 -4.86902058e-01
-1.32116699e+00 -1.01990592e+00 -1.98886290e-01 1.18758643e+00
-8.46234336e-03 -7.68190145e-01 -9.72197950e-01 -6.44656420e-01
-3.84011748e-03 6.98431432e-01 -6.13165081e-01 -6.28126636e-02
-8.58841181e-01 -6.53428257e-01 9.13648844e-01 5.94983041e-01
6.86219513e-01 -1.18298483e+00 8.22759122e-02 3.41586885e-04
-3.74384731e-01 -1.67326057e+00 -7.57286668e-01 2.34461486e-01
-6.75064027e-01 -8.14499140e-01 -5.70272624e-01 -7.46673405e-01
2.77940035e-01 8.41861870e-03 1.75637102e+00 -3.02206669e-02
-3.74804705e-01 2.78599322e-01 3.11676040e-03 -2.94576108e-01
-5.03814101e-01 3.37667108e-01 -1.41224280e-01 -4.60959524e-02
1.74780667e-01 -2.40551397e-01 -1.31903797e-01 2.43151337e-02
-9.00783896e-01 1.33376926e-01 8.04045558e-01 7.68686414e-01
1.01644397e+00 -2.10786700e-01 1.55064568e-01 -1.05705178e+00
4.01972175e-01 2.08960511e-02 -9.13363695e-01 7.05772102e-01
-7.30017781e-01 2.85409421e-01 7.66773760e-01 -1.96728081e-01
-9.32353258e-01 3.45104009e-01 -3.12490016e-01 -6.08818293e-01
-2.65634060e-01 2.37192705e-01 -4.90269274e-01 1.71058968e-01
5.61682999e-01 8.16751063e-01 -3.60559434e-01 -8.73783290e-01
3.31480682e-01 7.70563126e-01 1.03593290e+00 -6.78639710e-01
7.82580793e-01 3.61586332e-01 1.62563827e-02 -6.87904954e-01
-1.10231614e+00 -5.01751542e-01 -7.88869381e-01 1.02427475e-01
8.77085984e-01 -1.04423296e+00 -1.43016422e+00 4.75584775e-01
-1.23896384e+00 -2.58670777e-01 -1.15651004e-01 -1.47101087e-02
-7.82315910e-01 3.33821386e-01 -6.93537414e-01 -5.13202786e-01
-7.90544808e-01 -1.12644088e+00 1.36410296e+00 -1.78598911e-01
-4.43037711e-02 -6.01328194e-01 -2.68770307e-01 8.31070840e-01
1.76809058e-01 3.25457193e-02 9.57125843e-01 -6.71401680e-01
-9.55115616e-01 -1.33571625e-01 -5.89404345e-01 2.13340715e-01
-3.59438390e-01 -8.31113011e-02 -9.72563624e-01 -1.31107092e-01
-1.53301552e-01 -4.95010257e-01 1.17936540e+00 2.97667027e-01
1.96370232e+00 -4.45851833e-01 4.36017476e-02 1.03302574e+00
1.35928261e+00 3.07997555e-01 1.02594757e+00 4.81613606e-01
9.89386201e-01 4.21555191e-01 3.17500323e-01 4.19309258e-01
3.35710406e-01 8.18864048e-01 1.34116575e-01 -3.17075737e-02
-3.73239219e-01 -2.32216567e-01 3.00482452e-01 7.48425424e-01
2.58360267e-01 -6.32208109e-01 -1.06529617e+00 2.43031997e-02
-1.68748665e+00 -8.89989674e-01 -2.21690536e-01 2.20367146e+00
8.31115901e-01 2.47135714e-01 7.22834170e-02 2.25905880e-01
6.18749857e-01 6.27285168e-02 -3.02969694e-01 -4.83977824e-01
-7.14565814e-01 1.72163203e-01 6.34038210e-01 2.72277594e-01
-1.18601549e+00 1.08650577e+00 6.16192722e+00 1.44809794e+00
-7.00639784e-01 -2.70682633e-01 1.11733782e+00 1.84342209e-02
4.44272421e-02 -5.91305852e-01 -1.23286009e+00 2.08201095e-01
1.07205153e+00 2.98412226e-04 8.08111072e-01 7.29928255e-01
-3.88761789e-01 1.50985733e-01 -1.23135555e+00 1.67580748e+00
4.29761201e-01 -2.06453633e+00 6.43993616e-01 -9.48285088e-02
5.55837691e-01 -2.15058103e-01 1.66021585e-01 4.80206758e-01
1.60774633e-01 -1.61741042e+00 9.09356356e-01 6.38821006e-01
1.15549541e+00 -7.76791334e-01 6.30300522e-01 1.49847165e-01
-1.25610173e+00 3.18215758e-01 -3.39525282e-01 4.10448849e-01
-3.97832990e-01 3.84429306e-01 -6.77613974e-01 5.37675798e-01
8.55098188e-01 7.95341969e-01 -9.33387280e-01 6.16873085e-01
2.02139601e-01 5.83735943e-01 -5.69649413e-02 1.75724700e-02
-1.17552191e-01 -7.36771151e-02 4.17498946e-01 1.56412971e+00
-7.32560605e-02 -2.33405128e-01 -1.48941241e-02 7.32478917e-01
-9.59226549e-01 2.91862220e-01 -4.78611201e-01 -3.22464108e-02
4.49111044e-01 1.10691202e+00 -6.72997177e-01 -7.90886223e-01
-7.08647966e-02 1.14125931e+00 4.14337695e-01 2.75309056e-01
-7.23125219e-01 -3.51473540e-01 3.34402174e-01 -8.73812959e-02
3.85243654e-01 8.25927779e-02 -1.05434036e+00 -1.53397441e+00
1.26239091e-01 -1.46256101e+00 5.69929361e-01 -8.66824746e-01
-1.13249397e+00 7.62242794e-01 -4.94361877e-01 -1.03915560e+00
1.61187444e-02 -8.19955766e-01 5.75214252e-02 7.98466384e-01
-1.00092661e+00 -9.94333804e-01 -1.38257384e-01 5.77219009e-01
6.64525330e-01 -6.40749454e-01 7.39583731e-01 6.00106955e-01
-9.41877484e-01 1.20305979e+00 5.74851811e-01 4.01604503e-01
7.24512577e-01 -1.35468686e+00 5.02424479e-01 7.40991235e-01
4.19843882e-01 4.55599070e-01 5.03784657e-01 -5.96240819e-01
-1.85193706e+00 -1.25810039e+00 1.05076051e+00 -8.04306448e-01
3.98614764e-01 -7.86219001e-01 -1.01417696e+00 5.24689615e-01
1.02168672e-01 -1.96756691e-01 3.59226435e-01 3.03531606e-02
-7.87429452e-01 -3.21036488e-01 -9.47261512e-01 7.16051579e-01
9.94644225e-01 -6.49465382e-01 -4.83867340e-02 5.07240772e-01
5.00832260e-01 -8.07701588e-01 -1.22093153e+00 3.68418425e-01
4.76196408e-01 -6.46402121e-01 1.39245927e+00 -5.77954590e-01
8.75020325e-01 -3.64092588e-02 -4.20931906e-01 -3.19720119e-01
-4.12711710e-01 -6.38754606e-01 -3.50683481e-01 1.30065155e+00
4.92733210e-01 2.49409005e-01 1.06260264e+00 1.29080713e-01
-1.46884590e-01 -7.42255688e-01 -8.91836405e-01 -9.25856411e-01
2.53889471e-01 -4.91359323e-01 6.23618960e-01 9.46294248e-01
-5.96512914e-01 8.44369531e-01 -5.77912390e-01 -8.11189711e-02
8.26090872e-01 2.71857679e-01 8.95662546e-01 -9.01005447e-01
-1.35547787e-01 -7.49214590e-01 1.37389228e-01 -1.12391293e+00
2.57088430e-02 -1.13907707e+00 -1.67862445e-01 -1.54668784e+00
6.11297846e-01 1.83687016e-01 -1.31532952e-01 2.89786905e-01
-5.98372854e-02 2.71854073e-01 3.72402161e-01 4.15979654e-01
-9.80115950e-01 4.91796136e-01 1.23637652e+00 -3.26345325e-01
1.32333875e-01 -1.09741122e-01 -5.37749350e-01 4.34141248e-01
7.54479468e-01 -5.30511498e-01 1.48246318e-01 -5.35662293e-01
1.36884227e-01 2.98435420e-01 8.26382637e-02 -9.30783451e-01
2.43326172e-01 -3.74208353e-02 8.76619279e-01 -1.21620834e+00
2.03175798e-01 -3.92761022e-01 -4.71000113e-02 3.60444933e-01
-8.30304325e-01 3.50571394e-01 4.46284145e-01 3.64086807e-01
-1.25076488e-01 -2.35212278e-02 8.94980252e-01 -2.12085936e-02
-4.27977771e-01 3.21347415e-01 -4.76757199e-01 3.61288846e-01
2.18359187e-01 -2.12224677e-01 -5.66565156e-01 -2.47599021e-01
-5.32293797e-01 6.15079887e-02 1.54756889e-01 4.90863115e-01
6.49172544e-01 -1.22905874e+00 -9.11178172e-01 1.86197963e-02
1.51134133e-01 -1.75258264e-01 1.84071600e-01 4.66367662e-01
-5.48062265e-01 8.70042145e-01 -7.78086409e-02 -4.77155894e-01
-1.44757974e+00 6.53132558e-01 1.68891951e-01 -8.35654795e-01
-4.88493085e-01 9.65830207e-01 2.40487710e-01 -6.12102628e-01
7.99341381e-01 -2.85131279e-02 -2.79145241e-01 -1.11888841e-01
5.41724741e-01 5.09541750e-01 5.10161579e-01 -4.49467719e-01
-4.29928869e-01 5.38378894e-01 -4.17181015e-01 1.04717813e-01
1.24046767e+00 1.02458552e-01 -1.69548944e-01 3.13973010e-01
1.42338896e+00 2.59230956e-02 -7.30968177e-01 -5.20734966e-01
1.42862335e-01 -1.59932002e-01 -4.86833462e-03 -1.21947122e+00
-9.01079476e-01 9.85798776e-01 5.85101187e-01 -3.21073472e-01
1.04971027e+00 -2.07356527e-01 7.55662680e-01 8.37572396e-01
-1.15208253e-01 -1.23580432e+00 3.54904830e-01 9.20305192e-01
1.10584235e+00 -1.00403082e+00 1.51228234e-01 -4.85485762e-01
-7.55391717e-01 1.13275266e+00 6.42606795e-01 1.85824841e-01
1.87522292e-01 6.56740248e-01 -2.08163083e-01 -1.30217865e-01
-9.32542384e-01 5.05726365e-03 8.75333190e-01 2.48942733e-01
6.33802772e-01 -1.46456778e-01 2.47163758e-01 7.31026649e-01
-4.99095351e-01 -1.15070492e-02 4.94377971e-01 9.27904963e-01
-1.23856246e-01 -8.29382837e-01 -4.78737175e-01 7.41761327e-01
-6.87333465e-01 -4.63684618e-01 -7.33428419e-01 4.99165803e-01
-1.52924716e-01 5.42115211e-01 1.55200914e-01 -6.55315936e-01
5.75240493e-01 2.20599491e-02 6.32533669e-01 -3.30113590e-01
-9.55109954e-01 2.48886779e-01 1.58180222e-01 -5.69509923e-01
2.42521942e-01 -5.52786291e-01 -1.01540387e+00 -7.39420116e-01
1.06983773e-01 -5.80141656e-02 6.25719190e-01 6.68859661e-01
5.25331140e-01 7.91364372e-01 2.75327593e-01 -4.85726409e-02
-6.08421624e-01 -9.31217015e-01 -4.12363321e-01 2.82262951e-01
1.02583602e-01 -1.34355947e-01 -7.67243281e-02 4.20273900e-01]
|
[11.692536354064941, 2.9895665645599365]
|
d9fc9c48-33ea-42ec-a5b9-198579cb82ae
|
a-heat-map-based-algorithm-for-recognizing
|
1502.06076
| null |
http://arxiv.org/abs/1502.06076v1
|
http://arxiv.org/pdf/1502.06076v1.pdf
|
A Heat-Map-based Algorithm for Recognizing Group Activities in Videos
|
In this paper, a new heat-map-based (HMB) algorithm is proposed for group
activity recognition. The proposed algorithm first models human trajectories as
series of "heat sources" and then applies a thermal diffusion process to create
a heat map (HM) for representing the group activities. Based on this heat map,
a new key-point based (KPB) method is used for handling the alignments among
heat maps with different scales and rotations. And a surface-fitting (SF)
method is also proposed for recognizing group activities. Our proposed HM
feature can efficiently embed the temporal motion information of the group
activities while the proposed KPB and SF methods can effectively utilize the
characteristics of the heat map for activity recognition. Experimental results
demonstrate the effectiveness of our proposed algorithms.
|
['Zhenzhong Chen', 'Bin Sheng', 'Weiyao Lin', 'Jianxin Wu', 'Hang Chu']
|
2015-02-21
| null | null | null | null |
['group-activity-recognition']
|
['computer-vision']
|
[ 1.91298261e-01 -5.84940910e-01 -2.94591606e-01 -5.24787158e-02
-3.95114273e-01 -2.35323414e-01 7.61114776e-01 4.54548784e-02
-7.20270351e-02 2.28503615e-01 3.23452771e-01 -2.22353600e-02
-1.23658098e-01 -8.30862105e-01 -2.77430773e-01 -9.55122054e-01
-1.22869231e-01 -5.34828939e-02 5.56413710e-01 1.13259949e-01
5.83341360e-01 5.20882964e-01 -1.14141822e+00 -5.21534635e-03
9.27688301e-01 8.57042074e-01 1.28670722e-01 7.30476320e-01
-1.54004455e-01 8.90019655e-01 -4.57816005e-01 3.12061846e-01
1.81289569e-01 -6.96498334e-01 -6.08817399e-01 3.03777337e-01
-1.94806099e-01 -2.08165050e-01 -5.00070751e-01 5.78249633e-01
1.21394202e-01 6.63600683e-01 8.70082557e-01 -1.47665167e+00
-3.23442876e-01 -2.93131113e-01 -7.45641708e-01 2.07018718e-01
8.34881127e-01 -7.63520151e-02 1.48844585e-01 -1.00148213e+00
5.66151798e-01 1.08129811e+00 6.02484286e-01 1.84204355e-01
-7.15257704e-01 -5.34832001e-01 4.47796248e-02 6.45385027e-01
-1.57485271e+00 -1.05067089e-01 1.06652796e+00 -5.19887686e-01
8.84857953e-01 5.53955555e-01 1.27481282e+00 6.28500581e-01
3.70172441e-01 1.02941239e+00 1.00913632e+00 -4.04189527e-01
4.61241841e-01 -3.97268027e-01 1.79800853e-01 7.37773478e-01
2.38263924e-02 -3.78530055e-01 -5.48636258e-01 -4.95365649e-01
1.06586349e+00 6.70098305e-01 -3.08495402e-01 -4.28231478e-01
-1.55288899e+00 2.88277596e-01 3.08617771e-01 4.28347141e-01
-3.56819779e-01 6.09975494e-02 1.61251888e-01 -1.24767609e-01
4.34390366e-01 -1.34317763e-02 2.17873752e-01 -6.39193475e-01
-1.11172473e+00 1.98140353e-01 6.20416641e-01 9.81307685e-01
1.04527664e+00 -2.06801683e-01 -2.50566155e-01 6.37811959e-01
4.53016698e-01 7.33382285e-01 5.97078145e-01 -7.80023634e-01
3.74607146e-01 1.11184680e+00 3.94297451e-01 -1.62697542e+00
-3.72530550e-01 3.93512398e-01 -1.13090777e+00 -2.88282812e-01
2.96017021e-01 2.11298391e-01 -8.80359650e-01 9.71867204e-01
5.82878530e-01 7.89998531e-01 -2.20620826e-01 7.75548637e-01
1.71731159e-01 1.28085089e+00 -1.10766403e-01 -4.44193870e-01
1.10751033e+00 -1.06946194e+00 -9.38157380e-01 8.04125965e-02
6.11347497e-01 -2.75843501e-01 7.67624080e-01 1.84590686e-02
-8.19773495e-01 -6.24652326e-01 -1.15908873e+00 1.12404458e-01
-3.53270143e-01 2.24153157e-02 3.29358935e-01 4.01777506e-01
-9.68561947e-01 4.00571316e-01 -1.41218686e+00 -6.11507595e-01
1.64383635e-01 3.59416269e-02 -1.84008181e-01 6.49994537e-02
-7.59862959e-01 5.37700236e-01 2.23314941e-01 3.05093765e-01
-6.89182401e-01 -2.60098606e-01 -8.53945732e-01 -2.07816884e-01
9.87120569e-02 -2.35242456e-01 6.34261072e-01 -7.10749984e-01
-1.44666994e+00 2.75413305e-01 -7.73658693e-01 -3.57484557e-02
5.84558785e-01 -6.16681576e-02 -6.48186982e-01 4.56189275e-01
-2.16720060e-01 1.55400604e-01 1.02025378e+00 -9.13526952e-01
-5.37077427e-01 -3.02521884e-01 -6.68209732e-01 3.72372001e-01
-3.59275639e-01 7.53378347e-02 -5.79592347e-01 -6.85128629e-01
5.33434987e-01 -1.14197743e+00 -1.89176068e-01 -1.94841530e-02
-7.58028552e-02 -1.85067460e-01 1.22888160e+00 -8.61696124e-01
1.84053361e+00 -2.15382051e+00 2.58941293e-01 7.30558038e-01
5.74345998e-02 5.79931550e-02 3.72537017e-01 7.33558595e-01
3.02523494e-01 -1.46890804e-01 -6.19603276e-01 -9.95283276e-02
-3.08956653e-01 2.08601475e-01 -5.74695282e-02 6.39982343e-01
-6.94389567e-02 7.82787025e-01 -1.03462255e+00 -6.93066239e-01
4.67822999e-01 5.27546942e-01 -8.33889376e-03 3.28237265e-01
4.70273703e-01 5.33539891e-01 -4.86750841e-01 5.95768571e-01
8.57251883e-01 -4.49224077e-02 1.23285770e-01 5.98679073e-02
-2.06063926e-01 -2.38135755e-01 -1.36156452e+00 1.62805080e+00
4.33575399e-02 4.88776326e-01 -3.78398716e-01 -6.58207834e-01
1.32722425e+00 1.94953501e-01 9.40087974e-01 -3.75351250e-01
-1.17045946e-01 3.46648656e-02 -5.03241897e-01 -4.55643922e-01
5.44048846e-01 4.09320056e-01 -3.52816656e-02 6.48827076e-01
-2.96079904e-01 3.03974777e-01 -1.61836687e-02 1.24369815e-01
1.27658653e+00 2.82811463e-01 4.16823298e-01 -2.76533276e-01
8.61340523e-01 -1.10886255e-02 7.76517630e-01 2.94022292e-01
-6.62106276e-01 5.00085771e-01 5.22830486e-02 -6.47342145e-01
-1.08339942e+00 -8.72611582e-01 3.03190082e-01 7.53594697e-01
6.04125857e-01 -6.76005542e-01 -1.01968241e+00 -6.06664419e-01
-2.48061836e-01 1.22788973e-01 -6.71863139e-01 -2.30946988e-01
-9.08407211e-01 -6.90073192e-01 3.38702530e-01 7.67580748e-01
1.06304753e+00 -9.84120250e-01 -6.79169953e-01 3.05897027e-01
-3.18544656e-01 -6.25174224e-01 -7.35397458e-01 -6.30995691e-01
-1.05514288e+00 -1.02361023e+00 -1.01770139e+00 -8.64499807e-01
1.02896237e+00 8.27006400e-01 2.13393971e-01 1.18985265e-01
-7.99308792e-02 4.76854235e-01 -5.63856006e-01 2.93515306e-02
-2.07621217e-01 -2.29931623e-01 1.44291781e-02 6.33359492e-01
4.93954062e-01 -5.12549877e-01 -1.03196728e+00 7.19072223e-01
-8.15764368e-01 2.32138991e-01 4.68858540e-01 4.30375069e-01
6.07111156e-01 1.91118587e-02 2.20498368e-01 -2.06442088e-01
6.54038548e-01 -4.45070088e-01 -2.98816502e-01 5.40268660e-01
-5.56976974e-01 -1.85374931e-01 2.99659640e-01 -6.18583262e-01
-1.18099105e+00 1.21953167e-01 4.19888943e-01 -4.66596931e-01
-1.20038189e-01 3.12199414e-01 -1.10753290e-01 -3.04754525e-01
2.96375215e-01 6.69071794e-01 5.44139631e-02 -4.27426010e-01
2.72094399e-01 7.38125563e-01 3.97647649e-01 -3.73245507e-01
8.11738729e-01 8.09650302e-01 1.54185653e-01 -1.04560900e+00
1.14389651e-01 -9.85179067e-01 -9.42634940e-01 -6.84790254e-01
1.10693669e+00 -5.64133763e-01 -7.07728803e-01 1.09518969e+00
-7.99720347e-01 -1.40976131e-01 9.50443298e-02 4.85823452e-01
-6.88169539e-01 8.37315023e-01 -6.75566554e-01 -1.10890687e+00
-2.77929246e-01 -8.12231898e-01 9.68729019e-01 4.10408854e-01
-1.61999494e-01 -1.10258985e+00 5.37935615e-01 1.33028761e-01
1.92090735e-01 4.70173687e-01 4.13211316e-01 -1.72134146e-01
-6.50246382e-01 -3.20529610e-01 1.20590487e-02 -1.29945472e-01
4.62921053e-01 3.37115079e-01 -5.52045166e-01 -2.49494299e-01
8.48900452e-02 4.22006726e-01 4.58917618e-01 2.95210898e-01
8.75665188e-01 -4.07693863e-01 -7.09991574e-01 6.98565364e-01
1.18495560e+00 8.18218291e-01 9.67895925e-01 4.45707738e-01
1.09741354e+00 3.88235897e-01 9.30716217e-01 6.01095796e-01
3.72520000e-01 7.55942464e-01 -2.68469781e-01 -1.10930637e-01
3.89035881e-01 -4.77146953e-01 5.92468083e-01 1.24389231e+00
-6.04616940e-01 1.33237526e-01 -1.12951291e+00 3.16123277e-01
-2.46470952e+00 -1.21275365e+00 -3.38122815e-01 2.17553902e+00
3.73256505e-01 -1.79596081e-01 4.18002158e-01 4.12988275e-01
8.34119737e-01 3.77622843e-01 -4.08382863e-01 -2.29120642e-01
2.72125006e-01 -2.53218383e-01 4.43034261e-01 3.11372876e-01
-1.10602629e+00 5.51672280e-01 6.60162020e+00 8.78413141e-01
-8.26646686e-01 -6.76807165e-02 1.39838263e-01 3.94635618e-01
5.17543964e-02 6.34891912e-02 -4.58687186e-01 7.13877738e-01
5.80824316e-01 -3.32005799e-01 3.03547680e-01 7.13276625e-01
5.59096873e-01 -4.29553121e-01 -4.87510920e-01 1.13458848e+00
5.95801212e-02 -9.07403946e-01 -4.80356999e-02 2.98477877e-02
6.83032036e-01 -6.59722388e-01 -1.18355043e-01 -5.11607602e-02
-6.44888580e-02 -4.39583898e-01 4.57265407e-01 1.05995512e+00
5.71944773e-01 -7.46807396e-01 3.34120810e-01 4.78301257e-01
-1.93462288e+00 -8.63720402e-02 -2.91697502e-01 -4.72443998e-02
3.33777040e-01 2.44684055e-01 -9.38554347e-01 6.91752493e-01
6.29288495e-01 1.11896408e+00 -7.64127970e-01 1.06337678e+00
-1.47915594e-02 6.57873511e-01 -3.07484478e-01 -3.56931269e-01
1.98495761e-01 -8.30076158e-01 5.24689496e-01 1.41678405e+00
6.86124682e-01 2.12520376e-01 1.67916656e-01 5.81259966e-01
6.34856284e-01 2.65038311e-01 -4.81309205e-01 -1.41393304e-01
4.59420472e-01 1.18482554e+00 -9.77191389e-01 -7.09334731e-01
-3.11383039e-01 1.23331165e+00 -7.21768811e-02 4.89813477e-01
-8.35156143e-01 -5.26802897e-01 3.44675332e-01 2.62296408e-01
-9.15078968e-02 -8.39699924e-01 2.63668522e-02 -1.31989670e+00
6.11694567e-02 -2.91765720e-01 4.67932582e-01 -9.48689520e-01
-7.41346061e-01 7.99584109e-03 3.45299989e-01 -1.63928902e+00
-1.70295894e-01 -1.02049947e-01 -8.20485175e-01 7.09333777e-01
-8.34163427e-01 -1.17113948e+00 -7.06157386e-01 9.02620494e-01
4.94355440e-01 1.97679192e-01 5.97656012e-01 -2.78016459e-02
-7.25602686e-01 1.11462940e-02 3.25225621e-01 2.64386177e-01
4.16749120e-01 -1.20625591e+00 5.53696871e-01 9.88496065e-01
-8.97880867e-02 7.90302813e-01 3.18887144e-01 -1.01804864e+00
-1.45835805e+00 -9.03853118e-01 6.10348761e-01 -2.31820628e-01
4.27984923e-01 -3.66984844e-01 -1.19913232e+00 6.74708486e-01
-4.47352305e-02 -2.47314841e-01 7.43539691e-01 -4.03320104e-01
2.20914725e-02 -1.31772175e-01 -9.59906995e-01 5.23820221e-01
1.11346161e+00 -5.36465943e-01 -6.63764298e-01 1.84865799e-02
3.30782264e-01 -1.80720285e-01 -9.85655367e-01 3.29113334e-01
7.15562224e-01 -7.53353655e-01 9.24355388e-01 -3.25238816e-02
-2.77768731e-01 -7.82414913e-01 2.93109298e-01 -1.20719886e+00
-7.18401492e-01 -7.51420200e-01 -4.31520611e-01 9.89236653e-01
-1.88286006e-01 -5.01320839e-01 7.75698841e-01 5.84013939e-01
7.47287869e-02 -5.90175927e-01 -7.67768085e-01 -8.11976254e-01
-7.76877940e-01 -1.54749155e-01 6.44795001e-01 1.06688249e+00
5.97234011e-01 -4.41657186e-01 -4.07483160e-01 -1.74215332e-01
5.06387115e-01 -1.17441364e-01 9.02556598e-01 -8.18826854e-01
9.59508047e-02 -1.98322907e-01 -8.50066543e-01 -1.06819105e+00
-4.39950168e-01 -6.18659735e-01 4.92776334e-02 -1.68271363e+00
2.61865705e-01 -1.99453950e-01 -4.15418565e-01 2.04274848e-01
-2.80054450e-01 -1.35151505e-01 2.14215845e-01 8.20169032e-01
-7.07415700e-01 5.77284038e-01 1.14612424e+00 7.45799914e-02
-6.34312928e-01 5.90113923e-03 1.50185317e-01 6.07062280e-01
7.50588655e-01 -1.24333262e-01 -4.00343359e-01 1.94931298e-01
-3.65461595e-02 1.18355609e-01 1.33295670e-01 -1.41036665e+00
4.84398574e-01 -4.57309157e-01 4.65568900e-01 -8.88246715e-01
2.85543352e-01 -8.27389121e-01 5.17665327e-01 6.13529563e-01
4.63091629e-03 2.76778787e-01 -1.23282745e-01 9.57990110e-01
-2.63478279e-01 2.97046155e-01 4.47629094e-01 1.75891846e-01
-7.85911679e-01 4.15472716e-01 -6.38990939e-01 -6.47889197e-01
1.44380856e+00 -7.37573147e-01 -2.09422678e-01 -4.40636724e-01
-5.94012141e-01 2.01973826e-01 6.78109229e-01 2.56915450e-01
9.83593166e-01 -1.87917328e+00 -1.39020666e-01 4.30138975e-01
2.84405977e-01 -3.02690193e-02 4.53154624e-01 1.16596866e+00
-9.62694228e-01 1.24254040e-01 -3.45501810e-01 -6.88899457e-01
-1.26913357e+00 6.35144651e-01 1.73677415e-01 -2.63255447e-01
-6.56463802e-01 3.16391945e-01 -5.06923981e-02 4.38852981e-02
-1.02703981e-01 -4.96402562e-01 -3.14453185e-01 -6.01180829e-02
7.82242596e-01 1.14066374e+00 -2.86040038e-01 -9.60265338e-01
-3.60742211e-01 9.77293730e-01 3.28432411e-01 -4.39592510e-01
9.65674102e-01 -2.53654242e-01 -8.13531354e-02 6.76338434e-01
1.20326614e+00 -1.88361481e-01 -1.46296108e+00 2.67746989e-02
2.44369462e-01 -7.04598010e-01 -4.98048306e-01 -2.05996424e-01
-7.19011009e-01 9.24843073e-01 4.92170662e-01 2.20007673e-01
1.41406333e+00 -4.69281495e-01 9.51162338e-01 5.80462776e-02
4.35722858e-01 -1.29898357e+00 1.96456596e-01 1.57578126e-01
5.34620881e-01 -6.28439546e-01 -2.04613097e-02 -4.74511713e-01
-6.72934771e-01 1.18345690e+00 3.55795443e-01 -2.31983945e-01
6.52580976e-01 1.46817220e-02 -1.58993363e-01 -8.47165659e-02
-2.56620497e-01 2.65752207e-02 3.52027416e-01 6.96879625e-01
-1.25695735e-01 1.67674288e-01 -5.23659408e-01 1.95792437e-01
2.77513862e-01 9.52735394e-02 3.23397756e-01 1.49217963e+00
-8.29929650e-01 -8.11518490e-01 -7.24165082e-01 1.12070762e-01
1.84214726e-01 5.86825907e-01 -5.44653296e-01 4.69169378e-01
-9.83810201e-02 9.34075356e-01 1.87957585e-01 -8.43416095e-01
2.12482274e-01 2.92474270e-01 4.34817612e-01 -5.75323664e-02
-2.08846971e-01 2.29685724e-01 -4.01519716e-01 -8.40151668e-01
-6.83959186e-01 -7.94092894e-01 -1.34656179e+00 -3.19106132e-01
-2.37747300e-02 6.54453635e-02 5.89328408e-01 7.51137912e-01
4.64349478e-01 -6.43632486e-02 9.54623401e-01 -8.44496787e-01
2.26840094e-01 -9.52492595e-01 -6.71367168e-01 8.29607189e-01
1.33389952e-02 -6.10989034e-01 -3.63430530e-01 2.66042888e-01]
|
[8.158425331115723, 0.37691929936408997]
|
049ac831-0f1a-4690-8d02-70625453f600
|
a-cognitive-account-of-the-puzzle-of
|
2305.00296
| null |
https://arxiv.org/abs/2305.00296v1
|
https://arxiv.org/pdf/2305.00296v1.pdf
|
A Cognitive Account of the Puzzle of Ideography
|
In this commentary article to 'The Puzzle of Ideography' by Morin, we put forth a new cognitive account of the puzzle of ideography, that complements the standardization account of Morin. Efficient standardization of spoken language is phenomenologically attributed to a modality effect coupled with chunking of cognitive representations, further aided by multi-sensory integration and the serialized nature of attention. These cognitive mechanisms are crucial for explaining why languages dominate graphic codes for general-purpose human communication.
|
['Xerxes D. Arsiwalla']
|
2023-04-29
| null | null | null | null |
['chunking']
|
['natural-language-processing']
|
[-8.81133154e-02 1.95659459e-01 -2.15823457e-01 4.96718325e-02
-3.84148918e-02 -7.52678216e-01 8.36692989e-01 -5.52031994e-02
-3.32344770e-01 2.65017897e-01 1.12258065e+00 -5.36788762e-01
-5.51898301e-01 -6.63147867e-02 -3.06989461e-01 -1.88304767e-01
3.33999991e-01 4.14134830e-01 -1.67218804e-01 -7.32278049e-01
9.69730377e-01 2.07575094e-02 -1.51266825e+00 5.50123632e-01
7.18362272e-01 5.02346218e-01 5.30545652e-01 5.39464653e-01
-4.12872076e-01 1.12505102e+00 -5.10109961e-01 -2.25158170e-01
-2.91153967e-01 -8.05487096e-01 -7.11166859e-01 1.07757114e-01
-1.64567530e-01 1.64023191e-01 -1.63866550e-01 9.88065898e-01
3.56489331e-01 -9.79251042e-02 6.09255075e-01 -5.21751881e-01
-1.25943756e+00 8.05991471e-01 -1.36184692e-01 5.70206702e-01
6.90712631e-01 -8.53940845e-02 7.27983475e-01 -9.60130215e-01
1.00353301e+00 1.44911540e+00 7.66946793e-01 5.55645704e-01
-1.00176978e+00 -3.39607090e-01 -7.73025816e-03 3.03239673e-01
-1.40520418e+00 -8.15701187e-01 5.44385433e-01 -8.68080378e-01
9.31594849e-01 2.98065543e-01 1.43775856e+00 1.02159750e+00
6.33123696e-01 2.43791267e-01 1.32123148e+00 -6.28862083e-01
-5.72204590e-04 6.58460334e-02 1.65572554e-01 4.11612421e-01
3.61240864e-01 1.20571695e-01 -1.24782562e+00 -2.03994866e-02
8.30949962e-01 -2.56478935e-01 -2.61383921e-01 2.18733907e-01
-1.46768606e+00 5.49749792e-01 -1.14756942e-01 7.25517452e-01
-6.55881405e-01 7.99288079e-02 5.61285317e-01 4.63476092e-01
1.21711738e-01 6.86712563e-01 -1.01623498e-01 -5.24666548e-01
-7.23997355e-01 7.21329451e-02 6.13155842e-01 3.35943073e-01
-1.60845116e-01 7.45677873e-02 3.50664735e-01 5.65490901e-01
7.47290432e-01 9.65705216e-01 4.65355486e-01 -1.01048625e+00
3.45801443e-01 3.70067447e-01 -2.33995035e-01 -9.37423527e-01
-5.94620049e-01 -4.17544961e-01 -4.10661072e-01 2.97032803e-01
5.16807973e-01 -1.64661035e-02 -1.76902637e-01 1.73548400e+00
-2.69700527e-01 -9.51468170e-01 2.31320217e-01 1.18540347e+00
5.04513204e-01 5.26564717e-01 4.90552157e-01 -2.97062159e-01
1.63488162e+00 -3.86901617e-01 -1.07544982e+00 -1.42387703e-01
3.79343122e-01 -9.90528405e-01 1.18214548e+00 4.59731668e-01
-1.33443892e+00 -1.48359641e-01 -1.20346534e+00 -3.85211378e-01
-1.58366382e-01 -1.66440025e-01 7.30175018e-01 6.79504573e-01
-1.08644426e+00 3.66007000e-01 -4.16765213e-01 -6.99755728e-01
3.96851420e-01 -2.42684106e-03 -1.34926319e-01 4.99417335e-01
-9.21933472e-01 1.37044489e+00 4.65577692e-01 2.79072076e-01
-9.71719250e-02 -2.49224961e-01 -4.79978472e-01 -2.10113093e-01
2.94507176e-01 -9.43467617e-01 1.12487710e+00 -1.54076946e+00
-1.64259899e+00 1.18957388e+00 1.96461063e-02 1.11596279e-01
-1.63079381e-01 -4.58118886e-01 -8.34859192e-01 2.88851559e-01
-5.64680248e-02 4.22260880e-01 6.89907849e-01 -1.15774643e+00
-6.20347746e-02 -3.77095550e-01 -1.26379758e-01 6.68424129e-01
1.74836755e-01 5.29862463e-01 1.50544181e-01 -6.62516654e-01
6.04889572e-01 -6.71142280e-01 3.63068134e-01 -4.04812574e-01
-2.04399213e-01 5.86731546e-02 2.13265672e-01 -1.03468561e+00
1.44949389e+00 -2.36219239e+00 5.30705869e-01 2.53761202e-01
7.86180973e-01 -3.40908408e-01 1.30977109e-01 9.73340034e-01
-1.94787666e-01 6.07519858e-02 3.03906035e-02 2.48170957e-01
5.04364192e-01 1.82952788e-02 -5.12917757e-01 5.95327139e-01
-5.94777167e-01 9.96230602e-01 -6.22255385e-01 -1.31389380e-01
-5.43512516e-02 4.52756166e-01 -2.23908797e-01 -5.09656668e-01
9.10738707e-02 5.60293555e-01 1.51465824e-02 8.34291160e-01
3.55815798e-01 -5.20475626e-01 7.84105122e-01 1.26712874e-01
-6.98615372e-01 7.29381323e-01 -4.88935322e-01 1.97958231e+00
-7.12567195e-03 8.84042740e-01 1.91306576e-01 -4.11674708e-01
4.53799695e-01 4.66696382e-01 -1.98666766e-01 -1.16405249e+00
3.63562912e-01 2.13192776e-01 7.95084059e-01 -7.66324222e-01
4.80029732e-01 -6.44728482e-01 -5.56164801e-01 1.11775029e+00
-2.42695212e-01 -1.07385390e-01 -1.03064753e-01 4.24916029e-01
4.37045813e-01 1.85297981e-01 7.40876019e-01 -8.67755413e-01
2.42933869e-01 -5.09159751e-02 1.80408999e-01 4.84953791e-01
7.27042332e-02 3.58645916e-01 4.03208822e-01 -6.21186316e-01
-1.21962881e+00 -1.14764178e+00 -6.41527697e-02 1.26132727e+00
3.10805887e-01 -7.58787632e-01 -9.02432442e-01 2.00132743e-01
-4.67354923e-01 1.10420620e+00 -6.19810760e-01 -1.11706555e-01
-4.24847841e-01 -4.59505647e-01 5.90091527e-01 1.93619251e-01
1.69056103e-01 -1.03266251e+00 -1.52887762e+00 1.56688288e-01
-3.78174216e-01 -6.43092692e-01 -3.82814884e-01 -2.21563205e-02
-5.68716049e-01 -1.08373344e+00 -4.24795032e-01 -2.50411659e-01
3.96273047e-01 3.28505009e-01 8.27310979e-01 4.08719689e-01
2.11737916e-01 4.41188425e-01 -2.02057004e-01 -4.03807074e-01
-6.31594419e-01 -5.92699237e-02 2.33169775e-02 -7.95235932e-01
-1.68240312e-02 -6.30141318e-01 -3.66635799e-01 -1.21436544e-01
-7.39041030e-01 7.72852838e-01 6.11510277e-01 3.73488754e-01
-3.70850414e-01 -7.57441700e-01 2.96403497e-01 -5.75957179e-01
1.08149183e+00 -4.95390415e-01 -1.05732948e-01 2.35363320e-01
-2.33692050e-01 6.63001463e-02 1.61089689e-01 -2.93401480e-01
-1.21603239e+00 -9.50532258e-01 9.38439071e-02 5.17255366e-01
4.30096895e-01 7.47400105e-01 -2.68049184e-02 -4.80457395e-02
7.37496436e-01 2.41928950e-01 5.40926814e-01 -1.52706817e-01
2.81228155e-01 4.04580921e-01 6.06934786e-01 -5.37647307e-01
2.25801468e-01 5.19275606e-01 -2.52797514e-01 -9.38551486e-01
-5.07380724e-01 2.79615134e-01 -4.50003952e-01 -5.65045118e-01
1.02486837e+00 -8.78293812e-01 -1.07282007e+00 1.22075953e-01
-1.36303747e+00 -2.01226711e-01 -4.19978857e-01 8.76857936e-01
-7.75212705e-01 3.10953856e-01 -5.58523715e-01 -8.68390501e-01
-1.49551615e-01 -6.80601418e-01 7.31037617e-01 4.82238829e-02
-1.06516206e+00 -9.92675781e-01 2.85130978e-01 1.04648083e-01
6.47003055e-01 -2.56739352e-02 1.19590545e+00 -6.78951025e-01
-3.47212106e-01 4.90638405e-01 -1.29375219e-01 -3.90428632e-01
-2.88648605e-01 -4.22479391e-01 -4.93146867e-01 4.58240539e-01
6.09476864e-01 -1.78735659e-01 2.79244989e-01 2.03577280e-01
1.43569782e-01 -3.78150314e-01 -6.81279749e-02 3.60680014e-01
1.36144590e+00 6.22834384e-01 7.79222786e-01 5.82658947e-01
3.62523466e-01 8.94158840e-01 -2.35288795e-02 4.47171807e-01
6.71812534e-01 5.31773388e-01 -3.31261396e-01 4.23745066e-01
-2.89464146e-01 -2.43885051e-02 3.43431950e-01 1.51671159e+00
-4.99537140e-01 1.11074083e-01 -1.25493062e+00 5.78256659e-02
-1.61481225e+00 -1.22283554e+00 -2.91277260e-01 1.89442253e+00
3.56913894e-01 1.35345951e-01 3.50161754e-02 1.71089292e-01
9.02203560e-01 1.69214487e-01 1.46207675e-01 -8.52539778e-01
-4.63703245e-01 -7.91436210e-02 1.47855744e-01 6.38710082e-01
-1.83879569e-01 9.85090017e-01 8.85771084e+00 4.36386734e-01
-1.02880287e+00 5.82709134e-01 -2.14988902e-01 -3.60786021e-02
-6.82313025e-01 -1.09290481e-01 -2.79728234e-01 7.06152260e-01
1.20852268e+00 -2.29604021e-01 6.81644857e-01 3.41631435e-02
1.83989868e-01 -5.10275960e-01 -5.32183111e-01 1.08842278e+00
5.45866787e-01 -1.53194046e+00 -1.38077915e-01 8.55529085e-02
2.75255829e-01 -3.83123547e-01 2.19229415e-01 -2.60279089e-01
-1.91687226e-01 -9.48124588e-01 1.44670212e+00 7.49635696e-01
8.79906952e-01 -3.61460686e-01 3.26994181e-01 1.40131101e-01
-4.97970194e-01 -1.15122460e-01 1.67153239e-01 -7.92160034e-01
7.74546921e-01 1.99589252e-01 -1.61998689e-01 2.98229232e-02
-8.13852176e-02 2.27977633e-01 -2.22687662e-01 4.97432113e-01
-3.48411530e-01 2.43952274e-01 -1.05109014e-01 1.88122445e-03
-2.61792094e-02 -3.28291543e-02 8.88594329e-01 8.05944920e-01
3.33906293e-01 3.09871584e-01 -7.85071075e-01 8.47423077e-01
6.64565206e-01 2.07258519e-02 -4.88433838e-01 -3.82924795e-01
5.95142543e-01 6.60743773e-01 -1.25628078e+00 -2.64593959e-01
-1.90920904e-01 8.66988361e-01 1.13626920e-01 1.52861759e-01
-8.53114784e-01 1.95987150e-01 2.58270830e-01 1.59863293e-01
8.85241330e-02 -6.46376193e-01 -9.05884445e-01 -9.43780482e-01
-2.49155164e-01 -7.22888470e-01 -2.38249317e-01 -7.41344273e-01
-7.20771134e-01 4.53359514e-01 5.84615208e-02 -6.11626208e-01
-1.04208075e-01 -4.02344674e-01 -2.16800824e-01 8.37881386e-01
-4.01677370e-01 -1.20585215e+00 2.21739113e-01 2.78906554e-01
4.02713865e-01 -1.66184604e-01 1.06568694e+00 -7.09661767e-02
-1.25051633e-01 -4.81983833e-02 5.57950623e-02 -7.11649477e-01
5.01760721e-01 -1.03946793e+00 4.99406964e-01 4.51341003e-01
-6.48906752e-02 9.14425313e-01 8.62267554e-01 -8.39530468e-01
-1.49368632e+00 2.31600925e-01 1.37974250e+00 -6.23872697e-01
7.67998874e-01 -5.30191123e-01 -7.49444306e-01 5.46939075e-01
1.01747763e+00 -9.45731878e-01 1.17908967e+00 1.56409889e-01
-4.45070356e-01 4.41298872e-01 -7.53456414e-01 7.15103924e-01
1.38356316e+00 -7.46515930e-01 -1.53840160e+00 1.04706101e-01
3.29993159e-01 -3.84582430e-01 -4.76937443e-01 -3.72129649e-01
1.22318411e+00 -1.02915370e+00 6.34831369e-01 -1.57399967e-01
4.38006639e-01 -2.26916239e-01 3.78778540e-02 -1.00295627e+00
-5.81666648e-01 -7.91173697e-01 3.63803655e-01 7.21855164e-01
1.68820187e-01 -9.16041672e-01 -1.29848629e-01 3.73925924e-01
-5.52206457e-01 -2.02660725e-01 -1.09868205e+00 -2.45051697e-01
-3.56711149e-01 -5.88284910e-01 3.33287716e-01 9.69851732e-01
1.00110519e+00 6.04640245e-01 -2.35081270e-01 -6.26978815e-01
4.02720213e-01 -4.18098330e-01 1.65719032e-01 -1.44235361e+00
-1.40501797e-01 -7.80675411e-01 -1.20783299e-01 -5.89920938e-01
1.25305131e-01 -9.93787467e-01 -2.04778120e-01 -1.32389581e+00
2.67560452e-01 1.37398168e-01 1.67815194e-01 2.47942638e-02
6.29663527e-01 3.32113028e-01 6.91718996e-01 8.54966462e-01
-8.91615391e-01 1.25372455e-01 1.29376876e+00 5.81364572e-01
-3.46894354e-01 -1.09199917e+00 -1.55825055e+00 8.98351848e-01
7.51089096e-01 -2.89218098e-01 -4.92326409e-01 -5.05342424e-01
1.35949850e+00 3.09663773e-01 4.11765844e-01 -8.95705700e-01
3.96458119e-01 -1.32920697e-01 2.85618514e-01 -3.52692038e-01
1.54363319e-01 -7.44065225e-01 6.45703137e-01 8.01701248e-01
-1.68575659e-01 8.01765919e-01 4.17892069e-01 2.34762952e-02
7.49324933e-02 1.31996036e-01 3.27473909e-01 -1.11815721e-01
-7.93439209e-01 -1.12352002e+00 -1.37829590e+00 1.07427001e-01
7.17395008e-01 -5.16664088e-01 -8.99618924e-01 -2.92015553e-01
-7.86036432e-01 -3.72887135e-01 5.79162896e-01 2.99981862e-01
4.79814887e-01 -1.43952215e+00 -3.83246750e-01 1.34427315e-02
-3.83525491e-01 -7.94273376e-01 5.94848573e-01 1.47381628e+00
-8.70919704e-01 8.12496901e-01 -7.26116002e-01 6.36197850e-02
-7.79758275e-01 4.81233448e-01 -7.42256641e-02 3.68739367e-01
-9.97812986e-01 4.15379643e-01 4.11597699e-01 1.70355275e-01
-6.00306273e-01 -7.81862438e-02 -1.89345360e-01 3.64821404e-01
8.40328038e-01 6.68258727e-01 -5.36975384e-01 -9.94947135e-01
-7.89709628e-01 7.04801738e-01 7.02982247e-02 -8.62074554e-01
7.06065834e-01 -8.01033080e-01 -6.06186092e-01 1.07193327e+00
3.51120502e-01 4.44102854e-01 -5.50488889e-01 4.61498648e-01
-9.17136967e-02 -2.52474099e-01 -5.53101778e-01 -1.43410432e+00
-1.10051334e-01 6.05282724e-01 2.16797903e-01 2.11030051e-01
6.52116954e-01 1.61059812e-01 4.17964607e-01 4.70247008e-02
-7.59377563e-03 -1.43037665e+00 1.79905444e-03 7.15602636e-01
1.38397932e+00 -3.13857317e-01 1.00149870e-01 -3.32720041e-01
-9.39556599e-01 1.22894740e+00 1.10245563e-01 -6.23997934e-02
3.29686493e-01 5.08649349e-01 7.34843090e-02 -6.52323902e-01
-9.28983271e-01 -1.45953698e-02 2.54191041e-01 6.49647176e-01
6.32996798e-01 5.86403906e-02 -1.40881252e+00 1.06692553e+00
-7.42476404e-01 4.54994477e-02 4.16720241e-01 5.72538912e-01
-7.34805584e-01 -8.27306926e-01 -1.05617011e+00 -1.59716055e-01
-2.39645049e-01 -2.47105017e-01 -4.50613052e-01 1.10976970e+00
4.91630077e-01 5.20566642e-01 4.18434113e-01 -2.92880982e-01
-1.11804731e-01 5.23751438e-01 9.98428822e-01 -2.15946063e-01
-9.54274416e-01 4.58524704e-01 3.71974677e-01 -3.32984895e-01
-6.75469100e-01 -1.10377884e+00 -1.31582177e+00 -5.10006249e-01
3.74089152e-01 2.08755150e-01 7.83387125e-01 1.42437208e+00
2.96770930e-01 6.80768669e-01 -4.63930130e-01 -7.27757275e-01
-1.43371537e-01 -8.51472914e-01 -7.42099762e-01 -1.95566174e-02
7.53129348e-02 -6.95883095e-01 -4.18621063e-01 -1.36270627e-01]
|
[10.080007553100586, 8.451208114624023]
|
f18a661c-3224-4bfc-aa6b-faec2676a35a
|
subjective-and-objective-quality-assessment-4
|
2303.08050
| null |
https://arxiv.org/abs/2303.08050v2
|
https://arxiv.org/pdf/2303.08050v2.pdf
|
Subjective and Objective Quality Assessment for in-the-Wild Computer Graphics Images
|
Computer graphics images (CGIs) are artificially generated by means of computer programs and are widely perceived under various scenarios, such as games, streaming media, etc. In practice, the quality of CGIs consistently suffers from poor rendering during production, inevitable compression artifacts during the transmission of multimedia applications, and low aesthetic quality resulting from poor composition and design. However, few works have been dedicated to dealing with the challenge of computer graphics image quality assessment (CGIQA). Most image quality assessment (IQA) metrics are developed for natural scene images (NSIs) and validated on databases consisting of NSIs with synthetic distortions, which are not suitable for in-the-wild CGIs. To bridge the gap between evaluating the quality of NSIs and CGIs, we construct a large-scale in-the-wild CGIQA database consisting of 6,000 CGIs (CGIQA-6k) and carry out the subjective experiment in a well-controlled laboratory environment to obtain the accurate perceptual ratings of the CGIs. Then, we propose an effective deep learning-based no-reference (NR) IQA model by utilizing both distortion and aesthetic quality representation. Experimental results show that the proposed method outperforms all other state-of-the-art NR IQA methods on the constructed CGIQA-6k database and other CGIQA-related databases. The database will be released to facilitate further research.
|
['Guangtao Zhai', 'Xiongkuo Min', 'Qiyuan Wang', 'Jun He', 'Quan Zhou', 'Wei Lu', 'Tao Wang', 'Wei Sun', 'ZiCheng Zhang']
|
2023-03-14
| null | null | null | null |
['image-quality-assessment']
|
['computer-vision']
|
[ 1.69256881e-01 -6.37221396e-01 4.18570846e-01 -2.95857519e-01
-6.10170722e-01 -1.07148513e-01 2.70210475e-01 7.04121143e-02
-1.53492674e-01 2.35578597e-01 -1.38305333e-02 -1.46720171e-01
-1.93912312e-01 -9.93460059e-01 -6.47016644e-01 -5.16586065e-01
-1.98051214e-01 -6.24180771e-02 1.43372595e-01 -3.87384921e-01
4.48889256e-01 4.06389415e-01 -1.80749536e+00 3.24845403e-01
1.06341386e+00 1.36550009e+00 2.68375009e-01 8.57586563e-01
4.77975830e-02 5.79807878e-01 -1.11235476e+00 -7.99861312e-01
3.01045746e-01 -5.59839725e-01 -2.14679882e-01 9.33539495e-02
4.83130664e-01 -7.72053659e-01 -3.28918070e-01 1.16215289e+00
9.11766648e-01 1.04674302e-01 5.06203175e-01 -1.48776865e+00
-9.16715205e-01 -2.07995847e-01 -5.56648076e-01 1.72774062e-01
4.21126425e-01 4.11280453e-01 7.84521878e-01 -8.95196676e-01
3.46791685e-01 1.51667058e+00 4.62235659e-01 1.64696723e-01
-7.42635012e-01 -7.89181232e-01 -2.31489733e-01 5.31624317e-01
-1.48014975e+00 -3.30266982e-01 8.23874295e-01 -2.01550126e-01
6.16728246e-01 3.68094891e-01 8.09153736e-01 8.38363469e-01
2.34544978e-01 6.38307750e-01 1.06158793e+00 -3.38988036e-01
5.15061557e-01 -1.82838500e-01 -4.37297940e-01 6.50500953e-01
1.58799917e-01 3.04808378e-01 -6.12045884e-01 7.44283199e-02
1.19904733e+00 -3.74818712e-01 -3.98781687e-01 -1.60566553e-01
-1.05187619e+00 5.91040492e-01 4.30724949e-01 -1.22949919e-02
-1.51443675e-01 -1.06562473e-01 5.21157146e-01 5.52068472e-01
2.28043884e-01 2.98659593e-01 -7.52648041e-02 -3.64854038e-01
-8.52406323e-01 4.92017299e-01 4.72211182e-01 1.02641582e+00
4.29595053e-01 3.51516038e-01 -1.08251721e-01 1.21285284e+00
2.33949006e-01 6.49137616e-01 5.78002334e-01 -1.29603863e+00
4.42771643e-01 3.27134043e-01 3.55674982e-01 -1.62802374e+00
-7.74279758e-02 -2.14039966e-01 -1.17327154e+00 7.67090082e-01
2.15749517e-01 2.70368606e-01 -5.25753140e-01 1.29323578e+00
4.74872859e-03 5.67083694e-02 -4.66611497e-02 1.07224059e+00
1.32725775e+00 1.03911734e+00 -1.27021357e-01 -2.05866754e-01
1.00811756e+00 -8.74382138e-01 -7.50663042e-01 2.55244046e-01
1.98218554e-01 -9.86897945e-01 1.83493984e+00 9.57322419e-01
-1.32654667e+00 -1.14900088e+00 -1.51313949e+00 -2.98801642e-02
-1.43871501e-01 1.17503189e-01 2.98059136e-01 9.57886696e-01
-1.26583886e+00 6.06036305e-01 -3.13811481e-01 -5.46142198e-02
4.60697412e-01 -9.46505144e-02 -1.98058635e-01 -2.28365973e-01
-1.16727555e+00 5.06842673e-01 9.52047333e-02 2.93062609e-02
-9.83692944e-01 -6.54144704e-01 -8.33286703e-01 9.15140063e-02
1.86913937e-01 -4.89466220e-01 1.03529477e+00 -1.11028838e+00
-1.76529694e+00 8.64462197e-01 1.71442553e-01 1.67175367e-01
6.02763474e-01 -1.64841905e-01 -9.19065654e-01 1.80902794e-01
-1.94000050e-01 4.90470737e-01 8.83869827e-01 -1.59010577e+00
-3.53558928e-01 -2.41806403e-01 2.30239496e-01 2.66226292e-01
-2.01090455e-01 1.76712379e-01 -7.53068388e-01 -8.48376513e-01
-8.68064091e-02 -3.77931833e-01 1.13671072e-01 5.83170712e-01
-2.78895348e-01 2.42345050e-01 6.40601754e-01 -6.99734271e-01
1.29276478e+00 -2.32796502e+00 -3.06987613e-01 6.39101416e-02
8.99882019e-02 7.80764699e-01 -5.71891606e-01 2.40205035e-01
-1.68726873e-02 2.20062748e-01 5.60802259e-02 -2.12281615e-01
-1.26443878e-02 7.87114203e-02 -5.12457378e-02 1.80013210e-01
9.83694419e-02 6.42761111e-01 -7.70530999e-01 -6.17436290e-01
4.69796568e-01 3.38677078e-01 -6.04335248e-01 6.88231647e-01
6.28313748e-03 3.19614232e-01 -2.24192590e-01 7.20006108e-01
1.20085764e+00 -7.84792304e-02 -3.82068813e-01 -5.08543134e-01
1.04291998e-01 -2.88146764e-01 -1.33883500e+00 1.69271719e+00
-5.76806903e-01 5.85725188e-01 -2.67209113e-02 -6.78536713e-01
9.92761791e-01 -2.22014729e-02 2.12218031e-01 -1.40745294e+00
1.48555487e-01 1.43975720e-01 -2.82672327e-02 -8.11345041e-01
6.50757313e-01 2.78073996e-02 1.94709942e-01 1.92542285e-01
-1.21496484e-01 -4.08135593e-01 1.88570008e-01 1.30421907e-01
8.09392869e-01 -1.56580359e-01 1.58222571e-01 2.31300406e-02
6.01095676e-01 -3.84337932e-01 5.39470971e-01 6.43256128e-01
-5.65853655e-01 1.14250326e+00 2.69783258e-01 -4.36507761e-01
-1.43580186e+00 -1.35062933e+00 -4.92381677e-02 8.20972502e-01
6.18595123e-01 -3.77571106e-01 -8.88206363e-01 5.71929943e-03
-4.60075706e-01 5.27632296e-01 -2.66288161e-01 -1.48516029e-01
-3.49206448e-01 -6.21201396e-01 5.75198293e-01 1.56583786e-01
1.13745391e+00 -1.32575679e+00 -4.49338317e-01 9.75463167e-02
-2.01393709e-01 -1.02475107e+00 -3.90463442e-01 -8.71813357e-01
-5.35946667e-01 -1.05390966e+00 -9.07598555e-01 -5.01409829e-01
2.12981164e-01 5.08748174e-01 1.51756644e+00 3.35111916e-01
-3.61857474e-01 4.40845266e-02 -4.89575684e-01 -2.81531841e-01
-6.09245539e-01 -7.33256519e-01 -1.91506073e-01 -5.77558428e-02
-1.17228538e-01 -5.78662574e-01 -9.86900687e-01 6.55616581e-01
-1.32614160e+00 3.13734800e-01 4.68695164e-01 7.02621341e-01
5.99956214e-01 6.40371978e-01 5.42164803e-01 -5.29637933e-01
8.38404655e-01 -9.99512374e-02 -5.77716291e-01 1.92766756e-01
-4.52914298e-01 -4.51983690e-01 7.83038735e-01 -2.50622958e-01
-1.30984700e+00 -9.19522882e-01 -5.32542288e-01 -2.74554580e-01
-1.67426452e-01 4.76146132e-01 -8.00933540e-01 -2.15721101e-01
6.58333719e-01 1.48816228e-01 -1.56295359e-01 -3.01954687e-01
1.68421090e-01 8.78806472e-01 7.63759255e-01 -5.40543139e-01
5.80313087e-01 2.43940204e-01 -1.64496526e-01 -8.50152314e-01
-3.89511526e-01 -3.84159274e-02 5.66488020e-02 -5.96772015e-01
4.76485997e-01 -8.69587362e-01 -7.97578812e-01 1.21432769e+00
-1.04983068e+00 -2.61286706e-01 -3.67051251e-02 2.85019040e-01
-7.22699225e-01 6.71087801e-01 -7.32949615e-01 -6.64601445e-01
-3.90614033e-01 -1.45385802e+00 1.09439468e+00 4.03010875e-01
1.40645638e-01 -4.39194739e-01 3.82576436e-02 5.20246267e-01
6.05136812e-01 2.71913618e-01 1.03013813e+00 3.87276173e-01
-5.04409850e-01 -7.40073100e-02 -7.11259604e-01 6.96583986e-01
7.53288269e-02 3.91792446e-01 -8.66255462e-01 -2.89757639e-01
6.43406287e-02 -5.80570161e-01 2.51716703e-01 4.34397995e-01
1.59032297e+00 -2.27778032e-01 6.18182421e-01 8.90622556e-01
1.67779183e+00 5.66021562e-01 1.30700922e+00 3.69319260e-01
5.22950053e-01 4.18676108e-01 7.96595752e-01 7.09974408e-01
2.20318511e-01 7.14488864e-01 5.80637693e-01 -2.02594534e-01
-2.03848660e-01 -1.59371033e-01 2.16112748e-01 1.21160841e+00
-2.22822383e-01 -7.09220409e-01 -5.94503045e-01 2.50130355e-01
-1.37313139e+00 -6.69523299e-01 -8.45885500e-02 2.19240236e+00
7.23972678e-01 1.33868665e-01 -8.71987492e-02 4.96446729e-01
6.77611530e-01 1.96575969e-01 -6.60145760e-01 -6.40052259e-01
-3.52463186e-01 1.89894050e-01 -2.74347793e-02 4.97783944e-02
-9.16470587e-01 4.04545814e-01 6.12198067e+00 1.27290058e+00
-1.07064116e+00 -4.14698757e-02 1.04310739e+00 6.25136271e-02
-2.70126343e-01 -5.51242650e-01 4.39698547e-02 7.05176532e-01
9.23291206e-01 -2.62192547e-01 5.38835526e-01 8.27790022e-01
6.11469090e-01 -1.57750279e-01 -6.51628137e-01 1.66448045e+00
9.69210863e-02 -1.15526772e+00 3.31366718e-01 -2.66865402e-01
7.79750466e-01 -4.08072293e-01 5.53054512e-01 9.14127380e-02
-3.42785716e-02 -1.14324057e+00 7.64572620e-01 4.58682656e-01
1.28215718e+00 -8.22441697e-01 7.77098417e-01 1.39883846e-01
-9.75477517e-01 1.71790123e-01 -7.74852872e-01 -1.70112103e-02
6.12588227e-02 6.14482045e-01 6.29934147e-02 6.18121147e-01
9.32449460e-01 5.47404349e-01 -7.94134080e-01 1.37922478e+00
1.97250098e-01 5.02785087e-01 2.11376458e-01 1.84198335e-01
5.45321032e-02 -4.17107165e-01 2.11749807e-01 9.06215847e-01
7.27301419e-01 2.47176930e-01 -3.29596162e-01 8.78340304e-01
-4.06436771e-02 4.42993075e-01 -3.89121622e-01 1.23045202e-02
2.27284819e-01 9.07286525e-01 -3.46519768e-01 -5.05894363e-01
-4.79063332e-01 1.13150918e+00 -2.37054199e-01 4.11628246e-01
-8.44892144e-01 -4.46565986e-01 7.27649152e-01 6.21383116e-02
-8.79619941e-02 -9.09437537e-02 -2.11015388e-01 -1.16573250e+00
2.41278961e-01 -1.31507027e+00 3.74592841e-02 -1.49939561e+00
-1.42675352e+00 7.57403314e-01 -8.51523131e-02 -1.67679238e+00
-1.07135765e-01 -5.44749737e-01 -7.00199246e-01 9.25007343e-01
-1.42976975e+00 -7.96838164e-01 -9.89406765e-01 6.93133593e-01
5.82054257e-01 -1.95620388e-01 7.65567183e-01 6.48888707e-01
-4.15732265e-01 8.14889967e-01 2.59304076e-01 -1.60322458e-01
6.04764760e-01 -7.88772166e-01 5.71264327e-01 6.55522108e-01
-2.71448404e-01 1.31719902e-01 6.66882336e-01 -2.24246159e-01
-1.30198216e+00 -9.89902854e-01 2.00841427e-01 1.82311580e-01
8.31464455e-02 -1.09784603e-01 -1.15725923e+00 -2.13941291e-01
2.65079886e-01 1.07124686e-01 5.17410576e-01 -5.59704244e-01
-2.74289221e-01 -3.41514915e-01 -1.40177894e+00 6.00311220e-01
1.11100745e+00 -3.57881874e-01 -2.56338827e-02 -6.84659034e-02
7.45131195e-01 -3.38974357e-01 -9.40255105e-01 5.02726078e-01
5.87252557e-01 -1.63043797e+00 1.18117261e+00 -5.24358787e-02
9.10600364e-01 -4.26071495e-01 -2.30804101e-01 -1.40482819e+00
-2.24664748e-01 -3.27952147e-01 2.39545807e-01 1.31675255e+00
-2.19652265e-01 -1.62866414e-01 6.37391925e-01 4.07830417e-01
-1.44533351e-01 -5.65818012e-01 -6.22820020e-01 -8.23118746e-01
-8.30662325e-02 -6.47856593e-01 9.79323685e-01 6.05264544e-01
-5.23693800e-01 -6.73268437e-02 -5.59405625e-01 -2.15096455e-02
7.06094682e-01 -4.11380939e-02 9.75894511e-01 -8.03180695e-01
-3.85795742e-01 -5.10138333e-01 -8.65625858e-01 -7.60592222e-01
-4.43221211e-01 -7.21977502e-02 -2.61039827e-02 -1.29300797e+00
1.64470837e-01 -4.51128781e-01 -2.70124763e-01 -3.65225852e-01
-2.86518484e-01 4.54520911e-01 2.47368678e-01 1.63761839e-01
-6.50695860e-01 9.69445407e-01 1.79750109e+00 -3.13971967e-01
6.95357546e-02 -2.52346963e-01 -5.38855731e-01 7.41631866e-01
7.02567816e-01 1.20378345e-01 -7.25669622e-01 -3.96547437e-01
3.22415590e-01 3.54494393e-01 2.71733969e-01 -1.37817931e+00
-1.68157935e-01 -2.90830314e-01 4.14182067e-01 -5.20941556e-01
3.30758780e-01 -5.30521095e-01 4.59851414e-01 1.71011105e-01
-2.34600678e-01 8.19218010e-02 5.04131801e-02 3.97092998e-01
-6.28376901e-01 -1.49536908e-01 1.04723418e+00 -1.32387534e-01
-1.05282843e+00 3.66905481e-01 -1.11980893e-01 1.83393672e-01
6.42436206e-01 -3.74026209e-01 -4.02147293e-01 -7.92364657e-01
-2.78813481e-01 -3.12769234e-01 7.75264382e-01 3.08291674e-01
1.24277663e+00 -1.57376337e+00 -8.68119657e-01 4.84513730e-01
4.64016050e-01 -1.19231485e-01 8.28675210e-01 -1.49885109e-02
-1.25161338e+00 -1.21703304e-01 -7.95168161e-01 -3.28277856e-01
-1.12406623e+00 6.51707232e-01 1.65962756e-01 -1.32555723e-01
-4.14347023e-01 6.78121805e-01 5.44741213e-01 -2.40017980e-01
2.65965909e-01 -2.42294744e-01 -2.28064880e-01 -4.93366271e-01
9.06068087e-01 5.59028029e-01 3.47749144e-01 -8.23446512e-01
1.08730160e-01 4.68430310e-01 2.96831250e-01 -8.11243951e-02
1.12126923e+00 -2.90037274e-01 9.18445215e-02 2.31621534e-01
1.37638867e+00 -3.45142722e-01 -1.20663583e+00 -1.51539743e-01
-6.94789231e-01 -1.23886490e+00 7.50446087e-03 -8.13609004e-01
-1.44785297e+00 1.06322920e+00 1.22062886e+00 5.16649894e-02
1.63587964e+00 -5.95678806e-01 1.14446700e+00 1.28519339e-02
6.56782568e-01 -1.01279771e+00 6.43304408e-01 1.39149958e-02
1.30170178e+00 -1.23711884e+00 -2.32679352e-01 -4.05813366e-01
-7.15718806e-01 9.04342771e-01 6.47653818e-01 -5.36463112e-02
5.99307179e-01 1.07713223e-01 4.26773399e-01 -3.23505886e-02
-4.09025937e-01 2.80715555e-01 1.64403334e-01 9.01455104e-01
2.64473736e-01 1.77838311e-01 -1.18640661e-01 5.55405915e-01
-4.17949885e-01 2.33587205e-01 7.03757524e-01 6.69102609e-01
-2.39866510e-01 -8.51128161e-01 -5.97506166e-01 4.74197268e-01
-4.08823520e-01 -7.45271891e-02 1.27620205e-01 4.74173367e-01
2.58415103e-01 1.27941632e+00 -4.95554097e-02 -5.75689554e-01
5.88125706e-01 -7.49885619e-01 3.80419940e-01 2.43078242e-03
-1.55400127e-01 -1.27130315e-01 -1.22110270e-01 -9.10081029e-01
-4.31019545e-01 -1.61910579e-01 -7.17354417e-01 -8.29496443e-01
-1.89260021e-01 -3.45825076e-01 6.99337304e-01 4.41057265e-01
1.57754079e-01 6.91254616e-01 6.34948909e-01 -9.33825076e-01
-1.44900456e-01 -7.65328228e-01 -1.04250658e+00 9.91455078e-01
-8.52568191e-04 -6.06660843e-01 -1.18264727e-01 7.20931366e-02]
|
[11.803017616271973, -1.8398901224136353]
|
7beaf2d8-0857-44c1-86c7-ccf3dd6c82ab
|
reconstructing-continuously-heterogeneous
|
1909.05215
| null |
https://arxiv.org/abs/1909.05215v3
|
https://arxiv.org/pdf/1909.05215v3.pdf
|
Reconstructing continuous distributions of 3D protein structure from cryo-EM images
|
Cryo-electron microscopy (cryo-EM) is a powerful technique for determining the structure of proteins and other macromolecular complexes at near-atomic resolution. In single particle cryo-EM, the central problem is to reconstruct the three-dimensional structure of a macromolecule from $10^{4-7}$ noisy and randomly oriented two-dimensional projections. However, the imaged protein complexes may exhibit structural variability, which complicates reconstruction and is typically addressed using discrete clustering approaches that fail to capture the full range of protein dynamics. Here, we introduce a novel method for cryo-EM reconstruction that extends naturally to modeling continuous generative factors of structural heterogeneity. This method encodes structures in Fourier space using coordinate-based deep neural networks, and trains these networks from unlabeled 2D cryo-EM images by combining exact inference over image orientation with variational inference for structural heterogeneity. We demonstrate that the proposed method, termed cryoDRGN, can perform ab initio reconstruction of 3D protein complexes from simulated and real 2D cryo-EM image data. To our knowledge, cryoDRGN is the first neural network-based approach for cryo-EM reconstruction and the first end-to-end method for directly reconstructing continuous ensembles of protein structures from cryo-EM images.
|
['Tristan Bepler', 'Joseph H. Davis', 'Ellen D. Zhong', 'Bonnie Berger']
|
2019-09-11
| null |
https://openreview.net/forum?id=SJxUjlBtwB
|
https://openreview.net/pdf?id=SJxUjlBtwB
|
iclr-2020-1
|
['3d-volumetric-reconstruction', 'cryogenic-electron-microscopy-cryo-em']
|
['computer-vision', 'computer-vision']
|
[ 9.94725749e-02 -1.71265319e-01 3.30210596e-01 -5.35452843e-01
-8.33578229e-01 -3.86192143e-01 3.94084454e-01 -1.36996478e-01
-6.21402442e-01 9.75324214e-01 -1.31387398e-01 -5.04665673e-01
-3.43887471e-02 -3.45649660e-01 -1.10035610e+00 -1.31694770e+00
4.67407219e-02 1.37421083e+00 -9.23646763e-02 1.26732618e-01
6.60908297e-02 7.79646635e-01 -1.22906804e+00 4.52373087e-01
4.39720571e-01 5.37693381e-01 8.35316122e-01 6.22514606e-01
-1.74039006e-01 5.30725300e-01 -1.16393469e-01 8.62637684e-02
-2.07040459e-01 -5.22135675e-01 -6.66822553e-01 1.75844714e-01
2.53180444e-01 1.90173443e-02 -5.93787916e-02 8.85393918e-01
5.86632013e-01 1.70215964e-01 1.00919223e+00 -1.60658613e-01
-7.52650738e-01 1.97636820e-02 -2.82272279e-01 2.38098755e-01
1.64367288e-01 4.19501990e-01 8.67722631e-01 -9.89395678e-01
1.28561425e+00 1.17051959e+00 5.02224267e-01 7.36475110e-01
-2.07545686e+00 -1.58491164e-01 3.04248817e-02 1.49383605e-01
-1.00786579e+00 -2.80018598e-01 7.47959197e-01 -6.62269115e-01
1.38664436e+00 -2.14111447e-01 5.49750566e-01 1.32583117e+00
5.03652394e-01 3.48727405e-01 1.30142176e+00 -2.37463027e-01
4.73135561e-01 -5.84159970e-01 2.80165076e-01 7.37550914e-01
-6.80394024e-02 1.41560556e-02 -8.50426108e-02 -6.61176622e-01
4.78438824e-01 4.84735459e-01 -2.04791203e-01 -7.61307895e-01
-1.02255309e+00 6.65170252e-01 3.87849547e-02 8.89604613e-02
-7.13367343e-01 -2.53051538e-02 2.26700127e-01 4.31496613e-02
3.47892433e-01 4.32430983e-01 -4.81856197e-01 -1.26645699e-01
-1.04551041e+00 5.42642117e-01 5.96249759e-01 2.92276889e-01
9.15458798e-01 -3.48703831e-01 7.17572927e-01 6.01559877e-01
3.53696495e-01 3.94378811e-01 3.87935609e-01 -1.19474185e+00
-7.05074891e-02 1.69565946e-01 3.04452747e-01 -3.07762295e-01
-5.80977142e-01 3.73201579e-01 -1.00938857e+00 5.10712028e-01
3.25937241e-01 -5.61879836e-02 -1.10958183e+00 1.80435956e+00
4.76909488e-01 -1.11178845e-01 -2.16711238e-01 9.73021030e-01
6.18549526e-01 7.08869338e-01 -1.46265730e-01 -6.80324316e-01
1.02946520e+00 -4.77528036e-01 -6.17783546e-01 1.63516298e-01
1.74378440e-01 -4.67395902e-01 5.59353471e-01 4.78216588e-01
-1.22711277e+00 -4.61340547e-01 -1.10831559e+00 -1.12896085e-01
-1.36196107e-01 -1.05949238e-01 3.91981483e-01 1.91183344e-01
-8.49509001e-01 9.94581759e-01 -1.65080249e+00 -1.77934855e-01
1.70512438e-01 5.62446773e-01 -7.64311910e-01 -9.62493643e-02
-5.79895079e-01 9.72337902e-01 3.60106081e-01 -1.75813492e-02
-1.11048138e+00 -4.74121213e-01 -4.99385983e-01 -1.98564649e-01
-9.49935708e-03 -7.58074105e-01 9.52161729e-01 -3.96696180e-01
-1.57455599e+00 1.06232858e+00 -9.69241500e-01 -3.90922308e-01
-1.76127806e-01 2.76109338e-01 3.83658037e-02 2.17645571e-01
-1.19450249e-01 4.49074566e-01 6.40397966e-01 -1.45940506e+00
1.91465572e-01 -9.01620626e-01 -5.36338389e-01 4.19120938e-02
7.51724303e-01 -1.74530908e-01 2.33440027e-01 1.34578511e-01
7.02873707e-01 -8.75115812e-01 -4.68847096e-01 -3.04237187e-01
-3.49161714e-01 2.85194255e-02 8.26134622e-01 -5.27580082e-01
4.55545276e-01 -1.76998389e+00 8.89782965e-01 -8.01036731e-05
7.09279060e-01 1.87755659e-01 1.56527564e-01 6.56155467e-01
-2.77081132e-01 -2.13396341e-01 -4.36657727e-01 -7.60236979e-01
8.42938200e-03 4.89515960e-01 -3.49221528e-02 8.17854404e-01
1.34105697e-01 8.57593238e-01 -4.80647862e-01 -1.93444952e-01
3.67623746e-01 8.93379211e-01 -4.51203436e-01 4.80749279e-01
-6.86271429e-01 9.22404706e-01 -1.20459095e-01 3.25978965e-01
1.09320045e+00 -6.94579124e-01 1.06780887e+00 -1.53968662e-01
-2.27921829e-01 3.65162522e-01 -7.75728285e-01 1.67638469e+00
3.00407745e-02 1.65744767e-01 6.40384912e-01 -1.41771448e+00
7.84897864e-01 3.52352500e-01 6.77018881e-01 -4.43309784e-01
-1.46915287e-01 -5.06250523e-02 3.72902192e-02 -4.45324659e-01
7.30472058e-02 -8.87532949e-01 2.71811277e-01 8.37950766e-01
4.28949028e-01 8.67703259e-02 -5.29559925e-02 8.30460489e-02
9.66386318e-01 4.53609675e-01 1.35122404e-01 -1.29020482e-01
2.11563170e-01 5.09989157e-04 5.49387038e-01 5.11840701e-01
-1.31896019e-01 8.14189017e-01 3.77166450e-01 -9.61886823e-01
-1.88926089e+00 -1.52697837e+00 -4.10712451e-01 7.57893801e-01
-9.99256074e-02 -2.56169021e-01 -9.50184882e-01 -1.62753731e-01
-5.93876578e-02 1.26002459e-02 -5.64588189e-01 9.63436663e-02
-5.76137483e-01 -1.42453074e+00 1.54162318e-01 -6.58880547e-02
-2.40152672e-01 -1.24415123e+00 -3.89911741e-01 5.28971016e-01
-2.37331912e-01 -9.05131161e-01 -1.11846291e-01 8.62253129e-01
-1.07960045e+00 -1.24523377e+00 -5.57192564e-01 -6.67010248e-01
8.74269366e-01 2.86925614e-01 1.19870532e+00 -1.48136839e-01
-6.43001735e-01 -1.18705451e-01 1.81464300e-01 2.26459354e-01
-7.14429200e-01 -1.85896695e-01 5.70000052e-01 -2.38727197e-01
7.90071785e-01 -1.19603980e+00 -5.70583999e-01 2.24795356e-01
-8.67576480e-01 1.29771665e-01 3.05536956e-01 1.21033418e+00
1.39010656e+00 -7.92151839e-02 -3.45894471e-02 -1.09280324e+00
6.95818245e-01 -2.36517042e-01 -6.20812476e-01 9.05892923e-02
-4.39950258e-01 5.58882415e-01 8.74954700e-01 -2.29318172e-01
-1.16179466e+00 4.51824009e-01 -4.41229880e-01 -5.69179237e-01
-8.25411737e-01 3.34637076e-01 -2.46148765e-01 1.86834484e-01
5.38262844e-01 8.43573511e-01 5.28464317e-01 -7.40776777e-01
1.76289439e-01 2.84276456e-01 5.06783605e-01 -1.01163542e+00
7.75993019e-02 9.31247056e-01 2.46891961e-01 -9.59126532e-01
-5.70726991e-01 -4.59422976e-01 -1.05813909e+00 3.24194640e-01
1.08913636e+00 -7.45193601e-01 -1.37491524e+00 3.93063098e-01
-1.24855447e+00 -2.51764327e-01 2.59154975e-01 5.13130069e-01
-1.06791389e+00 9.30877209e-01 -9.56241429e-01 -7.41985381e-01
-2.17709377e-01 -1.67751133e+00 1.21274674e+00 -2.95475006e-01
-6.63445219e-02 -7.90701985e-01 7.26315618e-01 3.52738202e-01
-4.52595856e-03 3.12396586e-01 1.18212938e+00 -7.31400773e-03
-4.76520568e-01 2.06473500e-01 1.16942532e-01 1.12169974e-01
6.97872713e-02 1.03362918e-01 -7.00599730e-01 -4.39422905e-01
3.31475616e-01 -5.56053698e-01 1.10051572e+00 9.88461077e-01
8.18045676e-01 6.34263679e-02 -5.21541059e-01 7.97760844e-01
1.44775724e+00 2.51595348e-01 5.95291495e-01 5.83164319e-02
6.56573057e-01 5.03331721e-01 1.10241048e-01 3.98316205e-01
6.00610338e-02 5.93145728e-01 4.99900877e-01 1.43451333e-01
5.89923918e-01 -2.04147577e-01 1.81262106e-01 8.29970956e-01
-5.35912931e-01 -9.64695588e-02 -6.83328331e-01 6.73552603e-02
-1.91937983e+00 -1.32517254e+00 1.45510808e-01 1.93747067e+00
8.45054030e-01 -3.86235230e-02 2.40894347e-01 -3.29264224e-01
7.26440609e-01 1.00075342e-01 -9.81362581e-01 -3.63315232e-02
-3.50209713e-01 3.63435537e-01 5.25116324e-02 6.97726786e-01
-8.92507136e-01 7.09061265e-01 7.11236763e+00 1.74426660e-01
-1.13011456e+00 1.61025584e-01 4.71288204e-01 -2.29912147e-01
-1.30361408e-01 1.65120512e-01 -1.00596809e+00 6.58483624e-01
1.18782687e+00 5.00122607e-01 6.81622028e-01 6.14993155e-01
4.54944015e-01 8.67763348e-03 -1.06865621e+00 8.48739743e-01
-3.71899635e-01 -1.90838337e+00 -9.36785433e-03 4.42060679e-01
4.77781981e-01 5.84657431e-01 -6.74497485e-02 -3.52006227e-01
6.88868761e-01 -1.15039539e+00 1.55791879e-01 7.12688684e-01
7.18117476e-01 -8.81802022e-01 4.39336747e-01 8.44856858e-01
-6.96614087e-01 4.25660759e-01 -9.82294738e-01 -1.94656365e-02
7.22516358e-01 8.18102419e-01 -8.02581966e-01 -1.95013490e-02
6.36481285e-01 6.08043253e-01 4.33991909e-01 4.12893593e-02
3.53209436e-01 4.15630758e-01 -3.37676227e-01 7.04815760e-02
9.30534229e-02 -9.37052369e-01 3.05158138e-01 1.01712739e+00
4.30223979e-02 2.59669095e-01 2.21197620e-01 1.38693869e+00
-1.10443667e-01 -6.34985328e-01 -5.27835310e-01 -3.77666861e-01
2.64912069e-01 1.05344594e+00 -5.17891884e-01 -1.20609574e-01
-7.20603988e-02 9.94349539e-01 8.55881453e-01 3.94843817e-01
-4.09202456e-01 7.33306631e-02 1.04604089e+00 4.63306546e-01
7.54101336e-01 -7.66584933e-01 4.01246578e-01 -1.26243329e+00
-1.01889677e-01 -8.85351121e-01 -1.25619397e-01 -7.19553411e-01
-1.48441672e+00 4.39293832e-01 -4.45592731e-01 -3.83777112e-01
-3.74845654e-01 -1.06024635e+00 -2.77366698e-01 1.18807089e+00
-9.50389147e-01 -6.97256684e-01 3.08396101e-01 2.34320000e-01
5.87144680e-02 -9.20744985e-02 1.22745919e+00 -1.76952556e-02
-4.71427977e-01 2.07136199e-02 9.45069313e-01 -3.46732169e-01
3.35672170e-01 -1.50592649e+00 6.71737075e-01 3.22828144e-01
6.21149736e-03 1.05735219e+00 9.62767363e-01 -7.73355126e-01
-1.70083928e+00 -7.39004254e-01 5.32345653e-01 -6.16769075e-01
2.74552107e-01 -7.02182233e-01 -1.23576283e+00 6.55297577e-01
9.24566388e-02 2.85452038e-01 9.60551441e-01 6.31997138e-02
-2.14169845e-01 7.32567310e-01 -1.36691940e+00 1.75338209e-01
8.58151436e-01 -9.28828895e-01 -5.78477144e-01 4.49844182e-01
5.55604756e-01 -2.53761977e-01 -1.25178289e+00 4.57353294e-02
5.83070815e-01 -1.29522538e+00 1.21832573e+00 -1.10498047e+00
3.53742748e-01 -4.59396571e-01 -4.57185954e-01 -1.14443684e+00
-6.54597938e-01 -5.87823749e-01 -2.82624483e-01 2.29926750e-01
2.74751872e-01 -3.62235904e-01 1.23131990e+00 4.41667944e-01
-1.62522241e-01 -8.31661880e-01 -1.01503956e+00 -3.56159627e-01
3.20072979e-01 -3.10031641e-02 2.53637731e-01 7.92541087e-01
-8.36098343e-02 4.50654328e-01 -3.57083917e-01 1.67576596e-01
1.05790377e+00 3.97432625e-01 6.42826378e-01 -1.27077079e+00
-8.31835210e-01 1.28655955e-01 -2.86271065e-01 -1.28618920e+00
5.30880630e-01 -6.38306677e-01 6.39628991e-02 -1.22449481e+00
7.65062988e-01 1.19943373e-01 5.72197735e-02 -2.33344063e-01
8.18706229e-02 -5.68131693e-02 -2.36741960e-01 5.44750929e-01
-7.58909702e-01 6.79619908e-01 1.21215093e+00 -1.21003781e-02
1.28621817e-01 -1.41673654e-01 -2.02014163e-01 4.40175682e-01
4.90587592e-01 -6.30721748e-01 -4.11258712e-02 -3.49362497e-03
6.14738204e-02 2.62274116e-01 2.96590298e-01 -6.76589549e-01
7.25465361e-03 -2.06693321e-01 4.93901134e-01 -1.05174935e+00
7.35246241e-01 -5.39482951e-01 6.00839078e-01 3.04544121e-01
-9.69602168e-02 2.66480088e-01 -2.19245136e-01 7.95944929e-01
-1.75806329e-01 -2.41198484e-02 1.13059509e+00 -8.52898240e-01
-1.03350624e-01 4.56819981e-01 -9.57370698e-01 -1.31422192e-01
5.09096205e-01 -1.26947150e-01 -5.48874848e-02 -1.01506807e-01
-1.43984592e+00 -2.23177657e-01 1.05436862e+00 -5.04601121e-01
8.22405875e-01 -9.32060361e-01 -3.97117257e-01 2.92025805e-01
-2.45747313e-01 2.95099318e-01 5.15419722e-01 5.48255920e-01
-7.28482902e-01 5.93180835e-01 -3.20059001e-01 -1.02612019e+00
-1.21903265e+00 7.76664495e-01 7.56353021e-01 -4.40592021e-01
-8.08872819e-01 4.51154798e-01 3.01077902e-01 -1.12365091e+00
-2.54402071e-01 -4.46818098e-02 2.30465844e-01 -6.06706381e-01
5.84980309e-01 -1.23588838e-01 1.02257095e-01 -7.71914542e-01
6.43976629e-02 5.18193603e-01 -5.13486743e-01 -2.99539156e-02
1.75683403e+00 -1.69929847e-01 -5.03787756e-01 5.66792548e-01
1.34201121e+00 -6.70667887e-01 -1.70348155e+00 -2.83795446e-01
-3.38988364e-01 1.59808412e-01 -2.14536279e-01 -3.04315418e-01
-5.78270197e-01 1.15968490e+00 5.13564587e-01 -2.20553145e-01
5.39617002e-01 2.59589314e-01 7.41656005e-01 1.02811456e+00
6.12787843e-01 -8.65750909e-01 -2.03332931e-01 5.48291743e-01
2.14078411e-01 -1.21476066e+00 2.02035662e-02 2.23918304e-01
-7.67935961e-02 1.02898514e+00 1.58333465e-01 -2.25865349e-01
5.94096482e-01 5.81503034e-01 -5.88763282e-02 -6.64889395e-01
-1.10195124e+00 2.88572580e-01 -3.74277711e-01 8.14732611e-01
6.82826102e-01 1.58648387e-01 1.08222649e-01 4.06477630e-01
1.49836034e-01 -1.49019912e-01 2.27823466e-01 1.10962987e+00
-6.83164060e-01 -1.36322165e+00 -4.04126108e-01 3.15055460e-01
-6.20799780e-01 2.54094452e-02 -5.44440627e-01 2.77920544e-01
-1.23908952e-01 2.89425790e-01 2.07020476e-01 -1.99009642e-01
-7.09564090e-02 4.92318273e-01 9.20415699e-01 -4.28283304e-01
1.44003974e-02 1.22698888e-01 -3.11479002e-01 -4.48994845e-01
-7.50239372e-01 -7.14446545e-01 -1.46866000e+00 -6.48101330e-01
-2.49447390e-01 5.81819713e-01 7.53275931e-01 9.96487200e-01
7.78811097e-01 1.85632855e-01 4.35384363e-01 -1.35371733e+00
-4.60366845e-01 -9.93524075e-01 -9.29932535e-01 6.46747708e-01
4.57098037e-01 -6.95959568e-01 -3.64513159e-01 1.71278909e-01]
|
[13.295774459838867, -3.069888114929199]
|
652408aa-ae3f-44b5-a3bf-9603408dd9d3
|
model-aided-federated-reinforcement-learning
|
2306.02029
| null |
https://arxiv.org/abs/2306.02029v1
|
https://arxiv.org/pdf/2306.02029v1.pdf
|
Model-aided Federated Reinforcement Learning for Multi-UAV Trajectory Planning in IoT Networks
|
Deploying teams of cooperative unmanned aerial vehicles (UAVs) to harvest data from distributed Internet of Things (IoT) devices requires efficient trajectory planning and coordination algorithms. Multi-agent reinforcement learning (MARL) has emerged as an effective solution, but often requires extensive and costly real-world training data. In this paper, we propose a novel model-aided federated MARL algorithm to coordinate multiple UAVs on a data harvesting mission with limited knowledge about the environment, significantly reducing the real-world training data demand. The proposed algorithm alternates between learning an environment model from real-world measurements and federated QMIX training in the simulated environment. Specifically, collected measurements from the real-world environment are used to learn the radio channel and estimate unknown IoT device locations to create a simulated environment. Each UAV agent trains a local QMIX model in its simulated environment and continuously consolidates it through federated learning with other agents, accelerating the learning process and further improving training sample efficiency. Simulation results demonstrate that our proposed model-aided FedQMIX algorithm substantially reduces the need for real-world training experiences while attaining similar data collection performance as standard MARL algorithms.
|
['Marco Caccamo', 'David Gesbert', 'Harald Bayerlein', 'Omid Esrafilian', 'Jichao Chen']
|
2023-06-03
| null | null | null | null |
['multi-agent-reinforcement-learning', 'trajectory-planning']
|
['methodology', 'robots']
|
[-1.25672385e-01 -1.18150547e-01 -2.45918017e-02 8.80517960e-02
-5.43034077e-01 -8.23223650e-01 2.44175762e-01 2.22787499e-01
-5.20290196e-01 1.20052838e+00 -5.36213577e-01 -3.94789368e-01
-7.34061658e-01 -1.09108722e+00 -7.57722020e-01 -1.12926769e+00
-6.51668966e-01 6.41170561e-01 -1.34447679e-01 -1.38428822e-01
-4.46021914e-01 4.79064822e-01 -1.45781434e+00 -6.50134504e-01
9.98944521e-01 1.21471238e+00 7.87128031e-01 9.21603739e-01
3.52304697e-01 9.48865354e-01 -9.51636493e-01 2.47542888e-01
6.04090571e-01 -1.12844273e-01 -4.07401264e-01 1.24310233e-01
-3.55039686e-01 -4.61948305e-01 -1.48175955e-01 8.66347253e-01
6.07650876e-01 2.00777113e-01 6.90086633e-02 -1.73800552e+00
-1.39451399e-01 5.18762469e-01 -3.34173620e-01 -1.94742650e-01
-7.49032423e-02 2.70492345e-01 4.00276959e-01 -4.20945249e-02
3.35370332e-01 7.07527995e-01 4.87052768e-01 3.00652146e-01
-5.78076243e-01 -8.16560507e-01 4.57769066e-01 4.53060464e-04
-1.19895792e+00 -1.79017372e-02 5.35764158e-01 1.15242347e-01
6.67907774e-01 5.56889325e-02 1.15551293e+00 5.92072964e-01
6.21491015e-01 6.09110117e-01 7.88411021e-01 -2.87152559e-01
7.93790400e-01 -1.47608995e-01 -7.48432994e-01 1.06857431e+00
5.92765272e-01 4.87659395e-01 -2.21024781e-01 -3.26890528e-01
4.71732736e-01 2.05261871e-01 6.10723980e-02 -4.84951228e-01
-1.43989980e+00 4.40254450e-01 7.31969655e-01 -5.25213331e-02
-1.21476471e+00 4.30174321e-01 8.69666338e-02 6.17669463e-01
1.09388061e-01 5.55104136e-01 -8.61125171e-01 2.16216981e-01
-4.21618521e-01 3.55176181e-02 7.07379401e-01 1.40118587e+00
9.93060112e-01 4.05909956e-01 4.17976797e-01 1.18936233e-01
4.24432129e-01 1.47274268e+00 1.97208628e-01 -1.24900162e+00
3.84242177e-01 5.74789941e-01 7.56442308e-01 -5.56547463e-01
-7.39044726e-01 -5.34645081e-01 -8.12297523e-01 2.20034778e-01
5.09613343e-02 -1.34620821e+00 -4.94821846e-01 1.60565054e+00
9.94853675e-01 3.87725741e-01 6.18364155e-01 8.26682746e-01
-1.63248852e-02 9.03469384e-01 -2.27090213e-02 -6.45551980e-01
8.28469515e-01 -7.06102014e-01 -5.00461876e-01 -1.93603590e-01
6.43125057e-01 -2.88368911e-01 2.60968864e-01 4.81138587e-01
-6.41390800e-01 -3.15348536e-01 -1.11373985e+00 1.07972503e+00
-4.45365131e-01 4.33749318e-01 1.05533648e+00 4.74553257e-01
-9.18059707e-01 2.89103359e-01 -7.87946820e-01 -3.50591600e-01
2.79347837e-01 8.63908291e-01 5.71617521e-02 -2.55076498e-01
-8.21774602e-01 6.22108877e-01 4.09735143e-01 -4.67832088e-02
-1.62998235e+00 -5.94046772e-01 -5.56570232e-01 -3.12283218e-01
7.20341682e-01 -8.98908079e-01 1.44255269e+00 -4.30647522e-01
-1.69878197e+00 -2.39944950e-01 4.66033101e-01 -5.78953505e-01
-4.24629189e-02 3.06118047e-03 -4.88773465e-01 7.97628462e-02
9.94927362e-02 5.54322302e-01 9.35922205e-01 -1.35782444e+00
-1.23222399e+00 -3.80182236e-01 3.54489088e-01 4.23757523e-01
-2.86825389e-01 -5.81678331e-01 3.90996575e-01 -1.02596857e-01
-1.43755183e-01 -1.08538651e+00 -7.13238060e-01 -7.15223104e-02
1.81104049e-01 -1.83727458e-01 1.10903609e+00 2.99892146e-02
4.88647521e-01 -1.65808117e+00 7.92560652e-02 1.65959388e-01
6.98439777e-02 -2.56271679e-02 -4.19420898e-01 8.14237058e-01
8.77518415e-01 -4.70836699e-01 2.82234550e-01 -2.51182258e-01
-7.85004199e-02 5.82361400e-01 -3.07697177e-01 3.41065824e-01
-2.71958798e-01 5.95210552e-01 -1.42562449e+00 -2.25994855e-01
3.08223605e-01 1.46337867e-01 -1.23577863e-01 3.06858063e-01
-7.35947430e-01 8.25396597e-01 -9.46375191e-01 8.90818715e-01
7.06638753e-01 -2.03247607e-01 5.78934133e-01 3.92786004e-02
-4.40520316e-01 -4.91032928e-01 -1.01664305e+00 1.85341990e+00
-9.84285414e-01 1.52860433e-01 6.59696102e-01 -8.61116290e-01
8.96756172e-01 2.97532111e-01 1.24041510e+00 -5.90066791e-01
5.18411875e-01 6.92156628e-02 -2.01156512e-01 -5.11417925e-01
3.65672171e-01 8.92109200e-02 -3.74488175e-01 6.02961183e-01
2.46483594e-01 -3.80798280e-01 -1.20192699e-01 1.55883804e-02
1.39649439e+00 -4.72745933e-02 3.11323196e-01 -5.02231792e-02
2.84347415e-01 3.85387033e-01 7.59190977e-01 8.37383091e-01
-2.81242132e-01 -7.53701806e-01 -5.85605025e-01 -8.02738488e-01
-5.58642387e-01 -1.08738041e+00 5.80535114e-01 9.83599365e-01
6.89143062e-01 -2.01015338e-01 -3.90273809e-01 -6.79478943e-01
1.17898621e-01 4.95415032e-01 -2.87308753e-01 -6.28714189e-02
-1.76557168e-01 -8.47261786e-01 2.42639765e-01 4.02616300e-02
8.82986009e-01 -7.40811348e-01 -1.41354549e+00 5.60924590e-01
6.03373069e-03 -1.10539448e+00 -4.37329672e-02 4.72550690e-01
-6.05075121e-01 -1.29908323e+00 -2.27196157e-01 -7.17835128e-01
9.44876075e-01 7.71995962e-01 6.33942902e-01 -3.18359099e-02
-3.44841242e-01 1.06233239e+00 -5.21705210e-01 -1.25490129e+00
-2.77359366e-01 -2.22099144e-02 6.73007309e-01 -1.80723704e-03
1.50914807e-02 -5.73313236e-01 -4.63012278e-01 3.73543650e-01
-6.57435596e-01 -2.21394166e-01 8.70609283e-01 7.08350718e-01
6.87570989e-01 8.32082331e-01 9.06135976e-01 -2.89124429e-01
5.04051328e-01 -6.32780731e-01 -1.31216145e+00 4.93621618e-01
-5.18888354e-01 -4.07022536e-02 1.06539476e+00 -5.28037488e-01
-7.63299465e-01 5.19885361e-01 6.04022443e-01 -5.01580000e-01
-9.19310525e-02 4.27335590e-01 -1.18429698e-02 -5.94822109e-01
3.44839126e-01 2.07495898e-01 -5.69106936e-02 6.60140738e-02
2.32817695e-01 7.24223375e-01 2.72452563e-01 -5.95288694e-01
1.24640489e+00 5.07487655e-01 3.87042791e-01 -7.82942951e-01
-8.60499680e-01 -2.21993685e-01 -3.54687750e-01 -6.66256249e-01
6.07594967e-01 -1.07911992e+00 -1.18631005e+00 3.32282633e-01
-9.75885272e-01 -7.03562737e-01 -5.79179287e-01 9.27763164e-01
-6.15754962e-01 -2.36121058e-01 1.81473210e-01 -1.12378168e+00
-4.36791331e-01 -9.29184139e-01 9.66608644e-01 3.37646723e-01
6.48610353e-01 -9.35972333e-01 3.84078264e-01 -1.01170860e-01
5.56459785e-01 5.05823195e-01 4.12719190e-01 7.80192763e-02
-9.61706281e-01 -2.65231490e-01 4.32107329e-01 -2.28002980e-01
4.17692125e-01 -5.02554893e-01 -5.56170940e-01 -9.49670076e-01
-8.75812098e-02 -4.43708122e-01 -4.41474132e-02 4.05867368e-01
7.18277752e-01 -5.37301481e-01 -6.28534853e-01 3.84872109e-01
1.55588949e+00 5.50303698e-01 -3.32169414e-01 1.29142046e-01
2.17761800e-01 1.70561522e-01 1.32483065e+00 1.13379514e+00
6.73836946e-01 3.65595222e-01 1.30745912e+00 6.59953356e-02
5.09626091e-01 -3.28358859e-01 4.67641532e-01 7.86143184e-01
1.25906870e-01 -4.66134757e-01 -5.30226588e-01 3.26994091e-01
-2.19502687e+00 -7.47298777e-01 4.95331287e-01 2.12683487e+00
-8.62232149e-02 -4.71948147e-01 1.34447858e-01 -1.97663128e-01
3.73724073e-01 -8.69600847e-02 -1.14351869e+00 2.48417422e-01
7.88158402e-02 -1.24249101e-01 1.10764968e+00 3.05593133e-01
-1.01890683e+00 8.19197536e-01 5.13363981e+00 1.53151631e-01
-1.16162229e+00 2.33048812e-01 -3.13734293e-01 -1.46012932e-01
1.29216254e-01 -3.17814238e-02 -3.93208444e-01 1.60782292e-01
1.11782265e+00 -3.36643904e-01 1.03730989e+00 9.49484468e-01
4.24059778e-01 -2.35565141e-01 -6.02311969e-01 1.11988950e+00
-3.26655686e-01 -1.34643030e+00 -2.83864945e-01 3.81138355e-01
8.85291636e-01 6.26332045e-01 -1.67770877e-01 3.93117189e-01
1.18976736e+00 -2.21621439e-01 6.02152526e-01 4.91801858e-01
4.74646866e-01 -9.34911966e-01 7.93067634e-01 8.96063507e-01
-1.60338628e+00 -9.92753267e-01 -6.36641800e-01 -1.94289386e-01
6.77637681e-02 4.26910490e-01 -1.11995888e+00 1.12284768e+00
6.72234058e-01 5.86191714e-01 -2.22549766e-01 1.17656291e+00
2.87556238e-02 4.63463247e-01 -4.92490351e-01 -6.00717187e-01
4.83050317e-01 -3.09093982e-01 8.28363538e-01 1.18291266e-01
6.86941087e-01 2.55915582e-01 8.42192054e-01 2.14980051e-01
-3.98224918e-03 -2.23666504e-01 -9.43461895e-01 7.39773437e-02
8.91367376e-01 1.48658073e+00 -3.88372183e-01 -6.74993992e-02
-2.88962543e-01 5.81458449e-01 9.89639834e-02 2.91527897e-01
-8.50450695e-01 -3.36561114e-01 7.51133442e-01 -5.02460480e-01
2.17511743e-01 -5.26771605e-01 5.46160042e-01 -7.51508653e-01
-3.39540094e-01 -5.50836921e-01 2.05501750e-01 -7.46185839e-01
-1.03043818e+00 6.72853708e-01 -2.26390377e-01 -1.77740633e+00
-3.10186893e-01 -3.85214329e-01 -4.84458327e-01 1.38965219e-01
-1.56164193e+00 -1.39645278e+00 -6.08113348e-01 8.19864988e-01
3.51905316e-01 -5.67622662e-01 1.14441645e+00 -1.29723176e-01
-4.85776067e-01 -6.27890676e-02 2.46477634e-01 -3.36752176e-01
3.04985762e-01 -1.04109800e+00 -1.88750952e-01 6.95401192e-01
1.59560546e-01 -1.08027291e-02 5.98749578e-01 -6.87249005e-01
-2.26117635e+00 -1.74225473e+00 -1.69057786e-01 -3.38276736e-02
6.03832722e-01 -8.55241567e-02 2.35722423e-01 5.34710169e-01
3.54122877e-01 1.60206527e-01 5.59534252e-01 -3.65252644e-01
4.35212076e-01 -8.16717684e-01 -1.47756648e+00 3.34559411e-01
8.26634526e-01 9.53971818e-02 3.57069105e-01 6.74149394e-01
8.98280084e-01 -1.44756138e-01 -9.78781343e-01 3.74028862e-01
2.95227528e-01 -1.28362730e-01 4.36265796e-01 -5.52479267e-01
-9.69966412e-01 -7.33641386e-01 -4.41366255e-01 -2.02076030e+00
-1.47348240e-01 -1.17422831e+00 -2.42857948e-01 7.54985869e-01
1.76912278e-01 -8.59811246e-01 7.69524693e-01 1.45056888e-01
-2.26250961e-01 -5.05617678e-01 -1.13303006e+00 -9.58994627e-01
-3.39413017e-01 -3.02797049e-01 1.16202891e+00 5.69024742e-01
-8.14540088e-02 1.97762504e-01 -5.34605861e-01 1.10067105e+00
1.18841314e+00 2.01183736e-01 1.14150083e+00 -1.41925490e+00
-2.36505240e-01 5.99490166e-01 -2.59115964e-01 -8.88780892e-01
1.95794061e-01 -5.74760675e-01 2.12366313e-01 -1.67627764e+00
-5.92819512e-01 -1.04805863e+00 -2.24056795e-01 5.54121077e-01
5.02665877e-01 -3.88538450e-01 2.39992037e-01 -6.92456812e-02
-1.19092703e+00 8.27366054e-01 1.19889557e+00 -5.42342663e-01
-3.22217613e-01 4.77578878e-01 -3.16408575e-01 4.25403655e-01
1.22310317e+00 -5.91271639e-01 -1.09904301e+00 -9.36196923e-01
1.62734166e-01 5.59237182e-01 3.08037102e-01 -1.78073978e+00
7.43614495e-01 -5.31018853e-01 3.02753180e-01 -6.35235548e-01
4.07954574e-01 -1.56340265e+00 3.04835647e-01 7.57579625e-01
2.25135610e-01 3.79699498e-01 2.39268437e-01 1.04311645e+00
1.92803130e-01 2.63198689e-02 3.83078068e-01 -1.64120555e-01
-7.41048276e-01 6.97570801e-01 -6.07233107e-01 -3.65698069e-01
1.67484844e+00 2.22585291e-01 -3.49359602e-01 -5.13451219e-01
-5.00951469e-01 9.88511384e-01 3.43990386e-01 2.31576577e-01
7.51841486e-01 -1.00648785e+00 -3.34266156e-01 2.44248971e-01
-8.47003236e-02 1.00263305e-01 2.25370541e-01 5.18816411e-01
-1.49308830e-01 3.24925452e-01 -1.89688161e-01 -5.02484322e-01
-9.05951142e-01 7.52857208e-01 4.47518557e-01 -2.06087351e-01
-1.70391142e-01 4.11821723e-01 -6.39551342e-01 -8.43574524e-01
2.07294285e-01 -3.47422779e-01 3.94759268e-01 -3.24334890e-01
5.02977788e-01 4.54624116e-01 -1.11304671e-01 -1.57175124e-01
-2.79241204e-01 4.38793629e-01 3.82971436e-01 -2.56567560e-02
1.55120957e+00 -4.59925234e-01 9.97977406e-02 3.80534567e-02
5.04778326e-01 -1.65943846e-01 -1.54293263e+00 -1.52911186e-01
-2.74242997e-01 -1.26627356e-01 4.25967067e-01 -9.48295832e-01
-1.09071672e+00 2.09422931e-01 7.49278963e-01 3.52565616e-01
1.33644235e+00 -1.42966837e-01 8.57398689e-01 1.13395488e+00
1.72301757e+00 -1.17566621e+00 9.31180418e-02 3.01472813e-01
3.85081977e-01 -1.09638512e+00 -7.40010068e-02 2.17038603e-03
-5.43770075e-01 9.76394534e-01 6.42224073e-01 -1.00742988e-01
8.54176164e-01 4.54560012e-01 2.00353369e-01 -2.83055782e-01
-1.05535555e+00 -3.77231270e-01 -7.33788192e-01 1.23783326e+00
-9.56033587e-01 4.22083676e-01 6.73578978e-01 3.68995786e-01
-8.26474875e-02 1.64620087e-01 4.76369947e-01 1.25271571e+00
-7.64851868e-01 -9.73377943e-01 -3.29260975e-01 5.54194927e-01
3.53019059e-01 5.12765110e-01 -4.24221121e-02 6.87983274e-01
4.28580642e-01 1.38713753e+00 -1.01666898e-01 -6.64914012e-01
2.03073174e-01 -6.78144574e-01 5.47297597e-01 -2.78134018e-01
-3.03003907e-01 -3.17480922e-01 -1.96084037e-01 -5.14559329e-01
-6.17740691e-01 -5.38455367e-01 -1.37100494e+00 -1.16479069e-01
-5.06036222e-01 6.51996076e-01 9.77050543e-01 9.10532951e-01
7.10162640e-01 6.95739985e-01 1.29750550e+00 -1.04584098e+00
-7.40485430e-01 -6.27245545e-01 -5.93244016e-01 -8.42106342e-01
6.25152707e-01 -8.26056659e-01 -2.85745617e-02 -4.09900546e-01]
|
[5.777977466583252, 1.5890781879425049]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.